Merge remote branch 'origin/x86/urgent' into x86/asm

11 files changed, 116 insertions, 98 deletions

diff --git a/arch/x86/include/asm/cmpxchg_32.h b/arch/x86/include/asm/cmpxchg_32.h
index 8859e12..c1cf59d 100644
--- a/arch/x86/include/asm/cmpxchg_32.h
+++ b/arch/x86/include/asm/cmpxchg_32.h
@@ -27,20 +27,20 @@ struct __xchg_dummy {
 	switch (size) {							\
 	case 1:								\
 		asm volatile("xchgb %b0,%1"				\
-			     : "=q" (__x)				\
-			     : "m" (*__xg(ptr)), "0" (__x)		\
+			     : "=q" (__x), "+m" (*__xg(ptr))		\
+			     : "0" (__x)				\
 			     : "memory");				\
 		break;							\
 	case 2:								\
 		asm volatile("xchgw %w0,%1"				\
-			     : "=r" (__x)				\
-			     : "m" (*__xg(ptr)), "0" (__x)		\
+			     : "=r" (__x), "+m" (*__xg(ptr))		\
+			     : "0" (__x)				\
 			     : "memory");				\
 		break;							\
 	case 4:								\
 		asm volatile("xchgl %0,%1"				\
-			     : "=r" (__x)				\
-			     : "m" (*__xg(ptr)), "0" (__x)		\
+			     : "=r" (__x), "+m" (*__xg(ptr))		\
+			     : "0" (__x)				\
 			     : "memory");				\
 		break;							\
 	default:							\
@@ -70,14 +70,14 @@ static inline void __set_64bit(unsigned long long *ptr,
 			       unsigned int low, unsigned int high)
 {
 	asm volatile("\n1:\t"
-		     "movl (%0), %%eax\n\t"
-		     "movl 4(%0), %%edx\n\t"
-		     LOCK_PREFIX "cmpxchg8b (%0)\n\t"
+		     "movl (%1), %%eax\n\t"
+		     "movl 4(%1), %%edx\n\t"
+		     LOCK_PREFIX "cmpxchg8b (%1)\n\t"
 		     "jnz 1b"
-		     : /* no outputs */
-		     : "D"(ptr),
-		       "b"(low),
-		       "c"(high)
+		     : "=m" (*ptr)
+		     : "D" (ptr),
+		       "b" (low),
+		       "c" (high)
 		     : "ax", "dx", "memory");
 }
 
@@ -121,21 +121,21 @@ extern void __cmpxchg_wrong_size(void);
 	__typeof__(*(ptr)) __new = (new);				\
 	switch (size) {							\
 	case 1:								\
-		asm volatile(lock "cmpxchgb %b1,%2"			\
-			     : "=a"(__ret)				\
-			     : "q"(__new), "m"(*__xg(ptr)), "0"(__old)	\
+		asm volatile(lock "cmpxchgb %b2,%1"			\
+			     : "=a" (__ret), "+m" (*__xg(ptr))		\
+			     : "q" (__new), "0" (__old)			\
 			     : "memory");				\
 		break;							\
 	case 2:								\
-		asm volatile(lock "cmpxchgw %w1,%2"			\
-			     : "=a"(__ret)				\
-			     : "r"(__new), "m"(*__xg(ptr)), "0"(__old)	\
+		asm volatile(lock "cmpxchgw %w2,%1"			\
+			     : "=a" (__ret), "+m" (*__xg(ptr))		\
+			     : "r" (__new), "0" (__old)			\
 			     : "memory");				\
 		break;							\
 	case 4:								\
-		asm volatile(lock "cmpxchgl %1,%2"			\
-			     : "=a"(__ret)				\
-			     : "r"(__new), "m"(*__xg(ptr)), "0"(__old)	\
+		asm volatile(lock "cmpxchgl %2,%1"			\
+			     : "=a" (__ret), "+m" (*__xg(ptr))		\
+			     : "r" (__new), "0" (__old)			\
 			     : "memory");				\
 		break;							\
 	default:							\
@@ -180,12 +180,12 @@ static inline unsigned long long __cmpxchg64(volatile void *ptr,
 					     unsigned long long new)
 {
 	unsigned long long prev;
-	asm volatile(LOCK_PREFIX "cmpxchg8b %3"
-		     : "=A"(prev)
-		     : "b"((unsigned long)new),
-		       "c"((unsigned long)(new >> 32)),
-		       "m"(*__xg(ptr)),
-		       "0"(old)
+	asm volatile(LOCK_PREFIX "cmpxchg8b %1"
+		     : "=A" (prev),
+		       "+m" (*__xg(ptr))
+		     : "b" ((unsigned long)new),
+		       "c" ((unsigned long)(new >> 32)),
+		       "0" (old)
 		     : "memory");
 	return prev;
 }
@@ -195,12 +195,12 @@ static inline unsigned long long __cmpxchg64_local(volatile void *ptr,
 						   unsigned long long new)
 {
 	unsigned long long prev;
-	asm volatile("cmpxchg8b %3"
-		     : "=A"(prev)
-		     : "b"((unsigned long)new),
-		       "c"((unsigned long)(new >> 32)),
-		       "m"(*__xg(ptr)),
-		       "0"(old)
+	asm volatile("cmpxchg8b %1"
+		     : "=A" (prev),
+		       "+m" (*__xg(ptr))
+		     : "b" ((unsigned long)new),
+		       "c" ((unsigned long)(new >> 32)),
+		       "0" (old)
 		     : "memory");
 	return prev;
 }
diff --git a/arch/x86/include/asm/cmpxchg_64.h b/arch/x86/include/asm/cmpxchg_64.h
index 485ae41..b92f147 100644
--- a/arch/x86/include/asm/cmpxchg_64.h
+++ b/arch/x86/include/asm/cmpxchg_64.h
@@ -26,26 +26,26 @@ extern void __cmpxchg_wrong_size(void);
 	switch (size) {							\
 	case 1:								\
 		asm volatile("xchgb %b0,%1"				\
-			     : "=q" (__x)				\
-			     : "m" (*__xg(ptr)), "0" (__x)		\
+			     : "=q" (__x), "+m" (*__xg(ptr))		\
+			     : "0" (__x)				\
 			     : "memory");				\
 		break;							\
 	case 2:								\
 		asm volatile("xchgw %w0,%1"				\
-			     : "=r" (__x)				\
-			     : "m" (*__xg(ptr)), "0" (__x)		\
+			     : "=r" (__x), "+m" (*__xg(ptr))		\
+			     : "0" (__x)				\
 			     : "memory");				\
 		break;							\
 	case 4:								\
 		asm volatile("xchgl %k0,%1"				\
-			     : "=r" (__x)				\
-			     : "m" (*__xg(ptr)), "0" (__x)		\
+			     : "=r" (__x), "+m" (*__xg(ptr))		\
+			     : "0" (__x)				\
 			     : "memory");				\
 		break;							\
 	case 8:								\
 		asm volatile("xchgq %0,%1"				\
-			     : "=r" (__x)				\
-			     : "m" (*__xg(ptr)), "0" (__x)		\
+			     : "=r" (__x), "+m" (*__xg(ptr))		\
+			     : "0" (__x)				\
 			     : "memory");				\
 		break;							\
 	default:							\
@@ -71,27 +71,27 @@ extern void __cmpxchg_wrong_size(void);
 	__typeof__(*(ptr)) __new = (new);				\
 	switch (size) {							\
 	case 1:								\
-		asm volatile(lock "cmpxchgb %b1,%2"			\
-			     : "=a"(__ret)				\
-			     : "q"(__new), "m"(*__xg(ptr)), "0"(__old)	\
+		asm volatile(lock "cmpxchgb %b2,%1"			\
+			     : "=a" (__ret), "+m" (*__xg(ptr))		\
+			     : "q" (__new), "0" (__old)			\
 			     : "memory");				\
 		break;							\
 	case 2:								\
-		asm volatile(lock "cmpxchgw %w1,%2"			\
-			     : "=a"(__ret)				\
-			     : "r"(__new), "m"(*__xg(ptr)), "0"(__old)	\
+		asm volatile(lock "cmpxchgw %w2,%1"			\
+			     : "=a" (__ret), "+m" (*__xg(ptr))		\
+			     : "r" (__new), "0" (__old)			\
 			     : "memory");				\
 		break;							\
 	case 4:								\
-		asm volatile(lock "cmpxchgl %k1,%2"			\
-			     : "=a"(__ret)				\
-			     : "r"(__new), "m"(*__xg(ptr)), "0"(__old)	\
+		asm volatile(lock "cmpxchgl %k2,%1"			\
+			     : "=a" (__ret), "+m" (*__xg(ptr))		\
+			     : "r" (__new), "0" (__old)			\
 			     : "memory");				\
 		break;							\
 	case 8:								\
-		asm volatile(lock "cmpxchgq %1,%2"			\
-			     : "=a"(__ret)				\
-			     : "r"(__new), "m"(*__xg(ptr)), "0"(__old)	\
+		asm volatile(lock "cmpxchgq %2,%1"			\
+			     : "=a" (__ret), "+m" (*__xg(ptr))		\
+			     : "r" (__new), "0" (__old)			\
 			     : "memory");				\
 		break;							\
 	default:							\
diff --git a/arch/x86/kernel/acpi/cstate.c b/arch/x86/kernel/acpi/cstate.c
index 2e837f5..fb7a5f0 100644
--- a/arch/x86/kernel/acpi/cstate.c
+++ b/arch/x86/kernel/acpi/cstate.c
@@ -145,6 +145,15 @@ int acpi_processor_ffh_cstate_probe(unsigned int cpu,
 		percpu_entry->states[cx->index].eax = cx->address;
 		percpu_entry->states[cx->index].ecx = MWAIT_ECX_INTERRUPT_BREAK;
 	}
+
+	/*
+	 * For _CST FFH on Intel, if GAS.access_size bit 1 is cleared,
+	 * then we should skip checking BM_STS for this C-state.
+	 * ref: "Intel Processor Vendor-Specific ACPI Interface Specification"
+	 */
+	if ((c->x86_vendor == X86_VENDOR_INTEL) && !(reg->access_size & 0x2))
+		cx->bm_sts_skip = 1;
+
 	return retval;
 }
 EXPORT_SYMBOL_GPL(acpi_processor_ffh_cstate_probe);
diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c
index 82e5086..fcc3c61f 100644
--- a/arch/x86/kernel/acpi/sleep.c
+++ b/arch/x86/kernel/acpi/sleep.c
@@ -157,9 +157,14 @@ static int __init acpi_sleep_setup(char *str)
 #ifdef CONFIG_HIBERNATION
 		if (strncmp(str, "s4_nohwsig", 10) == 0)
 			acpi_no_s4_hw_signature();
-		if (strncmp(str, "s4_nonvs", 8) == 0)
-			acpi_s4_no_nvs();
+		if (strncmp(str, "s4_nonvs", 8) == 0) {
+			pr_warning("ACPI: acpi_sleep=s4_nonvs is deprecated, "
+					"please use acpi_sleep=nonvs instead");
+			acpi_nvs_nosave();
+		}
 #endif
+		if (strncmp(str, "nonvs", 5) == 0)
+			acpi_nvs_nosave();
 		if (strncmp(str, "old_ordering", 12) == 0)
 			acpi_old_suspend_ordering();
 		str = strchr(str, ',');
diff --git a/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c
index ce7cde7..a36de5b 100644
--- a/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c
+++ b/arch/x86/kernel/cpu/cpufreq/pcc-cpufreq.c
@@ -368,22 +368,16 @@ static int __init pcc_cpufreq_do_osc(acpi_handle *handle)
 		return -ENODEV;
 
 	out_obj = output.pointer;
-	if (out_obj->type != ACPI_TYPE_BUFFER) {
-		ret = -ENODEV;
-		goto out_free;
-	}
+	if (out_obj->type != ACPI_TYPE_BUFFER)
+		return -ENODEV;
 
 	errors = *((u32 *)out_obj->buffer.pointer) & ~(1 << 0);
-	if (errors) {
-		ret = -ENODEV;
-		goto out_free;
-	}
+	if (errors)
+		return -ENODEV;
 
 	supported = *((u32 *)(out_obj->buffer.pointer + 4));
-	if (!(supported & 0x1)) {
-		ret = -ENODEV;
-		goto out_free;
-	}
+	if (!(supported & 0x1))
+		return -ENODEV;
 
 out_free:
 	kfree(output.pointer);
@@ -397,13 +391,17 @@ static int __init pcc_cpufreq_probe(void)
 	struct pcc_memory_resource *mem_resource;
 	struct pcc_register_resource *reg_resource;
 	union acpi_object *out_obj, *member;
-	acpi_handle handle, osc_handle;
+	acpi_handle handle, osc_handle, pcch_handle;
 	int ret = 0;
 
 	status = acpi_get_handle(NULL, "\\_SB", &handle);
 	if (ACPI_FAILURE(status))
 		return -ENODEV;
 
+	status = acpi_get_handle(handle, "PCCH", &pcch_handle);
+	if (ACPI_FAILURE(status))
+		return -ENODEV;
+
 	status = acpi_get_handle(handle, "_OSC", &osc_handle);
 	if (ACPI_SUCCESS(status)) {
 		ret = pcc_cpufreq_do_osc(&osc_handle);
@@ -543,13 +541,13 @@ static int pcc_cpufreq_cpu_init(struct cpufreq_policy *policy)
 
 	if (!pcch_virt_addr) {
 		result = -1;
-		goto pcch_null;
+		goto out;
 	}
 
 	result = pcc_get_offset(cpu);
 	if (result) {
 		dprintk("init: PCCP evaluation failed\n");
-		goto free;
+		goto out;
 	}
 
 	policy->max = policy->cpuinfo.max_freq =
@@ -558,14 +556,15 @@ static int pcc_cpufreq_cpu_init(struct cpufreq_policy *policy)
 		ioread32(&pcch_hdr->minimum_frequency) * 1000;
 	policy->cur = pcc_get_freq(cpu);
 
+	if (!policy->cur) {
+		dprintk("init: Unable to get current CPU frequency\n");
+		result = -EINVAL;
+		goto out;
+	}
+
 	dprintk("init: policy->max is %d, policy->min is %d\n",
 		policy->max, policy->min);
-
-	return 0;
-free:
-	pcc_clear_mapping();
-	free_percpu(pcc_cpu_info);
-pcch_null:
+out:
 	return result;
 }
 
diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
index 7ec2123..3e90cce 100644
--- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
+++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c
@@ -1023,13 +1023,12 @@ static int get_transition_latency(struct powernow_k8_data *data)
 	}
 	if (max_latency == 0) {
 		/*
-		 * Fam 11h always returns 0 as transition latency.
-		 * This is intended and means "very fast". While cpufreq core
-		 * and governors currently can handle that gracefully, better
-		 * set it to 1 to avoid problems in the future.
-		 * For all others it's a BIOS bug.
+		 * Fam 11h and later may return 0 as transition latency. This
+		 * is intended and means "very fast". While cpufreq core and
+		 * governors currently can handle that gracefully, better set it
+		 * to 1 to avoid problems in the future.
 		 */
-		if (boot_cpu_data.x86 != 0x11)
+		if (boot_cpu_data.x86 < 0x11)
 			printk(KERN_ERR FW_WARN PFX "Invalid zero transition "
 				"latency\n");
 		max_latency = 1;
diff --git a/arch/x86/kernel/entry_64.S b/arch/x86/kernel/entry_64.S
index 0697ff1..4db7c4d 100644
--- a/arch/x86/kernel/entry_64.S
+++ b/arch/x86/kernel/entry_64.S
@@ -571,8 +571,8 @@ auditsys:
 	 * masked off.
 	 */
 sysret_audit:
-	movq %rax,%rsi		/* second arg, syscall return value */
-	cmpq $0,%rax		/* is it < 0? */
+	movq RAX-ARGOFFSET(%rsp),%rsi	/* second arg, syscall return value */
+	cmpq $0,%rsi		/* is it < 0? */
 	setl %al		/* 1 if so, 0 if not */
 	movzbl %al,%edi		/* zero-extend that into %edi */
 	inc %edi /* first arg, 0->1(AUDITSC_SUCCESS), 1->2(AUDITSC_FAILURE) */
diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c
index a198b7c..ba390d7 100644
--- a/arch/x86/kernel/hpet.c
+++ b/arch/x86/kernel/hpet.c
@@ -964,7 +964,7 @@ fs_initcall(hpet_late_init);
 
 void hpet_disable(void)
 {
-	if (is_hpet_capable()) {
+	if (is_hpet_capable() && hpet_virt_address) {
 		unsigned int cfg = hpet_readl(HPET_CFG);
 
 		if (hpet_legacy_int_enabled) {
diff --git a/arch/x86/kernel/i8259.c b/arch/x86/kernel/i8259.c
index 7c9f02c..cafa7c8 100644
--- a/arch/x86/kernel/i8259.c
+++ b/arch/x86/kernel/i8259.c
@@ -276,16 +276,6 @@ static struct sys_device device_i8259A = {
 	.cls	= &i8259_sysdev_class,
 };
 
-static int __init i8259A_init_sysfs(void)
-{
-	int error = sysdev_class_register(&i8259_sysdev_class);
-	if (!error)
-		error = sysdev_register(&device_i8259A);
-	return error;
-}
-
-device_initcall(i8259A_init_sysfs);
-
 static void mask_8259A(void)
 {
 	unsigned long flags;
@@ -407,3 +397,18 @@ struct legacy_pic default_legacy_pic = {
 };
 
 struct legacy_pic *legacy_pic = &default_legacy_pic;
+
+static int __init i8259A_init_sysfs(void)
+{
+	int error;
+
+	if (legacy_pic != &default_legacy_pic)
+		return 0;
+
+	error = sysdev_class_register(&i8259_sysdev_class);
+	if (!error)
+		error = sysdev_register(&device_i8259A);
+	return error;
+}
+
+device_initcall(i8259A_init_sysfs);
diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h
index 89d66ca..2331bdc 100644
--- a/arch/x86/kvm/paging_tmpl.h
+++ b/arch/x86/kvm/paging_tmpl.h
@@ -342,6 +342,7 @@ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr,
 			/* advance table_gfn when emulating 1gb pages with 4k */
 			if (delta == 0)
 				table_gfn += PT_INDEX(addr, level);
+			access &= gw->pte_access;
 		} else {
 			direct = 0;
 			table_gfn = gw->table_gfn[level - 2];
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 6127468..33f0a84 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -1562,7 +1562,7 @@ static int msr_io(struct kvm_vcpu *vcpu, struct kvm_msrs __user *user_msrs,
 
 	r = -ENOMEM;
 	size = sizeof(struct kvm_msr_entry) * msrs.nmsrs;
-	entries = vmalloc(size);
+	entries = kmalloc(size, GFP_KERNEL);
 	if (!entries)
 		goto out;
 
@@ -1581,7 +1581,7 @@ static int msr_io(struct kvm_vcpu *vcpu, struct kvm_msrs __user *user_msrs,
 	r = n;
 
 out_free:
-	vfree(entries);
+	kfree(entries);
 out:
 	return r;
 }