Merge tag 'ceph-for-6.1-rc1' of https://github.com/ceph/ceph-client

Pull ceph updates from Ilya Dryomov:
 "A quiet round this time: several assorted filesystem fixes, the most
  noteworthy one being some additional wakeups in cap handling code, and
  a messenger cleanup"

* tag 'ceph-for-6.1-rc1' of https://github.com/ceph/ceph-client:
  ceph: remove Sage's git tree from documentation
  ceph: fix incorrectly showing the .snap size for stat
  ceph: fail the open_by_handle_at() if the dentry is being unlinked
  ceph: increment i_version when doing a setattr with caps
  ceph: Use kcalloc for allocating multiple elements
  ceph: no need to wait for transition RDCACHE|RD -> RD
  ceph: fail the request if the peer MDS doesn't support getvxattr op
  ceph: wake up the waiters if any new caps comes
  libceph: drop last_piece flag from ceph_msg_data_cursor

5 files changed, 54 insertions, 9 deletions

diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index 53cfe026b3ea..fb023f9fafcb 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -754,6 +754,7 @@ void ceph_add_cap(struct inode *inode,
 	cap->issue_seq = seq;
 	cap->mseq = mseq;
 	cap->cap_gen = gen;
+	wake_up_all(&ci->i_cap_wq);
 }
 
 /*
@@ -2285,7 +2286,7 @@ retry:
 		struct ceph_mds_request *req;
 		int i;
 
-		sessions = kzalloc(max_sessions * sizeof(s), GFP_KERNEL);
+		sessions = kcalloc(max_sessions, sizeof(s), GFP_KERNEL);
 		if (!sessions) {
 			err = -ENOMEM;
 			goto out;
@@ -2759,13 +2760,17 @@ again:
 		 * on transition from wanted -> needed caps.  This is needed
 		 * for WRBUFFER|WR -> WR to avoid a new WR sync write from
 		 * going before a prior buffered writeback happens.
+		 *
+		 * For RDCACHE|RD -> RD, there is not need to wait and we can
+		 * just exclude the revoking caps and force to sync read.
 		 */
 		int not = want & ~(have & need);
 		int revoking = implemented & ~have;
+		int exclude = revoking & not;
 		dout("get_cap_refs %p have %s but not %s (revoking %s)\n",
 		     inode, ceph_cap_string(have), ceph_cap_string(not),
 		     ceph_cap_string(revoking));
-		if ((revoking & not) == 0) {
+		if (!exclude || !(exclude & CEPH_CAP_FILE_BUFFER)) {
 			if (!snap_rwsem_locked &&
 			    !ci->i_head_snapc &&
 			    (need & CEPH_CAP_FILE_WR)) {
@@ -2787,7 +2792,7 @@ again:
 				snap_rwsem_locked = true;
 			}
 			if ((have & want) == want)
-				*got = need | want;
+				*got = need | (want & ~exclude);
 			else
 				*got = need;
 			ceph_take_cap_refs(ci, *got, true);
@@ -3550,6 +3555,9 @@ static void handle_cap_grant(struct inode *inode,
 			check_caps = 1; /* check auth cap only */
 		else
 			check_caps = 2; /* check all caps */
+		/* If there is new caps, try to wake up the waiters */
+		if (~cap->issued & newcaps)
+			wake = true;
 		cap->issued = newcaps;
 		cap->implemented |= newcaps;
 	} else if (cap->issued == newcaps) {
diff --git a/fs/ceph/export.c b/fs/ceph/export.c
index e0fa66ac8b9f..f780e4e0d062 100644
--- a/fs/ceph/export.c
+++ b/fs/ceph/export.c
@@ -181,6 +181,7 @@ struct inode *ceph_lookup_inode(struct super_block *sb, u64 ino)
 static struct dentry *__fh_to_dentry(struct super_block *sb, u64 ino)
 {
 	struct inode *inode = __lookup_inode(sb, ino);
+	struct ceph_inode_info *ci = ceph_inode(inode);
 	int err;
 
 	if (IS_ERR(inode))
@@ -192,7 +193,7 @@ static struct dentry *__fh_to_dentry(struct super_block *sb, u64 ino)
 		return ERR_PTR(err);
 	}
 	/* -ESTALE if inode as been unlinked and no file is open */
-	if ((inode->i_nlink == 0) && (atomic_read(&inode->i_count) == 1)) {
+	if ((inode->i_nlink == 0) && !__ceph_is_file_opened(ci)) {
 		iput(inode);
 		return ERR_PTR(-ESTALE);
 	}
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index 42351d7a0dd6..9ebb7cee7978 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -2192,6 +2192,7 @@ int __ceph_setattr(struct inode *inode, struct iattr *attr)
 		inode_dirty_flags = __ceph_mark_dirty_caps(ci, dirtied,
 							   &prealloc_cf);
 		inode->i_ctime = attr->ia_ctime;
+		inode_inc_iversion_raw(inode);
 	}
 
 	release &= issued;
@@ -2356,6 +2357,7 @@ int ceph_do_getvxattr(struct inode *inode, const char *name, void *value,
 		goto out;
 	}
 
+	req->r_feature_needed = CEPHFS_FEATURE_OP_GETVXATTR;
 	req->r_path2 = kstrdup(name, GFP_NOFS);
 	if (!req->r_path2) {
 		err = -ENOMEM;
@@ -2447,6 +2449,7 @@ int ceph_getattr(struct user_namespace *mnt_userns, const struct path *path,
 		 struct kstat *stat, u32 request_mask, unsigned int flags)
 {
 	struct inode *inode = d_inode(path->dentry);
+	struct super_block *sb = inode->i_sb;
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	u32 valid_mask = STATX_BASIC_STATS;
 	int err = 0;
@@ -2476,16 +2479,34 @@ int ceph_getattr(struct user_namespace *mnt_userns, const struct path *path,
 	}
 
 	if (ceph_snap(inode) == CEPH_NOSNAP)
-		stat->dev = inode->i_sb->s_dev;
+		stat->dev = sb->s_dev;
 	else
 		stat->dev = ci->i_snapid_map ? ci->i_snapid_map->dev : 0;
 
 	if (S_ISDIR(inode->i_mode)) {
-		if (ceph_test_mount_opt(ceph_sb_to_client(inode->i_sb),
-					RBYTES))
+		if (ceph_test_mount_opt(ceph_sb_to_client(sb), RBYTES)) {
 			stat->size = ci->i_rbytes;
-		else
+		} else if (ceph_snap(inode) == CEPH_SNAPDIR) {
+			struct ceph_inode_info *pci;
+			struct ceph_snap_realm *realm;
+			struct inode *parent;
+
+			parent = ceph_lookup_inode(sb, ceph_ino(inode));
+			if (!parent)
+				return PTR_ERR(parent);
+
+			pci = ceph_inode(parent);
+			spin_lock(&pci->i_ceph_lock);
+			realm = pci->i_snap_realm;
+			if (realm)
+				stat->size = realm->num_snaps;
+			else
+				stat->size = 0;
+			spin_unlock(&pci->i_ceph_lock);
+			iput(parent);
+		} else {
 			stat->size = ci->i_files + ci->i_subdirs;
+		}
 		stat->blocks = 0;
 		stat->blksize = 65536;
 		/*
diff --git a/fs/ceph/mds_client.c b/fs/ceph/mds_client.c
index 80f8b9ec1a31..26a0a8b9975e 100644
--- a/fs/ceph/mds_client.c
+++ b/fs/ceph/mds_client.c
@@ -2318,6 +2318,7 @@ ceph_mdsc_create_request(struct ceph_mds_client *mdsc, int op, int mode)
 	INIT_LIST_HEAD(&req->r_unsafe_dir_item);
 	INIT_LIST_HEAD(&req->r_unsafe_target_item);
 	req->r_fmode = -1;
+	req->r_feature_needed = -1;
 	kref_init(&req->r_kref);
 	RB_CLEAR_NODE(&req->r_node);
 	INIT_LIST_HEAD(&req->r_wait);
@@ -2916,6 +2917,16 @@ static void __do_request(struct ceph_mds_client *mdsc,
 
 	dout("do_request mds%d session %p state %s\n", mds, session,
 	     ceph_session_state_name(session->s_state));
+
+	/*
+	 * The old ceph will crash the MDSs when see unknown OPs
+	 */
+	if (req->r_feature_needed > 0 &&
+	    !test_bit(req->r_feature_needed, &session->s_features)) {
+		err = -EOPNOTSUPP;
+		goto out_session;
+	}
+
 	if (session->s_state != CEPH_MDS_SESSION_OPEN &&
 	    session->s_state != CEPH_MDS_SESSION_HUNG) {
 		/*
diff --git a/fs/ceph/mds_client.h b/fs/ceph/mds_client.h
index 256e3eada6c1..0598faa50e2e 100644
--- a/fs/ceph/mds_client.h
+++ b/fs/ceph/mds_client.h
@@ -31,8 +31,9 @@ enum ceph_feature_type {
 	CEPHFS_FEATURE_METRIC_COLLECT,
 	CEPHFS_FEATURE_ALTERNATE_NAME,
 	CEPHFS_FEATURE_NOTIFY_SESSION_STATE,
+	CEPHFS_FEATURE_OP_GETVXATTR,
 
-	CEPHFS_FEATURE_MAX = CEPHFS_FEATURE_NOTIFY_SESSION_STATE,
+	CEPHFS_FEATURE_MAX = CEPHFS_FEATURE_OP_GETVXATTR,
 };
 
 #define CEPHFS_FEATURES_CLIENT_SUPPORTED {	\
@@ -44,6 +45,7 @@ enum ceph_feature_type {
 	CEPHFS_FEATURE_DELEG_INO,		\
 	CEPHFS_FEATURE_METRIC_COLLECT,		\
 	CEPHFS_FEATURE_NOTIFY_SESSION_STATE,	\
+	CEPHFS_FEATURE_OP_GETVXATTR,		\
 }
 
 /*
@@ -336,6 +338,8 @@ struct ceph_mds_request {
 	long long	  r_dir_ordered_cnt;
 	int		  r_readdir_cache_idx;
 
+	int		  r_feature_needed;
+
 	struct ceph_cap_reservation r_caps_reservation;
 };