diff options
Diffstat (limited to 'arch/x86/kernel')
34 files changed, 1887 insertions, 383 deletions
diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index 95f216b..339ce35 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -111,7 +111,7 @@ obj-$(CONFIG_SWIOTLB) += pci-swiotlb_64.o # NB rename without _64 ### # 64 bit specific files ifeq ($(CONFIG_X86_64),y) - obj-$(CONFIG_X86_UV) += tlb_uv.o bios_uv.o uv_irq.o uv_sysfs.o + obj-$(CONFIG_X86_UV) += tlb_uv.o bios_uv.o uv_irq.o uv_sysfs.o uv_time.o obj-$(CONFIG_X86_PM_TIMER) += pmtimer_64.o obj-$(CONFIG_AUDIT) += audit_64.o diff --git a/arch/x86/kernel/alternative.c b/arch/x86/kernel/alternative.c index 6907b8e..4c80f15 100644 --- a/arch/x86/kernel/alternative.c +++ b/arch/x86/kernel/alternative.c @@ -414,9 +414,17 @@ void __init alternative_instructions(void) that might execute the to be patched code. Other CPUs are not running. */ stop_nmi(); -#ifdef CONFIG_X86_MCE - stop_mce(); -#endif + + /* + * Don't stop machine check exceptions while patching. + * MCEs only happen when something got corrupted and in this + * case we must do something about the corruption. + * Ignoring it is worse than a unlikely patching race. + * Also machine checks tend to be broadcast and if one CPU + * goes into machine check the others follow quickly, so we don't + * expect a machine check to cause undue problems during to code + * patching. + */ apply_alternatives(__alt_instructions, __alt_instructions_end); @@ -456,9 +464,6 @@ void __init alternative_instructions(void) (unsigned long)__smp_locks_end); restart_nmi(); -#ifdef CONFIG_X86_MCE - restart_mce(); -#endif } /** diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c index f9cecdf..30909a2 100644 --- a/arch/x86/kernel/apic/apic.c +++ b/arch/x86/kernel/apic/apic.c @@ -46,6 +46,7 @@ #include <asm/idle.h> #include <asm/mtrr.h> #include <asm/smp.h> +#include <asm/mce.h> unsigned int num_processors; @@ -842,6 +843,14 @@ void clear_local_APIC(void) apic_write(APIC_LVTTHMR, v | APIC_LVT_MASKED); } #endif +#ifdef CONFIG_X86_MCE_INTEL + if (maxlvt >= 6) { + v = apic_read(APIC_LVTCMCI); + if (!(v & APIC_LVT_MASKED)) + apic_write(APIC_LVTCMCI, v | APIC_LVT_MASKED); + } +#endif + /* * Clean APIC state for other OSs: */ @@ -1241,6 +1250,12 @@ void __cpuinit setup_local_APIC(void) apic_write(APIC_LVT1, value); preempt_enable(); + +#ifdef CONFIG_X86_MCE_INTEL + /* Recheck CMCI information after local APIC is up on CPU #0 */ + if (smp_processor_id() == 0) + cmci_recheck(); +#endif } void __cpuinit end_local_APIC_setup(void) diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c index 25423a5..f47df59 100644 --- a/arch/x86/kernel/cpu/amd.c +++ b/arch/x86/kernel/cpu/amd.c @@ -5,6 +5,7 @@ #include <asm/io.h> #include <asm/processor.h> #include <asm/apic.h> +#include <asm/cpu.h> #ifdef CONFIG_X86_64 # include <asm/numa_64.h> @@ -141,6 +142,55 @@ static void __cpuinit init_amd_k6(struct cpuinfo_x86 *c) } } +static void __cpuinit amd_k7_smp_check(struct cpuinfo_x86 *c) +{ +#ifdef CONFIG_SMP + /* calling is from identify_secondary_cpu() ? */ + if (c->cpu_index == boot_cpu_id) + return; + + /* + * Certain Athlons might work (for various values of 'work') in SMP + * but they are not certified as MP capable. + */ + /* Athlon 660/661 is valid. */ + if ((c->x86_model == 6) && ((c->x86_mask == 0) || + (c->x86_mask == 1))) + goto valid_k7; + + /* Duron 670 is valid */ + if ((c->x86_model == 7) && (c->x86_mask == 0)) + goto valid_k7; + + /* + * Athlon 662, Duron 671, and Athlon >model 7 have capability + * bit. It's worth noting that the A5 stepping (662) of some + * Athlon XP's have the MP bit set. + * See http://www.heise.de/newsticker/data/jow-18.10.01-000 for + * more. + */ + if (((c->x86_model == 6) && (c->x86_mask >= 2)) || + ((c->x86_model == 7) && (c->x86_mask >= 1)) || + (c->x86_model > 7)) + if (cpu_has_mp) + goto valid_k7; + + /* If we get here, not a certified SMP capable AMD system. */ + + /* + * Don't taint if we are running SMP kernel on a single non-MP + * approved Athlon + */ + WARN_ONCE(1, "WARNING: This combination of AMD" + "processors is not suitable for SMP.\n"); + if (!test_taint(TAINT_UNSAFE_SMP)) + add_taint(TAINT_UNSAFE_SMP); + +valid_k7: + ; +#endif +} + static void __cpuinit init_amd_k7(struct cpuinfo_x86 *c) { u32 l, h; @@ -175,6 +225,8 @@ static void __cpuinit init_amd_k7(struct cpuinfo_x86 *c) } set_cpu_cap(c, X86_FEATURE_K7); + + amd_k7_smp_check(c); } #endif diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c index 4b1c319..22590cf 100644 --- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c +++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c @@ -601,7 +601,7 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy) if (!data) return -ENOMEM; - data->acpi_data = percpu_ptr(acpi_perf_data, cpu); + data->acpi_data = per_cpu_ptr(acpi_perf_data, cpu); per_cpu(drv_data, cpu) = data; if (cpu_has(c, X86_FEATURE_CONSTANT_TSC)) diff --git a/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c b/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c index b585e04c..3178c3a 100644 --- a/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c +++ b/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c @@ -277,7 +277,6 @@ static struct cpufreq_driver p4clockmod_driver = { .name = "p4-clockmod", .owner = THIS_MODULE, .attr = p4clockmod_attr, - .hide_interface = 1, }; diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c index 25c559b..191117f 100644 --- a/arch/x86/kernel/cpu/intel.c +++ b/arch/x86/kernel/cpu/intel.c @@ -13,6 +13,7 @@ #include <asm/uaccess.h> #include <asm/ds.h> #include <asm/bugs.h> +#include <asm/cpu.h> #ifdef CONFIG_X86_64 #include <asm/topology.h> @@ -110,6 +111,28 @@ static void __cpuinit trap_init_f00f_bug(void) } #endif +static void __cpuinit intel_smp_check(struct cpuinfo_x86 *c) +{ +#ifdef CONFIG_SMP + /* calling is from identify_secondary_cpu() ? */ + if (c->cpu_index == boot_cpu_id) + return; + + /* + * Mask B, Pentium, but not Pentium MMX + */ + if (c->x86 == 5 && + c->x86_mask >= 1 && c->x86_mask <= 4 && + c->x86_model <= 3) { + /* + * Remember we have B step Pentia with bugs + */ + WARN_ONCE(1, "WARNING: SMP operation may be unreliable" + "with B stepping processors.\n"); + } +#endif +} + static void __cpuinit intel_workarounds(struct cpuinfo_x86 *c) { unsigned long lo, hi; @@ -186,6 +209,8 @@ static void __cpuinit intel_workarounds(struct cpuinfo_x86 *c) #ifdef CONFIG_X86_NUMAQ numaq_tsc_disable(); #endif + + intel_smp_check(c); } #else static void __cpuinit intel_workarounds(struct cpuinfo_x86 *c) diff --git a/arch/x86/kernel/cpu/mcheck/Makefile b/arch/x86/kernel/cpu/mcheck/Makefile index d7d2323..b2f8982 100644 --- a/arch/x86/kernel/cpu/mcheck/Makefile +++ b/arch/x86/kernel/cpu/mcheck/Makefile @@ -4,3 +4,4 @@ obj-$(CONFIG_X86_32) += k7.o p4.o p5.o p6.o winchip.o obj-$(CONFIG_X86_MCE_INTEL) += mce_intel_64.o obj-$(CONFIG_X86_MCE_AMD) += mce_amd_64.o obj-$(CONFIG_X86_MCE_NONFATAL) += non-fatal.o +obj-$(CONFIG_X86_MCE_THRESHOLD) += threshold.o diff --git a/arch/x86/kernel/cpu/mcheck/mce_32.c b/arch/x86/kernel/cpu/mcheck/mce_32.c index dfaebce..3552119 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_32.c +++ b/arch/x86/kernel/cpu/mcheck/mce_32.c @@ -60,20 +60,6 @@ void mcheck_init(struct cpuinfo_x86 *c) } } -static unsigned long old_cr4 __initdata; - -void __init stop_mce(void) -{ - old_cr4 = read_cr4(); - clear_in_cr4(X86_CR4_MCE); -} - -void __init restart_mce(void) -{ - if (old_cr4 & X86_CR4_MCE) - set_in_cr4(X86_CR4_MCE); -} - static int __init mcheck_disable(char *str) { mce_disabled = 1; diff --git a/arch/x86/kernel/cpu/mcheck/mce_64.c b/arch/x86/kernel/cpu/mcheck/mce_64.c index fe79985..ca14604 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_64.c +++ b/arch/x86/kernel/cpu/mcheck/mce_64.c @@ -3,6 +3,8 @@ * K8 parts Copyright 2002,2003 Andi Kleen, SuSE Labs. * Rest from unknown author(s). * 2004 Andi Kleen. Rewrote most of it. + * Copyright 2008 Intel Corporation + * Author: Andi Kleen */ #include <linux/init.h> @@ -24,6 +26,9 @@ #include <linux/ctype.h> #include <linux/kmod.h> #include <linux/kdebug.h> +#include <linux/kobject.h> +#include <linux/sysfs.h> +#include <linux/ratelimit.h> #include <asm/processor.h> #include <asm/msr.h> #include <asm/mce.h> @@ -32,7 +37,6 @@ #include <asm/idle.h> #define MISC_MCELOG_MINOR 227 -#define NR_SYSFS_BANKS 6 atomic_t mce_entry; @@ -47,7 +51,7 @@ static int mce_dont_init; */ static int tolerant = 1; static int banks; -static unsigned long bank[NR_SYSFS_BANKS] = { [0 ... NR_SYSFS_BANKS-1] = ~0UL }; +static u64 *bank; static unsigned long notify_user; static int rip_msr; static int mce_bootlog = -1; @@ -58,6 +62,19 @@ static char *trigger_argv[2] = { trigger, NULL }; static DECLARE_WAIT_QUEUE_HEAD(mce_wait); +/* MCA banks polled by the period polling timer for corrected events */ +DEFINE_PER_CPU(mce_banks_t, mce_poll_banks) = { + [0 ... BITS_TO_LONGS(MAX_NR_BANKS)-1] = ~0UL +}; + +/* Do initial initialization of a struct mce */ +void mce_setup(struct mce *m) +{ + memset(m, 0, sizeof(struct mce)); + m->cpu = smp_processor_id(); + rdtscll(m->tsc); +} + /* * Lockless MCE logging infrastructure. * This avoids deadlocks on printk locks without having to break locks. Also @@ -119,11 +136,11 @@ static void print_mce(struct mce *m) print_symbol("{%s}", m->ip); printk("\n"); } - printk(KERN_EMERG "TSC %Lx ", m->tsc); + printk(KERN_EMERG "TSC %llx ", m->tsc); if (m->addr) - printk("ADDR %Lx ", m->addr); + printk("ADDR %llx ", m->addr); if (m->misc) - printk("MISC %Lx ", m->misc); + printk("MISC %llx ", m->misc); printk("\n"); printk(KERN_EMERG "This is not a software problem!\n"); printk(KERN_EMERG "Run through mcelog --ascii to decode " @@ -149,8 +166,10 @@ static void mce_panic(char *msg, struct mce *backup, unsigned long start) panic(msg); } -static int mce_available(struct cpuinfo_x86 *c) +int mce_available(struct cpuinfo_x86 *c) { + if (mce_dont_init) + return 0; return cpu_has(c, X86_FEATURE_MCE) && cpu_has(c, X86_FEATURE_MCA); } @@ -172,7 +191,77 @@ static inline void mce_get_rip(struct mce *m, struct pt_regs *regs) } /* - * The actual machine check handler + * Poll for corrected events or events that happened before reset. + * Those are just logged through /dev/mcelog. + * + * This is executed in standard interrupt context. + */ +void machine_check_poll(enum mcp_flags flags, mce_banks_t *b) +{ + struct mce m; + int i; + + mce_setup(&m); + + rdmsrl(MSR_IA32_MCG_STATUS, m.mcgstatus); + for (i = 0; i < banks; i++) { + if (!bank[i] || !test_bit(i, *b)) + continue; + + m.misc = 0; + m.addr = 0; + m.bank = i; + m.tsc = 0; + + barrier(); + rdmsrl(MSR_IA32_MC0_STATUS + i*4, m.status); + if (!(m.status & MCI_STATUS_VAL)) + continue; + + /* + * Uncorrected events are handled by the exception handler + * when it is enabled. But when the exception is disabled log + * everything. + * + * TBD do the same check for MCI_STATUS_EN here? + */ + if ((m.status & MCI_STATUS_UC) && !(flags & MCP_UC)) + continue; + + if (m.status & MCI_STATUS_MISCV) + rdmsrl(MSR_IA32_MC0_MISC + i*4, m.misc); + if (m.status & MCI_STATUS_ADDRV) + rdmsrl(MSR_IA32_MC0_ADDR + i*4, m.addr); + + if (!(flags & MCP_TIMESTAMP)) + m.tsc = 0; + /* + * Don't get the IP here because it's unlikely to + * have anything to do with the actual error location. + */ + + mce_log(&m); + add_taint(TAINT_MACHINE_CHECK); + + /* + * Clear state for this bank. + */ + wrmsrl(MSR_IA32_MC0_STATUS+4*i, 0); + } + + /* + * Don't clear MCG_STATUS here because it's only defined for + * exceptions. + */ +} + +/* + * The actual machine check handler. This only handles real + * exceptions when something got corrupted coming in through int 18. + * + * This is executed in NMI context not subject to normal locking rules. This + * implies that most kernel services cannot be safely used. Don't even + * think about putting a printk in there! */ void do_machine_check(struct pt_regs * regs, long error_code) { @@ -190,17 +279,18 @@ void do_machine_check(struct pt_regs * regs, long error_code) * error. */ int kill_it = 0; + DECLARE_BITMAP(toclear, MAX_NR_BANKS); atomic_inc(&mce_entry); - if ((regs - && notify_die(DIE_NMI, "machine check", regs, error_code, + if (notify_die(DIE_NMI, "machine check", regs, error_code, 18, SIGKILL) == NOTIFY_STOP) - || !banks) + goto out2; + if (!banks) goto out2; - memset(&m, 0, sizeof(struct mce)); - m.cpu = smp_processor_id(); + mce_setup(&m); + rdmsrl(MSR_IA32_MCG_STATUS, m.mcgstatus); /* if the restart IP is not valid, we're done for */ if (!(m.mcgstatus & MCG_STATUS_RIPV)) @@ -210,18 +300,32 @@ void do_machine_check(struct pt_regs * regs, long error_code) barrier(); for (i = 0; i < banks; i++) { - if (i < NR_SYSFS_BANKS && !bank[i]) + __clear_bit(i, toclear); + if (!bank[i]) continue; m.misc = 0; m.addr = 0; m.bank = i; - m.tsc = 0; rdmsrl(MSR_IA32_MC0_STATUS + i*4, m.status); if ((m.status & MCI_STATUS_VAL) == 0) continue; + /* + * Non uncorrected errors are handled by machine_check_poll + * Leave them alone. + */ + if ((m.status & MCI_STATUS_UC) == 0) + continue; + + /* + * Set taint even when machine check was not enabled. + */ + add_taint(TAINT_MACHINE_CHECK); + + __set_bit(i, toclear); + if (m.status & MCI_STATUS_EN) { /* if PCC was set, there's no way out */ no_way_out |= !!(m.status & MCI_STATUS_PCC); @@ -235,6 +339,12 @@ void do_machine_check(struct pt_regs * regs, long error_code) no_way_out = 1; kill_it = 1; } + } else { + /* + * Machine check event was not enabled. Clear, but + * ignore. + */ + continue; } if (m.status & MCI_STATUS_MISCV) @@ -243,10 +353,7 @@ void do_machine_check(struct pt_regs * regs, long error_code) rdmsrl(MSR_IA32_MC0_ADDR + i*4, m.addr); mce_get_rip(&m, regs); - if (error_code >= 0) - rdtscll(m.tsc); - if (error_code != -2) - mce_log(&m); + mce_log(&m); /* Did this bank cause the exception? */ /* Assume that the bank with uncorrectable errors did it, @@ -255,14 +362,8 @@ void do_machine_check(struct pt_regs * regs, long error_code) panicm = m; panicm_found = 1; } - - add_taint(TAINT_MACHINE_CHECK); } - /* Never do anything final in the polling timer */ - if (!regs) - goto out; - /* If we didn't find an uncorrectable error, pick the last one (shouldn't happen, just being safe). */ if (!panicm_found) @@ -309,10 +410,11 @@ void do_machine_check(struct pt_regs * regs, long error_code) /* notify userspace ASAP */ set_thread_flag(TIF_MCE_NOTIFY); - out: /* the last thing we do is clear state */ - for (i = 0; i < banks; i++) - wrmsrl(MSR_IA32_MC0_STATUS+4*i, 0); + for (i = 0; i < banks; i++) { + if (test_bit(i, toclear)) + wrmsrl(MSR_IA32_MC0_STATUS+4*i, 0); + } wrmsrl(MSR_IA32_MCG_STATUS, 0); out2: atomic_dec(&mce_entry); @@ -332,15 +434,13 @@ void do_machine_check(struct pt_regs * regs, long error_code) * and historically has been the register value of the * MSR_IA32_THERMAL_STATUS (Intel) msr. */ -void mce_log_therm_throt_event(unsigned int cpu, __u64 status) +void mce_log_therm_throt_event(__u64 status) { struct mce m; - memset(&m, 0, sizeof(m)); - m.cpu = cpu; + mce_setup(&m); m.bank = MCE_THERMAL_BANK; m.status = status; - rdtscll(m.tsc); mce_log(&m); } #endif /* CONFIG_X86_MCE_INTEL */ @@ -353,18 +453,18 @@ void mce_log_therm_throt_event(unsigned int cpu, __u64 status) static int check_interval = 5 * 60; /* 5 minutes */ static int next_interval; /* in jiffies */ -static void mcheck_timer(struct work_struct *work); -static DECLARE_DELAYED_WORK(mcheck_work, mcheck_timer); +static void mcheck_timer(unsigned long); +static DEFINE_PER_CPU(struct timer_list, mce_timer); -static void mcheck_check_cpu(void *info) +static void mcheck_timer(unsigned long data) { - if (mce_available(¤t_cpu_data)) - do_machine_check(NULL, 0); -} + struct timer_list *t = &per_cpu(mce_timer, data); -static void mcheck_timer(struct work_struct *work) -{ - on_each_cpu(mcheck_check_cpu, NULL, 1); + WARN_ON(smp_processor_id() != data); + + if (mce_available(¤t_cpu_data)) + machine_check_poll(MCP_TIMESTAMP, + &__get_cpu_var(mce_poll_banks)); /* * Alert userspace if needed. If we logged an MCE, reduce the @@ -377,31 +477,41 @@ static void mcheck_timer(struct work_struct *work) (int)round_jiffies_relative(check_interval*HZ)); } - schedule_delayed_work(&mcheck_work, next_interval); + t->expires = jiffies + next_interval; + add_timer(t); +} + +static void mce_do_trigger(struct work_struct *work) +{ + call_usermodehelper(trigger, trigger_argv, NULL, UMH_NO_WAIT); } +static DECLARE_WORK(mce_trigger_work, mce_do_trigger); + /* - * This is only called from process context. This is where we do - * anything we need to alert userspace about new MCEs. This is called - * directly from the poller and also from entry.S and idle, thanks to - * TIF_MCE_NOTIFY. + * Notify the user(s) about new machine check events. + * Can be called from interrupt context, but not from machine check/NMI + * context. */ int mce_notify_user(void) { + /* Not more than two messages every minute */ + static DEFINE_RATELIMIT_STATE(ratelimit, 60*HZ, 2); + clear_thread_flag(TIF_MCE_NOTIFY); if (test_and_clear_bit(0, ¬ify_user)) { - static unsigned long last_print; - unsigned long now = jiffies; - wake_up_interruptible(&mce_wait); - if (trigger[0]) - call_usermodehelper(trigger, trigger_argv, NULL, - UMH_NO_WAIT); - if (time_after_eq(now, last_print + (check_interval*HZ))) { - last_print = now; + /* + * There is no risk of missing notifications because + * work_pending is always cleared before the function is + * executed. + */ + if (trigger[0] && !work_pending(&mce_trigger_work)) + schedule_work(&mce_trigger_work); + + if (__ratelimit(&ratelimit)) printk(KERN_INFO "Machine check events logged\n"); - } return 1; } @@ -425,63 +535,78 @@ static struct notifier_block mce_idle_notifier = { static __init int periodic_mcheck_init(void) { - next_interval = check_interval * HZ; - if (next_interval) - schedule_delayed_work(&mcheck_work, - round_jiffies_relative(next_interval)); - idle_notifier_register(&mce_idle_notifier); - return 0; + idle_notifier_register(&mce_idle_notifier); + return 0; } __initcall(periodic_mcheck_init); - /* * Initialize Machine Checks for a CPU. */ -static void mce_init(void *dummy) +static int mce_cap_init(void) { u64 cap; - int i; + unsigned b; rdmsrl(MSR_IA32_MCG_CAP, cap); - banks = cap & 0xff; - if (banks > MCE_EXTENDED_BANK) { - banks = MCE_EXTENDED_BANK; - printk(KERN_INFO "MCE: warning: using only %d banks\n", - MCE_EXTENDED_BANK); + b = cap & 0xff; + if (b > MAX_NR_BANKS) { + printk(KERN_WARNING + "MCE: Using only %u machine check banks out of %u\n", + MAX_NR_BANKS, b); + b = MAX_NR_BANKS; } + + /* Don't support asymmetric configurations today */ + WARN_ON(banks != 0 && b != banks); + banks = b; + if (!bank) { + bank = kmalloc(banks * sizeof(u64), GFP_KERNEL); + if (!bank) + return -ENOMEM; + memset(bank, 0xff, banks * sizeof(u64)); + } + /* Use accurate RIP reporting if available. */ if ((cap & (1<<9)) && ((cap >> 16) & 0xff) >= 9) rip_msr = MSR_IA32_MCG_EIP; - /* Log the machine checks left over from the previous reset. - This also clears all registers */ - do_machine_check(NULL, mce_bootlog ? -1 : -2); + return 0; +} + +static void mce_init(void *dummy) +{ + u64 cap; + int i; + mce_banks_t all_banks; + + /* + * Log the machine checks left over from the previous reset. + */ + bitmap_fill(all_banks, MAX_NR_BANKS); + machine_check_poll(MCP_UC, &all_banks); set_in_cr4(X86_CR4_MCE); + rdmsrl(MSR_IA32_MCG_CAP, cap); if (cap & MCG_CTL_P) wrmsr(MSR_IA32_MCG_CTL, 0xffffffff, 0xffffffff); for (i = 0; i < banks; i++) { - if (i < NR_SYSFS_BANKS) - wrmsrl(MSR_IA32_MC0_CTL+4*i, bank[i]); - else - wrmsrl(MSR_IA32_MC0_CTL+4*i, ~0UL); - + wrmsrl(MSR_IA32_MC0_CTL+4*i, bank[i]); wrmsrl(MSR_IA32_MC0_STATUS+4*i, 0); } } /* Add per CPU specific workarounds here */ -static void __cpuinit mce_cpu_quirks(struct cpuinfo_x86 *c) +static void mce_cpu_quirks(struct cpuinfo_x86 *c) { /* This should be disabled by the BIOS, but isn't always */ if (c->x86_vendor == X86_VENDOR_AMD) { - if(c->x86 == 15) + if (c->x86 == 15 && banks > 4) /* disable GART TBL walk error reporting, which trips off incorrectly with the IOMMU & 3ware & Cerberus. */ - clear_bit(10, &bank[4]); + clear_bit(10, (unsigned long *)&bank[4]); if(c->x86 <= 17 && mce_bootlog < 0) /* Lots of broken BIOS around that don't clear them by default and leave crap in there. Don't log. */ @@ -504,20 +629,38 @@ static void mce_cpu_features(struct cpuinfo_x86 *c) } } +static void mce_init_timer(void) +{ + struct timer_list *t = &__get_cpu_var(mce_timer); + + /* data race harmless because everyone sets to the same value */ + if (!next_interval) + next_interval = check_interval * HZ; + if (!next_interval) + return; + setup_timer(t, mcheck_timer, smp_processor_id()); + t->expires = round_jiffies(jiffies + next_interval); + add_timer(t); +} + /* * Called for each booted CPU to set up machine checks. * Must be called with preempt off. */ void __cpuinit mcheck_init(struct cpuinfo_x86 *c) { - mce_cpu_quirks(c); + if (!mce_available(c)) + return; - if (mce_dont_init || - !mce_available(c)) + if (mce_cap_init() < 0) { + mce_dont_init = 1; return; + } + mce_cpu_quirks(c); mce_init(NULL); mce_cpu_features(c); + mce_init_timer(); } /* @@ -573,7 +716,7 @@ static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize, { unsigned long *cpu_tsc; static DEFINE_MUTEX(mce_read_mutex); - unsigned next; + unsigned prev, next; char __user *buf = ubuf; int i, err; @@ -592,25 +735,32 @@ static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize, } err = 0; - for (i = 0; i < next; i++) { - unsigned long start = jiffies; - - while (!mcelog.entry[i].finished) { - if (time_after_eq(jiffies, start + 2)) { - memset(mcelog.entry + i,0, sizeof(struct mce)); - goto timeout; + prev = 0; + do { + for (i = prev; i < next; i++) { + unsigned long start = jiffies; + + while (!mcelog.entry[i].finished) { + if (time_after_eq(jiffies, start + 2)) { + memset(mcelog.entry + i, 0, + sizeof(struct mce)); + goto timeout; + } + cpu_relax(); } - cpu_relax(); + smp_rmb(); + err |= copy_to_user(buf, mcelog.entry + i, + sizeof(struct mce)); + buf += sizeof(struct mce); +timeout: + ; } - smp_rmb(); - err |= copy_to_user(buf, mcelog.entry + i, sizeof(struct mce)); - buf += sizeof(struct mce); - timeout: - ; - } - memset(mcelog.entry, 0, next * sizeof(struct mce)); - mcelog.next = 0; + memset(mcelog.entry + prev, 0, + (next - prev) * sizeof(struct mce)); + prev = next; + next = cmpxchg(&mcelog.next, prev, 0); + } while (next != prev); synchronize_sched(); @@ -680,20 +830,6 @@ static struct miscdevice mce_log_device = { &mce_chrdev_ops, }; -static unsigned long old_cr4 __initdata; - -void __init stop_mce(void) -{ - old_cr4 = read_cr4(); - clear_in_cr4(X86_CR4_MCE); -} - -void __init restart_mce(void) -{ - if (old_cr4 & X86_CR4_MCE) - set_in_cr4(X86_CR4_MCE); -} - /* * Old style boot options parsing. Only for compatibility. */ @@ -703,8 +839,7 @@ static int __init mcheck_disable(char *str) return 1; } -/* mce=off disables machine check. Note you can re-enable it later - using sysfs. +/* mce=off disables machine check. mce=TOLERANCELEVEL (number, see above) mce=bootlog Log MCEs from before booting. Disabled by default on AMD. mce=nobootlog Don't log MCEs from before booting. */ @@ -728,6 +863,29 @@ __setup("mce=", mcheck_enable); * Sysfs support */ +/* + * Disable machine checks on suspend and shutdown. We can't really handle + * them later. + */ +static int mce_disable(void) +{ + int i; + + for (i = 0; i < banks; i++) + wrmsrl(MSR_IA32_MC0_CTL + i*4, 0); + return 0; +} + +static int mce_suspend(struct sys_device *dev, pm_message_t state) +{ + return mce_disable(); +} + +static int mce_shutdown(struct sys_device *dev) +{ + return mce_disable(); +} + /* On resume clear all MCE state. Don't want to see leftovers from the BIOS. Only one CPU is active at this time, the others get readded later using CPU hotplug. */ @@ -738,20 +896,24 @@ static int mce_resume(struct sys_device *dev) return 0; } +static void mce_cpu_restart(void *data) +{ + del_timer_sync(&__get_cpu_var(mce_timer)); + if (mce_available(¤t_cpu_data)) + mce_init(NULL); + mce_init_timer(); +} + /* Reinit MCEs after user configuration changes */ static void mce_restart(void) { - if (next_interval) - cancel_delayed_work(&mcheck_work); - /* Timer race is harmless here */ - on_each_cpu(mce_init, NULL, 1); next_interval = check_interval * HZ; - if (next_interval) - schedule_delayed_work(&mcheck_work, - round_jiffies_relative(next_interval)); + on_each_cpu(mce_cpu_restart, NULL, 1); } static struct sysdev_class mce_sysclass = { + .suspend = mce_suspend, + .shutdown = mce_shutdown, .resume = mce_resume, .name = "machinecheck", }; @@ -778,16 +940,26 @@ void (*threshold_cpu_callback)(unsigned long action, unsigned int cpu) __cpuinit } \ static SYSDEV_ATTR(name, 0644, show_ ## name, set_ ## name); -/* - * TBD should generate these dynamically based on number of available banks. - * Have only 6 contol banks in /sysfs until then. - */ -ACCESSOR(bank0ctl,bank[0],mce_restart()) -ACCESSOR(bank1ctl,bank[1],mce_restart()) -ACCESSOR(bank2ctl,bank[2],mce_restart()) -ACCESSOR(bank3ctl,bank[3],mce_restart()) -ACCESSOR(bank4ctl,bank[4],mce_restart()) -ACCESSOR(bank5ctl,bank[5],mce_restart()) +static struct sysdev_attribute *bank_attrs; + +static ssize_t show_bank(struct sys_device *s, struct sysdev_attribute *attr, + char *buf) +{ + u64 b = bank[attr - bank_attrs]; + return sprintf(buf, "%llx\n", b); +} + +static ssize_t set_bank(struct sys_device *s, struct sysdev_attribute *attr, + const char *buf, size_t siz) +{ + char *end; + u64 new = simple_strtoull(buf, &end, 0); + if (end == buf) + return -EINVAL; + bank[attr - bank_attrs] = new; + mce_restart(); + return end-buf; +} static ssize_t show_trigger(struct sys_device *s, struct sysdev_attribute *attr, char *buf) @@ -814,8 +986,6 @@ static SYSDEV_ATTR(trigger, 0644, show_trigger, set_trigger); static SYSDEV_INT_ATTR(tolerant, 0644, tolerant); ACCESSOR(check_interval,check_interval,mce_restart()) static struct sysdev_attribute *mce_attributes[] = { - &attr_bank0ctl, &attr_bank1ctl, &attr_bank2ctl, - &attr_bank3ctl, &attr_bank4ctl, &attr_bank5ctl, &attr_tolerant.attr, &attr_check_interval, &attr_trigger, NULL }; @@ -845,11 +1015,22 @@ static __cpuinit int mce_create_device(unsigned int cpu) if (err) goto error; } + for (i = 0; i < banks; i++) { + err = sysdev_create_file(&per_cpu(device_mce, cpu), + &bank_attrs[i]); + if (err) + goto error2; + } cpu_set(cpu, mce_device_initialized); return 0; +error2: + while (--i >= 0) { + sysdev_remove_file(&per_cpu(device_mce, cpu), + &bank_attrs[i]); + } error: - while (i--) { + while (--i >= 0) { sysdev_remove_file(&per_cpu(device_mce,cpu), mce_attributes[i]); } @@ -868,15 +1049,46 @@ static __cpuinit void mce_remove_device(unsigned int cpu) for (i = 0; mce_attributes[i]; i++) sysdev_remove_file(&per_cpu(device_mce,cpu), mce_attributes[i]); + for (i = 0; i < banks; i++) + sysdev_remove_file(&per_cpu(device_mce, cpu), + &bank_attrs[i]); sysdev_unregister(&per_cpu(device_mce,cpu)); cpu_clear(cpu, mce_device_initialized); } +/* Make sure there are no machine checks on offlined CPUs. */ +static void mce_disable_cpu(void *h) +{ + int i; + unsigned long action = *(unsigned long *)h; + + if (!mce_available(¤t_cpu_data)) + return; + if (!(action & CPU_TASKS_FROZEN)) + cmci_clear(); + for (i = 0; i < banks; i++) + wrmsrl(MSR_IA32_MC0_CTL + i*4, 0); +} + +static void mce_reenable_cpu(void *h) +{ + int i; + unsigned long action = *(unsigned long *)h; + + if (!mce_available(¤t_cpu_data)) + return; + if (!(action & CPU_TASKS_FROZEN)) + cmci_reenable(); + for (i = 0; i < banks; i++) + wrmsrl(MSR_IA32_MC0_CTL + i*4, bank[i]); +} + /* Get notified when a cpu comes on/off. Be hotplug friendly. */ static int __cpuinit mce_cpu_callback(struct notifier_block *nfb, unsigned long action, void *hcpu) { unsigned int cpu = (unsigned long)hcpu; + struct timer_list *t = &per_cpu(mce_timer, cpu); switch (action) { case CPU_ONLINE: @@ -891,6 +1103,21 @@ static int __cpuinit mce_cpu_callback(struct notifier_block *nfb, threshold_cpu_callback(action, cpu); mce_remove_device(cpu); break; + case CPU_DOWN_PREPARE: + case CPU_DOWN_PREPARE_FROZEN: + del_timer_sync(t); + smp_call_function_single(cpu, mce_disable_cpu, &action, 1); + break; + case CPU_DOWN_FAILED: + case CPU_DOWN_FAILED_FROZEN: + t->expires = round_jiffies(jiffies + next_interval); + add_timer_on(t, cpu); + smp_call_function_single(cpu, mce_reenable_cpu, &action, 1); + break; + case CPU_POST_DEAD: + /* intentionally ignoring frozen here */ + cmci_rediscover(cpu); + break; } return NOTIFY_OK; } @@ -899,6 +1126,34 @@ static struct notifier_block mce_cpu_notifier __cpuinitdata = { .notifier_call = mce_cpu_callback, }; +static __init int mce_init_banks(void) +{ + int i; + + bank_attrs = kzalloc(sizeof(struct sysdev_attribute) * banks, + GFP_KERNEL); + if (!bank_attrs) + return -ENOMEM; + + for (i = 0; i < banks; i++) { + struct sysdev_attribute *a = &bank_attrs[i]; + a->attr.name = kasprintf(GFP_KERNEL, "bank%d", i); + if (!a->attr.name) + goto nomem; + a->attr.mode = 0644; + a->show = show_bank; + a->store = set_bank; + } + return 0; + +nomem: + while (--i >= 0) + kfree(bank_attrs[i].attr.name); + kfree(bank_attrs); + bank_attrs = NULL; + return -ENOMEM; +} + static __init int mce_init_device(void) { int err; @@ -906,6 +1161,11 @@ static __init int mce_init_device(void) if (!mce_available(&boot_cpu_data)) return -EIO; + + err = mce_init_banks(); + if (err) + return err; + err = sysdev_class_register(&mce_sysclass); if (err) return err; diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd_64.c b/arch/x86/kernel/cpu/mcheck/mce_amd_64.c index 9817506..c5a32f9 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_amd_64.c +++ b/arch/x86/kernel/cpu/mcheck/mce_amd_64.c @@ -79,6 +79,8 @@ static unsigned char shared_bank[NR_BANKS] = { static DEFINE_PER_CPU(unsigned char, bank_map); /* see which banks are on */ +static void amd_threshold_interrupt(void); + /* * CPU Initialization */ @@ -174,6 +176,8 @@ void mce_amd_feature_init(struct cpuinfo_x86 *c) tr.reset = 0; tr.old_limit = 0; threshold_restart_bank(&tr); + + mce_threshold_vector = amd_threshold_interrupt; } } } @@ -187,19 +191,13 @@ void mce_amd_feature_init(struct cpuinfo_x86 *c) * the interrupt goes off when error_count reaches threshold_limit. * the handler will simply log mcelog w/ software defined bank number. */ -asmlinkage void mce_threshold_interrupt(void) +static void amd_threshold_interrupt(void) { unsigned int bank, block; struct mce m; u32 low = 0, high = 0, address = 0; - ack_APIC_irq(); - exit_idle(); - irq_enter(); - - memset(&m, 0, sizeof(m)); - rdtscll(m.tsc); - m.cpu = smp_processor_id(); + mce_setup(&m); /* assume first bank caused it */ for (bank = 0; bank < NR_BANKS; ++bank) { @@ -233,7 +231,8 @@ asmlinkage void mce_threshold_interrupt(void) /* Log the machine check that caused the threshold event. */ - do_machine_check(NULL, 0); + machine_check_poll(MCP_TIMESTAMP, + &__get_cpu_var(mce_poll_banks)); if (high & MASK_OVERFLOW_HI) { rdmsrl(address, m.misc); @@ -243,13 +242,10 @@ asmlinkage void mce_threshold_interrupt(void) + bank * NR_BLOCKS + block; mce_log(&m); - goto out; + return; } } } -out: - inc_irq_stat(irq_threshold_count); - irq_exit(); } /* diff --git a/arch/x86/kernel/cpu/mcheck/mce_intel_64.c b/arch/x86/kernel/cpu/mcheck/mce_intel_64.c index aa5e287..aaa7d97 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_intel_64.c +++ b/arch/x86/kernel/cpu/mcheck/mce_intel_64.c @@ -1,6 +1,8 @@ /* * Intel specific MCE features. * Copyright 2004 Zwane Mwaikambo <zwane@linuxpower.ca> + * Copyright (C) 2008, 2009 Intel Corporation + * Author: Andi Kleen */ #include <linux/init.h> @@ -13,6 +15,7 @@ #include <asm/hw_irq.h> #include <asm/idle.h> #include <asm/therm_throt.h> +#include <asm/apic.h> asmlinkage void smp_thermal_interrupt(void) { @@ -25,7 +28,7 @@ asmlinkage void smp_thermal_interrupt(void) rdmsrl(MSR_IA32_THERM_STATUS, msr_val); if (therm_throt_process(msr_val & 1)) - mce_log_therm_throt_event(smp_processor_id(), msr_val); + mce_log_therm_throt_event(msr_val); inc_irq_stat(irq_thermal_count); irq_exit(); @@ -85,7 +88,209 @@ static void intel_init_thermal(struct cpuinfo_x86 *c) return; } +/* + * Support for Intel Correct Machine Check Interrupts. This allows + * the CPU to raise an interrupt when a corrected machine check happened. + * Normally we pick those up using a regular polling timer. + * Also supports reliable discovery of shared banks. + */ + +static DEFINE_PER_CPU(mce_banks_t, mce_banks_owned); + +/* + * cmci_discover_lock protects against parallel discovery attempts + * which could race against each other. + */ +static DEFINE_SPINLOCK(cmci_discover_lock); + +#define CMCI_THRESHOLD 1 + +static int cmci_supported(int *banks) +{ + u64 cap; + + /* + * Vendor check is not strictly needed, but the initial + * initialization is vendor keyed and this + * makes sure none of the backdoors are entered otherwise. + */ + if (boot_cpu_data.x86_vendor != X86_VENDOR_INTEL) + return 0; + if (!cpu_has_apic || lapic_get_maxlvt() < 6) + return 0; + rdmsrl(MSR_IA32_MCG_CAP, cap); + *banks = min_t(unsigned, MAX_NR_BANKS, cap & 0xff); + return !!(cap & MCG_CMCI_P); +} + +/* + * The interrupt handler. This is called on every event. + * Just call the poller directly to log any events. + * This could in theory increase the threshold under high load, + * but doesn't for now. + */ +static void intel_threshold_interrupt(void) +{ + machine_check_poll(MCP_TIMESTAMP, &__get_cpu_var(mce_banks_owned)); + mce_notify_user(); +} + +static void print_update(char *type, int *hdr, int num) +{ + if (*hdr == 0) + printk(KERN_INFO "CPU %d MCA banks", smp_processor_id()); + *hdr = 1; + printk(KERN_CONT " %s:%d", type, num); +} + +/* + * Enable CMCI (Corrected Machine Check Interrupt) for available MCE banks + * on this CPU. Use the algorithm recommended in the SDM to discover shared + * banks. + */ +static void cmci_discover(int banks, int boot) +{ + unsigned long *owned = (void *)&__get_cpu_var(mce_banks_owned); + int hdr = 0; + int i; + + spin_lock(&cmci_discover_lock); + for (i = 0; i < banks; i++) { + u64 val; + + if (test_bit(i, owned)) + continue; + + rdmsrl(MSR_IA32_MC0_CTL2 + i, val); + + /* Already owned by someone else? */ + if (val & CMCI_EN) { + if (test_and_clear_bit(i, owned) || boot) + print_update("SHD", &hdr, i); + __clear_bit(i, __get_cpu_var(mce_poll_banks)); + continue; + } + + val |= CMCI_EN | CMCI_THRESHOLD; + wrmsrl(MSR_IA32_MC0_CTL2 + i, val); + rdmsrl(MSR_IA32_MC0_CTL2 + i, val); + + /* Did the enable bit stick? -- the bank supports CMCI */ + if (val & CMCI_EN) { + if (!test_and_set_bit(i, owned) || boot) + print_update("CMCI", &hdr, i); + __clear_bit(i, __get_cpu_var(mce_poll_banks)); + } else { + WARN_ON(!test_bit(i, __get_cpu_var(mce_poll_banks))); + } + } + spin_unlock(&cmci_discover_lock); + if (hdr) + printk(KERN_CONT "\n"); +} + +/* + * Just in case we missed an event during initialization check + * all the CMCI owned banks. + */ +void cmci_recheck(void) +{ + unsigned long flags; + int banks; + + if (!mce_available(¤t_cpu_data) || !cmci_supported(&banks)) + return; + local_irq_save(flags); + machine_check_poll(MCP_TIMESTAMP, &__get_cpu_var(mce_banks_owned)); + local_irq_restore(flags); +} + +/* + * Disable CMCI on this CPU for all banks it owns when it goes down. + * This allows other CPUs to claim the banks on rediscovery. + */ +void cmci_clear(void) +{ + int i; + int banks; + u64 val; + + if (!cmci_supported(&banks)) + return; + spin_lock(&cmci_discover_lock); + for (i = 0; i < banks; i++) { + if (!test_bit(i, __get_cpu_var(mce_banks_owned))) + continue; + /* Disable CMCI */ + rdmsrl(MSR_IA32_MC0_CTL2 + i, val); + val &= ~(CMCI_EN|CMCI_THRESHOLD_MASK); + wrmsrl(MSR_IA32_MC0_CTL2 + i, val); + __clear_bit(i, __get_cpu_var(mce_banks_owned)); + } + spin_unlock(&cmci_discover_lock); +} + +/* + * After a CPU went down cycle through all the others and rediscover + * Must run in process context. + */ +void cmci_rediscover(int dying) +{ + int banks; + int cpu; + cpumask_var_t old; + + if (!cmci_supported(&banks)) + return; + if (!alloc_cpumask_var(&old, GFP_KERNEL)) + return; + cpumask_copy(old, ¤t->cpus_allowed); + + for_each_online_cpu (cpu) { + if (cpu == dying) + continue; + if (set_cpus_allowed_ptr(current, &cpumask_of_cpu(cpu))) + continue; + /* Recheck banks in case CPUs don't all have the same */ + if (cmci_supported(&banks)) + cmci_discover(banks, 0); + } + + set_cpus_allowed_ptr(current, old); + free_cpumask_var(old); +} + +/* + * Reenable CMCI on this CPU in case a CPU down failed. + */ +void cmci_reenable(void) +{ + int banks; + if (cmci_supported(&banks)) + cmci_discover(banks, 0); +} + +static __cpuinit void intel_init_cmci(void) +{ + int banks; + + if (!cmci_supported(&banks)) + return; + + mce_threshold_vector = intel_threshold_interrupt; + cmci_discover(banks, 1); + /* + * For CPU #0 this runs with still disabled APIC, but that's + * ok because only the vector is set up. We still do another + * check for the banks later for CPU #0 just to make sure + * to not miss any events. + */ + apic_write(APIC_LVTCMCI, THRESHOLD_APIC_VECTOR|APIC_DM_FIXED); + cmci_recheck(); +} + void mce_intel_feature_init(struct cpuinfo_x86 *c) { intel_init_thermal(c); + intel_init_cmci(); } diff --git a/arch/x86/kernel/cpu/mcheck/threshold.c b/arch/x86/kernel/cpu/mcheck/threshold.c new file mode 100644 index 0000000..23ee9e7 --- /dev/null +++ b/arch/x86/kernel/cpu/mcheck/threshold.c @@ -0,0 +1,29 @@ +/* + * Common corrected MCE threshold handler code: + */ +#include <linux/interrupt.h> +#include <linux/kernel.h> + +#include <asm/irq_vectors.h> +#include <asm/apic.h> +#include <asm/idle.h> +#include <asm/mce.h> + +static void default_threshold_interrupt(void) +{ + printk(KERN_ERR "Unexpected threshold interrupt at vector %x\n", + THRESHOLD_APIC_VECTOR); +} + +void (*mce_threshold_vector)(void) = default_threshold_interrupt; + +asmlinkage void mce_threshold_interrupt(void) +{ + exit_idle(); + irq_enter(); + inc_irq_stat(irq_threshold_count); + mce_threshold_vector(); + irq_exit(); + /* Ack only at the end to avoid potential reentry */ + ack_APIC_irq(); +} diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index 169a120..87b67e3 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -729,7 +729,7 @@ struct pebs_tracer *ds_request_pebs(struct task_struct *task, spin_unlock_irqrestore(&ds_lock, irq); - ds_write_config(tracer->ds.context, &tracer->trace.ds, ds_bts); + ds_write_config(tracer->ds.context, &tracer->trace.ds, ds_pebs); ds_resume_pebs(tracer); return tracer; @@ -1029,5 +1029,4 @@ void ds_copy_thread(struct task_struct *tsk, struct task_struct *father) void ds_exit_thread(struct task_struct *tsk) { - WARN_ON(tsk->thread.ds_ctx); } diff --git a/arch/x86/kernel/efi.c b/arch/x86/kernel/efi.c index b205272..1736acc 100644 --- a/arch/x86/kernel/efi.c +++ b/arch/x86/kernel/efi.c @@ -469,7 +469,7 @@ void __init efi_enter_virtual_mode(void) efi_memory_desc_t *md; efi_status_t status; unsigned long size; - u64 end, systab, addr, npages; + u64 end, systab, addr, npages, end_pfn; void *p, *va; efi.systab = NULL; @@ -481,7 +481,10 @@ void __init efi_enter_virtual_mode(void) size = md->num_pages << EFI_PAGE_SHIFT; end = md->phys_addr + size; - if (PFN_UP(end) <= max_low_pfn_mapped) + end_pfn = PFN_UP(end); + if (end_pfn <= max_low_pfn_mapped + || (end_pfn > (1UL << (32 - PAGE_SHIFT)) + && end_pfn <= max_pfn_mapped)) va = __va(md->phys_addr); else va = efi_ioremap(md->phys_addr, size); diff --git a/arch/x86/kernel/efi_64.c b/arch/x86/kernel/efi_64.c index a4ee291..22c3b78 100644 --- a/arch/x86/kernel/efi_64.c +++ b/arch/x86/kernel/efi_64.c @@ -100,24 +100,11 @@ void __init efi_call_phys_epilog(void) void __iomem *__init efi_ioremap(unsigned long phys_addr, unsigned long size) { - static unsigned pages_mapped __initdata; - unsigned i, pages; - unsigned long offset; + unsigned long last_map_pfn; - pages = PFN_UP(phys_addr + size) - PFN_DOWN(phys_addr); - offset = phys_addr & ~PAGE_MASK; - phys_addr &= PAGE_MASK; - - if (pages_mapped + pages > MAX_EFI_IO_PAGES) + last_map_pfn = init_memory_mapping(phys_addr, phys_addr + size); + if ((last_map_pfn << PAGE_SHIFT) < phys_addr + size) return NULL; - for (i = 0; i < pages; i++) { - __set_fixmap(FIX_EFI_IO_MAP_FIRST_PAGE - pages_mapped, - phys_addr, PAGE_KERNEL); - phys_addr += PAGE_SIZE; - pages_mapped++; - } - - return (void __iomem *)__fix_to_virt(FIX_EFI_IO_MAP_FIRST_PAGE - \ - (pages_mapped - pages)) + offset; + return (void __iomem *)__va(phys_addr); } diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S index 83d1836..7ba4621 100644 --- a/arch/x86/kernel/entry_64.S +++ b/arch/x86/kernel/entry_64.S @@ -984,6 +984,8 @@ apicinterrupt UV_BAU_MESSAGE \ #endif apicinterrupt LOCAL_TIMER_VECTOR \ apic_timer_interrupt smp_apic_timer_interrupt +apicinterrupt GENERIC_INTERRUPT_VECTOR \ + generic_interrupt smp_generic_interrupt #ifdef CONFIG_SMP apicinterrupt INVALIDATE_TLB_VECTOR_START+0 \ diff --git a/arch/x86/kernel/i387.c b/arch/x86/kernel/i387.c index b0f61f0..f2f8540 100644 --- a/arch/x86/kernel/i387.c +++ b/arch/x86/kernel/i387.c @@ -136,7 +136,7 @@ int init_fpu(struct task_struct *tsk) #ifdef CONFIG_X86_32 if (!HAVE_HWFP) { memset(tsk->thread.xstate, 0, xstate_size); - finit(); + finit_task(tsk); set_stopped_child_used_math(tsk); return 0; } diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c index f13ca16..b864341 100644 --- a/arch/x86/kernel/irq.c +++ b/arch/x86/kernel/irq.c @@ -15,6 +15,9 @@ atomic_t irq_err_count; +/* Function pointer for generic interrupt vector handling */ +void (*generic_interrupt_extension)(void) = NULL; + /* * 'what should we do if we get a hw irq event on an illegal vector'. * each architecture has to answer this themselves. @@ -56,6 +59,12 @@ static int show_other_interrupts(struct seq_file *p) seq_printf(p, "%10u ", irq_stats(j)->apic_timer_irqs); seq_printf(p, " Local timer interrupts\n"); #endif + if (generic_interrupt_extension) { + seq_printf(p, "PLT: "); + for_each_online_cpu(j) + seq_printf(p, "%10u ", irq_stats(j)->generic_irqs); + seq_printf(p, " Platform interrupts\n"); + } #ifdef CONFIG_SMP seq_printf(p, "RES: "); for_each_online_cpu(j) @@ -163,6 +172,8 @@ u64 arch_irq_stat_cpu(unsigned int cpu) #ifdef CONFIG_X86_LOCAL_APIC sum += irq_stats(cpu)->apic_timer_irqs; #endif + if (generic_interrupt_extension) + sum += irq_stats(cpu)->generic_irqs; #ifdef CONFIG_SMP sum += irq_stats(cpu)->irq_resched_count; sum += irq_stats(cpu)->irq_call_count; @@ -226,4 +237,27 @@ unsigned int __irq_entry do_IRQ(struct pt_regs *regs) return 1; } +/* + * Handler for GENERIC_INTERRUPT_VECTOR. + */ +void smp_generic_interrupt(struct pt_regs *regs) +{ + struct pt_regs *old_regs = set_irq_regs(regs); + + ack_APIC_irq(); + + exit_idle(); + + irq_enter(); + + inc_irq_stat(generic_irqs); + + if (generic_interrupt_extension) + generic_interrupt_extension(); + + irq_exit(); + + set_irq_regs(old_regs); +} + EXPORT_SYMBOL_GPL(vector_used_by_percpu_irq); diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index 9dc6b2b..3b09634 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -16,6 +16,7 @@ #include <linux/cpu.h> #include <linux/delay.h> #include <linux/uaccess.h> +#include <linux/percpu.h> #include <asm/apic.h> @@ -55,13 +56,13 @@ static inline void print_stack_overflow(void) { } union irq_ctx { struct thread_info tinfo; u32 stack[THREAD_SIZE/sizeof(u32)]; -}; +} __attribute__((aligned(PAGE_SIZE))); -static union irq_ctx *hardirq_ctx[NR_CPUS] __read_mostly; -static union irq_ctx *softirq_ctx[NR_CPUS] __read_mostly; +static DEFINE_PER_CPU(union irq_ctx *, hardirq_ctx); +static DEFINE_PER_CPU(union irq_ctx *, softirq_ctx); -static char softirq_stack[NR_CPUS * THREAD_SIZE] __page_aligned_bss; -static char hardirq_stack[NR_CPUS * THREAD_SIZE] __page_aligned_bss; +static DEFINE_PER_CPU_PAGE_ALIGNED(union irq_ctx, hardirq_stack); +static DEFINE_PER_CPU_PAGE_ALIGNED(union irq_ctx, softirq_stack); static void call_on_stack(void *func, void *stack) { @@ -81,7 +82,7 @@ execute_on_irq_stack(int overflow, struct irq_desc *desc, int irq) u32 *isp, arg1, arg2; curctx = (union irq_ctx *) current_thread_info(); - irqctx = hardirq_ctx[smp_processor_id()]; + irqctx = __get_cpu_var(hardirq_ctx); /* * this is where we switch to the IRQ stack. However, if we are @@ -125,34 +126,34 @@ void __cpuinit irq_ctx_init(int cpu) { union irq_ctx *irqctx; - if (hardirq_ctx[cpu]) + if (per_cpu(hardirq_ctx, cpu)) return; - irqctx = (union irq_ctx*) &hardirq_stack[cpu*THREAD_SIZE]; + irqctx = &per_cpu(hardirq_stack, cpu); irqctx->tinfo.task = NULL; irqctx->tinfo.exec_domain = NULL; irqctx->tinfo.cpu = cpu; irqctx->tinfo.preempt_count = HARDIRQ_OFFSET; irqctx->tinfo.addr_limit = MAKE_MM_SEG(0); - hardirq_ctx[cpu] = irqctx; + per_cpu(hardirq_ctx, cpu) = irqctx; - irqctx = (union irq_ctx *) &softirq_stack[cpu*THREAD_SIZE]; + irqctx = &per_cpu(softirq_stack, cpu); irqctx->tinfo.task = NULL; irqctx->tinfo.exec_domain = NULL; irqctx->tinfo.cpu = cpu; irqctx->tinfo.preempt_count = 0; irqctx->tinfo.addr_limit = MAKE_MM_SEG(0); - softirq_ctx[cpu] = irqctx; + per_cpu(softirq_ctx, cpu) = irqctx; printk(KERN_DEBUG "CPU %u irqstacks, hard=%p soft=%p\n", - cpu, hardirq_ctx[cpu], softirq_ctx[cpu]); + cpu, per_cpu(hardirq_ctx, cpu), per_cpu(softirq_ctx, cpu)); } void irq_ctx_exit(int cpu) { - hardirq_ctx[cpu] = NULL; + per_cpu(hardirq_ctx, cpu) = NULL; } asmlinkage void do_softirq(void) @@ -169,7 +170,7 @@ asmlinkage void do_softirq(void) if (local_softirq_pending()) { curctx = current_thread_info(); - irqctx = softirq_ctx[smp_processor_id()]; + irqctx = __get_cpu_var(softirq_ctx); irqctx->tinfo.task = curctx->task; irqctx->tinfo.previous_esp = current_stack_pointer; diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c index 50b8c3a..bc13261 100644 --- a/arch/x86/kernel/irqinit_32.c +++ b/arch/x86/kernel/irqinit_32.c @@ -175,6 +175,9 @@ void __init native_init_IRQ(void) /* self generated IPI for local APIC timer */ alloc_intr_gate(LOCAL_TIMER_VECTOR, apic_timer_interrupt); + /* generic IPI for platform specific use */ + alloc_intr_gate(GENERIC_INTERRUPT_VECTOR, generic_interrupt); + /* IPI vectors for APIC spurious and error interrupts */ alloc_intr_gate(SPURIOUS_APIC_VECTOR, spurious_interrupt); alloc_intr_gate(ERROR_APIC_VECTOR, error_interrupt); diff --git a/arch/x86/kernel/irqinit_64.c b/arch/x86/kernel/irqinit_64.c index da481a1..c7a49e0 100644 --- a/arch/x86/kernel/irqinit_64.c +++ b/arch/x86/kernel/irqinit_64.c @@ -147,6 +147,9 @@ static void __init apic_intr_init(void) /* self generated IPI for local APIC timer */ alloc_intr_gate(LOCAL_TIMER_VECTOR, apic_timer_interrupt); + /* generic IPI for platform specific use */ + alloc_intr_gate(GENERIC_INTERRUPT_VECTOR, generic_interrupt); + /* IPI vectors for APIC spurious and error interrupts */ alloc_intr_gate(SPURIOUS_APIC_VECTOR, spurious_interrupt); alloc_intr_gate(ERROR_APIC_VECTOR, error_interrupt); diff --git a/arch/x86/kernel/machine_kexec_32.c b/arch/x86/kernel/machine_kexec_32.c index f5fc8c7..e7368c1 100644 --- a/arch/x86/kernel/machine_kexec_32.c +++ b/arch/x86/kernel/machine_kexec_32.c @@ -14,12 +14,12 @@ #include <linux/ftrace.h> #include <linux/suspend.h> #include <linux/gfp.h> +#include <linux/io.h> #include <asm/pgtable.h> #include <asm/pgalloc.h> #include <asm/tlbflush.h> #include <asm/mmu_context.h> -#include <asm/io.h> #include <asm/apic.h> #include <asm/cpufeature.h> #include <asm/desc.h> @@ -63,7 +63,7 @@ static void load_segments(void) "\tmovl %%eax,%%fs\n" "\tmovl %%eax,%%gs\n" "\tmovl %%eax,%%ss\n" - ::: "eax", "memory"); + : : : "eax", "memory"); #undef STR #undef __STR } @@ -205,7 +205,8 @@ void machine_kexec(struct kimage *image) if (image->preserve_context) { #ifdef CONFIG_X86_IO_APIC - /* We need to put APICs in legacy mode so that we can + /* + * We need to put APICs in legacy mode so that we can * get timer interrupts in second kernel. kexec/kdump * paths already have calls to disable_IO_APIC() in * one form or other. kexec jump path also need @@ -227,7 +228,8 @@ void machine_kexec(struct kimage *image) page_list[PA_SWAP_PAGE] = (page_to_pfn(image->swap_page) << PAGE_SHIFT); - /* The segment registers are funny things, they have both a + /* + * The segment registers are funny things, they have both a * visible and an invisible part. Whenever the visible part is * set to a specific selector, the invisible part is loaded * with from a table in memory. At no other time is the @@ -237,11 +239,12 @@ void machine_kexec(struct kimage *image) * segments, before I zap the gdt with an invalid value. */ load_segments(); - /* The gdt & idt are now invalid. + /* + * The gdt & idt are now invalid. * If you want to load them you must set up your own idt & gdt. */ - set_gdt(phys_to_virt(0),0); - set_idt(phys_to_virt(0),0); + set_gdt(phys_to_virt(0), 0); + set_idt(phys_to_virt(0), 0); /* now call it */ image->start = relocate_kernel_ptr((unsigned long)image->head, diff --git a/arch/x86/kernel/machine_kexec_64.c b/arch/x86/kernel/machine_kexec_64.c index 6993d51..89cea4d 100644 --- a/arch/x86/kernel/machine_kexec_64.c +++ b/arch/x86/kernel/machine_kexec_64.c @@ -12,11 +12,47 @@ #include <linux/reboot.h> #include <linux/numa.h> #include <linux/ftrace.h> +#include <linux/io.h> +#include <linux/suspend.h> #include <asm/pgtable.h> #include <asm/tlbflush.h> #include <asm/mmu_context.h> -#include <asm/io.h> + +static int init_one_level2_page(struct kimage *image, pgd_t *pgd, + unsigned long addr) +{ + pud_t *pud; + pmd_t *pmd; + struct page *page; + int result = -ENOMEM; + + addr &= PMD_MASK; + pgd += pgd_index(addr); + if (!pgd_present(*pgd)) { + page = kimage_alloc_control_pages(image, 0); + if (!page) + goto out; + pud = (pud_t *)page_address(page); + memset(pud, 0, PAGE_SIZE); + set_pgd(pgd, __pgd(__pa(pud) | _KERNPG_TABLE)); + } + pud = pud_offset(pgd, addr); + if (!pud_present(*pud)) { + page = kimage_alloc_control_pages(image, 0); + if (!page) + goto out; + pmd = (pmd_t *)page_address(page); + memset(pmd, 0, PAGE_SIZE); + set_pud(pud, __pud(__pa(pmd) | _KERNPG_TABLE)); + } + pmd = pmd_offset(pud, addr); + if (!pmd_present(*pmd)) + set_pmd(pmd, __pmd(addr | __PAGE_KERNEL_LARGE_EXEC)); + result = 0; +out: + return result; +} static void init_level2_page(pmd_t *level2p, unsigned long addr) { @@ -83,9 +119,8 @@ static int init_level4_page(struct kimage *image, pgd_t *level4p, } level3p = (pud_t *)page_address(page); result = init_level3_page(image, level3p, addr, last_addr); - if (result) { + if (result) goto out; - } set_pgd(level4p++, __pgd(__pa(level3p) | _KERNPG_TABLE)); addr += PGDIR_SIZE; } @@ -156,6 +191,13 @@ static int init_pgtable(struct kimage *image, unsigned long start_pgtable) result = init_level4_page(image, level4p, 0, max_pfn << PAGE_SHIFT); if (result) return result; + /* + * image->start may be outside 0 ~ max_pfn, for example when + * jump back to original kernel from kexeced kernel + */ + result = init_one_level2_page(image, level4p, image->start); + if (result) + return result; return init_transition_pgtable(image, level4p); } @@ -229,20 +271,45 @@ void machine_kexec(struct kimage *image) { unsigned long page_list[PAGES_NR]; void *control_page; + int save_ftrace_enabled; - tracer_disable(); +#ifdef CONFIG_KEXEC_JUMP + if (kexec_image->preserve_context) + save_processor_state(); +#endif + + save_ftrace_enabled = __ftrace_enabled_save(); /* Interrupts aren't acceptable while we reboot */ local_irq_disable(); + if (image->preserve_context) { +#ifdef CONFIG_X86_IO_APIC + /* + * We need to put APICs in legacy mode so that we can + * get timer interrupts in second kernel. kexec/kdump + * paths already have calls to disable_IO_APIC() in + * one form or other. kexec jump path also need + * one. + */ + disable_IO_APIC(); +#endif + } + control_page = page_address(image->control_code_page) + PAGE_SIZE; - memcpy(control_page, relocate_kernel, PAGE_SIZE); + memcpy(control_page, relocate_kernel, KEXEC_CONTROL_CODE_MAX_SIZE); page_list[PA_CONTROL_PAGE] = virt_to_phys(control_page); + page_list[VA_CONTROL_PAGE] = (unsigned long)control_page; page_list[PA_TABLE_PAGE] = (unsigned long)__pa(page_address(image->control_code_page)); - /* The segment registers are funny things, they have both a + if (image->type == KEXEC_TYPE_DEFAULT) + page_list[PA_SWAP_PAGE] = (page_to_pfn(image->swap_page) + << PAGE_SHIFT); + + /* + * The segment registers are funny things, they have both a * visible and an invisible part. Whenever the visible part is * set to a specific selector, the invisible part is loaded * with from a table in memory. At no other time is the @@ -252,15 +319,25 @@ void machine_kexec(struct kimage *image) * segments, before I zap the gdt with an invalid value. */ load_segments(); - /* The gdt & idt are now invalid. + /* + * The gdt & idt are now invalid. * If you want to load them you must set up your own idt & gdt. */ - set_gdt(phys_to_virt(0),0); - set_idt(phys_to_virt(0),0); + set_gdt(phys_to_virt(0), 0); + set_idt(phys_to_virt(0), 0); /* now call it */ - relocate_kernel((unsigned long)image->head, (unsigned long)page_list, - image->start); + image->start = relocate_kernel((unsigned long)image->head, + (unsigned long)page_list, + image->start, + image->preserve_context); + +#ifdef CONFIG_KEXEC_JUMP + if (kexec_image->preserve_context) + restore_processor_state(); +#endif + + __ftrace_enabled_restore(save_ftrace_enabled); } void arch_crash_save_vmcoreinfo(void) diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c index 37cb1bd..e819240 100644 --- a/arch/x86/kernel/mpparse.c +++ b/arch/x86/kernel/mpparse.c @@ -558,6 +558,19 @@ static inline void __init construct_default_ISA_mptable(int mpc_default_type) static struct mpf_intel *mpf_found; +static unsigned long __init get_mpc_size(unsigned long physptr) +{ + struct mpc_table *mpc; + unsigned long size; + + mpc = early_ioremap(physptr, PAGE_SIZE); + size = mpc->length; + early_iounmap(mpc, PAGE_SIZE); + apic_printk(APIC_VERBOSE, " mpc: %lx-%lx\n", physptr, physptr + size); + + return size; +} + /* * Scan the memory blocks for an SMP configuration block. */ @@ -611,12 +624,16 @@ static void __init __get_smp_config(unsigned int early) construct_default_ISA_mptable(mpf->feature1); } else if (mpf->physptr) { + struct mpc_table *mpc; + unsigned long size; + size = get_mpc_size(mpf->physptr); + mpc = early_ioremap(mpf->physptr, size); /* * Read the physical hardware table. Anything here will * override the defaults. */ - if (!smp_read_mpc(phys_to_virt(mpf->physptr), early)) { + if (!smp_read_mpc(mpc, early)) { #ifdef CONFIG_X86_LOCAL_APIC smp_found_config = 0; #endif @@ -624,8 +641,10 @@ static void __init __get_smp_config(unsigned int early) "BIOS bug, MP table errors detected!...\n"); printk(KERN_ERR "... disabling SMP support. " "(tell your hw vendor)\n"); + early_iounmap(mpc, size); return; } + early_iounmap(mpc, size); if (early) return; @@ -697,10 +716,10 @@ static int __init smp_scan_config(unsigned long base, unsigned long length, if (!reserve) return 1; - reserve_bootmem_generic(virt_to_phys(mpf), PAGE_SIZE, + reserve_bootmem_generic(virt_to_phys(mpf), sizeof(*mpf), BOOTMEM_DEFAULT); if (mpf->physptr) { - unsigned long size = PAGE_SIZE; + unsigned long size = get_mpc_size(mpf->physptr); #ifdef CONFIG_X86_32 /* * We cannot access to MPC table to compute diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index 1cc18d4..2aef36d 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -216,6 +216,14 @@ static struct dmi_system_id __initdata reboot_dmi_table[] = { DMI_MATCH(DMI_PRODUCT_NAME, "HP Compaq"), }, }, + { /* Handle problems with rebooting on Dell XPS710 */ + .callback = set_bios_reboot, + .ident = "Dell XPS710", + .matches = { + DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), + DMI_MATCH(DMI_PRODUCT_NAME, "Dell XPS710"), + }, + }, { } }; diff --git a/arch/x86/kernel/relocate_kernel_32.S b/arch/x86/kernel/relocate_kernel_32.S index 2064d0a..4123553 100644 --- a/arch/x86/kernel/relocate_kernel_32.S +++ b/arch/x86/kernel/relocate_kernel_32.S @@ -17,7 +17,8 @@ #define PTR(x) (x << 2) -/* control_page + KEXEC_CONTROL_CODE_MAX_SIZE +/* + * control_page + KEXEC_CONTROL_CODE_MAX_SIZE * ~ control_page + PAGE_SIZE are used as data storage and stack for * jumping back */ @@ -76,8 +77,10 @@ relocate_kernel: movl %eax, CP_PA_SWAP_PAGE(%edi) movl %ebx, CP_PA_BACKUP_PAGES_MAP(%edi) - /* get physical address of control page now */ - /* this is impossible after page table switch */ + /* + * get physical address of control page now + * this is impossible after page table switch + */ movl PTR(PA_CONTROL_PAGE)(%ebp), %edi /* switch to new set of page tables */ @@ -97,7 +100,8 @@ identity_mapped: /* store the start address on the stack */ pushl %edx - /* Set cr0 to a known state: + /* + * Set cr0 to a known state: * - Paging disabled * - Alignment check disabled * - Write protect disabled @@ -113,7 +117,8 @@ identity_mapped: /* clear cr4 if applicable */ testl %ecx, %ecx jz 1f - /* Set cr4 to a known state: + /* + * Set cr4 to a known state: * Setting everything to zero seems safe. */ xorl %eax, %eax @@ -132,15 +137,18 @@ identity_mapped: call swap_pages addl $8, %esp - /* To be certain of avoiding problems with self-modifying code + /* + * To be certain of avoiding problems with self-modifying code * I need to execute a serializing instruction here. * So I flush the TLB, it's handy, and not processor dependent. */ xorl %eax, %eax movl %eax, %cr3 - /* set all of the registers to known values */ - /* leave %esp alone */ + /* + * set all of the registers to known values + * leave %esp alone + */ testl %esi, %esi jnz 1f diff --git a/arch/x86/kernel/relocate_kernel_64.S b/arch/x86/kernel/relocate_kernel_64.S index d32cfb2..4de8f5b 100644 --- a/arch/x86/kernel/relocate_kernel_64.S +++ b/arch/x86/kernel/relocate_kernel_64.S @@ -19,29 +19,77 @@ #define PTR(x) (x << 3) #define PAGE_ATTR (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY) +/* + * control_page + KEXEC_CONTROL_CODE_MAX_SIZE + * ~ control_page + PAGE_SIZE are used as data storage and stack for + * jumping back + */ +#define DATA(offset) (KEXEC_CONTROL_CODE_MAX_SIZE+(offset)) + +/* Minimal CPU state */ +#define RSP DATA(0x0) +#define CR0 DATA(0x8) +#define CR3 DATA(0x10) +#define CR4 DATA(0x18) + +/* other data */ +#define CP_PA_TABLE_PAGE DATA(0x20) +#define CP_PA_SWAP_PAGE DATA(0x28) +#define CP_PA_BACKUP_PAGES_MAP DATA(0x30) + .text .align PAGE_SIZE .code64 .globl relocate_kernel relocate_kernel: - /* %rdi indirection_page + /* + * %rdi indirection_page * %rsi page_list * %rdx start address + * %rcx preserve_context */ + /* Save the CPU context, used for jumping back */ + pushq %rbx + pushq %rbp + pushq %r12 + pushq %r13 + pushq %r14 + pushq %r15 + pushf + + movq PTR(VA_CONTROL_PAGE)(%rsi), %r11 + movq %rsp, RSP(%r11) + movq %cr0, %rax + movq %rax, CR0(%r11) + movq %cr3, %rax + movq %rax, CR3(%r11) + movq %cr4, %rax + movq %rax, CR4(%r11) + /* zero out flags, and disable interrupts */ pushq $0 popfq - /* get physical address of control page now */ - /* this is impossible after page table switch */ + /* + * get physical address of control page now + * this is impossible after page table switch + */ movq PTR(PA_CONTROL_PAGE)(%rsi), %r8 /* get physical address of page table now too */ - movq PTR(PA_TABLE_PAGE)(%rsi), %rcx + movq PTR(PA_TABLE_PAGE)(%rsi), %r9 + + /* get physical address of swap page now */ + movq PTR(PA_SWAP_PAGE)(%rsi), %r10 + + /* save some information for jumping back */ + movq %r9, CP_PA_TABLE_PAGE(%r11) + movq %r10, CP_PA_SWAP_PAGE(%r11) + movq %rdi, CP_PA_BACKUP_PAGES_MAP(%r11) /* Switch to the identity mapped page tables */ - movq %rcx, %cr3 + movq %r9, %cr3 /* setup a new stack at the end of the physical control page */ lea PAGE_SIZE(%r8), %rsp @@ -55,7 +103,8 @@ identity_mapped: /* store the start address on the stack */ pushq %rdx - /* Set cr0 to a known state: + /* + * Set cr0 to a known state: * - Paging enabled * - Alignment check disabled * - Write protect disabled @@ -68,7 +117,8 @@ identity_mapped: orl $(X86_CR0_PG | X86_CR0_PE), %eax movq %rax, %cr0 - /* Set cr4 to a known state: + /* + * Set cr4 to a known state: * - physical address extension enabled */ movq $X86_CR4_PAE, %rax @@ -78,9 +128,87 @@ identity_mapped: 1: /* Flush the TLB (needed?) */ - movq %rcx, %cr3 + movq %r9, %cr3 + + movq %rcx, %r11 + call swap_pages + + /* + * To be certain of avoiding problems with self-modifying code + * I need to execute a serializing instruction here. + * So I flush the TLB by reloading %cr3 here, it's handy, + * and not processor dependent. + */ + movq %cr3, %rax + movq %rax, %cr3 + + /* + * set all of the registers to known values + * leave %rsp alone + */ + + testq %r11, %r11 + jnz 1f + xorq %rax, %rax + xorq %rbx, %rbx + xorq %rcx, %rcx + xorq %rdx, %rdx + xorq %rsi, %rsi + xorq %rdi, %rdi + xorq %rbp, %rbp + xorq %r8, %r8 + xorq %r9, %r9 + xorq %r10, %r9 + xorq %r11, %r11 + xorq %r12, %r12 + xorq %r13, %r13 + xorq %r14, %r14 + xorq %r15, %r15 + + ret + +1: + popq %rdx + leaq PAGE_SIZE(%r10), %rsp + call *%rdx + + /* get the re-entry point of the peer system */ + movq 0(%rsp), %rbp + call 1f +1: + popq %r8 + subq $(1b - relocate_kernel), %r8 + movq CP_PA_SWAP_PAGE(%r8), %r10 + movq CP_PA_BACKUP_PAGES_MAP(%r8), %rdi + movq CP_PA_TABLE_PAGE(%r8), %rax + movq %rax, %cr3 + lea PAGE_SIZE(%r8), %rsp + call swap_pages + movq $virtual_mapped, %rax + pushq %rax + ret + +virtual_mapped: + movq RSP(%r8), %rsp + movq CR4(%r8), %rax + movq %rax, %cr4 + movq CR3(%r8), %rax + movq CR0(%r8), %r8 + movq %rax, %cr3 + movq %r8, %cr0 + movq %rbp, %rax + + popf + popq %r15 + popq %r14 + popq %r13 + popq %r12 + popq %rbp + popq %rbx + ret /* Do the copies */ +swap_pages: movq %rdi, %rcx /* Put the page_list in %rcx */ xorq %rdi, %rdi xorq %rsi, %rsi @@ -112,36 +240,27 @@ identity_mapped: movq %rcx, %rsi /* For ever source page do a copy */ andq $0xfffffffffffff000, %rsi + movq %rdi, %rdx + movq %rsi, %rax + + movq %r10, %rdi movq $512, %rcx rep ; movsq - jmp 0b -3: - - /* To be certain of avoiding problems with self-modifying code - * I need to execute a serializing instruction here. - * So I flush the TLB by reloading %cr3 here, it's handy, - * and not processor dependent. - */ - movq %cr3, %rax - movq %rax, %cr3 - /* set all of the registers to known values */ - /* leave %rsp alone */ + movq %rax, %rdi + movq %rdx, %rsi + movq $512, %rcx + rep ; movsq - xorq %rax, %rax - xorq %rbx, %rbx - xorq %rcx, %rcx - xorq %rdx, %rdx - xorq %rsi, %rsi - xorq %rdi, %rdi - xorq %rbp, %rbp - xorq %r8, %r8 - xorq %r9, %r9 - xorq %r10, %r9 - xorq %r11, %r11 - xorq %r12, %r12 - xorq %r13, %r13 - xorq %r14, %r14 - xorq %r15, %r15 + movq %rdx, %rdi + movq %r10, %rsi + movq $512, %rcx + rep ; movsq + lea PAGE_SIZE(%rax), %rsi + jmp 0b +3: ret + + .globl kexec_control_code_size +.set kexec_control_code_size, . - relocate_kernel diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 4c54bc0..f28c56e 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -202,7 +202,9 @@ struct ist_info ist_info; #endif #else -struct cpuinfo_x86 boot_cpu_data __read_mostly; +struct cpuinfo_x86 boot_cpu_data __read_mostly = { + .x86_phys_bits = MAX_PHYSMEM_BITS, +}; EXPORT_SYMBOL(boot_cpu_data); #endif @@ -770,6 +772,9 @@ void __init setup_arch(char **cmdline_p) finish_e820_parsing(); + if (efi_enabled) + efi_init(); + dmi_scan_machine(); dmi_check_system(bad_bios_dmi_table); @@ -789,8 +794,6 @@ void __init setup_arch(char **cmdline_p) insert_resource(&iomem_resource, &data_resource); insert_resource(&iomem_resource, &bss_resource); - if (efi_enabled) - efi_init(); #ifdef CONFIG_X86_32 if (ppro_with_ram_bug()) { diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index d992e6c..efa615f 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -7,6 +7,7 @@ #include <linux/crash_dump.h> #include <linux/smp.h> #include <linux/topology.h> +#include <linux/pfn.h> #include <asm/sections.h> #include <asm/processor.h> #include <asm/setup.h> @@ -41,6 +42,352 @@ unsigned long __per_cpu_offset[NR_CPUS] __read_mostly = { }; EXPORT_SYMBOL(__per_cpu_offset); +/* + * On x86_64 symbols referenced from code should be reachable using + * 32bit relocations. Reserve space for static percpu variables in + * modules so that they are always served from the first chunk which + * is located at the percpu segment base. On x86_32, anything can + * address anywhere. No need to reserve space in the first chunk. + */ +#ifdef CONFIG_X86_64 +#define PERCPU_FIRST_CHUNK_RESERVE PERCPU_MODULE_RESERVE +#else +#define PERCPU_FIRST_CHUNK_RESERVE 0 +#endif + +/** + * pcpu_need_numa - determine percpu allocation needs to consider NUMA + * + * If NUMA is not configured or there is only one NUMA node available, + * there is no reason to consider NUMA. This function determines + * whether percpu allocation should consider NUMA or not. + * + * RETURNS: + * true if NUMA should be considered; otherwise, false. + */ +static bool __init pcpu_need_numa(void) +{ +#ifdef CONFIG_NEED_MULTIPLE_NODES + pg_data_t *last = NULL; + unsigned int cpu; + + for_each_possible_cpu(cpu) { + int node = early_cpu_to_node(cpu); + + if (node_online(node) && NODE_DATA(node) && + last && last != NODE_DATA(node)) + return true; + + last = NODE_DATA(node); + } +#endif + return false; +} + +/** + * pcpu_alloc_bootmem - NUMA friendly alloc_bootmem wrapper for percpu + * @cpu: cpu to allocate for + * @size: size allocation in bytes + * @align: alignment + * + * Allocate @size bytes aligned at @align for cpu @cpu. This wrapper + * does the right thing for NUMA regardless of the current + * configuration. + * + * RETURNS: + * Pointer to the allocated area on success, NULL on failure. + */ +static void * __init pcpu_alloc_bootmem(unsigned int cpu, unsigned long size, + unsigned long align) +{ + const unsigned long goal = __pa(MAX_DMA_ADDRESS); +#ifdef CONFIG_NEED_MULTIPLE_NODES + int node = early_cpu_to_node(cpu); + void *ptr; + + if (!node_online(node) || !NODE_DATA(node)) { + ptr = __alloc_bootmem_nopanic(size, align, goal); + pr_info("cpu %d has no node %d or node-local memory\n", + cpu, node); + pr_debug("per cpu data for cpu%d %lu bytes at %016lx\n", + cpu, size, __pa(ptr)); + } else { + ptr = __alloc_bootmem_node_nopanic(NODE_DATA(node), + size, align, goal); + pr_debug("per cpu data for cpu%d %lu bytes on node%d at " + "%016lx\n", cpu, size, node, __pa(ptr)); + } + return ptr; +#else + return __alloc_bootmem_nopanic(size, align, goal); +#endif +} + +/* + * Remap allocator + * + * This allocator uses PMD page as unit. A PMD page is allocated for + * each cpu and each is remapped into vmalloc area using PMD mapping. + * As PMD page is quite large, only part of it is used for the first + * chunk. Unused part is returned to the bootmem allocator. + * + * So, the PMD pages are mapped twice - once to the physical mapping + * and to the vmalloc area for the first percpu chunk. The double + * mapping does add one more PMD TLB entry pressure but still is much + * better than only using 4k mappings while still being NUMA friendly. + */ +#ifdef CONFIG_NEED_MULTIPLE_NODES +static size_t pcpur_size __initdata; +static void **pcpur_ptrs __initdata; + +static struct page * __init pcpur_get_page(unsigned int cpu, int pageno) +{ + size_t off = (size_t)pageno << PAGE_SHIFT; + + if (off >= pcpur_size) + return NULL; + + return virt_to_page(pcpur_ptrs[cpu] + off); +} + +static ssize_t __init setup_pcpu_remap(size_t static_size) +{ + static struct vm_struct vm; + pg_data_t *last; + size_t ptrs_size, dyn_size; + unsigned int cpu; + ssize_t ret; + + /* + * If large page isn't supported, there's no benefit in doing + * this. Also, on non-NUMA, embedding is better. + */ + if (!cpu_has_pse || pcpu_need_numa()) + return -EINVAL; + + last = NULL; + for_each_possible_cpu(cpu) { + int node = early_cpu_to_node(cpu); + + if (node_online(node) && NODE_DATA(node) && + last && last != NODE_DATA(node)) + goto proceed; + + last = NODE_DATA(node); + } + return -EINVAL; + +proceed: + /* + * Currently supports only single page. Supporting multiple + * pages won't be too difficult if it ever becomes necessary. + */ + pcpur_size = PFN_ALIGN(static_size + PERCPU_MODULE_RESERVE + + PERCPU_DYNAMIC_RESERVE); + if (pcpur_size > PMD_SIZE) { + pr_warning("PERCPU: static data is larger than large page, " + "can't use large page\n"); + return -EINVAL; + } + dyn_size = pcpur_size - static_size - PERCPU_FIRST_CHUNK_RESERVE; + + /* allocate pointer array and alloc large pages */ + ptrs_size = PFN_ALIGN(num_possible_cpus() * sizeof(pcpur_ptrs[0])); + pcpur_ptrs = alloc_bootmem(ptrs_size); + + for_each_possible_cpu(cpu) { + pcpur_ptrs[cpu] = pcpu_alloc_bootmem(cpu, PMD_SIZE, PMD_SIZE); + if (!pcpur_ptrs[cpu]) + goto enomem; + + /* + * Only use pcpur_size bytes and give back the rest. + * + * Ingo: The 2MB up-rounding bootmem is needed to make + * sure the partial 2MB page is still fully RAM - it's + * not well-specified to have a PAT-incompatible area + * (unmapped RAM, device memory, etc.) in that hole. + */ + free_bootmem(__pa(pcpur_ptrs[cpu] + pcpur_size), + PMD_SIZE - pcpur_size); + + memcpy(pcpur_ptrs[cpu], __per_cpu_load, static_size); + } + + /* allocate address and map */ + vm.flags = VM_ALLOC; + vm.size = num_possible_cpus() * PMD_SIZE; + vm_area_register_early(&vm, PMD_SIZE); + + for_each_possible_cpu(cpu) { + pmd_t *pmd; + + pmd = populate_extra_pmd((unsigned long)vm.addr + + cpu * PMD_SIZE); + set_pmd(pmd, pfn_pmd(page_to_pfn(virt_to_page(pcpur_ptrs[cpu])), + PAGE_KERNEL_LARGE)); + } + + /* we're ready, commit */ + pr_info("PERCPU: Remapped at %p with large pages, static data " + "%zu bytes\n", vm.addr, static_size); + + ret = pcpu_setup_first_chunk(pcpur_get_page, static_size, + PERCPU_FIRST_CHUNK_RESERVE, + PMD_SIZE, dyn_size, vm.addr, NULL); + goto out_free_ar; + +enomem: + for_each_possible_cpu(cpu) + if (pcpur_ptrs[cpu]) + free_bootmem(__pa(pcpur_ptrs[cpu]), PMD_SIZE); + ret = -ENOMEM; +out_free_ar: + free_bootmem(__pa(pcpur_ptrs), ptrs_size); + return ret; +} +#else +static ssize_t __init setup_pcpu_remap(size_t static_size) +{ + return -EINVAL; +} +#endif + +/* + * Embedding allocator + * + * The first chunk is sized to just contain the static area plus + * module and dynamic reserves, and allocated as a contiguous area + * using bootmem allocator and used as-is without being mapped into + * vmalloc area. This enables the first chunk to piggy back on the + * linear physical PMD mapping and doesn't add any additional pressure + * to TLB. Note that if the needed size is smaller than the minimum + * unit size, the leftover is returned to the bootmem allocator. + */ +static void *pcpue_ptr __initdata; +static size_t pcpue_size __initdata; +static size_t pcpue_unit_size __initdata; + +static struct page * __init pcpue_get_page(unsigned int cpu, int pageno) +{ + size_t off = (size_t)pageno << PAGE_SHIFT; + + if (off >= pcpue_size) + return NULL; + + return virt_to_page(pcpue_ptr + cpu * pcpue_unit_size + off); +} + +static ssize_t __init setup_pcpu_embed(size_t static_size) +{ + unsigned int cpu; + size_t dyn_size; + + /* + * If large page isn't supported, there's no benefit in doing + * this. Also, embedding allocation doesn't play well with + * NUMA. + */ + if (!cpu_has_pse || pcpu_need_numa()) + return -EINVAL; + + /* allocate and copy */ + pcpue_size = PFN_ALIGN(static_size + PERCPU_MODULE_RESERVE + + PERCPU_DYNAMIC_RESERVE); + pcpue_unit_size = max_t(size_t, pcpue_size, PCPU_MIN_UNIT_SIZE); + dyn_size = pcpue_size - static_size - PERCPU_FIRST_CHUNK_RESERVE; + + pcpue_ptr = pcpu_alloc_bootmem(0, num_possible_cpus() * pcpue_unit_size, + PAGE_SIZE); + if (!pcpue_ptr) + return -ENOMEM; + + for_each_possible_cpu(cpu) { + void *ptr = pcpue_ptr + cpu * pcpue_unit_size; + + free_bootmem(__pa(ptr + pcpue_size), + pcpue_unit_size - pcpue_size); + memcpy(ptr, __per_cpu_load, static_size); + } + + /* we're ready, commit */ + pr_info("PERCPU: Embedded %zu pages at %p, static data %zu bytes\n", + pcpue_size >> PAGE_SHIFT, pcpue_ptr, static_size); + + return pcpu_setup_first_chunk(pcpue_get_page, static_size, + PERCPU_FIRST_CHUNK_RESERVE, + pcpue_unit_size, dyn_size, + pcpue_ptr, NULL); +} + +/* + * 4k page allocator + * + * This is the basic allocator. Static percpu area is allocated + * page-by-page and most of initialization is done by the generic + * setup function. + */ +static struct page **pcpu4k_pages __initdata; +static int pcpu4k_nr_static_pages __initdata; + +static struct page * __init pcpu4k_get_page(unsigned int cpu, int pageno) +{ + if (pageno < pcpu4k_nr_static_pages) + return pcpu4k_pages[cpu * pcpu4k_nr_static_pages + pageno]; + return NULL; +} + +static void __init pcpu4k_populate_pte(unsigned long addr) +{ + populate_extra_pte(addr); +} + +static ssize_t __init setup_pcpu_4k(size_t static_size) +{ + size_t pages_size; + unsigned int cpu; + int i, j; + ssize_t ret; + + pcpu4k_nr_static_pages = PFN_UP(static_size); + + /* unaligned allocations can't be freed, round up to page size */ + pages_size = PFN_ALIGN(pcpu4k_nr_static_pages * num_possible_cpus() + * sizeof(pcpu4k_pages[0])); + pcpu4k_pages = alloc_bootmem(pages_size); + + /* allocate and copy */ + j = 0; + for_each_possible_cpu(cpu) + for (i = 0; i < pcpu4k_nr_static_pages; i++) { + void *ptr; + + ptr = pcpu_alloc_bootmem(cpu, PAGE_SIZE, PAGE_SIZE); + if (!ptr) + goto enomem; + + memcpy(ptr, __per_cpu_load + i * PAGE_SIZE, PAGE_SIZE); + pcpu4k_pages[j++] = virt_to_page(ptr); + } + + /* we're ready, commit */ + pr_info("PERCPU: Allocated %d 4k pages, static data %zu bytes\n", + pcpu4k_nr_static_pages, static_size); + + ret = pcpu_setup_first_chunk(pcpu4k_get_page, static_size, + PERCPU_FIRST_CHUNK_RESERVE, -1, -1, NULL, + pcpu4k_populate_pte); + goto out_free_ar; + +enomem: + while (--j >= 0) + free_bootmem(__pa(page_address(pcpu4k_pages[j])), PAGE_SIZE); + ret = -ENOMEM; +out_free_ar: + free_bootmem(__pa(pcpu4k_pages), pages_size); + return ret; +} + static inline void setup_percpu_segment(int cpu) { #ifdef CONFIG_X86_32 @@ -61,38 +408,35 @@ static inline void setup_percpu_segment(int cpu) */ void __init setup_per_cpu_areas(void) { - ssize_t size; - char *ptr; - int cpu; - - /* Copy section for each CPU (we discard the original) */ - size = roundup(PERCPU_ENOUGH_ROOM, PAGE_SIZE); + size_t static_size = __per_cpu_end - __per_cpu_start; + unsigned int cpu; + unsigned long delta; + size_t pcpu_unit_size; + ssize_t ret; pr_info("NR_CPUS:%d nr_cpumask_bits:%d nr_cpu_ids:%d nr_node_ids:%d\n", NR_CPUS, nr_cpumask_bits, nr_cpu_ids, nr_node_ids); - pr_info("PERCPU: Allocating %zd bytes of per cpu data\n", size); + /* + * Allocate percpu area. If PSE is supported, try to make use + * of large page mappings. Please read comments on top of + * each allocator for details. + */ + ret = setup_pcpu_remap(static_size); + if (ret < 0) + ret = setup_pcpu_embed(static_size); + if (ret < 0) + ret = setup_pcpu_4k(static_size); + if (ret < 0) + panic("cannot allocate static percpu area (%zu bytes, err=%zd)", + static_size, ret); - for_each_possible_cpu(cpu) { -#ifndef CONFIG_NEED_MULTIPLE_NODES - ptr = alloc_bootmem_pages(size); -#else - int node = early_cpu_to_node(cpu); - if (!node_online(node) || !NODE_DATA(node)) { - ptr = alloc_bootmem_pages(size); - pr_info("cpu %d has no node %d or node-local memory\n", - cpu, node); - pr_debug("per cpu data for cpu%d at %016lx\n", - cpu, __pa(ptr)); - } else { - ptr = alloc_bootmem_pages_node(NODE_DATA(node), size); - pr_debug("per cpu data for cpu%d on node%d at %016lx\n", - cpu, node, __pa(ptr)); - } -#endif + pcpu_unit_size = ret; - memcpy(ptr, __per_cpu_load, __per_cpu_end - __per_cpu_start); - per_cpu_offset(cpu) = ptr - __per_cpu_start; + /* alrighty, percpu areas up and running */ + delta = (unsigned long)pcpu_base_addr - (unsigned long)__per_cpu_start; + for_each_possible_cpu(cpu) { + per_cpu_offset(cpu) = delta + cpu * pcpu_unit_size; per_cpu(this_cpu_off, cpu) = per_cpu_offset(cpu); per_cpu(cpu_number, cpu) = cpu; setup_percpu_segment(cpu); @@ -125,8 +469,6 @@ void __init setup_per_cpu_areas(void) */ if (cpu == boot_cpu_id) switch_to_new_gdt(cpu); - - DBG("PERCPU: cpu %4d %p\n", cpu, ptr); } /* indicate the early static arrays will soon be gone */ diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 249334f..ef7d101 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -114,10 +114,6 @@ EXPORT_PER_CPU_SYMBOL(cpu_info); atomic_t init_deasserted; - -/* Set if we find a B stepping CPU */ -static int __cpuinitdata smp_b_stepping; - #if defined(CONFIG_NUMA) && defined(CONFIG_X86_32) /* which logical CPUs are on which nodes */ @@ -271,8 +267,6 @@ static void __cpuinit smp_callin(void) cpumask_set_cpu(cpuid, cpu_callin_mask); } -static int __cpuinitdata unsafe_smp; - /* * Activate a secondary processor. */ @@ -340,76 +334,6 @@ notrace static void __cpuinit start_secondary(void *unused) cpu_idle(); } -static void __cpuinit smp_apply_quirks(struct cpuinfo_x86 *c) -{ - /* - * Mask B, Pentium, but not Pentium MMX - */ - if (c->x86_vendor == X86_VENDOR_INTEL && - c->x86 == 5 && - c->x86_mask >= 1 && c->x86_mask <= 4 && - c->x86_model <= 3) - /* - * Remember we have B step Pentia with bugs - */ - smp_b_stepping = 1; - - /* - * Certain Athlons might work (for various values of 'work') in SMP - * but they are not certified as MP capable. - */ - if ((c->x86_vendor == X86_VENDOR_AMD) && (c->x86 == 6)) { - - if (num_possible_cpus() == 1) - goto valid_k7; - - /* Athlon 660/661 is valid. */ - if ((c->x86_model == 6) && ((c->x86_mask == 0) || - (c->x86_mask == 1))) - goto valid_k7; - - /* Duron 670 is valid */ - if ((c->x86_model == 7) && (c->x86_mask == 0)) - goto valid_k7; - - /* - * Athlon 662, Duron 671, and Athlon >model 7 have capability - * bit. It's worth noting that the A5 stepping (662) of some - * Athlon XP's have the MP bit set. - * See http://www.heise.de/newsticker/data/jow-18.10.01-000 for - * more. - */ - if (((c->x86_model == 6) && (c->x86_mask >= 2)) || - ((c->x86_model == 7) && (c->x86_mask >= 1)) || - (c->x86_model > 7)) - if (cpu_has_mp) - goto valid_k7; - - /* If we get here, not a certified SMP capable AMD system. */ - unsafe_smp = 1; - } - -valid_k7: - ; -} - -static void __cpuinit smp_checks(void) -{ - if (smp_b_stepping) - printk(KERN_WARNING "WARNING: SMP operation may be unreliable" - "with B stepping processors.\n"); - - /* - * Don't taint if we are running SMP kernel on a single non-MP - * approved Athlon - */ - if (unsafe_smp && num_online_cpus() > 1) { - printk(KERN_INFO "WARNING: This combination of AMD" - "processors is not suitable for SMP.\n"); - add_taint(TAINT_UNSAFE_SMP); - } -} - /* * The bootstrap kernel entry code has set these up. Save them for * a given CPU @@ -423,7 +347,6 @@ void __cpuinit smp_store_cpu_info(int id) c->cpu_index = id; if (id != 0) identify_secondary_cpu(c); - smp_apply_quirks(c); } @@ -1193,7 +1116,6 @@ void __init native_smp_cpus_done(unsigned int max_cpus) pr_debug("Boot done.\n"); impress_friends(); - smp_checks(); #ifdef CONFIG_X86_IO_APIC setup_ioapic_dest(); #endif diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c index f04549a..d038b9c 100644 --- a/arch/x86/kernel/tlb_uv.c +++ b/arch/x86/kernel/tlb_uv.c @@ -314,8 +314,6 @@ const struct cpumask *uv_flush_tlb_others(const struct cpumask *cpumask, int locals = 0; struct bau_desc *bau_desc; - WARN_ON(!in_atomic()); - cpumask_andnot(flush_mask, cpumask, cpumask_of(cpu)); uv_cpu = uv_blade_processor_id(); diff --git a/arch/x86/kernel/uv_time.c b/arch/x86/kernel/uv_time.c new file mode 100644 index 0000000..2ffb6c5 --- /dev/null +++ b/arch/x86/kernel/uv_time.c @@ -0,0 +1,393 @@ +/* + * SGI RTC clock/timer routines. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + * + * Copyright (c) 2009 Silicon Graphics, Inc. All Rights Reserved. + * Copyright (c) Dimitri Sivanich + */ +#include <linux/clockchips.h> + +#include <asm/uv/uv_mmrs.h> +#include <asm/uv/uv_hub.h> +#include <asm/uv/bios.h> +#include <asm/uv/uv.h> +#include <asm/apic.h> +#include <asm/cpu.h> + +#define RTC_NAME "sgi_rtc" + +static cycle_t uv_read_rtc(void); +static int uv_rtc_next_event(unsigned long, struct clock_event_device *); +static void uv_rtc_timer_setup(enum clock_event_mode, + struct clock_event_device *); + +static struct clocksource clocksource_uv = { + .name = RTC_NAME, + .rating = 400, + .read = uv_read_rtc, + .mask = (cycle_t)UVH_RTC_REAL_TIME_CLOCK_MASK, + .shift = 10, + .flags = CLOCK_SOURCE_IS_CONTINUOUS, +}; + +static struct clock_event_device clock_event_device_uv = { + .name = RTC_NAME, + .features = CLOCK_EVT_FEAT_ONESHOT, + .shift = 20, + .rating = 400, + .irq = -1, + .set_next_event = uv_rtc_next_event, + .set_mode = uv_rtc_timer_setup, + .event_handler = NULL, +}; + +static DEFINE_PER_CPU(struct clock_event_device, cpu_ced); + +/* There is one of these allocated per node */ +struct uv_rtc_timer_head { + spinlock_t lock; + /* next cpu waiting for timer, local node relative: */ + int next_cpu; + /* number of cpus on this node: */ + int ncpus; + struct { + int lcpu; /* systemwide logical cpu number */ + u64 expires; /* next timer expiration for this cpu */ + } cpu[1]; +}; + +/* + * Access to uv_rtc_timer_head via blade id. + */ +static struct uv_rtc_timer_head **blade_info __read_mostly; + +static int uv_rtc_enable; + +/* + * Hardware interface routines + */ + +/* Send IPIs to another node */ +static void uv_rtc_send_IPI(int cpu) +{ + unsigned long apicid, val; + int pnode; + + apicid = cpu_physical_id(cpu); + pnode = uv_apicid_to_pnode(apicid); + val = (1UL << UVH_IPI_INT_SEND_SHFT) | + (apicid << UVH_IPI_INT_APIC_ID_SHFT) | + (GENERIC_INTERRUPT_VECTOR << UVH_IPI_INT_VECTOR_SHFT); + + uv_write_global_mmr64(pnode, UVH_IPI_INT, val); +} + +/* Check for an RTC interrupt pending */ +static int uv_intr_pending(int pnode) +{ + return uv_read_global_mmr64(pnode, UVH_EVENT_OCCURRED0) & + UVH_EVENT_OCCURRED0_RTC1_MASK; +} + +/* Setup interrupt and return non-zero if early expiration occurred. */ +static int uv_setup_intr(int cpu, u64 expires) +{ + u64 val; + int pnode = uv_cpu_to_pnode(cpu); + + uv_write_global_mmr64(pnode, UVH_RTC1_INT_CONFIG, + UVH_RTC1_INT_CONFIG_M_MASK); + uv_write_global_mmr64(pnode, UVH_INT_CMPB, -1L); + + uv_write_global_mmr64(pnode, UVH_EVENT_OCCURRED0_ALIAS, + UVH_EVENT_OCCURRED0_RTC1_MASK); + + val = (GENERIC_INTERRUPT_VECTOR << UVH_RTC1_INT_CONFIG_VECTOR_SHFT) | + ((u64)cpu_physical_id(cpu) << UVH_RTC1_INT_CONFIG_APIC_ID_SHFT); + + /* Set configuration */ + uv_write_global_mmr64(pnode, UVH_RTC1_INT_CONFIG, val); + /* Initialize comparator value */ + uv_write_global_mmr64(pnode, UVH_INT_CMPB, expires); + + return (expires < uv_read_rtc() && !uv_intr_pending(pnode)); +} + +/* + * Per-cpu timer tracking routines + */ + +static __init void uv_rtc_deallocate_timers(void) +{ + int bid; + + for_each_possible_blade(bid) { + kfree(blade_info[bid]); + } + kfree(blade_info); +} + +/* Allocate per-node list of cpu timer expiration times. */ +static __init int uv_rtc_allocate_timers(void) +{ + int cpu; + + blade_info = kmalloc(uv_possible_blades * sizeof(void *), GFP_KERNEL); + if (!blade_info) + return -ENOMEM; + memset(blade_info, 0, uv_possible_blades * sizeof(void *)); + + for_each_present_cpu(cpu) { + int nid = cpu_to_node(cpu); + int bid = uv_cpu_to_blade_id(cpu); + int bcpu = uv_cpu_hub_info(cpu)->blade_processor_id; + struct uv_rtc_timer_head *head = blade_info[bid]; + + if (!head) { + head = kmalloc_node(sizeof(struct uv_rtc_timer_head) + + (uv_blade_nr_possible_cpus(bid) * + 2 * sizeof(u64)), + GFP_KERNEL, nid); + if (!head) { + uv_rtc_deallocate_timers(); + return -ENOMEM; + } + spin_lock_init(&head->lock); + head->ncpus = uv_blade_nr_possible_cpus(bid); + head->next_cpu = -1; + blade_info[bid] = head; + } + + head->cpu[bcpu].lcpu = cpu; + head->cpu[bcpu].expires = ULLONG_MAX; + } + + return 0; +} + +/* Find and set the next expiring timer. */ +static void uv_rtc_find_next_timer(struct uv_rtc_timer_head *head, int pnode) +{ + u64 lowest = ULLONG_MAX; + int c, bcpu = -1; + + head->next_cpu = -1; + for (c = 0; c < head->ncpus; c++) { + u64 exp = head->cpu[c].expires; + if (exp < lowest) { + bcpu = c; + lowest = exp; + } + } + if (bcpu >= 0) { + head->next_cpu = bcpu; + c = head->cpu[bcpu].lcpu; + if (uv_setup_intr(c, lowest)) + /* If we didn't set it up in time, trigger */ + uv_rtc_send_IPI(c); + } else { + uv_write_global_mmr64(pnode, UVH_RTC1_INT_CONFIG, + UVH_RTC1_INT_CONFIG_M_MASK); + } +} + +/* + * Set expiration time for current cpu. + * + * Returns 1 if we missed the expiration time. + */ +static int uv_rtc_set_timer(int cpu, u64 expires) +{ + int pnode = uv_cpu_to_pnode(cpu); + int bid = uv_cpu_to_blade_id(cpu); + struct uv_rtc_timer_head *head = blade_info[bid]; + int bcpu = uv_cpu_hub_info(cpu)->blade_processor_id; + u64 *t = &head->cpu[bcpu].expires; + unsigned long flags; + int next_cpu; + + spin_lock_irqsave(&head->lock, flags); + + next_cpu = head->next_cpu; + *t = expires; + /* Will this one be next to go off? */ + if (next_cpu < 0 || bcpu == next_cpu || + expires < head->cpu[next_cpu].expires) { + head->next_cpu = bcpu; + if (uv_setup_intr(cpu, expires)) { + *t = ULLONG_MAX; + uv_rtc_find_next_timer(head, pnode); + spin_unlock_irqrestore(&head->lock, flags); + return 1; + } + } + + spin_unlock_irqrestore(&head->lock, flags); + return 0; +} + +/* + * Unset expiration time for current cpu. + * + * Returns 1 if this timer was pending. + */ +static int uv_rtc_unset_timer(int cpu) +{ + int pnode = uv_cpu_to_pnode(cpu); + int bid = uv_cpu_to_blade_id(cpu); + struct uv_rtc_timer_head *head = blade_info[bid]; + int bcpu = uv_cpu_hub_info(cpu)->blade_processor_id; + u64 *t = &head->cpu[bcpu].expires; + unsigned long flags; + int rc = 0; + + spin_lock_irqsave(&head->lock, flags); + + if (head->next_cpu == bcpu && uv_read_rtc() >= *t) + rc = 1; + + *t = ULLONG_MAX; + + /* Was the hardware setup for this timer? */ + if (head->next_cpu == bcpu) + uv_rtc_find_next_timer(head, pnode); + + spin_unlock_irqrestore(&head->lock, flags); + + return rc; +} + + +/* + * Kernel interface routines. + */ + +/* + * Read the RTC. + */ +static cycle_t uv_read_rtc(void) +{ + return (cycle_t)uv_read_local_mmr(UVH_RTC); +} + +/* + * Program the next event, relative to now + */ +static int uv_rtc_next_event(unsigned long delta, + struct clock_event_device *ced) +{ + int ced_cpu = cpumask_first(ced->cpumask); + + return uv_rtc_set_timer(ced_cpu, delta + uv_read_rtc()); +} + +/* + * Setup the RTC timer in oneshot mode + */ +static void uv_rtc_timer_setup(enum clock_event_mode mode, + struct clock_event_device *evt) +{ + int ced_cpu = cpumask_first(evt->cpumask); + + switch (mode) { + case CLOCK_EVT_MODE_PERIODIC: + case CLOCK_EVT_MODE_ONESHOT: + case CLOCK_EVT_MODE_RESUME: + /* Nothing to do here yet */ + break; + case CLOCK_EVT_MODE_UNUSED: + case CLOCK_EVT_MODE_SHUTDOWN: + uv_rtc_unset_timer(ced_cpu); + break; + } +} + +static void uv_rtc_interrupt(void) +{ + struct clock_event_device *ced = &__get_cpu_var(cpu_ced); + int cpu = smp_processor_id(); + + if (!ced || !ced->event_handler) + return; + + if (uv_rtc_unset_timer(cpu) != 1) + return; + + ced->event_handler(ced); +} + +static int __init uv_enable_rtc(char *str) +{ + uv_rtc_enable = 1; + + return 1; +} +__setup("uvrtc", uv_enable_rtc); + +static __init void uv_rtc_register_clockevents(struct work_struct *dummy) +{ + struct clock_event_device *ced = &__get_cpu_var(cpu_ced); + + *ced = clock_event_device_uv; + ced->cpumask = cpumask_of(smp_processor_id()); + clockevents_register_device(ced); +} + +static __init int uv_rtc_setup_clock(void) +{ + int rc; + + if (!uv_rtc_enable || !is_uv_system() || generic_interrupt_extension) + return -ENODEV; + + generic_interrupt_extension = uv_rtc_interrupt; + + clocksource_uv.mult = clocksource_hz2mult(sn_rtc_cycles_per_second, + clocksource_uv.shift); + + rc = clocksource_register(&clocksource_uv); + if (rc) { + generic_interrupt_extension = NULL; + return rc; + } + + /* Setup and register clockevents */ + rc = uv_rtc_allocate_timers(); + if (rc) { + clocksource_unregister(&clocksource_uv); + generic_interrupt_extension = NULL; + return rc; + } + + clock_event_device_uv.mult = div_sc(sn_rtc_cycles_per_second, + NSEC_PER_SEC, clock_event_device_uv.shift); + + clock_event_device_uv.min_delta_ns = NSEC_PER_SEC / + sn_rtc_cycles_per_second; + + clock_event_device_uv.max_delta_ns = clocksource_uv.mask * + (NSEC_PER_SEC / sn_rtc_cycles_per_second); + + rc = schedule_on_each_cpu(uv_rtc_register_clockevents); + if (rc) { + clocksource_unregister(&clocksource_uv); + generic_interrupt_extension = NULL; + uv_rtc_deallocate_timers(); + } + + return rc; +} +arch_initcall(uv_rtc_setup_clock); diff --git a/arch/x86/kernel/vmlinux_64.lds.S b/arch/x86/kernel/vmlinux_64.lds.S index fbfced6..5bf54e4 100644 --- a/arch/x86/kernel/vmlinux_64.lds.S +++ b/arch/x86/kernel/vmlinux_64.lds.S @@ -275,3 +275,10 @@ ASSERT((_end - _text <= KERNEL_IMAGE_SIZE), ASSERT((per_cpu__irq_stack_union == 0), "irq_stack_union is not at start of per-cpu area"); #endif + +#ifdef CONFIG_KEXEC +#include <asm/kexec.h> + +ASSERT(kexec_control_code_size <= KEXEC_CONTROL_CODE_MAX_SIZE, + "kexec control code size is too big") +#endif |
