Merge tag 'ceph-for-4.11-rc1' of git://github.com/ceph/ceph-client

Pull ceph updates from Ilya Dryomov:
 "This time around we have:

   - support for rbd data-pool feature, which enables rbd images on
     erasure-coded pools (myself). CEPH_PG_MAX_SIZE has been bumped to
     allow erasure-coded profiles with k+m up to 32.

   - a patch for ceph_d_revalidate() performance regression introduced
     in 4.9, along with some cleanups in the area (Jeff Layton)

   - a set of fixes for unsafe ->d_parent accesses in CephFS (Jeff
     Layton)

   - buffered reads are now processed in rsize windows instead of rasize
     windows (Andreas Gerstmayr). The new default for rsize mount option
     is 64M.

   - ack vs commit distinction is gone, greatly simplifying ->fsync()
     and MOSDOpReply handling code (myself)

  ... also a few filesystem bug fixes from Zheng, a CRUSH sync up (CRUSH
  computations are still serialized though) and several minor fixes and
  cleanups all over"

* tag 'ceph-for-4.11-rc1' of git://github.com/ceph/ceph-client: (52 commits)
  libceph, rbd, ceph: WRITE | ONDISK -> WRITE
  libceph: get rid of ack vs commit
  ceph: remove special ack vs commit behavior
  ceph: tidy some white space in get_nonsnap_parent()
  crush: fix dprintk compilation
  crush: do is_out test only if we do not collide
  ceph: remove req from unsafe list when unregistering it
  rbd: constify device_type structure
  rbd: kill obj_request->object_name and rbd_segment_name_cache
  rbd: store and use obj_request->object_no
  rbd: RBD_V{1,2}_DATA_FORMAT macros
  rbd: factor out __rbd_osd_req_create()
  rbd: set offset and length outside of rbd_obj_request_create()
  rbd: support for data-pool feature
  rbd: introduce rbd_init_layout()
  rbd: use rbd_obj_bytes() more
  rbd: remove now unused rbd_obj_request_wait() and helpers
  rbd: switch rbd_obj_method_sync() to ceph_osdc_call()
  libceph: pass reply buffer length through ceph_osdc_call()
  rbd: do away with obj_request in rbd_obj_read_sync()
  ...

13 files changed, 287 insertions, 306 deletions

diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index 7ce35aec8c76..f297a9e18642 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -391,6 +391,7 @@ static int start_read(struct inode *inode, struct list_head *page_list, int max)
 			nr_pages = i;
 			if (nr_pages > 0) {
 				len = nr_pages << PAGE_SHIFT;
+				osd_req_op_extent_update(req, 0, len);
 				break;
 			}
 			goto out_pages;
@@ -771,7 +772,7 @@ static int ceph_writepages_start(struct address_space *mapping,
 	     wbc->sync_mode == WB_SYNC_NONE ? "NONE" :
 	     (wbc->sync_mode == WB_SYNC_ALL ? "ALL" : "HOLD"));
 
-	if (ACCESS_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
+	if (READ_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
 		if (ci->i_wrbuffer_ref > 0) {
 			pr_warn_ratelimited(
 				"writepage_start %p %lld forced umount\n",
@@ -1017,8 +1018,7 @@ new_request:
 					&ci->i_layout, vino,
 					offset, &len, 0, num_ops,
 					CEPH_OSD_OP_WRITE,
-					CEPH_OSD_FLAG_WRITE |
-					CEPH_OSD_FLAG_ONDISK,
+					CEPH_OSD_FLAG_WRITE,
 					snapc, truncate_seq,
 					truncate_size, false);
 		if (IS_ERR(req)) {
@@ -1028,8 +1028,7 @@ new_request:
 						min(num_ops,
 						    CEPH_OSD_SLAB_OPS),
 						CEPH_OSD_OP_WRITE,
-						CEPH_OSD_FLAG_WRITE |
-						CEPH_OSD_FLAG_ONDISK,
+						CEPH_OSD_FLAG_WRITE,
 						snapc, truncate_seq,
 						truncate_size, true);
 			BUG_ON(IS_ERR(req));
@@ -1194,7 +1193,7 @@ 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) {
+	if (READ_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
 		dout(" page %p forced umount\n", page);
 		unlock_page(page);
 		return -EIO;
@@ -1681,8 +1680,7 @@ int ceph_uninline_data(struct file *filp, struct page *locked_page)
 
 	req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
 				    ceph_vino(inode), 0, &len, 0, 1,
-				    CEPH_OSD_OP_CREATE,
-				    CEPH_OSD_FLAG_ONDISK | CEPH_OSD_FLAG_WRITE,
+				    CEPH_OSD_OP_CREATE, CEPH_OSD_FLAG_WRITE,
 				    NULL, 0, 0, false);
 	if (IS_ERR(req)) {
 		err = PTR_ERR(req);
@@ -1699,8 +1697,7 @@ int ceph_uninline_data(struct file *filp, struct page *locked_page)
 
 	req = ceph_osdc_new_request(&fsc->client->osdc, &ci->i_layout,
 				    ceph_vino(inode), 0, &len, 1, 3,
-				    CEPH_OSD_OP_WRITE,
-				    CEPH_OSD_FLAG_ONDISK | CEPH_OSD_FLAG_WRITE,
+				    CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_WRITE,
 				    NULL, ci->i_truncate_seq,
 				    ci->i_truncate_size, false);
 	if (IS_ERR(req)) {
@@ -1873,7 +1870,7 @@ static int __ceph_pool_perm_get(struct ceph_inode_info *ci,
 		goto out_unlock;
 	}
 
-	wr_req->r_flags = CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ACK;
+	wr_req->r_flags = CEPH_OSD_FLAG_WRITE;
 	osd_req_op_init(wr_req, 0, CEPH_OSD_OP_CREATE, CEPH_OSD_OP_FLAG_EXCL);
 	ceph_oloc_copy(&wr_req->r_base_oloc, &rd_req->r_base_oloc);
 	ceph_oid_copy(&wr_req->r_base_oid, &rd_req->r_base_oid);
diff --git a/fs/ceph/cache.c b/fs/ceph/cache.c
index 5bc5d37b1217..4e7421caf380 100644
--- a/fs/ceph/cache.c
+++ b/fs/ceph/cache.c
@@ -234,7 +234,7 @@ void ceph_fscache_file_set_cookie(struct inode *inode, struct file *filp)
 		fscache_enable_cookie(ci->fscache, ceph_fscache_can_enable,
 				inode);
 		if (fscache_cookie_enabled(ci->fscache)) {
-			dout("fscache_file_set_cookie %p %p enabing cache\n",
+			dout("fscache_file_set_cookie %p %p enabling cache\n",
 			     inode, filp);
 		}
 	}
diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index 94fd76d04683..cd966f276a8d 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -867,7 +867,7 @@ int __ceph_caps_file_wanted(struct ceph_inode_info *ci)
 /*
  * Return caps we have registered with the MDS(s) as 'wanted'.
  */
-int __ceph_caps_mds_wanted(struct ceph_inode_info *ci)
+int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check)
 {
 	struct ceph_cap *cap;
 	struct rb_node *p;
@@ -875,7 +875,7 @@ int __ceph_caps_mds_wanted(struct ceph_inode_info *ci)
 
 	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))
+		if (check && !__cap_is_valid(cap))
 			continue;
 		if (cap == ci->i_auth_cap)
 			mds_wanted |= cap->mds_wanted;
@@ -1184,6 +1184,13 @@ static int __send_cap(struct ceph_mds_client *mdsc, struct ceph_cap *cap,
 		delayed = 1;
 	}
 	ci->i_ceph_flags &= ~(CEPH_I_NODELAY | CEPH_I_FLUSH);
+	if (want & ~cap->mds_wanted) {
+		/* user space may open/close single file frequently.
+		 * This avoids droping mds_wanted immediately after
+		 * requesting new mds_wanted.
+		 */
+		__cap_set_timeouts(mdsc, ci);
+	}
 
 	cap->issued &= retain;  /* drop bits we don't want */
 	if (cap->implemented & ~cap->issued) {
@@ -2084,8 +2091,6 @@ int ceph_fsync(struct file *file, loff_t start, loff_t end, int datasync)
 
 	dout("fsync %p%s\n", inode, datasync ? " datasync" : "");
 
-	ceph_sync_write_wait(inode);
-
 	ret = filemap_write_and_wait_range(inode->i_mapping, start, end);
 	if (ret < 0)
 		goto out;
@@ -2477,23 +2482,22 @@ again:
 
 		if (ci->i_ceph_flags & CEPH_I_CAP_DROPPED) {
 			int mds_wanted;
-			if (ACCESS_ONCE(mdsc->fsc->mount_state) ==
+			if (READ_ONCE(mdsc->fsc->mount_state) ==
 			    CEPH_MOUNT_SHUTDOWN) {
 				dout("get_cap_refs %p forced umount\n", inode);
 				*err = -EIO;
 				ret = 1;
 				goto out_unlock;
 			}
-			mds_wanted = __ceph_caps_mds_wanted(ci);
-			if ((mds_wanted & need) != need) {
+			mds_wanted = __ceph_caps_mds_wanted(ci, false);
+			if (need & ~(mds_wanted & need)) {
 				dout("get_cap_refs %p caps were dropped"
 				     " (session killed?)\n", inode);
 				*err = -ESTALE;
 				ret = 1;
 				goto out_unlock;
 			}
-			if ((mds_wanted & file_wanted) ==
-			    (file_wanted & (CEPH_CAP_FILE_RD|CEPH_CAP_FILE_WR)))
+			if (!(file_wanted & ~mds_wanted))
 				ci->i_ceph_flags &= ~CEPH_I_CAP_DROPPED;
 		}
 
@@ -3404,6 +3408,7 @@ retry:
 			tcap->implemented |= issued;
 			if (cap == ci->i_auth_cap)
 				ci->i_auth_cap = tcap;
+
 			if (!list_empty(&ci->i_cap_flush_list) &&
 			    ci->i_auth_cap == tcap) {
 				spin_lock(&mdsc->cap_dirty_lock);
@@ -3417,9 +3422,18 @@ retry:
 	} else if (tsession) {
 		/* add placeholder for the export tagert */
 		int flag = (cap == ci->i_auth_cap) ? CEPH_CAP_FLAG_AUTH : 0;
+		tcap = new_cap;
 		ceph_add_cap(inode, tsession, t_cap_id, -1, issued, 0,
 			     t_seq - 1, t_mseq, (u64)-1, flag, &new_cap);
 
+		if (!list_empty(&ci->i_cap_flush_list) &&
+		    ci->i_auth_cap == tcap) {
+			spin_lock(&mdsc->cap_dirty_lock);
+			list_move_tail(&ci->i_flushing_item,
+				       &tcap->session->s_cap_flushing);
+			spin_unlock(&mdsc->cap_dirty_lock);
+		}
+
 		__ceph_remove_cap(cap, false);
 		goto out_unlock;
 	}
@@ -3924,9 +3938,10 @@ int ceph_encode_inode_release(void **p, struct inode *inode,
 }
 
 int ceph_encode_dentry_release(void **p, struct dentry *dentry,
+			       struct inode *dir,
 			       int mds, int drop, int unless)
 {
-	struct inode *dir = d_inode(dentry->d_parent);
+	struct dentry *parent = NULL;
 	struct ceph_mds_request_release *rel = *p;
 	struct ceph_dentry_info *di = ceph_dentry(dentry);
 	int force = 0;
@@ -3941,9 +3956,14 @@ int ceph_encode_dentry_release(void **p, struct dentry *dentry,
 	spin_lock(&dentry->d_lock);
 	if (di->lease_session && di->lease_session->s_mds == mds)
 		force = 1;
+	if (!dir) {
+		parent = dget(dentry->d_parent);
+		dir = d_inode(parent);
+	}
 	spin_unlock(&dentry->d_lock);
 
 	ret = ceph_encode_inode_release(p, dir, mds, drop, unless, force);
+	dput(parent);
 
 	spin_lock(&dentry->d_lock);
 	if (ret && di->lease_session && di->lease_session->s_mds == mds) {
diff --git a/fs/ceph/debugfs.c b/fs/ceph/debugfs.c
index 39ff678e567f..f2ae393e2c31 100644
--- a/fs/ceph/debugfs.c
+++ b/fs/ceph/debugfs.c
@@ -70,7 +70,7 @@ static int mdsc_show(struct seq_file *s, void *p)
 
 		seq_printf(s, "%s", ceph_mds_op_name(req->r_op));
 
-		if (req->r_got_unsafe)
+		if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
 			seq_puts(s, "\t(unsafe)");
 		else
 			seq_puts(s, "\t");
diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c
index 8ab1fdf0bd49..3e9ad501addf 100644
--- a/fs/ceph/dir.c
+++ b/fs/ceph/dir.c
@@ -371,7 +371,7 @@ more:
 		/* hints to request -> mds selection code */
 		req->r_direct_mode = USE_AUTH_MDS;
 		req->r_direct_hash = ceph_frag_value(frag);
-		req->r_direct_is_hash = true;
+		__set_bit(CEPH_MDS_R_DIRECT_IS_HASH, &req->r_req_flags);
 		if (fi->last_name) {
 			req->r_path2 = kstrdup(fi->last_name, GFP_KERNEL);
 			if (!req->r_path2) {
@@ -417,7 +417,7 @@ more:
 		fi->frag = frag;
 		fi->last_readdir = req;
 
-		if (req->r_did_prepopulate) {
+		if (test_bit(CEPH_MDS_R_DID_PREPOPULATE, &req->r_req_flags)) {
 			fi->readdir_cache_idx = req->r_readdir_cache_idx;
 			if (fi->readdir_cache_idx < 0) {
 				/* preclude from marking dir ordered */
@@ -752,7 +752,8 @@ static struct dentry *ceph_lookup(struct inode *dir, struct dentry *dentry,
 		mask |= CEPH_CAP_XATTR_SHARED;
 	req->r_args.getattr.mask = cpu_to_le32(mask);
 
-	req->r_locked_dir = dir;
+	req->r_parent = dir;
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 	err = ceph_mdsc_do_request(mdsc, NULL, req);
 	err = ceph_handle_snapdir(req, dentry, err);
 	dentry = ceph_finish_lookup(req, dentry, err);
@@ -813,7 +814,8 @@ static int ceph_mknod(struct inode *dir, struct dentry *dentry,
 	}
 	req->r_dentry = dget(dentry);
 	req->r_num_caps = 2;
-	req->r_locked_dir = dir;
+	req->r_parent = dir;
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 	req->r_args.mknod.mode = cpu_to_le32(mode);
 	req->r_args.mknod.rdev = cpu_to_le32(rdev);
 	req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
@@ -864,7 +866,8 @@ static int ceph_symlink(struct inode *dir, struct dentry *dentry,
 		ceph_mdsc_put_request(req);
 		goto out;
 	}
-	req->r_locked_dir = dir;
+	req->r_parent = dir;
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 	req->r_dentry = dget(dentry);
 	req->r_num_caps = 2;
 	req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
@@ -913,7 +916,8 @@ static int ceph_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
 
 	req->r_dentry = dget(dentry);
 	req->r_num_caps = 2;
-	req->r_locked_dir = dir;
+	req->r_parent = dir;
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 	req->r_args.mkdir.mode = cpu_to_le32(mode);
 	req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
 	req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
@@ -957,7 +961,8 @@ static int ceph_link(struct dentry *old_dentry, struct inode *dir,
 	req->r_dentry = dget(dentry);
 	req->r_num_caps = 2;
 	req->r_old_dentry = dget(old_dentry);
-	req->r_locked_dir = dir;
+	req->r_parent = dir;
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 	req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
 	req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
 	/* release LINK_SHARED on source inode (mds will lock it) */
@@ -1023,7 +1028,8 @@ static int ceph_unlink(struct inode *dir, struct dentry *dentry)
 	}
 	req->r_dentry = dget(dentry);
 	req->r_num_caps = 2;
-	req->r_locked_dir = dir;
+	req->r_parent = dir;
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 	req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
 	req->r_dentry_unless = CEPH_CAP_FILE_EXCL;
 	req->r_inode_drop = drop_caps_for_unlink(inode);
@@ -1066,7 +1072,8 @@ static int ceph_rename(struct inode *old_dir, struct dentry *old_dentry,
 	req->r_num_caps = 2;
 	req->r_old_dentry = dget(old_dentry);
 	req->r_old_dentry_dir = old_dir;
-	req->r_locked_dir = new_dir;
+	req->r_parent = new_dir;
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 	req->r_old_dentry_drop = CEPH_CAP_FILE_SHARED;
 	req->r_old_dentry_unless = CEPH_CAP_FILE_EXCL;
 	req->r_dentry_drop = CEPH_CAP_FILE_SHARED;
@@ -1194,7 +1201,7 @@ static int ceph_d_revalidate(struct dentry *dentry, unsigned int flags)
 	struct inode *dir;
 
 	if (flags & LOOKUP_RCU) {
-		parent = ACCESS_ONCE(dentry->d_parent);
+		parent = READ_ONCE(dentry->d_parent);
 		dir = d_inode_rcu(parent);
 		if (!dir)
 			return -ECHILD;
@@ -1237,11 +1244,12 @@ static int ceph_d_revalidate(struct dentry *dentry, unsigned int flags)
 			return -ECHILD;
 
 		op = ceph_snap(dir) == CEPH_SNAPDIR ?
-			CEPH_MDS_OP_LOOKUPSNAP : CEPH_MDS_OP_GETATTR;
+			CEPH_MDS_OP_LOOKUPSNAP : CEPH_MDS_OP_LOOKUP;
 		req = ceph_mdsc_create_request(mdsc, op, USE_ANY_MDS);
 		if (!IS_ERR(req)) {
 			req->r_dentry = dget(dentry);
-			req->r_num_caps = op == CEPH_MDS_OP_GETATTR ? 1 : 2;
+			req->r_num_caps = 2;
+			req->r_parent = dir;
 
 			mask = CEPH_STAT_CAP_INODE | CEPH_CAP_AUTH_SHARED;
 			if (ceph_security_xattr_wanted(dir))
diff --git a/fs/ceph/export.c b/fs/ceph/export.c
index 180bbef760f2..e8f11fa565c5 100644
--- a/fs/ceph/export.c
+++ b/fs/ceph/export.c
@@ -207,7 +207,8 @@ static int ceph_get_name(struct dentry *parent, char *name,
 	req->r_inode = d_inode(child);
 	ihold(d_inode(child));
 	req->r_ino2 = ceph_vino(d_inode(parent));
-	req->r_locked_dir = d_inode(parent);
+	req->r_parent = d_inode(parent);
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 	req->r_num_caps = 2;
 	err = ceph_mdsc_do_request(mdsc, NULL, req);
 
diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index 045d30d26624..26cc95421cca 100644
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -283,7 +283,7 @@ int ceph_open(struct inode *inode, struct file *file)
 	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);
+		int mds_wanted = __ceph_caps_mds_wanted(ci, true);
 		int issued = __ceph_caps_issued(ci, NULL);
 
 		dout("open %p fmode %d want %s issued %s using existing\n",
@@ -379,7 +379,8 @@ int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
                mask |= CEPH_CAP_XATTR_SHARED;
        req->r_args.open.mask = cpu_to_le32(mask);
 
-	req->r_locked_dir = dir;           /* caller holds dir->i_mutex */
+	req->r_parent = dir;
+	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);
 	err = ceph_mdsc_do_request(mdsc,
 				   (flags & (O_CREAT|O_TRUNC)) ? dir : NULL,
 				   req);
@@ -758,9 +759,7 @@ static void ceph_aio_retry_work(struct work_struct *work)
 		goto out;
 	}
 
-	req->r_flags =	CEPH_OSD_FLAG_ORDERSNAP |
-			CEPH_OSD_FLAG_ONDISK |
-			CEPH_OSD_FLAG_WRITE;
+	req->r_flags = CEPH_OSD_FLAG_ORDERSNAP | CEPH_OSD_FLAG_WRITE;
 	ceph_oloc_copy(&req->r_base_oloc, &orig_req->r_base_oloc);
 	ceph_oid_copy(&req->r_base_oid, &orig_req->r_base_oid);
 
@@ -794,89 +793,6 @@ out:
 	kfree(aio_work);
 }
 
-/*
- * Write commit request unsafe callback, called to tell us when a
- * request is unsafe (that is, in flight--has been handed to the
- * messenger to send to its target osd).  It is called again when
- * we've received a response message indicating the request is
- * "safe" (its CEPH_OSD_FLAG_ONDISK flag is set), or when a request
- * is completed early (and unsuccessfully) due to a timeout or
- * interrupt.
- *
- * This is used if we requested both an ACK and ONDISK commit reply
- * from the OSD.
- */
-static void ceph_sync_write_unsafe(struct ceph_osd_request *req, bool unsafe)
-{
-	struct ceph_inode_info *ci = ceph_inode(req->r_inode);
-
-	dout("%s %p tid %llu %ssafe\n", __func__, req, req->r_tid,
-		unsafe ? "un" : "");
-	if (unsafe) {
-		ceph_get_cap_refs(ci, CEPH_CAP_FILE_WR);
-		spin_lock(&ci->i_unsafe_lock);
-		list_add_tail(&req->r_unsafe_item,
-			      &ci->i_unsafe_writes);
-		spin_unlock(&ci->i_unsafe_lock);
-
-		complete_all(&req->r_completion);
-	} else {
-		spin_lock(&ci->i_unsafe_lock);
-		list_del_init(&req->r_unsafe_item);
-		spin_unlock(&ci->i_unsafe_lock);
-		ceph_put_cap_refs(ci, CEPH_CAP_FILE_WR);
-	}
-}
-
-/*
- * Wait on any unsafe replies for the given inode.  First wait on the
- * newest request, and make that the upper bound.  Then, if there are
- * more requests, keep waiting on the oldest as long as it is still older
- * than the original request.
- */
-void ceph_sync_write_wait(struct inode *inode)
-{
-	struct ceph_inode_info *ci = ceph_inode(inode);
-	struct list_head *head = &ci->i_unsafe_writes;
-	struct ceph_osd_request *req;
-	u64 last_tid;
-
-	if (!S_ISREG(inode->i_mode))
-		return;
-
-	spin_lock(&ci->i_unsafe_lock);
-	if (list_empty(head))
-		goto out;
-
-	/* set upper bound as _last_ entry in chain */
-
-	req = list_last_entry(head, struct ceph_osd_request,
-			      r_unsafe_item);
-	last_tid = req->r_tid;
-
-	do {
-		ceph_osdc_get_request(req);
-		spin_unlock(&ci->i_unsafe_lock);
-
-		dout("sync_write_wait on tid %llu (until %llu)\n",
-		     req->r_tid, last_tid);
-		wait_for_completion(&req->r_done_completion);
-		ceph_osdc_put_request(req);
-
-		spin_lock(&ci->i_unsafe_lock);
-		/*
-		 * from here on look at first entry in chain, since we
-		 * only want to wait for anything older than last_tid
-		 */
-		if (list_empty(head))
-			break;
-		req = list_first_entry(head, struct ceph_osd_request,
-				       r_unsafe_item);
-	} while (req->r_tid < last_tid);
-out:
-	spin_unlock(&ci->i_unsafe_lock);
-}
-
 static ssize_t
 ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter,
 		       struct ceph_snap_context *snapc,
@@ -915,9 +831,7 @@ ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter,
 		if (ret2 < 0)
 			dout("invalidate_inode_pages2_range returned %d\n", ret2);
 
-		flags = CEPH_OSD_FLAG_ORDERSNAP |
-			CEPH_OSD_FLAG_ONDISK |
-			CEPH_OSD_FLAG_WRITE;
+		flags = CEPH_OSD_FLAG_ORDERSNAP | CEPH_OSD_FLAG_WRITE;
 	} else {
 		flags = CEPH_OSD_FLAG_READ;
 	}
@@ -1116,10 +1030,7 @@ ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos,
 	if (ret < 0)
 		dout("invalidate_inode_pages2_range returned %d\n", ret);
 
-	flags = CEPH_OSD_FLAG_ORDERSNAP |
-		CEPH_OSD_FLAG_ONDISK |
-		CEPH_OSD_FLAG_WRITE |
-		CEPH_OSD_FLAG_ACK;
+	flags = CEPH_OSD_FLAG_ORDERSNAP | CEPH_OSD_FLAG_WRITE;
 
 	while ((len = iov_iter_count(from)) > 0) {
 		size_t left;
@@ -1165,8 +1076,6 @@ ceph_sync_write(struct kiocb *iocb, struct iov_iter *from, loff_t pos,
 			goto out;
 		}
 
-		/* get a second commit callback */
-		req->r_unsafe_callback = ceph_sync_write_unsafe;
 		req->r_inode = inode;
 
 		osd_req_op_extent_osd_data_pages(req, 0, pages, len, 0,
@@ -1616,8 +1525,7 @@ static int ceph_zero_partial_object(struct inode *inode,
 					ceph_vino(inode),
 					offset, length,
 					0, 1, op,
-					CEPH_OSD_FLAG_WRITE |
-					CEPH_OSD_FLAG_ONDISK,
+					CEPH_OSD_FLAG_WRITE,
 					NULL, 0, 0, false);
 	if (IS_ERR(req)) {
 		ret = PTR_ERR(req);
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index 5e659d054b40..fd8f771f99b7 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -499,7 +499,6 @@ struct inode *ceph_alloc_inode(struct super_block *sb)
 	ci->i_rdcache_gen = 0;
 	ci->i_rdcache_revoking = 0;
 
-	INIT_LIST_HEAD(&ci->i_unsafe_writes);
 	INIT_LIST_HEAD(&ci->i_unsafe_dirops);
 	INIT_LIST_HEAD(&ci->i_unsafe_iops);
 	spin_lock_init(&ci->i_unsafe_lock);
@@ -583,14 +582,6 @@ int ceph_drop_inode(struct inode *inode)
 	return 1;
 }
 
-void ceph_evict_inode(struct inode *inode)
-{
-	/* wait unsafe sync writes */
-	ceph_sync_write_wait(inode);
-	truncate_inode_pages_final(&inode->i_data);
-	clear_inode(inode);
-}
-
 static inline blkcnt_t calc_inode_blocks(u64 size)
 {
 	return (size + (1<<9) - 1) >> 9;
@@ -1016,7 +1007,9 @@ out:
 static void update_dentry_lease(struct dentry *dentry,
 				struct ceph_mds_reply_lease *lease,
 				struct ceph_mds_session *session,
-				unsigned long from_time)
+				unsigned long from_time,
+				struct ceph_vino *tgt_vino,
+				struct ceph_vino *dir_vino)
 {
 	struct ceph_dentry_info *di = ceph_dentry(dentry);
 	long unsigned duration = le32_to_cpu(lease->duration_ms);
@@ -1024,13 +1017,27 @@ static void update_dentry_lease(struct dentry *dentry,
 	long unsigned half_ttl = from_time + (duration * HZ / 2) / 1000;
 	struct inode *dir;
 
+	/*
+	 * Make sure dentry's inode matches tgt_vino. NULL tgt_vino means that
+	 * we expect a negative dentry.
+	 */
+	if (!tgt_vino && d_really_is_positive(dentry))
+		return;
+
+	if (tgt_vino && (d_really_is_negative(dentry) ||
+			!ceph_ino_compare(d_inode(dentry), tgt_vino)))
+		return;
+
 	spin_lock(&dentry->d_lock);
 	dout("update_dentry_lease %p duration %lu ms ttl %lu\n",
 	     dentry, duration, ttl);
 
-	/* make lease_rdcache_gen match directory */
 	dir = d_inode(dentry->d_parent);
 
+	/* make sure parent matches dir_vino */
+	if (!ceph_ino_compare(dir, dir_vino))
+		goto out_unlock;
+
 	/* only track leases on regular dentries */
 	if (ceph_snap(dir) != CEPH_NOSNAP)
 		goto out_unlock;
@@ -1108,61 +1115,27 @@ out:
  *
  * Called with snap_rwsem (read).
  */
-int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req,
-		    struct ceph_mds_session *session)
+int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req)
 {
+	struct ceph_mds_session *session = req->r_session;
 	struct ceph_mds_reply_info_parsed *rinfo = &req->r_reply_info;
 	struct inode *in = NULL;
-	struct ceph_vino vino;
+	struct ceph_vino tvino, dvino;
 	struct ceph_fs_client *fsc = ceph_sb_to_client(sb);
 	int err = 0;
 
 	dout("fill_trace %p is_dentry %d is_target %d\n", req,
 	     rinfo->head->is_dentry, rinfo->head->is_target);
 
-#if 0
-	/*
-	 * Debugging hook:
-	 *
-	 * If we resend completed ops to a recovering mds, we get no
-	 * trace.  Since that is very rare, pretend this is the case
-	 * to ensure the 'no trace' handlers in the callers behave.
-	 *
-	 * Fill in inodes unconditionally to avoid breaking cap
-	 * invariants.
-	 */
-	if (rinfo->head->op & CEPH_MDS_OP_WRITE) {
-		pr_info("fill_trace faking empty trace on %lld %s\n",
-			req->r_tid, ceph_mds_op_name(rinfo->head->op));
-		if (rinfo->head->is_dentry) {
-			rinfo->head->is_dentry = 0;
-			err = fill_inode(req->r_locked_dir,
-					 &rinfo->diri, rinfo->dirfrag,
-					 session, req->r_request_started, -1);
-		}
-		if (rinfo->head->is_target) {
-			rinfo->head->is_target = 0;
-			ininfo = rinfo->targeti.in;
-			vino.ino = le64_to_cpu(ininfo->ino);
-			vino.snap = le64_to_cpu(ininfo->snapid);
-			in = ceph_get_inode(sb, vino);
-			err = fill_inode(in, &rinfo->targeti, NULL,
-					 session, req->r_request_started,
-					 req->r_fmode);
-			iput(in);
-		}
-	}
-#endif
-
 	if (!rinfo->head->is_target && !rinfo->head->is_dentry) {
 		dout("fill_trace reply is empty!\n");
-		if (rinfo->head->result == 0 && req->r_locked_dir)
+		if (rinfo->head->result == 0 && req->r_parent)
 			ceph_invalidate_dir_request(req);
 		return 0;
 	}
 
 	if (rinfo->head->is_dentry) {
-		struct inode *dir = req->r_locked_dir;
+		struct inode *dir = req->r_parent;
 
 		if (dir) {
 			err = fill_inode(dir, NULL,
@@ -1188,8 +1161,8 @@ int ceph_fill_trace(struct super_block *sb, struct ceph_mds_request *req,
 			dname.name = rinfo->dname;
 			dname.len = rinfo->dname_len;
 			dname.hash = full_name_hash(parent, dname.name, dname.len);
-			vino.ino = le64_to_cpu(rinfo->targeti.in->ino);
-			vino.snap = le64_to_cpu(rinfo->targeti.in->snapid);
+			tvino.ino = le64_to_cpu(rinfo->targeti.in->ino);
+			tvino.snap = le64_to_cpu(rinfo->targeti.in->snapid);
 retry_lookup:
 			dn = d_lookup(parent, &dname);
 			dout("d_lookup on parent=%p name=%.*s got %p\n",
@@ -1206,8 +1179,8 @@ retry_lookup:
 				}
 				err = 0;
 			} else if (d_really_is_positive(dn) &&
-				   (ceph_ino(d_inode(dn)) != vino.ino ||
-				    ceph_snap(d_inode(dn)) != vino.snap)) {
+				   (ceph_ino(d_inode(dn)) != tvino.ino ||
+				    ceph_snap(d_inode(dn)) != tvino.snap)) {
 				dout(" dn %p points to wrong inode %p\n",
 				     dn, d_inode(dn));
 				d_delete(dn);
@@ -1221,10 +1194,10 @@ retry_lookup:
 	}
 
 	if (rinfo->head->is_target) {
-		vino.ino = le64_to_cpu(rinfo->targeti.in->ino);
-		vino.snap = le64_to_cpu(rinfo->targeti.in->snapid);
+		tvino.ino = le64_to_cpu(rinfo->targeti.in->ino);
+		tvino.snap = le64_to_cpu(rinfo->targeti.in->snapid);
 
-		in = ceph_get_inode(sb, vino);
+		in = ceph_get_inode(sb, tvino);
 		if (IS_ERR(in)) {
 			err = PTR_ERR(in);
 			goto done;
@@ -1233,8 +1206,8 @@ retry_lookup:
 
 		err = fill_inode(in, req->r_locked_page, &rinfo->targeti, NULL,
 				session, req->r_request_started,
-				(!req->r_aborted && rinfo->head->result == 0) ?
-				req->r_fmode : -1,
+				(!test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags) &&
+				rinfo->head->result == 0) ?  req->r_fmode : -1,
 				&req->r_caps_reservation);
 		if (err < 0) {
 			pr_err("fill_inode badness %p %llx.%llx\n",
@@ -1247,8 +1220,9 @@ retry_lookup:
 	 * ignore null lease/binding on snapdir ENOENT, or else we
 	 * will have trouble splicing in the virtual snapdir later
 	 */
-	if (rinfo->head->is_dentry && !req->r_aborted &&
-	    req->r_locked_dir &&
+	if (rinfo->head->is_dentry &&
+            !test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags) &&
+	    test_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags) &&
 	    (rinfo->head->is_target || strncmp(req->r_dentry->d_name.name,
 					       fsc->mount_options->snapdir_name,
 					       req->r_dentry->d_name.len))) {
@@ -1257,17 +1231,19 @@ retry_lookup:
 		 * mknod symlink mkdir  : null -> new inode
 		 * unlink               : linked -> null
 		 */
-		struct inode *dir = req->r_locked_dir;
+		struct inode *dir = req->r_parent;
 		struct dentry *dn = req->r_dentry;
 		bool have_dir_cap, have_lease;
 
 		BUG_ON(!dn);
 		BUG_ON(!dir);
 		BUG_ON(d_inode(dn->d_parent) != dir);
-		BUG_ON(ceph_ino(dir) !=
-		       le64_to_cpu(rinfo->diri.in->ino));
-		BUG_ON(ceph_snap(dir) !=
-		       le64_to_cpu(rinfo->diri.in->snapid));
+
+		dvino.ino = le64_to_cpu(rinfo->diri.in->ino);
+		dvino.snap = le64_to_cpu(rinfo->diri.in->snapid);
+
+		BUG_ON(ceph_ino(dir) != dvino.ino);
+		BUG_ON(ceph_snap(dir) != dvino.snap);
 
 		/* do we have a lease on the whole dir? */
 		have_dir_cap =
@@ -1319,12 +1295,13 @@ retry_lookup:
 				ceph_dir_clear_ordered(dir);
 				dout("d_delete %p\n", dn);
 				d_delete(dn);
-			} else {
-				if (have_lease && d_unhashed(dn))
+			} else if (have_lease) {
+				if (d_unhashed(dn))
 					d_add(dn, NULL);
 				update_dentry_lease(dn, rinfo->dlease,
 						    session,
-						    req->r_request_started);
+						    req->r_request_started,
+						    NULL, &dvino);
 			}
 			goto done;
 		}
@@ -1347,15 +1324,19 @@ retry_lookup:
 			have_lease = false;
 		}
 
-		if (have_lease)
+		if (have_lease) {
+			tvino.ino = le64_to_cpu(rinfo->targeti.in->ino);
+			tvino.snap = le64_to_cpu(rinfo->targeti.in->snapid);
 			update_dentry_lease(dn, rinfo->dlease, session,
-					    req->r_request_started);
+					    req->r_request_started,
+					    &tvino, &dvino);
+		}
 		dout(" final dn %p\n", dn);
-	} else if (!req->r_aborted &&
-		   (req->r_op == CEPH_MDS_OP_LOOKUPSNAP ||
-		    req->r_op == CEPH_MDS_OP_MKSNAP)) {
+	} else if ((req->r_op == CEPH_MDS_OP_LOOKUPSNAP ||
+		    req->r_op == CEPH_MDS_OP_MKSNAP) &&
+		   !test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) {
 		struct dentry *dn = req->r_dentry;
-		struct inode *dir = req->r_locked_dir;
+		struct inode *dir = req->r_parent;
 
 		/* fill out a snapdir LOOKUPSNAP dentry */
 		BUG_ON(!dn);
@@ -1370,6 +1351,26 @@ retry_lookup:
 			goto done;
 		}
 		req->r_dentry = dn;  /* may have spliced */
+	} else if (rinfo->head->is_dentry) {
+		struct ceph_vino *ptvino = NULL;
+
+		if ((le32_to_cpu(rinfo->diri.in->cap.caps) & CEPH_CAP_FILE_SHARED) ||
+		    le32_to_cpu(rinfo->dlease->duration_ms)) {
+			dvino.ino = le64_to_cpu(rinfo->diri.in->ino);
+			dvino.snap = le64_to_cpu(rinfo->diri.in->snapid);
+
+			if (rinfo->head->is_target) {
+				tvino.ino = le64_to_cpu(rinfo->targeti.in->ino);
+				tvino.snap = le64_to_cpu(rinfo->targeti.in->snapid);
+				ptvino = &tvino;
+			}
+
+			update_dentry_lease(req->r_dentry, rinfo->dlease,
+				session, req->r_request_started, ptvino,
+				&dvino);
+		} else {
+			dout("%s: no dentry lease or dir cap\n", __func__);
+		}
 	}
 done:
 	dout("fill_trace done err=%d\n", err);
@@ -1478,7 +1479,7 @@ int ceph_readdir_prepopulate(struct ceph_mds_request *req,
 	u32 fpos_offset;
 	struct ceph_readdir_cache_control cache_ctl = {};
 
-	if (req->r_aborted)
+	if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags))
 		return readdir_prepopulate_inodes_only(req, session);
 
 	if (rinfo->hash_order && req->r_path2) {
@@ -1523,14 +1524,14 @@ int ceph_readdir_prepopulate(struct ceph_mds_request *req,
 	/* FIXME: release caps/leases if error occurs */
 	for (i = 0; i < rinfo->dir_nr; i++) {
 		struct ceph_mds_reply_dir_entry *rde = rinfo->dir_entries + i;
-		struct ceph_vino vino;
+		struct ceph_vino tvino, dvino;
 
 		dname.name = rde->name;
 		dname.len = rde->name_len;
 		dname.hash = full_name_hash(parent, dname.name, dname.len);
 
-		vino.ino = le64_to_cpu(rde->inode.in->ino);
-		vino.snap = le64_to_cpu(rde->inode.in->snapid);
+		tvino.ino = le64_to_cpu(rde->inode.in->ino);
+		tvino.snap = le64_to_cpu(rde->inode.in->snapid);
 
 		if (rinfo->hash_order) {
 			u32 hash = ceph_str_hash(ci->i_dir_layout.dl_dir_hash,
@@ -1559,8 +1560,8 @@ retry_lookup:
 				goto out;
 			}
 		} else if (d_really_is_positive(dn) &&
-			   (ceph_ino(d_inode(dn)) != vino.ino ||
-			    ceph_snap(d_inode(dn)) != vino.snap)) {
+			   (ceph_ino(d_inode(dn)) != tvino.ino ||
+			    ceph_snap(d_inode(dn)) != tvino.snap)) {
 			dout(" dn %p points to wrong inode %p\n",
 			     dn, d_inode(dn));
 			d_delete(dn);
@@ -1572,7 +1573,7 @@ retry_lookup:
 		if (d_really_is_positive(dn)) {
 			in = d_inode(dn);
 		} else {
-			in = ceph_get_inode(parent->d_sb, vino);
+			in = ceph_get_inode(parent->d_sb, tvino);
 			if (IS_ERR(in)) {
 				dout("new_inode badness\n");
 				d_drop(dn);
@@ -1617,8 +1618,9 @@ retry_lookup:
 
 		ceph_dentry(dn)->offset = rde->offset;
 
+		dvino = ceph_vino(d_inode(parent));
 		update_dentry_lease(dn, rde->lease, req->r_session,
-				    req->r_request_started);
+				    req->r_request_started, &tvino, &dvino);
 
 		if (err == 0 && skipped == 0 && cache_ctl.index >= 0) {
 			ret = fill_readdir_cache(d_inode(parent), dn,
@@ -1632,7 +1634,7 @@ next_item:
 	}
 out:
 	if (err == 0 && skipped == 0) {
-		req->r_did_prepopulate = true;
+		set_bit(CEPH_MDS_R_DID_PREPOPULATE, &req->r_req_flags);
 		req->r_readdir_cache_idx = cache_ctl.index;
 	}
 	ceph_readdir_cache_release(&cache_ctl);
@@ -1720,7 +1722,7 @@ static void ceph_invalidate_work(struct work_struct *work)
 
 	mutex_lock(&ci->i_truncate_mutex);
 
-	if (ACCESS_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
+	if (READ_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
 		pr_warn_ratelimited("invalidate_pages %p %lld forced umount\n",
 				    inode, ceph_ino(inode));
 		mapping_set_error(inode->i_mapping, -EIO);
diff --git a/fs/ceph/ioctl.c b/fs/ceph/ioctl.c
index 7d752d53353a..4c9c72f26eb9 100644
--- a/fs/ceph/ioctl.c
+++ b/fs/ceph/ioctl.c
@@ -25,7 +25,7 @@ static long ceph_ioctl_get_layout(struct file *file, void __user *arg)
 		l.stripe_count = ci->i_layout.stripe_count;
 		l.object_size = ci->i_layout.object_size;
 		l.data_pool = ci->i_layout.pool_id;
-		l.preferred_osd = (s32)-1;
+		l.preferred_osd = -1;
 		if (copy_to_user(arg, &l, sizeof(l)))
 			return -EFAULT;
 	}
@@ -97,7 +97,7 @@ static long ceph_ioctl_set_layout(struct file *file, void __user *arg)
 		nl.data_pool = ci->i_layout.pool_id;
 
 	/* this is obsolete, and always -1 */
-	nl.preferred_osd = le64_to_cpu(-1);
+	nl.preferred_osd = -1;
 
 	err = __validate_layout(mdsc, &nl);
 	if (err)
diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index c9d2e553a6c4..c681762d76e6 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -547,8 +547,8 @@ void ceph_mdsc_release_request(struct kref *kref)
 		ceph_put_cap_refs(ceph_inode(req->r_inode), CEPH_CAP_PIN);
 		iput(req->r_inode);
 	}
-	if (req->r_locked_dir)
-		ceph_put_cap_refs(ceph_inode(req->r_locked_dir), CEPH_CAP_PIN);
+	if (req->r_parent)
+		ceph_put_cap_refs(ceph_inode(req->r_parent), CEPH_CAP_PIN);
 	iput(req->r_target_inode);
 	if (req->r_dentry)
 		dput(req->r_dentry);
@@ -628,6 +628,9 @@ static void __unregister_request(struct ceph_mds_client *mdsc,
 {
 	dout("__unregister_request %p tid %lld\n", req, req->r_tid);
 
+	/* Never leave an unregistered request on an unsafe list! */
+	list_del_init(&req->r_unsafe_item);
+
 	if (req->r_tid == mdsc->oldest_tid) {
 		struct rb_node *p = rb_next(&req->r_node);
 		mdsc->oldest_tid = 0;
@@ -644,13 +647,15 @@ static void __unregister_request(struct ceph_mds_client *mdsc,
 
 	erase_request(&mdsc->request_tree, req);
 
-	if (req->r_unsafe_dir && req->r_got_unsafe) {
+	if (req->r_unsafe_dir  &&
+	    test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
 		struct ceph_inode_info *ci = ceph_inode(req->r_unsafe_dir);
 		spin_lock(&ci->i_unsafe_lock);
 		list_del_init(&req->r_unsafe_dir_item);
 		spin_unlock(&ci->i_unsafe_lock);
 	}
-	if (req->r_target_inode && req->r_got_unsafe) {
+	if (req->r_target_inode &&
+	    test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
 		struct ceph_inode_info *ci = ceph_inode(req->r_target_inode);
 		spin_lock(&ci->i_unsafe_lock);
 		list_del_init(&req->r_unsafe_target_item);
@@ -668,6 +673,28 @@ static void __unregister_request(struct ceph_mds_client *mdsc,
 }
 
 /*
+ * Walk back up the dentry tree until we hit a dentry representing a
+ * non-snapshot inode. We do this using the rcu_read_lock (which must be held
+ * when calling this) to ensure that the objects won't disappear while we're
+ * working with them. Once we hit a candidate dentry, we attempt to take a
+ * reference to it, and return that as the result.
+ */
+static struct inode *get_nonsnap_parent(struct dentry *dentry)
+{
+	struct inode *inode = NULL;
+
+	while (dentry && !IS_ROOT(dentry)) {
+		inode = d_inode_rcu(dentry);
+		if (!inode || ceph_snap(inode) == CEPH_NOSNAP)
+			break;
+		dentry = dentry->d_parent;
+	}
+	if (inode)
+		inode = igrab(inode);
+	return inode;
+}
+
+/*
  * Choose mds to send request to next.  If there is a hint set in the
  * request (e.g., due to a prior forward hint from the mds), use that.
  * Otherwise, consult frag tree and/or caps to identify the
@@ -675,19 +702,6 @@ static void __unregister_request(struct ceph_mds_client *mdsc,
  *
  * Called under mdsc->mutex.
  */
-static struct dentry *get_nonsnap_parent(struct dentry *dentry)
-{
-	/*
-	 * we don't need to worry about protecting the d_parent access
-	 * here because we never renaming inside the snapped namespace
-	 * except to resplice to another snapdir, and either the old or new
-	 * result is a valid result.
-	 */
-	while (!IS_ROOT(dentry) && ceph_snap(d_inode(dentry)) != CEPH_NOSNAP)
-		dentry = dentry->d_parent;
-	return dentry;
-}
-
 static int __choose_mds(struct ceph_mds_client *mdsc,
 			struct ceph_mds_request *req)
 {
@@ -697,7 +711,7 @@ static int __choose_mds(struct ceph_mds_client *mdsc,
 	int mode = req->r_direct_mode;
 	int mds = -1;
 	u32 hash = req->r_direct_hash;
-	bool is_hash = req->r_direct_is_hash;
+	bool is_hash = test_bit(CEPH_MDS_R_DIRECT_IS_HASH, &req->r_req_flags);
 
 	/*
 	 * is there a specific mds we should try?  ignore hint if we have
@@ -717,30 +731,39 @@ static int __choose_mds(struct ceph_mds_client *mdsc,
 	inode = NULL;
 	if (req->r_inode) {
 		inode = req->r_inode;
+		ihold(inode);
 	} else if (req->r_dentry) {
 		/* ignore race with rename; old or new d_parent is okay */
-		struct dentry *parent = req->r_dentry->d_parent;
-		struct inode *dir = d_inode(parent);
+		struct dentry *parent;
+		struct inode *dir;
+
+		rcu_read_lock();
+		parent = req->r_dentry->d_parent;
+		dir = req->r_parent ? : d_inode_rcu(parent);
 
-		if (dir->i_sb != mdsc->fsc->sb) {
-			/* not this fs! */
+		if (!dir || dir->i_sb != mdsc->fsc->sb) {
+			/*  not this fs or parent went negative */
 			inode = d_inode(req->r_dentry);
+			if (inode)
+				ihold(inode);
 		} else if (ceph_snap(dir) != CEPH_NOSNAP) {
 			/* direct snapped/virtual snapdir requests
 			 * based on parent dir inode */
-			struct dentry *dn = get_nonsnap_parent(parent);
-			inode = d_inode(dn);
+			inode = get_nonsnap_parent(parent);
 			dout("__choose_mds using nonsnap parent %p\n", inode);
 		} else {
 			/* dentry target */
 			inode = d_inode(req->r_dentry);
 			if (!inode || mode == USE_AUTH_MDS) {
 				/* dir + name */
-				inode = dir;
+				inode = igrab(dir);
 				hash = ceph_dentry_hash(dir, req->r_dentry);
 				is_hash = true;
+			} else {
+				ihold(inode);
 			}
 		}
+		rcu_read_unlock();
 	}
 
 	dout("__choose_mds %p is_hash=%d (%d) mode %d\n", inode, (int)is_hash,
@@ -769,7 +792,7 @@ static int __choose_mds(struct ceph_mds_client *mdsc,
 				     (int)r, frag.ndist);
 				if (ceph_mdsmap_get_state(mdsc->mdsmap, mds) >=
 				    CEPH_MDS_STATE_ACTIVE)
-					return mds;
+					goto out;
 			}
 
 			/* since this file/dir wasn't known to be
@@ -784,7 +807,7 @@ static int __choose_mds(struct ceph_mds_client *mdsc,
 				     inode, ceph_vinop(inode), frag.frag, mds);
 				if (ceph_mdsmap_get_state(mdsc->mdsmap, mds) >=
 				    CEPH_MDS_STATE_ACTIVE)
-					return mds;
+					goto out;
 			}
 		}
 	}
@@ -797,6 +820,7 @@ static int __choose_mds(struct ceph_mds_client *mdsc,
 		cap = rb_entry(rb_first(&ci->i_caps), struct ceph_cap, ci_node);
 	if (!cap) {
 		spin_unlock(&ci->i_ceph_lock);
+		iput(inode);
 		goto random;
 	}
 	mds = cap->session->s_mds;
@@ -804,6 +828,8 @@ static int __choose_mds(struct ceph_mds_client *mdsc,
 	     inode, ceph_vinop(inode), mds,
 	     cap == ci->i_auth_cap ? "auth " : "", cap);
 	spin_unlock(&ci->i_ceph_lock);
+out:
+	iput(inode);
 	return mds;
 
 random:
@@ -1036,7 +1062,6 @@ static void cleanup_session_requests(struct ceph_mds_client *mdsc,
 	while (!list_empty(&session->s_unsafe)) {
 		req = list_first_entry(&session->s_unsafe,
 				       struct ceph_mds_request, r_unsafe_item);
-		list_del_init(&req->r_unsafe_item);
 		pr_warn_ratelimited(" dropping unsafe request %llu\n",
 				    req->r_tid);
 		__unregister_request(mdsc, req);
@@ -1146,7 +1171,7 @@ static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,
 		ci->i_ceph_flags |= CEPH_I_CAP_DROPPED;
 
 		if (ci->i_wrbuffer_ref > 0 &&
-		    ACCESS_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
+		    READ_ONCE(fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
 			invalidate = true;
 
 		while (!list_empty(&ci->i_cap_flush_list)) {
@@ -1775,18 +1800,23 @@ retry:
 	return path;
 }
 
-static int build_dentry_path(struct dentry *dentry,
+static int build_dentry_path(struct dentry *dentry, struct inode *dir,
 			     const char **ppath, int *ppathlen, u64 *pino,
 			     int *pfreepath)
 {
 	char *path;
 
-	if (ceph_snap(d_inode(dentry->d_parent)) == CEPH_NOSNAP) {
-		*pino = ceph_ino(d_inode(dentry->d_parent));
+	rcu_read_lock();
+	if (!dir)
+		dir = d_inode_rcu(dentry->d_parent);
+	if (dir && ceph_snap(dir) == CEPH_NOSNAP) {
+		*pino = ceph_ino(dir);
+		rcu_read_unlock();
 		*ppath = dentry->d_name.name;
 		*ppathlen = dentry->d_name.len;
 		return 0;
 	}
+	rcu_read_unlock();
 	path = ceph_mdsc_build_path(dentry, ppathlen, pino, 1);
 	if (IS_ERR(path))
 		return PTR_ERR(path);
@@ -1822,8 +1852,8 @@ static int build_inode_path(struct inode *inode,
  * an explicit ino+path.
  */
 static int set_request_path_attr(struct inode *rinode, struct dentry *rdentry,
-				  const char *rpath, u64 rino,
-				  const char **ppath, int *pathlen,
+				  struct inode *rdiri, const char *rpath,
+				  u64 rino, const char **ppath, int *pathlen,
 				  u64 *ino, int *freepath)
 {
 	int r = 0;
@@ -1833,7 +1863,8 @@ static int set_request_path_attr(struct inode *rinode, struct dentry *rdentry,
 		dout(" inode %p %llx.%llx\n", rinode, ceph_ino(rinode),
 		     ceph_snap(rinode));
 	} else if (rdentry) {
-		r = build_dentry_path(rdentry, ppath, pathlen, ino, freepath);
+		r = build_dentry_path(rdentry, rdiri, ppath, pathlen, ino,
+					freepath);
 		dout(" dentry %p %llx/%.*s\n", rdentry, *ino, *pathlen,
 		     *ppath);
 	} else if (rpath || rino) {
@@ -1866,7 +1897,7 @@ static struct ceph_msg *create_request_message(struct ceph_mds_client *mdsc,
 	int ret;
 
 	ret = set_request_path_attr(req->r_inode, req->r_dentry,
-			      req->r_path1, req->r_ino1.ino,
+			      req->r_parent, req->r_path1, req->r_ino1.ino,
 			      &path1, &pathlen1, &ino1, &freepath1);
 	if (ret < 0) {
 		msg = ERR_PTR(ret);
@@ -1874,6 +1905,7 @@ static struct ceph_msg *create_request_message(struct ceph_mds_client *mdsc,
 	}
 
 	ret = set_request_path_attr(NULL, req->r_old_dentry,
+			      req->r_old_dentry_dir,
 			      req->r_path2, req->r_ino2.ino,
 			      &path2, &pathlen2, &ino2, &freepath2);
 	if (ret < 0) {
@@ -1927,10 +1959,13 @@ static struct ceph_msg *create_request_message(struct ceph_mds_client *mdsc,
 		      mds, req->r_inode_drop, req->r_inode_unless, 0);
 	if (req->r_dentry_drop)
 		releases += ceph_encode_dentry_release(&p, req->r_dentry,
-		       mds, req->r_dentry_drop, req->r_dentry_unless);
+				req->r_parent, mds, req->r_dentry_drop,
+				req->r_dentry_unless);
 	if (req->r_old_dentry_drop)
 		releases += ceph_encode_dentry_release(&p, req->r_old_dentry,
-		       mds, req->r_old_dentry_drop, req->r_old_dentry_unless);
+				req->r_old_dentry_dir, mds,
+				req->r_old_dentry_drop,
+				req->r_old_dentry_unless);
 	if (req->r_old_inode_drop)
 		releases += ceph_encode_inode_release(&p,
 		      d_inode(req->r_old_dentry),
@@ -2012,7 +2047,7 @@ static int __prepare_send_request(struct ceph_mds_client *mdsc,
 	dout("prepare_send_request %p tid %lld %s (attempt %d)\n", req,
 	     req->r_tid, ceph_mds_op_name(req->r_op), req->r_attempts);
 
-	if (req->r_got_unsafe) {
+	if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
 		void *p;
 		/*
 		 * Replay.  Do not regenerate message (and rebuild
@@ -2061,16 +2096,16 @@ static int __prepare_send_request(struct ceph_mds_client *mdsc,
 
 	rhead = msg->front.iov_base;
 	rhead->oldest_client_tid = cpu_to_le64(__get_oldest_tid(mdsc));
-	if (req->r_got_unsafe)
+	if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
 		flags |= CEPH_MDS_FLAG_REPLAY;
-	if (req->r_locked_dir)
+	if (req->r_parent)
 		flags |= CEPH_MDS_FLAG_WANT_DENTRY;
 	rhead->flags = cpu_to_le32(flags);
 	rhead->num_fwd = req->r_num_fwd;
 	rhead->num_retry = req->r_attempts - 1;
 	rhead->ino = 0;
 
-	dout(" r_locked_dir = %p\n", req->r_locked_dir);
+	dout(" r_parent = %p\n", req->r_parent);
 	return 0;
 }
 
@@ -2084,8 +2119,8 @@ static int __do_request(struct ceph_mds_client *mdsc,
 	int mds = -1;
 	int err = 0;
 
-	if (req->r_err || req->r_got_result) {
-		if (req->r_aborted)
+	if (req->r_err || test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
+		if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags))
 			__unregister_request(mdsc, req);
 		goto out;
 	}
@@ -2096,12 +2131,12 @@ static int __do_request(struct ceph_mds_client *mdsc,
 		err = -EIO;
 		goto finish;
 	}
-	if (ACCESS_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
+	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN) {
 		dout("do_request forced umount\n");
 		err = -EIO;
 		goto finish;
 	}
-	if (ACCESS_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_MOUNTING) {
+	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_MOUNTING) {
 		if (mdsc->mdsmap_err) {
 			err = mdsc->mdsmap_err;
 			dout("do_request mdsmap err %d\n", err);
@@ -2215,7 +2250,7 @@ static void kick_requests(struct ceph_mds_client *mdsc, int mds)
 	while (p) {
 		req = rb_entry(p, struct ceph_mds_request, r_node);
 		p = rb_next(p);
-		if (req->r_got_unsafe)
+		if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
 			continue;
 		if (req->r_attempts > 0)
 			continue; /* only new requests */
@@ -2250,11 +2285,11 @@ int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
 
 	dout("do_request on %p\n", req);
 
-	/* take CAP_PIN refs for r_inode, r_locked_dir, r_old_dentry */
+	/* take CAP_PIN refs for r_inode, r_parent, r_old_dentry */
 	if (req->r_inode)
 		ceph_get_cap_refs(ceph_inode(req->r_inode), CEPH_CAP_PIN);
-	if (req->r_locked_dir)
-		ceph_get_cap_refs(ceph_inode(req->r_locked_dir), CEPH_CAP_PIN);
+	if (req->r_parent)
+		ceph_get_cap_refs(ceph_inode(req->r_parent), CEPH_CAP_PIN);
 	if (req->r_old_dentry_dir)
 		ceph_get_cap_refs(ceph_inode(req->r_old_dentry_dir),
 				  CEPH_CAP_PIN);
@@ -2289,7 +2324,7 @@ int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
 	mutex_lock(&mdsc->mutex);
 
 	/* only abort if we didn't race with a real reply */
-	if (req->r_got_result) {
+	if (test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags)) {
 		err = le32_to_cpu(req->r_reply_info.head->result);
 	} else if (err < 0) {
 		dout("aborted request %lld with %d\n", req->r_tid, err);
@@ -2301,10 +2336,10 @@ int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
 		 */
 		mutex_lock(&req->r_fill_mutex);
 		req->r_err = err;
-		req->r_aborted = true;
+		set_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags);
 		mutex_unlock(&req->r_fill_mutex);
 
-		if (req->r_locked_dir &&
+		if (req->r_parent &&
 		    (req->r_op & CEPH_MDS_OP_WRITE))
 			ceph_invalidate_dir_request(req);
 	} else {
@@ -2323,7 +2358,7 @@ out:
  */
 void ceph_invalidate_dir_request(struct ceph_mds_request *req)
 {
-	struct inode *inode = req->r_locked_dir;
+	struct inode *inode = req->r_parent;
 
 	dout("invalidate_dir_request %p (complete, lease(s))\n", inode);
 
@@ -2379,14 +2414,14 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
 	}
 
 	/* dup? */
-	if ((req->r_got_unsafe && !head->safe) ||
-	    (req->r_got_safe && head->safe)) {
+	if ((test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags) && !head->safe) ||
+	    (test_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags) && head->safe)) {
 		pr_warn("got a dup %s reply on %llu from mds%d\n",
 			   head->safe ? "safe" : "unsafe", tid, mds);
 		mutex_unlock(&mdsc->mutex);
 		goto out;
 	}
-	if (req->r_got_safe) {
+	if (test_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags)) {
 		pr_warn("got unsafe after safe on %llu from mds%d\n",
 			   tid, mds);
 		mutex_unlock(&mdsc->mutex);
@@ -2425,10 +2460,10 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
 
 
 	if (head->safe) {
-		req->r_got_safe = true;
+		set_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags);
 		__unregister_request(mdsc, req);
 
-		if (req->r_got_unsafe) {
+		if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
 			/*
 			 * We already handled the unsafe response, now do the
 			 * cleanup.  No need to examine the response; the MDS
@@ -2437,7 +2472,6 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
 			 * useful we could do with a revised return value.
 			 */
 			dout("got safe reply %llu, mds%d\n", tid, mds);
-			list_del_init(&req->r_unsafe_item);
 
 			/* last unsafe request during umount? */
 			if (mdsc->stopping && !__get_oldest_req(mdsc))
@@ -2446,7 +2480,7 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
 			goto out;
 		}
 	} else {
-		req->r_got_unsafe = true;
+		set_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags);
 		list_add_tail(&req->r_unsafe_item, &req->r_session->s_unsafe);
 		if (req->r_unsafe_dir) {
 			struct ceph_inode_info *ci =
@@ -2486,7 +2520,7 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
 	/* insert trace into our cache */
 	mutex_lock(&req->r_fill_mutex);
 	current->journal_info = req;
-	err = ceph_fill_trace(mdsc->fsc->sb, req, req->r_session);
+	err = ceph_fill_trace(mdsc->fsc->sb, req);
 	if (err == 0) {
 		if (result == 0 && (req->r_op == CEPH_MDS_OP_READDIR ||
 				    req->r_op == CEPH_MDS_OP_LSSNAP))
@@ -2500,7 +2534,8 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
 	if (realm)
 		ceph_put_snap_realm(mdsc, realm);
 
-	if (err == 0 && req->r_got_unsafe && req->r_target_inode) {
+	if (err == 0 && req->r_target_inode &&
+	    test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags)) {
 		struct ceph_inode_info *ci = ceph_inode(req->r_target_inode);
 		spin_lock(&ci->i_unsafe_lock);
 		list_add_tail(&req->r_unsafe_target_item, &ci->i_unsafe_iops);
@@ -2508,12 +2543,12 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
 	}
 out_err:
 	mutex_lock(&mdsc->mutex);
-	if (!req->r_aborted) {
+	if (!test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) {
 		if (err) {
 			req->r_err = err;
 		} else {
 			req->r_reply =  ceph_msg_get(msg);
-			req->r_got_result = true;
+			set_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags);
 		}
 	} else {
 		dout("reply arrived after request %lld was aborted\n", tid);
@@ -2557,7 +2592,7 @@ static void handle_forward(struct ceph_mds_client *mdsc,
 		goto out;  /* dup reply? */
 	}
 
-	if (req->r_aborted) {
+	if (test_bit(CEPH_MDS_R_ABORTED, &req->r_req_flags)) {
 		dout("forward tid %llu aborted, unregistering\n", tid);
 		__unregister_request(mdsc, req);
 	} else if (fwd_seq <= req->r_num_fwd) {
@@ -2567,7 +2602,7 @@ static void handle_forward(struct ceph_mds_client *mdsc,
 		/* resend. forward race not possible; mds would drop */
 		dout("forward tid %llu to mds%d (we resend)\n", tid, next_mds);
 		BUG_ON(req->r_err);
-		BUG_ON(req->r_got_result);
+		BUG_ON(test_bit(CEPH_MDS_R_GOT_RESULT, &req->r_req_flags));
 		req->r_attempts = 0;
 		req->r_num_fwd = fwd_seq;
 		req->r_resend_mds = next_mds;
@@ -2732,7 +2767,7 @@ static void replay_unsafe_requests(struct ceph_mds_client *mdsc,
 	while (p) {
 		req = rb_entry(p, struct ceph_mds_request, r_node);
 		p = rb_next(p);
-		if (req->r_got_unsafe)
+		if (test_bit(CEPH_MDS_R_GOT_UNSAFE, &req->r_req_flags))
 			continue;
 		if (req->r_attempts == 0)
 			continue; /* only old requests */
@@ -3556,7 +3591,7 @@ void ceph_mdsc_sync(struct ceph_mds_client *mdsc)
 {
 	u64 want_tid, want_flush;
 
-	if (ACCESS_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
+	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
 		return;
 
 	dout("sync\n");
@@ -3587,7 +3622,7 @@ void ceph_mdsc_sync(struct ceph_mds_client *mdsc)
  */
 static bool done_closing_sessions(struct ceph_mds_client *mdsc, int skipped)
 {
-	if (ACCESS_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
+	if (READ_ONCE(mdsc->fsc->mount_state) == CEPH_MOUNT_SHUTDOWN)
 		return true;
 	return atomic_read(&mdsc->num_sessions) <= skipped;
 }
diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h
index 3c6f77b7bb02..ac0475a2daa7 100644
--- a/fs/ceph/mds_client.h
+++ b/fs/ceph/mds_client.h
@@ -202,9 +202,18 @@ struct ceph_mds_request {
 	char *r_path1, *r_path2;
 	struct ceph_vino r_ino1, r_ino2;
 
-	struct inode *r_locked_dir; /* dir (if any) i_mutex locked by vfs */
+	struct inode *r_parent;		    /* parent dir inode */
 	struct inode *r_target_inode;       /* resulting inode */
 
+#define CEPH_MDS_R_DIRECT_IS_HASH	(1) /* r_direct_hash is valid */
+#define CEPH_MDS_R_ABORTED		(2) /* call was aborted */
+#define CEPH_MDS_R_GOT_UNSAFE		(3) /* got an unsafe reply */
+#define CEPH_MDS_R_GOT_SAFE		(4) /* got a safe reply */
+#define CEPH_MDS_R_GOT_RESULT		(5) /* got a result */
+#define CEPH_MDS_R_DID_PREPOPULATE	(6) /* prepopulated readdir */
+#define CEPH_MDS_R_PARENT_LOCKED	(7) /* is r_parent->i_rwsem wlocked? */
+	unsigned long	r_req_flags;
+
 	struct mutex r_fill_mutex;
 
 	union ceph_mds_request_args r_args;
@@ -216,7 +225,6 @@ struct ceph_mds_request {
 	/* for choosing which mds to send this request to */
 	int r_direct_mode;
 	u32 r_direct_hash;      /* choose dir frag based on this dentry hash */
-	bool r_direct_is_hash;  /* true if r_direct_hash is valid */
 
 	/* data payload is used for xattr ops */
 	struct ceph_pagelist *r_pagelist;
@@ -234,7 +242,6 @@ struct ceph_mds_request {
 	struct ceph_mds_reply_info_parsed r_reply_info;
 	struct page *r_locked_page;
 	int r_err;
-	bool r_aborted;
 
 	unsigned long r_timeout;  /* optional.  jiffies, 0 is "wait forever" */
 	unsigned long r_started;  /* start time to measure timeout against */
@@ -262,9 +269,7 @@ struct ceph_mds_request {
 	ceph_mds_request_callback_t r_callback;
 	ceph_mds_request_wait_callback_t r_wait_for_completion;
 	struct list_head  r_unsafe_item;  /* per-session unsafe list item */
-	bool		  r_got_unsafe, r_got_safe, r_got_result;
 
-	bool              r_did_prepopulate;
 	long long	  r_dir_release_cnt;
 	long long	  r_dir_ordered_cnt;
 	int		  r_readdir_cache_idx;
diff --git a/fs/ceph/super.c b/fs/ceph/super.c
index 6bd20d707bfd..0ec8d0114e57 100644
--- a/fs/ceph/super.c
+++ b/fs/ceph/super.c
@@ -757,7 +757,6 @@ static const struct super_operations ceph_super_ops = {
 	.destroy_inode	= ceph_destroy_inode,
 	.write_inode    = ceph_write_inode,
 	.drop_inode	= ceph_drop_inode,
-	.evict_inode	= ceph_evict_inode,
 	.sync_fs        = ceph_sync_fs,
 	.put_super	= ceph_put_super,
 	.show_options   = ceph_show_options,
@@ -952,6 +951,14 @@ static int ceph_register_bdi(struct super_block *sb,
 		fsc->backing_dev_info.ra_pages =
 			VM_MAX_READAHEAD * 1024 / PAGE_SIZE;
 
+	if (fsc->mount_options->rsize > fsc->mount_options->rasize &&
+	    fsc->mount_options->rsize >= PAGE_SIZE)
+		fsc->backing_dev_info.io_pages =
+			(fsc->mount_options->rsize + PAGE_SIZE - 1)
+			>> PAGE_SHIFT;
+	else if (fsc->mount_options->rsize == 0)
+		fsc->backing_dev_info.io_pages = ULONG_MAX;
+
 	err = bdi_register(&fsc->backing_dev_info, NULL, "ceph-%ld",
 			   atomic_long_inc_return(&bdi_seq));
 	if (!err)
diff --git a/fs/ceph/super.h b/fs/ceph/super.h
index 3373b61faefd..e9410bcf4113 100644
--- a/fs/ceph/super.h
+++ b/fs/ceph/super.h
@@ -45,8 +45,8 @@
 #define ceph_test_mount_opt(fsc, opt) \
 	(!!((fsc)->mount_options->flags & CEPH_MOUNT_OPT_##opt))
 
-#define CEPH_RSIZE_DEFAULT             0           /* max read size */
-#define CEPH_RASIZE_DEFAULT            (8192*1024) /* readahead */
+#define CEPH_RSIZE_DEFAULT              (64*1024*1024) /* max read size */
+#define CEPH_RASIZE_DEFAULT             (8192*1024)    /* max readahead */
 #define CEPH_MAX_READDIR_DEFAULT        1024
 #define CEPH_MAX_READDIR_BYTES_DEFAULT  (512*1024)
 #define CEPH_SNAPDIRNAME_DEFAULT        ".snap"
@@ -343,7 +343,6 @@ struct ceph_inode_info {
 	u32 i_rdcache_gen;      /* incremented each time we get FILE_CACHE. */
 	u32 i_rdcache_revoking; /* RDCACHE gen to async invalidate, if any */
 
-	struct list_head i_unsafe_writes; /* uncommitted sync writes */
 	struct list_head i_unsafe_dirops; /* uncommitted mds dir ops */
 	struct list_head i_unsafe_iops;   /* uncommitted mds inode ops */
 	spinlock_t i_unsafe_lock;
@@ -602,7 +601,7 @@ static inline int __ceph_caps_wanted(struct ceph_inode_info *ci)
 }
 
 /* what the mds thinks we want */
-extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci);
+extern int __ceph_caps_mds_wanted(struct ceph_inode_info *ci, bool check);
 
 extern void ceph_caps_init(struct ceph_mds_client *mdsc);
 extern void ceph_caps_finalize(struct ceph_mds_client *mdsc);
@@ -753,7 +752,6 @@ extern const struct inode_operations ceph_file_iops;
 extern struct inode *ceph_alloc_inode(struct super_block *sb);
 extern void ceph_destroy_inode(struct inode *inode);
 extern int ceph_drop_inode(struct inode *inode);
-extern void ceph_evict_inode(struct inode *inode);
 
 extern struct inode *ceph_get_inode(struct super_block *sb,
 				    struct ceph_vino vino);
@@ -764,8 +762,7 @@ extern void ceph_fill_file_time(struct inode *inode, int issued,
 				u64 time_warp_seq, struct timespec *ctime,
 				struct timespec *mtime, struct timespec *atime);
 extern int ceph_fill_trace(struct super_block *sb,
-			   struct ceph_mds_request *req,
-			   struct ceph_mds_session *session);
+			   struct ceph_mds_request *req);
 extern int ceph_readdir_prepopulate(struct ceph_mds_request *req,
 				    struct ceph_mds_session *session);
 
@@ -904,6 +901,7 @@ extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
 extern int ceph_encode_inode_release(void **p, struct inode *inode,
 				     int mds, int drop, int unless, int force);
 extern int ceph_encode_dentry_release(void **p, struct dentry *dn,
+				      struct inode *dir,
 				      int mds, int drop, int unless);
 
 extern int ceph_get_caps(struct ceph_inode_info *ci, int need, int want,
@@ -933,7 +931,7 @@ extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry,
 extern int ceph_release(struct inode *inode, struct file *filp);
 extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,
 				  char *data, size_t len);
-extern void ceph_sync_write_wait(struct inode *inode);
+
 /* dir.c */
 extern const struct file_operations ceph_dir_fops;
 extern const struct file_operations ceph_snapdir_fops;