It doesn't make sense for a perf event to be configured as a CHAIN event in isolation, so extend the arm_pmu structure with a ->filter_match() function to allow the backend PMU implementation to reject CHAIN events early.
Cc: stable@vger.kernel.org Signed-off-by: Will Deacon will.deacon@arm.com --- arch/arm64/kernel/perf_event.c | 7 +++++++ drivers/perf/arm_pmu.c | 8 +++++++- include/linux/perf/arm_pmu.h | 1 + 3 files changed, 15 insertions(+), 1 deletion(-)
diff --git a/arch/arm64/kernel/perf_event.c b/arch/arm64/kernel/perf_event.c index 8e38d5267f22..e213f8e867f6 100644 --- a/arch/arm64/kernel/perf_event.c +++ b/arch/arm64/kernel/perf_event.c @@ -966,6 +966,12 @@ static int armv8pmu_set_event_filter(struct hw_perf_event *event, return 0; }
+static int armv8pmu_filter_match(struct perf_event *event) +{ + unsigned long evtype = event->hw.config_base & ARMV8_PMU_EVTYPE_EVENT; + return evtype != ARMV8_PMUV3_PERFCTR_CHAIN; +} + static void armv8pmu_reset(void *info) { struct arm_pmu *cpu_pmu = (struct arm_pmu *)info; @@ -1114,6 +1120,7 @@ static int armv8_pmu_init(struct arm_pmu *cpu_pmu) cpu_pmu->stop = armv8pmu_stop, cpu_pmu->reset = armv8pmu_reset, cpu_pmu->set_event_filter = armv8pmu_set_event_filter; + cpu_pmu->filter_match = armv8pmu_filter_match;
return 0; } diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c index 7f01f6f60b87..d0b7dd8fb184 100644 --- a/drivers/perf/arm_pmu.c +++ b/drivers/perf/arm_pmu.c @@ -485,7 +485,13 @@ 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); + int ret; + + ret = cpumask_test_cpu(cpu, &armpmu->supported_cpus); + if (ret && armpmu->filter_match) + return armpmu->filter_match(event); + + return ret; }
static ssize_t armpmu_cpumask_show(struct device *dev, diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h index 10f92e1d8e7b..bf309ff6f244 100644 --- a/include/linux/perf/arm_pmu.h +++ b/include/linux/perf/arm_pmu.h @@ -99,6 +99,7 @@ struct arm_pmu { void (*stop)(struct arm_pmu *); void (*reset)(void *); int (*map_event)(struct perf_event *event); + int (*filter_match)(struct perf_event *event); int num_events; bool secure_access; /* 32-bit ARM only */ #define ARMV8_PMUV3_MAX_COMMON_EVENTS 0x40
Hi Will
On 10/10/18 17:31, Will Deacon wrote:
It doesn't make sense for a perf event to be configured as a CHAIN event in isolation, so extend the arm_pmu structure with a ->filter_match() function to allow the backend PMU implementation to reject CHAIN events early.
Cc: stable@vger.kernel.org Signed-off-by: Will Deacon will.deacon@arm.com
arch/arm64/kernel/perf_event.c | 7 +++++++ drivers/perf/arm_pmu.c | 8 +++++++- include/linux/perf/arm_pmu.h | 1 + 3 files changed, 15 insertions(+), 1 deletion(-)
diff --git a/arch/arm64/kernel/perf_event.c b/arch/arm64/kernel/perf_event.c index 8e38d5267f22..e213f8e867f6 100644 --- a/arch/arm64/kernel/perf_event.c +++ b/arch/arm64/kernel/perf_event.c @@ -966,6 +966,12 @@ static int armv8pmu_set_event_filter(struct hw_perf_event *event, return 0; } +static int armv8pmu_filter_match(struct perf_event *event) +{
- unsigned long evtype = event->hw.config_base & ARMV8_PMU_EVTYPE_EVENT;
- return evtype != ARMV8_PMUV3_PERFCTR_CHAIN;
+}
The patch looks correct. I guess we could handle it via the existing map_event(), avoiding another arch specific callback.
Either way,
Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com
On Wed, Oct 10, 2018 at 06:30:13PM +0100, Suzuki K Poulose wrote:
On 10/10/18 17:31, Will Deacon wrote:
It doesn't make sense for a perf event to be configured as a CHAIN event in isolation, so extend the arm_pmu structure with a ->filter_match() function to allow the backend PMU implementation to reject CHAIN events early.
Cc: stable@vger.kernel.org Signed-off-by: Will Deacon will.deacon@arm.com
arch/arm64/kernel/perf_event.c | 7 +++++++ drivers/perf/arm_pmu.c | 8 +++++++- include/linux/perf/arm_pmu.h | 1 + 3 files changed, 15 insertions(+), 1 deletion(-)
diff --git a/arch/arm64/kernel/perf_event.c b/arch/arm64/kernel/perf_event.c index 8e38d5267f22..e213f8e867f6 100644 --- a/arch/arm64/kernel/perf_event.c +++ b/arch/arm64/kernel/perf_event.c @@ -966,6 +966,12 @@ static int armv8pmu_set_event_filter(struct hw_perf_event *event, return 0; } +static int armv8pmu_filter_match(struct perf_event *event) +{
- unsigned long evtype = event->hw.config_base & ARMV8_PMU_EVTYPE_EVENT;
- return evtype != ARMV8_PMUV3_PERFCTR_CHAIN;
+}
The patch looks correct. I guess we could handle it via the existing map_event(), avoiding another arch specific callback.
I initially implemented it using ->map_event(), but that has weird interactions with groups where other events in the group following the CHAIN event will not count, but previous events will. This means that the perf behaviour depends on the order in which you specify the events, which is really confusing!
Either way,
Reviewed-by: Suzuki K Poulose suzuki.poulose@arm.com
Thanks.
Will
linux-stable-mirror@lists.linaro.org
-
Suzuki K Poulose -
Will Deacon