[81/81] thread-pool: add missing qemu_bh_cancel in completion function

Submitted by Michael Roth on March 20, 2017, 11:08 p.m.

Details

Message ID 1490051325-3770-82-git-send-email-mdroth@linux.vnet.ibm.com
State New
Headers show

Commit Message

Michael Roth March 20, 2017, 11:08 p.m.
From: Peter Lieven <pl@kamp.de>

commit 3c80ca15 fixed a deadlock scenarion with nested aio_poll invocations.

However, the rescheduling of the completion BH introcuded unnecessary spinning
in the main-loop. On very fast file backends this can even lead to the
"WARNING: I/O thread spun for 1000 iterations" message popping up.

Callgrind reports about 3-4% less instructions with this patch running
qemu-img bench on a ramdisk based VMDK file.

Fixes: 3c80ca158c96ff902a30883a8933e755988948b1
Cc: qemu-stable@nongnu.org
Signed-off-by: Peter Lieven <pl@kamp.de>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
(cherry picked from commit b7a745dc33a18377bb4a8dfe54d1df01ea60bf66)
* drop context dep on b9e413d
Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
---
 thread-pool.c | 7 +++++++
 1 file changed, 7 insertions(+)

Patch hide | download patch | download mbox

diff --git a/thread-pool.c b/thread-pool.c
index 6fba913..4157210 100644
--- a/thread-pool.c
+++ b/thread-pool.c
@@ -185,6 +185,13 @@  restart:
             qemu_bh_schedule(pool->completion_bh);
 
             elem->common.cb(elem->common.opaque, elem->ret);
+
+            /* We can safely cancel the completion_bh here regardless of someone
+             * else having scheduled it meanwhile because we reenter the
+             * completion function anyway (goto restart).
+             */
+            qemu_bh_cancel(pool->completion_bh);
+
             qemu_aio_unref(elem);
             goto restart;
         } else {