ceph: use i_ceph_lock instead of i_lock

We have been using i_lock to protect all kinds of data structures in the
ceph_inode_info struct, including lists of inodes that we need to iterate
over while avoiding races with inode destruction.  That requires grabbing
a reference to the inode with the list lock protected, but igrab() now
takes i_lock to check the inode flags.

Changing the list lock ordering would be a painful process.

However, using a ceph-specific i_ceph_lock in the ceph inode instead of
i_lock is a simple mechanical change and avoids the ordering constraints
imposed by igrab().

Reported-by: Amon Ott <a.ott@m-privacy.de>
Signed-off-by: Sage Weil <sage@newdream.net>

fs/ceph/addr.c

@@ -87,7 +87,7 @@ static int ceph_set_page_dirty(struct page *page)
 	snapc = ceph_get_snap_context(ci->i_snap_realm->cached_context);
 
 	/* dirty the head */
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	if (ci->i_head_snapc == NULL)
 		ci->i_head_snapc = ceph_get_snap_context(snapc);
 	++ci->i_wrbuffer_ref_head;
@@ -100,7 +100,7 @@ static int ceph_set_page_dirty(struct page *page)
 	     ci->i_wrbuffer_ref-1, ci->i_wrbuffer_ref_head-1,
 	     ci->i_wrbuffer_ref, ci->i_wrbuffer_ref_head,
 	     snapc, snapc->seq, snapc->num_snaps);
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	/* now adjust page */
 	spin_lock_irq(&mapping->tree_lock);
@@ -391,7 +391,7 @@ static struct ceph_snap_context *get_oldest_context(struct inode *inode,
 	struct ceph_snap_context *snapc = NULL;
 	struct ceph_cap_snap *capsnap = NULL;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) {
 		dout(" cap_snap %p snapc %p has %d dirty pages\n", capsnap,
 		     capsnap->context, capsnap->dirty_pages);
@@ -407,7 +407,7 @@ static struct ceph_snap_context *get_oldest_context(struct inode *inode,
 		dout(" head snapc %p has %d dirty pages\n",
 		     snapc, ci->i_wrbuffer_ref_head);
 	}
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return snapc;
 }
 

fs/ceph/caps.c

@@ -309,7 +309,7 @@ void ceph_reservation_status(struct ceph_fs_client *fsc,
 /*
  * Find ceph_cap for given mds, if any.
  *
- * Called with i_lock held.
+ * Called with i_ceph_lock held.
  */
 static struct ceph_cap *__get_cap_for_mds(struct ceph_inode_info *ci, int mds)
 {
@@ -332,9 +332,9 @@ struct ceph_cap *ceph_get_cap_for_mds(struct ceph_inode_info *ci, int mds)
 {
 	struct ceph_cap *cap;
 
-	spin_lock(&ci->vfs_inode.i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	cap = __get_cap_for_mds(ci, mds);
-	spin_unlock(&ci->vfs_inode.i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return cap;
 }
 
@@ -361,15 +361,16 @@ static int __ceph_get_cap_mds(struct ceph_inode_info *ci)
 
 int ceph_get_cap_mds(struct inode *inode)
 {
+	struct ceph_inode_info *ci = ceph_inode(inode);
 	int mds;
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	mds = __ceph_get_cap_mds(ceph_inode(inode));
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return mds;
 }
 
 /*
- * Called under i_lock.
+ * Called under i_ceph_lock.
  */
 static void __insert_cap_node(struct ceph_inode_info *ci,
 			      struct ceph_cap *new)
@@ -415,7 +416,7 @@ static void __cap_set_timeouts(struct ceph_mds_client *mdsc,
  *
  * If I_FLUSH is set, leave the inode at the front of the list.
  *
- * Caller holds i_lock
+ * Caller holds i_ceph_lock
  *    -> we take mdsc->cap_delay_lock
  */
 static void __cap_delay_requeue(struct ceph_mds_client *mdsc,
@@ -457,7 +458,7 @@ static void __cap_delay_requeue_front(struct ceph_mds_client *mdsc,
 /*
  * Cancel delayed work on cap.
  *
- * Caller must hold i_lock.
+ * Caller must hold i_ceph_lock.
  */
 static void __cap_delay_cancel(struct ceph_mds_client *mdsc,
 			       struct ceph_inode_info *ci)
@@ -532,14 +533,14 @@ int ceph_add_cap(struct inode *inode,
 		wanted |= ceph_caps_for_mode(fmode);
 
 retry:
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	cap = __get_cap_for_mds(ci, mds);
 	if (!cap) {
 		if (new_cap) {
 			cap = new_cap;
 			new_cap = NULL;
 		} else {
-			spin_unlock(&inode->i_lock);
+			spin_unlock(&ci->i_ceph_lock);
 			new_cap = get_cap(mdsc, caps_reservation);
 			if (new_cap == NULL)
 				return -ENOMEM;
@@ -625,7 +626,7 @@ retry:
 
 	if (fmode >= 0)
 		__ceph_get_fmode(ci, fmode);
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	wake_up_all(&ci->i_cap_wq);
 	return 0;
 }
@@ -792,7 +793,7 @@ int ceph_caps_revoking(struct ceph_inode_info *ci, int mask)
 	struct rb_node *p;
 	int ret = 0;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) {
 		cap = rb_entry(p, struct ceph_cap, ci_node);
 		if (__cap_is_valid(cap) &&
@@ -801,7 +802,7 @@ int ceph_caps_revoking(struct ceph_inode_info *ci, int mask)
 			break;
 		}
 	}
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	dout("ceph_caps_revoking %p %s = %d\n", inode,
 	     ceph_cap_string(mask), ret);
 	return ret;
@@ -855,7 +856,7 @@ int __ceph_caps_mds_wanted(struct ceph_inode_info *ci)
 }
 
 /*
- * called under i_lock
+ * called under i_ceph_lock
  */
 static int __ceph_is_any_caps(struct ceph_inode_info *ci)
 {
@@ -865,7 +866,7 @@ static int __ceph_is_any_caps(struct ceph_inode_info *ci)
 /*
  * Remove a cap.  Take steps to deal with a racing iterate_session_caps.
  *
- * caller should hold i_lock.
+ * caller should hold i_ceph_lock.
  * caller will not hold session s_mutex if called from destroy_inode.
  */
 void __ceph_remove_cap(struct ceph_cap *cap)
@@ -1028,7 +1029,7 @@ static void __queue_cap_release(struct ceph_mds_session *session,
 
 /*
  * Queue cap releases when an inode is dropped from our cache.  Since
- * inode is about to be destroyed, there is no need for i_lock.
+ * inode is about to be destroyed, there is no need for i_ceph_lock.
  */
 void ceph_queue_caps_release(struct inode *inode)
 {
@@ -1049,7 +1050,7 @@ void ceph_queue_caps_release(struct inode *inode)
 
 /*
  * Send a cap msg on the given inode.  Update our caps state, then
- * drop i_lock and send the message.
+ * drop i_ceph_lock and send the message.
  *
  * Make note of max_size reported/requested from mds, revoked caps
  * that have now been implemented.
@@ -1061,13 +1062,13 @@ void ceph_queue_caps_release(struct inode *inode)
  * Return non-zero if delayed release, or we experienced an error
  * such that the caller should requeue + retry later.
  *
- * called with i_lock, then drops it.
+ * called with i_ceph_lock, then drops it.
  * caller should hold snap_rwsem (read), s_mutex.
  */
 static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap,
 		      int op, int used, int want, int retain, int flushing,
 		      unsigned *pflush_tid)
-	__releases(cap->ci->vfs_inode->i_lock)
+	__releases(cap->ci->i_ceph_lock)
 {
 	struct ceph_inode_info *ci = cap->ci;
 	struct inode *inode = &ci->vfs_inode;
@@ -1170,7 +1171,7 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap,
 		xattr_version = ci->i_xattrs.version;
 	}
 
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	ret = send_cap_msg(session, ceph_vino(inode).ino, cap_id,
 		op, keep, want, flushing, seq, flush_tid, issue_seq, mseq,
@@ -1198,13 +1199,13 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap,
  * Unless @again is true, skip cap_snaps that were already sent to
  * the MDS (i.e., during this session).
  *
- * Called under i_lock.  Takes s_mutex as needed.
+ * Called under i_ceph_lock.  Takes s_mutex as needed.
  */
 void __ceph_flush_snaps(struct ceph_inode_info *ci,
 			struct ceph_mds_session **psession,
 			int again)
-		__releases(ci->vfs_inode->i_lock)
-		__acquires(ci->vfs_inode->i_lock)
+		__releases(ci->i_ceph_lock)
+		__acquires(ci->i_ceph_lock)
 {
 	struct inode *inode = &ci->vfs_inode;
 	int mds;
@@ -1261,7 +1262,7 @@ retry:
 			session = NULL;
 		}
 		if (!session) {
-			spin_unlock(&inode->i_lock);
+			spin_unlock(&ci->i_ceph_lock);
 			mutex_lock(&mdsc->mutex);
 			session = __ceph_lookup_mds_session(mdsc, mds);
 			mutex_unlock(&mdsc->mutex);
@@ -1275,7 +1276,7 @@ retry:
 			 * deletion or migration.  retry, and we'll
 			 * get a better @mds value next time.
 			 */
-			spin_lock(&inode->i_lock);
+			spin_lock(&ci->i_ceph_lock);
 			goto retry;
 		}
 
@@ -1285,7 +1286,7 @@ retry:
 			list_del_init(&capsnap->flushing_item);
 		list_add_tail(&capsnap->flushing_item,
 			      &session->s_cap_snaps_flushing);
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 
 		dout("flush_snaps %p cap_snap %p follows %lld tid %llu\n",
 		     inode, capsnap, capsnap->follows, capsnap->flush_tid);
@@ -1302,7 +1303,7 @@ retry:
 		next_follows = capsnap->follows + 1;
 		ceph_put_cap_snap(capsnap);
 
-		spin_lock(&inode->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		goto retry;
 	}
 
@@ -1322,11 +1323,9 @@ out:
 
 static void ceph_flush_snaps(struct ceph_inode_info *ci)
 {
-	struct inode *inode = &ci->vfs_inode;
-
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	__ceph_flush_snaps(ci, NULL, 0);
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 }
 
 /*
@@ -1373,7 +1372,7 @@ int __ceph_mark_dirty_caps(struct ceph_inode_info *ci, int mask)
  * Add dirty inode to the flushing list.  Assigned a seq number so we
  * can wait for caps to flush without starving.
  *
- * Called under i_lock.
+ * Called under i_ceph_lock.
  */
 static int __mark_caps_flushing(struct inode *inode,
 				 struct ceph_mds_session *session)
@@ -1421,9 +1420,9 @@ static int try_nonblocking_invalidate(struct inode *inode)
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	u32 invalidating_gen = ci->i_rdcache_gen;
 
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	invalidate_mapping_pages(&inode->i_data, 0, -1);
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 
 	if (inode->i_data.nrpages == 0 &&
 	    invalidating_gen == ci->i_rdcache_gen) {
@@ -1470,7 +1469,7 @@ void ceph_check_caps(struct ceph_inode_info *ci, int flags,
 	if (mdsc->stopping)
 		is_delayed = 1;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 
 	if (ci->i_ceph_flags & CEPH_I_FLUSH)
 		flags |= CHECK_CAPS_FLUSH;
@@ -1480,7 +1479,7 @@ void ceph_check_caps(struct ceph_inode_info *ci, int flags,
 		__ceph_flush_snaps(ci, &session, 0);
 	goto retry_locked;
 retry:
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 retry_locked:
 	file_wanted = __ceph_caps_file_wanted(ci);
 	used = __ceph_caps_used(ci);
@@ -1634,7 +1633,7 @@ ack:
 			if (mutex_trylock(&session->s_mutex) == 0) {
 				dout("inverting session/ino locks on %p\n",
 				     session);
-				spin_unlock(&inode->i_lock);
+				spin_unlock(&ci->i_ceph_lock);
 				if (took_snap_rwsem) {
 					up_read(&mdsc->snap_rwsem);
 					took_snap_rwsem = 0;
@@ -1648,7 +1647,7 @@ ack:
 			if (down_read_trylock(&mdsc->snap_rwsem) == 0) {
 				dout("inverting snap/in locks on %p\n",
 				     inode);
-				spin_unlock(&inode->i_lock);
+				spin_unlock(&ci->i_ceph_lock);
 				down_read(&mdsc->snap_rwsem);
 				took_snap_rwsem = 1;
 				goto retry;
@@ -1664,10 +1663,10 @@ ack:
 		mds = cap->mds;  /* remember mds, so we don't repeat */
 		sent++;
 
-		/* __send_cap drops i_lock */
+		/* __send_cap drops i_ceph_lock */
 		delayed += __send_cap(mdsc, cap, CEPH_CAP_OP_UPDATE, used, want,
 				      retain, flushing, NULL);
-		goto retry; /* retake i_lock and restart our cap scan. */
+		goto retry; /* retake i_ceph_lock and restart our cap scan. */
 	}
 
 	/*
@@ -1681,7 +1680,7 @@ ack:
 	else if (!is_delayed || force_requeue)
 		__cap_delay_requeue(mdsc, ci);
 
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	if (queue_invalidate)
 		ceph_queue_invalidate(inode);
@@ -1704,7 +1703,7 @@ static int try_flush_caps(struct inode *inode, struct ceph_mds_session *session,
 	int flushing = 0;
 
 retry:
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	if (ci->i_ceph_flags & CEPH_I_NOFLUSH) {
 		dout("try_flush_caps skipping %p I_NOFLUSH set\n", inode);
 		goto out;
@@ -1716,7 +1715,7 @@ retry:
 		int delayed;
 
 		if (!session) {
-			spin_unlock(&inode->i_lock);
+			spin_unlock(&ci->i_ceph_lock);
 			session = cap->session;
 			mutex_lock(&session->s_mutex);
 			goto retry;
@@ -1727,18 +1726,18 @@ retry:
 
 		flushing = __mark_caps_flushing(inode, session);
 
-		/* __send_cap drops i_lock */
+		/* __send_cap drops i_ceph_lock */
 		delayed = __send_cap(mdsc, cap, CEPH_CAP_OP_FLUSH, used, want,
 				     cap->issued | cap->implemented, flushing,
 				     flush_tid);
 		if (!delayed)
 			goto out_unlocked;
 
-		spin_lock(&inode->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		__cap_delay_requeue(mdsc, ci);
 	}
 out:
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 out_unlocked:
 	if (session && unlock_session)
 		mutex_unlock(&session->s_mutex);
@@ -1753,7 +1752,7 @@ static int caps_are_flushed(struct inode *inode, unsigned tid)
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	int i, ret = 1;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	for (i = 0; i < CEPH_CAP_BITS; i++)
 		if ((ci->i_flushing_caps & (1 << i)) &&
 		    ci->i_cap_flush_tid[i] <= tid) {
@@ -1761,7 +1760,7 @@ static int caps_are_flushed(struct inode *inode, unsigned tid)
 			ret = 0;
 			break;
 		}
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return ret;
 }
 
@@ -1868,10 +1867,10 @@ int ceph_write_inode(struct inode *inode, struct writeback_control *wbc)
 		struct ceph_mds_client *mdsc =
 			ceph_sb_to_client(inode->i_sb)->mdsc;
 
-		spin_lock(&inode->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		if (__ceph_caps_dirty(ci))
 			__cap_delay_requeue_front(mdsc, ci);
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 	}
 	return err;
 }
@@ -1894,7 +1893,7 @@ static void kick_flushing_capsnaps(struct ceph_mds_client *mdsc,
 		struct inode *inode = &ci->vfs_inode;
 		struct ceph_cap *cap;
 
-		spin_lock(&inode->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		cap = ci->i_auth_cap;
 		if (cap && cap->session == session) {
 			dout("kick_flushing_caps %p cap %p capsnap %p\n", inode,
@@ -1904,7 +1903,7 @@ static void kick_flushing_capsnaps(struct ceph_mds_client *mdsc,
 			pr_err("%p auth cap %p not mds%d ???\n", inode,
 			       cap, session->s_mds);
 		}
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 	}
 }
 
@@ -1921,7 +1920,7 @@ void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
 		struct ceph_cap *cap;
 		int delayed = 0;
 
-		spin_lock(&inode->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		cap = ci->i_auth_cap;
 		if (cap && cap->session == session) {
 			dout("kick_flushing_caps %p cap %p %s\n", inode,
@@ -1932,14 +1931,14 @@ void ceph_kick_flushing_caps(struct ceph_mds_client *mdsc,
 					     cap->issued | cap->implemented,
 					     ci->i_flushing_caps, NULL);
 			if (delayed) {
-				spin_lock(&inode->i_lock);
+				spin_lock(&ci->i_ceph_lock);
 				__cap_delay_requeue(mdsc, ci);
-				spin_unlock(&inode->i_lock);
+				spin_unlock(&ci->i_ceph_lock);
 			}
 		} else {
 			pr_err("%p auth cap %p not mds%d ???\n", inode,
 			       cap, session->s_mds);
-			spin_unlock(&inode->i_lock);
+			spin_unlock(&ci->i_ceph_lock);
 		}
 	}
 }
@@ -1952,7 +1951,7 @@ static void kick_flushing_inode_caps(struct ceph_mds_client *mdsc,
 	struct ceph_cap *cap;
 	int delayed = 0;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	cap = ci->i_auth_cap;
 	dout("kick_flushing_inode_caps %p flushing %s flush_seq %lld\n", inode,
 	     ceph_cap_string(ci->i_flushing_caps), ci->i_cap_flush_seq);
@@ -1964,12 +1963,12 @@ static void kick_flushing_inode_caps(struct ceph_mds_client *mdsc,
 				     cap->issued | cap->implemented,
 				     ci->i_flushing_caps, NULL);
 		if (delayed) {
-			spin_lock(&inode->i_lock);
+			spin_lock(&ci->i_ceph_lock);
 			__cap_delay_requeue(mdsc, ci);
-			spin_unlock(&inode->i_lock);
+			spin_unlock(&ci->i_ceph_lock);
 		}
 	} else {
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 	}
 }
 
@@ -1978,7 +1977,7 @@ static void kick_flushing_inode_caps(struct ceph_mds_client *mdsc,
  * Take references to capabilities we hold, so that we don't release
  * them to the MDS prematurely.
  *
- * Protected by i_lock.
+ * Protected by i_ceph_lock.
  */
 static void __take_cap_refs(struct ceph_inode_info *ci, int got)
 {
@@ -2016,7 +2015,7 @@ static int try_get_cap_refs(struct ceph_inode_info *ci, int need, int want,
 
 	dout("get_cap_refs %p need %s want %s\n", inode,
 	     ceph_cap_string(need), ceph_cap_string(want));
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 
 	/* make sure file is actually open */
 	file_wanted = __ceph_caps_file_wanted(ci);
@@ -2077,7 +2076,7 @@ static int try_get_cap_refs(struct ceph_inode_info *ci, int need, int want,
 		     ceph_cap_string(have), ceph_cap_string(need));
 	}
 out:
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	dout("get_cap_refs %p ret %d got %s\n", inode,
 	     ret, ceph_cap_string(*got));
 	return ret;
@@ -2094,7 +2093,7 @@ static void check_max_size(struct inode *inode, loff_t endoff)
 	int check = 0;
 
 	/* do we need to explicitly request a larger max_size? */
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	if ((endoff >= ci->i_max_size ||
 	     endoff > (inode->i_size << 1)) &&
 	    endoff > ci->i_wanted_max_size) {
@@ -2103,7 +2102,7 @@ static void check_max_size(struct inode *inode, loff_t endoff)
 		ci->i_wanted_max_size = endoff;
 		check = 1;
 	}
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	if (check)
 		ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL);
 }
@@ -2140,9 +2139,9 @@ retry:
  */
 void ceph_get_cap_refs(struct ceph_inode_info *ci, int caps)
 {
-	spin_lock(&ci->vfs_inode.i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	__take_cap_refs(ci, caps);
-	spin_unlock(&ci->vfs_inode.i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 }
 
 /*
@@ -2160,7 +2159,7 @@ void ceph_put_cap_refs(struct ceph_inode_info *ci, int had)
 	int last = 0, put = 0, flushsnaps = 0, wake = 0;
 	struct ceph_cap_snap *capsnap;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	if (had & CEPH_CAP_PIN)
 		--ci->i_pin_ref;
 	if (had & CEPH_CAP_FILE_RD)
@@ -2193,7 +2192,7 @@ void ceph_put_cap_refs(struct ceph_inode_info *ci, int had)
 				}
 			}
 		}
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	dout("put_cap_refs %p had %s%s%s\n", inode, ceph_cap_string(had),
 	     last ? " last" : "", put ? " put" : "");
@@ -2225,7 +2224,7 @@ void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
 	int found = 0;
 	struct ceph_cap_snap *capsnap = NULL;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	ci->i_wrbuffer_ref -= nr;
 	last = !ci->i_wrbuffer_ref;
 
@@ -2274,7 +2273,7 @@ void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
 		}
 	}
 
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	if (last) {
 		ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL);
@@ -2291,7 +2290,7 @@ void ceph_put_wrbuffer_cap_refs(struct ceph_inode_info *ci, int nr,
  * Handle a cap GRANT message from the MDS.  (Note that a GRANT may
  * actually be a revocation if it specifies a smaller cap set.)
  *
- * caller holds s_mutex and i_lock, we drop both.
+ * caller holds s_mutex and i_ceph_lock, we drop both.
  *
  * return value:
  *  0 - ok
@@ -2302,7 +2301,7 @@ static void handle_cap_grant(struct inode *inode, struct ceph_mds_caps *grant,
 			     struct ceph_mds_session *session,
 			     struct ceph_cap *cap,
 			     struct ceph_buffer *xattr_buf)
-		__releases(inode->i_lock)
+		__releases(ci->i_ceph_lock)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	int mds = session->s_mds;
@@ -2453,7 +2452,7 @@ static void handle_cap_grant(struct inode *inode, struct ceph_mds_caps *grant,
 	}
 	BUG_ON(cap->issued & ~cap->implemented);
 
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	if (writeback)
 		/*
 		 * queue inode for writeback: we can't actually call
@@ -2483,7 +2482,7 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid,
 				 struct ceph_mds_caps *m,
 				 struct ceph_mds_session *session,
 				 struct ceph_cap *cap)
-	__releases(inode->i_lock)
+	__releases(ci->i_ceph_lock)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_mds_client *mdsc = ceph_sb_to_client(inode->i_sb)->mdsc;
@@ -2539,7 +2538,7 @@ static void handle_cap_flush_ack(struct inode *inode, u64 flush_tid,
 	wake_up_all(&ci->i_cap_wq);
 
 out:
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	if (drop)
 		iput(inode);
 }
@@ -2562,7 +2561,7 @@ static void handle_cap_flushsnap_ack(struct inode *inode, u64 flush_tid,
 	dout("handle_cap_flushsnap_ack inode %p ci %p mds%d follows %lld\n",
 	     inode, ci, session->s_mds, follows);
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) {
 		if (capsnap->follows == follows) {
 			if (capsnap->flush_tid != flush_tid) {
@@ -2585,7 +2584,7 @@ static void handle_cap_flushsnap_ack(struct inode *inode, u64 flush_tid,
 			     capsnap, capsnap->follows);
 		}
 	}
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	if (drop)
 		iput(inode);
 }
@@ -2598,7 +2597,7 @@ static void handle_cap_flushsnap_ack(struct inode *inode, u64 flush_tid,
 static void handle_cap_trunc(struct inode *inode,
 			     struct ceph_mds_caps *trunc,
 			     struct ceph_mds_session *session)
-	__releases(inode->i_lock)
+	__releases(ci->i_ceph_lock)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	int mds = session->s_mds;
@@ -2617,7 +2616,7 @@ static void handle_cap_trunc(struct inode *inode,
 	     inode, mds, seq, truncate_size, truncate_seq);
 	queue_trunc = ceph_fill_file_size(inode, issued,
 					  truncate_seq, truncate_size, size);
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	if (queue_trunc)
 		ceph_queue_vmtruncate(inode);
@@ -2646,7 +2645,7 @@ static void handle_cap_export(struct inode *inode, struct ceph_mds_caps *ex,
 	dout("handle_cap_export inode %p ci %p mds%d mseq %d\n",
 	     inode, ci, mds, mseq);
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 
 	/* make sure we haven't seen a higher mseq */
 	for (p = rb_first(&ci->i_caps); p; p = rb_next(p)) {
@@ -2690,7 +2689,7 @@ static void handle_cap_export(struct inode *inode, struct ceph_mds_caps *ex,
 	}
 	/* else, we already released it */
 
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 }
 
 /*
@@ -2745,9 +2744,9 @@ static void handle_cap_import(struct ceph_mds_client *mdsc,
 	up_read(&mdsc->snap_rwsem);
 
 	/* make sure we re-request max_size, if necessary */
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	ci->i_requested_max_size = 0;
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 }
 
 /*
@@ -2762,6 +2761,7 @@ void ceph_handle_caps(struct ceph_mds_session *session,
 	struct ceph_mds_client *mdsc = session->s_mdsc;
 	struct super_block *sb = mdsc->fsc->sb;
 	struct inode *inode;
+	struct ceph_inode_info *ci;
 	struct ceph_cap *cap;
 	struct ceph_mds_caps *h;
 	int mds = session->s_mds;
@@ -2815,6 +2815,7 @@ void ceph_handle_caps(struct ceph_mds_session *session,
 
 	/* lookup ino */
 	inode = ceph_find_inode(sb, vino);
+	ci = ceph_inode(inode);
 	dout(" op %s ino %llx.%llx inode %p\n", ceph_cap_op_name(op), vino.ino,
 	     vino.snap, inode);
 	if (!inode) {
@@ -2844,16 +2845,16 @@ void ceph_handle_caps(struct ceph_mds_session *session,
 	}
 
 	/* the rest require a cap */
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	cap = __get_cap_for_mds(ceph_inode(inode), mds);
 	if (!cap) {
 		dout(" no cap on %p ino %llx.%llx from mds%d\n",
 		     inode, ceph_ino(inode), ceph_snap(inode), mds);
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		goto flush_cap_releases;
 	}
 
-	/* note that each of these drops i_lock for us */
+	/* note that each of these drops i_ceph_lock for us */
 	switch (op) {
 	case CEPH_CAP_OP_REVOKE:
 	case CEPH_CAP_OP_GRANT:
@@ -2869,7 +2870,7 @@ void ceph_handle_caps(struct ceph_mds_session *session,
 		break;
 
 	default:
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		pr_err("ceph_handle_caps: unknown cap op %d %s\n", op,
 		       ceph_cap_op_name(op));
 	}
@@ -2962,13 +2963,13 @@ void ceph_put_fmode(struct ceph_inode_info *ci, int fmode)
 	struct inode *inode = &ci->vfs_inode;
 	int last = 0;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	dout("put_fmode %p fmode %d %d -> %d\n", inode, fmode,
 	     ci->i_nr_by_mode[fmode], ci->i_nr_by_mode[fmode]-1);
 	BUG_ON(ci->i_nr_by_mode[fmode] == 0);
 	if (--ci->i_nr_by_mode[fmode] == 0)
 		last++;
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	if (last && ci->i_vino.snap == CEPH_NOSNAP)
 		ceph_check_caps(ci, 0, NULL);
@@ -2991,7 +2992,7 @@ int ceph_encode_inode_release(void **p, struct inode *inode,
 	int used, dirty;
 	int ret = 0;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	used = __ceph_caps_used(ci);
 	dirty = __ceph_caps_dirty(ci);
 
@@ -3046,7 +3047,7 @@ int ceph_encode_inode_release(void **p, struct inode *inode,
 			     inode, cap, ceph_cap_string(cap->issued));
 		}
 	}
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return ret;
 }
 
@@ -3061,7 +3062,7 @@ int ceph_encode_dentry_release(void **p, struct dentry *dentry,
 
 	/*
 	 * force an record for the directory caps if we have a dentry lease.
-	 * this is racy (can't take i_lock and d_lock together), but it
+	 * this is racy (can't take i_ceph_lock and d_lock together), but it
 	 * doesn't have to be perfect; the mds will revoke anything we don't
 	 * release.
 	 */

fs/ceph/dir.c

@@ -281,18 +281,18 @@ static int ceph_readdir(struct file *filp, void *dirent, filldir_t filldir)
 	}
 
 	/* can we use the dcache? */
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	if ((filp->f_pos == 2 || fi->dentry) &&
 	    !ceph_test_mount_opt(fsc, NOASYNCREADDIR) &&
 	    ceph_snap(inode) != CEPH_SNAPDIR &&
 	    ceph_dir_test_complete(inode) &&
 	    __ceph_caps_issued_mask(ci, CEPH_CAP_FILE_SHARED, 1)) {
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		err = __dcache_readdir(filp, dirent, filldir);
 		if (err != -EAGAIN)
 			return err;
 	} else {
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 	}
 	if (fi->dentry) {
 		err = note_last_dentry(fi, fi->dentry->d_name.name,
@@ -428,12 +428,12 @@ more:
 	 * were released during the whole readdir, and we should have
 	 * the complete dir contents in our cache.
 	 */
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	if (ci->i_release_count == fi->dir_release_count) {
 		ceph_dir_set_complete(inode);
 		ci->i_max_offset = filp->f_pos;
 	}
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	dout("readdir %p filp %p done.\n", inode, filp);
 	return 0;
@@ -607,7 +607,7 @@ static struct dentry *ceph_lookup(struct inode *dir, struct dentry *dentry,
 		struct ceph_inode_info *ci = ceph_inode(dir);
 		struct ceph_dentry_info *di = ceph_dentry(dentry);
 
-		spin_lock(&dir->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		dout(" dir %p flags are %d\n", dir, ci->i_ceph_flags);
 		if (strncmp(dentry->d_name.name,
 			    fsc->mount_options->snapdir_name,
@@ -615,13 +615,13 @@ static struct dentry *ceph_lookup(struct inode *dir, struct dentry *dentry,
 		    !is_root_ceph_dentry(dir, dentry) &&
 		    ceph_dir_test_complete(dir) &&
 		    (__ceph_caps_issued_mask(ci, CEPH_CAP_FILE_SHARED, 1))) {
-			spin_unlock(&dir->i_lock);
+			spin_unlock(&ci->i_ceph_lock);
 			dout(" dir %p complete, -ENOENT\n", dir);
 			d_add(dentry, NULL);
 			di->lease_shared_gen = ci->i_shared_gen;
 			return NULL;
 		}
-		spin_unlock(&dir->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 	}
 
 	op = ceph_snap(dir) == CEPH_SNAPDIR ?
@@ -841,12 +841,12 @@ static int drop_caps_for_unlink(struct inode *inode)
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	int drop = CEPH_CAP_LINK_SHARED | CEPH_CAP_LINK_EXCL;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	if (inode->i_nlink == 1) {
 		drop |= ~(__ceph_caps_wanted(ci) | CEPH_CAP_PIN);
 		ci->i_ceph_flags |= CEPH_I_NODELAY;
 	}
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return drop;
 }
 
@@ -1015,10 +1015,10 @@ static int dir_lease_is_valid(struct inode *dir, struct dentry *dentry)
 	struct ceph_dentry_info *di = ceph_dentry(dentry);
 	int valid = 0;
 
-	spin_lock(&dir->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	if (ci->i_shared_gen == di->lease_shared_gen)
 		valid = __ceph_caps_issued_mask(ci, CEPH_CAP_FILE_SHARED, 1);
-	spin_unlock(&dir->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	dout("dir_lease_is_valid dir %p v%u dentry %p v%u = %d\n",
 	     dir, (unsigned)ci->i_shared_gen, dentry,
 	     (unsigned)di->lease_shared_gen, valid);

fs/ceph/file.c

@@ -147,9 +147,9 @@ int ceph_open(struct inode *inode, struct file *file)
 
 	/* trivially open snapdir */
 	if (ceph_snap(inode) == CEPH_SNAPDIR) {
-		spin_lock(&inode->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		__ceph_get_fmode(ci, fmode);
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		return ceph_init_file(inode, file, fmode);
 	}
 
@@ -158,7 +158,7 @@ int ceph_open(struct inode *inode, struct file *file)
 	 * write) or any MDS (for read).  Update wanted set
 	 * asynchronously.
 	 */
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	if (__ceph_is_any_real_caps(ci) &&
 	    (((fmode & CEPH_FILE_MODE_WR) == 0) || ci->i_auth_cap)) {
 		int mds_wanted = __ceph_caps_mds_wanted(ci);
@@ -168,7 +168,7 @@ int ceph_open(struct inode *inode, struct file *file)
 		     inode, fmode, ceph_cap_string(wanted),
 		     ceph_cap_string(issued));
 		__ceph_get_fmode(ci, fmode);
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 
 		/* adjust wanted? */
 		if ((issued & wanted) != wanted &&
@@ -180,10 +180,10 @@ int ceph_open(struct inode *inode, struct file *file)
 	} else if (ceph_snap(inode) != CEPH_NOSNAP &&
 		   (ci->i_snap_caps & wanted) == wanted) {
 		__ceph_get_fmode(ci, fmode);
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		return ceph_init_file(inode, file, fmode);
 	}
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	dout("open fmode %d wants %s\n", fmode, ceph_cap_string(wanted));
 	req = prepare_open_request(inode->i_sb, flags, 0);
@@ -743,9 +743,9 @@ retry_snap:
 		 */
 		int dirty;
 
-		spin_lock(&inode->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR);
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		ceph_put_cap_refs(ci, got);
 
 		ret = generic_file_aio_write(iocb, iov, nr_segs, pos);
@@ -764,9 +764,9 @@ retry_snap:
 
 	if (ret >= 0) {
 		int dirty;
-		spin_lock(&inode->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR);
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		if (dirty)
 			__mark_inode_dirty(inode, dirty);
 	}

fs/ceph/inode.c

@@ -297,6 +297,8 @@ struct inode *ceph_alloc_inode(struct super_block *sb)
 
 	dout("alloc_inode %p\n", &ci->vfs_inode);
 
+	spin_lock_init(&ci->i_ceph_lock);
+
 	ci->i_version = 0;
 	ci->i_time_warp_seq = 0;
 	ci->i_ceph_flags = 0;
@@ -583,7 +585,7 @@ static int fill_inode(struct inode *inode,
 			       iinfo->xattr_len);
 	}
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 
 	/*
 	 * provided version will be odd if inode value is projected,
@@ -680,7 +682,7 @@ static int fill_inode(struct inode *inode,
 			char *sym;
 
 			BUG_ON(symlen != inode->i_size);
-			spin_unlock(&inode->i_lock);
+			spin_unlock(&ci->i_ceph_lock);
 
 			err = -ENOMEM;
 			sym = kmalloc(symlen+1, GFP_NOFS);
@@ -689,7 +691,7 @@ static int fill_inode(struct inode *inode,
 			memcpy(sym, iinfo->symlink, symlen);
 			sym[symlen] = 0;
 
-			spin_lock(&inode->i_lock);
+			spin_lock(&ci->i_ceph_lock);
 			if (!ci->i_symlink)
 				ci->i_symlink = sym;
 			else
@@ -715,7 +717,7 @@ static int fill_inode(struct inode *inode,
 	}
 
 no_change:
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	/* queue truncate if we saw i_size decrease */
 	if (queue_trunc)
@@ -750,13 +752,13 @@ no_change:
 				     info->cap.flags,
 				     caps_reservation);
 		} else {
-			spin_lock(&inode->i_lock);
+			spin_lock(&ci->i_ceph_lock);
 			dout(" %p got snap_caps %s\n", inode,
 			     ceph_cap_string(le32_to_cpu(info->cap.caps)));
 			ci->i_snap_caps |= le32_to_cpu(info->cap.caps);
 			if (cap_fmode >= 0)
 				__ceph_get_fmode(ci, cap_fmode);
-			spin_unlock(&inode->i_lock);
+			spin_unlock(&ci->i_ceph_lock);
 		}
 	} else if (cap_fmode >= 0) {
 		pr_warning("mds issued no caps on %llx.%llx\n",
@@ -849,19 +851,20 @@ static void ceph_set_dentry_offset(struct dentry *dn)
 {
 	struct dentry *dir = dn->d_parent;
 	struct inode *inode = dir->d_inode;
+	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_dentry_info *di;
 
 	BUG_ON(!inode);
 
 	di = ceph_dentry(dn);
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	if (!ceph_dir_test_complete(inode)) {
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		return;
 	}
 	di->offset = ceph_inode(inode)->i_max_offset++;
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	spin_lock(&dir->d_lock);
 	spin_lock_nested(&dn->d_lock, DENTRY_D_LOCK_NESTED);
@@ -1308,7 +1311,7 @@ int ceph_inode_set_size(struct inode *inode, loff_t size)
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	int ret = 0;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	dout("set_size %p %llu -> %llu\n", inode, inode->i_size, size);
 	inode->i_size = size;
 	inode->i_blocks = (size + (1 << 9) - 1) >> 9;
@@ -1318,7 +1321,7 @@ int ceph_inode_set_size(struct inode *inode, loff_t size)
 	    (ci->i_reported_size << 1) < ci->i_max_size)
 		ret = 1;
 
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return ret;
 }
 
@@ -1376,20 +1379,20 @@ static void ceph_invalidate_work(struct work_struct *work)
 	u32 orig_gen;
 	int check = 0;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	dout("invalidate_pages %p gen %d revoking %d\n", inode,
 	     ci->i_rdcache_gen, ci->i_rdcache_revoking);
 	if (ci->i_rdcache_revoking != ci->i_rdcache_gen) {
 		/* nevermind! */
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		goto out;
 	}
 	orig_gen = ci->i_rdcache_gen;
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	truncate_inode_pages(&inode->i_data, 0);
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	if (orig_gen == ci->i_rdcache_gen &&
 	    orig_gen == ci->i_rdcache_revoking) {
 		dout("invalidate_pages %p gen %d successful\n", inode,
@@ -1401,7 +1404,7 @@ static void ceph_invalidate_work(struct work_struct *work)
 		     inode, orig_gen, ci->i_rdcache_gen,
 		     ci->i_rdcache_revoking);
 	}
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	if (check)
 		ceph_check_caps(ci, 0, NULL);
@@ -1460,10 +1463,10 @@ void __ceph_do_pending_vmtruncate(struct inode *inode)
 	int wrbuffer_refs, wake = 0;
 
 retry:
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	if (ci->i_truncate_pending == 0) {
 		dout("__do_pending_vmtruncate %p none pending\n", inode);
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		return;
 	}
 
@@ -1474,7 +1477,7 @@ retry:
 	if (ci->i_wrbuffer_ref_head < ci->i_wrbuffer_ref) {
 		dout("__do_pending_vmtruncate %p flushing snaps first\n",
 		     inode);
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		filemap_write_and_wait_range(&inode->i_data, 0,
 					     inode->i_sb->s_maxbytes);
 		goto retry;
@@ -1484,15 +1487,15 @@ retry:
 	wrbuffer_refs = ci->i_wrbuffer_ref;
 	dout("__do_pending_vmtruncate %p (%d) to %lld\n", inode,
 	     ci->i_truncate_pending, to);
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	truncate_inode_pages(inode->i_mapping, to);
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	ci->i_truncate_pending--;
 	if (ci->i_truncate_pending == 0)
 		wake = 1;
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	if (wrbuffer_refs == 0)
 		ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL);
@@ -1547,7 +1550,7 @@ int ceph_setattr(struct dentry *dentry, struct iattr *attr)
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	issued = __ceph_caps_issued(ci, NULL);
 	dout("setattr %p issued %s\n", inode, ceph_cap_string(issued));
 
@@ -1695,7 +1698,7 @@ int ceph_setattr(struct dentry *dentry, struct iattr *attr)
 	}
 
 	release &= issued;
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	if (inode_dirty_flags)
 		__mark_inode_dirty(inode, inode_dirty_flags);
@@ -1717,7 +1720,7 @@ int ceph_setattr(struct dentry *dentry, struct iattr *attr)
 	__ceph_do_pending_vmtruncate(inode);
 	return err;
 out:
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	ceph_mdsc_put_request(req);
 	return err;
 }

fs/ceph/ioctl.c

@@ -241,11 +241,11 @@ static long ceph_ioctl_lazyio(struct file *file)
 	struct ceph_inode_info *ci = ceph_inode(inode);
 
 	if ((fi->fmode & CEPH_FILE_MODE_LAZY) == 0) {
-		spin_lock(&inode->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		ci->i_nr_by_mode[fi->fmode]--;
 		fi->fmode |= CEPH_FILE_MODE_LAZY;
 		ci->i_nr_by_mode[fi->fmode]++;
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		dout("ioctl_layzio: file %p marked lazy\n", file);
 
 		ceph_check_caps(ci, 0, NULL);

fs/ceph/mds_client.c

@@ -732,21 +732,21 @@ static int __choose_mds(struct ceph_mds_client *mdsc,
 		}
 	}
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	cap = NULL;
 	if (mode == USE_AUTH_MDS)
 		cap = ci->i_auth_cap;
 	if (!cap && !RB_EMPTY_ROOT(&ci->i_caps))
 		cap = rb_entry(rb_first(&ci->i_caps), struct ceph_cap, ci_node);
 	if (!cap) {
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		goto random;
 	}
 	mds = cap->session->s_mds;
 	dout("choose_mds %p %llx.%llx mds%d (%scap %p)\n",
 	     inode, ceph_vinop(inode), mds,
 	     cap == ci->i_auth_cap ? "auth " : "", cap);
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return mds;
 
 random:
@@ -951,7 +951,7 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
 
 	dout("removing cap %p, ci is %p, inode is %p\n",
 	     cap, ci, &ci->vfs_inode);
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	__ceph_remove_cap(cap);
 	if (!__ceph_is_any_real_caps(ci)) {
 		struct ceph_mds_client *mdsc =
@@ -984,7 +984,7 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
 		}
 		spin_unlock(&mdsc->cap_dirty_lock);
 	}
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	while (drop--)
 		iput(inode);
 	return 0;
@@ -1015,10 +1015,10 @@ static int wake_up_session_cb(struct inode *inode, struct ceph_cap *cap,
 
 	wake_up_all(&ci->i_cap_wq);
 	if (arg) {
-		spin_lock(&inode->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		ci->i_wanted_max_size = 0;
 		ci->i_requested_max_size = 0;
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 	}
 	return 0;
 }
@@ -1151,7 +1151,7 @@ static int trim_caps_cb(struct inode *inode, struct ceph_cap *cap, void *arg)
 	if (session->s_trim_caps <= 0)
 		return -1;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	mine = cap->issued | cap->implemented;
 	used = __ceph_caps_used(ci);
 	oissued = __ceph_caps_issued_other(ci, cap);
@@ -1170,7 +1170,7 @@ static int trim_caps_cb(struct inode *inode, struct ceph_cap *cap, void *arg)
 		__ceph_remove_cap(cap);
 	} else {
 		/* try to drop referring dentries */
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		d_prune_aliases(inode);
 		dout("trim_caps_cb %p cap %p  pruned, count now %d\n",
 		     inode, cap, atomic_read(&inode->i_count));
@@ -1178,7 +1178,7 @@ static int trim_caps_cb(struct inode *inode, struct ceph_cap *cap, void *arg)
 	}
 
 out:
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return 0;
 }
 
@@ -1296,7 +1296,7 @@ static int check_cap_flush(struct ceph_mds_client *mdsc, u64 want_flush_seq)
 					   i_flushing_item);
 			struct inode *inode = &ci->vfs_inode;
 
-			spin_lock(&inode->i_lock);
+			spin_lock(&ci->i_ceph_lock);
 			if (ci->i_cap_flush_seq <= want_flush_seq) {
 				dout("check_cap_flush still flushing %p "
 				     "seq %lld <= %lld to mds%d\n", inode,
@@ -1304,7 +1304,7 @@ static int check_cap_flush(struct ceph_mds_client *mdsc, u64 want_flush_seq)
 				     session->s_mds);
 				ret = 0;
 			}
-			spin_unlock(&inode->i_lock);
+			spin_unlock(&ci->i_ceph_lock);
 		}
 		mutex_unlock(&session->s_mutex);
 		ceph_put_mds_session(session);
@@ -2011,10 +2011,10 @@ void ceph_invalidate_dir_request(struct ceph_mds_request *req)
 	struct ceph_inode_info *ci = ceph_inode(inode);
 
 	dout("invalidate_dir_request %p (D_COMPLETE, lease(s))\n", inode);
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	ceph_dir_clear_complete(inode);
 	ci->i_release_count++;
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	if (req->r_dentry)
 		ceph_invalidate_dentry_lease(req->r_dentry);
@@ -2422,7 +2422,7 @@ static int encode_caps_cb(struct inode *inode, struct ceph_cap *cap,
 	if (err)
 		goto out_free;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	cap->seq = 0;        /* reset cap seq */
 	cap->issue_seq = 0;  /* and issue_seq */
 
@@ -2445,7 +2445,7 @@ static int encode_caps_cb(struct inode *inode, struct ceph_cap *cap,
 		rec.v1.pathbase = cpu_to_le64(pathbase);
 		reclen = sizeof(rec.v1);
 	}
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 
 	if (recon_state->flock) {
 		int num_fcntl_locks, num_flock_locks;

fs/ceph/mds_client.h

@@ -20,7 +20,7 @@
  *
  *         mdsc->snap_rwsem
  *
- *         inode->i_lock
+ *         ci->i_ceph_lock
  *                 mdsc->snap_flush_lock
  *                 mdsc->cap_delay_lock
  *

fs/ceph/snap.c

@@ -446,7 +446,7 @@ void ceph_queue_cap_snap(struct ceph_inode_info *ci)
 		return;
 	}
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	used = __ceph_caps_used(ci);
 	dirty = __ceph_caps_dirty(ci);
 
@@ -528,7 +528,7 @@ void ceph_queue_cap_snap(struct ceph_inode_info *ci)
 		kfree(capsnap);
 	}
 
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 }
 
 /*
@@ -537,7 +537,7 @@ void ceph_queue_cap_snap(struct ceph_inode_info *ci)
  *
  * If capsnap can now be flushed, add to snap_flush list, and return 1.
  *
- * Caller must hold i_lock.
+ * Caller must hold i_ceph_lock.
  */
 int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
 			    struct ceph_cap_snap *capsnap)
@@ -739,9 +739,9 @@ static void flush_snaps(struct ceph_mds_client *mdsc)
 		inode = &ci->vfs_inode;
 		ihold(inode);
 		spin_unlock(&mdsc->snap_flush_lock);
-		spin_lock(&inode->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		__ceph_flush_snaps(ci, &session, 0);
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		iput(inode);
 		spin_lock(&mdsc->snap_flush_lock);
 	}
@@ -847,7 +847,7 @@ void ceph_handle_snap(struct ceph_mds_client *mdsc,
 				continue;
 			ci = ceph_inode(inode);
 
-			spin_lock(&inode->i_lock);
+			spin_lock(&ci->i_ceph_lock);
 			if (!ci->i_snap_realm)
 				goto skip_inode;
 			/*
@@ -876,7 +876,7 @@ void ceph_handle_snap(struct ceph_mds_client *mdsc,
 			oldrealm = ci->i_snap_realm;
 			ci->i_snap_realm = realm;
 			spin_unlock(&realm->inodes_with_caps_lock);
-			spin_unlock(&inode->i_lock);
+			spin_unlock(&ci->i_ceph_lock);
 
 			ceph_get_snap_realm(mdsc, realm);
 			ceph_put_snap_realm(mdsc, oldrealm);
@@ -885,7 +885,7 @@ void ceph_handle_snap(struct ceph_mds_client *mdsc,
 			continue;
 
 skip_inode:
-			spin_unlock(&inode->i_lock);
+			spin_unlock(&ci->i_ceph_lock);
 			iput(inode);
 		}
 

fs/ceph/super.h

@@ -220,7 +220,7 @@ struct ceph_dentry_info {
  * The locking for D_COMPLETE is a bit odd:
  *  - we can clear it at almost any time (see ceph_d_prune)
  *  - it is only meaningful if:
- *    - we hold dir inode i_lock
+ *    - we hold dir inode i_ceph_lock
  *    - we hold dir FILE_SHARED caps
  *    - the dentry D_COMPLETE is set
  */
@@ -250,6 +250,8 @@ struct ceph_inode_xattrs_info {
 struct ceph_inode_info {
 	struct ceph_vino i_vino;   /* ceph ino + snap */
 
+	spinlock_t i_ceph_lock;
+
 	u64 i_version;
 	u32 i_time_warp_seq;
 
@@ -271,7 +273,7 @@ struct ceph_inode_info {
 
 	struct ceph_inode_xattrs_info i_xattrs;
 
-	/* capabilities.  protected _both_ by i_lock and cap->session's
+	/* capabilities.  protected _both_ by i_ceph_lock and cap->session's
 	 * s_mutex. */
 	struct rb_root i_caps;           /* cap list */
 	struct ceph_cap *i_auth_cap;     /* authoritative cap, if any */
@@ -437,18 +439,18 @@ static inline void ceph_i_clear(struct inode *inode, unsigned mask)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	ci->i_ceph_flags &= ~mask;
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 }
 
 static inline void ceph_i_set(struct inode *inode, unsigned mask)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	ci->i_ceph_flags |= mask;
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 }
 
 static inline bool ceph_i_test(struct inode *inode, unsigned mask)
@@ -456,9 +458,9 @@ static inline bool ceph_i_test(struct inode *inode, unsigned mask)
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	bool r;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	r = (ci->i_ceph_flags & mask) == mask;
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return r;
 }
 
@@ -508,9 +510,9 @@ extern int __ceph_caps_issued_other(struct ceph_inode_info *ci,
 static inline int ceph_caps_issued(struct ceph_inode_info *ci)
 {
 	int issued;
-	spin_lock(&ci->vfs_inode.i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	issued = __ceph_caps_issued(ci, NULL);
-	spin_unlock(&ci->vfs_inode.i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return issued;
 }
 
@@ -518,9 +520,9 @@ static inline int ceph_caps_issued_mask(struct ceph_inode_info *ci, int mask,
 					int touch)
 {
 	int r;
-	spin_lock(&ci->vfs_inode.i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	r = __ceph_caps_issued_mask(ci, mask, touch);
-	spin_unlock(&ci->vfs_inode.i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return r;
 }
 
@@ -743,10 +745,9 @@ extern int ceph_add_cap(struct inode *inode,
 extern void __ceph_remove_cap(struct ceph_cap *cap);
 static inline void ceph_remove_cap(struct ceph_cap *cap)
 {
-	struct inode *inode = &cap->ci->vfs_inode;
-	spin_lock(&inode->i_lock);
+	spin_lock(&cap->ci->i_ceph_lock);
 	__ceph_remove_cap(cap);
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&cap->ci->i_ceph_lock);
 }
 extern void ceph_put_cap(struct ceph_mds_client *mdsc,
 			 struct ceph_cap *cap);

fs/ceph/xattr.c

@@ -343,8 +343,8 @@ void __ceph_destroy_xattrs(struct ceph_inode_info *ci)
 }
 
 static int __build_xattrs(struct inode *inode)
-	__releases(inode->i_lock)
-	__acquires(inode->i_lock)
+	__releases(ci->i_ceph_lock)
+	__acquires(ci->i_ceph_lock)
 {
 	u32 namelen;
 	u32 numattr = 0;
@@ -372,7 +372,7 @@ start:
 		end = p + ci->i_xattrs.blob->vec.iov_len;
 		ceph_decode_32_safe(&p, end, numattr, bad);
 		xattr_version = ci->i_xattrs.version;
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 
 		xattrs = kcalloc(numattr, sizeof(struct ceph_xattr *),
 				 GFP_NOFS);
@@ -387,7 +387,7 @@ start:
 				goto bad_lock;
 		}
 
-		spin_lock(&inode->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		if (ci->i_xattrs.version != xattr_version) {
 			/* lost a race, retry */
 			for (i = 0; i < numattr; i++)
@@ -418,7 +418,7 @@ start:
 
 	return err;
 bad_lock:
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 bad:
 	if (xattrs) {
 		for (i = 0; i < numattr; i++)
@@ -512,7 +512,7 @@ ssize_t ceph_getxattr(struct dentry *dentry, const char *name, void *value,
 	if (vxattrs)
 		vxattr = ceph_match_vxattr(vxattrs, name);
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	dout("getxattr %p ver=%lld index_ver=%lld\n", inode,
 	     ci->i_xattrs.version, ci->i_xattrs.index_version);
 
@@ -520,14 +520,14 @@ ssize_t ceph_getxattr(struct dentry *dentry, const char *name, void *value,
 	    (ci->i_xattrs.index_version >= ci->i_xattrs.version)) {
 		goto get_xattr;
 	} else {
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		/* get xattrs from mds (if we don't already have them) */
 		err = ceph_do_getattr(inode, CEPH_STAT_CAP_XATTR);
 		if (err)
 			return err;
 	}
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 
 	if (vxattr && vxattr->readonly) {
 		err = vxattr->getxattr_cb(ci, value, size);
@@ -558,7 +558,7 @@ get_xattr:
 	memcpy(value, xattr->val, xattr->val_len);
 
 out:
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return err;
 }
 
@@ -573,7 +573,7 @@ ssize_t ceph_listxattr(struct dentry *dentry, char *names, size_t size)
 	u32 len;
 	int i;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	dout("listxattr %p ver=%lld index_ver=%lld\n", inode,
 	     ci->i_xattrs.version, ci->i_xattrs.index_version);
 
@@ -581,13 +581,13 @@ ssize_t ceph_listxattr(struct dentry *dentry, char *names, size_t size)
 	    (ci->i_xattrs.index_version >= ci->i_xattrs.version)) {
 		goto list_xattr;
 	} else {
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		err = ceph_do_getattr(inode, CEPH_STAT_CAP_XATTR);
 		if (err)
 			return err;
 	}
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 
 	err = __build_xattrs(inode);
 	if (err < 0)
@@ -619,7 +619,7 @@ list_xattr:
 		}
 
 out:
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	return err;
 }
 
@@ -739,7 +739,7 @@ int ceph_setxattr(struct dentry *dentry, const char *name,
 	if (!xattr)
 		goto out;
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 retry:
 	issued = __ceph_caps_issued(ci, NULL);
 	if (!(issued & CEPH_CAP_XATTR_EXCL))
@@ -752,12 +752,12 @@ retry:
 	    required_blob_size > ci->i_xattrs.prealloc_blob->alloc_len) {
 		struct ceph_buffer *blob = NULL;
 
-		spin_unlock(&inode->i_lock);
+		spin_unlock(&ci->i_ceph_lock);
 		dout(" preaallocating new blob size=%d\n", required_blob_size);
 		blob = ceph_buffer_new(required_blob_size, GFP_NOFS);
 		if (!blob)
 			goto out;
-		spin_lock(&inode->i_lock);
+		spin_lock(&ci->i_ceph_lock);
 		if (ci->i_xattrs.prealloc_blob)
 			ceph_buffer_put(ci->i_xattrs.prealloc_blob);
 		ci->i_xattrs.prealloc_blob = blob;
@@ -770,13 +770,13 @@ retry:
 	dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_XATTR_EXCL);
 	ci->i_xattrs.dirty = true;
 	inode->i_ctime = CURRENT_TIME;
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	if (dirty)
 		__mark_inode_dirty(inode, dirty);
 	return err;
 
 do_sync:
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	err = ceph_sync_setxattr(dentry, name, value, size, flags);
 out:
 	kfree(newname);
@@ -833,7 +833,7 @@ int ceph_removexattr(struct dentry *dentry, const char *name)
 			return -EOPNOTSUPP;
 	}
 
-	spin_lock(&inode->i_lock);
+	spin_lock(&ci->i_ceph_lock);
 	__build_xattrs(inode);
 	issued = __ceph_caps_issued(ci, NULL);
 	dout("removexattr %p issued %s\n", inode, ceph_cap_string(issued));
@@ -846,12 +846,12 @@ int ceph_removexattr(struct dentry *dentry, const char *name)
 	ci->i_xattrs.dirty = true;
 	inode->i_ctime = CURRENT_TIME;
 
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	if (dirty)
 		__mark_inode_dirty(inode, dirty);
 	return err;
 do_sync:
-	spin_unlock(&inode->i_lock);
+	spin_unlock(&ci->i_ceph_lock);
 	err = ceph_send_removexattr(dentry, name);
 	return err;
 }