Merge branch 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull x86 perf updates from Ingo Molnar:
 "This series tightens up RDPMC permissions: currently even highly
  sandboxed x86 execution environments (such as seccomp) have permission
  to execute RDPMC, which may leak various perf events / PMU state such
  as timing information and other CPU execution details.

  This 'all is allowed' RDPMC mode is still preserved as the
  (non-default) /sys/devices/cpu/rdpmc=2 setting.  The new default is
  that RDPMC access is only allowed if a perf event is mmap-ed (which is
  needed to correctly interpret RDPMC counter values in any case).

  As a side effect of these changes CR4 handling is cleaned up in the
  x86 code and a shadow copy of the CR4 value is added.

  The extra CR4 manipulation adds ~ <50ns to the context switch cost
  between rdpmc-capable and rdpmc-non-capable mms"

* 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf/x86: Add /sys/devices/cpu/rdpmc=2 to allow rdpmc for all tasks
  perf/x86: Only allow rdpmc if a perf_event is mapped
  perf: Pass the event to arch_perf_update_userpage()
  perf: Add pmu callbacks to track event mapping and unmapping
  x86: Add a comment clarifying LDT context switching
  x86: Store a per-cpu shadow copy of CR4
  x86: Clean up cr4 manipulation

35 files changed, 253 insertions, 120 deletions

diff --git a/arch/x86/include/asm/mmu.h b/arch/x86/include/asm/mmu.h
index 876e74e8eec7..09b9620a73b4 100644
--- a/arch/x86/include/asm/mmu.h
+++ b/arch/x86/include/asm/mmu.h
@@ -19,6 +19,8 @@ typedef struct {
 
 	struct mutex lock;
 	void __user *vdso;
+
+	atomic_t perf_rdpmc_allowed;	/* nonzero if rdpmc is allowed */
 } mm_context_t;
 
 #ifdef CONFIG_SMP
diff --git a/arch/x86/include/asm/mmu_context.h b/arch/x86/include/asm/mmu_context.h
index 4b75d591eb5e..883f6b933fa4 100644
--- a/arch/x86/include/asm/mmu_context.h
+++ b/arch/x86/include/asm/mmu_context.h
@@ -18,6 +18,21 @@ static inline void paravirt_activate_mm(struct mm_struct *prev,
 }
 #endif	/* !CONFIG_PARAVIRT */
 
+#ifdef CONFIG_PERF_EVENTS
+extern struct static_key rdpmc_always_available;
+
+static inline void load_mm_cr4(struct mm_struct *mm)
+{
+	if (static_key_true(&rdpmc_always_available) ||
+	    atomic_read(&mm->context.perf_rdpmc_allowed))
+		cr4_set_bits(X86_CR4_PCE);
+	else
+		cr4_clear_bits(X86_CR4_PCE);
+}
+#else
+static inline void load_mm_cr4(struct mm_struct *mm) {}
+#endif
+
 /*
  * Used for LDT copy/destruction.
  */
@@ -52,15 +67,20 @@ static inline void switch_mm(struct mm_struct *prev, struct mm_struct *next,
 		/* Stop flush ipis for the previous mm */
 		cpumask_clear_cpu(cpu, mm_cpumask(prev));
 
+		/* Load per-mm CR4 state */
+		load_mm_cr4(next);
+
 		/*
 		 * Load the LDT, if the LDT is different.
 		 *
-		 * It's possible leave_mm(prev) has been called.  If so,
-		 * then prev->context.ldt could be out of sync with the
-		 * LDT descriptor or the LDT register.  This can only happen
-		 * if prev->context.ldt is non-null, since we never free
-		 * an LDT.  But LDTs can't be shared across mms, so
-		 * prev->context.ldt won't be equal to next->context.ldt.
+		 * It's possible that prev->context.ldt doesn't match
+		 * the LDT register.  This can happen if leave_mm(prev)
+		 * was called and then modify_ldt changed
+		 * prev->context.ldt but suppressed an IPI to this CPU.
+		 * In this case, prev->context.ldt != NULL, because we
+		 * never free an LDT while the mm still exists.  That
+		 * means that next->context.ldt != prev->context.ldt,
+		 * because mms never share an LDT.
 		 */
 		if (unlikely(prev->context.ldt != next->context.ldt))
 			load_LDT_nolock(&next->context);
@@ -85,6 +105,7 @@ static inline void switch_mm(struct mm_struct *prev, struct mm_struct *next,
 			 */
 			load_cr3(next->pgd);
 			trace_tlb_flush(TLB_FLUSH_ON_TASK_SWITCH, TLB_FLUSH_ALL);
+			load_mm_cr4(next);
 			load_LDT_nolock(&next->context);
 		}
 	}
diff --git a/arch/x86/include/asm/paravirt.h b/arch/x86/include/asm/paravirt.h
index 32444ae939ca..965c47d254aa 100644
--- a/arch/x86/include/asm/paravirt.h
+++ b/arch/x86/include/asm/paravirt.h
@@ -80,16 +80,16 @@ static inline void write_cr3(unsigned long x)
 	PVOP_VCALL1(pv_mmu_ops.write_cr3, x);
 }
 
-static inline unsigned long read_cr4(void)
+static inline unsigned long __read_cr4(void)
 {
 	return PVOP_CALL0(unsigned long, pv_cpu_ops.read_cr4);
 }
-static inline unsigned long read_cr4_safe(void)
+static inline unsigned long __read_cr4_safe(void)
 {
 	return PVOP_CALL0(unsigned long, pv_cpu_ops.read_cr4_safe);
 }
 
-static inline void write_cr4(unsigned long x)
+static inline void __write_cr4(unsigned long x)
 {
 	PVOP_VCALL1(pv_cpu_ops.write_cr4, x);
 }
diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index a092a0cce0b7..ec1c93588cef 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -579,39 +579,6 @@ static inline void load_sp0(struct tss_struct *tss,
 #define set_iopl_mask native_set_iopl_mask
 #endif /* CONFIG_PARAVIRT */
 
-/*
- * Save the cr4 feature set we're using (ie
- * Pentium 4MB enable and PPro Global page
- * enable), so that any CPU's that boot up
- * after us can get the correct flags.
- */
-extern unsigned long mmu_cr4_features;
-extern u32 *trampoline_cr4_features;
-
-static inline void set_in_cr4(unsigned long mask)
-{
-	unsigned long cr4;
-
-	mmu_cr4_features |= mask;
-	if (trampoline_cr4_features)
-		*trampoline_cr4_features = mmu_cr4_features;
-	cr4 = read_cr4();
-	cr4 |= mask;
-	write_cr4(cr4);
-}
-
-static inline void clear_in_cr4(unsigned long mask)
-{
-	unsigned long cr4;
-
-	mmu_cr4_features &= ~mask;
-	if (trampoline_cr4_features)
-		*trampoline_cr4_features = mmu_cr4_features;
-	cr4 = read_cr4();
-	cr4 &= ~mask;
-	write_cr4(cr4);
-}
-
 typedef struct {
 	unsigned long		seg;
 } mm_segment_t;
diff --git a/arch/x86/include/asm/special_insns.h b/arch/x86/include/asm/special_insns.h
index e820c080a4e9..6a4b00fafb00 100644
--- a/arch/x86/include/asm/special_insns.h
+++ b/arch/x86/include/asm/special_insns.h
@@ -137,17 +137,17 @@ static inline void write_cr3(unsigned long x)
 	native_write_cr3(x);
 }
 
-static inline unsigned long read_cr4(void)
+static inline unsigned long __read_cr4(void)
 {
 	return native_read_cr4();
 }
 
-static inline unsigned long read_cr4_safe(void)
+static inline unsigned long __read_cr4_safe(void)
 {
 	return native_read_cr4_safe();
 }
 
-static inline void write_cr4(unsigned long x)
+static inline void __write_cr4(unsigned long x)
 {
 	native_write_cr4(x);
 }
diff --git a/arch/x86/include/asm/tlbflush.h b/arch/x86/include/asm/tlbflush.h
index 04905bfc508b..cd791948b286 100644
--- a/arch/x86/include/asm/tlbflush.h
+++ b/arch/x86/include/asm/tlbflush.h
@@ -15,6 +15,75 @@
 #define __flush_tlb_single(addr) __native_flush_tlb_single(addr)
 #endif
 
+struct tlb_state {
+#ifdef CONFIG_SMP
+	struct mm_struct *active_mm;
+	int state;
+#endif
+
+	/*
+	 * Access to this CR4 shadow and to H/W CR4 is protected by
+	 * disabling interrupts when modifying either one.
+	 */
+	unsigned long cr4;
+};
+DECLARE_PER_CPU_SHARED_ALIGNED(struct tlb_state, cpu_tlbstate);
+
+/* Initialize cr4 shadow for this CPU. */
+static inline void cr4_init_shadow(void)
+{
+	this_cpu_write(cpu_tlbstate.cr4, __read_cr4());
+}
+
+/* Set in this cpu's CR4. */
+static inline void cr4_set_bits(unsigned long mask)
+{
+	unsigned long cr4;
+
+	cr4 = this_cpu_read(cpu_tlbstate.cr4);
+	if ((cr4 | mask) != cr4) {
+		cr4 |= mask;
+		this_cpu_write(cpu_tlbstate.cr4, cr4);
+		__write_cr4(cr4);
+	}
+}
+
+/* Clear in this cpu's CR4. */
+static inline void cr4_clear_bits(unsigned long mask)
+{
+	unsigned long cr4;
+
+	cr4 = this_cpu_read(cpu_tlbstate.cr4);
+	if ((cr4 & ~mask) != cr4) {
+		cr4 &= ~mask;
+		this_cpu_write(cpu_tlbstate.cr4, cr4);
+		__write_cr4(cr4);
+	}
+}
+
+/* Read the CR4 shadow. */
+static inline unsigned long cr4_read_shadow(void)
+{
+	return this_cpu_read(cpu_tlbstate.cr4);
+}
+
+/*
+ * Save some of cr4 feature set we're using (e.g.  Pentium 4MB
+ * enable and PPro Global page enable), so that any CPU's that boot
+ * up after us can get the correct flags.  This should only be used
+ * during boot on the boot cpu.
+ */
+extern unsigned long mmu_cr4_features;
+extern u32 *trampoline_cr4_features;
+
+static inline void cr4_set_bits_and_update_boot(unsigned long mask)
+{
+	mmu_cr4_features |= mask;
+	if (trampoline_cr4_features)
+		*trampoline_cr4_features = mmu_cr4_features;
+	cr4_set_bits(mask);
+}
+
 static inline void __native_flush_tlb(void)
 {
 	native_write_cr3(native_read_cr3());
@@ -24,7 +93,7 @@ static inline void __native_flush_tlb_global_irq_disabled(void)
 {
 	unsigned long cr4;
 
-	cr4 = native_read_cr4();
+	cr4 = this_cpu_read(cpu_tlbstate.cr4);
 	/* clear PGE */
 	native_write_cr4(cr4 & ~X86_CR4_PGE);
 	/* write old PGE again and flush TLBs */
@@ -184,12 +253,6 @@ void native_flush_tlb_others(const struct cpumask *cpumask,
 #define TLBSTATE_OK	1
 #define TLBSTATE_LAZY	2
 
-struct tlb_state {
-	struct mm_struct *active_mm;
-	int state;
-};
-DECLARE_PER_CPU_SHARED_ALIGNED(struct tlb_state, cpu_tlbstate);
-
 static inline void reset_lazy_tlbstate(void)
 {
 	this_cpu_write(cpu_tlbstate.state, 0);
diff --git a/arch/x86/include/asm/virtext.h b/arch/x86/include/asm/virtext.h
index 5da71c27cc59..cce9ee68e335 100644
--- a/arch/x86/include/asm/virtext.h
+++ b/arch/x86/include/asm/virtext.h
@@ -19,6 +19,7 @@
 
 #include <asm/vmx.h>
 #include <asm/svm.h>
+#include <asm/tlbflush.h>
 
 /*
  * VMX functions:
@@ -40,12 +41,12 @@ static inline int cpu_has_vmx(void)
 static inline void cpu_vmxoff(void)
 {
 	asm volatile (ASM_VMX_VMXOFF : : : "cc");
-	write_cr4(read_cr4() & ~X86_CR4_VMXE);
+	cr4_clear_bits(X86_CR4_VMXE);
 }
 
 static inline int cpu_vmx_enabled(void)
 {
-	return read_cr4() & X86_CR4_VMXE;
+	return __read_cr4() & X86_CR4_VMXE;
 }
 
 /** Disable VMX if it is enabled on the current CPU
diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c
index 31368207837c..d1daead5fcdd 100644
--- a/arch/x86/kernel/acpi/sleep.c
+++ b/arch/x86/kernel/acpi/sleep.c
@@ -78,7 +78,7 @@ int x86_acpi_suspend_lowlevel(void)
 
 	header->pmode_cr0 = read_cr0();
 	if (__this_cpu_read(cpu_info.cpuid_level) >= 0) {
-		header->pmode_cr4 = read_cr4();
+		header->pmode_cr4 = __read_cr4();
 		header->pmode_behavior |= (1 << WAKEUP_BEHAVIOR_RESTORE_CR4);
 	}
 	if (!rdmsr_safe(MSR_IA32_MISC_ENABLE,
diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c
index b15bffcaba6d..b5c8ff5e9dfc 100644
--- a/arch/x86/kernel/cpu/common.c
+++ b/arch/x86/kernel/cpu/common.c
@@ -19,6 +19,7 @@
 #include <asm/archrandom.h>
 #include <asm/hypervisor.h>
 #include <asm/processor.h>
+#include <asm/tlbflush.h>
 #include <asm/debugreg.h>
 #include <asm/sections.h>
 #include <asm/vsyscall.h>
@@ -278,7 +279,7 @@ __setup("nosmep", setup_disable_smep);
 static __always_inline void setup_smep(struct cpuinfo_x86 *c)
 {
 	if (cpu_has(c, X86_FEATURE_SMEP))
-		set_in_cr4(X86_CR4_SMEP);
+		cr4_set_bits(X86_CR4_SMEP);
 }
 
 static __init int setup_disable_smap(char *arg)
@@ -298,9 +299,9 @@ static __always_inline void setup_smap(struct cpuinfo_x86 *c)
 
 	if (cpu_has(c, X86_FEATURE_SMAP)) {
 #ifdef CONFIG_X86_SMAP
-		set_in_cr4(X86_CR4_SMAP);
+		cr4_set_bits(X86_CR4_SMAP);
 #else
-		clear_in_cr4(X86_CR4_SMAP);
+		cr4_clear_bits(X86_CR4_SMAP);
 #endif
 	}
 }
@@ -1295,6 +1296,12 @@ void cpu_init(void)
 	wait_for_master_cpu(cpu);
 
 	/*
+	 * Initialize the CR4 shadow before doing anything that could
+	 * try to read it.
+	 */
+	cr4_init_shadow();
+
+	/*
 	 * Load microcode on this cpu if a valid microcode is available.
 	 * This is early microcode loading procedure.
 	 */
@@ -1313,7 +1320,7 @@ void cpu_init(void)
 
 	pr_debug("Initializing CPU#%d\n", cpu);
 
-	clear_in_cr4(X86_CR4_VME|X86_CR4_PVI|X86_CR4_TSD|X86_CR4_DE);
+	cr4_clear_bits(X86_CR4_VME|X86_CR4_PVI|X86_CR4_TSD|X86_CR4_DE);
 
 	/*
 	 * Initialize the per-CPU GDT with the boot GDT,
@@ -1394,7 +1401,7 @@ void cpu_init(void)
 	printk(KERN_INFO "Initializing CPU#%d\n", cpu);
 
 	if (cpu_feature_enabled(X86_FEATURE_VME) || cpu_has_tsc || cpu_has_de)
-		clear_in_cr4(X86_CR4_VME|X86_CR4_PVI|X86_CR4_TSD|X86_CR4_DE);
+		cr4_clear_bits(X86_CR4_VME|X86_CR4_PVI|X86_CR4_TSD|X86_CR4_DE);
 
 	load_current_idt();
 	switch_to_new_gdt(cpu);
diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c
index cdfed7953963..3be9fa69f875 100644
--- a/arch/x86/kernel/cpu/mcheck/mce.c
+++ b/arch/x86/kernel/cpu/mcheck/mce.c
@@ -44,6 +44,7 @@
 
 #include <asm/processor.h>
 #include <asm/traps.h>
+#include <asm/tlbflush.h>
 #include <asm/mce.h>
 #include <asm/msr.h>
 
@@ -1452,7 +1453,7 @@ static void __mcheck_cpu_init_generic(void)
 	bitmap_fill(all_banks, MAX_NR_BANKS);
 	machine_check_poll(MCP_UC | m_fl, &all_banks);
 
-	set_in_cr4(X86_CR4_MCE);
+	cr4_set_bits(X86_CR4_MCE);
 
 	rdmsrl(MSR_IA32_MCG_CAP, cap);
 	if (cap & MCG_CTL_P)
diff --git a/arch/x86/kernel/cpu/mcheck/p5.c b/arch/x86/kernel/cpu/mcheck/p5.c
index ec2663a708e4..737b0ad4e61a 100644
--- a/arch/x86/kernel/cpu/mcheck/p5.c
+++ b/arch/x86/kernel/cpu/mcheck/p5.c
@@ -9,6 +9,7 @@
 
 #include <asm/processor.h>
 #include <asm/traps.h>
+#include <asm/tlbflush.h>
 #include <asm/mce.h>
 #include <asm/msr.h>
 
@@ -65,7 +66,7 @@ void intel_p5_mcheck_init(struct cpuinfo_x86 *c)
 	       "Intel old style machine check architecture supported.\n");
 
 	/* Enable MCE: */
-	set_in_cr4(X86_CR4_MCE);
+	cr4_set_bits(X86_CR4_MCE);
 	printk(KERN_INFO
 	       "Intel old style machine check reporting enabled on CPU#%d.\n",
 	       smp_processor_id());
diff --git a/arch/x86/kernel/cpu/mcheck/winchip.c b/arch/x86/kernel/cpu/mcheck/winchip.c
index bd5d46a32210..44f138296fbe 100644
--- a/arch/x86/kernel/cpu/mcheck/winchip.c
+++ b/arch/x86/kernel/cpu/mcheck/winchip.c
@@ -8,6 +8,7 @@
 
 #include <asm/processor.h>
 #include <asm/traps.h>
+#include <asm/tlbflush.h>
 #include <asm/mce.h>
 #include <asm/msr.h>
 
@@ -36,7 +37,7 @@ void winchip_mcheck_init(struct cpuinfo_x86 *c)
 	lo &= ~(1<<4);	/* Enable MCE */
 	wrmsr(MSR_IDT_FCR1, lo, hi);
 
-	set_in_cr4(X86_CR4_MCE);
+	cr4_set_bits(X86_CR4_MCE);
 
 	printk(KERN_INFO
 	       "Winchip machine check reporting enabled on CPU#0.\n");
diff --git a/arch/x86/kernel/cpu/mtrr/cyrix.c b/arch/x86/kernel/cpu/mtrr/cyrix.c
index 9e451b0876b5..f8c81ba0b465 100644
--- a/arch/x86/kernel/cpu/mtrr/cyrix.c
+++ b/arch/x86/kernel/cpu/mtrr/cyrix.c
@@ -138,8 +138,8 @@ static void prepare_set(void)
 
 	/*  Save value of CR4 and clear Page Global Enable (bit 7)  */
 	if (cpu_has_pge) {
-		cr4 = read_cr4();
-		write_cr4(cr4 & ~X86_CR4_PGE);
+		cr4 = __read_cr4();
+		__write_cr4(cr4 & ~X86_CR4_PGE);
 	}
 
 	/*
@@ -171,7 +171,7 @@ static void post_set(void)
 
 	/* Restore value of CR4 */
 	if (cpu_has_pge)
-		write_cr4(cr4);
+		__write_cr4(cr4);
 }
 
 static void cyrix_set_arr(unsigned int reg, unsigned long base,
diff --git a/arch/x86/kernel/cpu/mtrr/generic.c b/arch/x86/kernel/cpu/mtrr/generic.c
index 0e25a1bc5ab5..7d74f7b3c6ba 100644
--- a/arch/x86/kernel/cpu/mtrr/generic.c
+++ b/arch/x86/kernel/cpu/mtrr/generic.c
@@ -678,8 +678,8 @@ static void prepare_set(void) __acquires(set_atomicity_lock)
 
 	/* Save value of CR4 and clear Page Global Enable (bit 7) */
 	if (cpu_has_pge) {
-		cr4 = read_cr4();
-		write_cr4(cr4 & ~X86_CR4_PGE);
+		cr4 = __read_cr4();
+		__write_cr4(cr4 & ~X86_CR4_PGE);
 	}
 
 	/* Flush all TLBs via a mov %cr3, %reg; mov %reg, %cr3 */
@@ -708,7 +708,7 @@ static void post_set(void) __releases(set_atomicity_lock)
 
 	/* Restore value of CR4 */
 	if (cpu_has_pge)
-		write_cr4(cr4);
+		__write_cr4(cr4);
 	raw_spin_unlock(&set_atomicity_lock);
 }
 
diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index 143e5f5dc855..b71a7f86d68a 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -31,6 +31,8 @@
 #include <asm/nmi.h>
 #include <asm/smp.h>
 #include <asm/alternative.h>
+#include <asm/mmu_context.h>
+#include <asm/tlbflush.h>
 #include <asm/timer.h>
 #include <asm/desc.h>
 #include <asm/ldt.h>
@@ -43,6 +45,8 @@ DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = {
 	.enabled = 1,
 };
 
+struct static_key rdpmc_always_available = STATIC_KEY_INIT_FALSE;
+
 u64 __read_mostly hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
@@ -1327,8 +1331,6 @@ x86_pmu_notifier(struct notifier_block *self, unsigned long action, void *hcpu)
 		break;
 
 	case CPU_STARTING:
-		if (x86_pmu.attr_rdpmc)
-			set_in_cr4(X86_CR4_PCE);
 		if (x86_pmu.cpu_starting)
 			x86_pmu.cpu_starting(cpu);
 		break;
@@ -1804,14 +1806,44 @@ static int x86_pmu_event_init(struct perf_event *event)
 			event->destroy(event);
 	}
 
+	if (ACCESS_ONCE(x86_pmu.attr_rdpmc))
+		event->hw.flags |= PERF_X86_EVENT_RDPMC_ALLOWED;
+
 	return err;
 }
 
+static void refresh_pce(void *ignored)
+{
+	if (current->mm)
+		load_mm_cr4(current->mm);
+}
+
+static void x86_pmu_event_mapped(struct perf_event *event)
+{
+	if (!(event->hw.flags & PERF_X86_EVENT_RDPMC_ALLOWED))
+		return;
+
+	if (atomic_inc_return(&current->mm->context.perf_rdpmc_allowed) == 1)
+		on_each_cpu_mask(mm_cpumask(current->mm), refresh_pce, NULL, 1);
+}
+
+static void x86_pmu_event_unmapped(struct perf_event *event)
+{
+	if (!current->mm)
+		return;
+
+	if (!(event->hw.flags & PERF_X86_EVENT_RDPMC_ALLOWED))
+		return;
+
+	if (atomic_dec_and_test(&current->mm->context.perf_rdpmc_allowed))
+		on_each_cpu_mask(mm_cpumask(current->mm), refresh_pce, NULL, 1);
+}
+
 static int x86_pmu_event_idx(struct perf_event *event)
 {
 	int idx = event->hw.idx;
 
-	if (!x86_pmu.attr_rdpmc)
+	if (!(event->hw.flags & PERF_X86_EVENT_RDPMC_ALLOWED))
 		return 0;
 
 	if (x86_pmu.num_counters_fixed && idx >= INTEL_PMC_IDX_FIXED) {
@@ -1829,16 +1861,6 @@ static ssize_t get_attr_rdpmc(struct device *cdev,
 	return snprintf(buf, 40, "%d\n", x86_pmu.attr_rdpmc);
 }
 
-static void change_rdpmc(void *info)
-{
-	bool enable = !!(unsigned long)info;
-
-	if (enable)
-		set_in_cr4(X86_CR4_PCE);
-	else
-		clear_in_cr4(X86_CR4_PCE);
-}
-
 static ssize_t set_attr_rdpmc(struct device *cdev,
 			      struct device_attribute *attr,
 			      const char *buf, size_t count)
@@ -1850,14 +1872,27 @@ static ssize_t set_attr_rdpmc(struct device *cdev,
 	if (ret)
 		return ret;
 
+	if (val > 2)
+		return -EINVAL;
+
 	if (x86_pmu.attr_rdpmc_broken)
 		return -ENOTSUPP;
 
-	if (!!val != !!x86_pmu.attr_rdpmc) {
-		x86_pmu.attr_rdpmc = !!val;
-		on_each_cpu(change_rdpmc, (void *)val, 1);
+	if ((val == 2) != (x86_pmu.attr_rdpmc == 2)) {
+		/*
+		 * Changing into or out of always available, aka
+		 * perf-event-bypassing mode.  This path is extremely slow,
+		 * but only root can trigger it, so it's okay.
+		 */
+		if (val == 2)
+			static_key_slow_inc(&rdpmc_always_available);
+		else
+			static_key_slow_dec(&rdpmc_always_available);
+		on_each_cpu(refresh_pce, NULL, 1);
 	}
 
+	x86_pmu.attr_rdpmc = val;
+
 	return count;
 }
 
@@ -1900,6 +1935,9 @@ static struct pmu pmu = {
 
 	.event_init		= x86_pmu_event_init,
 
+	.event_mapped		= x86_pmu_event_mapped,
+	.event_unmapped		= x86_pmu_event_unmapped,
+
 	.add			= x86_pmu_add,
 	.del			= x86_pmu_del,
 	.start			= x86_pmu_start,
@@ -1914,13 +1952,15 @@ static struct pmu pmu = {
 	.flush_branch_stack	= x86_pmu_flush_branch_stack,
 };
 
-void arch_perf_update_userpage(struct perf_event_mmap_page *userpg, u64 now)
+void arch_perf_update_userpage(struct perf_event *event,
+			       struct perf_event_mmap_page *userpg, u64 now)
 {
 	struct cyc2ns_data *data;
 
 	userpg->cap_user_time = 0;
 	userpg->cap_user_time_zero = 0;
-	userpg->cap_user_rdpmc = x86_pmu.attr_rdpmc;
+	userpg->cap_user_rdpmc =
+		!!(event->hw.flags & PERF_X86_EVENT_RDPMC_ALLOWED);
 	userpg->pmc_width = x86_pmu.cntval_bits;
 
 	if (!sched_clock_stable())
diff --git a/arch/x86/kernel/cpu/perf_event.h b/arch/x86/kernel/cpu/perf_event.h
index 4e6cdb0ddc70..df525d2be1e8 100644
--- a/arch/x86/kernel/cpu/perf_event.h
+++ b/arch/x86/kernel/cpu/perf_event.h
@@ -71,6 +71,8 @@ struct event_constraint {
 #define PERF_X86_EVENT_COMMITTED	0x8 /* event passed commit_txn */
 #define PERF_X86_EVENT_PEBS_LD_HSW	0x10 /* haswell style datala, load */
 #define PERF_X86_EVENT_PEBS_NA_HSW	0x20 /* haswell style datala, unknown */
+#define PERF_X86_EVENT_RDPMC_ALLOWED	0x40 /* grant rdpmc permission */
+
 
 struct amd_nb {
 	int nb_id;  /* NorthBridge id */
diff --git a/arch/x86/kernel/head32.c b/arch/x86/kernel/head32.c
index d6c1b9836995..2911ef3a9f1c 100644
--- a/arch/x86/kernel/head32.c
+++ b/arch/x86/kernel/head32.c
@@ -31,6 +31,7 @@ static void __init i386_default_early_setup(void)
 
 asmlinkage __visible void __init i386_start_kernel(void)
 {
+	cr4_init_shadow();
 	sanitize_boot_params(&boot_params);
 
 	/* Call the subarch specific early setup function */
diff --git a/arch/x86/kernel/head64.c b/arch/x86/kernel/head64.c
index efcddfaf05f9..c4f8d4659070 100644
--- a/arch/x86/kernel/head64.c
+++ b/arch/x86/kernel/head64.c
@@ -156,6 +156,8 @@ asmlinkage __visible void __init x86_64_start_kernel(char * real_mode_data)
 				(__START_KERNEL & PGDIR_MASK)));
 	BUILD_BUG_ON(__fix_to_virt(__end_of_fixed_addresses) <= MODULES_END);
 
+	cr4_init_shadow();
+
 	/* Kill off the identity-map trampoline */
 	reset_early_page_tables();
 
diff --git a/arch/x86/kernel/i387.c b/arch/x86/kernel/i387.c
index 81049ffab2d6..d5651fce0b71 100644
--- a/arch/x86/kernel/i387.c
+++ b/arch/x86/kernel/i387.c
@@ -13,6 +13,7 @@
 #include <asm/sigcontext.h>
 #include <asm/processor.h>
 #include <asm/math_emu.h>
+#include <asm/tlbflush.h>
 #include <asm/uaccess.h>
 #include <asm/ptrace.h>
 #include <asm/i387.h>
@@ -193,7 +194,7 @@ void fpu_init(void)
 	if (cpu_has_xmm)
 		cr4_mask |= X86_CR4_OSXMMEXCPT;
 	if (cr4_mask)
-		set_in_cr4(cr4_mask);
+		cr4_set_bits(cr4_mask);
 
 	cr0 = read_cr0();
 	cr0 &= ~(X86_CR0_TS|X86_CR0_EM); /* clear TS and EM */
diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
index e127ddaa2d5a..046e2d620bbe 100644
--- a/arch/x86/kernel/process.c
+++ b/arch/x86/kernel/process.c
@@ -28,6 +28,7 @@
 #include <asm/fpu-internal.h>
 #include <asm/debugreg.h>
 #include <asm/nmi.h>
+#include <asm/tlbflush.h>
 
 /*
  * per-CPU TSS segments. Threads are completely 'soft' on Linux,
@@ -141,7 +142,7 @@ void flush_thread(void)
 
 static void hard_disable_TSC(void)
 {
-	write_cr4(read_cr4() | X86_CR4_TSD);
+	cr4_set_bits(X86_CR4_TSD);
 }
 
 void disable_TSC(void)
@@ -158,7 +159,7 @@ void disable_TSC(void)
 
 static void hard_enable_TSC(void)
 {
-	write_cr4(read_cr4() & ~X86_CR4_TSD);
+	cr4_clear_bits(X86_CR4_TSD);
 }
 
 static void enable_TSC(void)
diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c
index 8f3ebfe710d0..603c4f99cb5a 100644
--- a/arch/x86/kernel/process_32.c
+++ b/arch/x86/kernel/process_32.c
@@ -101,7 +101,7 @@ void __show_regs(struct pt_regs *regs, int all)
 	cr0 = read_cr0();
 	cr2 = read_cr2();
 	cr3 = read_cr3();
-	cr4 = read_cr4_safe();
+	cr4 = __read_cr4_safe();
 	printk(KERN_DEFAULT "CR0: %08lx CR2: %08lx CR3: %08lx CR4: %08lx\n",
 			cr0, cr2, cr3, cr4);
 
diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c
index 5a2c02913af3..67fcc43577d2 100644
--- a/arch/x86/kernel/process_64.c
+++ b/arch/x86/kernel/process_64.c
@@ -93,7 +93,7 @@ void __show_regs(struct pt_regs *regs, int all)
 	cr0 = read_cr0();
 	cr2 = read_cr2();
 	cr3 = read_cr3();
-	cr4 = read_cr4();
+	cr4 = __read_cr4();
 
 	printk(KERN_DEFAULT "FS:  %016lx(%04x) GS:%016lx(%04x) knlGS:%016lx\n",
 	       fs, fsindex, gs, gsindex, shadowgs);
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index 27d200929864..0a2421cca01f 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -1179,7 +1179,7 @@ void __init setup_arch(char **cmdline_p)
 
 	if (boot_cpu_data.cpuid_level >= 0) {
 		/* A CPU has %cr4 if and only if it has CPUID */
-		mmu_cr4_features = read_cr4();
+		mmu_cr4_features = __read_cr4();
 		if (trampoline_cr4_features)
 			*trampoline_cr4_features = mmu_cr4_features;
 	}
diff --git a/arch/x86/kernel/xsave.c b/arch/x86/kernel/xsave.c
index 0de1fae2bdf0..34f66e58a896 100644
--- a/arch/x86/kernel/xsave.c
+++ b/arch/x86/kernel/xsave.c
@@ -12,6 +12,7 @@
 #include <asm/i387.h>
 #include <asm/fpu-internal.h>
 #include <asm/sigframe.h>
+#include <asm/tlbflush.h>
 #include <asm/xcr.h>
 
 /*
@@ -453,7 +454,7 @@ static void prepare_fx_sw_frame(void)
  */
 static inline void xstate_enable(void)
 {
-	set_in_cr4(X86_CR4_OSXSAVE);
+	cr4_set_bits(X86_CR4_OSXSAVE);
 	xsetbv(XCR_XFEATURE_ENABLED_MASK, pcntxt_mask);
 }
 
diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
index a17d848c6d42..d319e0c24758 100644
--- a/arch/x86/kvm/svm.c
+++ b/arch/x86/kvm/svm.c
@@ -1583,7 +1583,7 @@ static void svm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0)
 
 static int svm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4)
 {
-	unsigned long host_cr4_mce = read_cr4() & X86_CR4_MCE;
+	unsigned long host_cr4_mce = cr4_read_shadow() & X86_CR4_MCE;
 	unsigned long old_cr4 = to_svm(vcpu)->vmcb->save.cr4;
 
 	if (cr4 & X86_CR4_VMXE)
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 3f73bfad0349..14c1a18d206a 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -2871,7 +2871,7 @@ static int hardware_enable(void)
 	u64 phys_addr = __pa(per_cpu(vmxarea, cpu));
 	u64 old, test_bits;
 
-	if (read_cr4() & X86_CR4_VMXE)
+	if (cr4_read_shadow() & X86_CR4_VMXE)
 		return -EBUSY;
 
 	INIT_LIST_HEAD(&per_cpu(loaded_vmcss_on_cpu, cpu));
@@ -2898,7 +2898,7 @@ static int hardware_enable(void)
 		/* enable and lock */
 		wrmsrl(MSR_IA32_FEATURE_CONTROL, old | test_bits);
 	}
-	write_cr4(read_cr4() | X86_CR4_VMXE); /* FIXME: not cpu hotplug safe */
+	cr4_set_bits(X86_CR4_VMXE);
 
 	if (vmm_exclusive) {
 		kvm_cpu_vmxon(phys_addr);
@@ -2935,7 +2935,7 @@ static void hardware_disable(void)
 		vmclear_local_loaded_vmcss();
 		kvm_cpu_vmxoff();
 	}
-	write_cr4(read_cr4() & ~X86_CR4_VMXE);
+	cr4_clear_bits(X86_CR4_VMXE);
 }
 
 static __init int adjust_vmx_controls(u32 ctl_min, u32 ctl_opt,
@@ -4450,7 +4450,7 @@ static void vmx_set_constant_host_state(struct vcpu_vmx *vmx)
 	vmcs_writel(HOST_CR3, read_cr3());  /* 22.2.3  FIXME: shadow tables */
 
 	/* Save the most likely value for this task's CR4 in the VMCS. */
-	cr4 = read_cr4();
+	cr4 = cr4_read_shadow();
 	vmcs_writel(HOST_CR4, cr4);			/* 22.2.3, 22.2.5 */
 	vmx->host_state.vmcs_host_cr4 = cr4;
 
@@ -8146,7 +8146,7 @@ static void __noclone vmx_vcpu_run(struct kvm_vcpu *vcpu)
 	if (test_bit(VCPU_REGS_RIP, (unsigned long *)&vcpu->arch.regs_dirty))
 		vmcs_writel(GUEST_RIP, vcpu->arch.regs[VCPU_REGS_RIP]);
 
-	cr4 = read_cr4();
+	cr4 = cr4_read_shadow();
 	if (unlikely(cr4 != vmx->host_state.vmcs_host_cr4)) {
 		vmcs_writel(HOST_CR4, cr4);
 		vmx->host_state.vmcs_host_cr4 = cr4;
diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c
index e3ff27a5b634..ede025fb46f1 100644
--- a/arch/x86/mm/fault.c
+++ b/arch/x86/mm/fault.c
@@ -600,7 +600,7 @@ show_fault_oops(struct pt_regs *regs, unsigned long error_code,
 			printk(nx_warning, from_kuid(&init_user_ns, current_uid()));
 		if (pte && pte_present(*pte) && pte_exec(*pte) &&
 				(pgd_flags(*pgd) & _PAGE_USER) &&
-				(read_cr4() & X86_CR4_SMEP))
+				(__read_cr4() & X86_CR4_SMEP))
 			printk(smep_warning, from_kuid(&init_user_ns, current_uid()));
 	}
 
diff --git a/arch/x86/mm/init.c b/arch/x86/mm/init.c
index 649da47d3827..553c094b9cd7 100644
--- a/arch/x86/mm/init.c
+++ b/arch/x86/mm/init.c
@@ -173,11 +173,11 @@ static void __init probe_page_size_mask(void)
 
 	/* Enable PSE if available */
 	if (cpu_has_pse)
-		set_in_cr4(X86_CR4_PSE);
+		cr4_set_bits_and_update_boot(X86_CR4_PSE);
 
 	/* Enable PGE if available */
 	if (cpu_has_pge) {
-		set_in_cr4(X86_CR4_PGE);
+		cr4_set_bits_and_update_boot(X86_CR4_PGE);
 		__supported_pte_mask |= _PAGE_GLOBAL;
 	}
 }
@@ -713,6 +713,15 @@ void __init zone_sizes_init(void)
 	free_area_init_nodes(max_zone_pfns);
 }
 
+DEFINE_PER_CPU_SHARED_ALIGNED(struct tlb_state, cpu_tlbstate) = {
+#ifdef CONFIG_SMP
+	.active_mm = &init_mm,
+	.state = 0,
+#endif
+	.cr4 = ~0UL,	/* fail hard if we screw up cr4 shadow initialization */
+};
+EXPORT_SYMBOL_GPL(cpu_tlbstate);
+
 void update_cache_mode_entry(unsigned entry, enum page_cache_mode cache)
 {
 	/* entry 0 MUST be WB (hardwired to speed up translations) */
diff --git a/arch/x86/mm/tlb.c b/arch/x86/mm/tlb.c
index ee61c36d64f8..3250f2371aea 100644
--- a/arch/x86/mm/tlb.c
+++ b/arch/x86/mm/tlb.c
@@ -14,9 +14,6 @@
 #include <asm/uv/uv.h>
 #include <linux/debugfs.h>
 
-DEFINE_PER_CPU_SHARED_ALIGNED(struct tlb_state, cpu_tlbstate)
-			= { &init_mm, 0, };
-
 /*
  *	Smarter SMP flushing macros.
  *		c/o Linus Torvalds.
diff --git a/arch/x86/power/cpu.c b/arch/x86/power/cpu.c
index 6ec7910f59bf..3e32ed5648a0 100644
--- a/arch/x86/power/cpu.c
+++ b/arch/x86/power/cpu.c
@@ -105,11 +105,8 @@ static void __save_processor_state(struct saved_context *ctxt)
 	ctxt->cr0 = read_cr0();
 	ctxt->cr2 = read_cr2();
 	ctxt->cr3 = read_cr3();
-#ifdef CONFIG_X86_32
-	ctxt->cr4 = read_cr4_safe();
-#else
-/* CONFIG_X86_64 */
-	ctxt->cr4 = read_cr4();
+	ctxt->cr4 = __read_cr4_safe();
+#ifdef CONFIG_X86_64
 	ctxt->cr8 = read_cr8();
 #endif
 	ctxt->misc_enable_saved = !rdmsrl_safe(MSR_IA32_MISC_ENABLE,
@@ -175,12 +172,12 @@ static void notrace __restore_processor_state(struct saved_context *ctxt)
 	/* cr4 was introduced in the Pentium CPU */
 #ifdef CONFIG_X86_32
 	if (ctxt->cr4)
-		write_cr4(ctxt->cr4);
+		__write_cr4(ctxt->cr4);
 #else
 /* CONFIG X86_64 */
 	wrmsrl(MSR_EFER, ctxt->efer);
 	write_cr8(ctxt->cr8);
-	write_cr4(ctxt->cr4);
+	__write_cr4(ctxt->cr4);
 #endif
 	write_cr3(ctxt->cr3);
 	write_cr2(ctxt->cr2);
diff --git a/arch/x86/realmode/init.c b/arch/x86/realmode/init.c
index bad628a620c4..0b7a63d98440 100644
--- a/arch/x86/realmode/init.c
+++ b/arch/x86/realmode/init.c
@@ -81,7 +81,7 @@ void __init setup_real_mode(void)
 
 	trampoline_header->start = (u64) secondary_startup_64;
 	trampoline_cr4_features = &trampoline_header->cr4;
-	*trampoline_cr4_features = read_cr4();
+	*trampoline_cr4_features = __read_cr4();
 
 	trampoline_pgd = (u64 *) __va(real_mode_header->trampoline_pgd);
 	trampoline_pgd[0] = init_level4_pgt[pgd_index(__PAGE_OFFSET)].pgd;
diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index 78a881b7fc41..bd8b8459c3d0 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -1494,10 +1494,10 @@ static void xen_pvh_set_cr_flags(int cpu)
 	 * set them here. For all, OSFXSR OSXMMEXCPT are set in fpu_init.
 	*/
 	if (cpu_has_pse)
-		set_in_cr4(X86_CR4_PSE);
+		cr4_set_bits_and_update_boot(X86_CR4_PSE);
 
 	if (cpu_has_pge)
-		set_in_cr4(X86_CR4_PGE);
+		cr4_set_bits_and_update_boot(X86_CR4_PGE);
 }
 
 /*
diff --git a/drivers/lguest/x86/core.c b/drivers/lguest/x86/core.c
index 922a1acbf652..6adfd7ba4c97 100644
--- a/drivers/lguest/x86/core.c
+++ b/drivers/lguest/x86/core.c
@@ -47,6 +47,7 @@
 #include <asm/lguest.h>
 #include <asm/uaccess.h>
 #include <asm/i387.h>
+#include <asm/tlbflush.h>
 #include "../lg.h"
 
 static int cpu_had_pge;
@@ -452,9 +453,9 @@ void lguest_arch_handle_trap(struct lg_cpu *cpu)
 static void adjust_pge(void *on)
 {
 	if (on)
-		write_cr4(read_cr4() | X86_CR4_PGE);
+		cr4_set_bits(X86_CR4_PGE);
 	else
-		write_cr4(read_cr4() & ~X86_CR4_PGE);
+		cr4_clear_bits(X86_CR4_PGE);
 }
 
 /*H:020
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index 2cdc9d422bed..2b621982938d 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -202,6 +202,13 @@ struct pmu {
 	 */
 	int (*event_init)		(struct perf_event *event);
 
+	/*
+	 * Notification that the event was mapped or unmapped.  Called
+	 * in the context of the mapping task.
+	 */
+	void (*event_mapped)		(struct perf_event *event); /*optional*/
+	void (*event_unmapped)		(struct perf_event *event); /*optional*/
+
 #define PERF_EF_START	0x01		/* start the counter when adding    */
 #define PERF_EF_RELOAD	0x02		/* reload the counter when starting */
 #define PERF_EF_UPDATE	0x04		/* update the counter when stopping */
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 8812d8e35f5b..f04daabfd1cf 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -4101,7 +4101,8 @@ unlock:
 	rcu_read_unlock();
 }
 
-void __weak arch_perf_update_userpage(struct perf_event_mmap_page *userpg, u64 now)
+void __weak arch_perf_update_userpage(
+	struct perf_event *event, struct perf_event_mmap_page *userpg, u64 now)
 {
 }
 
@@ -4151,7 +4152,7 @@ void perf_event_update_userpage(struct perf_event *event)
 	userpg->time_running = running +
 			atomic64_read(&event->child_total_time_running);
 
-	arch_perf_update_userpage(userpg, now);
+	arch_perf_update_userpage(event, userpg, now);
 
 	barrier();
 	++userpg->lock;
@@ -4293,6 +4294,9 @@ static void perf_mmap_open(struct vm_area_struct *vma)
 
 	atomic_inc(&event->mmap_count);
 	atomic_inc(&event->rb->mmap_count);
+
+	if (event->pmu->event_mapped)
+		event->pmu->event_mapped(event);
 }
 
 /*
@@ -4312,6 +4316,9 @@ static void perf_mmap_close(struct vm_area_struct *vma)
 	int mmap_locked = rb->mmap_locked;
 	unsigned long size = perf_data_size(rb);
 
+	if (event->pmu->event_unmapped)
+		event->pmu->event_unmapped(event);
+
 	atomic_dec(&rb->mmap_count);
 
 	if (!atomic_dec_and_mutex_lock(&event->mmap_count, &event->mmap_mutex))
@@ -4513,6 +4520,9 @@ unlock:
 	vma->vm_flags |= VM_DONTCOPY | VM_DONTEXPAND | VM_DONTDUMP;
 	vma->vm_ops = &perf_mmap_vmops;
 
+	if (event->pmu->event_mapped)
+		event->pmu->event_mapped(event);
+
 	return ret;
 }