nfs4proc.c 235.2 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
/*
 *  fs/nfs/nfs4proc.c
 *
 *  Client-side procedure declarations for NFSv4.
 *
 *  Copyright (c) 2002 The Regents of the University of Michigan.
 *  All rights reserved.
 *
 *  Kendrick Smith <kmsmith@umich.edu>
 *  Andy Adamson   <andros@umich.edu>
 *
 *  Redistribution and use in source and binary forms, with or without
 *  modification, are permitted provided that the following conditions
 *  are met:
 *
 *  1. Redistributions of source code must retain the above copyright
 *     notice, this list of conditions and the following disclaimer.
 *  2. Redistributions in binary form must reproduce the above copyright
 *     notice, this list of conditions and the following disclaimer in the
 *     documentation and/or other materials provided with the distribution.
 *  3. Neither the name of the University nor the names of its
 *     contributors may be used to endorse or promote products derived
 *     from this software without specific prior written permission.
 *
 *  THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
 *  WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
 *  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 *  DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 *  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
 *  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
 *  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
 *  BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
 *  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
 *  NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 *  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */

#include <linux/mm.h>
#include <linux/delay.h>
#include <linux/errno.h>
41
#include <linux/file.h>
L
Linus Torvalds 已提交
42
#include <linux/string.h>
43 44
#include <linux/ratelimit.h>
#include <linux/printk.h>
45
#include <linux/slab.h>
L
Linus Torvalds 已提交
46 47 48 49 50
#include <linux/sunrpc/clnt.h>
#include <linux/nfs.h>
#include <linux/nfs4.h>
#include <linux/nfs_fs.h>
#include <linux/nfs_page.h>
51
#include <linux/nfs_mount.h>
L
Linus Torvalds 已提交
52
#include <linux/namei.h>
53
#include <linux/mount.h>
B
Benny Halevy 已提交
54
#include <linux/module.h>
55
#include <linux/xattr.h>
56
#include <linux/utsname.h>
57
#include <linux/freezer.h>
L
Linus Torvalds 已提交
58

59
#include "nfs4_fs.h"
L
Linus Torvalds 已提交
60
#include "delegation.h"
61
#include "internal.h"
62
#include "iostat.h"
A
Andy Adamson 已提交
63
#include "callback.h"
64
#include "pnfs.h"
65
#include "netns.h"
66
#include "nfs4idmap.h"
67
#include "nfs4session.h"
D
David Howells 已提交
68
#include "fscache.h"
L
Linus Torvalds 已提交
69

70 71
#include "nfs4trace.h"

L
Linus Torvalds 已提交
72 73
#define NFSDBG_FACILITY		NFSDBG_PROC

74
#define NFS4_POLL_RETRY_MIN	(HZ/10)
L
Linus Torvalds 已提交
75 76
#define NFS4_POLL_RETRY_MAX	(15*HZ)

77
struct nfs4_opendata;
78
static int _nfs4_proc_open(struct nfs4_opendata *data);
79
static int _nfs4_recover_proc_open(struct nfs4_opendata *data);
L
Linus Torvalds 已提交
80
static int nfs4_do_fsinfo(struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *);
81
static int nfs4_async_handle_error(struct rpc_task *, const struct nfs_server *, struct nfs4_state *, long *);
82
static void nfs_fixup_referral_attributes(struct nfs_fattr *fattr);
83 84
static int nfs4_proc_getattr(struct nfs_server *, struct nfs_fh *, struct nfs_fattr *, struct nfs4_label *label);
static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fattr *fattr, struct nfs4_label *label);
85 86
static int nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
			    struct nfs_fattr *fattr, struct iattr *sattr,
87 88
			    struct nfs4_state *state, struct nfs4_label *ilabel,
			    struct nfs4_label *olabel);
89
#ifdef CONFIG_NFS_V4_1
90 91 92 93
static int nfs41_test_stateid(struct nfs_server *, nfs4_stateid *,
		struct rpc_cred *);
static int nfs41_free_stateid(struct nfs_server *, nfs4_stateid *,
		struct rpc_cred *);
94
#endif
95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141

#ifdef CONFIG_NFS_V4_SECURITY_LABEL
static inline struct nfs4_label *
nfs4_label_init_security(struct inode *dir, struct dentry *dentry,
	struct iattr *sattr, struct nfs4_label *label)
{
	int err;

	if (label == NULL)
		return NULL;

	if (nfs_server_capable(dir, NFS_CAP_SECURITY_LABEL) == 0)
		return NULL;

	err = security_dentry_init_security(dentry, sattr->ia_mode,
				&dentry->d_name, (void **)&label->label, &label->len);
	if (err == 0)
		return label;

	return NULL;
}
static inline void
nfs4_label_release_security(struct nfs4_label *label)
{
	if (label)
		security_release_secctx(label->label, label->len);
}
static inline u32 *nfs4_bitmask(struct nfs_server *server, struct nfs4_label *label)
{
	if (label)
		return server->attr_bitmask;

	return server->attr_bitmask_nl;
}
#else
static inline struct nfs4_label *
nfs4_label_init_security(struct inode *dir, struct dentry *dentry,
	struct iattr *sattr, struct nfs4_label *l)
{ return NULL; }
static inline void
nfs4_label_release_security(struct nfs4_label *label)
{ return; }
static inline u32 *
nfs4_bitmask(struct nfs_server *server, struct nfs4_label *label)
{ return server->attr_bitmask; }
#endif

L
Linus Torvalds 已提交
142
/* Prevent leaks of NFSv4 errors into userland */
143
static int nfs4_map_errors(int err)
L
Linus Torvalds 已提交
144
{
145 146 147 148
	if (err >= -1000)
		return err;
	switch (err) {
	case -NFS4ERR_RESOURCE:
149 150
	case -NFS4ERR_LAYOUTTRYLATER:
	case -NFS4ERR_RECALLCONFLICT:
151
		return -EREMOTEIO;
152
	case -NFS4ERR_WRONGSEC:
153
	case -NFS4ERR_WRONG_CRED:
154
		return -EPERM;
155 156 157
	case -NFS4ERR_BADOWNER:
	case -NFS4ERR_BADNAME:
		return -EINVAL;
158 159
	case -NFS4ERR_SHARE_DENIED:
		return -EACCES;
160 161
	case -NFS4ERR_MINOR_VERS_MISMATCH:
		return -EPROTONOSUPPORT;
162 163
	case -NFS4ERR_FILE_OPEN:
		return -EBUSY;
164
	default:
L
Linus Torvalds 已提交
165
		dprintk("%s could not handle NFSv4 error %d\n",
166
				__func__, -err);
167
		break;
L
Linus Torvalds 已提交
168
	}
169
	return -EIO;
L
Linus Torvalds 已提交
170 171 172 173 174
}

/*
 * This is our standard bitmap for GETATTR requests.
 */
175
const u32 nfs4_fattr_bitmap[3] = {
L
Linus Torvalds 已提交
176 177 178 179 180 181 182 183 184 185 186 187 188
	FATTR4_WORD0_TYPE
	| FATTR4_WORD0_CHANGE
	| FATTR4_WORD0_SIZE
	| FATTR4_WORD0_FSID
	| FATTR4_WORD0_FILEID,
	FATTR4_WORD1_MODE
	| FATTR4_WORD1_NUMLINKS
	| FATTR4_WORD1_OWNER
	| FATTR4_WORD1_OWNER_GROUP
	| FATTR4_WORD1_RAWDEV
	| FATTR4_WORD1_SPACE_USED
	| FATTR4_WORD1_TIME_ACCESS
	| FATTR4_WORD1_TIME_METADATA
189 190
	| FATTR4_WORD1_TIME_MODIFY
	| FATTR4_WORD1_MOUNTED_ON_FILEID,
191 192 193
#ifdef CONFIG_NFS_V4_SECURITY_LABEL
	FATTR4_WORD2_SECURITY_LABEL
#endif
L
Linus Torvalds 已提交
194 195
};

196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213
static const u32 nfs4_pnfs_open_bitmap[3] = {
	FATTR4_WORD0_TYPE
	| FATTR4_WORD0_CHANGE
	| FATTR4_WORD0_SIZE
	| FATTR4_WORD0_FSID
	| FATTR4_WORD0_FILEID,
	FATTR4_WORD1_MODE
	| FATTR4_WORD1_NUMLINKS
	| FATTR4_WORD1_OWNER
	| FATTR4_WORD1_OWNER_GROUP
	| FATTR4_WORD1_RAWDEV
	| FATTR4_WORD1_SPACE_USED
	| FATTR4_WORD1_TIME_ACCESS
	| FATTR4_WORD1_TIME_METADATA
	| FATTR4_WORD1_TIME_MODIFY,
	FATTR4_WORD2_MDSTHRESHOLD
};

214 215 216 217 218 219
static const u32 nfs4_open_noattr_bitmap[3] = {
	FATTR4_WORD0_TYPE
	| FATTR4_WORD0_CHANGE
	| FATTR4_WORD0_FILEID,
};

220
const u32 nfs4_statfs_bitmap[3] = {
L
Linus Torvalds 已提交
221 222 223 224 225 226 227 228
	FATTR4_WORD0_FILES_AVAIL
	| FATTR4_WORD0_FILES_FREE
	| FATTR4_WORD0_FILES_TOTAL,
	FATTR4_WORD1_SPACE_AVAIL
	| FATTR4_WORD1_SPACE_FREE
	| FATTR4_WORD1_SPACE_TOTAL
};

229
const u32 nfs4_pathconf_bitmap[3] = {
L
Linus Torvalds 已提交
230 231 232 233 234
	FATTR4_WORD0_MAXLINK
	| FATTR4_WORD0_MAXNAME,
	0
};

235
const u32 nfs4_fsinfo_bitmap[3] = { FATTR4_WORD0_MAXFILESIZE
L
Linus Torvalds 已提交
236 237 238
			| FATTR4_WORD0_MAXREAD
			| FATTR4_WORD0_MAXWRITE
			| FATTR4_WORD0_LEASE_TIME,
239
			FATTR4_WORD1_TIME_DELTA
240 241
			| FATTR4_WORD1_FS_LAYOUT_TYPES,
			FATTR4_WORD2_LAYOUT_BLKSIZE
L
Linus Torvalds 已提交
242 243
};

244
const u32 nfs4_fs_locations_bitmap[3] = {
245 246 247 248 249 250 251 252 253 254 255 256 257 258 259
	FATTR4_WORD0_TYPE
	| FATTR4_WORD0_CHANGE
	| FATTR4_WORD0_SIZE
	| FATTR4_WORD0_FSID
	| FATTR4_WORD0_FILEID
	| FATTR4_WORD0_FS_LOCATIONS,
	FATTR4_WORD1_MODE
	| FATTR4_WORD1_NUMLINKS
	| FATTR4_WORD1_OWNER
	| FATTR4_WORD1_OWNER_GROUP
	| FATTR4_WORD1_RAWDEV
	| FATTR4_WORD1_SPACE_USED
	| FATTR4_WORD1_TIME_ACCESS
	| FATTR4_WORD1_TIME_METADATA
	| FATTR4_WORD1_TIME_MODIFY
260
	| FATTR4_WORD1_MOUNTED_ON_FILEID,
261 262
};

A
Al Viro 已提交
263
static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dentry,
L
Linus Torvalds 已提交
264 265
		struct nfs4_readdir_arg *readdir)
{
266
	__be32 *start, *p;
L
Linus Torvalds 已提交
267 268

	if (cookie > 2) {
269
		readdir->cookie = cookie;
L
Linus Torvalds 已提交
270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285
		memcpy(&readdir->verifier, verifier, sizeof(readdir->verifier));
		return;
	}

	readdir->cookie = 0;
	memset(&readdir->verifier, 0, sizeof(readdir->verifier));
	if (cookie == 2)
		return;
	
	/*
	 * NFSv4 servers do not return entries for '.' and '..'
	 * Therefore, we fake these entries here.  We let '.'
	 * have cookie 0 and '..' have cookie 1.  Note that
	 * when talking to the server, we always send cookie 0
	 * instead of 1 or 2.
	 */
286
	start = p = kmap_atomic(*readdir->pages);
L
Linus Torvalds 已提交
287 288 289 290 291 292 293 294 295 296 297
	
	if (cookie == 0) {
		*p++ = xdr_one;                                  /* next */
		*p++ = xdr_zero;                   /* cookie, first word */
		*p++ = xdr_one;                   /* cookie, second word */
		*p++ = xdr_one;                             /* entry len */
		memcpy(p, ".\0\0\0", 4);                        /* entry */
		p++;
		*p++ = xdr_one;                         /* bitmap length */
		*p++ = htonl(FATTR4_WORD0_FILEID);             /* bitmap */
		*p++ = htonl(8);              /* attribute buffer length */
298
		p = xdr_encode_hyper(p, NFS_FILEID(d_inode(dentry)));
L
Linus Torvalds 已提交
299 300 301 302 303 304 305 306 307 308 309
	}
	
	*p++ = xdr_one;                                  /* next */
	*p++ = xdr_zero;                   /* cookie, first word */
	*p++ = xdr_two;                   /* cookie, second word */
	*p++ = xdr_two;                             /* entry len */
	memcpy(p, "..\0\0", 4);                         /* entry */
	p++;
	*p++ = xdr_one;                         /* bitmap length */
	*p++ = htonl(FATTR4_WORD0_FILEID);             /* bitmap */
	*p++ = htonl(8);              /* attribute buffer length */
310
	p = xdr_encode_hyper(p, NFS_FILEID(d_inode(dentry->d_parent)));
L
Linus Torvalds 已提交
311 312 313

	readdir->pgbase = (char *)p - (char *)start;
	readdir->count -= readdir->pgbase;
314
	kunmap_atomic(start);
L
Linus Torvalds 已提交
315 316
}

317 318 319 320 321 322 323 324 325 326 327 328 329 330
static long nfs4_update_delay(long *timeout)
{
	long ret;
	if (!timeout)
		return NFS4_POLL_RETRY_MAX;
	if (*timeout <= 0)
		*timeout = NFS4_POLL_RETRY_MIN;
	if (*timeout > NFS4_POLL_RETRY_MAX)
		*timeout = NFS4_POLL_RETRY_MAX;
	ret = *timeout;
	*timeout <<= 1;
	return ret;
}

331 332 333 334 335 336
static int nfs4_delay(struct rpc_clnt *clnt, long *timeout)
{
	int res = 0;

	might_sleep();

337 338
	freezable_schedule_timeout_killable_unsafe(
		nfs4_update_delay(timeout));
339 340 341 342 343 344 345 346
	if (fatal_signal_pending(current))
		res = -ERESTARTSYS;
	return res;
}

/* This is the error handling routine for processes that are allowed
 * to sleep.
 */
A
Anna Schumaker 已提交
347
int nfs4_handle_exception(struct nfs_server *server, int errorcode, struct nfs4_exception *exception)
348 349
{
	struct nfs_client *clp = server->nfs_client;
350
	struct nfs4_state *state = exception->state;
351
	struct inode *inode = exception->inode;
352 353 354 355 356 357
	int ret = errorcode;

	exception->retry = 0;
	switch(errorcode) {
		case 0:
			return 0;
358
		case -NFS4ERR_OPENMODE:
359 360 361
		case -NFS4ERR_DELEG_REVOKED:
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
362
			if (inode && nfs4_have_delegation(inode, FMODE_READ)) {
363
				nfs4_inode_return_delegation(inode);
364 365 366 367 368
				exception->retry = 1;
				return 0;
			}
			if (state == NULL)
				break;
369 370 371
			ret = nfs4_schedule_stateid_recovery(server, state);
			if (ret < 0)
				break;
372
			goto wait_on_recovery;
373
		case -NFS4ERR_EXPIRED:
374 375 376 377 378
			if (state != NULL) {
				ret = nfs4_schedule_stateid_recovery(server, state);
				if (ret < 0)
					break;
			}
379
		case -NFS4ERR_STALE_STATEID:
380
		case -NFS4ERR_STALE_CLIENTID:
381 382
			nfs4_schedule_lease_recovery(clp);
			goto wait_on_recovery;
383 384 385 386 387
		case -NFS4ERR_MOVED:
			ret = nfs4_schedule_migration_recovery(server);
			if (ret < 0)
				break;
			goto wait_on_recovery;
388 389 390
		case -NFS4ERR_LEASE_MOVED:
			nfs4_schedule_lease_moved_recovery(clp);
			goto wait_on_recovery;
391
#if defined(CONFIG_NFS_V4_1)
392 393 394 395 396 397 398 399 400
		case -NFS4ERR_BADSESSION:
		case -NFS4ERR_BADSLOT:
		case -NFS4ERR_BAD_HIGH_SLOT:
		case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
		case -NFS4ERR_DEADSESSION:
		case -NFS4ERR_SEQ_FALSE_RETRY:
		case -NFS4ERR_SEQ_MISORDERED:
			dprintk("%s ERROR: %d Reset session\n", __func__,
				errorcode);
401
			nfs4_schedule_session_recovery(clp->cl_session, errorcode);
402
			goto wait_on_recovery;
403
#endif /* defined(CONFIG_NFS_V4_1) */
404
		case -NFS4ERR_FILE_OPEN:
405 406 407 408 409 410 411
			if (exception->timeout > HZ) {
				/* We have retried a decent amount, time to
				 * fail
				 */
				ret = -EBUSY;
				break;
			}
412 413 414 415 416
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			ret = nfs4_delay(server->client, &exception->timeout);
			if (ret != 0)
				break;
417
		case -NFS4ERR_RETRY_UNCACHED_REP:
418 419
		case -NFS4ERR_OLD_STATEID:
			exception->retry = 1;
420 421 422 423 424 425 426 427 428 429 430 431 432
			break;
		case -NFS4ERR_BADOWNER:
			/* The following works around a Linux server bug! */
		case -NFS4ERR_BADNAME:
			if (server->caps & NFS_CAP_UIDGID_NOMAP) {
				server->caps &= ~NFS_CAP_UIDGID_NOMAP;
				exception->retry = 1;
				printk(KERN_WARNING "NFS: v4 server %s "
						"does not accept raw "
						"uid/gids. "
						"Reenabling the idmapper.\n",
						server->nfs_client->cl_hostname);
			}
433 434 435
	}
	/* We failed to handle the error */
	return nfs4_map_errors(ret);
436
wait_on_recovery:
437
	ret = nfs4_wait_clnt_recover(clp);
438 439
	if (test_bit(NFS_MIG_FAILED, &server->mig_status))
		return -EIO;
440 441 442
	if (ret == 0)
		exception->retry = 1;
	return ret;
443 444
}

445 446 447 448 449 450 451 452 453 454 455 456 457 458
/*
 * Return 'true' if 'clp' is using an rpc_client that is integrity protected
 * or 'false' otherwise.
 */
static bool _nfs4_is_integrity_protected(struct nfs_client *clp)
{
	rpc_authflavor_t flavor = clp->cl_rpcclient->cl_auth->au_flavor;

	if (flavor == RPC_AUTH_GSS_KRB5I ||
	    flavor == RPC_AUTH_GSS_KRB5P)
		return true;

	return false;
}
459

460
static void do_renew_lease(struct nfs_client *clp, unsigned long timestamp)
L
Linus Torvalds 已提交
461 462 463 464 465 466 467
{
	spin_lock(&clp->cl_lock);
	if (time_before(clp->cl_last_renewal,timestamp))
		clp->cl_last_renewal = timestamp;
	spin_unlock(&clp->cl_lock);
}

468 469 470 471 472
static void renew_lease(const struct nfs_server *server, unsigned long timestamp)
{
	do_renew_lease(server->nfs_client, timestamp);
}

473 474 475 476 477 478
struct nfs4_call_sync_data {
	const struct nfs_server *seq_server;
	struct nfs4_sequence_args *seq_args;
	struct nfs4_sequence_res *seq_res;
};

479 480
void nfs4_init_sequence(struct nfs4_sequence_args *args,
			struct nfs4_sequence_res *res, int cache_reply)
481 482 483 484 485 486 487 488 489 490 491 492 493
{
	args->sa_slot = NULL;
	args->sa_cache_this = cache_reply;
	args->sa_privileged = 0;

	res->sr_slot = NULL;
}

static void nfs4_set_sequence_privileged(struct nfs4_sequence_args *args)
{
	args->sa_privileged = 1;
}

494 495 496 497
int nfs40_setup_sequence(struct nfs4_slot_table *tbl,
			 struct nfs4_sequence_args *args,
			 struct nfs4_sequence_res *res,
			 struct rpc_task *task)
C
Chuck Lever 已提交
498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532
{
	struct nfs4_slot *slot;

	/* slot already allocated? */
	if (res->sr_slot != NULL)
		goto out_start;

	spin_lock(&tbl->slot_tbl_lock);
	if (nfs4_slot_tbl_draining(tbl) && !args->sa_privileged)
		goto out_sleep;

	slot = nfs4_alloc_slot(tbl);
	if (IS_ERR(slot)) {
		if (slot == ERR_PTR(-ENOMEM))
			task->tk_timeout = HZ >> 2;
		goto out_sleep;
	}
	spin_unlock(&tbl->slot_tbl_lock);

	args->sa_slot = slot;
	res->sr_slot = slot;

out_start:
	rpc_call_start(task);
	return 0;

out_sleep:
	if (args->sa_privileged)
		rpc_sleep_on_priority(&tbl->slot_tbl_waitq, task,
				NULL, RPC_PRIORITY_PRIVILEGED);
	else
		rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
	spin_unlock(&tbl->slot_tbl_lock);
	return -EAGAIN;
}
533
EXPORT_SYMBOL_GPL(nfs40_setup_sequence);
C
Chuck Lever 已提交
534 535 536 537 538 539 540

static int nfs40_sequence_done(struct rpc_task *task,
			       struct nfs4_sequence_res *res)
{
	struct nfs4_slot *slot = res->sr_slot;
	struct nfs4_slot_table *tbl;

541
	if (slot == NULL)
C
Chuck Lever 已提交
542 543 544 545 546 547 548 549 550 551 552 553 554
		goto out;

	tbl = slot->table;
	spin_lock(&tbl->slot_tbl_lock);
	if (!nfs41_wake_and_assign_slot(tbl, slot))
		nfs4_free_slot(tbl, slot);
	spin_unlock(&tbl->slot_tbl_lock);

	res->sr_slot = NULL;
out:
	return 1;
}

A
Andy Adamson 已提交
555 556
#if defined(CONFIG_NFS_V4_1)

557
static void nfs41_sequence_free_slot(struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
558
{
559
	struct nfs4_session *session;
A
Andy Adamson 已提交
560
	struct nfs4_slot_table *tbl;
561
	struct nfs4_slot *slot = res->sr_slot;
562
	bool send_new_highest_used_slotid = false;
A
Andy Adamson 已提交
563

564
	tbl = slot->table;
565
	session = tbl->session;
566

567
	spin_lock(&tbl->slot_tbl_lock);
568 569 570 571 572 573
	/* Be nice to the server: try to ensure that the last transmitted
	 * value for highest_user_slotid <= target_highest_slotid
	 */
	if (tbl->highest_used_slotid > tbl->target_highest_slotid)
		send_new_highest_used_slotid = true;

574
	if (nfs41_wake_and_assign_slot(tbl, slot)) {
575 576 577
		send_new_highest_used_slotid = false;
		goto out_unlock;
	}
578
	nfs4_free_slot(tbl, slot);
579 580 581

	if (tbl->highest_used_slotid != NFS4_NO_SLOT)
		send_new_highest_used_slotid = false;
582
out_unlock:
583
	spin_unlock(&tbl->slot_tbl_lock);
584
	res->sr_slot = NULL;
585 586
	if (send_new_highest_used_slotid)
		nfs41_server_notify_highest_slotid_update(session->clp);
A
Andy Adamson 已提交
587 588
}

589
int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
590
{
591
	struct nfs4_session *session;
592
	struct nfs4_slot *slot = res->sr_slot;
593
	struct nfs_client *clp;
594
	bool interrupted = false;
595
	int ret = 1;
A
Andy Adamson 已提交
596

597 598
	if (slot == NULL)
		goto out_noaction;
599 600
	/* don't increment the sequence number if the task wasn't sent */
	if (!RPC_WAS_SENT(task))
A
Andy Adamson 已提交
601 602
		goto out;

603
	session = slot->table->session;
604

605 606 607 608 609
	if (slot->interrupted) {
		slot->interrupted = 0;
		interrupted = true;
	}

610
	trace_nfs4_sequence_done(session, res);
A
Andy Adamson 已提交
611
	/* Check the SEQUENCE operation status */
612 613
	switch (res->sr_status) {
	case 0:
A
Andy Adamson 已提交
614
		/* Update the slot's sequence and clientid lease timer */
615
		++slot->seq_nr;
616
		clp = session->clp;
617
		do_renew_lease(clp, res->sr_timestamp);
618
		/* Check sequence flags */
619 620
		if (res->sr_status_flags != 0)
			nfs4_schedule_lease_recovery(clp);
621
		nfs41_update_target_slotid(slot->table, slot, res);
622
		break;
623 624 625 626 627 628 629 630 631
	case 1:
		/*
		 * sr_status remains 1 if an RPC level error occurred.
		 * The server may or may not have processed the sequence
		 * operation..
		 * Mark the slot as having hosted an interrupted RPC call.
		 */
		slot->interrupted = 1;
		goto out;
632 633 634 635 636
	case -NFS4ERR_DELAY:
		/* The server detected a resend of the RPC call and
		 * returned NFS4ERR_DELAY as per Section 2.10.6.2
		 * of RFC5661.
		 */
637
		dprintk("%s: slot=%u seq=%u: Operation in progress\n",
638
			__func__,
639
			slot->slot_nr,
640
			slot->seq_nr);
641
		goto out_retry;
642 643 644 645 646
	case -NFS4ERR_BADSLOT:
		/*
		 * The slot id we used was probably retired. Try again
		 * using a different slot id.
		 */
647 648
		goto retry_nowait;
	case -NFS4ERR_SEQ_MISORDERED:
649 650 651 652 653 654 655 656
		/*
		 * Was the last operation on this sequence interrupted?
		 * If so, retry after bumping the sequence number.
		 */
		if (interrupted) {
			++slot->seq_nr;
			goto retry_nowait;
		}
657 658 659 660
		/*
		 * Could this slot have been previously retired?
		 * If so, then the server may be expecting seq_nr = 1!
		 */
661 662 663 664 665
		if (slot->seq_nr != 1) {
			slot->seq_nr = 1;
			goto retry_nowait;
		}
		break;
666 667 668
	case -NFS4ERR_SEQ_FALSE_RETRY:
		++slot->seq_nr;
		goto retry_nowait;
669 670
	default:
		/* Just update the slot sequence no. */
671
		++slot->seq_nr;
A
Andy Adamson 已提交
672 673 674 675
	}
out:
	/* The session may be reset by one of the error handlers. */
	dprintk("%s: Error %d free the slot \n", __func__, res->sr_status);
676
	nfs41_sequence_free_slot(res);
677
out_noaction:
678
	return ret;
679 680 681 682 683 684
retry_nowait:
	if (rpc_restart_call_prepare(task)) {
		task->tk_status = 0;
		ret = 0;
	}
	goto out;
685
out_retry:
686
	if (!rpc_restart_call(task))
687 688 689
		goto out;
	rpc_delay(task, NFS4_POLL_RETRY_MAX);
	return 0;
A
Andy Adamson 已提交
690
}
691
EXPORT_SYMBOL_GPL(nfs41_sequence_done);
A
Andy Adamson 已提交
692

P
Peng Tao 已提交
693
int nfs4_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res)
694
{
695
	if (res->sr_slot == NULL)
696
		return 1;
C
Chuck Lever 已提交
697 698
	if (!res->sr_slot->table->session)
		return nfs40_sequence_done(task, res);
699
	return nfs41_sequence_done(task, res);
700
}
P
Peng Tao 已提交
701
EXPORT_SYMBOL_GPL(nfs4_sequence_done);
702

A
Andy Adamson 已提交
703
int nfs41_setup_sequence(struct nfs4_session *session,
A
Andy Adamson 已提交
704 705 706 707
				struct nfs4_sequence_args *args,
				struct nfs4_sequence_res *res,
				struct rpc_task *task)
{
A
Andy Adamson 已提交
708 709 710 711
	struct nfs4_slot *slot;
	struct nfs4_slot_table *tbl;

	dprintk("--> %s\n", __func__);
A
Andy Adamson 已提交
712
	/* slot already allocated? */
713
	if (res->sr_slot != NULL)
714
		goto out_success;
A
Andy Adamson 已提交
715

A
Andy Adamson 已提交
716 717
	tbl = &session->fc_slot_table;

718 719
	task->tk_timeout = 0;

A
Andy Adamson 已提交
720
	spin_lock(&tbl->slot_tbl_lock);
721
	if (test_bit(NFS4_SLOT_TBL_DRAINING, &tbl->slot_tbl_state) &&
722
	    !args->sa_privileged) {
723 724
		/* The state manager will wait until the slot table is empty */
		dprintk("%s session is draining\n", __func__);
725
		goto out_sleep;
726 727
	}

728
	slot = nfs4_alloc_slot(tbl);
729 730 731 732
	if (IS_ERR(slot)) {
		/* If out of memory, try again in 1/4 second */
		if (slot == ERR_PTR(-ENOMEM))
			task->tk_timeout = HZ >> 2;
A
Andy Adamson 已提交
733
		dprintk("<-- %s: no free slots\n", __func__);
734
		goto out_sleep;
A
Andy Adamson 已提交
735 736 737
	}
	spin_unlock(&tbl->slot_tbl_lock);

738
	args->sa_slot = slot;
A
Andy Adamson 已提交
739

740
	dprintk("<-- %s slotid=%u seqid=%u\n", __func__,
741
			slot->slot_nr, slot->seq_nr);
A
Andy Adamson 已提交
742

743
	res->sr_slot = slot;
744
	res->sr_timestamp = jiffies;
745
	res->sr_status_flags = 0;
A
Andy Adamson 已提交
746 747 748 749 750
	/*
	 * sr_status is only set in decode_sequence, and so will remain
	 * set to 1 if an rpc level failure occurs.
	 */
	res->sr_status = 1;
751
	trace_nfs4_setup_sequence(session, args);
752 753
out_success:
	rpc_call_start(task);
A
Andy Adamson 已提交
754
	return 0;
755
out_sleep:
756 757
	/* Privileged tasks are queued with top priority */
	if (args->sa_privileged)
758 759 760 761
		rpc_sleep_on_priority(&tbl->slot_tbl_waitq, task,
				NULL, RPC_PRIORITY_PRIVILEGED);
	else
		rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
762 763
	spin_unlock(&tbl->slot_tbl_lock);
	return -EAGAIN;
A
Andy Adamson 已提交
764
}
A
Andy Adamson 已提交
765
EXPORT_SYMBOL_GPL(nfs41_setup_sequence);
A
Andy Adamson 已提交
766

767 768 769 770
static int nfs4_setup_sequence(const struct nfs_server *server,
			       struct nfs4_sequence_args *args,
			       struct nfs4_sequence_res *res,
			       struct rpc_task *task)
A
Andy Adamson 已提交
771
{
772
	struct nfs4_session *session = nfs4_get_session(server);
A
Andy Adamson 已提交
773 774
	int ret = 0;

C
Chuck Lever 已提交
775
	if (!session)
776 777
		return nfs40_setup_sequence(server->nfs_client->cl_slot_tbl,
					    args, res, task);
778

779
	dprintk("--> %s clp %p session %p sr_slot %u\n",
780
		__func__, session->clp, session, res->sr_slot ?
781
			res->sr_slot->slot_nr : NFS4_NO_SLOT);
A
Andy Adamson 已提交
782

783
	ret = nfs41_setup_sequence(session, args, res, task);
C
Chuck Lever 已提交
784

A
Andy Adamson 已提交
785 786 787 788 789 790
	dprintk("<-- %s status=%d\n", __func__, ret);
	return ret;
}

static void nfs41_call_sync_prepare(struct rpc_task *task, void *calldata)
{
791
	struct nfs4_call_sync_data *data = calldata;
792
	struct nfs4_session *session = nfs4_get_session(data->seq_server);
A
Andy Adamson 已提交
793

794 795
	dprintk("--> %s data->seq_server %p\n", __func__, data->seq_server);

796
	nfs41_setup_sequence(session, data->seq_args, data->seq_res, task);
A
Andy Adamson 已提交
797 798
}

A
Andy Adamson 已提交
799 800
static void nfs41_call_sync_done(struct rpc_task *task, void *calldata)
{
801
	struct nfs4_call_sync_data *data = calldata;
A
Andy Adamson 已提交
802

803
	nfs41_sequence_done(task, data->seq_res);
A
Andy Adamson 已提交
804 805
}

806
static const struct rpc_call_ops nfs41_call_sync_ops = {
A
Andy Adamson 已提交
807
	.rpc_call_prepare = nfs41_call_sync_prepare,
A
Andy Adamson 已提交
808
	.rpc_call_done = nfs41_call_sync_done,
A
Andy Adamson 已提交
809 810
};

C
Chuck Lever 已提交
811 812
#else	/* !CONFIG_NFS_V4_1 */

813 814 815 816 817
static int nfs4_setup_sequence(const struct nfs_server *server,
			       struct nfs4_sequence_args *args,
			       struct nfs4_sequence_res *res,
			       struct rpc_task *task)
{
818 819
	return nfs40_setup_sequence(server->nfs_client->cl_slot_tbl,
				    args, res, task);
820 821
}

P
Peng Tao 已提交
822 823
int nfs4_sequence_done(struct rpc_task *task,
		       struct nfs4_sequence_res *res)
824
{
C
Chuck Lever 已提交
825
	return nfs40_sequence_done(task, res);
826
}
P
Peng Tao 已提交
827
EXPORT_SYMBOL_GPL(nfs4_sequence_done);
C
Chuck Lever 已提交
828 829

#endif	/* !CONFIG_NFS_V4_1 */
830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848

static void nfs40_call_sync_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_call_sync_data *data = calldata;
	nfs4_setup_sequence(data->seq_server,
				data->seq_args, data->seq_res, task);
}

static void nfs40_call_sync_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_call_sync_data *data = calldata;
	nfs4_sequence_done(task, data->seq_res);
}

static const struct rpc_call_ops nfs40_call_sync_ops = {
	.rpc_call_prepare = nfs40_call_sync_prepare,
	.rpc_call_done = nfs40_call_sync_done,
};

849 850
static int nfs4_call_sync_sequence(struct rpc_clnt *clnt,
				   struct nfs_server *server,
A
Andy Adamson 已提交
851 852
				   struct rpc_message *msg,
				   struct nfs4_sequence_args *args,
853
				   struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
854 855 856
{
	int ret;
	struct rpc_task *task;
857
	struct nfs_client *clp = server->nfs_client;
858
	struct nfs4_call_sync_data data = {
859
		.seq_server = server,
A
Andy Adamson 已提交
860 861 862 863
		.seq_args = args,
		.seq_res = res,
	};
	struct rpc_task_setup task_setup = {
864
		.rpc_client = clnt,
A
Andy Adamson 已提交
865
		.rpc_message = msg,
866
		.callback_ops = clp->cl_mvops->call_sync_ops,
A
Andy Adamson 已提交
867 868 869 870 871 872 873 874 875 876 877 878 879
		.callback_data = &data
	};

	task = rpc_run_task(&task_setup);
	if (IS_ERR(task))
		ret = PTR_ERR(task);
	else {
		ret = task->tk_status;
		rpc_put_task(task);
	}
	return ret;
}

880 881
int nfs4_call_sync(struct rpc_clnt *clnt,
		   struct nfs_server *server,
882 883 884 885 886
		   struct rpc_message *msg,
		   struct nfs4_sequence_args *args,
		   struct nfs4_sequence_res *res,
		   int cache_reply)
{
887
	nfs4_init_sequence(args, res, cache_reply);
888
	return nfs4_call_sync_sequence(clnt, server, msg, args, res);
889
}
A
Andy Adamson 已提交
890

891
static void update_changeattr(struct inode *dir, struct nfs4_change_info *cinfo)
L
Linus Torvalds 已提交
892
{
893
	struct nfs_inode *nfsi = NFS_I(dir);
L
Linus Torvalds 已提交
894

895
	spin_lock(&dir->i_lock);
896
	nfsi->cache_validity |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA;
897
	if (!cinfo->atomic || cinfo->before != dir->i_version)
898
		nfs_force_lookup_revalidate(dir);
899
	dir->i_version = cinfo->after;
900
	nfsi->attr_gencount = nfs_inc_attr_generation_counter();
D
David Howells 已提交
901
	nfs_fscache_invalidate(dir);
902
	spin_unlock(&dir->i_lock);
L
Linus Torvalds 已提交
903 904
}

905
struct nfs4_opendata {
906
	struct kref kref;
907 908
	struct nfs_openargs o_arg;
	struct nfs_openres o_res;
909 910
	struct nfs_open_confirmargs c_arg;
	struct nfs_open_confirmres c_res;
911 912
	struct nfs4_string owner_name;
	struct nfs4_string group_name;
913
	struct nfs_fattr f_attr;
914
	struct nfs4_label *f_label;
915
	struct dentry *dir;
916
	struct dentry *dentry;
917
	struct nfs4_state_owner *owner;
918
	struct nfs4_state *state;
919
	struct iattr attrs;
920
	unsigned long timestamp;
921
	unsigned int rpc_done : 1;
922
	unsigned int file_created : 1;
923
	unsigned int is_recover : 1;
924 925
	int rpc_status;
	int cancelled;
926 927
};

928 929 930 931 932 933 934 935 936 937 938 939
static bool nfs4_clear_cap_atomic_open_v1(struct nfs_server *server,
		int err, struct nfs4_exception *exception)
{
	if (err != -EINVAL)
		return false;
	if (!(server->caps & NFS_CAP_ATOMIC_OPEN_V1))
		return false;
	server->caps &= ~NFS_CAP_ATOMIC_OPEN_V1;
	exception->retry = 1;
	return true;
}

940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964
static u32
nfs4_map_atomic_open_share(struct nfs_server *server,
		fmode_t fmode, int openflags)
{
	u32 res = 0;

	switch (fmode & (FMODE_READ | FMODE_WRITE)) {
	case FMODE_READ:
		res = NFS4_SHARE_ACCESS_READ;
		break;
	case FMODE_WRITE:
		res = NFS4_SHARE_ACCESS_WRITE;
		break;
	case FMODE_READ|FMODE_WRITE:
		res = NFS4_SHARE_ACCESS_BOTH;
	}
	if (!(server->caps & NFS_CAP_ATOMIC_OPEN_V1))
		goto out;
	/* Want no delegation if we're using O_DIRECT */
	if (openflags & O_DIRECT)
		res |= NFS4_SHARE_WANT_NO_DELEG;
out:
	return res;
}

965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981
static enum open_claim_type4
nfs4_map_atomic_open_claim(struct nfs_server *server,
		enum open_claim_type4 claim)
{
	if (server->caps & NFS_CAP_ATOMIC_OPEN_V1)
		return claim;
	switch (claim) {
	default:
		return claim;
	case NFS4_OPEN_CLAIM_FH:
		return NFS4_OPEN_CLAIM_NULL;
	case NFS4_OPEN_CLAIM_DELEG_CUR_FH:
		return NFS4_OPEN_CLAIM_DELEGATE_CUR;
	case NFS4_OPEN_CLAIM_DELEG_PREV_FH:
		return NFS4_OPEN_CLAIM_DELEGATE_PREV;
	}
}
982 983 984 985

static void nfs4_init_opendata_res(struct nfs4_opendata *p)
{
	p->o_res.f_attr = &p->f_attr;
986
	p->o_res.f_label = p->f_label;
987 988
	p->o_res.seqid = p->o_arg.seqid;
	p->c_res.seqid = p->c_arg.seqid;
989
	p->o_res.server = p->o_arg.server;
990
	p->o_res.access_request = p->o_arg.access;
991
	nfs_fattr_init(&p->f_attr);
992
	nfs_fattr_init_names(&p->f_attr, &p->owner_name, &p->group_name);
993 994
}

995
static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry,
996
		struct nfs4_state_owner *sp, fmode_t fmode, int flags,
997
		const struct iattr *attrs,
998
		struct nfs4_label *label,
999
		enum open_claim_type4 claim,
1000
		gfp_t gfp_mask)
1001
{
1002
	struct dentry *parent = dget_parent(dentry);
1003
	struct inode *dir = d_inode(parent);
1004
	struct nfs_server *server = NFS_SERVER(dir);
1005
	struct nfs_seqid *(*alloc_seqid)(struct nfs_seqid_counter *, gfp_t);
1006 1007
	struct nfs4_opendata *p;

1008
	p = kzalloc(sizeof(*p), gfp_mask);
1009 1010
	if (p == NULL)
		goto err;
1011 1012 1013 1014 1015

	p->f_label = nfs4_label_alloc(server, gfp_mask);
	if (IS_ERR(p->f_label))
		goto err_free_p;

1016 1017
	alloc_seqid = server->nfs_client->cl_mvops->alloc_seqid;
	p->o_arg.seqid = alloc_seqid(&sp->so_seqid, gfp_mask);
1018
	if (IS_ERR(p->o_arg.seqid))
1019
		goto err_free_label;
1020 1021
	nfs_sb_active(dentry->d_sb);
	p->dentry = dget(dentry);
1022 1023 1024
	p->dir = parent;
	p->owner = sp;
	atomic_inc(&sp->so_count);
1025 1026
	p->o_arg.open_flags = flags;
	p->o_arg.fmode = fmode & (FMODE_READ|FMODE_WRITE);
1027 1028
	p->o_arg.share_access = nfs4_map_atomic_open_share(server,
			fmode, flags);
1029 1030 1031 1032 1033 1034 1035 1036
	/* don't put an ACCESS op in OPEN compound if O_EXCL, because ACCESS
	 * will return permission denied for all bits until close */
	if (!(flags & O_EXCL)) {
		/* ask server to check for all possible rights as results
		 * are cached */
		p->o_arg.access = NFS4_ACCESS_READ | NFS4_ACCESS_MODIFY |
				  NFS4_ACCESS_EXTEND | NFS4_ACCESS_EXECUTE;
	}
1037
	p->o_arg.clientid = server->nfs_client->cl_clientid;
1038 1039
	p->o_arg.id.create_time = ktime_to_ns(sp->so_seqid.create_time);
	p->o_arg.id.uniquifier = sp->so_seqid.owner_id;
1040
	p->o_arg.name = &dentry->d_name;
1041
	p->o_arg.server = server;
1042
	p->o_arg.bitmask = nfs4_bitmask(server, label);
1043
	p->o_arg.open_bitmap = &nfs4_fattr_bitmap[0];
1044
	p->o_arg.label = label;
1045 1046
	p->o_arg.claim = nfs4_map_atomic_open_claim(server, claim);
	switch (p->o_arg.claim) {
1047 1048 1049 1050 1051 1052 1053 1054 1055
	case NFS4_OPEN_CLAIM_NULL:
	case NFS4_OPEN_CLAIM_DELEGATE_CUR:
	case NFS4_OPEN_CLAIM_DELEGATE_PREV:
		p->o_arg.fh = NFS_FH(dir);
		break;
	case NFS4_OPEN_CLAIM_PREVIOUS:
	case NFS4_OPEN_CLAIM_FH:
	case NFS4_OPEN_CLAIM_DELEG_CUR_FH:
	case NFS4_OPEN_CLAIM_DELEG_PREV_FH:
1056
		p->o_arg.fh = NFS_FH(d_inode(dentry));
1057
	}
1058
	if (attrs != NULL && attrs->ia_valid != 0) {
1059
		__u32 verf[2];
1060

1061 1062
		p->o_arg.u.attrs = &p->attrs;
		memcpy(&p->attrs, attrs, sizeof(p->attrs));
1063 1064 1065 1066 1067

		verf[0] = jiffies;
		verf[1] = current->pid;
		memcpy(p->o_arg.u.verifier.data, verf,
				sizeof(p->o_arg.u.verifier.data));
1068
	}
1069 1070 1071
	p->c_arg.fh = &p->o_res.fh;
	p->c_arg.stateid = &p->o_res.stateid;
	p->c_arg.seqid = p->o_arg.seqid;
1072
	nfs4_init_opendata_res(p);
1073
	kref_init(&p->kref);
1074
	return p;
1075 1076 1077 1078

err_free_label:
	nfs4_label_free(p->f_label);
err_free_p:
1079 1080 1081 1082 1083 1084
	kfree(p);
err:
	dput(parent);
	return NULL;
}

1085
static void nfs4_opendata_free(struct kref *kref)
1086
{
1087 1088
	struct nfs4_opendata *p = container_of(kref,
			struct nfs4_opendata, kref);
1089
	struct super_block *sb = p->dentry->d_sb;
1090 1091

	nfs_free_seqid(p->o_arg.seqid);
1092 1093
	if (p->state != NULL)
		nfs4_put_open_state(p->state);
1094
	nfs4_put_state_owner(p->owner);
1095 1096 1097

	nfs4_label_free(p->f_label);

1098
	dput(p->dir);
1099 1100
	dput(p->dentry);
	nfs_sb_deactive(sb);
1101
	nfs_fattr_free_names(&p->f_attr);
1102
	kfree(p->f_attr.mdsthreshold);
1103 1104 1105 1106 1107 1108 1109
	kfree(p);
}

static void nfs4_opendata_put(struct nfs4_opendata *p)
{
	if (p != NULL)
		kref_put(&p->kref, nfs4_opendata_free);
1110 1111
}

1112 1113 1114 1115 1116 1117 1118 1119
static int nfs4_wait_for_completion_rpc_task(struct rpc_task *task)
{
	int ret;

	ret = rpc_wait_for_completion_task(task);
	return ret;
}

1120
static int can_open_cached(struct nfs4_state *state, fmode_t mode, int open_mode)
1121 1122
{
	int ret = 0;
1123

1124
	if (open_mode & (O_EXCL|O_TRUNC))
1125 1126
		goto out;
	switch (mode & (FMODE_READ|FMODE_WRITE)) {
1127
		case FMODE_READ:
1128 1129
			ret |= test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0
				&& state->n_rdonly != 0;
1130 1131
			break;
		case FMODE_WRITE:
1132 1133
			ret |= test_bit(NFS_O_WRONLY_STATE, &state->flags) != 0
				&& state->n_wronly != 0;
1134 1135
			break;
		case FMODE_READ|FMODE_WRITE:
1136 1137
			ret |= test_bit(NFS_O_RDWR_STATE, &state->flags) != 0
				&& state->n_rdwr != 0;
1138
	}
1139
out:
1140 1141 1142
	return ret;
}

1143
static int can_open_delegated(struct nfs_delegation *delegation, fmode_t fmode)
1144
{
1145 1146
	if (delegation == NULL)
		return 0;
1147
	if ((delegation->type & fmode) != fmode)
1148
		return 0;
1149 1150
	if (test_bit(NFS_DELEGATION_RETURNING, &delegation->flags))
		return 0;
1151
	nfs_mark_delegation_referenced(delegation);
1152 1153 1154
	return 1;
}

1155
static void update_open_stateflags(struct nfs4_state *state, fmode_t fmode)
1156
{
1157
	switch (fmode) {
1158 1159 1160 1161 1162 1163 1164 1165 1166
		case FMODE_WRITE:
			state->n_wronly++;
			break;
		case FMODE_READ:
			state->n_rdonly++;
			break;
		case FMODE_READ|FMODE_WRITE:
			state->n_rdwr++;
	}
1167
	nfs4_state_set_mode_locked(state, state->state | fmode);
1168 1169
}

1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184
static void nfs_test_and_clear_all_open_stateid(struct nfs4_state *state)
{
	struct nfs_client *clp = state->owner->so_server->nfs_client;
	bool need_recover = false;

	if (test_and_clear_bit(NFS_O_RDONLY_STATE, &state->flags) && state->n_rdonly)
		need_recover = true;
	if (test_and_clear_bit(NFS_O_WRONLY_STATE, &state->flags) && state->n_wronly)
		need_recover = true;
	if (test_and_clear_bit(NFS_O_RDWR_STATE, &state->flags) && state->n_rdwr)
		need_recover = true;
	if (need_recover)
		nfs4_state_mark_reclaim_nograce(clp, state);
}

1185 1186 1187 1188 1189
static bool nfs_need_update_open_stateid(struct nfs4_state *state,
		nfs4_stateid *stateid)
{
	if (test_and_set_bit(NFS_OPEN_STATE, &state->flags) == 0)
		return true;
1190 1191
	if (!nfs4_stateid_match_other(stateid, &state->open_stateid)) {
		nfs_test_and_clear_all_open_stateid(state);
1192
		return true;
1193
	}
1194 1195 1196 1197 1198
	if (nfs4_stateid_is_newer(stateid, &state->open_stateid))
		return true;
	return false;
}

T
Trond Myklebust 已提交
1199 1200 1201 1202 1203 1204 1205 1206 1207 1208
static void nfs_resync_open_stateid_locked(struct nfs4_state *state)
{
	if (state->n_wronly)
		set_bit(NFS_O_WRONLY_STATE, &state->flags);
	if (state->n_rdonly)
		set_bit(NFS_O_RDONLY_STATE, &state->flags);
	if (state->n_rdwr)
		set_bit(NFS_O_RDWR_STATE, &state->flags);
}

1209 1210
static void nfs_clear_open_stateid_locked(struct nfs4_state *state,
		nfs4_stateid *stateid, fmode_t fmode)
1211
{
1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226
	clear_bit(NFS_O_RDWR_STATE, &state->flags);
	switch (fmode & (FMODE_READ|FMODE_WRITE)) {
	case FMODE_WRITE:
		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
		break;
	case FMODE_READ:
		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
		break;
	case 0:
		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
		clear_bit(NFS_OPEN_STATE, &state->flags);
	}
	if (stateid == NULL)
		return;
T
Trond Myklebust 已提交
1227 1228 1229 1230
	/* Handle races with OPEN */
	if (!nfs4_stateid_match_other(stateid, &state->open_stateid) ||
	    !nfs4_stateid_is_newer(stateid, &state->open_stateid)) {
		nfs_resync_open_stateid_locked(state);
1231
		return;
T
Trond Myklebust 已提交
1232
	}
1233
	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
1234 1235
		nfs4_stateid_copy(&state->stateid, stateid);
	nfs4_stateid_copy(&state->open_stateid, stateid);
1236 1237 1238 1239 1240 1241 1242
}

static void nfs_clear_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
{
	write_seqlock(&state->seqlock);
	nfs_clear_open_stateid_locked(state, stateid, fmode);
	write_sequnlock(&state->seqlock);
1243 1244
	if (test_bit(NFS_STATE_RECLAIM_NOGRACE, &state->flags))
		nfs4_schedule_state_manager(state->owner->so_server->nfs_client);
1245 1246
}

1247
static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
1248
{
1249
	switch (fmode) {
1250 1251 1252 1253 1254 1255 1256 1257 1258
		case FMODE_READ:
			set_bit(NFS_O_RDONLY_STATE, &state->flags);
			break;
		case FMODE_WRITE:
			set_bit(NFS_O_WRONLY_STATE, &state->flags);
			break;
		case FMODE_READ|FMODE_WRITE:
			set_bit(NFS_O_RDWR_STATE, &state->flags);
	}
1259 1260 1261 1262 1263
	if (!nfs_need_update_open_stateid(state, stateid))
		return;
	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
		nfs4_stateid_copy(&state->stateid, stateid);
	nfs4_stateid_copy(&state->open_stateid, stateid);
1264 1265
}

1266
static void __update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, const nfs4_stateid *deleg_stateid, fmode_t fmode)
L
Linus Torvalds 已提交
1267
{
1268 1269 1270 1271 1272
	/*
	 * Protect the call to nfs4_state_set_mode_locked and
	 * serialise the stateid update
	 */
	write_seqlock(&state->seqlock);
1273
	if (deleg_stateid != NULL) {
1274
		nfs4_stateid_copy(&state->stateid, deleg_stateid);
1275 1276 1277
		set_bit(NFS_DELEGATED_STATE, &state->flags);
	}
	if (open_stateid != NULL)
1278
		nfs_set_open_stateid_locked(state, open_stateid, fmode);
1279 1280
	write_sequnlock(&state->seqlock);
	spin_lock(&state->owner->so_lock);
1281
	update_open_stateflags(state, fmode);
1282
	spin_unlock(&state->owner->so_lock);
L
Linus Torvalds 已提交
1283 1284
}

1285
static int update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *delegation, fmode_t fmode)
1286 1287 1288 1289 1290
{
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *deleg_cur;
	int ret = 0;

1291
	fmode &= (FMODE_READ|FMODE_WRITE);
1292 1293 1294 1295 1296 1297 1298

	rcu_read_lock();
	deleg_cur = rcu_dereference(nfsi->delegation);
	if (deleg_cur == NULL)
		goto no_delegation;

	spin_lock(&deleg_cur->lock);
1299
	if (rcu_dereference(nfsi->delegation) != deleg_cur ||
1300
	   test_bit(NFS_DELEGATION_RETURNING, &deleg_cur->flags) ||
1301
	    (deleg_cur->type & fmode) != fmode)
1302 1303 1304 1305
		goto no_delegation_unlock;

	if (delegation == NULL)
		delegation = &deleg_cur->stateid;
1306
	else if (!nfs4_stateid_match(&deleg_cur->stateid, delegation))
1307 1308
		goto no_delegation_unlock;

1309
	nfs_mark_delegation_referenced(deleg_cur);
1310
	__update_open_stateid(state, open_stateid, &deleg_cur->stateid, fmode);
1311 1312 1313 1314 1315 1316 1317
	ret = 1;
no_delegation_unlock:
	spin_unlock(&deleg_cur->lock);
no_delegation:
	rcu_read_unlock();

	if (!ret && open_stateid != NULL) {
1318
		__update_open_stateid(state, open_stateid, NULL, fmode);
1319 1320
		ret = 1;
	}
1321 1322
	if (test_bit(NFS_STATE_RECLAIM_NOGRACE, &state->flags))
		nfs4_schedule_state_manager(state->owner->so_server->nfs_client);
1323 1324 1325 1326

	return ret;
}

1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343
static bool nfs4_update_lock_stateid(struct nfs4_lock_state *lsp,
		const nfs4_stateid *stateid)
{
	struct nfs4_state *state = lsp->ls_state;
	bool ret = false;

	spin_lock(&state->state_lock);
	if (!nfs4_stateid_match_other(stateid, &lsp->ls_stateid))
		goto out_noupdate;
	if (!nfs4_stateid_is_newer(stateid, &lsp->ls_stateid))
		goto out_noupdate;
	nfs4_stateid_copy(&lsp->ls_stateid, stateid);
	ret = true;
out_noupdate:
	spin_unlock(&state->state_lock);
	return ret;
}
1344

1345
static void nfs4_return_incompatible_delegation(struct inode *inode, fmode_t fmode)
1346 1347 1348 1349 1350
{
	struct nfs_delegation *delegation;

	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(inode)->delegation);
1351
	if (delegation == NULL || (delegation->type & fmode) == fmode) {
1352 1353 1354 1355
		rcu_read_unlock();
		return;
	}
	rcu_read_unlock();
1356
	nfs4_inode_return_delegation(inode);
1357 1358
}

1359
static struct nfs4_state *nfs4_try_open_cached(struct nfs4_opendata *opendata)
1360 1361 1362 1363
{
	struct nfs4_state *state = opendata->state;
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *delegation;
1364
	int open_mode = opendata->o_arg.open_flags;
1365
	fmode_t fmode = opendata->o_arg.fmode;
1366 1367 1368 1369
	nfs4_stateid stateid;
	int ret = -EAGAIN;

	for (;;) {
1370
		spin_lock(&state->owner->so_lock);
1371
		if (can_open_cached(state, fmode, open_mode)) {
1372
			update_open_stateflags(state, fmode);
1373
			spin_unlock(&state->owner->so_lock);
1374
			goto out_return_state;
1375
		}
1376
		spin_unlock(&state->owner->so_lock);
1377 1378
		rcu_read_lock();
		delegation = rcu_dereference(nfsi->delegation);
1379
		if (!can_open_delegated(delegation, fmode)) {
1380
			rcu_read_unlock();
1381
			break;
1382
		}
1383
		/* Save the delegation */
1384
		nfs4_stateid_copy(&stateid, &delegation->stateid);
1385
		rcu_read_unlock();
1386
		nfs_release_seqid(opendata->o_arg.seqid);
1387 1388 1389 1390 1391
		if (!opendata->is_recover) {
			ret = nfs_may_open(state->inode, state->owner->so_cred, open_mode);
			if (ret != 0)
				goto out;
		}
1392
		ret = -EAGAIN;
1393 1394

		/* Try to update the stateid using the delegation */
1395
		if (update_open_stateid(state, NULL, &stateid, fmode))
1396
			goto out_return_state;
1397 1398 1399 1400 1401 1402 1403 1404
	}
out:
	return ERR_PTR(ret);
out_return_state:
	atomic_inc(&state->count);
	return state;
}

1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442
static void
nfs4_opendata_check_deleg(struct nfs4_opendata *data, struct nfs4_state *state)
{
	struct nfs_client *clp = NFS_SERVER(state->inode)->nfs_client;
	struct nfs_delegation *delegation;
	int delegation_flags = 0;

	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
	if (delegation)
		delegation_flags = delegation->flags;
	rcu_read_unlock();
	if (data->o_arg.claim == NFS4_OPEN_CLAIM_DELEGATE_CUR) {
		pr_err_ratelimited("NFS: Broken NFSv4 server %s is "
				   "returning a delegation for "
				   "OPEN(CLAIM_DELEGATE_CUR)\n",
				   clp->cl_hostname);
	} else if ((delegation_flags & 1UL<<NFS_DELEGATION_NEED_RECLAIM) == 0)
		nfs_inode_set_delegation(state->inode,
					 data->owner->so_cred,
					 &data->o_res);
	else
		nfs_inode_reclaim_delegation(state->inode,
					     data->owner->so_cred,
					     &data->o_res);
}

/*
 * Check the inode attributes against the CLAIM_PREVIOUS returned attributes
 * and update the nfs4_state.
 */
static struct nfs4_state *
_nfs4_opendata_reclaim_to_nfs4_state(struct nfs4_opendata *data)
{
	struct inode *inode = data->state->inode;
	struct nfs4_state *state = data->state;
	int ret;

1443 1444 1445 1446 1447 1448 1449
	if (!data->rpc_done) {
		if (data->rpc_status) {
			ret = data->rpc_status;
			goto err;
		}
		/* cached opens have already been processed */
		goto update;
1450 1451 1452 1453 1454 1455 1456 1457
	}

	ret = nfs_refresh_inode(inode, &data->f_attr);
	if (ret)
		goto err;

	if (data->o_res.delegation_type != 0)
		nfs4_opendata_check_deleg(data, state);
1458
update:
1459 1460
	update_open_stateid(state, &data->o_res.stateid, NULL,
			    data->o_arg.fmode);
1461
	atomic_inc(&state->count);
1462 1463 1464 1465 1466 1467 1468 1469 1470

	return state;
err:
	return ERR_PTR(ret);

}

static struct nfs4_state *
_nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data)
1471 1472 1473
{
	struct inode *inode;
	struct nfs4_state *state = NULL;
1474
	int ret;
1475

1476
	if (!data->rpc_done) {
1477
		state = nfs4_try_open_cached(data);
1478 1479 1480
		goto out;
	}

1481
	ret = -EAGAIN;
1482
	if (!(data->f_attr.valid & NFS_ATTR_FATTR))
1483
		goto err;
1484
	inode = nfs_fhget(data->dir->d_sb, &data->o_res.fh, &data->f_attr, data->f_label);
1485
	ret = PTR_ERR(inode);
1486
	if (IS_ERR(inode))
1487 1488
		goto err;
	ret = -ENOMEM;
1489 1490
	state = nfs4_get_open_state(inode, data->owner);
	if (state == NULL)
1491
		goto err_put_inode;
1492 1493
	if (data->o_res.delegation_type != 0)
		nfs4_opendata_check_deleg(data, state);
1494
	update_open_stateid(state, &data->o_res.stateid, NULL,
1495
			data->o_arg.fmode);
1496
	iput(inode);
1497
out:
1498
	nfs_release_seqid(data->o_arg.seqid);
1499
	return state;
1500 1501 1502 1503
err_put_inode:
	iput(inode);
err:
	return ERR_PTR(ret);
1504 1505
}

1506 1507 1508 1509 1510 1511 1512 1513
static struct nfs4_state *
nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data)
{
	if (data->o_arg.claim == NFS4_OPEN_CLAIM_PREVIOUS)
		return _nfs4_opendata_reclaim_to_nfs4_state(data);
	return _nfs4_opendata_to_nfs4_state(data);
}

1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530
static struct nfs_open_context *nfs4_state_find_open_context(struct nfs4_state *state)
{
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_open_context *ctx;

	spin_lock(&state->inode->i_lock);
	list_for_each_entry(ctx, &nfsi->open_files, list) {
		if (ctx->state != state)
			continue;
		get_nfs_open_context(ctx);
		spin_unlock(&state->inode->i_lock);
		return ctx;
	}
	spin_unlock(&state->inode->i_lock);
	return ERR_PTR(-ENOENT);
}

1531 1532
static struct nfs4_opendata *nfs4_open_recoverdata_alloc(struct nfs_open_context *ctx,
		struct nfs4_state *state, enum open_claim_type4 claim)
T
Trond Myklebust 已提交
1533 1534 1535
{
	struct nfs4_opendata *opendata;

1536
	opendata = nfs4_opendata_alloc(ctx->dentry, state->owner, 0, 0,
1537
			NULL, NULL, claim, GFP_NOFS);
T
Trond Myklebust 已提交
1538 1539 1540 1541 1542 1543 1544
	if (opendata == NULL)
		return ERR_PTR(-ENOMEM);
	opendata->state = state;
	atomic_inc(&state->count);
	return opendata;
}

1545
static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, fmode_t fmode, struct nfs4_state **res)
1546
{
1547
	struct nfs4_state *newstate;
1548 1549
	int ret;

1550 1551
	opendata->o_arg.open_flags = 0;
	opendata->o_arg.fmode = fmode;
1552 1553 1554
	opendata->o_arg.share_access = nfs4_map_atomic_open_share(
			NFS_SB(opendata->dentry->d_sb),
			fmode, 0);
1555 1556 1557
	memset(&opendata->o_res, 0, sizeof(opendata->o_res));
	memset(&opendata->c_res, 0, sizeof(opendata->c_res));
	nfs4_init_opendata_res(opendata);
1558
	ret = _nfs4_recover_proc_open(opendata);
1559 1560
	if (ret != 0)
		return ret; 
1561
	newstate = nfs4_opendata_to_nfs4_state(opendata);
1562 1563
	if (IS_ERR(newstate))
		return PTR_ERR(newstate);
1564
	nfs4_close_state(newstate, fmode);
1565
	*res = newstate;
1566 1567 1568 1569 1570 1571 1572 1573
	return 0;
}

static int nfs4_open_recover(struct nfs4_opendata *opendata, struct nfs4_state *state)
{
	struct nfs4_state *newstate;
	int ret;

1574 1575 1576 1577
	/* Don't trigger recovery in nfs_test_and_clear_all_open_stateid */
	clear_bit(NFS_O_RDWR_STATE, &state->flags);
	clear_bit(NFS_O_WRONLY_STATE, &state->flags);
	clear_bit(NFS_O_RDONLY_STATE, &state->flags);
1578
	/* memory barrier prior to reading state->n_* */
1579
	clear_bit(NFS_DELEGATED_STATE, &state->flags);
1580
	clear_bit(NFS_OPEN_STATE, &state->flags);
1581 1582
	smp_rmb();
	if (state->n_rdwr != 0) {
1583
		ret = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE, &newstate);
1584 1585
		if (ret != 0)
			return ret;
1586 1587
		if (newstate != state)
			return -ESTALE;
1588 1589
	}
	if (state->n_wronly != 0) {
1590
		ret = nfs4_open_recover_helper(opendata, FMODE_WRITE, &newstate);
1591 1592
		if (ret != 0)
			return ret;
1593 1594
		if (newstate != state)
			return -ESTALE;
1595 1596
	}
	if (state->n_rdonly != 0) {
1597
		ret = nfs4_open_recover_helper(opendata, FMODE_READ, &newstate);
1598 1599
		if (ret != 0)
			return ret;
1600 1601
		if (newstate != state)
			return -ESTALE;
1602
	}
1603 1604 1605 1606 1607
	/*
	 * We may have performed cached opens for all three recoveries.
	 * Check if we need to update the current stateid.
	 */
	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0 &&
1608
	    !nfs4_stateid_match(&state->stateid, &state->open_stateid)) {
1609
		write_seqlock(&state->seqlock);
1610
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
1611
			nfs4_stateid_copy(&state->stateid, &state->open_stateid);
1612
		write_sequnlock(&state->seqlock);
1613
	}
1614 1615 1616
	return 0;
}

L
Linus Torvalds 已提交
1617 1618 1619 1620
/*
 * OPEN_RECLAIM:
 * 	reclaim state on the server after a reboot.
 */
1621
static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1622
{
1623
	struct nfs_delegation *delegation;
1624
	struct nfs4_opendata *opendata;
1625
	fmode_t delegation_type = 0;
L
Linus Torvalds 已提交
1626 1627
	int status;

1628 1629
	opendata = nfs4_open_recoverdata_alloc(ctx, state,
			NFS4_OPEN_CLAIM_PREVIOUS);
T
Trond Myklebust 已提交
1630 1631
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1632 1633
	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
1634
	if (delegation != NULL && test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) != 0)
1635
		delegation_type = delegation->type;
1636
	rcu_read_unlock();
1637 1638
	opendata->o_arg.u.delegation_type = delegation_type;
	status = nfs4_open_recover(opendata, state);
1639
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
1640 1641 1642
	return status;
}

1643
static int nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1644 1645 1646 1647 1648
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;
	do {
1649
		err = _nfs4_do_open_reclaim(ctx, state);
1650
		trace_nfs4_open_reclaim(ctx, 0, err);
1651 1652
		if (nfs4_clear_cap_atomic_open_v1(server, err, &exception))
			continue;
1653
		if (err != -NFS4ERR_DELAY)
1654 1655
			break;
		nfs4_handle_exception(server, err, &exception);
L
Linus Torvalds 已提交
1656 1657 1658 1659
	} while (exception.retry);
	return err;
}

1660 1661 1662 1663 1664 1665 1666
static int nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
	struct nfs_open_context *ctx;
	int ret;

	ctx = nfs4_state_find_open_context(state);
	if (IS_ERR(ctx))
1667
		return -EAGAIN;
1668
	ret = nfs4_do_open_reclaim(ctx, state);
1669 1670 1671 1672
	put_nfs_open_context(ctx);
	return ret;
}

1673
static int nfs4_handle_delegation_recall_error(struct nfs_server *server, struct nfs4_state *state, const nfs4_stateid *stateid, int err)
L
Linus Torvalds 已提交
1674
{
1675 1676 1677 1678 1679 1680
	switch (err) {
		default:
			printk(KERN_ERR "NFS: %s: unhandled error "
					"%d.\n", __func__, err);
		case 0:
		case -ENOENT:
1681
		case -EAGAIN:
1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698
		case -ESTALE:
			break;
		case -NFS4ERR_BADSESSION:
		case -NFS4ERR_BADSLOT:
		case -NFS4ERR_BAD_HIGH_SLOT:
		case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
		case -NFS4ERR_DEADSESSION:
			set_bit(NFS_DELEGATED_STATE, &state->flags);
			nfs4_schedule_session_recovery(server->nfs_client->cl_session, err);
			return -EAGAIN;
		case -NFS4ERR_STALE_CLIENTID:
		case -NFS4ERR_STALE_STATEID:
			set_bit(NFS_DELEGATED_STATE, &state->flags);
		case -NFS4ERR_EXPIRED:
			/* Don't recall a delegation if it was lost */
			nfs4_schedule_lease_recovery(server->nfs_client);
			return -EAGAIN;
1699 1700 1701
		case -NFS4ERR_MOVED:
			nfs4_schedule_migration_recovery(server);
			return -EAGAIN;
1702 1703 1704
		case -NFS4ERR_LEASE_MOVED:
			nfs4_schedule_lease_moved_recovery(server->nfs_client);
			return -EAGAIN;
1705 1706 1707
		case -NFS4ERR_DELEG_REVOKED:
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
1708
		case -NFS4ERR_OPENMODE:
1709 1710 1711
			nfs_inode_find_state_and_recover(state->inode,
					stateid);
			nfs4_schedule_stateid_recovery(server, state);
1712
			return -EAGAIN;
1713 1714 1715 1716 1717
		case -NFS4ERR_DELAY:
		case -NFS4ERR_GRACE:
			set_bit(NFS_DELEGATED_STATE, &state->flags);
			ssleep(1);
			return -EAGAIN;
1718 1719 1720 1721
		case -ENOMEM:
		case -NFS4ERR_DENIED:
			/* kill_proc(fl->fl_pid, SIGLOST, 1); */
			return 0;
1722
	}
L
Linus Torvalds 已提交
1723 1724 1725
	return err;
}

1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741
int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid)
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_opendata *opendata;
	int err;

	opendata = nfs4_open_recoverdata_alloc(ctx, state,
			NFS4_OPEN_CLAIM_DELEG_CUR_FH);
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
	nfs4_stateid_copy(&opendata->o_arg.u.delegation, stateid);
	err = nfs4_open_recover(opendata, state);
	nfs4_opendata_put(opendata);
	return nfs4_handle_delegation_recall_error(server, state, stateid, err);
}

1742 1743 1744 1745
static void nfs4_open_confirm_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;

1746 1747
	nfs40_setup_sequence(data->o_arg.server->nfs_client->cl_slot_tbl,
			     &data->c_arg.seq_args, &data->c_res.seq_res, task);
1748 1749
}

1750 1751 1752 1753
static void nfs4_open_confirm_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;

1754
	nfs40_sequence_done(task, &data->c_res.seq_res);
1755

1756
	data->rpc_status = task->tk_status;
1757
	if (data->rpc_status == 0) {
1758
		nfs4_stateid_copy(&data->o_res.stateid, &data->c_res.stateid);
1759
		nfs_confirm_seqid(&data->owner->so_seqid, 0);
1760
		renew_lease(data->o_res.server, data->timestamp);
1761
		data->rpc_done = 1;
1762
	}
1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773
}

static void nfs4_open_confirm_release(void *calldata)
{
	struct nfs4_opendata *data = calldata;
	struct nfs4_state *state = NULL;

	/* If this request hasn't been cancelled, do nothing */
	if (data->cancelled == 0)
		goto out_free;
	/* In case of error, no cleanup! */
1774
	if (!data->rpc_done)
1775 1776
		goto out_free;
	state = nfs4_opendata_to_nfs4_state(data);
1777
	if (!IS_ERR(state))
1778
		nfs4_close_state(state, data->o_arg.fmode);
1779
out_free:
1780
	nfs4_opendata_put(data);
1781 1782 1783
}

static const struct rpc_call_ops nfs4_open_confirm_ops = {
1784
	.rpc_call_prepare = nfs4_open_confirm_prepare,
1785 1786 1787 1788 1789 1790 1791 1792 1793
	.rpc_call_done = nfs4_open_confirm_done,
	.rpc_release = nfs4_open_confirm_release,
};

/*
 * Note: On error, nfs4_proc_open_confirm will free the struct nfs4_opendata
 */
static int _nfs4_proc_open_confirm(struct nfs4_opendata *data)
{
1794
	struct nfs_server *server = NFS_SERVER(d_inode(data->dir));
1795
	struct rpc_task *task;
1796 1797 1798 1799 1800 1801
	struct  rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_CONFIRM],
		.rpc_argp = &data->c_arg,
		.rpc_resp = &data->c_res,
		.rpc_cred = data->owner->so_cred,
	};
T
Trond Myklebust 已提交
1802 1803
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1804
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1805 1806
		.callback_ops = &nfs4_open_confirm_ops,
		.callback_data = data,
1807
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1808 1809
		.flags = RPC_TASK_ASYNC,
	};
L
Linus Torvalds 已提交
1810 1811
	int status;

1812
	nfs4_init_sequence(&data->c_arg.seq_args, &data->c_res.seq_res, 1);
1813
	kref_get(&data->kref);
1814 1815
	data->rpc_done = 0;
	data->rpc_status = 0;
1816
	data->timestamp = jiffies;
T
Trond Myklebust 已提交
1817
	task = rpc_run_task(&task_setup_data);
1818
	if (IS_ERR(task))
1819 1820 1821 1822 1823 1824 1825
		return PTR_ERR(task);
	status = nfs4_wait_for_completion_rpc_task(task);
	if (status != 0) {
		data->cancelled = 1;
		smp_wmb();
	} else
		status = data->rpc_status;
1826
	rpc_put_task(task);
L
Linus Torvalds 已提交
1827 1828 1829
	return status;
}

1830
static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
L
Linus Torvalds 已提交
1831
{
1832 1833
	struct nfs4_opendata *data = calldata;
	struct nfs4_state_owner *sp = data->owner;
1834
	struct nfs_client *clp = sp->so_server->nfs_client;
1835

1836
	if (nfs_wait_on_sequence(data->o_arg.seqid, task) != 0)
1837
		goto out_wait;
1838 1839 1840 1841 1842 1843 1844
	/*
	 * Check if we still need to send an OPEN call, or if we can use
	 * a delegation instead.
	 */
	if (data->state != NULL) {
		struct nfs_delegation *delegation;

1845
		if (can_open_cached(data->state, data->o_arg.fmode, data->o_arg.open_flags))
1846
			goto out_no_action;
1847 1848
		rcu_read_lock();
		delegation = rcu_dereference(NFS_I(data->state->inode)->delegation);
1849
		if (data->o_arg.claim != NFS4_OPEN_CLAIM_DELEGATE_CUR &&
1850
		    data->o_arg.claim != NFS4_OPEN_CLAIM_DELEG_CUR_FH &&
1851 1852
		    can_open_delegated(delegation, data->o_arg.fmode))
			goto unlock_no_action;
1853 1854
		rcu_read_unlock();
	}
1855
	/* Update client id. */
1856
	data->o_arg.clientid = clp->cl_clientid;
1857 1858 1859 1860
	switch (data->o_arg.claim) {
	case NFS4_OPEN_CLAIM_PREVIOUS:
	case NFS4_OPEN_CLAIM_DELEG_CUR_FH:
	case NFS4_OPEN_CLAIM_DELEG_PREV_FH:
1861
		data->o_arg.open_bitmap = &nfs4_open_noattr_bitmap[0];
1862 1863
	case NFS4_OPEN_CLAIM_FH:
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR];
T
Trond Myklebust 已提交
1864 1865
		nfs_copy_fh(&data->o_res.fh, data->o_arg.fh);
	}
1866
	data->timestamp = jiffies;
1867
	if (nfs4_setup_sequence(data->o_arg.server,
1868
				&data->o_arg.seq_args,
1869 1870 1871
				&data->o_res.seq_res,
				task) != 0)
		nfs_release_seqid(data->o_arg.seqid);
1872 1873 1874 1875 1876 1877 1878 1879 1880 1881

	/* Set the create mode (note dependency on the session type) */
	data->o_arg.createmode = NFS4_CREATE_UNCHECKED;
	if (data->o_arg.open_flags & O_EXCL) {
		data->o_arg.createmode = NFS4_CREATE_EXCLUSIVE;
		if (nfs4_has_persistent_session(clp))
			data->o_arg.createmode = NFS4_CREATE_GUARDED;
		else if (clp->cl_mvops->minor_version > 0)
			data->o_arg.createmode = NFS4_CREATE_EXCLUSIVE4_1;
	}
1882
	return;
1883 1884
unlock_no_action:
	rcu_read_unlock();
1885 1886
out_no_action:
	task->tk_action = NULL;
1887
out_wait:
1888
	nfs4_sequence_done(task, &data->o_res.seq_res);
1889 1890
}

1891 1892 1893
static void nfs4_open_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;
L
Linus Torvalds 已提交
1894

1895
	data->rpc_status = task->tk_status;
1896

1897 1898
	if (!nfs4_sequence_done(task, &data->o_res.seq_res))
		return;
1899

1900
	if (task->tk_status == 0) {
1901 1902
		if (data->o_res.f_attr->valid & NFS_ATTR_FATTR_TYPE) {
			switch (data->o_res.f_attr->mode & S_IFMT) {
1903 1904 1905
			case S_IFREG:
				break;
			case S_IFLNK:
1906
				data->rpc_status = -ELOOP;
1907 1908
				break;
			case S_IFDIR:
1909
				data->rpc_status = -EISDIR;
1910 1911
				break;
			default:
1912
				data->rpc_status = -ENOTDIR;
1913
			}
1914
		}
1915
		renew_lease(data->o_res.server, data->timestamp);
1916 1917
		if (!(data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM))
			nfs_confirm_seqid(&data->owner->so_seqid, 0);
1918
	}
1919
	data->rpc_done = 1;
1920
}
1921

1922 1923 1924 1925 1926 1927 1928 1929 1930
static void nfs4_open_release(void *calldata)
{
	struct nfs4_opendata *data = calldata;
	struct nfs4_state *state = NULL;

	/* If this request hasn't been cancelled, do nothing */
	if (data->cancelled == 0)
		goto out_free;
	/* In case of error, no cleanup! */
1931
	if (data->rpc_status != 0 || !data->rpc_done)
1932 1933 1934 1935 1936
		goto out_free;
	/* In case we need an open_confirm, no cleanup! */
	if (data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM)
		goto out_free;
	state = nfs4_opendata_to_nfs4_state(data);
1937
	if (!IS_ERR(state))
1938
		nfs4_close_state(state, data->o_arg.fmode);
1939
out_free:
1940
	nfs4_opendata_put(data);
1941 1942 1943 1944 1945 1946 1947 1948
}

static const struct rpc_call_ops nfs4_open_ops = {
	.rpc_call_prepare = nfs4_open_prepare,
	.rpc_call_done = nfs4_open_done,
	.rpc_release = nfs4_open_release,
};

1949
static int nfs4_run_open_task(struct nfs4_opendata *data, int isrecover)
1950
{
1951
	struct inode *dir = d_inode(data->dir);
1952 1953 1954 1955
	struct nfs_server *server = NFS_SERVER(dir);
	struct nfs_openargs *o_arg = &data->o_arg;
	struct nfs_openres *o_res = &data->o_res;
	struct rpc_task *task;
1956 1957 1958 1959 1960 1961
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN],
		.rpc_argp = o_arg,
		.rpc_resp = o_res,
		.rpc_cred = data->owner->so_cred,
	};
T
Trond Myklebust 已提交
1962 1963
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1964
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1965 1966
		.callback_ops = &nfs4_open_ops,
		.callback_data = data,
1967
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1968 1969
		.flags = RPC_TASK_ASYNC,
	};
1970 1971
	int status;

1972
	nfs4_init_sequence(&o_arg->seq_args, &o_res->seq_res, 1);
1973
	kref_get(&data->kref);
1974 1975
	data->rpc_done = 0;
	data->rpc_status = 0;
1976
	data->cancelled = 0;
1977 1978
	data->is_recover = 0;
	if (isrecover) {
1979
		nfs4_set_sequence_privileged(&o_arg->seq_args);
1980 1981
		data->is_recover = 1;
	}
T
Trond Myklebust 已提交
1982
	task = rpc_run_task(&task_setup_data);
1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997
        if (IS_ERR(task))
                return PTR_ERR(task);
        status = nfs4_wait_for_completion_rpc_task(task);
        if (status != 0) {
                data->cancelled = 1;
                smp_wmb();
        } else
                status = data->rpc_status;
        rpc_put_task(task);

	return status;
}

static int _nfs4_recover_proc_open(struct nfs4_opendata *data)
{
1998
	struct inode *dir = d_inode(data->dir);
1999 2000 2001 2002 2003 2004 2005
	struct nfs_openres *o_res = &data->o_res;
        int status;

	status = nfs4_run_open_task(data, 1);
	if (status != 0 || !data->rpc_done)
		return status;

2006 2007
	nfs_fattr_map_and_free_names(NFS_SERVER(dir), &data->f_attr);

2008 2009 2010 2011 2012 2013 2014 2015 2016
	if (o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
		status = _nfs4_proc_open_confirm(data);
		if (status != 0)
			return status;
	}

	return status;
}

2017 2018 2019 2020 2021 2022 2023 2024
/*
 * Additional permission checks in order to distinguish between an
 * open for read, and an open for execute. This works around the
 * fact that NFSv4 OPEN treats read and execute permissions as being
 * the same.
 * Note that in the non-execute case, we want to turn off permission
 * checking if we just created a new file (POSIX open() semantics).
 */
2025 2026
static int nfs4_opendata_access(struct rpc_cred *cred,
				struct nfs4_opendata *opendata,
2027 2028
				struct nfs4_state *state, fmode_t fmode,
				int openflags)
2029 2030 2031 2032 2033 2034 2035 2036 2037 2038
{
	struct nfs_access_entry cache;
	u32 mask;

	/* access call failed or for some reason the server doesn't
	 * support any access modes -- defer access call until later */
	if (opendata->o_res.access_supported == 0)
		return 0;

	mask = 0;
2039 2040 2041 2042
	/*
	 * Use openflags to check for exec, because fmode won't
	 * always have FMODE_EXEC set when file open for exec.
	 */
2043 2044 2045
	if (openflags & __FMODE_EXEC) {
		/* ONLY check for exec rights */
		mask = MAY_EXEC;
2046
	} else if ((fmode & FMODE_READ) && !opendata->file_created)
2047
		mask = MAY_READ;
2048 2049 2050 2051 2052 2053

	cache.cred = cred;
	cache.jiffies = jiffies;
	nfs_access_set_mask(&cache, opendata->o_res.access_result);
	nfs_access_add_cache(state->inode, &cache);

2054
	if ((mask & ~cache.mask & (MAY_READ | MAY_EXEC)) == 0)
2055 2056 2057 2058
		return 0;

	/* even though OPEN succeeded, access is denied. Close the file */
	nfs4_close_state(state, fmode);
2059
	return -EACCES;
2060 2061
}

2062 2063 2064 2065 2066
/*
 * Note: On error, nfs4_proc_open will free the struct nfs4_opendata
 */
static int _nfs4_proc_open(struct nfs4_opendata *data)
{
2067
	struct inode *dir = d_inode(data->dir);
2068 2069 2070 2071 2072 2073
	struct nfs_server *server = NFS_SERVER(dir);
	struct nfs_openargs *o_arg = &data->o_arg;
	struct nfs_openres *o_res = &data->o_res;
	int status;

	status = nfs4_run_open_task(data, 0);
2074 2075 2076 2077 2078 2079
	if (!data->rpc_done)
		return status;
	if (status != 0) {
		if (status == -NFS4ERR_BADNAME &&
				!(o_arg->open_flags & O_CREAT))
			return -ENOENT;
2080
		return status;
2081
	}
2082

2083 2084
	nfs_fattr_map_and_free_names(server, &data->f_attr);

2085
	if (o_arg->open_flags & O_CREAT) {
2086
		update_changeattr(dir, &o_res->cinfo);
2087 2088 2089 2090 2091
		if (o_arg->open_flags & O_EXCL)
			data->file_created = 1;
		else if (o_res->cinfo.before != o_res->cinfo.after)
			data->file_created = 1;
	}
T
Trond Myklebust 已提交
2092 2093
	if ((o_res->rflags & NFS4_OPEN_RESULT_LOCKTYPE_POSIX) == 0)
		server->caps &= ~NFS_CAP_POSIX_LOCK;
L
Linus Torvalds 已提交
2094
	if(o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
2095
		status = _nfs4_proc_open_confirm(data);
L
Linus Torvalds 已提交
2096
		if (status != 0)
2097
			return status;
L
Linus Torvalds 已提交
2098 2099
	}
	if (!(o_res->f_attr->valid & NFS_ATTR_FATTR))
2100
		nfs4_proc_getattr(server, &o_res->fh, o_res->f_attr, o_res->f_label);
2101
	return 0;
L
Linus Torvalds 已提交
2102 2103
}

A
Andy Adamson 已提交
2104 2105 2106 2107 2108
static int nfs4_recover_expired_lease(struct nfs_server *server)
{
	return nfs4_client_recover_expired_lease(server->nfs_client);
}

L
Linus Torvalds 已提交
2109 2110 2111 2112 2113
/*
 * OPEN_EXPIRED:
 * 	reclaim state on the server after a network partition.
 * 	Assumes caller holds the appropriate lock
 */
2114
static int _nfs4_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
2115
{
2116
	struct nfs4_opendata *opendata;
2117
	int ret;
L
Linus Torvalds 已提交
2118

2119
	opendata = nfs4_open_recoverdata_alloc(ctx, state,
2120
			NFS4_OPEN_CLAIM_FH);
T
Trond Myklebust 已提交
2121 2122
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
2123
	ret = nfs4_open_recover(opendata, state);
2124
	if (ret == -ESTALE)
2125
		d_drop(ctx->dentry);
2126
	nfs4_opendata_put(opendata);
2127
	return ret;
L
Linus Torvalds 已提交
2128 2129
}

2130
static int nfs4_do_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
2131
{
2132
	struct nfs_server *server = NFS_SERVER(state->inode);
2133 2134 2135 2136
	struct nfs4_exception exception = { };
	int err;

	do {
2137
		err = _nfs4_open_expired(ctx, state);
2138
		trace_nfs4_open_expired(ctx, 0, err);
2139 2140
		if (nfs4_clear_cap_atomic_open_v1(server, err, &exception))
			continue;
2141 2142 2143 2144 2145 2146 2147 2148
		switch (err) {
		default:
			goto out;
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			nfs4_handle_exception(server, err, &exception);
			err = 0;
		}
2149
	} while (exception.retry);
2150
out:
2151 2152 2153
	return err;
}

L
Linus Torvalds 已提交
2154 2155 2156
static int nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
	struct nfs_open_context *ctx;
2157
	int ret;
L
Linus Torvalds 已提交
2158

2159 2160
	ctx = nfs4_state_find_open_context(state);
	if (IS_ERR(ctx))
2161
		return -EAGAIN;
2162
	ret = nfs4_do_open_expired(ctx, state);
2163 2164
	put_nfs_open_context(ctx);
	return ret;
L
Linus Torvalds 已提交
2165 2166
}

2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188
static void nfs_finish_clear_delegation_stateid(struct nfs4_state *state)
{
	nfs_remove_bad_delegation(state->inode);
	write_seqlock(&state->seqlock);
	nfs4_stateid_copy(&state->stateid, &state->open_stateid);
	write_sequnlock(&state->seqlock);
	clear_bit(NFS_DELEGATED_STATE, &state->flags);
}

static void nfs40_clear_delegation_stateid(struct nfs4_state *state)
{
	if (rcu_access_pointer(NFS_I(state->inode)->delegation) != NULL)
		nfs_finish_clear_delegation_stateid(state);
}

static int nfs40_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
	/* NFSv4.0 doesn't allow for delegation recovery on open expire */
	nfs40_clear_delegation_stateid(state);
	return nfs4_open_expired(sp, state);
}

2189
#if defined(CONFIG_NFS_V4_1)
2190
static void nfs41_check_delegation_stateid(struct nfs4_state *state)
2191 2192
{
	struct nfs_server *server = NFS_SERVER(state->inode);
2193
	nfs4_stateid stateid;
2194
	struct nfs_delegation *delegation;
2195 2196
	struct rpc_cred *cred;
	int status;
2197

2198 2199 2200
	/* Get the delegation credential for use by test/free_stateid */
	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
2201
	if (delegation == NULL) {
2202
		rcu_read_unlock();
2203 2204 2205 2206 2207 2208 2209 2210
		return;
	}

	nfs4_stateid_copy(&stateid, &delegation->stateid);
	cred = get_rpccred(delegation->cred);
	rcu_read_unlock();
	status = nfs41_test_stateid(server, &stateid, cred);
	trace_nfs4_test_delegation_stateid(state, NULL, status);
2211

2212 2213 2214 2215
	if (status != NFS_OK) {
		/* Free the stateid unless the server explicitly
		 * informs us the stateid is unrecognized. */
		if (status != -NFS4ERR_BAD_STATEID)
2216 2217
			nfs41_free_stateid(server, &stateid, cred);
		nfs_finish_clear_delegation_stateid(state);
2218
	}
2219

2220
	put_rpccred(cred);
2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233
}

/**
 * nfs41_check_open_stateid - possibly free an open stateid
 *
 * @state: NFSv4 state for an inode
 *
 * Returns NFS_OK if recovery for this stateid is now finished.
 * Otherwise a negative NFS4ERR value is returned.
 */
static int nfs41_check_open_stateid(struct nfs4_state *state)
{
	struct nfs_server *server = NFS_SERVER(state->inode);
2234
	nfs4_stateid *stateid = &state->open_stateid;
2235
	struct rpc_cred *cred = state->owner->so_cred;
2236 2237 2238 2239 2240 2241 2242 2243
	int status;

	/* If a state reset has been done, test_stateid is unneeded */
	if ((test_bit(NFS_O_RDONLY_STATE, &state->flags) == 0) &&
	    (test_bit(NFS_O_WRONLY_STATE, &state->flags) == 0) &&
	    (test_bit(NFS_O_RDWR_STATE, &state->flags) == 0))
		return -NFS4ERR_BAD_STATEID;

2244
	status = nfs41_test_stateid(server, stateid, cred);
2245
	trace_nfs4_test_open_stateid(state, NULL, status);
2246 2247 2248 2249
	if (status != NFS_OK) {
		/* Free the stateid unless the server explicitly
		 * informs us the stateid is unrecognized. */
		if (status != -NFS4ERR_BAD_STATEID)
2250
			nfs41_free_stateid(server, stateid, cred);
2251 2252 2253 2254

		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
		clear_bit(NFS_O_RDWR_STATE, &state->flags);
2255
		clear_bit(NFS_OPEN_STATE, &state->flags);
2256 2257 2258 2259 2260 2261
	}
	return status;
}

static int nfs41_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
2262
	int status;
2263

2264
	nfs41_check_delegation_stateid(state);
2265
	status = nfs41_check_open_stateid(state);
2266 2267 2268
	if (status != NFS_OK)
		status = nfs4_open_expired(sp, state);
	return status;
2269 2270 2271
}
#endif

2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287
/*
 * on an EXCLUSIVE create, the server should send back a bitmask with FATTR4-*
 * fields corresponding to attributes that were used to store the verifier.
 * Make sure we clobber those fields in the later setattr call
 */
static inline void nfs4_exclusive_attrset(struct nfs4_opendata *opendata, struct iattr *sattr)
{
	if ((opendata->o_res.attrset[1] & FATTR4_WORD1_TIME_ACCESS) &&
	    !(sattr->ia_valid & ATTR_ATIME_SET))
		sattr->ia_valid |= ATTR_ATIME;

	if ((opendata->o_res.attrset[1] & FATTR4_WORD1_TIME_MODIFY) &&
	    !(sattr->ia_valid & ATTR_MTIME_SET))
		sattr->ia_valid |= ATTR_MTIME;
}

2288 2289 2290
static int _nfs4_open_and_get_state(struct nfs4_opendata *opendata,
		fmode_t fmode,
		int flags,
2291
		struct nfs_open_context *ctx)
2292 2293 2294
{
	struct nfs4_state_owner *sp = opendata->owner;
	struct nfs_server *server = sp->so_server;
2295
	struct dentry *dentry;
2296 2297 2298 2299 2300 2301 2302
	struct nfs4_state *state;
	unsigned int seq;
	int ret;

	seq = raw_seqcount_begin(&sp->so_reclaim_seqcount);

	ret = _nfs4_proc_open(opendata);
2303
	if (ret != 0)
2304 2305 2306 2307 2308 2309 2310 2311 2312
		goto out;

	state = nfs4_opendata_to_nfs4_state(opendata);
	ret = PTR_ERR(state);
	if (IS_ERR(state))
		goto out;
	if (server->caps & NFS_CAP_POSIX_LOCK)
		set_bit(NFS_STATE_POSIX_LOCKS, &state->flags);

2313
	dentry = opendata->dentry;
2314
	if (d_really_is_negative(dentry)) {
2315 2316 2317 2318 2319 2320 2321 2322 2323 2324
		/* FIXME: Is this d_drop() ever needed? */
		d_drop(dentry);
		dentry = d_add_unique(dentry, igrab(state->inode));
		if (dentry == NULL) {
			dentry = opendata->dentry;
		} else if (dentry != ctx->dentry) {
			dput(ctx->dentry);
			ctx->dentry = dget(dentry);
		}
		nfs_set_verifier(dentry,
2325
				nfs_save_change_attribute(d_inode(opendata->dir)));
2326 2327
	}

2328 2329 2330 2331
	ret = nfs4_opendata_access(sp->so_cred, opendata, state, fmode, flags);
	if (ret != 0)
		goto out;

2332
	ctx->state = state;
2333
	if (d_inode(dentry) == state->inode) {
2334 2335 2336 2337
		nfs_inode_attach_open_context(ctx);
		if (read_seqcount_retry(&sp->so_reclaim_seqcount, seq))
			nfs4_schedule_stateid_recovery(server, state);
	}
2338 2339 2340 2341
out:
	return ret;
}

L
Linus Torvalds 已提交
2342
/*
2343
 * Returns a referenced nfs4_state
L
Linus Torvalds 已提交
2344
 */
2345
static int _nfs4_do_open(struct inode *dir,
2346
			struct nfs_open_context *ctx,
2347 2348
			int flags,
			struct iattr *sattr,
2349 2350
			struct nfs4_label *label,
			int *opened)
L
Linus Torvalds 已提交
2351 2352 2353 2354
{
	struct nfs4_state_owner  *sp;
	struct nfs4_state     *state = NULL;
	struct nfs_server       *server = NFS_SERVER(dir);
2355
	struct nfs4_opendata *opendata;
2356 2357 2358 2359
	struct dentry *dentry = ctx->dentry;
	struct rpc_cred *cred = ctx->cred;
	struct nfs4_threshold **ctx_th = &ctx->mdsthreshold;
	fmode_t fmode = ctx->mode & (FMODE_READ|FMODE_WRITE|FMODE_EXEC);
2360
	enum open_claim_type4 claim = NFS4_OPEN_CLAIM_NULL;
2361
	struct nfs4_label *olabel = NULL;
2362
	int status;
L
Linus Torvalds 已提交
2363 2364 2365

	/* Protect against reboot recovery conflicts */
	status = -ENOMEM;
2366 2367
	sp = nfs4_get_state_owner(server, cred, GFP_KERNEL);
	if (sp == NULL) {
L
Linus Torvalds 已提交
2368 2369 2370
		dprintk("nfs4_do_open: nfs4_get_state_owner failed!\n");
		goto out_err;
	}
2371 2372
	status = nfs4_recover_expired_lease(server);
	if (status != 0)
2373
		goto err_put_state_owner;
2374 2375
	if (d_really_is_positive(dentry))
		nfs4_return_incompatible_delegation(d_inode(dentry), fmode);
2376
	status = -ENOMEM;
2377
	if (d_really_is_positive(dentry))
2378
		claim = NFS4_OPEN_CLAIM_FH;
2379
	opendata = nfs4_opendata_alloc(dentry, sp, fmode, flags, sattr,
2380
			label, claim, GFP_KERNEL);
2381
	if (opendata == NULL)
T
Trond Myklebust 已提交
2382
		goto err_put_state_owner;
L
Linus Torvalds 已提交
2383

2384 2385 2386 2387 2388 2389 2390 2391
	if (label) {
		olabel = nfs4_label_alloc(server, GFP_KERNEL);
		if (IS_ERR(olabel)) {
			status = PTR_ERR(olabel);
			goto err_opendata_put;
		}
	}

2392 2393 2394 2395 2396 2397
	if (server->attr_bitmask[2] & FATTR4_WORD2_MDSTHRESHOLD) {
		if (!opendata->f_attr.mdsthreshold) {
			opendata->f_attr.mdsthreshold = pnfs_mdsthreshold_alloc();
			if (!opendata->f_attr.mdsthreshold)
				goto err_free_label;
		}
2398
		opendata->o_arg.open_bitmap = &nfs4_pnfs_open_bitmap[0];
2399
	}
2400 2401
	if (d_really_is_positive(dentry))
		opendata->state = nfs4_get_open_state(d_inode(dentry), sp);
2402

2403
	status = _nfs4_open_and_get_state(opendata, fmode, flags, ctx);
2404
	if (status != 0)
2405
		goto err_free_label;
2406
	state = ctx->state;
2407

2408 2409
	if ((opendata->o_arg.open_flags & O_EXCL) &&
	    (opendata->o_arg.createmode != NFS4_CREATE_GUARDED)) {
2410 2411 2412 2413 2414
		nfs4_exclusive_attrset(opendata, sattr);

		nfs_fattr_init(opendata->o_res.f_attr);
		status = nfs4_do_setattr(state->inode, cred,
				opendata->o_res.f_attr, sattr,
2415 2416
				state, label, olabel);
		if (status == 0) {
2417 2418
			nfs_setattr_update_inode(state->inode, sattr,
					opendata->o_res.f_attr);
2419
			nfs_setsecurity(state->inode, opendata->o_res.f_attr, olabel);
2420
		}
2421
	}
2422 2423
	if (opendata->file_created)
		*opened |= FILE_CREATED;
2424

2425
	if (pnfs_use_threshold(ctx_th, opendata->f_attr.mdsthreshold, server)) {
2426
		*ctx_th = opendata->f_attr.mdsthreshold;
2427 2428
		opendata->f_attr.mdsthreshold = NULL;
	}
2429

2430 2431
	nfs4_label_free(olabel);

2432
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
2433 2434
	nfs4_put_state_owner(sp);
	return 0;
2435 2436
err_free_label:
	nfs4_label_free(olabel);
2437 2438
err_opendata_put:
	nfs4_opendata_put(opendata);
2439 2440
err_put_state_owner:
	nfs4_put_state_owner(sp);
L
Linus Torvalds 已提交
2441 2442 2443 2444 2445
out_err:
	return status;
}


2446
static struct nfs4_state *nfs4_do_open(struct inode *dir,
2447
					struct nfs_open_context *ctx,
2448 2449
					int flags,
					struct iattr *sattr,
2450 2451
					struct nfs4_label *label,
					int *opened)
L
Linus Torvalds 已提交
2452
{
2453
	struct nfs_server *server = NFS_SERVER(dir);
L
Linus Torvalds 已提交
2454 2455 2456 2457 2458
	struct nfs4_exception exception = { };
	struct nfs4_state *res;
	int status;

	do {
2459
		status = _nfs4_do_open(dir, ctx, flags, sattr, label, opened);
2460
		res = ctx->state;
2461
		trace_nfs4_open_file(ctx, flags, status);
L
Linus Torvalds 已提交
2462 2463 2464 2465 2466 2467 2468 2469
		if (status == 0)
			break;
		/* NOTE: BAD_SEQID means the server and client disagree about the
		 * book-keeping w.r.t. state-changing operations
		 * (OPEN/CLOSE/LOCK/LOCKU...)
		 * It is actually a sign of a bug on the client or on the server.
		 *
		 * If we receive a BAD_SEQID error in the particular case of
2470
		 * doing an OPEN, we assume that nfs_increment_open_seqid() will
L
Linus Torvalds 已提交
2471 2472 2473 2474 2475
		 * have unhashed the old state_owner for us, and that we can
		 * therefore safely retry using a new one. We should still warn
		 * the user though...
		 */
		if (status == -NFS4ERR_BAD_SEQID) {
2476
			pr_warn_ratelimited("NFS: v4 server %s "
2477 2478
					" returned a bad sequence-id error!\n",
					NFS_SERVER(dir)->nfs_client->cl_hostname);
L
Linus Torvalds 已提交
2479 2480 2481
			exception.retry = 1;
			continue;
		}
2482 2483 2484 2485 2486 2487 2488 2489 2490 2491
		/*
		 * BAD_STATEID on OPEN means that the server cancelled our
		 * state before it received the OPEN_CONFIRM.
		 * Recover by retrying the request as per the discussion
		 * on Page 181 of RFC3530.
		 */
		if (status == -NFS4ERR_BAD_STATEID) {
			exception.retry = 1;
			continue;
		}
2492 2493 2494 2495 2496
		if (status == -EAGAIN) {
			/* We must have found a delegation */
			exception.retry = 1;
			continue;
		}
2497 2498 2499
		if (nfs4_clear_cap_atomic_open_v1(server, status, &exception))
			continue;
		res = ERR_PTR(nfs4_handle_exception(server,
L
Linus Torvalds 已提交
2500 2501 2502 2503 2504
					status, &exception));
	} while (exception.retry);
	return res;
}

2505 2506
static int _nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
			    struct nfs_fattr *fattr, struct iattr *sattr,
2507 2508
			    struct nfs4_state *state, struct nfs4_label *ilabel,
			    struct nfs4_label *olabel)
L
Linus Torvalds 已提交
2509
{
2510
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
2511
        struct nfs_setattrargs  arg = {
2512
                .fh             = NFS_FH(inode),
L
Linus Torvalds 已提交
2513 2514 2515
                .iap            = sattr,
		.server		= server,
		.bitmask = server->attr_bitmask,
2516
		.label		= ilabel,
L
Linus Torvalds 已提交
2517 2518 2519
        };
        struct nfs_setattrres  res = {
		.fattr		= fattr,
2520
		.label		= olabel,
L
Linus Torvalds 已提交
2521 2522 2523
		.server		= server,
        };
        struct rpc_message msg = {
2524 2525 2526 2527
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETATTR],
		.rpc_argp	= &arg,
		.rpc_resp	= &res,
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
2528
        };
2529
	unsigned long timestamp = jiffies;
2530 2531
	fmode_t fmode;
	bool truncate;
2532
	int status;
L
Linus Torvalds 已提交
2533

2534 2535 2536 2537
	arg.bitmask = nfs4_bitmask(server, ilabel);
	if (ilabel)
		arg.bitmask = nfs4_bitmask(server, olabel);

2538
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
2539

2540 2541 2542 2543 2544 2545
	/* Servers should only apply open mode checks for file size changes */
	truncate = (sattr->ia_valid & ATTR_SIZE) ? true : false;
	fmode = truncate ? FMODE_WRITE : FMODE_READ;

	if (nfs4_copy_delegation_stateid(&arg.stateid, inode, fmode)) {
		/* Use that stateid */
2546
	} else if (truncate && state != NULL) {
2547 2548 2549 2550
		struct nfs_lockowner lockowner = {
			.l_owner = current->files,
			.l_pid = current->tgid,
		};
2551 2552 2553 2554 2555
		if (!nfs4_valid_open_stateid(state))
			return -EBADF;
		if (nfs4_select_rw_stateid(&arg.stateid, state, FMODE_WRITE,
				&lockowner) == -EIO)
			return -EBADF;
2556
	} else
2557
		nfs4_stateid_copy(&arg.stateid, &zero_stateid);
L
Linus Torvalds 已提交
2558

2559
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
2560 2561
	if (status == 0 && state != NULL)
		renew_lease(server, timestamp);
2562
	return status;
L
Linus Torvalds 已提交
2563 2564
}

2565 2566
static int nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
			   struct nfs_fattr *fattr, struct iattr *sattr,
2567 2568
			   struct nfs4_state *state, struct nfs4_label *ilabel,
			   struct nfs4_label *olabel)
L
Linus Torvalds 已提交
2569
{
2570
	struct nfs_server *server = NFS_SERVER(inode);
2571 2572
	struct nfs4_exception exception = {
		.state = state,
2573
		.inode = inode,
2574
	};
L
Linus Torvalds 已提交
2575 2576
	int err;
	do {
2577
		err = _nfs4_do_setattr(inode, cred, fattr, sattr, state, ilabel, olabel);
2578
		trace_nfs4_setattr(inode, err);
2579 2580
		switch (err) {
		case -NFS4ERR_OPENMODE:
2581 2582 2583 2584 2585 2586 2587
			if (!(sattr->ia_valid & ATTR_SIZE)) {
				pr_warn_once("NFSv4: server %s is incorrectly "
						"applying open mode checks to "
						"a SETATTR that is not "
						"changing file size.\n",
						server->nfs_client->cl_hostname);
			}
2588 2589 2590 2591 2592 2593 2594 2595
			if (state && !(state->state & FMODE_WRITE)) {
				err = -EBADF;
				if (sattr->ia_valid & ATTR_OPEN)
					err = -EACCES;
				goto out;
			}
		}
		err = nfs4_handle_exception(server, err, &exception);
L
Linus Torvalds 已提交
2596
	} while (exception.retry);
2597
out:
L
Linus Torvalds 已提交
2598 2599 2600 2601 2602 2603 2604 2605
	return err;
}

struct nfs4_closedata {
	struct inode *inode;
	struct nfs4_state *state;
	struct nfs_closeargs arg;
	struct nfs_closeres res;
2606
	struct nfs_fattr fattr;
2607
	unsigned long timestamp;
F
Fred Isaman 已提交
2608 2609
	bool roc;
	u32 roc_barrier;
L
Linus Torvalds 已提交
2610 2611
};

2612
static void nfs4_free_closedata(void *data)
2613
{
2614 2615
	struct nfs4_closedata *calldata = data;
	struct nfs4_state_owner *sp = calldata->state->owner;
2616
	struct super_block *sb = calldata->state->inode->i_sb;
2617

F
Fred Isaman 已提交
2618 2619
	if (calldata->roc)
		pnfs_roc_release(calldata->state->inode);
2620 2621 2622
	nfs4_put_open_state(calldata->state);
	nfs_free_seqid(calldata->arg.seqid);
	nfs4_put_state_owner(sp);
2623
	nfs_sb_deactive(sb);
2624 2625 2626
	kfree(calldata);
}

2627
static void nfs4_close_done(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
2628
{
2629
	struct nfs4_closedata *calldata = data;
L
Linus Torvalds 已提交
2630 2631
	struct nfs4_state *state = calldata->state;
	struct nfs_server *server = NFS_SERVER(calldata->inode);
2632
	nfs4_stateid *res_stateid = NULL;
L
Linus Torvalds 已提交
2633

2634
	dprintk("%s: begin!\n", __func__);
2635 2636
	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
		return;
2637
	trace_nfs4_close(state, &calldata->arg, &calldata->res, task->tk_status);
L
Linus Torvalds 已提交
2638 2639 2640 2641 2642
        /* hmm. we are done with the inode, and in the process of freeing
	 * the state_owner. we keep this around to process errors
	 */
	switch (task->tk_status) {
		case 0:
2643 2644
			res_stateid = &calldata->res.stateid;
			if (calldata->arg.fmode == 0 && calldata->roc)
F
Fred Isaman 已提交
2645 2646
				pnfs_roc_set_barrier(state->inode,
						     calldata->roc_barrier);
2647
			renew_lease(server, calldata->timestamp);
2648
			break;
2649
		case -NFS4ERR_ADMIN_REVOKED:
L
Linus Torvalds 已提交
2650
		case -NFS4ERR_STALE_STATEID:
2651 2652
		case -NFS4ERR_OLD_STATEID:
		case -NFS4ERR_BAD_STATEID:
L
Linus Torvalds 已提交
2653
		case -NFS4ERR_EXPIRED:
2654
			if (!nfs4_stateid_match(&calldata->arg.stateid,
2655
						&state->open_stateid)) {
2656 2657 2658
				rpc_restart_call_prepare(task);
				goto out_release;
			}
2659
			if (calldata->arg.fmode == 0)
2660
				break;
L
Linus Torvalds 已提交
2661
		default:
2662
			if (nfs4_async_handle_error(task, server, state, NULL) == -EAGAIN) {
2663
				rpc_restart_call_prepare(task);
2664 2665
				goto out_release;
			}
L
Linus Torvalds 已提交
2666
	}
2667
	nfs_clear_open_stateid(state, res_stateid, calldata->arg.fmode);
2668
out_release:
2669
	nfs_release_seqid(calldata->arg.seqid);
2670
	nfs_refresh_inode(calldata->inode, calldata->res.fattr);
2671
	dprintk("%s: done, ret = %d!\n", __func__, task->tk_status);
L
Linus Torvalds 已提交
2672 2673
}

T
Trond Myklebust 已提交
2674
static void nfs4_close_prepare(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
2675
{
T
Trond Myklebust 已提交
2676
	struct nfs4_closedata *calldata = data;
2677
	struct nfs4_state *state = calldata->state;
2678
	struct inode *inode = calldata->inode;
2679
	bool is_rdonly, is_wronly, is_rdwr;
2680
	int call_close = 0;
2681

2682
	dprintk("%s: begin!\n", __func__);
2683
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
2684
		goto out_wait;
2685

2686
	task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_DOWNGRADE];
2687
	spin_lock(&state->owner->so_lock);
2688 2689 2690
	is_rdwr = test_bit(NFS_O_RDWR_STATE, &state->flags);
	is_rdonly = test_bit(NFS_O_RDONLY_STATE, &state->flags);
	is_wronly = test_bit(NFS_O_WRONLY_STATE, &state->flags);
2691
	nfs4_stateid_copy(&calldata->arg.stateid, &state->open_stateid);
2692
	/* Calculate the change in open mode */
2693
	calldata->arg.fmode = 0;
2694
	if (state->n_rdwr == 0) {
2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708
		if (state->n_rdonly == 0)
			call_close |= is_rdonly;
		else if (is_rdonly)
			calldata->arg.fmode |= FMODE_READ;
		if (state->n_wronly == 0)
			call_close |= is_wronly;
		else if (is_wronly)
			calldata->arg.fmode |= FMODE_WRITE;
	} else if (is_rdwr)
		calldata->arg.fmode |= FMODE_READ|FMODE_WRITE;

	if (calldata->arg.fmode == 0)
		call_close |= is_rdwr;

2709 2710
	if (!nfs4_valid_open_stateid(state))
		call_close = 0;
2711
	spin_unlock(&state->owner->so_lock);
2712 2713

	if (!call_close) {
2714
		/* Note: exit _without_ calling nfs4_close_done */
2715
		goto out_no_action;
2716
	}
2717

F
Fred Isaman 已提交
2718
	if (calldata->arg.fmode == 0) {
2719
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE];
F
Fred Isaman 已提交
2720
		if (calldata->roc &&
2721 2722
		    pnfs_roc_drain(inode, &calldata->roc_barrier, task)) {
			nfs_release_seqid(calldata->arg.seqid);
2723
			goto out_wait;
2724
		    }
F
Fred Isaman 已提交
2725
	}
2726 2727 2728
	calldata->arg.share_access =
		nfs4_map_atomic_open_share(NFS_SERVER(inode),
				calldata->arg.fmode, 0);
2729

2730
	nfs_fattr_init(calldata->res.fattr);
2731
	calldata->timestamp = jiffies;
2732
	if (nfs4_setup_sequence(NFS_SERVER(inode),
2733 2734
				&calldata->arg.seq_args,
				&calldata->res.seq_res,
2735 2736
				task) != 0)
		nfs_release_seqid(calldata->arg.seqid);
2737
	dprintk("%s: done!\n", __func__);
2738 2739 2740 2741 2742
	return;
out_no_action:
	task->tk_action = NULL;
out_wait:
	nfs4_sequence_done(task, &calldata->res.seq_res);
L
Linus Torvalds 已提交
2743 2744
}

2745
static const struct rpc_call_ops nfs4_close_ops = {
T
Trond Myklebust 已提交
2746
	.rpc_call_prepare = nfs4_close_prepare,
2747 2748 2749 2750
	.rpc_call_done = nfs4_close_done,
	.rpc_release = nfs4_free_closedata,
};

P
Peng Tao 已提交
2751 2752
static bool nfs4_roc(struct inode *inode)
{
2753
	if (!nfs_have_layout(inode))
P
Peng Tao 已提交
2754 2755 2756 2757
		return false;
	return pnfs_roc(inode);
}

L
Linus Torvalds 已提交
2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768
/* 
 * It is possible for data to be read/written from a mem-mapped file 
 * after the sys_close call (which hits the vfs layer as a flush).
 * This means that we can't safely call nfsv4 close on a file until 
 * the inode is cleared. This in turn means that we are not good
 * NFSv4 citizens - we do not indicate to the server to update the file's 
 * share state even when we are done with one of the three share 
 * stateid's in the inode.
 *
 * NOTE: Caller must be holding the sp->so_owner semaphore!
 */
2769
int nfs4_do_close(struct nfs4_state *state, gfp_t gfp_mask, int wait)
L
Linus Torvalds 已提交
2770
{
2771
	struct nfs_server *server = NFS_SERVER(state->inode);
2772
	struct nfs_seqid *(*alloc_seqid)(struct nfs_seqid_counter *, gfp_t);
L
Linus Torvalds 已提交
2773
	struct nfs4_closedata *calldata;
2774 2775
	struct nfs4_state_owner *sp = state->owner;
	struct rpc_task *task;
2776 2777 2778 2779
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
2780 2781
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
2782
		.rpc_message = &msg,
T
Trond Myklebust 已提交
2783
		.callback_ops = &nfs4_close_ops,
2784
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
2785 2786
		.flags = RPC_TASK_ASYNC,
	};
2787
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2788

2789 2790 2791
	nfs4_state_protect(server->nfs_client, NFS_SP4_MACH_CRED_CLEANUP,
		&task_setup_data.rpc_client, &msg);

2792
	calldata = kzalloc(sizeof(*calldata), gfp_mask);
L
Linus Torvalds 已提交
2793
	if (calldata == NULL)
2794
		goto out;
2795
	nfs4_init_sequence(&calldata->arg.seq_args, &calldata->res.seq_res, 1);
2796
	calldata->inode = state->inode;
L
Linus Torvalds 已提交
2797
	calldata->state = state;
2798
	calldata->arg.fh = NFS_FH(state->inode);
L
Linus Torvalds 已提交
2799
	/* Serialization for the sequence id */
2800 2801
	alloc_seqid = server->nfs_client->cl_mvops->alloc_seqid;
	calldata->arg.seqid = alloc_seqid(&state->owner->so_seqid, gfp_mask);
2802
	if (IS_ERR(calldata->arg.seqid))
2803
		goto out_free_calldata;
2804
	calldata->arg.fmode = 0;
2805
	calldata->arg.bitmask = server->cache_consistency_bitmask;
2806
	calldata->res.fattr = &calldata->fattr;
2807
	calldata->res.seqid = calldata->arg.seqid;
2808
	calldata->res.server = server;
P
Peng Tao 已提交
2809
	calldata->roc = nfs4_roc(state->inode);
2810
	nfs_sb_active(calldata->inode->i_sb);
2811

2812 2813
	msg.rpc_argp = &calldata->arg;
	msg.rpc_resp = &calldata->res;
T
Trond Myklebust 已提交
2814 2815
	task_setup_data.callback_data = calldata;
	task = rpc_run_task(&task_setup_data);
2816 2817
	if (IS_ERR(task))
		return PTR_ERR(task);
2818 2819 2820
	status = 0;
	if (wait)
		status = rpc_wait_for_completion_task(task);
2821
	rpc_put_task(task);
2822
	return status;
2823 2824 2825
out_free_calldata:
	kfree(calldata);
out:
2826 2827
	nfs4_put_open_state(state);
	nfs4_put_state_owner(sp);
2828
	return status;
L
Linus Torvalds 已提交
2829 2830
}

2831
static struct inode *
2832 2833
nfs4_atomic_open(struct inode *dir, struct nfs_open_context *ctx,
		int open_flags, struct iattr *attr, int *opened)
L
Linus Torvalds 已提交
2834 2835
{
	struct nfs4_state *state;
2836 2837 2838
	struct nfs4_label l = {0, 0, 0, NULL}, *label = NULL;

	label = nfs4_label_init_security(dir, ctx->dentry, attr, &l);
L
Linus Torvalds 已提交
2839

T
Trond Myklebust 已提交
2840
	/* Protect against concurrent sillydeletes */
2841
	state = nfs4_do_open(dir, ctx, open_flags, attr, label, opened);
2842 2843 2844

	nfs4_label_release_security(label);

2845 2846
	if (IS_ERR(state))
		return ERR_CAST(state);
2847
	return state->inode;
L
Linus Torvalds 已提交
2848 2849
}

2850
static void nfs4_close_context(struct nfs_open_context *ctx, int is_sync)
T
Trond Myklebust 已提交
2851 2852 2853 2854
{
	if (ctx->state == NULL)
		return;
	if (is_sync)
2855
		nfs4_close_sync(ctx->state, ctx->mode);
T
Trond Myklebust 已提交
2856
	else
2857
		nfs4_close_state(ctx->state, ctx->mode);
T
Trond Myklebust 已提交
2858
}
L
Linus Torvalds 已提交
2859

2860 2861
#define FATTR4_WORD1_NFS40_MASK (2*FATTR4_WORD1_MOUNTED_ON_FILEID - 1UL)
#define FATTR4_WORD2_NFS41_MASK (2*FATTR4_WORD2_SUPPATTR_EXCLCREAT - 1UL)
2862
#define FATTR4_WORD2_NFS42_MASK (2*FATTR4_WORD2_SECURITY_LABEL - 1UL)
2863

L
Linus Torvalds 已提交
2864 2865
static int _nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
{
B
Benny Halevy 已提交
2866 2867 2868
	struct nfs4_server_caps_arg args = {
		.fhandle = fhandle,
	};
L
Linus Torvalds 已提交
2869 2870 2871
	struct nfs4_server_caps_res res = {};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SERVER_CAPS],
B
Benny Halevy 已提交
2872
		.rpc_argp = &args,
L
Linus Torvalds 已提交
2873 2874 2875 2876
		.rpc_resp = &res,
	};
	int status;

2877
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
2878
	if (status == 0) {
2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890
		/* Sanity check the server answers */
		switch (server->nfs_client->cl_minorversion) {
		case 0:
			res.attr_bitmask[1] &= FATTR4_WORD1_NFS40_MASK;
			res.attr_bitmask[2] = 0;
			break;
		case 1:
			res.attr_bitmask[2] &= FATTR4_WORD2_NFS41_MASK;
			break;
		case 2:
			res.attr_bitmask[2] &= FATTR4_WORD2_NFS42_MASK;
		}
L
Linus Torvalds 已提交
2891
		memcpy(server->attr_bitmask, res.attr_bitmask, sizeof(server->attr_bitmask));
2892 2893 2894 2895
		server->caps &= ~(NFS_CAP_ACLS|NFS_CAP_HARDLINKS|
				NFS_CAP_SYMLINKS|NFS_CAP_FILEID|
				NFS_CAP_MODE|NFS_CAP_NLINK|NFS_CAP_OWNER|
				NFS_CAP_OWNER_GROUP|NFS_CAP_ATIME|
2896 2897
				NFS_CAP_CTIME|NFS_CAP_MTIME|
				NFS_CAP_SECURITY_LABEL);
2898 2899
		if (res.attr_bitmask[0] & FATTR4_WORD0_ACL &&
				res.acl_bitmask & ACL4_SUPPORT_ALLOW_ACL)
L
Linus Torvalds 已提交
2900 2901 2902 2903 2904
			server->caps |= NFS_CAP_ACLS;
		if (res.has_links != 0)
			server->caps |= NFS_CAP_HARDLINKS;
		if (res.has_symlinks != 0)
			server->caps |= NFS_CAP_SYMLINKS;
2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920
		if (res.attr_bitmask[0] & FATTR4_WORD0_FILEID)
			server->caps |= NFS_CAP_FILEID;
		if (res.attr_bitmask[1] & FATTR4_WORD1_MODE)
			server->caps |= NFS_CAP_MODE;
		if (res.attr_bitmask[1] & FATTR4_WORD1_NUMLINKS)
			server->caps |= NFS_CAP_NLINK;
		if (res.attr_bitmask[1] & FATTR4_WORD1_OWNER)
			server->caps |= NFS_CAP_OWNER;
		if (res.attr_bitmask[1] & FATTR4_WORD1_OWNER_GROUP)
			server->caps |= NFS_CAP_OWNER_GROUP;
		if (res.attr_bitmask[1] & FATTR4_WORD1_TIME_ACCESS)
			server->caps |= NFS_CAP_ATIME;
		if (res.attr_bitmask[1] & FATTR4_WORD1_TIME_METADATA)
			server->caps |= NFS_CAP_CTIME;
		if (res.attr_bitmask[1] & FATTR4_WORD1_TIME_MODIFY)
			server->caps |= NFS_CAP_MTIME;
2921 2922 2923 2924 2925 2926
#ifdef CONFIG_NFS_V4_SECURITY_LABEL
		if (res.attr_bitmask[2] & FATTR4_WORD2_SECURITY_LABEL)
			server->caps |= NFS_CAP_SECURITY_LABEL;
#endif
		memcpy(server->attr_bitmask_nl, res.attr_bitmask,
				sizeof(server->attr_bitmask));
2927
		server->attr_bitmask_nl[2] &= ~FATTR4_WORD2_SECURITY_LABEL;
2928

2929 2930 2931
		memcpy(server->cache_consistency_bitmask, res.attr_bitmask, sizeof(server->cache_consistency_bitmask));
		server->cache_consistency_bitmask[0] &= FATTR4_WORD0_CHANGE|FATTR4_WORD0_SIZE;
		server->cache_consistency_bitmask[1] &= FATTR4_WORD1_TIME_METADATA|FATTR4_WORD1_TIME_MODIFY;
2932
		server->cache_consistency_bitmask[2] = 0;
L
Linus Torvalds 已提交
2933
		server->acl_bitmask = res.acl_bitmask;
2934
		server->fh_expire_type = res.fh_expire_type;
L
Linus Torvalds 已提交
2935
	}
A
Andy Adamson 已提交
2936

L
Linus Torvalds 已提交
2937 2938 2939
	return status;
}

2940
int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
L
Linus Torvalds 已提交
2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(server,
				_nfs4_server_capabilities(server, fhandle),
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle,
		struct nfs_fsinfo *info)
{
2955
	u32 bitmask[3];
L
Linus Torvalds 已提交
2956
	struct nfs4_lookup_root_arg args = {
2957
		.bitmask = bitmask,
L
Linus Torvalds 已提交
2958 2959 2960
	};
	struct nfs4_lookup_res res = {
		.server = server,
2961
		.fattr = info->fattr,
L
Linus Torvalds 已提交
2962 2963 2964 2965 2966 2967 2968
		.fh = fhandle,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP_ROOT],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
2969

2970 2971 2972 2973 2974 2975 2976
	bitmask[0] = nfs4_fattr_bitmap[0];
	bitmask[1] = nfs4_fattr_bitmap[1];
	/*
	 * Process the label in the upcoming getfattr
	 */
	bitmask[2] = nfs4_fattr_bitmap[2] & ~FATTR4_WORD2_SECURITY_LABEL;

2977
	nfs_fattr_init(info->fattr);
2978
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
2979 2980 2981 2982 2983 2984 2985 2986
}

static int nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle,
		struct nfs_fsinfo *info)
{
	struct nfs4_exception exception = { };
	int err;
	do {
2987
		err = _nfs4_lookup_root(server, fhandle, info);
2988
		trace_nfs4_lookup_root(server, fhandle, info->fattr, err);
2989 2990 2991
		switch (err) {
		case 0:
		case -NFS4ERR_WRONGSEC:
2992
			goto out;
2993 2994 2995
		default:
			err = nfs4_handle_exception(server, err, &exception);
		}
L
Linus Torvalds 已提交
2996
	} while (exception.retry);
2997
out:
L
Linus Torvalds 已提交
2998 2999 3000
	return err;
}

3001 3002 3003
static int nfs4_lookup_root_sec(struct nfs_server *server, struct nfs_fh *fhandle,
				struct nfs_fsinfo *info, rpc_authflavor_t flavor)
{
3004 3005 3006
	struct rpc_auth_create_args auth_args = {
		.pseudoflavor = flavor,
	};
3007 3008 3009
	struct rpc_auth *auth;
	int ret;

3010
	auth = rpcauth_create(&auth_args, server->client);
3011
	if (IS_ERR(auth)) {
3012
		ret = -EACCES;
3013 3014 3015 3016 3017 3018 3019
		goto out;
	}
	ret = nfs4_lookup_root(server, fhandle, info);
out:
	return ret;
}

3020 3021 3022 3023 3024 3025 3026 3027 3028
/*
 * Retry pseudoroot lookup with various security flavors.  We do this when:
 *
 *   NFSv4.0: the PUTROOTFH operation returns NFS4ERR_WRONGSEC
 *   NFSv4.1: the server does not support the SECINFO_NO_NAME operation
 *
 * Returns zero on success, or a negative NFS4ERR value, or a
 * negative errno value.
 */
3029
static int nfs4_find_root_sec(struct nfs_server *server, struct nfs_fh *fhandle,
3030
			      struct nfs_fsinfo *info)
L
Linus Torvalds 已提交
3031
{
3032 3033 3034 3035 3036
	/* Per 3530bis 15.33.5 */
	static const rpc_authflavor_t flav_array[] = {
		RPC_AUTH_GSS_KRB5P,
		RPC_AUTH_GSS_KRB5I,
		RPC_AUTH_GSS_KRB5,
3037
		RPC_AUTH_UNIX,			/* courtesy */
3038 3039 3040 3041
		RPC_AUTH_NULL,
	};
	int status = -EPERM;
	size_t i;
C
Chuck Lever 已提交
3042

3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060
	if (server->auth_info.flavor_len > 0) {
		/* try each flavor specified by user */
		for (i = 0; i < server->auth_info.flavor_len; i++) {
			status = nfs4_lookup_root_sec(server, fhandle, info,
						server->auth_info.flavors[i]);
			if (status == -NFS4ERR_WRONGSEC || status == -EACCES)
				continue;
			break;
		}
	} else {
		/* no flavors specified by user, try default list */
		for (i = 0; i < ARRAY_SIZE(flav_array); i++) {
			status = nfs4_lookup_root_sec(server, fhandle, info,
						      flav_array[i]);
			if (status == -NFS4ERR_WRONGSEC || status == -EACCES)
				continue;
			break;
		}
3061
	}
3062

3063 3064 3065 3066 3067 3068 3069 3070 3071
	/*
	 * -EACCESS could mean that the user doesn't have correct permissions
	 * to access the mount.  It could also mean that we tried to mount
	 * with a gss auth flavor, but rpc.gssd isn't running.  Either way,
	 * existing mount programs don't handle -EACCES very well so it should
	 * be mapped to -EPERM instead.
	 */
	if (status == -EACCES)
		status = -EPERM;
3072 3073 3074
	return status;
}

C
Chuck Lever 已提交
3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086
static int nfs4_do_find_root_sec(struct nfs_server *server,
		struct nfs_fh *fhandle, struct nfs_fsinfo *info)
{
	int mv = server->nfs_client->cl_minorversion;
	return nfs_v4_minor_ops[mv]->find_root_sec(server, fhandle, info);
}

/**
 * nfs4_proc_get_rootfh - get file handle for server's pseudoroot
 * @server: initialized nfs_server handle
 * @fhandle: we fill in the pseudo-fs root file handle
 * @info: we fill in an FSINFO struct
3087
 * @auth_probe: probe the auth flavours
C
Chuck Lever 已提交
3088 3089
 *
 * Returns zero on success, or a negative errno.
3090
 */
3091
int nfs4_proc_get_rootfh(struct nfs_server *server, struct nfs_fh *fhandle,
3092 3093
			 struct nfs_fsinfo *info,
			 bool auth_probe)
3094
{
3095
	int status = 0;
C
Chuck Lever 已提交
3096

3097
	if (!auth_probe)
3098
		status = nfs4_lookup_root(server, fhandle, info);
3099 3100

	if (auth_probe || status == NFS4ERR_WRONGSEC)
C
Chuck Lever 已提交
3101 3102
		status = nfs4_do_find_root_sec(server, fhandle, info);

L
Linus Torvalds 已提交
3103 3104 3105 3106
	if (status == 0)
		status = nfs4_server_capabilities(server, fhandle);
	if (status == 0)
		status = nfs4_do_fsinfo(server, fhandle, info);
C
Chuck Lever 已提交
3107

3108
	return nfs4_map_errors(status);
L
Linus Torvalds 已提交
3109 3110
}

3111 3112 3113 3114 3115
static int nfs4_proc_get_root(struct nfs_server *server, struct nfs_fh *mntfh,
			      struct nfs_fsinfo *info)
{
	int error;
	struct nfs_fattr *fattr = info->fattr;
3116
	struct nfs4_label *label = NULL;
3117 3118 3119 3120 3121 3122 3123

	error = nfs4_server_capabilities(server, mntfh);
	if (error < 0) {
		dprintk("nfs4_get_root: getcaps error = %d\n", -error);
		return error;
	}

3124 3125 3126 3127
	label = nfs4_label_alloc(server, GFP_KERNEL);
	if (IS_ERR(label))
		return PTR_ERR(label);

3128
	error = nfs4_proc_getattr(server, mntfh, fattr, label);
3129 3130
	if (error < 0) {
		dprintk("nfs4_get_root: getattr error = %d\n", -error);
3131
		goto err_free_label;
3132 3133 3134 3135 3136 3137
	}

	if (fattr->valid & NFS_ATTR_FATTR_FSID &&
	    !nfs_fsid_equal(&server->fsid, &fattr->fsid))
		memcpy(&server->fsid, &fattr->fsid, sizeof(server->fsid));

3138 3139 3140
err_free_label:
	nfs4_label_free(label);

3141 3142 3143
	return error;
}

M
Manoj Naik 已提交
3144 3145 3146 3147 3148
/*
 * Get locations and (maybe) other attributes of a referral.
 * Note that we'll actually follow the referral later when
 * we detect fsid mismatch in inode revalidation
 */
3149 3150 3151
static int nfs4_get_referral(struct rpc_clnt *client, struct inode *dir,
			     const struct qstr *name, struct nfs_fattr *fattr,
			     struct nfs_fh *fhandle)
M
Manoj Naik 已提交
3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163
{
	int status = -ENOMEM;
	struct page *page = NULL;
	struct nfs4_fs_locations *locations = NULL;

	page = alloc_page(GFP_KERNEL);
	if (page == NULL)
		goto out;
	locations = kmalloc(sizeof(struct nfs4_fs_locations), GFP_KERNEL);
	if (locations == NULL)
		goto out;

3164
	status = nfs4_proc_fs_locations(client, dir, name, locations, page);
M
Manoj Naik 已提交
3165 3166
	if (status != 0)
		goto out;
3167 3168 3169 3170 3171 3172

	/*
	 * If the fsid didn't change, this is a migration event, not a
	 * referral.  Cause us to drop into the exception handler, which
	 * will kick off migration recovery.
	 */
M
Manoj Naik 已提交
3173
	if (nfs_fsid_equal(&NFS_SERVER(dir)->fsid, &locations->fattr.fsid)) {
3174 3175
		dprintk("%s: server did not return a different fsid for"
			" a referral at %s\n", __func__, name->name);
3176
		status = -NFS4ERR_MOVED;
M
Manoj Naik 已提交
3177 3178
		goto out;
	}
3179 3180
	/* Fixup attributes for the nfs_lookup() call to nfs_fhget() */
	nfs_fixup_referral_attributes(&locations->fattr);
M
Manoj Naik 已提交
3181

3182
	/* replace the lookup nfs_fattr with the locations nfs_fattr */
M
Manoj Naik 已提交
3183 3184 3185 3186 3187
	memcpy(fattr, &locations->fattr, sizeof(struct nfs_fattr));
	memset(fhandle, 0, sizeof(struct nfs_fh));
out:
	if (page)
		__free_page(page);
3188
	kfree(locations);
M
Manoj Naik 已提交
3189 3190 3191
	return status;
}

3192 3193
static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle,
				struct nfs_fattr *fattr, struct nfs4_label *label)
L
Linus Torvalds 已提交
3194 3195 3196 3197 3198 3199 3200
{
	struct nfs4_getattr_arg args = {
		.fh = fhandle,
		.bitmask = server->attr_bitmask,
	};
	struct nfs4_getattr_res res = {
		.fattr = fattr,
3201
		.label = label,
L
Linus Torvalds 已提交
3202 3203 3204 3205 3206 3207 3208
		.server = server,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETATTR],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
3209 3210 3211

	args.bitmask = nfs4_bitmask(server, label);

3212
	nfs_fattr_init(fattr);
3213
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3214 3215
}

3216 3217
static int nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle,
				struct nfs_fattr *fattr, struct nfs4_label *label)
L
Linus Torvalds 已提交
3218 3219 3220 3221
{
	struct nfs4_exception exception = { };
	int err;
	do {
3222 3223 3224
		err = _nfs4_proc_getattr(server, fhandle, fattr, label);
		trace_nfs4_getattr(server, fhandle, fattr, err);
		err = nfs4_handle_exception(server, err,
L
Linus Torvalds 已提交
3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250
				&exception);
	} while (exception.retry);
	return err;
}

/* 
 * The file is not closed if it is opened due to the a request to change
 * the size of the file. The open call will not be needed once the
 * VFS layer lookup-intents are implemented.
 *
 * Close is called when the inode is destroyed.
 * If we haven't opened the file for O_WRONLY, we
 * need to in the size_change case to obtain a stateid.
 *
 * Got race?
 * Because OPEN is always done by name in nfsv4, it is
 * possible that we opened a different file by the same
 * name.  We can recognize this race condition, but we
 * can't do anything about it besides returning an error.
 *
 * This will be fixed with VFS changes (lookup-intent).
 */
static int
nfs4_proc_setattr(struct dentry *dentry, struct nfs_fattr *fattr,
		  struct iattr *sattr)
{
3251
	struct inode *inode = d_inode(dentry);
3252
	struct rpc_cred *cred = NULL;
3253
	struct nfs4_state *state = NULL;
3254
	struct nfs4_label *label = NULL;
L
Linus Torvalds 已提交
3255 3256
	int status;

3257 3258 3259
	if (pnfs_ld_layoutret_on_setattr(inode) &&
	    sattr->ia_valid & ATTR_SIZE &&
	    sattr->ia_size < i_size_read(inode))
3260
		pnfs_commit_and_return_layout(inode);
B
Benny Halevy 已提交
3261

3262
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
3263
	
3264 3265
	/* Deal with open(O_TRUNC) */
	if (sattr->ia_valid & ATTR_OPEN)
3266
		sattr->ia_valid &= ~(ATTR_MTIME|ATTR_CTIME);
3267 3268

	/* Optimization: if the end result is no change, don't RPC */
3269
	if ((sattr->ia_valid & ~(ATTR_FILE|ATTR_OPEN)) == 0)
3270 3271
		return 0;

3272
	/* Search for an existing open(O_WRITE) file */
3273 3274 3275 3276
	if (sattr->ia_valid & ATTR_FILE) {
		struct nfs_open_context *ctx;

		ctx = nfs_file_open_context(sattr->ia_file);
N
Neil Brown 已提交
3277 3278 3279 3280
		if (ctx) {
			cred = ctx->cred;
			state = ctx->state;
		}
3281
	}
3282

3283 3284 3285 3286 3287
	label = nfs4_label_alloc(NFS_SERVER(inode), GFP_KERNEL);
	if (IS_ERR(label))
		return PTR_ERR(label);

	status = nfs4_do_setattr(inode, cred, fattr, sattr, state, NULL, label);
3288
	if (status == 0) {
3289
		nfs_setattr_update_inode(inode, sattr, fattr);
3290 3291
		nfs_setsecurity(inode, fattr, label);
	}
3292
	nfs4_label_free(label);
L
Linus Torvalds 已提交
3293 3294 3295
	return status;
}

3296 3297
static int _nfs4_proc_lookup(struct rpc_clnt *clnt, struct inode *dir,
		const struct qstr *name, struct nfs_fh *fhandle,
3298
		struct nfs_fattr *fattr, struct nfs4_label *label)
D
David Howells 已提交
3299
{
3300
	struct nfs_server *server = NFS_SERVER(dir);
D
David Howells 已提交
3301 3302 3303
	int		       status;
	struct nfs4_lookup_arg args = {
		.bitmask = server->attr_bitmask,
3304
		.dir_fh = NFS_FH(dir),
D
David Howells 已提交
3305 3306 3307 3308 3309
		.name = name,
	};
	struct nfs4_lookup_res res = {
		.server = server,
		.fattr = fattr,
3310
		.label = label,
D
David Howells 已提交
3311 3312 3313 3314 3315 3316 3317 3318
		.fh = fhandle,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};

3319 3320
	args.bitmask = nfs4_bitmask(server, label);

D
David Howells 已提交
3321 3322
	nfs_fattr_init(fattr);

L
Linus Torvalds 已提交
3323
	dprintk("NFS call  lookup %s\n", name->name);
3324
	status = nfs4_call_sync(clnt, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3325 3326 3327 3328
	dprintk("NFS reply lookup: %d\n", status);
	return status;
}

3329
static void nfs_fixup_secinfo_attributes(struct nfs_fattr *fattr)
3330 3331
{
	fattr->valid |= NFS_ATTR_FATTR_TYPE | NFS_ATTR_FATTR_MODE |
3332
		NFS_ATTR_FATTR_NLINK | NFS_ATTR_FATTR_MOUNTPOINT;
3333 3334 3335 3336
	fattr->mode = S_IFDIR | S_IRUGO | S_IXUGO;
	fattr->nlink = 2;
}

3337 3338
static int nfs4_proc_lookup_common(struct rpc_clnt **clnt, struct inode *dir,
				   struct qstr *name, struct nfs_fh *fhandle,
3339
				   struct nfs_fattr *fattr, struct nfs4_label *label)
L
Linus Torvalds 已提交
3340 3341
{
	struct nfs4_exception exception = { };
3342
	struct rpc_clnt *client = *clnt;
L
Linus Torvalds 已提交
3343 3344
	int err;
	do {
3345
		err = _nfs4_proc_lookup(client, dir, name, fhandle, fattr, label);
3346
		trace_nfs4_lookup(dir, name, err);
3347
		switch (err) {
3348
		case -NFS4ERR_BADNAME:
3349 3350
			err = -ENOENT;
			goto out;
3351
		case -NFS4ERR_MOVED:
3352
			err = nfs4_get_referral(client, dir, name, fattr, fhandle);
3353 3354
			if (err == -NFS4ERR_MOVED)
				err = nfs4_handle_exception(NFS_SERVER(dir), err, &exception);
3355
			goto out;
3356
		case -NFS4ERR_WRONGSEC:
3357 3358 3359
			err = -EPERM;
			if (client != *clnt)
				goto out;
3360
			client = nfs4_negotiate_security(client, dir, name);
3361 3362 3363 3364 3365 3366 3367
			if (IS_ERR(client))
				return PTR_ERR(client);

			exception.retry = 1;
			break;
		default:
			err = nfs4_handle_exception(NFS_SERVER(dir), err, &exception);
3368
		}
L
Linus Torvalds 已提交
3369
	} while (exception.retry);
3370 3371 3372 3373 3374 3375 3376

out:
	if (err == 0)
		*clnt = client;
	else if (client != *clnt)
		rpc_shutdown_client(client);

L
Linus Torvalds 已提交
3377 3378 3379
	return err;
}

3380
static int nfs4_proc_lookup(struct inode *dir, struct qstr *name,
3381 3382
			    struct nfs_fh *fhandle, struct nfs_fattr *fattr,
			    struct nfs4_label *label)
3383 3384 3385 3386
{
	int status;
	struct rpc_clnt *client = NFS_CLIENT(dir);

3387
	status = nfs4_proc_lookup_common(&client, dir, name, fhandle, fattr, label);
3388 3389 3390 3391 3392 3393 3394
	if (client != NFS_CLIENT(dir)) {
		rpc_shutdown_client(client);
		nfs_fixup_secinfo_attributes(fattr);
	}
	return status;
}

3395 3396 3397 3398
struct rpc_clnt *
nfs4_proc_lookup_mountpoint(struct inode *dir, struct qstr *name,
			    struct nfs_fh *fhandle, struct nfs_fattr *fattr)
{
3399
	struct rpc_clnt *client = NFS_CLIENT(dir);
3400 3401
	int status;

3402
	status = nfs4_proc_lookup_common(&client, dir, name, fhandle, fattr, NULL);
3403
	if (status < 0)
3404
		return ERR_PTR(status);
3405
	return (client == NFS_CLIENT(dir)) ? rpc_clone_client(client) : client;
3406 3407
}

L
Linus Torvalds 已提交
3408 3409
static int _nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
{
3410
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
3411 3412
	struct nfs4_accessargs args = {
		.fh = NFS_FH(inode),
3413
		.bitmask = server->cache_consistency_bitmask,
3414 3415 3416
	};
	struct nfs4_accessres res = {
		.server = server,
L
Linus Torvalds 已提交
3417 3418 3419 3420 3421 3422 3423 3424
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_ACCESS],
		.rpc_argp = &args,
		.rpc_resp = &res,
		.rpc_cred = entry->cred,
	};
	int mode = entry->mask;
3425
	int status = 0;
L
Linus Torvalds 已提交
3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442

	/*
	 * Determine which access bits we want to ask for...
	 */
	if (mode & MAY_READ)
		args.access |= NFS4_ACCESS_READ;
	if (S_ISDIR(inode->i_mode)) {
		if (mode & MAY_WRITE)
			args.access |= NFS4_ACCESS_MODIFY | NFS4_ACCESS_EXTEND | NFS4_ACCESS_DELETE;
		if (mode & MAY_EXEC)
			args.access |= NFS4_ACCESS_LOOKUP;
	} else {
		if (mode & MAY_WRITE)
			args.access |= NFS4_ACCESS_MODIFY | NFS4_ACCESS_EXTEND;
		if (mode & MAY_EXEC)
			args.access |= NFS4_ACCESS_EXECUTE;
	}
3443 3444 3445 3446 3447

	res.fattr = nfs_alloc_fattr();
	if (res.fattr == NULL)
		return -ENOMEM;

3448
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3449
	if (!status) {
3450
		nfs_access_set_mask(entry, res.access);
3451
		nfs_refresh_inode(inode, res.fattr);
L
Linus Torvalds 已提交
3452
	}
3453
	nfs_free_fattr(res.fattr);
L
Linus Torvalds 已提交
3454 3455 3456 3457 3458 3459 3460 3461
	return status;
}

static int nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
{
	struct nfs4_exception exception = { };
	int err;
	do {
3462 3463 3464
		err = _nfs4_proc_access(inode, entry);
		trace_nfs4_access(inode, err);
		err = nfs4_handle_exception(NFS_SERVER(inode), err,
L
Linus Torvalds 已提交
3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486
				&exception);
	} while (exception.retry);
	return err;
}

/*
 * TODO: For the time being, we don't try to get any attributes
 * along with any of the zero-copy operations READ, READDIR,
 * READLINK, WRITE.
 *
 * In the case of the first three, we want to put the GETATTR
 * after the read-type operation -- this is because it is hard
 * to predict the length of a GETATTR response in v4, and thus
 * align the READ data correctly.  This means that the GETATTR
 * may end up partially falling into the page cache, and we should
 * shift it into the 'tail' of the xdr_buf before processing.
 * To do this efficiently, we need to know the total length
 * of data received, which doesn't seem to be available outside
 * of the RPC layer.
 *
 * In the case of WRITE, we also want to put the GETATTR after
 * the operation -- in this case because we want to make sure
3487
 * we get the post-operation mtime and size.
L
Linus Torvalds 已提交
3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500
 *
 * Both of these changes to the XDR layer would in fact be quite
 * minor, but I decided to leave them for a subsequent patch.
 */
static int _nfs4_proc_readlink(struct inode *inode, struct page *page,
		unsigned int pgbase, unsigned int pglen)
{
	struct nfs4_readlink args = {
		.fh       = NFS_FH(inode),
		.pgbase	  = pgbase,
		.pglen    = pglen,
		.pages    = &page,
	};
B
Benny Halevy 已提交
3501
	struct nfs4_readlink_res res;
L
Linus Torvalds 已提交
3502 3503 3504
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READLINK],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3505
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3506 3507
	};

3508
	return nfs4_call_sync(NFS_SERVER(inode)->client, NFS_SERVER(inode), &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3509 3510 3511 3512 3513 3514 3515 3516
}

static int nfs4_proc_readlink(struct inode *inode, struct page *page,
		unsigned int pgbase, unsigned int pglen)
{
	struct nfs4_exception exception = { };
	int err;
	do {
3517 3518 3519
		err = _nfs4_proc_readlink(inode, page, pgbase, pglen);
		trace_nfs4_readlink(inode, err);
		err = nfs4_handle_exception(NFS_SERVER(inode), err,
L
Linus Torvalds 已提交
3520 3521 3522 3523 3524 3525
				&exception);
	} while (exception.retry);
	return err;
}

/*
3526
 * This is just for mknod.  open(O_CREAT) will always do ->open_context().
L
Linus Torvalds 已提交
3527 3528 3529
 */
static int
nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
3530
		 int flags)
L
Linus Torvalds 已提交
3531
{
3532
	struct nfs4_label l, *ilabel = NULL;
3533
	struct nfs_open_context *ctx;
L
Linus Torvalds 已提交
3534
	struct nfs4_state *state;
3535
	int opened = 0;
L
Linus Torvalds 已提交
3536 3537
	int status = 0;

3538 3539 3540 3541
	ctx = alloc_nfs_open_context(dentry, FMODE_READ);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);

3542 3543
	ilabel = nfs4_label_init_security(dir, dentry, sattr, &l);

A
Aneesh Kumar K.V 已提交
3544
	sattr->ia_mode &= ~current_umask();
3545
	state = nfs4_do_open(dir, ctx, flags, sattr, ilabel, &opened);
L
Linus Torvalds 已提交
3546 3547
	if (IS_ERR(state)) {
		status = PTR_ERR(state);
3548
		goto out;
L
Linus Torvalds 已提交
3549 3550
	}
out:
3551
	nfs4_label_release_security(ilabel);
3552
	put_nfs_open_context(ctx);
L
Linus Torvalds 已提交
3553 3554 3555 3556 3557
	return status;
}

static int _nfs4_proc_remove(struct inode *dir, struct qstr *name)
{
3558
	struct nfs_server *server = NFS_SERVER(dir);
3559
	struct nfs_removeargs args = {
L
Linus Torvalds 已提交
3560
		.fh = NFS_FH(dir),
3561
		.name = *name,
3562
	};
3563
	struct nfs_removeres res = {
3564
		.server = server,
L
Linus Torvalds 已提交
3565 3566
	};
	struct rpc_message msg = {
3567 3568 3569
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE],
		.rpc_argp = &args,
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3570
	};
3571
	int status;
L
Linus Torvalds 已提交
3572

3573
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 1);
3574
	if (status == 0)
3575
		update_changeattr(dir, &res.cinfo);
L
Linus Torvalds 已提交
3576 3577 3578 3579 3580 3581 3582 3583
	return status;
}

static int nfs4_proc_remove(struct inode *dir, struct qstr *name)
{
	struct nfs4_exception exception = { };
	int err;
	do {
3584 3585 3586
		err = _nfs4_proc_remove(dir, name);
		trace_nfs4_remove(dir, name, err);
		err = nfs4_handle_exception(NFS_SERVER(dir), err,
L
Linus Torvalds 已提交
3587 3588 3589 3590 3591
				&exception);
	} while (exception.retry);
	return err;
}

3592
static void nfs4_proc_unlink_setup(struct rpc_message *msg, struct inode *dir)
L
Linus Torvalds 已提交
3593
{
3594 3595 3596
	struct nfs_server *server = NFS_SERVER(dir);
	struct nfs_removeargs *args = msg->rpc_argp;
	struct nfs_removeres *res = msg->rpc_resp;
L
Linus Torvalds 已提交
3597

3598
	res->server = server;
L
Linus Torvalds 已提交
3599
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE];
3600
	nfs4_init_sequence(&args->seq_args, &res->seq_res, 1);
3601 3602

	nfs_fattr_init(res->dir_attr);
L
Linus Torvalds 已提交
3603 3604
}

3605 3606
static void nfs4_proc_unlink_rpc_prepare(struct rpc_task *task, struct nfs_unlinkdata *data)
{
3607 3608 3609 3610
	nfs4_setup_sequence(NFS_SERVER(data->dir),
			&data->args.seq_args,
			&data->res.seq_res,
			task);
L
Linus Torvalds 已提交
3611 3612
}

3613
static int nfs4_proc_unlink_done(struct rpc_task *task, struct inode *dir)
L
Linus Torvalds 已提交
3614
{
3615 3616
	struct nfs_unlinkdata *data = task->tk_calldata;
	struct nfs_removeres *res = &data->res;
3617

3618 3619
	if (!nfs4_sequence_done(task, &res->seq_res))
		return 0;
3620 3621
	if (nfs4_async_handle_error(task, res->server, NULL,
				    &data->timeout) == -EAGAIN)
3622 3623 3624
		return 0;
	update_changeattr(dir, &res->cinfo);
	return 1;
L
Linus Torvalds 已提交
3625 3626
}

3627 3628 3629 3630 3631 3632 3633 3634
static void nfs4_proc_rename_setup(struct rpc_message *msg, struct inode *dir)
{
	struct nfs_server *server = NFS_SERVER(dir);
	struct nfs_renameargs *arg = msg->rpc_argp;
	struct nfs_renameres *res = msg->rpc_resp;

	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RENAME];
	res->server = server;
3635
	nfs4_init_sequence(&arg->seq_args, &res->seq_res, 1);
3636 3637
}

3638 3639
static void nfs4_proc_rename_rpc_prepare(struct rpc_task *task, struct nfs_renamedata *data)
{
3640 3641 3642 3643
	nfs4_setup_sequence(NFS_SERVER(data->old_dir),
			&data->args.seq_args,
			&data->res.seq_res,
			task);
3644 3645 3646 3647 3648
}

static int nfs4_proc_rename_done(struct rpc_task *task, struct inode *old_dir,
				 struct inode *new_dir)
{
3649 3650
	struct nfs_renamedata *data = task->tk_calldata;
	struct nfs_renameres *res = &data->res;
3651 3652 3653

	if (!nfs4_sequence_done(task, &res->seq_res))
		return 0;
3654
	if (nfs4_async_handle_error(task, res->server, NULL, &data->timeout) == -EAGAIN)
3655 3656 3657 3658 3659 3660 3661
		return 0;

	update_changeattr(old_dir, &res->old_cinfo);
	update_changeattr(new_dir, &res->new_cinfo);
	return 1;
}

L
Linus Torvalds 已提交
3662 3663
static int _nfs4_proc_link(struct inode *inode, struct inode *dir, struct qstr *name)
{
3664
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
3665 3666 3667 3668
	struct nfs4_link_arg arg = {
		.fh     = NFS_FH(inode),
		.dir_fh = NFS_FH(dir),
		.name   = name,
3669 3670 3671 3672
		.bitmask = server->attr_bitmask,
	};
	struct nfs4_link_res res = {
		.server = server,
3673
		.label = NULL,
L
Linus Torvalds 已提交
3674 3675 3676 3677
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LINK],
		.rpc_argp = &arg,
3678
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3679
	};
3680 3681 3682
	int status = -ENOMEM;

	res.fattr = nfs_alloc_fattr();
3683
	if (res.fattr == NULL)
3684
		goto out;
L
Linus Torvalds 已提交
3685

3686 3687 3688 3689 3690
	res.label = nfs4_label_alloc(server, GFP_KERNEL);
	if (IS_ERR(res.label)) {
		status = PTR_ERR(res.label);
		goto out;
	}
3691
	arg.bitmask = nfs4_bitmask(server, res.label);
3692

3693
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
3694 3695
	if (!status) {
		update_changeattr(dir, &res.cinfo);
3696 3697 3698
		status = nfs_post_op_update_inode(inode, res.fattr);
		if (!status)
			nfs_setsecurity(inode, res.fattr, res.label);
3699
	}
3700 3701 3702 3703


	nfs4_label_free(res.label);

3704 3705
out:
	nfs_free_fattr(res.fattr);
L
Linus Torvalds 已提交
3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720
	return status;
}

static int nfs4_proc_link(struct inode *inode, struct inode *dir, struct qstr *name)
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(NFS_SERVER(inode),
				_nfs4_proc_link(inode, dir, name),
				&exception);
	} while (exception.retry);
	return err;
}

3721 3722 3723 3724 3725 3726
struct nfs4_createdata {
	struct rpc_message msg;
	struct nfs4_create_arg arg;
	struct nfs4_create_res res;
	struct nfs_fh fh;
	struct nfs_fattr fattr;
3727
	struct nfs4_label *label;
3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738
};

static struct nfs4_createdata *nfs4_alloc_createdata(struct inode *dir,
		struct qstr *name, struct iattr *sattr, u32 ftype)
{
	struct nfs4_createdata *data;

	data = kzalloc(sizeof(*data), GFP_KERNEL);
	if (data != NULL) {
		struct nfs_server *server = NFS_SERVER(dir);

3739 3740 3741 3742
		data->label = nfs4_label_alloc(server, GFP_KERNEL);
		if (IS_ERR(data->label))
			goto out_free;

3743 3744 3745 3746 3747 3748 3749 3750
		data->msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CREATE];
		data->msg.rpc_argp = &data->arg;
		data->msg.rpc_resp = &data->res;
		data->arg.dir_fh = NFS_FH(dir);
		data->arg.server = server;
		data->arg.name = name;
		data->arg.attrs = sattr;
		data->arg.ftype = ftype;
3751
		data->arg.bitmask = nfs4_bitmask(server, data->label);
3752 3753 3754
		data->res.server = server;
		data->res.fh = &data->fh;
		data->res.fattr = &data->fattr;
3755
		data->res.label = data->label;
3756 3757 3758
		nfs_fattr_init(data->res.fattr);
	}
	return data;
3759 3760 3761
out_free:
	kfree(data);
	return NULL;
3762 3763 3764 3765
}

static int nfs4_do_create(struct inode *dir, struct dentry *dentry, struct nfs4_createdata *data)
{
3766
	int status = nfs4_call_sync(NFS_SERVER(dir)->client, NFS_SERVER(dir), &data->msg,
3767
				    &data->arg.seq_args, &data->res.seq_res, 1);
3768 3769
	if (status == 0) {
		update_changeattr(dir, &data->res.dir_cinfo);
3770
		status = nfs_instantiate(dentry, data->res.fh, data->res.fattr, data->res.label);
3771 3772 3773 3774 3775 3776
	}
	return status;
}

static void nfs4_free_createdata(struct nfs4_createdata *data)
{
3777
	nfs4_label_free(data->label);
3778 3779 3780
	kfree(data);
}

3781
static int _nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
3782 3783
		struct page *page, unsigned int len, struct iattr *sattr,
		struct nfs4_label *label)
L
Linus Torvalds 已提交
3784
{
3785 3786
	struct nfs4_createdata *data;
	int status = -ENAMETOOLONG;
L
Linus Torvalds 已提交
3787

3788
	if (len > NFS4_MAXPATHLEN)
3789
		goto out;
3790

3791 3792 3793 3794 3795 3796 3797 3798
	status = -ENOMEM;
	data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4LNK);
	if (data == NULL)
		goto out;

	data->msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SYMLINK];
	data->arg.u.symlink.pages = &page;
	data->arg.u.symlink.len = len;
3799
	data->arg.label = label;
L
Linus Torvalds 已提交
3800
	
3801 3802 3803 3804
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
3805 3806 3807
	return status;
}

3808
static int nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
3809
		struct page *page, unsigned int len, struct iattr *sattr)
L
Linus Torvalds 已提交
3810 3811
{
	struct nfs4_exception exception = { };
3812
	struct nfs4_label l, *label = NULL;
L
Linus Torvalds 已提交
3813
	int err;
3814 3815 3816

	label = nfs4_label_init_security(dir, dentry, sattr, &l);

L
Linus Torvalds 已提交
3817
	do {
3818 3819 3820
		err = _nfs4_proc_symlink(dir, dentry, page, len, sattr, label);
		trace_nfs4_symlink(dir, &dentry->d_name, err);
		err = nfs4_handle_exception(NFS_SERVER(dir), err,
L
Linus Torvalds 已提交
3821 3822
				&exception);
	} while (exception.retry);
3823 3824

	nfs4_label_release_security(label);
L
Linus Torvalds 已提交
3825 3826 3827 3828
	return err;
}

static int _nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
3829
		struct iattr *sattr, struct nfs4_label *label)
L
Linus Torvalds 已提交
3830
{
3831 3832
	struct nfs4_createdata *data;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
3833

3834 3835 3836 3837
	data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4DIR);
	if (data == NULL)
		goto out;

3838
	data->arg.label = label;
3839 3840 3841 3842
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
3843 3844 3845 3846 3847 3848 3849
	return status;
}

static int nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr)
{
	struct nfs4_exception exception = { };
3850
	struct nfs4_label l, *label = NULL;
L
Linus Torvalds 已提交
3851
	int err;
A
Aneesh Kumar K.V 已提交
3852

3853 3854
	label = nfs4_label_init_security(dir, dentry, sattr, &l);

A
Aneesh Kumar K.V 已提交
3855
	sattr->ia_mode &= ~current_umask();
L
Linus Torvalds 已提交
3856
	do {
3857 3858 3859
		err = _nfs4_proc_mkdir(dir, dentry, sattr, label);
		trace_nfs4_mkdir(dir, &dentry->d_name, err);
		err = nfs4_handle_exception(NFS_SERVER(dir), err,
L
Linus Torvalds 已提交
3860 3861
				&exception);
	} while (exception.retry);
3862 3863
	nfs4_label_release_security(label);

L
Linus Torvalds 已提交
3864 3865 3866 3867
	return err;
}

static int _nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
B
Bryan Schumaker 已提交
3868
		u64 cookie, struct page **pages, unsigned int count, int plus)
L
Linus Torvalds 已提交
3869
{
3870
	struct inode		*dir = d_inode(dentry);
L
Linus Torvalds 已提交
3871 3872
	struct nfs4_readdir_arg args = {
		.fh = NFS_FH(dir),
B
Bryan Schumaker 已提交
3873
		.pages = pages,
L
Linus Torvalds 已提交
3874 3875
		.pgbase = 0,
		.count = count,
3876
		.bitmask = NFS_SERVER(d_inode(dentry))->attr_bitmask,
B
Bryan Schumaker 已提交
3877
		.plus = plus,
L
Linus Torvalds 已提交
3878 3879 3880 3881 3882 3883 3884 3885 3886 3887
	};
	struct nfs4_readdir_res res;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READDIR],
		.rpc_argp = &args,
		.rpc_resp = &res,
		.rpc_cred = cred,
	};
	int			status;

3888 3889
	dprintk("%s: dentry = %pd2, cookie = %Lu\n", __func__,
			dentry,
3890
			(unsigned long long)cookie);
3891
	nfs4_setup_readdir(cookie, NFS_I(dir)->cookieverf, dentry, &args);
L
Linus Torvalds 已提交
3892
	res.pgbase = args.pgbase;
3893
	status = nfs4_call_sync(NFS_SERVER(dir)->client, NFS_SERVER(dir), &msg, &args.seq_args, &res.seq_res, 0);
3894
	if (status >= 0) {
3895
		memcpy(NFS_I(dir)->cookieverf, res.verifier.data, NFS4_VERIFIER_SIZE);
3896 3897
		status += args.pgbase;
	}
3898 3899 3900

	nfs_invalidate_atime(dir);

3901
	dprintk("%s: returns %d\n", __func__, status);
L
Linus Torvalds 已提交
3902 3903 3904 3905
	return status;
}

static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
B
Bryan Schumaker 已提交
3906
		u64 cookie, struct page **pages, unsigned int count, int plus)
L
Linus Torvalds 已提交
3907 3908 3909 3910
{
	struct nfs4_exception exception = { };
	int err;
	do {
3911 3912
		err = _nfs4_proc_readdir(dentry, cred, cookie,
				pages, count, plus);
3913 3914
		trace_nfs4_readdir(d_inode(dentry), err);
		err = nfs4_handle_exception(NFS_SERVER(d_inode(dentry)), err,
L
Linus Torvalds 已提交
3915 3916 3917 3918 3919 3920
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
3921
		struct iattr *sattr, struct nfs4_label *label, dev_t rdev)
L
Linus Torvalds 已提交
3922
{
3923 3924 3925
	struct nfs4_createdata *data;
	int mode = sattr->ia_mode;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
3926

3927 3928 3929 3930
	data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4SOCK);
	if (data == NULL)
		goto out;

L
Linus Torvalds 已提交
3931
	if (S_ISFIFO(mode))
3932
		data->arg.ftype = NF4FIFO;
L
Linus Torvalds 已提交
3933
	else if (S_ISBLK(mode)) {
3934 3935 3936
		data->arg.ftype = NF4BLK;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
L
Linus Torvalds 已提交
3937 3938
	}
	else if (S_ISCHR(mode)) {
3939 3940 3941
		data->arg.ftype = NF4CHR;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
3942 3943 3944
	} else if (!S_ISSOCK(mode)) {
		status = -EINVAL;
		goto out_free;
L
Linus Torvalds 已提交
3945
	}
3946

3947
	data->arg.label = label;
3948
	status = nfs4_do_create(dir, dentry, data);
3949
out_free:
3950 3951
	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
3952 3953 3954 3955 3956 3957 3958
	return status;
}

static int nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr, dev_t rdev)
{
	struct nfs4_exception exception = { };
3959
	struct nfs4_label l, *label = NULL;
L
Linus Torvalds 已提交
3960
	int err;
A
Aneesh Kumar K.V 已提交
3961

3962 3963
	label = nfs4_label_init_security(dir, dentry, sattr, &l);

A
Aneesh Kumar K.V 已提交
3964
	sattr->ia_mode &= ~current_umask();
L
Linus Torvalds 已提交
3965
	do {
3966 3967 3968
		err = _nfs4_proc_mknod(dir, dentry, sattr, label, rdev);
		trace_nfs4_mknod(dir, &dentry->d_name, err);
		err = nfs4_handle_exception(NFS_SERVER(dir), err,
L
Linus Torvalds 已提交
3969 3970
				&exception);
	} while (exception.retry);
3971 3972 3973

	nfs4_label_release_security(label);

L
Linus Torvalds 已提交
3974 3975 3976 3977 3978 3979 3980 3981 3982 3983
	return err;
}

static int _nfs4_proc_statfs(struct nfs_server *server, struct nfs_fh *fhandle,
		 struct nfs_fsstat *fsstat)
{
	struct nfs4_statfs_arg args = {
		.fh = fhandle,
		.bitmask = server->attr_bitmask,
	};
B
Benny Halevy 已提交
3984 3985 3986
	struct nfs4_statfs_res res = {
		.fsstat = fsstat,
	};
L
Linus Torvalds 已提交
3987 3988 3989
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_STATFS],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3990
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3991 3992
	};

3993
	nfs_fattr_init(fsstat->fattr);
3994
	return  nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015
}

static int nfs4_proc_statfs(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsstat *fsstat)
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(server,
				_nfs4_proc_statfs(server, fhandle, fsstat),
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle,
		struct nfs_fsinfo *fsinfo)
{
	struct nfs4_fsinfo_arg args = {
		.fh = fhandle,
		.bitmask = server->attr_bitmask,
	};
B
Benny Halevy 已提交
4016 4017 4018
	struct nfs4_fsinfo_res res = {
		.fsinfo = fsinfo,
	};
L
Linus Torvalds 已提交
4019 4020 4021
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FSINFO],
		.rpc_argp = &args,
B
Benny Halevy 已提交
4022
		.rpc_resp = &res,
L
Linus Torvalds 已提交
4023 4024
	};

4025
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
4026 4027 4028 4029 4030
}

static int nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo)
{
	struct nfs4_exception exception = { };
4031
	unsigned long now = jiffies;
L
Linus Torvalds 已提交
4032 4033 4034
	int err;

	do {
4035
		err = _nfs4_do_fsinfo(server, fhandle, fsinfo);
4036
		trace_nfs4_fsinfo(server, fhandle, fsinfo->fattr, err);
4037 4038 4039 4040 4041 4042 4043 4044 4045 4046
		if (err == 0) {
			struct nfs_client *clp = server->nfs_client;

			spin_lock(&clp->cl_lock);
			clp->cl_lease_time = fsinfo->lease_time * HZ;
			clp->cl_last_renewal = now;
			spin_unlock(&clp->cl_lock);
			break;
		}
		err = nfs4_handle_exception(server, err, &exception);
L
Linus Torvalds 已提交
4047 4048 4049 4050 4051 4052
	} while (exception.retry);
	return err;
}

static int nfs4_proc_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo)
{
4053 4054
	int error;

4055
	nfs_fattr_init(fsinfo->fattr);
4056
	error = nfs4_do_fsinfo(server, fhandle, fsinfo);
4057 4058 4059
	if (error == 0) {
		/* block layout checks this! */
		server->pnfs_blksize = fsinfo->blksize;
4060
		set_pnfs_layoutdriver(server, fhandle, fsinfo->layouttype);
4061
	}
4062 4063

	return error;
L
Linus Torvalds 已提交
4064 4065 4066 4067 4068 4069 4070 4071 4072
}

static int _nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle,
		struct nfs_pathconf *pathconf)
{
	struct nfs4_pathconf_arg args = {
		.fh = fhandle,
		.bitmask = server->attr_bitmask,
	};
B
Benny Halevy 已提交
4073 4074 4075
	struct nfs4_pathconf_res res = {
		.pathconf = pathconf,
	};
L
Linus Torvalds 已提交
4076 4077 4078
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_PATHCONF],
		.rpc_argp = &args,
B
Benny Halevy 已提交
4079
		.rpc_resp = &res,
L
Linus Torvalds 已提交
4080 4081 4082 4083 4084 4085 4086 4087
	};

	/* None of the pathconf attributes are mandatory to implement */
	if ((args.bitmask[0] & nfs4_pathconf_bitmap[0]) == 0) {
		memset(pathconf, 0, sizeof(*pathconf));
		return 0;
	}

4088
	nfs_fattr_init(pathconf->fattr);
4089
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105
}

static int nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle,
		struct nfs_pathconf *pathconf)
{
	struct nfs4_exception exception = { };
	int err;

	do {
		err = nfs4_handle_exception(server,
				_nfs4_proc_pathconf(server, fhandle, pathconf),
				&exception);
	} while (exception.retry);
	return err;
}

4106
int nfs4_set_rw_stateid(nfs4_stateid *stateid,
4107 4108 4109 4110 4111 4112 4113 4114
		const struct nfs_open_context *ctx,
		const struct nfs_lock_context *l_ctx,
		fmode_t fmode)
{
	const struct nfs_lockowner *lockowner = NULL;

	if (l_ctx != NULL)
		lockowner = &l_ctx->lockowner;
4115
	return nfs4_select_rw_stateid(stateid, ctx->state, fmode, lockowner);
4116 4117 4118
}
EXPORT_SYMBOL_GPL(nfs4_set_rw_stateid);

4119 4120 4121 4122 4123 4124 4125
static bool nfs4_stateid_is_current(nfs4_stateid *stateid,
		const struct nfs_open_context *ctx,
		const struct nfs_lock_context *l_ctx,
		fmode_t fmode)
{
	nfs4_stateid current_stateid;

4126 4127 4128
	/* If the current stateid represents a lost lock, then exit */
	if (nfs4_set_rw_stateid(&current_stateid, ctx, l_ctx, fmode) == -EIO)
		return true;
4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146
	return nfs4_stateid_match(stateid, &current_stateid);
}

static bool nfs4_error_stateid_expired(int err)
{
	switch (err) {
	case -NFS4ERR_DELEG_REVOKED:
	case -NFS4ERR_ADMIN_REVOKED:
	case -NFS4ERR_BAD_STATEID:
	case -NFS4ERR_STALE_STATEID:
	case -NFS4ERR_OLD_STATEID:
	case -NFS4ERR_OPENMODE:
	case -NFS4ERR_EXPIRED:
		return true;
	}
	return false;
}

4147
void __nfs4_read_done_cb(struct nfs_pgio_header *hdr)
4148
{
4149
	nfs_invalidate_atime(hdr->inode);
4150 4151
}

4152
static int nfs4_read_done_cb(struct rpc_task *task, struct nfs_pgio_header *hdr)
L
Linus Torvalds 已提交
4153
{
4154
	struct nfs_server *server = NFS_SERVER(hdr->inode);
L
Linus Torvalds 已提交
4155

4156 4157
	trace_nfs4_read(hdr, task->tk_status);
	if (nfs4_async_handle_error(task, server,
4158 4159
				    hdr->args.context->state,
				    NULL) == -EAGAIN) {
4160
		rpc_restart_call_prepare(task);
T
Trond Myklebust 已提交
4161
		return -EAGAIN;
L
Linus Torvalds 已提交
4162
	}
4163

4164
	__nfs4_read_done_cb(hdr);
L
Linus Torvalds 已提交
4165
	if (task->tk_status > 0)
4166
		renew_lease(server, hdr->timestamp);
T
Trond Myklebust 已提交
4167
	return 0;
L
Linus Torvalds 已提交
4168 4169
}

4170
static bool nfs4_read_stateid_changed(struct rpc_task *task,
4171
		struct nfs_pgio_args *args)
4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183
{

	if (!nfs4_error_stateid_expired(task->tk_status) ||
		nfs4_stateid_is_current(&args->stateid,
				args->context,
				args->lock_context,
				FMODE_READ))
		return false;
	rpc_restart_call_prepare(task);
	return true;
}

4184
static int nfs4_read_done(struct rpc_task *task, struct nfs_pgio_header *hdr)
4185 4186 4187 4188
{

	dprintk("--> %s\n", __func__);

4189
	if (!nfs4_sequence_done(task, &hdr->res.seq_res))
4190
		return -EAGAIN;
4191
	if (nfs4_read_stateid_changed(task, &hdr->args))
4192
		return -EAGAIN;
4193 4194
	return hdr->pgio_done_cb ? hdr->pgio_done_cb(task, hdr) :
				    nfs4_read_done_cb(task, hdr);
4195 4196
}

4197 4198
static void nfs4_proc_read_setup(struct nfs_pgio_header *hdr,
				 struct rpc_message *msg)
L
Linus Torvalds 已提交
4199
{
4200 4201
	hdr->timestamp   = jiffies;
	hdr->pgio_done_cb = nfs4_read_done_cb;
4202
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READ];
4203
	nfs4_init_sequence(&hdr->args.seq_args, &hdr->res.seq_res, 0);
L
Linus Torvalds 已提交
4204 4205
}

4206 4207
static int nfs4_proc_pgio_rpc_prepare(struct rpc_task *task,
				      struct nfs_pgio_header *hdr)
4208
{
4209 4210 4211
	if (nfs4_setup_sequence(NFS_SERVER(hdr->inode),
			&hdr->args.seq_args,
			&hdr->res.seq_res,
4212
			task))
4213
		return 0;
4214 4215 4216
	if (nfs4_set_rw_stateid(&hdr->args.stateid, hdr->args.context,
				hdr->args.lock_context,
				hdr->rw_ops->rw_mode) == -EIO)
4217
		return -EIO;
4218
	if (unlikely(test_bit(NFS_CONTEXT_BAD, &hdr->args.context->flags)))
4219 4220
		return -EIO;
	return 0;
L
Linus Torvalds 已提交
4221 4222
}

4223 4224
static int nfs4_write_done_cb(struct rpc_task *task,
			      struct nfs_pgio_header *hdr)
L
Linus Torvalds 已提交
4225
{
4226
	struct inode *inode = hdr->inode;
4227

4228 4229
	trace_nfs4_write(hdr, task->tk_status);
	if (nfs4_async_handle_error(task, NFS_SERVER(inode),
4230 4231
				    hdr->args.context->state,
				    NULL) == -EAGAIN) {
4232
		rpc_restart_call_prepare(task);
4233
		return -EAGAIN;
L
Linus Torvalds 已提交
4234
	}
4235
	if (task->tk_status >= 0) {
4236
		renew_lease(NFS_SERVER(inode), hdr->timestamp);
4237
		nfs_writeback_update_inode(hdr);
4238
	}
4239
	return 0;
L
Linus Torvalds 已提交
4240 4241
}

4242
static bool nfs4_write_stateid_changed(struct rpc_task *task,
4243
		struct nfs_pgio_args *args)
4244 4245 4246 4247 4248 4249 4250 4251 4252 4253 4254 4255
{

	if (!nfs4_error_stateid_expired(task->tk_status) ||
		nfs4_stateid_is_current(&args->stateid,
				args->context,
				args->lock_context,
				FMODE_WRITE))
		return false;
	rpc_restart_call_prepare(task);
	return true;
}

4256
static int nfs4_write_done(struct rpc_task *task, struct nfs_pgio_header *hdr)
4257
{
4258
	if (!nfs4_sequence_done(task, &hdr->res.seq_res))
4259
		return -EAGAIN;
4260
	if (nfs4_write_stateid_changed(task, &hdr->args))
4261
		return -EAGAIN;
4262 4263
	return hdr->pgio_done_cb ? hdr->pgio_done_cb(task, hdr) :
		nfs4_write_done_cb(task, hdr);
4264 4265
}

4266
static
4267
bool nfs4_write_need_cache_consistency_data(struct nfs_pgio_header *hdr)
4268
{
4269
	/* Don't request attributes for pNFS or O_DIRECT writes */
4270
	if (hdr->ds_clp != NULL || hdr->dreq != NULL)
4271 4272 4273 4274
		return false;
	/* Otherwise, request attributes if and only if we don't hold
	 * a delegation
	 */
4275
	return nfs4_have_delegation(hdr->inode, FMODE_READ) == 0;
4276 4277
}

4278 4279
static void nfs4_proc_write_setup(struct nfs_pgio_header *hdr,
				  struct rpc_message *msg)
L
Linus Torvalds 已提交
4280
{
4281
	struct nfs_server *server = NFS_SERVER(hdr->inode);
4282

4283 4284 4285
	if (!nfs4_write_need_cache_consistency_data(hdr)) {
		hdr->args.bitmask = NULL;
		hdr->res.fattr = NULL;
4286
	} else
4287
		hdr->args.bitmask = server->cache_consistency_bitmask;
4288

4289 4290 4291 4292
	if (!hdr->pgio_done_cb)
		hdr->pgio_done_cb = nfs4_write_done_cb;
	hdr->res.server = server;
	hdr->timestamp   = jiffies;
L
Linus Torvalds 已提交
4293

4294
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_WRITE];
4295
	nfs4_init_sequence(&hdr->args.seq_args, &hdr->res.seq_res, 1);
L
Linus Torvalds 已提交
4296 4297
}

4298
static void nfs4_proc_commit_rpc_prepare(struct rpc_task *task, struct nfs_commit_data *data)
4299
{
4300 4301 4302 4303
	nfs4_setup_sequence(NFS_SERVER(data->inode),
			&data->args.seq_args,
			&data->res.seq_res,
			task);
L
Linus Torvalds 已提交
4304 4305
}

4306
static int nfs4_commit_done_cb(struct rpc_task *task, struct nfs_commit_data *data)
L
Linus Torvalds 已提交
4307 4308
{
	struct inode *inode = data->inode;
4309

4310
	trace_nfs4_commit(data, task->tk_status);
4311 4312
	if (nfs4_async_handle_error(task, NFS_SERVER(inode),
				    NULL, NULL) == -EAGAIN) {
4313
		rpc_restart_call_prepare(task);
4314
		return -EAGAIN;
L
Linus Torvalds 已提交
4315
	}
4316
	return 0;
L
Linus Torvalds 已提交
4317 4318
}

4319
static int nfs4_commit_done(struct rpc_task *task, struct nfs_commit_data *data)
4320 4321 4322
{
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return -EAGAIN;
4323
	return data->commit_done_cb(task, data);
4324 4325
}

4326
static void nfs4_proc_commit_setup(struct nfs_commit_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
4327
{
4328
	struct nfs_server *server = NFS_SERVER(data->inode);
4329

4330 4331
	if (data->commit_done_cb == NULL)
		data->commit_done_cb = nfs4_commit_done_cb;
4332
	data->res.server = server;
4333
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_COMMIT];
4334
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
L
Linus Torvalds 已提交
4335 4336
}

4337 4338 4339 4340 4341
struct nfs4_renewdata {
	struct nfs_client	*client;
	unsigned long		timestamp;
};

L
Linus Torvalds 已提交
4342 4343 4344 4345
/*
 * nfs4_proc_async_renew(): This is not one of the nfs_rpc_ops; it is a special
 * standalone procedure for queueing an asynchronous RENEW.
 */
4346
static void nfs4_renew_release(void *calldata)
4347
{
4348 4349
	struct nfs4_renewdata *data = calldata;
	struct nfs_client *clp = data->client;
4350

4351 4352 4353
	if (atomic_read(&clp->cl_count) > 1)
		nfs4_schedule_state_renewal(clp);
	nfs_put_client(clp);
4354
	kfree(data);
4355 4356
}

4357
static void nfs4_renew_done(struct rpc_task *task, void *calldata)
L
Linus Torvalds 已提交
4358
{
4359 4360 4361
	struct nfs4_renewdata *data = calldata;
	struct nfs_client *clp = data->client;
	unsigned long timestamp = data->timestamp;
L
Linus Torvalds 已提交
4362

4363
	trace_nfs4_renew_async(clp, task->tk_status);
4364 4365 4366 4367 4368 4369 4370
	switch (task->tk_status) {
	case 0:
		break;
	case -NFS4ERR_LEASE_MOVED:
		nfs4_schedule_lease_moved_recovery(clp);
		break;
	default:
4371
		/* Unless we're shutting down, schedule state recovery! */
4372 4373 4374
		if (test_bit(NFS_CS_RENEWD, &clp->cl_res_state) == 0)
			return;
		if (task->tk_status != NFS4ERR_CB_PATH_DOWN) {
4375
			nfs4_schedule_lease_recovery(clp);
4376 4377 4378
			return;
		}
		nfs4_schedule_path_down_recovery(clp);
L
Linus Torvalds 已提交
4379
	}
4380
	do_renew_lease(clp, timestamp);
L
Linus Torvalds 已提交
4381 4382
}

4383 4384
static const struct rpc_call_ops nfs4_renew_ops = {
	.rpc_call_done = nfs4_renew_done,
4385
	.rpc_release = nfs4_renew_release,
4386 4387
};

4388
static int nfs4_proc_async_renew(struct nfs_client *clp, struct rpc_cred *cred, unsigned renew_flags)
L
Linus Torvalds 已提交
4389 4390 4391 4392
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
4393
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
4394
	};
4395
	struct nfs4_renewdata *data;
L
Linus Torvalds 已提交
4396

4397 4398
	if (renew_flags == 0)
		return 0;
4399 4400
	if (!atomic_inc_not_zero(&clp->cl_count))
		return -EIO;
4401
	data = kmalloc(sizeof(*data), GFP_NOFS);
4402 4403 4404 4405
	if (data == NULL)
		return -ENOMEM;
	data->client = clp;
	data->timestamp = jiffies;
4406
	return rpc_call_async(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT,
4407
			&nfs4_renew_ops, data);
L
Linus Torvalds 已提交
4408 4409
}

4410
static int nfs4_proc_renew(struct nfs_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
4411 4412 4413 4414
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
4415
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
4416 4417 4418 4419
	};
	unsigned long now = jiffies;
	int status;

4420
	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
L
Linus Torvalds 已提交
4421 4422
	if (status < 0)
		return status;
4423
	do_renew_lease(clp, now);
L
Linus Torvalds 已提交
4424 4425 4426
	return 0;
}

4427 4428
static inline int nfs4_server_supports_acls(struct nfs_server *server)
{
4429
	return server->caps & NFS_CAP_ACLS;
4430 4431
}

4432 4433
/* Assuming that XATTR_SIZE_MAX is a multiple of PAGE_SIZE, and that
 * it's OK to put sizeof(void) * (XATTR_SIZE_MAX/PAGE_SIZE) bytes on
4434 4435
 * the stack.
 */
4436
#define NFS4ACL_MAXPAGES DIV_ROUND_UP(XATTR_SIZE_MAX, PAGE_SIZE)
4437

4438 4439 4440 4441 4442 4443 4444 4445 4446
static int buf_to_pages_noslab(const void *buf, size_t buflen,
		struct page **pages, unsigned int *pgbase)
{
	struct page *newpage, **spages;
	int rc = 0;
	size_t len;
	spages = pages;

	do {
4447
		len = min_t(size_t, PAGE_SIZE, buflen);
4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466
		newpage = alloc_page(GFP_KERNEL);

		if (newpage == NULL)
			goto unwind;
		memcpy(page_address(newpage), buf, len);
                buf += len;
                buflen -= len;
		*pages++ = newpage;
		rc++;
	} while (buflen != 0);

	return rc;

unwind:
	for(; rc > 0; rc--)
		__free_page(spages[rc-1]);
	return -ENOMEM;
}

4467 4468 4469
struct nfs4_cached_acl {
	int cached;
	size_t len;
A
Andrew Morton 已提交
4470
	char data[0];
4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512
};

static void nfs4_set_cached_acl(struct inode *inode, struct nfs4_cached_acl *acl)
{
	struct nfs_inode *nfsi = NFS_I(inode);

	spin_lock(&inode->i_lock);
	kfree(nfsi->nfs4_acl);
	nfsi->nfs4_acl = acl;
	spin_unlock(&inode->i_lock);
}

static void nfs4_zap_acl_attr(struct inode *inode)
{
	nfs4_set_cached_acl(inode, NULL);
}

static inline ssize_t nfs4_read_cached_acl(struct inode *inode, char *buf, size_t buflen)
{
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs4_cached_acl *acl;
	int ret = -ENOENT;

	spin_lock(&inode->i_lock);
	acl = nfsi->nfs4_acl;
	if (acl == NULL)
		goto out;
	if (buf == NULL) /* user is just asking for length */
		goto out_len;
	if (acl->cached == 0)
		goto out;
	ret = -ERANGE; /* see getxattr(2) man page */
	if (acl->len > buflen)
		goto out;
	memcpy(buf, acl->data, acl->len);
out_len:
	ret = acl->len;
out:
	spin_unlock(&inode->i_lock);
	return ret;
}

4513
static void nfs4_write_cached_acl(struct inode *inode, struct page **pages, size_t pgbase, size_t acl_len)
4514 4515
{
	struct nfs4_cached_acl *acl;
4516
	size_t buflen = sizeof(*acl) + acl_len;
4517

4518
	if (buflen <= PAGE_SIZE) {
4519
		acl = kmalloc(buflen, GFP_KERNEL);
4520 4521 4522
		if (acl == NULL)
			goto out;
		acl->cached = 1;
4523
		_copy_from_pages(acl->data, pages, pgbase, acl_len);
4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534
	} else {
		acl = kmalloc(sizeof(*acl), GFP_KERNEL);
		if (acl == NULL)
			goto out;
		acl->cached = 0;
	}
	acl->len = acl_len;
out:
	nfs4_set_cached_acl(inode, acl);
}

4535 4536 4537 4538 4539 4540 4541 4542 4543 4544
/*
 * The getxattr API returns the required buffer length when called with a
 * NULL buf. The NFSv4 acl tool then calls getxattr again after allocating
 * the required buf.  On a NULL buf, we send a page of data to the server
 * guessing that the ACL request can be serviced by a page. If so, we cache
 * up to the page of ACL data, and the 2nd call to getxattr is serviced by
 * the cache. If not so, we throw away the page, and cache the required
 * length. The next getxattr call will then produce another round trip to
 * the server, this time with the input buf of the required size.
 */
4545
static ssize_t __nfs4_get_acl_uncached(struct inode *inode, void *buf, size_t buflen)
4546
{
4547
	struct page *pages[NFS4ACL_MAXPAGES] = {NULL, };
4548 4549 4550 4551 4552
	struct nfs_getaclargs args = {
		.fh = NFS_FH(inode),
		.acl_pages = pages,
		.acl_len = buflen,
	};
B
Benny Halevy 已提交
4553 4554 4555
	struct nfs_getaclres res = {
		.acl_len = buflen,
	};
4556 4557 4558
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETACL],
		.rpc_argp = &args,
B
Benny Halevy 已提交
4559
		.rpc_resp = &res,
4560
	};
4561 4562
	unsigned int npages = DIV_ROUND_UP(buflen, PAGE_SIZE);
	int ret = -ENOMEM, i;
4563

4564 4565 4566 4567
	/* As long as we're doing a round trip to the server anyway,
	 * let's be prepared for a page of acl data. */
	if (npages == 0)
		npages = 1;
4568 4569
	if (npages > ARRAY_SIZE(pages))
		return -ERANGE;
4570

4571 4572 4573 4574
	for (i = 0; i < npages; i++) {
		pages[i] = alloc_page(GFP_KERNEL);
		if (!pages[i])
			goto out_free;
4575
	}
4576 4577 4578 4579 4580 4581

	/* for decoding across pages */
	res.acl_scratch = alloc_page(GFP_KERNEL);
	if (!res.acl_scratch)
		goto out_free;

4582 4583
	args.acl_len = npages * PAGE_SIZE;
	args.acl_pgbase = 0;
4584

4585
	dprintk("%s  buf %p buflen %zu npages %d args.acl_len %zu\n",
4586 4587 4588
		__func__, buf, buflen, npages, args.acl_len);
	ret = nfs4_call_sync(NFS_SERVER(inode)->client, NFS_SERVER(inode),
			     &msg, &args.seq_args, &res.seq_res, 0);
4589 4590
	if (ret)
		goto out_free;
4591

4592 4593 4594 4595 4596
	/* Handle the case where the passed-in buffer is too short */
	if (res.acl_flags & NFS4_ACL_TRUNC) {
		/* Did the user only issue a request for the acl length? */
		if (buf == NULL)
			goto out_ok;
4597
		ret = -ERANGE;
4598
		goto out_free;
4599
	}
4600
	nfs4_write_cached_acl(inode, pages, res.acl_data_offset, res.acl_len);
4601 4602 4603 4604 4605
	if (buf) {
		if (res.acl_len > buflen) {
			ret = -ERANGE;
			goto out_free;
		}
4606
		_copy_from_pages(buf, pages, res.acl_data_offset, res.acl_len);
4607
	}
4608 4609
out_ok:
	ret = res.acl_len;
4610
out_free:
4611 4612 4613
	for (i = 0; i < npages; i++)
		if (pages[i])
			__free_page(pages[i]);
4614 4615
	if (res.acl_scratch)
		__free_page(res.acl_scratch);
4616 4617 4618
	return ret;
}

4619 4620 4621 4622 4623 4624
static ssize_t nfs4_get_acl_uncached(struct inode *inode, void *buf, size_t buflen)
{
	struct nfs4_exception exception = { };
	ssize_t ret;
	do {
		ret = __nfs4_get_acl_uncached(inode, buf, buflen);
4625
		trace_nfs4_get_acl(inode, ret);
4626 4627 4628 4629 4630 4631 4632
		if (ret >= 0)
			break;
		ret = nfs4_handle_exception(NFS_SERVER(inode), ret, &exception);
	} while (exception.retry);
	return ret;
}

4633 4634 4635 4636 4637 4638 4639 4640 4641 4642
static ssize_t nfs4_proc_get_acl(struct inode *inode, void *buf, size_t buflen)
{
	struct nfs_server *server = NFS_SERVER(inode);
	int ret;

	if (!nfs4_server_supports_acls(server))
		return -EOPNOTSUPP;
	ret = nfs_revalidate_inode(server, inode);
	if (ret < 0)
		return ret;
4643 4644
	if (NFS_I(inode)->cache_validity & NFS_INO_INVALID_ACL)
		nfs_zap_acl_cache(inode);
4645 4646
	ret = nfs4_read_cached_acl(inode, buf, buflen);
	if (ret != -ENOENT)
4647 4648
		/* -ENOENT is returned if there is no ACL or if there is an ACL
		 * but no cached acl data, just the acl length */
4649 4650 4651 4652
		return ret;
	return nfs4_get_acl_uncached(inode, buf, buflen);
}

4653
static int __nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen)
4654 4655 4656 4657 4658 4659 4660 4661
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct page *pages[NFS4ACL_MAXPAGES];
	struct nfs_setaclargs arg = {
		.fh		= NFS_FH(inode),
		.acl_pages	= pages,
		.acl_len	= buflen,
	};
B
Benny Halevy 已提交
4662
	struct nfs_setaclres res;
4663 4664 4665
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETACL],
		.rpc_argp	= &arg,
B
Benny Halevy 已提交
4666
		.rpc_resp	= &res,
4667
	};
4668
	unsigned int npages = DIV_ROUND_UP(buflen, PAGE_SIZE);
4669
	int ret, i;
4670 4671 4672

	if (!nfs4_server_supports_acls(server))
		return -EOPNOTSUPP;
4673 4674
	if (npages > ARRAY_SIZE(pages))
		return -ERANGE;
4675 4676 4677
	i = buf_to_pages_noslab(buf, buflen, arg.acl_pages, &arg.acl_pgbase);
	if (i < 0)
		return i;
4678
	nfs4_inode_return_delegation(inode);
4679
	ret = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
4680 4681 4682 4683 4684 4685 4686 4687

	/*
	 * Free each page after tx, so the only ref left is
	 * held by the network stack
	 */
	for (; i > 0; i--)
		put_page(pages[i-1]);

4688 4689 4690 4691 4692 4693 4694
	/*
	 * Acl update can result in inode attribute update.
	 * so mark the attribute cache invalid.
	 */
	spin_lock(&inode->i_lock);
	NFS_I(inode)->cache_validity |= NFS_INO_INVALID_ATTR;
	spin_unlock(&inode->i_lock);
4695 4696
	nfs_access_zap_cache(inode);
	nfs_zap_acl_cache(inode);
4697 4698 4699
	return ret;
}

4700 4701 4702 4703 4704
static int nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen)
{
	struct nfs4_exception exception = { };
	int err;
	do {
4705 4706 4707
		err = __nfs4_proc_set_acl(inode, buf, buflen);
		trace_nfs4_set_acl(inode, err);
		err = nfs4_handle_exception(NFS_SERVER(inode), err,
4708 4709 4710 4711 4712
				&exception);
	} while (exception.retry);
	return err;
}

4713 4714 4715 4716 4717 4718 4719 4720 4721
#ifdef CONFIG_NFS_V4_SECURITY_LABEL
static int _nfs4_get_security_label(struct inode *inode, void *buf,
					size_t buflen)
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs_fattr fattr;
	struct nfs4_label label = {0, 0, buflen, buf};

	u32 bitmask[3] = { 0, 0, FATTR4_WORD2_SECURITY_LABEL };
4722
	struct nfs4_getattr_arg arg = {
4723 4724 4725 4726 4727 4728 4729 4730 4731 4732
		.fh		= NFS_FH(inode),
		.bitmask	= bitmask,
	};
	struct nfs4_getattr_res res = {
		.fattr		= &fattr,
		.label		= &label,
		.server		= server,
	};
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_GETATTR],
4733
		.rpc_argp	= &arg,
4734 4735 4736 4737 4738 4739
		.rpc_resp	= &res,
	};
	int ret;

	nfs_fattr_init(&fattr);

4740
	ret = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 0);
4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759
	if (ret)
		return ret;
	if (!(fattr.valid & NFS_ATTR_FATTR_V4_SECURITY_LABEL))
		return -ENOENT;
	if (buflen < label.len)
		return -ERANGE;
	return 0;
}

static int nfs4_get_security_label(struct inode *inode, void *buf,
					size_t buflen)
{
	struct nfs4_exception exception = { };
	int err;

	if (!nfs_server_capable(inode, NFS_CAP_SECURITY_LABEL))
		return -EOPNOTSUPP;

	do {
4760 4761 4762
		err = _nfs4_get_security_label(inode, buf, buflen);
		trace_nfs4_get_security_label(inode, err);
		err = nfs4_handle_exception(NFS_SERVER(inode), err,
4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_do_set_security_label(struct inode *inode,
		struct nfs4_label *ilabel,
		struct nfs_fattr *fattr,
		struct nfs4_label *olabel)
{

	struct iattr sattr = {0};
	struct nfs_server *server = NFS_SERVER(inode);
	const u32 bitmask[3] = { 0, 0, FATTR4_WORD2_SECURITY_LABEL };
4777
	struct nfs_setattrargs arg = {
4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790
		.fh             = NFS_FH(inode),
		.iap            = &sattr,
		.server		= server,
		.bitmask	= bitmask,
		.label		= ilabel,
	};
	struct nfs_setattrres res = {
		.fattr		= fattr,
		.label		= olabel,
		.server		= server,
	};
	struct rpc_message msg = {
		.rpc_proc       = &nfs4_procedures[NFSPROC4_CLNT_SETATTR],
4791
		.rpc_argp       = &arg,
4792 4793 4794 4795
		.rpc_resp       = &res,
	};
	int status;

4796
	nfs4_stateid_copy(&arg.stateid, &zero_stateid);
4797

4798
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813
	if (status)
		dprintk("%s failed: %d\n", __func__, status);

	return status;
}

static int nfs4_do_set_security_label(struct inode *inode,
		struct nfs4_label *ilabel,
		struct nfs_fattr *fattr,
		struct nfs4_label *olabel)
{
	struct nfs4_exception exception = { };
	int err;

	do {
4814 4815 4816 4817
		err = _nfs4_do_set_security_label(inode, ilabel,
				fattr, olabel);
		trace_nfs4_set_security_label(inode, err);
		err = nfs4_handle_exception(NFS_SERVER(inode), err,
4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828
				&exception);
	} while (exception.retry);
	return err;
}

static int
nfs4_set_security_label(struct dentry *dentry, const void *buf, size_t buflen)
{
	struct nfs4_label ilabel, *olabel = NULL;
	struct nfs_fattr fattr;
	struct rpc_cred *cred;
4829
	struct inode *inode = d_inode(dentry);
4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863
	int status;

	if (!nfs_server_capable(inode, NFS_CAP_SECURITY_LABEL))
		return -EOPNOTSUPP;

	nfs_fattr_init(&fattr);

	ilabel.pi = 0;
	ilabel.lfs = 0;
	ilabel.label = (char *)buf;
	ilabel.len = buflen;

	cred = rpc_lookup_cred();
	if (IS_ERR(cred))
		return PTR_ERR(cred);

	olabel = nfs4_label_alloc(NFS_SERVER(inode), GFP_KERNEL);
	if (IS_ERR(olabel)) {
		status = -PTR_ERR(olabel);
		goto out;
	}

	status = nfs4_do_set_security_label(inode, &ilabel, &fattr, olabel);
	if (status == 0)
		nfs_setsecurity(inode, &fattr, olabel);

	nfs4_label_free(olabel);
out:
	put_rpccred(cred);
	return status;
}
#endif	/* CONFIG_NFS_V4_SECURITY_LABEL */


L
Linus Torvalds 已提交
4864
static int
4865 4866
nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server,
			struct nfs4_state *state, long *timeout)
L
Linus Torvalds 已提交
4867
{
4868 4869 4870
	struct nfs_client *clp = server->nfs_client;

	if (task->tk_status >= 0)
L
Linus Torvalds 已提交
4871 4872
		return 0;
	switch(task->tk_status) {
4873
		case -NFS4ERR_DELEG_REVOKED:
4874 4875 4876 4877 4878
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OPENMODE:
			if (state == NULL)
				break;
4879
			if (nfs4_schedule_stateid_recovery(server, state) < 0)
C
Chuck Lever 已提交
4880
				goto recovery_failed;
4881
			goto wait_on_recovery;
4882
		case -NFS4ERR_EXPIRED:
4883 4884
			if (state != NULL) {
				if (nfs4_schedule_stateid_recovery(server, state) < 0)
C
Chuck Lever 已提交
4885
					goto recovery_failed;
4886
			}
L
Linus Torvalds 已提交
4887
		case -NFS4ERR_STALE_STATEID:
4888
		case -NFS4ERR_STALE_CLIENTID:
4889 4890
			nfs4_schedule_lease_recovery(clp);
			goto wait_on_recovery;
4891 4892 4893 4894
		case -NFS4ERR_MOVED:
			if (nfs4_schedule_migration_recovery(server) < 0)
				goto recovery_failed;
			goto wait_on_recovery;
4895 4896 4897
		case -NFS4ERR_LEASE_MOVED:
			nfs4_schedule_lease_moved_recovery(clp);
			goto wait_on_recovery;
4898 4899 4900 4901 4902 4903 4904 4905 4906 4907
#if defined(CONFIG_NFS_V4_1)
		case -NFS4ERR_BADSESSION:
		case -NFS4ERR_BADSLOT:
		case -NFS4ERR_BAD_HIGH_SLOT:
		case -NFS4ERR_DEADSESSION:
		case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
		case -NFS4ERR_SEQ_FALSE_RETRY:
		case -NFS4ERR_SEQ_MISORDERED:
			dprintk("%s ERROR %d, Reset session\n", __func__,
				task->tk_status);
4908
			nfs4_schedule_session_recovery(clp->cl_session, task->tk_status);
4909
			goto wait_on_recovery;
4910
#endif /* CONFIG_NFS_V4_1 */
L
Linus Torvalds 已提交
4911
		case -NFS4ERR_DELAY:
4912
			nfs_inc_server_stats(server, NFSIOS_DELAY);
4913 4914
			rpc_delay(task, nfs4_update_delay(timeout));
			goto restart_call;
4915
		case -NFS4ERR_GRACE:
L
Linus Torvalds 已提交
4916
			rpc_delay(task, NFS4_POLL_RETRY_MAX);
4917
		case -NFS4ERR_RETRY_UNCACHED_REP:
L
Linus Torvalds 已提交
4918
		case -NFS4ERR_OLD_STATEID:
4919
			goto restart_call;
L
Linus Torvalds 已提交
4920 4921 4922
	}
	task->tk_status = nfs4_map_errors(task->tk_status);
	return 0;
C
Chuck Lever 已提交
4923
recovery_failed:
4924 4925
	task->tk_status = -EIO;
	return 0;
4926
wait_on_recovery:
4927 4928 4929
	rpc_sleep_on(&clp->cl_rpcwaitq, task, NULL);
	if (test_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state) == 0)
		rpc_wake_up_queued_task(&clp->cl_rpcwaitq, task);
4930 4931
	if (test_bit(NFS_MIG_FAILED, &server->mig_status))
		goto recovery_failed;
4932
restart_call:
4933 4934
	task->tk_status = 0;
	return -EAGAIN;
L
Linus Torvalds 已提交
4935 4936
}

4937 4938
static void nfs4_init_boot_verifier(const struct nfs_client *clp,
				    nfs4_verifier *bootverf)
4939 4940 4941
{
	__be32 verf[2];

4942 4943 4944 4945
	if (test_bit(NFS4CLNT_PURGE_STATE, &clp->cl_state)) {
		/* An impossible timestamp guarantees this value
		 * will never match a generated boot time. */
		verf[0] = 0;
4946
		verf[1] = cpu_to_be32(NSEC_PER_SEC + 1);
4947
	} else {
4948
		struct nfs_net *nn = net_generic(clp->cl_net, nfs_net_id);
4949 4950
		verf[0] = cpu_to_be32(nn->boot_time.tv_sec);
		verf[1] = cpu_to_be32(nn->boot_time.tv_nsec);
4951
	}
4952 4953 4954
	memcpy(bootverf->data, verf, sizeof(bootverf->data));
}

4955 4956
static int
nfs4_init_nonuniform_client_string(struct nfs_client *clp)
4957
{
4958 4959 4960 4961
	int result;
	size_t len;
	char *str;
	bool retried = false;
4962

4963
	if (clp->cl_owner_id != NULL)
4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984
		return 0;
retry:
	rcu_read_lock();
	len = 10 + strlen(clp->cl_ipaddr) + 1 +
		strlen(rpc_peeraddr2str(clp->cl_rpcclient, RPC_DISPLAY_ADDR)) +
		1 +
		strlen(rpc_peeraddr2str(clp->cl_rpcclient, RPC_DISPLAY_PROTO)) +
		1;
	rcu_read_unlock();

	if (len > NFS4_OPAQUE_LIMIT + 1)
		return -EINVAL;

	/*
	 * Since this string is allocated at mount time, and held until the
	 * nfs_client is destroyed, we can use GFP_KERNEL here w/o worrying
	 * about a memory-reclaim deadlock.
	 */
	str = kmalloc(len, GFP_KERNEL);
	if (!str)
		return -ENOMEM;
4985

4986
	rcu_read_lock();
4987 4988 4989 4990
	result = scnprintf(str, len, "Linux NFSv4.0 %s/%s %s",
			clp->cl_ipaddr,
			rpc_peeraddr2str(clp->cl_rpcclient, RPC_DISPLAY_ADDR),
			rpc_peeraddr2str(clp->cl_rpcclient, RPC_DISPLAY_PROTO));
4991
	rcu_read_unlock();
4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002

	/* Did something change? */
	if (result >= len) {
		kfree(str);
		if (retried)
			return -EINVAL;
		retried = true;
		goto retry;
	}
	clp->cl_owner_id = str;
	return 0;
5003 5004
}

5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041
static int
nfs4_init_uniquifier_client_string(struct nfs_client *clp)
{
	int result;
	size_t len;
	char *str;

	len = 10 + 10 + 1 + 10 + 1 +
		strlen(nfs4_client_id_uniquifier) + 1 +
		strlen(clp->cl_rpcclient->cl_nodename) + 1;

	if (len > NFS4_OPAQUE_LIMIT + 1)
		return -EINVAL;

	/*
	 * Since this string is allocated at mount time, and held until the
	 * nfs_client is destroyed, we can use GFP_KERNEL here w/o worrying
	 * about a memory-reclaim deadlock.
	 */
	str = kmalloc(len, GFP_KERNEL);
	if (!str)
		return -ENOMEM;

	result = scnprintf(str, len, "Linux NFSv%u.%u %s/%s",
			clp->rpc_ops->version, clp->cl_minorversion,
			nfs4_client_id_uniquifier,
			clp->cl_rpcclient->cl_nodename);
	if (result >= len) {
		kfree(str);
		return -EINVAL;
	}
	clp->cl_owner_id = str;
	return 0;
}

static int
nfs4_init_uniform_client_string(struct nfs_client *clp)
5042
{
5043 5044 5045
	int result;
	size_t len;
	char *str;
5046 5047

	if (clp->cl_owner_id != NULL)
5048
		return 0;
5049 5050

	if (nfs4_client_id_uniquifier[0] != '\0')
5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076
		return nfs4_init_uniquifier_client_string(clp);

	len = 10 + 10 + 1 + 10 + 1 +
		strlen(clp->cl_rpcclient->cl_nodename) + 1;

	if (len > NFS4_OPAQUE_LIMIT + 1)
		return -EINVAL;

	/*
	 * Since this string is allocated at mount time, and held until the
	 * nfs_client is destroyed, we can use GFP_KERNEL here w/o worrying
	 * about a memory-reclaim deadlock.
	 */
	str = kmalloc(len, GFP_KERNEL);
	if (!str)
		return -ENOMEM;

	result = scnprintf(str, len, "Linux NFSv%u.%u %s",
			clp->rpc_ops->version, clp->cl_minorversion,
			clp->cl_rpcclient->cl_nodename);
	if (result >= len) {
		kfree(str);
		return -EINVAL;
	}
	clp->cl_owner_id = str;
	return 0;
5077 5078
}

5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092
/*
 * nfs4_callback_up_net() starts only "tcp" and "tcp6" callback
 * services.  Advertise one based on the address family of the
 * clientaddr.
 */
static unsigned int
nfs4_init_callback_netid(const struct nfs_client *clp, char *buf, size_t len)
{
	if (strchr(clp->cl_ipaddr, ':') != NULL)
		return scnprintf(buf, len, "tcp6");
	else
		return scnprintf(buf, len, "tcp");
}

5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104
static void nfs4_setclientid_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_setclientid *sc = calldata;

	if (task->tk_status == 0)
		sc->sc_cred = get_rpccred(task->tk_rqstp->rq_cred);
}

static const struct rpc_call_ops nfs4_setclientid_ops = {
	.rpc_call_done = nfs4_setclientid_done,
};

5105 5106 5107 5108 5109 5110 5111 5112 5113 5114
/**
 * nfs4_proc_setclientid - Negotiate client ID
 * @clp: state data structure
 * @program: RPC program for NFSv4 callback service
 * @port: IP port number for NFS4 callback service
 * @cred: RPC credential to use for this call
 * @res: where to place the result
 *
 * Returns zero, a negative errno, or a negative NFS4ERR status code.
 */
5115 5116 5117
int nfs4_proc_setclientid(struct nfs_client *clp, u32 program,
		unsigned short port, struct rpc_cred *cred,
		struct nfs4_setclientid_res *res)
L
Linus Torvalds 已提交
5118 5119 5120 5121 5122
{
	nfs4_verifier sc_verifier;
	struct nfs4_setclientid setclientid = {
		.sc_verifier = &sc_verifier,
		.sc_prog = program,
5123
		.sc_clnt = clp,
L
Linus Torvalds 已提交
5124 5125 5126 5127
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID],
		.rpc_argp = &setclientid,
5128
		.rpc_resp = res,
5129
		.rpc_cred = cred,
L
Linus Torvalds 已提交
5130
	};
5131 5132 5133 5134 5135 5136 5137 5138
	struct rpc_task *task;
	struct rpc_task_setup task_setup_data = {
		.rpc_client = clp->cl_rpcclient,
		.rpc_message = &msg,
		.callback_ops = &nfs4_setclientid_ops,
		.callback_data = &setclientid,
		.flags = RPC_TASK_TIMEOUT,
	};
5139
	int status;
L
Linus Torvalds 已提交
5140

5141
	/* nfs_client_id4 */
5142
	nfs4_init_boot_verifier(clp, &sc_verifier);
5143 5144 5145 5146

	if (test_bit(NFS_CS_MIGRATION, &clp->cl_flags))
		status = nfs4_init_uniform_client_string(clp);
	else
5147
		status = nfs4_init_nonuniform_client_string(clp);
5148 5149 5150

	if (status)
		goto out;
5151

5152
	/* cb_client4 */
5153 5154 5155 5156
	setclientid.sc_netid_len =
				nfs4_init_callback_netid(clp,
						setclientid.sc_netid,
						sizeof(setclientid.sc_netid));
5157
	setclientid.sc_uaddr_len = scnprintf(setclientid.sc_uaddr,
5158
				sizeof(setclientid.sc_uaddr), "%s.%u.%u",
L
Linus Torvalds 已提交
5159 5160
				clp->cl_ipaddr, port >> 8, port & 255);

5161
	dprintk("NFS call  setclientid auth=%s, '%s'\n",
5162
		clp->cl_rpcclient->cl_auth->au_ops->au_name,
5163
		clp->cl_owner_id);
5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175
	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task)) {
		status = PTR_ERR(task);
		goto out;
	}
	status = task->tk_status;
	if (setclientid.sc_cred) {
		clp->cl_acceptor = rpcauth_stringify_acceptor(setclientid.sc_cred);
		put_rpccred(setclientid.sc_cred);
	}
	rpc_put_task(task);
out:
5176
	trace_nfs4_setclientid(clp, status);
5177 5178
	dprintk("NFS reply setclientid: %d\n", status);
	return status;
L
Linus Torvalds 已提交
5179 5180
}

5181 5182 5183 5184 5185 5186 5187 5188
/**
 * nfs4_proc_setclientid_confirm - Confirm client ID
 * @clp: state data structure
 * @res: result of a previous SETCLIENTID
 * @cred: RPC credential to use for this call
 *
 * Returns zero, a negative errno, or a negative NFS4ERR status code.
 */
5189
int nfs4_proc_setclientid_confirm(struct nfs_client *clp,
5190 5191
		struct nfs4_setclientid_res *arg,
		struct rpc_cred *cred)
L
Linus Torvalds 已提交
5192 5193 5194
{
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID_CONFIRM],
5195
		.rpc_argp = arg,
5196
		.rpc_cred = cred,
L
Linus Torvalds 已提交
5197 5198 5199
	};
	int status;

5200 5201 5202
	dprintk("NFS call  setclientid_confirm auth=%s, (client ID %llx)\n",
		clp->cl_rpcclient->cl_auth->au_ops->au_name,
		clp->cl_clientid);
5203
	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
5204
	trace_nfs4_setclientid_confirm(clp, status);
5205
	dprintk("NFS reply setclientid_confirm: %d\n", status);
L
Linus Torvalds 已提交
5206 5207 5208
	return status;
}

5209 5210
struct nfs4_delegreturndata {
	struct nfs4_delegreturnargs args;
5211
	struct nfs4_delegreturnres res;
5212 5213
	struct nfs_fh fh;
	nfs4_stateid stateid;
5214
	unsigned long timestamp;
5215
	struct nfs_fattr fattr;
5216
	int rpc_status;
5217 5218 5219
	struct inode *inode;
	bool roc;
	u32 roc_barrier;
5220 5221 5222 5223 5224
};

static void nfs4_delegreturn_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_delegreturndata *data = calldata;
5225

5226 5227
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return;
5228

5229
	trace_nfs4_delegreturn_exit(&data->args, &data->res, task->tk_status);
5230 5231
	switch (task->tk_status) {
	case 0:
5232
		renew_lease(data->res.server, data->timestamp);
5233 5234 5235 5236 5237 5238 5239
	case -NFS4ERR_ADMIN_REVOKED:
	case -NFS4ERR_DELEG_REVOKED:
	case -NFS4ERR_BAD_STATEID:
	case -NFS4ERR_OLD_STATEID:
	case -NFS4ERR_STALE_STATEID:
	case -NFS4ERR_EXPIRED:
		task->tk_status = 0;
5240 5241
		if (data->roc)
			pnfs_roc_set_barrier(data->inode, data->roc_barrier);
5242
		break;
5243
	default:
5244 5245
		if (nfs4_async_handle_error(task, data->res.server,
					    NULL, NULL) == -EAGAIN) {
5246
			rpc_restart_call_prepare(task);
5247 5248 5249 5250
			return;
		}
	}
	data->rpc_status = task->tk_status;
5251 5252 5253 5254
}

static void nfs4_delegreturn_release(void *calldata)
{
5255
	struct nfs4_delegreturndata *data = calldata;
5256
	struct inode *inode = data->inode;
5257

5258 5259 5260 5261 5262
	if (inode) {
		if (data->roc)
			pnfs_roc_release(inode);
		nfs_iput_and_deactive(inode);
	}
5263 5264 5265
	kfree(calldata);
}

5266 5267 5268 5269 5270 5271
static void nfs4_delegreturn_prepare(struct rpc_task *task, void *data)
{
	struct nfs4_delegreturndata *d_data;

	d_data = (struct nfs4_delegreturndata *)data;

5272 5273 5274 5275
	if (d_data->roc &&
	    pnfs_roc_drain(d_data->inode, &d_data->roc_barrier, task))
		return;

5276 5277 5278 5279
	nfs4_setup_sequence(d_data->res.server,
			&d_data->args.seq_args,
			&d_data->res.seq_res,
			task);
5280 5281
}

5282
static const struct rpc_call_ops nfs4_delegreturn_ops = {
5283
	.rpc_call_prepare = nfs4_delegreturn_prepare,
5284 5285 5286 5287
	.rpc_call_done = nfs4_delegreturn_done,
	.rpc_release = nfs4_delegreturn_release,
};

5288
static int _nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
5289 5290
{
	struct nfs4_delegreturndata *data;
5291
	struct nfs_server *server = NFS_SERVER(inode);
5292
	struct rpc_task *task;
5293 5294 5295 5296
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DELEGRETURN],
		.rpc_cred = cred,
	};
T
Trond Myklebust 已提交
5297 5298
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
5299
		.rpc_message = &msg,
T
Trond Myklebust 已提交
5300 5301 5302
		.callback_ops = &nfs4_delegreturn_ops,
		.flags = RPC_TASK_ASYNC,
	};
5303
	int status = 0;
5304

5305
	data = kzalloc(sizeof(*data), GFP_NOFS);
5306 5307
	if (data == NULL)
		return -ENOMEM;
5308
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
5309 5310
	data->args.fhandle = &data->fh;
	data->args.stateid = &data->stateid;
5311
	data->args.bitmask = server->cache_consistency_bitmask;
5312
	nfs_copy_fh(&data->fh, NFS_FH(inode));
5313
	nfs4_stateid_copy(&data->stateid, stateid);
5314 5315
	data->res.fattr = &data->fattr;
	data->res.server = server;
5316
	nfs_fattr_init(data->res.fattr);
5317
	data->timestamp = jiffies;
5318
	data->rpc_status = 0;
5319 5320 5321
	data->inode = nfs_igrab_and_active(inode);
	if (data->inode)
		data->roc = nfs4_roc(inode);
5322

T
Trond Myklebust 已提交
5323
	task_setup_data.callback_data = data;
5324 5325
	msg.rpc_argp = &data->args;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
5326
	task = rpc_run_task(&task_setup_data);
5327
	if (IS_ERR(task))
5328
		return PTR_ERR(task);
5329 5330
	if (!issync)
		goto out;
5331
	status = nfs4_wait_for_completion_rpc_task(task);
5332 5333 5334
	if (status != 0)
		goto out;
	status = data->rpc_status;
5335 5336 5337 5338
	if (status == 0)
		nfs_post_op_update_inode_force_wcc(inode, &data->fattr);
	else
		nfs_refresh_inode(inode, &data->fattr);
5339
out:
5340
	rpc_put_task(task);
5341
	return status;
L
Linus Torvalds 已提交
5342 5343
}

5344
int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
L
Linus Torvalds 已提交
5345 5346 5347 5348 5349
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs4_exception exception = { };
	int err;
	do {
5350
		err = _nfs4_proc_delegreturn(inode, cred, stateid, issync);
5351
		trace_nfs4_delegreturn(inode, err);
L
Linus Torvalds 已提交
5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371
		switch (err) {
			case -NFS4ERR_STALE_STATEID:
			case -NFS4ERR_EXPIRED:
			case 0:
				return 0;
		}
		err = nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
	return err;
}

#define NFS4_LOCK_MINTIMEOUT (1 * HZ)
#define NFS4_LOCK_MAXTIMEOUT (30 * HZ)

/* 
 * sleep, with exponential backoff, and retry the LOCK operation. 
 */
static unsigned long
nfs4_set_lock_task_retry(unsigned long timeout)
{
5372
	freezable_schedule_timeout_killable_unsafe(timeout);
L
Linus Torvalds 已提交
5373 5374 5375 5376 5377 5378 5379 5380 5381 5382
	timeout <<= 1;
	if (timeout > NFS4_LOCK_MAXTIMEOUT)
		return NFS4_LOCK_MAXTIMEOUT;
	return timeout;
}

static int _nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
	struct inode *inode = state->inode;
	struct nfs_server *server = NFS_SERVER(inode);
5383
	struct nfs_client *clp = server->nfs_client;
T
Trond Myklebust 已提交
5384
	struct nfs_lockt_args arg = {
L
Linus Torvalds 已提交
5385
		.fh = NFS_FH(inode),
T
Trond Myklebust 已提交
5386
		.fl = request,
L
Linus Torvalds 已提交
5387
	};
T
Trond Myklebust 已提交
5388 5389
	struct nfs_lockt_res res = {
		.denied = request,
L
Linus Torvalds 已提交
5390 5391 5392 5393 5394 5395 5396 5397 5398 5399
	};
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_LOCKT],
		.rpc_argp       = &arg,
		.rpc_resp       = &res,
		.rpc_cred	= state->owner->so_cred,
	};
	struct nfs4_lock_state *lsp;
	int status;

T
Trond Myklebust 已提交
5400
	arg.lock_owner.clientid = clp->cl_clientid;
5401 5402 5403 5404
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
	lsp = request->fl_u.nfs4_fl.owner;
5405
	arg.lock_owner.id = lsp->ls_seqid.owner_id;
5406
	arg.lock_owner.s_dev = server->s_dev;
5407
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
T
Trond Myklebust 已提交
5408 5409 5410 5411 5412 5413
	switch (status) {
		case 0:
			request->fl_type = F_UNLCK;
			break;
		case -NFS4ERR_DENIED:
			status = 0;
L
Linus Torvalds 已提交
5414
	}
5415
	request->fl_ops->fl_release_private(request);
5416
	request->fl_ops = NULL;
5417
out:
L
Linus Torvalds 已提交
5418 5419 5420 5421 5422 5423 5424 5425 5426
	return status;
}

static int nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
	struct nfs4_exception exception = { };
	int err;

	do {
5427 5428 5429
		err = _nfs4_proc_getlk(state, cmd, request);
		trace_nfs4_get_lock(request, state, cmd, err);
		err = nfs4_handle_exception(NFS_SERVER(state->inode), err,
L
Linus Torvalds 已提交
5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450
				&exception);
	} while (exception.retry);
	return err;
}

static int do_vfs_lock(struct file *file, struct file_lock *fl)
{
	int res = 0;
	switch (fl->fl_flags & (FL_POSIX|FL_FLOCK)) {
		case FL_POSIX:
			res = posix_lock_file_wait(file, fl);
			break;
		case FL_FLOCK:
			res = flock_lock_file_wait(file, fl);
			break;
		default:
			BUG();
	}
	return res;
}

5451
struct nfs4_unlockdata {
T
Trond Myklebust 已提交
5452 5453
	struct nfs_locku_args arg;
	struct nfs_locku_res res;
5454 5455
	struct nfs4_lock_state *lsp;
	struct nfs_open_context *ctx;
T
Trond Myklebust 已提交
5456 5457
	struct file_lock fl;
	const struct nfs_server *server;
5458
	unsigned long timestamp;
5459 5460
};

T
Trond Myklebust 已提交
5461 5462 5463 5464 5465 5466 5467 5468
static struct nfs4_unlockdata *nfs4_alloc_unlockdata(struct file_lock *fl,
		struct nfs_open_context *ctx,
		struct nfs4_lock_state *lsp,
		struct nfs_seqid *seqid)
{
	struct nfs4_unlockdata *p;
	struct inode *inode = lsp->ls_state->inode;

5469
	p = kzalloc(sizeof(*p), GFP_NOFS);
T
Trond Myklebust 已提交
5470 5471 5472 5473 5474
	if (p == NULL)
		return NULL;
	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
	p->arg.seqid = seqid;
5475
	p->res.seqid = seqid;
T
Trond Myklebust 已提交
5476 5477 5478 5479 5480 5481 5482 5483 5484
	p->lsp = lsp;
	atomic_inc(&lsp->ls_count);
	/* Ensure we don't close file until we're done freeing locks! */
	p->ctx = get_nfs_open_context(ctx);
	memcpy(&p->fl, fl, sizeof(p->fl));
	p->server = NFS_SERVER(inode);
	return p;
}

5485
static void nfs4_locku_release_calldata(void *data)
5486
{
5487
	struct nfs4_unlockdata *calldata = data;
T
Trond Myklebust 已提交
5488
	nfs_free_seqid(calldata->arg.seqid);
5489 5490 5491
	nfs4_put_lock_state(calldata->lsp);
	put_nfs_open_context(calldata->ctx);
	kfree(calldata);
5492 5493
}

5494
static void nfs4_locku_done(struct rpc_task *task, void *data)
5495
{
5496
	struct nfs4_unlockdata *calldata = data;
5497

5498 5499
	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
		return;
5500 5501
	switch (task->tk_status) {
		case 0:
5502
			renew_lease(calldata->server, calldata->timestamp);
5503 5504 5505 5506
			do_vfs_lock(calldata->fl.fl_file, &calldata->fl);
			if (nfs4_update_lock_stateid(calldata->lsp,
					&calldata->res.stateid))
				break;
5507 5508
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OLD_STATEID:
5509 5510
		case -NFS4ERR_STALE_STATEID:
		case -NFS4ERR_EXPIRED:
5511 5512 5513
			if (!nfs4_stateid_match(&calldata->arg.stateid,
						&calldata->lsp->ls_stateid))
				rpc_restart_call_prepare(task);
5514 5515
			break;
		default:
5516 5517
			if (nfs4_async_handle_error(task, calldata->server,
						    NULL, NULL) == -EAGAIN)
5518
				rpc_restart_call_prepare(task);
5519
	}
5520
	nfs_release_seqid(calldata->arg.seqid);
5521 5522
}

T
Trond Myklebust 已提交
5523
static void nfs4_locku_prepare(struct rpc_task *task, void *data)
5524
{
T
Trond Myklebust 已提交
5525
	struct nfs4_unlockdata *calldata = data;
5526

T
Trond Myklebust 已提交
5527
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
5528
		goto out_wait;
5529
	nfs4_stateid_copy(&calldata->arg.stateid, &calldata->lsp->ls_stateid);
5530
	if (test_bit(NFS_LOCK_INITIALIZED, &calldata->lsp->ls_flags) == 0) {
5531
		/* Note: exit _without_ running nfs4_locku_done */
5532
		goto out_no_action;
5533
	}
5534
	calldata->timestamp = jiffies;
5535
	if (nfs4_setup_sequence(calldata->server,
5536
				&calldata->arg.seq_args,
5537 5538 5539
				&calldata->res.seq_res,
				task) != 0)
		nfs_release_seqid(calldata->arg.seqid);
5540 5541 5542 5543 5544
	return;
out_no_action:
	task->tk_action = NULL;
out_wait:
	nfs4_sequence_done(task, &calldata->res.seq_res);
5545 5546
}

5547
static const struct rpc_call_ops nfs4_locku_ops = {
T
Trond Myklebust 已提交
5548
	.rpc_call_prepare = nfs4_locku_prepare,
5549
	.rpc_call_done = nfs4_locku_done,
5550
	.rpc_release = nfs4_locku_release_calldata,
5551 5552
};

5553 5554 5555 5556 5557 5558
static struct rpc_task *nfs4_do_unlck(struct file_lock *fl,
		struct nfs_open_context *ctx,
		struct nfs4_lock_state *lsp,
		struct nfs_seqid *seqid)
{
	struct nfs4_unlockdata *data;
5559 5560 5561 5562
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCKU],
		.rpc_cred = ctx->cred,
	};
T
Trond Myklebust 已提交
5563 5564
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(lsp->ls_state->inode),
5565
		.rpc_message = &msg,
T
Trond Myklebust 已提交
5566
		.callback_ops = &nfs4_locku_ops,
5567
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
5568 5569
		.flags = RPC_TASK_ASYNC,
	};
5570

5571 5572 5573
	nfs4_state_protect(NFS_SERVER(lsp->ls_state->inode)->nfs_client,
		NFS_SP4_MACH_CRED_CLEANUP, &task_setup_data.rpc_client, &msg);

5574 5575 5576 5577 5578
	/* Ensure this is an unlock - when canceling a lock, the
	 * canceled lock is passed in, and it won't be an unlock.
	 */
	fl->fl_type = F_UNLCK;

5579 5580 5581 5582 5583 5584
	data = nfs4_alloc_unlockdata(fl, ctx, lsp, seqid);
	if (data == NULL) {
		nfs_free_seqid(seqid);
		return ERR_PTR(-ENOMEM);
	}

5585
	nfs4_init_sequence(&data->arg.seq_args, &data->res.seq_res, 1);
5586 5587
	msg.rpc_argp = &data->arg;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
5588 5589
	task_setup_data.callback_data = data;
	return rpc_run_task(&task_setup_data);
5590 5591
}

5592 5593
static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
{
5594 5595 5596
	struct inode *inode = state->inode;
	struct nfs4_state_owner *sp = state->owner;
	struct nfs_inode *nfsi = NFS_I(inode);
T
Trond Myklebust 已提交
5597
	struct nfs_seqid *seqid;
L
Linus Torvalds 已提交
5598
	struct nfs4_lock_state *lsp;
5599
	struct rpc_task *task;
5600
	struct nfs_seqid *(*alloc_seqid)(struct nfs_seqid_counter *, gfp_t);
5601
	int status = 0;
5602
	unsigned char fl_flags = request->fl_flags;
5603

5604
	status = nfs4_set_lock_state(state, request);
5605 5606
	/* Unlock _before_ we do the RPC call */
	request->fl_flags |= FL_EXISTS;
5607 5608 5609
	/* Exclude nfs_delegation_claim_locks() */
	mutex_lock(&sp->so_delegreturn_mutex);
	/* Exclude nfs4_reclaim_open_stateid() - note nesting! */
5610 5611 5612
	down_read(&nfsi->rwsem);
	if (do_vfs_lock(request->fl_file, request) == -ENOENT) {
		up_read(&nfsi->rwsem);
5613
		mutex_unlock(&sp->so_delegreturn_mutex);
5614
		goto out;
5615 5616
	}
	up_read(&nfsi->rwsem);
5617
	mutex_unlock(&sp->so_delegreturn_mutex);
5618
	if (status != 0)
5619 5620
		goto out;
	/* Is this a delegated lock? */
5621
	lsp = request->fl_u.nfs4_fl.owner;
5622 5623
	if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) == 0)
		goto out;
5624 5625
	alloc_seqid = NFS_SERVER(inode)->nfs_client->cl_mvops->alloc_seqid;
	seqid = alloc_seqid(&lsp->ls_seqid, GFP_KERNEL);
5626
	status = -ENOMEM;
5627
	if (IS_ERR(seqid))
5628
		goto out;
5629
	task = nfs4_do_unlck(request, nfs_file_open_context(request->fl_file), lsp, seqid);
5630 5631
	status = PTR_ERR(task);
	if (IS_ERR(task))
5632
		goto out;
5633
	status = nfs4_wait_for_completion_rpc_task(task);
5634
	rpc_put_task(task);
5635
out:
5636
	request->fl_flags = fl_flags;
5637
	trace_nfs4_unlock(request, state, F_SETLK, status);
L
Linus Torvalds 已提交
5638 5639 5640
	return status;
}

5641 5642 5643 5644 5645 5646
struct nfs4_lockdata {
	struct nfs_lock_args arg;
	struct nfs_lock_res res;
	struct nfs4_lock_state *lsp;
	struct nfs_open_context *ctx;
	struct file_lock fl;
5647
	unsigned long timestamp;
5648 5649
	int rpc_status;
	int cancelled;
5650
	struct nfs_server *server;
5651 5652 5653
};

static struct nfs4_lockdata *nfs4_alloc_lockdata(struct file_lock *fl,
5654 5655
		struct nfs_open_context *ctx, struct nfs4_lock_state *lsp,
		gfp_t gfp_mask)
L
Linus Torvalds 已提交
5656
{
5657 5658
	struct nfs4_lockdata *p;
	struct inode *inode = lsp->ls_state->inode;
L
Linus Torvalds 已提交
5659
	struct nfs_server *server = NFS_SERVER(inode);
5660
	struct nfs_seqid *(*alloc_seqid)(struct nfs_seqid_counter *, gfp_t);
5661

5662
	p = kzalloc(sizeof(*p), gfp_mask);
5663 5664 5665 5666 5667
	if (p == NULL)
		return NULL;

	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
5668
	p->arg.open_seqid = nfs_alloc_seqid(&lsp->ls_state->owner->so_seqid, gfp_mask);
5669
	if (IS_ERR(p->arg.open_seqid))
5670
		goto out_free;
5671 5672
	alloc_seqid = server->nfs_client->cl_mvops->alloc_seqid;
	p->arg.lock_seqid = alloc_seqid(&lsp->ls_seqid, gfp_mask);
5673
	if (IS_ERR(p->arg.lock_seqid))
5674
		goto out_free_seqid;
5675
	p->arg.lock_owner.clientid = server->nfs_client->cl_clientid;
5676
	p->arg.lock_owner.id = lsp->ls_seqid.owner_id;
5677
	p->arg.lock_owner.s_dev = server->s_dev;
5678
	p->res.lock_seqid = p->arg.lock_seqid;
5679
	p->lsp = lsp;
5680
	p->server = server;
5681 5682
	atomic_inc(&lsp->ls_count);
	p->ctx = get_nfs_open_context(ctx);
5683
	get_file(fl->fl_file);
5684 5685
	memcpy(&p->fl, fl, sizeof(p->fl));
	return p;
5686 5687
out_free_seqid:
	nfs_free_seqid(p->arg.open_seqid);
5688 5689 5690 5691 5692 5693 5694 5695 5696
out_free:
	kfree(p);
	return NULL;
}

static void nfs4_lock_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_lockdata *data = calldata;
	struct nfs4_state *state = data->lsp->ls_state;
5697

5698
	dprintk("%s: begin!\n", __func__);
5699
	if (nfs_wait_on_sequence(data->arg.lock_seqid, task) != 0)
5700
		goto out_wait;
5701
	/* Do we need to do an open_to_lock_owner? */
5702
	if (!test_bit(NFS_LOCK_INITIALIZED, &data->lsp->ls_flags)) {
5703
		if (nfs_wait_on_sequence(data->arg.open_seqid, task) != 0) {
5704
			goto out_release_lock_seqid;
5705
		}
5706 5707
		nfs4_stateid_copy(&data->arg.open_stateid,
				&state->open_stateid);
5708
		data->arg.new_lock_owner = 1;
5709
		data->res.open_seqid = data->arg.open_seqid;
5710
	} else {
5711
		data->arg.new_lock_owner = 0;
5712 5713 5714
		nfs4_stateid_copy(&data->arg.lock_stateid,
				&data->lsp->ls_stateid);
	}
5715 5716 5717 5718 5719
	if (!nfs4_valid_open_stateid(state)) {
		data->rpc_status = -EBADF;
		task->tk_action = NULL;
		goto out_release_open_seqid;
	}
5720
	data->timestamp = jiffies;
5721 5722
	if (nfs4_setup_sequence(data->server,
				&data->arg.seq_args,
5723
				&data->res.seq_res,
5724
				task) == 0)
5725
		return;
5726
out_release_open_seqid:
5727 5728 5729
	nfs_release_seqid(data->arg.open_seqid);
out_release_lock_seqid:
	nfs_release_seqid(data->arg.lock_seqid);
5730 5731
out_wait:
	nfs4_sequence_done(task, &data->res.seq_res);
5732
	dprintk("%s: done!, ret = %d\n", __func__, data->rpc_status);
5733 5734
}

5735 5736 5737
static void nfs4_lock_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_lockdata *data = calldata;
5738
	struct nfs4_lock_state *lsp = data->lsp;
5739

5740
	dprintk("%s: begin!\n", __func__);
5741

5742 5743
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return;
5744

5745
	data->rpc_status = task->tk_status;
5746 5747
	switch (task->tk_status) {
	case 0:
5748
		renew_lease(NFS_SERVER(d_inode(data->ctx->dentry)),
5749
				data->timestamp);
5750 5751 5752 5753 5754 5755 5756
		if (data->arg.new_lock) {
			data->fl.fl_flags &= ~(FL_SLEEP | FL_ACCESS);
			if (do_vfs_lock(data->fl.fl_file, &data->fl) < 0) {
				rpc_restart_call_prepare(task);
				break;
			}
		}
5757 5758 5759 5760 5761 5762
		if (data->arg.new_lock_owner != 0) {
			nfs_confirm_seqid(&lsp->ls_seqid, 0);
			nfs4_stateid_copy(&lsp->ls_stateid, &data->res.stateid);
			set_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags);
		} else if (!nfs4_update_lock_stateid(lsp, &data->res.stateid))
			rpc_restart_call_prepare(task);
5763 5764 5765 5766 5767 5768 5769 5770 5771 5772 5773 5774
		break;
	case -NFS4ERR_BAD_STATEID:
	case -NFS4ERR_OLD_STATEID:
	case -NFS4ERR_STALE_STATEID:
	case -NFS4ERR_EXPIRED:
		if (data->arg.new_lock_owner != 0) {
			if (!nfs4_stateid_match(&data->arg.open_stateid,
						&lsp->ls_state->open_stateid))
				rpc_restart_call_prepare(task);
		} else if (!nfs4_stateid_match(&data->arg.lock_stateid,
						&lsp->ls_stateid))
				rpc_restart_call_prepare(task);
5775
	}
5776
	dprintk("%s: done, ret = %d!\n", __func__, data->rpc_status);
5777 5778 5779 5780 5781 5782
}

static void nfs4_lock_release(void *calldata)
{
	struct nfs4_lockdata *data = calldata;

5783
	dprintk("%s: begin!\n", __func__);
5784
	nfs_free_seqid(data->arg.open_seqid);
5785 5786 5787 5788 5789
	if (data->cancelled != 0) {
		struct rpc_task *task;
		task = nfs4_do_unlck(&data->fl, data->ctx, data->lsp,
				data->arg.lock_seqid);
		if (!IS_ERR(task))
5790
			rpc_put_task_async(task);
5791
		dprintk("%s: cancelling lock!\n", __func__);
5792 5793 5794 5795
	} else
		nfs_free_seqid(data->arg.lock_seqid);
	nfs4_put_lock_state(data->lsp);
	put_nfs_open_context(data->ctx);
5796
	fput(data->fl.fl_file);
5797
	kfree(data);
5798
	dprintk("%s: done!\n", __func__);
5799 5800 5801 5802 5803 5804 5805 5806
}

static const struct rpc_call_ops nfs4_lock_ops = {
	.rpc_call_prepare = nfs4_lock_prepare,
	.rpc_call_done = nfs4_lock_done,
	.rpc_release = nfs4_lock_release,
};

5807 5808 5809 5810 5811
static void nfs4_handle_setlk_error(struct nfs_server *server, struct nfs4_lock_state *lsp, int new_lock_owner, int error)
{
	switch (error) {
	case -NFS4ERR_ADMIN_REVOKED:
	case -NFS4ERR_BAD_STATEID:
5812
		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
5813
		if (new_lock_owner != 0 ||
5814
		   test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) != 0)
5815
			nfs4_schedule_stateid_recovery(server, lsp->ls_state);
5816 5817 5818
		break;
	case -NFS4ERR_STALE_STATEID:
		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
5819 5820
	case -NFS4ERR_EXPIRED:
		nfs4_schedule_lease_recovery(server->nfs_client);
5821 5822 5823
	};
}

5824
static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *fl, int recovery_type)
5825 5826 5827
{
	struct nfs4_lockdata *data;
	struct rpc_task *task;
5828 5829 5830 5831
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCK],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
5832 5833
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(state->inode),
5834
		.rpc_message = &msg,
T
Trond Myklebust 已提交
5835
		.callback_ops = &nfs4_lock_ops,
5836
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
5837 5838
		.flags = RPC_TASK_ASYNC,
	};
5839 5840
	int ret;

5841
	dprintk("%s: begin!\n", __func__);
5842
	data = nfs4_alloc_lockdata(fl, nfs_file_open_context(fl->fl_file),
5843 5844
			fl->fl_u.nfs4_fl.owner,
			recovery_type == NFS_LOCK_NEW ? GFP_KERNEL : GFP_NOFS);
5845 5846 5847 5848
	if (data == NULL)
		return -ENOMEM;
	if (IS_SETLKW(cmd))
		data->arg.block = 1;
5849
	nfs4_init_sequence(&data->arg.seq_args, &data->res.seq_res, 1);
5850 5851
	msg.rpc_argp = &data->arg;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
5852
	task_setup_data.callback_data = data;
5853 5854 5855 5856
	if (recovery_type > NFS_LOCK_NEW) {
		if (recovery_type == NFS_LOCK_RECLAIM)
			data->arg.reclaim = NFS_LOCK_RECLAIM;
		nfs4_set_sequence_privileged(&data->arg.seq_args);
5857 5858
	} else
		data->arg.new_lock = 1;
T
Trond Myklebust 已提交
5859
	task = rpc_run_task(&task_setup_data);
5860
	if (IS_ERR(task))
5861 5862 5863 5864
		return PTR_ERR(task);
	ret = nfs4_wait_for_completion_rpc_task(task);
	if (ret == 0) {
		ret = data->rpc_status;
5865 5866 5867
		if (ret)
			nfs4_handle_setlk_error(data->server, data->lsp,
					data->arg.new_lock_owner, ret);
5868 5869
	} else
		data->cancelled = 1;
5870
	rpc_put_task(task);
5871
	dprintk("%s: done, ret = %d!\n", __func__, ret);
5872
	return ret;
L
Linus Torvalds 已提交
5873 5874 5875 5876
}

static int nfs4_lock_reclaim(struct nfs4_state *state, struct file_lock *request)
{
5877
	struct nfs_server *server = NFS_SERVER(state->inode);
5878 5879 5880
	struct nfs4_exception exception = {
		.inode = state->inode,
	};
5881 5882 5883
	int err;

	do {
5884 5885 5886
		/* Cache the lock if possible... */
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
5887
		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_RECLAIM);
5888
		trace_nfs4_lock_reclaim(request, state, F_SETLK, err);
5889
		if (err != -NFS4ERR_DELAY)
5890 5891 5892 5893
			break;
		nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
	return err;
L
Linus Torvalds 已提交
5894 5895 5896 5897
}

static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request)
{
5898
	struct nfs_server *server = NFS_SERVER(state->inode);
5899 5900 5901
	struct nfs4_exception exception = {
		.inode = state->inode,
	};
5902 5903
	int err;

5904 5905 5906
	err = nfs4_set_lock_state(state, request);
	if (err != 0)
		return err;
5907
	if (!recover_lost_locks) {
5908 5909 5910
		set_bit(NFS_LOCK_LOST, &request->fl_u.nfs4_fl.owner->ls_flags);
		return 0;
	}
5911
	do {
5912 5913
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
5914
		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_EXPIRED);
5915
		trace_nfs4_lock_expired(request, state, F_SETLK, err);
5916 5917 5918 5919 5920 5921 5922 5923
		switch (err) {
		default:
			goto out;
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			nfs4_handle_exception(server, err, &exception);
			err = 0;
		}
5924
	} while (exception.retry);
5925
out:
5926
	return err;
L
Linus Torvalds 已提交
5927 5928
}

5929
#if defined(CONFIG_NFS_V4_1)
5930 5931 5932 5933 5934 5935 5936 5937
/**
 * nfs41_check_expired_locks - possibly free a lock stateid
 *
 * @state: NFSv4 state for an inode
 *
 * Returns NFS_OK if recovery for this stateid is now finished.
 * Otherwise a negative NFS4ERR value is returned.
 */
5938
static int nfs41_check_expired_locks(struct nfs4_state *state)
5939
{
5940
	int status, ret = -NFS4ERR_BAD_STATEID;
5941
	struct nfs4_lock_state *lsp;
5942 5943
	struct nfs_server *server = NFS_SERVER(state->inode);

5944
	list_for_each_entry(lsp, &state->lock_states, ls_locks) {
5945
		if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags)) {
5946 5947 5948 5949 5950
			struct rpc_cred *cred = lsp->ls_state->owner->so_cred;

			status = nfs41_test_stateid(server,
					&lsp->ls_stateid,
					cred);
5951
			trace_nfs4_test_lock_stateid(state, lsp, status);
5952
			if (status != NFS_OK) {
5953 5954
				/* Free the stateid unless the server
				 * informs us the stateid is unrecognized. */
5955 5956
				if (status != -NFS4ERR_BAD_STATEID)
					nfs41_free_stateid(server,
5957 5958
							&lsp->ls_stateid,
							cred);
5959
				clear_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags);
5960 5961 5962 5963 5964 5965 5966 5967 5968 5969 5970 5971 5972 5973
				ret = status;
			}
		}
	};

	return ret;
}

static int nfs41_lock_expired(struct nfs4_state *state, struct file_lock *request)
{
	int status = NFS_OK;

	if (test_bit(LK_STATE_IN_USE, &state->flags))
		status = nfs41_check_expired_locks(state);
5974 5975 5976
	if (status != NFS_OK)
		status = nfs4_lock_expired(state, request);
	return status;
5977 5978 5979
}
#endif

L
Linus Torvalds 已提交
5980 5981
static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
5982
	struct nfs_inode *nfsi = NFS_I(state->inode);
5983
	unsigned char fl_flags = request->fl_flags;
5984
	int status = -ENOLCK;
L
Linus Torvalds 已提交
5985

5986 5987 5988
	if ((fl_flags & FL_POSIX) &&
			!test_bit(NFS_STATE_POSIX_LOCKS, &state->flags))
		goto out;
5989 5990 5991 5992
	/* Is this a delegated open? */
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
5993 5994 5995 5996
	request->fl_flags |= FL_ACCESS;
	status = do_vfs_lock(request->fl_file, request);
	if (status < 0)
		goto out;
5997
	down_read(&nfsi->rwsem);
5998 5999 6000
	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
		/* Yes: cache locks! */
		/* ...but avoid races with delegation recall... */
6001 6002
		request->fl_flags = fl_flags & ~FL_SLEEP;
		status = do_vfs_lock(request->fl_file, request);
6003
		up_read(&nfsi->rwsem);
6004 6005
		goto out;
	}
6006
	up_read(&nfsi->rwsem);
6007
	status = _nfs4_do_setlk(state, cmd, request, NFS_LOCK_NEW);
6008 6009
out:
	request->fl_flags = fl_flags;
L
Linus Torvalds 已提交
6010 6011 6012 6013 6014
	return status;
}

static int nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
6015 6016
	struct nfs4_exception exception = {
		.state = state,
6017
		.inode = state->inode,
6018
	};
L
Linus Torvalds 已提交
6019 6020 6021
	int err;

	do {
6022
		err = _nfs4_proc_setlk(state, cmd, request);
6023
		trace_nfs4_set_lock(request, state, cmd, err);
6024 6025
		if (err == -NFS4ERR_DENIED)
			err = -EAGAIN;
L
Linus Torvalds 已提交
6026
		err = nfs4_handle_exception(NFS_SERVER(state->inode),
6027
				err, &exception);
L
Linus Torvalds 已提交
6028 6029 6030 6031 6032 6033 6034 6035 6036 6037 6038 6039 6040
	} while (exception.retry);
	return err;
}

static int
nfs4_proc_lock(struct file *filp, int cmd, struct file_lock *request)
{
	struct nfs_open_context *ctx;
	struct nfs4_state *state;
	unsigned long timeout = NFS4_LOCK_MINTIMEOUT;
	int status;

	/* verify open state */
6041
	ctx = nfs_file_open_context(filp);
L
Linus Torvalds 已提交
6042 6043 6044 6045 6046
	state = ctx->state;

	if (request->fl_start < 0 || request->fl_end < 0)
		return -EINVAL;

6047 6048 6049 6050 6051
	if (IS_GETLK(cmd)) {
		if (state != NULL)
			return nfs4_proc_getlk(state, F_GETLK, request);
		return 0;
	}
L
Linus Torvalds 已提交
6052 6053 6054 6055

	if (!(IS_SETLK(cmd) || IS_SETLKW(cmd)))
		return -EINVAL;

6056 6057 6058 6059 6060
	if (request->fl_type == F_UNLCK) {
		if (state != NULL)
			return nfs4_proc_unlck(state, cmd, request);
		return 0;
	}
L
Linus Torvalds 已提交
6061

6062 6063
	if (state == NULL)
		return -ENOLCK;
6064 6065 6066 6067
	/*
	 * Don't rely on the VFS having checked the file open mode,
	 * since it won't do this for flock() locks.
	 */
6068
	switch (request->fl_type) {
6069 6070 6071 6072 6073 6074 6075 6076 6077
	case F_RDLCK:
		if (!(filp->f_mode & FMODE_READ))
			return -EBADF;
		break;
	case F_WRLCK:
		if (!(filp->f_mode & FMODE_WRITE))
			return -EBADF;
	}

L
Linus Torvalds 已提交
6078 6079 6080 6081 6082 6083 6084 6085 6086 6087 6088 6089
	do {
		status = nfs4_proc_setlk(state, cmd, request);
		if ((status != -EAGAIN) || IS_SETLK(cmd))
			break;
		timeout = nfs4_set_lock_task_retry(timeout);
		status = -ERESTARTSYS;
		if (signalled())
			break;
	} while(status < 0);
	return status;
}

6090
int nfs4_lock_delegation_recall(struct file_lock *fl, struct nfs4_state *state, const nfs4_stateid *stateid)
6091 6092 6093 6094 6095 6096
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	int err;

	err = nfs4_set_lock_state(state, fl);
	if (err != 0)
6097
		return err;
6098
	err = _nfs4_do_setlk(state, F_SETLK, fl, NFS_LOCK_NEW);
6099
	return nfs4_handle_delegation_recall_error(server, state, stateid, err);
6100
}
6101

6102 6103
struct nfs_release_lockowner_data {
	struct nfs4_lock_state *lsp;
6104
	struct nfs_server *server;
6105
	struct nfs_release_lockowner_args args;
6106
	struct nfs_release_lockowner_res res;
6107
	unsigned long timestamp;
6108 6109
};

6110 6111 6112
static void nfs4_release_lockowner_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs_release_lockowner_data *data = calldata;
6113
	struct nfs_server *server = data->server;
6114 6115
	nfs40_setup_sequence(server->nfs_client->cl_slot_tbl,
			     &data->args.seq_args, &data->res.seq_res, task);
6116
	data->args.lock_owner.clientid = server->nfs_client->cl_clientid;
6117
	data->timestamp = jiffies;
6118 6119 6120 6121 6122
}

static void nfs4_release_lockowner_done(struct rpc_task *task, void *calldata)
{
	struct nfs_release_lockowner_data *data = calldata;
6123 6124
	struct nfs_server *server = data->server;

6125
	nfs40_sequence_done(task, &data->res.seq_res);
6126 6127 6128 6129 6130 6131 6132

	switch (task->tk_status) {
	case 0:
		renew_lease(server, data->timestamp);
		break;
	case -NFS4ERR_STALE_CLIENTID:
	case -NFS4ERR_EXPIRED:
6133 6134
		nfs4_schedule_lease_recovery(server->nfs_client);
		break;
6135 6136
	case -NFS4ERR_LEASE_MOVED:
	case -NFS4ERR_DELAY:
6137 6138
		if (nfs4_async_handle_error(task, server,
					    NULL, NULL) == -EAGAIN)
6139 6140
			rpc_restart_call_prepare(task);
	}
6141 6142
}

6143 6144
static void nfs4_release_lockowner_release(void *calldata)
{
6145
	struct nfs_release_lockowner_data *data = calldata;
6146
	nfs4_free_lock_state(data->server, data->lsp);
6147 6148 6149
	kfree(calldata);
}

6150
static const struct rpc_call_ops nfs4_release_lockowner_ops = {
6151 6152
	.rpc_call_prepare = nfs4_release_lockowner_prepare,
	.rpc_call_done = nfs4_release_lockowner_done,
6153 6154 6155
	.rpc_release = nfs4_release_lockowner_release,
};

6156 6157
static void
nfs4_release_lockowner(struct nfs_server *server, struct nfs4_lock_state *lsp)
6158
{
6159
	struct nfs_release_lockowner_data *data;
6160 6161 6162 6163 6164
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RELEASE_LOCKOWNER],
	};

	if (server->nfs_client->cl_mvops->minor_version != 0)
6165
		return;
6166

6167 6168
	data = kmalloc(sizeof(*data), GFP_NOFS);
	if (!data)
6169
		return;
6170
	data->lsp = lsp;
6171
	data->server = server;
6172 6173 6174
	data->args.lock_owner.clientid = server->nfs_client->cl_clientid;
	data->args.lock_owner.id = lsp->ls_seqid.owner_id;
	data->args.lock_owner.s_dev = server->s_dev;
6175

6176
	msg.rpc_argp = &data->args;
6177 6178
	msg.rpc_resp = &data->res;
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 0);
6179
	rpc_call_async(server->client, &msg, 0, &nfs4_release_lockowner_ops, data);
6180 6181
}

6182 6183
#define XATTR_NAME_NFSV4_ACL "system.nfs4_acl"

6184 6185 6186
static int nfs4_xattr_set_nfs4_acl(struct dentry *dentry, const char *key,
				   const void *buf, size_t buflen,
				   int flags, int type)
6187
{
6188 6189
	if (strcmp(key, "") != 0)
		return -EINVAL;
6190

6191
	return nfs4_proc_set_acl(d_inode(dentry), buf, buflen);
6192 6193
}

6194 6195
static int nfs4_xattr_get_nfs4_acl(struct dentry *dentry, const char *key,
				   void *buf, size_t buflen, int type)
6196
{
6197 6198
	if (strcmp(key, "") != 0)
		return -EINVAL;
6199

6200
	return nfs4_proc_get_acl(d_inode(dentry), buf, buflen);
6201 6202
}

6203 6204 6205
static size_t nfs4_xattr_list_nfs4_acl(struct dentry *dentry, char *list,
				       size_t list_len, const char *name,
				       size_t name_len, int type)
6206
{
6207
	size_t len = sizeof(XATTR_NAME_NFSV4_ACL);
6208

6209
	if (!nfs4_server_supports_acls(NFS_SERVER(d_inode(dentry))))
6210
		return 0;
6211 6212 6213

	if (list && len <= list_len)
		memcpy(list, XATTR_NAME_NFSV4_ACL, len);
6214
	return len;
6215 6216
}

6217 6218 6219 6220 6221 6222 6223 6224 6225 6226 6227 6228 6229 6230 6231 6232 6233 6234 6235 6236
#ifdef CONFIG_NFS_V4_SECURITY_LABEL
static inline int nfs4_server_supports_labels(struct nfs_server *server)
{
	return server->caps & NFS_CAP_SECURITY_LABEL;
}

static int nfs4_xattr_set_nfs4_label(struct dentry *dentry, const char *key,
				   const void *buf, size_t buflen,
				   int flags, int type)
{
	if (security_ismaclabel(key))
		return nfs4_set_security_label(dentry, buf, buflen);

	return -EOPNOTSUPP;
}

static int nfs4_xattr_get_nfs4_label(struct dentry *dentry, const char *key,
				   void *buf, size_t buflen, int type)
{
	if (security_ismaclabel(key))
6237
		return nfs4_get_security_label(d_inode(dentry), buf, buflen);
6238 6239 6240 6241 6242 6243 6244 6245 6246
	return -EOPNOTSUPP;
}

static size_t nfs4_xattr_list_nfs4_label(struct dentry *dentry, char *list,
				       size_t list_len, const char *name,
				       size_t name_len, int type)
{
	size_t len = 0;

6247 6248
	if (nfs_server_capable(d_inode(dentry), NFS_CAP_SECURITY_LABEL)) {
		len = security_inode_listsecurity(d_inode(dentry), NULL, 0);
6249
		if (list && len <= list_len)
6250
			security_inode_listsecurity(d_inode(dentry), list, len);
6251 6252 6253 6254 6255 6256 6257 6258 6259 6260 6261 6262 6263
	}
	return len;
}

static const struct xattr_handler nfs4_xattr_nfs4_label_handler = {
	.prefix = XATTR_SECURITY_PREFIX,
	.list	= nfs4_xattr_list_nfs4_label,
	.get	= nfs4_xattr_get_nfs4_label,
	.set	= nfs4_xattr_set_nfs4_label,
};
#endif


6264 6265 6266
/*
 * nfs_fhget will use either the mounted_on_fileid or the fileid
 */
6267 6268
static void nfs_fixup_referral_attributes(struct nfs_fattr *fattr)
{
6269 6270 6271
	if (!(((fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID) ||
	       (fattr->valid & NFS_ATTR_FATTR_FILEID)) &&
	      (fattr->valid & NFS_ATTR_FATTR_FSID) &&
6272
	      (fattr->valid & NFS_ATTR_FATTR_V4_LOCATIONS)))
6273 6274 6275
		return;

	fattr->valid |= NFS_ATTR_FATTR_TYPE | NFS_ATTR_FATTR_MODE |
6276
		NFS_ATTR_FATTR_NLINK | NFS_ATTR_FATTR_V4_REFERRAL;
6277 6278 6279 6280
	fattr->mode = S_IFDIR | S_IRUGO | S_IXUGO;
	fattr->nlink = 2;
}

6281 6282 6283 6284
static int _nfs4_proc_fs_locations(struct rpc_clnt *client, struct inode *dir,
				   const struct qstr *name,
				   struct nfs4_fs_locations *fs_locations,
				   struct page *page)
6285 6286
{
	struct nfs_server *server = NFS_SERVER(dir);
6287
	u32 bitmask[3] = {
6288
		[0] = FATTR4_WORD0_FSID | FATTR4_WORD0_FS_LOCATIONS,
6289 6290 6291
	};
	struct nfs4_fs_locations_arg args = {
		.dir_fh = NFS_FH(dir),
6292
		.name = name,
6293 6294 6295
		.page = page,
		.bitmask = bitmask,
	};
B
Benny Halevy 已提交
6296 6297 6298
	struct nfs4_fs_locations_res res = {
		.fs_locations = fs_locations,
	};
6299 6300 6301
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FS_LOCATIONS],
		.rpc_argp = &args,
B
Benny Halevy 已提交
6302
		.rpc_resp = &res,
6303 6304 6305
	};
	int status;

6306
	dprintk("%s: start\n", __func__);
6307 6308 6309 6310 6311 6312 6313 6314

	/* Ask for the fileid of the absent filesystem if mounted_on_fileid
	 * is not supported */
	if (NFS_SERVER(dir)->attr_bitmask[1] & FATTR4_WORD1_MOUNTED_ON_FILEID)
		bitmask[1] |= FATTR4_WORD1_MOUNTED_ON_FILEID;
	else
		bitmask[0] |= FATTR4_WORD0_FILEID;

6315
	nfs_fattr_init(&fs_locations->fattr);
6316
	fs_locations->server = server;
6317
	fs_locations->nlocations = 0;
6318
	status = nfs4_call_sync(client, server, &msg, &args.seq_args, &res.seq_res, 0);
6319
	dprintk("%s: returned status = %d\n", __func__, status);
6320 6321 6322
	return status;
}

6323 6324 6325 6326
int nfs4_proc_fs_locations(struct rpc_clnt *client, struct inode *dir,
			   const struct qstr *name,
			   struct nfs4_fs_locations *fs_locations,
			   struct page *page)
6327 6328 6329 6330
{
	struct nfs4_exception exception = { };
	int err;
	do {
6331 6332 6333 6334
		err = _nfs4_proc_fs_locations(client, dir, name,
				fs_locations, page);
		trace_nfs4_get_fs_locations(dir, name, err);
		err = nfs4_handle_exception(NFS_SERVER(dir), err,
6335 6336 6337 6338 6339
				&exception);
	} while (exception.retry);
	return err;
}

6340 6341 6342 6343 6344 6345 6346 6347 6348 6349 6350 6351 6352 6353 6354 6355 6356 6357 6358 6359 6360 6361 6362 6363 6364 6365 6366 6367 6368 6369 6370 6371 6372 6373 6374 6375 6376 6377 6378 6379 6380 6381 6382 6383 6384 6385 6386 6387 6388 6389 6390 6391 6392 6393 6394 6395 6396 6397 6398 6399 6400 6401 6402 6403 6404 6405 6406 6407 6408 6409 6410 6411 6412 6413 6414 6415 6416 6417 6418 6419 6420 6421 6422 6423 6424 6425 6426 6427 6428 6429 6430 6431 6432 6433 6434 6435 6436 6437 6438 6439 6440 6441 6442 6443 6444 6445 6446 6447 6448 6449 6450 6451 6452 6453 6454 6455 6456 6457 6458 6459 6460 6461 6462 6463 6464 6465 6466 6467 6468 6469 6470 6471 6472 6473 6474 6475 6476 6477 6478 6479 6480 6481 6482 6483 6484 6485 6486 6487 6488 6489 6490
/*
 * This operation also signals the server that this client is
 * performing migration recovery.  The server can stop returning
 * NFS4ERR_LEASE_MOVED to this client.  A RENEW operation is
 * appended to this compound to identify the client ID which is
 * performing recovery.
 */
static int _nfs40_proc_get_locations(struct inode *inode,
				     struct nfs4_fs_locations *locations,
				     struct page *page, struct rpc_cred *cred)
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct rpc_clnt *clnt = server->client;
	u32 bitmask[2] = {
		[0] = FATTR4_WORD0_FSID | FATTR4_WORD0_FS_LOCATIONS,
	};
	struct nfs4_fs_locations_arg args = {
		.clientid	= server->nfs_client->cl_clientid,
		.fh		= NFS_FH(inode),
		.page		= page,
		.bitmask	= bitmask,
		.migration	= 1,		/* skip LOOKUP */
		.renew		= 1,		/* append RENEW */
	};
	struct nfs4_fs_locations_res res = {
		.fs_locations	= locations,
		.migration	= 1,
		.renew		= 1,
	};
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_FS_LOCATIONS],
		.rpc_argp	= &args,
		.rpc_resp	= &res,
		.rpc_cred	= cred,
	};
	unsigned long now = jiffies;
	int status;

	nfs_fattr_init(&locations->fattr);
	locations->server = server;
	locations->nlocations = 0;

	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0);
	nfs4_set_sequence_privileged(&args.seq_args);
	status = nfs4_call_sync_sequence(clnt, server, &msg,
					&args.seq_args, &res.seq_res);
	if (status)
		return status;

	renew_lease(server, now);
	return 0;
}

#ifdef CONFIG_NFS_V4_1

/*
 * This operation also signals the server that this client is
 * performing migration recovery.  The server can stop asserting
 * SEQ4_STATUS_LEASE_MOVED for this client.  The client ID
 * performing this operation is identified in the SEQUENCE
 * operation in this compound.
 *
 * When the client supports GETATTR(fs_locations_info), it can
 * be plumbed in here.
 */
static int _nfs41_proc_get_locations(struct inode *inode,
				     struct nfs4_fs_locations *locations,
				     struct page *page, struct rpc_cred *cred)
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct rpc_clnt *clnt = server->client;
	u32 bitmask[2] = {
		[0] = FATTR4_WORD0_FSID | FATTR4_WORD0_FS_LOCATIONS,
	};
	struct nfs4_fs_locations_arg args = {
		.fh		= NFS_FH(inode),
		.page		= page,
		.bitmask	= bitmask,
		.migration	= 1,		/* skip LOOKUP */
	};
	struct nfs4_fs_locations_res res = {
		.fs_locations	= locations,
		.migration	= 1,
	};
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_FS_LOCATIONS],
		.rpc_argp	= &args,
		.rpc_resp	= &res,
		.rpc_cred	= cred,
	};
	int status;

	nfs_fattr_init(&locations->fattr);
	locations->server = server;
	locations->nlocations = 0;

	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0);
	nfs4_set_sequence_privileged(&args.seq_args);
	status = nfs4_call_sync_sequence(clnt, server, &msg,
					&args.seq_args, &res.seq_res);
	if (status == NFS4_OK &&
	    res.seq_res.sr_status_flags & SEQ4_STATUS_LEASE_MOVED)
		status = -NFS4ERR_LEASE_MOVED;
	return status;
}

#endif	/* CONFIG_NFS_V4_1 */

/**
 * nfs4_proc_get_locations - discover locations for a migrated FSID
 * @inode: inode on FSID that is migrating
 * @locations: result of query
 * @page: buffer
 * @cred: credential to use for this operation
 *
 * Returns NFS4_OK on success, a negative NFS4ERR status code if the
 * operation failed, or a negative errno if a local error occurred.
 *
 * On success, "locations" is filled in, but if the server has
 * no locations information, NFS_ATTR_FATTR_V4_LOCATIONS is not
 * asserted.
 *
 * -NFS4ERR_LEASE_MOVED is returned if the server still has leases
 * from this client that require migration recovery.
 */
int nfs4_proc_get_locations(struct inode *inode,
			    struct nfs4_fs_locations *locations,
			    struct page *page, struct rpc_cred *cred)
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs_client *clp = server->nfs_client;
	const struct nfs4_mig_recovery_ops *ops =
					clp->cl_mvops->mig_recovery_ops;
	struct nfs4_exception exception = { };
	int status;

	dprintk("%s: FSID %llx:%llx on \"%s\"\n", __func__,
		(unsigned long long)server->fsid.major,
		(unsigned long long)server->fsid.minor,
		clp->cl_hostname);
	nfs_display_fhandle(NFS_FH(inode), __func__);

	do {
		status = ops->get_locations(inode, locations, page, cred);
		if (status != -NFS4ERR_DELAY)
			break;
		nfs4_handle_exception(server, status, &exception);
	} while (exception.retry);
	return status;
}

6491 6492 6493 6494 6495 6496 6497 6498 6499 6500 6501 6502 6503 6504 6505 6506 6507 6508 6509 6510 6511 6512 6513 6514 6515 6516 6517 6518 6519 6520 6521 6522 6523 6524 6525 6526 6527 6528 6529 6530 6531 6532 6533 6534 6535 6536 6537 6538 6539 6540 6541 6542 6543 6544 6545 6546 6547 6548 6549 6550 6551 6552 6553 6554 6555 6556 6557 6558 6559 6560 6561 6562 6563 6564 6565 6566 6567 6568 6569 6570 6571 6572 6573 6574 6575 6576 6577 6578 6579 6580 6581 6582 6583 6584 6585 6586 6587 6588 6589 6590 6591 6592 6593 6594 6595 6596 6597 6598 6599 6600 6601 6602 6603 6604 6605 6606 6607 6608 6609 6610 6611 6612 6613 6614 6615 6616
/*
 * This operation also signals the server that this client is
 * performing "lease moved" recovery.  The server can stop
 * returning NFS4ERR_LEASE_MOVED to this client.  A RENEW operation
 * is appended to this compound to identify the client ID which is
 * performing recovery.
 */
static int _nfs40_proc_fsid_present(struct inode *inode, struct rpc_cred *cred)
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
	struct rpc_clnt *clnt = server->client;
	struct nfs4_fsid_present_arg args = {
		.fh		= NFS_FH(inode),
		.clientid	= clp->cl_clientid,
		.renew		= 1,		/* append RENEW */
	};
	struct nfs4_fsid_present_res res = {
		.renew		= 1,
	};
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_FSID_PRESENT],
		.rpc_argp	= &args,
		.rpc_resp	= &res,
		.rpc_cred	= cred,
	};
	unsigned long now = jiffies;
	int status;

	res.fh = nfs_alloc_fhandle();
	if (res.fh == NULL)
		return -ENOMEM;

	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0);
	nfs4_set_sequence_privileged(&args.seq_args);
	status = nfs4_call_sync_sequence(clnt, server, &msg,
						&args.seq_args, &res.seq_res);
	nfs_free_fhandle(res.fh);
	if (status)
		return status;

	do_renew_lease(clp, now);
	return 0;
}

#ifdef CONFIG_NFS_V4_1

/*
 * This operation also signals the server that this client is
 * performing "lease moved" recovery.  The server can stop asserting
 * SEQ4_STATUS_LEASE_MOVED for this client.  The client ID performing
 * this operation is identified in the SEQUENCE operation in this
 * compound.
 */
static int _nfs41_proc_fsid_present(struct inode *inode, struct rpc_cred *cred)
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct rpc_clnt *clnt = server->client;
	struct nfs4_fsid_present_arg args = {
		.fh		= NFS_FH(inode),
	};
	struct nfs4_fsid_present_res res = {
	};
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_FSID_PRESENT],
		.rpc_argp	= &args,
		.rpc_resp	= &res,
		.rpc_cred	= cred,
	};
	int status;

	res.fh = nfs_alloc_fhandle();
	if (res.fh == NULL)
		return -ENOMEM;

	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0);
	nfs4_set_sequence_privileged(&args.seq_args);
	status = nfs4_call_sync_sequence(clnt, server, &msg,
						&args.seq_args, &res.seq_res);
	nfs_free_fhandle(res.fh);
	if (status == NFS4_OK &&
	    res.seq_res.sr_status_flags & SEQ4_STATUS_LEASE_MOVED)
		status = -NFS4ERR_LEASE_MOVED;
	return status;
}

#endif	/* CONFIG_NFS_V4_1 */

/**
 * nfs4_proc_fsid_present - Is this FSID present or absent on server?
 * @inode: inode on FSID to check
 * @cred: credential to use for this operation
 *
 * Server indicates whether the FSID is present, moved, or not
 * recognized.  This operation is necessary to clear a LEASE_MOVED
 * condition for this client ID.
 *
 * Returns NFS4_OK if the FSID is present on this server,
 * -NFS4ERR_MOVED if the FSID is no longer present, a negative
 *  NFS4ERR code if some error occurred on the server, or a
 *  negative errno if a local failure occurred.
 */
int nfs4_proc_fsid_present(struct inode *inode, struct rpc_cred *cred)
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs_client *clp = server->nfs_client;
	const struct nfs4_mig_recovery_ops *ops =
					clp->cl_mvops->mig_recovery_ops;
	struct nfs4_exception exception = { };
	int status;

	dprintk("%s: FSID %llx:%llx on \"%s\"\n", __func__,
		(unsigned long long)server->fsid.major,
		(unsigned long long)server->fsid.minor,
		clp->cl_hostname);
	nfs_display_fhandle(NFS_FH(inode), __func__);

	do {
		status = ops->fsid_present(inode, cred);
		if (status != -NFS4ERR_DELAY)
			break;
		nfs4_handle_exception(server, status, &exception);
	} while (exception.retry);
	return status;
}

6617
/**
6618 6619 6620 6621 6622
 * If 'use_integrity' is true and the state managment nfs_client
 * cl_rpcclient is using krb5i/p, use the integrity protected cl_rpcclient
 * and the machine credential as per RFC3530bis and RFC5661 Security
 * Considerations sections. Otherwise, just use the user cred with the
 * filesystem's rpc_client.
6623
 */
6624
static int _nfs4_proc_secinfo(struct inode *dir, const struct qstr *name, struct nfs4_secinfo_flavors *flavors, bool use_integrity)
B
Bryan Schumaker 已提交
6625 6626 6627 6628 6629 6630 6631 6632 6633 6634 6635 6636 6637 6638
{
	int status;
	struct nfs4_secinfo_arg args = {
		.dir_fh = NFS_FH(dir),
		.name   = name,
	};
	struct nfs4_secinfo_res res = {
		.flavors     = flavors,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SECINFO],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
6639
	struct rpc_clnt *clnt = NFS_SERVER(dir)->client;
6640
	struct rpc_cred *cred = NULL;
6641 6642 6643

	if (use_integrity) {
		clnt = NFS_SERVER(dir)->nfs_client->cl_rpcclient;
6644 6645
		cred = nfs4_get_clid_cred(NFS_SERVER(dir)->nfs_client);
		msg.rpc_cred = cred;
6646
	}
B
Bryan Schumaker 已提交
6647 6648

	dprintk("NFS call  secinfo %s\n", name->name);
6649 6650 6651 6652

	nfs4_state_protect(NFS_SERVER(dir)->nfs_client,
		NFS_SP4_MACH_CRED_SECINFO, &clnt, &msg);

6653 6654
	status = nfs4_call_sync(clnt, NFS_SERVER(dir), &msg, &args.seq_args,
				&res.seq_res, 0);
B
Bryan Schumaker 已提交
6655
	dprintk("NFS reply  secinfo: %d\n", status);
6656

6657 6658
	if (cred)
		put_rpccred(cred);
6659

B
Bryan Schumaker 已提交
6660 6661 6662
	return status;
}

6663 6664
int nfs4_proc_secinfo(struct inode *dir, const struct qstr *name,
		      struct nfs4_secinfo_flavors *flavors)
B
Bryan Schumaker 已提交
6665 6666 6667 6668
{
	struct nfs4_exception exception = { };
	int err;
	do {
6669 6670 6671 6672 6673 6674 6675 6676 6677 6678 6679 6680 6681 6682 6683
		err = -NFS4ERR_WRONGSEC;

		/* try to use integrity protection with machine cred */
		if (_nfs4_is_integrity_protected(NFS_SERVER(dir)->nfs_client))
			err = _nfs4_proc_secinfo(dir, name, flavors, true);

		/*
		 * if unable to use integrity protection, or SECINFO with
		 * integrity protection returns NFS4ERR_WRONGSEC (which is
		 * disallowed by spec, but exists in deployed servers) use
		 * the current filesystem's rpc_client and the user cred.
		 */
		if (err == -NFS4ERR_WRONGSEC)
			err = _nfs4_proc_secinfo(dir, name, flavors, false);

6684 6685
		trace_nfs4_secinfo(dir, name, err);
		err = nfs4_handle_exception(NFS_SERVER(dir), err,
B
Bryan Schumaker 已提交
6686 6687 6688 6689 6690
				&exception);
	} while (exception.retry);
	return err;
}

6691
#ifdef CONFIG_NFS_V4_1
6692 6693 6694 6695 6696 6697 6698 6699 6700 6701 6702 6703 6704 6705 6706 6707 6708 6709 6710
/*
 * Check the exchange flags returned by the server for invalid flags, having
 * both PNFS and NON_PNFS flags set, and not having one of NON_PNFS, PNFS, or
 * DS flags set.
 */
static int nfs4_check_cl_exchange_flags(u32 flags)
{
	if (flags & ~EXCHGID4_FLAG_MASK_R)
		goto out_inval;
	if ((flags & EXCHGID4_FLAG_USE_PNFS_MDS) &&
	    (flags & EXCHGID4_FLAG_USE_NON_PNFS))
		goto out_inval;
	if (!(flags & (EXCHGID4_FLAG_MASK_PNFS)))
		goto out_inval;
	return NFS_OK;
out_inval:
	return -NFS4ERR_INVAL;
}

6711
static bool
6712 6713
nfs41_same_server_scope(struct nfs41_server_scope *a,
			struct nfs41_server_scope *b)
6714 6715 6716 6717 6718 6719 6720 6721
{
	if (a->server_scope_sz == b->server_scope_sz &&
	    memcmp(a->server_scope, b->server_scope, a->server_scope_sz) == 0)
		return true;

	return false;
}

6722 6723 6724 6725 6726 6727
/*
 * nfs4_proc_bind_conn_to_session()
 *
 * The 4.1 client currently uses the same TCP connection for the
 * fore and backchannel.
 */
6728
int nfs4_proc_bind_conn_to_session(struct nfs_client *clp, struct rpc_cred *cred)
6729 6730
{
	int status;
6731 6732 6733 6734
	struct nfs41_bind_conn_to_session_args args = {
		.client = clp,
		.dir = NFS4_CDFC4_FORE_OR_BOTH,
	};
6735 6736 6737 6738
	struct nfs41_bind_conn_to_session_res res;
	struct rpc_message msg = {
		.rpc_proc =
			&nfs4_procedures[NFSPROC4_CLNT_BIND_CONN_TO_SESSION],
6739
		.rpc_argp = &args,
6740
		.rpc_resp = &res,
6741
		.rpc_cred = cred,
6742 6743 6744 6745
	};

	dprintk("--> %s\n", __func__);

6746 6747 6748
	nfs4_copy_sessionid(&args.sessionid, &clp->cl_session->sess_id);
	if (!(clp->cl_session->flags & SESSION4_BACK_CHAN))
		args.dir = NFS4_CDFC4_FORE;
6749 6750

	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
6751
	trace_nfs4_bind_conn_to_session(clp, status);
6752
	if (status == 0) {
6753
		if (memcmp(res.sessionid.data,
6754 6755 6756
		    clp->cl_session->sess_id.data, NFS4_MAX_SESSIONID_LEN)) {
			dprintk("NFS: %s: Session ID mismatch\n", __func__);
			status = -EIO;
6757
			goto out;
6758
		}
6759
		if ((res.dir & args.dir) != res.dir || res.dir == 0) {
6760 6761 6762
			dprintk("NFS: %s: Unexpected direction from server\n",
				__func__);
			status = -EIO;
6763
			goto out;
6764
		}
6765
		if (res.use_conn_in_rdma_mode != args.use_conn_in_rdma_mode) {
6766 6767 6768
			dprintk("NFS: %s: Server returned RDMA mode = true\n",
				__func__);
			status = -EIO;
6769
			goto out;
6770 6771 6772 6773 6774 6775 6776
		}
	}
out:
	dprintk("<-- %s status= %d\n", __func__, status);
	return status;
}

B
Benny Halevy 已提交
6777
/*
6778 6779
 * Minimum set of SP4_MACH_CRED operations from RFC 5661 in the enforce map
 * and operations we'd like to see to enable certain features in the allow map
6780 6781 6782 6783 6784 6785 6786 6787 6788
 */
static const struct nfs41_state_protection nfs4_sp4_mach_cred_request = {
	.how = SP4_MACH_CRED,
	.enforce.u.words = {
		[1] = 1 << (OP_BIND_CONN_TO_SESSION - 32) |
		      1 << (OP_EXCHANGE_ID - 32) |
		      1 << (OP_CREATE_SESSION - 32) |
		      1 << (OP_DESTROY_SESSION - 32) |
		      1 << (OP_DESTROY_CLIENTID - 32)
6789 6790 6791
	},
	.allow.u.words = {
		[0] = 1 << (OP_CLOSE) |
6792 6793
		      1 << (OP_LOCKU) |
		      1 << (OP_COMMIT),
6794
		[1] = 1 << (OP_SECINFO - 32) |
6795 6796
		      1 << (OP_SECINFO_NO_NAME - 32) |
		      1 << (OP_TEST_STATEID - 32) |
6797 6798
		      1 << (OP_FREE_STATEID - 32) |
		      1 << (OP_WRITE - 32)
6799 6800 6801 6802 6803 6804
	}
};

/*
 * Select the state protection mode for client `clp' given the server results
 * from exchange_id in `sp'.
B
Benny Halevy 已提交
6805
 *
6806 6807 6808 6809 6810 6811 6812 6813 6814 6815 6816 6817 6818 6819 6820 6821 6822 6823 6824 6825 6826 6827 6828 6829 6830 6831 6832 6833 6834 6835 6836 6837 6838 6839 6840 6841 6842 6843 6844 6845 6846 6847 6848 6849 6850 6851 6852 6853 6854 6855 6856 6857
 * Returns 0 on success, negative errno otherwise.
 */
static int nfs4_sp4_select_mode(struct nfs_client *clp,
				 struct nfs41_state_protection *sp)
{
	static const u32 supported_enforce[NFS4_OP_MAP_NUM_WORDS] = {
		[1] = 1 << (OP_BIND_CONN_TO_SESSION - 32) |
		      1 << (OP_EXCHANGE_ID - 32) |
		      1 << (OP_CREATE_SESSION - 32) |
		      1 << (OP_DESTROY_SESSION - 32) |
		      1 << (OP_DESTROY_CLIENTID - 32)
	};
	unsigned int i;

	if (sp->how == SP4_MACH_CRED) {
		/* Print state protect result */
		dfprintk(MOUNT, "Server SP4_MACH_CRED support:\n");
		for (i = 0; i <= LAST_NFS4_OP; i++) {
			if (test_bit(i, sp->enforce.u.longs))
				dfprintk(MOUNT, "  enforce op %d\n", i);
			if (test_bit(i, sp->allow.u.longs))
				dfprintk(MOUNT, "  allow op %d\n", i);
		}

		/* make sure nothing is on enforce list that isn't supported */
		for (i = 0; i < NFS4_OP_MAP_NUM_WORDS; i++) {
			if (sp->enforce.u.words[i] & ~supported_enforce[i]) {
				dfprintk(MOUNT, "sp4_mach_cred: disabled\n");
				return -EINVAL;
			}
		}

		/*
		 * Minimal mode - state operations are allowed to use machine
		 * credential.  Note this already happens by default, so the
		 * client doesn't have to do anything more than the negotiation.
		 *
		 * NOTE: we don't care if EXCHANGE_ID is in the list -
		 *       we're already using the machine cred for exchange_id
		 *       and will never use a different cred.
		 */
		if (test_bit(OP_BIND_CONN_TO_SESSION, sp->enforce.u.longs) &&
		    test_bit(OP_CREATE_SESSION, sp->enforce.u.longs) &&
		    test_bit(OP_DESTROY_SESSION, sp->enforce.u.longs) &&
		    test_bit(OP_DESTROY_CLIENTID, sp->enforce.u.longs)) {
			dfprintk(MOUNT, "sp4_mach_cred:\n");
			dfprintk(MOUNT, "  minimal mode enabled\n");
			set_bit(NFS_SP4_MACH_CRED_MINIMAL, &clp->cl_sp4_flags);
		} else {
			dfprintk(MOUNT, "sp4_mach_cred: disabled\n");
			return -EINVAL;
		}
6858 6859 6860 6861 6862 6863

		if (test_bit(OP_CLOSE, sp->allow.u.longs) &&
		    test_bit(OP_LOCKU, sp->allow.u.longs)) {
			dfprintk(MOUNT, "  cleanup mode enabled\n");
			set_bit(NFS_SP4_MACH_CRED_CLEANUP, &clp->cl_sp4_flags);
		}
6864 6865 6866 6867 6868 6869

		if (test_bit(OP_SECINFO, sp->allow.u.longs) &&
		    test_bit(OP_SECINFO_NO_NAME, sp->allow.u.longs)) {
			dfprintk(MOUNT, "  secinfo mode enabled\n");
			set_bit(NFS_SP4_MACH_CRED_SECINFO, &clp->cl_sp4_flags);
		}
6870 6871 6872 6873 6874 6875

		if (test_bit(OP_TEST_STATEID, sp->allow.u.longs) &&
		    test_bit(OP_FREE_STATEID, sp->allow.u.longs)) {
			dfprintk(MOUNT, "  stateid mode enabled\n");
			set_bit(NFS_SP4_MACH_CRED_STATEID, &clp->cl_sp4_flags);
		}
6876 6877 6878 6879 6880 6881 6882 6883 6884 6885

		if (test_bit(OP_WRITE, sp->allow.u.longs)) {
			dfprintk(MOUNT, "  write mode enabled\n");
			set_bit(NFS_SP4_MACH_CRED_WRITE, &clp->cl_sp4_flags);
		}

		if (test_bit(OP_COMMIT, sp->allow.u.longs)) {
			dfprintk(MOUNT, "  commit mode enabled\n");
			set_bit(NFS_SP4_MACH_CRED_COMMIT, &clp->cl_sp4_flags);
		}
6886 6887 6888 6889 6890 6891 6892
	}

	return 0;
}

/*
 * _nfs4_proc_exchange_id()
6893
 *
6894
 * Wrapper for EXCHANGE_ID operation.
B
Benny Halevy 已提交
6895
 */
6896 6897
static int _nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred,
	u32 sp4_how)
B
Benny Halevy 已提交
6898 6899 6900
{
	nfs4_verifier verifier;
	struct nfs41_exchange_id_args args = {
6901
		.verifier = &verifier,
B
Benny Halevy 已提交
6902
		.client = clp,
6903
#ifdef CONFIG_NFS_V4_1_MIGRATION
6904
		.flags = EXCHGID4_FLAG_SUPP_MOVED_REFER |
6905 6906 6907 6908 6909 6910
			 EXCHGID4_FLAG_BIND_PRINC_STATEID |
			 EXCHGID4_FLAG_SUPP_MOVED_MIGR,
#else
		.flags = EXCHGID4_FLAG_SUPP_MOVED_REFER |
			 EXCHGID4_FLAG_BIND_PRINC_STATEID,
#endif
B
Benny Halevy 已提交
6911 6912
	};
	struct nfs41_exchange_id_res res = {
6913
		0
B
Benny Halevy 已提交
6914 6915 6916 6917 6918 6919 6920 6921 6922
	};
	int status;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_EXCHANGE_ID],
		.rpc_argp = &args,
		.rpc_resp = &res,
		.rpc_cred = cred,
	};

6923
	nfs4_init_boot_verifier(clp, &verifier);
6924 6925 6926

	status = nfs4_init_uniform_client_string(clp);
	if (status)
6927 6928 6929
		goto out;

	dprintk("NFS call  exchange_id auth=%s, '%s'\n",
6930
		clp->cl_rpcclient->cl_auth->au_ops->au_name,
6931
		clp->cl_owner_id);
B
Benny Halevy 已提交
6932

6933
	res.server_owner = kzalloc(sizeof(struct nfs41_server_owner),
6934
					GFP_NOFS);
6935
	if (unlikely(res.server_owner == NULL)) {
6936 6937 6938
		status = -ENOMEM;
		goto out;
	}
6939

6940
	res.server_scope = kzalloc(sizeof(struct nfs41_server_scope),
6941
					GFP_NOFS);
6942
	if (unlikely(res.server_scope == NULL)) {
6943
		status = -ENOMEM;
6944
		goto out_server_owner;
6945
	}
6946

6947
	res.impl_id = kzalloc(sizeof(struct nfs41_impl_id), GFP_NOFS);
6948
	if (unlikely(res.impl_id == NULL)) {
6949 6950 6951 6952
		status = -ENOMEM;
		goto out_server_scope;
	}

6953 6954 6955 6956 6957 6958 6959 6960 6961 6962 6963 6964 6965 6966 6967 6968
	switch (sp4_how) {
	case SP4_NONE:
		args.state_protect.how = SP4_NONE;
		break;

	case SP4_MACH_CRED:
		args.state_protect = nfs4_sp4_mach_cred_request;
		break;

	default:
		/* unsupported! */
		WARN_ON_ONCE(1);
		status = -EINVAL;
		goto out_server_scope;
	}

6969
	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
6970
	trace_nfs4_exchange_id(clp, status);
6971
	if (status == 0)
6972
		status = nfs4_check_cl_exchange_flags(res.flags);
6973

6974 6975 6976
	if (status == 0)
		status = nfs4_sp4_select_mode(clp, &res.state_protect);

6977
	if (status == 0) {
6978
		clp->cl_clientid = res.clientid;
6979 6980 6981 6982 6983
		clp->cl_exchange_flags = res.flags;
		/* Client ID is not confirmed */
		if (!(res.flags & EXCHGID4_FLAG_CONFIRMED_R)) {
			clear_bit(NFS4_SESSION_ESTABLISHED,
					&clp->cl_session->session_state);
6984
			clp->cl_seqid = res.seqid;
6985
		}
6986

6987 6988 6989
		kfree(clp->cl_serverowner);
		clp->cl_serverowner = res.server_owner;
		res.server_owner = NULL;
6990

6991
		/* use the most recent implementation id */
6992 6993
		kfree(clp->cl_implid);
		clp->cl_implid = res.impl_id;
6994

6995
		if (clp->cl_serverscope != NULL &&
6996
		    !nfs41_same_server_scope(clp->cl_serverscope,
6997 6998 6999 7000
					     res.server_scope)) {
			dprintk("%s: server_scope mismatch detected\n",
				__func__);
			set_bit(NFS4CLNT_SERVER_SCOPE_MISMATCH, &clp->cl_state);
7001 7002
			kfree(clp->cl_serverscope);
			clp->cl_serverscope = NULL;
7003 7004
		}

7005
		if (clp->cl_serverscope == NULL) {
7006
			clp->cl_serverscope = res.server_scope;
7007 7008
			goto out;
		}
7009 7010
	} else
		kfree(res.impl_id);
7011

7012 7013
out_server_owner:
	kfree(res.server_owner);
7014
out_server_scope:
7015 7016
	kfree(res.server_scope);
out:
7017
	if (clp->cl_implid != NULL)
7018
		dprintk("NFS reply exchange_id: Server Implementation ID: "
7019
			"domain: %s, name: %s, date: %llu,%u\n",
7020
			clp->cl_implid->domain, clp->cl_implid->name,
7021 7022
			clp->cl_implid->date.seconds,
			clp->cl_implid->date.nseconds);
7023
	dprintk("NFS reply exchange_id: %d\n", status);
B
Benny Halevy 已提交
7024 7025 7026
	return status;
}

7027 7028 7029 7030 7031 7032 7033 7034 7035 7036 7037 7038 7039 7040 7041 7042 7043 7044 7045 7046 7047 7048 7049 7050 7051 7052 7053 7054 7055
/*
 * nfs4_proc_exchange_id()
 *
 * Returns zero, a negative errno, or a negative NFS4ERR status code.
 *
 * Since the clientid has expired, all compounds using sessions
 * associated with the stale clientid will be returning
 * NFS4ERR_BADSESSION in the sequence operation, and will therefore
 * be in some phase of session reset.
 *
 * Will attempt to negotiate SP4_MACH_CRED if krb5i / krb5p auth is used.
 */
int nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred)
{
	rpc_authflavor_t authflavor = clp->cl_rpcclient->cl_auth->au_flavor;
	int status;

	/* try SP4_MACH_CRED if krb5i/p	*/
	if (authflavor == RPC_AUTH_GSS_KRB5I ||
	    authflavor == RPC_AUTH_GSS_KRB5P) {
		status = _nfs4_proc_exchange_id(clp, cred, SP4_MACH_CRED);
		if (!status)
			return 0;
	}

	/* try SP4_NONE */
	return _nfs4_proc_exchange_id(clp, cred, SP4_NONE);
}

T
Trond Myklebust 已提交
7056 7057 7058 7059 7060 7061 7062 7063 7064 7065 7066
static int _nfs4_proc_destroy_clientid(struct nfs_client *clp,
		struct rpc_cred *cred)
{
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DESTROY_CLIENTID],
		.rpc_argp = clp,
		.rpc_cred = cred,
	};
	int status;

	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
7067
	trace_nfs4_destroy_clientid(clp, status);
T
Trond Myklebust 已提交
7068
	if (status)
7069
		dprintk("NFS: Got error %d from the server %s on "
T
Trond Myklebust 已提交
7070 7071 7072 7073 7074 7075 7076 7077 7078 7079 7080 7081 7082 7083 7084 7085 7086 7087 7088 7089 7090 7091 7092 7093 7094 7095 7096 7097 7098 7099 7100 7101 7102
			"DESTROY_CLIENTID.", status, clp->cl_hostname);
	return status;
}

static int nfs4_proc_destroy_clientid(struct nfs_client *clp,
		struct rpc_cred *cred)
{
	unsigned int loop;
	int ret;

	for (loop = NFS4_MAX_LOOP_ON_RECOVER; loop != 0; loop--) {
		ret = _nfs4_proc_destroy_clientid(clp, cred);
		switch (ret) {
		case -NFS4ERR_DELAY:
		case -NFS4ERR_CLIENTID_BUSY:
			ssleep(1);
			break;
		default:
			return ret;
		}
	}
	return 0;
}

int nfs4_destroy_clientid(struct nfs_client *clp)
{
	struct rpc_cred *cred;
	int ret = 0;

	if (clp->cl_mvops->minor_version < 1)
		goto out;
	if (clp->cl_exchange_flags == 0)
		goto out;
7103 7104
	if (clp->cl_preserve_clid)
		goto out;
7105
	cred = nfs4_get_clid_cred(clp);
T
Trond Myklebust 已提交
7106 7107 7108 7109 7110 7111 7112 7113 7114 7115 7116 7117
	ret = nfs4_proc_destroy_clientid(clp, cred);
	if (cred)
		put_rpccred(cred);
	switch (ret) {
	case 0:
	case -NFS4ERR_STALE_CLIENTID:
		clp->cl_exchange_flags = 0;
	}
out:
	return ret;
}

A
Andy Adamson 已提交
7118 7119 7120 7121 7122 7123 7124 7125 7126 7127 7128 7129 7130 7131 7132
struct nfs4_get_lease_time_data {
	struct nfs4_get_lease_time_args *args;
	struct nfs4_get_lease_time_res *res;
	struct nfs_client *clp;
};

static void nfs4_get_lease_time_prepare(struct rpc_task *task,
					void *calldata)
{
	struct nfs4_get_lease_time_data *data =
			(struct nfs4_get_lease_time_data *)calldata;

	dprintk("--> %s\n", __func__);
	/* just setup sequence, do not trigger session recovery
	   since we're invoked within one */
7133 7134 7135 7136
	nfs41_setup_sequence(data->clp->cl_session,
			&data->args->la_seq_args,
			&data->res->lr_seq_res,
			task);
A
Andy Adamson 已提交
7137 7138 7139 7140 7141 7142 7143 7144 7145 7146 7147 7148 7149
	dprintk("<-- %s\n", __func__);
}

/*
 * Called from nfs4_state_manager thread for session setup, so don't recover
 * from sequence operation or clientid errors.
 */
static void nfs4_get_lease_time_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_get_lease_time_data *data =
			(struct nfs4_get_lease_time_data *)calldata;

	dprintk("--> %s\n", __func__);
7150 7151
	if (!nfs41_sequence_done(task, &data->res->lr_seq_res))
		return;
A
Andy Adamson 已提交
7152 7153 7154 7155 7156 7157
	switch (task->tk_status) {
	case -NFS4ERR_DELAY:
	case -NFS4ERR_GRACE:
		dprintk("%s Retry: tk_status %d\n", __func__, task->tk_status);
		rpc_delay(task, NFS4_POLL_RETRY_MIN);
		task->tk_status = 0;
7158 7159
		/* fall through */
	case -NFS4ERR_RETRY_UNCACHED_REP:
7160
		rpc_restart_call_prepare(task);
A
Andy Adamson 已提交
7161 7162 7163 7164 7165
		return;
	}
	dprintk("<-- %s\n", __func__);
}

7166
static const struct rpc_call_ops nfs4_get_lease_time_ops = {
A
Andy Adamson 已提交
7167 7168 7169 7170 7171 7172 7173 7174 7175 7176 7177 7178 7179 7180 7181 7182 7183 7184 7185 7186 7187 7188 7189 7190 7191
	.rpc_call_prepare = nfs4_get_lease_time_prepare,
	.rpc_call_done = nfs4_get_lease_time_done,
};

int nfs4_proc_get_lease_time(struct nfs_client *clp, struct nfs_fsinfo *fsinfo)
{
	struct rpc_task *task;
	struct nfs4_get_lease_time_args args;
	struct nfs4_get_lease_time_res res = {
		.lr_fsinfo = fsinfo,
	};
	struct nfs4_get_lease_time_data data = {
		.args = &args,
		.res = &res,
		.clp = clp,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GET_LEASE_TIME],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
	struct rpc_task_setup task_setup = {
		.rpc_client = clp->cl_rpcclient,
		.rpc_message = &msg,
		.callback_ops = &nfs4_get_lease_time_ops,
7192 7193
		.callback_data = &data,
		.flags = RPC_TASK_TIMEOUT,
A
Andy Adamson 已提交
7194 7195 7196
	};
	int status;

7197
	nfs4_init_sequence(&args.la_seq_args, &res.lr_seq_res, 0);
7198
	nfs4_set_sequence_privileged(&args.la_seq_args);
A
Andy Adamson 已提交
7199 7200 7201 7202 7203 7204 7205 7206 7207 7208 7209 7210 7211 7212
	dprintk("--> %s\n", __func__);
	task = rpc_run_task(&task_setup);

	if (IS_ERR(task))
		status = PTR_ERR(task);
	else {
		status = task->tk_status;
		rpc_put_task(task);
	}
	dprintk("<-- %s return %d\n", __func__, status);

	return status;
}

A
Andy Adamson 已提交
7213 7214 7215 7216 7217 7218 7219 7220 7221 7222 7223
/*
 * Initialize the values to be used by the client in CREATE_SESSION
 * If nfs4_init_session set the fore channel request and response sizes,
 * use them.
 *
 * Set the back channel max_resp_sz_cached to zero to force the client to
 * always set csa_cachethis to FALSE because the current implementation
 * of the back channel DRC only supports caching the CB_SEQUENCE operation.
 */
static void nfs4_init_channel_attrs(struct nfs41_create_session_args *args)
{
7224 7225 7226 7227
	unsigned int max_rqst_sz, max_resp_sz;

	max_rqst_sz = NFS_MAX_FILE_IO_SIZE + nfs41_maxwrite_overhead;
	max_resp_sz = NFS_MAX_FILE_IO_SIZE + nfs41_maxread_overhead;
A
Andy Adamson 已提交
7228 7229

	/* Fore channel attributes */
7230 7231
	args->fc_attrs.max_rqst_sz = max_rqst_sz;
	args->fc_attrs.max_resp_sz = max_resp_sz;
A
Andy Adamson 已提交
7232
	args->fc_attrs.max_ops = NFS4_MAX_OPS;
7233
	args->fc_attrs.max_reqs = max_session_slots;
A
Andy Adamson 已提交
7234 7235

	dprintk("%s: Fore Channel : max_rqst_sz=%u max_resp_sz=%u "
7236
		"max_ops=%u max_reqs=%u\n",
A
Andy Adamson 已提交
7237 7238
		__func__,
		args->fc_attrs.max_rqst_sz, args->fc_attrs.max_resp_sz,
7239
		args->fc_attrs.max_ops, args->fc_attrs.max_reqs);
A
Andy Adamson 已提交
7240 7241 7242 7243 7244 7245 7246 7247 7248 7249 7250 7251 7252 7253 7254 7255

	/* Back channel attributes */
	args->bc_attrs.max_rqst_sz = PAGE_SIZE;
	args->bc_attrs.max_resp_sz = PAGE_SIZE;
	args->bc_attrs.max_resp_sz_cached = 0;
	args->bc_attrs.max_ops = NFS4_MAX_BACK_CHANNEL_OPS;
	args->bc_attrs.max_reqs = 1;

	dprintk("%s: Back Channel : max_rqst_sz=%u max_resp_sz=%u "
		"max_resp_sz_cached=%u max_ops=%u max_reqs=%u\n",
		__func__,
		args->bc_attrs.max_rqst_sz, args->bc_attrs.max_resp_sz,
		args->bc_attrs.max_resp_sz_cached, args->bc_attrs.max_ops,
		args->bc_attrs.max_reqs);
}

7256 7257
static int nfs4_verify_fore_channel_attrs(struct nfs41_create_session_args *args,
		struct nfs41_create_session_res *res)
7258
{
7259
	struct nfs4_channel_attrs *sent = &args->fc_attrs;
7260
	struct nfs4_channel_attrs *rcvd = &res->fc_attrs;
7261 7262 7263 7264 7265 7266 7267 7268 7269 7270 7271 7272 7273

	if (rcvd->max_resp_sz > sent->max_resp_sz)
		return -EINVAL;
	/*
	 * Our requested max_ops is the minimum we need; we're not
	 * prepared to break up compounds into smaller pieces than that.
	 * So, no point even trying to continue if the server won't
	 * cooperate:
	 */
	if (rcvd->max_ops < sent->max_ops)
		return -EINVAL;
	if (rcvd->max_reqs == 0)
		return -EINVAL;
7274 7275
	if (rcvd->max_reqs > NFS4_MAX_SLOT_TABLE)
		rcvd->max_reqs = NFS4_MAX_SLOT_TABLE;
7276
	return 0;
7277 7278
}

7279 7280
static int nfs4_verify_back_channel_attrs(struct nfs41_create_session_args *args,
		struct nfs41_create_session_res *res)
7281 7282
{
	struct nfs4_channel_attrs *sent = &args->bc_attrs;
7283
	struct nfs4_channel_attrs *rcvd = &res->bc_attrs;
7284

7285 7286
	if (!(res->flags & SESSION4_BACK_CHAN))
		goto out;
7287 7288 7289 7290 7291 7292 7293
	if (rcvd->max_rqst_sz > sent->max_rqst_sz)
		return -EINVAL;
	if (rcvd->max_resp_sz < sent->max_resp_sz)
		return -EINVAL;
	if (rcvd->max_resp_sz_cached > sent->max_resp_sz_cached)
		return -EINVAL;
	/* These would render the backchannel useless: */
7294
	if (rcvd->max_ops != sent->max_ops)
7295
		return -EINVAL;
7296
	if (rcvd->max_reqs != sent->max_reqs)
7297
		return -EINVAL;
7298
out:
7299 7300
	return 0;
}
7301 7302

static int nfs4_verify_channel_attrs(struct nfs41_create_session_args *args,
7303
				     struct nfs41_create_session_res *res)
7304
{
7305
	int ret;
7306

7307
	ret = nfs4_verify_fore_channel_attrs(args, res);
7308 7309
	if (ret)
		return ret;
7310 7311 7312 7313 7314 7315 7316
	return nfs4_verify_back_channel_attrs(args, res);
}

static void nfs4_update_session(struct nfs4_session *session,
		struct nfs41_create_session_res *res)
{
	nfs4_copy_sessionid(&session->sess_id, &res->sessionid);
7317 7318 7319
	/* Mark client id and session as being confirmed */
	session->clp->cl_exchange_flags |= EXCHGID4_FLAG_CONFIRMED_R;
	set_bit(NFS4_SESSION_ESTABLISHED, &session->session_state);
7320 7321
	session->flags = res->flags;
	memcpy(&session->fc_attrs, &res->fc_attrs, sizeof(session->fc_attrs));
7322 7323 7324
	if (res->flags & SESSION4_BACK_CHAN)
		memcpy(&session->bc_attrs, &res->bc_attrs,
				sizeof(session->bc_attrs));
7325 7326
}

7327 7328
static int _nfs4_proc_create_session(struct nfs_client *clp,
		struct rpc_cred *cred)
A
Andy Adamson 已提交
7329 7330 7331 7332
{
	struct nfs4_session *session = clp->cl_session;
	struct nfs41_create_session_args args = {
		.client = clp,
7333 7334
		.clientid = clp->cl_clientid,
		.seqid = clp->cl_seqid,
A
Andy Adamson 已提交
7335 7336
		.cb_program = NFS4_CALLBACK,
	};
7337 7338
	struct nfs41_create_session_res res;

A
Andy Adamson 已提交
7339 7340 7341 7342
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CREATE_SESSION],
		.rpc_argp = &args,
		.rpc_resp = &res,
7343
		.rpc_cred = cred,
A
Andy Adamson 已提交
7344 7345 7346 7347
	};
	int status;

	nfs4_init_channel_attrs(&args);
7348
	args.flags = (SESSION4_PERSIST | SESSION4_BACK_CHAN);
A
Andy Adamson 已提交
7349

7350
	status = rpc_call_sync(session->clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
7351
	trace_nfs4_create_session(clp, status);
A
Andy Adamson 已提交
7352

7353
	if (!status) {
7354
		/* Verify the session's negotiated channel_attrs values */
7355
		status = nfs4_verify_channel_attrs(&args, &res);
A
Andy Adamson 已提交
7356
		/* Increment the clientid slot sequence id */
7357 7358 7359 7360 7361
		if (clp->cl_seqid == res.seqid)
			clp->cl_seqid++;
		if (status)
			goto out;
		nfs4_update_session(session, &res);
A
Andy Adamson 已提交
7362
	}
7363
out:
A
Andy Adamson 已提交
7364 7365 7366 7367 7368 7369 7370 7371
	return status;
}

/*
 * Issues a CREATE_SESSION operation to the server.
 * It is the responsibility of the caller to verify the session is
 * expired before calling this routine.
 */
7372
int nfs4_proc_create_session(struct nfs_client *clp, struct rpc_cred *cred)
A
Andy Adamson 已提交
7373 7374 7375 7376 7377 7378 7379
{
	int status;
	unsigned *ptr;
	struct nfs4_session *session = clp->cl_session;

	dprintk("--> %s clp=%p session=%p\n", __func__, clp, session);

7380
	status = _nfs4_proc_create_session(clp, cred);
A
Andy Adamson 已提交
7381 7382 7383
	if (status)
		goto out;

7384 7385 7386
	/* Init or reset the session slot tables */
	status = nfs4_setup_session_slot_tables(session);
	dprintk("slot table setup returned %d\n", status);
A
Andy Adamson 已提交
7387 7388 7389 7390 7391 7392 7393 7394 7395 7396 7397
	if (status)
		goto out;

	ptr = (unsigned *)&session->sess_id.data[0];
	dprintk("%s client>seqid %d sessionid %u:%u:%u:%u\n", __func__,
		clp->cl_seqid, ptr[0], ptr[1], ptr[2], ptr[3]);
out:
	dprintk("<-- %s\n", __func__);
	return status;
}

A
Andy Adamson 已提交
7398 7399 7400 7401
/*
 * Issue the over-the-wire RPC DESTROY_SESSION.
 * The caller must serialize access to this routine.
 */
7402 7403
int nfs4_proc_destroy_session(struct nfs4_session *session,
		struct rpc_cred *cred)
A
Andy Adamson 已提交
7404
{
7405 7406 7407 7408 7409
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DESTROY_SESSION],
		.rpc_argp = session,
		.rpc_cred = cred,
	};
A
Andy Adamson 已提交
7410 7411 7412 7413 7414
	int status = 0;

	dprintk("--> nfs4_proc_destroy_session\n");

	/* session is still being setup */
7415 7416
	if (!test_and_clear_bit(NFS4_SESSION_ESTABLISHED, &session->session_state))
		return 0;
A
Andy Adamson 已提交
7417

7418
	status = rpc_call_sync(session->clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
7419
	trace_nfs4_destroy_session(session->clp, status);
A
Andy Adamson 已提交
7420 7421

	if (status)
7422
		dprintk("NFS: Got error %d from the server on DESTROY_SESSION. "
A
Andy Adamson 已提交
7423 7424 7425 7426 7427 7428
			"Session has been destroyed regardless...\n", status);

	dprintk("<-- nfs4_proc_destroy_session\n");
	return status;
}

A
Andy Adamson 已提交
7429 7430 7431
/*
 * Renew the cl_session lease.
 */
7432 7433 7434 7435 7436 7437
struct nfs4_sequence_data {
	struct nfs_client *clp;
	struct nfs4_sequence_args args;
	struct nfs4_sequence_res res;
};

7438 7439
static void nfs41_sequence_release(void *data)
{
7440 7441
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
7442

7443 7444 7445
	if (atomic_read(&clp->cl_count) > 1)
		nfs4_schedule_state_renewal(clp);
	nfs_put_client(clp);
7446
	kfree(calldata);
7447 7448
}

7449 7450 7451 7452 7453 7454 7455
static int nfs41_sequence_handle_errors(struct rpc_task *task, struct nfs_client *clp)
{
	switch(task->tk_status) {
	case -NFS4ERR_DELAY:
		rpc_delay(task, NFS4_POLL_RETRY_MAX);
		return -EAGAIN;
	default:
7456
		nfs4_schedule_lease_recovery(clp);
7457 7458 7459 7460
	}
	return 0;
}

7461
static void nfs41_sequence_call_done(struct rpc_task *task, void *data)
A
Andy Adamson 已提交
7462
{
7463 7464
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
A
Andy Adamson 已提交
7465

7466 7467
	if (!nfs41_sequence_done(task, task->tk_msg.rpc_resp))
		return;
A
Andy Adamson 已提交
7468

7469
	trace_nfs4_sequence(clp, task->tk_status);
A
Andy Adamson 已提交
7470 7471
	if (task->tk_status < 0) {
		dprintk("%s ERROR %d\n", __func__, task->tk_status);
7472 7473
		if (atomic_read(&clp->cl_count) == 1)
			goto out;
A
Andy Adamson 已提交
7474

7475 7476
		if (nfs41_sequence_handle_errors(task, clp) == -EAGAIN) {
			rpc_restart_call_prepare(task);
A
Andy Adamson 已提交
7477 7478 7479 7480
			return;
		}
	}
	dprintk("%s rpc_cred %p\n", __func__, task->tk_msg.rpc_cred);
7481
out:
A
Andy Adamson 已提交
7482 7483 7484 7485 7486
	dprintk("<-- %s\n", __func__);
}

static void nfs41_sequence_prepare(struct rpc_task *task, void *data)
{
7487 7488
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
A
Andy Adamson 已提交
7489 7490 7491 7492 7493 7494
	struct nfs4_sequence_args *args;
	struct nfs4_sequence_res *res;

	args = task->tk_msg.rpc_argp;
	res = task->tk_msg.rpc_resp;

7495
	nfs41_setup_sequence(clp->cl_session, args, res, task);
A
Andy Adamson 已提交
7496 7497 7498 7499 7500
}

static const struct rpc_call_ops nfs41_sequence_ops = {
	.rpc_call_done = nfs41_sequence_call_done,
	.rpc_call_prepare = nfs41_sequence_prepare,
7501
	.rpc_release = nfs41_sequence_release,
A
Andy Adamson 已提交
7502 7503
};

7504 7505 7506
static struct rpc_task *_nfs41_proc_sequence(struct nfs_client *clp,
		struct rpc_cred *cred,
		bool is_privileged)
A
Andy Adamson 已提交
7507
{
7508
	struct nfs4_sequence_data *calldata;
A
Andy Adamson 已提交
7509 7510 7511 7512
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SEQUENCE],
		.rpc_cred = cred,
	};
7513 7514 7515
	struct rpc_task_setup task_setup_data = {
		.rpc_client = clp->cl_rpcclient,
		.rpc_message = &msg,
7516
		.callback_ops = &nfs41_sequence_ops,
7517
		.flags = RPC_TASK_ASYNC | RPC_TASK_TIMEOUT,
7518
	};
A
Andy Adamson 已提交
7519

7520
	if (!atomic_inc_not_zero(&clp->cl_count))
7521
		return ERR_PTR(-EIO);
7522
	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
7523
	if (calldata == NULL) {
7524
		nfs_put_client(clp);
7525
		return ERR_PTR(-ENOMEM);
A
Andy Adamson 已提交
7526
	}
7527
	nfs4_init_sequence(&calldata->args, &calldata->res, 0);
7528 7529
	if (is_privileged)
		nfs4_set_sequence_privileged(&calldata->args);
7530 7531 7532
	msg.rpc_argp = &calldata->args;
	msg.rpc_resp = &calldata->res;
	calldata->clp = clp;
7533
	task_setup_data.callback_data = calldata;
A
Andy Adamson 已提交
7534

7535 7536 7537
	return rpc_run_task(&task_setup_data);
}

7538
static int nfs41_proc_async_sequence(struct nfs_client *clp, struct rpc_cred *cred, unsigned renew_flags)
7539 7540 7541 7542
{
	struct rpc_task *task;
	int ret = 0;

7543
	if ((renew_flags & NFS4_RENEW_TIMEOUT) == 0)
7544
		return -EAGAIN;
7545
	task = _nfs41_proc_sequence(clp, cred, false);
7546 7547 7548
	if (IS_ERR(task))
		ret = PTR_ERR(task);
	else
7549
		rpc_put_task_async(task);
7550 7551 7552 7553 7554 7555 7556 7557 7558
	dprintk("<-- %s status=%d\n", __func__, ret);
	return ret;
}

static int nfs4_proc_sequence(struct nfs_client *clp, struct rpc_cred *cred)
{
	struct rpc_task *task;
	int ret;

7559
	task = _nfs41_proc_sequence(clp, cred, true);
7560 7561 7562 7563 7564
	if (IS_ERR(task)) {
		ret = PTR_ERR(task);
		goto out;
	}
	ret = rpc_wait_for_completion_task(task);
7565 7566 7567 7568 7569
	if (!ret) {
		struct nfs4_sequence_res *res = task->tk_msg.rpc_resp;

		if (task->tk_status == 0)
			nfs41_handle_sequence_flag_errors(clp, res->sr_status_flags);
7570
		ret = task->tk_status;
7571
	}
7572 7573 7574 7575
	rpc_put_task(task);
out:
	dprintk("<-- %s status=%d\n", __func__, ret);
	return ret;
A
Andy Adamson 已提交
7576 7577
}

7578 7579 7580 7581 7582 7583 7584 7585 7586 7587
struct nfs4_reclaim_complete_data {
	struct nfs_client *clp;
	struct nfs41_reclaim_complete_args arg;
	struct nfs41_reclaim_complete_res res;
};

static void nfs4_reclaim_complete_prepare(struct rpc_task *task, void *data)
{
	struct nfs4_reclaim_complete_data *calldata = data;

7588 7589 7590 7591
	nfs41_setup_sequence(calldata->clp->cl_session,
			&calldata->arg.seq_args,
			&calldata->res.seq_res,
			task);
7592 7593
}

7594 7595 7596 7597 7598 7599 7600 7601 7602
static int nfs41_reclaim_complete_handle_errors(struct rpc_task *task, struct nfs_client *clp)
{
	switch(task->tk_status) {
	case 0:
	case -NFS4ERR_COMPLETE_ALREADY:
	case -NFS4ERR_WRONG_CRED: /* What to do here? */
		break;
	case -NFS4ERR_DELAY:
		rpc_delay(task, NFS4_POLL_RETRY_MAX);
7603 7604
		/* fall through */
	case -NFS4ERR_RETRY_UNCACHED_REP:
7605 7606
		return -EAGAIN;
	default:
7607
		nfs4_schedule_lease_recovery(clp);
7608 7609 7610 7611
	}
	return 0;
}

7612 7613 7614 7615 7616 7617 7618
static void nfs4_reclaim_complete_done(struct rpc_task *task, void *data)
{
	struct nfs4_reclaim_complete_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
	struct nfs4_sequence_res *res = &calldata->res.seq_res;

	dprintk("--> %s\n", __func__);
7619 7620
	if (!nfs41_sequence_done(task, res))
		return;
7621

7622
	trace_nfs4_reclaim_complete(clp, task->tk_status);
7623 7624 7625 7626
	if (nfs41_reclaim_complete_handle_errors(task, clp) == -EAGAIN) {
		rpc_restart_call_prepare(task);
		return;
	}
7627 7628 7629 7630 7631 7632 7633 7634 7635 7636 7637 7638 7639 7640 7641 7642 7643 7644 7645
	dprintk("<-- %s\n", __func__);
}

static void nfs4_free_reclaim_complete_data(void *data)
{
	struct nfs4_reclaim_complete_data *calldata = data;

	kfree(calldata);
}

static const struct rpc_call_ops nfs4_reclaim_complete_call_ops = {
	.rpc_call_prepare = nfs4_reclaim_complete_prepare,
	.rpc_call_done = nfs4_reclaim_complete_done,
	.rpc_release = nfs4_free_reclaim_complete_data,
};

/*
 * Issue a global reclaim complete.
 */
7646 7647
static int nfs41_proc_reclaim_complete(struct nfs_client *clp,
		struct rpc_cred *cred)
7648 7649 7650 7651 7652
{
	struct nfs4_reclaim_complete_data *calldata;
	struct rpc_task *task;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RECLAIM_COMPLETE],
7653
		.rpc_cred = cred,
7654 7655 7656 7657 7658 7659 7660 7661 7662 7663
	};
	struct rpc_task_setup task_setup_data = {
		.rpc_client = clp->cl_rpcclient,
		.rpc_message = &msg,
		.callback_ops = &nfs4_reclaim_complete_call_ops,
		.flags = RPC_TASK_ASYNC,
	};
	int status = -ENOMEM;

	dprintk("--> %s\n", __func__);
7664
	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
7665 7666 7667 7668 7669
	if (calldata == NULL)
		goto out;
	calldata->clp = clp;
	calldata->arg.one_fs = 0;

7670
	nfs4_init_sequence(&calldata->arg.seq_args, &calldata->res.seq_res, 0);
7671
	nfs4_set_sequence_privileged(&calldata->arg.seq_args);
7672 7673 7674 7675
	msg.rpc_argp = &calldata->arg;
	msg.rpc_resp = &calldata->res;
	task_setup_data.callback_data = calldata;
	task = rpc_run_task(&task_setup_data);
7676
	if (IS_ERR(task)) {
7677
		status = PTR_ERR(task);
7678 7679
		goto out;
	}
7680 7681 7682
	status = nfs4_wait_for_completion_rpc_task(task);
	if (status == 0)
		status = task->tk_status;
7683
	rpc_put_task(task);
7684
	return 0;
7685 7686 7687 7688
out:
	dprintk("<-- %s status=%d\n", __func__, status);
	return status;
}
7689 7690 7691 7692 7693

static void
nfs4_layoutget_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;
F
Fred Isaman 已提交
7694
	struct nfs_server *server = NFS_SERVER(lgp->args.inode);
7695
	struct nfs4_session *session = nfs4_get_session(server);
7696 7697

	dprintk("--> %s\n", __func__);
F
Fred Isaman 已提交
7698 7699 7700 7701 7702
	/* Note the is a race here, where a CB_LAYOUTRECALL can come in
	 * right now covering the LAYOUTGET we are about to send.
	 * However, that is not so catastrophic, and there seems
	 * to be no way to prevent it completely.
	 */
7703
	if (nfs41_setup_sequence(session, &lgp->args.seq_args,
7704
				&lgp->res.seq_res, task))
7705
		return;
7706 7707
	if (pnfs_choose_layoutget_stateid(&lgp->args.stateid,
					  NFS_I(lgp->args.inode)->layout,
7708
					  &lgp->args.range,
7709 7710 7711
					  lgp->args.ctx->state)) {
		rpc_exit(task, NFS4_OK);
	}
7712 7713 7714 7715 7716
}

static void nfs4_layoutget_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;
7717 7718 7719 7720
	struct inode *inode = lgp->args.inode;
	struct nfs_server *server = NFS_SERVER(inode);
	struct pnfs_layout_hdr *lo;
	struct nfs4_state *state = NULL;
7721
	unsigned long timeo, now, giveup;
7722

7723
	dprintk("--> %s tk_status => %d\n", __func__, -task->tk_status);
7724

7725
	if (!nfs41_sequence_done(task, &lgp->res.seq_res))
7726
		goto out;
7727 7728 7729

	switch (task->tk_status) {
	case 0:
7730
		goto out;
7731 7732 7733 7734
	/*
	 * NFS4ERR_LAYOUTTRYLATER is a conflict with another client
	 * (or clients) writing to the same RAID stripe
	 */
7735
	case -NFS4ERR_LAYOUTTRYLATER:
7736 7737 7738 7739
	/*
	 * NFS4ERR_RECALLCONFLICT is when conflict with self (must recall
	 * existing layout before getting a new one).
	 */
7740
	case -NFS4ERR_RECALLCONFLICT:
7741 7742
		timeo = rpc_get_timeout(task->tk_client);
		giveup = lgp->args.timestamp + timeo;
7743 7744 7745 7746 7747 7748 7749 7750 7751 7752 7753 7754 7755 7756 7757 7758 7759 7760 7761 7762
		now = jiffies;
		if (time_after(giveup, now)) {
			unsigned long delay;

			/* Delay for:
			 * - Not less then NFS4_POLL_RETRY_MIN.
			 * - One last time a jiffie before we give up
			 * - exponential backoff (time_now minus start_attempt)
			 */
			delay = max_t(unsigned long, NFS4_POLL_RETRY_MIN,
				    min((giveup - now - 1),
					now - lgp->args.timestamp));

			dprintk("%s: NFS4ERR_RECALLCONFLICT waiting %lu\n",
				__func__, delay);
			rpc_delay(task, delay);
			task->tk_status = 0;
			rpc_restart_call_prepare(task);
			goto out; /* Do not call nfs4_async_handle_error() */
		}
7763 7764 7765 7766 7767 7768 7769 7770 7771 7772 7773 7774
		break;
	case -NFS4ERR_EXPIRED:
	case -NFS4ERR_BAD_STATEID:
		spin_lock(&inode->i_lock);
		lo = NFS_I(inode)->layout;
		if (!lo || list_empty(&lo->plh_segs)) {
			spin_unlock(&inode->i_lock);
			/* If the open stateid was bad, then recover it. */
			state = lgp->args.ctx->state;
		} else {
			LIST_HEAD(head);

7775 7776 7777 7778
			/*
			 * Mark the bad layout state as invalid, then retry
			 * with the current stateid.
			 */
7779 7780 7781
			pnfs_mark_matching_lsegs_invalid(lo, &head, NULL);
			spin_unlock(&inode->i_lock);
			pnfs_free_lseg_list(&head);
7782 7783 7784
	
			task->tk_status = 0;
			rpc_restart_call_prepare(task);
7785 7786
		}
	}
7787
	if (nfs4_async_handle_error(task, server, state, NULL) == -EAGAIN)
7788 7789
		rpc_restart_call_prepare(task);
out:
7790 7791 7792
	dprintk("<-- %s\n", __func__);
}

7793 7794 7795 7796 7797 7798 7799 7800 7801 7802 7803 7804 7805 7806 7807 7808 7809 7810 7811 7812 7813 7814 7815 7816 7817 7818 7819 7820 7821 7822 7823 7824 7825 7826 7827 7828 7829 7830 7831 7832 7833 7834 7835 7836
static size_t max_response_pages(struct nfs_server *server)
{
	u32 max_resp_sz = server->nfs_client->cl_session->fc_attrs.max_resp_sz;
	return nfs_page_array_len(0, max_resp_sz);
}

static void nfs4_free_pages(struct page **pages, size_t size)
{
	int i;

	if (!pages)
		return;

	for (i = 0; i < size; i++) {
		if (!pages[i])
			break;
		__free_page(pages[i]);
	}
	kfree(pages);
}

static struct page **nfs4_alloc_pages(size_t size, gfp_t gfp_flags)
{
	struct page **pages;
	int i;

	pages = kcalloc(size, sizeof(struct page *), gfp_flags);
	if (!pages) {
		dprintk("%s: can't alloc array of %zu pages\n", __func__, size);
		return NULL;
	}

	for (i = 0; i < size; i++) {
		pages[i] = alloc_page(gfp_flags);
		if (!pages[i]) {
			dprintk("%s: failed to allocate page\n", __func__);
			nfs4_free_pages(pages, size);
			return NULL;
		}
	}

	return pages;
}

7837 7838 7839
static void nfs4_layoutget_release(void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;
7840 7841
	struct inode *inode = lgp->args.inode;
	struct nfs_server *server = NFS_SERVER(inode);
7842
	size_t max_pages = max_response_pages(server);
7843 7844

	dprintk("--> %s\n", __func__);
7845
	nfs4_free_pages(lgp->args.layout.pages, max_pages);
7846
	pnfs_put_layout_hdr(NFS_I(inode)->layout);
7847 7848 7849 7850 7851 7852 7853 7854 7855 7856 7857
	put_nfs_open_context(lgp->args.ctx);
	kfree(calldata);
	dprintk("<-- %s\n", __func__);
}

static const struct rpc_call_ops nfs4_layoutget_call_ops = {
	.rpc_call_prepare = nfs4_layoutget_prepare,
	.rpc_call_done = nfs4_layoutget_done,
	.rpc_release = nfs4_layoutget_release,
};

7858 7859
struct pnfs_layout_segment *
nfs4_proc_layoutget(struct nfs4_layoutget *lgp, gfp_t gfp_flags)
7860
{
7861 7862
	struct inode *inode = lgp->args.inode;
	struct nfs_server *server = NFS_SERVER(inode);
7863
	size_t max_pages = max_response_pages(server);
7864 7865 7866 7867 7868
	struct rpc_task *task;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LAYOUTGET],
		.rpc_argp = &lgp->args,
		.rpc_resp = &lgp->res,
7869
		.rpc_cred = lgp->cred,
7870 7871 7872 7873 7874 7875 7876 7877
	};
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
		.rpc_message = &msg,
		.callback_ops = &nfs4_layoutget_call_ops,
		.callback_data = lgp,
		.flags = RPC_TASK_ASYNC,
	};
7878
	struct pnfs_layout_segment *lseg = NULL;
7879 7880 7881 7882
	int status = 0;

	dprintk("--> %s\n", __func__);

7883 7884 7885
	/* nfs4_layoutget_release calls pnfs_put_layout_hdr */
	pnfs_get_layout_hdr(NFS_I(inode)->layout);

7886 7887 7888
	lgp->args.layout.pages = nfs4_alloc_pages(max_pages, gfp_flags);
	if (!lgp->args.layout.pages) {
		nfs4_layoutget_release(lgp);
7889
		return ERR_PTR(-ENOMEM);
7890 7891
	}
	lgp->args.layout.pglen = max_pages * PAGE_SIZE;
7892
	lgp->args.timestamp = jiffies;
7893

7894
	lgp->res.layoutp = &lgp->args.layout;
7895
	lgp->res.seq_res.sr_slot = NULL;
7896
	nfs4_init_sequence(&lgp->args.seq_args, &lgp->res.seq_res, 0);
7897

7898 7899
	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task))
7900
		return ERR_CAST(task);
7901
	status = nfs4_wait_for_completion_rpc_task(task);
F
Fred Isaman 已提交
7902 7903
	if (status == 0)
		status = task->tk_status;
7904 7905 7906 7907
	trace_nfs4_layoutget(lgp->args.ctx,
			&lgp->args.range,
			&lgp->res.range,
			status);
7908 7909
	/* if layoutp->len is 0, nfs4_layoutget_prepare called rpc_exit */
	if (status == 0 && lgp->res.layoutp->len)
7910
		lseg = pnfs_layout_process(lgp);
7911 7912
	rpc_put_task(task);
	dprintk("<-- %s status=%d\n", __func__, status);
7913 7914 7915
	if (status)
		return ERR_PTR(status);
	return lseg;
7916 7917
}

B
Benny Halevy 已提交
7918 7919 7920 7921 7922 7923
static void
nfs4_layoutreturn_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutreturn *lrp = calldata;

	dprintk("--> %s\n", __func__);
7924 7925 7926 7927
	nfs41_setup_sequence(lrp->clp->cl_session,
			&lrp->args.seq_args,
			&lrp->res.seq_res,
			task);
B
Benny Halevy 已提交
7928 7929 7930 7931 7932 7933 7934 7935 7936
}

static void nfs4_layoutreturn_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutreturn *lrp = calldata;
	struct nfs_server *server;

	dprintk("--> %s\n", __func__);

7937
	if (!nfs41_sequence_done(task, &lrp->res.seq_res))
B
Benny Halevy 已提交
7938 7939 7940
		return;

	server = NFS_SERVER(lrp->args.inode);
7941 7942 7943 7944 7945 7946
	switch (task->tk_status) {
	default:
		task->tk_status = 0;
	case 0:
		break;
	case -NFS4ERR_DELAY:
7947
		if (nfs4_async_handle_error(task, server, NULL, NULL) != -EAGAIN)
7948
			break;
7949
		rpc_restart_call_prepare(task);
B
Benny Halevy 已提交
7950 7951 7952 7953 7954 7955 7956 7957
		return;
	}
	dprintk("<-- %s\n", __func__);
}

static void nfs4_layoutreturn_release(void *calldata)
{
	struct nfs4_layoutreturn *lrp = calldata;
7958
	struct pnfs_layout_hdr *lo = lrp->args.layout;
B
Benny Halevy 已提交
7959 7960

	dprintk("--> %s\n", __func__);
7961 7962 7963
	spin_lock(&lo->plh_inode->i_lock);
	if (lrp->res.lrs_present)
		pnfs_set_layout_stateid(lo, &lrp->res.stateid, true);
7964
	pnfs_clear_layoutreturn_waitbit(lo);
7965 7966
	clear_bit(NFS_LAYOUT_RETURN_BEFORE_CLOSE, &lo->plh_flags);
	rpc_wake_up(&NFS_SERVER(lo->plh_inode)->roc_rpcwaitq);
7967 7968
	lo->plh_block_lgets--;
	spin_unlock(&lo->plh_inode->i_lock);
7969
	pnfs_put_layout_hdr(lrp->args.layout);
7970
	nfs_iput_and_deactive(lrp->inode);
B
Benny Halevy 已提交
7971 7972 7973 7974 7975 7976 7977 7978 7979 7980
	kfree(calldata);
	dprintk("<-- %s\n", __func__);
}

static const struct rpc_call_ops nfs4_layoutreturn_call_ops = {
	.rpc_call_prepare = nfs4_layoutreturn_prepare,
	.rpc_call_done = nfs4_layoutreturn_done,
	.rpc_release = nfs4_layoutreturn_release,
};

7981
int nfs4_proc_layoutreturn(struct nfs4_layoutreturn *lrp, bool sync)
B
Benny Halevy 已提交
7982 7983 7984 7985 7986 7987
{
	struct rpc_task *task;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LAYOUTRETURN],
		.rpc_argp = &lrp->args,
		.rpc_resp = &lrp->res,
7988
		.rpc_cred = lrp->cred,
B
Benny Halevy 已提交
7989 7990
	};
	struct rpc_task_setup task_setup_data = {
7991
		.rpc_client = NFS_SERVER(lrp->args.inode)->client,
B
Benny Halevy 已提交
7992 7993 7994 7995
		.rpc_message = &msg,
		.callback_ops = &nfs4_layoutreturn_call_ops,
		.callback_data = lrp,
	};
7996
	int status = 0;
B
Benny Halevy 已提交
7997 7998

	dprintk("--> %s\n", __func__);
7999 8000 8001 8002 8003 8004 8005 8006
	if (!sync) {
		lrp->inode = nfs_igrab_and_active(lrp->args.inode);
		if (!lrp->inode) {
			nfs4_layoutreturn_release(lrp);
			return -EAGAIN;
		}
		task_setup_data.flags |= RPC_TASK_ASYNC;
	}
8007
	nfs4_init_sequence(&lrp->args.seq_args, &lrp->res.seq_res, 1);
B
Benny Halevy 已提交
8008 8009 8010
	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task))
		return PTR_ERR(task);
8011 8012
	if (sync)
		status = task->tk_status;
8013
	trace_nfs4_layoutreturn(lrp->args.inode, status);
B
Benny Halevy 已提交
8014 8015 8016 8017 8018
	dprintk("<-- %s status=%d\n", __func__, status);
	rpc_put_task(task);
	return status;
}

8019
static int
8020 8021 8022
_nfs4_proc_getdeviceinfo(struct nfs_server *server,
		struct pnfs_device *pdev,
		struct rpc_cred *cred)
8023 8024 8025
{
	struct nfs4_getdeviceinfo_args args = {
		.pdev = pdev,
8026 8027
		.notify_types = NOTIFY_DEVICEID4_CHANGE |
			NOTIFY_DEVICEID4_DELETE,
8028 8029 8030 8031 8032 8033 8034 8035
	};
	struct nfs4_getdeviceinfo_res res = {
		.pdev = pdev,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETDEVICEINFO],
		.rpc_argp = &args,
		.rpc_resp = &res,
8036
		.rpc_cred = cred,
8037 8038 8039 8040
	};
	int status;

	dprintk("--> %s\n", __func__);
8041
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
8042 8043
	if (res.notification & ~args.notify_types)
		dprintk("%s: unsupported notification\n", __func__);
8044 8045
	if (res.notification != args.notify_types)
		pdev->nocache = 1;
8046

8047 8048 8049 8050 8051
	dprintk("<-- %s status=%d\n", __func__, status);

	return status;
}

8052 8053 8054
int nfs4_proc_getdeviceinfo(struct nfs_server *server,
		struct pnfs_device *pdev,
		struct rpc_cred *cred)
8055 8056 8057 8058 8059 8060
{
	struct nfs4_exception exception = { };
	int err;

	do {
		err = nfs4_handle_exception(server,
8061
					_nfs4_proc_getdeviceinfo(server, pdev, cred),
8062 8063 8064 8065 8066 8067
					&exception);
	} while (exception.retry);
	return err;
}
EXPORT_SYMBOL_GPL(nfs4_proc_getdeviceinfo);

A
Andy Adamson 已提交
8068 8069 8070 8071
static void nfs4_layoutcommit_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutcommit_data *data = calldata;
	struct nfs_server *server = NFS_SERVER(data->args.inode);
8072
	struct nfs4_session *session = nfs4_get_session(server);
A
Andy Adamson 已提交
8073

8074 8075 8076 8077
	nfs41_setup_sequence(session,
			&data->args.seq_args,
			&data->res.seq_res,
			task);
A
Andy Adamson 已提交
8078 8079 8080 8081 8082 8083 8084 8085
}

static void
nfs4_layoutcommit_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutcommit_data *data = calldata;
	struct nfs_server *server = NFS_SERVER(data->args.inode);

8086
	if (!nfs41_sequence_done(task, &data->res.seq_res))
A
Andy Adamson 已提交
8087 8088 8089
		return;

	switch (task->tk_status) { /* Just ignore these failures */
8090 8091 8092 8093
	case -NFS4ERR_DELEG_REVOKED: /* layout was recalled */
	case -NFS4ERR_BADIOMODE:     /* no IOMODE_RW layout for range */
	case -NFS4ERR_BADLAYOUT:     /* no layout */
	case -NFS4ERR_GRACE:	    /* loca_recalim always false */
A
Andy Adamson 已提交
8094
		task->tk_status = 0;
8095 8096 8097
	case 0:
		break;
	default:
8098
		if (nfs4_async_handle_error(task, server, NULL, NULL) == -EAGAIN) {
8099 8100 8101 8102
			rpc_restart_call_prepare(task);
			return;
		}
	}
A
Andy Adamson 已提交
8103 8104 8105 8106 8107 8108
}

static void nfs4_layoutcommit_release(void *calldata)
{
	struct nfs4_layoutcommit_data *data = calldata;

A
Andy Adamson 已提交
8109
	pnfs_cleanup_layoutcommit(data);
8110 8111
	nfs_post_op_update_inode_force_wcc(data->args.inode,
					   data->res.fattr);
A
Andy Adamson 已提交
8112
	put_rpccred(data->cred);
8113
	nfs_iput_and_deactive(data->inode);
A
Andy Adamson 已提交
8114 8115 8116 8117 8118 8119 8120 8121 8122 8123
	kfree(data);
}

static const struct rpc_call_ops nfs4_layoutcommit_ops = {
	.rpc_call_prepare = nfs4_layoutcommit_prepare,
	.rpc_call_done = nfs4_layoutcommit_done,
	.rpc_release = nfs4_layoutcommit_release,
};

int
8124
nfs4_proc_layoutcommit(struct nfs4_layoutcommit_data *data, bool sync)
A
Andy Adamson 已提交
8125 8126 8127 8128 8129 8130 8131 8132 8133 8134 8135 8136 8137 8138 8139 8140 8141 8142 8143 8144 8145 8146 8147
{
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LAYOUTCOMMIT],
		.rpc_argp = &data->args,
		.rpc_resp = &data->res,
		.rpc_cred = data->cred,
	};
	struct rpc_task_setup task_setup_data = {
		.task = &data->task,
		.rpc_client = NFS_CLIENT(data->args.inode),
		.rpc_message = &msg,
		.callback_ops = &nfs4_layoutcommit_ops,
		.callback_data = data,
	};
	struct rpc_task *task;
	int status = 0;

	dprintk("NFS: %4d initiating layoutcommit call. sync %d "
		"lbw: %llu inode %lu\n",
		data->task.tk_pid, sync,
		data->args.lastbytewritten,
		data->args.inode->i_ino);

8148 8149 8150 8151 8152 8153 8154 8155
	if (!sync) {
		data->inode = nfs_igrab_and_active(data->args.inode);
		if (data->inode == NULL) {
			nfs4_layoutcommit_release(data);
			return -EAGAIN;
		}
		task_setup_data.flags = RPC_TASK_ASYNC;
	}
8156
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
A
Andy Adamson 已提交
8157 8158 8159
	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task))
		return PTR_ERR(task);
8160 8161
	if (sync)
		status = task->tk_status;
8162
	trace_nfs4_layoutcommit(data->args.inode, status);
A
Andy Adamson 已提交
8163 8164 8165 8166
	dprintk("%s: status %d\n", __func__, status);
	rpc_put_task(task);
	return status;
}
8167

8168 8169 8170 8171
/**
 * Use the state managment nfs_client cl_rpcclient, which uses krb5i (if
 * possible) as per RFC3530bis and RFC5661 Security Considerations sections
 */
8172 8173
static int
_nfs41_proc_secinfo_no_name(struct nfs_server *server, struct nfs_fh *fhandle,
8174 8175
		    struct nfs_fsinfo *info,
		    struct nfs4_secinfo_flavors *flavors, bool use_integrity)
8176 8177 8178 8179 8180 8181 8182 8183 8184 8185 8186 8187
{
	struct nfs41_secinfo_no_name_args args = {
		.style = SECINFO_STYLE_CURRENT_FH,
	};
	struct nfs4_secinfo_res res = {
		.flavors = flavors,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SECINFO_NO_NAME],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
8188
	struct rpc_clnt *clnt = server->client;
8189
	struct rpc_cred *cred = NULL;
8190 8191 8192 8193
	int status;

	if (use_integrity) {
		clnt = server->nfs_client->cl_rpcclient;
8194 8195
		cred = nfs4_get_clid_cred(server->nfs_client);
		msg.rpc_cred = cred;
8196 8197 8198 8199 8200 8201 8202
	}

	dprintk("--> %s\n", __func__);
	status = nfs4_call_sync(clnt, server, &msg, &args.seq_args,
				&res.seq_res, 0);
	dprintk("<-- %s status=%d\n", __func__, status);

8203 8204
	if (cred)
		put_rpccred(cred);
8205 8206

	return status;
8207 8208 8209 8210 8211 8212 8213 8214 8215
}

static int
nfs41_proc_secinfo_no_name(struct nfs_server *server, struct nfs_fh *fhandle,
			   struct nfs_fsinfo *info, struct nfs4_secinfo_flavors *flavors)
{
	struct nfs4_exception exception = { };
	int err;
	do {
8216 8217 8218 8219 8220 8221 8222 8223 8224 8225 8226 8227 8228 8229 8230 8231 8232 8233
		/* first try using integrity protection */
		err = -NFS4ERR_WRONGSEC;

		/* try to use integrity protection with machine cred */
		if (_nfs4_is_integrity_protected(server->nfs_client))
			err = _nfs41_proc_secinfo_no_name(server, fhandle, info,
							  flavors, true);

		/*
		 * if unable to use integrity protection, or SECINFO with
		 * integrity protection returns NFS4ERR_WRONGSEC (which is
		 * disallowed by spec, but exists in deployed servers) use
		 * the current filesystem's rpc_client and the user cred.
		 */
		if (err == -NFS4ERR_WRONGSEC)
			err = _nfs41_proc_secinfo_no_name(server, fhandle, info,
							  flavors, false);

8234 8235 8236
		switch (err) {
		case 0:
		case -NFS4ERR_WRONGSEC:
8237
		case -ENOTSUPP:
8238
			goto out;
8239 8240 8241 8242
		default:
			err = nfs4_handle_exception(server, err, &exception);
		}
	} while (exception.retry);
8243
out:
8244 8245 8246 8247 8248 8249 8250 8251 8252
	return err;
}

static int
nfs41_find_root_sec(struct nfs_server *server, struct nfs_fh *fhandle,
		    struct nfs_fsinfo *info)
{
	int err;
	struct page *page;
8253
	rpc_authflavor_t flavor = RPC_AUTH_MAXFLAVOR;
8254
	struct nfs4_secinfo_flavors *flavors;
8255 8256
	struct nfs4_secinfo4 *secinfo;
	int i;
8257 8258 8259 8260 8261 8262 8263 8264 8265 8266 8267 8268 8269 8270

	page = alloc_page(GFP_KERNEL);
	if (!page) {
		err = -ENOMEM;
		goto out;
	}

	flavors = page_address(page);
	err = nfs41_proc_secinfo_no_name(server, fhandle, info, flavors);

	/*
	 * Fall back on "guess and check" method if
	 * the server doesn't support SECINFO_NO_NAME
	 */
8271
	if (err == -NFS4ERR_WRONGSEC || err == -ENOTSUPP) {
8272 8273 8274 8275 8276 8277
		err = nfs4_find_root_sec(server, fhandle, info);
		goto out_freepage;
	}
	if (err)
		goto out_freepage;

8278 8279 8280 8281 8282 8283 8284 8285 8286 8287 8288 8289 8290 8291 8292
	for (i = 0; i < flavors->num_flavors; i++) {
		secinfo = &flavors->flavors[i];

		switch (secinfo->flavor) {
		case RPC_AUTH_NULL:
		case RPC_AUTH_UNIX:
		case RPC_AUTH_GSS:
			flavor = rpcauth_get_pseudoflavor(secinfo->flavor,
					&secinfo->flavor_info);
			break;
		default:
			flavor = RPC_AUTH_MAXFLAVOR;
			break;
		}

8293 8294 8295
		if (!nfs_auth_info_match(&server->auth_info, flavor))
			flavor = RPC_AUTH_MAXFLAVOR;

8296 8297 8298 8299 8300 8301 8302 8303 8304 8305
		if (flavor != RPC_AUTH_MAXFLAVOR) {
			err = nfs4_lookup_root_sec(server, fhandle,
						   info, flavor);
			if (!err)
				break;
		}
	}

	if (flavor == RPC_AUTH_MAXFLAVOR)
		err = -EPERM;
8306 8307 8308 8309 8310 8311 8312 8313

out_freepage:
	put_page(page);
	if (err == -EACCES)
		return -EPERM;
out:
	return err;
}
8314

8315 8316 8317
static int _nfs41_test_stateid(struct nfs_server *server,
		nfs4_stateid *stateid,
		struct rpc_cred *cred)
B
Bryan Schumaker 已提交
8318 8319 8320
{
	int status;
	struct nfs41_test_stateid_args args = {
8321
		.stateid = stateid,
B
Bryan Schumaker 已提交
8322 8323 8324 8325 8326 8327
	};
	struct nfs41_test_stateid_res res;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_TEST_STATEID],
		.rpc_argp = &args,
		.rpc_resp = &res,
8328
		.rpc_cred = cred,
B
Bryan Schumaker 已提交
8329
	};
8330 8331 8332 8333
	struct rpc_clnt *rpc_client = server->client;

	nfs4_state_protect(server->nfs_client, NFS_SP4_MACH_CRED_STATEID,
		&rpc_client, &msg);
8334

8335
	dprintk("NFS call  test_stateid %p\n", stateid);
8336
	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0);
8337
	nfs4_set_sequence_privileged(&args.seq_args);
8338
	status = nfs4_call_sync_sequence(rpc_client, server, &msg,
8339
			&args.seq_args, &res.seq_res);
8340 8341
	if (status != NFS_OK) {
		dprintk("NFS reply test_stateid: failed, %d\n", status);
8342
		return status;
8343 8344
	}
	dprintk("NFS reply test_stateid: succeeded, %d\n", -res.status);
8345
	return -res.status;
B
Bryan Schumaker 已提交
8346 8347
}

8348 8349 8350 8351 8352
/**
 * nfs41_test_stateid - perform a TEST_STATEID operation
 *
 * @server: server / transport on which to perform the operation
 * @stateid: state ID to test
8353
 * @cred: credential
8354 8355 8356 8357 8358
 *
 * Returns NFS_OK if the server recognizes that "stateid" is valid.
 * Otherwise a negative NFS4ERR value is returned if the operation
 * failed or the state ID is not currently valid.
 */
8359 8360 8361
static int nfs41_test_stateid(struct nfs_server *server,
		nfs4_stateid *stateid,
		struct rpc_cred *cred)
B
Bryan Schumaker 已提交
8362 8363 8364 8365
{
	struct nfs4_exception exception = { };
	int err;
	do {
8366
		err = _nfs41_test_stateid(server, stateid, cred);
8367 8368 8369
		if (err != -NFS4ERR_DELAY)
			break;
		nfs4_handle_exception(server, err, &exception);
B
Bryan Schumaker 已提交
8370 8371 8372
	} while (exception.retry);
	return err;
}
B
Bryan Schumaker 已提交
8373

8374 8375 8376
struct nfs_free_stateid_data {
	struct nfs_server *server;
	struct nfs41_free_stateid_args args;
B
Bryan Schumaker 已提交
8377
	struct nfs41_free_stateid_res res;
8378 8379 8380 8381 8382 8383 8384 8385 8386 8387 8388 8389 8390 8391 8392 8393 8394 8395 8396
};

static void nfs41_free_stateid_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs_free_stateid_data *data = calldata;
	nfs41_setup_sequence(nfs4_get_session(data->server),
			&data->args.seq_args,
			&data->res.seq_res,
			task);
}

static void nfs41_free_stateid_done(struct rpc_task *task, void *calldata)
{
	struct nfs_free_stateid_data *data = calldata;

	nfs41_sequence_done(task, &data->res.seq_res);

	switch (task->tk_status) {
	case -NFS4ERR_DELAY:
8397
		if (nfs4_async_handle_error(task, data->server, NULL, NULL) == -EAGAIN)
8398 8399 8400 8401 8402 8403 8404 8405 8406
			rpc_restart_call_prepare(task);
	}
}

static void nfs41_free_stateid_release(void *calldata)
{
	kfree(calldata);
}

8407
static const struct rpc_call_ops nfs41_free_stateid_ops = {
8408 8409 8410 8411 8412 8413 8414
	.rpc_call_prepare = nfs41_free_stateid_prepare,
	.rpc_call_done = nfs41_free_stateid_done,
	.rpc_release = nfs41_free_stateid_release,
};

static struct rpc_task *_nfs41_free_stateid(struct nfs_server *server,
		nfs4_stateid *stateid,
8415
		struct rpc_cred *cred,
8416 8417
		bool privileged)
{
B
Bryan Schumaker 已提交
8418 8419
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FREE_STATEID],
8420
		.rpc_cred = cred,
B
Bryan Schumaker 已提交
8421
	};
8422 8423 8424 8425 8426 8427 8428
	struct rpc_task_setup task_setup = {
		.rpc_client = server->client,
		.rpc_message = &msg,
		.callback_ops = &nfs41_free_stateid_ops,
		.flags = RPC_TASK_ASYNC,
	};
	struct nfs_free_stateid_data *data;
B
Bryan Schumaker 已提交
8429

8430 8431 8432
	nfs4_state_protect(server->nfs_client, NFS_SP4_MACH_CRED_STATEID,
		&task_setup.rpc_client, &msg);

8433
	dprintk("NFS call  free_stateid %p\n", stateid);
8434 8435 8436 8437 8438 8439 8440 8441 8442 8443
	data = kmalloc(sizeof(*data), GFP_NOFS);
	if (!data)
		return ERR_PTR(-ENOMEM);
	data->server = server;
	nfs4_stateid_copy(&data->args.stateid, stateid);

	task_setup.callback_data = data;

	msg.rpc_argp = &data->args;
	msg.rpc_resp = &data->res;
8444
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 0);
8445 8446 8447 8448
	if (privileged)
		nfs4_set_sequence_privileged(&data->args.seq_args);

	return rpc_run_task(&task_setup);
B
Bryan Schumaker 已提交
8449 8450
}

8451 8452 8453 8454 8455
/**
 * nfs41_free_stateid - perform a FREE_STATEID operation
 *
 * @server: server / transport on which to perform the operation
 * @stateid: state ID to release
8456
 * @cred: credential
8457 8458 8459 8460
 *
 * Returns NFS_OK if the server freed "stateid".  Otherwise a
 * negative NFS4ERR value is returned.
 */
8461 8462 8463
static int nfs41_free_stateid(struct nfs_server *server,
		nfs4_stateid *stateid,
		struct rpc_cred *cred)
B
Bryan Schumaker 已提交
8464
{
8465 8466 8467
	struct rpc_task *task;
	int ret;

8468
	task = _nfs41_free_stateid(server, stateid, cred, true);
8469 8470 8471 8472 8473 8474 8475
	if (IS_ERR(task))
		return PTR_ERR(task);
	ret = rpc_wait_for_completion_task(task);
	if (!ret)
		ret = task->tk_status;
	rpc_put_task(task);
	return ret;
B
Bryan Schumaker 已提交
8476
}
8477

8478 8479
static void
nfs41_free_lock_state(struct nfs_server *server, struct nfs4_lock_state *lsp)
8480 8481
{
	struct rpc_task *task;
8482
	struct rpc_cred *cred = lsp->ls_state->owner->so_cred;
8483

8484
	task = _nfs41_free_stateid(server, &lsp->ls_stateid, cred, false);
8485 8486
	nfs4_free_lock_state(server, lsp);
	if (IS_ERR(task))
8487
		return;
8488 8489 8490
	rpc_put_task(task);
}

8491 8492 8493
static bool nfs41_match_stateid(const nfs4_stateid *s1,
		const nfs4_stateid *s2)
{
8494
	if (memcmp(s1->other, s2->other, sizeof(s1->other)) != 0)
8495 8496
		return false;

8497
	if (s1->seqid == s2->seqid)
8498
		return true;
8499
	if (s1->seqid == 0 || s2->seqid == 0)
8500 8501 8502 8503 8504
		return true;

	return false;
}

8505 8506
#endif /* CONFIG_NFS_V4_1 */

8507 8508 8509
static bool nfs4_match_stateid(const nfs4_stateid *s1,
		const nfs4_stateid *s2)
{
8510
	return nfs4_stateid_match(s1, s2);
8511 8512 8513
}


8514
static const struct nfs4_state_recovery_ops nfs40_reboot_recovery_ops = {
8515
	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
8516
	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
L
Linus Torvalds 已提交
8517 8518
	.recover_open	= nfs4_open_reclaim,
	.recover_lock	= nfs4_lock_reclaim,
8519
	.establish_clid = nfs4_init_clientid,
8520
	.detect_trunking = nfs40_discover_server_trunking,
L
Linus Torvalds 已提交
8521 8522
};

8523
#if defined(CONFIG_NFS_V4_1)
8524
static const struct nfs4_state_recovery_ops nfs41_reboot_recovery_ops = {
8525 8526 8527 8528
	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
	.recover_open	= nfs4_open_reclaim,
	.recover_lock	= nfs4_lock_reclaim,
8529
	.establish_clid = nfs41_init_clientid,
8530
	.reclaim_complete = nfs41_proc_reclaim_complete,
8531
	.detect_trunking = nfs41_discover_server_trunking,
8532 8533 8534
};
#endif /* CONFIG_NFS_V4_1 */

8535
static const struct nfs4_state_recovery_ops nfs40_nograce_recovery_ops = {
8536 8537
	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
8538
	.recover_open	= nfs40_open_expired,
8539 8540 8541 8542 8543
	.recover_lock	= nfs4_lock_expired,
	.establish_clid = nfs4_init_clientid,
};

#if defined(CONFIG_NFS_V4_1)
8544
static const struct nfs4_state_recovery_ops nfs41_nograce_recovery_ops = {
8545
	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
8546
	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
8547 8548
	.recover_open	= nfs41_open_expired,
	.recover_lock	= nfs41_lock_expired,
8549
	.establish_clid = nfs41_init_clientid,
L
Linus Torvalds 已提交
8550
};
8551
#endif /* CONFIG_NFS_V4_1 */
L
Linus Torvalds 已提交
8552

8553
static const struct nfs4_state_maintenance_ops nfs40_state_renewal_ops = {
B
Benny Halevy 已提交
8554
	.sched_state_renewal = nfs4_proc_async_renew,
8555
	.get_state_renewal_cred_locked = nfs4_get_renew_cred_locked,
8556
	.renew_lease = nfs4_proc_renew,
B
Benny Halevy 已提交
8557 8558 8559
};

#if defined(CONFIG_NFS_V4_1)
8560
static const struct nfs4_state_maintenance_ops nfs41_state_renewal_ops = {
B
Benny Halevy 已提交
8561
	.sched_state_renewal = nfs41_proc_async_sequence,
8562
	.get_state_renewal_cred_locked = nfs4_get_machine_cred_locked,
8563
	.renew_lease = nfs4_proc_sequence,
B
Benny Halevy 已提交
8564 8565 8566
};
#endif

8567
static const struct nfs4_mig_recovery_ops nfs40_mig_recovery_ops = {
8568
	.get_locations = _nfs40_proc_get_locations,
8569
	.fsid_present = _nfs40_proc_fsid_present,
8570 8571 8572 8573
};

#if defined(CONFIG_NFS_V4_1)
static const struct nfs4_mig_recovery_ops nfs41_mig_recovery_ops = {
8574
	.get_locations = _nfs41_proc_get_locations,
8575
	.fsid_present = _nfs41_proc_fsid_present,
8576 8577 8578
};
#endif	/* CONFIG_NFS_V4_1 */

8579 8580
static const struct nfs4_minor_version_ops nfs_v4_0_minor_ops = {
	.minor_version = 0,
8581 8582 8583 8584
	.init_caps = NFS_CAP_READDIRPLUS
		| NFS_CAP_ATOMIC_OPEN
		| NFS_CAP_CHANGE_ATTR
		| NFS_CAP_POSIX_LOCK,
8585 8586
	.init_client = nfs40_init_client,
	.shutdown_client = nfs40_shutdown_client,
8587
	.match_stateid = nfs4_match_stateid,
8588
	.find_root_sec = nfs4_find_root_sec,
8589
	.free_lock_state = nfs4_release_lockowner,
8590
	.alloc_seqid = nfs_alloc_seqid,
8591
	.call_sync_ops = &nfs40_call_sync_ops,
8592 8593 8594
	.reboot_recovery_ops = &nfs40_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs40_nograce_recovery_ops,
	.state_renewal_ops = &nfs40_state_renewal_ops,
8595
	.mig_recovery_ops = &nfs40_mig_recovery_ops,
8596 8597 8598
};

#if defined(CONFIG_NFS_V4_1)
8599 8600 8601 8602 8603 8604
static struct nfs_seqid *
nfs_alloc_no_seqid(struct nfs_seqid_counter *arg1, gfp_t arg2)
{
	return NULL;
}

8605 8606
static const struct nfs4_minor_version_ops nfs_v4_1_minor_ops = {
	.minor_version = 1,
8607 8608 8609
	.init_caps = NFS_CAP_READDIRPLUS
		| NFS_CAP_ATOMIC_OPEN
		| NFS_CAP_CHANGE_ATTR
8610
		| NFS_CAP_POSIX_LOCK
8611
		| NFS_CAP_STATEID_NFSV41
A
Anna Schumaker 已提交
8612
		| NFS_CAP_ATOMIC_OPEN_V1,
8613 8614
	.init_client = nfs41_init_client,
	.shutdown_client = nfs41_shutdown_client,
8615
	.match_stateid = nfs41_match_stateid,
8616
	.find_root_sec = nfs41_find_root_sec,
8617
	.free_lock_state = nfs41_free_lock_state,
8618
	.alloc_seqid = nfs_alloc_no_seqid,
8619
	.call_sync_ops = &nfs41_call_sync_ops,
8620 8621 8622
	.reboot_recovery_ops = &nfs41_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs41_nograce_recovery_ops,
	.state_renewal_ops = &nfs41_state_renewal_ops,
8623
	.mig_recovery_ops = &nfs41_mig_recovery_ops,
8624 8625 8626
};
#endif

8627 8628 8629
#if defined(CONFIG_NFS_V4_2)
static const struct nfs4_minor_version_ops nfs_v4_2_minor_ops = {
	.minor_version = 2,
8630 8631 8632 8633 8634
	.init_caps = NFS_CAP_READDIRPLUS
		| NFS_CAP_ATOMIC_OPEN
		| NFS_CAP_CHANGE_ATTR
		| NFS_CAP_POSIX_LOCK
		| NFS_CAP_STATEID_NFSV41
A
Anna Schumaker 已提交
8635
		| NFS_CAP_ATOMIC_OPEN_V1
A
Anna Schumaker 已提交
8636
		| NFS_CAP_ALLOCATE
A
Anna Schumaker 已提交
8637
		| NFS_CAP_DEALLOCATE
A
Anna Schumaker 已提交
8638
		| NFS_CAP_SEEK,
8639 8640
	.init_client = nfs41_init_client,
	.shutdown_client = nfs41_shutdown_client,
8641 8642
	.match_stateid = nfs41_match_stateid,
	.find_root_sec = nfs41_find_root_sec,
8643
	.free_lock_state = nfs41_free_lock_state,
8644
	.call_sync_ops = &nfs41_call_sync_ops,
8645
	.alloc_seqid = nfs_alloc_no_seqid,
8646 8647 8648 8649 8650 8651
	.reboot_recovery_ops = &nfs41_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs41_nograce_recovery_ops,
	.state_renewal_ops = &nfs41_state_renewal_ops,
};
#endif

8652 8653 8654 8655 8656
const struct nfs4_minor_version_ops *nfs_v4_minor_ops[] = {
	[0] = &nfs_v4_0_minor_ops,
#if defined(CONFIG_NFS_V4_1)
	[1] = &nfs_v4_1_minor_ops,
#endif
8657 8658 8659
#if defined(CONFIG_NFS_V4_2)
	[2] = &nfs_v4_2_minor_ops,
#endif
8660 8661
};

8662
static const struct inode_operations nfs4_dir_inode_operations = {
8663 8664 8665 8666 8667 8668 8669 8670 8671 8672 8673 8674 8675 8676 8677 8678 8679 8680 8681
	.create		= nfs_create,
	.lookup		= nfs_lookup,
	.atomic_open	= nfs_atomic_open,
	.link		= nfs_link,
	.unlink		= nfs_unlink,
	.symlink	= nfs_symlink,
	.mkdir		= nfs_mkdir,
	.rmdir		= nfs_rmdir,
	.mknod		= nfs_mknod,
	.rename		= nfs_rename,
	.permission	= nfs_permission,
	.getattr	= nfs_getattr,
	.setattr	= nfs_setattr,
	.getxattr	= generic_getxattr,
	.setxattr	= generic_setxattr,
	.listxattr	= generic_listxattr,
	.removexattr	= generic_removexattr,
};

8682
static const struct inode_operations nfs4_file_inode_operations = {
8683 8684 8685
	.permission	= nfs_permission,
	.getattr	= nfs_getattr,
	.setattr	= nfs_setattr,
8686 8687 8688 8689
	.getxattr	= generic_getxattr,
	.setxattr	= generic_setxattr,
	.listxattr	= generic_listxattr,
	.removexattr	= generic_removexattr,
8690 8691
};

D
David Howells 已提交
8692
const struct nfs_rpc_ops nfs_v4_clientops = {
L
Linus Torvalds 已提交
8693 8694 8695
	.version	= 4,			/* protocol version */
	.dentry_ops	= &nfs4_dentry_operations,
	.dir_inode_ops	= &nfs4_dir_inode_operations,
8696
	.file_inode_ops	= &nfs4_file_inode_operations,
8697
	.file_ops	= &nfs4_file_operations,
L
Linus Torvalds 已提交
8698
	.getroot	= nfs4_proc_get_root,
B
Bryan Schumaker 已提交
8699
	.submount	= nfs4_submount,
B
Bryan Schumaker 已提交
8700
	.try_mount	= nfs4_try_mount,
L
Linus Torvalds 已提交
8701 8702 8703 8704 8705 8706 8707 8708
	.getattr	= nfs4_proc_getattr,
	.setattr	= nfs4_proc_setattr,
	.lookup		= nfs4_proc_lookup,
	.access		= nfs4_proc_access,
	.readlink	= nfs4_proc_readlink,
	.create		= nfs4_proc_create,
	.remove		= nfs4_proc_remove,
	.unlink_setup	= nfs4_proc_unlink_setup,
8709
	.unlink_rpc_prepare = nfs4_proc_unlink_rpc_prepare,
L
Linus Torvalds 已提交
8710
	.unlink_done	= nfs4_proc_unlink_done,
8711
	.rename_setup	= nfs4_proc_rename_setup,
8712
	.rename_rpc_prepare = nfs4_proc_rename_rpc_prepare,
8713
	.rename_done	= nfs4_proc_rename_done,
L
Linus Torvalds 已提交
8714 8715 8716 8717 8718 8719 8720 8721 8722
	.link		= nfs4_proc_link,
	.symlink	= nfs4_proc_symlink,
	.mkdir		= nfs4_proc_mkdir,
	.rmdir		= nfs4_proc_remove,
	.readdir	= nfs4_proc_readdir,
	.mknod		= nfs4_proc_mknod,
	.statfs		= nfs4_proc_statfs,
	.fsinfo		= nfs4_proc_fsinfo,
	.pathconf	= nfs4_proc_pathconf,
8723
	.set_capabilities = nfs4_server_capabilities,
L
Linus Torvalds 已提交
8724
	.decode_dirent	= nfs4_decode_dirent,
8725
	.pgio_rpc_prepare = nfs4_proc_pgio_rpc_prepare,
L
Linus Torvalds 已提交
8726
	.read_setup	= nfs4_proc_read_setup,
T
Trond Myklebust 已提交
8727
	.read_done	= nfs4_read_done,
L
Linus Torvalds 已提交
8728
	.write_setup	= nfs4_proc_write_setup,
8729
	.write_done	= nfs4_write_done,
L
Linus Torvalds 已提交
8730
	.commit_setup	= nfs4_proc_commit_setup,
8731
	.commit_rpc_prepare = nfs4_proc_commit_rpc_prepare,
8732
	.commit_done	= nfs4_commit_done,
L
Linus Torvalds 已提交
8733
	.lock		= nfs4_proc_lock,
8734
	.clear_acl_cache = nfs4_zap_acl_attr,
T
Trond Myklebust 已提交
8735
	.close_context  = nfs4_close_context,
8736
	.open_context	= nfs4_atomic_open,
8737
	.have_delegation = nfs4_have_delegation,
8738
	.return_delegation = nfs4_inode_return_delegation,
8739
	.alloc_client	= nfs4_alloc_client,
8740
	.init_client	= nfs4_init_client,
8741
	.free_client	= nfs4_free_client,
8742 8743
	.create_server	= nfs4_create_server,
	.clone_server	= nfs_clone_server,
L
Linus Torvalds 已提交
8744 8745
};

8746 8747 8748 8749 8750 8751 8752 8753 8754
static const struct xattr_handler nfs4_xattr_nfs4_acl_handler = {
	.prefix	= XATTR_NAME_NFSV4_ACL,
	.list	= nfs4_xattr_list_nfs4_acl,
	.get	= nfs4_xattr_get_nfs4_acl,
	.set	= nfs4_xattr_set_nfs4_acl,
};

const struct xattr_handler *nfs4_xattr_handlers[] = {
	&nfs4_xattr_nfs4_acl_handler,
8755 8756 8757
#ifdef CONFIG_NFS_V4_SECURITY_LABEL
	&nfs4_xattr_nfs4_label_handler,
#endif
8758 8759 8760
	NULL
};

L
Linus Torvalds 已提交
8761 8762 8763 8764 8765
/*
 * Local variables:
 *  c-basic-offset: 8
 * End:
 */