Backup jobs may yield prior to installing their handler, because of the job_co_entry shim which guarantees that a job won't begin work until we are ready to start an entire transaction.
Unfortunately, this makes proving correctness about transactional points-in-time for backup hard to reason about. Make it explicitly clear by moving the handler registration to creation time, and changing the write notifier to a no-op until the job is started. Reported-by: Vladimir Sementsov-Ogievskiy <vsement...@virtuozzo.com> Signed-off-by: John Snow <js...@redhat.com> --- block/backup.c | 32 +++++++++++++++++++++++--------- include/qemu/job.h | 5 +++++ job.c | 2 +- 3 files changed, 29 insertions(+), 10 deletions(-) diff --git a/block/backup.c b/block/backup.c index 07d751aea4..4df5b95415 100644 --- a/block/backup.c +++ b/block/backup.c @@ -344,6 +344,13 @@ static int coroutine_fn backup_before_write_notify( assert(QEMU_IS_ALIGNED(req->offset, BDRV_SECTOR_SIZE)); assert(QEMU_IS_ALIGNED(req->bytes, BDRV_SECTOR_SIZE)); + /* The handler is installed at creation time; the actual point-in-time + * starts at job_start(). Transactions guarantee those two points are + * the same point in time. */ + if (!job_started(&job->common.job)) { + return 0; + } + return backup_do_cow(job, req->offset, req->bytes, NULL, true); } @@ -398,6 +405,12 @@ static void backup_clean(Job *job) BackupBlockJob *s = container_of(job, BackupBlockJob, common.job); BlockDriverState *bs = blk_bs(s->common.blk); + /* cancelled before job_start: remove write_notifier */ + if (s->before_write.notify) { + notifier_with_return_remove(&s->before_write); + s->before_write.notify = NULL; + } + if (s->copy_bitmap) { bdrv_release_dirty_bitmap(bs, s->copy_bitmap); s->copy_bitmap = NULL; @@ -527,17 +540,8 @@ static void backup_init_copy_bitmap(BackupBlockJob *job) static int coroutine_fn backup_run(Job *job, Error **errp) { BackupBlockJob *s = container_of(job, BackupBlockJob, common.job); - BlockDriverState *bs = blk_bs(s->common.blk); int ret = 0; - QLIST_INIT(&s->inflight_reqs); - qemu_co_rwlock_init(&s->flush_rwlock); - - backup_init_copy_bitmap(s); - - s->before_write.notify = backup_before_write_notify; - bdrv_add_before_write_notifier(bs, &s->before_write); - if (s->sync_mode == MIRROR_SYNC_MODE_TOP) { int64_t offset = 0; int64_t count; @@ -572,6 +576,7 @@ static int coroutine_fn backup_run(Job *job, Error **errp) out: notifier_with_return_remove(&s->before_write); + s->before_write.notify = NULL; /* wait until pending backup_do_cow() calls have completed */ qemu_co_rwlock_wrlock(&s->flush_rwlock); @@ -767,6 +772,15 @@ BlockJob *backup_job_create(const char *job_id, BlockDriverState *bs, &error_abort); job->len = len; + /* Finally, install a write notifier that takes effect after job_start() */ + backup_init_copy_bitmap(job); + + QLIST_INIT(&job->inflight_reqs); + qemu_co_rwlock_init(&job->flush_rwlock); + + job->before_write.notify = backup_before_write_notify; + bdrv_add_before_write_notifier(bs, &job->before_write); + return &job->common; error: diff --git a/include/qemu/job.h b/include/qemu/job.h index 9e7cd1e4a0..733afb696b 100644 --- a/include/qemu/job.h +++ b/include/qemu/job.h @@ -394,6 +394,11 @@ void job_enter_cond(Job *job, bool(*fn)(Job *job)); */ void job_start(Job *job); +/** + * job_started returns true if the @job has started. + */ +bool job_started(Job *job); + /** * @job: The job to enter. * diff --git a/job.c b/job.c index 28dd48f8a5..745af659ff 100644 --- a/job.c +++ b/job.c @@ -243,7 +243,7 @@ bool job_is_completed(Job *job) return false; } -static bool job_started(Job *job) +bool job_started(Job *job) { return job->co; } -- 2.21.0