Merge branch 'linus' into perfcounters/core

Merge reason: This brach was on -rc1, refresh it to almost-rc4 to pick up
              the latest upstream fixes.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

4 files changed, 54 insertions, 49 deletions

diff --git a/arch/x86/kernel/cpu/addon_cpuid_features.c b/arch/x86/kernel/cpu/addon_cpuid_features.c
index 8220ae6..c965e52 100644
--- a/arch/x86/kernel/cpu/addon_cpuid_features.c
+++ b/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 }
 	};
 
diff --git a/arch/x86/kernel/cpu/cpu_debug.c b/arch/x86/kernel/cpu/cpu_debug.c
index 46e29ab..46e29ab 100755..100644
--- a/arch/x86/kernel/cpu/cpu_debug.c
+++ b/arch/x86/kernel/cpu/cpu_debug.c
diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
index 19f6b9d..208ecf6 100644
--- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
+++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c
@@ -65,13 +65,18 @@ enum {
 struct acpi_cpufreq_data {
 	struct acpi_processor_performance *acpi_data;
 	struct cpufreq_frequency_table *freq_table;
-	unsigned int max_freq;
 	unsigned int resume;
 	unsigned int cpu_feature;
 };
 
 static DEFINE_PER_CPU(struct acpi_cpufreq_data *, drv_data);
 
+struct acpi_msr_data {
+	u64 saved_aperf, saved_mperf;
+};
+
+static DEFINE_PER_CPU(struct acpi_msr_data, msr_data);
+
 DEFINE_TRACE(power_mark);
 
 /* acpi_perf_data is a pointer to percpu data. */
@@ -152,7 +157,8 @@ struct drv_cmd {
 	u32 val;
 };
 
-static long do_drv_read(void *_cmd)
+/* Called via smp_call_function_single(), on the target CPU */
+static void do_drv_read(void *_cmd)
 {
 	struct drv_cmd *cmd = _cmd;
 	u32 h;
@@ -169,10 +175,10 @@ static long do_drv_read(void *_cmd)
 	default:
 		break;
 	}
-	return 0;
 }
 
-static long do_drv_write(void *_cmd)
+/* Called via smp_call_function_many(), on the target CPUs */
+static void do_drv_write(void *_cmd)
 {
 	struct drv_cmd *cmd = _cmd;
 	u32 lo, hi;
@@ -191,23 +197,24 @@ static long do_drv_write(void *_cmd)
 	default:
 		break;
 	}
-	return 0;
 }
 
 static void drv_read(struct drv_cmd *cmd)
 {
 	cmd->val = 0;
 
-	work_on_cpu(cpumask_any(cmd->mask), do_drv_read, cmd);
+	smp_call_function_single(cpumask_any(cmd->mask), do_drv_read, cmd, 1);
 }
 
 static void drv_write(struct drv_cmd *cmd)
 {
-	unsigned int i;
+	int this_cpu;
 
-	for_each_cpu(i, cmd->mask) {
-		work_on_cpu(i, do_drv_write, cmd);
-	}
+	this_cpu = get_cpu();
+	if (cpumask_test_cpu(this_cpu, cmd->mask))
+		do_drv_write(cmd);
+	smp_call_function_many(cmd->mask, do_drv_write, cmd, 1);
+	put_cpu();
 }
 
 static u32 get_cur_val(const struct cpumask *mask)
@@ -241,28 +248,23 @@ static u32 get_cur_val(const struct cpumask *mask)
 	return cmd.val;
 }
 
-struct perf_cur {
+struct perf_pair {
 	union {
 		struct {
 			u32 lo;
 			u32 hi;
 		} split;
 		u64 whole;
-	} aperf_cur, mperf_cur;
+	} aperf, mperf;
 };
 
-
-static long read_measured_perf_ctrs(void *_cur)
+/* Called via smp_call_function_single(), on the target CPU */
+static void read_measured_perf_ctrs(void *_cur)
 {
-	struct perf_cur *cur = _cur;
-
-	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);
+	struct perf_pair *cur = _cur;
 
-	wrmsr(MSR_IA32_APERF, 0, 0);
-	wrmsr(MSR_IA32_MPERF, 0, 0);
-
-	return 0;
+	rdmsr(MSR_IA32_APERF, cur->aperf.split.lo, cur->aperf.split.hi);
+	rdmsr(MSR_IA32_MPERF, cur->mperf.split.lo, cur->mperf.split.hi);
 }
 
 /*
@@ -281,58 +283,63 @@ static long read_measured_perf_ctrs(void *_cur)
 static unsigned int get_measured_perf(struct cpufreq_policy *policy,
 				      unsigned int cpu)
 {
-	struct perf_cur cur;
+	struct perf_pair readin, cur;
 	unsigned int perf_percent;
 	unsigned int retval;
 
-	if (!work_on_cpu(cpu, read_measured_perf_ctrs, &cur))
+	if (smp_call_function_single(cpu, read_measured_perf_ctrs, &readin, 1))
 		return 0;
 
+	cur.aperf.whole = readin.aperf.whole -
+				per_cpu(msr_data, cpu).saved_aperf;
+	cur.mperf.whole = readin.mperf.whole -
+				per_cpu(msr_data, cpu).saved_mperf;
+	per_cpu(msr_data, cpu).saved_aperf = readin.aperf.whole;
+	per_cpu(msr_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_cur.split.hi || cur.mperf_cur.split.hi)) {
+	if (unlikely(cur.aperf.split.hi || cur.mperf.split.hi)) {
 		int shift_count;
 		u32 h;
 
-		h = max_t(u32, cur.aperf_cur.split.hi, cur.mperf_cur.split.hi);
+		h = max_t(u32, cur.aperf.split.hi, cur.mperf.split.hi);
 		shift_count = fls(h);
 
-		cur.aperf_cur.whole >>= shift_count;
-		cur.mperf_cur.whole >>= shift_count;
+		cur.aperf.whole >>= shift_count;
+		cur.mperf.whole >>= shift_count;
 	}
 
-	if (((unsigned long)(-1) / 100) < cur.aperf_cur.split.lo) {
+	if (((unsigned long)(-1) / 100) < cur.aperf.split.lo) {
 		int shift_count = 7;
-		cur.aperf_cur.split.lo >>= shift_count;
-		cur.mperf_cur.split.lo >>= shift_count;
+		cur.aperf.split.lo >>= shift_count;
+		cur.mperf.split.lo >>= shift_count;
 	}
 
-	if (cur.aperf_cur.split.lo && cur.mperf_cur.split.lo)
-		perf_percent = (cur.aperf_cur.split.lo * 100) /
-				cur.mperf_cur.split.lo;
+	if (cur.aperf.split.lo && cur.mperf.split.lo)
+		perf_percent = (cur.aperf.split.lo * 100) / cur.mperf.split.lo;
 	else
 		perf_percent = 0;
 
 #else
-	if (unlikely(((unsigned long)(-1) / 100) < cur.aperf_cur.whole)) {
+	if (unlikely(((unsigned long)(-1) / 100) < cur.aperf.whole)) {
 		int shift_count = 7;
-		cur.aperf_cur.whole >>= shift_count;
-		cur.mperf_cur.whole >>= shift_count;
+		cur.aperf.whole >>= shift_count;
+		cur.mperf.whole >>= shift_count;
 	}
 
-	if (cur.aperf_cur.whole && cur.mperf_cur.whole)
-		perf_percent = (cur.aperf_cur.whole * 100) /
-				cur.mperf_cur.whole;
+	if (cur.aperf.whole && cur.mperf.whole)
+		perf_percent = (cur.aperf.whole * 100) / cur.mperf.whole;
 	else
 		perf_percent = 0;
 
 #endif
 
-	retval = per_cpu(drv_data, policy->cpu)->max_freq * perf_percent / 100;
+	retval = (policy->cpuinfo.max_freq * perf_percent) / 100;
 
 	return retval;
 }
@@ -685,16 +692,11 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy)
 	/* Check for high latency (>20uS) from buggy BIOSes, like on T42 */
 	if (perf->control_register.space_id == ACPI_ADR_SPACE_FIXED_HARDWARE &&
 	    policy->cpuinfo.transition_latency > 20 * 1000) {
-		static int print_once;
 		policy->cpuinfo.transition_latency = 20 * 1000;
-		if (!print_once) {
-			print_once = 1;
-			printk(KERN_INFO "Capping off P-state tranision latency"
-				" at 20 uS\n");
-		}
+			printk_once(KERN_INFO "Capping off P-state tranision"
+				    " latency at 20 uS\n");
 	}
 
-	data->max_freq = perf->states[0].core_frequency * 1000;
 	/* table init */
 	for (i = 0; i < perf->state_count; i++) {
 		if (i > 0 && perf->states[i].core_frequency >=
@@ -713,6 +715,9 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy)
 	if (result)
 		goto err_freqfree;
 
+	if (perf->states[0].core_frequency * 1000 != policy->cpuinfo.max_freq)
+		printk(KERN_WARNING FW_WARN "P-state 0 is not max freq\n");
+
 	switch (perf->control_register.space_id) {
 	case ACPI_ADR_SPACE_SYSTEM_IO:
 		/* Current speed is unknown and not detectable by IO port */
diff --git a/arch/x86/kernel/cpu/cpufreq/longhaul.c b/arch/x86/kernel/cpu/cpufreq/longhaul.c
index 0bd48e6..ce2ed3e 100644
--- a/arch/x86/kernel/cpu/cpufreq/longhaul.c
+++ b/arch/x86/kernel/cpu/cpufreq/longhaul.c
@@ -33,7 +33,6 @@
 #include <linux/timex.h>
 #include <linux/io.h>
 #include <linux/acpi.h>
-#include <linux/kernel.h>
 
 #include <asm/msr.h>
 #include <acpi/processor.h>