]> www.infradead.org Git - users/jedix/linux-maple.git/commit
perf/x86/intel/ds: Fix non 0 retire latency on Raptorlake
authorKan Liang <kan.liang@linux.intel.com>
Mon, 8 Jul 2024 19:33:36 +0000 (12:33 -0700)
committerPeter Zijlstra <peterz@infradead.org>
Tue, 9 Jul 2024 11:26:39 +0000 (13:26 +0200)
commite5f32ad56b22ebe384a6e7ddad6e9520c5495563
treeddc98776788318bca8a68b58f541342698364600
parent556a7c039a52c21da33eaae9269984a1ef59189b
perf/x86/intel/ds: Fix non 0 retire latency on Raptorlake

A non-0 retire latency can be observed on a Raptorlake which doesn't
support the retire latency feature.
By design, the retire latency shares the PERF_SAMPLE_WEIGHT_STRUCT
sample type with other types of latency. That could avoid adding too
many different sample types to support all kinds of latency. For the
machine which doesn't support some kind of latency, 0 should be
returned.

Perf doesn’t clear/init all the fields of a sample data for the sake
of performance. It expects the later perf_{prepare,output}_sample() to
update the uninitialized field. However, the current implementation
doesn't touch the field of the retire latency if the feature is not
supported. The memory garbage is dumped into the perf data.

Clear the retire latency if the feature is not supported.

Fixes: c87a31093c70 ("perf/x86: Support Retire Latency")
Reported-by: "Bayduraev, Alexey V" <alexey.v.bayduraev@intel.com>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: "Bayduraev, Alexey V" <alexey.v.bayduraev@intel.com>
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/20240708193336.1192217-4-kan.liang@linux.intel.com
arch/x86/events/intel/ds.c