Commit 36cddd10 authored by Namhyung Kim's avatar Namhyung Kim Committed by Arnaldo Carvalho de Melo
Browse files

perf lock contention: Do not fail EEXIST for update 



When it updates the lock stat for the first time, it needs to create an
element in the BPF hash map.

But if there's a concurrent thread waiting for the same lock (like for
rwsem or rwlock), it might race with the thread and possibly fail to
update with -EEXIST.

In that case, it can lookup the map again and put the data there instead
of failing.

Signed-off-by: default avatarNamhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Song Liu <song@kernel.org>
Cc: bpf@vger.kernel.org
Link: https://lore.kernel.org/r/20240830065150.1758962-3-namhyung@kernel.org


Signed-off-by: default avatarArnaldo Carvalho de Melo <acme@redhat.com>
parent 05a5dd1d

tools/perf/util/bpf_skel/lock_contention.bpf.c

+7 −0
Original line number Diff line number Diff line
@@ -491,6 +491,12 @@ int contention_end(u64 *ctx) 


		err = bpf_map_update_elem(&lock_stat, &key, &first, BPF_NOEXIST);

		if (err < 0) {

			if (err == -EEXIST) {

				/* it lost the race, try to get it again */

				data = bpf_map_lookup_elem(&lock_stat, &key);

				if (data != NULL)

					goto found;

			}

			if (err == -E2BIG)

				data_map_full = 1;

			__sync_fetch_and_add(&data_fail, 1);
@@ -498,6 +504,7 @@ int contention_end(u64 *ctx) 
		goto out;

	}



found:

	__sync_fetch_and_add(&data->total_time, duration);

	__sync_fetch_and_add(&data->count, 1);