Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  ceph: keep reference to parent inode on ceph_dentry
  ceph: queue cap_snaps once per realm
  libceph: fix socket write error handling
  libceph: fix socket read error handling

fs/ceph/dir.c

@@ -60,6 +60,7 @@ int ceph_init_dentry(struct dentry *dentry)
 	}
 	di->dentry = dentry;
 	di->lease_session = NULL;
+	di->parent_inode = igrab(dentry->d_parent->d_inode);
 	dentry->d_fsdata = di;
 	dentry->d_time = jiffies;
 	ceph_dentry_lru_add(dentry);
@@ -1033,7 +1034,7 @@ static void ceph_dentry_release(struct dentry *dentry)
 	u64 snapid = CEPH_NOSNAP;
 
 	if (!IS_ROOT(dentry)) {
-		parent_inode = dentry->d_parent->d_inode;
+		parent_inode = di->parent_inode;
 		if (parent_inode)
 			snapid = ceph_snap(parent_inode);
 	}
@@ -1058,6 +1059,8 @@ static void ceph_dentry_release(struct dentry *dentry)
 		kmem_cache_free(ceph_dentry_cachep, di);
 		dentry->d_fsdata = NULL;
 	}
+	if (parent_inode)
+		iput(parent_inode);
 }
 
 static int ceph_snapdir_d_revalidate(struct dentry *dentry,

fs/ceph/snap.c

@@ -584,10 +584,14 @@ static void queue_realm_cap_snaps(struct ceph_snap_realm *realm)
 	if (lastinode)
 		iput(lastinode);
 
-	dout("queue_realm_cap_snaps %p %llx children\n", realm, realm->ino);
-	list_for_each_entry(child, &realm->children, child_item)
-		queue_realm_cap_snaps(child);
+	list_for_each_entry(child, &realm->children, child_item) {
+		dout("queue_realm_cap_snaps %p %llx queue child %p %llx\n",
+		     realm, realm->ino, child, child->ino);
+		list_del_init(&child->dirty_item);
+		list_add(&child->dirty_item, &realm->dirty_item);
+	}
 
+	list_del_init(&realm->dirty_item);
 	dout("queue_realm_cap_snaps %p %llx done\n", realm, realm->ino);
 }
 
@@ -683,7 +687,9 @@ more:
 	 * queue cap snaps _after_ we've built the new snap contexts,
 	 * so that i_head_snapc can be set appropriately.
 	 */
-	list_for_each_entry(realm, &dirty_realms, dirty_item) {
+	while (!list_empty(&dirty_realms)) {
+		realm = list_first_entry(&dirty_realms, struct ceph_snap_realm,
+					 dirty_item);
 		queue_realm_cap_snaps(realm);
 	}
 

fs/ceph/super.h

@@ -207,6 +207,7 @@ struct ceph_dentry_info {
 	struct dentry *dentry;
 	u64 time;
 	u64 offset;
+	struct inode *parent_inode;
 };
 
 struct ceph_inode_xattrs_info {

net/ceph/messenger.c

@@ -252,8 +252,12 @@ static int ceph_tcp_recvmsg(struct socket *sock, void *buf, size_t len)
 {
 	struct kvec iov = {buf, len};
 	struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL };
+	int r;
 
-	return kernel_recvmsg(sock, &msg, &iov, 1, len, msg.msg_flags);
+	r = kernel_recvmsg(sock, &msg, &iov, 1, len, msg.msg_flags);
+	if (r == -EAGAIN)
+		r = 0;
+	return r;
 }
 
 /*
@@ -264,13 +268,17 @@ static int ceph_tcp_sendmsg(struct socket *sock, struct kvec *iov,
 		     size_t kvlen, size_t len, int more)
 {
 	struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL };
+	int r;
 
 	if (more)
 		msg.msg_flags |= MSG_MORE;
 	else
 		msg.msg_flags |= MSG_EOR;  /* superfluous, but what the hell */
 
-	return kernel_sendmsg(sock, &msg, iov, kvlen, len);
+	r = kernel_sendmsg(sock, &msg, iov, kvlen, len);
+	if (r == -EAGAIN)
+		r = 0;
+	return r;
 }
 
 
@@ -847,6 +855,8 @@ static int write_partial_msg_pages(struct ceph_connection *con)
 		    (msg->pages || msg->pagelist || msg->bio || in_trail))
 			kunmap(page);
 
+		if (ret == -EAGAIN)
+			ret = 0;
 		if (ret <= 0)
 			goto out;
 
@@ -1737,16 +1747,12 @@ more_kvec:
 	if (con->out_skip) {
 		ret = write_partial_skip(con);
 		if (ret <= 0)
-			goto done;
-		if (ret < 0) {
-			dout("try_write write_partial_skip err %d\n", ret);
-			goto done;
-		}
+			goto out;
 	}
 	if (con->out_kvec_left) {
 		ret = write_partial_kvec(con);
 		if (ret <= 0)
-			goto done;
+			goto out;
 	}
 
 	/* msg pages? */
@@ -1761,11 +1767,11 @@ more_kvec:
 		if (ret == 1)
 			goto more_kvec;  /* we need to send the footer, too! */
 		if (ret == 0)
-			goto done;
+			goto out;
 		if (ret < 0) {
 			dout("try_write write_partial_msg_pages err %d\n",
 			     ret);
-			goto done;
+			goto out;
 		}
 	}
 
@@ -1789,10 +1795,9 @@ do_next:
 	/* Nothing to do! */
 	clear_bit(WRITE_PENDING, &con->state);
 	dout("try_write nothing else to write.\n");
-done:
 	ret = 0;
 out:
-	dout("try_write done on %p\n", con);
+	dout("try_write done on %p ret %d\n", con, ret);
 	return ret;
 }
 
@@ -1821,19 +1826,17 @@ more:
 			dout("try_read connecting\n");
 			ret = read_partial_banner(con);
 			if (ret <= 0)
-				goto done;
-			if (process_banner(con) < 0) {
-				ret = -1;
 				goto out;
-			}
+			ret = process_banner(con);
+			if (ret < 0)
+				goto out;
 		}
 		ret = read_partial_connect(con);
 		if (ret <= 0)
-			goto done;
-		if (process_connect(con) < 0) {
-			ret = -1;
 			goto out;
-		}
+		ret = process_connect(con);
+		if (ret < 0)
+			goto out;
 		goto more;
 	}
 
@@ -1848,7 +1851,7 @@ more:
 		dout("skipping %d / %d bytes\n", skip, -con->in_base_pos);
 		ret = ceph_tcp_recvmsg(con->sock, buf, skip);
 		if (ret <= 0)
-			goto done;
+			goto out;
 		con->in_base_pos += ret;
 		if (con->in_base_pos)
 			goto more;
@@ -1859,7 +1862,7 @@ more:
 		 */
 		ret = ceph_tcp_recvmsg(con->sock, &con->in_tag, 1);
 		if (ret <= 0)
-			goto done;
+			goto out;
 		dout("try_read got tag %d\n", (int)con->in_tag);
 		switch (con->in_tag) {
 		case CEPH_MSGR_TAG_MSG:
@@ -1870,7 +1873,7 @@ more:
 			break;
 		case CEPH_MSGR_TAG_CLOSE:
 			set_bit(CLOSED, &con->state);   /* fixme */
-			goto done;
+			goto out;
 		default:
 			goto bad_tag;
 		}
@@ -1882,13 +1885,12 @@ more:
 			case -EBADMSG:
 				con->error_msg = "bad crc";
 				ret = -EIO;
-				goto out;
+				break;
 			case -EIO:
 				con->error_msg = "io error";
-				goto out;
-			default:
-				goto done;
+				break;
 			}
+			goto out;
 		}
 		if (con->in_tag == CEPH_MSGR_TAG_READY)
 			goto more;
@@ -1898,15 +1900,13 @@ more:
 	if (con->in_tag == CEPH_MSGR_TAG_ACK) {
 		ret = read_partial_ack(con);
 		if (ret <= 0)
-			goto done;
+			goto out;
 		process_ack(con);
 		goto more;
 	}
 
-done:
-	ret = 0;
 out:
-	dout("try_read done on %p\n", con);
+	dout("try_read done on %p ret %d\n", con, ret);
 	return ret;
 
 bad_tag: