diff options
| author | Kajol Jain <kjain@linux.ibm.com> | 2024-11-18 12:41:14 +0100 |
|---|---|---|
| committer | Michael Ellerman <mpe@ellerman.id.au> | 2024-11-19 04:11:30 +0100 |
| commit | f26f9933e3e31b2117b804b6b8932388db88a131 (patch) | |
| tree | cca45a6dd146b18e165bf36d208c10b8b0895748 /arch/powerpc/perf | |
| parent | powerpc/kvm: Add vpa latency counters to kvm_vcpu_arch (diff) | |
| download | linux-f26f9933e3e31b2117b804b6b8932388db88a131.tar.xz linux-f26f9933e3e31b2117b804b6b8932388db88a131.zip | |
powerpc/perf: Add per-task/process monitoring to vpa_pmu driver
Enhance the vpa_pmu driver with a feature to observe context switch
latency event for both per-task (tid) and per-pid (pid) option.
Couple of new helper functions are added to hide the abstraction of
reading the context switch latency counter from kvm_vcpu_arch struct
and these helper functions are defined in the "kvm/book3s_hv.c".
"PERF_ATTACH_TASK" flag is used to decide whether to read the counter
values from lppaca or kvm_vcpu_arch struct.
Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
Co-developed-by: Madhavan Srinivasan <maddy@linux.ibm.com>
Signed-off-by: Madhavan Srinivasan <maddy@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://patch.msgid.link/20241118114114.208964-4-kjain@linux.ibm.com
Diffstat (limited to 'arch/powerpc/perf')
| -rw-r--r-- | arch/powerpc/perf/vpa-pmu.c | 15 |
1 files changed, 12 insertions, 3 deletions
diff --git a/arch/powerpc/perf/vpa-pmu.c b/arch/powerpc/perf/vpa-pmu.c index c143c626b4a0..6a5bfd2a13b5 100644 --- a/arch/powerpc/perf/vpa-pmu.c +++ b/arch/powerpc/perf/vpa-pmu.c @@ -97,13 +97,22 @@ static unsigned long get_counter_data(struct perf_event *event) switch (config) { case L1_TO_L2_CS_LAT: - data = kvmhv_get_l1_to_l2_cs_time(); + if (event->attach_state & PERF_ATTACH_TASK) + data = kvmhv_get_l1_to_l2_cs_time_vcpu(); + else + data = kvmhv_get_l1_to_l2_cs_time(); break; case L2_TO_L1_CS_LAT: - data = kvmhv_get_l2_to_l1_cs_time(); + if (event->attach_state & PERF_ATTACH_TASK) + data = kvmhv_get_l2_to_l1_cs_time_vcpu(); + else + data = kvmhv_get_l2_to_l1_cs_time(); break; case L2_RUNTIME_AGG: - data = kvmhv_get_l2_runtime_agg(); + if (event->attach_state & PERF_ATTACH_TASK) + data = kvmhv_get_l2_runtime_agg_vcpu(); + else + data = kvmhv_get_l2_runtime_agg(); break; default: data = 0; |