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

Pull ceph fixes from Sage Weil:
 "There is an RBD fix for a crash due to the immutable bio changes, an
  error path fix, and a locking fix in the recent redirect support"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  libceph: do not dereference a NULL bio pointer
  libceph: take map_sem for read in handle_reply()
  libceph: factor out logic from ceph_osdc_start_request()
  libceph: fix error handling in ceph_osdc_init()

2 files changed, 60 insertions, 32 deletions

diff --git a/net/ceph/messenger.c b/net/ceph/messenger.c
index 0e478a0f420..30efc5c1862 100644
--- a/net/ceph/messenger.c
+++ b/net/ceph/messenger.c
@@ -840,9 +840,13 @@ static bool ceph_msg_data_bio_advance(struct ceph_msg_data_cursor *cursor,
 
 	if (!cursor->bvec_iter.bi_size) {
 		bio = bio->bi_next;
-		cursor->bvec_iter = bio->bi_iter;
+		cursor->bio = bio;
+		if (bio)
+			cursor->bvec_iter = bio->bi_iter;
+		else
+			memset(&cursor->bvec_iter, 0,
+			       sizeof(cursor->bvec_iter));
 	}
-	cursor->bio = bio;
 
 	if (!cursor->last_piece) {
 		BUG_ON(!cursor->resid);
diff --git a/net/ceph/osd_client.c b/net/ceph/osd_client.c
index 010ff3bd58a..0676f2b199d 100644
--- a/net/ceph/osd_client.c
+++ b/net/ceph/osd_client.c
@@ -1427,6 +1427,40 @@ static void __send_queued(struct ceph_osd_client *osdc)
 }
 
 /*
+ * Caller should hold map_sem for read and request_mutex.
+ */
+static int __ceph_osdc_start_request(struct ceph_osd_client *osdc,
+				     struct ceph_osd_request *req,
+				     bool nofail)
+{
+	int rc;
+
+	__register_request(osdc, req);
+	req->r_sent = 0;
+	req->r_got_reply = 0;
+	rc = __map_request(osdc, req, 0);
+	if (rc < 0) {
+		if (nofail) {
+			dout("osdc_start_request failed map, "
+				" will retry %lld\n", req->r_tid);
+			rc = 0;
+		} else {
+			__unregister_request(osdc, req);
+		}
+		return rc;
+	}
+
+	if (req->r_osd == NULL) {
+		dout("send_request %p no up osds in pg\n", req);
+		ceph_monc_request_next_osdmap(&osdc->client->monc);
+	} else {
+		__send_queued(osdc);
+	}
+
+	return 0;
+}
+
+/*
  * Timeout callback, called every N seconds when 1 or more osd
  * requests has been active for more than N seconds.  When this
  * happens, we ping all OSDs with requests who have timed out to
@@ -1653,6 +1687,7 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
 	osdmap_epoch = ceph_decode_32(&p);
 
 	/* lookup */
+	down_read(&osdc->map_sem);
 	mutex_lock(&osdc->request_mutex);
 	req = __lookup_request(osdc, tid);
 	if (req == NULL) {
@@ -1709,7 +1744,6 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
 		dout("redirect pool %lld\n", redir.oloc.pool);
 
 		__unregister_request(osdc, req);
-		mutex_unlock(&osdc->request_mutex);
 
 		req->r_target_oloc = redir.oloc; /* struct */
 
@@ -1721,10 +1755,10 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
 		 * successfully.  In the future we might want to follow
 		 * original request's nofail setting here.
 		 */
-		err = ceph_osdc_start_request(osdc, req, true);
+		err = __ceph_osdc_start_request(osdc, req, true);
 		BUG_ON(err);
 
-		goto done;
+		goto out_unlock;
 	}
 
 	already_completed = req->r_got_reply;
@@ -1742,8 +1776,7 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
 		req->r_got_reply = 1;
 	} else if ((flags & CEPH_OSD_FLAG_ONDISK) == 0) {
 		dout("handle_reply tid %llu dup ack\n", tid);
-		mutex_unlock(&osdc->request_mutex);
-		goto done;
+		goto out_unlock;
 	}
 
 	dout("handle_reply tid %llu flags %d\n", tid, flags);
@@ -1758,6 +1791,7 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
 		__unregister_request(osdc, req);
 
 	mutex_unlock(&osdc->request_mutex);
+	up_read(&osdc->map_sem);
 
 	if (!already_completed) {
 		if (req->r_unsafe_callback &&
@@ -1775,10 +1809,14 @@ static void handle_reply(struct ceph_osd_client *osdc, struct ceph_msg *msg,
 		complete_request(req);
 	}
 
-done:
+out:
 	dout("req=%p req->r_linger=%d\n", req, req->r_linger);
 	ceph_osdc_put_request(req);
 	return;
+out_unlock:
+	mutex_unlock(&osdc->request_mutex);
+	up_read(&osdc->map_sem);
+	goto out;
 
 bad_put:
 	req->r_result = -EIO;
@@ -1791,6 +1829,7 @@ bad_put:
 	ceph_osdc_put_request(req);
 bad_mutex:
 	mutex_unlock(&osdc->request_mutex);
+	up_read(&osdc->map_sem);
 bad:
 	pr_err("corrupt osd_op_reply got %d %d\n",
 	       (int)msg->front.iov_len, le32_to_cpu(msg->hdr.front_len));
@@ -2351,34 +2390,16 @@ int ceph_osdc_start_request(struct ceph_osd_client *osdc,
 			    struct ceph_osd_request *req,
 			    bool nofail)
 {
-	int rc = 0;
+	int rc;
 
 	down_read(&osdc->map_sem);
 	mutex_lock(&osdc->request_mutex);
-	__register_request(osdc, req);
-	req->r_sent = 0;
-	req->r_got_reply = 0;
-	rc = __map_request(osdc, req, 0);
-	if (rc < 0) {
-		if (nofail) {
-			dout("osdc_start_request failed map, "
-				" will retry %lld\n", req->r_tid);
-			rc = 0;
-		} else {
-			__unregister_request(osdc, req);
-		}
-		goto out_unlock;
-	}
-	if (req->r_osd == NULL) {
-		dout("send_request %p no up osds in pg\n", req);
-		ceph_monc_request_next_osdmap(&osdc->client->monc);
-	} else {
-		__send_queued(osdc);
-	}
-	rc = 0;
-out_unlock:
+
+	rc = __ceph_osdc_start_request(osdc, req, nofail);
+
 	mutex_unlock(&osdc->request_mutex);
 	up_read(&osdc->map_sem);
+
 	return rc;
 }
 EXPORT_SYMBOL(ceph_osdc_start_request);
@@ -2504,9 +2525,12 @@ int ceph_osdc_init(struct ceph_osd_client *osdc, struct ceph_client *client)
 	err = -ENOMEM;
 	osdc->notify_wq = create_singlethread_workqueue("ceph-watch-notify");
 	if (!osdc->notify_wq)
-		goto out_msgpool;
+		goto out_msgpool_reply;
+
 	return 0;
 
+out_msgpool_reply:
+	ceph_msgpool_destroy(&osdc->msgpool_op_reply);
 out_msgpool:
 	ceph_msgpool_destroy(&osdc->msgpool_op);
 out_mempool: