lkml.org 
[lkml]   [2022]   [Feb]   [4]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
Subject[PATCH] perf/arm64: fix mapping for HW_BRANCH_INSTRUCTIONS on PMUv3
From
With the existing code, the following command:

$ perf stat -e branches sleep 0
Performance counter stats for 'sleep 0':
<not supported> branches

on N1 core (pmuv3).

This is due to the fact that the mapping for the generic event is wrong.
It is using ARMV8_PMUV3_PERFCTR_PC_WRITE_RETIRED which is not implemented
on N1 (and most likely on any PMUv3 implementations). However, there is
another supported event ARMV8_PMUV3_PERFCTR_BR_RETIRED measuring the same
condition.

This patch switches the mapping to ARMV8_PMUV3_PERFCTR_BR_RETIRED so that
the perf stat command above works.

Signed-off-by: Stephane Eranian <eranian@google.com>
---
arch/arm64/kernel/perf_event.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm64/kernel/perf_event.c b/arch/arm64/kernel/perf_event.c
index cab678ed6618..ec2b98343a0b 100644
--- a/arch/arm64/kernel/perf_event.c
+++ b/arch/arm64/kernel/perf_event.c
@@ -45,7 +45,7 @@ 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_PC_WRITE_RETIRED,
+ [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,
--
2.35.0.263.gb82422642f-goog
\
 
 \ /
  Last update: 2022-02-04 08:40    [W:0.036 / U:0.336 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site