Merge tag 'ceph-for-6.2-rc1' of https://github.com/ceph/ceph-client

Pull cph update from Ilya Dryomov:
 "A fix to facilitate prompt cap releases on async creates from Xiubo.

  This should address sporadic "client isn't responding to mclientcaps
  (revoke) ..." warnings and potential associated MDS hangs"

* tag 'ceph-for-6.2-rc1' of https://github.com/ceph/ceph-client:
  ceph: try to check caps immediately after async creating finishes
  ceph: remove useless session parameter for check_caps()

6 files changed, 35 insertions, 31 deletions

diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index 61f47debec5a..8c74871e37c9 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -1367,7 +1367,7 @@ out:
 	folio_put(folio);
 
 	if (check_cap)
-		ceph_check_caps(ceph_inode(inode), CHECK_CAPS_AUTHONLY, NULL);
+		ceph_check_caps(ceph_inode(inode), CHECK_CAPS_AUTHONLY);
 
 	return copied;
 }
diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index e54814d0c2f7..4b159f97fe7b 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -1898,8 +1898,7 @@ bool __ceph_should_report_size(struct ceph_inode_info *ci)
  *  CHECK_CAPS_FLUSH - we should flush any dirty caps immediately, without
  *    further delay.
  */
-void ceph_check_caps(struct ceph_inode_info *ci, int flags,
-		     struct ceph_mds_session *session)
+void ceph_check_caps(struct ceph_inode_info *ci, int flags)
 {
 	struct inode *inode = &ci->netfs.inode;
 	struct ceph_mds_client *mdsc = ceph_sb_to_mdsc(inode->i_sb);
@@ -1913,15 +1912,14 @@ void ceph_check_caps(struct ceph_inode_info *ci, int flags,
 	bool queue_invalidate = false;
 	bool tried_invalidate = false;
 	bool queue_writeback = false;
-
-	if (session)
-		ceph_get_mds_session(session);
+	struct ceph_mds_session *session = NULL;
 
 	spin_lock(&ci->i_ceph_lock);
 	if (ci->i_ceph_flags & CEPH_I_ASYNC_CREATE) {
+		ci->i_ceph_flags |= CEPH_I_ASYNC_CHECK_CAPS;
+
 		/* Don't send messages until we get async create reply */
 		spin_unlock(&ci->i_ceph_lock);
-		ceph_put_mds_session(session);
 		return;
 	}
 
@@ -2851,7 +2849,7 @@ static void check_max_size(struct inode *inode, loff_t endoff)
 		check = 1;
 	spin_unlock(&ci->i_ceph_lock);
 	if (check)
-		ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL);
+		ceph_check_caps(ci, CHECK_CAPS_AUTHONLY);
 }
 
 static inline int get_used_fmode(int caps)
@@ -3140,7 +3138,7 @@ static void __ceph_put_cap_refs(struct ceph_inode_info *ci, int had,
 	switch (mode) {
 	case PUT_CAP_REFS_SYNC:
 		if (last)
-			ceph_check_caps(ci, 0, NULL);
+			ceph_check_caps(ci, 0);
 		else if (flushsnaps)
 			ceph_flush_snaps(ci, NULL);
 		break;
@@ -3255,7 +3253,7 @@ unlock:
 	spin_unlock(&ci->i_ceph_lock);
 
 	if (last) {
-		ceph_check_caps(ci, 0, NULL);
+		ceph_check_caps(ci, 0);
 	} else if (flush_snaps) {
 		ceph_flush_snaps(ci, NULL);
 	}
@@ -3604,10 +3602,9 @@ static void handle_cap_grant(struct inode *inode,
 
 	mutex_unlock(&session->s_mutex);
 	if (check_caps == 1)
-		ceph_check_caps(ci, CHECK_CAPS_AUTHONLY | CHECK_CAPS_NOINVAL,
-				session);
+		ceph_check_caps(ci, CHECK_CAPS_AUTHONLY | CHECK_CAPS_NOINVAL);
 	else if (check_caps == 2)
-		ceph_check_caps(ci, CHECK_CAPS_NOINVAL, session);
+		ceph_check_caps(ci, CHECK_CAPS_NOINVAL);
 }
 
 /*
@@ -4333,7 +4330,7 @@ unsigned long ceph_check_delayed_caps(struct ceph_mds_client *mdsc)
 		if (inode) {
 			spin_unlock(&mdsc->cap_delay_lock);
 			dout("check_delayed_caps on %p\n", inode);
-			ceph_check_caps(ci, 0, NULL);
+			ceph_check_caps(ci, 0);
 			iput(inode);
 			spin_lock(&mdsc->cap_delay_lock);
 		}
@@ -4362,7 +4359,7 @@ static void flush_dirty_session_caps(struct ceph_mds_session *s)
 		dout("flush_dirty_caps %llx.%llx\n", ceph_vinop(inode));
 		spin_unlock(&mdsc->cap_dirty_lock);
 		ceph_wait_on_async_create(inode);
-		ceph_check_caps(ci, CHECK_CAPS_FLUSH, NULL);
+		ceph_check_caps(ci, CHECK_CAPS_FLUSH);
 		iput(inode);
 		spin_lock(&mdsc->cap_dirty_lock);
 	}
diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index 6f9580defb2b..764598e1efd9 100644
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -313,7 +313,7 @@ int ceph_renew_caps(struct inode *inode, int fmode)
 		spin_unlock(&ci->i_ceph_lock);
 		dout("renew caps %p want %s issued %s updating mds_wanted\n",
 		     inode, ceph_cap_string(wanted), ceph_cap_string(issued));
-		ceph_check_caps(ci, 0, NULL);
+		ceph_check_caps(ci, 0);
 		return 0;
 	}
 	spin_unlock(&ci->i_ceph_lock);
@@ -408,7 +408,7 @@ int ceph_open(struct inode *inode, struct file *file)
 		if ((issued & wanted) != wanted &&
 		    (mds_wanted & wanted) != wanted &&
 		    ceph_snap(inode) != CEPH_SNAPDIR)
-			ceph_check_caps(ci, 0, NULL);
+			ceph_check_caps(ci, 0);
 
 		return ceph_init_file(inode, file, fmode);
 	} else if (ceph_snap(inode) != CEPH_NOSNAP &&
@@ -534,14 +534,23 @@ static void wake_async_create_waiters(struct inode *inode,
 				      struct ceph_mds_session *session)
 {
 	struct ceph_inode_info *ci = ceph_inode(inode);
+	bool check_cap = false;
 
 	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);
+
+		if (ci->i_ceph_flags & CEPH_I_ASYNC_CHECK_CAPS) {
+			ci->i_ceph_flags &= ~CEPH_I_ASYNC_CHECK_CAPS;
+			check_cap = true;
+		}
 	}
 	ceph_kick_flushing_inode_caps(session, ci);
 	spin_unlock(&ci->i_ceph_lock);
+
+	if (check_cap)
+		ceph_check_caps(ci, CHECK_CAPS_FLUSH);
 }
 
 static void ceph_async_create_cb(struct ceph_mds_client *mdsc,
@@ -1092,7 +1101,7 @@ static void ceph_aio_complete(struct inode *inode,
 		loff_t endoff = aio_req->iocb->ki_pos + aio_req->total_len;
 		if (endoff > i_size_read(inode)) {
 			if (ceph_inode_set_size(inode, endoff))
-				ceph_check_caps(ci, CHECK_CAPS_AUTHONLY, NULL);
+				ceph_check_caps(ci, CHECK_CAPS_AUTHONLY);
 		}
 
 		spin_lock(&ci->i_ceph_lock);
@@ -1421,8 +1430,7 @@ ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter,
 		if (write && pos > size) {
 			if (ceph_inode_set_size(inode, pos))
 				ceph_check_caps(ceph_inode(inode),
-						CHECK_CAPS_AUTHONLY,
-						NULL);
+						CHECK_CAPS_AUTHONLY);
 		}
 	}
 
@@ -1577,8 +1585,7 @@ out:
 			check_caps = ceph_inode_set_size(inode, pos);
 			if (check_caps)
 				ceph_check_caps(ceph_inode(inode),
-						CHECK_CAPS_AUTHONLY,
-						NULL);
+						CHECK_CAPS_AUTHONLY);
 		}
 
 	}
@@ -1906,7 +1913,7 @@ retry_snap:
 		if (dirty)
 			__mark_inode_dirty(inode, dirty);
 		if (ceph_quota_is_max_bytes_approaching(inode, iocb->ki_pos))
-			ceph_check_caps(ci, CHECK_CAPS_FLUSH, NULL);
+			ceph_check_caps(ci, CHECK_CAPS_FLUSH);
 	}
 
 	dout("aio_write %p %llx.%llx %llu~%u  dropping cap refs on %s\n",
@@ -2521,8 +2528,7 @@ static ssize_t __ceph_copy_file_range(struct file *src_file, loff_t src_off,
 		/* Let the MDS know about dst file size change */
 		if (ceph_inode_set_size(dst_inode, dst_off) ||
 		    ceph_quota_is_max_bytes_approaching(dst_inode, dst_off))
-			ceph_check_caps(dst_ci, CHECK_CAPS_AUTHONLY | CHECK_CAPS_FLUSH,
-					NULL);
+			ceph_check_caps(dst_ci, CHECK_CAPS_AUTHONLY | CHECK_CAPS_FLUSH);
 	}
 	/* Mark Fw dirty */
 	spin_lock(&dst_ci->i_ceph_lock);
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index f23c5a6edc6f..23d05ec87fcc 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -1909,7 +1909,7 @@ static void ceph_do_invalidate_pages(struct inode *inode)
 	mutex_unlock(&ci->i_truncate_mutex);
 out:
 	if (check)
-		ceph_check_caps(ci, 0, NULL);
+		ceph_check_caps(ci, 0);
 }
 
 /*
@@ -1969,7 +1969,7 @@ retry:
 	mutex_unlock(&ci->i_truncate_mutex);
 
 	if (wrbuffer_refs == 0)
-		ceph_check_caps(ci, 0, NULL);
+		ceph_check_caps(ci, 0);
 
 	wake_up_all(&ci->i_cap_wq);
 }
@@ -1991,7 +1991,7 @@ static void ceph_inode_work(struct work_struct *work)
 		__ceph_do_pending_vmtruncate(inode);
 
 	if (test_and_clear_bit(CEPH_I_WORK_CHECK_CAPS, &ci->i_work_mask))
-		ceph_check_caps(ci, 0, NULL);
+		ceph_check_caps(ci, 0);
 
 	if (test_and_clear_bit(CEPH_I_WORK_FLUSH_SNAPS, &ci->i_work_mask))
 		ceph_flush_snaps(ci, NULL);
diff --git a/fs/ceph/ioctl.c b/fs/ceph/ioctl.c
index 6e061bf62ad4..deac817647eb 100644
--- a/fs/ceph/ioctl.c
+++ b/fs/ceph/ioctl.c
@@ -253,7 +253,7 @@ static long ceph_ioctl_lazyio(struct file *file)
 		spin_unlock(&ci->i_ceph_lock);
 		dout("ioctl_layzio: file %p marked lazy\n", file);
 
-		ceph_check_caps(ci, 0, NULL);
+		ceph_check_caps(ci, 0);
 	} else {
 		dout("ioctl_layzio: file %p already lazy\n", file);
 	}
diff --git a/fs/ceph/super.h b/fs/ceph/super.h
index 50e57a1fa32f..30bdb391a0dc 100644
--- a/fs/ceph/super.h
+++ b/fs/ceph/super.h
@@ -593,6 +593,8 @@ static inline struct inode *ceph_find_inode(struct super_block *sb,
 #define CEPH_ASYNC_CREATE_BIT	(12)	  /* async create in flight for this */
 #define CEPH_I_ASYNC_CREATE	(1 << CEPH_ASYNC_CREATE_BIT)
 #define CEPH_I_SHUTDOWN		(1 << 13) /* inode is no longer usable */
+#define CEPH_I_ASYNC_CHECK_CAPS	(1 << 14) /* check caps immediately after async
+					     creating finishes */
 
 /*
  * Masks of ceph inode work.
@@ -1200,8 +1202,7 @@ extern void ceph_remove_capsnap(struct inode *inode,
 extern void ceph_flush_snaps(struct ceph_inode_info *ci,
 			     struct ceph_mds_session **psession);
 extern bool __ceph_should_report_size(struct ceph_inode_info *ci);
-extern void ceph_check_caps(struct ceph_inode_info *ci, int flags,
-			    struct ceph_mds_session *session);
+extern void ceph_check_caps(struct ceph_inode_info *ci, int flags);
 extern unsigned long ceph_check_delayed_caps(struct ceph_mds_client *mdsc);
 extern void ceph_flush_dirty_caps(struct ceph_mds_client *mdsc);
 extern int  ceph_drop_caps_for_unlink(struct inode *inode);