Merge tag 'nfs-for-3.8-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client bugfixes from Trond Myklebust: - Error reporting in nfs_xdev_mount incorrectly maps all errors to ENOMEM - Fix an NFSv4 refcounting issue - Fix a mount failure when the server reboots during NFSv4 trunking discovery - NFSv4.1 mounts may need to run the lease recovery thread. - Don't silently fail setattr() requests on mountpoints - Fix a SUNRPC socket/transport livelock and priority queue issue - We must handle NFS4ERR_DELAY when resetting the NFSv4.1 session. * tag 'nfs-for-3.8-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: NFSv4.1: Handle NFS4ERR_DELAY when resetting the NFSv4.1 session SUNRPC: When changing the queue priority, ensure that we change the owner NFS: Don't silently fail setattr() requests on mountpoints NFSv4.1: Ensure that nfs41_walk_client_list() does start lease recovery NFSv4: Fix NFSv4 trunking discovery NFSv4: Fix NFSv4 reference counting for trunked sessions NFS: Fix error reporting in nfs_xdev_mount

Merge tag 'nfs-for-3.8-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Pull NFS client bugfixes from Trond Myklebust: - Error reporting in nfs_xdev_mount incorrectly maps all errors to ENOMEM - Fix an NFSv4 refcounting issue - Fix a mount failure when the server reboots during NFSv4 trunking discovery - NFSv4.1 mounts may need to run the lease recovery thread. - Don't silently fail setattr() requests on mountpoints - Fix a SUNRPC socket/transport livelock and priority queue issue - We must handle NFS4ERR_DELAY when resetting the NFSv4.1 session. * tag 'nfs-for-3.8-4' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: NFSv4.1: Handle NFS4ERR_DELAY when resetting the NFSv4.1 session SUNRPC: When changing the queue priority, ensure that we change the owner NFS: Don't silently fail setattr() requests on mountpoints NFSv4.1: Ensure that nfs41_walk_client_list() does start lease recovery NFSv4: Fix NFSv4 trunking discovery NFSv4: Fix NFSv4 reference counting for trunked sessions NFS: Fix error reporting in nfs_xdev_mount
Linus Torvalds
2 parents aeb8eede8e c489ee290b
Showing 5 changed files Side-by-side Diff
fs/nfs/namespace.c
fs/nfs/nfs4client.c
fs/nfs/nfs4state.c
fs/nfs/super.c
net/sunrpc/sched.c
@@ -177,11 +177,31 @@
 	return mnt;
 }
  
+static int
+nfs_namespace_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
+{
+	if (NFS_FH(dentry->d_inode)->size != 0)
+		return nfs_getattr(mnt, dentry, stat);
+	generic_fillattr(dentry->d_inode, stat);
+	return 0;
+}
+
+static int
+nfs_namespace_setattr(struct dentry *dentry, struct iattr *attr)
+{
+	if (NFS_FH(dentry->d_inode)->size != 0)
+		return nfs_setattr(dentry, attr);
+	return -EACCES;
+}
+
 const struct inode_operations nfs_mountpoint_inode_operations = {
 	.getattr	= nfs_getattr,
+	.setattr	= nfs_setattr,
 };
  
 const struct inode_operations nfs_referral_inode_operations = {
+	.getattr	= nfs_namespace_getattr,
+	.setattr	= nfs_namespace_setattr,
 };
  
 static void nfs_expire_automounts(struct work_struct *work)
@@ -236,11 +236,10 @@
 	error = nfs4_discover_server_trunking(clp, &old);
 	if (error < 0)
 		goto error;
+	nfs_put_client(clp);
 	if (clp != old) {
 		clp->cl_preserve_clid = true;
-		nfs_put_client(clp);
 		clp = old;
-		atomic_inc(&clp->cl_count);
 	}
  
 	return clp;
@@ -306,7 +305,7 @@
 		.clientid	= new->cl_clientid,
 		.confirm	= new->cl_confirm,
 	};
-	int status;
+	int status = -NFS4ERR_STALE_CLIENTID;
  
 	spin_lock(&nn->nfs_client_lock);
 	list_for_each_entry_safe(pos, n, &nn->nfs_client_list, cl_share_link) {
  
  
  
  
  
  
  
  
@@ -332,40 +331,33 @@
  
 		if (prev)
 			nfs_put_client(prev);
+		prev = pos;
  
 		status = nfs4_proc_setclientid_confirm(pos, &clid, cred);
-		if (status == 0) {
+		switch (status) {
+		case -NFS4ERR_STALE_CLIENTID:
+			break;
+		case 0:
 			nfs4_swap_callback_idents(pos, new);
  
-			nfs_put_client(pos);
+			prev = NULL;
 			*result = pos;
 			dprintk("NFS: <-- %s using nfs_client = %p ({%d})\n",
 				__func__, pos, atomic_read(&pos->cl_count));
-			return 0;
+		default:
+			goto out;
 		}
-		if (status != -NFS4ERR_STALE_CLIENTID) {
-			nfs_put_client(pos);
-			dprintk("NFS: <-- %s status = %d, no result\n",
-				__func__, status);
-			return status;
-		}
  
 		spin_lock(&nn->nfs_client_lock);
-		prev = pos;
 	}
+	spin_unlock(&nn->nfs_client_lock);
  
-	/*
-	 * No matching nfs_client found.  This should be impossible,
-	 * because the new nfs_client has already been added to
-	 * nfs_client_list by nfs_get_client().
-	 *
-	 * Don't BUG(), since the caller is holding a mutex.
-	 */
+	/* No match found. The server lost our clientid */
+out:
 	if (prev)
 		nfs_put_client(prev);
-	spin_unlock(&nn->nfs_client_lock);
-	pr_err("NFS: %s Error: no matching nfs_client found\n", __func__);
-	return -NFS4ERR_STALE_CLIENTID;
+	dprintk("NFS: <-- %s status = %d\n", __func__, status);
+	return status;
 }
  
 #ifdef CONFIG_NFS_V4_1
@@ -432,7 +424,7 @@
 {
 	struct nfs_net *nn = net_generic(new->cl_net, nfs_net_id);
 	struct nfs_client *pos, *n, *prev = NULL;
-	int error;
+	int status = -NFS4ERR_STALE_CLIENTID;
  
 	spin_lock(&nn->nfs_client_lock);
 	list_for_each_entry_safe(pos, n, &nn->nfs_client_list, cl_share_link) {
  
  
@@ -448,14 +440,17 @@
 				nfs_put_client(prev);
 			prev = pos;
  
-			error = nfs_wait_client_init_complete(pos);
-			if (error < 0) {
+			nfs4_schedule_lease_recovery(pos);
+			status = nfs_wait_client_init_complete(pos);
+			if (status < 0) {
 				nfs_put_client(pos);
 				spin_lock(&nn->nfs_client_lock);
 				continue;
 			}
-
+			status = pos->cl_cons_state;
 			spin_lock(&nn->nfs_client_lock);
+			if (status < 0)
+				continue;
 		}
  
 		if (pos->rpc_ops != new->rpc_ops)
@@ -473,6 +468,7 @@
 		if (!nfs4_match_serverowners(pos, new))
 			continue;
  
+		atomic_inc(&pos->cl_count);
 		spin_unlock(&nn->nfs_client_lock);
 		dprintk("NFS: <-- %s using nfs_client = %p ({%d})\n",
 			__func__, pos, atomic_read(&pos->cl_count));
  
@@ -481,16 +477,10 @@
 		return 0;
 	}
  
-	/*
-	 * No matching nfs_client found.  This should be impossible,
-	 * because the new nfs_client has already been added to
-	 * nfs_client_list by nfs_get_client().
-	 *
-	 * Don't BUG(), since the caller is holding a mutex.
-	 */
+	/* No matching nfs_client found. */
 	spin_unlock(&nn->nfs_client_lock);
-	pr_err("NFS: %s Error: no matching nfs_client found\n", __func__);
-	return -NFS4ERR_STALE_CLIENTID;
+	dprintk("NFS: <-- %s status = %d\n", __func__, status);
+	return status;
 }
 #endif	/* CONFIG_NFS_V4_1 */
  
@@ -136,16 +136,11 @@
 	clp->cl_confirm = clid.confirm;
  
 	status = nfs40_walk_client_list(clp, result, cred);
-	switch (status) {
-	case -NFS4ERR_STALE_CLIENTID:
-		set_bit(NFS4CLNT_LEASE_CONFIRM, &clp->cl_state);
-	case 0:
+	if (status == 0) {
 		/* Sustain the lease, even if it's empty.  If the clientid4
 		 * goes stale it's of no use for trunking discovery. */
 		nfs4_schedule_state_renewal(*result);
-		break;
 	}
-
 out:
 	return status;
 }
@@ -1863,6 +1858,7 @@
 	case -ETIMEDOUT:
 	case -EAGAIN:
 		ssleep(1);
+	case -NFS4ERR_STALE_CLIENTID:
 		dprintk("NFS: %s after status %d, retrying\n",
 			__func__, status);
 		goto again;
@@ -2022,8 +2018,18 @@
 	nfs4_begin_drain_session(clp);
 	cred = nfs4_get_exchange_id_cred(clp);
 	status = nfs4_proc_destroy_session(clp->cl_session, cred);
-	if (status && status != -NFS4ERR_BADSESSION &&
-	    status != -NFS4ERR_DEADSESSION) {
+	switch (status) {
+	case 0:
+	case -NFS4ERR_BADSESSION:
+	case -NFS4ERR_DEADSESSION:
+		break;
+	case -NFS4ERR_BACK_CHAN_BUSY:
+	case -NFS4ERR_DELAY:
+		set_bit(NFS4CLNT_SESSION_RESET, &clp->cl_state);
+		status = 0;
+		ssleep(1);
+		goto out;
+	default:
 		status = nfs4_recovery_handle_error(clp, status);
 		goto out;
 	}
@@ -2589,27 +2589,23 @@
 	struct nfs_server *server;
 	struct dentry *mntroot = ERR_PTR(-ENOMEM);
 	struct nfs_subversion *nfs_mod = NFS_SB(data->sb)->nfs_client->cl_nfs_mod;
-	int error;
  
-	dprintk("--> nfs_xdev_mount_common()\n");
+	dprintk("--> nfs_xdev_mount()\n");
  
 	mount_info.mntfh = mount_info.cloned->fh;
  
 	/* create a new volume representation */
 	server = nfs_mod->rpc_ops->clone_server(NFS_SB(data->sb), data->fh, data->fattr, data->authflavor);
-	if (IS_ERR(server)) {
-		error = PTR_ERR(server);
-		goto out_err;
-	}
  
-	mntroot = nfs_fs_mount_common(server, flags, dev_name, &mount_info, nfs_mod);
-	dprintk("<-- nfs_xdev_mount_common() = 0\n");
-out:
-	return mntroot;
+	if (IS_ERR(server))
+		mntroot = ERR_CAST(server);
+	else
+		mntroot = nfs_fs_mount_common(server, flags,
+				dev_name, &mount_info, nfs_mod);
  
-out_err:
-	dprintk("<-- nfs_xdev_mount_common() = %d [error]\n", error);
-	goto out;
+	dprintk("<-- nfs_xdev_mount() = %ld\n",
+			IS_ERR(mntroot) ? PTR_ERR(mntroot) : 0L);
+	return mntroot;
 }
  
 #if IS_ENABLED(CONFIG_NFS_V4)
@@ -98,9 +98,25 @@
 	list_add(&task->u.tk_wait.timer_list, &queue->timer_list.list);
 }
  
+static void rpc_rotate_queue_owner(struct rpc_wait_queue *queue)
+{
+	struct list_head *q = &queue->tasks[queue->priority];
+	struct rpc_task *task;
+
+	if (!list_empty(q)) {
+		task = list_first_entry(q, struct rpc_task, u.tk_wait.list);
+		if (task->tk_owner == queue->owner)
+			list_move_tail(&task->u.tk_wait.list, q);
+	}
+}
+
 static void rpc_set_waitqueue_priority(struct rpc_wait_queue *queue, int priority)
 {
-	queue->priority = priority;
+	if (queue->priority != priority) {
+		/* Fairness: rotate the list when changing priority */
+		rpc_rotate_queue_owner(queue);
+		queue->priority = priority;
+	}
 }
  
 static void rpc_set_waitqueue_owner(struct rpc_wait_queue *queue, pid_t pid)
...	...	@@ -177,11 +177,31 @@
177	177	return mnt;
178	178	}
179	179
	180	+static int
	181	+nfs_namespace_getattr(struct vfsmount mnt, struct dentry dentry, struct kstat *stat)
	182	+{
	183	+ if (NFS_FH(dentry->d_inode)->size != 0)
	184	+ return nfs_getattr(mnt, dentry, stat);
	185	+ generic_fillattr(dentry->d_inode, stat);
	186	+ return 0;
	187	+}
	188	+
	189	+static int
	190	+nfs_namespace_setattr(struct dentry dentry, struct iattr attr)
	191	+{
	192	+ if (NFS_FH(dentry->d_inode)->size != 0)
	193	+ return nfs_setattr(dentry, attr);
	194	+ return -EACCES;
	195	+}
	196	+
180	197	const struct inode_operations nfs_mountpoint_inode_operations = {
181	198	.getattr = nfs_getattr,
	199	+ .setattr = nfs_setattr,
182	200	};
183	201
184	202	const struct inode_operations nfs_referral_inode_operations = {
	203	+ .getattr = nfs_namespace_getattr,
	204	+ .setattr = nfs_namespace_setattr,
185	205	};
186	206
187	207	static void nfs_expire_automounts(struct work_struct *work)
...	...	@@ -236,11 +236,10 @@
236	236	error = nfs4_discover_server_trunking(clp, &old);
237	237	if (error < 0)
238	238	goto error;
	239	+ nfs_put_client(clp);
239	240	if (clp != old) {
240	241	clp->cl_preserve_clid = true;
241		- nfs_put_client(clp);
242	242	clp = old;
243		- atomic_inc(&clp->cl_count);
244	243	}
245	244
246	245	return clp;
...	...	@@ -306,7 +305,7 @@
306	305	.clientid = new->cl_clientid,
307	306	.confirm = new->cl_confirm,
308	307	};
309		- int status;
	308	+ int status = -NFS4ERR_STALE_CLIENTID;
310	309
311	310	spin_lock(&nn->nfs_client_lock);
312	311	list_for_each_entry_safe(pos, n, &nn->nfs_client_list, cl_share_link) {
313	312
314	313
315	314
316	315
317	316
318	317
319	318
320	319
...	...	@@ -332,40 +331,33 @@
332	331
333	332	if (prev)
334	333	nfs_put_client(prev);
	334	+ prev = pos;
335	335
336	336	status = nfs4_proc_setclientid_confirm(pos, &clid, cred);
337		- if (status == 0) {
	337	+ switch (status) {
	338	+ case -NFS4ERR_STALE_CLIENTID:
	339	+ break;
	340	+ case 0:
338	341	nfs4_swap_callback_idents(pos, new);
339	342
340		- nfs_put_client(pos);
	343	+ prev = NULL;
341	344	*result = pos;
342	345	dprintk("NFS: <-- %s using nfs_client = %p ({%d})\n",
343	346	__func__, pos, atomic_read(&pos->cl_count));
344		- return 0;
	347	+ default:
	348	+ goto out;
345	349	}
346		- if (status != -NFS4ERR_STALE_CLIENTID) {
347		- nfs_put_client(pos);
348		- dprintk("NFS: <-- %s status = %d, no result\n",
349		- __func__, status);
350		- return status;
351		- }
352	350
353	351	spin_lock(&nn->nfs_client_lock);
354		- prev = pos;
355	352	}
	353	+ spin_unlock(&nn->nfs_client_lock);
356	354
357		- /*
358		- * No matching nfs_client found. This should be impossible,
359		- * because the new nfs_client has already been added to
360		- * nfs_client_list by nfs_get_client().
361		- *
362		- * Don't BUG(), since the caller is holding a mutex.
363		- */
	355	+ /* No match found. The server lost our clientid */
	356	+out:
364	357	if (prev)
365	358	nfs_put_client(prev);
366		- spin_unlock(&nn->nfs_client_lock);
367		- pr_err("NFS: %s Error: no matching nfs_client found\n", __func__);
368		- return -NFS4ERR_STALE_CLIENTID;
	359	+ dprintk("NFS: <-- %s status = %d\n", __func__, status);
	360	+ return status;
369	361	}
370	362
371	363	#ifdef CONFIG_NFS_V4_1
...	...	@@ -432,7 +424,7 @@
432	424	{
433	425	struct nfs_net *nn = net_generic(new->cl_net, nfs_net_id);
434	426	struct nfs_client pos, n, *prev = NULL;
435		- int error;
	427	+ int status = -NFS4ERR_STALE_CLIENTID;
436	428
437	429	spin_lock(&nn->nfs_client_lock);
438	430	list_for_each_entry_safe(pos, n, &nn->nfs_client_list, cl_share_link) {
439	431
440	432
...	...	@@ -448,14 +440,17 @@
448	440	nfs_put_client(prev);
449	441	prev = pos;
450	442
451		- error = nfs_wait_client_init_complete(pos);
452		- if (error < 0) {
	443	+ nfs4_schedule_lease_recovery(pos);
	444	+ status = nfs_wait_client_init_complete(pos);
	445	+ if (status < 0) {
453	446	nfs_put_client(pos);
454	447	spin_lock(&nn->nfs_client_lock);
455	448	continue;
456	449	}
457		-
	450	+ status = pos->cl_cons_state;
458	451	spin_lock(&nn->nfs_client_lock);
	452	+ if (status < 0)
	453	+ continue;
459	454	}
460	455
461	456	if (pos->rpc_ops != new->rpc_ops)
...	...	@@ -473,6 +468,7 @@
473	468	if (!nfs4_match_serverowners(pos, new))
474	469	continue;
475	470
	471	+ atomic_inc(&pos->cl_count);
476	472	spin_unlock(&nn->nfs_client_lock);
477	473	dprintk("NFS: <-- %s using nfs_client = %p ({%d})\n",
478	474	__func__, pos, atomic_read(&pos->cl_count));
479	475
...	...	@@ -481,16 +477,10 @@
481	477	return 0;
482	478	}
483	479
484		- /*
485		- * No matching nfs_client found. This should be impossible,
486		- * because the new nfs_client has already been added to
487		- * nfs_client_list by nfs_get_client().
488		- *
489		- * Don't BUG(), since the caller is holding a mutex.
490		- */
	480	+ /* No matching nfs_client found. */
491	481	spin_unlock(&nn->nfs_client_lock);
492		- pr_err("NFS: %s Error: no matching nfs_client found\n", __func__);
493		- return -NFS4ERR_STALE_CLIENTID;
	482	+ dprintk("NFS: <-- %s status = %d\n", __func__, status);
	483	+ return status;
494	484	}
495	485	#endif /* CONFIG_NFS_V4_1 */
496	486
...	...	@@ -136,16 +136,11 @@
136	136	clp->cl_confirm = clid.confirm;
137	137
138	138	status = nfs40_walk_client_list(clp, result, cred);
139		- switch (status) {
140		- case -NFS4ERR_STALE_CLIENTID:
141		- set_bit(NFS4CLNT_LEASE_CONFIRM, &clp->cl_state);
142		- case 0:
	139	+ if (status == 0) {
143	140	/* Sustain the lease, even if it's empty. If the clientid4
144	141	* goes stale it's of no use for trunking discovery. */
145	142	nfs4_schedule_state_renewal(*result);
146		- break;
147	143	}
148		-
149	144	out:
150	145	return status;
151	146	}
...	...	@@ -1863,6 +1858,7 @@
1863	1858	case -ETIMEDOUT:
1864	1859	case -EAGAIN:
1865	1860	ssleep(1);
	1861	+ case -NFS4ERR_STALE_CLIENTID:
1866	1862	dprintk("NFS: %s after status %d, retrying\n",
1867	1863	__func__, status);
1868	1864	goto again;
...	...	@@ -2022,8 +2018,18 @@
2022	2018	nfs4_begin_drain_session(clp);
2023	2019	cred = nfs4_get_exchange_id_cred(clp);
2024	2020	status = nfs4_proc_destroy_session(clp->cl_session, cred);
2025		- if (status && status != -NFS4ERR_BADSESSION &&
2026		- status != -NFS4ERR_DEADSESSION) {
	2021	+ switch (status) {
	2022	+ case 0:
	2023	+ case -NFS4ERR_BADSESSION:
	2024	+ case -NFS4ERR_DEADSESSION:
	2025	+ break;
	2026	+ case -NFS4ERR_BACK_CHAN_BUSY:
	2027	+ case -NFS4ERR_DELAY:
	2028	+ set_bit(NFS4CLNT_SESSION_RESET, &clp->cl_state);
	2029	+ status = 0;
	2030	+ ssleep(1);
	2031	+ goto out;
	2032	+ default:
2027	2033	status = nfs4_recovery_handle_error(clp, status);
2028	2034	goto out;
2029	2035	}
...	...	@@ -2589,27 +2589,23 @@
2589	2589	struct nfs_server *server;
2590	2590	struct dentry *mntroot = ERR_PTR(-ENOMEM);
2591	2591	struct nfs_subversion *nfs_mod = NFS_SB(data->sb)->nfs_client->cl_nfs_mod;
2592		- int error;
2593	2592
2594		- dprintk("--> nfs_xdev_mount_common()\n");
	2593	+ dprintk("--> nfs_xdev_mount()\n");
2595	2594
2596	2595	mount_info.mntfh = mount_info.cloned->fh;
2597	2596
2598	2597	/* create a new volume representation */
2599	2598	server = nfs_mod->rpc_ops->clone_server(NFS_SB(data->sb), data->fh, data->fattr, data->authflavor);
2600		- if (IS_ERR(server)) {
2601		- error = PTR_ERR(server);
2602		- goto out_err;
2603		- }
2604	2599
2605		- mntroot = nfs_fs_mount_common(server, flags, dev_name, &mount_info, nfs_mod);
2606		- dprintk("<-- nfs_xdev_mount_common() = 0\n");
2607		-out:
2608		- return mntroot;
	2600	+ if (IS_ERR(server))
	2601	+ mntroot = ERR_CAST(server);
	2602	+ else
	2603	+ mntroot = nfs_fs_mount_common(server, flags,
	2604	+ dev_name, &mount_info, nfs_mod);
2609	2605
2610		-out_err:
2611		- dprintk("<-- nfs_xdev_mount_common() = %d [error]\n", error);
2612		- goto out;
	2606	+ dprintk("<-- nfs_xdev_mount() = %ld\n",
	2607	+ IS_ERR(mntroot) ? PTR_ERR(mntroot) : 0L);
	2608	+ return mntroot;
2613	2609	}
2614	2610
2615	2611	#if IS_ENABLED(CONFIG_NFS_V4)
...	...	@@ -98,9 +98,25 @@
98	98	list_add(&task->u.tk_wait.timer_list, &queue->timer_list.list);
99	99	}
100	100
	101	+static void rpc_rotate_queue_owner(struct rpc_wait_queue *queue)
	102	+{
	103	+ struct list_head *q = &queue->tasks[queue->priority];
	104	+ struct rpc_task *task;
	105	+
	106	+ if (!list_empty(q)) {
	107	+ task = list_first_entry(q, struct rpc_task, u.tk_wait.list);
	108	+ if (task->tk_owner == queue->owner)
	109	+ list_move_tail(&task->u.tk_wait.list, q);
	110	+ }
	111	+}
	112	+
101	113	static void rpc_set_waitqueue_priority(struct rpc_wait_queue *queue, int priority)
102	114	{
103		- queue->priority = priority;
	115	+ if (queue->priority != priority) {
	116	+ /* Fairness: rotate the list when changing priority */
	117	+ rpc_rotate_queue_owner(queue);
	118	+ queue->priority = priority;
	119	+ }
104	120	}
105	121
106	122	static void rpc_set_waitqueue_owner(struct rpc_wait_queue *queue, pid_t pid)