Merge tag 'ceph-for-4.18-rc1' of git://github.com/ceph/ceph-client

Pull ceph updates from Ilya Dryomov:
 "The main piece is a set of libceph changes that revamps how OSD
  requests are aborted, improving CephFS ENOSPC handling and making
  "umount -f" actually work (Zheng and myself).

  The rest is mostly mount option handling cleanups from Chengguang and
  assorted fixes from Zheng, Luis and Dongsheng.

* tag 'ceph-for-4.18-rc1' of git://github.com/ceph/ceph-client: (31 commits)
  rbd: flush rbd_dev->watch_dwork after watch is unregistered
  ceph: update description of some mount options
  ceph: show ino32 if the value is different with default
  ceph: strengthen rsize/wsize/readdir_max_bytes validation
  ceph: fix alignment of rasize
  ceph: fix use-after-free in ceph_statfs()
  ceph: prevent i_version from going back
  ceph: fix wrong check for the case of updating link count
  libceph: allocate the locator string with GFP_NOFAIL
  libceph: make abort_on_full a per-osdc setting
  libceph: don't abort reads in ceph_osdc_abort_on_full()
  libceph: avoid a use-after-free during map check
  libceph: don't warn if req->r_abort_on_full is set
  libceph: use for_each_request() in ceph_osdc_abort_on_full()
  libceph: defer __complete_request() to a workqueue
  libceph: move more code into __complete_request()
  libceph: no need to call flush_workqueue() before destruction
  ceph: flush pending works before shutdown super
  ceph: abort osd requests on force umount
  libceph: introduce ceph_osdc_abort_requests()
  ...

Documentation/filesystems/ceph.txt

@@ -105,15 +105,13 @@ Mount Options
 	address its connection to the monitor originates from.
 
   wsize=X
-	Specify the maximum write size in bytes.  By default there is no
-	maximum.  Ceph will normally size writes based on the file stripe
-	size.
+	Specify the maximum write size in bytes.  Default: 16 MB.
 
   rsize=X
-	Specify the maximum read size in bytes.  Default: 64 MB.
+	Specify the maximum read size in bytes.  Default: 16 MB.
 
   rasize=X
-	Specify the maximum readahead.  Default: 8 MB.
+	Specify the maximum readahead size in bytes.  Default: 8 MB.
 
   mount_timeout=X
 	Specify the timeout value for mount (in seconds), in the case

drivers/block/rbd.c

@@ -2339,6 +2339,7 @@ static bool is_zero_bvecs(struct bio_vec *bvecs, u32 bytes)
 static int rbd_obj_issue_copyup(struct rbd_obj_request *obj_req, u32 bytes)
 {
 	unsigned int num_osd_ops = obj_req->osd_req->r_num_ops;
+	int ret;
 
 	dout("%s obj_req %p bytes %u\n", __func__, obj_req, bytes);
 	rbd_assert(obj_req->osd_req->r_ops[0].op == CEPH_OSD_OP_STAT);
@@ -2353,6 +2354,11 @@ static int rbd_obj_issue_copyup(struct rbd_obj_request *obj_req, u32 bytes)
 	if (!obj_req->osd_req)
 		return -ENOMEM;
 
+	ret = osd_req_op_cls_init(obj_req->osd_req, 0, CEPH_OSD_OP_CALL, "rbd",
+				  "copyup");
+	if (ret)
+		return ret;
+
 	/*
 	 * Only send non-zero copyup data to save some I/O and network
 	 * bandwidth -- zero copyup data is equivalent to the object not
@@ -2362,9 +2368,6 @@ static int rbd_obj_issue_copyup(struct rbd_obj_request *obj_req, u32 bytes)
 		dout("%s obj_req %p detected zeroes\n", __func__, obj_req);
 		bytes = 0;
 	}
-
-	osd_req_op_cls_init(obj_req->osd_req, 0, CEPH_OSD_OP_CALL, "rbd",
-			    "copyup");
 	osd_req_op_cls_request_data_bvecs(obj_req->osd_req, 0,
 					  obj_req->copyup_bvecs,
 					  obj_req->copyup_bvec_count,
@@ -3397,7 +3400,6 @@ static void cancel_tasks_sync(struct rbd_device *rbd_dev)
 {
 	dout("%s rbd_dev %p\n", __func__, rbd_dev);
 
-	cancel_delayed_work_sync(&rbd_dev->watch_dwork);
 	cancel_work_sync(&rbd_dev->acquired_lock_work);
 	cancel_work_sync(&rbd_dev->released_lock_work);
 	cancel_delayed_work_sync(&rbd_dev->lock_dwork);
@@ -3415,6 +3417,7 @@ static void rbd_unregister_watch(struct rbd_device *rbd_dev)
 	rbd_dev->watch_state = RBD_WATCH_STATE_UNREGISTERED;
 	mutex_unlock(&rbd_dev->watch_mutex);
 
+	cancel_delayed_work_sync(&rbd_dev->watch_dwork);
 	ceph_osdc_flush_notifies(&rbd_dev->rbd_client->client->osdc);
 }
 

fs/ceph/addr.c

@@ -1936,7 +1936,6 @@ static int __ceph_pool_perm_get(struct ceph_inode_info *ci,
 	err = ceph_osdc_start_request(&fsc->client->osdc, rd_req, false);
 
 	wr_req->r_mtime = ci->vfs_inode.i_mtime;
-	wr_req->r_abort_on_full = true;
 	err2 = ceph_osdc_start_request(&fsc->client->osdc, wr_req, false);
 
 	if (!err)

fs/ceph/dir.c

@@ -1486,6 +1486,8 @@ const struct file_operations ceph_dir_fops = {
 	.release = ceph_release,
 	.unlocked_ioctl = ceph_ioctl,
 	.fsync = ceph_fsync,
+	.lock = ceph_lock,
+	.flock = ceph_flock,
 };
 
 const struct file_operations ceph_snapdir_fops = {

fs/ceph/file.c

@@ -895,7 +895,6 @@ static void ceph_aio_retry_work(struct work_struct *work)
 	req->r_callback = ceph_aio_complete_req;
 	req->r_inode = inode;
 	req->r_priv = aio_req;
-	req->r_abort_on_full = true;
 
 	ret = ceph_osdc_start_request(req->r_osdc, req, false);
 out:

fs/ceph/inode.c

@@ -739,7 +739,7 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 	struct ceph_mds_client *mdsc = ceph_inode_to_client(inode)->mdsc;
 	struct ceph_mds_reply_inode *info = iinfo->in;
 	struct ceph_inode_info *ci = ceph_inode(inode);
-	int issued = 0, implemented, new_issued;
+	int issued, new_issued, info_caps;
 	struct timespec mtime, atime, ctime;
 	struct ceph_buffer *xattr_blob = NULL;
 	struct ceph_string *pool_ns = NULL;
@@ -754,8 +754,10 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 	     inode, ceph_vinop(inode), le64_to_cpu(info->version),
 	     ci->i_version);
 
+	info_caps = le32_to_cpu(info->cap.caps);
+
 	/* prealloc new cap struct */
-	if (info->cap.caps && ceph_snap(inode) == CEPH_NOSNAP)
+	if (info_caps && ceph_snap(inode) == CEPH_NOSNAP)
 		new_cap = ceph_get_cap(mdsc, caps_reservation);
 
 	/*
@@ -792,9 +794,9 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 	     le64_to_cpu(info->version) > (ci->i_version & ~1)))
 		new_version = true;
 
-	issued = __ceph_caps_issued(ci, &implemented);
-	issued |= implemented | __ceph_caps_dirty(ci);
-	new_issued = ~issued & le32_to_cpu(info->cap.caps);
+	__ceph_caps_issued(ci, &issued);
+	issued |= __ceph_caps_dirty(ci);
+	new_issued = ~issued & info_caps;
 
 	/* update inode */
 	inode->i_rdev = le32_to_cpu(info->rdev);
@@ -826,6 +828,11 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 				&ctime, &mtime, &atime);
 	}
 
+	if (new_version || (info_caps & CEPH_CAP_FILE_SHARED)) {
+		ci->i_files = le64_to_cpu(info->files);
+		ci->i_subdirs = le64_to_cpu(info->subdirs);
+	}
+
 	if (new_version ||
 	    (new_issued & (CEPH_CAP_ANY_FILE_RD | CEPH_CAP_ANY_FILE_WR))) {
 		s64 old_pool = ci->i_layout.pool_id;
@@ -854,6 +861,18 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 		}
 	}
 
+	/* layout and rstat are not tracked by capability, update them if
+	 * the inode info is from auth mds */
+	if (new_version || (info->cap.flags & CEPH_CAP_FLAG_AUTH)) {
+		if (S_ISDIR(inode->i_mode)) {
+			ci->i_dir_layout = iinfo->dir_layout;
+			ci->i_rbytes = le64_to_cpu(info->rbytes);
+			ci->i_rfiles = le64_to_cpu(info->rfiles);
+			ci->i_rsubdirs = le64_to_cpu(info->rsubdirs);
+			ceph_decode_timespec(&ci->i_rctime, &info->rctime);
+		}
+	}
+
 	/* xattrs */
 	/* note that if i_xattrs.len <= 4, i_xattrs.data will still be NULL. */
 	if ((ci->i_xattrs.version == 0 || !(issued & CEPH_CAP_XATTR_EXCL))  &&
@@ -870,6 +889,7 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 	}
 
 	/* finally update i_version */
+	if (le64_to_cpu(info->version) > ci->i_version)
 		ci->i_version = le64_to_cpu(info->version);
 
 	inode->i_mapping->a_ops = &ceph_aops;
@@ -918,15 +938,6 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 	case S_IFDIR:
 		inode->i_op = &ceph_dir_iops;
 		inode->i_fop = &ceph_dir_fops;
-
-		ci->i_dir_layout = iinfo->dir_layout;
-
-		ci->i_files = le64_to_cpu(info->files);
-		ci->i_subdirs = le64_to_cpu(info->subdirs);
-		ci->i_rbytes = le64_to_cpu(info->rbytes);
-		ci->i_rfiles = le64_to_cpu(info->rfiles);
-		ci->i_rsubdirs = le64_to_cpu(info->rsubdirs);
-		ceph_decode_timespec(&ci->i_rctime, &info->rctime);
 		break;
 	default:
 		pr_err("fill_inode %llx.%llx BAD mode 0%o\n",
@@ -934,12 +945,11 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 	}
 
 	/* were we issued a capability? */
-	if (info->cap.caps) {
+	if (info_caps) {
 		if (ceph_snap(inode) == CEPH_NOSNAP) {
-			unsigned caps = le32_to_cpu(info->cap.caps);
 			ceph_add_cap(inode, session,
 				     le64_to_cpu(info->cap.cap_id),
-				     cap_fmode, caps,
+				     cap_fmode, info_caps,
 				     le32_to_cpu(info->cap.wanted),
 				     le32_to_cpu(info->cap.seq),
 				     le32_to_cpu(info->cap.mseq),
@@ -949,7 +959,7 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 			/* set dir completion flag? */
 			if (S_ISDIR(inode->i_mode) &&
 			    ci->i_files == 0 && ci->i_subdirs == 0 &&
-			    (caps & CEPH_CAP_FILE_SHARED) &&
+			    (info_caps & CEPH_CAP_FILE_SHARED) &&
 			    (issued & CEPH_CAP_FILE_EXCL) == 0 &&
 			    !__ceph_dir_is_complete(ci)) {
 				dout(" marking %p complete (empty)\n", inode);
@@ -962,8 +972,8 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 			wake = true;
 		} else {
 			dout(" %p got snap_caps %s\n", inode,
-			     ceph_cap_string(le32_to_cpu(info->cap.caps)));
-			ci->i_snap_caps |= le32_to_cpu(info->cap.caps);
+			     ceph_cap_string(info_caps));
+			ci->i_snap_caps |= info_caps;
 			if (cap_fmode >= 0)
 				__ceph_get_fmode(ci, cap_fmode);
 		}
@@ -978,8 +988,7 @@ static int fill_inode(struct inode *inode, struct page *locked_page,
 		int cache_caps = CEPH_CAP_FILE_CACHE | CEPH_CAP_FILE_LAZYIO;
 		ci->i_inline_version = iinfo->inline_version;
 		if (ci->i_inline_version != CEPH_INLINE_NONE &&
-		    (locked_page ||
-		     (le32_to_cpu(info->cap.caps) & cache_caps)))
+		    (locked_page || (info_caps & cache_caps)))
 			fill_inline = true;
 	}
 
@@ -2178,6 +2187,7 @@ int __ceph_do_getattr(struct inode *inode, struct page *locked_page,
 	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;
 	int err;
 
 	if (ceph_snap(inode) == CEPH_SNAPDIR) {
@@ -2190,7 +2200,8 @@ int __ceph_do_getattr(struct inode *inode, struct page *locked_page,
 	if (!force && ceph_caps_issued_mask(ceph_inode(inode), mask, 1))
 		return 0;
 
-	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_GETATTR, USE_ANY_MDS);
+	mode = (mask & CEPH_STAT_RSTAT) ? USE_AUTH_MDS : USE_ANY_MDS;
+	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_GETATTR, mode);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
 	req->r_inode = inode;
@@ -2261,6 +2272,14 @@ int ceph_getattr(const struct path *path, struct kstat *stat,
 				stat->size = ci->i_files + ci->i_subdirs;
 			stat->blocks = 0;
 			stat->blksize = 65536;
+			/*
+			 * Some applications rely on the number of st_nlink
+			 * value on directories to be either 0 (if unlinked)
+			 * or 2 + number of subdirectories.
+			 */
+			if (stat->nlink == 1)
+				/* '.' + '..' + subdirs */
+				stat->nlink = 1 + 1 + ci->i_subdirs;
 		}
 	}
 	return err;

fs/ceph/super.c

@@ -45,7 +45,7 @@ static void ceph_put_super(struct super_block *s)
 static int ceph_statfs(struct dentry *dentry, struct kstatfs *buf)
 {
 	struct ceph_fs_client *fsc = ceph_inode_to_client(d_inode(dentry));
-	struct ceph_monmap *monmap = fsc->client->monc.monmap;
+	struct ceph_mon_client *monc = &fsc->client->monc;
 	struct ceph_statfs st;
 	u64 fsid;
 	int err;
@@ -58,7 +58,7 @@ static int ceph_statfs(struct dentry *dentry, struct kstatfs *buf)
 	}
 
 	dout("statfs\n");
-	err = ceph_monc_do_statfs(&fsc->client->monc, data_pool, &st);
+	err = ceph_monc_do_statfs(monc, data_pool, &st);
 	if (err < 0)
 		return err;
 
@@ -94,8 +94,11 @@ static int ceph_statfs(struct dentry *dentry, struct kstatfs *buf)
 	buf->f_namelen = NAME_MAX;
 
 	/* Must convert the fsid, for consistent values across arches */
-	fsid = le64_to_cpu(*(__le64 *)(&monmap->fsid)) ^
-	       le64_to_cpu(*((__le64 *)&monmap->fsid + 1));
+	mutex_lock(&monc->mutex);
+	fsid = le64_to_cpu(*(__le64 *)(&monc->monmap->fsid)) ^
+	       le64_to_cpu(*((__le64 *)&monc->monmap->fsid + 1));
+	mutex_unlock(&monc->mutex);
+
 	buf->f_fsid.val[0] = fsid & 0xffffffff;
 	buf->f_fsid.val[1] = fsid >> 32;
 
@@ -256,19 +259,19 @@ static int parse_fsopt_token(char *c, void *private)
 		break;
 		/* misc */
 	case Opt_wsize:
-		if (intval < PAGE_SIZE || intval > CEPH_MAX_WRITE_SIZE)
+		if (intval < (int)PAGE_SIZE || intval > CEPH_MAX_WRITE_SIZE)
 			return -EINVAL;
 		fsopt->wsize = ALIGN(intval, PAGE_SIZE);
 		break;
 	case Opt_rsize:
-		if (intval < PAGE_SIZE || intval > CEPH_MAX_READ_SIZE)
+		if (intval < (int)PAGE_SIZE || intval > CEPH_MAX_READ_SIZE)
 			return -EINVAL;
 		fsopt->rsize = ALIGN(intval, PAGE_SIZE);
 		break;
 	case Opt_rasize:
 		if (intval < 0)
 			return -EINVAL;
-		fsopt->rasize = ALIGN(intval + PAGE_SIZE - 1, PAGE_SIZE);
+		fsopt->rasize = ALIGN(intval, PAGE_SIZE);
 		break;
 	case Opt_caps_wanted_delay_min:
 		if (intval < 1)
@@ -286,7 +289,7 @@ static int parse_fsopt_token(char *c, void *private)
 		fsopt->max_readdir = intval;
 		break;
 	case Opt_readdir_max_bytes:
-		if (intval < PAGE_SIZE && intval != 0)
+		if (intval < (int)PAGE_SIZE && intval != 0)
 			return -EINVAL;
 		fsopt->max_readdir_bytes = intval;
 		break;
@@ -534,6 +537,8 @@ static int ceph_show_options(struct seq_file *m, struct dentry *root)
 		seq_puts(m, ",noasyncreaddir");
 	if ((fsopt->flags & CEPH_MOUNT_OPT_DCACHE) == 0)
 		seq_puts(m, ",nodcache");
+	if (fsopt->flags & CEPH_MOUNT_OPT_INO32)
+		seq_puts(m, ",ino32");
 	if (fsopt->flags & CEPH_MOUNT_OPT_FSCACHE) {
 		seq_show_option(m, "fsc", fsopt->fscache_uniq);
 	}
@@ -551,7 +556,7 @@ static int ceph_show_options(struct seq_file *m, struct dentry *root)
 
 	if (fsopt->mds_namespace)
 		seq_show_option(m, "mds_namespace", fsopt->mds_namespace);
-	if (fsopt->wsize)
+	if (fsopt->wsize != CEPH_MAX_WRITE_SIZE)
 		seq_printf(m, ",wsize=%d", fsopt->wsize);
 	if (fsopt->rsize != CEPH_MAX_READ_SIZE)
 		seq_printf(m, ",rsize=%d", fsopt->rsize);
@@ -616,7 +621,9 @@ static struct ceph_fs_client *create_fs_client(struct ceph_mount_options *fsopt,
 		err = PTR_ERR(fsc->client);
 		goto fail;
 	}
+
 	fsc->client->extra_mon_dispatch = extra_mon_dispatch;
+	fsc->client->osdc.abort_on_full = true;
 
 	if (!fsopt->mds_namespace) {
 		ceph_monc_want_map(&fsc->client->monc, CEPH_SUB_MDSMAP,
@@ -674,6 +681,13 @@ static struct ceph_fs_client *create_fs_client(struct ceph_mount_options *fsopt,
 	return ERR_PTR(err);
 }
 
+static void flush_fs_workqueues(struct ceph_fs_client *fsc)
+{
+	flush_workqueue(fsc->wb_wq);
+	flush_workqueue(fsc->pg_inv_wq);
+	flush_workqueue(fsc->trunc_wq);
+}
+
 static void destroy_fs_client(struct ceph_fs_client *fsc)
 {
 	dout("destroy_fs_client %p\n", fsc);
@@ -793,6 +807,7 @@ static void ceph_umount_begin(struct super_block *sb)
 	if (!fsc)
 		return;
 	fsc->mount_state = CEPH_MOUNT_SHUTDOWN;
+	ceph_osdc_abort_requests(&fsc->client->osdc, -EIO);
 	ceph_mdsc_force_umount(fsc->mdsc);
 	return;
 }
@@ -1088,6 +1103,8 @@ static void ceph_kill_sb(struct super_block *s)
 	dout("kill_sb %p\n", s);
 
 	ceph_mdsc_pre_umount(fsc->mdsc);
+	flush_fs_workqueues(fsc);
+
 	generic_shutdown_super(s);
 
 	fsc->client->extra_mon_dispatch = NULL;

fs/ceph/xattr.c

@@ -50,10 +50,14 @@ struct ceph_vxattr {
 	size_t name_size;	/* strlen(name) + 1 (for '\0') */
 	size_t (*getxattr_cb)(struct ceph_inode_info *ci, char *val,
 			      size_t size);
-	bool readonly, hidden;
 	bool (*exists_cb)(struct ceph_inode_info *ci);
+	unsigned int flags;
 };
 
+#define VXATTR_FLAG_READONLY		(1<<0)
+#define VXATTR_FLAG_HIDDEN		(1<<1)
+#define VXATTR_FLAG_RSTAT		(1<<2)
+
 /* layouts */
 
 static bool ceph_vxattrcb_layout_exists(struct ceph_inode_info *ci)
@@ -262,32 +266,31 @@ static size_t ceph_vxattrcb_quota_max_files(struct ceph_inode_info *ci,
 #define CEPH_XATTR_NAME2(_type, _name, _name2)	\
 	XATTR_CEPH_PREFIX #_type "." #_name "." #_name2
 
-#define XATTR_NAME_CEPH(_type, _name)					\
+#define XATTR_NAME_CEPH(_type, _name, _flags)				\
 	{								\
 		.name = CEPH_XATTR_NAME(_type, _name),			\
 		.name_size = sizeof (CEPH_XATTR_NAME(_type, _name)), \
 		.getxattr_cb = ceph_vxattrcb_ ## _type ## _ ## _name, \
-		.readonly = true,				\
-		.hidden = false,				\
 		.exists_cb = NULL,					\
+		.flags = (VXATTR_FLAG_READONLY | _flags),		\
 	}
+#define XATTR_RSTAT_FIELD(_type, _name)			\
+	XATTR_NAME_CEPH(_type, _name, VXATTR_FLAG_RSTAT)
 #define XATTR_LAYOUT_FIELD(_type, _name, _field)			\
 	{								\
 		.name = CEPH_XATTR_NAME2(_type, _name, _field),	\
 		.name_size = sizeof (CEPH_XATTR_NAME2(_type, _name, _field)), \
 		.getxattr_cb = ceph_vxattrcb_ ## _name ## _ ## _field, \
-		.readonly = false,				\
-		.hidden = true,			\
 		.exists_cb = ceph_vxattrcb_layout_exists,	\
+		.flags = VXATTR_FLAG_HIDDEN,			\
 	}
 #define XATTR_QUOTA_FIELD(_type, _name)					\
 	{								\
 		.name = CEPH_XATTR_NAME(_type, _name),			\
 		.name_size = sizeof(CEPH_XATTR_NAME(_type, _name)),	\
 		.getxattr_cb = ceph_vxattrcb_ ## _type ## _ ## _name,	\
-		.readonly = false,					\
-		.hidden = true,						\
 		.exists_cb = ceph_vxattrcb_quota_exists,		\
+		.flags = VXATTR_FLAG_HIDDEN,				\
 	}
 
 static struct ceph_vxattr ceph_dir_vxattrs[] = {
@@ -295,30 +298,28 @@ static struct ceph_vxattr ceph_dir_vxattrs[] = {
 		.name = "ceph.dir.layout",
 		.name_size = sizeof("ceph.dir.layout"),
 		.getxattr_cb = ceph_vxattrcb_layout,
-		.readonly = false,
-		.hidden = true,
 		.exists_cb = ceph_vxattrcb_layout_exists,
+		.flags = VXATTR_FLAG_HIDDEN,
 	},
 	XATTR_LAYOUT_FIELD(dir, layout, stripe_unit),
 	XATTR_LAYOUT_FIELD(dir, layout, stripe_count),
 	XATTR_LAYOUT_FIELD(dir, layout, object_size),
 	XATTR_LAYOUT_FIELD(dir, layout, pool),
 	XATTR_LAYOUT_FIELD(dir, layout, pool_namespace),
-	XATTR_NAME_CEPH(dir, entries),
-	XATTR_NAME_CEPH(dir, files),
-	XATTR_NAME_CEPH(dir, subdirs),
-	XATTR_NAME_CEPH(dir, rentries),
-	XATTR_NAME_CEPH(dir, rfiles),
-	XATTR_NAME_CEPH(dir, rsubdirs),
-	XATTR_NAME_CEPH(dir, rbytes),
-	XATTR_NAME_CEPH(dir, rctime),
+	XATTR_NAME_CEPH(dir, entries, 0),
+	XATTR_NAME_CEPH(dir, files, 0),
+	XATTR_NAME_CEPH(dir, subdirs, 0),
+	XATTR_RSTAT_FIELD(dir, rentries),
+	XATTR_RSTAT_FIELD(dir, rfiles),
+	XATTR_RSTAT_FIELD(dir, rsubdirs),
+	XATTR_RSTAT_FIELD(dir, rbytes),
+	XATTR_RSTAT_FIELD(dir, rctime),
 	{
 		.name = "ceph.quota",
 		.name_size = sizeof("ceph.quota"),
 		.getxattr_cb = ceph_vxattrcb_quota,
-		.readonly = false,
-		.hidden = true,
 		.exists_cb = ceph_vxattrcb_quota_exists,
+		.flags = VXATTR_FLAG_HIDDEN,
 	},
 	XATTR_QUOTA_FIELD(quota, max_bytes),
 	XATTR_QUOTA_FIELD(quota, max_files),
@@ -333,9 +334,8 @@ static struct ceph_vxattr ceph_file_vxattrs[] = {
 		.name = "ceph.file.layout",
 		.name_size = sizeof("ceph.file.layout"),
 		.getxattr_cb = ceph_vxattrcb_layout,
-		.readonly = false,
-		.hidden = true,
 		.exists_cb = ceph_vxattrcb_layout_exists,
+		.flags = VXATTR_FLAG_HIDDEN,
 	},
 	XATTR_LAYOUT_FIELD(file, layout, stripe_unit),
 	XATTR_LAYOUT_FIELD(file, layout, stripe_count),
@@ -374,9 +374,10 @@ static size_t __init vxattrs_name_size(struct ceph_vxattr *vxattrs)
 	struct ceph_vxattr *vxattr;
 	size_t size = 0;
 
-	for (vxattr = vxattrs; vxattr->name; vxattr++)
-		if (!vxattr->hidden)
+	for (vxattr = vxattrs; vxattr->name; vxattr++) {
+		if (!(vxattr->flags & VXATTR_FLAG_HIDDEN))
 			size += vxattr->name_size;
+	}
 
 	return size;
 }
@@ -809,7 +810,10 @@ ssize_t __ceph_getxattr(struct inode *inode, const char *name, void *value,
 	/* let's see if a virtual xattr was requested */
 	vxattr = ceph_match_vxattr(inode, name);
 	if (vxattr) {
-		err = ceph_do_getattr(inode, 0, true);
+		int mask = 0;
+		if (vxattr->flags & VXATTR_FLAG_RSTAT)
+			mask |= CEPH_STAT_RSTAT;
+		err = ceph_do_getattr(inode, mask, true);
 		if (err)
 			return err;
 		err = -ENODATA;
@@ -919,7 +923,7 @@ ssize_t ceph_listxattr(struct dentry *dentry, char *names, size_t size)
 	err = namelen;
 	if (vxattrs) {
 		for (i = 0; vxattrs[i].name; i++) {
-			if (!vxattrs[i].hidden &&
+			if (!(vxattrs[i].flags & VXATTR_FLAG_HIDDEN) &&
 			    !(vxattrs[i].exists_cb &&
 			      !vxattrs[i].exists_cb(ci))) {
 				len = sprintf(names, "%s", vxattrs[i].name);
@@ -1024,7 +1028,7 @@ int __ceph_setxattr(struct inode *inode, const char *name,
 
 	vxattr = ceph_match_vxattr(inode, name);
 	if (vxattr) {
-		if (vxattr->readonly)
+		if (vxattr->flags & VXATTR_FLAG_READONLY)
 			return -EOPNOTSUPP;
 		if (value && !strncmp(vxattr->name, "ceph.quota", 10))
 			check_realm = true;

include/linux/ceph/ceph_fs.h

@@ -628,6 +628,7 @@ int ceph_flags_to_mode(int flags);
 				 CEPH_CAP_XATTR_SHARED)
 #define CEPH_STAT_CAP_INLINE_DATA (CEPH_CAP_FILE_SHARED | \
 				   CEPH_CAP_FILE_RD)
+#define CEPH_STAT_RSTAT CEPH_CAP_FILE_WREXTEND
 
 #define CEPH_CAP_ANY_SHARED (CEPH_CAP_AUTH_SHARED |			\
 			      CEPH_CAP_LINK_SHARED |			\

include/linux/ceph/osd_client.h

@@ -170,6 +170,7 @@ struct ceph_osd_request {
 	u64             r_tid;              /* unique for this client */
 	struct rb_node  r_node;
 	struct rb_node  r_mc_node;          /* map check */
+	struct work_struct r_complete_work;
 	struct ceph_osd *r_osd;
 
 	struct ceph_osd_request_target r_t;
@@ -201,7 +202,6 @@ struct ceph_osd_request {
 	struct timespec r_mtime;              /* ditto */
 	u64 r_data_offset;                    /* ditto */
 	bool r_linger;                        /* don't resend on failure */
-	bool r_abort_on_full;		      /* return ENOSPC when full */
 
 	/* internal */
 	unsigned long r_stamp;                /* jiffies, send or check time */
@@ -347,6 +347,8 @@ struct ceph_osd_client {
 	struct rb_root         linger_map_checks;
 	atomic_t               num_requests;
 	atomic_t               num_homeless;
+	bool                   abort_on_full; /* abort w/ ENOSPC when full */
+	int                    abort_err;
 	struct delayed_work    timeout_work;
 	struct delayed_work    osds_timeout_work;
 #ifdef CONFIG_DEBUG_FS
@@ -359,6 +361,7 @@ struct ceph_osd_client {
 	struct ceph_msgpool	msgpool_op_reply;
 
 	struct workqueue_struct	*notify_wq;
+	struct workqueue_struct	*completion_wq;
 };
 
 static inline bool ceph_osdmap_flag(struct ceph_osd_client *osdc, int flag)
@@ -378,6 +381,7 @@ extern void ceph_osdc_handle_reply(struct ceph_osd_client *osdc,
 extern void ceph_osdc_handle_map(struct ceph_osd_client *osdc,
 				 struct ceph_msg *msg);
 void ceph_osdc_update_epoch_barrier(struct ceph_osd_client *osdc, u32 eb);
+void ceph_osdc_abort_requests(struct ceph_osd_client *osdc, int err);
 
 extern void osd_req_op_init(struct ceph_osd_request *osd_req,
 			    unsigned int which, u16 opcode, u32 flags);
@@ -440,7 +444,7 @@ extern void osd_req_op_cls_response_data_pages(struct ceph_osd_request *,
 					struct page **pages, u64 length,
 					u32 alignment, bool pages_from_pool,
 					bool own_pages);
-extern void osd_req_op_cls_init(struct ceph_osd_request *osd_req,
+extern int osd_req_op_cls_init(struct ceph_osd_request *osd_req,
 					unsigned int which, u16 opcode,
 					const char *class, const char *method);
 extern int osd_req_op_xattr_init(struct ceph_osd_request *osd_req, unsigned int which,

include/linux/ceph/osdmap.h

@@ -279,7 +279,7 @@ bool ceph_osds_changed(const struct ceph_osds *old_acting,
 		       const struct ceph_osds *new_acting,
 		       bool any_change);
 
-int __ceph_object_locator_to_pg(struct ceph_pg_pool_info *pi,
+void __ceph_object_locator_to_pg(struct ceph_pg_pool_info *pi,
 				 const struct ceph_object_id *oid,
 				 const struct ceph_object_locator *oloc,
 				 struct ceph_pg *raw_pgid);

net/ceph/messenger.c

@@ -168,12 +168,6 @@ static char tag_keepalive2 = CEPH_MSGR_TAG_KEEPALIVE2;
 static struct lock_class_key socket_class;
 #endif
 
-/*
- * When skipping (ignoring) a block of input we read it into a "skip
- * buffer," which is this many bytes in size.
- */
-#define SKIP_BUF_SIZE	1024
-
 static void queue_con(struct ceph_connection *con);
 static void cancel_con(struct ceph_connection *con);
 static void ceph_con_workfn(struct work_struct *);
@@ -520,12 +514,18 @@ static int ceph_tcp_connect(struct ceph_connection *con)
 	return 0;
 }
 
+/*
+ * If @buf is NULL, discard up to @len bytes.
+ */
 static int ceph_tcp_recvmsg(struct socket *sock, void *buf, size_t len)
 {
 	struct kvec iov = {buf, len};
 	struct msghdr msg = { .msg_flags = MSG_DONTWAIT | MSG_NOSIGNAL };
 	int r;
 
+	if (!buf)
+		msg.msg_flags |= MSG_TRUNC;
+
 	iov_iter_kvec(&msg.msg_iter, READ | ITER_KVEC, &iov, 1, len);
 	r = sock_recvmsg(sock, &msg, msg.msg_flags);
 	if (r == -EAGAIN)
@@ -2575,9 +2575,6 @@ static int try_write(struct ceph_connection *con)
 	    con->state != CON_STATE_OPEN)
 		return 0;
 
-more:
-	dout("try_write out_kvec_bytes %d\n", con->out_kvec_bytes);
-
 	/* open the socket first? */
 	if (con->state == CON_STATE_PREOPEN) {
 		BUG_ON(con->sock);
@@ -2598,7 +2595,8 @@ static int try_write(struct ceph_connection *con)
 		}
 	}
 
-more_kvec:
+more:
+	dout("try_write out_kvec_bytes %d\n", con->out_kvec_bytes);
 	BUG_ON(!con->sock);
 
 	/* kvec data queued? */
@@ -2623,7 +2621,7 @@ static int try_write(struct ceph_connection *con)
 
 		ret = write_partial_message_data(con);
 		if (ret == 1)
-			goto more_kvec;  /* we need to send the footer, too! */
+			goto more;  /* we need to send the footer, too! */
 		if (ret == 0)
 			goto out;
 		if (ret < 0) {
@@ -2659,8 +2657,6 @@ static int try_write(struct ceph_connection *con)
 	return ret;
 }
 
-
-
 /*
  * Read what we can from the socket.
  */
@@ -2721,16 +2717,11 @@ static int try_read(struct ceph_connection *con)
 	if (con->in_base_pos < 0) {
 		/*
 		 * skipping + discarding content.
-		 *
-		 * FIXME: there must be a better way to do this!
 		 */
-		static char buf[SKIP_BUF_SIZE];
-		int skip = min((int) sizeof (buf), -con->in_base_pos);
-
-		dout("skipping %d / %d bytes\n", skip, -con->in_base_pos);
-		ret = ceph_tcp_recvmsg(con->sock, buf, skip);
+		ret = ceph_tcp_recvmsg(con->sock, NULL, -con->in_base_pos);
 		if (ret <= 0)
 			goto out;
+		dout("skipped %d / %d bytes\n", ret, -con->in_base_pos);
 		con->in_base_pos += ret;
 		if (con->in_base_pos)
 			goto more;

net/ceph/osdmap.c

@@ -2146,7 +2146,7 @@ bool ceph_osds_changed(const struct ceph_osds *old_acting,
  * Should only be called with target_oid and target_oloc (as opposed to
  * base_oid and base_oloc), since tiering isn't taken into account.
  */
-int __ceph_object_locator_to_pg(struct ceph_pg_pool_info *pi,
+void __ceph_object_locator_to_pg(struct ceph_pg_pool_info *pi,
 				 const struct ceph_object_id *oid,
 				 const struct ceph_object_locator *oloc,
 				 struct ceph_pg *raw_pgid)
@@ -2165,11 +2165,8 @@ int __ceph_object_locator_to_pg(struct ceph_pg_pool_info *pi,
 		int nsl = oloc->pool_ns->len;
 		size_t total = nsl + 1 + oid->name_len;
 
-		if (total > sizeof(stack_buf)) {
-			buf = kmalloc(total, GFP_NOIO);
-			if (!buf)
-				return -ENOMEM;
-		}
+		if (total > sizeof(stack_buf))
+			buf = kmalloc(total, GFP_NOIO | __GFP_NOFAIL);
 		memcpy(buf, oloc->pool_ns->str, nsl);
 		buf[nsl] = '\037';
 		memcpy(buf + nsl + 1, oid->name, oid->name_len);
@@ -2181,7 +2178,6 @@ int __ceph_object_locator_to_pg(struct ceph_pg_pool_info *pi,
 		     oid->name, nsl, oloc->pool_ns->str,
 		     raw_pgid->pool, raw_pgid->seed);
 	}
-	return 0;
 }
 
 int ceph_object_locator_to_pg(struct ceph_osdmap *osdmap,
@@ -2195,7 +2191,8 @@ int ceph_object_locator_to_pg(struct ceph_osdmap *osdmap,
 	if (!pi)
 		return -ENOENT;
 
-	return __ceph_object_locator_to_pg(pi, oid, oloc, raw_pgid);
+	__ceph_object_locator_to_pg(pi, oid, oloc, raw_pgid);
+	return 0;
 }
 EXPORT_SYMBOL(ceph_object_locator_to_pg);