From: Stefan Reiter <s.rei...@proxmox.com> All code-paths leading to backup_clean (via job_clean) have the job's context already acquired. The job's context is guaranteed to be the same as the one used by backup_top via backup_job_create.
Since the previous logic effectively acquired the lock twice, this broke cleanup of backups for disks using IO threads, since the BDRV_POLL_WHILE in bdrv_backup_top_drop -> bdrv_do_drained_begin would only release the lock once, thus deadlocking with the IO thread. This is a partial revert of 0abf2581717a19. Signed-off-by: Stefan Reiter <s.rei...@proxmox.com> Reviewed-by: Max Reitz <mre...@redhat.com> Message-Id: <20200407115651.69472-4-s.rei...@proxmox.com> Signed-off-by: Kevin Wolf <kw...@redhat.com> --- block/backup.c | 4 ---- 1 file changed, 4 deletions(-) diff --git a/block/backup.c b/block/backup.c index 7430ca5883..a7a7dcaf4c 100644 --- a/block/backup.c +++ b/block/backup.c @@ -126,11 +126,7 @@ static void backup_abort(Job *job) static void backup_clean(Job *job) { BackupBlockJob *s = container_of(job, BackupBlockJob, common.job); - AioContext *aio_context = bdrv_get_aio_context(s->backup_top); - - aio_context_acquire(aio_context); bdrv_backup_top_drop(s->backup_top); - aio_context_release(aio_context); } void backup_do_checkpoint(BlockJob *job, Error **errp) -- 2.20.1