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

Pull Ceph updates from Sage Weil:
 "This changeset has a few main parts:

   - Ilya has finished a huge refactoring effort to sync up the
     client-side logic in libceph with the user-space client code, which
     has evolved significantly over the last couple years, with lots of
     additional behaviors (e.g., how requests are handled when cluster
     is full and transitions from full to non-full).

     This structure of the code is more closely aligned with userspace
     now such that it will be much easier to maintain going forward when
     behavior changes take place.  There are some locking improvements
     bundled in as well.

   - Zheng adds multi-filesystem support (multiple namespaces within the
     same Ceph cluster)

   - Zheng has changed the readdir offsets and directory enumeration so
     that dentry offsets are hash-based and therefore stable across
     directory fragmentation events on the MDS.

   - Zheng has a smorgasbord of bug fixes across fs/ceph"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client: (71 commits)
  ceph: fix wake_up_session_cb()
  ceph: don't use truncate_pagecache() to invalidate read cache
  ceph: SetPageError() for writeback pages if writepages fails
  ceph: handle interrupted ceph_writepage()
  ceph: make ceph_update_writeable_page() uninterruptible
  libceph: make ceph_osdc_wait_request() uninterruptible
  ceph: handle -EAGAIN returned by ceph_update_writeable_page()
  ceph: make fault/page_mkwrite return VM_FAULT_OOM for -ENOMEM
  ceph: block non-fatal signals for fault/page_mkwrite
  ceph: make logical calculation functions return bool
  ceph: tolerate bad i_size for symlink inode
  ceph: improve fragtree change detection
  ceph: keep leaf frag when updating fragtree
  ceph: fix dir_auth check in ceph_fill_dirfrag()
  ceph: don't assume frag tree splits in mds reply are sorted
  ceph: fix inode reference leak
  ceph: using hash value to compose dentry offset
  ceph: don't forbid marking directory complete after forward seek
  ceph: record 'offset' for each entry of readdir result
  ceph: define 'end/complete' in readdir reply as bit flags
  ...

29 files changed, 4758 insertions, 2508 deletions

diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c
index 0ede6d7e2568..81666a56415e 100644
--- a/drivers/block/rbd.c
+++ b/drivers/block/rbd.c
@@ -350,12 +350,12 @@ struct rbd_device {
 	struct rbd_spec		*spec;
 	struct rbd_options	*opts;
 
-	char			*header_name;
+	struct ceph_object_id	header_oid;
+	struct ceph_object_locator header_oloc;
 
 	struct ceph_file_layout	layout;
 
-	struct ceph_osd_event   *watch_event;
-	struct rbd_obj_request	*watch_request;
+	struct ceph_osd_linger_request *watch_handle;
 
 	struct rbd_spec		*parent_spec;
 	u64			parent_overlap;
@@ -1596,12 +1596,6 @@ static int rbd_obj_request_wait(struct rbd_obj_request *obj_request)
 	return __rbd_obj_request_wait(obj_request, 0);
 }
 
-static int rbd_obj_request_wait_timeout(struct rbd_obj_request *obj_request,
-					unsigned long timeout)
-{
-	return __rbd_obj_request_wait(obj_request, timeout);
-}
-
 static void rbd_img_request_complete(struct rbd_img_request *img_request)
 {
 
@@ -1751,12 +1745,6 @@ static void rbd_obj_request_complete(struct rbd_obj_request *obj_request)
 		complete_all(&obj_request->completion);
 }
 
-static void rbd_osd_trivial_callback(struct rbd_obj_request *obj_request)
-{
-	dout("%s: obj %p\n", __func__, obj_request);
-	obj_request_done_set(obj_request);
-}
-
 static void rbd_osd_read_callback(struct rbd_obj_request *obj_request)
 {
 	struct rbd_img_request *img_request = NULL;
@@ -1828,13 +1816,12 @@ static void rbd_osd_call_callback(struct rbd_obj_request *obj_request)
 		obj_request_done_set(obj_request);
 }
 
-static void rbd_osd_req_callback(struct ceph_osd_request *osd_req,
-				struct ceph_msg *msg)
+static void rbd_osd_req_callback(struct ceph_osd_request *osd_req)
 {
 	struct rbd_obj_request *obj_request = osd_req->r_priv;
 	u16 opcode;
 
-	dout("%s: osd_req %p msg %p\n", __func__, osd_req, msg);
+	dout("%s: osd_req %p\n", __func__, osd_req);
 	rbd_assert(osd_req == obj_request->osd_req);
 	if (obj_request_img_data_test(obj_request)) {
 		rbd_assert(obj_request->img_request);
@@ -1878,10 +1865,6 @@ static void rbd_osd_req_callback(struct ceph_osd_request *osd_req,
 	case CEPH_OSD_OP_CALL:
 		rbd_osd_call_callback(obj_request);
 		break;
-	case CEPH_OSD_OP_NOTIFY_ACK:
-	case CEPH_OSD_OP_WATCH:
-		rbd_osd_trivial_callback(obj_request);
-		break;
 	default:
 		rbd_warn(NULL, "%s: unsupported op %hu",
 			obj_request->object_name, (unsigned short) opcode);
@@ -1896,27 +1879,17 @@ static void rbd_osd_req_format_read(struct rbd_obj_request *obj_request)
 {
 	struct rbd_img_request *img_request = obj_request->img_request;
 	struct ceph_osd_request *osd_req = obj_request->osd_req;
-	u64 snap_id;
 
-	rbd_assert(osd_req != NULL);
-
-	snap_id = img_request ? img_request->snap_id : CEPH_NOSNAP;
-	ceph_osdc_build_request(osd_req, obj_request->offset,
-			NULL, snap_id, NULL);
+	if (img_request)
+		osd_req->r_snapid = img_request->snap_id;
 }
 
 static void rbd_osd_req_format_write(struct rbd_obj_request *obj_request)
 {
-	struct rbd_img_request *img_request = obj_request->img_request;
 	struct ceph_osd_request *osd_req = obj_request->osd_req;
-	struct ceph_snap_context *snapc;
-	struct timespec mtime = CURRENT_TIME;
 
-	rbd_assert(osd_req != NULL);
-
-	snapc = img_request ? img_request->snapc : NULL;
-	ceph_osdc_build_request(osd_req, obj_request->offset,
-			snapc, CEPH_NOSNAP, &mtime);
+	osd_req->r_mtime = CURRENT_TIME;
+	osd_req->r_data_offset = obj_request->offset;
 }
 
 /*
@@ -1954,7 +1927,7 @@ static struct ceph_osd_request *rbd_osd_req_create(
 	osd_req = ceph_osdc_alloc_request(osdc, snapc, num_ops, false,
 					  GFP_NOIO);
 	if (!osd_req)
-		return NULL;	/* ENOMEM */
+		goto fail;
 
 	if (op_type == OBJ_OP_WRITE || op_type == OBJ_OP_DISCARD)
 		osd_req->r_flags = CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK;
@@ -1965,9 +1938,18 @@ static struct ceph_osd_request *rbd_osd_req_create(
 	osd_req->r_priv = obj_request;
 
 	osd_req->r_base_oloc.pool = ceph_file_layout_pg_pool(rbd_dev->layout);
-	ceph_oid_set_name(&osd_req->r_base_oid, obj_request->object_name);
+	if (ceph_oid_aprintf(&osd_req->r_base_oid, GFP_NOIO, "%s",
+			     obj_request->object_name))
+		goto fail;
+
+	if (ceph_osdc_alloc_messages(osd_req, GFP_NOIO))
+		goto fail;
 
 	return osd_req;
+
+fail:
+	ceph_osdc_put_request(osd_req);
+	return NULL;
 }
 
 /*
@@ -2003,16 +1985,25 @@ rbd_osd_req_create_copyup(struct rbd_obj_request *obj_request)
 	osd_req = ceph_osdc_alloc_request(osdc, snapc, num_osd_ops,
 						false, GFP_NOIO);
 	if (!osd_req)
-		return NULL;	/* ENOMEM */
+		goto fail;
 
 	osd_req->r_flags = CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK;
 	osd_req->r_callback = rbd_osd_req_callback;
 	osd_req->r_priv = obj_request;
 
 	osd_req->r_base_oloc.pool = ceph_file_layout_pg_pool(rbd_dev->layout);
-	ceph_oid_set_name(&osd_req->r_base_oid, obj_request->object_name);
+	if (ceph_oid_aprintf(&osd_req->r_base_oid, GFP_NOIO, "%s",
+			     obj_request->object_name))
+		goto fail;
+
+	if (ceph_osdc_alloc_messages(osd_req, GFP_NOIO))
+		goto fail;
 
 	return osd_req;
+
+fail:
+	ceph_osdc_put_request(osd_req);
+	return NULL;
 }
 
 
@@ -2973,17 +2964,20 @@ static int rbd_img_request_submit(struct rbd_img_request *img_request)
 {
 	struct rbd_obj_request *obj_request;
 	struct rbd_obj_request *next_obj_request;
+	int ret = 0;
 
 	dout("%s: img %p\n", __func__, img_request);
-	for_each_obj_request_safe(img_request, obj_request, next_obj_request) {
-		int ret;
 
+	rbd_img_request_get(img_request);
+	for_each_obj_request_safe(img_request, obj_request, next_obj_request) {
 		ret = rbd_img_obj_request_submit(obj_request);
 		if (ret)
-			return ret;
+			goto out_put_ireq;
 	}
 
-	return 0;
+out_put_ireq:
+	rbd_img_request_put(img_request);
+	return ret;
 }
 
 static void rbd_img_parent_read_callback(struct rbd_img_request *img_request)
@@ -3090,45 +3084,18 @@ out_err:
 	obj_request_done_set(obj_request);
 }
 
-static int rbd_obj_notify_ack_sync(struct rbd_device *rbd_dev, u64 notify_id)
-{
-	struct rbd_obj_request *obj_request;
-	struct ceph_osd_client *osdc = &rbd_dev->rbd_client->client->osdc;
-	int ret;
-
-	obj_request = rbd_obj_request_create(rbd_dev->header_name, 0, 0,
-							OBJ_REQUEST_NODATA);
-	if (!obj_request)
-		return -ENOMEM;
-
-	ret = -ENOMEM;
-	obj_request->osd_req = rbd_osd_req_create(rbd_dev, OBJ_OP_READ, 1,
-						  obj_request);
-	if (!obj_request->osd_req)
-		goto out;
-
-	osd_req_op_watch_init(obj_request->osd_req, 0, CEPH_OSD_OP_NOTIFY_ACK,
-					notify_id, 0, 0);
-	rbd_osd_req_format_read(obj_request);
+static int rbd_dev_header_watch_sync(struct rbd_device *rbd_dev);
+static void __rbd_dev_header_unwatch_sync(struct rbd_device *rbd_dev);
 
-	ret = rbd_obj_request_submit(osdc, obj_request);
-	if (ret)
-		goto out;
-	ret = rbd_obj_request_wait(obj_request);
-out:
-	rbd_obj_request_put(obj_request);
-
-	return ret;
-}
-
-static void rbd_watch_cb(u64 ver, u64 notify_id, u8 opcode, void *data)
+static void rbd_watch_cb(void *arg, u64 notify_id, u64 cookie,
+			 u64 notifier_id, void *data, size_t data_len)
 {
-	struct rbd_device *rbd_dev = (struct rbd_device *)data;
+	struct rbd_device *rbd_dev = arg;
+	struct ceph_osd_client *osdc = &rbd_dev->rbd_client->client->osdc;
 	int ret;
 
-	dout("%s: \"%s\" notify_id %llu opcode %u\n", __func__,
-		rbd_dev->header_name, (unsigned long long)notify_id,
-		(unsigned int)opcode);
+	dout("%s rbd_dev %p cookie %llu notify_id %llu\n", __func__, rbd_dev,
+	     cookie, notify_id);
 
 	/*
 	 * Until adequate refresh error handling is in place, there is
@@ -3140,63 +3107,31 @@ static void rbd_watch_cb(u64 ver, u64 notify_id, u8 opcode, void *data)
 	if (ret)
 		rbd_warn(rbd_dev, "refresh failed: %d", ret);
 
-	ret = rbd_obj_notify_ack_sync(rbd_dev, notify_id);
+	ret = ceph_osdc_notify_ack(osdc, &rbd_dev->header_oid,
+				   &rbd_dev->header_oloc, notify_id, cookie,
+				   NULL, 0);
 	if (ret)
 		rbd_warn(rbd_dev, "notify_ack ret %d", ret);
 }
 
-/*
- * Send a (un)watch request and wait for the ack.  Return a request
- * with a ref held on success or error.
- */
-static struct rbd_obj_request *rbd_obj_watch_request_helper(
-						struct rbd_device *rbd_dev,
-						bool watch)
+static void rbd_watch_errcb(void *arg, u64 cookie, int err)
 {
-	struct ceph_osd_client *osdc = &rbd_dev->rbd_client->client->osdc;
-	struct ceph_options *opts = osdc->client->options;
-	struct rbd_obj_request *obj_request;
+	struct rbd_device *rbd_dev = arg;
 	int ret;
 
-	obj_request = rbd_obj_request_create(rbd_dev->header_name, 0, 0,
-					     OBJ_REQUEST_NODATA);
-	if (!obj_request)
-		return ERR_PTR(-ENOMEM);
-
-	obj_request->osd_req = rbd_osd_req_create(rbd_dev, OBJ_OP_WRITE, 1,
-						  obj_request);
-	if (!obj_request->osd_req) {
-		ret = -ENOMEM;
-		goto out;
-	}
-
-	osd_req_op_watch_init(obj_request->osd_req, 0, CEPH_OSD_OP_WATCH,
-			      rbd_dev->watch_event->cookie, 0, watch);
-	rbd_osd_req_format_write(obj_request);
+	rbd_warn(rbd_dev, "encountered watch error: %d", err);
 
-	if (watch)
-		ceph_osdc_set_request_linger(osdc, obj_request->osd_req);
-
-	ret = rbd_obj_request_submit(osdc, obj_request);
-	if (ret)
-		goto out;
+	__rbd_dev_header_unwatch_sync(rbd_dev);
 
-	ret = rbd_obj_request_wait_timeout(obj_request, opts->mount_timeout);
-	if (ret)
-		goto out;
-
-	ret = obj_request->result;
+	ret = rbd_dev_header_watch_sync(rbd_dev);
 	if (ret) {
-		if (watch)
-			rbd_obj_request_end(obj_request);
-		goto out;
+		rbd_warn(rbd_dev, "failed to reregister watch: %d", ret);
+		return;
 	}
 
-	return obj_request;
-
-out:
-	rbd_obj_request_put(obj_request);
-	return ERR_PTR(ret);
+	ret = rbd_dev_refresh(rbd_dev);
+	if (ret)
+		rbd_warn(rbd_dev, "reregisteration refresh failed: %d", ret);
 }
 
 /*
@@ -3205,35 +3140,33 @@ out:
 static int rbd_dev_header_watch_sync(struct rbd_device *rbd_dev)
 {
 	struct ceph_osd_client *osdc = &rbd_dev->rbd_client->client->osdc;
-	struct rbd_obj_request *obj_request;
-	int ret;
+	struct ceph_osd_linger_request *handle;
 
-	rbd_assert(!rbd_dev->watch_event);
-	rbd_assert(!rbd_dev->watch_request);
+	rbd_assert(!rbd_dev->watch_handle);
 
-	ret = ceph_osdc_create_event(osdc, rbd_watch_cb, rbd_dev,
-				     &rbd_dev->watch_event);
-	if (ret < 0)
-		return ret;
+	handle = ceph_osdc_watch(osdc, &rbd_dev->header_oid,
+				 &rbd_dev->header_oloc, rbd_watch_cb,
+				 rbd_watch_errcb, rbd_dev);
+	if (IS_ERR(handle))
+		return PTR_ERR(handle);
 
-	obj_request = rbd_obj_watch_request_helper(rbd_dev, true);
-	if (IS_ERR(obj_request)) {
-		ceph_osdc_cancel_event(rbd_dev->watch_event);
-		rbd_dev->watch_event = NULL;
-		return PTR_ERR(obj_request);
-	}
+	rbd_dev->watch_handle = handle;
+	return 0;
+}
 
-	/*
-	 * A watch request is set to linger, so the underlying osd
-	 * request won't go away until we unregister it.  We retain
-	 * a pointer to the object request during that time (in
-	 * rbd_dev->watch_request), so we'll keep a reference to it.
-	 * We'll drop that reference after we've unregistered it in
-	 * rbd_dev_header_unwatch_sync().
-	 */
-	rbd_dev->watch_request = obj_request;
+static void __rbd_dev_header_unwatch_sync(struct rbd_device *rbd_dev)
+{
+	struct ceph_osd_client *osdc = &rbd_dev->rbd_client->client->osdc;
+	int ret;
 
-	return 0;
+	if (!rbd_dev->watch_handle)
+		return;
+
+	ret = ceph_osdc_unwatch(osdc, rbd_dev->watch_handle);
+	if (ret)
+		rbd_warn(rbd_dev, "failed to unwatch: %d", ret);
+
+	rbd_dev->watch_handle = NULL;
 }
 
 /*
@@ -3241,24 +3174,7 @@ static int rbd_dev_header_watch_sync(struct rbd_device *rbd_dev)
  */
 static void rbd_dev_header_unwatch_sync(struct rbd_device *rbd_dev)
 {
-	struct rbd_obj_request *obj_request;
-
-	rbd_assert(rbd_dev->watch_event);
-	rbd_assert(rbd_dev->watch_request);
-
-	rbd_obj_request_end(rbd_dev->watch_request);
-	rbd_obj_request_put(rbd_dev->watch_request);
-	rbd_dev->watch_request = NULL;
-
-	obj_request = rbd_obj_watch_request_helper(rbd_dev, false);
-	if (!IS_ERR(obj_request))
-		rbd_obj_request_put(obj_request);
-	else
-		rbd_warn(rbd_dev, "unable to tear down watch request (%ld)",
-			 PTR_ERR(obj_request));
-
-	ceph_osdc_cancel_event(rbd_dev->watch_event);
-	rbd_dev->watch_event = NULL;
+	__rbd_dev_header_unwatch_sync(rbd_dev);
 
 	dout("%s flushing notifies\n", __func__);
 	ceph_osdc_flush_notifies(&rbd_dev->rbd_client->client->osdc);
@@ -3591,7 +3507,7 @@ static int rbd_dev_v1_header_info(struct rbd_device *rbd_dev)
 		if (!ondisk)
 			return -ENOMEM;
 
-		ret = rbd_obj_read_sync(rbd_dev, rbd_dev->header_name,
+		ret = rbd_obj_read_sync(rbd_dev, rbd_dev->header_oid.name,
 				       0, size, ondisk);
 		if (ret < 0)
 			goto out;
@@ -4033,6 +3949,8 @@ static void rbd_dev_release(struct device *dev)
 	struct rbd_device *rbd_dev = dev_to_rbd_dev(dev);
 	bool need_put = !!rbd_dev->opts;
 
+	ceph_oid_destroy(&rbd_dev->header_oid);
+
 	rbd_put_client(rbd_dev->rbd_client);
 	rbd_spec_put(rbd_dev->spec);
 	kfree(rbd_dev->opts);
@@ -4063,6 +3981,9 @@ static struct rbd_device *rbd_dev_create(struct rbd_client *rbdc,
 	INIT_LIST_HEAD(&rbd_dev->node);
 	init_rwsem(&rbd_dev->header_rwsem);
 
+	ceph_oid_init(&rbd_dev->header_oid);
+	ceph_oloc_init(&rbd_dev->header_oloc);
+
 	rbd_dev->dev.bus = &rbd_bus_type;
 	rbd_dev->dev.type = &rbd_device_type;
 	rbd_dev->dev.parent = &rbd_root_dev;
@@ -4111,7 +4032,7 @@ static int _rbd_dev_v2_snap_size(struct rbd_device *rbd_dev, u64 snap_id,
 		__le64 size;
 	} __attribute__ ((packed)) size_buf = { 0 };
 
-	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_name,
+	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_oid.name,
 				"rbd", "get_size",
 				&snapid, sizeof (snapid),
 				&size_buf, sizeof (size_buf));
@@ -4151,7 +4072,7 @@ static int rbd_dev_v2_object_prefix(struct rbd_device *rbd_dev)
 	if (!reply_buf)
 		return -ENOMEM;
 
-	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_name,
+	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_oid.name,
 				"rbd", "get_object_prefix", NULL, 0,
 				reply_buf, RBD_OBJ_PREFIX_LEN_MAX);
 	dout("%s: rbd_obj_method_sync returned %d\n", __func__, ret);
@@ -4186,7 +4107,7 @@ static int _rbd_dev_v2_snap_features(struct rbd_device *rbd_dev, u64 snap_id,
 	u64 unsup;
 	int ret;
 
-	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_name,
+	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_oid.name,
 				"rbd", "get_features",
 				&snapid, sizeof (snapid),
 				&features_buf, sizeof (features_buf));
@@ -4248,7 +4169,7 @@ static int rbd_dev_v2_parent_info(struct rbd_device *rbd_dev)
 	}
 
 	snapid = cpu_to_le64(rbd_dev->spec->snap_id);
-	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_name,
+	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_oid.name,
 				"rbd", "get_parent",
 				&snapid, sizeof (snapid),
 				reply_buf, size);
@@ -4351,7 +4272,7 @@ static int rbd_dev_v2_striping_info(struct rbd_device *rbd_dev)
 	u64 stripe_count;
 	int ret;
 
-	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_name,
+	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_oid.name,
 				"rbd", "get_stripe_unit_count", NULL, 0,
 				(char *)&striping_info_buf, size);
 	dout("%s: rbd_obj_method_sync returned %d\n", __func__, ret);
@@ -4599,7 +4520,7 @@ static int rbd_dev_v2_snap_context(struct rbd_device *rbd_dev)
 	if (!reply_buf)
 		return -ENOMEM;
 
-	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_name,
+	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_oid.name,
 				"rbd", "get_snapcontext", NULL, 0,
 				reply_buf, size);
 	dout("%s: rbd_obj_method_sync returned %d\n", __func__, ret);
@@ -4664,7 +4585,7 @@ static const char *rbd_dev_v2_snap_name(struct rbd_device *rbd_dev,
 		return ERR_PTR(-ENOMEM);
 
 	snapid = cpu_to_le64(snap_id);
-	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_name,
+	ret = rbd_obj_method_sync(rbd_dev, rbd_dev->header_oid.name,
 				"rbd", "get_snapshot_name",
 				&snapid, sizeof (snapid),
 				reply_buf, size);
@@ -4975,13 +4896,13 @@ static int rbd_add_get_pool_id(struct rbd_client *rbdc, const char *pool_name)
 again:
 	ret = ceph_pg_poolid_by_name(rbdc->client->osdc.osdmap, pool_name);
 	if (ret == -ENOENT && tries++ < 1) {
-		ret = ceph_monc_do_get_version(&rbdc->client->monc, "osdmap",
-					       &newest_epoch);
+		ret = ceph_monc_get_version(&rbdc->client->monc, "osdmap",
+					    &newest_epoch);
 		if (ret < 0)
 			return ret;
 
 		if (rbdc->client->osdc.osdmap->epoch < newest_epoch) {
-			ceph_monc_request_next_osdmap(&rbdc->client->monc);
+			ceph_osdc_maybe_request_map(&rbdc->client->osdc);
 			(void) ceph_monc_wait_osdmap(&rbdc->client->monc,
 						     newest_epoch,
 						     opts->mount_timeout);
@@ -5260,35 +5181,26 @@ err_out_unlock:
 static int rbd_dev_header_name(struct rbd_device *rbd_dev)
 {
 	struct rbd_spec *spec = rbd_dev->spec;
-	size_t size;
+	int ret;
 
 	/* Record the header object name for this rbd image. */
 
 	rbd_assert(rbd_image_format_valid(rbd_dev->image_format));
 
+	rbd_dev->header_oloc.pool = ceph_file_layout_pg_pool(rbd_dev->layout);
 	if (rbd_dev->image_format == 1)
-		size = strlen(spec->image_name) + sizeof (RBD_SUFFIX);
+		ret = ceph_oid_aprintf(&rbd_dev->header_oid, GFP_KERNEL, "%s%s",
+				       spec->image_name, RBD_SUFFIX);
 	else
-		size = sizeof (RBD_HEADER_PREFIX) + strlen(spec->image_id);
-
-	rbd_dev->header_name = kmalloc(size, GFP_KERNEL);
-	if (!rbd_dev->header_name)
-		return -ENOMEM;
+		ret = ceph_oid_aprintf(&rbd_dev->header_oid, GFP_KERNEL, "%s%s",
+				       RBD_HEADER_PREFIX, spec->image_id);
 
-	if (rbd_dev->image_format == 1)
-		sprintf(rbd_dev->header_name, "%s%s",
-			spec->image_name, RBD_SUFFIX);
-	else
-		sprintf(rbd_dev->header_name, "%s%s",
-			RBD_HEADER_PREFIX, spec->image_id);
-	return 0;
+	return ret;
 }
 
 static void rbd_dev_image_release(struct rbd_device *rbd_dev)
 {
 	rbd_dev_unprobe(rbd_dev);
-	kfree(rbd_dev->header_name);
-	rbd_dev->header_name = NULL;
 	rbd_dev->image_format = 0;
 	kfree(rbd_dev->spec->image_id);
 	rbd_dev->spec->image_id = NULL;
@@ -5327,7 +5239,7 @@ static int rbd_dev_image_probe(struct rbd_device *rbd_dev, int depth)
 				pr_info("image %s/%s does not exist\n",
 					rbd_dev->spec->pool_name,
 					rbd_dev->spec->image_name);
-			goto out_header_name;
+			goto err_out_format;
 		}
 	}
 
@@ -5373,7 +5285,7 @@ static int rbd_dev_image_probe(struct rbd_device *rbd_dev, int depth)
 		goto err_out_probe;
 
 	dout("discovered format %u image, header name is %s\n",
-		rbd_dev->image_format, rbd_dev->header_name);
+		rbd_dev->image_format, rbd_dev->header_oid.name);
 	return 0;
 
 err_out_probe:
@@ -5381,9 +5293,6 @@ err_out_probe:
 err_out_watch:
 	if (!depth)
 		rbd_dev_header_unwatch_sync(rbd_dev);
-out_header_name:
-	kfree(rbd_dev->header_name);
-	rbd_dev->header_name = NULL;
 err_out_format:
 	rbd_dev->image_format = 0;
 	kfree(rbd_dev->spec->image_id);
diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index 43098cd9602b..eeb71e5de27a 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -257,12 +257,12 @@ static int ceph_readpage(struct file *filp, struct page *page)
 /*
  * Finish an async read(ahead) op.
  */
-static void finish_read(struct ceph_osd_request *req, struct ceph_msg *msg)
+static void finish_read(struct ceph_osd_request *req)
 {
 	struct inode *inode = req->r_inode;
 	struct ceph_osd_data *osd_data;
-	int rc = req->r_result;
-	int bytes = le32_to_cpu(msg->hdr.data_len);
+	int rc = req->r_result <= 0 ? req->r_result : 0;
+	int bytes = req->r_result >= 0 ? req->r_result : 0;
 	int num_pages;
 	int i;
 
@@ -376,8 +376,6 @@ static int start_read(struct inode *inode, struct list_head *page_list, int max)
 	req->r_callback = finish_read;
 	req->r_inode = inode;
 
-	ceph_osdc_build_request(req, off, NULL, vino.snap, NULL);
-
 	dout("start_read %p starting %p %lld~%lld\n", inode, req, off, len);
 	ret = ceph_osdc_start_request(osdc, req, false);
 	if (ret < 0)
@@ -546,11 +544,21 @@ static int writepage_nounlock(struct page *page, struct writeback_control *wbc)
 				   truncate_seq, truncate_size,
 				   &inode->i_mtime, &page, 1);
 	if (err < 0) {
-		dout("writepage setting page/mapping error %d %p\n", err, page);
+		struct writeback_control tmp_wbc;
+		if (!wbc)
+			wbc = &tmp_wbc;
+		if (err == -ERESTARTSYS) {
+			/* killed by SIGKILL */
+			dout("writepage interrupted page %p\n", page);
+			redirty_page_for_writepage(wbc, page);
+			end_page_writeback(page);
+			goto out;
+		}
+		dout("writepage setting page/mapping error %d %p\n",
+		     err, page);
 		SetPageError(page);
 		mapping_set_error(&inode->i_data, err);
-		if (wbc)
-			wbc->pages_skipped++;
+		wbc->pages_skipped++;
 	} else {
 		dout("writepage cleaned page %p\n", page);
 		err = 0;  /* vfs expects us to return 0 */
@@ -571,12 +579,16 @@ static int ceph_writepage(struct page *page, struct writeback_control *wbc)
 	BUG_ON(!inode);
 	ihold(inode);
 	err = writepage_nounlock(page, wbc);
+	if (err == -ERESTARTSYS) {
+		/* direct memory reclaimer was killed by SIGKILL. return 0
+		 * to prevent caller from setting mapping/page error */
+		err = 0;
+	}
 	unlock_page(page);
 	iput(inode);
 	return err;
 }
 
-
 /*
  * lame release_pages helper.  release_pages() isn't exported to
  * modules.
@@ -600,8 +612,7 @@ static void ceph_release_pages(struct page **pages, int num)
  * If we get an error, set the mapping error bit, but not the individual
  * page error bits.
  */
-static void writepages_finish(struct ceph_osd_request *req,
-			      struct ceph_msg *msg)
+static void writepages_finish(struct ceph_osd_request *req)
 {
 	struct inode *inode = req->r_inode;
 	struct ceph_inode_info *ci = ceph_inode(inode);
@@ -615,7 +626,6 @@ static void writepages_finish(struct ceph_osd_request *req,
 	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 	bool remove_page;
 
-
 	dout("writepages_finish %p rc %d\n", inode, rc);
 	if (rc < 0)
 		mapping_set_error(mapping, rc);
@@ -650,6 +660,9 @@ static void writepages_finish(struct ceph_osd_request *req,
 				clear_bdi_congested(&fsc->backing_dev_info,
 						    BLK_RW_ASYNC);
 
+			if (rc < 0)
+				SetPageError(page);
+
 			ceph_put_snap_context(page_snap_context(page));
 			page->private = 0;
 			ClearPagePrivate(page);
@@ -718,8 +731,11 @@ static int ceph_writepages_start(struct address_space *mapping,
 	     (wbc->sync_mode == WB_SYNC_ALL ? "ALL" : "HOLD"));
 
 	if (ACCESS_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
-		pr_warn("writepage_start %p on forced umount\n", inode);
-		truncate_pagecache(inode, 0);
+		if (ci->i_wrbuffer_ref > 0) {
+			pr_warn_ratelimited(
+				"writepage_start %p %lld forced umount\n",
+				inode, ceph_ino(inode));
+		}
 		mapping_set_error(mapping, -EIO);
 		return -EIO; /* we're in a forced umount, don't write! */
 	}
@@ -1063,10 +1079,7 @@ new_request:
 			pages = NULL;
 		}
 
-		vino = ceph_vino(inode);
-		ceph_osdc_build_request(req, offset, snapc, vino.snap,
-					&inode->i_mtime);
-
+		req->r_mtime = inode->i_mtime;
 		rc = ceph_osdc_start_request(&fsc->client->osdc, req, true);
 		BUG_ON(rc);
 		req = NULL;
@@ -1099,8 +1112,7 @@ release_pvec_pages:
 		mapping->writeback_index = index;
 
 out:
-	if (req)
-		ceph_osdc_put_request(req);
+	ceph_osdc_put_request(req);
 	ceph_put_snap_context(snapc);
 	dout("writepages done, rc = %d\n", rc);
 	return rc;
@@ -1134,6 +1146,7 @@ static int ceph_update_writeable_page(struct file *file,
 			    struct page *page)
 {
 	struct inode *inode = file_inode(file);
+	struct ceph_fs_client *fsc = ceph_inode_to_client(inode);
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	loff_t page_off = pos & PAGE_MASK;
 	int pos_in_page = pos & ~PAGE_MASK;
@@ -1142,6 +1155,12 @@ static int ceph_update_writeable_page(struct file *file,
 	int r;
 	struct ceph_snap_context *snapc, *oldest;
 
+	if (ACCESS_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
+		dout(" page %p forced umount\n", page);
+		unlock_page(page);
+		return -EIO;
+	}
+
 retry_locked:
 	/* writepages currently holds page lock, but if we change that later, */
 	wait_on_page_writeback(page);
@@ -1165,7 +1184,7 @@ retry_locked:
 			snapc = ceph_get_snap_context(snapc);
 			unlock_page(page);
 			ceph_queue_writeback(inode);
-			r = wait_event_interruptible(ci->i_cap_wq,
+			r = wait_event_killable(ci->i_cap_wq,
 			       context_is_writeable_or_written(inode, snapc));
 			ceph_put_snap_context(snapc);
 			if (r == -ERESTARTSYS)
@@ -1311,6 +1330,17 @@ const struct address_space_operations ceph_aops = {
 	.direct_IO = ceph_direct_io,
 };
 
+static void ceph_block_sigs(sigset_t *oldset)
+{
+	sigset_t mask;
+	siginitsetinv(&mask, sigmask(SIGKILL));
+	sigprocmask(SIG_BLOCK, &mask, oldset);
+}
+
+static void ceph_restore_sigs(sigset_t *oldset)
+{
+	sigprocmask(SIG_SETMASK, oldset, NULL);
+}
 
 /*
  * vm ops
@@ -1323,6 +1353,9 @@ static int ceph_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 	struct page *pinned_page = NULL;
 	loff_t off = vmf->pgoff << PAGE_SHIFT;
 	int want, got, ret;
+	sigset_t oldset;
+
+	ceph_block_sigs(&oldset);
 
 	dout("filemap_fault %p %llx.%llx %llu~%zd trying to get caps\n",
 	     inode, ceph_vinop(inode), off, (size_t)PAGE_SIZE);
@@ -1330,17 +1363,12 @@ static int ceph_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 		want = CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_LAZYIO;
 	else
 		want = CEPH_CAP_FILE_CACHE;
-	while (1) {
-		got = 0;
-		ret = ceph_get_caps(ci, CEPH_CAP_FILE_RD, want,
-				    -1, &got, &pinned_page);
-		if (ret == 0)
-			break;
-		if (ret != -ERESTARTSYS) {
-			WARN_ON(1);
-			return VM_FAULT_SIGBUS;
-		}
-	}
+
+	got = 0;
+	ret = ceph_get_caps(ci, CEPH_CAP_FILE_RD, want, -1, &got, &pinned_page);
+	if (ret < 0)
+		goto out_restore;
+
 	dout("filemap_fault %p %llu~%zd got cap refs on %s\n",
 	     inode, off, (size_t)PAGE_SIZE, ceph_cap_string(got));
 
@@ -1357,7 +1385,7 @@ static int ceph_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 	ceph_put_cap_refs(ci, got);
 
 	if (ret != -EAGAIN)
-		return ret;
+		goto out_restore;
 
 	/* read inline data */
 	if (off >= PAGE_SIZE) {
@@ -1371,15 +1399,18 @@ static int ceph_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 						~__GFP_FS));
 		if (!page) {
 			ret = VM_FAULT_OOM;
-			goto out;
+			goto out_inline;
 		}
 		ret1 = __ceph_do_getattr(inode, page,
 					 CEPH_STAT_CAP_INLINE_DATA, true);
 		if (ret1 < 0 || off >= i_size_read(inode)) {
 			unlock_page(page);
 			put_page(page);
-			ret = VM_FAULT_SIGBUS;
-			goto out;
+			if (ret1 < 0)
+				ret = ret1;
+			else
+				ret = VM_FAULT_SIGBUS;
+			goto out_inline;
 		}
 		if (ret1 < PAGE_SIZE)
 			zero_user_segment(page, ret1, PAGE_SIZE);
@@ -1388,10 +1419,15 @@ static int ceph_filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
 		SetPageUpt