diff options
| author | Ian Rogers <[email protected]> | 2025-01-13 19:43:45 +0000 |
|---|---|---|
| committer | Namhyung Kim <[email protected]> | 2025-02-13 04:06:11 +0000 |
| commit | dc6d2bc2d893a878e7b58578ff01b4738708deb4 (patch) | |
| tree | b5524ff4429f1e88533f518753091462bc4bd8b4 /tools/perf/tests/switch-tracking.c | |
| parent | perf test stat_all_metrics: Ensure missing events fail test (diff) | |
| download | kernel-dc6d2bc2d893a878e7b58578ff01b4738708deb4.tar.gz kernel-dc6d2bc2d893a878e7b58578ff01b4738708deb4.zip | |
perf sample: Make user_regs and intr_regs optional
The struct dump_regs contains 512 bytes of cache_regs, meaning the two
values in perf_sample contribute 1088 bytes of its total 1384 bytes
size. Initializing this much memory has a cost reported by Tavian
Barnes <[email protected]> as about 2.5% when running `perf
script --itrace=i0`:
https://lore.kernel.org/lkml/d841b97b3ad2ca8bcab07e4293375fb7c32dfce7.1736618095.git.tavianator@tavianator.com/
Adrian Hunter <[email protected]> replied that the zero
initialization was necessary and couldn't simply be removed.
This patch aims to strike a middle ground of still zeroing the
perf_sample, but removing 79% of its size by make user_regs and
intr_regs optional pointers to zalloc-ed memory. To support the
allocation accessors are created for user_regs and intr_regs. To
support correct cleanup perf_sample__init and perf_sample__exit
functions are created and added throughout the code base.
Signed-off-by: Ian Rogers <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Namhyung Kim <[email protected]>
Diffstat (limited to 'tools/perf/tests/switch-tracking.c')
| -rw-r--r-- | tools/perf/tests/switch-tracking.c | 14 |
1 files changed, 10 insertions, 4 deletions
diff --git a/tools/perf/tests/switch-tracking.c b/tools/perf/tests/switch-tracking.c index 576f82a15015..8df3f9d9ffd2 100644 --- a/tools/perf/tests/switch-tracking.c +++ b/tools/perf/tests/switch-tracking.c @@ -131,9 +131,11 @@ static int process_sample_event(struct evlist *evlist, pid_t next_tid, prev_tid; int cpu, err; + perf_sample__init(&sample, /*all=*/false); if (evlist__parse_sample(evlist, event, &sample)) { pr_debug("evlist__parse_sample failed\n"); - return -1; + err = -1; + goto out; } evsel = evlist__id2evsel(evlist, sample.id); @@ -145,7 +147,7 @@ static int process_sample_event(struct evlist *evlist, cpu, prev_tid, next_tid); err = check_cpu(switch_tracking, cpu); if (err) - return err; + goto out; /* * Check for no missing sched_switch events i.e. that the * evsel->core.system_wide flag has worked. @@ -153,7 +155,8 @@ static int process_sample_event(struct evlist *evlist, if (switch_tracking->tids[cpu] != -1 && switch_tracking->tids[cpu] != prev_tid) { pr_debug("Missing sched_switch events\n"); - return -1; + err = -1; + goto out; } switch_tracking->tids[cpu] = next_tid; } @@ -169,7 +172,10 @@ static int process_sample_event(struct evlist *evlist, switch_tracking->cycles_after_comm_4 = 1; } - return 0; + err = 0; +out: + perf_sample__exit(&sample); + return err; } static int process_event(struct evlist *evlist, union perf_event *event, |