4 files changed, 106 insertions, 35 deletions

diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index ad1f46c66f..7fb4aae974 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -2156,6 +2156,30 @@ retry:
 		      ceph_cap_string(cap->implemented),
 		      ceph_cap_string(revoking));
 
+		/* completed revocation? going down and there are no caps? */
+		if (revoking) {
+			if ((revoking & cap_used) == 0) {
+				doutc(cl, "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;
+		}
+
 		if (cap == ci->i_auth_cap &&
 		    (cap->issued & CEPH_CAP_FILE_WR)) {
 			/* request larger max_size from MDS? */
@@ -2183,30 +2207,6 @@ retry:
 			}
 		}
 
-		/* completed revocation? going down and there are no caps? */
-		if (revoking) {
-			if ((revoking & cap_used) == 0) {
-				doutc(cl, "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? */
 		if (want & ~cap->mds_wanted) {
 			if (want & ~(cap->mds_wanted | cap->issued))
@@ -4772,7 +4772,22 @@ int ceph_drop_caps_for_unlink(struct inode *inode)
 		if (__ceph_caps_dirty(ci)) {
 			struct ceph_mds_client *mdsc =
 				ceph_inode_to_fs_client(inode)->mdsc;
-			__cap_delay_requeue_front(mdsc, ci);
+
+			doutc(mdsc->fsc->client, "%p %llx.%llx\n", inode,
+			      ceph_vinop(inode));
+			spin_lock(&mdsc->cap_unlink_delay_lock);
+			ci->i_ceph_flags |= CEPH_I_FLUSH;
+			if (!list_empty(&ci->i_cap_delay_list))
+				list_del_init(&ci->i_cap_delay_list);
+			list_add_tail(&ci->i_cap_delay_list,
+				      &mdsc->cap_unlink_delay_list);
+			spin_unlock(&mdsc->cap_unlink_delay_lock);
+
+			/*
+			 * Fire the work immediately, because the MDS maybe
+			 * waiting for caps release.
+			 */
+			ceph_queue_cap_unlink_work(mdsc);
 		}
 	}
 	spin_unlock(&ci->i_ceph_lock);
diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index 3b5aae29e9..523debc6f2 100644
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -1135,7 +1135,12 @@ ssize_t __ceph_sync_read(struct inode *inode, loff_t *ki_pos,
 		}
 
 		idx = 0;
-		left = ret > 0 ? ret : 0;
+		if (ret <= 0)
+			left = 0;
+		else if (off + ret > i_size)
+			left = i_size - off;
+		else
+			left = ret;
 		while (left > 0) {
 			size_t plen, copied;
 
@@ -1164,15 +1169,13 @@ ssize_t __ceph_sync_read(struct inode *inode, loff_t *ki_pos,
 	}
 
 	if (ret > 0) {
-		if (off > *ki_pos) {
-			if (off >= i_size) {
-				*retry_op = CHECK_EOF;
-				ret = i_size - *ki_pos;
-				*ki_pos = i_size;
-			} else {
-				ret = off - *ki_pos;
-				*ki_pos = off;
-			}
+		if (off >= i_size) {
+			*retry_op = CHECK_EOF;
+			ret = i_size - *ki_pos;
+			*ki_pos = i_size;
+		} else {
+			ret = off - *ki_pos;
+			*ki_pos = off;
 		}
 
 		if (last_objver)
diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index 2eb66dd7d0..950360b075 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -2470,6 +2470,50 @@ void ceph_reclaim_caps_nr(struct ceph_mds_client *mdsc, int nr)
 	}
 }
 
+void ceph_queue_cap_unlink_work(struct ceph_mds_client *mdsc)
+{
+	struct ceph_client *cl = mdsc->fsc->client;
+	if (mdsc->stopping)
+		return;
+
+        if (queue_work(mdsc->fsc->cap_wq, &mdsc->cap_unlink_work)) {
+                doutc(cl, "caps unlink work queued\n");
+        } else {
+                doutc(cl, "failed to queue caps unlink work\n");
+        }
+}
+
+static void ceph_cap_unlink_work(struct work_struct *work)
+{
+	struct ceph_mds_client *mdsc =
+		container_of(work, struct ceph_mds_client, cap_unlink_work);
+	struct ceph_client *cl = mdsc->fsc->client;
+
+	doutc(cl, "begin\n");
+	spin_lock(&mdsc->cap_unlink_delay_lock);
+	while (!list_empty(&mdsc->cap_unlink_delay_list)) {
+		struct ceph_inode_info *ci;
+		struct inode *inode;
+
+		ci = list_first_entry(&mdsc->cap_unlink_delay_list,
+				      struct ceph_inode_info,
+				      i_cap_delay_list);
+		list_del_init(&ci->i_cap_delay_list);
+
+		inode = igrab(&ci->netfs.inode);
+		if (inode) {
+			spin_unlock(&mdsc->cap_unlink_delay_lock);
+			doutc(cl, "on %p %llx.%llx\n", inode,
+			      ceph_vinop(inode));
+			ceph_check_caps(ci, CHECK_CAPS_FLUSH);
+			iput(inode);
+			spin_lock(&mdsc->cap_unlink_delay_lock);
+		}
+	}
+	spin_unlock(&mdsc->cap_unlink_delay_lock);
+	doutc(cl, "done\n");
+}
+
 /*
  * requests
  */
@@ -5345,6 +5389,8 @@ int ceph_mdsc_init(struct ceph_fs_client *fsc)
 	INIT_LIST_HEAD(&mdsc->cap_delay_list);
 	INIT_LIST_HEAD(&mdsc->cap_wait_list);
 	spin_lock_init(&mdsc->cap_delay_lock);
+	INIT_LIST_HEAD(&mdsc->cap_unlink_delay_list);
+	spin_lock_init(&mdsc->cap_unlink_delay_lock);
 	INIT_LIST_HEAD(&mdsc->snap_flush_list);
 	spin_lock_init(&mdsc->snap_flush_lock);
 	mdsc->last_cap_flush_tid = 1;
@@ -5353,6 +5399,7 @@ int ceph_mdsc_init(struct ceph_fs_client *fsc)
 	spin_lock_init(&mdsc->cap_dirty_lock);
 	init_waitqueue_head(&mdsc->cap_flushing_wq);
 	INIT_WORK(&mdsc->cap_reclaim_work, ceph_cap_reclaim_work);
+	INIT_WORK(&mdsc->cap_unlink_work, ceph_cap_unlink_work);
 	err = ceph_metric_init(&mdsc->metric);
 	if (err)
 		goto err_mdsmap;
@@ -5626,6 +5673,7 @@ void ceph_mdsc_close_sessions(struct ceph_mds_client *mdsc)
 	ceph_cleanup_global_and_empty_realms(mdsc);
 
 	cancel_work_sync(&mdsc->cap_reclaim_work);
+	cancel_work_sync(&mdsc->cap_unlink_work);
 	cancel_delayed_work_sync(&mdsc->delayed_work); /* cancel timer */
 
 	doutc(cl, "done\n");
diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h
index 40560af388..03f8ff0087 100644
--- a/fs/ceph/mds_client.h
+++ b/fs/ceph/mds_client.h
@@ -462,6 +462,8 @@ struct ceph_mds_client {
 	unsigned long    last_renew_caps;  /* last time we renewed our caps */
 	struct list_head cap_delay_list;   /* caps with delayed release */
 	spinlock_t       cap_delay_lock;   /* protects cap_delay_list */
+	struct list_head cap_unlink_delay_list;  /* caps with delayed release for unlink */
+	spinlock_t       cap_unlink_delay_lock;  /* protects cap_unlink_delay_list */
 	struct list_head snap_flush_list;  /* cap_snaps ready to flush */
 	spinlock_t       snap_flush_lock;
 
@@ -475,6 +477,8 @@ struct ceph_mds_client {
 	struct work_struct cap_reclaim_work;
 	atomic_t	   cap_reclaim_pending;
 
+	struct work_struct cap_unlink_work;
+
 	/*
 	 * Cap reservations
 	 *
@@ -574,6 +578,7 @@ extern void ceph_flush_cap_releases(struct ceph_mds_client *mdsc,
 				    struct ceph_mds_session *session);
 extern void ceph_queue_cap_reclaim_work(struct ceph_mds_client *mdsc);
 extern void ceph_reclaim_caps_nr(struct ceph_mds_client *mdsc, int nr);
+extern void ceph_queue_cap_unlink_work(struct ceph_mds_client *mdsc);
 extern int ceph_iterate_session_caps(struct ceph_mds_session *session,
 				     int (*cb)(struct inode *, int mds, void *),
 				     void *arg);