* [PATCH] perf: arm_pmuv3: Use BR_RETIRED for HW branch event if enabled
@ 2024-09-05 20:47 Ilkka Koskinen
2024-09-06 4:58 ` Anshuman Khandual
0 siblings, 1 reply; 3+ messages in thread
From: Ilkka Koskinen @ 2024-09-05 20:47 UTC (permalink / raw)
To: Will Deacon, Mark Rutland; +Cc: linux-arm-kernel, linux-kernel, ilkka
The PMU driver attempts to use PC_WRITE_RETIRED for the HW branch event,
if enabled. However, PC_WRITE_RETIRED counts only taken branches,
whereas BR_RETIRED counts also non-taken ones.
Furthermore, perf uses HW branch event to calculate branch misses ratio,
implying BR_RETIRED is the correct event to count.
Signed-off-by: Ilkka Koskinen <ilkka@os.amperecomputing.com>
---
drivers/perf/arm_pmuv3.c | 27 ++++-----------------------
1 file changed, 4 insertions(+), 23 deletions(-)
diff --git a/drivers/perf/arm_pmuv3.c b/drivers/perf/arm_pmuv3.c
index d246840797b6..a8ed08df1411 100644
--- a/drivers/perf/arm_pmuv3.c
+++ b/drivers/perf/arm_pmuv3.c
@@ -46,6 +46,7 @@ static const unsigned armv8_pmuv3_perf_map[PERF_COUNT_HW_MAX] = {
[PERF_COUNT_HW_INSTRUCTIONS] = ARMV8_PMUV3_PERFCTR_INST_RETIRED,
[PERF_COUNT_HW_CACHE_REFERENCES] = ARMV8_PMUV3_PERFCTR_L1D_CACHE,
[PERF_COUNT_HW_CACHE_MISSES] = ARMV8_PMUV3_PERFCTR_L1D_CACHE_REFILL,
+ [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = ARMV8_PMUV3_PERFCTR_BR_RETIRED,
[PERF_COUNT_HW_BRANCH_MISSES] = ARMV8_PMUV3_PERFCTR_BR_MIS_PRED,
[PERF_COUNT_HW_BUS_CYCLES] = ARMV8_PMUV3_PERFCTR_BUS_CYCLES,
[PERF_COUNT_HW_STALLED_CYCLES_FRONTEND] = ARMV8_PMUV3_PERFCTR_STALL_FRONTEND,
@@ -1083,28 +1084,6 @@ static void armv8pmu_reset(void *info)
armv8pmu_pmcr_write(pmcr);
}
-static int __armv8_pmuv3_map_event_id(struct arm_pmu *armpmu,
- struct perf_event *event)
-{
- if (event->attr.type == PERF_TYPE_HARDWARE &&
- event->attr.config == PERF_COUNT_HW_BRANCH_INSTRUCTIONS) {
-
- if (test_bit(ARMV8_PMUV3_PERFCTR_PC_WRITE_RETIRED,
- armpmu->pmceid_bitmap))
- return ARMV8_PMUV3_PERFCTR_PC_WRITE_RETIRED;
-
- if (test_bit(ARMV8_PMUV3_PERFCTR_BR_RETIRED,
- armpmu->pmceid_bitmap))
- return ARMV8_PMUV3_PERFCTR_BR_RETIRED;
-
- return HW_OP_UNSUPPORTED;
- }
-
- return armpmu_map_event(event, &armv8_pmuv3_perf_map,
- &armv8_pmuv3_perf_cache_map,
- ARMV8_PMU_EVTYPE_EVENT);
-}
-
static int __armv8_pmuv3_map_event(struct perf_event *event,
const unsigned (*extra_event_map)
[PERF_COUNT_HW_MAX],
@@ -1116,7 +1095,9 @@ static int __armv8_pmuv3_map_event(struct perf_event *event,
int hw_event_id;
struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
- hw_event_id = __armv8_pmuv3_map_event_id(armpmu, event);
+ hw_event_id = armpmu_map_event(event, &armv8_pmuv3_perf_map,
+ &armv8_pmuv3_perf_cache_map,
+ ARMV8_PMU_EVTYPE_EVENT);
/*
* CHAIN events only work when paired with an adjacent counter, and it
--
2.42.0
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] perf: arm_pmuv3: Use BR_RETIRED for HW branch event if enabled
2024-09-05 20:47 [PATCH] perf: arm_pmuv3: Use BR_RETIRED for HW branch event if enabled Ilkka Koskinen
@ 2024-09-06 4:58 ` Anshuman Khandual
2024-09-06 18:28 ` Ilkka Koskinen
0 siblings, 1 reply; 3+ messages in thread
From: Anshuman Khandual @ 2024-09-06 4:58 UTC (permalink / raw)
To: Ilkka Koskinen, Will Deacon, Mark Rutland; +Cc: linux-arm-kernel, linux-kernel
On 9/6/24 02:17, Ilkka Koskinen wrote:
> The PMU driver attempts to use PC_WRITE_RETIRED for the HW branch event,
> if enabled. However, PC_WRITE_RETIRED counts only taken branches,
> whereas BR_RETIRED counts also non-taken ones.
>
> Furthermore, perf uses HW branch event to calculate branch misses ratio,
> implying BR_RETIRED is the correct event to count.
But is the event BR_RETIRED always guaranteed to be available. Should not
armpmu->pmceid_bitmap be checked first ?
>
> Signed-off-by: Ilkka Koskinen <ilkka@os.amperecomputing.com>
> ---
> drivers/perf/arm_pmuv3.c | 27 ++++-----------------------
> 1 file changed, 4 insertions(+), 23 deletions(-)
>
> diff --git a/drivers/perf/arm_pmuv3.c b/drivers/perf/arm_pmuv3.c
> index d246840797b6..a8ed08df1411 100644
> --- a/drivers/perf/arm_pmuv3.c
> +++ b/drivers/perf/arm_pmuv3.c
> @@ -46,6 +46,7 @@ static const unsigned armv8_pmuv3_perf_map[PERF_COUNT_HW_MAX] = {
> [PERF_COUNT_HW_INSTRUCTIONS] = ARMV8_PMUV3_PERFCTR_INST_RETIRED,
> [PERF_COUNT_HW_CACHE_REFERENCES] = ARMV8_PMUV3_PERFCTR_L1D_CACHE,
> [PERF_COUNT_HW_CACHE_MISSES] = ARMV8_PMUV3_PERFCTR_L1D_CACHE_REFILL,
> + [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = ARMV8_PMUV3_PERFCTR_BR_RETIRED,
> [PERF_COUNT_HW_BRANCH_MISSES] = ARMV8_PMUV3_PERFCTR_BR_MIS_PRED,
> [PERF_COUNT_HW_BUS_CYCLES] = ARMV8_PMUV3_PERFCTR_BUS_CYCLES,
> [PERF_COUNT_HW_STALLED_CYCLES_FRONTEND] = ARMV8_PMUV3_PERFCTR_STALL_FRONTEND,
> @@ -1083,28 +1084,6 @@ static void armv8pmu_reset(void *info)
> armv8pmu_pmcr_write(pmcr);
> }
>
> -static int __armv8_pmuv3_map_event_id(struct arm_pmu *armpmu,
> - struct perf_event *event)
> -{
> - if (event->attr.type == PERF_TYPE_HARDWARE &&
> - event->attr.config == PERF_COUNT_HW_BRANCH_INSTRUCTIONS) {
> -
> - if (test_bit(ARMV8_PMUV3_PERFCTR_PC_WRITE_RETIRED,
> - armpmu->pmceid_bitmap))
> - return ARMV8_PMUV3_PERFCTR_PC_WRITE_RETIRED;
> -
> - if (test_bit(ARMV8_PMUV3_PERFCTR_BR_RETIRED,
> - armpmu->pmceid_bitmap))
> - return ARMV8_PMUV3_PERFCTR_BR_RETIRED;
If BR_RETIRED event is absent on the platform, PC_WRITE_RETIRED still remains
a good alternative to fallback on. Hence wondering if the above order could
just be changed to use BR_RETIRED first when available.
> -
> - return HW_OP_UNSUPPORTED;
> - }
> -
> - return armpmu_map_event(event, &armv8_pmuv3_perf_map,
> - &armv8_pmuv3_perf_cache_map,
> - ARMV8_PMU_EVTYPE_EVENT);
> -}
> -
> static int __armv8_pmuv3_map_event(struct perf_event *event,
> const unsigned (*extra_event_map)
> [PERF_COUNT_HW_MAX],
> @@ -1116,7 +1095,9 @@ static int __armv8_pmuv3_map_event(struct perf_event *event,
> int hw_event_id;
> struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
>
> - hw_event_id = __armv8_pmuv3_map_event_id(armpmu, event);
> + hw_event_id = armpmu_map_event(event, &armv8_pmuv3_perf_map,
> + &armv8_pmuv3_perf_cache_map,
> + ARMV8_PMU_EVTYPE_EVENT);
>
> /*
> * CHAIN events only work when paired with an adjacent counter, and it
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] perf: arm_pmuv3: Use BR_RETIRED for HW branch event if enabled
2024-09-06 4:58 ` Anshuman Khandual
@ 2024-09-06 18:28 ` Ilkka Koskinen
0 siblings, 0 replies; 3+ messages in thread
From: Ilkka Koskinen @ 2024-09-06 18:28 UTC (permalink / raw)
To: Anshuman Khandual
Cc: Ilkka Koskinen, Will Deacon, Mark Rutland, linux-arm-kernel,
linux-kernel
On Fri, 6 Sep 2024, Anshuman Khandual wrote:
> On 9/6/24 02:17, Ilkka Koskinen wrote:
>> The PMU driver attempts to use PC_WRITE_RETIRED for the HW branch event,
>> if enabled. However, PC_WRITE_RETIRED counts only taken branches,
>> whereas BR_RETIRED counts also non-taken ones.
>>
>> Furthermore, perf uses HW branch event to calculate branch misses ratio,
>> implying BR_RETIRED is the correct event to count.
> But is the event BR_RETIRED always guaranteed to be available. Should not
> armpmu->pmceid_bitmap be checked first ?
>
>>
>> Signed-off-by: Ilkka Koskinen <ilkka@os.amperecomputing.com>
>> ---
>> drivers/perf/arm_pmuv3.c | 27 ++++-----------------------
>> 1 file changed, 4 insertions(+), 23 deletions(-)
>>
>> diff --git a/drivers/perf/arm_pmuv3.c b/drivers/perf/arm_pmuv3.c
>> index d246840797b6..a8ed08df1411 100644
>> --- a/drivers/perf/arm_pmuv3.c
>> +++ b/drivers/perf/arm_pmuv3.c
>> @@ -46,6 +46,7 @@ static const unsigned armv8_pmuv3_perf_map[PERF_COUNT_HW_MAX] = {
>> [PERF_COUNT_HW_INSTRUCTIONS] = ARMV8_PMUV3_PERFCTR_INST_RETIRED,
>> [PERF_COUNT_HW_CACHE_REFERENCES] = ARMV8_PMUV3_PERFCTR_L1D_CACHE,
>> [PERF_COUNT_HW_CACHE_MISSES] = ARMV8_PMUV3_PERFCTR_L1D_CACHE_REFILL,
>> + [PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = ARMV8_PMUV3_PERFCTR_BR_RETIRED,
>> [PERF_COUNT_HW_BRANCH_MISSES] = ARMV8_PMUV3_PERFCTR_BR_MIS_PRED,
>> [PERF_COUNT_HW_BUS_CYCLES] = ARMV8_PMUV3_PERFCTR_BUS_CYCLES,
>> [PERF_COUNT_HW_STALLED_CYCLES_FRONTEND] = ARMV8_PMUV3_PERFCTR_STALL_FRONTEND,
>> @@ -1083,28 +1084,6 @@ static void armv8pmu_reset(void *info)
>> armv8pmu_pmcr_write(pmcr);
>> }
>>
>> -static int __armv8_pmuv3_map_event_id(struct arm_pmu *armpmu,
>> - struct perf_event *event)
>> -{
>> - if (event->attr.type == PERF_TYPE_HARDWARE &&
>> - event->attr.config == PERF_COUNT_HW_BRANCH_INSTRUCTIONS) {
>> -
>> - if (test_bit(ARMV8_PMUV3_PERFCTR_PC_WRITE_RETIRED,
>> - armpmu->pmceid_bitmap))
>> - return ARMV8_PMUV3_PERFCTR_PC_WRITE_RETIRED;
>> -
>> - if (test_bit(ARMV8_PMUV3_PERFCTR_BR_RETIRED,
>> - armpmu->pmceid_bitmap))
>> - return ARMV8_PMUV3_PERFCTR_BR_RETIRED;
>
> If BR_RETIRED event is absent on the platform, PC_WRITE_RETIRED still remains
> a good alternative to fallback on. Hence wondering if the above order could
> just be changed to use BR_RETIRED first when available.
If PC_WRITE_RETIRED is a good alternative, then I guess, it's just better
change the order of those two. I send a new version just doing that.
Cheers, Ilkka
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2024-09-06 18:28 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-09-05 20:47 [PATCH] perf: arm_pmuv3: Use BR_RETIRED for HW branch event if enabled Ilkka Koskinen
2024-09-06 4:58 ` Anshuman Khandual
2024-09-06 18:28 ` Ilkka Koskinen
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®