On Thu, Jun 09, 2022 at 05:47:11PM +0100, Stefan Hajnoczi wrote:
Every laio_io_plug() call has a matching laio_io_unplug() call. There is
a plugged counter that tracks the number of levels of plugging and
allows for nesting.
The plugged counter must reflect the balance between laio_io_plug() and
laio_io_unplug() calls accurately. Otherwise I/O stalls occur since
io_submit(2) calls are skipped while plugged.
We can add a Fixes tag:
Fixes: 68d7946648 ("linux-aio: add `dev_max_batch` parameter to
laio_io_unplug()")
Reported-by: Nikolay Tenev <n...@storpool.com>
Cc: Stefano Garzarella <sgarz...@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefa...@redhat.com>
---
block/linux-aio.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/block/linux-aio.c b/block/linux-aio.c
index 4c423fcccf..6078da7e42 100644
--- a/block/linux-aio.c
+++ b/block/linux-aio.c
@@ -363,8 +363,10 @@ void laio_io_unplug(BlockDriverState *bs, LinuxAioState *s,
uint64_t dev_max_batch)
{
assert(s->io_q.plugged);
+ s->io_q.plugged--;
+
if (s->io_q.in_queue >= laio_max_batch(s, dev_max_batch) ||
- (--s->io_q.plugged == 0 &&
+ (!s->io_q.plugged &&
!s->io_q.blocked && !QSIMPLEQ_EMPTY(&s->io_q.pending))) {
ioq_submit(s);
}
--
2.36.1
Thanks for fixing this issue!
Reviewed-by: Stefano Garzarella <sgarz...@redhat.com>
