From: Ilya Dryomov <idryomov@xxxxxxxxx> commit 870611e4877eff1e8413c3fb92a585e45d5291f6 upstream. Move capturing the snapshot context into the image request state machine, after exclusive lock is ensured to be held for the duration of dealing with the image request. This is needed to ensure correctness of fast-diff states (OBJECT_EXISTS vs OBJECT_EXISTS_CLEAN) and object deltas computed based off of them. Otherwise the object map that is forked for the snapshot isn't guaranteed to accurately reflect the contents of the snapshot when the snapshot is taken under I/O. This breaks differential backup and snapshot-based mirroring use cases with fast-diff enabled: since some object deltas may be incomplete, the destination image may get corrupted. Cc: stable@xxxxxxxxxxxxxxx Link: https://tracker.ceph.com/issues/61472 Signed-off-by: Ilya Dryomov <idryomov@xxxxxxxxx> Reviewed-by: Dongsheng Yang <dongsheng.yang@xxxxxxxxxxxx> Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> --- drivers/block/rbd.c | 30 +++++++++++++++++++++++------- 1 file changed, 23 insertions(+), 7 deletions(-) --- a/drivers/block/rbd.c +++ b/drivers/block/rbd.c @@ -1337,6 +1337,8 @@ static bool rbd_obj_is_tail(struct rbd_o */ static void rbd_obj_set_copyup_enabled(struct rbd_obj_request *obj_req) { + rbd_assert(obj_req->img_request->snapc); + if (obj_req->img_request->op_type == OBJ_OP_DISCARD) { dout("%s %p objno %llu discard\n", __func__, obj_req, obj_req->ex.oe_objno); @@ -1457,6 +1459,7 @@ __rbd_obj_add_osd_request(struct rbd_obj static struct ceph_osd_request * rbd_obj_add_osd_request(struct rbd_obj_request *obj_req, int num_ops) { + rbd_assert(obj_req->img_request->snapc); return __rbd_obj_add_osd_request(obj_req, obj_req->img_request->snapc, num_ops); } @@ -1593,15 +1596,18 @@ static void rbd_img_request_init(struct mutex_init(&img_request->state_mutex); } +/* + * Only snap_id is captured here, for reads. For writes, snapshot + * context is captured in rbd_img_object_requests() after exclusive + * lock is ensured to be held. + */ static void rbd_img_capture_header(struct rbd_img_request *img_req) { struct rbd_device *rbd_dev = img_req->rbd_dev; lockdep_assert_held(&rbd_dev->header_rwsem); - if (rbd_img_is_write(img_req)) - img_req->snapc = ceph_get_snap_context(rbd_dev->header.snapc); - else + if (!rbd_img_is_write(img_req)) img_req->snap_id = rbd_dev->spec->snap_id; if (rbd_dev_parent_get(rbd_dev)) @@ -3484,9 +3490,19 @@ static int rbd_img_exclusive_lock(struct static void rbd_img_object_requests(struct rbd_img_request *img_req) { + struct rbd_device *rbd_dev = img_req->rbd_dev; struct rbd_obj_request *obj_req; rbd_assert(!img_req->pending.result && !img_req->pending.num_pending); + rbd_assert(!need_exclusive_lock(img_req) || + __rbd_is_lock_owner(rbd_dev)); + + if (rbd_img_is_write(img_req)) { + rbd_assert(!img_req->snapc); + down_read(&rbd_dev->header_rwsem); + img_req->snapc = ceph_get_snap_context(rbd_dev->header.snapc); + up_read(&rbd_dev->header_rwsem); + } for_each_obj_request(img_req, obj_req) { int result = 0; @@ -3504,7 +3520,6 @@ static void rbd_img_object_requests(stru static bool rbd_img_advance(struct rbd_img_request *img_req, int *result) { - struct rbd_device *rbd_dev = img_req->rbd_dev; int ret; again: @@ -3525,9 +3540,6 @@ again: if (*result) return true; - rbd_assert(!need_exclusive_lock(img_req) || - __rbd_is_lock_owner(rbd_dev)); - rbd_img_object_requests(img_req); if (!img_req->pending.num_pending) { *result = img_req->pending.result; @@ -3989,6 +4001,10 @@ static int rbd_post_acquire_action(struc { int ret; + ret = rbd_dev_refresh(rbd_dev); + if (ret) + return ret; + if (rbd_dev->header.features & RBD_FEATURE_OBJECT_MAP) { ret = rbd_object_map_open(rbd_dev); if (ret)
