diff options
| author | Fam Zheng <famz@redhat.com> | 2014-11-06 20:44:36 +0800 |
|---|---|---|
| committer | Benjamin LaHaise <bcrl@kvack.org> | 2014-12-13 17:50:20 -0500 |
| commit | 5f785de588735306ec4d7c875caf9d28481c8b21 (patch) | |
| tree | 7f405dcbe7e4b53d554ee6e4040a331a505dcd92 | |
| parent | e4a0d3e720e7e508749c1439b5ba3aff56c92976 (diff) | |
aio: Skip timer for io_getevents if timeout=0
In this case, it is basically a polling. Let's not involve timer at all
because that would hurt performance for application event loops.
In an arbitrary test I've done, io_getevents syscall elapsed time
reduces from 50000+ nanoseconds to a few hundereds.
Signed-off-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Benjamin LaHaise <bcrl@kvack.org>
| -rw-r--r-- | fs/aio.c | 8 |
1 files changed, 6 insertions, 2 deletions
@@ -1253,8 +1253,12 @@ static long read_events(struct kioctx *ctx, long min_nr, long nr, * the ringbuffer empty. So in practice we should be ok, but it's * something to be aware of when touching this code. */ - wait_event_interruptible_hrtimeout(ctx->wait, - aio_read_events(ctx, min_nr, nr, event, &ret), until); + if (until.tv64 == 0) + aio_read_events(ctx, min_nr, nr, event, &ret); + else + wait_event_interruptible_hrtimeout(ctx->wait, + aio_read_events(ctx, min_nr, nr, event, &ret), + until); if (!ret && signal_pending(current)) ret = -EINTR; |