]> xenbits.xensource.com Git - qemu-xen-unstable.git/commitdiff
block/mirror: Make cancel always cancel pre-READY
authorMax Reitz <mreitz@redhat.com>
Tue, 1 May 2018 22:05:08 +0000 (00:05 +0200)
committerJeff Cody <jcody@redhat.com>
Tue, 8 May 2018 14:47:27 +0000 (10:47 -0400)
Commit b76e4458b1eb3c32e9824fe6aa51f67d2b251748 made the mirror block
job respect block-job-cancel's @force flag: With that flag set, it would
now always really cancel, even post-READY.

Unfortunately, it had a side effect: Without that flag set, it would now
never cancel, not even before READY.  Considering that is an
incompatible change and not noted anywhere in the commit or the
description of block-job-cancel's @force parameter, this seems
unintentional and we should revert to the previous behavior, which is to
immediately cancel the job when block-job-cancel is called before source
and target are in sync (i.e. before the READY event).

Cc: qemu-stable@nongnu.org
Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=1572856
Reported-by: Yanan Fu <yfu@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-id: 20180501220509.14152-2-mreitz@redhat.com
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Jeff Cody <jcody@redhat.com>
block/mirror.c

index 9436a8d5ee0d4fe0e79a44448c02efdf71b67725..99da9c0858b8951b6e9385584ec2699cff0af1ed 100644 (file)
@@ -874,7 +874,9 @@ static void coroutine_fn mirror_run(void *opaque)
         }
         trace_mirror_before_sleep(s, cnt, s->synced, delay_ns);
         block_job_sleep_ns(&s->common, delay_ns);
-        if (block_job_is_cancelled(&s->common) && s->common.force) {
+        if (block_job_is_cancelled(&s->common) &&
+            (!s->synced || s->common.force))
+        {
             break;
         }
         s->last_pause_ns = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);