From: Jeremy Fitzhardinge <jeremy@goop.org>
To: Rusty Russell <rusty@rustcorp.com.au>,
Zachary Amsden <zach@vmware.com>,
Chris Wright <chrisw@sous-sol.org>
Cc: Virtualization Mailing List <virtualization@lists.osdl.org>
Subject: rough sketch of revised patching infrastructure
Date: Wed, 21 Feb 2007 18:09:50 -0800 [thread overview]
Message-ID: <45DCFB6E.2010806@goop.org> (raw)
[-- Attachment #1: Type: text/plain, Size: 697 bytes --]
Here's the new patching patch. It compiles, but it doesn't, you know,
boot, as such.
The basic idea is to make the incremental cost of patching a new hook as
small as possible. If a backend wants pure default patch handling, then
it can simply call paravirt_patcher with a NULL patch table; this will
end up patching calls to be direct calls to the paravirt op, unless the
op is NULL or points to native_nop. Otherwise, a backend can set up a
patch table to set specific handling for entrypoints, though the default
is still to generate a direct call to the op.
The patch type is now derived from offsetof(paravirt_ops, op), so that
there's no need to maintain a specific enumeration.
J
[-- Attachment #2: paravirt-generic-patching.patch --]
[-- Type: text/x-patch, Size: 25143 bytes --]
diff -r d83f67d4b2b8 arch/i386/kernel/alternative.c
--- a/arch/i386/kernel/alternative.c Wed Feb 21 12:12:31 2007 -0800
+++ b/arch/i386/kernel/alternative.c Wed Feb 21 18:08:17 2007 -0800
@@ -350,9 +350,9 @@ void alternatives_smp_switch(int smp)
#endif
#ifdef CONFIG_PARAVIRT
-void apply_paravirt(struct paravirt_patch *start, struct paravirt_patch *end)
-{
- struct paravirt_patch *p;
+void apply_paravirt(struct paravirt_patch_site *start, struct paravirt_patch_site *end)
+{
+ struct paravirt_patch_site *p;
for (p = start; p < end; p++) {
unsigned int used;
@@ -379,8 +379,6 @@ void apply_paravirt(struct paravirt_patc
/* Sync to be conservative, in case we patched following instructions */
sync_core();
}
-extern struct paravirt_patch __start_parainstructions[],
- __stop_parainstructions[];
#endif /* CONFIG_PARAVIRT */
void __init alternative_instructions(void)
diff -r d83f67d4b2b8 arch/i386/kernel/asm-offsets.c
--- a/arch/i386/kernel/asm-offsets.c Wed Feb 21 12:12:31 2007 -0800
+++ b/arch/i386/kernel/asm-offsets.c Wed Feb 21 18:08:17 2007 -0800
@@ -110,5 +110,11 @@ void foo(void)
OFFSET(PARAVIRT_irq_enable_sysexit, paravirt_ops, irq_enable_sysexit);
OFFSET(PARAVIRT_iret, paravirt_ops, iret);
OFFSET(PARAVIRT_read_cr0, paravirt_ops, read_cr0);
+
+ DEFINE(PARAVIRT_IRQ_DISABLE, PARAVIRT_IRQ_DISABLE);
+ DEFINE(PARAVIRT_IRQ_ENABLE, PARAVIRT_IRQ_ENABLE);
+ DEFINE(PARAVIRT_STI_SYSEXIT, PARAVIRT_STI_SYSEXIT);
+ DEFINE(PARAVIRT_IRQ_DISABLE, PARAVIRT_IRQ_DISABLE);
+ DEFINE(PARAVIRT_INTERRUPT_RETURN, PARAVIRT_INTERRUPT_RETURN);
#endif
}
diff -r d83f67d4b2b8 arch/i386/kernel/paravirt.c
--- a/arch/i386/kernel/paravirt.c Wed Feb 21 12:12:31 2007 -0800
+++ b/arch/i386/kernel/paravirt.c Wed Feb 21 18:08:17 2007 -0800
@@ -61,34 +61,106 @@ DEF_NATIVE(iret, "iret");
DEF_NATIVE(iret, "iret");
DEF_NATIVE(sti_sysexit, "sti; sysexit");
-static const struct native_insns
-{
- const char *start, *end;
-} native_insns[] = {
- [PARAVIRT_IRQ_DISABLE] = { start_cli, end_cli },
- [PARAVIRT_IRQ_ENABLE] = { start_sti, end_sti },
- [PARAVIRT_RESTORE_FLAGS] = { start_popf, end_popf },
- [PARAVIRT_SAVE_FLAGS] = { start_pushf, end_pushf },
- [PARAVIRT_SAVE_FLAGS_IRQ_DISABLE] = { start_pushf_cli, end_pushf_cli },
- [PARAVIRT_INTERRUPT_RETURN] = { start_iret, end_iret },
- [PARAVIRT_STI_SYSEXIT] = { start_sti_sysexit, end_sti_sysexit },
+static const struct paravirt_patch native_patches[256] =
+{
+ [PARAVIRT_IRQ_DISABLE] = { PVP_INSTR, { { start_cli, end_cli } } },
+ [PARAVIRT_IRQ_ENABLE] = { PVP_INSTR, { { start_sti, end_sti } } },
+ [PARAVIRT_RESTORE_FLAGS] = { PVP_INSTR, { { start_popf, end_popf } } },
+ [PARAVIRT_SAVE_FLAGS] = { PVP_INSTR, { { start_pushf, end_pushf } } },
+ [PARAVIRT_SAVE_FLAGS_IRQ_DISABLE] = { PVP_INSTR, { { start_pushf_cli, end_pushf_cli } } },
+ [PARAVIRT_INTERRUPT_RETURN] = { PVP_INSTR, {{ start_iret, end_iret }} },
+ [PARAVIRT_STI_SYSEXIT] = { PVP_INSTR, {{ start_sti_sysexit, end_sti_sysexit }} },
+
+ [PARAVIRT_PATCH(pmd_val)] = { PVP_NOP },
+ [PARAVIRT_PATCH(pgd_val)] = { PVP_NOP },
};
static unsigned native_patch(u8 type, u16 clobbers, void *insns, unsigned len)
{
- unsigned int insn_len;
-
- /* Don't touch it if we don't have a replacement */
- if (type >= ARRAY_SIZE(native_insns) || !native_insns[type].start)
- return len;
-
- insn_len = native_insns[type].end - native_insns[type].start;
-
- /* Similarly if we can't fit replacement. */
- if (len < insn_len)
- return len;
-
- memcpy(insns, native_insns[type].start, insn_len);
+ return paravirt_patcher(type, clobbers, insns, len, native_patches);
+}
+
+static unsigned gen_call(unsigned char *call, void *target)
+{
+ unsigned long delta = (unsigned long)target - (unsigned long)(call+5);
+
+ *call++ = 0xe8; /* call */
+ *(unsigned long *)call = delta;
+
+ return 5;
+}
+
+static unsigned gen_jmp(unsigned char *call, void *target)
+{
+ unsigned long delta = (unsigned long)target - (unsigned long)(call+5);
+
+ *call++ = 0xe9; /* jmp */
+ *(unsigned long *)call = delta;
+
+ return 5;
+}
+
+unsigned paravirt_patcher(u8 type, u16 clobbers, void *insns, unsigned len,
+ const struct paravirt_patch *patch_table)
+{
+ unsigned insn_len;
+ static const struct paravirt_patch default_call = { PVP_OPCALL };
+ static const struct paravirt_patch default_jmp = { PVP_OPJMP };
+ void *opfunc = *((void **)¶virt_ops + type);
+ const struct paravirt_patch *p;
+
+ if (patch_table != NULL)
+ p = &patch_table[type];
+ else {
+ /* If there's no patch table, then do some sensable defaults */
+ if (type == PARAVIRT_PATCH(iret) ||
+ type == PARAVIRT_PATCH(irq_enable_sysexit))
+ p = &default_jmp;
+ else
+ p = &default_call;
+ }
+
+ switch(p->action) {
+ default:
+ case PVP_NULL:
+ /* do nothing */
+ insn_len = len;
+ break;
+
+ case PVP_NOP:
+ /* replace with nops */
+ insn_len = 0;
+ break;
+
+ case PVP_OPCALL:
+ /* generate direct call to paravirt_op (or NOPs if
+ there's no function) */
+ if (opfunc == NULL || opfunc == (void *)native_nop)
+ insn_len = 0; /* missing or nop function */
+ else
+ insn_len = gen_call(insns, opfunc);
+ break;
+
+ case PVP_OPJMP:
+ BUG_ON(opfunc == NULL);
+ insn_len = gen_jmp(insns, opfunc);
+ break;
+
+ case PVP_INSTR:
+ insn_len = p->instr.end - p->instr.start;
+ if (p->instr.start == NULL || insn_len > len)
+ insn_len = len;
+ else
+ memcpy(insns, p->instr.start, insn_len);
+ break;
+
+ case PVP_SPECIAL:
+ insn_len = (*p->fn)(type, clobbers, insns, len);
+ break;
+ }
+
+ BUG_ON(insn_len > len);
+
return insn_len;
}
diff -r d83f67d4b2b8 arch/i386/kernel/vmi.c
--- a/arch/i386/kernel/vmi.c Wed Feb 21 12:12:31 2007 -0800
+++ b/arch/i386/kernel/vmi.c Wed Feb 21 18:08:17 2007 -0800
@@ -70,10 +70,6 @@ struct {
void (*set_initial_ap_state)(int, int);
void (*halt)(void);
} vmi_ops;
-
-/* XXX move this to alternative.h */
-extern struct paravirt_patch __start_parainstructions[],
- __stop_parainstructions[];
/*
* VMI patching routines.
diff -r d83f67d4b2b8 include/asm-i386/alternative.h
--- a/include/asm-i386/alternative.h Wed Feb 21 12:12:31 2007 -0800
+++ b/include/asm-i386/alternative.h Wed Feb 21 18:08:17 2007 -0800
@@ -118,12 +118,12 @@ static inline void alternatives_smp_swit
#define LOCK_PREFIX ""
#endif
-struct paravirt_patch;
+struct paravirt_patch_site;
#ifdef CONFIG_PARAVIRT
-void apply_paravirt(struct paravirt_patch *start, struct paravirt_patch *end);
+void apply_paravirt(struct paravirt_patch_site *start, struct paravirt_patch_site *end);
#else
static inline void
-apply_paravirt(struct paravirt_patch *start, struct paravirt_patch *end)
+apply_paravirt(struct paravirt_patch_site *start, struct paravirt_patch_site *end)
{}
#define __start_parainstructions NULL
#define __stop_parainstructions NULL
diff -r d83f67d4b2b8 include/asm-i386/paravirt.h
--- a/include/asm-i386/paravirt.h Wed Feb 21 12:12:31 2007 -0800
+++ b/include/asm-i386/paravirt.h Wed Feb 21 18:08:17 2007 -0800
@@ -9,14 +9,6 @@
#ifdef CONFIG_PARAVIRT
/* These are the most performance critical ops, so we want to be able to patch
* callers */
-#define PARAVIRT_IRQ_DISABLE 0
-#define PARAVIRT_IRQ_ENABLE 1
-#define PARAVIRT_RESTORE_FLAGS 2
-#define PARAVIRT_SAVE_FLAGS 3
-#define PARAVIRT_SAVE_FLAGS_IRQ_DISABLE 4
-#define PARAVIRT_INTERRUPT_RETURN 5
-#define PARAVIRT_STI_SYSEXIT 6
-
/* Bitmask of what can be clobbered: usually at least eax. */
#define CLBR_NONE 0x0
#define CLBR_EAX 0x1
@@ -238,21 +230,109 @@ pgd_t native_make_pgd(unsigned long pgd)
#define paravirt_enabled() (paravirt_ops.paravirt_enabled)
+#define paravirt_type(type) [paravirt_typenum] "i" (type)
+#define paravirt_clobber(clobber) [paravirt_clobber] "i" (clobber)
+
+#define PARAVIRT_PATCH(x) (offsetof(struct paravirt_ops, x) / sizeof(void *))
+#define PARAVIRT_PATCH_SPECIAL(x) (0x80+x)
+
+#define _paravirt_alt(insn_string, type, clobber) \
+ "771:\n\t" insn_string "\n" "772:\n" \
+ ".pushsection .parainstructions,\"a\"\n" \
+ " .long 771b\n" \
+ " .byte " type "\n" \
+ " .byte 772b-771b\n" \
+ " .short " clobber "\n" \
+ ".popsection"
+
+#define paravirt_alt(insn_string) \
+ _paravirt_alt(insn_string, "%c[paravirt_typenum]", "%c[paravirt_clobber]")
+
+#define PVOP_CALL0(__rettype, op) \
+ ({ \
+ unsigned long long __ret; \
+ asm(paravirt_alt("call *%1") \
+ : "=A" (__ret) \
+ : "m" (paravirt_ops.op), \
+ paravirt_type(PARAVIRT_PATCH(op)), \
+ paravirt_clobber(CLBR_EAX|CLBR_ECX|CLBR_EDX) \
+ : "ecx"); \
+ (__rettype)__ret; \
+ })
+
+#define PVOP_CALL1(__rettype, op, arg1) \
+ ({ \
+ unsigned long long __ret; \
+ asm(paravirt_alt("call *%1") \
+ : "=A" (__ret) \
+ : "m" (paravirt_ops.op), \
+ "a" ((u32)(arg1)), \
+ paravirt_type(PARAVIRT_PATCH(op)), \
+ paravirt_clobber(CLBR_EAX|CLBR_ECX|CLBR_EDX) \
+ : "ecx"); \
+ (__rettype)__ret; \
+ })
+
+#define PVOP_CALL2(__rettype, op, arg1, arg2) \
+ ({ \
+ unsigned long long __ret; \
+ asm(paravirt_alt("call *%1") \
+ : "=A" (__ret) \
+ : "m" (paravirt_ops.op), \
+ "a" ((u32)(arg1)), \
+ "d" ((u32)(arg2)), \
+ paravirt_type(PARAVIRT_PATCH(op)), \
+ paravirt_clobber(CLBR_EAX|CLBR_ECX|CLBR_EDX) \
+ : "ecx"); \
+ (__rettype)__ret; \
+ })
+
+#define PVOP_CALL3(__rettype, op, arg1, arg2, arg3) \
+ ({ \
+ unsigned long long __ret; \
+ asm(paravirt_alt("call *%1") \
+ : "=A" (__ret) \
+ : "m" (paravirt_ops.op), \
+ "a" ((u32)(arg1)), \
+ "d" ((u32)(arg2)), \
+ "c" ((u32)(arg3)), \
+ paravirt_type(PARAVIRT_PATCH(op)), \
+ paravirt_clobber(CLBR_EAX|CLBR_ECX|CLBR_EDX) \
+ : "ecx"); \
+ (__rettype)__ret; \
+ })
+
+#define PVOP_CALL4(__rettype, op, arg1, arg2, arg3, arg4) \
+ ({ \
+ unsigned long long __ret; \
+ asm(paravirt_alt("push %5; call *%1; add $4,%%esp") \
+ : "=A" (__ret) \
+ : "m" (paravirt_ops.op), \
+ "a" ((u32)(arg1)), \
+ "d" ((u32)(arg2)), \
+ "c" ((u32)(arg3)), \
+ "mr" ((u32)(arg4)), \
+ paravirt_type(PARAVIRT_PATCH(op)), \
+ paravirt_clobber(CLBR_EAX|CLBR_ECX|CLBR_EDX) \
+ : "ecx"); \
+ (__rettype)__ret; \
+ })
+
static inline void load_esp0(struct tss_struct *tss,
struct thread_struct *thread)
{
- paravirt_ops.load_esp0(tss, thread);
+ PVOP_CALL2(void, load_esp0, tss, thread);
}
#define ARCH_SETUP paravirt_ops.arch_setup();
static inline unsigned long get_wallclock(void)
{
- return paravirt_ops.get_wallclock();
+ return PVOP_CALL0(unsigned long, get_wallclock);
}
static inline int set_wallclock(unsigned long nowtime)
{
- return paravirt_ops.set_wallclock(nowtime);
+ return PVOP_CALL1(int, set_wallclock, nowtime);
}
static inline void do_time_init(void)
@@ -264,42 +344,47 @@ static inline void __cpuid(unsigned int
static inline void __cpuid(unsigned int *eax, unsigned int *ebx,
unsigned int *ecx, unsigned int *edx)
{
- paravirt_ops.cpuid(eax, ebx, ecx, edx);
+ PVOP_CALL4(void, cpuid, eax, ebx, ecx, edx);
}
/*
* These special macros can be used to get or set a debugging register
*/
-#define get_debugreg(var, reg) var = paravirt_ops.get_debugreg(reg)
-#define set_debugreg(val, reg) paravirt_ops.set_debugreg(reg, val)
-
-#define clts() paravirt_ops.clts()
-
-#define read_cr0() paravirt_ops.read_cr0()
-#define write_cr0(x) paravirt_ops.write_cr0(x)
-
-#define read_cr2() paravirt_ops.read_cr2()
-#define write_cr2(x) paravirt_ops.write_cr2(x)
-
-#define read_cr3() paravirt_ops.read_cr3()
-#define write_cr3(x) paravirt_ops.write_cr3(x)
-
-#define read_cr4() paravirt_ops.read_cr4()
-#define read_cr4_safe(x) paravirt_ops.read_cr4_safe()
-#define write_cr4(x) paravirt_ops.write_cr4(x)
-
-#define raw_ptep_get_and_clear(xp) (paravirt_ops.ptep_get_and_clear(xp))
+#define get_debugreg(var, reg) var = PVOP_CALL1(unsigned long, get_debugreg, reg)
+#define set_debugreg(val, reg) PVOP_CALL2(void, set_debugreg, reg, val)
+
+#define clts() PVOP_CALL0(void, clts)
+
+#define read_cr0() PVOP_CALL0(unsigned long, read_cr0)
+#define write_cr0(x) PVOP_CALL1(void, write_cr0, x)
+
+#define read_cr2() PVOP_CALL0(unsigned long, read_cr2)
+#define write_cr2(x) PVOP_CALL1(void, write_cr2, x)
+
+#define read_cr3() PVOP_CALL0(unsigned long, read_cr3)
+#define write_cr3(x) PVOP_CALL1(void, write_cr3, x)
+
+#define read_cr4() PVOP_CALL0(unsigned long, read_cr4)
+#define read_cr4_safe(x) PVOP_CALL0(unsigned long, read_cr4_safe)
+#define write_cr4(x) PVOP_CALL1(void, write_cr4, x)
+
+static inline pte_t raw_ptep_get_and_clear(pte_t *p)
+{
+ pte_t ret;
+ PVOP_CALL2(void, ptep_get_and_clear, &ret, p);
+ return ret;
+}
static inline void raw_safe_halt(void)
{
- paravirt_ops.safe_halt();
+ PVOP_CALL0(void, safe_halt);
}
static inline void halt(void)
{
- paravirt_ops.safe_halt();
-}
-#define wbinvd() paravirt_ops.wbinvd()
+ PVOP_CALL0(void, safe_halt);
+}
+#define wbinvd() PVOP_CALL0(void, wbinvd)
#define get_kernel_rpl() (paravirt_ops.kernel_rpl)
@@ -335,22 +420,22 @@ static inline void halt(void)
_err; })
#define rdtsc(low,high) do { \
- u64 _l = paravirt_ops.read_tsc(); \
+ u64 _l = PVOP_CALL0(u64, read_tsc); \
low = (u32)_l; \
high = _l >> 32; \
} while(0)
#define rdtscl(low) do { \
- u64 _l = paravirt_ops.read_tsc(); \
+ u64 _l = PVOP_CALL0(u64, read_tsc); \
low = (int)_l; \
} while(0)
-#define rdtscll(val) (val = paravirt_ops.read_tsc())
+#define rdtscll(val) (val = PVOP_CALL0(u64, read_tsc))
#define write_tsc(val1,val2) wrmsr(0x10, val1, val2)
#define rdpmc(counter,low,high) do { \
- u64 _l = paravirt_ops.read_pmc(); \
+ u64 _l = PVOP_CALL0(u64, read_pmc); \
low = (u32)_l; \
high = _l >> 32; \
} while(0)
@@ -371,15 +456,66 @@ static inline void halt(void)
(paravirt_ops.write_idt_entry((dt), (entry), (low), (high)))
#define set_iopl_mask(mask) (paravirt_ops.set_iopl_mask(mask))
-#define __pte(x) paravirt_ops.make_pte(x)
-#define __pgd(x) paravirt_ops.make_pgd(x)
-
-#define pte_val(x) paravirt_ops.pte_val(x)
-#define pgd_val(x) paravirt_ops.pgd_val(x)
-
#ifdef CONFIG_X86_PAE
-#define __pmd(x) paravirt_ops.make_pmd(x)
-#define pmd_val(x) paravirt_ops.pmd_val(x)
+static inline pte_t __pte(unsigned long long val)
+{
+ pte_t ret;
+ PVOP_CALL3(void, make_pte, &ret, val, val >> 32);
+ return ret;
+}
+
+static inline pmd_t __pmd(unsigned long long val)
+{
+ pmd_t ret;
+ PVOP_CALL3(void, make_pmd, &ret, val, val >> 32);
+ return ret;
+}
+
+static inline pgd_t __pgd(unsigned long long val)
+{
+ pgd_t ret;
+ PVOP_CALL3(void, make_pgd, &ret, val, val >> 32);
+ return ret;
+}
+
+static inline unsigned long long pte_val(pte_t x)
+{
+ return PVOP_CALL2(unsigned long long, pte_val, x.pte_low, x.pte_high);
+}
+
+static inline unsigned long long pmd_val(pmd_t x)
+{
+ return PVOP_CALL2(unsigned long long, pmd_val, x.pmd, x.pmd >> 32);
+}
+
+static inline unsigned long long pgd_val(pgd_t x)
+{
+ return PVOP_CALL2(unsigned long long, pgd_val, x.pgd, x.pgd >> 32);
+}
+#else
+static inline pte_t __pte(unsigned long val)
+{
+ pte_t ret;
+ PVOP_CALL2(void, make_pte, &ret, val);
+ return ret;
+}
+
+static inline pgd_t __pte(unsigned long val)
+{
+ pgd_t ret;
+ PVOP_CALL2(void, make_pgd, &ret, val);
+ return ret;
+}
+
+static inline unsigned long pte_val(pte_t x)
+{
+ PVOP_CALL1(unsigned long, pte_val, x.pte_low);
+}
+
+static inline unsigned long pgd_val(pgd_t x)
+{
+ PVOP_CALL2(unsigned long, pgd_val, x.pgd);
+}
#endif
/* The paravirtualized I/O functions */
@@ -456,18 +592,18 @@ static inline void paravirt_activate_mm(
static inline void paravirt_activate_mm(struct mm_struct *prev,
struct mm_struct *next)
{
- paravirt_ops.activate_mm(prev, next);
+ PVOP_CALL2(void, activate_mm, prev, next);
}
static inline void arch_dup_mmap(struct mm_struct *oldmm,
struct mm_struct *mm)
{
- paravirt_ops.dup_mmap(oldmm, mm);
+ PVOP_CALL2(void, dup_mmap, oldmm, mm);
}
static inline void arch_exit_mmap(struct mm_struct *mm)
{
- paravirt_ops.exit_mmap(mm);
+ PVOP_CALL1(void, exit_mmap, mm);
}
static inline void paravirt_init_pda(struct i386_pda *pda, int cpu)
@@ -497,33 +633,41 @@ unsigned long native_store_tr(void);
static inline void set_pte(pte_t *ptep, pte_t pteval)
{
- paravirt_ops.set_pte(ptep, pteval);
+#ifdef CONFIG_X86_PAE
+ PVOP_CALL3(void, set_pte, ptep, pteval.pte_low, pteval.pte_high);
+#else
+ PVOP_CALL2(void, set_pte, ptep, pteval.pte_low);
+#endif
}
static inline void set_pte_at(struct mm_struct *mm, u32 addr, pte_t *ptep, pte_t pteval)
{
+#ifndef CONFIG_X86_PAE
+ PVOP_CALL4(void, set_pte_at, mm, addr, ptep, pteval.pte_low);
+#else
paravirt_ops.set_pte_at(mm, addr, ptep, pteval);
+#endif
}
static inline void set_pmd(pmd_t *pmdp, pmd_t pmdval)
{
- paravirt_ops.set_pmd(pmdp, pmdval);
+ PVOP_CALL3(void, set_pmd, pmdp, pmdval.pmd, pmdval.pmd >> 32);
}
static inline void pte_update(struct mm_struct *mm, u32 addr, pte_t *ptep)
{
- paravirt_ops.pte_update(mm, addr, ptep);
+ PVOP_CALL3(void, pte_update, mm, addr, ptep);
}
static inline void pte_update_defer(struct mm_struct *mm, u32 addr, pte_t *ptep)
{
- paravirt_ops.pte_update_defer(mm, addr, ptep);
+ PVOP_CALL3(void, pte_update_defer, mm, addr, ptep);
}
#ifdef CONFIG_X86_PAE
static inline void set_pte_atomic(pte_t *ptep, pte_t pteval)
{
- paravirt_ops.set_pte_atomic(ptep, pteval);
+ PVOP_CALL3(void, set_pte_atomic, ptep, pteval.pte_low, pteval.pte_high);
}
static inline void set_pte_present(struct mm_struct *mm, unsigned long addr, pte_t *ptep, pte_t pte)
@@ -533,17 +677,17 @@ static inline void set_pte_present(struc
static inline void set_pud(pud_t *pudp, pud_t pudval)
{
- paravirt_ops.set_pud(pudp, pudval);
+ PVOP_CALL3(void, set_pud, pudp, pudval.pgd.pgd, pudval.pgd.pgd >> 32);
}
static inline void pte_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
{
- paravirt_ops.pte_clear(mm, addr, ptep);
+ PVOP_CALL3(void, pte_clear, mm, addr, ptep);
}
static inline void pmd_clear(pmd_t *pmdp)
{
- paravirt_ops.pmd_clear(pmdp);
+ PVOP_CALL1(void, pmd_clear, pmdp);
}
#endif
@@ -553,29 +697,32 @@ static inline void pmd_clear(pmd_t *pmdp
#define PARAVIRT_LAZY_CPU 2
#define __HAVE_ARCH_ENTER_LAZY_CPU_MODE
-#define arch_enter_lazy_cpu_mode() paravirt_ops.set_lazy_mode(PARAVIRT_LAZY_CPU)
-#define arch_leave_lazy_cpu_mode() paravirt_ops.set_lazy_mode(PARAVIRT_LAZY_NONE)
+#define arch_enter_lazy_cpu_mode() PVOP_CALL1(void, set_lazy_mode, PARAVIRT_LAZY_CPU)
+#define arch_leave_lazy_cpu_mode() PVOP_CALL1(void, set_lazy_mode, PARAVIRT_LAZY_NONE)
#define __HAVE_ARCH_ENTER_LAZY_MMU_MODE
-#define arch_enter_lazy_mmu_mode() paravirt_ops.set_lazy_mode(PARAVIRT_LAZY_MMU)
-#define arch_leave_lazy_mmu_mode() paravirt_ops.set_lazy_mode(PARAVIRT_LAZY_NONE)
+#define arch_enter_lazy_mmu_mode() PVOP_CALL1(void, set_lazy_mode, PARAVIRT_LAZY_MMU)
+#define arch_leave_lazy_mmu_mode() PVOP_CALL1(void, set_lazy_mode, PARAVIRT_LAZY_NONE)
+
+#define PARAVIRT_IRQ_DISABLE PARAVIRT_PATCH(irq_disable)
+#define PARAVIRT_IRQ_ENABLE PARAVIRT_PATCH(irq_enable)
+#define PARAVIRT_RESTORE_FLAGS PARAVIRT_PATCH(restore_fl)
+#define PARAVIRT_SAVE_FLAGS PARAVIRT_PATCH(save_fl)
+#define PARAVIRT_SAVE_FLAGS_IRQ_DISABLE PARAVIRT_PATCH_SPECIAL(0)
+#define PARAVIRT_INTERRUPT_RETURN PARAVIRT_PATCH(iret)
+#define PARAVIRT_STI_SYSEXIT PARAVIRT_PATCH(irq_enable_sysexit)
+
/* These all sit in the .parainstructions section to tell us what to patch. */
-struct paravirt_patch {
+struct paravirt_patch_site {
u8 *instr; /* original instructions */
u8 instrtype; /* type of this instruction */
u8 len; /* length of original instruction */
u16 clobbers; /* what registers you may clobber */
};
-#define paravirt_alt(insn_string, typenum, clobber) \
- "771:\n\t" insn_string "\n" "772:\n" \
- ".pushsection .parainstructions,\"a\"\n" \
- " .long 771b\n" \
- " .byte " __stringify(typenum) "\n" \
- " .byte 772b-771b\n" \
- " .short " __stringify(clobber) "\n" \
- ".popsection"
+extern struct paravirt_patch_site __start_parainstructions[],
+ __stop_parainstructions[];
static inline unsigned long __raw_local_save_flags(void)
{
@@ -583,9 +730,10 @@ static inline unsigned long __raw_local_
__asm__ __volatile__(paravirt_alt( "pushl %%ecx; pushl %%edx;"
"call *%1;"
- "popl %%edx; popl %%ecx",
- PARAVIRT_SAVE_FLAGS, CLBR_NONE)
- : "=a"(f): "m"(paravirt_ops.save_fl)
+ "popl %%edx; popl %%ecx")
+ : "=a"(f): "m"(paravirt_ops.save_fl),
+ paravirt_type(PARAVIRT_SAVE_FLAGS),
+ paravirt_clobber(CLBR_NONE)
: "memory", "cc");
return f;
}
@@ -594,9 +742,10 @@ static inline void raw_local_irq_restore
{
__asm__ __volatile__(paravirt_alt( "pushl %%ecx; pushl %%edx;"
"call *%1;"
- "popl %%edx; popl %%ecx",
- PARAVIRT_RESTORE_FLAGS, CLBR_EAX)
- : "=a"(f) : "m" (paravirt_ops.restore_fl), "0"(f)
+ "popl %%edx; popl %%ecx")
+ : "=a"(f) : "m" (paravirt_ops.restore_fl), "0"(f),
+ paravirt_type(PARAVIRT_RESTORE_FLAGS),
+ paravirt_clobber(CLBR_EAX)
: "memory", "cc");
}
@@ -604,9 +753,10 @@ static inline void raw_local_irq_disable
{
__asm__ __volatile__(paravirt_alt( "pushl %%ecx; pushl %%edx;"
"call *%0;"
- "popl %%edx; popl %%ecx",
- PARAVIRT_IRQ_DISABLE, CLBR_EAX)
- : : "m" (paravirt_ops.irq_disable)
+ "popl %%edx; popl %%ecx")
+ : : "m" (paravirt_ops.irq_disable),
+ paravirt_type(PARAVIRT_IRQ_DISABLE),
+ paravirt_clobber(CLBR_EAX)
: "memory", "eax", "cc");
}
@@ -614,9 +764,10 @@ static inline void raw_local_irq_enable(
{
__asm__ __volatile__(paravirt_alt( "pushl %%ecx; pushl %%edx;"
"call *%0;"
- "popl %%edx; popl %%ecx",
- PARAVIRT_IRQ_ENABLE, CLBR_EAX)
- : : "m" (paravirt_ops.irq_enable)
+ "popl %%edx; popl %%ecx")
+ : : "m" (paravirt_ops.irq_enable),
+ paravirt_type(PARAVIRT_IRQ_ENABLE),
+ paravirt_clobber(CLBR_EAX)
: "memory", "eax", "cc");
}
@@ -627,30 +778,63 @@ static inline unsigned long __raw_local_
__asm__ __volatile__(paravirt_alt( "pushl %%ecx; pushl %%edx;"
"call *%1; pushl %%eax;"
"call *%2; popl %%eax;"
- "popl %%edx; popl %%ecx",
- PARAVIRT_SAVE_FLAGS_IRQ_DISABLE,
- CLBR_NONE)
+ "popl %%edx; popl %%ecx")
: "=a"(f)
: "m" (paravirt_ops.save_fl),
- "m" (paravirt_ops.irq_disable)
+ "m" (paravirt_ops.irq_disable),
+ paravirt_type(PARAVIRT_SAVE_FLAGS_IRQ_DISABLE),
+ paravirt_clobber(CLBR_NONE)
: "memory", "cc");
return f;
}
-#define CLI_STRING paravirt_alt("pushl %%ecx; pushl %%edx;" \
- "call *paravirt_ops+%c[irq_disable];" \
- "popl %%edx; popl %%ecx", \
- PARAVIRT_IRQ_DISABLE, CLBR_EAX)
-
-#define STI_STRING paravirt_alt("pushl %%ecx; pushl %%edx;" \
- "call *paravirt_ops+%c[irq_enable];" \
- "popl %%edx; popl %%ecx", \
- PARAVIRT_IRQ_ENABLE, CLBR_EAX)
+#define CLI_STRING _paravirt_alt("pushl %%ecx; pushl %%edx;" \
+ "call *paravirt_ops+%c[irq_disable];" \
+ "popl %%edx; popl %%ecx", \
+ "%c[paravirt_cli_type]", "%c[paravirt_clobber]")
+
+#define STI_STRING _paravirt_alt("pushl %%ecx; pushl %%edx;" \
+ "call *paravirt_ops+%c[irq_enable];" \
+ "popl %%edx; popl %%ecx", \
+ "%c[paravirt_cli_type]", "%c[paravirt_clobber]")
+
#define CLI_STI_CLOBBERS , "%eax"
-#define CLI_STI_INPUT_ARGS \
+#define CLI_STI_INPUT_ARGS \
, \
- [irq_disable] "i" (offsetof(struct paravirt_ops, irq_disable)), \
- [irq_enable] "i" (offsetof(struct paravirt_ops, irq_enable))
+ [irq_disable] "i" (offsetof(struct paravirt_ops, irq_disable)), \
+ [irq_enable] "i" (offsetof(struct paravirt_ops, irq_enable)), \
+ [paravirt_cli_type] "i" (PARAVIRT_IRQ_DISABLE), \
+ [paravirt_sti_type] "i" (PARAVIRT_IRQ_DISABLE), \
+ paravirt_clobber(CLBR_NONE)
+
+/* Specify to the paravirt patcher how a particular patch should be
+ dealt with. PVP_OPCALL is the default (=0) - it just replaces the
+ indirect call via paravirt_ops with a direct call to the
+ function. */
+enum pvpatch {
+ PVP_OPCALL, /* direct call to pv_op */
+ PVP_OPJMP, /* direct jump to pv_op */
+ PVP_NULL, /* do nothing */
+ PVP_NOP, /* patch with nops */
+ PVP_INSTR, /* replace with literal code */
+ PVP_SPECIAL, /* backend-specific special actions */
+};
+struct paravirt_patch
+{
+ enum pvpatch action;
+ union {
+ struct pvp_instructions {
+ const char *start, *end;
+ } instr;
+ void *p;
+ unsigned long ul;
+ unsigned (*fn)(u8 type, u16 clobbers, void *insns, unsigned len);
+ };
+};
+
+/* generic patcher */
+unsigned paravirt_patcher(u8 type, u16 clobbers, void *site, unsigned site_len,
+ const struct paravirt_patch *patch_table);
#else /* __ASSEMBLY__ */
[-- Attachment #3: Type: text/plain, Size: 165 bytes --]
_______________________________________________
Virtualization mailing list
Virtualization@lists.osdl.org
https://lists.osdl.org/mailman/listinfo/virtualization
next reply other threads:[~2007-02-22 2:09 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-02-22 2:09 Jeremy Fitzhardinge [this message]
2007-02-22 3:43 ` rough sketch of revised patching infrastructure Rusty Russell
2007-02-22 10:22 ` Jeremy Fitzhardinge
2007-02-22 10:55 ` Rusty Russell
2007-02-22 18:13 ` Jeremy Fitzhardinge
2007-02-22 22:14 ` Zachary Amsden
2007-02-22 22:49 ` Jeremy Fitzhardinge
2007-02-22 22:51 ` Ian Campbell
2007-02-22 22:54 ` Jeremy Fitzhardinge
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=45DCFB6E.2010806@goop.org \
--to=jeremy@goop.org \
--cc=chrisw@sous-sol.org \
--cc=rusty@rustcorp.com.au \
--cc=virtualization@lists.osdl.org \
--cc=zach@vmware.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).