Cleanup: rationalize paravirt wrappers

PATCH In review Rusty Russell 3 participants 8 messages Mar 22, 2007 – Mar 23, 2007 PARAVIRT_OPS interface

Share & tools

Rusty Russell

author

2007-03-22 02:42 UTC #0 lore raw

PATCH 1/1 Cleanup: rationalize paravirt wrappers

paravirt.c used to implement native versions of all low-level functions. Far cleaner is to have the native versions exposed in the headers and as inline native_XXX, and if !CONFIG_PARAVIRT, then simply #define XXX native_XXX.

There are four nice side effects:

1) write_dt_entry() now takes the correct "struct Xgt_desc_struct *" not "void *".

2) load_TLS is reintroduced to the for loop, not manually unrolled with a #error in case the bounds ever change.

3) Macros become inlines, with type checking.

4) Access to the native versions is trivial for KVM, lguest, Xen and others who might want it.

Signed-off-by Jeremy Fitzhardinge <[email protected]>

Signed-off-by Rusty Russell <[email protected]>

---
 arch/i386/kernel/paravirt.c  |  289 ------------------------------------------
 include/asm-i386/desc.h      |   90 ++++++++-----
 include/asm-i386/io.h        |   17 +-
 include/asm-i386/irqflags.h  |   61 ++++++--
 include/asm-i386/msr.h       |  156 ++++++++++++++--------
 include/asm-i386/paravirt.h  |   16 +-
 include/asm-i386/processor.h |   96 ++++++++++---
 include/asm-i386/system.h    |  139 ++++++++++++--------
 8 files changed, 383 insertions(+), 481 deletions(-)

===================================================================

arch/i386/kernel/paravirt.c +3 ~~-286~~

@@ -93,292 +93,9 @@ static unsigned native_patch(u8 type, u1
 	return insn_len;
 }
 
-static unsigned long native_get_debugreg(int regno)
-{
-	unsigned long val = 0; 	/* Damn you, gcc! */
-
-	switch (regno) {
-	case 0:
-		asm("movl %%db0, %0" :"=r" (val)); break;
-	case 1:
-		asm("movl %%db1, %0" :"=r" (val)); break;
-	case 2:
-		asm("movl %%db2, %0" :"=r" (val)); break;
-	case 3:
-		asm("movl %%db3, %0" :"=r" (val)); break;
-	case 6:
-		asm("movl %%db6, %0" :"=r" (val)); break;
-	case 7:
-		asm("movl %%db7, %0" :"=r" (val)); break;
-	default:
-		BUG();
-	}
-	return val;
-}
-
-static void native_set_debugreg(int regno, unsigned long value)
-{
-	switch (regno) {
-	case 0:
-		asm("movl %0,%%db0"	: /* no output */ :"r" (value));
-		break;
-	case 1:
-		asm("movl %0,%%db1"	: /* no output */ :"r" (value));
-		break;
-	case 2:
-		asm("movl %0,%%db2"	: /* no output */ :"r" (value));
-		break;
-	case 3:
-		asm("movl %0,%%db3"	: /* no output */ :"r" (value));
-		break;
-	case 6:
-		asm("movl %0,%%db6"	: /* no output */ :"r" (value));
-		break;
-	case 7:
-		asm("movl %0,%%db7"	: /* no output */ :"r" (value));
-		break;
-	default:
-		BUG();
-	}
-}
-
 void init_IRQ(void)
 {
 	paravirt_ops.init_IRQ();
-}
-
-static void native_clts(void)
-{
-	asm volatile ("clts");
-}
-
-static unsigned long native_read_cr0(void)
-{
-	unsigned long val;
-	asm volatile("movl %%cr0,%0\n\t" :"=r" (val));
-	return val;
-}
-
-static void native_write_cr0(unsigned long val)
-{
-	asm volatile("movl %0,%%cr0": :"r" (val));
-}
-
-static unsigned long native_read_cr2(void)
-{
-	unsigned long val;
-	asm volatile("movl %%cr2,%0\n\t" :"=r" (val));
-	return val;
-}
-
-static void native_write_cr2(unsigned long val)
-{
-	asm volatile("movl %0,%%cr2": :"r" (val));
-}
-
-static unsigned long native_read_cr3(void)
-{
-	unsigned long val;
-	asm volatile("movl %%cr3,%0\n\t" :"=r" (val));
-	return val;
-}
-
-static void native_write_cr3(unsigned long val)
-{
-	asm volatile("movl %0,%%cr3": :"r" (val));
-}
-
-static unsigned long native_read_cr4(void)
-{
-	unsigned long val;
-	asm volatile("movl %%cr4,%0\n\t" :"=r" (val));
-	return val;
-}
-
-static unsigned long native_read_cr4_safe(void)
-{
-	unsigned long val;
-	/* This could fault if %cr4 does not exist */
-	asm("1: movl %%cr4, %0		\n"
-		"2:				\n"
-		".section __ex_table,\"a\"	\n"
-		".long 1b,2b			\n"
-		".previous			\n"
-		: "=r" (val): "0" (0));
-	return val;
-}
-
-static void native_write_cr4(unsigned long val)
-{
-	asm volatile("movl %0,%%cr4": :"r" (val));
-}
-
-static unsigned long native_save_fl(void)
-{
-	unsigned long f;
-	asm volatile("pushfl ; popl %0":"=g" (f): /* no input */);
-	return f;
-}
-
-static void native_restore_fl(unsigned long f)
-{
-	asm volatile("pushl %0 ; popfl": /* no output */
-			     :"g" (f)
-			     :"memory", "cc");
-}
-
-static void native_irq_disable(void)
-{
-	asm volatile("cli": : :"memory");
-}
-
-static void native_irq_enable(void)
-{
-	asm volatile("sti": : :"memory");
-}
-
-static void native_safe_halt(void)
-{
-	asm volatile("sti; hlt": : :"memory");
-}
-
-static void native_halt(void)
-{
-	asm volatile("hlt": : :"memory");
-}
-
-static void native_wbinvd(void)
-{
-	asm volatile("wbinvd": : :"memory");
-}
-
-static unsigned long long native_read_msr(unsigned int msr, int *err)
-{
-	unsigned long long val;
-
-	asm volatile("2: rdmsr ; xorl %0,%0\n"
-		     "1:\n\t"
-		     ".section .fixup,\"ax\"\n\t"
-		     "3:  movl %3,%0 ; jmp 1b\n\t"
-		     ".previous\n\t"
- 		     ".section __ex_table,\"a\"\n"
-		     "   .align 4\n\t"
-		     "   .long 	2b,3b\n\t"
-		     ".previous"
-		     : "=r" (*err), "=A" (val)
-		     : "c" (msr), "i" (-EFAULT));
-
-	return val;
-}
-
-static int native_write_msr(unsigned int msr, unsigned long long val)
-{
-	int err;
-	asm volatile("2: wrmsr ; xorl %0,%0\n"
-		     "1:\n\t"
-		     ".section .fixup,\"ax\"\n\t"
-		     "3:  movl %4,%0 ; jmp 1b\n\t"
-		     ".previous\n\t"
- 		     ".section __ex_table,\"a\"\n"
-		     "   .align 4\n\t"
-		     "   .long 	2b,3b\n\t"
-		     ".previous"
-		     : "=a" (err)
-		     : "c" (msr), "0" ((u32)val), "d" ((u32)(val>>32)),
-		       "i" (-EFAULT));
-	return err;
-}
-
-static unsigned long long native_read_tsc(void)
-{
-	unsigned long long val;
-	asm volatile("rdtsc" : "=A" (val));
-	return val;
-}
-
-static unsigned long long native_read_pmc(void)
-{
-	unsigned long long val;
-	asm volatile("rdpmc" : "=A" (val));
-	return val;
-}
-
-static void native_load_tr_desc(void)
-{
-	asm volatile("ltr %w0"::"q" (GDT_ENTRY_TSS*8));
-}
-
-static void native_load_gdt(const struct Xgt_desc_struct *dtr)
-{
-	asm volatile("lgdt %0"::"m" (*dtr));
-}
-
-static void native_load_idt(const struct Xgt_desc_struct *dtr)
-{
-	asm volatile("lidt %0"::"m" (*dtr));
-}
-
-static void native_store_gdt(struct Xgt_desc_struct *dtr)
-{
-	asm ("sgdt %0":"=m" (*dtr));
-}
-
-static void native_store_idt(struct Xgt_desc_struct *dtr)
-{
-	asm ("sidt %0":"=m" (*dtr));
-}
-
-static unsigned long native_store_tr(void)
-{
-	unsigned long tr;
-	asm ("str %0":"=r" (tr));
-	return tr;
-}
-
-static void native_load_tls(struct thread_struct *t, unsigned int cpu)
-{
-#define C(i) get_cpu_gdt_table(cpu)[GDT_ENTRY_TLS_MIN + i] = t->tls_array[i]
-	C(0); C(1); C(2);
-#undef C
-}
-
-static inline void native_write_dt_entry(void *dt, int entry, u32 entry_low, u32 entry_high)
-{
-	u32 *lp = (u32 *)((char *)dt + entry*8);
-	lp[0] = entry_low;
-	lp[1] = entry_high;
-}
-
-static void native_write_ldt_entry(void *dt, int entrynum, u32 low, u32 high)
-{
-	native_write_dt_entry(dt, entrynum, low, high);
-}
-
-static void native_write_gdt_entry(void *dt, int entrynum, u32 low, u32 high)
-{
-	native_write_dt_entry(dt, entrynum, low, high);
-}
-
-static void native_write_idt_entry(void *dt, int entrynum, u32 low, u32 high)
-{
-	native_write_dt_entry(dt, entrynum, low, high);
-}
-
-static void native_load_esp0(struct tss_struct *tss,
-				      struct thread_struct *thread)
-{
-	tss->esp0 = thread->esp0;
-
-	/* This can only happen when SEP is enabled, no need to test "SEP"arately */
-	if (unlikely(tss->ss1 != thread->sysenter_cs)) {
-		tss->ss1 = thread->sysenter_cs;
-		wrmsr(MSR_IA32_SYSENTER_CS, thread->sysenter_cs, 0);
-	}
-}
-
-static void native_io_delay(void)
-{
-	asm volatile("outb %al,$0x80");
 }
 
 static void native_flush_tlb(void)

@@ -531,9 +248,9 @@ struct paravirt_ops paravirt_ops = {
 	.store_idt = native_store_idt,
 	.store_tr = native_store_tr,
 	.load_tls = native_load_tls,
-	.write_ldt_entry = native_write_ldt_entry,
-	.write_gdt_entry = native_write_gdt_entry,
-	.write_idt_entry = native_write_idt_entry,
+	.write_ldt_entry = write_dt_entry,
+	.write_gdt_entry = write_dt_entry,
+	.write_idt_entry = write_dt_entry,
 	.load_esp0 = native_load_esp0,
 
 	.set_iopl_mask = native_set_iopl_mask,

===================================================================

include/asm-i386/desc.h +60 ~~-30~~

111 lines not on this page · raw message

===================================================================

include/asm-i386/io.h +10 -7

@@ -250,19 +250,22 @@ static inline void flush_write_buffers(v
 
 #endif /* __KERNEL__ */
 
+static inline void native_io_delay(void)
+{
+	asm volatile("outb %%al,$0x80" : : : "memory");
+}
+
 #if defined(CONFIG_PARAVIRT)
 #include <asm/paravirt.h>
 #else
 
-#define __SLOW_DOWN_IO "outb %%al,$0x80;"
-
 static inline void slow_down_io(void) {
-	__asm__ __volatile__(
-		__SLOW_DOWN_IO
+	native_io_delay();
 #ifdef REALLY_SLOW_IO
-		__SLOW_DOWN_IO __SLOW_DOWN_IO __SLOW_DOWN_IO
-#endif
-		: : );
+	native_io_delay();
+	native_io_delay();
+	native_io_delay();
+#endif
 }
 
 #endif

===================================================================

include/asm-i386/irqflags.h +42 ~~-19~~

97 lines not on this page · raw message

===================================================================

include/asm-i386/msr.h +101 ~~-55~~

178 lines not on this page · raw message

===================================================================

include/asm-i386/paravirt.h +8 -8

@@ -29,6 +29,7 @@ struct Xgt_desc_struct;
 struct Xgt_desc_struct;
 struct tss_struct;
 struct mm_struct;
+struct desc_struct;
 struct paravirt_ops
 {
 	unsigned int kernel_rpl;

@@ -105,14 +106,13 @@ struct paravirt_ops
 	void (*set_ldt)(const void *desc, unsigned entries);
 	unsigned long (*store_tr)(void);
 	void (*load_tls)(struct thread_struct *t, unsigned int cpu);
-	void (*write_ldt_entry)(void *dt, int entrynum,
-					 u32 low, u32 high);
-	void (*write_gdt_entry)(void *dt, int entrynum,
-					 u32 low, u32 high);
-	void (*write_idt_entry)(void *dt, int entrynum,
-					 u32 low, u32 high);
-	void (*load_esp0)(struct tss_struct *tss,
-				   struct thread_struct *thread);
+	void (*write_ldt_entry)(struct desc_struct *,
+				int entrynum, u32 low, u32 high);
+	void (*write_gdt_entry)(struct desc_struct *,
+				int entrynum, u32 low, u32 high);
+	void (*write_idt_entry)(struct desc_struct *,
+				int entrynum, u32 low, u32 high);
+	void (*load_esp0)(struct tss_struct *tss, struct thread_struct *t);
 
 	void (*set_iopl_mask)(unsigned mask);

===================================================================

include/asm-i386/processor.h +74 ~~-22~~

124 lines not on this page · raw message

===================================================================

include/asm-i386/system.h +85 ~~-54~~

150 lines not on this page · raw message

Avi Kivity

2007-03-22 07:30 UTC #1 lore raw

Quoted 20 lines · Rusty Russell

include/asm-i386/msr.h +18 -0

+static inline unsigned long long native_read_msr(unsigned int msr, int *err)
+{
+	unsigned long long val;
+
+	asm volatile("2: rdmsr ; xorl %0,%0\n"
+		     "1:\n\t"
+		     ".section .fixup,\"ax\"\n\t"
+		     "3:  movl %3,%0 ; jmp 1b\n\t"
+		     ".previous\n\t"
+ 		     ".section __ex_table,\"a\"\n"
+		     "   .align 4\n\t"
+		     "   .long 	2b,3b\n\t"
+		     ".previous"
+		     : "=r" (*err), "=A" (val)
+		     : "c" (msr), "i" (-EFAULT));
+
+	return val;
+}

include/asm-i386/msr.h

Quoted 10 lines

include/asm-i386/msr.h +8 -0

 
 
+#define rdmsr(msr,val1,val2)						\
+	do {								\
+		int __err;						\
+		unsigned long long __val = native_read_msr(msr, &__err);	\
+		val1 = __val;						\
+		val2 = __val >> 32;					\
+	} while(0)
+

You're silently changing the behavior (as well as the prototype) here, rdmsr() used to fail loudly, now it fails silently, with no way for the caller to check.

Signature · 1 line

Do not meddle in the internals of kernels, for they are subtle and quick to panic.

Rusty Russell

author

2007-03-22 08:01 UTC #2 lore raw

Quoted 22 lines · Avi Kivity

Quoted 19 lines · Rusty Russell

include/asm-i386/msr.h +18 -0

+static inline unsigned long long native_read_msr(unsigned int msr, int *err)
+{
+	unsigned long long val;
+
+	asm volatile("2: rdmsr ; xorl %0,%0\n"
+		     "1:\n\t"
+		     ".section .fixup,\"ax\"\n\t"
+		     "3:  movl %3,%0 ; jmp 1b\n\t"
+		     ".previous\n\t"
+ 		     ".section __ex_table,\"a\"\n"
+		     "   .align 4\n\t"
+		     "   .long 	2b,3b\n\t"
+		     ".previous"
+		     : "=r" (*err), "=A" (val)
+		     : "c" (msr), "i" (-EFAULT));
+
+	return val;
+}

Quoted 13 lines

include/asm-i386/msr.h line 74
Quoted 9 lines
include/asm-i386/msr.h +8 -0
+#define rdmsr(msr,val1,val2)						\
+	do {								\
+		int __err;						\
+		unsigned long long __val = native_read_msr(msr, &__err);	\
+		val1 = __val;						\
+		val2 = __val >> 32;					\
+	} while(0)
+
You're silently changing the behavior (as well as the prototype) here, rdmsr() used to fail loudly, now it fails silently, with no way for the caller to check.

Hi Avi!

Not sure what you're saying about the prototype: the old macro was:

-#define rdmsr(msr,val1,val2) \
- __asm__ __volatile__("rdmsr" \
- : "=a" (val1), "=d" (val2) \
- : "c" (msr))
-

Which doesn't seem to be a great difference to me.

The behaviour change (don't oops when an invalid rdmsr is used) was there with CONFIG_PARAVIRT=y, the cleanup just made !CONFIG_PARAVIRT the same. Is it important?

Thanks,
Rusty.

Jeremy Fitzhardinge

2007-03-22 08:09 UTC #3 lore raw

Quoted 4 lines · Rusty Russell

The behaviour change (don't oops when an invalid rdmsr is used) was there with CONFIG_PARAVIRT=y, the cleanup just made !CONFIG_PARAVIRT the same. Is it important?

We could always fling a BUG_ON in the non-safe versions, which would probably be more informative than a GPF or whatever a bad MSR gives you.

Avi Kivity

2007-03-22 08:10 UTC #4 lore raw

In reply to Rusty Russell

Quoted 24 lines · Rusty Russell

Quoted 13 lines
include/asm-i386/msr.h line 74
Quoted 11 lines
include/asm-i386/msr.h +8 -0
+#define rdmsr(msr,val1,val2)						\
+	do {								\
+		int __err;						\
+		unsigned long long __val = native_read_msr(msr, &__err);	\
+		val1 = __val;						\
+		val2 = __val >> 32;					\
+	} while(0)
+
You're silently changing the behavior (as well as the prototype) here, rdmsr() used to fail loudly, now it fails silently, with no way for the caller to check.
Hi Avi!
Not sure what you're saying about the prototype: the old macro was:
include/asm-i386/msr.h +0 -5
-#define rdmsr(msr,val1,val2) \
-	__asm__ __volatile__("rdmsr" \
-			  : "=a" (val1), "=d" (val2) \
-			  : "c" (msr))
-

Quoted 2 lines

Which doesn't seem to be a great difference to me.

Right, scratch that. Was confused by rdmsr_safe().

Quoted 3 lines

The behaviour change (don't oops when an invalid rdmsr is used) was there with CONFIG_PARAVIRT=y, the cleanup just made !CONFIG_PARAVIRT the same. Is it important?

I think so. A function should either never fail, or indicate that it has failed (panic, error return, debug message). With the kinder, gentler rdmsr() one can code the wrong msr value and not notice that something is wrong.

Signature · 1 line

Do not meddle in the internals of kernels, for they are subtle and quick to panic.

Rusty Russell

author

2007-03-22 09:55 UTC #5 lore raw

Quoted 1 line · Avi Kivity

Right, scratch that. Was confused by rdmsr_safe().

Heh... me too 8)

Quoted 8 lines

Quoted 3 lines
The behaviour change (don't oops when an invalid rdmsr is used) was there with CONFIG_PARAVIRT=y, the cleanup just made !CONFIG_PARAVIRT the same. Is it important?
I think so. A function should either never fail, or indicate that it has failed (panic, error return, debug message). With the kinder, gentler rdmsr() one can code the wrong msr value and not notice that something is wrong.

This is persuasive. I think restoring native_read_msr & native_write_msr make sense, and Jeremy's BUG_ON suggestion will cover the CONFIG_PARAVIRT case.

Something like this (I'll roll back into the main patch, retest, and resend tomorrow).

Thanks!
Rusty.

include/asm-i386/msr.h +22 ~~-11~~

diff -r 3ddb2898e72a include/asm-i386/msr.h
@@ -3,7 +3,16 @@
 
 #include <asm/errno.h>
 
-static inline unsigned long long native_read_msr(unsigned int msr, int *err)
+static inline unsigned long long native_read_msr(unsigned int msr)
+{
+	unsigned long long val;
+
+	asm volatile("rdmsr" : "=A" (val) : "c" (msr));
+	return val;
+}
+
+static inline unsigned long long native_read_msr_safe(unsigned int msr,
+						      int *err)
 {
 	unsigned long long val;

@@ -22,7 +31,13 @@ static inline unsigned long long native_
 	return val;
 }
 
-static inline int native_write_msr(unsigned int msr, unsigned long long val)
+static inline void native_write_msr(unsigned int msr, unsigned long long val)
+{
+	asm volatile("wrmsr" : : "c" (msr), "A"(val));
+}
+
+static inline int native_write_msr_safe(unsigned int msr,
+					unsigned long long val)
 {
 	int err;
 	asm volatile("2: wrmsr ; xorl %0,%0\n"

@@ -66,21 +81,17 @@ static inline unsigned long long native_
 
 #define rdmsr(msr,val1,val2)						\
 	do {								\
-		int __err;						\
-		unsigned long long __val = native_read_msr(msr, &__err);	\
+		unsigned long long __val = native_read_msr(msr);	\
 		val1 = __val;						\
 		val2 = __val >> 32;					\
 	} while(0)
 
 #define wrmsr(msr,val1,val2)						\
-	do {								\
-		native_write_msr(msr, ((unsigned long long)val2 << 32) | val1);	\
-	} while(0)
+	native_write_msr(msr, ((unsigned long long)val2 << 32) | val1)
 
 #define rdmsrl(msr,val)					\
 	do {						\
-		int __err;				\
-		(val) = native_read_msr(msr, &__err);	\
+		(val) = native_read_msr(msr);		\
 	} while(0)
 
 static inline void wrmsrl (unsigned long msr, unsigned long long val)

@@ -93,13 +104,13 @@ static inline void wrmsrl (unsigned long
 
 /* wrmsr with exception handling */
 #define wrmsr_safe(msr,val1,val2)						\
-	(native_write_msr(msr, ((unsigned long long)val2 << 32) | val1))
+	(native_write_msr_safe(msr, ((unsigned long long)val2 << 32) | val1))
 
 /* rdmsr with exception handling */
 #define rdmsr_safe(msr,p1,p2)						\
 	({								\
 		int __err;						\
-		unsigned long long __val = native_read_msr(msr, &__err);\
+		unsigned long long __val = native_read_msr_safe(msr, &__err);\
 		(*p1) = __val;						\
 		(*p2) = __val >> 32;					\
 		__err;							\

include/asm-i386/paravirt.h +11 -4

diff -r 3ddb2898e72a include/asm-i386/paravirt.h
@@ -237,19 +237,26 @@ static inline void halt(void)
 	u64 _l = paravirt_ops.read_msr(msr,&_err);		\
 	val1 = (u32)_l;						\
 	val2 = _l >> 32;					\
+	BUG_ON(_err);						\
 } while(0)
 
 #define wrmsr(msr,val1,val2) do {				\
 	u64 _l = ((u64)(val2) << 32) | (val1);			\
-	paravirt_ops.write_msr((msr), _l);			\
+	int _err = paravirt_ops.write_msr((msr), _l);		\
+	BUG_ON(_err);						\
 } while(0)
 
 #define rdmsrl(msr,val) do {					\
 	int _err;						\
 	val = paravirt_ops.read_msr((msr),&_err);		\
-} while(0)
-
-#define wrmsrl(msr,val) (paravirt_ops.write_msr((msr),(val)))
+	BUG_ON(_err);						\
+} while(0)
+
+#define wrmsrl(msr,val) do {					\
+	int _err = paravirt_ops.write_msr((msr),(val));		\
+	BUG_ON(_err);						\
+} while(0)
+
 #define wrmsr_safe(msr,a,b) ({					\
 	u64 _l = ((u64)(b) << 32) | (a);			\
 	paravirt_ops.write_msr((msr),_l);			\

Rusty Russell

author

2007-03-23 11:09 UTC #6 lore raw

PATCH 1/1 Cleanup: rationalize paravirt wrappers

In reply to Rusty Russell

[ Andrew: This is the fixed-up, better-tested, with-Avi-suggestions version. Please tell me if it still breaks something. Passes allmodconfig with and without CONFIG_PARAVIRT, SMP and PREMPT here against 2.6.21-rc4-mm1. ]

There are several nice side effects:

1) write_dt_entry() now takes the correct "struct Xgt_desc_struct *" not "void *".

2) load_TLS is reintroduced to the for loop, not manually unrolled with a #error in case the bounds ever change.

3) Macros become inlines, with type checking.

4) Access to the native versions is trivial for KVM, lguest, Xen and others who might want it.

Signed-off-by Jeremy Fitzhardinge <[email protected]>

Signed-off-by Rusty Russell <[email protected]>

---
 arch/i386/kernel/paravirt.c  |  293 ------------------------------------------
 include/asm-i386/desc.h      |   90 ++++++++----
 include/asm-i386/io.h        |   17 +-
 include/asm-i386/irqflags.h  |   61 ++++++--
 include/asm-i386/msr.h       |  167 ++++++++++++++++-------
 include/asm-i386/paravirt.h  |   17 +-
 include/asm-i386/processor.h |   96 ++++++++++---
 include/asm-i386/system.h    |  139 ++++++++++++-------
 8 files changed, 397 insertions(+), 483 deletions(-)

arch/i386/kernel/paravirt.c +5 ~~-288~~

diff -r fe852f159745 arch/i386/kernel/paravirt.c
@@ -93,292 +93,9 @@ static unsigned native_patch(u8 type, u1
 	return insn_len;
 }
 
-static unsigned long native_get_debugreg(int regno)
-{
-	unsigned long val = 0; 	/* Damn you, gcc! */
-
-	switch (regno) {
-	case 0:
-		asm("movl %%db0, %0" :"=r" (val)); break;
-	case 1:
-		asm("movl %%db1, %0" :"=r" (val)); break;
-	case 2:
-		asm("movl %%db2, %0" :"=r" (val)); break;
-	case 3:
-		asm("movl %%db3, %0" :"=r" (val)); break;
-	case 6:
-		asm("movl %%db6, %0" :"=r" (val)); break;
-	case 7:
-		asm("movl %%db7, %0" :"=r" (val)); break;
-	default:
-		BUG();
-	}
-	return val;
-}
-
-static void native_set_debugreg(int regno, unsigned long value)
-{
-	switch (regno) {
-	case 0:
-		asm("movl %0,%%db0"	: /* no output */ :"r" (value));
-		break;
-	case 1:
-		asm("movl %0,%%db1"	: /* no output */ :"r" (value));
-		break;
-	case 2:
-		asm("movl %0,%%db2"	: /* no output */ :"r" (value));
-		break;
-	case 3:
-		asm("movl %0,%%db3"	: /* no output */ :"r" (value));
-		break;
-	case 6:
-		asm("movl %0,%%db6"	: /* no output */ :"r" (value));
-		break;
-	case 7:
-		asm("movl %0,%%db7"	: /* no output */ :"r" (value));
-		break;
-	default:
-		BUG();
-	}
-}
-
 void init_IRQ(void)
 {
 	paravirt_ops.init_IRQ();
-}
-
-static void native_clts(void)
-{
-	asm volatile ("clts");
-}
-
-static unsigned long native_read_cr0(void)
-{
-	unsigned long val;
-	asm volatile("movl %%cr0,%0\n\t" :"=r" (val));
-	return val;
-}
-
-static void native_write_cr0(unsigned long val)
-{
-	asm volatile("movl %0,%%cr0": :"r" (val));
-}
-
-static unsigned long native_read_cr2(void)
-{
-	unsigned long val;
-	asm volatile("movl %%cr2,%0\n\t" :"=r" (val));
-	return val;
-}
-
-static void native_write_cr2(unsigned long val)
-{
-	asm volatile("movl %0,%%cr2": :"r" (val));
-}
-
-static unsigned long native_read_cr3(void)
-{
-	unsigned long val;
-	asm volatile("movl %%cr3,%0\n\t" :"=r" (val));
-	return val;
-}
-
-static void native_write_cr3(unsigned long val)
-{
-	asm volatile("movl %0,%%cr3": :"r" (val));
-}
-
-static unsigned long native_read_cr4(void)
-{
-	unsigned long val;
-	asm volatile("movl %%cr4,%0\n\t" :"=r" (val));
-	return val;
-}
-
-static unsigned long native_read_cr4_safe(void)
-{
-	unsigned long val;
-	/* This could fault if %cr4 does not exist */
-	asm("1: movl %%cr4, %0		\n"
-		"2:				\n"
-		".section __ex_table,\"a\"	\n"
-		".long 1b,2b			\n"
-		".previous			\n"
-		: "=r" (val): "0" (0));
-	return val;
-}
-
-static void native_write_cr4(unsigned long val)
-{
-	asm volatile("movl %0,%%cr4": :"r" (val));
-}
-
-static unsigned long native_save_fl(void)
-{
-	unsigned long f;
-	asm volatile("pushfl ; popl %0":"=g" (f): /* no input */);
-	return f;
-}
-
-static void native_restore_fl(unsigned long f)
-{
-	asm volatile("pushl %0 ; popfl": /* no output */
-			     :"g" (f)
-			     :"memory", "cc");
-}
-
-static void native_irq_disable(void)
-{
-	asm volatile("cli": : :"memory");
-}
-
-static void native_irq_enable(void)
-{
-	asm volatile("sti": : :"memory");
-}
-
-static void native_safe_halt(void)
-{
-	asm volatile("sti; hlt": : :"memory");
-}
-
-static void native_halt(void)
-{
-	asm volatile("hlt": : :"memory");
-}
-
-static void native_wbinvd(void)
-{
-	asm volatile("wbinvd": : :"memory");
-}
-
-static unsigned long long native_read_msr(unsigned int msr, int *err)
-{
-	unsigned long long val;
-
-	asm volatile("2: rdmsr ; xorl %0,%0\n"
-		     "1:\n\t"
-		     ".section .fixup,\"ax\"\n\t"
-		     "3:  movl %3,%0 ; jmp 1b\n\t"
-		     ".previous\n\t"
- 		     ".section __ex_table,\"a\"\n"
-		     "   .align 4\n\t"
-		     "   .long 	2b,3b\n\t"
-		     ".previous"
-		     : "=r" (*err), "=A" (val)
-		     : "c" (msr), "i" (-EFAULT));
-
-	return val;
-}
-
-static int native_write_msr(unsigned int msr, unsigned long long val)
-{
-	int err;
-	asm volatile("2: wrmsr ; xorl %0,%0\n"
-		     "1:\n\t"
-		     ".section .fixup,\"ax\"\n\t"
-		     "3:  movl %4,%0 ; jmp 1b\n\t"
-		     ".previous\n\t"
- 		     ".section __ex_table,\"a\"\n"
-		     "   .align 4\n\t"
-		     "   .long 	2b,3b\n\t"
-		     ".previous"
-		     : "=a" (err)
-		     : "c" (msr), "0" ((u32)val), "d" ((u32)(val>>32)),
-		       "i" (-EFAULT));
-	return err;
-}
-
-static unsigned long long native_read_tsc(void)
-{
-	unsigned long long val;
-	asm volatile("rdtsc" : "=A" (val));
-	return val;
-}
-
-static unsigned long long native_read_pmc(void)
-{
-	unsigned long long val;
-	asm volatile("rdpmc" : "=A" (val));
-	return val;
-}
-
-static void native_load_tr_desc(void)
-{
-	asm volatile("ltr %w0"::"q" (GDT_ENTRY_TSS*8));
-}
-
-static void native_load_gdt(const struct Xgt_desc_struct *dtr)
-{
-	asm volatile("lgdt %0"::"m" (*dtr));
-}
-
-static void native_load_idt(const struct Xgt_desc_struct *dtr)
-{
-	asm volatile("lidt %0"::"m" (*dtr));
-}
-
-static void native_store_gdt(struct Xgt_desc_struct *dtr)
-{
-	asm ("sgdt %0":"=m" (*dtr));
-}
-
-static void native_store_idt(struct Xgt_desc_struct *dtr)
-{
-	asm ("sidt %0":"=m" (*dtr));
-}
-
-static unsigned long native_store_tr(void)
-{
-	unsigned long tr;
-	asm ("str %0":"=r" (tr));
-	return tr;
-}
-
-static void native_load_tls(struct thread_struct *t, unsigned int cpu)
-{
-#define C(i) get_cpu_gdt_table(cpu)[GDT_ENTRY_TLS_MIN + i] = t->tls_array[i]
-	C(0); C(1); C(2);
-#undef C
-}
-
-static inline void native_write_dt_entry(void *dt, int entry, u32 entry_low, u32 entry_high)
-{
-	u32 *lp = (u32 *)((char *)dt + entry*8);
-	lp[0] = entry_low;
-	lp[1] = entry_high;
-}
-
-static void native_write_ldt_entry(void *dt, int entrynum, u32 low, u32 high)
-{
-	native_write_dt_entry(dt, entrynum, low, high);
-}
-
-static void native_write_gdt_entry(void *dt, int entrynum, u32 low, u32 high)
-{
-	native_write_dt_entry(dt, entrynum, low, high);
-}
-
-static void native_write_idt_entry(void *dt, int entrynum, u32 low, u32 high)
-{
-	native_write_dt_entry(dt, entrynum, low, high);
-}
-
-static void native_load_esp0(struct tss_struct *tss,
-				      struct thread_struct *thread)
-{
-	tss->esp0 = thread->esp0;
-
-	/* This can only happen when SEP is enabled, no need to test "SEP"arately */
-	if (unlikely(tss->ss1 != thread->sysenter_cs)) {
-		tss->ss1 = thread->sysenter_cs;
-		wrmsr(MSR_IA32_SYSENTER_CS, thread->sysenter_cs, 0);
-	}
-}
-
-static void native_io_delay(void)
-{
-	asm volatile("outb %al,$0x80");
 }
 
 static void native_flush_tlb(void)

@@ -517,8 +234,8 @@ struct paravirt_ops paravirt_ops = {
 	.safe_halt = native_safe_halt,
 	.halt = native_halt,
 	.wbinvd = native_wbinvd,
-	.read_msr = native_read_msr,
-	.write_msr = native_write_msr,
+	.read_msr = native_read_msr_safe,
+	.write_msr = native_write_msr_safe,
 	.read_tsc = native_read_tsc,
 	.read_pmc = native_read_pmc,
 	.get_scheduled_cycles = native_read_tsc,

@@ -531,9 +248,9 @@ struct paravirt_ops paravirt_ops = {
 	.store_idt = native_store_idt,
 	.store_tr = native_store_tr,
 	.load_tls = native_load_tls,
-	.write_ldt_entry = native_write_ldt_entry,
-	.write_gdt_entry = native_write_gdt_entry,
-	.write_idt_entry = native_write_idt_entry,
+	.write_ldt_entry = write_dt_entry,
+	.write_gdt_entry = write_dt_entry,
+	.write_idt_entry = write_dt_entry,
 	.load_esp0 = native_load_esp0,
 
 	.set_iopl_mask = native_set_iopl_mask,

include/asm-i386/desc.h +60 ~~-30~~

111 lines not on this page · raw message

include/asm-i386/io.h +10 -7

diff -r fe852f159745 include/asm-i386/io.h
@@ -250,19 +250,22 @@ static inline void flush_write_buffers(v
 
 #endif /* __KERNEL__ */
 
+static inline void native_io_delay(void)
+{
+	asm volatile("outb %%al,$0x80" : : : "memory");
+}
+
 #if defined(CONFIG_PARAVIRT)
 #include <asm/paravirt.h>
 #else
 
-#define __SLOW_DOWN_IO "outb %%al,$0x80;"
-
 static inline void slow_down_io(void) {
-	__asm__ __volatile__(
-		__SLOW_DOWN_IO
+	native_io_delay();
 #ifdef REALLY_SLOW_IO
-		__SLOW_DOWN_IO __SLOW_DOWN_IO __SLOW_DOWN_IO
-#endif
-		: : );
+	native_io_delay();
+	native_io_delay();
+	native_io_delay();
+#endif
 }
 
 #endif

include/asm-i386/irqflags.h +42 ~~-19~~

97 lines not on this page · raw message

include/asm-i386/msr.h +112 ~~-55~~

189 lines not on this page · raw message

include/asm-i386/paravirt.h +9 -8

diff -r fe852f159745 include/asm-i386/paravirt.h
@@ -29,6 +29,7 @@ struct Xgt_desc_struct;
 struct Xgt_desc_struct;
 struct tss_struct;
 struct mm_struct;
+struct desc_struct;
 struct paravirt_ops
 {
 	unsigned int kernel_rpl;

@@ -105,14 +106,13 @@ struct paravirt_ops
 	void (*set_ldt)(const void *desc, unsigned entries);
 	unsigned long (*store_tr)(void);
 	void (*load_tls)(struct thread_struct *t, unsigned int cpu);
-	void (*write_ldt_entry)(void *dt, int entrynum,
-					 u32 low, u32 high);
-	void (*write_gdt_entry)(void *dt, int entrynum,
-					 u32 low, u32 high);
-	void (*write_idt_entry)(void *dt, int entrynum,
-					 u32 low, u32 high);
-	void (*load_esp0)(struct tss_struct *tss,
-				   struct thread_struct *thread);
+	void (*write_ldt_entry)(struct desc_struct *,
+				int entrynum, u32 low, u32 high);
+	void (*write_gdt_entry)(struct desc_struct *,
+				int entrynum, u32 low, u32 high);
+	void (*write_idt_entry)(struct desc_struct *,
+				int entrynum, u32 low, u32 high);
+	void (*load_esp0)(struct tss_struct *tss, struct thread_struct *t);
 
 	void (*set_iopl_mask)(unsigned mask);

@@ -232,6 +232,7 @@ static inline void halt(void)
 
 #define get_kernel_rpl()  (paravirt_ops.kernel_rpl)
 
+/* These should all do BUG_ON(_err), but our headers are too tangled. */
 #define rdmsr(msr,val1,val2) do {				\
 	int _err;						\
 	u64 _l = paravirt_ops.read_msr(msr,&_err);		\

include/asm-i386/processor.h +74 ~~-22~~

124 lines not on this page · raw message

include/asm-i386/system.h +85 ~~-54~~

150 lines not on this page · raw message

Avi Kivity

2007-03-23 11:20 UTC #7 lore raw

Quoted 3 lines · Rusty Russell

4) Access to the native versions is trivial for KVM, lguest, Xen and others who might want it.

kvm will be quite happy with it, as long as the x86_64 version has a similar API.

include/asm-x86 anyone? ;-)

Signature · 1 line

Do not meddle in the internals of kernels, for they are subtle and quick to panic.