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()
  ...

Documentation/filesystems/ceph.txt

@@ -98,11 +98,10 @@ Mount Options
 	size.
 
   rsize=X
-	Specify the maximum read size in bytes.  By default there is no
-	maximum.
+	Specify the maximum read size in bytes.  Default: 64 MB.
 
   rasize=X
-	Specify the maximum readahead.
+	Specify the maximum readahead.  Default: 8 MB.
 
   mount_timeout=X
 	Specify the timeout value for mount (in seconds), in the case

drivers/block/rbd_types.h

@@ -25,8 +25,8 @@
  */
 
 #define RBD_HEADER_PREFIX      "rbd_header."
-#define RBD_DATA_PREFIX        "rbd_data."
 #define RBD_ID_PREFIX          "rbd_id."
+#define RBD_V2_DATA_FORMAT     "%s.%016llx"
 
 #define RBD_LOCK_NAME          "rbd_lock"
 #define RBD_LOCK_TAG           "internal"
@@ -42,13 +42,14 @@ enum rbd_notify_op {
 /*
  * For format version 1, rbd image 'foo' consists of objects
  *   foo.rbd		- image metadata
- *   rb.<idhi>.<idlo>.00000000
- *   rb.<idhi>.<idlo>.00000001
+ *   rb.<idhi>.<idlo>.<extra>.000000000000
+ *   rb.<idhi>.<idlo>.<extra>.000000000001
  *   ...		- data
  * There is no notion of a persistent image id in rbd format 1.
  */
 
 #define RBD_SUFFIX		".rbd"
+#define RBD_V1_DATA_FORMAT	"%s.%012llx"
 
 #define RBD_DIRECTORY           "rbd_directory"
 #define RBD_INFO                "rbd_info"
@@ -57,9 +58,6 @@ enum rbd_notify_op {
 #define RBD_MIN_OBJ_ORDER       16
 #define RBD_MAX_OBJ_ORDER       30
 
-#define RBD_COMP_NONE		0
-#define RBD_CRYPT_NONE		0
-
 #define RBD_HEADER_TEXT		"<<< Rados Block Device Image >>>\n"
 #define RBD_HEADER_SIGNATURE	"RBD"
 #define RBD_HEADER_VERSION	"001.005"

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 @@ static int ceph_writepages_start(struct address_space *mapping,
 					&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 @@ static int ceph_writepages_start(struct address_space *mapping,
 						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);

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);
 		}
 	}

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 @@ static int try_get_cap_refs(struct ceph_inode_info *ci, int need, int want,
 
 		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 @@ static void handle_cap_export(struct inode *inode, struct ceph_mds_caps *ex,
 			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 @@ static void handle_cap_export(struct inode *inode, struct ceph_mds_caps *ex,
 	} 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) {

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");

fs/ceph/dir.c

@@ -371,7 +371,7 @@ static int ceph_readdir(struct file *file, struct dir_context *ctx)
 		/* 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 @@ static int ceph_readdir(struct file *file, struct dir_context *ctx)
 		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))

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);
 

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 @@ static void ceph_aio_retry_work(struct work_struct *work)
 	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);

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)

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;

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)

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;

include/linux/ceph/osd_client.h

@@ -22,7 +22,6 @@ struct ceph_osd_client;
  * completion callback for async writepages
  */
 typedef void (*ceph_osdc_callback_t)(struct ceph_osd_request *);
-typedef void (*ceph_osdc_unsafe_callback_t)(struct ceph_osd_request *, bool);
 
 #define CEPH_HOMELESS_OSD	-1
 
@@ -170,15 +169,12 @@ struct ceph_osd_request {
 	unsigned int		r_num_ops;
 
 	int               r_result;
-	bool              r_got_reply;
 
 	struct ceph_osd_client *r_osdc;
 	struct kref       r_kref;
 	bool              r_mempool;
-	struct completion r_completion;
-	struct completion r_done_completion;  /* fsync waiter */
+	struct completion r_completion;       /* private to osd_client.c */
 	ceph_osdc_callback_t r_callback;
-	ceph_osdc_unsafe_callback_t r_unsafe_callback;
 	struct list_head  r_unsafe_item;
 
 	struct inode *r_inode;         	      /* for use by callbacks */

include/linux/ceph/osdmap.h

@@ -57,7 +57,7 @@ static inline bool ceph_can_shift_osds(struct ceph_pg_pool_info *pool)
 	case CEPH_POOL_TYPE_EC:
 		return false;
 	default:
-		BUG_ON(1);
+		BUG();
 	}
 }
 
@@ -81,13 +81,6 @@ void ceph_oloc_copy(struct ceph_object_locator *dest,
 		    const struct ceph_object_locator *src);
 void ceph_oloc_destroy(struct ceph_object_locator *oloc);
 
-/*
- * Maximum supported by kernel client object name length
- *
- * (probably outdated: must be >= RBD_MAX_MD_NAME_LEN -- currently 100)
- */
-#define CEPH_MAX_OID_NAME_LEN 100
-
 /*
  * 51-char inline_name is long enough for all cephfs and all but one
  * rbd requests: <imgname> in "<imgname>.rbd"/"rbd_id.<imgname>" can be
@@ -173,8 +166,8 @@ struct ceph_osdmap {
 	 * the list of osds that store+replicate them. */
 	struct crush_map *crush;
 
-	struct mutex crush_scratch_mutex;
-	int crush_scratch_ary[CEPH_PG_MAX_SIZE * 3];
+	struct mutex crush_workspace_mutex;
+	void *crush_workspace;
 };
 
 static inline bool ceph_osd_exists(struct ceph_osdmap *map, int osd)

include/linux/ceph/rados.h

@@ -50,7 +50,7 @@ struct ceph_timespec {
 #define CEPH_PG_LAYOUT_LINEAR 2
 #define CEPH_PG_LAYOUT_HYBRID 3
 
-#define CEPH_PG_MAX_SIZE      16  /* max # osds in a single pg */
+#define CEPH_PG_MAX_SIZE      32  /* max # osds in a single pg */
 
 /*
  * placement group.

include/linux/crush/crush.h

@@ -135,13 +135,6 @@ struct crush_bucket {
 	__u32 size;      /* num items */
 	__s32 *items;
 
-	/*
-	 * cached random permutation: used for uniform bucket and for
-	 * the linear search fallback for the other bucket types.
-	 */
-	__u32 perm_x;  /* @x for which *perm is defined */
-	__u32 perm_n;  /* num elements of *perm that are permuted/defined */
-	__u32 *perm;
 };
 
 struct crush_bucket_uniform {
@@ -211,6 +204,21 @@ struct crush_map {
 	 * device fails. */
 	__u8 chooseleaf_stable;
 
+	/*
+	 * This value is calculated after decode or construction by
+	 * the builder. It is exposed here (rather than having a
+	 * 'build CRUSH working space' function) so that callers can
+	 * reserve a static buffer, allocate space on the stack, or
+	 * otherwise avoid calling into the heap allocator if they
+	 * want to. The size of the working space depends on the map,
+	 * while the size of the scratch vector passed to the mapper
+	 * depends on the size of the desired result set.
+	 *
+	 * Nothing stops the caller from allocating both in one swell
+	 * foop and passing in two points, though.
+	 */
+	size_t working_size;
+
 #ifndef __KERNEL__
 	/*
 	 * version 0 (original) of straw_calc has various flaws.  version 1
@@ -248,4 +256,23 @@ static inline int crush_calc_tree_node(int i)
 	return ((i+1) << 1)-1;
 }
 
+/*
+ * These data structures are private to the CRUSH implementation. They
+ * are exposed in this header file because builder needs their
+ * definitions to calculate the total working size.
+ *
+ * Moving this out of the crush map allow us to treat the CRUSH map as
+ * immutable within the mapper and removes the requirement for a CRUSH
+ * map lock.
+ */
+struct crush_work_bucket {
+	__u32 perm_x; /* @x for which *perm is defined */
+	__u32 perm_n; /* num elements of *perm that are permuted/defined */
+	__u32 *perm;  /* Permutation of the bucket's items */
+};
+
+struct crush_work {
+	struct crush_work_bucket **work; /* Per-bucket working store */
+};
+
 #endif

include/linux/crush/mapper.h

@@ -15,6 +15,20 @@ extern int crush_do_rule(const struct crush_map *map,
 			 int ruleno,
 			 int x, int *result, int result_max,
 			 const __u32 *weights, int weight_max,
-			 int *scratch);
+			 void *cwin);
+
+/*
+ * Returns the exact amount of workspace that will need to be used
+ * for a given combination of crush_map and result_max. The caller can
+ * then allocate this much on its own, either on the stack, in a
+ * per-thread long-lived buffer, or however it likes.
+ */
+static inline size_t crush_work_size(const struct crush_map *map,
+				     int result_max)
+{
+	return map->working_size + result_max * 3 * sizeof(__u32);
+}
+
+void crush_init_workspace(const struct crush_map *map, void *v);
 
 #endif

net/ceph/cls_lock_client.c

@@ -69,8 +69,8 @@ int ceph_cls_lock(struct ceph_osd_client *osdc,
 	dout("%s lock_name %s type %d cookie %s tag %s desc %s flags 0x%x\n",
 	     __func__, lock_name, type, cookie, tag, desc, flags);
 	ret = ceph_osdc_call(osdc, oid, oloc, "lock", "lock",
-			     CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK,
-			     lock_op_page, lock_op_buf_size, NULL, NULL);
+			     CEPH_OSD_FLAG_WRITE, lock_op_page,
+			     lock_op_buf_size, NULL, NULL);
 
 	dout("%s: status %d\n", __func__, ret);
 	__free_page(lock_op_page);
@@ -117,8 +117,8 @@ int ceph_cls_unlock(struct ceph_osd_client *osdc,
 
 	dout("%s lock_name %s cookie %s\n", __func__, lock_name, cookie);
 	ret = ceph_osdc_call(osdc, oid, oloc, "lock", "unlock",
-			     CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK,
-			     unlock_op_page, unlock_op_buf_size, NULL, NULL);
+			     CEPH_OSD_FLAG_WRITE, unlock_op_page,
+			     unlock_op_buf_size, NULL, NULL);
 
 	dout("%s: status %d\n", __func__, ret);
 	__free_page(unlock_op_page);
@@ -170,8 +170,8 @@ int ceph_cls_break_lock(struct ceph_osd_client *osdc,
 	dout("%s lock_name %s cookie %s locker %s%llu\n", __func__, lock_name,
 	     cookie, ENTITY_NAME(*locker));
 	ret = ceph_osdc_call(osdc, oid, oloc, "lock", "break_lock",
-			     CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK,
-			     break_op_page, break_op_buf_size, NULL, NULL);
+			     CEPH_OSD_FLAG_WRITE, break_op_page,
+			     break_op_buf_size, NULL, NULL);
 
 	dout("%s: status %d\n", __func__, ret);
 	__free_page(break_op_page);
@@ -278,7 +278,7 @@ int ceph_cls_lock_info(struct ceph_osd_client *osdc,
 	int get_info_op_buf_size;
 	int name_len = strlen(lock_name);
 	struct page *get_info_op_page, *reply_page;
-	size_t reply_len;
+	size_t reply_len = PAGE_SIZE;
 	void *p, *end;
 	int ret;
 

net/ceph/crush/crush.c

@@ -45,7 +45,6 @@ int crush_get_bucket_item_weight(const struct crush_bucket *b, int p)
 
 void crush_destroy_bucket_uniform(struct crush_bucket_uniform *b)
 {
-	kfree(b->h.perm);
 	kfree(b->h.items);
 	kfree(b);
 }
@@ -54,14 +53,12 @@ void crush_destroy_bucket_list(struct crush_bucket_list *b)
 {
 	kfree(b->item_weights);
 	kfree(b->sum_weights);
-	kfree(b->h.perm);
 	kfree(b->h.items);
 	kfree(b);
 }
 
 void crush_destroy_bucket_tree(struct crush_bucket_tree *b)
 {
-	kfree(b->h.perm);
 	kfree(b->h.items);
 	kfree(b->node_weights);
 	kfree(b);
@@ -71,7 +68,6 @@ void crush_destroy_bucket_straw(struct crush_bucket_straw *b)
 {
 	kfree(b->straws);
 	kfree(b->item_weights);
-	kfree(b->h.perm);
 	kfree(b->h.items);
 	kfree(b);
 }
@@ -79,7 +75,6 @@ void crush_destroy_bucket_straw(struct crush_bucket_straw *b)
 void crush_destroy_bucket_straw2(struct crush_bucket_straw2 *b)
 {
 	kfree(b->item_weights);
-	kfree(b->h.perm);
 	kfree(b->h.items);
 	kfree(b);
 }

net/ceph/crypto.c

@@ -3,6 +3,7 @@
 
 #include <linux/err.h>
 #include <linux/scatterlist.h>
+#include <linux/sched.h>
 #include <linux/slab.h>
 #include <crypto/aes.h>
 #include <crypto/skcipher.h>

net/ceph/osd_client.c

@@ -460,7 +460,6 @@ static void request_init(struct ceph_osd_request *req)
 
 	kref_init(&req->r_kref);
 	init_completion(&req->r_completion);
-	init_completion(&req->r_done_completion);
 	RB_CLEAR_NODE(&req->r_node);
 	RB_CLEAR_NODE(&req->r_mc_node);
 	INIT_LIST_HEAD(&req->r_unsafe_item);
@@ -672,6 +671,7 @@ void osd_req_op_extent_update(struct ceph_osd_request *osd_req,
 	BUG_ON(length > previous);
 
 	op->extent.length = length;
+	if (op->op == CEPH_OSD_OP_WRITE || op->op == CEPH_OSD_OP_WRITEFULL)
 		op->indata_len -= previous - length;
 }
 EXPORT_SYMBOL(osd_req_op_extent_update);
@@ -1636,7 +1636,7 @@ static void __submit_request(struct ceph_osd_request *req, bool wrlocked)
 	bool need_send = false;
 	bool promoted = false;
 
-	WARN_ON(req->r_tid || req->r_got_reply);
+	WARN_ON(req->r_tid);
 	dout("%s req %p wrlocked %d\n", __func__, req, wrlocked);
 
 again:
@@ -1704,17 +1704,10 @@ static void __submit_request(struct ceph_osd_request *req, bool wrlocked)
 
 static void account_request(struct ceph_osd_request *req)
 {
-	unsigned int mask = CEPH_OSD_FLAG_ACK | CEPH_OSD_FLAG_ONDISK;
+	WARN_ON(req->r_flags & (CEPH_OSD_FLAG_ACK | CEPH_OSD_FLAG_ONDISK));
+	WARN_ON(!(req->r_flags & (CEPH_OSD_FLAG_READ | CEPH_OSD_FLAG_WRITE)));
 
-	if (req->r_flags & CEPH_OSD_FLAG_READ) {
-		WARN_ON(req->r_flags & mask);
-		req->r_flags |= CEPH_OSD_FLAG_ACK;
-	} else if (req->r_flags & CEPH_OSD_FLAG_WRITE)
-		WARN_ON(!(req->r_flags & mask));
-	else
-		WARN_ON(1);
-
-	WARN_ON(req->r_unsafe_callback && (req->r_flags & mask) != mask);
+	req->r_flags |= CEPH_OSD_FLAG_ONDISK;
 	atomic_inc(&req->r_osdc->num_requests);
 }
 
@@ -1749,15 +1742,15 @@ static void finish_request(struct ceph_osd_request *req)
 
 static void __complete_request(struct ceph_osd_request *req)
 {
-	if (req->r_callback)
+	if (req->r_callback) {
+		dout("%s req %p tid %llu cb %pf result %d\n", __func__, req,
+		     req->r_tid, req->r_callback, req->r_result);
 		req->r_callback(req);
-	else
-		complete_all(&req->r_completion);
+	}
 }
 
 /*
- * Note that this is open-coded in handle_reply(), which has to deal
- * with ack vs commit, dup acks, etc.
+ * This is open-coded in handle_reply().
  */
 static void complete_request(struct ceph_osd_request *req, int err)
 {
@@ -1766,7 +1759,7 @@ static void complete_request(struct ceph_osd_request *req, int err)
 	req->r_result = err;
 	finish_request(req);
 	__complete_request(req);
-	complete_all(&req->r_done_completion);
+	complete_all(&req->r_completion);
 	ceph_osdc_put_request(req);
 }
 
@@ -1792,7 +1785,7 @@ static void cancel_request(struct ceph_osd_request *req)
 
 	cancel_map_check(req);
 	finish_request(req);
-	complete_all(&req->r_done_completion);
+	complete_all(&req->r_completion);
 	ceph_osdc_put_request(req);
 }
 
@@ -2169,7 +2162,6 @@ static void linger_commit_cb(struct ceph_osd_request *req)
 	mutex_lock(&lreq->lock);
 	dout("%s lreq %p linger_id %llu result %d\n", __func__, lreq,
 	     lreq->linger_id, req->r_result);
-	WARN_ON(!__linger_registered(lreq));
 	linger_reg_commit_complete(lreq, req->r_result);
 	lreq->committed = true;
 
@@ -2785,31 +2777,8 @@ static int decode_MOSDOpReply(const struct ceph_msg *msg, struct MOSDOpReply *m)
 }
 
 /*
- * We are done with @req if
- *   - @m is a safe reply, or
- *   - @m is an unsafe reply and we didn't want a safe one
- */
-static bool done_request(const struct ceph_osd_request *req,
-			 const struct MOSDOpReply *m)
-{
-	return (m->result < 0 ||
-		(m->flags & CEPH_OSD_FLAG_ONDISK) ||
-		!(req->r_flags & CEPH_OSD_FLAG_ONDISK));
-}
-
-/*
- * handle osd op reply.  either call the callback if it is specified,
- * or do the completion to wake up the waiting thread.
- *
- * ->r_unsafe_callback is set?	yes			no
- *
- * first reply is OK (needed	r_cb/r_completion,	r_cb/r_completion,
- * any or needed/got safe)	r_done_completion	r_done_completion
- *
- * first reply is unsafe	r_unsafe_cb(true)	(nothing)
- *
- * when we get the safe reply	r_unsafe_cb(false),	r_cb/r_completion,
- *				r_done_completion	r_done_completion
+ * Handle MOSDOpReply.  Set ->r_result and call the callback if it is
+ * specified.
  */
 static void handle_reply(struct ceph_osd *osd, struct ceph_msg *msg)
 {
@@ -2818,7 +2787,6 @@ static void handle_reply(struct ceph_osd *osd, struct ceph_msg *msg)
 	struct MOSDOpReply m;
 	u64 tid = le64_to_cpu(msg->hdr.tid);
 	u32 data_len = 0;
-	bool already_acked;
 	int ret;
 	int i;
 
@@ -2897,50 +2865,22 @@ static void handle_reply(struct ceph_osd *osd, struct ceph_msg *msg)
 		       le32_to_cpu(msg->hdr.data_len), req->r_tid);
 		goto fail_request;
 	}
-	dout("%s req %p tid %llu acked %d result %d data_len %u\n", __func__,
-	     req, req->r_tid, req->r_got_reply, m.result, data_len);
+	dout("%s req %p tid %llu result %d data_len %u\n", __func__,
+	     req, req->r_tid, m.result, data_len);
 
-	already_acked = req->r_got_reply;
-	if (!already_acked) {
+	/*
+	 * Since we only ever request ONDISK, we should only ever get
+	 * one (type of) reply back.
+	 */
+	WARN_ON(!(m.flags & CEPH_OSD_FLAG_ONDISK));
 	req->r_result = m.result ?: data_len;
-		req->r_replay_version = m.replay_version; /* struct */
-		req->r_got_reply = true;
-	} else if (!(m.flags & CEPH_OSD_FLAG_ONDISK)) {
-		dout("req %p tid %llu dup ack\n", req, req->r_tid);
-		goto out_unlock_session;
-	}
-
-	if (done_request(req, &m)) {
 	finish_request(req);
-		if (req->r_linger) {
-			WARN_ON(req->r_unsafe_callback);
-			dout("req %p tid %llu cb (locked)\n", req, req->r_tid);
-			__complete_request(req);
-		}
-	}
-
 	mutex_unlock(&osd->lock);
 	up_read(&osdc->lock);
 
-	if (done_request(req, &m)) {
-		if (already_acked && req->r_unsafe_callback) {
-			dout("req %p tid %llu safe-cb\n", req, req->r_tid);
-			req->r_unsafe_callback(req, false);
-		} else if (!req->r_linger) {
-			dout("req %p tid %llu cb\n", req, req->r_tid);
 	__complete_request(req);
-		}
-		complete_all(&req->r_done_completion);
+	complete_all(&req->r_completion);
 	ceph_osdc_put_request(req);
-	} else {
-		if (req->r_unsafe_callback) {
-			dout("req %p tid %llu unsafe-cb\n", req, req->r_tid);
-			req->r_unsafe_callback(req, true);
-		} else {
-			WARN_ON(1);
-		}
-	}
-
 	return;
 
 fail_request:
@@ -3540,7 +3480,7 @@ void ceph_osdc_sync(struct ceph_osd_client *osdc)
 			up_read(&osdc->lock);
 			dout("%s waiting on req %p tid %llu last_tid %llu\n",
 			     __func__, req, req->r_tid, last_tid);
-			wait_for_completion(&req->r_done_completion);
+			wait_for_completion(&req->r_completion);
 			ceph_osdc_put_request(req);
 			goto again;
 		}
@@ -3599,7 +3539,7 @@ ceph_osdc_watch(struct ceph_osd_client *osdc,
 
 	ceph_oid_copy(&lreq->t.base_oid, oid);
 	ceph_oloc_copy(&lreq->t.base_oloc, oloc);
-	lreq->t.flags = CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK;
+	lreq->t.flags = CEPH_OSD_FLAG_WRITE;
 	lreq->mtime = CURRENT_TIME;
 
 	lreq->reg_req = alloc_linger_request(lreq);
@@ -3657,7 +3597,7 @@ int ceph_osdc_unwatch(struct ceph_osd_client *osdc,
 
 	ceph_oid_copy(&req->r_base_oid, &lreq->t.base_oid);
 	ceph_oloc_copy(&req->r_base_oloc, &lreq->t.base_oloc);
-	req->r_flags = CEPH_OSD_FLAG_WRITE | CEPH_OSD_FLAG_ONDISK;
+	req->r_flags = CEPH_OSD_FLAG_WRITE;
 	req->r_mtime = CURRENT_TIME;
 	osd_req_op_watch_init(req, 0, lreq->linger_id,
 			      CEPH_OSD_WATCH_OP_UNWATCH);
@@ -4022,7 +3962,7 @@ EXPORT_SYMBOL(ceph_osdc_maybe_request_map);
  * Execute an OSD class method on an object.
  *
  * @flags: CEPH_OSD_FLAG_*
- * @resp_len: out param for reply length
+ * @resp_len: in/out param for reply length
  */
 int ceph_osdc_call(struct ceph_osd_client *osdc,
 		   struct ceph_object_id *oid,
@@ -4035,6 +3975,9 @@ int ceph_osdc_call(struct ceph_osd_client *osdc,
 	struct ceph_osd_request *req;
 	int ret;
 
+	if (req_len > PAGE_SIZE || (resp_page && *resp_len > PAGE_SIZE))
+		return -E2BIG;
+
 	req = ceph_osdc_alloc_request(osdc, NULL, 1, false, GFP_NOIO);
 	if (!req)
 		return -ENOMEM;
@@ -4053,7 +3996,7 @@ int ceph_osdc_call(struct ceph_osd_client *osdc,
 						  0, false, false);
 	if (resp_page)
 		osd_req_op_cls_response_data_pages(req, 0, &resp_page,
-						   PAGE_SIZE, 0, false, false);
+						   *resp_len, 0, false, false);
 
 	ceph_osdc_start_request(osdc, req, false);
 	ret = ceph_osdc_wait_request(osdc, req);
@@ -4220,8 +4163,7 @@ int ceph_osdc_writepages(struct ceph_osd_client *osdc, struct ceph_vino vino,
 	int page_align = off & ~PAGE_MASK;
 
 	req = ceph_osdc_new_request(osdc, layout, vino, off, &len, 0, 1,
-				    CEPH_OSD_OP_WRITE,
-				    CEPH_OSD_FLAG_ONDISK | CEPH_OSD_FLAG_WRITE,
+				    CEPH_OSD_OP_WRITE, CEPH_OSD_FLAG_WRITE,
 				    snapc, truncate_seq, truncate_size,
 				    true);
 	if (IS_ERR(req))

net/ceph/osdmap.c

@@ -153,6 +153,32 @@ static int skip_name_map(void **p, void *end)
         return -EINVAL;
 }
 
+static void crush_finalize(struct crush_map *c)
+{
+	__s32 b;
+
+	/* Space for the array of pointers to per-bucket workspace */
+	c->working_size = sizeof(struct crush_work) +
+	    c->max_buckets * sizeof(struct crush_work_bucket *);
+
+	for (b = 0; b < c->max_buckets; b++) {
+		if (!c->buckets[b])
+			continue;
+
+		switch (c->buckets[b]->alg) {
+		default:
+			/*
+			 * The base case, permutation variables and
+			 * the pointer to the permutation array.
+			 */
+			c->working_size += sizeof(struct crush_work_bucket);
+			break;
+		}
+		/* Every bucket has a permutation array. */
+		c->working_size += c->buckets[b]->size * sizeof(__u32);
+	}
+}
+
 static struct crush_map *crush_decode(void *pbyval, void *end)
 {
 	struct crush_map *c;
@@ -246,10 +272,6 @@ static struct crush_map *crush_decode(void *pbyval, void *end)
 		b->items = kcalloc(b->size, sizeof(__s32), GFP_NOFS);
 		if (b->items == NULL)
 			goto badmem;
-		b->perm = kcalloc(b->size, sizeof(u32), GFP_NOFS);
-		if (b->perm == NULL)
-			goto badmem;
-		b->perm_n = 0;
 
 		ceph_decode_need(p, end, b->size*sizeof(u32), bad);
 		for (j = 0; j < b->size; j++)
@@ -368,6 +390,8 @@ static struct crush_map *crush_decode(void *pbyval, void *end)
 	dout("crush decode tunable chooseleaf_stable = %d\n",
 	     c->chooseleaf_stable);
 
+	crush_finalize(c);
+
 done:
 	dout("crush_decode success\n");
 	return c;
@@ -719,7 +743,7 @@ struct ceph_osdmap *ceph_osdmap_alloc(void)
 	map->pool_max = -1;
 	map->pg_temp = RB_ROOT;
 	map->primary_temp = RB_ROOT;
-	mutex_init(&map->crush_scratch_mutex);
+	mutex_init(&map->crush_workspace_mutex);
 
 	return map;
 }
@@ -753,6 +777,7 @@ void ceph_osdmap_destroy(struct ceph_osdmap *map)
 	kfree(map->osd_weight);
 	kfree(map->osd_addr);
 	kfree(map->osd_primary_affinity);
+	kfree(map->crush_workspace);
 	kfree(map);
 }
 
@@ -808,6 +833,31 @@ static int osdmap_set_max_osd(struct ceph_osdmap *map, int max)
 	return 0;
 }
 
+static int osdmap_set_crush(struct ceph_osdmap *map, struct crush_map *crush)
+{
+	void *workspace;
+	size_t work_size;
+
+	if (IS_ERR(crush))
+		return PTR_ERR(crush);
+
+	work_size = crush_work_size(crush, CEPH_PG_MAX_SIZE);
+	dout("%s work_size %zu bytes\n", __func__, work_size);
+	workspace = kmalloc(work_size, GFP_NOIO);
+	if (!workspace) {
+		crush_destroy(crush);
+		return -ENOMEM;
+	}
+	crush_init_workspace(crush, workspace);
+
+	if (map->crush)
+		crush_destroy(map->crush);
+	kfree(map->crush_workspace);
+	map->crush = crush;
+	map->crush_workspace = workspace;
+	return 0;
+}
+
 #define OSDMAP_WRAPPER_COMPAT_VER	7
 #define OSDMAP_CLIENT_DATA_COMPAT_VER	1
 
@@ -1214,13 +1264,9 @@ static int osdmap_decode(void **p, void *end, struct ceph_osdmap *map)
 
 	/* crush */
 	ceph_decode_32_safe(p, end, len, e_inval);
-	map->crush = crush_decode(*p, min(*p + len, end));
-	if (IS_ERR(map->crush)) {
-		err = PTR_ERR(map->crush);
-		map->crush = NULL;
+	err = osdmap_set_crush(map, crush_decode(*p, min(*p + len, end)));
+	if (err)
 		goto bad;
-	}
-	*p += len;
 
 	/* ignore the rest */
 	*p = end;
@@ -1375,7 +1421,6 @@ static int decode_new_up_state_weight(void **p, void *end,
 struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end,
 					     struct ceph_osdmap *map)
 {
-	struct crush_map *newcrush = NULL;
 	struct ceph_fsid fsid;
 	u32 epoch = 0;
 	struct ceph_timespec modified;
@@ -1414,12 +1459,10 @@ struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end,
 	/* new crush? */
 	ceph_decode_32_safe(p, end, len, e_inval);
 	if (len > 0) {
-		newcrush = crush_decode(*p, min(*p+len, end));
-		if (IS_ERR(newcrush)) {
-			err = PTR_ERR(newcrush);
-			newcrush = NULL;
+		err = osdmap_set_crush(map,
+				       crush_decode(*p, min(*p + len, end)));
+		if (err)
 			goto bad;
-		}
 		*p += len;
 	}
 
@@ -1439,12 +1482,6 @@ struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end,
 
 	map->epoch++;
 	map->modified = modified;
-	if (newcrush) {
-		if (map->crush)
-			crush_destroy(map->crush);
-		map->crush = newcrush;
-		newcrush = NULL;
-	}
 
 	/* new_pools */
 	err = decode_new_pools(p, end, map);
@@ -1505,8 +1542,6 @@ struct ceph_osdmap *osdmap_apply_incremental(void **p, void *end,
 	print_hex_dump(KERN_DEBUG, "osdmap: ",
 		       DUMP_PREFIX_OFFSET, 16, 1,
 		       start, end - start, true);
-	if (newcrush)
-		crush_destroy(newcrush);
 	return ERR_PTR(err);
 }
 
@@ -1942,10 +1977,10 @@ static int do_crush(struct ceph_osdmap *map, int ruleno, int x,
 
 	BUG_ON(result_max > CEPH_PG_MAX_SIZE);
 
-	mutex_lock(&map->crush_scratch_mutex);
+	mutex_lock(&map->crush_workspace_mutex);
 	r = crush_do_rule(map->crush, ruleno, x, result, result_max,
-			  weight, weight_max, map->crush_scratch_ary);
-	mutex_unlock(&map->crush_scratch_mutex);
+			  weight, weight_max, map->crush_workspace);
+	mutex_unlock(&map->crush_workspace_mutex);
 
 	return r;
 }
@@ -1978,8 +2013,14 @@ static void pg_to_raw_osds(struct ceph_osdmap *osdmap,
 		return;
 	}
 
-	len = do_crush(osdmap, ruleno, pps, raw->osds,
-		       min_t(int, pi->size, ARRAY_SIZE(raw->osds)),
+	if (pi->size > ARRAY_SIZE(raw->osds)) {
+		pr_err_ratelimited("pool %lld ruleset %d type %d too wide: size %d > %zu\n",
+		       pi->id, pi->crush_ruleset, pi->type, pi->size,
+		       ARRAY_SIZE(raw->osds));
+		return;
+	}
+
+	len = do_crush(osdmap, ruleno, pps, raw->osds, pi->size,
 		       osdmap->osd_weight, osdmap->max_osd);
 	if (len < 0) {
 		pr_err("error %d from crush rule %d: pool %lld ruleset %d type %d size %d\n",

net/ceph/snapshot.c

@@ -18,8 +18,6 @@
  * 02110-1301, USA.
  */
 
-#include <stddef.h>
-
 #include <linux/types.h>
 #include <linux/export.h>
 #include <linux/ceph/libceph.h>