paravirt: refactor struct paravirt_ops into smaller pv_*_ops

This patch refactors the paravirt_ops structure into groups of
functionally related ops:

pv_info - random info, rather than function entrypoints
pv_init_ops - functions used at boot time (some for module_init too)
pv_misc_ops - lazy mode, which didn't fit well anywhere else
pv_time_ops - time-related functions
pv_cpu_ops - various privileged instruction ops
pv_irq_ops - operations for managing interrupt state
pv_apic_ops - APIC operations
pv_mmu_ops - operations for managing pagetables

There are several motivations for this:

1. Some of these ops will be general to all x86, and some will be
   i386/x86-64 specific.  This makes it easier to share common stuff
   while allowing separate implementations where needed.

2. At the moment we must export all of paravirt_ops, but modules only
   need selected parts of it.  This allows us to export on a case by case
   basis (and also choose which export license we want to apply).

3. Functional groupings make things a bit more readable.

Struct paravirt_ops is now only used as a template to generate
patch-site identifiers, and to extract function pointers for inserting
into jmp/calls when patching.  It is only instantiated when needed.

Signed-off-by: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Cc: Andi Kleen <ak@suse.de>
Cc: Zach Amsden <zach@vmware.com>
Cc: Avi Kivity <avi@qumranet.com>
Cc: Anthony Liguory <aliguori@us.ibm.com>
Cc: "Glauber de Oliveira Costa" <glommer@gmail.com>
Cc: Jun Nakajima <jun.nakajima@intel.com>

6 files changed, 257 insertions, 199 deletions

diff --git a/arch/x86/kernel/alternative.c b/arch/x86/kernel/alternative.c
index bd72d94e713..63c55148dd0 100644
--- a/arch/x86/kernel/alternative.c
+++ b/arch/x86/kernel/alternative.c
@@ -368,8 +368,8 @@ void apply_paravirt(struct paravirt_patch_site *start,
 		BUG_ON(p->len > MAX_PATCH_LEN);
 		/* prep the buffer with the original instructions */
 		memcpy(insnbuf, p->instr, p->len);
-		used = paravirt_ops.patch(p->instrtype, p->clobbers, insnbuf,
-					  (unsigned long)p->instr, p->len);
+		used = pv_init_ops.patch(p->instrtype, p->clobbers, insnbuf,
+					 (unsigned long)p->instr, p->len);
 
 		BUG_ON(used > p->len);
 
diff --git a/arch/x86/kernel/asm-offsets_32.c b/arch/x86/kernel/asm-offsets_32.c
index 8029742c0fc..f1b7cdda82b 100644
--- a/arch/x86/kernel/asm-offsets_32.c
+++ b/arch/x86/kernel/asm-offsets_32.c
@@ -116,12 +116,14 @@ void foo(void)
 
 #ifdef CONFIG_PARAVIRT
 	BLANK();
-	OFFSET(PARAVIRT_enabled, paravirt_ops, paravirt_enabled);
-	OFFSET(PARAVIRT_irq_disable, paravirt_ops, irq_disable);
-	OFFSET(PARAVIRT_irq_enable, paravirt_ops, irq_enable);
-	OFFSET(PARAVIRT_irq_enable_sysexit, paravirt_ops, irq_enable_sysexit);
-	OFFSET(PARAVIRT_iret, paravirt_ops, iret);
-	OFFSET(PARAVIRT_read_cr0, paravirt_ops, read_cr0);
+	OFFSET(PARAVIRT_enabled, pv_info, paravirt_enabled);
+	OFFSET(PARAVIRT_PATCH_pv_cpu_ops, paravirt_patch_template, pv_cpu_ops);
+	OFFSET(PARAVIRT_PATCH_pv_irq_ops, paravirt_patch_template, pv_irq_ops);
+	OFFSET(PV_IRQ_irq_disable, pv_irq_ops, irq_disable);
+	OFFSET(PV_IRQ_irq_enable, pv_irq_ops, irq_enable);
+	OFFSET(PV_CPU_iret, pv_cpu_ops, iret);
+	OFFSET(PV_CPU_irq_enable_sysexit, pv_cpu_ops, irq_enable_sysexit);
+	OFFSET(PV_CPU_read_cr0, pv_cpu_ops, read_cr0);
 #endif
 
 #ifdef CONFIG_XEN
diff --git a/arch/x86/kernel/entry_32.S b/arch/x86/kernel/entry_32.S
index 290b7bc82da..1f2062e94d8 100644
--- a/arch/x86/kernel/entry_32.S
+++ b/arch/x86/kernel/entry_32.S
@@ -434,7 +434,7 @@ ldt_ss:
 	 * is still available to implement the setting of the high
 	 * 16-bits in the INTERRUPT_RETURN paravirt-op.
 	 */
-	cmpl $0, paravirt_ops+PARAVIRT_enabled
+	cmpl $0, pv_info+PARAVIRT_enabled
 	jne restore_nocheck
 #endif
 
diff --git a/arch/x86/kernel/paravirt_32.c b/arch/x86/kernel/paravirt_32.c
index 739cfb207dd..fa412515af7 100644
--- a/arch/x86/kernel/paravirt_32.c
+++ b/arch/x86/kernel/paravirt_32.c
@@ -42,32 +42,33 @@ void _paravirt_nop(void)
 static void __init default_banner(void)
 {
 	printk(KERN_INFO "Booting paravirtualized kernel on %s\n",
-	       paravirt_ops.name);
+	       pv_info.name);
 }
 
 char *memory_setup(void)
 {
-	return paravirt_ops.memory_setup();
+	return pv_init_ops.memory_setup();
 }
 
 /* Simple instruction patching code. */
-#define DEF_NATIVE(name, code)					\
-	extern const char start_##name[], end_##name[];		\
-	asm("start_" #name ": " code "; end_" #name ":")
-
-DEF_NATIVE(irq_disable, "cli");
-DEF_NATIVE(irq_enable, "sti");
-DEF_NATIVE(restore_fl, "push %eax; popf");
-DEF_NATIVE(save_fl, "pushf; pop %eax");
-DEF_NATIVE(iret, "iret");
-DEF_NATIVE(irq_enable_sysexit, "sti; sysexit");
-DEF_NATIVE(read_cr2, "mov %cr2, %eax");
-DEF_NATIVE(write_cr3, "mov %eax, %cr3");
-DEF_NATIVE(read_cr3, "mov %cr3, %eax");
-DEF_NATIVE(clts, "clts");
-DEF_NATIVE(read_tsc, "rdtsc");
-
-DEF_NATIVE(ud2a, "ud2a");
+#define DEF_NATIVE(ops, name, code)					\
+	extern const char start_##ops##_##name[], end_##ops##_##name[];	\
+	asm("start_" #ops "_" #name ": " code "; end_" #ops "_" #name ":")
+
+DEF_NATIVE(pv_irq_ops, irq_disable, "cli");
+DEF_NATIVE(pv_irq_ops, irq_enable, "sti");
+DEF_NATIVE(pv_irq_ops, restore_fl, "push %eax; popf");
+DEF_NATIVE(pv_irq_ops, save_fl, "pushf; pop %eax");
+DEF_NATIVE(pv_cpu_ops, iret, "iret");
+DEF_NATIVE(pv_cpu_ops, irq_enable_sysexit, "sti; sysexit");
+DEF_NATIVE(pv_mmu_ops, read_cr2, "mov %cr2, %eax");
+DEF_NATIVE(pv_mmu_ops, write_cr3, "mov %eax, %cr3");
+DEF_NATIVE(pv_mmu_ops, read_cr3, "mov %cr3, %eax");
+DEF_NATIVE(pv_cpu_ops, clts, "clts");
+DEF_NATIVE(pv_cpu_ops, read_tsc, "rdtsc");
+
+/* Undefined instruction for dealing with missing ops pointers. */
+static const unsigned char ud2a[] = { 0x0f, 0x0b };
 
 static unsigned native_patch(u8 type, u16 clobbers, void *ibuf,
 			     unsigned long addr, unsigned len)
@@ -76,37 +77,29 @@ static unsigned native_patch(u8 type, u16 clobbers, void *ibuf,
 	unsigned ret;
 
 	switch(type) {
-#define SITE(x)	case PARAVIRT_PATCH(x):	start = start_##x; end = end_##x; goto patch_site
-		SITE(irq_disable);
-		SITE(irq_enable);
-		SITE(restore_fl);
-		SITE(save_fl);
-		SITE(iret);
-		SITE(irq_enable_sysexit);
-		SITE(read_cr2);
-		SITE(read_cr3);
-		SITE(write_cr3);
-		SITE(clts);
-		SITE(read_tsc);
+#define SITE(ops, x)						\
+	case PARAVIRT_PATCH(ops.x):				\
+		start = start_##ops##_##x;			\
+		end = end_##ops##_##x;				\
+		goto patch_site
+
+	SITE(pv_irq_ops, irq_disable);
+	SITE(pv_irq_ops, irq_enable);
+	SITE(pv_irq_ops, restore_fl);
+	SITE(pv_irq_ops, save_fl);
+	SITE(pv_cpu_ops, iret);
+	SITE(pv_cpu_ops, irq_enable_sysexit);
+	SITE(pv_mmu_ops, read_cr2);
+	SITE(pv_mmu_ops, read_cr3);
+	SITE(pv_mmu_ops, write_cr3);
+	SITE(pv_cpu_ops, clts);
+	SITE(pv_cpu_ops, read_tsc);
 #undef SITE
 
 	patch_site:
 		ret = paravirt_patch_insns(ibuf, len, start, end);
 		break;
 
-	case PARAVIRT_PATCH(make_pgd):
-	case PARAVIRT_PATCH(make_pte):
-	case PARAVIRT_PATCH(pgd_val):
-	case PARAVIRT_PATCH(pte_val):
-#ifdef CONFIG_X86_PAE
-	case PARAVIRT_PATCH(make_pmd):
-	case PARAVIRT_PATCH(pmd_val):
-#endif
-		/* These functions end up returning exactly what
-		   they're passed, in the same registers. */
-		ret = paravirt_patch_nop();
-		break;
-
 	default:
 		ret = paravirt_patch_default(type, clobbers, ibuf, addr, len);
 		break;
@@ -150,7 +143,7 @@ unsigned paravirt_patch_call(void *insnbuf,
 	return 5;
 }
 
-unsigned paravirt_patch_jmp(const void *target, void *insnbuf,
+unsigned paravirt_patch_jmp(void *insnbuf, const void *target,
 			    unsigned long addr, unsigned len)
 {
 	struct branch *b = insnbuf;
@@ -165,22 +158,38 @@ unsigned paravirt_patch_jmp(const void *target, void *insnbuf,
 	return 5;
 }
 
+/* Neat trick to map patch type back to the call within the
+ * corresponding structure. */
+static void *get_call_destination(u8 type)
+{
+	struct paravirt_patch_template tmpl = {
+		.pv_init_ops = pv_init_ops,
+		.pv_misc_ops = pv_misc_ops,
+		.pv_time_ops = pv_time_ops,
+		.pv_cpu_ops = pv_cpu_ops,
+		.pv_irq_ops = pv_irq_ops,
+		.pv_apic_ops = pv_apic_ops,
+		.pv_mmu_ops = pv_mmu_ops,
+	};
+	return *((void **)&tmpl + type);
+}
+
 unsigned paravirt_patch_default(u8 type, u16 clobbers, void *insnbuf,
 				unsigned long addr, unsigned len)
 {
-	void *opfunc = *((void **)&paravirt_ops + type);
+	void *opfunc = get_call_destination(type);
 	unsigned ret;
 
 	if (opfunc == NULL)
 		/* If there's no function, patch it with a ud2a (BUG) */
-		ret = paravirt_patch_insns(insnbuf, len, start_ud2a, end_ud2a);
+		ret = paravirt_patch_insns(insnbuf, len, ud2a, ud2a+sizeof(ud2a));
 	else if (opfunc == paravirt_nop)
 		/* If the operation is a nop, then nop the callsite */
 		ret = paravirt_patch_nop();
-	else if (type == PARAVIRT_PATCH(iret) ||
-		 type == PARAVIRT_PATCH(irq_enable_sysexit))
+	else if (type == PARAVIRT_PATCH(pv_cpu_ops.iret) ||
+		 type == PARAVIRT_PATCH(pv_cpu_ops.irq_enable_sysexit))
 		/* If operation requires a jmp, then jmp */
-		ret = paravirt_patch_jmp(opfunc, insnbuf, addr, len);
+		ret = paravirt_patch_jmp(insnbuf, opfunc, addr, len);
 	else
 		/* Otherwise call the function; assume target could
 		   clobber any caller-save reg */
@@ -205,7 +214,7 @@ unsigned paravirt_patch_insns(void *insnbuf, unsigned len,
 
 void init_IRQ(void)
 {
-	paravirt_ops.init_IRQ();
+	pv_irq_ops.init_IRQ();
 }
 
 static void native_flush_tlb(void)
@@ -233,7 +242,7 @@ extern void native_irq_enable_sysexit(void);
 
 static int __init print_banner(void)
 {
-	paravirt_ops.banner();
+	pv_init_ops.banner();
 	return 0;
 }
 core_initcall(print_banner);
@@ -273,47 +282,53 @@ int paravirt_disable_iospace(void)
 	return ret;
 }
 
-struct paravirt_ops paravirt_ops = {
+struct pv_info pv_info = {
 	.name = "bare hardware",
 	.paravirt_enabled = 0,
 	.kernel_rpl = 0,
 	.shared_kernel_pmd = 1,	/* Only used when CONFIG_X86_PAE is set */
+};
 
- 	.patch = native_patch,
+struct pv_init_ops pv_init_ops = {
+	.patch = native_patch,
 	.banner = default_banner,
 	.arch_setup = paravirt_nop,
 	.memory_setup = machine_specific_memory_setup,
+};
+
+struct pv_time_ops pv_time_ops = {
+	.time_init = hpet_time_init,
 	.get_wallclock = native_get_wallclock,
 	.set_wallclock = native_set_wallclock,
-	.time_init = hpet_time_init,
+	.sched_clock = native_sched_clock,
+	.get_cpu_khz = native_calculate_cpu_khz,
+};
+
+struct pv_irq_ops pv_irq_ops = {
 	.init_IRQ = native_init_IRQ,
+	.save_fl = native_save_fl,
+	.restore_fl = native_restore_fl,
+	.irq_disable = native_irq_disable,
+	.irq_enable = native_irq_enable,
+	.safe_halt = native_safe_halt,
+	.halt = native_halt,
+};
 
+struct pv_cpu_ops pv_cpu_ops = {
 	.cpuid = native_cpuid,
 	.get_debugreg = native_get_debugreg,
 	.set_debugreg = native_set_debugreg,
 	.clts = native_clts,
 	.read_cr0 = native_read_cr0,
 	.write_cr0 = native_write_cr0,
-	.read_cr2 = native_read_cr2,
-	.write_cr2 = native_write_cr2,
-	.read_cr3 = native_read_cr3,
-	.write_cr3 = native_write_cr3,
 	.read_cr4 = native_read_cr4,
 	.read_cr4_safe = native_read_cr4_safe,
 	.write_cr4 = native_write_cr4,
-	.save_fl = native_save_fl,
-	.restore_fl = native_restore_fl,
-	.irq_disable = native_irq_disable,
-	.irq_enable = native_irq_enable,
-	.safe_halt = native_safe_halt,
-	.halt = native_halt,
 	.wbinvd = native_wbinvd,
 	.read_msr = native_read_msr_safe,
 	.write_msr = native_write_msr_safe,
 	.read_tsc = native_read_tsc,
 	.read_pmc = native_read_pmc,
-	.sched_clock = native_sched_clock,
-	.get_cpu_khz = native_calculate_cpu_khz,
 	.load_tr_desc = native_load_tr_desc,
 	.set_ldt = native_set_ldt,
 	.load_gdt = native_load_gdt,
@@ -327,9 +342,14 @@ struct paravirt_ops paravirt_ops = {
 	.write_idt_entry = write_dt_entry,
 	.load_esp0 = native_load_esp0,
 
+	.irq_enable_sysexit = native_irq_enable_sysexit,
+	.iret = native_iret,
+
 	.set_iopl_mask = native_set_iopl_mask,
 	.io_delay = native_io_delay,
+};
 
+struct pv_apic_ops pv_apic_ops = {
 #ifdef CONFIG_X86_LOCAL_APIC
 	.apic_write = native_apic_write,
 	.apic_write_atomic = native_apic_write_atomic,
@@ -338,11 +358,21 @@ struct paravirt_ops paravirt_ops = {
 	.setup_secondary_clock = setup_secondary_APIC_clock,
 	.startup_ipi_hook = paravirt_nop,
 #endif
+};
+
+struct pv_misc_ops pv_misc_ops = {
 	.set_lazy_mode = paravirt_nop,
+};
 
+struct pv_mmu_ops pv_mmu_ops = {
 	.pagetable_setup_start = native_pagetable_setup_start,
 	.pagetable_setup_done = native_pagetable_setup_done,
 
+	.read_cr2 = native_read_cr2,
+	.write_cr2 = native_write_cr2,
+	.read_cr3 = native_read_cr3,
+	.write_cr3 = native_write_cr3,
+
 	.flush_tlb_user = native_flush_tlb,
 	.flush_tlb_kernel = native_flush_tlb_global,
 	.flush_tlb_single = native_flush_tlb_single,
@@ -381,12 +411,14 @@ struct paravirt_ops paravirt_ops = {
 	.make_pte = native_make_pte,
 	.make_pgd = native_make_pgd,
 
-	.irq_enable_sysexit = native_irq_enable_sysexit,
-	.iret = native_iret,
-
 	.dup_mmap = paravirt_nop,
 	.exit_mmap = paravirt_nop,
 	.activate_mm = paravirt_nop,
 };
 
-EXPORT_SYMBOL(paravirt_ops);
+EXPORT_SYMBOL_GPL(pv_time_ops);
+EXPORT_SYMBOL_GPL(pv_cpu_ops);
+EXPORT_SYMBOL_GPL(pv_mmu_ops);
+EXPORT_SYMBOL_GPL(pv_apic_ops);
+EXPORT_SYMBOL_GPL(pv_info);
+EXPORT_SYMBOL    (pv_irq_ops);
diff --git a/arch/x86/kernel/vmi_32.c b/arch/x86/kernel/vmi_32.c
index 18673e0f193..67cea5c2e3e 100644
--- a/arch/x86/kernel/vmi_32.c
+++ b/arch/x86/kernel/vmi_32.c
@@ -134,21 +134,21 @@ static unsigned vmi_patch(u8 type, u16 clobbers, void *insns,
 			  unsigned long eip, unsigned len)
 {
 	switch (type) {
-		case PARAVIRT_PATCH(irq_disable):
+		case PARAVIRT_PATCH(pv_irq_ops.irq_disable):
 			return patch_internal(VMI_CALL_DisableInterrupts, len,
 					      insns, eip);
-		case PARAVIRT_PATCH(irq_enable):
+		case PARAVIRT_PATCH(pv_irq_ops.irq_enable):
 			return patch_internal(VMI_CALL_EnableInterrupts, len,
 					      insns, eip);
-		case PARAVIRT_PATCH(restore_fl):
+		case PARAVIRT_PATCH(pv_irq_ops.restore_fl):
 			return patch_internal(VMI_CALL_SetInterruptMask, len,
 					      insns, eip);
-		case PARAVIRT_PATCH(save_fl):
+		case PARAVIRT_PATCH(pv_irq_ops.save_fl):
 			return patch_internal(VMI_CALL_GetInterruptMask, len,
 					      insns, eip);
-		case PARAVIRT_PATCH(iret):
+		case PARAVIRT_PATCH(pv_cpu_ops.iret):
 			return patch_internal(VMI_CALL_IRET, len, insns, eip);
-		case PARAVIRT_PATCH(irq_enable_sysexit):
+		case PARAVIRT_PATCH(pv_cpu_ops.irq_enable_sysexit):
 			return patch_internal(VMI_CALL_SYSEXIT, len, insns, eip);
 		default:
 			break;
@@ -690,9 +690,9 @@ do {								\
 	reloc = call_vrom_long_func(vmi_rom, get_reloc,		\
 				    VMI_CALL_##vmicall);	\
 	if (rel->type == VMI_RELOCATION_CALL_REL) 		\
-		paravirt_ops.opname = (void *)rel->eip;		\
+		opname = (void *)rel->eip;			\
 	else if (rel->type == VMI_RELOCATION_NOP) 		\
-		paravirt_ops.opname = (void *)vmi_nop;		\
+		opname = (void *)vmi_nop;			\
 	else if (rel->type != VMI_RELOCATION_NONE)		\
 		printk(KERN_WARNING "VMI: Unknown relocation "	\
 				    "type %d for " #vmicall"\n",\
@@ -712,7 +712,7 @@ do {								\
 				    VMI_CALL_##vmicall);	\
 	BUG_ON(rel->type == VMI_RELOCATION_JUMP_REL);		\
 	if (rel->type == VMI_RELOCATION_CALL_REL) {		\
-		paravirt_ops.opname = wrapper;			\
+		opname = wrapper;				\
 		vmi_ops.cache = (void *)rel->eip;		\
 	}							\
 } while (0)
@@ -732,11 +732,11 @@ static inline int __init activate_vmi(void)
 	}
 	savesegment(cs, kernel_cs);
 
-	paravirt_ops.paravirt_enabled = 1;
-	paravirt_ops.kernel_rpl = kernel_cs & SEGMENT_RPL_MASK;
+	pv_info.paravirt_enabled = 1;
+	pv_info.kernel_rpl = kernel_cs & SEGMENT_RPL_MASK;
+	pv_info.name = "vmi";
 
-	paravirt_ops.patch = vmi_patch;
-	paravirt_ops.name = "vmi";
+	pv_init_ops.patch = vmi_patch;
 
 	/*
 	 * Many of these operations are ABI compatible with VMI.
@@ -754,26 +754,26 @@ static inline int __init activate_vmi(void)
 	 */
 
 	/* CPUID is special, so very special it gets wrapped like a present */
-	para_wrap(cpuid, vmi_cpuid, cpuid, CPUID);
-
-	para_fill(clts, CLTS);
-	para_fill(get_debugreg, GetDR);
-	para_fill(set_debugreg, SetDR);
-	para_fill(read_cr0, GetCR0);
-	para_fill(read_cr2, GetCR2);
-	para_fill(read_cr3, GetCR3);
-	para_fill(read_cr4, GetCR4);
-	para_fill(write_cr0, SetCR0);
-	para_fill(write_cr2, SetCR2);
-	para_fill(write_cr3, SetCR3);
-	para_fill(write_cr4, SetCR4);
-	para_fill(save_fl, GetInterruptMask);
-	para_fill(restore_fl, SetInterruptMask);
-	para_fill(irq_disable, DisableInterrupts);
-	para_fill(irq_enable, EnableInterrupts);
-
-	para_fill(wbinvd, WBINVD);
-	para_fill(read_tsc, RDTSC);
+	para_wrap(pv_cpu_ops.cpuid, vmi_cpuid, cpuid, CPUID);
+
+	para_fill(pv_cpu_ops.clts, CLTS);
+	para_fill(pv_cpu_ops.get_debugreg, GetDR);
+	para_fill(pv_cpu_ops.set_debugreg, SetDR);
+	para_fill(pv_cpu_ops.read_cr0, GetCR0);
+	para_fill(pv_mmu_ops.read_cr2, GetCR2);
+	para_fill(pv_mmu_ops.read_cr3, GetCR3);
+	para_fill(pv_cpu_ops.read_cr4, GetCR4);
+	para_fill(pv_cpu_ops.write_cr0, SetCR0);
+	para_fill(pv_mmu_ops.write_cr2, SetCR2);
+	para_fill(pv_mmu_ops.write_cr3, SetCR3);
+	para_fill(pv_cpu_ops.write_cr4, SetCR4);
+	para_fill(pv_irq_ops.save_fl, GetInterruptMask);
+	para_fill(pv_irq_ops.restore_fl, SetInterruptMask);
+	para_fill(pv_irq_ops.irq_disable, DisableInterrupts);
+	para_fill(pv_irq_ops.irq_enable, EnableInterrupts);
+
+	para_fill(pv_cpu_ops.wbinvd, WBINVD);
+	para_fill(pv_cpu_ops.read_tsc, RDTSC);
 
 	/* The following we emulate with trap and emulate for now */
 	/* paravirt_ops.read_msr = vmi_rdmsr */
@@ -781,29 +781,29 @@ static inline int __init activate_vmi(void)
 	/* paravirt_ops.rdpmc = vmi_rdpmc */
 
 	/* TR interface doesn't pass TR value, wrap */
-	para_wrap(load_tr_desc, vmi_set_tr, set_tr, SetTR);
+	para_wrap(pv_cpu_ops.load_tr_desc, vmi_set_tr, set_tr, SetTR);
 
 	/* LDT is special, too */
-	para_wrap(set_ldt, vmi_set_ldt, _set_ldt, SetLDT);
-
-	para_fill(load_gdt, SetGDT);
-	para_fill(load_idt, SetIDT);
-	para_fill(store_gdt, GetGDT);
-	para_fill(store_idt, GetIDT);
-	para_fill(store_tr, GetTR);
-	paravirt_ops.load_tls = vmi_load_tls;
-	para_fill(write_ldt_entry, WriteLDTEntry);
-	para_fill(write_gdt_entry, WriteGDTEntry);
-	para_fill(write_idt_entry, WriteIDTEntry);
-	para_wrap(load_esp0, vmi_load_esp0, set_kernel_stack, UpdateKernelStack);
-	para_fill(set_iopl_mask, SetIOPLMask);
-	para_fill(io_delay, IODelay);
-	para_wrap(set_lazy_mode, vmi_set_lazy_mode, set_lazy_mode, SetLazyMode);
+	para_wrap(pv_cpu_ops.set_ldt, vmi_set_ldt, _set_ldt, SetLDT);
+
+	para_fill(pv_cpu_ops.load_gdt, SetGDT);
+	para_fill(pv_cpu_ops.load_idt, SetIDT);
+	para_fill(pv_cpu_ops.store_gdt, GetGDT);
+	para_fill(pv_cpu_ops.store_idt, GetIDT);
+	para_fill(pv_cpu_ops.store_tr, GetTR);
+	pv_cpu_ops.load_tls = vmi_load_tls;
+	para_fill(pv_cpu_ops.write_ldt_entry, WriteLDTEntry);
+	para_fill(pv_cpu_ops.write_gdt_entry, WriteGDTEntry);
+	para_fill(pv_cpu_ops.write_idt_entry, WriteIDTEntry);
+	para_wrap(pv_cpu_ops.load_esp0, vmi_load_esp0, set_kernel_stack, UpdateKernelStack);
+	para_fill(pv_cpu_ops.set_iopl_mask, SetIOPLMask);
+	para_fill(pv_cpu_ops.io_delay, IODelay);
+	para_wrap(pv_misc_ops.set_lazy_mode, vmi_set_lazy_mode, set_lazy_mode, SetLazyMode);
 
 	/* user and kernel flush are just handled with different flags to FlushTLB */
-	para_wrap(flush_tlb_user, vmi_flush_tlb_user, _flush_tlb, FlushTLB);
-	para_wrap(flush_tlb_kernel, vmi_flush_tlb_kernel, _flush_tlb, FlushTLB);
-	para_fill(flush_tlb_single, InvalPage);
+	para_wrap(pv_mmu_ops.flush_tlb_user, vmi_flush_tlb_user, _flush_tlb, FlushTLB);
+	para_wrap(pv_mmu_ops.flush_tlb_kernel, vmi_flush_tlb_kernel, _flush_tlb, FlushTLB);
+	para_fill(pv_mmu_ops.flush_tlb_single, InvalPage);
 
 	/*
 	 * Until a standard flag format can be agreed on, we need to
@@ -819,41 +819,41 @@ static inline int __init activate_vmi(void)
 #endif
 
 	if (vmi_ops.set_pte) {
-		paravirt_ops.set_pte = vmi_set_pte;
-		paravirt_ops.set_pte_at = vmi_set_pte_at;
-		paravirt_ops.set_pmd = vmi_set_pmd;
+		pv_mmu_ops.set_pte = vmi_set_pte;
+		pv_mmu_ops.set_pte_at = vmi_set_pte_at;
+		pv_mmu_ops.set_pmd = vmi_set_pmd;
 #ifdef CONFIG_X86_PAE
-		paravirt_ops.set_pte_atomic = vmi_set_pte_atomic;
-		paravirt_ops.set_pte_present = vmi_set_pte_present;
-		paravirt_ops.set_pud = vmi_set_pud;
-		paravirt_ops.pte_clear = vmi_pte_clear;
-		paravirt_ops.pmd_clear = vmi_pmd_clear;
+		pv_mmu_ops.set_pte_atomic = vmi_set_pte_atomic;
+		pv_mmu_ops.set_pte_present = vmi_set_pte_present;
+		pv_mmu_ops.set_pud = vmi_set_pud;
+		pv_mmu_ops.pte_clear = vmi_pte_clear;
+		pv_mmu_ops.pmd_clear = vmi_pmd_clear;
 #endif
 	}
 
 	if (vmi_ops.update_pte) {
-		paravirt_ops.pte_update = vmi_update_pte;
-		paravirt_ops.pte_update_defer = vmi_update_pte_defer;
+		pv_mmu_ops.pte_update = vmi_update_pte;
+		pv_mmu_ops.pte_update_defer = vmi_update_pte_defer;
 	}
 
 	vmi_ops.allocate_page = vmi_get_function(VMI_CALL_AllocatePage);
 	if (vmi_ops.allocate_page) {
-		paravirt_ops.alloc_pt = vmi_allocate_pt;
-		paravirt_ops.alloc_pd = vmi_allocate_pd;
-		paravirt_ops.alloc_pd_clone = vmi_allocate_pd_clone;
+		pv_mmu_ops.alloc_pt = vmi_allocate_pt;
+		pv_mmu_ops.alloc_pd = vmi_allocate_pd;
+		pv_mmu_ops.alloc_pd_clone = vmi_allocate_pd_clone;
 	}
 
 	vmi_ops.release_page = vmi_get_function(VMI_CALL_ReleasePage);
 	if (vmi_ops.release_page) {
-		paravirt_ops.release_pt = vmi_release_pt;
-		paravirt_ops.release_pd = vmi_release_pd;
+		pv_mmu_ops.release_pt = vmi_release_pt;
+		pv_mmu_ops.release_pd = vmi_release_pd;
 	}
 
 	/* Set linear is needed in all cases */
 	vmi_ops.set_linear_mapping = vmi_get_function(VMI_CALL_SetLinearMapping);
 #ifdef CONFIG_HIGHPTE
 	if (vmi_ops.set_linear_mapping)
-		paravirt_ops.kmap_atomic_pte = vmi_kmap_atomic_pte;
+		pv_mmu_ops.kmap_atomic_pte = vmi_kmap_atomic_pte;
 #endif
 
 	/*
@@ -863,17 +863,17 @@ static inline int __init activate_vmi(void)
 	 * the backend.  They are performance critical anyway, so requiring
 	 * a patch is not a big problem.
 	 */
-	paravirt_ops.irq_enable_sysexit = (void *)0xfeedbab0;
-	paravirt_ops.iret = (void *)0xbadbab0;
+	pv_cpu_ops.irq_enable_sysexit = (void *)0xfeedbab0;
+	pv_cpu_ops.iret = (void *)0xbadbab0;
 
 #ifdef CONFIG_SMP
-	para_wrap(startup_ipi_hook, vmi_startup_ipi_hook, set_initial_ap_state, SetInitialAPState);
+	para_wrap(pv_apic_ops.startup_ipi_hook, vmi_startup_ipi_hook, set_initial_ap_state, SetInitialAPState);
 #endif
 
 #ifdef CONFIG_X86_LOCAL_APIC
-	para_fill(apic_read, APICRead);
-	para_fill(apic_write, APICWrite);
-	para_fill(apic_write_atomic, APICWrite);
+	para_fill(pv_apic_ops.apic_read, APICRead);
+	para_fill(pv_apic_ops.apic_write, APICWrite);
+	para_fill(pv_apic_ops.apic_write_atomic, APICWrite);
 #endif
 
 	/*
@@ -891,15 +891,15 @@ static inline int __init activate_vmi(void)
 		vmi_timer_ops.set_alarm = vmi_get_function(VMI_CALL_SetAlarm);
 		vmi_timer_ops.cancel_alarm =
 			 vmi_get_function(VMI_CALL_CancelAlarm);
-		paravirt_ops.time_init = vmi_time_init;
-		paravirt_ops.get_wallclock = vmi_get_wallclock;
-		paravirt_ops.set_wallclock = vmi_set_wallclock;
+		pv_time_ops.time_init = vmi_time_init;
+		pv_time_ops.get_wallclock = vmi_get_wallclock;
+		pv_time_ops.set_wallclock = vmi_set_wallclock;
 #ifdef CONFIG_X86_LOCAL_APIC
-		paravirt_ops.setup_boot_clock = vmi_time_bsp_init;
-		paravirt_ops.setup_secondary_clock = vmi_time_ap_init;
+		pv_apic_ops.setup_boot_clock = vmi_time_bsp_init;
+		pv_apic_ops.setup_secondary_clock = vmi_time_ap_init;
 #endif
-		paravirt_ops.sched_clock = vmi_sched_clock;
- 		paravirt_ops.get_cpu_khz = vmi_cpu_khz;
+		pv_time_ops.sched_clock = vmi_sched_clock;
+ 		pv_time_ops.get_cpu_khz = vmi_cpu_khz;
 
 		/* We have true wallclock functions; disable CMOS clock sync */
 		no_sync_cmos_clock = 1;
@@ -908,7 +908,7 @@ static inline int __init activate_vmi(void)
 		disable_vmi_timer = 1;
 	}
 
-	para_fill(safe_halt, Halt);
+	para_fill(pv_irq_ops.safe_halt, Halt);
 
 	/*
 	 * Alternative instruction rewriting doesn't happen soon enough
diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index f01bfcd4bde..3d3bf05dec7 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -124,7 +124,7 @@ static void __init xen_vcpu_setup(int cpu)
 static void __init xen_banner(void)
 {
 	printk(KERN_INFO "Booting paravirtualized kernel on %s\n",
-	       paravirt_ops.name);
+	       pv_info.name);
 	printk(KERN_INFO "Hypervisor signature: %s\n", xen_start_info->magic);
 }
 
@@ -738,7 +738,7 @@ static __init void xen_pagetable_setup_start(pgd_t *base)
 	pgd_t *xen_pgd = (pgd_t *)xen_start_info->pt_base;
 
 	/* special set_pte for pagetable initialization */
-	paravirt_ops.set_pte = xen_set_pte_init;
+	pv_mmu_ops.set_pte = xen_set_pte_init;
 
 	init_mm.pgd = base;
 	/*
@@ -785,8 +785,8 @@ static __init void xen_pagetable_setup_done(pgd_t *base)
 {
 	/* This will work as long as patching hasn't happened yet
 	   (which it hasn't) */
-	paravirt_ops.alloc_pt = xen_alloc_pt;
-	paravirt_ops.set_pte = xen_set_pte;
+	pv_mmu_ops.alloc_pt = xen_alloc_pt;
+	pv_mmu_ops.set_pte = xen_set_pte;
 
 	if (!xen_feature(XENFEAT_auto_translated_physmap)) {
 		/*
@@ -833,12 +833,12 @@ void __init xen_setup_vcpu_info_placement(void)
 	if (have_vcpu_info_placement) {
 		printk(KERN_INFO "Xen: using vcpu_info placement\n");
 
-		paravirt_ops.save_fl = xen_save_fl_direct;
-		paravirt_ops.restore_fl = xen_restore_fl_direct;
-		paravirt_ops.irq_disable = xen_irq_disable_direct;
-		paravirt_ops.irq_enable = xen_irq_enable_direct;
-		paravirt_ops.read_cr2 = xen_read_cr2_direct;
-		paravirt_ops.iret = xen_iret_direct;
+		pv_irq_ops.save_fl = xen_save_fl_direct;
+		pv_irq_ops.restore_fl = xen_restore_fl_direct;
+		pv_irq_ops.irq_disable = xen_irq_disable_direct;
+		pv_irq_ops.irq_enable = xen_irq_enable_direct;
+		pv_mmu_ops.read_cr2 = xen_read_cr2_direct;
+		pv_cpu_ops.iret = xen_iret_direct;
 	}
 }
 
@@ -850,8 +850,8 @@ static unsigned xen_patch(u8 type, u16 clobbers, void *insnbuf,
 
 	start = end = reloc = NULL;
 
-#define SITE(x)								\
-	case PARAVIRT_PATCH(x):						\
+#define SITE(op, x)							\
+	case PARAVIRT_PATCH(op.x):					\
 	if (have_vcpu_info_placement) {					\
 		start = (char *)xen_##x##_direct;			\
 		end = xen_##x##_direct_end;				\
@@ -860,10 +860,10 @@ static unsigned xen_patch(u8 type, u16 clobbers, void *insnbuf,
 	goto patch_site
 
 	switch (type) {
-		SITE(irq_enable);
-		SITE(irq_disable);
-		SITE(save_fl);
-		SITE(restore_fl);
+		SITE(pv_irq_ops, irq_enable);
+		SITE(pv_irq_ops, irq_disable);
+		SITE(pv_irq_ops, save_fl);
+		SITE(pv_irq_ops, restore_fl);
 #undef SITE
 
 	patch_site:
@@ -895,26 +895,32 @@ static unsigned xen_patch(u8 type, u16 clobbers, void *insnbuf,
 	return ret;
 }
 
-static const struct paravirt_ops xen_paravirt_ops __initdata = {
+static const struct pv_info xen_info __initdata = {
 	.paravirt_enabled = 1,
 	.shared_kernel_pmd = 0,
 
 	.name = "Xen",
-	.banner = xen_banner,
+};
 
+static const struct pv_init_ops xen_init_ops __initdata = {
 	.patch = xen_patch,
 
+	.banner = xen_banner,
 	.memory_setup = xen_memory_setup,
 	.arch_setup = xen_arch_setup,
-	.init_IRQ = xen_init_IRQ,
 	.post_allocator_init = xen_mark_init_mm_pinned,
+};
 
+static const struct pv_time_ops xen_time_ops __initdata = {
 	.time_init = xen_time_init,
+
 	.set_wallclock = xen_set_wallclock,
 	.get_wallclock = xen_get_wallclock,
 	.get_cpu_khz = xen_cpu_khz,
 	.sched_clock = xen_sched_clock,
+};
 
+static const struct pv_cpu_ops xen_cpu_ops __initdata = {
 	.cpuid = xen_cpuid,
 
 	.set_debugreg = xen_set_debugreg,
@@ -925,22 +931,10 @@ static const struct paravirt_ops xen_paravirt_ops __initdata = {
 	.read_cr0 = native_read_cr0,
 	.write_cr0 = native_write_cr0,
 
-	.read_cr2 = xen_read_cr2,
-	.write_cr2 = xen_write_cr2,
-
-	.read_cr3 = xen_read_cr3,
-	.write_cr3 = xen_write_cr3,
-
 	.read_cr4 = native_read_cr4,
 	.read_cr4_safe = native_read_cr4_safe,
 	.write_cr4 = xen_write_cr4,
 
-	.save_fl = xen_save_fl,
-	.restore_fl = xen_restore_fl,
-	.irq_disable = xen_irq_disable,
-	.irq_enable = xen_irq_enable,
-	.safe_halt = xen_safe_halt,
-	.halt = xen_halt,
 	.wbinvd = native_wbinvd,
 
 	.read_msr = native_read_msr_safe,
@@ -968,7 +962,19 @@ static const struct paravirt_ops xen_paravirt_ops __initdata = {
 
 	.set_iopl_mask = xen_set_iopl_mask,
 	.io_delay = xen_io_delay,
+};
+
+static const struct pv_irq_ops xen_irq_ops __initdata = {
+	.init_IRQ = xen_init_IRQ,
+	.save_fl = xen_save_fl,
+	.restore_fl = xen_restore_fl,
+	.irq_disable = xen_irq_disable,
+	.irq_enable = xen_irq_enable,
+	.safe_halt = xen_safe_halt,
+	.halt = xen_halt,
+};
 
+static const struct pv_apic_ops xen_apic_ops __initdata = {
 #ifdef CONFIG_X86_LOCAL_APIC
 	.apic_write = xen_apic_write,
 	.apic_write_atomic = xen_apic_write,
@@ -977,6 +983,17 @@ static const struct paravirt_ops xen_paravirt_ops __initdata = {
 	.setup_secondary_clock = paravirt_nop,
 	.startup_ipi_hook = paravirt_nop,
 #endif
+};
+
+static const struct pv_mmu_ops xen_mmu_ops __initdata = {
+	.pagetable_setup_start = xen_pagetable_setup_start,
+	.pagetable_setup_done = xen_pagetable_setup_done,
+
+	.read_cr2 = xen_read_cr2,
+	.write_cr2 = xen_write_cr2,
+
+	.read_cr3 = xen_read_cr3,
+	.write_cr3 = xen_write_cr3,
 
 	.flush_tlb_user = xen_flush_tlb,
 	.flush_tlb_kernel = xen_flush_tlb,
@@ -986,9 +1003,6 @@ static const struct paravirt_ops xen_paravirt_ops __initdata = {
 	.pte_update = paravirt_nop,
 	.pte_update_defer = paravirt_nop,
 
-	.pagetable_setup_start = xen_pagetable_setup_start,
-	.pagetable_setup_done = xen_pagetable_setup_done,
-
 	.alloc_pt = xen_alloc_pt_init,
 	.release_pt = xen_release_pt,
 	.alloc_pd = paravirt_nop,
@@ -1023,7 +1037,9 @@ static const struct paravirt_ops xen_paravirt_ops __initdata = {
 	.activate_mm = xen_activate_mm,
 	.dup_mmap = xen_dup_mmap,
 	.exit_mmap = xen_exit_mmap,
+};
 
+static const struct pv_misc_ops xen_misc_ops __initdata = {
 	.set_lazy_mode = xen_set_lazy_mode,
 };
 
@@ -1091,7 +1107,15 @@ asmlinkage void __init xen_start_kernel(void)
 	BUG_ON(memcmp(xen_start_info->magic, "xen-3.0", 7) != 0);
 
 	/* Install Xen paravirt ops */
-	paravirt_ops = xen_paravirt_ops;
+	pv_info = xen_info;
+	pv_init_ops = xen_init_ops;
+	pv_time_ops = xen_time_ops;
+	pv_cpu_ops = xen_cpu_ops;
+	pv_irq_ops = xen_irq_ops;
+	pv_apic_ops = xen_apic_ops;
+	pv_mmu_ops = xen_mmu_ops;
+	pv_misc_ops = xen_misc_ops;
+
 	machine_ops = xen_machine_ops;
 
 #ifdef CONFIG_SMP
@@ -1124,9 +1148,9 @@ asmlinkage void __init xen_start_kernel(void)
 	xen_setup_vcpu_info_placement();
 #endif
 
-	paravirt_ops.kernel_rpl = 1;
+	pv_info.kernel_rpl = 1;
 	if (xen_feature(XENFEAT_supervisor_mode_kernel))
-		paravirt_ops.kernel_rpl = 0;
+		pv_info.kernel_rpl = 0;
 
 	/* set the limit of our address space */
 	reserve_top_address(-HYPERVISOR_VIRT_START + 2 * PAGE_SIZE);