Use %fs to access per-cpu variables in uni-processor kernels the same

as multi-processor kernels.  The old way made it difficult for kernel
modules to be portable between uni-processor and multi-processor
kernels.  It is no longer necessary to jump through hoops.

- always load %fs with the private segment on entry to the kernel
- change the type of the self referntial pointer from struct privatespace
  to struct globaldata
- make the globaldata symbol have value 0 in all cases, so the symbols
  in globals.s are always offsets, not aliases for fields in globaldata
- define the globaldata space used for uniprocessor kernels in C, rather
  than assembler
- change the assmebly language accessors to use %fs, add a macro
  PCPU_ADDR(member, reg), which loads the register reg with the address
  of the per-cpu variable member

12 files changed, 29 insertions, 50 deletions

diff --git a/sys/i386/i386/exception.s b/sys/i386/i386/exception.s
index 6ba916f..07f4a29 100644
--- a/sys/i386/i386/exception.s
+++ b/sys/i386/i386/exception.s
@@ -47,11 +47,6 @@
 
 #include "assym.s"
 
-#ifdef SMP
-#define	MOVL_KPSEL_EAX	movl	$KPSEL,%eax
-#else
-#define	MOVL_KPSEL_EAX
-#endif
 #define	SEL_RPL_MASK	0x0003
 
 	.text
@@ -172,7 +167,7 @@ IDTVEC(fpu)
 	mov	$KDSEL,%ax
 	mov	%ax,%ds
 	mov	%ax,%es
-	MOVL_KPSEL_EAX
+	mov	$KPSEL,%ax
 	mov	%ax,%fs
 	FAKE_MCOUNT(13*4(%esp))
 
@@ -213,7 +208,7 @@ alltraps_with_regs_pushed:
 	mov	$KDSEL,%ax
 	mov	%ax,%ds
 	mov	%ax,%es
-	MOVL_KPSEL_EAX
+	mov	$KPSEL,%ax
 	mov	%ax,%fs
 	FAKE_MCOUNT(13*4(%esp))
 calltrap:
@@ -253,7 +248,7 @@ IDTVEC(syscall)
 	mov	$KDSEL,%ax		/* switch to kernel segments */
 	mov	%ax,%ds
 	mov	%ax,%es
-	MOVL_KPSEL_EAX
+	mov	$KPSEL,%ax
 	mov	%ax,%fs
 	movl	TF_ERR(%esp),%eax	/* copy saved eflags to final spot */
 	movl	%eax,TF_EFLAGS(%esp)
@@ -287,7 +282,7 @@ IDTVEC(int0x80_syscall)
 	mov	$KDSEL,%ax		/* switch to kernel segments */
 	mov	%ax,%ds
 	mov	%ax,%es
-	MOVL_KPSEL_EAX
+	mov	$KPSEL,%ax
 	mov	%ax,%fs
 	movl	$2,TF_ERR(%esp)		/* sizeof "int 0x80" */
 	FAKE_MCOUNT(13*4(%esp))
@@ -308,8 +303,7 @@ ENTRY(fork_trampoline)
 #ifdef SMP
 	cmpl	$0,PCPU(SWITCHTIME)
 	jne	1f
-	movl	$GD_SWITCHTIME,%eax
-	addl	%fs:0,%eax
+	PCPU_ADDR(SWITCHTIME, %eax)
 	pushl	%eax
 	call	_microuptime
 	popl	%edx
diff --git a/sys/i386/i386/genassym.c b/sys/i386/i386/genassym.c
index 6f04558..233f2f7 100644
--- a/sys/i386/i386/genassym.c
+++ b/sys/i386/i386/genassym.c
@@ -169,6 +169,7 @@ ASSYM(BI_SYMTAB, offsetof(struct bootinfo, bi_symtab));
 ASSYM(BI_ESYMTAB, offsetof(struct bootinfo, bi_esymtab));
 ASSYM(BI_KERNEND, offsetof(struct bootinfo, bi_kernend));
 ASSYM(GD_SIZEOF, sizeof(struct globaldata));
+ASSYM(GD_PRVSPACE, offsetof(struct globaldata, gd_prvspace));
 ASSYM(GD_CURPROC, offsetof(struct globaldata, gd_curproc));
 ASSYM(GD_NPXPROC, offsetof(struct globaldata, gd_npxproc));
 ASSYM(GD_IDLEPROC, offsetof(struct globaldata, gd_idleproc));
@@ -223,10 +224,7 @@ ASSYM(LA_ICR_HI, offsetof(struct LAPIC, icr_hi));
 
 ASSYM(KCSEL, GSEL(GCODE_SEL, SEL_KPL));
 ASSYM(KDSEL, GSEL(GDATA_SEL, SEL_KPL));
-
-#ifdef SMP
 ASSYM(KPSEL, GSEL(GPRIV_SEL, SEL_KPL));
-#endif
 
 ASSYM(BC32SEL, GSEL(GBIOSCODE32_SEL, SEL_KPL));
 ASSYM(GPROC0_SEL, GPROC0_SEL);
diff --git a/sys/i386/i386/globals.s b/sys/i386/i386/globals.s
index aaa6f81..fcc4cdd 100644
--- a/sys/i386/i386/globals.s
+++ b/sys/i386/i386/globals.s
@@ -48,19 +48,9 @@
 	.set    gd_idlestack_top,PS_IDLESTACK_TOP
 #endif
 
-	/*
-	 * Define layout of the global data.  On SMP this lives in
-	 * the per-cpu address space, otherwise it's in the data segment.
-	 */
 	.globl	globaldata
-#ifndef SMP
-	.data
-	ALIGN_DATA
-globaldata:
-	.space	GD_SIZEOF		/* in data segment */
-#else
 	.set	globaldata,0
-#endif
+
 	.globl	gd_curproc, gd_curpcb, gd_npxproc, gd_idleproc
 	.globl	gd_astpending, gd_common_tss, gd_switchtime, gd_switchticks
 	.globl	gd_intr_nesting_level
diff --git a/sys/i386/i386/machdep.c b/sys/i386/i386/machdep.c
index 7b537ff..c1c6f29 100644
--- a/sys/i386/i386/machdep.c
+++ b/sys/i386/i386/machdep.c
@@ -244,6 +244,9 @@ static vm_offset_t buffer_sva, buffer_eva;
 vm_offset_t clean_sva, clean_eva;
 static vm_offset_t pager_sva, pager_eva;
 static struct trapframe proc0_tf;
+#ifndef SMP
+static struct globaldata __globaldata;
+#endif
 
 struct cpuhead cpuhead;
 
@@ -1850,10 +1853,14 @@ init386(first)
 	gdt_segs[GPRIV_SEL].ssd_base = (int) &SMP_prvspace[0];
 	gdt_segs[GPROC0_SEL].ssd_base =
 		(int) &SMP_prvspace[0].globaldata.gd_common_tss;
-	SMP_prvspace[0].globaldata.gd_prvspace = &SMP_prvspace[0];
+	SMP_prvspace[0].globaldata.gd_prvspace = &SMP_prvspace[0].globaldata;
 #else
-	gdt_segs[GPRIV_SEL].ssd_limit = i386_btop(0) - 1;
-	gdt_segs[GPROC0_SEL].ssd_base = (int) &common_tss;
+	gdt_segs[GPRIV_SEL].ssd_limit =
+		i386_btop(sizeof(struct globaldata)) - 1;
+	gdt_segs[GPRIV_SEL].ssd_base = (int) &__globaldata;
+	gdt_segs[GPROC0_SEL].ssd_base =
+		(int) &__globaldata.gd_common_tss;
+	__globaldata.gd_prvspace = &__globaldata;
 #endif
 
 	for (x = 0; x < NGDT; x++) {
diff --git a/sys/i386/i386/support.s b/sys/i386/i386/support.s
index e240de4..3f0f51f 100644
--- a/sys/i386/i386/support.s
+++ b/sys/i386/i386/support.s
@@ -1516,9 +1516,7 @@ ENTRY(lgdt)
 	mov	%ax,%es
 	mov	%ax,%gs
 	mov	%ax,%ss
-#ifdef SMP
 	movl	$KPSEL,%eax
-#endif
 	mov	%ax,%fs
 
 	/* reload code selector by turning return into intersegmental return */
diff --git a/sys/i386/i386/swtch.s b/sys/i386/i386/swtch.s
index 6d2e30d..8d5cade 100644
--- a/sys/i386/i386/swtch.s
+++ b/sys/i386/i386/swtch.s
@@ -224,12 +224,7 @@ sw1b:
 
 	btrl	%esi, _private_tss
 	jae	3f
-#ifdef SMP
-	movl	$GD_COMMON_TSSD, %edi
-	addl	%fs:0, %edi
-#else
-	movl	$PCPU(COMMON_TSSD), %edi
-#endif
+	PCPU_ADDR(COMMON_TSSD, %edi)
 2:
 	/* move correct tss descriptor into GDT slot, then reload tr */
 	movl	PCPU(TSS_GDT), %ebx		/* entry in GDT */
diff --git a/sys/i386/include/asmacros.h b/sys/i386/include/asmacros.h
index c2d89b8..0931404 100644
--- a/sys/i386/include/asmacros.h
+++ b/sys/i386/include/asmacros.h
@@ -69,10 +69,10 @@
 #define NON_GPROF_ENTRY(name)	GEN_ENTRY(name)
 #define NON_GPROF_RET		.byte 0xc3	/* opcode for `ret' */
 
-#ifdef SMP
+#ifdef LOCORE
 #define	PCPU(member)	%fs:GD_ ## member
-#else
-#define	PCPU(member)	CNAME(globaldata) + GD_ ## member
+#define	PCPU_ADDR(member, reg)	movl %fs:GD_PRVSPACE,reg; \
+			addl $GD_ ## member,reg
 #endif
 
 #ifdef GPROF
diff --git a/sys/i386/include/globaldata.h b/sys/i386/include/globaldata.h
index 2788499..90c3bd7 100644
--- a/sys/i386/include/globaldata.h
+++ b/sys/i386/include/globaldata.h
@@ -53,7 +53,7 @@
  * other processors"
  */
 struct globaldata {
-	struct privatespace *gd_prvspace;	/* self-reference */
+	struct globaldata *gd_prvspace;		/* self-reference */
 	struct proc	*gd_curproc;
 	struct proc	*gd_npxproc;
 	struct pcb	*gd_curpcb;
@@ -92,8 +92,6 @@ struct globaldata {
 #endif
 };
 
-extern struct globaldata globaldata;
-
 SLIST_HEAD(cpuhead, globaldata);
 extern struct cpuhead cpuhead;
 
diff --git a/sys/i386/include/globals.h b/sys/i386/include/globals.h
index a20d83b..386810c 100644
--- a/sys/i386/include/globals.h
+++ b/sys/i386/include/globals.h
@@ -82,7 +82,8 @@ _global_globaldata(void)
 	return (val);
 }
 
-#if defined(SMP) || defined(KLD_MODULE)
+/* #if defined(SMP) || defined(KLD_MODULE) */
+#if 1
 /*
  * The following set of macros works for UP kernel as well, but for maximum
  * performance we allow the global variables to be accessed directly. On the
@@ -179,11 +180,7 @@ GLOBAL_FUNC(prv_PADDR1)
 
 GLOBAL_FUNC(witness_spin_check)
 
-#ifdef SMP
 #define	GLOBALDATA		GLOBAL_RVALUE(globaldata, struct globaldata *)
-#else
-#define	GLOBALDATA		(&globaldata)
-#endif
 
 #define	CURTHD			curproc
 #define	CURPROC			curproc
diff --git a/sys/i386/include/pcpu.h b/sys/i386/include/pcpu.h
index 2788499..90c3bd7 100644
--- a/sys/i386/include/pcpu.h
+++ b/sys/i386/include/pcpu.h
@@ -53,7 +53,7 @@
  * other processors"
  */
 struct globaldata {
-	struct privatespace *gd_prvspace;	/* self-reference */
+	struct globaldata *gd_prvspace;		/* self-reference */
 	struct proc	*gd_curproc;
 	struct proc	*gd_npxproc;
 	struct pcb	*gd_curpcb;
@@ -92,8 +92,6 @@ struct globaldata {
 #endif
 };
 
-extern struct globaldata globaldata;
-
 SLIST_HEAD(cpuhead, globaldata);
 extern struct cpuhead cpuhead;
 
diff --git a/sys/i386/isa/atpic_vector.s b/sys/i386/isa/atpic_vector.s
index e28c375..635d2ad 100644
--- a/sys/i386/isa/atpic_vector.s
+++ b/sys/i386/isa/atpic_vector.s
@@ -58,6 +58,7 @@ IDTVEC(vec_name) ; \
 	mov	$KDSEL,%ax ; \
 	mov	%ax,%ds ; \
 	mov	%ax,%es ; \
+	mov	$KPSEL,%ax ; \
 	mov	%ax,%fs ; \
 	FAKE_MCOUNT((12+ACTUALLY_PUSHED)*4(%esp)) ; \
 	incb	PCPU(INTR_NESTING_LEVEL) ; \
@@ -93,6 +94,7 @@ IDTVEC(vec_name) ; \
 	mov	$KDSEL,%ax ;	/* load kernel ds, es and fs */ \
 	mov	%ax,%ds ; \
 	mov	%ax,%es ; \
+	mov	$KPSEL,%ax ; \
 	mov	%ax,%fs ; \
 	maybe_extra_ipending ; \
 	movb	_imen + IRQ_BYTE(irq_num),%al ; \
diff --git a/sys/i386/isa/icu_vector.s b/sys/i386/isa/icu_vector.s
index e28c375..635d2ad 100644
--- a/sys/i386/isa/icu_vector.s
+++ b/sys/i386/isa/icu_vector.s
@@ -58,6 +58,7 @@ IDTVEC(vec_name) ; \
 	mov	$KDSEL,%ax ; \
 	mov	%ax,%ds ; \
 	mov	%ax,%es ; \
+	mov	$KPSEL,%ax ; \
 	mov	%ax,%fs ; \
 	FAKE_MCOUNT((12+ACTUALLY_PUSHED)*4(%esp)) ; \
 	incb	PCPU(INTR_NESTING_LEVEL) ; \
@@ -93,6 +94,7 @@ IDTVEC(vec_name) ; \
 	mov	$KDSEL,%ax ;	/* load kernel ds, es and fs */ \
 	mov	%ax,%ds ; \
 	mov	%ax,%es ; \
+	mov	$KPSEL,%ax ; \
 	mov	%ax,%fs ; \
 	maybe_extra_ipending ; \
 	movb	_imen + IRQ_BYTE(irq_num),%al ; \