diff mbox series

[PULL,3/5] aio-posix: do skip system call if ctx->notifier polling succeeds

Message ID 20180926025526.26961-4-famz@redhat.com
State New
Headers show
Series [PULL,1/5] aio-posix: fix concurrent access to poll_disable_cnt | expand

Commit Message

Fam Zheng Sept. 26, 2018, 2:55 a.m. UTC 
From: Paolo Bonzini <pbonzini@redhat.com>

Commit 70232b5253 ("aio-posix: Don't count ctx->notifier as progress when
2018-08-15), by not reporting progress, causes aio_poll to execute the
system call when polling succeeds because of ctx->notifier.  This introduces
latency before the call to aio_bh_poll() and negates the advantages of
polling, unfortunately.

The fix builds on the previous patch, separating the effect of polling on
the timeout from the progress reported to aio_poll().  ctx->notifier
does zero the timeout, causing the caller to skip the system call,
but it does not report progress, so that the bug fix of commit 70232b5253
still stands.

Fixes: 70232b5253a3c4e03ed1ac47ef9246a8ac66c6fa
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20180912171040.1732-4-pbonzini@redhat.com>
Reviewed-by: Fam Zheng <famz@redhat.com>
Signed-off-by: Fam Zheng <famz@redhat.com>
---
 util/aio-posix.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
diff mbox series

Patch

diff --git a/util/aio-posix.c b/util/aio-posix.c
index c54d1ebfb1..621b3025d8 100644
--- a/util/aio-posix.c
+++ b/util/aio-posix.c
@@ -498,10 +498,11 @@  static bool run_poll_handlers_once(AioContext *ctx, int64_t *timeout)
     QLIST_FOREACH_RCU(node, &ctx->aio_handlers, node) {
         if (!node->deleted && node->io_poll &&
             aio_node_check(ctx, node->is_external) &&
-            node->io_poll(node->opaque) &&
-            node->opaque != &ctx->notifier) {
+            node->io_poll(node->opaque)) {
             *timeout = 0;
-            progress = true;
+            if (node->opaque != &ctx->notifier) {
+                progress = true;
+            }
         }
 
         /* Caller handles freeing deleted nodes.  Don't do it here. */