---

yaml
---
r: 142831
b: refs/heads/master
c: db954b5
h: refs/heads/master
i:
  142829: f852ea8
  142827: bf2d851
  142823: 00c2f4f
  142815: 61d1d0d
v: v3

[refs]

@@ -1,2 +1,2 @@
 ---
-refs/heads/master: 18b2646fe3babeb40b34a0c1751e0bf5adfdc64c
+refs/heads/master: db954b5898dd3ef3ef93f4144158ea8f97deb058

trunk/arch/x86/include/asm/cpufeature.h

@@ -154,6 +154,7 @@
  * CPUID levels like 0x6, 0xA etc
  */
 #define X86_FEATURE_IDA		(7*32+ 0) /* Intel Dynamic Acceleration */
+#define X86_FEATURE_ARAT	(7*32+ 1) /* Always Running APIC Timer */
 
 /* Virtualization flags: Linux defined */
 #define X86_FEATURE_TPR_SHADOW  (8*32+ 0) /* Intel TPR Shadow */

trunk/arch/x86/kernel/apic/apic.c

@@ -431,6 +431,12 @@ static void __cpuinit setup_APIC_timer(void)
 {
 	struct clock_event_device *levt = &__get_cpu_var(lapic_events);
 
+	if (cpu_has(&current_cpu_data, X86_FEATURE_ARAT)) {
+		lapic_clockevent.features &= ~CLOCK_EVT_FEAT_C3STOP;
+		/* Make LAPIC timer preferrable over percpu HPET */
+		lapic_clockevent.rating = 150;
+	}
+
 	memcpy(levt, &lapic_clockevent, sizeof(*levt));
 	levt->cpumask = cpumask_of(smp_processor_id());
 

trunk/arch/x86/kernel/cpu/addon_cpuid_features.c

@@ -31,6 +31,7 @@ void __cpuinit init_scattered_cpuid_features(struct cpuinfo_x86 *c)
 
 	static const struct cpuid_bit __cpuinitconst cpuid_bits[] = {
 		{ X86_FEATURE_IDA, CR_EAX, 1, 0x00000006 },
+		{ X86_FEATURE_ARAT, CR_EAX, 2, 0x00000006 },
 		{ 0, 0, 0, 0 }
 	};
 

trunk/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c

@@ -68,7 +68,6 @@ struct acpi_cpufreq_data {
 	unsigned int max_freq;
 	unsigned int resume;
 	unsigned int cpu_feature;
-	u64 saved_aperf, saved_mperf;
 };
 
 static DEFINE_PER_CPU(struct acpi_cpufreq_data *, drv_data);
@@ -242,23 +241,26 @@ static u32 get_cur_val(const struct cpumask *mask)
 	return cmd.val;
 }
 
-struct perf_pair {
+struct perf_cur {
 	union {
 		struct {
 			u32 lo;
 			u32 hi;
 		} split;
 		u64 whole;
-	} aperf, mperf;
+	} aperf_cur, mperf_cur;
 };
 
 
 static long read_measured_perf_ctrs(void *_cur)
 {
-	struct perf_pair *cur = _cur;
+	struct perf_cur *cur = _cur;
 
-	rdmsr(MSR_IA32_APERF, cur->aperf.split.lo, cur->aperf.split.hi);
-	rdmsr(MSR_IA32_MPERF, cur->mperf.split.lo, cur->mperf.split.hi);
+	rdmsr(MSR_IA32_APERF, cur->aperf_cur.split.lo, cur->aperf_cur.split.hi);
+	rdmsr(MSR_IA32_MPERF, cur->mperf_cur.split.lo, cur->mperf_cur.split.hi);
+
+	wrmsr(MSR_IA32_APERF, 0, 0);
+	wrmsr(MSR_IA32_MPERF, 0, 0);
 
 	return 0;
 }
@@ -279,57 +281,52 @@ static long read_measured_perf_ctrs(void *_cur)
 static unsigned int get_measured_perf(struct cpufreq_policy *policy,
 				      unsigned int cpu)
 {
-	struct perf_pair readin, cur;
+	struct perf_cur cur;
 	unsigned int perf_percent;
 	unsigned int retval;
 
-	if (!work_on_cpu(cpu, read_measured_perf_ctrs, &readin))
+	if (!work_on_cpu(cpu, read_measured_perf_ctrs, &cur))
 		return 0;
 
-	cur.aperf.whole = readin.aperf.whole -
-				per_cpu(drv_data, cpu)->saved_aperf;
-	cur.mperf.whole = readin.mperf.whole -
-				per_cpu(drv_data, cpu)->saved_mperf;
-	per_cpu(drv_data, cpu)->saved_aperf = readin.aperf.whole;
-	per_cpu(drv_data, cpu)->saved_mperf = readin.mperf.whole;
-
 #ifdef __i386__
 	/*
 	 * We dont want to do 64 bit divide with 32 bit kernel
 	 * Get an approximate value. Return failure in case we cannot get
 	 * an approximate value.
 	 */
-	if (unlikely(cur.aperf.split.hi || cur.mperf.split.hi)) {
+	if (unlikely(cur.aperf_cur.split.hi || cur.mperf_cur.split.hi)) {
 		int shift_count;
 		u32 h;
 
-		h = max_t(u32, cur.aperf.split.hi, cur.mperf.split.hi);
+		h = max_t(u32, cur.aperf_cur.split.hi, cur.mperf_cur.split.hi);
 		shift_count = fls(h);
 
-		cur.aperf.whole >>= shift_count;
-		cur.mperf.whole >>= shift_count;
+		cur.aperf_cur.whole >>= shift_count;
+		cur.mperf_cur.whole >>= shift_count;
 	}
 
-	if (((unsigned long)(-1) / 100) < cur.aperf.split.lo) {
+	if (((unsigned long)(-1) / 100) < cur.aperf_cur.split.lo) {
 		int shift_count = 7;
-		cur.aperf.split.lo >>= shift_count;
-		cur.mperf.split.lo >>= shift_count;
+		cur.aperf_cur.split.lo >>= shift_count;
+		cur.mperf_cur.split.lo >>= shift_count;
 	}
 
-	if (cur.aperf.split.lo && cur.mperf.split.lo)
-		perf_percent = (cur.aperf.split.lo * 100) / cur.mperf.split.lo;
+	if (cur.aperf_cur.split.lo && cur.mperf_cur.split.lo)
+		perf_percent = (cur.aperf_cur.split.lo * 100) /
+				cur.mperf_cur.split.lo;
 	else
 		perf_percent = 0;
 
 #else
-	if (unlikely(((unsigned long)(-1) / 100) < cur.aperf.whole)) {
+	if (unlikely(((unsigned long)(-1) / 100) < cur.aperf_cur.whole)) {
 		int shift_count = 7;
-		cur.aperf.whole >>= shift_count;
-		cur.mperf.whole >>= shift_count;
+		cur.aperf_cur.whole >>= shift_count;
+		cur.mperf_cur.whole >>= shift_count;
 	}
 
-	if (cur.aperf.whole && cur.mperf.whole)
-		perf_percent = (cur.aperf.whole * 100) / cur.mperf.whole;
+	if (cur.aperf_cur.whole && cur.mperf_cur.whole)
+		perf_percent = (cur.aperf_cur.whole * 100) /
+				cur.mperf_cur.whole;
 	else
 		perf_percent = 0;
 

trunk/drivers/acpi/processor_idle.c

@@ -145,6 +145,9 @@ static void acpi_timer_check_state(int state, struct acpi_processor *pr,
 	struct acpi_processor_power *pwr = &pr->power;
 	u8 type = local_apic_timer_c2_ok ? ACPI_STATE_C3 : ACPI_STATE_C2;
 
+	if (cpu_has(&cpu_data(pr->id), X86_FEATURE_ARAT))
+		return;
+
 	/*
 	 * Check, if one of the previous states already marked the lapic
 	 * unstable