linux/drivers/perf/arm_pmu.c

889 lines
21 KiB
C
Raw Normal View History

#undef DEBUG
/*
* ARM performance counter support.
*
* Copyright (C) 2009 picoChip Designs, Ltd., Jamie Iles
* Copyright (C) 2010 ARM Ltd., Will Deacon <will.deacon@arm.com>
*
* This code is based on the sparc64 perf event code, which is in turn based
* on the x86 code.
*/
#define pr_fmt(fmt) "hw perfevents: " fmt
#include <linux/bitmap.h>
#include <linux/cpumask.h>
#include <linux/cpu_pm.h>
#include <linux/export.h>
#include <linux/kernel.h>
#include <linux/perf/arm_pmu.h>
#include <linux/slab.h>
#include <linux/sched/clock.h>
#include <linux/spinlock.h>
#include <linux/irq.h>
#include <linux/irqdesc.h>
#include <asm/irq_regs.h>
static DEFINE_PER_CPU(struct arm_pmu *, cpu_armpmu);
static DEFINE_PER_CPU(int, cpu_irq);
static int
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;
cache_type = (config >> 0) & 0xff;
if (cache_type >= PERF_COUNT_HW_CACHE_MAX)
return -EINVAL;
cache_op = (config >> 8) & 0xff;
if (cache_op >= PERF_COUNT_HW_CACHE_OP_MAX)
return -EINVAL;
cache_result = (config >> 16) & 0xff;
if (cache_result >= PERF_COUNT_HW_CACHE_RESULT_MAX)
return -EINVAL;
if (!cache_map)
return -ENOENT;
ret = (int)(*cache_map)[cache_type][cache_op][cache_result];
if (ret == CACHE_OP_UNSUPPORTED)
return -ENOENT;
return ret;
}
static int
armpmu_map_hw_event(const unsigned (*event_map)[PERF_COUNT_HW_MAX], u64 config)
{
ARM: 7810/1: perf: Fix array out of bounds access in armpmu_map_hw_event() Vince Weaver reports an oops in the ARM perf event code while running his perf_fuzzer tool on a pandaboard running v3.11-rc4. Unable to handle kernel paging request at virtual address 73fd14cc pgd = eca6c000 [73fd14cc] *pgd=00000000 Internal error: Oops: 5 [#1] SMP ARM Modules linked in: snd_soc_omap_hdmi omapdss snd_soc_omap_abe_twl6040 snd_soc_twl6040 snd_soc_omap snd_soc_omap_hdmi_card snd_soc_omap_mcpdm snd_soc_omap_mcbsp snd_soc_core snd_compress regmap_spi snd_pcm snd_page_alloc snd_timer snd soundcore CPU: 1 PID: 2790 Comm: perf_fuzzer Not tainted 3.11.0-rc4 #6 task: eddcab80 ti: ed892000 task.ti: ed892000 PC is at armpmu_map_event+0x20/0x88 LR is at armpmu_event_init+0x38/0x280 pc : [<c001c3e4>] lr : [<c001c17c>] psr: 60000013 sp : ed893e40 ip : ecececec fp : edfaec00 r10: 00000000 r9 : 00000000 r8 : ed8c3ac0 r7 : ed8c3b5c r6 : edfaec00 r5 : 00000000 r4 : 00000000 r3 : 000000ff r2 : c0496144 r1 : c049611c r0 : edfaec00 Flags: nZCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment user Control: 10c5387d Table: aca6c04a DAC: 00000015 Process perf_fuzzer (pid: 2790, stack limit = 0xed892240) Stack: (0xed893e40 to 0xed894000) 3e40: 00000800 c001c17c 00000002 c008a748 00000001 00000000 00000000 c00bf078 3e60: 00000000 edfaee50 00000000 00000000 00000000 edfaec00 ed8c3ac0 edfaec00 3e80: 00000000 c073ffac ed893f20 c00bf180 00000001 00000000 c00bf078 ed893f20 3ea0: 00000000 ed8c3ac0 00000000 00000000 00000000 c0cb0818 eddcab80 c00bf440 3ec0: ed893f20 00000000 eddcab80 eca76800 00000000 eca76800 00000000 00000000 3ee0: 00000000 ec984c80 eddcab80 c00bfe68 00000000 00000000 00000000 00000080 3f00: 00000000 ed892000 00000000 ed892030 00000004 ecc7e3c8 ecc7e3c8 00000000 3f20: 00000000 00000048 ecececec 00000000 00000000 00000000 00000000 00000000 3f40: 00000000 00000000 00297810 00000000 00000000 00000000 00000000 00000000 3f60: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 3f80: 00000002 00000002 000103a4 00000002 0000016c c00128e8 ed892000 00000000 3fa0: 00090998 c0012700 00000002 000103a4 00090ab8 00000000 00000000 0000000f 3fc0: 00000002 000103a4 00000002 0000016c 00090ab0 00090ab8 000107a0 00090998 3fe0: bed92be0 bed92bd0 0000b785 b6e8f6d0 40000010 00090ab8 00000000 00000000 [<c001c3e4>] (armpmu_map_event+0x20/0x88) from [<c001c17c>] (armpmu_event_init+0x38/0x280) [<c001c17c>] (armpmu_event_init+0x38/0x280) from [<c00bf180>] (perf_init_event+0x108/0x180) [<c00bf180>] (perf_init_event+0x108/0x180) from [<c00bf440>] (perf_event_alloc+0x248/0x40c) [<c00bf440>] (perf_event_alloc+0x248/0x40c) from [<c00bfe68>] (SyS_perf_event_open+0x4f4/0x8fc) [<c00bfe68>] (SyS_perf_event_open+0x4f4/0x8fc) from [<c0012700>] (ret_fast_syscall+0x0/0x48) Code: 0a000005 e3540004 0a000016 e3540000 (0791010c) This is because event->attr.config in armpmu_event_init() contains a very large number copied directly from userspace and is never checked against the size of the array indexed in armpmu_map_hw_event(). Fix the problem by checking the value of config before indexing the array and rejecting invalid config values. Reported-by: Vince Weaver <vincent.weaver@maine.edu> Tested-by: Vince Weaver <vincent.weaver@maine.edu> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Stephen Boyd <sboyd@codeaurora.org> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
2013-08-08 19:41:59 +02:00
int mapping;
if (config >= PERF_COUNT_HW_MAX)
return -EINVAL;
if (!event_map)
return -ENOENT;
ARM: 7810/1: perf: Fix array out of bounds access in armpmu_map_hw_event() Vince Weaver reports an oops in the ARM perf event code while running his perf_fuzzer tool on a pandaboard running v3.11-rc4. Unable to handle kernel paging request at virtual address 73fd14cc pgd = eca6c000 [73fd14cc] *pgd=00000000 Internal error: Oops: 5 [#1] SMP ARM Modules linked in: snd_soc_omap_hdmi omapdss snd_soc_omap_abe_twl6040 snd_soc_twl6040 snd_soc_omap snd_soc_omap_hdmi_card snd_soc_omap_mcpdm snd_soc_omap_mcbsp snd_soc_core snd_compress regmap_spi snd_pcm snd_page_alloc snd_timer snd soundcore CPU: 1 PID: 2790 Comm: perf_fuzzer Not tainted 3.11.0-rc4 #6 task: eddcab80 ti: ed892000 task.ti: ed892000 PC is at armpmu_map_event+0x20/0x88 LR is at armpmu_event_init+0x38/0x280 pc : [<c001c3e4>] lr : [<c001c17c>] psr: 60000013 sp : ed893e40 ip : ecececec fp : edfaec00 r10: 00000000 r9 : 00000000 r8 : ed8c3ac0 r7 : ed8c3b5c r6 : edfaec00 r5 : 00000000 r4 : 00000000 r3 : 000000ff r2 : c0496144 r1 : c049611c r0 : edfaec00 Flags: nZCv IRQs on FIQs on Mode SVC_32 ISA ARM Segment user Control: 10c5387d Table: aca6c04a DAC: 00000015 Process perf_fuzzer (pid: 2790, stack limit = 0xed892240) Stack: (0xed893e40 to 0xed894000) 3e40: 00000800 c001c17c 00000002 c008a748 00000001 00000000 00000000 c00bf078 3e60: 00000000 edfaee50 00000000 00000000 00000000 edfaec00 ed8c3ac0 edfaec00 3e80: 00000000 c073ffac ed893f20 c00bf180 00000001 00000000 c00bf078 ed893f20 3ea0: 00000000 ed8c3ac0 00000000 00000000 00000000 c0cb0818 eddcab80 c00bf440 3ec0: ed893f20 00000000 eddcab80 eca76800 00000000 eca76800 00000000 00000000 3ee0: 00000000 ec984c80 eddcab80 c00bfe68 00000000 00000000 00000000 00000080 3f00: 00000000 ed892000 00000000 ed892030 00000004 ecc7e3c8 ecc7e3c8 00000000 3f20: 00000000 00000048 ecececec 00000000 00000000 00000000 00000000 00000000 3f40: 00000000 00000000 00297810 00000000 00000000 00000000 00000000 00000000 3f60: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000 3f80: 00000002 00000002 000103a4 00000002 0000016c c00128e8 ed892000 00000000 3fa0: 00090998 c0012700 00000002 000103a4 00090ab8 00000000 00000000 0000000f 3fc0: 00000002 000103a4 00000002 0000016c 00090ab0 00090ab8 000107a0 00090998 3fe0: bed92be0 bed92bd0 0000b785 b6e8f6d0 40000010 00090ab8 00000000 00000000 [<c001c3e4>] (armpmu_map_event+0x20/0x88) from [<c001c17c>] (armpmu_event_init+0x38/0x280) [<c001c17c>] (armpmu_event_init+0x38/0x280) from [<c00bf180>] (perf_init_event+0x108/0x180) [<c00bf180>] (perf_init_event+0x108/0x180) from [<c00bf440>] (perf_event_alloc+0x248/0x40c) [<c00bf440>] (perf_event_alloc+0x248/0x40c) from [<c00bfe68>] (SyS_perf_event_open+0x4f4/0x8fc) [<c00bfe68>] (SyS_perf_event_open+0x4f4/0x8fc) from [<c0012700>] (ret_fast_syscall+0x0/0x48) Code: 0a000005 e3540004 0a000016 e3540000 (0791010c) This is because event->attr.config in armpmu_event_init() contains a very large number copied directly from userspace and is never checked against the size of the array indexed in armpmu_map_hw_event(). Fix the problem by checking the value of config before indexing the array and rejecting invalid config values. Reported-by: Vince Weaver <vincent.weaver@maine.edu> Tested-by: Vince Weaver <vincent.weaver@maine.edu> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Stephen Boyd <sboyd@codeaurora.org> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
2013-08-08 19:41:59 +02:00
mapping = (*event_map)[config];
return mapping == HW_OP_UNSUPPORTED ? -ENOENT : mapping;
}
static int
armpmu_map_raw_event(u32 raw_event_mask, u64 config)
{
return (int)(config & raw_event_mask);
}
int
armpmu_map_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;
int type = event->attr.type;
if (type == event->pmu->type)
return armpmu_map_raw_event(raw_event_mask, config);
switch (type) {
case PERF_TYPE_HARDWARE:
return armpmu_map_hw_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;
}
int armpmu_event_set_period(struct perf_event *event)
{
struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
struct hw_perf_event *hwc = &event->hw;
s64 left = local64_read(&hwc->period_left);
s64 period = hwc->sample_period;
int ret = 0;
if (unlikely(left <= -period)) {
left = period;
local64_set(&hwc->period_left, left);
hwc->last_period = period;
ret = 1;
}
if (unlikely(left <= 0)) {
left += period;
local64_set(&hwc->period_left, left);
hwc->last_period = period;
ret = 1;
}
/*
* Limit the maximum period to prevent the counter value
* from overtaking the one we are about to program. In
* effect we are reducing max_period to account for
* interrupt latency (and we are being very conservative).
*/
if (left > (armpmu->max_period >> 1))
left = armpmu->max_period >> 1;
local64_set(&hwc->prev_count, (u64)-left);
armpmu->write_counter(event, (u64)(-left) & 0xffffffff);
perf_event_update_userpage(event);
return ret;
}
u64 armpmu_event_update(struct perf_event *event)
{
struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
struct hw_perf_event *hwc = &event->hw;
u64 delta, prev_raw_count, new_raw_count;
again:
prev_raw_count = local64_read(&hwc->prev_count);
new_raw_count = armpmu->read_counter(event);
if (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
new_raw_count) != prev_raw_count)
goto again;
delta = (new_raw_count - prev_raw_count) & armpmu->max_period;
local64_add(delta, &event->count);
local64_sub(delta, &hwc->period_left);
return new_raw_count;
}
static void
2010-06-16 14:37:10 +02:00
armpmu_read(struct perf_event *event)
{
armpmu_event_update(event);
}
static void
2010-06-16 14:37:10 +02:00
armpmu_stop(struct perf_event *event, int flags)
{
struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
struct hw_perf_event *hwc = &event->hw;
2010-06-16 14:37:10 +02:00
/*
* ARM pmu always has to update the counter, so ignore
* PERF_EF_UPDATE, see comments in armpmu_start().
*/
if (!(hwc->state & PERF_HES_STOPPED)) {
armpmu->disable(event);
armpmu_event_update(event);
2010-06-16 14:37:10 +02:00
hwc->state |= PERF_HES_STOPPED | PERF_HES_UPTODATE;
}
}
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;
2010-06-16 14:37:10 +02:00
/*
* ARM pmu always has to reprogram the period, so ignore
* PERF_EF_RELOAD, see the comment below.
*/
if (flags & PERF_EF_RELOAD)
WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
hwc->state = 0;
/*
* Set the period again. Some counters can't be stopped, so when we
2010-06-16 14:37:10 +02:00
* were stopped we simply disabled the IRQ source and the counter
* may have been left counting. If we don't do this step then we may
* get an interrupt too soon or *way* too late if the overflow has
* happened since disabling.
*/
armpmu_event_set_period(event);
armpmu->enable(event);
}
2010-06-16 14:37:10 +02:00
static void
armpmu_del(struct perf_event *event, int flags)
{
struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
struct pmu_hw_events *hw_events = this_cpu_ptr(armpmu->hw_events);
2010-06-16 14:37:10 +02:00
struct hw_perf_event *hwc = &event->hw;
int idx = hwc->idx;
armpmu_stop(event, PERF_EF_UPDATE);
hw_events->events[idx] = NULL;
clear_bit(idx, hw_events->used_mask);
if (armpmu->clear_event_idx)
armpmu->clear_event_idx(hw_events, event);
2010-06-16 14:37:10 +02:00
perf_event_update_userpage(event);
}
static int
2010-06-16 14:37:10 +02:00
armpmu_add(struct perf_event *event, int flags)
{
struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
struct pmu_hw_events *hw_events = this_cpu_ptr(armpmu->hw_events);
struct hw_perf_event *hwc = &event->hw;
int idx;
/* An event following a process won't be stopped earlier */
if (!cpumask_test_cpu(smp_processor_id(), &armpmu->supported_cpus))
return -ENOENT;
/* If we don't have a space for the counter then finish early. */
idx = armpmu->get_event_idx(hw_events, event);
if (idx < 0)
return idx;
/*
* If there is an event in the counter we are going to use then make
* sure it is disabled.
*/
event->hw.idx = idx;
armpmu->disable(event);
hw_events->events[idx] = event;
2010-06-16 14:37:10 +02:00
hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE;
if (flags & PERF_EF_START)
armpmu_start(event, PERF_EF_RELOAD);
/* Propagate our changes to the userspace mapping. */
perf_event_update_userpage(event);
return 0;
}
static int
ARM: perf: reject groups spanning multiple hardware PMUs The perf core implicitly rejects events spanning multiple HW PMUs, as in these cases the event->ctx will differ. However this validation is performed after pmu::event_init() is called in perf_init_event(), and thus pmu::event_init() may be called with a group leader from a different HW PMU. The ARM PMU driver does not take this fact into account, and when validating groups assumes that it can call to_arm_pmu(event->pmu) for any HW event. When the event in question is from another HW PMU this is wrong, and results in dereferencing garbage. This patch updates the ARM PMU driver to first test for and reject events from other PMUs, moving the to_arm_pmu and related logic after this test. Fixes a crash triggered by perf_fuzzer on Linux-4.0-rc2, with a CCI PMU present: --- CPU: 0 PID: 1527 Comm: perf_fuzzer Not tainted 4.0.0-rc2 #57 Hardware name: ARM-Versatile Express task: bd8484c0 ti: be676000 task.ti: be676000 PC is at 0xbf1bbc90 LR is at validate_event+0x34/0x5c pc : [<bf1bbc90>] lr : [<80016060>] psr: 00000013 ... [<80016060>] (validate_event) from [<80016198>] (validate_group+0x28/0x90) [<80016198>] (validate_group) from [<80016398>] (armpmu_event_init+0x150/0x218) [<80016398>] (armpmu_event_init) from [<800882e4>] (perf_try_init_event+0x30/0x48) [<800882e4>] (perf_try_init_event) from [<8008f544>] (perf_init_event+0x5c/0xf4) [<8008f544>] (perf_init_event) from [<8008f8a8>] (perf_event_alloc+0x2cc/0x35c) [<8008f8a8>] (perf_event_alloc) from [<8009015c>] (SyS_perf_event_open+0x498/0xa70) [<8009015c>] (SyS_perf_event_open) from [<8000e420>] (ret_fast_syscall+0x0/0x34) Code: bf1be000 bf1bb380 802a2664 00000000 (00000002) ---[ end trace 01aff0ff00926a0a ]--- Also cleans up the code to use the arm_pmu only when we know that we are dealing with an arm pmu event. Cc: Will Deacon <will.deacon@arm.com> Acked-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Peter Ziljstra (Intel) <peterz@infradead.org> Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-03-17 19:14:58 +01:00
validate_event(struct pmu *pmu, struct pmu_hw_events *hw_events,
struct perf_event *event)
{
ARM: perf: reject groups spanning multiple hardware PMUs The perf core implicitly rejects events spanning multiple HW PMUs, as in these cases the event->ctx will differ. However this validation is performed after pmu::event_init() is called in perf_init_event(), and thus pmu::event_init() may be called with a group leader from a different HW PMU. The ARM PMU driver does not take this fact into account, and when validating groups assumes that it can call to_arm_pmu(event->pmu) for any HW event. When the event in question is from another HW PMU this is wrong, and results in dereferencing garbage. This patch updates the ARM PMU driver to first test for and reject events from other PMUs, moving the to_arm_pmu and related logic after this test. Fixes a crash triggered by perf_fuzzer on Linux-4.0-rc2, with a CCI PMU present: --- CPU: 0 PID: 1527 Comm: perf_fuzzer Not tainted 4.0.0-rc2 #57 Hardware name: ARM-Versatile Express task: bd8484c0 ti: be676000 task.ti: be676000 PC is at 0xbf1bbc90 LR is at validate_event+0x34/0x5c pc : [<bf1bbc90>] lr : [<80016060>] psr: 00000013 ... [<80016060>] (validate_event) from [<80016198>] (validate_group+0x28/0x90) [<80016198>] (validate_group) from [<80016398>] (armpmu_event_init+0x150/0x218) [<80016398>] (armpmu_event_init) from [<800882e4>] (perf_try_init_event+0x30/0x48) [<800882e4>] (perf_try_init_event) from [<8008f544>] (perf_init_event+0x5c/0xf4) [<8008f544>] (perf_init_event) from [<8008f8a8>] (perf_event_alloc+0x2cc/0x35c) [<8008f8a8>] (perf_event_alloc) from [<8009015c>] (SyS_perf_event_open+0x498/0xa70) [<8009015c>] (SyS_perf_event_open) from [<8000e420>] (ret_fast_syscall+0x0/0x34) Code: bf1be000 bf1bb380 802a2664 00000000 (00000002) ---[ end trace 01aff0ff00926a0a ]--- Also cleans up the code to use the arm_pmu only when we know that we are dealing with an arm pmu event. Cc: Will Deacon <will.deacon@arm.com> Acked-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Peter Ziljstra (Intel) <peterz@infradead.org> Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-03-17 19:14:58 +01:00
struct arm_pmu *armpmu;
if (is_software_event(event))
return 1;
ARM: perf: reject groups spanning multiple hardware PMUs The perf core implicitly rejects events spanning multiple HW PMUs, as in these cases the event->ctx will differ. However this validation is performed after pmu::event_init() is called in perf_init_event(), and thus pmu::event_init() may be called with a group leader from a different HW PMU. The ARM PMU driver does not take this fact into account, and when validating groups assumes that it can call to_arm_pmu(event->pmu) for any HW event. When the event in question is from another HW PMU this is wrong, and results in dereferencing garbage. This patch updates the ARM PMU driver to first test for and reject events from other PMUs, moving the to_arm_pmu and related logic after this test. Fixes a crash triggered by perf_fuzzer on Linux-4.0-rc2, with a CCI PMU present: --- CPU: 0 PID: 1527 Comm: perf_fuzzer Not tainted 4.0.0-rc2 #57 Hardware name: ARM-Versatile Express task: bd8484c0 ti: be676000 task.ti: be676000 PC is at 0xbf1bbc90 LR is at validate_event+0x34/0x5c pc : [<bf1bbc90>] lr : [<80016060>] psr: 00000013 ... [<80016060>] (validate_event) from [<80016198>] (validate_group+0x28/0x90) [<80016198>] (validate_group) from [<80016398>] (armpmu_event_init+0x150/0x218) [<80016398>] (armpmu_event_init) from [<800882e4>] (perf_try_init_event+0x30/0x48) [<800882e4>] (perf_try_init_event) from [<8008f544>] (perf_init_event+0x5c/0xf4) [<8008f544>] (perf_init_event) from [<8008f8a8>] (perf_event_alloc+0x2cc/0x35c) [<8008f8a8>] (perf_event_alloc) from [<8009015c>] (SyS_perf_event_open+0x498/0xa70) [<8009015c>] (SyS_perf_event_open) from [<8000e420>] (ret_fast_syscall+0x0/0x34) Code: bf1be000 bf1bb380 802a2664 00000000 (00000002) ---[ end trace 01aff0ff00926a0a ]--- Also cleans up the code to use the arm_pmu only when we know that we are dealing with an arm pmu event. Cc: Will Deacon <will.deacon@arm.com> Acked-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Peter Ziljstra (Intel) <peterz@infradead.org> Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-03-17 19:14:58 +01:00
/*
* Reject groups spanning multiple HW PMUs (e.g. CPU + CCI). The
* core perf code won't check that the pmu->ctx == leader->ctx
* until after pmu->event_init(event).
*/
if (event->pmu != pmu)
return 0;
if (event->state < PERF_EVENT_STATE_OFF)
return 1;
if (event->state == PERF_EVENT_STATE_OFF && !event->attr.enable_on_exec)
return 1;
ARM: perf: reject groups spanning multiple hardware PMUs The perf core implicitly rejects events spanning multiple HW PMUs, as in these cases the event->ctx will differ. However this validation is performed after pmu::event_init() is called in perf_init_event(), and thus pmu::event_init() may be called with a group leader from a different HW PMU. The ARM PMU driver does not take this fact into account, and when validating groups assumes that it can call to_arm_pmu(event->pmu) for any HW event. When the event in question is from another HW PMU this is wrong, and results in dereferencing garbage. This patch updates the ARM PMU driver to first test for and reject events from other PMUs, moving the to_arm_pmu and related logic after this test. Fixes a crash triggered by perf_fuzzer on Linux-4.0-rc2, with a CCI PMU present: --- CPU: 0 PID: 1527 Comm: perf_fuzzer Not tainted 4.0.0-rc2 #57 Hardware name: ARM-Versatile Express task: bd8484c0 ti: be676000 task.ti: be676000 PC is at 0xbf1bbc90 LR is at validate_event+0x34/0x5c pc : [<bf1bbc90>] lr : [<80016060>] psr: 00000013 ... [<80016060>] (validate_event) from [<80016198>] (validate_group+0x28/0x90) [<80016198>] (validate_group) from [<80016398>] (armpmu_event_init+0x150/0x218) [<80016398>] (armpmu_event_init) from [<800882e4>] (perf_try_init_event+0x30/0x48) [<800882e4>] (perf_try_init_event) from [<8008f544>] (perf_init_event+0x5c/0xf4) [<8008f544>] (perf_init_event) from [<8008f8a8>] (perf_event_alloc+0x2cc/0x35c) [<8008f8a8>] (perf_event_alloc) from [<8009015c>] (SyS_perf_event_open+0x498/0xa70) [<8009015c>] (SyS_perf_event_open) from [<8000e420>] (ret_fast_syscall+0x0/0x34) Code: bf1be000 bf1bb380 802a2664 00000000 (00000002) ---[ end trace 01aff0ff00926a0a ]--- Also cleans up the code to use the arm_pmu only when we know that we are dealing with an arm pmu event. Cc: Will Deacon <will.deacon@arm.com> Acked-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Peter Ziljstra (Intel) <peterz@infradead.org> Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-03-17 19:14:58 +01:00
armpmu = to_arm_pmu(event->pmu);
return armpmu->get_event_idx(hw_events, event) >= 0;
}
static int
validate_group(struct perf_event *event)
{
struct perf_event *sibling, *leader = event->group_leader;
struct pmu_hw_events fake_pmu;
/*
* Initialise the fake PMU. We only need to populate the
* used_mask for the purposes of validation.
*/
memset(&fake_pmu.used_mask, 0, sizeof(fake_pmu.used_mask));
ARM: perf: reject groups spanning multiple hardware PMUs The perf core implicitly rejects events spanning multiple HW PMUs, as in these cases the event->ctx will differ. However this validation is performed after pmu::event_init() is called in perf_init_event(), and thus pmu::event_init() may be called with a group leader from a different HW PMU. The ARM PMU driver does not take this fact into account, and when validating groups assumes that it can call to_arm_pmu(event->pmu) for any HW event. When the event in question is from another HW PMU this is wrong, and results in dereferencing garbage. This patch updates the ARM PMU driver to first test for and reject events from other PMUs, moving the to_arm_pmu and related logic after this test. Fixes a crash triggered by perf_fuzzer on Linux-4.0-rc2, with a CCI PMU present: --- CPU: 0 PID: 1527 Comm: perf_fuzzer Not tainted 4.0.0-rc2 #57 Hardware name: ARM-Versatile Express task: bd8484c0 ti: be676000 task.ti: be676000 PC is at 0xbf1bbc90 LR is at validate_event+0x34/0x5c pc : [<bf1bbc90>] lr : [<80016060>] psr: 00000013 ... [<80016060>] (validate_event) from [<80016198>] (validate_group+0x28/0x90) [<80016198>] (validate_group) from [<80016398>] (armpmu_event_init+0x150/0x218) [<80016398>] (armpmu_event_init) from [<800882e4>] (perf_try_init_event+0x30/0x48) [<800882e4>] (perf_try_init_event) from [<8008f544>] (perf_init_event+0x5c/0xf4) [<8008f544>] (perf_init_event) from [<8008f8a8>] (perf_event_alloc+0x2cc/0x35c) [<8008f8a8>] (perf_event_alloc) from [<8009015c>] (SyS_perf_event_open+0x498/0xa70) [<8009015c>] (SyS_perf_event_open) from [<8000e420>] (ret_fast_syscall+0x0/0x34) Code: bf1be000 bf1bb380 802a2664 00000000 (00000002) ---[ end trace 01aff0ff00926a0a ]--- Also cleans up the code to use the arm_pmu only when we know that we are dealing with an arm pmu event. Cc: Will Deacon <will.deacon@arm.com> Acked-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Peter Ziljstra (Intel) <peterz@infradead.org> Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-03-17 19:14:58 +01:00
if (!validate_event(event->pmu, &fake_pmu, leader))
return -EINVAL;
for_each_sibling_event(sibling, leader) {
ARM: perf: reject groups spanning multiple hardware PMUs The perf core implicitly rejects events spanning multiple HW PMUs, as in these cases the event->ctx will differ. However this validation is performed after pmu::event_init() is called in perf_init_event(), and thus pmu::event_init() may be called with a group leader from a different HW PMU. The ARM PMU driver does not take this fact into account, and when validating groups assumes that it can call to_arm_pmu(event->pmu) for any HW event. When the event in question is from another HW PMU this is wrong, and results in dereferencing garbage. This patch updates the ARM PMU driver to first test for and reject events from other PMUs, moving the to_arm_pmu and related logic after this test. Fixes a crash triggered by perf_fuzzer on Linux-4.0-rc2, with a CCI PMU present: --- CPU: 0 PID: 1527 Comm: perf_fuzzer Not tainted 4.0.0-rc2 #57 Hardware name: ARM-Versatile Express task: bd8484c0 ti: be676000 task.ti: be676000 PC is at 0xbf1bbc90 LR is at validate_event+0x34/0x5c pc : [<bf1bbc90>] lr : [<80016060>] psr: 00000013 ... [<80016060>] (validate_event) from [<80016198>] (validate_group+0x28/0x90) [<80016198>] (validate_group) from [<80016398>] (armpmu_event_init+0x150/0x218) [<80016398>] (armpmu_event_init) from [<800882e4>] (perf_try_init_event+0x30/0x48) [<800882e4>] (perf_try_init_event) from [<8008f544>] (perf_init_event+0x5c/0xf4) [<8008f544>] (perf_init_event) from [<8008f8a8>] (perf_event_alloc+0x2cc/0x35c) [<8008f8a8>] (perf_event_alloc) from [<8009015c>] (SyS_perf_event_open+0x498/0xa70) [<8009015c>] (SyS_perf_event_open) from [<8000e420>] (ret_fast_syscall+0x0/0x34) Code: bf1be000 bf1bb380 802a2664 00000000 (00000002) ---[ end trace 01aff0ff00926a0a ]--- Also cleans up the code to use the arm_pmu only when we know that we are dealing with an arm pmu event. Cc: Will Deacon <will.deacon@arm.com> Acked-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Peter Ziljstra (Intel) <peterz@infradead.org> Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-03-17 19:14:58 +01:00
if (!validate_event(event->pmu, &fake_pmu, sibling))
return -EINVAL;
}
ARM: perf: reject groups spanning multiple hardware PMUs The perf core implicitly rejects events spanning multiple HW PMUs, as in these cases the event->ctx will differ. However this validation is performed after pmu::event_init() is called in perf_init_event(), and thus pmu::event_init() may be called with a group leader from a different HW PMU. The ARM PMU driver does not take this fact into account, and when validating groups assumes that it can call to_arm_pmu(event->pmu) for any HW event. When the event in question is from another HW PMU this is wrong, and results in dereferencing garbage. This patch updates the ARM PMU driver to first test for and reject events from other PMUs, moving the to_arm_pmu and related logic after this test. Fixes a crash triggered by perf_fuzzer on Linux-4.0-rc2, with a CCI PMU present: --- CPU: 0 PID: 1527 Comm: perf_fuzzer Not tainted 4.0.0-rc2 #57 Hardware name: ARM-Versatile Express task: bd8484c0 ti: be676000 task.ti: be676000 PC is at 0xbf1bbc90 LR is at validate_event+0x34/0x5c pc : [<bf1bbc90>] lr : [<80016060>] psr: 00000013 ... [<80016060>] (validate_event) from [<80016198>] (validate_group+0x28/0x90) [<80016198>] (validate_group) from [<80016398>] (armpmu_event_init+0x150/0x218) [<80016398>] (armpmu_event_init) from [<800882e4>] (perf_try_init_event+0x30/0x48) [<800882e4>] (perf_try_init_event) from [<8008f544>] (perf_init_event+0x5c/0xf4) [<8008f544>] (perf_init_event) from [<8008f8a8>] (perf_event_alloc+0x2cc/0x35c) [<8008f8a8>] (perf_event_alloc) from [<8009015c>] (SyS_perf_event_open+0x498/0xa70) [<8009015c>] (SyS_perf_event_open) from [<8000e420>] (ret_fast_syscall+0x0/0x34) Code: bf1be000 bf1bb380 802a2664 00000000 (00000002) ---[ end trace 01aff0ff00926a0a ]--- Also cleans up the code to use the arm_pmu only when we know that we are dealing with an arm pmu event. Cc: Will Deacon <will.deacon@arm.com> Acked-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Peter Ziljstra (Intel) <peterz@infradead.org> Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2015-03-17 19:14:58 +01:00
if (!validate_event(event->pmu, &fake_pmu, event))
return -EINVAL;
return 0;
}
static irqreturn_t armpmu_dispatch_irq(int irq, void *dev)
{
struct arm_pmu *armpmu;
int ret;
u64 start_clock, finish_clock;
/*
* we request the IRQ with a (possibly percpu) struct arm_pmu**, but
* the handlers expect a struct arm_pmu*. The percpu_irq framework will
* do any necessary shifting, we just need to perform the first
* dereference.
*/
armpmu = *(void **)dev;
if (WARN_ON_ONCE(!armpmu))
return IRQ_NONE;
start_clock = sched_clock();
ret = armpmu->handle_irq(irq, armpmu);
finish_clock = sched_clock();
perf_sample_event_took(finish_clock - start_clock);
return ret;
}
static int
event_requires_mode_exclusion(struct perf_event_attr *attr)
{
return attr->exclude_idle || attr->exclude_user ||
attr->exclude_kernel || attr->exclude_hv;
}
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;
mapping = armpmu->map_event(event);
if (mapping < 0) {
pr_debug("event %x:%llx not supported\n", event->attr.type,
event->attr.config);
return mapping;
}
/*
* We don't assign an index until we actually place the event onto
* hardware. Use -1 to signify that we haven't decided where to put it
* yet. For SMP systems, each core has it's own PMU so we can't do any
* clever allocation or constraints checking at this point.
*/
hwc->idx = -1;
hwc->config_base = 0;
hwc->config = 0;
hwc->event_base = 0;
/*
* Check whether we need to exclude the counter from certain modes.
*/
if ((!armpmu->set_event_filter ||
armpmu->set_event_filter(hwc, &event->attr)) &&
event_requires_mode_exclusion(&event->attr)) {
pr_debug("ARM performance counters do not support "
"mode exclusion\n");
return -EOPNOTSUPP;
}
/*
* Store the event encoding into the config_base field.
*/
hwc->config_base |= (unsigned long)mapping;
if (!is_sampling_event(event)) {
/*
* For non-sampling runs, limit the sample_period to half
* of the counter width. That way, the new counter value
* is far less likely to overtake the previous one unless
* you have some serious IRQ latency issues.
*/
hwc->sample_period = armpmu->max_period >> 1;
hwc->last_period = hwc->sample_period;
local64_set(&hwc->period_left, hwc->sample_period);
}
if (event->group_leader != event) {
if (validate_group(event) != 0)
return -EINVAL;
}
return 0;
}
static int armpmu_event_init(struct perf_event *event)
{
struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
/*
* Reject CPU-affine events for CPUs that are of a different class to
* that which this PMU handles. Process-following events (where
* event->cpu == -1) can be migrated between CPUs, and thus we have to
* reject them later (in armpmu_add) if they're scheduled on a
* different class of CPU.
*/
if (event->cpu != -1 &&
!cpumask_test_cpu(event->cpu, &armpmu->supported_cpus))
return -ENOENT;
/* does not support taken branch sampling */
if (has_branch_stack(event))
return -EOPNOTSUPP;
if (armpmu->map_event(event) == -ENOENT)
return -ENOENT;
return __hw_perf_event_init(event);
}
2010-06-16 14:37:10 +02:00
static void armpmu_enable(struct pmu *pmu)
{
struct arm_pmu *armpmu = to_arm_pmu(pmu);
struct pmu_hw_events *hw_events = this_cpu_ptr(armpmu->hw_events);
int enabled = bitmap_weight(hw_events->used_mask, armpmu->num_events);
/* For task-bound events we may be called on other CPUs */
if (!cpumask_test_cpu(smp_processor_id(), &armpmu->supported_cpus))
return;
if (enabled)
armpmu->start(armpmu);
}
2010-06-16 14:37:10 +02:00
static void armpmu_disable(struct pmu *pmu)
{
struct arm_pmu *armpmu = to_arm_pmu(pmu);
/* For task-bound events we may be called on other CPUs */
if (!cpumask_test_cpu(smp_processor_id(), &armpmu->supported_cpus))
return;
armpmu->stop(armpmu);
}
/*
* In heterogeneous systems, events are specific to a particular
* microarchitecture, and aren't suitable for another. Thus, only match CPUs of
* the same microarchitecture.
*/
static int armpmu_filter_match(struct perf_event *event)
{
struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
unsigned int cpu = smp_processor_id();
return cpumask_test_cpu(cpu, &armpmu->supported_cpus);
}
static ssize_t armpmu_cpumask_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct arm_pmu *armpmu = to_arm_pmu(dev_get_drvdata(dev));
return cpumap_print_to_pagebuf(true, buf, &armpmu->supported_cpus);
}
static DEVICE_ATTR(cpus, S_IRUGO, armpmu_cpumask_show, NULL);
static struct attribute *armpmu_common_attrs[] = {
&dev_attr_cpus.attr,
NULL,
};
static struct attribute_group armpmu_common_attr_group = {
.attrs = armpmu_common_attrs,
};
/* Set at runtime when we know what CPU type we are. */
static struct arm_pmu *__oprofile_cpu_pmu;
/*
* Despite the names, these two functions are CPU-specific and are used
* by the OProfile/perf code.
*/
const char *perf_pmu_name(void)
{
if (!__oprofile_cpu_pmu)
return NULL;
return __oprofile_cpu_pmu->name;
}
EXPORT_SYMBOL_GPL(perf_pmu_name);
int perf_num_counters(void)
{
int max_events = 0;
if (__oprofile_cpu_pmu != NULL)
max_events = __oprofile_cpu_pmu->num_events;
return max_events;
}
EXPORT_SYMBOL_GPL(perf_num_counters);
static int armpmu_count_irq_users(const int irq)
{
int cpu, count = 0;
for_each_possible_cpu(cpu) {
if (per_cpu(cpu_irq, cpu) == irq)
count++;
}
return count;
}
void armpmu_free_irq(int irq, int cpu)
{
if (per_cpu(cpu_irq, cpu) == 0)
return;
if (WARN_ON(irq != per_cpu(cpu_irq, cpu)))
return;
if (!irq_is_percpu_devid(irq))
free_irq(irq, per_cpu_ptr(&cpu_armpmu, cpu));
else if (armpmu_count_irq_users(irq) == 1)
free_percpu_irq(irq, &cpu_armpmu);
per_cpu(cpu_irq, cpu) = 0;
}
int armpmu_request_irq(int irq, int cpu)
{
int err = 0;
const irq_handler_t handler = armpmu_dispatch_irq;
if (!irq)
return 0;
if (!irq_is_percpu_devid(irq)) {
unsigned long irq_flags;
err = irq_force_affinity(irq, cpumask_of(cpu));
if (err && num_possible_cpus() > 1) {
pr_warn("unable to set irq affinity (irq=%d, cpu=%u)\n",
irq, cpu);
goto err_out;
}
irq_flags = IRQF_PERCPU |
IRQF_NOBALANCING |
IRQF_NO_THREAD;
irq_set_status_flags(irq, IRQ_NOAUTOEN);
err = request_irq(irq, handler, irq_flags, "arm-pmu",
per_cpu_ptr(&cpu_armpmu, cpu));
} else if (armpmu_count_irq_users(irq) == 0) {
err = request_percpu_irq(irq, handler, "arm-pmu",
&cpu_armpmu);
}
if (err)
goto err_out;
per_cpu(cpu_irq, cpu) = irq;
return 0;
err_out:
pr_err("unable to request IRQ%d for ARM PMU counters\n", irq);
return err;
}
static int armpmu_get_cpu_irq(struct arm_pmu *pmu, int cpu)
{
struct pmu_hw_events __percpu *hw_events = pmu->hw_events;
return per_cpu(hw_events->irq, cpu);
}
/*
* PMU hardware loses all context when a CPU goes offline.
* When a CPU is hotplugged back in, since some hardware registers are
* UNKNOWN at reset, the PMU must be explicitly reset to avoid reading
* junk values out of them.
*/
static int arm_perf_starting_cpu(unsigned int cpu, struct hlist_node *node)
{
struct arm_pmu *pmu = hlist_entry_safe(node, struct arm_pmu, node);
int irq;
if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
return 0;
if (pmu->reset)
pmu->reset(pmu);
per_cpu(cpu_armpmu, cpu) = pmu;
irq = armpmu_get_cpu_irq(pmu, cpu);
if (irq) {
if (irq_is_percpu_devid(irq))
enable_percpu_irq(irq, IRQ_TYPE_NONE);
else
enable_irq(irq);
}
return 0;
}
static int arm_perf_teardown_cpu(unsigned int cpu, struct hlist_node *node)
{
struct arm_pmu *pmu = hlist_entry_safe(node, struct arm_pmu, node);
int irq;
if (!cpumask_test_cpu(cpu, &pmu->supported_cpus))
return 0;
irq = armpmu_get_cpu_irq(pmu, cpu);
if (irq) {
if (irq_is_percpu_devid(irq))
disable_percpu_irq(irq);
else
arm_pmu: Use disable_irq_nosync when disabling SPI in CPU teardown hook Commit 6de3f79112cc ("arm_pmu: explicitly enable/disable SPIs at hotplug") moved all of the arm_pmu IRQ enable/disable calls to the CPU hotplug hooks, regardless of whether they are implemented as PPIs or SPIs. This can lead to us sleeping from atomic context due to disable_irq blocking: | BUG: sleeping function called from invalid context at kernel/irq/manage.c:112 | in_atomic(): 1, irqs_disabled(): 128, pid: 15, name: migration/1 | no locks held by migration/1/15. | irq event stamp: 192 | hardirqs last enabled at (191): [<00000000803c2507>] | _raw_spin_unlock_irq+0x2c/0x4c | hardirqs last disabled at (192): [<000000007f57ad28>] multi_cpu_stop+0x9c/0x140 | softirqs last enabled at (0): [<0000000004ee1b58>] | copy_process.isra.77.part.78+0x43c/0x1504 | softirqs last disabled at (0): [< (null)>] (null) | CPU: 1 PID: 15 Comm: migration/1 Not tainted 4.16.0-rc3-salvator-x #1651 | Hardware name: Renesas Salvator-X board based on r8a7796 (DT) | Call trace: | dump_backtrace+0x0/0x140 | show_stack+0x14/0x1c | dump_stack+0xb4/0xf0 | ___might_sleep+0x1fc/0x218 | __might_sleep+0x70/0x80 | synchronize_irq+0x40/0xa8 | disable_irq+0x20/0x2c | arm_perf_teardown_cpu+0x80/0xac Since the interrupt is always CPU-affine and this code is running with interrupts disabled, we can just use disable_irq_nosync as we know there isn't a concurrent invocation of the handler to worry about. Fixes: 6de3f79112cc ("arm_pmu: explicitly enable/disable SPIs at hotplug") Reported-by: Geert Uytterhoeven <geert@linux-m68k.org> Tested-by: Geert Uytterhoeven <geert+renesas@glider.be> Acked-by: Mark Rutland <mark.rutland@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2018-02-26 17:10:56 +01:00
disable_irq_nosync(irq);
}
per_cpu(cpu_armpmu, cpu) = NULL;
return 0;
}
#ifdef CONFIG_CPU_PM
static void cpu_pm_pmu_setup(struct arm_pmu *armpmu, unsigned long cmd)
{
struct pmu_hw_events *hw_events = this_cpu_ptr(armpmu->hw_events);
struct perf_event *event;
int idx;
for (idx = 0; idx < armpmu->num_events; idx++) {
/*
* If the counter is not used skip it, there is no
* need of stopping/restarting it.
*/
if (!test_bit(idx, hw_events->used_mask))
continue;
event = hw_events->events[idx];
switch (cmd) {
case CPU_PM_ENTER:
/*
* Stop and update the counter
*/
armpmu_stop(event, PERF_EF_UPDATE);
break;
case CPU_PM_EXIT:
case CPU_PM_ENTER_FAILED:
drivers/perf: arm-pmu: fix RCU usage on pmu resume from low-power Commit da4e4f18afe0 ("drivers/perf: arm_pmu: implement CPU_PM notifier") added code in the arm perf infrastructure that allows the kernel to save/restore perf counters whenever the CPU enters a low-power state. The kernel saves/restores the counters for each active event through the armpmu_{stop/start} ARM pmu API, so that the low-power state enter/exit cycle is emulated through pmu start/stop operations for each event in use. However, calling armpmu_start() for each active event on power up executes code that requires RCU locking (perf_event_update_userpage()) to be functional, so, given that the core may call the CPU_PM notifiers while running the idle thread in an quiescent RCU state this is not allowed as detected through the following splat when kernel is run with CONFIG_PROVE_LOCKING enabled: [ 49.293286] [ 49.294761] =============================== [ 49.298895] [ INFO: suspicious RCU usage. ] [ 49.303031] 4.6.0-rc3+ #421 Not tainted [ 49.306821] ------------------------------- [ 49.310956] include/linux/rcupdate.h:872 rcu_read_lock() used illegally while idle! [ 49.318530] [ 49.318530] other info that might help us debug this: [ 49.318530] [ 49.326451] [ 49.326451] RCU used illegally from idle CPU! [ 49.326451] rcu_scheduler_active = 1, debug_locks = 0 [ 49.337209] RCU used illegally from extended quiescent state! [ 49.342892] 2 locks held by swapper/2/0: [ 49.346768] #0: (cpu_pm_notifier_lock){......}, at: [<ffffff8008163c28>] cpu_pm_exit+0x18/0x80 [ 49.355492] #1: (rcu_read_lock){......}, at: [<ffffff800816dc38>] perf_event_update_userpage+0x0/0x260 This patch wraps the armpmu_start() call (that indirectly calls perf_event_update_userpage()) on CPU_PM notifier power state exit (or failed entry) within the RCU_NONIDLE() macro so that the RCU subsystem is made aware the calling cpu is not idle from an RCU perspective for the armpmu_start() call duration, therefore fixing the issue. Fixes: da4e4f18afe0 ("drivers/perf: arm_pmu: implement CPU_PM notifier") Signed-off-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com> Reported-by: James Morse <james.morse@arm.com> Suggested-by: Kevin Hilman <khilman@baylibre.com> Cc: Ashwin Chaugule <ashwin.chaugule@linaro.org> Cc: Kevin Hilman <khilman@baylibre.com> Cc: Sudeep Holla <sudeep.holla@arm.com> Cc: Daniel Lezcano <daniel.lezcano@linaro.org> Cc: Mathieu Poirier <mathieu.poirier@linaro.org> Acked-by: Mark Rutland <mark.rutland@arm.com> Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Acked-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
2016-04-21 11:24:34 +02:00
/*
* Restore and enable the counter.
* armpmu_start() indirectly calls
*
* perf_event_update_userpage()
*
* that requires RCU read locking to be functional,
* wrap the call within RCU_NONIDLE to make the
* RCU subsystem aware this cpu is not idle from
* an RCU perspective for the armpmu_start() call
* duration.
*/
RCU_NONIDLE(armpmu_start(event, PERF_EF_RELOAD));
break;
default:
break;
}
}
}
static int cpu_pm_pmu_notify(struct notifier_block *b, unsigned long cmd,
void *v)
{
struct arm_pmu *armpmu = container_of(b, struct arm_pmu, cpu_pm_nb);
struct pmu_hw_events *hw_events = this_cpu_ptr(armpmu->hw_events);
int enabled = bitmap_weight(hw_events->used_mask, armpmu->num_events);
if (!cpumask_test_cpu(smp_processor_id(), &armpmu->supported_cpus))
return NOTIFY_DONE;
/*
* Always reset the PMU registers on power-up even if
* there are no events running.
*/
if (cmd == CPU_PM_EXIT && armpmu->reset)
armpmu->reset(armpmu);
if (!enabled)
return NOTIFY_OK;
switch (cmd) {
case CPU_PM_ENTER:
armpmu->stop(armpmu);
cpu_pm_pmu_setup(armpmu, cmd);
break;
case CPU_PM_EXIT:
cpu_pm_pmu_setup(armpmu, cmd);
case CPU_PM_ENTER_FAILED:
armpmu->start(armpmu);
break;
default:
return NOTIFY_DONE;
}
return NOTIFY_OK;
}
static int cpu_pm_pmu_register(struct arm_pmu *cpu_pmu)
{
cpu_pmu->cpu_pm_nb.notifier_call = cpu_pm_pmu_notify;
return cpu_pm_register_notifier(&cpu_pmu->cpu_pm_nb);
}
static void cpu_pm_pmu_unregister(struct arm_pmu *cpu_pmu)
{
cpu_pm_unregister_notifier(&cpu_pmu->cpu_pm_nb);
}
#else
static inline int cpu_pm_pmu_register(struct arm_pmu *cpu_pmu) { return 0; }
static inline void cpu_pm_pmu_unregister(struct arm_pmu *cpu_pmu) { }
#endif
static int cpu_pmu_init(struct arm_pmu *cpu_pmu)
{
int err;
err = cpuhp_state_add_instance(CPUHP_AP_PERF_ARM_STARTING,
&cpu_pmu->node);
if (err)
goto out;
err = cpu_pm_pmu_register(cpu_pmu);
if (err)
goto out_unregister;
return 0;
out_unregister:
cpuhp_state_remove_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
&cpu_pmu->node);
out:
return err;
}
static void cpu_pmu_destroy(struct arm_pmu *cpu_pmu)
{
cpu_pm_pmu_unregister(cpu_pmu);
cpuhp_state_remove_instance_nocalls(CPUHP_AP_PERF_ARM_STARTING,
&cpu_pmu->node);
}
static struct arm_pmu *__armpmu_alloc(gfp_t flags)
{
struct arm_pmu *pmu;
int cpu;
pmu = kzalloc(sizeof(*pmu), flags);
if (!pmu) {
pr_info("failed to allocate PMU device!\n");
goto out;
}
pmu->hw_events = alloc_percpu_gfp(struct pmu_hw_events, flags);
if (!pmu->hw_events) {
pr_info("failed to allocate per-cpu PMU data.\n");
goto out_free_pmu;
}
pmu->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,
.filter_match = armpmu_filter_match,
.attr_groups = pmu->attr_groups,
/*
* This is a CPU PMU potentially in a heterogeneous
* configuration (e.g. big.LITTLE). This is not an uncore PMU,
* and we have taken ctx sharing into account (e.g. with our
* pmu::filter_match callback and pmu::event_init group
* validation).
*/
.capabilities = PERF_PMU_CAP_HETEROGENEOUS_CPUS,
};
pmu->attr_groups[ARMPMU_ATTR_GROUP_COMMON] =
&armpmu_common_attr_group;
for_each_possible_cpu(cpu) {
struct pmu_hw_events *events;
events = per_cpu_ptr(pmu->hw_events, cpu);
raw_spin_lock_init(&events->pmu_lock);
events->percpu_pmu = pmu;
}
return pmu;
out_free_pmu:
kfree(pmu);
out:
return NULL;
}
struct arm_pmu *armpmu_alloc(void)
{
return __armpmu_alloc(GFP_KERNEL);
}
struct arm_pmu *armpmu_alloc_atomic(void)
{
return __armpmu_alloc(GFP_ATOMIC);
}
void armpmu_free(struct arm_pmu *pmu)
{
free_percpu(pmu->hw_events);
kfree(pmu);
}
int armpmu_register(struct arm_pmu *pmu)
{
int ret;
ret = cpu_pmu_init(pmu);
if (ret)
return ret;
ret = perf_pmu_register(&pmu->pmu, pmu->name, -1);
if (ret)
goto out_destroy;
if (!__oprofile_cpu_pmu)
__oprofile_cpu_pmu = pmu;
pr_info("enabled with %s PMU driver, %d counters available\n",
pmu->name, pmu->num_events);
return 0;
out_destroy:
cpu_pmu_destroy(pmu);
return ret;
}
static int arm_pmu_hp_init(void)
{
int ret;
ret = cpuhp_setup_state_multi(CPUHP_AP_PERF_ARM_STARTING,
"perf/arm/pmu:starting",
arm_perf_starting_cpu,
arm_perf_teardown_cpu);
if (ret)
pr_err("CPU hotplug notifier for ARM PMU could not be registered: %d\n",
ret);
return ret;
}
subsys_initcall(arm_pmu_hp_init);