Merge git://git.linux-nfs.org/pub/linux/nfs-2.6

* git://git.linux-nfs.org/pub/linux/nfs-2.6:
  NLM,NFSv4: Wait on local locks before we put RPC calls on the wire
  VFS: Add support for the FL_ACCESS flag to flock_lock_file()
  NFSv4: Ensure nfs4_lock_expired() caches delegated locks
  NLM,NFSv4: Don't put UNLOCK requests on the wire unless we hold a lock
  VFS: Allow caller to determine if BSD or posix locks were actually freed
  NFS: Optimise away an excessive GETATTR call when a file is symlinked
  This fixes a panic doing the first READDIR or READDIRPLUS call when:
  NFS: Fix NFS page_state usage
  Revert "Merge branch 'odirect'"

fs/lockd/clntproc.c

@@ -454,7 +454,7 @@ static void nlmclnt_locks_init_private(struct file_lock *fl, struct nlm_host *ho
 	fl->fl_ops = &nlmclnt_lock_ops;
 }
 
-static void do_vfs_lock(struct file_lock *fl)
+static int do_vfs_lock(struct file_lock *fl)
 {
 	int res = 0;
 	switch (fl->fl_flags & (FL_POSIX|FL_FLOCK)) {
@@ -467,9 +467,7 @@ static void do_vfs_lock(struct file_lock *fl)
 		default:
 			BUG();
 	}
-	if (res < 0)
-		printk(KERN_WARNING "%s: VFS is out of sync with lock manager!\n",
-				__FUNCTION__);
+	return res;
 }
 
 /*
@@ -498,6 +496,7 @@ nlmclnt_lock(struct nlm_rqst *req, struct file_lock *fl)
 	struct nlm_host	*host = req->a_host;
 	struct nlm_res	*resp = &req->a_res;
 	struct nlm_wait *block = NULL;
+	unsigned char fl_flags = fl->fl_flags;
 	int status = -ENOLCK;
 
 	if (!host->h_monitored && nsm_monitor(host) < 0) {
@@ -505,6 +504,10 @@ nlmclnt_lock(struct nlm_rqst *req, struct file_lock *fl)
 					host->h_name);
 		goto out;
 	}
+	fl->fl_flags |= FL_ACCESS;
+	status = do_vfs_lock(fl);
+	if (status < 0)
+		goto out;
 
 	block = nlmclnt_prepare_block(host, fl);
 again:
@@ -539,9 +542,10 @@ again:
 			up_read(&host->h_rwsem);
 			goto again;
 		}
-		fl->fl_flags |= FL_SLEEP;
 		/* Ensure the resulting lock will get added to granted list */
-		do_vfs_lock(fl);
+		fl->fl_flags = fl_flags | FL_SLEEP;
+		if (do_vfs_lock(fl) < 0)
+			printk(KERN_WARNING "%s: VFS is out of sync with lock manager!\n", __FUNCTION__);
 		up_read(&host->h_rwsem);
 	}
 	status = nlm_stat_to_errno(resp->status);
@@ -552,6 +556,7 @@ out_unblock:
 		nlmclnt_cancel(host, req->a_args.block, fl);
 out:
 	nlm_release_call(req);
+	fl->fl_flags = fl_flags;
 	return status;
 }
 
@@ -606,15 +611,19 @@ nlmclnt_unlock(struct nlm_rqst *req, struct file_lock *fl)
 {
 	struct nlm_host	*host = req->a_host;
 	struct nlm_res	*resp = &req->a_res;
-	int		status;
+	int status = 0;
 
 	/*
 	 * Note: the server is supposed to either grant us the unlock
 	 * request, or to deny it with NLM_LCK_DENIED_GRACE_PERIOD. In either
 	 * case, we want to unlock.
 	 */
+	fl->fl_flags |= FL_EXISTS;
 	down_read(&host->h_rwsem);
-	do_vfs_lock(fl);
+	if (do_vfs_lock(fl) == -ENOENT) {
+		up_read(&host->h_rwsem);
+		goto out;
+	}
 	up_read(&host->h_rwsem);
 
 	if (req->a_flags & RPC_TASK_ASYNC)
@@ -624,7 +633,6 @@ nlmclnt_unlock(struct nlm_rqst *req, struct file_lock *fl)
 	if (status < 0)
 		goto out;
 
-	status = 0;
 	if (resp->status == NLM_LCK_GRANTED)
 		goto out;
 

fs/locks.c

@@ -725,6 +725,10 @@ next_task:
 /* Try to create a FLOCK lock on filp. We always insert new FLOCK locks
  * at the head of the list, but that's secret knowledge known only to
  * flock_lock_file and posix_lock_file.
+ *
+ * Note that if called with an FL_EXISTS argument, the caller may determine
+ * whether or not a lock was successfully freed by testing the return
+ * value for -ENOENT.
  */
 static int flock_lock_file(struct file *filp, struct file_lock *request)
 {
@@ -735,6 +739,8 @@ static int flock_lock_file(struct file *filp, struct file_lock *request)
 	int found = 0;
 
 	lock_kernel();
+	if (request->fl_flags & FL_ACCESS)
+		goto find_conflict;
 	for_each_lock(inode, before) {
 		struct file_lock *fl = *before;
 		if (IS_POSIX(fl))
@@ -750,8 +756,11 @@ static int flock_lock_file(struct file *filp, struct file_lock *request)
 		break;
 	}
 
-	if (request->fl_type == F_UNLCK)
+	if (request->fl_type == F_UNLCK) {
+		if ((request->fl_flags & FL_EXISTS) && !found)
+			error = -ENOENT;
 		goto out;
+	}
 
 	error = -ENOMEM;
 	new_fl = locks_alloc_lock();
@@ -764,6 +773,7 @@ static int flock_lock_file(struct file *filp, struct file_lock *request)
 	if (found)
 		cond_resched();
 
+find_conflict:
 	for_each_lock(inode, before) {
 		struct file_lock *fl = *before;
 		if (IS_POSIX(fl))
@@ -777,6 +787,8 @@ static int flock_lock_file(struct file *filp, struct file_lock *request)
 			locks_insert_block(fl, request);
 		goto out;
 	}
+	if (request->fl_flags & FL_ACCESS)
+		goto out;
 	locks_copy_lock(new_fl, request);
 	locks_insert_lock(&inode->i_flock, new_fl);
 	new_fl = NULL;
@@ -948,8 +960,11 @@ static int __posix_lock_file_conf(struct inode *inode, struct file_lock *request
 
 	error = 0;
 	if (!added) {
-		if (request->fl_type == F_UNLCK)
+		if (request->fl_type == F_UNLCK) {
+			if (request->fl_flags & FL_EXISTS)
+				error = -ENOENT;
 			goto out;
+		}
 
 		if (!new_fl) {
 			error = -ENOLCK;
@@ -996,6 +1011,10 @@ static int __posix_lock_file_conf(struct inode *inode, struct file_lock *request
  * Add a POSIX style lock to a file.
  * We merge adjacent & overlapping locks whenever possible.
  * POSIX locks are sorted by owner task, then by starting address
+ *
+ * Note that if called with an FL_EXISTS argument, the caller may determine
+ * whether or not a lock was successfully freed by testing the return
+ * value for -ENOENT.
  */
 int posix_lock_file(struct file *filp, struct file_lock *fl)
 {

fs/nfs/dir.c

@@ -690,7 +690,9 @@ int nfs_lookup_verify_inode(struct inode *inode, struct nameidata *nd)
 			goto out_force;
 		/* This is an open(2) */
 		if (nfs_lookup_check_intent(nd, LOOKUP_OPEN) != 0 &&
-				!(server->flags & NFS_MOUNT_NOCTO))
+				!(server->flags & NFS_MOUNT_NOCTO) &&
+				(S_ISREG(inode->i_mode) ||
+				 S_ISDIR(inode->i_mode)))
 			goto out_force;
 	}
 	return nfs_revalidate_inode(server, inode);

fs/nfs/nfs4proc.c

@@ -3144,9 +3144,6 @@ static int do_vfs_lock(struct file *file, struct file_lock *fl)
 		default:
 			BUG();
 	}
-	if (res < 0)
-		printk(KERN_WARNING "%s: VFS is out of sync with lock manager!\n",
-				__FUNCTION__);
 	return res;
 }
 
@@ -3258,8 +3255,6 @@ static struct rpc_task *nfs4_do_unlck(struct file_lock *fl,
 		return ERR_PTR(-ENOMEM);
 	}
 
-	/* Unlock _before_ we do the RPC call */
-	do_vfs_lock(fl->fl_file, fl);
 	return rpc_run_task(NFS_CLIENT(lsp->ls_state->inode), RPC_TASK_ASYNC, &nfs4_locku_ops, data);
 }
 
@@ -3270,30 +3265,28 @@ static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *
 	struct rpc_task *task;
 	int status = 0;
 
-	/* Is this a delegated lock? */
-	if (test_bit(NFS_DELEGATED_STATE, &state->flags))
-		goto out_unlock;
-	/* Is this open_owner holding any locks on the server? */
-	if (test_bit(LK_STATE_IN_USE, &state->flags) == 0)
-		goto out_unlock;
-
 	status = nfs4_set_lock_state(state, request);
+	/* Unlock _before_ we do the RPC call */
+	request->fl_flags |= FL_EXISTS;
+	if (do_vfs_lock(request->fl_file, request) == -ENOENT)
+		goto out;
 	if (status != 0)
-		goto out_unlock;
+		goto out;
+	/* Is this a delegated lock? */
+	if (test_bit(NFS_DELEGATED_STATE, &state->flags))
+		goto out;
 	lsp = request->fl_u.nfs4_fl.owner;
-	status = -ENOMEM;
 	seqid = nfs_alloc_seqid(&lsp->ls_seqid);
+	status = -ENOMEM;
 	if (seqid == NULL)
-		goto out_unlock;
+		goto out;
 	task = nfs4_do_unlck(request, request->fl_file->private_data, lsp, seqid);
 	status = PTR_ERR(task);
 	if (IS_ERR(task))
-		goto out_unlock;
+		goto out;
 	status = nfs4_wait_for_completion_rpc_task(task);
 	rpc_release_task(task);
-	return status;
-out_unlock:
-	do_vfs_lock(request->fl_file, request);
+out:
 	return status;
 }
 
@@ -3461,10 +3454,10 @@ static int nfs4_lock_reclaim(struct nfs4_state *state, struct file_lock *request
 	struct nfs4_exception exception = { };
 	int err;
 
-	/* Cache the lock if possible... */
-	if (test_bit(NFS_DELEGATED_STATE, &state->flags))
-		return 0;
 	do {
+		/* Cache the lock if possible... */
+		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
+			return 0;
 		err = _nfs4_do_setlk(state, F_SETLK, request, 1);
 		if (err != -NFS4ERR_DELAY)
 			break;
@@ -3483,6 +3476,8 @@ static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request
 	if (err != 0)
 		return err;
 	do {
+		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
+			return 0;
 		err = _nfs4_do_setlk(state, F_SETLK, request, 0);
 		if (err != -NFS4ERR_DELAY)
 			break;
@@ -3494,29 +3489,42 @@ static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request
 static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
 {
 	struct nfs4_client *clp = state->owner->so_client;
+	unsigned char fl_flags = request->fl_flags;
 	int status;
 
 	/* Is this a delegated open? */
-	if (NFS_I(state->inode)->delegation_state != 0) {
-		/* Yes: cache locks! */
-		status = do_vfs_lock(request->fl_file, request);
-		/* ...but avoid races with delegation recall... */
-		if (status < 0 || test_bit(NFS_DELEGATED_STATE, &state->flags))
-			return status;
-	}
-	down_read(&clp->cl_sem);
 	status = nfs4_set_lock_state(state, request);
 	if (status != 0)
 		goto out;
+	request->fl_flags |= FL_ACCESS;
+	status = do_vfs_lock(request->fl_file, request);
+	if (status < 0)
+		goto out;
+	down_read(&clp->cl_sem);
+	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
+		struct nfs_inode *nfsi = NFS_I(state->inode);
+		/* Yes: cache locks! */
+		down_read(&nfsi->rwsem);
+		/* ...but avoid races with delegation recall... */
+		if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
+			request->fl_flags = fl_flags & ~FL_SLEEP;
+			status = do_vfs_lock(request->fl_file, request);
+			up_read(&nfsi->rwsem);
+			goto out_unlock;
+		}
+		up_read(&nfsi->rwsem);
+	}
 	status = _nfs4_do_setlk(state, cmd, request, 0);
 	if (status != 0)
-		goto out;
+		goto out_unlock;
 	/* Note: we always want to sleep here! */
-	request->fl_flags |= FL_SLEEP;
+	request->fl_flags = fl_flags | FL_SLEEP;
 	if (do_vfs_lock(request->fl_file, request) < 0)
 		printk(KERN_WARNING "%s: VFS is out of sync with lock manager!\n", __FUNCTION__);
-out:
+out_unlock:
 	up_read(&clp->cl_sem);
+out:
+	request->fl_flags = fl_flags;
 	return status;
 }
 

fs/nfs/write.c

@@ -578,7 +578,7 @@ static int nfs_wait_on_requests(struct inode *inode, unsigned long idx_start, un
 	return ret;
 }
 
-static void nfs_cancel_requests(struct list_head *head)
+static void nfs_cancel_dirty_list(struct list_head *head)
 {
 	struct nfs_page *req;
 	while(!list_empty(head)) {
@@ -589,6 +589,19 @@ static void nfs_cancel_requests(struct list_head *head)
 	}
 }
 
+static void nfs_cancel_commit_list(struct list_head *head)
+{
+	struct nfs_page *req;
+
+	while(!list_empty(head)) {
+		req = nfs_list_entry(head->next);
+		nfs_list_remove_request(req);
+		nfs_inode_remove_request(req);
+		nfs_clear_page_writeback(req);
+		dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
+	}
+}
+
 /*
  * nfs_scan_dirty - Scan an inode for dirty requests
  * @inode: NFS inode to scan
@@ -1381,6 +1394,7 @@ nfs_commit_list(struct inode *inode, struct list_head *head, int how)
 		nfs_list_remove_request(req);
 		nfs_mark_request_commit(req);
 		nfs_clear_page_writeback(req);
+		dec_zone_page_state(req->wb_page, NR_UNSTABLE_NFS);
 	}
 	return -ENOMEM;
 }
@@ -1499,7 +1513,7 @@ int nfs_sync_inode_wait(struct inode *inode, unsigned long idx_start,
 		if (pages != 0) {
 			spin_unlock(&nfsi->req_lock);
 			if (how & FLUSH_INVALIDATE)
-				nfs_cancel_requests(&head);
+				nfs_cancel_dirty_list(&head);
 			else
 				ret = nfs_flush_list(inode, &head, pages, how);
 			spin_lock(&nfsi->req_lock);
@@ -1512,7 +1526,7 @@ int nfs_sync_inode_wait(struct inode *inode, unsigned long idx_start,
 			break;
 		if (how & FLUSH_INVALIDATE) {
 			spin_unlock(&nfsi->req_lock);
-			nfs_cancel_requests(&head);
+			nfs_cancel_commit_list(&head);
 			spin_lock(&nfsi->req_lock);
 			continue;
 		}

include/linux/fs.h

@@ -716,6 +716,7 @@ extern spinlock_t files_lock;
 #define FL_POSIX	1
 #define FL_FLOCK	2
 #define FL_ACCESS	8	/* not trying to lock, just looking */
+#define FL_EXISTS	16	/* when unlocking, test for existence */
 #define FL_LEASE	32	/* lease held on this file */
 #define FL_CLOSE	64	/* unlock on close */
 #define FL_SLEEP	128	/* A blocking lock */

include/linux/nfs_xdr.h

@@ -729,6 +729,7 @@ struct nfs_read_data {
 	struct list_head	pages;	/* Coalesced read requests */
 	struct nfs_page		*req;	/* multi ops per nfs_page */
 	struct page		**pagevec;
+	unsigned int		npages;	/* active pages in pagevec */
 	struct nfs_readargs args;
 	struct nfs_readres  res;
 #ifdef CONFIG_NFS_V4
@@ -747,6 +748,7 @@ struct nfs_write_data {
 	struct list_head	pages;		/* Coalesced requests we wish to flush */
 	struct nfs_page		*req;		/* multi ops per nfs_page */
 	struct page		**pagevec;
+	unsigned int		npages;		/* active pages in pagevec */
 	struct nfs_writeargs	args;		/* argument struct */
 	struct nfs_writeres	res;		/* result struct */
 #ifdef CONFIG_NFS_V4

net/sunrpc/xdr.c

@@ -191,7 +191,6 @@ _shift_data_right_pages(struct page **pages, size_t pgto_base,
 	do {
 		/* Are any pointers crossing a page boundary? */
 		if (pgto_base == 0) {
-			flush_dcache_page(*pgto);
 			pgto_base = PAGE_CACHE_SIZE;
 			pgto--;
 		}
@@ -211,11 +210,11 @@ _shift_data_right_pages(struct page **pages, size_t pgto_base,
 		vto = kmap_atomic(*pgto, KM_USER0);
 		vfrom = kmap_atomic(*pgfrom, KM_USER1);
 		memmove(vto + pgto_base, vfrom + pgfrom_base, copy);
+		flush_dcache_page(*pgto);
 		kunmap_atomic(vfrom, KM_USER1);
 		kunmap_atomic(vto, KM_USER0);
 
 	} while ((len -= copy) != 0);
-	flush_dcache_page(*pgto);
 }
 
 /*