xref: /kernel/linux/linux-6.6/fs/nfs/nfs4proc.c (revision 62306a36)
1/*
2 *  fs/nfs/nfs4proc.c
3 *
4 *  Client-side procedure declarations for NFSv4.
5 *
6 *  Copyright (c) 2002 The Regents of the University of Michigan.
7 *  All rights reserved.
8 *
9 *  Kendrick Smith <kmsmith@umich.edu>
10 *  Andy Adamson   <andros@umich.edu>
11 *
12 *  Redistribution and use in source and binary forms, with or without
13 *  modification, are permitted provided that the following conditions
14 *  are met:
15 *
16 *  1. Redistributions of source code must retain the above copyright
17 *     notice, this list of conditions and the following disclaimer.
18 *  2. Redistributions in binary form must reproduce the above copyright
19 *     notice, this list of conditions and the following disclaimer in the
20 *     documentation and/or other materials provided with the distribution.
21 *  3. Neither the name of the University nor the names of its
22 *     contributors may be used to endorse or promote products derived
23 *     from this software without specific prior written permission.
24 *
25 *  THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
26 *  WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
27 *  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
28 *  DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29 *  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30 *  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31 *  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
32 *  BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
33 *  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
34 *  NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
35 *  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
36 */
37
38#include <linux/mm.h>
39#include <linux/delay.h>
40#include <linux/errno.h>
41#include <linux/string.h>
42#include <linux/ratelimit.h>
43#include <linux/printk.h>
44#include <linux/slab.h>
45#include <linux/sunrpc/clnt.h>
46#include <linux/nfs.h>
47#include <linux/nfs4.h>
48#include <linux/nfs_fs.h>
49#include <linux/nfs_page.h>
50#include <linux/nfs_mount.h>
51#include <linux/namei.h>
52#include <linux/mount.h>
53#include <linux/module.h>
54#include <linux/xattr.h>
55#include <linux/utsname.h>
56#include <linux/freezer.h>
57#include <linux/iversion.h>
58
59#include "nfs4_fs.h"
60#include "delegation.h"
61#include "internal.h"
62#include "iostat.h"
63#include "callback.h"
64#include "pnfs.h"
65#include "netns.h"
66#include "sysfs.h"
67#include "nfs4idmap.h"
68#include "nfs4session.h"
69#include "fscache.h"
70#include "nfs42.h"
71
72#include "nfs4trace.h"
73
74#define NFSDBG_FACILITY		NFSDBG_PROC
75
76#define NFS4_BITMASK_SZ		3
77
78#define NFS4_POLL_RETRY_MIN	(HZ/10)
79#define NFS4_POLL_RETRY_MAX	(15*HZ)
80
81/* file attributes which can be mapped to nfs attributes */
82#define NFS4_VALID_ATTRS (ATTR_MODE \
83	| ATTR_UID \
84	| ATTR_GID \
85	| ATTR_SIZE \
86	| ATTR_ATIME \
87	| ATTR_MTIME \
88	| ATTR_CTIME \
89	| ATTR_ATIME_SET \
90	| ATTR_MTIME_SET)
91
92struct nfs4_opendata;
93static int _nfs4_recover_proc_open(struct nfs4_opendata *data);
94static int nfs4_do_fsinfo(struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *);
95static void nfs_fixup_referral_attributes(struct nfs_fattr *fattr);
96static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle,
97			      struct nfs_fattr *fattr, struct inode *inode);
98static int nfs4_do_setattr(struct inode *inode, const struct cred *cred,
99			    struct nfs_fattr *fattr, struct iattr *sattr,
100			    struct nfs_open_context *ctx, struct nfs4_label *ilabel);
101#ifdef CONFIG_NFS_V4_1
102static struct rpc_task *_nfs41_proc_sequence(struct nfs_client *clp,
103		const struct cred *cred,
104		struct nfs4_slot *slot,
105		bool is_privileged);
106static int nfs41_test_stateid(struct nfs_server *, nfs4_stateid *,
107		const struct cred *);
108static int nfs41_free_stateid(struct nfs_server *, const nfs4_stateid *,
109		const struct cred *, bool);
110#endif
111
112#ifdef CONFIG_NFS_V4_SECURITY_LABEL
113static inline struct nfs4_label *
114nfs4_label_init_security(struct inode *dir, struct dentry *dentry,
115	struct iattr *sattr, struct nfs4_label *label)
116{
117	int err;
118
119	if (label == NULL)
120		return NULL;
121
122	if (nfs_server_capable(dir, NFS_CAP_SECURITY_LABEL) == 0)
123		return NULL;
124
125	label->lfs = 0;
126	label->pi = 0;
127	label->len = 0;
128	label->label = NULL;
129
130	err = security_dentry_init_security(dentry, sattr->ia_mode,
131				&dentry->d_name, NULL,
132				(void **)&label->label, &label->len);
133	if (err == 0)
134		return label;
135
136	return NULL;
137}
138static inline void
139nfs4_label_release_security(struct nfs4_label *label)
140{
141	if (label)
142		security_release_secctx(label->label, label->len);
143}
144static inline u32 *nfs4_bitmask(struct nfs_server *server, struct nfs4_label *label)
145{
146	if (label)
147		return server->attr_bitmask;
148
149	return server->attr_bitmask_nl;
150}
151#else
152static inline struct nfs4_label *
153nfs4_label_init_security(struct inode *dir, struct dentry *dentry,
154	struct iattr *sattr, struct nfs4_label *l)
155{ return NULL; }
156static inline void
157nfs4_label_release_security(struct nfs4_label *label)
158{ return; }
159static inline u32 *
160nfs4_bitmask(struct nfs_server *server, struct nfs4_label *label)
161{ return server->attr_bitmask; }
162#endif
163
164/* Prevent leaks of NFSv4 errors into userland */
165static int nfs4_map_errors(int err)
166{
167	if (err >= -1000)
168		return err;
169	switch (err) {
170	case -NFS4ERR_RESOURCE:
171	case -NFS4ERR_LAYOUTTRYLATER:
172	case -NFS4ERR_RECALLCONFLICT:
173	case -NFS4ERR_RETURNCONFLICT:
174		return -EREMOTEIO;
175	case -NFS4ERR_WRONGSEC:
176	case -NFS4ERR_WRONG_CRED:
177		return -EPERM;
178	case -NFS4ERR_BADOWNER:
179	case -NFS4ERR_BADNAME:
180		return -EINVAL;
181	case -NFS4ERR_SHARE_DENIED:
182		return -EACCES;
183	case -NFS4ERR_MINOR_VERS_MISMATCH:
184		return -EPROTONOSUPPORT;
185	case -NFS4ERR_FILE_OPEN:
186		return -EBUSY;
187	case -NFS4ERR_NOT_SAME:
188		return -ENOTSYNC;
189	default:
190		dprintk("%s could not handle NFSv4 error %d\n",
191				__func__, -err);
192		break;
193	}
194	return -EIO;
195}
196
197/*
198 * This is our standard bitmap for GETATTR requests.
199 */
200const u32 nfs4_fattr_bitmap[3] = {
201	FATTR4_WORD0_TYPE
202	| FATTR4_WORD0_CHANGE
203	| FATTR4_WORD0_SIZE
204	| FATTR4_WORD0_FSID
205	| FATTR4_WORD0_FILEID,
206	FATTR4_WORD1_MODE
207	| FATTR4_WORD1_NUMLINKS
208	| FATTR4_WORD1_OWNER
209	| FATTR4_WORD1_OWNER_GROUP
210	| FATTR4_WORD1_RAWDEV
211	| FATTR4_WORD1_SPACE_USED
212	| FATTR4_WORD1_TIME_ACCESS
213	| FATTR4_WORD1_TIME_METADATA
214	| FATTR4_WORD1_TIME_MODIFY
215	| FATTR4_WORD1_MOUNTED_ON_FILEID,
216#ifdef CONFIG_NFS_V4_SECURITY_LABEL
217	FATTR4_WORD2_SECURITY_LABEL
218#endif
219};
220
221static const u32 nfs4_pnfs_open_bitmap[3] = {
222	FATTR4_WORD0_TYPE
223	| FATTR4_WORD0_CHANGE
224	| FATTR4_WORD0_SIZE
225	| FATTR4_WORD0_FSID
226	| FATTR4_WORD0_FILEID,
227	FATTR4_WORD1_MODE
228	| FATTR4_WORD1_NUMLINKS
229	| FATTR4_WORD1_OWNER
230	| FATTR4_WORD1_OWNER_GROUP
231	| FATTR4_WORD1_RAWDEV
232	| FATTR4_WORD1_SPACE_USED
233	| FATTR4_WORD1_TIME_ACCESS
234	| FATTR4_WORD1_TIME_METADATA
235	| FATTR4_WORD1_TIME_MODIFY,
236	FATTR4_WORD2_MDSTHRESHOLD
237#ifdef CONFIG_NFS_V4_SECURITY_LABEL
238	| FATTR4_WORD2_SECURITY_LABEL
239#endif
240};
241
242static const u32 nfs4_open_noattr_bitmap[3] = {
243	FATTR4_WORD0_TYPE
244	| FATTR4_WORD0_FILEID,
245};
246
247const u32 nfs4_statfs_bitmap[3] = {
248	FATTR4_WORD0_FILES_AVAIL
249	| FATTR4_WORD0_FILES_FREE
250	| FATTR4_WORD0_FILES_TOTAL,
251	FATTR4_WORD1_SPACE_AVAIL
252	| FATTR4_WORD1_SPACE_FREE
253	| FATTR4_WORD1_SPACE_TOTAL
254};
255
256const u32 nfs4_pathconf_bitmap[3] = {
257	FATTR4_WORD0_MAXLINK
258	| FATTR4_WORD0_MAXNAME,
259	0
260};
261
262const u32 nfs4_fsinfo_bitmap[3] = { FATTR4_WORD0_MAXFILESIZE
263			| FATTR4_WORD0_MAXREAD
264			| FATTR4_WORD0_MAXWRITE
265			| FATTR4_WORD0_LEASE_TIME,
266			FATTR4_WORD1_TIME_DELTA
267			| FATTR4_WORD1_FS_LAYOUT_TYPES,
268			FATTR4_WORD2_LAYOUT_BLKSIZE
269			| FATTR4_WORD2_CLONE_BLKSIZE
270			| FATTR4_WORD2_CHANGE_ATTR_TYPE
271			| FATTR4_WORD2_XATTR_SUPPORT
272};
273
274const u32 nfs4_fs_locations_bitmap[3] = {
275	FATTR4_WORD0_CHANGE
276	| FATTR4_WORD0_SIZE
277	| FATTR4_WORD0_FSID
278	| FATTR4_WORD0_FILEID
279	| FATTR4_WORD0_FS_LOCATIONS,
280	FATTR4_WORD1_OWNER
281	| FATTR4_WORD1_OWNER_GROUP
282	| FATTR4_WORD1_RAWDEV
283	| FATTR4_WORD1_SPACE_USED
284	| FATTR4_WORD1_TIME_ACCESS
285	| FATTR4_WORD1_TIME_METADATA
286	| FATTR4_WORD1_TIME_MODIFY
287	| FATTR4_WORD1_MOUNTED_ON_FILEID,
288};
289
290static void nfs4_bitmap_copy_adjust(__u32 *dst, const __u32 *src,
291				    struct inode *inode, unsigned long flags)
292{
293	unsigned long cache_validity;
294
295	memcpy(dst, src, NFS4_BITMASK_SZ*sizeof(*dst));
296	if (!inode || !nfs4_have_delegation(inode, FMODE_READ))
297		return;
298
299	cache_validity = READ_ONCE(NFS_I(inode)->cache_validity) | flags;
300
301	/* Remove the attributes over which we have full control */
302	dst[1] &= ~FATTR4_WORD1_RAWDEV;
303	if (!(cache_validity & NFS_INO_INVALID_SIZE))
304		dst[0] &= ~FATTR4_WORD0_SIZE;
305
306	if (!(cache_validity & NFS_INO_INVALID_CHANGE))
307		dst[0] &= ~FATTR4_WORD0_CHANGE;
308
309	if (!(cache_validity & NFS_INO_INVALID_MODE))
310		dst[1] &= ~FATTR4_WORD1_MODE;
311	if (!(cache_validity & NFS_INO_INVALID_OTHER))
312		dst[1] &= ~(FATTR4_WORD1_OWNER | FATTR4_WORD1_OWNER_GROUP);
313}
314
315static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dentry,
316		struct nfs4_readdir_arg *readdir)
317{
318	unsigned int attrs = FATTR4_WORD0_FILEID | FATTR4_WORD0_TYPE;
319	__be32 *start, *p;
320
321	if (cookie > 2) {
322		readdir->cookie = cookie;
323		memcpy(&readdir->verifier, verifier, sizeof(readdir->verifier));
324		return;
325	}
326
327	readdir->cookie = 0;
328	memset(&readdir->verifier, 0, sizeof(readdir->verifier));
329	if (cookie == 2)
330		return;
331
332	/*
333	 * NFSv4 servers do not return entries for '.' and '..'
334	 * Therefore, we fake these entries here.  We let '.'
335	 * have cookie 0 and '..' have cookie 1.  Note that
336	 * when talking to the server, we always send cookie 0
337	 * instead of 1 or 2.
338	 */
339	start = p = kmap_atomic(*readdir->pages);
340
341	if (cookie == 0) {
342		*p++ = xdr_one;                                  /* next */
343		*p++ = xdr_zero;                   /* cookie, first word */
344		*p++ = xdr_one;                   /* cookie, second word */
345		*p++ = xdr_one;                             /* entry len */
346		memcpy(p, ".\0\0\0", 4);                        /* entry */
347		p++;
348		*p++ = xdr_one;                         /* bitmap length */
349		*p++ = htonl(attrs);                           /* bitmap */
350		*p++ = htonl(12);             /* attribute buffer length */
351		*p++ = htonl(NF4DIR);
352		p = xdr_encode_hyper(p, NFS_FILEID(d_inode(dentry)));
353	}
354
355	*p++ = xdr_one;                                  /* next */
356	*p++ = xdr_zero;                   /* cookie, first word */
357	*p++ = xdr_two;                   /* cookie, second word */
358	*p++ = xdr_two;                             /* entry len */
359	memcpy(p, "..\0\0", 4);                         /* entry */
360	p++;
361	*p++ = xdr_one;                         /* bitmap length */
362	*p++ = htonl(attrs);                           /* bitmap */
363	*p++ = htonl(12);             /* attribute buffer length */
364	*p++ = htonl(NF4DIR);
365	p = xdr_encode_hyper(p, NFS_FILEID(d_inode(dentry->d_parent)));
366
367	readdir->pgbase = (char *)p - (char *)start;
368	readdir->count -= readdir->pgbase;
369	kunmap_atomic(start);
370}
371
372static void nfs4_fattr_set_prechange(struct nfs_fattr *fattr, u64 version)
373{
374	if (!(fattr->valid & NFS_ATTR_FATTR_PRECHANGE)) {
375		fattr->pre_change_attr = version;
376		fattr->valid |= NFS_ATTR_FATTR_PRECHANGE;
377	}
378}
379
380static void nfs4_test_and_free_stateid(struct nfs_server *server,
381		nfs4_stateid *stateid,
382		const struct cred *cred)
383{
384	const struct nfs4_minor_version_ops *ops = server->nfs_client->cl_mvops;
385
386	ops->test_and_free_expired(server, stateid, cred);
387}
388
389static void __nfs4_free_revoked_stateid(struct nfs_server *server,
390		nfs4_stateid *stateid,
391		const struct cred *cred)
392{
393	stateid->type = NFS4_REVOKED_STATEID_TYPE;
394	nfs4_test_and_free_stateid(server, stateid, cred);
395}
396
397static void nfs4_free_revoked_stateid(struct nfs_server *server,
398		const nfs4_stateid *stateid,
399		const struct cred *cred)
400{
401	nfs4_stateid tmp;
402
403	nfs4_stateid_copy(&tmp, stateid);
404	__nfs4_free_revoked_stateid(server, &tmp, cred);
405}
406
407static long nfs4_update_delay(long *timeout)
408{
409	long ret;
410	if (!timeout)
411		return NFS4_POLL_RETRY_MAX;
412	if (*timeout <= 0)
413		*timeout = NFS4_POLL_RETRY_MIN;
414	if (*timeout > NFS4_POLL_RETRY_MAX)
415		*timeout = NFS4_POLL_RETRY_MAX;
416	ret = *timeout;
417	*timeout <<= 1;
418	return ret;
419}
420
421static int nfs4_delay_killable(long *timeout)
422{
423	might_sleep();
424
425	__set_current_state(TASK_KILLABLE|TASK_FREEZABLE_UNSAFE);
426	schedule_timeout(nfs4_update_delay(timeout));
427	if (!__fatal_signal_pending(current))
428		return 0;
429	return -EINTR;
430}
431
432static int nfs4_delay_interruptible(long *timeout)
433{
434	might_sleep();
435
436	__set_current_state(TASK_INTERRUPTIBLE|TASK_FREEZABLE_UNSAFE);
437	schedule_timeout(nfs4_update_delay(timeout));
438	if (!signal_pending(current))
439		return 0;
440	return __fatal_signal_pending(current) ? -EINTR :-ERESTARTSYS;
441}
442
443static int nfs4_delay(long *timeout, bool interruptible)
444{
445	if (interruptible)
446		return nfs4_delay_interruptible(timeout);
447	return nfs4_delay_killable(timeout);
448}
449
450static const nfs4_stateid *
451nfs4_recoverable_stateid(const nfs4_stateid *stateid)
452{
453	if (!stateid)
454		return NULL;
455	switch (stateid->type) {
456	case NFS4_OPEN_STATEID_TYPE:
457	case NFS4_LOCK_STATEID_TYPE:
458	case NFS4_DELEGATION_STATEID_TYPE:
459		return stateid;
460	default:
461		break;
462	}
463	return NULL;
464}
465
466/* This is the error handling routine for processes that are allowed
467 * to sleep.
468 */
469static int nfs4_do_handle_exception(struct nfs_server *server,
470		int errorcode, struct nfs4_exception *exception)
471{
472	struct nfs_client *clp = server->nfs_client;
473	struct nfs4_state *state = exception->state;
474	const nfs4_stateid *stateid;
475	struct inode *inode = exception->inode;
476	int ret = errorcode;
477
478	exception->delay = 0;
479	exception->recovering = 0;
480	exception->retry = 0;
481
482	stateid = nfs4_recoverable_stateid(exception->stateid);
483	if (stateid == NULL && state != NULL)
484		stateid = nfs4_recoverable_stateid(&state->stateid);
485
486	switch(errorcode) {
487		case 0:
488			return 0;
489		case -NFS4ERR_BADHANDLE:
490		case -ESTALE:
491			if (inode != NULL && S_ISREG(inode->i_mode))
492				pnfs_destroy_layout(NFS_I(inode));
493			break;
494		case -NFS4ERR_DELEG_REVOKED:
495		case -NFS4ERR_ADMIN_REVOKED:
496		case -NFS4ERR_EXPIRED:
497		case -NFS4ERR_BAD_STATEID:
498		case -NFS4ERR_PARTNER_NO_AUTH:
499			if (inode != NULL && stateid != NULL) {
500				nfs_inode_find_state_and_recover(inode,
501						stateid);
502				goto wait_on_recovery;
503			}
504			fallthrough;
505		case -NFS4ERR_OPENMODE:
506			if (inode) {
507				int err;
508
509				err = nfs_async_inode_return_delegation(inode,
510						stateid);
511				if (err == 0)
512					goto wait_on_recovery;
513				if (stateid != NULL && stateid->type == NFS4_DELEGATION_STATEID_TYPE) {
514					exception->retry = 1;
515					break;
516				}
517			}
518			if (state == NULL)
519				break;
520			ret = nfs4_schedule_stateid_recovery(server, state);
521			if (ret < 0)
522				break;
523			goto wait_on_recovery;
524		case -NFS4ERR_STALE_STATEID:
525		case -NFS4ERR_STALE_CLIENTID:
526			nfs4_schedule_lease_recovery(clp);
527			goto wait_on_recovery;
528		case -NFS4ERR_MOVED:
529			ret = nfs4_schedule_migration_recovery(server);
530			if (ret < 0)
531				break;
532			goto wait_on_recovery;
533		case -NFS4ERR_LEASE_MOVED:
534			nfs4_schedule_lease_moved_recovery(clp);
535			goto wait_on_recovery;
536#if defined(CONFIG_NFS_V4_1)
537		case -NFS4ERR_BADSESSION:
538		case -NFS4ERR_BADSLOT:
539		case -NFS4ERR_BAD_HIGH_SLOT:
540		case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
541		case -NFS4ERR_DEADSESSION:
542		case -NFS4ERR_SEQ_FALSE_RETRY:
543		case -NFS4ERR_SEQ_MISORDERED:
544			/* Handled in nfs41_sequence_process() */
545			goto wait_on_recovery;
546#endif /* defined(CONFIG_NFS_V4_1) */
547		case -NFS4ERR_FILE_OPEN:
548			if (exception->timeout > HZ) {
549				/* We have retried a decent amount, time to
550				 * fail
551				 */
552				ret = -EBUSY;
553				break;
554			}
555			fallthrough;
556		case -NFS4ERR_DELAY:
557			nfs_inc_server_stats(server, NFSIOS_DELAY);
558			fallthrough;
559		case -NFS4ERR_GRACE:
560		case -NFS4ERR_LAYOUTTRYLATER:
561		case -NFS4ERR_RECALLCONFLICT:
562		case -NFS4ERR_RETURNCONFLICT:
563			exception->delay = 1;
564			return 0;
565
566		case -NFS4ERR_RETRY_UNCACHED_REP:
567		case -NFS4ERR_OLD_STATEID:
568			exception->retry = 1;
569			break;
570		case -NFS4ERR_BADOWNER:
571			/* The following works around a Linux server bug! */
572		case -NFS4ERR_BADNAME:
573			if (server->caps & NFS_CAP_UIDGID_NOMAP) {
574				server->caps &= ~NFS_CAP_UIDGID_NOMAP;
575				exception->retry = 1;
576				printk(KERN_WARNING "NFS: v4 server %s "
577						"does not accept raw "
578						"uid/gids. "
579						"Reenabling the idmapper.\n",
580						server->nfs_client->cl_hostname);
581			}
582	}
583	/* We failed to handle the error */
584	return nfs4_map_errors(ret);
585wait_on_recovery:
586	exception->recovering = 1;
587	return 0;
588}
589
590/* This is the error handling routine for processes that are allowed
591 * to sleep.
592 */
593int nfs4_handle_exception(struct nfs_server *server, int errorcode, struct nfs4_exception *exception)
594{
595	struct nfs_client *clp = server->nfs_client;
596	int ret;
597
598	ret = nfs4_do_handle_exception(server, errorcode, exception);
599	if (exception->delay) {
600		ret = nfs4_delay(&exception->timeout,
601				exception->interruptible);
602		goto out_retry;
603	}
604	if (exception->recovering) {
605		if (exception->task_is_privileged)
606			return -EDEADLOCK;
607		ret = nfs4_wait_clnt_recover(clp);
608		if (test_bit(NFS_MIG_FAILED, &server->mig_status))
609			return -EIO;
610		goto out_retry;
611	}
612	return ret;
613out_retry:
614	if (ret == 0)
615		exception->retry = 1;
616	return ret;
617}
618
619static int
620nfs4_async_handle_exception(struct rpc_task *task, struct nfs_server *server,
621		int errorcode, struct nfs4_exception *exception)
622{
623	struct nfs_client *clp = server->nfs_client;
624	int ret;
625
626	ret = nfs4_do_handle_exception(server, errorcode, exception);
627	if (exception->delay) {
628		rpc_delay(task, nfs4_update_delay(&exception->timeout));
629		goto out_retry;
630	}
631	if (exception->recovering) {
632		if (exception->task_is_privileged)
633			return -EDEADLOCK;
634		rpc_sleep_on(&clp->cl_rpcwaitq, task, NULL);
635		if (test_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state) == 0)
636			rpc_wake_up_queued_task(&clp->cl_rpcwaitq, task);
637		goto out_retry;
638	}
639	if (test_bit(NFS_MIG_FAILED, &server->mig_status))
640		ret = -EIO;
641	return ret;
642out_retry:
643	if (ret == 0) {
644		exception->retry = 1;
645		/*
646		 * For NFS4ERR_MOVED, the client transport will need to
647		 * be recomputed after migration recovery has completed.
648		 */
649		if (errorcode == -NFS4ERR_MOVED)
650			rpc_task_release_transport(task);
651	}
652	return ret;
653}
654
655int
656nfs4_async_handle_error(struct rpc_task *task, struct nfs_server *server,
657			struct nfs4_state *state, long *timeout)
658{
659	struct nfs4_exception exception = {
660		.state = state,
661	};
662
663	if (task->tk_status >= 0)
664		return 0;
665	if (timeout)
666		exception.timeout = *timeout;
667	task->tk_status = nfs4_async_handle_exception(task, server,
668			task->tk_status,
669			&exception);
670	if (exception.delay && timeout)
671		*timeout = exception.timeout;
672	if (exception.retry)
673		return -EAGAIN;
674	return 0;
675}
676
677/*
678 * Return 'true' if 'clp' is using an rpc_client that is integrity protected
679 * or 'false' otherwise.
680 */
681static bool _nfs4_is_integrity_protected(struct nfs_client *clp)
682{
683	rpc_authflavor_t flavor = clp->cl_rpcclient->cl_auth->au_flavor;
684	return (flavor == RPC_AUTH_GSS_KRB5I) || (flavor == RPC_AUTH_GSS_KRB5P);
685}
686
687static void do_renew_lease(struct nfs_client *clp, unsigned long timestamp)
688{
689	spin_lock(&clp->cl_lock);
690	if (time_before(clp->cl_last_renewal,timestamp))
691		clp->cl_last_renewal = timestamp;
692	spin_unlock(&clp->cl_lock);
693}
694
695static void renew_lease(const struct nfs_server *server, unsigned long timestamp)
696{
697	struct nfs_client *clp = server->nfs_client;
698
699	if (!nfs4_has_session(clp))
700		do_renew_lease(clp, timestamp);
701}
702
703struct nfs4_call_sync_data {
704	const struct nfs_server *seq_server;
705	struct nfs4_sequence_args *seq_args;
706	struct nfs4_sequence_res *seq_res;
707};
708
709void nfs4_init_sequence(struct nfs4_sequence_args *args,
710			struct nfs4_sequence_res *res, int cache_reply,
711			int privileged)
712{
713	args->sa_slot = NULL;
714	args->sa_cache_this = cache_reply;
715	args->sa_privileged = privileged;
716
717	res->sr_slot = NULL;
718}
719
720static void nfs40_sequence_free_slot(struct nfs4_sequence_res *res)
721{
722	struct nfs4_slot *slot = res->sr_slot;
723	struct nfs4_slot_table *tbl;
724
725	tbl = slot->table;
726	spin_lock(&tbl->slot_tbl_lock);
727	if (!nfs41_wake_and_assign_slot(tbl, slot))
728		nfs4_free_slot(tbl, slot);
729	spin_unlock(&tbl->slot_tbl_lock);
730
731	res->sr_slot = NULL;
732}
733
734static int nfs40_sequence_done(struct rpc_task *task,
735			       struct nfs4_sequence_res *res)
736{
737	if (res->sr_slot != NULL)
738		nfs40_sequence_free_slot(res);
739	return 1;
740}
741
742#if defined(CONFIG_NFS_V4_1)
743
744static void nfs41_release_slot(struct nfs4_slot *slot)
745{
746	struct nfs4_session *session;
747	struct nfs4_slot_table *tbl;
748	bool send_new_highest_used_slotid = false;
749
750	if (!slot)
751		return;
752	tbl = slot->table;
753	session = tbl->session;
754
755	/* Bump the slot sequence number */
756	if (slot->seq_done)
757		slot->seq_nr++;
758	slot->seq_done = 0;
759
760	spin_lock(&tbl->slot_tbl_lock);
761	/* Be nice to the server: try to ensure that the last transmitted
762	 * value for highest_user_slotid <= target_highest_slotid
763	 */
764	if (tbl->highest_used_slotid > tbl->target_highest_slotid)
765		send_new_highest_used_slotid = true;
766
767	if (nfs41_wake_and_assign_slot(tbl, slot)) {
768		send_new_highest_used_slotid = false;
769		goto out_unlock;
770	}
771	nfs4_free_slot(tbl, slot);
772
773	if (tbl->highest_used_slotid != NFS4_NO_SLOT)
774		send_new_highest_used_slotid = false;
775out_unlock:
776	spin_unlock(&tbl->slot_tbl_lock);
777	if (send_new_highest_used_slotid)
778		nfs41_notify_server(session->clp);
779	if (waitqueue_active(&tbl->slot_waitq))
780		wake_up_all(&tbl->slot_waitq);
781}
782
783static void nfs41_sequence_free_slot(struct nfs4_sequence_res *res)
784{
785	nfs41_release_slot(res->sr_slot);
786	res->sr_slot = NULL;
787}
788
789static void nfs4_slot_sequence_record_sent(struct nfs4_slot *slot,
790		u32 seqnr)
791{
792	if ((s32)(seqnr - slot->seq_nr_highest_sent) > 0)
793		slot->seq_nr_highest_sent = seqnr;
794}
795static void nfs4_slot_sequence_acked(struct nfs4_slot *slot, u32 seqnr)
796{
797	nfs4_slot_sequence_record_sent(slot, seqnr);
798	slot->seq_nr_last_acked = seqnr;
799}
800
801static void nfs4_probe_sequence(struct nfs_client *client, const struct cred *cred,
802				struct nfs4_slot *slot)
803{
804	struct rpc_task *task = _nfs41_proc_sequence(client, cred, slot, true);
805	if (!IS_ERR(task))
806		rpc_put_task_async(task);
807}
808
809static int nfs41_sequence_process(struct rpc_task *task,
810		struct nfs4_sequence_res *res)
811{
812	struct nfs4_session *session;
813	struct nfs4_slot *slot = res->sr_slot;
814	struct nfs_client *clp;
815	int status;
816	int ret = 1;
817
818	if (slot == NULL)
819		goto out_noaction;
820	/* don't increment the sequence number if the task wasn't sent */
821	if (!RPC_WAS_SENT(task) || slot->seq_done)
822		goto out;
823
824	session = slot->table->session;
825	clp = session->clp;
826
827	trace_nfs4_sequence_done(session, res);
828
829	status = res->sr_status;
830	if (task->tk_status == -NFS4ERR_DEADSESSION)
831		status = -NFS4ERR_DEADSESSION;
832
833	/* Check the SEQUENCE operation status */
834	switch (status) {
835	case 0:
836		/* Mark this sequence number as having been acked */
837		nfs4_slot_sequence_acked(slot, slot->seq_nr);
838		/* Update the slot's sequence and clientid lease timer */
839		slot->seq_done = 1;
840		do_renew_lease(clp, res->sr_timestamp);
841		/* Check sequence flags */
842		nfs41_handle_sequence_flag_errors(clp, res->sr_status_flags,
843				!!slot->privileged);
844		nfs41_update_target_slotid(slot->table, slot, res);
845		break;
846	case 1:
847		/*
848		 * sr_status remains 1 if an RPC level error occurred.
849		 * The server may or may not have processed the sequence
850		 * operation..
851		 */
852		nfs4_slot_sequence_record_sent(slot, slot->seq_nr);
853		slot->seq_done = 1;
854		goto out;
855	case -NFS4ERR_DELAY:
856		/* The server detected a resend of the RPC call and
857		 * returned NFS4ERR_DELAY as per Section 2.10.6.2
858		 * of RFC5661.
859		 */
860		dprintk("%s: slot=%u seq=%u: Operation in progress\n",
861			__func__,
862			slot->slot_nr,
863			slot->seq_nr);
864		goto out_retry;
865	case -NFS4ERR_RETRY_UNCACHED_REP:
866	case -NFS4ERR_SEQ_FALSE_RETRY:
867		/*
868		 * The server thinks we tried to replay a request.
869		 * Retry the call after bumping the sequence ID.
870		 */
871		nfs4_slot_sequence_acked(slot, slot->seq_nr);
872		goto retry_new_seq;
873	case -NFS4ERR_BADSLOT:
874		/*
875		 * The slot id we used was probably retired. Try again
876		 * using a different slot id.
877		 */
878		if (slot->slot_nr < slot->table->target_highest_slotid)
879			goto session_recover;
880		goto retry_nowait;
881	case -NFS4ERR_SEQ_MISORDERED:
882		nfs4_slot_sequence_record_sent(slot, slot->seq_nr);
883		/*
884		 * Were one or more calls using this slot interrupted?
885		 * If the server never received the request, then our
886		 * transmitted slot sequence number may be too high. However,
887		 * if the server did receive the request then it might
888		 * accidentally give us a reply with a mismatched operation.
889		 * We can sort this out by sending a lone sequence operation
890		 * to the server on the same slot.
891		 */
892		if ((s32)(slot->seq_nr - slot->seq_nr_last_acked) > 1) {
893			slot->seq_nr--;
894			if (task->tk_msg.rpc_proc != &nfs4_procedures[NFSPROC4_CLNT_SEQUENCE]) {
895				nfs4_probe_sequence(clp, task->tk_msg.rpc_cred, slot);
896				res->sr_slot = NULL;
897			}
898			goto retry_nowait;
899		}
900		/*
901		 * RFC5661:
902		 * A retry might be sent while the original request is
903		 * still in progress on the replier. The replier SHOULD
904		 * deal with the issue by returning NFS4ERR_DELAY as the
905		 * reply to SEQUENCE or CB_SEQUENCE operation, but
906		 * implementations MAY return NFS4ERR_SEQ_MISORDERED.
907		 *
908		 * Restart the search after a delay.
909		 */
910		slot->seq_nr = slot->seq_nr_highest_sent;
911		goto out_retry;
912	case -NFS4ERR_BADSESSION:
913	case -NFS4ERR_DEADSESSION:
914	case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
915		goto session_recover;
916	default:
917		/* Just update the slot sequence no. */
918		slot->seq_done = 1;
919	}
920out:
921	/* The session may be reset by one of the error handlers. */
922	dprintk("%s: Error %d free the slot \n", __func__, res->sr_status);
923out_noaction:
924	return ret;
925session_recover:
926	set_bit(NFS4_SLOT_TBL_DRAINING, &session->fc_slot_table.slot_tbl_state);
927	nfs4_schedule_session_recovery(session, status);
928	dprintk("%s ERROR: %d Reset session\n", __func__, status);
929	nfs41_sequence_free_slot(res);
930	goto out;
931retry_new_seq:
932	++slot->seq_nr;
933retry_nowait:
934	if (rpc_restart_call_prepare(task)) {
935		nfs41_sequence_free_slot(res);
936		task->tk_status = 0;
937		ret = 0;
938	}
939	goto out;
940out_retry:
941	if (!rpc_restart_call(task))
942		goto out;
943	rpc_delay(task, NFS4_POLL_RETRY_MAX);
944	return 0;
945}
946
947int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res)
948{
949	if (!nfs41_sequence_process(task, res))
950		return 0;
951	if (res->sr_slot != NULL)
952		nfs41_sequence_free_slot(res);
953	return 1;
954
955}
956EXPORT_SYMBOL_GPL(nfs41_sequence_done);
957
958static int nfs4_sequence_process(struct rpc_task *task, struct nfs4_sequence_res *res)
959{
960	if (res->sr_slot == NULL)
961		return 1;
962	if (res->sr_slot->table->session != NULL)
963		return nfs41_sequence_process(task, res);
964	return nfs40_sequence_done(task, res);
965}
966
967static void nfs4_sequence_free_slot(struct nfs4_sequence_res *res)
968{
969	if (res->sr_slot != NULL) {
970		if (res->sr_slot->table->session != NULL)
971			nfs41_sequence_free_slot(res);
972		else
973			nfs40_sequence_free_slot(res);
974	}
975}
976
977int nfs4_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res)
978{
979	if (res->sr_slot == NULL)
980		return 1;
981	if (!res->sr_slot->table->session)
982		return nfs40_sequence_done(task, res);
983	return nfs41_sequence_done(task, res);
984}
985EXPORT_SYMBOL_GPL(nfs4_sequence_done);
986
987static void nfs41_call_sync_prepare(struct rpc_task *task, void *calldata)
988{
989	struct nfs4_call_sync_data *data = calldata;
990
991	dprintk("--> %s data->seq_server %p\n", __func__, data->seq_server);
992
993	nfs4_setup_sequence(data->seq_server->nfs_client,
994			    data->seq_args, data->seq_res, task);
995}
996
997static void nfs41_call_sync_done(struct rpc_task *task, void *calldata)
998{
999	struct nfs4_call_sync_data *data = calldata;
1000
1001	nfs41_sequence_done(task, data->seq_res);
1002}
1003
1004static const struct rpc_call_ops nfs41_call_sync_ops = {
1005	.rpc_call_prepare = nfs41_call_sync_prepare,
1006	.rpc_call_done = nfs41_call_sync_done,
1007};
1008
1009#else	/* !CONFIG_NFS_V4_1 */
1010
1011static int nfs4_sequence_process(struct rpc_task *task, struct nfs4_sequence_res *res)
1012{
1013	return nfs40_sequence_done(task, res);
1014}
1015
1016static void nfs4_sequence_free_slot(struct nfs4_sequence_res *res)
1017{
1018	if (res->sr_slot != NULL)
1019		nfs40_sequence_free_slot(res);
1020}
1021
1022int nfs4_sequence_done(struct rpc_task *task,
1023		       struct nfs4_sequence_res *res)
1024{
1025	return nfs40_sequence_done(task, res);
1026}
1027EXPORT_SYMBOL_GPL(nfs4_sequence_done);
1028
1029#endif	/* !CONFIG_NFS_V4_1 */
1030
1031static void nfs41_sequence_res_init(struct nfs4_sequence_res *res)
1032{
1033	res->sr_timestamp = jiffies;
1034	res->sr_status_flags = 0;
1035	res->sr_status = 1;
1036}
1037
1038static
1039void nfs4_sequence_attach_slot(struct nfs4_sequence_args *args,
1040		struct nfs4_sequence_res *res,
1041		struct nfs4_slot *slot)
1042{
1043	if (!slot)
1044		return;
1045	slot->privileged = args->sa_privileged ? 1 : 0;
1046	args->sa_slot = slot;
1047
1048	res->sr_slot = slot;
1049}
1050
1051int nfs4_setup_sequence(struct nfs_client *client,
1052			struct nfs4_sequence_args *args,
1053			struct nfs4_sequence_res *res,
1054			struct rpc_task *task)
1055{
1056	struct nfs4_session *session = nfs4_get_session(client);
1057	struct nfs4_slot_table *tbl  = client->cl_slot_tbl;
1058	struct nfs4_slot *slot;
1059
1060	/* slot already allocated? */
1061	if (res->sr_slot != NULL)
1062		goto out_start;
1063
1064	if (session)
1065		tbl = &session->fc_slot_table;
1066
1067	spin_lock(&tbl->slot_tbl_lock);
1068	/* The state manager will wait until the slot table is empty */
1069	if (nfs4_slot_tbl_draining(tbl) && !args->sa_privileged)
1070		goto out_sleep;
1071
1072	slot = nfs4_alloc_slot(tbl);
1073	if (IS_ERR(slot)) {
1074		if (slot == ERR_PTR(-ENOMEM))
1075			goto out_sleep_timeout;
1076		goto out_sleep;
1077	}
1078	spin_unlock(&tbl->slot_tbl_lock);
1079
1080	nfs4_sequence_attach_slot(args, res, slot);
1081
1082	trace_nfs4_setup_sequence(session, args);
1083out_start:
1084	nfs41_sequence_res_init(res);
1085	rpc_call_start(task);
1086	return 0;
1087out_sleep_timeout:
1088	/* Try again in 1/4 second */
1089	if (args->sa_privileged)
1090		rpc_sleep_on_priority_timeout(&tbl->slot_tbl_waitq, task,
1091				jiffies + (HZ >> 2), RPC_PRIORITY_PRIVILEGED);
1092	else
1093		rpc_sleep_on_timeout(&tbl->slot_tbl_waitq, task,
1094				NULL, jiffies + (HZ >> 2));
1095	spin_unlock(&tbl->slot_tbl_lock);
1096	return -EAGAIN;
1097out_sleep:
1098	if (args->sa_privileged)
1099		rpc_sleep_on_priority(&tbl->slot_tbl_waitq, task,
1100				RPC_PRIORITY_PRIVILEGED);
1101	else
1102		rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
1103	spin_unlock(&tbl->slot_tbl_lock);
1104	return -EAGAIN;
1105}
1106EXPORT_SYMBOL_GPL(nfs4_setup_sequence);
1107
1108static void nfs40_call_sync_prepare(struct rpc_task *task, void *calldata)
1109{
1110	struct nfs4_call_sync_data *data = calldata;
1111	nfs4_setup_sequence(data->seq_server->nfs_client,
1112				data->seq_args, data->seq_res, task);
1113}
1114
1115static void nfs40_call_sync_done(struct rpc_task *task, void *calldata)
1116{
1117	struct nfs4_call_sync_data *data = calldata;
1118	nfs4_sequence_done(task, data->seq_res);
1119}
1120
1121static const struct rpc_call_ops nfs40_call_sync_ops = {
1122	.rpc_call_prepare = nfs40_call_sync_prepare,
1123	.rpc_call_done = nfs40_call_sync_done,
1124};
1125
1126static int nfs4_call_sync_custom(struct rpc_task_setup *task_setup)
1127{
1128	int ret;
1129	struct rpc_task *task;
1130
1131	task = rpc_run_task(task_setup);
1132	if (IS_ERR(task))
1133		return PTR_ERR(task);
1134
1135	ret = task->tk_status;
1136	rpc_put_task(task);
1137	return ret;
1138}
1139
1140static int nfs4_do_call_sync(struct rpc_clnt *clnt,
1141			     struct nfs_server *server,
1142			     struct rpc_message *msg,
1143			     struct nfs4_sequence_args *args,
1144			     struct nfs4_sequence_res *res,
1145			     unsigned short task_flags)
1146{
1147	struct nfs_client *clp = server->nfs_client;
1148	struct nfs4_call_sync_data data = {
1149		.seq_server = server,
1150		.seq_args = args,
1151		.seq_res = res,
1152	};
1153	struct rpc_task_setup task_setup = {
1154		.rpc_client = clnt,
1155		.rpc_message = msg,
1156		.callback_ops = clp->cl_mvops->call_sync_ops,
1157		.callback_data = &data,
1158		.flags = task_flags,
1159	};
1160
1161	return nfs4_call_sync_custom(&task_setup);
1162}
1163
1164static int nfs4_call_sync_sequence(struct rpc_clnt *clnt,
1165				   struct nfs_server *server,
1166				   struct rpc_message *msg,
1167				   struct nfs4_sequence_args *args,
1168				   struct nfs4_sequence_res *res)
1169{
1170	unsigned short task_flags = 0;
1171
1172	if (server->caps & NFS_CAP_MOVEABLE)
1173		task_flags = RPC_TASK_MOVEABLE;
1174	return nfs4_do_call_sync(clnt, server, msg, args, res, task_flags);
1175}
1176
1177
1178int nfs4_call_sync(struct rpc_clnt *clnt,
1179		   struct nfs_server *server,
1180		   struct rpc_message *msg,
1181		   struct nfs4_sequence_args *args,
1182		   struct nfs4_sequence_res *res,
1183		   int cache_reply)
1184{
1185	nfs4_init_sequence(args, res, cache_reply, 0);
1186	return nfs4_call_sync_sequence(clnt, server, msg, args, res);
1187}
1188
1189static void
1190nfs4_inc_nlink_locked(struct inode *inode)
1191{
1192	nfs_set_cache_invalid(inode, NFS_INO_INVALID_CHANGE |
1193					     NFS_INO_INVALID_CTIME |
1194					     NFS_INO_INVALID_NLINK);
1195	inc_nlink(inode);
1196}
1197
1198static void
1199nfs4_inc_nlink(struct inode *inode)
1200{
1201	spin_lock(&inode->i_lock);
1202	nfs4_inc_nlink_locked(inode);
1203	spin_unlock(&inode->i_lock);
1204}
1205
1206static void
1207nfs4_dec_nlink_locked(struct inode *inode)
1208{
1209	nfs_set_cache_invalid(inode, NFS_INO_INVALID_CHANGE |
1210					     NFS_INO_INVALID_CTIME |
1211					     NFS_INO_INVALID_NLINK);
1212	drop_nlink(inode);
1213}
1214
1215static void
1216nfs4_update_changeattr_locked(struct inode *inode,
1217		struct nfs4_change_info *cinfo,
1218		unsigned long timestamp, unsigned long cache_validity)
1219{
1220	struct nfs_inode *nfsi = NFS_I(inode);
1221	u64 change_attr = inode_peek_iversion_raw(inode);
1222
1223	cache_validity |= NFS_INO_INVALID_CTIME | NFS_INO_INVALID_MTIME;
1224	if (S_ISDIR(inode->i_mode))
1225		cache_validity |= NFS_INO_INVALID_DATA;
1226
1227	switch (NFS_SERVER(inode)->change_attr_type) {
1228	case NFS4_CHANGE_TYPE_IS_UNDEFINED:
1229		if (cinfo->after == change_attr)
1230			goto out;
1231		break;
1232	default:
1233		if ((s64)(change_attr - cinfo->after) >= 0)
1234			goto out;
1235	}
1236
1237	inode_set_iversion_raw(inode, cinfo->after);
1238	if (!cinfo->atomic || cinfo->before != change_attr) {
1239		if (S_ISDIR(inode->i_mode))
1240			nfs_force_lookup_revalidate(inode);
1241
1242		if (!NFS_PROTO(inode)->have_delegation(inode, FMODE_READ))
1243			cache_validity |=
1244				NFS_INO_INVALID_ACCESS | NFS_INO_INVALID_ACL |
1245				NFS_INO_INVALID_SIZE | NFS_INO_INVALID_OTHER |
1246				NFS_INO_INVALID_BLOCKS | NFS_INO_INVALID_NLINK |
1247				NFS_INO_INVALID_MODE | NFS_INO_INVALID_XATTR;
1248		nfsi->attrtimeo = NFS_MINATTRTIMEO(inode);
1249	}
1250	nfsi->attrtimeo_timestamp = jiffies;
1251	nfsi->read_cache_jiffies = timestamp;
1252	nfsi->attr_gencount = nfs_inc_attr_generation_counter();
1253	nfsi->cache_validity &= ~NFS_INO_INVALID_CHANGE;
1254out:
1255	nfs_set_cache_invalid(inode, cache_validity);
1256}
1257
1258void
1259nfs4_update_changeattr(struct inode *dir, struct nfs4_change_info *cinfo,
1260		unsigned long timestamp, unsigned long cache_validity)
1261{
1262	spin_lock(&dir->i_lock);
1263	nfs4_update_changeattr_locked(dir, cinfo, timestamp, cache_validity);
1264	spin_unlock(&dir->i_lock);
1265}
1266
1267struct nfs4_open_createattrs {
1268	struct nfs4_label *label;
1269	struct iattr *sattr;
1270	const __u32 verf[2];
1271};
1272
1273static bool nfs4_clear_cap_atomic_open_v1(struct nfs_server *server,
1274		int err, struct nfs4_exception *exception)
1275{
1276	if (err != -EINVAL)
1277		return false;
1278	if (!(server->caps & NFS_CAP_ATOMIC_OPEN_V1))
1279		return false;
1280	server->caps &= ~NFS_CAP_ATOMIC_OPEN_V1;
1281	exception->retry = 1;
1282	return true;
1283}
1284
1285static fmode_t _nfs4_ctx_to_accessmode(const struct nfs_open_context *ctx)
1286{
1287	 return ctx->mode & (FMODE_READ|FMODE_WRITE|FMODE_EXEC);
1288}
1289
1290static fmode_t _nfs4_ctx_to_openmode(const struct nfs_open_context *ctx)
1291{
1292	fmode_t ret = ctx->mode & (FMODE_READ|FMODE_WRITE);
1293
1294	return (ctx->mode & FMODE_EXEC) ? FMODE_READ | ret : ret;
1295}
1296
1297static u32
1298nfs4_map_atomic_open_share(struct nfs_server *server,
1299		fmode_t fmode, int openflags)
1300{
1301	u32 res = 0;
1302
1303	switch (fmode & (FMODE_READ | FMODE_WRITE)) {
1304	case FMODE_READ:
1305		res = NFS4_SHARE_ACCESS_READ;
1306		break;
1307	case FMODE_WRITE:
1308		res = NFS4_SHARE_ACCESS_WRITE;
1309		break;
1310	case FMODE_READ|FMODE_WRITE:
1311		res = NFS4_SHARE_ACCESS_BOTH;
1312	}
1313	if (!(server->caps & NFS_CAP_ATOMIC_OPEN_V1))
1314		goto out;
1315	/* Want no delegation if we're using O_DIRECT */
1316	if (openflags & O_DIRECT)
1317		res |= NFS4_SHARE_WANT_NO_DELEG;
1318out:
1319	return res;
1320}
1321
1322static enum open_claim_type4
1323nfs4_map_atomic_open_claim(struct nfs_server *server,
1324		enum open_claim_type4 claim)
1325{
1326	if (server->caps & NFS_CAP_ATOMIC_OPEN_V1)
1327		return claim;
1328	switch (claim) {
1329	default:
1330		return claim;
1331	case NFS4_OPEN_CLAIM_FH:
1332		return NFS4_OPEN_CLAIM_NULL;
1333	case NFS4_OPEN_CLAIM_DELEG_CUR_FH:
1334		return NFS4_OPEN_CLAIM_DELEGATE_CUR;
1335	case NFS4_OPEN_CLAIM_DELEG_PREV_FH:
1336		return NFS4_OPEN_CLAIM_DELEGATE_PREV;
1337	}
1338}
1339
1340static void nfs4_init_opendata_res(struct nfs4_opendata *p)
1341{
1342	p->o_res.f_attr = &p->f_attr;
1343	p->o_res.seqid = p->o_arg.seqid;
1344	p->c_res.seqid = p->c_arg.seqid;
1345	p->o_res.server = p->o_arg.server;
1346	p->o_res.access_request = p->o_arg.access;
1347	nfs_fattr_init(&p->f_attr);
1348	nfs_fattr_init_names(&p->f_attr, &p->owner_name, &p->group_name);
1349}
1350
1351static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry,
1352		struct nfs4_state_owner *sp, fmode_t fmode, int flags,
1353		const struct nfs4_open_createattrs *c,
1354		enum open_claim_type4 claim,
1355		gfp_t gfp_mask)
1356{
1357	struct dentry *parent = dget_parent(dentry);
1358	struct inode *dir = d_inode(parent);
1359	struct nfs_server *server = NFS_SERVER(dir);
1360	struct nfs_seqid *(*alloc_seqid)(struct nfs_seqid_counter *, gfp_t);
1361	struct nfs4_label *label = (c != NULL) ? c->label : NULL;
1362	struct nfs4_opendata *p;
1363
1364	p = kzalloc(sizeof(*p), gfp_mask);
1365	if (p == NULL)
1366		goto err;
1367
1368	p->f_attr.label = nfs4_label_alloc(server, gfp_mask);
1369	if (IS_ERR(p->f_attr.label))
1370		goto err_free_p;
1371
1372	p->a_label = nfs4_label_alloc(server, gfp_mask);
1373	if (IS_ERR(p->a_label))
1374		goto err_free_f;
1375
1376	alloc_seqid = server->nfs_client->cl_mvops->alloc_seqid;
1377	p->o_arg.seqid = alloc_seqid(&sp->so_seqid, gfp_mask);
1378	if (IS_ERR(p->o_arg.seqid))
1379		goto err_free_label;
1380	nfs_sb_active(dentry->d_sb);
1381	p->dentry = dget(dentry);
1382	p->dir = parent;
1383	p->owner = sp;
1384	atomic_inc(&sp->so_count);
1385	p->o_arg.open_flags = flags;
1386	p->o_arg.fmode = fmode & (FMODE_READ|FMODE_WRITE);
1387	p->o_arg.claim = nfs4_map_atomic_open_claim(server, claim);
1388	p->o_arg.share_access = nfs4_map_atomic_open_share(server,
1389			fmode, flags);
1390	if (flags & O_CREAT) {
1391		p->o_arg.umask = current_umask();
1392		p->o_arg.label = nfs4_label_copy(p->a_label, label);
1393		if (c->sattr != NULL && c->sattr->ia_valid != 0) {
1394			p->o_arg.u.attrs = &p->attrs;
1395			memcpy(&p->attrs, c->sattr, sizeof(p->attrs));
1396
1397			memcpy(p->o_arg.u.verifier.data, c->verf,
1398					sizeof(p->o_arg.u.verifier.data));
1399		}
1400	}
1401	/* ask server to check for all possible rights as results
1402	 * are cached */
1403	switch (p->o_arg.claim) {
1404	default:
1405		break;
1406	case NFS4_OPEN_CLAIM_NULL:
1407	case NFS4_OPEN_CLAIM_FH:
1408		p->o_arg.access = NFS4_ACCESS_READ | NFS4_ACCESS_MODIFY |
1409				  NFS4_ACCESS_EXTEND | NFS4_ACCESS_DELETE |
1410				  NFS4_ACCESS_EXECUTE |
1411				  nfs_access_xattr_mask(server);
1412	}
1413	p->o_arg.clientid = server->nfs_client->cl_clientid;
1414	p->o_arg.id.create_time = ktime_to_ns(sp->so_seqid.create_time);
1415	p->o_arg.id.uniquifier = sp->so_seqid.owner_id;
1416	p->o_arg.name = &dentry->d_name;
1417	p->o_arg.server = server;
1418	p->o_arg.bitmask = nfs4_bitmask(server, label);
1419	p->o_arg.open_bitmap = &nfs4_fattr_bitmap[0];
1420	switch (p->o_arg.claim) {
1421	case NFS4_OPEN_CLAIM_NULL:
1422	case NFS4_OPEN_CLAIM_DELEGATE_CUR:
1423	case NFS4_OPEN_CLAIM_DELEGATE_PREV:
1424		p->o_arg.fh = NFS_FH(dir);
1425		break;
1426	case NFS4_OPEN_CLAIM_PREVIOUS:
1427	case NFS4_OPEN_CLAIM_FH:
1428	case NFS4_OPEN_CLAIM_DELEG_CUR_FH:
1429	case NFS4_OPEN_CLAIM_DELEG_PREV_FH:
1430		p->o_arg.fh = NFS_FH(d_inode(dentry));
1431	}
1432	p->c_arg.fh = &p->o_res.fh;
1433	p->c_arg.stateid = &p->o_res.stateid;
1434	p->c_arg.seqid = p->o_arg.seqid;
1435	nfs4_init_opendata_res(p);
1436	kref_init(&p->kref);
1437	return p;
1438
1439err_free_label:
1440	nfs4_label_free(p->a_label);
1441err_free_f:
1442	nfs4_label_free(p->f_attr.label);
1443err_free_p:
1444	kfree(p);
1445err:
1446	dput(parent);
1447	return NULL;
1448}
1449
1450static void nfs4_opendata_free(struct kref *kref)
1451{
1452	struct nfs4_opendata *p = container_of(kref,
1453			struct nfs4_opendata, kref);
1454	struct super_block *sb = p->dentry->d_sb;
1455
1456	nfs4_lgopen_release(p->lgp);
1457	nfs_free_seqid(p->o_arg.seqid);
1458	nfs4_sequence_free_slot(&p->o_res.seq_res);
1459	if (p->state != NULL)
1460		nfs4_put_open_state(p->state);
1461	nfs4_put_state_owner(p->owner);
1462
1463	nfs4_label_free(p->a_label);
1464	nfs4_label_free(p->f_attr.label);
1465
1466	dput(p->dir);
1467	dput(p->dentry);
1468	nfs_sb_deactive(sb);
1469	nfs_fattr_free_names(&p->f_attr);
1470	kfree(p->f_attr.mdsthreshold);
1471	kfree(p);
1472}
1473
1474static void nfs4_opendata_put(struct nfs4_opendata *p)
1475{
1476	if (p != NULL)
1477		kref_put(&p->kref, nfs4_opendata_free);
1478}
1479
1480static bool nfs4_mode_match_open_stateid(struct nfs4_state *state,
1481		fmode_t fmode)
1482{
1483	switch(fmode & (FMODE_READ|FMODE_WRITE)) {
1484	case FMODE_READ|FMODE_WRITE:
1485		return state->n_rdwr != 0;
1486	case FMODE_WRITE:
1487		return state->n_wronly != 0;
1488	case FMODE_READ:
1489		return state->n_rdonly != 0;
1490	}
1491	WARN_ON_ONCE(1);
1492	return false;
1493}
1494
1495static int can_open_cached(struct nfs4_state *state, fmode_t mode,
1496		int open_mode, enum open_claim_type4 claim)
1497{
1498	int ret = 0;
1499
1500	if (open_mode & (O_EXCL|O_TRUNC))
1501		goto out;
1502	switch (claim) {
1503	case NFS4_OPEN_CLAIM_NULL:
1504	case NFS4_OPEN_CLAIM_FH:
1505		goto out;
1506	default:
1507		break;
1508	}
1509	switch (mode & (FMODE_READ|FMODE_WRITE)) {
1510		case FMODE_READ:
1511			ret |= test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0
1512				&& state->n_rdonly != 0;
1513			break;
1514		case FMODE_WRITE:
1515			ret |= test_bit(NFS_O_WRONLY_STATE, &state->flags) != 0
1516				&& state->n_wronly != 0;
1517			break;
1518		case FMODE_READ|FMODE_WRITE:
1519			ret |= test_bit(NFS_O_RDWR_STATE, &state->flags) != 0
1520				&& state->n_rdwr != 0;
1521	}
1522out:
1523	return ret;
1524}
1525
1526static int can_open_delegated(struct nfs_delegation *delegation, fmode_t fmode,
1527		enum open_claim_type4 claim)
1528{
1529	if (delegation == NULL)
1530		return 0;
1531	if ((delegation->type & fmode) != fmode)
1532		return 0;
1533	switch (claim) {
1534	case NFS4_OPEN_CLAIM_NULL:
1535	case NFS4_OPEN_CLAIM_FH:
1536		break;
1537	case NFS4_OPEN_CLAIM_PREVIOUS:
1538		if (!test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags))
1539			break;
1540		fallthrough;
1541	default:
1542		return 0;
1543	}
1544	nfs_mark_delegation_referenced(delegation);
1545	return 1;
1546}
1547
1548static void update_open_stateflags(struct nfs4_state *state, fmode_t fmode)
1549{
1550	switch (fmode) {
1551		case FMODE_WRITE:
1552			state->n_wronly++;
1553			break;
1554		case FMODE_READ:
1555			state->n_rdonly++;
1556			break;
1557		case FMODE_READ|FMODE_WRITE:
1558			state->n_rdwr++;
1559	}
1560	nfs4_state_set_mode_locked(state, state->state | fmode);
1561}
1562
1563#ifdef CONFIG_NFS_V4_1
1564static bool nfs_open_stateid_recover_openmode(struct nfs4_state *state)
1565{
1566	if (state->n_rdonly && !test_bit(NFS_O_RDONLY_STATE, &state->flags))
1567		return true;
1568	if (state->n_wronly && !test_bit(NFS_O_WRONLY_STATE, &state->flags))
1569		return true;
1570	if (state->n_rdwr && !test_bit(NFS_O_RDWR_STATE, &state->flags))
1571		return true;
1572	return false;
1573}
1574#endif /* CONFIG_NFS_V4_1 */
1575
1576static void nfs_state_log_update_open_stateid(struct nfs4_state *state)
1577{
1578	if (test_and_clear_bit(NFS_STATE_CHANGE_WAIT, &state->flags))
1579		wake_up_all(&state->waitq);
1580}
1581
1582static void nfs_test_and_clear_all_open_stateid(struct nfs4_state *state)
1583{
1584	struct nfs_client *clp = state->owner->so_server->nfs_client;
1585	bool need_recover = false;
1586
1587	if (test_and_clear_bit(NFS_O_RDONLY_STATE, &state->flags) && state->n_rdonly)
1588		need_recover = true;
1589	if (test_and_clear_bit(NFS_O_WRONLY_STATE, &state->flags) && state->n_wronly)
1590		need_recover = true;
1591	if (test_and_clear_bit(NFS_O_RDWR_STATE, &state->flags) && state->n_rdwr)
1592		need_recover = true;
1593	if (need_recover)
1594		nfs4_state_mark_reclaim_nograce(clp, state);
1595}
1596
1597/*
1598 * Check for whether or not the caller may update the open stateid
1599 * to the value passed in by stateid.
1600 *
1601 * Note: This function relies heavily on the server implementing
1602 * RFC7530 Section 9.1.4.2, and RFC5661 Section 8.2.2
1603 * correctly.
1604 * i.e. The stateid seqids have to be initialised to 1, and
1605 * are then incremented on every state transition.
1606 */
1607static bool nfs_stateid_is_sequential(struct nfs4_state *state,
1608		const nfs4_stateid *stateid)
1609{
1610	if (test_bit(NFS_OPEN_STATE, &state->flags)) {
1611		/* The common case - we're updating to a new sequence number */
1612		if (nfs4_stateid_match_other(stateid, &state->open_stateid)) {
1613			if (nfs4_stateid_is_next(&state->open_stateid, stateid))
1614				return true;
1615			return false;
1616		}
1617		/* The server returned a new stateid */
1618	}
1619	/* This is the first OPEN in this generation */
1620	if (stateid->seqid == cpu_to_be32(1))
1621		return true;
1622	return false;
1623}
1624
1625static void nfs_resync_open_stateid_locked(struct nfs4_state *state)
1626{
1627	if (!(state->n_wronly || state->n_rdonly || state->n_rdwr))
1628		return;
1629	if (state->n_wronly)
1630		set_bit(NFS_O_WRONLY_STATE, &state->flags);
1631	if (state->n_rdonly)
1632		set_bit(NFS_O_RDONLY_STATE, &state->flags);
1633	if (state->n_rdwr)
1634		set_bit(NFS_O_RDWR_STATE, &state->flags);
1635	set_bit(NFS_OPEN_STATE, &state->flags);
1636}
1637
1638static void nfs_clear_open_stateid_locked(struct nfs4_state *state,
1639		nfs4_stateid *stateid, fmode_t fmode)
1640{
1641	clear_bit(NFS_O_RDWR_STATE, &state->flags);
1642	switch (fmode & (FMODE_READ|FMODE_WRITE)) {
1643	case FMODE_WRITE:
1644		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
1645		break;
1646	case FMODE_READ:
1647		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
1648		break;
1649	case 0:
1650		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
1651		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
1652		clear_bit(NFS_OPEN_STATE, &state->flags);
1653	}
1654	if (stateid == NULL)
1655		return;
1656	/* Handle OPEN+OPEN_DOWNGRADE races */
1657	if (nfs4_stateid_match_other(stateid, &state->open_stateid) &&
1658	    !nfs4_stateid_is_newer(stateid, &state->open_stateid)) {
1659		nfs_resync_open_stateid_locked(state);
1660		goto out;
1661	}
1662	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
1663		nfs4_stateid_copy(&state->stateid, stateid);
1664	nfs4_stateid_copy(&state->open_stateid, stateid);
1665	trace_nfs4_open_stateid_update(state->inode, stateid, 0);
1666out:
1667	nfs_state_log_update_open_stateid(state);
1668}
1669
1670static void nfs_clear_open_stateid(struct nfs4_state *state,
1671	nfs4_stateid *arg_stateid,
1672	nfs4_stateid *stateid, fmode_t fmode)
1673{
1674	write_seqlock(&state->seqlock);
1675	/* Ignore, if the CLOSE argment doesn't match the current stateid */
1676	if (nfs4_state_match_open_stateid_other(state, arg_stateid))
1677		nfs_clear_open_stateid_locked(state, stateid, fmode);
1678	write_sequnlock(&state->seqlock);
1679	if (test_bit(NFS_STATE_RECLAIM_NOGRACE, &state->flags))
1680		nfs4_schedule_state_manager(state->owner->so_server->nfs_client);
1681}
1682
1683static void nfs_set_open_stateid_locked(struct nfs4_state *state,
1684		const nfs4_stateid *stateid, nfs4_stateid *freeme)
1685	__must_hold(&state->owner->so_lock)
1686	__must_hold(&state->seqlock)
1687	__must_hold(RCU)
1688
1689{
1690	DEFINE_WAIT(wait);
1691	int status = 0;
1692	for (;;) {
1693
1694		if (nfs_stateid_is_sequential(state, stateid))
1695			break;
1696
1697		if (status)
1698			break;
1699		/* Rely on seqids for serialisation with NFSv4.0 */
1700		if (!nfs4_has_session(NFS_SERVER(state->inode)->nfs_client))
1701			break;
1702
1703		set_bit(NFS_STATE_CHANGE_WAIT, &state->flags);
1704		prepare_to_wait(&state->waitq, &wait, TASK_KILLABLE);
1705		/*
1706		 * Ensure we process the state changes in the same order
1707		 * in which the server processed them by delaying the
1708		 * update of the stateid until we are in sequence.
1709		 */
1710		write_sequnlock(&state->seqlock);
1711		spin_unlock(&state->owner->so_lock);
1712		rcu_read_unlock();
1713		trace_nfs4_open_stateid_update_wait(state->inode, stateid, 0);
1714
1715		if (!fatal_signal_pending(current)) {
1716			if (schedule_timeout(5*HZ) == 0)
1717				status = -EAGAIN;
1718			else
1719				status = 0;
1720		} else
1721			status = -EINTR;
1722		finish_wait(&state->waitq, &wait);
1723		rcu_read_lock();
1724		spin_lock(&state->owner->so_lock);
1725		write_seqlock(&state->seqlock);
1726	}
1727
1728	if (test_bit(NFS_OPEN_STATE, &state->flags) &&
1729	    !nfs4_stateid_match_other(stateid, &state->open_stateid)) {
1730		nfs4_stateid_copy(freeme, &state->open_stateid);
1731		nfs_test_and_clear_all_open_stateid(state);
1732	}
1733
1734	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
1735		nfs4_stateid_copy(&state->stateid, stateid);
1736	nfs4_stateid_copy(&state->open_stateid, stateid);
1737	trace_nfs4_open_stateid_update(state->inode, stateid, status);
1738	nfs_state_log_update_open_stateid(state);
1739}
1740
1741static void nfs_state_set_open_stateid(struct nfs4_state *state,
1742		const nfs4_stateid *open_stateid,
1743		fmode_t fmode,
1744		nfs4_stateid *freeme)
1745{
1746	/*
1747	 * Protect the call to nfs4_state_set_mode_locked and
1748	 * serialise the stateid update
1749	 */
1750	write_seqlock(&state->seqlock);
1751	nfs_set_open_stateid_locked(state, open_stateid, freeme);
1752	switch (fmode) {
1753	case FMODE_READ:
1754		set_bit(NFS_O_RDONLY_STATE, &state->flags);
1755		break;
1756	case FMODE_WRITE:
1757		set_bit(NFS_O_WRONLY_STATE, &state->flags);
1758		break;
1759	case FMODE_READ|FMODE_WRITE:
1760		set_bit(NFS_O_RDWR_STATE, &state->flags);
1761	}
1762	set_bit(NFS_OPEN_STATE, &state->flags);
1763	write_sequnlock(&state->seqlock);
1764}
1765
1766static void nfs_state_clear_open_state_flags(struct nfs4_state *state)
1767{
1768	clear_bit(NFS_O_RDWR_STATE, &state->flags);
1769	clear_bit(NFS_O_WRONLY_STATE, &state->flags);
1770	clear_bit(NFS_O_RDONLY_STATE, &state->flags);
1771	clear_bit(NFS_OPEN_STATE, &state->flags);
1772}
1773
1774static void nfs_state_set_delegation(struct nfs4_state *state,
1775		const nfs4_stateid *deleg_stateid,
1776		fmode_t fmode)
1777{
1778	/*
1779	 * Protect the call to nfs4_state_set_mode_locked and
1780	 * serialise the stateid update
1781	 */
1782	write_seqlock(&state->seqlock);
1783	nfs4_stateid_copy(&state->stateid, deleg_stateid);
1784	set_bit(NFS_DELEGATED_STATE, &state->flags);
1785	write_sequnlock(&state->seqlock);
1786}
1787
1788static void nfs_state_clear_delegation(struct nfs4_state *state)
1789{
1790	write_seqlock(&state->seqlock);
1791	nfs4_stateid_copy(&state->stateid, &state->open_stateid);
1792	clear_bit(NFS_DELEGATED_STATE, &state->flags);
1793	write_sequnlock(&state->seqlock);
1794}
1795
1796int update_open_stateid(struct nfs4_state *state,
1797		const nfs4_stateid *open_stateid,
1798		const nfs4_stateid *delegation,
1799		fmode_t fmode)
1800{
1801	struct nfs_server *server = NFS_SERVER(state->inode);
1802	struct nfs_client *clp = server->nfs_client;
1803	struct nfs_inode *nfsi = NFS_I(state->inode);
1804	struct nfs_delegation *deleg_cur;
1805	nfs4_stateid freeme = { };
1806	int ret = 0;
1807
1808	fmode &= (FMODE_READ|FMODE_WRITE);
1809
1810	rcu_read_lock();
1811	spin_lock(&state->owner->so_lock);
1812	if (open_stateid != NULL) {
1813		nfs_state_set_open_stateid(state, open_stateid, fmode, &freeme);
1814		ret = 1;
1815	}
1816
1817	deleg_cur = nfs4_get_valid_delegation(state->inode);
1818	if (deleg_cur == NULL)
1819		goto no_delegation;
1820
1821	spin_lock(&deleg_cur->lock);
1822	if (rcu_dereference(nfsi->delegation) != deleg_cur ||
1823	   test_bit(NFS_DELEGATION_RETURNING, &deleg_cur->flags) ||
1824	    (deleg_cur->type & fmode) != fmode)
1825		goto no_delegation_unlock;
1826
1827	if (delegation == NULL)
1828		delegation = &deleg_cur->stateid;
1829	else if (!nfs4_stateid_match_other(&deleg_cur->stateid, delegation))
1830		goto no_delegation_unlock;
1831
1832	nfs_mark_delegation_referenced(deleg_cur);
1833	nfs_state_set_delegation(state, &deleg_cur->stateid, fmode);
1834	ret = 1;
1835no_delegation_unlock:
1836	spin_unlock(&deleg_cur->lock);
1837no_delegation:
1838	if (ret)
1839		update_open_stateflags(state, fmode);
1840	spin_unlock(&state->owner->so_lock);
1841	rcu_read_unlock();
1842
1843	if (test_bit(NFS_STATE_RECLAIM_NOGRACE, &state->flags))
1844		nfs4_schedule_state_manager(clp);
1845	if (freeme.type != 0)
1846		nfs4_test_and_free_stateid(server, &freeme,
1847				state->owner->so_cred);
1848
1849	return ret;
1850}
1851
1852static bool nfs4_update_lock_stateid(struct nfs4_lock_state *lsp,
1853		const nfs4_stateid *stateid)
1854{
1855	struct nfs4_state *state = lsp->ls_state;
1856	bool ret = false;
1857
1858	spin_lock(&state->state_lock);
1859	if (!nfs4_stateid_match_other(stateid, &lsp->ls_stateid))
1860		goto out_noupdate;
1861	if (!nfs4_stateid_is_newer(stateid, &lsp->ls_stateid))
1862		goto out_noupdate;
1863	nfs4_stateid_copy(&lsp->ls_stateid, stateid);
1864	ret = true;
1865out_noupdate:
1866	spin_unlock(&state->state_lock);
1867	return ret;
1868}
1869
1870static void nfs4_return_incompatible_delegation(struct inode *inode, fmode_t fmode)
1871{
1872	struct nfs_delegation *delegation;
1873
1874	fmode &= FMODE_READ|FMODE_WRITE;
1875	rcu_read_lock();
1876	delegation = nfs4_get_valid_delegation(inode);
1877	if (delegation == NULL || (delegation->type & fmode) == fmode) {
1878		rcu_read_unlock();
1879		return;
1880	}
1881	rcu_read_unlock();
1882	nfs4_inode_return_delegation(inode);
1883}
1884
1885static struct nfs4_state *nfs4_try_open_cached(struct nfs4_opendata *opendata)
1886{
1887	struct nfs4_state *state = opendata->state;
1888	struct nfs_delegation *delegation;
1889	int open_mode = opendata->o_arg.open_flags;
1890	fmode_t fmode = opendata->o_arg.fmode;
1891	enum open_claim_type4 claim = opendata->o_arg.claim;
1892	nfs4_stateid stateid;
1893	int ret = -EAGAIN;
1894
1895	for (;;) {
1896		spin_lock(&state->owner->so_lock);
1897		if (can_open_cached(state, fmode, open_mode, claim)) {
1898			update_open_stateflags(state, fmode);
1899			spin_unlock(&state->owner->so_lock);
1900			goto out_return_state;
1901		}
1902		spin_unlock(&state->owner->so_lock);
1903		rcu_read_lock();
1904		delegation = nfs4_get_valid_delegation(state->inode);
1905		if (!can_open_delegated(delegation, fmode, claim)) {
1906			rcu_read_unlock();
1907			break;
1908		}
1909		/* Save the delegation */
1910		nfs4_stateid_copy(&stateid, &delegation->stateid);
1911		rcu_read_unlock();
1912		nfs_release_seqid(opendata->o_arg.seqid);
1913		if (!opendata->is_recover) {
1914			ret = nfs_may_open(state->inode, state->owner->so_cred, open_mode);
1915			if (ret != 0)
1916				goto out;
1917		}
1918		ret = -EAGAIN;
1919
1920		/* Try to update the stateid using the delegation */
1921		if (update_open_stateid(state, NULL, &stateid, fmode))
1922			goto out_return_state;
1923	}
1924out:
1925	return ERR_PTR(ret);
1926out_return_state:
1927	refcount_inc(&state->count);
1928	return state;
1929}
1930
1931static void
1932nfs4_opendata_check_deleg(struct nfs4_opendata *data, struct nfs4_state *state)
1933{
1934	struct nfs_client *clp = NFS_SERVER(state->inode)->nfs_client;
1935	struct nfs_delegation *delegation;
1936	int delegation_flags = 0;
1937
1938	rcu_read_lock();
1939	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
1940	if (delegation)
1941		delegation_flags = delegation->flags;
1942	rcu_read_unlock();
1943	switch (data->o_arg.claim) {
1944	default:
1945		break;
1946	case NFS4_OPEN_CLAIM_DELEGATE_CUR:
1947	case NFS4_OPEN_CLAIM_DELEG_CUR_FH:
1948		pr_err_ratelimited("NFS: Broken NFSv4 server %s is "
1949				   "returning a delegation for "
1950				   "OPEN(CLAIM_DELEGATE_CUR)\n",
1951				   clp->cl_hostname);
1952		return;
1953	}
1954	if ((delegation_flags & 1UL<<NFS_DELEGATION_NEED_RECLAIM) == 0)
1955		nfs_inode_set_delegation(state->inode,
1956				data->owner->so_cred,
1957				data->o_res.delegation_type,
1958				&data->o_res.delegation,
1959				data->o_res.pagemod_limit);
1960	else
1961		nfs_inode_reclaim_delegation(state->inode,
1962				data->owner->so_cred,
1963				data->o_res.delegation_type,
1964				&data->o_res.delegation,
1965				data->o_res.pagemod_limit);
1966
1967	if (data->o_res.do_recall)
1968		nfs_async_inode_return_delegation(state->inode,
1969						  &data->o_res.delegation);
1970}
1971
1972/*
1973 * Check the inode attributes against the CLAIM_PREVIOUS returned attributes
1974 * and update the nfs4_state.
1975 */
1976static struct nfs4_state *
1977_nfs4_opendata_reclaim_to_nfs4_state(struct nfs4_opendata *data)
1978{
1979	struct inode *inode = data->state->inode;
1980	struct nfs4_state *state = data->state;
1981	int ret;
1982
1983	if (!data->rpc_done) {
1984		if (data->rpc_status)
1985			return ERR_PTR(data->rpc_status);
1986		return nfs4_try_open_cached(data);
1987	}
1988
1989	ret = nfs_refresh_inode(inode, &data->f_attr);
1990	if (ret)
1991		return ERR_PTR(ret);
1992
1993	if (data->o_res.delegation_type != 0)
1994		nfs4_opendata_check_deleg(data, state);
1995
1996	if (!update_open_stateid(state, &data->o_res.stateid,
1997				NULL, data->o_arg.fmode))
1998		return ERR_PTR(-EAGAIN);
1999	refcount_inc(&state->count);
2000
2001	return state;
2002}
2003
2004static struct inode *
2005nfs4_opendata_get_inode(struct nfs4_opendata *data)
2006{
2007	struct inode *inode;
2008
2009	switch (data->o_arg.claim) {
2010	case NFS4_OPEN_CLAIM_NULL:
2011	case NFS4_OPEN_CLAIM_DELEGATE_CUR:
2012	case NFS4_OPEN_CLAIM_DELEGATE_PREV:
2013		if (!(data->f_attr.valid & NFS_ATTR_FATTR))
2014			return ERR_PTR(-EAGAIN);
2015		inode = nfs_fhget(data->dir->d_sb, &data->o_res.fh,
2016				&data->f_attr);
2017		break;
2018	default:
2019		inode = d_inode(data->dentry);
2020		ihold(inode);
2021		nfs_refresh_inode(inode, &data->f_attr);
2022	}
2023	return inode;
2024}
2025
2026static struct nfs4_state *
2027nfs4_opendata_find_nfs4_state(struct nfs4_opendata *data)
2028{
2029	struct nfs4_state *state;
2030	struct inode *inode;
2031
2032	inode = nfs4_opendata_get_inode(data);
2033	if (IS_ERR(inode))
2034		return ERR_CAST(inode);
2035	if (data->state != NULL && data->state->inode == inode) {
2036		state = data->state;
2037		refcount_inc(&state->count);
2038	} else
2039		state = nfs4_get_open_state(inode, data->owner);
2040	iput(inode);
2041	if (state == NULL)
2042		state = ERR_PTR(-ENOMEM);
2043	return state;
2044}
2045
2046static struct nfs4_state *
2047_nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data)
2048{
2049	struct nfs4_state *state;
2050
2051	if (!data->rpc_done) {
2052		state = nfs4_try_open_cached(data);
2053		trace_nfs4_cached_open(data->state);
2054		goto out;
2055	}
2056
2057	state = nfs4_opendata_find_nfs4_state(data);
2058	if (IS_ERR(state))
2059		goto out;
2060
2061	if (data->o_res.delegation_type != 0)
2062		nfs4_opendata_check_deleg(data, state);
2063	if (!update_open_stateid(state, &data->o_res.stateid,
2064				NULL, data->o_arg.fmode)) {
2065		nfs4_put_open_state(state);
2066		state = ERR_PTR(-EAGAIN);
2067	}
2068out:
2069	nfs_release_seqid(data->o_arg.seqid);
2070	return state;
2071}
2072
2073static struct nfs4_state *
2074nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data)
2075{
2076	struct nfs4_state *ret;
2077
2078	if (data->o_arg.claim == NFS4_OPEN_CLAIM_PREVIOUS)
2079		ret =_nfs4_opendata_reclaim_to_nfs4_state(data);
2080	else
2081		ret = _nfs4_opendata_to_nfs4_state(data);
2082	nfs4_sequence_free_slot(&data->o_res.seq_res);
2083	return ret;
2084}
2085
2086static struct nfs_open_context *
2087nfs4_state_find_open_context_mode(struct nfs4_state *state, fmode_t mode)
2088{
2089	struct nfs_inode *nfsi = NFS_I(state->inode);
2090	struct nfs_open_context *ctx;
2091
2092	rcu_read_lock();
2093	list_for_each_entry_rcu(ctx, &nfsi->open_files, list) {
2094		if (ctx->state != state)
2095			continue;
2096		if ((ctx->mode & mode) != mode)
2097			continue;
2098		if (!get_nfs_open_context(ctx))
2099			continue;
2100		rcu_read_unlock();
2101		return ctx;
2102	}
2103	rcu_read_unlock();
2104	return ERR_PTR(-ENOENT);
2105}
2106
2107static struct nfs_open_context *
2108nfs4_state_find_open_context(struct nfs4_state *state)
2109{
2110	struct nfs_open_context *ctx;
2111
2112	ctx = nfs4_state_find_open_context_mode(state, FMODE_READ|FMODE_WRITE);
2113	if (!IS_ERR(ctx))
2114		return ctx;
2115	ctx = nfs4_state_find_open_context_mode(state, FMODE_WRITE);
2116	if (!IS_ERR(ctx))
2117		return ctx;
2118	return nfs4_state_find_open_context_mode(state, FMODE_READ);
2119}
2120
2121static struct nfs4_opendata *nfs4_open_recoverdata_alloc(struct nfs_open_context *ctx,
2122		struct nfs4_state *state, enum open_claim_type4 claim)
2123{
2124	struct nfs4_opendata *opendata;
2125
2126	opendata = nfs4_opendata_alloc(ctx->dentry, state->owner, 0, 0,
2127			NULL, claim, GFP_NOFS);
2128	if (opendata == NULL)
2129		return ERR_PTR(-ENOMEM);
2130	opendata->state = state;
2131	refcount_inc(&state->count);
2132	return opendata;
2133}
2134
2135static int nfs4_open_recover_helper(struct nfs4_opendata *opendata,
2136				    fmode_t fmode)
2137{
2138	struct nfs4_state *newstate;
2139	struct nfs_server *server = NFS_SB(opendata->dentry->d_sb);
2140	int openflags = opendata->o_arg.open_flags;
2141	int ret;
2142
2143	if (!nfs4_mode_match_open_stateid(opendata->state, fmode))
2144		return 0;
2145	opendata->o_arg.fmode = fmode;
2146	opendata->o_arg.share_access =
2147		nfs4_map_atomic_open_share(server, fmode, openflags);
2148	memset(&opendata->o_res, 0, sizeof(opendata->o_res));
2149	memset(&opendata->c_res, 0, sizeof(opendata->c_res));
2150	nfs4_init_opendata_res(opendata);
2151	ret = _nfs4_recover_proc_open(opendata);
2152	if (ret != 0)
2153		return ret;
2154	newstate = nfs4_opendata_to_nfs4_state(opendata);
2155	if (IS_ERR(newstate))
2156		return PTR_ERR(newstate);
2157	if (newstate != opendata->state)
2158		ret = -ESTALE;
2159	nfs4_close_state(newstate, fmode);
2160	return ret;
2161}
2162
2163static int nfs4_open_recover(struct nfs4_opendata *opendata, struct nfs4_state *state)
2164{
2165	int ret;
2166
2167	/* memory barrier prior to reading state->n_* */
2168	smp_rmb();
2169	ret = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE);
2170	if (ret != 0)
2171		return ret;
2172	ret = nfs4_open_recover_helper(opendata, FMODE_WRITE);
2173	if (ret != 0)
2174		return ret;
2175	ret = nfs4_open_recover_helper(opendata, FMODE_READ);
2176	if (ret != 0)
2177		return ret;
2178	/*
2179	 * We may have performed cached opens for all three recoveries.
2180	 * Check if we need to update the current stateid.
2181	 */
2182	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0 &&
2183	    !nfs4_stateid_match(&state->stateid, &state->open_stateid)) {
2184		write_seqlock(&state->seqlock);
2185		if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
2186			nfs4_stateid_copy(&state->stateid, &state->open_stateid);
2187		write_sequnlock(&state->seqlock);
2188	}
2189	return 0;
2190}
2191
2192/*
2193 * OPEN_RECLAIM:
2194 * 	reclaim state on the server after a reboot.
2195 */
2196static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
2197{
2198	struct nfs_delegation *delegation;
2199	struct nfs4_opendata *opendata;
2200	fmode_t delegation_type = 0;
2201	int status;
2202
2203	opendata = nfs4_open_recoverdata_alloc(ctx, state,
2204			NFS4_OPEN_CLAIM_PREVIOUS);
2205	if (IS_ERR(opendata))
2206		return PTR_ERR(opendata);
2207	rcu_read_lock();
2208	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
2209	if (delegation != NULL && test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) != 0)
2210		delegation_type = delegation->type;
2211	rcu_read_unlock();
2212	opendata->o_arg.u.delegation_type = delegation_type;
2213	status = nfs4_open_recover(opendata, state);
2214	nfs4_opendata_put(opendata);
2215	return status;
2216}
2217
2218static int nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
2219{
2220	struct nfs_server *server = NFS_SERVER(state->inode);
2221	struct nfs4_exception exception = { };
2222	int err;
2223	do {
2224		err = _nfs4_do_open_reclaim(ctx, state);
2225		trace_nfs4_open_reclaim(ctx, 0, err);
2226		if (nfs4_clear_cap_atomic_open_v1(server, err, &exception))
2227			continue;
2228		if (err != -NFS4ERR_DELAY)
2229			break;
2230		nfs4_handle_exception(server, err, &exception);
2231	} while (exception.retry);
2232	return err;
2233}
2234
2235static int nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *state)
2236{
2237	struct nfs_open_context *ctx;
2238	int ret;
2239
2240	ctx = nfs4_state_find_open_context(state);
2241	if (IS_ERR(ctx))
2242		return -EAGAIN;
2243	clear_bit(NFS_DELEGATED_STATE, &state->flags);
2244	nfs_state_clear_open_state_flags(state);
2245	ret = nfs4_do_open_reclaim(ctx, state);
2246	put_nfs_open_context(ctx);
2247	return ret;
2248}
2249
2250static int nfs4_handle_delegation_recall_error(struct nfs_server *server, struct nfs4_state *state, const nfs4_stateid *stateid, struct file_lock *fl, int err)
2251{
2252	switch (err) {
2253		default:
2254			printk(KERN_ERR "NFS: %s: unhandled error "
2255					"%d.\n", __func__, err);
2256			fallthrough;
2257		case 0:
2258		case -ENOENT:
2259		case -EAGAIN:
2260		case -ESTALE:
2261		case -ETIMEDOUT:
2262			break;
2263		case -NFS4ERR_BADSESSION:
2264		case -NFS4ERR_BADSLOT:
2265		case -NFS4ERR_BAD_HIGH_SLOT:
2266		case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
2267		case -NFS4ERR_DEADSESSION:
2268			return -EAGAIN;
2269		case -NFS4ERR_STALE_CLIENTID:
2270		case -NFS4ERR_STALE_STATEID:
2271			/* Don't recall a delegation if it was lost */
2272			nfs4_schedule_lease_recovery(server->nfs_client);
2273			return -EAGAIN;
2274		case -NFS4ERR_MOVED:
2275			nfs4_schedule_migration_recovery(server);
2276			return -EAGAIN;
2277		case -NFS4ERR_LEASE_MOVED:
2278			nfs4_schedule_lease_moved_recovery(server->nfs_client);
2279			return -EAGAIN;
2280		case -NFS4ERR_DELEG_REVOKED:
2281		case -NFS4ERR_ADMIN_REVOKED:
2282		case -NFS4ERR_EXPIRED:
2283		case -NFS4ERR_BAD_STATEID:
2284		case -NFS4ERR_OPENMODE:
2285			nfs_inode_find_state_and_recover(state->inode,
2286					stateid);
2287			nfs4_schedule_stateid_recovery(server, state);
2288			return -EAGAIN;
2289		case -NFS4ERR_DELAY:
2290		case -NFS4ERR_GRACE:
2291			ssleep(1);
2292			return -EAGAIN;
2293		case -ENOMEM:
2294		case -NFS4ERR_DENIED:
2295			if (fl) {
2296				struct nfs4_lock_state *lsp = fl->fl_u.nfs4_fl.owner;
2297				if (lsp)
2298					set_bit(NFS_LOCK_LOST, &lsp->ls_flags);
2299			}
2300			return 0;
2301	}
2302	return err;
2303}
2304
2305int nfs4_open_delegation_recall(struct nfs_open_context *ctx,
2306		struct nfs4_state *state, const nfs4_stateid *stateid)
2307{
2308	struct nfs_server *server = NFS_SERVER(state->inode);
2309	struct nfs4_opendata *opendata;
2310	int err = 0;
2311
2312	opendata = nfs4_open_recoverdata_alloc(ctx, state,
2313			NFS4_OPEN_CLAIM_DELEG_CUR_FH);
2314	if (IS_ERR(opendata))
2315		return PTR_ERR(opendata);
2316	nfs4_stateid_copy(&opendata->o_arg.u.delegation, stateid);
2317	if (!test_bit(NFS_O_RDWR_STATE, &state->flags)) {
2318		err = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE);
2319		if (err)
2320			goto out;
2321	}
2322	if (!test_bit(NFS_O_WRONLY_STATE, &state->flags)) {
2323		err = nfs4_open_recover_helper(opendata, FMODE_WRITE);
2324		if (err)
2325			goto out;
2326	}
2327	if (!test_bit(NFS_O_RDONLY_STATE, &state->flags)) {
2328		err = nfs4_open_recover_helper(opendata, FMODE_READ);
2329		if (err)
2330			goto out;
2331	}
2332	nfs_state_clear_delegation(state);
2333out:
2334	nfs4_opendata_put(opendata);
2335	return nfs4_handle_delegation_recall_error(server, state, stateid, NULL, err);
2336}
2337
2338static void nfs4_open_confirm_prepare(struct rpc_task *task, void *calldata)
2339{
2340	struct nfs4_opendata *data = calldata;
2341
2342	nfs4_setup_sequence(data->o_arg.server->nfs_client,
2343			   &data->c_arg.seq_args, &data->c_res.seq_res, task);
2344}
2345
2346static void nfs4_open_confirm_done(struct rpc_task *task, void *calldata)
2347{
2348	struct nfs4_opendata *data = calldata;
2349
2350	nfs40_sequence_done(task, &data->c_res.seq_res);
2351
2352	data->rpc_status = task->tk_status;
2353	if (data->rpc_status == 0) {
2354		nfs4_stateid_copy(&data->o_res.stateid, &data->c_res.stateid);
2355		nfs_confirm_seqid(&data->owner->so_seqid, 0);
2356		renew_lease(data->o_res.server, data->timestamp);
2357		data->rpc_done = true;
2358	}
2359}
2360
2361static void nfs4_open_confirm_release(void *calldata)
2362{
2363	struct nfs4_opendata *data = calldata;
2364	struct nfs4_state *state = NULL;
2365
2366	/* If this request hasn't been cancelled, do nothing */
2367	if (!data->cancelled)
2368		goto out_free;
2369	/* In case of error, no cleanup! */
2370	if (!data->rpc_done)
2371		goto out_free;
2372	state = nfs4_opendata_to_nfs4_state(data);
2373	if (!IS_ERR(state))
2374		nfs4_close_state(state, data->o_arg.fmode);
2375out_free:
2376	nfs4_opendata_put(data);
2377}
2378
2379static const struct rpc_call_ops nfs4_open_confirm_ops = {
2380	.rpc_call_prepare = nfs4_open_confirm_prepare,
2381	.rpc_call_done = nfs4_open_confirm_done,
2382	.rpc_release = nfs4_open_confirm_release,
2383};
2384
2385/*
2386 * Note: On error, nfs4_proc_open_confirm will free the struct nfs4_opendata
2387 */
2388static int _nfs4_proc_open_confirm(struct nfs4_opendata *data)
2389{
2390	struct nfs_server *server = NFS_SERVER(d_inode(data->dir));
2391	struct rpc_task *task;
2392	struct  rpc_message msg = {
2393		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_CONFIRM],
2394		.rpc_argp = &data->c_arg,
2395		.rpc_resp = &data->c_res,
2396		.rpc_cred = data->owner->so_cred,
2397	};
2398	struct rpc_task_setup task_setup_data = {
2399		.rpc_client = server->client,
2400		.rpc_message = &msg,
2401		.callback_ops = &nfs4_open_confirm_ops,
2402		.callback_data = data,
2403		.workqueue = nfsiod_workqueue,
2404		.flags = RPC_TASK_ASYNC | RPC_TASK_CRED_NOREF,
2405	};
2406	int status;
2407
2408	nfs4_init_sequence(&data->c_arg.seq_args, &data->c_res.seq_res, 1,
2409				data->is_recover);
2410	kref_get(&data->kref);
2411	data->rpc_done = false;
2412	data->rpc_status = 0;
2413	data->timestamp = jiffies;
2414	task = rpc_run_task(&task_setup_data);
2415	if (IS_ERR(task))
2416		return PTR_ERR(task);
2417	status = rpc_wait_for_completion_task(task);
2418	if (status != 0) {
2419		data->cancelled = true;
2420		smp_wmb();
2421	} else
2422		status = data->rpc_status;
2423	rpc_put_task(task);
2424	return status;
2425}
2426
2427static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
2428{
2429	struct nfs4_opendata *data = calldata;
2430	struct nfs4_state_owner *sp = data->owner;
2431	struct nfs_client *clp = sp->so_server->nfs_client;
2432	enum open_claim_type4 claim = data->o_arg.claim;
2433
2434	if (nfs_wait_on_sequence(data->o_arg.seqid, task) != 0)
2435		goto out_wait;
2436	/*
2437	 * Check if we still need to send an OPEN call, or if we can use
2438	 * a delegation instead.
2439	 */
2440	if (data->state != NULL) {
2441		struct nfs_delegation *delegation;
2442
2443		if (can_open_cached(data->state, data->o_arg.fmode,
2444					data->o_arg.open_flags, claim))
2445			goto out_no_action;
2446		rcu_read_lock();
2447		delegation = nfs4_get_valid_delegation(data->state->inode);
2448		if (can_open_delegated(delegation, data->o_arg.fmode, claim))
2449			goto unlock_no_action;
2450		rcu_read_unlock();
2451	}
2452	/* Update client id. */
2453	data->o_arg.clientid = clp->cl_clientid;
2454	switch (claim) {
2455	default:
2456		break;
2457	case NFS4_OPEN_CLAIM_PREVIOUS:
2458	case NFS4_OPEN_CLAIM_DELEG_CUR_FH:
2459	case NFS4_OPEN_CLAIM_DELEG_PREV_FH:
2460		data->o_arg.open_bitmap = &nfs4_open_noattr_bitmap[0];
2461		fallthrough;
2462	case NFS4_OPEN_CLAIM_FH:
2463		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR];
2464	}
2465	data->timestamp = jiffies;
2466	if (nfs4_setup_sequence(data->o_arg.server->nfs_client,
2467				&data->o_arg.seq_args,
2468				&data->o_res.seq_res,
2469				task) != 0)
2470		nfs_release_seqid(data->o_arg.seqid);
2471
2472	/* Set the create mode (note dependency on the session type) */
2473	data->o_arg.createmode = NFS4_CREATE_UNCHECKED;
2474	if (data->o_arg.open_flags & O_EXCL) {
2475		data->o_arg.createmode = NFS4_CREATE_EXCLUSIVE4_1;
2476		if (clp->cl_mvops->minor_version == 0) {
2477			data->o_arg.createmode = NFS4_CREATE_EXCLUSIVE;
2478			/* don't put an ACCESS op in OPEN compound if O_EXCL,
2479			 * because ACCESS will return permission denied for
2480			 * all bits until close */
2481			data->o_res.access_request = data->o_arg.access = 0;
2482		} else if (nfs4_has_persistent_session(clp))
2483			data->o_arg.createmode = NFS4_CREATE_GUARDED;
2484	}
2485	return;
2486unlock_no_action:
2487	trace_nfs4_cached_open(data->state);
2488	rcu_read_unlock();
2489out_no_action:
2490	task->tk_action = NULL;
2491out_wait:
2492	nfs4_sequence_done(task, &data->o_res.seq_res);
2493}
2494
2495static void nfs4_open_done(struct rpc_task *task, void *calldata)
2496{
2497	struct nfs4_opendata *data = calldata;
2498
2499	data->rpc_status = task->tk_status;
2500
2501	if (!nfs4_sequence_process(task, &data->o_res.seq_res))
2502		return;
2503
2504	if (task->tk_status == 0) {
2505		if (data->o_res.f_attr->valid & NFS_ATTR_FATTR_TYPE) {
2506			switch (data->o_res.f_attr->mode & S_IFMT) {
2507			case S_IFREG:
2508				break;
2509			case S_IFLNK:
2510				data->rpc_status = -ELOOP;
2511				break;
2512			case S_IFDIR:
2513				data->rpc_status = -EISDIR;
2514				break;
2515			default:
2516				data->rpc_status = -ENOTDIR;
2517			}
2518		}
2519		renew_lease(data->o_res.server, data->timestamp);
2520		if (!(data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM))
2521			nfs_confirm_seqid(&data->owner->so_seqid, 0);
2522	}
2523	data->rpc_done = true;
2524}
2525
2526static void nfs4_open_release(void *calldata)
2527{
2528	struct nfs4_opendata *data = calldata;
2529	struct nfs4_state *state = NULL;
2530
2531	/* If this request hasn't been cancelled, do nothing */
2532	if (!data->cancelled)
2533		goto out_free;
2534	/* In case of error, no cleanup! */
2535	if (data->rpc_status != 0 || !data->rpc_done)
2536		goto out_free;
2537	/* In case we need an open_confirm, no cleanup! */
2538	if (data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM)
2539		goto out_free;
2540	state = nfs4_opendata_to_nfs4_state(data);
2541	if (!IS_ERR(state))
2542		nfs4_close_state(state, data->o_arg.fmode);
2543out_free:
2544	nfs4_opendata_put(data);
2545}
2546
2547static const struct rpc_call_ops nfs4_open_ops = {
2548	.rpc_call_prepare = nfs4_open_prepare,
2549	.rpc_call_done = nfs4_open_done,
2550	.rpc_release = nfs4_open_release,
2551};
2552
2553static int nfs4_run_open_task(struct nfs4_opendata *data,
2554			      struct nfs_open_context *ctx)
2555{
2556	struct inode *dir = d_inode(data->dir);
2557	struct nfs_server *server = NFS_SERVER(dir);
2558	struct nfs_openargs *o_arg = &data->o_arg;
2559	struct nfs_openres *o_res = &data->o_res;
2560	struct rpc_task *task;
2561	struct rpc_message msg = {
2562		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN],
2563		.rpc_argp = o_arg,
2564		.rpc_resp = o_res,
2565		.rpc_cred = data->owner->so_cred,
2566	};
2567	struct rpc_task_setup task_setup_data = {
2568		.rpc_client = server->client,
2569		.rpc_message = &msg,
2570		.callback_ops = &nfs4_open_ops,
2571		.callback_data = data,
2572		.workqueue = nfsiod_workqueue,
2573		.flags = RPC_TASK_ASYNC | RPC_TASK_CRED_NOREF,
2574	};
2575	int status;
2576
2577	if (nfs_server_capable(dir, NFS_CAP_MOVEABLE))
2578		task_setup_data.flags |= RPC_TASK_MOVEABLE;
2579
2580	kref_get(&data->kref);
2581	data->rpc_done = false;
2582	data->rpc_status = 0;
2583	data->cancelled = false;
2584	data->is_recover = false;
2585	if (!ctx) {
2586		nfs4_init_sequence(&o_arg->seq_args, &o_res->seq_res, 1, 1);
2587		data->is_recover = true;
2588		task_setup_data.flags |= RPC_TASK_TIMEOUT;
2589	} else {
2590		nfs4_init_sequence(&o_arg->seq_args, &o_res->seq_res, 1, 0);
2591		pnfs_lgopen_prepare(data, ctx);
2592	}
2593	task = rpc_run_task(&task_setup_data);
2594	if (IS_ERR(task))
2595		return PTR_ERR(task);
2596	status = rpc_wait_for_completion_task(task);
2597	if (status != 0) {
2598		data->cancelled = true;
2599		smp_wmb();
2600	} else
2601		status = data->rpc_status;
2602	rpc_put_task(task);
2603
2604	return status;
2605}
2606
2607static int _nfs4_recover_proc_open(struct nfs4_opendata *data)
2608{
2609	struct inode *dir = d_inode(data->dir);
2610	struct nfs_openres *o_res = &data->o_res;
2611	int status;
2612
2613	status = nfs4_run_open_task(data, NULL);
2614	if (status != 0 || !data->rpc_done)
2615		return status;
2616
2617	nfs_fattr_map_and_free_names(NFS_SERVER(dir), &data->f_attr);
2618
2619	if (o_res->rflags & NFS4_OPEN_RESULT_CONFIRM)
2620		status = _nfs4_proc_open_confirm(data);
2621
2622	return status;
2623}
2624
2625/*
2626 * Additional permission checks in order to distinguish between an
2627 * open for read, and an open for execute. This works around the
2628 * fact that NFSv4 OPEN treats read and execute permissions as being
2629 * the same.
2630 * Note that in the non-execute case, we want to turn off permission
2631 * checking if we just created a new file (POSIX open() semantics).
2632 */
2633static int nfs4_opendata_access(const struct cred *cred,
2634				struct nfs4_opendata *opendata,
2635				struct nfs4_state *state, fmode_t fmode)
2636{
2637	struct nfs_access_entry cache;
2638	u32 mask, flags;
2639
2640	/* access call failed or for some reason the server doesn't
2641	 * support any access modes -- defer access call until later */
2642	if (opendata->o_res.access_supported == 0)
2643		return 0;
2644
2645	mask = 0;
2646	if (fmode & FMODE_EXEC) {
2647		/* ONLY check for exec rights */
2648		if (S_ISDIR(state->inode->i_mode))
2649			mask = NFS4_ACCESS_LOOKUP;
2650		else
2651			mask = NFS4_ACCESS_EXECUTE;
2652	} else if ((fmode & FMODE_READ) && !opendata->file_created)
2653		mask = NFS4_ACCESS_READ;
2654
2655	nfs_access_set_mask(&cache, opendata->o_res.access_result);
2656	nfs_access_add_cache(state->inode, &cache, cred);
2657
2658	flags = NFS4_ACCESS_READ | NFS4_ACCESS_EXECUTE | NFS4_ACCESS_LOOKUP;
2659	if ((mask & ~cache.mask & flags) == 0)
2660		return 0;
2661
2662	return -EACCES;
2663}
2664
2665/*
2666 * Note: On error, nfs4_proc_open will free the struct nfs4_opendata
2667 */
2668static int _nfs4_proc_open(struct nfs4_opendata *data,
2669			   struct nfs_open_context *ctx)
2670{
2671	struct inode *dir = d_inode(data->dir);
2672	struct nfs_server *server = NFS_SERVER(dir);
2673	struct nfs_openargs *o_arg = &data->o_arg;
2674	struct nfs_openres *o_res = &data->o_res;
2675	int status;
2676
2677	status = nfs4_run_open_task(data, ctx);
2678	if (!data->rpc_done)
2679		return status;
2680	if (status != 0) {
2681		if (status == -NFS4ERR_BADNAME &&
2682				!(o_arg->open_flags & O_CREAT))
2683			return -ENOENT;
2684		return status;
2685	}
2686
2687	nfs_fattr_map_and_free_names(server, &data->f_attr);
2688
2689	if (o_arg->open_flags & O_CREAT) {
2690		if (o_arg->open_flags & O_EXCL)
2691			data->file_created = true;
2692		else if (o_res->cinfo.before != o_res->cinfo.after)
2693			data->file_created = true;
2694		if (data->file_created ||
2695		    inode_peek_iversion_raw(dir) != o_res->cinfo.after)
2696			nfs4_update_changeattr(dir, &o_res->cinfo,
2697					o_res->f_attr->time_start,
2698					NFS_INO_INVALID_DATA);
2699	}
2700	if ((o_res->rflags & NFS4_OPEN_RESULT_LOCKTYPE_POSIX) == 0)
2701		server->caps &= ~NFS_CAP_POSIX_LOCK;
2702	if(o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
2703		status = _nfs4_proc_open_confirm(data);
2704		if (status != 0)
2705			return status;
2706	}
2707	if (!(o_res->f_attr->valid & NFS_ATTR_FATTR)) {
2708		struct nfs_fh *fh = &o_res->fh;
2709
2710		nfs4_sequence_free_slot(&o_res->seq_res);
2711		if (o_arg->claim == NFS4_OPEN_CLAIM_FH)
2712			fh = NFS_FH(d_inode(data->dentry));
2713		nfs4_proc_getattr(server, fh, o_res->f_attr, NULL);
2714	}
2715	return 0;
2716}
2717
2718/*
2719 * OPEN_EXPIRED:
2720 * 	reclaim state on the server after a network partition.
2721 * 	Assumes caller holds the appropriate lock
2722 */
2723static int _nfs4_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
2724{
2725	struct nfs4_opendata *opendata;
2726	int ret;
2727
2728	opendata = nfs4_open_recoverdata_alloc(ctx, state, NFS4_OPEN_CLAIM_FH);
2729	if (IS_ERR(opendata))
2730		return PTR_ERR(opendata);
2731	/*
2732	 * We're not recovering a delegation, so ask for no delegation.
2733	 * Otherwise the recovery thread could deadlock with an outstanding
2734	 * delegation return.
2735	 */
2736	opendata->o_arg.open_flags = O_DIRECT;
2737	ret = nfs4_open_recover(opendata, state);
2738	if (ret == -ESTALE)
2739		d_drop(ctx->dentry);
2740	nfs4_opendata_put(opendata);
2741	return ret;
2742}
2743
2744static int nfs4_do_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
2745{
2746	struct nfs_server *server = NFS_SERVER(state->inode);
2747	struct nfs4_exception exception = { };
2748	int err;
2749
2750	do {
2751		err = _nfs4_open_expired(ctx, state);
2752		trace_nfs4_open_expired(ctx, 0, err);
2753		if (nfs4_clear_cap_atomic_open_v1(server, err, &exception))
2754			continue;
2755		switch (err) {
2756		default:
2757			goto out;
2758		case -NFS4ERR_GRACE:
2759		case -NFS4ERR_DELAY:
2760			nfs4_handle_exception(server, err, &exception);
2761			err = 0;
2762		}
2763	} while (exception.retry);
2764out:
2765	return err;
2766}
2767
2768static int nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
2769{
2770	struct nfs_open_context *ctx;
2771	int ret;
2772
2773	ctx = nfs4_state_find_open_context(state);
2774	if (IS_ERR(ctx))
2775		return -EAGAIN;
2776	ret = nfs4_do_open_expired(ctx, state);
2777	put_nfs_open_context(ctx);
2778	return ret;
2779}
2780
2781static void nfs_finish_clear_delegation_stateid(struct nfs4_state *state,
2782		const nfs4_stateid *stateid)
2783{
2784	nfs_remove_bad_delegation(state->inode, stateid);
2785	nfs_state_clear_delegation(state);
2786}
2787
2788static void nfs40_clear_delegation_stateid(struct nfs4_state *state)
2789{
2790	if (rcu_access_pointer(NFS_I(state->inode)->delegation) != NULL)
2791		nfs_finish_clear_delegation_stateid(state, NULL);
2792}
2793
2794static int nfs40_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
2795{
2796	/* NFSv4.0 doesn't allow for delegation recovery on open expire */
2797	nfs40_clear_delegation_stateid(state);
2798	nfs_state_clear_open_state_flags(state);
2799	return nfs4_open_expired(sp, state);
2800}
2801
2802static int nfs40_test_and_free_expired_stateid(struct nfs_server *server,
2803		nfs4_stateid *stateid,
2804		const struct cred *cred)
2805{
2806	return -NFS4ERR_BAD_STATEID;
2807}
2808
2809#if defined(CONFIG_NFS_V4_1)
2810static int nfs41_test_and_free_expired_stateid(struct nfs_server *server,
2811		nfs4_stateid *stateid,
2812		const struct cred *cred)
2813{
2814	int status;
2815
2816	switch (stateid->type) {
2817	default:
2818		break;
2819	case NFS4_INVALID_STATEID_TYPE:
2820	case NFS4_SPECIAL_STATEID_TYPE:
2821		return -NFS4ERR_BAD_STATEID;
2822	case NFS4_REVOKED_STATEID_TYPE:
2823		goto out_free;
2824	}
2825
2826	status = nfs41_test_stateid(server, stateid, cred);
2827	switch (status) {
2828	case -NFS4ERR_EXPIRED:
2829	case -NFS4ERR_ADMIN_REVOKED:
2830	case -NFS4ERR_DELEG_REVOKED:
2831		break;
2832	default:
2833		return status;
2834	}
2835out_free:
2836	/* Ack the revoked state to the server */
2837	nfs41_free_stateid(server, stateid, cred, true);
2838	return -NFS4ERR_EXPIRED;
2839}
2840
2841static int nfs41_check_delegation_stateid(struct nfs4_state *state)
2842{
2843	struct nfs_server *server = NFS_SERVER(state->inode);
2844	nfs4_stateid stateid;
2845	struct nfs_delegation *delegation;
2846	const struct cred *cred = NULL;
2847	int status, ret = NFS_OK;
2848
2849	/* Get the delegation credential for use by test/free_stateid */
2850	rcu_read_lock();
2851	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
2852	if (delegation == NULL) {
2853		rcu_read_unlock();
2854		nfs_state_clear_delegation(state);
2855		return NFS_OK;
2856	}
2857
2858	spin_lock(&delegation->lock);
2859	nfs4_stateid_copy(&stateid, &delegation->stateid);
2860
2861	if (!test_and_clear_bit(NFS_DELEGATION_TEST_EXPIRED,
2862				&delegation->flags)) {
2863		spin_unlock(&delegation->lock);
2864		rcu_read_unlock();
2865		return NFS_OK;
2866	}
2867
2868	if (delegation->cred)
2869		cred = get_cred(delegation->cred);
2870	spin_unlock(&delegation->lock);
2871	rcu_read_unlock();
2872	status = nfs41_test_and_free_expired_stateid(server, &stateid, cred);
2873	trace_nfs4_test_delegation_stateid(state, NULL, status);
2874	if (status == -NFS4ERR_EXPIRED || status == -NFS4ERR_BAD_STATEID)
2875		nfs_finish_clear_delegation_stateid(state, &stateid);
2876	else
2877		ret = status;
2878
2879	put_cred(cred);
2880	return ret;
2881}
2882
2883static void nfs41_delegation_recover_stateid(struct nfs4_state *state)
2884{
2885	nfs4_stateid tmp;
2886
2887	if (test_bit(NFS_DELEGATED_STATE, &state->flags) &&
2888	    nfs4_copy_delegation_stateid(state->inode, state->state,
2889				&tmp, NULL) &&
2890	    nfs4_stateid_match_other(&state->stateid, &tmp))
2891		nfs_state_set_delegation(state, &tmp, state->state);
2892	else
2893		nfs_state_clear_delegation(state);
2894}
2895
2896/**
2897 * nfs41_check_expired_locks - possibly free a lock stateid
2898 *
2899 * @state: NFSv4 state for an inode
2900 *
2901 * Returns NFS_OK if recovery for this stateid is now finished.
2902 * Otherwise a negative NFS4ERR value is returned.
2903 */
2904static int nfs41_check_expired_locks(struct nfs4_state *state)
2905{
2906	int status, ret = NFS_OK;
2907	struct nfs4_lock_state *lsp, *prev = NULL;
2908	struct nfs_server *server = NFS_SERVER(state->inode);
2909
2910	if (!test_bit(LK_STATE_IN_USE, &state->flags))
2911		goto out;
2912
2913	spin_lock(&state->state_lock);
2914	list_for_each_entry(lsp, &state->lock_states, ls_locks) {
2915		if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags)) {
2916			const struct cred *cred = lsp->ls_state->owner->so_cred;
2917
2918			refcount_inc(&lsp->ls_count);
2919			spin_unlock(&state->state_lock);
2920
2921			nfs4_put_lock_state(prev);
2922			prev = lsp;
2923
2924			status = nfs41_test_and_free_expired_stateid(server,
2925					&lsp->ls_stateid,
2926					cred);
2927			trace_nfs4_test_lock_stateid(state, lsp, status);
2928			if (status == -NFS4ERR_EXPIRED ||
2929			    status == -NFS4ERR_BAD_STATEID) {
2930				clear_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags);
2931				lsp->ls_stateid.type = NFS4_INVALID_STATEID_TYPE;
2932				if (!recover_lost_locks)
2933					set_bit(NFS_LOCK_LOST, &lsp->ls_flags);
2934			} else if (status != NFS_OK) {
2935				ret = status;
2936				nfs4_put_lock_state(prev);
2937				goto out;
2938			}
2939			spin_lock(&state->state_lock);
2940		}
2941	}
2942	spin_unlock(&state->state_lock);
2943	nfs4_put_lock_state(prev);
2944out:
2945	return ret;
2946}
2947
2948/**
2949 * nfs41_check_open_stateid - possibly free an open stateid
2950 *
2951 * @state: NFSv4 state for an inode
2952 *
2953 * Returns NFS_OK if recovery for this stateid is now finished.
2954 * Otherwise a negative NFS4ERR value is returned.
2955 */
2956static int nfs41_check_open_stateid(struct nfs4_state *state)
2957{
2958	struct nfs_server *server = NFS_SERVER(state->inode);
2959	nfs4_stateid *stateid = &state->open_stateid;
2960	const struct cred *cred = state->owner->so_cred;
2961	int status;
2962
2963	if (test_bit(NFS_OPEN_STATE, &state->flags) == 0)
2964		return -NFS4ERR_BAD_STATEID;
2965	status = nfs41_test_and_free_expired_stateid(server, stateid, cred);
2966	trace_nfs4_test_open_stateid(state, NULL, status);
2967	if (status == -NFS4ERR_EXPIRED || status == -NFS4ERR_BAD_STATEID) {
2968		nfs_state_clear_open_state_flags(state);
2969		stateid->type = NFS4_INVALID_STATEID_TYPE;
2970		return status;
2971	}
2972	if (nfs_open_stateid_recover_openmode(state))
2973		return -NFS4ERR_OPENMODE;
2974	return NFS_OK;
2975}
2976
2977static int nfs41_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
2978{
2979	int status;
2980
2981	status = nfs41_check_delegation_stateid(state);
2982	if (status != NFS_OK)
2983		return status;
2984	nfs41_delegation_recover_stateid(state);
2985
2986	status = nfs41_check_expired_locks(state);
2987	if (status != NFS_OK)
2988		return status;
2989	status = nfs41_check_open_stateid(state);
2990	if (status != NFS_OK)
2991		status = nfs4_open_expired(sp, state);
2992	return status;
2993}
2994#endif
2995
2996/*
2997 * on an EXCLUSIVE create, the server should send back a bitmask with FATTR4-*
2998 * fields corresponding to attributes that were used to store the verifier.
2999 * Make sure we clobber those fields in the later setattr call
3000 */
3001static unsigned nfs4_exclusive_attrset(struct nfs4_opendata *opendata,
3002				struct iattr *sattr, struct nfs4_label **label)
3003{
3004	const __u32 *bitmask = opendata->o_arg.server->exclcreat_bitmask;
3005	__u32 attrset[3];
3006	unsigned ret;
3007	unsigned i;
3008
3009	for (i = 0; i < ARRAY_SIZE(attrset); i++) {
3010		attrset[i] = opendata->o_res.attrset[i];
3011		if (opendata->o_arg.createmode == NFS4_CREATE_EXCLUSIVE4_1)
3012			attrset[i] &= ~bitmask[i];
3013	}
3014
3015	ret = (opendata->o_arg.createmode == NFS4_CREATE_EXCLUSIVE) ?
3016		sattr->ia_valid : 0;
3017
3018	if ((attrset[1] & (FATTR4_WORD1_TIME_ACCESS|FATTR4_WORD1_TIME_ACCESS_SET))) {
3019		if (sattr->ia_valid & ATTR_ATIME_SET)
3020			ret |= ATTR_ATIME_SET;
3021		else
3022			ret |= ATTR_ATIME;
3023	}
3024
3025	if ((attrset[1] & (FATTR4_WORD1_TIME_MODIFY|FATTR4_WORD1_TIME_MODIFY_SET))) {
3026		if (sattr->ia_valid & ATTR_MTIME_SET)
3027			ret |= ATTR_MTIME_SET;
3028		else
3029			ret |= ATTR_MTIME;
3030	}
3031
3032	if (!(attrset[2] & FATTR4_WORD2_SECURITY_LABEL))
3033		*label = NULL;
3034	return ret;
3035}
3036
3037static int _nfs4_open_and_get_state(struct nfs4_opendata *opendata,
3038		struct nfs_open_context *ctx)
3039{
3040	struct nfs4_state_owner *sp = opendata->owner;
3041	struct nfs_server *server = sp->so_server;
3042	struct dentry *dentry;
3043	struct nfs4_state *state;
3044	fmode_t acc_mode = _nfs4_ctx_to_accessmode(ctx);
3045	struct inode *dir = d_inode(opendata->dir);
3046	unsigned long dir_verifier;
3047	unsigned int seq;
3048	int ret;
3049
3050	seq = raw_seqcount_begin(&sp->so_reclaim_seqcount);
3051	dir_verifier = nfs_save_change_attribute(dir);
3052
3053	ret = _nfs4_proc_open(opendata, ctx);
3054	if (ret != 0)
3055		goto out;
3056
3057	state = _nfs4_opendata_to_nfs4_state(opendata);
3058	ret = PTR_ERR(state);
3059	if (IS_ERR(state))
3060		goto out;
3061	ctx->state = state;
3062	if (server->caps & NFS_CAP_POSIX_LOCK)
3063		set_bit(NFS_STATE_POSIX_LOCKS, &state->flags);
3064	if (opendata->o_res.rflags & NFS4_OPEN_RESULT_MAY_NOTIFY_LOCK)
3065		set_bit(NFS_STATE_MAY_NOTIFY_LOCK, &state->flags);
3066	if (opendata->o_res.rflags & NFS4_OPEN_RESULT_PRESERVE_UNLINKED)
3067		set_bit(NFS_INO_PRESERVE_UNLINKED, &NFS_I(state->inode)->flags);
3068
3069	dentry = opendata->dentry;
3070	if (d_really_is_negative(dentry)) {
3071		struct dentry *alias;
3072		d_drop(dentry);
3073		alias = d_exact_alias(dentry, state->inode);
3074		if (!alias)
3075			alias = d_splice_alias(igrab(state->inode), dentry);
3076		/* d_splice_alias() can't fail here - it's a non-directory */
3077		if (alias) {
3078			dput(ctx->dentry);
3079			ctx->dentry = dentry = alias;
3080		}
3081	}
3082
3083	switch(opendata->o_arg.claim) {
3084	default:
3085		break;
3086	case NFS4_OPEN_CLAIM_NULL:
3087	case NFS4_OPEN_CLAIM_DELEGATE_CUR:
3088	case NFS4_OPEN_CLAIM_DELEGATE_PREV:
3089		if (!opendata->rpc_done)
3090			break;
3091		if (opendata->o_res.delegation_type != 0)
3092			dir_verifier = nfs_save_change_attribute(dir);
3093		nfs_set_verifier(dentry, dir_verifier);
3094	}
3095
3096	/* Parse layoutget results before we check for access */
3097	pnfs_parse_lgopen(state->inode, opendata->lgp, ctx);
3098
3099	ret = nfs4_opendata_access(sp->so_cred, opendata, state, acc_mode);
3100	if (ret != 0)
3101		goto out;
3102
3103	if (d_inode(dentry) == state->inode) {
3104		nfs_inode_attach_open_context(ctx);
3105		if (read_seqcount_retry(&sp->so_reclaim_seqcount, seq))
3106			nfs4_schedule_stateid_recovery(server, state);
3107	}
3108
3109out:
3110	if (!opendata->cancelled) {
3111		if (opendata->lgp) {
3112			nfs4_lgopen_release(opendata->lgp);
3113			opendata->lgp = NULL;
3114		}
3115		nfs4_sequence_free_slot(&opendata->o_res.seq_res);
3116	}
3117	return ret;
3118}
3119
3120/*
3121 * Returns a referenced nfs4_state
3122 */
3123static int _nfs4_do_open(struct inode *dir,
3124			struct nfs_open_context *ctx,
3125			int flags,
3126			const struct nfs4_open_createattrs *c,
3127			int *opened)
3128{
3129	struct nfs4_state_owner  *sp;
3130	struct nfs4_state     *state = NULL;
3131	struct nfs_server       *server = NFS_SERVER(dir);
3132	struct nfs4_opendata *opendata;
3133	struct dentry *dentry = ctx->dentry;
3134	const struct cred *cred = ctx->cred;
3135	struct nfs4_threshold **ctx_th = &ctx->mdsthreshold;
3136	fmode_t fmode = _nfs4_ctx_to_openmode(ctx);
3137	enum open_claim_type4 claim = NFS4_OPEN_CLAIM_NULL;
3138	struct iattr *sattr = c->sattr;
3139	struct nfs4_label *label = c->label;
3140	int status;
3141
3142	/* Protect against reboot recovery conflicts */
3143	status = -ENOMEM;
3144	sp = nfs4_get_state_owner(server, cred, GFP_KERNEL);
3145	if (sp == NULL) {
3146		dprintk("nfs4_do_open: nfs4_get_state_owner failed!\n");
3147		goto out_err;
3148	}
3149	status = nfs4_client_recover_expired_lease(server->nfs_client);
3150	if (status != 0)
3151		goto err_put_state_owner;
3152	if (d_really_is_positive(dentry))
3153		nfs4_return_incompatible_delegation(d_inode(dentry), fmode);
3154	status = -ENOMEM;
3155	if (d_really_is_positive(dentry))
3156		claim = NFS4_OPEN_CLAIM_FH;
3157	opendata = nfs4_opendata_alloc(dentry, sp, fmode, flags,
3158			c, claim, GFP_KERNEL);
3159	if (opendata == NULL)
3160		goto err_put_state_owner;
3161
3162	if (server->attr_bitmask[2] & FATTR4_WORD2_MDSTHRESHOLD) {
3163		if (!opendata->f_attr.mdsthreshold) {
3164			opendata->f_attr.mdsthreshold = pnfs_mdsthreshold_alloc();
3165			if (!opendata->f_attr.mdsthreshold)
3166				goto err_opendata_put;
3167		}
3168		opendata->o_arg.open_bitmap = &nfs4_pnfs_open_bitmap[0];
3169	}
3170	if (d_really_is_positive(dentry))
3171		opendata->state = nfs4_get_open_state(d_inode(dentry), sp);
3172
3173	status = _nfs4_open_and_get_state(opendata, ctx);
3174	if (status != 0)
3175		goto err_opendata_put;
3176	state = ctx->state;
3177
3178	if ((opendata->o_arg.open_flags & (O_CREAT|O_EXCL)) == (O_CREAT|O_EXCL) &&
3179	    (opendata->o_arg.createmode != NFS4_CREATE_GUARDED)) {
3180		unsigned attrs = nfs4_exclusive_attrset(opendata, sattr, &label);
3181		/*
3182		 * send create attributes which was not set by open
3183		 * with an extra setattr.
3184		 */
3185		if (attrs || label) {
3186			unsigned ia_old = sattr->ia_valid;
3187
3188			sattr->ia_valid = attrs;
3189			nfs_fattr_init(opendata->o_res.f_attr);
3190			status = nfs4_do_setattr(state->inode, cred,
3191					opendata->o_res.f_attr, sattr,
3192					ctx, label);
3193			if (status == 0) {
3194				nfs_setattr_update_inode(state->inode, sattr,
3195						opendata->o_res.f_attr);
3196				nfs_setsecurity(state->inode, opendata->o_res.f_attr);
3197			}
3198			sattr->ia_valid = ia_old;
3199		}
3200	}
3201	if (opened && opendata->file_created)
3202		*opened = 1;
3203
3204	if (pnfs_use_threshold(ctx_th, opendata->f_attr.mdsthreshold, server)) {
3205		*ctx_th = opendata->f_attr.mdsthreshold;
3206		opendata->f_attr.mdsthreshold = NULL;
3207	}
3208
3209	nfs4_opendata_put(opendata);
3210	nfs4_put_state_owner(sp);
3211	return 0;
3212err_opendata_put:
3213	nfs4_opendata_put(opendata);
3214err_put_state_owner:
3215	nfs4_put_state_owner(sp);
3216out_err:
3217	return status;
3218}
3219
3220
3221static struct nfs4_state *nfs4_do_open(struct inode *dir,
3222					struct nfs_open_context *ctx,
3223					int flags,
3224					struct iattr *sattr,
3225					struct nfs4_label *label,
3226					int *opened)
3227{
3228	struct nfs_server *server = NFS_SERVER(dir);
3229	struct nfs4_exception exception = {
3230		.interruptible = true,
3231	};
3232	struct nfs4_state *res;
3233	struct nfs4_open_createattrs c = {
3234		.label = label,
3235		.sattr = sattr,
3236		.verf = {
3237			[0] = (__u32)jiffies,
3238			[1] = (__u32)current->pid,
3239		},
3240	};
3241	int status;
3242
3243	do {
3244		status = _nfs4_do_open(dir, ctx, flags, &c, opened);
3245		res = ctx->state;
3246		trace_nfs4_open_file(ctx, flags, status);
3247		if (status == 0)
3248			break;
3249		/* NOTE: BAD_SEQID means the server and client disagree about the
3250		 * book-keeping w.r.t. state-changing operations
3251		 * (OPEN/CLOSE/LOCK/LOCKU...)
3252		 * It is actually a sign of a bug on the client or on the server.
3253		 *
3254		 * If we receive a BAD_SEQID error in the particular case of
3255		 * doing an OPEN, we assume that nfs_increment_open_seqid() will
3256		 * have unhashed the old state_owner for us, and that we can
3257		 * therefore safely retry using a new one. We should still warn
3258		 * the user though...
3259		 */
3260		if (status == -NFS4ERR_BAD_SEQID) {
3261			pr_warn_ratelimited("NFS: v4 server %s "
3262					" returned a bad sequence-id error!\n",
3263					NFS_SERVER(dir)->nfs_client->cl_hostname);
3264			exception.retry = 1;
3265			continue;
3266		}
3267		/*
3268		 * BAD_STATEID on OPEN means that the server cancelled our
3269		 * state before it received the OPEN_CONFIRM.
3270		 * Recover by retrying the request as per the discussion
3271		 * on Page 181 of RFC3530.
3272		 */
3273		if (status == -NFS4ERR_BAD_STATEID) {
3274			exception.retry = 1;
3275			continue;
3276		}
3277		if (status == -NFS4ERR_EXPIRED) {
3278			nfs4_schedule_lease_recovery(server->nfs_client);
3279			exception.retry = 1;
3280			continue;
3281		}
3282		if (status == -EAGAIN) {
3283			/* We must have found a delegation */
3284			exception.retry = 1;
3285			continue;
3286		}
3287		if (nfs4_clear_cap_atomic_open_v1(server, status, &exception))
3288			continue;
3289		res = ERR_PTR(nfs4_handle_exception(server,
3290					status, &exception));
3291	} while (exception.retry);
3292	return res;
3293}
3294
3295static int _nfs4_do_setattr(struct inode *inode,
3296			    struct nfs_setattrargs *arg,
3297			    struct nfs_setattrres *res,
3298			    const struct cred *cred,
3299			    struct nfs_open_context *ctx)
3300{
3301	struct nfs_server *server = NFS_SERVER(inode);
3302	struct rpc_message msg = {
3303		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETATTR],
3304		.rpc_argp	= arg,
3305		.rpc_resp	= res,
3306		.rpc_cred	= cred,
3307	};
3308	const struct cred *delegation_cred = NULL;
3309	unsigned long timestamp = jiffies;
3310	bool truncate;
3311	int status;
3312
3313	nfs_fattr_init(res->fattr);
3314
3315	/* Servers should only apply open mode checks for file size changes */
3316	truncate = (arg->iap->ia_valid & ATTR_SIZE) ? true : false;
3317	if (!truncate) {
3318		nfs4_inode_make_writeable(inode);
3319		goto zero_stateid;
3320	}
3321
3322	if (nfs4_copy_delegation_stateid(inode, FMODE_WRITE, &arg->stateid, &delegation_cred)) {
3323		/* Use that stateid */
3324	} else if (ctx != NULL && ctx->state) {
3325		struct nfs_lock_context *l_ctx;
3326		if (!nfs4_valid_open_stateid(ctx->state))
3327			return -EBADF;
3328		l_ctx = nfs_get_lock_context(ctx);
3329		if (IS_ERR(l_ctx))
3330			return PTR_ERR(l_ctx);
3331		status = nfs4_select_rw_stateid(ctx->state, FMODE_WRITE, l_ctx,
3332						&arg->stateid, &delegation_cred);
3333		nfs_put_lock_context(l_ctx);
3334		if (status == -EIO)
3335			return -EBADF;
3336		else if (status == -EAGAIN)
3337			goto zero_stateid;
3338	} else {
3339zero_stateid:
3340		nfs4_stateid_copy(&arg->stateid, &zero_stateid);
3341	}
3342	if (delegation_cred)
3343		msg.rpc_cred = delegation_cred;
3344
3345	status = nfs4_call_sync(server->client, server, &msg, &arg->seq_args, &res->seq_res, 1);
3346
3347	put_cred(delegation_cred);
3348	if (status == 0 && ctx != NULL)
3349		renew_lease(server, timestamp);
3350	trace_nfs4_setattr(inode, &arg->stateid, status);
3351	return status;
3352}
3353
3354static int nfs4_do_setattr(struct inode *inode, const struct cred *cred,
3355			   struct nfs_fattr *fattr, struct iattr *sattr,
3356			   struct nfs_open_context *ctx, struct nfs4_label *ilabel)
3357{
3358	struct nfs_server *server = NFS_SERVER(inode);
3359	__u32 bitmask[NFS4_BITMASK_SZ];
3360	struct nfs4_state *state = ctx ? ctx->state : NULL;
3361	struct nfs_setattrargs	arg = {
3362		.fh		= NFS_FH(inode),
3363		.iap		= sattr,
3364		.server		= server,
3365		.bitmask = bitmask,
3366		.label		= ilabel,
3367	};
3368	struct nfs_setattrres  res = {
3369		.fattr		= fattr,
3370		.server		= server,
3371	};
3372	struct nfs4_exception exception = {
3373		.state = state,
3374		.inode = inode,
3375		.stateid = &arg.stateid,
3376	};
3377	unsigned long adjust_flags = NFS_INO_INVALID_CHANGE;
3378	int err;
3379
3380	if (sattr->ia_valid & (ATTR_MODE | ATTR_KILL_SUID | ATTR_KILL_SGID))
3381		adjust_flags |= NFS_INO_INVALID_MODE;
3382	if (sattr->ia_valid & (ATTR_UID | ATTR_GID))
3383		adjust_flags |= NFS_INO_INVALID_OTHER;
3384
3385	do {
3386		nfs4_bitmap_copy_adjust(bitmask, nfs4_bitmask(server, fattr->label),
3387					inode, adjust_flags);
3388
3389		err = _nfs4_do_setattr(inode, &arg, &res, cred, ctx);
3390		switch (err) {
3391		case -NFS4ERR_OPENMODE:
3392			if (!(sattr->ia_valid & ATTR_SIZE)) {
3393				pr_warn_once("NFSv4: server %s is incorrectly "
3394						"applying open mode checks to "
3395						"a SETATTR that is not "
3396						"changing file size.\n",
3397						server->nfs_client->cl_hostname);
3398			}
3399			if (state && !(state->state & FMODE_WRITE)) {
3400				err = -EBADF;
3401				if (sattr->ia_valid & ATTR_OPEN)
3402					err = -EACCES;
3403				goto out;
3404			}
3405		}
3406		err = nfs4_handle_exception(server, err, &exception);
3407	} while (exception.retry);
3408out:
3409	return err;
3410}
3411
3412static bool
3413nfs4_wait_on_layoutreturn(struct inode *inode, struct rpc_task *task)
3414{
3415	if (inode == NULL || !nfs_have_layout(inode))
3416		return false;
3417
3418	return pnfs_wait_on_layoutreturn(inode, task);
3419}
3420
3421/*
3422 * Update the seqid of an open stateid
3423 */
3424static void nfs4_sync_open_stateid(nfs4_stateid *dst,
3425		struct nfs4_state *state)
3426{
3427	__be32 seqid_open;
3428	u32 dst_seqid;
3429	int seq;
3430
3431	for (;;) {
3432		if (!nfs4_valid_open_stateid(state))
3433			break;
3434		seq = read_seqbegin(&state->seqlock);
3435		if (!nfs4_state_match_open_stateid_other(state, dst)) {
3436			nfs4_stateid_copy(dst, &state->open_stateid);
3437			if (read_seqretry(&state->seqlock, seq))
3438				continue;
3439			break;
3440		}
3441		seqid_open = state->open_stateid.seqid;
3442		if (read_seqretry(&state->seqlock, seq))
3443			continue;
3444
3445		dst_seqid = be32_to_cpu(dst->seqid);
3446		if ((s32)(dst_seqid - be32_to_cpu(seqid_open)) < 0)
3447			dst->seqid = seqid_open;
3448		break;
3449	}
3450}
3451
3452/*
3453 * Update the seqid of an open stateid after receiving
3454 * NFS4ERR_OLD_STATEID
3455 */
3456static bool nfs4_refresh_open_old_stateid(nfs4_stateid *dst,
3457		struct nfs4_state *state)
3458{
3459	__be32 seqid_open;
3460	u32 dst_seqid;
3461	bool ret;
3462	int seq, status = -EAGAIN;
3463	DEFINE_WAIT(wait);
3464
3465	for (;;) {
3466		ret = false;
3467		if (!nfs4_valid_open_stateid(state))
3468			break;
3469		seq = read_seqbegin(&state->seqlock);
3470		if (!nfs4_state_match_open_stateid_other(state, dst)) {
3471			if (read_seqretry(&state->seqlock, seq))
3472				continue;
3473			break;
3474		}
3475
3476		write_seqlock(&state->seqlock);
3477		seqid_open = state->open_stateid.seqid;
3478
3479		dst_seqid = be32_to_cpu(dst->seqid);
3480
3481		/* Did another OPEN bump the state's seqid?  try again: */
3482		if ((s32)(be32_to_cpu(seqid_open) - dst_seqid) > 0) {
3483			dst->seqid = seqid_open;
3484			write_sequnlock(&state->seqlock);
3485			ret = true;
3486			break;
3487		}
3488
3489		/* server says we're behind but we haven't seen the update yet */
3490		set_bit(NFS_STATE_CHANGE_WAIT, &state->flags);
3491		prepare_to_wait(&state->waitq, &wait, TASK_KILLABLE);
3492		write_sequnlock(&state->seqlock);
3493		trace_nfs4_close_stateid_update_wait(state->inode, dst, 0);
3494
3495		if (fatal_signal_pending(current))
3496			status = -EINTR;
3497		else
3498			if (schedule_timeout(5*HZ) != 0)
3499				status = 0;
3500
3501		finish_wait(&state->waitq, &wait);
3502
3503		if (!status)
3504			continue;
3505		if (status == -EINTR)
3506			break;
3507
3508		/* we slept the whole 5 seconds, we must have lost a seqid */
3509		dst->seqid = cpu_to_be32(dst_seqid + 1);
3510		ret = true;
3511		break;
3512	}
3513
3514	return ret;
3515}
3516
3517struct nfs4_closedata {
3518	struct inode *inode;
3519	struct nfs4_state *state;
3520	struct nfs_closeargs arg;
3521	struct nfs_closeres res;
3522	struct {
3523		struct nfs4_layoutreturn_args arg;
3524		struct nfs4_layoutreturn_res res;
3525		struct nfs4_xdr_opaque_data ld_private;
3526		u32 roc_barrier;
3527		bool roc;
3528	} lr;
3529	struct nfs_fattr fattr;
3530	unsigned long timestamp;
3531};
3532
3533static void nfs4_free_closedata(void *data)
3534{
3535	struct nfs4_closedata *calldata = data;
3536	struct nfs4_state_owner *sp = calldata->state->owner;
3537	struct super_block *sb = calldata->state->inode->i_sb;
3538
3539	if (calldata->lr.roc)
3540		pnfs_roc_release(&calldata->lr.arg, &calldata->lr.res,
3541				calldata->res.lr_ret);
3542	nfs4_put_open_state(calldata->state);
3543	nfs_free_seqid(calldata->arg.seqid);
3544	nfs4_put_state_owner(sp);
3545	nfs_sb_deactive(sb);
3546	kfree(calldata);
3547}
3548
3549static void nfs4_close_done(struct rpc_task *task, void *data)
3550{
3551	struct nfs4_closedata *calldata = data;
3552	struct nfs4_state *state = calldata->state;
3553	struct nfs_server *server = NFS_SERVER(calldata->inode);
3554	nfs4_stateid *res_stateid = NULL;
3555	struct nfs4_exception exception = {
3556		.state = state,
3557		.inode = calldata->inode,
3558		.stateid = &calldata->arg.stateid,
3559	};
3560
3561	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
3562		return;
3563	trace_nfs4_close(state, &calldata->arg, &calldata->res, task->tk_status);
3564
3565	/* Handle Layoutreturn errors */
3566	if (pnfs_roc_done(task, &calldata->arg.lr_args, &calldata->res.lr_res,
3567			  &calldata->res.lr_ret) == -EAGAIN)
3568		goto out_restart;
3569
3570	/* hmm. we are done with the inode, and in the process of freeing
3571	 * the state_owner. we keep this around to process errors
3572	 */
3573	switch (task->tk_status) {
3574		case 0:
3575			res_stateid = &calldata->res.stateid;
3576			renew_lease(server, calldata->timestamp);
3577			break;
3578		case -NFS4ERR_ACCESS:
3579			if (calldata->arg.bitmask != NULL) {
3580				calldata->arg.bitmask = NULL;
3581				calldata->res.fattr = NULL;
3582				goto out_restart;
3583
3584			}
3585			break;
3586		case -NFS4ERR_OLD_STATEID:
3587			/* Did we race with OPEN? */
3588			if (nfs4_refresh_open_old_stateid(&calldata->arg.stateid,
3589						state))
3590				goto out_restart;
3591			goto out_release;
3592		case -NFS4ERR_ADMIN_REVOKED:
3593		case -NFS4ERR_STALE_STATEID:
3594		case -NFS4ERR_EXPIRED:
3595			nfs4_free_revoked_stateid(server,
3596					&calldata->arg.stateid,
3597					task->tk_msg.rpc_cred);
3598			fallthrough;
3599		case -NFS4ERR_BAD_STATEID:
3600			if (calldata->arg.fmode == 0)
3601				break;
3602			fallthrough;
3603		default:
3604			task->tk_status = nfs4_async_handle_exception(task,
3605					server, task->tk_status, &exception);
3606			if (exception.retry)
3607				goto out_restart;
3608	}
3609	nfs_clear_open_stateid(state, &calldata->arg.stateid,
3610			res_stateid, calldata->arg.fmode);
3611out_release:
3612	task->tk_status = 0;
3613	nfs_release_seqid(calldata->arg.seqid);
3614	nfs_refresh_inode(calldata->inode, &calldata->fattr);
3615	dprintk("%s: ret = %d\n", __func__, task->tk_status);
3616	return;
3617out_restart:
3618	task->tk_status = 0;
3619	rpc_restart_call_prepare(task);
3620	goto out_release;
3621}
3622
3623static void nfs4_close_prepare(struct rpc_task *task, void *data)
3624{
3625	struct nfs4_closedata *calldata = data;
3626	struct nfs4_state *state = calldata->state;
3627	struct inode *inode = calldata->inode;
3628	struct nfs_server *server = NFS_SERVER(inode);
3629	struct pnfs_layout_hdr *lo;
3630	bool is_rdonly, is_wronly, is_rdwr;
3631	int call_close = 0;
3632
3633	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
3634		goto out_wait;
3635
3636	task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_DOWNGRADE];
3637	spin_lock(&state->owner->so_lock);
3638	is_rdwr = test_bit(NFS_O_RDWR_STATE, &state->flags);
3639	is_rdonly = test_bit(NFS_O_RDONLY_STATE, &state->flags);
3640	is_wronly = test_bit(NFS_O_WRONLY_STATE, &state->flags);
3641	/* Calculate the change in open mode */
3642	calldata->arg.fmode = 0;
3643	if (state->n_rdwr == 0) {
3644		if (state->n_rdonly == 0)
3645			call_close |= is_rdonly;
3646		else if (is_rdonly)
3647			calldata->arg.fmode |= FMODE_READ;
3648		if (state->n_wronly == 0)
3649			call_close |= is_wronly;
3650		else if (is_wronly)
3651			calldata->arg.fmode |= FMODE_WRITE;
3652		if (calldata->arg.fmode != (FMODE_READ|FMODE_WRITE))
3653			call_close |= is_rdwr;
3654	} else if (is_rdwr)
3655		calldata->arg.fmode |= FMODE_READ|FMODE_WRITE;
3656
3657	nfs4_sync_open_stateid(&calldata->arg.stateid, state);
3658	if (!nfs4_valid_open_stateid(state))
3659		call_close = 0;
3660	spin_unlock(&state->owner->so_lock);
3661
3662	if (!call_close) {
3663		/* Note: exit _without_ calling nfs4_close_done */
3664		goto out_no_action;
3665	}
3666
3667	if (!calldata->lr.roc && nfs4_wait_on_layoutreturn(inode, task)) {
3668		nfs_release_seqid(calldata->arg.seqid);
3669		goto out_wait;
3670	}
3671
3672	lo = calldata->arg.lr_args ? calldata->arg.lr_args->layout : NULL;
3673	if (lo && !pnfs_layout_is_valid(lo)) {
3674		calldata->arg.lr_args = NULL;
3675		calldata->res.lr_res = NULL;
3676	}
3677
3678	if (calldata->arg.fmode == 0)
3679		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE];
3680
3681	if (calldata->arg.fmode == 0 || calldata->arg.fmode == FMODE_READ) {
3682		/* Close-to-open cache consistency revalidation */
3683		if (!nfs4_have_delegation(inode, FMODE_READ)) {
3684			nfs4_bitmask_set(calldata->arg.bitmask_store,
3685					 server->cache_consistency_bitmask,
3686					 inode, 0);
3687			calldata->arg.bitmask = calldata->arg.bitmask_store;
3688		} else
3689			calldata->arg.bitmask = NULL;
3690	}
3691
3692	calldata->arg.share_access =
3693		nfs4_map_atomic_open_share(NFS_SERVER(inode),
3694				calldata->arg.fmode, 0);
3695
3696	if (calldata->res.fattr == NULL)
3697		calldata->arg.bitmask = NULL;
3698	else if (calldata->arg.bitmask == NULL)
3699		calldata->res.fattr = NULL;
3700	calldata->timestamp = jiffies;
3701	if (nfs4_setup_sequence(NFS_SERVER(inode)->nfs_client,
3702				&calldata->arg.seq_args,
3703				&calldata->res.seq_res,
3704				task) != 0)
3705		nfs_release_seqid(calldata->arg.seqid);
3706	return;
3707out_no_action:
3708	task->tk_action = NULL;
3709out_wait:
3710	nfs4_sequence_done(task, &calldata->res.seq_res);
3711}
3712
3713static const struct rpc_call_ops nfs4_close_ops = {
3714	.rpc_call_prepare = nfs4_close_prepare,
3715	.rpc_call_done = nfs4_close_done,
3716	.rpc_release = nfs4_free_closedata,
3717};
3718
3719/*
3720 * It is possible for data to be read/written from a mem-mapped file
3721 * after the sys_close call (which hits the vfs layer as a flush).
3722 * This means that we can't safely call nfsv4 close on a file until
3723 * the inode is cleared. This in turn means that we are not good
3724 * NFSv4 citizens - we do not indicate to the server to update the file's
3725 * share state even when we are done with one of the three share
3726 * stateid's in the inode.
3727 *
3728 * NOTE: Caller must be holding the sp->so_owner semaphore!
3729 */
3730int nfs4_do_close(struct nfs4_state *state, gfp_t gfp_mask, int wait)
3731{
3732	struct nfs_server *server = NFS_SERVER(state->inode);
3733	struct nfs_seqid *(*alloc_seqid)(struct nfs_seqid_counter *, gfp_t);
3734	struct nfs4_closedata *calldata;
3735	struct nfs4_state_owner *sp = state->owner;
3736	struct rpc_task *task;
3737	struct rpc_message msg = {
3738		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE],
3739		.rpc_cred = state->owner->so_cred,
3740	};
3741	struct rpc_task_setup task_setup_data = {
3742		.rpc_client = server->client,
3743		.rpc_message = &msg,
3744		.callback_ops = &nfs4_close_ops,
3745		.workqueue = nfsiod_workqueue,
3746		.flags = RPC_TASK_ASYNC | RPC_TASK_CRED_NOREF,
3747	};
3748	int status = -ENOMEM;
3749
3750	if (nfs_server_capable(state->inode, NFS_CAP_MOVEABLE))
3751		task_setup_data.flags |= RPC_TASK_MOVEABLE;
3752
3753	nfs4_state_protect(server->nfs_client, NFS_SP4_MACH_CRED_CLEANUP,
3754		&task_setup_data.rpc_client, &msg);
3755
3756	calldata = kzalloc(sizeof(*calldata), gfp_mask);
3757	if (calldata == NULL)
3758		goto out;
3759	nfs4_init_sequence(&calldata->arg.seq_args, &calldata->res.seq_res, 1, 0);
3760	calldata->inode = state->inode;
3761	calldata->state = state;
3762	calldata->arg.fh = NFS_FH(state->inode);
3763	if (!nfs4_copy_open_stateid(&calldata->arg.stateid, state))
3764		goto out_free_calldata;
3765	/* Serialization for the sequence id */
3766	alloc_seqid = server->nfs_client->cl_mvops->alloc_seqid;
3767	calldata->arg.seqid = alloc_seqid(&state->owner->so_seqid, gfp_mask);
3768	if (IS_ERR(calldata->arg.seqid))
3769		goto out_free_calldata;
3770	nfs_fattr_init(&calldata->fattr);
3771	calldata->arg.fmode = 0;
3772	calldata->lr.arg.ld_private = &calldata->lr.ld_private;
3773	calldata->res.fattr = &calldata->fattr;
3774	calldata->res.seqid = calldata->arg.seqid;
3775	calldata->res.server = server;
3776	calldata->res.lr_ret = -NFS4ERR_NOMATCHING_LAYOUT;
3777	calldata->lr.roc = pnfs_roc(state->inode,
3778			&calldata->lr.arg, &calldata->lr.res, msg.rpc_cred);
3779	if (calldata->lr.roc) {
3780		calldata->arg.lr_args = &calldata->lr.arg;
3781		calldata->res.lr_res = &calldata->lr.res;
3782	}
3783	nfs_sb_active(calldata->inode->i_sb);
3784
3785	msg.rpc_argp = &calldata->arg;
3786	msg.rpc_resp = &calldata->res;
3787	task_setup_data.callback_data = calldata;
3788	task = rpc_run_task(&task_setup_data);
3789	if (IS_ERR(task))
3790		return PTR_ERR(task);
3791	status = 0;
3792	if (wait)
3793		status = rpc_wait_for_completion_task(task);
3794	rpc_put_task(task);
3795	return status;
3796out_free_calldata:
3797	kfree(calldata);
3798out:
3799	nfs4_put_open_state(state);
3800	nfs4_put_state_owner(sp);
3801	return status;
3802}
3803
3804static struct inode *
3805nfs4_atomic_open(struct inode *dir, struct nfs_open_context *ctx,
3806		int open_flags, struct iattr *attr, int *opened)
3807{
3808	struct nfs4_state *state;
3809	struct nfs4_label l, *label;
3810
3811	label = nfs4_label_init_security(dir, ctx->dentry, attr, &l);
3812
3813	/* Protect against concurrent sillydeletes */
3814	state = nfs4_do_open(dir, ctx, open_flags, attr, label, opened);
3815
3816	nfs4_label_release_security(label);
3817
3818	if (IS_ERR(state))
3819		return ERR_CAST(state);
3820	return state->inode;
3821}
3822
3823static void nfs4_close_context(struct nfs_open_context *ctx, int is_sync)
3824{
3825	if (ctx->state == NULL)
3826		return;
3827	if (is_sync)
3828		nfs4_close_sync(ctx->state, _nfs4_ctx_to_openmode(ctx));
3829	else
3830		nfs4_close_state(ctx->state, _nfs4_ctx_to_openmode(ctx));
3831}
3832
3833#define FATTR4_WORD1_NFS40_MASK (2*FATTR4_WORD1_MOUNTED_ON_FILEID - 1UL)
3834#define FATTR4_WORD2_NFS41_MASK (2*FATTR4_WORD2_SUPPATTR_EXCLCREAT - 1UL)
3835#define FATTR4_WORD2_NFS42_MASK (2*FATTR4_WORD2_XATTR_SUPPORT - 1UL)
3836
3837static int _nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
3838{
3839	u32 bitmask[3] = {}, minorversion = server->nfs_client->cl_minorversion;
3840	struct nfs4_server_caps_arg args = {
3841		.fhandle = fhandle,
3842		.bitmask = bitmask,
3843	};
3844	struct nfs4_server_caps_res res = {};
3845	struct rpc_message msg = {
3846		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SERVER_CAPS],
3847		.rpc_argp = &args,
3848		.rpc_resp = &res,
3849	};
3850	int status;
3851	int i;
3852
3853	bitmask[0] = FATTR4_WORD0_SUPPORTED_ATTRS |
3854		     FATTR4_WORD0_FH_EXPIRE_TYPE |
3855		     FATTR4_WORD0_LINK_SUPPORT |
3856		     FATTR4_WORD0_SYMLINK_SUPPORT |
3857		     FATTR4_WORD0_ACLSUPPORT |
3858		     FATTR4_WORD0_CASE_INSENSITIVE |
3859		     FATTR4_WORD0_CASE_PRESERVING;
3860	if (minorversion)
3861		bitmask[2] = FATTR4_WORD2_SUPPATTR_EXCLCREAT;
3862
3863	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
3864	if (status == 0) {
3865		/* Sanity check the server answers */
3866		switch (minorversion) {
3867		case 0:
3868			res.attr_bitmask[1] &= FATTR4_WORD1_NFS40_MASK;
3869			res.attr_bitmask[2] = 0;
3870			break;
3871		case 1:
3872			res.attr_bitmask[2] &= FATTR4_WORD2_NFS41_MASK;
3873			break;
3874		case 2:
3875			res.attr_bitmask[2] &= FATTR4_WORD2_NFS42_MASK;
3876		}
3877		memcpy(server->attr_bitmask, res.attr_bitmask, sizeof(server->attr_bitmask));
3878		server->caps &= ~(NFS_CAP_ACLS | NFS_CAP_HARDLINKS |
3879				  NFS_CAP_SYMLINKS| NFS_CAP_SECURITY_LABEL);
3880		server->fattr_valid = NFS_ATTR_FATTR_V4;
3881		if (res.attr_bitmask[0] & FATTR4_WORD0_ACL &&
3882				res.acl_bitmask & ACL4_SUPPORT_ALLOW_ACL)
3883			server->caps |= NFS_CAP_ACLS;
3884		if (res.has_links != 0)
3885			server->caps |= NFS_CAP_HARDLINKS;
3886		if (res.has_symlinks != 0)
3887			server->caps |= NFS_CAP_SYMLINKS;
3888		if (res.case_insensitive)
3889			server->caps |= NFS_CAP_CASE_INSENSITIVE;
3890		if (res.case_preserving)
3891			server->caps |= NFS_CAP_CASE_PRESERVING;
3892#ifdef CONFIG_NFS_V4_SECURITY_LABEL
3893		if (res.attr_bitmask[2] & FATTR4_WORD2_SECURITY_LABEL)
3894			server->caps |= NFS_CAP_SECURITY_LABEL;
3895#endif
3896		if (res.attr_bitmask[0] & FATTR4_WORD0_FS_LOCATIONS)
3897			server->caps |= NFS_CAP_FS_LOCATIONS;
3898		if (!(res.attr_bitmask[0] & FATTR4_WORD0_FILEID))
3899			server->fattr_valid &= ~NFS_ATTR_FATTR_FILEID;
3900		if (!(res.attr_bitmask[1] & FATTR4_WORD1_MODE))
3901			server->fattr_valid &= ~NFS_ATTR_FATTR_MODE;
3902		if (!(res.attr_bitmask[1] & FATTR4_WORD1_NUMLINKS))
3903			server->fattr_valid &= ~NFS_ATTR_FATTR_NLINK;
3904		if (!(res.attr_bitmask[1] & FATTR4_WORD1_OWNER))
3905			server->fattr_valid &= ~(NFS_ATTR_FATTR_OWNER |
3906				NFS_ATTR_FATTR_OWNER_NAME);
3907		if (!(res.attr_bitmask[1] & FATTR4_WORD1_OWNER_GROUP))
3908			server->fattr_valid &= ~(NFS_ATTR_FATTR_GROUP |
3909				NFS_ATTR_FATTR_GROUP_NAME);
3910		if (!(res.attr_bitmask[1] & FATTR4_WORD1_SPACE_USED))
3911			server->fattr_valid &= ~NFS_ATTR_FATTR_SPACE_USED;
3912		if (!(res.attr_bitmask[1] & FATTR4_WORD1_TIME_ACCESS))
3913			server->fattr_valid &= ~NFS_ATTR_FATTR_ATIME;
3914		if (!(res.attr_bitmask[1] & FATTR4_WORD1_TIME_METADATA))
3915			server->fattr_valid &= ~NFS_ATTR_FATTR_CTIME;
3916		if (!(res.attr_bitmask[1] & FATTR4_WORD1_TIME_MODIFY))
3917			server->fattr_valid &= ~NFS_ATTR_FATTR_MTIME;
3918		memcpy(server->attr_bitmask_nl, res.attr_bitmask,
3919				sizeof(server->attr_bitmask));
3920		server->attr_bitmask_nl[2] &= ~FATTR4_WORD2_SECURITY_LABEL;
3921
3922		memcpy(server->cache_consistency_bitmask, res.attr_bitmask, sizeof(server->cache_consistency_bitmask));
3923		server->cache_consistency_bitmask[0] &= FATTR4_WORD0_CHANGE|FATTR4_WORD0_SIZE;
3924		server->cache_consistency_bitmask[1] &= FATTR4_WORD1_TIME_METADATA|FATTR4_WORD1_TIME_MODIFY;
3925		server->cache_consistency_bitmask[2] = 0;
3926
3927		/* Avoid a regression due to buggy server */
3928		for (i = 0; i < ARRAY_SIZE(res.exclcreat_bitmask); i++)
3929			res.exclcreat_bitmask[i] &= res.attr_bitmask[i];
3930		memcpy(server->exclcreat_bitmask, res.exclcreat_bitmask,
3931			sizeof(server->exclcreat_bitmask));
3932
3933		server->acl_bitmask = res.acl_bitmask;
3934		server->fh_expire_type = res.fh_expire_type;
3935	}
3936
3937	return status;
3938}
3939
3940int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
3941{
3942	struct nfs4_exception exception = {
3943		.interruptible = true,
3944	};
3945	int err;
3946
3947	nfs4_server_set_init_caps(server);
3948	do {
3949		err = nfs4_handle_exception(server,
3950				_nfs4_server_capabilities(server, fhandle),
3951				&exception);
3952	} while (exception.retry);
3953	return err;
3954}
3955
3956static void test_fs_location_for_trunking(struct nfs4_fs_location *location,
3957					  struct nfs_client *clp,
3958					  struct nfs_server *server)
3959{
3960	int i;
3961
3962	for (i = 0; i < location->nservers; i++) {
3963		struct nfs4_string *srv_loc = &location->servers[i];
3964		struct sockaddr_storage addr;
3965		size_t addrlen;
3966		struct xprt_create xprt_args = {
3967			.ident = 0,
3968			.net = clp->cl_net,
3969		};
3970		struct nfs4_add_xprt_data xprtdata = {
3971			.clp = clp,
3972		};
3973		struct rpc_add_xprt_test rpcdata = {
3974			.add_xprt_test = clp->cl_mvops->session_trunk,
3975			.data = &xprtdata,
3976		};
3977		char *servername = NULL;
3978
3979		if (!srv_loc->len)
3980			continue;
3981
3982		addrlen = nfs_parse_server_name(srv_loc->data, srv_loc->len,
3983						&addr, sizeof(addr),
3984						clp->cl_net, server->port);
3985		if (!addrlen)
3986			return;
3987		xprt_args.dstaddr = (struct sockaddr *)&addr;
3988		xprt_args.addrlen = addrlen;
3989		servername = kmalloc(srv_loc->len + 1, GFP_KERNEL);
3990		if (!servername)
3991			return;
3992		memcpy(servername, srv_loc->data, srv_loc->len);
3993		servername[srv_loc->len] = '\0';
3994		xprt_args.servername = servername;
3995
3996		xprtdata.cred = nfs4_get_clid_cred(clp);
3997		rpc_clnt_add_xprt(clp->cl_rpcclient, &xprt_args,
3998				  rpc_clnt_setup_test_and_add_xprt,
3999				  &rpcdata);
4000		if (xprtdata.cred)
4001			put_cred(xprtdata.cred);
4002		kfree(servername);
4003	}
4004}
4005
4006static int _nfs4_discover_trunking(struct nfs_server *server,
4007				   struct nfs_fh *fhandle)
4008{
4009	struct nfs4_fs_locations *locations = NULL;
4010	struct page *page;
4011	const struct cred *cred;
4012	struct nfs_client *clp = server->nfs_client;
4013	const struct nfs4_state_maintenance_ops *ops =
4014		clp->cl_mvops->state_renewal_ops;
4015	int status = -ENOMEM, i;
4016
4017	cred = ops->get_state_renewal_cred(clp);
4018	if (cred == NULL) {
4019		cred = nfs4_get_clid_cred(clp);
4020		if (cred == NULL)
4021			return -ENOKEY;
4022	}
4023
4024	page = alloc_page(GFP_KERNEL);
4025	if (!page)
4026		goto out_put_cred;
4027	locations = kmalloc(sizeof(struct nfs4_fs_locations), GFP_KERNEL);
4028	if (!locations)
4029		goto out_free;
4030	locations->fattr = nfs_alloc_fattr();
4031	if (!locations->fattr)
4032		goto out_free_2;
4033
4034	status = nfs4_proc_get_locations(server, fhandle, locations, page,
4035					 cred);
4036	if (status)
4037		goto out_free_3;
4038
4039	for (i = 0; i < locations->nlocations; i++)
4040		test_fs_location_for_trunking(&locations->locations[i], clp,
4041					      server);
4042out_free_3:
4043	kfree(locations->fattr);
4044out_free_2:
4045	kfree(locations);
4046out_free:
4047	__free_page(page);
4048out_put_cred:
4049	put_cred(cred);
4050	return status;
4051}
4052
4053static int nfs4_discover_trunking(struct nfs_server *server,
4054				  struct nfs_fh *fhandle)
4055{
4056	struct nfs4_exception exception = {
4057		.interruptible = true,
4058	};
4059	struct nfs_client *clp = server->nfs_client;
4060	int err = 0;
4061
4062	if (!nfs4_has_session(clp))
4063		goto out;
4064	do {
4065		err = nfs4_handle_exception(server,
4066				_nfs4_discover_trunking(server, fhandle),
4067				&exception);
4068	} while (exception.retry);
4069out:
4070	return err;
4071}
4072
4073static int _nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle,
4074		struct nfs_fsinfo *info)
4075{
4076	u32 bitmask[3];
4077	struct nfs4_lookup_root_arg args = {
4078		.bitmask = bitmask,
4079	};
4080	struct nfs4_lookup_res res = {
4081		.server = server,
4082		.fattr = info->fattr,
4083		.fh = fhandle,
4084	};
4085	struct rpc_message msg = {
4086		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP_ROOT],
4087		.rpc_argp = &args,
4088		.rpc_resp = &res,
4089	};
4090
4091	bitmask[0] = nfs4_fattr_bitmap[0];
4092	bitmask[1] = nfs4_fattr_bitmap[1];
4093	/*
4094	 * Process the label in the upcoming getfattr
4095	 */
4096	bitmask[2] = nfs4_fattr_bitmap[2] & ~FATTR4_WORD2_SECURITY_LABEL;
4097
4098	nfs_fattr_init(info->fattr);
4099	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
4100}
4101
4102static int nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle,
4103		struct nfs_fsinfo *info)
4104{
4105	struct nfs4_exception exception = {
4106		.interruptible = true,
4107	};
4108	int err;
4109	do {
4110		err = _nfs4_lookup_root(server, fhandle, info);
4111		trace_nfs4_lookup_root(server, fhandle, info->fattr, err);
4112		switch (err) {
4113		case 0:
4114		case -NFS4ERR_WRONGSEC:
4115			goto out;
4116		default:
4117			err = nfs4_handle_exception(server, err, &exception);
4118		}
4119	} while (exception.retry);
4120out:
4121	return err;
4122}
4123
4124static int nfs4_lookup_root_sec(struct nfs_server *server, struct nfs_fh *fhandle,
4125				struct nfs_fsinfo *info, rpc_authflavor_t flavor)
4126{
4127	struct rpc_auth_create_args auth_args = {
4128		.pseudoflavor = flavor,
4129	};
4130	struct rpc_auth *auth;
4131
4132	auth = rpcauth_create(&auth_args, server->client);
4133	if (IS_ERR(auth))
4134		return -EACCES;
4135	return nfs4_lookup_root(server, fhandle, info);
4136}
4137
4138/*
4139 * Retry pseudoroot lookup with various security flavors.  We do this when:
4140 *
4141 *   NFSv4.0: the PUTROOTFH operation returns NFS4ERR_WRONGSEC
4142 *   NFSv4.1: the server does not support the SECINFO_NO_NAME operation
4143 *
4144 * Returns zero on success, or a negative NFS4ERR value, or a
4145 * negative errno value.
4146 */
4147static int nfs4_find_root_sec(struct nfs_server *server, struct nfs_fh *fhandle,
4148			      struct nfs_fsinfo *info)
4149{
4150	/* Per 3530bis 15.33.5 */
4151	static const rpc_authflavor_t flav_array[] = {
4152		RPC_AUTH_GSS_KRB5P,
4153		RPC_AUTH_GSS_KRB5I,
4154		RPC_AUTH_GSS_KRB5,
4155		RPC_AUTH_UNIX,			/* courtesy */
4156		RPC_AUTH_NULL,
4157	};
4158	int status = -EPERM;
4159	size_t i;
4160
4161	if (server->auth_info.flavor_len > 0) {
4162		/* try each flavor specified by user */
4163		for (i = 0; i < server->auth_info.flavor_len; i++) {
4164			status = nfs4_lookup_root_sec(server, fhandle, info,
4165						server->auth_info.flavors[i]);
4166			if (status == -NFS4ERR_WRONGSEC || status == -EACCES)
4167				continue;
4168			break;
4169		}
4170	} else {
4171		/* no flavors specified by user, try default list */
4172		for (i = 0; i < ARRAY_SIZE(flav_array); i++) {
4173			status = nfs4_lookup_root_sec(server, fhandle, info,
4174						      flav_array[i]);
4175			if (status == -NFS4ERR_WRONGSEC || status == -EACCES)
4176				continue;
4177			break;
4178		}
4179	}
4180
4181	/*
4182	 * -EACCES could mean that the user doesn't have correct permissions
4183	 * to access the mount.  It could also mean that we tried to mount
4184	 * with a gss auth flavor, but rpc.gssd isn't running.  Either way,
4185	 * existing mount programs don't handle -EACCES very well so it should
4186	 * be mapped to -EPERM instead.
4187	 */
4188	if (status == -EACCES)
4189		status = -EPERM;
4190	return status;
4191}
4192
4193/**
4194 * nfs4_proc_get_rootfh - get file handle for server's pseudoroot
4195 * @server: initialized nfs_server handle
4196 * @fhandle: we fill in the pseudo-fs root file handle
4197 * @info: we fill in an FSINFO struct
4198 * @auth_probe: probe the auth flavours
4199 *
4200 * Returns zero on success, or a negative errno.
4201 */
4202int nfs4_proc_get_rootfh(struct nfs_server *server, struct nfs_fh *fhandle,
4203			 struct nfs_fsinfo *info,
4204			 bool auth_probe)
4205{
4206	int status = 0;
4207
4208	if (!auth_probe)
4209		status = nfs4_lookup_root(server, fhandle, info);
4210
4211	if (auth_probe || status == NFS4ERR_WRONGSEC)
4212		status = server->nfs_client->cl_mvops->find_root_sec(server,
4213				fhandle, info);
4214
4215	if (status == 0)
4216		status = nfs4_server_capabilities(server, fhandle);
4217	if (status == 0)
4218		status = nfs4_do_fsinfo(server, fhandle, info);
4219
4220	return nfs4_map_errors(status);
4221}
4222
4223static int nfs4_proc_get_root(struct nfs_server *server, struct nfs_fh *mntfh,
4224			      struct nfs_fsinfo *info)
4225{
4226	int error;
4227	struct nfs_fattr *fattr = info->fattr;
4228
4229	error = nfs4_server_capabilities(server, mntfh);
4230	if (error < 0) {
4231		dprintk("nfs4_get_root: getcaps error = %d\n", -error);
4232		return error;
4233	}
4234
4235	error = nfs4_proc_getattr(server, mntfh, fattr, NULL);
4236	if (error < 0) {
4237		dprintk("nfs4_get_root: getattr error = %d\n", -error);
4238		goto out;
4239	}
4240
4241	if (fattr->valid & NFS_ATTR_FATTR_FSID &&
4242	    !nfs_fsid_equal(&server->fsid, &fattr->fsid))
4243		memcpy(&server->fsid, &fattr->fsid, sizeof(server->fsid));
4244
4245out:
4246	return error;
4247}
4248
4249/*
4250 * Get locations and (maybe) other attributes of a referral.
4251 * Note that we'll actually follow the referral later when
4252 * we detect fsid mismatch in inode revalidation
4253 */
4254static int nfs4_get_referral(struct rpc_clnt *client, struct inode *dir,
4255			     const struct qstr *name, struct nfs_fattr *fattr,
4256			     struct nfs_fh *fhandle)
4257{
4258	int status = -ENOMEM;
4259	struct page *page = NULL;
4260	struct nfs4_fs_locations *locations = NULL;
4261
4262	page = alloc_page(GFP_KERNEL);
4263	if (page == NULL)
4264		goto out;
4265	locations = kmalloc(sizeof(struct nfs4_fs_locations), GFP_KERNEL);
4266	if (locations == NULL)
4267		goto out;
4268
4269	locations->fattr = fattr;
4270
4271	status = nfs4_proc_fs_locations(client, dir, name, locations, page);
4272	if (status != 0)
4273		goto out;
4274
4275	/*
4276	 * If the fsid didn't change, this is a migration event, not a
4277	 * referral.  Cause us to drop into the exception handler, which
4278	 * will kick off migration recovery.
4279	 */
4280	if (nfs_fsid_equal(&NFS_SERVER(dir)->fsid, &fattr->fsid)) {
4281		dprintk("%s: server did not return a different fsid for"
4282			" a referral at %s\n", __func__, name->name);
4283		status = -NFS4ERR_MOVED;
4284		goto out;
4285	}
4286	/* Fixup attributes for the nfs_lookup() call to nfs_fhget() */
4287	nfs_fixup_referral_attributes(fattr);
4288	memset(fhandle, 0, sizeof(struct nfs_fh));
4289out:
4290	if (page)
4291		__free_page(page);
4292	kfree(locations);
4293	return status;
4294}
4295
4296static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle,
4297				struct nfs_fattr *fattr, struct inode *inode)
4298{
4299	__u32 bitmask[NFS4_BITMASK_SZ];
4300	struct nfs4_getattr_arg args = {
4301		.fh = fhandle,
4302		.bitmask = bitmask,
4303	};
4304	struct nfs4_getattr_res res = {
4305		.fattr = fattr,
4306		.server = server,
4307	};
4308	struct rpc_message msg = {
4309		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETATTR],
4310		.rpc_argp = &args,
4311		.rpc_resp = &res,
4312	};
4313	unsigned short task_flags = 0;
4314
4315	if (nfs4_has_session(server->nfs_client))
4316		task_flags = RPC_TASK_MOVEABLE;
4317
4318	/* Is this is an attribute revalidation, subject to softreval? */
4319	if (inode && (server->flags & NFS_MOUNT_SOFTREVAL))
4320		task_flags |= RPC_TASK_TIMEOUT;
4321
4322	nfs4_bitmap_copy_adjust(bitmask, nfs4_bitmask(server, fattr->label), inode, 0);
4323	nfs_fattr_init(fattr);
4324	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0, 0);
4325	return nfs4_do_call_sync(server->client, server, &msg,
4326			&args.seq_args, &res.seq_res, task_flags);
4327}
4328
4329int nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle,
4330				struct nfs_fattr *fattr, struct inode *inode)
4331{
4332	struct nfs4_exception exception = {
4333		.interruptible = true,
4334	};
4335	int err;
4336	do {
4337		err = _nfs4_proc_getattr(server, fhandle, fattr, inode);
4338		trace_nfs4_getattr(server, fhandle, fattr, err);
4339		err = nfs4_handle_exception(server, err,
4340				&exception);
4341	} while (exception.retry);
4342	return err;
4343}
4344
4345/*
4346 * The file is not closed if it is opened due to the a request to change
4347 * the size of the file. The open call will not be needed once the
4348 * VFS layer lookup-intents are implemented.
4349 *
4350 * Close is called when the inode is destroyed.
4351 * If we haven't opened the file for O_WRONLY, we
4352 * need to in the size_change case to obtain a stateid.
4353 *
4354 * Got race?
4355 * Because OPEN is always done by name in nfsv4, it is
4356 * possible that we opened a different file by the same
4357 * name.  We can recognize this race condition, but we
4358 * can't do anything about it besides returning an error.
4359 *
4360 * This will be fixed with VFS changes (lookup-intent).
4361 */
4362static int
4363nfs4_proc_setattr(struct dentry *dentry, struct nfs_fattr *fattr,
4364		  struct iattr *sattr)
4365{
4366	struct inode *inode = d_inode(dentry);
4367	const struct cred *cred = NULL;
4368	struct nfs_open_context *ctx = NULL;
4369	int status;
4370
4371	if (pnfs_ld_layoutret_on_setattr(inode) &&
4372	    sattr->ia_valid & ATTR_SIZE &&
4373	    sattr->ia_size < i_size_read(inode))
4374		pnfs_commit_and_return_layout(inode);
4375
4376	nfs_fattr_init(fattr);
4377
4378	/* Deal with open(O_TRUNC) */
4379	if (sattr->ia_valid & ATTR_OPEN)
4380		sattr->ia_valid &= ~(ATTR_MTIME|ATTR_CTIME);
4381
4382	/* Optimization: if the end result is no change, don't RPC */
4383	if ((sattr->ia_valid & ~(ATTR_FILE|ATTR_OPEN)) == 0)
4384		return 0;
4385
4386	/* Search for an existing open(O_WRITE) file */
4387	if (sattr->ia_valid & ATTR_FILE) {
4388
4389		ctx = nfs_file_open_context(sattr->ia_file);
4390		if (ctx)
4391			cred = ctx->cred;
4392	}
4393
4394	/* Return any delegations if we're going to change ACLs */
4395	if ((sattr->ia_valid & (ATTR_MODE|ATTR_UID|ATTR_GID)) != 0)
4396		nfs4_inode_make_writeable(inode);
4397
4398	status = nfs4_do_setattr(inode, cred, fattr, sattr, ctx, NULL);
4399	if (status == 0) {
4400		nfs_setattr_update_inode(inode, sattr, fattr);
4401		nfs_setsecurity(inode, fattr);
4402	}
4403	return status;
4404}
4405
4406static int _nfs4_proc_lookup(struct rpc_clnt *clnt, struct inode *dir,
4407		struct dentry *dentry, struct nfs_fh *fhandle,
4408		struct nfs_fattr *fattr)
4409{
4410	struct nfs_server *server = NFS_SERVER(dir);
4411	int		       status;
4412	struct nfs4_lookup_arg args = {
4413		.bitmask = server->attr_bitmask,
4414		.dir_fh = NFS_FH(dir),
4415		.name = &dentry->d_name,
4416	};
4417	struct nfs4_lookup_res res = {
4418		.server = server,
4419		.fattr = fattr,
4420		.fh = fhandle,
4421	};
4422	struct rpc_message msg = {
4423		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP],
4424		.rpc_argp = &args,
4425		.rpc_resp = &res,
4426	};
4427	unsigned short task_flags = 0;
4428
4429	if (nfs_server_capable(dir, NFS_CAP_MOVEABLE))
4430		task_flags = RPC_TASK_MOVEABLE;
4431
4432	/* Is this is an attribute revalidation, subject to softreval? */
4433	if (nfs_lookup_is_soft_revalidate(dentry))
4434		task_flags |= RPC_TASK_TIMEOUT;
4435
4436	args.bitmask = nfs4_bitmask(server, fattr->label);
4437
4438	nfs_fattr_init(fattr);
4439
4440	dprintk("NFS call  lookup %pd2\n", dentry);
4441	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0, 0);
4442	status = nfs4_do_call_sync(clnt, server, &msg,
4443			&args.seq_args, &res.seq_res, task_flags);
4444	dprintk("NFS reply lookup: %d\n", status);
4445	return status;
4446}
4447
4448static void nfs_fixup_secinfo_attributes(struct nfs_fattr *fattr)
4449{
4450	fattr->valid |= NFS_ATTR_FATTR_TYPE | NFS_ATTR_FATTR_MODE |
4451		NFS_ATTR_FATTR_NLINK | NFS_ATTR_FATTR_MOUNTPOINT;
4452	fattr->mode = S_IFDIR | S_IRUGO | S_IXUGO;
4453	fattr->nlink = 2;
4454}
4455
4456static int nfs4_proc_lookup_common(struct rpc_clnt **clnt, struct inode *dir,
4457				   struct dentry *dentry, struct nfs_fh *fhandle,
4458				   struct nfs_fattr *fattr)
4459{
4460	struct nfs4_exception exception = {
4461		.interruptible = true,
4462	};
4463	struct rpc_clnt *client = *clnt;
4464	const struct qstr *name = &dentry->d_name;
4465	int err;
4466	do {
4467		err = _nfs4_proc_lookup(client, dir, dentry, fhandle, fattr);
4468		trace_nfs4_lookup(dir, name, err);
4469		switch (err) {
4470		case -NFS4ERR_BADNAME:
4471			err = -ENOENT;
4472			goto out;
4473		case -NFS4ERR_MOVED:
4474			err = nfs4_get_referral(client, dir, name, fattr, fhandle);
4475			if (err == -NFS4ERR_MOVED)
4476				err = nfs4_handle_exception(NFS_SERVER(dir), err, &exception);
4477			goto out;
4478		case -NFS4ERR_WRONGSEC:
4479			err = -EPERM;
4480			if (client != *clnt)
4481				goto out;
4482			client = nfs4_negotiate_security(client, dir, name);
4483			if (IS_ERR(client))
4484				return PTR_ERR(client);
4485
4486			exception.retry = 1;
4487			break;
4488		default:
4489			err = nfs4_handle_exception(NFS_SERVER(dir), err, &exception);
4490		}
4491	} while (exception.retry);
4492
4493out:
4494	if (err == 0)
4495		*clnt = client;
4496	else if (client != *clnt)
4497		rpc_shutdown_client(client);
4498
4499	return err;
4500}
4501
4502static int nfs4_proc_lookup(struct inode *dir, struct dentry *dentry,
4503			    struct nfs_fh *fhandle, struct nfs_fattr *fattr)
4504{
4505	int status;
4506	struct rpc_clnt *client = NFS_CLIENT(dir);
4507
4508	status = nfs4_proc_lookup_common(&client, dir, dentry, fhandle, fattr);
4509	if (client != NFS_CLIENT(dir)) {
4510		rpc_shutdown_client(client);
4511		nfs_fixup_secinfo_attributes(fattr);
4512	}
4513	return status;
4514}
4515
4516struct rpc_clnt *
4517nfs4_proc_lookup_mountpoint(struct inode *dir, struct dentry *dentry,
4518			    struct nfs_fh *fhandle, struct nfs_fattr *fattr)
4519{
4520	struct rpc_clnt *client = NFS_CLIENT(dir);
4521	int status;
4522
4523	status = nfs4_proc_lookup_common(&client, dir, dentry, fhandle, fattr);
4524	if (status < 0)
4525		return ERR_PTR(status);
4526	return (client == NFS_CLIENT(dir)) ? rpc_clone_client(client) : client;
4527}
4528
4529static int _nfs4_proc_lookupp(struct inode *inode,
4530		struct nfs_fh *fhandle, struct nfs_fattr *fattr)
4531{
4532	struct rpc_clnt *clnt = NFS_CLIENT(inode);
4533	struct nfs_server *server = NFS_SERVER(inode);
4534	int		       status;
4535	struct nfs4_lookupp_arg args = {
4536		.bitmask = server->attr_bitmask,
4537		.fh = NFS_FH(inode),
4538	};
4539	struct nfs4_lookupp_res res = {
4540		.server = server,
4541		.fattr = fattr,
4542		.fh = fhandle,
4543	};
4544	struct rpc_message msg = {
4545		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUPP],
4546		.rpc_argp = &args,
4547		.rpc_resp = &res,
4548	};
4549	unsigned short task_flags = 0;
4550
4551	if (NFS_SERVER(inode)->flags & NFS_MOUNT_SOFTREVAL)
4552		task_flags |= RPC_TASK_TIMEOUT;
4553
4554	args.bitmask = nfs4_bitmask(server, fattr->label);
4555
4556	nfs_fattr_init(fattr);
4557
4558	dprintk("NFS call  lookupp ino=0x%lx\n", inode->i_ino);
4559	status = nfs4_call_sync(clnt, server, &msg, &args.seq_args,
4560				&res.seq_res, task_flags);
4561	dprintk("NFS reply lookupp: %d\n", status);
4562	return status;
4563}
4564
4565static int nfs4_proc_lookupp(struct inode *inode, struct nfs_fh *fhandle,
4566			     struct nfs_fattr *fattr)
4567{
4568	struct nfs4_exception exception = {
4569		.interruptible = true,
4570	};
4571	int err;
4572	do {
4573		err = _nfs4_proc_lookupp(inode, fhandle, fattr);
4574		trace_nfs4_lookupp(inode, err);
4575		err = nfs4_handle_exception(NFS_SERVER(inode), err,
4576				&exception);
4577	} while (exception.retry);
4578	return err;
4579}
4580
4581static int _nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry,
4582			     const struct cred *cred)
4583{
4584	struct nfs_server *server = NFS_SERVER(inode);
4585	struct nfs4_accessargs args = {
4586		.fh = NFS_FH(inode),
4587		.access = entry->mask,
4588	};
4589	struct nfs4_accessres res = {
4590		.server = server,
4591	};
4592	struct rpc_message msg = {
4593		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_ACCESS],
4594		.rpc_argp = &args,
4595		.rpc_resp = &res,
4596		.rpc_cred = cred,
4597	};
4598	int status = 0;
4599
4600	if (!nfs4_have_delegation(inode, FMODE_READ)) {
4601		res.fattr = nfs_alloc_fattr();
4602		if (res.fattr == NULL)
4603			return -ENOMEM;
4604		args.bitmask = server->cache_consistency_bitmask;
4605	}
4606	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
4607	if (!status) {
4608		nfs_access_set_mask(entry, res.access);
4609		if (res.fattr)
4610			nfs_refresh_inode(inode, res.fattr);
4611	}
4612	nfs_free_fattr(res.fattr);
4613	return status;
4614}
4615
4616static int nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry,
4617			    const struct cred *cred)
4618{
4619	struct nfs4_exception exception = {
4620		.interruptible = true,
4621	};
4622	int err;
4623	do {
4624		err = _nfs4_proc_access(inode, entry, cred);
4625		trace_nfs4_access(inode, err);
4626		err = nfs4_handle_exception(NFS_SERVER(inode), err,
4627				&exception);
4628	} while (exception.retry);
4629	return err;
4630}
4631
4632/*
4633 * TODO: For the time being, we don't try to get any attributes
4634 * along with any of the zero-copy operations READ, READDIR,
4635 * READLINK, WRITE.
4636 *
4637 * In the case of the first three, we want to put the GETATTR
4638 * after the read-type operation -- this is because it is hard
4639 * to predict the length of a GETATTR response in v4, and thus
4640 * align the READ data correctly.  This means that the GETATTR
4641 * may end up partially falling into the page cache, and we should
4642 * shift it into the 'tail' of the xdr_buf before processing.
4643 * To do this efficiently, we need to know the total length
4644 * of data received, which doesn't seem to be available outside
4645 * of the RPC layer.
4646 *
4647 * In the case of WRITE, we also want to put the GETATTR after
4648 * the operation -- in this case because we want to make sure
4649 * we get the post-operation mtime and size.
4650 *
4651 * Both of these changes to the XDR layer would in fact be quite
4652 * minor, but I decided to leave them for a subsequent patch.
4653 */
4654static int _nfs4_proc_readlink(struct inode *inode, struct page *page,
4655		unsigned int pgbase, unsigned int pglen)
4656{
4657	struct nfs4_readlink args = {
4658		.fh       = NFS_FH(inode),
4659		.pgbase	  = pgbase,
4660		.pglen    = pglen,
4661		.pages    = &page,
4662	};
4663	struct nfs4_readlink_res res;
4664	struct rpc_message msg = {
4665		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READLINK],
4666		.rpc_argp = &args,
4667		.rpc_resp = &res,
4668	};
4669
4670	return nfs4_call_sync(NFS_SERVER(inode)->client, NFS_SERVER(inode), &msg, &args.seq_args, &res.seq_res, 0);
4671}
4672
4673static int nfs4_proc_readlink(struct inode *inode, struct page *page,
4674		unsigned int pgbase, unsigned int pglen)
4675{
4676	struct nfs4_exception exception = {
4677		.interruptible = true,
4678	};
4679	int err;
4680	do {
4681		err = _nfs4_proc_readlink(inode, page, pgbase, pglen);
4682		trace_nfs4_readlink(inode, err);
4683		err = nfs4_handle_exception(NFS_SERVER(inode), err,
4684				&exception);
4685	} while (exception.retry);
4686	return err;
4687}
4688
4689/*
4690 * This is just for mknod.  open(O_CREAT) will always do ->open_context().
4691 */
4692static int
4693nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
4694		 int flags)
4695{
4696	struct nfs_server *server = NFS_SERVER(dir);
4697	struct nfs4_label l, *ilabel;
4698	struct nfs_open_context *ctx;
4699	struct nfs4_state *state;
4700	int status = 0;
4701
4702	ctx = alloc_nfs_open_context(dentry, FMODE_READ, NULL);
4703	if (IS_ERR(ctx))
4704		return PTR_ERR(ctx);
4705
4706	ilabel = nfs4_label_init_security(dir, dentry, sattr, &l);
4707
4708	if (!(server->attr_bitmask[2] & FATTR4_WORD2_MODE_UMASK))
4709		sattr->ia_mode &= ~current_umask();
4710	state = nfs4_do_open(dir, ctx, flags, sattr, ilabel, NULL);
4711	if (IS_ERR(state)) {
4712		status = PTR_ERR(state);
4713		goto out;
4714	}
4715out:
4716	nfs4_label_release_security(ilabel);
4717	put_nfs_open_context(ctx);
4718	return status;
4719}
4720
4721static int
4722_nfs4_proc_remove(struct inode *dir, const struct qstr *name, u32 ftype)
4723{
4724	struct nfs_server *server = NFS_SERVER(dir);
4725	struct nfs_removeargs args = {
4726		.fh = NFS_FH(dir),
4727		.name = *name,
4728	};
4729	struct nfs_removeres res = {
4730		.server = server,
4731	};
4732	struct rpc_message msg = {
4733		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE],
4734		.rpc_argp = &args,
4735		.rpc_resp = &res,
4736	};
4737	unsigned long timestamp = jiffies;
4738	int status;
4739
4740	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 1);
4741	if (status == 0) {
4742		spin_lock(&dir->i_lock);
4743		/* Removing a directory decrements nlink in the parent */
4744		if (ftype == NF4DIR && dir->i_nlink > 2)
4745			nfs4_dec_nlink_locked(dir);
4746		nfs4_update_changeattr_locked(dir, &res.cinfo, timestamp,
4747					      NFS_INO_INVALID_DATA);
4748		spin_unlock(&dir->i_lock);
4749	}
4750	return status;
4751}
4752
4753static int nfs4_proc_remove(struct inode *dir, struct dentry *dentry)
4754{
4755	struct nfs4_exception exception = {
4756		.interruptible = true,
4757	};
4758	struct inode *inode = d_inode(dentry);
4759	int err;
4760
4761	if (inode) {
4762		if (inode->i_nlink == 1)
4763			nfs4_inode_return_delegation(inode);
4764		else
4765			nfs4_inode_make_writeable(inode);
4766	}
4767	do {
4768		err = _nfs4_proc_remove(dir, &dentry->d_name, NF4REG);
4769		trace_nfs4_remove(dir, &dentry->d_name, err);
4770		err = nfs4_handle_exception(NFS_SERVER(dir), err,
4771				&exception);
4772	} while (exception.retry);
4773	return err;
4774}
4775
4776static int nfs4_proc_rmdir(struct inode *dir, const struct qstr *name)
4777{
4778	struct nfs4_exception exception = {
4779		.interruptible = true,
4780	};
4781	int err;
4782
4783	do {
4784		err = _nfs4_proc_remove(dir, name, NF4DIR);
4785		trace_nfs4_remove(dir, name, err);
4786		err = nfs4_handle_exception(NFS_SERVER(dir), err,
4787				&exception);
4788	} while (exception.retry);
4789	return err;
4790}
4791
4792static void nfs4_proc_unlink_setup(struct rpc_message *msg,
4793		struct dentry *dentry,
4794		struct inode *inode)
4795{
4796	struct nfs_removeargs *args = msg->rpc_argp;
4797	struct nfs_removeres *res = msg->rpc_resp;
4798
4799	res->server = NFS_SB(dentry->d_sb);
4800	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE];
4801	nfs4_init_sequence(&args->seq_args, &res->seq_res, 1, 0);
4802
4803	nfs_fattr_init(res->dir_attr);
4804
4805	if (inode) {
4806		nfs4_inode_return_delegation(inode);
4807		nfs_d_prune_case_insensitive_aliases(inode);
4808	}
4809}
4810
4811static void nfs4_proc_unlink_rpc_prepare(struct rpc_task *task, struct nfs_unlinkdata *data)
4812{
4813	nfs4_setup_sequence(NFS_SB(data->dentry->d_sb)->nfs_client,
4814			&data->args.seq_args,
4815			&data->res.seq_res,
4816			task);
4817}
4818
4819static int nfs4_proc_unlink_done(struct rpc_task *task, struct inode *dir)
4820{
4821	struct nfs_unlinkdata *data = task->tk_calldata;
4822	struct nfs_removeres *res = &data->res;
4823
4824	if (!nfs4_sequence_done(task, &res->seq_res))
4825		return 0;
4826	if (nfs4_async_handle_error(task, res->server, NULL,
4827				    &data->timeout) == -EAGAIN)
4828		return 0;
4829	if (task->tk_status == 0)
4830		nfs4_update_changeattr(dir, &res->cinfo,
4831				res->dir_attr->time_start,
4832				NFS_INO_INVALID_DATA);
4833	return 1;
4834}
4835
4836static void nfs4_proc_rename_setup(struct rpc_message *msg,
4837		struct dentry *old_dentry,
4838		struct dentry *new_dentry)
4839{
4840	struct nfs_renameargs *arg = msg->rpc_argp;
4841	struct nfs_renameres *res = msg->rpc_resp;
4842	struct inode *old_inode = d_inode(old_dentry);
4843	struct inode *new_inode = d_inode(new_dentry);
4844
4845	if (old_inode)
4846		nfs4_inode_make_writeable(old_inode);
4847	if (new_inode)
4848		nfs4_inode_return_delegation(new_inode);
4849	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RENAME];
4850	res->server = NFS_SB(old_dentry->d_sb);
4851	nfs4_init_sequence(&arg->seq_args, &res->seq_res, 1, 0);
4852}
4853
4854static void nfs4_proc_rename_rpc_prepare(struct rpc_task *task, struct nfs_renamedata *data)
4855{
4856	nfs4_setup_sequence(NFS_SERVER(data->old_dir)->nfs_client,
4857			&data->args.seq_args,
4858			&data->res.seq_res,
4859			task);
4860}
4861
4862static int nfs4_proc_rename_done(struct rpc_task *task, struct inode *old_dir,
4863				 struct inode *new_dir)
4864{
4865	struct nfs_renamedata *data = task->tk_calldata;
4866	struct nfs_renameres *res = &data->res;
4867
4868	if (!nfs4_sequence_done(task, &res->seq_res))
4869		return 0;
4870	if (nfs4_async_handle_error(task, res->server, NULL, &data->timeout) == -EAGAIN)
4871		return 0;
4872
4873	if (task->tk_status == 0) {
4874		nfs_d_prune_case_insensitive_aliases(d_inode(data->old_dentry));
4875		if (new_dir != old_dir) {
4876			/* Note: If we moved a directory, nlink will change */
4877			nfs4_update_changeattr(old_dir, &res->old_cinfo,
4878					res->old_fattr->time_start,
4879					NFS_INO_INVALID_NLINK |
4880					    NFS_INO_INVALID_DATA);
4881			nfs4_update_changeattr(new_dir, &res->new_cinfo,
4882					res->new_fattr->time_start,
4883					NFS_INO_INVALID_NLINK |
4884					    NFS_INO_INVALID_DATA);
4885		} else
4886			nfs4_update_changeattr(old_dir, &res->old_cinfo,
4887					res->old_fattr->time_start,
4888					NFS_INO_INVALID_DATA);
4889	}
4890	return 1;
4891}
4892
4893static int _nfs4_proc_link(struct inode *inode, struct inode *dir, const struct qstr *name)
4894{
4895	struct nfs_server *server = NFS_SERVER(inode);
4896	__u32 bitmask[NFS4_BITMASK_SZ];
4897	struct nfs4_link_arg arg = {
4898		.fh     = NFS_FH(inode),
4899		.dir_fh = NFS_FH(dir),
4900		.name   = name,
4901		.bitmask = bitmask,
4902	};
4903	struct nfs4_link_res res = {
4904		.server = server,
4905	};
4906	struct rpc_message msg = {
4907		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LINK],
4908		.rpc_argp = &arg,
4909		.rpc_resp = &res,
4910	};
4911	int status = -ENOMEM;
4912
4913	res.fattr = nfs_alloc_fattr_with_label(server);
4914	if (res.fattr == NULL)
4915		goto out;
4916
4917	nfs4_inode_make_writeable(inode);
4918	nfs4_bitmap_copy_adjust(bitmask, nfs4_bitmask(server, res.fattr->label), inode,
4919				NFS_INO_INVALID_CHANGE);
4920	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
4921	if (!status) {
4922		nfs4_update_changeattr(dir, &res.cinfo, res.fattr->time_start,
4923				       NFS_INO_INVALID_DATA);
4924		nfs4_inc_nlink(inode);
4925		status = nfs_post_op_update_inode(inode, res.fattr);
4926		if (!status)
4927			nfs_setsecurity(inode, res.fattr);
4928	}
4929
4930out:
4931	nfs_free_fattr(res.fattr);
4932	return status;
4933}
4934
4935static int nfs4_proc_link(struct inode *inode, struct inode *dir, const struct qstr *name)
4936{
4937	struct nfs4_exception exception = {
4938		.interruptible = true,
4939	};
4940	int err;
4941	do {
4942		err = nfs4_handle_exception(NFS_SERVER(inode),
4943				_nfs4_proc_link(inode, dir, name),
4944				&exception);
4945	} while (exception.retry);
4946	return err;
4947}
4948
4949struct nfs4_createdata {
4950	struct rpc_message msg;
4951	struct nfs4_create_arg arg;
4952	struct nfs4_create_res res;
4953	struct nfs_fh fh;
4954	struct nfs_fattr fattr;
4955};
4956
4957static struct nfs4_createdata *nfs4_alloc_createdata(struct inode *dir,
4958		const struct qstr *name, struct iattr *sattr, u32 ftype)
4959{
4960	struct nfs4_createdata *data;
4961
4962	data = kzalloc(sizeof(*data), GFP_KERNEL);
4963	if (data != NULL) {
4964		struct nfs_server *server = NFS_SERVER(dir);
4965
4966		data->fattr.label = nfs4_label_alloc(server, GFP_KERNEL);
4967		if (IS_ERR(data->fattr.label))
4968			goto out_free;
4969
4970		data->msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CREATE];
4971		data->msg.rpc_argp = &data->arg;
4972		data->msg.rpc_resp = &data->res;
4973		data->arg.dir_fh = NFS_FH(dir);
4974		data->arg.server = server;
4975		data->arg.name = name;
4976		data->arg.attrs = sattr;
4977		data->arg.ftype = ftype;
4978		data->arg.bitmask = nfs4_bitmask(server, data->fattr.label);
4979		data->arg.umask = current_umask();
4980		data->res.server = server;
4981		data->res.fh = &data->fh;
4982		data->res.fattr = &data->fattr;
4983		nfs_fattr_init(data->res.fattr);
4984	}
4985	return data;
4986out_free:
4987	kfree(data);
4988	return NULL;
4989}
4990
4991static int nfs4_do_create(struct inode *dir, struct dentry *dentry, struct nfs4_createdata *data)
4992{
4993	int status = nfs4_call_sync(NFS_SERVER(dir)->client, NFS_SERVER(dir), &data->msg,
4994				    &data->arg.seq_args, &data->res.seq_res, 1);
4995	if (status == 0) {
4996		spin_lock(&dir->i_lock);
4997		/* Creating a directory bumps nlink in the parent */
4998		if (data->arg.ftype == NF4DIR)
4999			nfs4_inc_nlink_locked(dir);
5000		nfs4_update_changeattr_locked(dir, &data->res.dir_cinfo,
5001					      data->res.fattr->time_start,
5002					      NFS_INO_INVALID_DATA);
5003		spin_unlock(&dir->i_lock);
5004		status = nfs_instantiate(dentry, data->res.fh, data->res.fattr);
5005	}
5006	return status;
5007}
5008
5009static void nfs4_free_createdata(struct nfs4_createdata *data)
5010{
5011	nfs4_label_free(data->fattr.label);
5012	kfree(data);
5013}
5014
5015static int _nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
5016		struct page *page, unsigned int len, struct iattr *sattr,
5017		struct nfs4_label *label)
5018{
5019	struct nfs4_createdata *data;
5020	int status = -ENAMETOOLONG;
5021
5022	if (len > NFS4_MAXPATHLEN)
5023		goto out;
5024
5025	status = -ENOMEM;
5026	data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4LNK);
5027	if (data == NULL)
5028		goto out;
5029
5030	data->msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SYMLINK];
5031	data->arg.u.symlink.pages = &page;
5032	data->arg.u.symlink.len = len;
5033	data->arg.label = label;
5034
5035	status = nfs4_do_create(dir, dentry, data);
5036
5037	nfs4_free_createdata(data);
5038out:
5039	return status;
5040}
5041
5042static int nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
5043		struct page *page, unsigned int len, struct iattr *sattr)
5044{
5045	struct nfs4_exception exception = {
5046		.interruptible = true,
5047	};
5048	struct nfs4_label l, *label;
5049	int err;
5050
5051	label = nfs4_label_init_security(dir, dentry, sattr, &l);
5052
5053	do {
5054		err = _nfs4_proc_symlink(dir, dentry, page, len, sattr, label);
5055		trace_nfs4_symlink(dir, &dentry->d_name, err);
5056		err = nfs4_handle_exception(NFS_SERVER(dir), err,
5057				&exception);
5058	} while (exception.retry);
5059
5060	nfs4_label_release_security(label);
5061	return err;
5062}
5063
5064static int _nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
5065		struct iattr *sattr, struct nfs4_label *label)
5066{
5067	struct nfs4_createdata *data;
5068	int status = -ENOMEM;
5069
5070	data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4DIR);
5071	if (data == NULL)
5072		goto out;
5073
5074	data->arg.label = label;
5075	status = nfs4_do_create(dir, dentry, data);
5076
5077	nfs4_free_createdata(data);
5078out:
5079	return status;
5080}
5081
5082static int nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
5083		struct iattr *sattr)
5084{
5085	struct nfs_server *server = NFS_SERVER(dir);
5086	struct nfs4_exception exception = {
5087		.interruptible = true,
5088	};
5089	struct nfs4_label l, *label;
5090	int err;
5091
5092	label = nfs4_label_init_security(dir, dentry, sattr, &l);
5093
5094	if (!(server->attr_bitmask[2] & FATTR4_WORD2_MODE_UMASK))
5095		sattr->ia_mode &= ~current_umask();
5096	do {
5097		err = _nfs4_proc_mkdir(dir, dentry, sattr, label);
5098		trace_nfs4_mkdir(dir, &dentry->d_name, err);
5099		err = nfs4_handle_exception(NFS_SERVER(dir), err,
5100				&exception);
5101	} while (exception.retry);
5102	nfs4_label_release_security(label);
5103
5104	return err;
5105}
5106
5107static int _nfs4_proc_readdir(struct nfs_readdir_arg *nr_arg,
5108			      struct nfs_readdir_res *nr_res)
5109{
5110	struct inode		*dir = d_inode(nr_arg->dentry);
5111	struct nfs_server	*server = NFS_SERVER(dir);
5112	struct nfs4_readdir_arg args = {
5113		.fh = NFS_FH(dir),
5114		.pages = nr_arg->pages,
5115		.pgbase = 0,
5116		.count = nr_arg->page_len,
5117		.plus = nr_arg->plus,
5118	};
5119	struct nfs4_readdir_res res;
5120	struct rpc_message msg = {
5121		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READDIR],
5122		.rpc_argp = &args,
5123		.rpc_resp = &res,
5124		.rpc_cred = nr_arg->cred,
5125	};
5126	int			status;
5127
5128	dprintk("%s: dentry = %pd2, cookie = %llu\n", __func__,
5129		nr_arg->dentry, (unsigned long long)nr_arg->cookie);
5130	if (!(server->caps & NFS_CAP_SECURITY_LABEL))
5131		args.bitmask = server->attr_bitmask_nl;
5132	else
5133		args.bitmask = server->attr_bitmask;
5134
5135	nfs4_setup_readdir(nr_arg->cookie, nr_arg->verf, nr_arg->dentry, &args);
5136	res.pgbase = args.pgbase;
5137	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args,
5138			&res.seq_res, 0);
5139	if (status >= 0) {
5140		memcpy(nr_res->verf, res.verifier.data, NFS4_VERIFIER_SIZE);
5141		status += args.pgbase;
5142	}
5143
5144	nfs_invalidate_atime(dir);
5145
5146	dprintk("%s: returns %d\n", __func__, status);
5147	return status;
5148}
5149
5150static int nfs4_proc_readdir(struct nfs_readdir_arg *arg,
5151			     struct nfs_readdir_res *res)
5152{
5153	struct nfs4_exception exception = {
5154		.interruptible = true,
5155	};
5156	int err;
5157	do {
5158		err = _nfs4_proc_readdir(arg, res);
5159		trace_nfs4_readdir(d_inode(arg->dentry), err);
5160		err = nfs4_handle_exception(NFS_SERVER(d_inode(arg->dentry)),
5161					    err, &exception);
5162	} while (exception.retry);
5163	return err;
5164}
5165
5166static int _nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
5167		struct iattr *sattr, struct nfs4_label *label, dev_t rdev)
5168{
5169	struct nfs4_createdata *data;
5170	int mode = sattr->ia_mode;
5171	int status = -ENOMEM;
5172
5173	data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4SOCK);
5174	if (data == NULL)
5175		goto out;
5176
5177	if (S_ISFIFO(mode))
5178		data->arg.ftype = NF4FIFO;
5179	else if (S_ISBLK(mode)) {
5180		data->arg.ftype = NF4BLK;
5181		data->arg.u.device.specdata1 = MAJOR(rdev);
5182		data->arg.u.device.specdata2 = MINOR(rdev);
5183	}
5184	else if (S_ISCHR(mode)) {
5185		data->arg.ftype = NF4CHR;
5186		data->arg.u.device.specdata1 = MAJOR(rdev);
5187		data->arg.u.device.specdata2 = MINOR(rdev);
5188	} else if (!S_ISSOCK(mode)) {
5189		status = -EINVAL;
5190		goto out_free;
5191	}
5192
5193	data->arg.label = label;
5194	status = nfs4_do_create(dir, dentry, data);
5195out_free:
5196	nfs4_free_createdata(data);
5197out:
5198	return status;
5199}
5200
5201static int nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
5202		struct iattr *sattr, dev_t rdev)
5203{
5204	struct nfs_server *server = NFS_SERVER(dir);
5205	struct nfs4_exception exception = {
5206		.interruptible = true,
5207	};
5208	struct nfs4_label l, *label;
5209	int err;
5210
5211	label = nfs4_label_init_security(dir, dentry, sattr, &l);
5212
5213	if (!(server->attr_bitmask[2] & FATTR4_WORD2_MODE_UMASK))
5214		sattr->ia_mode &= ~current_umask();
5215	do {
5216		err = _nfs4_proc_mknod(dir, dentry, sattr, label, rdev);
5217		trace_nfs4_mknod(dir, &dentry->d_name, err);
5218		err = nfs4_handle_exception(NFS_SERVER(dir), err,
5219				&exception);
5220	} while (exception.retry);
5221
5222	nfs4_label_release_security(label);
5223
5224	return err;
5225}
5226
5227static int _nfs4_proc_statfs(struct nfs_server *server, struct nfs_fh *fhandle,
5228		 struct nfs_fsstat *fsstat)
5229{
5230	struct nfs4_statfs_arg args = {
5231		.fh = fhandle,
5232		.bitmask = server->attr_bitmask,
5233	};
5234	struct nfs4_statfs_res res = {
5235		.fsstat = fsstat,
5236	};
5237	struct rpc_message msg = {
5238		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_STATFS],
5239		.rpc_argp = &args,
5240		.rpc_resp = &res,
5241	};
5242
5243	nfs_fattr_init(fsstat->fattr);
5244	return  nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
5245}
5246
5247static int nfs4_proc_statfs(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsstat *fsstat)
5248{
5249	struct nfs4_exception exception = {
5250		.interruptible = true,
5251	};
5252	int err;
5253	do {
5254		err = nfs4_handle_exception(server,
5255				_nfs4_proc_statfs(server, fhandle, fsstat),
5256				&exception);
5257	} while (exception.retry);
5258	return err;
5259}
5260
5261static int _nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle,
5262		struct nfs_fsinfo *fsinfo)
5263{
5264	struct nfs4_fsinfo_arg args = {
5265		.fh = fhandle,
5266		.bitmask = server->attr_bitmask,
5267	};
5268	struct nfs4_fsinfo_res res = {
5269		.fsinfo = fsinfo,
5270	};
5271	struct rpc_message msg = {
5272		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FSINFO],
5273		.rpc_argp = &args,
5274		.rpc_resp = &res,
5275	};
5276
5277	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
5278}
5279
5280static int nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo)
5281{
5282	struct nfs4_exception exception = {
5283		.interruptible = true,
5284	};
5285	int err;
5286
5287	do {
5288		err = _nfs4_do_fsinfo(server, fhandle, fsinfo);
5289		trace_nfs4_fsinfo(server, fhandle, fsinfo->fattr, err);
5290		if (err == 0) {
5291			nfs4_set_lease_period(server->nfs_client, fsinfo->lease_time * HZ);
5292			break;
5293		}
5294		err = nfs4_handle_exception(server, err, &exception);
5295	} while (exception.retry);
5296	return err;
5297}
5298
5299static int nfs4_proc_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo)
5300{
5301	int error;
5302
5303	nfs_fattr_init(fsinfo->fattr);
5304	error = nfs4_do_fsinfo(server, fhandle, fsinfo);
5305	if (error == 0) {
5306		/* block layout checks this! */
5307		server->pnfs_blksize = fsinfo->blksize;
5308		set_pnfs_layoutdriver(server, fhandle, fsinfo);
5309	}
5310
5311	return error;
5312}
5313
5314static int _nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle,
5315		struct nfs_pathconf *pathconf)
5316{
5317	struct nfs4_pathconf_arg args = {
5318		.fh = fhandle,
5319		.bitmask = server->attr_bitmask,
5320	};
5321	struct nfs4_pathconf_res res = {
5322		.pathconf = pathconf,
5323	};
5324	struct rpc_message msg = {
5325		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_PATHCONF],
5326		.rpc_argp = &args,
5327		.rpc_resp = &res,
5328	};
5329
5330	/* None of the pathconf attributes are mandatory to implement */
5331	if ((args.bitmask[0] & nfs4_pathconf_bitmap[0]) == 0) {
5332		memset(pathconf, 0, sizeof(*pathconf));
5333		return 0;
5334	}
5335
5336	nfs_fattr_init(pathconf->fattr);
5337	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
5338}
5339
5340static int nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle,
5341		struct nfs_pathconf *pathconf)
5342{
5343	struct nfs4_exception exception = {
5344		.interruptible = true,
5345	};
5346	int err;
5347
5348	do {
5349		err = nfs4_handle_exception(server,
5350				_nfs4_proc_pathconf(server, fhandle, pathconf),
5351				&exception);
5352	} while (exception.retry);
5353	return err;
5354}
5355
5356int nfs4_set_rw_stateid(nfs4_stateid *stateid,
5357		const struct nfs_open_context *ctx,
5358		const struct nfs_lock_context *l_ctx,
5359		fmode_t fmode)
5360{
5361	return nfs4_select_rw_stateid(ctx->state, fmode, l_ctx, stateid, NULL);
5362}
5363EXPORT_SYMBOL_GPL(nfs4_set_rw_stateid);
5364
5365static bool nfs4_stateid_is_current(nfs4_stateid *stateid,
5366		const struct nfs_open_context *ctx,
5367		const struct nfs_lock_context *l_ctx,
5368		fmode_t fmode)
5369{
5370	nfs4_stateid _current_stateid;
5371
5372	/* If the current stateid represents a lost lock, then exit */
5373	if (nfs4_set_rw_stateid(&_current_stateid, ctx, l_ctx, fmode) == -EIO)
5374		return true;
5375	return nfs4_stateid_match(stateid, &_current_stateid);
5376}
5377
5378static bool nfs4_error_stateid_expired(int err)
5379{
5380	switch (err) {
5381	case -NFS4ERR_DELEG_REVOKED:
5382	case -NFS4ERR_ADMIN_REVOKED:
5383	case -NFS4ERR_BAD_STATEID:
5384	case -NFS4ERR_STALE_STATEID:
5385	case -NFS4ERR_OLD_STATEID:
5386	case -NFS4ERR_OPENMODE:
5387	case -NFS4ERR_EXPIRED:
5388		return true;
5389	}
5390	return false;
5391}
5392
5393static int nfs4_read_done_cb(struct rpc_task *task, struct nfs_pgio_header *hdr)
5394{
5395	struct nfs_server *server = NFS_SERVER(hdr->inode);
5396
5397	trace_nfs4_read(hdr, task->tk_status);
5398	if (task->tk_status < 0) {
5399		struct nfs4_exception exception = {
5400			.inode = hdr->inode,
5401			.state = hdr->args.context->state,
5402			.stateid = &hdr->args.stateid,
5403		};
5404		task->tk_status = nfs4_async_handle_exception(task,
5405				server, task->tk_status, &exception);
5406		if (exception.retry) {
5407			rpc_restart_call_prepare(task);
5408			return -EAGAIN;
5409		}
5410	}
5411
5412	if (task->tk_status > 0)
5413		renew_lease(server, hdr->timestamp);
5414	return 0;
5415}
5416
5417static bool nfs4_read_stateid_changed(struct rpc_task *task,
5418		struct nfs_pgio_args *args)
5419{
5420
5421	if (!nfs4_error_stateid_expired(task->tk_status) ||
5422		nfs4_stateid_is_current(&args->stateid,
5423				args->context,
5424				args->lock_context,
5425				FMODE_READ))
5426		return false;
5427	rpc_restart_call_prepare(task);
5428	return true;
5429}
5430
5431static bool nfs4_read_plus_not_supported(struct rpc_task *task,
5432					 struct nfs_pgio_header *hdr)
5433{
5434	struct nfs_server *server = NFS_SERVER(hdr->inode);
5435	struct rpc_message *msg = &task->tk_msg;
5436
5437	if (msg->rpc_proc == &nfs4_procedures[NFSPROC4_CLNT_READ_PLUS] &&
5438	    server->caps & NFS_CAP_READ_PLUS && task->tk_status == -ENOTSUPP) {
5439		server->caps &= ~NFS_CAP_READ_PLUS;
5440		msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READ];
5441		rpc_restart_call_prepare(task);
5442		return true;
5443	}
5444	return false;
5445}
5446
5447static int nfs4_read_done(struct rpc_task *task, struct nfs_pgio_header *hdr)
5448{
5449	if (!nfs4_sequence_done(task, &hdr->res.seq_res))
5450		return -EAGAIN;
5451	if (nfs4_read_stateid_changed(task, &hdr->args))
5452		return -EAGAIN;
5453	if (nfs4_read_plus_not_supported(task, hdr))
5454		return -EAGAIN;
5455	if (task->tk_status > 0)
5456		nfs_invalidate_atime(hdr->inode);
5457	return hdr->pgio_done_cb ? hdr->pgio_done_cb(task, hdr) :
5458				    nfs4_read_done_cb(task, hdr);
5459}
5460
5461#if defined CONFIG_NFS_V4_2 && defined CONFIG_NFS_V4_2_READ_PLUS
5462static bool nfs42_read_plus_support(struct nfs_pgio_header *hdr,
5463				    struct rpc_message *msg)
5464{
5465	/* Note: We don't use READ_PLUS with pNFS yet */
5466	if (nfs_server_capable(hdr->inode, NFS_CAP_READ_PLUS) && !hdr->ds_clp) {
5467		msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READ_PLUS];
5468		return nfs_read_alloc_scratch(hdr, READ_PLUS_SCRATCH_SIZE);
5469	}
5470	return false;
5471}
5472#else
5473static bool nfs42_read_plus_support(struct nfs_pgio_header *hdr,
5474				    struct rpc_message *msg)
5475{
5476	return false;
5477}
5478#endif /* CONFIG_NFS_V4_2 */
5479
5480static void nfs4_proc_read_setup(struct nfs_pgio_header *hdr,
5481				 struct rpc_message *msg)
5482{
5483	hdr->timestamp   = jiffies;
5484	if (!hdr->pgio_done_cb)
5485		hdr->pgio_done_cb = nfs4_read_done_cb;
5486	if (!nfs42_read_plus_support(hdr, msg))
5487		msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READ];
5488	nfs4_init_sequence(&hdr->args.seq_args, &hdr->res.seq_res, 0, 0);
5489}
5490
5491static int nfs4_proc_pgio_rpc_prepare(struct rpc_task *task,
5492				      struct nfs_pgio_header *hdr)
5493{
5494	if (nfs4_setup_sequence(NFS_SERVER(hdr->inode)->nfs_client,
5495			&hdr->args.seq_args,
5496			&hdr->res.seq_res,
5497			task))
5498		return 0;
5499	if (nfs4_set_rw_stateid(&hdr->args.stateid, hdr->args.context,
5500				hdr->args.lock_context,
5501				hdr->rw_mode) == -EIO)
5502		return -EIO;
5503	if (unlikely(test_bit(NFS_CONTEXT_BAD, &hdr->args.context->flags)))
5504		return -EIO;
5505	return 0;
5506}
5507
5508static int nfs4_write_done_cb(struct rpc_task *task,
5509			      struct nfs_pgio_header *hdr)
5510{
5511	struct inode *inode = hdr->inode;
5512
5513	trace_nfs4_write(hdr, task->tk_status);
5514	if (task->tk_status < 0) {
5515		struct nfs4_exception exception = {
5516			.inode = hdr->inode,
5517			.state = hdr->args.context->state,
5518			.stateid = &hdr->args.stateid,
5519		};
5520		task->tk_status = nfs4_async_handle_exception(task,
5521				NFS_SERVER(inode), task->tk_status,
5522				&exception);
5523		if (exception.retry) {
5524			rpc_restart_call_prepare(task);
5525			return -EAGAIN;
5526		}
5527	}
5528	if (task->tk_status >= 0) {
5529		renew_lease(NFS_SERVER(inode), hdr->timestamp);
5530		nfs_writeback_update_inode(hdr);
5531	}
5532	return 0;
5533}
5534
5535static bool nfs4_write_stateid_changed(struct rpc_task *task,
5536		struct nfs_pgio_args *args)
5537{
5538
5539	if (!nfs4_error_stateid_expired(task->tk_status) ||
5540		nfs4_stateid_is_current(&args->stateid,
5541				args->context,
5542				args->lock_context,
5543				FMODE_WRITE))
5544		return false;
5545	rpc_restart_call_prepare(task);
5546	return true;
5547}
5548
5549static int nfs4_write_done(struct rpc_task *task, struct nfs_pgio_header *hdr)
5550{
5551	if (!nfs4_sequence_done(task, &hdr->res.seq_res))
5552		return -EAGAIN;
5553	if (nfs4_write_stateid_changed(task, &hdr->args))
5554		return -EAGAIN;
5555	return hdr->pgio_done_cb ? hdr->pgio_done_cb(task, hdr) :
5556		nfs4_write_done_cb(task, hdr);
5557}
5558
5559static
5560bool nfs4_write_need_cache_consistency_data(struct nfs_pgio_header *hdr)
5561{
5562	/* Don't request attributes for pNFS or O_DIRECT writes */
5563	if (hdr->ds_clp != NULL || hdr->dreq != NULL)
5564		return false;
5565	/* Otherwise, request attributes if and only if we don't hold
5566	 * a delegation
5567	 */
5568	return nfs4_have_delegation(hdr->inode, FMODE_READ) == 0;
5569}
5570
5571void nfs4_bitmask_set(__u32 bitmask[], const __u32 src[],
5572		      struct inode *inode, unsigned long cache_validity)
5573{
5574	struct nfs_server *server = NFS_SERVER(inode);
5575	unsigned int i;
5576
5577	memcpy(bitmask, src, sizeof(*bitmask) * NFS4_BITMASK_SZ);
5578	cache_validity |= READ_ONCE(NFS_I(inode)->cache_validity);
5579
5580	if (cache_validity & NFS_INO_INVALID_CHANGE)
5581		bitmask[0] |= FATTR4_WORD0_CHANGE;
5582	if (cache_validity & NFS_INO_INVALID_ATIME)
5583		bitmask[1] |= FATTR4_WORD1_TIME_ACCESS;
5584	if (cache_validity & NFS_INO_INVALID_MODE)
5585		bitmask[1] |= FATTR4_WORD1_MODE;
5586	if (cache_validity & NFS_INO_INVALID_OTHER)
5587		bitmask[1] |= FATTR4_WORD1_OWNER | FATTR4_WORD1_OWNER_GROUP;
5588	if (cache_validity & NFS_INO_INVALID_NLINK)
5589		bitmask[1] |= FATTR4_WORD1_NUMLINKS;
5590	if (cache_validity & NFS_INO_INVALID_CTIME)
5591		bitmask[1] |= FATTR4_WORD1_TIME_METADATA;
5592	if (cache_validity & NFS_INO_INVALID_MTIME)
5593		bitmask[1] |= FATTR4_WORD1_TIME_MODIFY;
5594	if (cache_validity & NFS_INO_INVALID_BLOCKS)
5595		bitmask[1] |= FATTR4_WORD1_SPACE_USED;
5596
5597	if (cache_validity & NFS_INO_INVALID_SIZE)
5598		bitmask[0] |= FATTR4_WORD0_SIZE;
5599
5600	for (i = 0; i < NFS4_BITMASK_SZ; i++)
5601		bitmask[i] &= server->attr_bitmask[i];
5602}
5603
5604static void nfs4_proc_write_setup(struct nfs_pgio_header *hdr,
5605				  struct rpc_message *msg,
5606				  struct rpc_clnt **clnt)
5607{
5608	struct nfs_server *server = NFS_SERVER(hdr->inode);
5609
5610	if (!nfs4_write_need_cache_consistency_data(hdr)) {
5611		hdr->args.bitmask = NULL;
5612		hdr->res.fattr = NULL;
5613	} else {
5614		nfs4_bitmask_set(hdr->args.bitmask_store,
5615				 server->cache_consistency_bitmask,
5616				 hdr->inode, NFS_INO_INVALID_BLOCKS);
5617		hdr->args.bitmask = hdr->args.bitmask_store;
5618	}
5619
5620	if (!hdr->pgio_done_cb)
5621		hdr->pgio_done_cb = nfs4_write_done_cb;
5622	hdr->res.server = server;
5623	hdr->timestamp   = jiffies;
5624
5625	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_WRITE];
5626	nfs4_init_sequence(&hdr->args.seq_args, &hdr->res.seq_res, 0, 0);
5627	nfs4_state_protect_write(hdr->ds_clp ? hdr->ds_clp : server->nfs_client, clnt, msg, hdr);
5628}
5629
5630static void nfs4_proc_commit_rpc_prepare(struct rpc_task *task, struct nfs_commit_data *data)
5631{
5632	nfs4_setup_sequence(NFS_SERVER(data->inode)->nfs_client,
5633			&data->args.seq_args,
5634			&data->res.seq_res,
5635			task);
5636}
5637
5638static int nfs4_commit_done_cb(struct rpc_task *task, struct nfs_commit_data *data)
5639{
5640	struct inode *inode = data->inode;
5641
5642	trace_nfs4_commit(data, task->tk_status);
5643	if (nfs4_async_handle_error(task, NFS_SERVER(inode),
5644				    NULL, NULL) == -EAGAIN) {
5645		rpc_restart_call_prepare(task);
5646		return -EAGAIN;
5647	}
5648	return 0;
5649}
5650
5651static int nfs4_commit_done(struct rpc_task *task, struct nfs_commit_data *data)
5652{
5653	if (!nfs4_sequence_done(task, &data->res.seq_res))
5654		return -EAGAIN;
5655	return data->commit_done_cb(task, data);
5656}
5657
5658static void nfs4_proc_commit_setup(struct nfs_commit_data *data, struct rpc_message *msg,
5659				   struct rpc_clnt **clnt)
5660{
5661	struct nfs_server *server = NFS_SERVER(data->inode);
5662
5663	if (data->commit_done_cb == NULL)
5664		data->commit_done_cb = nfs4_commit_done_cb;
5665	data->res.server = server;
5666	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_COMMIT];
5667	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1, 0);
5668	nfs4_state_protect(data->ds_clp ? data->ds_clp : server->nfs_client,
5669			NFS_SP4_MACH_CRED_COMMIT, clnt, msg);
5670}
5671
5672static int _nfs4_proc_commit(struct file *dst, struct nfs_commitargs *args,
5673				struct nfs_commitres *res)
5674{
5675	struct inode *dst_inode = file_inode(dst);
5676	struct nfs_server *server = NFS_SERVER(dst_inode);
5677	struct rpc_message msg = {
5678		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_COMMIT],
5679		.rpc_argp = args,
5680		.rpc_resp = res,
5681	};
5682
5683	args->fh = NFS_FH(dst_inode);
5684	return nfs4_call_sync(server->client, server, &msg,
5685			&args->seq_args, &res->seq_res, 1);
5686}
5687
5688int nfs4_proc_commit(struct file *dst, __u64 offset, __u32 count, struct nfs_commitres *res)
5689{
5690	struct nfs_commitargs args = {
5691		.offset = offset,
5692		.count = count,
5693	};
5694	struct nfs_server *dst_server = NFS_SERVER(file_inode(dst));
5695	struct nfs4_exception exception = { };
5696	int status;
5697
5698	do {
5699		status = _nfs4_proc_commit(dst, &args, res);
5700		status = nfs4_handle_exception(dst_server, status, &exception);
5701	} while (exception.retry);
5702
5703	return status;
5704}
5705
5706struct nfs4_renewdata {
5707	struct nfs_client	*client;
5708	unsigned long		timestamp;
5709};
5710
5711/*
5712 * nfs4_proc_async_renew(): This is not one of the nfs_rpc_ops; it is a special
5713 * standalone procedure for queueing an asynchronous RENEW.
5714 */
5715static void nfs4_renew_release(void *calldata)
5716{
5717	struct nfs4_renewdata *data = calldata;
5718	struct nfs_client *clp = data->client;
5719
5720	if (refcount_read(&clp->cl_count) > 1)
5721		nfs4_schedule_state_renewal(clp);
5722	nfs_put_client(clp);
5723	kfree(data);
5724}
5725
5726static void nfs4_renew_done(struct rpc_task *task, void *calldata)
5727{
5728	struct nfs4_renewdata *data = calldata;
5729	struct nfs_client *clp = data->client;
5730	unsigned long timestamp = data->timestamp;
5731
5732	trace_nfs4_renew_async(clp, task->tk_status);
5733	switch (task->tk_status) {
5734	case 0:
5735		break;
5736	case -NFS4ERR_LEASE_MOVED:
5737		nfs4_schedule_lease_moved_recovery(clp);
5738		break;
5739	default:
5740		/* Unless we're shutting down, schedule state recovery! */
5741		if (test_bit(NFS_CS_RENEWD, &clp->cl_res_state) == 0)
5742			return;
5743		if (task->tk_status != NFS4ERR_CB_PATH_DOWN) {
5744			nfs4_schedule_lease_recovery(clp);
5745			return;
5746		}
5747		nfs4_schedule_path_down_recovery(clp);
5748	}
5749	do_renew_lease(clp, timestamp);
5750}
5751
5752static const struct rpc_call_ops nfs4_renew_ops = {
5753	.rpc_call_done = nfs4_renew_done,
5754	.rpc_release = nfs4_renew_release,
5755};
5756
5757static int nfs4_proc_async_renew(struct nfs_client *clp, const struct cred *cred, unsigned renew_flags)
5758{
5759	struct rpc_message msg = {
5760		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
5761		.rpc_argp	= clp,
5762		.rpc_cred	= cred,
5763	};
5764	struct nfs4_renewdata *data;
5765
5766	if (renew_flags == 0)
5767		return 0;
5768	if (!refcount_inc_not_zero(&clp->cl_count))
5769		return -EIO;
5770	data = kmalloc(sizeof(*data), GFP_NOFS);
5771	if (data == NULL) {
5772		nfs_put_client(clp);
5773		return -ENOMEM;
5774	}
5775	data->client = clp;
5776	data->timestamp = jiffies;
5777	return rpc_call_async(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT,
5778			&nfs4_renew_ops, data);
5779}
5780
5781static int nfs4_proc_renew(struct nfs_client *clp, const struct cred *cred)
5782{
5783	struct rpc_message msg = {
5784		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
5785		.rpc_argp	= clp,
5786		.rpc_cred	= cred,
5787	};
5788	unsigned long now = jiffies;
5789	int status;
5790
5791	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
5792	if (status < 0)
5793		return status;
5794	do_renew_lease(clp, now);
5795	return 0;
5796}
5797
5798static bool nfs4_server_supports_acls(const struct nfs_server *server,
5799				      enum nfs4_acl_type type)
5800{
5801	switch (type) {
5802	default:
5803		return server->attr_bitmask[0] & FATTR4_WORD0_ACL;
5804	case NFS4ACL_DACL:
5805		return server->attr_bitmask[1] & FATTR4_WORD1_DACL;
5806	case NFS4ACL_SACL:
5807		return server->attr_bitmask[1] & FATTR4_WORD1_SACL;
5808	}
5809}
5810
5811/* Assuming that XATTR_SIZE_MAX is a multiple of PAGE_SIZE, and that
5812 * it's OK to put sizeof(void) * (XATTR_SIZE_MAX/PAGE_SIZE) bytes on
5813 * the stack.
5814 */
5815#define NFS4ACL_MAXPAGES DIV_ROUND_UP(XATTR_SIZE_MAX, PAGE_SIZE)
5816
5817int nfs4_buf_to_pages_noslab(const void *buf, size_t buflen,
5818		struct page **pages)
5819{
5820	struct page *newpage, **spages;
5821	int rc = 0;
5822	size_t len;
5823	spages = pages;
5824
5825	do {
5826		len = min_t(size_t, PAGE_SIZE, buflen);
5827		newpage = alloc_page(GFP_KERNEL);
5828
5829		if (newpage == NULL)
5830			goto unwind;
5831		memcpy(page_address(newpage), buf, len);
5832		buf += len;
5833		buflen -= len;
5834		*pages++ = newpage;
5835		rc++;
5836	} while (buflen != 0);
5837
5838	return rc;
5839
5840unwind:
5841	for(; rc > 0; rc--)
5842		__free_page(spages[rc-1]);
5843	return -ENOMEM;
5844}
5845
5846struct nfs4_cached_acl {
5847	enum nfs4_acl_type type;
5848	int cached;
5849	size_t len;
5850	char data[];
5851};
5852
5853static void nfs4_set_cached_acl(struct inode *inode, struct nfs4_cached_acl *acl)
5854{
5855	struct nfs_inode *nfsi = NFS_I(inode);
5856
5857	spin_lock(&inode->i_lock);
5858	kfree(nfsi->nfs4_acl);
5859	nfsi->nfs4_acl = acl;
5860	spin_unlock(&inode->i_lock);
5861}
5862
5863static void nfs4_zap_acl_attr(struct inode *inode)
5864{
5865	nfs4_set_cached_acl(inode, NULL);
5866}
5867
5868static ssize_t nfs4_read_cached_acl(struct inode *inode, char *buf,
5869				    size_t buflen, enum nfs4_acl_type type)
5870{
5871	struct nfs_inode *nfsi = NFS_I(inode);
5872	struct nfs4_cached_acl *acl;
5873	int ret = -ENOENT;
5874
5875	spin_lock(&inode->i_lock);
5876	acl = nfsi->nfs4_acl;
5877	if (acl == NULL)
5878		goto out;
5879	if (acl->type != type)
5880		goto out;
5881	if (buf == NULL) /* user is just asking for length */
5882		goto out_len;
5883	if (acl->cached == 0)
5884		goto out;
5885	ret = -ERANGE; /* see getxattr(2) man page */
5886	if (acl->len > buflen)
5887		goto out;
5888	memcpy(buf, acl->data, acl->len);
5889out_len:
5890	ret = acl->len;
5891out:
5892	spin_unlock(&inode->i_lock);
5893	return ret;
5894}
5895
5896static void nfs4_write_cached_acl(struct inode *inode, struct page **pages,
5897				  size_t pgbase, size_t acl_len,
5898				  enum nfs4_acl_type type)
5899{
5900	struct nfs4_cached_acl *acl;
5901	size_t buflen = sizeof(*acl) + acl_len;
5902
5903	if (buflen <= PAGE_SIZE) {
5904		acl = kmalloc(buflen, GFP_KERNEL);
5905		if (acl == NULL)
5906			goto out;
5907		acl->cached = 1;
5908		_copy_from_pages(acl->data, pages, pgbase, acl_len);
5909	} else {
5910		acl = kmalloc(sizeof(*acl), GFP_KERNEL);
5911		if (acl == NULL)
5912			goto out;
5913		acl->cached = 0;
5914	}
5915	acl->type = type;
5916	acl->len = acl_len;
5917out:
5918	nfs4_set_cached_acl(inode, acl);
5919}
5920
5921/*
5922 * The getxattr API returns the required buffer length when called with a
5923 * NULL buf. The NFSv4 acl tool then calls getxattr again after allocating
5924 * the required buf.  On a NULL buf, we send a page of data to the server
5925 * guessing that the ACL request can be serviced by a page. If so, we cache
5926 * up to the page of ACL data, and the 2nd call to getxattr is serviced by
5927 * the cache. If not so, we throw away the page, and cache the required
5928 * length. The next getxattr call will then produce another round trip to
5929 * the server, this time with the input buf of the required size.
5930 */
5931static ssize_t __nfs4_get_acl_uncached(struct inode *inode, void *buf,
5932				       size_t buflen, enum nfs4_acl_type type)
5933{
5934	struct page **pages;
5935	struct nfs_getaclargs args = {
5936		.fh = NFS_FH(inode),
5937		.acl_type = type,
5938		.acl_len = buflen,
5939	};
5940	struct nfs_getaclres res = {
5941		.acl_type = type,
5942		.acl_len = buflen,
5943	};
5944	struct rpc_message msg = {
5945		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETACL],
5946		.rpc_argp = &args,
5947		.rpc_resp = &res,
5948	};
5949	unsigned int npages;
5950	int ret = -ENOMEM, i;
5951	struct nfs_server *server = NFS_SERVER(inode);
5952
5953	if (buflen == 0)
5954		buflen = server->rsize;
5955
5956	npages = DIV_ROUND_UP(buflen, PAGE_SIZE) + 1;
5957	pages = kmalloc_array(npages, sizeof(struct page *), GFP_KERNEL);
5958	if (!pages)
5959		return -ENOMEM;
5960
5961	args.acl_pages = pages;
5962
5963	for (i = 0; i < npages; i++) {
5964		pages[i] = alloc_page(GFP_KERNEL);
5965		if (!pages[i])
5966			goto out_free;
5967	}
5968
5969	/* for decoding across pages */
5970	res.acl_scratch = alloc_page(GFP_KERNEL);
5971	if (!res.acl_scratch)
5972		goto out_free;
5973
5974	args.acl_len = npages * PAGE_SIZE;
5975
5976	dprintk("%s  buf %p buflen %zu npages %d args.acl_len %zu\n",
5977		__func__, buf, buflen, npages, args.acl_len);
5978	ret = nfs4_call_sync(NFS_SERVER(inode)->client, NFS_SERVER(inode),
5979			     &msg, &args.seq_args, &res.seq_res, 0);
5980	if (ret)
5981		goto out_free;
5982
5983	/* Handle the case where the passed-in buffer is too short */
5984	if (res.acl_flags & NFS4_ACL_TRUNC) {
5985		/* Did the user only issue a request for the acl length? */
5986		if (buf == NULL)
5987			goto out_ok;
5988		ret = -ERANGE;
5989		goto out_free;
5990	}
5991	nfs4_write_cached_acl(inode, pages, res.acl_data_offset, res.acl_len,
5992			      type);
5993	if (buf) {
5994		if (res.acl_len > buflen) {
5995			ret = -ERANGE;
5996			goto out_free;
5997		}
5998		_copy_from_pages(buf, pages, res.acl_data_offset, res.acl_len);
5999	}
6000out_ok:
6001	ret = res.acl_len;
6002out_free:
6003	while (--i >= 0)
6004		__free_page(pages[i]);
6005	if (res.acl_scratch)
6006		__free_page(res.acl_scratch);
6007	kfree(pages);
6008	return ret;
6009}
6010
6011static ssize_t nfs4_get_acl_uncached(struct inode *inode, void *buf,
6012				     size_t buflen, enum nfs4_acl_type type)
6013{
6014	struct nfs4_exception exception = {
6015		.interruptible = true,
6016	};
6017	ssize_t ret;
6018	do {
6019		ret = __nfs4_get_acl_uncached(inode, buf, buflen, type);
6020		trace_nfs4_get_acl(inode, ret);
6021		if (ret >= 0)
6022			break;
6023		ret = nfs4_handle_exception(NFS_SERVER(inode), ret, &exception);
6024	} while (exception.retry);
6025	return ret;
6026}
6027
6028static ssize_t nfs4_proc_get_acl(struct inode *inode, void *buf, size_t buflen,
6029				 enum nfs4_acl_type type)
6030{
6031	struct nfs_server *server = NFS_SERVER(inode);
6032	int ret;
6033
6034	if (!nfs4_server_supports_acls(server, type))
6035		return -EOPNOTSUPP;
6036	ret = nfs_revalidate_inode(inode, NFS_INO_INVALID_CHANGE);
6037	if (ret < 0)
6038		return ret;
6039	if (NFS_I(inode)->cache_validity & NFS_INO_INVALID_ACL)
6040		nfs_zap_acl_cache(inode);
6041	ret = nfs4_read_cached_acl(inode, buf, buflen, type);
6042	if (ret != -ENOENT)
6043		/* -ENOENT is returned if there is no ACL or if there is an ACL
6044		 * but no cached acl data, just the acl length */
6045		return ret;
6046	return nfs4_get_acl_uncached(inode, buf, buflen, type);
6047}
6048
6049static int __nfs4_proc_set_acl(struct inode *inode, const void *buf,
6050			       size_t buflen, enum nfs4_acl_type type)
6051{
6052	struct nfs_server *server = NFS_SERVER(inode);
6053	struct page *pages[NFS4ACL_MAXPAGES];
6054	struct nfs_setaclargs arg = {
6055		.fh = NFS_FH(inode),
6056		.acl_type = type,
6057		.acl_len = buflen,
6058		.acl_pages = pages,
6059	};
6060	struct nfs_setaclres res;
6061	struct rpc_message msg = {
6062		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETACL],
6063		.rpc_argp	= &arg,
6064		.rpc_resp	= &res,
6065	};
6066	unsigned int npages = DIV_ROUND_UP(buflen, PAGE_SIZE);
6067	int ret, i;
6068
6069	/* You can't remove system.nfs4_acl: */
6070	if (buflen == 0)
6071		return -EINVAL;
6072	if (!nfs4_server_supports_acls(server, type))
6073		return -EOPNOTSUPP;
6074	if (npages > ARRAY_SIZE(pages))
6075		return -ERANGE;
6076	i = nfs4_buf_to_pages_noslab(buf, buflen, arg.acl_pages);
6077	if (i < 0)
6078		return i;
6079	nfs4_inode_make_writeable(inode);
6080	ret = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
6081
6082	/*
6083	 * Free each page after tx, so the only ref left is
6084	 * held by the network stack
6085	 */
6086	for (; i > 0; i--)
6087		put_page(pages[i-1]);
6088
6089	/*
6090	 * Acl update can result in inode attribute update.
6091	 * so mark the attribute cache invalid.
6092	 */
6093	spin_lock(&inode->i_lock);
6094	nfs_set_cache_invalid(inode, NFS_INO_INVALID_CHANGE |
6095					     NFS_INO_INVALID_CTIME |
6096					     NFS_INO_REVAL_FORCED);
6097	spin_unlock(&inode->i_lock);
6098	nfs_access_zap_cache(inode);
6099	nfs_zap_acl_cache(inode);
6100	return ret;
6101}
6102
6103static int nfs4_proc_set_acl(struct inode *inode, const void *buf,
6104			     size_t buflen, enum nfs4_acl_type type)
6105{
6106	struct nfs4_exception exception = { };
6107	int err;
6108	do {
6109		err = __nfs4_proc_set_acl(inode, buf, buflen, type);
6110		trace_nfs4_set_acl(inode, err);
6111		if (err == -NFS4ERR_BADOWNER || err == -NFS4ERR_BADNAME) {
6112			/*
6113			 * no need to retry since the kernel
6114			 * isn't involved in encoding the ACEs.
6115			 */
6116			err = -EINVAL;
6117			break;
6118		}
6119		err = nfs4_handle_exception(NFS_SERVER(inode), err,
6120				&exception);
6121	} while (exception.retry);
6122	return err;
6123}
6124
6125#ifdef CONFIG_NFS_V4_SECURITY_LABEL
6126static int _nfs4_get_security_label(struct inode *inode, void *buf,
6127					size_t buflen)
6128{
6129	struct nfs_server *server = NFS_SERVER(inode);
6130	struct nfs4_label label = {0, 0, buflen, buf};
6131
6132	u32 bitmask[3] = { 0, 0, FATTR4_WORD2_SECURITY_LABEL };
6133	struct nfs_fattr fattr = {
6134		.label = &label,
6135	};
6136	struct nfs4_getattr_arg arg = {
6137		.fh		= NFS_FH(inode),
6138		.bitmask	= bitmask,
6139	};
6140	struct nfs4_getattr_res res = {
6141		.fattr		= &fattr,
6142		.server		= server,
6143	};
6144	struct rpc_message msg = {
6145		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_GETATTR],
6146		.rpc_argp	= &arg,
6147		.rpc_resp	= &res,
6148	};
6149	int ret;
6150
6151	nfs_fattr_init(&fattr);
6152
6153	ret = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 0);
6154	if (ret)
6155		return ret;
6156	if (!(fattr.valid & NFS_ATTR_FATTR_V4_SECURITY_LABEL))
6157		return -ENOENT;
6158	return label.len;
6159}
6160
6161static int nfs4_get_security_label(struct inode *inode, void *buf,
6162					size_t buflen)
6163{
6164	struct nfs4_exception exception = {
6165		.interruptible = true,
6166	};
6167	int err;
6168
6169	if (!nfs_server_capable(inode, NFS_CAP_SECURITY_LABEL))
6170		return -EOPNOTSUPP;
6171
6172	do {
6173		err = _nfs4_get_security_label(inode, buf, buflen);
6174		trace_nfs4_get_security_label(inode, err);
6175		err = nfs4_handle_exception(NFS_SERVER(inode), err,
6176				&exception);
6177	} while (exception.retry);
6178	return err;
6179}
6180
6181static int _nfs4_do_set_security_label(struct inode *inode,
6182		struct nfs4_label *ilabel,
6183		struct nfs_fattr *fattr)
6184{
6185
6186	struct iattr sattr = {0};
6187	struct nfs_server *server = NFS_SERVER(inode);
6188	const u32 bitmask[3] = { 0, 0, FATTR4_WORD2_SECURITY_LABEL };
6189	struct nfs_setattrargs arg = {
6190		.fh		= NFS_FH(inode),
6191		.iap		= &sattr,
6192		.server		= server,
6193		.bitmask	= bitmask,
6194		.label		= ilabel,
6195	};
6196	struct nfs_setattrres res = {
6197		.fattr		= fattr,
6198		.server		= server,
6199	};
6200	struct rpc_message msg = {
6201		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETATTR],
6202		.rpc_argp	= &arg,
6203		.rpc_resp	= &res,
6204	};
6205	int status;
6206
6207	nfs4_stateid_copy(&arg.stateid, &zero_stateid);
6208
6209	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
6210	if (status)
6211		dprintk("%s failed: %d\n", __func__, status);
6212
6213	return status;
6214}
6215
6216static int nfs4_do_set_security_label(struct inode *inode,
6217		struct nfs4_label *ilabel,
6218		struct nfs_fattr *fattr)
6219{
6220	struct nfs4_exception exception = { };
6221	int err;
6222
6223	do {
6224		err = _nfs4_do_set_security_label(inode, ilabel, fattr);
6225		trace_nfs4_set_security_label(inode, err);
6226		err = nfs4_handle_exception(NFS_SERVER(inode), err,
6227				&exception);
6228	} while (exception.retry);
6229	return err;
6230}
6231
6232static int
6233nfs4_set_security_label(struct inode *inode, const void *buf, size_t buflen)
6234{
6235	struct nfs4_label ilabel = {0, 0, buflen, (char *)buf };
6236	struct nfs_fattr *fattr;
6237	int status;
6238
6239	if (!nfs_server_capable(inode, NFS_CAP_SECURITY_LABEL))
6240		return -EOPNOTSUPP;
6241
6242	fattr = nfs_alloc_fattr_with_label(NFS_SERVER(inode));
6243	if (fattr == NULL)
6244		return -ENOMEM;
6245
6246	status = nfs4_do_set_security_label(inode, &ilabel, fattr);
6247	if (status == 0)
6248		nfs_setsecurity(inode, fattr);
6249
6250	return status;
6251}
6252#endif	/* CONFIG_NFS_V4_SECURITY_LABEL */
6253
6254
6255static void nfs4_init_boot_verifier(const struct nfs_client *clp,
6256				    nfs4_verifier *bootverf)
6257{
6258	__be32 verf[2];
6259
6260	if (test_bit(NFS4CLNT_PURGE_STATE, &clp->cl_state)) {
6261		/* An impossible timestamp guarantees this value
6262		 * will never match a generated boot time. */
6263		verf[0] = cpu_to_be32(U32_MAX);
6264		verf[1] = cpu_to_be32(U32_MAX);
6265	} else {
6266		struct nfs_net *nn = net_generic(clp->cl_net, nfs_net_id);
6267		u64 ns = ktime_to_ns(nn->boot_time);
6268
6269		verf[0] = cpu_to_be32(ns >> 32);
6270		verf[1] = cpu_to_be32(ns);
6271	}
6272	memcpy(bootverf->data, verf, sizeof(bootverf->data));
6273}
6274
6275static size_t
6276nfs4_get_uniquifier(struct nfs_client *clp, char *buf, size_t buflen)
6277{
6278	struct nfs_net *nn = net_generic(clp->cl_net, nfs_net_id);
6279	struct nfs_netns_client *nn_clp = nn->nfs_client;
6280	const char *id;
6281
6282	buf[0] = '\0';
6283
6284	if (nn_clp) {
6285		rcu_read_lock();
6286		id = rcu_dereference(nn_clp->identifier);
6287		if (id)
6288			strscpy(buf, id, buflen);
6289		rcu_read_unlock();
6290	}
6291
6292	if (nfs4_client_id_uniquifier[0] != '\0' && buf[0] == '\0')
6293		strscpy(buf, nfs4_client_id_uniquifier, buflen);
6294
6295	return strlen(buf);
6296}
6297
6298static int
6299nfs4_init_nonuniform_client_string(struct nfs_client *clp)
6300{
6301	char buf[NFS4_CLIENT_ID_UNIQ_LEN];
6302	size_t buflen;
6303	size_t len;
6304	char *str;
6305
6306	if (clp->cl_owner_id != NULL)
6307		return 0;
6308
6309	rcu_read_lock();
6310	len = 14 +
6311		strlen(clp->cl_rpcclient->cl_nodename) +
6312		1 +
6313		strlen(rpc_peeraddr2str(clp->cl_rpcclient, RPC_DISPLAY_ADDR)) +
6314		1;
6315	rcu_read_unlock();
6316
6317	buflen = nfs4_get_uniquifier(clp, buf, sizeof(buf));
6318	if (buflen)
6319		len += buflen + 1;
6320
6321	if (len > NFS4_OPAQUE_LIMIT + 1)
6322		return -EINVAL;
6323
6324	/*
6325	 * Since this string is allocated at mount time, and held until the
6326	 * nfs_client is destroyed, we can use GFP_KERNEL here w/o worrying
6327	 * about a memory-reclaim deadlock.
6328	 */
6329	str = kmalloc(len, GFP_KERNEL);
6330	if (!str)
6331		return -ENOMEM;
6332
6333	rcu_read_lock();
6334	if (buflen)
6335		scnprintf(str, len, "Linux NFSv4.0 %s/%s/%s",
6336			  clp->cl_rpcclient->cl_nodename, buf,
6337			  rpc_peeraddr2str(clp->cl_rpcclient,
6338					   RPC_DISPLAY_ADDR));
6339	else
6340		scnprintf(str, len, "Linux NFSv4.0 %s/%s",
6341			  clp->cl_rpcclient->cl_nodename,
6342			  rpc_peeraddr2str(clp->cl_rpcclient,
6343					   RPC_DISPLAY_ADDR));
6344	rcu_read_unlock();
6345
6346	clp->cl_owner_id = str;
6347	return 0;
6348}
6349
6350static int
6351nfs4_init_uniform_client_string(struct nfs_client *clp)
6352{
6353	char buf[NFS4_CLIENT_ID_UNIQ_LEN];
6354	size_t buflen;
6355	size_t len;
6356	char *str;
6357
6358	if (clp->cl_owner_id != NULL)
6359		return 0;
6360
6361	len = 10 + 10 + 1 + 10 + 1 +
6362		strlen(clp->cl_rpcclient->cl_nodename) + 1;
6363
6364	buflen = nfs4_get_uniquifier(clp, buf, sizeof(buf));
6365	if (buflen)
6366		len += buflen + 1;
6367
6368	if (len > NFS4_OPAQUE_LIMIT + 1)
6369		return -EINVAL;
6370
6371	/*
6372	 * Since this string is allocated at mount time, and held until the
6373	 * nfs_client is destroyed, we can use GFP_KERNEL here w/o worrying
6374	 * about a memory-reclaim deadlock.
6375	 */
6376	str = kmalloc(len, GFP_KERNEL);
6377	if (!str)
6378		return -ENOMEM;
6379
6380	if (buflen)
6381		scnprintf(str, len, "Linux NFSv%u.%u %s/%s",
6382			  clp->rpc_ops->version, clp->cl_minorversion,
6383			  buf, clp->cl_rpcclient->cl_nodename);
6384	else
6385		scnprintf(str, len, "Linux NFSv%u.%u %s",
6386			  clp->rpc_ops->version, clp->cl_minorversion,
6387			  clp->cl_rpcclient->cl_nodename);
6388	clp->cl_owner_id = str;
6389	return 0;
6390}
6391
6392/*
6393 * nfs4_callback_up_net() starts only "tcp" and "tcp6" callback
6394 * services.  Advertise one based on the address family of the
6395 * clientaddr.
6396 */
6397static unsigned int
6398nfs4_init_callback_netid(const struct nfs_client *clp, char *buf, size_t len)
6399{
6400	if (strchr(clp->cl_ipaddr, ':') != NULL)
6401		return scnprintf(buf, len, "tcp6");
6402	else
6403		return scnprintf(buf, len, "tcp");
6404}
6405
6406static void nfs4_setclientid_done(struct rpc_task *task, void *calldata)
6407{
6408	struct nfs4_setclientid *sc = calldata;
6409
6410	if (task->tk_status == 0)
6411		sc->sc_cred = get_rpccred(task->tk_rqstp->rq_cred);
6412}
6413
6414static const struct rpc_call_ops nfs4_setclientid_ops = {
6415	.rpc_call_done = nfs4_setclientid_done,
6416};
6417
6418/**
6419 * nfs4_proc_setclientid - Negotiate client ID
6420 * @clp: state data structure
6421 * @program: RPC program for NFSv4 callback service
6422 * @port: IP port number for NFS4 callback service
6423 * @cred: credential to use for this call
6424 * @res: where to place the result
6425 *
6426 * Returns zero, a negative errno, or a negative NFS4ERR status code.
6427 */
6428int nfs4_proc_setclientid(struct nfs_client *clp, u32 program,
6429		unsigned short port, const struct cred *cred,
6430		struct nfs4_setclientid_res *res)
6431{
6432	nfs4_verifier sc_verifier;
6433	struct nfs4_setclientid setclientid = {
6434		.sc_verifier = &sc_verifier,
6435		.sc_prog = program,
6436		.sc_clnt = clp,
6437	};
6438	struct rpc_message msg = {
6439		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID],
6440		.rpc_argp = &setclientid,
6441		.rpc_resp = res,
6442		.rpc_cred = cred,
6443	};
6444	struct rpc_task_setup task_setup_data = {
6445		.rpc_client = clp->cl_rpcclient,
6446		.rpc_message = &msg,
6447		.callback_ops = &nfs4_setclientid_ops,
6448		.callback_data = &setclientid,
6449		.flags = RPC_TASK_TIMEOUT | RPC_TASK_NO_ROUND_ROBIN,
6450	};
6451	unsigned long now = jiffies;
6452	int status;
6453
6454	/* nfs_client_id4 */
6455	nfs4_init_boot_verifier(clp, &sc_verifier);
6456
6457	if (test_bit(NFS_CS_MIGRATION, &clp->cl_flags))
6458		status = nfs4_init_uniform_client_string(clp);
6459	else
6460		status = nfs4_init_nonuniform_client_string(clp);
6461
6462	if (status)
6463		goto out;
6464
6465	/* cb_client4 */
6466	setclientid.sc_netid_len =
6467				nfs4_init_callback_netid(clp,
6468						setclientid.sc_netid,
6469						sizeof(setclientid.sc_netid));
6470	setclientid.sc_uaddr_len = scnprintf(setclientid.sc_uaddr,
6471				sizeof(setclientid.sc_uaddr), "%s.%u.%u",
6472				clp->cl_ipaddr, port >> 8, port & 255);
6473
6474	dprintk("NFS call  setclientid auth=%s, '%s'\n",
6475		clp->cl_rpcclient->cl_auth->au_ops->au_name,
6476		clp->cl_owner_id);
6477
6478	status = nfs4_call_sync_custom(&task_setup_data);
6479	if (setclientid.sc_cred) {
6480		kfree(clp->cl_acceptor);
6481		clp->cl_acceptor = rpcauth_stringify_acceptor(setclientid.sc_cred);
6482		put_rpccred(setclientid.sc_cred);
6483	}
6484
6485	if (status == 0)
6486		do_renew_lease(clp, now);
6487out:
6488	trace_nfs4_setclientid(clp, status);
6489	dprintk("NFS reply setclientid: %d\n", status);
6490	return status;
6491}
6492
6493/**
6494 * nfs4_proc_setclientid_confirm - Confirm client ID
6495 * @clp: state data structure
6496 * @arg: result of a previous SETCLIENTID
6497 * @cred: credential to use for this call
6498 *
6499 * Returns zero, a negative errno, or a negative NFS4ERR status code.
6500 */
6501int nfs4_proc_setclientid_confirm(struct nfs_client *clp,
6502		struct nfs4_setclientid_res *arg,
6503		const struct cred *cred)
6504{
6505	struct rpc_message msg = {
6506		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID_CONFIRM],
6507		.rpc_argp = arg,
6508		.rpc_cred = cred,
6509	};
6510	int status;
6511
6512	dprintk("NFS call  setclientid_confirm auth=%s, (client ID %llx)\n",
6513		clp->cl_rpcclient->cl_auth->au_ops->au_name,
6514		clp->cl_clientid);
6515	status = rpc_call_sync(clp->cl_rpcclient, &msg,
6516			       RPC_TASK_TIMEOUT | RPC_TASK_NO_ROUND_ROBIN);
6517	trace_nfs4_setclientid_confirm(clp, status);
6518	dprintk("NFS reply setclientid_confirm: %d\n", status);
6519	return status;
6520}
6521
6522struct nfs4_delegreturndata {
6523	struct nfs4_delegreturnargs args;
6524	struct nfs4_delegreturnres res;
6525	struct nfs_fh fh;
6526	nfs4_stateid stateid;
6527	unsigned long timestamp;
6528	struct {
6529		struct nfs4_layoutreturn_args arg;
6530		struct nfs4_layoutreturn_res res;
6531		struct nfs4_xdr_opaque_data ld_private;
6532		u32 roc_barrier;
6533		bool roc;
6534	} lr;
6535	struct nfs_fattr fattr;
6536	int rpc_status;
6537	struct inode *inode;
6538};
6539
6540static void nfs4_delegreturn_done(struct rpc_task *task, void *calldata)
6541{
6542	struct nfs4_delegreturndata *data = calldata;
6543	struct nfs4_exception exception = {
6544		.inode = data->inode,
6545		.stateid = &data->stateid,
6546		.task_is_privileged = data->args.seq_args.sa_privileged,
6547	};
6548
6549	if (!nfs4_sequence_done(task, &data->res.seq_res))
6550		return;
6551
6552	trace_nfs4_delegreturn_exit(&data->args, &data->res, task->tk_status);
6553
6554	/* Handle Layoutreturn errors */
6555	if (pnfs_roc_done(task, &data->args.lr_args, &data->res.lr_res,
6556			  &data->res.lr_ret) == -EAGAIN)
6557		goto out_restart;
6558
6559	switch (task->tk_status) {
6560	case 0:
6561		renew_lease(data->res.server, data->timestamp);
6562		break;
6563	case -NFS4ERR_ADMIN_REVOKED:
6564	case -NFS4ERR_DELEG_REVOKED:
6565	case -NFS4ERR_EXPIRED:
6566		nfs4_free_revoked_stateid(data->res.server,
6567				data->args.stateid,
6568				task->tk_msg.rpc_cred);
6569		fallthrough;
6570	case -NFS4ERR_BAD_STATEID:
6571	case -NFS4ERR_STALE_STATEID:
6572	case -ETIMEDOUT:
6573		task->tk_status = 0;
6574		break;
6575	case -NFS4ERR_OLD_STATEID:
6576		if (!nfs4_refresh_delegation_stateid(&data->stateid, data->inode))
6577			nfs4_stateid_seqid_inc(&data->stateid);
6578		if (data->args.bitmask) {
6579			data->args.bitmask = NULL;
6580			data->res.fattr = NULL;
6581		}
6582		goto out_restart;
6583	case -NFS4ERR_ACCESS:
6584		if (data->args.bitmask) {
6585			data->args.bitmask = NULL;
6586			data->res.fattr = NULL;
6587			goto out_restart;
6588		}
6589		fallthrough;
6590	default:
6591		task->tk_status = nfs4_async_handle_exception(task,
6592				data->res.server, task->tk_status,
6593				&exception);
6594		if (exception.retry)
6595			goto out_restart;
6596	}
6597	nfs_delegation_mark_returned(data->inode, data->args.stateid);
6598	data->rpc_status = task->tk_status;
6599	return;
6600out_restart:
6601	task->tk_status = 0;
6602	rpc_restart_call_prepare(task);
6603}
6604
6605static void nfs4_delegreturn_release(void *calldata)
6606{
6607	struct nfs4_delegreturndata *data = calldata;
6608	struct inode *inode = data->inode;
6609
6610	if (data->lr.roc)
6611		pnfs_roc_release(&data->lr.arg, &data->lr.res,
6612				 data->res.lr_ret);
6613	if (inode) {
6614		nfs4_fattr_set_prechange(&data->fattr,
6615					 inode_peek_iversion_raw(inode));
6616		nfs_refresh_inode(inode, &data->fattr);
6617		nfs_iput_and_deactive(inode);
6618	}
6619	kfree(calldata);
6620}
6621
6622static void nfs4_delegreturn_prepare(struct rpc_task *task, void *data)
6623{
6624	struct nfs4_delegreturndata *d_data;
6625	struct pnfs_layout_hdr *lo;
6626
6627	d_data = data;
6628
6629	if (!d_data->lr.roc && nfs4_wait_on_layoutreturn(d_data->inode, task)) {
6630		nfs4_sequence_done(task, &d_data->res.seq_res);
6631		return;
6632	}
6633
6634	lo = d_data->args.lr_args ? d_data->args.lr_args->layout : NULL;
6635	if (lo && !pnfs_layout_is_valid(lo)) {
6636		d_data->args.lr_args = NULL;
6637		d_data->res.lr_res = NULL;
6638	}
6639
6640	nfs4_setup_sequence(d_data->res.server->nfs_client,
6641			&d_data->args.seq_args,
6642			&d_data->res.seq_res,
6643			task);
6644}
6645
6646static const struct rpc_call_ops nfs4_delegreturn_ops = {
6647	.rpc_call_prepare = nfs4_delegreturn_prepare,
6648	.rpc_call_done = nfs4_delegreturn_done,
6649	.rpc_release = nfs4_delegreturn_release,
6650};
6651
6652static int _nfs4_proc_delegreturn(struct inode *inode, const struct cred *cred, const nfs4_stateid *stateid, int issync)
6653{
6654	struct nfs4_delegreturndata *data;
6655	struct nfs_server *server = NFS_SERVER(inode);
6656	struct rpc_task *task;
6657	struct rpc_message msg = {
6658		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DELEGRETURN],
6659		.rpc_cred = cred,
6660	};
6661	struct rpc_task_setup task_setup_data = {
6662		.rpc_client = server->client,
6663		.rpc_message = &msg,
6664		.callback_ops = &nfs4_delegreturn_ops,
6665		.flags = RPC_TASK_ASYNC | RPC_TASK_TIMEOUT,
6666	};
6667	int status = 0;
6668
6669	if (nfs_server_capable(inode, NFS_CAP_MOVEABLE))
6670		task_setup_data.flags |= RPC_TASK_MOVEABLE;
6671
6672	data = kzalloc(sizeof(*data), GFP_KERNEL);
6673	if (data == NULL)
6674		return -ENOMEM;
6675
6676	nfs4_state_protect(server->nfs_client,
6677			NFS_SP4_MACH_CRED_CLEANUP,
6678			&task_setup_data.rpc_client, &msg);
6679
6680	data->args.fhandle = &data->fh;
6681	data->args.stateid = &data->stateid;
6682	nfs4_bitmask_set(data->args.bitmask_store,
6683			 server->cache_consistency_bitmask, inode, 0);
6684	data->args.bitmask = data->args.bitmask_store;
6685	nfs_copy_fh(&data->fh, NFS_FH(inode));
6686	nfs4_stateid_copy(&data->stateid, stateid);
6687	data->res.fattr = &data->fattr;
6688	data->res.server = server;
6689	data->res.lr_ret = -NFS4ERR_NOMATCHING_LAYOUT;
6690	data->lr.arg.ld_private = &data->lr.ld_private;
6691	nfs_fattr_init(data->res.fattr);
6692	data->timestamp = jiffies;
6693	data->rpc_status = 0;
6694	data->inode = nfs_igrab_and_active(inode);
6695	if (data->inode || issync) {
6696		data->lr.roc = pnfs_roc(inode, &data->lr.arg, &data->lr.res,
6697					cred);
6698		if (data->lr.roc) {
6699			data->args.lr_args = &data->lr.arg;
6700			data->res.lr_res = &data->lr.res;
6701		}
6702	}
6703
6704	if (!data->inode)
6705		nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1,
6706				   1);
6707	else
6708		nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1,
6709				   0);
6710	task_setup_data.callback_data = data;
6711	msg.rpc_argp = &data->args;
6712	msg.rpc_resp = &data->res;
6713	task = rpc_run_task(&task_setup_data);
6714	if (IS_ERR(task))
6715		return PTR_ERR(task);
6716	if (!issync)
6717		goto out;
6718	status = rpc_wait_for_completion_task(task);
6719	if (status != 0)
6720		goto out;
6721	status = data->rpc_status;
6722out:
6723	rpc_put_task(task);
6724	return status;
6725}
6726
6727int nfs4_proc_delegreturn(struct inode *inode, const struct cred *cred, const nfs4_stateid *stateid, int issync)
6728{
6729	struct nfs_server *server = NFS_SERVER(inode);
6730	struct nfs4_exception exception = { };
6731	int err;
6732	do {
6733		err = _nfs4_proc_delegreturn(inode, cred, stateid, issync);
6734		trace_nfs4_delegreturn(inode, stateid, err);
6735		switch (err) {
6736			case -NFS4ERR_STALE_STATEID:
6737			case -NFS4ERR_EXPIRED:
6738			case 0:
6739				return 0;
6740		}
6741		err = nfs4_handle_exception(server, err, &exception);
6742	} while (exception.retry);
6743	return err;
6744}
6745
6746static int _nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *request)
6747{
6748	struct inode *inode = state->inode;
6749	struct nfs_server *server = NFS_SERVER(inode);
6750	struct nfs_client *clp = server->nfs_client;
6751	struct nfs_lockt_args arg = {
6752		.fh = NFS_FH(inode),
6753		.fl = request,
6754	};
6755	struct nfs_lockt_res res = {
6756		.denied = request,
6757	};
6758	struct rpc_message msg = {
6759		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_LOCKT],
6760		.rpc_argp	= &arg,
6761		.rpc_resp	= &res,
6762		.rpc_cred	= state->owner->so_cred,
6763	};
6764	struct nfs4_lock_state *lsp;
6765	int status;
6766
6767	arg.lock_owner.clientid = clp->cl_clientid;
6768	status = nfs4_set_lock_state(state, request);
6769	if (status != 0)
6770		goto out;
6771	lsp = request->fl_u.nfs4_fl.owner;
6772	arg.lock_owner.id = lsp->ls_seqid.owner_id;
6773	arg.lock_owner.s_dev = server->s_dev;
6774	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
6775	switch (status) {
6776		case 0:
6777			request->fl_type = F_UNLCK;
6778			break;
6779		case -NFS4ERR_DENIED:
6780			status = 0;
6781	}
6782	request->fl_ops->fl_release_private(request);
6783	request->fl_ops = NULL;
6784out:
6785	return status;
6786}
6787
6788static int nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *request)
6789{
6790	struct nfs4_exception exception = {
6791		.interruptible = true,
6792	};
6793	int err;
6794
6795	do {
6796		err = _nfs4_proc_getlk(state, cmd, request);
6797		trace_nfs4_get_lock(request, state, cmd, err);
6798		err = nfs4_handle_exception(NFS_SERVER(state->inode), err,
6799				&exception);
6800	} while (exception.retry);
6801	return err;
6802}
6803
6804/*
6805 * Update the seqid of a lock stateid after receiving
6806 * NFS4ERR_OLD_STATEID
6807 */
6808static bool nfs4_refresh_lock_old_stateid(nfs4_stateid *dst,
6809		struct nfs4_lock_state *lsp)
6810{
6811	struct nfs4_state *state = lsp->ls_state;
6812	bool ret = false;
6813
6814	spin_lock(&state->state_lock);
6815	if (!nfs4_stateid_match_other(dst, &lsp->ls_stateid))
6816		goto out;
6817	if (!nfs4_stateid_is_newer(&lsp->ls_stateid, dst))
6818		nfs4_stateid_seqid_inc(dst);
6819	else
6820		dst->seqid = lsp->ls_stateid.seqid;
6821	ret = true;
6822out:
6823	spin_unlock(&state->state_lock);
6824	return ret;
6825}
6826
6827static bool nfs4_sync_lock_stateid(nfs4_stateid *dst,
6828		struct nfs4_lock_state *lsp)
6829{
6830	struct nfs4_state *state = lsp->ls_state;
6831	bool ret;
6832
6833	spin_lock(&state->state_lock);
6834	ret = !nfs4_stateid_match_other(dst, &lsp->ls_stateid);
6835	nfs4_stateid_copy(dst, &lsp->ls_stateid);
6836	spin_unlock(&state->state_lock);
6837	return ret;
6838}
6839
6840struct nfs4_unlockdata {
6841	struct nfs_locku_args arg;
6842	struct nfs_locku_res res;
6843	struct nfs4_lock_state *lsp;
6844	struct nfs_open_context *ctx;
6845	struct nfs_lock_context *l_ctx;
6846	struct file_lock fl;
6847	struct nfs_server *server;
6848	unsigned long timestamp;
6849};
6850
6851static struct nfs4_unlockdata *nfs4_alloc_unlockdata(struct file_lock *fl,
6852		struct nfs_open_context *ctx,
6853		struct nfs4_lock_state *lsp,
6854		struct nfs_seqid *seqid)
6855{
6856	struct nfs4_unlockdata *p;
6857	struct nfs4_state *state = lsp->ls_state;
6858	struct inode *inode = state->inode;
6859
6860	p = kzalloc(sizeof(*p), GFP_KERNEL);
6861	if (p == NULL)
6862		return NULL;
6863	p->arg.fh = NFS_FH(inode);
6864	p->arg.fl = &p->fl;
6865	p->arg.seqid = seqid;
6866	p->res.seqid = seqid;
6867	p->lsp = lsp;
6868	/* Ensure we don't close file until we're done freeing locks! */
6869	p->ctx = get_nfs_open_context(ctx);
6870	p->l_ctx = nfs_get_lock_context(ctx);
6871	locks_init_lock(&p->fl);
6872	locks_copy_lock(&p->fl, fl);
6873	p->server = NFS_SERVER(inode);
6874	spin_lock(&state->state_lock);
6875	nfs4_stateid_copy(&p->arg.stateid, &lsp->ls_stateid);
6876	spin_unlock(&state->state_lock);
6877	return p;
6878}
6879
6880static void nfs4_locku_release_calldata(void *data)
6881{
6882	struct nfs4_unlockdata *calldata = data;
6883	nfs_free_seqid(calldata->arg.seqid);
6884	nfs4_put_lock_state(calldata->lsp);
6885	nfs_put_lock_context(calldata->l_ctx);
6886	put_nfs_open_context(calldata->ctx);
6887	kfree(calldata);
6888}
6889
6890static void nfs4_locku_done(struct rpc_task *task, void *data)
6891{
6892	struct nfs4_unlockdata *calldata = data;
6893	struct nfs4_exception exception = {
6894		.inode = calldata->lsp->ls_state->inode,
6895		.stateid = &calldata->arg.stateid,
6896	};
6897
6898	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
6899		return;
6900	switch (task->tk_status) {
6901		case 0:
6902			renew_lease(calldata->server, calldata->timestamp);
6903			locks_lock_inode_wait(calldata->lsp->ls_state->inode, &calldata->fl);
6904			if (nfs4_update_lock_stateid(calldata->lsp,
6905					&calldata->res.stateid))
6906				break;
6907			fallthrough;
6908		case -NFS4ERR_ADMIN_REVOKED:
6909		case -NFS4ERR_EXPIRED:
6910			nfs4_free_revoked_stateid(calldata->server,
6911					&calldata->arg.stateid,
6912					task->tk_msg.rpc_cred);
6913			fallthrough;
6914		case -NFS4ERR_BAD_STATEID:
6915		case -NFS4ERR_STALE_STATEID:
6916			if (nfs4_sync_lock_stateid(&calldata->arg.stateid,
6917						calldata->lsp))
6918				rpc_restart_call_prepare(task);
6919			break;
6920		case -NFS4ERR_OLD_STATEID:
6921			if (nfs4_refresh_lock_old_stateid(&calldata->arg.stateid,
6922						calldata->lsp))
6923				rpc_restart_call_prepare(task);
6924			break;
6925		default:
6926			task->tk_status = nfs4_async_handle_exception(task,
6927					calldata->server, task->tk_status,
6928					&exception);
6929			if (exception.retry)
6930				rpc_restart_call_prepare(task);
6931	}
6932	nfs_release_seqid(calldata->arg.seqid);
6933}
6934
6935static void nfs4_locku_prepare(struct rpc_task *task, void *data)
6936{
6937	struct nfs4_unlockdata *calldata = data;
6938
6939	if (test_bit(NFS_CONTEXT_UNLOCK, &calldata->l_ctx->open_context->flags) &&
6940		nfs_async_iocounter_wait(task, calldata->l_ctx))
6941		return;
6942
6943	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
6944		goto out_wait;
6945	if (test_bit(NFS_LOCK_INITIALIZED, &calldata->lsp->ls_flags) == 0) {
6946		/* Note: exit _without_ running nfs4_locku_done */
6947		goto out_no_action;
6948	}
6949	calldata->timestamp = jiffies;
6950	if (nfs4_setup_sequence(calldata->server->nfs_client,
6951				&calldata->arg.seq_args,
6952				&calldata->res.seq_res,
6953				task) != 0)
6954		nfs_release_seqid(calldata->arg.seqid);
6955	return;
6956out_no_action:
6957	task->tk_action = NULL;
6958out_wait:
6959	nfs4_sequence_done(task, &calldata->res.seq_res);
6960}
6961
6962static const struct rpc_call_ops nfs4_locku_ops = {
6963	.rpc_call_prepare = nfs4_locku_prepare,
6964	.rpc_call_done = nfs4_locku_done,
6965	.rpc_release = nfs4_locku_release_calldata,
6966};
6967
6968static struct rpc_task *nfs4_do_unlck(struct file_lock *fl,
6969		struct nfs_open_context *ctx,
6970		struct nfs4_lock_state *lsp,
6971		struct nfs_seqid *seqid)
6972{
6973	struct nfs4_unlockdata *data;
6974	struct rpc_message msg = {
6975		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCKU],
6976		.rpc_cred = ctx->cred,
6977	};
6978	struct rpc_task_setup task_setup_data = {
6979		.rpc_client = NFS_CLIENT(lsp->ls_state->inode),
6980		.rpc_message = &msg,
6981		.callback_ops = &nfs4_locku_ops,
6982		.workqueue = nfsiod_workqueue,
6983		.flags = RPC_TASK_ASYNC,
6984	};
6985
6986	if (nfs_server_capable(lsp->ls_state->inode, NFS_CAP_MOVEABLE))
6987		task_setup_data.flags |= RPC_TASK_MOVEABLE;
6988
6989	nfs4_state_protect(NFS_SERVER(lsp->ls_state->inode)->nfs_client,
6990		NFS_SP4_MACH_CRED_CLEANUP, &task_setup_data.rpc_client, &msg);
6991
6992	/* Ensure this is an unlock - when canceling a lock, the
6993	 * canceled lock is passed in, and it won't be an unlock.
6994	 */
6995	fl->fl_type = F_UNLCK;
6996	if (fl->fl_flags & FL_CLOSE)
6997		set_bit(NFS_CONTEXT_UNLOCK, &ctx->flags);
6998
6999	data = nfs4_alloc_unlockdata(fl, ctx, lsp, seqid);
7000	if (data == NULL) {
7001		nfs_free_seqid(seqid);
7002		return ERR_PTR(-ENOMEM);
7003	}
7004
7005	nfs4_init_sequence(&data->arg.seq_args, &data->res.seq_res, 1, 0);
7006	msg.rpc_argp = &data->arg;
7007	msg.rpc_resp = &data->res;
7008	task_setup_data.callback_data = data;
7009	return rpc_run_task(&task_setup_data);
7010}
7011
7012static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
7013{
7014	struct inode *inode = state->inode;
7015	struct nfs4_state_owner *sp = state->owner;
7016	struct nfs_inode *nfsi = NFS_I(inode);
7017	struct nfs_seqid *seqid;
7018	struct nfs4_lock_state *lsp;
7019	struct rpc_task *task;
7020	struct nfs_seqid *(*alloc_seqid)(struct nfs_seqid_counter *, gfp_t);
7021	int status = 0;
7022	unsigned char fl_flags = request->fl_flags;
7023
7024	status = nfs4_set_lock_state(state, request);
7025	/* Unlock _before_ we do the RPC call */
7026	request->fl_flags |= FL_EXISTS;
7027	/* Exclude nfs_delegation_claim_locks() */
7028	mutex_lock(&sp->so_delegreturn_mutex);
7029	/* Exclude nfs4_reclaim_open_stateid() - note nesting! */
7030	down_read(&nfsi->rwsem);
7031	if (locks_lock_inode_wait(inode, request) == -ENOENT) {
7032		up_read(&nfsi->rwsem);
7033		mutex_unlock(&sp->so_delegreturn_mutex);
7034		goto out;
7035	}
7036	lsp = request->fl_u.nfs4_fl.owner;
7037	set_bit(NFS_LOCK_UNLOCKING, &lsp->ls_flags);
7038	up_read(&nfsi->rwsem);
7039	mutex_unlock(&sp->so_delegreturn_mutex);
7040	if (status != 0)
7041		goto out;
7042	/* Is this a delegated lock? */
7043	if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) == 0)
7044		goto out;
7045	alloc_seqid = NFS_SERVER(inode)->nfs_client->cl_mvops->alloc_seqid;
7046	seqid = alloc_seqid(&lsp->ls_seqid, GFP_KERNEL);
7047	status = -ENOMEM;
7048	if (IS_ERR(seqid))
7049		goto out;
7050	task = nfs4_do_unlck(request, nfs_file_open_context(request->fl_file), lsp, seqid);
7051	status = PTR_ERR(task);
7052	if (IS_ERR(task))
7053		goto out;
7054	status = rpc_wait_for_completion_task(task);
7055	rpc_put_task(task);
7056out:
7057	request->fl_flags = fl_flags;
7058	trace_nfs4_unlock(request, state, F_SETLK, status);
7059	return status;
7060}
7061
7062struct nfs4_lockdata {
7063	struct nfs_lock_args arg;
7064	struct nfs_lock_res res;
7065	struct nfs4_lock_state *lsp;
7066	struct nfs_open_context *ctx;
7067	struct file_lock fl;
7068	unsigned long timestamp;
7069	int rpc_status;
7070	int cancelled;
7071	struct nfs_server *server;
7072};
7073
7074static struct nfs4_lockdata *nfs4_alloc_lockdata(struct file_lock *fl,
7075		struct nfs_open_context *ctx, struct nfs4_lock_state *lsp,
7076		gfp_t gfp_mask)
7077{
7078	struct nfs4_lockdata *p;
7079	struct inode *inode = lsp->ls_state->inode;
7080	struct nfs_server *server = NFS_SERVER(inode);
7081	struct nfs_seqid *(*alloc_seqid)(struct nfs_seqid_counter *, gfp_t);
7082
7083	p = kzalloc(sizeof(*p), gfp_mask);
7084	if (p == NULL)
7085		return NULL;
7086
7087	p->arg.fh = NFS_FH(inode);
7088	p->arg.fl = &p->fl;
7089	p->arg.open_seqid = nfs_alloc_seqid(&lsp->ls_state->owner->so_seqid, gfp_mask);
7090	if (IS_ERR(p->arg.open_seqid))
7091		goto out_free;
7092	alloc_seqid = server->nfs_client->cl_mvops->alloc_seqid;
7093	p->arg.lock_seqid = alloc_seqid(&lsp->ls_seqid, gfp_mask);
7094	if (IS_ERR(p->arg.lock_seqid))
7095		goto out_free_seqid;
7096	p->arg.lock_owner.clientid = server->nfs_client->cl_clientid;
7097	p->arg.lock_owner.id = lsp->ls_seqid.owner_id;
7098	p->arg.lock_owner.s_dev = server->s_dev;
7099	p->res.lock_seqid = p->arg.lock_seqid;
7100	p->lsp = lsp;
7101	p->server = server;
7102	p->ctx = get_nfs_open_context(ctx);
7103	locks_init_lock(&p->fl);
7104	locks_copy_lock(&p->fl, fl);
7105	return p;
7106out_free_seqid:
7107	nfs_free_seqid(p->arg.open_seqid);
7108out_free:
7109	kfree(p);
7110	return NULL;
7111}
7112
7113static void nfs4_lock_prepare(struct rpc_task *task, void *calldata)
7114{
7115	struct nfs4_lockdata *data = calldata;
7116	struct nfs4_state *state = data->lsp->ls_state;
7117
7118	if (nfs_wait_on_sequence(data->arg.lock_seqid, task) != 0)
7119		goto out_wait;
7120	/* Do we need to do an open_to_lock_owner? */
7121	if (!test_bit(NFS_LOCK_INITIALIZED, &data->lsp->ls_flags)) {
7122		if (nfs_wait_on_sequence(data->arg.open_seqid, task) != 0) {
7123			goto out_release_lock_seqid;
7124		}
7125		nfs4_stateid_copy(&data->arg.open_stateid,
7126				&state->open_stateid);
7127		data->arg.new_lock_owner = 1;
7128		data->res.open_seqid = data->arg.open_seqid;
7129	} else {
7130		data->arg.new_lock_owner = 0;
7131		nfs4_stateid_copy(&data->arg.lock_stateid,
7132				&data->lsp->ls_stateid);
7133	}
7134	if (!nfs4_valid_open_stateid(state)) {
7135		data->rpc_status = -EBADF;
7136		task->tk_action = NULL;
7137		goto out_release_open_seqid;
7138	}
7139	data->timestamp = jiffies;
7140	if (nfs4_setup_sequence(data->server->nfs_client,
7141				&data->arg.seq_args,
7142				&data->res.seq_res,
7143				task) == 0)
7144		return;
7145out_release_open_seqid:
7146	nfs_release_seqid(data->arg.open_seqid);
7147out_release_lock_seqid:
7148	nfs_release_seqid(data->arg.lock_seqid);
7149out_wait:
7150	nfs4_sequence_done(task, &data->res.seq_res);
7151	dprintk("%s: ret = %d\n", __func__, data->rpc_status);
7152}
7153
7154static void nfs4_lock_done(struct rpc_task *task, void *calldata)
7155{
7156	struct nfs4_lockdata *data = calldata;
7157	struct nfs4_lock_state *lsp = data->lsp;
7158
7159	if (!nfs4_sequence_done(task, &data->res.seq_res))
7160		return;
7161
7162	data->rpc_status = task->tk_status;
7163	switch (task->tk_status) {
7164	case 0:
7165		renew_lease(NFS_SERVER(d_inode(data->ctx->dentry)),
7166				data->timestamp);
7167		if (data->arg.new_lock && !data->cancelled) {
7168			data->fl.fl_flags &= ~(FL_SLEEP | FL_ACCESS);
7169			if (locks_lock_inode_wait(lsp->ls_state->inode, &data->fl) < 0)
7170				goto out_restart;
7171		}
7172		if (data->arg.new_lock_owner != 0) {
7173			nfs_confirm_seqid(&lsp->ls_seqid, 0);
7174			nfs4_stateid_copy(&lsp->ls_stateid, &data->res.stateid);
7175			set_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags);
7176		} else if (!nfs4_update_lock_stateid(lsp, &data->res.stateid))
7177			goto out_restart;
7178		break;
7179	case -NFS4ERR_OLD_STATEID:
7180		if (data->arg.new_lock_owner != 0 &&
7181			nfs4_refresh_open_old_stateid(&data->arg.open_stateid,
7182					lsp->ls_state))
7183			goto out_restart;
7184		if (nfs4_refresh_lock_old_stateid(&data->arg.lock_stateid, lsp))
7185			goto out_restart;
7186		fallthrough;
7187	case -NFS4ERR_BAD_STATEID:
7188	case -NFS4ERR_STALE_STATEID:
7189	case -NFS4ERR_EXPIRED:
7190		if (data->arg.new_lock_owner != 0) {
7191			if (!nfs4_stateid_match(&data->arg.open_stateid,
7192						&lsp->ls_state->open_stateid))
7193				goto out_restart;
7194		} else if (!nfs4_stateid_match(&data->arg.lock_stateid,
7195						&lsp->ls_stateid))
7196				goto out_restart;
7197	}
7198out_done:
7199	dprintk("%s: ret = %d!\n", __func__, data->rpc_status);
7200	return;
7201out_restart:
7202	if (!data->cancelled)
7203		rpc_restart_call_prepare(task);
7204	goto out_done;
7205}
7206
7207static void nfs4_lock_release(void *calldata)
7208{
7209	struct nfs4_lockdata *data = calldata;
7210
7211	nfs_free_seqid(data->arg.open_seqid);
7212	if (data->cancelled && data->rpc_status == 0) {
7213		struct rpc_task *task;
7214		task = nfs4_do_unlck(&data->fl, data->ctx, data->lsp,
7215				data->arg.lock_seqid);
7216		if (!IS_ERR(task))
7217			rpc_put_task_async(task);
7218		dprintk("%s: cancelling lock!\n", __func__);
7219	} else
7220		nfs_free_seqid(data->arg.lock_seqid);
7221	nfs4_put_lock_state(data->lsp);
7222	put_nfs_open_context(data->ctx);
7223	kfree(data);
7224}
7225
7226static const struct rpc_call_ops nfs4_lock_ops = {
7227	.rpc_call_prepare = nfs4_lock_prepare,
7228	.rpc_call_done = nfs4_lock_done,
7229	.rpc_release = nfs4_lock_release,
7230};
7231
7232static void nfs4_handle_setlk_error(struct nfs_server *server, struct nfs4_lock_state *lsp, int new_lock_owner, int error)
7233{
7234	switch (error) {
7235	case -NFS4ERR_ADMIN_REVOKED:
7236	case -NFS4ERR_EXPIRED:
7237	case -NFS4ERR_BAD_STATEID:
7238		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
7239		if (new_lock_owner != 0 ||
7240		   test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) != 0)
7241			nfs4_schedule_stateid_recovery(server, lsp->ls_state);
7242		break;
7243	case -NFS4ERR_STALE_STATEID:
7244		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
7245		nfs4_schedule_lease_recovery(server->nfs_client);
7246	}
7247}
7248
7249static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *fl, int recovery_type)
7250{
7251	struct nfs4_lockdata *data;
7252	struct rpc_task *task;
7253	struct rpc_message msg = {
7254		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCK],
7255		.rpc_cred = state->owner->so_cred,
7256	};
7257	struct rpc_task_setup task_setup_data = {
7258		.rpc_client = NFS_CLIENT(state->inode),
7259		.rpc_message = &msg,
7260		.callback_ops = &nfs4_lock_ops,
7261		.workqueue = nfsiod_workqueue,
7262		.flags = RPC_TASK_ASYNC | RPC_TASK_CRED_NOREF,
7263	};
7264	int ret;
7265
7266	if (nfs_server_capable(state->inode, NFS_CAP_MOVEABLE))
7267		task_setup_data.flags |= RPC_TASK_MOVEABLE;
7268
7269	data = nfs4_alloc_lockdata(fl, nfs_file_open_context(fl->fl_file),
7270				   fl->fl_u.nfs4_fl.owner, GFP_KERNEL);
7271	if (data == NULL)
7272		return -ENOMEM;
7273	if (IS_SETLKW(cmd))
7274		data->arg.block = 1;
7275	nfs4_init_sequence(&data->arg.seq_args, &data->res.seq_res, 1,
7276				recovery_type > NFS_LOCK_NEW);
7277	msg.rpc_argp = &data->arg;
7278	msg.rpc_resp = &data->res;
7279	task_setup_data.callback_data = data;
7280	if (recovery_type > NFS_LOCK_NEW) {
7281		if (recovery_type == NFS_LOCK_RECLAIM)
7282			data->arg.reclaim = NFS_LOCK_RECLAIM;
7283	} else
7284		data->arg.new_lock = 1;
7285	task = rpc_run_task(&task_setup_data);
7286	if (IS_ERR(task))
7287		return PTR_ERR(task);
7288	ret = rpc_wait_for_completion_task(task);
7289	if (ret == 0) {
7290		ret = data->rpc_status;
7291		if (ret)
7292			nfs4_handle_setlk_error(data->server, data->lsp,
7293					data->arg.new_lock_owner, ret);
7294	} else
7295		data->cancelled = true;
7296	trace_nfs4_set_lock(fl, state, &data->res.stateid, cmd, ret);
7297	rpc_put_task(task);
7298	dprintk("%s: ret = %d\n", __func__, ret);
7299	return ret;
7300}
7301
7302static int nfs4_lock_reclaim(struct nfs4_state *state, struct file_lock *request)
7303{
7304	struct nfs_server *server = NFS_SERVER(state->inode);
7305	struct nfs4_exception exception = {
7306		.inode = state->inode,
7307	};
7308	int err;
7309
7310	do {
7311		/* Cache the lock if possible... */
7312		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
7313			return 0;
7314		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_RECLAIM);
7315		if (err != -NFS4ERR_DELAY)
7316			break;
7317		nfs4_handle_exception(server, err, &exception);
7318	} while (exception.retry);
7319	return err;
7320}
7321
7322static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request)
7323{
7324	struct nfs_server *server = NFS_SERVER(state->inode);
7325	struct nfs4_exception exception = {
7326		.inode = state->inode,
7327	};
7328	int err;
7329
7330	err = nfs4_set_lock_state(state, request);
7331	if (err != 0)
7332		return err;
7333	if (!recover_lost_locks) {
7334		set_bit(NFS_LOCK_LOST, &request->fl_u.nfs4_fl.owner->ls_flags);
7335		return 0;
7336	}
7337	do {
7338		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
7339			return 0;
7340		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_EXPIRED);
7341		switch (err) {
7342		default:
7343			goto out;
7344		case -NFS4ERR_GRACE:
7345		case -NFS4ERR_DELAY:
7346			nfs4_handle_exception(server, err, &exception);
7347			err = 0;
7348		}
7349	} while (exception.retry);
7350out:
7351	return err;
7352}
7353
7354#if defined(CONFIG_NFS_V4_1)
7355static int nfs41_lock_expired(struct nfs4_state *state, struct file_lock *request)
7356{
7357	struct nfs4_lock_state *lsp;
7358	int status;
7359
7360	status = nfs4_set_lock_state(state, request);
7361	if (status != 0)
7362		return status;
7363	lsp = request->fl_u.nfs4_fl.owner;
7364	if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) ||
7365	    test_bit(NFS_LOCK_LOST, &lsp->ls_flags))
7366		return 0;
7367	return nfs4_lock_expired(state, request);
7368}
7369#endif
7370
7371static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
7372{
7373	struct nfs_inode *nfsi = NFS_I(state->inode);
7374	struct nfs4_state_owner *sp = state->owner;
7375	unsigned char fl_flags = request->fl_flags;
7376	int status;
7377
7378	request->fl_flags |= FL_ACCESS;
7379	status = locks_lock_inode_wait(state->inode, request);
7380	if (status < 0)
7381		goto out;
7382	mutex_lock(&sp->so_delegreturn_mutex);
7383	down_read(&nfsi->rwsem);
7384	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
7385		/* Yes: cache locks! */
7386		/* ...but avoid races with delegation recall... */
7387		request->fl_flags = fl_flags & ~FL_SLEEP;
7388		status = locks_lock_inode_wait(state->inode, request);
7389		up_read(&nfsi->rwsem);
7390		mutex_unlock(&sp->so_delegreturn_mutex);
7391		goto out;
7392	}
7393	up_read(&nfsi->rwsem);
7394	mutex_unlock(&sp->so_delegreturn_mutex);
7395	status = _nfs4_do_setlk(state, cmd, request, NFS_LOCK_NEW);
7396out:
7397	request->fl_flags = fl_flags;
7398	return status;
7399}
7400
7401static int nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
7402{
7403	struct nfs4_exception exception = {
7404		.state = state,
7405		.inode = state->inode,
7406		.interruptible = true,
7407	};
7408	int err;
7409
7410	do {
7411		err = _nfs4_proc_setlk(state, cmd, request);
7412		if (err == -NFS4ERR_DENIED)
7413			err = -EAGAIN;
7414		err = nfs4_handle_exception(NFS_SERVER(state->inode),
7415				err, &exception);
7416	} while (exception.retry);
7417	return err;
7418}
7419
7420#define NFS4_LOCK_MINTIMEOUT (1 * HZ)
7421#define NFS4_LOCK_MAXTIMEOUT (30 * HZ)
7422
7423static int
7424nfs4_retry_setlk_simple(struct nfs4_state *state, int cmd,
7425			struct file_lock *request)
7426{
7427	int		status = -ERESTARTSYS;
7428	unsigned long	timeout = NFS4_LOCK_MINTIMEOUT;
7429
7430	while(!signalled()) {
7431		status = nfs4_proc_setlk(state, cmd, request);
7432		if ((status != -EAGAIN) || IS_SETLK(cmd))
7433			break;
7434		__set_current_state(TASK_INTERRUPTIBLE|TASK_FREEZABLE);
7435		schedule_timeout(timeout);
7436		timeout *= 2;
7437		timeout = min_t(unsigned long, NFS4_LOCK_MAXTIMEOUT, timeout);
7438		status = -ERESTARTSYS;
7439	}
7440	return status;
7441}
7442
7443#ifdef CONFIG_NFS_V4_1
7444struct nfs4_lock_waiter {
7445	struct inode		*inode;
7446	struct nfs_lowner	owner;
7447	wait_queue_entry_t	wait;
7448};
7449
7450static int
7451nfs4_wake_lock_waiter(wait_queue_entry_t *wait, unsigned int mode, int flags, void *key)
7452{
7453	struct nfs4_lock_waiter	*waiter	=
7454		container_of(wait, struct nfs4_lock_waiter, wait);
7455
7456	/* NULL key means to wake up everyone */
7457	if (key) {
7458		struct cb_notify_lock_args	*cbnl = key;
7459		struct nfs_lowner		*lowner = &cbnl->cbnl_owner,
7460						*wowner = &waiter->owner;
7461
7462		/* Only wake if the callback was for the same owner. */
7463		if (lowner->id != wowner->id || lowner->s_dev != wowner->s_dev)
7464			return 0;
7465
7466		/* Make sure it's for the right inode */
7467		if (nfs_compare_fh(NFS_FH(waiter->inode), &cbnl->cbnl_fh))
7468			return 0;
7469	}
7470
7471	return woken_wake_function(wait, mode, flags, key);
7472}
7473
7474static int
7475nfs4_retry_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
7476{
7477	struct nfs4_lock_state *lsp = request->fl_u.nfs4_fl.owner;
7478	struct nfs_server *server = NFS_SERVER(state->inode);
7479	struct nfs_client *clp = server->nfs_client;
7480	wait_queue_head_t *q = &clp->cl_lock_waitq;
7481	struct nfs4_lock_waiter waiter = {
7482		.inode = state->inode,
7483		.owner = { .clientid = clp->cl_clientid,
7484			   .id = lsp->ls_seqid.owner_id,
7485			   .s_dev = server->s_dev },
7486	};
7487	int status;
7488
7489	/* Don't bother with waitqueue if we don't expect a callback */
7490	if (!test_bit(NFS_STATE_MAY_NOTIFY_LOCK, &state->flags))
7491		return nfs4_retry_setlk_simple(state, cmd, request);
7492
7493	init_wait(&waiter.wait);
7494	waiter.wait.func = nfs4_wake_lock_waiter;
7495	add_wait_queue(q, &waiter.wait);
7496
7497	do {
7498		status = nfs4_proc_setlk(state, cmd, request);
7499		if (status != -EAGAIN || IS_SETLK(cmd))
7500			break;
7501
7502		status = -ERESTARTSYS;
7503		wait_woken(&waiter.wait, TASK_INTERRUPTIBLE|TASK_FREEZABLE,
7504			   NFS4_LOCK_MAXTIMEOUT);
7505	} while (!signalled());
7506
7507	remove_wait_queue(q, &waiter.wait);
7508
7509	return status;
7510}
7511#else /* !CONFIG_NFS_V4_1 */
7512static inline int
7513nfs4_retry_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
7514{
7515	return nfs4_retry_setlk_simple(state, cmd, request);
7516}
7517#endif
7518
7519static int
7520nfs4_proc_lock(struct file *filp, int cmd, struct file_lock *request)
7521{
7522	struct nfs_open_context *ctx;
7523	struct nfs4_state *state;
7524	int status;
7525
7526	/* verify open state */
7527	ctx = nfs_file_open_context(filp);
7528	state = ctx->state;
7529
7530	if (IS_GETLK(cmd)) {
7531		if (state != NULL)
7532			return nfs4_proc_getlk(state, F_GETLK, request);
7533		return 0;
7534	}
7535
7536	if (!(IS_SETLK(cmd) || IS_SETLKW(cmd)))
7537		return -EINVAL;
7538
7539	if (request->fl_type == F_UNLCK) {
7540		if (state != NULL)
7541			return nfs4_proc_unlck(state, cmd, request);
7542		return 0;
7543	}
7544
7545	if (state == NULL)
7546		return -ENOLCK;
7547
7548	if ((request->fl_flags & FL_POSIX) &&
7549	    !test_bit(NFS_STATE_POSIX_LOCKS, &state->flags))
7550		return -ENOLCK;
7551
7552	/*
7553	 * Don't rely on the VFS having checked the file open mode,
7554	 * since it won't do this for flock() locks.
7555	 */
7556	switch (request->fl_type) {
7557	case F_RDLCK:
7558		if (!(filp->f_mode & FMODE_READ))
7559			return -EBADF;
7560		break;
7561	case F_WRLCK:
7562		if (!(filp->f_mode & FMODE_WRITE))
7563			return -EBADF;
7564	}
7565
7566	status = nfs4_set_lock_state(state, request);
7567	if (status != 0)
7568		return status;
7569
7570	return nfs4_retry_setlk(state, cmd, request);
7571}
7572
7573static int nfs4_delete_lease(struct file *file, void **priv)
7574{
7575	return generic_setlease(file, F_UNLCK, NULL, priv);
7576}
7577
7578static int nfs4_add_lease(struct file *file, int arg, struct file_lock **lease,
7579			  void **priv)
7580{
7581	struct inode *inode = file_inode(file);
7582	fmode_t type = arg == F_RDLCK ? FMODE_READ : FMODE_WRITE;
7583	int ret;
7584
7585	/* No delegation, no lease */
7586	if (!nfs4_have_delegation(inode, type))
7587		return -EAGAIN;
7588	ret = generic_setlease(file, arg, lease, priv);
7589	if (ret || nfs4_have_delegation(inode, type))
7590		return ret;
7591	/* We raced with a delegation return */
7592	nfs4_delete_lease(file, priv);
7593	return -EAGAIN;
7594}
7595
7596int nfs4_proc_setlease(struct file *file, int arg, struct file_lock **lease,
7597		       void **priv)
7598{
7599	switch (arg) {
7600	case F_RDLCK:
7601	case F_WRLCK:
7602		return nfs4_add_lease(file, arg, lease, priv);
7603	case F_UNLCK:
7604		return nfs4_delete_lease(file, priv);
7605	default:
7606		return -EINVAL;
7607	}
7608}
7609
7610int nfs4_lock_delegation_recall(struct file_lock *fl, struct nfs4_state *state, const nfs4_stateid *stateid)
7611{
7612	struct nfs_server *server = NFS_SERVER(state->inode);
7613	int err;
7614
7615	err = nfs4_set_lock_state(state, fl);
7616	if (err != 0)
7617		return err;
7618	do {
7619		err = _nfs4_do_setlk(state, F_SETLK, fl, NFS_LOCK_NEW);
7620		if (err != -NFS4ERR_DELAY)
7621			break;
7622		ssleep(1);
7623	} while (err == -NFS4ERR_DELAY);
7624	return nfs4_handle_delegation_recall_error(server, state, stateid, fl, err);
7625}
7626
7627struct nfs_release_lockowner_data {
7628	struct nfs4_lock_state *lsp;
7629	struct nfs_server *server;
7630	struct nfs_release_lockowner_args args;
7631	struct nfs_release_lockowner_res res;
7632	unsigned long timestamp;
7633};
7634
7635static void nfs4_release_lockowner_prepare(struct rpc_task *task, void *calldata)
7636{
7637	struct nfs_release_lockowner_data *data = calldata;
7638	struct nfs_server *server = data->server;
7639	nfs4_setup_sequence(server->nfs_client, &data->args.seq_args,
7640			   &data->res.seq_res, task);
7641	data->args.lock_owner.clientid = server->nfs_client->cl_clientid;
7642	data->timestamp = jiffies;
7643}
7644
7645static void nfs4_release_lockowner_done(struct rpc_task *task, void *calldata)
7646{
7647	struct nfs_release_lockowner_data *data = calldata;
7648	struct nfs_server *server = data->server;
7649
7650	nfs40_sequence_done(task, &data->res.seq_res);
7651
7652	switch (task->tk_status) {
7653	case 0:
7654		renew_lease(server, data->timestamp);
7655		break;
7656	case -NFS4ERR_STALE_CLIENTID:
7657	case -NFS4ERR_EXPIRED:
7658		nfs4_schedule_lease_recovery(server->nfs_client);
7659		break;
7660	case -NFS4ERR_LEASE_MOVED:
7661	case -NFS4ERR_DELAY:
7662		if (nfs4_async_handle_error(task, server,
7663					    NULL, NULL) == -EAGAIN)
7664			rpc_restart_call_prepare(task);
7665	}
7666}
7667
7668static void nfs4_release_lockowner_release(void *calldata)
7669{
7670	struct nfs_release_lockowner_data *data = calldata;
7671	nfs4_free_lock_state(data->server, data->lsp);
7672	kfree(calldata);
7673}
7674
7675static const struct rpc_call_ops nfs4_release_lockowner_ops = {
7676	.rpc_call_prepare = nfs4_release_lockowner_prepare,
7677	.rpc_call_done = nfs4_release_lockowner_done,
7678	.rpc_release = nfs4_release_lockowner_release,
7679};
7680
7681static void
7682nfs4_release_lockowner(struct nfs_server *server, struct nfs4_lock_state *lsp)
7683{
7684	struct nfs_release_lockowner_data *data;
7685	struct rpc_message msg = {
7686		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RELEASE_LOCKOWNER],
7687	};
7688
7689	if (server->nfs_client->cl_mvops->minor_version != 0)
7690		return;
7691
7692	data = kmalloc(sizeof(*data), GFP_KERNEL);
7693	if (!data)
7694		return;
7695	data->lsp = lsp;
7696	data->server = server;
7697	data->args.lock_owner.clientid = server->nfs_client->cl_clientid;
7698	data->args.lock_owner.id = lsp->ls_seqid.owner_id;
7699	data->args.lock_owner.s_dev = server->s_dev;
7700
7701	msg.rpc_argp = &data->args;
7702	msg.rpc_resp = &data->res;
7703	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 0, 0);
7704	rpc_call_async(server->client, &msg, 0, &nfs4_release_lockowner_ops, data);
7705}
7706
7707#define XATTR_NAME_NFSV4_ACL "system.nfs4_acl"
7708
7709static int nfs4_xattr_set_nfs4_acl(const struct xattr_handler *handler,
7710				   struct mnt_idmap *idmap,
7711				   struct dentry *unused, struct inode *inode,
7712				   const char *key, const void *buf,
7713				   size_t buflen, int flags)
7714{
7715	return nfs4_proc_set_acl(inode, buf, buflen, NFS4ACL_ACL);
7716}
7717
7718static int nfs4_xattr_get_nfs4_acl(const struct xattr_handler *handler,
7719				   struct dentry *unused, struct inode *inode,
7720				   const char *key, void *buf, size_t buflen)
7721{
7722	return nfs4_proc_get_acl(inode, buf, buflen, NFS4ACL_ACL);
7723}
7724
7725static bool nfs4_xattr_list_nfs4_acl(struct dentry *dentry)
7726{
7727	return nfs4_server_supports_acls(NFS_SB(dentry->d_sb), NFS4ACL_ACL);
7728}
7729
7730#if defined(CONFIG_NFS_V4_1)
7731#define XATTR_NAME_NFSV4_DACL "system.nfs4_dacl"
7732
7733static int nfs4_xattr_set_nfs4_dacl(const struct xattr_handler *handler,
7734				    struct mnt_idmap *idmap,
7735				    struct dentry *unused, struct inode *inode,
7736				    const char *key, const void *buf,
7737				    size_t buflen, int flags)
7738{
7739	return nfs4_proc_set_acl(inode, buf, buflen, NFS4ACL_DACL);
7740}
7741
7742static int nfs4_xattr_get_nfs4_dacl(const struct xattr_handler *handler,
7743				    struct dentry *unused, struct inode *inode,
7744				    const char *key, void *buf, size_t buflen)
7745{
7746	return nfs4_proc_get_acl(inode, buf, buflen, NFS4ACL_DACL);
7747}
7748
7749static bool nfs4_xattr_list_nfs4_dacl(struct dentry *dentry)
7750{
7751	return nfs4_server_supports_acls(NFS_SB(dentry->d_sb), NFS4ACL_DACL);
7752}
7753
7754#define XATTR_NAME_NFSV4_SACL "system.nfs4_sacl"
7755
7756static int nfs4_xattr_set_nfs4_sacl(const struct xattr_handler *handler,
7757				    struct mnt_idmap *idmap,
7758				    struct dentry *unused, struct inode *inode,
7759				    const char *key, const void *buf,
7760				    size_t buflen, int flags)
7761{
7762	return nfs4_proc_set_acl(inode, buf, buflen, NFS4ACL_SACL);
7763}
7764
7765static int nfs4_xattr_get_nfs4_sacl(const struct xattr_handler *handler,
7766				    struct dentry *unused, struct inode *inode,
7767				    const char *key, void *buf, size_t buflen)
7768{
7769	return nfs4_proc_get_acl(inode, buf, buflen, NFS4ACL_SACL);
7770}
7771
7772static bool nfs4_xattr_list_nfs4_sacl(struct dentry *dentry)
7773{
7774	return nfs4_server_supports_acls(NFS_SB(dentry->d_sb), NFS4ACL_SACL);
7775}
7776
7777#endif
7778
7779#ifdef CONFIG_NFS_V4_SECURITY_LABEL
7780
7781static int nfs4_xattr_set_nfs4_label(const struct xattr_handler *handler,
7782				     struct mnt_idmap *idmap,
7783				     struct dentry *unused, struct inode *inode,
7784				     const char *key, const void *buf,
7785				     size_t buflen, int flags)
7786{
7787	if (security_ismaclabel(key))
7788		return nfs4_set_security_label(inode, buf, buflen);
7789
7790	return -EOPNOTSUPP;
7791}
7792
7793static int nfs4_xattr_get_nfs4_label(const struct xattr_handler *handler,
7794				     struct dentry *unused, struct inode *inode,
7795				     const char *key, void *buf, size_t buflen)
7796{
7797	if (security_ismaclabel(key))
7798		return nfs4_get_security_label(inode, buf, buflen);
7799	return -EOPNOTSUPP;
7800}
7801
7802static ssize_t
7803nfs4_listxattr_nfs4_label(struct inode *inode, char *list, size_t list_len)
7804{
7805	int len = 0;
7806
7807	if (nfs_server_capable(inode, NFS_CAP_SECURITY_LABEL)) {
7808		len = security_inode_listsecurity(inode, list, list_len);
7809		if (len >= 0 && list_len && len > list_len)
7810			return -ERANGE;
7811	}
7812	return len;
7813}
7814
7815static const struct xattr_handler nfs4_xattr_nfs4_label_handler = {
7816	.prefix = XATTR_SECURITY_PREFIX,
7817	.get	= nfs4_xattr_get_nfs4_label,
7818	.set	= nfs4_xattr_set_nfs4_label,
7819};
7820
7821#else
7822
7823static ssize_t
7824nfs4_listxattr_nfs4_label(struct inode *inode, char *list, size_t list_len)
7825{
7826	return 0;
7827}
7828
7829#endif
7830
7831#ifdef CONFIG_NFS_V4_2
7832static int nfs4_xattr_set_nfs4_user(const struct xattr_handler *handler,
7833				    struct mnt_idmap *idmap,
7834				    struct dentry *unused, struct inode *inode,
7835				    const char *key, const void *buf,
7836				    size_t buflen, int flags)
7837{
7838	u32 mask;
7839	int ret;
7840
7841	if (!nfs_server_capable(inode, NFS_CAP_XATTR))
7842		return -EOPNOTSUPP;
7843
7844	/*
7845	 * There is no mapping from the MAY_* flags to the NFS_ACCESS_XA*
7846	 * flags right now. Handling of xattr operations use the normal
7847	 * file read/write permissions.
7848	 *
7849	 * Just in case the server has other ideas (which RFC 8276 allows),
7850	 * do a cached access check for the XA* flags to possibly avoid
7851	 * doing an RPC and getting EACCES back.
7852	 */
7853	if (!nfs_access_get_cached(inode, current_cred(), &mask, true)) {
7854		if (!(mask & NFS_ACCESS_XAWRITE))
7855			return -EACCES;
7856	}
7857
7858	if (buf == NULL) {
7859		ret = nfs42_proc_removexattr(inode, key);
7860		if (!ret)
7861			nfs4_xattr_cache_remove(inode, key);
7862	} else {
7863		ret = nfs42_proc_setxattr(inode, key, buf, buflen, flags);
7864		if (!ret)
7865			nfs4_xattr_cache_add(inode, key, buf, NULL, buflen);
7866	}
7867
7868	return ret;
7869}
7870
7871static int nfs4_xattr_get_nfs4_user(const struct xattr_handler *handler,
7872				    struct dentry *unused, struct inode *inode,
7873				    const char *key, void *buf, size_t buflen)
7874{
7875	u32 mask;
7876	ssize_t ret;
7877
7878	if (!nfs_server_capable(inode, NFS_CAP_XATTR))
7879		return -EOPNOTSUPP;
7880
7881	if (!nfs_access_get_cached(inode, current_cred(), &mask, true)) {
7882		if (!(mask & NFS_ACCESS_XAREAD))
7883			return -EACCES;
7884	}
7885
7886	ret = nfs_revalidate_inode(inode, NFS_INO_INVALID_CHANGE);
7887	if (ret)
7888		return ret;
7889
7890	ret = nfs4_xattr_cache_get(inode, key, buf, buflen);
7891	if (ret >= 0 || (ret < 0 && ret != -ENOENT))
7892		return ret;
7893
7894	ret = nfs42_proc_getxattr(inode, key, buf, buflen);
7895
7896	return ret;
7897}
7898
7899static ssize_t
7900nfs4_listxattr_nfs4_user(struct inode *inode, char *list, size_t list_len)
7901{
7902	u64 cookie;
7903	bool eof;
7904	ssize_t ret, size;
7905	char *buf;
7906	size_t buflen;
7907	u32 mask;
7908
7909	if (!nfs_server_capable(inode, NFS_CAP_XATTR))
7910		return 0;
7911
7912	if (!nfs_access_get_cached(inode, current_cred(), &mask, true)) {
7913		if (!(mask & NFS_ACCESS_XALIST))
7914			return 0;
7915	}
7916
7917	ret = nfs_revalidate_inode(inode, NFS_INO_INVALID_CHANGE);
7918	if (ret)
7919		return ret;
7920
7921	ret = nfs4_xattr_cache_list(inode, list, list_len);
7922	if (ret >= 0 || (ret < 0 && ret != -ENOENT))
7923		return ret;
7924
7925	cookie = 0;
7926	eof = false;
7927	buflen = list_len ? list_len : XATTR_LIST_MAX;
7928	buf = list_len ? list : NULL;
7929	size = 0;
7930
7931	while (!eof) {
7932		ret = nfs42_proc_listxattrs(inode, buf, buflen,
7933		    &cookie, &eof);
7934		if (ret < 0)
7935			return ret;
7936
7937		if (list_len) {
7938			buf += ret;
7939			buflen -= ret;
7940		}
7941		size += ret;
7942	}
7943
7944	if (list_len)
7945		nfs4_xattr_cache_set_list(inode, list, size);
7946
7947	return size;
7948}
7949
7950#else
7951
7952static ssize_t
7953nfs4_listxattr_nfs4_user(struct inode *inode, char *list, size_t list_len)
7954{
7955	return 0;
7956}
7957#endif /* CONFIG_NFS_V4_2 */
7958
7959/*
7960 * nfs_fhget will use either the mounted_on_fileid or the fileid
7961 */
7962static void nfs_fixup_referral_attributes(struct nfs_fattr *fattr)
7963{
7964	if (!(((fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID) ||
7965	       (fattr->valid & NFS_ATTR_FATTR_FILEID)) &&
7966	      (fattr->valid & NFS_ATTR_FATTR_FSID) &&
7967	      (fattr->valid & NFS_ATTR_FATTR_V4_LOCATIONS)))
7968		return;
7969
7970	fattr->valid |= NFS_ATTR_FATTR_TYPE | NFS_ATTR_FATTR_MODE |
7971		NFS_ATTR_FATTR_NLINK | NFS_ATTR_FATTR_V4_REFERRAL;
7972	fattr->mode = S_IFDIR | S_IRUGO | S_IXUGO;
7973	fattr->nlink = 2;
7974}
7975
7976static int _nfs4_proc_fs_locations(struct rpc_clnt *client, struct inode *dir,
7977				   const struct qstr *name,
7978				   struct nfs4_fs_locations *fs_locations,
7979				   struct page *page)
7980{
7981	struct nfs_server *server = NFS_SERVER(dir);
7982	u32 bitmask[3];
7983	struct nfs4_fs_locations_arg args = {
7984		.dir_fh = NFS_FH(dir),
7985		.name = name,
7986		.page = page,
7987		.bitmask = bitmask,
7988	};
7989	struct nfs4_fs_locations_res res = {
7990		.fs_locations = fs_locations,
7991	};
7992	struct rpc_message msg = {
7993		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FS_LOCATIONS],
7994		.rpc_argp = &args,
7995		.rpc_resp = &res,
7996	};
7997	int status;
7998
7999	dprintk("%s: start\n", __func__);
8000
8001	bitmask[0] = nfs4_fattr_bitmap[0] | FATTR4_WORD0_FS_LOCATIONS;
8002	bitmask[1] = nfs4_fattr_bitmap[1];
8003
8004	/* Ask for the fileid of the absent filesystem if mounted_on_fileid
8005	 * is not supported */
8006	if (NFS_SERVER(dir)->attr_bitmask[1] & FATTR4_WORD1_MOUNTED_ON_FILEID)
8007		bitmask[0] &= ~FATTR4_WORD0_FILEID;
8008	else
8009		bitmask[1] &= ~FATTR4_WORD1_MOUNTED_ON_FILEID;
8010
8011	nfs_fattr_init(fs_locations->fattr);
8012	fs_locations->server = server;
8013	fs_locations->nlocations = 0;
8014	status = nfs4_call_sync(client, server, &msg, &args.seq_args, &res.seq_res, 0);
8015	dprintk("%s: returned status = %d\n", __func__, status);
8016	return status;
8017}
8018
8019int nfs4_proc_fs_locations(struct rpc_clnt *client, struct inode *dir,
8020			   const struct qstr *name,
8021			   struct nfs4_fs_locations *fs_locations,
8022			   struct page *page)
8023{
8024	struct nfs4_exception exception = {
8025		.interruptible = true,
8026	};
8027	int err;
8028	do {
8029		err = _nfs4_proc_fs_locations(client, dir, name,
8030				fs_locations, page);
8031		trace_nfs4_get_fs_locations(dir, name, err);
8032		err = nfs4_handle_exception(NFS_SERVER(dir), err,
8033				&exception);
8034	} while (exception.retry);
8035	return err;
8036}
8037
8038/*
8039 * This operation also signals the server that this client is
8040 * performing migration recovery.  The server can stop returning
8041 * NFS4ERR_LEASE_MOVED to this client.  A RENEW operation is
8042 * appended to this compound to identify the client ID which is
8043 * performing recovery.
8044 */
8045static int _nfs40_proc_get_locations(struct nfs_server *server,
8046				     struct nfs_fh *fhandle,
8047				     struct nfs4_fs_locations *locations,
8048				     struct page *page, const struct cred *cred)
8049{
8050	struct rpc_clnt *clnt = server->client;
8051	u32 bitmask[2] = {
8052		[0] = FATTR4_WORD0_FSID | FATTR4_WORD0_FS_LOCATIONS,
8053	};
8054	struct nfs4_fs_locations_arg args = {
8055		.clientid	= server->nfs_client->cl_clientid,
8056		.fh		= fhandle,
8057		.page		= page,
8058		.bitmask	= bitmask,
8059		.migration	= 1,		/* skip LOOKUP */
8060		.renew		= 1,		/* append RENEW */
8061	};
8062	struct nfs4_fs_locations_res res = {
8063		.fs_locations	= locations,
8064		.migration	= 1,
8065		.renew		= 1,
8066	};
8067	struct rpc_message msg = {
8068		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_FS_LOCATIONS],
8069		.rpc_argp	= &args,
8070		.rpc_resp	= &res,
8071		.rpc_cred	= cred,
8072	};
8073	unsigned long now = jiffies;
8074	int status;
8075
8076	nfs_fattr_init(locations->fattr);
8077	locations->server = server;
8078	locations->nlocations = 0;
8079
8080	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0, 1);
8081	status = nfs4_call_sync_sequence(clnt, server, &msg,
8082					&args.seq_args, &res.seq_res);
8083	if (status)
8084		return status;
8085
8086	renew_lease(server, now);
8087	return 0;
8088}
8089
8090#ifdef CONFIG_NFS_V4_1
8091
8092/*
8093 * This operation also signals the server that this client is
8094 * performing migration recovery.  The server can stop asserting
8095 * SEQ4_STATUS_LEASE_MOVED for this client.  The client ID
8096 * performing this operation is identified in the SEQUENCE
8097 * operation in this compound.
8098 *
8099 * When the client supports GETATTR(fs_locations_info), it can
8100 * be plumbed in here.
8101 */
8102static int _nfs41_proc_get_locations(struct nfs_server *server,
8103				     struct nfs_fh *fhandle,
8104				     struct nfs4_fs_locations *locations,
8105				     struct page *page, const struct cred *cred)
8106{
8107	struct rpc_clnt *clnt = server->client;
8108	u32 bitmask[2] = {
8109		[0] = FATTR4_WORD0_FSID | FATTR4_WORD0_FS_LOCATIONS,
8110	};
8111	struct nfs4_fs_locations_arg args = {
8112		.fh		= fhandle,
8113		.page		= page,
8114		.bitmask	= bitmask,
8115		.migration	= 1,		/* skip LOOKUP */
8116	};
8117	struct nfs4_fs_locations_res res = {
8118		.fs_locations	= locations,
8119		.migration	= 1,
8120	};
8121	struct rpc_message msg = {
8122		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_FS_LOCATIONS],
8123		.rpc_argp	= &args,
8124		.rpc_resp	= &res,
8125		.rpc_cred	= cred,
8126	};
8127	struct nfs4_call_sync_data data = {
8128		.seq_server = server,
8129		.seq_args = &args.seq_args,
8130		.seq_res = &res.seq_res,
8131	};
8132	struct rpc_task_setup task_setup_data = {
8133		.rpc_client = clnt,
8134		.rpc_message = &msg,
8135		.callback_ops = server->nfs_client->cl_mvops->call_sync_ops,
8136		.callback_data = &data,
8137		.flags = RPC_TASK_NO_ROUND_ROBIN,
8138	};
8139	int status;
8140
8141	nfs_fattr_init(locations->fattr);
8142	locations->server = server;
8143	locations->nlocations = 0;
8144
8145	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0, 1);
8146	status = nfs4_call_sync_custom(&task_setup_data);
8147	if (status == NFS4_OK &&
8148	    res.seq_res.sr_status_flags & SEQ4_STATUS_LEASE_MOVED)
8149		status = -NFS4ERR_LEASE_MOVED;
8150	return status;
8151}
8152
8153#endif	/* CONFIG_NFS_V4_1 */
8154
8155/**
8156 * nfs4_proc_get_locations - discover locations for a migrated FSID
8157 * @server: pointer to nfs_server to process
8158 * @fhandle: pointer to the kernel NFS client file handle
8159 * @locations: result of query
8160 * @page: buffer
8161 * @cred: credential to use for this operation
8162 *
8163 * Returns NFS4_OK on success, a negative NFS4ERR status code if the
8164 * operation failed, or a negative errno if a local error occurred.
8165 *
8166 * On success, "locations" is filled in, but if the server has
8167 * no locations information, NFS_ATTR_FATTR_V4_LOCATIONS is not
8168 * asserted.
8169 *
8170 * -NFS4ERR_LEASE_MOVED is returned if the server still has leases
8171 * from this client that require migration recovery.
8172 */
8173int nfs4_proc_get_locations(struct nfs_server *server,
8174			    struct nfs_fh *fhandle,
8175			    struct nfs4_fs_locations *locations,
8176			    struct page *page, const struct cred *cred)
8177{
8178	struct nfs_client *clp = server->nfs_client;
8179	const struct nfs4_mig_recovery_ops *ops =
8180					clp->cl_mvops->mig_recovery_ops;
8181	struct nfs4_exception exception = {
8182		.interruptible = true,
8183	};
8184	int status;
8185
8186	dprintk("%s: FSID %llx:%llx on \"%s\"\n", __func__,
8187		(unsigned long long)server->fsid.major,
8188		(unsigned long long)server->fsid.minor,
8189		clp->cl_hostname);
8190	nfs_display_fhandle(fhandle, __func__);
8191
8192	do {
8193		status = ops->get_locations(server, fhandle, locations, page,
8194					    cred);
8195		if (status != -NFS4ERR_DELAY)
8196			break;
8197		nfs4_handle_exception(server, status, &exception);
8198	} while (exception.retry);
8199	return status;
8200}
8201
8202/*
8203 * This operation also signals the server that this client is
8204 * performing "lease moved" recovery.  The server can stop
8205 * returning NFS4ERR_LEASE_MOVED to this client.  A RENEW operation
8206 * is appended to this compound to identify the client ID which is
8207 * performing recovery.
8208 */
8209static int _nfs40_proc_fsid_present(struct inode *inode, const struct cred *cred)
8210{
8211	struct nfs_server *server = NFS_SERVER(inode);
8212	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
8213	struct rpc_clnt *clnt = server->client;
8214	struct nfs4_fsid_present_arg args = {
8215		.fh		= NFS_FH(inode),
8216		.clientid	= clp->cl_clientid,
8217		.renew		= 1,		/* append RENEW */
8218	};
8219	struct nfs4_fsid_present_res res = {
8220		.renew		= 1,
8221	};
8222	struct rpc_message msg = {
8223		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_FSID_PRESENT],
8224		.rpc_argp	= &args,
8225		.rpc_resp	= &res,
8226		.rpc_cred	= cred,
8227	};
8228	unsigned long now = jiffies;
8229	int status;
8230
8231	res.fh = nfs_alloc_fhandle();
8232	if (res.fh == NULL)
8233		return -ENOMEM;
8234
8235	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0, 1);
8236	status = nfs4_call_sync_sequence(clnt, server, &msg,
8237						&args.seq_args, &res.seq_res);
8238	nfs_free_fhandle(res.fh);
8239	if (status)
8240		return status;
8241
8242	do_renew_lease(clp, now);
8243	return 0;
8244}
8245
8246#ifdef CONFIG_NFS_V4_1
8247
8248/*
8249 * This operation also signals the server that this client is
8250 * performing "lease moved" recovery.  The server can stop asserting
8251 * SEQ4_STATUS_LEASE_MOVED for this client.  The client ID performing
8252 * this operation is identified in the SEQUENCE operation in this
8253 * compound.
8254 */
8255static int _nfs41_proc_fsid_present(struct inode *inode, const struct cred *cred)
8256{
8257	struct nfs_server *server = NFS_SERVER(inode);
8258	struct rpc_clnt *clnt = server->client;
8259	struct nfs4_fsid_present_arg args = {
8260		.fh		= NFS_FH(inode),
8261	};
8262	struct nfs4_fsid_present_res res = {
8263	};
8264	struct rpc_message msg = {
8265		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_FSID_PRESENT],
8266		.rpc_argp	= &args,
8267		.rpc_resp	= &res,
8268		.rpc_cred	= cred,
8269	};
8270	int status;
8271
8272	res.fh = nfs_alloc_fhandle();
8273	if (res.fh == NULL)
8274		return -ENOMEM;
8275
8276	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0, 1);
8277	status = nfs4_call_sync_sequence(clnt, server, &msg,
8278						&args.seq_args, &res.seq_res);
8279	nfs_free_fhandle(res.fh);
8280	if (status == NFS4_OK &&
8281	    res.seq_res.sr_status_flags & SEQ4_STATUS_LEASE_MOVED)
8282		status = -NFS4ERR_LEASE_MOVED;
8283	return status;
8284}
8285
8286#endif	/* CONFIG_NFS_V4_1 */
8287
8288/**
8289 * nfs4_proc_fsid_present - Is this FSID present or absent on server?
8290 * @inode: inode on FSID to check
8291 * @cred: credential to use for this operation
8292 *
8293 * Server indicates whether the FSID is present, moved, or not
8294 * recognized.  This operation is necessary to clear a LEASE_MOVED
8295 * condition for this client ID.
8296 *
8297 * Returns NFS4_OK if the FSID is present on this server,
8298 * -NFS4ERR_MOVED if the FSID is no longer present, a negative
8299 *  NFS4ERR code if some error occurred on the server, or a
8300 *  negative errno if a local failure occurred.
8301 */
8302int nfs4_proc_fsid_present(struct inode *inode, const struct cred *cred)
8303{
8304	struct nfs_server *server = NFS_SERVER(inode);
8305	struct nfs_client *clp = server->nfs_client;
8306	const struct nfs4_mig_recovery_ops *ops =
8307					clp->cl_mvops->mig_recovery_ops;
8308	struct nfs4_exception exception = {
8309		.interruptible = true,
8310	};
8311	int status;
8312
8313	dprintk("%s: FSID %llx:%llx on \"%s\"\n", __func__,
8314		(unsigned long long)server->fsid.major,
8315		(unsigned long long)server->fsid.minor,
8316		clp->cl_hostname);
8317	nfs_display_fhandle(NFS_FH(inode), __func__);
8318
8319	do {
8320		status = ops->fsid_present(inode, cred);
8321		if (status != -NFS4ERR_DELAY)
8322			break;
8323		nfs4_handle_exception(server, status, &exception);
8324	} while (exception.retry);
8325	return status;
8326}
8327
8328/*
8329 * If 'use_integrity' is true and the state managment nfs_client
8330 * cl_rpcclient is using krb5i/p, use the integrity protected cl_rpcclient
8331 * and the machine credential as per RFC3530bis and RFC5661 Security
8332 * Considerations sections. Otherwise, just use the user cred with the
8333 * filesystem's rpc_client.
8334 */
8335static int _nfs4_proc_secinfo(struct inode *dir, const struct qstr *name, struct nfs4_secinfo_flavors *flavors, bool use_integrity)
8336{
8337	int status;
8338	struct rpc_clnt *clnt = NFS_SERVER(dir)->client;
8339	struct nfs_client *clp = NFS_SERVER(dir)->nfs_client;
8340	struct nfs4_secinfo_arg args = {
8341		.dir_fh = NFS_FH(dir),
8342		.name   = name,
8343	};
8344	struct nfs4_secinfo_res res = {
8345		.flavors     = flavors,
8346	};
8347	struct rpc_message msg = {
8348		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SECINFO],
8349		.rpc_argp = &args,
8350		.rpc_resp = &res,
8351	};
8352	struct nfs4_call_sync_data data = {
8353		.seq_server = NFS_SERVER(dir),
8354		.seq_args = &args.seq_args,
8355		.seq_res = &res.seq_res,
8356	};
8357	struct rpc_task_setup task_setup = {
8358		.rpc_client = clnt,
8359		.rpc_message = &msg,
8360		.callback_ops = clp->cl_mvops->call_sync_ops,
8361		.callback_data = &data,
8362		.flags = RPC_TASK_NO_ROUND_ROBIN,
8363	};
8364	const struct cred *cred = NULL;
8365
8366	if (use_integrity) {
8367		clnt = clp->cl_rpcclient;
8368		task_setup.rpc_client = clnt;
8369
8370		cred = nfs4_get_clid_cred(clp);
8371		msg.rpc_cred = cred;
8372	}
8373
8374	dprintk("NFS call  secinfo %s\n", name->name);
8375
8376	nfs4_state_protect(clp, NFS_SP4_MACH_CRED_SECINFO, &clnt, &msg);
8377	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0, 0);
8378	status = nfs4_call_sync_custom(&task_setup);
8379
8380	dprintk("NFS reply  secinfo: %d\n", status);
8381
8382	put_cred(cred);
8383	return status;
8384}
8385
8386int nfs4_proc_secinfo(struct inode *dir, const struct qstr *name,
8387		      struct nfs4_secinfo_flavors *flavors)
8388{
8389	struct nfs4_exception exception = {
8390		.interruptible = true,
8391	};
8392	int err;
8393	do {
8394		err = -NFS4ERR_WRONGSEC;
8395
8396		/* try to use integrity protection with machine cred */
8397		if (_nfs4_is_integrity_protected(NFS_SERVER(dir)->nfs_client))
8398			err = _nfs4_proc_secinfo(dir, name, flavors, true);
8399
8400		/*
8401		 * if unable to use integrity protection, or SECINFO with
8402		 * integrity protection returns NFS4ERR_WRONGSEC (which is
8403		 * disallowed by spec, but exists in deployed servers) use
8404		 * the current filesystem's rpc_client and the user cred.
8405		 */
8406		if (err == -NFS4ERR_WRONGSEC)
8407			err = _nfs4_proc_secinfo(dir, name, flavors, false);
8408
8409		trace_nfs4_secinfo(dir, name, err);
8410		err = nfs4_handle_exception(NFS_SERVER(dir), err,
8411				&exception);
8412	} while (exception.retry);
8413	return err;
8414}
8415
8416#ifdef CONFIG_NFS_V4_1
8417/*
8418 * Check the exchange flags returned by the server for invalid flags, having
8419 * both PNFS and NON_PNFS flags set, and not having one of NON_PNFS, PNFS, or
8420 * DS flags set.
8421 */
8422static int nfs4_check_cl_exchange_flags(u32 flags, u32 version)
8423{
8424	if (version >= 2 && (flags & ~EXCHGID4_2_FLAG_MASK_R))
8425		goto out_inval;
8426	else if (version < 2 && (flags & ~EXCHGID4_FLAG_MASK_R))
8427		goto out_inval;
8428	if ((flags & EXCHGID4_FLAG_USE_PNFS_MDS) &&
8429	    (flags & EXCHGID4_FLAG_USE_NON_PNFS))
8430		goto out_inval;
8431	if (!(flags & (EXCHGID4_FLAG_MASK_PNFS)))
8432		goto out_inval;
8433	return NFS_OK;
8434out_inval:
8435	return -NFS4ERR_INVAL;
8436}
8437
8438static bool
8439nfs41_same_server_scope(struct nfs41_server_scope *a,
8440			struct nfs41_server_scope *b)
8441{
8442	if (a->server_scope_sz != b->server_scope_sz)
8443		return false;
8444	return memcmp(a->server_scope, b->server_scope, a->server_scope_sz) == 0;
8445}
8446
8447static void
8448nfs4_bind_one_conn_to_session_done(struct rpc_task *task, void *calldata)
8449{
8450	struct nfs41_bind_conn_to_session_args *args = task->tk_msg.rpc_argp;
8451	struct nfs41_bind_conn_to_session_res *res = task->tk_msg.rpc_resp;
8452	struct nfs_client *clp = args->client;
8453
8454	switch (task->tk_status) {
8455	case -NFS4ERR_BADSESSION:
8456	case -NFS4ERR_DEADSESSION:
8457		nfs4_schedule_session_recovery(clp->cl_session,
8458				task->tk_status);
8459		return;
8460	}
8461	if (args->dir == NFS4_CDFC4_FORE_OR_BOTH &&
8462			res->dir != NFS4_CDFS4_BOTH) {
8463		rpc_task_close_connection(task);
8464		if (args->retries++ < MAX_BIND_CONN_TO_SESSION_RETRIES)
8465			rpc_restart_call(task);
8466	}
8467}
8468
8469static const struct rpc_call_ops nfs4_bind_one_conn_to_session_ops = {
8470	.rpc_call_done =  nfs4_bind_one_conn_to_session_done,
8471};
8472
8473/*
8474 * nfs4_proc_bind_one_conn_to_session()
8475 *
8476 * The 4.1 client currently uses the same TCP connection for the
8477 * fore and backchannel.
8478 */
8479static
8480int nfs4_proc_bind_one_conn_to_session(struct rpc_clnt *clnt,
8481		struct rpc_xprt *xprt,
8482		struct nfs_client *clp,
8483		const struct cred *cred)
8484{
8485	int status;
8486	struct nfs41_bind_conn_to_session_args args = {
8487		.client = clp,
8488		.dir = NFS4_CDFC4_FORE_OR_BOTH,
8489		.retries = 0,
8490	};
8491	struct nfs41_bind_conn_to_session_res res;
8492	struct rpc_message msg = {
8493		.rpc_proc =
8494			&nfs4_procedures[NFSPROC4_CLNT_BIND_CONN_TO_SESSION],
8495		.rpc_argp = &args,
8496		.rpc_resp = &res,
8497		.rpc_cred = cred,
8498	};
8499	struct rpc_task_setup task_setup_data = {
8500		.rpc_client = clnt,
8501		.rpc_xprt = xprt,
8502		.callback_ops = &nfs4_bind_one_conn_to_session_ops,
8503		.rpc_message = &msg,
8504		.flags = RPC_TASK_TIMEOUT,
8505	};
8506	struct rpc_task *task;
8507
8508	nfs4_copy_sessionid(&args.sessionid, &clp->cl_session->sess_id);
8509	if (!(clp->cl_session->flags & SESSION4_BACK_CHAN))
8510		args.dir = NFS4_CDFC4_FORE;
8511
8512	/* Do not set the backchannel flag unless this is clnt->cl_xprt */
8513	if (xprt != rcu_access_pointer(clnt->cl_xprt))
8514		args.dir = NFS4_CDFC4_FORE;
8515
8516	task = rpc_run_task(&task_setup_data);
8517	if (!IS_ERR(task)) {
8518		status = task->tk_status;
8519		rpc_put_task(task);
8520	} else
8521		status = PTR_ERR(task);
8522	trace_nfs4_bind_conn_to_session(clp, status);
8523	if (status == 0) {
8524		if (memcmp(res.sessionid.data,
8525		    clp->cl_session->sess_id.data, NFS4_MAX_SESSIONID_LEN)) {
8526			dprintk("NFS: %s: Session ID mismatch\n", __func__);
8527			return -EIO;
8528		}
8529		if ((res.dir & args.dir) != res.dir || res.dir == 0) {
8530			dprintk("NFS: %s: Unexpected direction from server\n",
8531				__func__);
8532			return -EIO;
8533		}
8534		if (res.use_conn_in_rdma_mode != args.use_conn_in_rdma_mode) {
8535			dprintk("NFS: %s: Server returned RDMA mode = true\n",
8536				__func__);
8537			return -EIO;
8538		}
8539	}
8540
8541	return status;
8542}
8543
8544struct rpc_bind_conn_calldata {
8545	struct nfs_client *clp;
8546	const struct cred *cred;
8547};
8548
8549static int
8550nfs4_proc_bind_conn_to_session_callback(struct rpc_clnt *clnt,
8551		struct rpc_xprt *xprt,
8552		void *calldata)
8553{
8554	struct rpc_bind_conn_calldata *p = calldata;
8555
8556	return nfs4_proc_bind_one_conn_to_session(clnt, xprt, p->clp, p->cred);
8557}
8558
8559int nfs4_proc_bind_conn_to_session(struct nfs_client *clp, const struct cred *cred)
8560{
8561	struct rpc_bind_conn_calldata data = {
8562		.clp = clp,
8563		.cred = cred,
8564	};
8565	return rpc_clnt_iterate_for_each_xprt(clp->cl_rpcclient,
8566			nfs4_proc_bind_conn_to_session_callback, &data);
8567}
8568
8569/*
8570 * Minimum set of SP4_MACH_CRED operations from RFC 5661 in the enforce map
8571 * and operations we'd like to see to enable certain features in the allow map
8572 */
8573static const struct nfs41_state_protection nfs4_sp4_mach_cred_request = {
8574	.how = SP4_MACH_CRED,
8575	.enforce.u.words = {
8576		[1] = 1 << (OP_BIND_CONN_TO_SESSION - 32) |
8577		      1 << (OP_EXCHANGE_ID - 32) |
8578		      1 << (OP_CREATE_SESSION - 32) |
8579		      1 << (OP_DESTROY_SESSION - 32) |
8580		      1 << (OP_DESTROY_CLIENTID - 32)
8581	},
8582	.allow.u.words = {
8583		[0] = 1 << (OP_CLOSE) |
8584		      1 << (OP_OPEN_DOWNGRADE) |
8585		      1 << (OP_LOCKU) |
8586		      1 << (OP_DELEGRETURN) |
8587		      1 << (OP_COMMIT),
8588		[1] = 1 << (OP_SECINFO - 32) |
8589		      1 << (OP_SECINFO_NO_NAME - 32) |
8590		      1 << (OP_LAYOUTRETURN - 32) |
8591		      1 << (OP_TEST_STATEID - 32) |
8592		      1 << (OP_FREE_STATEID - 32) |
8593		      1 << (OP_WRITE - 32)
8594	}
8595};
8596
8597/*
8598 * Select the state protection mode for client `clp' given the server results
8599 * from exchange_id in `sp'.
8600 *
8601 * Returns 0 on success, negative errno otherwise.
8602 */
8603static int nfs4_sp4_select_mode(struct nfs_client *clp,
8604				 struct nfs41_state_protection *sp)
8605{
8606	static const u32 supported_enforce[NFS4_OP_MAP_NUM_WORDS] = {
8607		[1] = 1 << (OP_BIND_CONN_TO_SESSION - 32) |
8608		      1 << (OP_EXCHANGE_ID - 32) |
8609		      1 << (OP_CREATE_SESSION - 32) |
8610		      1 << (OP_DESTROY_SESSION - 32) |
8611		      1 << (OP_DESTROY_CLIENTID - 32)
8612	};
8613	unsigned long flags = 0;
8614	unsigned int i;
8615	int ret = 0;
8616
8617	if (sp->how == SP4_MACH_CRED) {
8618		/* Print state protect result */
8619		dfprintk(MOUNT, "Server SP4_MACH_CRED support:\n");
8620		for (i = 0; i <= LAST_NFS4_OP; i++) {
8621			if (test_bit(i, sp->enforce.u.longs))
8622				dfprintk(MOUNT, "  enforce op %d\n", i);
8623			if (test_bit(i, sp->allow.u.longs))
8624				dfprintk(MOUNT, "  allow op %d\n", i);
8625		}
8626
8627		/* make sure nothing is on enforce list that isn't supported */
8628		for (i = 0; i < NFS4_OP_MAP_NUM_WORDS; i++) {
8629			if (sp->enforce.u.words[i] & ~supported_enforce[i]) {
8630				dfprintk(MOUNT, "sp4_mach_cred: disabled\n");
8631				ret = -EINVAL;
8632				goto out;
8633			}
8634		}
8635
8636		/*
8637		 * Minimal mode - state operations are allowed to use machine
8638		 * credential.  Note this already happens by default, so the
8639		 * client doesn't have to do anything more than the negotiation.
8640		 *
8641		 * NOTE: we don't care if EXCHANGE_ID is in the list -
8642		 *       we're already using the machine cred for exchange_id
8643		 *       and will never use a different cred.
8644		 */
8645		if (test_bit(OP_BIND_CONN_TO_SESSION, sp->enforce.u.longs) &&
8646		    test_bit(OP_CREATE_SESSION, sp->enforce.u.longs) &&
8647		    test_bit(OP_DESTROY_SESSION, sp->enforce.u.longs) &&
8648		    test_bit(OP_DESTROY_CLIENTID, sp->enforce.u.longs)) {
8649			dfprintk(MOUNT, "sp4_mach_cred:\n");
8650			dfprintk(MOUNT, "  minimal mode enabled\n");
8651			__set_bit(NFS_SP4_MACH_CRED_MINIMAL, &flags);
8652		} else {
8653			dfprintk(MOUNT, "sp4_mach_cred: disabled\n");
8654			ret = -EINVAL;
8655			goto out;
8656		}
8657
8658		if (test_bit(OP_CLOSE, sp->allow.u.longs) &&
8659		    test_bit(OP_OPEN_DOWNGRADE, sp->allow.u.longs) &&
8660		    test_bit(OP_DELEGRETURN, sp->allow.u.longs) &&
8661		    test_bit(OP_LOCKU, sp->allow.u.longs)) {
8662			dfprintk(MOUNT, "  cleanup mode enabled\n");
8663			__set_bit(NFS_SP4_MACH_CRED_CLEANUP, &flags);
8664		}
8665
8666		if (test_bit(OP_LAYOUTRETURN, sp->allow.u.longs)) {
8667			dfprintk(MOUNT, "  pnfs cleanup mode enabled\n");
8668			__set_bit(NFS_SP4_MACH_CRED_PNFS_CLEANUP, &flags);
8669		}
8670
8671		if (test_bit(OP_SECINFO, sp->allow.u.longs) &&
8672		    test_bit(OP_SECINFO_NO_NAME, sp->allow.u.longs)) {
8673			dfprintk(MOUNT, "  secinfo mode enabled\n");
8674			__set_bit(NFS_SP4_MACH_CRED_SECINFO, &flags);
8675		}
8676
8677		if (test_bit(OP_TEST_STATEID, sp->allow.u.longs) &&
8678		    test_bit(OP_FREE_STATEID, sp->allow.u.longs)) {
8679			dfprintk(MOUNT, "  stateid mode enabled\n");
8680			__set_bit(NFS_SP4_MACH_CRED_STATEID, &flags);
8681		}
8682
8683		if (test_bit(OP_WRITE, sp->allow.u.longs)) {
8684			dfprintk(MOUNT, "  write mode enabled\n");
8685			__set_bit(NFS_SP4_MACH_CRED_WRITE, &flags);
8686		}
8687
8688		if (test_bit(OP_COMMIT, sp->allow.u.longs)) {
8689			dfprintk(MOUNT, "  commit mode enabled\n");
8690			__set_bit(NFS_SP4_MACH_CRED_COMMIT, &flags);
8691		}
8692	}
8693out:
8694	clp->cl_sp4_flags = flags;
8695	return ret;
8696}
8697
8698struct nfs41_exchange_id_data {
8699	struct nfs41_exchange_id_res res;
8700	struct nfs41_exchange_id_args args;
8701};
8702
8703static void nfs4_exchange_id_release(void *data)
8704{
8705	struct nfs41_exchange_id_data *cdata =
8706					(struct nfs41_exchange_id_data *)data;
8707
8708	nfs_put_client(cdata->args.client);
8709	kfree(cdata->res.impl_id);
8710	kfree(cdata->res.server_scope);
8711	kfree(cdata->res.server_owner);
8712	kfree(cdata);
8713}
8714
8715static const struct rpc_call_ops nfs4_exchange_id_call_ops = {
8716	.rpc_release = nfs4_exchange_id_release,
8717};
8718
8719/*
8720 * _nfs4_proc_exchange_id()
8721 *
8722 * Wrapper for EXCHANGE_ID operation.
8723 */
8724static struct rpc_task *
8725nfs4_run_exchange_id(struct nfs_client *clp, const struct cred *cred,
8726			u32 sp4_how, struct rpc_xprt *xprt)
8727{
8728	struct rpc_message msg = {
8729		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_EXCHANGE_ID],
8730		.rpc_cred = cred,
8731	};
8732	struct rpc_task_setup task_setup_data = {
8733		.rpc_client = clp->cl_rpcclient,
8734		.callback_ops = &nfs4_exchange_id_call_ops,
8735		.rpc_message = &msg,
8736		.flags = RPC_TASK_TIMEOUT | RPC_TASK_NO_ROUND_ROBIN,
8737	};
8738	struct nfs41_exchange_id_data *calldata;
8739	int status;
8740
8741	if (!refcount_inc_not_zero(&clp->cl_count))
8742		return ERR_PTR(-EIO);
8743
8744	status = -ENOMEM;
8745	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
8746	if (!calldata)
8747		goto out;
8748
8749	nfs4_init_boot_verifier(clp, &calldata->args.verifier);
8750
8751	status = nfs4_init_uniform_client_string(clp);
8752	if (status)
8753		goto out_calldata;
8754
8755	calldata->res.server_owner = kzalloc(sizeof(struct nfs41_server_owner),
8756						GFP_NOFS);
8757	status = -ENOMEM;
8758	if (unlikely(calldata->res.server_owner == NULL))
8759		goto out_calldata;
8760
8761	calldata->res.server_scope = kzalloc(sizeof(struct nfs41_server_scope),
8762					GFP_NOFS);
8763	if (unlikely(calldata->res.server_scope == NULL))
8764		goto out_server_owner;
8765
8766	calldata->res.impl_id = kzalloc(sizeof(struct nfs41_impl_id), GFP_NOFS);
8767	if (unlikely(calldata->res.impl_id == NULL))
8768		goto out_server_scope;
8769
8770	switch (sp4_how) {
8771	case SP4_NONE:
8772		calldata->args.state_protect.how = SP4_NONE;
8773		break;
8774
8775	case SP4_MACH_CRED:
8776		calldata->args.state_protect = nfs4_sp4_mach_cred_request;
8777		break;
8778
8779	default:
8780		/* unsupported! */
8781		WARN_ON_ONCE(1);
8782		status = -EINVAL;
8783		goto out_impl_id;
8784	}
8785	if (xprt) {
8786		task_setup_data.rpc_xprt = xprt;
8787		task_setup_data.flags |= RPC_TASK_SOFTCONN;
8788		memcpy(calldata->args.verifier.data, clp->cl_confirm.data,
8789				sizeof(calldata->args.verifier.data));
8790	}
8791	calldata->args.client = clp;
8792	calldata->args.flags = EXCHGID4_FLAG_SUPP_MOVED_REFER |
8793	EXCHGID4_FLAG_BIND_PRINC_STATEID;
8794#ifdef CONFIG_NFS_V4_1_MIGRATION
8795	calldata->args.flags |= EXCHGID4_FLAG_SUPP_MOVED_MIGR;
8796#endif
8797	if (test_bit(NFS_CS_DS, &clp->cl_flags))
8798		calldata->args.flags |= EXCHGID4_FLAG_USE_PNFS_DS;
8799	msg.rpc_argp = &calldata->args;
8800	msg.rpc_resp = &calldata->res;
8801	task_setup_data.callback_data = calldata;
8802
8803	return rpc_run_task(&task_setup_data);
8804
8805out_impl_id:
8806	kfree(calldata->res.impl_id);
8807out_server_scope:
8808	kfree(calldata->res.server_scope);
8809out_server_owner:
8810	kfree(calldata->res.server_owner);
8811out_calldata:
8812	kfree(calldata);
8813out:
8814	nfs_put_client(clp);
8815	return ERR_PTR(status);
8816}
8817
8818/*
8819 * _nfs4_proc_exchange_id()
8820 *
8821 * Wrapper for EXCHANGE_ID operation.
8822 */
8823static int _nfs4_proc_exchange_id(struct nfs_client *clp, const struct cred *cred,
8824			u32 sp4_how)
8825{
8826	struct rpc_task *task;
8827	struct nfs41_exchange_id_args *argp;
8828	struct nfs41_exchange_id_res *resp;
8829	unsigned long now = jiffies;
8830	int status;
8831
8832	task = nfs4_run_exchange_id(clp, cred, sp4_how, NULL);
8833	if (IS_ERR(task))
8834		return PTR_ERR(task);
8835
8836	argp = task->tk_msg.rpc_argp;
8837	resp = task->tk_msg.rpc_resp;
8838	status = task->tk_status;
8839	if (status  != 0)
8840		goto out;
8841
8842	status = nfs4_check_cl_exchange_flags(resp->flags,
8843			clp->cl_mvops->minor_version);
8844	if (status  != 0)
8845		goto out;
8846
8847	status = nfs4_sp4_select_mode(clp, &resp->state_protect);
8848	if (status != 0)
8849		goto out;
8850
8851	do_renew_lease(clp, now);
8852
8853	clp->cl_clientid = resp->clientid;
8854	clp->cl_exchange_flags = resp->flags;
8855	clp->cl_seqid = resp->seqid;
8856	/* Client ID is not confirmed */
8857	if (!(resp->flags & EXCHGID4_FLAG_CONFIRMED_R))
8858		clear_bit(NFS4_SESSION_ESTABLISHED,
8859			  &clp->cl_session->session_state);
8860
8861	if (clp->cl_serverscope != NULL &&
8862	    !nfs41_same_server_scope(clp->cl_serverscope,
8863				resp->server_scope)) {
8864		dprintk("%s: server_scope mismatch detected\n",
8865			__func__);
8866		set_bit(NFS4CLNT_SERVER_SCOPE_MISMATCH, &clp->cl_state);
8867	}
8868
8869	swap(clp->cl_serverowner, resp->server_owner);
8870	swap(clp->cl_serverscope, resp->server_scope);
8871	swap(clp->cl_implid, resp->impl_id);
8872
8873	/* Save the EXCHANGE_ID verifier session trunk tests */
8874	memcpy(clp->cl_confirm.data, argp->verifier.data,
8875	       sizeof(clp->cl_confirm.data));
8876out:
8877	trace_nfs4_exchange_id(clp, status);
8878	rpc_put_task(task);
8879	return status;
8880}
8881
8882/*
8883 * nfs4_proc_exchange_id()
8884 *
8885 * Returns zero, a negative errno, or a negative NFS4ERR status code.
8886 *
8887 * Since the clientid has expired, all compounds using sessions
8888 * associated with the stale clientid will be returning
8889 * NFS4ERR_BADSESSION in the sequence operation, and will therefore
8890 * be in some phase of session reset.
8891 *
8892 * Will attempt to negotiate SP4_MACH_CRED if krb5i / krb5p auth is used.
8893 */
8894int nfs4_proc_exchange_id(struct nfs_client *clp, const struct cred *cred)
8895{
8896	rpc_authflavor_t authflavor = clp->cl_rpcclient->cl_auth->au_flavor;
8897	int status;
8898
8899	/* try SP4_MACH_CRED if krb5i/p	*/
8900	if (authflavor == RPC_AUTH_GSS_KRB5I ||
8901	    authflavor == RPC_AUTH_GSS_KRB5P) {
8902		status = _nfs4_proc_exchange_id(clp, cred, SP4_MACH_CRED);
8903		if (!status)
8904			return 0;
8905	}
8906
8907	/* try SP4_NONE */
8908	return _nfs4_proc_exchange_id(clp, cred, SP4_NONE);
8909}
8910
8911/**
8912 * nfs4_test_session_trunk
8913 *
8914 * This is an add_xprt_test() test function called from
8915 * rpc_clnt_setup_test_and_add_xprt.
8916 *
8917 * The rpc_xprt_switch is referrenced by rpc_clnt_setup_test_and_add_xprt
8918 * and is dereferrenced in nfs4_exchange_id_release
8919 *
8920 * Upon success, add the new transport to the rpc_clnt
8921 *
8922 * @clnt: struct rpc_clnt to get new transport
8923 * @xprt: the rpc_xprt to test
8924 * @data: call data for _nfs4_proc_exchange_id.
8925 */
8926void nfs4_test_session_trunk(struct rpc_clnt *clnt, struct rpc_xprt *xprt,
8927			    void *data)
8928{
8929	struct nfs4_add_xprt_data *adata = data;
8930	struct rpc_task *task;
8931	int status;
8932
8933	u32 sp4_how;
8934
8935	dprintk("--> %s try %s\n", __func__,
8936		xprt->address_strings[RPC_DISPLAY_ADDR]);
8937
8938	sp4_how = (adata->clp->cl_sp4_flags == 0 ? SP4_NONE : SP4_MACH_CRED);
8939
8940try_again:
8941	/* Test connection for session trunking. Async exchange_id call */
8942	task = nfs4_run_exchange_id(adata->clp, adata->cred, sp4_how, xprt);
8943	if (IS_ERR(task))
8944		return;
8945
8946	status = task->tk_status;
8947	if (status == 0)
8948		status = nfs4_detect_session_trunking(adata->clp,
8949				task->tk_msg.rpc_resp, xprt);
8950
8951	if (status == 0)
8952		rpc_clnt_xprt_switch_add_xprt(clnt, xprt);
8953	else if (status != -NFS4ERR_DELAY && rpc_clnt_xprt_switch_has_addr(clnt,
8954				(struct sockaddr *)&xprt->addr))
8955		rpc_clnt_xprt_switch_remove_xprt(clnt, xprt);
8956
8957	rpc_put_task(task);
8958	if (status == -NFS4ERR_DELAY) {
8959		ssleep(1);
8960		goto try_again;
8961	}
8962}
8963EXPORT_SYMBOL_GPL(nfs4_test_session_trunk);
8964
8965static int _nfs4_proc_destroy_clientid(struct nfs_client *clp,
8966		const struct cred *cred)
8967{
8968	struct rpc_message msg = {
8969		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DESTROY_CLIENTID],
8970		.rpc_argp = clp,
8971		.rpc_cred = cred,
8972	};
8973	int status;
8974
8975	status = rpc_call_sync(clp->cl_rpcclient, &msg,
8976			       RPC_TASK_TIMEOUT | RPC_TASK_NO_ROUND_ROBIN);
8977	trace_nfs4_destroy_clientid(clp, status);
8978	if (status)
8979		dprintk("NFS: Got error %d from the server %s on "
8980			"DESTROY_CLIENTID.", status, clp->cl_hostname);
8981	return status;
8982}
8983
8984static int nfs4_proc_destroy_clientid(struct nfs_client *clp,
8985		const struct cred *cred)
8986{
8987	unsigned int loop;
8988	int ret;
8989
8990	for (loop = NFS4_MAX_LOOP_ON_RECOVER; loop != 0; loop--) {
8991		ret = _nfs4_proc_destroy_clientid(clp, cred);
8992		switch (ret) {
8993		case -NFS4ERR_DELAY:
8994		case -NFS4ERR_CLIENTID_BUSY:
8995			ssleep(1);
8996			break;
8997		default:
8998			return ret;
8999		}
9000	}
9001	return 0;
9002}
9003
9004int nfs4_destroy_clientid(struct nfs_client *clp)
9005{
9006	const struct cred *cred;
9007	int ret = 0;
9008
9009	if (clp->cl_mvops->minor_version < 1)
9010		goto out;
9011	if (clp->cl_exchange_flags == 0)
9012		goto out;
9013	if (clp->cl_preserve_clid)
9014		goto out;
9015	cred = nfs4_get_clid_cred(clp);
9016	ret = nfs4_proc_destroy_clientid(clp, cred);
9017	put_cred(cred);
9018	switch (ret) {
9019	case 0:
9020	case -NFS4ERR_STALE_CLIENTID:
9021		clp->cl_exchange_flags = 0;
9022	}
9023out:
9024	return ret;
9025}
9026
9027#endif /* CONFIG_NFS_V4_1 */
9028
9029struct nfs4_get_lease_time_data {
9030	struct nfs4_get_lease_time_args *args;
9031	struct nfs4_get_lease_time_res *res;
9032	struct nfs_client *clp;
9033};
9034
9035static void nfs4_get_lease_time_prepare(struct rpc_task *task,
9036					void *calldata)
9037{
9038	struct nfs4_get_lease_time_data *data =
9039			(struct nfs4_get_lease_time_data *)calldata;
9040
9041	/* just setup sequence, do not trigger session recovery
9042	   since we're invoked within one */
9043	nfs4_setup_sequence(data->clp,
9044			&data->args->la_seq_args,
9045			&data->res->lr_seq_res,
9046			task);
9047}
9048
9049/*
9050 * Called from nfs4_state_manager thread for session setup, so don't recover
9051 * from sequence operation or clientid errors.
9052 */
9053static void nfs4_get_lease_time_done(struct rpc_task *task, void *calldata)
9054{
9055	struct nfs4_get_lease_time_data *data =
9056			(struct nfs4_get_lease_time_data *)calldata;
9057
9058	if (!nfs4_sequence_done(task, &data->res->lr_seq_res))
9059		return;
9060	switch (task->tk_status) {
9061	case -NFS4ERR_DELAY:
9062	case -NFS4ERR_GRACE:
9063		rpc_delay(task, NFS4_POLL_RETRY_MIN);
9064		task->tk_status = 0;
9065		fallthrough;
9066	case -NFS4ERR_RETRY_UNCACHED_REP:
9067		rpc_restart_call_prepare(task);
9068		return;
9069	}
9070}
9071
9072static const struct rpc_call_ops nfs4_get_lease_time_ops = {
9073	.rpc_call_prepare = nfs4_get_lease_time_prepare,
9074	.rpc_call_done = nfs4_get_lease_time_done,
9075};
9076
9077int nfs4_proc_get_lease_time(struct nfs_client *clp, struct nfs_fsinfo *fsinfo)
9078{
9079	struct nfs4_get_lease_time_args args;
9080	struct nfs4_get_lease_time_res res = {
9081		.lr_fsinfo = fsinfo,
9082	};
9083	struct nfs4_get_lease_time_data data = {
9084		.args = &args,
9085		.res = &res,
9086		.clp = clp,
9087	};
9088	struct rpc_message msg = {
9089		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GET_LEASE_TIME],
9090		.rpc_argp = &args,
9091		.rpc_resp = &res,
9092	};
9093	struct rpc_task_setup task_setup = {
9094		.rpc_client = clp->cl_rpcclient,
9095		.rpc_message = &msg,
9096		.callback_ops = &nfs4_get_lease_time_ops,
9097		.callback_data = &data,
9098		.flags = RPC_TASK_TIMEOUT,
9099	};
9100
9101	nfs4_init_sequence(&args.la_seq_args, &res.lr_seq_res, 0, 1);
9102	return nfs4_call_sync_custom(&task_setup);
9103}
9104
9105#ifdef CONFIG_NFS_V4_1
9106
9107/*
9108 * Initialize the values to be used by the client in CREATE_SESSION
9109 * If nfs4_init_session set the fore channel request and response sizes,
9110 * use them.
9111 *
9112 * Set the back channel max_resp_sz_cached to zero to force the client to
9113 * always set csa_cachethis to FALSE because the current implementation
9114 * of the back channel DRC only supports caching the CB_SEQUENCE operation.
9115 */
9116static void nfs4_init_channel_attrs(struct nfs41_create_session_args *args,
9117				    struct rpc_clnt *clnt)
9118{
9119	unsigned int max_rqst_sz, max_resp_sz;
9120	unsigned int max_bc_payload = rpc_max_bc_payload(clnt);
9121	unsigned int max_bc_slots = rpc_num_bc_slots(clnt);
9122
9123	max_rqst_sz = NFS_MAX_FILE_IO_SIZE + nfs41_maxwrite_overhead;
9124	max_resp_sz = NFS_MAX_FILE_IO_SIZE + nfs41_maxread_overhead;
9125
9126	/* Fore channel attributes */
9127	args->fc_attrs.max_rqst_sz = max_rqst_sz;
9128	args->fc_attrs.max_resp_sz = max_resp_sz;
9129	args->fc_attrs.max_ops = NFS4_MAX_OPS;
9130	args->fc_attrs.max_reqs = max_session_slots;
9131
9132	dprintk("%s: Fore Channel : max_rqst_sz=%u max_resp_sz=%u "
9133		"max_ops=%u max_reqs=%u\n",
9134		__func__,
9135		args->fc_attrs.max_rqst_sz, args->fc_attrs.max_resp_sz,
9136		args->fc_attrs.max_ops, args->fc_attrs.max_reqs);
9137
9138	/* Back channel attributes */
9139	args->bc_attrs.max_rqst_sz = max_bc_payload;
9140	args->bc_attrs.max_resp_sz = max_bc_payload;
9141	args->bc_attrs.max_resp_sz_cached = 0;
9142	args->bc_attrs.max_ops = NFS4_MAX_BACK_CHANNEL_OPS;
9143	args->bc_attrs.max_reqs = max_t(unsigned short, max_session_cb_slots, 1);
9144	if (args->bc_attrs.max_reqs > max_bc_slots)
9145		args->bc_attrs.max_reqs = max_bc_slots;
9146
9147	dprintk("%s: Back Channel : max_rqst_sz=%u max_resp_sz=%u "
9148		"max_resp_sz_cached=%u max_ops=%u max_reqs=%u\n",
9149		__func__,
9150		args->bc_attrs.max_rqst_sz, args->bc_attrs.max_resp_sz,
9151		args->bc_attrs.max_resp_sz_cached, args->bc_attrs.max_ops,
9152		args->bc_attrs.max_reqs);
9153}
9154
9155static int nfs4_verify_fore_channel_attrs(struct nfs41_create_session_args *args,
9156		struct nfs41_create_session_res *res)
9157{
9158	struct nfs4_channel_attrs *sent = &args->fc_attrs;
9159	struct nfs4_channel_attrs *rcvd = &res->fc_attrs;
9160
9161	if (rcvd->max_resp_sz > sent->max_resp_sz)
9162		return -EINVAL;
9163	/*
9164	 * Our requested max_ops is the minimum we need; we're not
9165	 * prepared to break up compounds into smaller pieces than that.
9166	 * So, no point even trying to continue if the server won't
9167	 * cooperate:
9168	 */
9169	if (rcvd->max_ops < sent->max_ops)
9170		return -EINVAL;
9171	if (rcvd->max_reqs == 0)
9172		return -EINVAL;
9173	if (rcvd->max_reqs > NFS4_MAX_SLOT_TABLE)
9174		rcvd->max_reqs = NFS4_MAX_SLOT_TABLE;
9175	return 0;
9176}
9177
9178static int nfs4_verify_back_channel_attrs(struct nfs41_create_session_args *args,
9179		struct nfs41_create_session_res *res)
9180{
9181	struct nfs4_channel_attrs *sent = &args->bc_attrs;
9182	struct nfs4_channel_attrs *rcvd = &res->bc_attrs;
9183
9184	if (!(res->flags & SESSION4_BACK_CHAN))
9185		goto out;
9186	if (rcvd->max_rqst_sz > sent->max_rqst_sz)
9187		return -EINVAL;
9188	if (rcvd->max_resp_sz < sent->max_resp_sz)
9189		return -EINVAL;
9190	if (rcvd->max_resp_sz_cached > sent->max_resp_sz_cached)
9191		return -EINVAL;
9192	if (rcvd->max_ops > sent->max_ops)
9193		return -EINVAL;
9194	if (rcvd->max_reqs > sent->max_reqs)
9195		return -EINVAL;
9196out:
9197	return 0;
9198}
9199
9200static int nfs4_verify_channel_attrs(struct nfs41_create_session_args *args,
9201				     struct nfs41_create_session_res *res)
9202{
9203	int ret;
9204
9205	ret = nfs4_verify_fore_channel_attrs(args, res);
9206	if (ret)
9207		return ret;
9208	return nfs4_verify_back_channel_attrs(args, res);
9209}
9210
9211static void nfs4_update_session(struct nfs4_session *session,
9212		struct nfs41_create_session_res *res)
9213{
9214	nfs4_copy_sessionid(&session->sess_id, &res->sessionid);
9215	/* Mark client id and session as being confirmed */
9216	session->clp->cl_exchange_flags |= EXCHGID4_FLAG_CONFIRMED_R;
9217	set_bit(NFS4_SESSION_ESTABLISHED, &session->session_state);
9218	session->flags = res->flags;
9219	memcpy(&session->fc_attrs, &res->fc_attrs, sizeof(session->fc_attrs));
9220	if (res->flags & SESSION4_BACK_CHAN)
9221		memcpy(&session->bc_attrs, &res->bc_attrs,
9222				sizeof(session->bc_attrs));
9223}
9224
9225static int _nfs4_proc_create_session(struct nfs_client *clp,
9226		const struct cred *cred)
9227{
9228	struct nfs4_session *session = clp->cl_session;
9229	struct nfs41_create_session_args args = {
9230		.client = clp,
9231		.clientid = clp->cl_clientid,
9232		.seqid = clp->cl_seqid,
9233		.cb_program = NFS4_CALLBACK,
9234	};
9235	struct nfs41_create_session_res res;
9236
9237	struct rpc_message msg = {
9238		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CREATE_SESSION],
9239		.rpc_argp = &args,
9240		.rpc_resp = &res,
9241		.rpc_cred = cred,
9242	};
9243	int status;
9244
9245	nfs4_init_channel_attrs(&args, clp->cl_rpcclient);
9246	args.flags = (SESSION4_PERSIST | SESSION4_BACK_CHAN);
9247
9248	status = rpc_call_sync(session->clp->cl_rpcclient, &msg,
9249			       RPC_TASK_TIMEOUT | RPC_TASK_NO_ROUND_ROBIN);
9250	trace_nfs4_create_session(clp, status);
9251
9252	switch (status) {
9253	case -NFS4ERR_STALE_CLIENTID:
9254	case -NFS4ERR_DELAY:
9255	case -ETIMEDOUT:
9256	case -EACCES:
9257	case -EAGAIN:
9258		goto out;
9259	}
9260
9261	clp->cl_seqid++;
9262	if (!status) {
9263		/* Verify the session's negotiated channel_attrs values */
9264		status = nfs4_verify_channel_attrs(&args, &res);
9265		/* Increment the clientid slot sequence id */
9266		if (status)
9267			goto out;
9268		nfs4_update_session(session, &res);
9269	}
9270out:
9271	return status;
9272}
9273
9274/*
9275 * Issues a CREATE_SESSION operation to the server.
9276 * It is the responsibility of the caller to verify the session is
9277 * expired before calling this routine.
9278 */
9279int nfs4_proc_create_session(struct nfs_client *clp, const struct cred *cred)
9280{
9281	int status;
9282	unsigned *ptr;
9283	struct nfs4_session *session = clp->cl_session;
9284	struct nfs4_add_xprt_data xprtdata = {
9285		.clp = clp,
9286	};
9287	struct rpc_add_xprt_test rpcdata = {
9288		.add_xprt_test = clp->cl_mvops->session_trunk,
9289		.data = &xprtdata,
9290	};
9291
9292	dprintk("--> %s clp=%p session=%p\n", __func__, clp, session);
9293
9294	status = _nfs4_proc_create_session(clp, cred);
9295	if (status)
9296		goto out;
9297
9298	/* Init or reset the session slot tables */
9299	status = nfs4_setup_session_slot_tables(session);
9300	dprintk("slot table setup returned %d\n", status);
9301	if (status)
9302		goto out;
9303
9304	ptr = (unsigned *)&session->sess_id.data[0];
9305	dprintk("%s client>seqid %d sessionid %u:%u:%u:%u\n", __func__,
9306		clp->cl_seqid, ptr[0], ptr[1], ptr[2], ptr[3]);
9307	rpc_clnt_probe_trunked_xprts(clp->cl_rpcclient, &rpcdata);
9308out:
9309	return status;
9310}
9311
9312/*
9313 * Issue the over-the-wire RPC DESTROY_SESSION.
9314 * The caller must serialize access to this routine.
9315 */
9316int nfs4_proc_destroy_session(struct nfs4_session *session,
9317		const struct cred *cred)
9318{
9319	struct rpc_message msg = {
9320		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DESTROY_SESSION],
9321		.rpc_argp = session,
9322		.rpc_cred = cred,
9323	};
9324	int status = 0;
9325
9326	/* session is still being setup */
9327	if (!test_and_clear_bit(NFS4_SESSION_ESTABLISHED, &session->session_state))
9328		return 0;
9329
9330	status = rpc_call_sync(session->clp->cl_rpcclient, &msg,
9331			       RPC_TASK_TIMEOUT | RPC_TASK_NO_ROUND_ROBIN);
9332	trace_nfs4_destroy_session(session->clp, status);
9333
9334	if (status)
9335		dprintk("NFS: Got error %d from the server on DESTROY_SESSION. "
9336			"Session has been destroyed regardless...\n", status);
9337	rpc_clnt_manage_trunked_xprts(session->clp->cl_rpcclient);
9338	return status;
9339}
9340
9341/*
9342 * Renew the cl_session lease.
9343 */
9344struct nfs4_sequence_data {
9345	struct nfs_client *clp;
9346	struct nfs4_sequence_args args;
9347	struct nfs4_sequence_res res;
9348};
9349
9350static void nfs41_sequence_release(void *data)
9351{
9352	struct nfs4_sequence_data *calldata = data;
9353	struct nfs_client *clp = calldata->clp;
9354
9355	if (refcount_read(&clp->cl_count) > 1)
9356		nfs4_schedule_state_renewal(clp);
9357	nfs_put_client(clp);
9358	kfree(calldata);
9359}
9360
9361static int nfs41_sequence_handle_errors(struct rpc_task *task, struct nfs_client *clp)
9362{
9363	switch(task->tk_status) {
9364	case -NFS4ERR_DELAY:
9365		rpc_delay(task, NFS4_POLL_RETRY_MAX);
9366		return -EAGAIN;
9367	default:
9368		nfs4_schedule_lease_recovery(clp);
9369	}
9370	return 0;
9371}
9372
9373static void nfs41_sequence_call_done(struct rpc_task *task, void *data)
9374{
9375	struct nfs4_sequence_data *calldata = data;
9376	struct nfs_client *clp = calldata->clp;
9377
9378	if (!nfs41_sequence_done(task, task->tk_msg.rpc_resp))
9379		return;
9380
9381	trace_nfs4_sequence(clp, task->tk_status);
9382	if (task->tk_status < 0 && !task->tk_client->cl_shutdown) {
9383		dprintk("%s ERROR %d\n", __func__, task->tk_status);
9384		if (refcount_read(&clp->cl_count) == 1)
9385			return;
9386
9387		if (nfs41_sequence_handle_errors(task, clp) == -EAGAIN) {
9388			rpc_restart_call_prepare(task);
9389			return;
9390		}
9391	}
9392	dprintk("%s rpc_cred %p\n", __func__, task->tk_msg.rpc_cred);
9393}
9394
9395static void nfs41_sequence_prepare(struct rpc_task *task, void *data)
9396{
9397	struct nfs4_sequence_data *calldata = data;
9398	struct nfs_client *clp = calldata->clp;
9399	struct nfs4_sequence_args *args;
9400	struct nfs4_sequence_res *res;
9401
9402	args = task->tk_msg.rpc_argp;
9403	res = task->tk_msg.rpc_resp;
9404
9405	nfs4_setup_sequence(clp, args, res, task);
9406}
9407
9408static const struct rpc_call_ops nfs41_sequence_ops = {
9409	.rpc_call_done = nfs41_sequence_call_done,
9410	.rpc_call_prepare = nfs41_sequence_prepare,
9411	.rpc_release = nfs41_sequence_release,
9412};
9413
9414static struct rpc_task *_nfs41_proc_sequence(struct nfs_client *clp,
9415		const struct cred *cred,
9416		struct nfs4_slot *slot,
9417		bool is_privileged)
9418{
9419	struct nfs4_sequence_data *calldata;
9420	struct rpc_message msg = {
9421		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SEQUENCE],
9422		.rpc_cred = cred,
9423	};
9424	struct rpc_task_setup task_setup_data = {
9425		.rpc_client = clp->cl_rpcclient,
9426		.rpc_message = &msg,
9427		.callback_ops = &nfs41_sequence_ops,
9428		.flags = RPC_TASK_ASYNC | RPC_TASK_TIMEOUT | RPC_TASK_MOVEABLE,
9429	};
9430	struct rpc_task *ret;
9431
9432	ret = ERR_PTR(-EIO);
9433	if (!refcount_inc_not_zero(&clp->cl_count))
9434		goto out_err;
9435
9436	ret = ERR_PTR(-ENOMEM);
9437	calldata = kzalloc(sizeof(*calldata), GFP_KERNEL);
9438	if (calldata == NULL)
9439		goto out_put_clp;
9440	nfs4_init_sequence(&calldata->args, &calldata->res, 0, is_privileged);
9441	nfs4_sequence_attach_slot(&calldata->args, &calldata->res, slot);
9442	msg.rpc_argp = &calldata->args;
9443	msg.rpc_resp = &calldata->res;
9444	calldata->clp = clp;
9445	task_setup_data.callback_data = calldata;
9446
9447	ret = rpc_run_task(&task_setup_data);
9448	if (IS_ERR(ret))
9449		goto out_err;
9450	return ret;
9451out_put_clp:
9452	nfs_put_client(clp);
9453out_err:
9454	nfs41_release_slot(slot);
9455	return ret;
9456}
9457
9458static int nfs41_proc_async_sequence(struct nfs_client *clp, const struct cred *cred, unsigned renew_flags)
9459{
9460	struct rpc_task *task;
9461	int ret = 0;
9462
9463	if ((renew_flags & NFS4_RENEW_TIMEOUT) == 0)
9464		return -EAGAIN;
9465	task = _nfs41_proc_sequence(clp, cred, NULL, false);
9466	if (IS_ERR(task))
9467		ret = PTR_ERR(task);
9468	else
9469		rpc_put_task_async(task);
9470	dprintk("<-- %s status=%d\n", __func__, ret);
9471	return ret;
9472}
9473
9474static int nfs4_proc_sequence(struct nfs_client *clp, const struct cred *cred)
9475{
9476	struct rpc_task *task;
9477	int ret;
9478
9479	task = _nfs41_proc_sequence(clp, cred, NULL, true);
9480	if (IS_ERR(task)) {
9481		ret = PTR_ERR(task);
9482		goto out;
9483	}
9484	ret = rpc_wait_for_completion_task(task);
9485	if (!ret)
9486		ret = task->tk_status;
9487	rpc_put_task(task);
9488out:
9489	dprintk("<-- %s status=%d\n", __func__, ret);
9490	return ret;
9491}
9492
9493struct nfs4_reclaim_complete_data {
9494	struct nfs_client *clp;
9495	struct nfs41_reclaim_complete_args arg;
9496	struct nfs41_reclaim_complete_res res;
9497};
9498
9499static void nfs4_reclaim_complete_prepare(struct rpc_task *task, void *data)
9500{
9501	struct nfs4_reclaim_complete_data *calldata = data;
9502
9503	nfs4_setup_sequence(calldata->clp,
9504			&calldata->arg.seq_args,
9505			&calldata->res.seq_res,
9506			task);
9507}
9508
9509static int nfs41_reclaim_complete_handle_errors(struct rpc_task *task, struct nfs_client *clp)
9510{
9511	switch(task->tk_status) {
9512	case 0:
9513		wake_up_all(&clp->cl_lock_waitq);
9514		fallthrough;
9515	case -NFS4ERR_COMPLETE_ALREADY:
9516	case -NFS4ERR_WRONG_CRED: /* What to do here? */
9517		break;
9518	case -NFS4ERR_DELAY:
9519		rpc_delay(task, NFS4_POLL_RETRY_MAX);
9520		fallthrough;
9521	case -NFS4ERR_RETRY_UNCACHED_REP:
9522	case -EACCES:
9523		dprintk("%s: failed to reclaim complete error %d for server %s, retrying\n",
9524			__func__, task->tk_status, clp->cl_hostname);
9525		return -EAGAIN;
9526	case -NFS4ERR_BADSESSION:
9527	case -NFS4ERR_DEADSESSION:
9528	case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
9529		break;
9530	default:
9531		nfs4_schedule_lease_recovery(clp);
9532	}
9533	return 0;
9534}
9535
9536static void nfs4_reclaim_complete_done(struct rpc_task *task, void *data)
9537{
9538	struct nfs4_reclaim_complete_data *calldata = data;
9539	struct nfs_client *clp = calldata->clp;
9540	struct nfs4_sequence_res *res = &calldata->res.seq_res;
9541
9542	if (!nfs41_sequence_done(task, res))
9543		return;
9544
9545	trace_nfs4_reclaim_complete(clp, task->tk_status);
9546	if (nfs41_reclaim_complete_handle_errors(task, clp) == -EAGAIN) {
9547		rpc_restart_call_prepare(task);
9548		return;
9549	}
9550}
9551
9552static void nfs4_free_reclaim_complete_data(void *data)
9553{
9554	struct nfs4_reclaim_complete_data *calldata = data;
9555
9556	kfree(calldata);
9557}
9558
9559static const struct rpc_call_ops nfs4_reclaim_complete_call_ops = {
9560	.rpc_call_prepare = nfs4_reclaim_complete_prepare,
9561	.rpc_call_done = nfs4_reclaim_complete_done,
9562	.rpc_release = nfs4_free_reclaim_complete_data,
9563};
9564
9565/*
9566 * Issue a global reclaim complete.
9567 */
9568static int nfs41_proc_reclaim_complete(struct nfs_client *clp,
9569		const struct cred *cred)
9570{
9571	struct nfs4_reclaim_complete_data *calldata;
9572	struct rpc_message msg = {
9573		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RECLAIM_COMPLETE],
9574		.rpc_cred = cred,
9575	};
9576	struct rpc_task_setup task_setup_data = {
9577		.rpc_client = clp->cl_rpcclient,
9578		.rpc_message = &msg,
9579		.callback_ops = &nfs4_reclaim_complete_call_ops,
9580		.flags = RPC_TASK_NO_ROUND_ROBIN,
9581	};
9582	int status = -ENOMEM;
9583
9584	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
9585	if (calldata == NULL)
9586		goto out;
9587	calldata->clp = clp;
9588	calldata->arg.one_fs = 0;
9589
9590	nfs4_init_sequence(&calldata->arg.seq_args, &calldata->res.seq_res, 0, 1);
9591	msg.rpc_argp = &calldata->arg;
9592	msg.rpc_resp = &calldata->res;
9593	task_setup_data.callback_data = calldata;
9594	status = nfs4_call_sync_custom(&task_setup_data);
9595out:
9596	dprintk("<-- %s status=%d\n", __func__, status);
9597	return status;
9598}
9599
9600static void
9601nfs4_layoutget_prepare(struct rpc_task *task, void *calldata)
9602{
9603	struct nfs4_layoutget *lgp = calldata;
9604	struct nfs_server *server = NFS_SERVER(lgp->args.inode);
9605
9606	nfs4_setup_sequence(server->nfs_client, &lgp->args.seq_args,
9607				&lgp->res.seq_res, task);
9608}
9609
9610static void nfs4_layoutget_done(struct rpc_task *task, void *calldata)
9611{
9612	struct nfs4_layoutget *lgp = calldata;
9613
9614	nfs41_sequence_process(task, &lgp->res.seq_res);
9615}
9616
9617static int
9618nfs4_layoutget_handle_exception(struct rpc_task *task,
9619		struct nfs4_layoutget *lgp, struct nfs4_exception *exception)
9620{
9621	struct inode *inode = lgp->args.inode;
9622	struct nfs_server *server = NFS_SERVER(inode);
9623	struct pnfs_layout_hdr *lo = lgp->lo;
9624	int nfs4err = task->tk_status;
9625	int err, status = 0;
9626	LIST_HEAD(head);
9627
9628	dprintk("--> %s tk_status => %d\n", __func__, -task->tk_status);
9629
9630	nfs4_sequence_free_slot(&lgp->res.seq_res);
9631
9632	switch (nfs4err) {
9633	case 0:
9634		goto out;
9635
9636	/*
9637	 * NFS4ERR_LAYOUTUNAVAILABLE means we are not supposed to use pnfs
9638	 * on the file. set tk_status to -ENODATA to tell upper layer to
9639	 * retry go inband.
9640	 */
9641	case -NFS4ERR_LAYOUTUNAVAILABLE:
9642		status = -ENODATA;
9643		goto out;
9644	/*
9645	 * NFS4ERR_BADLAYOUT means the MDS cannot return a layout of
9646	 * length lgp->args.minlength != 0 (see RFC5661 section 18.43.3).
9647	 */
9648	case -NFS4ERR_BADLAYOUT:
9649		status = -EOVERFLOW;
9650		goto out;
9651	/*
9652	 * NFS4ERR_LAYOUTTRYLATER is a conflict with another client
9653	 * (or clients) writing to the same RAID stripe except when
9654	 * the minlength argument is 0 (see RFC5661 section 18.43.3).
9655	 *
9656	 * Treat it like we would RECALLCONFLICT -- we retry for a little
9657	 * while, and then eventually give up.
9658	 */
9659	case -NFS4ERR_LAYOUTTRYLATER:
9660		if (lgp->args.minlength == 0) {
9661			status = -EOVERFLOW;
9662			goto out;
9663		}
9664		status = -EBUSY;
9665		break;
9666	case -NFS4ERR_RECALLCONFLICT:
9667	case -NFS4ERR_RETURNCONFLICT:
9668		status = -ERECALLCONFLICT;
9669		break;
9670	case -NFS4ERR_DELEG_REVOKED:
9671	case -NFS4ERR_ADMIN_REVOKED:
9672	case -NFS4ERR_EXPIRED:
9673	case -NFS4ERR_BAD_STATEID:
9674		exception->timeout = 0;
9675		spin_lock(&inode->i_lock);
9676		/* If the open stateid was bad, then recover it. */
9677		if (!lo || test_bit(NFS_LAYOUT_INVALID_STID, &lo->plh_flags) ||
9678		    !nfs4_stateid_match_other(&lgp->args.stateid, &lo->plh_stateid)) {
9679			spin_unlock(&inode->i_lock);
9680			exception->state = lgp->args.ctx->state;
9681			exception->stateid = &lgp->args.stateid;
9682			break;
9683		}
9684
9685		/*
9686		 * Mark the bad layout state as invalid, then retry
9687		 */
9688		pnfs_mark_layout_stateid_invalid(lo, &head);
9689		spin_unlock(&inode->i_lock);
9690		nfs_commit_inode(inode, 0);
9691		pnfs_free_lseg_list(&head);
9692		status = -EAGAIN;
9693		goto out;
9694	}
9695
9696	err = nfs4_handle_exception(server, nfs4err, exception);
9697	if (!status) {
9698		if (exception->retry)
9699			status = -EAGAIN;
9700		else
9701			status = err;
9702	}
9703out:
9704	return status;
9705}
9706
9707size_t max_response_pages(struct nfs_server *server)
9708{
9709	u32 max_resp_sz = server->nfs_client->cl_session->fc_attrs.max_resp_sz;
9710	return nfs_page_array_len(0, max_resp_sz);
9711}
9712
9713static void nfs4_layoutget_release(void *calldata)
9714{
9715	struct nfs4_layoutget *lgp = calldata;
9716
9717	nfs4_sequence_free_slot(&lgp->res.seq_res);
9718	pnfs_layoutget_free(lgp);
9719}
9720
9721static const struct rpc_call_ops nfs4_layoutget_call_ops = {
9722	.rpc_call_prepare = nfs4_layoutget_prepare,
9723	.rpc_call_done = nfs4_layoutget_done,
9724	.rpc_release = nfs4_layoutget_release,
9725};
9726
9727struct pnfs_layout_segment *
9728nfs4_proc_layoutget(struct nfs4_layoutget *lgp, long *timeout)
9729{
9730	struct inode *inode = lgp->args.inode;
9731	struct nfs_server *server = NFS_SERVER(inode);
9732	struct rpc_task *task;
9733	struct rpc_message msg = {
9734		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LAYOUTGET],
9735		.rpc_argp = &lgp->args,
9736		.rpc_resp = &lgp->res,
9737		.rpc_cred = lgp->cred,
9738	};
9739	struct rpc_task_setup task_setup_data = {
9740		.rpc_client = server->client,
9741		.rpc_message = &msg,
9742		.callback_ops = &nfs4_layoutget_call_ops,
9743		.callback_data = lgp,
9744		.flags = RPC_TASK_ASYNC | RPC_TASK_CRED_NOREF |
9745			 RPC_TASK_MOVEABLE,
9746	};
9747	struct pnfs_layout_segment *lseg = NULL;
9748	struct nfs4_exception exception = {
9749		.inode = inode,
9750		.timeout = *timeout,
9751	};
9752	int status = 0;
9753
9754	nfs4_init_sequence(&lgp->args.seq_args, &lgp->res.seq_res, 0, 0);
9755
9756	task = rpc_run_task(&task_setup_data);
9757	if (IS_ERR(task))
9758		return ERR_CAST(task);
9759
9760	status = rpc_wait_for_completion_task(task);
9761	if (status != 0)
9762		goto out;
9763
9764	if (task->tk_status < 0) {
9765		status = nfs4_layoutget_handle_exception(task, lgp, &exception);
9766		*timeout = exception.timeout;
9767	} else if (lgp->res.layoutp->len == 0) {
9768		status = -EAGAIN;
9769		*timeout = nfs4_update_delay(&exception.timeout);
9770	} else
9771		lseg = pnfs_layout_process(lgp);
9772out:
9773	trace_nfs4_layoutget(lgp->args.ctx,
9774			&lgp->args.range,
9775			&lgp->res.range,
9776			&lgp->res.stateid,
9777			status);
9778
9779	rpc_put_task(task);
9780	dprintk("<-- %s status=%d\n", __func__, status);
9781	if (status)
9782		return ERR_PTR(status);
9783	return lseg;
9784}
9785
9786static void
9787nfs4_layoutreturn_prepare(struct rpc_task *task, void *calldata)
9788{
9789	struct nfs4_layoutreturn *lrp = calldata;
9790
9791	nfs4_setup_sequence(lrp->clp,
9792			&lrp->args.seq_args,
9793			&lrp->res.seq_res,
9794			task);
9795	if (!pnfs_layout_is_valid(lrp->args.layout))
9796		rpc_exit(task, 0);
9797}
9798
9799static void nfs4_layoutreturn_done(struct rpc_task *task, void *calldata)
9800{
9801	struct nfs4_layoutreturn *lrp = calldata;
9802	struct nfs_server *server;
9803
9804	if (!nfs41_sequence_process(task, &lrp->res.seq_res))
9805		return;
9806
9807	/*
9808	 * Was there an RPC level error? Assume the call succeeded,
9809	 * and that we need to release the layout
9810	 */
9811	if (task->tk_rpc_status != 0 && RPC_WAS_SENT(task)) {
9812		lrp->res.lrs_present = 0;
9813		return;
9814	}
9815
9816	server = NFS_SERVER(lrp->args.inode);
9817	switch (task->tk_status) {
9818	case -NFS4ERR_OLD_STATEID:
9819		if (nfs4_layout_refresh_old_stateid(&lrp->args.stateid,
9820					&lrp->args.range,
9821					lrp->args.inode))
9822			goto out_restart;
9823		fallthrough;
9824	default:
9825		task->tk_status = 0;
9826		fallthrough;
9827	case 0:
9828		break;
9829	case -NFS4ERR_DELAY:
9830		if (nfs4_async_handle_error(task, server, NULL, NULL) != -EAGAIN)
9831			break;
9832		goto out_restart;
9833	}
9834	return;
9835out_restart:
9836	task->tk_status = 0;
9837	nfs4_sequence_free_slot(&lrp->res.seq_res);
9838	rpc_restart_call_prepare(task);
9839}
9840
9841static void nfs4_layoutreturn_release(void *calldata)
9842{
9843	struct nfs4_layoutreturn *lrp = calldata;
9844	struct pnfs_layout_hdr *lo = lrp->args.layout;
9845
9846	pnfs_layoutreturn_free_lsegs(lo, &lrp->args.stateid, &lrp->args.range,
9847			lrp->res.lrs_present ? &lrp->res.stateid : NULL);
9848	nfs4_sequence_free_slot(&lrp->res.seq_res);
9849	if (lrp->ld_private.ops && lrp->ld_private.ops->free)
9850		lrp->ld_private.ops->free(&lrp->ld_private);
9851	pnfs_put_layout_hdr(lrp->args.layout);
9852	nfs_iput_and_deactive(lrp->inode);
9853	put_cred(lrp->cred);
9854	kfree(calldata);
9855}
9856
9857static const struct rpc_call_ops nfs4_layoutreturn_call_ops = {
9858	.rpc_call_prepare = nfs4_layoutreturn_prepare,
9859	.rpc_call_done = nfs4_layoutreturn_done,
9860	.rpc_release = nfs4_layoutreturn_release,
9861};
9862
9863int nfs4_proc_layoutreturn(struct nfs4_layoutreturn *lrp, bool sync)
9864{
9865	struct rpc_task *task;
9866	struct rpc_message msg = {
9867		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LAYOUTRETURN],
9868		.rpc_argp = &lrp->args,
9869		.rpc_resp = &lrp->res,
9870		.rpc_cred = lrp->cred,
9871	};
9872	struct rpc_task_setup task_setup_data = {
9873		.rpc_client = NFS_SERVER(lrp->args.inode)->client,
9874		.rpc_message = &msg,
9875		.callback_ops = &nfs4_layoutreturn_call_ops,
9876		.callback_data = lrp,
9877		.flags = RPC_TASK_MOVEABLE,
9878	};
9879	int status = 0;
9880
9881	nfs4_state_protect(NFS_SERVER(lrp->args.inode)->nfs_client,
9882			NFS_SP4_MACH_CRED_PNFS_CLEANUP,
9883			&task_setup_data.rpc_client, &msg);
9884
9885	lrp->inode = nfs_igrab_and_active(lrp->args.inode);
9886	if (!sync) {
9887		if (!lrp->inode) {
9888			nfs4_layoutreturn_release(lrp);
9889			return -EAGAIN;
9890		}
9891		task_setup_data.flags |= RPC_TASK_ASYNC;
9892	}
9893	if (!lrp->inode)
9894		nfs4_init_sequence(&lrp->args.seq_args, &lrp->res.seq_res, 1,
9895				   1);
9896	else
9897		nfs4_init_sequence(&lrp->args.seq_args, &lrp->res.seq_res, 1,
9898				   0);
9899	task = rpc_run_task(&task_setup_data);
9900	if (IS_ERR(task))
9901		return PTR_ERR(task);
9902	if (sync)
9903		status = task->tk_status;
9904	trace_nfs4_layoutreturn(lrp->args.inode, &lrp->args.stateid, status);
9905	dprintk("<-- %s status=%d\n", __func__, status);
9906	rpc_put_task(task);
9907	return status;
9908}
9909
9910static int
9911_nfs4_proc_getdeviceinfo(struct nfs_server *server,
9912		struct pnfs_device *pdev,
9913		const struct cred *cred)
9914{
9915	struct nfs4_getdeviceinfo_args args = {
9916		.pdev = pdev,
9917		.notify_types = NOTIFY_DEVICEID4_CHANGE |
9918			NOTIFY_DEVICEID4_DELETE,
9919	};
9920	struct nfs4_getdeviceinfo_res res = {
9921		.pdev = pdev,
9922	};
9923	struct rpc_message msg = {
9924		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETDEVICEINFO],
9925		.rpc_argp = &args,
9926		.rpc_resp = &res,
9927		.rpc_cred = cred,
9928	};
9929	int status;
9930
9931	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
9932	if (res.notification & ~args.notify_types)
9933		dprintk("%s: unsupported notification\n", __func__);
9934	if (res.notification != args.notify_types)
9935		pdev->nocache = 1;
9936
9937	trace_nfs4_getdeviceinfo(server, &pdev->dev_id, status);
9938
9939	dprintk("<-- %s status=%d\n", __func__, status);
9940
9941	return status;
9942}
9943
9944int nfs4_proc_getdeviceinfo(struct nfs_server *server,
9945		struct pnfs_device *pdev,
9946		const struct cred *cred)
9947{
9948	struct nfs4_exception exception = { };
9949	int err;
9950
9951	do {
9952		err = nfs4_handle_exception(server,
9953					_nfs4_proc_getdeviceinfo(server, pdev, cred),
9954					&exception);
9955	} while (exception.retry);
9956	return err;
9957}
9958EXPORT_SYMBOL_GPL(nfs4_proc_getdeviceinfo);
9959
9960static void nfs4_layoutcommit_prepare(struct rpc_task *task, void *calldata)
9961{
9962	struct nfs4_layoutcommit_data *data = calldata;
9963	struct nfs_server *server = NFS_SERVER(data->args.inode);
9964
9965	nfs4_setup_sequence(server->nfs_client,
9966			&data->args.seq_args,
9967			&data->res.seq_res,
9968			task);
9969}
9970
9971static void
9972nfs4_layoutcommit_done(struct rpc_task *task, void *calldata)
9973{
9974	struct nfs4_layoutcommit_data *data = calldata;
9975	struct nfs_server *server = NFS_SERVER(data->args.inode);
9976
9977	if (!nfs41_sequence_done(task, &data->res.seq_res))
9978		return;
9979
9980	switch (task->tk_status) { /* Just ignore these failures */
9981	case -NFS4ERR_DELEG_REVOKED: /* layout was recalled */
9982	case -NFS4ERR_BADIOMODE:     /* no IOMODE_RW layout for range */
9983	case -NFS4ERR_BADLAYOUT:     /* no layout */
9984	case -NFS4ERR_GRACE:	    /* loca_recalim always false */
9985		task->tk_status = 0;
9986		break;
9987	case 0:
9988		break;
9989	default:
9990		if (nfs4_async_handle_error(task, server, NULL, NULL) == -EAGAIN) {
9991			rpc_restart_call_prepare(task);
9992			return;
9993		}
9994	}
9995}
9996
9997static void nfs4_layoutcommit_release(void *calldata)
9998{
9999	struct nfs4_layoutcommit_data *data = calldata;
10000
10001	pnfs_cleanup_layoutcommit(data);
10002	nfs_post_op_update_inode_force_wcc(data->args.inode,
10003					   data->res.fattr);
10004	put_cred(data->cred);
10005	nfs_iput_and_deactive(data->inode);
10006	kfree(data);
10007}
10008
10009static const struct rpc_call_ops nfs4_layoutcommit_ops = {
10010	.rpc_call_prepare = nfs4_layoutcommit_prepare,
10011	.rpc_call_done = nfs4_layoutcommit_done,
10012	.rpc_release = nfs4_layoutcommit_release,
10013};
10014
10015int
10016nfs4_proc_layoutcommit(struct nfs4_layoutcommit_data *data, bool sync)
10017{
10018	struct rpc_message msg = {
10019		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LAYOUTCOMMIT],
10020		.rpc_argp = &data->args,
10021		.rpc_resp = &data->res,
10022		.rpc_cred = data->cred,
10023	};
10024	struct rpc_task_setup task_setup_data = {
10025		.task = &data->task,
10026		.rpc_client = NFS_CLIENT(data->args.inode),
10027		.rpc_message = &msg,
10028		.callback_ops = &nfs4_layoutcommit_ops,
10029		.callback_data = data,
10030		.flags = RPC_TASK_MOVEABLE,
10031	};
10032	struct rpc_task *task;
10033	int status = 0;
10034
10035	dprintk("NFS: initiating layoutcommit call. sync %d "
10036		"lbw: %llu inode %lu\n", sync,
10037		data->args.lastbytewritten,
10038		data->args.inode->i_ino);
10039
10040	if (!sync) {
10041		data->inode = nfs_igrab_and_active(data->args.inode);
10042		if (data->inode == NULL) {
10043			nfs4_layoutcommit_release(data);
10044			return -EAGAIN;
10045		}
10046		task_setup_data.flags = RPC_TASK_ASYNC;
10047	}
10048	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1, 0);
10049	task = rpc_run_task(&task_setup_data);
10050	if (IS_ERR(task))
10051		return PTR_ERR(task);
10052	if (sync)
10053		status = task->tk_status;
10054	trace_nfs4_layoutcommit(data->args.inode, &data->args.stateid, status);
10055	dprintk("%s: status %d\n", __func__, status);
10056	rpc_put_task(task);
10057	return status;
10058}
10059
10060/*
10061 * Use the state managment nfs_client cl_rpcclient, which uses krb5i (if
10062 * possible) as per RFC3530bis and RFC5661 Security Considerations sections
10063 */
10064static int
10065_nfs41_proc_secinfo_no_name(struct nfs_server *server, struct nfs_fh *fhandle,
10066		    struct nfs_fsinfo *info,
10067		    struct nfs4_secinfo_flavors *flavors, bool use_integrity)
10068{
10069	struct nfs41_secinfo_no_name_args args = {
10070		.style = SECINFO_STYLE_CURRENT_FH,
10071	};
10072	struct nfs4_secinfo_res res = {
10073		.flavors = flavors,
10074	};
10075	struct rpc_message msg = {
10076		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SECINFO_NO_NAME],
10077		.rpc_argp = &args,
10078		.rpc_resp = &res,
10079	};
10080	struct nfs4_call_sync_data data = {
10081		.seq_server = server,
10082		.seq_args = &args.seq_args,
10083		.seq_res = &res.seq_res,
10084	};
10085	struct rpc_task_setup task_setup = {
10086		.rpc_client = server->client,
10087		.rpc_message = &msg,
10088		.callback_ops = server->nfs_client->cl_mvops->call_sync_ops,
10089		.callback_data = &data,
10090		.flags = RPC_TASK_NO_ROUND_ROBIN,
10091	};
10092	const struct cred *cred = NULL;
10093	int status;
10094
10095	if (use_integrity) {
10096		task_setup.rpc_client = server->nfs_client->cl_rpcclient;
10097
10098		cred = nfs4_get_clid_cred(server->nfs_client);
10099		msg.rpc_cred = cred;
10100	}
10101
10102	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0, 0);
10103	status = nfs4_call_sync_custom(&task_setup);
10104	dprintk("<-- %s status=%d\n", __func__, status);
10105
10106	put_cred(cred);
10107
10108	return status;
10109}
10110
10111static int
10112nfs41_proc_secinfo_no_name(struct nfs_server *server, struct nfs_fh *fhandle,
10113			   struct nfs_fsinfo *info, struct nfs4_secinfo_flavors *flavors)
10114{
10115	struct nfs4_exception exception = {
10116		.interruptible = true,
10117	};
10118	int err;
10119	do {
10120		/* first try using integrity protection */
10121		err = -NFS4ERR_WRONGSEC;
10122
10123		/* try to use integrity protection with machine cred */
10124		if (_nfs4_is_integrity_protected(server->nfs_client))
10125			err = _nfs41_proc_secinfo_no_name(server, fhandle, info,
10126							  flavors, true);
10127
10128		/*
10129		 * if unable to use integrity protection, or SECINFO with
10130		 * integrity protection returns NFS4ERR_WRONGSEC (which is
10131		 * disallowed by spec, but exists in deployed servers) use
10132		 * the current filesystem's rpc_client and the user cred.
10133		 */
10134		if (err == -NFS4ERR_WRONGSEC)
10135			err = _nfs41_proc_secinfo_no_name(server, fhandle, info,
10136							  flavors, false);
10137
10138		switch (err) {
10139		case 0:
10140		case -NFS4ERR_WRONGSEC:
10141		case -ENOTSUPP:
10142			goto out;
10143		default:
10144			err = nfs4_handle_exception(server, err, &exception);
10145		}
10146	} while (exception.retry);
10147out:
10148	return err;
10149}
10150
10151static int
10152nfs41_find_root_sec(struct nfs_server *server, struct nfs_fh *fhandle,
10153		    struct nfs_fsinfo *info)
10154{
10155	int err;
10156	struct page *page;
10157	rpc_authflavor_t flavor = RPC_AUTH_MAXFLAVOR;
10158	struct nfs4_secinfo_flavors *flavors;
10159	struct nfs4_secinfo4 *secinfo;
10160	int i;
10161
10162	page = alloc_page(GFP_KERNEL);
10163	if (!page) {
10164		err = -ENOMEM;
10165		goto out;
10166	}
10167
10168	flavors = page_address(page);
10169	err = nfs41_proc_secinfo_no_name(server, fhandle, info, flavors);
10170
10171	/*
10172	 * Fall back on "guess and check" method if
10173	 * the server doesn't support SECINFO_NO_NAME
10174	 */
10175	if (err == -NFS4ERR_WRONGSEC || err == -ENOTSUPP) {
10176		err = nfs4_find_root_sec(server, fhandle, info);
10177		goto out_freepage;
10178	}
10179	if (err)
10180		goto out_freepage;
10181
10182	for (i = 0; i < flavors->num_flavors; i++) {
10183		secinfo = &flavors->flavors[i];
10184
10185		switch (secinfo->flavor) {
10186		case RPC_AUTH_NULL:
10187		case RPC_AUTH_UNIX:
10188		case RPC_AUTH_GSS:
10189			flavor = rpcauth_get_pseudoflavor(secinfo->flavor,
10190					&secinfo->flavor_info);
10191			break;
10192		default:
10193			flavor = RPC_AUTH_MAXFLAVOR;
10194			break;
10195		}
10196
10197		if (!nfs_auth_info_match(&server->auth_info, flavor))
10198			flavor = RPC_AUTH_MAXFLAVOR;
10199
10200		if (flavor != RPC_AUTH_MAXFLAVOR) {
10201			err = nfs4_lookup_root_sec(server, fhandle,
10202						   info, flavor);
10203			if (!err)
10204				break;
10205		}
10206	}
10207
10208	if (flavor == RPC_AUTH_MAXFLAVOR)
10209		err = -EPERM;
10210
10211out_freepage:
10212	put_page(page);
10213	if (err == -EACCES)
10214		return -EPERM;
10215out:
10216	return err;
10217}
10218
10219static int _nfs41_test_stateid(struct nfs_server *server,
10220		nfs4_stateid *stateid,
10221		const struct cred *cred)
10222{
10223	int status;
10224	struct nfs41_test_stateid_args args = {
10225		.stateid = stateid,
10226	};
10227	struct nfs41_test_stateid_res res;
10228	struct rpc_message msg = {
10229		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_TEST_STATEID],
10230		.rpc_argp = &args,
10231		.rpc_resp = &res,
10232		.rpc_cred = cred,
10233	};
10234	struct rpc_clnt *rpc_client = server->client;
10235
10236	nfs4_state_protect(server->nfs_client, NFS_SP4_MACH_CRED_STATEID,
10237		&rpc_client, &msg);
10238
10239	dprintk("NFS call  test_stateid %p\n", stateid);
10240	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0, 1);
10241	status = nfs4_call_sync_sequence(rpc_client, server, &msg,
10242			&args.seq_args, &res.seq_res);
10243	if (status != NFS_OK) {
10244		dprintk("NFS reply test_stateid: failed, %d\n", status);
10245		return status;
10246	}
10247	dprintk("NFS reply test_stateid: succeeded, %d\n", -res.status);
10248	return -res.status;
10249}
10250
10251static void nfs4_handle_delay_or_session_error(struct nfs_server *server,
10252		int err, struct nfs4_exception *exception)
10253{
10254	exception->retry = 0;
10255	switch(err) {
10256	case -NFS4ERR_DELAY:
10257	case -NFS4ERR_RETRY_UNCACHED_REP:
10258		nfs4_handle_exception(server, err, exception);
10259		break;
10260	case -NFS4ERR_BADSESSION:
10261	case -NFS4ERR_BADSLOT:
10262	case -NFS4ERR_BAD_HIGH_SLOT:
10263	case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
10264	case -NFS4ERR_DEADSESSION:
10265		nfs4_do_handle_exception(server, err, exception);
10266	}
10267}
10268
10269/**
10270 * nfs41_test_stateid - perform a TEST_STATEID operation
10271 *
10272 * @server: server / transport on which to perform the operation
10273 * @stateid: state ID to test
10274 * @cred: credential
10275 *
10276 * Returns NFS_OK if the server recognizes that "stateid" is valid.
10277 * Otherwise a negative NFS4ERR value is returned if the operation
10278 * failed or the state ID is not currently valid.
10279 */
10280static int nfs41_test_stateid(struct nfs_server *server,
10281		nfs4_stateid *stateid,
10282		const struct cred *cred)
10283{
10284	struct nfs4_exception exception = {
10285		.interruptible = true,
10286	};
10287	int err;
10288	do {
10289		err = _nfs41_test_stateid(server, stateid, cred);
10290		nfs4_handle_delay_or_session_error(server, err, &exception);
10291	} while (exception.retry);
10292	return err;
10293}
10294
10295struct nfs_free_stateid_data {
10296	struct nfs_server *server;
10297	struct nfs41_free_stateid_args args;
10298	struct nfs41_free_stateid_res res;
10299};
10300
10301static void nfs41_free_stateid_prepare(struct rpc_task *task, void *calldata)
10302{
10303	struct nfs_free_stateid_data *data = calldata;
10304	nfs4_setup_sequence(data->server->nfs_client,
10305			&data->args.seq_args,
10306			&data->res.seq_res,
10307			task);
10308}
10309
10310static void nfs41_free_stateid_done(struct rpc_task *task, void *calldata)
10311{
10312	struct nfs_free_stateid_data *data = calldata;
10313
10314	nfs41_sequence_done(task, &data->res.seq_res);
10315
10316	switch (task->tk_status) {
10317	case -NFS4ERR_DELAY:
10318		if (nfs4_async_handle_error(task, data->server, NULL, NULL) == -EAGAIN)
10319			rpc_restart_call_prepare(task);
10320	}
10321}
10322
10323static void nfs41_free_stateid_release(void *calldata)
10324{
10325	struct nfs_free_stateid_data *data = calldata;
10326	struct nfs_client *clp = data->server->nfs_client;
10327
10328	nfs_put_client(clp);
10329	kfree(calldata);
10330}
10331
10332static const struct rpc_call_ops nfs41_free_stateid_ops = {
10333	.rpc_call_prepare = nfs41_free_stateid_prepare,
10334	.rpc_call_done = nfs41_free_stateid_done,
10335	.rpc_release = nfs41_free_stateid_release,
10336};
10337
10338/**
10339 * nfs41_free_stateid - perform a FREE_STATEID operation
10340 *
10341 * @server: server / transport on which to perform the operation
10342 * @stateid: state ID to release
10343 * @cred: credential
10344 * @privileged: set to true if this call needs to be privileged
10345 *
10346 * Note: this function is always asynchronous.
10347 */
10348static int nfs41_free_stateid(struct nfs_server *server,
10349		const nfs4_stateid *stateid,
10350		const struct cred *cred,
10351		bool privileged)
10352{
10353	struct rpc_message msg = {
10354		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FREE_STATEID],
10355		.rpc_cred = cred,
10356	};
10357	struct rpc_task_setup task_setup = {
10358		.rpc_client = server->client,
10359		.rpc_message = &msg,
10360		.callback_ops = &nfs41_free_stateid_ops,
10361		.flags = RPC_TASK_ASYNC | RPC_TASK_MOVEABLE,
10362	};
10363	struct nfs_free_stateid_data *data;
10364	struct rpc_task *task;
10365	struct nfs_client *clp = server->nfs_client;
10366
10367	if (!refcount_inc_not_zero(&clp->cl_count))
10368		return -EIO;
10369
10370	nfs4_state_protect(server->nfs_client, NFS_SP4_MACH_CRED_STATEID,
10371		&task_setup.rpc_client, &msg);
10372
10373	dprintk("NFS call  free_stateid %p\n", stateid);
10374	data = kmalloc(sizeof(*data), GFP_KERNEL);
10375	if (!data)
10376		return -ENOMEM;
10377	data->server = server;
10378	nfs4_stateid_copy(&data->args.stateid, stateid);
10379
10380	task_setup.callback_data = data;
10381
10382	msg.rpc_argp = &data->args;
10383	msg.rpc_resp = &data->res;
10384	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1, privileged);
10385	task = rpc_run_task(&task_setup);
10386	if (IS_ERR(task))
10387		return PTR_ERR(task);
10388	rpc_put_task(task);
10389	return 0;
10390}
10391
10392static void
10393nfs41_free_lock_state(struct nfs_server *server, struct nfs4_lock_state *lsp)
10394{
10395	const struct cred *cred = lsp->ls_state->owner->so_cred;
10396
10397	nfs41_free_stateid(server, &lsp->ls_stateid, cred, false);
10398	nfs4_free_lock_state(server, lsp);
10399}
10400
10401static bool nfs41_match_stateid(const nfs4_stateid *s1,
10402		const nfs4_stateid *s2)
10403{
10404	if (s1->type != s2->type)
10405		return false;
10406
10407	if (memcmp(s1->other, s2->other, sizeof(s1->other)) != 0)
10408		return false;
10409
10410	if (s1->seqid == s2->seqid)
10411		return true;
10412
10413	return s1->seqid == 0 || s2->seqid == 0;
10414}
10415
10416#endif /* CONFIG_NFS_V4_1 */
10417
10418static bool nfs4_match_stateid(const nfs4_stateid *s1,
10419		const nfs4_stateid *s2)
10420{
10421	return nfs4_stateid_match(s1, s2);
10422}
10423
10424
10425static const struct nfs4_state_recovery_ops nfs40_reboot_recovery_ops = {
10426	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
10427	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
10428	.recover_open	= nfs4_open_reclaim,
10429	.recover_lock	= nfs4_lock_reclaim,
10430	.establish_clid = nfs4_init_clientid,
10431	.detect_trunking = nfs40_discover_server_trunking,
10432};
10433
10434#if defined(CONFIG_NFS_V4_1)
10435static const struct nfs4_state_recovery_ops nfs41_reboot_recovery_ops = {
10436	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
10437	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
10438	.recover_open	= nfs4_open_reclaim,
10439	.recover_lock	= nfs4_lock_reclaim,
10440	.establish_clid = nfs41_init_clientid,
10441	.reclaim_complete = nfs41_proc_reclaim_complete,
10442	.detect_trunking = nfs41_discover_server_trunking,
10443};
10444#endif /* CONFIG_NFS_V4_1 */
10445
10446static const struct nfs4_state_recovery_ops nfs40_nograce_recovery_ops = {
10447	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
10448	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
10449	.recover_open	= nfs40_open_expired,
10450	.recover_lock	= nfs4_lock_expired,
10451	.establish_clid = nfs4_init_clientid,
10452};
10453
10454#if defined(CONFIG_NFS_V4_1)
10455static const struct nfs4_state_recovery_ops nfs41_nograce_recovery_ops = {
10456	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
10457	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
10458	.recover_open	= nfs41_open_expired,
10459	.recover_lock	= nfs41_lock_expired,
10460	.establish_clid = nfs41_init_clientid,
10461};
10462#endif /* CONFIG_NFS_V4_1 */
10463
10464static const struct nfs4_state_maintenance_ops nfs40_state_renewal_ops = {
10465	.sched_state_renewal = nfs4_proc_async_renew,
10466	.get_state_renewal_cred = nfs4_get_renew_cred,
10467	.renew_lease = nfs4_proc_renew,
10468};
10469
10470#if defined(CONFIG_NFS_V4_1)
10471static const struct nfs4_state_maintenance_ops nfs41_state_renewal_ops = {
10472	.sched_state_renewal = nfs41_proc_async_sequence,
10473	.get_state_renewal_cred = nfs4_get_machine_cred,
10474	.renew_lease = nfs4_proc_sequence,
10475};
10476#endif
10477
10478static const struct nfs4_mig_recovery_ops nfs40_mig_recovery_ops = {
10479	.get_locations = _nfs40_proc_get_locations,
10480	.fsid_present = _nfs40_proc_fsid_present,
10481};
10482
10483#if defined(CONFIG_NFS_V4_1)
10484static const struct nfs4_mig_recovery_ops nfs41_mig_recovery_ops = {
10485	.get_locations = _nfs41_proc_get_locations,
10486	.fsid_present = _nfs41_proc_fsid_present,
10487};
10488#endif	/* CONFIG_NFS_V4_1 */
10489
10490static const struct nfs4_minor_version_ops nfs_v4_0_minor_ops = {
10491	.minor_version = 0,
10492	.init_caps = NFS_CAP_READDIRPLUS
10493		| NFS_CAP_ATOMIC_OPEN
10494		| NFS_CAP_POSIX_LOCK,
10495	.init_client = nfs40_init_client,
10496	.shutdown_client = nfs40_shutdown_client,
10497	.match_stateid = nfs4_match_stateid,
10498	.find_root_sec = nfs4_find_root_sec,
10499	.free_lock_state = nfs4_release_lockowner,
10500	.test_and_free_expired = nfs40_test_and_free_expired_stateid,
10501	.alloc_seqid = nfs_alloc_seqid,
10502	.call_sync_ops = &nfs40_call_sync_ops,
10503	.reboot_recovery_ops = &nfs40_reboot_recovery_ops,
10504	.nograce_recovery_ops = &nfs40_nograce_recovery_ops,
10505	.state_renewal_ops = &nfs40_state_renewal_ops,
10506	.mig_recovery_ops = &nfs40_mig_recovery_ops,
10507};
10508
10509#if defined(CONFIG_NFS_V4_1)
10510static struct nfs_seqid *
10511nfs_alloc_no_seqid(struct nfs_seqid_counter *arg1, gfp_t arg2)
10512{
10513	return NULL;
10514}
10515
10516static const struct nfs4_minor_version_ops nfs_v4_1_minor_ops = {
10517	.minor_version = 1,
10518	.init_caps = NFS_CAP_READDIRPLUS
10519		| NFS_CAP_ATOMIC_OPEN
10520		| NFS_CAP_POSIX_LOCK
10521		| NFS_CAP_STATEID_NFSV41
10522		| NFS_CAP_ATOMIC_OPEN_V1
10523		| NFS_CAP_LGOPEN
10524		| NFS_CAP_MOVEABLE,
10525	.init_client = nfs41_init_client,
10526	.shutdown_client = nfs41_shutdown_client,
10527	.match_stateid = nfs41_match_stateid,
10528	.find_root_sec = nfs41_find_root_sec,
10529	.free_lock_state = nfs41_free_lock_state,
10530	.test_and_free_expired = nfs41_test_and_free_expired_stateid,
10531	.alloc_seqid = nfs_alloc_no_seqid,
10532	.session_trunk = nfs4_test_session_trunk,
10533	.call_sync_ops = &nfs41_call_sync_ops,
10534	.reboot_recovery_ops = &nfs41_reboot_recovery_ops,
10535	.nograce_recovery_ops = &nfs41_nograce_recovery_ops,
10536	.state_renewal_ops = &nfs41_state_renewal_ops,
10537	.mig_recovery_ops = &nfs41_mig_recovery_ops,
10538};
10539#endif
10540
10541#if defined(CONFIG_NFS_V4_2)
10542static const struct nfs4_minor_version_ops nfs_v4_2_minor_ops = {
10543	.minor_version = 2,
10544	.init_caps = NFS_CAP_READDIRPLUS
10545		| NFS_CAP_ATOMIC_OPEN
10546		| NFS_CAP_POSIX_LOCK
10547		| NFS_CAP_STATEID_NFSV41
10548		| NFS_CAP_ATOMIC_OPEN_V1
10549		| NFS_CAP_LGOPEN
10550		| NFS_CAP_ALLOCATE
10551		| NFS_CAP_COPY
10552		| NFS_CAP_OFFLOAD_CANCEL
10553		| NFS_CAP_COPY_NOTIFY
10554		| NFS_CAP_DEALLOCATE
10555		| NFS_CAP_SEEK
10556		| NFS_CAP_LAYOUTSTATS
10557		| NFS_CAP_CLONE
10558		| NFS_CAP_LAYOUTERROR
10559		| NFS_CAP_READ_PLUS
10560		| NFS_CAP_MOVEABLE,
10561	.init_client = nfs41_init_client,
10562	.shutdown_client = nfs41_shutdown_client,
10563	.match_stateid = nfs41_match_stateid,
10564	.find_root_sec = nfs41_find_root_sec,
10565	.free_lock_state = nfs41_free_lock_state,
10566	.call_sync_ops = &nfs41_call_sync_ops,
10567	.test_and_free_expired = nfs41_test_and_free_expired_stateid,
10568	.alloc_seqid = nfs_alloc_no_seqid,
10569	.session_trunk = nfs4_test_session_trunk,
10570	.reboot_recovery_ops = &nfs41_reboot_recovery_ops,
10571	.nograce_recovery_ops = &nfs41_nograce_recovery_ops,
10572	.state_renewal_ops = &nfs41_state_renewal_ops,
10573	.mig_recovery_ops = &nfs41_mig_recovery_ops,
10574};
10575#endif
10576
10577const struct nfs4_minor_version_ops *nfs_v4_minor_ops[] = {
10578	[0] = &nfs_v4_0_minor_ops,
10579#if defined(CONFIG_NFS_V4_1)
10580	[1] = &nfs_v4_1_minor_ops,
10581#endif
10582#if defined(CONFIG_NFS_V4_2)
10583	[2] = &nfs_v4_2_minor_ops,
10584#endif
10585};
10586
10587static ssize_t nfs4_listxattr(struct dentry *dentry, char *list, size_t size)
10588{
10589	ssize_t error, error2, error3;
10590	size_t left = size;
10591
10592	error = generic_listxattr(dentry, list, left);
10593	if (error < 0)
10594		return error;
10595	if (list) {
10596		list += error;
10597		left -= error;
10598	}
10599
10600	error2 = nfs4_listxattr_nfs4_label(d_inode(dentry), list, left);
10601	if (error2 < 0)
10602		return error2;
10603
10604	if (list) {
10605		list += error2;
10606		left -= error2;
10607	}
10608
10609	error3 = nfs4_listxattr_nfs4_user(d_inode(dentry), list, left);
10610	if (error3 < 0)
10611		return error3;
10612
10613	error += error2 + error3;
10614	if (size && error > size)
10615		return -ERANGE;
10616	return error;
10617}
10618
10619static void nfs4_enable_swap(struct inode *inode)
10620{
10621	/* The state manager thread must always be running.
10622	 * It will notice the client is a swapper, and stay put.
10623	 */
10624	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
10625
10626	nfs4_schedule_state_manager(clp);
10627}
10628
10629static void nfs4_disable_swap(struct inode *inode)
10630{
10631	/* The state manager thread will now exit once it is
10632	 * woken.
10633	 */
10634	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
10635
10636	set_bit(NFS4CLNT_RUN_MANAGER, &clp->cl_state);
10637	clear_bit(NFS4CLNT_MANAGER_AVAILABLE, &clp->cl_state);
10638	wake_up_var(&clp->cl_state);
10639}
10640
10641static const struct inode_operations nfs4_dir_inode_operations = {
10642	.create		= nfs_create,
10643	.lookup		= nfs_lookup,
10644	.atomic_open	= nfs_atomic_open,
10645	.link		= nfs_link,
10646	.unlink		= nfs_unlink,
10647	.symlink	= nfs_symlink,
10648	.mkdir		= nfs_mkdir,
10649	.rmdir		= nfs_rmdir,
10650	.mknod		= nfs_mknod,
10651	.rename		= nfs_rename,
10652	.permission	= nfs_permission,
10653	.getattr	= nfs_getattr,
10654	.setattr	= nfs_setattr,
10655	.listxattr	= nfs4_listxattr,
10656};
10657
10658static const struct inode_operations nfs4_file_inode_operations = {
10659	.permission	= nfs_permission,
10660	.getattr	= nfs_getattr,
10661	.setattr	= nfs_setattr,
10662	.listxattr	= nfs4_listxattr,
10663};
10664
10665const struct nfs_rpc_ops nfs_v4_clientops = {
10666	.version	= 4,			/* protocol version */
10667	.dentry_ops	= &nfs4_dentry_operations,
10668	.dir_inode_ops	= &nfs4_dir_inode_operations,
10669	.file_inode_ops	= &nfs4_file_inode_operations,
10670	.file_ops	= &nfs4_file_operations,
10671	.getroot	= nfs4_proc_get_root,
10672	.submount	= nfs4_submount,
10673	.try_get_tree	= nfs4_try_get_tree,
10674	.getattr	= nfs4_proc_getattr,
10675	.setattr	= nfs4_proc_setattr,
10676	.lookup		= nfs4_proc_lookup,
10677	.lookupp	= nfs4_proc_lookupp,
10678	.access		= nfs4_proc_access,
10679	.readlink	= nfs4_proc_readlink,
10680	.create		= nfs4_proc_create,
10681	.remove		= nfs4_proc_remove,
10682	.unlink_setup	= nfs4_proc_unlink_setup,
10683	.unlink_rpc_prepare = nfs4_proc_unlink_rpc_prepare,
10684	.unlink_done	= nfs4_proc_unlink_done,
10685	.rename_setup	= nfs4_proc_rename_setup,
10686	.rename_rpc_prepare = nfs4_proc_rename_rpc_prepare,
10687	.rename_done	= nfs4_proc_rename_done,
10688	.link		= nfs4_proc_link,
10689	.symlink	= nfs4_proc_symlink,
10690	.mkdir		= nfs4_proc_mkdir,
10691	.rmdir		= nfs4_proc_rmdir,
10692	.readdir	= nfs4_proc_readdir,
10693	.mknod		= nfs4_proc_mknod,
10694	.statfs		= nfs4_proc_statfs,
10695	.fsinfo		= nfs4_proc_fsinfo,
10696	.pathconf	= nfs4_proc_pathconf,
10697	.set_capabilities = nfs4_server_capabilities,
10698	.decode_dirent	= nfs4_decode_dirent,
10699	.pgio_rpc_prepare = nfs4_proc_pgio_rpc_prepare,
10700	.read_setup	= nfs4_proc_read_setup,
10701	.read_done	= nfs4_read_done,
10702	.write_setup	= nfs4_proc_write_setup,
10703	.write_done	= nfs4_write_done,
10704	.commit_setup	= nfs4_proc_commit_setup,
10705	.commit_rpc_prepare = nfs4_proc_commit_rpc_prepare,
10706	.commit_done	= nfs4_commit_done,
10707	.lock		= nfs4_proc_lock,
10708	.clear_acl_cache = nfs4_zap_acl_attr,
10709	.close_context  = nfs4_close_context,
10710	.open_context	= nfs4_atomic_open,
10711	.have_delegation = nfs4_have_delegation,
10712	.alloc_client	= nfs4_alloc_client,
10713	.init_client	= nfs4_init_client,
10714	.free_client	= nfs4_free_client,
10715	.create_server	= nfs4_create_server,
10716	.clone_server	= nfs_clone_server,
10717	.discover_trunking = nfs4_discover_trunking,
10718	.enable_swap	= nfs4_enable_swap,
10719	.disable_swap	= nfs4_disable_swap,
10720};
10721
10722static const struct xattr_handler nfs4_xattr_nfs4_acl_handler = {
10723	.name	= XATTR_NAME_NFSV4_ACL,
10724	.list	= nfs4_xattr_list_nfs4_acl,
10725	.get	= nfs4_xattr_get_nfs4_acl,
10726	.set	= nfs4_xattr_set_nfs4_acl,
10727};
10728
10729#if defined(CONFIG_NFS_V4_1)
10730static const struct xattr_handler nfs4_xattr_nfs4_dacl_handler = {
10731	.name	= XATTR_NAME_NFSV4_DACL,
10732	.list	= nfs4_xattr_list_nfs4_dacl,
10733	.get	= nfs4_xattr_get_nfs4_dacl,
10734	.set	= nfs4_xattr_set_nfs4_dacl,
10735};
10736
10737static const struct xattr_handler nfs4_xattr_nfs4_sacl_handler = {
10738	.name	= XATTR_NAME_NFSV4_SACL,
10739	.list	= nfs4_xattr_list_nfs4_sacl,
10740	.get	= nfs4_xattr_get_nfs4_sacl,
10741	.set	= nfs4_xattr_set_nfs4_sacl,
10742};
10743#endif
10744
10745#ifdef CONFIG_NFS_V4_2
10746static const struct xattr_handler nfs4_xattr_nfs4_user_handler = {
10747	.prefix	= XATTR_USER_PREFIX,
10748	.get	= nfs4_xattr_get_nfs4_user,
10749	.set	= nfs4_xattr_set_nfs4_user,
10750};
10751#endif
10752
10753const struct xattr_handler *nfs4_xattr_handlers[] = {
10754	&nfs4_xattr_nfs4_acl_handler,
10755#if defined(CONFIG_NFS_V4_1)
10756	&nfs4_xattr_nfs4_dacl_handler,
10757	&nfs4_xattr_nfs4_sacl_handler,
10758#endif
10759#ifdef CONFIG_NFS_V4_SECURITY_LABEL
10760	&nfs4_xattr_nfs4_label_handler,
10761#endif
10762#ifdef CONFIG_NFS_V4_2
10763	&nfs4_xattr_nfs4_user_handler,
10764#endif
10765	NULL
10766};
10767