Merge branch 'xen-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/jeremy/xen

* 'xen-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/jeremy/xen:
  xfs: eagerly remove vmap mappings to avoid upsetting Xen
  xen: add some debug output for failed multicalls
  xen: fix incorrect vcpu_register_vcpu_info hypercall argument
  xen: ask the hypervisor how much space it needs reserved
  xen: lock pte pages while pinning/unpinning
  xen: deal with stale cr3 values when unpinning pagetables
  xen: add batch completion callbacks
  xen: yield to IPI target if necessary
  Clean up duplicate includes in arch/i386/xen/
  remove dead code in pgtable_cache_init
  paravirt: clean up lazy mode handling
  paravirt: refactor struct paravirt_ops into smaller pv_*_ops

7 files changed, 324 insertions, 141 deletions

diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index 493a083f688..94c39aaf695 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -25,7 +25,6 @@
 #include <linux/mm.h>
 #include <linux/page-flags.h>
 #include <linux/highmem.h>
-#include <linux/smp.h>
 
 #include <xen/interface/xen.h>
 #include <xen/interface/physdev.h>
@@ -52,11 +51,25 @@
 
 EXPORT_SYMBOL_GPL(hypercall_page);
 
-DEFINE_PER_CPU(enum paravirt_lazy_mode, xen_lazy_mode);
-
 DEFINE_PER_CPU(struct vcpu_info *, xen_vcpu);
 DEFINE_PER_CPU(struct vcpu_info, xen_vcpu_info);
-DEFINE_PER_CPU(unsigned long, xen_cr3);
+
+/*
+ * Note about cr3 (pagetable base) values:
+ *
+ * xen_cr3 contains the current logical cr3 value; it contains the
+ * last set cr3.  This may not be the current effective cr3, because
+ * its update may be being lazily deferred.  However, a vcpu looking
+ * at its own cr3 can use this value knowing that it everything will
+ * be self-consistent.
+ *
+ * xen_current_cr3 contains the actual vcpu cr3; it is set once the
+ * hypercall to set the vcpu cr3 is complete (so it may be a little
+ * out of date, but it will never be set early).  If one vcpu is
+ * looking at another vcpu's cr3 value, it should use this variable.
+ */
+DEFINE_PER_CPU(unsigned long, xen_cr3);	 /* cr3 stored as physaddr */
+DEFINE_PER_CPU(unsigned long, xen_current_cr3);	 /* actual vcpu cr3 */
 
 struct start_info *xen_start_info;
 EXPORT_SYMBOL_GPL(xen_start_info);
@@ -100,7 +113,7 @@ static void __init xen_vcpu_setup(int cpu)
 	info.mfn = virt_to_mfn(vcpup);
 	info.offset = offset_in_page(vcpup);
 
-	printk(KERN_DEBUG "trying to map vcpu_info %d at %p, mfn %x, offset %d\n",
+	printk(KERN_DEBUG "trying to map vcpu_info %d at %p, mfn %llx, offset %d\n",
 	       cpu, vcpup, info.mfn, info.offset);
 
 	/* Check to see if the hypervisor will put the vcpu_info
@@ -124,7 +137,7 @@ 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);
 }
 
@@ -249,29 +262,10 @@ static void xen_halt(void)
 		xen_safe_halt();
 }
 
-static void xen_set_lazy_mode(enum paravirt_lazy_mode mode)
+static void xen_leave_lazy(void)
 {
-	BUG_ON(preemptible());
-
-	switch (mode) {
-	case PARAVIRT_LAZY_NONE:
-		BUG_ON(x86_read_percpu(xen_lazy_mode) == PARAVIRT_LAZY_NONE);
-		break;
-
-	case PARAVIRT_LAZY_MMU:
-	case PARAVIRT_LAZY_CPU:
-		BUG_ON(x86_read_percpu(xen_lazy_mode) != PARAVIRT_LAZY_NONE);
-		break;
-
-	case PARAVIRT_LAZY_FLUSH:
-		/* flush if necessary, but don't change state */
-		if (x86_read_percpu(xen_lazy_mode) != PARAVIRT_LAZY_NONE)
-			xen_mc_flush();
-		return;
-	}
-
+	paravirt_leave_lazy(paravirt_get_lazy_mode());
 	xen_mc_flush();
-	x86_write_percpu(xen_lazy_mode, mode);
 }
 
 static unsigned long xen_store_tr(void)
@@ -358,7 +352,7 @@ static void xen_load_tls(struct thread_struct *t, unsigned int cpu)
 	 * loaded properly.  This will go away as soon as Xen has been
 	 * modified to not save/restore %gs for normal hypercalls.
 	 */
-	if (xen_get_lazy_mode() == PARAVIRT_LAZY_CPU)
+	if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_CPU)
 		loadsegment(gs, 0);
 }
 
@@ -632,32 +626,36 @@ static unsigned long xen_read_cr3(void)
 	return x86_read_percpu(xen_cr3);
 }
 
+static void set_current_cr3(void *v)
+{
+	x86_write_percpu(xen_current_cr3, (unsigned long)v);
+}
+
 static void xen_write_cr3(unsigned long cr3)
 {
+	struct mmuext_op *op;
+	struct multicall_space mcs;
+	unsigned long mfn = pfn_to_mfn(PFN_DOWN(cr3));
+
 	BUG_ON(preemptible());
 
-	if (cr3 == x86_read_percpu(xen_cr3)) {
-		/* just a simple tlb flush */
-		xen_flush_tlb();
-		return;
-	}
+	mcs = xen_mc_entry(sizeof(*op));  /* disables interrupts */
 
+	/* Update while interrupts are disabled, so its atomic with
+	   respect to ipis */
 	x86_write_percpu(xen_cr3, cr3);
 
+	op = mcs.args;
+	op->cmd = MMUEXT_NEW_BASEPTR;
+	op->arg1.mfn = mfn;
 
-	{
-		struct mmuext_op *op;
-		struct multicall_space mcs = xen_mc_entry(sizeof(*op));
-		unsigned long mfn = pfn_to_mfn(PFN_DOWN(cr3));
-
-		op = mcs.args;
-		op->cmd = MMUEXT_NEW_BASEPTR;
-		op->arg1.mfn = mfn;
+	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
 
-		MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
+	/* Update xen_update_cr3 once the batch has actually
+	   been submitted. */
+	xen_mc_callback(set_current_cr3, (void *)cr3);
 
-		xen_mc_issue(PARAVIRT_LAZY_CPU);
-	}
+	xen_mc_issue(PARAVIRT_LAZY_CPU);  /* interrupts restored */
 }
 
 /* Early in boot, while setting up the initial pagetable, assume
@@ -668,6 +666,15 @@ static __init void xen_alloc_pt_init(struct mm_struct *mm, u32 pfn)
 	make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
 }
 
+static void pin_pagetable_pfn(unsigned level, unsigned long pfn)
+{
+	struct mmuext_op op;
+	op.cmd = level;
+	op.arg1.mfn = pfn_to_mfn(pfn);
+	if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF))
+		BUG();
+}
+
 /* This needs to make sure the new pte page is pinned iff its being
    attached to a pinned pagetable. */
 static void xen_alloc_pt(struct mm_struct *mm, u32 pfn)
@@ -677,9 +684,10 @@ static void xen_alloc_pt(struct mm_struct *mm, u32 pfn)
 	if (PagePinned(virt_to_page(mm->pgd))) {
 		SetPagePinned(page);
 
-		if (!PageHighMem(page))
+		if (!PageHighMem(page)) {
 			make_lowmem_page_readonly(__va(PFN_PHYS(pfn)));
-		else
+			pin_pagetable_pfn(MMUEXT_PIN_L1_TABLE, pfn);
+		} else
 			/* make sure there are no stray mappings of
 			   this page */
 			kmap_flush_unused();
@@ -692,8 +700,10 @@ static void xen_release_pt(u32 pfn)
 	struct page *page = pfn_to_page(pfn);
 
 	if (PagePinned(page)) {
-		if (!PageHighMem(page))
+		if (!PageHighMem(page)) {
+			pin_pagetable_pfn(MMUEXT_UNPIN_TABLE, pfn);
 			make_lowmem_page_readwrite(__va(PFN_PHYS(pfn)));
+		}
 	}
 }
 
@@ -738,7 +748,7 @@ 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 +795,8 @@ 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)) {
 		/*
@@ -808,15 +818,15 @@ static __init void xen_pagetable_setup_done(pgd_t *base)
 	/* Actually pin the pagetable down, but we can't set PG_pinned
 	   yet because the page structures don't exist yet. */
 	{
-		struct mmuext_op op;
+		unsigned level;
+
 #ifdef CONFIG_X86_PAE
-		op.cmd = MMUEXT_PIN_L3_TABLE;
+		level = MMUEXT_PIN_L3_TABLE;
 #else
-		op.cmd = MMUEXT_PIN_L3_TABLE;
+		level = MMUEXT_PIN_L2_TABLE;
 #endif
-		op.arg1.mfn = pfn_to_mfn(PFN_DOWN(__pa(base)));
-		if (HYPERVISOR_mmuext_op(&op, 1, NULL, DOMID_SELF))
-			BUG();
+
+		pin_pagetable_pfn(level, PFN_DOWN(__pa(base)));
 	}
 }
 
@@ -833,12 +843,12 @@ 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 +860,8 @@ 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 +870,10 @@ 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 +905,32 @@ 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 +941,10 @@ 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,
@@ -969,6 +973,23 @@ static const struct paravirt_ops xen_paravirt_ops __initdata = {
 	.set_iopl_mask = xen_set_iopl_mask,
 	.io_delay = xen_io_delay,
 
+	.lazy_mode = {
+		.enter = paravirt_enter_lazy_cpu,
+		.leave = xen_leave_lazy,
+	},
+};
+
+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 +998,17 @@ 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 +1018,6 @@ 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,
@@ -1024,7 +1053,10 @@ static const struct paravirt_ops xen_paravirt_ops __initdata = {
 	.dup_mmap = xen_dup_mmap,
 	.exit_mmap = xen_exit_mmap,
 
-	.set_lazy_mode = xen_set_lazy_mode,
+	.lazy_mode = {
+		.enter = paravirt_enter_lazy_mmu,
+		.leave = xen_leave_lazy,
+	},
 };
 
 #ifdef CONFIG_SMP
@@ -1080,6 +1112,17 @@ static const struct machine_ops __initdata xen_machine_ops = {
 };
 
 
+static void __init xen_reserve_top(void)
+{
+	unsigned long top = HYPERVISOR_VIRT_START;
+	struct xen_platform_parameters pp;
+
+	if (HYPERVISOR_xen_version(XENVER_platform_parameters, &pp) == 0)
+		top = pp.virt_start;
+
+	reserve_top_address(-top + 2 * PAGE_SIZE);
+}
+
 /* First C function to be called on Xen boot */
 asmlinkage void __init xen_start_kernel(void)
 {
@@ -1091,7 +1134,14 @@ 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;
+
 	machine_ops = xen_machine_ops;
 
 #ifdef CONFIG_SMP
@@ -1113,6 +1163,7 @@ asmlinkage void __init xen_start_kernel(void)
 	/* keep using Xen gdt for now; no urgent need to change it */
 
 	x86_write_percpu(xen_cr3, __pa(pgd));
+	x86_write_percpu(xen_current_cr3, __pa(pgd));
 
 #ifdef CONFIG_SMP
 	/* Don't do the full vcpu_info placement stuff until we have a
@@ -1124,12 +1175,12 @@ 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);
+	xen_reserve_top();
 
 	/* set up basic CPUID stuff */
 	cpu_detect(&new_cpu_data);
diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c
index 874db0cd1d2..b2e32f9d007 100644
--- a/arch/x86/xen/mmu.c
+++ b/arch/x86/xen/mmu.c
@@ -41,7 +41,6 @@
 #include <linux/sched.h>
 #include <linux/highmem.h>
 #include <linux/bug.h>
-#include <linux/sched.h>
 
 #include <asm/pgtable.h>
 #include <asm/tlbflush.h>
@@ -155,7 +154,7 @@ void xen_set_pte_at(struct mm_struct *mm, unsigned long addr,
 		    pte_t *ptep, pte_t pteval)
 {
 	if (mm == current->mm || mm == &init_mm) {
-		if (xen_get_lazy_mode() == PARAVIRT_LAZY_MMU) {
+		if (paravirt_get_lazy_mode() == PARAVIRT_LAZY_MMU) {
 			struct multicall_space mcs;
 			mcs = xen_mc_entry(0);
 
@@ -304,7 +303,12 @@ pgd_t xen_make_pgd(unsigned long pgd)
 }
 #endif	/* CONFIG_X86_PAE */
 
-
+enum pt_level {
+	PT_PGD,
+	PT_PUD,
+	PT_PMD,
+	PT_PTE
+};
 
 /*
   (Yet another) pagetable walker.  This one is intended for pinning a
@@ -316,7 +320,7 @@ pgd_t xen_make_pgd(unsigned long pgd)
   FIXADDR_TOP.  But the important bit is that we don't pin beyond
   there, because then we start getting into Xen's ptes.
 */
-static int pgd_walk(pgd_t *pgd_base, int (*func)(struct page *, unsigned),
+static int pgd_walk(pgd_t *pgd_base, int (*func)(struct page *, enum pt_level),
 		    unsigned long limit)
 {
 	pgd_t *pgd = pgd_base;
@@ -341,7 +345,7 @@ static int pgd_walk(pgd_t *pgd_base, int (*func)(struct page *, unsigned),
 		pud = pud_offset(pgd, 0);
 
 		if (PTRS_PER_PUD > 1) /* not folded */
-			flush |= (*func)(virt_to_page(pud), 0);
+			flush |= (*func)(virt_to_page(pud), PT_PUD);
 
 		for (; addr != pud_limit; pud++, addr = pud_next) {
 			pmd_t *pmd;
@@ -360,7 +364,7 @@ static int pgd_walk(pgd_t *pgd_base, int (*func)(struct page *, unsigned),
 			pmd = pmd_offset(pud, 0);
 
 			if (PTRS_PER_PMD > 1) /* not folded */
-				flush |= (*func)(virt_to_page(pmd), 0);
+				flush |= (*func)(virt_to_page(pmd), PT_PMD);
 
 			for (; addr != pmd_limit; pmd++) {
 				addr += (PAGE_SIZE * PTRS_PER_PTE);
@@ -372,17 +376,47 @@ static int pgd_walk(pgd_t *pgd_base, int (*func)(struct page *, unsigned),
 				if (pmd_none(*pmd))
 					continue;
 
-				flush |= (*func)(pmd_page(*pmd), 0);
+				flush |= (*func)(pmd_page(*pmd), PT_PTE);
 			}
 		}
 	}
 
-	flush |= (*func)(virt_to_page(pgd_base), UVMF_TLB_FLUSH);
+	flush |= (*func)(virt_to_page(pgd_base), PT_PGD);
 
 	return flush;
 }
 
-static int pin_page(struct page *page, unsigned flags)
+static spinlock_t *lock_pte(struct page *page)
+{
+	spinlock_t *ptl = NULL;
+
+#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
+	ptl = __pte_lockptr(page);
+	spin_lock(ptl);
+#endif
+
+	return ptl;
+}
+
+static void do_unlock(void *v)
+{
+	spinlock_t *ptl = v;
+	spin_unlock(ptl);
+}
+
+static void xen_do_pin(unsigned level, unsigned long pfn)
+{
+	struct mmuext_op *op;
+	struct multicall_space mcs;
+
+	mcs = __xen_mc_entry(sizeof(*op));
+	op = mcs.args;
+	op->cmd = level;
+	op->arg1.mfn = pfn_to_mfn(pfn);
+	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
+}
+
+static int pin_page(struct page *page, enum pt_level level)
 {
 	unsigned pgfl = test_and_set_bit(PG_pinned, &page->flags);
 	int flush;
@@ -397,12 +431,26 @@ static int pin_page(struct page *page, unsigned flags)
 		void *pt = lowmem_page_address(page);
 		unsigned long pfn = page_to_pfn(page);
 		struct multicall_space mcs = __xen_mc_entry(0);
+		spinlock_t *ptl;
 
 		flush = 0;
 
+		ptl = NULL;
+		if (level == PT_PTE)
+			ptl = lock_pte(page);
+
 		MULTI_update_va_mapping(mcs.mc, (unsigned long)pt,
 					pfn_pte(pfn, PAGE_KERNEL_RO),
-					flags);
+					level == PT_PGD ? UVMF_TLB_FLUSH : 0);
+
+		if (level == PT_PTE)
+			xen_do_pin(MMUEXT_PIN_L1_TABLE, pfn);
+
+		if (ptl) {
+			/* Queue a deferred unlock for when this batch
+			   is completed. */
+			xen_mc_callback(do_unlock, ptl);
+		}
 	}
 
 	return flush;
@@ -413,8 +461,7 @@ static int pin_page(struct page *page, unsigned flags)
    read-only, and can be pinned. */
 void xen_pgd_pin(pgd_t *pgd)
 {
-	struct multicall_space mcs;
-	struct mmuext_op *op;
+	unsigned level;
 
 	xen_mc_batch();
 
@@ -425,16 +472,13 @@ void xen_pgd_pin(pgd_t *pgd)
 		xen_mc_batch();
 	}
 
-	mcs = __xen_mc_entry(sizeof(*op));
-	op = mcs.args;
-
 #ifdef CONFIG_X86_PAE
-	op->cmd = MMUEXT_PIN_L3_TABLE;
+	level = MMUEXT_PIN_L3_TABLE;
 #else
-	op->cmd = MMUEXT_PIN_L2_TABLE;
+	level = MMUEXT_PIN_L2_TABLE;
 #endif
-	op->arg1.mfn = pfn_to_mfn(PFN_DOWN(__pa(pgd)));
-	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
+
+	xen_do_pin(level, PFN_DOWN(__pa(pgd)));
 
 	xen_mc_issue(0);
 }
@@ -442,7 +486,7 @@ void xen_pgd_pin(pgd_t *pgd)
 /* The init_mm pagetable is really pinned as soon as its created, but
    that's before we have page structures to store the bits.  So do all
    the book-keeping now. */
-static __init int mark_pinned(struct page *page, unsigned flags)
+static __init int mark_pinned(struct page *page, enum pt_level level)
 {
 	SetPagePinned(page);
 	return 0;
@@ -453,18 +497,32 @@ void __init xen_mark_init_mm_pinned(void)
 	pgd_walk(init_mm.pgd, mark_pinned, FIXADDR_TOP);
 }
 
-static int unpin_page(struct page *page, unsigned flags)
+static int unpin_page(struct page *page, enum pt_level level)
 {
 	unsigned pgfl = test_and_clear_bit(PG_pinned, &page->flags);
 
 	if (pgfl && !PageHighMem(page)) {
 		void *pt = lowmem_page_address(page);
 		unsigned long pfn = page_to_pfn(page);
-		struct multicall_space mcs = __xen_mc_entry(0);
+		spinlock_t *ptl = NULL;
+		struct multicall_space mcs;
+
+		if (level == PT_PTE) {
+			ptl = lock_pte(page);
+
+			xen_do_pin(MMUEXT_UNPIN_TABLE, pfn);
+		}
+
+		mcs = __xen_mc_entry(0);
 
 		MULTI_update_va_mapping(mcs.mc, (unsigned long)pt,
 					pfn_pte(pfn, PAGE_KERNEL),
-					flags);
+					level == PT_PGD ? UVMF_TLB_FLUSH : 0);
+
+		if (ptl) {
+			/* unlock when batch completed */
+			xen_mc_callback(do_unlock, ptl);
+		}
 	}
 
 	return 0;		/* never need to flush on unpin */
@@ -473,18 +531,9 @@ static int unpin_page(struct page *page, unsigned flags)
 /* Release a pagetables pages back as normal RW */
 static void xen_pgd_unpin(pgd_t *pgd)
 {
-	struct mmuext_op *op;
-	struct multicall_space mcs;
-
 	xen_mc_batch();
 
-	mcs = __xen_mc_entry(sizeof(*op));
-
-	op = mcs.args;
-	op->cmd = MMUEXT_UNPIN_TABLE;
-	op->arg1.mfn = pfn_to_mfn(PFN_DOWN(__pa(pgd)));
-
-	MULTI_mmuext_op(mcs.mc, op, 1, NULL, DOMID_SELF);
+	xen_do_pin(MMUEXT_UNPIN_TABLE, PFN_DOWN(__pa(pgd)));
 
 	pgd_walk(pgd, unpin_page, TASK_SIZE);
 
@@ -515,20 +564,43 @@ static void drop_other_mm_ref(void *info)
 
 	if (__get_cpu_var(cpu_tlbstate).active_mm == mm)
 		leave_mm(smp_processor_id());
+
+	/* If this cpu still has a stale cr3 reference, then make sure
+	   it has been flushed. */
+	if (x86_read_percpu(xen_current_cr3) == __pa(mm->pgd)) {
+		load_cr3(swapper_pg_dir);
+		arch_flush_lazy_cpu_mode();
+	}
 }
 
 static void drop_mm_ref(struct mm_struct *mm)
 {
+	cpumask_t mask;
+	unsigned cpu;
+
 	if (current->active_mm == mm) {
 		if (current->mm == mm)
 			load_cr3(swapper_pg_dir);
 		else
 			leave_mm(smp_processor_id());
+		arch_flush_lazy_cpu_mode();
 	}
 
-	if (!cpus_empty(mm->cpu_vm_mask))
-		xen_smp_call_function_mask(mm->cpu_vm_mask, drop_other_mm_ref,
-					   mm, 1);
+	/* Get the "official" set of cpus referring to our pagetable. */
+	mask = mm->cpu_vm_mask;
+
+	/* It's possible that a vcpu may have a stale reference to our
+	   cr3, because its in lazy mode, and it hasn't yet flushed
+	   its set of pending hypercalls yet.  In this case, we can
+	   look at its actual current cr3 value, and force it to flush
+	   if needed. */
+	for_each_online_cpu(cpu) {
+		if (per_cpu(xen_current_cr3, cpu) == __pa(mm->pgd))
+			cpu_set(cpu, mask);
+	}
+
+	if (!cpus_empty(mask))
+		xen_smp_call_function_mask(mask, drop_other_mm_ref, mm, 1);
 }
 #else
 static void drop_mm_ref(struct mm_struct *mm)
@@ -563,5 +635,6 @@ void xen_exit_mmap(struct mm_struct *mm)
 	/* pgd may not be pinned in the error exit path of execve */
 	if (PagePinned(virt_to_page(mm->pgd)))
 		xen_pgd_unpin(mm->pgd);
+
 	spin_unlock(&mm->page_table_lock);
 }
diff --git a/arch/x86/xen/multicalls.c b/arch/x86/xen/multicalls.c
index c837e8e463d..5e6f36f6d87 100644
--- a/arch/x86/xen/multicalls.c
+++ b/arch/x86/xen/multicalls.c
@@ -26,13 +26,22 @@
 
 #include "multicalls.h"
 
+#define MC_DEBUG	1
+
 #define MC_BATCH	32
 #define MC_ARGS		(MC_BATCH * 16 / sizeof(u64))
 
 struct mc_buffer {
 	struct multicall_entry entries[MC_BATCH];
+#if MC_DEBUG
+	struct multicall_entry debug[MC_BATCH];
+#endif
 	u64 args[MC_ARGS];
-	unsigned mcidx, argidx;
+	struct callback {
+		void (*fn)(void *);
+		void *data;
+	} callbacks[MC_BATCH];
+	unsigned mcidx, argidx, cbidx;
 };
 
 static DEFINE_PER_CPU(struct mc_buffer, mc_buffer);
@@ -43,6 +52,7 @@ void xen_mc_flush(void)
 	struct mc_buffer *b = &__get_cpu_var(mc_buffer);
 	int ret = 0;
 	unsigned long flags;
+	int i;
 
 	BUG_ON(preemptible());
 
@@ -51,13 +61,31 @@ void xen_mc_flush(void)
 	local_irq_save(flags);
 
 	if (b->mcidx) {
-		int i;
+#if MC_DEBUG
+		memcpy(b->debug, b->entries,
+		       b->mcidx * sizeof(struct multicall_entry));
+#endif
 
 		if (HYPERVISOR_multicall(b->entries, b->mcidx) != 0)
 			BUG();
 		for (i = 0; i < b->mcidx; i++)
 			if (b->entries[i].result < 0)
 				ret++;
+
+#if MC_DEBUG
+		if (ret) {
+			printk(KERN_ERR "%d multicall(s) failed: cpu %d\n",
+			       ret, smp_processor_id());
+			for(i = 0; i < b->mcidx; i++) {
+				printk("  call %2d/%d: op=%lu arg=[%lx] result=%ld\n",
+				       i+1, b->mcidx,
+				       b->debug[i].op,
+				       b->debug[i].args[0],
+				       b->entries[i].result);
+			}
+		}
+#endif
+
 		b->mcidx = 0;
 		b->argidx = 0;
 	} else
@@ -65,6 +93,13 @@ void xen_mc_flush(void)
 
 	local_irq_restore(flags);
 
+	for(i = 0; i < b->cbidx; i++) {
+		struct callback *cb = &b->callbacks[i];
+
+		(*cb->fn)(cb->data);
+	}
+	b->cbidx = 0;
+
 	BUG_ON(ret);
 }
 
@@ -88,3 +123,16 @@ struct multicall_space __xen_mc_entry(size_t args)
 
 	return ret;
 }
+
+void xen_mc_callback(void (*fn)(void *), void *data)
+{
+	struct mc_buffer *b = &__get_cpu_var(mc_buffer);
+	struct callback *cb;
+
+	if (b->cbidx == MC_BATCH)
+		xen_mc_flush();
+
+	cb = &b->callbacks[b->cbidx++];
+	cb->fn = fn;
+	cb->data = data;
+}
diff --git a/arch/x86/xen/multicalls.h b/arch/x86/xen/multicalls.h
index e6f7530b156..8bae996d99a 100644
--- a/arch/x86/xen/multicalls.h
+++ b/arch/x86/xen/multicalls.h
@@ -35,11 +35,14 @@ void xen_mc_flush(void);
 /* Issue a multicall if we're not in a lazy mode */
 static inline void xen_mc_issue(unsigned mode)
 {
-	if ((xen_get_lazy_mode() & mode) == 0)
+	if ((paravirt_get_lazy_mode() & mode) == 0)
 		xen_mc_flush();
 
 	/* restore flags saved in xen_mc_batch */
 	local_irq_restore(x86_read_percpu(xen_mc_irq_flags));
 }
 
+/* Set up a callback to be called when the current batch is flushed */
+void xen_mc_callback(void (*fn)(void *), void *data);
+
 #endif /* _XEN_MULTICALLS_H */
diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
index 4fa33c27ccb..d53bf9d8a72 100644
--- a/arch/x86/xen/smp.c
+++ b/arch/x86/xen/smp.c
@@ -370,7 +370,8 @@ int xen_smp_call_function_mask(cpumask_t mask, void (*func)(void *),
 			       void *info, int wait)
 {
 	struct call_data_struct data;
-	int cpus;
+	int cpus, cpu;
+	bool yield;
 
 	/* Holding any lock stops cpus from going down. */
 	spin_lock(&call_lock);
@@ -399,9 +400,14 @@ int xen_smp_call_function_mask(cpumask_t mask, void (*func)(void *),
 	/* Send a message to other CPUs and wait for them to respond */
 	xen_send_IPI_mask(mask, XEN_CALL_FUNCTION_VECTOR);
 
-	/* Make sure other vcpus get a chance to run.
-	   XXX too severe?  Maybe we should check the other CPU's states? */
-	HYPERVISOR_sched_op(SCHEDOP_yield, 0);
+	/* Make sure other vcpus get a chance to run if they need to. */
+	yield = false;
+	for_each_cpu_mask(cpu, mask)
+		if (xen_vcpu_stolen(cpu))
+			yield = true;
+
+	if (yield)
+		HYPERVISOR_sched_op(SCHEDOP_yield, 0);
 
 	/* Wait for response */
 	while (atomic_read(&data.started) != cpus ||
diff --git a/arch/x86/xen/time.c b/arch/x86/xen/time.c
index dfd6db69ead..d083ff5ef08 100644
--- a/arch/x86/xen/time.c
+++ b/arch/x86/xen/time.c
@@ -105,6 +105,12 @@ static void get_runstate_snapshot(struct vcpu_runstate_info *res)
 	} while (get64(&state->state_entry_time) != state_time);
 }
 
+/* return true when a vcpu could run but has no real cpu to run on */
+bool xen_vcpu_stolen(int vcpu)
+{
+	return per_cpu(runstate, vcpu).state == RUNSTATE_runnable;
+}
+
 static void setup_runstate_info(int cpu)
 {
 	struct vcpu_register_runstate_memory_area area;
diff --git a/arch/x86/xen/xen-ops.h b/arch/x86/xen/xen-ops.h
index b9aaea45f07..b02a909bfd4 100644
--- a/arch/x86/xen/xen-ops.h
+++ b/arch/x86/xen/xen-ops.h
@@ -11,6 +11,7 @@ void xen_copy_trap_info(struct trap_info *traps);
 
 DECLARE_PER_CPU(struct vcpu_info *, xen_vcpu);
 DECLARE_PER_CPU(unsigned long, xen_cr3);
+DECLARE_PER_CPU(unsigned long, xen_current_cr3);
 
 extern struct start_info *xen_start_info;
 extern struct shared_info *HYPERVISOR_shared_info;
@@ -27,14 +28,9 @@ unsigned long xen_get_wallclock(void);
 int xen_set_wallclock(unsigned long time);
 unsigned long long xen_sched_clock(void);
 
-void xen_mark_init_mm_pinned(void);
-
-DECLARE_PER_CPU(enum paravirt_lazy_mode, xen_lazy_mode);
+bool xen_vcpu_stolen(int vcpu);
 
-static inline unsigned xen_get_lazy_mode(void)
-{
-	return x86_read_percpu(xen_lazy_mode);
-}
+void xen_mark_init_mm_pinned(void);
 
 void __init xen_fill_possible_map(void);