perf/ring_buffer: Use local_try_cmpxchg in __perf_output_begin

Use local_try_cmpxchg instead of local_cmpxchg (*ptr, old, new) == old
in __perf_output_begin.  x86 CMPXCHG instruction returns success in ZF
flag, so this change saves a compare after cmpxchg (and related move
instruction in front of cmpxchg).

Also, try_cmpxchg implicitly assigns old *ptr value to "old" when cmpxchg
fails. There is no need to re-read the value in the loop.

No functional change intended.

Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20230708090048.63046-2-ubizjak@gmail.com
This commit is contained in:
Uros Bizjak 2023-07-08 11:00:37 +02:00 committed by Peter Zijlstra
parent d6b45484c1
commit 1af61adb3a
1 changed files with 3 additions and 2 deletions

View File

@ -191,9 +191,10 @@ __perf_output_begin(struct perf_output_handle *handle,
perf_output_get_handle(handle); perf_output_get_handle(handle);
offset = local_read(&rb->head);
do { do {
head = offset;
tail = READ_ONCE(rb->user_page->data_tail); tail = READ_ONCE(rb->user_page->data_tail);
offset = head = local_read(&rb->head);
if (!rb->overwrite) { if (!rb->overwrite) {
if (unlikely(!ring_buffer_has_space(head, tail, if (unlikely(!ring_buffer_has_space(head, tail,
perf_data_size(rb), perf_data_size(rb),
@ -217,7 +218,7 @@ __perf_output_begin(struct perf_output_handle *handle,
head += size; head += size;
else else
head -= size; head -= size;
} while (local_cmpxchg(&rb->head, offset, head) != offset); } while (!local_try_cmpxchg(&rb->head, &offset, head));
if (backward) { if (backward) {
offset = head; offset = head;