[PATCH 0/4] perf/x86/power: Introduce AMD accumlated power reporting mechanism

[PATCH 0/4] perf/x86/power: Introduce AMD accumlated power reporting mechanism

[Huang Rui]

Hi,

This series of patches introduces the perf implementation of
accumulated power reporting algorithm. It will calculate the average
power consumption for the processor. The CPU feature flag is
CPUID.8000_0007H:EDX[12].

Simple example:

root@hr-zp:/home/ray/tip# ./tools/perf/perf stat -a -e 'power/power-pkg/' make -j4
  CHK     include/config/kernel.release
  CHK     include/generated/uapi/linux/version.h
  CHK     include/generated/utsrelease.h
  CHK     include/generated/timeconst.h
  CHK     include/generated/bounds.h
  CHK     include/generated/asm-offsets.h
  CALL    scripts/checksyscalls.sh
  CHK     include/generated/compile.h
  SKIPPED include/generated/compile.h
  Building modules, stage 2.
Kernel: arch/x86/boot/bzImage is ready  (#40)
  MODPOST 4225 modules

 Performance counter stats for 'system wide':

            183.44 mWatts power/power-pkg/

     341.837270111 seconds time elapsed

root@hr-zp:/home/ray/tip# ./tools/perf/perf stat -a -e 'power/power-pkg/' sleep 10

 Performance counter stats for 'system wide':

              0.18 mWatts power/power-pkg/

      10.012551815 seconds time elapsed

Reference:
http://lkml.kernel.org/r/20150831160622.GA29830@nazgul.tnic

Thanks,
Rui

Huang Rui (4):
  x86/amd: move nodes_per_socket into bsp_init_amd
  x86/amd: add accessor for number of cores per compute unit
  x86/cpufeature: add AMD Accumulated Power Mechanism feature flag
  perf/x86/amd/power: Add AMD accumulated power reporting mechanism

 arch/x86/include/asm/cpufeature.h          |   2 +-
 arch/x86/include/asm/processor.h           |   1 +
 arch/x86/kernel/cpu/Makefile               |   1 +
 arch/x86/kernel/cpu/amd.c                  |  31 +-
 arch/x86/kernel/cpu/perf_event_amd_power.c | 531 +++++++++++++++++++++++++++++
 5 files changed, 561 insertions(+), 5 deletions(-)
 create mode 100644 arch/x86/kernel/cpu/perf_event_amd_power.c

-- 
1.9.1

[PATCH 1/4] x86/amd: move nodes_per_socket into bsp_init_amd

[Huang Rui]

nodes_per_socket is static and it needn't be initialized many times
during every cpu core init. So move nodes_per_socket initialization
into bsp_init_amd.

Signed-off-by: Huang Rui <ray.huang@amd.com>
---
 arch/x86/kernel/cpu/amd.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index e229640..97db1b6 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -308,7 +308,6 @@ static void amd_get_topology(struct cpuinfo_x86 *c)
 		u32 eax, ebx, ecx, edx;
 
 		cpuid(0x8000001e, &eax, &ebx, &ecx, &edx);
-		nodes_per_socket = ((ecx >> 8) & 7) + 1;
 		node_id = ecx & 7;
 
 		/* get compute unit information */
@@ -319,7 +318,6 @@ static void amd_get_topology(struct cpuinfo_x86 *c)
 		u64 value;
 
 		rdmsrl(MSR_FAM10H_NODE_ID, value);
-		nodes_per_socket = ((value >> 3) & 7) + 1;
 		node_id = value & 7;
 	} else
 		return;
@@ -523,6 +521,18 @@ static void bsp_init_amd(struct cpuinfo_x86 *c)
 
 	if (cpu_has(c, X86_FEATURE_MWAITX))
 		use_mwaitx_delay();
+
+	if (boot_cpu_has(X86_FEATURE_TOPOEXT)) {
+		u32 ecx;
+
+		ecx = cpuid_ecx(0x8000001e);
+		nodes_per_socket = ((ecx >> 8) & 7) + 1;
+	} else if (boot_cpu_has(X86_FEATURE_NODEID_MSR)) {
+		u64 value;
+
+		rdmsrl(MSR_FAM10H_NODE_ID, value);
+		nodes_per_socket = ((value >> 3) & 7) + 1;
+	}
 }
 
 static void early_init_amd(struct cpuinfo_x86 *c)
-- 
1.9.1

[PATCH 2/4] x86/amd: add accessor for number of cores per compute unit

[Huang Rui]

Add an accessor function amd_get_cores_per_cu() which returns the
number of cores per compute unit.

In a subsequent patch, we will use this function in both x86 perf and
fam15h driver.

Signed-off-by: Huang Rui <ray.huang@amd.com>
---
 arch/x86/include/asm/processor.h |  1 +
 arch/x86/kernel/cpu/amd.c        | 17 +++++++++++++----
 2 files changed, 14 insertions(+), 4 deletions(-)

diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index 6752225..9bf48b2 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -810,6 +810,7 @@ static inline int mpx_disable_management(void)
 
 extern u16 amd_get_nb_id(int cpu);
 extern u32 amd_get_nodes_per_socket(void);
+extern u32 amd_get_cores_per_cu(void);
 
 static inline uint32_t hypervisor_cpuid_base(const char *sig, uint32_t leaves)
 {
diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index 97db1b6..d6e320f 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -27,6 +27,9 @@
  */
 static u32 nodes_per_socket = 1;
 
+/* cores_per_cu: stores the number of cores per compute unit */
+static u32 cores_per_cu = 1;
+
 static inline int rdmsrl_amd_safe(unsigned msr, unsigned long long *p)
 {
 	u32 gprs[8] = { 0 };
@@ -299,7 +302,6 @@ static int nearby_node(int apicid)
 #ifdef CONFIG_SMP
 static void amd_get_topology(struct cpuinfo_x86 *c)
 {
-	u32 cores_per_cu = 1;
 	u8 node_id;
 	int cpu = smp_processor_id();
 
@@ -313,7 +315,6 @@ static void amd_get_topology(struct cpuinfo_x86 *c)
 		/* get compute unit information */
 		smp_num_siblings = ((ebx >> 8) & 3) + 1;
 		c->compute_unit_id = ebx & 0xff;
-		cores_per_cu += ((ebx >> 8) & 3);
 	} else if (cpu_has(c, X86_FEATURE_NODEID_MSR)) {
 		u64 value;
 
@@ -391,6 +392,13 @@ u32 amd_get_nodes_per_socket(void)
 }
 EXPORT_SYMBOL_GPL(amd_get_nodes_per_socket);
 
+/* this function returns the number of cores per compute unit */
+u32 amd_get_cores_per_cu(void)
+{
+	return cores_per_cu;
+}
+EXPORT_SYMBOL_GPL(amd_get_cores_per_cu);
+
 static void srat_detect_node(struct cpuinfo_x86 *c)
 {
 #ifdef CONFIG_NUMA
@@ -523,10 +531,11 @@ static void bsp_init_amd(struct cpuinfo_x86 *c)
 		use_mwaitx_delay();
 
 	if (boot_cpu_has(X86_FEATURE_TOPOEXT)) {
-		u32 ecx;
+		u32 eax, ebx, ecx, edx;
 
-		ecx = cpuid_ecx(0x8000001e);
+		cpuid(0x8000001e, &eax, &ebx, &ecx, &edx);
 		nodes_per_socket = ((ecx >> 8) & 7) + 1;
+		cores_per_cu = ((ebx >> 8) & 3) + 1;
 	} else if (boot_cpu_has(X86_FEATURE_NODEID_MSR)) {
 		u64 value;
 
-- 
1.9.1

[PATCH 3/4] x86/cpufeature: add AMD Accumulated Power Mechanism feature flag

[Huang Rui]

AMD CPU family 15h model 0x60 introduces accumulated power mechanism.
It is used to report the processor power consumption and indicated by
CPUID Fn8000_0007_EDX[12].

Signed-off-by: Huang Rui <ray.huang@amd.com>
---
 arch/x86/include/asm/cpufeature.h | 2 +-
 arch/x86/kernel/cpu/amd.c         | 4 ++++
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/cpufeature.h b/arch/x86/include/asm/cpufeature.h
index f7ba9fb..0adce8f 100644
--- a/arch/x86/include/asm/cpufeature.h
+++ b/arch/x86/include/asm/cpufeature.h
@@ -85,7 +85,7 @@
 #define X86_FEATURE_P4		( 3*32+ 7) /* "" P4 */
 #define X86_FEATURE_CONSTANT_TSC ( 3*32+ 8) /* TSC ticks at a constant rate */
 #define X86_FEATURE_UP		( 3*32+ 9) /* smp kernel running on up */
-/* free, was #define X86_FEATURE_FXSAVE_LEAK ( 3*32+10) * "" FXSAVE leaks FOP/FIP/FOP */
+#define X86_FEATURE_ACC_POWER	( 3*32+10) /* AMD Accumulated Power Mechanism */
 #define X86_FEATURE_ARCH_PERFMON ( 3*32+11) /* Intel Architectural PerfMon */
 #define X86_FEATURE_PEBS	( 3*32+12) /* Precise-Event Based Sampling */
 #define X86_FEATURE_BTS		( 3*32+13) /* Branch Trace Store */
diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index d6e320f..97d4ce2 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -559,6 +559,10 @@ static void early_init_amd(struct cpuinfo_x86 *c)
 			set_sched_clock_stable();
 	}
 
+	/* Bit 12 of 8000_0007 edx is accumulated power mechanism. */
+	if (c->x86_power & BIT(12))
+		set_cpu_cap(c, X86_FEATURE_ACC_POWER);
+
 #ifdef CONFIG_X86_64
 	set_cpu_cap(c, X86_FEATURE_SYSCALL32);
 #else
-- 
1.9.1

[PATCH 4/4] perf/x86/amd/power: Add AMD accumulated power reporting mechanism

[Huang Rui]

Introduce an AMD accumlated power reporting mechanism for Carrizo
(Family 15h, Model 60h) processor that should be used to calculate the
average power consumed by a processor during a measurement interval.
The feature of accumulated power mechanism is indicated by CPUID
Fn8000_0007_EDX[12].

---------------------------------------------------------------------
* Tsample: compute unit power accumulator sample period
* Tref: the PTSC counter period
* PTSC: performance timestamp counter
* N: the ratio of compute unit power accumulator sample period to the
  PTSC period
* Jmax: max compute unit accumulated power which is indicated by
  MaxCpuSwPwrAcc MSR C001007b
* Jx/Jy: compute unit accumulated power which is indicated by
  CpuSwPwrAcc MSR C001007a
* Tx/Ty: the value of performance timestamp counter which is indicated
  by CU_PTSC MSR C0010280
* PwrCPUave: CPU average power

i. Determine the ratio of Tsample to Tref by executing CPUID Fn8000_0007.
	N = value of CPUID Fn8000_0007_ECX[CpuPwrSampleTimeRatio[15:0]].

ii. Read the full range of the cumulative energy value from the new
MSR MaxCpuSwPwrAcc.
	Jmax = value returned.
iii. At time x, SW reads CpuSwPwrAcc MSR and samples the PTSC.
	Jx = value read from CpuSwPwrAcc and Tx = value read from
PTSC.

iv. At time y, SW reads CpuSwPwrAcc MSR and samples the PTSC.
	Jy = value read from CpuSwPwrAcc and Ty = value read from
PTSC.

v. Calculate the average power consumption for a compute unit over
time period (y-x). Unit of result is uWatt.
	if (Jy < Jx) // Rollover has occurred
		Jdelta = (Jy + Jmax) - Jx
	else
		Jdelta = Jy - Jx
	PwrCPUave = N * Jdelta * 1000 / (Ty - Tx)
----------------------------------------------------------------------

This feature will be implemented both on hwmon and perf that discussed
in mail list before. At current design, it provides one event to
report per package/processor power consumption by counting each
compute unit power value.

Simple example:

root@hr-zp:/home/ray/tip# ./tools/perf/perf stat -a -e 'power/power-pkg/' make -j4
  CHK     include/config/kernel.release
  CHK     include/generated/uapi/linux/version.h
  CHK     include/generated/utsrelease.h
  CHK     include/generated/timeconst.h
  CHK     include/generated/bounds.h
  CHK     include/generated/asm-offsets.h
  CALL    scripts/checksyscalls.sh
  CHK     include/generated/compile.h
  SKIPPED include/generated/compile.h
  Building modules, stage 2.
Kernel: arch/x86/boot/bzImage is ready  (#40)
  MODPOST 4225 modules

 Performance counter stats for 'system wide':

            183.44 mWatts power/power-pkg/

     341.837270111 seconds time elapsed

root@hr-zp:/home/ray/tip# ./tools/perf/perf stat -a -e 'power/power-pkg/' sleep 10

 Performance counter stats for 'system wide':

              0.18 mWatts power/power-pkg/

      10.012551815 seconds time elapsed

Reference:
http://lkml.kernel.org/r/20150831160622.GA29830@nazgul.tnic

Suggested-by: Peter Zijlstra <peterz@infradead.org>
Suggested-by: Ingo Molnar <mingo@kernel.org>
Suggested-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Huang Rui <ray.huang@amd.com>
Cc: Guenter Roeck <linux@roeck-us.net>
---
 arch/x86/kernel/cpu/Makefile               |   1 +
 arch/x86/kernel/cpu/perf_event_amd_power.c | 531 +++++++++++++++++++++++++++++
 2 files changed, 532 insertions(+)
 create mode 100644 arch/x86/kernel/cpu/perf_event_amd_power.c

diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile
index 5803130..97f3413 100644
--- a/arch/x86/kernel/cpu/Makefile
+++ b/arch/x86/kernel/cpu/Makefile
@@ -34,6 +34,7 @@ obj-$(CONFIG_PERF_EVENTS)		+= perf_event.o
 
 ifdef CONFIG_PERF_EVENTS
 obj-$(CONFIG_CPU_SUP_AMD)		+= perf_event_amd.o perf_event_amd_uncore.o
+obj-$(CONFIG_CPU_SUP_AMD)		+= perf_event_amd_power.o
 ifdef CONFIG_AMD_IOMMU
 obj-$(CONFIG_CPU_SUP_AMD)		+= perf_event_amd_iommu.o
 endif
diff --git a/arch/x86/kernel/cpu/perf_event_amd_power.c b/arch/x86/kernel/cpu/perf_event_amd_power.c
new file mode 100644
index 0000000..6932755
--- /dev/null
+++ b/arch/x86/kernel/cpu/perf_event_amd_power.c
@@ -0,0 +1,531 @@
+/*
+ * Performance events - AMD Processor Power Reporting Mechanism
+ *
+ * Copyright (C) 2016 Advanced Micro Devices, Inc.
+ *
+ * Author: Huang Rui <ray.huang@amd.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/perf_event.h>
+#include <asm/cpu_device_id.h>
+#include "perf_event.h"
+
+#define MSR_F15H_CU_PWR_ACCUMULATOR     0xc001007a
+#define MSR_F15H_CU_MAX_PWR_ACCUMULATOR 0xc001007b
+#define MSR_F15H_PTSC			0xc0010280
+
+/*
+ * event code: LSB 8 bits, passed in attr->config
+ * any other bit is reserved
+ */
+#define AMD_POWER_EVENT_MASK	0xFFULL
+
+#define MAX_CUS	8
+
+/*
+ * Acc power status counters
+ */
+#define AMD_POWER_PKG_ID		0
+#define AMD_POWER_EVENTSEL_PKG		1
+
+/*
+ * the ratio of compute unit power accumulator sample period to the
+ * PTSC period
+ */
+static unsigned int cpu_pwr_sample_ratio;
+static unsigned int cores_per_cu;
+static unsigned int cu_num;
+
+/* maximum accumulated power of a compute unit */
+static u64 max_cu_acc_power;
+
+struct power_pmu {
+	spinlock_t		lock;
+	struct list_head	active_list;
+	struct pmu		*pmu; /* pointer to power_pmu_class */
+	local64_t		cpu_sw_pwr_ptsc;
+};
+
+static struct pmu pmu_class;
+
+/*
+ * Accumulated power is to measure the sum of each compute unit's
+ * power consumption. So it picks only one core from each compute unit
+ * to get the power with MSR_F15H_CU_PWR_ACCUMULATOR. The cpu_mask
+ * represents CPU bit map of all cores which are picked to measure the
+ * power for the compute units that they belong to.
+ */
+static cpumask_t cpu_mask;
+
+static DEFINE_PER_CPU(struct power_pmu *, amd_power_pmu);
+
+static u64 event_update(struct perf_event *event, struct power_pmu *pmu)
+{
+	struct hw_perf_event *hwc = &event->hw;
+	u64 prev_raw_count, new_raw_count, prev_ptsc, new_ptsc;
+	u64 delta, tdelta;
+
+again:
+	prev_raw_count = local64_read(&hwc->prev_count);
+	prev_ptsc = local64_read(&pmu->cpu_sw_pwr_ptsc);
+	rdmsrl(event->hw.event_base, new_raw_count);
+	rdmsrl(MSR_F15H_PTSC, new_ptsc);
+
+	if (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
+			    new_raw_count) != prev_raw_count) {
+		cpu_relax();
+		goto again;
+	}
+
+	/*
+	 * calculate the power comsumption for each compute unit over
+	 * a time period, the unit of final value (delta) is
+	 * micro-Watts. Then add it into event count.
+	 */
+	if (new_raw_count < prev_raw_count) {
+		delta = max_cu_acc_power + new_raw_count;
+		delta -= prev_raw_count;
+	} else
+		delta = new_raw_count - prev_raw_count;
+
+	delta *= cpu_pwr_sample_ratio * 1000;
+	tdelta = new_ptsc - prev_ptsc;
+
+	do_div(delta, tdelta);
+	local64_add(delta, &event->count);
+
+	return new_raw_count;
+}
+
+static void __pmu_event_start(struct power_pmu *pmu,
+			      struct perf_event *event)
+{
+	u64 ptsc, counts;
+
+	if (WARN_ON_ONCE(!(event->hw.state & PERF_HES_STOPPED)))
+		return;
+
+	event->hw.state = 0;
+
+	list_add_tail(&event->active_entry, &pmu->active_list);
+
+	rdmsrl(MSR_F15H_PTSC, ptsc);
+	local64_set(&pmu->cpu_sw_pwr_ptsc, ptsc);
+	rdmsrl(event->hw.event_base, counts);
+	local64_set(&event->hw.prev_count, counts);
+}
+
+static void pmu_event_start(struct perf_event *event, int mode)
+{
+	struct power_pmu *pmu = __this_cpu_read(amd_power_pmu);
+	unsigned long flags;
+
+	spin_lock_irqsave(&pmu->lock, flags);
+	__pmu_event_start(pmu, event);
+	spin_unlock_irqrestore(&pmu->lock, flags);
+}
+
+static void pmu_event_stop(struct perf_event *event, int mode)
+{
+	struct power_pmu *pmu = __this_cpu_read(amd_power_pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	unsigned long flags;
+
+	spin_lock_irqsave(&pmu->lock, flags);
+
+	/* mark event as deactivated and stopped */
+	if (!(hwc->state & PERF_HES_STOPPED)) {
+		list_del(&event->active_entry);
+		hwc->state |= PERF_HES_STOPPED;
+	}
+
+	/* check if update of sw counter is necessary */
+	if ((mode & PERF_EF_UPDATE) && !(hwc->state & PERF_HES_UPTODATE)) {
+		/*
+		 * Drain the remaining delta count out of a event
+		 * that we are disabling:
+		 */
+		event_update(event, pmu);
+		hwc->state |= PERF_HES_UPTODATE;
+	}
+
+	spin_unlock_irqrestore(&pmu->lock, flags);
+}
+
+static int pmu_event_add(struct perf_event *event, int mode)
+{
+	struct power_pmu *pmu = __this_cpu_read(amd_power_pmu);
+	struct hw_perf_event *hwc = &event->hw;
+	unsigned long flags;
+
+	spin_lock_irqsave(&pmu->lock, flags);
+
+	hwc->state = PERF_HES_UPTODATE | PERF_HES_STOPPED;
+
+	if (mode & PERF_EF_START)
+		__pmu_event_start(pmu, event);
+
+	spin_unlock_irqrestore(&pmu->lock, flags);
+
+	return 0;
+}
+
+static void pmu_event_del(struct perf_event *event, int flags)
+{
+	pmu_event_stop(event, PERF_EF_UPDATE);
+}
+
+static int pmu_event_init(struct perf_event *event)
+{
+	u64 cfg = event->attr.config & AMD_POWER_EVENT_MASK;
+	int bit, ret = 0;
+
+	/* only look at AMD power events */
+	if (event->attr.type != pmu_class.type)
+		return -ENOENT;
+
+	/* unsupported modes and filters */
+	if (event->attr.exclude_user   ||
+	    event->attr.exclude_kernel ||
+	    event->attr.exclude_hv     ||
+	    event->attr.exclude_idle   ||
+	    event->attr.exclude_host   ||
+	    event->attr.exclude_guest  ||
+	    event->attr.sample_period) /* no sampling */
+		return -EINVAL;
+
+	if (cfg == AMD_POWER_EVENTSEL_PKG)
+		bit = AMD_POWER_PKG_ID;
+	else
+		return -EINVAL;
+
+	event->hw.event_base = MSR_F15H_CU_PWR_ACCUMULATOR;
+	event->hw.config = cfg;
+	event->hw.idx = bit;
+
+	return ret;
+}
+
+static void pmu_event_read(struct perf_event *event)
+{
+	struct power_pmu *pmu = __this_cpu_read(amd_power_pmu);
+
+	event_update(event, pmu);
+}
+
+static ssize_t get_attr_cpumask(struct device *dev,
+				      struct device_attribute *attr, char *buf)
+{
+	return cpumap_print_to_pagebuf(true, buf, &cpu_mask);
+}
+
+static DEVICE_ATTR(cpumask, S_IRUGO, get_attr_cpumask, NULL);
+
+static struct attribute *pmu_attrs[] = {
+	&dev_attr_cpumask.attr,
+	NULL,
+};
+
+static struct attribute_group pmu_attr_group = {
+	.attrs = pmu_attrs,
+};
+
+
+/*
+ * at current, it only supports to report the power of each
+ * processor/package
+ */
+EVENT_ATTR_STR(power-pkg, power_pkg, "event=0x01");
+
+EVENT_ATTR_STR(power-pkg.unit, power_pkg_unit, "mWatts");
+
+/* convert the count from micro-Watts to milli-Watts */
+EVENT_ATTR_STR(power-pkg.scale, power_pkg_scale, "1.000000e-3");
+
+
+static struct attribute *events_attr[] = {
+	EVENT_PTR(power_pkg),
+	EVENT_PTR(power_pkg_unit),
+	EVENT_PTR(power_pkg_scale),
+	NULL,
+};
+
+static struct attribute_group pmu_events_group = {
+	.name = "events",
+	.attrs = events_attr,
+};
+
+PMU_FORMAT_ATTR(event, "config:0-7");
+
+static struct attribute *formats_attr[] = {
+	&format_attr_event.attr,
+	NULL,
+};
+
+static struct attribute_group pmu_format_group = {
+	.name = "format",
+	.attrs = formats_attr,
+};
+
+static const struct attribute_group *attr_groups[] = {
+	&pmu_attr_group,
+	&pmu_format_group,
+	&pmu_events_group,
+	NULL,
+};
+
+static struct pmu pmu_class = {
+	.attr_groups	= attr_groups,
+	.task_ctx_nr	= perf_invalid_context, /* system-wide only */
+	.event_init	= pmu_event_init,
+	.add		= pmu_event_add, /* must have */
+	.del		= pmu_event_del, /* must have */
+	.start		= pmu_event_start,
+	.stop		= pmu_event_stop,
+	.read		= pmu_event_read,
+};
+
+
+static int power_cpu_exit(int cpu)
+{
+	struct power_pmu *pmu = per_cpu(amd_power_pmu, cpu);
+	int i, cu, ret = 0;
+	int target = nr_cpumask_bits;
+	cpumask_var_t mask, tmp_mask;
+
+	cu = cpu / cores_per_cu;
+
+	if (!zalloc_cpumask_var(&mask, GFP_KERNEL))
+		return -ENOMEM;
+
+	if (!zalloc_cpumask_var(&tmp_mask, GFP_KERNEL)) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	for (i = 0; i < cores_per_cu; i++)
+		cpumask_set_cpu(i, mask);
+
+	cpumask_shift_left(mask, mask, cu * cores_per_cu);
+
+	cpumask_clear_cpu(cpu, &cpu_mask);
+	cpumask_clear_cpu(cpu, mask);
+
+	if (!cpumask_and(tmp_mask, mask, cpu_online_mask))
+		goto out1;
+
+	/*
+	 * find a new CPU on same compute unit, if was set in cpumask
+	 * and still some CPUs on compute unit, then move to the new
+	 * CPU
+	 */
+	target = cpumask_any(tmp_mask);
+	if (target < nr_cpumask_bits && target != cpu)
+		cpumask_set_cpu(target, &cpu_mask);
+
+	WARN_ON(cpumask_empty(&cpu_mask));
+
+out1:
+	/*
+	 * migrate events and context to new CPU
+	 */
+	if (target < nr_cpumask_bits)
+		perf_pmu_migrate_context(pmu->pmu, cpu, target);
+
+	free_cpumask_var(tmp_mask);
+out:
+	free_cpumask_var(mask);
+
+	return ret;
+
+}
+
+static int power_cpu_init(int cpu)
+{
+	int i, cu, ret = 0;
+	cpumask_var_t mask, dummy_mask;
+
+	cu = cpu / cores_per_cu;
+
+	if (!zalloc_cpumask_var(&mask, GFP_KERNEL))
+		return -ENOMEM;
+
+	if (!zalloc_cpumask_var(&dummy_mask, GFP_KERNEL)) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	for (i = 0; i < cores_per_cu; i++)
+		cpumask_set_cpu(i, mask);
+
+	cpumask_shift_left(mask, mask, cu * cores_per_cu);
+
+	if (!cpumask_and(dummy_mask, mask, &cpu_mask))
+		cpumask_set_cpu(cpu, &cpu_mask);
+
+	free_cpumask_var(dummy_mask);
+out:
+	free_cpumask_var(mask);
+
+	return ret;
+}
+
+static int power_cpu_prepare(int cpu)
+{
+	struct power_pmu *pmu = per_cpu(amd_power_pmu, cpu);
+	int phys_id = topology_physical_package_id(cpu);
+
+	if (pmu)
+		return 0;
+
+	if (phys_id < 0)
+		return -EINVAL;
+
+	pmu = kzalloc_node(sizeof(*pmu), GFP_KERNEL, cpu_to_node(cpu));
+	if (!pmu)
+		return -ENOMEM;
+
+	spin_lock_init(&pmu->lock);
+
+	INIT_LIST_HEAD(&pmu->active_list);
+
+	pmu->pmu = &pmu_class;
+
+	per_cpu(amd_power_pmu, cpu) = pmu;
+
+	return 0;
+}
+
+static void power_cpu_kfree(int cpu)
+{
+	struct power_pmu *pmu = per_cpu(amd_power_pmu, cpu);
+
+	kfree(pmu);
+}
+
+static int power_cpu_notifier(struct notifier_block *self,
+			      unsigned long action, void *hcpu)
+{
+	unsigned int cpu = (long)hcpu;
+
+	switch (action & ~CPU_TASKS_FROZEN) {
+	case CPU_UP_PREPARE:
+		if (power_cpu_prepare(cpu))
+			return NOTIFY_BAD;
+		break;
+	case CPU_STARTING:
+		if (power_cpu_init(cpu))
+			return NOTIFY_BAD;
+		break;
+	case CPU_ONLINE:
+	case CPU_DEAD:
+		power_cpu_kfree(cpu);
+		break;
+	case CPU_DOWN_PREPARE:
+		if (power_cpu_exit(cpu))
+			return NOTIFY_BAD;
+		break;
+	default:
+		break;
+	}
+
+	return NOTIFY_OK;
+}
+
+static const struct x86_cpu_id cpu_match[] = {
+	{ .vendor = X86_VENDOR_AMD, .family = 0x15 },
+	{},
+};
+
+static int __init amd_power_pmu_init(void)
+{
+	int i, ret;
+	u64 tmp;
+	cpumask_var_t tmp_mask, res_mask;
+
+	if (!x86_match_cpu(cpu_match))
+		return 0;
+
+	if (!boot_cpu_has(X86_FEATURE_ACC_POWER))
+		return -ENODEV;
+
+	cores_per_cu = amd_get_cores_per_cu();
+	cu_num = boot_cpu_data.x86_max_cores / cores_per_cu;
+
+	if (WARN_ON_ONCE(cu_num > MAX_CUS))
+		return -EINVAL;
+
+	cpu_pwr_sample_ratio = cpuid_ecx(0x80000007);
+
+	if (rdmsrl_safe(MSR_F15H_CU_MAX_PWR_ACCUMULATOR, &tmp)) {
+		pr_err("Failed to read max compute unit power accumulator MSR\n");
+		return -ENODEV;
+	}
+	max_cu_acc_power = tmp;
+
+	if (!zalloc_cpumask_var(&tmp_mask, GFP_KERNEL))
+		return -ENOMEM;
+
+	if (!zalloc_cpumask_var(&res_mask, GFP_KERNEL)) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	for (i = 0; i < cores_per_cu; i++)
+		cpumask_set_cpu(i, tmp_mask);
+
+	cpu_notifier_register_begin();
+
+	/*
+	 * Choose the one online core of each compute unit
+	 */
+	for (i = 0; i < cu_num; i++) {
+		/* WARN_ON for empty CU masks */
+		WARN_ON(!cpumask_and(res_mask, tmp_mask, cpu_online_mask));
+		cpumask_set_cpu(cpumask_any(res_mask), &cpu_mask);
+		cpumask_shift_left(tmp_mask, tmp_mask, cores_per_cu);
+	}
+
+	for_each_present_cpu(i) {
+		ret = power_cpu_prepare(i);
+		if (ret) {
+			/* unwind on [0 ... i-1] CPUs */
+			while (i--)
+				power_cpu_kfree(i);
+			goto out1;
+		}
+		ret = power_cpu_init(i);
+		if (ret) {
+			/* unwind on [0 ... i] CPUs */
+			while (i >= 0)
+				power_cpu_kfree(i--);
+			goto out1;
+		}
+	}
+
+	__perf_cpu_notifier(power_cpu_notifier);
+
+	ret = perf_pmu_register(&pmu_class, "power", -1);
+	if (WARN_ON(ret)) {
+		pr_warn("AMD Power PMU registration failed\n");
+		goto out1;
+	}
+
+	pr_info("AMD Power PMU detected, %d compute units\n", cu_num);
+
+out1:
+	cpu_notifier_register_done();
+
+	free_cpumask_var(res_mask);
+out:
+	free_cpumask_var(tmp_mask);
+
+	return ret;
+}
+device_initcall(amd_power_pmu_init);
-- 
1.9.1

Re: [PATCH 4/4] perf/x86/amd/power: Add AMD accumulated power reporting mechanism

[kbuild test robot]

[-- Attachment #1: Type: text/plain, Size: 2492 bytes --]

Hi Huang,

[auto build test ERROR on bp/for-next]
[also build test ERROR on v4.4-rc8 next-20160106]
[cannot apply to tip/x86/core]
[if your patch is applied to the wrong git tree, please drop us a note to help improving the system]

url:    https://github.com/0day-ci/linux/commits/Huang-Rui/perf-x86-power-Introduce-AMD-accumlated-power-reporting-mechanism/20160107-103309
base:   https://git.kernel.org/pub/scm/linux/kernel/git/bp/bp for-next
config: i386-randconfig-s1-201601 (attached as .config)
reproduce:
        # save the attached .config to linux build tree
        make ARCH=i386 

All error/warnings (new ones prefixed by >>):

   In file included from include/linux/kobject.h:21:0,
                    from include/linux/module.h:17,
                    from arch/x86/kernel/cpu/perf_event_amd_power.c:13:
>> arch/x86/kernel/cpu/perf_event.h:663:31: error: 'events_sysfs_show' undeclared here (not in a function)
     .attr  = __ATTR(_name, 0444, events_sysfs_show, NULL), \
                                  ^
   include/linux/sysfs.h:103:10: note: in definition of macro '__ATTR'
     .show = _show,      \
             ^
>> arch/x86/kernel/cpu/perf_event_amd_power.c:244:1: note: in expansion of macro 'EVENT_ATTR_STR'
    EVENT_ATTR_STR(power-pkg, power_pkg, "event=0x01");
    ^

vim +/events_sysfs_show +663 arch/x86/kernel/cpu/perf_event.h

3a54aaa0 Stephane Eranian 2013-01-24  657  	.id		= PERF_COUNT_HW_##_id,				\
3a54aaa0 Stephane Eranian 2013-01-24  658  	.event_str	= NULL,						\
3a54aaa0 Stephane Eranian 2013-01-24  659  };
3a54aaa0 Stephane Eranian 2013-01-24  660  
3a54aaa0 Stephane Eranian 2013-01-24  661  #define EVENT_ATTR_STR(_name, v, str)					\
3a54aaa0 Stephane Eranian 2013-01-24  662  static struct perf_pmu_events_attr event_attr_##v = {			\
3a54aaa0 Stephane Eranian 2013-01-24 @663  	.attr		= __ATTR(_name, 0444, events_sysfs_show, NULL),	\
3a54aaa0 Stephane Eranian 2013-01-24  664  	.id		= 0,						\
3a54aaa0 Stephane Eranian 2013-01-24  665  	.event_str	= str,						\
3a54aaa0 Stephane Eranian 2013-01-24  666  };

:::::: The code at line 663 was first introduced by commit
:::::: 3a54aaa0a3ddb2cf2ec1b94a94024e9a8a8af962 perf/x86: Improve sysfs event mapping with event string

:::::: TO: Stephane Eranian <eranian@google.com>
:::::: CC: Arnaldo Carvalho de Melo <acme@redhat.com>

---
0-DAY kernel test infrastructure                Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all                   Intel Corporation

[-- Attachment #2: .config.gz --]
[-- Type: application/octet-stream, Size: 22266 bytes --]

Re: [PATCH 4/4] perf/x86/amd/power: Add AMD accumulated power reporting mechanism

[Huang Rui]

On Thu, Jan 07, 2016 at 11:14:50AM +0800, kbuild test robot wrote:
> Hi Huang,
> 
> [auto build test ERROR on bp/for-next]
> [also build test ERROR on v4.4-rc8 next-20160106]
> [cannot apply to tip/x86/core]
> [if your patch is applied to the wrong git tree, please drop us a note to help improving the system]
> 
> url:    https://github.com/0day-ci/linux/commits/Huang-Rui/perf-x86-power-Introduce-AMD-accumlated-power-reporting-mechanism/20160107-103309
> base:   https://git.kernel.org/pub/scm/linux/kernel/git/bp/bp for-next
> config: i386-randconfig-s1-201601 (attached as .config)
> reproduce:
>         # save the attached .config to linux build tree
>         make ARCH=i386 
> 
> All error/warnings (new ones prefixed by >>):
> 
>    In file included from include/linux/kobject.h:21:0,
>                     from include/linux/module.h:17,
>                     from arch/x86/kernel/cpu/perf_event_amd_power.c:13:
> >> arch/x86/kernel/cpu/perf_event.h:663:31: error: 'events_sysfs_show' undeclared here (not in a function)
>      .attr  = __ATTR(_name, 0444, events_sysfs_show, NULL), \
>                                   ^
>    include/linux/sysfs.h:103:10: note: in definition of macro '__ATTR'
>      .show = _show,      \
>              ^
> >> arch/x86/kernel/cpu/perf_event_amd_power.c:244:1: note: in expansion of macro 'EVENT_ATTR_STR'
>     EVENT_ATTR_STR(power-pkg, power_pkg, "event=0x01");
>     ^
> 
> vim +/events_sysfs_show +663 arch/x86/kernel/cpu/perf_event.h
> 

Thanks to report this issue.

That's because events_sysfs_show is declared under
CONFIG_CPU_SUP_INTEL, I will add a patch to move it out.

Thanks,
Rui