Messages in this thread | | | Date | Thu, 1 Jun 2023 09:59:20 -0400 | Subject | Re: [PATCH v2 0/4] Hybrid event parsing fixes | From | "Liang, Kan" <> |
| |
On 2023-06-01 4:29 a.m., Ian Rogers wrote: > Correct various issues that have come up in mailing list discussions, > the most impactful of which is the wildcard opening of events of type > PERF_TYPE_HARDWARE and PERF_TYPE_HW_CACHE. But also try to avoid a > parse-error without a warning, have more fall-back behaviors and be > liberal with the definition of software events. > > The majority of the change is in correcting the parse-event test to > match the new hybrid parsing. > > Ian Rogers (4): > perf pmu: Correct perf_pmu__auto_merge_stats affecting hybrid > perf evsel: Add verbose 3 print of evsel name when opening > perf parse-events: Wildcard most "numeric" events > perf test: Update parse-events expectations
The patch set fixes the issues I found on a hybrid machine.
Tested-by: Kan Liang <kan.liang@linux.intel.com>
It's better to fold the examples (especially for the patch 1) in the description. So we can understand what's it trying to fix and what's the expected result now.
Thanks, Kan
> > tools/perf/tests/parse-events.c | 1108 ++++++++++++++++--------------- > tools/perf/util/evsel.c | 1 + > tools/perf/util/parse-events.c | 104 ++- > tools/perf/util/parse-events.y | 10 +- > tools/perf/util/pmu.c | 18 +- > tools/perf/util/pmu.h | 5 + > tools/perf/util/pmus.c | 5 + > tools/perf/util/pmus.h | 1 + > 8 files changed, 698 insertions(+), 554 deletions(-) >
| |