[RFC,00/15] ARM: perf: support multiple PMUs

Commit Message

Mark Rutland Aug. 15, 2011, 1:42 p.m. UTC

System (AKA nest or uncore) PMUs exist on devices which are not affine
to a single CPU. They usually cannot be directly associated with
individual tasks and are asynchronous with respect to the current
execution. Examples of devices which could have system PMUs include L2
cache controllers, GPUs and memory buses.

The following patch series refactors the ARM PMU backend, enabling
new PMUs to reuse the existing code. This should allow for system PMUs
to be supported in future. Further work will be required to get perf to
fully understand system PMUs, but this provides something usable.

The framework is intended to be used by system PMUs which hang off core
platform components (e.g. L2 cache, AXI bus). If a device is complex
enough or separate enough from core functionality to have its own
driver, it should implement its own PMU handling using the core perf
API directly.

The first patch ("perf: provide PMU when initing events") is currently
sitting in the tip tree, but as it's required for event initialization
to function (and hence for the PMU to be usable), it's provided here
for convenience.

The series is based on Will Deacon's perf-updates branch at:
	git://linux-arm.org/linux-2.6-wd.git perf-updates

An example driver using the framework (supporting the PMU present in
L220/PL310 level 2 cache controllers) can be found at:
	git://linux-arm.org/linux-2.6-wd.git perf-l2x0-wip

Any comments would be welcome.

Thanks,
Mark.

Mark Rutland (15):
  perf: provide PMU when initing events
  ARM: perf: only register a CPU PMU when present
  ARM: perf: clean up event group validation
  ARM: perf: remove active_mask
  ARM: perf: move active_events into struct arm_pmu
  ARM: perf: move platform device to struct arm_pmu
  ARM: perf: indirect access to cpu_hw_events
  ARM: perf: remove unnecessary armpmu->stop
  ARM: perf: lock PMU registers per-CPU
  ARM: perf: add type field to struct arm_pmu
  ARM: perf: refactor event mapping
  ARM: perf: add support for multiple PMUs
  ARM: perf: remove event limit from pmu_hw_events
  ARM: perf: remove cpu-related misnomers
  ARM: perf: move arm_pmu into <asm/pmu.h>

 arch/arm/include/asm/pmu.h          |   64 +++++++
 arch/arm/kernel/perf_event.c        |  318 +++++++++++++++++------------------
 arch/arm/kernel/perf_event_v6.c     |   73 ++++++---
 arch/arm/kernel/perf_event_v7.c     |   74 +++++---
 arch/arm/kernel/perf_event_xscale.c |   76 +++++----
 kernel/events/core.c                |    4 +-
 6 files changed, 359 insertions(+), 250 deletions(-)

From 1fdff971bb15ab62e70cbb111c4588c3d469843d Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Tue, 26 Apr 2011 13:00:07 +0100
Subject: [RFC PATCH 01/15] perf: provide PMU when initing events

Currently, an event's 'pmu' field is set after pmu::event_init() is
called. This means that pmu::event_init() must figure out which struct
pmu the event was initialised from. This makes it difficult to
consolidate common event initialisation code for similar PMUs, and
very difficult to implement drivers for PMUs which can have multiple
instances (e.g. a USB controller PMU, a GPU PMU, etc).

This patch sets the 'pmu' field before initialising the event, allowing
event init code to identify the struct pmu instance easily. In the
event of failure to initialise an event, the event is destroyed via
kfree() without calling perf_event::destroy(), so this shouldn't
result in bad behaviour even if the destroy field was set before
failure to initialise was noted.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 kernel/events/core.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

Comments

Mark Rutland Aug. 15, 2011, 1:45 p.m. UTC | #1

Hi all,

Sorry for the spam.

Evidently git send-email doesn't automatically split mboxes.

I'll resend this in a more manageable form.

Thanks,
Mark.

Patch

diff --git a/kernel/events/core.c b/kernel/events/core.c
index b8785e2..68c8017 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -5715,6 +5715,7 @@  struct pmu *perf_init_event(struct perf_event *event)
 	pmu = idr_find(&pmu_idr, event->attr.type);
 	rcu_read_unlock();
 	if (pmu) {
+		event->pmu = pmu;
 		ret = pmu->event_init(event);
 		if (ret)
 			pmu = ERR_PTR(ret);
@@ -5722,6 +5723,7 @@  struct pmu *perf_init_event(struct perf_event *event)
 	}
 
 	list_for_each_entry_rcu(pmu, &pmus, entry) {
+		event->pmu = pmu;
 		ret = pmu->event_init(event);
 		if (!ret)
 			goto unlock;
@@ -5848,8 +5850,6 @@  done:
 		return ERR_PTR(err);
 	}
 
-	event->pmu = pmu;
-
 	if (!event->parent) {
 		if (event->attach_state & PERF_ATTACH_TASK)
 			jump_label_inc(&perf_sched_events);
-- 
1.7.0.4


From 77d45290e0cdfa31909596bdfa2e3ee71da912aa Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Wed, 27 Apr 2011 10:31:51 +0100
Subject: [RFC PATCH 02/15] ARM: perf: only register a CPU PMU when present

Currently, an "empty" struct pmu is registered as the CPU PMU,
regardless of whether there is a physical PMU. This burdens the
accessor functions with checks to see whether a PMU is actually
present.

This patch changes initialisation to register a PMU only if there is a
supported PMU present, and removes the checks that this change makes
redundant.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/kernel/perf_event.c |   18 ++----------------
 1 files changed, 2 insertions(+), 16 deletions(-)

diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index 5d60c9c..753648c 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -250,9 +250,6 @@  armpmu_stop(struct perf_event *event, int flags)
 {
 	struct hw_perf_event *hwc = &event->hw;
 
-	if (!armpmu)
-		return;
-
 	/*
 	 * ARM pmu always has to update the counter, so ignore
 	 * PERF_EF_UPDATE, see comments in armpmu_start().
@@ -270,9 +267,6 @@  armpmu_start(struct perf_event *event, int flags)
 {
 	struct hw_perf_event *hwc = &event->hw;
 
-	if (!armpmu)
-		return;
-
 	/*
 	 * ARM pmu always has to reprogram the period, so ignore
 	 * PERF_EF_RELOAD, see the comment below.
@@ -567,9 +561,6 @@  static int armpmu_event_init(struct perf_event *event)
 		return -ENOENT;
 	}
 
-	if (!armpmu)
-		return -ENODEV;
-
 	event->destroy = hw_perf_event_destroy;
 
 	if (!atomic_inc_not_zero(&active_events)) {
@@ -599,9 +590,6 @@  static void armpmu_enable(struct pmu *pmu)
 	int idx, enabled = 0;
 	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
 
-	if (!armpmu)
-		return;
-
 	for (idx = 0; idx < armpmu->num_events; ++idx) {
 		struct perf_event *event = cpuc->events[idx];
 
@@ -618,8 +606,7 @@  static void armpmu_enable(struct pmu *pmu)
 
 static void armpmu_disable(struct pmu *pmu)
 {
-	if (armpmu)
-		armpmu->stop();
+	armpmu->stop();
 }
 
 static struct pmu pmu = {
@@ -738,12 +725,11 @@  init_hw_perf_events(void)
 	if (armpmu) {
 		pr_info("enabled with %s PMU driver, %d counters available\n",
 			armpmu->name, armpmu->num_events);
+		perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
 	} else {
 		pr_info("no hardware support available\n");
 	}
 
-	perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
-
 	return 0;
 }
 early_initcall(init_hw_perf_events);
-- 
1.7.0.4


From 33050bce4782ea40902a4d27492664e7b5a1cf15 Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Wed, 27 Apr 2011 16:22:21 +0100
Subject: [RFC PATCH 03/15] ARM: perf: clean up event group validation

Currently, event group validation compares each event's 'pmu' pointer
against the static 'pmu' pointer. This limits the code to supporting
only 1 PMU.

This patch changes the behaviour to consider an event's group leader's
'pmu' pointer as canonical for validation. This should ease later
generalisation of the code to support multiple PMUs at once.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/kernel/perf_event.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index 753648c..dfde928 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -348,8 +348,9 @@  validate_event(struct cpu_hw_events *cpuc,
 	       struct perf_event *event)
 {
 	struct hw_perf_event fake_event = event->hw;
+	struct pmu *leader_pmu = event->group_leader->pmu;
 
-	if (event->pmu != &pmu || event->state <= PERF_EVENT_STATE_OFF)
+	if (event->pmu != leader_pmu || event->state <= PERF_EVENT_STATE_OFF)
 		return 1;
 
 	return armpmu->get_event_idx(cpuc, &fake_event) >= 0;
-- 
1.7.0.4


From cabd32d06c3bd5092a34b7c37d19f21374d2f30b Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Tue, 19 Jul 2011 09:37:10 +0100
Subject: [RFC PATCH 04/15] ARM: perf: remove active_mask

Currently, pmu_hw_events::active_mask is used to keep track of which
events are active in hardware. As we can stop counters and their
interrupts, this is unnecessary.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/kernel/perf_event.c        |    8 --------
 arch/arm/kernel/perf_event_v6.c     |   19 ++++++++++++++++++-
 arch/arm/kernel/perf_event_v7.c     |    3 ---
 arch/arm/kernel/perf_event_xscale.c |    6 ------
 4 files changed, 18 insertions(+), 18 deletions(-)

diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index dfde928..438482f 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -57,12 +57,6 @@  struct cpu_hw_events {
 	 * an event. A 0 means that the counter can be used.
 	 */
 	unsigned long		used_mask[BITS_TO_LONGS(ARMPMU_MAX_HWEVENTS)];
-
-	/*
-	 * A 1 bit for an index indicates that the counter is actively being
-	 * used.
-	 */
-	unsigned long		active_mask[BITS_TO_LONGS(ARMPMU_MAX_HWEVENTS)];
 };
 static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events);
 
@@ -295,7 +289,6 @@  armpmu_del(struct perf_event *event, int flags)
 
 	WARN_ON(idx < 0);
 
-	clear_bit(idx, cpuc->active_mask);
 	armpmu_stop(event, PERF_EF_UPDATE);
 	cpuc->events[idx] = NULL;
 	clear_bit(idx, cpuc->used_mask);
@@ -327,7 +320,6 @@  armpmu_add(struct perf_event *event, int flags)
 	event->hw.idx = idx;
 	armpmu->disable(hwc, idx);
 	cpuc->events[idx] = event;
-	set_bit(idx, cpuc->active_mask);
 
 	hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE;
 	if (flags & PERF_EF_START)
diff --git a/arch/arm/kernel/perf_event_v6.c b/arch/arm/kernel/perf_event_v6.c
index 87f29b5..8390128 100644
--- a/arch/arm/kernel/perf_event_v6.c
+++ b/arch/arm/kernel/perf_event_v6.c
@@ -462,6 +462,23 @@  armv6pmu_enable_event(struct hw_perf_event *hwc,
 	raw_spin_unlock_irqrestore(&pmu_lock, flags);
 }
 
+static int counter_is_active(unsigned long pmcr, int idx)
+{
+	unsigned long mask = 0;
+	if (idx == ARMV6_CYCLE_COUNTER)
+		mask = ARMV6_PMCR_CCOUNT_IEN;
+	else if (idx == ARMV6_COUNTER0)
+		mask = ARMV6_PMCR_COUNT0_IEN;
+	else if (idx == ARMV6_COUNTER1)
+		mask = ARMV6_PMCR_COUNT1_IEN;
+
+	if (mask)
+		return pmcr & mask;
+
+	WARN_ONCE(1, "invalid counter number (%d)\n", idx);
+	return 0;
+}
+
 static irqreturn_t
 armv6pmu_handle_irq(int irq_num,
 		    void *dev)
@@ -491,7 +508,7 @@  armv6pmu_handle_irq(int irq_num,
 		struct perf_event *event = cpuc->events[idx];
 		struct hw_perf_event *hwc;
 
-		if (!test_bit(idx, cpuc->active_mask))
+		if (!counter_is_active(pmcr, idx))
 			continue;
 
 		/*
diff --git a/arch/arm/kernel/perf_event_v7.c b/arch/arm/kernel/perf_event_v7.c
index fe6c931..f4170fc 100644
--- a/arch/arm/kernel/perf_event_v7.c
+++ b/arch/arm/kernel/perf_event_v7.c
@@ -1022,9 +1022,6 @@  static irqreturn_t armv7pmu_handle_irq(int irq_num, void *dev)
 		struct perf_event *event = cpuc->events[idx];
 		struct hw_perf_event *hwc;
 
-		if (!test_bit(idx, cpuc->active_mask))
-			continue;
-
 		/*
 		 * We have a single interrupt for all counters. Check that
 		 * each counter has overflowed before we process it.
diff --git a/arch/arm/kernel/perf_event_xscale.c b/arch/arm/kernel/perf_event_xscale.c
index 54312fc..ca89a06 100644
--- a/arch/arm/kernel/perf_event_xscale.c
+++ b/arch/arm/kernel/perf_event_xscale.c
@@ -253,9 +253,6 @@  xscale1pmu_handle_irq(int irq_num, void *dev)
 		struct perf_event *event = cpuc->events[idx];
 		struct hw_perf_event *hwc;
 
-		if (!test_bit(idx, cpuc->active_mask))
-			continue;
-
 		if (!xscale1_pmnc_counter_has_overflowed(pmnc, idx))
 			continue;
 
@@ -585,9 +582,6 @@  xscale2pmu_handle_irq(int irq_num, void *dev)
 		struct perf_event *event = cpuc->events[idx];
 		struct hw_perf_event *hwc;
 
-		if (!test_bit(idx, cpuc->active_mask))
-			continue;
-
 		if (!xscale2_pmnc_counter_has_overflowed(pmnc, idx))
 			continue;
 
-- 
1.7.0.4


From d227015b5edc20cd671ea5212ab273b9619f3d7a Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Wed, 27 Apr 2011 11:20:11 +0100
Subject: [RFC PATCH 05/15] ARM: perf: move active_events into struct arm_pmu

This patch moves the active_events counter into struct arm_pmu, in
preparation for supporting multiple PMUs. This also moves
pmu_reserve_mutex, as it is used to guard accesses to active_events.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/kernel/perf_event.c |   31 ++++++++++++++++++++-----------
 1 files changed, 20 insertions(+), 11 deletions(-)

diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index 438482f..9874395 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -82,6 +82,8 @@  struct arm_pmu {
 	const unsigned	(*event_map)[PERF_COUNT_HW_MAX];
 	u32		raw_event_mask;
 	int		num_events;
+	atomic_t	active_events;
+	struct mutex	reserve_mutex;
 	u64		max_period;
 };
 
@@ -454,15 +456,15 @@  armpmu_reserve_hardware(void)
 	return 0;
 }
 
-static atomic_t active_events = ATOMIC_INIT(0);
-static DEFINE_MUTEX(pmu_reserve_mutex);
-
 static void
 hw_perf_event_destroy(struct perf_event *event)
 {
-	if (atomic_dec_and_mutex_lock(&active_events, &pmu_reserve_mutex)) {
+	atomic_t *active_events	 = &armpmu->active_events;
+	struct mutex *pmu_reserve_mutex = &armpmu->reserve_mutex;
+
+	if (atomic_dec_and_mutex_lock(active_events, pmu_reserve_mutex)) {
 		armpmu_release_hardware();
-		mutex_unlock(&pmu_reserve_mutex);
+		mutex_unlock(pmu_reserve_mutex);
 	}
 }
 
@@ -543,6 +545,7 @@  __hw_perf_event_init(struct perf_event *event)
 static int armpmu_event_init(struct perf_event *event)
 {
 	int err = 0;
+	atomic_t *active_events = &armpmu->active_events;
 
 	switch (event->attr.type) {
 	case PERF_TYPE_RAW:
@@ -556,15 +559,14 @@  static int armpmu_event_init(struct perf_event *event)
 
 	event->destroy = hw_perf_event_destroy;
 
-	if (!atomic_inc_not_zero(&active_events)) {
-		mutex_lock(&pmu_reserve_mutex);
-		if (atomic_read(&active_events) == 0) {
+	if (!atomic_inc_not_zero(active_events)) {
+		mutex_lock(&armpmu->reserve_mutex);
+		if (atomic_read(active_events) == 0)
 			err = armpmu_reserve_hardware();
-		}
 
 		if (!err)
-			atomic_inc(&active_events);
-		mutex_unlock(&pmu_reserve_mutex);
+			atomic_inc(active_events);
+		mutex_unlock(&armpmu->reserve_mutex);
 	}
 
 	if (err)
@@ -613,6 +615,12 @@  static struct pmu pmu = {
 	.read		= armpmu_read,
 };
 
+static void __init armpmu_init(struct arm_pmu *armpmu)
+{
+	atomic_set(&armpmu->active_events, 0);
+	mutex_init(&armpmu->reserve_mutex);
+}
+
 /* Include the PMU-specific implementations. */
 #include "perf_event_xscale.c"
 #include "perf_event_v6.c"
@@ -718,6 +726,7 @@  init_hw_perf_events(void)
 	if (armpmu) {
 		pr_info("enabled with %s PMU driver, %d counters available\n",
 			armpmu->name, armpmu->num_events);
+		armpmu_init(armpmu);
 		perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
 	} else {
 		pr_info("no hardware support available\n");
-- 
1.7.0.4


From 782b2a29e2677b92dc90d5df734db141c4ae5bf8 Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Wed, 4 May 2011 09:23:15 +0100
Subject: [RFC PATCH 06/15] ARM: perf: move platform device to struct arm_pmu

Currently the ARM perf code supports having a single struct
platform_device to supply IRQ numbers, limiting it to supporting a
single PMU.

This patch makes a platform_device instance variable on struct arm_pmu.
This should allow for multiple PMUs to be supported in future.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/kernel/perf_event.c |   10 ++++++----
 1 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index 9874395..1d648d1 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -26,8 +26,6 @@ 
 #include <asm/pmu.h>
 #include <asm/stacktrace.h>
 
-static struct platform_device *pmu_device;
-
 /*
  * Hardware lock to serialize accesses to PMU registers. Needed for the
  * read/modify/write sequences.
@@ -85,6 +83,7 @@  struct arm_pmu {
 	atomic_t	active_events;
 	struct mutex	reserve_mutex;
 	u64		max_period;
+	struct platform_device	*plat_device;
 };
 
 /* Set at runtime when we know what CPU type we are. */
@@ -374,7 +373,8 @@  validate_group(struct perf_event *event)
 
 static irqreturn_t armpmu_platform_irq(int irq, void *dev)
 {
-	struct arm_pmu_platdata *plat = dev_get_platdata(&pmu_device->dev);
+	struct platform_device *plat_device = armpmu->plat_device;
+	struct arm_pmu_platdata *plat = dev_get_platdata(&plat_device->dev);
 
 	return plat->handle_irq(irq, dev, armpmu->handle_irq);
 }
@@ -383,6 +383,7 @@  static void
 armpmu_release_hardware(void)
 {
 	int i, irq, irqs;
+	struct platform_device *pmu_device = armpmu->plat_device;
 
 	irqs = min(pmu_device->num_resources, num_possible_cpus());
 
@@ -404,6 +405,7 @@  armpmu_reserve_hardware(void)
 	struct arm_pmu_platdata *plat;
 	irq_handler_t handle_irq;
 	int i, err, irq, irqs;
+	struct platform_device *pmu_device = armpmu->plat_device;
 
 	err = reserve_pmu(ARM_PMU_DEVICE_CPU);
 	if (err) {
@@ -657,7 +659,7 @@  static struct platform_device_id armpmu_plat_device_ids[] = {
 
 static int __devinit armpmu_device_probe(struct platform_device *pdev)
 {
-	pmu_device = pdev;
+	armpmu->plat_device = pdev;
 	return 0;
 }
 
-- 
1.7.0.4


From e63510b8059280c0b4a6fd7c59b43b3435feb309 Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Wed, 4 May 2011 09:23:51 +0100
Subject: [RFC PATCH 07/15] ARM: perf: indirect access to cpu_hw_events

Currently, cpu_hw_events is a global per-CPU variable. To enable
support for multiple PMUs, there needs to be a mapping from an instance
of arm_pmu to its cpu_hw_events. Additionally, as system PMUs are not
CPU-affine, they should not have this stored per-CPU.

This patch moves access to the hardware events data behind an accessor
function (arm_pmu::get_hw_events). This allows each instance to have
its own hardware event data, which can be stored per-CPU or globally as
required.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/kernel/perf_event.c |   18 +++++++++++++++---
 1 files changed, 15 insertions(+), 3 deletions(-)

diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index 1d648d1..9d6ac99 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -84,6 +84,7 @@  struct arm_pmu {
 	struct mutex	reserve_mutex;
 	u64		max_period;
 	struct platform_device	*plat_device;
+	struct cpu_hw_events	*(*get_hw_events)(void);
 };
 
 /* Set at runtime when we know what CPU type we are. */
@@ -284,7 +285,7 @@  armpmu_start(struct perf_event *event, int flags)
 static void
 armpmu_del(struct perf_event *event, int flags)
 {
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct cpu_hw_events *cpuc = armpmu->get_hw_events();
 	struct hw_perf_event *hwc = &event->hw;
 	int idx = hwc->idx;
 
@@ -300,7 +301,7 @@  armpmu_del(struct perf_event *event, int flags)
 static int
 armpmu_add(struct perf_event *event, int flags)
 {
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct cpu_hw_events *cpuc = armpmu->get_hw_events();
 	struct hw_perf_event *hwc = &event->hw;
 	int idx;
 	int err = 0;
@@ -585,7 +586,7 @@  static void armpmu_enable(struct pmu *pmu)
 {
 	/* Enable all of the perf events on hardware. */
 	int idx, enabled = 0;
-	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
+	struct cpu_hw_events *cpuc = armpmu->get_hw_events();
 
 	for (idx = 0; idx < armpmu->num_events; ++idx) {
 		struct perf_event *event = cpuc->events[idx];
@@ -678,6 +679,16 @@  static int __init register_pmu_driver(void)
 }
 device_initcall(register_pmu_driver);
 
+static struct cpu_hw_events *armpmu_get_cpu_events(void)
+{
+	return &__get_cpu_var(cpu_hw_events);
+}
+
+static void __init cpu_pmu_init(struct arm_pmu *armpmu)
+{
+	armpmu->get_hw_events = armpmu_get_cpu_events;
+}
+
 /*
  * CPU PMU identification and registration.
  */
@@ -728,6 +739,7 @@  init_hw_perf_events(void)
 	if (armpmu) {
 		pr_info("enabled with %s PMU driver, %d counters available\n",
 			armpmu->name, armpmu->num_events);
+		cpu_pmu_init(armpmu);
 		armpmu_init(armpmu);
 		perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
 	} else {
-- 
1.7.0.4


From 1ffac3c1163c0154a5dd2a7b4d5051d0ac64b1f7 Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Mon, 8 Aug 2011 14:56:02 +0100
Subject: [RFC PATCH 08/15] ARM: perf: remove unnecessary armpmu->stop

As armpmu_disable will call armpmu->stop when the last event has been
removed, this is pointless and simply adds to the noise when debugging.
Additionally, due to this call occurring in a preemptible context, this
is problematic for per-cpu locking of PMU registers (where we will
attempt to access per-cpu spinlock for use with raw_spin_lock_irqsave).

This patch removes the call to armpmu->stop.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/kernel/perf_event.c |    1 -
 1 files changed, 0 insertions(+), 1 deletions(-)

diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index 9d6ac99..5ce6c33 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -396,7 +396,6 @@  armpmu_release_hardware(void)
 			free_irq(irq, NULL);
 	}
 
-	armpmu->stop();
 	release_pmu(ARM_PMU_DEVICE_CPU);
 }
 
-- 
1.7.0.4


From 1b8a488707f44bf75ef2169335a3b2b1cabcc9ff Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Thu, 28 Apr 2011 10:17:04 +0100
Subject: [RFC PATCH 09/15] ARM: perf: lock PMU registers per-CPU

Currently, a single lock serialises access to CPU PMU registers. This
global locking is unnecessary as PMU registers are local to the CPU
they monitor.

This patch replaces the global lock with a per-CPU lock. As the lock is
in struct cpu_hw_events, PMUs providing a single cpu_hw_events instance
can be locked globally.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/kernel/perf_event.c        |   17 +++++++++-----
 arch/arm/kernel/perf_event_v6.c     |   25 +++++++++++++--------
 arch/arm/kernel/perf_event_v7.c     |   20 ++++++++++-------
 arch/arm/kernel/perf_event_xscale.c |   40 +++++++++++++++++++++--------------
 4 files changed, 62 insertions(+), 40 deletions(-)

diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index 5ce6c33..9331d57 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -27,12 +27,6 @@ 
 #include <asm/stacktrace.h>
 
 /*
- * Hardware lock to serialize accesses to PMU registers. Needed for the
- * read/modify/write sequences.
- */
-static DEFINE_RAW_SPINLOCK(pmu_lock);
-
-/*
  * ARMv6 supports a maximum of 3 events, starting from index 0. If we add
  * another platform that supports more, we need to increase this to be the
  * largest of all platforms.
@@ -55,6 +49,12 @@  struct cpu_hw_events {
 	 * an event. A 0 means that the counter can be used.
 	 */
 	unsigned long		used_mask[BITS_TO_LONGS(ARMPMU_MAX_HWEVENTS)];
+
+	/*
+	 * Hardware lock to serialize accesses to PMU registers. Needed for the
+	 * read/modify/write sequences.
+	 */
+	raw_spinlock_t		pmu_lock;
 };
 static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events);
 
@@ -685,6 +685,11 @@  static struct cpu_hw_events *armpmu_get_cpu_events(void)
 
 static void __init cpu_pmu_init(struct arm_pmu *armpmu)
 {
+	int cpu;
+	for_each_possible_cpu(cpu) {
+		struct cpu_hw_events *events = &per_cpu(cpu_hw_events, cpu);
+		raw_spin_lock_init(&events->pmu_lock);
+	}
 	armpmu->get_hw_events = armpmu_get_cpu_events;
 }
 
diff --git a/arch/arm/kernel/perf_event_v6.c b/arch/arm/kernel/perf_event_v6.c
index 8390128..68cf704 100644
--- a/arch/arm/kernel/perf_event_v6.c
+++ b/arch/arm/kernel/perf_event_v6.c
@@ -433,6 +433,7 @@  armv6pmu_enable_event(struct hw_perf_event *hwc,
 		      int idx)
 {
 	unsigned long val, mask, evt, flags;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
 	if (ARMV6_CYCLE_COUNTER == idx) {
 		mask	= 0;
@@ -454,12 +455,12 @@  armv6pmu_enable_event(struct hw_perf_event *hwc,
 	 * Mask out the current event and set the counter to count the event
 	 * that we're interested in.
 	 */
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = armv6_pmcr_read();
 	val &= ~mask;
 	val |= evt;
 	armv6_pmcr_write(val);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static int counter_is_active(unsigned long pmcr, int idx)
@@ -544,24 +545,26 @@  static void
 armv6pmu_start(void)
 {
 	unsigned long flags, val;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = armv6_pmcr_read();
 	val |= ARMV6_PMCR_ENABLE;
 	armv6_pmcr_write(val);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static void
 armv6pmu_stop(void)
 {
 	unsigned long flags, val;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = armv6_pmcr_read();
 	val &= ~ARMV6_PMCR_ENABLE;
 	armv6_pmcr_write(val);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static int
@@ -595,6 +598,7 @@  armv6pmu_disable_event(struct hw_perf_event *hwc,
 		       int idx)
 {
 	unsigned long val, mask, evt, flags;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
 	if (ARMV6_CYCLE_COUNTER == idx) {
 		mask	= ARMV6_PMCR_CCOUNT_IEN;
@@ -615,12 +619,12 @@  armv6pmu_disable_event(struct hw_perf_event *hwc,
 	 * of ETM bus signal assertion cycles. The external reporting should
 	 * be disabled and so this should never increment.
 	 */
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = armv6_pmcr_read();
 	val &= ~mask;
 	val |= evt;
 	armv6_pmcr_write(val);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static void
@@ -628,6 +632,7 @@  armv6mpcore_pmu_disable_event(struct hw_perf_event *hwc,
 			      int idx)
 {
 	unsigned long val, mask, flags, evt = 0;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
 	if (ARMV6_CYCLE_COUNTER == idx) {
 		mask	= ARMV6_PMCR_CCOUNT_IEN;
@@ -644,12 +649,12 @@  armv6mpcore_pmu_disable_event(struct hw_perf_event *hwc,
 	 * Unlike UP ARMv6, we don't have a way of stopping the counters. We
 	 * simply disable the interrupt reporting.
 	 */
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = armv6_pmcr_read();
 	val &= ~mask;
 	val |= evt;
 	armv6_pmcr_write(val);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static struct arm_pmu armv6pmu = {
diff --git a/arch/arm/kernel/perf_event_v7.c b/arch/arm/kernel/perf_event_v7.c
index f4170fc..68ac522 100644
--- a/arch/arm/kernel/perf_event_v7.c
+++ b/arch/arm/kernel/perf_event_v7.c
@@ -936,12 +936,13 @@  static void armv7_pmnc_dump_regs(void)
 static void armv7pmu_enable_event(struct hw_perf_event *hwc, int idx)
 {
 	unsigned long flags;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
 	/*
 	 * Enable counter and interrupt, and set the counter to count
 	 * the event that we're interested in.
 	 */
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 
 	/*
 	 * Disable counter
@@ -966,17 +967,18 @@  static void armv7pmu_enable_event(struct hw_perf_event *hwc, int idx)
 	 */
 	armv7_pmnc_enable_counter(idx);
 
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static void armv7pmu_disable_event(struct hw_perf_event *hwc, int idx)
 {
 	unsigned long flags;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
 	/*
 	 * Disable counter and interrupt
 	 */
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 
 	/*
 	 * Disable counter
@@ -988,7 +990,7 @@  static void armv7pmu_disable_event(struct hw_perf_event *hwc, int idx)
 	 */
 	armv7_pmnc_disable_intens(idx);
 
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static irqreturn_t armv7pmu_handle_irq(int irq_num, void *dev)
@@ -1054,21 +1056,23 @@  static irqreturn_t armv7pmu_handle_irq(int irq_num, void *dev)
 static void armv7pmu_start(void)
 {
 	unsigned long flags;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	/* Enable all counters */
 	armv7_pmnc_write(armv7_pmnc_read() | ARMV7_PMNC_E);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static void armv7pmu_stop(void)
 {
 	unsigned long flags;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	/* Disable all counters */
 	armv7_pmnc_write(armv7_pmnc_read() & ~ARMV7_PMNC_E);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static int armv7pmu_get_event_idx(struct cpu_hw_events *cpuc,
diff --git a/arch/arm/kernel/perf_event_xscale.c b/arch/arm/kernel/perf_event_xscale.c
index ca89a06..18e4823 100644
--- a/arch/arm/kernel/perf_event_xscale.c
+++ b/arch/arm/kernel/perf_event_xscale.c
@@ -281,6 +281,7 @@  static void
 xscale1pmu_enable_event(struct hw_perf_event *hwc, int idx)
 {
 	unsigned long val, mask, evt, flags;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
 	switch (idx) {
 	case XSCALE_CYCLE_COUNTER:
@@ -302,18 +303,19 @@  xscale1pmu_enable_event(struct hw_perf_event *hwc, int idx)
 		return;
 	}
 
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = xscale1pmu_read_pmnc();
 	val &= ~mask;
 	val |= evt;
 	xscale1pmu_write_pmnc(val);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static void
 xscale1pmu_disable_event(struct hw_perf_event *hwc, int idx)
 {
 	unsigned long val, mask, evt, flags;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
 	switch (idx) {
 	case XSCALE_CYCLE_COUNTER:
@@ -333,12 +335,12 @@  xscale1pmu_disable_event(struct hw_perf_event *hwc, int idx)
 		return;
 	}
 
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = xscale1pmu_read_pmnc();
 	val &= ~mask;
 	val |= evt;
 	xscale1pmu_write_pmnc(val);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static int
@@ -365,24 +367,26 @@  static void
 xscale1pmu_start(void)
 {
 	unsigned long flags, val;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = xscale1pmu_read_pmnc();
 	val |= XSCALE_PMU_ENABLE;
 	xscale1pmu_write_pmnc(val);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static void
 xscale1pmu_stop(void)
 {
 	unsigned long flags, val;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = xscale1pmu_read_pmnc();
 	val &= ~XSCALE_PMU_ENABLE;
 	xscale1pmu_write_pmnc(val);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static inline u32
@@ -610,6 +614,7 @@  static void
 xscale2pmu_enable_event(struct hw_perf_event *hwc, int idx)
 {
 	unsigned long flags, ien, evtsel;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
 	ien = xscale2pmu_read_int_enable();
 	evtsel = xscale2pmu_read_event_select();
@@ -643,16 +648,17 @@  xscale2pmu_enable_event(struct hw_perf_event *hwc, int idx)
 		return;
 	}
 
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	xscale2pmu_write_event_select(evtsel);
 	xscale2pmu_write_int_enable(ien);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static void
 xscale2pmu_disable_event(struct hw_perf_event *hwc, int idx)
 {
 	unsigned long flags, ien, evtsel;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
 	ien = xscale2pmu_read_int_enable();
 	evtsel = xscale2pmu_read_event_select();
@@ -686,10 +692,10 @@  xscale2pmu_disable_event(struct hw_perf_event *hwc, int idx)
 		return;
 	}
 
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	xscale2pmu_write_event_select(evtsel);
 	xscale2pmu_write_int_enable(ien);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static int
@@ -712,24 +718,26 @@  static void
 xscale2pmu_start(void)
 {
 	unsigned long flags, val;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = xscale2pmu_read_pmnc() & ~XSCALE_PMU_CNT64;
 	val |= XSCALE_PMU_ENABLE;
 	xscale2pmu_write_pmnc(val);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static void
 xscale2pmu_stop(void)
 {
 	unsigned long flags, val;
+	struct cpu_hw_events *events = armpmu->get_hw_events();
 
-	raw_spin_lock_irqsave(&pmu_lock, flags);
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = xscale2pmu_read_pmnc();
 	val &= ~XSCALE_PMU_ENABLE;
 	xscale2pmu_write_pmnc(val);
-	raw_spin_unlock_irqrestore(&pmu_lock, flags);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
 static inline u32
-- 
1.7.0.4


From a37500b3166b550e115772dec7192fe89511fa7b Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Mon, 6 Jun 2011 10:37:50 +0100
Subject: [RFC PATCH 10/15] ARM: perf: add type field to struct arm_pmu

Currently, the ARM perf code assumes all PMUs it will handle are
CPU PMUs, having ARM_PMU_DEVICE_CPU hardcoded when reserving or
releasing hardware. This means that currently, the ARM perf code can't
support system PMUs.

This patch adds a 'type' field to struct arm_pmu, which allows the code
to reserve & release the hardware regardless of the PMU type.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/kernel/perf_event.c |    6 ++++--
 1 files changed, 4 insertions(+), 2 deletions(-)

diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index 9331d57..1a2ebbf 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -60,6 +60,7 @@  static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events);
 
 struct arm_pmu {
 	enum arm_perf_pmu_ids id;
+	enum arm_pmu_type type;
 	cpumask_t	active_irqs;
 	const char	*name;
 	irqreturn_t	(*handle_irq)(int irq_num, void *dev);
@@ -396,7 +397,7 @@  armpmu_release_hardware(void)
 			free_irq(irq, NULL);
 	}
 
-	release_pmu(ARM_PMU_DEVICE_CPU);
+	release_pmu(armpmu->type);
 }
 
 static int
@@ -407,7 +408,7 @@  armpmu_reserve_hardware(void)
 	int i, err, irq, irqs;
 	struct platform_device *pmu_device = armpmu->plat_device;
 
-	err = reserve_pmu(ARM_PMU_DEVICE_CPU);
+	err = reserve_pmu(armpmu->type);
 	if (err) {
 		pr_warning("unable to reserve pmu\n");
 		return err;
@@ -691,6 +692,7 @@  static void __init cpu_pmu_init(struct arm_pmu *armpmu)
 		raw_spin_lock_init(&events->pmu_lock);
 	}
 	armpmu->get_hw_events = armpmu_get_cpu_events;
+	armpmu->type = ARM_PMU_DEVICE_CPU;
 }
 
 /*
-- 
1.7.0.4


From 1bb54fbf1c3d279d464a01ad0b9906e297e9a600 Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Thu, 28 Apr 2011 15:47:10 +0100
Subject: [RFC PATCH 11/15] ARM: perf: refactor event mapping

Currently mapping an event type to a hardware configuration value
depends on the data being pointed to from struct arm_pmu. These fields
(cache_map, event_map, raw_event_mask) are currently specific to CPU
PMUs, and do not serve the general case well.

This patch replaces the event map pointers on struct arm_pmu with a new
'map_event' function pointer. Small shim functions are used to reuse
the existing common code.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/kernel/perf_event.c        |   67 +++++++++++++++++++----------------
 arch/arm/kernel/perf_event_v6.c     |   21 ++++++++---
 arch/arm/kernel/perf_event_v7.c     |   37 ++++++++++++++-----
 arch/arm/kernel/perf_event_xscale.c |   14 ++++---
 4 files changed, 87 insertions(+), 52 deletions(-)

diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index 1a2ebbf..b13bf23 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -75,11 +75,7 @@  struct arm_pmu {
 	void		(*start)(void);
 	void		(*stop)(void);
 	void		(*reset)(void *);
-	const unsigned	(*cache_map)[PERF_COUNT_HW_CACHE_MAX]
-				    [PERF_COUNT_HW_CACHE_OP_MAX]
-				    [PERF_COUNT_HW_CACHE_RESULT_MAX];
-	const unsigned	(*event_map)[PERF_COUNT_HW_MAX];
-	u32		raw_event_mask;
+	int		(*map_event)(struct perf_event *event);
 	int		num_events;
 	atomic_t	active_events;
 	struct mutex	reserve_mutex;
@@ -129,7 +125,11 @@  EXPORT_SYMBOL_GPL(perf_num_counters);
 #define CACHE_OP_UNSUPPORTED		0xFFFF
 
 static int
-armpmu_map_cache_event(u64 config)
+armpmu_map_cache_event(const unsigned (*cache_map)
+				      [PERF_COUNT_HW_CACHE_MAX]
+				      [PERF_COUNT_HW_CACHE_OP_MAX]
+				      [PERF_COUNT_HW_CACHE_RESULT_MAX],
+		       u64 config)
 {
 	unsigned int cache_type, cache_op, cache_result, ret;
 
@@ -145,7 +145,7 @@  armpmu_map_cache_event(u64 config)
 	if (cache_result >= PERF_COUNT_HW_CACHE_RESULT_MAX)
 		return -EINVAL;
 
-	ret = (int)(*armpmu->cache_map)[cache_type][cache_op][cache_result];
+	ret = (int)(*cache_map)[cache_type][cache_op][cache_result];
 
 	if (ret == CACHE_OP_UNSUPPORTED)
 		return -ENOENT;
@@ -154,16 +154,38 @@  armpmu_map_cache_event(u64 config)
 }
 
 static int
-armpmu_map_event(u64 config)
+armpmu_map_event(const unsigned (*event_map)[PERF_COUNT_HW_MAX], u64 config)
 {
-	int mapping = (*armpmu->event_map)[config];
-	return mapping == HW_OP_UNSUPPORTED ? -EOPNOTSUPP : mapping;
+	int mapping = (*event_map)[config];
+	return mapping == HW_OP_UNSUPPORTED ? -ENOENT : mapping;
 }
 
 static int
-armpmu_map_raw_event(u64 config)
+armpmu_map_raw_event(u32 raw_event_mask, u64 config)
 {
-	return (int)(config & armpmu->raw_event_mask);
+	return (int)(config & raw_event_mask);
+}
+
+static int map_cpu_event(struct perf_event *event,
+			 const unsigned (*event_map)[PERF_COUNT_HW_MAX],
+			 const unsigned (*cache_map)
+					[PERF_COUNT_HW_CACHE_MAX]
+					[PERF_COUNT_HW_CACHE_OP_MAX]
+					[PERF_COUNT_HW_CACHE_RESULT_MAX],
+			 u32 raw_event_mask)
+{
+	u64 config = event->attr.config;
+
+	switch (event->attr.type) {
+	case PERF_TYPE_HARDWARE:
+		return armpmu_map_event(event_map, config);
+	case PERF_TYPE_HW_CACHE:
+		return armpmu_map_cache_event(cache_map, config);
+	case PERF_TYPE_RAW:
+		return armpmu_map_raw_event(raw_event_mask, config);
+	}
+
+	return -ENOENT;
 }
 
 static int
@@ -484,17 +506,7 @@  __hw_perf_event_init(struct perf_event *event)
 	struct hw_perf_event *hwc = &event->hw;
 	int mapping, err;
 
-	/* Decode the generic type into an ARM event identifier. */
-	if (PERF_TYPE_HARDWARE == event->attr.type) {
-		mapping = armpmu_map_event(event->attr.config);
-	} else if (PERF_TYPE_HW_CACHE == event->attr.type) {
-		mapping = armpmu_map_cache_event(event->attr.config);
-	} else if (PERF_TYPE_RAW == event->attr.type) {
-		mapping = armpmu_map_raw_event(event->attr.config);
-	} else {
-		pr_debug("event type %x not supported\n", event->attr.type);
-		return -EOPNOTSUPP;
-	}
+	mapping = armpmu->map_event(event);
 
 	if (mapping < 0) {
 		pr_debug("event %x:%llx not supported\n", event->attr.type,
@@ -550,15 +562,8 @@  static int armpmu_event_init(struct perf_event *event)
 	int err = 0;
 	atomic_t *active_events = &armpmu->active_events;
 
-	switch (event->attr.type) {
-	case PERF_TYPE_RAW:
-	case PERF_TYPE_HARDWARE:
-	case PERF_TYPE_HW_CACHE:
-		break;
-
-	default:
+	if (armpmu->map_event(event) == -ENOENT)
 		return -ENOENT;
-	}
 
 	event->destroy = hw_perf_event_destroy;
 
diff --git a/arch/arm/kernel/perf_event_v6.c b/arch/arm/kernel/perf_event_v6.c
index 68cf704..a4c5aa9 100644
--- a/arch/arm/kernel/perf_event_v6.c
+++ b/arch/arm/kernel/perf_event_v6.c
@@ -657,6 +657,12 @@  armv6mpcore_pmu_disable_event(struct hw_perf_event *hwc,
 	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
+static int armv6_map_event(struct perf_event *event)
+{
+	return map_cpu_event(event, &armv6_perf_map,
+				&armv6_perf_cache_map, 0xFF);
+}
+
 static struct arm_pmu armv6pmu = {
 	.id			= ARM_PERF_PMU_ID_V6,
 	.name			= "v6",
@@ -668,9 +674,7 @@  static struct arm_pmu armv6pmu = {
 	.get_event_idx		= armv6pmu_get_event_idx,
 	.start			= armv6pmu_start,
 	.stop			= armv6pmu_stop,
-	.cache_map		= &armv6_perf_cache_map,
-	.event_map		= &armv6_perf_map,
-	.raw_event_mask		= 0xFF,
+	.map_event		= armv6_map_event,
 	.num_events		= 3,
 	.max_period		= (1LLU << 32) - 1,
 };
@@ -687,6 +691,13 @@  static struct arm_pmu *__init armv6pmu_init(void)
  * disable the interrupt reporting and update the event. When unthrottling we
  * reset the period and enable the interrupt reporting.
  */
+
+static int armv6mpcore_map_event(struct perf_event *event)
+{
+	return map_cpu_event(event, &armv6mpcore_perf_map,
+				&armv6mpcore_perf_cache_map, 0xFF);
+}
+
 static struct arm_pmu armv6mpcore_pmu = {
 	.id			= ARM_PERF_PMU_ID_V6MP,
 	.name			= "v6mpcore",
@@ -698,9 +709,7 @@  static struct arm_pmu armv6mpcore_pmu = {
 	.get_event_idx		= armv6pmu_get_event_idx,
 	.start			= armv6pmu_start,
 	.stop			= armv6pmu_stop,
-	.cache_map		= &armv6mpcore_perf_cache_map,
-	.event_map		= &armv6mpcore_perf_map,
-	.raw_event_mask		= 0xFF,
+	.map_event		= armv6mpcore_map_event,
 	.num_events		= 3,
 	.max_period		= (1LLU << 32) - 1,
 };
diff --git a/arch/arm/kernel/perf_event_v7.c b/arch/arm/kernel/perf_event_v7.c
index 68ac522..be7b58a 100644
--- a/arch/arm/kernel/perf_event_v7.c
+++ b/arch/arm/kernel/perf_event_v7.c
@@ -1140,6 +1140,30 @@  static void armv7pmu_reset(void *info)
 	armv7_pmnc_write(ARMV7_PMNC_P | ARMV7_PMNC_C);
 }
 
+static int armv7_a8_map_event(struct perf_event *event)
+{
+	return map_cpu_event(event, &armv7_a8_perf_map,
+				&armv7_a8_perf_cache_map, 0xFF);
+}
+
+static int armv7_a9_map_event(struct perf_event *event)
+{
+	return map_cpu_event(event, &armv7_a9_perf_map,
+				&armv7_a9_perf_cache_map, 0xFF);
+}
+
+static int armv7_a5_map_event(struct perf_event *event)
+{
+	return map_cpu_event(event, &armv7_a5_perf_map,
+				&armv7_a5_perf_cache_map, 0xFF);
+}
+
+static int armv7_a15_map_event(struct perf_event *event)
+{
+	return map_cpu_event(event, &armv7_a15_perf_map,
+				&armv7_a15_perf_cache_map, 0xFF);
+}
+
 static struct arm_pmu armv7pmu = {
 	.handle_irq		= armv7pmu_handle_irq,
 	.enable			= armv7pmu_enable_event,
@@ -1150,7 +1174,6 @@  static struct arm_pmu armv7pmu = {
 	.start			= armv7pmu_start,
 	.stop			= armv7pmu_stop,
 	.reset			= armv7pmu_reset,
-	.raw_event_mask		= 0xFF,
 	.max_period		= (1LLU << 32) - 1,
 };
 
@@ -1169,8 +1192,7 @@  static struct arm_pmu *__init armv7_a8_pmu_init(void)
 {
 	armv7pmu.id		= ARM_PERF_PMU_ID_CA8;
 	armv7pmu.name		= "ARMv7 Cortex-A8";
-	armv7pmu.cache_map	= &armv7_a8_perf_cache_map;
-	armv7pmu.event_map	= &armv7_a8_perf_map;
+	armv7pmu.map_event	= armv7_a8_map_event;
 	armv7pmu.num_events	= armv7_read_num_pmnc_events();
 	return &armv7pmu;
 }
@@ -1179,8 +1201,7 @@  static struct arm_pmu *__init armv7_a9_pmu_init(void)
 {
 	armv7pmu.id		= ARM_PERF_PMU_ID_CA9;
 	armv7pmu.name		= "ARMv7 Cortex-A9";
-	armv7pmu.cache_map	= &armv7_a9_perf_cache_map;
-	armv7pmu.event_map	= &armv7_a9_perf_map;
+	armv7pmu.map_event	= armv7_a9_map_event;
 	armv7pmu.num_events	= armv7_read_num_pmnc_events();
 	return &armv7pmu;
 }
@@ -1189,8 +1210,7 @@  static struct arm_pmu *__init armv7_a5_pmu_init(void)
 {
 	armv7pmu.id		= ARM_PERF_PMU_ID_CA5;
 	armv7pmu.name		= "ARMv7 Cortex-A5";
-	armv7pmu.cache_map	= &armv7_a5_perf_cache_map;
-	armv7pmu.event_map	= &armv7_a5_perf_map;
+	armv7pmu.map_event	= armv7_a5_map_event;
 	armv7pmu.num_events	= armv7_read_num_pmnc_events();
 	return &armv7pmu;
 }
@@ -1199,8 +1219,7 @@  static struct arm_pmu *__init armv7_a15_pmu_init(void)
 {
 	armv7pmu.id		= ARM_PERF_PMU_ID_CA15;
 	armv7pmu.name		= "ARMv7 Cortex-A15";
-	armv7pmu.cache_map	= &armv7_a15_perf_cache_map;
-	armv7pmu.event_map	= &armv7_a15_perf_map;
+	armv7pmu.map_event	= armv7_a15_map_event;
 	armv7pmu.num_events	= armv7_read_num_pmnc_events();
 	armv7pmu.set_event_filter = armv7pmu_set_event_filter;
 	return &armv7pmu;
diff --git a/arch/arm/kernel/perf_event_xscale.c b/arch/arm/kernel/perf_event_xscale.c
index 18e4823..d4c7610 100644
--- a/arch/arm/kernel/perf_event_xscale.c
+++ b/arch/arm/kernel/perf_event_xscale.c
@@ -425,6 +425,12 @@  xscale1pmu_write_counter(int counter, u32 val)
 	}
 }
 
+static int xscale_map_event(struct perf_event *event)
+{
+	return map_cpu_event(event, &xscale_perf_map,
+				&xscale_perf_cache_map, 0xFF);
+}
+
 static struct arm_pmu xscale1pmu = {
 	.id		= ARM_PERF_PMU_ID_XSCALE1,
 	.name		= "xscale1",
@@ -436,9 +442,7 @@  static struct arm_pmu xscale1pmu = {
 	.get_event_idx	= xscale1pmu_get_event_idx,
 	.start		= xscale1pmu_start,
 	.stop		= xscale1pmu_stop,
-	.cache_map	= &xscale_perf_cache_map,
-	.event_map	= &xscale_perf_map,
-	.raw_event_mask	= 0xFF,
+	.map_event	= xscale_map_event,
 	.num_events	= 3,
 	.max_period	= (1LLU << 32) - 1,
 };
@@ -799,9 +803,7 @@  static struct arm_pmu xscale2pmu = {
 	.get_event_idx	= xscale2pmu_get_event_idx,
 	.start		= xscale2pmu_start,
 	.stop		= xscale2pmu_stop,
-	.cache_map	= &xscale_perf_cache_map,
-	.event_map	= &xscale_perf_map,
-	.raw_event_mask	= 0xFF,
+	.map_event	= xscale_map_event,
 	.num_events	= 5,
 	.max_period	= (1LLU << 32) - 1,
 };
-- 
1.7.0.4


From 37247db4c2f399dace9ec32d73e7568f1965eac8 Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Thu, 28 Apr 2011 16:27:54 +0100
Subject: [RFC PATCH 12/15] ARM: perf: add support for multiple PMUs

Currently, a single static instance of struct pmu is used when
registering an ARM PMU with the main perf subsystem. This limits
the ARM perf code to supporting a single PMU.

This patch replaces the static struct pmu instance with a member
variable on struct arm_pmu. This provides bidirectional mapping
between the two structs, and therefore allows for support of multiple
PMUs. The function 'to_arm_pmu' is provided for convenience.

PMU-generic functions are also updated to use the new mapping, and
PMU-generic initialisation of the member variables is moved into a new
function: armpmu_init.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/kernel/perf_event.c |   63 +++++++++++++++++++++++++++--------------
 1 files changed, 41 insertions(+), 22 deletions(-)

diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index b13bf23..7f31eff 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -59,6 +59,7 @@  struct cpu_hw_events {
 static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events);
 
 struct arm_pmu {
+	struct pmu	pmu;
 	enum arm_perf_pmu_ids id;
 	enum arm_pmu_type type;
 	cpumask_t	active_irqs;
@@ -84,6 +85,8 @@  struct arm_pmu {
 	struct cpu_hw_events	*(*get_hw_events)(void);
 };
 
+#define to_arm_pmu(p) (container_of(p, struct arm_pmu, pmu))
+
 /* Set at runtime when we know what CPU type we are. */
 static struct arm_pmu *armpmu;
 
@@ -193,6 +196,7 @@  armpmu_event_set_period(struct perf_event *event,
 			struct hw_perf_event *hwc,
 			int idx)
 {
+	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
 	s64 left = local64_read(&hwc->period_left);
 	s64 period = hwc->sample_period;
 	int ret = 0;
@@ -228,6 +232,7 @@  armpmu_event_update(struct perf_event *event,
 		    struct hw_perf_event *hwc,
 		    int idx, int overflow)
 {
+	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
 	u64 delta, prev_raw_count, new_raw_count;
 
 again:
@@ -267,6 +272,7 @@  armpmu_read(struct perf_event *event)
 static void
 armpmu_stop(struct perf_event *event, int flags)
 {
+	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
 	struct hw_perf_event *hwc = &event->hw;
 
 	/*
@@ -284,6 +290,7 @@  armpmu_stop(struct perf_event *event, int flags)
 static void
 armpmu_start(struct perf_event *event, int flags)
 {
+	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
 	struct hw_perf_event *hwc = &event->hw;
 
 	/*
@@ -308,6 +315,7 @@  armpmu_start(struct perf_event *event, int flags)
 static void
 armpmu_del(struct perf_event *event, int flags)
 {
+	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
 	struct cpu_hw_events *cpuc = armpmu->get_hw_events();
 	struct hw_perf_event *hwc = &event->hw;
 	int idx = hwc->idx;
@@ -324,6 +332,7 @@  armpmu_del(struct perf_event *event, int flags)
 static int
 armpmu_add(struct perf_event *event, int flags)
 {
+	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
 	struct cpu_hw_events *cpuc = armpmu->get_hw_events();
 	struct hw_perf_event *hwc = &event->hw;
 	int idx;
@@ -358,12 +367,11 @@  out:
 	return err;
 }
 
-static struct pmu pmu;
-
 static int
 validate_event(struct cpu_hw_events *cpuc,
 	       struct perf_event *event)
 {
+	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
 	struct hw_perf_event fake_event = event->hw;
 	struct pmu *leader_pmu = event->group_leader->pmu;
 
@@ -397,6 +405,7 @@  validate_group(struct perf_event *event)
 
 static irqreturn_t armpmu_platform_irq(int irq, void *dev)
 {
+	struct arm_pmu *armpmu = (struct arm_pmu *) dev;
 	struct platform_device *plat_device = armpmu->plat_device;
 	struct arm_pmu_platdata *plat = dev_get_platdata(&plat_device->dev);
 
@@ -404,7 +413,7 @@  static irqreturn_t armpmu_platform_irq(int irq, void *dev)
 }
 
 static void
-armpmu_release_hardware(void)
+armpmu_release_hardware(struct arm_pmu *armpmu)
 {
 	int i, irq, irqs;
 	struct platform_device *pmu_device = armpmu->plat_device;
@@ -416,14 +425,14 @@  armpmu_release_hardware(void)
 			continue;
 		irq = platform_get_irq(pmu_device, i);
 		if (irq >= 0)
-			free_irq(irq, NULL);
+			free_irq(irq, armpmu);
 	}
 
 	release_pmu(armpmu->type);
 }
 
 static int
-armpmu_reserve_hardware(void)
+armpmu_reserve_hardware(struct arm_pmu *armpmu)
 {
 	struct arm_pmu_platdata *plat;
 	irq_handler_t handle_irq;
@@ -467,11 +476,11 @@  armpmu_reserve_hardware(void)
 
 		err = request_irq(irq, handle_irq,
 				  IRQF_DISABLED | IRQF_NOBALANCING,
-				  "arm-pmu", NULL);
+				  "arm-pmu", armpmu);
 		if (err) {
 			pr_err("unable to request IRQ%d for ARM PMU counters\n",
 				irq);
-			armpmu_release_hardware();
+			armpmu_release_hardware(armpmu);
 			return err;
 		}
 
@@ -484,11 +493,12 @@  armpmu_reserve_hardware(void)
 static void
 hw_perf_event_destroy(struct perf_event *event)
 {
+	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
 	atomic_t *active_events	 = &armpmu->active_events;
 	struct mutex *pmu_reserve_mutex = &armpmu->reserve_mutex;
 
 	if (atomic_dec_and_mutex_lock(active_events, pmu_reserve_mutex)) {
-		armpmu_release_hardware();
+		armpmu_release_hardware(armpmu);
 		mutex_unlock(pmu_reserve_mutex);
 	}
 }
@@ -503,6 +513,7 @@  event_requires_mode_exclusion(struct perf_event_attr *attr)
 static int
 __hw_perf_event_init(struct perf_event *event)
 {
+	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
 	struct hw_perf_event *hwc = &event->hw;
 	int mapping, err;
 
@@ -559,6 +570,7 @@  __hw_perf_event_init(struct perf_event *event)
 
 static int armpmu_event_init(struct perf_event *event)
 {
+	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
 	int err = 0;
 	atomic_t *active_events = &armpmu->active_events;
 
@@ -570,7 +582,7 @@  static int armpmu_event_init(struct perf_event *event)
 	if (!atomic_inc_not_zero(active_events)) {
 		mutex_lock(&armpmu->reserve_mutex);
 		if (atomic_read(active_events) == 0)
-			err = armpmu_reserve_hardware();
+			err = armpmu_reserve_hardware(armpmu);
 
 		if (!err)
 			atomic_inc(active_events);
@@ -589,6 +601,7 @@  static int armpmu_event_init(struct perf_event *event)
 
 static void armpmu_enable(struct pmu *pmu)
 {
+	struct arm_pmu *armpmu = to_arm_pmu(pmu);
 	/* Enable all of the perf events on hardware. */
 	int idx, enabled = 0;
 	struct cpu_hw_events *cpuc = armpmu->get_hw_events();
@@ -609,24 +622,31 @@  static void armpmu_enable(struct pmu *pmu)
 
 static void armpmu_disable(struct pmu *pmu)
 {
+	struct arm_pmu *armpmu = to_arm_pmu(pmu);
 	armpmu->stop();
 }
 
-static struct pmu pmu = {
-	.pmu_enable	= armpmu_enable,
-	.pmu_disable	= armpmu_disable,
-	.event_init	= armpmu_event_init,
-	.add		= armpmu_add,
-	.del		= armpmu_del,
-	.start		= armpmu_start,
-	.stop		= armpmu_stop,
-	.read		= armpmu_read,
-};
-
 static void __init armpmu_init(struct arm_pmu *armpmu)
 {
 	atomic_set(&armpmu->active_events, 0);
 	mutex_init(&armpmu->reserve_mutex);
+
+	armpmu->pmu = (struct pmu) {
+		.pmu_enable	= armpmu_enable,
+		.pmu_disable	= armpmu_disable,
+		.event_init	= armpmu_event_init,
+		.add		= armpmu_add,
+		.del		= armpmu_del,
+		.start		= armpmu_start,
+		.stop		= armpmu_stop,
+		.read		= armpmu_read,
+	};
+}
+
+static int __init armpmu_register(struct arm_pmu *armpmu, char *name, int type)
+{
+	armpmu_init(armpmu);
+	return perf_pmu_register(&armpmu->pmu, name, type);
 }
 
 /* Include the PMU-specific implementations. */
@@ -751,8 +771,7 @@  init_hw_perf_events(void)
 		pr_info("enabled with %s PMU driver, %d counters available\n",
 			armpmu->name, armpmu->num_events);
 		cpu_pmu_init(armpmu);
-		armpmu_init(armpmu);
-		perf_pmu_register(&pmu, "cpu", PERF_TYPE_RAW);
+		armpmu_register(armpmu, "cpu", PERF_TYPE_RAW);
 	} else {
 		pr_info("no hardware support available\n");
 	}
-- 
1.7.0.4


From 958e3b73e917a5ff5e37200b50b1619e84d8ee04 Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Fri, 24 Jun 2011 11:30:59 +0100
Subject: [RFC PATCH 13/15] ARM: perf: remove event limit from pmu_hw_events

Currently the event accounting data in pmu_hw_events is stored in
fixed-sized arrays within the structure.

This patch refactors the accounting data to allow any number of events
to be managed.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/kernel/perf_event.c |    9 +++++++--
 1 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index 7f31eff..e1db555 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -42,13 +42,13 @@  struct cpu_hw_events {
 	/*
 	 * The events that are active on the CPU for the given index.
 	 */
-	struct perf_event	*events[ARMPMU_MAX_HWEVENTS];
+	struct perf_event	**events;
 
 	/*
 	 * A 1 bit for an index indicates that the counter is being used for
 	 * an event. A 0 means that the counter can be used.
 	 */
-	unsigned long		used_mask[BITS_TO_LONGS(ARMPMU_MAX_HWEVENTS)];
+	unsigned long           *used_mask;
 
 	/*
 	 * Hardware lock to serialize accesses to PMU registers. Needed for the
@@ -56,6 +56,9 @@  struct cpu_hw_events {
 	 */
 	raw_spinlock_t		pmu_lock;
 };
+
+static DEFINE_PER_CPU(struct perf_event * [ARMPMU_MAX_HWEVENTS], hw_events);
+static DEFINE_PER_CPU(unsigned long [BITS_TO_LONGS(ARMPMU_MAX_HWEVENTS)], used_mask);
 static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events);
 
 struct arm_pmu {
@@ -714,6 +717,8 @@  static void __init cpu_pmu_init(struct arm_pmu *armpmu)
 	int cpu;
 	for_each_possible_cpu(cpu) {
 		struct cpu_hw_events *events = &per_cpu(cpu_hw_events, cpu);
+		events->events = per_cpu(hw_events, cpu);
+		events->used_mask = per_cpu(used_mask, cpu);
 		raw_spin_lock_init(&events->pmu_lock);
 	}
 	armpmu->get_hw_events = armpmu_get_cpu_events;
-- 
1.7.0.4


From c8be2941098c5ad25623ab8d5f9e633384f2cf54 Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Tue, 17 May 2011 11:20:11 +0100
Subject: [RFC PATCH 14/15] ARM: perf: remove cpu-related misnomers

Currently struct cpu_hw_events stores data on events running on a
PMU associated with a CPU. As this data is general enough to be used
for system PMUs, this name is a misnomer, and may cause confusion when
it is used for system PMUs.

Additionally, 'armpmu' is commonly used as a parameter name for an
instance of struct arm_pmu. The name is also used for a global instance
which represents the CPU's PMU.

As cpu_hw_events is now not tied to CPU PMUs, it is renamed to
pmu_hw_events, with instances of it renamed similarly. As the global
'armpmu' is CPU-specfic, it is renamed to cpu_pmu. This should make it
clearer which code is generic, and which is coupled with the CPU.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/kernel/perf_event.c        |   84 +++++++++++++++++-----------------
 arch/arm/kernel/perf_event_v6.c     |   18 ++++----
 arch/arm/kernel/perf_event_v7.c     |   22 +++++-----
 arch/arm/kernel/perf_event_xscale.c |   32 +++++++-------
 4 files changed, 78 insertions(+), 78 deletions(-)

diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index e1db555..8315133 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -37,10 +37,10 @@ 
  */
 #define ARMPMU_MAX_HWEVENTS		32
 
-/* The events for a given CPU. */
-struct cpu_hw_events {
+/* The events for a given PMU register set. */
+struct pmu_hw_events {
 	/*
-	 * The events that are active on the CPU for the given index.
+	 * The events that are active on the PMU for the given index.
 	 */
 	struct perf_event	**events;
 
@@ -59,7 +59,7 @@  struct cpu_hw_events {
 
 static DEFINE_PER_CPU(struct perf_event * [ARMPMU_MAX_HWEVENTS], hw_events);
 static DEFINE_PER_CPU(unsigned long [BITS_TO_LONGS(ARMPMU_MAX_HWEVENTS)], used_mask);
-static DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events);
+static DEFINE_PER_CPU(struct pmu_hw_events, cpu_hw_events);
 
 struct arm_pmu {
 	struct pmu	pmu;
@@ -70,7 +70,7 @@  struct arm_pmu {
 	irqreturn_t	(*handle_irq)(int irq_num, void *dev);
 	void		(*enable)(struct hw_perf_event *evt, int idx);
 	void		(*disable)(struct hw_perf_event *evt, int idx);
-	int		(*get_event_idx)(struct cpu_hw_events *cpuc,
+	int		(*get_event_idx)(struct pmu_hw_events *hw_events,
 					 struct hw_perf_event *hwc);
 	int		(*set_event_filter)(struct hw_perf_event *evt,
 					    struct perf_event_attr *attr);
@@ -85,21 +85,21 @@  struct arm_pmu {
 	struct mutex	reserve_mutex;
 	u64		max_period;
 	struct platform_device	*plat_device;
-	struct cpu_hw_events	*(*get_hw_events)(void);
+	struct pmu_hw_events	*(*get_hw_events)(void);
 };
 
 #define to_arm_pmu(p) (container_of(p, struct arm_pmu, pmu))
 
 /* Set at runtime when we know what CPU type we are. */
-static struct arm_pmu *armpmu;
+static struct arm_pmu *cpu_pmu;
 
 enum arm_perf_pmu_ids
 armpmu_get_pmu_id(void)
 {
 	int id = -ENODEV;
 
-	if (armpmu != NULL)
-		id = armpmu->id;
+	if (cpu_pmu != NULL)
+		id = cpu_pmu->id;
 
 	return id;
 }
@@ -110,8 +110,8 @@  armpmu_get_max_events(void)
 {
 	int max_events = 0;
 
-	if (armpmu != NULL)
-		max_events = armpmu->num_events;
+	if (cpu_pmu != NULL)
+		max_events = cpu_pmu->num_events;
 
 	return max_events;
 }
@@ -319,15 +319,15 @@  static void
 armpmu_del(struct perf_event *event, int flags)
 {
 	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
-	struct cpu_hw_events *cpuc = armpmu->get_hw_events();
+	struct pmu_hw_events *hw_events = armpmu->get_hw_events();
 	struct hw_perf_event *hwc = &event->hw;
 	int idx = hwc->idx;
 
 	WARN_ON(idx < 0);
 
 	armpmu_stop(event, PERF_EF_UPDATE);
-	cpuc->events[idx] = NULL;
-	clear_bit(idx, cpuc->used_mask);
+	hw_events->events[idx] = NULL;
+	clear_bit(idx, hw_events->used_mask);
 
 	perf_event_update_userpage(event);
 }
@@ -336,7 +336,7 @@  static int
 armpmu_add(struct perf_event *event, int flags)
 {
 	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
-	struct cpu_hw_events *cpuc = armpmu->get_hw_events();
+	struct pmu_hw_events *hw_events = armpmu->get_hw_events();
 	struct hw_perf_event *hwc = &event->hw;
 	int idx;
 	int err = 0;
@@ -344,7 +344,7 @@  armpmu_add(struct perf_event *event, int flags)
 	perf_pmu_disable(event->pmu);
 
 	/* If we don't have a space for the counter then finish early. */
-	idx = armpmu->get_event_idx(cpuc, hwc);
+	idx = armpmu->get_event_idx(hw_events, hwc);
 	if (idx < 0) {
 		err = idx;
 		goto out;
@@ -356,7 +356,7 @@  armpmu_add(struct perf_event *event, int flags)
 	 */
 	event->hw.idx = idx;
 	armpmu->disable(hwc, idx);
-	cpuc->events[idx] = event;
+	hw_events->events[idx] = event;
 
 	hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE;
 	if (flags & PERF_EF_START)
@@ -371,7 +371,7 @@  out:
 }
 
 static int
-validate_event(struct cpu_hw_events *cpuc,
+validate_event(struct pmu_hw_events *hw_events,
 	       struct perf_event *event)
 {
 	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
@@ -381,14 +381,14 @@  validate_event(struct cpu_hw_events *cpuc,
 	if (event->pmu != leader_pmu || event->state <= PERF_EVENT_STATE_OFF)
 		return 1;
 
-	return armpmu->get_event_idx(cpuc, &fake_event) >= 0;
+	return armpmu->get_event_idx(hw_events, &fake_event) >= 0;
 }
 
 static int
 validate_group(struct perf_event *event)
 {
 	struct perf_event *sibling, *leader = event->group_leader;
-	struct cpu_hw_events fake_pmu;
+	struct pmu_hw_events fake_pmu;
 
 	memset(&fake_pmu, 0, sizeof(fake_pmu));
 
@@ -604,13 +604,13 @@  static int armpmu_event_init(struct perf_event *event)
 
 static void armpmu_enable(struct pmu *pmu)
 {
-	struct arm_pmu *armpmu = to_arm_pmu(pmu);
 	/* Enable all of the perf events on hardware. */
+	struct arm_pmu *armpmu = to_arm_pmu(pmu);
 	int idx, enabled = 0;
-	struct cpu_hw_events *cpuc = armpmu->get_hw_events();
+	struct pmu_hw_events *hw_events = armpmu->get_hw_events();
 
 	for (idx = 0; idx < armpmu->num_events; ++idx) {
-		struct perf_event *event = cpuc->events[idx];
+		struct perf_event *event = hw_events->events[idx];
 
 		if (!event)
 			continue;
@@ -662,13 +662,13 @@  static int __init armpmu_register(struct arm_pmu *armpmu, char *name, int type)
  * This requires SMP to be available, so exists as a separate initcall.
  */
 static int __init
-armpmu_reset(void)
+cpu_pmu_reset(void)
 {
-	if (armpmu && armpmu->reset)
-		return on_each_cpu(armpmu->reset, NULL, 1);
+	if (cpu_pmu && cpu_pmu->reset)
+		return on_each_cpu(cpu_pmu->reset, NULL, 1);
 	return 0;
 }
-arch_initcall(armpmu_reset);
+arch_initcall(cpu_pmu_reset);
 
 /*
  * PMU platform driver and devicetree bindings.
@@ -688,7 +688,7 @@  static struct platform_device_id armpmu_plat_device_ids[] = {
 
 static int __devinit armpmu_device_probe(struct platform_device *pdev)
 {
-	armpmu->plat_device = pdev;
+	cpu_pmu->plat_device = pdev;
 	return 0;
 }
 
@@ -707,7 +707,7 @@  static int __init register_pmu_driver(void)
 }
 device_initcall(register_pmu_driver);
 
-static struct cpu_hw_events *armpmu_get_cpu_events(void)
+static struct pmu_hw_events *armpmu_get_cpu_events(void)
 {
 	return &__get_cpu_var(cpu_hw_events);
 }
@@ -716,7 +716,7 @@  static void __init cpu_pmu_init(struct arm_pmu *armpmu)
 {
 	int cpu;
 	for_each_possible_cpu(cpu) {
-		struct cpu_hw_events *events = &per_cpu(cpu_hw_events, cpu);
+		struct pmu_hw_events *events = &per_cpu(cpu_hw_events, cpu);
 		events->events = per_cpu(hw_events, cpu);
 		events->used_mask = per_cpu(used_mask, cpu);
 		raw_spin_lock_init(&events->pmu_lock);
@@ -741,22 +741,22 @@  init_hw_perf_events(void)
 		case 0xB360:	/* ARM1136 */
 		case 0xB560:	/* ARM1156 */
 		case 0xB760:	/* ARM1176 */
-			armpmu = armv6pmu_init();
+			cpu_pmu = armv6pmu_init();
 			break;
 		case 0xB020:	/* ARM11mpcore */
-			armpmu = armv6mpcore_pmu_init();
+			cpu_pmu = armv6mpcore_pmu_init();
 			break;
 		case 0xC080:	/* Cortex-A8 */
-			armpmu = armv7_a8_pmu_init();
+			cpu_pmu = armv7_a8_pmu_init();
 			break;
 		case 0xC090:	/* Cortex-A9 */
-			armpmu = armv7_a9_pmu_init();
+			cpu_pmu = armv7_a9_pmu_init();
 			break;
 		case 0xC050:	/* Cortex-A5 */
-			armpmu = armv7_a5_pmu_init();
+			cpu_pmu = armv7_a5_pmu_init();
 			break;
 		case 0xC0F0:	/* Cortex-A15 */
-			armpmu = armv7_a15_pmu_init();
+			cpu_pmu = armv7_a15_pmu_init();
 			break;
 		}
 	/* Intel CPUs [xscale]. */
@@ -764,19 +764,19 @@  init_hw_perf_events(void)
 		part_number = (cpuid >> 13) & 0x7;
 		switch (part_number) {
 		case 1:
-			armpmu = xscale1pmu_init();
+			cpu_pmu = xscale1pmu_init();
 			break;
 		case 2:
-			armpmu = xscale2pmu_init();
+			cpu_pmu = xscale2pmu_init();
 			break;
 		}
 	}
 
-	if (armpmu) {
+	if (cpu_pmu) {
 		pr_info("enabled with %s PMU driver, %d counters available\n",
-			armpmu->name, armpmu->num_events);
-		cpu_pmu_init(armpmu);
-		armpmu_register(armpmu, "cpu", PERF_TYPE_RAW);
+			cpu_pmu->name, cpu_pmu->num_events);
+		cpu_pmu_init(cpu_pmu);
+		armpmu_register(cpu_pmu, "cpu", PERF_TYPE_RAW);
 	} else {
 		pr_info("no hardware support available\n");
 	}
diff --git a/arch/arm/kernel/perf_event_v6.c b/arch/arm/kernel/perf_event_v6.c
index a4c5aa9..e63d811 100644
--- a/arch/arm/kernel/perf_event_v6.c
+++ b/arch/arm/kernel/perf_event_v6.c
@@ -433,7 +433,7 @@  armv6pmu_enable_event(struct hw_perf_event *hwc,
 		      int idx)
 {
 	unsigned long val, mask, evt, flags;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	if (ARMV6_CYCLE_COUNTER == idx) {
 		mask	= 0;
@@ -486,7 +486,7 @@  armv6pmu_handle_irq(int irq_num,
 {
 	unsigned long pmcr = armv6_pmcr_read();
 	struct perf_sample_data data;
-	struct cpu_hw_events *cpuc;
+	struct pmu_hw_events *cpuc;
 	struct pt_regs *regs;
 	int idx;
 
@@ -505,7 +505,7 @@  armv6pmu_handle_irq(int irq_num,
 	perf_sample_data_init(&data, 0);
 
 	cpuc = &__get_cpu_var(cpu_hw_events);
-	for (idx = 0; idx < armpmu->num_events; ++idx) {
+	for (idx = 0; idx < cpu_pmu->num_events; ++idx) {
 		struct perf_event *event = cpuc->events[idx];
 		struct hw_perf_event *hwc;
 
@@ -526,7 +526,7 @@  armv6pmu_handle_irq(int irq_num,
 			continue;
 
 		if (perf_event_overflow(event, &data, regs))
-			armpmu->disable(hwc, idx);
+			cpu_pmu->disable(hwc, idx);
 	}
 
 	/*
@@ -545,7 +545,7 @@  static void
 armv6pmu_start(void)
 {
 	unsigned long flags, val;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = armv6_pmcr_read();
@@ -558,7 +558,7 @@  static void
 armv6pmu_stop(void)
 {
 	unsigned long flags, val;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = armv6_pmcr_read();
@@ -568,7 +568,7 @@  armv6pmu_stop(void)
 }
 
 static int
-armv6pmu_get_event_idx(struct cpu_hw_events *cpuc,
+armv6pmu_get_event_idx(struct pmu_hw_events *cpuc,
 		       struct hw_perf_event *event)
 {
 	/* Always place a cycle counter into the cycle counter. */
@@ -598,7 +598,7 @@  armv6pmu_disable_event(struct hw_perf_event *hwc,
 		       int idx)
 {
 	unsigned long val, mask, evt, flags;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	if (ARMV6_CYCLE_COUNTER == idx) {
 		mask	= ARMV6_PMCR_CCOUNT_IEN;
@@ -632,7 +632,7 @@  armv6mpcore_pmu_disable_event(struct hw_perf_event *hwc,
 			      int idx)
 {
 	unsigned long val, mask, flags, evt = 0;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	if (ARMV6_CYCLE_COUNTER == idx) {
 		mask	= ARMV6_PMCR_CCOUNT_IEN;
diff --git a/arch/arm/kernel/perf_event_v7.c b/arch/arm/kernel/perf_event_v7.c
index be7b58a..98b7573 100644
--- a/arch/arm/kernel/perf_event_v7.c
+++ b/arch/arm/kernel/perf_event_v7.c
@@ -683,7 +683,7 @@  static const unsigned armv7_a15_perf_cache_map[PERF_COUNT_HW_CACHE_MAX]
  */
 #define	ARMV7_IDX_CYCLE_COUNTER	0
 #define	ARMV7_IDX_COUNTER0	1
-#define	ARMV7_IDX_COUNTER_LAST	(ARMV7_IDX_CYCLE_COUNTER + armpmu->num_events - 1)
+#define	ARMV7_IDX_COUNTER_LAST	(ARMV7_IDX_CYCLE_COUNTER + cpu_pmu->num_events - 1)
 
 #define	ARMV7_MAX_COUNTERS	32
 #define	ARMV7_COUNTER_MASK	(ARMV7_MAX_COUNTERS - 1)
@@ -936,7 +936,7 @@  static void armv7_pmnc_dump_regs(void)
 static void armv7pmu_enable_event(struct hw_perf_event *hwc, int idx)
 {
 	unsigned long flags;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	/*
 	 * Enable counter and interrupt, and set the counter to count
@@ -973,7 +973,7 @@  static void armv7pmu_enable_event(struct hw_perf_event *hwc, int idx)
 static void armv7pmu_disable_event(struct hw_perf_event *hwc, int idx)
 {
 	unsigned long flags;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	/*
 	 * Disable counter and interrupt
@@ -997,7 +997,7 @@  static irqreturn_t armv7pmu_handle_irq(int irq_num, void *dev)
 {
 	u32 pmnc;
 	struct perf_sample_data data;
-	struct cpu_hw_events *cpuc;
+	struct pmu_hw_events *cpuc;
 	struct pt_regs *regs;
 	int idx;
 
@@ -1020,7 +1020,7 @@  static irqreturn_t armv7pmu_handle_irq(int irq_num, void *dev)
 	perf_sample_data_init(&data, 0);
 
 	cpuc = &__get_cpu_var(cpu_hw_events);
-	for (idx = 0; idx < armpmu->num_events; ++idx) {
+	for (idx = 0; idx < cpu_pmu->num_events; ++idx) {
 		struct perf_event *event = cpuc->events[idx];
 		struct hw_perf_event *hwc;
 
@@ -1038,7 +1038,7 @@  static irqreturn_t armv7pmu_handle_irq(int irq_num, void *dev)
 			continue;
 
 		if (perf_event_overflow(event, &data, regs))
-			armpmu->disable(hwc, idx);
+			cpu_pmu->disable(hwc, idx);
 	}
 
 	/*
@@ -1056,7 +1056,7 @@  static irqreturn_t armv7pmu_handle_irq(int irq_num, void *dev)
 static void armv7pmu_start(void)
 {
 	unsigned long flags;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	/* Enable all counters */
@@ -1067,7 +1067,7 @@  static void armv7pmu_start(void)
 static void armv7pmu_stop(void)
 {
 	unsigned long flags;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	/* Disable all counters */
@@ -1075,7 +1075,7 @@  static void armv7pmu_stop(void)
 	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
-static int armv7pmu_get_event_idx(struct cpu_hw_events *cpuc,
+static int armv7pmu_get_event_idx(struct pmu_hw_events *cpuc,
 				  struct hw_perf_event *event)
 {
 	int idx;
@@ -1093,7 +1093,7 @@  static int armv7pmu_get_event_idx(struct cpu_hw_events *cpuc,
 	 * For anything other than a cycle counter, try and use
 	 * the events counters
 	 */
-	for (idx = ARMV7_IDX_COUNTER0; idx < armpmu->num_events; ++idx) {
+	for (idx = ARMV7_IDX_COUNTER0; idx < cpu_pmu->num_events; ++idx) {
 		if (!test_and_set_bit(idx, cpuc->used_mask))
 			return idx;
 	}
@@ -1130,7 +1130,7 @@  static int armv7pmu_set_event_filter(struct hw_perf_event *event,
 
 static void armv7pmu_reset(void *info)
 {
-	u32 idx, nb_cnt = armpmu->num_events;
+	u32 idx, nb_cnt = cpu_pmu->num_events;
 
 	/* The counter and interrupt enable registers are unknown at reset. */
 	for (idx = ARMV7_IDX_CYCLE_COUNTER; idx < nb_cnt; ++idx)
diff --git a/arch/arm/kernel/perf_event_xscale.c b/arch/arm/kernel/perf_event_xscale.c
index d4c7610..e0cca10 100644
--- a/arch/arm/kernel/perf_event_xscale.c
+++ b/arch/arm/kernel/perf_event_xscale.c
@@ -222,7 +222,7 @@  xscale1pmu_handle_irq(int irq_num, void *dev)
 {
 	unsigned long pmnc;
 	struct perf_sample_data data;
-	struct cpu_hw_events *cpuc;
+	struct pmu_hw_events *cpuc;
 	struct pt_regs *regs;
 	int idx;
 
@@ -249,7 +249,7 @@  xscale1pmu_handle_irq(int irq_num, void *dev)
 	perf_sample_data_init(&data, 0);
 
 	cpuc = &__get_cpu_var(cpu_hw_events);
-	for (idx = 0; idx < armpmu->num_events; ++idx) {
+	for (idx = 0; idx < cpu_pmu->num_events; ++idx) {
 		struct perf_event *event = cpuc->events[idx];
 		struct hw_perf_event *hwc;
 
@@ -263,7 +263,7 @@  xscale1pmu_handle_irq(int irq_num, void *dev)
 			continue;
 
 		if (perf_event_overflow(event, &data, regs))
-			armpmu->disable(hwc, idx);
+			cpu_pmu->disable(hwc, idx);
 	}
 
 	irq_work_run();
@@ -281,7 +281,7 @@  static void
 xscale1pmu_enable_event(struct hw_perf_event *hwc, int idx)
 {
 	unsigned long val, mask, evt, flags;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	switch (idx) {
 	case XSCALE_CYCLE_COUNTER:
@@ -315,7 +315,7 @@  static void
 xscale1pmu_disable_event(struct hw_perf_event *hwc, int idx)
 {
 	unsigned long val, mask, evt, flags;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	switch (idx) {
 	case XSCALE_CYCLE_COUNTER:
@@ -344,7 +344,7 @@  xscale1pmu_disable_event(struct hw_perf_event *hwc, int idx)
 }
 
 static int
-xscale1pmu_get_event_idx(struct cpu_hw_events *cpuc,
+xscale1pmu_get_event_idx(struct pmu_hw_events *cpuc,
 			struct hw_perf_event *event)
 {
 	if (XSCALE_PERFCTR_CCNT == event->config_base) {
@@ -367,7 +367,7 @@  static void
 xscale1pmu_start(void)
 {
 	unsigned long flags, val;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = xscale1pmu_read_pmnc();
@@ -380,7 +380,7 @@  static void
 xscale1pmu_stop(void)
 {
 	unsigned long flags, val;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = xscale1pmu_read_pmnc();
@@ -565,7 +565,7 @@  xscale2pmu_handle_irq(int irq_num, void *dev)
 {
 	unsigned long pmnc, of_flags;
 	struct perf_sample_data data;
-	struct cpu_hw_events *cpuc;
+	struct pmu_hw_events *cpuc;
 	struct pt_regs *regs;
 	int idx;
 
@@ -586,7 +586,7 @@  xscale2pmu_handle_irq(int irq_num, void *dev)
 	perf_sample_data_init(&data, 0);
 
 	cpuc = &__get_cpu_var(cpu_hw_events);
-	for (idx = 0; idx < armpmu->num_events; ++idx) {
+	for (idx = 0; idx < cpu_pmu->num_events; ++idx) {
 		struct perf_event *event = cpuc->events[idx];
 		struct hw_perf_event *hwc;
 
@@ -600,7 +600,7 @@  xscale2pmu_handle_irq(int irq_num, void *dev)
 			continue;
 
 		if (perf_event_overflow(event, &data, regs))
-			armpmu->disable(hwc, idx);
+			cpu_pmu->disable(hwc, idx);
 	}
 
 	irq_work_run();
@@ -618,7 +618,7 @@  static void
 xscale2pmu_enable_event(struct hw_perf_event *hwc, int idx)
 {
 	unsigned long flags, ien, evtsel;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	ien = xscale2pmu_read_int_enable();
 	evtsel = xscale2pmu_read_event_select();
@@ -662,7 +662,7 @@  static void
 xscale2pmu_disable_event(struct hw_perf_event *hwc, int idx)
 {
 	unsigned long flags, ien, evtsel;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	ien = xscale2pmu_read_int_enable();
 	evtsel = xscale2pmu_read_event_select();
@@ -703,7 +703,7 @@  xscale2pmu_disable_event(struct hw_perf_event *hwc, int idx)
 }
 
 static int
-xscale2pmu_get_event_idx(struct cpu_hw_events *cpuc,
+xscale2pmu_get_event_idx(struct pmu_hw_events *cpuc,
 			struct hw_perf_event *event)
 {
 	int idx = xscale1pmu_get_event_idx(cpuc, event);
@@ -722,7 +722,7 @@  static void
 xscale2pmu_start(void)
 {
 	unsigned long flags, val;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = xscale2pmu_read_pmnc() & ~XSCALE_PMU_CNT64;
@@ -735,7 +735,7 @@  static void
 xscale2pmu_stop(void)
 {
 	unsigned long flags, val;
-	struct cpu_hw_events *events = armpmu->get_hw_events();
+	struct pmu_hw_events *events = cpu_pmu->get_hw_events();
 
 	raw_spin_lock_irqsave(&events->pmu_lock, flags);
 	val = xscale2pmu_read_pmnc();
-- 
1.7.0.4


From 5403a10564f6c6c212ae1d3844a6b87175c65388 Mon Sep 17 00:00:00 2001
From: Mark Rutland <[e-mail]>
Date: Thu, 19 May 2011 10:07:57 +0100
Subject: [RFC PATCH 15/15] ARM: perf: move arm_pmu into <asm/pmu.h>

Currently, struct arm_pmu and related functions are only visible to
{,arch/arm/}/kernel/perf_event.c. This prevents new drivers from using
the framework.

This patch  moves declarations to asm/pmu.h, allowing new PMU drivers
to use the framework.

Signed-off-by: Mark Rutland <[e-mail]>
Reviewed-by: Will Deacon <[e-mail]>
---
 arch/arm/include/asm/pmu.h   |   64 ++++++++++++++++++++++++++++++++++++++++++
 arch/arm/kernel/perf_event.c |   53 ++--------------------------------
 2 files changed, 67 insertions(+), 50 deletions(-)

diff --git a/arch/arm/include/asm/pmu.h b/arch/arm/include/asm/pmu.h
index f210708..9c2c877 100644
--- a/arch/arm/include/asm/pmu.h
+++ b/arch/arm/include/asm/pmu.h
@@ -13,6 +13,7 @@ 
 #define __ARM_PMU_H__
 
 #include <linux/interrupt.h>
+#include <linux/perf_event.h>
 
 /*
  * Types of PMUs that can be accessed directly and require mutual
@@ -69,4 +70,67 @@  release_pmu(struct platform_device *pdev)	{ }
 
 #endif /* CONFIG_CPU_HAS_PMU */
 
+#ifdef CONFIG_HW_PERF_EVENTS
+
+/* The events for a given PMU register set. */
+struct pmu_hw_events {
+	/*
+	 * The events that are active on the PMU for the given index.
+	 */
+	struct perf_event	**events;
+
+	/*
+	 * A 1 bit for an index indicates that the counter is being used for
+	 * an event. A 0 means that the counter can be used.
+	 */
+	unsigned long           *used_mask;
+
+	/*
+	 * Hardware lock to serialize accesses to PMU registers. Needed for the
+	 * read/modify/write sequences.
+	 */
+	raw_spinlock_t		pmu_lock;
+};
+
+struct arm_pmu {
+	struct pmu	pmu;
+	enum arm_perf_pmu_ids id;
+	enum arm_pmu_type type;
+	cpumask_t	active_irqs;
+	const char	*name;
+	irqreturn_t	(*handle_irq)(int irq_num, void *dev);
+	void		(*enable)(struct hw_perf_event *evt, int idx);
+	void		(*disable)(struct hw_perf_event *evt, int idx);
+	int		(*get_event_idx)(struct pmu_hw_events *hw_events,
+					 struct hw_perf_event *hwc);
+	int		(*set_event_filter)(struct hw_perf_event *evt,
+					    struct perf_event_attr *attr);
+	u32		(*read_counter)(int idx);
+	void		(*write_counter)(int idx, u32 val);
+	void		(*start)(void);
+	void		(*stop)(void);
+	void		(*reset)(void *);
+	int		(*map_event)(struct perf_event *event);
+	int		num_events;
+	atomic_t	active_events;
+	struct mutex	reserve_mutex;
+	u64		max_period;
+	struct platform_device	*plat_device;
+	struct pmu_hw_events	*(*get_hw_events)(void);
+};
+
+#define to_arm_pmu(p) (container_of(p, struct arm_pmu, pmu))
+
+int __init armpmu_register(struct arm_pmu *armpmu, char *name, int type);
+
+u64 armpmu_event_update(struct perf_event *event,
+			struct hw_perf_event *hwc,
+			int idx, int overflow);
+
+int armpmu_event_set_period(struct perf_event *event,
+			    struct hw_perf_event *hwc,
+			    int idx);
+
+#endif /* CONFIG_HW_PERF_EVENTS */
+
 #endif /* __ARM_PMU_H__ */
diff --git a/arch/arm/kernel/perf_event.c b/arch/arm/kernel/perf_event.c
index 8315133..aaa631b 100644
--- a/arch/arm/kernel/perf_event.c
+++ b/arch/arm/kernel/perf_event.c
@@ -37,57 +37,10 @@ 
  */
 #define ARMPMU_MAX_HWEVENTS		32
 
-/* The events for a given PMU register set. */
-struct pmu_hw_events {
-	/*
-	 * The events that are active on the PMU for the given index.
-	 */
-	struct perf_event	**events;
-
-	/*
-	 * A 1 bit for an index indicates that the counter is being used for
-	 * an event. A 0 means that the counter can be used.
-	 */
-	unsigned long           *used_mask;
-
-	/*
-	 * Hardware lock to serialize accesses to PMU registers. Needed for the
-	 * read/modify/write sequences.
-	 */
-	raw_spinlock_t		pmu_lock;
-};
-
 static DEFINE_PER_CPU(struct perf_event * [ARMPMU_MAX_HWEVENTS], hw_events);
 static DEFINE_PER_CPU(unsigned long [BITS_TO_LONGS(ARMPMU_MAX_HWEVENTS)], used_mask);
 static DEFINE_PER_CPU(struct pmu_hw_events, cpu_hw_events);
 
-struct arm_pmu {
-	struct pmu	pmu;
-	enum arm_perf_pmu_ids id;
-	enum arm_pmu_type type;
-	cpumask_t	active_irqs;
-	const char	*name;
-	irqreturn_t	(*handle_irq)(int irq_num, void *dev);
-	void		(*enable)(struct hw_perf_event *evt, int idx);
-	void		(*disable)(struct hw_perf_event *evt, int idx);
-	int		(*get_event_idx)(struct pmu_hw_events *hw_events,
-					 struct hw_perf_event *hwc);
-	int		(*set_event_filter)(struct hw_perf_event *evt,
-					    struct perf_event_attr *attr);
-	u32		(*read_counter)(int idx);
-	void		(*write_counter)(int idx, u32 val);
-	void		(*start)(void);
-	void		(*stop)(void);
-	void		(*reset)(void *);
-	int		(*map_event)(struct perf_event *event);
-	int		num_events;
-	atomic_t	active_events;
-	struct mutex	reserve_mutex;
-	u64		max_period;
-	struct platform_device	*plat_device;
-	struct pmu_hw_events	*(*get_hw_events)(void);
-};
-
 #define to_arm_pmu(p) (container_of(p, struct arm_pmu, pmu))
 
 /* Set at runtime when we know what CPU type we are. */
@@ -194,7 +147,7 @@  static int map_cpu_event(struct perf_event *event,
 	return -ENOENT;
 }
 
-static int
+int
 armpmu_event_set_period(struct perf_event *event,
 			struct hw_perf_event *hwc,
 			int idx)
@@ -230,7 +183,7 @@  armpmu_event_set_period(struct perf_event *event,
 	return ret;
 }
 
-static u64
+u64
 armpmu_event_update(struct perf_event *event,
 		    struct hw_perf_event *hwc,
 		    int idx, int overflow)
@@ -646,7 +599,7 @@  static void __init armpmu_init(struct arm_pmu *armpmu)
 	};
 }
 
-static int __init armpmu_register(struct arm_pmu *armpmu, char *name, int type)
+int __init armpmu_register(struct arm_pmu *armpmu, char *name, int type)
 {
 	armpmu_init(armpmu);
 	return perf_pmu_register(&armpmu->pmu, name, type);