diff options
Diffstat (limited to 'sys')
269 files changed, 6023 insertions, 22507 deletions
diff --git a/sys/amd64/amd64/trap.c b/sys/amd64/amd64/trap.c index 4880c91..67842cf 100644 --- a/sys/amd64/amd64/trap.c +++ b/sys/amd64/amd64/trap.c @@ -614,7 +614,8 @@ trap_check(struct trapframe *frame) { #ifdef KDTRACE_HOOKS - if (dtrace_trap_func != NULL && (*dtrace_trap_func)(frame)) + if (dtrace_trap_func != NULL && + (*dtrace_trap_func)(frame, frame->tf_trapno) != 0) return; #endif trap(frame); diff --git a/sys/amd64/conf/NOTES b/sys/amd64/conf/NOTES index d1220b6..9b697f0 100644 --- a/sys/amd64/conf/NOTES +++ b/sys/amd64/conf/NOTES @@ -215,7 +215,7 @@ hint.atkbd.0.irq="1" # Options for atkbd: options ATKBD_DFLT_KEYMAP # specify the built-in keymap -makeoptions ATKBD_DFLT_KEYMAP=jp.106 +makeoptions ATKBD_DFLT_KEYMAP=fr.dvorak # `flags' for atkbd: # 0x01 Force detection of keyboard, else we always assume a keyboard diff --git a/sys/amd64/include/vmm.h b/sys/amd64/include/vmm.h index 0f19110..8a8c3f4 100644 --- a/sys/amd64/include/vmm.h +++ b/sys/amd64/include/vmm.h @@ -358,6 +358,8 @@ void vm_copyin(struct vm *vm, int vcpuid, struct vm_copyinfo *copyinfo, void *kaddr, size_t len); void vm_copyout(struct vm *vm, int vcpuid, const void *kaddr, struct vm_copyinfo *copyinfo, size_t len); + +int vcpu_trace_exceptions(struct vm *vm, int vcpuid); #endif /* KERNEL */ #define VM_MAXCPU 16 /* maximum virtual cpus */ diff --git a/sys/amd64/vmm/amd/svm.c b/sys/amd64/vmm/amd/svm.c index 753799a..7d75046 100644 --- a/sys/amd64/vmm/amd/svm.c +++ b/sys/amd64/vmm/amd/svm.c @@ -46,6 +46,7 @@ __FBSDID("$FreeBSD$"); #include <machine/specialreg.h> #include <machine/smp.h> #include <machine/vmm.h> +#include <machine/vmm_dev.h> #include <machine/vmm_instruction_emul.h> #include "vmm_lapic.h" @@ -429,8 +430,24 @@ vmcb_init(struct svm_softc *sc, int vcpu, uint64_t iopm_base_pa, svm_enable_intercept(sc, vcpu, VMCB_CR_INTCPT, mask); } - /* Intercept Machine Check exceptions. */ - svm_enable_intercept(sc, vcpu, VMCB_EXC_INTCPT, BIT(IDT_MC)); + + /* + * Intercept everything when tracing guest exceptions otherwise + * just intercept machine check exception. + */ + if (vcpu_trace_exceptions(sc->vm, vcpu)) { + for (n = 0; n < 32; n++) { + /* + * Skip unimplemented vectors in the exception bitmap. + */ + if (n == 2 || n == 9) { + continue; + } + svm_enable_intercept(sc, vcpu, VMCB_EXC_INTCPT, BIT(n)); + } + } else { + svm_enable_intercept(sc, vcpu, VMCB_EXC_INTCPT, BIT(IDT_MC)); + } /* Intercept various events (for e.g. I/O, MSR and CPUID accesses) */ svm_enable_intercept(sc, vcpu, VMCB_CTRL1_INTCPT, VMCB_INTCPT_IO); @@ -1176,9 +1193,10 @@ svm_vmexit(struct svm_softc *svm_sc, int vcpu, struct vm_exit *vmexit) struct vmcb_state *state; struct vmcb_ctrl *ctrl; struct svm_regctx *ctx; + struct vm_exception exception; uint64_t code, info1, info2, val; uint32_t eax, ecx, edx; - int handled; + int error, errcode_valid, handled, idtvec, reflect; bool retu; ctx = svm_get_guest_regctx(svm_sc, vcpu); @@ -1237,8 +1255,78 @@ svm_vmexit(struct svm_softc *svm_sc, int vcpu, struct vm_exit *vmexit) case VMCB_EXIT_NMI: /* external NMI */ handled = 1; break; - case VMCB_EXIT_MC: /* machine check */ + case 0x40 ... 0x5F: vmm_stat_incr(svm_sc->vm, vcpu, VMEXIT_EXCEPTION, 1); + reflect = 1; + idtvec = code - 0x40; + switch (idtvec) { + case IDT_MC: + /* + * Call the machine check handler by hand. Also don't + * reflect the machine check back into the guest. + */ + reflect = 0; + VCPU_CTR0(svm_sc->vm, vcpu, "Vectoring to MCE handler"); + __asm __volatile("int $18"); + break; + case IDT_PF: + error = svm_setreg(svm_sc, vcpu, VM_REG_GUEST_CR2, + info2); + KASSERT(error == 0, ("%s: error %d updating cr2", + __func__, error)); + /* fallthru */ + case IDT_NP: + case IDT_SS: + case IDT_GP: + case IDT_AC: + case IDT_TS: + errcode_valid = 1; + break; + + case IDT_DF: + errcode_valid = 1; + info1 = 0; + break; + + case IDT_BP: + case IDT_OF: + case IDT_BR: + /* + * The 'nrip' field is populated for INT3, INTO and + * BOUND exceptions and this also implies that + * 'inst_length' is non-zero. + * + * Reset 'inst_length' to zero so the guest %rip at + * event injection is identical to what it was when + * the exception originally happened. + */ + VCPU_CTR2(svm_sc->vm, vcpu, "Reset inst_length from %d " + "to zero before injecting exception %d", + vmexit->inst_length, idtvec); + vmexit->inst_length = 0; + /* fallthru */ + default: + errcode_valid = 0; + break; + } + KASSERT(vmexit->inst_length == 0, ("invalid inst_length (%d) " + "when reflecting exception %d into guest", + vmexit->inst_length, idtvec)); + + if (reflect) { + /* Reflect the exception back into the guest */ + exception.vector = idtvec; + exception.error_code_valid = errcode_valid; + exception.error_code = errcode_valid ? info1 : 0; + VCPU_CTR2(svm_sc->vm, vcpu, "Reflecting exception " + "%d/%#x into the guest", exception.vector, + exception.error_code); + error = vm_inject_exception(svm_sc->vm, vcpu, + &exception); + KASSERT(error == 0, ("%s: vm_inject_exception error %d", + __func__, error)); + } + handled = 1; break; case VMCB_EXIT_MSR: /* MSR access. */ eax = state->rax; diff --git a/sys/amd64/vmm/intel/vmcs.c b/sys/amd64/vmm/intel/vmcs.c index 51e5c2c..ae4d9db 100644 --- a/sys/amd64/vmm/intel/vmcs.c +++ b/sys/amd64/vmm/intel/vmcs.c @@ -332,7 +332,6 @@ vmcs_init(struct vmcs *vmcs) int error, codesel, datasel, tsssel; u_long cr0, cr4, efer; uint64_t pat, fsbase, idtrbase; - uint32_t exc_bitmap; codesel = vmm_get_host_codesel(); datasel = vmm_get_host_datasel(); @@ -417,11 +416,6 @@ vmcs_init(struct vmcs *vmcs) if ((error = vmwrite(VMCS_HOST_RIP, (u_long)vmx_exit_guest)) != 0) goto done; - /* exception bitmap */ - exc_bitmap = 1 << IDT_MC; - if ((error = vmwrite(VMCS_EXCEPTION_BITMAP, exc_bitmap)) != 0) - goto done; - /* link pointer */ if ((error = vmwrite(VMCS_LINK_POINTER, ~0)) != 0) goto done; diff --git a/sys/amd64/vmm/intel/vmcs.h b/sys/amd64/vmm/intel/vmcs.h index 6122de5..6d78a69 100644 --- a/sys/amd64/vmm/intel/vmcs.h +++ b/sys/amd64/vmm/intel/vmcs.h @@ -321,7 +321,7 @@ vmcs_write(uint32_t encoding, uint64_t val) #define EXIT_REASON_MTF 37 #define EXIT_REASON_MONITOR 39 #define EXIT_REASON_PAUSE 40 -#define EXIT_REASON_MCE 41 +#define EXIT_REASON_MCE_DURING_ENTRY 41 #define EXIT_REASON_TPR 43 #define EXIT_REASON_APIC_ACCESS 44 #define EXIT_REASON_VIRTUALIZED_EOI 45 diff --git a/sys/amd64/vmm/intel/vmx.c b/sys/amd64/vmm/intel/vmx.c index c855697..c3dd04e 100644 --- a/sys/amd64/vmm/intel/vmx.c +++ b/sys/amd64/vmm/intel/vmx.c @@ -283,8 +283,8 @@ exit_reason_to_str(int reason) return "monitor"; case EXIT_REASON_PAUSE: return "pause"; - case EXIT_REASON_MCE: - return "mce"; + case EXIT_REASON_MCE_DURING_ENTRY: + return "mce-during-entry"; case EXIT_REASON_TPR: return "tpr"; case EXIT_REASON_APIC_ACCESS: @@ -821,6 +821,7 @@ vmx_vminit(struct vm *vm, pmap_t pmap) int i, error; struct vmx *vmx; struct vmcs *vmcs; + uint32_t exc_bitmap; vmx = malloc(sizeof(struct vmx), M_VMX, M_WAITOK | M_ZERO); if ((uintptr_t)vmx & PAGE_MASK) { @@ -911,6 +912,14 @@ vmx_vminit(struct vm *vm, pmap_t pmap) error += vmwrite(VMCS_ENTRY_CTLS, entry_ctls); error += vmwrite(VMCS_MSR_BITMAP, vtophys(vmx->msr_bitmap)); error += vmwrite(VMCS_VPID, vpid[i]); + + /* exception bitmap */ + if (vcpu_trace_exceptions(vm, i)) + exc_bitmap = 0xffffffff; + else + exc_bitmap = 1 << IDT_MC; + error += vmwrite(VMCS_EXCEPTION_BITMAP, exc_bitmap); + if (virtual_interrupt_delivery) { error += vmwrite(VMCS_APIC_ACCESS, APIC_ACCESS_ADDRESS); error += vmwrite(VMCS_VIRTUAL_APIC, @@ -2056,8 +2065,9 @@ vmx_exit_process(struct vmx *vmx, int vcpu, struct vm_exit *vmexit) struct vlapic *vlapic; struct vm_inout_str *vis; struct vm_task_switch *ts; + struct vm_exception vmexc; uint32_t eax, ecx, edx, idtvec_info, idtvec_err, intr_info, inst_info; - uint32_t intr_type, reason; + uint32_t intr_type, intr_vec, reason; uint64_t exitintinfo, qual, gpa; bool retu; @@ -2074,6 +2084,18 @@ vmx_exit_process(struct vmx *vmx, int vcpu, struct vm_exit *vmexit) vmm_stat_incr(vmx->vm, vcpu, VMEXIT_COUNT, 1); /* + * VM-entry failures during or after loading guest state. + * + * These VM-exits are uncommon but must be handled specially + * as most VM-exit fields are not populated as usual. + */ + if (__predict_false(reason == EXIT_REASON_MCE_DURING_ENTRY)) { + VCPU_CTR0(vmx->vm, vcpu, "Handling MCE during VM-entry"); + __asm __volatile("int $18"); + return (1); + } + + /* * VM exits that can be triggered during event delivery need to * be handled specially by re-injecting the event if the IDT * vectoring information field's valid bit is set. @@ -2305,6 +2327,9 @@ vmx_exit_process(struct vmx *vmx, int vcpu, struct vm_exit *vmexit) KASSERT((intr_info & VMCS_INTR_VALID) != 0, ("VM exit interruption info invalid: %#x", intr_info)); + intr_vec = intr_info & 0xff; + intr_type = intr_info & VMCS_INTR_T_MASK; + /* * If Virtual NMIs control is 1 and the VM-exit is due to a * fault encountered during the execution of IRET then we must @@ -2315,16 +2340,55 @@ vmx_exit_process(struct vmx *vmx, int vcpu, struct vm_exit *vmexit) * See "Information for VM Exits Due to Vectored Events". */ if ((idtvec_info & VMCS_IDT_VEC_VALID) == 0 && - (intr_info & 0xff) != IDT_DF && + (intr_vec != IDT_DF) && (intr_info & EXIT_QUAL_NMIUDTI) != 0) vmx_restore_nmi_blocking(vmx, vcpu); /* * The NMI has already been handled in vmx_exit_handle_nmi(). */ - if ((intr_info & VMCS_INTR_T_MASK) == VMCS_INTR_T_NMI) + if (intr_type == VMCS_INTR_T_NMI) return (1); - break; + + /* + * Call the machine check handler by hand. Also don't reflect + * the machine check back into the guest. + */ + if (intr_vec == IDT_MC) { + VCPU_CTR0(vmx->vm, vcpu, "Vectoring to MCE handler"); + __asm __volatile("int $18"); + return (1); + } + + if (intr_vec == IDT_PF) { + error = vmxctx_setreg(vmxctx, VM_REG_GUEST_CR2, qual); + KASSERT(error == 0, ("%s: vmxctx_setreg(cr2) error %d", + __func__, error)); + } + + /* + * Software exceptions exhibit trap-like behavior. This in + * turn requires populating the VM-entry instruction length + * so that the %rip in the trap frame is past the INT3/INTO + * instruction. + */ + if (intr_type == VMCS_INTR_T_SWEXCEPTION) + vmcs_write(VMCS_ENTRY_INST_LENGTH, vmexit->inst_length); + + /* Reflect all other exceptions back into the guest */ + bzero(&vmexc, sizeof(struct vm_exception)); + vmexc.vector = intr_vec; + if (intr_info & VMCS_INTR_DEL_ERRCODE) { + vmexc.error_code_valid = 1; + vmexc.error_code = vmcs_read(VMCS_EXIT_INTR_ERRCODE); + } + VCPU_CTR2(vmx->vm, vcpu, "Reflecting exception %d/%#x into " + "the guest", vmexc.vector, vmexc.error_code); + error = vm_inject_exception(vmx->vm, vcpu, &vmexc); + KASSERT(error == 0, ("%s: vm_inject_exception error %d", + __func__, error)); + return (1); + case EXIT_REASON_EPT_FAULT: /* * If 'gpa' lies within the address space allocated to diff --git a/sys/amd64/vmm/intel/vmx_msr.c b/sys/amd64/vmm/intel/vmx_msr.c index 746ca73..f6bbf2a 100644 --- a/sys/amd64/vmm/intel/vmx_msr.c +++ b/sys/amd64/vmm/intel/vmx_msr.c @@ -376,9 +376,31 @@ vmx_rdmsr(struct vmx *vmx, int vcpuid, u_int num, uint64_t *val, bool *retu) int vmx_wrmsr(struct vmx *vmx, int vcpuid, u_int num, uint64_t val, bool *retu) { - int error = 0; - + uint64_t changed; + int error; + + error = 0; switch (num) { + case MSR_IA32_MISC_ENABLE: + changed = val ^ misc_enable; + /* + * If the host has disabled the NX feature then the guest + * also cannot use it. However, a Linux guest will try to + * enable the NX feature by writing to the MISC_ENABLE MSR. + * + * This can be safely ignored because the memory management + * code looks at CPUID.80000001H:EDX.NX to check if the + * functionality is actually enabled. + */ + changed &= ~(1UL << 34); + + /* + * Punt to userspace if any other bits are being modified. + */ + if (changed) + error = EINVAL; + + break; default: error = EINVAL; break; diff --git a/sys/amd64/vmm/io/vatpic.c b/sys/amd64/vmm/io/vatpic.c index c6d4590..74a7027 100644 --- a/sys/amd64/vmm/io/vatpic.c +++ b/sys/amd64/vmm/io/vatpic.c @@ -112,6 +112,16 @@ struct vatpic { static void vatpic_set_pinstate(struct vatpic *vatpic, int pin, bool newstate); +static __inline bool +master_atpic(struct vatpic *vatpic, struct atpic *atpic) +{ + + if (atpic == &vatpic->atpic[0]) + return (true); + else + return (false); +} + static __inline int vatpic_get_highest_isrpin(struct atpic *atpic) { @@ -250,6 +260,7 @@ vatpic_icw1(struct vatpic *vatpic, struct atpic *atpic, uint8_t val) atpic->mask = 0; atpic->lowprio = 7; atpic->rd_cmd_reg = 0; + atpic->poll = 0; if ((val & ICW1_SNGL) != 0) { VATPIC_CTR0(vatpic, "vatpic cascade mode required"); @@ -301,6 +312,15 @@ vatpic_icw4(struct vatpic *vatpic, struct atpic *atpic, uint8_t val) if ((val & ICW4_AEOI) != 0) atpic->aeoi = true; + if ((val & ICW4_SFNM) != 0) { + if (master_atpic(vatpic, atpic)) { + atpic->sfn = true; + } else { + VATPIC_CTR1(vatpic, "Ignoring special fully nested " + "mode on slave atpic: %#x", val); + } + } + atpic->icw_num = 0; atpic->ready = true; @@ -354,11 +374,17 @@ vatpic_ocw3(struct vatpic *vatpic, struct atpic *atpic, uint8_t val) { VATPIC_CTR1(vatpic, "atpic ocw3 0x%x", val); - atpic->poll = ((val & OCW3_P) != 0); + if (val & OCW3_ESMM) { + VATPIC_CTR0(vatpic, "atpic special mask mode not implemented"); + return (-1); + } if (val & OCW3_RR) { /* read register command */ atpic->rd_cmd_reg = val & OCW3_RIS; + + /* Polling mode */ + atpic->poll = ((val & OCW3_P) != 0); } return (0); @@ -578,12 +604,19 @@ static int vatpic_read(struct vatpic *vatpic, struct atpic *atpic, bool in, int port, int bytes, uint32_t *eax) { + int pin; + VATPIC_LOCK(vatpic); if (atpic->poll) { - VATPIC_CTR0(vatpic, "vatpic polled mode not supported"); - VATPIC_UNLOCK(vatpic); - return (-1); + atpic->poll = 0; + pin = vatpic_get_highest_irrpin(atpic); + if (pin >= 0) { + vatpic_pin_accepted(atpic, pin); + *eax = 0x80 | pin; + } else { + *eax = 0; + } } else { if (port & ICU_IMR_OFFSET) { /* read interrrupt mask register */ diff --git a/sys/amd64/vmm/vmm.c b/sys/amd64/vmm/vmm.c index e7fbada..d9cb6f3 100644 --- a/sys/amd64/vmm/vmm.c +++ b/sys/amd64/vmm/vmm.c @@ -207,6 +207,11 @@ static int vmm_ipinum; SYSCTL_INT(_hw_vmm, OID_AUTO, ipinum, CTLFLAG_RD, &vmm_ipinum, 0, "IPI vector used for vcpu notifications"); +static int trace_guest_exceptions; +SYSCTL_INT(_hw_vmm, OID_AUTO, trace_guest_exceptions, CTLFLAG_RDTUN, + &trace_guest_exceptions, 0, + "Trap into hypervisor on all guest exceptions and reflect them back"); + static void vcpu_cleanup(struct vm *vm, int i, bool destroy) { @@ -250,6 +255,13 @@ vcpu_init(struct vm *vm, int vcpu_id, bool create) vmm_stat_init(vcpu->stats); } +int +vcpu_trace_exceptions(struct vm *vm, int vcpuid) +{ + + return (trace_guest_exceptions); +} + struct vm_exit * vm_exitinfo(struct vm *vm, int cpuid) { diff --git a/sys/arm/allwinner/a20/std.a20 b/sys/arm/allwinner/a20/std.a20 index f582e91..7beb2bd 100644 --- a/sys/arm/allwinner/a20/std.a20 +++ b/sys/arm/allwinner/a20/std.a20 @@ -3,6 +3,7 @@ cpu CPU_CORTEXA machine arm armv6 +makeoptions CONF_CFLAGS="-march=armv7a -Wa,-march=armv7a" makeoption ARM_LITTLE_ENDIAN # Physical memory starts at 0x40200000. We assume images are loaded at diff --git a/sys/arm/allwinner/std.a10 b/sys/arm/allwinner/std.a10 index 11ef732..338395d 100644 --- a/sys/arm/allwinner/std.a10 +++ b/sys/arm/allwinner/std.a10 @@ -3,6 +3,7 @@ cpu CPU_CORTEXA machine arm armv6 +makeoptions CONF_CFLAGS="-march=armv7a -Wa,-march=armv7a" makeoption ARM_LITTLE_ENDIAN # Physical memory starts at 0x40200000. We assume images are loaded at diff --git a/sys/arm/altera/socfpga/std.socfpga b/sys/arm/altera/socfpga/std.socfpga index c6607a5..2bc45b9 100644 --- a/sys/arm/altera/socfpga/std.socfpga +++ b/sys/arm/altera/socfpga/std.socfpga @@ -4,6 +4,7 @@ makeoption ARM_LITTLE_ENDIAN cpu CPU_CORTEXA machine arm armv6 +makeoptions CONF_CFLAGS="-march=armv7a -Wa,-march=armv7a" options PHYSADDR=0x00000000 diff --git a/sys/arm/arm/db_trace.c b/sys/arm/arm/db_trace.c index e0d7552..04ab565 100644 --- a/sys/arm/arm/db_trace.c +++ b/sys/arm/arm/db_trace.c @@ -345,9 +345,16 @@ db_unwind_tab(struct unwind_state *state) /* * The program counter was not updated, load it from the link register. */ - if (state->registers[PC] == 0) + if (state->registers[PC] == 0) { state->registers[PC] = state->registers[LR]; + /* + * If the program counter changed, flag it in the update mask. + */ + if (state->start_pc != state->registers[PC]) + state->update_mask |= 1 << PC; + } + return 0; } @@ -602,14 +609,14 @@ db_trace_thread(struct thread *thr, int count) ctx = kdb_thr_ctx(thr); #ifdef __ARM_EABI__ - state.registers[FP] = ctx->un_32.pcb32_r11; - state.registers[SP] = ctx->un_32.pcb32_sp; - state.registers[LR] = ctx->un_32.pcb32_lr; - state.registers[PC] = ctx->un_32.pcb32_pc; + state.registers[FP] = ctx->pcb_regs.sf_r11; + state.registers[SP] = ctx->pcb_regs.sf_sp; + state.registers[LR] = ctx->pcb_regs.sf_lr; + state.registers[PC] = ctx->pcb_regs.sf_pc; db_stack_trace_cmd(&state); #else - db_stack_trace_cmd(ctx->un_32.pcb32_r11, -1, TRUE); + db_stack_trace_cmd(ctx->pcb_regs.sf_r11, -1, TRUE); #endif } else db_trace_self(); diff --git a/sys/arm/arm/dump_machdep.c b/sys/arm/arm/dump_machdep.c index 19c19c90..61d6eb5 100644 --- a/sys/arm/arm/dump_machdep.c +++ b/sys/arm/arm/dump_machdep.c @@ -160,11 +160,13 @@ static int cb_dumpdata(struct md_pa *mdp, int seqnr, void *arg) { struct dumperinfo *di = (struct dumperinfo*)arg; - vm_paddr_t pa; + vm_paddr_t a, pa; + void *va; uint32_t pgs; size_t counter, sz, chunk; - int c, error; + int i, c, error; + va = 0; error = 0; /* catch case in which chunk size is 0 */ counter = 0; pgs = mdp->md_size / PAGE_SIZE; @@ -194,16 +196,14 @@ cb_dumpdata(struct md_pa *mdp, int seqnr, void *arg) printf(" %d", pgs * PAGE_SIZE); counter &= (1<<24) - 1; } - if (pa == (pa & L1_ADDR_BITS)) { - pmap_kenter_section(0, pa & L1_ADDR_BITS, 0); - cpu_tlb_flushID_SE(0); - cpu_cpwait(); + for (i = 0; i < chunk; i++) { + a = pa + i * PAGE_SIZE; + va = pmap_kenter_temporary(trunc_page(a), i); } #ifdef SW_WATCHDOG wdog_kern_pat(WD_LASTVAL); #endif - error = dump_write(di, - (void *)(pa - (pa & L1_ADDR_BITS)),0, dumplo, sz); + error = dump_write(di, va, 0, dumplo, sz); if (error) break; dumplo += sz; diff --git a/sys/arm/arm/exception.S b/sys/arm/arm/exception.S index 5cef965..3886968 100644 --- a/sys/arm/arm/exception.S +++ b/sys/arm/arm/exception.S @@ -241,26 +241,26 @@ __FBSDID("$FreeBSD$"); #define UNWINDSVCFRAME #endif -#define DO_AST \ - ldr r0, [sp] /* Get the SPSR from stack */ ;\ - mrs r4, cpsr /* save CPSR */ ;\ - orr r1, r4, #(PSR_I|PSR_F) ;\ - msr cpsr_c, r1 /* Disable interrupts */ ;\ - and r0, r0, #(PSR_MODE) /* Returning to USR mode? */ ;\ - teq r0, #(PSR_USR32_MODE) ;\ - bne 2f /* Nope, get out now */ ;\ - bic r4, r4, #(PSR_I|PSR_F) ;\ -1: GET_CURTHREAD_PTR(r5) ;\ - ldr r1, [r5, #(TD_FLAGS)] ;\ - and r1, r1, #(TDF_ASTPENDING|TDF_NEEDRESCHED) ;\ - teq r1, #0x00000000 ;\ - beq 2f /* Nope. Just bail */ ;\ - msr cpsr_c, r4 /* Restore interrupts */ ;\ - mov r0, sp ;\ - bl _C_LABEL(ast) /* ast(frame) */ ;\ - orr r0, r4, #(PSR_I|PSR_F) ;\ - msr cpsr_c, r0 ;\ - b 1b ;\ +#define DO_AST \ + ldr r0, [sp]; /* Get the SPSR from stack */ \ + mrs r4, cpsr; /* save CPSR */ \ + orr r1, r4, #(PSR_I|PSR_F); \ + msr cpsr_c, r1; /* Disable interrupts */ \ + and r0, r0, #(PSR_MODE); /* Returning to USR mode? */ \ + teq r0, #(PSR_USR32_MODE); \ + bne 2f; /* Nope, get out now */ \ + bic r4, r4, #(PSR_I|PSR_F); \ +1: GET_CURTHREAD_PTR(r5); \ + ldr r1, [r5, #(TD_FLAGS)]; \ + and r1, r1, #(TDF_ASTPENDING|TDF_NEEDRESCHED); \ + teq r1, #0; \ + beq 2f; /* Nope. Just bail */ \ + msr cpsr_c, r4; /* Restore interrupts */ \ + mov r0, sp; \ + bl _C_LABEL(ast); /* ast(frame) */ \ + orr r0, r4, #(PSR_I|PSR_F); \ + msr cpsr_c, r0; \ + b 1b; \ 2: @@ -320,7 +320,8 @@ ASENTRY_NP(prefetch_abort_entry) PUSHFRAMEINSVC /* mode stack, build trapframe there. */ adr lr, exception_exit /* Return from handler via standard */ mov r0, sp /* exception exit routine. Pass the */ - b prefetch_abort_handler /* trapframe to the handler. */ + mov r1, #1 /* Type flag */ + b _C_LABEL(abort_handler) END(prefetch_abort_entry) /* @@ -337,9 +338,10 @@ ASENTRY_NP(data_abort_entry) #endif sub lr, lr, #8 /* Adjust the lr. Transition to scv32 */ PUSHFRAMEINSVC /* mode stack, build trapframe there. */ - adr lr, exception_exit /* Return from handler via standard */ - mov r0, sp /* exception exit routine. Pass the */ - b data_abort_handler /* trapframe to the handler. */ + adr lr, exception_exit /* Exception exit routine */ + mov r0, sp /* Trapframe to the handler */ + mov r1, #0 /* Type flag */ + b _C_LABEL(abort_handler) END(data_abort_entry) /* diff --git a/sys/arm/arm/gdb_machdep.c b/sys/arm/arm/gdb_machdep.c index 11b9c0d..8cc7a19 100644 --- a/sys/arm/arm/gdb_machdep.c +++ b/sys/arm/arm/gdb_machdep.c @@ -67,22 +67,26 @@ gdb_cpu_getreg(int regnum, size_t *regsz) } switch (regnum) { - case 8: return (&kdb_thrctx->un_32.pcb32_r8); - case 9: return (&kdb_thrctx->un_32.pcb32_r9); - case 10: return (&kdb_thrctx->un_32.pcb32_r10); - case 11: return (&kdb_thrctx->un_32.pcb32_r11); - case 12: return (&kdb_thrctx->un_32.pcb32_r12); - case 13: stacktest = kdb_thrctx->un_32.pcb32_sp + 5 * 4; + case 4: return (&kdb_thrctx->pcb_regs.sf_r4); + case 5: return (&kdb_thrctx->pcb_regs.sf_r5); + case 6: return (&kdb_thrctx->pcb_regs.sf_r6); + case 7: return (&kdb_thrctx->pcb_regs.sf_r7); + case 8: return (&kdb_thrctx->pcb_regs.sf_r8); + case 9: return (&kdb_thrctx->pcb_regs.sf_r9); + case 10: return (&kdb_thrctx->pcb_regs.sf_r10); + case 11: return (&kdb_thrctx->pcb_regs.sf_r11); + case 12: return (&kdb_thrctx->pcb_regs.sf_r12); + case 13: stacktest = kdb_thrctx->pcb_regs.sf_sp + 5 * 4; return (&stacktest); case 15: /* * On context switch, the PC is not put in the PCB, but * we can retrieve it from the stack. */ - if (kdb_thrctx->un_32.pcb32_sp > KERNBASE) { - kdb_thrctx->un_32.pcb32_pc = *(register_t *) - (kdb_thrctx->un_32.pcb32_sp + 4 * 4); - return (&kdb_thrctx->un_32.pcb32_pc); + if (kdb_thrctx->pcb_regs.sf_sp > KERNBASE) { + kdb_thrctx->pcb_regs.sf_pc = *(register_t *) + (kdb_thrctx->pcb_regs.sf_sp + 4 * 4); + return (&kdb_thrctx->pcb_regs.sf_pc); } } diff --git a/sys/arm/arm/genassym.c b/sys/arm/arm/genassym.c index ab0be4c..ce7b88d 100644 --- a/sys/arm/arm/genassym.c +++ b/sys/arm/arm/genassym.c @@ -63,13 +63,18 @@ ASSYM(PCB_FLAGS, offsetof(struct pcb, pcb_flags)); ASSYM(PCB_PAGEDIR, offsetof(struct pcb, pcb_pagedir)); ASSYM(PCB_L1VEC, offsetof(struct pcb, pcb_l1vec)); ASSYM(PCB_PL1VEC, offsetof(struct pcb, pcb_pl1vec)); -ASSYM(PCB_R8, offsetof(struct pcb, un_32.pcb32_r8)); -ASSYM(PCB_R9, offsetof(struct pcb, un_32.pcb32_r9)); -ASSYM(PCB_R10, offsetof(struct pcb, un_32.pcb32_r10)); -ASSYM(PCB_R11, offsetof(struct pcb, un_32.pcb32_r11)); -ASSYM(PCB_R12, offsetof(struct pcb, un_32.pcb32_r12)); -ASSYM(PCB_PC, offsetof(struct pcb, un_32.pcb32_pc)); -ASSYM(PCB_SP, offsetof(struct pcb, un_32.pcb32_sp)); +ASSYM(PCB_R4, offsetof(struct pcb, pcb_regs.sf_r4)); +ASSYM(PCB_R5, offsetof(struct pcb, pcb_regs.sf_r5)); +ASSYM(PCB_R6, offsetof(struct pcb, pcb_regs.sf_r6)); +ASSYM(PCB_R7, offsetof(struct pcb, pcb_regs.sf_r7)); +ASSYM(PCB_R8, offsetof(struct pcb, pcb_regs.sf_r8)); +ASSYM(PCB_R9, offsetof(struct pcb, pcb_regs.sf_r9)); +ASSYM(PCB_R10, offsetof(struct pcb, pcb_regs.sf_r10)); +ASSYM(PCB_R11, offsetof(struct pcb, pcb_regs.sf_r11)); +ASSYM(PCB_R12, offsetof(struct pcb, pcb_regs.sf_r12)); +ASSYM(PCB_SP, offsetof(struct pcb, pcb_regs.sf_sp)); +ASSYM(PCB_LR, offsetof(struct pcb, pcb_regs.sf_lr)); +ASSYM(PCB_PC, offsetof(struct pcb, pcb_regs.sf_pc)); ASSYM(PC_CURPCB, offsetof(struct pcpu, pc_curpcb)); ASSYM(PC_CURTHREAD, offsetof(struct pcpu, pc_curthread)); diff --git a/sys/arm/arm/gic.c b/sys/arm/arm/gic.c index f1bb014..af667ca 100644 --- a/sys/arm/arm/gic.c +++ b/sys/arm/arm/gic.c @@ -99,13 +99,13 @@ __FBSDID("$FreeBSD$"); #define GICD_ICFGR_TRIG_MASK 0x2 struct arm_gic_softc { + device_t gic_dev; struct resource * gic_res[3]; bus_space_tag_t gic_c_bst; bus_space_tag_t gic_d_bst; bus_space_handle_t gic_c_bsh; bus_space_handle_t gic_d_bsh; uint8_t ver; - device_t dev; struct mtx mutex; uint32_t nirqs; }; @@ -118,14 +118,14 @@ static struct resource_spec arm_gic_spec[] = { static struct arm_gic_softc *arm_gic_sc = NULL; -#define gic_c_read_4(reg) \ - bus_space_read_4(arm_gic_sc->gic_c_bst, arm_gic_sc->gic_c_bsh, reg) -#define gic_c_write_4(reg, val) \ - bus_space_write_4(arm_gic_sc->gic_c_bst, arm_gic_sc->gic_c_bsh, reg, val) -#define gic_d_read_4(reg) \ - bus_space_read_4(arm_gic_sc->gic_d_bst, arm_gic_sc->gic_d_bsh, reg) -#define gic_d_write_4(reg, val) \ - bus_space_write_4(arm_gic_sc->gic_d_bst, arm_gic_sc->gic_d_bsh, reg, val) +#define gic_c_read_4(_sc, _reg) \ + bus_space_read_4((_sc)->gic_c_bst, (_sc)->gic_c_bsh, (_reg)) +#define gic_c_write_4(_sc, _reg, _val) \ + bus_space_write_4((_sc)->gic_c_bst, (_sc)->gic_c_bsh, (_reg), (_val)) +#define gic_d_read_4(_sc, _reg) \ + bus_space_read_4((_sc)->gic_d_bst, (_sc)->gic_d_bsh, (_reg)) +#define gic_d_write_4(_sc, _reg, _val) \ + bus_space_write_4((_sc)->gic_d_bst, (_sc)->gic_d_bsh, (_reg), (_val)) static int gic_config_irq(int irq, enum intr_trigger trig, enum intr_polarity pol); @@ -158,35 +158,32 @@ arm_gic_probe(device_t dev) void gic_init_secondary(void) { - int i, nirqs; + struct arm_gic_softc *sc = arm_gic_sc; + int i; - /* Get the number of interrupts */ - nirqs = gic_d_read_4(GICD_TYPER); - nirqs = 32 * ((nirqs & 0x1f) + 1); - - for (i = 0; i < nirqs; i += 4) - gic_d_write_4(GICD_IPRIORITYR(i >> 2), 0); + for (i = 0; i < sc->nirqs; i += 4) + gic_d_write_4(sc, GICD_IPRIORITYR(i >> 2), 0); /* Set all the interrupts to be in Group 0 (secure) */ - for (i = 0; i < nirqs; i += 32) { - gic_d_write_4(GICD_IGROUPR(i >> 5), 0); + for (i = 0; i < sc->nirqs; i += 32) { + gic_d_write_4(sc, GICD_IGROUPR(i >> 5), 0); } /* Enable CPU interface */ - gic_c_write_4(GICC_CTLR, 1); + gic_c_write_4(sc, GICC_CTLR, 1); /* Set priority mask register. */ - gic_c_write_4(GICC_PMR, 0xff); + gic_c_write_4(sc, GICC_PMR, 0xff); /* Enable interrupt distribution */ - gic_d_write_4(GICD_CTLR, 0x01); + gic_d_write_4(sc, GICD_CTLR, 0x01); /* * Activate the timer interrupts: virtual, secure, and non-secure. */ - gic_d_write_4(GICD_ISENABLER(27 >> 5), (1UL << (27 & 0x1F))); - gic_d_write_4(GICD_ISENABLER(29 >> 5), (1UL << (29 & 0x1F))); - gic_d_write_4(GICD_ISENABLER(30 >> 5), (1UL << (30 & 0x1F))); + gic_d_write_4(sc, GICD_ISENABLER(27 >> 5), (1UL << (27 & 0x1F))); + gic_d_write_4(sc, GICD_ISENABLER(29 >> 5), (1UL << (29 & 0x1F))); + gic_d_write_4(sc, GICD_ISENABLER(30 >> 5), (1UL << (30 & 0x1F))); } int @@ -245,13 +242,15 @@ arm_gic_attach(device_t dev) return (ENXIO); sc = device_get_softc(dev); - sc->dev = dev; if (bus_alloc_resources(dev, arm_gic_spec, sc->gic_res)) { device_printf(dev, "could not allocate resources\n"); return (ENXIO); } + sc->gic_dev = dev; + arm_gic_sc = sc; + /* Initialize mutex */ mtx_init(&sc->mutex, "GIC lock", "", MTX_SPIN); @@ -263,100 +262,81 @@ arm_gic_attach(device_t dev) sc->gic_c_bst = rman_get_bustag(sc->gic_res[1]); sc->gic_c_bsh = rman_get_bushandle(sc->gic_res[1]); - arm_gic_sc = sc; - /* Disable interrupt forwarding to the CPU interface */ - gic_d_write_4(GICD_CTLR, 0x00); + gic_d_write_4(sc, GICD_CTLR, 0x00); /* Get the number of interrupts */ - sc->nirqs = gic_d_read_4(GICD_TYPER); + sc->nirqs = gic_d_read_4(sc, GICD_TYPER); sc->nirqs = 32 * ((sc->nirqs & 0x1f) + 1); /* Set up function pointers */ arm_post_filter = gic_post_filter; arm_config_irq = gic_config_irq; - icciidr = gic_c_read_4(GICC_IIDR); + icciidr = gic_c_read_4(sc, GICC_IIDR); device_printf(dev,"pn 0x%x, arch 0x%x, rev 0x%x, implementer 0x%x irqs %u\n", icciidr>>20, (icciidr>>16) & 0xF, (icciidr>>12) & 0xf, (icciidr & 0xfff), sc->nirqs); /* Set all global interrupts to be level triggered, active low. */ for (i = 32; i < sc->nirqs; i += 16) { - gic_d_write_4(GICD_ICFGR(i >> 4), 0x00000000); + gic_d_write_4(sc, GICD_ICFGR(i >> 4), 0x00000000); } /* Disable all interrupts. */ for (i = 32; i < sc->nirqs; i += 32) { - gic_d_write_4(GICD_ICENABLER(i >> 5), 0xFFFFFFFF); + gic_d_write_4(sc, GICD_ICENABLER(i >> 5), 0xFFFFFFFF); } for (i = 0; i < sc->nirqs; i += 4) { - gic_d_write_4(GICD_IPRIORITYR(i >> 2), 0); - gic_d_write_4(GICD_ITARGETSR(i >> 2), 1 << 0 | 1 << 8 | 1 << 16 | 1 << 24); + gic_d_write_4(sc, GICD_IPRIORITYR(i >> 2), 0); + gic_d_write_4(sc, GICD_ITARGETSR(i >> 2), + 1 << 0 | 1 << 8 | 1 << 16 | 1 << 24); } /* Set all the interrupts to be in Group 0 (secure) */ for (i = 0; i < sc->nirqs; i += 32) { - gic_d_write_4(GICD_IGROUPR(i >> 5), 0); + gic_d_write_4(sc, GICD_IGROUPR(i >> 5), 0); } /* Enable CPU interface */ - gic_c_write_4(GICC_CTLR, 1); + gic_c_write_4(sc, GICC_CTLR, 1); /* Set priority mask register. */ - gic_c_write_4(GICC_PMR, 0xff); + gic_c_write_4(sc, GICC_PMR, 0xff); /* Enable interrupt distribution */ - gic_d_write_4(GICD_CTLR, 0x01); + gic_d_write_4(sc, GICD_CTLR, 0x01); return (0); } -static device_method_t arm_gic_methods[] = { - DEVMETHOD(device_probe, arm_gic_probe), - DEVMETHOD(device_attach, arm_gic_attach), - { 0, 0 } -}; - -static driver_t arm_gic_driver = { - "gic", - arm_gic_methods, - sizeof(struct arm_gic_softc), -}; - -static devclass_t arm_gic_devclass; - -EARLY_DRIVER_MODULE(gic, simplebus, arm_gic_driver, arm_gic_devclass, 0, 0, - BUS_PASS_INTERRUPT + BUS_PASS_ORDER_MIDDLE); -EARLY_DRIVER_MODULE(gic, ofwbus, arm_gic_driver, arm_gic_devclass, 0, 0, - BUS_PASS_INTERRUPT + BUS_PASS_ORDER_MIDDLE); - static void gic_post_filter(void *arg) { + struct arm_gic_softc *sc = arm_gic_sc; uintptr_t irq = (uintptr_t) arg; if (irq > GIC_LAST_IPI) arm_irq_memory_barrier(irq); - gic_c_write_4(GICC_EOIR, irq); + gic_c_write_4(sc, GICC_EOIR, irq); } int arm_get_next_irq(int last_irq) { + struct arm_gic_softc *sc = arm_gic_sc; uint32_t active_irq; - active_irq = gic_c_read_4(GICC_IAR); + active_irq = gic_c_read_4(sc, GICC_IAR); /* * Immediatly EOIR the SGIs, because doing so requires the other * bits (ie CPU number), not just the IRQ number, and we do not * have this information later. */ - if ((active_irq & 0x3ff) <= GIC_LAST_IPI) - gic_c_write_4(GICC_EOIR, active_irq); + gic_c_write_4(sc, GICC_EOIR, active_irq); active_irq &= 0x3FF; if (active_irq == 0x3FF) { @@ -371,29 +351,33 @@ arm_get_next_irq(int last_irq) void arm_mask_irq(uintptr_t nb) { + struct arm_gic_softc *sc = arm_gic_sc; - gic_d_write_4(GICD_ICENABLER(nb >> 5), (1UL << (nb & 0x1F))); - gic_c_write_4(GICC_EOIR, nb); + gic_d_write_4(sc, GICD_ICENABLER(nb >> 5), (1UL << (nb & 0x1F))); + gic_c_write_4(sc, GICC_EOIR, nb); } void arm_unmask_irq(uintptr_t nb) { + struct arm_gic_softc *sc = arm_gic_sc; if (nb > GIC_LAST_IPI) arm_irq_memory_barrier(nb); - gic_d_write_4(GICD_ISENABLER(nb >> 5), (1UL << (nb & 0x1F))); + gic_d_write_4(sc, GICD_ISENABLER(nb >> 5), (1UL << (nb & 0x1F))); } static int gic_config_irq(int irq, enum intr_trigger trig, enum intr_polarity pol) { + struct arm_gic_softc *sc = arm_gic_sc; + device_t dev = sc->gic_dev; uint32_t reg; uint32_t mask; /* Function is public-accessible, so validate input arguments */ - if ((irq < 0) || (irq >= arm_gic_sc->nirqs)) + if ((irq < 0) || (irq >= sc->nirqs)) goto invalid_args; if ((trig != INTR_TRIGGER_EDGE) && (trig != INTR_TRIGGER_LEVEL) && (trig != INTR_TRIGGER_CONFORM)) @@ -402,9 +386,9 @@ gic_config_irq(int irq, enum intr_trigger trig, (pol != INTR_POLARITY_CONFORM)) goto invalid_args; - mtx_lock_spin(&arm_gic_sc->mutex); + mtx_lock_spin(&sc->mutex); - reg = gic_d_read_4(GICD_ICFGR(irq >> 4)); + reg = gic_d_read_4(sc, GICD_ICFGR(irq >> 4)); mask = (reg >> 2*(irq % 16)) & 0x3; if (pol == INTR_POLARITY_LOW) { @@ -426,14 +410,14 @@ gic_config_irq(int irq, enum intr_trigger trig, /* Set mask */ reg = reg & ~(0x3 << 2*(irq % 16)); reg = reg | (mask << 2*(irq % 16)); - gic_d_write_4(GICD_ICFGR(irq >> 4), reg); + gic_d_write_4(sc, GICD_ICFGR(irq >> 4), reg); - mtx_unlock_spin(&arm_gic_sc->mutex); + mtx_unlock_spin(&sc->mutex); return (0); invalid_args: - device_printf(arm_gic_sc->dev, "gic_config_irg, invalid parameters\n"); + device_printf(dev, "gic_config_irg, invalid parameters\n"); return (EINVAL); } @@ -441,17 +425,18 @@ invalid_args: void pic_ipi_send(cpuset_t cpus, u_int ipi) { + struct arm_gic_softc *sc = arm_gic_sc; uint32_t val = 0, i; for (i = 0; i < MAXCPU; i++) if (CPU_ISSET(i, &cpus)) val |= 1 << (16 + i); - gic_d_write_4(GICD_SGIR(0), val | ipi); + gic_d_write_4(sc, GICD_SGIR(0), val | ipi); } int -pic_ipi_get(int i) +pic_ipi_read(int i) { if (i != -1) { @@ -463,6 +448,7 @@ pic_ipi_get(int i) return (0); return (i); } + return (0x3ff); } @@ -472,3 +458,22 @@ pic_ipi_clear(int ipi) } #endif +static device_method_t arm_gic_methods[] = { + /* Device interface */ + DEVMETHOD(device_probe, arm_gic_probe), + DEVMETHOD(device_attach, arm_gic_attach), + { 0, 0 } +}; + +static driver_t arm_gic_driver = { + "gic", + arm_gic_methods, + sizeof(struct arm_gic_softc), +}; + +static devclass_t arm_gic_devclass; + +EARLY_DRIVER_MODULE(gic, simplebus, arm_gic_driver, arm_gic_devclass, 0, 0, + BUS_PASS_INTERRUPT + BUS_PASS_ORDER_MIDDLE); +EARLY_DRIVER_MODULE(gic, ofwbus, arm_gic_driver, arm_gic_devclass, 0, 0, + BUS_PASS_INTERRUPT + BUS_PASS_ORDER_MIDDLE); diff --git a/sys/arm/arm/identcpu.c b/sys/arm/arm/identcpu.c index 955b9f5..1b808ae 100644 --- a/sys/arm/arm/identcpu.c +++ b/sys/arm/arm/identcpu.c @@ -457,7 +457,7 @@ identify_arm_cpu(void) if (arm_cache_level) { printf("LoUU:%d LoC:%d LoUIS:%d \n", CPU_CLIDR_LOUU(arm_cache_level) + 1, - arm_cache_loc, CPU_CLIDR_LOUIS(arm_cache_level) + 1); + arm_cache_loc + 1, CPU_CLIDR_LOUIS(arm_cache_level) + 1); i = 0; while (((type = CPU_CLIDR_CTYPE(arm_cache_level, i)) != 0) && i < 7) { printf("Cache level %d: \n", i + 1); diff --git a/sys/arm/arm/intr.c b/sys/arm/arm/intr.c index 83eca07..030407a 100644 --- a/sys/arm/arm/intr.c +++ b/sys/arm/arm/intr.c @@ -36,8 +36,11 @@ * Soft interrupt and other generic interrupt functions. */ +#include "opt_platform.h" + #include <sys/cdefs.h> __FBSDID("$FreeBSD$"); + #include <sys/param.h> #include <sys/systm.h> #include <sys/syslog.h> @@ -47,10 +50,16 @@ __FBSDID("$FreeBSD$"); #include <sys/bus.h> #include <sys/interrupt.h> #include <sys/conf.h> + #include <machine/atomic.h> #include <machine/intr.h> #include <machine/cpu.h> +#ifdef FDT +#include <dev/fdt/fdt_common.h> +#include <machine/fdt.h> +#endif + #define INTRNAME_LEN (MAXCOMLEN + 1) typedef void (*mask_fn)(void *); @@ -89,6 +98,36 @@ intr_init(void *unused) SYSINIT(intr_init, SI_SUB_INTR, SI_ORDER_FIRST, intr_init, NULL); +#ifdef FDT +int +arm_fdt_map_irq(phandle_t iparent, pcell_t *intr, int icells) +{ + fdt_pic_decode_t intr_decode; + phandle_t intr_parent; + int i, rv, interrupt, trig, pol; + + intr_parent = OF_node_from_xref(iparent); + for (i = 0; i < icells; i++) + intr[i] = cpu_to_fdt32(intr[i]); + + for (i = 0; fdt_pic_table[i] != NULL; i++) { + intr_decode = fdt_pic_table[i]; + rv = intr_decode(intr_parent, intr, &interrupt, &trig, &pol); + + if (rv == 0) { + /* This was recognized as our PIC and decoded. */ + interrupt = FDT_MAP_IRQ(intr_parent, interrupt); + return (interrupt); + } + } + + /* Not in table, so guess */ + interrupt = FDT_MAP_IRQ(intr_parent, fdt32_to_cpu(intr[0])); + + return (interrupt); +} +#endif + void arm_setup_irqhandler(const char *name, driver_filter_t *filt, void (*hand)(void*), void *arg, int irq, int flags, void **cookiep) diff --git a/sys/arm/arm/machdep.c b/sys/arm/arm/machdep.c index 341ea4a..f3e75d5 100644 --- a/sys/arm/arm/machdep.c +++ b/sys/arm/arm/machdep.c @@ -378,7 +378,7 @@ cpu_startup(void *dummy) bufinit(); vm_pager_bufferinit(); - pcb->un_32.pcb32_sp = (u_int)thread0.td_kstack + + pcb->pcb_regs.sf_sp = (u_int)thread0.td_kstack + USPACE_SVC_STACK_TOP; vector_page_setprot(VM_PROT_READ); pmap_set_pcb_pagedir(pmap_kernel(), pcb); @@ -770,14 +770,18 @@ sys_sigreturn(td, uap) void makectx(struct trapframe *tf, struct pcb *pcb) { - pcb->un_32.pcb32_r8 = tf->tf_r8; - pcb->un_32.pcb32_r9 = tf->tf_r9; - pcb->un_32.pcb32_r10 = tf->tf_r10; - pcb->un_32.pcb32_r11 = tf->tf_r11; - pcb->un_32.pcb32_r12 = tf->tf_r12; - pcb->un_32.pcb32_pc = tf->tf_pc; - pcb->un_32.pcb32_lr = tf->tf_usr_lr; - pcb->un_32.pcb32_sp = tf->tf_usr_sp; + pcb->pcb_regs.sf_r4 = tf->tf_r4; + pcb->pcb_regs.sf_r5 = tf->tf_r5; + pcb->pcb_regs.sf_r6 = tf->tf_r6; + pcb->pcb_regs.sf_r7 = tf->tf_r7; + pcb->pcb_regs.sf_r8 = tf->tf_r8; + pcb->pcb_regs.sf_r9 = tf->tf_r9; + pcb->pcb_regs.sf_r10 = tf->tf_r10; + pcb->pcb_regs.sf_r11 = tf->tf_r11; + pcb->pcb_regs.sf_r12 = tf->tf_r12; + pcb->pcb_regs.sf_pc = tf->tf_pc; + pcb->pcb_regs.sf_lr = tf->tf_usr_lr; + pcb->pcb_regs.sf_sp = tf->tf_usr_sp; } /* diff --git a/sys/arm/arm/mp_machdep.c b/sys/arm/arm/mp_machdep.c index a1b29c1..32dce4d 100644 --- a/sys/arm/arm/mp_machdep.c +++ b/sys/arm/arm/mp_machdep.c @@ -266,7 +266,7 @@ ipi_handler(void *arg) cpu = PCPU_GET(cpuid); - ipi = pic_ipi_get((int)arg); + ipi = pic_ipi_read((int)arg); while ((ipi != 0x3ff)) { switch (ipi) { @@ -329,7 +329,7 @@ ipi_handler(void *arg) } pic_ipi_clear(ipi); - ipi = pic_ipi_get(-1); + ipi = pic_ipi_read(-1); } return (FILTER_HANDLED); diff --git a/sys/arm/arm/nexus.c b/sys/arm/arm/nexus.c index 6678d5d..06f2d6f 100644 --- a/sys/arm/arm/nexus.c +++ b/sys/arm/arm/nexus.c @@ -39,6 +39,8 @@ * and I/O memory address space. */ +#include "opt_platform.h" + #include <sys/cdefs.h> __FBSDID("$FreeBSD$"); @@ -60,10 +62,7 @@ __FBSDID("$FreeBSD$"); #include <machine/resource.h> #include <machine/intr.h> -#include "opt_platform.h" - #ifdef FDT -#include <dev/fdt/fdt_common.h> #include <machine/fdt.h> #include "ofw_bus_if.h" #endif @@ -351,28 +350,7 @@ static int nexus_ofw_map_intr(device_t dev, device_t child, phandle_t iparent, int icells, pcell_t *intr) { - fdt_pic_decode_t intr_decode; - phandle_t intr_parent; - int i, rv, interrupt, trig, pol; - - intr_parent = OF_node_from_xref(iparent); - for (i = 0; i < icells; i++) - intr[i] = cpu_to_fdt32(intr[i]); - - for (i = 0; fdt_pic_table[i] != NULL; i++) { - intr_decode = fdt_pic_table[i]; - rv = intr_decode(intr_parent, intr, &interrupt, &trig, &pol); - - if (rv == 0) { - /* This was recognized as our PIC and decoded. */ - interrupt = FDT_MAP_IRQ(intr_parent, interrupt); - return (interrupt); - } - } - - /* Not in table, so guess */ - interrupt = FDT_MAP_IRQ(intr_parent, fdt32_to_cpu(intr[0])); - return (interrupt); + return (arm_fdt_map_irq(iparent, intr, icells)); } #endif diff --git a/sys/arm/arm/stack_machdep.c b/sys/arm/arm/stack_machdep.c index f1708f5..9c984a9 100644 --- a/sys/arm/arm/stack_machdep.c +++ b/sys/arm/arm/stack_machdep.c @@ -76,7 +76,7 @@ stack_save_td(struct stack *st, struct thread *td) * as it doesn't have a frame pointer, however it's value is not used * when building for EABI. */ - frame = (u_int32_t *)td->td_pcb->un_32.pcb32_r11; + frame = (u_int32_t *)td->td_pcb->pcb_regs.sf_r11; stack_zero(st); stack_capture(st, frame); } diff --git a/sys/arm/arm/swtch.S b/sys/arm/arm/swtch.S index 610d575..e9d6f61 100644 --- a/sys/arm/arm/swtch.S +++ b/sys/arm/arm/swtch.S @@ -116,6 +116,14 @@ __FBSDID("$FreeBSD$"); .Lblocked_lock: .word _C_LABEL(blocked_lock) +/* + * cpu_throw(oldtd, newtd) + * + * Remove current thread state, then select the next thread to run + * and load its state. + * r0 = oldtd + * r1 = newtd + */ ENTRY(cpu_throw) mov r5, r1 @@ -144,7 +152,6 @@ ENTRY(cpu_throw) * r0 = Pointer to L1 slot for vector_page (or NULL) * r1 = lwp0's DACR * r5 = lwp0 - * r6 = exit func * r7 = lwp0's PCB * r9 = cpufuncs */ @@ -181,25 +188,11 @@ ENTRY(cpu_throw) mov lr, pc ldr pc, [r9, #CF_CONTEXT_SWITCH] - /* Restore all the save registers */ -#ifndef _ARM_ARCH_5E - add r1, r7, #PCB_R8 - ldmia r1, {r8-r13} -#else - ldr r8, [r7, #(PCB_R8)] - ldr r9, [r7, #(PCB_R9)] - ldr r10, [r7, #(PCB_R10)] - ldr r11, [r7, #(PCB_R11)] - ldr r12, [r7, #(PCB_R12)] - ldr r13, [r7, #(PCB_SP)] -#endif - GET_PCPU(r6, r4) /* Hook in a new pcb */ str r7, [r6, #PC_CURPCB] /* We have a new curthread now so make a note it */ - add r6, r6, #PC_CURTHREAD - str r5, [r6] + str r5, [r6, #PC_CURTHREAD] #ifndef ARM_TP_ADDRESS mcr p15, 0, r5, c13, c0, 4 #endif @@ -215,22 +208,31 @@ ENTRY(cpu_throw) #else mcr p15, 0, r6, c13, c0, 3 #endif - - add sp, sp, #4; - ldmfd sp!, {r4-r7, pc} + /* Restore all the saved registers and exit */ + add r3, r7, #PCB_R4 + ldmia r3, {r4-r12, sp, pc} END(cpu_throw) +/* + * cpu_switch(oldtd, newtd, lock) + * + * Save the current thread state, then select the next thread to run + * and load its state. + * r0 = oldtd + * r1 = newtd + * r2 = lock (new lock for old thread) + */ ENTRY(cpu_switch) - stmfd sp!, {r4-r7, lr} - sub sp, sp, #4; -#ifdef __ARM_EABI__ - .save {r4-r7, lr} - .pad #4 -#endif + /* Interrupts are disabled. */ + /* Save all the registers in the old thread's pcb. */ + ldr r3, [r0, #(TD_PCB)] + + /* Restore all the saved registers and exit */ + add r3, #(PCB_R4) + stmia r3, {r4-r12, sp, lr, pc} mov r6, r2 /* Save the mutex */ -.Lswitch_resume: /* rem: r0 = old lwp */ /* rem: interrupts are disabled */ @@ -246,30 +248,12 @@ ENTRY(cpu_switch) ldr r2, [r1, #TD_PCB] str r2, [r7, #PC_CURPCB] - /* rem: r1 = new process */ - /* rem: interrupts are enabled */ - /* Stage two : Save old context */ /* Get the user structure for the old thread. */ ldr r2, [r0, #(TD_PCB)] mov r4, r0 /* Save the old thread. */ - /* Save all the registers in the old thread's pcb */ -#ifndef _ARM_ARCH_5E - add r7, r2, #(PCB_R8) - stmia r7, {r8-r13} -#else - strd r8, [r2, #(PCB_R8)] - strd r10, [r2, #(PCB_R10)] - strd r12, [r2, #(PCB_R12)] -#endif - str pc, [r2, #(PCB_PC)] - - /* - * NOTE: We can now use r8-r13 until it is time to restore - * them for the new process. - */ #ifdef ARM_TP_ADDRESS /* Store the old tp */ ldr r3, =ARM_TP_ADDRESS @@ -318,7 +302,6 @@ ENTRY(cpu_switch) /* rem: r2 = old PCB */ /* rem: r9 = new PCB */ - /* rem: interrupts are enabled */ ldr r5, [r9, #(PCB_DACR)] /* r5 = new DACR */ mov r2, #DOMAIN_CLIENT @@ -336,7 +319,6 @@ ENTRY(cpu_switch) mrc p15, 0, r10, c2, c0, 0 /* r10 = old L1 */ ldr r11, [r9, #(PCB_PAGEDIR)] /* r11 = new L1 */ - teq r10, r11 /* Same L1? */ cmpeq r0, r5 /* Same DACR? */ beq .Lcs_context_switched /* yes! */ @@ -426,54 +408,18 @@ ENTRY(cpu_switch) /* rem: r9 = new PCB */ - /* Restore all the save registers */ -#ifndef _ARM_ARCH_5E - add r7, r9, #PCB_R8 - ldmia r7, {r8-r13} - sub r7, r7, #PCB_R8 /* restore PCB pointer */ -#else - mov r7, r9 - ldr r8, [r7, #(PCB_R8)] - ldr r9, [r7, #(PCB_R9)] - ldr r10, [r7, #(PCB_R10)] - ldr r11, [r7, #(PCB_R11)] - ldr r12, [r7, #(PCB_R12)] - ldr r13, [r7, #(PCB_SP)] -#endif - - /* rem: r5 = new lwp's proc */ - /* rem: r6 = lock */ - /* rem: r7 = new PCB */ - -.Lswitch_return: - - /* - * Pull the registers that got pushed when either savectx() or - * cpu_switch() was called and return. - */ - add sp, sp, #4; - ldmfd sp!, {r4-r7, pc} -#ifdef DIAGNOSTIC -.Lswitch_bogons: - adr r0, .Lswitch_panic_str - bl _C_LABEL(panic) -1: nop - b 1b - -.Lswitch_panic_str: - .asciz "cpu_switch: sched_qs empty with non-zero sched_whichqs!\n" -#endif + /* Restore all the saved registers and exit */ + add r3, r9, #PCB_R4 + ldmia r3, {r4-r12, sp, pc} END(cpu_switch) ENTRY(savectx) - stmfd sp!, {r4-r7, lr} + stmfd sp!, {lr} sub sp, sp, #4 - /* - * r0 = pcb - */ - /* Store all the registers in the process's pcb */ - add r2, r0, #(PCB_R8) - stmia r2, {r8-r13} + + /* Store all the registers in the thread's pcb */ + add r3, r0, #(PCB_R4) + stmia r3, {r4-r12, sp, lr, pc} #ifdef VFP fmrx r2, fpexc /* If the VFP is enabled */ tst r2, #(VFPEXC_EN) /* the current thread has */ @@ -482,7 +428,7 @@ ENTRY(savectx) blne _C_LABEL(vfp_store) /* and disable the VFP. */ #endif add sp, sp, #4; - ldmfd sp!, {r4-r7, pc} + ldmfd sp!, {pc} END(savectx) ENTRY(fork_trampoline) diff --git a/sys/arm/arm/trap.c b/sys/arm/arm/trap.c index ec14feb..081cfaa 100644 --- a/sys/arm/arm/trap.c +++ b/sys/arm/arm/trap.c @@ -127,6 +127,7 @@ static int dab_align(struct trapframe *, u_int, u_int, struct thread *, struct ksig *); static int dab_buserr(struct trapframe *, u_int, u_int, struct thread *, struct ksig *); +static void prefetch_abort_handler(struct trapframe *); static const struct data_abort data_aborts[] = { {dab_fatal, "Vector Exception"}, @@ -171,7 +172,7 @@ call_trapsignal(struct thread *td, int sig, u_long code) } void -data_abort_handler(struct trapframe *tf) +abort_handler(struct trapframe *tf, int type) { struct vm_map *map; struct pcb *pcb; @@ -184,6 +185,8 @@ data_abort_handler(struct trapframe *tf) struct ksig ksig; struct proc *p; + if (type == 1) + return (prefetch_abort_handler(tf)); /* Grab FAR/FSR before enabling interrupts */ far = cpu_faultaddress(); @@ -545,7 +548,7 @@ dab_buserr(struct trapframe *tf, u_int fsr, u_int far, struct thread *td, * If the current trapframe is at the top of the kernel stack, * the fault _must_ have come from user mode. */ - if (tf != ((struct trapframe *)pcb->un_32.pcb32_sp) - 1) { + if (tf != ((struct trapframe *)pcb->pcb_regs.sf_sp) - 1) { /* * Kernel mode. We're either about to die a * spectacular death, or pcb_onfault will come @@ -605,7 +608,7 @@ dab_buserr(struct trapframe *tf, u_int fsr, u_int far, struct thread *td, * does no have read permission so send it a signal. * Otherwise fault the page in and try again. */ -void +static void prefetch_abort_handler(struct trapframe *tf) { struct thread *td; diff --git a/sys/arm/arm/vm_machdep.c b/sys/arm/arm/vm_machdep.c index 7758a22..42587f5 100644 --- a/sys/arm/arm/vm_machdep.c +++ b/sys/arm/arm/vm_machdep.c @@ -79,7 +79,7 @@ __FBSDID("$FreeBSD$"); * struct switchframe and trapframe must both be a multiple of 8 * for correct stack alignment. */ -CTASSERT(sizeof(struct switchframe) == 24); +CTASSERT(sizeof(struct switchframe) == 48); CTASSERT(sizeof(struct trapframe) == 80); /* @@ -93,43 +93,55 @@ cpu_fork(register struct thread *td1, register struct proc *p2, { struct pcb *pcb2; struct trapframe *tf; - struct switchframe *sf; struct mdproc *mdp2; if ((flags & RFPROC) == 0) return; - pcb2 = (struct pcb *)(td2->td_kstack + td2->td_kstack_pages * PAGE_SIZE) - 1; + + /* Point the pcb to the top of the stack */ + pcb2 = (struct pcb *) + (td2->td_kstack + td2->td_kstack_pages * PAGE_SIZE) - 1; #ifdef __XSCALE__ #ifndef CPU_XSCALE_CORE3 pmap_use_minicache(td2->td_kstack, td2->td_kstack_pages * PAGE_SIZE); #endif #endif td2->td_pcb = pcb2; + + /* Clone td1's pcb */ bcopy(td1->td_pcb, pcb2, sizeof(*pcb2)); + + /* Point to mdproc and then copy over td1's contents */ mdp2 = &p2->p_md; bcopy(&td1->td_proc->p_md, mdp2, sizeof(*mdp2)); - pcb2->un_32.pcb32_sp = td2->td_kstack + - USPACE_SVC_STACK_TOP - sizeof(*pcb2); + + /* Point the frame to the stack in front of pcb and copy td1's frame */ + td2->td_frame = (struct trapframe *)pcb2 - 1; + *td2->td_frame = *td1->td_frame; + + /* + * Create a new fresh stack for the new process. + * Copy the trap frame for the return to user mode as if from a + * syscall. This copies most of the user mode register values. + */ + pmap_set_pcb_pagedir(vmspace_pmap(p2->p_vmspace), pcb2); + pcb2->pcb_regs.sf_r4 = (register_t)fork_return; + pcb2->pcb_regs.sf_r5 = (register_t)td2; + pcb2->pcb_regs.sf_lr = (register_t)fork_trampoline; + pcb2->pcb_regs.sf_sp = STACKALIGN(td2->td_frame); + pcb2->pcb_vfpcpu = -1; pcb2->pcb_vfpstate.fpscr = VFPSCR_DN | VFPSCR_FZ; - pmap_activate(td2); - td2->td_frame = tf = (struct trapframe *)STACKALIGN( - pcb2->un_32.pcb32_sp - sizeof(struct trapframe)); - *tf = *td1->td_frame; - sf = (struct switchframe *)tf - 1; - sf->sf_r4 = (u_int)fork_return; - sf->sf_r5 = (u_int)td2; - sf->sf_pc = (u_int)fork_trampoline; + + tf = td2->td_frame; tf->tf_spsr &= ~PSR_C; tf->tf_r0 = 0; tf->tf_r1 = 0; - pcb2->un_32.pcb32_sp = (u_int)sf; - KASSERT((pcb2->un_32.pcb32_sp & 7) == 0, - ("cpu_fork: Incorrect stack alignment")); + /* Setup to release spin count in fork_exit(). */ td2->td_md.md_spinlock_count = 1; - td2->td_md.md_saved_cspr = 0; + td2->td_md.md_saved_cspr = PSR_SVC32_MODE;; #ifdef ARM_TP_ADDRESS td2->td_md.md_tp = *(register_t *)ARM_TP_ADDRESS; #else @@ -218,25 +230,21 @@ cpu_set_syscall_retval(struct thread *td, int error) void cpu_set_upcall(struct thread *td, struct thread *td0) { - struct trapframe *tf; - struct switchframe *sf; bcopy(td0->td_frame, td->td_frame, sizeof(struct trapframe)); bcopy(td0->td_pcb, td->td_pcb, sizeof(struct pcb)); - tf = td->td_frame; - sf = (struct switchframe *)tf - 1; - sf->sf_r4 = (u_int)fork_return; - sf->sf_r5 = (u_int)td; - sf->sf_pc = (u_int)fork_trampoline; - tf->tf_spsr &= ~PSR_C; - tf->tf_r0 = 0; - td->td_pcb->un_32.pcb32_sp = (u_int)sf; - KASSERT((td->td_pcb->un_32.pcb32_sp & 7) == 0, - ("cpu_set_upcall: Incorrect stack alignment")); + + td->td_pcb->pcb_regs.sf_r4 = (register_t)fork_return; + td->td_pcb->pcb_regs.sf_r5 = (register_t)td; + td->td_pcb->pcb_regs.sf_lr = (register_t)fork_trampoline; + td->td_pcb->pcb_regs.sf_sp = STACKALIGN(td->td_frame); + + td->td_frame->tf_spsr &= ~PSR_C; + td->td_frame->tf_r0 = 0; /* Setup to release spin count in fork_exit(). */ td->td_md.md_spinlock_count = 1; - td->td_md.md_saved_cspr = 0; + td->td_md.md_saved_cspr = PSR_SVC32_MODE; } /* @@ -250,8 +258,7 @@ cpu_set_upcall_kse(struct thread *td, void (*entry)(void *), void *arg, { struct trapframe *tf = td->td_frame; - tf->tf_usr_sp = STACKALIGN((int)stack->ss_sp + stack->ss_size - - sizeof(struct trapframe)); + tf->tf_usr_sp = STACKALIGN((int)stack->ss_sp + stack->ss_size); tf->tf_pc = (int)entry; tf->tf_r0 = (int)arg; tf->tf_spsr = PSR_USR32_MODE; @@ -289,9 +296,8 @@ cpu_thread_alloc(struct thread *td) * placed into the stack pointer which must be 8 byte aligned in * the ARM EABI. */ - td->td_frame = (struct trapframe *)STACKALIGN((u_int)td->td_kstack + - USPACE_SVC_STACK_TOP - sizeof(struct pcb) - - sizeof(struct trapframe)); + td->td_frame = (struct trapframe *)((caddr_t)td->td_pcb) - 1; + #ifdef __XSCALE__ #ifndef CPU_XSCALE_CORE3 pmap_use_minicache(td->td_kstack, td->td_kstack_pages * PAGE_SIZE); @@ -318,16 +324,8 @@ cpu_thread_clean(struct thread *td) void cpu_set_fork_handler(struct thread *td, void (*func)(void *), void *arg) { - struct switchframe *sf; - struct trapframe *tf; - - tf = td->td_frame; - sf = (struct switchframe *)tf - 1; - sf->sf_r4 = (u_int)func; - sf->sf_r5 = (u_int)arg; - td->td_pcb->un_32.pcb32_sp = (u_int)sf; - KASSERT((td->td_pcb->un_32.pcb32_sp & 7) == 0, - ("cpu_set_fork_handler: Incorrect stack alignment")); + td->td_pcb->pcb_regs.sf_r4 = (register_t)func; /* function */ + td->td_pcb->pcb_regs.sf_r5 = (register_t)arg; /* first arg */ } /* diff --git a/sys/arm/at91/at91_machdep.c b/sys/arm/at91/at91_machdep.c index 0f6b79b..293a352 100644 --- a/sys/arm/at91/at91_machdep.c +++ b/sys/arm/at91/at91_machdep.c @@ -117,8 +117,6 @@ extern struct bus_space at91_bs_tag; struct pv_addr kernel_pt_table[NUM_KERNEL_PTS]; -extern uint32_t at91_master_clock; - /* Static device mappings. */ const struct arm_devmap_entry at91_devmap[] = { /* diff --git a/sys/arm/broadcom/bcm2835/bcm2835_cpufreq.c b/sys/arm/broadcom/bcm2835/bcm2835_cpufreq.c new file mode 100644 index 0000000..686e80c --- /dev/null +++ b/sys/arm/broadcom/bcm2835/bcm2835_cpufreq.c @@ -0,0 +1,1818 @@ +/*- + * Copyright (C) 2013-2014 Daisuke Aoyama <aoyama@peach.ne.jp> + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + */ + +#include <sys/cdefs.h> +__FBSDID("$FreeBSD$"); + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/bus.h> +#include <sys/cpu.h> +#include <sys/kernel.h> +#include <sys/lock.h> +#include <sys/malloc.h> +#include <sys/module.h> +#include <sys/mutex.h> +#include <sys/sema.h> +#include <sys/sysctl.h> + +#include <machine/bus.h> +#include <machine/cpu.h> +#include <machine/intr.h> + +#include <arm/broadcom/bcm2835/bcm2835_mbox.h> +#include <arm/broadcom/bcm2835/bcm2835_mbox_prop.h> +#include <arm/broadcom/bcm2835/bcm2835_vcbus.h> + +#include "cpufreq_if.h" +#include "mbox_if.h" + +#ifdef DEBUG +#define DPRINTF(fmt, ...) do { \ + printf("%s:%u: ", __func__, __LINE__); \ + printf(fmt, ##__VA_ARGS__); \ +} while (0) +#else +#define DPRINTF(fmt, ...) +#endif + +#define HZ2MHZ(freq) ((freq) / (1000 * 1000)) +#define MHZ2HZ(freq) ((freq) * (1000 * 1000)) +#define OFFSET2MVOLT(val) (1200 + ((val) * 25)) +#define MVOLT2OFFSET(val) (((val) - 1200) / 25) +#define RAW2K(temp) (((temp) + 273150) / 1000) +#define K2RAW(temp) (((temp) * 1000) - 273150) + +#define DEFAULT_ARM_FREQUENCY 700 +#define DEFAULT_CORE_FREQUENCY 250 +#define DEFAULT_SDRAM_FREQUENCY 400 +#define DEFAULT_LOWEST_FREQ 300 +#define TRANSITION_LATENCY 1000 +#define MIN_OVER_VOLTAGE -16 +#define MAX_OVER_VOLTAGE 6 +#define MSG_ERROR -999999999 +#define MHZSTEP 100 +#define HZSTEP (MHZ2HZ(MHZSTEP)) + +#define VC_LOCK(sc) do { \ + sema_wait(&vc_sema); \ + } while (0) +#define VC_UNLOCK(sc) do { \ + sema_post(&vc_sema); \ + } while (0) + +/* ARM->VC mailbox property semaphore */ +static struct sema vc_sema; + +static struct sysctl_ctx_list bcm2835_sysctl_ctx; + +struct bcm2835_cpufreq_softc { + device_t dev; + int arm_max_freq; + int arm_min_freq; + int core_max_freq; + int core_min_freq; + int sdram_max_freq; + int sdram_min_freq; + int max_voltage_core; + int min_voltage_core; + + /* the values written in mbox */ + int voltage_core; + int voltage_sdram; + int voltage_sdram_c; + int voltage_sdram_i; + int voltage_sdram_p; + int turbo_mode; + + /* mbox buffer (physical address) */ + bus_dma_tag_t dma_tag; + bus_dmamap_t dma_map; + bus_size_t dma_size; + void *dma_buf; + bus_addr_t dma_phys; + + /* initial hook for waiting mbox intr */ + struct intr_config_hook init_hook; +}; + +static int cpufreq_verbose = 0; +TUNABLE_INT("hw.bcm2835.cpufreq.verbose", &cpufreq_verbose); +static int cpufreq_lowest_freq = DEFAULT_LOWEST_FREQ; +TUNABLE_INT("hw.bcm2835.cpufreq.lowest_freq", &cpufreq_lowest_freq); + +#ifdef PROP_DEBUG +static void +bcm2835_dump(const void *data, int len) +{ + const uint8_t *p = (const uint8_t*)data; + int i; + + printf("dump @ %p:\n", data); + for (i = 0; i < len; i++) { + printf("%2.2x ", p[i]); + if ((i % 4) == 3) + printf(" "); + if ((i % 16) == 15) + printf("\n"); + } + printf("\n"); +} +#endif + +static int +bcm2835_mbox_call_prop(struct bcm2835_cpufreq_softc *sc) +{ + struct bcm2835_mbox_hdr *msg = (struct bcm2835_mbox_hdr *)sc->dma_buf; + struct bcm2835_mbox_tag_hdr *tag, *last; + uint8_t *up; + device_t mbox; + size_t hdr_size; + int idx; + int err; + + /* + * For multiple calls, locking is not here. The caller must have + * VC semaphore. + */ + + /* get mbox device */ + mbox = devclass_get_device(devclass_find("mbox"), 0); + if (mbox == NULL) { + device_printf(sc->dev, "can't find mbox\n"); + return (-1); + } + + /* go mailbox property */ +#ifdef PROP_DEBUG + bcm2835_dump(msg, 64); +#endif + bus_dmamap_sync(sc->dma_tag, sc->dma_map, + BUS_DMASYNC_PREWRITE | BUS_DMASYNC_PREREAD); + MBOX_WRITE(mbox, BCM2835_MBOX_CHAN_PROP, (uint32_t)sc->dma_phys); + MBOX_READ(mbox, BCM2835_MBOX_CHAN_PROP, &err); + bus_dmamap_sync(sc->dma_tag, sc->dma_map, BUS_DMASYNC_POSTREAD); +#ifdef PROP_DEBUG + bcm2835_dump(msg, 64); +#endif + + /* check response code */ + if (msg->code != BCM2835_MBOX_CODE_RESP_SUCCESS) { + device_printf(sc->dev, "mbox response error\n"); + return (-1); + } + + /* tag = first tag */ + up = (uint8_t *)msg; + hdr_size = sizeof(struct bcm2835_mbox_hdr); + tag = (struct bcm2835_mbox_tag_hdr *)(up + hdr_size); + /* last = end of buffer specified by header */ + last = (struct bcm2835_mbox_tag_hdr *)(up + msg->buf_size); + + /* loop unitl end tag (=0x0) */ + hdr_size = sizeof(struct bcm2835_mbox_tag_hdr); + for (idx = 0; tag->tag != 0; idx++) { + if ((tag->val_len & BCM2835_MBOX_TAG_VAL_LEN_RESPONSE) == 0) { + device_printf(sc->dev, "tag%d response error\n", idx); + return (-1); + } + /* clear response bit */ + tag->val_len &= ~BCM2835_MBOX_TAG_VAL_LEN_RESPONSE; + + /* get next tag */ + up = (uint8_t *)tag; + tag = (struct bcm2835_mbox_tag_hdr *)(up + hdr_size + + tag->val_buf_size); + + /* check buffer size of header */ + if (tag > last) { + device_printf(sc->dev, "mbox buffer size error\n"); + return (-1); + } + } + + return (0); +} + +static int +bcm2835_cpufreq_get_clock_rate(struct bcm2835_cpufreq_softc *sc, + uint32_t clock_id) +{ + struct msg_get_clock_rate *msg; + int rate; + int err; + + /* + * Get clock rate + * Tag: 0x00030002 + * Request: + * Length: 4 + * Value: + * u32: clock id + * Response: + * Length: 8 + * Value: + * u32: clock id + * u32: rate (in Hz) + */ + + /* using DMA buffer for VC */ + msg = (struct msg_get_clock_rate *)sc->dma_buf; + if (sizeof(*msg) > sc->dma_size) { + device_printf(sc->dev, "DMA size overflow (%zu>%lu)\n", + sizeof(*msg), sc->dma_size); + return (MSG_ERROR); + } + + /* setup single tag buffer */ + memset(msg, 0, sizeof(*msg)); + msg->hdr.buf_size = sizeof(*msg); + msg->hdr.code = BCM2835_MBOX_CODE_REQ; + msg->tag_hdr.tag = BCM2835_MBOX_TAG_GET_CLOCK_RATE; + msg->tag_hdr.val_buf_size = sizeof(msg->body); + msg->tag_hdr.val_len = sizeof(msg->body.req); + msg->body.req.clock_id = clock_id; + msg->end_tag = 0; + + /* call mailbox property */ + err = bcm2835_mbox_call_prop(sc); + if (err) { + device_printf(sc->dev, "can't get clock rate (id=%u)\n", + clock_id); + return (MSG_ERROR); + } + + /* result (Hz) */ + rate = (int)msg->body.resp.rate_hz; + DPRINTF("clock = %d(Hz)\n", rate); + return (rate); +} + +static int +bcm2835_cpufreq_get_max_clock_rate(struct bcm2835_cpufreq_softc *sc, + uint32_t clock_id) +{ + struct msg_get_max_clock_rate *msg; + int rate; + int err; + + /* + * Get max clock rate + * Tag: 0x00030004 + * Request: + * Length: 4 + * Value: + * u32: clock id + * Response: + * Length: 8 + * Value: + * u32: clock id + * u32: rate (in Hz) + */ + + /* using DMA buffer for VC */ + msg = (struct msg_get_max_clock_rate *)sc->dma_buf; + if (sizeof(*msg) > sc->dma_size) { + device_printf(sc->dev, "DMA size overflow (%zu>%lu)\n", + sizeof(*msg), sc->dma_size); + return (MSG_ERROR); + } + + /* setup single tag buffer */ + memset(msg, 0, sizeof(*msg)); + msg->hdr.buf_size = sizeof(*msg); + msg->hdr.code = BCM2835_MBOX_CODE_REQ; + msg->tag_hdr.tag = BCM2835_MBOX_TAG_GET_MAX_CLOCK_RATE; + msg->tag_hdr.val_buf_size = sizeof(msg->body); + msg->tag_hdr.val_len = sizeof(msg->body.req); + msg->body.req.clock_id = clock_id; + msg->end_tag = 0; + + /* call mailbox property */ + err = bcm2835_mbox_call_prop(sc); + if (err) { + device_printf(sc->dev, "can't get max clock rate (id=%u)\n", + clock_id); + return (MSG_ERROR); + } + + /* result (Hz) */ + rate = (int)msg->body.resp.rate_hz; + DPRINTF("clock = %d(Hz)\n", rate); + return (rate); +} + +static int +bcm2835_cpufreq_get_min_clock_rate(struct bcm2835_cpufreq_softc *sc, + uint32_t clock_id) +{ + struct msg_get_min_clock_rate *msg; + int rate; + int err; + + /* + * Get min clock rate + * Tag: 0x00030007 + * Request: + * Length: 4 + * Value: + * u32: clock id + * Response: + * Length: 8 + * Value: + * u32: clock id + * u32: rate (in Hz) + */ + + /* using DMA buffer for VC */ + msg = (struct msg_get_min_clock_rate *)sc->dma_buf; + if (sizeof(*msg) > sc->dma_size) { + device_printf(sc->dev, "DMA size overflow (%zu>%lu)\n", + sizeof(*msg), sc->dma_size); + return (MSG_ERROR); + } + + /* setup single tag buffer */ + memset(msg, 0, sizeof(*msg)); + msg->hdr.buf_size = sizeof(*msg); + msg->hdr.code = BCM2835_MBOX_CODE_REQ; + msg->tag_hdr.tag = BCM2835_MBOX_TAG_GET_MIN_CLOCK_RATE; + msg->tag_hdr.val_buf_size = sizeof(msg->body); + msg->tag_hdr.val_len = sizeof(msg->body.req); + msg->body.req.clock_id = clock_id; + msg->end_tag = 0; + + /* call mailbox property */ + err = bcm2835_mbox_call_prop(sc); + if (err) { + device_printf(sc->dev, "can't get min clock rate (id=%u)\n", + clock_id); + return (MSG_ERROR); + } + + /* result (Hz) */ + rate = (int)msg->body.resp.rate_hz; + DPRINTF("clock = %d(Hz)\n", rate); + return (rate); +} + +static int +bcm2835_cpufreq_set_clock_rate(struct bcm2835_cpufreq_softc *sc, + uint32_t clock_id, uint32_t rate_hz) +{ + struct msg_set_clock_rate *msg; + int rate; + int err; + + /* + * Set clock rate + * Tag: 0x00038002 + * Request: + * Length: 8 + * Value: + * u32: clock id + * u32: rate (in Hz) + * Response: + * Length: 8 + * Value: + * u32: clock id + * u32: rate (in Hz) + */ + + /* using DMA buffer for VC */ + msg = (struct msg_set_clock_rate *)sc->dma_buf; + if (sizeof(*msg) > sc->dma_size) { + device_printf(sc->dev, "DMA size overflow (%zu>%lu)\n", + sizeof(*msg), sc->dma_size); + return (MSG_ERROR); + } + + /* setup single tag buffer */ + memset(msg, 0, sizeof(*msg)); + msg->hdr.buf_size = sizeof(*msg); + msg->hdr.code = BCM2835_MBOX_CODE_REQ; + msg->tag_hdr.tag = BCM2835_MBOX_TAG_SET_CLOCK_RATE; + msg->tag_hdr.val_buf_size = sizeof(msg->body); + msg->tag_hdr.val_len = sizeof(msg->body.req); + msg->body.req.clock_id = clock_id; + msg->body.req.rate_hz = rate_hz; + msg->end_tag = 0; + + /* call mailbox property */ + err = bcm2835_mbox_call_prop(sc); + if (err) { + device_printf(sc->dev, "can't set clock rate (id=%u)\n", + clock_id); + return (MSG_ERROR); + } + + /* workaround for core clock */ + if (clock_id == BCM2835_MBOX_CLOCK_ID_CORE) { + /* for safety (may change voltage without changing clock) */ + DELAY(TRANSITION_LATENCY); + + /* + * XXX: the core clock is unable to change at once, + * to change certainly, write it twice now. + */ + + /* setup single tag buffer */ + memset(msg, 0, sizeof(*msg)); + msg->hdr.buf_size = sizeof(*msg); + msg->hdr.code = BCM2835_MBOX_CODE_REQ; + msg->tag_hdr.tag = BCM2835_MBOX_TAG_SET_CLOCK_RATE; + msg->tag_hdr.val_buf_size = sizeof(msg->body); + msg->tag_hdr.val_len = sizeof(msg->body.req); + msg->body.req.clock_id = clock_id; + msg->body.req.rate_hz = rate_hz; + msg->end_tag = 0; + + /* call mailbox property */ + err = bcm2835_mbox_call_prop(sc); + if (err) { + device_printf(sc->dev, + "can't set clock rate (id=%u)\n", clock_id); + return (MSG_ERROR); + } + } + + /* result (Hz) */ + rate = (int)msg->body.resp.rate_hz; + DPRINTF("clock = %d(Hz)\n", rate); + return (rate); +} + +static int +bcm2835_cpufreq_get_turbo(struct bcm2835_cpufreq_softc *sc) +{ + struct msg_get_turbo *msg; + int level; + int err; + + /* + * Get turbo + * Tag: 0x00030009 + * Request: + * Length: 4 + * Value: + * u32: id + * Response: + * Length: 8 + * Value: + * u32: id + * u32: level + */ + + /* using DMA buffer for VC */ + msg = (struct msg_get_turbo *)sc->dma_buf; + if (sizeof(*msg) > sc->dma_size) { + device_printf(sc->dev, "DMA size overflow (%zu>%lu)\n", + sizeof(*msg), sc->dma_size); + return (MSG_ERROR); + } + + /* setup single tag buffer */ + memset(msg, 0, sizeof(*msg)); + msg->hdr.buf_size = sizeof(*msg); + msg->hdr.code = BCM2835_MBOX_CODE_REQ; + msg->tag_hdr.tag = BCM2835_MBOX_TAG_GET_TURBO; + msg->tag_hdr.val_buf_size = sizeof(msg->body); + msg->tag_hdr.val_len = sizeof(msg->body.req); + msg->body.req.id = 0; + msg->end_tag = 0; + + /* call mailbox property */ + err = bcm2835_mbox_call_prop(sc); + if (err) { + device_printf(sc->dev, "can't get turbo\n"); + return (MSG_ERROR); + } + + /* result 0=non-turbo, 1=turbo */ + level = (int)msg->body.resp.level; + DPRINTF("level = %d\n", level); + return (level); +} + +static int +bcm2835_cpufreq_set_turbo(struct bcm2835_cpufreq_softc *sc, uint32_t level) +{ + struct msg_set_turbo *msg; + int value; + int err; + + /* + * Set turbo + * Tag: 0x00038009 + * Request: + * Length: 8 + * Value: + * u32: id + * u32: level + * Response: + * Length: 8 + * Value: + * u32: id + * u32: level + */ + + /* using DMA buffer for VC */ + msg = (struct msg_set_turbo *)sc->dma_buf; + if (sizeof(*msg) > sc->dma_size) { + device_printf(sc->dev, "DMA size overflow (%zu>%lu)\n", + sizeof(*msg), sc->dma_size); + return (MSG_ERROR); + } + + /* replace unknown value to OFF */ + if (level != BCM2835_MBOX_TURBO_ON && level != BCM2835_MBOX_TURBO_OFF) + level = BCM2835_MBOX_TURBO_OFF; + + /* setup single tag buffer */ + memset(msg, 0, sizeof(*msg)); + msg->hdr.buf_size = sizeof(*msg); + msg->hdr.code = BCM2835_MBOX_CODE_REQ; + msg->tag_hdr.tag = BCM2835_MBOX_TAG_SET_TURBO; + msg->tag_hdr.val_buf_size = sizeof(msg->body); + msg->tag_hdr.val_len = sizeof(msg->body.req); + msg->body.req.id = 0; + msg->body.req.level = level; + msg->end_tag = 0; + + /* call mailbox property */ + err = bcm2835_mbox_call_prop(sc); + if (err) { + device_printf(sc->dev, "can't set turbo\n"); + return (MSG_ERROR); + } + + /* result 0=non-turbo, 1=turbo */ + value = (int)msg->body.resp.level; + DPRINTF("level = %d\n", value); + return (value); +} + +static int +bcm2835_cpufreq_get_voltage(struct bcm2835_cpufreq_softc *sc, + uint32_t voltage_id) +{ + struct msg_get_voltage *msg; + int value; + int err; + + /* + * Get voltage + * Tag: 0x00030003 + * Request: + * Length: 4 + * Value: + * u32: voltage id + * Response: + * Length: 8 + * Value: + * u32: voltage id + * u32: value (offset from 1.2V in units of 0.025V) + */ + + /* using DMA buffer for VC */ + msg = (struct msg_get_voltage *)sc->dma_buf; + if (sizeof(*msg) > sc->dma_size) { + device_printf(sc->dev, "DMA size overflow (%zu>%lu)\n", + sizeof(*msg), sc->dma_size); + return (MSG_ERROR); + } + + /* setup single tag buffer */ + memset(msg, 0, sizeof(*msg)); + msg->hdr.buf_size = sizeof(*msg); + msg->hdr.code = BCM2835_MBOX_CODE_REQ; + msg->tag_hdr.tag = BCM2835_MBOX_TAG_GET_VOLTAGE; + msg->tag_hdr.val_buf_size = sizeof(msg->body); + msg->tag_hdr.val_len = sizeof(msg->body.req); + msg->body.req.voltage_id = voltage_id; + msg->end_tag = 0; + + /* call mailbox property */ + err = bcm2835_mbox_call_prop(sc); + if (err) { + device_printf(sc->dev, "can't get voltage\n"); + return (MSG_ERROR); + } + + /* result (offset from 1.2V) */ + value = (int)msg->body.resp.value; + DPRINTF("value = %d\n", value); + return (value); +} + +static int +bcm2835_cpufreq_get_max_voltage(struct bcm2835_cpufreq_softc *sc, + uint32_t voltage_id) +{ + struct msg_get_max_voltage *msg; + int value; + int err; + + /* + * Get voltage + * Tag: 0x00030005 + * Request: + * Length: 4 + * Value: + * u32: voltage id + * Response: + * Length: 8 + * Value: + * u32: voltage id + * u32: value (offset from 1.2V in units of 0.025V) + */ + + /* using DMA buffer for VC */ + msg = (struct msg_get_max_voltage *)sc->dma_buf; + if (sizeof(*msg) > sc->dma_size) { + device_printf(sc->dev, "DMA size overflow (%zu>%lu)\n", + sizeof(*msg), sc->dma_size); + return (MSG_ERROR); + } + + /* setup single tag buffer */ + memset(msg, 0, sizeof(*msg)); + msg->hdr.buf_size = sizeof(*msg); + msg->hdr.code = BCM2835_MBOX_CODE_REQ; + msg->tag_hdr.tag = BCM2835_MBOX_TAG_GET_MAX_VOLTAGE; + msg->tag_hdr.val_buf_size = sizeof(msg->body); + msg->tag_hdr.val_len = sizeof(msg->body.req); + msg->body.req.voltage_id = voltage_id; + msg->end_tag = 0; + + /* call mailbox property */ + err = bcm2835_mbox_call_prop(sc); + if (err) { + device_printf(sc->dev, "can't get max voltage\n"); + return (MSG_ERROR); + } + + /* result (offset from 1.2V) */ + value = (int)msg->body.resp.value; + DPRINTF("value = %d\n", value); + return (value); +} +static int +bcm2835_cpufreq_get_min_voltage(struct bcm2835_cpufreq_softc *sc, + uint32_t voltage_id) +{ + struct msg_get_min_voltage *msg; + int value; + int err; + + /* + * Get voltage + * Tag: 0x00030008 + * Request: + * Length: 4 + * Value: + * u32: voltage id + * Response: + * Length: 8 + * Value: + * u32: voltage id + * u32: value (offset from 1.2V in units of 0.025V) + */ + + /* using DMA buffer for VC */ + msg = (struct msg_get_min_voltage *)sc->dma_buf; + if (sizeof(*msg) > sc->dma_size) { + device_printf(sc->dev, "DMA size overflow (%zu>%lu)\n", + sizeof(*msg), sc->dma_size); + return (MSG_ERROR); + } + + /* setup single tag buffer */ + memset(msg, 0, sizeof(*msg)); + msg->hdr.buf_size = sizeof(*msg); + msg->hdr.code = BCM2835_MBOX_CODE_REQ; + msg->tag_hdr.tag = BCM2835_MBOX_TAG_GET_MIN_VOLTAGE; + msg->tag_hdr.val_buf_size = sizeof(msg->body); + msg->tag_hdr.val_len = sizeof(msg->body.req); + msg->body.req.voltage_id = voltage_id; + msg->end_tag = 0; + + /* call mailbox property */ + err = bcm2835_mbox_call_prop(sc); + if (err) { + device_printf(sc->dev, "can't get min voltage\n"); + return (MSG_ERROR); + } + + /* result (offset from 1.2V) */ + value = (int)msg->body.resp.value; + DPRINTF("value = %d\n", value); + return (value); +} + +static int +bcm2835_cpufreq_set_voltage(struct bcm2835_cpufreq_softc *sc, + uint32_t voltage_id, int32_t value) +{ + struct msg_set_voltage *msg; + int err; + + /* + * Set voltage + * Tag: 0x00038003 + * Request: + * Length: 4 + * Value: + * u32: voltage id + * u32: value (offset from 1.2V in units of 0.025V) + * Response: + * Length: 8 + * Value: + * u32: voltage id + * u32: value (offset from 1.2V in units of 0.025V) + */ + + /* + * over_voltage: + * 0 (1.2 V). Values above 6 are only allowed when force_turbo or + * current_limit_override are specified (which set the warranty bit). + */ + if (value > MAX_OVER_VOLTAGE || value < MIN_OVER_VOLTAGE) { + /* currently not supported */ + device_printf(sc->dev, "not supported voltage: %d\n", value); + return (MSG_ERROR); + } + + /* using DMA buffer for VC */ + msg = (struct msg_set_voltage *)sc->dma_buf; + if (sizeof(*msg) > sc->dma_size) { + device_printf(sc->dev, "DMA size overflow (%zu>%lu)\n", + sizeof(*msg), sc->dma_size); + return (MSG_ERROR); + } + + /* setup single tag buffer */ + memset(msg, 0, sizeof(*msg)); + msg->hdr.buf_size = sizeof(*msg); + msg->hdr.code = BCM2835_MBOX_CODE_REQ; + msg->tag_hdr.tag = BCM2835_MBOX_TAG_SET_VOLTAGE; + msg->tag_hdr.val_buf_size = sizeof(msg->body); + msg->tag_hdr.val_len = sizeof(msg->body.req); + msg->body.req.voltage_id = voltage_id; + msg->body.req.value = (uint32_t)value; + msg->end_tag = 0; + + /* call mailbox property */ + err = bcm2835_mbox_call_prop(sc); + if (err) { + device_printf(sc->dev, "can't set voltage\n"); + return (MSG_ERROR); + } + + /* result (offset from 1.2V) */ + value = (int)msg->body.resp.value; + DPRINTF("value = %d\n", value); + return (value); +} + +static int +bcm2835_cpufreq_get_temperature(struct bcm2835_cpufreq_softc *sc) +{ + struct msg_get_temperature *msg; + int value; + int err; + + /* + * Get temperature + * Tag: 0x00030006 + * Request: + * Length: 4 + * Value: + * u32: temperature id + * Response: + * Length: 8 + * Value: + * u32: temperature id + * u32: value + */ + + /* using DMA buffer for VC */ + msg = (struct msg_get_temperature *)sc->dma_buf; + if (sizeof(*msg) > sc->dma_size) { + device_printf(sc->dev, "DMA size overflow (%zu>%lu)\n", + sizeof(*msg), sc->dma_size); + return (MSG_ERROR); + } + + /* setup single tag buffer */ + memset(msg, 0, sizeof(*msg)); + msg->hdr.buf_size = sizeof(*msg); + msg->hdr.code = BCM2835_MBOX_CODE_REQ; + msg->tag_hdr.tag = BCM2835_MBOX_TAG_GET_TEMPERATURE; + msg->tag_hdr.val_buf_size = sizeof(msg->body); + msg->tag_hdr.val_len = sizeof(msg->body.req); + msg->body.req.temperature_id = 0; + msg->end_tag = 0; + + /* call mailbox property */ + err = bcm2835_mbox_call_prop(sc); + if (err) { + device_printf(sc->dev, "can't get temperature\n"); + return (MSG_ERROR); + } + + /* result (temperature of degree C) */ + value = (int)msg->body.resp.value; + DPRINTF("value = %d\n", value); + return (value); +} + + + +static int +sysctl_bcm2835_cpufreq_arm_freq(SYSCTL_HANDLER_ARGS) +{ + struct bcm2835_cpufreq_softc *sc = arg1; + int val; + int err; + + /* get realtime value */ + VC_LOCK(sc); + val = bcm2835_cpufreq_get_clock_rate(sc, BCM2835_MBOX_CLOCK_ID_ARM); + VC_UNLOCK(sc); + if (val == MSG_ERROR) + return (EIO); + + err = sysctl_handle_int(oidp, &val, 0, req); + if (err || !req->newptr) /* error || read request */ + return (err); + + /* write request */ + VC_LOCK(sc); + err = bcm2835_cpufreq_set_clock_rate(sc, BCM2835_MBOX_CLOCK_ID_ARM, + val); + VC_UNLOCK(sc); + if (err == MSG_ERROR) { + device_printf(sc->dev, "set clock arm_freq error\n"); + return (EIO); + } + DELAY(TRANSITION_LATENCY); + + return (0); +} + +static int +sysctl_bcm2835_cpufreq_core_freq(SYSCTL_HANDLER_ARGS) +{ + struct bcm2835_cpufreq_softc *sc = arg1; + int val; + int err; + + /* get realtime value */ + VC_LOCK(sc); + val = bcm2835_cpufreq_get_clock_rate(sc, BCM2835_MBOX_CLOCK_ID_CORE); + VC_UNLOCK(sc); + if (val == MSG_ERROR) + return (EIO); + + err = sysctl_handle_int(oidp, &val, 0, req); + if (err || !req->newptr) /* error || read request */ + return (err); + + /* write request */ + VC_LOCK(sc); + err = bcm2835_cpufreq_set_clock_rate(sc, BCM2835_MBOX_CLOCK_ID_CORE, + val); + if (err == MSG_ERROR) { + VC_UNLOCK(sc); + device_printf(sc->dev, "set clock core_freq error\n"); + return (EIO); + } + VC_UNLOCK(sc); + DELAY(TRANSITION_LATENCY); + + return (0); +} + +static int +sysctl_bcm2835_cpufreq_sdram_freq(SYSCTL_HANDLER_ARGS) +{ + struct bcm2835_cpufreq_softc *sc = arg1; + int val; + int err; + + /* get realtime value */ + VC_LOCK(sc); + val = bcm2835_cpufreq_get_clock_rate(sc, BCM2835_MBOX_CLOCK_ID_SDRAM); + VC_UNLOCK(sc); + if (val == MSG_ERROR) + return (EIO); + + err = sysctl_handle_int(oidp, &val, 0, req); + if (err || !req->newptr) /* error || read request */ + return (err); + + /* write request */ + VC_LOCK(sc); + err = bcm2835_cpufreq_set_clock_rate(sc, BCM2835_MBOX_CLOCK_ID_SDRAM, + val); + VC_UNLOCK(sc); + if (err == MSG_ERROR) { + device_printf(sc->dev, "set clock sdram_freq error\n"); + return (EIO); + } + DELAY(TRANSITION_LATENCY); + + return (0); +} + +static int +sysctl_bcm2835_cpufreq_turbo(SYSCTL_HANDLER_ARGS) +{ + struct bcm2835_cpufreq_softc *sc = arg1; + int val; + int err; + + /* get realtime value */ + VC_LOCK(sc); + val = bcm2835_cpufreq_get_turbo(sc); + VC_UNLOCK(sc); + if (val == MSG_ERROR) + return (EIO); + + err = sysctl_handle_int(oidp, &val, 0, req); + if (err || !req->newptr) /* error || read request */ + return (err); + + /* write request */ + if (val > 0) + sc->turbo_mode = BCM2835_MBOX_TURBO_ON; + else + sc->turbo_mode = BCM2835_MBOX_TURBO_OFF; + + VC_LOCK(sc); + err = bcm2835_cpufreq_set_turbo(sc, sc->turbo_mode); + VC_UNLOCK(sc); + if (err == MSG_ERROR) { + device_printf(sc->dev, "set turbo error\n"); + return (EIO); + } + DELAY(TRANSITION_LATENCY); + + return (0); +} + +static int +sysctl_bcm2835_cpufreq_voltage_core(SYSCTL_HANDLER_ARGS) +{ + struct bcm2835_cpufreq_softc *sc = arg1; + int val; + int err; + + /* get realtime value */ + VC_LOCK(sc); + val = bcm2835_cpufreq_get_voltage(sc, BCM2835_MBOX_VOLTAGE_ID_CORE); + VC_UNLOCK(sc); + if (val == MSG_ERROR) + return (EIO); + + err = sysctl_handle_int(oidp, &val, 0, req); + if (err || !req->newptr) /* error || read request */ + return (err); + + /* write request */ + if (val > MAX_OVER_VOLTAGE || val < MIN_OVER_VOLTAGE) + return (EINVAL); + sc->voltage_core = val; + + VC_LOCK(sc); + err = bcm2835_cpufreq_set_voltage(sc, BCM2835_MBOX_VOLTAGE_ID_CORE, + sc->voltage_core); + VC_UNLOCK(sc); + if (err == MSG_ERROR) { + device_printf(sc->dev, "set voltage core error\n"); + return (EIO); + } + DELAY(TRANSITION_LATENCY); + + return (0); +} + +static int +sysctl_bcm2835_cpufreq_voltage_sdram_c(SYSCTL_HANDLER_ARGS) +{ + struct bcm2835_cpufreq_softc *sc = arg1; + int val; + int err; + + /* get realtime value */ + VC_LOCK(sc); + val = bcm2835_cpufreq_get_voltage(sc, BCM2835_MBOX_VOLTAGE_ID_SDRAM_C); + VC_UNLOCK(sc); + if (val == MSG_ERROR) + return (EIO); + + err = sysctl_handle_int(oidp, &val, 0, req); + if (err || !req->newptr) /* error || read request */ + return (err); + + /* write request */ + if (val > MAX_OVER_VOLTAGE || val < MIN_OVER_VOLTAGE) + return (EINVAL); + sc->voltage_sdram_c = val; + + VC_LOCK(sc); + err = bcm2835_cpufreq_set_voltage(sc, BCM2835_MBOX_VOLTAGE_ID_SDRAM_C, + sc->voltage_sdram_c); + VC_UNLOCK(sc); + if (err == MSG_ERROR) { + device_printf(sc->dev, "set voltage sdram_c error\n"); + return (EIO); + } + DELAY(TRANSITION_LATENCY); + + return (0); +} + +static int +sysctl_bcm2835_cpufreq_voltage_sdram_i(SYSCTL_HANDLER_ARGS) +{ + struct bcm2835_cpufreq_softc *sc = arg1; + int val; + int err; + + /* get realtime value */ + VC_LOCK(sc); + val = bcm2835_cpufreq_get_voltage(sc, BCM2835_MBOX_VOLTAGE_ID_SDRAM_I); + VC_UNLOCK(sc); + if (val == MSG_ERROR) + return (EIO); + + err = sysctl_handle_int(oidp, &val, 0, req); + if (err || !req->newptr) /* error || read request */ + return (err); + + /* write request */ + if (val > MAX_OVER_VOLTAGE || val < MIN_OVER_VOLTAGE) + return (EINVAL); + sc->voltage_sdram_i = val; + + VC_LOCK(sc); + err = bcm2835_cpufreq_set_voltage(sc, BCM2835_MBOX_VOLTAGE_ID_SDRAM_I, + sc->voltage_sdram_i); + VC_UNLOCK(sc); + if (err == MSG_ERROR) { + device_printf(sc->dev, "set voltage sdram_i error\n"); + return (EIO); + } + DELAY(TRANSITION_LATENCY); + + return (0); +} + +static int +sysctl_bcm2835_cpufreq_voltage_sdram_p(SYSCTL_HANDLER_ARGS) +{ + struct bcm2835_cpufreq_softc *sc = arg1; + int val; + int err; + + /* get realtime value */ + VC_LOCK(sc); + val = bcm2835_cpufreq_get_voltage(sc, BCM2835_MBOX_VOLTAGE_ID_SDRAM_P); + VC_UNLOCK(sc); + if (val == MSG_ERROR) + return (EIO); + + err = sysctl_handle_int(oidp, &val, 0, req); + if (err || !req->newptr) /* error || read request */ + return (err); + + /* write request */ + if (val > MAX_OVER_VOLTAGE || val < MIN_OVER_VOLTAGE) + return (EINVAL); + sc->voltage_sdram_p = val; + + VC_LOCK(sc); + err = bcm2835_cpufreq_set_voltage(sc, BCM2835_MBOX_VOLTAGE_ID_SDRAM_P, + sc->voltage_sdram_p); + VC_UNLOCK(sc); + if (err == MSG_ERROR) { + device_printf(sc->dev, "set voltage sdram_p error\n"); + return (EIO); + } + DELAY(TRANSITION_LATENCY); + + return (0); +} + +static int +sysctl_bcm2835_cpufreq_voltage_sdram(SYSCTL_HANDLER_ARGS) +{ + struct bcm2835_cpufreq_softc *sc = arg1; + int val; + int err; + + /* multiple write only */ + if (!req->newptr) + return (EINVAL); + val = 0; + err = sysctl_handle_int(oidp, &val, 0, req); + if (err) + return (err); + + /* write request */ + if (val > MAX_OVER_VOLTAGE || val < MIN_OVER_VOLTAGE) + return (EINVAL); + sc->voltage_sdram = val; + + VC_LOCK(sc); + err = bcm2835_cpufreq_set_voltage(sc, BCM2835_MBOX_VOLTAGE_ID_SDRAM_C, + val); + if (err == MSG_ERROR) { + VC_UNLOCK(sc); + device_printf(sc->dev, "set voltage sdram_c error\n"); + return (EIO); + } + err = bcm2835_cpufreq_set_voltage(sc, BCM2835_MBOX_VOLTAGE_ID_SDRAM_I, + val); + if (err == MSG_ERROR) { + VC_UNLOCK(sc); + device_printf(sc->dev, "set voltage sdram_i error\n"); + return (EIO); + } + err = bcm2835_cpufreq_set_voltage(sc, BCM2835_MBOX_VOLTAGE_ID_SDRAM_P, + val); + if (err == MSG_ERROR) { + VC_UNLOCK(sc); + device_printf(sc->dev, "set voltage sdram_p error\n"); + return (EIO); + } + VC_UNLOCK(sc); + DELAY(TRANSITION_LATENCY); + + return (0); +} + +static int +sysctl_bcm2835_cpufreq_temperature(SYSCTL_HANDLER_ARGS) +{ + struct bcm2835_cpufreq_softc *sc = arg1; + int val; + int err; + + /* get realtime value */ + VC_LOCK(sc); + val = bcm2835_cpufreq_get_temperature(sc); + VC_UNLOCK(sc); + if (val == MSG_ERROR) + return (EIO); + + err = sysctl_handle_int(oidp, &val, 0, req); + if (err || !req->newptr) /* error || read request */ + return (err); + + /* write request */ + return (EINVAL); +} + +static int +sysctl_bcm2835_devcpu_temperature(SYSCTL_HANDLER_ARGS) +{ + struct bcm2835_cpufreq_softc *sc = arg1; + int val; + int err; + + /* get realtime value */ + VC_LOCK(sc); + val = bcm2835_cpufreq_get_temperature(sc); + VC_UNLOCK(sc); + if (val == MSG_ERROR) + return (EIO); + + /* 1/1000 celsius (raw) to 1/10 kelvin */ + val = RAW2K(val) * 10; + + err = sysctl_handle_int(oidp, &val, 0, req); + if (err || !req->newptr) /* error || read request */ + return (err); + + /* write request */ + return (EINVAL); +} + + +static void +bcm2835_cpufreq_init(void *arg) +{ + struct bcm2835_cpufreq_softc *sc = arg; + struct sysctl_ctx_list *ctx; + device_t cpu; + int arm_freq, core_freq, sdram_freq; + int arm_max_freq, arm_min_freq, core_max_freq, core_min_freq; + int sdram_max_freq, sdram_min_freq; + int voltage_core, voltage_sdram_c, voltage_sdram_i, voltage_sdram_p; + int max_voltage_core, min_voltage_core; + int max_voltage_sdram_c, min_voltage_sdram_c; + int max_voltage_sdram_i, min_voltage_sdram_i; + int max_voltage_sdram_p, min_voltage_sdram_p; + int turbo, temperature; + + VC_LOCK(sc); + + /* current clock */ + arm_freq = bcm2835_cpufreq_get_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_ARM); + core_freq = bcm2835_cpufreq_get_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_CORE); + sdram_freq = bcm2835_cpufreq_get_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_SDRAM); + + /* max/min clock */ + arm_max_freq = bcm2835_cpufreq_get_max_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_ARM); + arm_min_freq = bcm2835_cpufreq_get_min_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_ARM); + core_max_freq = bcm2835_cpufreq_get_max_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_CORE); + core_min_freq = bcm2835_cpufreq_get_min_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_CORE); + sdram_max_freq = bcm2835_cpufreq_get_max_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_SDRAM); + sdram_min_freq = bcm2835_cpufreq_get_min_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_SDRAM); + + /* turbo mode */ + turbo = bcm2835_cpufreq_get_turbo(sc); + if (turbo > 0) + sc->turbo_mode = BCM2835_MBOX_TURBO_ON; + else + sc->turbo_mode = BCM2835_MBOX_TURBO_OFF; + + /* voltage */ + voltage_core = bcm2835_cpufreq_get_voltage(sc, + BCM2835_MBOX_VOLTAGE_ID_CORE); + voltage_sdram_c = bcm2835_cpufreq_get_voltage(sc, + BCM2835_MBOX_VOLTAGE_ID_SDRAM_C); + voltage_sdram_i = bcm2835_cpufreq_get_voltage(sc, + BCM2835_MBOX_VOLTAGE_ID_SDRAM_I); + voltage_sdram_p = bcm2835_cpufreq_get_voltage(sc, + BCM2835_MBOX_VOLTAGE_ID_SDRAM_P); + + /* current values (offset from 1.2V) */ + sc->voltage_core = voltage_core; + sc->voltage_sdram = voltage_sdram_c; + sc->voltage_sdram_c = voltage_sdram_c; + sc->voltage_sdram_i = voltage_sdram_i; + sc->voltage_sdram_p = voltage_sdram_p; + + /* max/min voltage */ + max_voltage_core = bcm2835_cpufreq_get_max_voltage(sc, + BCM2835_MBOX_VOLTAGE_ID_CORE); + min_voltage_core = bcm2835_cpufreq_get_min_voltage(sc, + BCM2835_MBOX_VOLTAGE_ID_CORE); + max_voltage_sdram_c = bcm2835_cpufreq_get_max_voltage(sc, + BCM2835_MBOX_VOLTAGE_ID_SDRAM_C); + max_voltage_sdram_i = bcm2835_cpufreq_get_max_voltage(sc, + BCM2835_MBOX_VOLTAGE_ID_SDRAM_I); + max_voltage_sdram_p = bcm2835_cpufreq_get_max_voltage(sc, + BCM2835_MBOX_VOLTAGE_ID_SDRAM_P); + min_voltage_sdram_c = bcm2835_cpufreq_get_min_voltage(sc, + BCM2835_MBOX_VOLTAGE_ID_SDRAM_C); + min_voltage_sdram_i = bcm2835_cpufreq_get_min_voltage(sc, + BCM2835_MBOX_VOLTAGE_ID_SDRAM_I); + min_voltage_sdram_p = bcm2835_cpufreq_get_min_voltage(sc, + BCM2835_MBOX_VOLTAGE_ID_SDRAM_P); + + /* temperature */ + temperature = bcm2835_cpufreq_get_temperature(sc); + + /* show result */ + if (cpufreq_verbose || bootverbose) { + device_printf(sc->dev, "Boot settings:\n"); + device_printf(sc->dev, + "current ARM %dMHz, Core %dMHz, SDRAM %dMHz, Turbo %s\n", + HZ2MHZ(arm_freq), HZ2MHZ(core_freq), HZ2MHZ(sdram_freq), + (sc->turbo_mode == BCM2835_MBOX_TURBO_ON) ? "ON" : "OFF"); + + device_printf(sc->dev, + "max/min ARM %d/%dMHz, Core %d/%dMHz, SDRAM %d/%dMHz\n", + HZ2MHZ(arm_max_freq), HZ2MHZ(arm_min_freq), + HZ2MHZ(core_max_freq), HZ2MHZ(core_min_freq), + HZ2MHZ(sdram_max_freq), HZ2MHZ(sdram_min_freq)); + + device_printf(sc->dev, + "current Core %dmV, SDRAM_C %dmV, SDRAM_I %dmV, " + "SDRAM_P %dmV\n", + OFFSET2MVOLT(voltage_core), OFFSET2MVOLT(voltage_sdram_c), + OFFSET2MVOLT(voltage_sdram_i), + OFFSET2MVOLT(voltage_sdram_p)); + + device_printf(sc->dev, + "max/min Core %d/%dmV, SDRAM_C %d/%dmV, SDRAM_I %d/%dmV, " + "SDRAM_P %d/%dmV\n", + OFFSET2MVOLT(max_voltage_core), + OFFSET2MVOLT(min_voltage_core), + OFFSET2MVOLT(max_voltage_sdram_c), + OFFSET2MVOLT(min_voltage_sdram_c), + OFFSET2MVOLT(max_voltage_sdram_i), + OFFSET2MVOLT(min_voltage_sdram_i), + OFFSET2MVOLT(max_voltage_sdram_p), + OFFSET2MVOLT(min_voltage_sdram_p)); + + device_printf(sc->dev, + "Temperature %d.%dC\n", (temperature / 1000), + (temperature % 1000) / 100); + } else { /* !cpufreq_verbose && !bootverbose */ + device_printf(sc->dev, + "ARM %dMHz, Core %dMHz, SDRAM %dMHz, Turbo %s\n", + HZ2MHZ(arm_freq), HZ2MHZ(core_freq), HZ2MHZ(sdram_freq), + (sc->turbo_mode == BCM2835_MBOX_TURBO_ON) ? "ON" : "OFF"); + } + + /* keep in softc (MHz/mV) */ + sc->arm_max_freq = HZ2MHZ(arm_max_freq); + sc->arm_min_freq = HZ2MHZ(arm_min_freq); + sc->core_max_freq = HZ2MHZ(core_max_freq); + sc->core_min_freq = HZ2MHZ(core_min_freq); + sc->sdram_max_freq = HZ2MHZ(sdram_max_freq); + sc->sdram_min_freq = HZ2MHZ(sdram_min_freq); + sc->max_voltage_core = OFFSET2MVOLT(max_voltage_core); + sc->min_voltage_core = OFFSET2MVOLT(min_voltage_core); + + /* if turbo is on, set to max values */ + if (sc->turbo_mode == BCM2835_MBOX_TURBO_ON) { + bcm2835_cpufreq_set_clock_rate(sc, BCM2835_MBOX_CLOCK_ID_ARM, + arm_max_freq); + DELAY(TRANSITION_LATENCY); + bcm2835_cpufreq_set_clock_rate(sc, BCM2835_MBOX_CLOCK_ID_CORE, + core_max_freq); + DELAY(TRANSITION_LATENCY); + bcm2835_cpufreq_set_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_SDRAM, sdram_max_freq); + DELAY(TRANSITION_LATENCY); + } else { + bcm2835_cpufreq_set_clock_rate(sc, BCM2835_MBOX_CLOCK_ID_ARM, + arm_min_freq); + DELAY(TRANSITION_LATENCY); + bcm2835_cpufreq_set_clock_rate(sc, BCM2835_MBOX_CLOCK_ID_CORE, + core_min_freq); + DELAY(TRANSITION_LATENCY); + bcm2835_cpufreq_set_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_SDRAM, sdram_min_freq); + DELAY(TRANSITION_LATENCY); + } + + VC_UNLOCK(sc); + + /* add human readable temperature to dev.cpu node */ + cpu = device_get_parent(sc->dev); + if (cpu != NULL) { + ctx = device_get_sysctl_ctx(cpu); + SYSCTL_ADD_PROC(ctx, + SYSCTL_CHILDREN(device_get_sysctl_tree(cpu)), OID_AUTO, + "temperature", CTLTYPE_INT | CTLFLAG_RD, sc, 0, + sysctl_bcm2835_devcpu_temperature, "IK", + "Current SoC temperature"); + } + + /* release this hook (continue boot) */ + config_intrhook_disestablish(&sc->init_hook); +} + +static void +bcm2835_cpufreq_identify(driver_t *driver, device_t parent) +{ + + DPRINTF("driver=%p, parent=%p\n", driver, parent); + if (device_find_child(parent, "bcm2835_cpufreq", -1) != NULL) + return; + if (BUS_ADD_CHILD(parent, 0, "bcm2835_cpufreq", -1) == NULL) + device_printf(parent, "add child failed\n"); +} + +static int +bcm2835_cpufreq_probe(device_t dev) +{ + + device_set_desc(dev, "CPU Frequency Control"); + return (0); +} + +static void +bcm2835_cpufreq_cb(void *arg, bus_dma_segment_t *segs, int nseg, int err) +{ + bus_addr_t *addr; + + if (err) + return; + addr = (bus_addr_t *)arg; + *addr = PHYS_TO_VCBUS(segs[0].ds_addr); +} + +static int +bcm2835_cpufreq_attach(device_t dev) +{ + struct bcm2835_cpufreq_softc *sc; + struct sysctl_oid *oid; + int err; + + /* set self dev */ + sc = device_get_softc(dev); + sc->dev = dev; + + /* initial values */ + sc->arm_max_freq = -1; + sc->arm_min_freq = -1; + sc->core_max_freq = -1; + sc->core_min_freq = -1; + sc->sdram_max_freq = -1; + sc->sdram_min_freq = -1; + sc->max_voltage_core = 0; + sc->min_voltage_core = 0; + + /* create VC mbox buffer */ + sc->dma_size = PAGE_SIZE; + err = bus_dma_tag_create( + bus_get_dma_tag(sc->dev), + PAGE_SIZE, 0, /* alignment, boundary */ + BUS_SPACE_MAXADDR_32BIT, /* lowaddr */ + BUS_SPACE_MAXADDR, /* highaddr */ + NULL, NULL, /* filter, filterarg */ + sc->dma_size, 1, /* maxsize, nsegments */ + sc->dma_size, 0, /* maxsegsize, flags */ + NULL, NULL, /* lockfunc, lockarg */ + &sc->dma_tag); + if (err) { + device_printf(dev, "can't create DMA tag\n"); + return (ENXIO); + } + + err = bus_dmamem_alloc(sc->dma_tag, (void **)&sc->dma_buf, 0, + &sc->dma_map); + if (err) { + bus_dma_tag_destroy(sc->dma_tag); + device_printf(dev, "can't allocate dmamem\n"); + return (ENXIO); + } + + err = bus_dmamap_load(sc->dma_tag, sc->dma_map, sc->dma_buf, + sc->dma_size, bcm2835_cpufreq_cb, &sc->dma_phys, 0); + if (err) { + bus_dmamem_free(sc->dma_tag, sc->dma_buf, sc->dma_map); + bus_dma_tag_destroy(sc->dma_tag); + device_printf(dev, "can't load DMA map\n"); + return (ENXIO); + } + /* OK, ready to use VC buffer */ + + /* setup sysctl at first device */ + if (device_get_unit(dev) == 0) { + sysctl_ctx_init(&bcm2835_sysctl_ctx); + /* create node for hw.cpufreq */ + oid = SYSCTL_ADD_NODE(&bcm2835_sysctl_ctx, + SYSCTL_STATIC_CHILDREN(_hw), OID_AUTO, "cpufreq", + CTLFLAG_RD, NULL, ""); + + /* Frequency (Hz) */ + SYSCTL_ADD_PROC(&bcm2835_sysctl_ctx, SYSCTL_CHILDREN(oid), + OID_AUTO, "arm_freq", CTLTYPE_INT | CTLFLAG_RW, sc, 0, + sysctl_bcm2835_cpufreq_arm_freq, "IU", + "ARM frequency (Hz)"); + SYSCTL_ADD_PROC(&bcm2835_sysctl_ctx, SYSCTL_CHILDREN(oid), + OID_AUTO, "core_freq", CTLTYPE_INT | CTLFLAG_RW, sc, 0, + sysctl_bcm2835_cpufreq_core_freq, "IU", + "Core frequency (Hz)"); + SYSCTL_ADD_PROC(&bcm2835_sysctl_ctx, SYSCTL_CHILDREN(oid), + OID_AUTO, "sdram_freq", CTLTYPE_INT | CTLFLAG_RW, sc, 0, + sysctl_bcm2835_cpufreq_sdram_freq, "IU", + "SDRAM frequency (Hz)"); + + /* Turbo state */ + SYSCTL_ADD_PROC(&bcm2835_sysctl_ctx, SYSCTL_CHILDREN(oid), + OID_AUTO, "turbo", CTLTYPE_INT | CTLFLAG_RW, sc, 0, + sysctl_bcm2835_cpufreq_turbo, "IU", + "Disables dynamic clocking"); + + /* Voltage (offset from 1.2V in units of 0.025V) */ + SYSCTL_ADD_PROC(&bcm2835_sysctl_ctx, SYSCTL_CHILDREN(oid), + OID_AUTO, "voltage_core", CTLTYPE_INT | CTLFLAG_RW, sc, 0, + sysctl_bcm2835_cpufreq_voltage_core, "I", + "ARM/GPU core voltage" + "(offset from 1.2V in units of 0.025V)"); + SYSCTL_ADD_PROC(&bcm2835_sysctl_ctx, SYSCTL_CHILDREN(oid), + OID_AUTO, "voltage_sdram", CTLTYPE_INT | CTLFLAG_WR, sc, + 0, sysctl_bcm2835_cpufreq_voltage_sdram, "I", + "SDRAM voltage (offset from 1.2V in units of 0.025V)"); + + /* Voltage individual SDRAM */ + SYSCTL_ADD_PROC(&bcm2835_sysctl_ctx, SYSCTL_CHILDREN(oid), + OID_AUTO, "voltage_sdram_c", CTLTYPE_INT | CTLFLAG_RW, sc, + 0, sysctl_bcm2835_cpufreq_voltage_sdram_c, "I", + "SDRAM controller voltage" + "(offset from 1.2V in units of 0.025V)"); + SYSCTL_ADD_PROC(&bcm2835_sysctl_ctx, SYSCTL_CHILDREN(oid), + OID_AUTO, "voltage_sdram_i", CTLTYPE_INT | CTLFLAG_RW, sc, + 0, sysctl_bcm2835_cpufreq_voltage_sdram_i, "I", + "SDRAM I/O voltage (offset from 1.2V in units of 0.025V)"); + SYSCTL_ADD_PROC(&bcm2835_sysctl_ctx, SYSCTL_CHILDREN(oid), + OID_AUTO, "voltage_sdram_p", CTLTYPE_INT | CTLFLAG_RW, sc, + 0, sysctl_bcm2835_cpufreq_voltage_sdram_p, "I", + "SDRAM phy voltage (offset from 1.2V in units of 0.025V)"); + + /* Temperature */ + SYSCTL_ADD_PROC(&bcm2835_sysctl_ctx, SYSCTL_CHILDREN(oid), + OID_AUTO, "temperature", CTLTYPE_INT | CTLFLAG_RD, sc, 0, + sysctl_bcm2835_cpufreq_temperature, "I", + "SoC temperature (thousandths of a degree C)"); + } + + /* ARM->VC lock */ + sema_init(&vc_sema, 1, "vcsema"); + + /* register callback for using mbox when interrupts are enabled */ + sc->init_hook.ich_func = bcm2835_cpufreq_init; + sc->init_hook.ich_arg = sc; + + if (config_intrhook_establish(&sc->init_hook) != 0) { + bus_dmamap_unload(sc->dma_tag, sc->dma_map); + bus_dmamem_free(sc->dma_tag, sc->dma_buf, sc->dma_map); + bus_dma_tag_destroy(sc->dma_tag); + device_printf(dev, "config_intrhook_establish failed\n"); + return (ENOMEM); + } + + /* this device is controlled by cpufreq(4) */ + cpufreq_register(dev); + + return (0); +} + +static int +bcm2835_cpufreq_detach(device_t dev) +{ + struct bcm2835_cpufreq_softc *sc; + + sc = device_get_softc(dev); + + sema_destroy(&vc_sema); + + if (sc->dma_phys != 0) + bus_dmamap_unload(sc->dma_tag, sc->dma_map); + if (sc->dma_buf != NULL) + bus_dmamem_free(sc->dma_tag, sc->dma_buf, sc->dma_map); + if (sc->dma_tag != NULL) + bus_dma_tag_destroy(sc->dma_tag); + + return (cpufreq_unregister(dev)); +} + +static int +bcm2835_cpufreq_set(device_t dev, const struct cf_setting *cf) +{ + struct bcm2835_cpufreq_softc *sc; + uint32_t rate_hz, rem; + int cur_freq, resp_freq, arm_freq, min_freq, core_freq; + + if (cf == NULL || cf->freq < 0) + return (EINVAL); + + sc = device_get_softc(dev); + + /* setting clock (Hz) */ + rate_hz = (uint32_t)MHZ2HZ(cf->freq); + rem = rate_hz % HZSTEP; + rate_hz -= rem; + if (rate_hz == 0) + return (EINVAL); + + /* adjust min freq */ + min_freq = sc->arm_min_freq; + if (sc->turbo_mode != BCM2835_MBOX_TURBO_ON) + if (min_freq > cpufreq_lowest_freq) + min_freq = cpufreq_lowest_freq; + + if (rate_hz < MHZ2HZ(min_freq) || rate_hz > MHZ2HZ(sc->arm_max_freq)) + return (EINVAL); + + /* set new value and verify it */ + VC_LOCK(sc); + cur_freq = bcm2835_cpufreq_get_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_ARM); + resp_freq = bcm2835_cpufreq_set_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_ARM, rate_hz); + DELAY(TRANSITION_LATENCY); + arm_freq = bcm2835_cpufreq_get_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_ARM); + + /* + * if non-turbo and lower than or equal min_freq, + * clock down core and sdram to default first. + */ + if (sc->turbo_mode != BCM2835_MBOX_TURBO_ON) { + core_freq = bcm2835_cpufreq_get_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_CORE); + if (rate_hz > MHZ2HZ(sc->arm_min_freq)) { + bcm2835_cpufreq_set_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_CORE, + MHZ2HZ(sc->core_max_freq)); + DELAY(TRANSITION_LATENCY); + bcm2835_cpufreq_set_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_SDRAM, + MHZ2HZ(sc->sdram_max_freq)); + DELAY(TRANSITION_LATENCY); + } else { + if (sc->core_min_freq < DEFAULT_CORE_FREQUENCY && + core_freq > DEFAULT_CORE_FREQUENCY) { + /* first, down to 250, then down to min */ + DELAY(TRANSITION_LATENCY); + bcm2835_cpufreq_set_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_CORE, + MHZ2HZ(DEFAULT_CORE_FREQUENCY)); + DELAY(TRANSITION_LATENCY); + /* reset core voltage */ + bcm2835_cpufreq_set_voltage(sc, + BCM2835_MBOX_VOLTAGE_ID_CORE, 0); + DELAY(TRANSITION_LATENCY); + } + bcm2835_cpufreq_set_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_CORE, + MHZ2HZ(sc->core_min_freq)); + DELAY(TRANSITION_LATENCY); + bcm2835_cpufreq_set_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_SDRAM, + MHZ2HZ(sc->sdram_min_freq)); + DELAY(TRANSITION_LATENCY); + } + } + + VC_UNLOCK(sc); + + if (resp_freq < 0 || arm_freq < 0 || resp_freq != arm_freq) { + device_printf(dev, "wrong freq\n"); + return (EIO); + } + DPRINTF("cpufreq: %d -> %d\n", cur_freq, arm_freq); + + return (0); +} + +static int +bcm2835_cpufreq_get(device_t dev, struct cf_setting *cf) +{ + struct bcm2835_cpufreq_softc *sc; + int arm_freq; + + if (cf == NULL) + return (EINVAL); + + sc = device_get_softc(dev); + memset(cf, CPUFREQ_VAL_UNKNOWN, sizeof(*cf)); + cf->dev = NULL; + + /* get cuurent value */ + VC_LOCK(sc); + arm_freq = bcm2835_cpufreq_get_clock_rate(sc, + BCM2835_MBOX_CLOCK_ID_ARM); + VC_UNLOCK(sc); + if (arm_freq < 0) { + device_printf(dev, "can't get clock\n"); + return (EINVAL); + } + + /* CPU clock in MHz or 100ths of a percent. */ + cf->freq = HZ2MHZ(arm_freq); + /* Voltage in mV. */ + cf->volts = CPUFREQ_VAL_UNKNOWN; + /* Power consumed in mW. */ + cf->power = CPUFREQ_VAL_UNKNOWN; + /* Transition latency in us. */ + cf->lat = TRANSITION_LATENCY; + /* Driver providing this setting. */ + cf->dev = dev; + + return (0); +} + +static int +bcm2835_cpufreq_make_freq_list(device_t dev, struct cf_setting *sets, + int *count) +{ + struct bcm2835_cpufreq_softc *sc; + int freq, min_freq, volts, rem; + int idx; + + sc = device_get_softc(dev); + freq = sc->arm_max_freq; + min_freq = sc->arm_min_freq; + + /* adjust head freq to STEP */ + rem = freq % MHZSTEP; + freq -= rem; + if (freq < min_freq) + freq = min_freq; + + /* if non-turbo, add extra low freq */ + if (sc->turbo_mode != BCM2835_MBOX_TURBO_ON) + if (min_freq > cpufreq_lowest_freq) + min_freq = cpufreq_lowest_freq; + + /* from freq to min_freq */ + for (idx = 0; idx < *count && freq >= min_freq; idx++) { + if (freq > sc->arm_min_freq) + volts = sc->max_voltage_core; + else + volts = sc->min_voltage_core; + sets[idx].freq = freq; + sets[idx].volts = volts; + sets[idx].lat = TRANSITION_LATENCY; + sets[idx].dev = dev; + freq -= MHZSTEP; + } + *count = ++idx; + + return (0); +} + +static int +bcm2835_cpufreq_settings(device_t dev, struct cf_setting *sets, int *count) +{ + struct bcm2835_cpufreq_softc *sc; + + if (sets == NULL || count == NULL) + return (EINVAL); + + sc = device_get_softc(dev); + if (sc->arm_min_freq < 0 || sc->arm_max_freq < 0) { + printf("device is not configured\n"); + return (EINVAL); + } + + /* fill data with unknown value */ + memset(sets, CPUFREQ_VAL_UNKNOWN, sizeof(*sets) * (*count)); + /* create new array up to count */ + bcm2835_cpufreq_make_freq_list(dev, sets, count); + + return (0); +} + +static int +bcm2835_cpufreq_type(device_t dev, int *type) +{ + + if (type == NULL) + return (EINVAL); + *type = CPUFREQ_TYPE_ABSOLUTE; + + return (0); +} + +static device_method_t bcm2835_cpufreq_methods[] = { + /* Device interface */ + DEVMETHOD(device_identify, bcm2835_cpufreq_identify), + DEVMETHOD(device_probe, bcm2835_cpufreq_probe), + DEVMETHOD(device_attach, bcm2835_cpufreq_attach), + DEVMETHOD(device_detach, bcm2835_cpufreq_detach), + + /* cpufreq interface */ + DEVMETHOD(cpufreq_drv_set, bcm2835_cpufreq_set), + DEVMETHOD(cpufreq_drv_get, bcm2835_cpufreq_get), + DEVMETHOD(cpufreq_drv_settings, bcm2835_cpufreq_settings), + DEVMETHOD(cpufreq_drv_type, bcm2835_cpufreq_type), + + DEVMETHOD_END +}; + +static devclass_t bcm2835_cpufreq_devclass; +static driver_t bcm2835_cpufreq_driver = { + "bcm2835_cpufreq", + bcm2835_cpufreq_methods, + sizeof(struct bcm2835_cpufreq_softc), +}; + +DRIVER_MODULE(bcm2835_cpufreq, cpu, bcm2835_cpufreq_driver, + bcm2835_cpufreq_devclass, 0, 0); diff --git a/sys/arm/broadcom/bcm2835/bcm2835_intr.c b/sys/arm/broadcom/bcm2835/bcm2835_intr.c index 7a06eb0..4d14892 100644 --- a/sys/arm/broadcom/bcm2835/bcm2835_intr.c +++ b/sys/arm/broadcom/bcm2835/bcm2835_intr.c @@ -83,10 +83,10 @@ struct bcm_intc_softc { static struct bcm_intc_softc *bcm_intc_sc = NULL; -#define intc_read_4(reg) \ - bus_space_read_4(bcm_intc_sc->intc_bst, bcm_intc_sc->intc_bsh, reg) -#define intc_write_4(reg, val) \ - bus_space_write_4(bcm_intc_sc->intc_bst, bcm_intc_sc->intc_bsh, reg, val) +#define intc_read_4(_sc, reg) \ + bus_space_read_4((_sc)->intc_bst, (_sc)->intc_bsh, (reg)) +#define intc_write_4(_sc, reg, val) \ + bus_space_write_4((_sc)->intc_bst, (_sc)->intc_bsh, (reg), (val)) static int bcm_intc_probe(device_t dev) @@ -145,6 +145,7 @@ DRIVER_MODULE(intc, simplebus, bcm_intc_driver, bcm_intc_devclass, 0, 0); int arm_get_next_irq(int last_irq) { + struct bcm_intc_softc *sc = bcm_intc_sc; uint32_t pending; int32_t irq = last_irq + 1; @@ -154,7 +155,7 @@ arm_get_next_irq(int last_irq) /* TODO: should we mask last_irq? */ if (irq < BANK1_START) { - pending = intc_read_4(INTC_PENDING_BASIC); + pending = intc_read_4(sc, INTC_PENDING_BASIC); if ((pending & 0xFF) == 0) { irq = BANK1_START; /* skip to next bank */ } else do { @@ -164,7 +165,7 @@ arm_get_next_irq(int last_irq) } while (irq < BANK1_START); } if (irq < BANK2_START) { - pending = intc_read_4(INTC_PENDING_BANK1); + pending = intc_read_4(sc, INTC_PENDING_BANK1); if (pending == 0) { irq = BANK2_START; /* skip to next bank */ } else do { @@ -174,7 +175,7 @@ arm_get_next_irq(int last_irq) } while (irq < BANK2_START); } if (irq < BANK3_START) { - pending = intc_read_4(INTC_PENDING_BANK2); + pending = intc_read_4(sc, INTC_PENDING_BANK2); if (pending != 0) do { if (pending & (1 << IRQ_BANK2(irq))) return irq; @@ -187,14 +188,15 @@ arm_get_next_irq(int last_irq) void arm_mask_irq(uintptr_t nb) { + struct bcm_intc_softc *sc = bcm_intc_sc; dprintf("%s: %d\n", __func__, nb); if (IS_IRQ_BASIC(nb)) - intc_write_4(INTC_DISABLE_BASIC, (1 << nb)); + intc_write_4(sc, INTC_DISABLE_BASIC, (1 << nb)); else if (IS_IRQ_BANK1(nb)) - intc_write_4(INTC_DISABLE_BANK1, (1 << IRQ_BANK1(nb))); + intc_write_4(sc, INTC_DISABLE_BANK1, (1 << IRQ_BANK1(nb))); else if (IS_IRQ_BANK2(nb)) - intc_write_4(INTC_DISABLE_BANK2, (1 << IRQ_BANK2(nb))); + intc_write_4(sc, INTC_DISABLE_BANK2, (1 << IRQ_BANK2(nb))); else printf("arm_mask_irq: Invalid IRQ number: %d\n", nb); } @@ -202,14 +204,15 @@ arm_mask_irq(uintptr_t nb) void arm_unmask_irq(uintptr_t nb) { + struct bcm_intc_softc *sc = bcm_intc_sc; dprintf("%s: %d\n", __func__, nb); if (IS_IRQ_BASIC(nb)) - intc_write_4(INTC_ENABLE_BASIC, (1 << nb)); + intc_write_4(sc, INTC_ENABLE_BASIC, (1 << nb)); else if (IS_IRQ_BANK1(nb)) - intc_write_4(INTC_ENABLE_BANK1, (1 << IRQ_BANK1(nb))); + intc_write_4(sc, INTC_ENABLE_BANK1, (1 << IRQ_BANK1(nb))); else if (IS_IRQ_BANK2(nb)) - intc_write_4(INTC_ENABLE_BANK2, (1 << IRQ_BANK2(nb))); + intc_write_4(sc, INTC_ENABLE_BANK2, (1 << IRQ_BANK2(nb))); else printf("arm_mask_irq: Invalid IRQ number: %d\n", nb); } diff --git a/sys/arm/broadcom/bcm2835/bcm2835_mbox.c b/sys/arm/broadcom/bcm2835/bcm2835_mbox.c index 7628eb6..f87003f 100644 --- a/sys/arm/broadcom/bcm2835/bcm2835_mbox.c +++ b/sys/arm/broadcom/bcm2835/bcm2835_mbox.c @@ -34,6 +34,7 @@ __FBSDID("$FreeBSD$"); #include <sys/module.h> #include <sys/malloc.h> #include <sys/rman.h> +#include <sys/sema.h> #include <sys/timeet.h> #include <sys/timetc.h> #include <sys/watchdog.h> @@ -88,8 +89,8 @@ struct bcm_mbox_softc { void* intr_hl; bus_space_tag_t bst; bus_space_handle_t bsh; - int valid[BCM2835_MBOX_CHANS]; int msg[BCM2835_MBOX_CHANS]; + struct sema sema[BCM2835_MBOX_CHANS]; }; #define mbox_read_4(sc, reg) \ @@ -105,23 +106,19 @@ bcm_mbox_intr(void *arg) uint32_t data; uint32_t msg; - MBOX_LOCK(sc); while (!(mbox_read_4(sc, REG_STATUS) & STATUS_EMPTY)) { msg = mbox_read_4(sc, REG_READ); dprintf("bcm_mbox_intr: raw data %08x\n", msg); chan = MBOX_CHAN(msg); data = MBOX_DATA(msg); - if (sc->valid[chan]) { + if (sc->msg[chan]) { printf("bcm_mbox_intr: channel %d oveflow\n", chan); continue; } dprintf("bcm_mbox_intr: chan %d, data %08x\n", chan, data); - sc->msg[chan] = data; - sc->valid[chan] = 1; - wakeup(&sc->msg[chan]); - + sc->msg[chan] = msg; + sema_post(&sc->sema[chan]); } - MBOX_UNLOCK(sc); } static int @@ -172,12 +169,13 @@ bcm_mbox_attach(device_t dev) mtx_init(&sc->lock, "vcio mbox", NULL, MTX_DEF); for (i = 0; i < BCM2835_MBOX_CHANS; i++) { - sc->valid[0] = 0; - sc->msg[0] = 0; + sc->msg[i] = 0; + sema_init(&sc->sema[i], 0, "mbox"); } /* Read all pending messages */ - bcm_mbox_intr(sc); + while ((mbox_read_4(sc, REG_STATUS) & STATUS_EMPTY) == 0) + (void)mbox_read_4(sc, REG_READ); mbox_write_4(sc, REG_CONFIG, CONFIG_DATA_IRQ); @@ -219,10 +217,18 @@ bcm_mbox_read(device_t dev, int chan, uint32_t *data) dprintf("bcm_mbox_read: chan %d\n", chan); MBOX_LOCK(sc); - while (!sc->valid[chan]) - msleep(&sc->msg[chan], &sc->lock, PZERO, "vcio mbox read", 0); - *data = sc->msg[chan]; - sc->valid[chan] = 0; + while (sema_trywait(&sc->sema[chan]) == 0) { + /* do not unlock sc while waiting for the mbox */ + if (sema_timedwait(&sc->sema[chan], 10*hz) == 0) + break; + printf("timeout sema for chan %d\n", chan); + } + /* + * get data from intr handler, the same channel is never coming + * because of holding sc lock. + */ + *data = MBOX_DATA(sc->msg[chan]); + sc->msg[chan] = 0; MBOX_UNLOCK(sc); dprintf("bcm_mbox_read: chan %d, data %08x\n", chan, *data); diff --git a/sys/arm/broadcom/bcm2835/bcm2835_mbox.h b/sys/arm/broadcom/bcm2835/bcm2835_mbox.h index 52f48e4..17e59eb 100644 --- a/sys/arm/broadcom/bcm2835/bcm2835_mbox.h +++ b/sys/arm/broadcom/bcm2835/bcm2835_mbox.h @@ -36,6 +36,7 @@ #define BCM2835_MBOX_CHAN_LEDS 4 #define BCM2835_MBOX_CHAN_BUTTONS 5 #define BCM2835_MBOX_CHAN_TS 6 -#define BCM2835_MBOX_CHANS 7 +#define BCM2835_MBOX_CHAN_PROP 8 +#define BCM2835_MBOX_CHANS 9 #endif /* _BCM2835_MBOX_H_ */ diff --git a/sys/arm/broadcom/bcm2835/bcm2835_mbox_prop.h b/sys/arm/broadcom/bcm2835/bcm2835_mbox_prop.h new file mode 100644 index 0000000..a2e212e --- /dev/null +++ b/sys/arm/broadcom/bcm2835/bcm2835_mbox_prop.h @@ -0,0 +1,273 @@ +/*- + * Copyright (C) 2013-2014 Daisuke Aoyama <aoyama@peach.ne.jp> + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND + * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + * + * $FreeBSD$ + */ + +#ifndef _BCM2835_MBOX_PROP_H_ +#define _BCM2835_MBOX_PROP_H_ + +#include <sys/cdefs.h> +#include <sys/types.h> + +/* + * Mailbox property interface: + * https://github.com/raspberrypi/firmware/wiki/Mailbox-property-interface + */ +#define BCM2835_MBOX_CODE_REQ 0 +#define BCM2835_MBOX_CODE_RESP_SUCCESS 0x80000000 +#define BCM2835_MBOX_CODE_RESP_ERROR 0x80000001 +#define BCM2835_MBOX_TAG_VAL_LEN_RESPONSE 0x80000000 + +struct bcm2835_mbox_hdr { + uint32_t buf_size; + uint32_t code; +}; + +struct bcm2835_mbox_tag_hdr { + uint32_t tag; + uint32_t val_buf_size; + uint32_t val_len; +}; + +#define BCM2835_MBOX_CLOCK_ID_EMMC 0x00000001 +#define BCM2835_MBOX_CLOCK_ID_UART 0x00000002 +#define BCM2835_MBOX_CLOCK_ID_ARM 0x00000003 +#define BCM2835_MBOX_CLOCK_ID_CORE 0x00000004 +#define BCM2835_MBOX_CLOCK_ID_V3D 0x00000005 +#define BCM2835_MBOX_CLOCK_ID_H264 0x00000006 +#define BCM2835_MBOX_CLOCK_ID_ISP 0x00000007 +#define BCM2835_MBOX_CLOCK_ID_SDRAM 0x00000008 +#define BCM2835_MBOX_CLOCK_ID_PIXEL 0x00000009 +#define BCM2835_MBOX_CLOCK_ID_PWM 0x0000000a + +#define BCM2835_MBOX_TAG_GET_CLOCK_RATE 0x00030002 +#define BCM2835_MBOX_TAG_SET_CLOCK_RATE 0x00038002 +#define BCM2835_MBOX_TAG_GET_MAX_CLOCK_RATE 0x00030004 +#define BCM2835_MBOX_TAG_GET_MIN_CLOCK_RATE 0x00030007 + +struct msg_get_clock_rate { + struct bcm2835_mbox_hdr hdr; + struct bcm2835_mbox_tag_hdr tag_hdr; + union { + struct { + uint32_t clock_id; + } req; + struct { + uint32_t clock_id; + uint32_t rate_hz; + } resp; + } body; + uint32_t end_tag; +}; + +struct msg_set_clock_rate { + struct bcm2835_mbox_hdr hdr; + struct bcm2835_mbox_tag_hdr tag_hdr; + union { + struct { + uint32_t clock_id; + uint32_t rate_hz; + } req; + struct { + uint32_t clock_id; + uint32_t rate_hz; + } resp; + } body; + uint32_t end_tag; +}; + +struct msg_get_max_clock_rate { + struct bcm2835_mbox_hdr hdr; + struct bcm2835_mbox_tag_hdr tag_hdr; + union { + struct { + uint32_t clock_id; + } req; + struct { + uint32_t clock_id; + uint32_t rate_hz; + } resp; + } body; + uint32_t end_tag; +}; + +struct msg_get_min_clock_rate { + struct bcm2835_mbox_hdr hdr; + struct bcm2835_mbox_tag_hdr tag_hdr; + union { + struct { + uint32_t clock_id; + } req; + struct { + uint32_t clock_id; + uint32_t rate_hz; + } resp; + } body; + uint32_t end_tag; +}; + +#define BCM2835_MBOX_TURBO_ON 1 +#define BCM2835_MBOX_TURBO_OFF 0 + +#define BCM2835_MBOX_TAG_GET_TURBO 0x00030009 +#define BCM2835_MBOX_TAG_SET_TURBO 0x00038009 + +struct msg_get_turbo { + struct bcm2835_mbox_hdr hdr; + struct bcm2835_mbox_tag_hdr tag_hdr; + union { + struct { + uint32_t id; + } req; + struct { + uint32_t id; + uint32_t level; + } resp; + } body; + uint32_t end_tag; +}; + +struct msg_set_turbo { + struct bcm2835_mbox_hdr hdr; + struct bcm2835_mbox_tag_hdr tag_hdr; + union { + struct { + uint32_t id; + uint32_t level; + } req; + struct { + uint32_t id; + uint32_t level; + } resp; + } body; + uint32_t end_tag; +}; + +#define BCM2835_MBOX_VOLTAGE_ID_CORE 0x00000001 +#define BCM2835_MBOX_VOLTAGE_ID_SDRAM_C 0x00000002 +#define BCM2835_MBOX_VOLTAGE_ID_SDRAM_P 0x00000003 +#define BCM2835_MBOX_VOLTAGE_ID_SDRAM_I 0x00000004 + +#define BCM2835_MBOX_TAG_GET_VOLTAGE 0x00030003 +#define BCM2835_MBOX_TAG_SET_VOLTAGE 0x00038003 +#define BCM2835_MBOX_TAG_GET_MAX_VOLTAGE 0x00030005 +#define BCM2835_MBOX_TAG_GET_MIN_VOLTAGE 0x00030008 + +struct msg_get_voltage { + struct bcm2835_mbox_hdr hdr; + struct bcm2835_mbox_tag_hdr tag_hdr; + union { + struct { + uint32_t voltage_id; + } req; + struct { + uint32_t voltage_id; + uint32_t value; + } resp; + } body; + uint32_t end_tag; +}; + +struct msg_set_voltage { + struct bcm2835_mbox_hdr hdr; + struct bcm2835_mbox_tag_hdr tag_hdr; + union { + struct { + uint32_t voltage_id; + uint32_t value; + } req; + struct { + uint32_t voltage_id; + uint32_t value; + } resp; + } body; + uint32_t end_tag; +}; + +struct msg_get_max_voltage { + struct bcm2835_mbox_hdr hdr; + struct bcm2835_mbox_tag_hdr tag_hdr; + union { + struct { + uint32_t voltage_id; + } req; + struct { + uint32_t voltage_id; + uint32_t value; + } resp; + } body; + uint32_t end_tag; +}; + +struct msg_get_min_voltage { + struct bcm2835_mbox_hdr hdr; + struct bcm2835_mbox_tag_hdr tag_hdr; + union { + struct { + uint32_t voltage_id; + } req; + struct { + uint32_t voltage_id; + uint32_t value; + } resp; + } body; + uint32_t end_tag; +}; + +#define BCM2835_MBOX_TAG_GET_TEMPERATURE 0x00030006 +#define BCM2835_MBOX_TAG_GET_MAX_TEMPERATURE 0x0003000a + +struct msg_get_temperature { + struct bcm2835_mbox_hdr hdr; + struct bcm2835_mbox_tag_hdr tag_hdr; + union { + struct { + uint32_t temperature_id; + } req; + struct { + uint32_t temperature_id; + uint32_t value; + } resp; + } body; + uint32_t end_tag; +}; + +struct msg_get_max_temperature { + struct bcm2835_mbox_hdr hdr; + struct bcm2835_mbox_tag_hdr tag_hdr; + union { + struct { + uint32_t temperature_id; + } req; + struct { + uint32_t temperature_id; + uint32_t value; + } resp; + } body; + uint32_t end_tag; +}; + +#endif /* _BCM2835_MBOX_PROP_H_ */ diff --git a/sys/arm/broadcom/bcm2835/files.bcm2835 b/sys/arm/broadcom/bcm2835/files.bcm2835 index 865f2d0..11158cb 100644 --- a/sys/arm/broadcom/bcm2835/files.bcm2835 +++ b/sys/arm/broadcom/bcm2835/files.bcm2835 @@ -2,6 +2,7 @@ arm/broadcom/bcm2835/bcm2835_bsc.c optional bcm2835_bsc arm/broadcom/bcm2835/bcm2835_common.c optional fdt +arm/broadcom/bcm2835/bcm2835_cpufreq.c standard arm/broadcom/bcm2835/bcm2835_dma.c standard arm/broadcom/bcm2835/bcm2835_fb.c optional sc arm/broadcom/bcm2835/bcm2835_fbd.c optional vt @@ -25,3 +26,4 @@ arm/arm/cpufunc_asm_armv6.S standard kern/kern_clocksource.c standard dev/mbox/mbox_if.m standard +dev/ofw/ofw_cpu.c standard diff --git a/sys/arm/broadcom/bcm2835/std.bcm2835 b/sys/arm/broadcom/bcm2835/std.bcm2835 index ebc1fb0..026f5b2 100644 --- a/sys/arm/broadcom/bcm2835/std.bcm2835 +++ b/sys/arm/broadcom/bcm2835/std.bcm2835 @@ -2,6 +2,7 @@ machine arm armv6 cpu CPU_ARM1176 +makeoptions CONF_CFLAGS="-mcpu=arm1176jzf-s -Wa,-mcpu=arm1176jzf-s" files "../broadcom/bcm2835/files.bcm2835" diff --git a/sys/arm/conf/APALIS-IMX6 b/sys/arm/conf/APALIS-IMX6 index 137677f..ceb21c7 100644 --- a/sys/arm/conf/APALIS-IMX6 +++ b/sys/arm/conf/APALIS-IMX6 @@ -26,6 +26,6 @@ makeoptions MODULES_OVERRIDE="" makeoptions WITHOUT_MODULES="ahc" # Flattened Device Tree -options FDT -options FDT_DTB_STATIC -makeoptions FDT_DTS_FILE=apalis-imx6.dts +options FDT +options FDT_DTB_STATIC +makeoptions FDT_DTS_FILE=apalis-imx6.dts diff --git a/sys/arm/conf/ARMADAXP b/sys/arm/conf/ARMADAXP index edfe75c..ce0f0d4 100644 --- a/sys/arm/conf/ARMADAXP +++ b/sys/arm/conf/ARMADAXP @@ -1,8 +1,22 @@ # # Custom kernel for Marvell Armada XP # -# $FreeBSD$ +# For more information on this file, please read the config(5) manual page, +# and/or the handbook section on Kernel Configuration Files: +# +# http://www.FreeBSD.org/doc/en_US.ISO8859-1/books/handbook/kernelconfig-config.html +# +# The handbook is also available locally in /usr/share/doc/handbook +# if you've installed the doc distribution, otherwise always see the +# FreeBSD World Wide Web server (http://www.FreeBSD.org/) for the +# latest information. # +# An exhaustive list of options and more detailed explanations of the +# device lines is also present in the ../../conf/NOTES and NOTES files. +# If you are in doubt as to the purpose or necessity of a line, check first +# in NOTES. +# +# $FreeBSD$ ident MV-88F78XX0 include "../mv/armadaxp/std.mv78x60" @@ -10,58 +24,75 @@ include "../mv/armadaxp/std.mv78x60" options SOC_MV_ARMADAXP makeoptions MODULES_OVERRIDE="" -makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols makeoptions WERROR="-Werror" +options HZ=1000 #options SCHED_ULE # ULE scheduler options SCHED_4BSD # 4BSD scheduler +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem -options NFSCL # Network Filesystem Client +options SOFTUPDATES # Enable FFS soft updates support +options UFS_ACL # Support for access control lists +options UFS_DIRHASH # Improve performance on big directories +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client options NFSLOCKD # Network Lock Manager -options NFS_ROOT # NFS usable as /, requires NFSCLIENT -options BOOTP -options BOOTP_NFSROOT -options BOOTP_NFSV3 -options BOOTP_WIRED_TO=mge0 - +options NFS_ROOT # NFS usable as /, requires NFSCL +options MSDOSFS # MSDOS Filesystem +options CD9660 # ISO 9660 Filesystem +options PROCFS # Process filesystem (requires PSEUDOFS) +options PSEUDOFS # Pseudo-filesystem framework options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables options GEOM_PART_BSD # BSD partition scheme options GEOM_PART_MBR # MBR partition scheme -options GEOM_PART_GPT -options ROOTDEVNAME=\"ufs:/dev/da0p1\" - +options KTRACE # ktrace(1) support options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # Posix P1003_1B real-time extensions -options MUTEX_NOINLINE -options RWLOCK_NOINLINE -options NO_FFS_SNAPSHOT -options NO_SWAPPING -options VFP - -options SMP +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions +options KBD_INSTALL_CDEV # install a CDEV entry in /dev +options VFP # Enable floating point hardware support +options SMP # Enable multiple cores -# Debugging -#options VERBOSE_SYSINIT +# Debugging for use in -current +makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols +#options VERBOSE_SYSINIT # Enable verbose sysinit messages options ALT_BREAK_TO_DEBUGGER -options DDB +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: +options DDB # Enable the kernel debugger options GDB -#options DIAGNOSTIC #options INVARIANTS # Enable calls of extra sanity checking #options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS -options KDB -options KDB_TRACE +#options WITNESS # Enable checks to detect deadlocks and cycles +#options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed +#options WITNESS_KDB +#options DIAGNOSTIC #options KTR #options KTR_VERBOSE=0 #options KTR_ENTRIES=16384 #options KTR_MASK=(KTR_SPARE2) #options KTR_COMPILE=KTR_ALL -#options WITNESS # Enable checks to detect deadlocks and cycles -#options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed -#options WITNESS_KDB + +# NFS root from boopt/dhcp +options BOOTP +options BOOTP_NFSROOT +options BOOTP_NFSV3 +options BOOTP_WIRED_TO=mge0 + +options ROOTDEVNAME=\"ufs:/dev/da0p1\" + +options MUTEX_NOINLINE +options RWLOCK_NOINLINE +options NO_FFS_SNAPSHOT +options NO_SWAPPING # Pseudo devices device random @@ -94,14 +125,13 @@ device mge # Marvell Gigabit Ethernet controller device mii device e1000phy device bpf -options HZ=1000 options DEVICE_POLLING device vlan #PCI/PCIE device pci -#FDT -options FDT +# Flattened Device Tree +options FDT # Configure using FDT/DTB data options FDT_DTB_STATIC makeoptions FDT_DTS_FILE=db78460.dts diff --git a/sys/arm/conf/BEAGLEBONE b/sys/arm/conf/BEAGLEBONE index 0363274..6f7e53e 100644 --- a/sys/arm/conf/BEAGLEBONE +++ b/sys/arm/conf/BEAGLEBONE @@ -1,10 +1,11 @@ +# # BEAGLEBONE -- Custom configuration for the BeagleBone ARM development # platforms, check out http://www.beagleboard.org/bone and # http://www.beagleboard.org/black. This kernel config file is used for the # original BeagleBone and the BeagleBone Black. # -# For more information on this file, please read the handbook section on -# Kernel Configuration Files: +# For more information on this file, please read the config(5) manual page, +# and/or the handbook section on Kernel Configuration Files: # # http://www.FreeBSD.org/doc/en_US.ISO8859-1/books/handbook/kernelconfig-config.html # @@ -28,37 +29,47 @@ makeoptions WITHOUT_MODULES="ahc" options HZ=100 options SCHED_4BSD # 4BSD scheduler +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories -options GEOM_PART_BSD # BSD partition scheme -options GEOM_PART_MBR # MBR partition scheme -options TMPFS # Efficient memory filesystem +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client +options NFSLOCKD # Network Lock Manager +options NFS_ROOT # NFS usable as /, requires NFSCL options MSDOSFS # MSDOS Filesystem options CD9660 # ISO 9660 Filesystem options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables +options GEOM_PART_BSD # BSD partition scheme +options GEOM_PART_MBR # MBR partition scheme options COMPAT_43 # Compatible with BSD 4.3 [KEEP THIS!] options SCSI_DELAY=5000 # Delay (in ms) before probing SCSI options KTRACE # ktrace(1) support options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # Posix P1003_1B real-time extensions +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions options KBD_INSTALL_CDEV # install a CDEV entry in /dev -options PREEMPTION options PLATFORM -options FREEBSD_BOOT_LOADER -options VFP # vfp/neon +options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +options VFP # Enable floating point hardware support # Debugging for use in -current makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options BREAK_TO_DEBUGGER #options VERBOSE_SYSINIT # Enable verbose sysinit messages -options KDB +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: options DDB # Enable the kernel debugger options INVARIANTS # Enable calls of extra sanity checking options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS @@ -66,19 +77,19 @@ options WITNESS # Enable checks to detect deadlocks and cycles options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed #options DIAGNOSTIC -# NFS support -options NFSCL -#options NFSD -options NFSLOCKD +# NFS server support +#options NFSD -# Uncomment this for NFS root -#options NFS_ROOT # NFS usable as /, requires NFSCL +# NFS root from boopt/dhcp +#options BOOTP #options BOOTP_NFSROOT #options BOOTP_COMPAT -#options BOOTP #options BOOTP_NFSV3 #options BOOTP_WIRED_TO=cpsw0 +# Boot device is 2nd slice on MMC/SD card +options ROOTDEVNAME=\"ufs:mmcsd0s2\" + # MMC/SD/SDIO Card slot support device mmc # mmc/sd bus device mmcsd # mmc/sd flash cards @@ -90,9 +101,6 @@ device iic device ti_i2c device am335x_pmic # AM335x Power Management IC (TPC65217) -# Boot device is 2nd slice on MMC/SD card -options ROOTDEVNAME=\"ufs:mmcsd0s2\" - # Console and misc device uart device uart_ns8250 @@ -148,6 +156,6 @@ device usb_template # Control of the gadget device usfs # Flattened Device Tree -options FDT -options FDT_DTB_STATIC -makeoptions FDT_DTS_FILE=beaglebone.dts +options FDT # Configure using FDT/DTB data +options FDT_DTB_STATIC +makeoptions FDT_DTS_FILE=beaglebone.dts diff --git a/sys/arm/conf/CHROMEBOOK-PEACH-PIT b/sys/arm/conf/CHROMEBOOK-PEACH-PIT index 7bcfed7..e01128c 100644 --- a/sys/arm/conf/CHROMEBOOK-PEACH-PIT +++ b/sys/arm/conf/CHROMEBOOK-PEACH-PIT @@ -33,15 +33,15 @@ device kbdmux device ukbd # Uncomment this for NFS root -#options NFS_ROOT # NFS usable as /, requires NFSCL -#options BOOTP_NFSROOT -#options BOOTP_COMPAT -#options BOOTP -#options BOOTP_NFSV3 -#options BOOTP_WIRED_TO=ue0 -#options ROOTDEVNAME=\"nfs:10.5.0.1:/tftpboot/root\" +#options NFS_ROOT # NFS usable as /, requires NFSCL +#options BOOTP_NFSROOT +#options BOOTP_COMPAT +#options BOOTP +#options BOOTP_NFSV3 +#options BOOTP_WIRED_TO=ue0 +#options ROOTDEVNAME=\"nfs:10.5.0.1:/tftpboot/root\" #FDT -options FDT -options FDT_DTB_STATIC +options FDT +options FDT_DTB_STATIC makeoptions FDT_DTS_FILE=exynos5420-peach-pit.dts diff --git a/sys/arm/conf/CNS11XXNAS b/sys/arm/conf/CNS11XXNAS index 53640ff..903fd05 100644 --- a/sys/arm/conf/CNS11XXNAS +++ b/sys/arm/conf/CNS11XXNAS @@ -103,7 +103,7 @@ device bpf device loop device md -device random # Entropy device +device random # Entropy device device usb @@ -114,12 +114,12 @@ device umass device scbus # SCSI bus (required for ATA/SCSI) device da # Direct Access (disks) device pass -device cfi +device cfi #device udav # Davicom DM9601E USB device geom_label device geom_journal -device geom_part_bsd +device geom_part_bsd options ROOTDEVNAME=\"ufs:da0s1a\" diff --git a/sys/arm/conf/CUBIEBOARD b/sys/arm/conf/CUBIEBOARD index a0b318e..40fe313 100644 --- a/sys/arm/conf/CUBIEBOARD +++ b/sys/arm/conf/CUBIEBOARD @@ -1,8 +1,9 @@ +# # CUBIEBOARD -- Custom configuration for the CUBIEBOARD ARM development # platform, check out http://www.cubieboard.org # -# For more information on this file, please read the handbook section on -# Kernel Configuration Files: +# For more information on this file, please read the config(5) manual page, +# and/or the handbook section on Kernel Configuration Files: # # http://www.FreeBSD.org/doc/en_US.ISO8859-1/books/handbook/kernelconfig-config.html # @@ -27,36 +28,46 @@ makeoptions WITHOUT_MODULES="ahc" options HZ=100 options SCHED_4BSD # 4BSD scheduler +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols -options GEOM_PART_BSD # BSD partition scheme -options GEOM_PART_MBR # MBR partition scheme -options TMPFS # Efficient memory filesystem +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client +options NFSLOCKD # Network Lock Manager +options NFS_ROOT # NFS usable as /, requires NFSCL options MSDOSFS # MSDOS Filesystem options CD9660 # ISO 9660 Filesystem options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables +options GEOM_PART_BSD # BSD partition scheme +options GEOM_PART_MBR # MBR partition scheme options COMPAT_43 # Compatible with BSD 4.3 [KEEP THIS!] options SCSI_DELAY=5000 # Delay (in ms) before probing SCSI options KTRACE # ktrace(1) support options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # Posix P1003_1B real-time extensions +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions options KBD_INSTALL_CDEV # install a CDEV entry in /dev -options PREEMPTION -options FREEBSD_BOOT_LOADER -options VFP # vfp/neon +options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +options VFP # Enable floating point hardware support -# Debugging +# Debugging for use in -current makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options BREAK_TO_DEBUGGER #options VERBOSE_SYSINIT # Enable verbose sysinit messages -options KDB +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: options DDB # Enable the kernel debugger options INVARIANTS # Enable calls of extra sanity checking options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS @@ -64,30 +75,24 @@ options WITNESS # Enable checks to detect deadlocks and cycles options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed #options DIAGNOSTIC -# NFS support -#options NFSCL -#options NFSSERVER # Network Filesystem Server -#options NFSCLIENT # Network Filesystem Client - -# Uncomment this for NFS root -#options NFS_ROOT # NFS usable as /, requires NFSCLIENT +# NFS root from boopt/dhcp +#options BOOTP #options BOOTP_NFSROOT #options BOOTP_COMPAT -#options BOOTP #options BOOTP_NFSV3 #options BOOTP_WIRED_TO=cpsw0 -# MMC/SD/SDIO card slot support -#device mmc # mmc/sd bus -#device mmcsd # mmc/sd flash cards - # Boot device is 2nd slice on MMC/SD card options ROOTDEVNAME=\"ufs:/dev/da0s2\" +# MMC/SD/SDIO Card slot support +#device mmc # mmc/sd bus +#device mmcsd # mmc/sd flash cards + # ATA controllers #device ahci # AHCI-compatible SATA controllers #device ata # Legacy ATA/SATA controllers -#options ATA_STATIC_ID # Static device numbering +#options ATA_STATIC_ID # Static device numbering # Console and misc device uart @@ -134,7 +139,7 @@ device emac device miibus # Flattened Device Tree -options FDT +options FDT # Configure using FDT/DTB data options FDT_DTB_STATIC makeoptions FDT_DTS_FILE=cubieboard.dts diff --git a/sys/arm/conf/CUBIEBOARD2 b/sys/arm/conf/CUBIEBOARD2 index 1c5184c..785364c 100644 --- a/sys/arm/conf/CUBIEBOARD2 +++ b/sys/arm/conf/CUBIEBOARD2 @@ -1,8 +1,9 @@ +# # CUBIEBOARD2 -- Custom configuration for the CUBIEBOARD2 ARM development # platform, check out http://www.cubieboard.org # -# For more information on this file, please read the handbook section on -# Kernel Configuration Files: +# For more information on this file, please read the config(5) manual page, +# and/or the handbook section on Kernel Configuration Files: # # http://www.FreeBSD.org/doc/en_US.ISO8859-1/books/handbook/kernelconfig-config.html # @@ -20,74 +21,79 @@ ident CUBIEBOARD2 -include "../allwinner/a20/std.a20" +include "../allwinner/a20/std.a20" makeoptions MODULES_OVERRIDE="" makeoptions WITHOUT_MODULES="ahc" options HZ=100 options SCHED_4BSD # 4BSD scheduler +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols -options GEOM_PART_BSD # BSD partition scheme -options GEOM_PART_MBR # MBR partition scheme -options TMPFS # Efficient memory filesystem +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem options SOFTUPDATES # Enable FFS soft updates support -options UFS_ACL # Support for access control lists +options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories -options MSDOSFS # MSDOS Filesystem +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client +options NFSLOCKD # Network Lock Manager +options NFS_ROOT # NFS usable as /, requires NFSCL +options MSDOSFS # MSDOS Filesystem options CD9660 # ISO 9660 Filesystem options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables +options GEOM_PART_BSD # BSD partition scheme +options GEOM_PART_MBR # MBR partition scheme options COMPAT_43 # Compatible with BSD 4.3 [KEEP THIS!] options SCSI_DELAY=5000 # Delay (in ms) before probing SCSI options KTRACE # ktrace(1) support -options SYSVSHM # SYSV-style shared memory -options SYSVMSG # SYSV-style message queues -options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # Posix P1003_1B real-time extensions +options SYSVSHM # SYSV-style shared memory +options SYSVMSG # SYSV-style message queues +options SYSVSEM # SYSV-style semaphores +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions options KBD_INSTALL_CDEV # install a CDEV entry in /dev -options PREEMPTION -options FREEBSD_BOOT_LOADER -options VFP # vfp/neon +options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +options VFP # Enable floating point hardware support +options SMP # Enable multiple cores -# Debugging +# Debugging for use in -current makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options BREAK_TO_DEBUGGER #options VERBOSE_SYSINIT # Enable verbose sysinit messages -options KDB +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: options DDB # Enable the kernel debugger options INVARIANTS # Enable calls of extra sanity checking options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS -options WITNESS # Enable checks to detect deadlocks and cycles +options WITNESS # Enable checks to detect deadlocks and cycles options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed #options DIAGNOSTIC -# NFS support -#options NFSCL -#options NFSSERVER # Network Filesystem Server -#options NFSCLIENT # Network Filesystem Client - -# Uncomment this for NFS root -#options NFS_ROOT # NFS usable as /, requires NFSCLIENT +# NFS root from boopt/dhcp +#options BOOTP #options BOOTP_NFSROOT #options BOOTP_COMPAT -#options BOOTP #options BOOTP_NFSV3 #options BOOTP_WIRED_TO=cpsw0 -# MMC/SD/SDIO card slot support -#device mmc # mmc/sd bus -#device mmcsd # mmc/sd flash cards - # Boot device is 2nd slice on MMC/SD card options ROOTDEVNAME=\"ufs:/dev/da0s2\" +# MMC/SD/SDIO Card slot support +#device mmc # mmc/sd bus +#device mmcsd # mmc/sd flash cards + # ATA controllers -#device ahci # AHCI-compatible SATA controllers -#device ata # Legacy ATA/SATA controllers -#options ATA_STATIC_ID # Static device numbering +#device ahci # AHCI-compatible SATA controllers +#device ata # Legacy ATA/SATA controllers +#options ATA_STATIC_ID # Static device numbering # Console and misc device uart @@ -98,8 +104,8 @@ device md device random # Entropy device # I2C support -#device iicbus -#device iic +#device iicbus +#device iic # GPIO device gpio @@ -114,8 +120,8 @@ device usb options USB_DEBUG #options USB_REQ_DEBUG #options USB_VERBOSE -#device uhci -#device ohci +#device uhci +#device ohci device ehci device umass @@ -125,7 +131,7 @@ device loop device ether device mii device smscphy -#device cpsw +#device cpsw device bpf device emac @@ -134,8 +140,7 @@ device emac device miibus # Flattened Device Tree -options FDT +options FDT # Configure using FDT/DTB data options FDT_DTB_STATIC makeoptions FDT_DTS_FILE=cubieboard2.dts -options SMP # Enable multiple cores diff --git a/sys/arm/conf/DIGI-CCWMX53 b/sys/arm/conf/DIGI-CCWMX53 index 7756558..20dec9c 100644 --- a/sys/arm/conf/DIGI-CCWMX53 +++ b/sys/arm/conf/DIGI-CCWMX53 @@ -22,7 +22,7 @@ include "IMX53" ident DIGI-CCWMX53 -makeoptions WITHOUT_MODULES="ahc" +makeoptions WITHOUT_MODULES="ahc" # required for netbooting #options BOOTP @@ -34,5 +34,5 @@ makeoptions WITHOUT_MODULES="ahc" #options ROOTDEVNAME=\"ufs:ada0s2a\" # Flattened Device Tree -options FDT_DTB_STATIC -makeoptions FDT_DTS_FILE=digi-ccwmx53.dts +options FDT_DTB_STATIC +makeoptions FDT_DTS_FILE=digi-ccwmx53.dts diff --git a/sys/arm/conf/DOCKSTAR b/sys/arm/conf/DOCKSTAR index a3c57a1..ad1e74d 100644 --- a/sys/arm/conf/DOCKSTAR +++ b/sys/arm/conf/DOCKSTAR @@ -54,15 +54,15 @@ options FDT_DTB_STATIC # Misc pseudo devices device bpf # Required for DHCP -device firmware # firmware(9) required for USB wlan -device gif # IPv6 and IPv4 tunneling +device firmware # firmware(9) required for USB wlan +device gif # IPv6 and IPv4 tunneling device loop # Network loopback -device md # Memory/malloc disk +device md # Memory/malloc disk device pty # BSD-style compatibility pseudo ttys device random # Entropy device -device tun # Packet tunnel. +device tun # Packet tunnel. device ether # Required for all ethernet devices -device vlan # 802.1Q VLAN support +device vlan # 802.1Q VLAN support device wlan # 802.11 WLAN support # cam support for umass and ahci @@ -80,66 +80,66 @@ device e1000phy # USB options USB_HOST_ALIGN=32 # Align DMA to cacheline -#options USB_DEBUG # Compile in USB debug support -device usb # Basic usb support -device ehci # USB host controller -device umass # Mass storage -device uhid # Human-interface devices -device rum # Ralink Technology RT2501USB wireless NICs -device uath # Atheros AR5523 wireless NICs -device ural # Ralink Technology RT2500USB wireless NICs -device zyd # ZyDAS zb1211/zb1211b wireless NICs -device urtw # Realtek RTL8187B/L USB -device upgt # Conexant/Intersil PrismGT SoftMAC USB -device u3g # USB-based 3G modems (Option, Huawei, Sierra) +#options USB_DEBUG # Compile in USB debug support +device usb # Basic usb support +device ehci # USB host controller +device umass # Mass storage +device uhid # Human-interface devices +device rum # Ralink Technology RT2501USB wireless NICs +device uath # Atheros AR5523 wireless NICs +device ural # Ralink Technology RT2500USB wireless NICs +device zyd # ZyDAS zb1211/zb1211b wireless NICs +device urtw # Realtek RTL8187B/L USB +device upgt # Conexant/Intersil PrismGT SoftMAC USB +device u3g # USB-based 3G modems (Option, Huawei, Sierra) # I2C (TWSI) device iic device iicbus # Sound -device sound -device snd_uaudio +device sound +device snd_uaudio #crypto -device cesa # Marvell security engine -device crypto -device cryptodev +device cesa # Marvell security engine +device crypto +device cryptodev # IPSec -device enc -options IPSEC -options IPSEC_NAT_T -options TCP_SIGNATURE # include support for RFC 2385 +device enc +options IPSEC +options IPSEC_NAT_T +options TCP_SIGNATURE # include support for RFC 2385 # IPFW -options IPFIREWALL -options IPFIREWALL_DEFAULT_TO_ACCEPT -options IPFIREWALL_VERBOSE -options IPFIREWALL_VERBOSE_LIMIT=100 -options IPFIREWALL_NAT -options LIBALIAS -options DUMMYNET -options IPDIVERT +options IPFIREWALL +options IPFIREWALL_DEFAULT_TO_ACCEPT +options IPFIREWALL_VERBOSE +options IPFIREWALL_VERBOSE_LIMIT=100 +options IPFIREWALL_NAT +options LIBALIAS +options DUMMYNET +options IPDIVERT #PF -device pf -device pflog -device pfsync +device pf +device pflog +device pfsync # ALTQ, required for PF -options ALTQ # Basic ALTQ support -options ALTQ_CBQ # Class Based Queueing -options ALTQ_RED # Random Early Detection -options ALTQ_RIO # RED In/Out -options ALTQ_HFSC # Hierarchical Packet Scheduler -options ALTQ_CDNR # Traffic conditioner -options ALTQ_PRIQ # Priority Queueing -options ALTQ_NOPCC # Required if the TSC is unusable +options ALTQ # Basic ALTQ support +options ALTQ_CBQ # Class Based Queueing +options ALTQ_RED # Random Early Detection +options ALTQ_RIO # RED In/Out +options ALTQ_HFSC # Hierarchical Packet Scheduler +options ALTQ_CDNR # Traffic conditioner +options ALTQ_PRIQ # Priority Queueing +options ALTQ_NOPCC # Required if the TSC is unusable #options ALTQ_DEBUG # Debugging -makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols +makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options BREAK_TO_DEBUGGER options ALT_BREAK_TO_DEBUGGER options DDB @@ -154,7 +154,7 @@ options INVARIANT_SUPPORT # Extra sanity checks of internal structures, require # Enable these options for nfs root configured via BOOTP. options NFSCL # Network Filesystem Client options NFSLOCKD # Network Lock Manager -#options NFS_ROOT # NFS usable as /, requires NFSCLIENT +#options NFS_ROOT # NFS usable as /, requires NFSCL #options BOOTP #options BOOTP_NFSROOT #options BOOTP_NFSV3 diff --git a/sys/arm/conf/DREAMPLUG-1001 b/sys/arm/conf/DREAMPLUG-1001 index 7369592..135cba0 100644 --- a/sys/arm/conf/DREAMPLUG-1001 +++ b/sys/arm/conf/DREAMPLUG-1001 @@ -57,22 +57,22 @@ options FDT_DTB_STATIC # Misc pseudo devices device bpf # Required for DHCP -device firmware # firmware(9) required for USB wlan -device gif # IPv6 and IPv4 tunneling +device firmware # firmware(9) required for USB wlan +device gif # IPv6 and IPv4 tunneling device loop # Network loopback -device md # Memory/malloc disk +device md # Memory/malloc disk device pty # BSD-style compatibility pseudo ttys device random # Entropy device -device tun # Packet tunnel. +device tun # Packet tunnel. device ether # Required for all ethernet devices -device vlan # 802.1Q VLAN support +device vlan # 802.1Q VLAN support device wlan # 802.11 WLAN support # cam support for umass and ahci device scbus device pass device da -device cd +device cd # Serial ports device uart @@ -84,18 +84,18 @@ device e1000phy # USB options USB_HOST_ALIGN=32 # Align DMA to cacheline -#options USB_DEBUG # Compile in USB debug support -device usb # Basic usb support -device ehci # USB host controller -device umass # Mass storage -device uhid # Human-interface devices -device rum # Ralink Technology RT2501USB wireless NICs -device uath # Atheros AR5523 wireless NICs -device ural # Ralink Technology RT2500USB wireless NICs -device zyd # ZyDAS zb1211/zb1211b wireless NICs -device urtw # Realtek RTL8187B/L USB -device upgt # Conexant/Intersil PrismGT SoftMAC USB -device u3g # USB-based 3G modems (Option, Huawei, Sierra) +#options USB_DEBUG # Compile in USB debug support +device usb # Basic usb support +device ehci # USB host controller +device umass # Mass storage +device uhid # Human-interface devices +device rum # Ralink Technology RT2501USB wireless NICs +device uath # Atheros AR5523 wireless NICs +device ural # Ralink Technology RT2500USB wireless NICs +device zyd # ZyDAS zb1211/zb1211b wireless NICs +device urtw # Realtek RTL8187B/L USB +device upgt # Conexant/Intersil PrismGT SoftMAC USB +device u3g # USB-based 3G modems (Option, Huawei, Sierra) # I2C (TWSI) device iic @@ -106,48 +106,48 @@ device mvs device ahci # Sound -device sound -device snd_uaudio +device sound +device snd_uaudio #crypto -device cesa # Marvell security engine -device crypto -device cryptodev +device cesa # Marvell security engine +device crypto +device cryptodev # IPSec -device enc -options IPSEC -options IPSEC_NAT_T -options TCP_SIGNATURE # include support for RFC 2385 +device enc +options IPSEC +options IPSEC_NAT_T +options TCP_SIGNATURE # include support for RFC 2385 # IPFW -options IPFIREWALL -options IPFIREWALL_DEFAULT_TO_ACCEPT -options IPFIREWALL_VERBOSE -options IPFIREWALL_VERBOSE_LIMIT=100 -options IPFIREWALL_NAT -options LIBALIAS -options DUMMYNET -options IPDIVERT +options IPFIREWALL +options IPFIREWALL_DEFAULT_TO_ACCEPT +options IPFIREWALL_VERBOSE +options IPFIREWALL_VERBOSE_LIMIT=100 +options IPFIREWALL_NAT +options LIBALIAS +options DUMMYNET +options IPDIVERT #PF -device pf -device pflog -device pfsync +device pf +device pflog +device pfsync # ALTQ, required for PF -options ALTQ # Basic ALTQ support -options ALTQ_CBQ # Class Based Queueing -options ALTQ_RED # Random Early Detection -options ALTQ_RIO # RED In/Out -options ALTQ_HFSC # Hierarchical Packet Scheduler -options ALTQ_CDNR # Traffic conditioner -options ALTQ_PRIQ # Priority Queueing -options ALTQ_NOPCC # Required if the TSC is unusable +options ALTQ # Basic ALTQ support +options ALTQ_CBQ # Class Based Queueing +options ALTQ_RED # Random Early Detection +options ALTQ_RIO # RED In/Out +options ALTQ_HFSC # Hierarchical Packet Scheduler +options ALTQ_CDNR # Traffic conditioner +options ALTQ_PRIQ # Priority Queueing +options ALTQ_NOPCC # Required if the TSC is unusable #options ALTQ_DEBUG # Debugging -makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols +makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options BREAK_TO_DEBUGGER options ALT_BREAK_TO_DEBUGGER options DDB @@ -162,7 +162,7 @@ options INVARIANT_SUPPORT # Extra sanity checks of internal structures, require # Enable these options for nfs root configured via BOOTP. options NFSCL # Network Filesystem Client options NFSLOCKD # Network Lock Manager -#options NFS_ROOT # NFS usable as /, requires NFSCLIENT +#options NFS_ROOT # NFS usable as /, requires NFSCL #options BOOTP #options BOOTP_NFSROOT #options BOOTP_NFSV3 @@ -178,7 +178,7 @@ options ROOTDEVNAME=\"ufs:/dev/da1s1a\" # create a kernel config file that looks like this: # # include DREAMPLUG-1001 -# nomakeoptions FDT_DTS_FILE +# nomakeoptions FDT_DTS_FILE # makeoptions FDT_DTS_FILE=dreamplug-1001N.dts # device nand diff --git a/sys/arm/conf/EA3250 b/sys/arm/conf/EA3250 index 82bc386..d2d691b 100644 --- a/sys/arm/conf/EA3250 +++ b/sys/arm/conf/EA3250 @@ -19,7 +19,7 @@ options INET6 # IPv6 communications protocols options FFS # Berkeley Fast Filesystem options NFSCL # Network Filesystem Client options NFSLOCKD # Network Lock Manager -options NFS_ROOT # NFS usable as /, requires NFSCLIENT +options NFS_ROOT # NFS usable as /, requires NFSCL options GEOM_PART_BSD # BSD partition scheme options GEOM_PART_MBR # MBR partition scheme options TMPFS # Efficient memory filesystem diff --git a/sys/arm/conf/EFIKA_MX b/sys/arm/conf/EFIKA_MX index 677d73d..459cd79 100644 --- a/sys/arm/conf/EFIKA_MX +++ b/sys/arm/conf/EFIKA_MX @@ -1,3 +1,4 @@ +# # Kernel configuration for Efika MX Smarttop/Smartbook boards # # For more information on this file, please read the config(5) manual page, @@ -21,34 +22,32 @@ ident EFIKA_MX include "../freescale/imx/std.imx51" -makeoptions WITHOUT_MODULES="ahc" - -makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols -#options DEBUG +makeoptions WITHOUT_MODULES="ahc" options SCHED_4BSD # 4BSD scheduler -#options PREEMPTION # Enable kernel thread preemption +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking -#options INET6 # IPv6 communications protocols -#options SCTP # Stream Control Transmission Protocol +options INET6 # IPv6 communications protocols +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS #options MD_ROOT # MD is a potential root device options NFSCL # New Network Filesystem Client #options NFSD # New Network Filesystem Server options NFSLOCKD # Network Lock Manager options NFS_ROOT # NFS usable as /, requires NFSCL -options TMPFS # Efficient memory filesystem options MSDOSFS # MSDOS Filesystem options CD9660 # ISO 9660 Filesystem -#options PROCFS # Process filesystem (requires PSEUDOFS) +options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables options GEOM_PART_BSD # BSD partition scheme options GEOM_PART_MBR # MBR partition scheme -options GEOM_PART_GPT # GUID Partition Tables. options GEOM_LABEL # Provides labelization #options COMPAT_FREEBSD5 # Compatible with FreeBSD5 #options COMPAT_FREEBSD6 # Compatible with FreeBSD6 @@ -59,16 +58,34 @@ options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions +options KBD_INSTALL_CDEV # install a CDEV entry in /dev +options PLATFORM options INCLUDE_CONFIG_FILE # Include this file in kernel -options VFP # vfp/neon +options VFP # Enable floating point hardware support + +# Debugging for use in -current +makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols +options BREAK_TO_DEBUGGER +#options VERBOSE_SYSINIT # Enable verbose sysinit messages +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: +options DDB # Enable the kernel debugger +#options GDB # Support remote GDB +options DEADLKRES # Enable the deadlock resolver +options INVARIANTS # Enable calls of extra sanity checking +options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS +options WITNESS # Enable checks to detect deadlocks and cycles +#options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed -# required for netbooting +# NFS root from boopt/dhcp #options BOOTP -#options BOOTP_COMPAT #options BOOTP_NFSROOT +#options BOOTP_COMPAT #options BOOTP_NFSV3 #options BOOTP_WIRED_TO=ue0 -# + options ROOTDEVNAME=\"ufs:ada0s2a\" @@ -79,18 +96,6 @@ options ROOTDEVNAME=\"ufs:ada0s2a\" #options NO_SYSCTL_DESCR #options RWLOCK_NOINLINE -# Debugging support. Always need this: -options KDB # Enable kernel debugger support. -# For minimum debugger support (stable branch) use: -#options KDB_TRACE # Print a stack trace for a panic. -# For full debugger support use this instead: -options DDB # Support DDB. -#options GDB # Support remote GDB. -options DEADLKRES # Enable the deadlock resolver -options INVARIANTS # Enable calls of extra sanity checking -options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS -options WITNESS # Enable checks to detect deadlocks and cycles - # The `bpf' device enables the Berkeley Packet Filter. # Be aware of the administrative consequences of enabling this! # Note that 'bpf' is required for DHCP. @@ -162,15 +167,15 @@ device wlan_tkip # 802.11 TKIP support device wlan_amrr # AMRR transmit rate control algorithm # Flattened Device Tree -options FDT -options FDT_DTB_STATIC -makeoptions FDT_DTS_FILE=efikamx.dts +options FDT # Configure using FDT/DTB data +options FDT_DTB_STATIC +makeoptions FDT_DTS_FILE=efikamx.dts # NOTE: serial console will be disabled if syscons enabled # Uncomment following lines for framebuffer/syscons support device sc device kbdmux -options SC_DFLT_FONT # compile font in -makeoptions SC_DFLT_FONT=cp437 +options SC_DFLT_FONT # compile font in +makeoptions SC_DFLT_FONT=cp437 device ukbd # Allow keyboard like HIDs to control console device ums diff --git a/sys/arm/conf/EXYNOS5.common b/sys/arm/conf/EXYNOS5.common index 7cd577d..0179d5b 100644 --- a/sys/arm/conf/EXYNOS5.common +++ b/sys/arm/conf/EXYNOS5.common @@ -1,3 +1,4 @@ +# # Kernel configuration for Samsung Exynos 5 SoC. # # For more information on this file, please read the config(5) manual page, @@ -20,70 +21,72 @@ makeoptions MODULES_OVERRIDE="" makeoptions WITHOUT_MODULES="ahc" -makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols makeoptions WERROR="-Werror" options HZ=100 options SCHED_4BSD # 4BSD scheduler +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols -options GEOM_PART_BSD # BSD partition scheme -options GEOM_PART_MBR # MBR partition scheme -options GEOM_PART_GPT # GUID partition tables -options TMPFS # Efficient memory filesystem +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem -options SOFTUPDATES +options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client +options NFSLOCKD # Network Lock Manager +options NFS_ROOT # NFS usable as /, requires NFSCL options MSDOSFS # MSDOS Filesystem options CD9660 # ISO 9660 Filesystem options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables +options GEOM_PART_BSD # BSD partition scheme +options GEOM_PART_MBR # MBR partition scheme options COMPAT_43 # Compatible with BSD 4.3 [KEEP THIS!] options SCSI_DELAY=5000 # Delay (in ms) before probing SCSI -options KTRACE +options KTRACE # ktrace(1) support options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # Posix P1003_1B real-time extensions -options KBD_INSTALL_CDEV -options PREEMPTION -options FREEBSD_BOOT_LOADER -options VFP # vfp/neon +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions +options KBD_INSTALL_CDEV # install a CDEV entry in /dev +options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +options VFP # Enable floating point hardware support +options SMP # Enable multiple cores -options SMP - -# Debugging +# Debugging for use in -current makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options BREAK_TO_DEBUGGER -#options VERBOSE_SYSINIT # Enable verbose sysinit messages -options KDB +#options VERBOSE_SYSINIT # Enable verbose sysinit messages +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: options DDB # Enable the kernel debugger #options INVARIANTS # Enable calls of extra sanity checking #options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS -#options WITNESS # Enable checks to detect deadlocks and cycles -#options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed +#options WITNESS # Enable checks to detect deadlocks and cycles +#options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed #options DIAGNOSTIC -# NFS support -options NFSCL # Network Filesystem Client -options NFSLOCKD # Network Lock Manager -options NFS_ROOT # NFS usable as /, requires NFSCLIENT +# NFS root from boopt/dhcp +#options BOOTP +#options BOOTP_NFSROOT +#options BOOTP_COMPAT +#options BOOTP_NFSV3 +#options BOOTP_WIRED_TO=ue0 -# Uncomment this for NFS root -#options NFS_ROOT # NFS usable as /, requires NFSCL -#options BOOTP_NFSROOT -#options BOOTP_COMPAT -#options BOOTP -#options BOOTP_NFSV3 -#options BOOTP_WIRED_TO=ue0 +options ROOTDEVNAME=\"ufs:/dev/da0\" +# MMC/SD/SDIO Card slot support device mmc # mmc/sd bus device mmcsd # mmc/sd flash cards device dwmmc -options ROOTDEVNAME=\"ufs:/dev/da0\" - # Pseudo devices device loop @@ -96,8 +99,8 @@ device gpio options USB_HOST_ALIGN=64 # Align usb buffers to cache line size. device usb options USB_DEBUG -#options USB_REQ_DEBUG -#options USB_VERBOSE +#options USB_REQ_DEBUG +#options USB_VERBOSE #device musb device ehci #device ohci diff --git a/sys/arm/conf/HL201 b/sys/arm/conf/HL201 index efdea82..8a15544 100644 --- a/sys/arm/conf/HL201 +++ b/sys/arm/conf/HL201 @@ -143,5 +143,5 @@ options FDT_DTB_STATIC makeoptions FDT_DTS_FILE=hl201.dts options EARLY_PRINTF -options SOCDEV_PA=0xfc000000 +options SOCDEV_PA=0xfc000000 options SOCDEV_VA=0xdc000000 diff --git a/sys/arm/conf/IMX53 b/sys/arm/conf/IMX53 index 5fd5697..12cd4cf 100644 --- a/sys/arm/conf/IMX53 +++ b/sys/arm/conf/IMX53 @@ -1,3 +1,4 @@ +# # Kernel configuration for i.MX53 boards # # For more information on this file, please read the config(5) manual page, @@ -21,32 +22,29 @@ ident IMX53 include "../freescale/imx/std.imx53" -makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols -#options DEBUG - options SCHED_4BSD # 4BSD scheduler -#options PREEMPTION # Enable kernel thread preemption +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols -#options SCTP # Stream Control Transmission Protocol +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories options UFS_GJOURNAL # Enable gjournal-based UFS journaling -#options MD_ROOT # MD is a potential root device +options QUOTA # Enable disk quotas for UFS options NFSCL # New Network Filesystem Client #options NFSD # New Network Filesystem Server options NFSLOCKD # Network Lock Manager options NFS_ROOT # NFS usable as /, requires NFSCL -options TMPFS # Efficient memory filesystem options MSDOSFS # MSDOS Filesystem options CD9660 # ISO 9660 Filesystem -#options PROCFS # Process filesystem (requires PSEUDOFS) +options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables options GEOM_PART_BSD # BSD partition scheme options GEOM_PART_MBR # MBR partition scheme -options GEOM_PART_GPT # GUID Partition Tables. options GEOM_LABEL # Provides labelization #options COMPAT_FREEBSD5 # Compatible with FreeBSD5 #options COMPAT_FREEBSD6 # Compatible with FreeBSD6 @@ -57,27 +55,31 @@ options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions +options KBD_INSTALL_CDEV # install a CDEV entry in /dev +options PLATFORM options INCLUDE_CONFIG_FILE # Include this file in kernel -options VFP # vfp/neon - -# kernel/memory size reduction -#options MUTEX_NOINLINE -#options NO_FFS_SNAPSHOT -#options NO_SWAPPING -#options NO_SYSCTL_DESCR -#options RWLOCK_NOINLINE +options VFP # Enable floating point hardware support -# Debugging support. Always need this: -options KDB # Enable kernel debugger support. +# Debugging for use in -current +makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols +options KDB # Enable kernel debugger support # For minimum debugger support (stable branch) use: -#options KDB_TRACE # Print a stack trace for a panic. +#options KDB_TRACE # Print a stack trace for a panic # For full debugger support use this instead: -options DDB # Support DDB. -#options GDB # Support remote GDB. +options DDB # Enable the kernel debugger +#options GDB # Support remote GDB options DEADLKRES # Enable the deadlock resolver options INVARIANTS # Enable calls of extra sanity checking options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS options WITNESS # Enable checks to detect deadlocks and cycles +#options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed + +# kernel/memory size reduction +#options MUTEX_NOINLINE +#options NO_FFS_SNAPSHOT +#options NO_SWAPPING +#options NO_SYSCTL_DESCR +#options RWLOCK_NOINLINE # The `bpf' device enables the Berkeley Packet Filter. # Be aware of the administrative consequences of enabling this! @@ -161,12 +163,12 @@ device wlan_amrr # AMRR transmit rate control algorithm # Flattened Device Tree -options FDT +options FDT # Configure using FDT/DTB data # NOTE: serial console will be disabled if syscons enabled # Uncomment following lines for framebuffer/syscons support #device sc #device vt #device kbdmux -#options SC_DFLT_FONT # compile font in -#makeoptions SC_DFLT_FONT=cp437 +#options SC_DFLT_FONT # compile font in +#makeoptions SC_DFLT_FONT=cp437 diff --git a/sys/arm/conf/IMX53-QSB b/sys/arm/conf/IMX53-QSB index fdaa4ec..fdde591 100644 --- a/sys/arm/conf/IMX53-QSB +++ b/sys/arm/conf/IMX53-QSB @@ -34,5 +34,5 @@ options HZ=250 # 4ms scheduling quantum #options ROOTDEVNAME=\"ufs:ada0s2a\" # Flattened Device Tree -options FDT_DTB_STATIC -makeoptions FDT_DTS_FILE=imx53-qsb.dts +options FDT_DTB_STATIC +makeoptions FDT_DTS_FILE=imx53-qsb.dts diff --git a/sys/arm/conf/IMX6 b/sys/arm/conf/IMX6 index 007862b..bdaab26 100644 --- a/sys/arm/conf/IMX6 +++ b/sys/arm/conf/IMX6 @@ -1,3 +1,4 @@ +# # Kernel configuration for Freescale i.MX6 systems. # # For more information on this file, please read the config(5) manual page, @@ -20,140 +21,143 @@ ident IMX6 include "../freescale/imx/std.imx6" -options HZ=500 # Scheduling quantum is 2 milliseconds. -options SCHED_ULE # ULE scheduler -options PREEMPTION # Enable kernel thread preemption -options INET # InterNETworking -options INET6 # IPv6 communications protocols -options SCTP # Stream Control Transmission Protocol -options FFS # Berkeley Fast Filesystem -options SOFTUPDATES # Enable FFS soft updates support -options UFS_ACL # Support for access control lists -options UFS_DIRHASH # Improve performance on big directories -options UFS_GJOURNAL # Enable gjournal-based UFS journaling -#options MD_ROOT # MD is a potential root device -options NFSCL # New Network Filesystem Client -#options NFSD # New Network Filesystem Server -options NFSLOCKD # Network Lock Manager -options NFS_ROOT # NFS usable as /, requires NFSCL +options HZ=500 # Scheduling quantum is 2 milliseconds. +options SCHED_ULE # ULE scheduler +options PREEMPTION # Enable kernel thread preemption +options INET # InterNETworking +options INET6 # IPv6 communications protocols +options SCTP # Stream Control Transmission Protocol +options FFS # Berkeley Fast Filesystem +options SOFTUPDATES # Enable FFS soft updates support +options UFS_ACL # Support for access control lists +options UFS_DIRHASH # Improve performance on big directories +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client +#options NFSD # New Network Filesystem Server +options NFSLOCKD # Network Lock Manager +options NFS_ROOT # NFS usable as /, requires NFSCL +options MSDOSFS # MSDOS Filesystem +options CD9660 # ISO 9660 Filesystem +options PROCFS # Process filesystem (requires PSEUDOFS) +options PSEUDOFS # Pseudo-filesystem framework options TMPFS # Efficient memory filesystem -options MSDOSFS # MSDOS Filesystem -options CD9660 # ISO 9660 Filesystem -#options PROCFS # Process filesystem (requires PSEUDOFS) -options PSEUDOFS # Pseudo-filesystem framework +options GEOM_PART_GPT # GUID Partition Tables options GEOM_PART_BSD # BSD partition scheme options GEOM_PART_MBR # MBR partition scheme -options GEOM_PART_GPT # GUID Partition Tables. -options GEOM_LABEL # Provides labelization -options KTRACE # ktrace(1) support -options SYSVSHM # SYSV-style shared memory -options SYSVMSG # SYSV-style message queues -options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions -options INCLUDE_CONFIG_FILE # Include this file in kernel - -# Debugging support. Always need this: -options KDB # Enable kernel debugger support. -# For minimum debugger support use KDB_TRACE, for interactive use DDB. -#options KDB_TRACE # Print a stack trace for a panic. -options DDB # Support DDB. +options GEOM_LABEL # Provides labelization +options KTRACE # ktrace(1) support +options SYSVSHM # SYSV-style shared memory +options SYSVMSG # SYSV-style message queues +options SYSVSEM # SYSV-style semaphores +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions +options KBD_INSTALL_CDEV # install a CDEV entry in /dev +options INCLUDE_CONFIG_FILE # Include this file in kernel +options PLATFORM +options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +options VFP # Enable floating point hardware support +options SMP # Enable multiple cores + +# Debugging for use in -current +makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic # For full debugger support use this instead: -#options GDB # Support remote GDB. +options DDB # Enable the kernel debugger +#options GDB # Support remote GDB. # Other debugging options... -makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols -options ALT_BREAK_TO_DEBUGGER # Use <CR><tilde><ctrl-b> to enter debugger. -#options DEBUG -#options DEADLKRES # Enable the deadlock resolver -#options INVARIANTS # Enable calls of extra sanity checking -#options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS -#options WITNESS # Enable checks to detect deadlocks and cycles +options ALT_BREAK_TO_DEBUGGER # Use <CR><tilde><ctrl-b> to enter debugger. +#options DEADLKRES # Enable the deadlock resolver +#options INVARIANTS # Enable calls of extra sanity checking +#options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS +#options WITNESS # Enable checks to detect deadlocks and cycles +#options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed +#options DIAGNOSTIC + +# NFS root from boopt/dhcp +#options BOOTP +#options BOOTP_NFSROOT +#options BOOTP_COMPAT +#options BOOTP_NFSV3 +#options BOOTP_WIRED_TO=ffec0 + +# U-Boot stuff lives on slice 1, FreeBSD on slice 2. +options ROOTDEVNAME=\"ufs:mmcsd0s2a\" # Pseudo devices. -device loop # Network loopback -device random # Entropy device -device vlan # 802.1Q VLAN support -device tun # Packet tunnel. -device md # Memory "disks" -#device gif # IPv6 and IPv4 tunneling -#device firmware # firmware assist module -device ether # Ethernet support -device miibus # Required for ethernet -device bpf # Berkeley packet filter (required for DHCP) +device loop # Network loopback +device random # Entropy device +device vlan # 802.1Q VLAN support +device tun # Packet tunnel. +device md # Memory "disks" +#device gif # IPv6 and IPv4 tunneling +#device firmware # firmware assist module +device ether # Ethernet support +device miibus # Required for ethernet +device bpf # Berkeley packet filter (required for DHCP) # General-purpose input/output -device gpio +device gpio # Serial (COM) ports -device uart # Multi-uart driver +device uart # Multi-uart driver # SDCard -device sdhci # SD controller -device mmc # SD/MMC protocol -device mmcsd # SDCard disk device +device sdhci # SD controller +device mmc # SD/MMC protocol +device mmcsd # SDCard disk device # SCSI peripherals -device scbus # SCSI bus (required for ATA/SCSI) -device da # Direct Access (disks) -device cd # CD -device pass # Passthrough device (direct ATA/SCSI access) +device scbus # SCSI bus (required for ATA/SCSI) +device da # Direct Access (disks) +device cd # CD +device pass # Passthrough device (direct ATA/SCSI access) # USB support -#options USB_DEBUG # enable debug msgs -device ehci # OHCI USB interface -device usb # USB Bus (required) -device umass # Disks/Mass storage - Requires scbus and da -device uhid # "Human Interface Devices" -device u3g # USB modems -#device ukbd # Allow keyboard like HIDs to control console -#device ums # USB mouse +#options USB_DEBUG # enable debug msgs +device ehci # OHCI USB interface +device usb # USB Bus (required) +device umass # Disks/Mass storage - Requires scbus and da +device uhid # "Human Interface Devices" +device u3g # USB modems +#device ukbd # Allow keyboard like HIDs to control console +#device ums # USB mouse # USB Ethernet, requires miibus -#device aue # ADMtek USB Ethernet -#device axe # ASIX Electronics USB Ethernet -#device cdce # Generic USB over Ethernet -#device cue # CATC USB Ethernet -#device kue # Kawasaki LSI USB Ethernet -#device rue # RealTek RTL8150 USB Ethernet -#device udav # Davicom DM9601E USB +#device aue # ADMtek USB Ethernet +#device axe # ASIX Electronics USB Ethernet +#device cdce # Generic USB over Ethernet +#device cue # CATC USB Ethernet +#device kue # Kawasaki LSI USB Ethernet +#device rue # RealTek RTL8150 USB Ethernet +#device udav # Davicom DM9601E USB # USB Wireless -#device rum # Ralink Technology RT2501USB wireless NICs +#device rum # Ralink Technology RT2501USB wireless NICs # Wireless NIC cards -#device wlan # 802.11 support -#device wlan_wep # 802.11 WEP support -#device wlan_ccmp # 802.11 CCMP support -#device wlan_tkip # 802.11 TKIP support -#device wlan_amrr # AMRR transmit rate control algorithm +#device wlan # 802.11 support +#device wlan_wep # 802.11 WEP support +#device wlan_ccmp # 802.11 CCMP support +#device wlan_tkip # 802.11 TKIP support +#device wlan_amrr # AMRR transmit rate control algorithm # NOTE: serial console will be disabled if syscons enabled # Uncomment following lines for framebuffer/syscons support # Wandboard has no video console support yet. -#device sc -#device kbdmux -#options SC_DFLT_FONT # compile font in -#makeoptions SC_DFLT_FONT=cp437 - -# required for netbooting -#options BOOTP -#options BOOTP_COMPAT -#options BOOTP_NFSROOT -#options BOOTP_NFSV3 -#options BOOTP_WIRED_TO=ffec0 - -# U-Boot stuff lives on slice 1, FreeBSD on slice 2. -options ROOTDEVNAME=\"ufs:mmcsd0s2a\" +#device sc +#device kbdmux +#options SC_DFLT_FONT # compile font in +#makeoptions SC_DFLT_FONT=cp437 -# ARM and SoC-specific options -options FDT # Configure using FDT/DTB data. -options SMP # Enable multiple cores -options VFP # Enable floating point hardware support -options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +# Flattened Device Tree +options FDT # Configure using FDT/DTB data # SoC-specific devices -device ffec # Freescale Fast Ethernet Controller -device fsliic # Freescale i2c/iic -device iic # iic protocol -device iicbus # iic bus -#device imxwdt # Watchdog. WARNING: can't be disabled!!! +device ffec # Freescale Fast Ethernet Controller +device fsliic # Freescale i2c/iic +device iic # iic protocol +device iicbus # iic bus +#device imxwdt # Watchdog. WARNING: can't be disabled!!! diff --git a/sys/arm/conf/PANDABOARD b/sys/arm/conf/PANDABOARD index cc345cb..7a71d12 100644 --- a/sys/arm/conf/PANDABOARD +++ b/sys/arm/conf/PANDABOARD @@ -1,8 +1,9 @@ +# # PANDABOARD -- Custom configuration for the PandaBoard ARM development # platform, check out www.pandaboard.org # -# For more information on this file, please read the handbook section on -# Kernel Configuration Files: +# For more information on this file, please read the config(5) manual page, +# and/or the handbook section on Kernel Configuration Files: # # http://www.FreeBSD.org/doc/en_US.ISO8859-1/books/handbook/kernelconfig-config.html # @@ -28,44 +29,53 @@ hints "PANDABOARD.hints" include "../ti/omap4/pandaboard/std.pandaboard" -#To statically compile in device wiring instead of /boot/device.hints makeoptions MODULES_OVERRIDE="" -makeoptions WITHOUT_MODULES="ahc" +makeoptions WITHOUT_MODULES="ahc" options HZ=100 options SCHED_4BSD # 4BSD scheduler +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories -options GEOM_PART_BSD # BSD partition scheme -options GEOM_PART_MBR # MBR partition scheme -options TMPFS # Efficient memory filesystem +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client +options NFSLOCKD # Network Lock Manager +options NFS_ROOT # NFS usable as /, requires NFSCL options MSDOSFS # MSDOS Filesystem options CD9660 # ISO 9660 Filesystem options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables +options GEOM_PART_BSD # BSD partition scheme +options GEOM_PART_MBR # MBR partition scheme options COMPAT_43 # Compatible with BSD 4.3 [KEEP THIS!] options SCSI_DELAY=5000 # Delay (in ms) before probing SCSI options KTRACE # ktrace(1) support options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # Posix P1003_1B real-time extensions +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions options KBD_INSTALL_CDEV # install a CDEV entry in /dev -options PREEMPTION options PLATFORM -options FREEBSD_BOOT_LOADER -options VFP # vfp/neon -options SMP # Enable multiple cores +options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +options VFP # Enable floating point hardware support +options SMP # Enable multiple cores # Debugging for use in -current makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options BREAK_TO_DEBUGGER #options VERBOSE_SYSINIT # Enable verbose sysinit messages -options KDB +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: options DDB # Enable the kernel debugger #options INVARIANTS # Enable calls of extra sanity checking #options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS @@ -73,14 +83,10 @@ options DDB # Enable the kernel debugger #options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed #options DIAGNOSTIC -# NFS support -options NFSCL - -# NFS root -options NFS_ROOT # NFS usable as /, requires NFSCLIENT +# NFS root from boopt/dhcp +#options BOOTP #options BOOTP_NFSROOT #options BOOTP_COMPAT -#options BOOTP #options BOOTP_NFSV3 #options BOOTP_WIRED_TO=ue0 @@ -146,6 +152,6 @@ device twl_vreg device twl_clks # Flattened Device Tree -options FDT -options FDT_DTB_STATIC -makeoptions FDT_DTS_FILE=pandaboard.dts +options FDT # Configure using FDT/DTB data +options FDT_DTB_STATIC +makeoptions FDT_DTS_FILE=pandaboard.dts diff --git a/sys/arm/conf/RK3188 b/sys/arm/conf/RK3188 index d6a26dd..ab9d20a 100644 --- a/sys/arm/conf/RK3188 +++ b/sys/arm/conf/RK3188 @@ -1,7 +1,8 @@ +# # Kernel configuration for Rockchip RK3188 systems. # -# For more information on this file, please read the handbook section on -# Kernel Configuration Files: +# For more information on this file, please read the config(5) manual page, +# and/or the handbook section on Kernel Configuration Files: # # http://www.FreeBSD.org/doc/en_US.ISO8859-1/books/handbook/kernelconfig-config.html # @@ -23,56 +24,61 @@ include "../rockchip/std.rk30xx" options HZ=100 options SCHED_4BSD # 4BSD scheduler +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories -options GEOM_PART_BSD # BSD partition scheme -options GEOM_PART_MBR # MBR partition scheme -options TMPFS # Efficient memory filesystem +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client +options NFSLOCKD # Network Lock Manager +options NFS_ROOT # NFS usable as /, requires NFSCL options MSDOSFS # MSDOS Filesystem options CD9660 # ISO 9660 Filesystem options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables +options GEOM_PART_BSD # BSD partition scheme +options GEOM_PART_MBR # MBR partition scheme options COMPAT_43 # Compatible with BSD 4.3 [KEEP THIS!] options SCSI_DELAY=5000 # Delay (in ms) before probing SCSI options KTRACE # ktrace(1) support options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # Posix P1003_1B real-time extensions +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions options KBD_INSTALL_CDEV # install a CDEV entry in /dev -options PREEMPTION -options FREEBSD_BOOT_LOADER -options VFP # vfp/neon -options SMP # Enable multiple cores +options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +options VFP # Enable floating point hardware support +options SMP # Enable multiple cores -# Debugging +# Debugging for use in -current makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options BREAK_TO_DEBUGGER -#options VERBOSE_SYSINIT # Enable verbose sysinit messages -options KDB +#options VERBOSE_SYSINIT # Enable verbose sysinit messages +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: options DDB # Enable the kernel debugger -#options INVARIANTS # Enable calls of extra sanity checking -#options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS +#options INVARIANTS # Enable calls of extra sanity checking +#options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS options WITNESS # Enable checks to detect deadlocks and cycles options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed options DIAGNOSTIC -# NFS support -#options NFSCL -#options NFSSERVER # Network Filesystem Server -#options NFSCLIENT # Network Filesystem Client +# Boot device is 2nd slice on USB +options ROOTDEVNAME=\"ufs:/dev/da0s2\" -# MMC/SD/SDIO card slot support +# MMC/SD/SDIO Card slot support #device mmc # mmc/sd bus #device mmcsd # mmc/sd flash cards -# Boot device is 2nd slice on USB -options ROOTDEVNAME=\"ufs:/dev/da0s2\" - # Console and misc device uart device uart_ns8250 @@ -96,8 +102,8 @@ device pass options USB_HOST_ALIGN=32 # Align usb buffers to cache line size. device usb options USB_DEBUG -#options USB_REQ_DEBUG -#options USB_VERBOSE +#options USB_REQ_DEBUG +#options USB_VERBOSE device dwcotg # DWC OTG controller device umass @@ -109,10 +115,10 @@ device mii device bpf # Wireless NIC cards -options IEEE80211_DEBUG -options IEEE80211_AMPDU_AGE -options IEEE80211_SUPPORT_MESH -options IEEE80211_SUPPORT_TDMA +options IEEE80211_DEBUG +options IEEE80211_AMPDU_AGE +options IEEE80211_SUPPORT_MESH +options IEEE80211_SUPPORT_TDMA device wlan # 802.11 support device wlan_wep # 802.11 WEP support device wlan_ccmp # 802.11 CCMP support @@ -121,8 +127,9 @@ device urtwn device urtwnfw device firmware # Used by the above -# USB ethernet support, requires miibus +# USB Ethernet support, requires miibus device miibus device udav -options FDT # Configure using FDT/DTB data. +# Flattened Device Tree +options FDT # Configure using FDT/DTB data diff --git a/sys/arm/conf/RPI-B b/sys/arm/conf/RPI-B index 322a4ec..c1013d1 100644 --- a/sys/arm/conf/RPI-B +++ b/sys/arm/conf/RPI-B @@ -1,7 +1,8 @@ +# # RPI-B -- Custom configuration for the Raspberry Pi # -# For more information on this file, please read the handbook section on -# Kernel Configuration Files: +# For more information on this file, please read the config(5) manual page, +# and/or the handbook section on Kernel Configuration Files: # # http://www.FreeBSD.org/doc/en_US.ISO8859-1/books/handbook/kernelconfig-config.html # @@ -11,8 +12,8 @@ # latest information. # # An exhaustive list of options and more detailed explanations of the -# device lines is also present in the ../../conf/NOTES and NOTES files. -# If you are in doubt as to the purpose or necessity of a line, check first +# device lines is also present in the ../../conf/NOTES and NOTES files. +# If you are in doubt as to the purpose or necessity of a line, check first # in NOTES. # # $FreeBSD$ @@ -21,10 +22,9 @@ ident RPI-B include "../broadcom/bcm2835/std.rpi" -makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options HZ=100 - options SCHED_4BSD # 4BSD scheduler +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols options SCTP # Stream Control Transmission Protocol @@ -32,42 +32,62 @@ options FFS # Berkeley Fast Filesystem options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories -options GEOM_PART_BSD # BSD partition scheme -options GEOM_PART_MBR # MBR partition scheme -options TMPFS # Efficient memory filesystem +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client +options NFSLOCKD # Network Lock Manager +options NFS_ROOT # NFS usable as /, requires NFSCL options MSDOSFS # MSDOS Filesystem -device snp - -options NFSCL # Network Filesystem Client - -#options NFS_ROOT # NFS usable as /, requires NFSCLIENT -#options BOOTP_NFSROOT -#options BOOTP_COMPAT -#options BOOTP -#options BOOTP_NFSV3 -#options BOOTP_WIRED_TO=ue0 - +options CD9660 # ISO 9660 Filesystem +options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables +options GEOM_PART_BSD # BSD partition scheme +options GEOM_PART_MBR # MBR partition scheme options COMPAT_43 # Compatible with BSD 4.3 [KEEP THIS!] options SCSI_DELAY=5000 # Delay (in ms) before probing SCSI options KTRACE # ktrace(1) support options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # Posix P1003_1B real-time extensions +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions options KBD_INSTALL_CDEV # install a CDEV entry in /dev -#options ROOTDEVNAME=\"ufs:mmcsd0s2\" - -options PREEMPTION options PLATFORM +options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +options VFP # Enable floating point hardware support + +# Debugging for use in -current +makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols +options BREAK_TO_DEBUGGER +#options VERBOSE_SYSINIT # Enable verbose sysinit messages +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: +options DDB # Enable the kernel debugger +options INVARIANTS # Enable calls of extra sanity checking +options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS +#options WITNESS # Enable checks to detect deadlocks and cycles +#options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed +#options DIAGNOSTIC + +# NFS root from boopt/dhcp +#options BOOTP +#options BOOTP_NFSROOT +#options BOOTP_COMPAT +#options BOOTP_NFSV3 +#options BOOTP_WIRED_TO=ue0 + +#options ROOTDEVNAME=\"ufs:mmcsd0s2\" device bpf device loop device ether device uart -device pl011 - device pty +device snp +device pl011 # Comment following lines for boot console on serial port device vt @@ -86,11 +106,6 @@ device iic device iicbus device bcm2835_bsc -options KDB -options DDB # Enable the kernel debugger -options INVARIANTS # Enable calls of extra sanity checking -options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS - device md device random # Entropy device @@ -114,10 +129,8 @@ device spibus device bcm2835_spi # Flattened Device Tree -options FDT +options FDT # Configure using FDT/DTB data # Note: DTB is normally loaded and modified by RPi boot loader, then # handed to kernel via U-Boot and ubldr. -#options FDT_DTB_STATIC -makeoptions FDT_DTS_FILE=rpi.dts - -options VFP # vfp/neon +#options FDT_DTB_STATIC +makeoptions FDT_DTS_FILE=rpi.dts diff --git a/sys/arm/conf/SAM9260EK b/sys/arm/conf/SAM9260EK index c3cba33..dd0d52e 100644 --- a/sys/arm/conf/SAM9260EK +++ b/sys/arm/conf/SAM9260EK @@ -180,5 +180,5 @@ device nand # NAND interface on CS3 #makeoptions FDT_DTS_FILE=sam9260ek.dts options EARLY_PRINTF -options SOCDEV_PA=0xfc000000 +options SOCDEV_PA=0xfc000000 options SOCDEV_VA=0xdc000000 diff --git a/sys/arm/conf/SOCKIT b/sys/arm/conf/SOCKIT index 5a9ade5..744559a 100644 --- a/sys/arm/conf/SOCKIT +++ b/sys/arm/conf/SOCKIT @@ -1,3 +1,4 @@ +# # Kernel configuration for Terasic SoCKit (Altera Cyclone V SoC). # # For more information on this file, please read the config(5) manual page, @@ -22,44 +23,51 @@ include "../altera/socfpga/std.socfpga" makeoptions MODULES_OVERRIDE="" -makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols makeoptions WERROR="-Werror" options HZ=100 options SCHED_4BSD # 4BSD scheduler +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols -options GEOM_PART_BSD # BSD partition scheme -options GEOM_PART_MBR # MBR partition scheme -options GEOM_PART_GPT # GUID partition tables -options TMPFS # Efficient memory filesystem +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem -options SOFTUPDATES +options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client +options NFSLOCKD # Network Lock Manager +options NFS_ROOT # NFS usable as /, requires NFSCL options MSDOSFS # MSDOS Filesystem options CD9660 # ISO 9660 Filesystem options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables +options GEOM_PART_BSD # BSD partition scheme +options GEOM_PART_MBR # MBR partition scheme options COMPAT_43 # Compatible with BSD 4.3 [KEEP THIS!] options SCSI_DELAY=5000 # Delay (in ms) before probing SCSI -options KTRACE +options KTRACE # ktrace(1) support options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # Posix P1003_1B real-time extensions -options KBD_INSTALL_CDEV -options PREEMPTION -options FREEBSD_BOOT_LOADER -options VFP # vfp/neon +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions +options KBD_INSTALL_CDEV # install a CDEV entry in /dev +options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +options VFP # Enable floating point hardware support +options SMP # Enable multiple cores -options SMP - -# Debugging +# Debugging for use in -current makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options BREAK_TO_DEBUGGER #options VERBOSE_SYSINIT # Enable verbose sysinit messages -options KDB +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: options DDB # Enable the kernel debugger options INVARIANTS # Enable calls of extra sanity checking options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS @@ -67,25 +75,20 @@ options INVARIANT_SUPPORT # Extra sanity checks of internal structures, require #options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed #options DIAGNOSTIC -# NFS support -options NFSCL # Network Filesystem Client -options NFSLOCKD # Network Lock Manager -options NFS_ROOT # NFS usable as /, requires NFSCLIENT - -# Uncomment this for NFS root -#options NFS_ROOT # NFS usable as /, requires NFSCL +# NFS root from boopt/dhcp +#options BOOTP #options BOOTP_NFSROOT #options BOOTP_COMPAT -#options BOOTP #options BOOTP_NFSV3 #options BOOTP_WIRED_TO=ue0 +options ROOTDEVNAME=\"ufs:/dev/da0\" + +# MMC/SD/SDIO Card slot support device mmc # mmc/sd bus device mmcsd # mmc/sd flash cards device dwmmc -options ROOTDEVNAME=\"ufs:/dev/da0\" - # Pseudo devices device loop @@ -132,7 +135,7 @@ device miibus device axe # ASIX Electronics USB Ethernet device bpf # Berkeley packet filter -#FDT -options FDT +# Flattened Device Tree +options FDT # Configure using FDT/DTB data options FDT_DTB_STATIC makeoptions FDT_DTS_FILE=socfpga-sockit.dts diff --git a/sys/arm/conf/SOCKIT-BERI b/sys/arm/conf/SOCKIT-BERI index 76306de..918da53 100644 --- a/sys/arm/conf/SOCKIT-BERI +++ b/sys/arm/conf/SOCKIT-BERI @@ -1,3 +1,4 @@ +# # Kernel configuration for Terasic SoCKit (Altera Cyclone V SoC). # # For more information on this file, please read the config(5) manual page, @@ -22,44 +23,51 @@ include "../altera/socfpga/std.socfpga" makeoptions MODULES_OVERRIDE="" -makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols makeoptions WERROR="-Werror" options HZ=100 options SCHED_4BSD # 4BSD scheduler +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols -options GEOM_PART_BSD # BSD partition scheme -options GEOM_PART_MBR # MBR partition scheme -options GEOM_PART_GPT # GUID partition tables -options TMPFS # Efficient memory filesystem +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem -options SOFTUPDATES +options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client +options NFSLOCKD # Network Lock Manager +options NFS_ROOT # NFS usable as /, requires NFSCL options MSDOSFS # MSDOS Filesystem options CD9660 # ISO 9660 Filesystem options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables +options GEOM_PART_BSD # BSD partition scheme +options GEOM_PART_MBR # MBR partition scheme options COMPAT_43 # Compatible with BSD 4.3 [KEEP THIS!] options SCSI_DELAY=5000 # Delay (in ms) before probing SCSI -options KTRACE +options KTRACE # ktrace(1) support options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # Posix P1003_1B real-time extensions -options KBD_INSTALL_CDEV -options PREEMPTION -options FREEBSD_BOOT_LOADER -options VFP # vfp/neon +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions +options KBD_INSTALL_CDEV # install a CDEV entry in /dev +options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +options VFP # Enable floating point hardware support +options SMP # Enable multiple cores -options SMP - -# Debugging +# Debugging for use in -current makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options BREAK_TO_DEBUGGER #options VERBOSE_SYSINIT # Enable verbose sysinit messages -options KDB +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: options DDB # Enable the kernel debugger options INVARIANTS # Enable calls of extra sanity checking options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS @@ -67,25 +75,20 @@ options INVARIANT_SUPPORT # Extra sanity checks of internal structures, require #options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed #options DIAGNOSTIC -# NFS support -options NFSCL # Network Filesystem Client -options NFSLOCKD # Network Lock Manager -options NFS_ROOT # NFS usable as /, requires NFSCLIENT - -# Uncomment this for NFS root -#options NFS_ROOT # NFS usable as /, requires NFSCL +# NFS root from boopt/dhcp +#options BOOTP #options BOOTP_NFSROOT #options BOOTP_COMPAT -#options BOOTP #options BOOTP_NFSV3 #options BOOTP_WIRED_TO=ue0 +options ROOTDEVNAME=\"ufs:/dev/mmcsd0s4\" + +# MMC/SD/SDIO Card slot support device mmc # mmc/sd bus device mmcsd # mmc/sd flash cards device dwmmc -options ROOTDEVNAME=\"ufs:/dev/mmcsd0s4\" - # Pseudo devices device loop @@ -139,7 +142,7 @@ device miibus device axe # ASIX Electronics USB Ethernet device bpf # Berkeley packet filter -#FDT -options FDT +# Flattened Device Tree +options FDT # Configure using FDT/DTB data options FDT_DTB_STATIC makeoptions FDT_DTS_FILE=socfpga-sockit-beri.dts diff --git a/sys/arm/conf/VERSATILEPB b/sys/arm/conf/VERSATILEPB index fb2087a..2132389 100644 --- a/sys/arm/conf/VERSATILEPB +++ b/sys/arm/conf/VERSATILEPB @@ -1,7 +1,8 @@ +# # VERSATILEPB - Configuration for QEMU version of Versatile Platform Board # -# For more information on this file, please read the handbook section on -# Kernel Configuration Files: +# For more information on this file, please read the config(5) manual page, +# and/or the handbook section on Kernel Configuration Files: # # http://www.FreeBSD.org/doc/en_US.ISO8859-1/books/handbook/kernelconfig-config.html # @@ -11,8 +12,8 @@ # latest information. # # An exhaustive list of options and more detailed explanations of the -# device lines is also present in the ../../conf/NOTES and NOTES files. -# If you are in doubt as to the purpose or necessity of a line, check first +# device lines is also present in the ../../conf/NOTES and NOTES files. +# If you are in doubt as to the purpose or necessity of a line, check first # in NOTES. # # $FreeBSD$ @@ -29,36 +30,53 @@ makeoptions KERNVIRTADDR=0xc0100000 options KERNPHYSADDR=0x00100000 makeoptions KERNPHYSADDR=0x00100000 options PHYSADDR=0x00000000 -options FREEBSD_BOOT_LOADER -options LINUX_BOOT_ABI -makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options HZ=100 - options SCHED_4BSD # 4BSD scheduler +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking +options INET6 # IPv6 communications protocols +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories -device snp - +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client +options NFSLOCKD # Network Lock Manager +options NFS_ROOT # NFS usable as /, requires NFSCL +options MSDOSFS # MSDOS Filesystem +options CD9660 # ISO 9660 Filesystem +options PROCFS # Process filesystem (requires PSEUDOFS) +options PSEUDOFS # Pseudo-filesystem framework +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables options GEOM_PART_BSD # BSD partition scheme options GEOM_PART_MBR # MBR partition scheme -options TMPFS # Efficient memory filesystem -options PSEUDOFS # Pseudo-filesystem framework options COMPAT_43 # Compatible with BSD 4.3 [KEEP THIS!] options SCSI_DELAY=5000 # Delay (in ms) before probing SCSI options KTRACE # ktrace(1) support options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # Posix P1003_1B real-time extensions +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions options KBD_INSTALL_CDEV # install a CDEV entry in /dev -options ROOTDEVNAME=\"ufs:da0s1a\" -options VFP # vfp/neon +options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +options LINUX_BOOT_ABI # Process metadata passed from Linux boot loaders +options VFP # Enable floating point hardware support -options PREEMPTION +# Debugging for use in -current +makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: +options DDB # Enable the kernel debugger +options INVARIANTS # Enable calls of extra sanity checking +options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS + +options ROOTDEVNAME=\"ufs:da0s1a\" device bpf device loop @@ -72,6 +90,7 @@ device pl011 device pl190 device pty +device snp device pci @@ -87,18 +106,13 @@ device pass # Passthrough device (direct ATA/SCSI access) # Comment following lines for headless setup device sc device kbdmux -options SC_DFLT_FONT # compile font in -makeoptions SC_DFLT_FONT=cp437 - -options KDB -options DDB # Enable the kernel debugger -options INVARIANTS # Enable calls of extra sanity checking -options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS +options SC_DFLT_FONT # compile font in +makeoptions SC_DFLT_FONT=cp437 device md device random # Entropy device # Flattened Device Tree -options FDT -options FDT_DTB_STATIC -makeoptions FDT_DTS_FILE=versatilepb.dts +options FDT # Configure using FDT/DTB data +options FDT_DTB_STATIC +makeoptions FDT_DTS_FILE=versatilepb.dts diff --git a/sys/arm/conf/VYBRID b/sys/arm/conf/VYBRID index e2e1def..7ff206e 100644 --- a/sys/arm/conf/VYBRID +++ b/sys/arm/conf/VYBRID @@ -1,3 +1,4 @@ +# # Kernel configuration for Vybrid Family boards. # # For more information on this file, please read the config(5) manual page, @@ -23,75 +24,79 @@ include "../freescale/vybrid/std.vybrid" makeoptions MODULES_OVERRIDE="" makeoptions WITHOUT_MODULES="ahc" -makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols makeoptions WERROR="-Werror" options HZ=100 options SCHED_4BSD # 4BSD scheduler +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem -options SOFTUPDATES +options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories -options GEOM_PART_BSD # BSD partition scheme -options GEOM_PART_MBR # MBR partition scheme -options TMPFS # Efficient memory filesystem +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client +options NFSLOCKD # Network Lock Manager +options NFS_ROOT # NFS usable as /, requires NFSCL options MSDOSFS # MSDOS Filesystem options CD9660 # ISO 9660 Filesystem +#options NANDFS # NAND Filesystem options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework -#options NANDFS # NAND Filesystem +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables +options GEOM_PART_BSD # BSD partition scheme +options GEOM_PART_MBR # MBR partition scheme options COMPAT_43 # Compatible with BSD 4.3 [KEEP THIS!] options SCSI_DELAY=5000 # Delay (in ms) before probing SCSI -options KTRACE +options KTRACE # ktrace(1) support options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # Posix P1003_1B real-time extensions -options KBD_INSTALL_CDEV -options PREEMPTION -options FREEBSD_BOOT_LOADER -options MUTEX_NOINLINE -options RWLOCK_NOINLINE -options NO_FFS_SNAPSHOT -options NO_SWAPPING -options VFP # vfp/neon +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions +options KBD_INSTALL_CDEV # install a CDEV entry in /dev +options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +options VFP # Enable floating point hardware support +#options SMP # Enable multiple cores -# Debugging +# Debugging for use in -current makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols options BREAK_TO_DEBUGGER -#options VERBOSE_SYSINIT # Enable verbose sysinit messages -options KDB +#options VERBOSE_SYSINIT # Enable verbose sysinit messages +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: options DDB # Enable the kernel debugger -#options INVARIANTS # Enable calls of extra sanity checking -#options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS -#options WITNESS # Enable checks to detect deadlocks and cycles -#options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed -#options DIAGNOSTIC - -# NFS support -options NFSCL # Network Filesystem Client -options NFSLOCKD # Network Lock Manager -options NFS_ROOT # NFS usable as /, requires NFSCLIENT - -# Uncomment this for NFS root -#options NFS_ROOT # NFS usable as /, requires NFSCL -#options BOOTP_NFSROOT -#options BOOTP_COMPAT -#options BOOTP -#options BOOTP_NFSV3 -#options BOOTP_WIRED_TO=ffec0 - -device mmc # mmc/sd bus -device mmcsd # mmc/sd flash cards -device sdhci # generic sdhci +#options INVARIANTS # Enable calls of extra sanity checking +#options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS +#options WITNESS # Enable checks to detect deadlocks and cycles +#options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed +#options DIAGNOSTIC + +# NFS root from boopt/dhcp +#options BOOTP +#options BOOTP_NFSROOT +#options BOOTP_COMPAT +#options BOOTP_NFSV3 +#options BOOTP_WIRED_TO=ffec0 #options ROOTDEVNAME=\"nfs:10.5.0.1:/tftpboot/cosmic\" #options ROOTDEVNAME=\"nandfs:/dev/gnand0s.root\" options ROOTDEVNAME=\"ufs:/dev/da0\" -#options SMP +options MUTEX_NOINLINE +options RWLOCK_NOINLINE +options NO_FFS_SNAPSHOT +options NO_SWAPPING + +# MMC/SD/SDIO Card slot support +device mmc # mmc/sd bus +device mmcsd # mmc/sd flash cards +device sdhci # generic sdhci # Pseudo devices @@ -105,8 +110,8 @@ device gpio options USB_HOST_ALIGN=32 # Align usb buffers to cache line size. device usb options USB_DEBUG -#options USB_REQ_DEBUG -#options USB_VERBOSE +#options USB_REQ_DEBUG +#options USB_VERBOSE #device musb device ehci #device ohci @@ -150,4 +155,5 @@ device vt device kbdmux device ukbd -options FDT +# Flattened Device Tree +options FDT # Configure using FDT/DTB data diff --git a/sys/arm/conf/WANDBOARD-DUAL b/sys/arm/conf/WANDBOARD-DUAL index 1e690c9..66e2535 100644 --- a/sys/arm/conf/WANDBOARD-DUAL +++ b/sys/arm/conf/WANDBOARD-DUAL @@ -23,6 +23,6 @@ include "IMX6" ident WANDBOARD-DUAL # Flattened Device Tree -options FDT_DTB_STATIC -makeoptions FDT_DTS_FILE=wandboard-dual.dts +options FDT_DTB_STATIC +makeoptions FDT_DTS_FILE=wandboard-dual.dts diff --git a/sys/arm/conf/WANDBOARD-QUAD b/sys/arm/conf/WANDBOARD-QUAD index 121e712..830a445 100644 --- a/sys/arm/conf/WANDBOARD-QUAD +++ b/sys/arm/conf/WANDBOARD-QUAD @@ -23,6 +23,6 @@ include "IMX6" ident WANDBOARD-QUAD # Flattened Device Tree -options FDT_DTB_STATIC -makeoptions FDT_DTS_FILE=wandboard-quad.dts +options FDT_DTB_STATIC +makeoptions FDT_DTS_FILE=wandboard-quad.dts diff --git a/sys/arm/conf/WANDBOARD-SOLO b/sys/arm/conf/WANDBOARD-SOLO index 424bc5f..39a0c7c 100644 --- a/sys/arm/conf/WANDBOARD-SOLO +++ b/sys/arm/conf/WANDBOARD-SOLO @@ -23,6 +23,6 @@ include "IMX6" ident WANDBOARD-SOLO # Flattened Device Tree -options FDT_DTB_STATIC -makeoptions FDT_DTS_FILE=wandboard-solo.dts +options FDT_DTB_STATIC +makeoptions FDT_DTS_FILE=wandboard-solo.dts diff --git a/sys/arm/conf/ZEDBOARD b/sys/arm/conf/ZEDBOARD index 8ec0df1..e1200bf 100644 --- a/sys/arm/conf/ZEDBOARD +++ b/sys/arm/conf/ZEDBOARD @@ -1,73 +1,86 @@ +# # ZEDBOARD -- Custom configuration for the Xilinx Zynq-7000 based # ZedBoard (www.zedboard.org) -# -# For more information on this file, please read the handbook section on -# Kernel Configuration Files: -# +# +# For more information on this file, please read the config(5) manual page, +# and/or the handbook section on Kernel Configuration Files: +# # http://www.FreeBSD.org/doc/en_US.ISO8859-1/books/handbook/kernelconfig-config.html -# +# # The handbook is also available locally in /usr/share/doc/handbook # if you've installed the doc distribution, otherwise always see the # FreeBSD World Wide Web server (http://www.FreeBSD.org/) for the # latest information. -# +# # An exhaustive list of options and more detailed explanations of the -# device lines is also present in the ../../conf/NOTES and NOTES files. -# If you are in doubt as to the purpose or necessity of a line, check first +# device lines is also present in the ../../conf/NOTES and NOTES files. +# If you are in doubt as to the purpose or necessity of a line, check first # in NOTES. -# +# # $FreeBSD$ ident ZEDBOARD -include "../xilinx/zedboard/std.zedboard" +include "../xilinx/zedboard/std.zedboard" makeoptions MODULES_OVERRIDE="" makeoptions WITHOUT_MODULES="ahc" options SCHED_4BSD # 4BSD scheduler +options PREEMPTION # Enable kernel thread preemption options INET # InterNETworking options INET6 # IPv6 communications protocols +options SCTP # Stream Control Transmission Protocol options FFS # Berkeley Fast Filesystem options SOFTUPDATES # Enable FFS soft updates support options UFS_ACL # Support for access control lists options UFS_DIRHASH # Improve performance on big directories -# options ROOTDEVNAME=\"ufs:mmcsd0s2a\" - -options NFSCL # Network Filesystem Client -# options NFSSD # Network Filesystem Server -# options NFSLOCKD # Network Lock Manager -# options NFS_ROOT # NFS usable as /, requires NFSCL -# options BOOTP_NFSROOT -# options BOOTP - -options GEOM_PART_BSD # BSD partition scheme -options GEOM_PART_MBR # MBR partition scheme -options TMPFS # Efficient memory filesystem +options UFS_GJOURNAL # Enable gjournal-based UFS journaling +options QUOTA # Enable disk quotas for UFS +options NFSCL # New Network Filesystem Client +#options NFSSD # Network Filesystem Server +options NFSLOCKD # Network Lock Manager +options NFS_ROOT # NFS usable as /, requires NFSCL options MSDOSFS # MSDOS Filesystem options CD9660 # ISO 9660 Filesystem options PROCFS # Process filesystem (requires PSEUDOFS) options PSEUDOFS # Pseudo-filesystem framework +options TMPFS # Efficient memory filesystem +options GEOM_PART_GPT # GUID Partition Tables +options GEOM_PART_BSD # BSD partition scheme +options GEOM_PART_MBR # MBR partition scheme options SCSI_DELAY=5000 # Delay (in ms) before probing SCSI options KTRACE # ktrace(1) support options SYSVSHM # SYSV-style shared memory options SYSVMSG # SYSV-style message queues options SYSVSEM # SYSV-style semaphores -options _KPOSIX_PRIORITY_SCHEDULING # Posix P1003_1B real-time extensions -options FREEBSD_BOOT_LOADER -options VFP # vfp/neon -options SMP # Symmetric MultiProcessor Kernel +options _KPOSIX_PRIORITY_SCHEDULING # POSIX P1003_1B real-time extensions +options KBD_INSTALL_CDEV # install a CDEV entry in /dev +options FREEBSD_BOOT_LOADER # Process metadata passed from loader(8) +options VFP # Enable floating point hardware support +options SMP # Enable multiple cores -# Debugging +# Debugging for use in -current makeoptions DEBUG=-g # Build kernel with gdb(1) debug symbols -options DDB -options KDB -# options BREAK_TO_DEBUGGER +#options BREAK_TO_DEBUGGER +options KDB # Enable kernel debugger support +# For minimum debugger support (stable branch) use: +#options KDB_TRACE # Print a stack trace for a panic +# For full debugger support use this instead: +options DDB # Enable the kernel debugger +#options INVARIANTS # Enable calls of extra sanity checking +#options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS +#options WITNESS # Enable checks to detect deadlocks and cycles +#options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed +#options DIAGNOSTIC + +# NFS root from boopt/dhcp +#options BOOTP +#options BOOTP_NFSROOT +#options BOOTP_COMPAT +#options BOOTP_NFSV3 -# options INVARIANTS # Enable calls of extra sanity checking -# options INVARIANT_SUPPORT # Extra sanity checks of internal structures, required by INVARIANTS -# options WITNESS # Enable checks to detect deadlocks and cycles -# options WITNESS_SKIPSPIN # Don't run witness on spinlocks for speed +options ROOTDEVNAME=\"ufs:mmcsd0s2a\" device loop device random @@ -98,7 +111,7 @@ device axe # USB-Ethernet # Flattened Device Tree -options FDT -# options FDT_DTB_STATIC -# makeoptions FDT_DTS_FILE=zedboard.dts +options FDT # Configure using FDT/DTB data +#options FDT_DTB_STATIC +#makeoptions FDT_DTS_FILE=zedboard.dts diff --git a/sys/arm/freescale/imx/imx51_machdep.c b/sys/arm/freescale/imx/imx51_machdep.c index 8ab6541..5d8f380 100644 --- a/sys/arm/freescale/imx/imx51_machdep.c +++ b/sys/arm/freescale/imx/imx51_machdep.c @@ -39,36 +39,28 @@ __FBSDID("$FreeBSD$"); #include <machine/bus.h> #include <machine/devmap.h> #include <machine/machdep.h> -#include <machine/platform.h> +#include <machine/platformvar.h> #include <arm/freescale/imx/imx_machdep.h> -vm_offset_t -platform_lastaddr(void) +#include "platform_if.h" + +static vm_offset_t +imx51_lastaddr(platform_t plat) { return (arm_devmap_lastaddr()); } -void -platform_probe_and_attach(void) +static int +imx51_attach(platform_t plat) { /* XXX - Get rid of this stuff soon. */ boothowto |= RB_VERBOSE|RB_MULTIPLE; bootverbose = 1; -} - -void -platform_gpio_init(void) -{ - -} - -void -platform_late_init(void) -{ + return (0); } /* @@ -78,8 +70,8 @@ platform_late_init(void) * * Notably missing are entries for GPU, IPU, in general anything video related. */ -int -platform_devmap_init(void) +static int +imx51_devmap_init(platform_t plat) { arm_devmap_add_entry(0x70000000, 0x00100000); @@ -101,3 +93,12 @@ u_int imx_soc_type() return (IMXSOC_51); } +static platform_method_t imx51_methods[] = { + PLATFORMMETHOD(platform_attach, imx51_attach), + PLATFORMMETHOD(platform_devmap_init, imx51_devmap_init), + PLATFORMMETHOD(platform_lastaddr, imx51_lastaddr), + + PLATFORMMETHOD_END, +}; + +FDT_PLATFORM_DEF(imx51, "i.MX51", 0, "fsl,imx51"); diff --git a/sys/arm/freescale/imx/imx53_machdep.c b/sys/arm/freescale/imx/imx53_machdep.c index ebf09d9..92b59e4 100644 --- a/sys/arm/freescale/imx/imx53_machdep.c +++ b/sys/arm/freescale/imx/imx53_machdep.c @@ -39,36 +39,28 @@ __FBSDID("$FreeBSD$"); #include <machine/bus.h> #include <machine/devmap.h> #include <machine/machdep.h> -#include <machine/platform.h> +#include <machine/platformvar.h> #include <arm/freescale/imx/imx_machdep.h> -vm_offset_t -platform_lastaddr(void) +#include "platform_if.h" + +static vm_offset_t +imx53_lastaddr(platform_t plat) { return (arm_devmap_lastaddr()); } -void -platform_probe_and_attach(void) +static int +imx53_attach(platform_t plat) { /* XXX - Get rid of this stuff soon. */ boothowto |= RB_VERBOSE|RB_MULTIPLE; bootverbose = 1; -} - -void -platform_gpio_init(void) -{ - -} - -void -platform_late_init(void) -{ + return (0); } /* @@ -78,8 +70,8 @@ platform_late_init(void) * * Notably missing are entries for GPU, IPU, in general anything video related. */ -int -platform_devmap_init(void) +static int +imx53_devmap_init(platform_t plat) { arm_devmap_add_entry(0x50000000, 0x00100000); @@ -101,4 +93,13 @@ u_int imx_soc_type() return (IMXSOC_53); } +static platform_method_t imx53_methods[] = { + PLATFORMMETHOD(platform_attach, imx53_attach), + PLATFORMMETHOD(platform_devmap_init, imx53_devmap_init), + PLATFORMMETHOD(platform_lastaddr, imx53_lastaddr), + + PLATFORMMETHOD_END, +}; + +FDT_PLATFORM_DEF(imx53, "i.MX53", 0, "fsl,imx53"); diff --git a/sys/arm/freescale/imx/imx6_machdep.c b/sys/arm/freescale/imx/imx6_machdep.c index 510a09b..2322cff 100644 --- a/sys/arm/freescale/imx/imx6_machdep.c +++ b/sys/arm/freescale/imx/imx6_machdep.c @@ -40,7 +40,7 @@ __FBSDID("$FreeBSD$"); #include <machine/devmap.h> #include <machine/intr.h> #include <machine/machdep.h> -#include <machine/platform.h> +#include <machine/platformvar.h> #include <arm/arm/mpcore_timervar.h> #include <arm/freescale/imx/imx6_anatopreg.h> @@ -50,6 +50,8 @@ __FBSDID("$FreeBSD$"); #include <dev/fdt/fdt_common.h> #include <dev/ofw/openfirm.h> +#include "platform_if.h" + struct fdt_fixup_entry fdt_fixup_table[] = { { NULL, NULL } }; @@ -90,29 +92,25 @@ fdt_pic_decode_t fdt_pic_table[] = { NULL }; -vm_offset_t -platform_lastaddr(void) +static vm_offset_t +imx6_lastaddr(platform_t plat) { return (arm_devmap_lastaddr()); } -void -platform_probe_and_attach(void) +static int +imx6_attach(platform_t plat) { /* Inform the MPCore timer driver that its clock is variable. */ arm_tmr_change_frequency(ARM_TMR_FREQUENCY_VARIES); -} - -void -platform_gpio_init(void) -{ + return (0); } -void -platform_late_init(void) +static void +imx6_late_init(platform_t plat) { /* Cache the gpio1 node handle for imx6_decode_fdt() workaround code. */ @@ -136,8 +134,8 @@ platform_late_init(void) * static map some of that area. Be careful with other things in that area such * as OCRAM that probably shouldn't be mapped as PTE_DEVICE memory. */ -int -platform_devmap_init(void) +static int +imx6_devmap_init(platform_t plat) { const uint32_t IMX6_ARMMP_PHYS = 0x00a00000; const uint32_t IMX6_ARMMP_SIZE = 0x00100000; @@ -271,3 +269,15 @@ imx6_early_putc(int c) early_putc_t *early_putc = imx6_early_putc; #endif +static platform_method_t imx6_methods[] = { + PLATFORMMETHOD(platform_attach, imx6_attach), + PLATFORMMETHOD(platform_lastaddr, imx6_lastaddr), + PLATFORMMETHOD(platform_devmap_init, imx6_devmap_init), + PLATFORMMETHOD(platform_late_init, imx6_late_init), + + PLATFORMMETHOD_END, +}; + +FDT_PLATFORM_DEF2(imx6, imx6s, "i.MX6 Solo", 0, "fsl,imx6s"); +FDT_PLATFORM_DEF2(imx6, imx6d, "i.MX6 Dual", 0, "fsl,imx6d"); +FDT_PLATFORM_DEF2(imx6, imx6q, "i.MX6 Quad", 0, "fsl,imx6q"); diff --git a/sys/arm/freescale/imx/std.imx51 b/sys/arm/freescale/imx/std.imx51 index ce9d7d2..4ffcac7 100644 --- a/sys/arm/freescale/imx/std.imx51 +++ b/sys/arm/freescale/imx/std.imx51 @@ -1,6 +1,7 @@ # $FreeBSD$ machine arm armv6 cpu CPU_CORTEXA +makeoptions CONF_CFLAGS="-march=armv7a -Wa,-march=armv7a" makeoptions ARM_LITTLE_ENDIAN options ARM_L2_PIPT diff --git a/sys/arm/freescale/imx/std.imx53 b/sys/arm/freescale/imx/std.imx53 index 09bdb2d..6bc96cf 100644 --- a/sys/arm/freescale/imx/std.imx53 +++ b/sys/arm/freescale/imx/std.imx53 @@ -1,6 +1,7 @@ # $FreeBSD$ machine arm armv6 cpu CPU_CORTEXA +makeoptions CONF_CFLAGS="-march=armv7a -Wa,-march=armv7a" makeoptions ARM_LITTLE_ENDIAN options ARM_L2_PIPT diff --git a/sys/arm/freescale/imx/std.imx6 b/sys/arm/freescale/imx/std.imx6 index a559f17..4249f9e 100644 --- a/sys/arm/freescale/imx/std.imx6 +++ b/sys/arm/freescale/imx/std.imx6 @@ -1,6 +1,7 @@ # $FreeBSD$ machine arm armv6 cpu CPU_CORTEXA +makeoptions CONF_CFLAGS="-march=armv7a -Wa,-march=armv7a" makeoptions ARM_LITTLE_ENDIAN options ARM_L2_PIPT diff --git a/sys/arm/freescale/vybrid/std.vybrid b/sys/arm/freescale/vybrid/std.vybrid index 2fbd85c..35a2a25 100644 --- a/sys/arm/freescale/vybrid/std.vybrid +++ b/sys/arm/freescale/vybrid/std.vybrid @@ -4,6 +4,7 @@ makeoption ARM_LITTLE_ENDIAN cpu CPU_CORTEXA machine arm armv6 +makeoptions CONF_CFLAGS="-march=armv7a -Wa,-march=armv7a" options PHYSADDR=0x80000000 diff --git a/sys/arm/include/asm.h b/sys/arm/include/asm.h index 73b7355..9122e6e 100644 --- a/sys/arm/include/asm.h +++ b/sys/arm/include/asm.h @@ -39,6 +39,8 @@ #ifndef _MACHINE_ASM_H_ #define _MACHINE_ASM_H_ #include <sys/cdefs.h> +#include <machine/acle-compat.h> +#include <machine/sysreg.h> #define _C_LABEL(x) x #define _ASM_LABEL(x) x @@ -58,7 +60,24 @@ #endif /* + * gas/arm uses @ as a single comment character and thus cannot be used here. + * It recognises the # instead of an @ symbol in .type directives. + */ +#define _ASM_TYPE_FUNCTION #function +#define _ASM_TYPE_OBJECT #object + +/* XXX Is this still the right prologue for profiling? */ +#ifdef GPROF +#define _PROF_PROLOGUE \ + mov ip, lr; \ + bl __mcount +#else +#define _PROF_PROLOGUE +#endif + +/* * EENTRY()/EEND() mark "extra" entry/exit points from a function. + * LEENTRY()/LEEND() are the the same for local symbols. * The unwind info cannot handle the concept of a nested function, or a function * with multiple .fnstart directives, but some of our assembler code is written * with multiple labels to allow entry at several points. The EENTRY() macro @@ -66,41 +85,36 @@ * basically just a label that you can jump to. The EEND() macro does nothing * at all, except document the exit point associated with the same-named entry. */ -#define _EENTRY(x) .globl x; .type x,_ASM_TYPE_FUNCTION; x: -#define _EEND(x) /* nothing */ +#define GLOBAL(x) .global x -/* - * gas/arm uses @ as a single comment character and thus cannot be used here - * Instead it recognised the # instead of an @ symbols in .type directives - * We define a couple of macros so that assembly code will not be dependent - * on one or the other. - */ -#define _ASM_TYPE_FUNCTION #function -#define _ASM_TYPE_OBJECT #object -#define GLOBAL(X) .globl x -#define _ENTRY(x) \ - .text; _ALIGN_TEXT; _EENTRY(x) _FNSTART -#define _END(x) .size x, . - x; _FNEND +#define _LEENTRY(x) .type x,_ASM_TYPE_FUNCTION; x: +#define _LEEND(x) /* nothing */ +#define _EENTRY(x) GLOBAL(x); _LEENTRY(x) +#define _EEND(x) _LEEND(x) -#ifdef GPROF -# define _PROF_PROLOGUE \ - mov ip, lr; bl __mcount -#else -# define _PROF_PROLOGUE -#endif +#define _LENTRY(x) .text; _ALIGN_TEXT; _LEENTRY(x); _FNSTART +#define _LEND(x) .size x, . - x; _FNEND +#define _ENTRY(x) .text; _ALIGN_TEXT; _EENTRY(x); _FNSTART +#define _END(x) _LEND(x) #define ENTRY(y) _ENTRY(_C_LABEL(y)); _PROF_PROLOGUE -#define EENTRY(y) _EENTRY(_C_LABEL(y)); _PROF_PROLOGUE +#define EENTRY(y) _EENTRY(_C_LABEL(y)); #define ENTRY_NP(y) _ENTRY(_C_LABEL(y)) #define EENTRY_NP(y) _EENTRY(_C_LABEL(y)) #define END(y) _END(_C_LABEL(y)) -#define EEND(y) +#define EEND(y) _EEND(_C_LABEL(y)) #define ASENTRY(y) _ENTRY(_ASM_LABEL(y)); _PROF_PROLOGUE -#define ASEENTRY(y) _EENTRY(_ASM_LABEL(y)); _PROF_PROLOGUE +#define ASLENTRY(y) _LENTRY(_ASM_LABEL(y)); _PROF_PROLOGUE +#define ASEENTRY(y) _EENTRY(_ASM_LABEL(y)); +#define ASLEENTRY(y) _LEENTRY(_ASM_LABEL(y)); #define ASENTRY_NP(y) _ENTRY(_ASM_LABEL(y)) +#define ASLENTRY_NP(y) _LENTRY(_ASM_LABEL(y)) #define ASEENTRY_NP(y) _EENTRY(_ASM_LABEL(y)) +#define ASLEENTRY_NP(y) _LEENTRY(_ASM_LABEL(y)) #define ASEND(y) _END(_ASM_LABEL(y)) -#define ASEEND(y) +#define ASLEND(y) _LEND(_ASM_LABEL(y)) +#define ASEEND(y) _EEND(_ASM_LABEL(y)) +#define ASLEEND(y) _LEEND(_ASM_LABEL(y)) #define ASMSTR .asciz @@ -209,4 +223,18 @@ # define RETc(c) mov##c pc, lr #endif +#if __ARM_ARCH >= 7 +#define ISB isb +#define DSB dsb +#define DMB dmb +#elif __ARM_ARCH == 6 +#define ISB mcr CP15_CP15ISB +#define DSB mcr CP15_CP15DSB +#define DMB mcr CP15_CP15DMB +#else +#define ISB mcr CP15_CP15ISB +#define DSB mcr CP15_CP15DSB /* DSB and DMB are the */ +#define DMB mcr CP15_CP15DSB /* same prior to v6.*/ +#endif + #endif /* !_MACHINE_ASM_H_ */ diff --git a/sys/arm/include/cpuconf.h b/sys/arm/include/cpuconf.h index da35e57..fb02153 100644 --- a/sys/arm/include/cpuconf.h +++ b/sys/arm/include/cpuconf.h @@ -99,6 +99,29 @@ #endif #define ARM_NARCH (ARM_ARCH_4 + ARM_ARCH_5 + ARM_ARCH_6 | ARM_ARCH_7A) + +/* + * Compatibility for userland builds that have no CPUTYPE defined. Use the ARCH + * constants predefined by the compiler to define our old-school arch constants. + * This is a stopgap measure to tide us over until the conversion of all code + * to the newer ACLE constants defined by ARM (see acle-compat.h). + */ +#if ARM_NARCH == 0 +#if defined(__ARM_ARCH_4T__) +#undef ARM_ARCH_4 +#undef ARM_NARCH +#define ARM_ARCH_4 1 +#define ARM_NARCH 1 +#define CPU_ARM9 1 +#elif defined(__ARM_ARCH_6ZK__) +#undef ARM_ARCH_6 +#undef ARM_NARCH +#define ARM_ARCH_6 1 +#define ARM_NARCH 1 +#define CPU_ARM1176 1 +#endif +#endif + #if ARM_NARCH == 0 && !defined(KLD_MODULE) && defined(_KERNEL) #error ARM_NARCH is 0 #endif diff --git a/sys/arm/include/db_machdep.h b/sys/arm/include/db_machdep.h index da3b30e..741cae9 100644 --- a/sys/arm/include/db_machdep.h +++ b/sys/arm/include/db_machdep.h @@ -38,7 +38,7 @@ typedef vm_offset_t db_addr_t; typedef int db_expr_t; -#define PC_REGS() ((db_addr_t)kdb_thrctx->un_32.pcb32_pc) +#define PC_REGS() ((db_addr_t)kdb_thrctx->pcb_regs.sf_pc) #define BKPT_INST (KERNEL_BREAKPOINT) #define BKPT_SIZE (INSN_SIZE) diff --git a/sys/arm/include/frame.h b/sys/arm/include/frame.h index 1e6f43a..7655d89 100644 --- a/sys/arm/include/frame.h +++ b/sys/arm/include/frame.h @@ -86,57 +86,16 @@ struct trapframe { #define tf_r13 tf_usr_sp #define tf_r14 tf_usr_lr #define tf_r15 tf_pc -/* - * * Scheduler activations upcall frame. Pushed onto user stack before - * * calling an SA upcall. - * */ - -struct saframe { -#if 0 /* in registers on entry to upcall */ - int sa_type; - struct sa_t ** sa_sas; - int sa_events; - int sa_interrupted; -#endif - void * sa_arg; -}; /* - * * Signal frame. Pushed onto user stack before calling sigcode. - * */ - -/* the pointers are use in the trampoline code to locate the ucontext */ + * Signal frame. Pushed onto user stack before calling sigcode. + * The pointers are used in the trampoline code to locate the ucontext. + */ struct sigframe { - siginfo_t sf_si; /* actual saved siginfo */ + siginfo_t sf_si; /* actual saved siginfo */ ucontext_t sf_uc; /* actual saved ucontext */ }; -/* - * System stack frames. - */ - - -typedef struct irqframe { - unsigned int if_spsr; - unsigned int if_r0; - unsigned int if_r1; - unsigned int if_r2; - unsigned int if_r3; - unsigned int if_r4; - unsigned int if_r5; - unsigned int if_r6; - unsigned int if_r7; - unsigned int if_r8; - unsigned int if_r9; - unsigned int if_r10; - unsigned int if_r11; - unsigned int if_r12; - unsigned int if_usr_sp; - unsigned int if_usr_lr; - unsigned int if_svc_sp; - unsigned int if_svc_lr; - unsigned int if_pc; -} irqframe_t; /* * Switch frame. @@ -144,16 +103,23 @@ typedef struct irqframe { * It is important this is a multiple of 8 bytes so the stack is correctly * aligned when we create new threads. */ - -struct switchframe { - u_int pad; /* Used to pad the struct to a multiple of 8-bytes */ - u_int sf_r4; - u_int sf_r5; - u_int sf_r6; - u_int sf_r7; - u_int sf_pc; +struct switchframe +{ + register_t sf_r4; + register_t sf_r5; + register_t sf_r6; + register_t sf_r7; + register_t sf_r8; + register_t sf_r9; + register_t sf_r10; + register_t sf_r11; + register_t sf_r12; + register_t sf_sp; + register_t sf_lr; + register_t sf_pc; }; + /* * Stack frame. Used during stack traces (db_trace.c) */ diff --git a/sys/arm/include/intr.h b/sys/arm/include/intr.h index 3509772..9ee86f3 100644 --- a/sys/arm/include/intr.h +++ b/sys/arm/include/intr.h @@ -39,6 +39,10 @@ #ifndef _MACHINE_INTR_H_ #define _MACHINE_INTR_H_ +#ifdef FDT +#include <dev/ofw/openfirm.h> +#endif + /* XXX move to std.* files? */ #ifdef CPU_XSCALE_81342 #define NIRQ 128 @@ -85,4 +89,8 @@ void gic_init_secondary(void); int gic_decode_fdt(uint32_t iparentnode, uint32_t *intrcells, int *interrupt, int *trig, int *pol); +#ifdef FDT +int arm_fdt_map_irq(phandle_t, pcell_t *, int); +#endif + #endif /* _MACHINE_INTR_H */ diff --git a/sys/arm/include/machdep.h b/sys/arm/include/machdep.h index 907270e..ee524a8 100644 --- a/sys/arm/include/machdep.h +++ b/sys/arm/include/machdep.h @@ -20,8 +20,7 @@ struct trapframe; void arm_lock_cache_line(vm_offset_t); void init_proc0(vm_offset_t kstack); void halt(void); -void data_abort_handler(struct trapframe *); -void prefetch_abort_handler(struct trapframe *); +void abort_handler(struct trapframe *, int ); void set_stackptrs(int cpu); void undefinedinstruction_bounce(struct trapframe *); diff --git a/sys/arm/include/pcb.h b/sys/arm/include/pcb.h index 252d94e..b5ed607 100644 --- a/sys/arm/include/pcb.h +++ b/sys/arm/include/pcb.h @@ -39,50 +39,29 @@ #define _MACHINE_PCB_H_ #include <machine/fp.h> +#include <machine/frame.h> -struct trapframe; - -struct pcb_arm32 { - vm_offset_t pcb32_pagedir; /* PT hooks */ - uint32_t *pcb32_pl1vec; /* PTR to vector_base L1 entry*/ - uint32_t pcb32_l1vec; /* Value to stuff on ctx sw */ - u_int pcb32_dacr; /* Domain Access Control Reg */ - /* - * WARNING! - * cpuswitch.S relies on pcb32_r8 being quad-aligned in struct pcb - * (due to the use of "strd" when compiled for XSCALE) - */ - u_int pcb32_r8; /* used */ - u_int pcb32_r9; /* used */ - u_int pcb32_r10; /* used */ - u_int pcb32_r11; /* used */ - u_int pcb32_r12; /* used */ - u_int pcb32_sp; /* used */ - u_int pcb32_lr; - u_int pcb32_pc; -}; -#define pcb_pagedir un_32.pcb32_pagedir -#define pcb_pl1vec un_32.pcb32_pl1vec -#define pcb_l1vec un_32.pcb32_l1vec -#define pcb_dacr un_32.pcb32_dacr -#define pcb_cstate un_32.pcb32_cstate - /* * WARNING! - * See warning for struct pcb_arm32, above, before changing struct pcb! + * Keep pcb_regs first for faster access in switch.S */ struct pcb { + struct switchframe pcb_regs; /* CPU state */ u_int pcb_flags; #define PCB_OWNFPU 0x00000001 #define PCB_NOALIGNFLT 0x00000002 caddr_t pcb_onfault; /* On fault handler */ - struct pcb_arm32 un_32; + vm_offset_t pcb_pagedir; /* PT hooks */ + uint32_t *pcb_pl1vec; /* PTR to vector_base L1 entry*/ + uint32_t pcb_l1vec; /* Value to stuff on ctx sw */ + u_int pcb_dacr; /* Domain Access Control Reg */ + struct vfp_state pcb_vfpstate; /* VP/NEON state */ u_int pcb_vfpcpu; /* VP/NEON last cpu */ } __aligned(8); /* * We need the PCB to be aligned on 8 bytes, as we may - * access it using ldrd/strd, and some CPUs require it + * access it using ldrd/strd, and ARM ABI require it * to by aligned on 8 bytes. */ diff --git a/sys/arm/include/smp.h b/sys/arm/include/smp.h index 6301c9a..3803674 100644 --- a/sys/arm/include/smp.h +++ b/sys/arm/include/smp.h @@ -24,7 +24,7 @@ void ipi_selected(cpuset_t cpus, u_int ipi); /* PIC interface */ void pic_ipi_send(cpuset_t cpus, u_int ipi); void pic_ipi_clear(int ipi); -int pic_ipi_get(int arg); +int pic_ipi_read(int arg); /* Platform interface */ void platform_mp_setmaxid(void); diff --git a/sys/arm/include/sysreg.h b/sys/arm/include/sysreg.h index 76b1f2b..b1c7fd3 100644 --- a/sys/arm/include/sysreg.h +++ b/sys/arm/include/sysreg.h @@ -99,12 +99,13 @@ #if __ARM_ARCH >= 6 /* From ARMv6: */ #define CP15_IFSR(rr) p15, 0, rr, c5, c0, 1 /* Instruction Fault Status Register */ +#endif +#if __ARM_ARCH >= 7 /* From ARMv7: */ #define CP15_ADFSR(rr) p15, 0, rr, c5, c1, 0 /* Auxiliary Data Fault Status Register */ #define CP15_AIFSR(rr) p15, 0, rr, c5, c1, 1 /* Auxiliary Instruction Fault Status Register */ #endif - /* * CP15 C6 registers */ @@ -118,7 +119,7 @@ /* * CP15 C7 registers */ -#if __ARM_ARCH >= 6 +#if __ARM_ARCH >= 7 && defined(SMP) /* From ARMv7: */ #define CP15_ICIALLUIS p15, 0, r0, c7, c1, 0 /* Instruction cache invalidate all PoU, IS */ #define CP15_BPIALLIS p15, 0, r0, c7, c1, 6 /* Branch predictor invalidate all IS */ @@ -128,14 +129,14 @@ #define CP15_ICIALLU p15, 0, r0, c7, c5, 0 /* Instruction cache invalidate all PoU */ #define CP15_ICIMVAU(rr) p15, 0, rr, c7, c5, 1 /* Instruction cache invalidate */ -#if __ARM_ARCH >= 6 +#if __ARM_ARCH == 6 /* Deprecated in ARMv7 */ #define CP15_CP15ISB p15, 0, r0, c7, c5, 4 /* ISB */ #endif #define CP15_BPIALL p15, 0, r0, c7, c5, 6 /* Branch predictor invalidate all */ #define CP15_BPIMVA p15, 0, rr, c7, c5, 7 /* Branch predictor invalidate by MVA */ -#if __ARM_ARCH >= 6 +#if __ARM_ARCH == 6 /* Only ARMv6: */ #define CP15_DCIALL p15, 0, r0, c7, c6, 0 /* Data cache invalidate all */ #endif @@ -147,7 +148,7 @@ #define CP15_ATS1CUR(rr) p15, 0, rr, c7, c8, 2 /* Stage 1 Current state unprivileged read */ #define CP15_ATS1CUW(rr) p15, 0, rr, c7, c8, 3 /* Stage 1 Current state unprivileged write */ -#if __ARM_ARCH >= 6 +#if __ARM_ARCH >= 7 /* From ARMv7: */ #define CP15_ATS12NSOPR(rr) p15, 0, rr, c7, c8, 4 /* Stages 1 and 2 Non-secure only PL1 read */ #define CP15_ATS12NSOPW(rr) p15, 0, rr, c7, c8, 5 /* Stages 1 and 2 Non-secure only PL1 write */ @@ -155,24 +156,24 @@ #define CP15_ATS12NSOUW(rr) p15, 0, rr, c7, c8, 7 /* Stages 1 and 2 Non-secure only unprivileged write */ #endif -#if __ARM_ARCH >= 6 +#if __ARM_ARCH == 6 /* Only ARMv6: */ #define CP15_DCCALL p15, 0, r0, c7, c10, 0 /* Data cache clean all */ #endif #define CP15_DCCMVAC(rr) p15, 0, rr, c7, c10, 1 /* Data cache clean by MVA PoC */ #define CP15_DCCSW(rr) p15, 0, rr, c7, c10, 2 /* Data cache clean by set/way */ -#if __ARM_ARCH >= 6 +#if __ARM_ARCH == 6 /* Only ARMv6: */ #define CP15_CP15DSB p15, 0, r0, c7, c10, 4 /* DSB */ #define CP15_CP15DMB p15, 0, r0, c7, c10, 5 /* DMB */ #endif -#if __ARM_ARCH >= 6 +#if __ARM_ARCH >= 7 /* From ARMv7: */ #define CP15_DCCMVAU(rr) p15, 0, rr, c7, c11, 1 /* Data cache clean by MVA PoU */ #endif -#if __ARM_ARCH >= 6 +#if __ARM_ARCH == 6 /* Only ARMv6: */ #define CP15_DCCIALL p15, 0, r0, c7, c14, 0 /* Data cache clean and invalidate all */ #endif @@ -182,7 +183,7 @@ /* * CP15 C8 registers */ -#if __ARM_ARCH >= 6 +#if __ARM_ARCH >= 7 && defined(SMP) /* From ARMv7: */ #define CP15_TLBIALLIS p15, 0, r0, c8, c3, 0 /* Invalidate entire unified TLB IS */ #define CP15_TLBIMVAIS(rr) p15, 0, rr, c8, c3, 1 /* Invalidate unified TLB by MVA IS */ @@ -229,4 +230,9 @@ #define CP15_TPIDRURO(rr) p15, 0, rr, c13, c0, 3 /* User Read-Only Thread ID Register */ #define CP15_TPIDRPRW(rr) p15, 0, rr, c13, c0, 4 /* PL1 only Thread ID Register */ +/* + * CP15 C15 registers + */ +#define CP15_CBAR(rr) p15, 4, rr, c15, c0, 0 /* Configuration Base Address Register */ + #endif /* !MACHINE_SYSREG_H */ diff --git a/sys/arm/lpc/lpc_intc.c b/sys/arm/lpc/lpc_intc.c index 3137e89..bf26645 100644 --- a/sys/arm/lpc/lpc_intc.c +++ b/sys/arm/lpc/lpc_intc.c @@ -59,10 +59,10 @@ static void lpc_intc_eoi(void *); static struct lpc_intc_softc *intc_softc = NULL; -#define intc_read_4(reg) \ - bus_space_read_4(intc_softc->li_bst, intc_softc->li_bsh, reg) -#define intc_write_4(reg, val) \ - bus_space_write_4(intc_softc->li_bst, intc_softc->li_bsh, reg, val) +#define intc_read_4(_sc, _reg) \ + bus_space_read_4((_sc)->li_bst, (_sc)->li_bsh, (_reg)) +#define intc_write_4(_sc, _reg, _val) \ + bus_space_write_4((_sc)->li_bst, (_sc)->li_bsh, (_reg), (_val)) static int lpc_intc_probe(device_t dev) @@ -100,12 +100,12 @@ lpc_intc_attach(device_t dev) arm_post_filter = lpc_intc_eoi; /* Clear interrupt status registers and disable all interrupts */ - intc_write_4(LPC_INTC_MIC_ER, 0); - intc_write_4(LPC_INTC_SIC1_ER, 0); - intc_write_4(LPC_INTC_SIC2_ER, 0); - intc_write_4(LPC_INTC_MIC_RSR, ~0); - intc_write_4(LPC_INTC_SIC1_RSR, ~0); - intc_write_4(LPC_INTC_SIC2_RSR, ~0); + intc_write_4(sc, LPC_INTC_MIC_ER, 0); + intc_write_4(sc, LPC_INTC_SIC1_ER, 0); + intc_write_4(sc, LPC_INTC_SIC2_ER, 0); + intc_write_4(sc, LPC_INTC_MIC_RSR, ~0); + intc_write_4(sc, LPC_INTC_SIC1_RSR, ~0); + intc_write_4(sc, LPC_INTC_SIC2_RSR, ~0); return (0); } @@ -128,25 +128,26 @@ DRIVER_MODULE(pic, simplebus, lpc_intc_driver, lpc_intc_devclass, 0, 0); int arm_get_next_irq(int last) { + struct lpc_intc_softc *sc = intc_softc; uint32_t value; int i; /* IRQs 0-31 are mapped to LPC_INTC_MIC_SR */ - value = intc_read_4(LPC_INTC_MIC_SR); + value = intc_read_4(sc, LPC_INTC_MIC_SR); for (i = 0; i < 32; i++) { if (value & (1 << i)) return (i); } /* IRQs 32-63 are mapped to LPC_INTC_SIC1_SR */ - value = intc_read_4(LPC_INTC_SIC1_SR); + value = intc_read_4(sc, LPC_INTC_SIC1_SR); for (i = 0; i < 32; i++) { if (value & (1 << i)) return (i + 32); } /* IRQs 64-95 are mapped to LPC_INTC_SIC2_SR */ - value = intc_read_4(LPC_INTC_SIC2_SR); + value = intc_read_4(sc, LPC_INTC_SIC2_SR); for (i = 0; i < 32; i++) { if (value & (1 << i)) return (i + 64); @@ -158,6 +159,7 @@ arm_get_next_irq(int last) void arm_mask_irq(uintptr_t nb) { + struct lpc_intc_softc *sc = intc_softc; int reg; uint32_t value; @@ -174,14 +176,15 @@ arm_mask_irq(uintptr_t nb) reg = LPC_INTC_MIC_ER; /* Clear bit in ER register */ - value = intc_read_4(reg); + value = intc_read_4(sc, reg); value &= ~(1 << nb); - intc_write_4(reg, value); + intc_write_4(sc, reg, value); } void arm_unmask_irq(uintptr_t nb) { + struct lpc_intc_softc *sc = intc_softc; int reg; uint32_t value; @@ -195,14 +198,15 @@ arm_unmask_irq(uintptr_t nb) reg = LPC_INTC_MIC_ER; /* Set bit in ER register */ - value = intc_read_4(reg); + value = intc_read_4(sc, reg); value |= (1 << nb); - intc_write_4(reg, value); + intc_write_4(sc, reg, value); } static void lpc_intc_eoi(void *data) { + struct lpc_intc_softc *sc = intc_softc; int reg; int nb = (int)data; uint32_t value; @@ -217,9 +221,9 @@ lpc_intc_eoi(void *data) reg = LPC_INTC_MIC_RSR; /* Set bit in RSR register */ - value = intc_read_4(reg); + value = intc_read_4(sc, reg); value |= (1 << nb); - intc_write_4(reg, value); + intc_write_4(sc, reg, value); } diff --git a/sys/arm/mv/mpic.c b/sys/arm/mv/mpic.c index 813caf4..bfb5554 100644 --- a/sys/arm/mv/mpic.c +++ b/sys/arm/mv/mpic.c @@ -375,7 +375,7 @@ pic_ipi_send(cpuset_t cpus, u_int ipi) } int -pic_ipi_get(int i __unused) +pic_ipi_read(int i __unused) { uint32_t val; diff --git a/sys/arm/mv/std-pj4b.mv b/sys/arm/mv/std-pj4b.mv index 7d779c2..1053ff0 100644 --- a/sys/arm/mv/std-pj4b.mv +++ b/sys/arm/mv/std-pj4b.mv @@ -3,5 +3,6 @@ files "../mv/files.mv" cpu CPU_MV_PJ4B machine arm armv6 +makeoptions CONF_CFLAGS="-march=armv7a -Wa,-march=armv7a" options VM_MAXUSER_ADDRESS="(KERNBASE-(1024*1024*1024))" diff --git a/sys/arm/rockchip/std.rk30xx b/sys/arm/rockchip/std.rk30xx index 79ddcc6..b62b77a 100644 --- a/sys/arm/rockchip/std.rk30xx +++ b/sys/arm/rockchip/std.rk30xx @@ -3,6 +3,7 @@ cpu CPU_CORTEXA machine arm armv6 +makeoptions CONF_CFLAGS="-march=armv7a -Wa,-march=armv7a" makeoption ARM_LITTLE_ENDIAN # Physical memory starts at 0x60400000. We assume images are loaded at diff --git a/sys/arm/samsung/exynos/std.exynos5250 b/sys/arm/samsung/exynos/std.exynos5250 index 5f59adc..39c378b 100644 --- a/sys/arm/samsung/exynos/std.exynos5250 +++ b/sys/arm/samsung/exynos/std.exynos5250 @@ -4,6 +4,7 @@ makeoption ARM_LITTLE_ENDIAN cpu CPU_CORTEXA machine arm armv6 +makeoptions CONF_CFLAGS="-march=armv7a -Wa,-march=armv7a" options PHYSADDR=0x40000000 diff --git a/sys/arm/samsung/exynos/std.exynos5420 b/sys/arm/samsung/exynos/std.exynos5420 index c6468e7..b70537d 100644 --- a/sys/arm/samsung/exynos/std.exynos5420 +++ b/sys/arm/samsung/exynos/std.exynos5420 @@ -4,6 +4,7 @@ makeoption ARM_LITTLE_ENDIAN cpu CPU_CORTEXA machine arm armv6 +makeoptions CONF_CFLAGS="-march=armv7a -Wa,-march=armv7a" options PHYSADDR=0x20000000 diff --git a/sys/arm/ti/aintc.c b/sys/arm/ti/aintc.c index 213a1d3..675a5f3 100644 --- a/sys/arm/ti/aintc.c +++ b/sys/arm/ti/aintc.c @@ -72,10 +72,10 @@ static struct resource_spec ti_aintc_spec[] = { static struct ti_aintc_softc *ti_aintc_sc = NULL; -#define aintc_read_4(reg) \ - bus_space_read_4(ti_aintc_sc->aintc_bst, ti_aintc_sc->aintc_bsh, reg) -#define aintc_write_4(reg, val) \ - bus_space_write_4(ti_aintc_sc->aintc_bst, ti_aintc_sc->aintc_bsh, reg, val) +#define aintc_read_4(_sc, reg) \ + bus_space_read_4((_sc)->aintc_bst, (_sc)->aintc_bsh, (reg)) +#define aintc_write_4(_sc, reg, val) \ + bus_space_write_4((_sc)->aintc_bst, (_sc)->aintc_bsh, (reg), (val)) static int @@ -112,17 +112,17 @@ ti_aintc_attach(device_t dev) ti_aintc_sc = sc; - x = aintc_read_4(INTC_REVISION); + x = aintc_read_4(sc, INTC_REVISION); device_printf(dev, "Revision %u.%u\n",(x >> 4) & 0xF, x & 0xF); /* SoftReset */ - aintc_write_4(INTC_SYSCONFIG, 2); + aintc_write_4(sc, INTC_SYSCONFIG, 2); /* Wait for reset to complete */ - while(!(aintc_read_4(INTC_SYSSTATUS) & 1)); + while(!(aintc_read_4(sc, INTC_SYSSTATUS) & 1)); /*Set Priority Threshold */ - aintc_write_4(INTC_THRESHOLD, 0xFF); + aintc_write_4(sc, INTC_THRESHOLD, 0xFF); return (0); } @@ -146,22 +146,23 @@ DRIVER_MODULE(aintc, simplebus, ti_aintc_driver, ti_aintc_devclass, 0, 0); int arm_get_next_irq(int last_irq) { + struct ti_aintc_softc *sc = ti_aintc_sc; uint32_t active_irq; if (last_irq != -1) { - aintc_write_4(INTC_ISR_CLEAR(last_irq >> 5), + aintc_write_4(sc, INTC_ISR_CLEAR(last_irq >> 5), 1UL << (last_irq & 0x1F)); - aintc_write_4(INTC_CONTROL,1); + aintc_write_4(sc, INTC_CONTROL, 1); } /* Get the next active interrupt */ - active_irq = aintc_read_4(INTC_SIR_IRQ); + active_irq = aintc_read_4(sc, INTC_SIR_IRQ); /* Check for spurious interrupt */ if ((active_irq & 0xffffff80)) { - device_printf(ti_aintc_sc->sc_dev, - "Spurious interrupt detected (0x%08x)\n", active_irq); - aintc_write_4(INTC_SIR_IRQ, 0); + device_printf(sc->sc_dev, + "Spurious interrupt detected (0x%08x)\n", active_irq); + aintc_write_4(sc, INTC_SIR_IRQ, 0); return -1; } @@ -174,13 +175,16 @@ arm_get_next_irq(int last_irq) void arm_mask_irq(uintptr_t nb) { - aintc_write_4(INTC_MIR_SET(nb >> 5), (1UL << (nb & 0x1F))); + struct ti_aintc_softc *sc = ti_aintc_sc; + + aintc_write_4(sc, INTC_MIR_SET(nb >> 5), (1UL << (nb & 0x1F))); } void arm_unmask_irq(uintptr_t nb) { + struct ti_aintc_softc *sc = ti_aintc_sc; arm_irq_memory_barrier(nb); - aintc_write_4(INTC_MIR_CLEAR(nb >> 5), (1UL << (nb & 0x1F))); + aintc_write_4(sc, INTC_MIR_CLEAR(nb >> 5), (1UL << (nb & 0x1F))); } diff --git a/sys/arm/ti/am335x/am335x_usbss.c b/sys/arm/ti/am335x/am335x_usbss.c index f1204f3..0173922 100644 --- a/sys/arm/ti/am335x/am335x_usbss.c +++ b/sys/arm/ti/am335x/am335x_usbss.c @@ -288,21 +288,30 @@ musbotg_attach(device_t dev) return (ENXIO); } + /* Enable device clocks. */ + ti_prcm_clk_enable(MUSB0_CLK); + /* - * Reset USBSS, USB0 and USB1 + * Reset USBSS, USB0 and USB1. + * The registers of USB subsystem must not be accessed while the + * reset pulse is active (200ns). */ + USBSS_WRITE4(sc, USBSS_SYSCONFIG, USBSS_SYSCONFIG_SRESET); + DELAY(100); + i = 10; + while (USBSS_READ4(sc, USBSS_SYSCONFIG) & USBSS_SYSCONFIG_SRESET) { + DELAY(100); + if (i-- == 0) { + device_printf(dev, "reset timeout.\n"); + return (ENXIO); + } + } + + /* Read the module revision. */ rev = USBSS_READ4(sc, USBSS_REVREG); device_printf(dev, "TI AM335X USBSS v%d.%d.%d\n", (rev >> 8) & 7, (rev >> 6) & 3, rev & 63); - ti_prcm_clk_enable(MUSB0_CLK); - - USBSS_WRITE4(sc, USBSS_SYSCONFIG, - USBSS_SYSCONFIG_SRESET); - while (USBSS_READ4(sc, USBSS_SYSCONFIG) & - USBSS_SYSCONFIG_SRESET) - ; - err = bus_setup_intr(dev, sc->sc_irq_res[0], INTR_TYPE_BIO | INTR_MPSAFE, NULL, (driver_intr_t *)musbotg_usbss_interrupt, sc, diff --git a/sys/arm/ti/std.ti b/sys/arm/ti/std.ti index 218d731..5e859b6 100644 --- a/sys/arm/ti/std.ti +++ b/sys/arm/ti/std.ti @@ -2,5 +2,6 @@ cpu CPU_CORTEXA machine arm armv6 +makeoptions CONF_CFLAGS="-march=armv7a -Wa,-march=armv7a" files "../ti/files.ti" diff --git a/sys/arm/ti/ti_gpio.c b/sys/arm/ti/ti_gpio.c index 612ceb9..5fe8094 100644 --- a/sys/arm/ti/ti_gpio.c +++ b/sys/arm/ti/ti_gpio.c @@ -1,6 +1,6 @@ /*- - * Copyright (c) 2011 - * Ben Gray <ben.r.gray@gmail.com>. + * Copyright (c) 2011 Ben Gray <ben.r.gray@gmail.com>. + * Copyright (c) 2014 Luiz Otavio O Souza <loos@FreeBSD.org>. * All rights reserved. * * Redistribution and use in source and binary forms, with or without @@ -26,17 +26,8 @@ */ /** - * Very simple GPIO (general purpose IO) driver module for TI OMAP SoC's. - * - * Currently this driver only does the basics, get a value on a pin & set a - * value on a pin. Hopefully over time I'll expand this to be a bit more generic - * and support interrupts and other various bits on the SoC can do ... in the - * meantime this is all you get. - * - * Beware the OMA datasheet(s) lists GPIO banks 1-6, whereas I've used 0-5 here - * in the code. - * - * + * Beware that the OMAP4 datasheet(s) lists GPIO banks 1-6, whereas the code + * here uses 0-5. */ #include <sys/cdefs.h> @@ -52,6 +43,7 @@ __FBSDID("$FreeBSD$"); #include <sys/lock.h> #include <sys/mutex.h> #include <sys/gpio.h> +#include <sys/interrupt.h> #include <machine/bus.h> #include <machine/resource.h> @@ -69,10 +61,13 @@ __FBSDID("$FreeBSD$"); #include "gpio_if.h" #include "ti_gpio_if.h" +#if !defined(SOC_OMAP4) && !defined(SOC_TI_AM335X) +#error "Unknown SoC" +#endif + /* Register definitions */ #define TI_GPIO_REVISION 0x0000 #define TI_GPIO_SYSCONFIG 0x0010 -#if defined(SOC_OMAP4) || defined(SOC_TI_AM335X) #define TI_GPIO_IRQSTATUS_RAW_0 0x0024 #define TI_GPIO_IRQSTATUS_RAW_1 0x0028 #define TI_GPIO_IRQSTATUS_0 0x002C @@ -103,9 +98,6 @@ __FBSDID("$FreeBSD$"); #define TI_GPIO_SETWKUENA 0x0184 #define TI_GPIO_CLEARDATAOUT 0x0190 #define TI_GPIO_SETDATAOUT 0x0194 -#else -#error "Unknown SoC" -#endif /* Other SoC Specific definitions */ #define OMAP4_MAX_GPIO_BANKS 6 @@ -117,6 +109,10 @@ __FBSDID("$FreeBSD$"); #define AM335X_INTR_PER_BANK 2 #define AM335X_GPIO_REV 0x50600801 #define PINS_PER_BANK 32 +#define TI_GPIO_BANK(p) ((p) / PINS_PER_BANK) +#define TI_GPIO_MASK(p) (1U << ((p) % PINS_PER_BANK)) + +static struct ti_gpio_softc *ti_gpio_sc = NULL; static u_int ti_max_gpio_banks(void) @@ -226,17 +222,17 @@ static struct resource_spec ti_gpio_irq_spec[] = { /** * Macros for driver mutex locking */ -#define TI_GPIO_LOCK(_sc) mtx_lock(&(_sc)->sc_mtx) -#define TI_GPIO_UNLOCK(_sc) mtx_unlock(&(_sc)->sc_mtx) +#define TI_GPIO_LOCK(_sc) mtx_lock_spin(&(_sc)->sc_mtx) +#define TI_GPIO_UNLOCK(_sc) mtx_unlock_spin(&(_sc)->sc_mtx) #define TI_GPIO_LOCK_INIT(_sc) \ - mtx_init(&_sc->sc_mtx, device_get_nameunit(_sc->sc_dev), \ - "ti_gpio", MTX_DEF) -#define TI_GPIO_LOCK_DESTROY(_sc) mtx_destroy(&_sc->sc_mtx) -#define TI_GPIO_ASSERT_LOCKED(_sc) mtx_assert(&_sc->sc_mtx, MA_OWNED) -#define TI_GPIO_ASSERT_UNLOCKED(_sc) mtx_assert(&_sc->sc_mtx, MA_NOTOWNED) + mtx_init(&_sc->sc_mtx, device_get_nameunit((_sc)->sc_dev), \ + "ti_gpio", MTX_SPIN) +#define TI_GPIO_LOCK_DESTROY(_sc) mtx_destroy(&(_sc)->sc_mtx) +#define TI_GPIO_ASSERT_LOCKED(_sc) mtx_assert(&(_sc)->sc_mtx, MA_OWNED) +#define TI_GPIO_ASSERT_UNLOCKED(_sc) mtx_assert(&(_sc)->sc_mtx, MA_NOTOWNED) /** - * ti_gpio_read_4 - reads a 16-bit value from one of the PADCONFS registers + * ti_gpio_read_4 - reads a 32-bit value from one of the GPIO registers * @sc: GPIO device context * @bank: The bank to read from * @off: The offset of a register from the GPIO register address range @@ -252,7 +248,7 @@ ti_gpio_read_4(struct ti_gpio_softc *sc, unsigned int bank, bus_size_t off) } /** - * ti_gpio_write_4 - writes a 32-bit value to one of the PADCONFS registers + * ti_gpio_write_4 - writes a 32-bit value to one of the GPIO registers * @sc: GPIO device context * @bank: The bank to write to * @off: The offset of a register from the GPIO register address range @@ -273,13 +269,43 @@ ti_gpio_intr_clr(struct ti_gpio_softc *sc, unsigned int bank, uint32_t mask) { /* We clear both set of registers. */ -#if defined(SOC_OMAP4) || defined(SOC_TI_AM335X) ti_gpio_write_4(sc, bank, TI_GPIO_IRQSTATUS_CLR_0, mask); ti_gpio_write_4(sc, bank, TI_GPIO_IRQSTATUS_CLR_1, mask); -#else - ti_gpio_write_4(sc, bank, TI_GPIO_CLEARIRQENABLE1, mask); - ti_gpio_write_4(sc, bank, TI_GPIO_CLEARIRQENABLE2, mask); -#endif +} + +static inline void +ti_gpio_intr_set(struct ti_gpio_softc *sc, unsigned int bank, uint32_t mask) +{ + + /* + * On OMAP4 we unmask only the MPU interrupt and on AM335x we + * also activate only the first interrupt. + */ + ti_gpio_write_4(sc, bank, TI_GPIO_IRQSTATUS_SET_0, mask); +} + +static inline void +ti_gpio_intr_ack(struct ti_gpio_softc *sc, unsigned int bank, uint32_t mask) +{ + + /* + * Acknowledge the interrupt on both registers even if we use only + * the first one. + */ + ti_gpio_write_4(sc, bank, TI_GPIO_IRQSTATUS_0, mask); + ti_gpio_write_4(sc, bank, TI_GPIO_IRQSTATUS_1, mask); +} + +static inline uint32_t +ti_gpio_intr_status(struct ti_gpio_softc *sc, unsigned int bank) +{ + uint32_t reg; + + /* Get the status from both registers. */ + reg = ti_gpio_read_4(sc, bank, TI_GPIO_IRQSTATUS_0); + reg |= ti_gpio_read_4(sc, bank, TI_GPIO_IRQSTATUS_1); + + return (reg); } /** @@ -290,7 +316,7 @@ ti_gpio_intr_clr(struct ti_gpio_softc *sc, unsigned int bank, uint32_t mask) * * * LOCKING: - * Internally locks the context + * No locking required, returns static data. * * RETURNS: * Returns 0 on success otherwise an error code @@ -298,23 +324,21 @@ ti_gpio_intr_clr(struct ti_gpio_softc *sc, unsigned int bank, uint32_t mask) static int ti_gpio_pin_max(device_t dev, int *maxpin) { - struct ti_gpio_softc *sc = device_get_softc(dev); - unsigned int i; - unsigned int banks = 0; - TI_GPIO_LOCK(sc); + *maxpin = ti_max_gpio_banks() * PINS_PER_BANK - 1; - /* Calculate how many valid banks we have and then multiply that by 32 to - * give use the total number of pins. - */ - for (i = 0; i < ti_max_gpio_banks(); i++) { - if (sc->sc_mem_res[i] != NULL) - banks++; - } + return (0); +} - *maxpin = (banks * PINS_PER_BANK) - 1; +static int +ti_gpio_valid_pin(struct ti_gpio_softc *sc, int pin) +{ - TI_GPIO_UNLOCK(sc); + if (pin > sc->sc_maxpin || + TI_GPIO_BANK(pin) >= ti_max_gpio_banks() || + sc->sc_mem_res[TI_GPIO_BANK(pin)] == NULL) { + return (EINVAL); + } return (0); } @@ -332,7 +356,7 @@ ti_gpio_pin_max(device_t dev, int *maxpin) * - GPIO_PIN_PULLDOWN * * LOCKING: - * Internally locks the context + * No locking required, returns static data. * * RETURNS: * Returns 0 on success otherwise an error code @@ -340,22 +364,15 @@ ti_gpio_pin_max(device_t dev, int *maxpin) static int ti_gpio_pin_getcaps(device_t dev, uint32_t pin, uint32_t *caps) { - struct ti_gpio_softc *sc = device_get_softc(dev); - uint32_t bank = (pin / PINS_PER_BANK); - - TI_GPIO_LOCK(sc); + struct ti_gpio_softc *sc; - /* Sanity check the pin number is valid */ - if ((bank >= ti_max_gpio_banks()) || (sc->sc_mem_res[bank] == NULL)) { - TI_GPIO_UNLOCK(sc); + sc = device_get_softc(dev); + if (ti_gpio_valid_pin(sc, pin) != 0) return (EINVAL); - } - *caps = (GPIO_PIN_INPUT | GPIO_PIN_OUTPUT |GPIO_PIN_PULLUP | + *caps = (GPIO_PIN_INPUT | GPIO_PIN_OUTPUT | GPIO_PIN_PULLUP | GPIO_PIN_PULLDOWN); - TI_GPIO_UNLOCK(sc); - return (0); } @@ -378,20 +395,15 @@ ti_gpio_pin_getcaps(device_t dev, uint32_t pin, uint32_t *caps) static int ti_gpio_pin_getflags(device_t dev, uint32_t pin, uint32_t *flags) { - struct ti_gpio_softc *sc = device_get_softc(dev); - uint32_t bank = (pin / PINS_PER_BANK); - - TI_GPIO_LOCK(sc); + struct ti_gpio_softc *sc; - /* Sanity check the pin number is valid */ - if ((bank >= ti_max_gpio_banks()) || (sc->sc_mem_res[bank] == NULL)) { - TI_GPIO_UNLOCK(sc); + sc = device_get_softc(dev); + if (ti_gpio_valid_pin(sc, pin) != 0) return (EINVAL); - } /* Get the current pin state */ + TI_GPIO_LOCK(sc); TI_GPIO_GET_FLAGS(dev, pin, flags); - TI_GPIO_UNLOCK(sc); return (0); @@ -407,7 +419,7 @@ ti_gpio_pin_getflags(device_t dev, uint32_t pin, uint32_t *flags) * of the pin. * * LOCKING: - * Internally locks the context + * No locking required, returns static data. * * RETURNS: * Returns 0 on success otherwise an error code @@ -415,23 +427,16 @@ ti_gpio_pin_getflags(device_t dev, uint32_t pin, uint32_t *flags) static int ti_gpio_pin_getname(device_t dev, uint32_t pin, char *name) { - struct ti_gpio_softc *sc = device_get_softc(dev); - uint32_t bank = (pin / PINS_PER_BANK); - - TI_GPIO_LOCK(sc); + struct ti_gpio_softc *sc; - /* Sanity check the pin number is valid */ - if ((bank >= ti_max_gpio_banks()) || (sc->sc_mem_res[bank] == NULL)) { - TI_GPIO_UNLOCK(sc); + sc = device_get_softc(dev); + if (ti_gpio_valid_pin(sc, pin) != 0) return (EINVAL); - } /* Set a very simple name */ snprintf(name, GPIOMAXNAME, "gpio_%u", pin); name[GPIOMAXNAME - 1] = '\0'; - TI_GPIO_UNLOCK(sc); - return (0); } @@ -457,33 +462,27 @@ ti_gpio_pin_getname(device_t dev, uint32_t pin, char *name) static int ti_gpio_pin_setflags(device_t dev, uint32_t pin, uint32_t flags) { - struct ti_gpio_softc *sc = device_get_softc(dev); - uint32_t bank = (pin / PINS_PER_BANK); - uint32_t mask = (1UL << (pin % PINS_PER_BANK)); - uint32_t reg_val; - - TI_GPIO_LOCK(sc); + struct ti_gpio_softc *sc; + uint32_t oe; - /* Sanity check the pin number is valid */ - if ((bank >= ti_max_gpio_banks()) || (sc->sc_mem_res[bank] == NULL)) { - TI_GPIO_UNLOCK(sc); + sc = device_get_softc(dev); + if (ti_gpio_valid_pin(sc, pin) != 0) return (EINVAL); - } /* Set the GPIO mode and state */ + TI_GPIO_LOCK(sc); if (TI_GPIO_SET_FLAGS(dev, pin, flags) != 0) { TI_GPIO_UNLOCK(sc); return (EINVAL); } /* If configuring as an output set the "output enable" bit */ - reg_val = ti_gpio_read_4(sc, bank, TI_GPIO_OE); + oe = ti_gpio_read_4(sc, TI_GPIO_BANK(pin), TI_GPIO_OE); if (flags & GPIO_PIN_INPUT) - reg_val |= mask; + oe |= TI_GPIO_MASK(pin); else - reg_val &= ~mask; - ti_gpio_write_4(sc, bank, TI_GPIO_OE, reg_val); - + oe &= ~TI_GPIO_MASK(pin); + ti_gpio_write_4(sc, TI_GPIO_BANK(pin), TI_GPIO_OE, oe); TI_GPIO_UNLOCK(sc); return (0); @@ -506,21 +505,19 @@ ti_gpio_pin_setflags(device_t dev, uint32_t pin, uint32_t flags) static int ti_gpio_pin_set(device_t dev, uint32_t pin, unsigned int value) { - struct ti_gpio_softc *sc = device_get_softc(dev); - uint32_t bank = (pin / PINS_PER_BANK); - uint32_t mask = (1UL << (pin % PINS_PER_BANK)); - - TI_GPIO_LOCK(sc); + struct ti_gpio_softc *sc; + uint32_t reg; - /* Sanity check the pin number is valid */ - if ((bank >= ti_max_gpio_banks()) || (sc->sc_mem_res[bank] == NULL)) { - TI_GPIO_UNLOCK(sc); + sc = device_get_softc(dev); + if (ti_gpio_valid_pin(sc, pin) != 0) return (EINVAL); - } - - ti_gpio_write_4(sc, bank, (value == GPIO_PIN_LOW) ? TI_GPIO_CLEARDATAOUT - : TI_GPIO_SETDATAOUT, mask); + TI_GPIO_LOCK(sc); + if (value == GPIO_PIN_LOW) + reg = TI_GPIO_CLEARDATAOUT; + else + reg = TI_GPIO_SETDATAOUT; + ti_gpio_write_4(sc, TI_GPIO_BANK(pin), reg, TI_GPIO_MASK(pin)); TI_GPIO_UNLOCK(sc); return (0); @@ -544,28 +541,25 @@ ti_gpio_pin_set(device_t dev, uint32_t pin, unsigned int value) static int ti_gpio_pin_get(device_t dev, uint32_t pin, unsigned int *value) { - struct ti_gpio_softc *sc = device_get_softc(dev); - uint32_t bank = (pin / PINS_PER_BANK); - uint32_t mask = (1UL << (pin % PINS_PER_BANK)); - uint32_t val = 0; - - TI_GPIO_LOCK(sc); + struct ti_gpio_softc *sc; + uint32_t oe, reg, val; - /* Sanity check the pin number is valid */ - if ((bank >= ti_max_gpio_banks()) || (sc->sc_mem_res[bank] == NULL)) { - TI_GPIO_UNLOCK(sc); + sc = device_get_softc(dev); + if (ti_gpio_valid_pin(sc, pin) != 0) return (EINVAL); - } - - /* Sanity check the pin is not configured as an output */ - val = ti_gpio_read_4(sc, bank, TI_GPIO_OE); - /* Read the value on the pin */ - if (val & mask) - *value = (ti_gpio_read_4(sc, bank, TI_GPIO_DATAIN) & mask) ? 1 : 0; + /* + * Return data from output latch when set as output and from the + * input register otherwise. + */ + TI_GPIO_LOCK(sc); + oe = ti_gpio_read_4(sc, TI_GPIO_BANK(pin), TI_GPIO_OE); + if (oe & TI_GPIO_MASK(pin)) + reg = TI_GPIO_DATAIN; else - *value = (ti_gpio_read_4(sc, bank, TI_GPIO_DATAOUT) & mask) ? 1 : 0; - + reg = TI_GPIO_DATAOUT; + val = ti_gpio_read_4(sc, TI_GPIO_BANK(pin), reg); + *value = (val & TI_GPIO_MASK(pin)) ? 1 : 0; TI_GPIO_UNLOCK(sc); return (0); @@ -586,26 +580,21 @@ ti_gpio_pin_get(device_t dev, uint32_t pin, unsigned int *value) static int ti_gpio_pin_toggle(device_t dev, uint32_t pin) { - struct ti_gpio_softc *sc = device_get_softc(dev); - uint32_t bank = (pin / PINS_PER_BANK); - uint32_t mask = (1UL << (pin % PINS_PER_BANK)); - uint32_t val; - - TI_GPIO_LOCK(sc); + struct ti_gpio_softc *sc; + uint32_t reg, val; - /* Sanity check the pin number is valid */ - if ((bank >= ti_max_gpio_banks()) || (sc->sc_mem_res[bank] == NULL)) { - TI_GPIO_UNLOCK(sc); + sc = device_get_softc(dev); + if (ti_gpio_valid_pin(sc, pin) != 0) return (EINVAL); - } /* Toggle the pin */ - val = ti_gpio_read_4(sc, bank, TI_GPIO_DATAOUT); - if (val & mask) - ti_gpio_write_4(sc, bank, TI_GPIO_CLEARDATAOUT, mask); + TI_GPIO_LOCK(sc); + val = ti_gpio_read_4(sc, TI_GPIO_BANK(pin), TI_GPIO_DATAOUT); + if (val & TI_GPIO_MASK(pin)) + reg = TI_GPIO_CLEARDATAOUT; else - ti_gpio_write_4(sc, bank, TI_GPIO_SETDATAOUT, mask); - + reg = TI_GPIO_SETDATAOUT; + ti_gpio_write_4(sc, TI_GPIO_BANK(pin), reg, TI_GPIO_MASK(pin)); TI_GPIO_UNLOCK(sc); return (0); @@ -615,20 +604,40 @@ ti_gpio_pin_toggle(device_t dev, uint32_t pin) * ti_gpio_intr - ISR for all GPIO modules * @arg: the soft context pointer * - * Unsused - * * LOCKING: * Internally locks the context * */ -static void +static int ti_gpio_intr(void *arg) { - struct ti_gpio_softc *sc = arg; + int bank_last, irq; + struct intr_event *event; + struct ti_gpio_softc *sc; + uint32_t reg; - TI_GPIO_LOCK(sc); - /* TODO: something useful */ - TI_GPIO_UNLOCK(sc); + sc = (struct ti_gpio_softc *)arg; + bank_last = -1; + reg = 0; /* squelch bogus gcc warning */ + for (irq = 0; irq < sc->sc_maxpin; irq++) { + + /* Read interrupt status only once for each bank. */ + if (TI_GPIO_BANK(irq) != bank_last) { + reg = ti_gpio_intr_status(sc, TI_GPIO_BANK(irq)); + bank_last = TI_GPIO_BANK(irq); + } + if ((reg & TI_GPIO_MASK(irq)) == 0) + continue; + event = sc->sc_events[irq]; + if (event != NULL && !TAILQ_EMPTY(&event->ie_handlers)) + intr_event_handle(event, NULL); + else + device_printf(sc->sc_dev, "Stray IRQ %d\n", irq); + /* Ack the IRQ Status bit. */ + ti_gpio_intr_ack(sc, TI_GPIO_BANK(irq), TI_GPIO_MASK(irq)); + } + + return (FILTER_HANDLED); } static int @@ -643,13 +652,13 @@ ti_gpio_attach_intr(device_t dev) break; /* - * Register our interrupt handler for each of the IRQ resources. + * Register our interrupt filter for each of the IRQ resources. */ if (bus_setup_intr(dev, sc->sc_irq_res[i], - INTR_TYPE_MISC | INTR_MPSAFE, NULL, ti_gpio_intr, sc, + INTR_TYPE_MISC | INTR_MPSAFE, ti_gpio_intr, NULL, sc, &sc->sc_irq_hdl[i]) != 0) { device_printf(dev, - "WARNING: unable to register interrupt handler\n"); + "WARNING: unable to register interrupt filter\n"); return (-1); } } @@ -663,7 +672,7 @@ ti_gpio_detach_intr(device_t dev) int i; struct ti_gpio_softc *sc; - /* Teardown our interrupt handlers. */ + /* Teardown our interrupt filters. */ sc = device_get_softc(dev); for (i = 0; i < ti_max_gpio_intrs(); i++) { if (sc->sc_irq_res[i] == NULL) @@ -683,7 +692,7 @@ ti_gpio_bank_init(device_t dev, int bank) { int pin; struct ti_gpio_softc *sc; - uint32_t flags, reg_oe; + uint32_t flags, reg_oe, rev; sc = device_get_softc(dev); @@ -695,13 +704,12 @@ ti_gpio_bank_init(device_t dev, int bank) * actual revision numbers, so instead the values have been * determined by experimentation. */ - sc->sc_revision[bank] = ti_gpio_read_4(sc, bank, TI_GPIO_REVISION); + rev = ti_gpio_read_4(sc, bank, TI_GPIO_REVISION); /* Check the revision. */ - if (sc->sc_revision[bank] != ti_gpio_rev()) { + if (rev != ti_gpio_rev()) { device_printf(dev, "Warning: could not determine the revision " - "of %u GPIO module (revision:0x%08x)\n", - bank, sc->sc_revision[bank]); + "of GPIO module %d (revision:0x%08x)\n", bank, rev); return (EINVAL); } @@ -740,10 +748,13 @@ ti_gpio_attach(device_t dev) unsigned int i; int err; - sc = device_get_softc(dev); - sc->sc_dev = dev; + if (ti_gpio_sc != NULL) + return (ENXIO); + ti_gpio_sc = sc = device_get_softc(dev); + sc->sc_dev = dev; TI_GPIO_LOCK_INIT(sc); + ti_gpio_pin_max(dev, &sc->sc_maxpin); /* There are up to 6 different GPIO register sets located in different * memory areas on the chip. The memory range should have been set for @@ -769,6 +780,24 @@ ti_gpio_attach(device_t dev) return (ENXIO); } + /* + * Initialize the interrupt settings. The default is active-low + * interrupts. + */ + sc->sc_irq_trigger = malloc( + sizeof(*sc->sc_irq_trigger) * sc->sc_maxpin, + M_DEVBUF, M_WAITOK | M_ZERO); + sc->sc_irq_polarity = malloc( + sizeof(*sc->sc_irq_polarity) * sc->sc_maxpin, + M_DEVBUF, M_WAITOK | M_ZERO); + for (i = 0; i < sc->sc_maxpin; i++) { + sc->sc_irq_trigger[i] = INTR_TRIGGER_LEVEL; + sc->sc_irq_polarity[i] = INTR_POLARITY_LOW; + } + + sc->sc_events = malloc(sizeof(struct intr_event *) * sc->sc_maxpin, + M_DEVBUF, M_WAITOK | M_ZERO); + /* We need to go through each block and ensure the clocks are running and * the module is enabled. It might be better to do this only when the * pins are configured which would result in less power used if the GPIO @@ -825,6 +854,10 @@ ti_gpio_detach(device_t dev) bus_generic_detach(dev); + free(sc->sc_events, M_DEVBUF); + free(sc->sc_irq_polarity, M_DEVBUF); + free(sc->sc_irq_trigger, M_DEVBUF); + /* Release the memory and IRQ resources. */ ti_gpio_detach_intr(dev); bus_release_resources(dev, ti_gpio_irq_spec, sc->sc_irq_res); @@ -835,6 +868,193 @@ ti_gpio_detach(device_t dev) return (0); } +static uint32_t +ti_gpio_intr_reg(struct ti_gpio_softc *sc, int irq) +{ + + if (ti_gpio_valid_pin(sc, irq) != 0) + return (0); + + if (sc->sc_irq_trigger[irq] == INTR_TRIGGER_LEVEL) { + if (sc->sc_irq_polarity[irq] == INTR_POLARITY_LOW) + return (TI_GPIO_LEVELDETECT0); + else if (sc->sc_irq_polarity[irq] == INTR_POLARITY_HIGH) + return (TI_GPIO_LEVELDETECT1); + } else if (sc->sc_irq_trigger[irq] == INTR_TRIGGER_EDGE) { + if (sc->sc_irq_polarity[irq] == INTR_POLARITY_LOW) + return (TI_GPIO_FALLINGDETECT); + else if (sc->sc_irq_polarity[irq] == INTR_POLARITY_HIGH) + return (TI_GPIO_RISINGDETECT); + } + + return (0); +} + +static void +ti_gpio_mask_irq(void *source) +{ + int irq; + uint32_t reg, val; + + irq = (int)source; + if (ti_gpio_valid_pin(ti_gpio_sc, irq) != 0) + return; + + TI_GPIO_LOCK(ti_gpio_sc); + ti_gpio_intr_clr(ti_gpio_sc, TI_GPIO_BANK(irq), TI_GPIO_MASK(irq)); + reg = ti_gpio_intr_reg(ti_gpio_sc, irq); + if (reg != 0) { + val = ti_gpio_read_4(ti_gpio_sc, TI_GPIO_BANK(irq), reg); + val &= ~TI_GPIO_MASK(irq); + ti_gpio_write_4(ti_gpio_sc, TI_GPIO_BANK(irq), reg, val); + } + TI_GPIO_UNLOCK(ti_gpio_sc); +} + +static void +ti_gpio_unmask_irq(void *source) +{ + int irq; + uint32_t reg, val; + + irq = (int)source; + if (ti_gpio_valid_pin(ti_gpio_sc, irq) != 0) + return; + + TI_GPIO_LOCK(ti_gpio_sc); + reg = ti_gpio_intr_reg(ti_gpio_sc, irq); + if (reg != 0) { + val = ti_gpio_read_4(ti_gpio_sc, TI_GPIO_BANK(irq), reg); + val |= TI_GPIO_MASK(irq); + ti_gpio_write_4(ti_gpio_sc, TI_GPIO_BANK(irq), reg, val); + ti_gpio_intr_set(ti_gpio_sc, TI_GPIO_BANK(irq), + TI_GPIO_MASK(irq)); + } + TI_GPIO_UNLOCK(ti_gpio_sc); +} + +static int +ti_gpio_activate_resource(device_t dev, device_t child, int type, int rid, + struct resource *res) +{ + int pin; + + if (type != SYS_RES_IRQ) + return (ENXIO); + + /* Unmask the interrupt. */ + pin = rman_get_start(res); + ti_gpio_unmask_irq((void *)(uintptr_t)pin); + + return (0); +} + +static int +ti_gpio_deactivate_resource(device_t dev, device_t child, int type, int rid, + struct resource *res) +{ + int pin; + + if (type != SYS_RES_IRQ) + return (ENXIO); + + /* Mask the interrupt. */ + pin = rman_get_start(res); + ti_gpio_mask_irq((void *)(uintptr_t)pin); + + return (0); +} + +static int +ti_gpio_config_intr(device_t dev, int irq, enum intr_trigger trig, + enum intr_polarity pol) +{ + struct ti_gpio_softc *sc; + uint32_t oldreg, reg, val; + + sc = device_get_softc(dev); + if (ti_gpio_valid_pin(sc, irq) != 0) + return (EINVAL); + + /* There is no standard trigger or polarity. */ + if (trig == INTR_TRIGGER_CONFORM || pol == INTR_POLARITY_CONFORM) + return (EINVAL); + + TI_GPIO_LOCK(sc); + /* + * TRM recommends add the new event before remove the old one to + * avoid losing interrupts. + */ + oldreg = ti_gpio_intr_reg(sc, irq); + sc->sc_irq_trigger[irq] = trig; + sc->sc_irq_polarity[irq] = pol; + reg = ti_gpio_intr_reg(sc, irq); + if (reg != 0) { + /* Apply the new settings. */ + val = ti_gpio_read_4(sc, TI_GPIO_BANK(irq), reg); + val |= TI_GPIO_MASK(irq); + ti_gpio_write_4(sc, TI_GPIO_BANK(irq), reg, val); + } + if (oldreg != 0) { + /* Remove the old settings. */ + val = ti_gpio_read_4(sc, TI_GPIO_BANK(irq), oldreg); + val &= ~TI_GPIO_MASK(irq); + ti_gpio_write_4(sc, TI_GPIO_BANK(irq), oldreg, val); + } + TI_GPIO_UNLOCK(sc); + + return (0); +} + +static int +ti_gpio_setup_intr(device_t dev, device_t child, struct resource *ires, + int flags, driver_filter_t *filt, driver_intr_t *handler, + void *arg, void **cookiep) +{ + struct ti_gpio_softc *sc; + struct intr_event *event; + int pin, error; + + sc = device_get_softc(dev); + pin = rman_get_start(ires); + if (ti_gpio_valid_pin(sc, pin) != 0) + panic("%s: bad pin %d", __func__, pin); + + event = sc->sc_events[pin]; + if (event == NULL) { + error = intr_event_create(&event, (void *)(uintptr_t)pin, 0, + pin, ti_gpio_mask_irq, ti_gpio_unmask_irq, NULL, NULL, + "gpio%d pin%d:", device_get_unit(dev), pin); + if (error != 0) + return (error); + sc->sc_events[pin] = event; + } + intr_event_add_handler(event, device_get_nameunit(child), filt, + handler, arg, intr_priority(flags), flags, cookiep); + + return (0); +} + +static int +ti_gpio_teardown_intr(device_t dev, device_t child, struct resource *ires, + void *cookie) +{ + struct ti_gpio_softc *sc; + int pin, err; + + sc = device_get_softc(dev); + pin = rman_get_start(ires); + if (ti_gpio_valid_pin(sc, pin) != 0) + panic("%s: bad pin %d", __func__, pin); + if (sc->sc_events[pin] == NULL) + panic("Trying to teardown unoccupied IRQ"); + err = intr_event_remove_handler(cookie); + if (!err) + sc->sc_events[pin] = NULL; + + return (err); +} + static phandle_t ti_gpio_get_node(device_t bus, device_t dev) { @@ -857,6 +1077,13 @@ static device_method_t ti_gpio_methods[] = { DEVMETHOD(gpio_pin_set, ti_gpio_pin_set), DEVMETHOD(gpio_pin_toggle, ti_gpio_pin_toggle), + /* Bus interface */ + DEVMETHOD(bus_activate_resource, ti_gpio_activate_resource), + DEVMETHOD(bus_deactivate_resource, ti_gpio_deactivate_resource), + DEVMETHOD(bus_config_intr, ti_gpio_config_intr), + DEVMETHOD(bus_setup_intr, ti_gpio_setup_intr), + DEVMETHOD(bus_teardown_intr, ti_gpio_teardown_intr), + /* ofw_bus interface */ DEVMETHOD(ofw_bus_get_node, ti_gpio_get_node), diff --git a/sys/arm/ti/ti_gpio.h b/sys/arm/ti/ti_gpio.h index 0e1199f..763fc8a 100644 --- a/sys/arm/ti/ti_gpio.h +++ b/sys/arm/ti/ti_gpio.h @@ -47,6 +47,13 @@ struct ti_gpio_softc { device_t sc_dev; + /* Interrupt trigger type and level. */ + enum intr_trigger *sc_irq_trigger; + enum intr_polarity *sc_irq_polarity; + + int sc_maxpin; + struct mtx sc_mtx; + /* * The memory resource(s) for the PRCM register set, when the device is * created the caller can assign up to 6 memory regions depending on @@ -55,15 +62,11 @@ struct ti_gpio_softc { struct resource *sc_mem_res[MAX_GPIO_BANKS]; struct resource *sc_irq_res[MAX_GPIO_INTRS]; + /* Interrupt events. */ + struct intr_event **sc_events; + /* The handle for the register IRQ handlers. */ void *sc_irq_hdl[MAX_GPIO_INTRS]; - - /* - * The following describes the H/W revision of each of the GPIO banks. - */ - uint32_t sc_revision[MAX_GPIO_BANKS]; - - struct mtx sc_mtx; }; #endif /* TI_GPIO_H */ diff --git a/sys/arm/ti/ti_i2c.c b/sys/arm/ti/ti_i2c.c index 8f9af0c..d304827 100644 --- a/sys/arm/ti/ti_i2c.c +++ b/sys/arm/ti/ti_i2c.c @@ -119,7 +119,7 @@ struct ti_i2c_clock_config static struct ti_i2c_clock_config ti_omap4_i2c_clock_configs[] = { { 100000, 23, 13, 15, 0, 0}, { 400000, 9, 5, 7, 0, 0}, - { 1000000, 5, 1, 3, 0, 0}, + { 1000000, 3, 5, 7, 0, 0}, /* { 3200000, 1, 113, 115, 7, 10}, - HS mode */ { 0 /* Table terminator */ } }; diff --git a/sys/arm/ti/ti_sdhci.c b/sys/arm/ti/ti_sdhci.c index 024c830..380a40e 100644 --- a/sys/arm/ti/ti_sdhci.c +++ b/sys/arm/ti/ti_sdhci.c @@ -112,6 +112,7 @@ static struct ofw_compat_data compat_data[] = { #define MMCHS_CON 0x02C #define MMCHS_CON_DW8 (1 << 5) #define MMCHS_CON_DVAL_8_4MS (3 << 9) +#define MMCHS_CON_OD (1 << 0) #define MMCHS_SYSCTL 0x12C #define MMCHS_SYSCTL_CLKD_MASK 0x3FF #define MMCHS_SYSCTL_CLKD_SHIFT 6 @@ -327,7 +328,7 @@ ti_sdhci_update_ios(device_t brdev, device_t reqdev) struct ti_sdhci_softc *sc = device_get_softc(brdev); struct sdhci_slot *slot; struct mmc_ios *ios; - uint32_t val32; + uint32_t val32, newval32; slot = device_get_ivars(reqdev); ios = &slot->host.ios; @@ -339,10 +340,20 @@ ti_sdhci_update_ios(device_t brdev, device_t reqdev) * requested, then let the standard driver handle everything else. */ val32 = ti_mmchs_read_4(sc, MMCHS_CON); + newval32 = val32; + if (ios->bus_width == bus_width_8) - ti_mmchs_write_4(sc, MMCHS_CON, val32 | MMCHS_CON_DW8); + newval32 |= MMCHS_CON_DW8; else - ti_mmchs_write_4(sc, MMCHS_CON, val32 & ~MMCHS_CON_DW8); + newval32 &= ~MMCHS_CON_DW8; + + if (ios->bus_mode == opendrain) + newval32 |= MMCHS_CON_OD; + else /* if (ios->bus_mode == pushpull) */ + newval32 &= ~MMCHS_CON_OD; + + if (newval32 != val32) + ti_mmchs_write_4(sc, MMCHS_CON, newval32); return (sdhci_generic_update_ios(brdev, reqdev)); } @@ -392,20 +403,42 @@ ti_sdhci_hw_init(device_t dev) /* Issue a softreset to the controller */ ti_mmchs_write_4(sc, MMCHS_SYSCONFIG, MMCHS_SYSCONFIG_RESET); timeout = 1000; - while (!(ti_mmchs_read_4(sc, MMCHS_SYSSTATUS) & MMCHS_SYSSTATUS_RESETDONE)) { + while (!(ti_mmchs_read_4(sc, MMCHS_SYSSTATUS) & + MMCHS_SYSSTATUS_RESETDONE)) { if (--timeout == 0) { - device_printf(dev, "Error: Controller reset operation timed out\n"); + device_printf(dev, + "Error: Controller reset operation timed out\n"); break; } DELAY(100); } - /* Reset both the command and data state machines */ + /* + * Reset the command and data state machines and also other aspects of + * the controller such as bus clock and power. + * + * If we read the software reset register too fast after writing it we + * can get back a zero that means the reset hasn't started yet rather + * than that the reset is complete. Per TI recommendations, work around + * it by reading until we see the reset bit asserted, then read until + * it's clear. We also set the SDHCI_QUIRK_WAITFOR_RESET_ASSERTED quirk + * so that the main sdhci driver uses this same logic in its resets. + */ ti_sdhci_write_1(dev, NULL, SDHCI_SOFTWARE_RESET, SDHCI_RESET_ALL); - timeout = 1000; - while ((ti_sdhci_read_1(dev, NULL, SDHCI_SOFTWARE_RESET) & SDHCI_RESET_ALL)) { + timeout = 10000; + while ((ti_sdhci_read_1(dev, NULL, SDHCI_SOFTWARE_RESET) & + SDHCI_RESET_ALL) != SDHCI_RESET_ALL) { if (--timeout == 0) { - device_printf(dev, "Error: Software reset operation timed out\n"); + break; + } + DELAY(1); + } + timeout = 10000; + while ((ti_sdhci_read_1(dev, NULL, SDHCI_SOFTWARE_RESET) & + SDHCI_RESET_ALL)) { + if (--timeout == 0) { + device_printf(dev, + "Error: Software reset operation timed out\n"); break; } DELAY(100); @@ -568,6 +601,12 @@ ti_sdhci_attach(device_t dev) sc->slot.quirks |= SDHCI_QUIRK_DONT_SHIFT_RESPONSE; /* + * Reset bits are broken, have to wait to see the bits asserted + * before waiting to see them de-asserted. + */ + sc->slot.quirks |= SDHCI_QUIRK_WAITFOR_RESET_ASSERTED; + + /* * DMA is not really broken, I just haven't implemented it yet. */ sc->slot.quirks |= SDHCI_QUIRK_BROKEN_DMA; diff --git a/sys/arm/ti/ti_wdt.c b/sys/arm/ti/ti_wdt.c index 7f50a61..7e8e5ae 100644 --- a/sys/arm/ti/ti_wdt.c +++ b/sys/arm/ti/ti_wdt.c @@ -95,7 +95,7 @@ static devclass_t ti_wdt_devclass; DRIVER_MODULE(ti_wdt, simplebus, ti_wdt_driver, ti_wdt_devclass, 0, 0); -static volatile __inline uint32_t +static __inline uint32_t ti_wdt_reg_read(struct ti_wdt_softc *sc, uint32_t reg) { diff --git a/sys/arm/xilinx/std.zynq7 b/sys/arm/xilinx/std.zynq7 index 1cee32d..34e6739 100644 --- a/sys/arm/xilinx/std.zynq7 +++ b/sys/arm/xilinx/std.zynq7 @@ -5,6 +5,7 @@ cpu CPU_CORTEXA machine arm armv6 +makeoptions CONF_CFLAGS="-march=armv7a -Wa,-march=armv7a" files "../xilinx/files.zynq7" diff --git a/sys/boot/amd64/boot1.efi/Makefile b/sys/boot/amd64/boot1.efi/Makefile index 3e3bc9f..dd8eaf9 100644 --- a/sys/boot/amd64/boot1.efi/Makefile +++ b/sys/boot/amd64/boot1.efi/Makefile @@ -51,7 +51,7 @@ boot1.efi: loader.sym ${OBJCOPY} -j .text -j .sdata -j .data \ -j .dynamic -j .dynsym -j .rel.dyn \ -j .rela.dyn -j .reloc -j .eh_frame -j set_Xcommand_set \ - --target=${EFI_TARGET} ${.ALLSRC} ${.TARGET} + --output-target=${EFI_TARGET} ${.ALLSRC} ${.TARGET} CFLAGS+= -I${.CURDIR}/../../common diff --git a/sys/boot/amd64/boot1.efi/generate-fat.sh b/sys/boot/amd64/boot1.efi/generate-fat.sh index eafd3e0..e8f9cde 100755 --- a/sys/boot/amd64/boot1.efi/generate-fat.sh +++ b/sys/boot/amd64/boot1.efi/generate-fat.sh @@ -34,10 +34,10 @@ umount stub mdconfig -d -u $DEVICE rmdir stub -# Locate the offsets of the two fake files +# Locate the offset of the fake file BOOT1_OFFSET=$(hd $OUTPUT_FILE | grep 'Boot1 START' | cut -f 1 -d ' ') -# Convert to numbers of blocks +# Convert to number of blocks BOOT1_OFFSET=$(echo 0x$BOOT1_OFFSET | awk '{printf("%x\n",$1/512);}') echo '# This file autogenerated by generate-fat.sh - DO NOT EDIT' > Makefile.fat diff --git a/sys/boot/amd64/efi/Makefile b/sys/boot/amd64/efi/Makefile index a9a2132..66b8cf8 100644 --- a/sys/boot/amd64/efi/Makefile +++ b/sys/boot/amd64/efi/Makefile @@ -81,7 +81,7 @@ loader.efi: loader.sym ${OBJCOPY} -j .text -j .sdata -j .data \ -j .dynamic -j .dynsym -j .rel.dyn \ -j .rela.dyn -j .reloc -j .eh_frame -j set_Xcommand_set \ - --target=${EFI_TARGET} ${.ALLSRC} ${.TARGET} + --output-target=${EFI_TARGET} ${.ALLSRC} ${.TARGET} LIBEFI= ${.OBJDIR}/../../efi/libefi/libefi.a CFLAGS+= -I${.CURDIR}/../../common diff --git a/sys/boot/arm/uboot/Makefile b/sys/boot/arm/uboot/Makefile index 743f189..c69d8aa 100644 --- a/sys/boot/arm/uboot/Makefile +++ b/sys/boot/arm/uboot/Makefile @@ -76,7 +76,7 @@ LIBUBOOT_FDT= ${.OBJDIR}/../../uboot/fdt/libuboot_fdt.a LIBFDT= ${.OBJDIR}/../../fdt/libfdt.a .endif -.if !defined(NO_FORTH) +.if ${MK_FORTH} != "no" # Enable BootForth BOOT_FORTH= yes CFLAGS+= -DBOOT_FORTH -I${.CURDIR}/../../ficl -I${.CURDIR}/../../ficl/arm diff --git a/sys/boot/common/console.c b/sys/boot/common/console.c index 6c1fdab..6656eab 100644 --- a/sys/boot/common/console.c +++ b/sys/boot/common/console.c @@ -39,6 +39,7 @@ static int cons_set(struct env_var *ev, int flags, const void *value); static int cons_find(const char *name); static int cons_check(const char *string); static void cons_change(const char *string); +static int twiddle_set(struct env_var *ev, int flags, const void *value); /* * Detect possible console(s) to use. If preferred console(s) have been @@ -52,6 +53,9 @@ cons_probe(void) int active; char *prefconsole; + /* We want a callback to install the new value when this var changes. */ + env_setenv("twiddle_divisor", EV_VOLATILE, "1", twiddle_set, env_nounset); + /* Do all console probes */ for (cons = 0; consoles[cons] != NULL; cons++) { consoles[cons]->c_flags = 0; @@ -232,3 +236,28 @@ cons_change(const char *string) free(dup); } + +/* + * Change the twiddle divisor. + * + * The user can set the twiddle_divisor variable to directly control how fast + * the progress twiddle spins, useful for folks with slow serial consoles. The + * code to monitor changes to the variable and propagate them to the twiddle + * routines has to live somewhere. Twiddling is console-related so it's here. + */ +static int +twiddle_set(struct env_var *ev, int flags, const void *value) +{ + u_long tdiv; + char * eptr; + + tdiv = strtoul(value, &eptr, 0); + if (*(const char *)value == 0 || *eptr != 0) { + printf("invalid twiddle_divisor '%s'\n", (const char *)value); + return (CMD_ERROR); + } + twiddle_divisor((u_int)tdiv); + env_setenv(ev->ev_name, flags | EV_NOHOOK, value, NULL, NULL); + + return(CMD_OK); +} diff --git a/sys/boot/common/loader.8 b/sys/boot/common/loader.8 index fe247ac..cac5674 100644 --- a/sys/boot/common/loader.8 +++ b/sys/boot/common/loader.8 @@ -24,7 +24,7 @@ .\" .\" $FreeBSD$ .\" -.Dd October 1, 2013 +.Dd December 22, 2014 .Dt LOADER 8 .Os .Sh NAME @@ -670,6 +670,12 @@ Overrides the compile-time set value of .Dv TCBHASHSIZE or the preset default of 512. Must be a power of 2. +.It Va twiddle_divisor +Throttles the output of the `twiddle' I/O progress indicator displayed +while loading the kernel and modules. +This is useful on slow serial consoles where the time spent waiting for +these characters to be written can add up to many seconds. +The default is 1 (full speed); a value of 2 spins half as fast, and so on. .It Va vm.kmem_size Sets the size of kernel memory (bytes). This overrides the value determined when the kernel was compiled. diff --git a/sys/boot/fdt/dts/arm/am335x.dtsi b/sys/boot/fdt/dts/arm/am335x.dtsi index 9378f34..feb32f2 100644 --- a/sys/boot/fdt/dts/arm/am335x.dtsi +++ b/sys/boot/fdt/dts/arm/am335x.dtsi @@ -99,6 +99,8 @@ 0x481AE000 0x1000 >; interrupts = < 96 97 98 99 32 33 62 63 >; interrupt-parent = <&AINTC>; + interrupt-controller; + #interrupt-cells = <1>; }; uart0: serial@44E09000 { diff --git a/sys/boot/fdt/dts/arm/rpi.dts b/sys/boot/fdt/dts/arm/rpi.dts index 58ceaa4..08d9d24 100644 --- a/sys/boot/fdt/dts/arm/rpi.dts +++ b/sys/boot/fdt/dts/arm/rpi.dts @@ -35,6 +35,17 @@ memreserve = <0x08000000 0x08000000>; /* Set by VideoCore */ + cpus { + #address-cells = <1>; + #size-cells = <0>; + cpu@0 { + compatible = "arm,1176jzf-s"; + device_type = "cpu"; + reg = <0>; /* CPU ID=0 */ + clock-frequency = <700000000>; /* 700MHz */ + }; + }; + memory { device_type = "memory"; reg = <0 0x8000000>; /* 128MB, Set by VideoCore */ diff --git a/sys/boot/forth/brand.4th.8 b/sys/boot/forth/brand.4th.8 index 64e6854..dfd188d 100644 --- a/sys/boot/forth/brand.4th.8 +++ b/sys/boot/forth/brand.4th.8 @@ -111,7 +111,7 @@ loader_brand="fbsd" .Ed .Sh SEE ALSO .Xr loader.conf 5 , -.Xr loader 8 , +.Xr loader 8 .Sh HISTORY The .Nm diff --git a/sys/boot/forth/delay.4th.8 b/sys/boot/forth/delay.4th.8 index dd1680e..af31fd0 100644 --- a/sys/boot/forth/delay.4th.8 +++ b/sys/boot/forth/delay.4th.8 @@ -110,8 +110,8 @@ delay_execute .Ed .Sh SEE ALSO .Xr loader.conf 5 , -.Xr loader 8 , .Xr beastie.4th 8 , +.Xr loader 8 , .Xr loader.4th 8 .Sh HISTORY The diff --git a/sys/boot/forth/loader.conf b/sys/boot/forth/loader.conf index bd7d296..fd0b0de 100644 --- a/sys/boot/forth/loader.conf +++ b/sys/boot/forth/loader.conf @@ -75,6 +75,7 @@ module_path="/boot/modules" # Set the module search path # the block size is set to 512. If the value # is out of range ( < 8 || > 9008 ) an error is # returned. +#twiddle_divisor="1" # >1 means slow down the progress indicator. ############################################################## diff --git a/sys/boot/forth/menu.4th.8 b/sys/boot/forth/menu.4th.8 index ab67d73..3673eec 100644 --- a/sys/boot/forth/menu.4th.8 +++ b/sys/boot/forth/menu.4th.8 @@ -336,9 +336,9 @@ menu-display .Ed .Sh SEE ALSO .Xr loader.conf 5 , +.Xr beastie.4th 8 , .Xr loader 8 , -.Xr loader.4th 8 , -.Xr beastie.4th 8 +.Xr loader.4th 8 .Sh HISTORY The .Nm diff --git a/sys/boot/forth/menusets.4th.8 b/sys/boot/forth/menusets.4th.8 index e05e4a7..f785ae1 100644 --- a/sys/boot/forth/menusets.4th.8 +++ b/sys/boot/forth/menusets.4th.8 @@ -355,10 +355,10 @@ set submenu_command[1]="1 goto_menu" .Ed .Sh SEE ALSO .Xr loader.conf 5 , +.Xr beastie.4th 8 , .Xr loader 8 , .Xr loader.4th 8 , -.Xr menu.4th 8 , -.Xr beastie.4th 8 +.Xr menu.4th 8 .Sh HISTORY The .Nm diff --git a/sys/boot/forth/version.4th.8 b/sys/boot/forth/version.4th.8 index fe8b618..256df1e 100644 --- a/sys/boot/forth/version.4th.8 +++ b/sys/boot/forth/version.4th.8 @@ -113,8 +113,8 @@ loader_version="loader 1.1" .Ed .Sh SEE ALSO .Xr loader.conf 5 , -.Xr loader 8 , -.Xr color.4th 8 +.Xr color.4th 8 , +.Xr loader 8 .Sh HISTORY The .Nm diff --git a/sys/boot/i386/libi386/spinconsole.c b/sys/boot/i386/libi386/spinconsole.c index 752c29f..161d810 100644 --- a/sys/boot/i386/libi386/spinconsole.c +++ b/sys/boot/i386/libi386/spinconsole.c @@ -86,9 +86,11 @@ spinc_putchar(int c) if (now < (lasttime + 1)) return; lasttime = now; +#ifdef TERM_EMU get_pos(&curx, &cury); if (curx > 0) curs_move(&curx, &cury, curx - 1, cury); +#endif vidc_biosputchar((char)tw_chars); tw_chars = (tw_chars >> 8) | ((tw_chars & (unsigned long)0xFF) << 24); } diff --git a/sys/boot/powerpc/uboot/Makefile b/sys/boot/powerpc/uboot/Makefile index a5a06a9..681249a 100644 --- a/sys/boot/powerpc/uboot/Makefile +++ b/sys/boot/powerpc/uboot/Makefile @@ -66,7 +66,7 @@ LIBUBOOT_FDT= ${.OBJDIR}/../../uboot/fdt/libuboot_fdt.a LIBFDT= ${.OBJDIR}/../../fdt/libfdt.a .endif -.if !defined(NO_FORTH) +.if ${MK_FORTH} != "no" # Enable BootForth BOOT_FORTH= yes CFLAGS+= -DBOOT_FORTH -I${.CURDIR}/../../ficl -I${.CURDIR}/../../ficl/powerpc diff --git a/sys/boot/uboot/common/main.c b/sys/boot/uboot/common/main.c index ae77766..f58d0c4 100644 --- a/sys/boot/uboot/common/main.c +++ b/sys/boot/uboot/common/main.c @@ -212,10 +212,11 @@ get_load_device(int *type, int *unit, int *slice, int *partition) p = get_device_type(devstr, type); - /* - * Empty device string, or unknown device name, or a bare, known - * device name. - */ + /* Ignore optional spaces after the device name. */ + while (*p == ' ') + p++; + + /* Unknown device name, or a known name without unit number. */ if ((*type == -1) || (*p == '\0')) { return; } diff --git a/sys/cam/ctl/ctl.c b/sys/cam/ctl/ctl.c index 8fccd02..e90a029 100644 --- a/sys/cam/ctl/ctl.c +++ b/sys/cam/ctl/ctl.c @@ -111,7 +111,7 @@ struct ctl_softc *control_softc = NULL; * Note that these are default values only. The actual values will be * filled in when the user does a mode sense. */ -static struct copan_debugconf_subpage debugconf_page_default = { +const static struct copan_debugconf_subpage debugconf_page_default = { DBGCNF_PAGE_CODE | SMPH_SPF, /* page_code */ DBGCNF_SUBPAGE_CODE, /* subpage */ {(sizeof(struct copan_debugconf_subpage) - 4) >> 8, @@ -121,7 +121,7 @@ static struct copan_debugconf_subpage debugconf_page_default = { CTL_TIME_IO_DEFAULT_SECS>>0}, /* ctl_time_io_secs */ }; -static struct copan_debugconf_subpage debugconf_page_changeable = { +const static struct copan_debugconf_subpage debugconf_page_changeable = { DBGCNF_PAGE_CODE | SMPH_SPF, /* page_code */ DBGCNF_SUBPAGE_CODE, /* subpage */ {(sizeof(struct copan_debugconf_subpage) - 4) >> 8, @@ -130,7 +130,7 @@ static struct copan_debugconf_subpage debugconf_page_changeable = { {0xff,0xff}, /* ctl_time_io_secs */ }; -static struct scsi_da_rw_recovery_page rw_er_page_default = { +const static struct scsi_da_rw_recovery_page rw_er_page_default = { /*page_code*/SMS_RW_ERROR_RECOVERY_PAGE, /*page_length*/sizeof(struct scsi_da_rw_recovery_page) - 2, /*byte3*/SMS_RWER_AWRE|SMS_RWER_ARRE, @@ -144,7 +144,7 @@ static struct scsi_da_rw_recovery_page rw_er_page_default = { /*recovery_time_limit*/{0, 0}, }; -static struct scsi_da_rw_recovery_page rw_er_page_changeable = { +const static struct scsi_da_rw_recovery_page rw_er_page_changeable = { /*page_code*/SMS_RW_ERROR_RECOVERY_PAGE, /*page_length*/sizeof(struct scsi_da_rw_recovery_page) - 2, /*byte3*/0, @@ -158,7 +158,7 @@ static struct scsi_da_rw_recovery_page rw_er_page_changeable = { /*recovery_time_limit*/{0, 0}, }; -static struct scsi_format_page format_page_default = { +const static struct scsi_format_page format_page_default = { /*page_code*/SMS_FORMAT_DEVICE_PAGE, /*page_length*/sizeof(struct scsi_format_page) - 2, /*tracks_per_zone*/ {0, 0}, @@ -175,7 +175,7 @@ static struct scsi_format_page format_page_default = { /*reserved*/ {0, 0, 0} }; -static struct scsi_format_page format_page_changeable = { +const static struct scsi_format_page format_page_changeable = { /*page_code*/SMS_FORMAT_DEVICE_PAGE, /*page_length*/sizeof(struct scsi_format_page) - 2, /*tracks_per_zone*/ {0, 0}, @@ -191,7 +191,7 @@ static struct scsi_format_page format_page_changeable = { /*reserved*/ {0, 0, 0} }; -static struct scsi_rigid_disk_page rigid_disk_page_default = { +const static struct scsi_rigid_disk_page rigid_disk_page_default = { /*page_code*/SMS_RIGID_DISK_PAGE, /*page_length*/sizeof(struct scsi_rigid_disk_page) - 2, /*cylinders*/ {0, 0, 0}, @@ -208,7 +208,7 @@ static struct scsi_rigid_disk_page rigid_disk_page_default = { /*reserved2*/ {0, 0} }; -static struct scsi_rigid_disk_page rigid_disk_page_changeable = { +const static struct scsi_rigid_disk_page rigid_disk_page_changeable = { /*page_code*/SMS_RIGID_DISK_PAGE, /*page_length*/sizeof(struct scsi_rigid_disk_page) - 2, /*cylinders*/ {0, 0, 0}, @@ -224,7 +224,7 @@ static struct scsi_rigid_disk_page rigid_disk_page_changeable = { /*reserved2*/ {0, 0} }; -static struct scsi_caching_page caching_page_default = { +const static struct scsi_caching_page caching_page_default = { /*page_code*/SMS_CACHING_PAGE, /*page_length*/sizeof(struct scsi_caching_page) - 2, /*flags1*/ SCP_DISC | SCP_WCE, @@ -240,7 +240,7 @@ static struct scsi_caching_page caching_page_default = { /*non_cache_seg_size*/ {0, 0, 0} }; -static struct scsi_caching_page caching_page_changeable = { +const static struct scsi_caching_page caching_page_changeable = { /*page_code*/SMS_CACHING_PAGE, /*page_length*/sizeof(struct scsi_caching_page) - 2, /*flags1*/ SCP_WCE | SCP_RCD, @@ -256,7 +256,7 @@ static struct scsi_caching_page caching_page_changeable = { /*non_cache_seg_size*/ {0, 0, 0} }; -static struct scsi_control_page control_page_default = { +const static struct scsi_control_page control_page_default = { /*page_code*/SMS_CONTROL_MODE_PAGE, /*page_length*/sizeof(struct scsi_control_page) - 2, /*rlec*/0, @@ -268,7 +268,7 @@ static struct scsi_control_page control_page_default = { /*extended_selftest_completion_time*/{0, 0} }; -static struct scsi_control_page control_page_changeable = { +const static struct scsi_control_page control_page_changeable = { /*page_code*/SMS_CONTROL_MODE_PAGE, /*page_length*/sizeof(struct scsi_control_page) - 2, /*rlec*/SCP_DSENSE, @@ -280,7 +280,7 @@ static struct scsi_control_page control_page_changeable = { /*extended_selftest_completion_time*/{0, 0} }; -static struct scsi_info_exceptions_page ie_page_default = { +const static struct scsi_info_exceptions_page ie_page_default = { /*page_code*/SMS_INFO_EXCEPTIONS_PAGE, /*page_length*/sizeof(struct scsi_info_exceptions_page) - 2, /*info_flags*/SIEP_FLAGS_DEXCPT, @@ -289,7 +289,7 @@ static struct scsi_info_exceptions_page ie_page_default = { /*report_count*/{0, 0, 0, 0} }; -static struct scsi_info_exceptions_page ie_page_changeable = { +const static struct scsi_info_exceptions_page ie_page_changeable = { /*page_code*/SMS_INFO_EXCEPTIONS_PAGE, /*page_length*/sizeof(struct scsi_info_exceptions_page) - 2, /*info_flags*/0, @@ -300,7 +300,7 @@ static struct scsi_info_exceptions_page ie_page_changeable = { #define CTL_LBPM_LEN (sizeof(struct ctl_logical_block_provisioning_page) - 4) -static struct ctl_logical_block_provisioning_page lbp_page_default = {{ +const static struct ctl_logical_block_provisioning_page lbp_page_default = {{ /*page_code*/SMS_INFO_EXCEPTIONS_PAGE | SMPH_SPF, /*subpage_code*/0x02, /*page_length*/{CTL_LBPM_LEN >> 8, CTL_LBPM_LEN}, @@ -326,7 +326,7 @@ static struct ctl_logical_block_provisioning_page lbp_page_default = {{ } }; -static struct ctl_logical_block_provisioning_page lbp_page_changeable = {{ +const static struct ctl_logical_block_provisioning_page lbp_page_changeable = {{ /*page_code*/SMS_INFO_EXCEPTIONS_PAGE | SMPH_SPF, /*subpage_code*/0x02, /*page_length*/{CTL_LBPM_LEN >> 8, CTL_LBPM_LEN}, @@ -398,8 +398,8 @@ static int ctl_ioctl_fill_ooa(struct ctl_lun *lun, uint32_t *cur_fill_num, struct ctl_ooa_entry *kern_entries); static int ctl_ioctl(struct cdev *dev, u_long cmd, caddr_t addr, int flag, struct thread *td); -static uint32_t ctl_map_lun(int port_num, uint32_t lun); -static uint32_t ctl_map_lun_back(int port_num, uint32_t lun); +static uint32_t ctl_map_lun(struct ctl_softc *softc, int port_num, uint32_t lun); +static uint32_t ctl_map_lun_back(struct ctl_softc *softc, int port_num, uint32_t lun); static int ctl_alloc_lun(struct ctl_softc *ctl_softc, struct ctl_lun *lun, struct ctl_be_lun *be_lun, struct ctl_id target_id); static int ctl_free_lun(struct ctl_lun *lun); @@ -441,8 +441,7 @@ static ctl_action ctl_check_for_blockage(struct ctl_lun *lun, static ctl_action ctl_check_ooa(struct ctl_lun *lun, union ctl_io *pending_io, union ctl_io *starting_io); static int ctl_check_blocked(struct ctl_lun *lun); -static int ctl_scsiio_lun_check(struct ctl_softc *ctl_softc, - struct ctl_lun *lun, +static int ctl_scsiio_lun_check(struct ctl_lun *lun, const struct ctl_cmd_entry *entry, struct ctl_scsiio *ctsio); //static int ctl_check_rtr(union ctl_io *pending_io, struct ctl_softc *softc); @@ -609,12 +608,12 @@ ctl_isc_handler_finish_ser_only(struct ctl_softc *ctl_softc, static void ctl_isc_event_handler(ctl_ha_channel channel, ctl_ha_event event, int param) { - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; union ctl_io *io; struct ctl_prio *presio; ctl_ha_status isc_status; - ctl_softc = control_softc; + softc = control_softc; io = NULL; @@ -640,7 +639,7 @@ ctl_isc_event_handler(ctl_ha_channel channel, ctl_ha_event event, int param) #if 0 printf("Serialize\n"); #endif - io = ctl_alloc_io_nowait(ctl_softc->othersc_pool); + io = ctl_alloc_io_nowait(softc->othersc_pool); if (io == NULL) { printf("ctl_isc_event_handler: can't allocate " "ctl_io!\n"); @@ -672,7 +671,7 @@ ctl_isc_event_handler(ctl_ha_channel channel, ctl_ha_event event, int param) * * XXX KDM add another flag that is more specific. */ - if (ctl_softc->ha_mode == CTL_HA_MODE_SER_ONLY) + if (softc->ha_mode == CTL_HA_MODE_SER_ONLY) io->io_hdr.flags |= CTL_FLAG_INT_COPY; io->io_hdr.nexus = msg_info.hdr.nexus; #if 0 @@ -686,7 +685,7 @@ ctl_isc_event_handler(ctl_ha_channel channel, ctl_ha_event event, int param) io->scsiio.tag_type = msg_info.scsi.tag_type; memcpy(io->scsiio.cdb, msg_info.scsi.cdb, CTL_MAX_CDBLEN); - if (ctl_softc->ha_mode == CTL_HA_MODE_XFER) { + if (softc->ha_mode == CTL_HA_MODE_XFER) { const struct ctl_cmd_entry *entry; entry = ctl_get_cmd_entry(&io->scsiio, NULL); @@ -852,11 +851,11 @@ ctl_isc_event_handler(ctl_ha_channel channel, ctl_ha_event event, int param) * mode */ case CTL_MSG_FINISH_IO: - if (ctl_softc->ha_mode == CTL_HA_MODE_XFER) - ctl_isc_handler_finish_xfer(ctl_softc, + if (softc->ha_mode == CTL_HA_MODE_XFER) + ctl_isc_handler_finish_xfer(softc, &msg_info); else - ctl_isc_handler_finish_ser_only(ctl_softc, + ctl_isc_handler_finish_ser_only(softc, &msg_info); break; @@ -886,7 +885,7 @@ ctl_isc_event_handler(ctl_ha_channel channel, ctl_ha_event event, int param) case CTL_MSG_MANAGE_TASKS: { struct ctl_taskio *taskio; taskio = (struct ctl_taskio *)ctl_alloc_io_nowait( - ctl_softc->othersc_pool); + softc->othersc_pool); if (taskio == NULL) { printf("ctl_isc_event_handler: can't allocate " "ctl_io!\n"); @@ -915,7 +914,7 @@ ctl_isc_event_handler(ctl_ha_channel channel, ctl_ha_event event, int param) /* Persistent Reserve action which needs attention */ case CTL_MSG_PERS_ACTION: presio = (struct ctl_prio *)ctl_alloc_io_nowait( - ctl_softc->othersc_pool); + softc->othersc_pool); if (presio == NULL) { printf("ctl_isc_event_handler: can't allocate " "ctl_io!\n"); @@ -1744,8 +1743,8 @@ ctl_ioctl_do_datamove(struct ctl_scsiio *ctsio) i < ext_sg_entries && j < kern_sg_entries;) { uint8_t *ext_ptr, *kern_ptr; - len_to_copy = ctl_min(ext_sglist[i].len - ext_watermark, - kern_sglist[j].len - kern_watermark); + len_to_copy = MIN(ext_sglist[i].len - ext_watermark, + kern_sglist[j].len - kern_watermark); ext_ptr = (uint8_t *)ext_sglist[i].addr; ext_ptr = ext_ptr + ext_watermark; @@ -1832,16 +1831,16 @@ bailout: static int ctl_serialize_other_sc_cmd(struct ctl_scsiio *ctsio) { - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; union ctl_ha_msg msg_info; struct ctl_lun *lun; int retval = 0; uint32_t targ_lun; - ctl_softc = control_softc; + softc = control_softc; targ_lun = ctsio->io_hdr.nexus.targ_mapped_lun; - lun = ctl_softc->ctl_luns[targ_lun]; + lun = softc->ctl_luns[targ_lun]; if (lun==NULL) { /* @@ -1886,7 +1885,7 @@ ctl_serialize_other_sc_cmd(struct ctl_scsiio *ctsio) break; case CTL_ACTION_PASS: case CTL_ACTION_SKIP: - if (ctl_softc->ha_mode == CTL_HA_MODE_XFER) { + if (softc->ha_mode == CTL_HA_MODE_XFER) { ctsio->io_hdr.flags |= CTL_FLAG_IS_WAS_ON_RTR; ctl_enqueue_rtr((union ctl_io *)ctsio); } else { @@ -2835,8 +2834,8 @@ ctl_ioctl(struct cdev *dev, u_long cmd, caddr_t addr, int flag, bbr_info->scsi_status = metatask->taskinfo.bbrread.scsi_status; memcpy(&bbr_info->sense_data, &metatask->taskinfo.bbrread.sense_data, - ctl_min(sizeof(bbr_info->sense_data), - sizeof(metatask->taskinfo.bbrread.sense_data))); + MIN(sizeof(bbr_info->sense_data), + sizeof(metatask->taskinfo.bbrread.sense_data))); cfi_free_metatask(metatask); @@ -3498,11 +3497,11 @@ ctl_ioctl(struct cdev *dev, u_long cmd, caddr_t addr, int flag, if (port->wwpn_iid[j].name != NULL) retval = sbuf_printf(sb, - "\t<initiator>%u %s</initiator>\n", + "\t<initiator id=\"%u\">%s</initiator>\n", j, port->wwpn_iid[j].name); else retval = sbuf_printf(sb, - "\t<initiator>%u naa.%08jx</initiator>\n", + "\t<initiator id=\"%u\">naa.%08jx</initiator>\n", j, port->wwpn_iid[j].wwpn); if (retval != 0) break; @@ -3602,11 +3601,11 @@ ctl_port_idx(int port_num) } static uint32_t -ctl_map_lun(int port_num, uint32_t lun_id) +ctl_map_lun(struct ctl_softc *softc, int port_num, uint32_t lun_id) { struct ctl_port *port; - port = control_softc->ctl_ports[ctl_port_idx(port_num)]; + port = softc->ctl_ports[ctl_port_idx(port_num)]; if (port == NULL) return (UINT32_MAX); if (port->lun_map == NULL) @@ -3615,12 +3614,12 @@ ctl_map_lun(int port_num, uint32_t lun_id) } static uint32_t -ctl_map_lun_back(int port_num, uint32_t lun_id) +ctl_map_lun_back(struct ctl_softc *softc, int port_num, uint32_t lun_id) { struct ctl_port *port; uint32_t i; - port = control_softc->ctl_ports[ctl_port_idx(port_num)]; + port = softc->ctl_ports[ctl_port_idx(port_num)]; if (port->lun_map == NULL) return (lun_id); for (i = 0; i < CTL_MAX_LUNS; i++) { @@ -3643,7 +3642,7 @@ ctl_ffz(uint32_t *mask, uint32_t size) num_chunks = (size >> 5); if (num_chunks == 0) num_chunks++; - num_pieces = ctl_min((sizeof(uint32_t) * 8), size); + num_pieces = MIN((sizeof(uint32_t) * 8), size); for (i = 0; i < num_chunks; i++) { for (j = 0; j < num_pieces; j++) { @@ -3890,7 +3889,7 @@ ctl_copy_io(union ctl_io *src, union ctl_io *dest) */ pool_ref = dest->io_hdr.pool; - memcpy(dest, src, ctl_min(sizeof(*src), sizeof(*dest))); + memcpy(dest, src, MIN(sizeof(*src), sizeof(*dest))); dest->io_hdr.pool = pool_ref; /* @@ -4777,25 +4776,25 @@ ctl_free_lun(struct ctl_lun *lun) static void ctl_create_lun(struct ctl_be_lun *be_lun) { - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; - ctl_softc = control_softc; + softc = control_softc; /* * ctl_alloc_lun() should handle all potential failure cases. */ - ctl_alloc_lun(ctl_softc, NULL, be_lun, ctl_softc->target); + ctl_alloc_lun(softc, NULL, be_lun, softc->target); } int ctl_add_lun(struct ctl_be_lun *be_lun) { - struct ctl_softc *ctl_softc = control_softc; + struct ctl_softc *softc = control_softc; - mtx_lock(&ctl_softc->ctl_lock); - STAILQ_INSERT_TAIL(&ctl_softc->pending_lun_queue, be_lun, links); - mtx_unlock(&ctl_softc->ctl_lock); - wakeup(&ctl_softc->pending_lun_queue); + mtx_lock(&softc->ctl_lock); + STAILQ_INSERT_TAIL(&softc->pending_lun_queue, be_lun, links); + mtx_unlock(&softc->ctl_lock); + wakeup(&softc->pending_lun_queue); return (0); } @@ -4803,16 +4802,15 @@ ctl_add_lun(struct ctl_be_lun *be_lun) int ctl_enable_lun(struct ctl_be_lun *be_lun) { - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; struct ctl_port *port, *nport; struct ctl_lun *lun; int retval; - ctl_softc = control_softc; - lun = (struct ctl_lun *)be_lun->ctl_lun; + softc = lun->ctl_softc; - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); mtx_lock(&lun->lun_lock); if ((lun->flags & CTL_LUN_DISABLED) == 0) { /* @@ -4820,13 +4818,13 @@ ctl_enable_lun(struct ctl_be_lun *be_lun) * enabled? */ mtx_unlock(&lun->lun_lock); - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (0); } lun->flags &= ~CTL_LUN_DISABLED; mtx_unlock(&lun->lun_lock); - for (port = STAILQ_FIRST(&ctl_softc->port_list); port != NULL; port = nport) { + for (port = STAILQ_FIRST(&softc->port_list); port != NULL; port = nport) { nport = STAILQ_NEXT(port, links); /* @@ -4834,9 +4832,9 @@ ctl_enable_lun(struct ctl_be_lun *be_lun) * This can lead to a callback into CTL (at least in the * case of the internal initiator frontend. */ - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); retval = port->lun_enable(port->targ_lun_arg, lun->target,lun->lun); - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); if (retval != 0) { printf("%s: FETD %s port %d returned error " "%d for lun_enable on target %ju lun %jd\n", @@ -4851,7 +4849,7 @@ ctl_enable_lun(struct ctl_be_lun *be_lun) #endif } - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (0); } @@ -4859,27 +4857,26 @@ ctl_enable_lun(struct ctl_be_lun *be_lun) int ctl_disable_lun(struct ctl_be_lun *be_lun) { - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; struct ctl_port *port; struct ctl_lun *lun; int retval; - ctl_softc = control_softc; - lun = (struct ctl_lun *)be_lun->ctl_lun; + softc = lun->ctl_softc; - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); mtx_lock(&lun->lun_lock); if (lun->flags & CTL_LUN_DISABLED) { mtx_unlock(&lun->lun_lock); - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (0); } lun->flags |= CTL_LUN_DISABLED; mtx_unlock(&lun->lun_lock); - STAILQ_FOREACH(port, &ctl_softc->port_list, links) { - mtx_unlock(&ctl_softc->ctl_lock); + STAILQ_FOREACH(port, &softc->port_list, links) { + mtx_unlock(&softc->ctl_lock); /* * Drop the lock before we call the frontend's disable * routine, to avoid lock order reversals. @@ -4889,7 +4886,7 @@ ctl_disable_lun(struct ctl_be_lun *be_lun) */ retval = port->lun_disable(port->targ_lun_arg, lun->target, lun->lun); - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); if (retval != 0) { printf("ctl_alloc_lun: FETD %s port %d returned error " "%d for lun_disable on target %ju lun %jd\n", @@ -4898,7 +4895,7 @@ ctl_disable_lun(struct ctl_be_lun *be_lun) } } - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (0); } @@ -4906,80 +4903,55 @@ ctl_disable_lun(struct ctl_be_lun *be_lun) int ctl_start_lun(struct ctl_be_lun *be_lun) { - struct ctl_softc *ctl_softc; - struct ctl_lun *lun; - - ctl_softc = control_softc; - - lun = (struct ctl_lun *)be_lun->ctl_lun; + struct ctl_lun *lun = (struct ctl_lun *)be_lun->ctl_lun; mtx_lock(&lun->lun_lock); lun->flags &= ~CTL_LUN_STOPPED; mtx_unlock(&lun->lun_lock); - return (0); } int ctl_stop_lun(struct ctl_be_lun *be_lun) { - struct ctl_softc *ctl_softc; - struct ctl_lun *lun; - - ctl_softc = control_softc; - - lun = (struct ctl_lun *)be_lun->ctl_lun; + struct ctl_lun *lun = (struct ctl_lun *)be_lun->ctl_lun; mtx_lock(&lun->lun_lock); lun->flags |= CTL_LUN_STOPPED; mtx_unlock(&lun->lun_lock); - return (0); } int ctl_lun_offline(struct ctl_be_lun *be_lun) { - struct ctl_softc *ctl_softc; - struct ctl_lun *lun; - - ctl_softc = control_softc; - - lun = (struct ctl_lun *)be_lun->ctl_lun; + struct ctl_lun *lun = (struct ctl_lun *)be_lun->ctl_lun; mtx_lock(&lun->lun_lock); lun->flags |= CTL_LUN_OFFLINE; mtx_unlock(&lun->lun_lock); - return (0); } int ctl_lun_online(struct ctl_be_lun *be_lun) { - struct ctl_softc *ctl_softc; - struct ctl_lun *lun; - - ctl_softc = control_softc; - - lun = (struct ctl_lun *)be_lun->ctl_lun; + struct ctl_lun *lun = (struct ctl_lun *)be_lun->ctl_lun; mtx_lock(&lun->lun_lock); lun->flags &= ~CTL_LUN_OFFLINE; mtx_unlock(&lun->lun_lock); - return (0); } int ctl_invalidate_lun(struct ctl_be_lun *be_lun) { - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; struct ctl_lun *lun; - ctl_softc = control_softc; - lun = (struct ctl_lun *)be_lun->ctl_lun; + softc = lun->ctl_softc; mtx_lock(&lun->lun_lock); @@ -5002,9 +4974,9 @@ ctl_invalidate_lun(struct ctl_be_lun *be_lun) */ if (TAILQ_EMPTY(&lun->ooa_queue)) { mtx_unlock(&lun->lun_lock); - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); ctl_free_lun(lun); - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); } else mtx_unlock(&lun->lun_lock); @@ -5014,32 +4986,22 @@ ctl_invalidate_lun(struct ctl_be_lun *be_lun) int ctl_lun_inoperable(struct ctl_be_lun *be_lun) { - struct ctl_softc *ctl_softc; - struct ctl_lun *lun; - - ctl_softc = control_softc; - lun = (struct ctl_lun *)be_lun->ctl_lun; + struct ctl_lun *lun = (struct ctl_lun *)be_lun->ctl_lun; mtx_lock(&lun->lun_lock); lun->flags |= CTL_LUN_INOPERABLE; mtx_unlock(&lun->lun_lock); - return (0); } int ctl_lun_operable(struct ctl_be_lun *be_lun) { - struct ctl_softc *ctl_softc; - struct ctl_lun *lun; - - ctl_softc = control_softc; - lun = (struct ctl_lun *)be_lun->ctl_lun; + struct ctl_lun *lun = (struct ctl_lun *)be_lun->ctl_lun; mtx_lock(&lun->lun_lock); lun->flags &= ~CTL_LUN_INOPERABLE; mtx_unlock(&lun->lun_lock); - return (0); } @@ -5225,7 +5187,6 @@ int ctl_scsi_release(struct ctl_scsiio *ctsio) { int length, longid, thirdparty_id, resv_id; - struct ctl_softc *ctl_softc; struct ctl_lun *lun; uint32_t residx; @@ -5236,7 +5197,6 @@ ctl_scsi_release(struct ctl_scsiio *ctsio) residx = ctl_get_resindex(&ctsio->io_hdr.nexus); lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; - ctl_softc = control_softc; switch (ctsio->cdb[0]) { case RELEASE_10: { @@ -5313,7 +5273,6 @@ ctl_scsi_reserve(struct ctl_scsiio *ctsio) int extent, thirdparty, longid; int resv_id, length; uint64_t thirdparty_id; - struct ctl_softc *ctl_softc; struct ctl_lun *lun; uint32_t residx; @@ -5328,7 +5287,6 @@ ctl_scsi_reserve(struct ctl_scsiio *ctsio) residx = ctl_get_resindex(&ctsio->io_hdr.nexus); lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; - ctl_softc = control_softc; switch (ctsio->cdb[0]) { case RESERVE_10: { @@ -5402,13 +5360,11 @@ ctl_start_stop(struct ctl_scsiio *ctsio) { struct scsi_start_stop_unit *cdb; struct ctl_lun *lun; - struct ctl_softc *ctl_softc; int retval; CTL_DEBUG_PRINT(("ctl_start_stop\n")); lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; - ctl_softc = control_softc; retval = 0; cdb = (struct scsi_start_stop_unit *)ctsio->cdb; @@ -5529,7 +5485,7 @@ int ctl_sync_cache(struct ctl_scsiio *ctsio) { struct ctl_lun *lun; - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; uint64_t starting_lba; uint32_t block_count; int retval; @@ -5537,7 +5493,7 @@ ctl_sync_cache(struct ctl_scsiio *ctsio) CTL_DEBUG_PRINT(("ctl_sync_cache\n")); lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; - ctl_softc = control_softc; + softc = lun->ctl_softc; retval = 0; switch (ctsio->cdb[0]) { @@ -5590,7 +5546,7 @@ ctl_sync_cache(struct ctl_scsiio *ctsio) * CACHE command directly to the back end. */ mtx_lock(&lun->lun_lock); - if ((ctl_softc->flags & CTL_FLAG_REAL_SYNC) + if ((softc->flags & CTL_FLAG_REAL_SYNC) && (++(lun->sync_count) >= lun->sync_interval)) { lun->sync_count = 0; mtx_unlock(&lun->lun_lock); @@ -5611,13 +5567,11 @@ ctl_format(struct ctl_scsiio *ctsio) { struct scsi_format *cdb; struct ctl_lun *lun; - struct ctl_softc *ctl_softc; int length, defect_list_len; CTL_DEBUG_PRINT(("ctl_format\n")); lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; - ctl_softc = control_softc; cdb = (struct scsi_format *)ctsio->cdb; @@ -6075,7 +6029,6 @@ ctl_control_page_handler(struct ctl_scsiio *ctsio, { struct scsi_control_page *current_cp, *saved_cp, *user_cp; struct ctl_lun *lun; - struct ctl_softc *softc; int set_ua; uint32_t initidx; @@ -6091,8 +6044,6 @@ ctl_control_page_handler(struct ctl_scsiio *ctsio, (page_index->page_data + (page_index->page_len * CTL_PAGE_SAVED)); - softc = control_softc; - mtx_lock(&lun->lun_lock); if (((current_cp->rlec & SCP_DSENSE) == 0) && ((user_cp->rlec & SCP_DSENSE) != 0)) { @@ -6801,7 +6752,7 @@ ctl_mode_sense(struct ctl_scsiio *ctsio) header = (struct scsi_mode_hdr_6 *)ctsio->kern_data_ptr; - header->datalen = ctl_min(total_len - 1, 254); + header->datalen = MIN(total_len - 1, 254); if (control_dev == 0) { header->dev_specific = 0x10; /* DPOFUA */ if ((lun->flags & CTL_LUN_READONLY) || @@ -6823,7 +6774,7 @@ ctl_mode_sense(struct ctl_scsiio *ctsio) header = (struct scsi_mode_hdr_10 *)ctsio->kern_data_ptr; - datalen = ctl_min(total_len - 2, 65533); + datalen = MIN(total_len - 2, 65533); scsi_ulto2b(datalen, header->datalen); if (control_dev == 0) { header->dev_specific = 0x10; /* DPOFUA */ @@ -7358,8 +7309,8 @@ ctl_report_tagret_port_groups(struct ctl_scsiio *ctsio) CTL_DEBUG_PRINT(("ctl_report_tagret_port_groups\n")); cdb = (struct scsi_maintenance_in *)ctsio->cdb; - softc = control_softc; lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; + softc = lun->ctl_softc; retval = CTL_RETVAL_COMPLETE; @@ -7390,7 +7341,8 @@ ctl_report_tagret_port_groups(struct ctl_scsiio *ctsio) STAILQ_FOREACH(port, &softc->port_list, links) { if ((port->status & CTL_PORT_STATUS_ONLINE) == 0) continue; - if (ctl_map_lun_back(port->targ_port, lun->lun) >= CTL_MAX_LUNS) + if (ctl_map_lun_back(softc, port->targ_port, lun->lun) >= + CTL_MAX_LUNS) continue; num_target_ports++; } @@ -7463,8 +7415,8 @@ ctl_report_tagret_port_groups(struct ctl_scsiio *ctsio) STAILQ_FOREACH(port, &softc->port_list, links) { if ((port->status & CTL_PORT_STATUS_ONLINE) == 0) continue; - if (ctl_map_lun_back(port->targ_port, lun->lun) >= - CTL_MAX_LUNS) + if (ctl_map_lun_back(softc, port->targ_port, lun->lun) + >= CTL_MAX_LUNS) continue; p = port->targ_port % CTL_MAX_PORTS + g * CTL_MAX_PORTS; scsi_ulto2b(p, tpg_desc->descriptors[pc]. @@ -7756,13 +7708,12 @@ ctl_persistent_reserve_in(struct ctl_scsiio *ctsio) CTL_DEBUG_PRINT(("ctl_persistent_reserve_in\n")); - softc = control_softc; - cdb = (struct scsi_per_res_in *)ctsio->cdb; alloc_len = scsi_2btoul(cdb->length); lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; + softc = lun->ctl_softc; retry: mtx_lock(&lun->lun_lock); @@ -8385,10 +8336,9 @@ ctl_persistent_reserve_out(struct ctl_scsiio *ctsio) retval = CTL_RETVAL_COMPLETE; - softc = control_softc; - cdb = (struct scsi_per_res_out *)ctsio->cdb; lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; + softc = lun->ctl_softc; /* * We only support whole-LUN scope. The scope & type are ignored for @@ -9304,6 +9254,7 @@ ctl_verify(struct ctl_scsiio *ctsio) int ctl_report_luns(struct ctl_scsiio *ctsio) { + struct ctl_softc *softc = control_softc; struct scsi_report_luns *cdb; struct scsi_report_luns_data *lun_data; struct ctl_lun *lun, *request_lun; @@ -9319,9 +9270,9 @@ ctl_report_luns(struct ctl_scsiio *ctsio) CTL_DEBUG_PRINT(("ctl_report_luns\n")); - mtx_lock(&control_softc->ctl_lock); - num_luns = control_softc->num_luns; - mtx_unlock(&control_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); + num_luns = softc->num_luns; + mtx_unlock(&softc->ctl_lock); switch (cdb->select_report) { case RPL_REPORT_DEFAULT: @@ -9373,12 +9324,13 @@ ctl_report_luns(struct ctl_scsiio *ctsio) initidx = ctl_get_initindex(&ctsio->io_hdr.nexus); - mtx_lock(&control_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); for (targ_lun_id = 0, num_filled = 0; targ_lun_id < CTL_MAX_LUNS && num_filled < num_luns; targ_lun_id++) { - lun_id = ctl_map_lun(ctsio->io_hdr.nexus.targ_port, targ_lun_id); + lun_id = ctl_map_lun(softc, ctsio->io_hdr.nexus.targ_port, + targ_lun_id); if (lun_id >= CTL_MAX_LUNS) continue; - lun = control_softc->ctl_luns[lun_id]; + lun = softc->ctl_luns[lun_id]; if (lun == NULL) continue; @@ -9434,7 +9386,7 @@ ctl_report_luns(struct ctl_scsiio *ctsio) mtx_unlock(&lun->lun_lock); } } - mtx_unlock(&control_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); /* * It's quite possible that we've returned fewer LUNs than we allocated @@ -9563,7 +9515,7 @@ ctl_request_sense(struct ctl_scsiio *ctsio) (struct scsi_sense_data_fixed *)sense_ptr); else memcpy(sense_ptr, &lun->pending_sense[initidx], - ctl_min(sizeof(*sense_ptr), + MIN(sizeof(*sense_ptr), sizeof(lun->pending_sense[initidx]))); ctl_clear_mask(lun->have_ca, initidx); @@ -9647,6 +9599,7 @@ ctl_inquiry_evpd_supported(struct ctl_scsiio *ctsio, int alloc_len) struct scsi_vpd_supported_pages *pages; int sup_page_size; struct ctl_lun *lun; + int p; lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; @@ -9680,27 +9633,30 @@ ctl_inquiry_evpd_supported(struct ctl_scsiio *ctsio, int alloc_len) else pages->device = (SID_QUAL_LU_OFFLINE << 5) | T_DIRECT; - pages->length = SCSI_EVPD_NUM_SUPPORTED_PAGES; + p = 0; /* Supported VPD pages */ - pages->page_list[0] = SVPD_SUPPORTED_PAGES; + pages->page_list[p++] = SVPD_SUPPORTED_PAGES; /* Serial Number */ - pages->page_list[1] = SVPD_UNIT_SERIAL_NUMBER; + pages->page_list[p++] = SVPD_UNIT_SERIAL_NUMBER; /* Device Identification */ - pages->page_list[2] = SVPD_DEVICE_ID; + pages->page_list[p++] = SVPD_DEVICE_ID; /* Extended INQUIRY Data */ - pages->page_list[3] = SVPD_EXTENDED_INQUIRY_DATA; + pages->page_list[p++] = SVPD_EXTENDED_INQUIRY_DATA; /* Mode Page Policy */ - pages->page_list[4] = SVPD_MODE_PAGE_POLICY; + pages->page_list[p++] = SVPD_MODE_PAGE_POLICY; /* SCSI Ports */ - pages->page_list[5] = SVPD_SCSI_PORTS; + pages->page_list[p++] = SVPD_SCSI_PORTS; /* Third-party Copy */ - pages->page_list[6] = SVPD_SCSI_TPC; - /* Block limits */ - pages->page_list[7] = SVPD_BLOCK_LIMITS; - /* Block Device Characteristics */ - pages->page_list[8] = SVPD_BDC; - /* Logical Block Provisioning */ - pages->page_list[9] = SVPD_LBP; + pages->page_list[p++] = SVPD_SCSI_TPC; + if (lun != NULL && lun->be_lun->lun_type == T_DIRECT) { + /* Block limits */ + pages->page_list[p++] = SVPD_BLOCK_LIMITS; + /* Block Device Characteristics */ + pages->page_list[p++] = SVPD_BDC; + /* Logical Block Provisioning */ + pages->page_list[p++] = SVPD_LBP; + } + pages->length = p; ctl_set_success(ctsio); ctsio->io_hdr.flags |= CTL_FLAG_ALLOCATED; @@ -9869,15 +9825,15 @@ ctl_inquiry_evpd_devid(struct ctl_scsiio *ctsio, int alloc_len) { struct scsi_vpd_device_id *devid_ptr; struct scsi_vpd_id_descriptor *desc; - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; struct ctl_lun *lun; struct ctl_port *port; int data_len; uint8_t proto; - ctl_softc = control_softc; + softc = control_softc; - port = ctl_softc->ctl_ports[ctl_port_idx(ctsio->io_hdr.nexus.targ_port)]; + port = softc->ctl_ports[ctl_port_idx(ctsio->io_hdr.nexus.targ_port)]; lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; data_len = sizeof(struct scsi_vpd_device_id) + @@ -10011,7 +9967,7 @@ ctl_inquiry_evpd_scsi_ports(struct ctl_scsiio *ctsio, int alloc_len) if ((port->status & CTL_PORT_STATUS_ONLINE) == 0) continue; if (lun != NULL && - ctl_map_lun_back(port->targ_port, lun->lun) >= + ctl_map_lun_back(softc, port->targ_port, lun->lun) >= CTL_MAX_LUNS) continue; num_target_ports++; @@ -10065,8 +10021,8 @@ ctl_inquiry_evpd_scsi_ports(struct ctl_scsiio *ctsio, int alloc_len) if ((port->status & CTL_PORT_STATUS_ONLINE) == 0) continue; if (lun != NULL && - ctl_map_lun_back(port->targ_port, lun->lun) >= - CTL_MAX_LUNS) + ctl_map_lun_back(softc, port->targ_port, lun->lun) + >= CTL_MAX_LUNS) continue; p = port->targ_port % CTL_MAX_PORTS + g * CTL_MAX_PORTS; scsi_ulto2b(p, pd->relative_port_id); @@ -10142,7 +10098,7 @@ ctl_inquiry_evpd_block_limits(struct ctl_scsiio *ctsio, int alloc_len) scsi_ulto4b(0xffffffff, bl_ptr->max_txfer_len); if (lun != NULL) { bs = lun->be_lun->blocksize; - scsi_ulto4b(MAXPHYS / bs, bl_ptr->opt_txfer_len); + scsi_ulto4b(lun->be_lun->opttxferlen, bl_ptr->opt_txfer_len); if (lun->be_lun->flags & CTL_LUN_FLAG_UNMAP) { scsi_ulto4b(0xffffffff, bl_ptr->max_unmap_lba_cnt); scsi_ulto4b(0xffffffff, bl_ptr->max_unmap_blk_cnt); @@ -10282,13 +10238,12 @@ ctl_inquiry_evpd_lbp(struct ctl_scsiio *ctsio, int alloc_len) static int ctl_inquiry_evpd(struct ctl_scsiio *ctsio) { + struct ctl_lun *lun; struct scsi_inquiry *cdb; int alloc_len, retval; + lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; cdb = (struct scsi_inquiry *)ctsio->cdb; - - retval = CTL_RETVAL_COMPLETE; - alloc_len = scsi_2btoul(cdb->length); switch (cdb->page_code) { @@ -10314,15 +10269,22 @@ ctl_inquiry_evpd(struct ctl_scsiio *ctsio) retval = ctl_inquiry_evpd_tpc(ctsio, alloc_len); break; case SVPD_BLOCK_LIMITS: + if (lun == NULL || lun->be_lun->lun_type != T_DIRECT) + goto err; retval = ctl_inquiry_evpd_block_limits(ctsio, alloc_len); break; case SVPD_BDC: + if (lun == NULL || lun->be_lun->lun_type != T_DIRECT) + goto err; retval = ctl_inquiry_evpd_bdc(ctsio, alloc_len); break; case SVPD_LBP: + if (lun == NULL || lun->be_lun->lun_type != T_DIRECT) + goto err; retval = ctl_inquiry_evpd_lbp(ctsio, alloc_len); break; default: +err: ctl_set_invalid_field(ctsio, /*sks_valid*/ 1, /*command*/ 1, @@ -10342,20 +10304,20 @@ ctl_inquiry_std(struct ctl_scsiio *ctsio) { struct scsi_inquiry_data *inq_ptr; struct scsi_inquiry *cdb; - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; struct ctl_lun *lun; char *val; uint32_t alloc_len, data_len; ctl_port_type port_type; - ctl_softc = control_softc; + softc = control_softc; /* * Figure out whether we're talking to a Fibre Channel port or not. * We treat the ioctl front end, and any SCSI adapters, as packetized * SCSI front ends. */ - port_type = ctl_softc->ctl_ports[ + port_type = softc->ctl_ports[ ctl_port_idx(ctsio->io_hdr.nexus.targ_port)]->port_type; if (port_type == CTL_PORT_IOCTL || port_type == CTL_PORT_INTERNAL) port_type = CTL_PORT_SCSI; @@ -10432,7 +10394,7 @@ ctl_inquiry_std(struct ctl_scsiio *ctsio) if (lun != NULL) inq_ptr->device = (SID_QUAL_LU_CONNECTED << 5) | lun->be_lun->lun_type; - else if (ctl_softc->inquiry_pq_no_lun == 0) + else if (softc->inquiry_pq_no_lun == 0) inq_ptr->device = (SID_QUAL_LU_OFFLINE << 5) | T_DIRECT; else inq_ptr->device = (SID_QUAL_BAD_LU << 5) | T_NODEVICE; @@ -11065,7 +11027,6 @@ ctl_check_blocked(struct ctl_lun *lun) break; case CTL_ACTION_PASS: case CTL_ACTION_SKIP: { - struct ctl_softc *softc; const struct ctl_cmd_entry *entry; int isc_retval; @@ -11104,7 +11065,6 @@ ctl_check_blocked(struct ctl_lun *lun) break; } entry = ctl_get_cmd_entry(&cur_blocked->scsiio, NULL); - softc = control_softc; /* * Check this I/O for LUN state changes that may @@ -11114,7 +11074,7 @@ ctl_check_blocked(struct ctl_lun *lun) * for any states that can be caused by SCSI * commands. */ - if (ctl_scsiio_lun_check(softc, lun, entry, + if (ctl_scsiio_lun_check(lun, entry, &cur_blocked->scsiio) == 0) { cur_blocked->io_hdr.flags |= CTL_FLAG_IS_WAS_ON_RTR; @@ -11148,9 +11108,10 @@ ctl_check_blocked(struct ctl_lun *lun) * careful attention to the placement of any new checks. */ static int -ctl_scsiio_lun_check(struct ctl_softc *ctl_softc, struct ctl_lun *lun, +ctl_scsiio_lun_check(struct ctl_lun *lun, const struct ctl_cmd_entry *entry, struct ctl_scsiio *ctsio) { + struct ctl_softc *softc = lun->ctl_softc; int retval; uint32_t residx; @@ -11162,7 +11123,7 @@ ctl_scsiio_lun_check(struct ctl_softc *ctl_softc, struct ctl_lun *lun, * If this shelf is a secondary shelf controller, we have to reject * any media access commands. */ - if ((ctl_softc->flags & CTL_FLAG_ACTIVE_SHELF) == 0 && + if ((softc->flags & CTL_FLAG_ACTIVE_SHELF) == 0 && (entry->flags & CTL_CMD_FLAG_OK_ON_SECONDARY) == 0) { ctl_set_lun_standby(ctsio); retval = 1; @@ -11268,14 +11229,14 @@ static void ctl_failover(void) { struct ctl_lun *lun; - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; union ctl_io *next_io, *pending_io; union ctl_io *io; int lun_idx; - ctl_softc = control_softc; + softc = control_softc; - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); /* * Remove any cmds from the other SC from the rtr queue. These * will obviously only be for LUNs for which we're the primary. @@ -11285,19 +11246,19 @@ ctl_failover(void) * which HA mode we're in. */ #ifdef notyet - mtx_lock(&ctl_softc->queue_lock); - for (io = (union ctl_io *)STAILQ_FIRST(&ctl_softc->rtr_queue); + mtx_lock(&softc->queue_lock); + for (io = (union ctl_io *)STAILQ_FIRST(&softc->rtr_queue); io != NULL; io = next_io) { next_io = (union ctl_io *)STAILQ_NEXT(&io->io_hdr, links); if (io->io_hdr.flags & CTL_FLAG_FROM_OTHER_SC) - STAILQ_REMOVE(&ctl_softc->rtr_queue, &io->io_hdr, + STAILQ_REMOVE(&softc->rtr_queue, &io->io_hdr, ctl_io_hdr, links); } - mtx_unlock(&ctl_softc->queue_lock); + mtx_unlock(&softc->queue_lock); #endif - for (lun_idx=0; lun_idx < ctl_softc->num_luns; lun_idx++) { - lun = ctl_softc->ctl_luns[lun_idx]; + for (lun_idx=0; lun_idx < softc->num_luns; lun_idx++) { + lun = softc->ctl_luns[lun_idx]; if (lun==NULL) continue; @@ -11309,7 +11270,7 @@ ctl_failover(void) continue; if ((lun->flags & CTL_LUN_PRIMARY_SC) - && (ctl_softc->ha_mode == CTL_HA_MODE_SER_ONLY)) { + && (softc->ha_mode == CTL_HA_MODE_SER_ONLY)) { printf("FAILOVER: primary lun %d\n", lun_idx); /* * Remove all commands from the other SC. First from the @@ -11351,7 +11312,7 @@ ctl_failover(void) } ctl_check_blocked(lun); } else if ((lun->flags & CTL_LUN_PRIMARY_SC) - && (ctl_softc->ha_mode == CTL_HA_MODE_XFER)) { + && (softc->ha_mode == CTL_HA_MODE_XFER)) { printf("FAILOVER: primary lun %d\n", lun_idx); /* @@ -11369,7 +11330,7 @@ ctl_failover(void) io->io_hdr.flags |= CTL_FLAG_ABORT; } } else if (((lun->flags & CTL_LUN_PRIMARY_SC) == 0) - && (ctl_softc->ha_mode == CTL_HA_MODE_XFER)) { + && (softc->ha_mode == CTL_HA_MODE_XFER)) { printf("FAILOVER: secondary lun %d\n", lun_idx); @@ -11408,7 +11369,7 @@ ctl_failover(void) ctl_est_ua_all(lun, -1, CTL_UA_ASYM_ACC_CHANGE); } else if (((lun->flags & CTL_LUN_PRIMARY_SC) == 0) - && (ctl_softc->ha_mode == CTL_HA_MODE_SER_ONLY)) { + && (softc->ha_mode == CTL_HA_MODE_SER_ONLY)) { printf("FAILOVER: secondary lun %d\n", lun_idx); /* * if the first io on the OOA is not on the RtR queue @@ -11498,15 +11459,15 @@ ctl_failover(void) ctl_est_ua_all(lun, -1, CTL_UA_ASYM_ACC_CHANGE); } else { panic("Unhandled HA mode failover, LUN flags = %#x, " - "ha_mode = #%x", lun->flags, ctl_softc->ha_mode); + "ha_mode = #%x", lun->flags, softc->ha_mode); } } ctl_pause_rtr = 0; - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); } static int -ctl_scsiio_precheck(struct ctl_softc *ctl_softc, struct ctl_scsiio *ctsio) +ctl_scsiio_precheck(struct ctl_softc *softc, struct ctl_scsiio *ctsio) { struct ctl_lun *lun; const struct ctl_cmd_entry *entry; @@ -11519,7 +11480,7 @@ ctl_scsiio_precheck(struct ctl_softc *ctl_softc, struct ctl_scsiio *ctsio) targ_lun = ctsio->io_hdr.nexus.targ_mapped_lun; if ((targ_lun < CTL_MAX_LUNS) - && ((lun = ctl_softc->ctl_luns[targ_lun]) != NULL)) { + && ((lun = softc->ctl_luns[targ_lun]) != NULL)) { /* * If the LUN is invalid, pretend that it doesn't exist. * It will go away as soon as all pending I/O has been @@ -11651,7 +11612,7 @@ ctl_scsiio_precheck(struct ctl_softc *ctl_softc, struct ctl_scsiio *ctsio) } - if (ctl_scsiio_lun_check(ctl_softc, lun, entry, ctsio) != 0) { + if (ctl_scsiio_lun_check(lun, entry, ctsio) != 0) { mtx_unlock(&lun->lun_lock); ctl_done((union ctl_io *)ctsio); return (retval); @@ -11859,13 +11820,13 @@ bailout: * our single target. */ static int -ctl_bus_reset(struct ctl_softc *ctl_softc, union ctl_io *io) +ctl_bus_reset(struct ctl_softc *softc, union ctl_io *io) { - return(ctl_target_reset(ctl_softc, io, CTL_UA_BUS_RESET)); + return(ctl_target_reset(softc, io, CTL_UA_BUS_RESET)); } static int -ctl_target_reset(struct ctl_softc *ctl_softc, union ctl_io *io, +ctl_target_reset(struct ctl_softc *softc, union ctl_io *io, ctl_ua_type ua_type) { struct ctl_lun *lun; @@ -11889,10 +11850,10 @@ ctl_target_reset(struct ctl_softc *ctl_softc, union ctl_io *io, } retval = 0; - mtx_lock(&ctl_softc->ctl_lock); - STAILQ_FOREACH(lun, &ctl_softc->lun_list, links) + mtx_lock(&softc->ctl_lock); + STAILQ_FOREACH(lun, &softc->lun_list, links) retval += ctl_lun_reset(lun, io, ua_type); - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (retval); } @@ -12078,7 +12039,7 @@ ctl_abort_task(union ctl_io *io) { union ctl_io *xio; struct ctl_lun *lun; - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; #if 0 struct sbuf sb; char printbuf[128]; @@ -12086,19 +12047,19 @@ ctl_abort_task(union ctl_io *io) int found; uint32_t targ_lun; - ctl_softc = control_softc; + softc = control_softc; found = 0; /* * Look up the LUN. */ targ_lun = io->io_hdr.nexus.targ_mapped_lun; - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); if ((targ_lun < CTL_MAX_LUNS) - && (ctl_softc->ctl_luns[targ_lun] != NULL)) - lun = ctl_softc->ctl_luns[targ_lun]; + && (softc->ctl_luns[targ_lun] != NULL)) + lun = softc->ctl_luns[targ_lun]; else { - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (1); } @@ -12108,7 +12069,7 @@ ctl_abort_task(union ctl_io *io) #endif mtx_lock(&lun->lun_lock); - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); /* * Run through the OOA queue and attempt to find the given I/O. * The target port, initiator ID, tag type and tag number have to @@ -12224,7 +12185,7 @@ ctl_abort_task(union ctl_io *io) static void ctl_run_task(union ctl_io *io) { - struct ctl_softc *ctl_softc = control_softc; + struct ctl_softc *softc = control_softc; int retval = 1; const char *task_desc; @@ -12279,12 +12240,12 @@ ctl_run_task(union ctl_io *io) uint32_t targ_lun; targ_lun = io->io_hdr.nexus.targ_mapped_lun; - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); if ((targ_lun < CTL_MAX_LUNS) - && (ctl_softc->ctl_luns[targ_lun] != NULL)) - lun = ctl_softc->ctl_luns[targ_lun]; + && (softc->ctl_luns[targ_lun] != NULL)) + lun = softc->ctl_luns[targ_lun]; else { - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); retval = 1; break; } @@ -12311,14 +12272,14 @@ ctl_run_task(union ctl_io *io) retval = ctl_lun_reset(lun, io, CTL_UA_LUN_RESET); - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); break; } case CTL_TASK_TARGET_RESET: - retval = ctl_target_reset(ctl_softc, io, CTL_UA_TARG_RESET); + retval = ctl_target_reset(softc, io, CTL_UA_TARG_RESET); break; case CTL_TASK_BUS_RESET: - retval = ctl_bus_reset(ctl_softc, io); + retval = ctl_bus_reset(softc, io); break; case CTL_TASK_PORT_LOGIN: break; @@ -12345,13 +12306,13 @@ ctl_handle_isc(union ctl_io *io) { int free_io; struct ctl_lun *lun; - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; uint32_t targ_lun; - ctl_softc = control_softc; + softc = control_softc; targ_lun = io->io_hdr.nexus.targ_mapped_lun; - lun = ctl_softc->ctl_luns[targ_lun]; + lun = softc->ctl_luns[targ_lun]; switch (io->io_hdr.msg_type) { case CTL_MSG_SERIALIZE: @@ -12366,7 +12327,7 @@ ctl_handle_isc(union ctl_io *io) free_io = 0; entry = ctl_get_cmd_entry(&io->scsiio, NULL); mtx_lock(&lun->lun_lock); - if (ctl_scsiio_lun_check(ctl_softc, lun, + if (ctl_scsiio_lun_check(lun, entry, (struct ctl_scsiio *)io) != 0) { mtx_unlock(&lun->lun_lock); ctl_done(io); @@ -12378,7 +12339,7 @@ ctl_handle_isc(union ctl_io *io) break; } case CTL_MSG_FINISH_IO: - if (ctl_softc->ha_mode == CTL_HA_MODE_XFER) { + if (softc->ha_mode == CTL_HA_MODE_XFER) { free_io = 0; ctl_done(io); } else { @@ -12523,8 +12484,8 @@ ctl_inject_error(struct ctl_lun *lun, union ctl_io *io) * checks. */ bcopy(&desc->custom_sense, &io->scsiio.sense_data, - ctl_min(sizeof(desc->custom_sense), - sizeof(io->scsiio.sense_data))); + MIN(sizeof(desc->custom_sense), + sizeof(io->scsiio.sense_data))); io->scsiio.scsi_status = SCSI_STATUS_CHECK_COND; io->scsiio.sense_len = SSD_FULL_SIZE; io->io_hdr.status = CTL_SCSI_ERROR | CTL_AUTOSENSE; @@ -12753,7 +12714,7 @@ ctl_datamove(union ctl_io *io) */ for (sg_entries_sent = 0; sg_entries_sent < msg.dt.kern_sg_entries; msg.dt.sg_sequence++) { - msg.dt.cur_sg_entries = ctl_min((sizeof(msg.dt.sg_list)/ + msg.dt.cur_sg_entries = MIN((sizeof(msg.dt.sg_list)/ sizeof(msg.dt.sg_list[0])), msg.dt.kern_sg_entries - sg_entries_sent); @@ -13116,7 +13077,7 @@ ctl_datamove_remote_sgl_setup(union ctl_io *io) for (i = 0; (i < sizeof(io->io_hdr.remote_sglist) / sizeof(io->io_hdr.remote_sglist[0])) && (len_to_go > 0); i++) { - local_sglist[i].len = ctl_min(len_to_go, 131072); + local_sglist[i].len = MIN(len_to_go, 131072); CTL_SIZE_8B(local_dma_sglist[i].len, local_sglist[i].len); local_sglist[i].addr = @@ -13252,8 +13213,8 @@ ctl_datamove_remote_xfer(union ctl_io *io, unsigned command, * also have enough slack left over at the end, though, * to round up to the next 8 byte boundary. */ - cur_len = ctl_min(local_sglist[i].len - local_used, - remote_sglist[j].len - remote_used); + cur_len = MIN(local_sglist[i].len - local_used, + remote_sglist[j].len - remote_used); /* * In this case, we have a size issue and need to decrease @@ -13477,14 +13438,13 @@ static int ctl_process_done(union ctl_io *io) { struct ctl_lun *lun; - struct ctl_softc *ctl_softc = control_softc; + struct ctl_softc *softc = control_softc; void (*fe_done)(union ctl_io *io); uint32_t targ_port = ctl_port_idx(io->io_hdr.nexus.targ_port); CTL_DEBUG_PRINT(("ctl_process_done\n")); - fe_done = - control_softc->ctl_ports[targ_port]->fe_done; + fe_done = softc->ctl_ports[targ_port]->fe_done; #ifdef CTL_TIME_IO if ((time_uptime - io->io_hdr.start_time) > ctl_time_io_secs) { @@ -13611,9 +13571,9 @@ ctl_process_done(union ctl_io *io) if ((lun->flags & CTL_LUN_INVALID) && TAILQ_EMPTY(&lun->ooa_queue)) { mtx_unlock(&lun->lun_lock); - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); ctl_free_lun(lun); - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); } else mtx_unlock(&lun->lun_lock); @@ -13665,7 +13625,7 @@ bailout: * if the frontend comes back in in this context to queue * something. */ - if ((ctl_softc->ha_mode == CTL_HA_MODE_XFER) + if ((softc->ha_mode == CTL_HA_MODE_XFER) && (io->io_hdr.flags & CTL_FLAG_FROM_OTHER_SC)) { union ctl_ha_msg msg; @@ -13713,10 +13673,10 @@ int ctl_queue_sense(union ctl_io *io) { struct ctl_lun *lun; - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; uint32_t initidx, targ_lun; - ctl_softc = control_softc; + softc = control_softc; CTL_DEBUG_PRINT(("ctl_queue_sense\n")); @@ -13727,17 +13687,17 @@ ctl_queue_sense(union ctl_io *io) * things like an INQUIRY to a LUN that we don't have enabled. We * can't deal with that right now. */ - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); /* * If we don't have a LUN for this, just toss the sense * information. */ targ_lun = io->io_hdr.nexus.targ_lun; - targ_lun = ctl_map_lun(io->io_hdr.nexus.targ_port, targ_lun); + targ_lun = ctl_map_lun(softc, io->io_hdr.nexus.targ_port, targ_lun); if ((targ_lun < CTL_MAX_LUNS) - && (ctl_softc->ctl_luns[targ_lun] != NULL)) - lun = ctl_softc->ctl_luns[targ_lun]; + && (softc->ctl_luns[targ_lun] != NULL)) + lun = softc->ctl_luns[targ_lun]; else goto bailout; @@ -13753,13 +13713,13 @@ ctl_queue_sense(union ctl_io *io) } memcpy(&lun->pending_sense[initidx], &io->scsiio.sense_data, - ctl_min(sizeof(lun->pending_sense[initidx]), + MIN(sizeof(lun->pending_sense[initidx]), sizeof(io->scsiio.sense_data))); ctl_set_mask(lun->have_ca, initidx); mtx_unlock(&lun->lun_lock); bailout: - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); ctl_free_io(io); @@ -13774,12 +13734,9 @@ bailout: int ctl_queue(union ctl_io *io) { - struct ctl_softc *ctl_softc; CTL_DEBUG_PRINT(("ctl_queue cdb[0]=%02X\n", io->scsiio.cdb[0])); - ctl_softc = control_softc; - #ifdef CTL_TIME_IO io->io_hdr.start_time = time_uptime; getbintime(&io->io_hdr.start_bt); @@ -13787,7 +13744,8 @@ ctl_queue(union ctl_io *io) /* Map FE-specific LUN ID into global one. */ io->io_hdr.nexus.targ_mapped_lun = - ctl_map_lun(io->io_hdr.nexus.targ_port, io->io_hdr.nexus.targ_lun); + ctl_map_lun(control_softc, io->io_hdr.nexus.targ_port, + io->io_hdr.nexus.targ_lun); switch (io->io_hdr.io_type) { case CTL_IO_SCSI: @@ -13818,9 +13776,6 @@ ctl_done_timer_wakeup(void *arg) void ctl_done(union ctl_io *io) { - struct ctl_softc *ctl_softc; - - ctl_softc = control_softc; /* * Enable this to catch duplicate completion issues. diff --git a/sys/cam/ctl/ctl.h b/sys/cam/ctl/ctl.h index ae6c583..9cf967c 100644 --- a/sys/cam/ctl/ctl.h +++ b/sys/cam/ctl/ctl.h @@ -40,7 +40,6 @@ #ifndef _CTL_H_ #define _CTL_H_ -#define ctl_min(x,y) (((x) < (y)) ? (x) : (y)) #define CTL_RETVAL_COMPLETE 0 #define CTL_RETVAL_QUEUED 1 #define CTL_RETVAL_ALLOCATED 2 diff --git a/sys/cam/ctl/ctl_backend.c b/sys/cam/ctl/ctl_backend.c index 0e1a76c..cabecb7 100644 --- a/sys/cam/ctl/ctl_backend.c +++ b/sys/cam/ctl/ctl_backend.c @@ -66,33 +66,33 @@ extern struct ctl_softc *control_softc; int ctl_backend_register(struct ctl_backend_driver *be) { - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; struct ctl_backend_driver *be_tmp; - ctl_softc = control_softc; + softc = control_softc; - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); /* * Sanity check, make sure this isn't a duplicate registration. */ - STAILQ_FOREACH(be_tmp, &ctl_softc->be_list, links) { + STAILQ_FOREACH(be_tmp, &softc->be_list, links) { if (strcmp(be_tmp->name, be->name) == 0) { - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (-1); } } - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); /* * Call the backend's initialization routine. */ be->init(); - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); - STAILQ_INSERT_TAIL(&ctl_softc->be_list, be, links); + STAILQ_INSERT_TAIL(&softc->be_list, be, links); - ctl_softc->num_backends++; + softc->num_backends++; /* * Don't want to increment the usage count for internal consumers, @@ -113,7 +113,7 @@ ctl_backend_register(struct ctl_backend_driver *be) atomic_set(&be->num_luns, 0); #endif - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (0); } @@ -121,24 +121,24 @@ ctl_backend_register(struct ctl_backend_driver *be) int ctl_backend_deregister(struct ctl_backend_driver *be) { - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; - ctl_softc = control_softc; + softc = control_softc; - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); #if 0 if (atomic_read(&be->num_luns) != 0) { #endif /* XXX KDM fix this! */ if (be->num_luns != 0) { - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (-1); } - STAILQ_REMOVE(&ctl_softc->be_list, be, ctl_backend_driver, links); + STAILQ_REMOVE(&softc->be_list, be, ctl_backend_driver, links); - ctl_softc->num_backends--; + softc->num_backends--; /* XXX KDM find a substitute for this? */ #if 0 @@ -146,7 +146,7 @@ ctl_backend_deregister(struct ctl_backend_driver *be) MOD_DEC_USE_COUNT; #endif - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (0); } @@ -154,21 +154,21 @@ ctl_backend_deregister(struct ctl_backend_driver *be) struct ctl_backend_driver * ctl_backend_find(char *backend_name) { - struct ctl_softc *ctl_softc; + struct ctl_softc *softc; struct ctl_backend_driver *be_tmp; - ctl_softc = control_softc; + softc = control_softc; - mtx_lock(&ctl_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); - STAILQ_FOREACH(be_tmp, &ctl_softc->be_list, links) { + STAILQ_FOREACH(be_tmp, &softc->be_list, links) { if (strcmp(be_tmp->name, backend_name) == 0) { - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (be_tmp); } } - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (NULL); } diff --git a/sys/cam/ctl/ctl_backend.h b/sys/cam/ctl/ctl_backend.h index 5e0af5c..93a530c 100644 --- a/sys/cam/ctl/ctl_backend.h +++ b/sys/cam/ctl/ctl_backend.h @@ -146,10 +146,16 @@ typedef void (*be_lun_config_t)(void *be_lun, * * pblockexp is the log2() of number of LBAs on the LUN per physical sector. * - * pblockoff is the lowest LBA on the LUN aligned ot physical sector. + * pblockoff is the lowest LBA on the LUN aligned to physical sector. + * + * ublockexp is the log2() of number of LBAs on the LUN per UNMAP block. + * + * ublockoff is the lowest LBA on the LUN aligned to UNMAP block. * * atomicblock is the number of blocks that can be written atomically. * + * opttxferlen is the number of blocks that can be written in one operation. + * * req_lun_id is the requested LUN ID. CTL only pays attention to this * field if the CTL_LUN_FLAG_ID_REQ flag is set. If the requested LUN ID is * not available, the LUN addition will fail. If a particular LUN ID isn't @@ -197,6 +203,7 @@ struct ctl_be_lun { uint16_t ublockexp; /* passed to CTL */ uint16_t ublockoff; /* passed to CTL */ uint32_t atomicblock; /* passed to CTL */ + uint32_t opttxferlen; /* passed to CTL */ uint32_t req_lun_id; /* passed to CTL */ uint32_t lun_id; /* returned from CTL */ uint8_t serial_num[CTL_SN_LEN]; /* passed to CTL */ diff --git a/sys/cam/ctl/ctl_backend_block.c b/sys/cam/ctl/ctl_backend_block.c index 9c3e8fd..a5e45024 100644 --- a/sys/cam/ctl/ctl_backend_block.c +++ b/sys/cam/ctl/ctl_backend_block.c @@ -175,6 +175,8 @@ struct ctl_be_block_lun { uint16_t pblockoff; uint16_t ublockexp; uint16_t ublockoff; + uint32_t atomicblock; + uint32_t opttxferlen; struct ctl_be_block_softc *softc; struct devstat *disk_stats; ctl_be_block_lun_flags flags; @@ -1845,6 +1847,8 @@ ctl_be_block_open_file(struct ctl_be_block_lun *be_lun, struct ctl_lun_req *req) "file %s size %ju < block size %u", be_lun->dev_path, (uintmax_t)be_lun->size_bytes, be_lun->blocksize); } + + be_lun->opttxferlen = CTLBLK_MAX_IO_SIZE / be_lun->blocksize; return (error); } @@ -1856,7 +1860,7 @@ ctl_be_block_open_dev(struct ctl_be_block_lun *be_lun, struct ctl_lun_req *req) struct cdev *dev; struct cdevsw *devsw; char *value; - int error; + int error, atomic, maxio; off_t ps, pss, po, pos, us, uss, uo, uos; params = &be_lun->params; @@ -1870,8 +1874,16 @@ ctl_be_block_open_dev(struct ctl_be_block_lun *be_lun, struct ctl_lun_req *req) if (strcmp(be_lun->backend.dev.csw->d_name, "zvol") == 0) { be_lun->dispatch = ctl_be_block_dispatch_zvol; be_lun->get_lba_status = ctl_be_block_gls_zvol; - } else + atomic = maxio = CTLBLK_MAX_IO_SIZE; + } else { be_lun->dispatch = ctl_be_block_dispatch_dev; + atomic = 0; + maxio = be_lun->backend.dev.cdev->si_iosize_max; + if (maxio <= 0) + maxio = DFLTPHYS; + if (maxio > CTLBLK_MAX_IO_SIZE) + maxio = CTLBLK_MAX_IO_SIZE; + } be_lun->lun_flush = ctl_be_block_flush_dev; be_lun->unmap = ctl_be_block_unmap_dev; be_lun->getattr = ctl_be_block_getattr_dev; @@ -2002,6 +2014,8 @@ ctl_be_block_open_dev(struct ctl_be_block_lun *be_lun, struct ctl_lun_req *req) be_lun->ublockoff = (uss - uos) % uss; } + be_lun->atomicblock = atomic / be_lun->blocksize; + be_lun->opttxferlen = maxio / be_lun->blocksize; return (0); } @@ -2268,10 +2282,8 @@ ctl_be_block_create(struct ctl_be_block_softc *softc, struct ctl_lun_req *req) be_lun->ctl_be_lun.pblockoff = be_lun->pblockoff; be_lun->ctl_be_lun.ublockexp = be_lun->ublockexp; be_lun->ctl_be_lun.ublockoff = be_lun->ublockoff; - if (be_lun->dispatch == ctl_be_block_dispatch_zvol && - be_lun->blocksize != 0) - be_lun->ctl_be_lun.atomicblock = CTLBLK_MAX_IO_SIZE / - be_lun->blocksize; + be_lun->ctl_be_lun.atomicblock = be_lun->atomicblock; + be_lun->ctl_be_lun.opttxferlen = be_lun->opttxferlen; /* Tell the user the blocksize we ended up using */ params->lun_size_bytes = be_lun->size_bytes; params->blocksize_bytes = be_lun->blocksize; @@ -2290,32 +2302,32 @@ ctl_be_block_create(struct ctl_be_block_softc *softc, struct ctl_lun_req *req) snprintf(tmpstr, sizeof(tmpstr), "MYSERIAL%4d", softc->num_luns); strncpy((char *)be_lun->ctl_be_lun.serial_num, tmpstr, - ctl_min(sizeof(be_lun->ctl_be_lun.serial_num), + MIN(sizeof(be_lun->ctl_be_lun.serial_num), sizeof(tmpstr))); /* Tell the user what we used for a serial number */ strncpy((char *)params->serial_num, tmpstr, - ctl_min(sizeof(params->serial_num), sizeof(tmpstr))); + MIN(sizeof(params->serial_num), sizeof(tmpstr))); } else { strncpy((char *)be_lun->ctl_be_lun.serial_num, params->serial_num, - ctl_min(sizeof(be_lun->ctl_be_lun.serial_num), + MIN(sizeof(be_lun->ctl_be_lun.serial_num), sizeof(params->serial_num))); } if ((params->flags & CTL_LUN_FLAG_DEVID) == 0) { snprintf(tmpstr, sizeof(tmpstr), "MYDEVID%4d", softc->num_luns); strncpy((char *)be_lun->ctl_be_lun.device_id, tmpstr, - ctl_min(sizeof(be_lun->ctl_be_lun.device_id), + MIN(sizeof(be_lun->ctl_be_lun.device_id), sizeof(tmpstr))); /* Tell the user what we used for a device ID */ strncpy((char *)params->device_id, tmpstr, - ctl_min(sizeof(params->device_id), sizeof(tmpstr))); + MIN(sizeof(params->device_id), sizeof(tmpstr))); } else { strncpy((char *)be_lun->ctl_be_lun.device_id, params->device_id, - ctl_min(sizeof(be_lun->ctl_be_lun.device_id), - sizeof(params->device_id))); + MIN(sizeof(be_lun->ctl_be_lun.device_id), + sizeof(params->device_id))); } TASK_INIT(&be_lun->io_task, /*priority*/0, ctl_be_block_worker, be_lun); @@ -2649,10 +2661,8 @@ ctl_be_block_modify(struct ctl_be_block_softc *softc, struct ctl_lun_req *req) be_lun->ctl_be_lun.pblockoff = be_lun->pblockoff; be_lun->ctl_be_lun.ublockexp = be_lun->ublockexp; be_lun->ctl_be_lun.ublockoff = be_lun->ublockoff; - if (be_lun->dispatch == ctl_be_block_dispatch_zvol && - be_lun->blocksize != 0) - be_lun->ctl_be_lun.atomicblock = CTLBLK_MAX_IO_SIZE / - be_lun->blocksize; + be_lun->ctl_be_lun.atomicblock = be_lun->atomicblock; + be_lun->ctl_be_lun.opttxferlen = be_lun->opttxferlen; ctl_lun_capacity_changed(&be_lun->ctl_be_lun); if (oldsize == 0 && be_lun->size_blocks != 0) ctl_lun_online(&be_lun->ctl_be_lun); diff --git a/sys/cam/ctl/ctl_backend_ramdisk.c b/sys/cam/ctl/ctl_backend_ramdisk.c index adace4f..d43cb60 100644 --- a/sys/cam/ctl/ctl_backend_ramdisk.c +++ b/sys/cam/ctl/ctl_backend_ramdisk.c @@ -313,8 +313,7 @@ ctl_backend_ramdisk_continue(union ctl_io *io) sg_entries = (struct ctl_sg_entry *)io->scsiio.kern_data_ptr; for (i = 0, len_filled = 0; i < sg_filled; i++) { sg_entries[i].addr = softc->ramdisk_pages[i]; - sg_entries[i].len = ctl_min(PAGE_SIZE, - len - len_filled); + sg_entries[i].len = MIN(PAGE_SIZE, len - len_filled); len_filled += sg_entries[i].len; } io->io_hdr.flags |= CTL_FLAG_KDPTR_SGLIST; @@ -570,6 +569,8 @@ ctl_backend_ramdisk_create(struct ctl_be_ramdisk_softc *softc, be_lun->size_bytes = be_lun->size_blocks * blocksize; be_lun->ctl_be_lun.maxlba = be_lun->size_blocks - 1; + be_lun->ctl_be_lun.atomicblock = UINT32_MAX; + be_lun->ctl_be_lun.opttxferlen = softc->rd_size / blocksize; } else { be_lun->ctl_be_lun.maxlba = 0; blocksize = 0; @@ -596,7 +597,6 @@ ctl_backend_ramdisk_create(struct ctl_be_ramdisk_softc *softc, be_lun->ctl_be_lun.flags = CTL_LUN_FLAG_PRIMARY; if (unmap) be_lun->ctl_be_lun.flags |= CTL_LUN_FLAG_UNMAP; - be_lun->ctl_be_lun.atomicblock = UINT32_MAX; be_lun->ctl_be_lun.be_lun = be_lun; if (params->flags & CTL_LUN_FLAG_ID_REQ) { @@ -613,32 +613,32 @@ ctl_backend_ramdisk_create(struct ctl_be_ramdisk_softc *softc, snprintf(tmpstr, sizeof(tmpstr), "MYSERIAL%4d", softc->num_luns); strncpy((char *)be_lun->ctl_be_lun.serial_num, tmpstr, - ctl_min(sizeof(be_lun->ctl_be_lun.serial_num), - sizeof(tmpstr))); + MIN(sizeof(be_lun->ctl_be_lun.serial_num), + sizeof(tmpstr))); /* Tell the user what we used for a serial number */ strncpy((char *)params->serial_num, tmpstr, - ctl_min(sizeof(params->serial_num), sizeof(tmpstr))); + MIN(sizeof(params->serial_num), sizeof(tmpstr))); } else { strncpy((char *)be_lun->ctl_be_lun.serial_num, params->serial_num, - ctl_min(sizeof(be_lun->ctl_be_lun.serial_num), - sizeof(params->serial_num))); + MIN(sizeof(be_lun->ctl_be_lun.serial_num), + sizeof(params->serial_num))); } if ((params->flags & CTL_LUN_FLAG_DEVID) == 0) { snprintf(tmpstr, sizeof(tmpstr), "MYDEVID%4d", softc->num_luns); strncpy((char *)be_lun->ctl_be_lun.device_id, tmpstr, - ctl_min(sizeof(be_lun->ctl_be_lun.device_id), - sizeof(tmpstr))); + MIN(sizeof(be_lun->ctl_be_lun.device_id), + sizeof(tmpstr))); /* Tell the user what we used for a device ID */ strncpy((char *)params->device_id, tmpstr, - ctl_min(sizeof(params->device_id), sizeof(tmpstr))); + MIN(sizeof(params->device_id), sizeof(tmpstr))); } else { strncpy((char *)be_lun->ctl_be_lun.device_id, params->device_id, - ctl_min(sizeof(be_lun->ctl_be_lun.device_id), - sizeof(params->device_id))); + MIN(sizeof(be_lun->ctl_be_lun.device_id), + sizeof(params->device_id))); } STAILQ_INIT(&be_lun->cont_queue); diff --git a/sys/cam/ctl/ctl_frontend.c b/sys/cam/ctl/ctl_frontend.c index 52dc31c..982675e 100644 --- a/sys/cam/ctl/ctl_frontend.c +++ b/sys/cam/ctl/ctl_frontend.c @@ -68,21 +68,22 @@ extern struct ctl_softc *control_softc; int ctl_frontend_register(struct ctl_frontend *fe) { + struct ctl_softc *softc = control_softc; struct ctl_frontend *fe_tmp; - KASSERT(control_softc != NULL, ("CTL is not initialized")); + KASSERT(softc != NULL, ("CTL is not initialized")); /* * Sanity check, make sure this isn't a duplicate registration. */ - mtx_lock(&control_softc->ctl_lock); - STAILQ_FOREACH(fe_tmp, &control_softc->fe_list, links) { + mtx_lock(&softc->ctl_lock); + STAILQ_FOREACH(fe_tmp, &softc->fe_list, links) { if (strcmp(fe_tmp->name, fe->name) == 0) { - mtx_unlock(&control_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (-1); } } - mtx_unlock(&control_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); STAILQ_INIT(&fe->port_list); /* @@ -91,24 +92,25 @@ ctl_frontend_register(struct ctl_frontend *fe) if (fe->init != NULL) fe->init(); - mtx_lock(&control_softc->ctl_lock); - control_softc->num_frontends++; - STAILQ_INSERT_TAIL(&control_softc->fe_list, fe, links); - mtx_unlock(&control_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); + softc->num_frontends++; + STAILQ_INSERT_TAIL(&softc->fe_list, fe, links); + mtx_unlock(&softc->ctl_lock); return (0); } int ctl_frontend_deregister(struct ctl_frontend *fe) { + struct ctl_softc *softc = control_softc; if (!STAILQ_EMPTY(&fe->port_list)) return (-1); - mtx_lock(&control_softc->ctl_lock); - STAILQ_REMOVE(&control_softc->fe_list, fe, ctl_frontend, links); - control_softc->num_frontends--; - mtx_unlock(&control_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); + STAILQ_REMOVE(&softc->fe_list, fe, ctl_frontend, links); + softc->num_frontends--; + mtx_unlock(&softc->ctl_lock); /* * Call the frontend's shutdown routine. @@ -121,41 +123,42 @@ ctl_frontend_deregister(struct ctl_frontend *fe) struct ctl_frontend * ctl_frontend_find(char *frontend_name) { - struct ctl_softc *ctl_softc = control_softc; + struct ctl_softc *softc = control_softc; struct ctl_frontend *fe; - mtx_lock(&ctl_softc->ctl_lock); - STAILQ_FOREACH(fe, &ctl_softc->fe_list, links) { + mtx_lock(&softc->ctl_lock); + STAILQ_FOREACH(fe, &softc->fe_list, links) { if (strcmp(fe->name, frontend_name) == 0) { - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (fe); } } - mtx_unlock(&ctl_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (NULL); } int ctl_port_register(struct ctl_port *port) { + struct ctl_softc *softc = control_softc; void *pool; int port_num; int retval; retval = 0; - KASSERT(control_softc != NULL, ("CTL is not initialized")); + KASSERT(softc != NULL, ("CTL is not initialized")); - mtx_lock(&control_softc->ctl_lock); - port_num = ctl_ffz(control_softc->ctl_port_mask, CTL_MAX_PORTS); + mtx_lock(&softc->ctl_lock); + port_num = ctl_ffz(softc->ctl_port_mask, CTL_MAX_PORTS); if ((port_num == -1) - || (ctl_set_mask(control_softc->ctl_port_mask, port_num) == -1)) { + || (ctl_set_mask(softc->ctl_port_mask, port_num) == -1)) { port->targ_port = -1; - mtx_unlock(&control_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); return (1); } - control_softc->num_ports++; - mtx_unlock(&control_softc->ctl_lock); + softc->num_ports++; + mtx_unlock(&softc->ctl_lock); /* * Initialize the initiator and portname mappings @@ -176,15 +179,15 @@ ctl_port_register(struct ctl_port *port) * pending sense queue on the next command, whether or not it is * a REQUEST SENSE. */ - retval = ctl_pool_create(control_softc, port->port_name, + retval = ctl_pool_create(softc, port->port_name, port->num_requested_ctl_io + 20, &pool); if (retval != 0) { free(port->wwpn_iid, M_CTL); error: port->targ_port = -1; - mtx_lock(&control_softc->ctl_lock); - ctl_clear_mask(control_softc->ctl_port_mask, port_num); - mtx_unlock(&control_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); + ctl_clear_mask(softc->ctl_port_mask, port_num); + mtx_unlock(&softc->ctl_lock); return (retval); } port->ctl_pool_ref = pool; @@ -192,12 +195,12 @@ error: if (port->options.stqh_first == NULL) STAILQ_INIT(&port->options); - mtx_lock(&control_softc->ctl_lock); - port->targ_port = port_num + control_softc->port_offset; + mtx_lock(&softc->ctl_lock); + port->targ_port = port_num + softc->port_offset; STAILQ_INSERT_TAIL(&port->frontend->port_list, port, fe_links); - STAILQ_INSERT_TAIL(&control_softc->port_list, port, links); - control_softc->ctl_ports[port_num] = port; - mtx_unlock(&control_softc->ctl_lock); + STAILQ_INSERT_TAIL(&softc->port_list, port, links); + softc->ctl_ports[port_num] = port; + mtx_unlock(&softc->ctl_lock); return (retval); } @@ -205,6 +208,7 @@ error: int ctl_port_deregister(struct ctl_port *port) { + struct ctl_softc *softc = control_softc; struct ctl_io_pool *pool; int port_num, retval, i; @@ -217,15 +221,15 @@ ctl_port_deregister(struct ctl_port *port) goto bailout; } - mtx_lock(&control_softc->ctl_lock); - STAILQ_REMOVE(&control_softc->port_list, port, ctl_port, links); + mtx_lock(&softc->ctl_lock); + STAILQ_REMOVE(&softc->port_list, port, ctl_port, links); STAILQ_REMOVE(&port->frontend->port_list, port, ctl_port, fe_links); - control_softc->num_ports--; + softc->num_ports--; port_num = (port->targ_port < CTL_MAX_PORTS) ? port->targ_port : port->targ_port - CTL_MAX_PORTS; - ctl_clear_mask(control_softc->ctl_port_mask, port_num); - control_softc->ctl_ports[port_num] = NULL; - mtx_unlock(&control_softc->ctl_lock); + ctl_clear_mask(softc->ctl_port_mask, port_num); + softc->ctl_ports[port_num] = NULL; + mtx_unlock(&softc->ctl_lock); ctl_pool_free(pool); ctl_free_opts(&port->options); diff --git a/sys/cam/ctl/ctl_frontend_cam_sim.c b/sys/cam/ctl/ctl_frontend_cam_sim.c index cb17aec..e779e48 100644 --- a/sys/cam/ctl/ctl_frontend_cam_sim.c +++ b/sys/cam/ctl/ctl_frontend_cam_sim.c @@ -401,8 +401,8 @@ cfcs_datamove(union ctl_io *io) i < cam_sg_count && j < ctl_sg_count;) { uint8_t *cam_ptr, *ctl_ptr; - len_to_copy = ctl_min(cam_sglist[i].ds_len - cam_watermark, - ctl_sglist[j].len - ctl_watermark); + len_to_copy = MIN(cam_sglist[i].ds_len - cam_watermark, + ctl_sglist[j].len - ctl_watermark); cam_ptr = (uint8_t *)cam_sglist[i].ds_addr; cam_ptr = cam_ptr + cam_watermark; diff --git a/sys/cam/ctl/ctl_frontend_internal.c b/sys/cam/ctl/ctl_frontend_internal.c index 7e7adda..23bd0cf 100644 --- a/sys/cam/ctl/ctl_frontend_internal.c +++ b/sys/cam/ctl/ctl_frontend_internal.c @@ -499,8 +499,8 @@ cfi_datamove(union ctl_io *io) i < ext_sg_entries && j < kern_sg_entries;) { uint8_t *ext_ptr, *kern_ptr; - len_to_copy = ctl_min(ext_sglist[i].len - ext_watermark, - kern_sglist[j].len - kern_watermark); + len_to_copy = MIN(ext_sglist[i].len - ext_watermark, + kern_sglist[j].len - kern_watermark); ext_ptr = (uint8_t *)ext_sglist[i].addr; ext_ptr = ext_ptr + ext_watermark; @@ -1103,8 +1103,8 @@ cfi_metatask_bbr_errorparse(struct cfi_metatask *metatask, union ctl_io *io) metatask->taskinfo.bbrread.scsi_status = io->scsiio.scsi_status; memcpy(&metatask->taskinfo.bbrread.sense_data, &io->scsiio.sense_data, - ctl_min(sizeof(metatask->taskinfo.bbrread.sense_data), - sizeof(io->scsiio.sense_data))); + MIN(sizeof(metatask->taskinfo.bbrread.sense_data), + sizeof(io->scsiio.sense_data))); if (io->scsiio.scsi_status == SCSI_STATUS_RESERV_CONFLICT) { metatask->status = CFI_MT_ERROR; diff --git a/sys/cam/ctl/ctl_tpc.c b/sys/cam/ctl/ctl_tpc.c index 90a31ae..4e67ce1 100644 --- a/sys/cam/ctl/ctl_tpc.c +++ b/sys/cam/ctl/ctl_tpc.c @@ -145,8 +145,6 @@ struct tpc_list { TAILQ_ENTRY(tpc_list) links; }; -extern struct ctl_softc *control_softc; - static void tpc_timeout(void *arg) { @@ -216,6 +214,7 @@ ctl_tpc_lun_init(struct ctl_lun *lun) void ctl_tpc_lun_shutdown(struct ctl_lun *lun) { + struct ctl_softc *softc = lun->ctl_softc; struct tpc_list *list; struct tpc_token *token, *ttoken; @@ -228,11 +227,11 @@ ctl_tpc_lun_shutdown(struct ctl_lun *lun) } /* Free ROD tokens for this LUN. */ - mtx_assert(&control_softc->ctl_lock, MA_OWNED); - TAILQ_FOREACH_SAFE(token, &control_softc->tpc_tokens, links, ttoken) { + mtx_assert(&softc->ctl_lock, MA_OWNED); + TAILQ_FOREACH_SAFE(token, &softc->tpc_tokens, links, ttoken) { if (token->lun != lun->lun || token->active) continue; - TAILQ_REMOVE(&control_softc->tpc_tokens, token, links); + TAILQ_REMOVE(&softc->tpc_tokens, token, links); free(token->params, M_CTL); free(token, M_CTL); } @@ -796,7 +795,8 @@ tpc_resolve(struct tpc_list *list, uint16_t idx, uint32_t *ss) } if (idx >= list->ncscd) return (UINT64_MAX); - return (tpcl_resolve(list->init_port, &list->cscd[idx], ss)); + return (tpcl_resolve(list->lun->ctl_softc, + list->init_port, &list->cscd[idx], ss)); } static int @@ -1296,6 +1296,7 @@ static void tpc_process(struct tpc_list *list) { struct ctl_lun *lun = list->lun; + struct ctl_softc *softc = lun->ctl_softc; struct scsi_ec_segment *seg; struct ctl_scsiio *ctsio = list->ctsio; int retval = CTL_RETVAL_COMPLETE; @@ -1349,10 +1350,10 @@ done: free(list->params, M_CTL); list->params = NULL; if (list->token) { - mtx_lock(&control_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); if (--list->token->active == 0) list->token->last_active = time_uptime; - mtx_unlock(&control_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); list->token = NULL; } mtx_lock(&lun->lun_lock); @@ -1831,6 +1832,7 @@ ctl_populate_token(struct ctl_scsiio *ctsio) { struct scsi_populate_token *cdb; struct scsi_populate_token_data *data; + struct ctl_softc *softc; struct ctl_lun *lun; struct ctl_port *port; struct tpc_list *list, *tlist; @@ -1840,7 +1842,8 @@ ctl_populate_token(struct ctl_scsiio *ctsio) CTL_DEBUG_PRINT(("ctl_populate_token\n")); lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; - port = control_softc->ctl_ports[ctl_port_idx(ctsio->io_hdr.nexus.targ_port)]; + softc = lun->ctl_softc; + port = softc->ctl_ports[ctl_port_idx(ctsio->io_hdr.nexus.targ_port)]; cdb = (struct scsi_populate_token *)ctsio->cdb; len = scsi_4btoul(cdb->length); @@ -1944,9 +1947,9 @@ ctl_populate_token(struct ctl_scsiio *ctsio) list->curseg = 0; list->completed = 1; list->last_active = time_uptime; - mtx_lock(&control_softc->ctl_lock); - TAILQ_INSERT_TAIL(&control_softc->tpc_tokens, token, links); - mtx_unlock(&control_softc->ctl_lock); + mtx_lock(&softc->ctl_lock); + TAILQ_INSERT_TAIL(&softc->tpc_tokens, token, links); + mtx_unlock(&softc->ctl_lock); ctl_set_success(ctsio); ctl_done((union ctl_io *)ctsio); return (CTL_RETVAL_COMPLETE); @@ -1965,6 +1968,7 @@ ctl_write_using_token(struct ctl_scsiio *ctsio) { struct scsi_write_using_token *cdb; struct scsi_write_using_token_data *data; + struct ctl_softc *softc; struct ctl_lun *lun; struct tpc_list *list, *tlist; struct tpc_token *token; @@ -1973,6 +1977,7 @@ ctl_write_using_token(struct ctl_scsiio *ctsio) CTL_DEBUG_PRINT(("ctl_write_using_token\n")); lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; + softc = lun->ctl_softc; cdb = (struct scsi_write_using_token *)ctsio->cdb; len = scsi_4btoul(cdb->length); @@ -2051,8 +2056,8 @@ ctl_write_using_token(struct ctl_scsiio *ctsio) return (CTL_RETVAL_COMPLETE); } - mtx_lock(&control_softc->ctl_lock); - TAILQ_FOREACH(token, &control_softc->tpc_tokens, links) { + mtx_lock(&softc->ctl_lock); + TAILQ_FOREACH(token, &softc->tpc_tokens, links) { if (memcmp(token->token, data->rod_token, sizeof(data->rod_token)) == 0) break; @@ -2063,7 +2068,7 @@ ctl_write_using_token(struct ctl_scsiio *ctsio) if (data->flags & EC_WUT_DEL_TKN) token->timeout = 0; } - mtx_unlock(&control_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); if (token == NULL) { mtx_lock(&lun->lun_lock); TAILQ_REMOVE(&lun->tpc_lists, list, links); @@ -2188,6 +2193,7 @@ ctl_receive_rod_token_information(struct ctl_scsiio *ctsio) int ctl_report_all_rod_tokens(struct ctl_scsiio *ctsio) { + struct ctl_softc *softc; struct ctl_lun *lun; struct scsi_report_all_rod_tokens *cdb; struct scsi_report_all_rod_tokens_data *data; @@ -2199,14 +2205,15 @@ ctl_report_all_rod_tokens(struct ctl_scsiio *ctsio) cdb = (struct scsi_report_all_rod_tokens *)ctsio->cdb; lun = (struct ctl_lun *)ctsio->io_hdr.ctl_private[CTL_PRIV_LUN].ptr; + softc = lun->ctl_softc; retval = CTL_RETVAL_COMPLETE; tokens = 0; - mtx_lock(&control_softc->ctl_lock); - TAILQ_FOREACH(token, &control_softc->tpc_tokens, links) + mtx_lock(&softc->ctl_lock); + TAILQ_FOREACH(token, &softc->tpc_tokens, links) tokens++; - mtx_unlock(&control_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); if (tokens > 512) tokens = 512; @@ -2231,15 +2238,15 @@ ctl_report_all_rod_tokens(struct ctl_scsiio *ctsio) data = (struct scsi_report_all_rod_tokens_data *)ctsio->kern_data_ptr; i = 0; - mtx_lock(&control_softc->ctl_lock); - TAILQ_FOREACH(token, &control_softc->tpc_tokens, links) { + mtx_lock(&softc->ctl_lock); + TAILQ_FOREACH(token, &softc->tpc_tokens, links) { if (i >= tokens) break; memcpy(&data->rod_management_token_list[i * 96], token->token, 96); i++; } - mtx_unlock(&control_softc->ctl_lock); + mtx_unlock(&softc->ctl_lock); scsi_ulto4b(sizeof(*data) - 4 + i * 96, data->available_data); /* printf("RART tokens=%d\n", i); diff --git a/sys/cam/ctl/ctl_tpc.h b/sys/cam/ctl/ctl_tpc.h index ecbaec1..ffdab5a 100644 --- a/sys/cam/ctl/ctl_tpc.h +++ b/sys/cam/ctl/ctl_tpc.h @@ -31,7 +31,8 @@ void tpc_done(union ctl_io *io); -uint64_t tpcl_resolve(int init_port, struct scsi_ec_cscd *cscd, uint32_t *ss); +uint64_t tpcl_resolve(struct ctl_softc *softc, int init_port, + struct scsi_ec_cscd *cscd, uint32_t *ss); union ctl_io * tpcl_alloc_io(void); int tpcl_queue(union ctl_io *io, uint64_t lun); diff --git a/sys/cam/ctl/ctl_tpc_local.c b/sys/cam/ctl/ctl_tpc_local.c index 97a5f98..63360fe 100644 --- a/sys/cam/ctl/ctl_tpc_local.c +++ b/sys/cam/ctl/ctl_tpc_local.c @@ -63,7 +63,6 @@ struct tpcl_softc { int cur_tag_num; }; -extern struct ctl_softc *control_softc; static struct tpcl_softc tpcl_softc; static int tpcl_init(void); @@ -309,9 +308,9 @@ tpcl_done(union ctl_io *io) } uint64_t -tpcl_resolve(int init_port, struct scsi_ec_cscd *cscd, uint32_t *ss) +tpcl_resolve(struct ctl_softc *softc, int init_port, + struct scsi_ec_cscd *cscd, uint32_t *ss) { - struct ctl_softc *softc = control_softc; struct scsi_ec_cscd_id *cscdid; struct ctl_port *port; struct ctl_lun *lun; diff --git a/sys/cam/ctl/scsi_ctl.c b/sys/cam/ctl/scsi_ctl.c index a1f39d2..b1dba8c 100644 --- a/sys/cam/ctl/scsi_ctl.c +++ b/sys/cam/ctl/scsi_ctl.c @@ -228,8 +228,6 @@ static struct ctl_frontend ctlfe_frontend = }; CTL_FRONTEND_DECLARE(ctlfe, ctlfe_frontend); -extern struct ctl_softc *control_softc; - void ctlfeshutdown(void) { diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dnode.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dnode.c index 66552c3..665165a 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dnode.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/dnode.c @@ -81,16 +81,14 @@ dbuf_compare(const void *x1, const void *x2) return (1); } - if (d1->db_state < d2->db_state) { + if (d1->db_state == DB_SEARCH) { + ASSERT3S(d2->db_state, !=, DB_SEARCH); return (-1); - } - if (d1->db_state > d2->db_state) { + } else if (d2->db_state == DB_SEARCH) { + ASSERT3S(d1->db_state, !=, DB_SEARCH); return (1); } - ASSERT3S(d1->db_state, !=, DB_SEARCH); - ASSERT3S(d2->db_state, !=, DB_SEARCH); - if ((uintptr_t)d1 < (uintptr_t)d2) { return (-1); } diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa_config.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa_config.c index 2d0cf23..c4ee741 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa_config.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/spa_config.c @@ -536,8 +536,7 @@ spa_config_update(spa_t *spa, int what) /* * Update the global config cache to reflect the new mosconfig. */ - if (!spa->spa_is_root) - spa_config_sync(spa, B_FALSE, what != SPA_CONFIG_UPDATE_POOL); + spa_config_sync(spa, B_FALSE, what != SPA_CONFIG_UPDATE_POOL); if (what == SPA_CONFIG_UPDATE_POOL) spa_config_update(spa, SPA_CONFIG_UPDATE_VDEVS); diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zvol.h b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zvol.h index 52b8f30..75c6b37 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zvol.h +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/sys/zvol.h @@ -43,7 +43,7 @@ extern void zvol_create_cb(objset_t *os, void *arg, cred_t *cr, dmu_tx_t *tx); extern int zvol_create_minor(const char *); extern int zvol_remove_minor(const char *); extern void zvol_remove_minors(const char *); -extern int zvol_set_volsize(const char *, major_t, uint64_t); +extern int zvol_set_volsize(const char *, uint64_t); #ifdef sun extern int zvol_open(dev_t *devp, int flag, int otyp, cred_t *cr); diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c index d730083..383b789 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_ioctl.c @@ -2482,8 +2482,7 @@ zfs_prop_set_special(const char *dsname, zprop_source_t source, err = dsl_dataset_set_refreservation(dsname, source, intval); break; case ZFS_PROP_VOLSIZE: - err = zvol_set_volsize(dsname, ddi_driver_major(zfs_dip), - intval); + err = zvol_set_volsize(dsname, intval); break; case ZFS_PROP_VERSION: { diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c index 5d7fa70..9b11652 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c @@ -6180,15 +6180,20 @@ zfs_freebsd_create(ap) { struct componentname *cnp = ap->a_cnp; vattr_t *vap = ap->a_vap; - int mode; + int error, mode; ASSERT(cnp->cn_flags & SAVENAME); vattr_init_mask(vap); mode = vap->va_mode & ALLPERMS; - return (zfs_create(ap->a_dvp, cnp->cn_nameptr, vap, !EXCL, mode, - ap->a_vpp, cnp->cn_cred, cnp->cn_thread)); + error = zfs_create(ap->a_dvp, cnp->cn_nameptr, vap, !EXCL, mode, + ap->a_vpp, cnp->cn_cred, cnp->cn_thread); +#ifdef FREEBSD_NAMECACHE + if (error == 0 && (cnp->cn_flags & MAKEENTRY) != 0) + cache_enter(ap->a_dvp, *ap->a_vpp, cnp); +#endif + return (error); } static int diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zio.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zio.c index 3925e90..afda3e4 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zio.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zio.c @@ -704,18 +704,20 @@ zfs_blkptr_verify(spa_t *spa, const blkptr_t *bp) zfs_panic_recover("blkptr at %p DVA %u has invalid " "VDEV %llu", bp, i, (longlong_t)vdevid); + continue; } vdev_t *vd = spa->spa_root_vdev->vdev_child[vdevid]; if (vd == NULL) { zfs_panic_recover("blkptr at %p DVA %u has invalid " "VDEV %llu", bp, i, (longlong_t)vdevid); + continue; } if (vd->vdev_ops == &vdev_hole_ops) { zfs_panic_recover("blkptr at %p DVA %u has hole " "VDEV %llu", bp, i, (longlong_t)vdevid); - + continue; } if (vd->vdev_ops == &vdev_missing_ops) { /* diff --git a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zvol.c b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zvol.c index 260822c..f9cc3eb 100644 --- a/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zvol.c +++ b/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zvol.c @@ -97,6 +97,7 @@ #include "zfs_namecheck.h" +#ifndef illumos struct g_class zfs_zvol_class = { .name = "ZFS::ZVOL", .version = G_VERSION, @@ -104,25 +105,37 @@ struct g_class zfs_zvol_class = { DECLARE_GEOM_CLASS(zfs_zvol_class, zfs_zvol); +#endif void *zfsdev_state; static char *zvol_tag = "zvol_tag"; #define ZVOL_DUMPSIZE "dumpsize" /* - * The spa_namespace_lock protects the zfsdev_state structure from being - * modified while it's being used, e.g. an open that comes in before a - * create finishes. It also protects temporary opens of the dataset so that, + * This lock protects the zfsdev_state structure from being modified + * while it's being used, e.g. an open that comes in before a create + * finishes. It also protects temporary opens of the dataset so that, * e.g., an open doesn't get a spurious EBUSY. */ +#ifdef illumos +kmutex_t zfsdev_state_lock; +#else +/* + * In FreeBSD we've replaced the upstream zfsdev_state_lock with the + * spa_namespace_lock in the ZVOL code. + */ +#define zfsdev_state_lock spa_namespace_lock +#endif static uint32_t zvol_minors; +#ifndef illumos SYSCTL_DECL(_vfs_zfs); SYSCTL_NODE(_vfs_zfs, OID_AUTO, vol, CTLFLAG_RW, 0, "ZFS VOLUME"); static int volmode = ZFS_VOLMODE_GEOM; SYSCTL_INT(_vfs_zfs_vol, OID_AUTO, mode, CTLFLAG_RWTUN, &volmode, 0, "Expose as GEOM providers (1), device files (2) or neither"); +#endif typedef struct zvol_extent { list_node_t ze_node; dva_t ze_dva; /* dva associated with this extent */ @@ -133,28 +146,40 @@ typedef struct zvol_extent { * The in-core state of each volume. */ typedef struct zvol_state { +#ifndef illumos LIST_ENTRY(zvol_state) zv_links; +#endif char zv_name[MAXPATHLEN]; /* pool/dd name */ uint64_t zv_volsize; /* amount of space we advertise */ uint64_t zv_volblocksize; /* volume block size */ +#ifdef illumos + minor_t zv_minor; /* minor number */ +#else struct cdev *zv_dev; /* non-GEOM device */ struct g_provider *zv_provider; /* GEOM provider */ +#endif uint8_t zv_min_bs; /* minimum addressable block shift */ uint8_t zv_flags; /* readonly, dumpified, etc. */ objset_t *zv_objset; /* objset handle */ +#ifdef illumos + uint32_t zv_open_count[OTYPCNT]; /* open counts */ +#endif uint32_t zv_total_opens; /* total open count */ zilog_t *zv_zilog; /* ZIL handle */ list_t zv_extents; /* List of extents for dump */ znode_t zv_znode; /* for range locking */ dmu_buf_t *zv_dbuf; /* bonus handle */ +#ifndef illumos int zv_state; int zv_volmode; /* Provide GEOM or cdev */ struct bio_queue_head zv_queue; struct mtx zv_queue_mtx; /* zv_queue mutex */ +#endif } zvol_state_t; +#ifndef illumos static LIST_HEAD(, zvol_state) all_zvols; - +#endif /* * zvol specific flags */ @@ -172,6 +197,7 @@ int zvol_maxphys = DMU_MAX_ACCESS/2; * Toggle unmap functionality. */ boolean_t zvol_unmap_enabled = B_TRUE; +#ifndef illumos SYSCTL_INT(_vfs_zfs_vol, OID_AUTO, unmap_enabled, CTLFLAG_RWTUN, &zvol_unmap_enabled, 0, "Enable UNMAP functionality"); @@ -195,28 +221,30 @@ static struct cdevsw zvol_cdevsw = { .d_flags = D_DISK | D_TRACKCLOSE, }; -extern int zfs_set_prop_nvlist(const char *, zprop_source_t, - nvlist_t *, nvlist_t *); +static void zvol_geom_run(zvol_state_t *zv); +static void zvol_geom_destroy(zvol_state_t *zv); +static int zvol_geom_access(struct g_provider *pp, int acr, int acw, int ace); +static void zvol_geom_start(struct bio *bp); +static void zvol_geom_worker(void *arg); static void zvol_log_truncate(zvol_state_t *zv, dmu_tx_t *tx, uint64_t off, uint64_t len, boolean_t sync); +#endif /* !illumos */ + +extern int zfs_set_prop_nvlist(const char *, zprop_source_t, + nvlist_t *, nvlist_t *); static int zvol_remove_zv(zvol_state_t *); static int zvol_get_data(void *arg, lr_write_t *lr, char *buf, zio_t *zio); static int zvol_dumpify(zvol_state_t *zv); static int zvol_dump_fini(zvol_state_t *zv); static int zvol_dump_init(zvol_state_t *zv, boolean_t resize); -static void zvol_geom_run(zvol_state_t *zv); -static void zvol_geom_destroy(zvol_state_t *zv); -static int zvol_geom_access(struct g_provider *pp, int acr, int acw, int ace); -static void zvol_geom_start(struct bio *bp); -static void zvol_geom_worker(void *arg); - static void -zvol_size_changed(zvol_state_t *zv) +zvol_size_changed(zvol_state_t *zv, uint64_t volsize) { -#ifdef sun - dev_t dev = makedevice(maj, min); +#ifdef illumos + dev_t dev = makedevice(ddi_driver_major(zfs_dip), zv->zv_minor); + zv->zv_volsize = volsize; VERIFY(ddi_prop_update_int64(dev, zfs_dip, "Size", volsize) == DDI_SUCCESS); VERIFY(ddi_prop_update_int64(dev, zfs_dip, @@ -225,7 +253,8 @@ zvol_size_changed(zvol_state_t *zv) /* Notify specfs to invalidate the cached size */ spec_size_invalidate(dev, VBLK); spec_size_invalidate(dev, VCHR); -#else /* !sun */ +#else /* !illumos */ + zv->zv_volsize = volsize; if (zv->zv_volmode == ZFS_VOLMODE_GEOM) { struct g_provider *pp; @@ -236,7 +265,7 @@ zvol_size_changed(zvol_state_t *zv) g_resize_provider(pp, zv->zv_volsize); g_topology_unlock(); } -#endif /* !sun */ +#endif /* illumos */ } int @@ -292,16 +321,26 @@ zvol_get_stats(objset_t *os, nvlist_t *nv) static zvol_state_t * zvol_minor_lookup(const char *name) { +#ifdef illumos + minor_t minor; +#endif zvol_state_t *zv; - ASSERT(MUTEX_HELD(&spa_namespace_lock)); + ASSERT(MUTEX_HELD(&zfsdev_state_lock)); +#ifdef illumos + for (minor = 1; minor <= ZFSDEV_MAX_MINOR; minor++) { + zv = zfsdev_get_soft_state(minor, ZSST_ZVOL); + if (zv == NULL) + continue; +#else LIST_FOREACH(zv, &all_zvols, zv_links) { +#endif if (strcmp(zv->zv_name, name) == 0) - break; + return (zv); } - return (zv); + return (NULL); } /* extent mapping arg */ @@ -517,20 +556,20 @@ zil_replay_func_t *zvol_replay_vector[TX_MAX_TYPE] = { zvol_replay_err, /* TX_WRITE2 */ }; -#ifdef sun +#ifdef illumos int zvol_name2minor(const char *name, minor_t *minor) { zvol_state_t *zv; - mutex_enter(&spa_namespace_lock); + mutex_enter(&zfsdev_state_lock); zv = zvol_minor_lookup(name); if (minor && zv) *minor = zv->zv_minor; - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (zv ? 0 : -1); } -#endif /* sun */ +#endif /* illumos */ /* * Create a minor node (plus a whole lot more) for the specified volume. @@ -541,19 +580,26 @@ zvol_create_minor(const char *name) zfs_soft_state_t *zs; zvol_state_t *zv; objset_t *os; + dmu_object_info_t doi; +#ifdef illumos + minor_t minor = 0; + char chrbuf[30], blkbuf[30]; +#else struct cdev *dev; struct g_provider *pp; struct g_geom *gp; - dmu_object_info_t doi; uint64_t volsize, mode; +#endif int error; +#ifndef illumos ZFS_LOG(1, "Creating ZVOL %s...", name); +#endif - mutex_enter(&spa_namespace_lock); + mutex_enter(&zfsdev_state_lock); if (zvol_minor_lookup(name) != NULL) { - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (SET_ERROR(EEXIST)); } @@ -561,20 +607,20 @@ zvol_create_minor(const char *name) error = dmu_objset_own(name, DMU_OST_ZVOL, B_TRUE, FTAG, &os); if (error) { - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (error); } -#ifdef sun +#ifdef illumos if ((minor = zfsdev_minor_alloc()) == 0) { dmu_objset_disown(os, FTAG); - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (SET_ERROR(ENXIO)); } if (ddi_soft_state_zalloc(zfsdev_state, minor) != DDI_SUCCESS) { dmu_objset_disown(os, FTAG); - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (SET_ERROR(EAGAIN)); } (void) ddi_prop_update_string(minor, zfs_dip, ZVOL_PROP_NAME, @@ -586,7 +632,7 @@ zvol_create_minor(const char *name) minor, DDI_PSEUDO, 0) == DDI_FAILURE) { ddi_soft_state_free(zfsdev_state, minor); dmu_objset_disown(os, FTAG); - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (SET_ERROR(EAGAIN)); } @@ -597,14 +643,14 @@ zvol_create_minor(const char *name) ddi_remove_minor_node(zfs_dip, chrbuf); ddi_soft_state_free(zfsdev_state, minor); dmu_objset_disown(os, FTAG); - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (SET_ERROR(EAGAIN)); } zs = ddi_get_soft_state(zfsdev_state, minor); zs->zss_type = ZSST_ZVOL; zv = zs->zss_data = kmem_zalloc(sizeof (zvol_state_t), KM_SLEEP); -#else /* !sun */ +#else /* !illumos */ zv = kmem_zalloc(sizeof(*zv), KM_SLEEP); zv->zv_state = 0; @@ -612,7 +658,7 @@ zvol_create_minor(const char *name) if (error) { kmem_free(zv, sizeof(*zv)); dmu_objset_disown(os, zvol_tag); - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (error); } error = dsl_prop_get_integer(name, @@ -643,7 +689,7 @@ zvol_create_minor(const char *name) 0640, "%s/%s", ZVOL_DRIVER, name) != 0) { kmem_free(zv, sizeof(*zv)); dmu_objset_disown(os, FTAG); - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (SET_ERROR(ENXIO)); } zv->zv_dev = dev; @@ -651,10 +697,13 @@ zvol_create_minor(const char *name) dev->si_drv2 = zv; } LIST_INSERT_HEAD(&all_zvols, zv, zv_links); -#endif /* !sun */ +#endif /* illumos */ (void) strlcpy(zv->zv_name, name, MAXPATHLEN); zv->zv_min_bs = DEV_BSHIFT; +#ifdef illumos + zv->zv_minor = minor; +#endif zv->zv_objset = os; if (dmu_objset_is_snapshot(os) || !spa_writeable(dmu_objset_spa(os))) zv->zv_flags |= ZVOL_RDONLY; @@ -679,17 +728,16 @@ zvol_create_minor(const char *name) zvol_minors++; - mutex_exit(&spa_namespace_lock); - -#ifndef sun + mutex_exit(&zfsdev_state_lock); +#ifndef illumos if (zv->zv_volmode == ZFS_VOLMODE_GEOM) { zvol_geom_run(zv); g_topology_unlock(); } PICKUP_GIANT(); -#endif ZFS_LOG(1, "ZVOL %s created.", name); +#endif return (0); } @@ -700,20 +748,24 @@ zvol_create_minor(const char *name) static int zvol_remove_zv(zvol_state_t *zv) { -#ifdef sun +#ifdef illumos + char nmbuf[20]; minor_t minor = zv->zv_minor; #endif - ASSERT(MUTEX_HELD(&spa_namespace_lock)); + ASSERT(MUTEX_HELD(&zfsdev_state_lock)); if (zv->zv_total_opens != 0) return (SET_ERROR(EBUSY)); - ZFS_LOG(1, "ZVOL %s destroyed.", zv->zv_name); - -#ifdef sun +#ifdef illumos (void) snprintf(nmbuf, sizeof (nmbuf), "%u,raw", minor); ddi_remove_minor_node(zfs_dip, nmbuf); + + (void) snprintf(nmbuf, sizeof (nmbuf), "%u", minor); + ddi_remove_minor_node(zfs_dip, nmbuf); #else + ZFS_LOG(1, "ZVOL %s destroyed.", zv->zv_name); + LIST_REMOVE(zv, zv_links); if (zv->zv_volmode == ZFS_VOLMODE_GEOM) { g_topology_lock(); @@ -721,13 +773,15 @@ zvol_remove_zv(zvol_state_t *zv) g_topology_unlock(); } else if (zv->zv_volmode == ZFS_VOLMODE_DEV) destroy_dev(zv->zv_dev); -#endif /* sun */ +#endif avl_destroy(&zv->zv_znode.z_range_avl); mutex_destroy(&zv->zv_znode.z_range_lock); - kmem_free(zv, sizeof(*zv)); - + kmem_free(zv, sizeof (zvol_state_t)); +#ifdef illumos + ddi_soft_state_free(zfsdev_state, minor); +#endif zvol_minors--; return (0); } @@ -738,13 +792,13 @@ zvol_remove_minor(const char *name) zvol_state_t *zv; int rc; - mutex_enter(&spa_namespace_lock); + mutex_enter(&zfsdev_state_lock); if ((zv = zvol_minor_lookup(name)) == NULL) { - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (SET_ERROR(ENXIO)); } rc = zvol_remove_zv(zv); - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (rc); } @@ -762,21 +816,22 @@ zvol_first_open(zvol_state_t *zv) if (error) return (error); + zv->zv_objset = os; error = zap_lookup(os, ZVOL_ZAP_OBJ, "size", 8, 1, &volsize); if (error) { ASSERT(error == 0); dmu_objset_disown(os, zvol_tag); return (error); } - zv->zv_objset = os; + error = dmu_bonus_hold(os, ZVOL_OBJ, zvol_tag, &zv->zv_dbuf); if (error) { dmu_objset_disown(os, zvol_tag); return (error); } - zv->zv_volsize = volsize; + + zvol_size_changed(zv, volsize); zv->zv_zilog = zil_open(os, zvol_get_data); - zvol_size_changed(zv); VERIFY(dsl_prop_get_integer(zv->zv_name, "readonly", &readonly, NULL) == 0); @@ -809,7 +864,7 @@ zvol_last_close(zvol_state_t *zv) zv->zv_objset = NULL; } -#ifdef sun +#ifdef illumos int zvol_prealloc(zvol_state_t *zv) { @@ -848,7 +903,7 @@ zvol_prealloc(zvol_state_t *zv) return (0); } -#endif /* sun */ +#endif /* illumos */ static int zvol_update_volsize(objset_t *os, uint64_t volsize) @@ -856,7 +911,7 @@ zvol_update_volsize(objset_t *os, uint64_t volsize) dmu_tx_t *tx; int error; - ASSERT(MUTEX_HELD(&spa_namespace_lock)); + ASSERT(MUTEX_HELD(&zfsdev_state_lock)); tx = dmu_tx_create(os); dmu_tx_hold_zap(tx, ZVOL_ZAP_OBJ, TRUE, NULL); @@ -880,13 +935,34 @@ zvol_update_volsize(objset_t *os, uint64_t volsize) void zvol_remove_minors(const char *name) { +#ifdef illumos + zvol_state_t *zv; + char *namebuf; + minor_t minor; + + namebuf = kmem_zalloc(strlen(name) + 2, KM_SLEEP); + (void) strncpy(namebuf, name, strlen(name)); + (void) strcat(namebuf, "/"); + mutex_enter(&zfsdev_state_lock); + for (minor = 1; minor <= ZFSDEV_MAX_MINOR; minor++) { + + zv = zfsdev_get_soft_state(minor, ZSST_ZVOL); + if (zv == NULL) + continue; + if (strncmp(namebuf, zv->zv_name, strlen(namebuf)) == 0) + (void) zvol_remove_zv(zv); + } + kmem_free(namebuf, strlen(name) + 2); + + mutex_exit(&zfsdev_state_lock); +#else /* !illumos */ zvol_state_t *zv, *tzv; size_t namelen; namelen = strlen(name); DROP_GIANT(); - mutex_enter(&spa_namespace_lock); + mutex_enter(&zfsdev_state_lock); LIST_FOREACH_SAFE(zv, &all_zvols, zv_links, tzv) { if (strcmp(zv->zv_name, name) == 0 || @@ -897,69 +973,48 @@ zvol_remove_minors(const char *name) } } - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); PICKUP_GIANT(); +#endif /* illumos */ } -int -zvol_set_volsize(const char *name, major_t maj, uint64_t volsize) +static int +zvol_update_live_volsize(zvol_state_t *zv, uint64_t volsize) { - zvol_state_t *zv = NULL; - objset_t *os; - int error; - dmu_object_info_t doi; uint64_t old_volsize = 0ULL; - uint64_t readonly; - - mutex_enter(&spa_namespace_lock); - zv = zvol_minor_lookup(name); - if ((error = dmu_objset_hold(name, FTAG, &os)) != 0) { - mutex_exit(&spa_namespace_lock); - return (error); - } + int error = 0; - if ((error = dmu_object_info(os, ZVOL_OBJ, &doi)) != 0 || - (error = zvol_check_volsize(volsize, - doi.doi_data_block_size)) != 0) - goto out; + ASSERT(MUTEX_HELD(&zfsdev_state_lock)); - VERIFY(dsl_prop_get_integer(name, "readonly", &readonly, - NULL) == 0); - if (readonly) { - error = EROFS; - goto out; - } - - error = zvol_update_volsize(os, volsize); /* * Reinitialize the dump area to the new size. If we * failed to resize the dump area then restore it back to - * its original size. + * its original size. We must set the new volsize prior + * to calling dumpvp_resize() to ensure that the devices' + * size(9P) is not visible by the dump subsystem. */ - if (zv && error == 0) { + old_volsize = zv->zv_volsize; + zvol_size_changed(zv, volsize); + #ifdef ZVOL_DUMP - if (zv->zv_flags & ZVOL_DUMPIFIED) { - old_volsize = zv->zv_volsize; - zv->zv_volsize = volsize; - if ((error = zvol_dumpify(zv)) != 0 || - (error = dumpvp_resize()) != 0) { - (void) zvol_update_volsize(os, old_volsize); - zv->zv_volsize = old_volsize; - error = zvol_dumpify(zv); - } - } -#endif /* ZVOL_DUMP */ - if (error == 0) { - zv->zv_volsize = volsize; - zvol_size_changed(zv); + if (zv->zv_flags & ZVOL_DUMPIFIED) { + if ((error = zvol_dumpify(zv)) != 0 || + (error = dumpvp_resize()) != 0) { + int dumpify_error; + + (void) zvol_update_volsize(zv->zv_objset, old_volsize); + zvol_size_changed(zv, old_volsize); + dumpify_error = zvol_dumpify(zv); + error = dumpify_error ? dumpify_error : error; } } +#endif /* ZVOL_DUMP */ -#ifdef sun +#ifdef illumos /* * Generate a LUN expansion event. */ - if (zv && error == 0) { + if (error == 0) { sysevent_id_t eid; nvlist_t *attr; char *physpath = kmem_zalloc(MAXPATHLEN, KM_SLEEP); @@ -976,22 +1031,89 @@ zvol_set_volsize(const char *name, major_t maj, uint64_t volsize) nvlist_free(attr); kmem_free(physpath, MAXPATHLEN); } -#endif /* sun */ +#endif /* illumos */ + return (error); +} -out: - dmu_objset_rele(os, FTAG); +int +zvol_set_volsize(const char *name, uint64_t volsize) +{ + zvol_state_t *zv = NULL; + objset_t *os; + int error; + dmu_object_info_t doi; + uint64_t readonly; + boolean_t owned = B_FALSE; + + error = dsl_prop_get_integer(name, + zfs_prop_to_name(ZFS_PROP_READONLY), &readonly, NULL); + if (error != 0) + return (error); + if (readonly) + return (SET_ERROR(EROFS)); + + mutex_enter(&zfsdev_state_lock); + zv = zvol_minor_lookup(name); + + if (zv == NULL || zv->zv_objset == NULL) { + if ((error = dmu_objset_own(name, DMU_OST_ZVOL, B_FALSE, + FTAG, &os)) != 0) { + mutex_exit(&zfsdev_state_lock); + return (error); + } + owned = B_TRUE; + if (zv != NULL) + zv->zv_objset = os; + } else { + os = zv->zv_objset; + } + + if ((error = dmu_object_info(os, ZVOL_OBJ, &doi)) != 0 || + (error = zvol_check_volsize(volsize, doi.doi_data_block_size)) != 0) + goto out; - mutex_exit(&spa_namespace_lock); + error = zvol_update_volsize(os, volsize); + if (error == 0 && zv != NULL) + error = zvol_update_live_volsize(zv, volsize); +out: + if (owned) { + dmu_objset_disown(os, FTAG); + if (zv != NULL) + zv->zv_objset = NULL; + } + mutex_exit(&zfsdev_state_lock); return (error); } /*ARGSUSED*/ +#ifdef illumos +int +zvol_open(dev_t *devp, int flag, int otyp, cred_t *cr) +#else static int zvol_open(struct g_provider *pp, int flag, int count) +#endif { zvol_state_t *zv; int err = 0; +#ifdef illumos + + mutex_enter(&zfsdev_state_lock); + + zv = zfsdev_get_soft_state(getminor(*devp), ZSST_ZVOL); + if (zv == NULL) { + mutex_exit(&zfsdev_state_lock); + return (SET_ERROR(ENXIO)); + } + + if (zv->zv_total_opens == 0) + err = zvol_first_open(zv); + if (err) { + mutex_exit(&zfsdev_state_lock); + return (err); + } +#else /* !illumos */ boolean_t locked = B_FALSE; /* @@ -1005,15 +1127,15 @@ zvol_open(struct g_provider *pp, int flag, int count) * recursively, but that function already has all the * necessary protection. */ - if (!MUTEX_HELD(&spa_namespace_lock)) { - mutex_enter(&spa_namespace_lock); + if (!MUTEX_HELD(&zfsdev_state_lock)) { + mutex_enter(&zfsdev_state_lock); locked = B_TRUE; } zv = pp->private; if (zv == NULL) { if (locked) - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (SET_ERROR(ENXIO)); } @@ -1021,13 +1143,14 @@ zvol_open(struct g_provider *pp, int flag, int count) err = zvol_first_open(zv); if (err) { if (locked) - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (err); } pp->mediasize = zv->zv_volsize; pp->stripeoffset = 0; pp->stripesize = zv->zv_volblocksize; } +#endif /* illumos */ if ((flag & FWRITE) && (zv->zv_flags & ZVOL_RDONLY)) { err = SET_ERROR(EROFS); goto out; @@ -1046,20 +1169,46 @@ zvol_open(struct g_provider *pp, int flag, int count) } #endif +#ifdef illumos + if (zv->zv_open_count[otyp] == 0 || otyp == OTYP_LYR) { + zv->zv_open_count[otyp]++; + zv->zv_total_opens++; + } + mutex_exit(&zfsdev_state_lock); +#else zv->zv_total_opens += count; if (locked) - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); +#endif return (err); out: if (zv->zv_total_opens == 0) zvol_last_close(zv); +#ifdef illumos + mutex_exit(&zfsdev_state_lock); +#else if (locked) - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); +#endif return (err); } /*ARGSUSED*/ +#ifdef illumos +int +zvol_close(dev_t dev, int flag, int otyp, cred_t *cr) +{ + minor_t minor = getminor(dev); + zvol_state_t *zv; + int error = 0; + + mutex_enter(&zfsdev_state_lock); + + zv = zfsdev_get_soft_state(minor, ZSST_ZVOL); + if (zv == NULL) { + mutex_exit(&zfsdev_state_lock); +#else /* !illumos */ static int zvol_close(struct g_provider *pp, int flag, int count) { @@ -1068,15 +1217,16 @@ zvol_close(struct g_provider *pp, int flag, int count) boolean_t locked = B_FALSE; /* See comment in zvol_open(). */ - if (!MUTEX_HELD(&spa_namespace_lock)) { - mutex_enter(&spa_namespace_lock); + if (!MUTEX_HELD(&zfsdev_state_lock)) { + mutex_enter(&zfsdev_state_lock); locked = B_TRUE; } zv = pp->private; if (zv == NULL) { if (locked) - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); +#endif /* illumos */ return (SET_ERROR(ENXIO)); } @@ -1089,18 +1239,30 @@ zvol_close(struct g_provider *pp, int flag, int count) * If the open count is zero, this is a spurious close. * That indicates a bug in the kernel / DDI framework. */ +#ifdef illumos + ASSERT(zv->zv_open_count[otyp] != 0); +#endif ASSERT(zv->zv_total_opens != 0); /* * You may get multiple opens, but only one close. */ +#ifdef illumos + zv->zv_open_count[otyp]--; + zv->zv_total_opens--; +#else zv->zv_total_opens -= count; +#endif if (zv->zv_total_opens == 0) zvol_last_close(zv); +#ifdef illumos + mutex_exit(&zfsdev_state_lock); +#else if (locked) - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); +#endif return (error); } @@ -1260,7 +1422,7 @@ zvol_log_write(zvol_state_t *zv, dmu_tx_t *tx, offset_t off, ssize_t resid, } } -#ifdef sun +#ifdef illumos static int zvol_dumpio_vdev(vdev_t *vd, void *addr, uint64_t offset, uint64_t origoffset, uint64_t size, boolean_t doread, boolean_t isdump) @@ -1353,11 +1515,16 @@ zvol_dumpio(zvol_state_t *zv, void *addr, uint64_t offset, uint64_t size, return (error); } -#endif /* sun */ +int +zvol_strategy(buf_t *bp) +{ + zfs_soft_state_t *zs = NULL; +#else /* !illumos */ void zvol_strategy(struct bio *bp) { +#endif /* illumos */ zvol_state_t *zv; uint64_t off, volsize; size_t resid; @@ -1365,22 +1532,53 @@ zvol_strategy(struct bio *bp) objset_t *os; rl_t *rl; int error = 0; +#ifdef illumos + boolean_t doread = bp->b_flags & B_READ; +#else boolean_t doread = 0; +#endif boolean_t is_dumpified; boolean_t sync; +#ifdef illumos + if (getminor(bp->b_edev) == 0) { + error = SET_ERROR(EINVAL); + } else { + zs = ddi_get_soft_state(zfsdev_state, getminor(bp->b_edev)); + if (zs == NULL) + error = SET_ERROR(ENXIO); + else if (zs->zss_type != ZSST_ZVOL) + error = SET_ERROR(EINVAL); + } + + if (error) { + bioerror(bp, error); + biodone(bp); + return (0); + } + + zv = zs->zss_data; + + if (!(bp->b_flags & B_READ) && (zv->zv_flags & ZVOL_RDONLY)) { + bioerror(bp, EROFS); + biodone(bp); + return (0); + } + + off = ldbtob(bp->b_blkno); +#else /* !illumos */ if (bp->bio_to) zv = bp->bio_to->private; else zv = bp->bio_dev->si_drv2; if (zv == NULL) { - error = ENXIO; + error = SET_ERROR(ENXIO); goto out; } if (bp->bio_cmd != BIO_READ && (zv->zv_flags & ZVOL_RDONLY)) { - error = EROFS; + error = SET_ERROR(EROFS); goto out; } @@ -1398,26 +1596,41 @@ zvol_strategy(struct bio *bp) } off = bp->bio_offset; +#endif /* illumos */ volsize = zv->zv_volsize; os = zv->zv_objset; ASSERT(os != NULL); +#ifdef illumos + bp_mapin(bp); + addr = bp->b_un.b_addr; + resid = bp->b_bcount; + + if (resid > 0 && (off < 0 || off >= volsize)) { + bioerror(bp, EIO); + biodone(bp); + return (0); + } + + is_dumpified = zv->zv_flags & ZVOL_DUMPIFIED; + sync = ((!(bp->b_flags & B_ASYNC) && + !(zv->zv_flags & ZVOL_WCE)) || + (zv->zv_objset->os_sync == ZFS_SYNC_ALWAYS)) && + !doread && !is_dumpified; +#else /* !illumos */ addr = bp->bio_data; resid = bp->bio_length; if (resid > 0 && (off < 0 || off >= volsize)) { - error = EIO; + error = SET_ERROR(EIO); goto out; } -#ifdef illumos - is_dumpified = zv->zv_flags & ZVOL_DUMPIFIED; -#else is_dumpified = B_FALSE; -#endif - sync = !doread && !is_dumpified && + sync = !doread && !is_dumpified && zv->zv_objset->os_sync == ZFS_SYNC_ALWAYS; +#endif /* illumos */ /* * There must be no buffer changes when doing a dmu_sync() because @@ -1426,6 +1639,7 @@ zvol_strategy(struct bio *bp) rl = zfs_range_lock(&zv->zv_znode, off, resid, doread ? RL_READER : RL_WRITER); +#ifndef illumos if (bp->bio_cmd == BIO_DELETE) { dmu_tx_t *tx = dmu_tx_create(zv->zv_objset); error = dmu_tx_assign(tx, TXG_WAIT); @@ -1440,7 +1654,7 @@ zvol_strategy(struct bio *bp) } goto unlock; } - +#endif while (resid != 0 && off < volsize) { size_t size = MIN(resid, zvol_maxphys); #ifdef illumos @@ -1476,9 +1690,21 @@ zvol_strategy(struct bio *bp) addr += size; resid -= size; } +#ifndef illumos unlock: +#endif zfs_range_unlock(rl); +#ifdef illumos + if ((bp->b_resid = resid) == bp->b_bcount) + bioerror(bp, off > volsize ? EINVAL : error); + + if (sync) + zil_commit(zv->zv_zilog, ZVOL_OBJ); + biodone(bp); + + return (0); +#else /* !illumos */ bp->bio_completed = bp->bio_length - resid; if (bp->bio_completed < bp->bio_length && off > volsize) error = EINVAL; @@ -1492,9 +1718,10 @@ out: g_io_deliver(bp, error); else biofinish(bp, NULL, error); +#endif /* illumos */ } -#ifdef sun +#ifdef illumos /* * Set the buffer count to the zvol maximum transfer. * Using our own routine instead of the default minphys() @@ -1550,17 +1777,17 @@ int zvol_read(dev_t dev, uio_t *uio, cred_t *cr) { minor_t minor = getminor(dev); -#else +#else /* !illumos */ int zvol_read(struct cdev *dev, struct uio *uio, int ioflag) { -#endif +#endif /* illumos */ zvol_state_t *zv; uint64_t volsize; rl_t *rl; int error = 0; -#ifdef sun +#ifdef illumos zv = zfsdev_get_soft_state(minor, ZSST_ZVOL); if (zv == NULL) return (SET_ERROR(ENXIO)); @@ -1569,6 +1796,7 @@ zvol_read(struct cdev *dev, struct uio *uio, int ioflag) #endif volsize = zv->zv_volsize; + /* uio_loffset == volsize isn't an error as its required for EOF processing. */ if (uio->uio_resid > 0 && (uio->uio_loffset < 0 || uio->uio_loffset > volsize)) return (SET_ERROR(EIO)); @@ -1602,24 +1830,24 @@ zvol_read(struct cdev *dev, struct uio *uio, int ioflag) return (error); } -#ifdef sun +#ifdef illumos /*ARGSUSED*/ int zvol_write(dev_t dev, uio_t *uio, cred_t *cr) { minor_t minor = getminor(dev); -#else +#else /* !illumos */ int zvol_write(struct cdev *dev, struct uio *uio, int ioflag) { -#endif +#endif /* illumos */ zvol_state_t *zv; uint64_t volsize; rl_t *rl; int error = 0; boolean_t sync; -#ifdef sun +#ifdef illumos zv = zfsdev_get_soft_state(minor, ZSST_ZVOL); if (zv == NULL) return (SET_ERROR(ENXIO)); @@ -1628,6 +1856,7 @@ zvol_write(struct cdev *dev, struct uio *uio, int ioflag) #endif volsize = zv->zv_volsize; + /* uio_loffset == volsize isn't an error as its required for EOF processing. */ if (uio->uio_resid > 0 && (uio->uio_loffset < 0 || uio->uio_loffset > volsize)) return (SET_ERROR(EIO)); @@ -1638,9 +1867,7 @@ zvol_write(struct cdev *dev, struct uio *uio, int ioflag) zvol_minphys, uio); return (error); } -#endif -#ifdef sun sync = !(zv->zv_flags & ZVOL_WCE) || #else sync = (ioflag & IO_SYNC) || @@ -1677,7 +1904,7 @@ zvol_write(struct cdev *dev, struct uio *uio, int ioflag) return (error); } -#ifdef sun +#ifdef illumos int zvol_getefi(void *arg, int flag, uint64_t vs, uint8_t bs) { @@ -1806,7 +2033,7 @@ zvol_log_write_minor(void *minor_hdl, dmu_tx_t *tx, offset_t off, ssize_t resid, /* * END entry points to allow external callers access to the volume. */ -#endif /* sun */ +#endif /* illumos */ /* * Log a DKIOCFREE/free-long-range to the ZIL with TX_TRUNCATE. @@ -1832,7 +2059,7 @@ zvol_log_truncate(zvol_state_t *zv, dmu_tx_t *tx, uint64_t off, uint64_t len, zil_itx_assign(zilog, itx, tx); } -#ifdef sun +#ifdef illumos /* * Dirtbag ioctls to support mkfs(1M) for UFS filesystems. See dkio(7I). * Also a dirtbag dkio ioctl for unmap/free-block functionality. @@ -1846,12 +2073,12 @@ zvol_ioctl(dev_t dev, int cmd, intptr_t arg, int flag, cred_t *cr, int *rvalp) int error = 0; rl_t *rl; - mutex_enter(&spa_namespace_lock); + mutex_enter(&zfsdev_state_lock); zv = zfsdev_get_soft_state(getminor(dev), ZSST_ZVOL); if (zv == NULL) { - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (SET_ERROR(ENXIO)); } ASSERT(zv->zv_total_opens > 0); @@ -1869,7 +2096,7 @@ zvol_ioctl(dev_t dev, int cmd, intptr_t arg, int flag, cred_t *cr, int *rvalp) dki.dki_unit = getminor(dev); dki.dki_maxtransfer = 1 << (SPA_OLD_MAXBLOCKSHIFT - zv->zv_min_bs); - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); if (ddi_copyout(&dki, (void *)arg, sizeof (dki), flag)) error = SET_ERROR(EFAULT); return (error); @@ -1883,7 +2110,7 @@ zvol_ioctl(dev_t dev, int cmd, intptr_t arg, int flag, cred_t *cr, int *rvalp) dkm.dki_lbsize = 1U << zv->zv_min_bs; dkm.dki_capacity = zv->zv_volsize >> zv->zv_min_bs; dkm.dki_media_type = DK_UNKNOWN; - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); if (ddi_copyout(&dkm, (void *)arg, sizeof (dkm), flag)) error = SET_ERROR(EFAULT); return (error); @@ -1898,7 +2125,7 @@ zvol_ioctl(dev_t dev, int cmd, intptr_t arg, int flag, cred_t *cr, int *rvalp) dkmext.dki_pbsize = zv->zv_volblocksize; dkmext.dki_capacity = zv->zv_volsize >> zv->zv_min_bs; dkmext.dki_media_type = DK_UNKNOWN; - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); if (ddi_copyout(&dkmext, (void *)arg, sizeof (dkmext), flag)) error = SET_ERROR(EFAULT); return (error); @@ -1909,14 +2136,14 @@ zvol_ioctl(dev_t dev, int cmd, intptr_t arg, int flag, cred_t *cr, int *rvalp) uint64_t vs = zv->zv_volsize; uint8_t bs = zv->zv_min_bs; - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); error = zvol_getefi((void *)arg, flag, vs, bs); return (error); } case DKIOCFLUSHWRITECACHE: dkc = (struct dk_callback *)arg; - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); zil_commit(zv->zv_zilog, ZVOL_OBJ); if ((flag & FKIOCTL) && dkc != NULL && dkc->dkc_callback) { (*dkc->dkc_callback)(dkc->dkc_cookie, error); @@ -1942,10 +2169,10 @@ zvol_ioctl(dev_t dev, int cmd, intptr_t arg, int flag, cred_t *cr, int *rvalp) } if (wce) { zv->zv_flags |= ZVOL_WCE; - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); } else { zv->zv_flags &= ~ZVOL_WCE; - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); zil_commit(zv->zv_zilog, ZVOL_OBJ); } return (0); @@ -1997,7 +2224,7 @@ zvol_ioctl(dev_t dev, int cmd, intptr_t arg, int flag, cred_t *cr, int *rvalp) if (df.df_start >= zv->zv_volsize) break; /* No need to do anything... */ - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); rl = zfs_range_lock(&zv->zv_znode, df.df_start, df.df_length, RL_WRITER); @@ -2044,10 +2271,10 @@ zvol_ioctl(dev_t dev, int cmd, intptr_t arg, int flag, cred_t *cr, int *rvalp) break; } - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (error); } -#endif /* sun */ +#endif /* illumos */ int zvol_busy(void) @@ -2060,17 +2287,24 @@ zvol_init(void) { VERIFY(ddi_soft_state_init(&zfsdev_state, sizeof (zfs_soft_state_t), 1) == 0); +#ifdef illumos + mutex_init(&zfsdev_state_lock, NULL, MUTEX_DEFAULT, NULL); +#else ZFS_LOG(1, "ZVOL Initialized."); +#endif } void zvol_fini(void) { +#ifdef illumos + mutex_destroy(&zfsdev_state_lock); +#endif ddi_soft_state_fini(&zfsdev_state); ZFS_LOG(1, "ZVOL Deinitialized."); } -#ifdef sun +#ifdef illumos /*ARGSUSED*/ static int zfs_mvdev_dump_feature_check(void *arg, dmu_tx_t *tx) @@ -2103,7 +2337,7 @@ zvol_dump_init(zvol_state_t *zv, boolean_t resize) uint64_t version = spa_version(spa); enum zio_checksum checksum; - ASSERT(MUTEX_HELD(&spa_namespace_lock)); + ASSERT(MUTEX_HELD(&zfsdev_state_lock)); ASSERT(vd->vdev_ops == &vdev_root_ops); error = dmu_free_long_range(zv->zv_objset, ZVOL_OBJ, 0, @@ -2355,7 +2589,7 @@ zvol_dump_fini(zvol_state_t *zv) return (0); } -#endif /* sun */ +#else /* !illumos */ static void zvol_geom_run(zvol_state_t *zv) @@ -2670,7 +2904,7 @@ zvol_rename_minor(zvol_state_t *zv, const char *newname) struct g_provider *pp; struct cdev *dev; - ASSERT(MUTEX_HELD(&spa_namespace_lock)); + ASSERT(MUTEX_HELD(&zfsdev_state_lock)); if (zv->zv_volmode == ZFS_VOLMODE_GEOM) { g_topology_lock(); @@ -2723,8 +2957,8 @@ zvol_rename_minors(const char *oldname, const char *newname) DROP_GIANT(); /* See comment in zvol_open(). */ - if (!MUTEX_HELD(&spa_namespace_lock)) { - mutex_enter(&spa_namespace_lock); + if (!MUTEX_HELD(&zfsdev_state_lock)) { + mutex_enter(&zfsdev_state_lock); locked = B_TRUE; } @@ -2742,7 +2976,7 @@ zvol_rename_minors(const char *oldname, const char *newname) } if (locked) - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); PICKUP_GIANT(); } @@ -2752,17 +2986,17 @@ zvol_d_open(struct cdev *dev, int flags, int fmt, struct thread *td) zvol_state_t *zv; int err = 0; - mutex_enter(&spa_namespace_lock); + mutex_enter(&zfsdev_state_lock); zv = dev->si_drv2; if (zv == NULL) { - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return(ENXIO); /* zvol_create_minor() not done yet */ } if (zv->zv_total_opens == 0) err = zvol_first_open(zv); if (err) { - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (err); } if ((flags & FWRITE) && (zv->zv_flags & ZVOL_RDONLY)) { @@ -2784,12 +3018,12 @@ zvol_d_open(struct cdev *dev, int flags, int fmt, struct thread *td) #endif zv->zv_total_opens++; - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (err); out: if (zv->zv_total_opens == 0) zvol_last_close(zv); - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (err); } @@ -2799,10 +3033,10 @@ zvol_d_close(struct cdev *dev, int flags, int fmt, struct thread *td) zvol_state_t *zv; int err = 0; - mutex_enter(&spa_namespace_lock); + mutex_enter(&zfsdev_state_lock); zv = dev->si_drv2; if (zv == NULL) { - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return(ENXIO); } @@ -2825,7 +3059,7 @@ zvol_d_close(struct cdev *dev, int flags, int fmt, struct thread *td) if (zv->zv_total_opens == 0) zvol_last_close(zv); - mutex_exit(&spa_namespace_lock); + mutex_exit(&zfsdev_state_lock); return (0); } @@ -2933,3 +3167,4 @@ zvol_d_ioctl(struct cdev *dev, u_long cmd, caddr_t data, int fflag, struct threa return (error); } +#endif /* illumos */ diff --git a/sys/cddl/dev/dtrace/amd64/dtrace_subr.c b/sys/cddl/dev/dtrace/amd64/dtrace_subr.c index c123cc6..805add2 100644 --- a/sys/cddl/dev/dtrace/amd64/dtrace_subr.c +++ b/sys/cddl/dev/dtrace/amd64/dtrace_subr.c @@ -464,7 +464,7 @@ dtrace_gethrestime(void) /* Function to handle DTrace traps during probes. See amd64/amd64/trap.c. */ int -dtrace_trap(struct trapframe *frame) +dtrace_trap(struct trapframe *frame, u_int type) { /* * A trap can occur while DTrace executes a probe. Before @@ -480,7 +480,7 @@ dtrace_trap(struct trapframe *frame) * There are only a couple of trap types that are expected. * All the rest will be handled in the usual way. */ - switch (frame->tf_trapno) { + switch (type) { /* General protection fault. */ case T_PROTFLT: /* Flag an illegal operation. */ diff --git a/sys/cddl/dev/dtrace/i386/dtrace_subr.c b/sys/cddl/dev/dtrace/i386/dtrace_subr.c index e620a8f..4073317 100644 --- a/sys/cddl/dev/dtrace/i386/dtrace_subr.c +++ b/sys/cddl/dev/dtrace/i386/dtrace_subr.c @@ -473,7 +473,7 @@ dtrace_gethrestime(void) /* Function to handle DTrace traps during probes. See i386/i386/trap.c */ int -dtrace_trap(struct trapframe *frame) +dtrace_trap(struct trapframe *frame, u_int type) { /* * A trap can occur while DTrace executes a probe. Before @@ -489,7 +489,7 @@ dtrace_trap(struct trapframe *frame) * There are only a couple of trap types that are expected. * All the rest will be handled in the usual way. */ - switch (frame->tf_trapno) { + switch (type) { /* General protection fault. */ case T_PROTFLT: /* Flag an illegal operation. */ diff --git a/sys/cddl/dev/dtrace/mips/dtrace_subr.c b/sys/cddl/dev/dtrace/mips/dtrace_subr.c index 5565a61..4f13b98 100644 --- a/sys/cddl/dev/dtrace/mips/dtrace_subr.c +++ b/sys/cddl/dev/dtrace/mips/dtrace_subr.c @@ -137,11 +137,8 @@ dtrace_gethrestime(void) /* Function to handle DTrace traps during probes. See amd64/amd64/trap.c */ int -dtrace_trap(struct trapframe *frame) +dtrace_trap(struct trapframe *frame, u_int type) { - u_int type; - - type = (trapframe->cause & MIPS_CR_EXC_CODE) >> MIPS_CR_EXC_CODE_SHIFT; /* * A trap can occur while DTrace executes a probe. Before diff --git a/sys/cddl/dev/dtrace/powerpc/dtrace_subr.c b/sys/cddl/dev/dtrace/powerpc/dtrace_subr.c index 5411ece..a89f1d3 100644 --- a/sys/cddl/dev/dtrace/powerpc/dtrace_subr.c +++ b/sys/cddl/dev/dtrace/powerpc/dtrace_subr.c @@ -262,8 +262,9 @@ dtrace_gethrestime(void) /* Function to handle DTrace traps during probes. See powerpc/powerpc/trap.c */ int -dtrace_trap(struct trapframe *frame) +dtrace_trap(struct trapframe *frame, u_int type) { + /* * A trap can occur while DTrace executes a probe. Before * executing the probe, DTrace blocks re-scheduling and sets @@ -278,7 +279,7 @@ dtrace_trap(struct trapframe *frame) * There are only a couple of trap types that are expected. * All the rest will be handled in the usual way. */ - switch (frame->exc) { + switch (type) { /* Page fault. */ case EXC_DSI: case EXC_DSE: diff --git a/sys/conf/NOTES b/sys/conf/NOTES index 1436a3b..66c1dbc 100644 --- a/sys/conf/NOTES +++ b/sys/conf/NOTES @@ -1019,7 +1019,7 @@ options DUMMYNET # One of these is mandatory: options FFS #Fast filesystem -options NFSCLIENT #Network File System client +options NFSCL #Network File System client # The rest are optional: options AUTOFS #Automounter filesystem @@ -1027,7 +1027,6 @@ options CD9660 #ISO 9660 filesystem options FDESCFS #File descriptor filesystem options FUSE #FUSE support module options MSDOSFS #MS DOS File System (FAT, FAT32) -options NFSSERVER #Network File System server options NFSLOCKD #Network Lock Manager options NFSCL #New Network Filesystem Client options NFSD #New Network Filesystem Server @@ -1099,8 +1098,6 @@ options NFS_MINATTRTIMO=3 # VREG attrib cache timeout in sec options NFS_MAXATTRTIMO=60 options NFS_MINDIRATTRTIMO=30 # VDIR attrib cache timeout in sec options NFS_MAXDIRATTRTIMO=60 -options NFS_GATHERDELAY=10 # Default write gather delay (msec) -options NFS_WDELAYHASHSIZ=16 # and with this options NFS_DEBUG # Enable NFS Debugging # @@ -2357,19 +2354,6 @@ options SND_PCM_64 options SND_OLDSTEREO # -# IEEE-488 hardware: -# pcii: PCIIA cards (uPD7210 based isa cards) -# tnt4882: National Instruments PCI-GPIB card. - -device pcii -hint.pcii.0.at="isa" -hint.pcii.0.port="0x2e1" -hint.pcii.0.irq="5" -hint.pcii.0.drq="1" - -device tnt4882 - -# # Miscellaneous hardware: # # scd: Sony CD-ROM using proprietary (non-ATAPI) interface @@ -2596,7 +2580,7 @@ device pcfclock # Kernel BOOTP support options BOOTP # Use BOOTP to obtain IP address/hostname - # Requires NFSCLIENT and NFS_ROOT + # Requires NFSCL and NFS_ROOT options BOOTP_NFSROOT # NFS mount root filesystem using BOOTP info options BOOTP_NFSV3 # Use NFS v3 to NFS mount root options BOOTP_COMPAT # Workaround for broken bootp daemons. @@ -2788,7 +2772,7 @@ options U3G_DEBUG # options for ukbd: options UKBD_DFLT_KEYMAP # specify the built-in keymap -makeoptions UKBD_DFLT_KEYMAP=it.iso +makeoptions UKBD_DFLT_KEYMAP=jp.pc98 # options for uplcom: options UPLCOM_INTR_INTERVAL=100 # interrupt pipe interval diff --git a/sys/conf/files b/sys/conf/files index 939b635..237390f 100644 --- a/sys/conf/files +++ b/sys/conf/files @@ -1448,10 +1448,6 @@ dev/ida/ida_eisa.c optional ida eisa dev/ida/ida_pci.c optional ida pci dev/ie/if_ie.c optional ie isa nowerror dev/ie/if_ie_isa.c optional ie isa -dev/ieee488/ibfoo.c optional pcii | tnt4882 -dev/ieee488/pcii.c optional pcii -dev/ieee488/tnt4882.c optional tnt4882 -dev/ieee488/upd7210.c optional pcii | tnt4882 dev/iicbus/ad7418.c optional ad7418 dev/iicbus/ds133x.c optional ds133x dev/iicbus/ds1374.c optional ds1374 @@ -3559,24 +3555,12 @@ netsmb/smb_smb.c optional netsmb netsmb/smb_subr.c optional netsmb netsmb/smb_trantcp.c optional netsmb netsmb/smb_usr.c optional netsmb -nfs/bootp_subr.c optional bootp nfsclient | bootp nfscl -nfs/krpc_subr.c optional bootp nfsclient | bootp nfscl -nfs/nfs_common.c optional nfsclient | nfsserver -nfs/nfs_diskless.c optional nfsclient nfs_root | nfscl nfs_root -nfs/nfs_fha.c optional nfsserver | nfsd -nfs/nfs_lock.c optional nfsclient | nfscl | nfslockd | nfsd -nfsclient/nfs_bio.c optional nfsclient -nfsclient/nfs_node.c optional nfsclient -nfsclient/nfs_krpc.c optional nfsclient -nfsclient/nfs_subs.c optional nfsclient -nfsclient/nfs_nfsiod.c optional nfsclient -nfsclient/nfs_vfsops.c optional nfsclient -nfsclient/nfs_vnops.c optional nfsclient -nfsserver/nfs_fha_old.c optional nfsserver -nfsserver/nfs_serv.c optional nfsserver -nfsserver/nfs_srvkrpc.c optional nfsserver -nfsserver/nfs_srvsubs.c optional nfsserver -nfs/nfs_nfssvc.c optional nfsserver | nfscl | nfsd +nfs/bootp_subr.c optional bootp nfscl +nfs/krpc_subr.c optional bootp nfscl +nfs/nfs_diskless.c optional nfscl nfs_root +nfs/nfs_fha.c optional nfsd +nfs/nfs_lock.c optional nfscl | nfslockd | nfsd +nfs/nfs_nfssvc.c optional nfscl | nfsd nlm/nlm_advlock.c optional nfslockd | nfsd nlm/nlm_prot_clnt.c optional nfslockd | nfsd nlm/nlm_prot_impl.c optional nfslockd | nfsd @@ -3887,26 +3871,26 @@ opencrypto/gfmult.c optional crypto opencrypto/rmd160.c optional crypto | ipsec opencrypto/skipjack.c optional crypto opencrypto/xform.c optional crypto -rpc/auth_none.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -rpc/auth_unix.c optional krpc | nfslockd | nfsclient | nfscl | nfsd -rpc/authunix_prot.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -rpc/clnt_bck.c optional krpc | nfslockd | nfsserver | nfscl | nfsd -rpc/clnt_dg.c optional krpc | nfslockd | nfsclient | nfscl | nfsd -rpc/clnt_rc.c optional krpc | nfslockd | nfsclient | nfscl | nfsd -rpc/clnt_vc.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -rpc/getnetconfig.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -rpc/replay.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -rpc/rpc_callmsg.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -rpc/rpc_generic.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -rpc/rpc_prot.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -rpc/rpcb_clnt.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -rpc/rpcb_prot.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -rpc/svc.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -rpc/svc_auth.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -rpc/svc_auth_unix.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -rpc/svc_dg.c optional krpc | nfslockd | nfsserver | nfscl | nfsd -rpc/svc_generic.c optional krpc | nfslockd | nfsserver | nfscl | nfsd -rpc/svc_vc.c optional krpc | nfslockd | nfsserver | nfscl | nfsd +rpc/auth_none.c optional krpc | nfslockd | nfscl | nfsd +rpc/auth_unix.c optional krpc | nfslockd | nfscl | nfsd +rpc/authunix_prot.c optional krpc | nfslockd | nfscl | nfsd +rpc/clnt_bck.c optional krpc | nfslockd | nfscl | nfsd +rpc/clnt_dg.c optional krpc | nfslockd | nfscl | nfsd +rpc/clnt_rc.c optional krpc | nfslockd | nfscl | nfsd +rpc/clnt_vc.c optional krpc | nfslockd | nfscl | nfsd +rpc/getnetconfig.c optional krpc | nfslockd | nfscl | nfsd +rpc/replay.c optional krpc | nfslockd | nfscl | nfsd +rpc/rpc_callmsg.c optional krpc | nfslockd | nfscl | nfsd +rpc/rpc_generic.c optional krpc | nfslockd | nfscl | nfsd +rpc/rpc_prot.c optional krpc | nfslockd | nfscl | nfsd +rpc/rpcb_clnt.c optional krpc | nfslockd | nfscl | nfsd +rpc/rpcb_prot.c optional krpc | nfslockd | nfscl | nfsd +rpc/svc.c optional krpc | nfslockd | nfscl | nfsd +rpc/svc_auth.c optional krpc | nfslockd | nfscl | nfsd +rpc/svc_auth_unix.c optional krpc | nfslockd | nfscl | nfsd +rpc/svc_dg.c optional krpc | nfslockd | nfscl | nfsd +rpc/svc_generic.c optional krpc | nfslockd | nfscl | nfsd +rpc/svc_vc.c optional krpc | nfslockd | nfscl | nfsd rpc/rpcsec_gss/rpcsec_gss.c optional krpc kgssapi | nfslockd kgssapi | nfscl kgssapi | nfsd kgssapi rpc/rpcsec_gss/rpcsec_gss_conf.c optional krpc kgssapi | nfslockd kgssapi | nfscl kgssapi | nfsd kgssapi rpc/rpcsec_gss/rpcsec_gss_misc.c optional krpc kgssapi | nfslockd kgssapi | nfscl kgssapi | nfsd kgssapi @@ -4012,9 +3996,9 @@ xen/xenbus/xenbusb_if.m optional xen | xenhvm xen/xenbus/xenbusb.c optional xen | xenhvm xen/xenbus/xenbusb_front.c optional xen | xenhvm xen/xenbus/xenbusb_back.c optional xen | xenhvm -xdr/xdr.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -xdr/xdr_array.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -xdr/xdr_mbuf.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -xdr/xdr_mem.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -xdr/xdr_reference.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd -xdr/xdr_sizeof.c optional krpc | nfslockd | nfsclient | nfsserver | nfscl | nfsd +xdr/xdr.c optional krpc | nfslockd | nfscl | nfsd +xdr/xdr_array.c optional krpc | nfslockd | nfscl | nfsd +xdr/xdr_mbuf.c optional krpc | nfslockd | nfscl | nfsd +xdr/xdr_mem.c optional krpc | nfslockd | nfscl | nfsd +xdr/xdr_reference.c optional krpc | nfslockd | nfscl | nfsd +xdr/xdr_sizeof.c optional krpc | nfslockd | nfscl | nfsd diff --git a/sys/conf/newvers.sh b/sys/conf/newvers.sh index 9b3b84b..c06eeaf 100644 --- a/sys/conf/newvers.sh +++ b/sys/conf/newvers.sh @@ -52,7 +52,11 @@ else fi b=share/examples/etc/bsd-style-copyright -year=$(sed -Ee '/^Copyright .* The FreeBSD Project/!d;s/^.*1992-([0-9]*) .*$/\1/g' ${SYSDIR}/../COPYRIGHT) +if [ -r "${SYSDIR}/../COPYRIGHT" ]; then + year=$(sed -Ee '/^Copyright .* The FreeBSD Project/!d;s/^.*1992-([0-9]*) .*$/\1/g' ${SYSDIR}/../COPYRIGHT) +else + year=$(date +%Y) +fi # look for copyright template for bsd_copyright in ../$b ../../$b ../../../$b /usr/src/$b /usr/$b do diff --git a/sys/conf/options b/sys/conf/options index ae0f852..52c44a1 100644 --- a/sys/conf/options +++ b/sys/conf/options @@ -253,11 +253,6 @@ KGSSAPI_DEBUG opt_kgssapi.h # sys/i386/i386/autoconf.c. If any of these filesystems are # statically compiled into the kernel, code for mounting them as root # filesystems will be enabled - but look below. -NFSCLIENT opt_nfs.h -NFSSERVER opt_nfs.h - -# Use these options to compile the experimental nfs client and/or -# server that supports NFSv4 into a kernel. # NFSCL - client # NFSD - server NFSCL opt_nfs.h @@ -624,8 +619,6 @@ NFS_MINATTRTIMO opt_nfs.h NFS_MAXATTRTIMO opt_nfs.h NFS_MINDIRATTRTIMO opt_nfs.h NFS_MAXDIRATTRTIMO opt_nfs.h -NFS_GATHERDELAY opt_nfs.h -NFS_WDELAYHASHSIZ opt_nfs.h NFS_DEBUG opt_nfs.h # For the Bt848/Bt848A/Bt849/Bt878/Bt879 driver diff --git a/sys/dev/ahci/ahci.c b/sys/dev/ahci/ahci.c index 583983b..31aab5d 100644 --- a/sys/dev/ahci/ahci.c +++ b/sys/dev/ahci/ahci.c @@ -142,7 +142,7 @@ ahci_ctlr_reset(device_t dev) } if (timeout == 0) { device_printf(dev, "AHCI controller reset failure\n"); - return ENXIO; + return (ENXIO); } /* Reenable AHCI mode */ ATA_OUTL(ctlr->r_mem, AHCI_GHC, AHCI_GHC_AE); @@ -225,18 +225,18 @@ ahci_attach(device_t dev) bus_release_resource(dev, SYS_RES_MEMORY, ctlr->r_rid, ctlr->r_mem); rman_fini(&ctlr->sc_iomem); - return ENXIO; + return (ENXIO); } ahci_ctlr_setup(dev); /* Setup interrupts. */ - if (ahci_setup_interrupt(dev)) { + if ((error = ahci_setup_interrupt(dev)) != 0) { bus_dma_tag_destroy(ctlr->dma_tag); bus_release_resource(dev, SYS_RES_MEMORY, ctlr->r_rid, ctlr->r_mem); rman_fini(&ctlr->sc_iomem); - return ENXIO; + return (error); } i = 0; @@ -315,7 +315,7 @@ ahci_attach(device_t dev) device_set_ivars(child, (void *)(intptr_t)-1); } bus_generic_attach(dev); - return 0; + return (0); } int @@ -356,6 +356,14 @@ ahci_setup_interrupt(device_t dev) device_printf(dev, "Falling back to one MSI\n"); ctlr->numirqs = 1; } + + /* Ensure we don't overrun irqs. */ + if (ctlr->numirqs > AHCI_MAX_IRQS) { + device_printf(dev, "Too many irqs %d > %d (clamping)\n", + ctlr->numirqs, AHCI_MAX_IRQS); + ctlr->numirqs = AHCI_MAX_IRQS; + } + /* Allocate all IRQs. */ for (i = 0; i < ctlr->numirqs; i++) { ctlr->irqs[i].ctlr = ctlr; @@ -372,7 +380,7 @@ ahci_setup_interrupt(device_t dev) if (!(ctlr->irqs[i].r_irq = bus_alloc_resource_any(dev, SYS_RES_IRQ, &ctlr->irqs[i].r_irq_rid, RF_SHAREABLE | RF_ACTIVE))) { device_printf(dev, "unable to map interrupt\n"); - return ENXIO; + return (ENXIO); } if ((bus_setup_intr(dev, ctlr->irqs[i].r_irq, ATA_INTR_FLAGS, NULL, (ctlr->irqs[i].mode != AHCI_IRQ_MODE_ONE) ? ahci_intr : @@ -381,7 +389,7 @@ ahci_setup_interrupt(device_t dev) &ctlr->irqs[i], &ctlr->irqs[i].handle))) { /* SOS XXX release r_irq */ device_printf(dev, "unable to setup interrupt\n"); - return ENXIO; + return (ENXIO); } if (ctlr->numirqs > 1) { bus_describe_intr(dev, ctlr->irqs[i].r_irq, @@ -529,7 +537,7 @@ ahci_release_resource(device_t dev, device_t child, int type, int rid, return (0); case SYS_RES_IRQ: if (rid != ATA_IRQ_RID) - return ENOENT; + return (ENOENT); return (0); } return (EINVAL); diff --git a/sys/dev/ahci/ahci.h b/sys/dev/ahci/ahci.h index f5a3ca8..462f204 100644 --- a/sys/dev/ahci/ahci.h +++ b/sys/dev/ahci/ahci.h @@ -143,6 +143,7 @@ #define AHCI_MAX_PORTS 32 #define AHCI_MAX_SLOTS 32 +#define AHCI_MAX_IRQS 16 /* SATA AHCI v1.0 register defines */ #define AHCI_CAP 0x00 @@ -494,7 +495,7 @@ struct ahci_controller { #define AHCI_IRQ_MODE_ALL 0 #define AHCI_IRQ_MODE_AFTER 1 #define AHCI_IRQ_MODE_ONE 2 - } irqs[16]; + } irqs[AHCI_MAX_IRQS]; uint32_t caps; /* Controller capabilities */ uint32_t caps2; /* Controller capabilities */ uint32_t capsem; /* Controller capabilities */ diff --git a/sys/dev/ath/if_ath_sysctl.c b/sys/dev/ath/if_ath_sysctl.c index 40a34d4..45c8ae4 100644 --- a/sys/dev/ath/if_ath_sysctl.c +++ b/sys/dev/ath/if_ath_sysctl.c @@ -446,7 +446,15 @@ ath_sysctl_rfsilent(SYSCTL_HANDLER_ARGS) return error; if (!ath_hal_setrfsilent(sc->sc_ah, rfsilent)) return EINVAL; - sc->sc_rfsilentpin = rfsilent & 0x1c; + /* + * Earlier chips (< AR5212) have up to 8 GPIO + * pins exposed. + * + * AR5416 and later chips have many more GPIO + * pins (up to 16) so the mask is expanded to + * four bits. + */ + sc->sc_rfsilentpin = rfsilent & 0x3c; sc->sc_rfsilentpol = (rfsilent & 0x2) != 0; return 0; } diff --git a/sys/dev/beri/virtio/virtio.c b/sys/dev/beri/virtio/virtio.c index 4cc25c2..cb7c293 100644 --- a/sys/dev/beri/virtio/virtio.c +++ b/sys/dev/beri/virtio/virtio.c @@ -38,6 +38,7 @@ __FBSDID("$FreeBSD$"); #include <sys/param.h> #include <sys/systm.h> #include <sys/bus.h> +#include <sys/cdefs.h> #include <sys/kernel.h> #include <sys/module.h> #include <sys/malloc.h> @@ -156,7 +157,7 @@ vq_getchain(uint32_t offs, struct vqueue_info *vq, break; next = be16toh(vp->next); } - paddr_unmap((void *)vindir, be32toh(vdir->len)); + paddr_unmap(__DEVOLATILE(void *, vindir), be32toh(vdir->len)); } if ((be16toh(vdir->flags) & VRING_DESC_F_NEXT) == 0) diff --git a/sys/dev/bge/if_bge.c b/sys/dev/bge/if_bge.c index 838be4f..58f529d 100644 --- a/sys/dev/bge/if_bge.c +++ b/sys/dev/bge/if_bge.c @@ -1946,11 +1946,9 @@ bge_chipinit(struct bge_softc *sc) /* * Disable memory write invalidate. Apparently it is not supported - * properly by these devices. Also ensure that INTx isn't disabled, - * as these chips need it even when using MSI. + * properly by these devices. */ - PCI_CLRBIT(sc->bge_dev, BGE_PCI_CMD, - PCIM_CMD_INTxDIS | PCIM_CMD_MWIEN, 4); + PCI_CLRBIT(sc->bge_dev, BGE_PCI_CMD, PCIM_CMD_MWIEN, 4); /* Set the timer prescaler (always 66 MHz). */ CSR_WRITE_4(sc, BGE_MISC_CFG, BGE_32BITTIME_66MHZ); diff --git a/sys/dev/cpuctl/cpuctl.c b/sys/dev/cpuctl/cpuctl.c index 96ceb0d..ec41a2e 100644 --- a/sys/dev/cpuctl/cpuctl.c +++ b/sys/dev/cpuctl/cpuctl.c @@ -63,7 +63,7 @@ static d_ioctl_t cpuctl_ioctl; # define DPRINTF(...) #endif -#define UCODE_SIZE_MAX (16 * 1024) +#define UCODE_SIZE_MAX (32 * 1024) static int cpuctl_do_msr(int cpu, cpuctl_msr_args_t *data, u_long cmd, struct thread *td); diff --git a/sys/dev/fdt/fdt_common.c b/sys/dev/fdt/fdt_common.c index d99fdf2..f00519e 100644 --- a/sys/dev/fdt/fdt_common.c +++ b/sys/dev/fdt/fdt_common.c @@ -75,6 +75,12 @@ fdt_get_range_by_busaddr(phandle_t node, u_long addr, u_long *base, u_long bus_addr, par_bus_addr, pbase, psize; int err, i, len, tuple_size, tuples; + if (node == 0) { + *base = 0; + *size = ULONG_MAX; + return (0); + } + if ((fdt_addrsize_cells(node, &addr_cells, &size_cells)) != 0) return (ENXIO); /* @@ -91,9 +97,8 @@ fdt_get_range_by_busaddr(phandle_t node, u_long addr, u_long *base, if (len > sizeof(ranges)) return (ENOMEM); if (len == 0) { - *base = 0; - *size = ULONG_MAX; - return (0); + return (fdt_get_range_by_busaddr(OF_parent(node), addr, + base, size)); } if (OF_getprop(node, "ranges", ranges, sizeof(ranges)) <= 0) diff --git a/sys/dev/fxp/if_fxp.c b/sys/dev/fxp/if_fxp.c index 947f057..cf0e6a2 100644 --- a/sys/dev/fxp/if_fxp.c +++ b/sys/dev/fxp/if_fxp.c @@ -157,52 +157,52 @@ static const u_char fxp_cb_config_template[] = { * them. */ static const struct fxp_ident fxp_ident_table[] = { - { 0x1029, -1, 0, "Intel 82559 PCI/CardBus Pro/100" }, - { 0x1030, -1, 0, "Intel 82559 Pro/100 Ethernet" }, - { 0x1031, -1, 3, "Intel 82801CAM (ICH3) Pro/100 VE Ethernet" }, - { 0x1032, -1, 3, "Intel 82801CAM (ICH3) Pro/100 VE Ethernet" }, - { 0x1033, -1, 3, "Intel 82801CAM (ICH3) Pro/100 VM Ethernet" }, - { 0x1034, -1, 3, "Intel 82801CAM (ICH3) Pro/100 VM Ethernet" }, - { 0x1035, -1, 3, "Intel 82801CAM (ICH3) Pro/100 Ethernet" }, - { 0x1036, -1, 3, "Intel 82801CAM (ICH3) Pro/100 Ethernet" }, - { 0x1037, -1, 3, "Intel 82801CAM (ICH3) Pro/100 Ethernet" }, - { 0x1038, -1, 3, "Intel 82801CAM (ICH3) Pro/100 VM Ethernet" }, - { 0x1039, -1, 4, "Intel 82801DB (ICH4) Pro/100 VE Ethernet" }, - { 0x103A, -1, 4, "Intel 82801DB (ICH4) Pro/100 Ethernet" }, - { 0x103B, -1, 4, "Intel 82801DB (ICH4) Pro/100 VM Ethernet" }, - { 0x103C, -1, 4, "Intel 82801DB (ICH4) Pro/100 Ethernet" }, - { 0x103D, -1, 4, "Intel 82801DB (ICH4) Pro/100 VE Ethernet" }, - { 0x103E, -1, 4, "Intel 82801DB (ICH4) Pro/100 VM Ethernet" }, - { 0x1050, -1, 5, "Intel 82801BA (D865) Pro/100 VE Ethernet" }, - { 0x1051, -1, 5, "Intel 82562ET (ICH5/ICH5R) Pro/100 VE Ethernet" }, - { 0x1059, -1, 0, "Intel 82551QM Pro/100 M Mobile Connection" }, - { 0x1064, -1, 6, "Intel 82562EZ (ICH6)" }, - { 0x1065, -1, 6, "Intel 82562ET/EZ/GT/GZ PRO/100 VE Ethernet" }, - { 0x1068, -1, 6, "Intel 82801FBM (ICH6-M) Pro/100 VE Ethernet" }, - { 0x1069, -1, 6, "Intel 82562EM/EX/GX Pro/100 Ethernet" }, - { 0x1091, -1, 7, "Intel 82562GX Pro/100 Ethernet" }, - { 0x1092, -1, 7, "Intel Pro/100 VE Network Connection" }, - { 0x1093, -1, 7, "Intel Pro/100 VM Network Connection" }, - { 0x1094, -1, 7, "Intel Pro/100 946GZ (ICH7) Network Connection" }, - { 0x1209, -1, 0, "Intel 82559ER Embedded 10/100 Ethernet" }, - { 0x1229, 0x01, 0, "Intel 82557 Pro/100 Ethernet" }, - { 0x1229, 0x02, 0, "Intel 82557 Pro/100 Ethernet" }, - { 0x1229, 0x03, 0, "Intel 82557 Pro/100 Ethernet" }, - { 0x1229, 0x04, 0, "Intel 82558 Pro/100 Ethernet" }, - { 0x1229, 0x05, 0, "Intel 82558 Pro/100 Ethernet" }, - { 0x1229, 0x06, 0, "Intel 82559 Pro/100 Ethernet" }, - { 0x1229, 0x07, 0, "Intel 82559 Pro/100 Ethernet" }, - { 0x1229, 0x08, 0, "Intel 82559 Pro/100 Ethernet" }, - { 0x1229, 0x09, 0, "Intel 82559ER Pro/100 Ethernet" }, - { 0x1229, 0x0c, 0, "Intel 82550 Pro/100 Ethernet" }, - { 0x1229, 0x0d, 0, "Intel 82550C Pro/100 Ethernet" }, - { 0x1229, 0x0e, 0, "Intel 82550 Pro/100 Ethernet" }, - { 0x1229, 0x0f, 0, "Intel 82551 Pro/100 Ethernet" }, - { 0x1229, 0x10, 0, "Intel 82551 Pro/100 Ethernet" }, - { 0x1229, -1, 0, "Intel 82557/8/9 Pro/100 Ethernet" }, - { 0x2449, -1, 2, "Intel 82801BA/CAM (ICH2/3) Pro/100 Ethernet" }, - { 0x27dc, -1, 7, "Intel 82801GB (ICH7) 10/100 Ethernet" }, - { 0, -1, 0, NULL }, + { 0x8086, 0x1029, -1, 0, "Intel 82559 PCI/CardBus Pro/100" }, + { 0x8086, 0x1030, -1, 0, "Intel 82559 Pro/100 Ethernet" }, + { 0x8086, 0x1031, -1, 3, "Intel 82801CAM (ICH3) Pro/100 VE Ethernet" }, + { 0x8086, 0x1032, -1, 3, "Intel 82801CAM (ICH3) Pro/100 VE Ethernet" }, + { 0x8086, 0x1033, -1, 3, "Intel 82801CAM (ICH3) Pro/100 VM Ethernet" }, + { 0x8086, 0x1034, -1, 3, "Intel 82801CAM (ICH3) Pro/100 VM Ethernet" }, + { 0x8086, 0x1035, -1, 3, "Intel 82801CAM (ICH3) Pro/100 Ethernet" }, + { 0x8086, 0x1036, -1, 3, "Intel 82801CAM (ICH3) Pro/100 Ethernet" }, + { 0x8086, 0x1037, -1, 3, "Intel 82801CAM (ICH3) Pro/100 Ethernet" }, + { 0x8086, 0x1038, -1, 3, "Intel 82801CAM (ICH3) Pro/100 VM Ethernet" }, + { 0x8086, 0x1039, -1, 4, "Intel 82801DB (ICH4) Pro/100 VE Ethernet" }, + { 0x8086, 0x103A, -1, 4, "Intel 82801DB (ICH4) Pro/100 Ethernet" }, + { 0x8086, 0x103B, -1, 4, "Intel 82801DB (ICH4) Pro/100 VM Ethernet" }, + { 0x8086, 0x103C, -1, 4, "Intel 82801DB (ICH4) Pro/100 Ethernet" }, + { 0x8086, 0x103D, -1, 4, "Intel 82801DB (ICH4) Pro/100 VE Ethernet" }, + { 0x8086, 0x103E, -1, 4, "Intel 82801DB (ICH4) Pro/100 VM Ethernet" }, + { 0x8086, 0x1050, -1, 5, "Intel 82801BA (D865) Pro/100 VE Ethernet" }, + { 0x8086, 0x1051, -1, 5, "Intel 82562ET (ICH5/ICH5R) Pro/100 VE Ethernet" }, + { 0x8086, 0x1059, -1, 0, "Intel 82551QM Pro/100 M Mobile Connection" }, + { 0x8086, 0x1064, -1, 6, "Intel 82562EZ (ICH6)" }, + { 0x8086, 0x1065, -1, 6, "Intel 82562ET/EZ/GT/GZ PRO/100 VE Ethernet" }, + { 0x8086, 0x1068, -1, 6, "Intel 82801FBM (ICH6-M) Pro/100 VE Ethernet" }, + { 0x8086, 0x1069, -1, 6, "Intel 82562EM/EX/GX Pro/100 Ethernet" }, + { 0x8086, 0x1091, -1, 7, "Intel 82562GX Pro/100 Ethernet" }, + { 0x8086, 0x1092, -1, 7, "Intel Pro/100 VE Network Connection" }, + { 0x8086, 0x1093, -1, 7, "Intel Pro/100 VM Network Connection" }, + { 0x8086, 0x1094, -1, 7, "Intel Pro/100 946GZ (ICH7) Network Connection" }, + { 0x8086, 0x1209, -1, 0, "Intel 82559ER Embedded 10/100 Ethernet" }, + { 0x8086, 0x1229, 0x01, 0, "Intel 82557 Pro/100 Ethernet" }, + { 0x8086, 0x1229, 0x02, 0, "Intel 82557 Pro/100 Ethernet" }, + { 0x8086, 0x1229, 0x03, 0, "Intel 82557 Pro/100 Ethernet" }, + { 0x8086, 0x1229, 0x04, 0, "Intel 82558 Pro/100 Ethernet" }, + { 0x8086, 0x1229, 0x05, 0, "Intel 82558 Pro/100 Ethernet" }, + { 0x8086, 0x1229, 0x06, 0, "Intel 82559 Pro/100 Ethernet" }, + { 0x8086, 0x1229, 0x07, 0, "Intel 82559 Pro/100 Ethernet" }, + { 0x8086, 0x1229, 0x08, 0, "Intel 82559 Pro/100 Ethernet" }, + { 0x8086, 0x1229, 0x09, 0, "Intel 82559ER Pro/100 Ethernet" }, + { 0x8086, 0x1229, 0x0c, 0, "Intel 82550 Pro/100 Ethernet" }, + { 0x8086, 0x1229, 0x0d, 0, "Intel 82550C Pro/100 Ethernet" }, + { 0x8086, 0x1229, 0x0e, 0, "Intel 82550 Pro/100 Ethernet" }, + { 0x8086, 0x1229, 0x0f, 0, "Intel 82551 Pro/100 Ethernet" }, + { 0x8086, 0x1229, 0x10, 0, "Intel 82551 Pro/100 Ethernet" }, + { 0x8086, 0x1229, -1, 0, "Intel 82557/8/9 Pro/100 Ethernet" }, + { 0x8086, 0x2449, -1, 2, "Intel 82801BA/CAM (ICH2/3) Pro/100 Ethernet" }, + { 0x8086, 0x27dc, -1, 7, "Intel 82801GB (ICH7) 10/100 Ethernet" }, + { 0, 0, -1, 0, NULL }, }; #ifdef FXP_IP_CSUM_WAR @@ -374,18 +374,18 @@ fxp_dma_wait(struct fxp_softc *sc, volatile uint16_t *status, static const struct fxp_ident * fxp_find_ident(device_t dev) { - uint16_t devid; + uint16_t vendor; + uint16_t device; uint8_t revid; const struct fxp_ident *ident; - if (pci_get_vendor(dev) == FXP_VENDORID_INTEL) { - devid = pci_get_device(dev); - revid = pci_get_revid(dev); - for (ident = fxp_ident_table; ident->name != NULL; ident++) { - if (ident->devid == devid && - (ident->revid == revid || ident->revid == -1)) { - return (ident); - } + vendor = pci_get_vendor(dev); + device = pci_get_device(dev); + revid = pci_get_revid(dev); + for (ident = fxp_ident_table; ident->name != NULL; ident++) { + if (ident->vendor == vendor && ident->device == device && + (ident->revid == revid || ident->revid == -1)) { + return (ident); } } return (NULL); @@ -628,7 +628,7 @@ fxp_attach(device_t dev) /* For 82559 or later chips, Rx checksum offload is supported. */ if (sc->revision >= FXP_REV_82559_A0) { /* 82559ER does not support Rx checksum offloading. */ - if (sc->ident->devid != 0x1209) + if (sc->ident->device != 0x1209) sc->flags |= FXP_FLAG_82559_RXCSUM; } /* diff --git a/sys/dev/fxp/if_fxpreg.h b/sys/dev/fxp/if_fxpreg.h index 7fd60af..7ee8588 100644 --- a/sys/dev/fxp/if_fxpreg.h +++ b/sys/dev/fxp/if_fxpreg.h @@ -28,8 +28,6 @@ * $FreeBSD$ */ -#define FXP_VENDORID_INTEL 0x8086 - #define FXP_PCI_MMBA 0x10 #define FXP_PCI_IOBA 0x14 diff --git a/sys/dev/fxp/if_fxpvar.h b/sys/dev/fxp/if_fxpvar.h index e0097a1..78200ce 100644 --- a/sys/dev/fxp/if_fxpvar.h +++ b/sys/dev/fxp/if_fxpvar.h @@ -143,7 +143,8 @@ struct fxp_desc_list { }; struct fxp_ident { - uint16_t devid; + uint16_t vendor; + uint16_t device; int16_t revid; /* -1 matches anything */ uint8_t ich; const char *name; diff --git a/sys/dev/gpio/gpio_if.m b/sys/dev/gpio/gpio_if.m index f0ead65..adc119e 100644 --- a/sys/dev/gpio/gpio_if.m +++ b/sys/dev/gpio/gpio_if.m @@ -56,11 +56,11 @@ HEADER { }; # -# Get total number of pins +# Get maximum pin number # METHOD int pin_max { device_t dev; - int *npins; + int *maxpin; }; # diff --git a/sys/dev/ieee488/ibfoo.c b/sys/dev/ieee488/ibfoo.c deleted file mode 100644 index 2fe72dc..0000000 --- a/sys/dev/ieee488/ibfoo.c +++ /dev/null @@ -1,1127 +0,0 @@ -/*- - * Copyright (c) 2005 Poul-Henning Kamp <phk@FreeBSD.org> - * Copyright (c) 2010 Joerg Wunsch <joerg@FreeBSD.org> - * All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * High-level driver for µPD7210 based GPIB cards. - * - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -# define IBDEBUG -# undef IBDEBUG - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/conf.h> -#include <sys/malloc.h> -#include <sys/kernel.h> -#include <sys/limits.h> -#include <sys/module.h> -#include <sys/rman.h> -#include <sys/bus.h> -#include <sys/lock.h> -#include <sys/mutex.h> -#include <sys/uio.h> -#include <sys/time.h> -#include <machine/bus.h> -#include <machine/resource.h> -#include <isa/isavar.h> - -#include <dev/ieee488/ugpib.h> - -#define UPD7210_SW_DRIVER -#include <dev/ieee488/upd7210.h> -#include <dev/ieee488/tnt4882.h> - -static MALLOC_DEFINE(M_IBFOO, "IBFOO", "IBFOO"); - - -/* ibfoo API */ - -#include <dev/ieee488/ibfoo_int.h> - -/* XXX: This is really a bitmap */ -enum h_kind { - H_DEV = 1, - H_BOARD = 2, - H_EITHER = 3 -}; - -struct handle { - LIST_ENTRY(handle) list; - int handle; - enum h_kind kind; - int pad; - int sad; - struct timeval timeout; - int eot; - int eos; - int dma; -}; - -struct ibfoo { - struct upd7210 *u; - LIST_HEAD(,handle) handles; - struct unrhdr *unrhdr; - struct callout callout; - struct handle *h; - struct ibarg *ap; - - enum { - IDLE, - BUSY, - PIO_IDATA, - PIO_ODATA, - PIO_CMD, - DMA_IDATA, - FIFO_IDATA, - FIFO_ODATA, - FIFO_CMD - } mode; - - struct timeval deadline; - - struct handle *rdh; /* addressed for read */ - struct handle *wrh; /* addressed for write */ - - int doeoi; - - u_char *buf; - u_int buflen; -}; - -typedef int ibhandler_t(struct ibfoo *ib); - -static struct timeval timeouts[] = { - [TNONE] = { 0, 0}, - [T10us] = { 0, 10}, - [T30us] = { 0, 30}, - [T100us] = { 0, 100}, - [T300us] = { 0, 300}, - [T1ms] = { 0, 1000}, - [T3ms] = { 0, 3000}, - [T10ms] = { 0, 10000}, - [T30ms] = { 0, 30000}, - [T100ms] = { 0, 100000}, - [T300ms] = { 0, 300000}, - [T1s] = { 1, 0}, - [T3s] = { 3, 0}, - [T10s] = { 10, 0}, - [T30s] = { 30, 0}, - [T100s] = { 100, 0}, - [T300s] = { 300, 0}, - [T1000s] = { 1000, 0} -}; - -static const u_int max_timeouts = sizeof timeouts / sizeof timeouts[0]; - -static int ibdebug; - -static int -ib_set_error(struct ibarg *ap, int error) -{ - - if (ap->__iberr == 0) - ap->__iberr = error; - ap->__ibsta |= ERR; - ap->__retval = ap->__ibsta; - return (0); -} - -static int -ib_had_timeout(struct ibarg *ap) -{ - - ib_set_error(ap, EABO); - ap->__ibsta |= TIMO; - ap->__retval = ap->__ibsta; - return (0); -} - -static int -ib_set_errno(struct ibarg *ap, int errno) -{ - - if (ap->__iberr == 0) { - ap->__iberr = EDVR; - ap->__ibcnt = errno; - } - ap->__ibsta |= ERR; - ap->__retval = ap->__ibsta; - return (0); -} - -static int -gpib_ib_irq(struct upd7210 *u, int isr_3) -{ - struct ibfoo *ib; - - ib = u->ibfoo; - - mtx_assert(&u->mutex, MA_OWNED); - switch (ib->mode) { - case PIO_CMD: - if (!(u->rreg[ISR2] & IXR2_CO)) - return (0); - if (ib->buflen == 0) - break; - upd7210_wr(u, CDOR, *ib->buf); - ib->buf++; - ib->buflen--; - return (1); - case PIO_IDATA: - if (!(u->rreg[ISR1] & IXR1_DI)) - return (0); - *ib->buf = upd7210_rd(u, DIR); - ib->buf++; - ib->buflen--; - if (ib->buflen == 0 || (u->rreg[ISR1] & IXR1_ENDRX)) - break; - return (1); - case PIO_ODATA: - if (!(u->rreg[ISR1] & IXR1_DO)) - return (0); - if (ib->buflen == 0) - break; - if (ib->buflen == 1 && ib->doeoi) - upd7210_wr(u, AUXMR, AUXMR_SEOI); - upd7210_wr(u, CDOR, *ib->buf); - ib->buf++; - ib->buflen--; - return (1); - case DMA_IDATA: - if (!(u->rreg[ISR1] & IXR1_ENDRX)) - return (0); - break; - case FIFO_IDATA: - if (!(isr_3 & 0x15)) - return (0); - while (ib->buflen != 0 && (isr_3 & 0x04 /* NEF */) != 0) { - *ib->buf = bus_read_1(u->reg_res[0], fifob); - ib->buf++; - ib->buflen--; - isr_3 = bus_read_1(u->reg_res[0], isr3); - } - if ((isr_3 & 0x01) != 0 /* xfr done */ || - (u->rreg[ISR1] & IXR1_ENDRX) != 0 || - ib->buflen == 0) - break; - if (isr_3 & 0x10) - /* xfr stopped */ - bus_write_1(u->reg_res[0], cmdr, 0x04); /* GO */ - upd7210_wr(u, AUXMR, AUXMR_RFD); - return (1); - case FIFO_CMD: - case FIFO_ODATA: - if (!(isr_3 & 0x19)) - return (0); - if (ib->buflen == 0) - /* xfr DONE */ - break; - while (ib->buflen != 0 && (isr_3 & 0x08 /* NFF */) != 0) { - bus_write_1(u->reg_res[0], fifob, *ib->buf); - ib->buf++; - ib->buflen--; - isr_3 = bus_read_1(u->reg_res[0], isr3); - } - if (isr_3 & 0x10) - /* xfr stopped */ - bus_write_1(u->reg_res[0], cmdr, 0x04); /* GO */ - if (ib->buflen == 0) - /* no more NFF interrupts wanted */ - bus_write_1(u->reg_res[0], imr3, 0x11); /* STOP IE, DONE IE */ - return (1); - default: - return (0); - } - upd7210_wr(u, IMR1, 0); - upd7210_wr(u, IMR2, 0); - if (u->use_fifo) { - bus_write_1(u->reg_res[0], imr3, 0x00); - bus_write_1(u->reg_res[0], cmdr, 0x22); /* soft RESET */ - } - ib->mode = BUSY; - wakeup(&ib->buflen); - return (1); -} - -static void -gpib_ib_timeout(void *arg) -{ - struct upd7210 *u; - struct ibfoo *ib; - struct timeval tv; - u_int isr_3; - - u = arg; - ib = u->ibfoo; - mtx_lock(&u->mutex); - if (ib->mode == DMA_IDATA && isa_dmatc(u->dmachan)) { - KASSERT(u->dmachan >= 0, ("Bogus dmachan = %d", u->dmachan)); - upd7210_wr(u, IMR1, 0); - upd7210_wr(u, IMR2, 0); - ib->mode = BUSY; - wakeup(&ib->buflen); - } - if (ib->mode > BUSY) { - upd7210_rd(u, ISR1); - upd7210_rd(u, ISR2); - if (u->use_fifo) - isr_3 = bus_read_1(u->reg_res[0], isr3); - else - isr_3 = 0; - gpib_ib_irq(u, isr_3); - } - if (ib->mode != IDLE && timevalisset(&ib->deadline)) { - getmicrouptime(&tv); - if (timevalcmp(&ib->deadline, &tv, <)) { - ib_had_timeout(ib->ap); - upd7210_wr(u, IMR1, 0); - upd7210_wr(u, IMR2, 0); - if (u->use_fifo) { - bus_write_1(u->reg_res[0], imr3, 0x00); - bus_write_1(u->reg_res[0], cmdr, 0x22); /* soft RESET */ - } - ib->mode = BUSY; - wakeup(&ib->buflen); - } - } - if (ib->mode != IDLE) - callout_reset(&ib->callout, hz / 5, gpib_ib_timeout, arg); - mtx_unlock(&u->mutex); -} - -static void -gpib_ib_wait_xfer(struct upd7210 *u, struct ibfoo *ib) -{ - int i; - - mtx_assert(&u->mutex, MA_OWNED); - while (ib->mode > BUSY) { - i = msleep(&ib->buflen, &u->mutex, - PZERO | PCATCH, "ibwxfr", 0); - if (i == EINTR) { - ib_set_errno(ib->ap, i); - break; - } - if (u->rreg[ISR1] & IXR1_ERR) { - ib_set_error(ib->ap, EABO); /* XXX ? */ - break; - } - } - if ((u->rreg[ISR1] & IXR1_ENDRX) != 0) { - ib->ap->__retval |= END; - ib->ap->__ibsta |= END; - } - if ((u->rreg[ISR2] & IXR2_SRQI) != 0) { - ib->ap->__retval |= SRQI; - ib->ap->__ibsta |= SRQI; - } - ib->mode = BUSY; - ib->buf = NULL; - upd7210_wr(u, IMR1, 0); - upd7210_wr(u, IMR2, 0); - if (u->use_fifo) - bus_write_1(u->reg_res[0], imr3, 0x00); -} - -static void -config_eos(struct upd7210 *u, struct handle *h) -{ - int i; - - i = 0; - if (h->eos & REOS) { - upd7210_wr(u, EOSR, h->eos & 0xff); - i |= AUXA_REOS; - } - if (h->eos & XEOS) { - upd7210_wr(u, EOSR, h->eos & 0xff); - i |= AUXA_XEOS; - } - if (h->eos & BIN) - i |= AUXA_BIN; - upd7210_wr(u, AUXRA, C_AUXA | i); -} - -/* - * Look up the handle, and set the deadline if the handle has a timeout. - */ -static int -gethandle(struct upd7210 *u, struct ibarg *ap, struct handle **hp) -{ - struct ibfoo *ib; - struct handle *h; - - KASSERT(ap->__field & __F_HANDLE, ("gethandle without __F_HANDLE")); - ib = u->ibfoo; - LIST_FOREACH(h, &ib->handles, list) { - if (h->handle == ap->handle) { - *hp = h; - return (0); - } - } - ib_set_error(ap, EARG); - return (1); -} - -static int -pio_cmd(struct upd7210 *u, u_char *cmd, int len) -{ - struct ibfoo *ib; - - ib = u->ibfoo; - - if (ib->rdh != NULL || ib->wrh != NULL) { - upd7210_take_ctrl_async(u); - ib->rdh = NULL; - ib->wrh = NULL; - } - mtx_lock(&u->mutex); - ib->buf = cmd; - ib->buflen = len; - if (u->use_fifo) { - /* TNT5004 or TNT4882 in FIFO mode */ - ib->mode = FIFO_CMD; - upd7210_wr(u, AUXMR, 0x51); /* holdoff immediately */ - bus_write_1(u->reg_res[0], cmdr, 0x10); /* reset FIFO */ - bus_write_1(u->reg_res[0], cfg, 0x80); /* CMD, xfer OUT, 8-bit FIFO */ - bus_write_1(u->reg_res[0], imr3, 0x19); /* STOP IE, NFF IE, DONE IE */ - bus_write_1(u->reg_res[0], cnt0, -len); - bus_write_1(u->reg_res[0], cnt1, (-len) >> 8); - bus_write_1(u->reg_res[0], cnt2, (-len) >> 16); - bus_write_1(u->reg_res[0], cnt3, (-len) >> 24); - bus_write_1(u->reg_res[0], cmdr, 0x04); /* GO */ - } else { - ib->mode = PIO_CMD; - upd7210_wr(u, IMR2, IXR2_CO); - gpib_ib_irq(u, 0); - } - - gpib_ib_wait_xfer(u, ib); - - if (u->use_fifo) - bus_write_1(u->reg_res[0], cmdr, 0x08); /* STOP */ - - mtx_unlock(&u->mutex); - return (len - ib->buflen); -} - -static int -pio_odata(struct upd7210 *u, u_char *data, int len) -{ - struct ibfoo *ib; - - ib = u->ibfoo; - - if (len == 0) - return (0); - mtx_lock(&u->mutex); - ib->buf = data; - ib->buflen = len; - if (u->use_fifo) { - /* TNT5004 or TNT4882 in FIFO mode */ - ib->mode = FIFO_ODATA; - bus_write_1(u->reg_res[0], cmdr, 0x10); /* reset FIFO */ - if (ib->doeoi) - bus_write_1(u->reg_res[0], cfg, 0x08); /* CCEN */ - else - bus_write_1(u->reg_res[0], cfg, 0x00); /* xfer OUT, 8-bit FIFO */ - bus_write_1(u->reg_res[0], imr3, 0x19); /* STOP IE, NFF IE, DONE IE */ - bus_write_1(u->reg_res[0], cnt0, -len); - bus_write_1(u->reg_res[0], cnt1, (-len) >> 8); - bus_write_1(u->reg_res[0], cnt2, (-len) >> 16); - bus_write_1(u->reg_res[0], cnt3, (-len) >> 24); - bus_write_1(u->reg_res[0], cmdr, 0x04); /* GO */ - } else { - ib->mode = PIO_ODATA; - upd7210_wr(u, IMR1, IXR1_DO); - } - - gpib_ib_wait_xfer(u, ib); - - if (u->use_fifo) - bus_write_1(u->reg_res[0], cmdr, 0x08); /* STOP */ - - mtx_unlock(&u->mutex); - return (len - ib->buflen); -} - -static int -pio_idata(struct upd7210 *u, u_char *data, int len) -{ - struct ibfoo *ib; - - ib = u->ibfoo; - - mtx_lock(&u->mutex); - ib->buf = data; - ib->buflen = len; - if (u->use_fifo) { - /* TNT5004 or TNT4882 in FIFO mode */ - ib->mode = FIFO_IDATA; - bus_write_1(u->reg_res[0], cmdr, 0x10); /* reset FIFO */ - bus_write_1(u->reg_res[0], cfg, 0x20); /* xfer IN, 8-bit FIFO */ - bus_write_1(u->reg_res[0], cnt0, -len); - bus_write_1(u->reg_res[0], cnt1, (-len) >> 8); - bus_write_1(u->reg_res[0], cnt2, (-len) >> 16); - bus_write_1(u->reg_res[0], cnt3, (-len) >> 24); - bus_write_1(u->reg_res[0], cmdr, 0x04); /* GO */ - upd7210_wr(u, AUXMR, AUXMR_RFD); - bus_write_1(u->reg_res[0], imr3, 0x15); /* STOP IE, NEF IE, DONE IE */ - } else { - ib->mode = PIO_IDATA; - upd7210_wr(u, IMR1, IXR1_DI); - } - - gpib_ib_wait_xfer(u, ib); - - if (u->use_fifo) - bus_write_1(u->reg_res[0], cmdr, 0x08); /* STOP */ - - mtx_unlock(&u->mutex); - return (len - ib->buflen); -} - -static int -dma_idata(struct upd7210 *u, u_char *data, int len) -{ - int j; - struct ibfoo *ib; - - KASSERT(u->dmachan >= 0, ("Bogus dmachan %d", u->dmachan)); - ib = u->ibfoo; - ib->mode = DMA_IDATA; - isa_dmastart(ISADMA_READ, data, len, u->dmachan); - mtx_lock(&u->mutex); - upd7210_wr(u, IMR1, IXR1_ENDRX); - upd7210_wr(u, IMR2, IMR2_DMAI); - gpib_ib_wait_xfer(u, ib); - mtx_unlock(&u->mutex); - j = isa_dmastatus(u->dmachan); - isa_dmadone(ISADMA_READ, data, len, u->dmachan); - return (len - j); -} - -static int -ib_send_msg(struct ibfoo *ib, int msg) -{ - u_char buf[10]; - int i, j; - - i = 0; - buf[i++] = UNT; - buf[i++] = UNL; - buf[i++] = LAD | ib->h->pad; - if (ib->h->sad) - buf[i++] = LAD | TAD | ib->h->sad; - buf[i++] = TAD | 0; - buf[i++] = msg; - j = pio_cmd(ib->u, buf, i); - if (i != j) - ib_set_error(ib->ap, EABO); /* XXX ? */ - return (0); -} - -static int -ibask(struct ibfoo *ib) -{ /* XXX */ - - ibdebug = ib->ap->option; - return (0); -} - -#define ibbna NULL -#define ibcac NULL - -static int -ibclr(struct ibfoo *ib) -{ - - return (ib_send_msg(ib, SDC)); -} - -#define ibcmd NULL -#define ibcmda NULL -#define ibconfig NULL - -static int -ibdev(struct ibfoo *ib) -{ /* TBD */ - struct handle *h; - - h = malloc(sizeof *h, M_IBFOO, M_ZERO | M_WAITOK); - h->handle = alloc_unr(ib->unrhdr); - h->kind = H_DEV; - h->pad = ib->ap->pad; - h->sad = ib->ap->sad; - h->timeout = timeouts[ib->ap->tmo]; - h->eot = ib->ap->eot; - h->eos = ib->ap->eos; - mtx_lock(&ib->u->mutex); - LIST_INSERT_HEAD(&ib->handles, h, list); - mtx_unlock(&ib->u->mutex); - ib->ap->__retval = h->handle; - return (0); -} - -#define ibdiag NULL - -static int -ibdma(struct ibfoo *ib) -{ - - if (ib->u->dmachan < 0 && ib->ap->v) - return (ib_set_error(ib->ap, EARG)); - ib->h->dma = ib->ap->v; - return (0); -} - -static int -ibeos(struct ibfoo *ib) -{ - - ib->ap->__iberr = ib->h->eos; - ib->h->eos = ib->ap->eos; - if (ib->rdh == ib->h) - config_eos(ib->u, ib->h); - return (0); -} - -static int -ibeot(struct ibfoo *ib) -{ - - ib->h->eot = ib->ap->eot; - return (0); -} - -#define ibevent NULL -#define ibfind NULL -#define ibgts NULL -#define ibist NULL -#define iblines NULL -#define ibllo NULL -#define ibln NULL - -static int -ibloc(struct ibfoo *ib) -{ /* XXX */ - - if (ib->h->kind == H_BOARD) - return (EOPNOTSUPP); /* XXX */ - return (ib_send_msg(ib, GTL)); -} - -static int -ibonl(struct ibfoo *ib) -{ /* XXX */ - - if (ib->ap->v) - return (EOPNOTSUPP); /* XXX */ - mtx_lock(&ib->u->mutex); - LIST_REMOVE(ib->h, list); - mtx_unlock(&ib->u->mutex); - free(ib->h, M_IBFOO); - ib->h = NULL; - return (0); -} - -static int -ibpad(struct ibfoo *ib) -{ - - ib->h->pad = ib->ap->pad; - return (0); -} - -#define ibpct NULL -#define ibpoke NULL -#define ibppc NULL - -static int -ibrd(struct ibfoo *ib) -{ /* TBD */ - u_char buf[10], *bp; - int i, j, error, bl, bc; - u_char *dp; - - if (ib->h->kind == H_BOARD) - return (EOPNOTSUPP); /* XXX */ - bl = ib->ap->cnt; - if (bl > PAGE_SIZE) - bl = PAGE_SIZE; - bp = malloc(bl, M_IBFOO, M_WAITOK); - - if (ib->rdh != ib->h) { - i = 0; - buf[i++] = UNT; - buf[i++] = UNL; - buf[i++] = LAD | 0; - buf[i++] = TAD | ib->h->pad; - if (ib->h->sad) - buf[i++] = ib->h->sad; - i = pio_cmd(ib->u, buf, i); - config_eos(ib->u, ib->h); - ib->rdh = ib->h; - ib->wrh = NULL; - } - upd7210_goto_standby(ib->u); - dp = ib->ap->buffer; - bc = ib->ap->cnt; - error = 0; - while (bc > 0 && ib->ap->__iberr == 0) { - j = imin(bc, PAGE_SIZE); - if (ib->h->dma) - i = dma_idata(ib->u, bp, j); - else - i = pio_idata(ib->u, bp, j); - error = copyout(bp, dp , i); - if (error) - break; - ib->ap->__ibcnt += i; - if (i != j) - break; - bc -= i; - dp += i; - } - upd7210_take_ctrl_async(ib->u); - free(bp, M_IBFOO); - return (error); -} - -#define ibrda NULL -#define ibrdf NULL -#define ibrdkey NULL -#define ibrpp NULL -#define ibrsc NULL -#define ibrsp NULL -#define ibrsv NULL - -static int -ibsad(struct ibfoo *ib) -{ - - ib->h->sad = ib->ap->sad; - return (0); -} - -#define ibsgnl NULL - -static int -ibsic(struct ibfoo *ib) -{ /* TBD */ - - upd7210_wr(ib->u, AUXMR, AUXMR_SIFC); - DELAY(100); - upd7210_wr(ib->u, AUXMR, AUXMR_CIFC); - return (0); -} - -#define ibsre NULL -#define ibsrq NULL -#define ibstop NULL - -static int -ibtmo(struct ibfoo *ib) -{ - - ib->h->timeout = timeouts[ib->ap->tmo]; - return (0); -} - -#define ibtrap NULL - -static int -ibtrg(struct ibfoo *ib) -{ - - return (ib_send_msg(ib, GET)); -} - -#define ibwait NULL - -static int -ibwrt(struct ibfoo *ib) -{ /* XXX */ - u_char buf[10], *bp; - int i; - - if (ib->h->kind == H_BOARD) - return (EOPNOTSUPP); - bp = malloc(ib->ap->cnt, M_IBFOO, M_WAITOK); - /* XXX: bigger than PAGE_SIZE handling */ - i = copyin(ib->ap->buffer, bp, ib->ap->cnt); - if (i) { - free(bp, M_IBFOO); - return (i); - } - if (ib->wrh != ib->h) { - i = 0; - buf[i++] = UNT; - buf[i++] = UNL; - buf[i++] = LAD | ib->h->pad; - if (ib->h->sad) - buf[i++] = LAD | TAD | ib->h->sad; - buf[i++] = TAD | 0; - i = pio_cmd(ib->u, buf, i); - ib->rdh = NULL; - ib->wrh = ib->h; - config_eos(ib->u, ib->h); - } - upd7210_goto_standby(ib->u); - ib->doeoi = ib->h->eot; - i = pio_odata(ib->u, bp, ib->ap->cnt); - upd7210_take_ctrl_async(ib->u); - ib->ap->__ibcnt = i; - free(bp, M_IBFOO); - return (0); -} - -#define ibwrta NULL -#define ibwrtf NULL -#define ibwrtkey NULL -#define ibxtrc NULL - -static struct ibhandler { - const char *name; - enum h_kind kind; - ibhandler_t *func; - u_int args; -} ibhandlers[] = { - [__ID_IBASK] = { "ibask", H_EITHER, ibask, __F_HANDLE | __F_OPTION | __F_RETVAL }, - [__ID_IBBNA] = { "ibbna", H_DEV, ibbna, __F_HANDLE | __F_BDNAME }, - [__ID_IBCAC] = { "ibcac", H_BOARD, ibcac, __F_HANDLE | __F_V }, - [__ID_IBCLR] = { "ibclr", H_DEV, ibclr, __F_HANDLE }, - [__ID_IBCMD] = { "ibcmd", H_BOARD, ibcmd, __F_HANDLE | __F_BUFFER | __F_CNT }, - [__ID_IBCMDA] = { "ibcmda", H_BOARD, ibcmda, __F_HANDLE | __F_BUFFER | __F_CNT }, - [__ID_IBCONFIG] = { "ibconfig", H_EITHER, ibconfig, __F_HANDLE | __F_OPTION | __F_VALUE }, - [__ID_IBDEV] = { "ibdev", 0, ibdev, __F_BOARDID | __F_PAD | __F_SAD | __F_TMO | __F_EOT | __F_EOS }, - [__ID_IBDIAG] = { "ibdiag", H_EITHER, ibdiag, __F_HANDLE | __F_BUFFER | __F_CNT }, - [__ID_IBDMA] = { "ibdma", H_EITHER, ibdma, __F_HANDLE | __F_V }, - [__ID_IBEOS] = { "ibeos", H_EITHER, ibeos, __F_HANDLE | __F_EOS }, - [__ID_IBEOT] = { "ibeot", H_EITHER, ibeot, __F_HANDLE | __F_EOT }, - [__ID_IBEVENT] = { "ibevent", H_BOARD, ibevent, __F_HANDLE | __F_EVENT }, - [__ID_IBFIND] = { "ibfind", 0, ibfind, __F_BDNAME }, - [__ID_IBGTS] = { "ibgts", H_BOARD, ibgts, __F_HANDLE | __F_V }, - [__ID_IBIST] = { "ibist", H_BOARD, ibist, __F_HANDLE | __F_V }, - [__ID_IBLINES] = { "iblines", H_BOARD, iblines, __F_HANDLE | __F_LINES }, - [__ID_IBLLO] = { "ibllo", H_EITHER, ibllo, __F_HANDLE }, - [__ID_IBLN] = { "ibln", H_BOARD, ibln, __F_HANDLE | __F_PADVAL | __F_SADVAL | __F_LISTENFLAG }, - [__ID_IBLOC] = { "ibloc", H_EITHER, ibloc, __F_HANDLE }, - [__ID_IBONL] = { "ibonl", H_EITHER, ibonl, __F_HANDLE | __F_V }, - [__ID_IBPAD] = { "ibpad", H_EITHER, ibpad, __F_HANDLE | __F_PAD }, - [__ID_IBPCT] = { "ibpct", H_DEV, ibpct, __F_HANDLE }, - [__ID_IBPOKE] = { "ibpoke", H_EITHER, ibpoke, __F_HANDLE | __F_OPTION | __F_VALUE }, - [__ID_IBPPC] = { "ibppc", H_EITHER, ibppc, __F_HANDLE | __F_V }, - [__ID_IBRD] = { "ibrd", H_EITHER, ibrd, __F_HANDLE | __F_BUFFER | __F_CNT }, - [__ID_IBRDA] = { "ibrda", H_EITHER, ibrda, __F_HANDLE | __F_BUFFER | __F_CNT }, - [__ID_IBRDF] = { "ibrdf", H_EITHER, ibrdf, __F_HANDLE | __F_FLNAME }, - [__ID_IBRDKEY] = { "ibrdkey", H_EITHER, ibrdkey, __F_HANDLE | __F_BUFFER | __F_CNT }, - [__ID_IBRPP] = { "ibrpp", H_EITHER, ibrpp, __F_HANDLE | __F_PPR }, - [__ID_IBRSC] = { "ibrsc", H_BOARD, ibrsc, __F_HANDLE | __F_V }, - [__ID_IBRSP] = { "ibrsp", H_DEV, ibrsp, __F_HANDLE | __F_SPR }, - [__ID_IBRSV] = { "ibrsv", H_EITHER, ibrsv, __F_HANDLE | __F_V }, - [__ID_IBSAD] = { "ibsad", H_EITHER, ibsad, __F_HANDLE | __F_SAD }, - [__ID_IBSGNL] = { "ibsgnl", H_EITHER, ibsgnl, __F_HANDLE | __F_V }, - [__ID_IBSIC] = { "ibsic", H_BOARD, ibsic, __F_HANDLE }, - [__ID_IBSRE] = { "ibsre", H_BOARD, ibsre, __F_HANDLE | __F_V }, - [__ID_IBSRQ] = { "ibsrq", H_EITHER, ibsrq, __F_FUNC }, - [__ID_IBSTOP] = { "ibstop", H_EITHER, ibstop, __F_HANDLE }, - [__ID_IBTMO] = { "ibtmo", H_EITHER, ibtmo, __F_HANDLE | __F_TMO }, - [__ID_IBTRAP] = { "ibtrap", H_EITHER, ibtrap, __F_MASK | __F_MODE }, - [__ID_IBTRG] = { "ibtrg", H_DEV, ibtrg, __F_HANDLE }, - [__ID_IBWAIT] = { "ibwait", H_EITHER, ibwait, __F_HANDLE | __F_MASK }, - [__ID_IBWRT] = { "ibwrt", H_EITHER, ibwrt, __F_HANDLE | __F_BUFFER | __F_CNT }, - [__ID_IBWRTA] = { "ibwrta", H_EITHER, ibwrta, __F_HANDLE | __F_BUFFER | __F_CNT }, - [__ID_IBWRTF] = { "ibwrtf", H_EITHER, ibwrtf, __F_HANDLE | __F_FLNAME }, - [__ID_IBWRTKEY] = { "ibwrtkey", H_EITHER, ibwrtkey, __F_HANDLE | __F_BUFFER | __F_CNT }, - [__ID_IBXTRC] = { "ibxtrc", H_EITHER, ibxtrc, __F_HANDLE | __F_BUFFER | __F_CNT }, -}; - -static const u_int max_ibhandler = sizeof ibhandlers / sizeof ibhandlers[0]; - -static void -ib_dump_args(struct ibhandler *ih, struct ibarg *ap) -{ - - if (ih->name != NULL) - printf("%s(", ih->name); - else - printf("ibinvalid("); - printf("[0x%x]", ap->__field); - if (ap->__field & __F_HANDLE) printf(" handle=%d", ap->handle); - if (ap->__field & __F_EOS) printf(" eos=0x%x", ap->eos); - if (ap->__field & __F_EOT) printf(" eot=%d", ap->eot); - if (ap->__field & __F_TMO) printf(" tmo=%d", ap->tmo); - if (ap->__field & __F_PAD) printf(" pad=0x%x", ap->pad); - if (ap->__field & __F_SAD) printf(" sad=0x%x", ap->sad); - if (ap->__field & __F_BUFFER) printf(" buffer=%p", ap->buffer); - if (ap->__field & __F_CNT) printf(" cnt=%ld", ap->cnt); - if (ap->__field & __F_V) printf(" v=%d/0x%x", ap->v, ap->v); - /* XXX more ... */ - printf(")\n"); -} - -static int -gpib_ib_open(struct cdev *dev, int oflags, int devtype, struct thread *td) -{ - struct upd7210 *u; - struct ibfoo *ib; - int error = 0; - - u = dev->si_drv1; - - mtx_lock(&u->mutex); - if (u->busy) { - mtx_unlock(&u->mutex); - return (EBUSY); - } - u->busy = 1; - mtx_unlock(&u->mutex); - - if (u->dmachan >= 0) { - error = isa_dma_acquire(u->dmachan); - if (!error) { - error = isa_dma_init(u->dmachan, PAGE_SIZE, M_WAITOK); - if (error) - isa_dma_release(u->dmachan); - } - } - - if (error) { - mtx_lock(&u->mutex); - u->busy = 0; - mtx_unlock(&u->mutex); - return (error); - } - - ib = malloc(sizeof *ib, M_IBFOO, M_WAITOK | M_ZERO); - LIST_INIT(&ib->handles); - callout_init(&ib->callout, CALLOUT_MPSAFE); - ib->unrhdr = new_unrhdr(0, INT_MAX, NULL); - dev->si_drv2 = ib; - ib->u = u; - u->ibfoo = ib; - u->irq = gpib_ib_irq; - - upd7210_wr(u, AUXMR, AUXMR_CRST); - DELAY(10000); - DELAY(1000); - upd7210_wr(u, IMR1, 0x00); - upd7210_wr(u, IMR2, 0x00); - upd7210_wr(u, SPMR, 0x00); - upd7210_wr(u, ADR, 0x00); - upd7210_wr(u, ADR, ADR_ARS | ADR_DL | ADR_DT); - upd7210_wr(u, ADMR, ADMR_ADM0 | ADMR_TRM0 | ADMR_TRM1); - upd7210_wr(u, EOSR, 0x00); - upd7210_wr(u, AUXMR, C_ICR | 8); - upd7210_wr(u, AUXMR, C_PPR | PPR_U); - upd7210_wr(u, AUXMR, C_AUXA); - upd7210_wr(u, AUXMR, C_AUXB + 3); - upd7210_wr(u, AUXMR, C_AUXE + 0); - upd7210_wr(u, AUXMR, AUXMR_PON); - if (u->use_fifo) { - bus_write_1(u->reg_res[0], imr3, 0x00); - bus_write_1(u->reg_res[0], cmdr, 0x22); /* soft reset */ - bus_write_1(u->reg_res[0], cmdr, 0x03); /* set system - * controller bit */ - } - upd7210_wr(u, AUXMR, AUXMR_CIFC); - DELAY(100); - upd7210_wr(u, AUXMR, AUXMR_SIFC); - upd7210_wr(u, AUXMR, AUXMR_SREN); - return (0); -} - -static int -gpib_ib_close(struct cdev *dev, int oflags, int devtype, struct thread *td) -{ - struct upd7210 *u; - struct ibfoo *ib; - - u = dev->si_drv1; - ib = dev->si_drv2; - /* XXX: assert pointer consistency */ - - u->ibfoo = NULL; - /* XXX: free handles */ - dev->si_drv2 = NULL; - free(ib, M_IBFOO); - - if (u->dmachan >= 0) { - isa_dma_release(u->dmachan); - } - mtx_lock(&u->mutex); - u->busy = 0; - ibdebug = 0; - upd7210_wr(u, IMR1, 0x00); - upd7210_wr(u, IMR2, 0x00); - if (u->use_fifo) { - bus_write_1(u->reg_res[0], imr3, 0x00); - bus_write_1(u->reg_res[0], cmdr, 0x02); /* clear system - * controller bit */ - } - upd7210_wr(u, AUXMR, AUXMR_CRST); - DELAY(10000); - mtx_unlock(&u->mutex); - return (0); -} - -static int -gpib_ib_ioctl(struct cdev *dev, u_long cmd, caddr_t data, int fflag, struct thread *td) -{ - struct ibarg *ap; - struct ibhandler *ih; - struct handle *h; - struct upd7210 *u; - struct ibfoo *ib; - int error; - struct timeval deadline, tv; - - u = dev->si_drv1; - ib = u->ibfoo; - - /* We only support a single ioctl, everything else is a mistake */ - if (cmd != GPIB_IBFOO) - return (ENOIOCTL); - - /* Check the identifier and field-bitmap in the arguments. */ - ap = (void *)data; - if (ap->__ident < 0 || ap->__ident >= max_ibhandler) - return (EINVAL); - ih = &ibhandlers[ap->__ident]; - if (ap->__field != ih->args) - return (EINVAL); - - if (ibdebug) - ib_dump_args(ih, ap); - - if (ih->func == NULL) - return (EOPNOTSUPP); - - ap->__iberr = 0; - ap->__ibsta = 0; - ap->__ibcnt = 0; - ap->__retval = 0; - - if (ap->__field & __F_TMO) { - if (ap->tmo < 0 || ap->tmo >= max_timeouts) - return (ib_set_error(ap, EARG)); - } - - if (ap->__field & __F_EOS) { - if ((ap->eos & ~(REOS | XEOS | BIN | 0xff)) || - ((ap->eos & (BIN | 0x80)) == 0x80)) - return (ib_set_error(ap, EARG)); - } - if (ap->__field & __F_PAD) { - if (ap->pad < 0 || ap->pad > 30) - return (ib_set_error(ap, EARG)); - } - if (ap->__field & __F_SAD) { - if (ap->sad != 0 && (ap->sad < 0x60 || ap->sad > 126)) - return (ib_set_error(ap, EARG)); - } - - - mtx_lock(&u->mutex); - - - /* Find the handle, if any */ - h = NULL; - if ((ap->__field & __F_HANDLE) && gethandle(u, ap, &h)) { - mtx_unlock(&u->mutex); - return (0); - } - - /* Check that the handle is the right kind */ - if (h != NULL && !(h->kind & ih->kind)) { - mtx_unlock(&u->mutex); - return (ib_set_error(ap, EARG)); - } - - /* Set up handle and deadline */ - if (h != NULL && timevalisset(&h->timeout)) { - getmicrouptime(&deadline); - timevaladd(&deadline, &h->timeout); - } else { - timevalclear(&deadline); - } - - /* Wait for the card to be(come) available, respect deadline */ - while(u->busy != 1) { - error = msleep(ib, &u->mutex, - PZERO | PCATCH, "gpib_ibioctl", hz / 10); - if (error == 0) - continue; - mtx_unlock(&u->mutex); - if (error == EINTR) - return(ib_set_error(ap, EABO)); - if (error == EWOULDBLOCK && timevalisset(&deadline)) { - getmicrouptime(&tv); - if (timevalcmp(&deadline, &tv, <)) - return(ib_had_timeout(ap)); - } - mtx_lock(&u->mutex); - } - u->busy = 2; - mtx_unlock(&u->mutex); - - /* Hand over deadline handling to the callout routine */ - ib->ap = ap; - ib->h = h; - ib->mode = BUSY; - ib->deadline = deadline; - callout_reset(&ib->callout, hz / 5, gpib_ib_timeout, u); - - error = ih->func(ib); - - /* Release card */ - ib->mode = IDLE; - ib->ap = NULL; - ib->h = NULL; - timevalclear(&deadline); - callout_stop(&ib->callout); - - mtx_lock(&u->mutex); - u->busy = 1; - wakeup(ib); - mtx_unlock(&u->mutex); - - if (error) - return(ib_set_errno(ap, error)); - return (0); -} - -struct cdevsw gpib_ib_cdevsw = { - .d_version = D_VERSION, - .d_name = "gpib_ib", - .d_open = gpib_ib_open, - .d_ioctl = gpib_ib_ioctl, - .d_close = gpib_ib_close, -}; diff --git a/sys/dev/ieee488/ibfoo_int.h b/sys/dev/ieee488/ibfoo_int.h deleted file mode 100644 index 28f7289..0000000 --- a/sys/dev/ieee488/ibfoo_int.h +++ /dev/null @@ -1,147 +0,0 @@ -/*- - * Copyright (c) 2005 Poul-Henning Kamp - * All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * This file defines the ABI between the userland gpib library and the - * kernel. This file should not be used anywhere else. - * - * $FreeBSD$ - */ - -#include <sys/ioccom.h> - -typedef void ibsrq_t(void); -enum ibfoo_id { - __ID_INVALID = 0, - __ID_IBASK, - __ID_IBBNA, - __ID_IBCAC, - __ID_IBCLR, - __ID_IBCMD, - __ID_IBCMDA, - __ID_IBCONFIG, - __ID_IBDEV, - __ID_IBDIAG, - __ID_IBDMA, - __ID_IBEOS, - __ID_IBEOT, - __ID_IBEVENT, - __ID_IBFIND, - __ID_IBGTS, - __ID_IBIST, - __ID_IBLINES, - __ID_IBLLO, - __ID_IBLN, - __ID_IBLOC, - __ID_IBONL, - __ID_IBPAD, - __ID_IBPCT, - __ID_IBPOKE, - __ID_IBPPC, - __ID_IBRD, - __ID_IBRDA, - __ID_IBRDF, - __ID_IBRDKEY, - __ID_IBRPP, - __ID_IBRSC, - __ID_IBRSP, - __ID_IBRSV, - __ID_IBSAD, - __ID_IBSGNL, - __ID_IBSIC, - __ID_IBSRE, - __ID_IBSRQ, - __ID_IBSTOP, - __ID_IBTMO, - __ID_IBTRAP, - __ID_IBTRG, - __ID_IBWAIT, - __ID_IBWRT, - __ID_IBWRTA, - __ID_IBWRTF, - __ID_IBWRTKEY, - __ID_IBXTRC -}; - -#define __F_HANDLE (1 << 0) -#define __F_SPR (1 << 1) -#define __F_BUFFER (1 << 2) -#define __F_RETVAL (1 << 3) -#define __F_BDNAME (1 << 4) -#define __F_MASK (1 << 5) -#define __F_PADVAL (1 << 6) -#define __F_SADVAL (1 << 7) -#define __F_CNT (1 << 8) -#define __F_TMO (1 << 9) -#define __F_EOS (1 << 10) -#define __F_PPR (1 << 11) -#define __F_EOT (1 << 12) -#define __F_V (1 << 13) -#define __F_VALUE (1 << 14) -#define __F_SAD (1 << 15) -#define __F_BOARDID (1 << 16) -#define __F_OPTION (1 << 17) -#define __F_FLNAME (1 << 18) -#define __F_FUNC (1 << 19) -#define __F_LINES (1 << 20) -#define __F_PAD (1 << 21) -#define __F_MODE (1 << 22) -#define __F_LISTENFLAG (1 << 23) -#define __F_EVENT (1 << 24) - -struct ibarg { - enum ibfoo_id __ident; - unsigned int __field; - int __retval; - int __ibsta; - int __iberr; - int __ibcnt; - int handle; - char * spr; - void * buffer; - int * retval; - char * bdname; - int mask; - int padval; - int sadval; - long cnt; - int tmo; - int eos; - char * ppr; - int eot; - int v; - int value; - int sad; - int boardID; - int option; - char * flname; - ibsrq_t * func; - short * lines; - int pad; - int mode; - short * listenflag; - short * event; -}; - -#define GPIB_IBFOO _IOWR(4, 0, struct ibarg) diff --git a/sys/dev/ieee488/pcii.c b/sys/dev/ieee488/pcii.c deleted file mode 100644 index 833ec91..0000000 --- a/sys/dev/ieee488/pcii.c +++ /dev/null @@ -1,255 +0,0 @@ -/*- - * Copyright (c) 2005 Poul-Henning Kamp <phk@FreeBSD.org> - * Copyright (c) 2010 Joerg Wunsch <joerg@FreeBSD.org> - * All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * Driver for GPIB cards based on NEC µPD7210 and compatibles. - * - * This driver just hooks up to the hardware and leaves all the interesting - * stuff to upd7210.c. - * - * Supported hardware: - * PCIIA compatible cards. - * - * Tested and known working: - * "B&C Microsystems PC488A-0" - * "National Instruments GPIB-PCII/PCIIA" (in PCIIa mode) - * "Axiom AX5488" - * - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/lock.h> -#include <sys/mutex.h> -#include <sys/kernel.h> -#include <sys/module.h> -#include <sys/bus.h> -#include <machine/bus.h> -#include <machine/resource.h> -#include <sys/rman.h> -#include <isa/isavar.h> - -#define UPD7210_HW_DRIVER -#include <dev/ieee488/upd7210.h> - -struct pcii_softc { - int foo; - struct resource *res[11]; - void *intr_handler; - struct upd7210 upd7210; -}; - -static devclass_t pcii_devclass; - -static int pcii_probe(device_t dev); -static int pcii_attach(device_t dev); - -static device_method_t pcii_methods[] = { - DEVMETHOD(device_probe, pcii_probe), - DEVMETHOD(device_attach, pcii_attach), - DEVMETHOD(device_suspend, bus_generic_suspend), - DEVMETHOD(device_resume, bus_generic_resume), - - { 0, 0 } -}; - -static struct resource_spec pcii_res_spec[] = { - { SYS_RES_IRQ, 0, RF_ACTIVE | RF_SHAREABLE}, - { SYS_RES_DRQ, 0, RF_ACTIVE | RF_SHAREABLE | RF_OPTIONAL}, - { SYS_RES_IOPORT, 0, RF_ACTIVE}, - { SYS_RES_IOPORT, 1, RF_ACTIVE}, - { SYS_RES_IOPORT, 2, RF_ACTIVE}, - { SYS_RES_IOPORT, 3, RF_ACTIVE}, - { SYS_RES_IOPORT, 4, RF_ACTIVE}, - { SYS_RES_IOPORT, 5, RF_ACTIVE}, - { SYS_RES_IOPORT, 6, RF_ACTIVE}, - { SYS_RES_IOPORT, 7, RF_ACTIVE}, - { SYS_RES_IOPORT, 8, RF_ACTIVE | RF_SHAREABLE}, - { -1, 0, 0 } -}; - -static driver_t pcii_driver = { - "pcii", - pcii_methods, - sizeof(struct pcii_softc), -}; - -static int -pcii_probe(device_t dev) -{ - int rid, i, j; - u_long start, count, addr; - int error = 0; - struct pcii_softc *sc; - - device_set_desc(dev, "PCII IEEE-4888 controller"); - sc = device_get_softc(dev); - - rid = 0; - if (bus_get_resource(dev, SYS_RES_IOPORT, rid, &start, &count) != 0) - return ENXIO; - /* - * The PCIIA decodes a fixed pattern of 0x2e1 for the lower 10 - * address bits A0 ... A9. Bits A10 through A12 are used by - * the µPD7210 register select lines. This makes the - * individual 7210 register being 0x400 bytes apart in the ISA - * bus address space. Address bits A13 and A14 are compared - * to a DIP switch setting on the card, allowing for up to 4 - * different cards being installed (at base addresses 0x2e1, - * 0x22e1, 0x42e1, and 0x62e1, respectively). A15 has been - * used to select an optional on-board time-of-day clock chip - * (MM58167A) on the original PCIIA rather than the µPD7210 - * (which is not implemented on later boards). The - * documentation states the respective addresses for that chip - * should be handled as reserved addresses, which we don't do - * (right now). Finally, the IO addresses 0x2f0 ... 0x2f7 for - * a "special interrupt handling feature" (re-enable - * interrupts so the IRQ can be shared). - * - * Usually, the user will only set the base address in the - * device hints, so we handle the rest here. - * - * (Source: GPIB-PCIIA Technical Reference Manual, September - * 1989 Edition, National Instruments.) - */ - if ((start & 0x3ff) != 0x2e1) { - if (bootverbose) - printf("pcii_probe: PCIIA base address 0x%lx not " - "0x2e1/0x22e1/0x42e1/0x62e1\n", - start); - return (ENXIO); - } - - for (rid = 0, addr = start; rid < 8; rid++, addr += 0x400) { - if (bus_set_resource(dev, SYS_RES_IOPORT, rid, addr, 1) != 0) { - printf("pcii_probe: could not set IO port 0x%lx\n", - addr); - return (ENXIO); - } - } - if (bus_get_resource(dev, SYS_RES_IRQ, 0, &start, &count) != 0) { - printf("pcii_probe: cannot obtain IRQ level\n"); - return ENXIO; - } - if (start > 7) { - printf("pcii_probe: IRQ level %lu too high\n", start); - return ENXIO; - } - - if (bus_set_resource(dev, SYS_RES_IOPORT, 8, 0x2f0 + start, 1) != 0) { - printf("pcii_probe: could not set IO port 0x%3lx\n", - 0x2f0 + start); - return (ENXIO); - } - - error = bus_alloc_resources(dev, pcii_res_spec, sc->res); - if (error) { - printf("pcii_probe: Could not allocate resources\n"); - return (error); - } - error = ENXIO; - /* - * Perform some basic tests on the µPD7210 registers. At - * least *some* register must read different from 0x00 or - * 0xff. - */ - for (i = 0; i < 8; i++) { - j = bus_read_1(sc->res[2 + i], 0); - if (j != 0x00 && j != 0xff) - error = 0; - } - /* SPSR/SPMR read/write test */ - if (!error) { - bus_write_1(sc->res[2 + 3], 0, 0x55); - if (bus_read_1(sc->res[2 + 3], 0) != 0x55) - error = ENXIO; - } - if (!error) { - bus_write_1(sc->res[2 + 3], 0, 0xaa); - if (bus_read_1(sc->res[2 + 3], 0) != 0xaa) - error = ENXIO; - } - if (error) - printf("pcii_probe: probe failure\n"); - - bus_release_resources(dev, pcii_res_spec, sc->res); - return (error); -} - -static int -pcii_attach(device_t dev) -{ - struct pcii_softc *sc; - u_long start, count; - int unit; - int rid; - int error = 0; - - unit = device_get_unit(dev); - sc = device_get_softc(dev); - memset(sc, 0, sizeof *sc); - - device_set_desc(dev, "PCII IEEE-4888 controller"); - - if (bus_get_resource(dev, SYS_RES_IRQ, 0, &start, &count) != 0) { - printf("pcii_attach: cannot obtain IRQ number\n"); - return ENXIO; - } - - error = bus_alloc_resources(dev, pcii_res_spec, sc->res); - if (error) - return (error); - - error = bus_setup_intr(dev, sc->res[0], - INTR_TYPE_MISC | INTR_MPSAFE, NULL, - upd7210intr, &sc->upd7210, &sc->intr_handler); - if (error) { - bus_release_resources(dev, pcii_res_spec, sc->res); - return (error); - } - - for (rid = 0; rid < 8; rid++) { - sc->upd7210.reg_res[rid] = sc->res[2 + rid]; - sc->upd7210.reg_offset[rid] = 0; - } - sc->upd7210.irq_clear_res = sc->res[10]; - sc->upd7210.use_fifo = 0; - - if (sc->res[1] == NULL) - sc->upd7210.dmachan = -1; - else - sc->upd7210.dmachan = rman_get_start(sc->res[1]); - - upd7210attach(&sc->upd7210); - device_printf(dev, "attached gpib%d\n", sc->upd7210.unit); - - return (0); -} - -DRIVER_MODULE(pcii, isa, pcii_driver, pcii_devclass, 0, 0); -DRIVER_MODULE(pcii, acpi, pcii_driver, pcii_devclass, 0, 0); diff --git a/sys/dev/ieee488/tnt4882.c b/sys/dev/ieee488/tnt4882.c deleted file mode 100644 index 4b69d0d..0000000 --- a/sys/dev/ieee488/tnt4882.c +++ /dev/null @@ -1,320 +0,0 @@ -/*- - * Copyright (c) 2005 Poul-Henning Kamp - * Copyright (c) 2010 Joerg Wunsch - * All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * $FreeBSD$ - */ - -#include <sys/param.h> -#include <sys/kernel.h> -#include <sys/systm.h> -#include <sys/module.h> -#include <sys/bus.h> -#include <dev/pci/pcireg.h> -#include <dev/pci/pcivar.h> - -#include <machine/bus.h> -#include <machine/resource.h> -#include <machine/stdarg.h> -#include <sys/rman.h> - -/* vtophys */ -#include <vm/vm.h> -#include <vm/pmap.h> -#include <machine/pmap.h> - -#define UPD7210_HW_DRIVER 1 -#include <dev/ieee488/upd7210.h> -#include <dev/ieee488/tnt4882.h> - -struct tnt_softc { - int foo; - struct upd7210 upd7210; - - struct resource *res[3]; - void *intr_handler; -}; - -static struct resource_spec tnt_res_spec[] = { - { SYS_RES_MEMORY, PCIR_BAR(0), RF_ACTIVE}, - { SYS_RES_MEMORY, PCIR_BAR(1), RF_ACTIVE}, - { SYS_RES_IRQ, 0, RF_ACTIVE | RF_SHAREABLE}, - { -1, 0 } -}; - -struct tst { - enum {RD, WT, xDELAY, END} - action; - enum tnt4882reg reg; - uint8_t val; -}; - -/* - * From NI Application note 095: - * Writing Functional Self-Tests for the TNT4882 GPIB Interface Chip - * XXX: fill in the rest ? - */ -static struct tst tst_reset[] = { - {WT, tauxcr, 0x80}, /* chip reset if in 9914 mode */ - {WT, auxmr, 0x80}, /* swrst if swapped */ - {WT, tauxcr, 0x99}, /* switch to 7210 mode */ - {WT, auxmr, 0x99}, /* switch to 7210 mode if swapped */ - {WT, auxmr, 0x02}, /* execute chip reset */ - {WT, keyreg, 0x00}, /* important! clear the swap bit */ - {WT, eosr, 0x00}, /* clear EOS register */ - {WT, cdor, 0x00}, /* clear data lines */ - {WT, imr1, 0x00}, /* disable all interrupts */ - {WT, imr2, 0x00}, - {WT, imr0, 0x80}, - {WT, adr, 0x80}, - {WT, adr, 0x00}, - {WT, admr, 0x00}, /* clear addressing modes */ - {WT, auxmr, 0x00}, /* release from idle state with pon */ - {WT, auxmr, 0x60}, /* reset ppr */ - {WT, bcr, 0x00}, /* reset bcr */ - {WT, misc, 0x04}, /* set wrap plug bit */ - {WT, cmdr, 0xB2}, /* issue soft reset */ - {WT, hssel, 0x00}, /* select two-chip mode */ - {END, 0, 0} -}; - -static struct tst tst_read_reg[] = { - {RD, isr1, 0x00}, /* Verify mask registers are clear */ - {RD, isr2, 0x00}, - {RD, adsr, 0x40}, /* Verify ATN is not asserted */ - {RD, adr0, 0x00}, /* Verify Primary address not set */ - {RD, adr1, 0x00}, /* Verify Secondary address not set */ - {RD, sts1, 0x8B}, /* Verify DONE, STOP, HALT, and GSYNC set */ - {RD, isr3, 0x19}, /* Verify STOP, Not Full FIFO, & DONE set */ - {RD, sts2, 0x9A}, /* Verify FIFO A/B is empty */ - {RD, sasr, 0x00}, /* Verify clear */ - {RD, isr0, 0x01}, /* Verify SYNC bit is set */ - {END, 0, 0} -}; - -static struct tst tst_bsr_dcr[] = { - {WT, bcr, 0x55}, /* Set DAV, NRFD, SRQ, and REN */ - {WT, dcr, 0xAA}, /* Write pattern to GPIB data lines */ - {RD, bsr, 0x55}, /* Verify DAV, NRFD, SRQ, and REN are set */ - {RD, dsr, 0xAA}, /* Verify data pattern written previously */ - {WT, bcr, 0xAA}, /* Set ATN, NDAC, EOI, & IFC */ - {WT, dcr, 0x55}, /* Write pattern to GPIB data lines */ - {RD, bsr, 0xAA}, /* Verify ATN, NDAC, EOI, & IFC are set */ - {RD, dsr, 0x55}, /* Verify data pattern written previously */ - {WT, bcr, 0x00}, /* Clear control lines */ - {WT, dcr, 0x00}, /* Clear data lines */ - {RD, bsr, 0x00}, /* Verify control lines are clear */ - {RD, dsr, 0x00}, /* Verify data lines are clear */ - {END, 0, 0} -}; - -static struct tst tst_adr0_1[] = { - {WT, adr, 0x55}, /* Set Primary talk address */ - {WT, adr, 0xAA}, /* Set Secondary listen address */ - {RD, adr0, 0x55}, /* Read Primary address */ - {RD, adr1, 0x2A}, /* Read Secondary address */ - {WT, adr, 0x2A}, /* Set Primay listen address */ - {WT, adr, 0xD5}, /* Set Secondary talk address */ - {RD, adr0, 0x2A}, /* Read Primary address */ - {RD, adr1, 0x55}, /* Read Secondary address */ - {END, 0, 0} -}; - -static struct tst tst_cdor_dir[] = { - {WT, admr, 0xF0}, /* program AT-GPIB as talker only and - * listener only */ - {RD, isr1, 0x02}, /* check DO bit set */ - {RD, adsr, 0x46}, /* check AT-GPIB is both talker active - * and listener active */ - {WT, cdor, 0xAA}, /* write out data byte */ - {xDELAY, 0, 1}, /* One ISA I/O Cycle (500-ns) */ - {RD, isr1, 0x03}, /* check DO and DI bits set */ - {RD, dir, 0xAA}, /* verify data received */ - {WT, cdor, 0x55}, /* write out data byte */ - {xDELAY, 0, 1}, /* One ISA I/O Cycle (500-ns) */ - {RD, dir, 0x55}, /* verify data received */ - {END, 0, 0} -}; - -static struct tst tst_spmr_spsr[] = { - {WT, spsr, 0x00}, /* Write pattern to SPSR register */ - {RD, spmr, 0x00}, /* Read back previously written pattern */ - {WT, spsr, 0xBF}, /* Write pattern to SPSR register */ - {RD, spmr, 0xBF}, /* Read back previously written pattern */ - {END, 0, 0} -}; - -static struct tst tst_count0_1[] = { - {WT, cnt0, 0x55}, /* Verify every other bit can be set */ - {WT, cnt1, 0xAA}, - {RD, cnt0, 0x55}, /* Read back previously written pattern */ - {RD, cnt1, 0xAA}, - {WT, cnt0, 0xAA}, /* Verify every other bit can be set */ - {WT, cnt1, 0x55}, - {RD, cnt0, 0xAA}, /* Read back previously written pattern */ - {RD, cnt1, 0x55}, - {END, 0, 0} -}; - -static int -tst_exec(struct tnt_softc *sc, struct tst *tp, const char *name) -{ - uint8_t u; - int step; - - for (step = 0; tp->action != END; tp++, step++) { - switch (tp->action) { - case WT: - bus_write_1(sc->res[1], tp->reg, tp->val); - break; - case RD: - u = bus_read_1(sc->res[1], tp->reg); - if (u != tp->val) { - printf( - "Test %s, step %d: reg(%02x) = %02x", - name, step, tp->reg, u); - printf( "should have been %02x\n", tp->val); - return (1); - } - break; - case xDELAY: - DELAY(tp->val); - break; - default: - printf("Unknown action in test %s, step %d: %d\n", - name, step, tp->action); - return (1); - } - } - if (bootverbose) - printf("Test %s passed\n", name); - return (0); -} - -static int -tnt_probe(device_t dev) -{ - - if (pci_get_vendor(dev) == 0x1093 && pci_get_device(dev) == 0xc801) { - device_set_desc(dev, "NI PCI-GPIB"); - return (BUS_PROBE_DEFAULT); - } - return (ENXIO); -} - -static int -tnt_attach(device_t dev) -{ - struct tnt_softc *sc; - int error, i; - uint8_t version; - - sc = device_get_softc(dev); - - error = bus_alloc_resources(dev, tnt_res_spec, sc->res); - if (error) - return (error); - - error = bus_setup_intr(dev, sc->res[2], INTR_TYPE_MISC | INTR_MPSAFE, - NULL, upd7210intr, &sc->upd7210, &sc->intr_handler); - - /* IO Device Window Base Size Register (IODWBSR) */ - bus_write_4(sc->res[0], 0xc0, rman_get_start(sc->res[1]) | 0x80); - - tst_exec(sc, tst_reset, "Reset"); - tst_exec(sc, tst_read_reg, "Read registers"); - tst_exec(sc, tst_bsr_dcr, "BSR & DCR"); - tst_exec(sc, tst_adr0_1, "ADR0,1"); - tst_exec(sc, tst_cdor_dir, "CDOR/DIR"); - tst_exec(sc, tst_spmr_spsr, "CPMR/SPSR"); - tst_exec(sc, tst_count0_1, "COUNT0:1"); - tst_exec(sc, tst_reset, "Reset"); - - version = bus_read_1(sc->res[1], csr); - version = (version >> 4) & 0x0f; - device_printf(dev, "Chip version 0x%02x (TNT%s)\n", - version, - version >= 4? "5004 or above": "4882"); - if (version >= 4) { - device_printf(dev, "Forcing FIFO mode\n"); - sc->upd7210.use_fifo = 1; - } else { - sc->upd7210.use_fifo = 0; - } - - /* pass 7210 interrupts through */ - bus_write_1(sc->res[1], imr3, 0x02); - - for (i = 0; i < 8; i++) { - sc->upd7210.reg_res[i] = sc->res[1]; - sc->upd7210.reg_offset[i] = i * 2; - } - - /* No DMA help */ - sc->upd7210.dmachan = -1; - - /* No "special interrupt handling" needed here. */ - sc->upd7210.irq_clear_res = NULL; - - upd7210attach(&sc->upd7210); - device_printf(dev, "attached gpib%d\n", sc->upd7210.unit); - - if (sc->upd7210.use_fifo) - bus_write_1(sc->res[0], hssel, 0x01); /* one-chip mode */ - - - return (0); -} - -static int -tnt_detach(device_t dev) -{ - struct tnt_softc *sc; - - sc = device_get_softc(dev); - bus_teardown_intr(dev, sc->res[2], sc->intr_handler); - upd7210detach(&sc->upd7210); - - bus_release_resources(dev, tnt_res_spec, sc->res); - - return (0); -} - -static device_method_t tnt4882_methods[] = { - DEVMETHOD(device_probe, tnt_probe), - DEVMETHOD(device_attach, tnt_attach), - DEVMETHOD(device_detach, tnt_detach), - { 0, 0 } -}; - -static driver_t pci_gpib_driver = { - "tnt4882", - tnt4882_methods, - sizeof(struct tnt_softc) -}; - -static devclass_t pci_gpib_devclass; - -DRIVER_MODULE(pci_gpib, pci, pci_gpib_driver, pci_gpib_devclass, 0, 0); diff --git a/sys/dev/ieee488/tnt4882.h b/sys/dev/ieee488/tnt4882.h deleted file mode 100644 index 221cfd9..0000000 --- a/sys/dev/ieee488/tnt4882.h +++ /dev/null @@ -1,77 +0,0 @@ -/*- - * Copyright (c) 2010 Joerg Wunsch - * All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * $FreeBSD$ - */ - -enum tnt4882reg { - dir = 0x00, - cdor = 0x00, - isr1 = 0x02, - imr1 = 0x02, - isr2 = 0x04, - imr2 = 0x04, - accwr = 0x05, - spsr = 0x06, - spmr = 0x06, - intr = 0x07, - adsr = 0x08, - admr = 0x08, - cnt2 = 0x09, - cptr = 0x0a, - auxmr = 0x0a, - tauxcr = 0x0a, /* 9914 mode register */ - cnt3 = 0x0b, - adr0 = 0x0c, - adr = 0x0c, - hssel = 0x0d, - adr1 = 0x0e, - eosr = 0x0e, - sts1 = 0x10, - cfg = 0x10, - dsr = 0x11, - sh_cnt = 0x11, - imr3 = 0x12, - hier = 0x13, - cnt0 = 0x14, - misc = 0x15, - cnt1 = 0x16, - csr = 0x17, - keyreg = 0x17, - fifob = 0x18, - fifoa = 0x19, - isr3 = 0x1a, - ccr = 0x1a, - sasr = 0x1b, - dcr = 0x1b, - sts2 = 0x1c, - cmdr = 0x1c, - isr0 = 0x1d, - imr0 = 0x1d, - timer = 0x1e, - bsr = 0x1f, - bcr = 0x1f -}; - diff --git a/sys/dev/ieee488/ugpib.h b/sys/dev/ieee488/ugpib.h deleted file mode 100644 index 726a702..0000000 --- a/sys/dev/ieee488/ugpib.h +++ /dev/null @@ -1,155 +0,0 @@ -/*- - * Copyright (c) 2005 Poul-Henning Kamp <phk@FreeBSD.org> - * All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * $FreeBSD$ - * - */ - -#ifndef _DEV_IEEE488_UGPIB_H_ -#define _DEV_IEEE488_UGPIB_H_ - -/* ibfoo() return values */ -#define EDVR 0 /* System error */ -#define ECIC 1 /* Not Active Controller */ -#define ENOL 2 /* Nobody listening */ -#define EADR 3 /* Controller not addressed */ -#define EARG 4 /* Invalid argument */ -#define ESAC 5 /* Not System Controller */ -#define EABO 6 /* I/O Aborted/Time out */ -#define ENEB 7 /* No such controller */ -#define EOIP 10 /* Async I/O in progress */ -#define ECAP 11 /* No such capability */ -#define EFSO 12 /* File system error */ -#define EBUS 14 /* Command byte xfer error */ -#define ESTB 15 /* Serial poll status byte lost */ -#define ESRQ 16 /* SRQ line stuck */ -#define ETAB 20 /* Table problem */ - -/* ibsta bits */ -#define ERR (1<<15) /* Error */ -#define TIMO (1<<14) /* Timeout */ -#define END (1<<13) /* EOI/EOS */ -#define SRQI (1<<12) /* SRQ */ -#define RQS (1<<11) /* Device requests service */ -#define SPOLL (1<<10) /* Serial Poll */ -#define EVENT (1<<9) /* Event occured */ -#define CMPL (1<<8) /* I/O complete */ -#define LOK (1<<7) /* Lockout */ -#define REM (1<<6) /* Remote */ -#define CIC (1<<5) /* CIC */ -#define ATN (1<<4) /* ATN */ -#define TACS (1<<3) /* Talker */ -#define LACS (1<<2) /* Listener */ -#define DTAS (1<<1) /* Device trigger status */ -#define DCAS (1<<0) /* Device clear state */ - -/* Timeouts */ -#define TNONE 0 -#define T10us 1 -#define T30us 2 -#define T100us 3 -#define T300us 4 -#define T1ms 5 -#define T3ms 6 -#define T10ms 7 -#define T30ms 8 -#define T100ms 9 -#define T300ms 10 -#define T1s 11 -#define T3s 12 -#define T10s 13 -#define T30s 14 -#define T100s 15 -#define T300s 16 -#define T1000s 17 - -/* EOS bits */ -#define REOS (1 << 10) -#define XEOS (1 << 11) -#define BIN (1 << 12) - -/* Bus commands */ -#define GTL 0x01 /* Go To Local */ -#define SDC 0x04 /* Selected Device Clear */ -#define GET 0x08 /* Group Execute Trigger */ -#define LAD 0x20 /* Listen address */ -#define UNL 0x3F /* Unlisten */ -#define TAD 0x40 /* Talk address */ -#define UNT 0x5F /* Untalk */ - -#ifndef _KERNEL - -extern int ibcnt, iberr, ibsta; - -int ibask(int handle, int option, int *retval); -int ibbna(int handle, char *bdname); -int ibcac(int handle, int v); -int ibclr(int handle); -int ibcmd(int handle, void *buffer, long cnt); -int ibcmda(int handle, void *buffer, long cnt); -int ibconfig(int handle, int option, int value); -int ibdev(int boardID, int pad, int sad, int tmo, int eot, int eos); -int ibdiag(int handle, void *buffer, long cnt); -int ibdma(int handle, int v); -int ibeos(int handle, int eos); -int ibeot(int handle, int eot); -int ibevent(int handle, short *event); -int ibfind(char *bdname); -int ibgts(int handle, int v); -int ibist(int handle, int v); -int iblines(int handle, short *lines); -int ibllo(int handle); -int ibln(int handle, int padval, int sadval, short *listenflag); -int ibloc(int handle); -int ibonl(int handle, int v); -int ibpad(int handle, int pad); -int ibpct(int handle); -int ibpoke(int handle, int option, int value); -int ibppc(int handle, int v); -int ibrd(int handle, void *buffer, long cnt); -int ibrda(int handle, void *buffer, long cnt); -int ibrdf(int handle, char *flname); -int ibrdkey(int handle, void *buffer, int cnt); -int ibrpp(int handle, char *ppr); -int ibrsc(int handle, int v); -int ibrsp(int handle, char *spr); -int ibrsv(int handle, int v); -int ibsad(int handle, int sad); -int ibsgnl(int handle, int v); -int ibsic(int handle); -int ibsre(int handle, int v); -int ibsrq(void (*func)(void)); -int ibstop(int handle); -int ibtmo(int handle, int tmo); -int ibtrap(int mask, int mode); -int ibtrg(int handle); -int ibwait(int handle, int mask); -int ibwrt(int handle, const void *buffer, long cnt); -int ibwrta(int handle, const void *buffer, long cnt); -int ibwrtf(int handle, const char *flname); -int ibwrtkey(int handle, const void *buffer, int cnt); -int ibxtrc(int handle, void *buffer, long cnt); -#endif /* _KERNEL */ -#endif /* _DEV_IEEE488_UGPIB_H_ */ diff --git a/sys/dev/ieee488/upd7210.c b/sys/dev/ieee488/upd7210.c deleted file mode 100644 index 0e76c83..0000000 --- a/sys/dev/ieee488/upd7210.c +++ /dev/null @@ -1,369 +0,0 @@ -/*- - * Copyright (c) 2005 Poul-Henning Kamp <phk@FreeBSD.org> - * Copyright (c) 2010 Joerg Wunsch <joerg@FreeBSD.org> - * All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * High-level driver for µPD7210 based GPIB cards. - * - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -# define GPIB_DEBUG -# undef GPIB_DEBUG - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/conf.h> -#include <sys/malloc.h> -#include <sys/kernel.h> -#include <sys/limits.h> -#include <sys/module.h> -#include <sys/rman.h> -#include <sys/bus.h> -#include <sys/lock.h> -#include <sys/mutex.h> -#include <sys/uio.h> -#include <sys/time.h> -#include <machine/bus.h> -#include <machine/resource.h> -#include <isa/isavar.h> - -#define UPD7210_HW_DRIVER -#define UPD7210_SW_DRIVER -#include <dev/ieee488/upd7210.h> -#include <dev/ieee488/tnt4882.h> - -static MALLOC_DEFINE(M_GPIB, "GPIB", "GPIB"); - -/* upd7210 generic stuff */ - -void -upd7210_print_isr(u_int isr1, u_int isr2) -{ - printf("isr1=0x%b isr2=0x%b", - isr1, "\20\10CPT\7APT\6DET\5ENDRX\4DEC\3ERR\2DO\1DI", - isr2, "\20\10INT\7SRQI\6LOK\5REM\4CO\3LOKC\2REMC\1ADSC"); -} - -u_int -upd7210_rd(struct upd7210 *u, enum upd7210_rreg reg) -{ - u_int r; - - r = bus_read_1(u->reg_res[reg], u->reg_offset[reg]); - u->rreg[reg] = r; - return (r); -} - -void -upd7210_wr(struct upd7210 *u, enum upd7210_wreg reg, u_int val) -{ - - bus_write_1(u->reg_res[reg], u->reg_offset[reg], val); - u->wreg[reg] = val; - if (reg == AUXMR) - u->wreg[8 + (val >> 5)] = val & 0x1f; -} - -void -upd7210intr(void *arg) -{ - u_int isr_1, isr_2, isr_3; - struct upd7210 *u; - - u = arg; - mtx_lock(&u->mutex); - isr_1 = upd7210_rd(u, ISR1); - isr_2 = upd7210_rd(u, ISR2); - if (u->use_fifo) { - isr_3 = bus_read_1(u->reg_res[0], isr3); - } else { - isr_3 = 0; - } - if (isr_1 != 0 || isr_2 != 0 || isr_3 != 0) { - if (u->busy == 0 || u->irq == NULL || !u->irq(u, isr_3)) { -#if 0 - printf("upd7210intr [%02x %02x %02x", - upd7210_rd(u, DIR), isr1, isr2); - printf(" %02x %02x %02x %02x %02x] ", - upd7210_rd(u, SPSR), - upd7210_rd(u, ADSR), - upd7210_rd(u, CPTR), - upd7210_rd(u, ADR0), - upd7210_rd(u, ADR1)); - upd7210_print_isr(isr1, isr2); - printf("\n"); -#endif - } - /* - * "special interrupt handling" - * - * In order to implement shared IRQs, the original - * PCIIa uses IO locations 0x2f0 + (IRQ#) as an output - * location. If an ISR for a particular card has - * detected this card triggered the IRQ, it must reset - * the card's IRQ by writing (anything) to that IO - * location. - * - * Some clones apparently don't implement this - * feature, but National Instrument cards do. - */ - if (u->irq_clear_res != NULL) - bus_write_1(u->irq_clear_res, 0, 42); - } - mtx_unlock(&u->mutex); -} - -int -upd7210_take_ctrl_async(struct upd7210 *u) -{ - int i; - - upd7210_wr(u, AUXMR, AUXMR_TCA); - - if (!(upd7210_rd(u, ADSR) & ADSR_ATN)) - return (0); - for (i = 0; i < 20; i++) { - DELAY(1); - if (!(upd7210_rd(u, ADSR) & ADSR_ATN)) - return (0); - } - return (1); -} - -int -upd7210_goto_standby(struct upd7210 *u) -{ - int i; - - upd7210_wr(u, AUXMR, AUXMR_GTS); - - if (upd7210_rd(u, ADSR) & ADSR_ATN) - return (0); - for (i = 0; i < 20; i++) { - DELAY(1); - if (upd7210_rd(u, ADSR) & ADSR_ATN) - return (0); - } - return (1); -} - -/* Unaddressed Listen Only mode */ - -static int -gpib_l_irq(struct upd7210 *u, int isr_3) -{ - int i; - int have_data = 0; - - if (u->use_fifo) { - /* TNT5004 or TNT4882 in FIFO mode */ - if (isr_3 & 0x04) { - /* FIFO not empty */ - i = bus_read_1(u->reg_res[0], fifob); - have_data = 1; - bus_write_1(u->reg_res[0], cnt0, -1); - bus_write_1(u->reg_res[0], cnt1, (-1) >> 8); - bus_write_1(u->reg_res[0], cnt2, (-1) >> 16); - bus_write_1(u->reg_res[0], cnt3, (-1) >> 24); - bus_write_1(u->reg_res[0], cmdr, 0x04); /* GO */ - } - } else if (u->rreg[ISR1] & 1) { - i = upd7210_rd(u, DIR); - have_data = 1; - } - - if (have_data) { - u->buf[u->buf_wp++] = i; - u->buf_wp &= (u->bufsize - 1); - i = (u->buf_rp + u->bufsize - u->buf_wp) & (u->bufsize - 1); - if (i < 8) { - if (u->use_fifo) - bus_write_1(u->reg_res[0], imr3, 0x00); - else - upd7210_wr(u, IMR1, 0); - } - wakeup(u->buf); - return (1); - } - return (0); -} - -static int -gpib_l_open(struct cdev *dev, int oflags, int devtype, struct thread *td) -{ - struct upd7210 *u; - - u = dev->si_drv1; - - mtx_lock(&u->mutex); - if (u->busy) { - mtx_unlock(&u->mutex); - return (EBUSY); - } - u->busy = 1; - u->irq = gpib_l_irq; - mtx_unlock(&u->mutex); - - u->buf = malloc(PAGE_SIZE, M_GPIB, M_WAITOK); - u->bufsize = PAGE_SIZE; - u->buf_wp = 0; - u->buf_rp = 0; - - upd7210_wr(u, AUXMR, AUXMR_CRST); /* chip reset */ - DELAY(10000); - upd7210_wr(u, AUXMR, C_ICR | 8); /* 8 MHz clock */ - DELAY(1000); - upd7210_wr(u, ADR, 0x60); /* ADR0: disable listener and talker 0 */ - upd7210_wr(u, ADR, 0xe0); /* ADR1: disable listener and talker 1 */ - upd7210_wr(u, ADMR, 0x70); /* listen-only (lon) */ - upd7210_wr(u, AUXMR, AUXMR_PON); /* immediate execute power-on (pon) */ - if (u->use_fifo) { - /* TNT5004 or TNT4882 in FIFO mode */ - bus_write_1(u->reg_res[0], cmdr, 0x10); /* reset FIFO */ - bus_write_1(u->reg_res[0], cfg, 0x20); /* xfer IN, 8-bit FIFO */ - bus_write_1(u->reg_res[0], cnt0, -1); - bus_write_1(u->reg_res[0], cnt1, (-1) >> 8); - bus_write_1(u->reg_res[0], cnt2, (-1) >> 16); - bus_write_1(u->reg_res[0], cnt3, (-1) >> 24); - bus_write_1(u->reg_res[0], cmdr, 0x04); /* GO */ - bus_write_1(u->reg_res[0], imr3, 0x04); /* NEF IE */ - } else { - /* µPD7210/NAT7210, or TNT4882 in non-FIFO mode */ - upd7210_wr(u, IMR1, 0x01); /* data in interrupt enable */ - } - return (0); -} - -static int -gpib_l_close(struct cdev *dev, int oflags, int devtype, struct thread *td) -{ - struct upd7210 *u; - - u = dev->si_drv1; - - mtx_lock(&u->mutex); - u->busy = 0; - if (u->use_fifo) { - /* TNT5004 or TNT4882 in FIFO mode */ - bus_write_1(u->reg_res[0], cmdr, 0x22); /* soft RESET */ - bus_write_1(u->reg_res[0], imr3, 0x00); - } - upd7210_wr(u, AUXMR, AUXMR_CRST); - DELAY(10000); - upd7210_wr(u, IMR1, 0x00); - upd7210_wr(u, IMR2, 0x00); - free(u->buf, M_GPIB); - u->buf = NULL; - mtx_unlock(&u->mutex); - return (0); -} - -static int -gpib_l_read(struct cdev *dev, struct uio *uio, int ioflag) -{ - struct upd7210 *u; - int error; - size_t z; - - u = dev->si_drv1; - error = 0; - - mtx_lock(&u->mutex); - while (u->buf_wp == u->buf_rp) { - error = msleep(u->buf, &u->mutex, PZERO | PCATCH, - "gpibrd", hz); - if (error && error != EWOULDBLOCK) { - mtx_unlock(&u->mutex); - return (error); - } - } - while (uio->uio_resid > 0 && u->buf_wp != u->buf_rp) { - if (u->buf_wp < u->buf_rp) - z = u->bufsize - u->buf_rp; - else - z = u->buf_wp - u->buf_rp; - if (z > uio->uio_resid) - z = uio->uio_resid; - mtx_unlock(&u->mutex); - error = uiomove(u->buf + u->buf_rp, z, uio); - mtx_lock(&u->mutex); - if (error) - break; - u->buf_rp += z; - u->buf_rp &= (u->bufsize - 1); - } - if (u->use_fifo) { - bus_write_1(u->reg_res[0], imr3, 0x04); /* NFF IE */ - } else { - if (u->wreg[IMR1] == 0) - upd7210_wr(u, IMR1, 0x01); - } - mtx_unlock(&u->mutex); - return (error); -} - -static struct cdevsw gpib_l_cdevsw = { - .d_version = D_VERSION, - .d_name = "gpib_l", - .d_open = gpib_l_open, - .d_close = gpib_l_close, - .d_read = gpib_l_read, -}; - -/* Housekeeping */ - -static struct unrhdr *units; - -void -upd7210attach(struct upd7210 *u) -{ - struct cdev *dev; - - if (units == NULL) - units = new_unrhdr(0, INT_MAX, NULL); - u->unit = alloc_unr(units); - mtx_init(&u->mutex, "gpib", NULL, MTX_DEF); - u->cdev = make_dev(&gpib_l_cdevsw, u->unit, - UID_ROOT, GID_WHEEL, 0444, - "gpib%ul", u->unit); - u->cdev->si_drv1 = u; - - dev = make_dev(&gpib_ib_cdevsw, u->unit, - UID_ROOT, GID_WHEEL, 0444, - "gpib%uib", u->unit); - dev->si_drv1 = u; - dev_depends(u->cdev, dev); -} - -void -upd7210detach(struct upd7210 *u) -{ - - destroy_dev(u->cdev); - mtx_destroy(&u->mutex); - free_unr(units, u->unit); -} diff --git a/sys/dev/ieee488/upd7210.h b/sys/dev/ieee488/upd7210.h deleted file mode 100644 index 27e3a7a..0000000 --- a/sys/dev/ieee488/upd7210.h +++ /dev/null @@ -1,237 +0,0 @@ -/*- - * Copyright (c) 2005 Poul-Henning Kamp <phk@FreeBSD.org> - * Copyright (c) 2010 Joerg Wunsch <joerg@FreeBSD.org> - * All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * $FreeBSD$ - * - * Locating an actual µPD7210 data book has proven quite impossible for me. - * There are a fair number of newer chips which are supersets of the µPD7210 - * but they are particular eager to comprehensively mark what the extensions - * are and what is in the base set. Some even give the registers and their - * bits new names. - * - * The following information is based on a description of the µPD7210 found - * in an old manual for a VME board which used the chip. - */ - -#ifndef _DEV_IEEE488_UPD7210_H_ -#define _DEV_IEEE488_UPD7210_H_ -#ifdef _KERNEL - -struct upd7210; -struct ibfoo; - -/* upd7210 interface definitions for HW drivers */ - -typedef int upd7210_irq_t(struct upd7210 *, int); - -struct upd7210 { - struct resource *reg_res[8]; - struct resource *irq_clear_res; - u_int reg_offset[8]; - int dmachan; - int unit; - int use_fifo; - - /* private stuff */ - struct mtx mutex; - uint8_t rreg[8]; - uint8_t wreg[8 + 8]; - - upd7210_irq_t *irq; - - int busy; - u_char *buf; - size_t bufsize; - u_int buf_wp; - u_int buf_rp; - struct cdev *cdev; - - struct ibfoo *ibfoo; -}; - -#ifdef UPD7210_HW_DRIVER -void upd7210intr(void *); -void upd7210attach(struct upd7210 *); -void upd7210detach(struct upd7210 *); -#endif - -#ifdef UPD7210_SW_DRIVER - -/* upd7210 hardware definitions. */ - -/* Write registers */ -enum upd7210_wreg { - CDOR = 0, /* Command/Data Out Register */ - IMR1 = 1, /* Interrupt Mask Register 1 */ - IMR2 = 2, /* Interrupt Mask Register 2 */ - SPMR = 3, /* Serial Poll Mode Register */ - ADMR = 4, /* ADdress Mode Register */ - AUXMR = 5, /* AUXilliary Mode Register */ - ICR = 5, /* Internal Counter Register */ - PPR = 5, /* Parallel Poll Register */ - AUXRA = 5, /* AUXilliary Register A */ - AUXRB = 5, /* AUXilliary Register B */ - AUXRE = 5, /* AUXilliary Register E */ - ADR = 6, /* ADdress Register */ - EOSR = 7, /* End-Of-String Register */ -}; - -/* Read registers */ -enum upd7210_rreg { - DIR = 0, /* Data In Register */ - ISR1 = 1, /* Interrupt Status Register 1 */ - ISR2 = 2, /* Interrupt Status Register 2 */ - SPSR = 3, /* Serial Poll Status Register */ - ADSR = 4, /* ADdress Status Register */ - CPTR = 5, /* Command Pass Though Register */ - ADR0 = 6, /* ADdress Register 0 */ - ADR1 = 7, /* ADdress Register 1 */ -}; - -/* Bits for ISR1 and IMR1 */ -#define IXR1_DI (1 << 0) /* Data In */ -#define IXR1_DO (1 << 1) /* Data Out */ -#define IXR1_ERR (1 << 2) /* Error */ -#define IXR1_DEC (1 << 3) /* Device Clear */ -#define IXR1_ENDRX (1 << 4) /* End Received */ -#define IXR1_DET (1 << 5) /* Device Execute Trigger */ -#define IXR1_APT (1 << 6) /* Address Pass-Through */ -#define IXR1_CPT (1 << 7) /* Command Pass-Through */ - -/* Bits for ISR2 and IMR2 */ -#define IXR2_ADSC (1 << 0) /* Addressed Status Change */ -#define IXR2_REMC (1 << 1) /* Remote Change */ -#define IXR2_LOKC (1 << 2) /* Lockout Change */ -#define IXR2_CO (1 << 3) /* Command Out */ -#define ISR2_REM (1 << 4) /* Remove */ -#define IMR2_DMAI (1 << 4) /* DMA In Enable */ -#define ISR2_LOK (1 << 5) /* Lockout */ -#define IMR2_DMAO (1 << 5) /* DMA Out Enable */ -#define IXR2_SRQI (1 << 6) /* Service Request Input */ -#define ISR2_INT (1 << 7) /* Interrupt */ - -#define SPSR_PEND (1 << 6) /* Pending */ -#define SPMR_RSV (1 << 6) /* Request SerVice */ - -#define ADSR_MJMN (1 << 0) /* MaJor MiNor */ -#define ADSR_TA (1 << 1) /* Talker Active */ -#define ADSR_LA (1 << 2) /* Listener Active */ -#define ADSR_TPAS (1 << 3) /* Talker Primary Addr. State */ -#define ADSR_LPAS (1 << 4) /* Listener Primary Addr. State */ -#define ADSR_SPMS (1 << 5) /* Serial Poll Mode State */ -#define ADSR_ATN (1 << 6) /* Attention */ -#define ADSR_CIC (1 << 7) /* Controller In Charge */ - -#define ADMR_ADM0 (1 << 0) /* Address Mode 0 */ -#define ADMR_ADM1 (1 << 1) /* Address Mode 1 */ -#define ADMR_TRM0 (1 << 4) /* Transmit/Receive Mode 0 */ -#define ADMR_TRM1 (1 << 5) /* Transmit/Receive Mode 1 */ -#define ADMR_LON (1 << 6) /* Listen Only */ -#define ADMR_TON (1 << 7) /* Talk Only */ - -/* Constant part of overloaded write registers */ -#define C_ICR 0x20 -#define C_PPR 0x60 -#define C_AUXA 0x80 -#define C_AUXB 0xa0 -#define C_AUXE 0xc0 - -#define AUXMR_PON 0x00 /* Immediate Execute pon */ -#define AUXMR_CPP 0x01 /* Clear Parallel Poll */ -#define AUXMR_CRST 0x02 /* Chip Reset */ -#define AUXMR_RFD 0x03 /* Finish Handshake */ -#define AUXMR_TRIG 0x04 /* Trigger */ -#define AUXMR_RTL 0x05 /* Return to local */ -#define AUXMR_SEOI 0x06 /* Send EOI */ -#define AUXMR_NVSA 0x07 /* Non-Valid Secondary cmd/addr */ - /* 0x08 undefined/unknown */ -#define AUXMR_SPP 0x09 /* Set Parallel Poll */ - /* 0x0a undefined/unknown */ - /* 0x0b undefined/unknown */ - /* 0x0c undefined/unknown */ - /* 0x0d undefined/unknown */ - /* 0x0e undefined/unknown */ -#define AUXMR_VSA 0x0f /* Valid Secondary cmd/addr */ -#define AUXMR_GTS 0x10 /* Go to Standby */ -#define AUXMR_TCA 0x11 /* Take Control Async (pulsed) */ -#define AUXMR_TCS 0x12 /* Take Control Synchronously */ -#define AUXMR_LISTEN 0x13 /* Listen */ -#define AUXMR_DSC 0x14 /* Disable System Control */ - /* 0x15 undefined/unknown */ -#define AUXMR_SIFC 0x16 /* Set IFC */ -#define AUXMR_CREN 0x17 /* Clear REN */ - /* 0x18 undefined/unknown */ - /* 0x19 undefined/unknown */ -#define AUXMR_TCSE 0x1a /* Take Control Sync on End */ -#define AUXMR_LCM 0x1b /* Listen Continuously Mode */ -#define AUXMR_LUNL 0x1c /* Local Unlisten */ -#define AUXMR_EPP 0x1d /* Execute Parallel Poll */ -#define AUXMR_CIFC 0x1e /* Clear IFC */ -#define AUXMR_SREN 0x1f /* Set REN */ - -#define PPR_U (1 << 4) /* Unconfigure */ -#define PPR_S (1 << 3) /* Status Polarity */ - -#define AUXA_HLDA (1 << 0) /* Holdoff on All */ -#define AUXA_HLDE (1 << 1) /* Holdoff on END */ -#define AUXA_REOS (1 << 2) /* End on EOS received */ -#define AUXA_XEOS (1 << 3) /* Transmit END with EOS */ -#define AUXA_BIN (1 << 4) /* Binary */ - -#define AUXB_CPTE (1 << 0) /* Cmd Pass Through Enable */ -#define AUXB_SPEOI (1 << 1) /* Send Serial Poll EOI */ -#define AUXB_TRI (1 << 2) /* Three-State Timing */ -#define AUXB_INV (1 << 3) /* Invert */ -#define AUXB_ISS (1 << 4) /* Individual Status Select */ - -#define AUXE_DHDT (1 << 0) /* DAC Holdoff on DTAS */ -#define AUXE_DHDC (1 << 1) /* DAC Holdoff on DCAS */ - -#define ADR0_DL0 (1 << 5) /* Disable Listener 0 */ -#define ADR0_DT0 (1 << 6) /* Disable Talker 0 */ - -#define ADR_DL (1 << 5) /* Disable Listener */ -#define ADR_DT (1 << 6) /* Disable Talker */ -#define ADR_ARS (1 << 7) /* Address Register Select */ - -#define ADR1_DL1 (1 << 5) /* Disable Listener 1 */ -#define ADR1_DT1 (1 << 6) /* Disable Talker 1 */ -#define ADR1_EOI (1 << 7) /* End or Identify */ - -/* Stuff from software drivers */ -extern struct cdevsw gpib_ib_cdevsw; - -/* Stuff from upd7210.c */ -void upd7210_print_isr(u_int isr1, u_int isr2); -u_int upd7210_rd(struct upd7210 *u, enum upd7210_rreg reg); -void upd7210_wr(struct upd7210 *u, enum upd7210_wreg reg, u_int val); -int upd7210_take_ctrl_async(struct upd7210 *u); -int upd7210_goto_standby(struct upd7210 *u); - -#endif /* UPD7210_SW_DRIVER */ - -#endif /* _KERNEL */ -#endif /* _DEV_IEEE488_UPD7210_H_ */ diff --git a/sys/dev/ipmi/ipmi_kcs.c b/sys/dev/ipmi/ipmi_kcs.c index 76adf8c..eb5884a 100644 --- a/sys/dev/ipmi/ipmi_kcs.c +++ b/sys/dev/ipmi/ipmi_kcs.c @@ -184,6 +184,8 @@ kcs_start_write(struct ipmi_softc *sc) for (retry = 0; retry < 10; retry++) { /* Wait for IBF = 0 */ status = kcs_wait_for_ibf(sc, 0); + if (status & KCS_STATUS_IBF) + return (0); /* Clear OBF */ kcs_clear_obf(sc, status); @@ -193,6 +195,9 @@ kcs_start_write(struct ipmi_softc *sc) /* Wait for IBF = 0 */ status = kcs_wait_for_ibf(sc, 0); + if (status & KCS_STATUS_IBF) + return (0); + if (KCS_STATUS_STATE(status) == KCS_STATUS_STATE_WRITE) break; DELAY(1000000); @@ -222,6 +227,8 @@ kcs_write_byte(struct ipmi_softc *sc, u_char data) /* Wait for IBF = 0 */ status = kcs_wait_for_ibf(sc, 0); + if (status & KCS_STATUS_IBF) + return (0); if (KCS_STATUS_STATE(status) != KCS_STATUS_STATE_WRITE) return (0); @@ -244,6 +251,8 @@ kcs_write_last_byte(struct ipmi_softc *sc, u_char data) /* Wait for IBF = 0 */ status = kcs_wait_for_ibf(sc, 0); + if (status & KCS_STATUS_IBF) + return (0); if (KCS_STATUS_STATE(status) != KCS_STATUS_STATE_WRITE) /* error state */ @@ -274,6 +283,8 @@ kcs_read_byte(struct ipmi_softc *sc, u_char *data) /* Wait for OBF = 1 */ status = kcs_wait_for_obf(sc, 1); + if ((status & KCS_STATUS_OBF) == 0) + return (0); /* Read Data_out */ *data = INB(sc, KCS_DATA); @@ -288,6 +299,8 @@ kcs_read_byte(struct ipmi_softc *sc, u_char *data) /* Wait for OBF = 1*/ status = kcs_wait_for_obf(sc, 1); + if ((status & KCS_STATUS_OBF) == 0) + return (0); /* Read Dummy */ dummy = INB(sc, KCS_DATA); diff --git a/sys/dev/isci/isci_controller.c b/sys/dev/isci/isci_controller.c index f3ff082..b0f4285 100644 --- a/sys/dev/isci/isci_controller.c +++ b/sys/dev/isci/isci_controller.c @@ -373,6 +373,8 @@ SCI_STATUS isci_controller_initialize(struct ISCI_CONTROLLER *controller) fail_on_timeout = 1; TUNABLE_INT_FETCH("hw.isci.fail_on_task_timeout", &fail_on_timeout); + controller->fail_on_task_timeout = fail_on_timeout; + /* Attach to CAM using xpt_bus_register now, then immediately freeze * the simq. It will get released later when initial domain discovery * is complete. diff --git a/sys/dev/isci/isci_sysctl.c b/sys/dev/isci/isci_sysctl.c index 62a10b9..a7c56c6 100644 --- a/sys/dev/isci/isci_sysctl.c +++ b/sys/dev/isci/isci_sysctl.c @@ -226,12 +226,13 @@ static int isci_sysctl_fail_on_task_timeout(SYSCTL_HANDLER_ARGS) { struct isci_softc *isci = (struct isci_softc *)arg1; - int32_t fail_on_timeout = 0; + int32_t fail_on_timeout; int error, i; + fail_on_timeout = isci->controllers[0].fail_on_task_timeout; error = sysctl_handle_int(oidp, &fail_on_timeout, 0, req); - if (error || fail_on_timeout == 0) + if (error || req->newptr == NULL) return (error); for (i = 0; i < isci->controller_count; i++) diff --git a/sys/dev/iscsi/iscsi.c b/sys/dev/iscsi/iscsi.c index f9edc82..0d762aa 100644 --- a/sys/dev/iscsi/iscsi.c +++ b/sys/dev/iscsi/iscsi.c @@ -2157,6 +2157,10 @@ iscsi_action_scsiio(struct iscsi_session *is, union ccb *ccb) ISCSI_SESSION_DEBUG(is, "len %zd -> %zd", len, is->is_first_burst_length); len = is->is_first_burst_length; } + if (len > is->is_max_data_segment_length) { + ISCSI_SESSION_DEBUG(is, "len %zd -> %zd", len, is->is_max_data_segment_length); + len = is->is_max_data_segment_length; + } error = icl_pdu_append_data(request, csio->data_ptr, len, M_NOWAIT); if (error != 0) { diff --git a/sys/dev/iwn/if_iwn.c b/sys/dev/iwn/if_iwn.c index bf8c98d..b287c6c 100644 --- a/sys/dev/iwn/if_iwn.c +++ b/sys/dev/iwn/if_iwn.c @@ -279,7 +279,6 @@ static int iwn_send_btcoex(struct iwn_softc *); static int iwn_send_advanced_btcoex(struct iwn_softc *); static int iwn5000_runtime_calib(struct iwn_softc *); static int iwn_config(struct iwn_softc *); -static uint8_t *ieee80211_add_ssid(uint8_t *, const uint8_t *, u_int); static int iwn_scan(struct iwn_softc *, struct ieee80211vap *, struct ieee80211_scan_state *, struct ieee80211_channel *); static int iwn_auth(struct iwn_softc *, struct ieee80211vap *vap); @@ -6527,18 +6526,6 @@ iwn_config(struct iwn_softc *sc) return 0; } -/* - * Add an ssid element to a frame. - */ -static uint8_t * -ieee80211_add_ssid(uint8_t *frm, const uint8_t *ssid, u_int len) -{ - *frm++ = IEEE80211_ELEMID_SSID; - *frm++ = len; - memcpy(frm, ssid, len); - return frm + len; -} - static uint16_t iwn_get_active_dwell_time(struct iwn_softc *sc, struct ieee80211_channel *c, uint8_t n_probes) diff --git a/sys/dev/mmc/mmc.c b/sys/dev/mmc/mmc.c index a3b9744..ec4d358 100644 --- a/sys/dev/mmc/mmc.c +++ b/sys/dev/mmc/mmc.c @@ -63,6 +63,7 @@ __FBSDID("$FreeBSD$"); #include <sys/bus.h> #include <sys/endian.h> #include <sys/sysctl.h> +#include <sys/time.h> #include <dev/mmc/mmcreg.h> #include <dev/mmc/mmcbrvar.h> @@ -76,8 +77,13 @@ struct mmc_softc { struct intr_config_hook config_intrhook; device_t owner; uint32_t last_rca; + int squelched; /* suppress reporting of (expected) errors */ + int log_count; + struct timeval log_time; }; +#define LOG_PPS 5 /* Log no more than 5 errors per second. */ + /* * Per-card data */ @@ -426,6 +432,13 @@ mmc_wait_for_cmd(struct mmc_softc *sc, struct mmc_command *cmd, int retries) err = cmd->error; } while (err != MMC_ERR_NONE && retries-- > 0); + if (err != MMC_ERR_NONE && sc->squelched == 0) { + if (ppsratecheck(&sc->log_time, &sc->log_count, LOG_PPS)) { + device_printf(sc->dev, "CMD%d failed, RESULT: %d\n", + cmd->opcode, err); + } + } + return (err); } @@ -436,6 +449,8 @@ mmc_wait_for_app_cmd(struct mmc_softc *sc, uint32_t rca, struct mmc_command appcmd; int err; + /* Squelch error reporting at lower levels, we report below. */ + sc->squelched++; do { memset(&appcmd, 0, sizeof(appcmd)); appcmd.opcode = MMC_APP_CMD; @@ -455,6 +470,14 @@ mmc_wait_for_app_cmd(struct mmc_softc *sc, uint32_t rca, err = cmd->error; } } while (err != MMC_ERR_NONE && retries-- > 0); + sc->squelched--; + + if (err != MMC_ERR_NONE && sc->squelched == 0) { + if (ppsratecheck(&sc->log_time, &sc->log_count, LOG_PPS)) { + device_printf(sc->dev, "ACMD%d failed, RESULT: %d\n", + cmd->opcode, err); + } + } return (err); } @@ -760,6 +783,7 @@ mmc_test_bus_width(struct mmc_softc *sc) mmcbr_set_bus_width(sc->dev, bus_width_8); mmcbr_update_ios(sc->dev); + sc->squelched++; /* Errors are expected, squelch reporting. */ memset(&cmd, 0, sizeof(cmd)); memset(&data, 0, sizeof(data)); cmd.opcode = MMC_BUSTEST_W; @@ -783,6 +807,7 @@ mmc_test_bus_width(struct mmc_softc *sc) data.len = 8; data.flags = MMC_DATA_READ; err = mmc_wait_for_cmd(sc, &cmd, 0); + sc->squelched--; mmcbr_set_bus_width(sc->dev, bus_width_1); mmcbr_update_ios(sc->dev); @@ -795,6 +820,7 @@ mmc_test_bus_width(struct mmc_softc *sc) mmcbr_set_bus_width(sc->dev, bus_width_4); mmcbr_update_ios(sc->dev); + sc->squelched++; /* Errors are expected, squelch reporting. */ memset(&cmd, 0, sizeof(cmd)); memset(&data, 0, sizeof(data)); cmd.opcode = MMC_BUSTEST_W; @@ -818,6 +844,7 @@ mmc_test_bus_width(struct mmc_softc *sc) data.len = 4; data.flags = MMC_DATA_READ; err = mmc_wait_for_cmd(sc, &cmd, 0); + sc->squelched--; mmcbr_set_bus_width(sc->dev, bus_width_1); mmcbr_update_ios(sc->dev); @@ -1270,7 +1297,9 @@ mmc_discover_cards(struct mmc_softc *sc) if (bootverbose || mmc_debug) device_printf(sc->dev, "Probing cards\n"); while (1) { + sc->squelched++; /* Errors are expected, squelch reporting. */ err = mmc_all_send_cid(sc, raw_cid); + sc->squelched--; if (err == MMC_ERR_TIMEOUT) break; if (err != MMC_ERR_NONE) { @@ -1417,10 +1446,10 @@ mmc_discover_cards(struct mmc_softc *sc) break; } + mmc_select_card(sc, ivar->rca); + /* Only MMC >= 4.x cards support EXT_CSD. */ if (ivar->csd.spec_vers >= 4) { - /* Card must be selected to fetch EXT_CSD. */ - mmc_select_card(sc, ivar->rca); mmc_send_ext_csd(sc, ivar->raw_ext_csd); /* Handle extended capacity from EXT_CSD */ sec_count = ivar->raw_ext_csd[EXT_CSD_SEC_CNT] + @@ -1450,7 +1479,6 @@ mmc_discover_cards(struct mmc_softc *sc) mmc_switch(sc, EXT_CSD_CMD_SET_NORMAL, EXT_CSD_ERASE_GRP_DEF, 1); } - mmc_select_card(sc, 0); } else { ivar->bus_width = bus_width_1; ivar->timing = bus_timing_normal; @@ -1477,6 +1505,7 @@ mmc_discover_cards(struct mmc_softc *sc) child = device_add_child(sc->dev, NULL, -1); device_set_ivars(child, ivar); } + mmc_select_card(sc, 0); } } @@ -1536,6 +1565,7 @@ mmc_go_discovery(struct mmc_softc *sc) /* * First, try SD modes */ + sc->squelched++; /* Errors are expected, squelch reporting. */ mmcbr_set_mode(dev, mode_sd); mmc_power_up(sc); mmcbr_set_bus_mode(dev, pushpull); @@ -1561,6 +1591,7 @@ mmc_go_discovery(struct mmc_softc *sc) "MMC probe: OK (OCR: 0x%08x)\n", ocr); } else if (bootverbose || mmc_debug) device_printf(sc->dev, "SD probe: OK (OCR: 0x%08x)\n", ocr); + sc->squelched--; mmcbr_set_ocr(dev, mmc_select_vdd(sc, ocr)); if (mmcbr_get_ocr(dev) != 0) diff --git a/sys/dev/ofw/ofw_cpu.c b/sys/dev/ofw/ofw_cpu.c index 2265140..1d01b63 100644 --- a/sys/dev/ofw/ofw_cpu.c +++ b/sys/dev/ofw/ofw_cpu.c @@ -171,7 +171,7 @@ ofw_cpu_probe(device_t dev) { const char *type = ofw_bus_get_type(dev); - if (strcmp(type, "cpu") != 0) + if (type == NULL || strcmp(type, "cpu") != 0) return (ENXIO); device_set_desc(dev, "Open Firmware CPU"); @@ -182,12 +182,20 @@ static int ofw_cpu_attach(device_t dev) { struct ofw_cpu_softc *sc; + phandle_t node; uint32_t cell; sc = device_get_softc(dev); - OF_getprop(ofw_bus_get_node(dev), "reg", &cell, sizeof(cell)); + node = ofw_bus_get_node(dev); + if (OF_getencprop(node, "reg", &cell, sizeof(cell)) < 0) { + cell = device_get_unit(dev); + device_printf(dev, "missing 'reg' property, using %u\n", cell); + } sc->sc_cpu_pcpu = pcpu_find(cell); - OF_getprop(ofw_bus_get_node(dev), "clock-frequency", &cell, sizeof(cell)); + if (OF_getencprop(node, "clock-frequency", &cell, sizeof(cell)) < 0) { + device_printf(dev, "missing 'clock-frequency' property\n"); + return (ENXIO); + } sc->sc_nominal_mhz = cell / 1000000; /* convert to MHz */ bus_generic_probe(dev); diff --git a/sys/dev/pci/pci.c b/sys/dev/pci/pci.c index d2ac111..8f87851 100644 --- a/sys/dev/pci/pci.c +++ b/sys/dev/pci/pci.c @@ -268,7 +268,7 @@ static const struct pci_quirk pci_quirks[] = { { 0x43851002, PCI_QUIRK_UNMAP_REG, 0x14, 0 }, /* - * Atheros AR8161/AR8162/E2200 ethernet controller has a bug that + * Atheros AR8161/AR8162/E2200 Ethernet controllers have a bug that * MSI interrupt does not assert if PCIM_CMD_INTxDIS bit of the * command register is set. */ @@ -276,6 +276,17 @@ static const struct pci_quirk pci_quirks[] = { { 0xE0911969, PCI_QUIRK_MSI_INTX_BUG, 0, 0 }, { 0x10901969, PCI_QUIRK_MSI_INTX_BUG, 0, 0 }, + /* + * Broadcom BCM5714(S)/BCM5715(S)/BCM5780(S) Ethernet MACs don't + * issue MSI interrupts with PCIM_CMD_INTxDIS set either. + */ + { 0x166814e4, PCI_QUIRK_MSI_INTX_BUG, 0, 0 }, /* BCM5714 */ + { 0x166914e4, PCI_QUIRK_MSI_INTX_BUG, 0, 0 }, /* BCM5714S */ + { 0x166a14e4, PCI_QUIRK_MSI_INTX_BUG, 0, 0 }, /* BCM5780 */ + { 0x166b14e4, PCI_QUIRK_MSI_INTX_BUG, 0, 0 }, /* BCM5780S */ + { 0x167814e4, PCI_QUIRK_MSI_INTX_BUG, 0, 0 }, /* BCM5715 */ + { 0x167914e4, PCI_QUIRK_MSI_INTX_BUG, 0, 0 }, /* BCM5715S */ + { 0 } }; @@ -3866,14 +3877,16 @@ pci_setup_intr(device_t dev, device_t child, struct resource *irq, int flags, mte->mte_handlers++; } - if (!pci_has_quirk(pci_get_devid(dev), - PCI_QUIRK_MSI_INTX_BUG)) { - /* - * Make sure that INTx is disabled if we are - * using MSI/MSIX - */ + /* + * Make sure that INTx is disabled if we are using MSI/MSI-X, + * unless the device is affected by PCI_QUIRK_MSI_INTX_BUG, + * in which case we "enable" INTx so MSI/MSI-X actually works. + */ + if (!pci_has_quirk(pci_get_devid(child), + PCI_QUIRK_MSI_INTX_BUG)) pci_set_command_bit(dev, child, PCIM_CMD_INTxDIS); - } + else + pci_clear_command_bit(dev, child, PCIM_CMD_INTxDIS); bad: if (error) { (void)bus_generic_teardown_intr(dev, child, irq, diff --git a/sys/dev/sdhci/sdhci.c b/sys/dev/sdhci/sdhci.c index f92d42b..98861ab6 100644 --- a/sys/dev/sdhci/sdhci.c +++ b/sys/dev/sdhci/sdhci.c @@ -149,7 +149,6 @@ static void sdhci_reset(struct sdhci_slot *slot, uint8_t mask) { int timeout; - uint8_t res; if (slot->quirks & SDHCI_QUIRK_NO_CARD_NO_RESET) { if (!(RD4(slot, SDHCI_PRESENT_STATE) & @@ -168,26 +167,43 @@ sdhci_reset(struct sdhci_slot *slot, uint8_t mask) sdhci_set_clock(slot, clock); } - WR1(slot, SDHCI_SOFTWARE_RESET, mask); - if (mask & SDHCI_RESET_ALL) { slot->clock = 0; slot->power = 0; } + WR1(slot, SDHCI_SOFTWARE_RESET, mask); + + if (slot->quirks & SDHCI_QUIRK_WAITFOR_RESET_ASSERTED) { + /* + * Resets on TI OMAPs and AM335x are incompatible with SDHCI + * specification. The reset bit has internal propagation delay, + * so a fast read after write returns 0 even if reset process is + * in progress. The workaround is to poll for 1 before polling + * for 0. In the worst case, if we miss seeing it asserted the + * time we spent waiting is enough to ensure the reset finishes. + */ + timeout = 10000; + while ((RD1(slot, SDHCI_SOFTWARE_RESET) & mask) != mask) { + if (timeout <= 0) + break; + timeout--; + DELAY(1); + } + } + /* Wait max 100 ms */ - timeout = 100; + timeout = 10000; /* Controller clears the bits when it's done */ - while ((res = RD1(slot, SDHCI_SOFTWARE_RESET)) & mask) { - if (timeout == 0) { - slot_printf(slot, - "Reset 0x%x never completed - 0x%x.\n", - (int)mask, (int)res); + while (RD1(slot, SDHCI_SOFTWARE_RESET) & mask) { + if (timeout <= 0) { + slot_printf(slot, "Reset 0x%x never completed.\n", + mask); sdhci_dumpregs(slot); return; } timeout--; - DELAY(1000); + DELAY(10); } } @@ -713,9 +729,13 @@ sdhci_timeout(void *arg) struct sdhci_slot *slot = arg; if (slot->curcmd != NULL) { + slot_printf(slot, " Controller timeout\n"); + sdhci_dumpregs(slot); sdhci_reset(slot, SDHCI_RESET_CMD|SDHCI_RESET_DATA); slot->curcmd->error = MMC_ERR_TIMEOUT; sdhci_req_done(slot); + } else { + slot_printf(slot, " Spurious timeout - no active command\n"); } } @@ -1274,7 +1294,9 @@ sdhci_generic_intr(struct sdhci_slot *slot) /* Handle data interrupts. */ if (intmask & SDHCI_INT_DATA_MASK) { WR4(slot, SDHCI_INT_STATUS, intmask & SDHCI_INT_DATA_MASK); - sdhci_data_irq(slot, intmask & SDHCI_INT_DATA_MASK); + /* Dont call data_irq in case of errored command */ + if ((intmask & SDHCI_INT_CMD_ERROR_MASK) == 0) + sdhci_data_irq(slot, intmask & SDHCI_INT_DATA_MASK); } /* Handle AutoCMD12 error interrupt. */ if (intmask & SDHCI_INT_ACMD12ERR) { diff --git a/sys/dev/sdhci/sdhci.h b/sys/dev/sdhci/sdhci.h index 5cde2b0..ff1576e 100644 --- a/sys/dev/sdhci/sdhci.h +++ b/sys/dev/sdhci/sdhci.h @@ -59,6 +59,8 @@ #define SDHCI_QUIRK_MISSING_CAPS (1<<12) /* Hardware shifts the 136-bit response, don't do it in software. */ #define SDHCI_QUIRK_DONT_SHIFT_RESPONSE (1<<13) +/* Wait to see reset bit asserted before waiting for de-asserted */ +#define SDHCI_QUIRK_WAITFOR_RESET_ASSERTED (1<<14) /* * Controller registers @@ -182,8 +184,11 @@ #define SDHCI_INT_NORMAL_MASK 0x00007FFF #define SDHCI_INT_ERROR_MASK 0xFFFF8000 -#define SDHCI_INT_CMD_MASK (SDHCI_INT_RESPONSE | SDHCI_INT_TIMEOUT | \ +#define SDHCI_INT_CMD_ERROR_MASK (SDHCI_INT_TIMEOUT | \ SDHCI_INT_CRC | SDHCI_INT_END_BIT | SDHCI_INT_INDEX) + +#define SDHCI_INT_CMD_MASK (SDHCI_INT_RESPONSE | SDHCI_INT_CMD_ERROR_MASK) + #define SDHCI_INT_DATA_MASK (SDHCI_INT_DATA_END | SDHCI_INT_DMA_END | \ SDHCI_INT_DATA_AVAIL | SDHCI_INT_SPACE_AVAIL | \ SDHCI_INT_DATA_TIMEOUT | SDHCI_INT_DATA_CRC | \ diff --git a/sys/dev/usb/quirk/usb_quirk.c b/sys/dev/usb/quirk/usb_quirk.c index 1799d47..6b2f6b8 100644 --- a/sys/dev/usb/quirk/usb_quirk.c +++ b/sys/dev/usb/quirk/usb_quirk.c @@ -449,6 +449,7 @@ static struct usb_quirk_entry usb_quirks[USB_DEV_QUIRKS_MAX] = { USB_QUIRK(WESTERN, MYPASSPORT_08, 0x0000, 0xffff, UQ_MSC_NO_SYNC_CACHE), USB_QUIRK(WESTERN, MYPASSPORT_09, 0x0000, 0xffff, UQ_MSC_NO_SYNC_CACHE), USB_QUIRK(WESTERN, MYPASSPORT_10, 0x0000, 0xffff, UQ_MSC_NO_SYNC_CACHE), + USB_QUIRK(WESTERN, MYPASSPORT_11, 0x0000, 0xffff, UQ_MSC_NO_SYNC_CACHE), USB_QUIRK(WESTERN, MYPASSPORTES_00, 0x0000, 0xffff, UQ_MSC_NO_SYNC_CACHE), USB_QUIRK(WESTERN, MYPASSPORTES_01, 0x0000, 0xffff, UQ_MSC_NO_SYNC_CACHE), USB_QUIRK(WESTERN, MYPASSPORTES_02, 0x0000, 0xffff, UQ_MSC_NO_SYNC_CACHE), diff --git a/sys/dev/usb/usbdevs b/sys/dev/usb/usbdevs index 2d86d1b..b9ea927 100644 --- a/sys/dev/usb/usbdevs +++ b/sys/dev/usb/usbdevs @@ -2411,6 +2411,8 @@ product INTEL EASYPC_CAMERA 0x0110 Easy PC Camera product INTEL TESTBOARD 0x9890 82930 test board product INTEL2 IRMH 0x0020 Integrated Rate Matching Hub product INTEL2 IRMH2 0x0024 Integrated Rate Matching Hub +product INTEL2 IRMH3 0x8000 Integrated Rate Matching Hub +product INTEL2 IRMH4 0x8008 Integrated Rate Matching Hub /* Interbiometric products */ product INTERBIOMETRICS IOBOARD 0x1002 FTDI compatible adapter @@ -4492,6 +4494,7 @@ product WESTERN EXTHDD 0x0400 External HDD product WESTERN HUB 0x0500 USB HUB product WESTERN MYBOOK 0x0901 MyBook External HDD product WESTERN MYPASSPORT_00 0x0704 MyPassport External HDD +product WESTERN MYPASSPORT_11 0x0741 MyPassport External HDD product WESTERN MYPASSPORT_01 0x0746 MyPassport External HDD product WESTERN MYPASSPORT_02 0x0748 MyPassport External HDD product WESTERN MYPASSPORT_03 0x074A MyPassport External HDD diff --git a/sys/dev/vt/hw/efifb/efifb.c b/sys/dev/vt/hw/efifb/efifb.c index 05b2d79..7695547 100644 --- a/sys/dev/vt/hw/efifb/efifb.c +++ b/sys/dev/vt/hw/efifb/efifb.c @@ -37,6 +37,9 @@ __FBSDID("$FreeBSD$"); #include <sys/kernel.h> #include <sys/fbio.h> #include <sys/linker.h> +#include <sys/bus.h> +#include <sys/module.h> +#include <sys/rman.h> #include "opt_platform.h" @@ -179,3 +182,53 @@ vt_efifb_remap(void *xinfo) info->fb_size, VM_MEMATTR_WRITE_COMBINING); } +/* Dummy NewBus functions to reserve the resources used by the efifb driver */ +static void +vtefifb_identify(driver_t *driver, device_t parent) +{ + + if (local_info.fb_pbase == 0 || local_info.fb_size == 0) + return; + + if (BUS_ADD_CHILD(parent, 0, driver->name, 0) == NULL) + panic("Unable to attach vt_efifb console"); +} + +static int +vtefifb_probe(device_t dev) +{ + + device_set_desc(dev, "vt_efifb driver"); + return (BUS_PROBE_NOWILDCARD); +} + +static int +vtefifb_attach(device_t dev) +{ + struct resource *pseudo_phys_res; + int res_id; + + res_id = 0; + pseudo_phys_res = bus_alloc_resource(dev, SYS_RES_MEMORY, + &res_id, local_info.fb_pbase, + local_info.fb_pbase + local_info.fb_size, + local_info.fb_size, RF_ACTIVE); + if (pseudo_phys_res == NULL) + panic("Unable to reserve vt_efifb memory"); + return (0); +} + +/*-------------------- Private Device Attachment Data -----------------------*/ +static device_method_t vtefifb_methods[] = { + /* Device interface */ + DEVMETHOD(device_identify, vtefifb_identify), + DEVMETHOD(device_probe, vtefifb_probe), + DEVMETHOD(device_attach, vtefifb_attach), + + DEVMETHOD_END +}; + +DEFINE_CLASS_0(vtefifb, vtefifb_driver, vtefifb_methods, 0); +devclass_t vtefifb_devclass; + +DRIVER_MODULE(vtefifb, nexus, vtefifb_driver, vtefifb_devclass, NULL, NULL); diff --git a/sys/dev/vt/hw/vga/vt_vga.c b/sys/dev/vt/hw/vga/vt_vga.c index abddbf5..7f35053 100644 --- a/sys/dev/vt/hw/vga/vt_vga.c +++ b/sys/dev/vt/hw/vga/vt_vga.c @@ -36,6 +36,9 @@ __FBSDID("$FreeBSD$"); #include <sys/param.h> #include <sys/kernel.h> #include <sys/systm.h> +#include <sys/bus.h> +#include <sys/module.h> +#include <sys/rman.h> #include <dev/vt/vt.h> #include <dev/vt/hw/vga/vt_vga_reg.h> @@ -56,6 +59,7 @@ struct vga_softc { bus_space_handle_t vga_reg_handle; int vga_wmode; term_color_t vga_curfg, vga_curbg; + boolean_t vga_enabled; }; /* Convenience macros. */ @@ -1228,6 +1232,7 @@ vga_init(struct vt_device *vd) vd->vd_height = VT_VGA_HEIGHT; } vga_initialize(vd, textmode); + sc->vga_enabled = true; return (CN_INTERNAL); } @@ -1241,3 +1246,53 @@ vga_postswitch(struct vt_device *vd) /* Ask vt(9) to update chars on visible area. */ vd->vd_flags |= VDF_INVALID; } + +/* Dummy NewBus functions to reserve the resources used by the vt_vga driver */ +static void +vtvga_identify(driver_t *driver, device_t parent) +{ + + if (!vga_conssoftc.vga_enabled) + return; + + if (BUS_ADD_CHILD(parent, 0, driver->name, 0) == NULL) + panic("Unable to attach vt_vga console"); +} + +static int +vtvga_probe(device_t dev) +{ + + device_set_desc(dev, "vt_vga driver"); + return (BUS_PROBE_NOWILDCARD); +} + +static int +vtvga_attach(device_t dev) +{ + struct resource *pseudo_phys_res; + int res_id; + + res_id = 0; + pseudo_phys_res = bus_alloc_resource(dev, SYS_RES_MEMORY, + &res_id, VGA_MEM_BASE, VGA_MEM_BASE + VGA_MEM_SIZE, + VGA_MEM_SIZE, RF_ACTIVE); + if (pseudo_phys_res == NULL) + panic("Unable to reserve vt_vga memory"); + return (0); +} + +/*-------------------- Private Device Attachment Data -----------------------*/ +static device_method_t vtvga_methods[] = { + /* Device interface */ + DEVMETHOD(device_identify, vtvga_identify), + DEVMETHOD(device_probe, vtvga_probe), + DEVMETHOD(device_attach, vtvga_attach), + + DEVMETHOD_END +}; + +DEFINE_CLASS_0(vtvga, vtvga_driver, vtvga_methods, 0); +devclass_t vtvga_devclass; + +DRIVER_MODULE(vtvga, nexus, vtvga_driver, vtvga_devclass, NULL, NULL); diff --git a/sys/dev/vt/vt.h b/sys/dev/vt/vt.h index 262ffaf..67fcf1e 100644 --- a/sys/dev/vt/vt.h +++ b/sys/dev/vt/vt.h @@ -155,6 +155,7 @@ struct vt_device { int vd_keyboard; /* (G) Keyboard index. */ unsigned int vd_kbstate; /* (?) Device unit. */ unsigned int vd_unit; /* (c) Device unit. */ + int vd_altbrk; /* (?) Alt break seq. state */ }; #define VD_PASTEBUF(vd) ((vd)->vd_pastebuf.vpb_buf) diff --git a/sys/dev/vt/vt_core.c b/sys/dev/vt/vt_core.c index 6d31fae..51c4623 100644 --- a/sys/dev/vt/vt_core.c +++ b/sys/dev/vt/vt_core.c @@ -825,7 +825,9 @@ vt_processkey(keyboard_t *kbd, struct vt_device *vd, int c) terminal_input_char(vw->vw_terminal, 0x1b); } #endif - +#if defined(KDB) + kdb_alt_break(c, &vd->vd_altbrk); +#endif terminal_input_char(vw->vw_terminal, KEYCHAR(c)); } else terminal_input_raw(vw->vw_terminal, c); diff --git a/sys/dev/wpi/if_wpi.c b/sys/dev/wpi/if_wpi.c index e94fa74..accb46c 100644 --- a/sys/dev/wpi/if_wpi.c +++ b/sys/dev/wpi/if_wpi.c @@ -2564,7 +2564,7 @@ wpi_scan(struct wpi_softc *sc) struct ieee80211_channel *c; enum ieee80211_phymode mode; uint8_t *frm; - int nrates, pktlen, error, i, nssid; + int pktlen, error, i, nssid; bus_addr_t physaddr; desc = &ring->desc[ring->cur]; @@ -2613,7 +2613,7 @@ wpi_scan(struct wpi_softc *sc) nssid = MIN(ss->ss_nssid, WPI_SCAN_MAX_ESSIDS); for (i = 0; i < nssid; i++) { hdr->scan_essids[i].id = IEEE80211_ELEMID_SSID; - hdr->scan_essids[i].esslen = MIN(ss->ss_ssid[i].len, 32); + hdr->scan_essids[i].esslen = MIN(ss->ss_ssid[i].len, IEEE80211_NWID_LEN); memcpy(hdr->scan_essids[i].essid, ss->ss_ssid[i].ssid, hdr->scan_essids[i].esslen); #ifdef WPI_DEBUG @@ -2630,7 +2630,7 @@ wpi_scan(struct wpi_softc *sc) * Build a probe request frame. Most of the following code is a * copy & paste of what is done in net80211. */ - wh = (struct ieee80211_frame *)&hdr->scan_essids[4]; + wh = (struct ieee80211_frame *)&hdr->scan_essids[WPI_SCAN_MAX_ESSIDS]; wh->i_fc[0] = IEEE80211_FC0_VERSION_0 | IEEE80211_FC0_TYPE_MGT | IEEE80211_FC0_SUBTYPE_PROBE_REQ; wh->i_fc[1] = IEEE80211_FC1_DIR_NODS; @@ -2642,30 +2642,12 @@ wpi_scan(struct wpi_softc *sc) frm = (uint8_t *)(wh + 1); - /* add essid IE, the hardware will fill this in for us */ - *frm++ = IEEE80211_ELEMID_SSID; - *frm++ = 0; - mode = ieee80211_chan2mode(ic->ic_curchan); rs = &ic->ic_sup_rates[mode]; - /* add supported rates IE */ - *frm++ = IEEE80211_ELEMID_RATES; - nrates = rs->rs_nrates; - if (nrates > IEEE80211_RATE_SIZE) - nrates = IEEE80211_RATE_SIZE; - *frm++ = nrates; - memcpy(frm, rs->rs_rates, nrates); - frm += nrates; - - /* add supported xrates IE */ - if (rs->rs_nrates > IEEE80211_RATE_SIZE) { - nrates = rs->rs_nrates - IEEE80211_RATE_SIZE; - *frm++ = IEEE80211_ELEMID_XRATES; - *frm++ = nrates; - memcpy(frm, rs->rs_rates + IEEE80211_RATE_SIZE, nrates); - frm += nrates; - } + frm = ieee80211_add_ssid(frm, NULL, 0); + frm = ieee80211_add_rates(frm, rs); + frm = ieee80211_add_xrates(frm, rs); /* setup length of probe request */ hdr->tx.len = htole16(frm - (uint8_t *)wh); diff --git a/sys/dev/wpi/if_wpireg.h b/sys/dev/wpi/if_wpireg.h index df71b3d..60d183a 100644 --- a/sys/dev/wpi/if_wpireg.h +++ b/sys/dev/wpi/if_wpireg.h @@ -511,7 +511,7 @@ struct { struct { uint8_t id; uint8_t esslen; - uint8_t essid[32]; + uint8_t essid[IEEE80211_NWID_LEN]; }scan_essids[WPI_SCAN_MAX_ESSIDS]; /* followed by probe request body */ /* followed by nchan x wpi_scan_chan */ diff --git a/sys/fs/ext2fs/ext2_vnops.c b/sys/fs/ext2fs/ext2_vnops.c index 899ae79..5baca5a 100644 --- a/sys/fs/ext2fs/ext2_vnops.c +++ b/sys/fs/ext2fs/ext2_vnops.c @@ -239,8 +239,10 @@ ext2_create(struct vop_create_args *ap) error = ext2_makeinode(MAKEIMODE(ap->a_vap->va_type, ap->a_vap->va_mode), ap->a_dvp, ap->a_vpp, ap->a_cnp); - if (error) + if (error != 0) return (error); + if ((ap->a_cnp->cn_flags & MAKEENTRY) != 0) + cache_enter(ap->a_dvp, *ap->a_vpp, ap->a_cnp); return (0); } diff --git a/sys/fs/msdosfs/msdosfs_vnops.c b/sys/fs/msdosfs/msdosfs_vnops.c index a60fa13..3294656 100644 --- a/sys/fs/msdosfs/msdosfs_vnops.c +++ b/sys/fs/msdosfs/msdosfs_vnops.c @@ -184,6 +184,8 @@ msdosfs_create(ap) if (error) goto bad; *ap->a_vpp = DETOV(dep); + if ((cnp->cn_flags & MAKEENTRY) != 0) + cache_enter(ap->a_dvp, *ap->a_vpp, cnp); return (0); bad: diff --git a/sys/fs/nandfs/nandfs_vnops.c b/sys/fs/nandfs/nandfs_vnops.c index 65dcf64..40dd855 100644 --- a/sys/fs/nandfs/nandfs_vnops.c +++ b/sys/fs/nandfs/nandfs_vnops.c @@ -1411,6 +1411,8 @@ nandfs_create(struct vop_create_args *ap) return (error); } *vpp = NTOV(node); + if ((cnp->cn_flags & MAKEENTRY) != 0) + cache_enter(dvp, *vpp, cnp); DPRINTF(VNCALL, ("created file vp %p nandnode %p ino %jx\n", *vpp, node, (uintmax_t)node->nn_ino)); diff --git a/sys/fs/nfs/nfs_commonkrpc.c b/sys/fs/nfs/nfs_commonkrpc.c index 363ff89..c1d50f4 100644 --- a/sys/fs/nfs/nfs_commonkrpc.c +++ b/sys/fs/nfs/nfs_commonkrpc.c @@ -260,7 +260,7 @@ newnfs_connect(struct nfsmount *nmp, struct nfssockreq *nrp, client = clnt_reconnect_create(nconf, saddr, nrp->nr_prog, nrp->nr_vers, sndreserve, rcvreserve); - CLNT_CONTROL(client, CLSET_WAITCHAN, "newnfsreq"); + CLNT_CONTROL(client, CLSET_WAITCHAN, "nfsreq"); if (nmp != NULL) { if ((nmp->nm_flag & NFSMNT_INT)) CLNT_CONTROL(client, CLSET_INTERRUPTIBLE, &one); @@ -1166,10 +1166,10 @@ nfs_msg(struct thread *td, const char *server, const char *msg, int error) p = td ? td->td_proc : NULL; if (error) { - tprintf(p, LOG_INFO, "newnfs server %s: %s, error %d\n", + tprintf(p, LOG_INFO, "nfs server %s: %s, error %d\n", server, msg, error); } else { - tprintf(p, LOG_INFO, "newnfs server %s: %s\n", server, msg); + tprintf(p, LOG_INFO, "nfs server %s: %s\n", server, msg); } return (0); } diff --git a/sys/fs/nfs/nfs_commonsubs.c b/sys/fs/nfs/nfs_commonsubs.c index d9f8436..0f6db32 100644 --- a/sys/fs/nfs/nfs_commonsubs.c +++ b/sys/fs/nfs/nfs_commonsubs.c @@ -219,7 +219,8 @@ nfsm_mbufuio(struct nfsrv_descript *nd, struct uio *uiop, int siz) } mbufcp = NFSMTOD(mp, caddr_t); len = mbuf_len(mp); - KASSERT(len > 0, ("len %d", len)); + KASSERT(len >= 0, + ("len %d, corrupted mbuf?", len)); } xfer = (left > len) ? len : left; #ifdef notdef diff --git a/sys/fs/nfs/nfsport.h b/sys/fs/nfs/nfsport.h index 88c7a34..9f08854 100644 --- a/sys/fs/nfs/nfsport.h +++ b/sys/fs/nfs/nfsport.h @@ -930,24 +930,6 @@ void nfsd_mntinit(void); int newnfs_iosize(struct nfsmount *); -#ifdef NFS_DEBUG - -extern int nfs_debug; -#define NFS_DEBUG_ASYNCIO 1 /* asynchronous i/o */ -#define NFS_DEBUG_WG 2 /* server write gathering */ -#define NFS_DEBUG_RC 4 /* server request caching */ - -#define NFS_DPF(cat, args) \ - do { \ - if (nfs_debug & NFS_DEBUG_##cat) printf args; \ - } while (0) - -#else - -#define NFS_DPF(cat, args) - -#endif - int newnfs_vncmpf(struct vnode *, void *); #ifndef NFS_MINDIRATTRTIMO diff --git a/sys/fs/nfsclient/nfs.h b/sys/fs/nfsclient/nfs.h index 183515c..e05e553 100644 --- a/sys/fs/nfsclient/nfs.h +++ b/sys/fs/nfsclient/nfs.h @@ -55,6 +55,24 @@ #define NFS_ISV34(v) \ (VFSTONFS((v)->v_mount)->nm_flag & (NFSMNT_NFSV3 | NFSMNT_NFSV4)) +#ifdef NFS_DEBUG + +extern int nfs_debug; +#define NFS_DEBUG_ASYNCIO 1 /* asynchronous i/o */ +#define NFS_DEBUG_WG 2 /* server write gathering */ +#define NFS_DEBUG_RC 4 /* server request caching */ + +#define NFS_DPF(cat, args) \ + do { \ + if (nfs_debug & NFS_DEBUG_##cat) printf args; \ + } while (0) + +#else + +#define NFS_DPF(cat, args) + +#endif + /* * NFS iod threads can be in one of these three states once spawned. * NFSIOD_NOT_AVAILABLE - Cannot be assigned an I/O operation at this time. diff --git a/sys/fs/nfsclient/nfs_clnode.c b/sys/fs/nfsclient/nfs_clnode.c index 5052261..b1a3451 100644 --- a/sys/fs/nfsclient/nfs_clnode.c +++ b/sys/fs/nfsclient/nfs_clnode.c @@ -122,7 +122,7 @@ ncl_nget(struct mount *mntp, u_int8_t *fhp, int fhsize, struct nfsnode **npp, } np = uma_zalloc(newnfsnode_zone, M_WAITOK | M_ZERO); - error = getnewvnode("newnfs", mntp, &newnfs_vnodeops, &nvp); + error = getnewvnode("nfs", mntp, &newnfs_vnodeops, &nvp); if (error) { uma_zfree(newnfsnode_zone, np); return (error); diff --git a/sys/fs/nfsclient/nfs_clport.c b/sys/fs/nfsclient/nfs_clport.c index 8529c76..efdc812 100644 --- a/sys/fs/nfsclient/nfs_clport.c +++ b/sys/fs/nfsclient/nfs_clport.c @@ -198,7 +198,7 @@ nfscl_nget(struct mount *mntp, struct vnode *dvp, struct nfsfh *nfhp, } np = uma_zalloc(newnfsnode_zone, M_WAITOK | M_ZERO); - error = getnewvnode("newnfs", mntp, &newnfs_vnodeops, &nvp); + error = getnewvnode("nfs", mntp, &newnfs_vnodeops, &nvp); if (error) { uma_zfree(newnfsnode_zone, np); FREE((caddr_t)nfhp, M_NFSFH); diff --git a/sys/fs/nfsclient/nfs_clvfsops.c b/sys/fs/nfsclient/nfs_clvfsops.c index 33f92c6..9758b4c 100644 --- a/sys/fs/nfsclient/nfs_clvfsops.c +++ b/sys/fs/nfsclient/nfs_clvfsops.c @@ -100,6 +100,11 @@ SYSCTL_INT(_vfs_nfs, NFS_TPRINTF_INITIAL_DELAY, static int nfs_tprintf_delay = NFS_TPRINTF_DELAY; SYSCTL_INT(_vfs_nfs, NFS_TPRINTF_DELAY, downdelayinterval, CTLFLAG_RW, &nfs_tprintf_delay, 0, ""); +#ifdef NFS_DEBUG +int nfs_debug; +SYSCTL_INT(_vfs_nfs, OID_AUTO, debug, CTLFLAG_RW, &nfs_debug, 0, + "Toggle debug flag"); +#endif static int nfs_mountroot(struct mount *); static void nfs_sec_name(char *, int *); @@ -152,7 +157,7 @@ MODULE_DEPEND(nfs, nfslock, 1, 1, 1); * will be defined for kernels built without NFS_ROOT, although it * isn't used in that case. */ -#if !defined(NFS_ROOT) && !defined(NFSCLIENT) +#if !defined(NFS_ROOT) struct nfs_diskless nfs_diskless = { { { 0 } } }; struct nfsv3_diskless nfsv3_diskless = { { { 0 } } }; int nfs_diskless_valid = 0; @@ -704,7 +709,7 @@ nfs_decode_args(struct mount *mp, struct nfsmount *nmp, struct nfs_args *argp, while (newnfs_connect(nmp, &nmp->nm_sockreq, cred, td, 0)) { printf("newnfs_args: retrying connect\n"); - (void) nfs_catnap(PSOCK, 0, "newnfscon"); + (void) nfs_catnap(PSOCK, 0, "nfscon"); } } } else { @@ -1063,7 +1068,7 @@ nfs_mount(struct mount *mp) * greater than NFS_MAXDGRAMDATA, those thread(s) will be * hung, retrying the RPC(s) forever. Usually these threads * will be seen doing an uninterruptible sleep on wait channel - * "newnfsreq" (truncated to "newnfsre" by procstat). + * "nfsreq". */ if (args.sotype == SOCK_DGRAM && nmp->nm_sotype == SOCK_STREAM) tprintf(td->td_proc, LOG_WARNING, diff --git a/sys/fs/nfsclient/nfs_clvnops.c b/sys/fs/nfsclient/nfs_clvnops.c index 818551f..513abf4 100644 --- a/sys/fs/nfsclient/nfs_clvnops.c +++ b/sys/fs/nfsclient/nfs_clvnops.c @@ -1606,20 +1606,6 @@ again: } } else if (NFS_ISV34(dvp) && (fmode & O_EXCL)) { if (nfscl_checksattr(vap, &nfsva)) { - /* - * We are normally called with only a partially - * initialized VAP. Since the NFSv3 spec says that - * the server may use the file attributes to - * store the verifier, the spec requires us to do a - * SETATTR RPC. FreeBSD servers store the verifier in - * atime, but we can't really assume that all servers - * will so we ensure that our SETATTR sets both atime - * and mtime. - */ - if (vap->va_mtime.tv_sec == VNOVAL) - vfs_timestamp(&vap->va_mtime); - if (vap->va_atime.tv_sec == VNOVAL) - vap->va_atime = vap->va_mtime; error = nfsrpc_setattr(newvp, vap, NULL, cnp->cn_cred, cnp->cn_thread, &nfsva, &attrflag, NULL); if (error && (vap->va_uid != (uid_t)VNOVAL || diff --git a/sys/fs/nfsserver/nfs_nfsdport.c b/sys/fs/nfsserver/nfs_nfsdport.c index 5bd4538..17ca5a6 100644 --- a/sys/fs/nfsserver/nfs_nfsdport.c +++ b/sys/fs/nfsserver/nfs_nfsdport.c @@ -2970,12 +2970,7 @@ nfsvno_advlock(struct vnode *vp, int ftype, u_int64_t first, if (nfsrv_dolocallocks == 0) goto out; - - /* Check for VI_DOOMED here, so that VOP_ADVLOCK() isn't performed. */ - if ((vp->v_iflag & VI_DOOMED) != 0) { - error = EPERM; - goto out; - } + ASSERT_VOP_UNLOCKED(vp, "nfsvno_advlock: vp locked"); fl.l_whence = SEEK_SET; fl.l_type = ftype; @@ -2999,14 +2994,12 @@ nfsvno_advlock(struct vnode *vp, int ftype, u_int64_t first, fl.l_pid = (pid_t)0; fl.l_sysid = (int)nfsv4_sysid; - NFSVOPUNLOCK(vp, 0); if (ftype == F_UNLCK) error = VOP_ADVLOCK(vp, (caddr_t)td->td_proc, F_UNLCK, &fl, (F_POSIX | F_REMOTE)); else error = VOP_ADVLOCK(vp, (caddr_t)td->td_proc, F_SETLK, &fl, (F_POSIX | F_REMOTE)); - NFSVOPLOCK(vp, LK_EXCLUSIVE | LK_RETRY); out: NFSEXITCODE(error); diff --git a/sys/fs/nfsserver/nfs_nfsdstate.c b/sys/fs/nfsserver/nfs_nfsdstate.c index 97f8fff..5e9beeb 100644 --- a/sys/fs/nfsserver/nfs_nfsdstate.c +++ b/sys/fs/nfsserver/nfs_nfsdstate.c @@ -1344,6 +1344,8 @@ nfsrv_freeallnfslocks(struct nfsstate *stp, vnode_t vp, int cansleep, vnode_t tvp = NULL; uint64_t first, end; + if (vp != NULL) + ASSERT_VOP_UNLOCKED(vp, "nfsrv_freeallnfslocks: vnode locked"); lop = LIST_FIRST(&stp->ls_lock); while (lop != LIST_END(&stp->ls_lock)) { nlop = LIST_NEXT(lop, lo_lckowner); @@ -1363,9 +1365,10 @@ nfsrv_freeallnfslocks(struct nfsstate *stp, vnode_t vp, int cansleep, if (gottvp == 0) { if (nfsrv_dolocallocks == 0) tvp = NULL; - else if (vp == NULL && cansleep != 0) + else if (vp == NULL && cansleep != 0) { tvp = nfsvno_getvp(&lfp->lf_fh); - else + NFSVOPUNLOCK(tvp, 0); + } else tvp = vp; gottvp = 1; } @@ -1386,7 +1389,7 @@ nfsrv_freeallnfslocks(struct nfsstate *stp, vnode_t vp, int cansleep, lop = nlop; } if (vp == NULL && tvp != NULL) - vput(tvp); + vrele(tvp); } /* @@ -1497,7 +1500,7 @@ nfsrv_lockctrl(vnode_t vp, struct nfsstate **new_stpp, struct nfsclient *clp = NULL; u_int32_t bits; int error = 0, haslock = 0, ret, reterr; - int getlckret, delegation = 0, filestruct_locked; + int getlckret, delegation = 0, filestruct_locked, vnode_unlocked = 0; fhandle_t nfh; uint64_t first, end; uint32_t lock_flags; @@ -1587,6 +1590,11 @@ tryagain: * locking rolled back. */ NFSUNLOCKSTATE(); + if (vnode_unlocked == 0) { + ASSERT_VOP_ELOCKED(vp, "nfsrv_lockctrl1"); + vnode_unlocked = 1; + NFSVOPUNLOCK(vp, 0); + } reterr = nfsrv_locallock(vp, lfp, (new_lop->lo_flags & (NFSLCK_READ | NFSLCK_WRITE)), new_lop->lo_first, new_lop->lo_end, cfp, p); @@ -1756,6 +1764,11 @@ tryagain: if (filestruct_locked != 0) { /* Roll back local locks. */ NFSUNLOCKSTATE(); + if (vnode_unlocked == 0) { + ASSERT_VOP_ELOCKED(vp, "nfsrv_lockctrl2"); + vnode_unlocked = 1; + NFSVOPUNLOCK(vp, 0); + } nfsrv_locallock_rollback(vp, lfp, p); NFSLOCKSTATE(); nfsrv_unlocklf(lfp); @@ -1833,6 +1846,12 @@ tryagain: if (filestruct_locked != 0) { /* Roll back local locks. */ NFSUNLOCKSTATE(); + if (vnode_unlocked == 0) { + ASSERT_VOP_ELOCKED(vp, + "nfsrv_lockctrl3"); + vnode_unlocked = 1; + NFSVOPUNLOCK(vp, 0); + } nfsrv_locallock_rollback(vp, lfp, p); NFSLOCKSTATE(); nfsrv_unlocklf(lfp); @@ -1852,6 +1871,8 @@ tryagain: bits = tstp->ls_flags; bits >>= NFSLCK_SHIFT; if (new_stp->ls_flags & bits & NFSLCK_ACCESSBITS) { + KASSERT(vnode_unlocked == 0, + ("nfsrv_lockctrl: vnode unlocked1")); ret = nfsrv_clientconflict(tstp->ls_clp, &haslock, vp, p); if (ret == 1) { @@ -1883,6 +1904,8 @@ tryagain: * For setattr, just get rid of all the Delegations for other clients. */ if (new_stp->ls_flags & NFSLCK_SETATTR) { + KASSERT(vnode_unlocked == 0, + ("nfsrv_lockctrl: vnode unlocked2")); ret = nfsrv_cleandeleg(vp, lfp, clp, &haslock, p); if (ret) { /* @@ -1933,14 +1956,26 @@ tryagain: (new_lop->lo_flags & NFSLCK_WRITE) && (clp != tstp->ls_clp || (tstp->ls_flags & NFSLCK_DELEGREAD)))) { + ret = 0; if (filestruct_locked != 0) { /* Roll back local locks. */ NFSUNLOCKSTATE(); + if (vnode_unlocked == 0) { + ASSERT_VOP_ELOCKED(vp, "nfsrv_lockctrl4"); + NFSVOPUNLOCK(vp, 0); + } nfsrv_locallock_rollback(vp, lfp, p); NFSLOCKSTATE(); nfsrv_unlocklf(lfp); + NFSUNLOCKSTATE(); + NFSVOPLOCK(vp, LK_EXCLUSIVE | LK_RETRY); + vnode_unlocked = 0; + if ((vp->v_iflag & VI_DOOMED) != 0) + ret = NFSERR_SERVERFAULT; + NFSLOCKSTATE(); } - ret = nfsrv_delegconflict(tstp, &haslock, p, vp); + if (ret == 0) + ret = nfsrv_delegconflict(tstp, &haslock, p, vp); if (ret) { /* * nfsrv_delegconflict unlocks state when it @@ -1979,6 +2014,11 @@ tryagain: stateidp->other[2] = stp->ls_stateid.other[2]; if (filestruct_locked != 0) { NFSUNLOCKSTATE(); + if (vnode_unlocked == 0) { + ASSERT_VOP_ELOCKED(vp, "nfsrv_lockctrl5"); + vnode_unlocked = 1; + NFSVOPUNLOCK(vp, 0); + } /* Update the local locks. */ nfsrv_localunlock(vp, lfp, first, end, p); NFSLOCKSTATE(); @@ -2009,14 +2049,29 @@ tryagain: FREE((caddr_t)other_lop, M_NFSDLOCK); other_lop = NULL; } - ret = nfsrv_clientconflict(lop->lo_stp->ls_clp,&haslock,vp,p); + if (vnode_unlocked != 0) + ret = nfsrv_clientconflict(lop->lo_stp->ls_clp, &haslock, + NULL, p); + else + ret = nfsrv_clientconflict(lop->lo_stp->ls_clp, &haslock, + vp, p); if (ret == 1) { if (filestruct_locked != 0) { + if (vnode_unlocked == 0) { + ASSERT_VOP_ELOCKED(vp, "nfsrv_lockctrl6"); + NFSVOPUNLOCK(vp, 0); + } /* Roll back local locks. */ nfsrv_locallock_rollback(vp, lfp, p); NFSLOCKSTATE(); nfsrv_unlocklf(lfp); NFSUNLOCKSTATE(); + NFSVOPLOCK(vp, LK_EXCLUSIVE | LK_RETRY); + vnode_unlocked = 0; + if ((vp->v_iflag & VI_DOOMED) != 0) { + error = NFSERR_SERVERFAULT; + goto out; + } } /* * nfsrv_clientconflict() unlocks state when it @@ -2050,6 +2105,11 @@ tryagain: if (filestruct_locked != 0 && ret == 0) { /* Roll back local locks. */ NFSUNLOCKSTATE(); + if (vnode_unlocked == 0) { + ASSERT_VOP_ELOCKED(vp, "nfsrv_lockctrl7"); + vnode_unlocked = 1; + NFSVOPUNLOCK(vp, 0); + } nfsrv_locallock_rollback(vp, lfp, p); NFSLOCKSTATE(); nfsrv_unlocklf(lfp); @@ -2128,6 +2188,11 @@ out: nfsv4_unlock(&nfsv4rootfs_lock, 1); NFSUNLOCKV4ROOTMUTEX(); } + if (vnode_unlocked != 0) { + NFSVOPLOCK(vp, LK_EXCLUSIVE | LK_RETRY); + if (error == 0 && (vp->v_iflag & VI_DOOMED) != 0) + error = NFSERR_SERVERFAULT; + } if (other_lop) FREE((caddr_t)other_lop, M_NFSDLOCK); NFSEXITCODE2(error, nd); @@ -3235,11 +3300,14 @@ nfsrv_openupdate(vnode_t vp, struct nfsstate *new_stp, nfsquad_t clientid, /* Get the lf lock */ nfsrv_locklf(lfp); NFSUNLOCKSTATE(); + ASSERT_VOP_ELOCKED(vp, "nfsrv_openupdate"); + NFSVOPUNLOCK(vp, 0); if (nfsrv_freeopen(stp, vp, 1, p) == 0) { NFSLOCKSTATE(); nfsrv_unlocklf(lfp); NFSUNLOCKSTATE(); } + NFSVOPLOCK(vp, LK_EXCLUSIVE | LK_RETRY); } else { (void) nfsrv_freeopen(stp, NULL, 0, p); NFSUNLOCKSTATE(); @@ -4627,7 +4695,7 @@ static int nfsrv_clientconflict(struct nfsclient *clp, int *haslockp, vnode_t vp, NFSPROC_T *p) { - int gotlock, lktype; + int gotlock, lktype = 0; /* * If lease hasn't expired, we can't fix it. @@ -4637,8 +4705,10 @@ nfsrv_clientconflict(struct nfsclient *clp, int *haslockp, vnode_t vp, return (0); if (*haslockp == 0) { NFSUNLOCKSTATE(); - lktype = NFSVOPISLOCKED(vp); - NFSVOPUNLOCK(vp, 0); + if (vp != NULL) { + lktype = NFSVOPISLOCKED(vp); + NFSVOPUNLOCK(vp, 0); + } NFSLOCKV4ROOTMUTEX(); nfsv4_relref(&nfsv4rootfs_lock); do { @@ -4647,11 +4717,12 @@ nfsrv_clientconflict(struct nfsclient *clp, int *haslockp, vnode_t vp, } while (!gotlock); NFSUNLOCKV4ROOTMUTEX(); *haslockp = 1; - NFSVOPLOCK(vp, lktype | LK_RETRY); - if ((vp->v_iflag & VI_DOOMED) != 0) - return (2); - else - return (1); + if (vp != NULL) { + NFSVOPLOCK(vp, lktype | LK_RETRY); + if ((vp->v_iflag & VI_DOOMED) != 0) + return (2); + } + return (1); } NFSUNLOCKSTATE(); @@ -4692,7 +4763,7 @@ nfsrv_delegconflict(struct nfsstate *stp, int *haslockp, NFSPROC_T *p, vnode_t vp) { struct nfsclient *clp = stp->ls_clp; - int gotlock, error, lktype, retrycnt, zapped_clp; + int gotlock, error, lktype = 0, retrycnt, zapped_clp; nfsv4stateid_t tstateid; fhandle_t tfh; @@ -4809,8 +4880,10 @@ nfsrv_delegconflict(struct nfsstate *stp, int *haslockp, NFSPROC_T *p, */ if (*haslockp == 0) { NFSUNLOCKSTATE(); - lktype = NFSVOPISLOCKED(vp); - NFSVOPUNLOCK(vp, 0); + if (vp != NULL) { + lktype = NFSVOPISLOCKED(vp); + NFSVOPUNLOCK(vp, 0); + } NFSLOCKV4ROOTMUTEX(); nfsv4_relref(&nfsv4rootfs_lock); do { @@ -4819,14 +4892,16 @@ nfsrv_delegconflict(struct nfsstate *stp, int *haslockp, NFSPROC_T *p, } while (!gotlock); NFSUNLOCKV4ROOTMUTEX(); *haslockp = 1; - NFSVOPLOCK(vp, lktype | LK_RETRY); - if ((vp->v_iflag & VI_DOOMED) != 0) { - *haslockp = 0; - NFSLOCKV4ROOTMUTEX(); - nfsv4_unlock(&nfsv4rootfs_lock, 1); - NFSUNLOCKV4ROOTMUTEX(); - error = NFSERR_PERM; - goto out; + if (vp != NULL) { + NFSVOPLOCK(vp, lktype | LK_RETRY); + if ((vp->v_iflag & VI_DOOMED) != 0) { + *haslockp = 0; + NFSLOCKV4ROOTMUTEX(); + nfsv4_unlock(&nfsv4rootfs_lock, 1); + NFSUNLOCKV4ROOTMUTEX(); + error = NFSERR_PERM; + goto out; + } } error = -1; goto out; diff --git a/sys/fs/tmpfs/tmpfs_vnops.c b/sys/fs/tmpfs/tmpfs_vnops.c index 29ee389..c811b9a 100644 --- a/sys/fs/tmpfs/tmpfs_vnops.c +++ b/sys/fs/tmpfs/tmpfs_vnops.c @@ -213,10 +213,14 @@ tmpfs_create(struct vop_create_args *v) struct vnode **vpp = v->a_vpp; struct componentname *cnp = v->a_cnp; struct vattr *vap = v->a_vap; + int error; MPASS(vap->va_type == VREG || vap->va_type == VSOCK); - return tmpfs_alloc_file(dvp, vpp, vap, cnp, NULL); + error = tmpfs_alloc_file(dvp, vpp, vap, cnp, NULL); + if (error == 0 && (cnp->cn_flags & MAKEENTRY) != 0) + cache_enter(dvp, *vpp, cnp); + return (error); } static int diff --git a/sys/geom/sched/README b/sys/geom/sched/README index 1b52d90..b62d468 100644 --- a/sys/geom/sched/README +++ b/sys/geom/sched/README @@ -39,37 +39,17 @@ with cvs, and lets cvs progress when competing with a writer. To try it out: -1. USERS OF FREEBSD 7, PLEASE READ CAREFULLY THE FOLLOWING: - - On loading, this module patches one kernel function (g_io_request()) - so that I/O requests ("bio's") carry a classification tag, useful - for scheduling purposes. - - ON FREEBSD 7, the tag is stored in an existing (though rarely used) - field of the "struct bio", a solution which makes this module - incompatible with other modules using it, such as ZFS and gjournal. - Additionally, g_io_request() is patched in-memory to add a call - to the function that initializes this field (i386/amd64 only; - for other architectures you need to manually patch sys/geom/geom_io.c). - See details in the file g_sched.c. - - On FreeBSD 8.0 and above, the above trick is not necessary, - as the struct bio contains dedicated fields for the classifier, - and hooks for request classifiers. - - If you don't like the above, don't run this code. - -2. PLEASE MAKE SURE THAT THE DISK THAT YOU WILL BE USING FOR TESTS +1. PLEASE MAKE SURE THAT THE DISK THAT YOU WILL BE USING FOR TESTS DOES NOT CONTAIN PRECIOUS DATA. This is experimental code, so we make no guarantees, though I am routinely using it on my desktop and laptop. -3. EXTRACT AND BUILD THE PROGRAMS +2. EXTRACT AND BUILD THE PROGRAMS A 'make install' in the directory should work (with root privs), or you can even try the binary modules. If you want to build the modules yourself, look at the Makefile. -4. LOAD THE MODULE, CREATE A GEOM NODE, RUN TESTS +3. LOAD THE MODULE, CREATE A GEOM NODE, RUN TESTS The scheduler's module must be loaded first: diff --git a/sys/geom/sched/g_sched.c b/sys/geom/sched/g_sched.c index 009a58c..f1c9a3d 100644 --- a/sys/geom/sched/g_sched.c +++ b/sys/geom/sched/g_sched.c @@ -346,17 +346,8 @@ static inline u_long g_sched_classify(struct bio *bp) { -#if __FreeBSD_version > 800098 /* we have classifier fields in the struct bio */ -#define HAVE_BIO_CLASSIFIER return ((u_long)bp->bio_classifier1); -#else -#warning old version!!! - while (bp->bio_parent != NULL) - bp = bp->bio_parent; - - return ((u_long)bp->bio_caller1); -#endif } /* Return the hash chain for the given key. */ @@ -705,7 +696,7 @@ g_gsched_global_init(void) G_SCHED_DEBUG(0, "Initializing global data."); mtx_init(&me.gs_mtx, "gsched", NULL, MTX_DEF); LIST_INIT(&me.gs_scheds); - gs_bioq_init(&me.gs_pending); + bioq_init(&me.gs_pending); me.gs_initialized = 1; } } @@ -914,7 +905,7 @@ g_sched_temporary_start(struct bio *bio) mtx_lock(&me.gs_mtx); me.gs_npending++; - gs_bioq_disksort(&me.gs_pending, bio); + bioq_disksort(&me.gs_pending, bio); mtx_unlock(&me.gs_mtx); } @@ -923,7 +914,7 @@ g_sched_flush_pending(g_start_t *start) { struct bio *bp; - while ((bp = gs_bioq_takefirst(&me.gs_pending))) + while ((bp = bioq_takefirst(&me.gs_pending))) start(bp); } @@ -1365,161 +1356,7 @@ g_sched_destroy_geom(struct gctl_req *req, struct g_class *mp, * to the issuer of a request in bp->bio_classifier1 as soon * as the bio is posted to the geom queue (and not later, because * requests are managed by the g_down thread afterwards). - * - * On older versions of the system (but this code is not used - * in any existing release), we [ab]use the caller1 field in the - * root element of the bio tree to store the classification info. - * The marking is done at the beginning of g_io_request() - * and only if we find that the field is NULL. - * - * To avoid rebuilding the kernel, this module will patch the - * initial part of g_io_request() so it jumps to some hand-coded - * assembly that does the marking and then executes the original - * body of g_io_request(). - * - * fake_ioreq[] is architecture-specific machine code - * that implements the above. CODE_SIZE, STORE_SIZE etc. - * are constants used in the patching routine. Look at the - * code in g_ioreq_patch() for the details. - */ - -#ifndef HAVE_BIO_CLASSIFIER -/* - * Support for old FreeBSD versions */ -#if defined(__i386__) -#define CODE_SIZE 29 -#define STORE_SIZE 5 -#define EPILOGUE 5 -#define SIZE (CODE_SIZE + STORE_SIZE + EPILOGUE) - -static u_char fake_ioreq[SIZE] = { - 0x8b, 0x44, 0x24, 0x04, /* mov bp, %eax */ - /* 1: */ - 0x89, 0xc2, /* mov %eax, %edx # edx = bp */ - 0x8b, 0x40, 0x64, /* mov bp->bio_parent, %eax */ - 0x85, 0xc0, /* test %eax, %eax */ - 0x75, 0xf7, /* jne 1b */ - 0x8b, 0x42, 0x30, /* mov bp->bp_caller1, %eax */ - 0x85, 0xc0, /* test %eax, %eax */ - 0x75, 0x09, /* jne 2f */ - 0x64, 0xa1, 0x00, 0x00, /* mov %fs:0, %eax */ - 0x00, 0x00, - 0x89, 0x42, 0x30, /* mov %eax, bp->bio_caller1 */ - /* 2: */ - 0x55, 0x89, 0xe5, 0x57, 0x56, - 0xe9, 0x00, 0x00, 0x00, 0x00, /* jmp back... */ -}; -#elif defined(__amd64) -#define CODE_SIZE 38 -#define STORE_SIZE 6 -#define EPILOGUE 5 -#define SIZE (CODE_SIZE + STORE_SIZE + EPILOGUE) - -static u_char fake_ioreq[SIZE] = { - 0x48, 0x89, 0xf8, /* mov bp, %rax */ - /* 1: */ - 0x48, 0x89, 0xc2, /* mov %rax, %rdx # rdx = bp */ - 0x48, 0x8b, 0x82, 0xa8, /* mov bp->bio_parent, %rax */ - 0x00, 0x00, 0x00, - 0x48, 0x85, 0xc0, /* test %rax, %rax */ - 0x75, 0xf1, /* jne 1b */ - 0x48, 0x83, 0x7a, 0x58, /* cmp $0, bp->bp_caller1 */ - 0x00, - 0x75, 0x0d, /* jne 2f */ - 0x65, 0x48, 0x8b, 0x04, /* mov %gs:0, %rax */ - 0x25, 0x00, 0x00, 0x00, - 0x00, - 0x48, 0x89, 0x42, 0x58, /* mov %rax, bp->bio_caller1 */ - /* 2: */ - 0x55, 0x48, 0x89, 0xe5, 0x41, 0x56, - 0xe9, 0x00, 0x00, 0x00, 0x00, /* jmp back... */ -}; -#else /* neither x86 nor amd64 */ -static void -g_new_io_request(struct bio *bp, struct g_consumer *cp) -{ - struct bio *top = bp; - - /* - * bio classification: if bio_caller1 is available in the - * root of the 'struct bio' tree, store there the thread id - * of the thread that originated the request. - * More sophisticated classification schemes can be used. - */ - while (top->bio_parent) - top = top->bio_parent; - - if (top->bio_caller1 == NULL) - top->bio_caller1 = curthread; -} - -#error please add the code above in g_new_io_request() to the beginning of \ - /sys/geom/geom_io.c::g_io_request(), and remove this line. -#endif /* end of arch-specific code */ - -static int -g_ioreq_patch(void) -{ - u_char *original; - u_long ofs; - int found; - - if (me.gs_patched) - return (-1); - - original = (u_char *)g_io_request; - - found = !bcmp(original, fake_ioreq + CODE_SIZE, STORE_SIZE); - if (!found) - return (-1); - - /* Jump back to the original + STORE_SIZE. */ - ofs = (original + STORE_SIZE) - (fake_ioreq + SIZE); - bcopy(&ofs, fake_ioreq + CODE_SIZE + STORE_SIZE + 1, 4); - - /* Patch the original address with a jump to the trampoline. */ - *original = 0xe9; /* jump opcode */ - ofs = fake_ioreq - (original + 5); - bcopy(&ofs, original + 1, 4); - - me.gs_patched = 1; - - return (0); -} - -/* - * Restore the original code, this is easy. - */ -static void -g_ioreq_restore(void) -{ - u_char *original; - - if (me.gs_patched) { - original = (u_char *)g_io_request; - bcopy(fake_ioreq + CODE_SIZE, original, STORE_SIZE); - me.gs_patched = 0; - } -} - -static inline void -g_classifier_ini(void) -{ - - g_ioreq_patch(); -} - -static inline void -g_classifier_fini(void) -{ - - g_ioreq_restore(); -} - -/*--- end of support code for older FreeBSD versions */ - -#else /* HAVE_BIO_CLASSIFIER */ /* * Classifier support for recent FreeBSD versions: we use @@ -1552,7 +1389,6 @@ g_classifier_fini(void) g_unregister_classifier(&g_sched_classifier); } -#endif /* HAVE_BIO_CLASSIFIER */ static void g_sched_init(struct g_class *mp) diff --git a/sys/geom/sched/g_sched.h b/sys/geom/sched/g_sched.h index 3a34e29..9fdadc4 100644 --- a/sys/geom/sched/g_sched.h +++ b/sys/geom/sched/g_sched.h @@ -120,19 +120,6 @@ struct g_sched_softc { #define G_SCHED_PROXYING 1 #define G_SCHED_FLUSHING 2 -/* - * Temporary- our own version of the disksort, because the - * version in 7.x and 8.x before march 2009 is buggy. - */ -void gs_bioq_init(struct bio_queue_head *); -void gs_bioq_remove(struct bio_queue_head *, struct bio *); -void gs_bioq_flush(struct bio_queue_head *, struct devstat *, int); -void gs_bioq_insert_head(struct bio_queue_head *, struct bio *); -void gs_bioq_insert_tail(struct bio_queue_head *, struct bio *); -struct bio *gs_bioq_first(struct bio_queue_head *); -struct bio *gs_bioq_takefirst(struct bio_queue_head *); -void gs_bioq_disksort(struct bio_queue_head *, struct bio *); - #endif /* _KERNEL */ #endif /* _G_SCHED_H_ */ diff --git a/sys/geom/sched/gs_rr.c b/sys/geom/sched/gs_rr.c index 6f26879..b9d5d1b 100644 --- a/sys/geom/sched/gs_rr.c +++ b/sys/geom/sched/gs_rr.c @@ -315,7 +315,7 @@ g_rr_init_class(void *data, void *priv) struct g_rr_softc *sc = data; struct g_rr_queue *qp = priv; - gs_bioq_init(&qp->q_bioq); + bioq_init(&qp->q_bioq); /* * Set the initial parameters for the client: @@ -350,7 +350,7 @@ g_rr_fini_class(void *data, void *priv) { struct g_rr_queue *qp = priv; - KASSERT(gs_bioq_first(&qp->q_bioq) == NULL, + KASSERT(bioq_first(&qp->q_bioq) == NULL, ("released nonempty queue")); qp->q_sc->sc_nqueues--; me.queues--; @@ -438,7 +438,7 @@ g_rr_next(void *data, int force) qp->q_flags &= ~G_FLAG_COMPLETED; } - bp = gs_bioq_takefirst(&qp->q_bioq); /* surely not NULL */ + bp = bioq_takefirst(&qp->q_bioq); /* surely not NULL */ qp->q_service += bp->bio_length; /* charge the service */ /* @@ -456,7 +456,7 @@ g_rr_next(void *data, int force) * on read or writes (e.g., anticipate only on reads). */ expired = g_rr_queue_expired(qp); /* are we expired ? */ - next = gs_bioq_first(&qp->q_bioq); /* do we have one more ? */ + next = bioq_first(&qp->q_bioq); /* do we have one more ? */ if (expired) { sc->sc_active = NULL; /* Either requeue or release reference. */ @@ -538,7 +538,7 @@ g_rr_start(void *data, struct bio *bp) if (qp == NULL) return (-1); /* allocation failed, tell upstream */ - if (gs_bioq_first(&qp->q_bioq) == NULL) { + if (bioq_first(&qp->q_bioq) == NULL) { /* * We are inserting into an empty queue. * Reset its state if it is sc_active, @@ -560,7 +560,7 @@ g_rr_start(void *data, struct bio *bp) /* Inherit the reference returned by g_rr_queue_get(). */ bp->bio_caller1 = qp; - gs_bioq_disksort(&qp->q_bioq, bp); + bioq_disksort(&qp->q_bioq, bp); return (0); } diff --git a/sys/geom/sched/subr_disk.c b/sys/geom/sched/subr_disk.c deleted file mode 100644 index db2a9ef..0000000 --- a/sys/geom/sched/subr_disk.c +++ /dev/null @@ -1,226 +0,0 @@ -/*- - * ---------------------------------------------------------------------------- - * "THE BEER-WARE LICENSE" (Revision 42): - * <phk@FreeBSD.ORG> wrote this file. As long as you retain this notice you - * can do whatever you want with this stuff. If we meet some day, and you think - * this stuff is worth it, you can buy me a beer in return. Poul-Henning Kamp - * ---------------------------------------------------------------------------- - * - * The bioq_disksort() (and the specification of the bioq API) - * have been written by Luigi Rizzo and Fabio Checconi under the same - * license as above. - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -//#include "opt_geom.h" - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/bio.h> -#include <sys/conf.h> -#include <sys/disk.h> -#include <geom/geom_disk.h> -#include "g_sched.h" - -/* - * BIO queue implementation - * - * Please read carefully the description below before making any change - * to the code, or you might change the behaviour of the data structure - * in undesirable ways. - * - * A bioq stores disk I/O request (bio), normally sorted according to - * the distance of the requested position (bio->bio_offset) from the - * current head position (bioq->last_offset) in the scan direction, i.e. - * - * (uoff_t)(bio_offset - last_offset) - * - * Note that the cast to unsigned (uoff_t) is fundamental to insure - * that the distance is computed in the scan direction. - * - * The main methods for manipulating the bioq are: - * - * bioq_disksort() performs an ordered insertion; - * - * bioq_first() return the head of the queue, without removing; - * - * bioq_takefirst() return and remove the head of the queue, - * updating the 'current head position' as - * bioq->last_offset = bio->bio_offset + bio->bio_length; - * - * When updating the 'current head position', we assume that the result of - * bioq_takefirst() is dispatched to the device, so bioq->last_offset - * represents the head position once the request is complete. - * - * If the bioq is manipulated using only the above calls, it starts - * with a sorted sequence of requests with bio_offset >= last_offset, - * possibly followed by another sorted sequence of requests with - * 0 <= bio_offset < bioq->last_offset - * - * NOTE: historical behaviour was to ignore bio->bio_length in the - * update, but its use tracks the head position in a better way. - * Historical behaviour was also to update the head position when - * the request under service is complete, rather than when the - * request is extracted from the queue. However, the current API - * has no method to update the head position; secondly, once - * a request has been submitted to the disk, we have no idea of - * the actual head position, so the final one is our best guess. - * - * --- Direct queue manipulation --- - * - * A bioq uses an underlying TAILQ to store requests, so we also - * export methods to manipulate the TAILQ, in particular: - * - * bioq_insert_tail() insert an entry at the end. - * It also creates a 'barrier' so all subsequent - * insertions through bioq_disksort() will end up - * after this entry; - * - * bioq_insert_head() insert an entry at the head, update - * bioq->last_offset = bio->bio_offset so that - * all subsequent insertions through bioq_disksort() - * will end up after this entry; - * - * bioq_remove() remove a generic element from the queue, act as - * bioq_takefirst() if invoked on the head of the queue. - * - * The semantic of these methods is the same as the operations - * on the underlying TAILQ, but with additional guarantees on - * subsequent bioq_disksort() calls. E.g. bioq_insert_tail() - * can be useful for making sure that all previous ops are flushed - * to disk before continuing. - * - * Updating bioq->last_offset on a bioq_insert_head() guarantees - * that the bio inserted with the last bioq_insert_head() will stay - * at the head of the queue even after subsequent bioq_disksort(). - * - * Note that when the direct queue manipulation functions are used, - * the queue may contain multiple inversion points (i.e. more than - * two sorted sequences of requests). - * - */ - -void -gs_bioq_init(struct bio_queue_head *head) -{ - - TAILQ_INIT(&head->queue); - head->last_offset = 0; - head->insert_point = NULL; -} - -void -gs_bioq_remove(struct bio_queue_head *head, struct bio *bp) -{ - - if (head->insert_point == NULL) { - if (bp == TAILQ_FIRST(&head->queue)) - head->last_offset = bp->bio_offset + bp->bio_length; - } else if (bp == head->insert_point) - head->insert_point = NULL; - - TAILQ_REMOVE(&head->queue, bp, bio_queue); -} - -void -gs_bioq_flush(struct bio_queue_head *head, struct devstat *stp, int error) -{ - struct bio *bp; - - while ((bp = gs_bioq_takefirst(head)) != NULL) - biofinish(bp, stp, error); -} - -void -gs_bioq_insert_head(struct bio_queue_head *head, struct bio *bp) -{ - - if (head->insert_point == NULL) - head->last_offset = bp->bio_offset; - TAILQ_INSERT_HEAD(&head->queue, bp, bio_queue); -} - -void -gs_bioq_insert_tail(struct bio_queue_head *head, struct bio *bp) -{ - - TAILQ_INSERT_TAIL(&head->queue, bp, bio_queue); - head->insert_point = bp; - head->last_offset = bp->bio_offset; -} - -struct bio * -gs_bioq_first(struct bio_queue_head *head) -{ - - return (TAILQ_FIRST(&head->queue)); -} - -struct bio * -gs_bioq_takefirst(struct bio_queue_head *head) -{ - struct bio *bp; - - bp = TAILQ_FIRST(&head->queue); - if (bp != NULL) - gs_bioq_remove(head, bp); - return (bp); -} - -/* - * Compute the sorting key. The cast to unsigned is - * fundamental for correctness, see the description - * near the beginning of the file. - */ -static inline uoff_t -gs_bioq_bio_key(struct bio_queue_head *head, struct bio *bp) -{ - - return ((uoff_t)(bp->bio_offset - head->last_offset)); -} - -/* - * Seek sort for disks. - * - * Sort all requests in a single queue while keeping - * track of the current position of the disk with last_offset. - * See above for details. - */ -void -gs_bioq_disksort(struct bio_queue_head *head, struct bio *bp) -{ - struct bio *cur, *prev; - uoff_t key; - - if ((bp->bio_flags & BIO_ORDERED) != 0) { - /* - * Ordered transactions can only be dispatched - * after any currently queued transactions. They - * also have barrier semantics - no transactions - * queued in the future can pass them. - */ - gs_bioq_insert_tail(head, bp); - return; - } - - prev = NULL; - key = gs_bioq_bio_key(head, bp); - cur = TAILQ_FIRST(&head->queue); - - if (head->insert_point) { - prev = head->insert_point; - cur = TAILQ_NEXT(head->insert_point, bio_queue); - } - - while (cur != NULL && key >= gs_bioq_bio_key(head, cur)) { - prev = cur; - cur = TAILQ_NEXT(cur, bio_queue); - } - - if (prev == NULL) - TAILQ_INSERT_HEAD(&head->queue, bp, bio_queue); - else - TAILQ_INSERT_AFTER(&head->queue, prev, bp, bio_queue); -} diff --git a/sys/i386/i386/trap.c b/sys/i386/i386/trap.c index 490d481..6d35c3a 100644 --- a/sys/i386/i386/trap.c +++ b/sys/i386/i386/trap.c @@ -246,7 +246,7 @@ trap(struct trapframe *frame) * flag is cleared and finally re-scheduling is enabled. */ if ((type == T_PROTFLT || type == T_PAGEFLT) && - dtrace_trap_func != NULL && (*dtrace_trap_func)(frame)) + dtrace_trap_func != NULL && (*dtrace_trap_func)(frame, type)) goto out; #endif diff --git a/sys/i386/xen/clock.c b/sys/i386/xen/clock.c index 0f5b0e5..26fafee 100644 --- a/sys/i386/xen/clock.c +++ b/sys/i386/xen/clock.c @@ -118,7 +118,7 @@ struct mtx clock_lock; #define RTC_UNLOCK mtx_unlock_spin(&clock_lock) #define NS_PER_TICK (1000000000ULL/hz) -int adjkerntz; /* local offset from GMT in seconds */ +int adjkerntz; /* local offset from UTC in seconds */ int clkintr_pending; int pscnt = 1; int psdiv = 1; diff --git a/sys/kern/kern_sig.c b/sys/kern/kern_sig.c index b34eb62..a4f0f88 100644 --- a/sys/kern/kern_sig.c +++ b/sys/kern/kern_sig.c @@ -3167,7 +3167,8 @@ corefile_open(const char *comm, uid_t uid, pid_t pid, struct thread *td, sbuf_delete(&sb); cmode = S_IRUSR | S_IWUSR; - oflags = VN_OPEN_NOAUDIT | (capmode_coredump ? VN_OPEN_NOCAPCHECK : 0); + oflags = VN_OPEN_NOAUDIT | VN_OPEN_NAMECACHE | + (capmode_coredump ? VN_OPEN_NOCAPCHECK : 0); /* * If the core format has a %I in it, then we need to check diff --git a/sys/kern/subr_fattime.c b/sys/kern/subr_fattime.c index e2a2195..099479c 100644 --- a/sys/kern/subr_fattime.c +++ b/sys/kern/subr_fattime.c @@ -45,7 +45,7 @@ * Later on again, in Windows NT, timestamps were defined relative to GMT. * * Purists will point out that UTC replaced GMT for such uses around - * a century ago, already then. Ironically "NT" was an abbreviation of + * half a century ago, already then. Ironically "NT" was an abbreviation of * "New Technology". Anyway... * * The 'utc' argument determines if the resulting FATTIME timestamp diff --git a/sys/kern/uipc_sockbuf.c b/sys/kern/uipc_sockbuf.c index 537f9c8..88952ed 100644 --- a/sys/kern/uipc_sockbuf.c +++ b/sys/kern/uipc_sockbuf.c @@ -579,7 +579,7 @@ sbappend_locked(struct sockbuf *sb, struct mbuf *m) if (m == 0) return; - + m_clrprotoflags(m); SBLASTRECORDCHK(sb); n = sb->sb_mb; if (n) { @@ -732,6 +732,7 @@ sbappendrecord_locked(struct sockbuf *sb, struct mbuf *m0) if (m0 == 0) return; + m_clrprotoflags(m0); /* * Put the first mbuf on the queue. Note this permits zero length * records. @@ -777,6 +778,8 @@ sbappendaddr_locked_internal(struct sockbuf *sb, const struct sockaddr *asa, return (0); m->m_len = asa->sa_len; bcopy(asa, mtod(m, caddr_t), asa->sa_len); + if (m0) + m_clrprotoflags(m0); if (ctrl_last) ctrl_last->m_next = m0; /* concatenate data to control */ else @@ -872,6 +875,7 @@ sbappendcontrol_locked(struct sockbuf *sb, struct mbuf *m0, if (space > sbspace(sb)) return (0); + m_clrprotoflags(m0); n->m_next = m0; /* concatenate data to control */ SBLASTRECORDCHK(sb); @@ -1062,6 +1066,21 @@ sbcut_internal(struct sockbuf *sb, int len) m = n; } } + /* + * Free any zero-length mbufs from the buffer. + * For SOCK_DGRAM sockets such mbufs represent empty records. + * XXX: For SOCK_STREAM sockets such mbufs can appear in the buffer, + * when sosend_generic() needs to send only control data. + */ + while (m && m->m_len == 0) { + struct mbuf *n; + + sbfree(sb, m); + n = m->m_next; + m->m_next = mfree; + mfree = m; + m = n; + } if (m) { sb->sb_mb = m; m->m_nextpkt = next; diff --git a/sys/kern/uipc_socket.c b/sys/kern/uipc_socket.c index f08036d..189a30f 100644 --- a/sys/kern/uipc_socket.c +++ b/sys/kern/uipc_socket.c @@ -1310,11 +1310,14 @@ restart: resid = 0; if (flags & MSG_EOR) top->m_flags |= M_EOR; - } else if (resid > 0) { + } else { /* * Copy the data from userland into a mbuf - * chain. If no data is to be copied in, - * a single empty mbuf is returned. + * chain. If resid is 0, which can happen + * only if we have control to send, then + * a single empty mbuf is returned. This + * is a workaround to prevent protocol send + * methods to panic. */ top = m_uiotombuf(uio, M_WAITOK, space, (atomic ? max_hdr : 0), diff --git a/sys/kern/vfs_default.c b/sys/kern/vfs_default.c index a34be50..d6b3765 100644 --- a/sys/kern/vfs_default.c +++ b/sys/kern/vfs_default.c @@ -401,17 +401,24 @@ int vop_stdadvlock(struct vop_advlock_args *ap) { struct vnode *vp; - struct ucred *cred; struct vattr vattr; int error; vp = ap->a_vp; - cred = curthread->td_ucred; - vn_lock(vp, LK_SHARED | LK_RETRY); - error = VOP_GETATTR(vp, &vattr, cred); - VOP_UNLOCK(vp, 0); - if (error) - return (error); + if (ap->a_fl->l_whence == SEEK_END) { + /* + * The NFSv4 server must avoid doing a vn_lock() here, since it + * can deadlock the nfsd threads, due to a LOR. Fortunately + * the NFSv4 server always uses SEEK_SET and this code is + * only required for the SEEK_END case. + */ + vn_lock(vp, LK_SHARED | LK_RETRY); + error = VOP_GETATTR(vp, &vattr, curthread->td_ucred); + VOP_UNLOCK(vp, 0); + if (error) + return (error); + } else + vattr.va_size = 0; return (lf_advlock(ap, &(vp->v_lockf), vattr.va_size)); } @@ -420,17 +427,19 @@ int vop_stdadvlockasync(struct vop_advlockasync_args *ap) { struct vnode *vp; - struct ucred *cred; struct vattr vattr; int error; vp = ap->a_vp; - cred = curthread->td_ucred; - vn_lock(vp, LK_SHARED | LK_RETRY); - error = VOP_GETATTR(vp, &vattr, cred); - VOP_UNLOCK(vp, 0); - if (error) - return (error); + if (ap->a_fl->l_whence == SEEK_END) { + /* The size argument is only needed for SEEK_END. */ + vn_lock(vp, LK_SHARED | LK_RETRY); + error = VOP_GETATTR(vp, &vattr, curthread->td_ucred); + VOP_UNLOCK(vp, 0); + if (error) + return (error); + } else + vattr.va_size = 0; return (lf_advlockasync(ap, &(vp->v_lockf), vattr.va_size)); } diff --git a/sys/kern/vfs_vnops.c b/sys/kern/vfs_vnops.c index 40a69bb..ed4ad4d 100644 --- a/sys/kern/vfs_vnops.c +++ b/sys/kern/vfs_vnops.c @@ -217,6 +217,8 @@ restart: return (error); goto restart; } + if ((vn_open_flags & VN_OPEN_NAMECACHE) != 0) + ndp->ni_cnd.cn_flags |= MAKEENTRY; #ifdef MAC error = mac_vnode_check_create(cred, ndp->ni_dvp, &ndp->ni_cnd, vap); diff --git a/sys/mips/beri/beri_machdep.c b/sys/mips/beri/beri_machdep.c index 714ee3d..e3cca28 100644 --- a/sys/mips/beri/beri_machdep.c +++ b/sys/mips/beri/beri_machdep.c @@ -117,13 +117,13 @@ mips_init(void) ("First region is not within FDT memory range")); /* Limit size of the first region */ - phys_avail[1] = MIN(mr[0].mr_size, ctob(realmem)); + phys_avail[1] = (mr[0].mr_start + MIN(mr[0].mr_size, ctob(realmem))); dump_avail[1] = phys_avail[1]; /* Add the rest of regions */ for (i = 1, j = 2; i < mr_cnt; i++, j+=2) { phys_avail[j] = mr[i].mr_start; - phys_avail[j+1] = mr[i].mr_size; + phys_avail[j+1] = (mr[i].mr_start + mr[i].mr_size); dump_avail[j] = phys_avail[j]; dump_avail[j+1] = phys_avail[j+1]; } diff --git a/sys/mips/mips/trap.c b/sys/mips/mips/trap.c index a4405ae..98fe812 100644 --- a/sys/mips/mips/trap.c +++ b/sys/mips/mips/trap.c @@ -617,7 +617,8 @@ trap(struct trapframe *trapframe) * XXXDTRACE: add pid probe handler here (if ever) */ if (!usermode) { - if (dtrace_trap_func != NULL && (*dtrace_trap_func)(trapframe)) + if (dtrace_trap_func != NULL && + (*dtrace_trap_func)(trapframe, type) != 0) return (trapframe->pc); } #endif diff --git a/sys/modules/Makefile b/sys/modules/Makefile index 54533f2..0b1252d 100644 --- a/sys/modules/Makefile +++ b/sys/modules/Makefile @@ -248,14 +248,11 @@ SUBDIR= \ netfpga10g \ ${_netgraph} \ ${_nfe} \ - nfs_common \ nfscl \ - nfsclient \ nfscommon \ nfsd \ nfslock \ nfslockd \ - nfsserver \ nfssvc \ nge \ nmdm \ diff --git a/sys/modules/cryptodev/Makefile b/sys/modules/cryptodev/Makefile index a82517d..cc5ca12 100644 --- a/sys/modules/cryptodev/Makefile +++ b/sys/modules/cryptodev/Makefile @@ -3,6 +3,6 @@ .PATH: ${.CURDIR}/../../opencrypto KMOD = cryptodev SRCS = cryptodev.c -SRCS += bus_if.h device_if.h opt_compat.h +SRCS += bus_if.h device_if.h opt_compat.h opt_kdtrace.h .include <bsd.kmod.mk> diff --git a/sys/modules/dtrace/Makefile b/sys/modules/dtrace/Makefile index b84c632..08b6937 100644 --- a/sys/modules/dtrace/Makefile +++ b/sys/modules/dtrace/Makefile @@ -4,7 +4,6 @@ SUBDIR= dtmalloc \ dtnfscl \ - dtnfsclient \ dtrace \ dtraceall \ dtrace_test \ diff --git a/sys/modules/dtrace/dtnfsclient/Makefile b/sys/modules/dtrace/dtnfsclient/Makefile deleted file mode 100644 index 1c5208f..0000000 --- a/sys/modules/dtrace/dtnfsclient/Makefile +++ /dev/null @@ -1,17 +0,0 @@ -# $FreeBSD$ - -SYSDIR?= ${.CURDIR}/../../.. - -.PATH: ${SYSDIR}/nfsclient - -KMOD= dtnfsclient -SRCS= nfs_kdtrace.c -SRCS+= vnode_if.h - -CFLAGS+= -I${SYSDIR}/cddl/compat/opensolaris \ - -I${SYSDIR}/cddl/contrib/opensolaris/uts/common \ - -I${SYSDIR} - -.include <bsd.kmod.mk> - -CFLAGS+= -include ${SYSDIR}/cddl/compat/opensolaris/sys/debug_compat.h diff --git a/sys/modules/dtrace/dtraceall/dtraceall.c b/sys/modules/dtrace/dtraceall/dtraceall.c index eda740e..c144d06 100644 --- a/sys/modules/dtrace/dtraceall/dtraceall.c +++ b/sys/modules/dtrace/dtraceall/dtraceall.c @@ -69,9 +69,6 @@ MODULE_DEPEND(dtraceall, dtmalloc, 1, 1, 1); #if defined(NFSCLIENT) MODULE_DEPEND(dtraceall, dtnfscl, 1, 1, 1); #endif -#if defined(NFSCLIENT) -MODULE_DEPEND(dtraceall, dtnfsclient, 1, 1, 1); -#endif #if defined(__amd64__) || defined(__i386__) || defined(__powerpc__) MODULE_DEPEND(dtraceall, fbt, 1, 1, 1); #endif diff --git a/sys/modules/geom/geom_sched/gs_sched/Makefile b/sys/modules/geom/geom_sched/gs_sched/Makefile index 5739365..13bb91b 100644 --- a/sys/modules/geom/geom_sched/gs_sched/Makefile +++ b/sys/modules/geom/geom_sched/gs_sched/Makefile @@ -1,6 +1,6 @@ # $FreeBSD$ KMOD= geom_sched -SRCS= g_sched.c subr_disk.c +SRCS= g_sched.c # ../Makefile.inc automatically included .include <bsd.kmod.mk> diff --git a/sys/modules/geom/geom_sched/gsched_rr/Makefile b/sys/modules/geom/geom_sched/gsched_rr/Makefile index 4209277..44cc566 100644 --- a/sys/modules/geom/geom_sched/gsched_rr/Makefile +++ b/sys/modules/geom/geom_sched/gsched_rr/Makefile @@ -2,8 +2,6 @@ KMOD= gsched_rr SRCS= gs_rr.c -# hash.h on 6.x has a (char *) cast on a const pointer -#CWARNFLAGS= # ../Makefile.inc automatically included .include <bsd.kmod.mk> diff --git a/sys/modules/if_gif/Makefile b/sys/modules/if_gif/Makefile index 663c679..3b3a172 100644 --- a/sys/modules/if_gif/Makefile +++ b/sys/modules/if_gif/Makefile @@ -6,7 +6,11 @@ SYSDIR?=${.CURDIR}/../.. .PATH: ${SYSDIR}/net ${SYSDIR}/netinet ${SYSDIR}/netinet6 KMOD= if_gif -SRCS= if_gif.c in_gif.c opt_inet.h opt_inet6.h +SRCS= if_gif.c opt_inet.h opt_inet6.h + +.if ${MK_INET_SUPPORT} != "no" +SRCS+= in_gif.c +.endif .if ${MK_INET6_SUPPORT} != "no" SRCS+= in6_gif.c diff --git a/sys/modules/nfs_common/Makefile b/sys/modules/nfs_common/Makefile deleted file mode 100644 index 8f51f18..0000000 --- a/sys/modules/nfs_common/Makefile +++ /dev/null @@ -1,8 +0,0 @@ -# $FreeBSD$ - -.PATH: ${.CURDIR}/../../nfs - -KMOD= nfs_common -SRCS= nfs_common.c opt_nfs.h vnode_if.h - -.include <bsd.kmod.mk> diff --git a/sys/modules/nfsclient/Makefile b/sys/modules/nfsclient/Makefile deleted file mode 100644 index a015f34..0000000 --- a/sys/modules/nfsclient/Makefile +++ /dev/null @@ -1,32 +0,0 @@ -# $FreeBSD$ - -.PATH: ${.CURDIR}/../../nfsclient ${.CURDIR}/../../nfs ${.CURDIR}/../../rpc - -KMOD= nfsclient -SRCS= vnode_if.h \ - nfs_bio.c nfs_node.c nfs_subs.c nfs_nfsiod.c \ - nfs_vfsops.c nfs_vnops.c nfs_krpc.c \ - opt_inet.h opt_nfs.h opt_bootp.h opt_nfsroot.h -SRCS+= opt_inet6.h opt_kgssapi.h - -.if !defined(KERNBUILDDIR) -NFS_ROOT?= 1 # 0/1 - requires NFS_ROOT to be configured in kernel - -.if ${NFS_ROOT} > 0 -opt_nfsroot.h: - echo "#define NFS_ROOT 1" > ${.TARGET} -.endif -.else -OPT_NFS_ROOT!= cat ${KERNBUILDDIR}/opt_nfsroot.h -.if empty(OPT_NFS_ROOT) -NFS_ROOT= 0 -.else -NFS_ROOT= 1 -.endif -.endif - -.if ${NFS_ROOT} > 0 -SRCS+= nfs_diskless.c -.endif - -.include <bsd.kmod.mk> diff --git a/sys/modules/nfsserver/Makefile b/sys/modules/nfsserver/Makefile deleted file mode 100644 index d00fe47..0000000 --- a/sys/modules/nfsserver/Makefile +++ /dev/null @@ -1,12 +0,0 @@ -# $FreeBSD$ - -.PATH: ${.CURDIR}/../../nfsserver ${.CURDIR}/../../nfs -KMOD= nfsserver -SRCS= vnode_if.h \ - nfs_fha.c nfs_fha_old.c nfs_serv.c nfs_srvkrpc.c nfs_srvsubs.c \ - opt_mac.h \ - opt_kgssapi.h \ - opt_nfs.h -SRCS+= opt_inet6.h - -.include <bsd.kmod.mk> diff --git a/sys/net/if_gif.c b/sys/net/if_gif.c index 674060dd..0a4139f 100644 --- a/sys/net/if_gif.c +++ b/sys/net/if_gif.c @@ -72,7 +72,6 @@ __FBSDID("$FreeBSD$"); #include <netinet/ip_ecn.h> #ifdef INET #include <netinet/in_var.h> -#include <netinet/in_gif.h> #include <netinet/ip_var.h> #endif /* INET */ @@ -85,7 +84,6 @@ __FBSDID("$FreeBSD$"); #include <netinet6/ip6_ecn.h> #include <netinet6/ip6_var.h> #include <netinet6/scope6_var.h> -#include <netinet6/in6_gif.h> #include <netinet6/ip6protosw.h> #endif /* INET6 */ diff --git a/sys/net/if_gif.h b/sys/net/if_gif.h index b5ebf15..b263850 100644 --- a/sys/net/if_gif.h +++ b/sys/net/if_gif.h @@ -114,6 +114,16 @@ void gif_input(struct mbuf *, struct ifnet *, int, uint8_t); int gif_output(struct ifnet *, struct mbuf *, const struct sockaddr *, struct route *); int gif_encapcheck(const struct mbuf *, int, int, void *); +#ifdef INET +int in_gif_output(struct ifnet *, struct mbuf *, int, uint8_t); +int in_gif_encapcheck(const struct mbuf *, int, int, void *); +int in_gif_attach(struct gif_softc *); +#endif +#ifdef INET6 +int in6_gif_output(struct ifnet *, struct mbuf *, int, uint8_t); +int in6_gif_encapcheck(const struct mbuf *, int, int, void *); +int in6_gif_attach(struct gif_softc *); +#endif #endif /* _KERNEL */ #define GIFGOPTS _IOWR('i', 150, struct ifreq) diff --git a/sys/net/if_gre.c b/sys/net/if_gre.c index 4d00ca2..23e55e2 100644 --- a/sys/net/if_gre.c +++ b/sys/net/if_gre.c @@ -119,16 +119,6 @@ static int gre_set_tunnel(struct ifnet *, struct sockaddr *, struct sockaddr *); static void gre_delete_tunnel(struct ifnet *); -int gre_input(struct mbuf **, int *, int); -#ifdef INET -extern int in_gre_attach(struct gre_softc *); -extern int in_gre_output(struct mbuf *, int, int); -#endif -#ifdef INET6 -extern int in6_gre_attach(struct gre_softc *); -extern int in6_gre_output(struct mbuf *, int, int); -#endif - SYSCTL_DECL(_net_link); static SYSCTL_NODE(_net_link, IFT_TUNNEL, gre, CTLFLAG_RW, 0, "Generic Routing Encapsulation"); diff --git a/sys/net/if_gre.h b/sys/net/if_gre.h index 3a48efe..806b0cb 100644 --- a/sys/net/if_gre.h +++ b/sys/net/if_gre.h @@ -100,6 +100,15 @@ struct gre_softc { #define gre_oip gre_gihdr->gi_ip #define gre_oip6 gre_gi6hdr->gi6_ip6 +int gre_input(struct mbuf **, int *, int); +#ifdef INET +int in_gre_attach(struct gre_softc *); +int in_gre_output(struct mbuf *, int, int); +#endif +#ifdef INET6 +int in6_gre_attach(struct gre_softc *); +int in6_gre_output(struct mbuf *, int, int); +#endif /* * CISCO uses special type for GRE tunnel created as part of WCCP * connection, while in fact those packets are just IPv4 encapsulated diff --git a/sys/net/if_stf.c b/sys/net/if_stf.c index f3a6042..aed2a37 100644 --- a/sys/net/if_stf.c +++ b/sys/net/if_stf.c @@ -99,7 +99,6 @@ #include <net/route.h> #include <net/netisr.h> #include <net/if_types.h> -#include <net/if_stf.h> #include <net/vnet.h> #include <netinet/in.h> @@ -156,7 +155,8 @@ static MALLOC_DEFINE(M_STF, stfname, "6to4 Tunnel Interface"); static const int ip_stf_ttl = 40; extern struct domain inetdomain; -struct protosw in_stf_protosw = { +static int in_stf_input(struct mbuf **, int *, int); +static struct protosw in_stf_protosw = { .pr_type = SOCK_RAW, .pr_domain = &inetdomain, .pr_protocol = IPPROTO_IPV6, @@ -620,7 +620,7 @@ stf_checkaddr6(sc, in6, inifp) return 0; } -int +static int in_stf_input(struct mbuf **mp, int *offp, int proto) { struct stf_softc *sc; diff --git a/sys/net/if_stf.h b/sys/net/if_stf.h deleted file mode 100644 index 07f585b..0000000 --- a/sys/net/if_stf.h +++ /dev/null @@ -1,38 +0,0 @@ -/* $FreeBSD$ */ -/* $KAME: if_stf.h,v 1.5 2001/10/12 10:09:17 keiichi Exp $ */ - -/*- - * Copyright (C) 2000 WIDE Project. - * All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 3. Neither the name of the project nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - */ - -#ifndef _NET_IF_STF_H_ -#define _NET_IF_STF_H_ - -int in_stf_input(struct mbuf **, int *, int); - -#endif /* _NET_IF_STF_H_ */ diff --git a/sys/net80211/ieee80211_output.c b/sys/net80211/ieee80211_output.c index a8d92f9..8cb1f51 100644 --- a/sys/net80211/ieee80211_output.c +++ b/sys/net80211/ieee80211_output.c @@ -1702,7 +1702,7 @@ ieee80211_add_xrates(uint8_t *frm, const struct ieee80211_rateset *rs) /* * Add an ssid element to a frame. */ -static uint8_t * +uint8_t * ieee80211_add_ssid(uint8_t *frm, const uint8_t *ssid, u_int len) { *frm++ = IEEE80211_ELEMID_SSID; diff --git a/sys/net80211/ieee80211_power.c b/sys/net80211/ieee80211_power.c index 04b3c3a..812cd70 100644 --- a/sys/net80211/ieee80211_power.c +++ b/sys/net80211/ieee80211_power.c @@ -560,10 +560,15 @@ ieee80211_sta_pwrsave(struct ieee80211vap *vap, int enable) * Handle being notified that we have data available for us in a TIM/ATIM. * * This may schedule a transition from _SLEEP -> _RUN if it's appropriate. + * + * In STA mode, we may have put to sleep during scan and need to be dragged + * back out of powersave mode. */ void ieee80211_sta_tim_notify(struct ieee80211vap *vap, int set) { + struct ieee80211com *ic = vap->iv_ic; + /* * Schedule the driver state change. It'll happen at some point soon. * Since the hardware shouldn't know that we're running just yet @@ -574,10 +579,24 @@ ieee80211_sta_tim_notify(struct ieee80211vap *vap, int set) * XXX TODO: verify that the transition to RUN will wake up the * BSS node! */ - IEEE80211_DPRINTF(vap, IEEE80211_MSG_POWER, "%s: TIM=%d\n", __func__, set); IEEE80211_LOCK(vap->iv_ic); if (set == 1 && vap->iv_state == IEEE80211_S_SLEEP) { ieee80211_new_state_locked(vap, IEEE80211_S_RUN, 0); + IEEE80211_DPRINTF(vap, IEEE80211_MSG_POWER, + "%s: TIM=%d; wakeup\n", __func__, set); + } else if ((set == 1) && (ic->ic_flags_ext & IEEE80211_FEXT_BGSCAN)) { + /* + * XXX only do this if we're in RUN state? + */ + IEEE80211_DPRINTF(vap, IEEE80211_MSG_POWER, + "%s: wake up from bgscan vap sleep\n", + __func__); + /* + * We may be in BGSCAN mode - this means the VAP is is in STA + * mode powersave. If it is, we need to wake it up so we + * can process outbound traffic. + */ + vap->iv_sta_ps(vap, 0); } IEEE80211_UNLOCK(vap->iv_ic); } diff --git a/sys/net80211/ieee80211_proto.h b/sys/net80211/ieee80211_proto.h index 8df5116..3b46ac4 100644 --- a/sys/net80211/ieee80211_proto.h +++ b/sys/net80211/ieee80211_proto.h @@ -149,6 +149,7 @@ struct mbuf *ieee80211_alloc_cts(struct ieee80211com *, uint8_t *ieee80211_add_rates(uint8_t *, const struct ieee80211_rateset *); uint8_t *ieee80211_add_xrates(uint8_t *, const struct ieee80211_rateset *); +uint8_t *ieee80211_add_ssid(uint8_t *, const uint8_t *, u_int); uint8_t *ieee80211_add_wpa(uint8_t *, const struct ieee80211vap *); uint8_t *ieee80211_add_rsn(uint8_t *, const struct ieee80211vap *); uint8_t *ieee80211_add_qos(uint8_t *, const struct ieee80211_node *); diff --git a/sys/net80211/ieee80211_scan.c b/sys/net80211/ieee80211_scan.c index 66e8430..d81ba7c 100644 --- a/sys/net80211/ieee80211_scan.c +++ b/sys/net80211/ieee80211_scan.c @@ -698,6 +698,13 @@ ieee80211_cancel_scan(struct ieee80211vap *vap) SCAN_PRIVATE(ss)->ss_iflags |= ISCAN_CANCEL; /* wake up the scan task */ scan_signal(ss); + } else { + IEEE80211_DPRINTF(vap, IEEE80211_MSG_SCAN, + "%s: called; F_SCAN=%d, vap=%s, CANCEL=%d\n", + __func__, + !! (ic->ic_flags & IEEE80211_F_SCAN), + (ss->ss_vap == vap ? "match" : "nomatch"), + !! (SCAN_PRIVATE(ss)->ss_iflags & ISCAN_CANCEL)); } IEEE80211_UNLOCK(ic); } @@ -724,6 +731,13 @@ ieee80211_cancel_anyscan(struct ieee80211vap *vap) SCAN_PRIVATE(ss)->ss_iflags |= ISCAN_CANCEL; /* wake up the scan task */ scan_signal(ss); + } else { + IEEE80211_DPRINTF(vap, IEEE80211_MSG_SCAN, + "%s: called; F_SCAN=%d, vap=%s, CANCEL=%d\n", + __func__, + !! (ic->ic_flags & IEEE80211_F_SCAN), + (ss->ss_vap == vap ? "match" : "nomatch"), + !! (SCAN_PRIVATE(ss)->ss_iflags & ISCAN_CANCEL)); } IEEE80211_UNLOCK(ic); } @@ -738,6 +752,8 @@ ieee80211_scan_next(struct ieee80211vap *vap) struct ieee80211com *ic = vap->iv_ic; struct ieee80211_scan_state *ss = ic->ic_scan; + IEEE80211_DPRINTF(vap, IEEE80211_MSG_SCAN, "%s: called\n", __func__); + /* wake up the scan task */ IEEE80211_LOCK(ic); scan_signal(ss); @@ -754,6 +770,8 @@ ieee80211_scan_done(struct ieee80211vap *vap) struct ieee80211com *ic = vap->iv_ic; struct ieee80211_scan_state *ss; + IEEE80211_DPRINTF(vap, IEEE80211_MSG_SCAN, "%s: called\n", __func__); + IEEE80211_LOCK(ic); ss = ic->ic_scan; ss->ss_next = ss->ss_last; /* all channels are complete */ @@ -807,6 +825,10 @@ scan_curchan(struct ieee80211_scan_state *ss, unsigned long maxdwell) { struct ieee80211vap *vap = ss->ss_vap; + IEEE80211_DPRINTF(vap, IEEE80211_MSG_SCAN, + "%s: calling; maxdwell=%lu\n", + __func__, + maxdwell); IEEE80211_LOCK(vap->iv_ic); if (ss->ss_flags & IEEE80211_SCAN_ACTIVE) ieee80211_probe_curchan(vap, 0); @@ -821,7 +843,6 @@ scan_signal(void *arg) struct ieee80211_scan_state *ss = (struct ieee80211_scan_state *) arg; IEEE80211_LOCK_ASSERT(ss->ss_ic); - cv_signal(&SCAN_PRIVATE(ss)->ss_scan_cv); } @@ -834,6 +855,8 @@ scan_mindwell(struct ieee80211_scan_state *ss) { struct ieee80211com *ic = ss->ss_ic; + IEEE80211_DPRINTF(ss->ss_vap, IEEE80211_MSG_SCAN, "%s: called\n", __func__); + IEEE80211_LOCK(ic); scan_signal(ss); IEEE80211_UNLOCK(ic); @@ -882,13 +905,23 @@ scan_task(void *arg, int pending) } scanend = ticks + SCAN_PRIVATE(ss)->ss_duration; + + /* XXX scan state can change! Re-validate scan state! */ + IEEE80211_UNLOCK(ic); ic->ic_scan_start(ic); /* notify driver */ IEEE80211_LOCK(ic); for (;;) { + scandone = (ss->ss_next >= ss->ss_last) || (SCAN_PRIVATE(ss)->ss_iflags & ISCAN_CANCEL) != 0; + + IEEE80211_DPRINTF(vap, IEEE80211_MSG_SCAN, + "%s: loop start; scandone=%d\n", + __func__, + scandone); + if (scandone || (ss->ss_flags & IEEE80211_SCAN_GOTPICK) || (SCAN_PRIVATE(ss)->ss_iflags & ISCAN_ABORT) || time_after(ticks + ss->ss_mindwell, scanend)) @@ -944,6 +977,8 @@ scan_task(void *arg, int pending) ic->ic_scan_curchan(ss, maxdwell); IEEE80211_LOCK(ic); + /* XXX scan state can change! Re-validate scan state! */ + SCAN_PRIVATE(ss)->ss_chanmindwell = ticks + ss->ss_mindwell; /* clear mindwell lock and initial channel change flush */ SCAN_PRIVATE(ss)->ss_iflags &= ~ISCAN_REP; @@ -951,15 +986,20 @@ scan_task(void *arg, int pending) if ((SCAN_PRIVATE(ss)->ss_iflags & (ISCAN_CANCEL|ISCAN_ABORT))) continue; + IEEE80211_DPRINTF(vap, IEEE80211_MSG_SCAN, "%s: waiting\n", __func__); /* Wait to be signalled to scan the next channel */ cv_wait(&SCAN_PRIVATE(ss)->ss_scan_cv, IEEE80211_LOCK_OBJ(ic)); } + + IEEE80211_DPRINTF(vap, IEEE80211_MSG_SCAN, "%s: out\n", __func__); + if (SCAN_PRIVATE(ss)->ss_iflags & ISCAN_ABORT) goto done; IEEE80211_UNLOCK(ic); ic->ic_scan_end(ic); /* notify driver */ IEEE80211_LOCK(ic); + /* XXX scan state can change! Re-validate scan state! */ /* * Since a cancellation may have occured during one of the @@ -969,10 +1009,10 @@ scan_task(void *arg, int pending) ((SCAN_PRIVATE(ss)->ss_iflags & ISCAN_CANCEL) != 0)) { /* XXX printf? */ if_printf(vap->iv_ifp, - "%s: OOPS! scan cancelled during driver call!\n", + "%s: OOPS! scan cancelled during driver call (1)!\n", __func__); + scandone = 1; } - scandone |= ((SCAN_PRIVATE(ss)->ss_iflags & ISCAN_CANCEL) != 0); /* * Record scan complete time. Note that we also do @@ -1032,6 +1072,19 @@ scan_task(void *arg, int pending) ticks, ss->ss_mindwell, scanend); /* + * Since a cancellation may have occured during one of the + * driver calls (whilst unlocked), update scandone. + */ + if (scandone == 0 && + ((SCAN_PRIVATE(ss)->ss_iflags & ISCAN_CANCEL) != 0)) { + /* XXX printf? */ + if_printf(vap->iv_ifp, + "%s: OOPS! scan cancelled during driver call (2)!\n", + __func__); + scandone = 1; + } + + /* * Clear the SCAN bit first in case frames are * pending on the station power save queue. If * we defer this then the dispatch of the frames diff --git a/sys/net80211/ieee80211_sta.c b/sys/net80211/ieee80211_sta.c index 89ed856..c3c2bc9 100644 --- a/sys/net80211/ieee80211_sta.c +++ b/sys/net80211/ieee80211_sta.c @@ -1405,6 +1405,7 @@ sta_recv_mgmt(struct ieee80211_node *ni, struct mbuf *m0, int ix = aid / NBBY; int min = tim->tim_bitctl &~ 1; int max = tim->tim_len + min - 4; + int tim_ucast = 0, tim_mcast = 0; /* * Only do this for unicast traffic in the TIM @@ -1414,20 +1415,42 @@ sta_recv_mgmt(struct ieee80211_node *ni, struct mbuf *m0, */ if (min <= ix && ix <= max && isset(tim->tim_bitmap - min, aid)) { - ieee80211_sta_tim_notify(vap, 1); - ic->ic_lastdata = ticks; + tim_ucast = 1; } /* - * XXX TODO: do a separate notification + * Do a separate notification * for the multicast bit being set. */ -#if 0 if (tim->tim_bitctl & 1) { + tim_mcast = 1; + } + + /* + * If the TIM indicates there's traffic for + * us then get us out of STA mode powersave. + */ + if (tim_ucast == 1) { + + /* + * Wake us out of SLEEP state if we're + * in it; and if we're doing bgscan + * then wake us out of STA powersave. + */ ieee80211_sta_tim_notify(vap, 1); + + /* + * This is preventing us from + * continuing a bgscan; because it + * tricks the contbgscan() + * routine to think there's always + * traffic for us. + * + * I think we need both an RX and + * TX ic_lastdata field. + */ ic->ic_lastdata = ticks; } -#endif ni->ni_dtim_count = tim->tim_count; ni->ni_dtim_period = tim->tim_period; diff --git a/sys/netinet/in_gif.c b/sys/netinet/in_gif.c index 46f68f5..cd412ad 100644 --- a/sys/netinet/in_gif.c +++ b/sys/netinet/in_gif.c @@ -57,7 +57,6 @@ __FBSDID("$FreeBSD$"); #include <netinet/in_systm.h> #include <netinet/ip.h> #include <netinet/ip_var.h> -#include <netinet/in_gif.h> #include <netinet/in_var.h> #include <netinet/ip_encap.h> #include <netinet/ip_ecn.h> @@ -70,9 +69,10 @@ __FBSDID("$FreeBSD$"); static int gif_validate4(const struct ip *, struct gif_softc *, struct ifnet *); +static int in_gif_input(struct mbuf **, int *, int); extern struct domain inetdomain; -struct protosw in_gif_protosw = { +static struct protosw in_gif_protosw = { .pr_type = SOCK_RAW, .pr_domain = &inetdomain, .pr_protocol = 0/* IPPROTO_IPV[46] */, @@ -83,7 +83,8 @@ struct protosw in_gif_protosw = { .pr_usrreqs = &rip_usrreqs }; -VNET_DEFINE(int, ip_gif_ttl) = GIF_TTL; +#define GIF_TTL 30 +static VNET_DEFINE(int, ip_gif_ttl) = GIF_TTL; #define V_ip_gif_ttl VNET(ip_gif_ttl) SYSCTL_INT(_net_inet_ip, IPCTL_GIF_TTL, gifttl, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip_gif_ttl), 0, ""); @@ -133,7 +134,7 @@ in_gif_output(struct ifnet *ifp, struct mbuf *m, int proto, uint8_t ecn) return (ip_output(m, NULL, NULL, 0, NULL, NULL)); } -int +static int in_gif_input(struct mbuf **mp, int *offp, int proto) { struct mbuf *m = *mp; diff --git a/sys/netinet/in_gif.h b/sys/netinet/in_gif.h deleted file mode 100644 index d48d881..0000000 --- a/sys/netinet/in_gif.h +++ /dev/null @@ -1,44 +0,0 @@ -/* $FreeBSD$ */ -/* $KAME: in_gif.h,v 1.5 2000/04/14 08:36:02 itojun Exp $ */ - -/*- - * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. - * All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 3. Neither the name of the project nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - */ - -#ifndef _NETINET_IN_GIF_H_ -#define _NETINET_IN_GIF_H_ - -#define GIF_TTL 30 - -struct gif_softc; -int in_gif_input(struct mbuf **, int *, int); -int in_gif_output(struct ifnet *, struct mbuf *, int, uint8_t); -int in_gif_encapcheck(const struct mbuf *, int, int, void *); -int in_gif_attach(struct gif_softc *); - -#endif /*_NETINET_IN_GIF_H_*/ diff --git a/sys/netinet/in_systm.h b/sys/netinet/in_systm.h index 4b34aa0..acd2e62 100644 --- a/sys/netinet/in_systm.h +++ b/sys/netinet/in_systm.h @@ -44,12 +44,15 @@ * Internally the system keeps counters in the headers with the bytes * swapped so that VAX instructions will work on them. It reverses * the bytes before transmission at each protocol level. The n_ types - * represent the types with the bytes in ``high-ender'' order. + * represent the types with the bytes in ``high-ender'' order. Network + * byte order is usually referered to as big-endian these days rather + * than high-ender, which sadly invokes an Orson Scott Card novel, or + * worse, the movie. */ typedef u_int16_t n_short; /* short as received from the net */ typedef u_int32_t n_long; /* long as received from the net */ -typedef u_int32_t n_time; /* ms since 00:00 GMT, byte rev */ +typedef u_int32_t n_time; /* ms since 00:00 UTC, byte rev */ #ifdef _KERNEL uint32_t iptime(void); diff --git a/sys/netinet/ip_gre.c b/sys/netinet/ip_gre.c index 9f28191..5942364 100644 --- a/sys/netinet/ip_gre.c +++ b/sys/netinet/ip_gre.c @@ -68,11 +68,6 @@ __FBSDID("$FreeBSD$"); #include <net/if_gre.h> extern struct domain inetdomain; -extern int gre_input(struct mbuf **, int *, int); - -int in_gre_attach(struct gre_softc *); -int in_gre_output(struct mbuf *, int, int); - static const struct protosw in_gre_protosw = { .pr_type = SOCK_RAW, .pr_domain = &inetdomain, diff --git a/sys/netinet/ip_icmp.c b/sys/netinet/ip_icmp.c index d134a2e..b6b967f 100644 --- a/sys/netinet/ip_icmp.c +++ b/sys/netinet/ip_icmp.c @@ -861,7 +861,7 @@ icmp_send(struct mbuf *m, struct mbuf *opts) } /* - * Return milliseconds since 00:00 GMT in network format. + * Return milliseconds since 00:00 UTC in network format. */ uint32_t iptime(void) diff --git a/sys/netinet/ip_icmp.h b/sys/netinet/ip_icmp.h index ca1e963..38d44d7 100644 --- a/sys/netinet/ip_icmp.h +++ b/sys/netinet/ip_icmp.h @@ -99,7 +99,7 @@ struct icmp { struct id_ts { /* ICMP Timestamp */ /* * The next 3 fields are in network format, - * milliseconds since 00:00 GMT + * milliseconds since 00:00 UTC */ uint32_t its_otime; /* Originate */ uint32_t its_rtime; /* Receive */ diff --git a/sys/netinet/sctp_usrreq.c b/sys/netinet/sctp_usrreq.c index 617e480..95955de 100644 --- a/sys/netinet/sctp_usrreq.c +++ b/sys/netinet/sctp_usrreq.c @@ -3639,16 +3639,10 @@ flags_out: policy = sprstat->sprstat_policy; #if defined(SCTP_DETAILED_STR_STATS) if ((stcb != NULL) && - (policy != SCTP_PR_SCTP_NONE) && (sid < stcb->asoc.streamoutcnt) && - ((policy == SCTP_PR_SCTP_ALL) || - (PR_SCTP_VALID_POLICY(policy)))) { -#else - if ((stcb != NULL) && (policy != SCTP_PR_SCTP_NONE) && - (sid < stcb->asoc.streamoutcnt) && - (policy == SCTP_PR_SCTP_ALL)) { -#endif + ((policy <= SCTP_PR_SCTP_MAX) || + (policy == SCTP_PR_SCTP_ALL))) { if (policy == SCTP_PR_SCTP_ALL) { sprstat->sprstat_abandoned_unsent = stcb->asoc.strmout[sid].abandoned_unsent[0]; sprstat->sprstat_abandoned_sent = stcb->asoc.strmout[sid].abandoned_sent[0]; @@ -3656,6 +3650,13 @@ flags_out: sprstat->sprstat_abandoned_unsent = stcb->asoc.strmout[sid].abandoned_unsent[policy]; sprstat->sprstat_abandoned_sent = stcb->asoc.strmout[sid].abandoned_sent[policy]; } +#else + if ((stcb != NULL) && + (sid < stcb->asoc.streamoutcnt) && + (policy == SCTP_PR_SCTP_ALL)) { + sprstat->sprstat_abandoned_unsent = stcb->asoc.strmout[sid].abandoned_unsent[0]; + sprstat->sprstat_abandoned_sent = stcb->asoc.strmout[sid].abandoned_sent[0]; +#endif SCTP_TCB_UNLOCK(stcb); *optsize = sizeof(struct sctp_prstatus); } else { @@ -3675,8 +3676,8 @@ flags_out: policy = sprstat->sprstat_policy; if ((stcb != NULL) && (policy != SCTP_PR_SCTP_NONE) && - ((policy == SCTP_PR_SCTP_ALL) || - (PR_SCTP_VALID_POLICY(policy)))) { + ((policy <= SCTP_PR_SCTP_MAX) || + (policy == SCTP_PR_SCTP_ALL))) { if (policy == SCTP_PR_SCTP_ALL) { sprstat->sprstat_abandoned_unsent = stcb->asoc.abandoned_unsent[0]; sprstat->sprstat_abandoned_sent = stcb->asoc.abandoned_sent[0]; @@ -6037,7 +6038,7 @@ sctp_setopt(struct socket *so, int optname, void *optval, size_t optsize, SCTP_CHECK_AND_CAST(info, optval, struct sctp_default_prinfo, optsize); SCTP_FIND_STCB(inp, stcb, info->pr_assoc_id); - if (PR_SCTP_INVALID_POLICY(info->pr_policy)) { + if (info->pr_policy > SCTP_PR_SCTP_MAX) { SCTP_LTRACE_ERR_RET(inp, NULL, NULL, SCTP_FROM_SCTP_USRREQ, EINVAL); error = EINVAL; break; diff --git a/sys/netinet/sctputil.c b/sys/netinet/sctputil.c index d066c74..edf6ae8 100644 --- a/sys/netinet/sctputil.c +++ b/sys/netinet/sctputil.c @@ -1949,7 +1949,7 @@ sctp_timer_start(int t_type, struct sctp_inpcb *inp, struct sctp_tcb *stcb, * though we use a different timer. We also add the HB timer * PLUS a random jitter. */ - if ((inp == NULL) || (stcb == NULL) || (net == NULL)) { + if ((stcb == NULL) || (net == NULL)) { return; } else { uint32_t rndval; @@ -2004,9 +2004,6 @@ sctp_timer_start(int t_type, struct sctp_inpcb *inp, struct sctp_tcb *stcb, * nothing needed but the endpoint here ususually about 60 * minutes. */ - if (inp == NULL) { - return; - } tmr = &inp->sctp_ep.signature_change; to_ticks = inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_SIGNATURE]; break; @@ -2023,9 +2020,6 @@ sctp_timer_start(int t_type, struct sctp_inpcb *inp, struct sctp_tcb *stcb, * timer since that has stopped and we are in the GONE * state. */ - if (inp == NULL) { - return; - } tmr = &inp->sctp_ep.signature_change; to_ticks = MSEC_TO_TICKS(SCTP_INP_KILL_TIMEOUT); break; @@ -2034,10 +2028,7 @@ sctp_timer_start(int t_type, struct sctp_inpcb *inp, struct sctp_tcb *stcb, * Here we use the value found in the EP for PMTU ususually * about 10 minutes. */ - if ((stcb == NULL) || (inp == NULL)) { - return; - } - if (net == NULL) { + if ((stcb == NULL) || (net == NULL)) { return; } if (net->dest_state & SCTP_ADDR_NO_PMTUD) { @@ -2063,7 +2054,7 @@ sctp_timer_start(int t_type, struct sctp_inpcb *inp, struct sctp_tcb *stcb, * Here we use the endpoints shutdown guard timer usually * about 3 minutes. */ - if ((inp == NULL) || (stcb == NULL)) { + if (stcb == NULL) { return; } to_ticks = inp->sctp_ep.sctp_timeoutticks[SCTP_TIMER_MAXSHUTDOWN]; diff --git a/sys/netinet6/in6_gif.c b/sys/netinet6/in6_gif.c index 10d5d7b..3db4b4f 100644 --- a/sys/netinet6/in6_gif.c +++ b/sys/netinet6/in6_gif.c @@ -64,7 +64,6 @@ __FBSDID("$FreeBSD$"); #ifdef INET6 #include <netinet/ip6.h> #include <netinet6/ip6_var.h> -#include <netinet6/in6_gif.h> #include <netinet6/in6_var.h> #endif #include <netinet/ip_ecn.h> @@ -74,7 +73,8 @@ __FBSDID("$FreeBSD$"); #include <net/if_gif.h> -VNET_DEFINE(int, ip6_gif_hlim) = GIF_HLIM; +#define GIF_HLIM 30 +static VNET_DEFINE(int, ip6_gif_hlim) = GIF_HLIM; #define V_ip6_gif_hlim VNET(ip6_gif_hlim) SYSCTL_DECL(_net_inet6_ip6); @@ -83,9 +83,10 @@ SYSCTL_INT(_net_inet6_ip6, IPV6CTL_GIF_HLIM, gifhlim, CTLFLAG_VNET | CTLFLAG_RW, static int gif_validate6(const struct ip6_hdr *, struct gif_softc *, struct ifnet *); +static int in6_gif_input(struct mbuf **, int *, int); extern struct domain inet6domain; -struct protosw in6_gif_protosw = { +static struct protosw in6_gif_protosw = { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, .pr_protocol = 0, /* IPPROTO_IPV[46] */ @@ -144,7 +145,7 @@ in6_gif_output(struct ifnet *ifp, struct mbuf *m, int proto, uint8_t ecn) return (ip6_output(m, 0, NULL, IPV6_MINMTU, 0, NULL, NULL)); } -int +static int in6_gif_input(struct mbuf **mp, int *offp, int proto) { struct mbuf *m = *mp; diff --git a/sys/netinet6/in6_gif.h b/sys/netinet6/in6_gif.h deleted file mode 100644 index 1246171..0000000 --- a/sys/netinet6/in6_gif.h +++ /dev/null @@ -1,44 +0,0 @@ -/*- - * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. - * All rights reserved. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 3. Neither the name of the project nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * $KAME: in6_gif.h,v 1.5 2000/04/14 08:36:03 itojun Exp $ - * $FreeBSD$ - */ - -#ifndef _NETINET6_IN6_GIF_H_ -#define _NETINET6_IN6_GIF_H_ - -#define GIF_HLIM 30 - -struct gif_softc; -int in6_gif_input(struct mbuf **, int *, int); -int in6_gif_output(struct ifnet *, struct mbuf *, int, uint8_t); -int in6_gif_encapcheck(const struct mbuf *, int, int, void *); -int in6_gif_attach(struct gif_softc *); - -#endif /* _NETINET6_IN6_GIF_H_ */ diff --git a/sys/netinet6/ip6_gre.c b/sys/netinet6/ip6_gre.c index 29a0725..095a1de 100644 --- a/sys/netinet6/ip6_gre.c +++ b/sys/netinet6/ip6_gre.c @@ -62,11 +62,6 @@ __FBSDID("$FreeBSD$"); #include <net/if_gre.h> extern struct domain inet6domain; -extern int gre_input(struct mbuf **, int *, int); - -int in6_gre_attach(struct gre_softc *); -int in6_gre_output(struct mbuf *, int, int); - struct protosw in6_gre_protosw = { .pr_type = SOCK_RAW, .pr_domain = &inet6domain, diff --git a/sys/netipsec/ipsec.c b/sys/netipsec/ipsec.c index 5e0cdbf..8a79052 100644 --- a/sys/netipsec/ipsec.c +++ b/sys/netipsec/ipsec.c @@ -118,11 +118,12 @@ VNET_DEFINE(int, ip4_esp_trans_deflev) = IPSEC_LEVEL_USE; VNET_DEFINE(int, ip4_esp_net_deflev) = IPSEC_LEVEL_USE; VNET_DEFINE(int, ip4_ah_trans_deflev) = IPSEC_LEVEL_USE; VNET_DEFINE(int, ip4_ah_net_deflev) = IPSEC_LEVEL_USE; -VNET_DEFINE(struct secpolicy, ip4_def_policy); /* ECN ignore(-1)/forbidden(0)/allowed(1) */ VNET_DEFINE(int, ip4_ipsec_ecn) = 0; VNET_DEFINE(int, ip4_esp_randpad) = -1; +static VNET_DEFINE(struct secpolicy, def_policy); +#define V_def_policy VNET(def_policy) /* * Crypto support requirements: * @@ -141,7 +142,7 @@ SYSCTL_DECL(_net_inet_ipsec); /* net.inet.ipsec */ SYSCTL_INT(_net_inet_ipsec, IPSECCTL_DEF_POLICY, def_policy, - CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip4_def_policy).policy, 0, + CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(def_policy).policy, 0, "IPsec default policy."); SYSCTL_INT(_net_inet_ipsec, IPSECCTL_DEF_ESP_TRANSLEV, esp_trans_deflev, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip4_esp_trans_deflev), 0, @@ -213,7 +214,7 @@ SYSCTL_DECL(_net_inet6_ipsec6); /* net.inet6.ipsec6 */ SYSCTL_INT(_net_inet6_ipsec6, IPSECCTL_DEF_POLICY, def_policy, - CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip4_def_policy).policy, 0, + CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(def_policy).policy, 0, "IPsec default policy."); SYSCTL_INT(_net_inet6_ipsec6, IPSECCTL_DEF_ESP_TRANSLEV, esp_trans_deflev, CTLFLAG_VNET | CTLFLAG_RW, &VNET_NAME(ip6_esp_trans_deflev), 0, @@ -262,7 +263,7 @@ key_allocsp_default(const char* where, int tag) KEYDEBUG(KEYDEBUG_IPSEC_STAMP, printf("DP key_allocsp_default from %s:%u\n", where, tag)); - sp = &V_ip4_def_policy; + sp = &V_def_policy; if (sp->policy != IPSEC_POLICY_DISCARD && sp->policy != IPSEC_POLICY_NONE) { ipseclog((LOG_INFO, "fixed system default policy: %d->%d\n", @@ -828,17 +829,13 @@ ipsec_init_policy(struct socket *so, struct inpcbpolicy **pcb_sp) ipsec_delpcbpolicy(new); return (ENOBUFS); } - new->sp_in->state = IPSEC_SPSTATE_ALIVE; new->sp_in->policy = IPSEC_POLICY_ENTRUST; - if ((new->sp_out = KEY_NEWSP()) == NULL) { KEY_FREESP(&new->sp_in); ipsec_delpcbpolicy(new); return (ENOBUFS); } - new->sp_out->state = IPSEC_SPSTATE_ALIVE; new->sp_out->policy = IPSEC_POLICY_ENTRUST; - *pcb_sp = new; return (0); @@ -927,7 +924,6 @@ ipsec_deepcopy_policy(struct secpolicy *src) } dst->req = newchain; - dst->state = src->state; dst->policy = src->policy; /* Do not touch the refcnt fields. */ @@ -979,8 +975,6 @@ ipsec_set_policy_internal(struct secpolicy **pcb_sp, int optname, if ((newsp = key_msg2sp(xpl, len, &error)) == NULL) return (error); - newsp->state = IPSEC_SPSTATE_ALIVE; - /* Clear old SP and set new SP. */ KEY_FREESP(pcb_sp); *pcb_sp = newsp; @@ -1693,14 +1687,15 @@ ipsec_dumpmbuf(struct mbuf *m) } static void -ipsec_init(const void *unused __unused) +def_policy_init(const void *unused __unused) { - SECPOLICY_LOCK_INIT(&V_ip4_def_policy); - V_ip4_def_policy.refcnt = 1; /* NB: disallow free. */ + bzero(&V_def_policy, sizeof(struct secpolicy)); + V_def_policy.policy = IPSEC_POLICY_NONE; + V_def_policy.refcnt = 1; } -VNET_SYSINIT(ipsec_init, SI_SUB_PROTO_DOMAININIT, SI_ORDER_ANY, ipsec_init, - NULL); +VNET_SYSINIT(def_policy_init, SI_SUB_PROTO_DOMAININIT, SI_ORDER_ANY, + def_policy_init, NULL); /* XXX This stuff doesn't belong here... */ diff --git a/sys/netipsec/ipsec.h b/sys/netipsec/ipsec.h index 8ed39fb..5e3e1c3 100644 --- a/sys/netipsec/ipsec.h +++ b/sys/netipsec/ipsec.h @@ -81,21 +81,15 @@ struct secpolicyindex { /* Security Policy Data Base */ struct secpolicy { - LIST_ENTRY(secpolicy) chain; - struct mtx lock; + TAILQ_ENTRY(secpolicy) chain; - u_int refcnt; /* reference count */ struct secpolicyindex spidx; /* selector */ - u_int32_t id; /* It's unique number on the system. */ - u_int state; /* 0: dead, others: alive */ -#define IPSEC_SPSTATE_DEAD 0 -#define IPSEC_SPSTATE_ALIVE 1 - u_int policy; /* policy_type per pfkeyv2.h */ - u_int16_t scangen; /* scan generation # */ struct ipsecrequest *req; /* pointer to the ipsec request tree, */ /* if policy == IPSEC else this value == NULL.*/ - + u_int refcnt; /* reference count */ + u_int policy; /* policy_type per pfkeyv2.h */ + u_int32_t id; /* It's unique number on the system. */ /* * lifetime handler. * the policy can be used without limitiation if both lifetime and @@ -109,13 +103,6 @@ struct secpolicy { long validtime; /* duration this policy is valid without use */ }; -#define SECPOLICY_LOCK_INIT(_sp) \ - mtx_init(&(_sp)->lock, "ipsec policy", NULL, MTX_DEF) -#define SECPOLICY_LOCK(_sp) mtx_lock(&(_sp)->lock) -#define SECPOLICY_UNLOCK(_sp) mtx_unlock(&(_sp)->lock) -#define SECPOLICY_LOCK_DESTROY(_sp) mtx_destroy(&(_sp)->lock) -#define SECPOLICY_LOCK_ASSERT(_sp) mtx_assert(&(_sp)->lock, MA_OWNED) - /* Request for IPsec */ struct ipsecrequest { struct ipsecrequest *next; @@ -279,7 +266,6 @@ VNET_DECLARE(int, ipsec_integrity); #endif VNET_PCPUSTAT_DECLARE(struct ipsecstat, ipsec4stat); -VNET_DECLARE(struct secpolicy, ip4_def_policy); VNET_DECLARE(int, ip4_esp_trans_deflev); VNET_DECLARE(int, ip4_esp_net_deflev); VNET_DECLARE(int, ip4_ah_trans_deflev); @@ -292,7 +278,6 @@ VNET_DECLARE(int, crypto_support); #define IPSECSTAT_INC(name) \ VNET_PCPUSTAT_ADD(struct ipsecstat, ipsec4stat, name, 1) -#define V_ip4_def_policy VNET(ip4_def_policy) #define V_ip4_esp_trans_deflev VNET(ip4_esp_trans_deflev) #define V_ip4_esp_net_deflev VNET(ip4_esp_net_deflev) #define V_ip4_ah_trans_deflev VNET(ip4_ah_trans_deflev) diff --git a/sys/netipsec/key.c b/sys/netipsec/key.c index 953f453..b9bcd59 100644 --- a/sys/netipsec/key.c +++ b/sys/netipsec/key.c @@ -48,6 +48,7 @@ #include <sys/domain.h> #include <sys/protosw.h> #include <sys/malloc.h> +#include <sys/rmlock.h> #include <sys/socket.h> #include <sys/socketvar.h> #include <sys/sysctl.h> @@ -141,16 +142,19 @@ static VNET_DEFINE(u_int32_t, acq_seq) = 0; #define V_acq_seq VNET(acq_seq) /* SPD */ -static VNET_DEFINE(LIST_HEAD(_sptree, secpolicy), sptree[IPSEC_DIR_MAX]); +static VNET_DEFINE(TAILQ_HEAD(_sptree, secpolicy), sptree[IPSEC_DIR_MAX]); +static struct rmlock sptree_lock; #define V_sptree VNET(sptree) -static struct mtx sptree_lock; -#define SPTREE_LOCK_INIT() \ - mtx_init(&sptree_lock, "sptree", \ - "fast ipsec security policy database", MTX_DEF) -#define SPTREE_LOCK_DESTROY() mtx_destroy(&sptree_lock) -#define SPTREE_LOCK() mtx_lock(&sptree_lock) -#define SPTREE_UNLOCK() mtx_unlock(&sptree_lock) -#define SPTREE_LOCK_ASSERT() mtx_assert(&sptree_lock, MA_OWNED) +#define SPTREE_LOCK_INIT() rm_init(&sptree_lock, "sptree") +#define SPTREE_LOCK_DESTROY() rm_destroy(&sptree_lock) +#define SPTREE_RLOCK_TRACKER struct rm_priotracker sptree_tracker +#define SPTREE_RLOCK() rm_rlock(&sptree_lock, &sptree_tracker) +#define SPTREE_RUNLOCK() rm_runlock(&sptree_lock, &sptree_tracker) +#define SPTREE_RLOCK_ASSERT() rm_assert(&sptree_lock, RA_RLOCKED) +#define SPTREE_WLOCK() rm_wlock(&sptree_lock) +#define SPTREE_WUNLOCK() rm_wunlock(&sptree_lock) +#define SPTREE_WLOCK_ASSERT() rm_assert(&sptree_lock, RA_WLOCKED) +#define SPTREE_UNLOCK_ASSERT() rm_assert(&sptree_lock, RA_UNLOCKED) static VNET_DEFINE(LIST_HEAD(_sahtree, secashead), sahtree); /* SAD */ #define V_sahtree VNET(sahtree) @@ -416,9 +420,8 @@ static struct callout key_timer; static struct secasvar *key_allocsa_policy(const struct secasindex *); static void key_freesp_so(struct secpolicy **); static struct secasvar *key_do_allocsa_policy(struct secashead *, u_int); -static void key_delsp(struct secpolicy *); +static void key_unlink(struct secpolicy *); static struct secpolicy *key_getsp(struct secpolicyindex *); -static void _key_delsp(struct secpolicy *sp); static struct secpolicy *key_getspbyid(u_int32_t); static u_int32_t key_newreqid(void); static struct mbuf *key_gather_mbuf(struct mbuf *, @@ -575,15 +578,8 @@ sa_delref(struct secasvar *sav) return (refcount_release(&sav->refcnt)); } -#define SP_ADDREF(p) do { \ - (p)->refcnt++; \ - IPSEC_ASSERT((p)->refcnt != 0, ("SP refcnt overflow")); \ -} while (0) -#define SP_DELREF(p) do { \ - IPSEC_ASSERT((p)->refcnt > 0, ("SP refcnt underflow")); \ - (p)->refcnt--; \ -} while (0) - +#define SP_ADDREF(p) refcount_acquire(&(p)->refcnt) +#define SP_DELREF(p) refcount_release(&(p)->refcnt) /* * Update the refcnt while holding the SPTREE lock. @@ -591,9 +587,8 @@ sa_delref(struct secasvar *sav) void key_addref(struct secpolicy *sp) { - SPTREE_LOCK(); + SP_ADDREF(sp); - SPTREE_UNLOCK(); } /* @@ -606,7 +601,7 @@ key_havesp(u_int dir) { return (dir == IPSEC_DIR_INBOUND || dir == IPSEC_DIR_OUTBOUND ? - LIST_FIRST(&V_sptree[dir]) != NULL : 1); + TAILQ_FIRST(&V_sptree[dir]) != NULL : 1); } /* %%% IPsec policy management */ @@ -620,6 +615,7 @@ struct secpolicy * key_allocsp(struct secpolicyindex *spidx, u_int dir, const char* where, int tag) { + SPTREE_RLOCK_TRACKER; struct secpolicy *sp; IPSEC_ASSERT(spidx != NULL, ("null spidx")); @@ -634,14 +630,11 @@ key_allocsp(struct secpolicyindex *spidx, u_int dir, const char* where, printf("*** objects\n"); kdebug_secpolicyindex(spidx)); - SPTREE_LOCK(); - LIST_FOREACH(sp, &V_sptree[dir], chain) { + SPTREE_RLOCK(); + TAILQ_FOREACH(sp, &V_sptree[dir], chain) { KEYDEBUG(KEYDEBUG_IPSEC_DATA, printf("*** in SPD\n"); kdebug_secpolicyindex(&sp->spidx)); - - if (sp->state == IPSEC_SPSTATE_DEAD) - continue; if (key_cmpspidx_withmask(&sp->spidx, spidx)) goto found; } @@ -655,7 +648,7 @@ found: sp->lastused = time_second; SP_ADDREF(sp); } - SPTREE_UNLOCK(); + SPTREE_RUNLOCK(); KEYDEBUG(KEYDEBUG_IPSEC_STAMP, printf("DP %s return SP:%p (ID=%u) refcnt %u\n", __func__, @@ -673,6 +666,7 @@ struct secpolicy * key_allocsp2(u_int32_t spi, union sockaddr_union *dst, u_int8_t proto, u_int dir, const char* where, int tag) { + SPTREE_RLOCK_TRACKER; struct secpolicy *sp; IPSEC_ASSERT(dst != NULL, ("null dst")); @@ -688,14 +682,11 @@ key_allocsp2(u_int32_t spi, union sockaddr_union *dst, u_int8_t proto, printf("spi %u proto %u dir %u\n", spi, proto, dir); kdebug_sockaddr(&dst->sa)); - SPTREE_LOCK(); - LIST_FOREACH(sp, &V_sptree[dir], chain) { + SPTREE_RLOCK(); + TAILQ_FOREACH(sp, &V_sptree[dir], chain) { KEYDEBUG(KEYDEBUG_IPSEC_DATA, printf("*** in SPD\n"); kdebug_secpolicyindex(&sp->spidx)); - - if (sp->state == IPSEC_SPSTATE_DEAD) - continue; /* compare simple values, then dst address */ if (sp->spidx.ul_proto != proto) continue; @@ -715,7 +706,7 @@ found: sp->lastused = time_second; SP_ADDREF(sp); } - SPTREE_UNLOCK(); + SPTREE_RUNLOCK(); KEYDEBUG(KEYDEBUG_IPSEC_STAMP, printf("DP %s return SP:%p (ID=%u) refcnt %u\n", __func__, @@ -1174,22 +1165,41 @@ done: void _key_freesp(struct secpolicy **spp, const char* where, int tag) { + struct ipsecrequest *isr, *nextisr; struct secpolicy *sp = *spp; IPSEC_ASSERT(sp != NULL, ("null sp")); - - SPTREE_LOCK(); - SP_DELREF(sp); - KEYDEBUG(KEYDEBUG_IPSEC_STAMP, printf("DP %s SP:%p (ID=%u) from %s:%u; refcnt now %u\n", __func__, sp, sp->id, where, tag, sp->refcnt)); - if (sp->refcnt == 0) { - *spp = NULL; - key_delsp(sp); + if (SP_DELREF(sp) == 0) + return; + *spp = NULL; + for (isr = sp->req; isr != NULL; isr = nextisr) { + if (isr->sav != NULL) { + KEY_FREESAV(&isr->sav); + isr->sav = NULL; + } + nextisr = isr->next; + ipsec_delisr(isr); } - SPTREE_UNLOCK(); + free(sp, M_IPSEC_SP); +} + +static void +key_unlink(struct secpolicy *sp) +{ + + IPSEC_ASSERT(sp != NULL, ("null sp")); + IPSEC_ASSERT(sp->spidx.dir == IPSEC_DIR_INBOUND || + sp->spidx.dir == IPSEC_DIR_OUTBOUND, + ("invalid direction %u", sp->spidx.dir)); + SPTREE_UNLOCK_ASSERT(); + + SPTREE_WLOCK(); + TAILQ_REMOVE(&V_sptree[sp->spidx.dir], sp, chain); + SPTREE_WUNLOCK(); } /* @@ -1278,38 +1288,6 @@ key_freesav(struct secasvar **psav, const char* where, int tag) /* %%% SPD management */ /* - * free security policy entry. - */ -static void -key_delsp(struct secpolicy *sp) -{ - struct ipsecrequest *isr, *nextisr; - - IPSEC_ASSERT(sp != NULL, ("null sp")); - SPTREE_LOCK_ASSERT(); - - sp->state = IPSEC_SPSTATE_DEAD; - - IPSEC_ASSERT(sp->refcnt == 0, - ("SP with references deleted (refcnt %u)", sp->refcnt)); - - /* remove from SP index */ - if (__LIST_CHAINED(sp)) - LIST_REMOVE(sp, chain); - - for (isr = sp->req; isr != NULL; isr = nextisr) { - if (isr->sav != NULL) { - KEY_FREESAV(&isr->sav); - isr->sav = NULL; - } - - nextisr = isr->next; - ipsec_delisr(isr); - } - _key_delsp(sp); -} - -/* * search SPD * OUT: NULL : not found * others : found, pointer to a SP. @@ -1317,20 +1295,19 @@ key_delsp(struct secpolicy *sp) static struct secpolicy * key_getsp(struct secpolicyindex *spidx) { + SPTREE_RLOCK_TRACKER; struct secpolicy *sp; IPSEC_ASSERT(spidx != NULL, ("null spidx")); - SPTREE_LOCK(); - LIST_FOREACH(sp, &V_sptree[spidx->dir], chain) { - if (sp->state == IPSEC_SPSTATE_DEAD) - continue; + SPTREE_RLOCK(); + TAILQ_FOREACH(sp, &V_sptree[spidx->dir], chain) { if (key_cmpspidx_exactly(spidx, &sp->spidx)) { SP_ADDREF(sp); break; } } - SPTREE_UNLOCK(); + SPTREE_RUNLOCK(); return sp; } @@ -1343,28 +1320,25 @@ key_getsp(struct secpolicyindex *spidx) static struct secpolicy * key_getspbyid(u_int32_t id) { + SPTREE_RLOCK_TRACKER; struct secpolicy *sp; - SPTREE_LOCK(); - LIST_FOREACH(sp, &V_sptree[IPSEC_DIR_INBOUND], chain) { - if (sp->state == IPSEC_SPSTATE_DEAD) - continue; + SPTREE_RLOCK(); + TAILQ_FOREACH(sp, &V_sptree[IPSEC_DIR_INBOUND], chain) { if (sp->id == id) { SP_ADDREF(sp); goto done; } } - LIST_FOREACH(sp, &V_sptree[IPSEC_DIR_OUTBOUND], chain) { - if (sp->state == IPSEC_SPSTATE_DEAD) - continue; + TAILQ_FOREACH(sp, &V_sptree[IPSEC_DIR_OUTBOUND], chain) { if (sp->id == id) { SP_ADDREF(sp); goto done; } } done: - SPTREE_UNLOCK(); + SPTREE_RUNLOCK(); return sp; } @@ -1376,11 +1350,8 @@ key_newsp(const char* where, int tag) newsp = (struct secpolicy *) malloc(sizeof(struct secpolicy), M_IPSEC_SP, M_NOWAIT|M_ZERO); - if (newsp) { - SECPOLICY_LOCK_INIT(newsp); - newsp->refcnt = 1; - newsp->req = NULL; - } + if (newsp) + refcount_init(&newsp->refcnt, 1); KEYDEBUG(KEYDEBUG_IPSEC_STAMP, printf("DP %s from %s:%u return SP:%p\n", __func__, @@ -1388,13 +1359,6 @@ key_newsp(const char* where, int tag) return newsp; } -static void -_key_delsp(struct secpolicy *sp) -{ - SECPOLICY_LOCK_DESTROY(sp); - free(sp, M_IPSEC_SP); -} - /* * create secpolicy structure from sadb_x_policy structure. * NOTE: `state', `secpolicyindex' in secpolicy structure are not set, @@ -1874,9 +1838,7 @@ key_spdadd(struct socket *so, struct mbuf *m, const struct sadb_msghdr *mhp) newsp = key_getsp(&spidx); if (mhp->msg->sadb_msg_type == SADB_X_SPDUPDATE) { if (newsp) { - SPTREE_LOCK(); - newsp->state = IPSEC_SPSTATE_DEAD; - SPTREE_UNLOCK(); + key_unlink(newsp); KEY_FREESP(&newsp); } } else { @@ -1888,13 +1850,15 @@ key_spdadd(struct socket *so, struct mbuf *m, const struct sadb_msghdr *mhp) } } + /* XXX: there is race between key_getsp and key_msg2sp. */ + /* allocation new SP entry */ if ((newsp = key_msg2sp(xpl0, PFKEY_EXTLEN(xpl0), &error)) == NULL) { return key_senderror(so, m, error); } if ((newsp->id = key_getnewspid()) == 0) { - _key_delsp(newsp); + KEY_FREESP(&newsp); return key_senderror(so, m, ENOBUFS); } @@ -1910,18 +1874,20 @@ key_spdadd(struct socket *so, struct mbuf *m, const struct sadb_msghdr *mhp) /* sanity check on addr pair */ if (((struct sockaddr *)(src0 + 1))->sa_family != ((struct sockaddr *)(dst0+ 1))->sa_family) { - _key_delsp(newsp); + KEY_FREESP(&newsp); return key_senderror(so, m, EINVAL); } if (((struct sockaddr *)(src0 + 1))->sa_len != ((struct sockaddr *)(dst0+ 1))->sa_len) { - _key_delsp(newsp); + KEY_FREESP(&newsp); return key_senderror(so, m, EINVAL); } #if 1 - if (newsp->req && newsp->req->saidx.src.sa.sa_family && newsp->req->saidx.dst.sa.sa_family) { - if (newsp->req->saidx.src.sa.sa_family != newsp->req->saidx.dst.sa.sa_family) { - _key_delsp(newsp); + if (newsp->req && newsp->req->saidx.src.sa.sa_family && + newsp->req->saidx.dst.sa.sa_family) { + if (newsp->req->saidx.src.sa.sa_family != + newsp->req->saidx.dst.sa.sa_family) { + KEY_FREESP(&newsp); return key_senderror(so, m, EINVAL); } } @@ -1932,9 +1898,9 @@ key_spdadd(struct socket *so, struct mbuf *m, const struct sadb_msghdr *mhp) newsp->lifetime = lft ? lft->sadb_lifetime_addtime : 0; newsp->validtime = lft ? lft->sadb_lifetime_usetime : 0; - newsp->refcnt = 1; /* do not reclaim until I say I do */ - newsp->state = IPSEC_SPSTATE_ALIVE; - LIST_INSERT_TAIL(&V_sptree[newsp->spidx.dir], newsp, secpolicy, chain); + SPTREE_WLOCK(); + TAILQ_INSERT_TAIL(&V_sptree[newsp->spidx.dir], newsp, chain); + SPTREE_WUNLOCK(); /* delete the entry in spacqtree */ if (mhp->msg->sadb_msg_type == SADB_X_SPDUPDATE) { @@ -2109,9 +2075,7 @@ key_spddelete(struct socket *so, struct mbuf *m, /* save policy id to buffer to be returned. */ xpl0->sadb_x_policy_id = sp->id; - SPTREE_LOCK(); - sp->state = IPSEC_SPSTATE_DEAD; - SPTREE_UNLOCK(); + key_unlink(sp); KEY_FREESP(&sp); { @@ -2176,9 +2140,7 @@ key_spddelete2(struct socket *so, struct mbuf *m, return key_senderror(so, m, EINVAL); } - SPTREE_LOCK(); - sp->state = IPSEC_SPSTATE_DEAD; - SPTREE_UNLOCK(); + key_unlink(sp); KEY_FREESP(&sp); { @@ -2356,8 +2318,9 @@ key_spdacquire(struct secpolicy *sp) static int key_spdflush(struct socket *so, struct mbuf *m, const struct sadb_msghdr *mhp) { + TAILQ_HEAD(, secpolicy) drainq; struct sadb_msg *newmsg; - struct secpolicy *sp; + struct secpolicy *sp, *nextsp; u_int dir; IPSEC_ASSERT(so != NULL, ("null socket")); @@ -2368,11 +2331,17 @@ key_spdflush(struct socket *so, struct mbuf *m, const struct sadb_msghdr *mhp) if (m->m_len != PFKEY_ALIGN8(sizeof(struct sadb_msg))) return key_senderror(so, m, EINVAL); + TAILQ_INIT(&drainq); + SPTREE_WLOCK(); for (dir = 0; dir < IPSEC_DIR_MAX; dir++) { - SPTREE_LOCK(); - LIST_FOREACH(sp, &V_sptree[dir], chain) - sp->state = IPSEC_SPSTATE_DEAD; - SPTREE_UNLOCK(); + TAILQ_CONCAT(&drainq, &V_sptree[dir], chain); + } + SPTREE_WUNLOCK(); + sp = TAILQ_FIRST(&drainq); + while (sp != NULL) { + nextsp = TAILQ_NEXT(sp, chain); + KEY_FREESP(&sp); + sp = nextsp; } if (sizeof(struct sadb_msg) > m->m_len + M_TRAILINGSPACE(m)) { @@ -2405,6 +2374,7 @@ key_spdflush(struct socket *so, struct mbuf *m, const struct sadb_msghdr *mhp) static int key_spddump(struct socket *so, struct mbuf *m, const struct sadb_msghdr *mhp) { + SPTREE_RLOCK_TRACKER; struct secpolicy *sp; int cnt; u_int dir; @@ -2417,20 +2387,20 @@ key_spddump(struct socket *so, struct mbuf *m, const struct sadb_msghdr *mhp) /* search SPD entry and get buffer size. */ cnt = 0; - SPTREE_LOCK(); + SPTREE_RLOCK(); for (dir = 0; dir < IPSEC_DIR_MAX; dir++) { - LIST_FOREACH(sp, &V_sptree[dir], chain) { + TAILQ_FOREACH(sp, &V_sptree[dir], chain) { cnt++; } } if (cnt == 0) { - SPTREE_UNLOCK(); + SPTREE_RUNLOCK(); return key_senderror(so, m, ENOENT); } for (dir = 0; dir < IPSEC_DIR_MAX; dir++) { - LIST_FOREACH(sp, &V_sptree[dir], chain) { + TAILQ_FOREACH(sp, &V_sptree[dir], chain) { --cnt; n = key_setdumpsp(sp, SADB_X_SPDDUMP, cnt, mhp->msg->sadb_msg_pid); @@ -2440,7 +2410,7 @@ key_spddump(struct socket *so, struct mbuf *m, const struct sadb_msghdr *mhp) } } - SPTREE_UNLOCK(); + SPTREE_RUNLOCK(); m_freem(m); return 0; } @@ -2452,6 +2422,8 @@ key_setdumpsp(struct secpolicy *sp, u_int8_t type, u_int32_t seq, struct mbuf *result = NULL, *m; struct seclifetime lt; + SPTREE_RLOCK_ASSERT(); + m = key_setsadbmsg(type, 0, SADB_SATYPE_UNSPEC, seq, pid, sp->refcnt); if (!m) goto fail; @@ -4226,47 +4198,29 @@ key_bbcmp(const void *a1, const void *a2, u_int bits) static void key_flush_spd(time_t now) { - static u_int16_t sptree_scangen = 0; - u_int16_t gen = sptree_scangen++; + SPTREE_RLOCK_TRACKER; struct secpolicy *sp; u_int dir; /* SPD */ for (dir = 0; dir < IPSEC_DIR_MAX; dir++) { restart: - SPTREE_LOCK(); - LIST_FOREACH(sp, &V_sptree[dir], chain) { - if (sp->scangen == gen) /* previously handled */ - continue; - sp->scangen = gen; - if (sp->state == IPSEC_SPSTATE_DEAD && - sp->refcnt == 1) { - /* - * Ensure that we only decrease refcnt once, - * when we're the last consumer. - * Directly call SP_DELREF/key_delsp instead - * of KEY_FREESP to avoid unlocking/relocking - * SPTREE_LOCK before key_delsp: may refcnt - * be increased again during that time ? - * NB: also clean entries created by - * key_spdflush - */ - SP_DELREF(sp); - key_delsp(sp); - SPTREE_UNLOCK(); - goto restart; - } + SPTREE_RLOCK(); + TAILQ_FOREACH(sp, &V_sptree[dir], chain) { if (sp->lifetime == 0 && sp->validtime == 0) continue; - if ((sp->lifetime && now - sp->created > sp->lifetime) - || (sp->validtime && now - sp->lastused > sp->validtime)) { - sp->state = IPSEC_SPSTATE_DEAD; - SPTREE_UNLOCK(); + if ((sp->lifetime && + now - sp->created > sp->lifetime) || + (sp->validtime && + now - sp->lastused > sp->validtime)) { + SPTREE_RUNLOCK(); + key_unlink(sp); key_spdexpire(sp); + KEY_FREESP(&sp); goto restart; } } - SPTREE_UNLOCK(); + SPTREE_RUNLOCK(); } } @@ -7609,7 +7563,7 @@ key_init(void) int i; for (i = 0; i < IPSEC_DIR_MAX; i++) - LIST_INIT(&V_sptree[i]); + TAILQ_INIT(&V_sptree[i]); LIST_INIT(&V_sahtree); @@ -7619,10 +7573,6 @@ key_init(void) LIST_INIT(&V_acqtree); LIST_INIT(&V_spacqtree); - /* system default */ - V_ip4_def_policy.policy = IPSEC_POLICY_NONE; - V_ip4_def_policy.refcnt++; /*never reclaim this*/ - if (!IS_DEFAULT_VNET(curvnet)) return; @@ -7647,6 +7597,7 @@ key_init(void) void key_destroy(void) { + TAILQ_HEAD(, secpolicy) drainq; struct secpolicy *sp, *nextsp; struct secacq *acq, *nextacq; struct secspacq *spacq, *nextspacq; @@ -7654,18 +7605,18 @@ key_destroy(void) struct secreg *reg; int i; - SPTREE_LOCK(); + TAILQ_INIT(&drainq); + SPTREE_WLOCK(); for (i = 0; i < IPSEC_DIR_MAX; i++) { - for (sp = LIST_FIRST(&V_sptree[i]); - sp != NULL; sp = nextsp) { - nextsp = LIST_NEXT(sp, chain); - if (__LIST_CHAINED(sp)) { - LIST_REMOVE(sp, chain); - free(sp, M_IPSEC_SP); - } - } + TAILQ_CONCAT(&drainq, &V_sptree[i], chain); + } + SPTREE_WUNLOCK(); + sp = TAILQ_FIRST(&drainq); + while (sp != NULL) { + nextsp = TAILQ_NEXT(sp, chain); + KEY_FREESP(&sp); + sp = nextsp; } - SPTREE_UNLOCK(); SAHTREE_LOCK(); for (sah = LIST_FIRST(&V_sahtree); sah != NULL; sah = nextsah) { diff --git a/sys/netipsec/key_debug.c b/sys/netipsec/key_debug.c index 5cbc1db..97ac061 100644 --- a/sys/netipsec/key_debug.c +++ b/sys/netipsec/key_debug.c @@ -463,8 +463,8 @@ kdebug_secpolicy(struct secpolicy *sp) if (sp == NULL) panic("%s: NULL pointer was passed.\n", __func__); - printf("secpolicy{ refcnt=%u state=%u policy=%u\n", - sp->refcnt, sp->state, sp->policy); + printf("secpolicy{ refcnt=%u policy=%u\n", + sp->refcnt, sp->policy); kdebug_secpolicyindex(&sp->spidx); diff --git a/sys/nfs/bootp_subr.c b/sys/nfs/bootp_subr.c index 5e3145e..c5bd4cb 100644 --- a/sys/nfs/bootp_subr.c +++ b/sys/nfs/bootp_subr.c @@ -1735,9 +1735,6 @@ retry: goto out; } rootdevnames[0] = "nfs:"; -#ifdef NFSCLIENT - rootdevnames[1] = "oldnfs:"; -#endif nfs_diskless_valid = 3; } diff --git a/sys/nfs/nfs_common.c b/sys/nfs/nfs_common.c deleted file mode 100644 index ad9e7a1..0000000 --- a/sys/nfs/nfs_common.c +++ /dev/null @@ -1,407 +0,0 @@ -/*- - * Copyright (c) 1989, 1993 - * The Regents of the University of California. All rights reserved. - * - * This code is derived from software contributed to Berkeley by - * Rick Macklem at The University of Guelph. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 4. Neither the name of the University nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * @(#)nfs_subs.c 8.8 (Berkeley) 5/22/95 - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -/* - * These functions support the macros and help fiddle mbuf chains for - * the nfs op functions. They do things like create the rpc header and - * copy data between mbuf chains and uio lists. - */ - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/kernel.h> -#include <sys/bio.h> -#include <sys/buf.h> -#include <sys/proc.h> -#include <sys/mount.h> -#include <sys/vnode.h> -#include <sys/namei.h> -#include <sys/mbuf.h> -#include <sys/socket.h> -#include <sys/stat.h> -#include <sys/malloc.h> -#include <sys/module.h> -#include <sys/sysent.h> -#include <sys/syscall.h> -#include <sys/sysctl.h> - -#include <vm/vm.h> -#include <vm/vm_object.h> -#include <vm/vm_extern.h> - -#include <nfs/nfsproto.h> -#include <nfsserver/nfs.h> -#include <nfs/xdr_subs.h> -#include <nfs/nfs_common.h> - -#include <netinet/in.h> - -enum vtype nv3tov_type[8]= { - VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO -}; -nfstype nfsv3_type[9] = { - NFNON, NFREG, NFDIR, NFBLK, NFCHR, NFLNK, NFSOCK, NFFIFO, NFNON -}; - -static void *nfsm_dissect_xx_sub(int s, struct mbuf **md, caddr_t *dpos, - int how); - -SYSCTL_NODE(_vfs, OID_AUTO, nfs_common, CTLFLAG_RD, 0, "NFS common support"); - -static int nfs_realign_test; -SYSCTL_INT(_vfs_nfs_common, OID_AUTO, realign_test, CTLFLAG_RD, - &nfs_realign_test, 0, "Number of realign tests done"); - -static int nfs_realign_count; -SYSCTL_INT(_vfs_nfs_common, OID_AUTO, realign_count, CTLFLAG_RD, - &nfs_realign_count, 0, "Number of mbuf realignments done"); - -/* - * copies mbuf chain to the uio scatter/gather list - */ -int -nfsm_mbuftouio(struct mbuf **mrep, struct uio *uiop, int siz, caddr_t *dpos) -{ - char *mbufcp, *uiocp; - int xfer, left, len; - struct mbuf *mp; - long uiosiz, rem; - int error = 0; - - mp = *mrep; - mbufcp = *dpos; - len = mtod(mp, caddr_t)+mp->m_len-mbufcp; - rem = nfsm_rndup(siz)-siz; - while (siz > 0) { - if (uiop->uio_iovcnt <= 0 || uiop->uio_iov == NULL) - return (EFBIG); - left = uiop->uio_iov->iov_len; - uiocp = uiop->uio_iov->iov_base; - if (left > siz) - left = siz; - uiosiz = left; - while (left > 0) { - while (len == 0) { - mp = mp->m_next; - if (mp == NULL) - return (EBADRPC); - mbufcp = mtod(mp, caddr_t); - len = mp->m_len; - } - xfer = (left > len) ? len : left; -#ifdef notdef - /* Not Yet.. */ - if (uiop->uio_iov->iov_op != NULL) - (*(uiop->uio_iov->iov_op)) - (mbufcp, uiocp, xfer); - else -#endif - if (uiop->uio_segflg == UIO_SYSSPACE) - bcopy(mbufcp, uiocp, xfer); - else - copyout(mbufcp, uiocp, xfer); - left -= xfer; - len -= xfer; - mbufcp += xfer; - uiocp += xfer; - uiop->uio_offset += xfer; - uiop->uio_resid -= xfer; - } - if (uiop->uio_iov->iov_len <= siz) { - uiop->uio_iovcnt--; - uiop->uio_iov++; - } else { - uiop->uio_iov->iov_base = - (char *)uiop->uio_iov->iov_base + uiosiz; - uiop->uio_iov->iov_len -= uiosiz; - } - siz -= uiosiz; - } - *dpos = mbufcp; - *mrep = mp; - if (rem > 0) { - if (len < rem) - error = nfs_adv(mrep, dpos, rem, len); - else - *dpos += rem; - } - return (error); -} - -/* - * Help break down an mbuf chain by setting the first siz bytes contiguous - * pointed to by returned val. - * This is used by the macros nfsm_dissect for tough - * cases. (The macros use the vars. dpos and dpos2) - */ -void * -nfsm_disct(struct mbuf **mdp, caddr_t *dposp, int siz, int left, int how) -{ - struct mbuf *mp, *mp2; - int siz2, xfer; - caddr_t ptr, npos = NULL; - void *ret; - - mp = *mdp; - while (left == 0) { - *mdp = mp = mp->m_next; - if (mp == NULL) - return (NULL); - left = mp->m_len; - *dposp = mtod(mp, caddr_t); - } - if (left >= siz) { - ret = *dposp; - *dposp += siz; - } else if (mp->m_next == NULL) { - return (NULL); - } else if (siz > MHLEN) { - panic("nfs S too big"); - } else { - mp2 = m_get(how, MT_DATA); - if (mp2 == NULL) - return (NULL); - mp2->m_len = siz; - mp2->m_next = mp->m_next; - mp->m_next = mp2; - mp->m_len -= left; - mp = mp2; - ptr = mtod(mp, caddr_t); - ret = ptr; - bcopy(*dposp, ptr, left); /* Copy what was left */ - siz2 = siz-left; - ptr += left; - mp2 = mp->m_next; - npos = mtod(mp2, caddr_t); - /* Loop around copying up the siz2 bytes */ - while (siz2 > 0) { - if (mp2 == NULL) - return (NULL); - xfer = (siz2 > mp2->m_len) ? mp2->m_len : siz2; - if (xfer > 0) { - bcopy(mtod(mp2, caddr_t), ptr, xfer); - mp2->m_data += xfer; - mp2->m_len -= xfer; - ptr += xfer; - siz2 -= xfer; - } - if (siz2 > 0) { - mp2 = mp2->m_next; - if (mp2 != NULL) - npos = mtod(mp2, caddr_t); - } - } - *mdp = mp2; - *dposp = mtod(mp2, caddr_t); - if (!nfsm_aligned(*dposp, u_int32_t)) { - bcopy(*dposp, npos, mp2->m_len); - mp2->m_data = npos; - *dposp = npos; - } - } - return (ret); -} - -/* - * Advance the position in the mbuf chain. - */ -int -nfs_adv(struct mbuf **mdp, caddr_t *dposp, int offs, int left) -{ - struct mbuf *m; - int s; - - m = *mdp; - s = left; - while (s < offs) { - offs -= s; - m = m->m_next; - if (m == NULL) - return (EBADRPC); - s = m->m_len; - } - *mdp = m; - *dposp = mtod(m, caddr_t)+offs; - return (0); -} - -void * -nfsm_build_xx(int s, struct mbuf **mb, caddr_t *bpos) -{ - struct mbuf *mb2; - void *ret; - - if (s > M_TRAILINGSPACE(*mb)) { - mb2 = m_get(M_WAITOK, MT_DATA); - if (s > MLEN) - panic("build > MLEN"); - (*mb)->m_next = mb2; - *mb = mb2; - (*mb)->m_len = 0; - *bpos = mtod(*mb, caddr_t); - } - ret = *bpos; - (*mb)->m_len += s; - *bpos += s; - return (ret); -} - -void * -nfsm_dissect_xx(int s, struct mbuf **md, caddr_t *dpos) -{ - - return (nfsm_dissect_xx_sub(s, md, dpos, M_WAITOK)); -} - -void * -nfsm_dissect_xx_nonblock(int s, struct mbuf **md, caddr_t *dpos) -{ - - return (nfsm_dissect_xx_sub(s, md, dpos, M_NOWAIT)); -} - -static void * -nfsm_dissect_xx_sub(int s, struct mbuf **md, caddr_t *dpos, int how) -{ - int t1; - char *cp2; - void *ret; - - t1 = mtod(*md, caddr_t) + (*md)->m_len - *dpos; - if (t1 >= s) { - ret = *dpos; - *dpos += s; - return (ret); - } - cp2 = nfsm_disct(md, dpos, s, t1, how); - return (cp2); -} - -int -nfsm_strsiz_xx(int *s, int m, struct mbuf **mb, caddr_t *bpos) -{ - u_int32_t *tl; - - tl = nfsm_dissect_xx(NFSX_UNSIGNED, mb, bpos); - if (tl == NULL) - return (EBADRPC); - *s = fxdr_unsigned(int32_t, *tl); - if (*s > m) - return (EBADRPC); - return (0); -} - -int -nfsm_adv_xx(int s, struct mbuf **md, caddr_t *dpos) -{ - int t1; - - t1 = mtod(*md, caddr_t) + (*md)->m_len - *dpos; - if (t1 >= s) { - *dpos += s; - return (0); - } - t1 = nfs_adv(md, dpos, s, t1); - if (t1) - return (t1); - return (0); -} - -/* - * Check for badly aligned mbuf data and realign by copying the unaligned - * portion of the data into a new mbuf chain and freeing the portions of the - * old chain that were replaced. - * - * We cannot simply realign the data within the existing mbuf chain because - * the underlying buffers may contain other rpc commands and we cannot afford - * to overwrite them. - * - * We would prefer to avoid this situation entirely. The situation does not - * occur with NFS/UDP and is supposed to only occassionally occur with TCP. - * Use vfs.nfs_common.realign_count and realign_test to check this. - */ -int -nfs_realign(struct mbuf **pm, int how) -{ - struct mbuf *m, *n; - int off; - - ++nfs_realign_test; - while ((m = *pm) != NULL) { - if (!nfsm_aligned(m->m_len, u_int32_t) || - !nfsm_aligned(mtod(m, intptr_t), u_int32_t)) { - /* - * NB: we can't depend on m_pkthdr.len to help us - * decide what to do here. May not be worth doing - * the m_length calculation as m_copyback will - * expand the mbuf chain below as needed. - */ - if (m_length(m, NULL) >= MINCLSIZE) { - /* NB: m_copyback handles space > MCLBYTES */ - n = m_getcl(how, MT_DATA, 0); - } else - n = m_get(how, MT_DATA); - if (n == NULL) - return (ENOMEM); - /* - * Align the remainder of the mbuf chain. - */ - n->m_len = 0; - off = 0; - while (m != NULL) { - m_copyback(n, off, m->m_len, mtod(m, caddr_t)); - off += m->m_len; - m = m->m_next; - } - m_freem(*pm); - *pm = n; - ++nfs_realign_count; - break; - } - pm = &m->m_next; - } - return (0); -} - -static moduledata_t nfs_common_mod = { - "nfs_common", - NULL, - NULL -}; - -DECLARE_MODULE(nfs_common, nfs_common_mod, SI_SUB_VFS, SI_ORDER_ANY); -MODULE_VERSION(nfs_common, 1); diff --git a/sys/nfsclient/nfs_bio.c b/sys/nfsclient/nfs_bio.c deleted file mode 100644 index 1a5b156..0000000 --- a/sys/nfsclient/nfs_bio.c +++ /dev/null @@ -1,1794 +0,0 @@ -/*- - * Copyright (c) 1989, 1993 - * The Regents of the University of California. All rights reserved. - * - * This code is derived from software contributed to Berkeley by - * Rick Macklem at The University of Guelph. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 4. Neither the name of the University nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * @(#)nfs_bio.c 8.9 (Berkeley) 3/30/95 - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/bio.h> -#include <sys/buf.h> -#include <sys/kernel.h> -#include <sys/mbuf.h> -#include <sys/mount.h> -#include <sys/proc.h> -#include <sys/rwlock.h> -#include <sys/vmmeter.h> -#include <sys/vnode.h> - -#include <vm/vm.h> -#include <vm/vm_param.h> -#include <vm/vm_extern.h> -#include <vm/vm_page.h> -#include <vm/vm_object.h> -#include <vm/vm_pager.h> -#include <vm/vnode_pager.h> - -#include <nfs/nfsproto.h> -#include <nfsclient/nfs.h> -#include <nfsclient/nfsmount.h> -#include <nfsclient/nfsnode.h> -#include <nfs/nfs_kdtrace.h> - -static struct buf *nfs_getcacheblk(struct vnode *vp, daddr_t bn, int size, - struct thread *td); -static int nfs_directio_write(struct vnode *vp, struct uio *uiop, - struct ucred *cred, int ioflag); - -extern int nfs_directio_enable; -extern int nfs_directio_allow_mmap; - -/* - * Vnode op for VM getpages. - */ -int -nfs_getpages(struct vop_getpages_args *ap) -{ - int i, error, nextoff, size, toff, count, npages; - struct uio uio; - struct iovec iov; - vm_offset_t kva; - struct buf *bp; - struct vnode *vp; - struct thread *td; - struct ucred *cred; - struct nfsmount *nmp; - vm_object_t object; - vm_page_t *pages; - struct nfsnode *np; - - vp = ap->a_vp; - np = VTONFS(vp); - td = curthread; /* XXX */ - cred = curthread->td_ucred; /* XXX */ - nmp = VFSTONFS(vp->v_mount); - pages = ap->a_m; - count = ap->a_count; - - if ((object = vp->v_object) == NULL) { - nfs_printf("nfs_getpages: called with non-merged cache vnode??\n"); - return (VM_PAGER_ERROR); - } - - if (nfs_directio_enable && !nfs_directio_allow_mmap) { - mtx_lock(&np->n_mtx); - if ((np->n_flag & NNONCACHE) && (vp->v_type == VREG)) { - mtx_unlock(&np->n_mtx); - nfs_printf("nfs_getpages: called on non-cacheable vnode??\n"); - return (VM_PAGER_ERROR); - } else - mtx_unlock(&np->n_mtx); - } - - mtx_lock(&nmp->nm_mtx); - if ((nmp->nm_flag & NFSMNT_NFSV3) != 0 && - (nmp->nm_state & NFSSTA_GOTFSINFO) == 0) { - mtx_unlock(&nmp->nm_mtx); - /* We'll never get here for v4, because we always have fsinfo */ - (void)nfs_fsinfo(nmp, vp, cred, td); - } else - mtx_unlock(&nmp->nm_mtx); - - npages = btoc(count); - - /* - * Since the caller has busied the requested page, that page's valid - * field will not be changed by other threads. - */ - vm_page_assert_xbusied(pages[ap->a_reqpage]); - - /* - * If the requested page is partially valid, just return it and - * allow the pager to zero-out the blanks. Partially valid pages - * can only occur at the file EOF. - */ - if (pages[ap->a_reqpage]->valid != 0) { - vm_pager_free_nonreq(object, pages, ap->a_reqpage, npages); - return (VM_PAGER_OK); - } - - /* - * We use only the kva address for the buffer, but this is extremely - * convienient and fast. - */ - bp = getpbuf(&nfs_pbuf_freecnt); - - kva = (vm_offset_t) bp->b_data; - pmap_qenter(kva, pages, npages); - PCPU_INC(cnt.v_vnodein); - PCPU_ADD(cnt.v_vnodepgsin, npages); - - iov.iov_base = (caddr_t) kva; - iov.iov_len = count; - uio.uio_iov = &iov; - uio.uio_iovcnt = 1; - uio.uio_offset = IDX_TO_OFF(pages[0]->pindex); - uio.uio_resid = count; - uio.uio_segflg = UIO_SYSSPACE; - uio.uio_rw = UIO_READ; - uio.uio_td = td; - - error = (nmp->nm_rpcops->nr_readrpc)(vp, &uio, cred); - pmap_qremove(kva, npages); - - relpbuf(bp, &nfs_pbuf_freecnt); - - if (error && (uio.uio_resid == count)) { - nfs_printf("nfs_getpages: error %d\n", error); - vm_pager_free_nonreq(object, pages, ap->a_reqpage, npages); - return (VM_PAGER_ERROR); - } - - /* - * Calculate the number of bytes read and validate only that number - * of bytes. Note that due to pending writes, size may be 0. This - * does not mean that the remaining data is invalid! - */ - - size = count - uio.uio_resid; - VM_OBJECT_WLOCK(object); - for (i = 0, toff = 0; i < npages; i++, toff = nextoff) { - vm_page_t m; - nextoff = toff + PAGE_SIZE; - m = pages[i]; - - if (nextoff <= size) { - /* - * Read operation filled an entire page - */ - m->valid = VM_PAGE_BITS_ALL; - KASSERT(m->dirty == 0, - ("nfs_getpages: page %p is dirty", m)); - } else if (size > toff) { - /* - * Read operation filled a partial page. - */ - m->valid = 0; - vm_page_set_valid_range(m, 0, size - toff); - KASSERT(m->dirty == 0, - ("nfs_getpages: page %p is dirty", m)); - } else { - /* - * Read operation was short. If no error - * occured we may have hit a zero-fill - * section. We leave valid set to 0, and page - * is freed by vm_page_readahead_finish() if - * its index is not equal to requested, or - * page is zeroed and set valid by - * vm_pager_get_pages() for requested page. - */ - ; - } - if (i != ap->a_reqpage) - vm_page_readahead_finish(m); - } - VM_OBJECT_WUNLOCK(object); - return (0); -} - -/* - * Vnode op for VM putpages. - */ -int -nfs_putpages(struct vop_putpages_args *ap) -{ - struct uio uio; - struct iovec iov; - vm_offset_t kva; - struct buf *bp; - int iomode, must_commit, i, error, npages, count; - off_t offset; - int *rtvals; - struct vnode *vp; - struct thread *td; - struct ucred *cred; - struct nfsmount *nmp; - struct nfsnode *np; - vm_page_t *pages; - - vp = ap->a_vp; - np = VTONFS(vp); - td = curthread; /* XXX */ - /* Set the cred to n_writecred for the write rpcs. */ - if (np->n_writecred != NULL) - cred = crhold(np->n_writecred); - else - cred = crhold(curthread->td_ucred); /* XXX */ - nmp = VFSTONFS(vp->v_mount); - pages = ap->a_m; - count = ap->a_count; - rtvals = ap->a_rtvals; - npages = btoc(count); - offset = IDX_TO_OFF(pages[0]->pindex); - - mtx_lock(&nmp->nm_mtx); - if ((nmp->nm_flag & NFSMNT_NFSV3) != 0 && - (nmp->nm_state & NFSSTA_GOTFSINFO) == 0) { - mtx_unlock(&nmp->nm_mtx); - (void)nfs_fsinfo(nmp, vp, cred, td); - } else - mtx_unlock(&nmp->nm_mtx); - - mtx_lock(&np->n_mtx); - if (nfs_directio_enable && !nfs_directio_allow_mmap && - (np->n_flag & NNONCACHE) && (vp->v_type == VREG)) { - mtx_unlock(&np->n_mtx); - nfs_printf("nfs_putpages: called on noncache-able vnode??\n"); - mtx_lock(&np->n_mtx); - } - - for (i = 0; i < npages; i++) - rtvals[i] = VM_PAGER_ERROR; - - /* - * When putting pages, do not extend file past EOF. - */ - if (offset + count > np->n_size) { - count = np->n_size - offset; - if (count < 0) - count = 0; - } - mtx_unlock(&np->n_mtx); - - /* - * We use only the kva address for the buffer, but this is extremely - * convienient and fast. - */ - bp = getpbuf(&nfs_pbuf_freecnt); - - kva = (vm_offset_t) bp->b_data; - pmap_qenter(kva, pages, npages); - PCPU_INC(cnt.v_vnodeout); - PCPU_ADD(cnt.v_vnodepgsout, count); - - iov.iov_base = (caddr_t) kva; - iov.iov_len = count; - uio.uio_iov = &iov; - uio.uio_iovcnt = 1; - uio.uio_offset = offset; - uio.uio_resid = count; - uio.uio_segflg = UIO_SYSSPACE; - uio.uio_rw = UIO_WRITE; - uio.uio_td = td; - - if ((ap->a_sync & VM_PAGER_PUT_SYNC) == 0) - iomode = NFSV3WRITE_UNSTABLE; - else - iomode = NFSV3WRITE_FILESYNC; - - error = (nmp->nm_rpcops->nr_writerpc)(vp, &uio, cred, &iomode, &must_commit); - crfree(cred); - - pmap_qremove(kva, npages); - relpbuf(bp, &nfs_pbuf_freecnt); - - if (!error) { - vnode_pager_undirty_pages(pages, rtvals, count - uio.uio_resid); - if (must_commit) { - nfs_clearcommit(vp->v_mount); - } - } - return rtvals[0]; -} - -/* - * For nfs, cache consistency can only be maintained approximately. - * Although RFC1094 does not specify the criteria, the following is - * believed to be compatible with the reference port. - * For nfs: - * If the file's modify time on the server has changed since the - * last read rpc or you have written to the file, - * you may have lost data cache consistency with the - * server, so flush all of the file's data out of the cache. - * Then force a getattr rpc to ensure that you have up to date - * attributes. - * NB: This implies that cache data can be read when up to - * NFS_ATTRTIMEO seconds out of date. If you find that you need current - * attributes this could be forced by setting n_attrstamp to 0 before - * the VOP_GETATTR() call. - */ -static inline int -nfs_bioread_check_cons(struct vnode *vp, struct thread *td, struct ucred *cred) -{ - int error = 0; - struct vattr vattr; - struct nfsnode *np = VTONFS(vp); - int old_lock; - struct nfsmount *nmp = VFSTONFS(vp->v_mount); - - /* - * Grab the exclusive lock before checking whether the cache is - * consistent. - * XXX - We can make this cheaper later (by acquiring cheaper locks). - * But for now, this suffices. - */ - old_lock = nfs_upgrade_vnlock(vp); - if (vp->v_iflag & VI_DOOMED) { - nfs_downgrade_vnlock(vp, old_lock); - return (EBADF); - } - - mtx_lock(&np->n_mtx); - if (np->n_flag & NMODIFIED) { - mtx_unlock(&np->n_mtx); - if (vp->v_type != VREG) { - if (vp->v_type != VDIR) - panic("nfs: bioread, not dir"); - (nmp->nm_rpcops->nr_invaldir)(vp); - error = nfs_vinvalbuf(vp, V_SAVE, td, 1); - if (error) - goto out; - } - np->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp); - error = VOP_GETATTR(vp, &vattr, cred); - if (error) - goto out; - mtx_lock(&np->n_mtx); - np->n_mtime = vattr.va_mtime; - mtx_unlock(&np->n_mtx); - } else { - mtx_unlock(&np->n_mtx); - error = VOP_GETATTR(vp, &vattr, cred); - if (error) - return (error); - mtx_lock(&np->n_mtx); - if ((np->n_flag & NSIZECHANGED) - || (NFS_TIMESPEC_COMPARE(&np->n_mtime, &vattr.va_mtime))) { - mtx_unlock(&np->n_mtx); - if (vp->v_type == VDIR) - (nmp->nm_rpcops->nr_invaldir)(vp); - error = nfs_vinvalbuf(vp, V_SAVE, td, 1); - if (error) - goto out; - mtx_lock(&np->n_mtx); - np->n_mtime = vattr.va_mtime; - np->n_flag &= ~NSIZECHANGED; - } - mtx_unlock(&np->n_mtx); - } -out: - nfs_downgrade_vnlock(vp, old_lock); - return error; -} - -/* - * Vnode op for read using bio - */ -int -nfs_bioread(struct vnode *vp, struct uio *uio, int ioflag, struct ucred *cred) -{ - struct nfsnode *np = VTONFS(vp); - int biosize, i; - struct buf *bp, *rabp; - struct thread *td; - struct nfsmount *nmp = VFSTONFS(vp->v_mount); - daddr_t lbn, rabn; - off_t end; - int bcount; - int seqcount; - int nra, error = 0, n = 0, on = 0; - - KASSERT(uio->uio_rw == UIO_READ, ("nfs_read mode")); - if (uio->uio_resid == 0) - return (0); - if (uio->uio_offset < 0) /* XXX VDIR cookies can be negative */ - return (EINVAL); - td = uio->uio_td; - - mtx_lock(&nmp->nm_mtx); - if ((nmp->nm_flag & NFSMNT_NFSV3) != 0 && - (nmp->nm_state & NFSSTA_GOTFSINFO) == 0) { - mtx_unlock(&nmp->nm_mtx); - (void)nfs_fsinfo(nmp, vp, cred, td); - } else - mtx_unlock(&nmp->nm_mtx); - - end = uio->uio_offset + uio->uio_resid; - if (vp->v_type != VDIR && - (end > nmp->nm_maxfilesize || end < uio->uio_offset)) - return (EFBIG); - - if (nfs_directio_enable && (ioflag & IO_DIRECT) && (vp->v_type == VREG)) - /* No caching/ no readaheads. Just read data into the user buffer */ - return nfs_readrpc(vp, uio, cred); - - biosize = vp->v_bufobj.bo_bsize; - seqcount = (int)((off_t)(ioflag >> IO_SEQSHIFT) * biosize / BKVASIZE); - - error = nfs_bioread_check_cons(vp, td, cred); - if (error) - return error; - - do { - u_quad_t nsize; - - mtx_lock(&np->n_mtx); - nsize = np->n_size; - mtx_unlock(&np->n_mtx); - - switch (vp->v_type) { - case VREG: - nfsstats.biocache_reads++; - lbn = uio->uio_offset / biosize; - on = uio->uio_offset - (lbn * biosize); - - /* - * Start the read ahead(s), as required. - */ - if (nmp->nm_readahead > 0) { - for (nra = 0; nra < nmp->nm_readahead && nra < seqcount && - (off_t)(lbn + 1 + nra) * biosize < nsize; nra++) { - rabn = lbn + 1 + nra; - if (incore(&vp->v_bufobj, rabn) == NULL) { - rabp = nfs_getcacheblk(vp, rabn, biosize, td); - if (!rabp) { - error = nfs_sigintr(nmp, td); - return (error ? error : EINTR); - } - if ((rabp->b_flags & (B_CACHE|B_DELWRI)) == 0) { - rabp->b_flags |= B_ASYNC; - rabp->b_iocmd = BIO_READ; - vfs_busy_pages(rabp, 0); - if (nfs_asyncio(nmp, rabp, cred, td)) { - rabp->b_flags |= B_INVAL; - rabp->b_ioflags |= BIO_ERROR; - vfs_unbusy_pages(rabp); - brelse(rabp); - break; - } - } else { - brelse(rabp); - } - } - } - } - - /* Note that bcount is *not* DEV_BSIZE aligned. */ - bcount = biosize; - if ((off_t)lbn * biosize >= nsize) { - bcount = 0; - } else if ((off_t)(lbn + 1) * biosize > nsize) { - bcount = nsize - (off_t)lbn * biosize; - } - bp = nfs_getcacheblk(vp, lbn, bcount, td); - - if (!bp) { - error = nfs_sigintr(nmp, td); - return (error ? error : EINTR); - } - - /* - * If B_CACHE is not set, we must issue the read. If this - * fails, we return an error. - */ - - if ((bp->b_flags & B_CACHE) == 0) { - bp->b_iocmd = BIO_READ; - vfs_busy_pages(bp, 0); - error = nfs_doio(vp, bp, cred, td); - if (error) { - brelse(bp); - return (error); - } - } - - /* - * on is the offset into the current bp. Figure out how many - * bytes we can copy out of the bp. Note that bcount is - * NOT DEV_BSIZE aligned. - * - * Then figure out how many bytes we can copy into the uio. - */ - - n = 0; - if (on < bcount) - n = MIN((unsigned)(bcount - on), uio->uio_resid); - break; - case VLNK: - nfsstats.biocache_readlinks++; - bp = nfs_getcacheblk(vp, (daddr_t)0, NFS_MAXPATHLEN, td); - if (!bp) { - error = nfs_sigintr(nmp, td); - return (error ? error : EINTR); - } - if ((bp->b_flags & B_CACHE) == 0) { - bp->b_iocmd = BIO_READ; - vfs_busy_pages(bp, 0); - error = nfs_doio(vp, bp, cred, td); - if (error) { - bp->b_ioflags |= BIO_ERROR; - brelse(bp); - return (error); - } - } - n = MIN(uio->uio_resid, NFS_MAXPATHLEN - bp->b_resid); - on = 0; - break; - case VDIR: - nfsstats.biocache_readdirs++; - if (np->n_direofoffset - && uio->uio_offset >= np->n_direofoffset) { - return (0); - } - lbn = (uoff_t)uio->uio_offset / NFS_DIRBLKSIZ; - on = uio->uio_offset & (NFS_DIRBLKSIZ - 1); - bp = nfs_getcacheblk(vp, lbn, NFS_DIRBLKSIZ, td); - if (!bp) { - error = nfs_sigintr(nmp, td); - return (error ? error : EINTR); - } - if ((bp->b_flags & B_CACHE) == 0) { - bp->b_iocmd = BIO_READ; - vfs_busy_pages(bp, 0); - error = nfs_doio(vp, bp, cred, td); - if (error) { - brelse(bp); - } - while (error == NFSERR_BAD_COOKIE) { - (nmp->nm_rpcops->nr_invaldir)(vp); - error = nfs_vinvalbuf(vp, 0, td, 1); - /* - * Yuck! The directory has been modified on the - * server. The only way to get the block is by - * reading from the beginning to get all the - * offset cookies. - * - * Leave the last bp intact unless there is an error. - * Loop back up to the while if the error is another - * NFSERR_BAD_COOKIE (double yuch!). - */ - for (i = 0; i <= lbn && !error; i++) { - if (np->n_direofoffset - && (i * NFS_DIRBLKSIZ) >= np->n_direofoffset) - return (0); - bp = nfs_getcacheblk(vp, i, NFS_DIRBLKSIZ, td); - if (!bp) { - error = nfs_sigintr(nmp, td); - return (error ? error : EINTR); - } - if ((bp->b_flags & B_CACHE) == 0) { - bp->b_iocmd = BIO_READ; - vfs_busy_pages(bp, 0); - error = nfs_doio(vp, bp, cred, td); - /* - * no error + B_INVAL == directory EOF, - * use the block. - */ - if (error == 0 && (bp->b_flags & B_INVAL)) - break; - } - /* - * An error will throw away the block and the - * for loop will break out. If no error and this - * is not the block we want, we throw away the - * block and go for the next one via the for loop. - */ - if (error || i < lbn) - brelse(bp); - } - } - /* - * The above while is repeated if we hit another cookie - * error. If we hit an error and it wasn't a cookie error, - * we give up. - */ - if (error) - return (error); - } - - /* - * If not eof and read aheads are enabled, start one. - * (You need the current block first, so that you have the - * directory offset cookie of the next block.) - */ - if (nmp->nm_readahead > 0 && - (bp->b_flags & B_INVAL) == 0 && - (np->n_direofoffset == 0 || - (lbn + 1) * NFS_DIRBLKSIZ < np->n_direofoffset) && - incore(&vp->v_bufobj, lbn + 1) == NULL) { - rabp = nfs_getcacheblk(vp, lbn + 1, NFS_DIRBLKSIZ, td); - if (rabp) { - if ((rabp->b_flags & (B_CACHE|B_DELWRI)) == 0) { - rabp->b_flags |= B_ASYNC; - rabp->b_iocmd = BIO_READ; - vfs_busy_pages(rabp, 0); - if (nfs_asyncio(nmp, rabp, cred, td)) { - rabp->b_flags |= B_INVAL; - rabp->b_ioflags |= BIO_ERROR; - vfs_unbusy_pages(rabp); - brelse(rabp); - } - } else { - brelse(rabp); - } - } - } - /* - * Unlike VREG files, whos buffer size ( bp->b_bcount ) is - * chopped for the EOF condition, we cannot tell how large - * NFS directories are going to be until we hit EOF. So - * an NFS directory buffer is *not* chopped to its EOF. Now, - * it just so happens that b_resid will effectively chop it - * to EOF. *BUT* this information is lost if the buffer goes - * away and is reconstituted into a B_CACHE state ( due to - * being VMIO ) later. So we keep track of the directory eof - * in np->n_direofoffset and chop it off as an extra step - * right here. - */ - n = lmin(uio->uio_resid, NFS_DIRBLKSIZ - bp->b_resid - on); - if (np->n_direofoffset && n > np->n_direofoffset - uio->uio_offset) - n = np->n_direofoffset - uio->uio_offset; - break; - default: - nfs_printf(" nfs_bioread: type %x unexpected\n", vp->v_type); - bp = NULL; - break; - }; - - if (n > 0) { - error = uiomove(bp->b_data + on, (int)n, uio); - } - if (vp->v_type == VLNK) - n = 0; - if (bp != NULL) - brelse(bp); - } while (error == 0 && uio->uio_resid > 0 && n > 0); - return (error); -} - -/* - * The NFS write path cannot handle iovecs with len > 1. So we need to - * break up iovecs accordingly (restricting them to wsize). - * For the SYNC case, we can do this with 1 copy (user buffer -> mbuf). - * For the ASYNC case, 2 copies are needed. The first a copy from the - * user buffer to a staging buffer and then a second copy from the staging - * buffer to mbufs. This can be optimized by copying from the user buffer - * directly into mbufs and passing the chain down, but that requires a - * fair amount of re-working of the relevant codepaths (and can be done - * later). - */ -static int -nfs_directio_write(vp, uiop, cred, ioflag) - struct vnode *vp; - struct uio *uiop; - struct ucred *cred; - int ioflag; -{ - int error; - struct nfsmount *nmp = VFSTONFS(vp->v_mount); - struct thread *td = uiop->uio_td; - int size; - int wsize; - - mtx_lock(&nmp->nm_mtx); - wsize = nmp->nm_wsize; - mtx_unlock(&nmp->nm_mtx); - if (ioflag & IO_SYNC) { - int iomode, must_commit; - struct uio uio; - struct iovec iov; -do_sync: - while (uiop->uio_resid > 0) { - size = MIN(uiop->uio_resid, wsize); - size = MIN(uiop->uio_iov->iov_len, size); - iov.iov_base = uiop->uio_iov->iov_base; - iov.iov_len = size; - uio.uio_iov = &iov; - uio.uio_iovcnt = 1; - uio.uio_offset = uiop->uio_offset; - uio.uio_resid = size; - uio.uio_segflg = UIO_USERSPACE; - uio.uio_rw = UIO_WRITE; - uio.uio_td = td; - iomode = NFSV3WRITE_FILESYNC; - error = (nmp->nm_rpcops->nr_writerpc)(vp, &uio, cred, - &iomode, &must_commit); - KASSERT((must_commit == 0), - ("nfs_directio_write: Did not commit write")); - if (error) - return (error); - uiop->uio_offset += size; - uiop->uio_resid -= size; - if (uiop->uio_iov->iov_len <= size) { - uiop->uio_iovcnt--; - uiop->uio_iov++; - } else { - uiop->uio_iov->iov_base = - (char *)uiop->uio_iov->iov_base + size; - uiop->uio_iov->iov_len -= size; - } - } - } else { - struct uio *t_uio; - struct iovec *t_iov; - struct buf *bp; - - /* - * Break up the write into blocksize chunks and hand these - * over to nfsiod's for write back. - * Unfortunately, this incurs a copy of the data. Since - * the user could modify the buffer before the write is - * initiated. - * - * The obvious optimization here is that one of the 2 copies - * in the async write path can be eliminated by copying the - * data here directly into mbufs and passing the mbuf chain - * down. But that will require a fair amount of re-working - * of the code and can be done if there's enough interest - * in NFS directio access. - */ - while (uiop->uio_resid > 0) { - size = MIN(uiop->uio_resid, wsize); - size = MIN(uiop->uio_iov->iov_len, size); - bp = getpbuf(&nfs_pbuf_freecnt); - t_uio = malloc(sizeof(struct uio), M_NFSDIRECTIO, M_WAITOK); - t_iov = malloc(sizeof(struct iovec), M_NFSDIRECTIO, M_WAITOK); - t_iov->iov_base = malloc(size, M_NFSDIRECTIO, M_WAITOK); - t_iov->iov_len = size; - t_uio->uio_iov = t_iov; - t_uio->uio_iovcnt = 1; - t_uio->uio_offset = uiop->uio_offset; - t_uio->uio_resid = size; - t_uio->uio_segflg = UIO_SYSSPACE; - t_uio->uio_rw = UIO_WRITE; - t_uio->uio_td = td; - KASSERT(uiop->uio_segflg == UIO_USERSPACE || - uiop->uio_segflg == UIO_SYSSPACE, - ("nfs_directio_write: Bad uio_segflg")); - if (uiop->uio_segflg == UIO_USERSPACE) { - error = copyin(uiop->uio_iov->iov_base, - t_iov->iov_base, size); - if (error != 0) - goto err_free; - } else - /* - * UIO_SYSSPACE may never happen, but handle - * it just in case it does. - */ - bcopy(uiop->uio_iov->iov_base, t_iov->iov_base, - size); - bp->b_flags |= B_DIRECT; - bp->b_iocmd = BIO_WRITE; - if (cred != NOCRED) { - crhold(cred); - bp->b_wcred = cred; - } else - bp->b_wcred = NOCRED; - bp->b_caller1 = (void *)t_uio; - bp->b_vp = vp; - error = nfs_asyncio(nmp, bp, NOCRED, td); -err_free: - if (error) { - free(t_iov->iov_base, M_NFSDIRECTIO); - free(t_iov, M_NFSDIRECTIO); - free(t_uio, M_NFSDIRECTIO); - bp->b_vp = NULL; - relpbuf(bp, &nfs_pbuf_freecnt); - if (error == EINTR) - return (error); - goto do_sync; - } - uiop->uio_offset += size; - uiop->uio_resid -= size; - if (uiop->uio_iov->iov_len <= size) { - uiop->uio_iovcnt--; - uiop->uio_iov++; - } else { - uiop->uio_iov->iov_base = - (char *)uiop->uio_iov->iov_base + size; - uiop->uio_iov->iov_len -= size; - } - } - } - return (0); -} - -/* - * Vnode op for write using bio - */ -int -nfs_write(struct vop_write_args *ap) -{ - int biosize; - struct uio *uio = ap->a_uio; - struct thread *td = uio->uio_td; - struct vnode *vp = ap->a_vp; - struct nfsnode *np = VTONFS(vp); - struct ucred *cred = ap->a_cred; - int ioflag = ap->a_ioflag; - struct buf *bp; - struct vattr vattr; - struct nfsmount *nmp = VFSTONFS(vp->v_mount); - daddr_t lbn; - off_t end; - int bcount; - int n, on, error = 0; - - KASSERT(uio->uio_rw == UIO_WRITE, ("nfs_write mode")); - KASSERT(uio->uio_segflg != UIO_USERSPACE || uio->uio_td == curthread, - ("nfs_write proc")); - if (vp->v_type != VREG) - return (EIO); - mtx_lock(&np->n_mtx); - if (np->n_flag & NWRITEERR) { - np->n_flag &= ~NWRITEERR; - mtx_unlock(&np->n_mtx); - return (np->n_error); - } else - mtx_unlock(&np->n_mtx); - mtx_lock(&nmp->nm_mtx); - if ((nmp->nm_flag & NFSMNT_NFSV3) != 0 && - (nmp->nm_state & NFSSTA_GOTFSINFO) == 0) { - mtx_unlock(&nmp->nm_mtx); - (void)nfs_fsinfo(nmp, vp, cred, td); - } else - mtx_unlock(&nmp->nm_mtx); - - /* - * Synchronously flush pending buffers if we are in synchronous - * mode or if we are appending. - */ - if (ioflag & (IO_APPEND | IO_SYNC)) { - mtx_lock(&np->n_mtx); - if (np->n_flag & NMODIFIED) { - mtx_unlock(&np->n_mtx); -#ifdef notyet /* Needs matching nonblock semantics elsewhere, too. */ - /* - * Require non-blocking, synchronous writes to - * dirty files to inform the program it needs - * to fsync(2) explicitly. - */ - if (ioflag & IO_NDELAY) - return (EAGAIN); -#endif -flush_and_restart: - np->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp); - error = nfs_vinvalbuf(vp, V_SAVE, td, 1); - if (error) - return (error); - } else - mtx_unlock(&np->n_mtx); - } - - /* - * If IO_APPEND then load uio_offset. We restart here if we cannot - * get the append lock. - */ - if (ioflag & IO_APPEND) { - np->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp); - error = VOP_GETATTR(vp, &vattr, cred); - if (error) - return (error); - mtx_lock(&np->n_mtx); - uio->uio_offset = np->n_size; - mtx_unlock(&np->n_mtx); - } - - if (uio->uio_offset < 0) - return (EINVAL); - end = uio->uio_offset + uio->uio_resid; - if (end > nmp->nm_maxfilesize || end < uio->uio_offset) - return (EFBIG); - if (uio->uio_resid == 0) - return (0); - - if (nfs_directio_enable && (ioflag & IO_DIRECT) && vp->v_type == VREG) - return nfs_directio_write(vp, uio, cred, ioflag); - - /* - * Maybe this should be above the vnode op call, but so long as - * file servers have no limits, i don't think it matters - */ - if (vn_rlimit_fsize(vp, uio, td)) - return (EFBIG); - - biosize = vp->v_bufobj.bo_bsize; - /* - * Find all of this file's B_NEEDCOMMIT buffers. If our writes - * would exceed the local maximum per-file write commit size when - * combined with those, we must decide whether to flush, - * go synchronous, or return error. We don't bother checking - * IO_UNIT -- we just make all writes atomic anyway, as there's - * no point optimizing for something that really won't ever happen. - */ - if (!(ioflag & IO_SYNC)) { - int nflag; - - mtx_lock(&np->n_mtx); - nflag = np->n_flag; - mtx_unlock(&np->n_mtx); - int needrestart = 0; - if (nmp->nm_wcommitsize < uio->uio_resid) { - /* - * If this request could not possibly be completed - * without exceeding the maximum outstanding write - * commit size, see if we can convert it into a - * synchronous write operation. - */ - if (ioflag & IO_NDELAY) - return (EAGAIN); - ioflag |= IO_SYNC; - if (nflag & NMODIFIED) - needrestart = 1; - } else if (nflag & NMODIFIED) { - int wouldcommit = 0; - BO_LOCK(&vp->v_bufobj); - if (vp->v_bufobj.bo_dirty.bv_cnt != 0) { - TAILQ_FOREACH(bp, &vp->v_bufobj.bo_dirty.bv_hd, - b_bobufs) { - if (bp->b_flags & B_NEEDCOMMIT) - wouldcommit += bp->b_bcount; - } - } - BO_UNLOCK(&vp->v_bufobj); - /* - * Since we're not operating synchronously and - * bypassing the buffer cache, we are in a commit - * and holding all of these buffers whether - * transmitted or not. If not limited, this - * will lead to the buffer cache deadlocking, - * as no one else can flush our uncommitted buffers. - */ - wouldcommit += uio->uio_resid; - /* - * If we would initially exceed the maximum - * outstanding write commit size, flush and restart. - */ - if (wouldcommit > nmp->nm_wcommitsize) - needrestart = 1; - } - if (needrestart) - goto flush_and_restart; - } - - do { - nfsstats.biocache_writes++; - lbn = uio->uio_offset / biosize; - on = uio->uio_offset - (lbn * biosize); - n = MIN((unsigned)(biosize - on), uio->uio_resid); -again: - /* - * Handle direct append and file extension cases, calculate - * unaligned buffer size. - */ - mtx_lock(&np->n_mtx); - if (uio->uio_offset == np->n_size && n) { - mtx_unlock(&np->n_mtx); - /* - * Get the buffer (in its pre-append state to maintain - * B_CACHE if it was previously set). Resize the - * nfsnode after we have locked the buffer to prevent - * readers from reading garbage. - */ - bcount = on; - bp = nfs_getcacheblk(vp, lbn, bcount, td); - - if (bp != NULL) { - long save; - - mtx_lock(&np->n_mtx); - np->n_size = uio->uio_offset + n; - np->n_flag |= NMODIFIED; - vnode_pager_setsize(vp, np->n_size); - mtx_unlock(&np->n_mtx); - - save = bp->b_flags & B_CACHE; - bcount += n; - allocbuf(bp, bcount); - bp->b_flags |= save; - } - } else { - /* - * Obtain the locked cache block first, and then - * adjust the file's size as appropriate. - */ - bcount = on + n; - if ((off_t)lbn * biosize + bcount < np->n_size) { - if ((off_t)(lbn + 1) * biosize < np->n_size) - bcount = biosize; - else - bcount = np->n_size - (off_t)lbn * biosize; - } - mtx_unlock(&np->n_mtx); - bp = nfs_getcacheblk(vp, lbn, bcount, td); - mtx_lock(&np->n_mtx); - if (uio->uio_offset + n > np->n_size) { - np->n_size = uio->uio_offset + n; - np->n_flag |= NMODIFIED; - vnode_pager_setsize(vp, np->n_size); - } - mtx_unlock(&np->n_mtx); - } - - if (!bp) { - error = nfs_sigintr(nmp, td); - if (!error) - error = EINTR; - break; - } - - /* - * Issue a READ if B_CACHE is not set. In special-append - * mode, B_CACHE is based on the buffer prior to the write - * op and is typically set, avoiding the read. If a read - * is required in special append mode, the server will - * probably send us a short-read since we extended the file - * on our end, resulting in b_resid == 0 and, thusly, - * B_CACHE getting set. - * - * We can also avoid issuing the read if the write covers - * the entire buffer. We have to make sure the buffer state - * is reasonable in this case since we will not be initiating - * I/O. See the comments in kern/vfs_bio.c's getblk() for - * more information. - * - * B_CACHE may also be set due to the buffer being cached - * normally. - */ - - if (on == 0 && n == bcount) { - bp->b_flags |= B_CACHE; - bp->b_flags &= ~B_INVAL; - bp->b_ioflags &= ~BIO_ERROR; - } - - if ((bp->b_flags & B_CACHE) == 0) { - bp->b_iocmd = BIO_READ; - vfs_busy_pages(bp, 0); - error = nfs_doio(vp, bp, cred, td); - if (error) { - brelse(bp); - break; - } - } - if (bp->b_wcred == NOCRED) - bp->b_wcred = crhold(cred); - mtx_lock(&np->n_mtx); - np->n_flag |= NMODIFIED; - mtx_unlock(&np->n_mtx); - - /* - * If dirtyend exceeds file size, chop it down. This should - * not normally occur but there is an append race where it - * might occur XXX, so we log it. - * - * If the chopping creates a reverse-indexed or degenerate - * situation with dirtyoff/end, we 0 both of them. - */ - - if (bp->b_dirtyend > bcount) { - nfs_printf("NFS append race @%lx:%d\n", - (long)bp->b_blkno * DEV_BSIZE, - bp->b_dirtyend - bcount); - bp->b_dirtyend = bcount; - } - - if (bp->b_dirtyoff >= bp->b_dirtyend) - bp->b_dirtyoff = bp->b_dirtyend = 0; - - /* - * If the new write will leave a contiguous dirty - * area, just update the b_dirtyoff and b_dirtyend, - * otherwise force a write rpc of the old dirty area. - * - * While it is possible to merge discontiguous writes due to - * our having a B_CACHE buffer ( and thus valid read data - * for the hole), we don't because it could lead to - * significant cache coherency problems with multiple clients, - * especially if locking is implemented later on. - * - * as an optimization we could theoretically maintain - * a linked list of discontinuous areas, but we would still - * have to commit them separately so there isn't much - * advantage to it except perhaps a bit of asynchronization. - */ - - if (bp->b_dirtyend > 0 && - (on > bp->b_dirtyend || (on + n) < bp->b_dirtyoff)) { - if (bwrite(bp) == EINTR) { - error = EINTR; - break; - } - goto again; - } - - error = uiomove((char *)bp->b_data + on, n, uio); - - /* - * Since this block is being modified, it must be written - * again and not just committed. Since write clustering does - * not work for the stage 1 data write, only the stage 2 - * commit rpc, we have to clear B_CLUSTEROK as well. - */ - bp->b_flags &= ~(B_NEEDCOMMIT | B_CLUSTEROK); - - if (error) { - bp->b_ioflags |= BIO_ERROR; - brelse(bp); - break; - } - - /* - * Only update dirtyoff/dirtyend if not a degenerate - * condition. - */ - if (n) { - if (bp->b_dirtyend > 0) { - bp->b_dirtyoff = min(on, bp->b_dirtyoff); - bp->b_dirtyend = max((on + n), bp->b_dirtyend); - } else { - bp->b_dirtyoff = on; - bp->b_dirtyend = on + n; - } - vfs_bio_set_valid(bp, on, n); - } - - /* - * If IO_SYNC do bwrite(). - * - * IO_INVAL appears to be unused. The idea appears to be - * to turn off caching in this case. Very odd. XXX - */ - if ((ioflag & IO_SYNC)) { - if (ioflag & IO_INVAL) - bp->b_flags |= B_NOCACHE; - error = bwrite(bp); - if (error) - break; - } else if ((n + on) == biosize) { - bp->b_flags |= B_ASYNC; - (void) (nmp->nm_rpcops->nr_writebp)(bp, 0, NULL); - } else { - bdwrite(bp); - } - } while (uio->uio_resid > 0 && n > 0); - - return (error); -} - -/* - * Get an nfs cache block. - * - * Allocate a new one if the block isn't currently in the cache - * and return the block marked busy. If the calling process is - * interrupted by a signal for an interruptible mount point, return - * NULL. - * - * The caller must carefully deal with the possible B_INVAL state of - * the buffer. nfs_doio() clears B_INVAL (and nfs_asyncio() clears it - * indirectly), so synchronous reads can be issued without worrying about - * the B_INVAL state. We have to be a little more careful when dealing - * with writes (see comments in nfs_write()) when extending a file past - * its EOF. - */ -static struct buf * -nfs_getcacheblk(struct vnode *vp, daddr_t bn, int size, struct thread *td) -{ - struct buf *bp; - struct mount *mp; - struct nfsmount *nmp; - - mp = vp->v_mount; - nmp = VFSTONFS(mp); - - if (nmp->nm_flag & NFSMNT_INT) { - sigset_t oldset; - - nfs_set_sigmask(td, &oldset); - bp = getblk(vp, bn, size, PCATCH, 0, 0); - nfs_restore_sigmask(td, &oldset); - while (bp == NULL) { - if (nfs_sigintr(nmp, td)) - return (NULL); - bp = getblk(vp, bn, size, 0, 2 * hz, 0); - } - } else { - bp = getblk(vp, bn, size, 0, 0, 0); - } - - if (vp->v_type == VREG) - bp->b_blkno = bn * (vp->v_bufobj.bo_bsize / DEV_BSIZE); - return (bp); -} - -/* - * Flush and invalidate all dirty buffers. If another process is already - * doing the flush, just wait for completion. - */ -int -nfs_vinvalbuf(struct vnode *vp, int flags, struct thread *td, int intrflg) -{ - struct nfsnode *np = VTONFS(vp); - struct nfsmount *nmp = VFSTONFS(vp->v_mount); - int error = 0, slpflag, slptimeo; - int old_lock = 0; - - ASSERT_VOP_LOCKED(vp, "nfs_vinvalbuf"); - - if ((nmp->nm_flag & NFSMNT_INT) == 0) - intrflg = 0; - if (intrflg) { - slpflag = PCATCH; - slptimeo = 2 * hz; - } else { - slpflag = 0; - slptimeo = 0; - } - - old_lock = nfs_upgrade_vnlock(vp); - if (vp->v_iflag & VI_DOOMED) { - /* - * Since vgonel() uses the generic vinvalbuf() to flush - * dirty buffers and it does not call this function, it - * is safe to just return OK when VI_DOOMED is set. - */ - nfs_downgrade_vnlock(vp, old_lock); - return (0); - } - - /* - * Now, flush as required. - */ - if ((flags & V_SAVE) && (vp->v_bufobj.bo_object != NULL)) { - VM_OBJECT_WLOCK(vp->v_bufobj.bo_object); - vm_object_page_clean(vp->v_bufobj.bo_object, 0, 0, OBJPC_SYNC); - VM_OBJECT_WUNLOCK(vp->v_bufobj.bo_object); - /* - * If the page clean was interrupted, fail the invalidation. - * Not doing so, we run the risk of losing dirty pages in the - * vinvalbuf() call below. - */ - if (intrflg && (error = nfs_sigintr(nmp, td))) - goto out; - } - - error = vinvalbuf(vp, flags, slpflag, 0); - while (error) { - if (intrflg && (error = nfs_sigintr(nmp, td))) - goto out; - error = vinvalbuf(vp, flags, 0, slptimeo); - } - mtx_lock(&np->n_mtx); - if (np->n_directio_asyncwr == 0) - np->n_flag &= ~NMODIFIED; - mtx_unlock(&np->n_mtx); -out: - nfs_downgrade_vnlock(vp, old_lock); - return error; -} - -/* - * Initiate asynchronous I/O. Return an error if no nfsiods are available. - * This is mainly to avoid queueing async I/O requests when the nfsiods - * are all hung on a dead server. - * - * Note: nfs_asyncio() does not clear (BIO_ERROR|B_INVAL) but when the bp - * is eventually dequeued by the async daemon, nfs_doio() *will*. - */ -int -nfs_asyncio(struct nfsmount *nmp, struct buf *bp, struct ucred *cred, struct thread *td) -{ - int iod; - int gotiod; - int slpflag = 0; - int slptimeo = 0; - int error, error2; - - /* - * Commits are usually short and sweet so lets save some cpu and - * leave the async daemons for more important rpc's (such as reads - * and writes). - * - * Readdirplus RPCs do vget()s to acquire the vnodes for entries - * in the directory in order to update attributes. This can deadlock - * with another thread that is waiting for async I/O to be done by - * an nfsiod thread while holding a lock on one of these vnodes. - * To avoid this deadlock, don't allow the async nfsiod threads to - * perform Readdirplus RPCs. - */ - mtx_lock(&nfs_iod_mtx); - if ((bp->b_iocmd == BIO_WRITE && (bp->b_flags & B_NEEDCOMMIT) && - (nmp->nm_bufqiods > nfs_numasync / 2)) || - (bp->b_vp->v_type == VDIR && (nmp->nm_flag & NFSMNT_RDIRPLUS))) { - mtx_unlock(&nfs_iod_mtx); - return(EIO); - } -again: - if (nmp->nm_flag & NFSMNT_INT) - slpflag = PCATCH; - gotiod = FALSE; - - /* - * Find a free iod to process this request. - */ - for (iod = 0; iod < nfs_numasync; iod++) - if (nfs_iodwant[iod] == NFSIOD_AVAILABLE) { - gotiod = TRUE; - break; - } - - /* - * Try to create one if none are free. - */ - if (!gotiod) - nfs_nfsiodnew(); - else { - /* - * Found one, so wake it up and tell it which - * mount to process. - */ - NFS_DPF(ASYNCIO, ("nfs_asyncio: waking iod %d for mount %p\n", - iod, nmp)); - nfs_iodwant[iod] = NFSIOD_NOT_AVAILABLE; - nfs_iodmount[iod] = nmp; - nmp->nm_bufqiods++; - wakeup(&nfs_iodwant[iod]); - } - - /* - * If none are free, we may already have an iod working on this mount - * point. If so, it will process our request. - */ - if (!gotiod) { - if (nmp->nm_bufqiods > 0) { - NFS_DPF(ASYNCIO, - ("nfs_asyncio: %d iods are already processing mount %p\n", - nmp->nm_bufqiods, nmp)); - gotiod = TRUE; - } - } - - /* - * If we have an iod which can process the request, then queue - * the buffer. - */ - if (gotiod) { - /* - * Ensure that the queue never grows too large. We still want - * to asynchronize so we block rather then return EIO. - */ - while (nmp->nm_bufqlen >= 2 * nfs_numasync) { - NFS_DPF(ASYNCIO, - ("nfs_asyncio: waiting for mount %p queue to drain\n", nmp)); - nmp->nm_bufqwant = TRUE; - error = nfs_msleep(td, &nmp->nm_bufq, &nfs_iod_mtx, - slpflag | PRIBIO, - "nfsaio", slptimeo); - if (error) { - error2 = nfs_sigintr(nmp, td); - if (error2) { - mtx_unlock(&nfs_iod_mtx); - return (error2); - } - if (slpflag == PCATCH) { - slpflag = 0; - slptimeo = 2 * hz; - } - } - /* - * We might have lost our iod while sleeping, - * so check and loop if nescessary. - */ - goto again; - } - - /* We might have lost our nfsiod */ - if (nmp->nm_bufqiods == 0) { - NFS_DPF(ASYNCIO, -("nfs_asyncio: no iods after mount %p queue was drained, looping\n", nmp)); - goto again; - } - - if (bp->b_iocmd == BIO_READ) { - if (bp->b_rcred == NOCRED && cred != NOCRED) - bp->b_rcred = crhold(cred); - } else { - if (bp->b_wcred == NOCRED && cred != NOCRED) - bp->b_wcred = crhold(cred); - } - - if (bp->b_flags & B_REMFREE) - bremfreef(bp); - BUF_KERNPROC(bp); - TAILQ_INSERT_TAIL(&nmp->nm_bufq, bp, b_freelist); - nmp->nm_bufqlen++; - if ((bp->b_flags & B_DIRECT) && bp->b_iocmd == BIO_WRITE) { - mtx_lock(&(VTONFS(bp->b_vp))->n_mtx); - VTONFS(bp->b_vp)->n_flag |= NMODIFIED; - VTONFS(bp->b_vp)->n_directio_asyncwr++; - mtx_unlock(&(VTONFS(bp->b_vp))->n_mtx); - } - mtx_unlock(&nfs_iod_mtx); - return (0); - } - - mtx_unlock(&nfs_iod_mtx); - - /* - * All the iods are busy on other mounts, so return EIO to - * force the caller to process the i/o synchronously. - */ - NFS_DPF(ASYNCIO, ("nfs_asyncio: no iods available, i/o is synchronous\n")); - return (EIO); -} - -void -nfs_doio_directwrite(struct buf *bp) -{ - int iomode, must_commit; - struct uio *uiop = (struct uio *)bp->b_caller1; - char *iov_base = uiop->uio_iov->iov_base; - struct nfsmount *nmp = VFSTONFS(bp->b_vp->v_mount); - - iomode = NFSV3WRITE_FILESYNC; - uiop->uio_td = NULL; /* NULL since we're in nfsiod */ - (nmp->nm_rpcops->nr_writerpc)(bp->b_vp, uiop, bp->b_wcred, &iomode, &must_commit); - KASSERT((must_commit == 0), ("nfs_doio_directwrite: Did not commit write")); - free(iov_base, M_NFSDIRECTIO); - free(uiop->uio_iov, M_NFSDIRECTIO); - free(uiop, M_NFSDIRECTIO); - if ((bp->b_flags & B_DIRECT) && bp->b_iocmd == BIO_WRITE) { - struct nfsnode *np = VTONFS(bp->b_vp); - mtx_lock(&np->n_mtx); - np->n_directio_asyncwr--; - if (np->n_directio_asyncwr == 0) { - VTONFS(bp->b_vp)->n_flag &= ~NMODIFIED; - if ((np->n_flag & NFSYNCWAIT)) { - np->n_flag &= ~NFSYNCWAIT; - wakeup((caddr_t)&np->n_directio_asyncwr); - } - } - mtx_unlock(&np->n_mtx); - } - bp->b_vp = NULL; - relpbuf(bp, &nfs_pbuf_freecnt); -} - -/* - * Do an I/O operation to/from a cache block. This may be called - * synchronously or from an nfsiod. - */ -int -nfs_doio(struct vnode *vp, struct buf *bp, struct ucred *cr, struct thread *td) -{ - struct uio *uiop; - struct nfsnode *np; - struct nfsmount *nmp; - int error = 0, iomode, must_commit = 0; - struct uio uio; - struct iovec io; - struct proc *p = td ? td->td_proc : NULL; - uint8_t iocmd; - - np = VTONFS(vp); - nmp = VFSTONFS(vp->v_mount); - uiop = &uio; - uiop->uio_iov = &io; - uiop->uio_iovcnt = 1; - uiop->uio_segflg = UIO_SYSSPACE; - uiop->uio_td = td; - - /* - * clear BIO_ERROR and B_INVAL state prior to initiating the I/O. We - * do this here so we do not have to do it in all the code that - * calls us. - */ - bp->b_flags &= ~B_INVAL; - bp->b_ioflags &= ~BIO_ERROR; - - KASSERT(!(bp->b_flags & B_DONE), ("nfs_doio: bp %p already marked done", bp)); - iocmd = bp->b_iocmd; - if (iocmd == BIO_READ) { - io.iov_len = uiop->uio_resid = bp->b_bcount; - io.iov_base = bp->b_data; - uiop->uio_rw = UIO_READ; - - switch (vp->v_type) { - case VREG: - uiop->uio_offset = ((off_t)bp->b_blkno) * DEV_BSIZE; - nfsstats.read_bios++; - error = (nmp->nm_rpcops->nr_readrpc)(vp, uiop, cr); - - if (!error) { - if (uiop->uio_resid) { - /* - * If we had a short read with no error, we must have - * hit a file hole. We should zero-fill the remainder. - * This can also occur if the server hits the file EOF. - * - * Holes used to be able to occur due to pending - * writes, but that is not possible any longer. - */ - int nread = bp->b_bcount - uiop->uio_resid; - int left = uiop->uio_resid; - - if (left > 0) - bzero((char *)bp->b_data + nread, left); - uiop->uio_resid = 0; - } - } - /* ASSERT_VOP_LOCKED(vp, "nfs_doio"); */ - if (p && (vp->v_vflag & VV_TEXT)) { - mtx_lock(&np->n_mtx); - if (NFS_TIMESPEC_COMPARE(&np->n_mtime, &np->n_vattr.va_mtime)) { - mtx_unlock(&np->n_mtx); - PROC_LOCK(p); - killproc(p, "text file modification"); - PROC_UNLOCK(p); - } else - mtx_unlock(&np->n_mtx); - } - break; - case VLNK: - uiop->uio_offset = (off_t)0; - nfsstats.readlink_bios++; - error = (nmp->nm_rpcops->nr_readlinkrpc)(vp, uiop, cr); - break; - case VDIR: - nfsstats.readdir_bios++; - uiop->uio_offset = ((u_quad_t)bp->b_lblkno) * NFS_DIRBLKSIZ; - if ((nmp->nm_flag & NFSMNT_RDIRPLUS) != 0) { - error = nfs_readdirplusrpc(vp, uiop, cr); - if (error == NFSERR_NOTSUPP) - nmp->nm_flag &= ~NFSMNT_RDIRPLUS; - } - if ((nmp->nm_flag & NFSMNT_RDIRPLUS) == 0) - error = nfs_readdirrpc(vp, uiop, cr); - /* - * end-of-directory sets B_INVAL but does not generate an - * error. - */ - if (error == 0 && uiop->uio_resid == bp->b_bcount) - bp->b_flags |= B_INVAL; - break; - default: - nfs_printf("nfs_doio: type %x unexpected\n", vp->v_type); - break; - }; - if (error) { - bp->b_ioflags |= BIO_ERROR; - bp->b_error = error; - } - } else { - /* - * If we only need to commit, try to commit - */ - if (bp->b_flags & B_NEEDCOMMIT) { - int retv; - off_t off; - - off = ((u_quad_t)bp->b_blkno) * DEV_BSIZE + bp->b_dirtyoff; - retv = (nmp->nm_rpcops->nr_commit)( - vp, off, bp->b_dirtyend-bp->b_dirtyoff, - bp->b_wcred, td); - if (retv == 0) { - bp->b_dirtyoff = bp->b_dirtyend = 0; - bp->b_flags &= ~(B_NEEDCOMMIT | B_CLUSTEROK); - bp->b_resid = 0; - bufdone(bp); - return (0); - } - if (retv == NFSERR_STALEWRITEVERF) { - nfs_clearcommit(vp->v_mount); - } - } - - /* - * Setup for actual write - */ - mtx_lock(&np->n_mtx); - if ((off_t)bp->b_blkno * DEV_BSIZE + bp->b_dirtyend > np->n_size) - bp->b_dirtyend = np->n_size - (off_t)bp->b_blkno * DEV_BSIZE; - mtx_unlock(&np->n_mtx); - - if (bp->b_dirtyend > bp->b_dirtyoff) { - io.iov_len = uiop->uio_resid = bp->b_dirtyend - - bp->b_dirtyoff; - uiop->uio_offset = (off_t)bp->b_blkno * DEV_BSIZE - + bp->b_dirtyoff; - io.iov_base = (char *)bp->b_data + bp->b_dirtyoff; - uiop->uio_rw = UIO_WRITE; - nfsstats.write_bios++; - - if ((bp->b_flags & (B_ASYNC | B_NEEDCOMMIT | B_NOCACHE | B_CLUSTER)) == B_ASYNC) - iomode = NFSV3WRITE_UNSTABLE; - else - iomode = NFSV3WRITE_FILESYNC; - - error = (nmp->nm_rpcops->nr_writerpc)(vp, uiop, cr, &iomode, &must_commit); - - /* - * When setting B_NEEDCOMMIT also set B_CLUSTEROK to try - * to cluster the buffers needing commit. This will allow - * the system to submit a single commit rpc for the whole - * cluster. We can do this even if the buffer is not 100% - * dirty (relative to the NFS blocksize), so we optimize the - * append-to-file-case. - * - * (when clearing B_NEEDCOMMIT, B_CLUSTEROK must also be - * cleared because write clustering only works for commit - * rpc's, not for the data portion of the write). - */ - - if (!error && iomode == NFSV3WRITE_UNSTABLE) { - bp->b_flags |= B_NEEDCOMMIT; - if (bp->b_dirtyoff == 0 - && bp->b_dirtyend == bp->b_bcount) - bp->b_flags |= B_CLUSTEROK; - } else { - bp->b_flags &= ~(B_NEEDCOMMIT | B_CLUSTEROK); - } - - /* - * For an interrupted write, the buffer is still valid - * and the write hasn't been pushed to the server yet, - * so we can't set BIO_ERROR and report the interruption - * by setting B_EINTR. For the B_ASYNC case, B_EINTR - * is not relevant, so the rpc attempt is essentially - * a noop. For the case of a V3 write rpc not being - * committed to stable storage, the block is still - * dirty and requires either a commit rpc or another - * write rpc with iomode == NFSV3WRITE_FILESYNC before - * the block is reused. This is indicated by setting - * the B_DELWRI and B_NEEDCOMMIT flags. - * - * If the buffer is marked B_PAGING, it does not reside on - * the vp's paging queues so we cannot call bdirty(). The - * bp in this case is not an NFS cache block so we should - * be safe. XXX - * - * The logic below breaks up errors into recoverable and - * unrecoverable. For the former, we clear B_INVAL|B_NOCACHE - * and keep the buffer around for potential write retries. - * For the latter (eg ESTALE), we toss the buffer away (B_INVAL) - * and save the error in the nfsnode. This is less than ideal - * but necessary. Keeping such buffers around could potentially - * cause buffer exhaustion eventually (they can never be written - * out, so will get constantly be re-dirtied). It also causes - * all sorts of vfs panics. For non-recoverable write errors, - * also invalidate the attrcache, so we'll be forced to go over - * the wire for this object, returning an error to user on next - * call (most of the time). - */ - if (error == EINTR || error == EIO || error == ETIMEDOUT - || (!error && (bp->b_flags & B_NEEDCOMMIT))) { - int s; - - s = splbio(); - bp->b_flags &= ~(B_INVAL|B_NOCACHE); - if ((bp->b_flags & B_PAGING) == 0) { - bdirty(bp); - bp->b_flags &= ~B_DONE; - } - if (error && (bp->b_flags & B_ASYNC) == 0) - bp->b_flags |= B_EINTR; - splx(s); - } else { - if (error) { - bp->b_ioflags |= BIO_ERROR; - bp->b_flags |= B_INVAL; - bp->b_error = np->n_error = error; - mtx_lock(&np->n_mtx); - np->n_flag |= NWRITEERR; - np->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp); - mtx_unlock(&np->n_mtx); - } - bp->b_dirtyoff = bp->b_dirtyend = 0; - } - } else { - bp->b_resid = 0; - bufdone(bp); - return (0); - } - } - bp->b_resid = uiop->uio_resid; - if (must_commit) - nfs_clearcommit(vp->v_mount); - bufdone(bp); - return (error); -} - -/* - * Used to aid in handling ftruncate() operations on the NFS client side. - * Truncation creates a number of special problems for NFS. We have to - * throw away VM pages and buffer cache buffers that are beyond EOF, and - * we have to properly handle VM pages or (potentially dirty) buffers - * that straddle the truncation point. - */ - -int -nfs_meta_setsize(struct vnode *vp, struct ucred *cred, struct thread *td, u_quad_t nsize) -{ - struct nfsnode *np = VTONFS(vp); - u_quad_t tsize; - int biosize = vp->v_bufobj.bo_bsize; - int error = 0; - - mtx_lock(&np->n_mtx); - tsize = np->n_size; - np->n_size = nsize; - mtx_unlock(&np->n_mtx); - - if (nsize < tsize) { - struct buf *bp; - daddr_t lbn; - int bufsize; - - /* - * vtruncbuf() doesn't get the buffer overlapping the - * truncation point. We may have a B_DELWRI and/or B_CACHE - * buffer that now needs to be truncated. - */ - error = vtruncbuf(vp, cred, nsize, biosize); - lbn = nsize / biosize; - bufsize = nsize - (lbn * biosize); - bp = nfs_getcacheblk(vp, lbn, bufsize, td); - if (!bp) - return EINTR; - if (bp->b_dirtyoff > bp->b_bcount) - bp->b_dirtyoff = bp->b_bcount; - if (bp->b_dirtyend > bp->b_bcount) - bp->b_dirtyend = bp->b_bcount; - bp->b_flags |= B_RELBUF; /* don't leave garbage around */ - brelse(bp); - } else { - vnode_pager_setsize(vp, nsize); - } - return(error); -} - diff --git a/sys/nfsclient/nfs_kdtrace.c b/sys/nfsclient/nfs_kdtrace.c deleted file mode 100644 index 429dbc3..0000000 --- a/sys/nfsclient/nfs_kdtrace.c +++ /dev/null @@ -1,542 +0,0 @@ -/*- - * Copyright (c) 2009 Robert N. M. Watson - * All rights reserved. - * - * This software was developed at the University of Cambridge Computer - * Laboratory with support from a grant from Google, Inc. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/conf.h> -#include <sys/kernel.h> -#include <sys/malloc.h> -#include <sys/module.h> - -#include <sys/dtrace.h> -#include <sys/dtrace_bsd.h> - -#include <nfs/nfsproto.h> - -/* - * dtnfsclient is a DTrace provider that tracks the intent to perform RPCs - * in the NFS client, as well as acess to and maintenance of the access and - * attribute caches. This is not quite the same as RPCs, because NFS may - * issue multiple RPC transactions in the event that authentication fails, - * there's a jukebox error, or none at all if the access or attribute cache - * hits. However, it cleanly represents the logical layer between RPC - * transmission and vnode/vfs operations, providing access to state linking - * the two. - */ - -static int dtnfsclient_unload(void); -static void dtnfsclient_getargdesc(void *, dtrace_id_t, void *, - dtrace_argdesc_t *); -static void dtnfsclient_provide(void *, dtrace_probedesc_t *); -static void dtnfsclient_destroy(void *, dtrace_id_t, void *); -static void dtnfsclient_enable(void *, dtrace_id_t, void *); -static void dtnfsclient_disable(void *, dtrace_id_t, void *); -static void dtnfsclient_load(void *); - -static dtrace_pattr_t dtnfsclient_attr = { -{ DTRACE_STABILITY_STABLE, DTRACE_STABILITY_STABLE, DTRACE_CLASS_COMMON }, -{ DTRACE_STABILITY_PRIVATE, DTRACE_STABILITY_PRIVATE, DTRACE_CLASS_UNKNOWN }, -{ DTRACE_STABILITY_PRIVATE, DTRACE_STABILITY_PRIVATE, DTRACE_CLASS_UNKNOWN }, -{ DTRACE_STABILITY_STABLE, DTRACE_STABILITY_STABLE, DTRACE_CLASS_COMMON }, -{ DTRACE_STABILITY_STABLE, DTRACE_STABILITY_STABLE, DTRACE_CLASS_COMMON }, -}; - -/* - * Description of NFSv3 and (optional) NFSv2 probes for a procedure. - */ -struct dtnfsclient_rpc { - char *nr_v3_name; - char *nr_v2_name; /* Or NULL if none. */ - - /* - * IDs for the start and done cases, for both NFSv2 and NFSv3. - */ - uint32_t nr_v2_id_start, nr_v2_id_done; - uint32_t nr_v3_id_start, nr_v3_id_done; -}; - -/* - * This table is indexed by NFSv3 procedure number, but also used for NFSv2 - * procedure names. - */ -static struct dtnfsclient_rpc dtnfsclient_rpcs[NFS_NPROCS] = { - { "null", "null" }, - { "getattr", "getattr" }, - { "setattr", "setattr" }, - { "lookup", "lookup" }, - { "access" }, - { "readlink", "readlink" }, - { "read", "read" }, - { "write", "write" }, - { "create", "create" }, - { "mkdir", "mkdir" }, - { "symlink", "symlink" }, - { "mknod" }, - { "remove", "remove" }, - { "rmdir", "rmdir" }, - { "rename", "rename" }, - { "link", "link" }, - { "readdir", "readdir" }, - { "readdirplus" }, - { "fsstat", "statfs" }, - { "fsinfo" }, - { "pathconf" }, - { "commit" }, - { "noop" }, -}; - -/* - * Module name strings. - */ -static char *dtnfsclient_accesscache_str = "accesscache"; -static char *dtnfsclient_attrcache_str = "attrcache"; -static char *dtnfsclient_nfs2_str = "nfs2"; -static char *dtnfsclient_nfs3_str = "nfs3"; - -/* - * Function name strings. - */ -static char *dtnfsclient_flush_str = "flush"; -static char *dtnfsclient_load_str = "load"; -static char *dtnfsclient_get_str = "get"; - -/* - * Name strings. - */ -static char *dtnfsclient_done_str = "done"; -static char *dtnfsclient_hit_str = "hit"; -static char *dtnfsclient_miss_str = "miss"; -static char *dtnfsclient_start_str = "start"; - -static dtrace_pops_t dtnfsclient_pops = { - dtnfsclient_provide, - NULL, - dtnfsclient_enable, - dtnfsclient_disable, - NULL, - NULL, - dtnfsclient_getargdesc, - NULL, - NULL, - dtnfsclient_destroy -}; - -static dtrace_provider_id_t dtnfsclient_id; - -/* - * Most probes are generated from the above RPC table, but for access and - * attribute caches, we have specific IDs we recognize and handle specially - * in various spots. - */ -extern uint32_t nfsclient_accesscache_flush_done_id; -extern uint32_t nfsclient_accesscache_get_hit_id; -extern uint32_t nfsclient_accesscache_get_miss_id; -extern uint32_t nfsclient_accesscache_load_done_id; - -extern uint32_t nfsclient_attrcache_flush_done_id; -extern uint32_t nfsclient_attrcache_get_hit_id; -extern uint32_t nfsclient_attrcache_get_miss_id; -extern uint32_t nfsclient_attrcache_load_done_id; - -/* - * When tracing on a procedure is enabled, the DTrace ID for an RPC event is - * stored in one of these two NFS client-allocated arrays; 0 indicates that - * the event is not being traced so probes should not be called. - * - * For simplicity, we allocate both v2 and v3 arrays as NFS_NPROCS, and the - * v2 array is simply sparse. - */ -extern uint32_t nfsclient_nfs2_start_probes[NFS_NPROCS]; -extern uint32_t nfsclient_nfs2_done_probes[NFS_NPROCS]; - -extern uint32_t nfsclient_nfs3_start_probes[NFS_NPROCS]; -extern uint32_t nfsclient_nfs3_done_probes[NFS_NPROCS]; - -/* - * Look up a DTrace probe ID to see if it's associated with a "done" event -- - * if so, we will return a fourth argument type of "int". - */ -static int -dtnfs23_isdoneprobe(dtrace_id_t id) -{ - int i; - - for (i = 0; i < NFS_NPROCS; i++) { - if (dtnfsclient_rpcs[i].nr_v3_id_done == id || - dtnfsclient_rpcs[i].nr_v2_id_done == id) - return (1); - } - return (0); -} - -static void -dtnfsclient_getargdesc(void *arg, dtrace_id_t id, void *parg, - dtrace_argdesc_t *desc) -{ - const char *p = NULL; - - if (id == nfsclient_accesscache_flush_done_id || - id == nfsclient_attrcache_flush_done_id || - id == nfsclient_attrcache_get_miss_id) { - switch (desc->dtargd_ndx) { - case 0: - p = "struct vnode *"; - break; - default: - desc->dtargd_ndx = DTRACE_ARGNONE; - break; - } - } else if (id == nfsclient_accesscache_get_hit_id || - id == nfsclient_accesscache_get_miss_id) { - switch (desc->dtargd_ndx) { - case 0: - p = "struct vnode *"; - break; - case 1: - p = "uid_t"; - break; - case 2: - p = "uint32_t"; - break; - default: - desc->dtargd_ndx = DTRACE_ARGNONE; - break; - } - } else if (id == nfsclient_accesscache_load_done_id) { - switch (desc->dtargd_ndx) { - case 0: - p = "struct vnode *"; - break; - case 1: - p = "uid_t"; - break; - case 2: - p = "uint32_t"; - break; - case 3: - p = "int"; - break; - default: - desc->dtargd_ndx = DTRACE_ARGNONE; - break; - } - } else if (id == nfsclient_attrcache_get_hit_id) { - switch (desc->dtargd_ndx) { - case 0: - p = "struct vnode *"; - break; - case 1: - p = "struct vattr *"; - break; - default: - desc->dtargd_ndx = DTRACE_ARGNONE; - break; - } - } else if (id == nfsclient_attrcache_load_done_id) { - switch (desc->dtargd_ndx) { - case 0: - p = "struct vnode *"; - break; - case 1: - p = "struct vattr *"; - break; - case 2: - p = "int"; - break; - default: - desc->dtargd_ndx = DTRACE_ARGNONE; - break; - } - } else { - switch (desc->dtargd_ndx) { - case 0: - p = "struct vnode *"; - break; - case 1: - p = "struct mbuf *"; - break; - case 2: - p = "struct ucred *"; - break; - case 3: - p = "int"; - break; - case 4: - if (dtnfs23_isdoneprobe(id)) { - p = "int"; - break; - } - /* FALLSTHROUGH */ - default: - desc->dtargd_ndx = DTRACE_ARGNONE; - break; - } - } - if (p != NULL) - strlcpy(desc->dtargd_native, p, sizeof(desc->dtargd_native)); -} - -static void -dtnfsclient_provide(void *arg, dtrace_probedesc_t *desc) -{ - int i; - - if (desc != NULL) - return; - - /* - * Register access cache probes. - */ - if (dtrace_probe_lookup(dtnfsclient_id, dtnfsclient_accesscache_str, - dtnfsclient_flush_str, dtnfsclient_done_str) == 0) { - nfsclient_accesscache_flush_done_id = dtrace_probe_create( - dtnfsclient_id, dtnfsclient_accesscache_str, - dtnfsclient_flush_str, dtnfsclient_done_str, 0, NULL); - } - if (dtrace_probe_lookup(dtnfsclient_id, dtnfsclient_accesscache_str, - dtnfsclient_get_str, dtnfsclient_hit_str) == 0) { - nfsclient_accesscache_get_hit_id = dtrace_probe_create( - dtnfsclient_id, dtnfsclient_accesscache_str, - dtnfsclient_get_str, dtnfsclient_hit_str, 0, NULL); - } - if (dtrace_probe_lookup(dtnfsclient_id, dtnfsclient_accesscache_str, - dtnfsclient_get_str, dtnfsclient_miss_str) == 0) { - nfsclient_accesscache_get_miss_id = dtrace_probe_create( - dtnfsclient_id, dtnfsclient_accesscache_str, - dtnfsclient_get_str, dtnfsclient_miss_str, 0, NULL); - } - if (dtrace_probe_lookup(dtnfsclient_id, dtnfsclient_accesscache_str, - dtnfsclient_load_str, dtnfsclient_done_str) == 0) { - nfsclient_accesscache_load_done_id = dtrace_probe_create( - dtnfsclient_id, dtnfsclient_accesscache_str, - dtnfsclient_load_str, dtnfsclient_done_str, 0, NULL); - } - - /* - * Register attribute cache probes. - */ - if (dtrace_probe_lookup(dtnfsclient_id, dtnfsclient_attrcache_str, - dtnfsclient_flush_str, dtnfsclient_done_str) == 0) { - nfsclient_attrcache_flush_done_id = dtrace_probe_create( - dtnfsclient_id, dtnfsclient_attrcache_str, - dtnfsclient_flush_str, dtnfsclient_done_str, 0, NULL); - } - if (dtrace_probe_lookup(dtnfsclient_id, dtnfsclient_attrcache_str, - dtnfsclient_get_str, dtnfsclient_hit_str) == 0) { - nfsclient_attrcache_get_hit_id = dtrace_probe_create( - dtnfsclient_id, dtnfsclient_attrcache_str, - dtnfsclient_get_str, dtnfsclient_hit_str, 0, NULL); - } - if (dtrace_probe_lookup(dtnfsclient_id, dtnfsclient_attrcache_str, - dtnfsclient_get_str, dtnfsclient_miss_str) == 0) { - nfsclient_attrcache_get_miss_id = dtrace_probe_create( - dtnfsclient_id, dtnfsclient_attrcache_str, - dtnfsclient_get_str, dtnfsclient_miss_str, 0, NULL); - } - if (dtrace_probe_lookup(dtnfsclient_id, dtnfsclient_attrcache_str, - dtnfsclient_load_str, dtnfsclient_done_str) == 0) { - nfsclient_attrcache_load_done_id = dtrace_probe_create( - dtnfsclient_id, dtnfsclient_attrcache_str, - dtnfsclient_load_str, dtnfsclient_done_str, 0, NULL); - } - - /* - * Register NFSv2 RPC procedures; note sparseness check for each slot - * in the NFSv3 procnum-indexed array. - */ - for (i = 0; i < NFS_NPROCS; i++) { - if (dtnfsclient_rpcs[i].nr_v2_name != NULL && - dtrace_probe_lookup(dtnfsclient_id, dtnfsclient_nfs2_str, - dtnfsclient_rpcs[i].nr_v2_name, dtnfsclient_start_str) == - 0) { - dtnfsclient_rpcs[i].nr_v2_id_start = - dtrace_probe_create(dtnfsclient_id, - dtnfsclient_nfs2_str, - dtnfsclient_rpcs[i].nr_v2_name, - dtnfsclient_start_str, 0, - &nfsclient_nfs2_start_probes[i]); - } - if (dtnfsclient_rpcs[i].nr_v2_name != NULL && - dtrace_probe_lookup(dtnfsclient_id, dtnfsclient_nfs2_str, - dtnfsclient_rpcs[i].nr_v2_name, dtnfsclient_done_str) == - 0) { - dtnfsclient_rpcs[i].nr_v2_id_done = - dtrace_probe_create(dtnfsclient_id, - dtnfsclient_nfs2_str, - dtnfsclient_rpcs[i].nr_v2_name, - dtnfsclient_done_str, 0, - &nfsclient_nfs2_done_probes[i]); - } - } - - /* - * Register NFSv3 RPC procedures. - */ - for (i = 0; i < NFS_NPROCS; i++) { - if (dtrace_probe_lookup(dtnfsclient_id, dtnfsclient_nfs3_str, - dtnfsclient_rpcs[i].nr_v3_name, dtnfsclient_start_str) == - 0) { - dtnfsclient_rpcs[i].nr_v3_id_start = - dtrace_probe_create(dtnfsclient_id, - dtnfsclient_nfs3_str, - dtnfsclient_rpcs[i].nr_v3_name, - dtnfsclient_start_str, 0, - &nfsclient_nfs3_start_probes[i]); - } - if (dtrace_probe_lookup(dtnfsclient_id, dtnfsclient_nfs3_str, - dtnfsclient_rpcs[i].nr_v3_name, dtnfsclient_done_str) == - 0) { - dtnfsclient_rpcs[i].nr_v3_id_done = - dtrace_probe_create(dtnfsclient_id, - dtnfsclient_nfs3_str, - dtnfsclient_rpcs[i].nr_v3_name, - dtnfsclient_done_str, 0, - &nfsclient_nfs3_done_probes[i]); - } - } -} - -static void -dtnfsclient_destroy(void *arg, dtrace_id_t id, void *parg) -{ -} - -static void -dtnfsclient_enable(void *arg, dtrace_id_t id, void *parg) -{ - uint32_t *p = parg; - void *f = dtrace_probe; - - if (id == nfsclient_accesscache_flush_done_id) - dtrace_nfsclient_accesscache_flush_done_probe = f; - else if (id == nfsclient_accesscache_get_hit_id) - dtrace_nfsclient_accesscache_get_hit_probe = f; - else if (id == nfsclient_accesscache_get_miss_id) - dtrace_nfsclient_accesscache_get_miss_probe = f; - else if (id == nfsclient_accesscache_load_done_id) - dtrace_nfsclient_accesscache_load_done_probe = f; - else if (id == nfsclient_attrcache_flush_done_id) - dtrace_nfsclient_attrcache_flush_done_probe = f; - else if (id == nfsclient_attrcache_get_hit_id) - dtrace_nfsclient_attrcache_get_hit_probe = f; - else if (id == nfsclient_attrcache_get_miss_id) - dtrace_nfsclient_attrcache_get_miss_probe = f; - else if (id == nfsclient_attrcache_load_done_id) - dtrace_nfsclient_attrcache_load_done_probe = f; - else - *p = id; -} - -static void -dtnfsclient_disable(void *arg, dtrace_id_t id, void *parg) -{ - uint32_t *p = parg; - - if (id == nfsclient_accesscache_flush_done_id) - dtrace_nfsclient_accesscache_flush_done_probe = NULL; - else if (id == nfsclient_accesscache_get_hit_id) - dtrace_nfsclient_accesscache_get_hit_probe = NULL; - else if (id == nfsclient_accesscache_get_miss_id) - dtrace_nfsclient_accesscache_get_miss_probe = NULL; - else if (id == nfsclient_accesscache_load_done_id) - dtrace_nfsclient_accesscache_load_done_probe = NULL; - else if (id == nfsclient_attrcache_flush_done_id) - dtrace_nfsclient_attrcache_flush_done_probe = NULL; - else if (id == nfsclient_attrcache_get_hit_id) - dtrace_nfsclient_attrcache_get_hit_probe = NULL; - else if (id == nfsclient_attrcache_get_miss_id) - dtrace_nfsclient_attrcache_get_miss_probe = NULL; - else if (id == nfsclient_attrcache_load_done_id) - dtrace_nfsclient_attrcache_load_done_probe = NULL; - else - *p = 0; -} - -static void -dtnfsclient_load(void *dummy) -{ - - if (dtrace_register("nfsclient", &dtnfsclient_attr, - DTRACE_PRIV_USER, NULL, &dtnfsclient_pops, NULL, - &dtnfsclient_id) != 0) - return; - - dtrace_nfsclient_nfs23_start_probe = - (dtrace_nfsclient_nfs23_start_probe_func_t)dtrace_probe; - dtrace_nfsclient_nfs23_done_probe = - (dtrace_nfsclient_nfs23_done_probe_func_t)dtrace_probe; -} - - -static int -dtnfsclient_unload() -{ - - dtrace_nfsclient_nfs23_start_probe = NULL; - dtrace_nfsclient_nfs23_done_probe = NULL; - - return (dtrace_unregister(dtnfsclient_id)); -} - -static int -dtnfsclient_modevent(module_t mod __unused, int type, void *data __unused) -{ - int error = 0; - - switch (type) { - case MOD_LOAD: - break; - - case MOD_UNLOAD: - break; - - case MOD_SHUTDOWN: - break; - - default: - error = EOPNOTSUPP; - break; - } - - return (error); -} - -SYSINIT(dtnfsclient_load, SI_SUB_DTRACE_PROVIDER, SI_ORDER_ANY, - dtnfsclient_load, NULL); -SYSUNINIT(dtnfsclient_unload, SI_SUB_DTRACE_PROVIDER, SI_ORDER_ANY, - dtnfsclient_unload, NULL); - -DEV_MODULE(dtnfsclient, dtnfsclient_modevent, NULL); -MODULE_VERSION(dtnfsclient, 1); -MODULE_DEPEND(dtnfsclient, dtrace, 1, 1, 1); -MODULE_DEPEND(dtnfsclient, opensolaris, 1, 1, 1); -MODULE_DEPEND(dtnfsclient, oldnfs, 1, 1, 1); diff --git a/sys/nfsclient/nfs_krpc.c b/sys/nfsclient/nfs_krpc.c deleted file mode 100644 index f88e47f..0000000 --- a/sys/nfsclient/nfs_krpc.c +++ /dev/null @@ -1,887 +0,0 @@ -/*- - * Copyright (c) 1989, 1991, 1993, 1995 - * The Regents of the University of California. All rights reserved. - * - * This code is derived from software contributed to Berkeley by - * Rick Macklem at The University of Guelph. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 4. Neither the name of the University nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * @(#)nfs_socket.c 8.5 (Berkeley) 3/30/95 - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -/* - * Socket operations for use by nfs - */ - -#include "opt_inet6.h" -#include "opt_kgssapi.h" - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/kernel.h> -#include <sys/limits.h> -#include <sys/lock.h> -#include <sys/malloc.h> -#include <sys/mbuf.h> -#include <sys/mount.h> -#include <sys/mutex.h> -#include <sys/proc.h> -#include <sys/signalvar.h> -#include <sys/syscallsubr.h> -#include <sys/sysctl.h> -#include <sys/syslog.h> -#include <sys/vnode.h> - -#include <rpc/rpc.h> - -#include <nfs/nfsproto.h> -#include <nfsclient/nfs.h> -#include <nfs/xdr_subs.h> -#include <nfsclient/nfsm_subs.h> -#include <nfsclient/nfsmount.h> -#include <nfsclient/nfsnode.h> - -#ifdef KDTRACE_HOOKS -#include <sys/dtrace_bsd.h> - -dtrace_nfsclient_nfs23_start_probe_func_t - dtrace_nfsclient_nfs23_start_probe; - -dtrace_nfsclient_nfs23_done_probe_func_t - dtrace_nfsclient_nfs23_done_probe; - -/* - * Registered probes by RPC type. - */ -uint32_t nfsclient_nfs2_start_probes[NFS_NPROCS]; -uint32_t nfsclient_nfs2_done_probes[NFS_NPROCS]; - -uint32_t nfsclient_nfs3_start_probes[NFS_NPROCS]; -uint32_t nfsclient_nfs3_done_probes[NFS_NPROCS]; -#endif - -static int nfs_bufpackets = 4; -static int nfs_reconnects; -static int nfs3_jukebox_delay = 10; -static int nfs_skip_wcc_data_onerr = 1; -static int fake_wchan; - -SYSCTL_DECL(_vfs_oldnfs); - -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, bufpackets, CTLFLAG_RW, &nfs_bufpackets, 0, - "Buffer reservation size 2 < x < 64"); -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, reconnects, CTLFLAG_RD, &nfs_reconnects, 0, - "Number of times the nfs client has had to reconnect"); -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, nfs3_jukebox_delay, CTLFLAG_RW, - &nfs3_jukebox_delay, 0, - "Number of seconds to delay a retry after receiving EJUKEBOX"); -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, skip_wcc_data_onerr, CTLFLAG_RW, - &nfs_skip_wcc_data_onerr, 0, - "Disable weak cache consistency checking when server returns an error"); - -static void nfs_down(struct nfsmount *, struct thread *, const char *, - int, int); -static void nfs_up(struct nfsmount *, struct thread *, const char *, - int, int); -static int nfs_msg(struct thread *, const char *, const char *, int); - -extern int nfsv2_procid[]; - -struct nfs_cached_auth { - int ca_refs; /* refcount, including 1 from the cache */ - uid_t ca_uid; /* uid that corresponds to this auth */ - AUTH *ca_auth; /* RPC auth handle */ -}; - -/* - * RTT estimator - */ - -static enum nfs_rto_timer_t nfs_proct[NFS_NPROCS] = { - NFS_DEFAULT_TIMER, /* NULL */ - NFS_GETATTR_TIMER, /* GETATTR */ - NFS_DEFAULT_TIMER, /* SETATTR */ - NFS_LOOKUP_TIMER, /* LOOKUP */ - NFS_GETATTR_TIMER, /* ACCESS */ - NFS_READ_TIMER, /* READLINK */ - NFS_READ_TIMER, /* READ */ - NFS_WRITE_TIMER, /* WRITE */ - NFS_DEFAULT_TIMER, /* CREATE */ - NFS_DEFAULT_TIMER, /* MKDIR */ - NFS_DEFAULT_TIMER, /* SYMLINK */ - NFS_DEFAULT_TIMER, /* MKNOD */ - NFS_DEFAULT_TIMER, /* REMOVE */ - NFS_DEFAULT_TIMER, /* RMDIR */ - NFS_DEFAULT_TIMER, /* RENAME */ - NFS_DEFAULT_TIMER, /* LINK */ - NFS_READ_TIMER, /* READDIR */ - NFS_READ_TIMER, /* READDIRPLUS */ - NFS_DEFAULT_TIMER, /* FSSTAT */ - NFS_DEFAULT_TIMER, /* FSINFO */ - NFS_DEFAULT_TIMER, /* PATHCONF */ - NFS_DEFAULT_TIMER, /* COMMIT */ - NFS_DEFAULT_TIMER, /* NOOP */ -}; - -/* - * Choose the correct RTT timer for this NFS procedure. - */ -static inline enum nfs_rto_timer_t -nfs_rto_timer(u_int32_t procnum) -{ - - return (nfs_proct[procnum]); -} - -/* - * Initialize the RTT estimator state for a new mount point. - */ -static void -nfs_init_rtt(struct nfsmount *nmp) -{ - int i; - - for (i = 0; i < NFS_MAX_TIMER; i++) { - nmp->nm_timers[i].rt_srtt = hz; - nmp->nm_timers[i].rt_deviate = 0; - nmp->nm_timers[i].rt_rtxcur = hz; - } -} - -/* - * Initialize sockets and congestion for a new NFS connection. - * We do not free the sockaddr if error. - */ -int -nfs_connect(struct nfsmount *nmp) -{ - int rcvreserve, sndreserve; - int pktscale; - struct sockaddr *saddr; - struct ucred *origcred; - struct thread *td = curthread; - CLIENT *client; - struct netconfig *nconf; - rpcvers_t vers; - int one = 1, retries; - struct timeval timo; - - /* - * We need to establish the socket using the credentials of - * the mountpoint. Some parts of this process (such as - * sobind() and soconnect()) will use the curent thread's - * credential instead of the socket credential. To work - * around this, temporarily change the current thread's - * credential to that of the mountpoint. - * - * XXX: It would be better to explicitly pass the correct - * credential to sobind() and soconnect(). - */ - origcred = td->td_ucred; - td->td_ucred = nmp->nm_mountp->mnt_cred; - saddr = nmp->nm_nam; - - vers = NFS_VER2; - if (nmp->nm_flag & NFSMNT_NFSV3) - vers = NFS_VER3; - else if (nmp->nm_flag & NFSMNT_NFSV4) - vers = NFS_VER4; - if (saddr->sa_family == AF_INET) - if (nmp->nm_sotype == SOCK_DGRAM) - nconf = getnetconfigent("udp"); - else - nconf = getnetconfigent("tcp"); - else - if (nmp->nm_sotype == SOCK_DGRAM) - nconf = getnetconfigent("udp6"); - else - nconf = getnetconfigent("tcp6"); - - /* - * Get buffer reservation size from sysctl, but impose reasonable - * limits. - */ - pktscale = nfs_bufpackets; - if (pktscale < 2) - pktscale = 2; - if (pktscale > 64) - pktscale = 64; - mtx_lock(&nmp->nm_mtx); - if (nmp->nm_sotype == SOCK_DGRAM) { - sndreserve = (nmp->nm_wsize + NFS_MAXPKTHDR) * pktscale; - rcvreserve = (max(nmp->nm_rsize, nmp->nm_readdirsize) + - NFS_MAXPKTHDR) * pktscale; - } else if (nmp->nm_sotype == SOCK_SEQPACKET) { - sndreserve = (nmp->nm_wsize + NFS_MAXPKTHDR) * pktscale; - rcvreserve = (max(nmp->nm_rsize, nmp->nm_readdirsize) + - NFS_MAXPKTHDR) * pktscale; - } else { - if (nmp->nm_sotype != SOCK_STREAM) - panic("nfscon sotype"); - sndreserve = (nmp->nm_wsize + NFS_MAXPKTHDR + - sizeof (u_int32_t)) * pktscale; - rcvreserve = (nmp->nm_rsize + NFS_MAXPKTHDR + - sizeof (u_int32_t)) * pktscale; - } - mtx_unlock(&nmp->nm_mtx); - - client = clnt_reconnect_create(nconf, saddr, NFS_PROG, vers, - sndreserve, rcvreserve); - CLNT_CONTROL(client, CLSET_WAITCHAN, "nfsreq"); - if (nmp->nm_flag & NFSMNT_INT) - CLNT_CONTROL(client, CLSET_INTERRUPTIBLE, &one); - if (nmp->nm_flag & NFSMNT_RESVPORT) - CLNT_CONTROL(client, CLSET_PRIVPORT, &one); - if ((nmp->nm_flag & NFSMNT_SOFT) != 0) { - if (nmp->nm_sotype == SOCK_DGRAM) - /* - * For UDP, the large timeout for a reconnect will - * be set to "nm_retry * nm_timeo / 2", so we only - * want to do 2 reconnect timeout retries. - */ - retries = 2; - else - retries = nmp->nm_retry; - } else - retries = INT_MAX; - CLNT_CONTROL(client, CLSET_RETRIES, &retries); - - /* - * For UDP, there are 2 timeouts: - * - CLSET_RETRY_TIMEOUT sets the initial timeout for the timer - * that does a retransmit of an RPC request using the same socket - * and xid. This is what you normally want to do, since NFS - * servers depend on "same xid" for their Duplicate Request Cache. - * - timeout specified in CLNT_CALL_MBUF(), which specifies when - * retransmits on the same socket should fail and a fresh socket - * created. Each of these timeouts counts as one CLSET_RETRIES, - * as set above. - * Set the initial retransmit timeout for UDP. This timeout doesn't - * exist for TCP and the following call just fails, which is ok. - */ - timo.tv_sec = nmp->nm_timeo / NFS_HZ; - timo.tv_usec = (nmp->nm_timeo % NFS_HZ) * 1000000 / NFS_HZ; - CLNT_CONTROL(client, CLSET_RETRY_TIMEOUT, &timo); - - mtx_lock(&nmp->nm_mtx); - if (nmp->nm_client) { - /* - * Someone else already connected. - */ - CLNT_RELEASE(client); - } else - nmp->nm_client = client; - - /* - * Protocols that do not require connections may be optionally left - * unconnected for servers that reply from a port other than NFS_PORT. - */ - if (!(nmp->nm_flag & NFSMNT_NOCONN)) { - mtx_unlock(&nmp->nm_mtx); - CLNT_CONTROL(client, CLSET_CONNECT, &one); - } else - mtx_unlock(&nmp->nm_mtx); - - /* Restore current thread's credentials. */ - td->td_ucred = origcred; - - mtx_lock(&nmp->nm_mtx); - /* Initialize other non-zero congestion variables. */ - nfs_init_rtt(nmp); - mtx_unlock(&nmp->nm_mtx); - return (0); -} - -/* - * NFS disconnect. Clean up and unlink. - */ -void -nfs_disconnect(struct nfsmount *nmp) -{ - CLIENT *client; - - mtx_lock(&nmp->nm_mtx); - if (nmp->nm_client) { - client = nmp->nm_client; - nmp->nm_client = NULL; - mtx_unlock(&nmp->nm_mtx); - rpc_gss_secpurge_call(client); - CLNT_CLOSE(client); - CLNT_RELEASE(client); - } else - mtx_unlock(&nmp->nm_mtx); -} - -void -nfs_safedisconnect(struct nfsmount *nmp) -{ - - nfs_disconnect(nmp); -} - -static AUTH * -nfs_getauth(struct nfsmount *nmp, struct ucred *cred) -{ - rpc_gss_service_t svc; - AUTH *auth; - - switch (nmp->nm_secflavor) { - case RPCSEC_GSS_KRB5: - case RPCSEC_GSS_KRB5I: - case RPCSEC_GSS_KRB5P: - if (!nmp->nm_mech_oid) - if (!rpc_gss_mech_to_oid_call("kerberosv5", - &nmp->nm_mech_oid)) - return (NULL); - if (nmp->nm_secflavor == RPCSEC_GSS_KRB5) - svc = rpc_gss_svc_none; - else if (nmp->nm_secflavor == RPCSEC_GSS_KRB5I) - svc = rpc_gss_svc_integrity; - else - svc = rpc_gss_svc_privacy; - auth = rpc_gss_secfind_call(nmp->nm_client, cred, - nmp->nm_principal, nmp->nm_mech_oid, svc); - if (auth) - return (auth); - /* fallthrough */ - case AUTH_SYS: - default: - return (authunix_create(cred)); - - } -} - -/* - * Callback from the RPC code to generate up/down notifications. - */ - -struct nfs_feedback_arg { - struct nfsmount *nf_mount; - int nf_lastmsg; /* last tprintf */ - int nf_tprintfmsg; - struct thread *nf_td; -}; - -static void -nfs_feedback(int type, int proc, void *arg) -{ - struct nfs_feedback_arg *nf = (struct nfs_feedback_arg *) arg; - struct nfsmount *nmp = nf->nf_mount; - time_t now; - - switch (type) { - case FEEDBACK_REXMIT2: - case FEEDBACK_RECONNECT: - now = time_uptime; - if (nf->nf_lastmsg + nmp->nm_tprintf_delay < now) { - nfs_down(nmp, nf->nf_td, - "not responding", 0, NFSSTA_TIMEO); - nf->nf_tprintfmsg = TRUE; - nf->nf_lastmsg = now; - } - break; - - case FEEDBACK_OK: - nfs_up(nf->nf_mount, nf->nf_td, - "is alive again", NFSSTA_TIMEO, nf->nf_tprintfmsg); - break; - } -} - -/* - * nfs_request - goes something like this - * - fill in request struct - * - links it into list - * - calls nfs_send() for first transmit - * - calls nfs_receive() to get reply - * - break down rpc header and return with nfs reply pointed to - * by mrep or error - * nb: always frees up mreq mbuf list - */ -int -nfs_request(struct vnode *vp, struct mbuf *mreq, int procnum, - struct thread *td, struct ucred *cred, struct mbuf **mrp, - struct mbuf **mdp, caddr_t *dposp) -{ - struct mbuf *mrep; - u_int32_t *tl; - struct nfsmount *nmp; - struct mbuf *md; - time_t waituntil; - caddr_t dpos; - int error = 0, timeo; - AUTH *auth = NULL; - enum nfs_rto_timer_t timer; - struct nfs_feedback_arg nf; - struct rpc_callextra ext; - enum clnt_stat stat; - struct timeval timo; - - /* Reject requests while attempting a forced unmount. */ - if (vp->v_mount->mnt_kern_flag & MNTK_UNMOUNTF) { - m_freem(mreq); - return (ESTALE); - } - nmp = VFSTONFS(vp->v_mount); - bzero(&nf, sizeof(struct nfs_feedback_arg)); - nf.nf_mount = nmp; - nf.nf_td = td; - nf.nf_lastmsg = time_uptime - - ((nmp->nm_tprintf_delay) - (nmp->nm_tprintf_initial_delay)); - - /* - * XXX if not already connected call nfs_connect now. Longer - * term, change nfs_mount to call nfs_connect unconditionally - * and let clnt_reconnect_create handle reconnects. - */ - if (!nmp->nm_client) - nfs_connect(nmp); - - auth = nfs_getauth(nmp, cred); - if (!auth) { - m_freem(mreq); - return (EACCES); - } - bzero(&ext, sizeof(ext)); - ext.rc_auth = auth; - - ext.rc_feedback = nfs_feedback; - ext.rc_feedback_arg = &nf; - - /* - * Use a conservative timeout for RPCs other than getattr, - * lookup, read or write. The justification for doing "other" - * this way is that these RPCs happen so infrequently that - * timer est. would probably be stale. Also, since many of - * these RPCs are non-idempotent, a conservative timeout is - * desired. - */ - timer = nfs_rto_timer(procnum); - if (timer != NFS_DEFAULT_TIMER) - ext.rc_timers = &nmp->nm_timers[timer - 1]; - else - ext.rc_timers = NULL; - -#ifdef KDTRACE_HOOKS - if (dtrace_nfsclient_nfs23_start_probe != NULL) { - uint32_t probe_id; - int probe_procnum; - - if (nmp->nm_flag & NFSMNT_NFSV3) { - probe_id = nfsclient_nfs3_start_probes[procnum]; - probe_procnum = procnum; - } else { - probe_id = nfsclient_nfs2_start_probes[procnum]; - probe_procnum = nfsv2_procid[procnum]; - } - if (probe_id != 0) - (dtrace_nfsclient_nfs23_start_probe)(probe_id, vp, - mreq, cred, probe_procnum); - } -#endif - - nfsstats.rpcrequests++; -tryagain: - /* - * This timeout specifies when a new socket should be created, - * along with new xid values. For UDP, this should be done - * infrequently, since retransmits of RPC requests should normally - * use the same xid. - */ - if (nmp->nm_sotype == SOCK_DGRAM) { - if ((nmp->nm_flag & NFSMNT_SOFT) != 0) { - /* - * CLSET_RETRIES is set to 2, so this should be half - * of the total timeout required. - */ - timeo = nmp->nm_retry * nmp->nm_timeo / 2; - if (timeo < 1) - timeo = 1; - timo.tv_sec = timeo / NFS_HZ; - timo.tv_usec = (timeo % NFS_HZ) * 1000000 / NFS_HZ; - } else { - /* For UDP hard mounts, use a large value. */ - timo.tv_sec = NFS_MAXTIMEO / NFS_HZ; - timo.tv_usec = 0; - } - } else { - timo.tv_sec = nmp->nm_timeo / NFS_HZ; - timo.tv_usec = (nmp->nm_timeo % NFS_HZ) * 1000000 / NFS_HZ; - } - mrep = NULL; - stat = CLNT_CALL_MBUF(nmp->nm_client, &ext, - (nmp->nm_flag & NFSMNT_NFSV3) ? procnum : nfsv2_procid[procnum], - mreq, &mrep, timo); - - /* - * If there was a successful reply and a tprintf msg. - * tprintf a response. - */ - if (stat == RPC_SUCCESS) - error = 0; - else if (stat == RPC_TIMEDOUT) { - nfsstats.rpctimeouts++; - error = ETIMEDOUT; - } else if (stat == RPC_VERSMISMATCH) { - nfsstats.rpcinvalid++; - error = EOPNOTSUPP; - } else if (stat == RPC_PROGVERSMISMATCH) { - nfsstats.rpcinvalid++; - error = EPROTONOSUPPORT; - } else if (stat == RPC_INTR) { - error = EINTR; - } else { - nfsstats.rpcinvalid++; - error = EACCES; - } - if (error) - goto nfsmout; - - KASSERT(mrep != NULL, ("mrep shouldn't be NULL if no error\n")); - - /* - * Search for any mbufs that are not a multiple of 4 bytes long - * or with m_data not longword aligned. - * These could cause pointer alignment problems, so copy them to - * well aligned mbufs. - */ - error = nfs_realign(&mrep, M_NOWAIT); - if (error == ENOMEM) { - m_freem(mrep); - AUTH_DESTROY(auth); - nfsstats.rpcinvalid++; - return (error); - } - - md = mrep; - dpos = mtod(mrep, caddr_t); - tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED); - if (*tl != 0) { - error = fxdr_unsigned(int, *tl); - if ((nmp->nm_flag & NFSMNT_NFSV3) && - error == NFSERR_TRYLATER) { - m_freem(mrep); - error = 0; - waituntil = time_second + nfs3_jukebox_delay; - while (time_second < waituntil) - (void)tsleep(&fake_wchan, PSOCK, "nqnfstry", - hz); - goto tryagain; - } - /* - * Make sure NFSERR_RETERR isn't bogusly set by a server - * such as amd. (No actual NFS error has bit 31 set.) - */ - error &= ~NFSERR_RETERR; - - /* - * If the File Handle was stale, invalidate the lookup - * cache, just in case. - */ - if (error == ESTALE) - nfs_purgecache(vp); - /* - * Skip wcc data on non-ENOENT NFS errors for now. - * NetApp filers return corrupt postop attrs in the - * wcc data for NFS err EROFS. Not sure if they could - * return corrupt postop attrs for others errors. - * Blocking ENOENT post-op attributes breaks negative - * name caching, so always allow it through. - */ - if ((nmp->nm_flag & NFSMNT_NFSV3) && - (!nfs_skip_wcc_data_onerr || error == ENOENT)) { - *mrp = mrep; - *mdp = md; - *dposp = dpos; - error |= NFSERR_RETERR; - } else - m_freem(mrep); - goto nfsmout; - } - -#ifdef KDTRACE_HOOKS - if (dtrace_nfsclient_nfs23_done_probe != NULL) { - uint32_t probe_id; - int probe_procnum; - - if (nmp->nm_flag & NFSMNT_NFSV3) { - probe_id = nfsclient_nfs3_done_probes[procnum]; - probe_procnum = procnum; - } else { - probe_id = nfsclient_nfs2_done_probes[procnum]; - probe_procnum = (nmp->nm_flag & NFSMNT_NFSV3) ? - procnum : nfsv2_procid[procnum]; - } - if (probe_id != 0) - (dtrace_nfsclient_nfs23_done_probe)(probe_id, vp, - mreq, cred, probe_procnum, 0); - } -#endif - m_freem(mreq); - *mrp = mrep; - *mdp = md; - *dposp = dpos; - AUTH_DESTROY(auth); - return (0); - -nfsmout: -#ifdef KDTRACE_HOOKS - if (dtrace_nfsclient_nfs23_done_probe != NULL) { - uint32_t probe_id; - int probe_procnum; - - if (nmp->nm_flag & NFSMNT_NFSV3) { - probe_id = nfsclient_nfs3_done_probes[procnum]; - probe_procnum = procnum; - } else { - probe_id = nfsclient_nfs2_done_probes[procnum]; - probe_procnum = (nmp->nm_flag & NFSMNT_NFSV3) ? - procnum : nfsv2_procid[procnum]; - } - if (probe_id != 0) - (dtrace_nfsclient_nfs23_done_probe)(probe_id, vp, - mreq, cred, probe_procnum, error); - } -#endif - m_freem(mreq); - if (auth) - AUTH_DESTROY(auth); - return (error); -} - -/* - * Mark all of an nfs mount's outstanding requests with R_SOFTTERM and - * wait for all requests to complete. This is used by forced unmounts - * to terminate any outstanding RPCs. - */ -int -nfs_nmcancelreqs(struct nfsmount *nmp) -{ - - if (nmp->nm_client) - CLNT_CLOSE(nmp->nm_client); - return (0); -} - -/* - * Any signal that can interrupt an NFS operation in an intr mount - * should be added to this set. SIGSTOP and SIGKILL cannot be masked. - */ -int nfs_sig_set[] = { - SIGINT, - SIGTERM, - SIGHUP, - SIGKILL, - SIGQUIT -}; - -/* - * Check to see if one of the signals in our subset is pending on - * the process (in an intr mount). - */ -static int -nfs_sig_pending(sigset_t set) -{ - int i; - - for (i = 0 ; i < sizeof(nfs_sig_set)/sizeof(int) ; i++) - if (SIGISMEMBER(set, nfs_sig_set[i])) - return (1); - return (0); -} - -/* - * The set/restore sigmask functions are used to (temporarily) overwrite - * the thread td_sigmask during an RPC call (for example). These are also - * used in other places in the NFS client that might tsleep(). - */ -void -nfs_set_sigmask(struct thread *td, sigset_t *oldset) -{ - sigset_t newset; - int i; - struct proc *p; - - SIGFILLSET(newset); - if (td == NULL) - td = curthread; /* XXX */ - p = td->td_proc; - /* Remove the NFS set of signals from newset. */ - PROC_LOCK(p); - mtx_lock(&p->p_sigacts->ps_mtx); - for (i = 0 ; i < sizeof(nfs_sig_set)/sizeof(int) ; i++) { - /* - * But make sure we leave the ones already masked - * by the process, i.e. remove the signal from the - * temporary signalmask only if it wasn't already - * in p_sigmask. - */ - if (!SIGISMEMBER(td->td_sigmask, nfs_sig_set[i]) && - !SIGISMEMBER(p->p_sigacts->ps_sigignore, nfs_sig_set[i])) - SIGDELSET(newset, nfs_sig_set[i]); - } - mtx_unlock(&p->p_sigacts->ps_mtx); - kern_sigprocmask(td, SIG_SETMASK, &newset, oldset, - SIGPROCMASK_PROC_LOCKED); - PROC_UNLOCK(p); -} - -void -nfs_restore_sigmask(struct thread *td, sigset_t *set) -{ - if (td == NULL) - td = curthread; /* XXX */ - kern_sigprocmask(td, SIG_SETMASK, set, NULL, 0); -} - -/* - * NFS wrapper to msleep(), that shoves a new p_sigmask and restores the - * old one after msleep() returns. - */ -int -nfs_msleep(struct thread *td, void *ident, struct mtx *mtx, int priority, - char *wmesg, int timo) -{ - sigset_t oldset; - int error; - struct proc *p; - - if ((priority & PCATCH) == 0) - return msleep(ident, mtx, priority, wmesg, timo); - if (td == NULL) - td = curthread; /* XXX */ - nfs_set_sigmask(td, &oldset); - error = msleep(ident, mtx, priority, wmesg, timo); - nfs_restore_sigmask(td, &oldset); - p = td->td_proc; - return (error); -} - -/* - * Test for a termination condition pending on the process. - * This is used for NFSMNT_INT mounts. - */ -int -nfs_sigintr(struct nfsmount *nmp, struct thread *td) -{ - struct proc *p; - sigset_t tmpset; - - /* Terminate all requests while attempting a forced unmount. */ - if (nmp->nm_mountp->mnt_kern_flag & MNTK_UNMOUNTF) - return (EIO); - if (!(nmp->nm_flag & NFSMNT_INT)) - return (0); - if (td == NULL) - return (0); - p = td->td_proc; - PROC_LOCK(p); - tmpset = p->p_siglist; - SIGSETOR(tmpset, td->td_siglist); - SIGSETNAND(tmpset, td->td_sigmask); - mtx_lock(&p->p_sigacts->ps_mtx); - SIGSETNAND(tmpset, p->p_sigacts->ps_sigignore); - mtx_unlock(&p->p_sigacts->ps_mtx); - if ((SIGNOTEMPTY(p->p_siglist) || SIGNOTEMPTY(td->td_siglist)) - && nfs_sig_pending(tmpset)) { - PROC_UNLOCK(p); - return (EINTR); - } - PROC_UNLOCK(p); - return (0); -} - -static int -nfs_msg(struct thread *td, const char *server, const char *msg, int error) -{ - struct proc *p; - - p = td ? td->td_proc : NULL; - if (error) - tprintf(p, LOG_INFO, "nfs server %s: %s, error %d\n", server, - msg, error); - else - tprintf(p, LOG_INFO, "nfs server %s: %s\n", server, msg); - return (0); -} - -static void -nfs_down(struct nfsmount *nmp, struct thread *td, const char *msg, - int error, int flags) -{ - if (nmp == NULL) - return; - mtx_lock(&nmp->nm_mtx); - if ((flags & NFSSTA_TIMEO) && !(nmp->nm_state & NFSSTA_TIMEO)) { - nmp->nm_state |= NFSSTA_TIMEO; - mtx_unlock(&nmp->nm_mtx); - vfs_event_signal(&nmp->nm_mountp->mnt_stat.f_fsid, - VQ_NOTRESP, 0); - } else - mtx_unlock(&nmp->nm_mtx); - mtx_lock(&nmp->nm_mtx); - if ((flags & NFSSTA_LOCKTIMEO) && - !(nmp->nm_state & NFSSTA_LOCKTIMEO)) { - nmp->nm_state |= NFSSTA_LOCKTIMEO; - mtx_unlock(&nmp->nm_mtx); - vfs_event_signal(&nmp->nm_mountp->mnt_stat.f_fsid, - VQ_NOTRESPLOCK, 0); - } else - mtx_unlock(&nmp->nm_mtx); - nfs_msg(td, nmp->nm_mountp->mnt_stat.f_mntfromname, msg, error); -} - -static void -nfs_up(struct nfsmount *nmp, struct thread *td, const char *msg, - int flags, int tprintfmsg) -{ - if (nmp == NULL) - return; - if (tprintfmsg) - nfs_msg(td, nmp->nm_mountp->mnt_stat.f_mntfromname, msg, 0); - - mtx_lock(&nmp->nm_mtx); - if ((flags & NFSSTA_TIMEO) && (nmp->nm_state & NFSSTA_TIMEO)) { - nmp->nm_state &= ~NFSSTA_TIMEO; - mtx_unlock(&nmp->nm_mtx); - vfs_event_signal(&nmp->nm_mountp->mnt_stat.f_fsid, - VQ_NOTRESP, 1); - } else - mtx_unlock(&nmp->nm_mtx); - - mtx_lock(&nmp->nm_mtx); - if ((flags & NFSSTA_LOCKTIMEO) && - (nmp->nm_state & NFSSTA_LOCKTIMEO)) { - nmp->nm_state &= ~NFSSTA_LOCKTIMEO; - mtx_unlock(&nmp->nm_mtx); - vfs_event_signal(&nmp->nm_mountp->mnt_stat.f_fsid, - VQ_NOTRESPLOCK, 1); - } else - mtx_unlock(&nmp->nm_mtx); -} diff --git a/sys/nfsclient/nfs_nfsiod.c b/sys/nfsclient/nfs_nfsiod.c deleted file mode 100644 index 2fb2c88..0000000 --- a/sys/nfsclient/nfs_nfsiod.c +++ /dev/null @@ -1,346 +0,0 @@ -/*- - * Copyright (c) 1989, 1993 - * The Regents of the University of California. All rights reserved. - * - * This code is derived from software contributed to Berkeley by - * Rick Macklem at The University of Guelph. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 4. Neither the name of the University nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * @(#)nfs_syscalls.c 8.5 (Berkeley) 3/30/95 - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/sysproto.h> -#include <sys/kernel.h> -#include <sys/sysctl.h> -#include <sys/file.h> -#include <sys/filedesc.h> -#include <sys/vnode.h> -#include <sys/malloc.h> -#include <sys/mount.h> -#include <sys/proc.h> -#include <sys/bio.h> -#include <sys/buf.h> -#include <sys/mbuf.h> -#include <sys/socket.h> -#include <sys/socketvar.h> -#include <sys/domain.h> -#include <sys/protosw.h> -#include <sys/namei.h> -#include <sys/unistd.h> -#include <sys/kthread.h> -#include <sys/fcntl.h> -#include <sys/lockf.h> -#include <sys/mutex.h> -#include <sys/taskqueue.h> - -#include <netinet/in.h> -#include <netinet/tcp.h> - -#include <nfs/xdr_subs.h> -#include <nfs/nfsproto.h> -#include <nfsclient/nfs.h> -#include <nfsclient/nfsm_subs.h> -#include <nfsclient/nfsmount.h> -#include <nfsclient/nfsnode.h> -#include <nfs/nfs_lock.h> - -static MALLOC_DEFINE(M_NFSSVC, "nfsclient_srvsock", "Nfs server structure"); - -static void nfssvc_iod(void *); - -static int nfs_asyncdaemon[NFS_MAXASYNCDAEMON]; - -SYSCTL_DECL(_vfs_oldnfs); - -/* Maximum number of seconds a nfsiod kthread will sleep before exiting */ -static unsigned int nfs_iodmaxidle = 120; -SYSCTL_UINT(_vfs_oldnfs, OID_AUTO, iodmaxidle, CTLFLAG_RW, &nfs_iodmaxidle, 0, - "Max number of seconds an nfsiod kthread will sleep before exiting"); - -/* Maximum number of nfsiod kthreads */ -unsigned int nfs_iodmax = 20; - -/* Minimum number of nfsiod kthreads to keep as spares */ -static unsigned int nfs_iodmin = 0; - -static int nfs_nfsiodnew_sync(void); - -static int -sysctl_iodmin(SYSCTL_HANDLER_ARGS) -{ - int error, i; - int newmin; - - newmin = nfs_iodmin; - error = sysctl_handle_int(oidp, &newmin, 0, req); - if (error || (req->newptr == NULL)) - return (error); - mtx_lock(&nfs_iod_mtx); - if (newmin > nfs_iodmax) { - error = EINVAL; - goto out; - } - nfs_iodmin = newmin; - if (nfs_numasync >= nfs_iodmin) - goto out; - /* - * If the current number of nfsiod is lower - * than the new minimum, create some more. - */ - for (i = nfs_iodmin - nfs_numasync; i > 0; i--) - nfs_nfsiodnew_sync(); -out: - mtx_unlock(&nfs_iod_mtx); - return (0); -} -SYSCTL_PROC(_vfs_oldnfs, OID_AUTO, iodmin, CTLTYPE_UINT | CTLFLAG_RW, 0, - sizeof (nfs_iodmin), sysctl_iodmin, "IU", - "Min number of nfsiod kthreads to keep as spares"); - - -static int -sysctl_iodmax(SYSCTL_HANDLER_ARGS) -{ - int error, i; - int iod, newmax; - - newmax = nfs_iodmax; - error = sysctl_handle_int(oidp, &newmax, 0, req); - if (error || (req->newptr == NULL)) - return (error); - if (newmax > NFS_MAXASYNCDAEMON) - return (EINVAL); - mtx_lock(&nfs_iod_mtx); - nfs_iodmax = newmax; - if (nfs_numasync <= nfs_iodmax) - goto out; - /* - * If there are some asleep nfsiods that should - * exit, wakeup() them so that they check nfs_iodmax - * and exit. Those who are active will exit as - * soon as they finish I/O. - */ - iod = nfs_numasync - 1; - for (i = 0; i < nfs_numasync - nfs_iodmax; i++) { - if (nfs_iodwant[iod] == NFSIOD_AVAILABLE) - wakeup(&nfs_iodwant[iod]); - iod--; - } -out: - mtx_unlock(&nfs_iod_mtx); - return (0); -} -SYSCTL_PROC(_vfs_oldnfs, OID_AUTO, iodmax, CTLTYPE_UINT | CTLFLAG_RW, 0, - sizeof (nfs_iodmax), sysctl_iodmax, "IU", - "Max number of nfsiod kthreads"); - -static int -nfs_nfsiodnew_sync(void) -{ - int error, i; - - mtx_assert(&nfs_iod_mtx, MA_OWNED); - for (i = 0; i < nfs_iodmax; i++) { - if (nfs_asyncdaemon[i] == 0) { - nfs_asyncdaemon[i] = 1; - break; - } - } - if (i == nfs_iodmax) - return (0); - mtx_unlock(&nfs_iod_mtx); - error = kproc_create(nfssvc_iod, nfs_asyncdaemon + i, NULL, - RFHIGHPID, 0, "nfsiod %d", i); - mtx_lock(&nfs_iod_mtx); - if (error == 0) { - nfs_numasync++; - nfs_iodwant[i] = NFSIOD_AVAILABLE; - } else - nfs_asyncdaemon[i] = 0; - return (error); -} - -void -nfs_nfsiodnew_tq(__unused void *arg, int pending) -{ - - mtx_lock(&nfs_iod_mtx); - while (pending > 0) { - pending--; - nfs_nfsiodnew_sync(); - } - mtx_unlock(&nfs_iod_mtx); -} - -void -nfs_nfsiodnew(void) -{ - - mtx_assert(&nfs_iod_mtx, MA_OWNED); - taskqueue_enqueue(taskqueue_thread, &nfs_nfsiodnew_task); -} - -static void -nfsiod_setup(void *dummy) -{ - int error; - - TUNABLE_INT_FETCH("vfs.oldnfs.iodmin", &nfs_iodmin); - mtx_lock(&nfs_iod_mtx); - /* Silently limit the start number of nfsiod's */ - if (nfs_iodmin > NFS_MAXASYNCDAEMON) - nfs_iodmin = NFS_MAXASYNCDAEMON; - - while (nfs_numasync < nfs_iodmin) { - error = nfs_nfsiodnew_sync(); - if (error == -1) - panic("nfsiod_setup: nfs_nfsiodnew failed"); - } - mtx_unlock(&nfs_iod_mtx); -} -SYSINIT(nfsiod, SI_SUB_KTHREAD_IDLE, SI_ORDER_ANY, nfsiod_setup, NULL); - -static int nfs_defect = 0; -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, defect, CTLFLAG_RW, &nfs_defect, 0, - "Allow nfsiods to migrate serving different mounts"); - -/* - * Asynchronous I/O daemons for client nfs. - * They do read-ahead and write-behind operations on the block I/O cache. - * Returns if we hit the timeout defined by the iodmaxidle sysctl. - */ -static void -nfssvc_iod(void *instance) -{ - struct buf *bp; - struct nfsmount *nmp; - int myiod, timo; - int error = 0; - - mtx_lock(&nfs_iod_mtx); - myiod = (int *)instance - nfs_asyncdaemon; - /* - * Main loop - */ - for (;;) { - while (((nmp = nfs_iodmount[myiod]) == NULL) - || !TAILQ_FIRST(&nmp->nm_bufq)) { - if (myiod >= nfs_iodmax) - goto finish; - if (nmp) - nmp->nm_bufqiods--; - if (nfs_iodwant[myiod] == NFSIOD_NOT_AVAILABLE) - nfs_iodwant[myiod] = NFSIOD_AVAILABLE; - nfs_iodmount[myiod] = NULL; - /* - * Always keep at least nfs_iodmin kthreads. - */ - timo = (myiod < nfs_iodmin) ? 0 : nfs_iodmaxidle * hz; - error = msleep(&nfs_iodwant[myiod], &nfs_iod_mtx, PWAIT | PCATCH, - "-", timo); - if (error) { - nmp = nfs_iodmount[myiod]; - /* - * Rechecking the nm_bufq closes a rare race where the - * nfsiod is woken up at the exact time the idle timeout - * fires - */ - if (nmp && TAILQ_FIRST(&nmp->nm_bufq)) - error = 0; - break; - } - } - if (error) - break; - while ((bp = TAILQ_FIRST(&nmp->nm_bufq)) != NULL) { - int giant_locked = 0; - - /* Take one off the front of the list */ - TAILQ_REMOVE(&nmp->nm_bufq, bp, b_freelist); - nmp->nm_bufqlen--; - if (nmp->nm_bufqwant && nmp->nm_bufqlen <= nfs_numasync) { - nmp->nm_bufqwant = 0; - wakeup(&nmp->nm_bufq); - } - mtx_unlock(&nfs_iod_mtx); - if (NFS_ISV4(bp->b_vp)) { - giant_locked = 1; - mtx_lock(&Giant); - } - if (bp->b_flags & B_DIRECT) { - KASSERT((bp->b_iocmd == BIO_WRITE), ("nfscvs_iod: BIO_WRITE not set")); - (void)nfs_doio_directwrite(bp); - } else { - if (bp->b_iocmd == BIO_READ) - (void) nfs_doio(bp->b_vp, bp, bp->b_rcred, NULL); - else - (void) nfs_doio(bp->b_vp, bp, bp->b_wcred, NULL); - } - if (giant_locked) - mtx_unlock(&Giant); - mtx_lock(&nfs_iod_mtx); - /* - * Make sure the nmp hasn't been dismounted as soon as - * nfs_doio() completes for the last buffer. - */ - nmp = nfs_iodmount[myiod]; - if (nmp == NULL) - break; - - /* - * If there are more than one iod on this mount, then defect - * so that the iods can be shared out fairly between the mounts - */ - if (nfs_defect && nmp->nm_bufqiods > 1) { - NFS_DPF(ASYNCIO, - ("nfssvc_iod: iod %d defecting from mount %p\n", - myiod, nmp)); - nfs_iodmount[myiod] = NULL; - nmp->nm_bufqiods--; - break; - } - } - } -finish: - nfs_asyncdaemon[myiod] = 0; - if (nmp) - nmp->nm_bufqiods--; - nfs_iodwant[myiod] = NFSIOD_NOT_AVAILABLE; - nfs_iodmount[myiod] = NULL; - /* Someone may be waiting for the last nfsiod to terminate. */ - if (--nfs_numasync == 0) - wakeup(&nfs_numasync); - mtx_unlock(&nfs_iod_mtx); - if ((error == 0) || (error == EWOULDBLOCK)) - kproc_exit(0); - /* Abnormal termination */ - kproc_exit(1); -} diff --git a/sys/nfsclient/nfs_node.c b/sys/nfsclient/nfs_node.c deleted file mode 100644 index cee4343..0000000 --- a/sys/nfsclient/nfs_node.c +++ /dev/null @@ -1,276 +0,0 @@ -/*- - * Copyright (c) 1989, 1993 - * The Regents of the University of California. All rights reserved. - * - * This code is derived from software contributed to Berkeley by - * Rick Macklem at The University of Guelph. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 4. Neither the name of the University nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * @(#)nfs_node.c 8.6 (Berkeley) 5/22/95 - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/fcntl.h> -#include <sys/fnv_hash.h> -#include <sys/lock.h> -#include <sys/malloc.h> -#include <sys/mbuf.h> -#include <sys/mount.h> -#include <sys/namei.h> -#include <sys/proc.h> -#include <sys/socket.h> -#include <sys/sysctl.h> -#include <sys/taskqueue.h> -#include <sys/vnode.h> - -#include <vm/uma.h> - -#include <nfs/nfsproto.h> -#include <nfs/nfs_lock.h> -#include <nfsclient/nfs.h> -#include <nfsclient/nfsnode.h> -#include <nfsclient/nfsmount.h> - -static uma_zone_t nfsnode_zone; - -static void nfs_freesillyrename(void *arg, __unused int pending); - -#define TRUE 1 -#define FALSE 0 - -void -nfs_nhinit(void) -{ - - nfsnode_zone = uma_zcreate("NFSNODE", sizeof(struct nfsnode), NULL, - NULL, NULL, NULL, UMA_ALIGN_PTR, 0); -} - -void -nfs_nhuninit(void) -{ - uma_zdestroy(nfsnode_zone); -} - -struct nfs_vncmp { - int fhsize; - void *fh; -}; - -static int -nfs_vncmpf(struct vnode *vp, void *arg) -{ - struct nfs_vncmp *a; - struct nfsnode *np; - - a = arg; - np = VTONFS(vp); - return (bcmp(a->fh, np->n_fhp, a->fhsize)); -} - -/* - * Look up a vnode/nfsnode by file handle. - * Callers must check for mount points!! - * In all cases, a pointer to a - * nfsnode structure is returned. - */ -int -nfs_nget(struct mount *mntp, nfsfh_t *fhp, int fhsize, struct nfsnode **npp, int flags) -{ - struct thread *td = curthread; /* XXX */ - struct nfsnode *np; - struct vnode *vp; - struct vnode *nvp; - int error; - u_int hash; - struct nfsmount *nmp; - struct nfs_vncmp ncmp; - - nmp = VFSTONFS(mntp); - *npp = NULL; - - hash = fnv_32_buf(fhp->fh_bytes, fhsize, FNV1_32_INIT); - ncmp.fhsize = fhsize; - ncmp.fh = fhp; - - error = vfs_hash_get(mntp, hash, flags, - td, &nvp, nfs_vncmpf, &ncmp); - if (error) - return (error); - if (nvp != NULL) { - *npp = VTONFS(nvp); - return (0); - } - np = uma_zalloc(nfsnode_zone, M_WAITOK | M_ZERO); - - error = getnewvnode("nfs", mntp, &nfs_vnodeops, &nvp); - if (error) { - uma_zfree(nfsnode_zone, np); - return (error); - } - vp = nvp; - vp->v_bufobj.bo_ops = &buf_ops_nfs; - vp->v_data = np; - np->n_vnode = vp; - /* - * Initialize the mutex even if the vnode is going to be a loser. - * This simplifies the logic in reclaim, which can then unconditionally - * destroy the mutex (in the case of the loser, or if hash_insert happened - * to return an error no special casing is needed). - */ - mtx_init(&np->n_mtx, "NFSnode lock", NULL, MTX_DEF); - /* - * NFS supports recursive and shared locking. - */ - lockmgr(vp->v_vnlock, LK_EXCLUSIVE | LK_NOWITNESS, NULL); - VN_LOCK_AREC(vp); - VN_LOCK_ASHARE(vp); - if (fhsize > NFS_SMALLFH) { - np->n_fhp = malloc(fhsize, M_NFSBIGFH, M_WAITOK); - } else - np->n_fhp = &np->n_fh; - bcopy((caddr_t)fhp, (caddr_t)np->n_fhp, fhsize); - np->n_fhsize = fhsize; - error = insmntque(vp, mntp); - if (error != 0) { - *npp = NULL; - if (np->n_fhsize > NFS_SMALLFH) { - free((caddr_t)np->n_fhp, M_NFSBIGFH); - } - mtx_destroy(&np->n_mtx); - uma_zfree(nfsnode_zone, np); - return (error); - } - error = vfs_hash_insert(vp, hash, flags, - td, &nvp, nfs_vncmpf, &ncmp); - if (error) - return (error); - if (nvp != NULL) { - *npp = VTONFS(nvp); - /* vfs_hash_insert() vput()'s the losing vnode */ - return (0); - } - *npp = np; - - return (0); -} - -/* - * Do the vrele(sp->s_dvp) as a separate task in order to avoid a - * deadlock because of a LOR when vrele() locks the directory vnode. - */ -static void -nfs_freesillyrename(void *arg, __unused int pending) -{ - struct sillyrename *sp; - - sp = arg; - vrele(sp->s_dvp); - free(sp, M_NFSREQ); -} - -int -nfs_inactive(struct vop_inactive_args *ap) -{ - struct nfsnode *np; - struct sillyrename *sp; - struct thread *td = curthread; /* XXX */ - - np = VTONFS(ap->a_vp); - mtx_lock(&np->n_mtx); - if (ap->a_vp->v_type != VDIR) { - sp = np->n_sillyrename; - np->n_sillyrename = NULL; - } else - sp = NULL; - if (sp) { - mtx_unlock(&np->n_mtx); - (void)nfs_vinvalbuf(ap->a_vp, 0, td, 1); - /* - * Remove the silly file that was rename'd earlier - */ - (sp->s_removeit)(sp); - crfree(sp->s_cred); - TASK_INIT(&sp->s_task, 0, nfs_freesillyrename, sp); - taskqueue_enqueue(taskqueue_thread, &sp->s_task); - mtx_lock(&np->n_mtx); - } - np->n_flag &= NMODIFIED; - mtx_unlock(&np->n_mtx); - return (0); -} - -/* - * Reclaim an nfsnode so that it can be used for other purposes. - */ -int -nfs_reclaim(struct vop_reclaim_args *ap) -{ - struct vnode *vp = ap->a_vp; - struct nfsnode *np = VTONFS(vp); - struct nfsdmap *dp, *dp2; - - /* - * If the NLM is running, give it a chance to abort pending - * locks. - */ - if (nfs_reclaim_p) - nfs_reclaim_p(ap); - - /* - * Destroy the vm object and flush associated pages. - */ - vnode_destroy_vobject(vp); - - vfs_hash_remove(vp); - - /* - * Free up any directory cookie structures and - * large file handle structures that might be associated with - * this nfs node. - */ - if (vp->v_type == VDIR) { - dp = LIST_FIRST(&np->n_cookies); - while (dp) { - dp2 = dp; - dp = LIST_NEXT(dp, ndm_list); - free((caddr_t)dp2, M_NFSDIROFF); - } - } - if (np->n_writecred != NULL) - crfree(np->n_writecred); - if (np->n_fhsize > NFS_SMALLFH) { - free((caddr_t)np->n_fhp, M_NFSBIGFH); - } - mtx_destroy(&np->n_mtx); - uma_zfree(nfsnode_zone, vp->v_data); - vp->v_data = NULL; - return (0); -} diff --git a/sys/nfsclient/nfs_subs.c b/sys/nfsclient/nfs_subs.c deleted file mode 100644 index d4fd5eb..0000000 --- a/sys/nfsclient/nfs_subs.c +++ /dev/null @@ -1,1140 +0,0 @@ -/*- - * Copyright (c) 1989, 1993 - * The Regents of the University of California. All rights reserved. - * - * This code is derived from software contributed to Berkeley by - * Rick Macklem at The University of Guelph. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 4. Neither the name of the University nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * @(#)nfs_subs.c 8.8 (Berkeley) 5/22/95 - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -/* - * These functions support the macros and help fiddle mbuf chains for - * the nfs op functions. They do things like create the rpc header and - * copy data between mbuf chains and uio lists. - */ - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/kernel.h> -#include <sys/bio.h> -#include <sys/buf.h> -#include <sys/proc.h> -#include <sys/mount.h> -#include <sys/vnode.h> -#include <sys/namei.h> -#include <sys/mbuf.h> -#include <sys/socket.h> -#include <sys/stat.h> -#include <sys/malloc.h> -#include <sys/rwlock.h> -#include <sys/sysent.h> -#include <sys/syscall.h> -#include <sys/sysproto.h> -#include <sys/taskqueue.h> - -#include <vm/vm.h> -#include <vm/vm_object.h> -#include <vm/vm_extern.h> -#include <vm/uma.h> - -#include <nfs/nfsproto.h> -#include <nfsclient/nfs.h> -#include <nfsclient/nfsnode.h> -#include <nfs/nfs_kdtrace.h> -#include <nfs/xdr_subs.h> -#include <nfsclient/nfsm_subs.h> -#include <nfsclient/nfsmount.h> - -#include <netinet/in.h> - -/* - * Note that stdarg.h and the ANSI style va_start macro is used for both - * ANSI and traditional C compilers. - */ -#include <machine/stdarg.h> - -#ifdef KDTRACE_HOOKS -dtrace_nfsclient_attrcache_flush_probe_func_t - dtrace_nfsclient_attrcache_flush_done_probe; -uint32_t nfsclient_attrcache_flush_done_id; - -dtrace_nfsclient_attrcache_get_hit_probe_func_t - dtrace_nfsclient_attrcache_get_hit_probe; -uint32_t nfsclient_attrcache_get_hit_id; - -dtrace_nfsclient_attrcache_get_miss_probe_func_t - dtrace_nfsclient_attrcache_get_miss_probe; -uint32_t nfsclient_attrcache_get_miss_id; - -dtrace_nfsclient_attrcache_load_probe_func_t - dtrace_nfsclient_attrcache_load_done_probe; -uint32_t nfsclient_attrcache_load_done_id; -#endif /* !KDTRACE_HOOKS */ - -/* - * Data items converted to xdr at startup, since they are constant - * This is kinda hokey, but may save a little time doing byte swaps - */ -u_int32_t nfs_xdrneg1; -u_int32_t nfs_true, nfs_false; - -/* And other global data */ -static u_int32_t nfs_xid = 0; -static enum vtype nv2tov_type[8]= { - VNON, VREG, VDIR, VBLK, VCHR, VLNK, VNON, VNON -}; - -int nfs_ticks; -int nfs_pbuf_freecnt = -1; /* start out unlimited */ - -struct nfs_bufq nfs_bufq; -static struct mtx nfs_xid_mtx; -struct task nfs_nfsiodnew_task; - -/* - * and the reverse mapping from generic to Version 2 procedure numbers - */ -int nfsv2_procid[NFS_NPROCS] = { - NFSV2PROC_NULL, - NFSV2PROC_GETATTR, - NFSV2PROC_SETATTR, - NFSV2PROC_LOOKUP, - NFSV2PROC_NOOP, - NFSV2PROC_READLINK, - NFSV2PROC_READ, - NFSV2PROC_WRITE, - NFSV2PROC_CREATE, - NFSV2PROC_MKDIR, - NFSV2PROC_SYMLINK, - NFSV2PROC_CREATE, - NFSV2PROC_REMOVE, - NFSV2PROC_RMDIR, - NFSV2PROC_RENAME, - NFSV2PROC_LINK, - NFSV2PROC_READDIR, - NFSV2PROC_NOOP, - NFSV2PROC_STATFS, - NFSV2PROC_NOOP, - NFSV2PROC_NOOP, - NFSV2PROC_NOOP, - NFSV2PROC_NOOP, -}; - -LIST_HEAD(nfsnodehashhead, nfsnode); - -u_int32_t -nfs_xid_gen(void) -{ - uint32_t xid; - - mtx_lock(&nfs_xid_mtx); - - /* Get a pretty random xid to start with */ - if (!nfs_xid) - nfs_xid = random(); - /* - * Skip zero xid if it should ever happen. - */ - if (++nfs_xid == 0) - nfs_xid++; - xid = nfs_xid; - mtx_unlock(&nfs_xid_mtx); - return xid; -} - -/* - * copies a uio scatter/gather list to an mbuf chain. - * NOTE: can ony handle iovcnt == 1 - */ -int -nfsm_uiotombuf(struct uio *uiop, struct mbuf **mq, int siz, caddr_t *bpos) -{ - char *uiocp; - struct mbuf *mp, *mp2; - int xfer, left, mlen; - int uiosiz, clflg, rem; - char *cp; - - KASSERT(uiop->uio_iovcnt == 1, ("nfsm_uiotombuf: iovcnt != 1")); - - if (siz > MLEN) /* or should it >= MCLBYTES ?? */ - clflg = 1; - else - clflg = 0; - rem = nfsm_rndup(siz)-siz; - mp = mp2 = *mq; - while (siz > 0) { - left = uiop->uio_iov->iov_len; - uiocp = uiop->uio_iov->iov_base; - if (left > siz) - left = siz; - uiosiz = left; - while (left > 0) { - mlen = M_TRAILINGSPACE(mp); - if (mlen == 0) { - if (clflg) - mp = m_getcl(M_WAITOK, MT_DATA, 0); - else - mp = m_get(M_WAITOK, MT_DATA); - mp2->m_next = mp; - mp2 = mp; - mlen = M_TRAILINGSPACE(mp); - } - xfer = (left > mlen) ? mlen : left; -#ifdef notdef - /* Not Yet.. */ - if (uiop->uio_iov->iov_op != NULL) - (*(uiop->uio_iov->iov_op)) - (uiocp, mtod(mp, caddr_t)+mp->m_len, xfer); - else -#endif - if (uiop->uio_segflg == UIO_SYSSPACE) - bcopy(uiocp, mtod(mp, caddr_t)+mp->m_len, xfer); - else - copyin(uiocp, mtod(mp, caddr_t)+mp->m_len, xfer); - mp->m_len += xfer; - left -= xfer; - uiocp += xfer; - uiop->uio_offset += xfer; - uiop->uio_resid -= xfer; - } - uiop->uio_iov->iov_base = - (char *)uiop->uio_iov->iov_base + uiosiz; - uiop->uio_iov->iov_len -= uiosiz; - siz -= uiosiz; - } - if (rem > 0) { - if (rem > M_TRAILINGSPACE(mp)) { - mp = m_get(M_WAITOK, MT_DATA); - mp2->m_next = mp; - } - cp = mtod(mp, caddr_t)+mp->m_len; - for (left = 0; left < rem; left++) - *cp++ = '\0'; - mp->m_len += rem; - *bpos = cp; - } else - *bpos = mtod(mp, caddr_t)+mp->m_len; - *mq = mp; - return (0); -} - -/* - * Copy a string into mbufs for the hard cases... - */ -int -nfsm_strtmbuf(struct mbuf **mb, char **bpos, const char *cp, long siz) -{ - struct mbuf *m1 = NULL, *m2; - long left, xfer, len, tlen; - u_int32_t *tl; - int putsize; - - putsize = 1; - m2 = *mb; - left = M_TRAILINGSPACE(m2); - if (left > 0) { - tl = ((u_int32_t *)(*bpos)); - *tl++ = txdr_unsigned(siz); - putsize = 0; - left -= NFSX_UNSIGNED; - m2->m_len += NFSX_UNSIGNED; - if (left > 0) { - bcopy(cp, (caddr_t) tl, left); - siz -= left; - cp += left; - m2->m_len += left; - left = 0; - } - } - /* Loop around adding mbufs */ - while (siz > 0) { - if (siz > MLEN) { - m1 = m_getcl(M_WAITOK, MT_DATA, 0); - m1->m_len = MCLBYTES; - } else { - m1 = m_get(M_WAITOK, MT_DATA); - m1->m_len = MLEN; - } - m2->m_next = m1; - m2 = m1; - tl = mtod(m1, u_int32_t *); - tlen = 0; - if (putsize) { - *tl++ = txdr_unsigned(siz); - m1->m_len -= NFSX_UNSIGNED; - tlen = NFSX_UNSIGNED; - putsize = 0; - } - if (siz < m1->m_len) { - len = nfsm_rndup(siz); - xfer = siz; - if (xfer < len) - *(tl+(xfer>>2)) = 0; - } else { - xfer = len = m1->m_len; - } - bcopy(cp, (caddr_t) tl, xfer); - m1->m_len = len+tlen; - siz -= xfer; - cp += xfer; - } - *mb = m1; - *bpos = mtod(m1, caddr_t)+m1->m_len; - return (0); -} - -/* - * Called once to initialize data structures... - */ -int -nfs_init(struct vfsconf *vfsp) -{ - int i; - - nfsmount_zone = uma_zcreate("NFSMOUNT", sizeof(struct nfsmount), - NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0); - nfs_true = txdr_unsigned(TRUE); - nfs_false = txdr_unsigned(FALSE); - nfs_xdrneg1 = txdr_unsigned(-1); - nfs_ticks = (hz * NFS_TICKINTVL + 500) / 1000; - if (nfs_ticks < 1) - nfs_ticks = 1; - /* Ensure async daemons disabled */ - for (i = 0; i < NFS_MAXASYNCDAEMON; i++) { - nfs_iodwant[i] = NFSIOD_NOT_AVAILABLE; - nfs_iodmount[i] = NULL; - } - nfs_nhinit(); /* Init the nfsnode table */ - - /* - * Initialize reply list and start timer - */ - mtx_init(&nfs_iod_mtx, "NFS iod lock", NULL, MTX_DEF); - mtx_init(&nfs_xid_mtx, "NFS xid lock", NULL, MTX_DEF); - TASK_INIT(&nfs_nfsiodnew_task, 0, nfs_nfsiodnew_tq, NULL); - - nfs_pbuf_freecnt = nswbuf / 2 + 1; - - return (0); -} - -int -nfs_uninit(struct vfsconf *vfsp) -{ - int i; - - /* - * Tell all nfsiod processes to exit. Clear nfs_iodmax, and wakeup - * any sleeping nfsiods so they check nfs_iodmax and exit. - * Drain nfsiodnew task before we wait for them to finish. - */ - mtx_lock(&nfs_iod_mtx); - nfs_iodmax = 0; - mtx_unlock(&nfs_iod_mtx); - taskqueue_drain(taskqueue_thread, &nfs_nfsiodnew_task); - mtx_lock(&nfs_iod_mtx); - for (i = 0; i < nfs_numasync; i++) - if (nfs_iodwant[i] == NFSIOD_AVAILABLE) - wakeup(&nfs_iodwant[i]); - /* The last nfsiod to exit will wake us up when nfs_numasync hits 0 */ - while (nfs_numasync) - msleep(&nfs_numasync, &nfs_iod_mtx, PWAIT, "ioddie", 0); - mtx_unlock(&nfs_iod_mtx); - nfs_nhuninit(); - uma_zdestroy(nfsmount_zone); - return (0); -} - -void -nfs_dircookie_lock(struct nfsnode *np) -{ - mtx_lock(&np->n_mtx); - while (np->n_flag & NDIRCOOKIELK) - (void) msleep(&np->n_flag, &np->n_mtx, PZERO, "nfsdirlk", 0); - np->n_flag |= NDIRCOOKIELK; - mtx_unlock(&np->n_mtx); -} - -void -nfs_dircookie_unlock(struct nfsnode *np) -{ - mtx_lock(&np->n_mtx); - np->n_flag &= ~NDIRCOOKIELK; - wakeup(&np->n_flag); - mtx_unlock(&np->n_mtx); -} - -int -nfs_upgrade_vnlock(struct vnode *vp) -{ - int old_lock; - - ASSERT_VOP_LOCKED(vp, "nfs_upgrade_vnlock"); - old_lock = VOP_ISLOCKED(vp); - if (old_lock != LK_EXCLUSIVE) { - KASSERT(old_lock == LK_SHARED, - ("nfs_upgrade_vnlock: wrong old_lock %d", old_lock)); - /* Upgrade to exclusive lock, this might block */ - vn_lock(vp, LK_UPGRADE | LK_RETRY); - } - return (old_lock); -} - -void -nfs_downgrade_vnlock(struct vnode *vp, int old_lock) -{ - if (old_lock != LK_EXCLUSIVE) { - KASSERT(old_lock == LK_SHARED, ("wrong old_lock %d", old_lock)); - /* Downgrade from exclusive lock. */ - vn_lock(vp, LK_DOWNGRADE | LK_RETRY); - } -} - -void -nfs_printf(const char *fmt, ...) -{ - va_list ap; - - mtx_lock(&Giant); - va_start(ap, fmt); - vprintf(fmt, ap); - va_end(ap); - mtx_unlock(&Giant); -} - -/* - * Attribute cache routines. - * nfs_loadattrcache() - loads or updates the cache contents from attributes - * that are on the mbuf list - * nfs_getattrcache() - returns valid attributes if found in cache, returns - * error otherwise - */ - -/* - * Load the attribute cache (that lives in the nfsnode entry) with - * the values on the mbuf list and - * Iff vap not NULL - * copy the attributes to *vaper - */ -int -nfs_loadattrcache(struct vnode **vpp, struct mbuf **mdp, caddr_t *dposp, - struct vattr *vaper, int dontshrink) -{ - struct vnode *vp = *vpp; - struct vattr *vap; - struct nfs_fattr *fp; - struct nfsnode *np = NULL; - int32_t t1; - caddr_t cp2; - int rdev; - struct mbuf *md; - enum vtype vtyp; - u_short vmode; - struct timespec mtime, mtime_save; - int v3 = NFS_ISV3(vp); - int error = 0; - u_quad_t nsize; - int setnsize; - - md = *mdp; - t1 = (mtod(md, caddr_t) + md->m_len) - *dposp; - cp2 = nfsm_disct(mdp, dposp, NFSX_FATTR(v3), t1, M_WAITOK); - if (cp2 == NULL) { - error = EBADRPC; - goto out; - } - fp = (struct nfs_fattr *)cp2; - if (v3) { - vtyp = nfsv3tov_type(fp->fa_type); - vmode = fxdr_unsigned(u_short, fp->fa_mode); - rdev = makedev(fxdr_unsigned(int, fp->fa3_rdev.specdata1), - fxdr_unsigned(int, fp->fa3_rdev.specdata2)); - fxdr_nfsv3time(&fp->fa3_mtime, &mtime); - } else { - vtyp = nfsv2tov_type(fp->fa_type); - vmode = fxdr_unsigned(u_short, fp->fa_mode); - /* - * XXX - * - * The duplicate information returned in fa_type and fa_mode - * is an ambiguity in the NFS version 2 protocol. - * - * VREG should be taken literally as a regular file. If a - * server intents to return some type information differently - * in the upper bits of the mode field (e.g. for sockets, or - * FIFOs), NFSv2 mandates fa_type to be VNON. Anyway, we - * leave the examination of the mode bits even in the VREG - * case to avoid breakage for bogus servers, but we make sure - * that there are actually type bits set in the upper part of - * fa_mode (and failing that, trust the va_type field). - * - * NFSv3 cleared the issue, and requires fa_mode to not - * contain any type information (while also introduing sockets - * and FIFOs for fa_type). - */ - if (vtyp == VNON || (vtyp == VREG && (vmode & S_IFMT) != 0)) - vtyp = IFTOVT(vmode); - rdev = fxdr_unsigned(int32_t, fp->fa2_rdev); - fxdr_nfsv2time(&fp->fa2_mtime, &mtime); - - /* - * Really ugly NFSv2 kludge. - */ - if (vtyp == VCHR && rdev == 0xffffffff) - vtyp = VFIFO; - } - - /* - * If v_type == VNON it is a new node, so fill in the v_type, - * n_mtime fields. Check to see if it represents a special - * device, and if so, check for a possible alias. Once the - * correct vnode has been obtained, fill in the rest of the - * information. - */ - np = VTONFS(vp); - mtx_lock(&np->n_mtx); - if (vp->v_type != vtyp) { - vp->v_type = vtyp; - if (vp->v_type == VFIFO) - vp->v_op = &nfs_fifoops; - np->n_mtime = mtime; - } - vap = &np->n_vattr; - vap->va_type = vtyp; - vap->va_mode = (vmode & 07777); - vap->va_rdev = rdev; - mtime_save = vap->va_mtime; - vap->va_mtime = mtime; - vap->va_fsid = vp->v_mount->mnt_stat.f_fsid.val[0]; - if (v3) { - vap->va_nlink = fxdr_unsigned(u_short, fp->fa_nlink); - vap->va_uid = fxdr_unsigned(uid_t, fp->fa_uid); - vap->va_gid = fxdr_unsigned(gid_t, fp->fa_gid); - vap->va_size = fxdr_hyper(&fp->fa3_size); - vap->va_blocksize = NFS_FABLKSIZE; - vap->va_bytes = fxdr_hyper(&fp->fa3_used); - vap->va_fileid = fxdr_unsigned(int32_t, - fp->fa3_fileid.nfsuquad[1]); - fxdr_nfsv3time(&fp->fa3_atime, &vap->va_atime); - fxdr_nfsv3time(&fp->fa3_ctime, &vap->va_ctime); - vap->va_flags = 0; - vap->va_filerev = 0; - } else { - vap->va_nlink = fxdr_unsigned(u_short, fp->fa_nlink); - vap->va_uid = fxdr_unsigned(uid_t, fp->fa_uid); - vap->va_gid = fxdr_unsigned(gid_t, fp->fa_gid); - vap->va_size = fxdr_unsigned(u_int32_t, fp->fa2_size); - vap->va_blocksize = fxdr_unsigned(int32_t, fp->fa2_blocksize); - vap->va_bytes = (u_quad_t)fxdr_unsigned(int32_t, fp->fa2_blocks) - * NFS_FABLKSIZE; - vap->va_fileid = fxdr_unsigned(int32_t, fp->fa2_fileid); - fxdr_nfsv2time(&fp->fa2_atime, &vap->va_atime); - vap->va_flags = 0; - vap->va_ctime.tv_sec = fxdr_unsigned(u_int32_t, - fp->fa2_ctime.nfsv2_sec); - vap->va_ctime.tv_nsec = 0; - vap->va_gen = fxdr_unsigned(u_int32_t, fp->fa2_ctime.nfsv2_usec); - vap->va_filerev = 0; - } - np->n_attrstamp = time_second; - setnsize = 0; - nsize = 0; - if (vap->va_size != np->n_size) { - if (vap->va_type == VREG) { - if (dontshrink && vap->va_size < np->n_size) { - /* - * We've been told not to shrink the file; - * zero np->n_attrstamp to indicate that - * the attributes are stale. - */ - vap->va_size = np->n_size; - np->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp); - vnode_pager_setsize(vp, np->n_size); - } else if (np->n_flag & NMODIFIED) { - /* - * We've modified the file: Use the larger - * of our size, and the server's size. - */ - if (vap->va_size < np->n_size) { - vap->va_size = np->n_size; - } else { - np->n_size = vap->va_size; - np->n_flag |= NSIZECHANGED; - } - vnode_pager_setsize(vp, np->n_size); - } else if (vap->va_size < np->n_size) { - /* - * When shrinking the size, the call to - * vnode_pager_setsize() cannot be done - * with the mutex held, so delay it until - * after the mtx_unlock call. - */ - nsize = np->n_size = vap->va_size; - np->n_flag |= NSIZECHANGED; - setnsize = 1; - } else { - np->n_size = vap->va_size; - np->n_flag |= NSIZECHANGED; - vnode_pager_setsize(vp, np->n_size); - } - } else { - np->n_size = vap->va_size; - } - } - /* - * The following checks are added to prevent a race between (say) - * a READDIR+ and a WRITE. - * READDIR+, WRITE requests sent out. - * READDIR+ resp, WRITE resp received on client. - * However, the WRITE resp was handled before the READDIR+ resp - * causing the post op attrs from the write to be loaded first - * and the attrs from the READDIR+ to be loaded later. If this - * happens, we have stale attrs loaded into the attrcache. - * We detect this by for the mtime moving back. We invalidate the - * attrcache when this happens. - */ - if (timespeccmp(&mtime_save, &vap->va_mtime, >)) { - /* Size changed or mtime went backwards */ - np->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp); - } - if (vaper != NULL) { - bcopy((caddr_t)vap, (caddr_t)vaper, sizeof(*vap)); - if (np->n_flag & NCHG) { - if (np->n_flag & NACC) - vaper->va_atime = np->n_atim; - if (np->n_flag & NUPD) - vaper->va_mtime = np->n_mtim; - } - } - -#ifdef KDTRACE_HOOKS - if (np->n_attrstamp != 0) - KDTRACE_NFS_ATTRCACHE_LOAD_DONE(vp, &np->n_vattr, 0); -#endif - mtx_unlock(&np->n_mtx); - if (setnsize) - vnode_pager_setsize(vp, nsize); -out: -#ifdef KDTRACE_HOOKS - if (error) - KDTRACE_NFS_ATTRCACHE_LOAD_DONE(vp, NULL, error); -#endif - return (error); -} - -#ifdef NFS_ACDEBUG -#include <sys/sysctl.h> -SYSCTL_DECL(_vfs_oldnfs); -static int nfs_acdebug; -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, acdebug, CTLFLAG_RW, &nfs_acdebug, 0, - "Toggle acdebug (attribute cache debug) flag"); -#endif - -/* - * Check the time stamp - * If the cache is valid, copy contents to *vap and return 0 - * otherwise return an error - */ -int -nfs_getattrcache(struct vnode *vp, struct vattr *vaper) -{ - struct nfsnode *np; - struct vattr *vap; - struct nfsmount *nmp; - int timeo; - - np = VTONFS(vp); - vap = &np->n_vattr; - nmp = VFSTONFS(vp->v_mount); -#ifdef NFS_ACDEBUG - mtx_lock(&Giant); /* nfs_printf() */ -#endif - mtx_lock(&np->n_mtx); - /* XXX n_mtime doesn't seem to be updated on a miss-and-reload */ - timeo = (time_second - np->n_mtime.tv_sec) / 10; - -#ifdef NFS_ACDEBUG - if (nfs_acdebug>1) - nfs_printf("nfs_getattrcache: initial timeo = %d\n", timeo); -#endif - - if (vap->va_type == VDIR) { - if ((np->n_flag & NMODIFIED) || timeo < nmp->nm_acdirmin) - timeo = nmp->nm_acdirmin; - else if (timeo > nmp->nm_acdirmax) - timeo = nmp->nm_acdirmax; - } else { - if ((np->n_flag & NMODIFIED) || timeo < nmp->nm_acregmin) - timeo = nmp->nm_acregmin; - else if (timeo > nmp->nm_acregmax) - timeo = nmp->nm_acregmax; - } - -#ifdef NFS_ACDEBUG - if (nfs_acdebug > 2) - nfs_printf("acregmin %d; acregmax %d; acdirmin %d; acdirmax %d\n", - nmp->nm_acregmin, nmp->nm_acregmax, - nmp->nm_acdirmin, nmp->nm_acdirmax); - - if (nfs_acdebug) - nfs_printf("nfs_getattrcache: age = %d; final timeo = %d\n", - (time_second - np->n_attrstamp), timeo); -#endif - - if ((time_second - np->n_attrstamp) >= timeo) { - nfsstats.attrcache_misses++; - mtx_unlock(&np->n_mtx); -#ifdef NFS_ACDEBUG - mtx_unlock(&Giant); /* nfs_printf() */ -#endif - KDTRACE_NFS_ATTRCACHE_GET_MISS(vp); - return (ENOENT); - } - nfsstats.attrcache_hits++; - if (vap->va_size != np->n_size) { - if (vap->va_type == VREG) { - if (np->n_flag & NMODIFIED) { - if (vap->va_size < np->n_size) - vap->va_size = np->n_size; - else - np->n_size = vap->va_size; - } else { - np->n_size = vap->va_size; - } - vnode_pager_setsize(vp, np->n_size); - } else { - np->n_size = vap->va_size; - } - } - bcopy((caddr_t)vap, (caddr_t)vaper, sizeof(struct vattr)); - if (np->n_flag & NCHG) { - if (np->n_flag & NACC) - vaper->va_atime = np->n_atim; - if (np->n_flag & NUPD) - vaper->va_mtime = np->n_mtim; - } - mtx_unlock(&np->n_mtx); -#ifdef NFS_ACDEBUG - mtx_unlock(&Giant); /* nfs_printf() */ -#endif - KDTRACE_NFS_ATTRCACHE_GET_HIT(vp, vap); - return (0); -} - -/* - * Purge all cached information about an NFS vnode including name - * cache entries, the attribute cache, and the access cache. This is - * called when an NFS request for a node fails with a stale - * filehandle. - */ -void -nfs_purgecache(struct vnode *vp) -{ - struct nfsnode *np; - int i; - - np = VTONFS(vp); - cache_purge(vp); - mtx_lock(&np->n_mtx); - np->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp); - for (i = 0; i < NFS_ACCESSCACHESIZE; i++) - np->n_accesscache[i].stamp = 0; - KDTRACE_NFS_ACCESSCACHE_FLUSH_DONE(vp); - mtx_unlock(&np->n_mtx); -} - -static nfsuint64 nfs_nullcookie = { { 0, 0 } }; -/* - * This function finds the directory cookie that corresponds to the - * logical byte offset given. - */ -nfsuint64 * -nfs_getcookie(struct nfsnode *np, off_t off, int add) -{ - struct nfsdmap *dp, *dp2; - int pos; - nfsuint64 *retval = NULL; - - pos = (uoff_t)off / NFS_DIRBLKSIZ; - if (pos == 0 || off < 0) { - KASSERT(!add, ("nfs getcookie add at <= 0")); - return (&nfs_nullcookie); - } - pos--; - dp = LIST_FIRST(&np->n_cookies); - if (!dp) { - if (add) { - dp = malloc(sizeof (struct nfsdmap), - M_NFSDIROFF, M_WAITOK); - dp->ndm_eocookie = 0; - LIST_INSERT_HEAD(&np->n_cookies, dp, ndm_list); - } else - goto out; - } - while (pos >= NFSNUMCOOKIES) { - pos -= NFSNUMCOOKIES; - if (LIST_NEXT(dp, ndm_list)) { - if (!add && dp->ndm_eocookie < NFSNUMCOOKIES && - pos >= dp->ndm_eocookie) - goto out; - dp = LIST_NEXT(dp, ndm_list); - } else if (add) { - dp2 = malloc(sizeof (struct nfsdmap), - M_NFSDIROFF, M_WAITOK); - dp2->ndm_eocookie = 0; - LIST_INSERT_AFTER(dp, dp2, ndm_list); - dp = dp2; - } else - goto out; - } - if (pos >= dp->ndm_eocookie) { - if (add) - dp->ndm_eocookie = pos + 1; - else - goto out; - } - retval = &dp->ndm_cookies[pos]; -out: - return (retval); -} - -/* - * Invalidate cached directory information, except for the actual directory - * blocks (which are invalidated separately). - * Done mainly to avoid the use of stale offset cookies. - */ -void -nfs_invaldir(struct vnode *vp) -{ - struct nfsnode *np = VTONFS(vp); - - KASSERT(vp->v_type == VDIR, ("nfs: invaldir not dir")); - nfs_dircookie_lock(np); - np->n_direofoffset = 0; - np->n_cookieverf.nfsuquad[0] = 0; - np->n_cookieverf.nfsuquad[1] = 0; - if (LIST_FIRST(&np->n_cookies)) - LIST_FIRST(&np->n_cookies)->ndm_eocookie = 0; - nfs_dircookie_unlock(np); -} - -/* - * The write verifier has changed (probably due to a server reboot), so all - * B_NEEDCOMMIT blocks will have to be written again. Since they are on the - * dirty block list as B_DELWRI, all this takes is clearing the B_NEEDCOMMIT - * and B_CLUSTEROK flags. Once done the new write verifier can be set for the - * mount point. - * - * B_CLUSTEROK must be cleared along with B_NEEDCOMMIT because stage 1 data - * writes are not clusterable. - */ -void -nfs_clearcommit(struct mount *mp) -{ - struct vnode *vp, *nvp; - struct buf *bp, *nbp; - struct bufobj *bo; - - MNT_VNODE_FOREACH_ALL(vp, mp, nvp) { - bo = &vp->v_bufobj; - vholdl(vp); - VI_UNLOCK(vp); - BO_LOCK(bo); - TAILQ_FOREACH_SAFE(bp, &bo->bo_dirty.bv_hd, b_bobufs, nbp) { - if (!BUF_ISLOCKED(bp) && - (bp->b_flags & (B_DELWRI | B_NEEDCOMMIT)) - == (B_DELWRI | B_NEEDCOMMIT)) - bp->b_flags &= ~(B_NEEDCOMMIT | B_CLUSTEROK); - } - BO_UNLOCK(bo); - vdrop(vp); - } -} - -/* - * Helper functions for former macros. Some of these should be - * moved to their callers. - */ - -int -nfsm_mtofh_xx(struct vnode *d, struct vnode **v, int v3, int *f, - struct mbuf **md, caddr_t *dpos) -{ - struct nfsnode *ttnp; - struct vnode *ttvp; - nfsfh_t *ttfhp; - u_int32_t *tl; - int ttfhsize; - int t1; - - if (v3) { - tl = nfsm_dissect_xx(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - *f = fxdr_unsigned(int, *tl); - } else - *f = 1; - if (*f) { - t1 = nfsm_getfh_xx(&ttfhp, &ttfhsize, (v3), md, dpos); - if (t1 != 0) - return t1; - t1 = nfs_nget(d->v_mount, ttfhp, ttfhsize, &ttnp, LK_EXCLUSIVE); - if (t1 != 0) - return t1; - *v = NFSTOV(ttnp); - } - if (v3) { - tl = nfsm_dissect_xx(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - if (*f) - *f = fxdr_unsigned(int, *tl); - else if (fxdr_unsigned(int, *tl)) - nfsm_adv_xx(NFSX_V3FATTR, md, dpos); - } - if (*f) { - ttvp = *v; - t1 = nfs_loadattrcache(&ttvp, md, dpos, NULL, 0); - if (t1) - return t1; - *v = ttvp; - } - return 0; -} - -int -nfsm_getfh_xx(nfsfh_t **f, int *s, int v3, struct mbuf **md, caddr_t *dpos) -{ - u_int32_t *tl; - - if (v3) { - tl = nfsm_dissect_xx(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - *s = fxdr_unsigned(int, *tl); - if (*s <= 0 || *s > NFSX_V3FHMAX) - return EBADRPC; - } else - *s = NFSX_V2FH; - *f = nfsm_dissect_xx(nfsm_rndup(*s), md, dpos); - if (*f == NULL) - return EBADRPC; - else - return 0; -} - - -int -nfsm_loadattr_xx(struct vnode **v, struct vattr *va, struct mbuf **md, - caddr_t *dpos) -{ - int t1; - - struct vnode *ttvp = *v; - t1 = nfs_loadattrcache(&ttvp, md, dpos, va, 0); - if (t1 != 0) - return t1; - *v = ttvp; - return 0; -} - -int -nfsm_postop_attr_xx(struct vnode **v, int *f, struct vattr *va, - struct mbuf **md, caddr_t *dpos) -{ - u_int32_t *tl; - int t1; - - struct vnode *ttvp = *v; - tl = nfsm_dissect_xx(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - *f = fxdr_unsigned(int, *tl); - if (*f != 0) { - t1 = nfs_loadattrcache(&ttvp, md, dpos, va, 1); - if (t1 != 0) { - *f = 0; - return t1; - } - *v = ttvp; - } - return 0; -} - -int -nfsm_wcc_data_xx(struct vnode **v, int *f, struct mbuf **md, caddr_t *dpos) -{ - u_int32_t *tl; - int ttattrf, ttretf = 0; - int t1; - - tl = nfsm_dissect_xx(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - if (*tl == nfs_true) { - tl = nfsm_dissect_xx(6 * NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - mtx_lock(&(VTONFS(*v))->n_mtx); - if (*f) - ttretf = (VTONFS(*v)->n_mtime.tv_sec == fxdr_unsigned(u_int32_t, *(tl + 2)) && - VTONFS(*v)->n_mtime.tv_nsec == fxdr_unsigned(u_int32_t, *(tl + 3))); - mtx_unlock(&(VTONFS(*v))->n_mtx); - } - t1 = nfsm_postop_attr_xx(v, &ttattrf, NULL, md, dpos); - if (t1) - return t1; - if (*f) - *f = ttretf; - else - *f = ttattrf; - return 0; -} - -int -nfsm_strtom_xx(const char *a, int s, int m, struct mbuf **mb, caddr_t *bpos) -{ - u_int32_t *tl; - int t1; - - if (s > m) - return ENAMETOOLONG; - t1 = nfsm_rndup(s) + NFSX_UNSIGNED; - if (t1 <= M_TRAILINGSPACE(*mb)) { - tl = nfsm_build_xx(t1, mb, bpos); - *tl++ = txdr_unsigned(s); - *(tl + ((t1 >> 2) - 2)) = 0; - bcopy(a, tl, s); - } else { - t1 = nfsm_strtmbuf(mb, bpos, a, s); - if (t1 != 0) - return t1; - } - return 0; -} - -int -nfsm_fhtom_xx(struct vnode *v, int v3, struct mbuf **mb, caddr_t *bpos) -{ - u_int32_t *tl; - int t1; - caddr_t cp; - - if (v3) { - t1 = nfsm_rndup(VTONFS(v)->n_fhsize) + NFSX_UNSIGNED; - if (t1 < M_TRAILINGSPACE(*mb)) { - tl = nfsm_build_xx(t1, mb, bpos); - *tl++ = txdr_unsigned(VTONFS(v)->n_fhsize); - *(tl + ((t1 >> 2) - 2)) = 0; - bcopy(VTONFS(v)->n_fhp, tl, VTONFS(v)->n_fhsize); - } else { - t1 = nfsm_strtmbuf(mb, bpos, - (const char *)VTONFS(v)->n_fhp, - VTONFS(v)->n_fhsize); - if (t1 != 0) - return t1; - } - } else { - cp = nfsm_build_xx(NFSX_V2FH, mb, bpos); - bcopy(VTONFS(v)->n_fhp, cp, NFSX_V2FH); - } - return 0; -} - -void -nfsm_v3attrbuild_xx(struct vattr *va, int full, struct mbuf **mb, - caddr_t *bpos) -{ - u_int32_t *tl; - - if (va->va_mode != (mode_t)VNOVAL) { - tl = nfsm_build_xx(2 * NFSX_UNSIGNED, mb, bpos); - *tl++ = nfs_true; - *tl = txdr_unsigned(va->va_mode); - } else { - tl = nfsm_build_xx(NFSX_UNSIGNED, mb, bpos); - *tl = nfs_false; - } - if (full && va->va_uid != (uid_t)VNOVAL) { - tl = nfsm_build_xx(2 * NFSX_UNSIGNED, mb, bpos); - *tl++ = nfs_true; - *tl = txdr_unsigned(va->va_uid); - } else { - tl = nfsm_build_xx(NFSX_UNSIGNED, mb, bpos); - *tl = nfs_false; - } - if (full && va->va_gid != (gid_t)VNOVAL) { - tl = nfsm_build_xx(2 * NFSX_UNSIGNED, mb, bpos); - *tl++ = nfs_true; - *tl = txdr_unsigned(va->va_gid); - } else { - tl = nfsm_build_xx(NFSX_UNSIGNED, mb, bpos); - *tl = nfs_false; - } - if (full && va->va_size != VNOVAL) { - tl = nfsm_build_xx(3 * NFSX_UNSIGNED, mb, bpos); - *tl++ = nfs_true; - txdr_hyper(va->va_size, tl); - } else { - tl = nfsm_build_xx(NFSX_UNSIGNED, mb, bpos); - *tl = nfs_false; - } - if (va->va_atime.tv_sec != VNOVAL) { - if ((va->va_vaflags & VA_UTIMES_NULL) == 0) { - tl = nfsm_build_xx(3 * NFSX_UNSIGNED, mb, bpos); - *tl++ = txdr_unsigned(NFSV3SATTRTIME_TOCLIENT); - txdr_nfsv3time(&va->va_atime, tl); - } else { - tl = nfsm_build_xx(NFSX_UNSIGNED, mb, bpos); - *tl = txdr_unsigned(NFSV3SATTRTIME_TOSERVER); - } - } else { - tl = nfsm_build_xx(NFSX_UNSIGNED, mb, bpos); - *tl = txdr_unsigned(NFSV3SATTRTIME_DONTCHANGE); - } - if (va->va_mtime.tv_sec != VNOVAL) { - if ((va->va_vaflags & VA_UTIMES_NULL) == 0) { - tl = nfsm_build_xx(3 * NFSX_UNSIGNED, mb, bpos); - *tl++ = txdr_unsigned(NFSV3SATTRTIME_TOCLIENT); - txdr_nfsv3time(&va->va_mtime, tl); - } else { - tl = nfsm_build_xx(NFSX_UNSIGNED, mb, bpos); - *tl = txdr_unsigned(NFSV3SATTRTIME_TOSERVER); - } - } else { - tl = nfsm_build_xx(NFSX_UNSIGNED, mb, bpos); - *tl = txdr_unsigned(NFSV3SATTRTIME_DONTCHANGE); - } -} diff --git a/sys/nfsclient/nfs_vfsops.c b/sys/nfsclient/nfs_vfsops.c deleted file mode 100644 index 0c002d2..0000000 --- a/sys/nfsclient/nfs_vfsops.c +++ /dev/null @@ -1,1582 +0,0 @@ -/*- - * Copyright (c) 1989, 1993, 1995 - * The Regents of the University of California. All rights reserved. - * - * This code is derived from software contributed to Berkeley by - * Rick Macklem at The University of Guelph. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 4. Neither the name of the University nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * @(#)nfs_vfsops.c 8.12 (Berkeley) 5/20/95 - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - - -#include "opt_bootp.h" -#include "opt_nfsroot.h" - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/kernel.h> -#include <sys/bio.h> -#include <sys/buf.h> -#include <sys/jail.h> -#include <sys/limits.h> -#include <sys/lock.h> -#include <sys/malloc.h> -#include <sys/mbuf.h> -#include <sys/module.h> -#include <sys/mount.h> -#include <sys/proc.h> -#include <sys/socket.h> -#include <sys/socketvar.h> -#include <sys/sockio.h> -#include <sys/sysctl.h> -#include <sys/syslog.h> -#include <sys/vnode.h> -#include <sys/signalvar.h> - -#include <vm/vm.h> -#include <vm/vm_extern.h> -#include <vm/uma.h> - -#include <net/if.h> -#include <net/if_var.h> -#include <net/route.h> -#include <net/vnet.h> - -#include <netinet/in.h> - -#include <rpc/rpc.h> - -#include <nfs/nfsproto.h> -#include <nfsclient/nfs.h> -#include <nfsclient/nfsnode.h> -#include <nfsclient/nfsmount.h> -#include <nfs/xdr_subs.h> -#include <nfsclient/nfsm_subs.h> -#include <nfs/nfsdiskless.h> - -FEATURE(nfsclient, "NFS client"); - -MALLOC_DEFINE(M_NFSREQ, "nfsclient_req", "NFS request header"); -MALLOC_DEFINE(M_NFSBIGFH, "nfsclient_bigfh", "NFS version 3 file handle"); -MALLOC_DEFINE(M_NFSDIROFF, "nfsclient_diroff", "NFS directory offset data"); -MALLOC_DEFINE(M_NFSHASH, "nfsclient_hash", "NFS hash tables"); -MALLOC_DEFINE(M_NFSDIRECTIO, "nfsclient_directio", "NFS Direct IO async write state"); - -uma_zone_t nfsmount_zone; - -struct nfsstats nfsstats; - -SYSCTL_NODE(_vfs, OID_AUTO, oldnfs, CTLFLAG_RW, 0, "Old NFS filesystem"); -SYSCTL_STRUCT(_vfs_oldnfs, NFS_NFSSTATS, nfsstats, CTLFLAG_RW, - &nfsstats, nfsstats, "S,nfsstats"); -static int nfs_ip_paranoia = 1; -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, nfs_ip_paranoia, CTLFLAG_RW, - &nfs_ip_paranoia, 0, - "Disallow accepting replies from IPs which differ from those sent"); -#ifdef NFS_DEBUG -int nfs_debug; -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, debug, CTLFLAG_RW, &nfs_debug, 0, - "Toggle debug flag"); -#endif -static int nfs_tprintf_initial_delay = NFS_TPRINTF_INITIAL_DELAY; -SYSCTL_INT(_vfs_oldnfs, NFS_TPRINTF_INITIAL_DELAY, - downdelayinitial, CTLFLAG_RW, &nfs_tprintf_initial_delay, 0, - "Delay before printing \"nfs server not responding\" messages"); -/* how long between console messages "nfs server foo not responding" */ -static int nfs_tprintf_delay = NFS_TPRINTF_DELAY; -SYSCTL_INT(_vfs_oldnfs, NFS_TPRINTF_DELAY, - downdelayinterval, CTLFLAG_RW, &nfs_tprintf_delay, 0, - "Delay between printing \"nfs server not responding\" messages"); - -static void nfs_decode_args(struct mount *mp, struct nfsmount *nmp, - struct nfs_args *argp, const char *hostname); -static int mountnfs(struct nfs_args *, struct mount *, - struct sockaddr *, char *, struct vnode **, - struct ucred *cred, int, int); -static void nfs_getnlminfo(struct vnode *, uint8_t *, size_t *, - struct sockaddr_storage *, int *, off_t *, - struct timeval *); -static vfs_mount_t nfs_mount; -static vfs_cmount_t nfs_cmount; -static vfs_unmount_t nfs_unmount; -static vfs_root_t nfs_root; -static vfs_statfs_t nfs_statfs; -static vfs_sync_t nfs_sync; -static vfs_sysctl_t nfs_sysctl; - -static int fake_wchan; - -/* - * nfs vfs operations. - */ -static struct vfsops nfs_vfsops = { - .vfs_init = nfs_init, - .vfs_mount = nfs_mount, - .vfs_cmount = nfs_cmount, - .vfs_root = nfs_root, - .vfs_statfs = nfs_statfs, - .vfs_sync = nfs_sync, - .vfs_uninit = nfs_uninit, - .vfs_unmount = nfs_unmount, - .vfs_sysctl = nfs_sysctl, -}; -VFS_SET(nfs_vfsops, oldnfs, VFCF_NETWORK | VFCF_SBDRY); - -/* So that loader and kldload(2) can find us, wherever we are.. */ -MODULE_VERSION(oldnfs, 1); -MODULE_DEPEND(oldnfs, krpc, 1, 1, 1); -#ifdef KGSSAPI -MODULE_DEPEND(oldnfs, kgssapi, 1, 1, 1); -#endif -MODULE_DEPEND(oldnfs, nfs_common, 1, 1, 1); -MODULE_DEPEND(oldnfs, nfslock, 1, 1, 1); - -static struct nfs_rpcops nfs_rpcops = { - nfs_readrpc, - nfs_writerpc, - nfs_writebp, - nfs_readlinkrpc, - nfs_invaldir, - nfs_commit, -}; - -/* - * This structure is now defined in sys/nfs/nfs_diskless.c so that it - * can be shared by both NFS clients. It is declared here so that it - * will be defined for kernels built without NFS_ROOT, although it - * isn't used in that case. - */ -#ifndef NFS_ROOT -struct nfs_diskless nfs_diskless = { { { 0 } } }; -struct nfsv3_diskless nfsv3_diskless = { { { 0 } } }; -int nfs_diskless_valid = 0; -#endif - -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, diskless_valid, CTLFLAG_RD, - &nfs_diskless_valid, 0, - "Has the diskless struct been filled correctly"); - -SYSCTL_STRING(_vfs_oldnfs, OID_AUTO, diskless_rootpath, CTLFLAG_RD, - nfsv3_diskless.root_hostnam, 0, "Path to nfs root"); - -SYSCTL_OPAQUE(_vfs_oldnfs, OID_AUTO, diskless_rootaddr, CTLFLAG_RD, - &nfsv3_diskless.root_saddr, sizeof nfsv3_diskless.root_saddr, - "%Ssockaddr_in", "Diskless root nfs address"); - - -void nfsargs_ntoh(struct nfs_args *); -static int nfs_mountdiskless(char *, - struct sockaddr_in *, struct nfs_args *, - struct thread *, struct vnode **, struct mount *); -static void nfs_convert_diskless(void); -static void nfs_convert_oargs(struct nfs_args *args, - struct onfs_args *oargs); - -int -nfs_iosize(struct nfsmount *nmp) -{ - int iosize; - - /* - * Calculate the size used for io buffers. Use the larger - * of the two sizes to minimise nfs requests but make sure - * that it is at least one VM page to avoid wasting buffer - * space. - */ - iosize = imax(nmp->nm_rsize, nmp->nm_wsize); - iosize = imax(iosize, PAGE_SIZE); - return (iosize); -} - -static void -nfs_convert_oargs(struct nfs_args *args, struct onfs_args *oargs) -{ - - args->version = NFS_ARGSVERSION; - args->addr = oargs->addr; - args->addrlen = oargs->addrlen; - args->sotype = oargs->sotype; - args->proto = oargs->proto; - args->fh = oargs->fh; - args->fhsize = oargs->fhsize; - args->flags = oargs->flags; - args->wsize = oargs->wsize; - args->rsize = oargs->rsize; - args->readdirsize = oargs->readdirsize; - args->timeo = oargs->timeo; - args->retrans = oargs->retrans; - args->maxgrouplist = oargs->maxgrouplist; - args->readahead = oargs->readahead; - args->deadthresh = oargs->deadthresh; - args->hostname = oargs->hostname; -} - -static void -nfs_convert_diskless(void) -{ - - bcopy(&nfs_diskless.myif, &nfsv3_diskless.myif, - sizeof(struct ifaliasreq)); - bcopy(&nfs_diskless.mygateway, &nfsv3_diskless.mygateway, - sizeof(struct sockaddr_in)); - nfs_convert_oargs(&nfsv3_diskless.root_args,&nfs_diskless.root_args); - if (nfsv3_diskless.root_args.flags & NFSMNT_NFSV3) { - nfsv3_diskless.root_fhsize = NFSX_V3FH; - bcopy(nfs_diskless.root_fh, nfsv3_diskless.root_fh, NFSX_V3FH); - } else { - nfsv3_diskless.root_fhsize = NFSX_V2FH; - bcopy(nfs_diskless.root_fh, nfsv3_diskless.root_fh, NFSX_V2FH); - } - bcopy(&nfs_diskless.root_saddr,&nfsv3_diskless.root_saddr, - sizeof(struct sockaddr_in)); - bcopy(nfs_diskless.root_hostnam, nfsv3_diskless.root_hostnam, MNAMELEN); - nfsv3_diskless.root_time = nfs_diskless.root_time; - bcopy(nfs_diskless.my_hostnam, nfsv3_diskless.my_hostnam, - MAXHOSTNAMELEN); - nfs_diskless_valid = 3; -} - -/* - * nfs statfs call - */ -static int -nfs_statfs(struct mount *mp, struct statfs *sbp) -{ - struct vnode *vp; - struct thread *td; - struct nfs_statfs *sfp; - caddr_t bpos, dpos; - struct nfsmount *nmp = VFSTONFS(mp); - int error = 0, v3 = (nmp->nm_flag & NFSMNT_NFSV3), retattr; - struct mbuf *mreq, *mrep, *md, *mb; - struct nfsnode *np; - u_quad_t tquad; - - td = curthread; -#ifndef nolint - sfp = NULL; -#endif - error = vfs_busy(mp, MBF_NOWAIT); - if (error) - return (error); - error = nfs_nget(mp, (nfsfh_t *)nmp->nm_fh, nmp->nm_fhsize, &np, LK_EXCLUSIVE); - if (error) { - vfs_unbusy(mp); - return (error); - } - vp = NFSTOV(np); - mtx_lock(&nmp->nm_mtx); - if (v3 && (nmp->nm_state & NFSSTA_GOTFSINFO) == 0) { - mtx_unlock(&nmp->nm_mtx); - (void)nfs_fsinfo(nmp, vp, td->td_ucred, td); - } else - mtx_unlock(&nmp->nm_mtx); - nfsstats.rpccnt[NFSPROC_FSSTAT]++; - mreq = m_get2(NFSX_FH(v3), M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(vp, v3); - nfsm_request(vp, NFSPROC_FSSTAT, td, td->td_ucred); - if (v3) - nfsm_postop_attr(vp, retattr); - if (error) { - if (mrep != NULL) - m_freem(mrep); - goto nfsmout; - } - sfp = nfsm_dissect(struct nfs_statfs *, NFSX_STATFS(v3)); - mtx_lock(&nmp->nm_mtx); - sbp->f_iosize = nfs_iosize(nmp); - mtx_unlock(&nmp->nm_mtx); - if (v3) { - sbp->f_bsize = NFS_FABLKSIZE; - tquad = fxdr_hyper(&sfp->sf_tbytes); - sbp->f_blocks = tquad / NFS_FABLKSIZE; - tquad = fxdr_hyper(&sfp->sf_fbytes); - sbp->f_bfree = tquad / NFS_FABLKSIZE; - tquad = fxdr_hyper(&sfp->sf_abytes); - sbp->f_bavail = tquad / NFS_FABLKSIZE; - sbp->f_files = (fxdr_unsigned(int32_t, - sfp->sf_tfiles.nfsuquad[1]) & 0x7fffffff); - sbp->f_ffree = (fxdr_unsigned(int32_t, - sfp->sf_ffiles.nfsuquad[1]) & 0x7fffffff); - } else { - sbp->f_bsize = fxdr_unsigned(int32_t, sfp->sf_bsize); - sbp->f_blocks = fxdr_unsigned(int32_t, sfp->sf_blocks); - sbp->f_bfree = fxdr_unsigned(int32_t, sfp->sf_bfree); - sbp->f_bavail = fxdr_unsigned(int32_t, sfp->sf_bavail); - sbp->f_files = 0; - sbp->f_ffree = 0; - } - m_freem(mrep); -nfsmout: - vput(vp); - vfs_unbusy(mp); - return (error); -} - -/* - * nfs version 3 fsinfo rpc call - */ -int -nfs_fsinfo(struct nfsmount *nmp, struct vnode *vp, struct ucred *cred, - struct thread *td) -{ - struct nfsv3_fsinfo *fsp; - u_int32_t pref, max; - caddr_t bpos, dpos; - int error = 0, retattr; - struct mbuf *mreq, *mrep, *md, *mb; - u_int64_t maxfsize; - - nfsstats.rpccnt[NFSPROC_FSINFO]++; - mreq = m_get2(NFSX_FH(1), M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(vp, 1); - nfsm_request(vp, NFSPROC_FSINFO, td, cred); - nfsm_postop_attr(vp, retattr); - if (!error) { - fsp = nfsm_dissect(struct nfsv3_fsinfo *, NFSX_V3FSINFO); - pref = fxdr_unsigned(u_int32_t, fsp->fs_wtpref); - mtx_lock(&nmp->nm_mtx); - if (pref < nmp->nm_wsize && pref >= NFS_FABLKSIZE) - nmp->nm_wsize = (pref + NFS_FABLKSIZE - 1) & - ~(NFS_FABLKSIZE - 1); - max = fxdr_unsigned(u_int32_t, fsp->fs_wtmax); - if (max < nmp->nm_wsize && max > 0) { - nmp->nm_wsize = max & ~(NFS_FABLKSIZE - 1); - if (nmp->nm_wsize == 0) - nmp->nm_wsize = max; - } - pref = fxdr_unsigned(u_int32_t, fsp->fs_rtpref); - if (pref < nmp->nm_rsize && pref >= NFS_FABLKSIZE) - nmp->nm_rsize = (pref + NFS_FABLKSIZE - 1) & - ~(NFS_FABLKSIZE - 1); - max = fxdr_unsigned(u_int32_t, fsp->fs_rtmax); - if (max < nmp->nm_rsize && max > 0) { - nmp->nm_rsize = max & ~(NFS_FABLKSIZE - 1); - if (nmp->nm_rsize == 0) - nmp->nm_rsize = max; - } - pref = fxdr_unsigned(u_int32_t, fsp->fs_dtpref); - if (pref < nmp->nm_readdirsize && pref >= NFS_DIRBLKSIZ) - nmp->nm_readdirsize = (pref + NFS_DIRBLKSIZ - 1) & - ~(NFS_DIRBLKSIZ - 1); - if (max < nmp->nm_readdirsize && max > 0) { - nmp->nm_readdirsize = max & ~(NFS_DIRBLKSIZ - 1); - if (nmp->nm_readdirsize == 0) - nmp->nm_readdirsize = max; - } - maxfsize = fxdr_hyper(&fsp->fs_maxfilesize); - if (maxfsize > 0 && maxfsize < nmp->nm_maxfilesize) - nmp->nm_maxfilesize = maxfsize; - nmp->nm_mountp->mnt_stat.f_iosize = nfs_iosize(nmp); - nmp->nm_state |= NFSSTA_GOTFSINFO; - mtx_unlock(&nmp->nm_mtx); - } - m_freem(mrep); -nfsmout: - return (error); -} - -/* - * Mount a remote root fs via. nfs. This depends on the info in the - * nfs_diskless structure that has been filled in properly by some primary - * bootstrap. - * It goes something like this: - * - do enough of "ifconfig" by calling ifioctl() so that the system - * can talk to the server - * - If nfs_diskless.mygateway is filled in, use that address as - * a default gateway. - * - build the rootfs mount point and call mountnfs() to do the rest. - * - * It is assumed to be safe to read, modify, and write the nfsv3_diskless - * structure, as well as other global NFS client variables here, as - * nfs_mountroot() will be called once in the boot before any other NFS - * client activity occurs. - */ -int -nfs_mountroot(struct mount *mp) -{ - struct thread *td = curthread; - struct nfsv3_diskless *nd = &nfsv3_diskless; - struct socket *so; - struct vnode *vp; - struct ifreq ir; - int error; - u_long l; - char buf[128]; - char *cp; - - -#if defined(BOOTP_NFSROOT) && defined(BOOTP) - bootpc_init(); /* use bootp to get nfs_diskless filled in */ -#elif defined(NFS_ROOT) - nfs_setup_diskless(); -#endif - - if (nfs_diskless_valid == 0) { - return (-1); - } - if (nfs_diskless_valid == 1) - nfs_convert_diskless(); - - /* - * XXX splnet, so networks will receive... - */ - splnet(); - - /* - * Do enough of ifconfig(8) so that the critical net interface can - * talk to the server. - */ - error = socreate(nd->myif.ifra_addr.sa_family, &so, nd->root_args.sotype, 0, - td->td_ucred, td); - if (error) - panic("nfs_mountroot: socreate(%04x): %d", - nd->myif.ifra_addr.sa_family, error); - -#if 0 /* XXX Bad idea */ - /* - * We might not have been told the right interface, so we pass - * over the first ten interfaces of the same kind, until we get - * one of them configured. - */ - - for (i = strlen(nd->myif.ifra_name) - 1; - nd->myif.ifra_name[i] >= '0' && - nd->myif.ifra_name[i] <= '9'; - nd->myif.ifra_name[i] ++) { - error = ifioctl(so, SIOCAIFADDR, (caddr_t)&nd->myif, td); - if(!error) - break; - } -#endif - - error = ifioctl(so, SIOCAIFADDR, (caddr_t)&nd->myif, td); - if (error) - panic("nfs_mountroot: SIOCAIFADDR: %d", error); - - if ((cp = kern_getenv("boot.netif.mtu")) != NULL) { - ir.ifr_mtu = strtol(cp, NULL, 10); - bcopy(nd->myif.ifra_name, ir.ifr_name, IFNAMSIZ); - freeenv(cp); - error = ifioctl(so, SIOCSIFMTU, (caddr_t)&ir, td); - if (error) - printf("nfs_mountroot: SIOCSIFMTU: %d", error); - } - soclose(so); - - /* - * If the gateway field is filled in, set it as the default route. - * Note that pxeboot will set a default route of 0 if the route - * is not set by the DHCP server. Check also for a value of 0 - * to avoid panicking inappropriately in that situation. - */ - if (nd->mygateway.sin_len != 0 && - nd->mygateway.sin_addr.s_addr != 0) { - struct sockaddr_in mask, sin; - - bzero((caddr_t)&mask, sizeof(mask)); - sin = mask; - sin.sin_family = AF_INET; - sin.sin_len = sizeof(sin); - /* XXX MRT use table 0 for this sort of thing */ - CURVNET_SET(TD_TO_VNET(td)); - error = rtrequest_fib(RTM_ADD, (struct sockaddr *)&sin, - (struct sockaddr *)&nd->mygateway, - (struct sockaddr *)&mask, - RTF_UP | RTF_GATEWAY, NULL, RT_DEFAULT_FIB); - CURVNET_RESTORE(); - if (error) - panic("nfs_mountroot: RTM_ADD: %d", error); - } - - /* - * Create the rootfs mount point. - */ - nd->root_args.fh = nd->root_fh; - nd->root_args.fhsize = nd->root_fhsize; - l = ntohl(nd->root_saddr.sin_addr.s_addr); - snprintf(buf, sizeof(buf), "%ld.%ld.%ld.%ld:%s", - (l >> 24) & 0xff, (l >> 16) & 0xff, - (l >> 8) & 0xff, (l >> 0) & 0xff, nd->root_hostnam); - printf("NFS ROOT: %s\n", buf); - nd->root_args.hostname = buf; - if ((error = nfs_mountdiskless(buf, - &nd->root_saddr, &nd->root_args, td, &vp, mp)) != 0) { - return (error); - } - - /* - * This is not really an nfs issue, but it is much easier to - * set hostname here and then let the "/etc/rc.xxx" files - * mount the right /var based upon its preset value. - */ - mtx_lock(&prison0.pr_mtx); - strlcpy(prison0.pr_hostname, nd->my_hostnam, - sizeof (prison0.pr_hostname)); - mtx_unlock(&prison0.pr_mtx); - inittodr(ntohl(nd->root_time)); - return (0); -} - -/* - * Internal version of mount system call for diskless setup. - */ -static int -nfs_mountdiskless(char *path, - struct sockaddr_in *sin, struct nfs_args *args, struct thread *td, - struct vnode **vpp, struct mount *mp) -{ - struct sockaddr *nam; - int error; - - nam = sodupsockaddr((struct sockaddr *)sin, M_WAITOK); - if ((error = mountnfs(args, mp, nam, path, vpp, td->td_ucred, - NFS_DEFAULT_NAMETIMEO, NFS_DEFAULT_NEGNAMETIMEO)) != 0) { - printf("nfs_mountroot: mount %s on /: %d\n", path, error); - return (error); - } - return (0); -} - -static int -nfs_sec_name_to_num(char *sec) -{ - if (!strcmp(sec, "krb5")) - return (RPCSEC_GSS_KRB5); - if (!strcmp(sec, "krb5i")) - return (RPCSEC_GSS_KRB5I); - if (!strcmp(sec, "krb5p")) - return (RPCSEC_GSS_KRB5P); - if (!strcmp(sec, "sys")) - return (AUTH_SYS); - /* - * Userland should validate the string but we will try and - * cope with unexpected values. - */ - return (AUTH_SYS); -} - -static void -nfs_decode_args(struct mount *mp, struct nfsmount *nmp, struct nfs_args *argp, - const char *hostname) -{ - int s; - int adjsock; - int maxio; - char *p; - char *secname; - char *principal; - - s = splnet(); - - /* - * Set read-only flag if requested; otherwise, clear it if this is - * an update. If this is not an update, then either the read-only - * flag is already clear, or this is a root mount and it was set - * intentionally at some previous point. - */ - if (vfs_getopt(mp->mnt_optnew, "ro", NULL, NULL) == 0) { - MNT_ILOCK(mp); - mp->mnt_flag |= MNT_RDONLY; - MNT_IUNLOCK(mp); - } else if (mp->mnt_flag & MNT_UPDATE) { - MNT_ILOCK(mp); - mp->mnt_flag &= ~MNT_RDONLY; - MNT_IUNLOCK(mp); - } - - /* - * Silently clear NFSMNT_NOCONN if it's a TCP mount, it makes - * no sense in that context. Also, set up appropriate retransmit - * and soft timeout behavior. - */ - if (argp->sotype == SOCK_STREAM) { - nmp->nm_flag &= ~NFSMNT_NOCONN; - nmp->nm_flag |= NFSMNT_DUMBTIMR; - nmp->nm_timeo = NFS_MAXTIMEO; - nmp->nm_retry = NFS_RETRANS_TCP; - } - - /* Also clear RDIRPLUS if not NFSv3, it crashes some servers */ - if ((argp->flags & NFSMNT_NFSV3) == 0) - nmp->nm_flag &= ~NFSMNT_RDIRPLUS; - - /* Re-bind if rsrvd port requested and wasn't on one */ - adjsock = !(nmp->nm_flag & NFSMNT_RESVPORT) - && (argp->flags & NFSMNT_RESVPORT); - /* Also re-bind if we're switching to/from a connected UDP socket */ - adjsock |= ((nmp->nm_flag & NFSMNT_NOCONN) != - (argp->flags & NFSMNT_NOCONN)); - - /* Update flags atomically. Don't change the lock bits. */ - nmp->nm_flag = argp->flags | nmp->nm_flag; - splx(s); - - if ((argp->flags & NFSMNT_TIMEO) && argp->timeo > 0) { - nmp->nm_timeo = (argp->timeo * NFS_HZ + 5) / 10; - if (nmp->nm_timeo < NFS_MINTIMEO) - nmp->nm_timeo = NFS_MINTIMEO; - else if (nmp->nm_timeo > NFS_MAXTIMEO) - nmp->nm_timeo = NFS_MAXTIMEO; - } - - if ((argp->flags & NFSMNT_RETRANS) && argp->retrans > 1) { - nmp->nm_retry = argp->retrans; - if (nmp->nm_retry > NFS_MAXREXMIT) - nmp->nm_retry = NFS_MAXREXMIT; - } - - if (argp->flags & NFSMNT_NFSV3) { - if (argp->sotype == SOCK_DGRAM) - maxio = NFS_MAXDGRAMDATA; - else - maxio = NFS_MAXDATA; - } else - maxio = NFS_V2MAXDATA; - - if ((argp->flags & NFSMNT_WSIZE) && argp->wsize > 0) { - nmp->nm_wsize = argp->wsize; - /* Round down to multiple of blocksize */ - nmp->nm_wsize &= ~(NFS_FABLKSIZE - 1); - if (nmp->nm_wsize <= 0) - nmp->nm_wsize = NFS_FABLKSIZE; - } - if (nmp->nm_wsize > maxio) - nmp->nm_wsize = maxio; - if (nmp->nm_wsize > MAXBSIZE) - nmp->nm_wsize = MAXBSIZE; - - if ((argp->flags & NFSMNT_RSIZE) && argp->rsize > 0) { - nmp->nm_rsize = argp->rsize; - /* Round down to multiple of blocksize */ - nmp->nm_rsize &= ~(NFS_FABLKSIZE - 1); - if (nmp->nm_rsize <= 0) - nmp->nm_rsize = NFS_FABLKSIZE; - } - if (nmp->nm_rsize > maxio) - nmp->nm_rsize = maxio; - if (nmp->nm_rsize > MAXBSIZE) - nmp->nm_rsize = MAXBSIZE; - - if ((argp->flags & NFSMNT_READDIRSIZE) && argp->readdirsize > 0) { - nmp->nm_readdirsize = argp->readdirsize; - } - if (nmp->nm_readdirsize > maxio) - nmp->nm_readdirsize = maxio; - if (nmp->nm_readdirsize > nmp->nm_rsize) - nmp->nm_readdirsize = nmp->nm_rsize; - - if ((argp->flags & NFSMNT_ACREGMIN) && argp->acregmin >= 0) - nmp->nm_acregmin = argp->acregmin; - else - nmp->nm_acregmin = NFS_MINATTRTIMO; - if ((argp->flags & NFSMNT_ACREGMAX) && argp->acregmax >= 0) - nmp->nm_acregmax = argp->acregmax; - else - nmp->nm_acregmax = NFS_MAXATTRTIMO; - if ((argp->flags & NFSMNT_ACDIRMIN) && argp->acdirmin >= 0) - nmp->nm_acdirmin = argp->acdirmin; - else - nmp->nm_acdirmin = NFS_MINDIRATTRTIMO; - if ((argp->flags & NFSMNT_ACDIRMAX) && argp->acdirmax >= 0) - nmp->nm_acdirmax = argp->acdirmax; - else - nmp->nm_acdirmax = NFS_MAXDIRATTRTIMO; - if (nmp->nm_acdirmin > nmp->nm_acdirmax) - nmp->nm_acdirmin = nmp->nm_acdirmax; - if (nmp->nm_acregmin > nmp->nm_acregmax) - nmp->nm_acregmin = nmp->nm_acregmax; - - if ((argp->flags & NFSMNT_MAXGRPS) && argp->maxgrouplist >= 0) { - if (argp->maxgrouplist <= NFS_MAXGRPS) - nmp->nm_numgrps = argp->maxgrouplist; - else - nmp->nm_numgrps = NFS_MAXGRPS; - } - if ((argp->flags & NFSMNT_READAHEAD) && argp->readahead >= 0) { - if (argp->readahead <= NFS_MAXRAHEAD) - nmp->nm_readahead = argp->readahead; - else - nmp->nm_readahead = NFS_MAXRAHEAD; - } - if ((argp->flags & NFSMNT_WCOMMITSIZE) && argp->wcommitsize >= 0) { - if (argp->wcommitsize < nmp->nm_wsize) - nmp->nm_wcommitsize = nmp->nm_wsize; - else - nmp->nm_wcommitsize = argp->wcommitsize; - } - if ((argp->flags & NFSMNT_DEADTHRESH) && argp->deadthresh >= 0) { - if (argp->deadthresh <= NFS_MAXDEADTHRESH) - nmp->nm_deadthresh = argp->deadthresh; - else - nmp->nm_deadthresh = NFS_MAXDEADTHRESH; - } - - adjsock |= ((nmp->nm_sotype != argp->sotype) || - (nmp->nm_soproto != argp->proto)); - nmp->nm_sotype = argp->sotype; - nmp->nm_soproto = argp->proto; - - if (nmp->nm_client && adjsock) { - nfs_safedisconnect(nmp); - if (nmp->nm_sotype == SOCK_DGRAM) - while (nfs_connect(nmp)) { - printf("nfs_args: retrying connect\n"); - (void) tsleep(&fake_wchan, PSOCK, "nfscon", hz); - } - } - - if (hostname) { - strlcpy(nmp->nm_hostname, hostname, - sizeof(nmp->nm_hostname)); - p = strchr(nmp->nm_hostname, ':'); - if (p) - *p = '\0'; - } - - if (vfs_getopt(mp->mnt_optnew, "sec", - (void **) &secname, NULL) == 0) { - nmp->nm_secflavor = nfs_sec_name_to_num(secname); - } else { - nmp->nm_secflavor = AUTH_SYS; - } - - if (vfs_getopt(mp->mnt_optnew, "principal", - (void **) &principal, NULL) == 0) { - strlcpy(nmp->nm_principal, principal, - sizeof(nmp->nm_principal)); - } else { - snprintf(nmp->nm_principal, sizeof(nmp->nm_principal), - "nfs@%s", nmp->nm_hostname); - } -} - -static const char *nfs_opts[] = { "from", "nfs_args", - "noatime", "noexec", "suiddir", "nosuid", "nosymfollow", "union", - "noclusterr", "noclusterw", "multilabel", "acls", "force", "update", - "async", "dumbtimer", "noconn", "nolockd", "intr", "rdirplus", "resvport", - "readahead", "readdirsize", "soft", "hard", "mntudp", "tcp", "udp", - "wsize", "rsize", "retrans", "acregmin", "acregmax", "acdirmin", - "acdirmax", "deadthresh", "hostname", "timeout", "addr", "fh", "nfsv3", - "sec", "maxgroups", "principal", "negnametimeo", "nocto", "wcommitsize", - "nametimeo", - NULL }; - -/* - * VFS Operations. - * - * mount system call - * It seems a bit dumb to copyinstr() the host and path here and then - * bcopy() them in mountnfs(), but I wanted to detect errors before - * doing the sockargs() call because sockargs() allocates an mbuf and - * an error after that means that I have to release the mbuf. - */ -/* ARGSUSED */ -static int -nfs_mount(struct mount *mp) -{ - struct nfs_args args = { - .version = NFS_ARGSVERSION, - .addr = NULL, - .addrlen = sizeof (struct sockaddr_in), - .sotype = SOCK_STREAM, - .proto = 0, - .fh = NULL, - .fhsize = 0, - .flags = NFSMNT_RESVPORT, - .wsize = NFS_WSIZE, - .rsize = NFS_RSIZE, - .readdirsize = NFS_READDIRSIZE, - .timeo = 10, - .retrans = NFS_RETRANS, - .maxgrouplist = NFS_MAXGRPS, - .readahead = NFS_DEFRAHEAD, - .wcommitsize = 0, /* was: NQ_DEFLEASE */ - .deadthresh = NFS_MAXDEADTHRESH, /* was: NQ_DEADTHRESH */ - .hostname = NULL, - /* args version 4 */ - .acregmin = NFS_MINATTRTIMO, - .acregmax = NFS_MAXATTRTIMO, - .acdirmin = NFS_MINDIRATTRTIMO, - .acdirmax = NFS_MAXDIRATTRTIMO, - }; - int error, ret, has_nfs_args_opt; - int has_addr_opt, has_fh_opt, has_hostname_opt; - struct sockaddr *nam; - struct vnode *vp; - char hst[MNAMELEN]; - size_t len; - u_char nfh[NFSX_V3FHMAX]; - char *opt; - int nametimeo = NFS_DEFAULT_NAMETIMEO; - int negnametimeo = NFS_DEFAULT_NEGNAMETIMEO; - - has_nfs_args_opt = 0; - has_addr_opt = 0; - has_fh_opt = 0; - has_hostname_opt = 0; - - if (vfs_filteropt(mp->mnt_optnew, nfs_opts)) { - error = EINVAL; - goto out; - } - - if ((mp->mnt_flag & (MNT_ROOTFS | MNT_UPDATE)) == MNT_ROOTFS) { - error = nfs_mountroot(mp); - goto out; - } - - /* - * The old mount_nfs program passed the struct nfs_args - * from userspace to kernel. The new mount_nfs program - * passes string options via nmount() from userspace to kernel - * and we populate the struct nfs_args in the kernel. - */ - if (vfs_getopt(mp->mnt_optnew, "nfs_args", NULL, NULL) == 0) { - error = vfs_copyopt(mp->mnt_optnew, "nfs_args", &args, - sizeof args); - if (error) - goto out; - - if (args.version != NFS_ARGSVERSION) { - error = EPROGMISMATCH; - goto out; - } - has_nfs_args_opt = 1; - } - - if (vfs_getopt(mp->mnt_optnew, "dumbtimer", NULL, NULL) == 0) - args.flags |= NFSMNT_DUMBTIMR; - if (vfs_getopt(mp->mnt_optnew, "noconn", NULL, NULL) == 0) - args.flags |= NFSMNT_NOCONN; - if (vfs_getopt(mp->mnt_optnew, "conn", NULL, NULL) == 0) - args.flags |= NFSMNT_NOCONN; - if (vfs_getopt(mp->mnt_optnew, "nolockd", NULL, NULL) == 0) - args.flags |= NFSMNT_NOLOCKD; - if (vfs_getopt(mp->mnt_optnew, "lockd", NULL, NULL) == 0) - args.flags &= ~NFSMNT_NOLOCKD; - if (vfs_getopt(mp->mnt_optnew, "intr", NULL, NULL) == 0) - args.flags |= NFSMNT_INT; - if (vfs_getopt(mp->mnt_optnew, "rdirplus", NULL, NULL) == 0) - args.flags |= NFSMNT_RDIRPLUS; - if (vfs_getopt(mp->mnt_optnew, "resvport", NULL, NULL) == 0) - args.flags |= NFSMNT_RESVPORT; - if (vfs_getopt(mp->mnt_optnew, "noresvport", NULL, NULL) == 0) - args.flags &= ~NFSMNT_RESVPORT; - if (vfs_getopt(mp->mnt_optnew, "soft", NULL, NULL) == 0) - args.flags |= NFSMNT_SOFT; - if (vfs_getopt(mp->mnt_optnew, "hard", NULL, NULL) == 0) - args.flags &= ~NFSMNT_SOFT; - if (vfs_getopt(mp->mnt_optnew, "mntudp", NULL, NULL) == 0) - args.sotype = SOCK_DGRAM; - if (vfs_getopt(mp->mnt_optnew, "udp", NULL, NULL) == 0) - args.sotype = SOCK_DGRAM; - if (vfs_getopt(mp->mnt_optnew, "tcp", NULL, NULL) == 0) - args.sotype = SOCK_STREAM; - if (vfs_getopt(mp->mnt_optnew, "nfsv3", NULL, NULL) == 0) - args.flags |= NFSMNT_NFSV3; - if (vfs_getopt(mp->mnt_optnew, "nocto", NULL, NULL) == 0) - args.flags |= NFSMNT_NOCTO; - if (vfs_getopt(mp->mnt_optnew, "readdirsize", (void **)&opt, NULL) == 0) { - if (opt == NULL) { - vfs_mount_error(mp, "illegal readdirsize"); - error = EINVAL; - goto out; - } - ret = sscanf(opt, "%d", &args.readdirsize); - if (ret != 1 || args.readdirsize <= 0) { - vfs_mount_error(mp, "illegal readdirsize: %s", - opt); - error = EINVAL; - goto out; - } - args.flags |= NFSMNT_READDIRSIZE; - } - if (vfs_getopt(mp->mnt_optnew, "readahead", (void **)&opt, NULL) == 0) { - if (opt == NULL) { - vfs_mount_error(mp, "illegal readahead"); - error = EINVAL; - goto out; - } - ret = sscanf(opt, "%d", &args.readahead); - if (ret != 1 || args.readahead <= 0) { - vfs_mount_error(mp, "illegal readahead: %s", - opt); - error = EINVAL; - goto out; - } - args.flags |= NFSMNT_READAHEAD; - } - if (vfs_getopt(mp->mnt_optnew, "wsize", (void **)&opt, NULL) == 0) { - if (opt == NULL) { - vfs_mount_error(mp, "illegal wsize"); - error = EINVAL; - goto out; - } - ret = sscanf(opt, "%d", &args.wsize); - if (ret != 1 || args.wsize <= 0) { - vfs_mount_error(mp, "illegal wsize: %s", - opt); - error = EINVAL; - goto out; - } - args.flags |= NFSMNT_WSIZE; - } - if (vfs_getopt(mp->mnt_optnew, "rsize", (void **)&opt, NULL) == 0) { - if (opt == NULL) { - vfs_mount_error(mp, "illegal rsize"); - error = EINVAL; - goto out; - } - ret = sscanf(opt, "%d", &args.rsize); - if (ret != 1 || args.rsize <= 0) { - vfs_mount_error(mp, "illegal wsize: %s", - opt); - error = EINVAL; - goto out; - } - args.flags |= NFSMNT_RSIZE; - } - if (vfs_getopt(mp->mnt_optnew, "retrans", (void **)&opt, NULL) == 0) { - if (opt == NULL) { - vfs_mount_error(mp, "illegal retrans"); - error = EINVAL; - goto out; - } - ret = sscanf(opt, "%d", &args.retrans); - if (ret != 1 || args.retrans <= 0) { - vfs_mount_error(mp, "illegal retrans: %s", - opt); - error = EINVAL; - goto out; - } - args.flags |= NFSMNT_RETRANS; - } - if (vfs_getopt(mp->mnt_optnew, "acregmin", (void **)&opt, NULL) == 0) { - ret = sscanf(opt, "%d", &args.acregmin); - if (ret != 1 || args.acregmin < 0) { - vfs_mount_error(mp, "illegal acregmin: %s", - opt); - error = EINVAL; - goto out; - } - args.flags |= NFSMNT_ACREGMIN; - } - if (vfs_getopt(mp->mnt_optnew, "acregmax", (void **)&opt, NULL) == 0) { - ret = sscanf(opt, "%d", &args.acregmax); - if (ret != 1 || args.acregmax < 0) { - vfs_mount_error(mp, "illegal acregmax: %s", - opt); - error = EINVAL; - goto out; - } - args.flags |= NFSMNT_ACREGMAX; - } - if (vfs_getopt(mp->mnt_optnew, "acdirmin", (void **)&opt, NULL) == 0) { - ret = sscanf(opt, "%d", &args.acdirmin); - if (ret != 1 || args.acdirmin < 0) { - vfs_mount_error(mp, "illegal acdirmin: %s", - opt); - error = EINVAL; - goto out; - } - args.flags |= NFSMNT_ACDIRMIN; - } - if (vfs_getopt(mp->mnt_optnew, "acdirmax", (void **)&opt, NULL) == 0) { - ret = sscanf(opt, "%d", &args.acdirmax); - if (ret != 1 || args.acdirmax < 0) { - vfs_mount_error(mp, "illegal acdirmax: %s", - opt); - error = EINVAL; - goto out; - } - args.flags |= NFSMNT_ACDIRMAX; - } - if (vfs_getopt(mp->mnt_optnew, "wcommitsize", (void **)&opt, NULL) == 0) { - ret = sscanf(opt, "%d", &args.wcommitsize); - if (ret != 1 || args.wcommitsize < 0) { - vfs_mount_error(mp, "illegal wcommitsize: %s", opt); - error = EINVAL; - goto out; - } - args.flags |= NFSMNT_WCOMMITSIZE; - } - if (vfs_getopt(mp->mnt_optnew, "deadthresh", (void **)&opt, NULL) == 0) { - ret = sscanf(opt, "%d", &args.deadthresh); - if (ret != 1 || args.deadthresh <= 0) { - vfs_mount_error(mp, "illegal deadthresh: %s", - opt); - error = EINVAL; - goto out; - } - args.flags |= NFSMNT_DEADTHRESH; - } - if (vfs_getopt(mp->mnt_optnew, "timeout", (void **)&opt, NULL) == 0) { - ret = sscanf(opt, "%d", &args.timeo); - if (ret != 1 || args.timeo <= 0) { - vfs_mount_error(mp, "illegal timeout: %s", - opt); - error = EINVAL; - goto out; - } - args.flags |= NFSMNT_TIMEO; - } - if (vfs_getopt(mp->mnt_optnew, "maxgroups", (void **)&opt, NULL) == 0) { - ret = sscanf(opt, "%d", &args.maxgrouplist); - if (ret != 1 || args.maxgrouplist <= 0) { - vfs_mount_error(mp, "illegal maxgroups: %s", - opt); - error = EINVAL; - goto out; - } - args.flags |= NFSMNT_MAXGRPS; - } - if (vfs_getopt(mp->mnt_optnew, "nametimeo", (void **)&opt, NULL) == 0) { - ret = sscanf(opt, "%d", &nametimeo); - if (ret != 1 || nametimeo < 0) { - vfs_mount_error(mp, "illegal nametimeo: %s", opt); - error = EINVAL; - goto out; - } - } - if (vfs_getopt(mp->mnt_optnew, "negnametimeo", (void **)&opt, NULL) - == 0) { - ret = sscanf(opt, "%d", &negnametimeo); - if (ret != 1 || negnametimeo < 0) { - vfs_mount_error(mp, "illegal negnametimeo: %s", - opt); - error = EINVAL; - goto out; - } - } - if (vfs_getopt(mp->mnt_optnew, "addr", (void **)&args.addr, - &args.addrlen) == 0) { - has_addr_opt = 1; - if (args.addrlen > SOCK_MAXADDRLEN) { - error = ENAMETOOLONG; - goto out; - } - nam = malloc(args.addrlen, M_SONAME, - M_WAITOK); - bcopy(args.addr, nam, args.addrlen); - nam->sa_len = args.addrlen; - } - if (vfs_getopt(mp->mnt_optnew, "fh", (void **)&args.fh, - &args.fhsize) == 0) { - has_fh_opt = 1; - } - if (vfs_getopt(mp->mnt_optnew, "hostname", (void **)&args.hostname, - NULL) == 0) { - has_hostname_opt = 1; - } - if (args.hostname == NULL) { - vfs_mount_error(mp, "Invalid hostname"); - error = EINVAL; - goto out; - } - if (args.fhsize < 0 || args.fhsize > NFSX_V3FHMAX) { - vfs_mount_error(mp, "Bad file handle"); - error = EINVAL; - goto out; - } - - if (mp->mnt_flag & MNT_UPDATE) { - struct nfsmount *nmp = VFSTONFS(mp); - - if (nmp == NULL) { - error = EIO; - goto out; - } - - /* - * If a change from TCP->UDP is done and there are thread(s) - * that have I/O RPC(s) in progress with a tranfer size - * greater than NFS_MAXDGRAMDATA, those thread(s) will be - * hung, retrying the RPC(s) forever. Usually these threads - * will be seen doing an uninterruptible sleep on wait channel - * "newnfsreq" (truncated to "newnfsre" by procstat). - */ - if (args.sotype == SOCK_DGRAM && nmp->nm_sotype == SOCK_STREAM) - tprintf(curthread->td_proc, LOG_WARNING, - "Warning: mount -u that changes TCP->UDP can result in hung threads\n"); - - /* - * When doing an update, we can't change from or to - * v3, switch lockd strategies or change cookie translation - */ - args.flags = (args.flags & - ~(NFSMNT_NFSV3 | NFSMNT_NOLOCKD /*|NFSMNT_XLATECOOKIE*/)) | - (nmp->nm_flag & - (NFSMNT_NFSV3 | NFSMNT_NOLOCKD /*|NFSMNT_XLATECOOKIE*/)); - nfs_decode_args(mp, nmp, &args, NULL); - goto out; - } - - /* - * Make the nfs_ip_paranoia sysctl serve as the default connection - * or no-connection mode for those protocols that support - * no-connection mode (the flag will be cleared later for protocols - * that do not support no-connection mode). This will allow a client - * to receive replies from a different IP then the request was - * sent to. Note: default value for nfs_ip_paranoia is 1 (paranoid), - * not 0. - */ - if (nfs_ip_paranoia == 0) - args.flags |= NFSMNT_NOCONN; - - if (has_nfs_args_opt) { - /* - * In the 'nfs_args' case, the pointers in the args - * structure are in userland - we copy them in here. - */ - if (!has_fh_opt) { - error = copyin((caddr_t)args.fh, (caddr_t)nfh, - args.fhsize); - if (error) { - goto out; - } - args.fh = nfh; - } - if (!has_hostname_opt) { - error = copyinstr(args.hostname, hst, MNAMELEN-1, &len); - if (error) { - goto out; - } - bzero(&hst[len], MNAMELEN - len); - args.hostname = hst; - } - if (!has_addr_opt) { - /* sockargs() call must be after above copyin() calls */ - error = getsockaddr(&nam, (caddr_t)args.addr, - args.addrlen); - if (error) { - goto out; - } - } - } else if (has_addr_opt == 0) { - vfs_mount_error(mp, "No server address"); - error = EINVAL; - goto out; - } - error = mountnfs(&args, mp, nam, args.hostname, &vp, - curthread->td_ucred, nametimeo, negnametimeo); -out: - if (!error) { - MNT_ILOCK(mp); - mp->mnt_kern_flag |= MNTK_LOOKUP_SHARED; - MNT_IUNLOCK(mp); - } - return (error); -} - - -/* - * VFS Operations. - * - * mount system call - * It seems a bit dumb to copyinstr() the host and path here and then - * bcopy() them in mountnfs(), but I wanted to detect errors before - * doing the sockargs() call because sockargs() allocates an mbuf and - * an error after that means that I have to release the mbuf. - */ -/* ARGSUSED */ -static int -nfs_cmount(struct mntarg *ma, void *data, uint64_t flags) -{ - int error; - struct nfs_args args; - - error = copyin(data, &args, sizeof (struct nfs_args)); - if (error) - return error; - - ma = mount_arg(ma, "nfs_args", &args, sizeof args); - - error = kernel_mount(ma, flags); - return (error); -} - -/* - * Common code for mount and mountroot - */ -static int -mountnfs(struct nfs_args *argp, struct mount *mp, struct sockaddr *nam, - char *hst, struct vnode **vpp, struct ucred *cred, int nametimeo, - int negnametimeo) -{ - struct nfsmount *nmp; - struct nfsnode *np; - int error; - struct vattr attrs; - - if (mp->mnt_flag & MNT_UPDATE) { - nmp = VFSTONFS(mp); - printf("%s: MNT_UPDATE is no longer handled here\n", __func__); - free(nam, M_SONAME); - return (0); - } else { - nmp = uma_zalloc(nfsmount_zone, M_WAITOK); - bzero((caddr_t)nmp, sizeof (struct nfsmount)); - TAILQ_INIT(&nmp->nm_bufq); - mp->mnt_data = nmp; - nmp->nm_getinfo = nfs_getnlminfo; - nmp->nm_vinvalbuf = nfs_vinvalbuf; - } - vfs_getnewfsid(mp); - nmp->nm_mountp = mp; - mtx_init(&nmp->nm_mtx, "NFSmount lock", NULL, MTX_DEF); - - /* - * V2 can only handle 32 bit filesizes. A 4GB-1 limit may be too - * high, depending on whether we end up with negative offsets in - * the client or server somewhere. 2GB-1 may be safer. - * - * For V3, nfs_fsinfo will adjust this as necessary. Assume maximum - * that we can handle until we find out otherwise. - */ - if ((argp->flags & NFSMNT_NFSV3) == 0) - nmp->nm_maxfilesize = 0xffffffffLL; - else - nmp->nm_maxfilesize = OFF_MAX; - - nmp->nm_timeo = NFS_TIMEO; - nmp->nm_retry = NFS_RETRANS; - if ((argp->flags & NFSMNT_NFSV3) && argp->sotype == SOCK_STREAM) { - nmp->nm_wsize = nmp->nm_rsize = NFS_MAXDATA; - } else { - nmp->nm_wsize = NFS_WSIZE; - nmp->nm_rsize = NFS_RSIZE; - } - nmp->nm_wcommitsize = hibufspace / (desiredvnodes / 1000); - nmp->nm_readdirsize = NFS_READDIRSIZE; - nmp->nm_numgrps = NFS_MAXGRPS; - nmp->nm_readahead = NFS_DEFRAHEAD; - nmp->nm_deadthresh = NFS_MAXDEADTHRESH; - nmp->nm_nametimeo = nametimeo; - nmp->nm_negnametimeo = negnametimeo; - nmp->nm_tprintf_delay = nfs_tprintf_delay; - if (nmp->nm_tprintf_delay < 0) - nmp->nm_tprintf_delay = 0; - nmp->nm_tprintf_initial_delay = nfs_tprintf_initial_delay; - if (nmp->nm_tprintf_initial_delay < 0) - nmp->nm_tprintf_initial_delay = 0; - nmp->nm_fhsize = argp->fhsize; - bcopy((caddr_t)argp->fh, (caddr_t)nmp->nm_fh, argp->fhsize); - bcopy(hst, mp->mnt_stat.f_mntfromname, MNAMELEN); - nmp->nm_nam = nam; - /* Set up the sockets and per-host congestion */ - nmp->nm_sotype = argp->sotype; - nmp->nm_soproto = argp->proto; - nmp->nm_rpcops = &nfs_rpcops; - - nfs_decode_args(mp, nmp, argp, hst); - - /* - * For Connection based sockets (TCP,...) defer the connect until - * the first request, in case the server is not responding. - */ - if (nmp->nm_sotype == SOCK_DGRAM && - (error = nfs_connect(nmp))) - goto bad; - - /* - * This is silly, but it has to be set so that vinifod() works. - * We do not want to do an nfs_statfs() here since we can get - * stuck on a dead server and we are holding a lock on the mount - * point. - */ - mtx_lock(&nmp->nm_mtx); - mp->mnt_stat.f_iosize = nfs_iosize(nmp); - mtx_unlock(&nmp->nm_mtx); - /* - * A reference count is needed on the nfsnode representing the - * remote root. If this object is not persistent, then backward - * traversals of the mount point (i.e. "..") will not work if - * the nfsnode gets flushed out of the cache. Ufs does not have - * this problem, because one can identify root inodes by their - * number == ROOTINO (2). - */ - error = nfs_nget(mp, (nfsfh_t *)nmp->nm_fh, nmp->nm_fhsize, &np, LK_EXCLUSIVE); - if (error) - goto bad; - *vpp = NFSTOV(np); - - /* - * Get file attributes and transfer parameters for the - * mountpoint. This has the side effect of filling in - * (*vpp)->v_type with the correct value. - */ - if (argp->flags & NFSMNT_NFSV3) - nfs_fsinfo(nmp, *vpp, curthread->td_ucred, curthread); - else - VOP_GETATTR(*vpp, &attrs, curthread->td_ucred); - - /* - * Lose the lock but keep the ref. - */ - VOP_UNLOCK(*vpp, 0); - - return (0); -bad: - nfs_disconnect(nmp); - mtx_destroy(&nmp->nm_mtx); - uma_zfree(nfsmount_zone, nmp); - free(nam, M_SONAME); - return (error); -} - -/* - * unmount system call - */ -static int -nfs_unmount(struct mount *mp, int mntflags) -{ - struct nfsmount *nmp; - int error, flags = 0, i; - - if (mntflags & MNT_FORCE) - flags |= FORCECLOSE; - nmp = VFSTONFS(mp); - /* - * Goes something like this.. - * - Call vflush() to clear out vnodes for this filesystem - * - Close the socket - * - Free up the data structures - */ - /* In the forced case, cancel any outstanding requests. */ - if (flags & FORCECLOSE) { - error = nfs_nmcancelreqs(nmp); - if (error) - goto out; - } - /* We hold 1 extra ref on the root vnode; see comment in mountnfs(). */ - error = vflush(mp, 1, flags, curthread); - if (error) - goto out; - - /* - * We are now committed to the unmount. - */ - /* Make sure no nfsiods are assigned to this mount. */ - mtx_lock(&nfs_iod_mtx); - for (i = 0; i < NFS_MAXASYNCDAEMON; i++) - if (nfs_iodmount[i] == nmp) { - nfs_iodwant[i] = NFSIOD_AVAILABLE; - nfs_iodmount[i] = NULL; - } - mtx_unlock(&nfs_iod_mtx); - nfs_disconnect(nmp); - free(nmp->nm_nam, M_SONAME); - - mtx_destroy(&nmp->nm_mtx); - uma_zfree(nfsmount_zone, nmp); -out: - return (error); -} - -/* - * Return root of a filesystem - */ -static int -nfs_root(struct mount *mp, int flags, struct vnode **vpp) -{ - struct vnode *vp; - struct nfsmount *nmp; - struct nfsnode *np; - int error; - - nmp = VFSTONFS(mp); - error = nfs_nget(mp, (nfsfh_t *)nmp->nm_fh, nmp->nm_fhsize, &np, flags); - if (error) - return error; - vp = NFSTOV(np); - /* - * Get transfer parameters and attributes for root vnode once. - */ - mtx_lock(&nmp->nm_mtx); - if ((nmp->nm_state & NFSSTA_GOTFSINFO) == 0 && - (nmp->nm_flag & NFSMNT_NFSV3)) { - mtx_unlock(&nmp->nm_mtx); - nfs_fsinfo(nmp, vp, curthread->td_ucred, curthread); - } else - mtx_unlock(&nmp->nm_mtx); - if (vp->v_type == VNON) - vp->v_type = VDIR; - vp->v_vflag |= VV_ROOT; - *vpp = vp; - return (0); -} - -/* - * Flush out the buffer cache - */ -/* ARGSUSED */ -static int -nfs_sync(struct mount *mp, int waitfor) -{ - struct vnode *vp, *mvp; - struct thread *td; - int error, allerror = 0; - - td = curthread; - - MNT_ILOCK(mp); - /* - * If a forced dismount is in progress, return from here so that - * the umount(2) syscall doesn't get stuck in VFS_SYNC() before - * calling VFS_UNMOUNT(). - */ - if ((mp->mnt_kern_flag & MNTK_UNMOUNTF) != 0) { - MNT_IUNLOCK(mp); - return (EBADF); - } - MNT_IUNLOCK(mp); - - /* - * Force stale buffer cache information to be flushed. - */ -loop: - MNT_VNODE_FOREACH_ALL(vp, mp, mvp) { - /* XXX Racy bv_cnt check. */ - if (VOP_ISLOCKED(vp) || vp->v_bufobj.bo_dirty.bv_cnt == 0 || - waitfor == MNT_LAZY) { - VI_UNLOCK(vp); - continue; - } - if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, td)) { - MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp); - goto loop; - } - error = VOP_FSYNC(vp, waitfor, td); - if (error) - allerror = error; - VOP_UNLOCK(vp, 0); - vrele(vp); - } - return (allerror); -} - -static int -nfs_sysctl(struct mount *mp, fsctlop_t op, struct sysctl_req *req) -{ - struct nfsmount *nmp = VFSTONFS(mp); - struct vfsquery vq; - int error; - - bzero(&vq, sizeof(vq)); - switch (op) { -#if 0 - case VFS_CTL_NOLOCKS: - val = (nmp->nm_flag & NFSMNT_NOLOCKS) ? 1 : 0; - if (req->oldptr != NULL) { - error = SYSCTL_OUT(req, &val, sizeof(val)); - if (error) - return (error); - } - if (req->newptr != NULL) { - error = SYSCTL_IN(req, &val, sizeof(val)); - if (error) - return (error); - if (val) - nmp->nm_flag |= NFSMNT_NOLOCKS; - else - nmp->nm_flag &= ~NFSMNT_NOLOCKS; - } - break; -#endif - case VFS_CTL_QUERY: - mtx_lock(&nmp->nm_mtx); - if (nmp->nm_state & NFSSTA_TIMEO) - vq.vq_flags |= VQ_NOTRESP; - mtx_unlock(&nmp->nm_mtx); -#if 0 - if (!(nmp->nm_flag & NFSMNT_NOLOCKS) && - (nmp->nm_state & NFSSTA_LOCKTIMEO)) - vq.vq_flags |= VQ_NOTRESPLOCK; -#endif - error = SYSCTL_OUT(req, &vq, sizeof(vq)); - break; - case VFS_CTL_TIMEO: - if (req->oldptr != NULL) { - error = SYSCTL_OUT(req, &nmp->nm_tprintf_initial_delay, - sizeof(nmp->nm_tprintf_initial_delay)); - if (error) - return (error); - } - if (req->newptr != NULL) { - error = vfs_suser(mp, req->td); - if (error) - return (error); - error = SYSCTL_IN(req, &nmp->nm_tprintf_initial_delay, - sizeof(nmp->nm_tprintf_initial_delay)); - if (error) - return (error); - if (nmp->nm_tprintf_initial_delay < 0) - nmp->nm_tprintf_initial_delay = 0; - } - break; - default: - return (ENOTSUP); - } - return (0); -} - -/* - * Extract the information needed by the nlm from the nfs vnode. - */ -static void -nfs_getnlminfo(struct vnode *vp, uint8_t *fhp, size_t *fhlenp, - struct sockaddr_storage *sp, int *is_v3p, off_t *sizep, - struct timeval *timeop) -{ - struct nfsmount *nmp; - struct nfsnode *np = VTONFS(vp); - - nmp = VFSTONFS(vp->v_mount); - if (fhlenp != NULL) - *fhlenp = (size_t)np->n_fhsize; - if (fhp != NULL) - bcopy(np->n_fhp, fhp, np->n_fhsize); - if (sp != NULL) - bcopy(nmp->nm_nam, sp, min(nmp->nm_nam->sa_len, sizeof(*sp))); - if (is_v3p != NULL) - *is_v3p = NFS_ISV3(vp); - if (sizep != NULL) - *sizep = np->n_size; - if (timeop != NULL) { - timeop->tv_sec = nmp->nm_timeo / NFS_HZ; - timeop->tv_usec = (nmp->nm_timeo % NFS_HZ) * (1000000 / NFS_HZ); - } -} - diff --git a/sys/nfsclient/nfs_vnops.c b/sys/nfsclient/nfs_vnops.c deleted file mode 100644 index 2516d7d..0000000 --- a/sys/nfsclient/nfs_vnops.c +++ /dev/null @@ -1,3544 +0,0 @@ -/*- - * Copyright (c) 1989, 1993 - * The Regents of the University of California. All rights reserved. - * - * This code is derived from software contributed to Berkeley by - * Rick Macklem at The University of Guelph. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 4. Neither the name of the University nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * @(#)nfs_vnops.c 8.16 (Berkeley) 5/27/95 - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -/* - * vnode op calls for Sun NFS version 2 and 3 - */ - -#include "opt_inet.h" - -#include <sys/param.h> -#include <sys/kernel.h> -#include <sys/systm.h> -#include <sys/resourcevar.h> -#include <sys/proc.h> -#include <sys/mount.h> -#include <sys/bio.h> -#include <sys/buf.h> -#include <sys/jail.h> -#include <sys/lock.h> -#include <sys/malloc.h> -#include <sys/mbuf.h> -#include <sys/namei.h> -#include <sys/socket.h> -#include <sys/vnode.h> -#include <sys/dirent.h> -#include <sys/fcntl.h> -#include <sys/lockf.h> -#include <sys/rwlock.h> -#include <sys/stat.h> -#include <sys/sysctl.h> -#include <sys/signalvar.h> - -#include <vm/vm.h> -#include <vm/vm_extern.h> -#include <vm/vm_object.h> - -#include <nfs/nfsproto.h> -#include <nfsclient/nfs.h> -#include <nfsclient/nfsnode.h> -#include <nfsclient/nfsmount.h> -#include <nfs/nfs_kdtrace.h> -#include <nfs/nfs_lock.h> -#include <nfs/xdr_subs.h> -#include <nfsclient/nfsm_subs.h> - -#include <net/if.h> -#include <net/if_var.h> -#include <net/vnet.h> -#include <netinet/in.h> -#include <netinet/in_var.h> - -#include <machine/stdarg.h> - -#ifdef KDTRACE_HOOKS -#include <sys/dtrace_bsd.h> - -dtrace_nfsclient_accesscache_flush_probe_func_t - dtrace_nfsclient_accesscache_flush_done_probe; -uint32_t nfsclient_accesscache_flush_done_id; - -dtrace_nfsclient_accesscache_get_probe_func_t - dtrace_nfsclient_accesscache_get_hit_probe, - dtrace_nfsclient_accesscache_get_miss_probe; -uint32_t nfsclient_accesscache_get_hit_id; -uint32_t nfsclient_accesscache_get_miss_id; - -dtrace_nfsclient_accesscache_load_probe_func_t - dtrace_nfsclient_accesscache_load_done_probe; -uint32_t nfsclient_accesscache_load_done_id; -#endif /* !KDTRACE_HOOKS */ - -/* Defs */ -#define TRUE 1 -#define FALSE 0 - -/* - * Ifdef for FreeBSD-current merged buffer cache. It is unfortunate that these - * calls are not in getblk() and brelse() so that they would not be necessary - * here. - */ -#ifndef B_VMIO -#define vfs_busy_pages(bp, f) -#endif - -static vop_read_t nfsfifo_read; -static vop_write_t nfsfifo_write; -static vop_close_t nfsfifo_close; -static int nfs_flush(struct vnode *, int, int); -static int nfs_setattrrpc(struct vnode *, struct vattr *, struct ucred *); -static vop_lookup_t nfs_lookup; -static vop_create_t nfs_create; -static vop_mknod_t nfs_mknod; -static vop_open_t nfs_open; -static vop_close_t nfs_close; -static vop_access_t nfs_access; -static vop_getattr_t nfs_getattr; -static vop_setattr_t nfs_setattr; -static vop_read_t nfs_read; -static vop_fsync_t nfs_fsync; -static vop_remove_t nfs_remove; -static vop_link_t nfs_link; -static vop_rename_t nfs_rename; -static vop_mkdir_t nfs_mkdir; -static vop_rmdir_t nfs_rmdir; -static vop_symlink_t nfs_symlink; -static vop_readdir_t nfs_readdir; -static vop_strategy_t nfs_strategy; -static int nfs_lookitup(struct vnode *, const char *, int, - struct ucred *, struct thread *, struct nfsnode **); -static int nfs_sillyrename(struct vnode *, struct vnode *, - struct componentname *); -static vop_access_t nfsspec_access; -static vop_readlink_t nfs_readlink; -static vop_print_t nfs_print; -static vop_advlock_t nfs_advlock; -static vop_advlockasync_t nfs_advlockasync; - -/* - * Global vfs data structures for nfs - */ -struct vop_vector nfs_vnodeops = { - .vop_default = &default_vnodeops, - .vop_access = nfs_access, - .vop_advlock = nfs_advlock, - .vop_advlockasync = nfs_advlockasync, - .vop_close = nfs_close, - .vop_create = nfs_create, - .vop_fsync = nfs_fsync, - .vop_getattr = nfs_getattr, - .vop_getpages = nfs_getpages, - .vop_putpages = nfs_putpages, - .vop_inactive = nfs_inactive, - .vop_link = nfs_link, - .vop_lookup = nfs_lookup, - .vop_mkdir = nfs_mkdir, - .vop_mknod = nfs_mknod, - .vop_open = nfs_open, - .vop_print = nfs_print, - .vop_read = nfs_read, - .vop_readdir = nfs_readdir, - .vop_readlink = nfs_readlink, - .vop_reclaim = nfs_reclaim, - .vop_remove = nfs_remove, - .vop_rename = nfs_rename, - .vop_rmdir = nfs_rmdir, - .vop_setattr = nfs_setattr, - .vop_strategy = nfs_strategy, - .vop_symlink = nfs_symlink, - .vop_write = nfs_write, -}; - -struct vop_vector nfs_fifoops = { - .vop_default = &fifo_specops, - .vop_access = nfsspec_access, - .vop_close = nfsfifo_close, - .vop_fsync = nfs_fsync, - .vop_getattr = nfs_getattr, - .vop_inactive = nfs_inactive, - .vop_print = nfs_print, - .vop_read = nfsfifo_read, - .vop_reclaim = nfs_reclaim, - .vop_setattr = nfs_setattr, - .vop_write = nfsfifo_write, -}; - -static int nfs_mknodrpc(struct vnode *dvp, struct vnode **vpp, - struct componentname *cnp, struct vattr *vap); -static int nfs_removerpc(struct vnode *dvp, const char *name, int namelen, - struct ucred *cred, struct thread *td); -static int nfs_renamerpc(struct vnode *fdvp, const char *fnameptr, - int fnamelen, struct vnode *tdvp, - const char *tnameptr, int tnamelen, - struct ucred *cred, struct thread *td); -static int nfs_renameit(struct vnode *sdvp, struct componentname *scnp, - struct sillyrename *sp); - -/* - * Global variables - */ -struct mtx nfs_iod_mtx; -enum nfsiod_state nfs_iodwant[NFS_MAXASYNCDAEMON]; -struct nfsmount *nfs_iodmount[NFS_MAXASYNCDAEMON]; -int nfs_numasync = 0; -#define DIRHDSIZ (sizeof (struct dirent) - (MAXNAMLEN + 1)) - -SYSCTL_DECL(_vfs_oldnfs); - -static int nfsaccess_cache_timeout = NFS_MAXATTRTIMO; -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, access_cache_timeout, CTLFLAG_RW, - &nfsaccess_cache_timeout, 0, "NFS ACCESS cache timeout"); - -static int nfs_prime_access_cache = 0; -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, prime_access_cache, CTLFLAG_RW, - &nfs_prime_access_cache, 0, - "Prime NFS ACCESS cache when fetching attributes"); - -static int nfsv3_commit_on_close = 0; -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, nfsv3_commit_on_close, CTLFLAG_RW, - &nfsv3_commit_on_close, 0, "write+commit on close, else only write"); - -static int nfs_clean_pages_on_close = 1; -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, clean_pages_on_close, CTLFLAG_RW, - &nfs_clean_pages_on_close, 0, "NFS clean dirty pages on close"); - -int nfs_directio_enable = 0; -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, nfs_directio_enable, CTLFLAG_RW, - &nfs_directio_enable, 0, "Enable NFS directio"); - -/* - * This sysctl allows other processes to mmap a file that has been opened - * O_DIRECT by a process. In general, having processes mmap the file while - * Direct IO is in progress can lead to Data Inconsistencies. But, we allow - * this by default to prevent DoS attacks - to prevent a malicious user from - * opening up files O_DIRECT preventing other users from mmap'ing these - * files. "Protected" environments where stricter consistency guarantees are - * required can disable this knob. The process that opened the file O_DIRECT - * cannot mmap() the file, because mmap'ed IO on an O_DIRECT open() is not - * meaningful. - */ -int nfs_directio_allow_mmap = 1; -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, nfs_directio_allow_mmap, CTLFLAG_RW, - &nfs_directio_allow_mmap, 0, "Enable mmaped IO on file with O_DIRECT opens"); - -#if 0 -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, access_cache_hits, CTLFLAG_RD, - &nfsstats.accesscache_hits, 0, "NFS ACCESS cache hit count"); - -SYSCTL_INT(_vfs_oldnfs, OID_AUTO, access_cache_misses, CTLFLAG_RD, - &nfsstats.accesscache_misses, 0, "NFS ACCESS cache miss count"); -#endif - -#define NFSV3ACCESS_ALL (NFSV3ACCESS_READ | NFSV3ACCESS_MODIFY \ - | NFSV3ACCESS_EXTEND | NFSV3ACCESS_EXECUTE \ - | NFSV3ACCESS_DELETE | NFSV3ACCESS_LOOKUP) - -/* - * SMP Locking Note : - * The list of locks after the description of the lock is the ordering - * of other locks acquired with the lock held. - * np->n_mtx : Protects the fields in the nfsnode. - VM Object Lock - VI_MTX (acquired indirectly) - * nmp->nm_mtx : Protects the fields in the nfsmount. - rep->r_mtx - * nfs_iod_mtx : Global lock, protects shared nfsiod state. - * nfs_reqq_mtx : Global lock, protects the nfs_reqq list. - nmp->nm_mtx - rep->r_mtx - * rep->r_mtx : Protects the fields in an nfsreq. - */ - -static int -nfs3_access_otw(struct vnode *vp, int wmode, struct thread *td, - struct ucred *cred, uint32_t *retmode) -{ - const int v3 = 1; - u_int32_t *tl; - int error = 0, attrflag, i, lrupos; - - struct mbuf *mreq, *mrep, *md, *mb; - caddr_t bpos, dpos; - u_int32_t rmode; - struct nfsnode *np = VTONFS(vp); - - nfsstats.rpccnt[NFSPROC_ACCESS]++; - mreq = m_get2(NFSX_FH(v3) + NFSX_UNSIGNED, M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(vp, v3); - tl = nfsm_build(u_int32_t *, NFSX_UNSIGNED); - *tl = txdr_unsigned(wmode); - nfsm_request(vp, NFSPROC_ACCESS, td, cred); - nfsm_postop_attr(vp, attrflag); - if (!error) { - lrupos = 0; - tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED); - rmode = fxdr_unsigned(u_int32_t, *tl); - mtx_lock(&np->n_mtx); - for (i = 0; i < NFS_ACCESSCACHESIZE; i++) { - if (np->n_accesscache[i].uid == cred->cr_uid) { - np->n_accesscache[i].mode = rmode; - np->n_accesscache[i].stamp = time_second; - break; - } - if (i > 0 && np->n_accesscache[i].stamp < - np->n_accesscache[lrupos].stamp) - lrupos = i; - } - if (i == NFS_ACCESSCACHESIZE) { - np->n_accesscache[lrupos].uid = cred->cr_uid; - np->n_accesscache[lrupos].mode = rmode; - np->n_accesscache[lrupos].stamp = time_second; - } - mtx_unlock(&np->n_mtx); - if (retmode != NULL) - *retmode = rmode; - KDTRACE_NFS_ACCESSCACHE_LOAD_DONE(vp, cred->cr_uid, rmode, 0); - } - m_freem(mrep); -nfsmout: -#ifdef KDTRACE_HOOKS - if (error) { - KDTRACE_NFS_ACCESSCACHE_LOAD_DONE(vp, cred->cr_uid, 0, - error); - } -#endif - return (error); -} - -/* - * nfs access vnode op. - * For nfs version 2, just return ok. File accesses may fail later. - * For nfs version 3, use the access rpc to check accessibility. If file modes - * are changed on the server, accesses might still fail later. - */ -static int -nfs_access(struct vop_access_args *ap) -{ - struct vnode *vp = ap->a_vp; - int error = 0, i, gotahit; - u_int32_t mode, rmode, wmode; - int v3 = NFS_ISV3(vp); - struct nfsnode *np = VTONFS(vp); - - /* - * Disallow write attempts on filesystems mounted read-only; - * unless the file is a socket, fifo, or a block or character - * device resident on the filesystem. - */ - if ((ap->a_accmode & VWRITE) && (vp->v_mount->mnt_flag & MNT_RDONLY)) { - switch (vp->v_type) { - case VREG: - case VDIR: - case VLNK: - return (EROFS); - default: - break; - } - } - /* - * For nfs v3, check to see if we have done this recently, and if - * so return our cached result instead of making an ACCESS call. - * If not, do an access rpc, otherwise you are stuck emulating - * ufs_access() locally using the vattr. This may not be correct, - * since the server may apply other access criteria such as - * client uid-->server uid mapping that we do not know about. - */ - if (v3) { - if (ap->a_accmode & VREAD) - mode = NFSV3ACCESS_READ; - else - mode = 0; - if (vp->v_type != VDIR) { - if (ap->a_accmode & VWRITE) - mode |= (NFSV3ACCESS_MODIFY | NFSV3ACCESS_EXTEND); - if (ap->a_accmode & VEXEC) - mode |= NFSV3ACCESS_EXECUTE; - } else { - if (ap->a_accmode & VWRITE) - mode |= (NFSV3ACCESS_MODIFY | NFSV3ACCESS_EXTEND | - NFSV3ACCESS_DELETE); - if (ap->a_accmode & VEXEC) - mode |= NFSV3ACCESS_LOOKUP; - } - /* XXX safety belt, only make blanket request if caching */ - if (nfsaccess_cache_timeout > 0) { - wmode = NFSV3ACCESS_READ | NFSV3ACCESS_MODIFY | - NFSV3ACCESS_EXTEND | NFSV3ACCESS_EXECUTE | - NFSV3ACCESS_DELETE | NFSV3ACCESS_LOOKUP; - } else { - wmode = mode; - } - - /* - * Does our cached result allow us to give a definite yes to - * this request? - */ - gotahit = 0; - mtx_lock(&np->n_mtx); - for (i = 0; i < NFS_ACCESSCACHESIZE; i++) { - if (ap->a_cred->cr_uid == np->n_accesscache[i].uid) { - if (time_second < (np->n_accesscache[i].stamp + - nfsaccess_cache_timeout) && - (np->n_accesscache[i].mode & mode) == mode) { - nfsstats.accesscache_hits++; - gotahit = 1; - } - break; - } - } - mtx_unlock(&np->n_mtx); -#ifdef KDTRACE_HOOKS - if (gotahit) - KDTRACE_NFS_ACCESSCACHE_GET_HIT(vp, - ap->a_cred->cr_uid, mode); - else - KDTRACE_NFS_ACCESSCACHE_GET_MISS(vp, - ap->a_cred->cr_uid, mode); -#endif - if (gotahit == 0) { - /* - * Either a no, or a don't know. Go to the wire. - */ - nfsstats.accesscache_misses++; - error = nfs3_access_otw(vp, wmode, ap->a_td, ap->a_cred, - &rmode); - if (!error) { - if ((rmode & mode) != mode) - error = EACCES; - } - } - return (error); - } else { - if ((error = nfsspec_access(ap)) != 0) { - return (error); - } - /* - * Attempt to prevent a mapped root from accessing a file - * which it shouldn't. We try to read a byte from the file - * if the user is root and the file is not zero length. - * After calling nfsspec_access, we should have the correct - * file size cached. - */ - mtx_lock(&np->n_mtx); - if (ap->a_cred->cr_uid == 0 && (ap->a_accmode & VREAD) - && VTONFS(vp)->n_size > 0) { - struct iovec aiov; - struct uio auio; - char buf[1]; - - mtx_unlock(&np->n_mtx); - aiov.iov_base = buf; - aiov.iov_len = 1; - auio.uio_iov = &aiov; - auio.uio_iovcnt = 1; - auio.uio_offset = 0; - auio.uio_resid = 1; - auio.uio_segflg = UIO_SYSSPACE; - auio.uio_rw = UIO_READ; - auio.uio_td = ap->a_td; - - if (vp->v_type == VREG) - error = nfs_readrpc(vp, &auio, ap->a_cred); - else if (vp->v_type == VDIR) { - char* bp; - bp = malloc(NFS_DIRBLKSIZ, M_TEMP, M_WAITOK); - aiov.iov_base = bp; - aiov.iov_len = auio.uio_resid = NFS_DIRBLKSIZ; - error = nfs_readdirrpc(vp, &auio, ap->a_cred); - free(bp, M_TEMP); - } else if (vp->v_type == VLNK) - error = nfs_readlinkrpc(vp, &auio, ap->a_cred); - else - error = EACCES; - } else - mtx_unlock(&np->n_mtx); - return (error); - } -} - -int nfs_otw_getattr_avoid = 0; - -/* - * nfs open vnode op - * Check to see if the type is ok - * and that deletion is not in progress. - * For paged in text files, you will need to flush the page cache - * if consistency is lost. - */ -/* ARGSUSED */ -static int -nfs_open(struct vop_open_args *ap) -{ - struct vnode *vp = ap->a_vp; - struct nfsnode *np = VTONFS(vp); - struct vattr vattr; - int error; - int fmode = ap->a_mode; - struct ucred *cred; - - if (vp->v_type != VREG && vp->v_type != VDIR && vp->v_type != VLNK) - return (EOPNOTSUPP); - - /* - * Get a valid lease. If cached data is stale, flush it. - */ - mtx_lock(&np->n_mtx); - if (np->n_flag & NMODIFIED) { - mtx_unlock(&np->n_mtx); - error = nfs_vinvalbuf(vp, V_SAVE, ap->a_td, 1); - if (error == EINTR || error == EIO) - return (error); - mtx_lock(&np->n_mtx); - np->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp); - if (vp->v_type == VDIR) - np->n_direofoffset = 0; - mtx_unlock(&np->n_mtx); - error = VOP_GETATTR(vp, &vattr, ap->a_cred); - if (error) - return (error); - mtx_lock(&np->n_mtx); - np->n_mtime = vattr.va_mtime; - } else { - mtx_unlock(&np->n_mtx); - error = VOP_GETATTR(vp, &vattr, ap->a_cred); - if (error) - return (error); - mtx_lock(&np->n_mtx); - if (NFS_TIMESPEC_COMPARE(&np->n_mtime, &vattr.va_mtime)) { - if (vp->v_type == VDIR) - np->n_direofoffset = 0; - mtx_unlock(&np->n_mtx); - error = nfs_vinvalbuf(vp, V_SAVE, ap->a_td, 1); - if (error == EINTR || error == EIO) { - return (error); - } - mtx_lock(&np->n_mtx); - np->n_mtime = vattr.va_mtime; - } - } - /* - * If the object has >= 1 O_DIRECT active opens, we disable caching. - */ - if (nfs_directio_enable && (fmode & O_DIRECT) && (vp->v_type == VREG)) { - if (np->n_directio_opens == 0) { - mtx_unlock(&np->n_mtx); - error = nfs_vinvalbuf(vp, V_SAVE, ap->a_td, 1); - if (error) - return (error); - mtx_lock(&np->n_mtx); - np->n_flag |= NNONCACHE; - } - np->n_directio_opens++; - } - - /* - * If this is an open for writing, capture a reference to the - * credentials, so they can be used by nfs_putpages(). Using - * these write credentials is preferable to the credentials of - * whatever thread happens to be doing the VOP_PUTPAGES() since - * the write RPCs are less likely to fail with EACCES. - */ - if ((fmode & FWRITE) != 0) { - cred = np->n_writecred; - np->n_writecred = crhold(ap->a_cred); - } else - cred = NULL; - mtx_unlock(&np->n_mtx); - if (cred != NULL) - crfree(cred); - vnode_create_vobject(vp, vattr.va_size, ap->a_td); - return (0); -} - -/* - * nfs close vnode op - * What an NFS client should do upon close after writing is a debatable issue. - * Most NFS clients push delayed writes to the server upon close, basically for - * two reasons: - * 1 - So that any write errors may be reported back to the client process - * doing the close system call. By far the two most likely errors are - * NFSERR_NOSPC and NFSERR_DQUOT to indicate space allocation failure. - * 2 - To put a worst case upper bound on cache inconsistency between - * multiple clients for the file. - * There is also a consistency problem for Version 2 of the protocol w.r.t. - * not being able to tell if other clients are writing a file concurrently, - * since there is no way of knowing if the changed modify time in the reply - * is only due to the write for this client. - * (NFS Version 3 provides weak cache consistency data in the reply that - * should be sufficient to detect and handle this case.) - * - * The current code does the following: - * for NFS Version 2 - play it safe and flush/invalidate all dirty buffers - * for NFS Version 3 - flush dirty buffers to the server but don't invalidate - * or commit them (this satisfies 1 and 2 except for the - * case where the server crashes after this close but - * before the commit RPC, which is felt to be "good - * enough". Changing the last argument to nfs_flush() to - * a 1 would force a commit operation, if it is felt a - * commit is necessary now. - */ -/* ARGSUSED */ -static int -nfs_close(struct vop_close_args *ap) -{ - struct vnode *vp = ap->a_vp; - struct nfsnode *np = VTONFS(vp); - int error = 0; - int fmode = ap->a_fflag; - - if (vp->v_type == VREG) { - /* - * Examine and clean dirty pages, regardless of NMODIFIED. - * This closes a major hole in close-to-open consistency. - * We want to push out all dirty pages (and buffers) on - * close, regardless of whether they were dirtied by - * mmap'ed writes or via write(). - */ - if (nfs_clean_pages_on_close && vp->v_object) { - VM_OBJECT_WLOCK(vp->v_object); - vm_object_page_clean(vp->v_object, 0, 0, 0); - VM_OBJECT_WUNLOCK(vp->v_object); - } - mtx_lock(&np->n_mtx); - if (np->n_flag & NMODIFIED) { - mtx_unlock(&np->n_mtx); - if (NFS_ISV3(vp)) { - /* - * Under NFSv3 we have dirty buffers to dispose of. We - * must flush them to the NFS server. We have the option - * of waiting all the way through the commit rpc or just - * waiting for the initial write. The default is to only - * wait through the initial write so the data is in the - * server's cache, which is roughly similar to the state - * a standard disk subsystem leaves the file in on close(). - * - * We cannot clear the NMODIFIED bit in np->n_flag due to - * potential races with other processes, and certainly - * cannot clear it if we don't commit. - */ - int cm = nfsv3_commit_on_close ? 1 : 0; - error = nfs_flush(vp, MNT_WAIT, cm); - /* np->n_flag &= ~NMODIFIED; */ - } else - error = nfs_vinvalbuf(vp, V_SAVE, ap->a_td, 1); - mtx_lock(&np->n_mtx); - } - if (np->n_flag & NWRITEERR) { - np->n_flag &= ~NWRITEERR; - error = np->n_error; - } - mtx_unlock(&np->n_mtx); - } - if (nfs_directio_enable) - KASSERT((np->n_directio_asyncwr == 0), - ("nfs_close: dirty unflushed (%d) directio buffers\n", - np->n_directio_asyncwr)); - if (nfs_directio_enable && (fmode & O_DIRECT) && (vp->v_type == VREG)) { - mtx_lock(&np->n_mtx); - KASSERT((np->n_directio_opens > 0), - ("nfs_close: unexpectedly value (0) of n_directio_opens\n")); - np->n_directio_opens--; - if (np->n_directio_opens == 0) - np->n_flag &= ~NNONCACHE; - mtx_unlock(&np->n_mtx); - } - return (error); -} - -/* - * nfs getattr call from vfs. - */ -static int -nfs_getattr(struct vop_getattr_args *ap) -{ - struct vnode *vp = ap->a_vp; - struct nfsnode *np = VTONFS(vp); - struct thread *td = curthread; - struct vattr *vap = ap->a_vap; - struct vattr vattr; - caddr_t bpos, dpos; - int error = 0; - struct mbuf *mreq, *mrep, *md, *mb; - int v3 = NFS_ISV3(vp); - - /* - * Update local times for special files. - */ - mtx_lock(&np->n_mtx); - if (np->n_flag & (NACC | NUPD)) - np->n_flag |= NCHG; - mtx_unlock(&np->n_mtx); - /* - * First look in the cache. - */ - if (nfs_getattrcache(vp, &vattr) == 0) - goto nfsmout; - if (v3 && nfs_prime_access_cache && nfsaccess_cache_timeout > 0) { - nfsstats.accesscache_misses++; - nfs3_access_otw(vp, NFSV3ACCESS_ALL, td, ap->a_cred, NULL); - if (nfs_getattrcache(vp, &vattr) == 0) - goto nfsmout; - } - nfsstats.rpccnt[NFSPROC_GETATTR]++; - mreq = m_get2(NFSX_FH(v3), M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(vp, v3); - nfsm_request(vp, NFSPROC_GETATTR, td, ap->a_cred); - if (!error) { - nfsm_loadattr(vp, &vattr); - } - m_freem(mrep); -nfsmout: - vap->va_type = vattr.va_type; - vap->va_mode = vattr.va_mode; - vap->va_nlink = vattr.va_nlink; - vap->va_uid = vattr.va_uid; - vap->va_gid = vattr.va_gid; - vap->va_fsid = vattr.va_fsid; - vap->va_fileid = vattr.va_fileid; - vap->va_size = vattr.va_size; - vap->va_blocksize = vattr.va_blocksize; - vap->va_atime = vattr.va_atime; - vap->va_mtime = vattr.va_mtime; - vap->va_ctime = vattr.va_ctime; - vap->va_gen = vattr.va_gen; - vap->va_flags = vattr.va_flags; - vap->va_rdev = vattr.va_rdev; - vap->va_bytes = vattr.va_bytes; - vap->va_filerev = vattr.va_filerev; - - return (error); -} - -/* - * nfs setattr call. - */ -static int -nfs_setattr(struct vop_setattr_args *ap) -{ - struct vnode *vp = ap->a_vp; - struct nfsnode *np = VTONFS(vp); - struct vattr *vap = ap->a_vap; - struct thread *td = curthread; - int error = 0; - u_quad_t tsize; - -#ifndef nolint - tsize = (u_quad_t)0; -#endif - - /* - * Setting of flags is not supported. - */ - if (vap->va_flags != VNOVAL) - return (EOPNOTSUPP); - - /* - * Disallow write attempts if the filesystem is mounted read-only. - */ - if ((vap->va_flags != VNOVAL || vap->va_uid != (uid_t)VNOVAL || - vap->va_gid != (gid_t)VNOVAL || vap->va_atime.tv_sec != VNOVAL || - vap->va_mtime.tv_sec != VNOVAL || vap->va_mode != (mode_t)VNOVAL) && - (vp->v_mount->mnt_flag & MNT_RDONLY)) { - error = EROFS; - goto out; - } - if (vap->va_size != VNOVAL) { - switch (vp->v_type) { - case VDIR: - return (EISDIR); - case VCHR: - case VBLK: - case VSOCK: - case VFIFO: - if (vap->va_mtime.tv_sec == VNOVAL && - vap->va_atime.tv_sec == VNOVAL && - vap->va_mode == (mode_t)VNOVAL && - vap->va_uid == (uid_t)VNOVAL && - vap->va_gid == (gid_t)VNOVAL) - return (0); - vap->va_size = VNOVAL; - break; - default: - /* - * Disallow write attempts if the filesystem is - * mounted read-only. - */ - if (vp->v_mount->mnt_flag & MNT_RDONLY) - return (EROFS); - /* - * We run vnode_pager_setsize() early (why?), - * we must set np->n_size now to avoid vinvalbuf - * V_SAVE races that might setsize a lower - * value. - */ - mtx_lock(&np->n_mtx); - tsize = np->n_size; - mtx_unlock(&np->n_mtx); - error = nfs_meta_setsize(vp, ap->a_cred, td, - vap->va_size); - mtx_lock(&np->n_mtx); - if (np->n_flag & NMODIFIED) { - tsize = np->n_size; - mtx_unlock(&np->n_mtx); - if (vap->va_size == 0) - error = nfs_vinvalbuf(vp, 0, td, 1); - else - error = nfs_vinvalbuf(vp, V_SAVE, td, 1); - if (error) { - vnode_pager_setsize(vp, tsize); - goto out; - } - } else - mtx_unlock(&np->n_mtx); - /* - * np->n_size has already been set to vap->va_size - * in nfs_meta_setsize(). We must set it again since - * nfs_loadattrcache() could be called through - * nfs_meta_setsize() and could modify np->n_size. - */ - mtx_lock(&np->n_mtx); - np->n_vattr.va_size = np->n_size = vap->va_size; - mtx_unlock(&np->n_mtx); - }; - } else { - mtx_lock(&np->n_mtx); - if ((vap->va_mtime.tv_sec != VNOVAL || vap->va_atime.tv_sec != VNOVAL) && - (np->n_flag & NMODIFIED) && vp->v_type == VREG) { - mtx_unlock(&np->n_mtx); - if ((error = nfs_vinvalbuf(vp, V_SAVE, td, 1)) != 0 && - (error == EINTR || error == EIO)) - return error; - } else - mtx_unlock(&np->n_mtx); - } - error = nfs_setattrrpc(vp, vap, ap->a_cred); - if (error && vap->va_size != VNOVAL) { - mtx_lock(&np->n_mtx); - np->n_size = np->n_vattr.va_size = tsize; - vnode_pager_setsize(vp, tsize); - mtx_unlock(&np->n_mtx); - } -out: - return (error); -} - -/* - * Do an nfs setattr rpc. - */ -static int -nfs_setattrrpc(struct vnode *vp, struct vattr *vap, struct ucred *cred) -{ - struct nfsv2_sattr *sp; - struct nfsnode *np = VTONFS(vp); - caddr_t bpos, dpos; - u_int32_t *tl; - int error = 0, i, wccflag = NFSV3_WCCRATTR; - struct mbuf *mreq, *mrep, *md, *mb; - int v3 = NFS_ISV3(vp); - - nfsstats.rpccnt[NFSPROC_SETATTR]++; - mreq = m_get2(NFSX_FH(v3) + NFSX_SATTR(v3), M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(vp, v3); - if (v3) { - nfsm_v3attrbuild(vap, TRUE); - tl = nfsm_build(u_int32_t *, NFSX_UNSIGNED); - *tl = nfs_false; - } else { - sp = nfsm_build(struct nfsv2_sattr *, NFSX_V2SATTR); - if (vap->va_mode == (mode_t)VNOVAL) - sp->sa_mode = nfs_xdrneg1; - else - sp->sa_mode = vtonfsv2_mode(vp->v_type, vap->va_mode); - if (vap->va_uid == (uid_t)VNOVAL) - sp->sa_uid = nfs_xdrneg1; - else - sp->sa_uid = txdr_unsigned(vap->va_uid); - if (vap->va_gid == (gid_t)VNOVAL) - sp->sa_gid = nfs_xdrneg1; - else - sp->sa_gid = txdr_unsigned(vap->va_gid); - sp->sa_size = txdr_unsigned(vap->va_size); - txdr_nfsv2time(&vap->va_atime, &sp->sa_atime); - txdr_nfsv2time(&vap->va_mtime, &sp->sa_mtime); - } - nfsm_request(vp, NFSPROC_SETATTR, curthread, cred); - if (v3) { - mtx_lock(&np->n_mtx); - for (i = 0; i < NFS_ACCESSCACHESIZE; i++) - np->n_accesscache[i].stamp = 0; - mtx_unlock(&np->n_mtx); - KDTRACE_NFS_ACCESSCACHE_FLUSH_DONE(vp); - nfsm_wcc_data(vp, wccflag); - } else - nfsm_loadattr(vp, NULL); - m_freem(mrep); -nfsmout: - return (error); -} - -/* - * nfs lookup call, one step at a time... - * First look in cache - * If not found, unlock the directory nfsnode and do the rpc - */ -static int -nfs_lookup(struct vop_lookup_args *ap) -{ - struct componentname *cnp = ap->a_cnp; - struct vnode *dvp = ap->a_dvp; - struct vnode **vpp = ap->a_vpp; - struct mount *mp = dvp->v_mount; - struct vattr dvattr, vattr; - struct timespec nctime; - int flags = cnp->cn_flags; - struct vnode *newvp; - struct nfsmount *nmp; - caddr_t bpos, dpos; - struct mbuf *mreq, *mrep, *md, *mb; - long len; - nfsfh_t *fhp; - struct nfsnode *np, *newnp; - int error = 0, attrflag, dattrflag, fhsize, ltype, ncticks; - int v3 = NFS_ISV3(dvp); - struct thread *td = cnp->cn_thread; - - *vpp = NULLVP; - if ((flags & ISLASTCN) && (mp->mnt_flag & MNT_RDONLY) && - (cnp->cn_nameiop == DELETE || cnp->cn_nameiop == RENAME)) - return (EROFS); - if (dvp->v_type != VDIR) - return (ENOTDIR); - nmp = VFSTONFS(mp); - np = VTONFS(dvp); - if ((error = VOP_ACCESS(dvp, VEXEC, cnp->cn_cred, td)) != 0) { - *vpp = NULLVP; - return (error); - } - error = cache_lookup(dvp, vpp, cnp, &nctime, &ncticks); - if (error > 0 && error != ENOENT) - return (error); - if (error == -1) { - /* - * Lookups of "." are special and always return the - * current directory. cache_lookup() already handles - * associated locking bookkeeping, etc. - */ - if (cnp->cn_namelen == 1 && cnp->cn_nameptr[0] == '.') { - /* XXX: Is this really correct? */ - if (cnp->cn_nameiop != LOOKUP && - (flags & ISLASTCN)) - cnp->cn_flags |= SAVENAME; - return (0); - } - - /* - * We only accept a positive hit in the cache if the - * change time of the file matches our cached copy. - * Otherwise, we discard the cache entry and fallback - * to doing a lookup RPC. We also only trust cache - * entries for less than nm_nametimeo seconds. - * - * To better handle stale file handles and attributes, - * clear the attribute cache of this node if it is a - * leaf component, part of an open() call, and not - * locally modified before fetching the attributes. - * This should allow stale file handles to be detected - * here where we can fall back to a LOOKUP RPC to - * recover rather than having nfs_open() detect the - * stale file handle and failing open(2) with ESTALE. - */ - newvp = *vpp; - newnp = VTONFS(newvp); - if (!(nmp->nm_flag & NFSMNT_NOCTO) && - (flags & (ISLASTCN | ISOPEN)) == (ISLASTCN | ISOPEN) && - !(newnp->n_flag & NMODIFIED)) { - mtx_lock(&newnp->n_mtx); - newnp->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(newvp); - mtx_unlock(&newnp->n_mtx); - } - if ((u_int)(ticks - ncticks) < (nmp->nm_nametimeo * hz) && - VOP_GETATTR(newvp, &vattr, cnp->cn_cred) == 0 && - timespeccmp(&vattr.va_ctime, &nctime, ==)) { - nfsstats.lookupcache_hits++; - if (cnp->cn_nameiop != LOOKUP && - (flags & ISLASTCN)) - cnp->cn_flags |= SAVENAME; - return (0); - } - cache_purge(newvp); - if (dvp != newvp) - vput(newvp); - else - vrele(newvp); - *vpp = NULLVP; - } else if (error == ENOENT) { - if (dvp->v_iflag & VI_DOOMED) - return (ENOENT); - /* - * We only accept a negative hit in the cache if the - * modification time of the parent directory matches - * the cached copy in the name cache entry. - * Otherwise, we discard all of the negative cache - * entries for this directory. We also only trust - * negative cache entries for up to nm_negnametimeo - * seconds. - */ - if ((u_int)(ticks - ncticks) < (nmp->nm_negnametimeo * hz) && - VOP_GETATTR(dvp, &vattr, cnp->cn_cred) == 0 && - timespeccmp(&vattr.va_mtime, &nctime, ==)) { - nfsstats.lookupcache_hits++; - return (ENOENT); - } - cache_purge_negative(dvp); - } - - attrflag = dattrflag = 0; - error = 0; - newvp = NULLVP; - nfsstats.lookupcache_misses++; - nfsstats.rpccnt[NFSPROC_LOOKUP]++; - len = cnp->cn_namelen; - mreq = m_get2(NFSX_FH(v3) + NFSX_UNSIGNED + nfsm_rndup(len), M_WAITOK, - MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(dvp, v3); - nfsm_strtom(cnp->cn_nameptr, len, NFS_MAXNAMLEN); - nfsm_request(dvp, NFSPROC_LOOKUP, cnp->cn_thread, cnp->cn_cred); - if (error) { - if (v3) { - nfsm_postop_attr_va(dvp, dattrflag, &vattr); - m_freem(mrep); - } - goto nfsmout; - } - nfsm_getfh(fhp, fhsize, v3); - - /* - * Handle RENAME case... - */ - if (cnp->cn_nameiop == RENAME && (flags & ISLASTCN)) { - if (NFS_CMPFH(np, fhp, fhsize)) { - m_freem(mrep); - return (EISDIR); - } - error = nfs_nget(mp, fhp, fhsize, &np, LK_EXCLUSIVE); - if (error) { - m_freem(mrep); - return (error); - } - newvp = NFSTOV(np); - if (v3) { - nfsm_postop_attr(newvp, attrflag); - nfsm_postop_attr(dvp, attrflag); - } else - nfsm_loadattr(newvp, NULL); - *vpp = newvp; - m_freem(mrep); - cnp->cn_flags |= SAVENAME; - return (0); - } - - if (flags & ISDOTDOT) { - ltype = VOP_ISLOCKED(dvp); - error = vfs_busy(mp, MBF_NOWAIT); - if (error != 0) { - vfs_ref(mp); - VOP_UNLOCK(dvp, 0); - error = vfs_busy(mp, 0); - vn_lock(dvp, ltype | LK_RETRY); - vfs_rel(mp); - if (error == 0 && (dvp->v_iflag & VI_DOOMED)) { - vfs_unbusy(mp); - error = ENOENT; - } - if (error != 0) { - m_freem(mrep); - return (error); - } - } - VOP_UNLOCK(dvp, 0); - error = nfs_nget(mp, fhp, fhsize, &np, cnp->cn_lkflags); - if (error == 0) - newvp = NFSTOV(np); - vfs_unbusy(mp); - if (newvp != dvp) - vn_lock(dvp, ltype | LK_RETRY); - if (dvp->v_iflag & VI_DOOMED) { - if (error == 0) { - if (newvp == dvp) - vrele(newvp); - else - vput(newvp); - } - error = ENOENT; - } - if (error) { - m_freem(mrep); - return (error); - } - } else if (NFS_CMPFH(np, fhp, fhsize)) { - VREF(dvp); - newvp = dvp; - } else { - error = nfs_nget(mp, fhp, fhsize, &np, cnp->cn_lkflags); - if (error) { - m_freem(mrep); - return (error); - } - newvp = NFSTOV(np); - - /* - * Flush the attribute cache when opening a leaf node - * to ensure that fresh attributes are fetched in - * nfs_open() if we are unable to fetch attributes - * from the LOOKUP reply. - */ - if ((flags & (ISLASTCN | ISOPEN)) == (ISLASTCN | ISOPEN) && - !(np->n_flag & NMODIFIED)) { - mtx_lock(&np->n_mtx); - np->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(newvp); - mtx_unlock(&np->n_mtx); - } - } - if (v3) { - nfsm_postop_attr_va(newvp, attrflag, &vattr); - nfsm_postop_attr_va(dvp, dattrflag, &dvattr); - } else { - nfsm_loadattr(newvp, &vattr); - attrflag = 1; - } - if (cnp->cn_nameiop != LOOKUP && (flags & ISLASTCN)) - cnp->cn_flags |= SAVENAME; - if ((cnp->cn_flags & MAKEENTRY) && - (cnp->cn_nameiop != DELETE || !(flags & ISLASTCN)) && - attrflag != 0 && (newvp->v_type != VDIR || dattrflag != 0)) - cache_enter_time(dvp, newvp, cnp, &vattr.va_ctime, - newvp->v_type != VDIR ? NULL : &dvattr.va_ctime); - *vpp = newvp; - m_freem(mrep); -nfsmout: - if (error) { - if (newvp != NULLVP) { - vput(newvp); - *vpp = NULLVP; - } - - if (error != ENOENT) - goto done; - - /* The requested file was not found. */ - if ((cnp->cn_nameiop == CREATE || cnp->cn_nameiop == RENAME) && - (flags & ISLASTCN)) { - /* - * XXX: UFS does a full VOP_ACCESS(dvp, - * VWRITE) here instead of just checking - * MNT_RDONLY. - */ - if (mp->mnt_flag & MNT_RDONLY) - return (EROFS); - cnp->cn_flags |= SAVENAME; - return (EJUSTRETURN); - } - - if ((cnp->cn_flags & MAKEENTRY) != 0 && dattrflag) { - /* - * Cache the modification time of the parent - * directory from the post-op attributes in - * the name cache entry. The negative cache - * entry will be ignored once the directory - * has changed. Don't bother adding the entry - * if the directory has already changed. - */ - mtx_lock(&np->n_mtx); - if (timespeccmp(&np->n_vattr.va_mtime, - &vattr.va_mtime, ==)) { - mtx_unlock(&np->n_mtx); - cache_enter_time(dvp, NULL, cnp, - &vattr.va_mtime, NULL); - } else - mtx_unlock(&np->n_mtx); - } - return (ENOENT); - } -done: - return (error); -} - -/* - * nfs read call. - * Just call nfs_bioread() to do the work. - */ -static int -nfs_read(struct vop_read_args *ap) -{ - struct vnode *vp = ap->a_vp; - - switch (vp->v_type) { - case VREG: - return (nfs_bioread(vp, ap->a_uio, ap->a_ioflag, ap->a_cred)); - case VDIR: - return (EISDIR); - default: - return (EOPNOTSUPP); - } -} - -/* - * nfs readlink call - */ -static int -nfs_readlink(struct vop_readlink_args *ap) -{ - struct vnode *vp = ap->a_vp; - - if (vp->v_type != VLNK) - return (EINVAL); - return (nfs_bioread(vp, ap->a_uio, 0, ap->a_cred)); -} - -/* - * Do a readlink rpc. - * Called by nfs_doio() from below the buffer cache. - */ -int -nfs_readlinkrpc(struct vnode *vp, struct uio *uiop, struct ucred *cred) -{ - caddr_t bpos, dpos; - int error = 0, len, attrflag; - struct mbuf *mreq, *mrep, *md, *mb; - int v3 = NFS_ISV3(vp); - - nfsstats.rpccnt[NFSPROC_READLINK]++; - mreq = m_get2(NFSX_FH(v3), M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(vp, v3); - nfsm_request(vp, NFSPROC_READLINK, uiop->uio_td, cred); - if (v3) - nfsm_postop_attr(vp, attrflag); - if (!error) { - nfsm_strsiz(len, NFS_MAXPATHLEN); - if (len == NFS_MAXPATHLEN) { - struct nfsnode *np = VTONFS(vp); - mtx_lock(&np->n_mtx); - if (np->n_size && np->n_size < NFS_MAXPATHLEN) - len = np->n_size; - mtx_unlock(&np->n_mtx); - } - nfsm_mtouio(uiop, len); - } - m_freem(mrep); -nfsmout: - return (error); -} - -/* - * nfs read rpc call - * Ditto above - */ -int -nfs_readrpc(struct vnode *vp, struct uio *uiop, struct ucred *cred) -{ - u_int32_t *tl; - caddr_t bpos, dpos; - struct mbuf *mreq, *mrep, *md, *mb; - struct nfsmount *nmp; - off_t end; - int error = 0, len, retlen, tsiz, eof, attrflag; - int v3 = NFS_ISV3(vp); - int rsize; - -#ifndef nolint - eof = 0; -#endif - nmp = VFSTONFS(vp->v_mount); - tsiz = uiop->uio_resid; - mtx_lock(&nmp->nm_mtx); - end = uiop->uio_offset + tsiz; - if (end > nmp->nm_maxfilesize || end < uiop->uio_offset) { - mtx_unlock(&nmp->nm_mtx); - return (EFBIG); - } - rsize = nmp->nm_rsize; - mtx_unlock(&nmp->nm_mtx); - while (tsiz > 0) { - nfsstats.rpccnt[NFSPROC_READ]++; - len = (tsiz > rsize) ? rsize : tsiz; - mreq = m_get2(NFSX_FH(v3) + NFSX_UNSIGNED * 3, M_WAITOK, - MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(vp, v3); - tl = nfsm_build(u_int32_t *, NFSX_UNSIGNED * 3); - if (v3) { - txdr_hyper(uiop->uio_offset, tl); - *(tl + 2) = txdr_unsigned(len); - } else { - *tl++ = txdr_unsigned(uiop->uio_offset); - *tl++ = txdr_unsigned(len); - *tl = 0; - } - nfsm_request(vp, NFSPROC_READ, uiop->uio_td, cred); - if (v3) { - nfsm_postop_attr(vp, attrflag); - if (error) { - m_freem(mrep); - goto nfsmout; - } - tl = nfsm_dissect(u_int32_t *, 2 * NFSX_UNSIGNED); - eof = fxdr_unsigned(int, *(tl + 1)); - } else { - nfsm_loadattr(vp, NULL); - } - nfsm_strsiz(retlen, rsize); - nfsm_mtouio(uiop, retlen); - m_freem(mrep); - tsiz -= retlen; - if (v3) { - if (eof || retlen == 0) { - tsiz = 0; - } - } else if (retlen < len) { - tsiz = 0; - } - } -nfsmout: - return (error); -} - -/* - * nfs write call - */ -int -nfs_writerpc(struct vnode *vp, struct uio *uiop, struct ucred *cred, - int *iomode, int *must_commit) -{ - u_int32_t *tl; - int32_t backup; - caddr_t bpos, dpos; - struct mbuf *mreq, *mrep, *md, *mb; - struct nfsmount *nmp = VFSTONFS(vp->v_mount); - off_t end; - int error = 0, len, tsiz, wccflag = NFSV3_WCCRATTR, rlen, commit; - int v3 = NFS_ISV3(vp), committed = NFSV3WRITE_FILESYNC; - int wsize; - - KASSERT(uiop->uio_iovcnt == 1, ("nfs: writerpc iovcnt > 1")); - *must_commit = 0; - tsiz = uiop->uio_resid; - mtx_lock(&nmp->nm_mtx); - end = uiop->uio_offset + tsiz; - if (end > nmp->nm_maxfilesize || end < uiop->uio_offset) { - mtx_unlock(&nmp->nm_mtx); - return (EFBIG); - } - wsize = nmp->nm_wsize; - mtx_unlock(&nmp->nm_mtx); - while (tsiz > 0) { - nfsstats.rpccnt[NFSPROC_WRITE]++; - len = (tsiz > wsize) ? wsize : tsiz; - mreq = m_get2(NFSX_FH(v3) + 5 * NFSX_UNSIGNED, M_WAITOK, - MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(vp, v3); - if (v3) { - tl = nfsm_build(u_int32_t *, 5 * NFSX_UNSIGNED); - txdr_hyper(uiop->uio_offset, tl); - tl += 2; - *tl++ = txdr_unsigned(len); - *tl++ = txdr_unsigned(*iomode); - *tl = txdr_unsigned(len); - } else { - u_int32_t x; - - tl = nfsm_build(u_int32_t *, 4 * NFSX_UNSIGNED); - /* Set both "begin" and "current" to non-garbage. */ - x = txdr_unsigned((u_int32_t)uiop->uio_offset); - *tl++ = x; /* "begin offset" */ - *tl++ = x; /* "current offset" */ - x = txdr_unsigned(len); - *tl++ = x; /* total to this offset */ - *tl = x; /* size of this write */ - } - nfsm_uiotom(uiop, len); - nfsm_request(vp, NFSPROC_WRITE, uiop->uio_td, cred); - if (v3) { - wccflag = NFSV3_WCCCHK; - nfsm_wcc_data(vp, wccflag); - if (!error) { - tl = nfsm_dissect(u_int32_t *, 2 * NFSX_UNSIGNED - + NFSX_V3WRITEVERF); - rlen = fxdr_unsigned(int, *tl++); - if (rlen == 0) { - error = NFSERR_IO; - m_freem(mrep); - break; - } else if (rlen < len) { - backup = len - rlen; - uiop->uio_iov->iov_base = - (char *)uiop->uio_iov->iov_base - - backup; - uiop->uio_iov->iov_len += backup; - uiop->uio_offset -= backup; - uiop->uio_resid += backup; - len = rlen; - } - commit = fxdr_unsigned(int, *tl++); - - /* - * Return the lowest committment level - * obtained by any of the RPCs. - */ - if (committed == NFSV3WRITE_FILESYNC) - committed = commit; - else if (committed == NFSV3WRITE_DATASYNC && - commit == NFSV3WRITE_UNSTABLE) - committed = commit; - mtx_lock(&nmp->nm_mtx); - if ((nmp->nm_state & NFSSTA_HASWRITEVERF) == 0){ - bcopy((caddr_t)tl, (caddr_t)nmp->nm_verf, - NFSX_V3WRITEVERF); - nmp->nm_state |= NFSSTA_HASWRITEVERF; - } else if (bcmp((caddr_t)tl, - (caddr_t)nmp->nm_verf, NFSX_V3WRITEVERF)) { - *must_commit = 1; - bcopy((caddr_t)tl, (caddr_t)nmp->nm_verf, - NFSX_V3WRITEVERF); - } - mtx_unlock(&nmp->nm_mtx); - } - } else { - nfsm_loadattr(vp, NULL); - } - if (wccflag) { - mtx_lock(&(VTONFS(vp))->n_mtx); - VTONFS(vp)->n_mtime = VTONFS(vp)->n_vattr.va_mtime; - mtx_unlock(&(VTONFS(vp))->n_mtx); - } - m_freem(mrep); - if (error) - break; - tsiz -= len; - } -nfsmout: - if (DOINGASYNC(vp)) - committed = NFSV3WRITE_FILESYNC; - *iomode = committed; - if (error) - uiop->uio_resid = tsiz; - return (error); -} - -/* - * nfs mknod rpc - * For NFS v2 this is a kludge. Use a create rpc but with the IFMT bits of the - * mode set to specify the file type and the size field for rdev. - */ -static int -nfs_mknodrpc(struct vnode *dvp, struct vnode **vpp, struct componentname *cnp, - struct vattr *vap) -{ - struct nfsv2_sattr *sp; - u_int32_t *tl; - struct vnode *newvp = NULL; - struct nfsnode *np = NULL; - struct vattr vattr; - caddr_t bpos, dpos; - int error = 0, wccflag = NFSV3_WCCRATTR, gotvp = 0; - struct mbuf *mreq, *mrep, *md, *mb; - u_int32_t rdev; - int v3 = NFS_ISV3(dvp); - - if (vap->va_type == VCHR || vap->va_type == VBLK) - rdev = txdr_unsigned(vap->va_rdev); - else if (vap->va_type == VFIFO || vap->va_type == VSOCK) - rdev = nfs_xdrneg1; - else { - return (EOPNOTSUPP); - } - if ((error = VOP_GETATTR(dvp, &vattr, cnp->cn_cred)) != 0) - return (error); - nfsstats.rpccnt[NFSPROC_MKNOD]++; - mreq = m_get2(NFSX_FH(v3) + 4 * NFSX_UNSIGNED + - nfsm_rndup(cnp->cn_namelen) + NFSX_SATTR(v3), M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(dvp, v3); - nfsm_strtom(cnp->cn_nameptr, cnp->cn_namelen, NFS_MAXNAMLEN); - if (v3) { - tl = nfsm_build(u_int32_t *, NFSX_UNSIGNED); - *tl++ = vtonfsv3_type(vap->va_type); - nfsm_v3attrbuild(vap, FALSE); - if (vap->va_type == VCHR || vap->va_type == VBLK) { - tl = nfsm_build(u_int32_t *, 2 * NFSX_UNSIGNED); - *tl++ = txdr_unsigned(major(vap->va_rdev)); - *tl = txdr_unsigned(minor(vap->va_rdev)); - } - } else { - sp = nfsm_build(struct nfsv2_sattr *, NFSX_V2SATTR); - sp->sa_mode = vtonfsv2_mode(vap->va_type, vap->va_mode); - sp->sa_uid = nfs_xdrneg1; - sp->sa_gid = nfs_xdrneg1; - sp->sa_size = rdev; - txdr_nfsv2time(&vap->va_atime, &sp->sa_atime); - txdr_nfsv2time(&vap->va_mtime, &sp->sa_mtime); - } - nfsm_request(dvp, NFSPROC_MKNOD, cnp->cn_thread, cnp->cn_cred); - if (!error) { - nfsm_mtofh(dvp, newvp, v3, gotvp); - if (!gotvp) { - if (newvp) { - vput(newvp); - newvp = NULL; - } - error = nfs_lookitup(dvp, cnp->cn_nameptr, - cnp->cn_namelen, cnp->cn_cred, cnp->cn_thread, &np); - if (!error) - newvp = NFSTOV(np); - } - } - if (v3) - nfsm_wcc_data(dvp, wccflag); - m_freem(mrep); -nfsmout: - if (error) { - if (newvp) - vput(newvp); - } else { - *vpp = newvp; - } - mtx_lock(&(VTONFS(dvp))->n_mtx); - VTONFS(dvp)->n_flag |= NMODIFIED; - if (!wccflag) { - VTONFS(dvp)->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(dvp); - } - mtx_unlock(&(VTONFS(dvp))->n_mtx); - return (error); -} - -/* - * nfs mknod vop - * just call nfs_mknodrpc() to do the work. - */ -/* ARGSUSED */ -static int -nfs_mknod(struct vop_mknod_args *ap) -{ - return (nfs_mknodrpc(ap->a_dvp, ap->a_vpp, ap->a_cnp, ap->a_vap)); -} - -static u_long create_verf; -/* - * nfs file create call - */ -static int -nfs_create(struct vop_create_args *ap) -{ - struct vnode *dvp = ap->a_dvp; - struct vattr *vap = ap->a_vap; - struct componentname *cnp = ap->a_cnp; - struct nfsv2_sattr *sp; - u_int32_t *tl; - struct nfsnode *np = NULL; - struct vnode *newvp = NULL; - caddr_t bpos, dpos; - int error = 0, wccflag = NFSV3_WCCRATTR, gotvp = 0, fmode = 0; - struct mbuf *mreq, *mrep, *md, *mb; - struct vattr vattr; - int v3 = NFS_ISV3(dvp); - - /* - * Oops, not for me.. - */ - if (vap->va_type == VSOCK) { - error = nfs_mknodrpc(dvp, ap->a_vpp, cnp, vap); - return (error); - } - - if ((error = VOP_GETATTR(dvp, &vattr, cnp->cn_cred)) != 0) { - return (error); - } - if (vap->va_vaflags & VA_EXCLUSIVE) - fmode |= O_EXCL; -again: - nfsstats.rpccnt[NFSPROC_CREATE]++; - mreq = m_get2(NFSX_FH(v3) + 2 * NFSX_UNSIGNED + - nfsm_rndup(cnp->cn_namelen) + NFSX_SATTR(v3), M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(dvp, v3); - nfsm_strtom(cnp->cn_nameptr, cnp->cn_namelen, NFS_MAXNAMLEN); - if (v3) { - tl = nfsm_build(u_int32_t *, NFSX_UNSIGNED); - if (fmode & O_EXCL) { - *tl = txdr_unsigned(NFSV3CREATE_EXCLUSIVE); - tl = nfsm_build(u_int32_t *, NFSX_V3CREATEVERF); -#ifdef INET - CURVNET_SET(CRED_TO_VNET(cnp->cn_cred)); - IN_IFADDR_RLOCK(); - if (!TAILQ_EMPTY(&V_in_ifaddrhead)) - *tl++ = IA_SIN(TAILQ_FIRST(&V_in_ifaddrhead))->sin_addr.s_addr; - else -#endif - *tl++ = create_verf; -#ifdef INET - IN_IFADDR_RUNLOCK(); - CURVNET_RESTORE(); -#endif - *tl = ++create_verf; - } else { - *tl = txdr_unsigned(NFSV3CREATE_UNCHECKED); - nfsm_v3attrbuild(vap, FALSE); - } - } else { - sp = nfsm_build(struct nfsv2_sattr *, NFSX_V2SATTR); - sp->sa_mode = vtonfsv2_mode(vap->va_type, vap->va_mode); - sp->sa_uid = nfs_xdrneg1; - sp->sa_gid = nfs_xdrneg1; - sp->sa_size = 0; - txdr_nfsv2time(&vap->va_atime, &sp->sa_atime); - txdr_nfsv2time(&vap->va_mtime, &sp->sa_mtime); - } - nfsm_request(dvp, NFSPROC_CREATE, cnp->cn_thread, cnp->cn_cred); - if (!error) { - nfsm_mtofh(dvp, newvp, v3, gotvp); - if (!gotvp) { - if (newvp) { - vput(newvp); - newvp = NULL; - } - error = nfs_lookitup(dvp, cnp->cn_nameptr, - cnp->cn_namelen, cnp->cn_cred, cnp->cn_thread, &np); - if (!error) - newvp = NFSTOV(np); - } - } - if (v3) - nfsm_wcc_data(dvp, wccflag); - m_freem(mrep); -nfsmout: - if (error) { - if (v3 && (fmode & O_EXCL) && error == NFSERR_NOTSUPP) { - fmode &= ~O_EXCL; - goto again; - } - if (newvp) - vput(newvp); - } else if (v3 && (fmode & O_EXCL)) { - /* - * We are normally called with only a partially initialized - * VAP. Since the NFSv3 spec says that server may use the - * file attributes to store the verifier, the spec requires - * us to do a SETATTR RPC. FreeBSD servers store the verifier - * in atime, but we can't really assume that all servers will - * so we ensure that our SETATTR sets both atime and mtime. - */ - if (vap->va_mtime.tv_sec == VNOVAL) - vfs_timestamp(&vap->va_mtime); - if (vap->va_atime.tv_sec == VNOVAL) - vap->va_atime = vap->va_mtime; - error = nfs_setattrrpc(newvp, vap, cnp->cn_cred); - if (error) - vput(newvp); - } - if (!error) { - *ap->a_vpp = newvp; - } - mtx_lock(&(VTONFS(dvp))->n_mtx); - VTONFS(dvp)->n_flag |= NMODIFIED; - if (!wccflag) { - VTONFS(dvp)->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(dvp); - } - mtx_unlock(&(VTONFS(dvp))->n_mtx); - return (error); -} - -/* - * nfs file remove call - * To try and make nfs semantics closer to ufs semantics, a file that has - * other processes using the vnode is renamed instead of removed and then - * removed later on the last close. - * - If v_usecount > 1 - * If a rename is not already in the works - * call nfs_sillyrename() to set it up - * else - * do the remove rpc - */ -static int -nfs_remove(struct vop_remove_args *ap) -{ - struct vnode *vp = ap->a_vp; - struct vnode *dvp = ap->a_dvp; - struct componentname *cnp = ap->a_cnp; - struct nfsnode *np = VTONFS(vp); - int error = 0; - struct vattr vattr; - - KASSERT((cnp->cn_flags & HASBUF) != 0, ("nfs_remove: no name")); - KASSERT(vrefcnt(vp) > 0, ("nfs_remove: bad v_usecount")); - if (vp->v_type == VDIR) - error = EPERM; - else if (vrefcnt(vp) == 1 || (np->n_sillyrename && - !VOP_GETATTR(vp, &vattr, cnp->cn_cred) && vattr.va_nlink > 1)) { - /* - * Purge the name cache so that the chance of a lookup for - * the name succeeding while the remove is in progress is - * minimized. Without node locking it can still happen, such - * that an I/O op returns ESTALE, but since you get this if - * another host removes the file.. - */ - cache_purge(vp); - /* - * throw away biocache buffers, mainly to avoid - * unnecessary delayed writes later. - */ - error = nfs_vinvalbuf(vp, 0, cnp->cn_thread, 1); - /* Do the rpc */ - if (error != EINTR && error != EIO) - error = nfs_removerpc(dvp, cnp->cn_nameptr, - cnp->cn_namelen, cnp->cn_cred, cnp->cn_thread); - /* - * Kludge City: If the first reply to the remove rpc is lost.. - * the reply to the retransmitted request will be ENOENT - * since the file was in fact removed - * Therefore, we cheat and return success. - */ - if (error == ENOENT) - error = 0; - } else if (!np->n_sillyrename) - error = nfs_sillyrename(dvp, vp, cnp); - mtx_lock(&np->n_mtx); - np->n_attrstamp = 0; - mtx_unlock(&np->n_mtx); - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp); - return (error); -} - -/* - * nfs file remove rpc called from nfs_inactive - */ -int -nfs_removeit(struct sillyrename *sp) -{ - /* - * Make sure that the directory vnode is still valid. - * XXX we should lock sp->s_dvp here. - */ - if (sp->s_dvp->v_type == VBAD) - return (0); - return (nfs_removerpc(sp->s_dvp, sp->s_name, sp->s_namlen, sp->s_cred, - NULL)); -} - -/* - * Nfs remove rpc, called from nfs_remove() and nfs_removeit(). - */ -static int -nfs_removerpc(struct vnode *dvp, const char *name, int namelen, - struct ucred *cred, struct thread *td) -{ - caddr_t bpos, dpos; - int error = 0, wccflag = NFSV3_WCCRATTR; - struct mbuf *mreq, *mrep, *md, *mb; - int v3 = NFS_ISV3(dvp); - - nfsstats.rpccnt[NFSPROC_REMOVE]++; - mreq = m_get2(NFSX_FH(v3) + NFSX_UNSIGNED + nfsm_rndup(namelen), - M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(dvp, v3); - nfsm_strtom(name, namelen, NFS_MAXNAMLEN); - nfsm_request(dvp, NFSPROC_REMOVE, td, cred); - if (v3) - nfsm_wcc_data(dvp, wccflag); - m_freem(mrep); -nfsmout: - mtx_lock(&(VTONFS(dvp))->n_mtx); - VTONFS(dvp)->n_flag |= NMODIFIED; - if (!wccflag) { - VTONFS(dvp)->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(dvp); - } - mtx_unlock(&(VTONFS(dvp))->n_mtx); - return (error); -} - -/* - * nfs file rename call - */ -static int -nfs_rename(struct vop_rename_args *ap) -{ - struct vnode *fvp = ap->a_fvp; - struct vnode *tvp = ap->a_tvp; - struct vnode *fdvp = ap->a_fdvp; - struct vnode *tdvp = ap->a_tdvp; - struct componentname *tcnp = ap->a_tcnp; - struct componentname *fcnp = ap->a_fcnp; - int error; - - KASSERT((tcnp->cn_flags & HASBUF) != 0 && - (fcnp->cn_flags & HASBUF) != 0, ("nfs_rename: no name")); - /* Check for cross-device rename */ - if ((fvp->v_mount != tdvp->v_mount) || - (tvp && (fvp->v_mount != tvp->v_mount))) { - error = EXDEV; - goto out; - } - - if (fvp == tvp) { - nfs_printf("nfs_rename: fvp == tvp (can't happen)\n"); - error = 0; - goto out; - } - if ((error = vn_lock(fvp, LK_EXCLUSIVE)) != 0) - goto out; - - /* - * We have to flush B_DELWRI data prior to renaming - * the file. If we don't, the delayed-write buffers - * can be flushed out later after the file has gone stale - * under NFSV3. NFSV2 does not have this problem because - * ( as far as I can tell ) it flushes dirty buffers more - * often. - * - * Skip the rename operation if the fsync fails, this can happen - * due to the server's volume being full, when we pushed out data - * that was written back to our cache earlier. Not checking for - * this condition can result in potential (silent) data loss. - */ - error = VOP_FSYNC(fvp, MNT_WAIT, fcnp->cn_thread); - VOP_UNLOCK(fvp, 0); - if (!error && tvp) - error = VOP_FSYNC(tvp, MNT_WAIT, tcnp->cn_thread); - if (error) - goto out; - - /* - * If the tvp exists and is in use, sillyrename it before doing the - * rename of the new file over it. - * XXX Can't sillyrename a directory. - */ - if (tvp && vrefcnt(tvp) > 1 && !VTONFS(tvp)->n_sillyrename && - tvp->v_type != VDIR && !nfs_sillyrename(tdvp, tvp, tcnp)) { - vput(tvp); - tvp = NULL; - } - - error = nfs_renamerpc(fdvp, fcnp->cn_nameptr, fcnp->cn_namelen, - tdvp, tcnp->cn_nameptr, tcnp->cn_namelen, tcnp->cn_cred, - tcnp->cn_thread); - - if (fvp->v_type == VDIR) { - if (tvp != NULL && tvp->v_type == VDIR) - cache_purge(tdvp); - cache_purge(fdvp); - } - -out: - if (tdvp == tvp) - vrele(tdvp); - else - vput(tdvp); - if (tvp) - vput(tvp); - vrele(fdvp); - vrele(fvp); - /* - * Kludge: Map ENOENT => 0 assuming that it is a reply to a retry. - */ - if (error == ENOENT) - error = 0; - return (error); -} - -/* - * nfs file rename rpc called from nfs_remove() above - */ -static int -nfs_renameit(struct vnode *sdvp, struct componentname *scnp, - struct sillyrename *sp) -{ - - return (nfs_renamerpc(sdvp, scnp->cn_nameptr, scnp->cn_namelen, sdvp, - sp->s_name, sp->s_namlen, scnp->cn_cred, scnp->cn_thread)); -} - -/* - * Do an nfs rename rpc. Called from nfs_rename() and nfs_renameit(). - */ -static int -nfs_renamerpc(struct vnode *fdvp, const char *fnameptr, int fnamelen, - struct vnode *tdvp, const char *tnameptr, int tnamelen, struct ucred *cred, - struct thread *td) -{ - caddr_t bpos, dpos; - int error = 0, fwccflag = NFSV3_WCCRATTR, twccflag = NFSV3_WCCRATTR; - struct mbuf *mreq, *mrep, *md, *mb; - int v3 = NFS_ISV3(fdvp); - - nfsstats.rpccnt[NFSPROC_RENAME]++; - mreq = m_get2((NFSX_FH(v3) + NFSX_UNSIGNED)*2 + nfsm_rndup(fnamelen) + - nfsm_rndup(tnamelen), M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(fdvp, v3); - nfsm_strtom(fnameptr, fnamelen, NFS_MAXNAMLEN); - nfsm_fhtom(tdvp, v3); - nfsm_strtom(tnameptr, tnamelen, NFS_MAXNAMLEN); - nfsm_request(fdvp, NFSPROC_RENAME, td, cred); - if (v3) { - nfsm_wcc_data(fdvp, fwccflag); - nfsm_wcc_data(tdvp, twccflag); - } - m_freem(mrep); -nfsmout: - mtx_lock(&(VTONFS(fdvp))->n_mtx); - VTONFS(fdvp)->n_flag |= NMODIFIED; - mtx_unlock(&(VTONFS(fdvp))->n_mtx); - mtx_lock(&(VTONFS(tdvp))->n_mtx); - VTONFS(tdvp)->n_flag |= NMODIFIED; - mtx_unlock(&(VTONFS(tdvp))->n_mtx); - if (!fwccflag) { - VTONFS(fdvp)->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(fdvp); - } - if (!twccflag) { - VTONFS(tdvp)->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(tdvp); - } - return (error); -} - -/* - * nfs hard link create call - */ -static int -nfs_link(struct vop_link_args *ap) -{ - struct vnode *vp = ap->a_vp; - struct vnode *tdvp = ap->a_tdvp; - struct componentname *cnp = ap->a_cnp; - caddr_t bpos, dpos; - int error = 0, wccflag = NFSV3_WCCRATTR, attrflag = 0; - struct mbuf *mreq, *mrep, *md, *mb; - int v3; - - if (vp->v_mount != tdvp->v_mount) { - return (EXDEV); - } - - /* - * Push all writes to the server, so that the attribute cache - * doesn't get "out of sync" with the server. - * XXX There should be a better way! - */ - VOP_FSYNC(vp, MNT_WAIT, cnp->cn_thread); - - v3 = NFS_ISV3(vp); - nfsstats.rpccnt[NFSPROC_LINK]++; - mreq = m_get2(NFSX_FH(v3)*2 + NFSX_UNSIGNED + - nfsm_rndup(cnp->cn_namelen), M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(vp, v3); - nfsm_fhtom(tdvp, v3); - nfsm_strtom(cnp->cn_nameptr, cnp->cn_namelen, NFS_MAXNAMLEN); - nfsm_request(vp, NFSPROC_LINK, cnp->cn_thread, cnp->cn_cred); - if (v3) { - nfsm_postop_attr(vp, attrflag); - nfsm_wcc_data(tdvp, wccflag); - } - m_freem(mrep); -nfsmout: - mtx_lock(&(VTONFS(tdvp))->n_mtx); - VTONFS(tdvp)->n_flag |= NMODIFIED; - mtx_unlock(&(VTONFS(tdvp))->n_mtx); - if (!attrflag) { - VTONFS(vp)->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(vp); - } - if (!wccflag) { - VTONFS(tdvp)->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(tdvp); - } - return (error); -} - -/* - * nfs symbolic link create call - */ -static int -nfs_symlink(struct vop_symlink_args *ap) -{ - struct vnode *dvp = ap->a_dvp; - struct vattr *vap = ap->a_vap; - struct componentname *cnp = ap->a_cnp; - struct nfsv2_sattr *sp; - caddr_t bpos, dpos; - int slen, error = 0, wccflag = NFSV3_WCCRATTR, gotvp; - struct mbuf *mreq, *mrep, *md, *mb; - struct vnode *newvp = NULL; - int v3 = NFS_ISV3(dvp); - - nfsstats.rpccnt[NFSPROC_SYMLINK]++; - slen = strlen(ap->a_target); - mreq = m_get2(NFSX_FH(v3) + 2*NFSX_UNSIGNED + - nfsm_rndup(cnp->cn_namelen) + nfsm_rndup(slen) + NFSX_SATTR(v3), - M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(dvp, v3); - nfsm_strtom(cnp->cn_nameptr, cnp->cn_namelen, NFS_MAXNAMLEN); - if (v3) { - nfsm_v3attrbuild(vap, FALSE); - } - nfsm_strtom(ap->a_target, slen, NFS_MAXPATHLEN); - if (!v3) { - sp = nfsm_build(struct nfsv2_sattr *, NFSX_V2SATTR); - sp->sa_mode = vtonfsv2_mode(VLNK, vap->va_mode); - sp->sa_uid = nfs_xdrneg1; - sp->sa_gid = nfs_xdrneg1; - sp->sa_size = nfs_xdrneg1; - txdr_nfsv2time(&vap->va_atime, &sp->sa_atime); - txdr_nfsv2time(&vap->va_mtime, &sp->sa_mtime); - } - - /* - * Issue the NFS request and get the rpc response. - * - * Only NFSv3 responses returning an error of 0 actually return - * a file handle that can be converted into newvp without having - * to do an extra lookup rpc. - */ - nfsm_request(dvp, NFSPROC_SYMLINK, cnp->cn_thread, cnp->cn_cred); - if (v3) { - if (error == 0) - nfsm_mtofh(dvp, newvp, v3, gotvp); - nfsm_wcc_data(dvp, wccflag); - } - - /* - * out code jumps -> here, mrep is also freed. - */ - - m_freem(mrep); -nfsmout: - - /* - * If we do not have an error and we could not extract the newvp from - * the response due to the request being NFSv2, we have to do a - * lookup in order to obtain a newvp to return. - */ - if (error == 0 && newvp == NULL) { - struct nfsnode *np = NULL; - - error = nfs_lookitup(dvp, cnp->cn_nameptr, cnp->cn_namelen, - cnp->cn_cred, cnp->cn_thread, &np); - if (!error) - newvp = NFSTOV(np); - } - if (error) { - if (newvp) - vput(newvp); - } else { - *ap->a_vpp = newvp; - } - mtx_lock(&(VTONFS(dvp))->n_mtx); - VTONFS(dvp)->n_flag |= NMODIFIED; - mtx_unlock(&(VTONFS(dvp))->n_mtx); - if (!wccflag) { - VTONFS(dvp)->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(dvp); - } - return (error); -} - -/* - * nfs make dir call - */ -static int -nfs_mkdir(struct vop_mkdir_args *ap) -{ - struct vnode *dvp = ap->a_dvp; - struct vattr *vap = ap->a_vap; - struct componentname *cnp = ap->a_cnp; - struct nfsv2_sattr *sp; - int len; - struct nfsnode *np = NULL; - struct vnode *newvp = NULL; - caddr_t bpos, dpos; - int error = 0, wccflag = NFSV3_WCCRATTR; - int gotvp = 0; - struct mbuf *mreq, *mrep, *md, *mb; - struct vattr vattr; - int v3 = NFS_ISV3(dvp); - - if ((error = VOP_GETATTR(dvp, &vattr, cnp->cn_cred)) != 0) - return (error); - len = cnp->cn_namelen; - nfsstats.rpccnt[NFSPROC_MKDIR]++; - mreq = m_get2(NFSX_FH(v3) + NFSX_UNSIGNED + nfsm_rndup(len) + - NFSX_SATTR(v3), M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(dvp, v3); - nfsm_strtom(cnp->cn_nameptr, len, NFS_MAXNAMLEN); - if (v3) { - nfsm_v3attrbuild(vap, FALSE); - } else { - sp = nfsm_build(struct nfsv2_sattr *, NFSX_V2SATTR); - sp->sa_mode = vtonfsv2_mode(VDIR, vap->va_mode); - sp->sa_uid = nfs_xdrneg1; - sp->sa_gid = nfs_xdrneg1; - sp->sa_size = nfs_xdrneg1; - txdr_nfsv2time(&vap->va_atime, &sp->sa_atime); - txdr_nfsv2time(&vap->va_mtime, &sp->sa_mtime); - } - nfsm_request(dvp, NFSPROC_MKDIR, cnp->cn_thread, cnp->cn_cred); - if (!error) - nfsm_mtofh(dvp, newvp, v3, gotvp); - if (v3) - nfsm_wcc_data(dvp, wccflag); - m_freem(mrep); -nfsmout: - mtx_lock(&(VTONFS(dvp))->n_mtx); - VTONFS(dvp)->n_flag |= NMODIFIED; - mtx_unlock(&(VTONFS(dvp))->n_mtx); - if (!wccflag) { - VTONFS(dvp)->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(dvp); - } - if (error == 0 && newvp == NULL) { - error = nfs_lookitup(dvp, cnp->cn_nameptr, len, cnp->cn_cred, - cnp->cn_thread, &np); - if (!error) { - newvp = NFSTOV(np); - if (newvp->v_type != VDIR) - error = EEXIST; - } - } - if (error) { - if (newvp) - vput(newvp); - } else - *ap->a_vpp = newvp; - return (error); -} - -/* - * nfs remove directory call - */ -static int -nfs_rmdir(struct vop_rmdir_args *ap) -{ - struct vnode *vp = ap->a_vp; - struct vnode *dvp = ap->a_dvp; - struct componentname *cnp = ap->a_cnp; - caddr_t bpos, dpos; - int error = 0, wccflag = NFSV3_WCCRATTR; - struct mbuf *mreq, *mrep, *md, *mb; - int v3 = NFS_ISV3(dvp); - - if (dvp == vp) - return (EINVAL); - nfsstats.rpccnt[NFSPROC_RMDIR]++; - mreq = m_get2(NFSX_FH(v3) + NFSX_UNSIGNED + - nfsm_rndup(cnp->cn_namelen), M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(dvp, v3); - nfsm_strtom(cnp->cn_nameptr, cnp->cn_namelen, NFS_MAXNAMLEN); - nfsm_request(dvp, NFSPROC_RMDIR, cnp->cn_thread, cnp->cn_cred); - if (v3) - nfsm_wcc_data(dvp, wccflag); - m_freem(mrep); -nfsmout: - mtx_lock(&(VTONFS(dvp))->n_mtx); - VTONFS(dvp)->n_flag |= NMODIFIED; - mtx_unlock(&(VTONFS(dvp))->n_mtx); - if (!wccflag) { - VTONFS(dvp)->n_attrstamp = 0; - KDTRACE_NFS_ATTRCACHE_FLUSH_DONE(dvp); - } - cache_purge(dvp); - cache_purge(vp); - /* - * Kludge: Map ENOENT => 0 assuming that you have a reply to a retry. - */ - if (error == ENOENT) - error = 0; - return (error); -} - -/* - * nfs readdir call - */ -static int -nfs_readdir(struct vop_readdir_args *ap) -{ - struct vnode *vp = ap->a_vp; - struct nfsnode *np = VTONFS(vp); - struct uio *uio = ap->a_uio; - int tresid, error = 0; - struct vattr vattr; - - if (vp->v_type != VDIR) - return(EPERM); - - /* - * First, check for hit on the EOF offset cache - */ - if (np->n_direofoffset > 0 && uio->uio_offset >= np->n_direofoffset && - (np->n_flag & NMODIFIED) == 0) { - if (VOP_GETATTR(vp, &vattr, ap->a_cred) == 0) { - mtx_lock(&np->n_mtx); - if (!NFS_TIMESPEC_COMPARE(&np->n_mtime, &vattr.va_mtime)) { - mtx_unlock(&np->n_mtx); - nfsstats.direofcache_hits++; - goto out; - } else - mtx_unlock(&np->n_mtx); - } - } - - /* - * Call nfs_bioread() to do the real work. - */ - tresid = uio->uio_resid; - error = nfs_bioread(vp, uio, 0, ap->a_cred); - - if (!error && uio->uio_resid == tresid) { - nfsstats.direofcache_misses++; - } -out: - return (error); -} - -/* - * Readdir rpc call. - * Called from below the buffer cache by nfs_doio(). - */ -int -nfs_readdirrpc(struct vnode *vp, struct uio *uiop, struct ucred *cred) -{ - int len, left; - struct dirent *dp = NULL; - u_int32_t *tl; - caddr_t cp; - nfsuint64 *cookiep; - caddr_t bpos, dpos; - struct mbuf *mreq, *mrep, *md, *mb; - nfsuint64 cookie; - struct nfsmount *nmp = VFSTONFS(vp->v_mount); - struct nfsnode *dnp = VTONFS(vp); - u_quad_t fileno; - int error = 0, tlen, more_dirs = 1, blksiz = 0, bigenough = 1; - int attrflag; - int v3 = NFS_ISV3(vp); - - KASSERT(uiop->uio_iovcnt == 1 && - (uiop->uio_offset & (DIRBLKSIZ - 1)) == 0 && - (uiop->uio_resid & (DIRBLKSIZ - 1)) == 0, - ("nfs readdirrpc bad uio")); - - /* - * If there is no cookie, assume directory was stale. - */ - nfs_dircookie_lock(dnp); - cookiep = nfs_getcookie(dnp, uiop->uio_offset, 0); - if (cookiep) { - cookie = *cookiep; - nfs_dircookie_unlock(dnp); - } else { - nfs_dircookie_unlock(dnp); - return (NFSERR_BAD_COOKIE); - } - - /* - * Loop around doing readdir rpc's of size nm_readdirsize - * truncated to a multiple of DIRBLKSIZ. - * The stopping criteria is EOF or buffer full. - */ - while (more_dirs && bigenough) { - nfsstats.rpccnt[NFSPROC_READDIR]++; - mreq = m_get2(NFSX_FH(v3) + NFSX_READDIR(v3), M_WAITOK, - MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(vp, v3); - if (v3) { - tl = nfsm_build(u_int32_t *, 5 * NFSX_UNSIGNED); - *tl++ = cookie.nfsuquad[0]; - *tl++ = cookie.nfsuquad[1]; - mtx_lock(&dnp->n_mtx); - *tl++ = dnp->n_cookieverf.nfsuquad[0]; - *tl++ = dnp->n_cookieverf.nfsuquad[1]; - mtx_unlock(&dnp->n_mtx); - } else { - tl = nfsm_build(u_int32_t *, 2 * NFSX_UNSIGNED); - *tl++ = cookie.nfsuquad[0]; - } - *tl = txdr_unsigned(nmp->nm_readdirsize); - nfsm_request(vp, NFSPROC_READDIR, uiop->uio_td, cred); - if (v3) { - nfsm_postop_attr(vp, attrflag); - if (!error) { - tl = nfsm_dissect(u_int32_t *, - 2 * NFSX_UNSIGNED); - mtx_lock(&dnp->n_mtx); - dnp->n_cookieverf.nfsuquad[0] = *tl++; - dnp->n_cookieverf.nfsuquad[1] = *tl; - mtx_unlock(&dnp->n_mtx); - } else { - m_freem(mrep); - goto nfsmout; - } - } - tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED); - more_dirs = fxdr_unsigned(int, *tl); - - /* loop thru the dir entries, doctoring them to 4bsd form */ - while (more_dirs && bigenough) { - if (v3) { - tl = nfsm_dissect(u_int32_t *, - 3 * NFSX_UNSIGNED); - fileno = fxdr_hyper(tl); - len = fxdr_unsigned(int, *(tl + 2)); - } else { - tl = nfsm_dissect(u_int32_t *, - 2 * NFSX_UNSIGNED); - fileno = fxdr_unsigned(u_quad_t, *tl++); - len = fxdr_unsigned(int, *tl); - } - if (len <= 0 || len > NFS_MAXNAMLEN) { - error = EBADRPC; - m_freem(mrep); - goto nfsmout; - } - tlen = nfsm_rndup(len); - if (tlen == len) - tlen += 4; /* To ensure null termination */ - left = DIRBLKSIZ - blksiz; - if ((tlen + DIRHDSIZ) > left) { - dp->d_reclen += left; - uiop->uio_iov->iov_base = - (char *)uiop->uio_iov->iov_base + left; - uiop->uio_iov->iov_len -= left; - uiop->uio_offset += left; - uiop->uio_resid -= left; - blksiz = 0; - } - if ((tlen + DIRHDSIZ) > uiop->uio_resid) - bigenough = 0; - if (bigenough) { - dp = (struct dirent *)uiop->uio_iov->iov_base; - dp->d_fileno = (int)fileno; - dp->d_namlen = len; - dp->d_reclen = tlen + DIRHDSIZ; - dp->d_type = DT_UNKNOWN; - blksiz += dp->d_reclen; - if (blksiz == DIRBLKSIZ) - blksiz = 0; - uiop->uio_offset += DIRHDSIZ; - uiop->uio_resid -= DIRHDSIZ; - uiop->uio_iov->iov_base = - (char *)uiop->uio_iov->iov_base + DIRHDSIZ; - uiop->uio_iov->iov_len -= DIRHDSIZ; - nfsm_mtouio(uiop, len); - cp = uiop->uio_iov->iov_base; - tlen -= len; - *cp = '\0'; /* null terminate */ - uiop->uio_iov->iov_base = - (char *)uiop->uio_iov->iov_base + tlen; - uiop->uio_iov->iov_len -= tlen; - uiop->uio_offset += tlen; - uiop->uio_resid -= tlen; - } else - nfsm_adv(nfsm_rndup(len)); - if (v3) { - tl = nfsm_dissect(u_int32_t *, - 3 * NFSX_UNSIGNED); - } else { - tl = nfsm_dissect(u_int32_t *, - 2 * NFSX_UNSIGNED); - } - if (bigenough) { - cookie.nfsuquad[0] = *tl++; - if (v3) - cookie.nfsuquad[1] = *tl++; - } else if (v3) - tl += 2; - else - tl++; - more_dirs = fxdr_unsigned(int, *tl); - } - /* - * If at end of rpc data, get the eof boolean - */ - if (!more_dirs) { - tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED); - more_dirs = (fxdr_unsigned(int, *tl) == 0); - } - m_freem(mrep); - } - /* - * Fill last record, iff any, out to a multiple of DIRBLKSIZ - * by increasing d_reclen for the last record. - */ - if (blksiz > 0) { - left = DIRBLKSIZ - blksiz; - dp->d_reclen += left; - uiop->uio_iov->iov_base = - (char *)uiop->uio_iov->iov_base + left; - uiop->uio_iov->iov_len -= left; - uiop->uio_offset += left; - uiop->uio_resid -= left; - } - - /* - * We are now either at the end of the directory or have filled the - * block. - */ - if (bigenough) - dnp->n_direofoffset = uiop->uio_offset; - else { - if (uiop->uio_resid > 0) - nfs_printf("EEK! readdirrpc resid > 0\n"); - nfs_dircookie_lock(dnp); - cookiep = nfs_getcookie(dnp, uiop->uio_offset, 1); - *cookiep = cookie; - nfs_dircookie_unlock(dnp); - } -nfsmout: - return (error); -} - -/* - * NFS V3 readdir plus RPC. Used in place of nfs_readdirrpc(). - */ -int -nfs_readdirplusrpc(struct vnode *vp, struct uio *uiop, struct ucred *cred) -{ - int len, left; - struct dirent *dp; - u_int32_t *tl; - caddr_t cp; - struct vnode *newvp; - nfsuint64 *cookiep; - caddr_t bpos, dpos, dpossav1, dpossav2; - struct mbuf *mreq, *mrep, *md, *mb, *mdsav1, *mdsav2; - struct nameidata nami, *ndp = &nami; - struct componentname *cnp = &ndp->ni_cnd; - nfsuint64 cookie; - struct nfsmount *nmp = VFSTONFS(vp->v_mount); - struct nfsnode *dnp = VTONFS(vp), *np; - struct vattr vattr, dvattr; - nfsfh_t *fhp; - u_quad_t fileno; - int error = 0, tlen, more_dirs = 1, blksiz = 0, doit, bigenough = 1, i; - int attrflag, dattrflag, fhsize; - -#ifndef nolint - dp = NULL; -#endif - KASSERT(uiop->uio_iovcnt == 1 && - (uiop->uio_offset & (DIRBLKSIZ - 1)) == 0 && - (uiop->uio_resid & (DIRBLKSIZ - 1)) == 0, - ("nfs readdirplusrpc bad uio")); - ndp->ni_dvp = vp; - newvp = NULLVP; - - /* - * If there is no cookie, assume directory was stale. - */ - nfs_dircookie_lock(dnp); - cookiep = nfs_getcookie(dnp, uiop->uio_offset, 0); - if (cookiep) { - cookie = *cookiep; - nfs_dircookie_unlock(dnp); - } else { - nfs_dircookie_unlock(dnp); - return (NFSERR_BAD_COOKIE); - } - /* - * Loop around doing readdir rpc's of size nm_readdirsize - * truncated to a multiple of DIRBLKSIZ. - * The stopping criteria is EOF or buffer full. - */ - while (more_dirs && bigenough) { - nfsstats.rpccnt[NFSPROC_READDIRPLUS]++; - mreq = m_get2(NFSX_FH(1) + 6 * NFSX_UNSIGNED, M_WAITOK, - MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(vp, 1); - tl = nfsm_build(u_int32_t *, 6 * NFSX_UNSIGNED); - *tl++ = cookie.nfsuquad[0]; - *tl++ = cookie.nfsuquad[1]; - mtx_lock(&dnp->n_mtx); - *tl++ = dnp->n_cookieverf.nfsuquad[0]; - *tl++ = dnp->n_cookieverf.nfsuquad[1]; - mtx_unlock(&dnp->n_mtx); - *tl++ = txdr_unsigned(nmp->nm_readdirsize); - *tl = txdr_unsigned(nmp->nm_rsize); - nfsm_request(vp, NFSPROC_READDIRPLUS, uiop->uio_td, cred); - nfsm_postop_attr_va(vp, dattrflag, &dvattr); - if (error) { - m_freem(mrep); - goto nfsmout; - } - tl = nfsm_dissect(u_int32_t *, 3 * NFSX_UNSIGNED); - mtx_lock(&dnp->n_mtx); - dnp->n_cookieverf.nfsuquad[0] = *tl++; - dnp->n_cookieverf.nfsuquad[1] = *tl++; - mtx_unlock(&dnp->n_mtx); - more_dirs = fxdr_unsigned(int, *tl); - - /* loop thru the dir entries, doctoring them to 4bsd form */ - while (more_dirs && bigenough) { - tl = nfsm_dissect(u_int32_t *, 3 * NFSX_UNSIGNED); - fileno = fxdr_hyper(tl); - len = fxdr_unsigned(int, *(tl + 2)); - if (len <= 0 || len > NFS_MAXNAMLEN) { - error = EBADRPC; - m_freem(mrep); - goto nfsmout; - } - tlen = nfsm_rndup(len); - if (tlen == len) - tlen += 4; /* To ensure null termination*/ - left = DIRBLKSIZ - blksiz; - if ((tlen + DIRHDSIZ) > left) { - dp->d_reclen += left; - uiop->uio_iov->iov_base = - (char *)uiop->uio_iov->iov_base + left; - uiop->uio_iov->iov_len -= left; - uiop->uio_offset += left; - uiop->uio_resid -= left; - blksiz = 0; - } - if ((tlen + DIRHDSIZ) > uiop->uio_resid) - bigenough = 0; - if (bigenough) { - dp = (struct dirent *)uiop->uio_iov->iov_base; - dp->d_fileno = (int)fileno; - dp->d_namlen = len; - dp->d_reclen = tlen + DIRHDSIZ; - dp->d_type = DT_UNKNOWN; - blksiz += dp->d_reclen; - if (blksiz == DIRBLKSIZ) - blksiz = 0; - uiop->uio_offset += DIRHDSIZ; - uiop->uio_resid -= DIRHDSIZ; - uiop->uio_iov->iov_base = - (char *)uiop->uio_iov->iov_base + DIRHDSIZ; - uiop->uio_iov->iov_len -= DIRHDSIZ; - cnp->cn_nameptr = uiop->uio_iov->iov_base; - cnp->cn_namelen = len; - nfsm_mtouio(uiop, len); - cp = uiop->uio_iov->iov_base; - tlen -= len; - *cp = '\0'; - uiop->uio_iov->iov_base = - (char *)uiop->uio_iov->iov_base + tlen; - uiop->uio_iov->iov_len -= tlen; - uiop->uio_offset += tlen; - uiop->uio_resid -= tlen; - } else - nfsm_adv(nfsm_rndup(len)); - tl = nfsm_dissect(u_int32_t *, 3 * NFSX_UNSIGNED); - if (bigenough) { - cookie.nfsuquad[0] = *tl++; - cookie.nfsuquad[1] = *tl++; - } else - tl += 2; - - /* - * Since the attributes are before the file handle - * (sigh), we must skip over the attributes and then - * come back and get them. - */ - attrflag = fxdr_unsigned(int, *tl); - if (attrflag) { - dpossav1 = dpos; - mdsav1 = md; - nfsm_adv(NFSX_V3FATTR); - tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED); - doit = fxdr_unsigned(int, *tl); - /* - * Skip loading the attrs for "..". There's a - * race between loading the attrs here and - * lookups that look for the directory currently - * being read (in the parent). We try to acquire - * the exclusive lock on ".." here, owning the - * lock on the directory being read. Lookup will - * hold the lock on ".." and try to acquire the - * lock on the directory being read. - * - * There are other ways of fixing this, one would - * be to do a trylock on the ".." vnode and skip - * loading the attrs on ".." if it happens to be - * locked by another process. But skipping the - * attrload on ".." seems the easiest option. - */ - if (strcmp(dp->d_name, "..") == 0) { - doit = 0; - /* - * We've already skipped over the attrs, - * skip over the filehandle. And store d_type - * as VDIR. - */ - tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED); - i = fxdr_unsigned(int, *tl); - nfsm_adv(nfsm_rndup(i)); - dp->d_type = IFTODT(VTTOIF(VDIR)); - } - if (doit) { - nfsm_getfh(fhp, fhsize, 1); - if (NFS_CMPFH(dnp, fhp, fhsize)) { - VREF(vp); - newvp = vp; - np = dnp; - } else { - error = nfs_nget(vp->v_mount, fhp, - fhsize, &np, LK_EXCLUSIVE); - if (error) - doit = 0; - else - newvp = NFSTOV(np); - } - } - if (doit && bigenough) { - dpossav2 = dpos; - dpos = dpossav1; - mdsav2 = md; - md = mdsav1; - nfsm_loadattr(newvp, &vattr); - dpos = dpossav2; - md = mdsav2; - dp->d_type = IFTODT(VTTOIF(vattr.va_type)); - ndp->ni_vp = newvp; - if (newvp->v_type != VDIR || dattrflag != 0) - cache_enter_time(ndp->ni_dvp, ndp->ni_vp, - cnp, &vattr.va_ctime, - newvp->v_type != VDIR ? NULL : - &dvattr.va_ctime); - } - } else { - /* Just skip over the file handle */ - tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED); - i = fxdr_unsigned(int, *tl); - if (i) { - tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED); - fhsize = fxdr_unsigned(int, *tl); - nfsm_adv(nfsm_rndup(fhsize)); - } - } - if (newvp != NULLVP) { - if (newvp == vp) - vrele(newvp); - else - vput(newvp); - newvp = NULLVP; - } - tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED); - more_dirs = fxdr_unsigned(int, *tl); - } - /* - * If at end of rpc data, get the eof boolean - */ - if (!more_dirs) { - tl = nfsm_dissect(u_int32_t *, NFSX_UNSIGNED); - more_dirs = (fxdr_unsigned(int, *tl) == 0); - } - m_freem(mrep); - } - /* - * Fill last record, iff any, out to a multiple of DIRBLKSIZ - * by increasing d_reclen for the last record. - */ - if (blksiz > 0) { - left = DIRBLKSIZ - blksiz; - dp->d_reclen += left; - uiop->uio_iov->iov_base = - (char *)uiop->uio_iov->iov_base + left; - uiop->uio_iov->iov_len -= left; - uiop->uio_offset += left; - uiop->uio_resid -= left; - } - - /* - * We are now either at the end of the directory or have filled the - * block. - */ - if (bigenough) - dnp->n_direofoffset = uiop->uio_offset; - else { - if (uiop->uio_resid > 0) - nfs_printf("EEK! readdirplusrpc resid > 0\n"); - nfs_dircookie_lock(dnp); - cookiep = nfs_getcookie(dnp, uiop->uio_offset, 1); - *cookiep = cookie; - nfs_dircookie_unlock(dnp); - } -nfsmout: - if (newvp != NULLVP) { - if (newvp == vp) - vrele(newvp); - else - vput(newvp); - newvp = NULLVP; - } - return (error); -} - -/* - * Silly rename. To make the NFS filesystem that is stateless look a little - * more like the "ufs" a remove of an active vnode is translated to a rename - * to a funny looking filename that is removed by nfs_inactive on the - * nfsnode. There is the potential for another process on a different client - * to create the same funny name between the nfs_lookitup() fails and the - * nfs_rename() completes, but... - */ -static int -nfs_sillyrename(struct vnode *dvp, struct vnode *vp, struct componentname *cnp) -{ - struct sillyrename *sp; - struct nfsnode *np; - int error; - short pid; - unsigned int lticks; - - cache_purge(dvp); - np = VTONFS(vp); - KASSERT(vp->v_type != VDIR, ("nfs: sillyrename dir")); - sp = malloc(sizeof (struct sillyrename), - M_NFSREQ, M_WAITOK); - sp->s_cred = crhold(cnp->cn_cred); - sp->s_dvp = dvp; - sp->s_removeit = nfs_removeit; - VREF(dvp); - - /* - * Fudge together a funny name. - * Changing the format of the funny name to accomodate more - * sillynames per directory. - * The name is now changed to .nfs.<ticks>.<pid>.4, where ticks is - * CPU ticks since boot. - */ - pid = cnp->cn_thread->td_proc->p_pid; - lticks = (unsigned int)ticks; - for ( ; ; ) { - sp->s_namlen = sprintf(sp->s_name, - ".nfs.%08x.%04x4.4", lticks, - pid); - if (nfs_lookitup(dvp, sp->s_name, sp->s_namlen, sp->s_cred, - cnp->cn_thread, NULL)) - break; - lticks++; - } - error = nfs_renameit(dvp, cnp, sp); - if (error) - goto bad; - error = nfs_lookitup(dvp, sp->s_name, sp->s_namlen, sp->s_cred, - cnp->cn_thread, &np); - np->n_sillyrename = sp; - return (0); -bad: - vrele(sp->s_dvp); - crfree(sp->s_cred); - free((caddr_t)sp, M_NFSREQ); - return (error); -} - -/* - * Look up a file name and optionally either update the file handle or - * allocate an nfsnode, depending on the value of npp. - * npp == NULL --> just do the lookup - * *npp == NULL --> allocate a new nfsnode and make sure attributes are - * handled too - * *npp != NULL --> update the file handle in the vnode - */ -static int -nfs_lookitup(struct vnode *dvp, const char *name, int len, struct ucred *cred, - struct thread *td, struct nfsnode **npp) -{ - struct vnode *newvp = NULL; - struct nfsnode *np, *dnp = VTONFS(dvp); - caddr_t bpos, dpos; - int error = 0, fhlen, attrflag; - struct mbuf *mreq, *mrep, *md, *mb; - nfsfh_t *nfhp; - int v3 = NFS_ISV3(dvp); - - nfsstats.rpccnt[NFSPROC_LOOKUP]++; - mreq = m_get2(NFSX_FH(v3) + NFSX_UNSIGNED + nfsm_rndup(len), - M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(dvp, v3); - nfsm_strtom(name, len, NFS_MAXNAMLEN); - nfsm_request(dvp, NFSPROC_LOOKUP, td, cred); - if (npp && !error) { - nfsm_getfh(nfhp, fhlen, v3); - if (*npp) { - np = *npp; - if (np->n_fhsize > NFS_SMALLFH && fhlen <= NFS_SMALLFH) { - free((caddr_t)np->n_fhp, M_NFSBIGFH); - np->n_fhp = &np->n_fh; - } else if (np->n_fhsize <= NFS_SMALLFH && fhlen>NFS_SMALLFH) - np->n_fhp =(nfsfh_t *)malloc(fhlen, M_NFSBIGFH, M_WAITOK); - bcopy((caddr_t)nfhp, (caddr_t)np->n_fhp, fhlen); - np->n_fhsize = fhlen; - newvp = NFSTOV(np); - } else if (NFS_CMPFH(dnp, nfhp, fhlen)) { - VREF(dvp); - newvp = dvp; - } else { - error = nfs_nget(dvp->v_mount, nfhp, fhlen, &np, LK_EXCLUSIVE); - if (error) { - m_freem(mrep); - return (error); - } - newvp = NFSTOV(np); - } - if (v3) { - nfsm_postop_attr(newvp, attrflag); - if (!attrflag && *npp == NULL) { - m_freem(mrep); - if (newvp == dvp) - vrele(newvp); - else - vput(newvp); - return (ENOENT); - } - } else - nfsm_loadattr(newvp, NULL); - } - m_freem(mrep); -nfsmout: - if (npp && *npp == NULL) { - if (error) { - if (newvp) { - if (newvp == dvp) - vrele(newvp); - else - vput(newvp); - } - } else - *npp = np; - } - return (error); -} - -/* - * Nfs Version 3 commit rpc - */ -int -nfs_commit(struct vnode *vp, u_quad_t offset, int cnt, struct ucred *cred, - struct thread *td) -{ - u_int32_t *tl; - struct nfsmount *nmp = VFSTONFS(vp->v_mount); - caddr_t bpos, dpos; - int error = 0, wccflag = NFSV3_WCCRATTR; - struct mbuf *mreq, *mrep, *md, *mb; - - mtx_lock(&nmp->nm_mtx); - if ((nmp->nm_state & NFSSTA_HASWRITEVERF) == 0) { - mtx_unlock(&nmp->nm_mtx); - return (0); - } - mtx_unlock(&nmp->nm_mtx); - nfsstats.rpccnt[NFSPROC_COMMIT]++; - mreq = m_get2(NFSX_FH(1), M_WAITOK, MT_DATA, 0); - mb = mreq; - bpos = mtod(mb, caddr_t); - nfsm_fhtom(vp, 1); - tl = nfsm_build(u_int32_t *, 3 * NFSX_UNSIGNED); - txdr_hyper(offset, tl); - tl += 2; - *tl = txdr_unsigned(cnt); - nfsm_request(vp, NFSPROC_COMMIT, td, cred); - nfsm_wcc_data(vp, wccflag); - if (!error) { - tl = nfsm_dissect(u_int32_t *, NFSX_V3WRITEVERF); - if (bcmp((caddr_t)nmp->nm_verf, (caddr_t)tl, - NFSX_V3WRITEVERF)) { - bcopy((caddr_t)tl, (caddr_t)nmp->nm_verf, - NFSX_V3WRITEVERF); - error = NFSERR_STALEWRITEVERF; - } - } - m_freem(mrep); -nfsmout: - return (error); -} - -/* - * Strategy routine. - * For async requests when nfsiod(s) are running, queue the request by - * calling nfs_asyncio(), otherwise just all nfs_doio() to do the - * request. - */ -static int -nfs_strategy(struct vop_strategy_args *ap) -{ - struct buf *bp = ap->a_bp; - struct ucred *cr; - - KASSERT(!(bp->b_flags & B_DONE), - ("nfs_strategy: buffer %p unexpectedly marked B_DONE", bp)); - BUF_ASSERT_HELD(bp); - - if (bp->b_iocmd == BIO_READ) - cr = bp->b_rcred; - else - cr = bp->b_wcred; - - /* - * If the op is asynchronous and an i/o daemon is waiting - * queue the request, wake it up and wait for completion - * otherwise just do it ourselves. - */ - if ((bp->b_flags & B_ASYNC) == 0 || - nfs_asyncio(VFSTONFS(ap->a_vp->v_mount), bp, NOCRED, curthread)) - (void)nfs_doio(ap->a_vp, bp, cr, curthread); - return (0); -} - -/* - * fsync vnode op. Just call nfs_flush() with commit == 1. - */ -/* ARGSUSED */ -static int -nfs_fsync(struct vop_fsync_args *ap) -{ - - return (nfs_flush(ap->a_vp, ap->a_waitfor, 1)); -} - -/* - * Flush all the blocks associated with a vnode. - * Walk through the buffer pool and push any dirty pages - * associated with the vnode. - */ -static int -nfs_flush(struct vnode *vp, int waitfor, int commit) -{ - struct nfsnode *np = VTONFS(vp); - struct buf *bp; - int i; - struct buf *nbp; - struct nfsmount *nmp = VFSTONFS(vp->v_mount); - int error = 0, slptimeo = 0, slpflag = 0, retv, bvecpos; - int passone = 1; - u_quad_t off, endoff, toff; - struct ucred* wcred = NULL; - struct buf **bvec = NULL; - struct bufobj *bo; - struct thread *td = curthread; -#ifndef NFS_COMMITBVECSIZ -#define NFS_COMMITBVECSIZ 20 -#endif - struct buf *bvec_on_stack[NFS_COMMITBVECSIZ]; - int bvecsize = 0, bveccount; - - if (nmp->nm_flag & NFSMNT_INT) - slpflag = PCATCH; - if (!commit) - passone = 0; - bo = &vp->v_bufobj; - /* - * A b_flags == (B_DELWRI | B_NEEDCOMMIT) block has been written to the - * server, but has not been committed to stable storage on the server - * yet. On the first pass, the byte range is worked out and the commit - * rpc is done. On the second pass, nfs_writebp() is called to do the - * job. - */ -again: - off = (u_quad_t)-1; - endoff = 0; - bvecpos = 0; - if (NFS_ISV3(vp) && commit) { - if (bvec != NULL && bvec != bvec_on_stack) - free(bvec, M_TEMP); - /* - * Count up how many buffers waiting for a commit. - */ - bveccount = 0; - BO_LOCK(bo); - TAILQ_FOREACH_SAFE(bp, &bo->bo_dirty.bv_hd, b_bobufs, nbp) { - if (!BUF_ISLOCKED(bp) && - (bp->b_flags & (B_DELWRI | B_NEEDCOMMIT)) - == (B_DELWRI | B_NEEDCOMMIT)) - bveccount++; - } - /* - * Allocate space to remember the list of bufs to commit. It is - * important to use M_NOWAIT here to avoid a race with nfs_write. - * If we can't get memory (for whatever reason), we will end up - * committing the buffers one-by-one in the loop below. - */ - if (bveccount > NFS_COMMITBVECSIZ) { - /* - * Release the vnode interlock to avoid a lock - * order reversal. - */ - BO_UNLOCK(bo); - bvec = (struct buf **) - malloc(bveccount * sizeof(struct buf *), - M_TEMP, M_NOWAIT); - BO_LOCK(bo); - if (bvec == NULL) { - bvec = bvec_on_stack; - bvecsize = NFS_COMMITBVECSIZ; - } else - bvecsize = bveccount; - } else { - bvec = bvec_on_stack; - bvecsize = NFS_COMMITBVECSIZ; - } - TAILQ_FOREACH_SAFE(bp, &bo->bo_dirty.bv_hd, b_bobufs, nbp) { - if (bvecpos >= bvecsize) - break; - if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_NOWAIT, NULL)) { - nbp = TAILQ_NEXT(bp, b_bobufs); - continue; - } - if ((bp->b_flags & (B_DELWRI | B_NEEDCOMMIT)) != - (B_DELWRI | B_NEEDCOMMIT)) { - BUF_UNLOCK(bp); - nbp = TAILQ_NEXT(bp, b_bobufs); - continue; - } - BO_UNLOCK(bo); - bremfree(bp); - /* - * Work out if all buffers are using the same cred - * so we can deal with them all with one commit. - * - * NOTE: we are not clearing B_DONE here, so we have - * to do it later on in this routine if we intend to - * initiate I/O on the bp. - * - * Note: to avoid loopback deadlocks, we do not - * assign b_runningbufspace. - */ - if (wcred == NULL) - wcred = bp->b_wcred; - else if (wcred != bp->b_wcred) - wcred = NOCRED; - vfs_busy_pages(bp, 1); - - BO_LOCK(bo); - /* - * bp is protected by being locked, but nbp is not - * and vfs_busy_pages() may sleep. We have to - * recalculate nbp. - */ - nbp = TAILQ_NEXT(bp, b_bobufs); - - /* - * A list of these buffers is kept so that the - * second loop knows which buffers have actually - * been committed. This is necessary, since there - * may be a race between the commit rpc and new - * uncommitted writes on the file. - */ - bvec[bvecpos++] = bp; - toff = ((u_quad_t)bp->b_blkno) * DEV_BSIZE + - bp->b_dirtyoff; - if (toff < off) - off = toff; - toff += (u_quad_t)(bp->b_dirtyend - bp->b_dirtyoff); - if (toff > endoff) - endoff = toff; - } - BO_UNLOCK(bo); - } - if (bvecpos > 0) { - /* - * Commit data on the server, as required. - * If all bufs are using the same wcred, then use that with - * one call for all of them, otherwise commit each one - * separately. - */ - if (wcred != NOCRED) - retv = nfs_commit(vp, off, (int)(endoff - off), - wcred, td); - else { - retv = 0; - for (i = 0; i < bvecpos; i++) { - off_t off, size; - bp = bvec[i]; - off = ((u_quad_t)bp->b_blkno) * DEV_BSIZE + - bp->b_dirtyoff; - size = (u_quad_t)(bp->b_dirtyend - - bp->b_dirtyoff); - retv = nfs_commit(vp, off, (int)size, - bp->b_wcred, td); - if (retv) break; - } - } - - if (retv == NFSERR_STALEWRITEVERF) - nfs_clearcommit(vp->v_mount); - - /* - * Now, either mark the blocks I/O done or mark the - * blocks dirty, depending on whether the commit - * succeeded. - */ - for (i = 0; i < bvecpos; i++) { - bp = bvec[i]; - bp->b_flags &= ~(B_NEEDCOMMIT | B_CLUSTEROK); - if (retv) { - /* - * Error, leave B_DELWRI intact - */ - vfs_unbusy_pages(bp); - brelse(bp); - } else { - /* - * Success, remove B_DELWRI ( bundirty() ). - * - * b_dirtyoff/b_dirtyend seem to be NFS - * specific. We should probably move that - * into bundirty(). XXX - */ - bufobj_wref(bo); - bp->b_flags |= B_ASYNC; - bundirty(bp); - bp->b_flags &= ~B_DONE; - bp->b_ioflags &= ~BIO_ERROR; - bp->b_dirtyoff = bp->b_dirtyend = 0; - bufdone(bp); - } - } - } - - /* - * Start/do any write(s) that are required. - */ -loop: - BO_LOCK(bo); - TAILQ_FOREACH_SAFE(bp, &bo->bo_dirty.bv_hd, b_bobufs, nbp) { - if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_NOWAIT, NULL)) { - if (waitfor != MNT_WAIT || passone) - continue; - - error = BUF_TIMELOCK(bp, - LK_EXCLUSIVE | LK_SLEEPFAIL | LK_INTERLOCK, - BO_LOCKPTR(bo), "nfsfsync", slpflag, slptimeo); - if (error == 0) { - BUF_UNLOCK(bp); - goto loop; - } - if (error == ENOLCK) { - error = 0; - goto loop; - } - if (nfs_sigintr(nmp, td)) { - error = EINTR; - goto done; - } - if (slpflag == PCATCH) { - slpflag = 0; - slptimeo = 2 * hz; - } - goto loop; - } - if ((bp->b_flags & B_DELWRI) == 0) - panic("nfs_fsync: not dirty"); - if ((passone || !commit) && (bp->b_flags & B_NEEDCOMMIT)) { - BUF_UNLOCK(bp); - continue; - } - BO_UNLOCK(bo); - bremfree(bp); - if (passone || !commit) - bp->b_flags |= B_ASYNC; - else - bp->b_flags |= B_ASYNC; - bwrite(bp); - if (nfs_sigintr(nmp, td)) { - error = EINTR; - goto done; - } - goto loop; - } - if (passone) { - passone = 0; - BO_UNLOCK(bo); - goto again; - } - if (waitfor == MNT_WAIT) { - while (bo->bo_numoutput) { - error = bufobj_wwait(bo, slpflag, slptimeo); - if (error) { - BO_UNLOCK(bo); - error = nfs_sigintr(nmp, td); - if (error) - goto done; - if (slpflag == PCATCH) { - slpflag = 0; - slptimeo = 2 * hz; - } - BO_LOCK(bo); - } - } - if (bo->bo_dirty.bv_cnt != 0 && commit) { - BO_UNLOCK(bo); - goto loop; - } - /* - * Wait for all the async IO requests to drain - */ - BO_UNLOCK(bo); - mtx_lock(&np->n_mtx); - while (np->n_directio_asyncwr > 0) { - np->n_flag |= NFSYNCWAIT; - error = nfs_msleep(td, (caddr_t)&np->n_directio_asyncwr, - &np->n_mtx, slpflag | (PRIBIO + 1), - "nfsfsync", 0); - if (error) { - if (nfs_sigintr(nmp, td)) { - mtx_unlock(&np->n_mtx); - error = EINTR; - goto done; - } - } - } - mtx_unlock(&np->n_mtx); - } else - BO_UNLOCK(bo); - mtx_lock(&np->n_mtx); - if (np->n_flag & NWRITEERR) { - error = np->n_error; - np->n_flag &= ~NWRITEERR; - } - if (commit && bo->bo_dirty.bv_cnt == 0 && - bo->bo_numoutput == 0 && np->n_directio_asyncwr == 0) - np->n_flag &= ~NMODIFIED; - mtx_unlock(&np->n_mtx); -done: - if (bvec != NULL && bvec != bvec_on_stack) - free(bvec, M_TEMP); - return (error); -} - -/* - * NFS advisory byte-level locks. - */ -static int -nfs_advlock(struct vop_advlock_args *ap) -{ - struct vnode *vp = ap->a_vp; - u_quad_t size; - int error; - - error = vn_lock(vp, LK_SHARED); - if (error) - return (error); - if ((VFSTONFS(vp->v_mount)->nm_flag & NFSMNT_NOLOCKD) != 0) { - size = VTONFS(vp)->n_size; - VOP_UNLOCK(vp, 0); - error = lf_advlock(ap, &(vp->v_lockf), size); - } else { - if (nfs_advlock_p) - error = nfs_advlock_p(ap); - else - error = ENOLCK; - } - - return (error); -} - -/* - * NFS advisory byte-level locks. - */ -static int -nfs_advlockasync(struct vop_advlockasync_args *ap) -{ - struct vnode *vp = ap->a_vp; - u_quad_t size; - int error; - - error = vn_lock(vp, LK_SHARED); - if (error) - return (error); - if ((VFSTONFS(vp->v_mount)->nm_flag & NFSMNT_NOLOCKD) != 0) { - size = VTONFS(vp)->n_size; - VOP_UNLOCK(vp, 0); - error = lf_advlockasync(ap, &(vp->v_lockf), size); - } else { - VOP_UNLOCK(vp, 0); - error = EOPNOTSUPP; - } - return (error); -} - -/* - * Print out the contents of an nfsnode. - */ -static int -nfs_print(struct vop_print_args *ap) -{ - struct vnode *vp = ap->a_vp; - struct nfsnode *np = VTONFS(vp); - - nfs_printf("\tfileid %ld fsid 0x%x", - np->n_vattr.va_fileid, np->n_vattr.va_fsid); - if (vp->v_type == VFIFO) - fifo_printinfo(vp); - printf("\n"); - return (0); -} - -/* - * This is the "real" nfs::bwrite(struct buf*). - * We set B_CACHE if this is a VMIO buffer. - */ -int -nfs_writebp(struct buf *bp, int force __unused, struct thread *td) -{ - int s; - int oldflags = bp->b_flags; -#if 0 - int retv = 1; - off_t off; -#endif - - BUF_ASSERT_HELD(bp); - - if (bp->b_flags & B_INVAL) { - brelse(bp); - return(0); - } - - bp->b_flags |= B_CACHE; - - /* - * Undirty the bp. We will redirty it later if the I/O fails. - */ - - s = splbio(); - bundirty(bp); - bp->b_flags &= ~B_DONE; - bp->b_ioflags &= ~BIO_ERROR; - bp->b_iocmd = BIO_WRITE; - - bufobj_wref(bp->b_bufobj); - curthread->td_ru.ru_oublock++; - splx(s); - - /* - * Note: to avoid loopback deadlocks, we do not - * assign b_runningbufspace. - */ - vfs_busy_pages(bp, 1); - - BUF_KERNPROC(bp); - bp->b_iooffset = dbtob(bp->b_blkno); - bstrategy(bp); - - if( (oldflags & B_ASYNC) == 0) { - int rtval = bufwait(bp); - - if (oldflags & B_DELWRI) { - s = splbio(); - reassignbuf(bp); - splx(s); - } - brelse(bp); - return (rtval); - } - - return (0); -} - -/* - * nfs special file access vnode op. - * Essentially just get vattr and then imitate iaccess() since the device is - * local to the client. - */ -static int -nfsspec_access(struct vop_access_args *ap) -{ - struct vattr *vap; - struct ucred *cred = ap->a_cred; - struct vnode *vp = ap->a_vp; - accmode_t accmode = ap->a_accmode; - struct vattr vattr; - int error; - - /* - * Disallow write attempts on filesystems mounted read-only; - * unless the file is a socket, fifo, or a block or character - * device resident on the filesystem. - */ - if ((accmode & VWRITE) && (vp->v_mount->mnt_flag & MNT_RDONLY)) { - switch (vp->v_type) { - case VREG: - case VDIR: - case VLNK: - return (EROFS); - default: - break; - } - } - vap = &vattr; - error = VOP_GETATTR(vp, vap, cred); - if (error) - goto out; - error = vaccess(vp->v_type, vap->va_mode, vap->va_uid, vap->va_gid, - accmode, cred, NULL); -out: - return error; -} - -/* - * Read wrapper for fifos. - */ -static int -nfsfifo_read(struct vop_read_args *ap) -{ - struct nfsnode *np = VTONFS(ap->a_vp); - int error; - - /* - * Set access flag. - */ - mtx_lock(&np->n_mtx); - np->n_flag |= NACC; - vfs_timestamp(&np->n_atim); - mtx_unlock(&np->n_mtx); - error = fifo_specops.vop_read(ap); - return error; -} - -/* - * Write wrapper for fifos. - */ -static int -nfsfifo_write(struct vop_write_args *ap) -{ - struct nfsnode *np = VTONFS(ap->a_vp); - - /* - * Set update flag. - */ - mtx_lock(&np->n_mtx); - np->n_flag |= NUPD; - vfs_timestamp(&np->n_mtim); - mtx_unlock(&np->n_mtx); - return(fifo_specops.vop_write(ap)); -} - -/* - * Close wrapper for fifos. - * - * Update the times on the nfsnode then do fifo close. - */ -static int -nfsfifo_close(struct vop_close_args *ap) -{ - struct vnode *vp = ap->a_vp; - struct nfsnode *np = VTONFS(vp); - struct vattr vattr; - struct timespec ts; - - mtx_lock(&np->n_mtx); - if (np->n_flag & (NACC | NUPD)) { - vfs_timestamp(&ts); - if (np->n_flag & NACC) - np->n_atim = ts; - if (np->n_flag & NUPD) - np->n_mtim = ts; - np->n_flag |= NCHG; - if (vrefcnt(vp) == 1 && - (vp->v_mount->mnt_flag & MNT_RDONLY) == 0) { - VATTR_NULL(&vattr); - if (np->n_flag & NACC) - vattr.va_atime = np->n_atim; - if (np->n_flag & NUPD) - vattr.va_mtime = np->n_mtim; - mtx_unlock(&np->n_mtx); - (void)VOP_SETATTR(vp, &vattr, ap->a_cred); - goto out; - } - } - mtx_unlock(&np->n_mtx); -out: - return (fifo_specops.vop_close(ap)); -} - -/* - * Just call nfs_writebp() with the force argument set to 1. - * - * NOTE: B_DONE may or may not be set in a_bp on call. - */ -static int -nfs_bwrite(struct buf *bp) -{ - - return (nfs_writebp(bp, 1, curthread)); -} - -struct buf_ops buf_ops_nfs = { - .bop_name = "buf_ops_nfs", - .bop_write = nfs_bwrite, - .bop_strategy = bufstrategy, - .bop_sync = bufsync, - .bop_bdflush = bufbdflush, -}; diff --git a/sys/nfsserver/nfs_fha_old.c b/sys/nfsserver/nfs_fha_old.c deleted file mode 100644 index a928278..0000000 --- a/sys/nfsserver/nfs_fha_old.c +++ /dev/null @@ -1,265 +0,0 @@ -/*- - * Copyright (c) 2008 Isilon Inc http://www.isilon.com/ - * Copyright (c) 2013 Spectra Logic Corporation - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * - * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/sysproto.h> -#include <sys/kernel.h> -#include <sys/vnode.h> -#include <sys/malloc.h> -#include <sys/mount.h> -#include <sys/mbuf.h> -#include <sys/sysctl.h> - -#include <rpc/rpc.h> -#include <nfs/xdr_subs.h> -#include <nfs/nfsproto.h> -#include <nfs/nfs_fha.h> -#include <nfsserver/nfs.h> -#include <nfsserver/nfsm_subs.h> -#include <nfsserver/nfs_fha_old.h> - -static void fhaold_init(void *foo); -static void fhaold_uninit(void *foo); -rpcproc_t fhaold_get_procnum(rpcproc_t procnum); -int fhaold_realign(struct mbuf **mb, int malloc_flags); -int fhaold_get_fh(uint64_t *fh, int v3, struct mbuf **md, caddr_t *dpos); -int fhaold_is_read(rpcproc_t procnum); -int fhaold_is_write(rpcproc_t procnum); -int fhaold_get_offset(struct mbuf **md, caddr_t *dpos, int v3, - struct fha_info *info); -int fhaold_no_offset(rpcproc_t procnum); -void fhaold_set_locktype(rpcproc_t procnum, struct fha_info *info); -static int fheold_stats_sysctl(SYSCTL_HANDLER_ARGS); - -static struct fha_params fhaold_softc; - -SYSCTL_DECL(_vfs_nfsrv); - -extern SVCPOOL *nfsrv_pool; - -SYSINIT(nfs_fhaold, SI_SUB_ROOT_CONF, SI_ORDER_ANY, fhaold_init, NULL); -SYSUNINIT(nfs_fhaold, SI_SUB_ROOT_CONF, SI_ORDER_ANY, fhaold_uninit, NULL); - -static void -fhaold_init(void *foo) -{ - struct fha_params *softc; - - softc = &fhaold_softc; - - bzero(softc, sizeof(*softc)); - - /* - * Setup the callbacks for this FHA personality. - */ - softc->callbacks.get_procnum = fhaold_get_procnum; - softc->callbacks.realign = fhaold_realign; - softc->callbacks.get_fh = fhaold_get_fh; - softc->callbacks.is_read = fhaold_is_read; - softc->callbacks.is_write = fhaold_is_write; - softc->callbacks.get_offset = fhaold_get_offset; - softc->callbacks.no_offset = fhaold_no_offset; - softc->callbacks.set_locktype = fhaold_set_locktype; - softc->callbacks.fhe_stats_sysctl = fheold_stats_sysctl; - - snprintf(softc->server_name, sizeof(softc->server_name), - FHAOLD_SERVER_NAME); - - softc->pool = &nfsrv_pool; - - /* - * Initialize the sysctl context list for the fha module. - */ - sysctl_ctx_init(&softc->sysctl_ctx); - softc->sysctl_tree = SYSCTL_ADD_NODE(&softc->sysctl_ctx, - SYSCTL_STATIC_CHILDREN(_vfs_nfsrv), OID_AUTO, "fha", CTLFLAG_RD, - 0, "fha node"); - if (softc->sysctl_tree == NULL) { - printf("%s: unable to allocate sysctl tree\n", __func__); - return; - } - fha_init(softc); -} - -static void -fhaold_uninit(void *foo) -{ - struct fha_params *softc; - - softc = &fhaold_softc; - - fha_uninit(softc); -} - - -rpcproc_t -fhaold_get_procnum(rpcproc_t procnum) -{ - if (procnum > NFSV2PROC_STATFS) - return (-1); - - return (nfsrv_nfsv3_procid[procnum]); -} - -int -fhaold_realign(struct mbuf **mb, int malloc_flags) -{ - return (nfs_realign(mb, malloc_flags)); -} - -int -fhaold_get_fh(uint64_t *fh, int v3, struct mbuf **md, caddr_t *dpos) -{ - u_int32_t *tl; - uint8_t *buf; - uint64_t t; - int fhlen, i; - - if (v3) { - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - fhlen = fxdr_unsigned(int, *tl); - if (fhlen != 0 && fhlen != NFSX_V3FH) - return EBADRPC; - } else { - fhlen = NFSX_V2FH; - } - t = 0; - if (fhlen != 0) { - buf = nfsm_dissect_xx_nonblock(fhlen, md, dpos); - if (buf == NULL) - return EBADRPC; - for (i = 0; i < fhlen; i++) - t ^= ((uint64_t)buf[i] << (i & 7) * 8); - } - *fh = t; - return 0; -} - -int -fhaold_is_read(rpcproc_t procnum) -{ - if (procnum == NFSPROC_READ) - return (1); - else - return (0); -} - -int -fhaold_is_write(rpcproc_t procnum) -{ - if (procnum == NFSPROC_WRITE) - return (1); - else - return (0); -} - -int -fhaold_get_offset(struct mbuf **md, caddr_t *dpos, int v3, - struct fha_info *info) -{ - uint32_t *tl; - - if (v3) { - tl = nfsm_dissect_xx_nonblock(2 * NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - goto out; - info->offset = fxdr_hyper(tl); - } else { - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - goto out; - info->offset = fxdr_unsigned(uint32_t, *tl); - } - - return (0); -out: - return (-1); -} - -int -fhaold_no_offset(rpcproc_t procnum) -{ - if (procnum == NFSPROC_FSSTAT || - procnum == NFSPROC_FSINFO || - procnum == NFSPROC_PATHCONF || - procnum == NFSPROC_NOOP || - procnum == NFSPROC_NULL) - return (1); - else - return (0); -} - -void -fhaold_set_locktype(rpcproc_t procnum, struct fha_info *info) -{ - switch (procnum) { - case NFSPROC_NULL: - case NFSPROC_GETATTR: - case NFSPROC_LOOKUP: - case NFSPROC_ACCESS: - case NFSPROC_READLINK: - case NFSPROC_READ: - case NFSPROC_READDIR: - case NFSPROC_READDIRPLUS: - case NFSPROC_WRITE: - info->locktype = LK_SHARED; - break; - case NFSPROC_SETATTR: - case NFSPROC_CREATE: - case NFSPROC_MKDIR: - case NFSPROC_SYMLINK: - case NFSPROC_MKNOD: - case NFSPROC_REMOVE: - case NFSPROC_RMDIR: - case NFSPROC_RENAME: - case NFSPROC_LINK: - case NFSPROC_FSSTAT: - case NFSPROC_FSINFO: - case NFSPROC_PATHCONF: - case NFSPROC_COMMIT: - case NFSPROC_NOOP: - info->locktype = LK_EXCLUSIVE; - break; - } -} - -static int -fheold_stats_sysctl(SYSCTL_HANDLER_ARGS) -{ - return (fhe_stats_sysctl(oidp, arg1, arg2, req, &fhaold_softc)); -} - -SVCTHREAD * -fhaold_assign(SVCTHREAD *this_thread, struct svc_req *req) -{ - return (fha_assign(this_thread, req, &fhaold_softc)); -} diff --git a/sys/nfsserver/nfs_serv.c b/sys/nfsserver/nfs_serv.c deleted file mode 100644 index 32fd3f5..0000000 --- a/sys/nfsserver/nfs_serv.c +++ /dev/null @@ -1,3767 +0,0 @@ -/*- - * Copyright (c) 1989, 1993 - * The Regents of the University of California. All rights reserved. - * - * This code is derived from software contributed to Berkeley by - * Rick Macklem at The University of Guelph. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 4. Neither the name of the University nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * @(#)nfs_serv.c 8.8 (Berkeley) 7/31/95 - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -/* - * nfs version 2 and 3 server calls to vnode ops - * - these routines generally have 3 phases - * 1 - break down and validate rpc request in mbuf list - * 2 - do the vnode ops for the request - * (surprisingly ?? many are very similar to syscalls in vfs_syscalls.c) - * 3 - build the rpc reply in an mbuf list - * nb: - * - do not mix the phases, since the nfsm_?? macros can return failures - * on a bad rpc or similar and do not do any vrele() or vput()'s - * - * - the nfsm_reply() macro generates an nfs rpc reply with the nfs - * error number iff error != 0 whereas - * returning an error from the server function implies a fatal error - * such as a badly constructed rpc request that should be dropped without - * a reply. - * For nfsm_reply(), the case where error == EBADRPC is treated - * specially; after constructing a reply, it does an immediate - * `goto nfsmout' to avoid getting any V3 post-op status appended. - * - * Other notes: - * Warning: always pay careful attention to resource cleanup on return - * and note that nfsm_*() macros can terminate a procedure on certain - * errors. - * - * lookup() and namei() - * may return garbage in various structural fields/return elements - * if an error is returned, and may garbage up nd.ni_dvp even if no - * error is returned and you did not request LOCKPARENT or WANTPARENT. - * - * We use the ni_cnd.cn_flags 'HASBUF' flag to track whether the name - * buffer has been freed or not. - */ - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/proc.h> -#include <sys/namei.h> -#include <sys/unistd.h> -#include <sys/vnode.h> -#include <sys/mount.h> -#include <sys/socket.h> -#include <sys/socketvar.h> -#include <sys/malloc.h> -#include <sys/mbuf.h> -#include <sys/priv.h> -#include <sys/dirent.h> -#include <sys/stat.h> -#include <sys/kernel.h> -#include <sys/sysctl.h> -#include <sys/bio.h> -#include <sys/buf.h> -#include <sys/rwlock.h> - -#include <vm/vm.h> -#include <vm/vm_extern.h> -#include <vm/vm_object.h> - -#include <nfs/nfsproto.h> -#include <nfsserver/nfs.h> -#include <nfs/xdr_subs.h> -#include <nfsserver/nfsm_subs.h> - -FEATURE(nfsserver, "NFS server"); - -#ifdef NFSRV_DEBUG -#define nfsdbprintf(info) printf info -#else -#define nfsdbprintf(info) -#endif - -#define MAX_COMMIT_COUNT (1024 * 1024) - -#define MAX_REORDERED_RPC 16 -#define NUM_HEURISTIC 1031 -#define NHUSE_INIT 64 -#define NHUSE_INC 16 -#define NHUSE_MAX 2048 - -static struct nfsheur { - struct vnode *nh_vp; /* vp to match (unreferenced pointer) */ - off_t nh_nextoff; /* next offset for sequential detection */ - int nh_use; /* use count for selection */ - int nh_seqcount; /* heuristic */ -} nfsheur[NUM_HEURISTIC]; - -/* Global vars */ - -int nfsrvw_procrastinate = NFS_GATHERDELAY * 1000; -int nfsrvw_procrastinate_v3 = 0; - -static struct timeval nfsver = { 0 }; - -SYSCTL_NODE(_vfs, OID_AUTO, nfsrv, CTLFLAG_RW, 0, "NFS server"); - -static int nfs_async; -static int nfs_commit_blks; -static int nfs_commit_miss; -SYSCTL_INT(_vfs_nfsrv, OID_AUTO, async, CTLFLAG_RW, &nfs_async, 0, - "Tell client that writes were synced even though they were not"); -SYSCTL_INT(_vfs_nfsrv, OID_AUTO, commit_blks, CTLFLAG_RW, &nfs_commit_blks, 0, - "Number of completed commits"); -SYSCTL_INT(_vfs_nfsrv, OID_AUTO, commit_miss, CTLFLAG_RW, &nfs_commit_miss, 0, ""); - -struct nfsrvstats nfsrvstats; -SYSCTL_STRUCT(_vfs_nfsrv, NFS_NFSRVSTATS, nfsrvstats, CTLFLAG_RW, - &nfsrvstats, nfsrvstats, "S,nfsrvstats"); - -static int nfsrv_access(struct vnode *, accmode_t, struct ucred *, - int, int); - -/* - * Clear nameidata fields that are tested in nsfmout cleanup code prior - * to using first nfsm macro (that might jump to the cleanup code). - */ - -static __inline void -ndclear(struct nameidata *nd) -{ - - nd->ni_cnd.cn_flags = 0; - nd->ni_vp = NULL; - nd->ni_dvp = NULL; - nd->ni_startdir = NULL; - nd->ni_strictrelative = 0; -} - -/* - * Heuristic to detect sequential operation. - */ -static struct nfsheur * -nfsrv_sequential_heuristic(struct uio *uio, struct vnode *vp) -{ - struct nfsheur *nh; - int hi, try; - - /* Locate best candidate. */ - try = 32; - hi = ((int)(vm_offset_t)vp / sizeof(struct vnode)) % NUM_HEURISTIC; - nh = &nfsheur[hi]; - while (try--) { - if (nfsheur[hi].nh_vp == vp) { - nh = &nfsheur[hi]; - break; - } - if (nfsheur[hi].nh_use > 0) - --nfsheur[hi].nh_use; - hi = (hi + 1) % NUM_HEURISTIC; - if (nfsheur[hi].nh_use < nh->nh_use) - nh = &nfsheur[hi]; - } - - /* Initialize hint if this is a new file. */ - if (nh->nh_vp != vp) { - nh->nh_vp = vp; - nh->nh_nextoff = uio->uio_offset; - nh->nh_use = NHUSE_INIT; - if (uio->uio_offset == 0) - nh->nh_seqcount = 4; - else - nh->nh_seqcount = 1; - } - - /* Calculate heuristic. */ - if ((uio->uio_offset == 0 && nh->nh_seqcount > 0) || - uio->uio_offset == nh->nh_nextoff) { - /* See comments in vfs_vnops.c:sequential_heuristic(). */ - nh->nh_seqcount += howmany(uio->uio_resid, 16384); - if (nh->nh_seqcount > IO_SEQMAX) - nh->nh_seqcount = IO_SEQMAX; - } else if (qabs(uio->uio_offset - nh->nh_nextoff) <= MAX_REORDERED_RPC * - imax(vp->v_mount->mnt_stat.f_iosize, uio->uio_resid)) { - /* Probably a reordered RPC, leave seqcount alone. */ - } else if (nh->nh_seqcount > 1) { - nh->nh_seqcount /= 2; - } else { - nh->nh_seqcount = 0; - } - nh->nh_use += NHUSE_INC; - if (nh->nh_use > NHUSE_MAX) - nh->nh_use = NHUSE_MAX; - return (nh); -} - -/* - * nfs v3 access service - */ -int -nfsrv3_access(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct vnode *vp = NULL; - nfsfh_t nfh; - fhandle_t *fhp; - u_int32_t *tl; - caddr_t bpos; - int error = 0, rdonly, getret; - struct mbuf *mb, *mreq; - struct vattr vattr, *vap = &vattr; - u_long testmode, nfsmode; - int v3 = (nfsd->nd_flag & ND_NFSV3); - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - if (!v3) - panic("nfsrv3_access: v3 proc called on a v2 connection"); - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - tl = nfsm_dissect_nonblock(u_int32_t *, NFSX_UNSIGNED); - error = nfsrv_fhtovp(fhp, 0, &vp, nfsd, slp, nam, &rdonly); - if (error) { - nfsm_reply(NFSX_UNSIGNED); - nfsm_srvpostop_attr(1, NULL); - error = 0; - goto nfsmout; - } - nfsmode = fxdr_unsigned(u_int32_t, *tl); - if ((nfsmode & NFSV3ACCESS_READ) && - nfsrv_access(vp, VREAD, cred, rdonly, 0)) - nfsmode &= ~NFSV3ACCESS_READ; - if (vp->v_type == VDIR) - testmode = (NFSV3ACCESS_MODIFY | NFSV3ACCESS_EXTEND | - NFSV3ACCESS_DELETE); - else - testmode = (NFSV3ACCESS_MODIFY | NFSV3ACCESS_EXTEND); - if ((nfsmode & testmode) && - nfsrv_access(vp, VWRITE, cred, rdonly, 0)) - nfsmode &= ~testmode; - if (vp->v_type == VDIR) - testmode = NFSV3ACCESS_LOOKUP; - else - testmode = NFSV3ACCESS_EXECUTE; - if ((nfsmode & testmode) && - nfsrv_access(vp, VEXEC, cred, rdonly, 0)) - nfsmode &= ~testmode; - getret = VOP_GETATTR(vp, vap, cred); - vput(vp); - vp = NULL; - nfsm_reply(NFSX_POSTOPATTR(1) + NFSX_UNSIGNED); - nfsm_srvpostop_attr(getret, vap); - tl = nfsm_build(u_int32_t *, NFSX_UNSIGNED); - *tl = txdr_unsigned(nfsmode); -nfsmout: - if (vp) - vput(vp); - return(error); -} - -/* - * nfs getattr service - */ -int -nfsrv_getattr(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct nfs_fattr *fp; - struct vattr va; - struct vattr *vap = &va; - struct vnode *vp = NULL; - nfsfh_t nfh; - fhandle_t *fhp; - caddr_t bpos; - int error = 0, rdonly; - struct mbuf *mb, *mreq; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - error = nfsrv_fhtovp(fhp, 0, &vp, nfsd, slp, nam, &rdonly); - if (error) { - nfsm_reply(0); - error = 0; - goto nfsmout; - } - error = VOP_GETATTR(vp, vap, cred); - vput(vp); - vp = NULL; - nfsm_reply(NFSX_FATTR(nfsd->nd_flag & ND_NFSV3)); - if (error) { - error = 0; - goto nfsmout; - } - fp = nfsm_build(struct nfs_fattr *, - NFSX_FATTR(nfsd->nd_flag & ND_NFSV3)); - nfsm_srvfillattr(vap, fp); - /* fall through */ - -nfsmout: - if (vp) - vput(vp); - return(error); -} - -/* - * nfs setattr service - */ -int -nfsrv_setattr(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct vattr va, preat; - struct vattr *vap = &va; - struct nfsv2_sattr *sp; - struct nfs_fattr *fp; - struct vnode *vp = NULL; - nfsfh_t nfh; - fhandle_t *fhp; - u_int32_t *tl; - caddr_t bpos; - int error = 0, rdonly, preat_ret = 1, postat_ret = 1; - int v3 = (nfsd->nd_flag & ND_NFSV3), gcheck = 0; - struct mbuf *mb, *mreq; - struct timespec guard = { 0, 0 }; - struct mount *mp = NULL; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) { - error = ESTALE; - goto out; - } - (void) vn_start_write(NULL, &mp, V_WAIT); - vfs_rel(mp); /* The write holds a ref. */ - VATTR_NULL(vap); - if (v3) { - nfsm_srvsattr(vap); - tl = nfsm_dissect_nonblock(u_int32_t *, NFSX_UNSIGNED); - gcheck = fxdr_unsigned(int, *tl); - if (gcheck) { - tl = nfsm_dissect_nonblock(u_int32_t *, 2 * NFSX_UNSIGNED); - fxdr_nfsv3time(tl, &guard); - } - } else { - sp = nfsm_dissect_nonblock(struct nfsv2_sattr *, NFSX_V2SATTR); - /* - * Nah nah nah nah na nah - * There is a bug in the Sun client that puts 0xffff in the mode - * field of sattr when it should put in 0xffffffff. The u_short - * doesn't sign extend. - * --> check the low order 2 bytes for 0xffff - */ - if ((fxdr_unsigned(int, sp->sa_mode) & 0xffff) != 0xffff) - vap->va_mode = nfstov_mode(sp->sa_mode); - if (sp->sa_uid != nfsrv_nfs_xdrneg1) - vap->va_uid = fxdr_unsigned(uid_t, sp->sa_uid); - if (sp->sa_gid != nfsrv_nfs_xdrneg1) - vap->va_gid = fxdr_unsigned(gid_t, sp->sa_gid); - if (sp->sa_size != nfsrv_nfs_xdrneg1) - vap->va_size = fxdr_unsigned(u_quad_t, sp->sa_size); - if (sp->sa_atime.nfsv2_sec != nfsrv_nfs_xdrneg1) { -#ifdef notyet - fxdr_nfsv2time(&sp->sa_atime, &vap->va_atime); -#else - vap->va_atime.tv_sec = - fxdr_unsigned(int32_t, sp->sa_atime.nfsv2_sec); - vap->va_atime.tv_nsec = 0; -#endif - } - if (sp->sa_mtime.nfsv2_sec != nfsrv_nfs_xdrneg1) - fxdr_nfsv2time(&sp->sa_mtime, &vap->va_mtime); - - } - - /* - * Now that we have all the fields, lets do it. - */ - error = nfsrv_fhtovp(fhp, 0, &vp, nfsd, slp, nam, &rdonly); - if (error) { - nfsm_reply(2 * NFSX_UNSIGNED); - if (v3) - nfsm_srvwcc_data(preat_ret, &preat, postat_ret, vap); - error = 0; - goto nfsmout; - } - - /* - * vp now an active resource, pay careful attention to cleanup - */ - if (v3) { - error = preat_ret = VOP_GETATTR(vp, &preat, cred); - if (!error && gcheck && - (preat.va_ctime.tv_sec != guard.tv_sec || - preat.va_ctime.tv_nsec != guard.tv_nsec)) - error = NFSERR_NOT_SYNC; - if (error) { - vput(vp); - vp = NULL; - nfsm_reply(NFSX_WCCDATA(v3)); - if (v3) - nfsm_srvwcc_data(preat_ret, &preat, postat_ret, vap); - error = 0; - goto nfsmout; - } - } - - /* - * If the size is being changed write acces is required, otherwise - * just check for a read only filesystem. - */ - if (vap->va_size == ((u_quad_t)((quad_t) -1))) { - if (rdonly || (vp->v_mount->mnt_flag & MNT_RDONLY)) { - error = EROFS; - goto out; - } - } else { - if (vp->v_type == VDIR) { - error = EISDIR; - goto out; - } else if ((error = nfsrv_access(vp, VWRITE, cred, rdonly, - 0)) != 0) - goto out; - } - error = VOP_SETATTR(vp, vap, cred); - postat_ret = VOP_GETATTR(vp, vap, cred); - if (!error) - error = postat_ret; -out: - if (vp != NULL) - vput(vp); - - vp = NULL; - nfsm_reply(NFSX_WCCORFATTR(v3)); - if (v3) { - nfsm_srvwcc_data(preat_ret, &preat, postat_ret, vap); - } else if (!error) { - /* v2 non-error case. */ - fp = nfsm_build(struct nfs_fattr *, NFSX_V2FATTR); - nfsm_srvfillattr(vap, fp); - } - error = 0; - /* fall through */ - -nfsmout: - if (vp) - vput(vp); - vn_finished_write(mp); - return(error); -} - -/* - * nfs lookup rpc - */ -int -nfsrv_lookup(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct nfs_fattr *fp; - struct nameidata nd, ind, *ndp = &nd; - struct vnode *vp, *dirp = NULL; - nfsfh_t nfh; - fhandle_t *fhp; - caddr_t bpos; - int error = 0, len, dirattr_ret = 1; - int v3 = (nfsd->nd_flag & ND_NFSV3), pubflag; - struct mbuf *mb, *mreq; - struct vattr va, dirattr, *vap = &va; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - ndclear(&nd); - - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - nfsm_srvnamesiz(len); - - pubflag = nfs_ispublicfh(fhp); - - nd.ni_cnd.cn_cred = cred; - nd.ni_cnd.cn_nameiop = LOOKUP; - nd.ni_cnd.cn_flags = LOCKLEAF | SAVESTART; - error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos, - &dirp, v3, &dirattr, &dirattr_ret, pubflag); - - /* - * namei failure, only dirp to cleanup. Clear out garbarge from - * structure in case macros jump to nfsmout. - */ - - if (error) { - if (dirp) { - vrele(dirp); - dirp = NULL; - } - nfsm_reply(NFSX_POSTOPATTR(v3)); - if (v3) - nfsm_srvpostop_attr(dirattr_ret, &dirattr); - error = 0; - goto nfsmout; - } - - /* - * Locate index file for public filehandle - * - * error is 0 on entry and 0 on exit from this block. - */ - - if (pubflag) { - if (nd.ni_vp->v_type == VDIR && nfs_pub.np_index != NULL) { - /* - * Setup call to lookup() to see if we can find - * the index file. Arguably, this doesn't belong - * in a kernel.. Ugh. If an error occurs, do not - * try to install an index file and then clear the - * error. - * - * When we replace nd with ind and redirect ndp, - * maintenance of ni_startdir and ni_vp shift to - * ind and we have to clean them up in the old nd. - * However, the cnd resource continues to be maintained - * via the original nd. Confused? You aren't alone! - */ - ind = nd; - VOP_UNLOCK(nd.ni_vp, 0); - ind.ni_pathlen = strlen(nfs_pub.np_index); - ind.ni_cnd.cn_nameptr = ind.ni_cnd.cn_pnbuf = - nfs_pub.np_index; - ind.ni_startdir = nd.ni_vp; - VREF(ind.ni_startdir); - error = lookup(&ind); - ind.ni_dvp = NULL; - - if (error == 0) { - /* - * Found an index file. Get rid of - * the old references. transfer nd.ni_vp' - */ - if (dirp) - vrele(dirp); - dirp = nd.ni_vp; - nd.ni_vp = NULL; - vrele(nd.ni_startdir); - nd.ni_startdir = NULL; - ndp = &ind; - } - error = 0; - } - /* - * If the public filehandle was used, check that this lookup - * didn't result in a filehandle outside the publicly exported - * filesystem. We clear the poor vp here to avoid lockups due - * to NFS I/O. - */ - - if (ndp->ni_vp->v_mount != nfs_pub.np_mount) { - vput(nd.ni_vp); - nd.ni_vp = NULL; - error = EPERM; - } - } - - /* - * Resources at this point: - * ndp->ni_vp may not be NULL - */ - - if (error) { - nfsm_reply(NFSX_POSTOPATTR(v3)); - if (v3) - nfsm_srvpostop_attr(dirattr_ret, &dirattr); - error = 0; - goto nfsmout; - } - - /* - * Get underlying attribute, then release remaining resources ( for - * the same potential blocking reason ) and reply. - */ - vp = ndp->ni_vp; - bzero((caddr_t)fhp, sizeof(nfh)); - fhp->fh_fsid = vp->v_mount->mnt_stat.f_fsid; - error = VOP_VPTOFH(vp, &fhp->fh_fid); - if (!error) - error = VOP_GETATTR(vp, vap, cred); - - vput(vp); - vrele(ndp->ni_startdir); - vrele(dirp); - ndp->ni_vp = NULL; - ndp->ni_startdir = NULL; - dirp = NULL; - nfsm_reply(NFSX_SRVFH(v3) + NFSX_POSTOPORFATTR(v3) + NFSX_POSTOPATTR(v3)); - if (error) { - if (v3) - nfsm_srvpostop_attr(dirattr_ret, &dirattr); - error = 0; - goto nfsmout; - } - nfsm_srvfhtom(fhp, v3); - if (v3) { - nfsm_srvpostop_attr(0, vap); - nfsm_srvpostop_attr(dirattr_ret, &dirattr); - } else { - fp = nfsm_build(struct nfs_fattr *, NFSX_V2FATTR); - nfsm_srvfillattr(vap, fp); - } - -nfsmout: - if (ndp->ni_vp || dirp || ndp->ni_startdir) { - if (ndp->ni_vp) - vput(ndp->ni_vp); - if (dirp) - vrele(dirp); - if (ndp->ni_startdir) - vrele(ndp->ni_startdir); - } - NDFREE(&nd, NDF_ONLY_PNBUF); - return (error); -} - -/* - * nfs readlink service - */ -int -nfsrv_readlink(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct iovec iv[(NFS_MAXPATHLEN+MLEN-1)/MLEN]; - struct iovec *ivp = iv; - struct mbuf *mp; - u_int32_t *tl; - caddr_t bpos; - int error = 0, rdonly, i, tlen, len, getret; - int v3 = (nfsd->nd_flag & ND_NFSV3); - struct mbuf *mb, *mp3, *nmp, *mreq; - struct vnode *vp = NULL; - struct vattr attr; - nfsfh_t nfh; - fhandle_t *fhp; - struct uio io, *uiop = &io; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); -#ifndef nolint - mp = NULL; -#endif - mp3 = NULL; - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - len = 0; - i = 0; - while (len < NFS_MAXPATHLEN) { - MGET(nmp, M_WAITOK, MT_DATA); - MCLGET(nmp, M_WAITOK); - nmp->m_len = NFSMSIZ(nmp); - if (len == 0) - mp3 = mp = nmp; - else { - mp->m_next = nmp; - mp = nmp; - } - if ((len + mp->m_len) > NFS_MAXPATHLEN) { - mp->m_len = NFS_MAXPATHLEN - len; - len = NFS_MAXPATHLEN; - } else - len += mp->m_len; - ivp->iov_base = mtod(mp, caddr_t); - ivp->iov_len = mp->m_len; - i++; - ivp++; - } - uiop->uio_iov = iv; - uiop->uio_iovcnt = i; - uiop->uio_offset = 0; - uiop->uio_resid = len; - uiop->uio_rw = UIO_READ; - uiop->uio_segflg = UIO_SYSSPACE; - uiop->uio_td = NULL; - error = nfsrv_fhtovp(fhp, 0, &vp, nfsd, slp, nam, &rdonly); - if (error) { - nfsm_reply(2 * NFSX_UNSIGNED); - if (v3) - nfsm_srvpostop_attr(1, NULL); - error = 0; - goto nfsmout; - } - if (vp->v_type != VLNK) { - if (v3) - error = EINVAL; - else - error = ENXIO; - } else - error = VOP_READLINK(vp, uiop, cred); - getret = VOP_GETATTR(vp, &attr, cred); - vput(vp); - vp = NULL; - nfsm_reply(NFSX_POSTOPATTR(v3) + NFSX_UNSIGNED); - if (v3) - nfsm_srvpostop_attr(getret, &attr); - if (error) { - error = 0; - goto nfsmout; - } - if (uiop->uio_resid > 0) { - len -= uiop->uio_resid; - tlen = nfsm_rndup(len); - nfsm_adj(mp3, NFS_MAXPATHLEN-tlen, tlen-len); - } - tl = nfsm_build(u_int32_t *, NFSX_UNSIGNED); - *tl = txdr_unsigned(len); - mb->m_next = mp3; - mp3 = NULL; -nfsmout: - if (mp3) - m_freem(mp3); - if (vp) - vput(vp); - return(error); -} - -/* - * nfs read service - */ -int -nfsrv_read(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct iovec *iv; - struct iovec *iv2; - struct mbuf *m; - struct nfs_fattr *fp; - u_int32_t *tl; - int i; - caddr_t bpos; - int error = 0, rdonly, cnt, len, left, siz, tlen, getret; - int v3 = (nfsd->nd_flag & ND_NFSV3), reqlen; - struct mbuf *mb, *mreq; - struct mbuf *m2; - struct vnode *vp = NULL; - nfsfh_t nfh; - fhandle_t *fhp; - struct uio io, *uiop = &io; - struct vattr va, *vap = &va; - struct nfsheur *nh; - off_t off; - int ioflag = 0; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - if (v3) { - tl = nfsm_dissect_nonblock(u_int32_t *, 2 * NFSX_UNSIGNED); - off = fxdr_hyper(tl); - } else { - tl = nfsm_dissect_nonblock(u_int32_t *, NFSX_UNSIGNED); - off = (off_t)fxdr_unsigned(u_int32_t, *tl); - } - nfsm_srvstrsiz(reqlen, NFS_SRVMAXDATA(nfsd)); - - /* - * Reference vp. If an error occurs, vp will be invalid, but we - * have to NULL it just in case. The macros might goto nfsmout - * as well. - */ - - error = nfsrv_fhtovp(fhp, 0, &vp, nfsd, slp, nam, &rdonly); - if (error) { - vp = NULL; - nfsm_reply(2 * NFSX_UNSIGNED); - if (v3) - nfsm_srvpostop_attr(1, NULL); - error = 0; - goto nfsmout; - } - - if (vp->v_type != VREG) { - if (v3) - error = EINVAL; - else - error = (vp->v_type == VDIR) ? EISDIR : EACCES; - } - if (!error) { - if ((error = nfsrv_access(vp, VREAD, cred, rdonly, 1)) != 0) - error = nfsrv_access(vp, VEXEC, cred, rdonly, 1); - } - getret = VOP_GETATTR(vp, vap, cred); - if (!error) - error = getret; - if (error) { - vput(vp); - vp = NULL; - nfsm_reply(NFSX_POSTOPATTR(v3)); - if (v3) - nfsm_srvpostop_attr(getret, vap); - error = 0; - goto nfsmout; - } - - /* - * Calculate byte count to read - */ - if (off >= vap->va_size) - cnt = 0; - else if ((off + reqlen) > vap->va_size) - cnt = vap->va_size - off; - else - cnt = reqlen; - - nfsm_reply(NFSX_POSTOPORFATTR(v3) + 3 * NFSX_UNSIGNED+nfsm_rndup(cnt)); - if (v3) { - tl = nfsm_build(u_int32_t *, NFSX_V3FATTR + 4 * NFSX_UNSIGNED); - *tl++ = nfsrv_nfs_true; - fp = (struct nfs_fattr *)tl; - tl += (NFSX_V3FATTR / sizeof (u_int32_t)); - } else { - tl = nfsm_build(u_int32_t *, NFSX_V2FATTR + NFSX_UNSIGNED); - fp = (struct nfs_fattr *)tl; - tl += (NFSX_V2FATTR / sizeof (u_int32_t)); - } - len = left = nfsm_rndup(cnt); - if (cnt > 0) { - /* - * Generate the mbuf list with the uio_iov ref. to it. - */ - i = 0; - m = m2 = mb; - while (left > 0) { - siz = min(M_TRAILINGSPACE(m), left); - if (siz > 0) { - left -= siz; - i++; - } - if (left > 0) { - MGET(m, M_WAITOK, MT_DATA); - MCLGET(m, M_WAITOK); - m->m_len = 0; - m2->m_next = m; - m2 = m; - } - } - iv = malloc(i * sizeof (struct iovec), - M_TEMP, M_WAITOK); - uiop->uio_iov = iv2 = iv; - m = mb; - left = len; - i = 0; - while (left > 0) { - if (m == NULL) - panic("nfsrv_read iov"); - siz = min(M_TRAILINGSPACE(m), left); - if (siz > 0) { - iv->iov_base = mtod(m, caddr_t) + m->m_len; - iv->iov_len = siz; - m->m_len += siz; - left -= siz; - iv++; - i++; - } - m = m->m_next; - } - uiop->uio_iovcnt = i; - uiop->uio_offset = off; - uiop->uio_resid = len; - uiop->uio_rw = UIO_READ; - uiop->uio_segflg = UIO_SYSSPACE; - uiop->uio_td = NULL; - nh = nfsrv_sequential_heuristic(uiop, vp); - ioflag |= nh->nh_seqcount << IO_SEQSHIFT; - error = VOP_READ(vp, uiop, IO_NODELOCKED | ioflag, cred); - if (error == 0) - nh->nh_nextoff = uiop->uio_offset; - free((caddr_t)iv2, M_TEMP); - if (error || (getret = VOP_GETATTR(vp, vap, cred))) { - if (!error) - error = getret; - m_freem(mreq); - vput(vp); - vp = NULL; - nfsm_reply(NFSX_POSTOPATTR(v3)); - if (v3) - nfsm_srvpostop_attr(getret, vap); - error = 0; - goto nfsmout; - } - } else - uiop->uio_resid = 0; - vput(vp); - vp = NULL; - nfsm_srvfillattr(vap, fp); - tlen = len - uiop->uio_resid; - cnt = cnt < tlen ? cnt : tlen; - tlen = nfsm_rndup(cnt); - if (len != tlen || tlen != cnt) - nfsm_adj(mb, len - tlen, tlen - cnt); - if (v3) { - *tl++ = txdr_unsigned(cnt); - if (cnt < reqlen) - *tl++ = nfsrv_nfs_true; - else - *tl++ = nfsrv_nfs_false; - } - *tl = txdr_unsigned(cnt); -nfsmout: - if (vp) - vput(vp); - return(error); -} - -/* - * nfs write service - */ -int -nfsrv_write(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct iovec *ivp; - int i, cnt; - struct mbuf *mp; - struct nfs_fattr *fp; - struct iovec *iv; - struct vattr va, forat; - struct vattr *vap = &va; - u_int32_t *tl; - caddr_t bpos; - int error = 0, rdonly, len, forat_ret = 1; - int ioflags, aftat_ret = 1, retlen = 0, zeroing, adjust; - int stable = NFSV3WRITE_FILESYNC; - int v3 = (nfsd->nd_flag & ND_NFSV3); - struct mbuf *mb, *mreq; - struct vnode *vp = NULL; - struct nfsheur *nh; - nfsfh_t nfh; - fhandle_t *fhp; - struct uio io, *uiop = &io; - off_t off; - struct mount *mntp = NULL; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - if (mrep == NULL) { - *mrq = NULL; - error = 0; - goto nfsmout; - } - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - if ((mntp = vfs_getvfs(&fhp->fh_fsid)) == NULL) { - error = ESTALE; - goto ereply; - } - (void) vn_start_write(NULL, &mntp, V_WAIT); - vfs_rel(mntp); /* The write holds a ref. */ - if (v3) { - tl = nfsm_dissect_nonblock(u_int32_t *, 5 * NFSX_UNSIGNED); - off = fxdr_hyper(tl); - tl += 3; - stable = fxdr_unsigned(int, *tl++); - } else { - tl = nfsm_dissect_nonblock(u_int32_t *, 4 * NFSX_UNSIGNED); - off = (off_t)fxdr_unsigned(u_int32_t, *++tl); - tl += 2; - if (nfs_async) - stable = NFSV3WRITE_UNSTABLE; - } - retlen = len = fxdr_unsigned(int32_t, *tl); - cnt = i = 0; - - /* - * For NFS Version 2, it is not obvious what a write of zero length - * should do, but I might as well be consistent with Version 3, - * which is to return ok so long as there are no permission problems. - */ - if (len > 0) { - zeroing = 1; - mp = mrep; - while (mp) { - if (mp == md) { - zeroing = 0; - adjust = dpos - mtod(mp, caddr_t); - mp->m_len -= adjust; - if (mp->m_len > 0 && adjust > 0) - mp->m_data += adjust; - } - if (zeroing) - mp->m_len = 0; - else if (mp->m_len > 0) { - i += mp->m_len; - if (i > len) { - mp->m_len -= (i - len); - zeroing = 1; - } - if (mp->m_len > 0) - cnt++; - } - mp = mp->m_next; - } - } - if (len > NFS_MAXDATA || len < 0 || i < len) { - error = EIO; - nfsm_reply(2 * NFSX_UNSIGNED); - if (v3) - nfsm_srvwcc_data(forat_ret, &forat, aftat_ret, vap); - error = 0; - goto nfsmout; - } - error = nfsrv_fhtovp(fhp, 0, &vp, nfsd, slp, nam, &rdonly); - if (error) { - vp = NULL; - nfsm_reply(2 * NFSX_UNSIGNED); - if (v3) - nfsm_srvwcc_data(forat_ret, &forat, aftat_ret, vap); - error = 0; - goto nfsmout; - } - if (v3) - forat_ret = VOP_GETATTR(vp, &forat, cred); - if (vp->v_type != VREG) { - if (v3) - error = EINVAL; - else - error = (vp->v_type == VDIR) ? EISDIR : EACCES; - } - if (!error) - error = nfsrv_access(vp, VWRITE, cred, rdonly, 1); - if (error) { - vput(vp); - vp = NULL; - nfsm_reply(NFSX_WCCDATA(v3)); - if (v3) - nfsm_srvwcc_data(forat_ret, &forat, aftat_ret, vap); - error = 0; - goto nfsmout; - } - - if (len > 0) { - ivp = malloc(cnt * sizeof (struct iovec), M_TEMP, - M_WAITOK); - uiop->uio_iov = iv = ivp; - uiop->uio_iovcnt = cnt; - mp = mrep; - while (mp) { - if (mp->m_len > 0) { - ivp->iov_base = mtod(mp, caddr_t); - ivp->iov_len = mp->m_len; - ivp++; - } - mp = mp->m_next; - } - - /* - * XXX - * The IO_METASYNC flag indicates that all metadata (and not just - * enough to ensure data integrity) mus be written to stable storage - * synchronously. - * (IO_METASYNC is not yet implemented in 4.4BSD-Lite.) - */ - if (stable == NFSV3WRITE_UNSTABLE) - ioflags = IO_NODELOCKED; - else if (stable == NFSV3WRITE_DATASYNC) - ioflags = (IO_SYNC | IO_NODELOCKED); - else - ioflags = (IO_METASYNC | IO_SYNC | IO_NODELOCKED); - uiop->uio_resid = len; - uiop->uio_rw = UIO_WRITE; - uiop->uio_segflg = UIO_SYSSPACE; - uiop->uio_td = NULL; - uiop->uio_offset = off; - nh = nfsrv_sequential_heuristic(uiop, vp); - ioflags |= nh->nh_seqcount << IO_SEQSHIFT; - error = VOP_WRITE(vp, uiop, ioflags, cred); - if (error == 0) - nh->nh_nextoff = uiop->uio_offset; - /* Unlocked write. */ - nfsrvstats.srvvop_writes++; - free((caddr_t)iv, M_TEMP); - } - aftat_ret = VOP_GETATTR(vp, vap, cred); - vput(vp); - vp = NULL; - if (!error) - error = aftat_ret; -ereply: - nfsm_reply(NFSX_PREOPATTR(v3) + NFSX_POSTOPORFATTR(v3) + - 2 * NFSX_UNSIGNED + NFSX_WRITEVERF(v3)); - if (v3) { - nfsm_srvwcc_data(forat_ret, &forat, aftat_ret, vap); - if (error) { - error = 0; - goto nfsmout; - } - tl = nfsm_build(u_int32_t *, 4 * NFSX_UNSIGNED); - *tl++ = txdr_unsigned(retlen); - /* - * If nfs_async is set, then pretend the write was FILESYNC. - */ - if (stable == NFSV3WRITE_UNSTABLE && !nfs_async) - *tl++ = txdr_unsigned(stable); - else - *tl++ = txdr_unsigned(NFSV3WRITE_FILESYNC); - /* - * Actually, there is no need to txdr these fields, - * but it may make the values more human readable, - * for debugging purposes. - */ - if (nfsver.tv_sec == 0) - nfsver = boottime; - *tl++ = txdr_unsigned(nfsver.tv_sec); - *tl = txdr_unsigned(nfsver.tv_usec); - } else if (!error) { - /* v2 non-error case. */ - fp = nfsm_build(struct nfs_fattr *, NFSX_V2FATTR); - nfsm_srvfillattr(vap, fp); - } - error = 0; -nfsmout: - if (vp) - vput(vp); - vn_finished_write(mntp); - return(error); -} - -/* - * nfs create service - * now does a truncate to 0 length via. setattr if it already exists - */ -int -nfsrv_create(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct nfs_fattr *fp; - struct vattr va, dirfor, diraft; - struct vattr *vap = &va; - struct nfsv2_sattr *sp; - u_int32_t *tl; - struct nameidata nd; - caddr_t bpos; - int error = 0, rdev, len, tsize, dirfor_ret = 1, diraft_ret = 1; - int v3 = (nfsd->nd_flag & ND_NFSV3), how, exclusive_flag = 0; - struct mbuf *mb, *mreq; - struct vnode *dirp = NULL; - nfsfh_t nfh; - fhandle_t *fhp; - u_quad_t tempsize; - struct timespec cverf; - struct mount *mp = NULL; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); -#ifndef nolint - rdev = 0; -#endif - ndclear(&nd); - - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) { - error = ESTALE; - goto ereply; - } - (void) vn_start_write(NULL, &mp, V_WAIT); - vfs_rel(mp); /* The write holds a ref. */ - nfsm_srvnamesiz(len); - - nd.ni_cnd.cn_cred = cred; - nd.ni_cnd.cn_nameiop = CREATE; - nd.ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF | SAVESTART | NOCACHE; - - /* - * Call namei and do initial cleanup to get a few things - * out of the way. If we get an initial error we cleanup - * and return here to avoid special-casing the invalid nd - * structure through the rest of the case. dirp may be - * set even if an error occurs, but the nd structure will not - * be valid at all if an error occurs so we have to invalidate it - * prior to calling nfsm_reply ( which might goto nfsmout ). - */ - error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos, - &dirp, v3, &dirfor, &dirfor_ret, FALSE); - if (dirp && !v3) { - vrele(dirp); - dirp = NULL; - } - if (error) { - nfsm_reply(NFSX_WCCDATA(v3)); - if (v3) - nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft); - error = 0; - goto nfsmout; - } - - /* - * No error. Continue. State: - * - * startdir is valid ( we release this immediately ) - * dirp may be valid - * nd.ni_vp may be valid - * nd.ni_dvp is valid - * - * The error state is set through the code and we may also do some - * opportunistic releasing of vnodes to avoid holding locks through - * NFS I/O. The cleanup at the end is a catch-all - */ - - VATTR_NULL(vap); - if (v3) { - tl = nfsm_dissect_nonblock(u_int32_t *, NFSX_UNSIGNED); - how = fxdr_unsigned(int, *tl); - switch (how) { - case NFSV3CREATE_GUARDED: - if (nd.ni_vp) { - error = EEXIST; - break; - } - /* fall through */ - case NFSV3CREATE_UNCHECKED: - nfsm_srvsattr(vap); - break; - case NFSV3CREATE_EXCLUSIVE: - tl = nfsm_dissect_nonblock(u_int32_t *, - NFSX_V3CREATEVERF); - /* Unique bytes, endianness is not important. */ - cverf.tv_sec = (int32_t)tl[0]; - cverf.tv_nsec = tl[1]; - exclusive_flag = 1; - break; - }; - vap->va_type = VREG; - } else { - sp = nfsm_dissect_nonblock(struct nfsv2_sattr *, NFSX_V2SATTR); - vap->va_type = IFTOVT(fxdr_unsigned(u_int32_t, sp->sa_mode)); - if (vap->va_type == VNON) - vap->va_type = VREG; - vap->va_mode = nfstov_mode(sp->sa_mode); - switch (vap->va_type) { - case VREG: - tsize = fxdr_unsigned(int32_t, sp->sa_size); - if (tsize != -1) - vap->va_size = (u_quad_t)tsize; - break; - case VCHR: - case VBLK: - case VFIFO: - rdev = fxdr_unsigned(long, sp->sa_size); - break; - default: - break; - }; - } - - /* - * Iff doesn't exist, create it - * otherwise just truncate to 0 length - * should I set the mode too ? - * - * The only possible error we can have at this point is EEXIST. - * nd.ni_vp will also be non-NULL in that case. - */ - if (nd.ni_vp == NULL) { - if (vap->va_mode == (mode_t)VNOVAL) - vap->va_mode = 0; - if (vap->va_type == VREG || vap->va_type == VSOCK) { - error = VOP_CREATE(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, vap); - if (error) - NDFREE(&nd, NDF_ONLY_PNBUF); - else { - if (exclusive_flag) { - exclusive_flag = 0; - VATTR_NULL(vap); - vap->va_atime = cverf; - error = VOP_SETATTR(nd.ni_vp, vap, - cred); - } - } - } else if (vap->va_type == VCHR || vap->va_type == VBLK || - vap->va_type == VFIFO) { - /* - * NFSv2-specific code for creating device nodes - * and fifos. - * - * Handle SysV FIFO node special cases. All other - * devices require super user to access. - */ - if (vap->va_type == VCHR && rdev == 0xffffffff) - vap->va_type = VFIFO; - if (vap->va_type != VFIFO && - (error = priv_check_cred(cred, PRIV_VFS_MKNOD_DEV, - 0))) { - goto ereply; - } - vap->va_rdev = rdev; - error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, vap); - if (error) { - NDFREE(&nd, NDF_ONLY_PNBUF); - goto ereply; - } - vput(nd.ni_vp); - nd.ni_vp = NULL; - - /* - * release dvp prior to lookup - */ - vput(nd.ni_dvp); - nd.ni_dvp = NULL; - /* - * Setup for lookup. - * - * Even though LOCKPARENT was cleared, ni_dvp may - * be garbage. - */ - nd.ni_cnd.cn_nameiop = LOOKUP; - nd.ni_cnd.cn_flags &= ~(LOCKPARENT); - nd.ni_cnd.cn_thread = curthread; - nd.ni_cnd.cn_cred = cred; - error = lookup(&nd); - nd.ni_dvp = NULL; - if (error) - goto ereply; - - if (nd.ni_cnd.cn_flags & ISSYMLINK) { - error = EINVAL; - goto ereply; - } - } else { - error = ENXIO; - } - } else { - if (vap->va_size != -1) { - error = nfsrv_access(nd.ni_vp, VWRITE, - cred, (nd.ni_cnd.cn_flags & RDONLY), 0); - if (!error) { - tempsize = vap->va_size; - VATTR_NULL(vap); - vap->va_size = tempsize; - error = VOP_SETATTR(nd.ni_vp, vap, cred); - } - } - } - - if (!error) { - bzero((caddr_t)fhp, sizeof(nfh)); - fhp->fh_fsid = nd.ni_vp->v_mount->mnt_stat.f_fsid; - error = VOP_VPTOFH(nd.ni_vp, &fhp->fh_fid); - if (!error) - error = VOP_GETATTR(nd.ni_vp, vap, cred); - } - if (v3) { - if (exclusive_flag && !error && - bcmp(&cverf, &vap->va_atime, sizeof (cverf))) - error = EEXIST; - if (dirp == nd.ni_dvp) - diraft_ret = VOP_GETATTR(dirp, &diraft, cred); - else { - /* Drop the other locks to avoid deadlock. */ - if (nd.ni_dvp) { - if (nd.ni_dvp == nd.ni_vp) - vrele(nd.ni_dvp); - else - vput(nd.ni_dvp); - } - if (nd.ni_vp) - vput(nd.ni_vp); - nd.ni_dvp = NULL; - nd.ni_vp = NULL; - - vn_lock(dirp, LK_EXCLUSIVE | LK_RETRY); - diraft_ret = VOP_GETATTR(dirp, &diraft, cred); - VOP_UNLOCK(dirp, 0); - } - } -ereply: - nfsm_reply(NFSX_SRVFH(v3) + NFSX_FATTR(v3) + NFSX_WCCDATA(v3)); - if (v3) { - if (!error) { - nfsm_srvpostop_fh(fhp); - nfsm_srvpostop_attr(0, vap); - } - nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft); - } else if (!error) { - /* v2 non-error case. */ - nfsm_srvfhtom(fhp, v3); - fp = nfsm_build(struct nfs_fattr *, NFSX_V2FATTR); - nfsm_srvfillattr(vap, fp); - } - error = 0; - -nfsmout: - if (nd.ni_dvp) { - if (nd.ni_dvp == nd.ni_vp) - vrele(nd.ni_dvp); - else - vput(nd.ni_dvp); - } - if (nd.ni_vp) - vput(nd.ni_vp); - if (nd.ni_startdir) { - vrele(nd.ni_startdir); - nd.ni_startdir = NULL; - } - if (dirp) - vrele(dirp); - NDFREE(&nd, NDF_ONLY_PNBUF); - vn_finished_write(mp); - return (error); -} - -/* - * nfs v3 mknod service - */ -int -nfsrv_mknod(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct vattr va, dirfor, diraft; - struct vattr *vap = &va; - struct thread *td = curthread; - u_int32_t *tl; - struct nameidata nd; - caddr_t bpos; - int error = 0, len, dirfor_ret = 1, diraft_ret = 1; - u_int32_t major, minor; - enum vtype vtyp; - struct mbuf *mb, *mreq; - struct vnode *vp, *dirp = NULL; - nfsfh_t nfh; - fhandle_t *fhp; - struct mount *mp = NULL; - int v3 = (nfsd->nd_flag & ND_NFSV3); - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - if (!v3) - panic("nfsrv_mknod: v3 proc called on a v2 connection"); - ndclear(&nd); - - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) { - error = ESTALE; - goto ereply; - } - (void) vn_start_write(NULL, &mp, V_WAIT); - vfs_rel(mp); /* The write holds a ref. */ - nfsm_srvnamesiz(len); - - nd.ni_cnd.cn_cred = cred; - nd.ni_cnd.cn_nameiop = CREATE; - nd.ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF | SAVESTART | NOCACHE; - - /* - * Handle nfs_namei() call. If an error occurs, the nd structure - * is not valid. However, nfsm_*() routines may still jump to - * nfsmout. - */ - - error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos, - &dirp, v3, &dirfor, &dirfor_ret, FALSE); - if (error) { - nfsm_reply(NFSX_WCCDATA(1)); - nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft); - error = 0; - goto nfsmout; - } - tl = nfsm_dissect_nonblock(u_int32_t *, NFSX_UNSIGNED); - vtyp = nfsv3tov_type(*tl); - if (vtyp != VCHR && vtyp != VBLK && vtyp != VSOCK && vtyp != VFIFO) { - error = NFSERR_BADTYPE; - goto out; - } - VATTR_NULL(vap); - nfsm_srvsattr(vap); - if (vtyp == VCHR || vtyp == VBLK) { - tl = nfsm_dissect_nonblock(u_int32_t *, 2 * NFSX_UNSIGNED); - major = fxdr_unsigned(u_int32_t, *tl++); - minor = fxdr_unsigned(u_int32_t, *tl); - vap->va_rdev = makedev(major, minor); - } - - /* - * Iff doesn't exist, create it. - */ - if (nd.ni_vp) { - error = EEXIST; - goto out; - } - vap->va_type = vtyp; - if (vap->va_mode == (mode_t)VNOVAL) - vap->va_mode = 0; - if (vtyp == VSOCK) { - vrele(nd.ni_startdir); - nd.ni_startdir = NULL; - error = VOP_CREATE(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, vap); - if (error) - NDFREE(&nd, NDF_ONLY_PNBUF); - } else { - if (vtyp != VFIFO && (error = priv_check_cred(cred, - PRIV_VFS_MKNOD_DEV, 0))) - goto out; - error = VOP_MKNOD(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, vap); - if (error) { - NDFREE(&nd, NDF_ONLY_PNBUF); - goto out; - } - vput(nd.ni_vp); - nd.ni_vp = NULL; - - /* - * Release dvp prior to lookup - */ - vput(nd.ni_dvp); - nd.ni_dvp = NULL; - - nd.ni_cnd.cn_nameiop = LOOKUP; - nd.ni_cnd.cn_flags &= ~(LOCKPARENT); - nd.ni_cnd.cn_thread = td; - nd.ni_cnd.cn_cred = td->td_ucred; - error = lookup(&nd); - nd.ni_dvp = NULL; - - if (error) - goto out; - if (nd.ni_cnd.cn_flags & ISSYMLINK) - error = EINVAL; - } - - /* - * send response, cleanup, return. - */ -out: - vp = nd.ni_vp; - if (!error) { - bzero((caddr_t)fhp, sizeof(nfh)); - fhp->fh_fsid = vp->v_mount->mnt_stat.f_fsid; - error = VOP_VPTOFH(vp, &fhp->fh_fid); - if (!error) - error = VOP_GETATTR(vp, vap, cred); - } - if (nd.ni_dvp) { - if (nd.ni_dvp == nd.ni_vp) - vrele(nd.ni_dvp); - else - vput(nd.ni_dvp); - nd.ni_dvp = NULL; - } - if (vp) { - vput(vp); - vp = NULL; - nd.ni_vp = NULL; - } - if (nd.ni_startdir) { - vrele(nd.ni_startdir); - nd.ni_startdir = NULL; - } - NDFREE(&nd, NDF_ONLY_PNBUF); - if (dirp) { - vn_lock(dirp, LK_EXCLUSIVE | LK_RETRY); - diraft_ret = VOP_GETATTR(dirp, &diraft, cred); - vput(dirp); - } -ereply: - nfsm_reply(NFSX_SRVFH(1) + NFSX_POSTOPATTR(1) + NFSX_WCCDATA(1)); - if (v3) { - if (!error) { - nfsm_srvpostop_fh(fhp); - nfsm_srvpostop_attr(0, vap); - } - nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft); - } - vn_finished_write(mp); - return (0); -nfsmout: - if (nd.ni_dvp) { - if (nd.ni_dvp == nd.ni_vp) - vrele(nd.ni_dvp); - else - vput(nd.ni_dvp); - } - if (nd.ni_vp) - vput(nd.ni_vp); - if (dirp) - vrele(dirp); - if (nd.ni_startdir) - vrele(nd.ni_startdir); - NDFREE(&nd, NDF_ONLY_PNBUF); - vn_finished_write(mp); - return (error); -} - -/* - * nfs remove service - */ -int -nfsrv_remove(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct nameidata nd; - caddr_t bpos; - int error = 0, len, dirfor_ret = 1, diraft_ret = 1; - int v3 = (nfsd->nd_flag & ND_NFSV3); - struct mbuf *mb, *mreq; - struct vnode *dirp; - struct vattr dirfor, diraft; - nfsfh_t nfh; - fhandle_t *fhp; - struct mount *mp = NULL; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - ndclear(&nd); - - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) { - error = ESTALE; - goto ereply; - } - (void) vn_start_write(NULL, &mp, V_WAIT); - vfs_rel(mp); /* The write holds a ref. */ - nfsm_srvnamesiz(len); - - nd.ni_cnd.cn_cred = cred; - nd.ni_cnd.cn_nameiop = DELETE; - nd.ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF; - error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos, - &dirp, v3, &dirfor, &dirfor_ret, FALSE); - if (dirp && !v3) { - vrele(dirp); - dirp = NULL; - } - if (error == 0) { - if (nd.ni_vp->v_type == VDIR) { - error = EPERM; /* POSIX */ - goto out; - } - /* - * The root of a mounted filesystem cannot be deleted. - */ - if (nd.ni_vp->v_vflag & VV_ROOT) { - error = EBUSY; - goto out; - } -out: - if (!error) { - error = VOP_REMOVE(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd); - NDFREE(&nd, NDF_ONLY_PNBUF); - } - } - if (dirp && v3) { - if (dirp == nd.ni_dvp) - diraft_ret = VOP_GETATTR(dirp, &diraft, cred); - else { - /* Drop the other locks to avoid deadlock. */ - if (nd.ni_dvp) { - if (nd.ni_dvp == nd.ni_vp) - vrele(nd.ni_dvp); - else - vput(nd.ni_dvp); - } - if (nd.ni_vp) - vput(nd.ni_vp); - nd.ni_dvp = NULL; - nd.ni_vp = NULL; - - vn_lock(dirp, LK_EXCLUSIVE | LK_RETRY); - diraft_ret = VOP_GETATTR(dirp, &diraft, cred); - VOP_UNLOCK(dirp, 0); - } - vrele(dirp); - dirp = NULL; - } -ereply: - nfsm_reply(NFSX_WCCDATA(v3)); - if (v3) - nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft); - error = 0; -nfsmout: - NDFREE(&nd, NDF_ONLY_PNBUF); - if (nd.ni_dvp) { - if (nd.ni_dvp == nd.ni_vp) - vrele(nd.ni_dvp); - else - vput(nd.ni_dvp); - } - if (nd.ni_vp) - vput(nd.ni_vp); - vn_finished_write(mp); - return(error); -} - -/* - * nfs rename service - */ -int -nfsrv_rename(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - caddr_t bpos; - int error = 0, len, len2, fdirfor_ret = 1, fdiraft_ret = 1; - int tdirfor_ret = 1, tdiraft_ret = 1; - int v3 = (nfsd->nd_flag & ND_NFSV3); - struct mbuf *mb, *mreq; - struct nameidata fromnd, tond; - struct vnode *fvp, *tvp, *tdvp, *fdirp = NULL; - struct vnode *tdirp = NULL; - struct vattr fdirfor, fdiraft, tdirfor, tdiraft; - nfsfh_t fnfh, tnfh; - fhandle_t *ffhp, *tfhp; - uid_t saved_uid; - struct mount *mp = NULL; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); -#ifndef nolint - fvp = NULL; -#endif - ffhp = &fnfh.fh_generic; - tfhp = &tnfh.fh_generic; - - /* - * Clear fields incase goto nfsmout occurs from macro. - */ - - ndclear(&fromnd); - ndclear(&tond); - - nfsm_srvmtofh(ffhp); - if ((mp = vfs_getvfs(&ffhp->fh_fsid)) == NULL) { - error = ESTALE; - goto out1; - } - (void) vn_start_write(NULL, &mp, V_WAIT); - vfs_rel(mp); /* The write holds a ref. */ - nfsm_srvnamesiz(len); - /* - * Remember our original uid so that we can reset cr_uid before - * the second nfs_namei() call, in case it is remapped. - */ - saved_uid = cred->cr_uid; - fromnd.ni_cnd.cn_cred = cred; - fromnd.ni_cnd.cn_nameiop = DELETE; - fromnd.ni_cnd.cn_flags = WANTPARENT | SAVESTART; - error = nfs_namei(&fromnd, nfsd, ffhp, len, slp, nam, &md, - &dpos, &fdirp, v3, &fdirfor, &fdirfor_ret, FALSE); - if (fdirp && !v3) { - vrele(fdirp); - fdirp = NULL; - } - if (error) { - nfsm_reply(2 * NFSX_WCCDATA(v3)); - if (v3) { - nfsm_srvwcc_data(fdirfor_ret, &fdirfor, fdiraft_ret, &fdiraft); - nfsm_srvwcc_data(tdirfor_ret, &tdirfor, tdiraft_ret, &tdiraft); - } - error = 0; - goto nfsmout; - } - fvp = fromnd.ni_vp; - nfsm_srvmtofh(tfhp); - nfsm_srvnamesiz(len2); - cred->cr_uid = saved_uid; - tond.ni_cnd.cn_cred = cred; - tond.ni_cnd.cn_nameiop = RENAME; - tond.ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF | NOCACHE | SAVESTART; - error = nfs_namei(&tond, nfsd, tfhp, len2, slp, nam, &md, - &dpos, &tdirp, v3, &tdirfor, &tdirfor_ret, FALSE); - if (tdirp && !v3) { - vrele(tdirp); - tdirp = NULL; - } - if (error) - goto out1; - - tdvp = tond.ni_dvp; - tvp = tond.ni_vp; - if (tvp != NULL) { - if (fvp->v_type == VDIR && tvp->v_type != VDIR) { - if (v3) - error = EEXIST; - else - error = EISDIR; - goto out; - } else if (fvp->v_type != VDIR && tvp->v_type == VDIR) { - if (v3) - error = EEXIST; - else - error = ENOTDIR; - goto out; - } - if (tvp->v_type == VDIR && tvp->v_mountedhere) { - if (v3) - error = EXDEV; - else - error = ENOTEMPTY; - goto out; - } - } - if (fvp->v_type == VDIR && fvp->v_mountedhere) { - if (v3) - error = EXDEV; - else - error = ENOTEMPTY; - goto out; - } - if (fvp->v_mount != tdvp->v_mount) { - if (v3) - error = EXDEV; - else - error = ENOTEMPTY; - goto out; - } - if (fvp == tdvp) { - if (v3) - error = EINVAL; - else - error = ENOTEMPTY; - } - /* - * If source is the same as the destination (that is the - * same vnode with the same name in the same directory), - * then there is nothing to do. - */ - if (fvp == tvp && fromnd.ni_dvp == tdvp && - fromnd.ni_cnd.cn_namelen == tond.ni_cnd.cn_namelen && - !bcmp(fromnd.ni_cnd.cn_nameptr, tond.ni_cnd.cn_nameptr, - fromnd.ni_cnd.cn_namelen)) - error = -1; -out: - if (!error) { - /* - * The VOP_RENAME function releases all vnode references & - * locks prior to returning so we need to clear the pointers - * to bypass cleanup code later on. - */ - error = VOP_RENAME(fromnd.ni_dvp, fromnd.ni_vp, &fromnd.ni_cnd, - tond.ni_dvp, tond.ni_vp, &tond.ni_cnd); - fromnd.ni_dvp = NULL; - fromnd.ni_vp = NULL; - tond.ni_dvp = NULL; - tond.ni_vp = NULL; - if (error) { - NDFREE(&fromnd, NDF_ONLY_PNBUF); - NDFREE(&tond, NDF_ONLY_PNBUF); - } - } else { - if (error == -1) - error = 0; - } - /* fall through */ -out1: - nfsm_reply(2 * NFSX_WCCDATA(v3)); - if (v3) { - /* Release existing locks to prevent deadlock. */ - if (tond.ni_dvp) { - if (tond.ni_dvp == tond.ni_vp) - vrele(tond.ni_dvp); - else - vput(tond.ni_dvp); - } - if (tond.ni_vp) - vput(tond.ni_vp); - tond.ni_dvp = NULL; - tond.ni_vp = NULL; - - if (fdirp) { - vn_lock(fdirp, LK_EXCLUSIVE | LK_RETRY); - fdiraft_ret = VOP_GETATTR(fdirp, &fdiraft, cred); - VOP_UNLOCK(fdirp, 0); - } - if (tdirp) { - vn_lock(tdirp, LK_EXCLUSIVE | LK_RETRY); - tdiraft_ret = VOP_GETATTR(tdirp, &tdiraft, cred); - VOP_UNLOCK(tdirp, 0); - } - nfsm_srvwcc_data(fdirfor_ret, &fdirfor, fdiraft_ret, &fdiraft); - nfsm_srvwcc_data(tdirfor_ret, &tdirfor, tdiraft_ret, &tdiraft); - } - error = 0; - /* fall through */ - -nfsmout: - /* - * Clear out tond related fields - */ - if (tond.ni_dvp) { - if (tond.ni_dvp == tond.ni_vp) - vrele(tond.ni_dvp); - else - vput(tond.ni_dvp); - } - if (tond.ni_vp) - vput(tond.ni_vp); - if (tdirp) - vrele(tdirp); - if (tond.ni_startdir) - vrele(tond.ni_startdir); - NDFREE(&tond, NDF_ONLY_PNBUF); - /* - * Clear out fromnd related fields - */ - if (fdirp) - vrele(fdirp); - if (fromnd.ni_startdir) - vrele(fromnd.ni_startdir); - NDFREE(&fromnd, NDF_ONLY_PNBUF); - if (fromnd.ni_dvp) - vrele(fromnd.ni_dvp); - if (fromnd.ni_vp) - vrele(fromnd.ni_vp); - - vn_finished_write(mp); - return (error); -} - -/* - * nfs link service - */ -int -nfsrv_link(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct nameidata nd; - caddr_t bpos; - int error = 0, rdonly, len, dirfor_ret = 1, diraft_ret = 1; - int getret = 1, v3 = (nfsd->nd_flag & ND_NFSV3); - struct mbuf *mb, *mreq; - struct vnode *vp = NULL, *xp, *dirp = NULL; - struct vattr dirfor, diraft, at; - nfsfh_t nfh, dnfh; - fhandle_t *fhp, *dfhp; - struct mount *mp = NULL; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - ndclear(&nd); - - fhp = &nfh.fh_generic; - dfhp = &dnfh.fh_generic; - nfsm_srvmtofh(fhp); - if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) { - error = ESTALE; - goto ereply; - } - (void) vn_start_write(NULL, &mp, V_WAIT); - vfs_rel(mp); /* The write holds a ref. */ - nfsm_srvmtofh(dfhp); - nfsm_srvnamesiz(len); - - error = nfsrv_fhtovp(fhp, 0, &vp, nfsd, slp, nam, &rdonly); - if (error) { - nfsm_reply(NFSX_POSTOPATTR(v3) + NFSX_WCCDATA(v3)); - if (v3) { - nfsm_srvpostop_attr(getret, &at); - nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft); - } - vp = NULL; - error = 0; - goto nfsmout; - } - if (v3) - getret = VOP_GETATTR(vp, &at, cred); - if (vp->v_type == VDIR) { - error = EPERM; /* POSIX */ - goto out1; - } - VOP_UNLOCK(vp, 0); - nd.ni_cnd.cn_cred = cred; - nd.ni_cnd.cn_nameiop = CREATE; - nd.ni_cnd.cn_flags = LOCKPARENT | NOCACHE; - error = nfs_namei(&nd, nfsd, dfhp, len, slp, nam, &md, &dpos, - &dirp, v3, &dirfor, &dirfor_ret, FALSE); - if (dirp && !v3) { - vrele(dirp); - dirp = NULL; - } - if (error) { - vrele(vp); - vp = NULL; - goto out2; - } - xp = nd.ni_vp; - if (xp != NULL) { - error = EEXIST; - vrele(vp); - vp = NULL; - goto out2; - } - xp = nd.ni_dvp; - if (vp->v_mount != xp->v_mount) { - error = EXDEV; - vrele(vp); - vp = NULL; - goto out2; - } - vn_lock(vp, LK_EXCLUSIVE | LK_RETRY); - error = VOP_LINK(nd.ni_dvp, vp, &nd.ni_cnd); - NDFREE(&nd, NDF_ONLY_PNBUF); - /* fall through */ - -out1: - if (v3) - getret = VOP_GETATTR(vp, &at, cred); -out2: - if (dirp) { - if (dirp == nd.ni_dvp) - diraft_ret = VOP_GETATTR(dirp, &diraft, cred); - else { - /* Release existing locks to prevent deadlock. */ - if (nd.ni_dvp) { - if (nd.ni_dvp == nd.ni_vp) - vrele(nd.ni_dvp); - else - vput(nd.ni_dvp); - } - if (nd.ni_vp) - vrele(nd.ni_vp); - nd.ni_dvp = NULL; - nd.ni_vp = NULL; - - vn_lock(dirp, LK_EXCLUSIVE | LK_RETRY); - diraft_ret = VOP_GETATTR(dirp, &diraft, cred); - VOP_UNLOCK(dirp, 0); - } - } -ereply: - nfsm_reply(NFSX_POSTOPATTR(v3) + NFSX_WCCDATA(v3)); - if (v3) { - nfsm_srvpostop_attr(getret, &at); - nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft); - } - error = 0; - /* fall through */ - -nfsmout: - NDFREE(&nd, NDF_ONLY_PNBUF); - if (vp) - vput(vp); - if (nd.ni_dvp) { - if (nd.ni_dvp == nd.ni_vp) - vrele(nd.ni_dvp); - else - vput(nd.ni_dvp); - } - if (dirp) - vrele(dirp); - if (nd.ni_vp) - vrele(nd.ni_vp); - vn_finished_write(mp); - return(error); -} - -/* - * nfs symbolic link service - */ -int -nfsrv_symlink(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct vattr va, dirfor, diraft; - struct nameidata nd; - struct vattr *vap = &va; - struct nfsv2_sattr *sp; - char *bpos, *pathcp = NULL; - struct uio io; - struct iovec iv; - int error = 0, len, len2, dirfor_ret = 1, diraft_ret = 1; - int v3 = (nfsd->nd_flag & ND_NFSV3); - struct mbuf *mb, *mreq; - struct vnode *dirp = NULL; - nfsfh_t nfh; - fhandle_t *fhp; - struct mount *mp = NULL; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - ndclear(&nd); - - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) { - error = ESTALE; - goto out; - } - (void) vn_start_write(NULL, &mp, V_WAIT); - vfs_rel(mp); /* The write holds a ref. */ - nfsm_srvnamesiz(len); - nd.ni_cnd.cn_cred = cred; - nd.ni_cnd.cn_nameiop = CREATE; - nd.ni_cnd.cn_flags = LOCKPARENT | SAVESTART | NOCACHE; - error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos, - &dirp, v3, &dirfor, &dirfor_ret, FALSE); - if (error == 0) { - VATTR_NULL(vap); - if (v3) - nfsm_srvsattr(vap); - nfsm_srvpathsiz(len2); - } - if (dirp && !v3) { - vrele(dirp); - dirp = NULL; - } - if (error) - goto out; - pathcp = malloc(len2 + 1, M_TEMP, M_WAITOK); - iv.iov_base = pathcp; - iv.iov_len = len2; - io.uio_resid = len2; - io.uio_offset = 0; - io.uio_iov = &iv; - io.uio_iovcnt = 1; - io.uio_segflg = UIO_SYSSPACE; - io.uio_rw = UIO_READ; - io.uio_td = NULL; - nfsm_mtouio(&io, len2); - if (!v3) { - sp = nfsm_dissect_nonblock(struct nfsv2_sattr *, NFSX_V2SATTR); - vap->va_mode = nfstov_mode(sp->sa_mode); - } - *(pathcp + len2) = '\0'; - if (nd.ni_vp) { - error = EEXIST; - goto out; - } - - /* - * issue symlink op. SAVESTART is set so the underlying path component - * is only freed by the VOP if an error occurs. - */ - if (vap->va_mode == (mode_t)VNOVAL) - vap->va_mode = 0; - error = VOP_SYMLINK(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, vap, pathcp); - if (error) - NDFREE(&nd, NDF_ONLY_PNBUF); - else - vput(nd.ni_vp); - nd.ni_vp = NULL; - /* - * releases directory prior to potential lookup op. - */ - vput(nd.ni_dvp); - nd.ni_dvp = NULL; - - if (error == 0) { - if (v3) { - /* - * Issue lookup. Leave SAVESTART set so we can easily free - * the name buffer later on. - * - * since LOCKPARENT is not set, ni_dvp will be garbage on - * return whether an error occurs or not. - */ - nd.ni_cnd.cn_nameiop = LOOKUP; - nd.ni_cnd.cn_flags &= ~(LOCKPARENT | FOLLOW); - nd.ni_cnd.cn_flags |= (NOFOLLOW | LOCKLEAF); - nd.ni_cnd.cn_thread = curthread; - nd.ni_cnd.cn_cred = cred; - error = lookup(&nd); - nd.ni_dvp = NULL; - - if (error == 0) { - bzero((caddr_t)fhp, sizeof(nfh)); - fhp->fh_fsid = nd.ni_vp->v_mount->mnt_stat.f_fsid; - error = VOP_VPTOFH(nd.ni_vp, &fhp->fh_fid); - if (!error) - error = VOP_GETATTR(nd.ni_vp, vap, cred); - vput(nd.ni_vp); - nd.ni_vp = NULL; - } - } - } -out: - /* - * These releases aren't strictly required, does even doing them - * make any sense? XXX can nfsm_reply() block? - */ - if (pathcp) { - free(pathcp, M_TEMP); - pathcp = NULL; - } - if (dirp) { - vn_lock(dirp, LK_EXCLUSIVE | LK_RETRY); - diraft_ret = VOP_GETATTR(dirp, &diraft, cred); - VOP_UNLOCK(dirp, 0); - } - if (nd.ni_startdir) { - vrele(nd.ni_startdir); - nd.ni_startdir = NULL; - } - nfsm_reply(NFSX_SRVFH(v3) + NFSX_POSTOPATTR(v3) + NFSX_WCCDATA(v3)); - if (v3) { - if (!error) { - nfsm_srvpostop_fh(fhp); - nfsm_srvpostop_attr(0, vap); - } - nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft); - } - error = 0; - /* fall through */ - -nfsmout: - NDFREE(&nd, NDF_ONLY_PNBUF); - if (nd.ni_dvp) { - if (nd.ni_dvp == nd.ni_vp) - vrele(nd.ni_dvp); - else - vput(nd.ni_dvp); - } - if (nd.ni_vp) - vrele(nd.ni_vp); - if (nd.ni_startdir) - vrele(nd.ni_startdir); - if (dirp) - vrele(dirp); - if (pathcp) - free(pathcp, M_TEMP); - - vn_finished_write(mp); - return (error); -} - -/* - * nfs mkdir service - */ -int -nfsrv_mkdir(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct vattr va, dirfor, diraft; - struct vattr *vap = &va; - struct nfs_fattr *fp; - struct nameidata nd; - u_int32_t *tl; - caddr_t bpos; - int error = 0, len, dirfor_ret = 1, diraft_ret = 1; - int v3 = (nfsd->nd_flag & ND_NFSV3); - struct mbuf *mb, *mreq; - struct vnode *dirp = NULL; - int vpexcl = 0; - nfsfh_t nfh; - fhandle_t *fhp; - struct mount *mp = NULL; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - ndclear(&nd); - - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) { - error = ESTALE; - goto out; - } - (void) vn_start_write(NULL, &mp, V_WAIT); - vfs_rel(mp); /* The write holds a ref. */ - nfsm_srvnamesiz(len); - nd.ni_cnd.cn_cred = cred; - nd.ni_cnd.cn_nameiop = CREATE; - nd.ni_cnd.cn_flags = LOCKPARENT | NOCACHE; - - error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos, - &dirp, v3, &dirfor, &dirfor_ret, FALSE); - if (dirp && !v3) { - vrele(dirp); - dirp = NULL; - } - if (error) { - nfsm_reply(NFSX_WCCDATA(v3)); - if (v3) - nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft); - error = 0; - goto nfsmout; - } - VATTR_NULL(vap); - if (v3) { - nfsm_srvsattr(vap); - } else { - tl = nfsm_dissect_nonblock(u_int32_t *, NFSX_UNSIGNED); - vap->va_mode = nfstov_mode(*tl++); - } - - /* - * At this point nd.ni_dvp is referenced and exclusively locked and - * nd.ni_vp, if it exists, is referenced but not locked. - */ - - vap->va_type = VDIR; - if (nd.ni_vp != NULL) { - NDFREE(&nd, NDF_ONLY_PNBUF); - error = EEXIST; - goto out; - } - - /* - * Issue mkdir op. Since SAVESTART is not set, the pathname - * component is freed by the VOP call. This will fill-in - * nd.ni_vp, reference, and exclusively lock it. - */ - if (vap->va_mode == (mode_t)VNOVAL) - vap->va_mode = 0; - error = VOP_MKDIR(nd.ni_dvp, &nd.ni_vp, &nd.ni_cnd, vap); - NDFREE(&nd, NDF_ONLY_PNBUF); - vpexcl = 1; - - vput(nd.ni_dvp); - nd.ni_dvp = NULL; - - if (!error) { - bzero((caddr_t)fhp, sizeof(nfh)); - fhp->fh_fsid = nd.ni_vp->v_mount->mnt_stat.f_fsid; - error = VOP_VPTOFH(nd.ni_vp, &fhp->fh_fid); - if (!error) - error = VOP_GETATTR(nd.ni_vp, vap, cred); - } -out: - if (dirp) { - if (dirp == nd.ni_dvp) { - diraft_ret = VOP_GETATTR(dirp, &diraft, cred); - } else { - /* Release existing locks to prevent deadlock. */ - if (nd.ni_dvp) { - NDFREE(&nd, NDF_ONLY_PNBUF); - if (nd.ni_dvp == nd.ni_vp && vpexcl) - vrele(nd.ni_dvp); - else - vput(nd.ni_dvp); - } - if (nd.ni_vp) { - if (vpexcl) - vput(nd.ni_vp); - else - vrele(nd.ni_vp); - } - nd.ni_dvp = NULL; - nd.ni_vp = NULL; - vn_lock(dirp, LK_EXCLUSIVE | LK_RETRY); - diraft_ret = VOP_GETATTR(dirp, &diraft, cred); - VOP_UNLOCK(dirp, 0); - } - } - nfsm_reply(NFSX_SRVFH(v3) + NFSX_POSTOPATTR(v3) + NFSX_WCCDATA(v3)); - if (v3) { - if (!error) { - nfsm_srvpostop_fh(fhp); - nfsm_srvpostop_attr(0, vap); - } - nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft); - } else if (!error) { - /* v2 non-error case. */ - nfsm_srvfhtom(fhp, v3); - fp = nfsm_build(struct nfs_fattr *, NFSX_V2FATTR); - nfsm_srvfillattr(vap, fp); - } - error = 0; - /* fall through */ - -nfsmout: - if (nd.ni_dvp) { - NDFREE(&nd, NDF_ONLY_PNBUF); - if (nd.ni_dvp == nd.ni_vp && vpexcl) - vrele(nd.ni_dvp); - else - vput(nd.ni_dvp); - } - if (nd.ni_vp) { - if (vpexcl) - vput(nd.ni_vp); - else - vrele(nd.ni_vp); - } - if (dirp) - vrele(dirp); - vn_finished_write(mp); - return (error); -} - -/* - * nfs rmdir service - */ -int -nfsrv_rmdir(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - caddr_t bpos; - int error = 0, len, dirfor_ret = 1, diraft_ret = 1; - int v3 = (nfsd->nd_flag & ND_NFSV3); - struct mbuf *mb, *mreq; - struct vnode *vp, *dirp = NULL; - struct vattr dirfor, diraft; - nfsfh_t nfh; - fhandle_t *fhp; - struct nameidata nd; - struct mount *mp = NULL; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - ndclear(&nd); - - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) { - error = ESTALE; - goto out; - } - (void) vn_start_write(NULL, &mp, V_WAIT); - vfs_rel(mp); /* The write holds a ref. */ - nfsm_srvnamesiz(len); - nd.ni_cnd.cn_cred = cred; - nd.ni_cnd.cn_nameiop = DELETE; - nd.ni_cnd.cn_flags = LOCKPARENT | LOCKLEAF; - error = nfs_namei(&nd, nfsd, fhp, len, slp, nam, &md, &dpos, - &dirp, v3, &dirfor, &dirfor_ret, FALSE); - if (dirp && !v3) { - vrele(dirp); - dirp = NULL; - } - if (error) { - nfsm_reply(NFSX_WCCDATA(v3)); - if (v3) - nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft); - error = 0; - goto nfsmout; - } - vp = nd.ni_vp; - if (vp->v_type != VDIR) { - error = ENOTDIR; - goto out; - } - /* - * No rmdir "." please. - */ - if (nd.ni_dvp == vp) { - error = EINVAL; - goto out; - } - /* - * The root of a mounted filesystem cannot be deleted. - */ - if (vp->v_vflag & VV_ROOT) - error = EBUSY; -out: - /* - * Issue or abort op. Since SAVESTART is not set, path name - * component is freed by the VOP after either. - */ - if (!error) - error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd); - NDFREE(&nd, NDF_ONLY_PNBUF); - - if (dirp) { - if (dirp == nd.ni_dvp) - diraft_ret = VOP_GETATTR(dirp, &diraft, cred); - else { - /* Release existing locks to prevent deadlock. */ - if (nd.ni_dvp) { - if (nd.ni_dvp == nd.ni_vp) - vrele(nd.ni_dvp); - else - vput(nd.ni_dvp); - } - if (nd.ni_vp) - vput(nd.ni_vp); - nd.ni_dvp = NULL; - nd.ni_vp = NULL; - vn_lock(dirp, LK_EXCLUSIVE | LK_RETRY); - diraft_ret = VOP_GETATTR(dirp, &diraft, cred); - VOP_UNLOCK(dirp, 0); - } - } - nfsm_reply(NFSX_WCCDATA(v3)); - error = 0; - if (v3) - nfsm_srvwcc_data(dirfor_ret, &dirfor, diraft_ret, &diraft); - /* fall through */ - -nfsmout: - NDFREE(&nd, NDF_ONLY_PNBUF); - if (nd.ni_dvp) { - if (nd.ni_dvp == nd.ni_vp) - vrele(nd.ni_dvp); - else - vput(nd.ni_dvp); - } - if (nd.ni_vp) - vput(nd.ni_vp); - if (dirp) - vrele(dirp); - - vn_finished_write(mp); - return(error); -} - -/* - * nfs readdir service - * - mallocs what it thinks is enough to read - * count rounded up to a multiple of NFS_DIRBLKSIZ <= NFS_MAXREADDIR - * - calls VOP_READDIR() - * - loops around building the reply - * if the output generated exceeds count break out of loop - * The nfsm_clget macro is used here so that the reply will be packed - * tightly in mbuf clusters. - * - it only knows that it has encountered eof when the VOP_READDIR() - * reads nothing - * - as such one readdir rpc will return eof false although you are there - * and then the next will return eof - * - it trims out records with d_fileno == 0 - * this doesn't matter for Unix clients, but they might confuse clients - * for other os'. - * NB: It is tempting to set eof to true if the VOP_READDIR() reads less - * than requested, but this may not apply to all filesystems. For - * example, client NFS does not { although it is never remote mounted - * anyhow } - * The alternate call nfsrv_readdirplus() does lookups as well. - * PS: The NFS protocol spec. does not clarify what the "count" byte - * argument is a count of.. just name strings and file id's or the - * entire reply rpc or ... - * I tried just file name and id sizes and it confused the Sun client, - * so I am using the full rpc size now. The "paranoia.." comment refers - * to including the status longwords that are not a part of the dir. - * "entry" structures, but are in the rpc. - */ -struct flrep { - nfsuint64 fl_off; - u_int32_t fl_postopok; - u_int32_t fl_fattr[NFSX_V3FATTR / sizeof (u_int32_t)]; - u_int32_t fl_fhok; - u_int32_t fl_fhsize; - u_int32_t fl_nfh[NFSX_V3FH / sizeof (u_int32_t)]; -}; - -int -nfsrv_readdir(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - char *bp, *be; - struct mbuf *mp; - struct dirent *dp; - caddr_t cp; - u_int32_t *tl; - caddr_t bpos; - struct mbuf *mb, *mreq; - char *cpos, *cend, *rbuf; - struct vnode *vp = NULL; - struct vattr at; - nfsfh_t nfh; - fhandle_t *fhp; - struct uio io; - struct iovec iv; - int len, nlen, rem, xfer, tsiz, i, error = 0, getret = 1; - int siz, cnt, fullsiz, eofflag, rdonly, ncookies; - int v3 = (nfsd->nd_flag & ND_NFSV3); - u_quad_t off, toff, verf; - u_long *cookies = NULL, *cookiep; /* needs to be int64_t or off_t */ - int is_ufs; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - if (v3) { - tl = nfsm_dissect_nonblock(u_int32_t *, 5 * NFSX_UNSIGNED); - toff = fxdr_hyper(tl); - tl += 2; - verf = fxdr_hyper(tl); - tl += 2; - } else { - tl = nfsm_dissect_nonblock(u_int32_t *, 2 * NFSX_UNSIGNED); - toff = fxdr_unsigned(u_quad_t, *tl++); - verf = 0; /* shut up gcc */ - } - off = toff; - cnt = fxdr_unsigned(int, *tl); - siz = ((cnt + DIRBLKSIZ - 1) & ~(DIRBLKSIZ - 1)); - xfer = NFS_SRVMAXDATA(nfsd); - if (cnt > xfer) - cnt = xfer; - if (siz > xfer) - siz = xfer; - fullsiz = siz; - error = nfsrv_fhtovp(fhp, 0, &vp, nfsd, slp, nam, &rdonly); - if (!error && vp->v_type != VDIR) { - error = ENOTDIR; - vput(vp); - vp = NULL; - } - if (error) { - nfsm_reply(NFSX_UNSIGNED); - if (v3) - nfsm_srvpostop_attr(getret, &at); - error = 0; - goto nfsmout; - } - - /* - * Obtain lock on vnode for this section of the code - */ - if (v3) { - error = getret = VOP_GETATTR(vp, &at, cred); -#if 0 - /* - * XXX This check may be too strict for Solaris 2.5 clients. - */ - if (!error && toff && verf && verf != at.va_filerev) - error = NFSERR_BAD_COOKIE; -#endif - } - if (!error) - error = nfsrv_access(vp, VEXEC, cred, rdonly, 0); - if (error) { - vput(vp); - vp = NULL; - nfsm_reply(NFSX_POSTOPATTR(v3)); - if (v3) - nfsm_srvpostop_attr(getret, &at); - error = 0; - goto nfsmout; - } - is_ufs = strcmp(vp->v_mount->mnt_vfc->vfc_name, "ufs") == 0; - VOP_UNLOCK(vp, 0); - - /* - * end section. Allocate rbuf and continue - */ - rbuf = malloc(siz, M_TEMP, M_WAITOK); -again: - iv.iov_base = rbuf; - iv.iov_len = fullsiz; - io.uio_iov = &iv; - io.uio_iovcnt = 1; - io.uio_offset = (off_t)off; - io.uio_resid = fullsiz; - io.uio_segflg = UIO_SYSSPACE; - io.uio_rw = UIO_READ; - io.uio_td = NULL; - eofflag = 0; - if (cookies) { - free((caddr_t)cookies, M_TEMP); - cookies = NULL; - } - vn_lock(vp, LK_SHARED | LK_RETRY); - error = VOP_READDIR(vp, &io, cred, &eofflag, &ncookies, &cookies); - off = (off_t)io.uio_offset; - if (!cookies && !error) - error = NFSERR_PERM; - if (v3) { - getret = VOP_GETATTR(vp, &at, cred); - if (!error) - error = getret; - } - VOP_UNLOCK(vp, 0); - if (error) { - vrele(vp); - vp = NULL; - free((caddr_t)rbuf, M_TEMP); - if (cookies) - free((caddr_t)cookies, M_TEMP); - nfsm_reply(NFSX_POSTOPATTR(v3)); - if (v3) - nfsm_srvpostop_attr(getret, &at); - error = 0; - goto nfsmout; - } - if (io.uio_resid) { - siz -= io.uio_resid; - - /* - * If nothing read, return eof - * rpc reply - */ - if (siz == 0) { - vrele(vp); - vp = NULL; - nfsm_reply(NFSX_POSTOPATTR(v3) + NFSX_COOKIEVERF(v3) + - 2 * NFSX_UNSIGNED); - if (v3) { - nfsm_srvpostop_attr(getret, &at); - tl = nfsm_build(u_int32_t *, 4 * NFSX_UNSIGNED); - txdr_hyper(at.va_filerev, tl); - tl += 2; - } else - tl = nfsm_build(u_int32_t *, 2 * NFSX_UNSIGNED); - *tl++ = nfsrv_nfs_false; - *tl = nfsrv_nfs_true; - free((caddr_t)rbuf, M_TEMP); - free((caddr_t)cookies, M_TEMP); - error = 0; - goto nfsmout; - } - } - - /* - * Check for degenerate cases of nothing useful read. - * If so go try again - */ - cpos = rbuf; - cend = rbuf + siz; - dp = (struct dirent *)cpos; - cookiep = cookies; - /* - * For some reason FreeBSD's ufs_readdir() chooses to back the - * directory offset up to a block boundary, so it is necessary to - * skip over the records that precede the requested offset. This - * requires the assumption that file offset cookies monotonically - * increase. - */ - while (cpos < cend && ncookies > 0 && - (dp->d_fileno == 0 || dp->d_type == DT_WHT || - (is_ufs == 1 && ((u_quad_t)(*cookiep)) <= toff))) { - cpos += dp->d_reclen; - dp = (struct dirent *)cpos; - cookiep++; - ncookies--; - } - if (cpos >= cend || ncookies == 0) { - toff = off; - siz = fullsiz; - goto again; - } - - len = 3 * NFSX_UNSIGNED; /* paranoia, probably can be 0 */ - nfsm_reply(NFSX_POSTOPATTR(v3) + NFSX_COOKIEVERF(v3) + siz); - if (v3) { - nfsm_srvpostop_attr(getret, &at); - tl = nfsm_build(u_int32_t *, 2 * NFSX_UNSIGNED); - txdr_hyper(at.va_filerev, tl); - } - mp = mb; - bp = bpos; - be = bp + M_TRAILINGSPACE(mp); - - /* Loop through the records and build reply */ - while (cpos < cend && ncookies > 0) { - if (dp->d_fileno != 0 && dp->d_type != DT_WHT) { - nlen = dp->d_namlen; - rem = nfsm_rndup(nlen) - nlen; - len += (4 * NFSX_UNSIGNED + nlen + rem); - if (v3) - len += 2 * NFSX_UNSIGNED; - if (len > cnt) { - eofflag = 0; - break; - } - /* - * Build the directory record xdr from - * the dirent entry. - */ - nfsm_clget; - *tl = nfsrv_nfs_true; - bp += NFSX_UNSIGNED; - if (v3) { - nfsm_clget; - *tl = 0; - bp += NFSX_UNSIGNED; - } - nfsm_clget; - *tl = txdr_unsigned(dp->d_fileno); - bp += NFSX_UNSIGNED; - nfsm_clget; - *tl = txdr_unsigned(nlen); - bp += NFSX_UNSIGNED; - - /* And loop around copying the name */ - xfer = nlen; - cp = dp->d_name; - while (xfer > 0) { - nfsm_clget; - if ((bp+xfer) > be) - tsiz = be-bp; - else - tsiz = xfer; - bcopy(cp, bp, tsiz); - bp += tsiz; - xfer -= tsiz; - if (xfer > 0) - cp += tsiz; - } - /* And null pad to an int32_t boundary. */ - for (i = 0; i < rem; i++) - *bp++ = '\0'; - nfsm_clget; - - /* Finish off the record */ - if (v3) { - *tl = 0; - bp += NFSX_UNSIGNED; - nfsm_clget; - } - *tl = txdr_unsigned(*cookiep); - bp += NFSX_UNSIGNED; - } - cpos += dp->d_reclen; - dp = (struct dirent *)cpos; - cookiep++; - ncookies--; - } - vrele(vp); - vp = NULL; - nfsm_clget; - *tl = nfsrv_nfs_false; - bp += NFSX_UNSIGNED; - nfsm_clget; - if (eofflag) - *tl = nfsrv_nfs_true; - else - *tl = nfsrv_nfs_false; - bp += NFSX_UNSIGNED; - if (mp != mb) { - if (bp < be) - mp->m_len = bp - mtod(mp, caddr_t); - } else - mp->m_len += bp - bpos; - free((caddr_t)rbuf, M_TEMP); - free((caddr_t)cookies, M_TEMP); - -nfsmout: - if (vp) - vrele(vp); - return(error); -} - -int -nfsrv_readdirplus(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - char *bp, *be; - struct mbuf *mp; - struct dirent *dp; - caddr_t cp; - u_int32_t *tl; - caddr_t bpos; - struct mbuf *mb, *mreq; - char *cpos, *cend, *rbuf; - struct vnode *vp = NULL, *nvp; - struct flrep fl; - nfsfh_t nfh; - fhandle_t *fhp, *nfhp = (fhandle_t *)fl.fl_nfh; - struct uio io; - struct iovec iv; - struct vattr va, at, *vap = &va; - struct nfs_fattr *fp; - int len, nlen, rem, xfer, tsiz, i, error = 0, error1, getret = 1; - int vp_locked; - int siz, cnt, fullsiz, eofflag, rdonly, dirlen, ncookies; - u_quad_t off, toff, verf; - u_long *cookies = NULL, *cookiep; /* needs to be int64_t or off_t */ - int v3 = (nfsd->nd_flag & ND_NFSV3); - int usevget = 1; - struct componentname cn; - struct mount *mntp = NULL; - int is_ufs; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - vp_locked = 0; - if (!v3) - panic("nfsrv_readdirplus: v3 proc called on a v2 connection"); - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - tl = nfsm_dissect_nonblock(u_int32_t *, 6 * NFSX_UNSIGNED); - toff = fxdr_hyper(tl); - tl += 2; - verf = fxdr_hyper(tl); - tl += 2; - siz = fxdr_unsigned(int, *tl++); - cnt = fxdr_unsigned(int, *tl); - off = toff; - siz = ((siz + DIRBLKSIZ - 1) & ~(DIRBLKSIZ - 1)); - xfer = NFS_SRVMAXDATA(nfsd); - if (cnt > xfer) - cnt = xfer; - if (siz > xfer) - siz = xfer; - fullsiz = siz; - error = nfsrv_fhtovp(fhp, NFSRV_FLAG_BUSY, &vp, nfsd, slp, - nam, &rdonly); - if (!error) { - vp_locked = 1; - mntp = vp->v_mount; - if (vp->v_type != VDIR) { - error = ENOTDIR; - vput(vp); - vp = NULL; - vp_locked = 0; - } - } - if (error) { - nfsm_reply(NFSX_UNSIGNED); - nfsm_srvpostop_attr(getret, &at); - error = 0; - goto nfsmout; - } - error = getret = VOP_GETATTR(vp, &at, cred); -#if 0 - /* - * XXX This check may be too strict for Solaris 2.5 clients. - */ - if (!error && toff && verf && verf != at.va_filerev) - error = NFSERR_BAD_COOKIE; -#endif - if (!error) - error = nfsrv_access(vp, VEXEC, cred, rdonly, 0); - if (error) { - vput(vp); - vp_locked = 0; - vp = NULL; - nfsm_reply(NFSX_V3POSTOPATTR); - nfsm_srvpostop_attr(getret, &at); - error = 0; - goto nfsmout; - } - is_ufs = strcmp(vp->v_mount->mnt_vfc->vfc_name, "ufs") == 0; - VOP_UNLOCK(vp, 0); - vp_locked = 0; - rbuf = malloc(siz, M_TEMP, M_WAITOK); -again: - iv.iov_base = rbuf; - iv.iov_len = fullsiz; - io.uio_iov = &iv; - io.uio_iovcnt = 1; - io.uio_offset = (off_t)off; - io.uio_resid = fullsiz; - io.uio_segflg = UIO_SYSSPACE; - io.uio_rw = UIO_READ; - io.uio_td = NULL; - eofflag = 0; - vp_locked = 1; - if (cookies) { - free((caddr_t)cookies, M_TEMP); - cookies = NULL; - } - vn_lock(vp, LK_SHARED | LK_RETRY); - error = VOP_READDIR(vp, &io, cred, &eofflag, &ncookies, &cookies); - off = (u_quad_t)io.uio_offset; - getret = VOP_GETATTR(vp, &at, cred); - VOP_UNLOCK(vp, 0); - vp_locked = 0; - if (!cookies && !error) - error = NFSERR_PERM; - if (!error) - error = getret; - if (error) { - vrele(vp); - vp = NULL; - if (cookies) - free((caddr_t)cookies, M_TEMP); - free((caddr_t)rbuf, M_TEMP); - nfsm_reply(NFSX_V3POSTOPATTR); - nfsm_srvpostop_attr(getret, &at); - error = 0; - goto nfsmout; - } - if (io.uio_resid) { - siz -= io.uio_resid; - - /* - * If nothing read, return eof - * rpc reply - */ - if (siz == 0) { - vrele(vp); - vp = NULL; - nfsm_reply(NFSX_V3POSTOPATTR + NFSX_V3COOKIEVERF + - 2 * NFSX_UNSIGNED); - nfsm_srvpostop_attr(getret, &at); - tl = nfsm_build(u_int32_t *, 4 * NFSX_UNSIGNED); - txdr_hyper(at.va_filerev, tl); - tl += 2; - *tl++ = nfsrv_nfs_false; - *tl = nfsrv_nfs_true; - free((caddr_t)cookies, M_TEMP); - free((caddr_t)rbuf, M_TEMP); - error = 0; - goto nfsmout; - } - } - - /* - * Check for degenerate cases of nothing useful read. - * If so go try again - */ - cpos = rbuf; - cend = rbuf + siz; - dp = (struct dirent *)cpos; - cookiep = cookies; - /* - * For some reason FreeBSD's ufs_readdir() chooses to back the - * directory offset up to a block boundary, so it is necessary to - * skip over the records that precede the requested offset. This - * requires the assumption that file offset cookies monotonically - * increase. - */ - while (cpos < cend && ncookies > 0 && - (dp->d_fileno == 0 || dp->d_type == DT_WHT || - (is_ufs == 1 && ((u_quad_t)(*cookiep)) <= toff))) { - cpos += dp->d_reclen; - dp = (struct dirent *)cpos; - cookiep++; - ncookies--; - } - if (cpos >= cend || ncookies == 0) { - toff = off; - siz = fullsiz; - goto again; - } - - dirlen = len = NFSX_V3POSTOPATTR + NFSX_V3COOKIEVERF + - 2 * NFSX_UNSIGNED; - nfsm_reply(cnt); - nfsm_srvpostop_attr(getret, &at); - tl = nfsm_build(u_int32_t *, 2 * NFSX_UNSIGNED); - txdr_hyper(at.va_filerev, tl); - mp = mb; - bp = bpos; - be = bp + M_TRAILINGSPACE(mp); - - /* Loop through the records and build reply */ - while (cpos < cend && ncookies > 0) { - if (dp->d_fileno != 0 && dp->d_type != DT_WHT) { - nlen = dp->d_namlen; - rem = nfsm_rndup(nlen)-nlen; - - if (usevget) { - /* - * For readdir_and_lookup get the vnode using - * the file number. - */ - error = VFS_VGET(mntp, dp->d_fileno, LK_SHARED, - &nvp); - if (error != 0 && error != EOPNOTSUPP) { - error = 0; - goto invalid; - } else if (error == EOPNOTSUPP) { - /* - * VFS_VGET() not supported? - * Let's switch to VOP_LOOKUP(). - */ - error = 0; - usevget = 0; - cn.cn_nameiop = LOOKUP; - cn.cn_flags = ISLASTCN | NOFOLLOW | \ - LOCKSHARED | LOCKLEAF; - cn.cn_lkflags = LK_SHARED | LK_RETRY; - cn.cn_cred = cred; - cn.cn_thread = curthread; - } - } - if (!usevget) { - cn.cn_nameptr = dp->d_name; - cn.cn_namelen = dp->d_namlen; - if (dp->d_namlen == 2 && - dp->d_name[0] == '.' && - dp->d_name[1] == '.') { - cn.cn_flags |= ISDOTDOT; - } else { - cn.cn_flags &= ~ISDOTDOT; - } - if (!vp_locked) { - vn_lock(vp, LK_SHARED | LK_RETRY); - vp_locked = 1; - } - if ((vp->v_vflag & VV_ROOT) != 0 && - (cn.cn_flags & ISDOTDOT) != 0) { - vref(vp); - nvp = vp; - } else if (VOP_LOOKUP(vp, &nvp, &cn) != 0) - goto invalid; - } - - bzero((caddr_t)nfhp, NFSX_V3FH); - nfhp->fh_fsid = nvp->v_mount->mnt_stat.f_fsid; - if ((error1 = VOP_VPTOFH(nvp, &nfhp->fh_fid)) == 0) - error1 = VOP_GETATTR(nvp, vap, cred); - if (!usevget && vp == nvp) - vunref(nvp); - else - vput(nvp); - nvp = NULL; - if (error1 != 0) - goto invalid; - - /* - * If either the dircount or maxcount will be - * exceeded, get out now. Both of these lengths - * are calculated conservatively, including all - * XDR overheads. - */ - len += (8 * NFSX_UNSIGNED + nlen + rem + NFSX_V3FH + - NFSX_V3POSTOPATTR); - dirlen += (6 * NFSX_UNSIGNED + nlen + rem); - if (len > cnt || dirlen > fullsiz) { - eofflag = 0; - break; - } - - /* - * Build the directory record xdr from - * the dirent entry. - */ - fp = (struct nfs_fattr *)&fl.fl_fattr; - nfsm_srvfillattr(vap, fp); - fl.fl_fhsize = txdr_unsigned(NFSX_V3FH); - fl.fl_fhok = nfsrv_nfs_true; - fl.fl_postopok = nfsrv_nfs_true; - fl.fl_off.nfsuquad[0] = 0; - fl.fl_off.nfsuquad[1] = txdr_unsigned(*cookiep); - - nfsm_clget; - *tl = nfsrv_nfs_true; - bp += NFSX_UNSIGNED; - nfsm_clget; - *tl = 0; - bp += NFSX_UNSIGNED; - nfsm_clget; - *tl = txdr_unsigned(dp->d_fileno); - bp += NFSX_UNSIGNED; - nfsm_clget; - *tl = txdr_unsigned(nlen); - bp += NFSX_UNSIGNED; - - /* And loop around copying the name */ - xfer = nlen; - cp = dp->d_name; - while (xfer > 0) { - nfsm_clget; - if ((bp + xfer) > be) - tsiz = be - bp; - else - tsiz = xfer; - bcopy(cp, bp, tsiz); - bp += tsiz; - xfer -= tsiz; - if (xfer > 0) - cp += tsiz; - } - /* And null pad to an int32_t boundary. */ - for (i = 0; i < rem; i++) - *bp++ = '\0'; - - /* - * Now copy the flrep structure out. - */ - xfer = sizeof (struct flrep); - cp = (caddr_t)&fl; - while (xfer > 0) { - nfsm_clget; - if ((bp + xfer) > be) - tsiz = be - bp; - else - tsiz = xfer; - bcopy(cp, bp, tsiz); - bp += tsiz; - xfer -= tsiz; - if (xfer > 0) - cp += tsiz; - } - } -invalid: - cpos += dp->d_reclen; - dp = (struct dirent *)cpos; - cookiep++; - ncookies--; - } - if (!usevget && vp_locked) - vput(vp); - else - vrele(vp); - vp = NULL; - nfsm_clget; - *tl = nfsrv_nfs_false; - bp += NFSX_UNSIGNED; - nfsm_clget; - if (eofflag) - *tl = nfsrv_nfs_true; - else - *tl = nfsrv_nfs_false; - bp += NFSX_UNSIGNED; - if (mp != mb) { - if (bp < be) - mp->m_len = bp - mtod(mp, caddr_t); - } else - mp->m_len += bp - bpos; - free((caddr_t)cookies, M_TEMP); - free((caddr_t)rbuf, M_TEMP); -nfsmout: - if (vp) - vrele(vp); - if (mntp) - vfs_unbusy(mntp); - return(error); -} - -/* - * nfs commit service - */ -int -nfsrv_commit(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct vattr bfor, aft; - struct vnode *vp = NULL; - nfsfh_t nfh; - fhandle_t *fhp; - u_int32_t *tl; - caddr_t bpos; - int error = 0, rdonly, for_ret = 1, aft_ret = 1, cnt; - struct mbuf *mb, *mreq; - u_quad_t off; - struct mount *mp = NULL; - int v3 = (nfsd->nd_flag & ND_NFSV3); - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - if (!v3) - panic("nfsrv_commit: v3 proc called on a v2 connection"); - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - if ((mp = vfs_getvfs(&fhp->fh_fsid)) == NULL) { - error = ESTALE; - goto ereply; - } - (void) vn_start_write(NULL, &mp, V_WAIT); - vfs_rel(mp); /* The write holds a ref. */ - tl = nfsm_dissect_nonblock(u_int32_t *, 3 * NFSX_UNSIGNED); - - /* - * XXX At this time VOP_FSYNC() does not accept offset and byte - * count parameters, so these arguments are useless (someday maybe). - */ - off = fxdr_hyper(tl); - tl += 2; - cnt = fxdr_unsigned(int, *tl); - error = nfsrv_fhtovp(fhp, 0, &vp, nfsd, slp, nam, &rdonly); - if (error) { - nfsm_reply(2 * NFSX_UNSIGNED); - nfsm_srvwcc_data(for_ret, &bfor, aft_ret, &aft); - error = 0; - goto nfsmout; - } - for_ret = VOP_GETATTR(vp, &bfor, cred); - - /* - * RFC 1813 3.3.21: if count is 0, a flush from offset to the end of file - * is done. At this time VOP_FSYNC does not accept offset and byte count - * parameters so call VOP_FSYNC the whole file for now. - */ - if (cnt == 0 || cnt > MAX_COMMIT_COUNT) { - /* - * Give up and do the whole thing - */ - if (vp->v_object && - (vp->v_object->flags & OBJ_MIGHTBEDIRTY)) { - VM_OBJECT_WLOCK(vp->v_object); - vm_object_page_clean(vp->v_object, 0, 0, OBJPC_SYNC); - VM_OBJECT_WUNLOCK(vp->v_object); - } - error = VOP_FSYNC(vp, MNT_WAIT, curthread); - } else { - /* - * Locate and synchronously write any buffers that fall - * into the requested range. Note: we are assuming that - * f_iosize is a power of 2. - */ - int iosize = vp->v_mount->mnt_stat.f_iosize; - int iomask = iosize - 1; - struct bufobj *bo; - daddr_t lblkno; - - /* - * Align to iosize boundry, super-align to page boundry. - */ - if (off & iomask) { - cnt += off & iomask; - off &= ~(u_quad_t)iomask; - } - if (off & PAGE_MASK) { - cnt += off & PAGE_MASK; - off &= ~(u_quad_t)PAGE_MASK; - } - lblkno = off / iosize; - - if (vp->v_object && - (vp->v_object->flags & OBJ_MIGHTBEDIRTY)) { - VM_OBJECT_WLOCK(vp->v_object); - vm_object_page_clean(vp->v_object, off, off + cnt, - OBJPC_SYNC); - VM_OBJECT_WUNLOCK(vp->v_object); - } - - bo = &vp->v_bufobj; - BO_LOCK(bo); - while (cnt > 0) { - struct buf *bp; - - /* - * If we have a buffer and it is marked B_DELWRI we - * have to lock and write it. Otherwise the prior - * write is assumed to have already been committed. - * - * gbincore() can return invalid buffers now so we - * have to check that bit as well (though B_DELWRI - * should not be set if B_INVAL is set there could be - * a race here since we haven't locked the buffer). - */ - if ((bp = gbincore(&vp->v_bufobj, lblkno)) != NULL) { - if (BUF_LOCK(bp, LK_EXCLUSIVE | LK_SLEEPFAIL | - LK_INTERLOCK, BO_LOCKPTR(bo)) == ENOLCK) { - BO_LOCK(bo); - continue; /* retry */ - } - if ((bp->b_flags & (B_DELWRI|B_INVAL)) == - B_DELWRI) { - bremfree(bp); - bp->b_flags &= ~B_ASYNC; - bwrite(bp); - ++nfs_commit_miss; - } else - BUF_UNLOCK(bp); - BO_LOCK(bo); - } - ++nfs_commit_blks; - if (cnt < iosize) - break; - cnt -= iosize; - ++lblkno; - } - BO_UNLOCK(bo); - } - - aft_ret = VOP_GETATTR(vp, &aft, cred); - vput(vp); - vp = NULL; -ereply: - nfsm_reply(NFSX_V3WCCDATA + NFSX_V3WRITEVERF); - nfsm_srvwcc_data(for_ret, &bfor, aft_ret, &aft); - if (!error) { - tl = nfsm_build(u_int32_t *, NFSX_V3WRITEVERF); - if (nfsver.tv_sec == 0) - nfsver = boottime; - *tl++ = txdr_unsigned(nfsver.tv_sec); - *tl = txdr_unsigned(nfsver.tv_usec); - } else { - error = 0; - } -nfsmout: - if (vp) - vput(vp); - vn_finished_write(mp); - return(error); -} - -/* - * nfs statfs service - */ -int -nfsrv_statfs(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct statfs *sf; - struct nfs_statfs *sfp; - caddr_t bpos; - int error = 0, rdonly, getret = 1; - int v3 = (nfsd->nd_flag & ND_NFSV3); - struct mbuf *mb, *mreq; - struct vnode *vp = NULL; - struct vattr at; - nfsfh_t nfh; - fhandle_t *fhp; - struct statfs statfs; - u_quad_t tval; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - error = nfsrv_fhtovp(fhp, 0, &vp, nfsd, slp, nam, &rdonly); - if (error) { - nfsm_reply(NFSX_UNSIGNED); - if (v3) - nfsm_srvpostop_attr(getret, &at); - error = 0; - goto nfsmout; - } - sf = &statfs; - error = VFS_STATFS(vp->v_mount, sf); - getret = VOP_GETATTR(vp, &at, cred); - vput(vp); - vp = NULL; - nfsm_reply(NFSX_POSTOPATTR(v3) + NFSX_STATFS(v3)); - if (v3) - nfsm_srvpostop_attr(getret, &at); - if (error) { - error = 0; - goto nfsmout; - } - sfp = nfsm_build(struct nfs_statfs *, NFSX_STATFS(v3)); - if (v3) { - tval = (u_quad_t)sf->f_blocks; - tval *= (u_quad_t)sf->f_bsize; - txdr_hyper(tval, &sfp->sf_tbytes); - tval = (u_quad_t)sf->f_bfree; - tval *= (u_quad_t)sf->f_bsize; - txdr_hyper(tval, &sfp->sf_fbytes); - /* - * Don't send negative values for available space, - * since this field is unsigned in the NFS protocol. - * Otherwise, the client would see absurdly high - * numbers for free space. - */ - if (sf->f_bavail < 0) - tval = 0; - else - tval = (u_quad_t)sf->f_bavail; - tval *= (u_quad_t)sf->f_bsize; - txdr_hyper(tval, &sfp->sf_abytes); - sfp->sf_tfiles.nfsuquad[0] = 0; - sfp->sf_tfiles.nfsuquad[1] = txdr_unsigned(sf->f_files); - sfp->sf_ffiles.nfsuquad[0] = 0; - sfp->sf_ffiles.nfsuquad[1] = txdr_unsigned(sf->f_ffree); - sfp->sf_afiles.nfsuquad[0] = 0; - sfp->sf_afiles.nfsuquad[1] = txdr_unsigned(sf->f_ffree); - sfp->sf_invarsec = 0; - } else { - sfp->sf_tsize = txdr_unsigned(NFS_MAXDGRAMDATA); - sfp->sf_bsize = txdr_unsigned(sf->f_bsize); - sfp->sf_blocks = txdr_unsigned(sf->f_blocks); - sfp->sf_bfree = txdr_unsigned(sf->f_bfree); - if (sf->f_bavail < 0) - sfp->sf_bavail = 0; - else - sfp->sf_bavail = txdr_unsigned(sf->f_bavail); - } -nfsmout: - if (vp) - vput(vp); - return(error); -} - -/* - * nfs fsinfo service - */ -int -nfsrv_fsinfo(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct nfsv3_fsinfo *sip; - caddr_t bpos; - int error = 0, rdonly, getret = 1, pref; - struct mbuf *mb, *mreq; - struct vnode *vp = NULL; - struct vattr at; - nfsfh_t nfh; - fhandle_t *fhp; - u_quad_t maxfsize; - struct statfs sb; - int v3 = (nfsd->nd_flag & ND_NFSV3); - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - if (!v3) - panic("nfsrv_fsinfo: v3 proc called on a v2 connection"); - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - error = nfsrv_fhtovp(fhp, 0, &vp, nfsd, slp, nam, &rdonly); - if (error) { - nfsm_reply(NFSX_UNSIGNED); - nfsm_srvpostop_attr(getret, &at); - error = 0; - goto nfsmout; - } - - /* XXX Try to make a guess on the max file size. */ - VFS_STATFS(vp->v_mount, &sb); - maxfsize = (u_quad_t)0x80000000 * sb.f_bsize - 1; - - getret = VOP_GETATTR(vp, &at, cred); - vput(vp); - vp = NULL; - nfsm_reply(NFSX_V3POSTOPATTR + NFSX_V3FSINFO); - nfsm_srvpostop_attr(getret, &at); - sip = nfsm_build(struct nfsv3_fsinfo *, NFSX_V3FSINFO); - - /* - * XXX - * There should be filesystem VFS OP(s) to get this information. - * For now, assume ufs. - */ - pref = NFS_SRVMAXDATA(nfsd); - sip->fs_rtmax = txdr_unsigned(pref); - sip->fs_rtpref = txdr_unsigned(pref); - sip->fs_rtmult = txdr_unsigned(NFS_FABLKSIZE); - sip->fs_wtmax = txdr_unsigned(pref); - sip->fs_wtpref = txdr_unsigned(pref); - sip->fs_wtmult = txdr_unsigned(NFS_FABLKSIZE); - sip->fs_dtpref = txdr_unsigned(pref); - txdr_hyper(maxfsize, &sip->fs_maxfilesize); - sip->fs_timedelta.nfsv3_sec = 0; - sip->fs_timedelta.nfsv3_nsec = txdr_unsigned(1); - sip->fs_properties = txdr_unsigned(NFSV3FSINFO_LINK | - NFSV3FSINFO_SYMLINK | NFSV3FSINFO_HOMOGENEOUS | - NFSV3FSINFO_CANSETTIME); -nfsmout: - if (vp) - vput(vp); - return(error); -} - -/* - * nfs pathconf service - */ -int -nfsrv_pathconf(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep, *md = nfsd->nd_md; - struct sockaddr *nam = nfsd->nd_nam; - caddr_t dpos = nfsd->nd_dpos; - struct ucred *cred = nfsd->nd_cr; - struct nfsv3_pathconf *pc; - caddr_t bpos; - int error = 0, rdonly, getret = 1; - register_t linkmax, namemax, chownres, notrunc; - struct mbuf *mb, *mreq; - struct vnode *vp = NULL; - struct vattr at; - nfsfh_t nfh; - fhandle_t *fhp; - int v3 = (nfsd->nd_flag & ND_NFSV3); - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - if (!v3) - panic("nfsrv_pathconf: v3 proc called on a v2 connection"); - fhp = &nfh.fh_generic; - nfsm_srvmtofh(fhp); - error = nfsrv_fhtovp(fhp, 0, &vp, nfsd, slp, nam, &rdonly); - if (error) { - nfsm_reply(NFSX_UNSIGNED); - nfsm_srvpostop_attr(getret, &at); - error = 0; - goto nfsmout; - } - error = VOP_PATHCONF(vp, _PC_LINK_MAX, &linkmax); - if (!error) - error = VOP_PATHCONF(vp, _PC_NAME_MAX, &namemax); - if (!error) - error = VOP_PATHCONF(vp, _PC_CHOWN_RESTRICTED, &chownres); - if (!error) - error = VOP_PATHCONF(vp, _PC_NO_TRUNC, ¬runc); - getret = VOP_GETATTR(vp, &at, cred); - vput(vp); - vp = NULL; - nfsm_reply(NFSX_V3POSTOPATTR + NFSX_V3PATHCONF); - nfsm_srvpostop_attr(getret, &at); - if (error) { - error = 0; - goto nfsmout; - } - pc = nfsm_build(struct nfsv3_pathconf *, NFSX_V3PATHCONF); - - pc->pc_linkmax = txdr_unsigned(linkmax); - pc->pc_namemax = txdr_unsigned(namemax); - pc->pc_notrunc = txdr_unsigned(notrunc); - pc->pc_chownrestricted = txdr_unsigned(chownres); - - /* - * These should probably be supported by VOP_PATHCONF(), but - * until msdosfs is exportable (why would you want to?), the - * Unix defaults should be ok. - */ - pc->pc_caseinsensitive = nfsrv_nfs_false; - pc->pc_casepreserving = nfsrv_nfs_true; -nfsmout: - if (vp) - vput(vp); - return(error); -} - -/* - * Null operation, used by clients to ping server - */ -/* ARGSUSED */ -int -nfsrv_null(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep; - caddr_t bpos; - int error = NFSERR_RETVOID; - struct mbuf *mb, *mreq; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - nfsm_reply(0); -nfsmout: - return (error); -} - -/* - * No operation, used for obsolete procedures - */ -/* ARGSUSED */ -int -nfsrv_noop(struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct mbuf **mrq) -{ - struct mbuf *mrep = nfsd->nd_mrep; - caddr_t bpos; - int error; - struct mbuf *mb, *mreq; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - if (nfsd->nd_repstat) - error = nfsd->nd_repstat; - else - error = EPROCUNAVAIL; - nfsm_reply(0); - error = 0; -nfsmout: - return (error); -} - -/* - * Perform access checking for vnodes obtained from file handles that would - * refer to files already opened by a Unix client. You cannot just use - * vn_writechk() and VOP_ACCESS() for two reasons. - * 1 - You must check for exported rdonly as well as MNT_RDONLY for the write - * case. - * 2 - The owner is to be given access irrespective of mode bits for some - * operations, so that processes that chmod after opening a file don't - * break. I don't like this because it opens a security hole, but since - * the nfs server opens a security hole the size of a barn door anyhow, - * what the heck. - * - * The exception to rule 2 is EPERM. If a file is IMMUTABLE, VOP_ACCESS() - * will return EPERM instead of EACCES. EPERM is always an error. - */ -static int -nfsrv_access(struct vnode *vp, accmode_t accmode, struct ucred *cred, - int rdonly, int override) -{ - struct vattr vattr; - int error; - - nfsdbprintf(("%s %d\n", __FILE__, __LINE__)); - - if (accmode & VWRITE) { - /* Just vn_writechk() changed to check rdonly */ - /* - * Disallow write attempts on read-only filesystems; - * unless the file is a socket or a block or character - * device resident on the filesystem. - */ - if (rdonly || (vp->v_mount->mnt_flag & MNT_RDONLY)) { - switch (vp->v_type) { - case VREG: - case VDIR: - case VLNK: - return (EROFS); - default: - break; - } - } - /* - * If there's shared text associated with - * the inode, we can't allow writing. - */ - if (VOP_IS_TEXT(vp)) - return (ETXTBSY); - } - - error = VOP_GETATTR(vp, &vattr, cred); - if (error) - return (error); - error = VOP_ACCESS(vp, accmode, cred, curthread); - /* - * Allow certain operations for the owner (reads and writes - * on files that are already open). - */ - if (override && error == EACCES && cred->cr_uid == vattr.va_uid) - error = 0; - return (error); -} diff --git a/sys/nfsserver/nfs_srvkrpc.c b/sys/nfsserver/nfs_srvkrpc.c deleted file mode 100644 index 8bd045a..0000000 --- a/sys/nfsserver/nfs_srvkrpc.c +++ /dev/null @@ -1,545 +0,0 @@ -/*- - * Copyright (c) 1989, 1993 - * The Regents of the University of California. All rights reserved. - * - * This code is derived from software contributed to Berkeley by - * Rick Macklem at The University of Guelph. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 4. Neither the name of the University nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * @(#)nfs_syscalls.c 8.5 (Berkeley) 3/30/95 - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -#include "opt_inet6.h" -#include "opt_kgssapi.h" - -#include <sys/param.h> -#include <sys/capsicum.h> -#include <sys/systm.h> -#include <sys/sysproto.h> -#include <sys/kernel.h> -#include <sys/sysctl.h> -#include <sys/file.h> -#include <sys/filedesc.h> -#include <sys/jail.h> -#include <sys/vnode.h> -#include <sys/malloc.h> -#include <sys/mount.h> -#include <sys/priv.h> -#include <sys/proc.h> -#include <sys/bio.h> -#include <sys/buf.h> -#include <sys/mbuf.h> -#include <sys/socket.h> -#include <sys/socketvar.h> -#include <sys/domain.h> -#include <sys/protosw.h> -#include <sys/namei.h> -#include <sys/fcntl.h> -#include <sys/lockf.h> -#include <sys/eventhandler.h> - -#include <netinet/in.h> -#include <netinet/tcp.h> -#ifdef INET6 -#include <net/if.h> -#include <net/if_var.h> /* XXX: for in6_var.h */ -#include <netinet6/in6_var.h> /* XXX: for ip6_sprintf */ -#endif - -#include <rpc/rpc.h> -#include <rpc/rpcsec_gss.h> -#include <rpc/replay.h> - -#include <nfs/xdr_subs.h> -#include <nfs/nfsproto.h> -#include <nfs/nfs_fha.h> -#include <nfsserver/nfs.h> -#include <nfsserver/nfsm_subs.h> -#include <nfsserver/nfsrvcache.h> -#include <nfsserver/nfs_fha_old.h> - -#include <security/mac/mac_framework.h> - -static MALLOC_DEFINE(M_NFSSVC, "nfss_srvsock", "Nfs server structure"); - -MALLOC_DEFINE(M_NFSRVDESC, "nfss_srvdesc", "NFS server socket descriptor"); -MALLOC_DEFINE(M_NFSD, "nfss_daemon", "Nfs server daemon structure"); - -#define TRUE 1 -#define FALSE 0 - -SYSCTL_DECL(_vfs_nfsrv); - -SVCPOOL *nfsrv_pool; -int nfsd_waiting = 0; -int nfsrv_numnfsd = 0; -struct callout nfsrv_callout; -static eventhandler_tag nfsrv_nmbclusters_tag; - -static int nfs_privport = 0; -SYSCTL_INT(_vfs_nfsrv, NFS_NFSPRIVPORT, nfs_privport, CTLFLAG_RW, - &nfs_privport, 0, - "Only allow clients using a privileged port"); -SYSCTL_INT(_vfs_nfsrv, OID_AUTO, gatherdelay, CTLFLAG_RW, - &nfsrvw_procrastinate, 0, - "Delay value for write gathering"); -SYSCTL_INT(_vfs_nfsrv, OID_AUTO, gatherdelay_v3, CTLFLAG_RW, - &nfsrvw_procrastinate_v3, 0, - "Delay in seconds for NFSv3 write gathering"); - -static int nfssvc_addsock(struct file *, struct thread *); -static int nfssvc_nfsd(struct thread *, struct nfsd_nfsd_args *); - -extern u_long sb_max_adj; - -int32_t (*nfsrv3_procs[NFS_NPROCS])(struct nfsrv_descript *nd, - struct nfssvc_sock *slp, struct mbuf **mreqp) = { - nfsrv_null, - nfsrv_getattr, - nfsrv_setattr, - nfsrv_lookup, - nfsrv3_access, - nfsrv_readlink, - nfsrv_read, - nfsrv_write, - nfsrv_create, - nfsrv_mkdir, - nfsrv_symlink, - nfsrv_mknod, - nfsrv_remove, - nfsrv_rmdir, - nfsrv_rename, - nfsrv_link, - nfsrv_readdir, - nfsrv_readdirplus, - nfsrv_statfs, - nfsrv_fsinfo, - nfsrv_pathconf, - nfsrv_commit, - nfsrv_noop -}; - -/* - * NFS server system calls - */ -/* - * This is now called from nfssvc() in nfs/nfs_nfssvc.c. - */ - -/* - * Nfs server psuedo system call for the nfsd's - * Based on the flag value it either: - * - adds a socket to the selection list - * - remains in the kernel as an nfsd - * - remains in the kernel as an nfsiod - * For INET6 we suppose that nfsd provides only IN6P_IPV6_V6ONLY sockets - * and that mountd provides - * - sockaddr with no IPv4-mapped addresses - * - mask for both INET and INET6 families if there is IPv4-mapped overlap - */ -int -nfssvc_nfsserver(struct thread *td, struct nfssvc_args *uap) -{ - struct file *fp; - struct nfsd_addsock_args addsockarg; - struct nfsd_nfsd_args nfsdarg; - cap_rights_t rights; - int error; - - if (uap->flag & NFSSVC_ADDSOCK) { - error = copyin(uap->argp, (caddr_t)&addsockarg, - sizeof(addsockarg)); - if (error) - return (error); - error = fget(td, addsockarg.sock, - cap_rights_init(&rights, CAP_SOCK_SERVER), &fp); - if (error) - return (error); - if (fp->f_type != DTYPE_SOCKET) { - fdrop(fp, td); - return (error); /* XXXRW: Should be EINVAL? */ - } - error = nfssvc_addsock(fp, td); - fdrop(fp, td); - } else if (uap->flag & NFSSVC_OLDNFSD) - error = nfssvc_nfsd(td, NULL); - else if (uap->flag & NFSSVC_NFSD) { - if (!uap->argp) - return (EINVAL); - error = copyin(uap->argp, (caddr_t)&nfsdarg, - sizeof(nfsdarg)); - if (error) - return (error); - error = nfssvc_nfsd(td, &nfsdarg); - } else - error = ENXIO; - return (error); -} - -/* - * Generate the rpc reply header - * siz arg. is used to decide if adding a cluster is worthwhile - */ -struct mbuf * -nfs_rephead(int siz, struct nfsrv_descript *nd, int err, - struct mbuf **mbp, caddr_t *bposp) -{ - u_int32_t *tl; - struct mbuf *mreq; - caddr_t bpos; - struct mbuf *mb; - - if (err == EBADRPC) - return (NULL); - - nd->nd_repstat = err; - if (err && (nd->nd_flag & ND_NFSV3) == 0) /* XXX recheck */ - siz = 0; - - MGET(mreq, M_WAITOK, MT_DATA); - - /* - * If this is a big reply, use a cluster - */ - mreq->m_len = 0; - if (siz >= MINCLSIZE) { - MCLGET(mreq, M_WAITOK); - } - mb = mreq; - bpos = mtod(mb, caddr_t); - - if (err != NFSERR_RETVOID) { - tl = nfsm_build(u_int32_t *, NFSX_UNSIGNED); - if (err) - *tl = txdr_unsigned(nfsrv_errmap(nd, err)); - else - *tl = 0; - } - - *mbp = mb; - *bposp = bpos; - if (err != 0 && err != NFSERR_RETVOID) - nfsrvstats.srvrpc_errs++; - - return (mreq); -} - -static void -nfssvc_program(struct svc_req *rqst, SVCXPRT *xprt) -{ - rpcproc_t procnum; - int32_t (*proc)(struct nfsrv_descript *nd, struct nfssvc_sock *slp, - struct mbuf **mreqp); - int flag; - struct nfsrv_descript nd; - struct mbuf *mreq, *mrep; - int error; - - if (rqst->rq_vers == NFS_VER2) { - if (rqst->rq_proc > NFSV2PROC_STATFS) { - svcerr_noproc(rqst); - svc_freereq(rqst); - return; - } - procnum = nfsrv_nfsv3_procid[rqst->rq_proc]; - flag = 0; - } else { - if (rqst->rq_proc >= NFS_NPROCS) { - svcerr_noproc(rqst); - svc_freereq(rqst); - return; - } - procnum = rqst->rq_proc; - flag = ND_NFSV3; - } - proc = nfsrv3_procs[procnum]; - - mreq = mrep = NULL; - mreq = rqst->rq_args; - rqst->rq_args = NULL; - (void)nfs_realign(&mreq, M_WAITOK); - - /* - * Note: we want rq_addr, not svc_getrpccaller for nd_nam2 - - * NFS_SRVMAXDATA uses a NULL value for nd_nam2 to detect TCP - * mounts. - */ - memset(&nd, 0, sizeof(nd)); - nd.nd_md = nd.nd_mrep = mreq; - nd.nd_dpos = mtod(mreq, caddr_t); - nd.nd_nam = svc_getrpccaller(rqst); - nd.nd_nam2 = rqst->rq_addr; - nd.nd_procnum = procnum; - nd.nd_cr = NULL; - nd.nd_flag = flag; - - if (nfs_privport) { - /* Check if source port is privileged */ - u_short port; - struct sockaddr *nam = nd.nd_nam; - struct sockaddr_in *sin; - - sin = (struct sockaddr_in *)nam; - /* - * INET/INET6 - same code: - * sin_port and sin6_port are at same offset - */ - port = ntohs(sin->sin_port); - if (port >= IPPORT_RESERVED && - nd.nd_procnum != NFSPROC_NULL) { -#ifdef INET6 - char b6[INET6_ADDRSTRLEN]; -#if defined(KLD_MODULE) - /* Do not use ip6_sprintf: the nfs module should work without INET6. */ -#define ip6_sprintf(buf, a) \ - (sprintf((buf), "%x:%x:%x:%x:%x:%x:%x:%x", \ - (a)->s6_addr16[0], (a)->s6_addr16[1], \ - (a)->s6_addr16[2], (a)->s6_addr16[3], \ - (a)->s6_addr16[4], (a)->s6_addr16[5], \ - (a)->s6_addr16[6], (a)->s6_addr16[7]), \ - (buf)) -#endif -#endif - printf("NFS request from unprivileged port (%s:%d)\n", -#ifdef INET6 - sin->sin_family == AF_INET6 ? - ip6_sprintf(b6, &satosin6(sin)->sin6_addr) : -#if defined(KLD_MODULE) -#undef ip6_sprintf -#endif -#endif - inet_ntoa(sin->sin_addr), port); - m_freem(mreq); - svcerr_weakauth(rqst); - svc_freereq(rqst); - return; - } - } - - if (proc != nfsrv_null) { - if (!svc_getcred(rqst, &nd.nd_cr, &nd.nd_credflavor)) { - m_freem(mreq); - svcerr_weakauth(rqst); - svc_freereq(rqst); - return; - } -#ifdef MAC - mac_cred_associate_nfsd(nd.nd_cr); -#endif - } - nfsrvstats.srvrpccnt[nd.nd_procnum]++; - - error = proc(&nd, NULL, &mrep); - - if (nd.nd_cr) - crfree(nd.nd_cr); - - if (mrep == NULL) { - svcerr_decode(rqst); - svc_freereq(rqst); - return; - } - if (error && error != NFSERR_RETVOID) { - svcerr_systemerr(rqst); - svc_freereq(rqst); - return; - } - if (nd.nd_repstat & NFSERR_AUTHERR) { - svcerr_auth(rqst, nd.nd_repstat & ~NFSERR_AUTHERR); - m_freem(mrep); - } else { - if (!svc_sendreply_mbuf(rqst, mrep)) - svcerr_systemerr(rqst); - } - svc_freereq(rqst); -} - -/* - * Adds a socket to the list for servicing by nfsds. - */ -static int -nfssvc_addsock(struct file *fp, struct thread *td) -{ - int siz; - struct socket *so; - int error; - SVCXPRT *xprt; - - so = fp->f_data; - - siz = sb_max_adj; - error = soreserve(so, siz, siz); - if (error) - return (error); - - /* - * Steal the socket from userland so that it doesn't close - * unexpectedly. - */ - if (so->so_type == SOCK_DGRAM) - xprt = svc_dg_create(nfsrv_pool, so, 0, 0); - else - xprt = svc_vc_create(nfsrv_pool, so, 0, 0); - if (xprt) { - fp->f_ops = &badfileops; - fp->f_data = NULL; - svc_reg(xprt, NFS_PROG, NFS_VER2, nfssvc_program, NULL); - svc_reg(xprt, NFS_PROG, NFS_VER3, nfssvc_program, NULL); - SVC_RELEASE(xprt); - } - - return (0); -} - -/* - * Called by nfssvc() for nfsds. Just loops around servicing rpc requests - * until it is killed by a signal. - */ -static int -nfssvc_nfsd(struct thread *td, struct nfsd_nfsd_args *args) -{ - char principal[128]; - int error; - - if (args) { - error = copyinstr(args->principal, principal, - sizeof(principal), NULL); - if (error) - return (error); - } else { - memcpy(principal, "nfs@", 4); - getcredhostname(td->td_ucred, principal + 4, - sizeof(principal) - 4); - } - - /* - * Only the first nfsd actually does any work. The RPC code - * adds threads to it as needed. Any extra processes offered - * by nfsd just exit. If nfsd is new enough, it will call us - * once with a structure that specifies how many threads to - * use. - */ - NFSD_LOCK(); - if (nfsrv_numnfsd == 0) { - nfsrv_numnfsd++; - - NFSD_UNLOCK(); - - rpc_gss_set_svc_name_call(principal, "kerberosv5", - GSS_C_INDEFINITE, NFS_PROG, NFS_VER2); - rpc_gss_set_svc_name_call(principal, "kerberosv5", - GSS_C_INDEFINITE, NFS_PROG, NFS_VER3); - - if (args) { - nfsrv_pool->sp_minthreads = args->minthreads; - nfsrv_pool->sp_maxthreads = args->maxthreads; - } else { - nfsrv_pool->sp_minthreads = 4; - nfsrv_pool->sp_maxthreads = 4; - } - - svc_run(nfsrv_pool); - - rpc_gss_clear_svc_name_call(NFS_PROG, NFS_VER2); - rpc_gss_clear_svc_name_call(NFS_PROG, NFS_VER3); - - NFSD_LOCK(); - nfsrv_numnfsd--; - nfsrv_init(TRUE); - } - NFSD_UNLOCK(); - - return (0); -} - -/* - * Size the NFS server's duplicate request cache at 1/2 the - * nmbclusters, floating within a (64, 2048) range. This is to - * prevent all mbuf clusters being tied up in the NFS dupreq - * cache for small values of nmbclusters. - */ -static size_t -nfsrv_replay_size(void) -{ - size_t replaysiz; - - replaysiz = nmbclusters / 2; - if (replaysiz > NFSRVCACHE_MAX_SIZE) - replaysiz = NFSRVCACHE_MAX_SIZE; - if (replaysiz < NFSRVCACHE_MIN_SIZE) - replaysiz = NFSRVCACHE_MIN_SIZE; - replaysiz *= MCLBYTES; - - return (replaysiz); -} - -/* - * Called when nmbclusters changes - we resize the replay cache - * accordingly. - */ -static void -nfsrv_nmbclusters_change(void *tag) -{ - - if (nfsrv_pool) - replay_setsize(nfsrv_pool->sp_rcache, nfsrv_replay_size()); -} - -/* - * Initialize the data structures for the server. - * Handshake with any new nfsds starting up to avoid any chance of - * corruption. - */ -void -nfsrv_init(int terminating) -{ - - NFSD_LOCK_ASSERT(); - - if (terminating) { - NFSD_UNLOCK(); - EVENTHANDLER_DEREGISTER(nmbclusters_change, - nfsrv_nmbclusters_tag); - svcpool_destroy(nfsrv_pool); - nfsrv_pool = NULL; - NFSD_LOCK(); - } else - nfs_pub.np_valid = 0; - - NFSD_UNLOCK(); - - nfsrv_pool = svcpool_create("nfsd", SYSCTL_STATIC_CHILDREN(_vfs_nfsrv)); - nfsrv_pool->sp_rcache = replay_newcache(nfsrv_replay_size()); - nfsrv_pool->sp_assign = fhaold_assign; - nfsrv_pool->sp_done = fha_nd_complete; - nfsrv_nmbclusters_tag = EVENTHANDLER_REGISTER(nmbclusters_change, - nfsrv_nmbclusters_change, NULL, EVENTHANDLER_PRI_FIRST); - - NFSD_LOCK(); -} diff --git a/sys/nfsserver/nfs_srvsubs.c b/sys/nfsserver/nfs_srvsubs.c deleted file mode 100644 index 04371da..0000000 --- a/sys/nfsserver/nfs_srvsubs.c +++ /dev/null @@ -1,1418 +0,0 @@ -/*- - * Copyright (c) 1989, 1993 - * The Regents of the University of California. All rights reserved. - * - * This code is derived from software contributed to Berkeley by - * Rick Macklem at The University of Guelph. - * - * Redistribution and use in source and binary forms, with or without - * modification, are permitted provided that the following conditions - * are met: - * 1. Redistributions of source code must retain the above copyright - * notice, this list of conditions and the following disclaimer. - * 2. Redistributions in binary form must reproduce the above copyright - * notice, this list of conditions and the following disclaimer in the - * documentation and/or other materials provided with the distribution. - * 4. Neither the name of the University nor the names of its contributors - * may be used to endorse or promote products derived from this software - * without specific prior written permission. - * - * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND - * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE - * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE - * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE - * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL - * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS - * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) - * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT - * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY - * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF - * SUCH DAMAGE. - * - * @(#)nfs_subs.c 8.8 (Berkeley) 5/22/95 - */ - -#include <sys/cdefs.h> -__FBSDID("$FreeBSD$"); - -/* - * These functions support the macros and help fiddle mbuf chains for - * the nfs op functions. They do things like create the rpc header and - * copy data between mbuf chains and uio lists. - */ - -#include "opt_inet6.h" - -#include <sys/param.h> -#include <sys/systm.h> -#include <sys/kernel.h> -#include <sys/bio.h> -#include <sys/buf.h> -#include <sys/proc.h> -#include <sys/mount.h> -#include <sys/vnode.h> -#include <sys/namei.h> -#include <sys/mbuf.h> -#include <sys/refcount.h> -#include <sys/socket.h> -#include <sys/stat.h> -#include <sys/malloc.h> -#include <sys/module.h> -#include <sys/sysent.h> -#include <sys/syscall.h> -#include <sys/sysproto.h> - -#include <vm/vm.h> -#include <vm/vm_object.h> -#include <vm/vm_extern.h> -#include <vm/uma.h> - -#include <rpc/rpc.h> - -#include <nfs/nfsproto.h> -#include <nfsserver/nfs.h> -#include <nfs/xdr_subs.h> -#include <nfsserver/nfsm_subs.h> - -#include <netinet/in.h> - -/* - * Data items converted to xdr at startup, since they are constant - * This is kinda hokey, but may save a little time doing byte swaps - */ -u_int32_t nfsrv_nfs_xdrneg1; -u_int32_t nfsrv_nfs_true, nfsrv_nfs_false; - -/* And other global data */ -static const nfstype nfsv2_type[9] = { NFNON, NFREG, NFDIR, NFBLK, NFCHR, - NFLNK, NFNON, NFCHR, NFNON }; -#define vtonfsv2_type(a) txdr_unsigned(nfsv2_type[((int32_t)(a))]) -#define vtonfsv3_mode(m) txdr_unsigned((m) & ALLPERMS) - -int nfsrv_ticks; - -struct mtx nfsd_mtx; - -/* - * Mapping of old NFS Version 2 RPC numbers to generic numbers. - */ -const int nfsrv_nfsv3_procid[NFS_NPROCS] = { - NFSPROC_NULL, - NFSPROC_GETATTR, - NFSPROC_SETATTR, - NFSPROC_NOOP, - NFSPROC_LOOKUP, - NFSPROC_READLINK, - NFSPROC_READ, - NFSPROC_NOOP, - NFSPROC_WRITE, - NFSPROC_CREATE, - NFSPROC_REMOVE, - NFSPROC_RENAME, - NFSPROC_LINK, - NFSPROC_SYMLINK, - NFSPROC_MKDIR, - NFSPROC_RMDIR, - NFSPROC_READDIR, - NFSPROC_FSSTAT, - NFSPROC_NOOP, - NFSPROC_NOOP, - NFSPROC_NOOP, - NFSPROC_NOOP, - NFSPROC_NOOP, -}; - -/* - * and the reverse mapping from generic to Version 2 procedure numbers - */ -const int nfsrvv2_procid[NFS_NPROCS] = { - NFSV2PROC_NULL, - NFSV2PROC_GETATTR, - NFSV2PROC_SETATTR, - NFSV2PROC_LOOKUP, - NFSV2PROC_NOOP, - NFSV2PROC_READLINK, - NFSV2PROC_READ, - NFSV2PROC_WRITE, - NFSV2PROC_CREATE, - NFSV2PROC_MKDIR, - NFSV2PROC_SYMLINK, - NFSV2PROC_CREATE, - NFSV2PROC_REMOVE, - NFSV2PROC_RMDIR, - NFSV2PROC_RENAME, - NFSV2PROC_LINK, - NFSV2PROC_READDIR, - NFSV2PROC_NOOP, - NFSV2PROC_STATFS, - NFSV2PROC_NOOP, - NFSV2PROC_NOOP, - NFSV2PROC_NOOP, - NFSV2PROC_NOOP, -}; - -/* - * Maps errno values to nfs error numbers. - * Use 0 (which gets converted to NFSERR_IO) as the catch all for ones not - * specifically defined in RFC 1094. - */ -static const u_char nfsrv_v2errmap[ELAST] = { - NFSERR_PERM, NFSERR_NOENT, 0, 0, 0, - NFSERR_NXIO, 0, 0, 0, 0, - 0, 0, NFSERR_ACCES, 0, 0, - 0, NFSERR_EXIST, 0, NFSERR_NODEV, NFSERR_NOTDIR, - NFSERR_ISDIR, 0, 0, 0, 0, - 0, NFSERR_FBIG, NFSERR_NOSPC, 0, NFSERR_ROFS, - 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, - 0, 0, 0, 0, 0, - 0, 0, NFSERR_NAMETOL, 0, 0, - NFSERR_NOTEMPTY, 0, 0, NFSERR_DQUOT, NFSERR_STALE, - 0 -}; - -/* - * Maps errno values to nfs error numbers. - * Although it is not obvious whether or not NFS clients really care if - * a returned error value is in the specified list for the procedure, the - * safest thing to do is filter them appropriately. For Version 2, the - * X/Open XNFS document is the only specification that defines error values - * for each RPC (The RFC simply lists all possible error values for all RPCs), - * so I have decided to not do this for Version 2. - * The first entry is the default error return and the rest are the valid - * errors for that RPC in increasing numeric order. - */ -static const short nfsv3err_null[] = { - 0, - 0, -}; - -static const short nfsv3err_getattr[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_setattr[] = { - NFSERR_IO, - NFSERR_PERM, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_INVAL, - NFSERR_NOSPC, - NFSERR_ROFS, - NFSERR_DQUOT, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_NOT_SYNC, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_lookup[] = { - NFSERR_IO, - NFSERR_NOENT, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_NOTDIR, - NFSERR_NAMETOL, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_access[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_readlink[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_INVAL, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_NOTSUPP, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_read[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_NXIO, - NFSERR_ACCES, - NFSERR_INVAL, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_write[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_INVAL, - NFSERR_FBIG, - NFSERR_NOSPC, - NFSERR_ROFS, - NFSERR_DQUOT, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_create[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_EXIST, - NFSERR_NOTDIR, - NFSERR_NOSPC, - NFSERR_ROFS, - NFSERR_NAMETOL, - NFSERR_DQUOT, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_NOTSUPP, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_mkdir[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_EXIST, - NFSERR_NOTDIR, - NFSERR_NOSPC, - NFSERR_ROFS, - NFSERR_NAMETOL, - NFSERR_DQUOT, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_NOTSUPP, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_symlink[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_EXIST, - NFSERR_NOTDIR, - NFSERR_NOSPC, - NFSERR_ROFS, - NFSERR_NAMETOL, - NFSERR_DQUOT, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_NOTSUPP, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_mknod[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_EXIST, - NFSERR_NOTDIR, - NFSERR_NOSPC, - NFSERR_ROFS, - NFSERR_NAMETOL, - NFSERR_DQUOT, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_NOTSUPP, - NFSERR_SERVERFAULT, - NFSERR_BADTYPE, - 0, -}; - -static const short nfsv3err_remove[] = { - NFSERR_IO, - NFSERR_NOENT, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_NOTDIR, - NFSERR_ROFS, - NFSERR_NAMETOL, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_rmdir[] = { - NFSERR_IO, - NFSERR_NOENT, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_EXIST, - NFSERR_NOTDIR, - NFSERR_INVAL, - NFSERR_ROFS, - NFSERR_NAMETOL, - NFSERR_NOTEMPTY, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_NOTSUPP, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_rename[] = { - NFSERR_IO, - NFSERR_NOENT, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_EXIST, - NFSERR_XDEV, - NFSERR_NOTDIR, - NFSERR_ISDIR, - NFSERR_INVAL, - NFSERR_NOSPC, - NFSERR_ROFS, - NFSERR_MLINK, - NFSERR_NAMETOL, - NFSERR_NOTEMPTY, - NFSERR_DQUOT, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_NOTSUPP, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_link[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_EXIST, - NFSERR_XDEV, - NFSERR_NOTDIR, - NFSERR_INVAL, - NFSERR_NOSPC, - NFSERR_ROFS, - NFSERR_MLINK, - NFSERR_NAMETOL, - NFSERR_DQUOT, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_NOTSUPP, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_readdir[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_NOTDIR, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_BAD_COOKIE, - NFSERR_TOOSMALL, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_readdirplus[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_ACCES, - NFSERR_NOTDIR, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_BAD_COOKIE, - NFSERR_NOTSUPP, - NFSERR_TOOSMALL, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_fsstat[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_fsinfo[] = { - NFSERR_STALE, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_pathconf[] = { - NFSERR_STALE, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_SERVERFAULT, - 0, -}; - -static const short nfsv3err_commit[] = { - NFSERR_IO, - NFSERR_IO, - NFSERR_STALE, - NFSERR_BADHANDLE, - NFSERR_SERVERFAULT, - 0, -}; - -static const short *nfsrv_v3errmap[] = { - nfsv3err_null, - nfsv3err_getattr, - nfsv3err_setattr, - nfsv3err_lookup, - nfsv3err_access, - nfsv3err_readlink, - nfsv3err_read, - nfsv3err_write, - nfsv3err_create, - nfsv3err_mkdir, - nfsv3err_symlink, - nfsv3err_mknod, - nfsv3err_remove, - nfsv3err_rmdir, - nfsv3err_rename, - nfsv3err_link, - nfsv3err_readdir, - nfsv3err_readdirplus, - nfsv3err_fsstat, - nfsv3err_fsinfo, - nfsv3err_pathconf, - nfsv3err_commit, -}; - -extern int (*nfsd_call_nfsserver)(struct thread *, struct nfssvc_args *); - -/* - * Called once to initialize data structures... - */ -static int -nfsrv_modevent(module_t mod, int type, void *data) -{ - int error = 0; - - switch (type) { - case MOD_LOAD: - mtx_init(&nfsd_mtx, "nfsd_mtx", NULL, MTX_DEF); - nfsrv_nfs_true = txdr_unsigned(TRUE); - nfsrv_nfs_false = txdr_unsigned(FALSE); - nfsrv_nfs_xdrneg1 = txdr_unsigned(-1); - nfsrv_ticks = (hz * NFS_TICKINTVL + 500) / 1000; - if (nfsrv_ticks < 1) - nfsrv_ticks = 1; - - NFSD_LOCK(); - nfsrv_init(0); /* Init server data structures */ - NFSD_UNLOCK(); - - nfsd_call_nfsserver = nfssvc_nfsserver; - break; - - case MOD_UNLOAD: - if (nfsrv_numnfsd != 0) { - error = EBUSY; - break; - } - - nfsd_call_nfsserver = NULL; - callout_drain(&nfsrv_callout); - mtx_destroy(&nfsd_mtx); - break; - default: - error = EOPNOTSUPP; - break; - } - return error; -} -static moduledata_t nfsserver_mod = { - "nfsserver", - nfsrv_modevent, - NULL, -}; -DECLARE_MODULE(nfsserver, nfsserver_mod, SI_SUB_VFS, SI_ORDER_ANY); - -/* So that loader and kldload(2) can find us, wherever we are.. */ -MODULE_VERSION(nfsserver, 1); -MODULE_DEPEND(nfsserver, nfssvc, 1, 1, 1); -MODULE_DEPEND(nfsserver, krpc, 1, 1, 1); -MODULE_DEPEND(nfsserver, nfs_common, 1, 1, 1); - -/* - * Set up nameidata for a lookup() call and do it. - * - * If pubflag is set, this call is done for a lookup operation on the - * public filehandle. In that case we allow crossing mountpoints and - * absolute pathnames. However, the caller is expected to check that - * the lookup result is within the public fs, and deny access if - * it is not. - * - * nfs_namei() clears out garbage fields that namei() might leave garbage. - * This is mainly ni_vp and ni_dvp when an error occurs, and ni_dvp when no - * error occurs but the parent was not requested. - * - * dirp may be set whether an error is returned or not, and must be - * released by the caller. - */ -int -nfs_namei(struct nameidata *ndp, struct nfsrv_descript *nfsd, - fhandle_t *fhp, int len, struct nfssvc_sock *slp, - struct sockaddr *nam, struct mbuf **mdp, - caddr_t *dposp, struct vnode **retdirp, int v3, struct vattr *retdirattrp, - int *retdirattr_retp, int pubflag) -{ - int i, rem; - struct mbuf *md; - char *fromcp, *tocp, *cp; - struct iovec aiov; - struct uio auio; - struct vnode *dp; - int error, rdonly, linklen; - struct componentname *cnp = &ndp->ni_cnd; - int lockleaf = (cnp->cn_flags & LOCKLEAF) != 0; - - *retdirp = NULL; - cnp->cn_flags |= NOMACCHECK; - cnp->cn_pnbuf = uma_zalloc(namei_zone, M_WAITOK); - - /* - * Copy the name from the mbuf list to ndp->ni_pnbuf - * and set the various ndp fields appropriately. - */ - fromcp = *dposp; - tocp = cnp->cn_pnbuf; - md = *mdp; - rem = mtod(md, caddr_t) + md->m_len - fromcp; - for (i = 0; i < len; i++) { - while (rem == 0) { - md = md->m_next; - if (md == NULL) { - error = EBADRPC; - goto out; - } - fromcp = mtod(md, caddr_t); - rem = md->m_len; - } - if (*fromcp == '\0' || (!pubflag && *fromcp == '/')) { - error = EACCES; - goto out; - } - *tocp++ = *fromcp++; - rem--; - } - *tocp = '\0'; - *mdp = md; - *dposp = fromcp; - len = nfsm_rndup(len)-len; - if (len > 0) { - if (rem >= len) - *dposp += len; - else if ((error = nfs_adv(mdp, dposp, len, rem)) != 0) - goto out; - } - - if (!pubflag && nfs_ispublicfh(fhp)) - return (ESTALE); - - /* - * Extract and set starting directory. - */ - error = nfsrv_fhtovp(fhp, 0, &dp, nfsd, slp, nam, &rdonly); - if (error) - goto out; - if (dp->v_type != VDIR) { - vput(dp); - error = ENOTDIR; - goto out; - } - - if (rdonly) - cnp->cn_flags |= RDONLY; - - /* - * Set return directory. Reference to dp is implicitly transfered - * to the returned pointer - */ - *retdirp = dp; - if (v3) { - *retdirattr_retp = VOP_GETATTR(dp, retdirattrp, - ndp->ni_cnd.cn_cred); - } - - VOP_UNLOCK(dp, 0); - - if (pubflag) { - /* - * Oh joy. For WebNFS, handle those pesky '%' escapes, - * and the 'native path' indicator. - */ - cp = uma_zalloc(namei_zone, M_WAITOK); - fromcp = cnp->cn_pnbuf; - tocp = cp; - if ((unsigned char)*fromcp >= WEBNFS_SPECCHAR_START) { - switch ((unsigned char)*fromcp) { - case WEBNFS_NATIVE_CHAR: - /* - * 'Native' path for us is the same - * as a path according to the NFS spec, - * just skip the escape char. - */ - fromcp++; - break; - /* - * More may be added in the future, range 0x80-0xff - */ - default: - error = EIO; - uma_zfree(namei_zone, cp); - goto out; - } - } - /* - * Translate the '%' escapes, URL-style. - */ - while (*fromcp != '\0') { - if (*fromcp == WEBNFS_ESC_CHAR) { - if (fromcp[1] != '\0' && fromcp[2] != '\0') { - fromcp++; - *tocp++ = HEXSTRTOI(fromcp); - fromcp += 2; - continue; - } else { - error = ENOENT; - uma_zfree(namei_zone, cp); - goto out; - } - } else - *tocp++ = *fromcp++; - } - *tocp = '\0'; - uma_zfree(namei_zone, cnp->cn_pnbuf); - cnp->cn_pnbuf = cp; - } - - ndp->ni_pathlen = (tocp - cnp->cn_pnbuf) + 1; - ndp->ni_segflg = UIO_SYSSPACE; - - if (pubflag) { - ndp->ni_rootdir = rootvnode; - ndp->ni_loopcnt = 0; - - if (cnp->cn_pnbuf[0] == '/') - dp = rootvnode; - } else { - cnp->cn_flags |= NOCROSSMOUNT; - } - - /* - * Initialize for scan, set ni_startdir and bump ref on dp again - * because lookup() will dereference ni_startdir. - */ - - cnp->cn_thread = curthread; - VREF(dp); - ndp->ni_startdir = dp; - - if (!lockleaf) - cnp->cn_flags |= LOCKLEAF; - for (;;) { - cnp->cn_nameptr = cnp->cn_pnbuf; - /* - * Call lookup() to do the real work. If an error occurs, - * ndp->ni_vp and ni_dvp are left uninitialized or NULL and - * we do not have to dereference anything before returning. - * In either case ni_startdir will be dereferenced and NULLed - * out. - */ - error = lookup(ndp); - if (error) - break; - - /* - * Check for encountering a symbolic link. Trivial - * termination occurs if no symlink encountered. - * Note: zfree is safe because error is 0, so we will - * not zfree it again when we break. - */ - if ((cnp->cn_flags & ISSYMLINK) == 0) { - if (cnp->cn_flags & (SAVENAME | SAVESTART)) - cnp->cn_flags |= HASBUF; - else - uma_zfree(namei_zone, cnp->cn_pnbuf); - if (ndp->ni_vp && !lockleaf) - VOP_UNLOCK(ndp->ni_vp, 0); - break; - } - - /* - * Validate symlink - */ - if ((cnp->cn_flags & LOCKPARENT) && ndp->ni_pathlen == 1) - VOP_UNLOCK(ndp->ni_dvp, 0); - if (!pubflag) { - error = EINVAL; - goto badlink2; - } - - if (ndp->ni_loopcnt++ >= MAXSYMLINKS) { - error = ELOOP; - goto badlink2; - } - if (ndp->ni_pathlen > 1) - cp = uma_zalloc(namei_zone, M_WAITOK); - else - cp = cnp->cn_pnbuf; - aiov.iov_base = cp; - aiov.iov_len = MAXPATHLEN; - auio.uio_iov = &aiov; - auio.uio_iovcnt = 1; - auio.uio_offset = 0; - auio.uio_rw = UIO_READ; - auio.uio_segflg = UIO_SYSSPACE; - auio.uio_td = NULL; - auio.uio_resid = MAXPATHLEN; - error = VOP_READLINK(ndp->ni_vp, &auio, cnp->cn_cred); - if (error) { - badlink1: - if (ndp->ni_pathlen > 1) - uma_zfree(namei_zone, cp); - badlink2: - vput(ndp->ni_vp); - vrele(ndp->ni_dvp); - break; - } - linklen = MAXPATHLEN - auio.uio_resid; - if (linklen == 0) { - error = ENOENT; - goto badlink1; - } - if (linklen + ndp->ni_pathlen >= MAXPATHLEN) { - error = ENAMETOOLONG; - goto badlink1; - } - - /* - * Adjust or replace path - */ - if (ndp->ni_pathlen > 1) { - bcopy(ndp->ni_next, cp + linklen, ndp->ni_pathlen); - uma_zfree(namei_zone, cnp->cn_pnbuf); - cnp->cn_pnbuf = cp; - } else - cnp->cn_pnbuf[linklen] = '\0'; - ndp->ni_pathlen += linklen; - - /* - * Cleanup refs for next loop and check if root directory - * should replace current directory. Normally ni_dvp - * becomes the new base directory and is cleaned up when - * we loop. Explicitly null pointers after invalidation - * to clarify operation. - */ - vput(ndp->ni_vp); - ndp->ni_vp = NULL; - - if (cnp->cn_pnbuf[0] == '/') { - vrele(ndp->ni_dvp); - ndp->ni_dvp = ndp->ni_rootdir; - VREF(ndp->ni_dvp); - } - ndp->ni_startdir = ndp->ni_dvp; - ndp->ni_dvp = NULL; - } - if (!lockleaf) - cnp->cn_flags &= ~LOCKLEAF; - - /* - * nfs_namei() guarentees that fields will not contain garbage - * whether an error occurs or not. This allows the caller to track - * cleanup state trivially. - */ -out: - if (error) { - uma_zfree(namei_zone, cnp->cn_pnbuf); - ndp->ni_vp = NULL; - ndp->ni_dvp = NULL; - ndp->ni_startdir = NULL; - cnp->cn_flags &= ~HASBUF; - } else if ((ndp->ni_cnd.cn_flags & (WANTPARENT|LOCKPARENT)) == 0) { - ndp->ni_dvp = NULL; - } - return (error); -} - -/* - * A fiddled version of m_adj() that ensures null fill to a long - * boundary and only trims off the back end - */ -void -nfsm_adj(struct mbuf *mp, int len, int nul) -{ - struct mbuf *m; - int count, i; - char *cp; - - /* - * Trim from tail. Scan the mbuf chain, - * calculating its length and finding the last mbuf. - * If the adjustment only affects this mbuf, then just - * adjust and return. Otherwise, rescan and truncate - * after the remaining size. - */ - count = 0; - m = mp; - for (;;) { - count += m->m_len; - if (m->m_next == NULL) - break; - m = m->m_next; - } - if (m->m_len > len) { - m->m_len -= len; - if (nul > 0) { - cp = mtod(m, caddr_t)+m->m_len-nul; - for (i = 0; i < nul; i++) - *cp++ = '\0'; - } - return; - } - count -= len; - if (count < 0) - count = 0; - /* - * Correct length for chain is "count". - * Find the mbuf with last data, adjust its length, - * and toss data from remaining mbufs on chain. - */ - for (m = mp; m; m = m->m_next) { - if (m->m_len >= count) { - m->m_len = count; - if (nul > 0) { - cp = mtod(m, caddr_t)+m->m_len-nul; - for (i = 0; i < nul; i++) - *cp++ = '\0'; - } - if (m->m_next != NULL) { - m_freem(m->m_next); - m->m_next = NULL; - } - break; - } - count -= m->m_len; - } -} - -/* - * Make these functions instead of macros, so that the kernel text size - * doesn't get too big... - */ -void -nfsm_srvwcc(struct nfsrv_descript *nfsd, int before_ret, - struct vattr *before_vap, int after_ret, struct vattr *after_vap, - struct mbuf **mbp, char **bposp) -{ - struct mbuf *mb = *mbp; - char *bpos = *bposp; - u_int32_t *tl; - - if (before_ret) { - tl = nfsm_build(u_int32_t *, NFSX_UNSIGNED); - *tl = nfsrv_nfs_false; - } else { - tl = nfsm_build(u_int32_t *, 7 * NFSX_UNSIGNED); - *tl++ = nfsrv_nfs_true; - txdr_hyper(before_vap->va_size, tl); - tl += 2; - txdr_nfsv3time(&(before_vap->va_mtime), tl); - tl += 2; - txdr_nfsv3time(&(before_vap->va_ctime), tl); - } - *bposp = bpos; - *mbp = mb; - nfsm_srvpostopattr(nfsd, after_ret, after_vap, mbp, bposp); -} - -void -nfsm_srvpostopattr(struct nfsrv_descript *nfsd, int after_ret, - struct vattr *after_vap, struct mbuf **mbp, char **bposp) -{ - struct mbuf *mb = *mbp; - char *bpos = *bposp; - u_int32_t *tl; - struct nfs_fattr *fp; - - if (after_ret) { - tl = nfsm_build(u_int32_t *, NFSX_UNSIGNED); - *tl = nfsrv_nfs_false; - } else { - tl = nfsm_build(u_int32_t *, NFSX_UNSIGNED + NFSX_V3FATTR); - *tl++ = nfsrv_nfs_true; - fp = (struct nfs_fattr *)tl; - nfsm_srvfattr(nfsd, after_vap, fp); - } - *mbp = mb; - *bposp = bpos; -} - -void -nfsm_srvfattr(struct nfsrv_descript *nfsd, struct vattr *vap, - struct nfs_fattr *fp) -{ - - fp->fa_nlink = txdr_unsigned(vap->va_nlink); - fp->fa_uid = txdr_unsigned(vap->va_uid); - fp->fa_gid = txdr_unsigned(vap->va_gid); - if (nfsd->nd_flag & ND_NFSV3) { - fp->fa_type = vtonfsv3_type(vap->va_type); - fp->fa_mode = vtonfsv3_mode(vap->va_mode); - txdr_hyper(vap->va_size, &fp->fa3_size); - txdr_hyper(vap->va_bytes, &fp->fa3_used); - fp->fa3_rdev.specdata1 = txdr_unsigned(major(vap->va_rdev)); - fp->fa3_rdev.specdata2 = txdr_unsigned(minor(vap->va_rdev)); - fp->fa3_fsid.nfsuquad[0] = 0; - fp->fa3_fsid.nfsuquad[1] = txdr_unsigned(vap->va_fsid); - fp->fa3_fileid.nfsuquad[0] = 0; - fp->fa3_fileid.nfsuquad[1] = txdr_unsigned(vap->va_fileid); - txdr_nfsv3time(&vap->va_atime, &fp->fa3_atime); - txdr_nfsv3time(&vap->va_mtime, &fp->fa3_mtime); - txdr_nfsv3time(&vap->va_ctime, &fp->fa3_ctime); - } else { - fp->fa_type = vtonfsv2_type(vap->va_type); - fp->fa_mode = vtonfsv2_mode(vap->va_type, vap->va_mode); - fp->fa2_size = txdr_unsigned(vap->va_size); - fp->fa2_blocksize = txdr_unsigned(vap->va_blocksize); - if (vap->va_type == VFIFO) - fp->fa2_rdev = 0xffffffff; - else - fp->fa2_rdev = txdr_unsigned(vap->va_rdev); - fp->fa2_blocks = txdr_unsigned(vap->va_bytes / NFS_FABLKSIZE); - fp->fa2_fsid = txdr_unsigned(vap->va_fsid); - fp->fa2_fileid = txdr_unsigned(vap->va_fileid); - txdr_nfsv2time(&vap->va_atime, &fp->fa2_atime); - txdr_nfsv2time(&vap->va_mtime, &fp->fa2_mtime); - txdr_nfsv2time(&vap->va_ctime, &fp->fa2_ctime); - } -} - -/* - * nfsrv_fhtovp() - convert a fh to a vnode ptr (optionally locked) - * - look up fsid in mount list (if not found ret error) - * - get vp and export rights by calling VFS_FHTOVP() - * - if cred->cr_uid == 0 or MNT_EXPORTANON set it to credanon - */ -int -nfsrv_fhtovp(fhandle_t *fhp, int flags, struct vnode **vpp, - struct nfsrv_descript *nfsd, struct nfssvc_sock *slp, - struct sockaddr *nam, int *rdonlyp) -{ - struct mount *mp; - int i; - struct ucred *cred, *credanon; - int error, exflags; -#ifdef MNT_EXNORESPORT /* XXX needs mountd and /etc/exports help yet */ - struct sockaddr_int *saddr; -#endif - int credflavor; - int numsecflavors, *secflavors; - int authsys; - int v3 = nfsd->nd_flag & ND_NFSV3; - int mountreq; - - *vpp = NULL; - - if (nfs_ispublicfh(fhp)) { - if (!nfs_pub.np_valid) - return (ESTALE); - fhp = &nfs_pub.np_handle; - } - - mp = vfs_busyfs(&fhp->fh_fsid); - if (!mp) - return (ESTALE); - error = VFS_CHECKEXP(mp, nam, &exflags, &credanon, - &numsecflavors, &secflavors); - if (error) { - vfs_unbusy(mp); - goto out; - } - if (numsecflavors == 0) { - /* - * This can happen if the system is running with an - * old mountd that doesn't pass in a secflavor list. - */ - numsecflavors = 1; - authsys = AUTH_SYS; - secflavors = &authsys; - } - credflavor = nfsd->nd_credflavor; - for (i = 0; i < numsecflavors; i++) { - if (secflavors[i] == credflavor) - break; - } - if (i == numsecflavors) { - /* - * RFC 2623 section 2.3.2 - allow certain procedures - * used at NFS client mount time even if they have - * weak authentication. - */ - mountreq = FALSE; - if (v3) { - if (nfsd->nd_procnum == NFSPROC_FSINFO - || nfsd->nd_procnum == NFSPROC_GETATTR) - mountreq = TRUE; - } else { - if (nfsd->nd_procnum == NFSPROC_FSSTAT - || nfsd->nd_procnum == NFSPROC_GETATTR) - mountreq = TRUE; - } - if (!mountreq) { - error = NFSERR_AUTHERR | AUTH_TOOWEAK; - vfs_unbusy(mp); - goto out; - } - } - error = VFS_FHTOVP(mp, &fhp->fh_fid, LK_EXCLUSIVE, vpp); - if (error) { - /* Make sure the server replies ESTALE to the client. */ - error = ESTALE; - vfs_unbusy(mp); - goto out; - } -#ifdef MNT_EXNORESPORT - if (!(exflags & (MNT_EXNORESPORT|MNT_EXPUBLIC))) { - saddr = (struct sockaddr_in *)nam; - if ((saddr->sin_family == AF_INET || - saddr->sin_family == AF_INET6) && - /* same code for INET and INET6: sin*_port at same offet */ - ntohs(saddr->sin_port) >= IPPORT_RESERVED) { - vput(*vpp); - *vpp = NULL; - error = NFSERR_AUTHERR | AUTH_TOOWEAK; - vfs_unbusy(mp); - goto out; - } - } -#endif - /* - * Check/setup credentials. - */ - cred = nfsd->nd_cr; - if (cred->cr_uid == 0 || (exflags & MNT_EXPORTANON)) { - cred->cr_uid = credanon->cr_uid; - crsetgroups(cred, credanon->cr_ngroups, credanon->cr_groups); - } - if (exflags & MNT_EXRDONLY) - *rdonlyp = 1; - else - *rdonlyp = 0; - - if (!(flags & NFSRV_FLAG_BUSY)) - vfs_unbusy(mp); -out: - if (credanon != NULL) - crfree(credanon); - - return (error); -} - - -/* - * WebNFS: check if a filehandle is a public filehandle. For v3, this - * means a length of 0, for v2 it means all zeroes. nfsm_srvmtofh has - * transformed this to all zeroes in both cases, so check for it. - */ -int -nfs_ispublicfh(fhandle_t *fhp) -{ - char *cp = (char *)fhp; - int i; - - NFSD_LOCK_DONTCARE(); - - for (i = 0; i < NFSX_V3FH; i++) - if (*cp++ != 0) - return (FALSE); - return (TRUE); -} - -/* - * Map errnos to NFS error numbers. For Version 3 also filter out error - * numbers not specified for the associated procedure. - */ -int -nfsrv_errmap(struct nfsrv_descript *nd, int err) -{ - const short *defaulterrp, *errp; - int e; - - - if (nd->nd_flag & ND_NFSV3) { - if (nd->nd_procnum <= NFSPROC_COMMIT) { - errp = defaulterrp = nfsrv_v3errmap[nd->nd_procnum]; - while (*++errp) { - if (*errp == err) - return (err); - else if (*errp > err) - break; - } - return ((int)*defaulterrp); - } else - return (err & 0xffff); - } - e = 0; - if (err <= ELAST) - e = nfsrv_v2errmap[err - 1]; - if (e != 0) - return (e); - return (NFSERR_IO); -} - -/* - * Sort the group list in increasing numerical order. - * (Insertion sort by Chris Torek, who was grossed out by the bubble sort - * that used to be here.) - */ -void -nfsrvw_sort(gid_t *list, int num) -{ - int i, j; - gid_t v; - - /* Insertion sort. */ - for (i = 1; i < num; i++) { - v = list[i]; - /* find correct slot for value v, moving others up */ - for (j = i; --j >= 0 && v < list[j];) - list[j + 1] = list[j]; - list[j + 1] = v; - } -} - -/* - * Helper functions for macros. - */ - -void -nfsm_srvfhtom_xx(fhandle_t *f, int v3, struct mbuf **mb, caddr_t *bpos) -{ - u_int32_t *tl; - - if (v3) { - tl = nfsm_build_xx(NFSX_UNSIGNED + NFSX_V3FH, mb, bpos); - *tl++ = txdr_unsigned(NFSX_V3FH); - bcopy(f, tl, NFSX_V3FH); - } else { - tl = nfsm_build_xx(NFSX_V2FH, mb, bpos); - bcopy(f, tl, NFSX_V2FH); - } -} - -void -nfsm_srvpostop_fh_xx(fhandle_t *f, struct mbuf **mb, caddr_t *bpos) -{ - u_int32_t *tl; - - tl = nfsm_build_xx(2 * NFSX_UNSIGNED + NFSX_V3FH, mb, bpos); - *tl++ = nfsrv_nfs_true; - *tl++ = txdr_unsigned(NFSX_V3FH); - bcopy(f, tl, NFSX_V3FH); -} - -int -nfsm_srvstrsiz_xx(int *s, int m, struct mbuf **md, caddr_t *dpos) -{ - u_int32_t *tl; - - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - *s = fxdr_unsigned(int32_t, *tl); - if (*s > m || *s <= 0) - return EBADRPC; - return 0; -} - -int -nfsm_srvnamesiz_xx(int *s, int m, struct mbuf **md, caddr_t *dpos) -{ - u_int32_t *tl; - - NFSD_LOCK_DONTCARE(); - - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - *s = fxdr_unsigned(int32_t, *tl); - if (*s > m) - return NFSERR_NAMETOL; - if (*s <= 0) - return EBADRPC; - return 0; -} - -int -nfsm_srvnamesiz0_xx(int *s, int m, struct mbuf **md, caddr_t *dpos) -{ - u_int32_t *tl; - - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - *s = fxdr_unsigned(int32_t, *tl); - if (*s > m) - return NFSERR_NAMETOL; - if (*s < 0) - return EBADRPC; - return 0; -} - -void -nfsm_clget_xx(u_int32_t **tl, struct mbuf *mb, struct mbuf **mp, - char **bp, char **be, caddr_t bpos) -{ - struct mbuf *nmp; - - NFSD_UNLOCK_ASSERT(); - - if (*bp >= *be) { - if (*mp == mb) - (*mp)->m_len += *bp - bpos; - MGET(nmp, M_WAITOK, MT_DATA); - MCLGET(nmp, M_WAITOK); - nmp->m_len = NFSMSIZ(nmp); - (*mp)->m_next = nmp; - *mp = nmp; - *bp = mtod(*mp, caddr_t); - *be = *bp + (*mp)->m_len; - } - *tl = (u_int32_t *)*bp; -} - -int -nfsm_srvmtofh_xx(fhandle_t *f, int v3, struct mbuf **md, caddr_t *dpos) -{ - u_int32_t *tl; - int fhlen; - - if (v3) { - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - fhlen = fxdr_unsigned(int, *tl); - if (fhlen != 0 && fhlen != NFSX_V3FH) - return EBADRPC; - } else { - fhlen = NFSX_V2FH; - } - if (fhlen != 0) { - tl = nfsm_dissect_xx_nonblock(fhlen, md, dpos); - if (tl == NULL) - return EBADRPC; - bcopy((caddr_t)tl, (caddr_t)(f), fhlen); - } else { - bzero((caddr_t)(f), NFSX_V3FH); - } - return 0; -} - -int -nfsm_srvsattr_xx(struct vattr *a, struct mbuf **md, caddr_t *dpos) -{ - u_int32_t *tl; - int toclient = 0; - - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - if (*tl == nfsrv_nfs_true) { - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - (a)->va_mode = nfstov_mode(*tl); - } - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - if (*tl == nfsrv_nfs_true) { - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - (a)->va_uid = fxdr_unsigned(uid_t, *tl); - } - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - if (*tl == nfsrv_nfs_true) { - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - (a)->va_gid = fxdr_unsigned(gid_t, *tl); - } - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - if (*tl == nfsrv_nfs_true) { - tl = nfsm_dissect_xx_nonblock(2 * NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - (a)->va_size = fxdr_hyper(tl); - } - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - switch (fxdr_unsigned(int, *tl)) { - case NFSV3SATTRTIME_TOCLIENT: - tl = nfsm_dissect_xx_nonblock(2 * NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - fxdr_nfsv3time(tl, &(a)->va_atime); - toclient = 1; - break; - case NFSV3SATTRTIME_TOSERVER: - vfs_timestamp(&a->va_atime); - a->va_vaflags |= VA_UTIMES_NULL; - break; - } - tl = nfsm_dissect_xx_nonblock(NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - switch (fxdr_unsigned(int, *tl)) { - case NFSV3SATTRTIME_TOCLIENT: - tl = nfsm_dissect_xx_nonblock(2 * NFSX_UNSIGNED, md, dpos); - if (tl == NULL) - return EBADRPC; - fxdr_nfsv3time(tl, &(a)->va_mtime); - a->va_vaflags &= ~VA_UTIMES_NULL; - break; - case NFSV3SATTRTIME_TOSERVER: - vfs_timestamp(&a->va_mtime); - if (toclient == 0) - a->va_vaflags |= VA_UTIMES_NULL; - break; - } - return 0; -} diff --git a/sys/powerpc/aim/trap.c b/sys/powerpc/aim/trap.c index 354c63b..2f30d6d 100644 --- a/sys/powerpc/aim/trap.c +++ b/sys/powerpc/aim/trap.c @@ -177,7 +177,7 @@ trap(struct trapframe *frame) * handled the trap and modified the trap frame so that this * function can return normally. */ - if (dtrace_trap_func != NULL && (*dtrace_trap_func)(frame)) + if (dtrace_trap_func != NULL && (*dtrace_trap_func)(frame, type) != 0) return; #endif diff --git a/sys/sys/dtrace_bsd.h b/sys/sys/dtrace_bsd.h index 6bcaf29..f46b900 100644 --- a/sys/sys/dtrace_bsd.h +++ b/sys/sys/dtrace_bsd.h @@ -39,14 +39,14 @@ struct vattr; struct vnode; struct reg; -int dtrace_trap(struct trapframe *); +int dtrace_trap(struct trapframe *, u_int); /* * The dtrace module handles traps that occur during a DTrace probe. * This type definition is used in the trap handler to provide a * hook for the dtrace module to register its handler with. */ -typedef int (*dtrace_trap_func_t)(struct trapframe *); +typedef int (*dtrace_trap_func_t)(struct trapframe *, u_int); extern dtrace_trap_func_t dtrace_trap_func; /* diff --git a/sys/sys/mbuf.h b/sys/sys/mbuf.h index ac3cfb2..cbb2a18 100644 --- a/sys/sys/mbuf.h +++ b/sys/sys/mbuf.h @@ -757,7 +757,10 @@ static __inline void m_clrprotoflags(struct mbuf *m) { - m->m_flags &= ~M_PROTOFLAGS; + while (m) { + m->m_flags &= ~M_PROTOFLAGS; + m = m->m_next; + } } static __inline struct mbuf * diff --git a/sys/sys/param.h b/sys/sys/param.h index d25cdef..a4922f1 100644 --- a/sys/sys/param.h +++ b/sys/sys/param.h @@ -58,7 +58,7 @@ * in the range 5 to 9. */ #undef __FreeBSD_version -#define __FreeBSD_version 1100050 /* Master, propagated to newvers */ +#define __FreeBSD_version 1100051 /* Master, propagated to newvers */ /* * __FreeBSD_kernel__ indicates that this system uses the kernel of FreeBSD, diff --git a/sys/sys/sem.h b/sys/sys/sem.h index f52bc0c..0ea259b 100644 --- a/sys/sys/sem.h +++ b/sys/sys/sem.h @@ -37,7 +37,7 @@ struct semid_ds_old { long sem_pad1; /* SVABI/386 says I need this here */ time_t sem_ctime; /* last change time */ /* Times measured in secs since */ - /* 00:00:00 GMT, Jan. 1, 1970 */ + /* 00:00:00 UTC, Jan. 1, 1970, without leap seconds */ long sem_pad2; /* SVABI/386 says I need this here */ long sem_pad3[4]; /* SVABI/386 says I need this here */ }; @@ -50,7 +50,7 @@ struct semid_ds { time_t sem_otime; /* last operation time */ time_t sem_ctime; /* last change time */ /* Times measured in secs since */ - /* 00:00:00 GMT, Jan. 1, 1970 */ + /* 00:00:00 UTC, Jan. 1, 1970, without leap seconds */ }; /* diff --git a/sys/sys/time.h b/sys/sys/time.h index 1f0a530..395e888 100644 --- a/sys/sys/time.h +++ b/sys/sys/time.h @@ -398,7 +398,7 @@ extern sbintime_t sbt_tickthreshold; * Functions containing "up" returns time relative to boot and * should be used for calculating time intervals. * - * Functions without "up" returns GMT time. + * Functions without "up" returns UTC time. * * Functions with the "get" prefix returns a less precise result * much faster than the functions without "get" prefix and should diff --git a/sys/sys/vnode.h b/sys/sys/vnode.h index 87382b8..8a14f69 100644 --- a/sys/sys/vnode.h +++ b/sys/sys/vnode.h @@ -580,6 +580,7 @@ typedef void vop_getpages_iodone_t(void *, vm_page_t *, int, int); /* vn_open_flags */ #define VN_OPEN_NOAUDIT 0x00000001 #define VN_OPEN_NOCAPCHECK 0x00000002 +#define VN_OPEN_NAMECACHE 0x00000004 /* * Public vnode manipulation functions. diff --git a/sys/teken/libteken/teken.3 b/sys/teken/libteken/teken.3 index 1c2ebbe..70bed07 100644 --- a/sys/teken/libteken/teken.3 +++ b/sys/teken/libteken/teken.3 @@ -188,7 +188,7 @@ prior to 9.0. .Sh SEE ALSO .Xr ncurses 3 , .Xr termcap 3 , -.Xr syscons 4 . +.Xr syscons 4 .Sh HISTORY The .Nm diff --git a/sys/ufs/ufs/ufs_vnops.c b/sys/ufs/ufs/ufs_vnops.c index 21d4ba4..3d7a21d 100644 --- a/sys/ufs/ufs/ufs_vnops.c +++ b/sys/ufs/ufs/ufs_vnops.c @@ -205,8 +205,10 @@ ufs_create(ap) error = ufs_makeinode(MAKEIMODE(ap->a_vap->va_type, ap->a_vap->va_mode), ap->a_dvp, ap->a_vpp, ap->a_cnp); - if (error) + if (error != 0) return (error); + if ((ap->a_cnp->cn_flags & MAKEENTRY) != 0) + cache_enter(ap->a_dvp, *ap->a_vpp, ap->a_cnp); return (0); } diff --git a/sys/vm/vm_page.c b/sys/vm/vm_page.c index b2877c1..79665ba 100644 --- a/sys/vm/vm_page.c +++ b/sys/vm/vm_page.c @@ -2711,6 +2711,8 @@ retrylookup: sleep = (allocflags & VM_ALLOC_IGN_SBUSY) != 0 ? vm_page_xbusied(m) : vm_page_busied(m); if (sleep) { + if ((allocflags & VM_ALLOC_NOWAIT) != 0) + return (NULL); /* * Reference the page before unlocking and * sleeping so that the page daemon is less @@ -2736,8 +2738,10 @@ retrylookup: return (m); } } - m = vm_page_alloc(object, pindex, allocflags & ~VM_ALLOC_IGN_SBUSY); + m = vm_page_alloc(object, pindex, allocflags); if (m == NULL) { + if ((allocflags & VM_ALLOC_NOWAIT) != 0) + return (NULL); VM_OBJECT_WUNLOCK(object); VM_WAIT; VM_OBJECT_WLOCK(object); diff --git a/sys/vm/vm_page.h b/sys/vm/vm_page.h index f12b76c..cd30772 100644 --- a/sys/vm/vm_page.h +++ b/sys/vm/vm_page.h @@ -376,22 +376,36 @@ extern long first_page; /* first physical page number */ vm_page_t PHYS_TO_VM_PAGE(vm_paddr_t pa); -/* page allocation classes: */ +/* + * Page allocation parameters for vm_page for the functions + * vm_page_alloc(), vm_page_grab(), vm_page_alloc_contig() and + * vm_page_alloc_freelist(). Some functions support only a subset + * of the flags, and ignore others, see the flags legend. + * + * Bits 0 - 1 define class. + * Bits 2 - 15 dedicated for flags. + * Legend: + * (a) - vm_page_alloc() supports the flag. + * (c) - vm_page_alloc_contig() supports the flag. + * (f) - vm_page_alloc_freelist() supports the flag. + * (g) - vm_page_grab() supports the flag. + * Bits above 15 define the count of additional pages that the caller + * intends to allocate. + */ #define VM_ALLOC_NORMAL 0 #define VM_ALLOC_INTERRUPT 1 #define VM_ALLOC_SYSTEM 2 #define VM_ALLOC_CLASS_MASK 3 -/* page allocation flags: */ -#define VM_ALLOC_WIRED 0x0020 /* non pageable */ -#define VM_ALLOC_ZERO 0x0040 /* Try to obtain a zeroed page */ -#define VM_ALLOC_NOOBJ 0x0100 /* No associated object */ -#define VM_ALLOC_NOBUSY 0x0200 /* Do not busy the page */ -#define VM_ALLOC_IFCACHED 0x0400 /* Fail if the page is not cached */ -#define VM_ALLOC_IFNOTCACHED 0x0800 /* Fail if the page is cached */ -#define VM_ALLOC_IGN_SBUSY 0x1000 /* vm_page_grab() only */ -#define VM_ALLOC_NODUMP 0x2000 /* don't include in dump */ -#define VM_ALLOC_SBUSY 0x4000 /* Shared busy the page */ - +#define VM_ALLOC_WIRED 0x0020 /* (acfg) Allocate non pageable page */ +#define VM_ALLOC_ZERO 0x0040 /* (acfg) Try to obtain a zeroed page */ +#define VM_ALLOC_NOOBJ 0x0100 /* (acg) No associated object */ +#define VM_ALLOC_NOBUSY 0x0200 /* (acg) Do not busy the page */ +#define VM_ALLOC_IFCACHED 0x0400 /* (ag) Fail if page is not cached */ +#define VM_ALLOC_IFNOTCACHED 0x0800 /* (ag) Fail if page is cached */ +#define VM_ALLOC_IGN_SBUSY 0x1000 /* (g) Ignore shared busy flag */ +#define VM_ALLOC_NODUMP 0x2000 /* (ag) don't include in dump */ +#define VM_ALLOC_SBUSY 0x4000 /* (acg) Shared busy the page */ +#define VM_ALLOC_NOWAIT 0x8000 /* (g) Do not sleep, return NULL */ #define VM_ALLOC_COUNT_SHIFT 16 #define VM_ALLOC_COUNT(count) ((count) << VM_ALLOC_COUNT_SHIFT) |