ceph: try to queue a writeback if revoking fails

If the pagecaches writeback just finished and the i_wrbuffer_ref
reaches zero it will try to trigger ceph_check_caps(). But if just
before ceph_check_caps() the i_wrbuffer_ref could be increased
again by mmap/cache write, then the Fwb revoke will fail.

We need to try to queue a writeback in this case instead of
triggering the writeback by BDI's delayed work per 5 seconds.

URL: https://tracker.ceph.com/issues/46904
URL: https://tracker.ceph.com/issues/55377
Signed-off-by: Xiubo Li <xiubli@redhat.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

1 files changed, 24 insertions, 4 deletions

diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index a051e6e4d7ca..bf2e94005598 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -1911,6 +1911,7 @@ void ceph_check_caps(struct ceph_inode_info *ci, int flags,
 	struct rb_node *p;
 	bool queue_invalidate = false;
 	bool tried_invalidate = false;
+	bool queue_writeback = false;
 
 	if (session)
 		ceph_get_mds_session(session);
@@ -2063,10 +2064,27 @@ retry:
 		}
 
 		/* completed revocation? going down and there are no caps? */
-		if (revoking && (revoking & cap_used) == 0) {
-			dout("completed revocation of %s\n",
-			     ceph_cap_string(cap->implemented & ~cap->issued));
-			goto ack;
+		if (revoking) {
+			if ((revoking & cap_used) == 0) {
+				dout("completed revocation of %s\n",
+				      ceph_cap_string(cap->implemented & ~cap->issued));
+				goto ack;
+			}
+
+			/*
+			 * If the "i_wrbuffer_ref" was increased by mmap or generic
+			 * cache write just before the ceph_check_caps() is called,
+			 * the Fb capability revoking will fail this time. Then we
+			 * must wait for the BDI's delayed work to flush the dirty
+			 * pages and to release the "i_wrbuffer_ref", which will cost
+			 * at most 5 seconds. That means the MDS needs to wait at
+			 * most 5 seconds to finished the Fb capability's revocation.
+			 *
+			 * Let's queue a writeback for it.
+			 */
+			if (S_ISREG(inode->i_mode) && ci->i_wrbuffer_ref &&
+			    (revoking & CEPH_CAP_FILE_BUFFER))
+				queue_writeback = true;
 		}
 
 		/* want more caps from mds? */
@@ -2136,6 +2154,8 @@ ack:
 	spin_unlock(&ci->i_ceph_lock);
 
 	ceph_put_mds_session(session);
+	if (queue_writeback)
+		ceph_queue_writeback(inode);
 	if (queue_invalidate)
 		ceph_queue_invalidate(inode);
 }