Merge tag 'ceph-for-5.18-rc1' of https://github.com/ceph/ceph-client

Pull ceph updates from Ilya Dryomov:
 "The highlights are:

   - several changes to how snap context and snap realms are tracked
     (Xiubo Li). In particular, this should resolve a long-standing
     issue of high kworker CPU usage and various stalls caused by
     needless iteration over all inodes in the snap realm.

   - async create fixes to address hangs in some edge cases (Jeff
     Layton)

   - support for getvxattr MDS op for querying server-side xattrs, such
     as file/directory layouts and ephemeral pins (Milind Changire)

   - average latency is now maintained for all metrics (Venky Shankar)

   - some tweaks around handling inline data to make it fit better with
     netfs helper library (David Howells)

  Also a couple of memory leaks got plugged along with a few assorted
  fixups. Last but not least, Xiubo has stepped up to serve as a CephFS
  co-maintainer"

* tag 'ceph-for-5.18-rc1' of https://github.com/ceph/ceph-client: (27 commits)
  ceph: fix memory leak in ceph_readdir when note_last_dentry returns error
  ceph: uninitialized variable in debug output
  ceph: use tracked average r/w/m latencies to display metrics in debugfs
  ceph: include average/stdev r/w/m latency in mds metrics
  ceph: track average r/w/m latency
  ceph: use ktime_to_timespec64() rather than jiffies_to_timespec64()
  ceph: assign the ci only when the inode isn't NULL
  ceph: fix inode reference leakage in ceph_get_snapdir()
  ceph: misc fix for code style and logs
  ceph: allocate capsnap memory outside of ceph_queue_cap_snap()
  ceph: do not release the global snaprealm until unmounting
  ceph: remove incorrect and unused CEPH_INO_DOTDOT macro
  MAINTAINERS: add Xiubo Li as cephfs co-maintainer
  ceph: eliminate the recursion when rebuilding the snap context
  ceph: do not update snapshot context when there is no new snapshot
  ceph: zero the dir_entries memory when allocating it
  ceph: move to a dedicated slabcache for ceph_cap_snap
  ceph: add getvxattr op
  libceph: drop else branches in prepare_read_data{,_cont}
  ceph: fix comments mentioning i_mutex
  ...

MAINTAINERS

@@ -4456,6 +4456,7 @@ F:	drivers/power/supply/cw2015_battery.c
 CEPH COMMON CODE (LIBCEPH)
 M:	Ilya Dryomov <idryomov@gmail.com>
 M:	Jeff Layton <jlayton@kernel.org>
+M:	Xiubo Li <xiubli@redhat.com>
 L:	ceph-devel@vger.kernel.org
 S:	Supported
 W:	http://ceph.com/
@@ -4466,6 +4467,7 @@ F:	net/ceph/
 
 CEPH DISTRIBUTED FILE SYSTEM CLIENT (CEPH)
 M:	Jeff Layton <jlayton@kernel.org>
+M:	Xiubo Li <xiubli@redhat.com>
 M:	Ilya Dryomov <idryomov@gmail.com>
 L:	ceph-devel@vger.kernel.org
 S:	Supported

fs/ceph/caps.c

@@ -1915,6 +1915,13 @@ void ceph_check_caps(struct ceph_inode_info *ci, int flags,
 		ceph_get_mds_session(session);
 
 	spin_lock(&ci->i_ceph_lock);
+	if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE) {
+		/* Don't send messages until we get async create reply */
+		spin_unlock(&ci->i_ceph_lock);
+		ceph_put_mds_session(session);
+		return;
+	}
+
 	if (ci->i_ceph_flags & CEPH_I_FLUSH)
 		flags |= CHECK_CAPS_FLUSH;
 retry:
@@ -2409,6 +2416,9 @@ int ceph_write_inode(struct inode *inode, struct writeback_control *wbc)
 	dout("write_inode %p wait=%d\n", inode, wait);
 	ceph_fscache_unpin_writeback(inode, wbc);
 	if (wait) {
+		err = ceph_wait_on_async_create(inode);
+		if (err)
+			return err;
 		dirty = try_flush_caps(inode, &flush_tid);
 		if (dirty)
 			err = wait_event_interruptible(ci->i_cap_wq,
@@ -2439,6 +2449,10 @@ static void __kick_flushing_caps(struct ceph_mds_client *mdsc,
 	u64 first_tid = 0;
 	u64 last_snap_flush = 0;
 
+	/* Don't do anything until create reply comes in */
+	if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE)
+		return;
+
 	ci->i_ceph_flags &= ~CEPH_I_KICK_FLUSH;
 
 	list_for_each_entry_reverse(cf, &ci->i_cap_flush_list, i_list) {
@@ -4152,7 +4166,6 @@ void ceph_handle_caps(struct ceph_mds_session *session,
 
 	/* lookup ino */
 	inode = ceph_find_inode(mdsc->fsc->sb, vino);
-	ci = ceph_inode(inode);
 	dout(" op %s ino %llx.%llx inode %p\n", ceph_cap_op_name(op), vino.ino,
 	     vino.snap, inode);
 
@@ -4178,6 +4191,7 @@ void ceph_handle_caps(struct ceph_mds_session *session,
 		}
 		goto flush_cap_releases;
 	}
+	ci = ceph_inode(inode);
 
 	/* these will work even if we don't have a cap yet */
 	switch (op) {

fs/ceph/debugfs.c

@@ -175,7 +175,7 @@ static int metrics_latency_show(struct seq_file *s, void *p)
 	struct ceph_fs_client *fsc = s->private;
 	struct ceph_client_metric *cm = &fsc->mdsc->metric;
 	struct ceph_metric *m;
-	s64 total, sum, avg, min, max, sq;
+	s64 total, avg, min, max, sq;
 	int i;
 
 	seq_printf(s, "item          total       avg_lat(us)     min_lat(us)     max_lat(us)     stdev(us)\n");
@@ -185,8 +185,7 @@ static int metrics_latency_show(struct seq_file *s, void *p)
 		m = &cm->metric[i];
 		spin_lock(&m->lock);
 		total = m->total;
-		sum = m->latency_sum;
-		avg = total > 0 ? DIV64_U64_ROUND_CLOSEST(sum, total) : 0;
+		avg = m->latency_avg;
 		min = m->latency_min;
 		max = m->latency_max;
 		sq = m->latency_sq_sum;

fs/ceph/dir.c

@@ -145,7 +145,7 @@ __dcache_find_get_entry(struct dentry *parent, u64 idx,
 			return ERR_PTR(-EAGAIN);
 		}
 		/* reading/filling the cache are serialized by
-		   i_mutex, no need to use page lock */
+		   i_rwsem, no need to use page lock */
 		unlock_page(cache_ctl->page);
 		cache_ctl->dentries = kmap(cache_ctl->page);
 	}
@@ -155,7 +155,7 @@ __dcache_find_get_entry(struct dentry *parent, u64 idx,
 	rcu_read_lock();
 	spin_lock(&parent->d_lock);
 	/* check i_size again here, because empty directory can be
-	 * marked as complete while not holding the i_mutex. */
+	 * marked as complete while not holding the i_rwsem. */
 	if (ceph_dir_is_complete_ordered(dir) && ptr_pos < i_size_read(dir))
 		dentry = cache_ctl->dentries[cache_ctl->index];
 	else
@@ -478,8 +478,11 @@ static int ceph_readdir(struct file *file, struct dir_context *ctx)
 					2 : (fpos_off(rde->offset) + 1);
 			err = note_last_dentry(dfi, rde->name, rde->name_len,
 					       next_offset);
-			if (err)
+			if (err) {
+				ceph_mdsc_put_request(dfi->last_readdir);
+				dfi->last_readdir = NULL;
 				return err;
+			}
 		} else if (req->r_reply_info.dir_end) {
 			dfi->next_offset = 2;
 			/* keep last name */
@@ -520,6 +523,12 @@ static int ceph_readdir(struct file *file, struct dir_context *ctx)
 		if (!dir_emit(ctx, rde->name, rde->name_len,
 			      ceph_present_ino(inode->i_sb, le64_to_cpu(rde->inode.in->ino)),
 			      le32_to_cpu(rde->inode.in->mode) >> 12)) {
+			/*
+			 * NOTE: Here no need to put the 'dfi->last_readdir',
+			 * because when dir_emit stops us it's most likely
+			 * doesn't have enough memory, etc. So for next readdir
+			 * it will continue.
+			 */
 			dout("filldir stopping us...\n");
 			return 0;
 		}
@@ -671,7 +680,7 @@ struct dentry *ceph_handle_snapdir(struct ceph_mds_request *req,
 				   struct dentry *dentry)
 {
 	struct ceph_fs_client *fsc = ceph_sb_to_client(dentry->d_sb);
-	struct inode *parent = d_inode(dentry->d_parent); /* we hold i_mutex */
+	struct inode *parent = d_inode(dentry->d_parent); /* we hold i_rwsem */
 
 	/* .snap dir? */
 	if (ceph_snap(parent) == CEPH_NOSNAP &&

fs/ceph/file.c

@@ -207,6 +207,7 @@ static int ceph_init_file_info(struct inode *inode, struct file *file,
 	struct ceph_mount_options *opt =
 		ceph_inode_to_client(&ci->vfs_inode)->mount_options;
 	struct ceph_file_info *fi;
+	int ret;
 
 	dout("%s %p %p 0%o (%s)\n", __func__, inode, file,
 			inode->i_mode, isdir ? "dir" : "regular");
@@ -240,7 +241,22 @@ static int ceph_init_file_info(struct inode *inode, struct file *file,
 	INIT_LIST_HEAD(&fi->rw_contexts);
 	fi->filp_gen = READ_ONCE(ceph_inode_to_client(inode)->filp_gen);
 
+	if ((file->f_mode & FMODE_WRITE) &&
+	    ci->i_inline_version != CEPH_INLINE_NONE) {
+		ret = ceph_uninline_data(file);
+		if (ret < 0)
+			goto error;
+	}
+
 	return 0;
+
+error:
+	ceph_fscache_unuse_cookie(inode, file->f_mode & FMODE_WRITE);
+	ceph_put_fmode(ci, fi->fmode, 1);
+	kmem_cache_free(ceph_file_cachep, fi);
+	/* wake up anyone waiting for caps on this inode */
+	wake_up_all(&ci->i_cap_wq);
+	return ret;
 }
 
 /*
@@ -516,52 +532,67 @@ static void restore_deleg_ino(struct inode *dir, u64 ino)
 	}
 }
 
+static void wake_async_create_waiters(struct inode *inode,
+				      struct ceph_mds_session *session)
+{
+	struct ceph_inode_info *ci = ceph_inode(inode);
+
+	spin_lock(&ci->i_ceph_lock);
+	if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE) {
+		ci->i_ceph_flags &= ~CEPH_I_ASYNC_CREATE;
+		wake_up_bit(&ci->i_ceph_flags, CEPH_ASYNC_CREATE_BIT);
+	}
+	ceph_kick_flushing_inode_caps(session, ci);
+	spin_unlock(&ci->i_ceph_lock);
+}
+
 static void ceph_async_create_cb(struct ceph_mds_client *mdsc,
                                  struct ceph_mds_request *req)
 {
+	struct dentry *dentry = req->r_dentry;
+	struct inode *dinode = d_inode(dentry);
+	struct inode *tinode = req->r_target_inode;
 	int result = req->r_err ? req->r_err :
 			le32_to_cpu(req->r_reply_info.head->result);
 
+	WARN_ON_ONCE(dinode && tinode && dinode != tinode);
+
+	/* MDS changed -- caller must resubmit */
 	if (result == -EJUKEBOX)
 		goto out;
 
 	mapping_set_error(req->r_parent->i_mapping, result);
 
 	if (result) {
-		struct dentry *dentry = req->r_dentry;
-		struct inode *inode = d_inode(dentry);
 		int pathlen = 0;
 		u64 base = 0;
 		char *path = ceph_mdsc_build_path(req->r_dentry, &pathlen,
 						  &base, 0);
 
+		pr_warn("ceph: async create failure path=(%llx)%s result=%d!\n",
+			base, IS_ERR(path) ? "<<bad>>" : path, result);
+		ceph_mdsc_free_path(path, pathlen);
+
 		ceph_dir_clear_complete(req->r_parent);
 		if (!d_unhashed(dentry))
 			d_drop(dentry);
 
-		ceph_inode_shutdown(inode);
-
-		pr_warn("ceph: async create failure path=(%llx)%s result=%d!\n",
-			base, IS_ERR(path) ? "<<bad>>" : path, result);
-		ceph_mdsc_free_path(path, pathlen);
+		if (dinode) {
+			mapping_set_error(dinode->i_mapping, result);
+			ceph_inode_shutdown(dinode);
+			wake_async_create_waiters(dinode, req->r_session);
+		}
 	}
 
-	if (req->r_target_inode) {
-		struct ceph_inode_info *ci = ceph_inode(req->r_target_inode);
-		u64 ino = ceph_vino(req->r_target_inode).ino;
+	if (tinode) {
+		u64 ino = ceph_vino(tinode).ino;
 
 		if (req->r_deleg_ino != ino)
 			pr_warn("%s: inode number mismatch! err=%d deleg_ino=0x%llx target=0x%llx\n",
 				__func__, req->r_err, req->r_deleg_ino, ino);
-		mapping_set_error(req->r_target_inode->i_mapping, result);
 
-		spin_lock(&ci->i_ceph_lock);
-		if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE) {
-			ci->i_ceph_flags &= ~CEPH_I_ASYNC_CREATE;
-			wake_up_bit(&ci->i_ceph_flags, CEPH_ASYNC_CREATE_BIT);
-		}
-		ceph_kick_flushing_inode_caps(req->r_session, ci);
-		spin_unlock(&ci->i_ceph_lock);
+		mapping_set_error(tinode->i_mapping, result);
+		wake_async_create_waiters(tinode, req->r_session);
 	} else if (!result) {
 		pr_warn("%s: no req->r_target_inode for 0x%llx\n", __func__,
 			req->r_deleg_ino);
@@ -1041,7 +1072,6 @@ static void ceph_aio_complete(struct inode *inode,
 		}
 
 		spin_lock(&ci->i_ceph_lock);
-		ci->i_inline_version = CEPH_INLINE_NONE;
 		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
 					       &aio_req->prealloc_cf);
 		spin_unlock(&ci->i_ceph_lock);
@@ -1778,12 +1808,6 @@ static ssize_t ceph_write_iter(struct kiocb *iocb, struct iov_iter *from)
 	if (err)
 		goto out;
 
-	if (ci->i_inline_version != CEPH_INLINE_NONE) {
-		err = ceph_uninline_data(file, NULL);
-		if (err < 0)
-			goto out;
-	}
-
 	dout("aio_write %p %llx.%llx %llu~%zd getting caps. i_size %llu\n",
 	     inode, ceph_vinop(inode), pos, count, i_size_read(inode));
 	if (!(fi->flags & CEPH_F_SYNC) && !direct_lock)
@@ -1855,7 +1879,6 @@ static ssize_t ceph_write_iter(struct kiocb *iocb, struct iov_iter *from)
 		int dirty;
 
 		spin_lock(&ci->i_ceph_lock);
-		ci->i_inline_version = CEPH_INLINE_NONE;
 		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
 					       &prealloc_cf);
 		spin_unlock(&ci->i_ceph_lock);
@@ -2109,12 +2132,6 @@ static long ceph_fallocate(struct file *file, int mode,
 		goto unlock;
 	}
 
-	if (ci->i_inline_version != CEPH_INLINE_NONE) {
-		ret = ceph_uninline_data(file, NULL);
-		if (ret < 0)
-			goto unlock;
-	}
-
 	size = i_size_read(inode);
 
 	/* Are we punching a hole beyond EOF? */
@@ -2139,7 +2156,6 @@ static long ceph_fallocate(struct file *file, int mode,
 
 	if (!ret) {
 		spin_lock(&ci->i_ceph_lock);
-		ci->i_inline_version = CEPH_INLINE_NONE;
 		dirty = __ceph_mark_dirty_caps(ci, CEPH_CAP_FILE_WR,
 					       &prealloc_cf);
 		spin_unlock(&ci->i_ceph_lock);
@@ -2532,7 +2548,6 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
 	}
 	/* Mark Fw dirty */
 	spin_lock(&dst_ci->i_ceph_lock);
-	dst_ci->i_inline_version = CEPH_INLINE_NONE;
 	dirty = __ceph_mark_dirty_caps(dst_ci, CEPH_CAP_FILE_WR, &prealloc_cf);
 	spin_unlock(&dst_ci->i_ceph_lock);
 	if (dirty)

fs/ceph/inode.c

@@ -87,13 +87,13 @@ struct inode *ceph_get_snapdir(struct inode *parent)
 	if (!S_ISDIR(parent->i_mode)) {
 		pr_warn_once("bad snapdir parent type (mode=0%o)\n",
 			     parent->i_mode);
-		return ERR_PTR(-ENOTDIR);
+		goto err;
 	}
 
 	if (!(inode->i_state & I_NEW) && !S_ISDIR(inode->i_mode)) {
 		pr_warn_once("bad snapdir inode type (mode=0%o)\n",
 			     inode->i_mode);
-		return ERR_PTR(-ENOTDIR);
+		goto err;
 	}
 
 	inode->i_mode = parent->i_mode;
@@ -113,6 +113,12 @@ struct inode *ceph_get_snapdir(struct inode *parent)
 	}
 
 	return inode;
+err:
+	if ((inode->i_state & I_NEW))
+		discard_new_inode(inode);
+	else
+		iput(inode);
+	return ERR_PTR(-ENOTDIR);
 }
 
 const struct inode_operations ceph_file_iops = {
@@ -1201,7 +1207,7 @@ static void update_dentry_lease_careful(struct dentry *dentry,
 
 /*
  * splice a dentry to an inode.
- * caller must hold directory i_mutex for this to be safe.
+ * caller must hold directory i_rwsem for this to be safe.
  */
 static int splice_dentry(struct dentry **pdn, struct inode *in)
 {
@@ -1598,7 +1604,7 @@ static int fill_readdir_cache(struct inode *dir, struct dentry *dn,
 			return idx == 0 ? -ENOMEM : 0;
 		}
 		/* reading/filling the cache are serialized by
-		 * i_mutex, no need to use page lock */
+		 * i_rwsem, no need to use page lock */
 		unlock_page(ctl->page);
 		ctl->dentries = kmap(ctl->page);
 		if (idx == 0)
@@ -2301,6 +2307,57 @@ int __ceph_do_getattr(struct inode *inode, struct page *locked_page,
 	return err;
 }
 
+int ceph_do_getvxattr(struct inode *inode, const char *name, void *value,
+		      size_t size)
+{
+	struct ceph_fs_client *fsc = ceph_sb_to_client(inode->i_sb);
+	struct ceph_mds_client *mdsc = fsc->mdsc;
+	struct ceph_mds_request *req;
+	int mode = USE_AUTH_MDS;
+	int err;
+	char *xattr_value;
+	size_t xattr_value_len;
+
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_GETVXATTR, mode);
+	if (IS_ERR(req)) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	req->r_path2 = kstrdup(name, GFP_NOFS);
+	if (!req->r_path2) {
+		err = -ENOMEM;
+		goto put;
+	}
+
+	ihold(inode);
+	req->r_inode = inode;
+	err = ceph_mdsc_do_request(mdsc, NULL, req);
+	if (err < 0)
+		goto put;
+
+	xattr_value = req->r_reply_info.xattr_info.xattr_value;
+	xattr_value_len = req->r_reply_info.xattr_info.xattr_value_len;
+
+	dout("do_getvxattr xattr_value_len:%zu, size:%zu\n", xattr_value_len, size);
+
+	err = (int)xattr_value_len;
+	if (size == 0)
+		goto put;
+
+	if (xattr_value_len > size) {
+		err = -ERANGE;
+		goto put;
+	}
+
+	memcpy(value, xattr_value, xattr_value_len);
+put:
+	ceph_mdsc_put_request(req);
+out:
+	dout("do_getvxattr result=%d\n", err);
+	return err;
+}
+
 
 /*
  * Check inode permissions.  We verify we have a valid value for

fs/ceph/locks.c

@@ -111,10 +111,10 @@ static int ceph_lock_message(u8 lock_type, u16 operation, struct inode *inode,
 	req->r_args.filelock_change.length = cpu_to_le64(length);
 	req->r_args.filelock_change.wait = wait;
 
-	if (wait)
-		req->r_wait_for_completion = ceph_lock_wait_for_completion;
-
-	err = ceph_mdsc_do_request(mdsc, inode, req);
+	err = ceph_mdsc_submit_request(mdsc, inode, req);
+	if (!err)
+		err = ceph_mdsc_wait_request(mdsc, req, wait ?
+					ceph_lock_wait_for_completion : NULL);
 	if (!err && operation == CEPH_MDS_OP_GETFILELOCK) {
 		fl->fl_pid = -le64_to_cpu(req->r_reply_info.filelock_reply->pid);
 		if (CEPH_LOCK_SHARED == req->r_reply_info.filelock_reply->type)

fs/ceph/mds_client.c

@@ -555,6 +555,28 @@ static int parse_reply_info_create(void **p, void *end,
 	return -EIO;
 }
 
+static int parse_reply_info_getvxattr(void **p, void *end,
+				      struct ceph_mds_reply_info_parsed *info,
+				      u64 features)
+{
+	u32 value_len;
+
+	ceph_decode_skip_8(p, end, bad); /* skip current version: 1 */
+	ceph_decode_skip_8(p, end, bad); /* skip first version: 1 */
+	ceph_decode_skip_32(p, end, bad); /* skip payload length */
+
+	ceph_decode_32_safe(p, end, value_len, bad);
+
+	if (value_len == end - *p) {
+	  info->xattr_info.xattr_value = *p;
+	  info->xattr_info.xattr_value_len = value_len;
+	  *p = end;
+	  return value_len;
+	}
+bad:
+	return -EIO;
+}
+
 /*
  * parse extra results
  */
@@ -570,6 +592,8 @@ static int parse_reply_info_extra(void **p, void *end,
 		return parse_reply_info_readdir(p, end, info, features);
 	else if (op == CEPH_MDS_OP_CREATE)
 		return parse_reply_info_create(p, end, info, features, s);
+	else if (op == CEPH_MDS_OP_GETVXATTR)
+		return parse_reply_info_getvxattr(p, end, info, features);
 	else
 		return -EIO;
 }
@@ -2178,7 +2202,8 @@ int ceph_alloc_readdir_reply_buffer(struct ceph_mds_request *req,
 	order = get_order(size * num_entries);
 	while (order >= 0) {
 		rinfo->dir_entries = (void*)__get_free_pages(GFP_KERNEL |
-							     __GFP_NOWARN,
+							     __GFP_NOWARN |
+							     __GFP_ZERO,
 							     order);
 		if (rinfo->dir_entries)
 			break;
@@ -2946,15 +2971,16 @@ int ceph_mdsc_submit_request(struct ceph_mds_client *mdsc, struct inode *dir,
 	return err;
 }
 
-static int ceph_mdsc_wait_request(struct ceph_mds_client *mdsc,
-				  struct ceph_mds_request *req)
+int ceph_mdsc_wait_request(struct ceph_mds_client *mdsc,
+			   struct ceph_mds_request *req,
+			   ceph_mds_request_wait_callback_t wait_func)
 {
 	int err;
 
 	/* wait */
 	dout("do_request waiting\n");
-	if (!req->r_timeout && req->r_wait_for_completion) {
-		err = req->r_wait_for_completion(mdsc, req);
+	if (wait_func) {
+		err = wait_func(mdsc, req);
 	} else {
 		long timeleft = wait_for_completion_killable_timeout(
 					&req->r_completion,
@@ -3011,7 +3037,7 @@ int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
 	/* issue */
 	err = ceph_mdsc_submit_request(mdsc, dir, req);
 	if (!err)
-		err = ceph_mdsc_wait_request(mdsc, req);
+		err = ceph_mdsc_wait_request(mdsc, req, NULL);
 	dout("do_request %p done, result %d\n", req, err);
 	return err;
 }
@@ -3097,35 +3123,6 @@ static void handle_reply(struct ceph_mds_session *session, struct ceph_msg *msg)
 
 	result = le32_to_cpu(head->result);
 
-	/*
-	 * Handle an ESTALE
-	 * if we're not talking to the authority, send to them
-	 * if the authority has changed while we weren't looking,
-	 * send to new authority
-	 * Otherwise we just have to return an ESTALE
-	 */
-	if (result == -ESTALE) {
-		dout("got ESTALE on request %llu\n", req->r_tid);
-		req->r_resend_mds = -1;
-		if (req->r_direct_mode != USE_AUTH_MDS) {
-			dout("not using auth, setting for that now\n");
-			req->r_direct_mode = USE_AUTH_MDS;
-			__do_request(mdsc, req);
-			mutex_unlock(&mdsc->mutex);
-			goto out;
-		} else  {
-			int mds = __choose_mds(mdsc, req, NULL);
-			if (mds >= 0 && mds != req->r_session->s_mds) {
-				dout("but auth changed, so resending\n");
-				__do_request(mdsc, req);
-				mutex_unlock(&mdsc->mutex);
-				goto out;
-			}
-		}
-		dout("have to return ESTALE on request %llu\n", req->r_tid);
-	}
-
-
 	if (head->safe) {
 		set_bit(CEPH_MDS_R_GOT_SAFE, &req->r_req_flags);
 		__unregister_request(mdsc, req);
@@ -4841,7 +4838,7 @@ void ceph_mdsc_close_sessions(struct ceph_mds_client *mdsc)
 	mutex_unlock(&mdsc->mutex);
 
 	ceph_cleanup_snapid_map(mdsc);
-	ceph_cleanup_empty_realms(mdsc);
+	ceph_cleanup_global_and_empty_realms(mdsc);
 
 	cancel_work_sync(&mdsc->cap_reclaim_work);
 	cancel_delayed_work_sync(&mdsc->delayed_work); /* cancel timer */

fs/ceph/mds_client.h

@@ -100,6 +100,11 @@ struct ceph_mds_reply_dir_entry {
 	loff_t			      offset;
 };
 
+struct ceph_mds_reply_xattr {
+	char *xattr_value;
+	size_t xattr_value_len;
+};
+
 /*
  * parsed info about an mds reply, including information about
  * either: 1) the target inode and/or its parent directory and dentry,
@@ -115,6 +120,7 @@ struct ceph_mds_reply_info_parsed {
 	char                          *dname;
 	u32                           dname_len;
 	struct ceph_mds_reply_lease   *dlease;
+	struct ceph_mds_reply_xattr   xattr_info;
 
 	/* extra */
 	union {
@@ -274,8 +280,8 @@ struct ceph_mds_request {
 
 	union ceph_mds_request_args r_args;
 	int r_fmode;        /* file mode, if expecting cap */
-	const struct cred *r_cred;
 	int r_request_release_offset;
+	const struct cred *r_cred;
 	struct timespec64 r_stamp;
 
 	/* for choosing which mds to send this request to */
@@ -296,12 +302,11 @@ struct ceph_mds_request {
 	struct ceph_msg  *r_reply;
 	struct ceph_mds_reply_info_parsed r_reply_info;
 	int r_err;
-
+	u32               r_readdir_offset;
 
 	struct page *r_locked_page;
 	int r_dir_caps;
 	int r_num_caps;
-	u32               r_readdir_offset;
 
 	unsigned long r_timeout;  /* optional.  jiffies, 0 is "wait forever" */
 	unsigned long r_started;  /* start time to measure timeout against */
@@ -329,7 +334,6 @@ struct ceph_mds_request {
 	struct completion r_completion;
 	struct completion r_safe_completion;
 	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 */
 
 	long long	  r_dir_release_cnt;
@@ -507,6 +511,9 @@ ceph_mdsc_create_request(struct ceph_mds_client *mdsc, int op, int mode);
 extern int ceph_mdsc_submit_request(struct ceph_mds_client *mdsc,
 				    struct inode *dir,
 				    struct ceph_mds_request *req);
+int ceph_mdsc_wait_request(struct ceph_mds_client *mdsc,
+			struct ceph_mds_request *req,
+			ceph_mds_request_wait_callback_t wait_func);
 extern int ceph_mdsc_do_request(struct ceph_mds_client *mdsc,
 				struct inode *dir,
 				struct ceph_mds_request *req);

fs/ceph/metric.c

@@ -8,6 +8,12 @@
 #include "metric.h"
 #include "mds_client.h"
 
+static void ktime_to_ceph_timespec(struct ceph_timespec *ts, ktime_t val)
+{
+	struct timespec64 t = ktime_to_timespec64(val);
+	ceph_encode_timespec64(ts, &t);
+}
+
 static bool ceph_mdsc_send_metrics(struct ceph_mds_client *mdsc,
 				   struct ceph_mds_session *s)
 {
@@ -26,7 +32,6 @@ static bool ceph_mdsc_send_metrics(struct ceph_mds_client *mdsc,
 	u64 nr_caps = atomic64_read(&m->total_caps);
 	u32 header_len = sizeof(struct ceph_metric_header);
 	struct ceph_msg *msg;
-	struct timespec64 ts;
 	s64 sum;
 	s32 items = 0;
 	s32 len;
@@ -59,37 +64,40 @@ static bool ceph_mdsc_send_metrics(struct ceph_mds_client *mdsc,
 	/* encode the read latency metric */
 	read = (struct ceph_metric_read_latency *)(cap + 1);
 	read->header.type = cpu_to_le32(CLIENT_METRIC_TYPE_READ_LATENCY);
-	read->header.ver = 1;
+	read->header.ver = 2;
 	read->header.compat = 1;
 	read->header.data_len = cpu_to_le32(sizeof(*read) - header_len);
 	sum = m->metric[METRIC_READ].latency_sum;
-	jiffies_to_timespec64(sum, &ts);
-	read->sec = cpu_to_le32(ts.tv_sec);
-	read->nsec = cpu_to_le32(ts.tv_nsec);
+	ktime_to_ceph_timespec(&read->lat, sum);
+	ktime_to_ceph_timespec(&read->avg, m->metric[METRIC_READ].latency_avg);
+	read->sq_sum = cpu_to_le64(m->metric[METRIC_READ].latency_sq_sum);
+	read->count = cpu_to_le64(m->metric[METRIC_READ].total);
 	items++;
 
 	/* encode the write latency metric */
 	write = (struct ceph_metric_write_latency *)(read + 1);
 	write->header.type = cpu_to_le32(CLIENT_METRIC_TYPE_WRITE_LATENCY);
-	write->header.ver = 1;
+	write->header.ver = 2;
 	write->header.compat = 1;
 	write->header.data_len = cpu_to_le32(sizeof(*write) - header_len);
 	sum = m->metric[METRIC_WRITE].latency_sum;
-	jiffies_to_timespec64(sum, &ts);
-	write->sec = cpu_to_le32(ts.tv_sec);
-	write->nsec = cpu_to_le32(ts.tv_nsec);
+	ktime_to_ceph_timespec(&write->lat, sum);
+	ktime_to_ceph_timespec(&write->avg, m->metric[METRIC_WRITE].latency_avg);
+	write->sq_sum = cpu_to_le64(m->metric[METRIC_WRITE].latency_sq_sum);
+	write->count = cpu_to_le64(m->metric[METRIC_WRITE].total);
 	items++;
 
 	/* encode the metadata latency metric */
 	meta = (struct ceph_metric_metadata_latency *)(write + 1);
 	meta->header.type = cpu_to_le32(CLIENT_METRIC_TYPE_METADATA_LATENCY);
-	meta->header.ver = 1;
+	meta->header.ver = 2;
 	meta->header.compat = 1;
 	meta->header.data_len = cpu_to_le32(sizeof(*meta) - header_len);
 	sum = m->metric[METRIC_METADATA].latency_sum;
-	jiffies_to_timespec64(sum, &ts);
-	meta->sec = cpu_to_le32(ts.tv_sec);
-	meta->nsec = cpu_to_le32(ts.tv_nsec);
+	ktime_to_ceph_timespec(&meta->lat, sum);
+	ktime_to_ceph_timespec(&meta->avg, m->metric[METRIC_METADATA].latency_avg);
+	meta->sq_sum = cpu_to_le64(m->metric[METRIC_METADATA].latency_sq_sum);
+	meta->count = cpu_to_le64(m->metric[METRIC_METADATA].total);
 	items++;
 
 	/* encode the dentry lease metric */
@@ -250,6 +258,7 @@ int ceph_metric_init(struct ceph_client_metric *m)
 		metric->size_max = 0;
 		metric->total = 0;
 		metric->latency_sum = 0;
+		metric->latency_avg = 0;
 		metric->latency_sq_sum = 0;
 		metric->latency_min = KTIME_MAX;
 		metric->latency_max = 0;
@@ -307,20 +316,19 @@ void ceph_metric_destroy(struct ceph_client_metric *m)
 		max = new;			\
 }
 
-static inline void __update_stdev(ktime_t total, ktime_t lsum,
+static inline void __update_mean_and_stdev(ktime_t total, ktime_t *lavg,
 					   ktime_t *sq_sump, ktime_t lat)
 {
-	ktime_t avg, sq;
-
-	if (unlikely(total == 1))
-		return;
+	ktime_t avg;
 
+	if (unlikely(total == 1)) {
+		*lavg = lat;
+	} else {
 		/* the sq is (lat - old_avg) * (lat - new_avg) */
-	avg = DIV64_U64_ROUND_CLOSEST((lsum - lat), (total - 1));
-	sq = lat - avg;
-	avg = DIV64_U64_ROUND_CLOSEST(lsum, total);
-	sq = sq * (lat - avg);
-	*sq_sump += sq;
+		avg = *lavg + div64_s64(lat - *lavg, total);
+		*sq_sump += (lat - *lavg)*(lat - avg);
+		*lavg = avg;
+	}
 }
 
 void ceph_update_metrics(struct ceph_metric *m,
@@ -339,6 +347,7 @@ void ceph_update_metrics(struct ceph_metric *m,
 	METRIC_UPDATE_MIN_MAX(m->size_min, m->size_max, size);
 	m->latency_sum += lat;
 	METRIC_UPDATE_MIN_MAX(m->latency_min, m->latency_max, lat);
-	__update_stdev(total, m->latency_sum, &m->latency_sq_sum, lat);
+	__update_mean_and_stdev(total, &m->latency_avg,	&m->latency_sq_sum,
+				lat);
 	spin_unlock(&m->lock);
 }

fs/ceph/metric.h

@@ -2,7 +2,7 @@
 #ifndef _FS_CEPH_MDS_METRIC_H
 #define _FS_CEPH_MDS_METRIC_H
 
-#include <linux/types.h>
+#include <linux/ceph/types.h>
 #include <linux/percpu_counter.h>
 #include <linux/ktime.h>
 
@@ -19,8 +19,14 @@ enum ceph_metric_type {
 	CLIENT_METRIC_TYPE_OPENED_INODES,
 	CLIENT_METRIC_TYPE_READ_IO_SIZES,
 	CLIENT_METRIC_TYPE_WRITE_IO_SIZES,
-
-	CLIENT_METRIC_TYPE_MAX = CLIENT_METRIC_TYPE_WRITE_IO_SIZES,
+	CLIENT_METRIC_TYPE_AVG_READ_LATENCY,
+	CLIENT_METRIC_TYPE_STDEV_READ_LATENCY,
+	CLIENT_METRIC_TYPE_AVG_WRITE_LATENCY,
+	CLIENT_METRIC_TYPE_STDEV_WRITE_LATENCY,
+	CLIENT_METRIC_TYPE_AVG_METADATA_LATENCY,
+	CLIENT_METRIC_TYPE_STDEV_METADATA_LATENCY,
+
+	CLIENT_METRIC_TYPE_MAX = CLIENT_METRIC_TYPE_STDEV_METADATA_LATENCY,
 };
 
 /*
@@ -38,6 +44,12 @@ enum ceph_metric_type {
 	CLIENT_METRIC_TYPE_OPENED_INODES,	   \
 	CLIENT_METRIC_TYPE_READ_IO_SIZES,	   \
 	CLIENT_METRIC_TYPE_WRITE_IO_SIZES,	   \
+	CLIENT_METRIC_TYPE_AVG_READ_LATENCY,	   \
+	CLIENT_METRIC_TYPE_STDEV_READ_LATENCY,	   \
+	CLIENT_METRIC_TYPE_AVG_WRITE_LATENCY,	   \
+	CLIENT_METRIC_TYPE_STDEV_WRITE_LATENCY,	   \
+	CLIENT_METRIC_TYPE_AVG_METADATA_LATENCY,   \
+	CLIENT_METRIC_TYPE_STDEV_METADATA_LATENCY, \
 						   \
 	CLIENT_METRIC_TYPE_MAX,			   \
 }
@@ -60,22 +72,28 @@ struct ceph_metric_cap {
 /* metric read latency header */
 struct ceph_metric_read_latency {
 	struct ceph_metric_header header;
-	__le32 sec;
-	__le32 nsec;
+	struct ceph_timespec lat;
+	struct ceph_timespec avg;
+	__le64 sq_sum;
+	__le64 count;
 } __packed;
 
 /* metric write latency header */
 struct ceph_metric_write_latency {
 	struct ceph_metric_header header;
-	__le32 sec;
-	__le32 nsec;
+	struct ceph_timespec lat;
+	struct ceph_timespec avg;
+	__le64 sq_sum;
+	__le64 count;
 } __packed;
 
 /* metric metadata latency header */
 struct ceph_metric_metadata_latency {
 	struct ceph_metric_header header;
-	__le32 sec;
-	__le32 nsec;
+	struct ceph_timespec lat;
+	struct ceph_timespec avg;
+	__le64 sq_sum;
+	__le64 count;
 } __packed;
 
 /* metric dentry lease header */
@@ -140,6 +158,7 @@ struct ceph_metric {
 	u64 size_min;
 	u64 size_max;
 	ktime_t latency_sum;
+	ktime_t latency_avg;
 	ktime_t latency_sq_sum;
 	ktime_t latency_min;
 	ktime_t latency_max;

fs/ceph/strings.c

@@ -60,6 +60,7 @@ const char *ceph_mds_op_name(int op)
 	case CEPH_MDS_OP_LOOKUPINO:  return "lookupino";
 	case CEPH_MDS_OP_LOOKUPNAME:  return "lookupname";
 	case CEPH_MDS_OP_GETATTR:  return "getattr";
+	case CEPH_MDS_OP_GETVXATTR:  return "getvxattr";
 	case CEPH_MDS_OP_SETXATTR: return "setxattr";
 	case CEPH_MDS_OP_SETATTR: return "setattr";
 	case CEPH_MDS_OP_RMXATTR: return "rmxattr";

fs/ceph/super.c

@@ -865,6 +865,7 @@ static void destroy_fs_client(struct ceph_fs_client *fsc)
  */
 struct kmem_cache *ceph_inode_cachep;
 struct kmem_cache *ceph_cap_cachep;
+struct kmem_cache *ceph_cap_snap_cachep;
 struct kmem_cache *ceph_cap_flush_cachep;
 struct kmem_cache *ceph_dentry_cachep;
 struct kmem_cache *ceph_file_cachep;
@@ -893,6 +894,9 @@ static int __init init_caches(void)
 	ceph_cap_cachep = KMEM_CACHE(ceph_cap, SLAB_MEM_SPREAD);
 	if (!ceph_cap_cachep)
 		goto bad_cap;
+	ceph_cap_snap_cachep = KMEM_CACHE(ceph_cap_snap, SLAB_MEM_SPREAD);
+	if (!ceph_cap_snap_cachep)
+		goto bad_cap_snap;
 	ceph_cap_flush_cachep = KMEM_CACHE(ceph_cap_flush,
 					   SLAB_RECLAIM_ACCOUNT|SLAB_MEM_SPREAD);
 	if (!ceph_cap_flush_cachep)
@@ -932,6 +936,8 @@ static int __init init_caches(void)
 bad_dentry:
 	kmem_cache_destroy(ceph_cap_flush_cachep);
 bad_cap_flush:
+	kmem_cache_destroy(ceph_cap_snap_cachep);
+bad_cap_snap:
 	kmem_cache_destroy(ceph_cap_cachep);
 bad_cap:
 	kmem_cache_destroy(ceph_inode_cachep);
@@ -948,6 +954,7 @@ static void destroy_caches(void)
 
 	kmem_cache_destroy(ceph_inode_cachep);
 	kmem_cache_destroy(ceph_cap_cachep);
+	kmem_cache_destroy(ceph_cap_snap_cachep);
 	kmem_cache_destroy(ceph_cap_flush_cachep);
 	kmem_cache_destroy(ceph_dentry_cachep);
 	kmem_cache_destroy(ceph_file_cachep);

fs/ceph/super.h

@@ -231,7 +231,7 @@ static inline void ceph_put_cap_snap(struct ceph_cap_snap *capsnap)
 	if (refcount_dec_and_test(&capsnap->nref)) {
 		if (capsnap->xattr_blob)
 			ceph_buffer_put(capsnap->xattr_blob);
-		kfree(capsnap);
+		kmem_cache_free(ceph_cap_snap_cachep, capsnap);
 	}
 }
 
@@ -884,6 +884,8 @@ struct ceph_snap_realm {
 
 	struct list_head dirty_item;     /* if realm needs new context */
 
+	struct list_head rebuild_item;   /* rebuild snap realms _downward_ in hierarchy */
+
 	/* the current set of snaps for this realm */
 	struct ceph_snap_context *cached_context;
 
@@ -939,7 +941,7 @@ extern void ceph_handle_snap(struct ceph_mds_client *mdsc,
 			     struct ceph_msg *msg);
 extern int __ceph_finish_cap_snap(struct ceph_inode_info *ci,
 				  struct ceph_cap_snap *capsnap);
-extern void ceph_cleanup_empty_realms(struct ceph_mds_client *mdsc);
+extern void ceph_cleanup_global_and_empty_realms(struct ceph_mds_client *mdsc);
 
 extern struct ceph_snapid_map *ceph_get_snapid_map(struct ceph_mds_client *mdsc,
 						   u64 snap);
@@ -1049,6 +1051,7 @@ static inline bool ceph_inode_is_shutdown(struct inode *inode)
 
 /* xattr.c */
 int __ceph_setxattr(struct inode *, const char *, const void *, size_t, int);
+int ceph_do_getvxattr(struct inode *inode, const char *name, void *value, size_t size);
 ssize_t __ceph_getxattr(struct inode *, const char *, void *, size_t);
 extern ssize_t ceph_listxattr(struct dentry *, char *, size_t);
 extern struct ceph_buffer *__ceph_build_xattrs_blob(struct ceph_inode_info *ci);
@@ -1214,7 +1217,7 @@ extern void __ceph_touch_fmode(struct ceph_inode_info *ci,
 /* addr.c */
 extern const struct address_space_operations ceph_aops;
 extern int ceph_mmap(struct file *file, struct vm_area_struct *vma);
-extern int ceph_uninline_data(struct file *filp, struct page *locked_page);
+extern int ceph_uninline_data(struct file *file);
 extern int ceph_pool_perm_check(struct inode *inode, int need);
 extern void ceph_pool_perm_destroy(struct ceph_mds_client* mdsc);
 int ceph_purge_inode_cap(struct inode *inode, struct ceph_cap *cap, bool *invalidate);

fs/ceph/xattr.c

@@ -923,10 +923,13 @@ ssize_t __ceph_getxattr(struct inode *inode, const char *name, void *value,
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_inode_xattr *xattr;
-	struct ceph_vxattr *vxattr = NULL;
+	struct ceph_vxattr *vxattr;
 	int req_mask;
 	ssize_t err;
 
+	if (strncmp(name, XATTR_CEPH_PREFIX, XATTR_CEPH_PREFIX_LEN))
+		goto handle_non_vxattrs;
+
 	/* let's see if a virtual xattr was requested */
 	vxattr = ceph_match_vxattr(inode, name);
 	if (vxattr) {
@@ -945,8 +948,14 @@ ssize_t __ceph_getxattr(struct inode *inode, const char *name, void *value,
 				err = -ERANGE;
 		}
 		return err;
+	} else {
+		err = ceph_do_getvxattr(inode, name, value, size);
+		/* this would happen with a new client and old server combo */
+		if (err == -EOPNOTSUPP)
+			err = -ENODATA;
+		return err;
 	}
-
+handle_non_vxattrs:
 	req_mask = __get_request_mask(inode);
 
 	spin_lock(&ci->i_ceph_lock);

include/linux/ceph/ceph_fs.h

@@ -29,7 +29,7 @@
 
 #define CEPH_INO_ROOT   1
 #define CEPH_INO_CEPH   2            /* hidden .ceph dir */
-#define CEPH_INO_DOTDOT 3	/* used by ceph fuse for parent (..) */
+#define CEPH_INO_GLOBAL_SNAPREALM  3 /* global dummy snaprealm */
 
 /* arbitrary limit on max # of monitors (cluster of 3 is typical) */
 #define CEPH_MAX_MON   31
@@ -328,6 +328,7 @@ enum {
 	CEPH_MDS_OP_LOOKUPPARENT = 0x00103,
 	CEPH_MDS_OP_LOOKUPINO  = 0x00104,
 	CEPH_MDS_OP_LOOKUPNAME = 0x00105,
+	CEPH_MDS_OP_GETVXATTR  = 0x00106,
 
 	CEPH_MDS_OP_SETXATTR   = 0x01105,
 	CEPH_MDS_OP_RMXATTR    = 0x01106,

include/linux/ceph/libceph.h

@@ -284,6 +284,7 @@ DEFINE_RB_LOOKUP_FUNC(name, type, keyfld, nodefld)
 
 extern struct kmem_cache *ceph_inode_cachep;
 extern struct kmem_cache *ceph_cap_cachep;
+extern struct kmem_cache *ceph_cap_snap_cachep;
 extern struct kmem_cache *ceph_cap_flush_cachep;
 extern struct kmem_cache *ceph_dentry_cachep;
 extern struct kmem_cache *ceph_file_cachep;

net/ceph/messenger_v2.c

@@ -1773,10 +1773,8 @@ static int prepare_read_data(struct ceph_connection *con)
 
 		bv.bv_page = con->bounce_page;
 		bv.bv_offset = 0;
-		set_in_bvec(con, &bv);
-	} else {
-		set_in_bvec(con, &bv);
 	}
+	set_in_bvec(con, &bv);
 	con->v2.in_state = IN_S_PREPARE_READ_DATA_CONT;
 	return 0;
 }
@@ -1807,10 +1805,8 @@ static void prepare_read_data_cont(struct ceph_connection *con)
 		if (ceph_test_opt(from_msgr(con->msgr), RXBOUNCE)) {
 			bv.bv_page = con->bounce_page;
 			bv.bv_offset = 0;
-			set_in_bvec(con, &bv);
-		} else {
-			set_in_bvec(con, &bv);
 		}
+		set_in_bvec(con, &bv);
 		WARN_ON(con->v2.in_state != IN_S_PREPARE_READ_DATA_CONT);
 		return;
 	}