nfs4proc.c 235.6 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
static void renew_lease(const struct nfs_server *server, unsigned long timestamp)
{
470 471 472 473
	struct nfs_client *clp = server->nfs_client;

	if (!nfs4_has_session(clp))
		do_renew_lease(clp, timestamp);
474 475
}

476 477 478 479 480 481
struct nfs4_call_sync_data {
	const struct nfs_server *seq_server;
	struct nfs4_sequence_args *seq_args;
	struct nfs4_sequence_res *seq_res;
};

482 483
void nfs4_init_sequence(struct nfs4_sequence_args *args,
			struct nfs4_sequence_res *res, int cache_reply)
484 485 486 487 488 489 490 491 492 493 494 495 496
{
	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;
}

497 498 499 500
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
501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535
{
	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;
}
536
EXPORT_SYMBOL_GPL(nfs40_setup_sequence);
C
Chuck Lever 已提交
537 538 539 540 541 542 543

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;

544
	if (slot == NULL)
C
Chuck Lever 已提交
545 546 547 548 549 550 551 552 553 554 555 556 557
		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
558 559
#if defined(CONFIG_NFS_V4_1)

560
static void nfs41_sequence_free_slot(struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
561
{
562
	struct nfs4_session *session;
A
Andy Adamson 已提交
563
	struct nfs4_slot_table *tbl;
564
	struct nfs4_slot *slot = res->sr_slot;
565
	bool send_new_highest_used_slotid = false;
A
Andy Adamson 已提交
566

567
	tbl = slot->table;
568
	session = tbl->session;
569

570
	spin_lock(&tbl->slot_tbl_lock);
571 572 573 574 575 576
	/* 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;

577
	if (nfs41_wake_and_assign_slot(tbl, slot)) {
578 579 580
		send_new_highest_used_slotid = false;
		goto out_unlock;
	}
581
	nfs4_free_slot(tbl, slot);
582 583 584

	if (tbl->highest_used_slotid != NFS4_NO_SLOT)
		send_new_highest_used_slotid = false;
585
out_unlock:
586
	spin_unlock(&tbl->slot_tbl_lock);
587
	res->sr_slot = NULL;
588
	if (send_new_highest_used_slotid)
589
		nfs41_notify_server(session->clp);
A
Andy Adamson 已提交
590 591
}

592
int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
593
{
594
	struct nfs4_session *session;
595
	struct nfs4_slot *slot = res->sr_slot;
596
	struct nfs_client *clp;
597
	bool interrupted = false;
598
	int ret = 1;
A
Andy Adamson 已提交
599

600 601
	if (slot == NULL)
		goto out_noaction;
602 603
	/* don't increment the sequence number if the task wasn't sent */
	if (!RPC_WAS_SENT(task))
A
Andy Adamson 已提交
604 605
		goto out;

606
	session = slot->table->session;
607

608 609 610 611 612
	if (slot->interrupted) {
		slot->interrupted = 0;
		interrupted = true;
	}

613
	trace_nfs4_sequence_done(session, res);
A
Andy Adamson 已提交
614
	/* Check the SEQUENCE operation status */
615 616
	switch (res->sr_status) {
	case 0:
A
Andy Adamson 已提交
617
		/* Update the slot's sequence and clientid lease timer */
618
		++slot->seq_nr;
619
		clp = session->clp;
620
		do_renew_lease(clp, res->sr_timestamp);
621
		/* Check sequence flags */
622
		nfs41_handle_sequence_flag_errors(clp, res->sr_status_flags);
623
		nfs41_update_target_slotid(slot->table, slot, res);
624
		break;
625 626 627 628 629 630 631 632 633
	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;
634 635 636 637 638
	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.
		 */
639
		dprintk("%s: slot=%u seq=%u: Operation in progress\n",
640
			__func__,
641
			slot->slot_nr,
642
			slot->seq_nr);
643
		goto out_retry;
644 645 646 647 648
	case -NFS4ERR_BADSLOT:
		/*
		 * The slot id we used was probably retired. Try again
		 * using a different slot id.
		 */
649 650
		goto retry_nowait;
	case -NFS4ERR_SEQ_MISORDERED:
651 652 653 654 655 656 657 658
		/*
		 * Was the last operation on this sequence interrupted?
		 * If so, retry after bumping the sequence number.
		 */
		if (interrupted) {
			++slot->seq_nr;
			goto retry_nowait;
		}
659 660 661 662
		/*
		 * Could this slot have been previously retired?
		 * If so, then the server may be expecting seq_nr = 1!
		 */
663 664 665 666 667
		if (slot->seq_nr != 1) {
			slot->seq_nr = 1;
			goto retry_nowait;
		}
		break;
668 669 670
	case -NFS4ERR_SEQ_FALSE_RETRY:
		++slot->seq_nr;
		goto retry_nowait;
671 672
	default:
		/* Just update the slot sequence no. */
673
		++slot->seq_nr;
A
Andy Adamson 已提交
674 675 676 677
	}
out:
	/* The session may be reset by one of the error handlers. */
	dprintk("%s: Error %d free the slot \n", __func__, res->sr_status);
678
	nfs41_sequence_free_slot(res);
679
out_noaction:
680
	return ret;
681 682 683 684 685 686
retry_nowait:
	if (rpc_restart_call_prepare(task)) {
		task->tk_status = 0;
		ret = 0;
	}
	goto out;
687
out_retry:
688
	if (!rpc_restart_call(task))
689 690 691
		goto out;
	rpc_delay(task, NFS4_POLL_RETRY_MAX);
	return 0;
A
Andy Adamson 已提交
692
}
693
EXPORT_SYMBOL_GPL(nfs41_sequence_done);
A
Andy Adamson 已提交
694

P
Peng Tao 已提交
695
int nfs4_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res)
696
{
697
	if (res->sr_slot == NULL)
698
		return 1;
C
Chuck Lever 已提交
699 700
	if (!res->sr_slot->table->session)
		return nfs40_sequence_done(task, res);
701
	return nfs41_sequence_done(task, res);
702
}
P
Peng Tao 已提交
703
EXPORT_SYMBOL_GPL(nfs4_sequence_done);
704

A
Andy Adamson 已提交
705
int nfs41_setup_sequence(struct nfs4_session *session,
A
Andy Adamson 已提交
706 707 708 709
				struct nfs4_sequence_args *args,
				struct nfs4_sequence_res *res,
				struct rpc_task *task)
{
A
Andy Adamson 已提交
710 711 712 713
	struct nfs4_slot *slot;
	struct nfs4_slot_table *tbl;

	dprintk("--> %s\n", __func__);
A
Andy Adamson 已提交
714
	/* slot already allocated? */
715
	if (res->sr_slot != NULL)
716
		goto out_success;
A
Andy Adamson 已提交
717

A
Andy Adamson 已提交
718 719
	tbl = &session->fc_slot_table;

720 721
	task->tk_timeout = 0;

A
Andy Adamson 已提交
722
	spin_lock(&tbl->slot_tbl_lock);
723
	if (test_bit(NFS4_SLOT_TBL_DRAINING, &tbl->slot_tbl_state) &&
724
	    !args->sa_privileged) {
725 726
		/* The state manager will wait until the slot table is empty */
		dprintk("%s session is draining\n", __func__);
727
		goto out_sleep;
728 729
	}

730
	slot = nfs4_alloc_slot(tbl);
731 732 733 734
	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
735
		dprintk("<-- %s: no free slots\n", __func__);
736
		goto out_sleep;
A
Andy Adamson 已提交
737 738 739
	}
	spin_unlock(&tbl->slot_tbl_lock);

740
	args->sa_slot = slot;
A
Andy Adamson 已提交
741

742
	dprintk("<-- %s slotid=%u seqid=%u\n", __func__,
743
			slot->slot_nr, slot->seq_nr);
A
Andy Adamson 已提交
744

745
	res->sr_slot = slot;
746
	res->sr_timestamp = jiffies;
747
	res->sr_status_flags = 0;
A
Andy Adamson 已提交
748 749 750 751 752
	/*
	 * 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;
753
	trace_nfs4_setup_sequence(session, args);
754 755
out_success:
	rpc_call_start(task);
A
Andy Adamson 已提交
756
	return 0;
757
out_sleep:
758 759
	/* Privileged tasks are queued with top priority */
	if (args->sa_privileged)
760 761 762 763
		rpc_sleep_on_priority(&tbl->slot_tbl_waitq, task,
				NULL, RPC_PRIORITY_PRIVILEGED);
	else
		rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
764 765
	spin_unlock(&tbl->slot_tbl_lock);
	return -EAGAIN;
A
Andy Adamson 已提交
766
}
A
Andy Adamson 已提交
767
EXPORT_SYMBOL_GPL(nfs41_setup_sequence);
A
Andy Adamson 已提交
768

769 770 771 772
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
773
{
774
	struct nfs4_session *session = nfs4_get_session(server);
A
Andy Adamson 已提交
775 776
	int ret = 0;

C
Chuck Lever 已提交
777
	if (!session)
778 779
		return nfs40_setup_sequence(server->nfs_client->cl_slot_tbl,
					    args, res, task);
780

781
	dprintk("--> %s clp %p session %p sr_slot %u\n",
782
		__func__, session->clp, session, res->sr_slot ?
783
			res->sr_slot->slot_nr : NFS4_NO_SLOT);
A
Andy Adamson 已提交
784

785
	ret = nfs41_setup_sequence(session, args, res, task);
C
Chuck Lever 已提交
786

A
Andy Adamson 已提交
787 788 789 790 791 792
	dprintk("<-- %s status=%d\n", __func__, ret);
	return ret;
}

static void nfs41_call_sync_prepare(struct rpc_task *task, void *calldata)
{
793
	struct nfs4_call_sync_data *data = calldata;
794
	struct nfs4_session *session = nfs4_get_session(data->seq_server);
A
Andy Adamson 已提交
795

796 797
	dprintk("--> %s data->seq_server %p\n", __func__, data->seq_server);

798
	nfs41_setup_sequence(session, data->seq_args, data->seq_res, task);
A
Andy Adamson 已提交
799 800
}

A
Andy Adamson 已提交
801 802
static void nfs41_call_sync_done(struct rpc_task *task, void *calldata)
{
803
	struct nfs4_call_sync_data *data = calldata;
A
Andy Adamson 已提交
804

805
	nfs41_sequence_done(task, data->seq_res);
A
Andy Adamson 已提交
806 807
}

808
static const struct rpc_call_ops nfs41_call_sync_ops = {
A
Andy Adamson 已提交
809
	.rpc_call_prepare = nfs41_call_sync_prepare,
A
Andy Adamson 已提交
810
	.rpc_call_done = nfs41_call_sync_done,
A
Andy Adamson 已提交
811 812
};

C
Chuck Lever 已提交
813 814
#else	/* !CONFIG_NFS_V4_1 */

815 816 817 818 819
static int nfs4_setup_sequence(const struct nfs_server *server,
			       struct nfs4_sequence_args *args,
			       struct nfs4_sequence_res *res,
			       struct rpc_task *task)
{
820 821
	return nfs40_setup_sequence(server->nfs_client->cl_slot_tbl,
				    args, res, task);
822 823
}

P
Peng Tao 已提交
824 825
int nfs4_sequence_done(struct rpc_task *task,
		       struct nfs4_sequence_res *res)
826
{
C
Chuck Lever 已提交
827
	return nfs40_sequence_done(task, res);
828
}
P
Peng Tao 已提交
829
EXPORT_SYMBOL_GPL(nfs4_sequence_done);
C
Chuck Lever 已提交
830 831

#endif	/* !CONFIG_NFS_V4_1 */
832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850

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,
};

851 852
static int nfs4_call_sync_sequence(struct rpc_clnt *clnt,
				   struct nfs_server *server,
A
Andy Adamson 已提交
853 854
				   struct rpc_message *msg,
				   struct nfs4_sequence_args *args,
855
				   struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
856 857 858
{
	int ret;
	struct rpc_task *task;
859
	struct nfs_client *clp = server->nfs_client;
860
	struct nfs4_call_sync_data data = {
861
		.seq_server = server,
A
Andy Adamson 已提交
862 863 864 865
		.seq_args = args,
		.seq_res = res,
	};
	struct rpc_task_setup task_setup = {
866
		.rpc_client = clnt,
A
Andy Adamson 已提交
867
		.rpc_message = msg,
868
		.callback_ops = clp->cl_mvops->call_sync_ops,
A
Andy Adamson 已提交
869 870 871 872 873 874 875 876 877 878 879 880 881
		.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;
}

882 883
int nfs4_call_sync(struct rpc_clnt *clnt,
		   struct nfs_server *server,
884 885 886 887 888
		   struct rpc_message *msg,
		   struct nfs4_sequence_args *args,
		   struct nfs4_sequence_res *res,
		   int cache_reply)
{
889
	nfs4_init_sequence(args, res, cache_reply);
890
	return nfs4_call_sync_sequence(clnt, server, msg, args, res);
891
}
A
Andy Adamson 已提交
892

893
static void update_changeattr(struct inode *dir, struct nfs4_change_info *cinfo)
L
Linus Torvalds 已提交
894
{
895
	struct nfs_inode *nfsi = NFS_I(dir);
L
Linus Torvalds 已提交
896

897
	spin_lock(&dir->i_lock);
898
	nfsi->cache_validity |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA;
899
	if (!cinfo->atomic || cinfo->before != dir->i_version)
900
		nfs_force_lookup_revalidate(dir);
901
	dir->i_version = cinfo->after;
902
	nfsi->attr_gencount = nfs_inc_attr_generation_counter();
D
David Howells 已提交
903
	nfs_fscache_invalidate(dir);
904
	spin_unlock(&dir->i_lock);
L
Linus Torvalds 已提交
905 906
}

907
struct nfs4_opendata {
908
	struct kref kref;
909 910
	struct nfs_openargs o_arg;
	struct nfs_openres o_res;
911 912
	struct nfs_open_confirmargs c_arg;
	struct nfs_open_confirmres c_res;
913 914
	struct nfs4_string owner_name;
	struct nfs4_string group_name;
915
	struct nfs4_label *a_label;
916
	struct nfs_fattr f_attr;
917
	struct nfs4_label *f_label;
918
	struct dentry *dir;
919
	struct dentry *dentry;
920
	struct nfs4_state_owner *owner;
921
	struct nfs4_state *state;
922
	struct iattr attrs;
923
	unsigned long timestamp;
924
	unsigned int rpc_done : 1;
925
	unsigned int file_created : 1;
926
	unsigned int is_recover : 1;
927 928
	int rpc_status;
	int cancelled;
929 930
};

931 932 933 934 935 936 937 938 939 940 941 942
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;
}

943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967
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;
}

968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984
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;
	}
}
985 986 987 988

static void nfs4_init_opendata_res(struct nfs4_opendata *p)
{
	p->o_res.f_attr = &p->f_attr;
989
	p->o_res.f_label = p->f_label;
990 991
	p->o_res.seqid = p->o_arg.seqid;
	p->c_res.seqid = p->c_arg.seqid;
992
	p->o_res.server = p->o_arg.server;
993
	p->o_res.access_request = p->o_arg.access;
994
	nfs_fattr_init(&p->f_attr);
995
	nfs_fattr_init_names(&p->f_attr, &p->owner_name, &p->group_name);
996 997
}

998
static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry,
999
		struct nfs4_state_owner *sp, fmode_t fmode, int flags,
1000
		const struct iattr *attrs,
1001
		struct nfs4_label *label,
1002
		enum open_claim_type4 claim,
1003
		gfp_t gfp_mask)
1004
{
1005
	struct dentry *parent = dget_parent(dentry);
1006
	struct inode *dir = d_inode(parent);
1007
	struct nfs_server *server = NFS_SERVER(dir);
1008
	struct nfs_seqid *(*alloc_seqid)(struct nfs_seqid_counter *, gfp_t);
1009 1010
	struct nfs4_opendata *p;

1011
	p = kzalloc(sizeof(*p), gfp_mask);
1012 1013
	if (p == NULL)
		goto err;
1014 1015 1016 1017 1018

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

1019 1020 1021 1022
	p->a_label = nfs4_label_alloc(server, gfp_mask);
	if (IS_ERR(p->a_label))
		goto err_free_f;

1023 1024
	alloc_seqid = server->nfs_client->cl_mvops->alloc_seqid;
	p->o_arg.seqid = alloc_seqid(&sp->so_seqid, gfp_mask);
1025
	if (IS_ERR(p->o_arg.seqid))
1026
		goto err_free_label;
1027 1028
	nfs_sb_active(dentry->d_sb);
	p->dentry = dget(dentry);
1029 1030 1031
	p->dir = parent;
	p->owner = sp;
	atomic_inc(&sp->so_count);
1032 1033
	p->o_arg.open_flags = flags;
	p->o_arg.fmode = fmode & (FMODE_READ|FMODE_WRITE);
1034 1035
	p->o_arg.share_access = nfs4_map_atomic_open_share(server,
			fmode, flags);
1036 1037 1038 1039 1040 1041 1042 1043
	/* 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;
	}
1044
	p->o_arg.clientid = server->nfs_client->cl_clientid;
1045 1046
	p->o_arg.id.create_time = ktime_to_ns(sp->so_seqid.create_time);
	p->o_arg.id.uniquifier = sp->so_seqid.owner_id;
1047
	p->o_arg.name = &dentry->d_name;
1048
	p->o_arg.server = server;
1049
	p->o_arg.bitmask = nfs4_bitmask(server, label);
1050
	p->o_arg.open_bitmap = &nfs4_fattr_bitmap[0];
1051
	p->o_arg.label = nfs4_label_copy(p->a_label, label);
1052 1053
	p->o_arg.claim = nfs4_map_atomic_open_claim(server, claim);
	switch (p->o_arg.claim) {
1054 1055 1056 1057 1058 1059 1060 1061 1062
	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:
1063
		p->o_arg.fh = NFS_FH(d_inode(dentry));
1064
	}
1065
	if (attrs != NULL && attrs->ia_valid != 0) {
1066
		__u32 verf[2];
1067

1068 1069
		p->o_arg.u.attrs = &p->attrs;
		memcpy(&p->attrs, attrs, sizeof(p->attrs));
1070 1071 1072 1073 1074

		verf[0] = jiffies;
		verf[1] = current->pid;
		memcpy(p->o_arg.u.verifier.data, verf,
				sizeof(p->o_arg.u.verifier.data));
1075
	}
1076 1077 1078
	p->c_arg.fh = &p->o_res.fh;
	p->c_arg.stateid = &p->o_res.stateid;
	p->c_arg.seqid = p->o_arg.seqid;
1079
	nfs4_init_opendata_res(p);
1080
	kref_init(&p->kref);
1081
	return p;
1082 1083

err_free_label:
1084 1085
	nfs4_label_free(p->a_label);
err_free_f:
1086 1087
	nfs4_label_free(p->f_label);
err_free_p:
1088 1089 1090 1091 1092 1093
	kfree(p);
err:
	dput(parent);
	return NULL;
}

1094
static void nfs4_opendata_free(struct kref *kref)
1095
{
1096 1097
	struct nfs4_opendata *p = container_of(kref,
			struct nfs4_opendata, kref);
1098
	struct super_block *sb = p->dentry->d_sb;
1099 1100

	nfs_free_seqid(p->o_arg.seqid);
1101 1102
	if (p->state != NULL)
		nfs4_put_open_state(p->state);
1103
	nfs4_put_state_owner(p->owner);
1104

1105
	nfs4_label_free(p->a_label);
1106 1107
	nfs4_label_free(p->f_label);

1108
	dput(p->dir);
1109 1110
	dput(p->dentry);
	nfs_sb_deactive(sb);
1111
	nfs_fattr_free_names(&p->f_attr);
1112
	kfree(p->f_attr.mdsthreshold);
1113 1114 1115 1116 1117 1118 1119
	kfree(p);
}

static void nfs4_opendata_put(struct nfs4_opendata *p)
{
	if (p != NULL)
		kref_put(&p->kref, nfs4_opendata_free);
1120 1121
}

1122 1123 1124 1125 1126 1127 1128 1129
static int nfs4_wait_for_completion_rpc_task(struct rpc_task *task)
{
	int ret;

	ret = rpc_wait_for_completion_task(task);
	return ret;
}

1130
static int can_open_cached(struct nfs4_state *state, fmode_t mode, int open_mode)
1131 1132
{
	int ret = 0;
1133

1134
	if (open_mode & (O_EXCL|O_TRUNC))
1135 1136
		goto out;
	switch (mode & (FMODE_READ|FMODE_WRITE)) {
1137
		case FMODE_READ:
1138 1139
			ret |= test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0
				&& state->n_rdonly != 0;
1140 1141
			break;
		case FMODE_WRITE:
1142 1143
			ret |= test_bit(NFS_O_WRONLY_STATE, &state->flags) != 0
				&& state->n_wronly != 0;
1144 1145
			break;
		case FMODE_READ|FMODE_WRITE:
1146 1147
			ret |= test_bit(NFS_O_RDWR_STATE, &state->flags) != 0
				&& state->n_rdwr != 0;
1148
	}
1149
out:
1150 1151 1152
	return ret;
}

1153
static int can_open_delegated(struct nfs_delegation *delegation, fmode_t fmode)
1154
{
1155 1156
	if (delegation == NULL)
		return 0;
1157
	if ((delegation->type & fmode) != fmode)
1158
		return 0;
1159 1160
	if (test_bit(NFS_DELEGATION_RETURNING, &delegation->flags))
		return 0;
1161
	nfs_mark_delegation_referenced(delegation);
1162 1163 1164
	return 1;
}

1165
static void update_open_stateflags(struct nfs4_state *state, fmode_t fmode)
1166
{
1167
	switch (fmode) {
1168 1169 1170 1171 1172 1173 1174 1175 1176
		case FMODE_WRITE:
			state->n_wronly++;
			break;
		case FMODE_READ:
			state->n_rdonly++;
			break;
		case FMODE_READ|FMODE_WRITE:
			state->n_rdwr++;
	}
1177
	nfs4_state_set_mode_locked(state, state->state | fmode);
1178 1179
}

1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194
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);
}

1195 1196 1197 1198 1199
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;
1200 1201
	if (!nfs4_stateid_match_other(stateid, &state->open_stateid)) {
		nfs_test_and_clear_all_open_stateid(state);
1202
		return true;
1203
	}
1204 1205 1206 1207 1208
	if (nfs4_stateid_is_newer(stateid, &state->open_stateid))
		return true;
	return false;
}

T
Trond Myklebust 已提交
1209 1210
static void nfs_resync_open_stateid_locked(struct nfs4_state *state)
{
1211 1212
	if (!(state->n_wronly || state->n_rdonly || state->n_rdwr))
		return;
T
Trond Myklebust 已提交
1213 1214 1215 1216 1217 1218
	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);
1219
	set_bit(NFS_OPEN_STATE, &state->flags);
T
Trond Myklebust 已提交
1220 1221
}

1222 1223
static void nfs_clear_open_stateid_locked(struct nfs4_state *state,
		nfs4_stateid *stateid, fmode_t fmode)
1224
{
1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239
	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
1240 1241 1242 1243
	/* 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);
1244
		return;
T
Trond Myklebust 已提交
1245
	}
1246
	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
1247 1248
		nfs4_stateid_copy(&state->stateid, stateid);
	nfs4_stateid_copy(&state->open_stateid, stateid);
1249 1250 1251 1252 1253 1254 1255
}

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);
1256 1257
	if (test_bit(NFS_STATE_RECLAIM_NOGRACE, &state->flags))
		nfs4_schedule_state_manager(state->owner->so_server->nfs_client);
1258 1259
}

1260
static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
1261
{
1262
	switch (fmode) {
1263 1264 1265 1266 1267 1268 1269 1270 1271
		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);
	}
1272 1273 1274 1275 1276
	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);
1277 1278
}

1279
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
1280
{
1281 1282 1283 1284 1285
	/*
	 * Protect the call to nfs4_state_set_mode_locked and
	 * serialise the stateid update
	 */
	write_seqlock(&state->seqlock);
1286
	if (deleg_stateid != NULL) {
1287
		nfs4_stateid_copy(&state->stateid, deleg_stateid);
1288 1289 1290
		set_bit(NFS_DELEGATED_STATE, &state->flags);
	}
	if (open_stateid != NULL)
1291
		nfs_set_open_stateid_locked(state, open_stateid, fmode);
1292 1293
	write_sequnlock(&state->seqlock);
	spin_lock(&state->owner->so_lock);
1294
	update_open_stateflags(state, fmode);
1295
	spin_unlock(&state->owner->so_lock);
L
Linus Torvalds 已提交
1296 1297
}

1298
static int update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *delegation, fmode_t fmode)
1299 1300 1301 1302 1303
{
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *deleg_cur;
	int ret = 0;

1304
	fmode &= (FMODE_READ|FMODE_WRITE);
1305 1306 1307 1308 1309 1310 1311

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

	spin_lock(&deleg_cur->lock);
1312
	if (rcu_dereference(nfsi->delegation) != deleg_cur ||
1313
	   test_bit(NFS_DELEGATION_RETURNING, &deleg_cur->flags) ||
1314
	    (deleg_cur->type & fmode) != fmode)
1315 1316 1317 1318
		goto no_delegation_unlock;

	if (delegation == NULL)
		delegation = &deleg_cur->stateid;
1319
	else if (!nfs4_stateid_match(&deleg_cur->stateid, delegation))
1320 1321
		goto no_delegation_unlock;

1322
	nfs_mark_delegation_referenced(deleg_cur);
1323
	__update_open_stateid(state, open_stateid, &deleg_cur->stateid, fmode);
1324 1325 1326 1327 1328 1329 1330
	ret = 1;
no_delegation_unlock:
	spin_unlock(&deleg_cur->lock);
no_delegation:
	rcu_read_unlock();

	if (!ret && open_stateid != NULL) {
1331
		__update_open_stateid(state, open_stateid, NULL, fmode);
1332 1333
		ret = 1;
	}
1334 1335
	if (test_bit(NFS_STATE_RECLAIM_NOGRACE, &state->flags))
		nfs4_schedule_state_manager(state->owner->so_server->nfs_client);
1336 1337 1338 1339

	return ret;
}

1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356
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;
}
1357

1358
static void nfs4_return_incompatible_delegation(struct inode *inode, fmode_t fmode)
1359 1360 1361 1362 1363
{
	struct nfs_delegation *delegation;

	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(inode)->delegation);
1364
	if (delegation == NULL || (delegation->type & fmode) == fmode) {
1365 1366 1367 1368
		rcu_read_unlock();
		return;
	}
	rcu_read_unlock();
1369
	nfs4_inode_return_delegation(inode);
1370 1371
}

1372
static struct nfs4_state *nfs4_try_open_cached(struct nfs4_opendata *opendata)
1373 1374 1375 1376
{
	struct nfs4_state *state = opendata->state;
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *delegation;
1377
	int open_mode = opendata->o_arg.open_flags;
1378
	fmode_t fmode = opendata->o_arg.fmode;
1379 1380 1381 1382
	nfs4_stateid stateid;
	int ret = -EAGAIN;

	for (;;) {
1383
		spin_lock(&state->owner->so_lock);
1384
		if (can_open_cached(state, fmode, open_mode)) {
1385
			update_open_stateflags(state, fmode);
1386
			spin_unlock(&state->owner->so_lock);
1387
			goto out_return_state;
1388
		}
1389
		spin_unlock(&state->owner->so_lock);
1390 1391
		rcu_read_lock();
		delegation = rcu_dereference(nfsi->delegation);
1392
		if (!can_open_delegated(delegation, fmode)) {
1393
			rcu_read_unlock();
1394
			break;
1395
		}
1396
		/* Save the delegation */
1397
		nfs4_stateid_copy(&stateid, &delegation->stateid);
1398
		rcu_read_unlock();
1399
		nfs_release_seqid(opendata->o_arg.seqid);
1400 1401 1402 1403 1404
		if (!opendata->is_recover) {
			ret = nfs_may_open(state->inode, state->owner->so_cred, open_mode);
			if (ret != 0)
				goto out;
		}
1405
		ret = -EAGAIN;
1406 1407

		/* Try to update the stateid using the delegation */
1408
		if (update_open_stateid(state, NULL, &stateid, fmode))
1409
			goto out_return_state;
1410 1411 1412 1413 1414 1415 1416 1417
	}
out:
	return ERR_PTR(ret);
out_return_state:
	atomic_inc(&state->count);
	return state;
}

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 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455
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;

1456 1457 1458 1459 1460 1461 1462
	if (!data->rpc_done) {
		if (data->rpc_status) {
			ret = data->rpc_status;
			goto err;
		}
		/* cached opens have already been processed */
		goto update;
1463 1464 1465 1466 1467 1468 1469 1470
	}

	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);
1471
update:
1472 1473
	update_open_stateid(state, &data->o_res.stateid, NULL,
			    data->o_arg.fmode);
1474
	atomic_inc(&state->count);
1475 1476 1477 1478 1479 1480 1481 1482 1483

	return state;
err:
	return ERR_PTR(ret);

}

static struct nfs4_state *
_nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data)
1484 1485 1486
{
	struct inode *inode;
	struct nfs4_state *state = NULL;
1487
	int ret;
1488

1489
	if (!data->rpc_done) {
1490
		state = nfs4_try_open_cached(data);
1491 1492 1493
		goto out;
	}

1494
	ret = -EAGAIN;
1495
	if (!(data->f_attr.valid & NFS_ATTR_FATTR))
1496
		goto err;
1497
	inode = nfs_fhget(data->dir->d_sb, &data->o_res.fh, &data->f_attr, data->f_label);
1498
	ret = PTR_ERR(inode);
1499
	if (IS_ERR(inode))
1500 1501
		goto err;
	ret = -ENOMEM;
1502 1503
	state = nfs4_get_open_state(inode, data->owner);
	if (state == NULL)
1504
		goto err_put_inode;
1505 1506
	if (data->o_res.delegation_type != 0)
		nfs4_opendata_check_deleg(data, state);
1507
	update_open_stateid(state, &data->o_res.stateid, NULL,
1508
			data->o_arg.fmode);
1509
	iput(inode);
1510
out:
1511
	nfs_release_seqid(data->o_arg.seqid);
1512
	return state;
1513 1514 1515 1516
err_put_inode:
	iput(inode);
err:
	return ERR_PTR(ret);
1517 1518
}

1519 1520 1521 1522 1523 1524 1525 1526
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);
}

1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543
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);
}

1544 1545
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
1546 1547 1548
{
	struct nfs4_opendata *opendata;

1549
	opendata = nfs4_opendata_alloc(ctx->dentry, state->owner, 0, 0,
1550
			NULL, NULL, claim, GFP_NOFS);
T
Trond Myklebust 已提交
1551 1552 1553 1554 1555 1556 1557
	if (opendata == NULL)
		return ERR_PTR(-ENOMEM);
	opendata->state = state;
	atomic_inc(&state->count);
	return opendata;
}

1558
static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, fmode_t fmode, struct nfs4_state **res)
1559
{
1560
	struct nfs4_state *newstate;
1561 1562
	int ret;

1563 1564 1565 1566 1567 1568 1569
	if ((opendata->o_arg.claim == NFS4_OPEN_CLAIM_DELEGATE_CUR ||
	     opendata->o_arg.claim == NFS4_OPEN_CLAIM_DELEG_CUR_FH) &&
	    (opendata->o_arg.u.delegation_type & fmode) != fmode)
		/* This mode can't have been delegated, so we must have
		 * a valid open_stateid to cover it - not need to reclaim.
		 */
		return 0;
1570 1571
	opendata->o_arg.open_flags = 0;
	opendata->o_arg.fmode = fmode;
1572 1573 1574
	opendata->o_arg.share_access = nfs4_map_atomic_open_share(
			NFS_SB(opendata->dentry->d_sb),
			fmode, 0);
1575 1576 1577
	memset(&opendata->o_res, 0, sizeof(opendata->o_res));
	memset(&opendata->c_res, 0, sizeof(opendata->c_res));
	nfs4_init_opendata_res(opendata);
1578
	ret = _nfs4_recover_proc_open(opendata);
1579 1580
	if (ret != 0)
		return ret; 
1581
	newstate = nfs4_opendata_to_nfs4_state(opendata);
1582 1583
	if (IS_ERR(newstate))
		return PTR_ERR(newstate);
1584
	nfs4_close_state(newstate, fmode);
1585
	*res = newstate;
1586 1587 1588 1589 1590 1591 1592 1593
	return 0;
}

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

1594 1595 1596 1597
	/* 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);
1598
	/* memory barrier prior to reading state->n_* */
1599
	clear_bit(NFS_DELEGATED_STATE, &state->flags);
1600
	clear_bit(NFS_OPEN_STATE, &state->flags);
1601 1602
	smp_rmb();
	if (state->n_rdwr != 0) {
1603
		ret = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE, &newstate);
1604 1605
		if (ret != 0)
			return ret;
1606 1607
		if (newstate != state)
			return -ESTALE;
1608 1609
	}
	if (state->n_wronly != 0) {
1610
		ret = nfs4_open_recover_helper(opendata, FMODE_WRITE, &newstate);
1611 1612
		if (ret != 0)
			return ret;
1613 1614
		if (newstate != state)
			return -ESTALE;
1615 1616
	}
	if (state->n_rdonly != 0) {
1617
		ret = nfs4_open_recover_helper(opendata, FMODE_READ, &newstate);
1618 1619
		if (ret != 0)
			return ret;
1620 1621
		if (newstate != state)
			return -ESTALE;
1622
	}
1623 1624 1625 1626 1627
	/*
	 * 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 &&
1628
	    !nfs4_stateid_match(&state->stateid, &state->open_stateid)) {
1629
		write_seqlock(&state->seqlock);
1630
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
1631
			nfs4_stateid_copy(&state->stateid, &state->open_stateid);
1632
		write_sequnlock(&state->seqlock);
1633
	}
1634 1635 1636
	return 0;
}

L
Linus Torvalds 已提交
1637 1638 1639 1640
/*
 * OPEN_RECLAIM:
 * 	reclaim state on the server after a reboot.
 */
1641
static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1642
{
1643
	struct nfs_delegation *delegation;
1644
	struct nfs4_opendata *opendata;
1645
	fmode_t delegation_type = 0;
L
Linus Torvalds 已提交
1646 1647
	int status;

1648 1649
	opendata = nfs4_open_recoverdata_alloc(ctx, state,
			NFS4_OPEN_CLAIM_PREVIOUS);
T
Trond Myklebust 已提交
1650 1651
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1652 1653
	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
1654
	if (delegation != NULL && test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) != 0)
1655
		delegation_type = delegation->type;
1656
	rcu_read_unlock();
1657 1658
	opendata->o_arg.u.delegation_type = delegation_type;
	status = nfs4_open_recover(opendata, state);
1659
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
1660 1661 1662
	return status;
}

1663
static int nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1664 1665 1666 1667 1668
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;
	do {
1669
		err = _nfs4_do_open_reclaim(ctx, state);
1670
		trace_nfs4_open_reclaim(ctx, 0, err);
1671 1672
		if (nfs4_clear_cap_atomic_open_v1(server, err, &exception))
			continue;
1673
		if (err != -NFS4ERR_DELAY)
1674 1675
			break;
		nfs4_handle_exception(server, err, &exception);
L
Linus Torvalds 已提交
1676 1677 1678 1679
	} while (exception.retry);
	return err;
}

1680 1681 1682 1683 1684 1685 1686
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))
1687
		return -EAGAIN;
1688
	ret = nfs4_do_open_reclaim(ctx, state);
1689 1690 1691 1692
	put_nfs_open_context(ctx);
	return ret;
}

1693
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
1694
{
1695 1696 1697 1698 1699 1700
	switch (err) {
		default:
			printk(KERN_ERR "NFS: %s: unhandled error "
					"%d.\n", __func__, err);
		case 0:
		case -ENOENT:
1701
		case -EAGAIN:
1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718
		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;
1719 1720 1721
		case -NFS4ERR_MOVED:
			nfs4_schedule_migration_recovery(server);
			return -EAGAIN;
1722 1723 1724
		case -NFS4ERR_LEASE_MOVED:
			nfs4_schedule_lease_moved_recovery(server->nfs_client);
			return -EAGAIN;
1725 1726 1727
		case -NFS4ERR_DELEG_REVOKED:
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
1728
		case -NFS4ERR_OPENMODE:
1729 1730 1731
			nfs_inode_find_state_and_recover(state->inode,
					stateid);
			nfs4_schedule_stateid_recovery(server, state);
1732
			return -EAGAIN;
1733 1734 1735 1736 1737
		case -NFS4ERR_DELAY:
		case -NFS4ERR_GRACE:
			set_bit(NFS_DELEGATED_STATE, &state->flags);
			ssleep(1);
			return -EAGAIN;
1738 1739 1740 1741
		case -ENOMEM:
		case -NFS4ERR_DENIED:
			/* kill_proc(fl->fl_pid, SIGLOST, 1); */
			return 0;
1742
	}
L
Linus Torvalds 已提交
1743 1744 1745
	return err;
}

1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761
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);
}

1762 1763 1764 1765
static void nfs4_open_confirm_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;

1766 1767
	nfs40_setup_sequence(data->o_arg.server->nfs_client->cl_slot_tbl,
			     &data->c_arg.seq_args, &data->c_res.seq_res, task);
1768 1769
}

1770 1771 1772 1773
static void nfs4_open_confirm_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;

1774
	nfs40_sequence_done(task, &data->c_res.seq_res);
1775

1776
	data->rpc_status = task->tk_status;
1777
	if (data->rpc_status == 0) {
1778
		nfs4_stateid_copy(&data->o_res.stateid, &data->c_res.stateid);
1779
		nfs_confirm_seqid(&data->owner->so_seqid, 0);
1780
		renew_lease(data->o_res.server, data->timestamp);
1781
		data->rpc_done = 1;
1782
	}
1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793
}

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! */
1794
	if (!data->rpc_done)
1795 1796
		goto out_free;
	state = nfs4_opendata_to_nfs4_state(data);
1797
	if (!IS_ERR(state))
1798
		nfs4_close_state(state, data->o_arg.fmode);
1799
out_free:
1800
	nfs4_opendata_put(data);
1801 1802 1803
}

static const struct rpc_call_ops nfs4_open_confirm_ops = {
1804
	.rpc_call_prepare = nfs4_open_confirm_prepare,
1805 1806 1807 1808 1809 1810 1811 1812 1813
	.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)
{
1814
	struct nfs_server *server = NFS_SERVER(d_inode(data->dir));
1815
	struct rpc_task *task;
1816 1817 1818 1819 1820 1821
	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
1822 1823
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1824
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1825 1826
		.callback_ops = &nfs4_open_confirm_ops,
		.callback_data = data,
1827
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1828 1829
		.flags = RPC_TASK_ASYNC,
	};
L
Linus Torvalds 已提交
1830 1831
	int status;

1832
	nfs4_init_sequence(&data->c_arg.seq_args, &data->c_res.seq_res, 1);
1833
	kref_get(&data->kref);
1834 1835
	data->rpc_done = 0;
	data->rpc_status = 0;
1836
	data->timestamp = jiffies;
T
Trond Myklebust 已提交
1837
	task = rpc_run_task(&task_setup_data);
1838
	if (IS_ERR(task))
1839 1840 1841 1842 1843 1844 1845
		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;
1846
	rpc_put_task(task);
L
Linus Torvalds 已提交
1847 1848 1849
	return status;
}

1850
static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
L
Linus Torvalds 已提交
1851
{
1852 1853
	struct nfs4_opendata *data = calldata;
	struct nfs4_state_owner *sp = data->owner;
1854
	struct nfs_client *clp = sp->so_server->nfs_client;
1855

1856
	if (nfs_wait_on_sequence(data->o_arg.seqid, task) != 0)
1857
		goto out_wait;
1858 1859 1860 1861 1862 1863 1864
	/*
	 * 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;

1865
		if (can_open_cached(data->state, data->o_arg.fmode, data->o_arg.open_flags))
1866
			goto out_no_action;
1867 1868
		rcu_read_lock();
		delegation = rcu_dereference(NFS_I(data->state->inode)->delegation);
1869
		if (data->o_arg.claim != NFS4_OPEN_CLAIM_DELEGATE_CUR &&
1870
		    data->o_arg.claim != NFS4_OPEN_CLAIM_DELEG_CUR_FH &&
1871 1872
		    can_open_delegated(delegation, data->o_arg.fmode))
			goto unlock_no_action;
1873 1874
		rcu_read_unlock();
	}
1875
	/* Update client id. */
1876
	data->o_arg.clientid = clp->cl_clientid;
1877 1878 1879 1880
	switch (data->o_arg.claim) {
	case NFS4_OPEN_CLAIM_PREVIOUS:
	case NFS4_OPEN_CLAIM_DELEG_CUR_FH:
	case NFS4_OPEN_CLAIM_DELEG_PREV_FH:
1881
		data->o_arg.open_bitmap = &nfs4_open_noattr_bitmap[0];
1882 1883
	case NFS4_OPEN_CLAIM_FH:
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR];
T
Trond Myklebust 已提交
1884 1885
		nfs_copy_fh(&data->o_res.fh, data->o_arg.fh);
	}
1886
	data->timestamp = jiffies;
1887
	if (nfs4_setup_sequence(data->o_arg.server,
1888
				&data->o_arg.seq_args,
1889 1890 1891
				&data->o_res.seq_res,
				task) != 0)
		nfs_release_seqid(data->o_arg.seqid);
1892 1893 1894 1895 1896 1897 1898 1899 1900 1901

	/* 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;
	}
1902
	return;
1903 1904
unlock_no_action:
	rcu_read_unlock();
1905 1906
out_no_action:
	task->tk_action = NULL;
1907
out_wait:
1908
	nfs4_sequence_done(task, &data->o_res.seq_res);
1909 1910
}

1911 1912 1913
static void nfs4_open_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;
L
Linus Torvalds 已提交
1914

1915
	data->rpc_status = task->tk_status;
1916

1917 1918
	if (!nfs4_sequence_done(task, &data->o_res.seq_res))
		return;
1919

1920
	if (task->tk_status == 0) {
1921 1922
		if (data->o_res.f_attr->valid & NFS_ATTR_FATTR_TYPE) {
			switch (data->o_res.f_attr->mode & S_IFMT) {
1923 1924 1925
			case S_IFREG:
				break;
			case S_IFLNK:
1926
				data->rpc_status = -ELOOP;
1927 1928
				break;
			case S_IFDIR:
1929
				data->rpc_status = -EISDIR;
1930 1931
				break;
			default:
1932
				data->rpc_status = -ENOTDIR;
1933
			}
1934
		}
1935
		renew_lease(data->o_res.server, data->timestamp);
1936 1937
		if (!(data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM))
			nfs_confirm_seqid(&data->owner->so_seqid, 0);
1938
	}
1939
	data->rpc_done = 1;
1940
}
1941

1942 1943 1944 1945 1946 1947 1948 1949 1950
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! */
1951
	if (data->rpc_status != 0 || !data->rpc_done)
1952 1953 1954 1955 1956
		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);
1957
	if (!IS_ERR(state))
1958
		nfs4_close_state(state, data->o_arg.fmode);
1959
out_free:
1960
	nfs4_opendata_put(data);
1961 1962 1963 1964 1965 1966 1967 1968
}

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,
};

1969
static int nfs4_run_open_task(struct nfs4_opendata *data, int isrecover)
1970
{
1971
	struct inode *dir = d_inode(data->dir);
1972 1973 1974 1975
	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;
1976 1977 1978 1979 1980 1981
	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
1982 1983
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1984
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1985 1986
		.callback_ops = &nfs4_open_ops,
		.callback_data = data,
1987
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1988 1989
		.flags = RPC_TASK_ASYNC,
	};
1990 1991
	int status;

1992
	nfs4_init_sequence(&o_arg->seq_args, &o_res->seq_res, 1);
1993
	kref_get(&data->kref);
1994 1995
	data->rpc_done = 0;
	data->rpc_status = 0;
1996
	data->cancelled = 0;
1997 1998
	data->is_recover = 0;
	if (isrecover) {
1999
		nfs4_set_sequence_privileged(&o_arg->seq_args);
2000 2001
		data->is_recover = 1;
	}
T
Trond Myklebust 已提交
2002
	task = rpc_run_task(&task_setup_data);
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
        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)
{
2018
	struct inode *dir = d_inode(data->dir);
2019 2020 2021 2022 2023 2024 2025
	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;

2026 2027
	nfs_fattr_map_and_free_names(NFS_SERVER(dir), &data->f_attr);

2028 2029 2030 2031 2032 2033 2034 2035 2036
	if (o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
		status = _nfs4_proc_open_confirm(data);
		if (status != 0)
			return status;
	}

	return status;
}

2037 2038 2039 2040 2041 2042 2043 2044
/*
 * 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).
 */
2045 2046
static int nfs4_opendata_access(struct rpc_cred *cred,
				struct nfs4_opendata *opendata,
2047 2048
				struct nfs4_state *state, fmode_t fmode,
				int openflags)
2049 2050 2051 2052 2053 2054 2055 2056 2057 2058
{
	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;
2059 2060 2061 2062
	/*
	 * Use openflags to check for exec, because fmode won't
	 * always have FMODE_EXEC set when file open for exec.
	 */
2063 2064 2065
	if (openflags & __FMODE_EXEC) {
		/* ONLY check for exec rights */
		mask = MAY_EXEC;
2066
	} else if ((fmode & FMODE_READ) && !opendata->file_created)
2067
		mask = MAY_READ;
2068 2069 2070 2071 2072 2073

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

2074
	if ((mask & ~cache.mask & (MAY_READ | MAY_EXEC)) == 0)
2075 2076 2077 2078
		return 0;

	/* even though OPEN succeeded, access is denied. Close the file */
	nfs4_close_state(state, fmode);
2079
	return -EACCES;
2080 2081
}

2082 2083 2084 2085 2086
/*
 * Note: On error, nfs4_proc_open will free the struct nfs4_opendata
 */
static int _nfs4_proc_open(struct nfs4_opendata *data)
{
2087
	struct inode *dir = d_inode(data->dir);
2088 2089 2090 2091 2092 2093
	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);
2094 2095 2096 2097 2098 2099
	if (!data->rpc_done)
		return status;
	if (status != 0) {
		if (status == -NFS4ERR_BADNAME &&
				!(o_arg->open_flags & O_CREAT))
			return -ENOENT;
2100
		return status;
2101
	}
2102

2103 2104
	nfs_fattr_map_and_free_names(server, &data->f_attr);

2105
	if (o_arg->open_flags & O_CREAT) {
2106
		update_changeattr(dir, &o_res->cinfo);
2107 2108 2109 2110 2111
		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
2112 2113
	if ((o_res->rflags & NFS4_OPEN_RESULT_LOCKTYPE_POSIX) == 0)
		server->caps &= ~NFS_CAP_POSIX_LOCK;
L
Linus Torvalds 已提交
2114
	if(o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
2115
		status = _nfs4_proc_open_confirm(data);
L
Linus Torvalds 已提交
2116
		if (status != 0)
2117
			return status;
L
Linus Torvalds 已提交
2118 2119
	}
	if (!(o_res->f_attr->valid & NFS_ATTR_FATTR))
2120
		nfs4_proc_getattr(server, &o_res->fh, o_res->f_attr, o_res->f_label);
2121
	return 0;
L
Linus Torvalds 已提交
2122 2123
}

A
Andy Adamson 已提交
2124 2125 2126 2127 2128
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
2129 2130 2131 2132 2133
/*
 * OPEN_EXPIRED:
 * 	reclaim state on the server after a network partition.
 * 	Assumes caller holds the appropriate lock
 */
2134
static int _nfs4_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
2135
{
2136
	struct nfs4_opendata *opendata;
2137
	int ret;
L
Linus Torvalds 已提交
2138

2139
	opendata = nfs4_open_recoverdata_alloc(ctx, state,
2140
			NFS4_OPEN_CLAIM_FH);
T
Trond Myklebust 已提交
2141 2142
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
2143
	ret = nfs4_open_recover(opendata, state);
2144
	if (ret == -ESTALE)
2145
		d_drop(ctx->dentry);
2146
	nfs4_opendata_put(opendata);
2147
	return ret;
L
Linus Torvalds 已提交
2148 2149
}

2150
static int nfs4_do_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
2151
{
2152
	struct nfs_server *server = NFS_SERVER(state->inode);
2153 2154 2155 2156
	struct nfs4_exception exception = { };
	int err;

	do {
2157
		err = _nfs4_open_expired(ctx, state);
2158
		trace_nfs4_open_expired(ctx, 0, err);
2159 2160
		if (nfs4_clear_cap_atomic_open_v1(server, err, &exception))
			continue;
2161 2162 2163 2164 2165 2166 2167 2168
		switch (err) {
		default:
			goto out;
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			nfs4_handle_exception(server, err, &exception);
			err = 0;
		}
2169
	} while (exception.retry);
2170
out:
2171 2172 2173
	return err;
}

L
Linus Torvalds 已提交
2174 2175 2176
static int nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
	struct nfs_open_context *ctx;
2177
	int ret;
L
Linus Torvalds 已提交
2178

2179 2180
	ctx = nfs4_state_find_open_context(state);
	if (IS_ERR(ctx))
2181
		return -EAGAIN;
2182
	ret = nfs4_do_open_expired(ctx, state);
2183 2184
	put_nfs_open_context(ctx);
	return ret;
L
Linus Torvalds 已提交
2185 2186
}

2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208
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);
}

2209
#if defined(CONFIG_NFS_V4_1)
2210
static void nfs41_check_delegation_stateid(struct nfs4_state *state)
2211 2212
{
	struct nfs_server *server = NFS_SERVER(state->inode);
2213
	nfs4_stateid stateid;
2214
	struct nfs_delegation *delegation;
2215 2216
	struct rpc_cred *cred;
	int status;
2217

2218 2219 2220
	/* Get the delegation credential for use by test/free_stateid */
	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
2221
	if (delegation == NULL) {
2222
		rcu_read_unlock();
2223 2224 2225 2226 2227 2228 2229 2230
		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);
2231

2232 2233 2234 2235
	if (status != NFS_OK) {
		/* Free the stateid unless the server explicitly
		 * informs us the stateid is unrecognized. */
		if (status != -NFS4ERR_BAD_STATEID)
2236 2237
			nfs41_free_stateid(server, &stateid, cred);
		nfs_finish_clear_delegation_stateid(state);
2238
	}
2239

2240
	put_rpccred(cred);
2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253
}

/**
 * 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);
2254
	nfs4_stateid *stateid = &state->open_stateid;
2255
	struct rpc_cred *cred = state->owner->so_cred;
2256 2257 2258 2259 2260 2261 2262 2263
	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;

2264
	status = nfs41_test_stateid(server, stateid, cred);
2265
	trace_nfs4_test_open_stateid(state, NULL, status);
2266 2267 2268 2269
	if (status != NFS_OK) {
		/* Free the stateid unless the server explicitly
		 * informs us the stateid is unrecognized. */
		if (status != -NFS4ERR_BAD_STATEID)
2270
			nfs41_free_stateid(server, stateid, cred);
2271 2272 2273 2274

		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
		clear_bit(NFS_O_RDWR_STATE, &state->flags);
2275
		clear_bit(NFS_OPEN_STATE, &state->flags);
2276 2277 2278 2279 2280 2281
	}
	return status;
}

static int nfs41_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
2282
	int status;
2283

2284
	nfs41_check_delegation_stateid(state);
2285
	status = nfs41_check_open_stateid(state);
2286 2287 2288
	if (status != NFS_OK)
		status = nfs4_open_expired(sp, state);
	return status;
2289 2290 2291
}
#endif

2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307
/*
 * 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;
}

2308 2309 2310
static int _nfs4_open_and_get_state(struct nfs4_opendata *opendata,
		fmode_t fmode,
		int flags,
2311
		struct nfs_open_context *ctx)
2312 2313 2314
{
	struct nfs4_state_owner *sp = opendata->owner;
	struct nfs_server *server = sp->so_server;
2315
	struct dentry *dentry;
2316 2317 2318 2319 2320 2321 2322
	struct nfs4_state *state;
	unsigned int seq;
	int ret;

	seq = raw_seqcount_begin(&sp->so_reclaim_seqcount);

	ret = _nfs4_proc_open(opendata);
2323
	if (ret != 0)
2324 2325 2326 2327 2328 2329 2330 2331 2332
		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);

2333
	dentry = opendata->dentry;
2334
	if (d_really_is_negative(dentry)) {
2335 2336 2337 2338 2339 2340 2341 2342 2343 2344
		/* 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,
2345
				nfs_save_change_attribute(d_inode(opendata->dir)));
2346 2347
	}

2348 2349 2350 2351
	ret = nfs4_opendata_access(sp->so_cred, opendata, state, fmode, flags);
	if (ret != 0)
		goto out;

2352
	ctx->state = state;
2353
	if (d_inode(dentry) == state->inode) {
2354 2355 2356 2357
		nfs_inode_attach_open_context(ctx);
		if (read_seqcount_retry(&sp->so_reclaim_seqcount, seq))
			nfs4_schedule_stateid_recovery(server, state);
	}
2358 2359 2360 2361
out:
	return ret;
}

L
Linus Torvalds 已提交
2362
/*
2363
 * Returns a referenced nfs4_state
L
Linus Torvalds 已提交
2364
 */
2365
static int _nfs4_do_open(struct inode *dir,
2366
			struct nfs_open_context *ctx,
2367 2368
			int flags,
			struct iattr *sattr,
2369 2370
			struct nfs4_label *label,
			int *opened)
L
Linus Torvalds 已提交
2371 2372 2373 2374
{
	struct nfs4_state_owner  *sp;
	struct nfs4_state     *state = NULL;
	struct nfs_server       *server = NFS_SERVER(dir);
2375
	struct nfs4_opendata *opendata;
2376 2377 2378 2379
	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);
2380
	enum open_claim_type4 claim = NFS4_OPEN_CLAIM_NULL;
2381
	struct nfs4_label *olabel = NULL;
2382
	int status;
L
Linus Torvalds 已提交
2383 2384 2385

	/* Protect against reboot recovery conflicts */
	status = -ENOMEM;
2386 2387
	sp = nfs4_get_state_owner(server, cred, GFP_KERNEL);
	if (sp == NULL) {
L
Linus Torvalds 已提交
2388 2389 2390
		dprintk("nfs4_do_open: nfs4_get_state_owner failed!\n");
		goto out_err;
	}
2391 2392
	status = nfs4_recover_expired_lease(server);
	if (status != 0)
2393
		goto err_put_state_owner;
2394 2395
	if (d_really_is_positive(dentry))
		nfs4_return_incompatible_delegation(d_inode(dentry), fmode);
2396
	status = -ENOMEM;
2397
	if (d_really_is_positive(dentry))
2398
		claim = NFS4_OPEN_CLAIM_FH;
2399
	opendata = nfs4_opendata_alloc(dentry, sp, fmode, flags, sattr,
2400
			label, claim, GFP_KERNEL);
2401
	if (opendata == NULL)
T
Trond Myklebust 已提交
2402
		goto err_put_state_owner;
L
Linus Torvalds 已提交
2403

2404 2405 2406 2407 2408 2409 2410 2411
	if (label) {
		olabel = nfs4_label_alloc(server, GFP_KERNEL);
		if (IS_ERR(olabel)) {
			status = PTR_ERR(olabel);
			goto err_opendata_put;
		}
	}

2412 2413 2414 2415 2416 2417
	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;
		}
2418
		opendata->o_arg.open_bitmap = &nfs4_pnfs_open_bitmap[0];
2419
	}
2420 2421
	if (d_really_is_positive(dentry))
		opendata->state = nfs4_get_open_state(d_inode(dentry), sp);
2422

2423
	status = _nfs4_open_and_get_state(opendata, fmode, flags, ctx);
2424
	if (status != 0)
2425
		goto err_free_label;
2426
	state = ctx->state;
2427

2428
	if ((opendata->o_arg.open_flags & (O_CREAT|O_EXCL)) == (O_CREAT|O_EXCL) &&
2429
	    (opendata->o_arg.createmode != NFS4_CREATE_GUARDED)) {
2430 2431 2432 2433 2434
		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,
2435 2436
				state, label, olabel);
		if (status == 0) {
2437 2438
			nfs_setattr_update_inode(state->inode, sattr,
					opendata->o_res.f_attr);
2439
			nfs_setsecurity(state->inode, opendata->o_res.f_attr, olabel);
2440
		}
2441
	}
2442 2443
	if (opendata->file_created)
		*opened |= FILE_CREATED;
2444

2445
	if (pnfs_use_threshold(ctx_th, opendata->f_attr.mdsthreshold, server)) {
2446
		*ctx_th = opendata->f_attr.mdsthreshold;
2447 2448
		opendata->f_attr.mdsthreshold = NULL;
	}
2449

2450 2451
	nfs4_label_free(olabel);

2452
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
2453 2454
	nfs4_put_state_owner(sp);
	return 0;
2455 2456
err_free_label:
	nfs4_label_free(olabel);
2457 2458
err_opendata_put:
	nfs4_opendata_put(opendata);
2459 2460
err_put_state_owner:
	nfs4_put_state_owner(sp);
L
Linus Torvalds 已提交
2461 2462 2463 2464 2465
out_err:
	return status;
}


2466
static struct nfs4_state *nfs4_do_open(struct inode *dir,
2467
					struct nfs_open_context *ctx,
2468 2469
					int flags,
					struct iattr *sattr,
2470 2471
					struct nfs4_label *label,
					int *opened)
L
Linus Torvalds 已提交
2472
{
2473
	struct nfs_server *server = NFS_SERVER(dir);
L
Linus Torvalds 已提交
2474 2475 2476 2477 2478
	struct nfs4_exception exception = { };
	struct nfs4_state *res;
	int status;

	do {
2479
		status = _nfs4_do_open(dir, ctx, flags, sattr, label, opened);
2480
		res = ctx->state;
2481
		trace_nfs4_open_file(ctx, flags, status);
L
Linus Torvalds 已提交
2482 2483 2484 2485 2486 2487 2488 2489
		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
2490
		 * doing an OPEN, we assume that nfs_increment_open_seqid() will
L
Linus Torvalds 已提交
2491 2492 2493 2494 2495
		 * 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) {
2496
			pr_warn_ratelimited("NFS: v4 server %s "
2497 2498
					" returned a bad sequence-id error!\n",
					NFS_SERVER(dir)->nfs_client->cl_hostname);
L
Linus Torvalds 已提交
2499 2500 2501
			exception.retry = 1;
			continue;
		}
2502 2503 2504 2505 2506 2507 2508 2509 2510 2511
		/*
		 * 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;
		}
2512 2513 2514 2515 2516
		if (status == -EAGAIN) {
			/* We must have found a delegation */
			exception.retry = 1;
			continue;
		}
2517 2518 2519
		if (nfs4_clear_cap_atomic_open_v1(server, status, &exception))
			continue;
		res = ERR_PTR(nfs4_handle_exception(server,
L
Linus Torvalds 已提交
2520 2521 2522 2523 2524
					status, &exception));
	} while (exception.retry);
	return res;
}

2525 2526
static int _nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
			    struct nfs_fattr *fattr, struct iattr *sattr,
2527 2528
			    struct nfs4_state *state, struct nfs4_label *ilabel,
			    struct nfs4_label *olabel)
L
Linus Torvalds 已提交
2529
{
2530
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
2531
        struct nfs_setattrargs  arg = {
2532
                .fh             = NFS_FH(inode),
L
Linus Torvalds 已提交
2533 2534 2535
                .iap            = sattr,
		.server		= server,
		.bitmask = server->attr_bitmask,
2536
		.label		= ilabel,
L
Linus Torvalds 已提交
2537 2538 2539
        };
        struct nfs_setattrres  res = {
		.fattr		= fattr,
2540
		.label		= olabel,
L
Linus Torvalds 已提交
2541 2542 2543
		.server		= server,
        };
        struct rpc_message msg = {
2544 2545 2546 2547
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETATTR],
		.rpc_argp	= &arg,
		.rpc_resp	= &res,
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
2548
        };
2549
	unsigned long timestamp = jiffies;
2550 2551
	fmode_t fmode;
	bool truncate;
2552
	int status;
L
Linus Torvalds 已提交
2553

2554 2555 2556 2557
	arg.bitmask = nfs4_bitmask(server, ilabel);
	if (ilabel)
		arg.bitmask = nfs4_bitmask(server, olabel);

2558
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
2559

2560 2561 2562 2563 2564 2565
	/* 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 */
2566
	} else if (truncate && state != NULL) {
2567 2568 2569 2570
		struct nfs_lockowner lockowner = {
			.l_owner = current->files,
			.l_pid = current->tgid,
		};
2571 2572 2573 2574 2575
		if (!nfs4_valid_open_stateid(state))
			return -EBADF;
		if (nfs4_select_rw_stateid(&arg.stateid, state, FMODE_WRITE,
				&lockowner) == -EIO)
			return -EBADF;
2576
	} else
2577
		nfs4_stateid_copy(&arg.stateid, &zero_stateid);
L
Linus Torvalds 已提交
2578

2579
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
2580 2581
	if (status == 0 && state != NULL)
		renew_lease(server, timestamp);
2582
	return status;
L
Linus Torvalds 已提交
2583 2584
}

2585 2586
static int nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
			   struct nfs_fattr *fattr, struct iattr *sattr,
2587 2588
			   struct nfs4_state *state, struct nfs4_label *ilabel,
			   struct nfs4_label *olabel)
L
Linus Torvalds 已提交
2589
{
2590
	struct nfs_server *server = NFS_SERVER(inode);
2591 2592
	struct nfs4_exception exception = {
		.state = state,
2593
		.inode = inode,
2594
	};
L
Linus Torvalds 已提交
2595 2596
	int err;
	do {
2597
		err = _nfs4_do_setattr(inode, cred, fattr, sattr, state, ilabel, olabel);
2598
		trace_nfs4_setattr(inode, err);
2599 2600
		switch (err) {
		case -NFS4ERR_OPENMODE:
2601 2602 2603 2604 2605 2606 2607
			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);
			}
2608 2609 2610 2611 2612 2613 2614 2615
			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
2616
	} while (exception.retry);
2617
out:
L
Linus Torvalds 已提交
2618 2619 2620 2621 2622 2623 2624 2625
	return err;
}

struct nfs4_closedata {
	struct inode *inode;
	struct nfs4_state *state;
	struct nfs_closeargs arg;
	struct nfs_closeres res;
2626
	struct nfs_fattr fattr;
2627
	unsigned long timestamp;
F
Fred Isaman 已提交
2628 2629
	bool roc;
	u32 roc_barrier;
L
Linus Torvalds 已提交
2630 2631
};

2632
static void nfs4_free_closedata(void *data)
2633
{
2634 2635
	struct nfs4_closedata *calldata = data;
	struct nfs4_state_owner *sp = calldata->state->owner;
2636
	struct super_block *sb = calldata->state->inode->i_sb;
2637

F
Fred Isaman 已提交
2638 2639
	if (calldata->roc)
		pnfs_roc_release(calldata->state->inode);
2640 2641 2642
	nfs4_put_open_state(calldata->state);
	nfs_free_seqid(calldata->arg.seqid);
	nfs4_put_state_owner(sp);
2643
	nfs_sb_deactive(sb);
2644 2645 2646
	kfree(calldata);
}

2647
static void nfs4_close_done(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
2648
{
2649
	struct nfs4_closedata *calldata = data;
L
Linus Torvalds 已提交
2650 2651
	struct nfs4_state *state = calldata->state;
	struct nfs_server *server = NFS_SERVER(calldata->inode);
2652
	nfs4_stateid *res_stateid = NULL;
L
Linus Torvalds 已提交
2653

2654
	dprintk("%s: begin!\n", __func__);
2655 2656
	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
		return;
2657
	trace_nfs4_close(state, &calldata->arg, &calldata->res, task->tk_status);
L
Linus Torvalds 已提交
2658 2659 2660 2661 2662
        /* 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:
2663 2664
			res_stateid = &calldata->res.stateid;
			if (calldata->arg.fmode == 0 && calldata->roc)
F
Fred Isaman 已提交
2665 2666
				pnfs_roc_set_barrier(state->inode,
						     calldata->roc_barrier);
2667
			renew_lease(server, calldata->timestamp);
2668
			break;
2669
		case -NFS4ERR_ADMIN_REVOKED:
L
Linus Torvalds 已提交
2670
		case -NFS4ERR_STALE_STATEID:
2671 2672
		case -NFS4ERR_OLD_STATEID:
		case -NFS4ERR_BAD_STATEID:
L
Linus Torvalds 已提交
2673
		case -NFS4ERR_EXPIRED:
2674
			if (!nfs4_stateid_match(&calldata->arg.stateid,
2675
						&state->open_stateid)) {
2676 2677 2678
				rpc_restart_call_prepare(task);
				goto out_release;
			}
2679
			if (calldata->arg.fmode == 0)
2680
				break;
L
Linus Torvalds 已提交
2681
		default:
2682
			if (nfs4_async_handle_error(task, server, state, NULL) == -EAGAIN) {
2683
				rpc_restart_call_prepare(task);
2684 2685
				goto out_release;
			}
L
Linus Torvalds 已提交
2686
	}
2687
	nfs_clear_open_stateid(state, res_stateid, calldata->arg.fmode);
2688
out_release:
2689
	nfs_release_seqid(calldata->arg.seqid);
2690
	nfs_refresh_inode(calldata->inode, calldata->res.fattr);
2691
	dprintk("%s: done, ret = %d!\n", __func__, task->tk_status);
L
Linus Torvalds 已提交
2692 2693
}

T
Trond Myklebust 已提交
2694
static void nfs4_close_prepare(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
2695
{
T
Trond Myklebust 已提交
2696
	struct nfs4_closedata *calldata = data;
2697
	struct nfs4_state *state = calldata->state;
2698
	struct inode *inode = calldata->inode;
2699
	bool is_rdonly, is_wronly, is_rdwr;
2700
	int call_close = 0;
2701

2702
	dprintk("%s: begin!\n", __func__);
2703
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
2704
		goto out_wait;
2705

2706
	task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_DOWNGRADE];
2707
	spin_lock(&state->owner->so_lock);
2708 2709 2710
	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);
2711
	nfs4_stateid_copy(&calldata->arg.stateid, &state->open_stateid);
2712
	/* Calculate the change in open mode */
2713
	calldata->arg.fmode = 0;
2714
	if (state->n_rdwr == 0) {
2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728
		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;

2729 2730
	if (!nfs4_valid_open_stateid(state))
		call_close = 0;
2731
	spin_unlock(&state->owner->so_lock);
2732 2733

	if (!call_close) {
2734
		/* Note: exit _without_ calling nfs4_close_done */
2735
		goto out_no_action;
2736
	}
2737

F
Fred Isaman 已提交
2738
	if (calldata->arg.fmode == 0) {
2739
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE];
2740 2741
		if (calldata->roc)
			pnfs_roc_get_barrier(inode, &calldata->roc_barrier);
F
Fred Isaman 已提交
2742
	}
2743 2744 2745
	calldata->arg.share_access =
		nfs4_map_atomic_open_share(NFS_SERVER(inode),
				calldata->arg.fmode, 0);
2746

2747
	nfs_fattr_init(calldata->res.fattr);
2748
	calldata->timestamp = jiffies;
2749
	if (nfs4_setup_sequence(NFS_SERVER(inode),
2750 2751
				&calldata->arg.seq_args,
				&calldata->res.seq_res,
2752 2753
				task) != 0)
		nfs_release_seqid(calldata->arg.seqid);
2754
	dprintk("%s: done!\n", __func__);
2755 2756 2757 2758 2759
	return;
out_no_action:
	task->tk_action = NULL;
out_wait:
	nfs4_sequence_done(task, &calldata->res.seq_res);
L
Linus Torvalds 已提交
2760 2761
}

2762
static const struct rpc_call_ops nfs4_close_ops = {
T
Trond Myklebust 已提交
2763
	.rpc_call_prepare = nfs4_close_prepare,
2764 2765 2766 2767
	.rpc_call_done = nfs4_close_done,
	.rpc_release = nfs4_free_closedata,
};

P
Peng Tao 已提交
2768 2769
static bool nfs4_roc(struct inode *inode)
{
2770
	if (!nfs_have_layout(inode))
P
Peng Tao 已提交
2771 2772 2773 2774
		return false;
	return pnfs_roc(inode);
}

L
Linus Torvalds 已提交
2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785
/* 
 * 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!
 */
2786
int nfs4_do_close(struct nfs4_state *state, gfp_t gfp_mask, int wait)
L
Linus Torvalds 已提交
2787
{
2788
	struct nfs_server *server = NFS_SERVER(state->inode);
2789
	struct nfs_seqid *(*alloc_seqid)(struct nfs_seqid_counter *, gfp_t);
L
Linus Torvalds 已提交
2790
	struct nfs4_closedata *calldata;
2791 2792
	struct nfs4_state_owner *sp = state->owner;
	struct rpc_task *task;
2793 2794 2795 2796
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
2797 2798
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
2799
		.rpc_message = &msg,
T
Trond Myklebust 已提交
2800
		.callback_ops = &nfs4_close_ops,
2801
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
2802 2803
		.flags = RPC_TASK_ASYNC,
	};
2804
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2805

2806 2807 2808
	nfs4_state_protect(server->nfs_client, NFS_SP4_MACH_CRED_CLEANUP,
		&task_setup_data.rpc_client, &msg);

2809
	calldata = kzalloc(sizeof(*calldata), gfp_mask);
L
Linus Torvalds 已提交
2810
	if (calldata == NULL)
2811
		goto out;
2812
	nfs4_init_sequence(&calldata->arg.seq_args, &calldata->res.seq_res, 1);
2813
	calldata->inode = state->inode;
L
Linus Torvalds 已提交
2814
	calldata->state = state;
2815
	calldata->arg.fh = NFS_FH(state->inode);
L
Linus Torvalds 已提交
2816
	/* Serialization for the sequence id */
2817 2818
	alloc_seqid = server->nfs_client->cl_mvops->alloc_seqid;
	calldata->arg.seqid = alloc_seqid(&state->owner->so_seqid, gfp_mask);
2819
	if (IS_ERR(calldata->arg.seqid))
2820
		goto out_free_calldata;
2821
	calldata->arg.fmode = 0;
2822
	calldata->arg.bitmask = server->cache_consistency_bitmask;
2823
	calldata->res.fattr = &calldata->fattr;
2824
	calldata->res.seqid = calldata->arg.seqid;
2825
	calldata->res.server = server;
P
Peng Tao 已提交
2826
	calldata->roc = nfs4_roc(state->inode);
2827
	nfs_sb_active(calldata->inode->i_sb);
2828

2829 2830
	msg.rpc_argp = &calldata->arg;
	msg.rpc_resp = &calldata->res;
T
Trond Myklebust 已提交
2831 2832
	task_setup_data.callback_data = calldata;
	task = rpc_run_task(&task_setup_data);
2833 2834
	if (IS_ERR(task))
		return PTR_ERR(task);
2835 2836 2837
	status = 0;
	if (wait)
		status = rpc_wait_for_completion_task(task);
2838
	rpc_put_task(task);
2839
	return status;
2840 2841 2842
out_free_calldata:
	kfree(calldata);
out:
2843 2844
	nfs4_put_open_state(state);
	nfs4_put_state_owner(sp);
2845
	return status;
L
Linus Torvalds 已提交
2846 2847
}

2848
static struct inode *
2849 2850
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
2851 2852
{
	struct nfs4_state *state;
2853 2854 2855
	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
2856

T
Trond Myklebust 已提交
2857
	/* Protect against concurrent sillydeletes */
2858
	state = nfs4_do_open(dir, ctx, open_flags, attr, label, opened);
2859 2860 2861

	nfs4_label_release_security(label);

2862 2863
	if (IS_ERR(state))
		return ERR_CAST(state);
2864
	return state->inode;
L
Linus Torvalds 已提交
2865 2866
}

2867
static void nfs4_close_context(struct nfs_open_context *ctx, int is_sync)
T
Trond Myklebust 已提交
2868 2869 2870 2871
{
	if (ctx->state == NULL)
		return;
	if (is_sync)
2872
		nfs4_close_sync(ctx->state, ctx->mode);
T
Trond Myklebust 已提交
2873
	else
2874
		nfs4_close_state(ctx->state, ctx->mode);
T
Trond Myklebust 已提交
2875
}
L
Linus Torvalds 已提交
2876

2877 2878
#define FATTR4_WORD1_NFS40_MASK (2*FATTR4_WORD1_MOUNTED_ON_FILEID - 1UL)
#define FATTR4_WORD2_NFS41_MASK (2*FATTR4_WORD2_SUPPATTR_EXCLCREAT - 1UL)
2879
#define FATTR4_WORD2_NFS42_MASK (2*FATTR4_WORD2_SECURITY_LABEL - 1UL)
2880

L
Linus Torvalds 已提交
2881 2882
static int _nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
{
B
Benny Halevy 已提交
2883 2884 2885
	struct nfs4_server_caps_arg args = {
		.fhandle = fhandle,
	};
L
Linus Torvalds 已提交
2886 2887 2888
	struct nfs4_server_caps_res res = {};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SERVER_CAPS],
B
Benny Halevy 已提交
2889
		.rpc_argp = &args,
L
Linus Torvalds 已提交
2890 2891 2892 2893
		.rpc_resp = &res,
	};
	int status;

2894
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
2895
	if (status == 0) {
2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907
		/* 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
2908
		memcpy(server->attr_bitmask, res.attr_bitmask, sizeof(server->attr_bitmask));
2909 2910 2911 2912
		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|
2913 2914
				NFS_CAP_CTIME|NFS_CAP_MTIME|
				NFS_CAP_SECURITY_LABEL);
2915 2916
		if (res.attr_bitmask[0] & FATTR4_WORD0_ACL &&
				res.acl_bitmask & ACL4_SUPPORT_ALLOW_ACL)
L
Linus Torvalds 已提交
2917 2918 2919 2920 2921
			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;
2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937
		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;
2938 2939 2940 2941 2942 2943
#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));
2944
		server->attr_bitmask_nl[2] &= ~FATTR4_WORD2_SECURITY_LABEL;
2945

2946 2947 2948
		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;
2949
		server->cache_consistency_bitmask[2] = 0;
L
Linus Torvalds 已提交
2950
		server->acl_bitmask = res.acl_bitmask;
2951
		server->fh_expire_type = res.fh_expire_type;
L
Linus Torvalds 已提交
2952
	}
A
Andy Adamson 已提交
2953

L
Linus Torvalds 已提交
2954 2955 2956
	return status;
}

2957
int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
L
Linus Torvalds 已提交
2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971
{
	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)
{
2972
	u32 bitmask[3];
L
Linus Torvalds 已提交
2973
	struct nfs4_lookup_root_arg args = {
2974
		.bitmask = bitmask,
L
Linus Torvalds 已提交
2975 2976 2977
	};
	struct nfs4_lookup_res res = {
		.server = server,
2978
		.fattr = info->fattr,
L
Linus Torvalds 已提交
2979 2980 2981 2982 2983 2984 2985
		.fh = fhandle,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP_ROOT],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
2986

2987 2988 2989 2990 2991 2992 2993
	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;

2994
	nfs_fattr_init(info->fattr);
2995
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
2996 2997 2998 2999 3000 3001 3002 3003
}

static int nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle,
		struct nfs_fsinfo *info)
{
	struct nfs4_exception exception = { };
	int err;
	do {
3004
		err = _nfs4_lookup_root(server, fhandle, info);
3005
		trace_nfs4_lookup_root(server, fhandle, info->fattr, err);
3006 3007 3008
		switch (err) {
		case 0:
		case -NFS4ERR_WRONGSEC:
3009
			goto out;
3010 3011 3012
		default:
			err = nfs4_handle_exception(server, err, &exception);
		}
L
Linus Torvalds 已提交
3013
	} while (exception.retry);
3014
out:
L
Linus Torvalds 已提交
3015 3016 3017
	return err;
}

3018 3019 3020
static int nfs4_lookup_root_sec(struct nfs_server *server, struct nfs_fh *fhandle,
				struct nfs_fsinfo *info, rpc_authflavor_t flavor)
{
3021 3022 3023
	struct rpc_auth_create_args auth_args = {
		.pseudoflavor = flavor,
	};
3024 3025 3026
	struct rpc_auth *auth;
	int ret;

3027
	auth = rpcauth_create(&auth_args, server->client);
3028
	if (IS_ERR(auth)) {
3029
		ret = -EACCES;
3030 3031 3032 3033 3034 3035 3036
		goto out;
	}
	ret = nfs4_lookup_root(server, fhandle, info);
out:
	return ret;
}

3037 3038 3039 3040 3041 3042 3043 3044 3045
/*
 * 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.
 */
3046
static int nfs4_find_root_sec(struct nfs_server *server, struct nfs_fh *fhandle,
3047
			      struct nfs_fsinfo *info)
L
Linus Torvalds 已提交
3048
{
3049 3050 3051 3052 3053
	/* Per 3530bis 15.33.5 */
	static const rpc_authflavor_t flav_array[] = {
		RPC_AUTH_GSS_KRB5P,
		RPC_AUTH_GSS_KRB5I,
		RPC_AUTH_GSS_KRB5,
3054
		RPC_AUTH_UNIX,			/* courtesy */
3055 3056 3057 3058
		RPC_AUTH_NULL,
	};
	int status = -EPERM;
	size_t i;
C
Chuck Lever 已提交
3059

3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077
	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;
		}
3078
	}
3079

3080 3081 3082 3083 3084 3085 3086 3087 3088
	/*
	 * -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;
3089 3090 3091
	return status;
}

C
Chuck Lever 已提交
3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103
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
3104
 * @auth_probe: probe the auth flavours
C
Chuck Lever 已提交
3105 3106
 *
 * Returns zero on success, or a negative errno.
3107
 */
3108
int nfs4_proc_get_rootfh(struct nfs_server *server, struct nfs_fh *fhandle,
3109 3110
			 struct nfs_fsinfo *info,
			 bool auth_probe)
3111
{
3112
	int status = 0;
C
Chuck Lever 已提交
3113

3114
	if (!auth_probe)
3115
		status = nfs4_lookup_root(server, fhandle, info);
3116 3117

	if (auth_probe || status == NFS4ERR_WRONGSEC)
C
Chuck Lever 已提交
3118 3119
		status = nfs4_do_find_root_sec(server, fhandle, info);

L
Linus Torvalds 已提交
3120 3121 3122 3123
	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
3124

3125
	return nfs4_map_errors(status);
L
Linus Torvalds 已提交
3126 3127
}

3128 3129 3130 3131 3132
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;
3133
	struct nfs4_label *label = NULL;
3134 3135 3136 3137 3138 3139 3140

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

3141 3142 3143 3144
	label = nfs4_label_alloc(server, GFP_KERNEL);
	if (IS_ERR(label))
		return PTR_ERR(label);

3145
	error = nfs4_proc_getattr(server, mntfh, fattr, label);
3146 3147
	if (error < 0) {
		dprintk("nfs4_get_root: getattr error = %d\n", -error);
3148
		goto err_free_label;
3149 3150 3151 3152 3153 3154
	}

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

3155 3156 3157
err_free_label:
	nfs4_label_free(label);

3158 3159 3160
	return error;
}

M
Manoj Naik 已提交
3161 3162 3163 3164 3165
/*
 * 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
 */
3166 3167 3168
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
3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180
{
	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;

3181
	status = nfs4_proc_fs_locations(client, dir, name, locations, page);
M
Manoj Naik 已提交
3182 3183
	if (status != 0)
		goto out;
3184 3185 3186 3187 3188 3189

	/*
	 * 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
3190
	if (nfs_fsid_equal(&NFS_SERVER(dir)->fsid, &locations->fattr.fsid)) {
3191 3192
		dprintk("%s: server did not return a different fsid for"
			" a referral at %s\n", __func__, name->name);
3193
		status = -NFS4ERR_MOVED;
M
Manoj Naik 已提交
3194 3195
		goto out;
	}
3196 3197
	/* Fixup attributes for the nfs_lookup() call to nfs_fhget() */
	nfs_fixup_referral_attributes(&locations->fattr);
M
Manoj Naik 已提交
3198

3199
	/* replace the lookup nfs_fattr with the locations nfs_fattr */
M
Manoj Naik 已提交
3200 3201 3202 3203 3204
	memcpy(fattr, &locations->fattr, sizeof(struct nfs_fattr));
	memset(fhandle, 0, sizeof(struct nfs_fh));
out:
	if (page)
		__free_page(page);
3205
	kfree(locations);
M
Manoj Naik 已提交
3206 3207 3208
	return status;
}

3209 3210
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
3211 3212 3213 3214 3215 3216 3217
{
	struct nfs4_getattr_arg args = {
		.fh = fhandle,
		.bitmask = server->attr_bitmask,
	};
	struct nfs4_getattr_res res = {
		.fattr = fattr,
3218
		.label = label,
L
Linus Torvalds 已提交
3219 3220 3221 3222 3223 3224 3225
		.server = server,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETATTR],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
3226 3227 3228

	args.bitmask = nfs4_bitmask(server, label);

3229
	nfs_fattr_init(fattr);
3230
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3231 3232
}

3233 3234
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
3235 3236 3237 3238
{
	struct nfs4_exception exception = { };
	int err;
	do {
3239 3240 3241
		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
3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267
				&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)
{
3268
	struct inode *inode = d_inode(dentry);
3269
	struct rpc_cred *cred = NULL;
3270
	struct nfs4_state *state = NULL;
3271
	struct nfs4_label *label = NULL;
L
Linus Torvalds 已提交
3272 3273
	int status;

3274 3275 3276
	if (pnfs_ld_layoutret_on_setattr(inode) &&
	    sattr->ia_valid & ATTR_SIZE &&
	    sattr->ia_size < i_size_read(inode))
3277
		pnfs_commit_and_return_layout(inode);
B
Benny Halevy 已提交
3278

3279
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
3280
	
3281 3282
	/* Deal with open(O_TRUNC) */
	if (sattr->ia_valid & ATTR_OPEN)
3283
		sattr->ia_valid &= ~(ATTR_MTIME|ATTR_CTIME);
3284 3285

	/* Optimization: if the end result is no change, don't RPC */
3286
	if ((sattr->ia_valid & ~(ATTR_FILE|ATTR_OPEN)) == 0)
3287 3288
		return 0;

3289
	/* Search for an existing open(O_WRITE) file */
3290 3291 3292 3293
	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
3294 3295 3296 3297
		if (ctx) {
			cred = ctx->cred;
			state = ctx->state;
		}
3298
	}
3299

3300 3301 3302 3303 3304
	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);
3305
	if (status == 0) {
3306
		nfs_setattr_update_inode(inode, sattr, fattr);
3307 3308
		nfs_setsecurity(inode, fattr, label);
	}
3309
	nfs4_label_free(label);
L
Linus Torvalds 已提交
3310 3311 3312
	return status;
}

3313 3314
static int _nfs4_proc_lookup(struct rpc_clnt *clnt, struct inode *dir,
		const struct qstr *name, struct nfs_fh *fhandle,
3315
		struct nfs_fattr *fattr, struct nfs4_label *label)
D
David Howells 已提交
3316
{
3317
	struct nfs_server *server = NFS_SERVER(dir);
D
David Howells 已提交
3318 3319 3320
	int		       status;
	struct nfs4_lookup_arg args = {
		.bitmask = server->attr_bitmask,
3321
		.dir_fh = NFS_FH(dir),
D
David Howells 已提交
3322 3323 3324 3325 3326
		.name = name,
	};
	struct nfs4_lookup_res res = {
		.server = server,
		.fattr = fattr,
3327
		.label = label,
D
David Howells 已提交
3328 3329 3330 3331 3332 3333 3334 3335
		.fh = fhandle,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};

3336 3337
	args.bitmask = nfs4_bitmask(server, label);

D
David Howells 已提交
3338 3339
	nfs_fattr_init(fattr);

L
Linus Torvalds 已提交
3340
	dprintk("NFS call  lookup %s\n", name->name);
3341
	status = nfs4_call_sync(clnt, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3342 3343 3344 3345
	dprintk("NFS reply lookup: %d\n", status);
	return status;
}

3346
static void nfs_fixup_secinfo_attributes(struct nfs_fattr *fattr)
3347 3348
{
	fattr->valid |= NFS_ATTR_FATTR_TYPE | NFS_ATTR_FATTR_MODE |
3349
		NFS_ATTR_FATTR_NLINK | NFS_ATTR_FATTR_MOUNTPOINT;
3350 3351 3352 3353
	fattr->mode = S_IFDIR | S_IRUGO | S_IXUGO;
	fattr->nlink = 2;
}

3354 3355
static int nfs4_proc_lookup_common(struct rpc_clnt **clnt, struct inode *dir,
				   struct qstr *name, struct nfs_fh *fhandle,
3356
				   struct nfs_fattr *fattr, struct nfs4_label *label)
L
Linus Torvalds 已提交
3357 3358
{
	struct nfs4_exception exception = { };
3359
	struct rpc_clnt *client = *clnt;
L
Linus Torvalds 已提交
3360 3361
	int err;
	do {
3362
		err = _nfs4_proc_lookup(client, dir, name, fhandle, fattr, label);
3363
		trace_nfs4_lookup(dir, name, err);
3364
		switch (err) {
3365
		case -NFS4ERR_BADNAME:
3366 3367
			err = -ENOENT;
			goto out;
3368
		case -NFS4ERR_MOVED:
3369
			err = nfs4_get_referral(client, dir, name, fattr, fhandle);
3370 3371
			if (err == -NFS4ERR_MOVED)
				err = nfs4_handle_exception(NFS_SERVER(dir), err, &exception);
3372
			goto out;
3373
		case -NFS4ERR_WRONGSEC:
3374 3375 3376
			err = -EPERM;
			if (client != *clnt)
				goto out;
3377
			client = nfs4_negotiate_security(client, dir, name);
3378 3379 3380 3381 3382 3383 3384
			if (IS_ERR(client))
				return PTR_ERR(client);

			exception.retry = 1;
			break;
		default:
			err = nfs4_handle_exception(NFS_SERVER(dir), err, &exception);
3385
		}
L
Linus Torvalds 已提交
3386
	} while (exception.retry);
3387 3388 3389 3390 3391 3392 3393

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

L
Linus Torvalds 已提交
3394 3395 3396
	return err;
}

3397
static int nfs4_proc_lookup(struct inode *dir, struct qstr *name,
3398 3399
			    struct nfs_fh *fhandle, struct nfs_fattr *fattr,
			    struct nfs4_label *label)
3400 3401 3402 3403
{
	int status;
	struct rpc_clnt *client = NFS_CLIENT(dir);

3404
	status = nfs4_proc_lookup_common(&client, dir, name, fhandle, fattr, label);
3405 3406 3407 3408 3409 3410 3411
	if (client != NFS_CLIENT(dir)) {
		rpc_shutdown_client(client);
		nfs_fixup_secinfo_attributes(fattr);
	}
	return status;
}

3412 3413 3414 3415
struct rpc_clnt *
nfs4_proc_lookup_mountpoint(struct inode *dir, struct qstr *name,
			    struct nfs_fh *fhandle, struct nfs_fattr *fattr)
{
3416
	struct rpc_clnt *client = NFS_CLIENT(dir);
3417 3418
	int status;

3419
	status = nfs4_proc_lookup_common(&client, dir, name, fhandle, fattr, NULL);
3420
	if (status < 0)
3421
		return ERR_PTR(status);
3422
	return (client == NFS_CLIENT(dir)) ? rpc_clone_client(client) : client;
3423 3424
}

L
Linus Torvalds 已提交
3425 3426
static int _nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
{
3427
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
3428 3429
	struct nfs4_accessargs args = {
		.fh = NFS_FH(inode),
3430
		.bitmask = server->cache_consistency_bitmask,
3431 3432 3433
	};
	struct nfs4_accessres res = {
		.server = server,
L
Linus Torvalds 已提交
3434 3435 3436 3437 3438 3439 3440 3441
	};
	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;
3442
	int status = 0;
L
Linus Torvalds 已提交
3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459

	/*
	 * 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;
	}
3460 3461 3462 3463 3464

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

3465
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3466
	if (!status) {
3467
		nfs_access_set_mask(entry, res.access);
3468
		nfs_refresh_inode(inode, res.fattr);
L
Linus Torvalds 已提交
3469
	}
3470
	nfs_free_fattr(res.fattr);
L
Linus Torvalds 已提交
3471 3472 3473 3474 3475 3476 3477 3478
	return status;
}

static int nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
{
	struct nfs4_exception exception = { };
	int err;
	do {
3479 3480 3481
		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
3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503
				&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
3504
 * we get the post-operation mtime and size.
L
Linus Torvalds 已提交
3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517
 *
 * 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
3518
	struct nfs4_readlink_res res;
L
Linus Torvalds 已提交
3519 3520 3521
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READLINK],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3522
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3523 3524
	};

3525
	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
3526 3527 3528 3529 3530 3531 3532 3533
}

static int nfs4_proc_readlink(struct inode *inode, struct page *page,
		unsigned int pgbase, unsigned int pglen)
{
	struct nfs4_exception exception = { };
	int err;
	do {
3534 3535 3536
		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
3537 3538 3539 3540 3541 3542
				&exception);
	} while (exception.retry);
	return err;
}

/*
3543
 * This is just for mknod.  open(O_CREAT) will always do ->open_context().
L
Linus Torvalds 已提交
3544 3545 3546
 */
static int
nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
3547
		 int flags)
L
Linus Torvalds 已提交
3548
{
3549
	struct nfs4_label l, *ilabel = NULL;
3550
	struct nfs_open_context *ctx;
L
Linus Torvalds 已提交
3551
	struct nfs4_state *state;
3552
	int opened = 0;
L
Linus Torvalds 已提交
3553 3554
	int status = 0;

3555 3556 3557 3558
	ctx = alloc_nfs_open_context(dentry, FMODE_READ);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);

3559 3560
	ilabel = nfs4_label_init_security(dir, dentry, sattr, &l);

A
Aneesh Kumar K.V 已提交
3561
	sattr->ia_mode &= ~current_umask();
3562
	state = nfs4_do_open(dir, ctx, flags, sattr, ilabel, &opened);
L
Linus Torvalds 已提交
3563 3564
	if (IS_ERR(state)) {
		status = PTR_ERR(state);
3565
		goto out;
L
Linus Torvalds 已提交
3566 3567
	}
out:
3568
	nfs4_label_release_security(ilabel);
3569
	put_nfs_open_context(ctx);
L
Linus Torvalds 已提交
3570 3571 3572 3573 3574
	return status;
}

static int _nfs4_proc_remove(struct inode *dir, struct qstr *name)
{
3575
	struct nfs_server *server = NFS_SERVER(dir);
3576
	struct nfs_removeargs args = {
L
Linus Torvalds 已提交
3577
		.fh = NFS_FH(dir),
3578
		.name = *name,
3579
	};
3580
	struct nfs_removeres res = {
3581
		.server = server,
L
Linus Torvalds 已提交
3582 3583
	};
	struct rpc_message msg = {
3584 3585 3586
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE],
		.rpc_argp = &args,
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3587
	};
3588
	int status;
L
Linus Torvalds 已提交
3589

3590
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 1);
3591
	if (status == 0)
3592
		update_changeattr(dir, &res.cinfo);
L
Linus Torvalds 已提交
3593 3594 3595 3596 3597 3598 3599 3600
	return status;
}

static int nfs4_proc_remove(struct inode *dir, struct qstr *name)
{
	struct nfs4_exception exception = { };
	int err;
	do {
3601 3602 3603
		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
3604 3605 3606 3607 3608
				&exception);
	} while (exception.retry);
	return err;
}

3609
static void nfs4_proc_unlink_setup(struct rpc_message *msg, struct inode *dir)
L
Linus Torvalds 已提交
3610
{
3611 3612 3613
	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
3614

3615
	res->server = server;
L
Linus Torvalds 已提交
3616
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE];
3617
	nfs4_init_sequence(&args->seq_args, &res->seq_res, 1);
3618 3619

	nfs_fattr_init(res->dir_attr);
L
Linus Torvalds 已提交
3620 3621
}

3622 3623
static void nfs4_proc_unlink_rpc_prepare(struct rpc_task *task, struct nfs_unlinkdata *data)
{
3624 3625 3626 3627
	nfs4_setup_sequence(NFS_SERVER(data->dir),
			&data->args.seq_args,
			&data->res.seq_res,
			task);
L
Linus Torvalds 已提交
3628 3629
}

3630
static int nfs4_proc_unlink_done(struct rpc_task *task, struct inode *dir)
L
Linus Torvalds 已提交
3631
{
3632 3633
	struct nfs_unlinkdata *data = task->tk_calldata;
	struct nfs_removeres *res = &data->res;
3634

3635 3636
	if (!nfs4_sequence_done(task, &res->seq_res))
		return 0;
3637 3638
	if (nfs4_async_handle_error(task, res->server, NULL,
				    &data->timeout) == -EAGAIN)
3639 3640 3641
		return 0;
	update_changeattr(dir, &res->cinfo);
	return 1;
L
Linus Torvalds 已提交
3642 3643
}

3644 3645 3646 3647 3648 3649 3650 3651
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;
3652
	nfs4_init_sequence(&arg->seq_args, &res->seq_res, 1);
3653 3654
}

3655 3656
static void nfs4_proc_rename_rpc_prepare(struct rpc_task *task, struct nfs_renamedata *data)
{
3657 3658 3659 3660
	nfs4_setup_sequence(NFS_SERVER(data->old_dir),
			&data->args.seq_args,
			&data->res.seq_res,
			task);
3661 3662 3663 3664 3665
}

static int nfs4_proc_rename_done(struct rpc_task *task, struct inode *old_dir,
				 struct inode *new_dir)
{
3666 3667
	struct nfs_renamedata *data = task->tk_calldata;
	struct nfs_renameres *res = &data->res;
3668 3669 3670

	if (!nfs4_sequence_done(task, &res->seq_res))
		return 0;
3671
	if (nfs4_async_handle_error(task, res->server, NULL, &data->timeout) == -EAGAIN)
3672 3673 3674 3675 3676 3677 3678
		return 0;

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

L
Linus Torvalds 已提交
3679 3680
static int _nfs4_proc_link(struct inode *inode, struct inode *dir, struct qstr *name)
{
3681
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
3682 3683 3684 3685
	struct nfs4_link_arg arg = {
		.fh     = NFS_FH(inode),
		.dir_fh = NFS_FH(dir),
		.name   = name,
3686 3687 3688 3689
		.bitmask = server->attr_bitmask,
	};
	struct nfs4_link_res res = {
		.server = server,
3690
		.label = NULL,
L
Linus Torvalds 已提交
3691 3692 3693 3694
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LINK],
		.rpc_argp = &arg,
3695
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3696
	};
3697 3698 3699
	int status = -ENOMEM;

	res.fattr = nfs_alloc_fattr();
3700
	if (res.fattr == NULL)
3701
		goto out;
L
Linus Torvalds 已提交
3702

3703 3704 3705 3706 3707
	res.label = nfs4_label_alloc(server, GFP_KERNEL);
	if (IS_ERR(res.label)) {
		status = PTR_ERR(res.label);
		goto out;
	}
3708
	arg.bitmask = nfs4_bitmask(server, res.label);
3709

3710
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
3711 3712
	if (!status) {
		update_changeattr(dir, &res.cinfo);
3713 3714 3715
		status = nfs_post_op_update_inode(inode, res.fattr);
		if (!status)
			nfs_setsecurity(inode, res.fattr, res.label);
3716
	}
3717 3718 3719 3720


	nfs4_label_free(res.label);

3721 3722
out:
	nfs_free_fattr(res.fattr);
L
Linus Torvalds 已提交
3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737
	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;
}

3738 3739 3740 3741 3742 3743
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;
3744
	struct nfs4_label *label;
3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755
};

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);

3756 3757 3758 3759
		data->label = nfs4_label_alloc(server, GFP_KERNEL);
		if (IS_ERR(data->label))
			goto out_free;

3760 3761 3762 3763 3764 3765 3766 3767
		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;
3768
		data->arg.bitmask = nfs4_bitmask(server, data->label);
3769 3770 3771
		data->res.server = server;
		data->res.fh = &data->fh;
		data->res.fattr = &data->fattr;
3772
		data->res.label = data->label;
3773 3774 3775
		nfs_fattr_init(data->res.fattr);
	}
	return data;
3776 3777 3778
out_free:
	kfree(data);
	return NULL;
3779 3780 3781 3782
}

static int nfs4_do_create(struct inode *dir, struct dentry *dentry, struct nfs4_createdata *data)
{
3783
	int status = nfs4_call_sync(NFS_SERVER(dir)->client, NFS_SERVER(dir), &data->msg,
3784
				    &data->arg.seq_args, &data->res.seq_res, 1);
3785 3786
	if (status == 0) {
		update_changeattr(dir, &data->res.dir_cinfo);
3787
		status = nfs_instantiate(dentry, data->res.fh, data->res.fattr, data->res.label);
3788 3789 3790 3791 3792 3793
	}
	return status;
}

static void nfs4_free_createdata(struct nfs4_createdata *data)
{
3794
	nfs4_label_free(data->label);
3795 3796 3797
	kfree(data);
}

3798
static int _nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
3799 3800
		struct page *page, unsigned int len, struct iattr *sattr,
		struct nfs4_label *label)
L
Linus Torvalds 已提交
3801
{
3802 3803
	struct nfs4_createdata *data;
	int status = -ENAMETOOLONG;
L
Linus Torvalds 已提交
3804

3805
	if (len > NFS4_MAXPATHLEN)
3806
		goto out;
3807

3808 3809 3810 3811 3812 3813 3814 3815
	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;
3816
	data->arg.label = label;
L
Linus Torvalds 已提交
3817
	
3818 3819 3820 3821
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
3822 3823 3824
	return status;
}

3825
static int nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
3826
		struct page *page, unsigned int len, struct iattr *sattr)
L
Linus Torvalds 已提交
3827 3828
{
	struct nfs4_exception exception = { };
3829
	struct nfs4_label l, *label = NULL;
L
Linus Torvalds 已提交
3830
	int err;
3831 3832 3833

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

L
Linus Torvalds 已提交
3834
	do {
3835 3836 3837
		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
3838 3839
				&exception);
	} while (exception.retry);
3840 3841

	nfs4_label_release_security(label);
L
Linus Torvalds 已提交
3842 3843 3844 3845
	return err;
}

static int _nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
3846
		struct iattr *sattr, struct nfs4_label *label)
L
Linus Torvalds 已提交
3847
{
3848 3849
	struct nfs4_createdata *data;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
3850

3851 3852 3853 3854
	data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4DIR);
	if (data == NULL)
		goto out;

3855
	data->arg.label = label;
3856 3857 3858 3859
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
3860 3861 3862 3863 3864 3865 3866
	return status;
}

static int nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr)
{
	struct nfs4_exception exception = { };
3867
	struct nfs4_label l, *label = NULL;
L
Linus Torvalds 已提交
3868
	int err;
A
Aneesh Kumar K.V 已提交
3869

3870 3871
	label = nfs4_label_init_security(dir, dentry, sattr, &l);

A
Aneesh Kumar K.V 已提交
3872
	sattr->ia_mode &= ~current_umask();
L
Linus Torvalds 已提交
3873
	do {
3874 3875 3876
		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
3877 3878
				&exception);
	} while (exception.retry);
3879 3880
	nfs4_label_release_security(label);

L
Linus Torvalds 已提交
3881 3882 3883 3884
	return err;
}

static int _nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
B
Bryan Schumaker 已提交
3885
		u64 cookie, struct page **pages, unsigned int count, int plus)
L
Linus Torvalds 已提交
3886
{
3887
	struct inode		*dir = d_inode(dentry);
L
Linus Torvalds 已提交
3888 3889
	struct nfs4_readdir_arg args = {
		.fh = NFS_FH(dir),
B
Bryan Schumaker 已提交
3890
		.pages = pages,
L
Linus Torvalds 已提交
3891 3892
		.pgbase = 0,
		.count = count,
3893
		.bitmask = NFS_SERVER(d_inode(dentry))->attr_bitmask,
B
Bryan Schumaker 已提交
3894
		.plus = plus,
L
Linus Torvalds 已提交
3895 3896 3897 3898 3899 3900 3901 3902 3903 3904
	};
	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;

3905 3906
	dprintk("%s: dentry = %pd2, cookie = %Lu\n", __func__,
			dentry,
3907
			(unsigned long long)cookie);
3908
	nfs4_setup_readdir(cookie, NFS_I(dir)->cookieverf, dentry, &args);
L
Linus Torvalds 已提交
3909
	res.pgbase = args.pgbase;
3910
	status = nfs4_call_sync(NFS_SERVER(dir)->client, NFS_SERVER(dir), &msg, &args.seq_args, &res.seq_res, 0);
3911
	if (status >= 0) {
3912
		memcpy(NFS_I(dir)->cookieverf, res.verifier.data, NFS4_VERIFIER_SIZE);
3913 3914
		status += args.pgbase;
	}
3915 3916 3917

	nfs_invalidate_atime(dir);

3918
	dprintk("%s: returns %d\n", __func__, status);
L
Linus Torvalds 已提交
3919 3920 3921 3922
	return status;
}

static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
B
Bryan Schumaker 已提交
3923
		u64 cookie, struct page **pages, unsigned int count, int plus)
L
Linus Torvalds 已提交
3924 3925 3926 3927
{
	struct nfs4_exception exception = { };
	int err;
	do {
3928 3929
		err = _nfs4_proc_readdir(dentry, cred, cookie,
				pages, count, plus);
3930 3931
		trace_nfs4_readdir(d_inode(dentry), err);
		err = nfs4_handle_exception(NFS_SERVER(d_inode(dentry)), err,
L
Linus Torvalds 已提交
3932 3933 3934 3935 3936 3937
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
3938
		struct iattr *sattr, struct nfs4_label *label, dev_t rdev)
L
Linus Torvalds 已提交
3939
{
3940 3941 3942
	struct nfs4_createdata *data;
	int mode = sattr->ia_mode;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
3943

3944 3945 3946 3947
	data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4SOCK);
	if (data == NULL)
		goto out;

L
Linus Torvalds 已提交
3948
	if (S_ISFIFO(mode))
3949
		data->arg.ftype = NF4FIFO;
L
Linus Torvalds 已提交
3950
	else if (S_ISBLK(mode)) {
3951 3952 3953
		data->arg.ftype = NF4BLK;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
L
Linus Torvalds 已提交
3954 3955
	}
	else if (S_ISCHR(mode)) {
3956 3957 3958
		data->arg.ftype = NF4CHR;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
3959 3960 3961
	} else if (!S_ISSOCK(mode)) {
		status = -EINVAL;
		goto out_free;
L
Linus Torvalds 已提交
3962
	}
3963

3964
	data->arg.label = label;
3965
	status = nfs4_do_create(dir, dentry, data);
3966
out_free:
3967 3968
	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
3969 3970 3971 3972 3973 3974 3975
	return status;
}

static int nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr, dev_t rdev)
{
	struct nfs4_exception exception = { };
3976
	struct nfs4_label l, *label = NULL;
L
Linus Torvalds 已提交
3977
	int err;
A
Aneesh Kumar K.V 已提交
3978

3979 3980
	label = nfs4_label_init_security(dir, dentry, sattr, &l);

A
Aneesh Kumar K.V 已提交
3981
	sattr->ia_mode &= ~current_umask();
L
Linus Torvalds 已提交
3982
	do {
3983 3984 3985
		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
3986 3987
				&exception);
	} while (exception.retry);
3988 3989 3990

	nfs4_label_release_security(label);

L
Linus Torvalds 已提交
3991 3992 3993 3994 3995 3996 3997 3998 3999 4000
	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
4001 4002 4003
	struct nfs4_statfs_res res = {
		.fsstat = fsstat,
	};
L
Linus Torvalds 已提交
4004 4005 4006
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_STATFS],
		.rpc_argp = &args,
B
Benny Halevy 已提交
4007
		.rpc_resp = &res,
L
Linus Torvalds 已提交
4008 4009
	};

4010
	nfs_fattr_init(fsstat->fattr);
4011
	return  nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032
}

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
4033 4034 4035
	struct nfs4_fsinfo_res res = {
		.fsinfo = fsinfo,
	};
L
Linus Torvalds 已提交
4036 4037 4038
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FSINFO],
		.rpc_argp = &args,
B
Benny Halevy 已提交
4039
		.rpc_resp = &res,
L
Linus Torvalds 已提交
4040 4041
	};

4042
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
4043 4044 4045 4046 4047
}

static int nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo)
{
	struct nfs4_exception exception = { };
4048
	unsigned long now = jiffies;
L
Linus Torvalds 已提交
4049 4050 4051
	int err;

	do {
4052
		err = _nfs4_do_fsinfo(server, fhandle, fsinfo);
4053
		trace_nfs4_fsinfo(server, fhandle, fsinfo->fattr, err);
4054 4055 4056 4057 4058 4059 4060 4061 4062 4063
		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
4064 4065 4066 4067 4068 4069
	} while (exception.retry);
	return err;
}

static int nfs4_proc_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo)
{
4070 4071
	int error;

4072
	nfs_fattr_init(fsinfo->fattr);
4073
	error = nfs4_do_fsinfo(server, fhandle, fsinfo);
4074 4075 4076
	if (error == 0) {
		/* block layout checks this! */
		server->pnfs_blksize = fsinfo->blksize;
4077
		set_pnfs_layoutdriver(server, fhandle, fsinfo->layouttype);
4078
	}
4079 4080

	return error;
L
Linus Torvalds 已提交
4081 4082 4083 4084 4085 4086 4087 4088 4089
}

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
4090 4091 4092
	struct nfs4_pathconf_res res = {
		.pathconf = pathconf,
	};
L
Linus Torvalds 已提交
4093 4094 4095
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_PATHCONF],
		.rpc_argp = &args,
B
Benny Halevy 已提交
4096
		.rpc_resp = &res,
L
Linus Torvalds 已提交
4097 4098 4099 4100 4101 4102 4103 4104
	};

	/* 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;
	}

4105
	nfs_fattr_init(pathconf->fattr);
4106
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122
}

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;
}

4123
int nfs4_set_rw_stateid(nfs4_stateid *stateid,
4124 4125 4126 4127 4128 4129 4130 4131
		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;
4132
	return nfs4_select_rw_stateid(stateid, ctx->state, fmode, lockowner);
4133 4134 4135
}
EXPORT_SYMBOL_GPL(nfs4_set_rw_stateid);

4136 4137 4138 4139 4140 4141 4142
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;

4143 4144 4145
	/* If the current stateid represents a lost lock, then exit */
	if (nfs4_set_rw_stateid(&current_stateid, ctx, l_ctx, fmode) == -EIO)
		return true;
4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163
	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;
}

4164
void __nfs4_read_done_cb(struct nfs_pgio_header *hdr)
4165
{
4166
	nfs_invalidate_atime(hdr->inode);
4167 4168
}

4169
static int nfs4_read_done_cb(struct rpc_task *task, struct nfs_pgio_header *hdr)
L
Linus Torvalds 已提交
4170
{
4171
	struct nfs_server *server = NFS_SERVER(hdr->inode);
L
Linus Torvalds 已提交
4172

4173 4174
	trace_nfs4_read(hdr, task->tk_status);
	if (nfs4_async_handle_error(task, server,
4175 4176
				    hdr->args.context->state,
				    NULL) == -EAGAIN) {
4177
		rpc_restart_call_prepare(task);
T
Trond Myklebust 已提交
4178
		return -EAGAIN;
L
Linus Torvalds 已提交
4179
	}
4180

4181
	__nfs4_read_done_cb(hdr);
L
Linus Torvalds 已提交
4182
	if (task->tk_status > 0)
4183
		renew_lease(server, hdr->timestamp);
T
Trond Myklebust 已提交
4184
	return 0;
L
Linus Torvalds 已提交
4185 4186
}

4187
static bool nfs4_read_stateid_changed(struct rpc_task *task,
4188
		struct nfs_pgio_args *args)
4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199 4200
{

	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;
}

4201
static int nfs4_read_done(struct rpc_task *task, struct nfs_pgio_header *hdr)
4202 4203 4204 4205
{

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

4206
	if (!nfs4_sequence_done(task, &hdr->res.seq_res))
4207
		return -EAGAIN;
4208
	if (nfs4_read_stateid_changed(task, &hdr->args))
4209
		return -EAGAIN;
4210 4211
	return hdr->pgio_done_cb ? hdr->pgio_done_cb(task, hdr) :
				    nfs4_read_done_cb(task, hdr);
4212 4213
}

4214 4215
static void nfs4_proc_read_setup(struct nfs_pgio_header *hdr,
				 struct rpc_message *msg)
L
Linus Torvalds 已提交
4216
{
4217 4218
	hdr->timestamp   = jiffies;
	hdr->pgio_done_cb = nfs4_read_done_cb;
4219
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READ];
4220
	nfs4_init_sequence(&hdr->args.seq_args, &hdr->res.seq_res, 0);
L
Linus Torvalds 已提交
4221 4222
}

4223 4224
static int nfs4_proc_pgio_rpc_prepare(struct rpc_task *task,
				      struct nfs_pgio_header *hdr)
4225
{
4226 4227 4228
	if (nfs4_setup_sequence(NFS_SERVER(hdr->inode),
			&hdr->args.seq_args,
			&hdr->res.seq_res,
4229
			task))
4230
		return 0;
4231 4232 4233
	if (nfs4_set_rw_stateid(&hdr->args.stateid, hdr->args.context,
				hdr->args.lock_context,
				hdr->rw_ops->rw_mode) == -EIO)
4234
		return -EIO;
4235
	if (unlikely(test_bit(NFS_CONTEXT_BAD, &hdr->args.context->flags)))
4236 4237
		return -EIO;
	return 0;
L
Linus Torvalds 已提交
4238 4239
}

4240 4241
static int nfs4_write_done_cb(struct rpc_task *task,
			      struct nfs_pgio_header *hdr)
L
Linus Torvalds 已提交
4242
{
4243
	struct inode *inode = hdr->inode;
4244

4245 4246
	trace_nfs4_write(hdr, task->tk_status);
	if (nfs4_async_handle_error(task, NFS_SERVER(inode),
4247 4248
				    hdr->args.context->state,
				    NULL) == -EAGAIN) {
4249
		rpc_restart_call_prepare(task);
4250
		return -EAGAIN;
L
Linus Torvalds 已提交
4251
	}
4252
	if (task->tk_status >= 0) {
4253
		renew_lease(NFS_SERVER(inode), hdr->timestamp);
4254
		nfs_writeback_update_inode(hdr);
4255
	}
4256
	return 0;
L
Linus Torvalds 已提交
4257 4258
}

4259
static bool nfs4_write_stateid_changed(struct rpc_task *task,
4260
		struct nfs_pgio_args *args)
4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272
{

	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;
}

4273
static int nfs4_write_done(struct rpc_task *task, struct nfs_pgio_header *hdr)
4274
{
4275
	if (!nfs4_sequence_done(task, &hdr->res.seq_res))
4276
		return -EAGAIN;
4277
	if (nfs4_write_stateid_changed(task, &hdr->args))
4278
		return -EAGAIN;
4279 4280
	return hdr->pgio_done_cb ? hdr->pgio_done_cb(task, hdr) :
		nfs4_write_done_cb(task, hdr);
4281 4282
}

4283
static
4284
bool nfs4_write_need_cache_consistency_data(struct nfs_pgio_header *hdr)
4285
{
4286
	/* Don't request attributes for pNFS or O_DIRECT writes */
4287
	if (hdr->ds_clp != NULL || hdr->dreq != NULL)
4288 4289 4290 4291
		return false;
	/* Otherwise, request attributes if and only if we don't hold
	 * a delegation
	 */
4292
	return nfs4_have_delegation(hdr->inode, FMODE_READ) == 0;
4293 4294
}

4295 4296
static void nfs4_proc_write_setup(struct nfs_pgio_header *hdr,
				  struct rpc_message *msg)
L
Linus Torvalds 已提交
4297
{
4298
	struct nfs_server *server = NFS_SERVER(hdr->inode);
4299

4300 4301 4302
	if (!nfs4_write_need_cache_consistency_data(hdr)) {
		hdr->args.bitmask = NULL;
		hdr->res.fattr = NULL;
4303
	} else
4304
		hdr->args.bitmask = server->cache_consistency_bitmask;
4305

4306 4307 4308 4309
	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
4310

4311
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_WRITE];
4312
	nfs4_init_sequence(&hdr->args.seq_args, &hdr->res.seq_res, 1);
L
Linus Torvalds 已提交
4313 4314
}

4315
static void nfs4_proc_commit_rpc_prepare(struct rpc_task *task, struct nfs_commit_data *data)
4316
{
4317 4318 4319 4320
	nfs4_setup_sequence(NFS_SERVER(data->inode),
			&data->args.seq_args,
			&data->res.seq_res,
			task);
L
Linus Torvalds 已提交
4321 4322
}

4323
static int nfs4_commit_done_cb(struct rpc_task *task, struct nfs_commit_data *data)
L
Linus Torvalds 已提交
4324 4325
{
	struct inode *inode = data->inode;
4326

4327
	trace_nfs4_commit(data, task->tk_status);
4328 4329
	if (nfs4_async_handle_error(task, NFS_SERVER(inode),
				    NULL, NULL) == -EAGAIN) {
4330
		rpc_restart_call_prepare(task);
4331
		return -EAGAIN;
L
Linus Torvalds 已提交
4332
	}
4333
	return 0;
L
Linus Torvalds 已提交
4334 4335
}

4336
static int nfs4_commit_done(struct rpc_task *task, struct nfs_commit_data *data)
4337 4338 4339
{
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return -EAGAIN;
4340
	return data->commit_done_cb(task, data);
4341 4342
}

4343
static void nfs4_proc_commit_setup(struct nfs_commit_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
4344
{
4345
	struct nfs_server *server = NFS_SERVER(data->inode);
4346

4347 4348
	if (data->commit_done_cb == NULL)
		data->commit_done_cb = nfs4_commit_done_cb;
4349
	data->res.server = server;
4350
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_COMMIT];
4351
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
L
Linus Torvalds 已提交
4352 4353
}

4354 4355 4356 4357 4358
struct nfs4_renewdata {
	struct nfs_client	*client;
	unsigned long		timestamp;
};

L
Linus Torvalds 已提交
4359 4360 4361 4362
/*
 * nfs4_proc_async_renew(): This is not one of the nfs_rpc_ops; it is a special
 * standalone procedure for queueing an asynchronous RENEW.
 */
4363
static void nfs4_renew_release(void *calldata)
4364
{
4365 4366
	struct nfs4_renewdata *data = calldata;
	struct nfs_client *clp = data->client;
4367

4368 4369 4370
	if (atomic_read(&clp->cl_count) > 1)
		nfs4_schedule_state_renewal(clp);
	nfs_put_client(clp);
4371
	kfree(data);
4372 4373
}

4374
static void nfs4_renew_done(struct rpc_task *task, void *calldata)
L
Linus Torvalds 已提交
4375
{
4376 4377 4378
	struct nfs4_renewdata *data = calldata;
	struct nfs_client *clp = data->client;
	unsigned long timestamp = data->timestamp;
L
Linus Torvalds 已提交
4379

4380
	trace_nfs4_renew_async(clp, task->tk_status);
4381 4382 4383 4384 4385 4386 4387
	switch (task->tk_status) {
	case 0:
		break;
	case -NFS4ERR_LEASE_MOVED:
		nfs4_schedule_lease_moved_recovery(clp);
		break;
	default:
4388
		/* Unless we're shutting down, schedule state recovery! */
4389 4390 4391
		if (test_bit(NFS_CS_RENEWD, &clp->cl_res_state) == 0)
			return;
		if (task->tk_status != NFS4ERR_CB_PATH_DOWN) {
4392
			nfs4_schedule_lease_recovery(clp);
4393 4394 4395
			return;
		}
		nfs4_schedule_path_down_recovery(clp);
L
Linus Torvalds 已提交
4396
	}
4397
	do_renew_lease(clp, timestamp);
L
Linus Torvalds 已提交
4398 4399
}

4400 4401
static const struct rpc_call_ops nfs4_renew_ops = {
	.rpc_call_done = nfs4_renew_done,
4402
	.rpc_release = nfs4_renew_release,
4403 4404
};

4405
static int nfs4_proc_async_renew(struct nfs_client *clp, struct rpc_cred *cred, unsigned renew_flags)
L
Linus Torvalds 已提交
4406 4407 4408 4409
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
4410
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
4411
	};
4412
	struct nfs4_renewdata *data;
L
Linus Torvalds 已提交
4413

4414 4415
	if (renew_flags == 0)
		return 0;
4416 4417
	if (!atomic_inc_not_zero(&clp->cl_count))
		return -EIO;
4418
	data = kmalloc(sizeof(*data), GFP_NOFS);
4419 4420 4421 4422
	if (data == NULL)
		return -ENOMEM;
	data->client = clp;
	data->timestamp = jiffies;
4423
	return rpc_call_async(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT,
4424
			&nfs4_renew_ops, data);
L
Linus Torvalds 已提交
4425 4426
}

4427
static int nfs4_proc_renew(struct nfs_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
4428 4429 4430 4431
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
4432
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
4433 4434 4435 4436
	};
	unsigned long now = jiffies;
	int status;

4437
	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
L
Linus Torvalds 已提交
4438 4439
	if (status < 0)
		return status;
4440
	do_renew_lease(clp, now);
L
Linus Torvalds 已提交
4441 4442 4443
	return 0;
}

4444 4445
static inline int nfs4_server_supports_acls(struct nfs_server *server)
{
4446
	return server->caps & NFS_CAP_ACLS;
4447 4448
}

4449 4450
/* 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
4451 4452
 * the stack.
 */
4453
#define NFS4ACL_MAXPAGES DIV_ROUND_UP(XATTR_SIZE_MAX, PAGE_SIZE)
4454

4455 4456 4457 4458 4459 4460 4461 4462 4463
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 {
4464
		len = min_t(size_t, PAGE_SIZE, buflen);
4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483
		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;
}

4484 4485 4486
struct nfs4_cached_acl {
	int cached;
	size_t len;
A
Andrew Morton 已提交
4487
	char data[0];
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 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529
};

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;
}

4530
static void nfs4_write_cached_acl(struct inode *inode, struct page **pages, size_t pgbase, size_t acl_len)
4531 4532
{
	struct nfs4_cached_acl *acl;
4533
	size_t buflen = sizeof(*acl) + acl_len;
4534

4535
	if (buflen <= PAGE_SIZE) {
4536
		acl = kmalloc(buflen, GFP_KERNEL);
4537 4538 4539
		if (acl == NULL)
			goto out;
		acl->cached = 1;
4540
		_copy_from_pages(acl->data, pages, pgbase, acl_len);
4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551
	} 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);
}

4552 4553 4554 4555 4556 4557 4558 4559 4560 4561
/*
 * 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.
 */
4562
static ssize_t __nfs4_get_acl_uncached(struct inode *inode, void *buf, size_t buflen)
4563
{
4564
	struct page *pages[NFS4ACL_MAXPAGES] = {NULL, };
4565 4566 4567 4568 4569
	struct nfs_getaclargs args = {
		.fh = NFS_FH(inode),
		.acl_pages = pages,
		.acl_len = buflen,
	};
B
Benny Halevy 已提交
4570 4571 4572
	struct nfs_getaclres res = {
		.acl_len = buflen,
	};
4573 4574 4575
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETACL],
		.rpc_argp = &args,
B
Benny Halevy 已提交
4576
		.rpc_resp = &res,
4577
	};
4578 4579
	unsigned int npages = DIV_ROUND_UP(buflen, PAGE_SIZE);
	int ret = -ENOMEM, i;
4580

4581 4582 4583 4584
	/* 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;
4585 4586
	if (npages > ARRAY_SIZE(pages))
		return -ERANGE;
4587

4588 4589 4590 4591
	for (i = 0; i < npages; i++) {
		pages[i] = alloc_page(GFP_KERNEL);
		if (!pages[i])
			goto out_free;
4592
	}
4593 4594 4595 4596 4597 4598

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

4599 4600
	args.acl_len = npages * PAGE_SIZE;
	args.acl_pgbase = 0;
4601

4602
	dprintk("%s  buf %p buflen %zu npages %d args.acl_len %zu\n",
4603 4604 4605
		__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);
4606 4607
	if (ret)
		goto out_free;
4608

4609 4610 4611 4612 4613
	/* 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;
4614
		ret = -ERANGE;
4615
		goto out_free;
4616
	}
4617
	nfs4_write_cached_acl(inode, pages, res.acl_data_offset, res.acl_len);
4618 4619 4620 4621 4622
	if (buf) {
		if (res.acl_len > buflen) {
			ret = -ERANGE;
			goto out_free;
		}
4623
		_copy_from_pages(buf, pages, res.acl_data_offset, res.acl_len);
4624
	}
4625 4626
out_ok:
	ret = res.acl_len;
4627
out_free:
4628 4629 4630
	for (i = 0; i < npages; i++)
		if (pages[i])
			__free_page(pages[i]);
4631 4632
	if (res.acl_scratch)
		__free_page(res.acl_scratch);
4633 4634 4635
	return ret;
}

4636 4637 4638 4639 4640 4641
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);
4642
		trace_nfs4_get_acl(inode, ret);
4643 4644 4645 4646 4647 4648 4649
		if (ret >= 0)
			break;
		ret = nfs4_handle_exception(NFS_SERVER(inode), ret, &exception);
	} while (exception.retry);
	return ret;
}

4650 4651 4652 4653 4654 4655 4656 4657 4658 4659
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;
4660 4661
	if (NFS_I(inode)->cache_validity & NFS_INO_INVALID_ACL)
		nfs_zap_acl_cache(inode);
4662 4663
	ret = nfs4_read_cached_acl(inode, buf, buflen);
	if (ret != -ENOENT)
4664 4665
		/* -ENOENT is returned if there is no ACL or if there is an ACL
		 * but no cached acl data, just the acl length */
4666 4667 4668 4669
		return ret;
	return nfs4_get_acl_uncached(inode, buf, buflen);
}

4670
static int __nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen)
4671 4672 4673 4674 4675 4676 4677 4678
{
	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
4679
	struct nfs_setaclres res;
4680 4681 4682
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETACL],
		.rpc_argp	= &arg,
B
Benny Halevy 已提交
4683
		.rpc_resp	= &res,
4684
	};
4685
	unsigned int npages = DIV_ROUND_UP(buflen, PAGE_SIZE);
4686
	int ret, i;
4687 4688 4689

	if (!nfs4_server_supports_acls(server))
		return -EOPNOTSUPP;
4690 4691
	if (npages > ARRAY_SIZE(pages))
		return -ERANGE;
4692 4693 4694
	i = buf_to_pages_noslab(buf, buflen, arg.acl_pages, &arg.acl_pgbase);
	if (i < 0)
		return i;
4695
	nfs4_inode_return_delegation(inode);
4696
	ret = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
4697 4698 4699 4700 4701 4702 4703 4704

	/*
	 * 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]);

4705 4706 4707 4708 4709 4710 4711
	/*
	 * 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);
4712 4713
	nfs_access_zap_cache(inode);
	nfs_zap_acl_cache(inode);
4714 4715 4716
	return ret;
}

4717 4718 4719 4720 4721
static int nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen)
{
	struct nfs4_exception exception = { };
	int err;
	do {
4722 4723 4724
		err = __nfs4_proc_set_acl(inode, buf, buflen);
		trace_nfs4_set_acl(inode, err);
		err = nfs4_handle_exception(NFS_SERVER(inode), err,
4725 4726 4727 4728 4729
				&exception);
	} while (exception.retry);
	return err;
}

4730 4731 4732 4733 4734 4735 4736 4737 4738
#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 };
4739
	struct nfs4_getattr_arg arg = {
4740 4741 4742 4743 4744 4745 4746 4747 4748 4749
		.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],
4750
		.rpc_argp	= &arg,
4751 4752 4753 4754 4755 4756
		.rpc_resp	= &res,
	};
	int ret;

	nfs_fattr_init(&fattr);

4757
	ret = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 0);
4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776
	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 {
4777 4778 4779
		err = _nfs4_get_security_label(inode, buf, buflen);
		trace_nfs4_get_security_label(inode, err);
		err = nfs4_handle_exception(NFS_SERVER(inode), err,
4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793
				&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 };
4794
	struct nfs_setattrargs arg = {
4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807
		.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],
4808
		.rpc_argp       = &arg,
4809 4810 4811 4812
		.rpc_resp       = &res,
	};
	int status;

4813
	nfs4_stateid_copy(&arg.stateid, &zero_stateid);
4814

4815
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830
	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 {
4831 4832 4833 4834
		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,
4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845
				&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;
4846
	struct inode *inode = d_inode(dentry);
4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880
	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
4881
static int
4882 4883
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
4884
{
4885 4886 4887
	struct nfs_client *clp = server->nfs_client;

	if (task->tk_status >= 0)
L
Linus Torvalds 已提交
4888 4889
		return 0;
	switch(task->tk_status) {
4890
		case -NFS4ERR_DELEG_REVOKED:
4891 4892 4893 4894 4895
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OPENMODE:
			if (state == NULL)
				break;
4896
			if (nfs4_schedule_stateid_recovery(server, state) < 0)
C
Chuck Lever 已提交
4897
				goto recovery_failed;
4898
			goto wait_on_recovery;
4899
		case -NFS4ERR_EXPIRED:
4900 4901
			if (state != NULL) {
				if (nfs4_schedule_stateid_recovery(server, state) < 0)
C
Chuck Lever 已提交
4902
					goto recovery_failed;
4903
			}
L
Linus Torvalds 已提交
4904
		case -NFS4ERR_STALE_STATEID:
4905
		case -NFS4ERR_STALE_CLIENTID:
4906 4907
			nfs4_schedule_lease_recovery(clp);
			goto wait_on_recovery;
4908 4909 4910 4911
		case -NFS4ERR_MOVED:
			if (nfs4_schedule_migration_recovery(server) < 0)
				goto recovery_failed;
			goto wait_on_recovery;
4912 4913 4914
		case -NFS4ERR_LEASE_MOVED:
			nfs4_schedule_lease_moved_recovery(clp);
			goto wait_on_recovery;
4915 4916 4917 4918 4919 4920 4921 4922 4923 4924
#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);
4925
			nfs4_schedule_session_recovery(clp->cl_session, task->tk_status);
4926
			goto wait_on_recovery;
4927
#endif /* CONFIG_NFS_V4_1 */
L
Linus Torvalds 已提交
4928
		case -NFS4ERR_DELAY:
4929
			nfs_inc_server_stats(server, NFSIOS_DELAY);
4930 4931
			rpc_delay(task, nfs4_update_delay(timeout));
			goto restart_call;
4932
		case -NFS4ERR_GRACE:
L
Linus Torvalds 已提交
4933
			rpc_delay(task, NFS4_POLL_RETRY_MAX);
4934
		case -NFS4ERR_RETRY_UNCACHED_REP:
L
Linus Torvalds 已提交
4935
		case -NFS4ERR_OLD_STATEID:
4936
			goto restart_call;
L
Linus Torvalds 已提交
4937 4938 4939
	}
	task->tk_status = nfs4_map_errors(task->tk_status);
	return 0;
C
Chuck Lever 已提交
4940
recovery_failed:
4941 4942
	task->tk_status = -EIO;
	return 0;
4943
wait_on_recovery:
4944 4945 4946
	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);
4947 4948
	if (test_bit(NFS_MIG_FAILED, &server->mig_status))
		goto recovery_failed;
4949
restart_call:
4950 4951
	task->tk_status = 0;
	return -EAGAIN;
L
Linus Torvalds 已提交
4952 4953
}

4954 4955
static void nfs4_init_boot_verifier(const struct nfs_client *clp,
				    nfs4_verifier *bootverf)
4956 4957 4958
{
	__be32 verf[2];

4959 4960 4961 4962
	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;
4963
		verf[1] = cpu_to_be32(NSEC_PER_SEC + 1);
4964
	} else {
4965
		struct nfs_net *nn = net_generic(clp->cl_net, nfs_net_id);
4966 4967
		verf[0] = cpu_to_be32(nn->boot_time.tv_sec);
		verf[1] = cpu_to_be32(nn->boot_time.tv_nsec);
4968
	}
4969 4970 4971
	memcpy(bootverf->data, verf, sizeof(bootverf->data));
}

4972 4973
static int
nfs4_init_nonuniform_client_string(struct nfs_client *clp)
4974
{
4975 4976 4977 4978
	int result;
	size_t len;
	char *str;
	bool retried = false;
4979

4980
	if (clp->cl_owner_id != NULL)
4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001
		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;
5002

5003
	rcu_read_lock();
5004 5005 5006 5007
	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));
5008
	rcu_read_unlock();
5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019

	/* Did something change? */
	if (result >= len) {
		kfree(str);
		if (retried)
			return -EINVAL;
		retried = true;
		goto retry;
	}
	clp->cl_owner_id = str;
	return 0;
5020 5021
}

5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058
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)
5059
{
5060 5061 5062
	int result;
	size_t len;
	char *str;
5063 5064

	if (clp->cl_owner_id != NULL)
5065
		return 0;
5066 5067

	if (nfs4_client_id_uniquifier[0] != '\0')
5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093
		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;
5094 5095
}

5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109
/*
 * 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");
}

5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121
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,
};

5122 5123 5124 5125 5126 5127 5128 5129 5130 5131
/**
 * 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.
 */
5132 5133 5134
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
5135 5136 5137 5138 5139
{
	nfs4_verifier sc_verifier;
	struct nfs4_setclientid setclientid = {
		.sc_verifier = &sc_verifier,
		.sc_prog = program,
5140
		.sc_clnt = clp,
L
Linus Torvalds 已提交
5141 5142 5143 5144
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID],
		.rpc_argp = &setclientid,
5145
		.rpc_resp = res,
5146
		.rpc_cred = cred,
L
Linus Torvalds 已提交
5147
	};
5148 5149 5150 5151 5152 5153 5154 5155
	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,
	};
5156
	int status;
L
Linus Torvalds 已提交
5157

5158
	/* nfs_client_id4 */
5159
	nfs4_init_boot_verifier(clp, &sc_verifier);
5160 5161 5162 5163

	if (test_bit(NFS_CS_MIGRATION, &clp->cl_flags))
		status = nfs4_init_uniform_client_string(clp);
	else
5164
		status = nfs4_init_nonuniform_client_string(clp);
5165 5166 5167

	if (status)
		goto out;
5168

5169
	/* cb_client4 */
5170 5171 5172 5173
	setclientid.sc_netid_len =
				nfs4_init_callback_netid(clp,
						setclientid.sc_netid,
						sizeof(setclientid.sc_netid));
5174
	setclientid.sc_uaddr_len = scnprintf(setclientid.sc_uaddr,
5175
				sizeof(setclientid.sc_uaddr), "%s.%u.%u",
L
Linus Torvalds 已提交
5176 5177
				clp->cl_ipaddr, port >> 8, port & 255);

5178
	dprintk("NFS call  setclientid auth=%s, '%s'\n",
5179
		clp->cl_rpcclient->cl_auth->au_ops->au_name,
5180
		clp->cl_owner_id);
5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192
	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:
5193
	trace_nfs4_setclientid(clp, status);
5194 5195
	dprintk("NFS reply setclientid: %d\n", status);
	return status;
L
Linus Torvalds 已提交
5196 5197
}

5198 5199 5200 5201 5202 5203 5204 5205
/**
 * 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.
 */
5206
int nfs4_proc_setclientid_confirm(struct nfs_client *clp,
5207 5208
		struct nfs4_setclientid_res *arg,
		struct rpc_cred *cred)
L
Linus Torvalds 已提交
5209 5210 5211
{
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID_CONFIRM],
5212
		.rpc_argp = arg,
5213
		.rpc_cred = cred,
L
Linus Torvalds 已提交
5214 5215 5216
	};
	int status;

5217 5218 5219
	dprintk("NFS call  setclientid_confirm auth=%s, (client ID %llx)\n",
		clp->cl_rpcclient->cl_auth->au_ops->au_name,
		clp->cl_clientid);
5220
	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
5221
	trace_nfs4_setclientid_confirm(clp, status);
5222
	dprintk("NFS reply setclientid_confirm: %d\n", status);
L
Linus Torvalds 已提交
5223 5224 5225
	return status;
}

5226 5227
struct nfs4_delegreturndata {
	struct nfs4_delegreturnargs args;
5228
	struct nfs4_delegreturnres res;
5229 5230
	struct nfs_fh fh;
	nfs4_stateid stateid;
5231
	unsigned long timestamp;
5232
	struct nfs_fattr fattr;
5233
	int rpc_status;
5234 5235 5236
	struct inode *inode;
	bool roc;
	u32 roc_barrier;
5237 5238 5239 5240 5241
};

static void nfs4_delegreturn_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_delegreturndata *data = calldata;
5242

5243 5244
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return;
5245

5246
	trace_nfs4_delegreturn_exit(&data->args, &data->res, task->tk_status);
5247 5248
	switch (task->tk_status) {
	case 0:
5249
		renew_lease(data->res.server, data->timestamp);
5250 5251 5252 5253 5254 5255 5256
	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;
5257 5258
		if (data->roc)
			pnfs_roc_set_barrier(data->inode, data->roc_barrier);
5259
		break;
5260
	default:
5261 5262
		if (nfs4_async_handle_error(task, data->res.server,
					    NULL, NULL) == -EAGAIN) {
5263
			rpc_restart_call_prepare(task);
5264 5265 5266 5267
			return;
		}
	}
	data->rpc_status = task->tk_status;
5268 5269 5270 5271
}

static void nfs4_delegreturn_release(void *calldata)
{
5272
	struct nfs4_delegreturndata *data = calldata;
5273
	struct inode *inode = data->inode;
5274

5275 5276 5277 5278 5279
	if (inode) {
		if (data->roc)
			pnfs_roc_release(inode);
		nfs_iput_and_deactive(inode);
	}
5280 5281 5282
	kfree(calldata);
}

5283 5284 5285 5286 5287 5288
static void nfs4_delegreturn_prepare(struct rpc_task *task, void *data)
{
	struct nfs4_delegreturndata *d_data;

	d_data = (struct nfs4_delegreturndata *)data;

5289 5290
	if (d_data->roc)
		pnfs_roc_get_barrier(d_data->inode, &d_data->roc_barrier);
5291

5292 5293 5294 5295
	nfs4_setup_sequence(d_data->res.server,
			&d_data->args.seq_args,
			&d_data->res.seq_res,
			task);
5296 5297
}

5298
static const struct rpc_call_ops nfs4_delegreturn_ops = {
5299
	.rpc_call_prepare = nfs4_delegreturn_prepare,
5300 5301 5302 5303
	.rpc_call_done = nfs4_delegreturn_done,
	.rpc_release = nfs4_delegreturn_release,
};

5304
static int _nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
5305 5306
{
	struct nfs4_delegreturndata *data;
5307
	struct nfs_server *server = NFS_SERVER(inode);
5308
	struct rpc_task *task;
5309 5310 5311 5312
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DELEGRETURN],
		.rpc_cred = cred,
	};
T
Trond Myklebust 已提交
5313 5314
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
5315
		.rpc_message = &msg,
T
Trond Myklebust 已提交
5316 5317 5318
		.callback_ops = &nfs4_delegreturn_ops,
		.flags = RPC_TASK_ASYNC,
	};
5319
	int status = 0;
5320

5321
	data = kzalloc(sizeof(*data), GFP_NOFS);
5322 5323
	if (data == NULL)
		return -ENOMEM;
5324
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
5325 5326
	data->args.fhandle = &data->fh;
	data->args.stateid = &data->stateid;
5327
	data->args.bitmask = server->cache_consistency_bitmask;
5328
	nfs_copy_fh(&data->fh, NFS_FH(inode));
5329
	nfs4_stateid_copy(&data->stateid, stateid);
5330 5331
	data->res.fattr = &data->fattr;
	data->res.server = server;
5332
	nfs_fattr_init(data->res.fattr);
5333
	data->timestamp = jiffies;
5334
	data->rpc_status = 0;
5335 5336 5337
	data->inode = nfs_igrab_and_active(inode);
	if (data->inode)
		data->roc = nfs4_roc(inode);
5338

T
Trond Myklebust 已提交
5339
	task_setup_data.callback_data = data;
5340 5341
	msg.rpc_argp = &data->args;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
5342
	task = rpc_run_task(&task_setup_data);
5343
	if (IS_ERR(task))
5344
		return PTR_ERR(task);
5345 5346
	if (!issync)
		goto out;
5347
	status = nfs4_wait_for_completion_rpc_task(task);
5348 5349 5350
	if (status != 0)
		goto out;
	status = data->rpc_status;
5351 5352 5353 5354
	if (status == 0)
		nfs_post_op_update_inode_force_wcc(inode, &data->fattr);
	else
		nfs_refresh_inode(inode, &data->fattr);
5355
out:
5356
	rpc_put_task(task);
5357
	return status;
L
Linus Torvalds 已提交
5358 5359
}

5360
int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
L
Linus Torvalds 已提交
5361 5362 5363 5364 5365
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs4_exception exception = { };
	int err;
	do {
5366
		err = _nfs4_proc_delegreturn(inode, cred, stateid, issync);
5367
		trace_nfs4_delegreturn(inode, err);
L
Linus Torvalds 已提交
5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387
		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)
{
5388
	freezable_schedule_timeout_killable_unsafe(timeout);
L
Linus Torvalds 已提交
5389 5390 5391 5392 5393 5394 5395 5396 5397 5398
	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);
5399
	struct nfs_client *clp = server->nfs_client;
T
Trond Myklebust 已提交
5400
	struct nfs_lockt_args arg = {
L
Linus Torvalds 已提交
5401
		.fh = NFS_FH(inode),
T
Trond Myklebust 已提交
5402
		.fl = request,
L
Linus Torvalds 已提交
5403
	};
T
Trond Myklebust 已提交
5404 5405
	struct nfs_lockt_res res = {
		.denied = request,
L
Linus Torvalds 已提交
5406 5407 5408 5409 5410 5411 5412 5413 5414 5415
	};
	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
5416
	arg.lock_owner.clientid = clp->cl_clientid;
5417 5418 5419 5420
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
	lsp = request->fl_u.nfs4_fl.owner;
5421
	arg.lock_owner.id = lsp->ls_seqid.owner_id;
5422
	arg.lock_owner.s_dev = server->s_dev;
5423
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
T
Trond Myklebust 已提交
5424 5425 5426 5427 5428 5429
	switch (status) {
		case 0:
			request->fl_type = F_UNLCK;
			break;
		case -NFS4ERR_DENIED:
			status = 0;
L
Linus Torvalds 已提交
5430
	}
5431
	request->fl_ops->fl_release_private(request);
5432
	request->fl_ops = NULL;
5433
out:
L
Linus Torvalds 已提交
5434 5435 5436 5437 5438 5439 5440 5441 5442
	return status;
}

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

	do {
5443 5444 5445
		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
5446 5447 5448 5449 5450
				&exception);
	} while (exception.retry);
	return err;
}

5451
static int do_vfs_lock(struct inode *inode, struct file_lock *fl)
L
Linus Torvalds 已提交
5452 5453 5454 5455
{
	int res = 0;
	switch (fl->fl_flags & (FL_POSIX|FL_FLOCK)) {
		case FL_POSIX:
5456
			res = posix_lock_inode_wait(inode, fl);
L
Linus Torvalds 已提交
5457 5458
			break;
		case FL_FLOCK:
5459
			res = flock_lock_inode_wait(inode, fl);
L
Linus Torvalds 已提交
5460 5461 5462 5463 5464 5465 5466
			break;
		default:
			BUG();
	}
	return res;
}

5467
struct nfs4_unlockdata {
T
Trond Myklebust 已提交
5468 5469
	struct nfs_locku_args arg;
	struct nfs_locku_res res;
5470 5471
	struct nfs4_lock_state *lsp;
	struct nfs_open_context *ctx;
T
Trond Myklebust 已提交
5472 5473
	struct file_lock fl;
	const struct nfs_server *server;
5474
	unsigned long timestamp;
5475 5476
};

T
Trond Myklebust 已提交
5477 5478 5479 5480 5481 5482 5483 5484
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;

5485
	p = kzalloc(sizeof(*p), GFP_NOFS);
T
Trond Myklebust 已提交
5486 5487 5488 5489 5490
	if (p == NULL)
		return NULL;
	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
	p->arg.seqid = seqid;
5491
	p->res.seqid = seqid;
T
Trond Myklebust 已提交
5492 5493 5494 5495 5496 5497 5498 5499 5500
	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;
}

5501
static void nfs4_locku_release_calldata(void *data)
5502
{
5503
	struct nfs4_unlockdata *calldata = data;
T
Trond Myklebust 已提交
5504
	nfs_free_seqid(calldata->arg.seqid);
5505 5506 5507
	nfs4_put_lock_state(calldata->lsp);
	put_nfs_open_context(calldata->ctx);
	kfree(calldata);
5508 5509
}

5510
static void nfs4_locku_done(struct rpc_task *task, void *data)
5511
{
5512
	struct nfs4_unlockdata *calldata = data;
5513

5514 5515
	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
		return;
5516 5517
	switch (task->tk_status) {
		case 0:
5518
			renew_lease(calldata->server, calldata->timestamp);
5519
			do_vfs_lock(calldata->lsp->ls_state->inode, &calldata->fl);
5520 5521 5522
			if (nfs4_update_lock_stateid(calldata->lsp,
					&calldata->res.stateid))
				break;
5523 5524
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OLD_STATEID:
5525 5526
		case -NFS4ERR_STALE_STATEID:
		case -NFS4ERR_EXPIRED:
5527 5528 5529
			if (!nfs4_stateid_match(&calldata->arg.stateid,
						&calldata->lsp->ls_stateid))
				rpc_restart_call_prepare(task);
5530 5531
			break;
		default:
5532 5533
			if (nfs4_async_handle_error(task, calldata->server,
						    NULL, NULL) == -EAGAIN)
5534
				rpc_restart_call_prepare(task);
5535
	}
5536
	nfs_release_seqid(calldata->arg.seqid);
5537 5538
}

T
Trond Myklebust 已提交
5539
static void nfs4_locku_prepare(struct rpc_task *task, void *data)
5540
{
T
Trond Myklebust 已提交
5541
	struct nfs4_unlockdata *calldata = data;
5542

T
Trond Myklebust 已提交
5543
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
5544
		goto out_wait;
5545
	nfs4_stateid_copy(&calldata->arg.stateid, &calldata->lsp->ls_stateid);
5546
	if (test_bit(NFS_LOCK_INITIALIZED, &calldata->lsp->ls_flags) == 0) {
5547
		/* Note: exit _without_ running nfs4_locku_done */
5548
		goto out_no_action;
5549
	}
5550
	calldata->timestamp = jiffies;
5551
	if (nfs4_setup_sequence(calldata->server,
5552
				&calldata->arg.seq_args,
5553 5554 5555
				&calldata->res.seq_res,
				task) != 0)
		nfs_release_seqid(calldata->arg.seqid);
5556 5557 5558 5559 5560
	return;
out_no_action:
	task->tk_action = NULL;
out_wait:
	nfs4_sequence_done(task, &calldata->res.seq_res);
5561 5562
}

5563
static const struct rpc_call_ops nfs4_locku_ops = {
T
Trond Myklebust 已提交
5564
	.rpc_call_prepare = nfs4_locku_prepare,
5565
	.rpc_call_done = nfs4_locku_done,
5566
	.rpc_release = nfs4_locku_release_calldata,
5567 5568
};

5569 5570 5571 5572 5573 5574
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;
5575 5576 5577 5578
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCKU],
		.rpc_cred = ctx->cred,
	};
T
Trond Myklebust 已提交
5579 5580
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(lsp->ls_state->inode),
5581
		.rpc_message = &msg,
T
Trond Myklebust 已提交
5582
		.callback_ops = &nfs4_locku_ops,
5583
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
5584 5585
		.flags = RPC_TASK_ASYNC,
	};
5586

5587 5588 5589
	nfs4_state_protect(NFS_SERVER(lsp->ls_state->inode)->nfs_client,
		NFS_SP4_MACH_CRED_CLEANUP, &task_setup_data.rpc_client, &msg);

5590 5591 5592 5593 5594
	/* 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;

5595 5596 5597 5598 5599 5600
	data = nfs4_alloc_unlockdata(fl, ctx, lsp, seqid);
	if (data == NULL) {
		nfs_free_seqid(seqid);
		return ERR_PTR(-ENOMEM);
	}

5601
	nfs4_init_sequence(&data->arg.seq_args, &data->res.seq_res, 1);
5602 5603
	msg.rpc_argp = &data->arg;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
5604 5605
	task_setup_data.callback_data = data;
	return rpc_run_task(&task_setup_data);
5606 5607
}

5608 5609
static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
{
5610 5611 5612
	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
5613
	struct nfs_seqid *seqid;
L
Linus Torvalds 已提交
5614
	struct nfs4_lock_state *lsp;
5615
	struct rpc_task *task;
5616
	struct nfs_seqid *(*alloc_seqid)(struct nfs_seqid_counter *, gfp_t);
5617
	int status = 0;
5618
	unsigned char fl_flags = request->fl_flags;
5619

5620
	status = nfs4_set_lock_state(state, request);
5621 5622
	/* Unlock _before_ we do the RPC call */
	request->fl_flags |= FL_EXISTS;
5623 5624 5625
	/* Exclude nfs_delegation_claim_locks() */
	mutex_lock(&sp->so_delegreturn_mutex);
	/* Exclude nfs4_reclaim_open_stateid() - note nesting! */
5626
	down_read(&nfsi->rwsem);
5627
	if (do_vfs_lock(inode, request) == -ENOENT) {
5628
		up_read(&nfsi->rwsem);
5629
		mutex_unlock(&sp->so_delegreturn_mutex);
5630
		goto out;
5631 5632
	}
	up_read(&nfsi->rwsem);
5633
	mutex_unlock(&sp->so_delegreturn_mutex);
5634
	if (status != 0)
5635 5636
		goto out;
	/* Is this a delegated lock? */
5637
	lsp = request->fl_u.nfs4_fl.owner;
5638 5639
	if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) == 0)
		goto out;
5640 5641
	alloc_seqid = NFS_SERVER(inode)->nfs_client->cl_mvops->alloc_seqid;
	seqid = alloc_seqid(&lsp->ls_seqid, GFP_KERNEL);
5642
	status = -ENOMEM;
5643
	if (IS_ERR(seqid))
5644
		goto out;
5645
	task = nfs4_do_unlck(request, nfs_file_open_context(request->fl_file), lsp, seqid);
5646 5647
	status = PTR_ERR(task);
	if (IS_ERR(task))
5648
		goto out;
5649
	status = nfs4_wait_for_completion_rpc_task(task);
5650
	rpc_put_task(task);
5651
out:
5652
	request->fl_flags = fl_flags;
5653
	trace_nfs4_unlock(request, state, F_SETLK, status);
L
Linus Torvalds 已提交
5654 5655 5656
	return status;
}

5657 5658 5659 5660 5661 5662
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;
5663
	unsigned long timestamp;
5664 5665
	int rpc_status;
	int cancelled;
5666
	struct nfs_server *server;
5667 5668 5669
};

static struct nfs4_lockdata *nfs4_alloc_lockdata(struct file_lock *fl,
5670 5671
		struct nfs_open_context *ctx, struct nfs4_lock_state *lsp,
		gfp_t gfp_mask)
L
Linus Torvalds 已提交
5672
{
5673 5674
	struct nfs4_lockdata *p;
	struct inode *inode = lsp->ls_state->inode;
L
Linus Torvalds 已提交
5675
	struct nfs_server *server = NFS_SERVER(inode);
5676
	struct nfs_seqid *(*alloc_seqid)(struct nfs_seqid_counter *, gfp_t);
5677

5678
	p = kzalloc(sizeof(*p), gfp_mask);
5679 5680 5681 5682 5683
	if (p == NULL)
		return NULL;

	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
5684
	p->arg.open_seqid = nfs_alloc_seqid(&lsp->ls_state->owner->so_seqid, gfp_mask);
5685
	if (IS_ERR(p->arg.open_seqid))
5686
		goto out_free;
5687 5688
	alloc_seqid = server->nfs_client->cl_mvops->alloc_seqid;
	p->arg.lock_seqid = alloc_seqid(&lsp->ls_seqid, gfp_mask);
5689
	if (IS_ERR(p->arg.lock_seqid))
5690
		goto out_free_seqid;
5691
	p->arg.lock_owner.clientid = server->nfs_client->cl_clientid;
5692
	p->arg.lock_owner.id = lsp->ls_seqid.owner_id;
5693
	p->arg.lock_owner.s_dev = server->s_dev;
5694
	p->res.lock_seqid = p->arg.lock_seqid;
5695
	p->lsp = lsp;
5696
	p->server = server;
5697 5698
	atomic_inc(&lsp->ls_count);
	p->ctx = get_nfs_open_context(ctx);
5699
	get_file(fl->fl_file);
5700 5701
	memcpy(&p->fl, fl, sizeof(p->fl));
	return p;
5702 5703
out_free_seqid:
	nfs_free_seqid(p->arg.open_seqid);
5704 5705 5706 5707 5708 5709 5710 5711 5712
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;
5713

5714
	dprintk("%s: begin!\n", __func__);
5715
	if (nfs_wait_on_sequence(data->arg.lock_seqid, task) != 0)
5716
		goto out_wait;
5717
	/* Do we need to do an open_to_lock_owner? */
5718
	if (!test_bit(NFS_LOCK_INITIALIZED, &data->lsp->ls_flags)) {
5719
		if (nfs_wait_on_sequence(data->arg.open_seqid, task) != 0) {
5720
			goto out_release_lock_seqid;
5721
		}
5722 5723
		nfs4_stateid_copy(&data->arg.open_stateid,
				&state->open_stateid);
5724
		data->arg.new_lock_owner = 1;
5725
		data->res.open_seqid = data->arg.open_seqid;
5726
	} else {
5727
		data->arg.new_lock_owner = 0;
5728 5729 5730
		nfs4_stateid_copy(&data->arg.lock_stateid,
				&data->lsp->ls_stateid);
	}
5731 5732 5733 5734 5735
	if (!nfs4_valid_open_stateid(state)) {
		data->rpc_status = -EBADF;
		task->tk_action = NULL;
		goto out_release_open_seqid;
	}
5736
	data->timestamp = jiffies;
5737 5738
	if (nfs4_setup_sequence(data->server,
				&data->arg.seq_args,
5739
				&data->res.seq_res,
5740
				task) == 0)
5741
		return;
5742
out_release_open_seqid:
5743 5744 5745
	nfs_release_seqid(data->arg.open_seqid);
out_release_lock_seqid:
	nfs_release_seqid(data->arg.lock_seqid);
5746 5747
out_wait:
	nfs4_sequence_done(task, &data->res.seq_res);
5748
	dprintk("%s: done!, ret = %d\n", __func__, data->rpc_status);
5749 5750
}

5751 5752 5753
static void nfs4_lock_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_lockdata *data = calldata;
5754
	struct nfs4_lock_state *lsp = data->lsp;
5755

5756
	dprintk("%s: begin!\n", __func__);
5757

5758 5759
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return;
5760

5761
	data->rpc_status = task->tk_status;
5762 5763
	switch (task->tk_status) {
	case 0:
5764
		renew_lease(NFS_SERVER(d_inode(data->ctx->dentry)),
5765
				data->timestamp);
5766 5767
		if (data->arg.new_lock) {
			data->fl.fl_flags &= ~(FL_SLEEP | FL_ACCESS);
5768
			if (do_vfs_lock(lsp->ls_state->inode, &data->fl) < 0) {
5769 5770 5771 5772
				rpc_restart_call_prepare(task);
				break;
			}
		}
5773 5774 5775 5776 5777 5778
		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);
5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790
		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);
5791
	}
5792
	dprintk("%s: done, ret = %d!\n", __func__, data->rpc_status);
5793 5794 5795 5796 5797 5798
}

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

5799
	dprintk("%s: begin!\n", __func__);
5800
	nfs_free_seqid(data->arg.open_seqid);
5801 5802 5803 5804 5805
	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))
5806
			rpc_put_task_async(task);
5807
		dprintk("%s: cancelling lock!\n", __func__);
5808 5809 5810 5811
	} else
		nfs_free_seqid(data->arg.lock_seqid);
	nfs4_put_lock_state(data->lsp);
	put_nfs_open_context(data->ctx);
5812
	fput(data->fl.fl_file);
5813
	kfree(data);
5814
	dprintk("%s: done!\n", __func__);
5815 5816 5817 5818 5819 5820 5821 5822
}

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,
};

5823 5824 5825 5826 5827
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:
5828
		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
5829
		if (new_lock_owner != 0 ||
5830
		   test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) != 0)
5831
			nfs4_schedule_stateid_recovery(server, lsp->ls_state);
5832 5833 5834
		break;
	case -NFS4ERR_STALE_STATEID:
		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
5835 5836
	case -NFS4ERR_EXPIRED:
		nfs4_schedule_lease_recovery(server->nfs_client);
5837 5838 5839
	};
}

5840
static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *fl, int recovery_type)
5841 5842 5843
{
	struct nfs4_lockdata *data;
	struct rpc_task *task;
5844 5845 5846 5847
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCK],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
5848 5849
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(state->inode),
5850
		.rpc_message = &msg,
T
Trond Myklebust 已提交
5851
		.callback_ops = &nfs4_lock_ops,
5852
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
5853 5854
		.flags = RPC_TASK_ASYNC,
	};
5855 5856
	int ret;

5857
	dprintk("%s: begin!\n", __func__);
5858
	data = nfs4_alloc_lockdata(fl, nfs_file_open_context(fl->fl_file),
5859 5860
			fl->fl_u.nfs4_fl.owner,
			recovery_type == NFS_LOCK_NEW ? GFP_KERNEL : GFP_NOFS);
5861 5862 5863 5864
	if (data == NULL)
		return -ENOMEM;
	if (IS_SETLKW(cmd))
		data->arg.block = 1;
5865
	nfs4_init_sequence(&data->arg.seq_args, &data->res.seq_res, 1);
5866 5867
	msg.rpc_argp = &data->arg;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
5868
	task_setup_data.callback_data = data;
5869 5870 5871 5872
	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);
5873 5874
	} else
		data->arg.new_lock = 1;
T
Trond Myklebust 已提交
5875
	task = rpc_run_task(&task_setup_data);
5876
	if (IS_ERR(task))
5877 5878 5879 5880
		return PTR_ERR(task);
	ret = nfs4_wait_for_completion_rpc_task(task);
	if (ret == 0) {
		ret = data->rpc_status;
5881 5882 5883
		if (ret)
			nfs4_handle_setlk_error(data->server, data->lsp,
					data->arg.new_lock_owner, ret);
5884 5885
	} else
		data->cancelled = 1;
5886
	rpc_put_task(task);
5887
	dprintk("%s: done, ret = %d!\n", __func__, ret);
5888
	return ret;
L
Linus Torvalds 已提交
5889 5890 5891 5892
}

static int nfs4_lock_reclaim(struct nfs4_state *state, struct file_lock *request)
{
5893
	struct nfs_server *server = NFS_SERVER(state->inode);
5894 5895 5896
	struct nfs4_exception exception = {
		.inode = state->inode,
	};
5897 5898 5899
	int err;

	do {
5900 5901 5902
		/* Cache the lock if possible... */
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
5903
		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_RECLAIM);
5904
		trace_nfs4_lock_reclaim(request, state, F_SETLK, err);
5905
		if (err != -NFS4ERR_DELAY)
5906 5907 5908 5909
			break;
		nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
	return err;
L
Linus Torvalds 已提交
5910 5911 5912 5913
}

static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request)
{
5914
	struct nfs_server *server = NFS_SERVER(state->inode);
5915 5916 5917
	struct nfs4_exception exception = {
		.inode = state->inode,
	};
5918 5919
	int err;

5920 5921 5922
	err = nfs4_set_lock_state(state, request);
	if (err != 0)
		return err;
5923
	if (!recover_lost_locks) {
5924 5925 5926
		set_bit(NFS_LOCK_LOST, &request->fl_u.nfs4_fl.owner->ls_flags);
		return 0;
	}
5927
	do {
5928 5929
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
5930
		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_EXPIRED);
5931
		trace_nfs4_lock_expired(request, state, F_SETLK, err);
5932 5933 5934 5935 5936 5937 5938 5939
		switch (err) {
		default:
			goto out;
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			nfs4_handle_exception(server, err, &exception);
			err = 0;
		}
5940
	} while (exception.retry);
5941
out:
5942
	return err;
L
Linus Torvalds 已提交
5943 5944
}

5945
#if defined(CONFIG_NFS_V4_1)
5946 5947 5948 5949 5950 5951 5952 5953
/**
 * 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.
 */
5954
static int nfs41_check_expired_locks(struct nfs4_state *state)
5955
{
5956
	int status, ret = -NFS4ERR_BAD_STATEID;
5957
	struct nfs4_lock_state *lsp;
5958 5959
	struct nfs_server *server = NFS_SERVER(state->inode);

5960
	list_for_each_entry(lsp, &state->lock_states, ls_locks) {
5961
		if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags)) {
5962 5963 5964 5965 5966
			struct rpc_cred *cred = lsp->ls_state->owner->so_cred;

			status = nfs41_test_stateid(server,
					&lsp->ls_stateid,
					cred);
5967
			trace_nfs4_test_lock_stateid(state, lsp, status);
5968
			if (status != NFS_OK) {
5969 5970
				/* Free the stateid unless the server
				 * informs us the stateid is unrecognized. */
5971 5972
				if (status != -NFS4ERR_BAD_STATEID)
					nfs41_free_stateid(server,
5973 5974
							&lsp->ls_stateid,
							cred);
5975
				clear_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags);
5976 5977 5978 5979 5980 5981 5982 5983 5984 5985 5986 5987 5988 5989
				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);
5990 5991 5992
	if (status != NFS_OK)
		status = nfs4_lock_expired(state, request);
	return status;
5993 5994 5995
}
#endif

L
Linus Torvalds 已提交
5996 5997
static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
5998
	struct nfs_inode *nfsi = NFS_I(state->inode);
5999
	unsigned char fl_flags = request->fl_flags;
6000
	int status = -ENOLCK;
L
Linus Torvalds 已提交
6001

6002 6003 6004
	if ((fl_flags & FL_POSIX) &&
			!test_bit(NFS_STATE_POSIX_LOCKS, &state->flags))
		goto out;
6005 6006 6007 6008
	/* Is this a delegated open? */
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
6009
	request->fl_flags |= FL_ACCESS;
6010
	status = do_vfs_lock(state->inode, request);
6011 6012
	if (status < 0)
		goto out;
6013
	down_read(&nfsi->rwsem);
6014 6015 6016
	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
		/* Yes: cache locks! */
		/* ...but avoid races with delegation recall... */
6017
		request->fl_flags = fl_flags & ~FL_SLEEP;
6018
		status = do_vfs_lock(state->inode, request);
6019
		up_read(&nfsi->rwsem);
6020 6021
		goto out;
	}
6022
	up_read(&nfsi->rwsem);
6023
	status = _nfs4_do_setlk(state, cmd, request, NFS_LOCK_NEW);
6024 6025
out:
	request->fl_flags = fl_flags;
L
Linus Torvalds 已提交
6026 6027 6028 6029 6030
	return status;
}

static int nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
6031 6032
	struct nfs4_exception exception = {
		.state = state,
6033
		.inode = state->inode,
6034
	};
L
Linus Torvalds 已提交
6035 6036 6037
	int err;

	do {
6038
		err = _nfs4_proc_setlk(state, cmd, request);
6039
		trace_nfs4_set_lock(request, state, cmd, err);
6040 6041
		if (err == -NFS4ERR_DENIED)
			err = -EAGAIN;
L
Linus Torvalds 已提交
6042
		err = nfs4_handle_exception(NFS_SERVER(state->inode),
6043
				err, &exception);
L
Linus Torvalds 已提交
6044 6045 6046 6047 6048 6049 6050 6051 6052 6053 6054 6055 6056
	} 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 */
6057
	ctx = nfs_file_open_context(filp);
L
Linus Torvalds 已提交
6058 6059 6060 6061 6062
	state = ctx->state;

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

6063 6064 6065 6066 6067
	if (IS_GETLK(cmd)) {
		if (state != NULL)
			return nfs4_proc_getlk(state, F_GETLK, request);
		return 0;
	}
L
Linus Torvalds 已提交
6068 6069 6070 6071

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

6072 6073 6074 6075 6076
	if (request->fl_type == F_UNLCK) {
		if (state != NULL)
			return nfs4_proc_unlck(state, cmd, request);
		return 0;
	}
L
Linus Torvalds 已提交
6077

6078 6079
	if (state == NULL)
		return -ENOLCK;
6080 6081 6082 6083
	/*
	 * Don't rely on the VFS having checked the file open mode,
	 * since it won't do this for flock() locks.
	 */
6084
	switch (request->fl_type) {
6085 6086 6087 6088 6089 6090 6091 6092 6093
	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
6094 6095 6096 6097 6098 6099 6100 6101 6102 6103 6104 6105
	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;
}

6106
int nfs4_lock_delegation_recall(struct file_lock *fl, struct nfs4_state *state, const nfs4_stateid *stateid)
6107 6108 6109 6110 6111 6112
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	int err;

	err = nfs4_set_lock_state(state, fl);
	if (err != 0)
6113
		return err;
6114
	err = _nfs4_do_setlk(state, F_SETLK, fl, NFS_LOCK_NEW);
6115
	return nfs4_handle_delegation_recall_error(server, state, stateid, err);
6116
}
6117

6118 6119
struct nfs_release_lockowner_data {
	struct nfs4_lock_state *lsp;
6120
	struct nfs_server *server;
6121
	struct nfs_release_lockowner_args args;
6122
	struct nfs_release_lockowner_res res;
6123
	unsigned long timestamp;
6124 6125
};

6126 6127 6128
static void nfs4_release_lockowner_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs_release_lockowner_data *data = calldata;
6129
	struct nfs_server *server = data->server;
6130 6131
	nfs40_setup_sequence(server->nfs_client->cl_slot_tbl,
			     &data->args.seq_args, &data->res.seq_res, task);
6132
	data->args.lock_owner.clientid = server->nfs_client->cl_clientid;
6133
	data->timestamp = jiffies;
6134 6135 6136 6137 6138
}

static void nfs4_release_lockowner_done(struct rpc_task *task, void *calldata)
{
	struct nfs_release_lockowner_data *data = calldata;
6139 6140
	struct nfs_server *server = data->server;

6141
	nfs40_sequence_done(task, &data->res.seq_res);
6142 6143 6144 6145 6146 6147 6148

	switch (task->tk_status) {
	case 0:
		renew_lease(server, data->timestamp);
		break;
	case -NFS4ERR_STALE_CLIENTID:
	case -NFS4ERR_EXPIRED:
6149 6150
		nfs4_schedule_lease_recovery(server->nfs_client);
		break;
6151 6152
	case -NFS4ERR_LEASE_MOVED:
	case -NFS4ERR_DELAY:
6153 6154
		if (nfs4_async_handle_error(task, server,
					    NULL, NULL) == -EAGAIN)
6155 6156
			rpc_restart_call_prepare(task);
	}
6157 6158
}

6159 6160
static void nfs4_release_lockowner_release(void *calldata)
{
6161
	struct nfs_release_lockowner_data *data = calldata;
6162
	nfs4_free_lock_state(data->server, data->lsp);
6163 6164 6165
	kfree(calldata);
}

6166
static const struct rpc_call_ops nfs4_release_lockowner_ops = {
6167 6168
	.rpc_call_prepare = nfs4_release_lockowner_prepare,
	.rpc_call_done = nfs4_release_lockowner_done,
6169 6170 6171
	.rpc_release = nfs4_release_lockowner_release,
};

6172 6173
static void
nfs4_release_lockowner(struct nfs_server *server, struct nfs4_lock_state *lsp)
6174
{
6175
	struct nfs_release_lockowner_data *data;
6176 6177 6178 6179 6180
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RELEASE_LOCKOWNER],
	};

	if (server->nfs_client->cl_mvops->minor_version != 0)
6181
		return;
6182

6183 6184
	data = kmalloc(sizeof(*data), GFP_NOFS);
	if (!data)
6185
		return;
6186
	data->lsp = lsp;
6187
	data->server = server;
6188 6189 6190
	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;
6191

6192
	msg.rpc_argp = &data->args;
6193 6194
	msg.rpc_resp = &data->res;
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 0);
6195
	rpc_call_async(server->client, &msg, 0, &nfs4_release_lockowner_ops, data);
6196 6197
}

6198 6199
#define XATTR_NAME_NFSV4_ACL "system.nfs4_acl"

6200 6201 6202
static int nfs4_xattr_set_nfs4_acl(struct dentry *dentry, const char *key,
				   const void *buf, size_t buflen,
				   int flags, int type)
6203
{
6204 6205
	if (strcmp(key, "") != 0)
		return -EINVAL;
6206

6207
	return nfs4_proc_set_acl(d_inode(dentry), buf, buflen);
6208 6209
}

6210 6211
static int nfs4_xattr_get_nfs4_acl(struct dentry *dentry, const char *key,
				   void *buf, size_t buflen, int type)
6212
{
6213 6214
	if (strcmp(key, "") != 0)
		return -EINVAL;
6215

6216
	return nfs4_proc_get_acl(d_inode(dentry), buf, buflen);
6217 6218
}

6219 6220 6221
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)
6222
{
6223
	size_t len = sizeof(XATTR_NAME_NFSV4_ACL);
6224

6225
	if (!nfs4_server_supports_acls(NFS_SERVER(d_inode(dentry))))
6226
		return 0;
6227 6228 6229

	if (list && len <= list_len)
		memcpy(list, XATTR_NAME_NFSV4_ACL, len);
6230
	return len;
6231 6232
}

6233 6234 6235 6236 6237 6238 6239 6240 6241 6242 6243 6244 6245 6246 6247 6248 6249 6250 6251 6252
#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))
6253
		return nfs4_get_security_label(d_inode(dentry), buf, buflen);
6254 6255 6256 6257 6258 6259 6260 6261 6262
	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;

6263 6264
	if (nfs_server_capable(d_inode(dentry), NFS_CAP_SECURITY_LABEL)) {
		len = security_inode_listsecurity(d_inode(dentry), NULL, 0);
6265
		if (list && len <= list_len)
6266
			security_inode_listsecurity(d_inode(dentry), list, len);
6267 6268 6269 6270 6271 6272 6273 6274 6275 6276 6277 6278 6279
	}
	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


6280 6281 6282
/*
 * nfs_fhget will use either the mounted_on_fileid or the fileid
 */
6283 6284
static void nfs_fixup_referral_attributes(struct nfs_fattr *fattr)
{
6285 6286 6287
	if (!(((fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID) ||
	       (fattr->valid & NFS_ATTR_FATTR_FILEID)) &&
	      (fattr->valid & NFS_ATTR_FATTR_FSID) &&
6288
	      (fattr->valid & NFS_ATTR_FATTR_V4_LOCATIONS)))
6289 6290 6291
		return;

	fattr->valid |= NFS_ATTR_FATTR_TYPE | NFS_ATTR_FATTR_MODE |
6292
		NFS_ATTR_FATTR_NLINK | NFS_ATTR_FATTR_V4_REFERRAL;
6293 6294 6295 6296
	fattr->mode = S_IFDIR | S_IRUGO | S_IXUGO;
	fattr->nlink = 2;
}

6297 6298 6299 6300
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)
6301 6302
{
	struct nfs_server *server = NFS_SERVER(dir);
6303
	u32 bitmask[3] = {
6304
		[0] = FATTR4_WORD0_FSID | FATTR4_WORD0_FS_LOCATIONS,
6305 6306 6307
	};
	struct nfs4_fs_locations_arg args = {
		.dir_fh = NFS_FH(dir),
6308
		.name = name,
6309 6310 6311
		.page = page,
		.bitmask = bitmask,
	};
B
Benny Halevy 已提交
6312 6313 6314
	struct nfs4_fs_locations_res res = {
		.fs_locations = fs_locations,
	};
6315 6316 6317
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FS_LOCATIONS],
		.rpc_argp = &args,
B
Benny Halevy 已提交
6318
		.rpc_resp = &res,
6319 6320 6321
	};
	int status;

6322
	dprintk("%s: start\n", __func__);
6323 6324 6325 6326 6327 6328 6329 6330

	/* 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;

6331
	nfs_fattr_init(&fs_locations->fattr);
6332
	fs_locations->server = server;
6333
	fs_locations->nlocations = 0;
6334
	status = nfs4_call_sync(client, server, &msg, &args.seq_args, &res.seq_res, 0);
6335
	dprintk("%s: returned status = %d\n", __func__, status);
6336 6337 6338
	return status;
}

6339 6340 6341 6342
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)
6343 6344 6345 6346
{
	struct nfs4_exception exception = { };
	int err;
	do {
6347 6348 6349 6350
		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,
6351 6352 6353 6354 6355
				&exception);
	} while (exception.retry);
	return err;
}

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 6491 6492 6493 6494 6495 6496 6497 6498 6499 6500 6501 6502 6503 6504 6505 6506
/*
 * 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;
}

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 6617 6618 6619 6620 6621 6622 6623 6624 6625 6626 6627 6628 6629 6630 6631 6632
/*
 * 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;
}

6633
/**
6634 6635 6636 6637 6638
 * 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.
6639
 */
6640
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
6641 6642 6643 6644 6645 6646 6647 6648 6649 6650 6651 6652 6653 6654
{
	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,
	};
6655
	struct rpc_clnt *clnt = NFS_SERVER(dir)->client;
6656
	struct rpc_cred *cred = NULL;
6657 6658 6659

	if (use_integrity) {
		clnt = NFS_SERVER(dir)->nfs_client->cl_rpcclient;
6660 6661
		cred = nfs4_get_clid_cred(NFS_SERVER(dir)->nfs_client);
		msg.rpc_cred = cred;
6662
	}
B
Bryan Schumaker 已提交
6663 6664

	dprintk("NFS call  secinfo %s\n", name->name);
6665 6666 6667 6668

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

6669 6670
	status = nfs4_call_sync(clnt, NFS_SERVER(dir), &msg, &args.seq_args,
				&res.seq_res, 0);
B
Bryan Schumaker 已提交
6671
	dprintk("NFS reply  secinfo: %d\n", status);
6672

6673 6674
	if (cred)
		put_rpccred(cred);
6675

B
Bryan Schumaker 已提交
6676 6677 6678
	return status;
}

6679 6680
int nfs4_proc_secinfo(struct inode *dir, const struct qstr *name,
		      struct nfs4_secinfo_flavors *flavors)
B
Bryan Schumaker 已提交
6681 6682 6683 6684
{
	struct nfs4_exception exception = { };
	int err;
	do {
6685 6686 6687 6688 6689 6690 6691 6692 6693 6694 6695 6696 6697 6698 6699
		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);

6700 6701
		trace_nfs4_secinfo(dir, name, err);
		err = nfs4_handle_exception(NFS_SERVER(dir), err,
B
Bryan Schumaker 已提交
6702 6703 6704 6705 6706
				&exception);
	} while (exception.retry);
	return err;
}

6707
#ifdef CONFIG_NFS_V4_1
6708 6709 6710 6711 6712 6713 6714 6715 6716 6717 6718 6719 6720 6721 6722 6723 6724 6725 6726
/*
 * 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;
}

6727
static bool
6728 6729
nfs41_same_server_scope(struct nfs41_server_scope *a,
			struct nfs41_server_scope *b)
6730 6731 6732 6733 6734 6735 6736 6737
{
	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;
}

6738 6739 6740 6741 6742 6743
/*
 * nfs4_proc_bind_conn_to_session()
 *
 * The 4.1 client currently uses the same TCP connection for the
 * fore and backchannel.
 */
6744
int nfs4_proc_bind_conn_to_session(struct nfs_client *clp, struct rpc_cred *cred)
6745 6746
{
	int status;
6747 6748 6749 6750
	struct nfs41_bind_conn_to_session_args args = {
		.client = clp,
		.dir = NFS4_CDFC4_FORE_OR_BOTH,
	};
6751 6752 6753 6754
	struct nfs41_bind_conn_to_session_res res;
	struct rpc_message msg = {
		.rpc_proc =
			&nfs4_procedures[NFSPROC4_CLNT_BIND_CONN_TO_SESSION],
6755
		.rpc_argp = &args,
6756
		.rpc_resp = &res,
6757
		.rpc_cred = cred,
6758 6759 6760 6761
	};

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

6762 6763 6764
	nfs4_copy_sessionid(&args.sessionid, &clp->cl_session->sess_id);
	if (!(clp->cl_session->flags & SESSION4_BACK_CHAN))
		args.dir = NFS4_CDFC4_FORE;
6765 6766

	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
6767
	trace_nfs4_bind_conn_to_session(clp, status);
6768
	if (status == 0) {
6769
		if (memcmp(res.sessionid.data,
6770 6771 6772
		    clp->cl_session->sess_id.data, NFS4_MAX_SESSIONID_LEN)) {
			dprintk("NFS: %s: Session ID mismatch\n", __func__);
			status = -EIO;
6773
			goto out;
6774
		}
6775
		if ((res.dir & args.dir) != res.dir || res.dir == 0) {
6776 6777 6778
			dprintk("NFS: %s: Unexpected direction from server\n",
				__func__);
			status = -EIO;
6779
			goto out;
6780
		}
6781
		if (res.use_conn_in_rdma_mode != args.use_conn_in_rdma_mode) {
6782 6783 6784
			dprintk("NFS: %s: Server returned RDMA mode = true\n",
				__func__);
			status = -EIO;
6785
			goto out;
6786 6787 6788 6789 6790 6791 6792
		}
	}
out:
	dprintk("<-- %s status= %d\n", __func__, status);
	return status;
}

B
Benny Halevy 已提交
6793
/*
6794 6795
 * 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
6796 6797 6798 6799 6800 6801 6802 6803 6804
 */
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)
6805 6806 6807
	},
	.allow.u.words = {
		[0] = 1 << (OP_CLOSE) |
6808 6809
		      1 << (OP_LOCKU) |
		      1 << (OP_COMMIT),
6810
		[1] = 1 << (OP_SECINFO - 32) |
6811 6812
		      1 << (OP_SECINFO_NO_NAME - 32) |
		      1 << (OP_TEST_STATEID - 32) |
6813 6814
		      1 << (OP_FREE_STATEID - 32) |
		      1 << (OP_WRITE - 32)
6815 6816 6817 6818 6819 6820
	}
};

/*
 * Select the state protection mode for client `clp' given the server results
 * from exchange_id in `sp'.
B
Benny Halevy 已提交
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 6858 6859 6860 6861 6862 6863 6864 6865 6866 6867 6868 6869 6870 6871 6872 6873
 * 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;
		}
6874 6875 6876 6877 6878 6879

		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);
		}
6880 6881 6882 6883 6884 6885

		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);
		}
6886 6887 6888 6889 6890 6891

		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);
		}
6892 6893 6894 6895 6896 6897 6898 6899 6900 6901

		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);
		}
6902 6903 6904 6905 6906 6907 6908
	}

	return 0;
}

/*
 * _nfs4_proc_exchange_id()
6909
 *
6910
 * Wrapper for EXCHANGE_ID operation.
B
Benny Halevy 已提交
6911
 */
6912 6913
static int _nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred,
	u32 sp4_how)
B
Benny Halevy 已提交
6914 6915 6916
{
	nfs4_verifier verifier;
	struct nfs41_exchange_id_args args = {
6917
		.verifier = &verifier,
B
Benny Halevy 已提交
6918
		.client = clp,
6919
#ifdef CONFIG_NFS_V4_1_MIGRATION
6920
		.flags = EXCHGID4_FLAG_SUPP_MOVED_REFER |
6921 6922 6923 6924 6925 6926
			 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
6927 6928
	};
	struct nfs41_exchange_id_res res = {
6929
		0
B
Benny Halevy 已提交
6930 6931 6932 6933 6934 6935 6936 6937 6938
	};
	int status;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_EXCHANGE_ID],
		.rpc_argp = &args,
		.rpc_resp = &res,
		.rpc_cred = cred,
	};

6939
	nfs4_init_boot_verifier(clp, &verifier);
6940 6941 6942

	status = nfs4_init_uniform_client_string(clp);
	if (status)
6943 6944 6945
		goto out;

	dprintk("NFS call  exchange_id auth=%s, '%s'\n",
6946
		clp->cl_rpcclient->cl_auth->au_ops->au_name,
6947
		clp->cl_owner_id);
B
Benny Halevy 已提交
6948

6949
	res.server_owner = kzalloc(sizeof(struct nfs41_server_owner),
6950
					GFP_NOFS);
6951
	if (unlikely(res.server_owner == NULL)) {
6952 6953 6954
		status = -ENOMEM;
		goto out;
	}
6955

6956
	res.server_scope = kzalloc(sizeof(struct nfs41_server_scope),
6957
					GFP_NOFS);
6958
	if (unlikely(res.server_scope == NULL)) {
6959
		status = -ENOMEM;
6960
		goto out_server_owner;
6961
	}
6962

6963
	res.impl_id = kzalloc(sizeof(struct nfs41_impl_id), GFP_NOFS);
6964
	if (unlikely(res.impl_id == NULL)) {
6965 6966 6967 6968
		status = -ENOMEM;
		goto out_server_scope;
	}

6969 6970 6971 6972 6973 6974 6975 6976 6977 6978 6979 6980 6981
	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;
6982
		goto out_impl_id;
6983 6984
	}

6985
	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
6986
	trace_nfs4_exchange_id(clp, status);
6987
	if (status == 0)
6988
		status = nfs4_check_cl_exchange_flags(res.flags);
6989

6990 6991 6992
	if (status == 0)
		status = nfs4_sp4_select_mode(clp, &res.state_protect);

6993
	if (status == 0) {
6994
		clp->cl_clientid = res.clientid;
6995 6996 6997 6998 6999
		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);
7000
			clp->cl_seqid = res.seqid;
7001
		}
7002

7003 7004 7005
		kfree(clp->cl_serverowner);
		clp->cl_serverowner = res.server_owner;
		res.server_owner = NULL;
7006

7007
		/* use the most recent implementation id */
7008 7009
		kfree(clp->cl_implid);
		clp->cl_implid = res.impl_id;
7010
		res.impl_id = NULL;
7011

7012
		if (clp->cl_serverscope != NULL &&
7013
		    !nfs41_same_server_scope(clp->cl_serverscope,
7014 7015 7016 7017
					     res.server_scope)) {
			dprintk("%s: server_scope mismatch detected\n",
				__func__);
			set_bit(NFS4CLNT_SERVER_SCOPE_MISMATCH, &clp->cl_state);
7018 7019
			kfree(clp->cl_serverscope);
			clp->cl_serverscope = NULL;
7020 7021
		}

7022
		if (clp->cl_serverscope == NULL) {
7023
			clp->cl_serverscope = res.server_scope;
7024
			res.server_scope = NULL;
7025
		}
7026
	}
7027

7028 7029
out_impl_id:
	kfree(res.impl_id);
7030
out_server_scope:
7031
	kfree(res.server_scope);
7032 7033
out_server_owner:
	kfree(res.server_owner);
7034
out:
7035
	if (clp->cl_implid != NULL)
7036
		dprintk("NFS reply exchange_id: Server Implementation ID: "
7037
			"domain: %s, name: %s, date: %llu,%u\n",
7038
			clp->cl_implid->domain, clp->cl_implid->name,
7039 7040
			clp->cl_implid->date.seconds,
			clp->cl_implid->date.nseconds);
7041
	dprintk("NFS reply exchange_id: %d\n", status);
B
Benny Halevy 已提交
7042 7043 7044
	return status;
}

7045 7046 7047 7048 7049 7050 7051 7052 7053 7054 7055 7056 7057 7058 7059 7060 7061 7062 7063 7064 7065 7066 7067 7068 7069 7070 7071 7072 7073
/*
 * 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
7074 7075 7076 7077 7078 7079 7080 7081 7082 7083 7084
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);
7085
	trace_nfs4_destroy_clientid(clp, status);
T
Trond Myklebust 已提交
7086
	if (status)
7087
		dprintk("NFS: Got error %d from the server %s on "
T
Trond Myklebust 已提交
7088 7089 7090 7091 7092 7093 7094 7095 7096 7097 7098 7099 7100 7101 7102 7103 7104 7105 7106 7107 7108 7109 7110 7111 7112 7113 7114 7115 7116 7117 7118 7119 7120
			"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;
7121 7122
	if (clp->cl_preserve_clid)
		goto out;
7123
	cred = nfs4_get_clid_cred(clp);
T
Trond Myklebust 已提交
7124 7125 7126 7127 7128 7129 7130 7131 7132 7133 7134 7135
	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
7136 7137 7138 7139 7140 7141 7142 7143 7144 7145 7146 7147 7148 7149 7150
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 */
7151 7152 7153 7154
	nfs41_setup_sequence(data->clp->cl_session,
			&data->args->la_seq_args,
			&data->res->lr_seq_res,
			task);
A
Andy Adamson 已提交
7155 7156 7157 7158 7159 7160 7161 7162 7163 7164 7165 7166 7167
	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__);
7168 7169
	if (!nfs41_sequence_done(task, &data->res->lr_seq_res))
		return;
A
Andy Adamson 已提交
7170 7171 7172 7173 7174 7175
	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;
7176 7177
		/* fall through */
	case -NFS4ERR_RETRY_UNCACHED_REP:
7178
		rpc_restart_call_prepare(task);
A
Andy Adamson 已提交
7179 7180 7181 7182 7183
		return;
	}
	dprintk("<-- %s\n", __func__);
}

7184
static const struct rpc_call_ops nfs4_get_lease_time_ops = {
A
Andy Adamson 已提交
7185 7186 7187 7188 7189 7190 7191 7192 7193 7194 7195 7196 7197 7198 7199 7200 7201 7202 7203 7204 7205 7206 7207 7208 7209
	.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,
7210 7211
		.callback_data = &data,
		.flags = RPC_TASK_TIMEOUT,
A
Andy Adamson 已提交
7212 7213 7214
	};
	int status;

7215
	nfs4_init_sequence(&args.la_seq_args, &res.lr_seq_res, 0);
7216
	nfs4_set_sequence_privileged(&args.la_seq_args);
A
Andy Adamson 已提交
7217 7218 7219 7220 7221 7222 7223 7224 7225 7226 7227 7228 7229 7230
	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
7231 7232 7233 7234 7235 7236 7237 7238 7239 7240 7241
/*
 * 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)
{
7242 7243 7244 7245
	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
7246 7247

	/* Fore channel attributes */
7248 7249
	args->fc_attrs.max_rqst_sz = max_rqst_sz;
	args->fc_attrs.max_resp_sz = max_resp_sz;
A
Andy Adamson 已提交
7250
	args->fc_attrs.max_ops = NFS4_MAX_OPS;
7251
	args->fc_attrs.max_reqs = max_session_slots;
A
Andy Adamson 已提交
7252 7253

	dprintk("%s: Fore Channel : max_rqst_sz=%u max_resp_sz=%u "
7254
		"max_ops=%u max_reqs=%u\n",
A
Andy Adamson 已提交
7255 7256
		__func__,
		args->fc_attrs.max_rqst_sz, args->fc_attrs.max_resp_sz,
7257
		args->fc_attrs.max_ops, args->fc_attrs.max_reqs);
A
Andy Adamson 已提交
7258 7259 7260 7261 7262 7263 7264 7265 7266 7267 7268 7269 7270 7271 7272 7273

	/* 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);
}

7274 7275
static int nfs4_verify_fore_channel_attrs(struct nfs41_create_session_args *args,
		struct nfs41_create_session_res *res)
7276
{
7277
	struct nfs4_channel_attrs *sent = &args->fc_attrs;
7278
	struct nfs4_channel_attrs *rcvd = &res->fc_attrs;
7279 7280 7281 7282 7283 7284 7285 7286 7287 7288 7289 7290 7291

	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;
7292 7293
	if (rcvd->max_reqs > NFS4_MAX_SLOT_TABLE)
		rcvd->max_reqs = NFS4_MAX_SLOT_TABLE;
7294
	return 0;
7295 7296
}

7297 7298
static int nfs4_verify_back_channel_attrs(struct nfs41_create_session_args *args,
		struct nfs41_create_session_res *res)
7299 7300
{
	struct nfs4_channel_attrs *sent = &args->bc_attrs;
7301
	struct nfs4_channel_attrs *rcvd = &res->bc_attrs;
7302

7303 7304
	if (!(res->flags & SESSION4_BACK_CHAN))
		goto out;
7305 7306 7307 7308 7309 7310 7311
	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: */
7312
	if (rcvd->max_ops != sent->max_ops)
7313
		return -EINVAL;
7314
	if (rcvd->max_reqs != sent->max_reqs)
7315
		return -EINVAL;
7316
out:
7317 7318
	return 0;
}
7319 7320

static int nfs4_verify_channel_attrs(struct nfs41_create_session_args *args,
7321
				     struct nfs41_create_session_res *res)
7322
{
7323
	int ret;
7324

7325
	ret = nfs4_verify_fore_channel_attrs(args, res);
7326 7327
	if (ret)
		return ret;
7328 7329 7330 7331 7332 7333 7334
	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);
7335 7336 7337
	/* 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);
7338 7339
	session->flags = res->flags;
	memcpy(&session->fc_attrs, &res->fc_attrs, sizeof(session->fc_attrs));
7340 7341 7342
	if (res->flags & SESSION4_BACK_CHAN)
		memcpy(&session->bc_attrs, &res->bc_attrs,
				sizeof(session->bc_attrs));
7343 7344
}

7345 7346
static int _nfs4_proc_create_session(struct nfs_client *clp,
		struct rpc_cred *cred)
A
Andy Adamson 已提交
7347 7348 7349 7350
{
	struct nfs4_session *session = clp->cl_session;
	struct nfs41_create_session_args args = {
		.client = clp,
7351 7352
		.clientid = clp->cl_clientid,
		.seqid = clp->cl_seqid,
A
Andy Adamson 已提交
7353 7354
		.cb_program = NFS4_CALLBACK,
	};
7355 7356
	struct nfs41_create_session_res res;

A
Andy Adamson 已提交
7357 7358 7359 7360
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CREATE_SESSION],
		.rpc_argp = &args,
		.rpc_resp = &res,
7361
		.rpc_cred = cred,
A
Andy Adamson 已提交
7362 7363 7364 7365
	};
	int status;

	nfs4_init_channel_attrs(&args);
7366
	args.flags = (SESSION4_PERSIST | SESSION4_BACK_CHAN);
A
Andy Adamson 已提交
7367

7368
	status = rpc_call_sync(session->clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
7369
	trace_nfs4_create_session(clp, status);
A
Andy Adamson 已提交
7370

7371
	if (!status) {
7372
		/* Verify the session's negotiated channel_attrs values */
7373
		status = nfs4_verify_channel_attrs(&args, &res);
A
Andy Adamson 已提交
7374
		/* Increment the clientid slot sequence id */
7375 7376 7377 7378 7379
		if (clp->cl_seqid == res.seqid)
			clp->cl_seqid++;
		if (status)
			goto out;
		nfs4_update_session(session, &res);
A
Andy Adamson 已提交
7380
	}
7381
out:
A
Andy Adamson 已提交
7382 7383 7384 7385 7386 7387 7388 7389
	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.
 */
7390
int nfs4_proc_create_session(struct nfs_client *clp, struct rpc_cred *cred)
A
Andy Adamson 已提交
7391 7392 7393 7394 7395 7396 7397
{
	int status;
	unsigned *ptr;
	struct nfs4_session *session = clp->cl_session;

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

7398
	status = _nfs4_proc_create_session(clp, cred);
A
Andy Adamson 已提交
7399 7400 7401
	if (status)
		goto out;

7402 7403 7404
	/* 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
7405 7406 7407 7408 7409 7410 7411 7412 7413 7414 7415
	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
7416 7417 7418 7419
/*
 * Issue the over-the-wire RPC DESTROY_SESSION.
 * The caller must serialize access to this routine.
 */
7420 7421
int nfs4_proc_destroy_session(struct nfs4_session *session,
		struct rpc_cred *cred)
A
Andy Adamson 已提交
7422
{
7423 7424 7425 7426 7427
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DESTROY_SESSION],
		.rpc_argp = session,
		.rpc_cred = cred,
	};
A
Andy Adamson 已提交
7428 7429 7430 7431 7432
	int status = 0;

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

	/* session is still being setup */
7433 7434
	if (!test_and_clear_bit(NFS4_SESSION_ESTABLISHED, &session->session_state))
		return 0;
A
Andy Adamson 已提交
7435

7436
	status = rpc_call_sync(session->clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
7437
	trace_nfs4_destroy_session(session->clp, status);
A
Andy Adamson 已提交
7438 7439

	if (status)
7440
		dprintk("NFS: Got error %d from the server on DESTROY_SESSION. "
A
Andy Adamson 已提交
7441 7442 7443 7444 7445 7446
			"Session has been destroyed regardless...\n", status);

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

A
Andy Adamson 已提交
7447 7448 7449
/*
 * Renew the cl_session lease.
 */
7450 7451 7452 7453 7454 7455
struct nfs4_sequence_data {
	struct nfs_client *clp;
	struct nfs4_sequence_args args;
	struct nfs4_sequence_res res;
};

7456 7457
static void nfs41_sequence_release(void *data)
{
7458 7459
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
7460

7461 7462 7463
	if (atomic_read(&clp->cl_count) > 1)
		nfs4_schedule_state_renewal(clp);
	nfs_put_client(clp);
7464
	kfree(calldata);
7465 7466
}

7467 7468 7469 7470 7471 7472 7473
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:
7474
		nfs4_schedule_lease_recovery(clp);
7475 7476 7477 7478
	}
	return 0;
}

7479
static void nfs41_sequence_call_done(struct rpc_task *task, void *data)
A
Andy Adamson 已提交
7480
{
7481 7482
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
A
Andy Adamson 已提交
7483

7484 7485
	if (!nfs41_sequence_done(task, task->tk_msg.rpc_resp))
		return;
A
Andy Adamson 已提交
7486

7487
	trace_nfs4_sequence(clp, task->tk_status);
A
Andy Adamson 已提交
7488 7489
	if (task->tk_status < 0) {
		dprintk("%s ERROR %d\n", __func__, task->tk_status);
7490 7491
		if (atomic_read(&clp->cl_count) == 1)
			goto out;
A
Andy Adamson 已提交
7492

7493 7494
		if (nfs41_sequence_handle_errors(task, clp) == -EAGAIN) {
			rpc_restart_call_prepare(task);
A
Andy Adamson 已提交
7495 7496 7497 7498
			return;
		}
	}
	dprintk("%s rpc_cred %p\n", __func__, task->tk_msg.rpc_cred);
7499
out:
A
Andy Adamson 已提交
7500 7501 7502 7503 7504
	dprintk("<-- %s\n", __func__);
}

static void nfs41_sequence_prepare(struct rpc_task *task, void *data)
{
7505 7506
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
A
Andy Adamson 已提交
7507 7508 7509 7510 7511 7512
	struct nfs4_sequence_args *args;
	struct nfs4_sequence_res *res;

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

7513
	nfs41_setup_sequence(clp->cl_session, args, res, task);
A
Andy Adamson 已提交
7514 7515 7516 7517 7518
}

static const struct rpc_call_ops nfs41_sequence_ops = {
	.rpc_call_done = nfs41_sequence_call_done,
	.rpc_call_prepare = nfs41_sequence_prepare,
7519
	.rpc_release = nfs41_sequence_release,
A
Andy Adamson 已提交
7520 7521
};

7522 7523 7524
static struct rpc_task *_nfs41_proc_sequence(struct nfs_client *clp,
		struct rpc_cred *cred,
		bool is_privileged)
A
Andy Adamson 已提交
7525
{
7526
	struct nfs4_sequence_data *calldata;
A
Andy Adamson 已提交
7527 7528 7529 7530
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SEQUENCE],
		.rpc_cred = cred,
	};
7531 7532 7533
	struct rpc_task_setup task_setup_data = {
		.rpc_client = clp->cl_rpcclient,
		.rpc_message = &msg,
7534
		.callback_ops = &nfs41_sequence_ops,
7535
		.flags = RPC_TASK_ASYNC | RPC_TASK_TIMEOUT,
7536
	};
A
Andy Adamson 已提交
7537

7538
	if (!atomic_inc_not_zero(&clp->cl_count))
7539
		return ERR_PTR(-EIO);
7540
	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
7541
	if (calldata == NULL) {
7542
		nfs_put_client(clp);
7543
		return ERR_PTR(-ENOMEM);
A
Andy Adamson 已提交
7544
	}
7545
	nfs4_init_sequence(&calldata->args, &calldata->res, 0);
7546 7547
	if (is_privileged)
		nfs4_set_sequence_privileged(&calldata->args);
7548 7549 7550
	msg.rpc_argp = &calldata->args;
	msg.rpc_resp = &calldata->res;
	calldata->clp = clp;
7551
	task_setup_data.callback_data = calldata;
A
Andy Adamson 已提交
7552

7553 7554 7555
	return rpc_run_task(&task_setup_data);
}

7556
static int nfs41_proc_async_sequence(struct nfs_client *clp, struct rpc_cred *cred, unsigned renew_flags)
7557 7558 7559 7560
{
	struct rpc_task *task;
	int ret = 0;

7561
	if ((renew_flags & NFS4_RENEW_TIMEOUT) == 0)
7562
		return -EAGAIN;
7563
	task = _nfs41_proc_sequence(clp, cred, false);
7564 7565 7566
	if (IS_ERR(task))
		ret = PTR_ERR(task);
	else
7567
		rpc_put_task_async(task);
7568 7569 7570 7571 7572 7573 7574 7575 7576
	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;

7577
	task = _nfs41_proc_sequence(clp, cred, true);
7578 7579 7580 7581 7582
	if (IS_ERR(task)) {
		ret = PTR_ERR(task);
		goto out;
	}
	ret = rpc_wait_for_completion_task(task);
7583
	if (!ret)
7584 7585 7586 7587 7588
		ret = task->tk_status;
	rpc_put_task(task);
out:
	dprintk("<-- %s status=%d\n", __func__, ret);
	return ret;
A
Andy Adamson 已提交
7589 7590
}

7591 7592 7593 7594 7595 7596 7597 7598 7599 7600
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;

7601 7602 7603 7604
	nfs41_setup_sequence(calldata->clp->cl_session,
			&calldata->arg.seq_args,
			&calldata->res.seq_res,
			task);
7605 7606
}

7607 7608 7609 7610 7611 7612 7613 7614 7615
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);
7616 7617
		/* fall through */
	case -NFS4ERR_RETRY_UNCACHED_REP:
7618 7619
		return -EAGAIN;
	default:
7620
		nfs4_schedule_lease_recovery(clp);
7621 7622 7623 7624
	}
	return 0;
}

7625 7626 7627 7628 7629 7630 7631
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__);
7632 7633
	if (!nfs41_sequence_done(task, res))
		return;
7634

7635
	trace_nfs4_reclaim_complete(clp, task->tk_status);
7636 7637 7638 7639
	if (nfs41_reclaim_complete_handle_errors(task, clp) == -EAGAIN) {
		rpc_restart_call_prepare(task);
		return;
	}
7640 7641 7642 7643 7644 7645 7646 7647 7648 7649 7650 7651 7652 7653 7654 7655 7656 7657 7658
	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.
 */
7659 7660
static int nfs41_proc_reclaim_complete(struct nfs_client *clp,
		struct rpc_cred *cred)
7661 7662 7663 7664 7665
{
	struct nfs4_reclaim_complete_data *calldata;
	struct rpc_task *task;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RECLAIM_COMPLETE],
7666
		.rpc_cred = cred,
7667 7668 7669 7670 7671 7672 7673 7674 7675 7676
	};
	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__);
7677
	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
7678 7679 7680 7681 7682
	if (calldata == NULL)
		goto out;
	calldata->clp = clp;
	calldata->arg.one_fs = 0;

7683
	nfs4_init_sequence(&calldata->arg.seq_args, &calldata->res.seq_res, 0);
7684
	nfs4_set_sequence_privileged(&calldata->arg.seq_args);
7685 7686 7687 7688
	msg.rpc_argp = &calldata->arg;
	msg.rpc_resp = &calldata->res;
	task_setup_data.callback_data = calldata;
	task = rpc_run_task(&task_setup_data);
7689
	if (IS_ERR(task)) {
7690
		status = PTR_ERR(task);
7691 7692
		goto out;
	}
7693 7694 7695
	status = nfs4_wait_for_completion_rpc_task(task);
	if (status == 0)
		status = task->tk_status;
7696
	rpc_put_task(task);
7697
	return 0;
7698 7699 7700 7701
out:
	dprintk("<-- %s status=%d\n", __func__, status);
	return status;
}
7702 7703 7704 7705 7706

static void
nfs4_layoutget_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;
F
Fred Isaman 已提交
7707
	struct nfs_server *server = NFS_SERVER(lgp->args.inode);
7708
	struct nfs4_session *session = nfs4_get_session(server);
7709 7710

	dprintk("--> %s\n", __func__);
F
Fred Isaman 已提交
7711 7712 7713 7714 7715
	/* 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.
	 */
7716
	if (nfs41_setup_sequence(session, &lgp->args.seq_args,
7717
				&lgp->res.seq_res, task))
7718
		return;
7719 7720
	if (pnfs_choose_layoutget_stateid(&lgp->args.stateid,
					  NFS_I(lgp->args.inode)->layout,
7721
					  &lgp->args.range,
7722 7723 7724
					  lgp->args.ctx->state)) {
		rpc_exit(task, NFS4_OK);
	}
7725 7726 7727 7728 7729
}

static void nfs4_layoutget_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;
7730 7731 7732 7733
	struct inode *inode = lgp->args.inode;
	struct nfs_server *server = NFS_SERVER(inode);
	struct pnfs_layout_hdr *lo;
	struct nfs4_state *state = NULL;
7734
	unsigned long timeo, now, giveup;
7735

7736
	dprintk("--> %s tk_status => %d\n", __func__, -task->tk_status);
7737

7738
	if (!nfs41_sequence_done(task, &lgp->res.seq_res))
7739
		goto out;
7740 7741 7742

	switch (task->tk_status) {
	case 0:
7743
		goto out;
7744 7745 7746 7747
	/*
	 * NFS4ERR_LAYOUTTRYLATER is a conflict with another client
	 * (or clients) writing to the same RAID stripe
	 */
7748
	case -NFS4ERR_LAYOUTTRYLATER:
7749 7750 7751 7752
	/*
	 * NFS4ERR_RECALLCONFLICT is when conflict with self (must recall
	 * existing layout before getting a new one).
	 */
7753
	case -NFS4ERR_RECALLCONFLICT:
7754 7755
		timeo = rpc_get_timeout(task->tk_client);
		giveup = lgp->args.timestamp + timeo;
7756 7757 7758 7759 7760 7761 7762 7763 7764 7765 7766 7767 7768 7769 7770 7771 7772 7773 7774 7775
		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() */
		}
7776 7777 7778 7779 7780 7781 7782 7783 7784 7785 7786 7787
		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);

7788 7789 7790 7791
			/*
			 * Mark the bad layout state as invalid, then retry
			 * with the current stateid.
			 */
7792 7793 7794
			pnfs_mark_matching_lsegs_invalid(lo, &head, NULL);
			spin_unlock(&inode->i_lock);
			pnfs_free_lseg_list(&head);
7795 7796 7797
	
			task->tk_status = 0;
			rpc_restart_call_prepare(task);
7798 7799
		}
	}
7800
	if (nfs4_async_handle_error(task, server, state, NULL) == -EAGAIN)
7801 7802
		rpc_restart_call_prepare(task);
out:
7803 7804 7805
	dprintk("<-- %s\n", __func__);
}

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 7837 7838 7839 7840 7841 7842 7843 7844 7845 7846 7847 7848 7849
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;
}

7850 7851 7852
static void nfs4_layoutget_release(void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;
7853 7854
	struct inode *inode = lgp->args.inode;
	struct nfs_server *server = NFS_SERVER(inode);
7855
	size_t max_pages = max_response_pages(server);
7856 7857

	dprintk("--> %s\n", __func__);
7858
	nfs4_free_pages(lgp->args.layout.pages, max_pages);
7859
	pnfs_put_layout_hdr(NFS_I(inode)->layout);
7860 7861 7862 7863 7864 7865 7866 7867 7868 7869 7870
	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,
};

7871 7872
struct pnfs_layout_segment *
nfs4_proc_layoutget(struct nfs4_layoutget *lgp, gfp_t gfp_flags)
7873
{
7874 7875
	struct inode *inode = lgp->args.inode;
	struct nfs_server *server = NFS_SERVER(inode);
7876
	size_t max_pages = max_response_pages(server);
7877 7878 7879 7880 7881
	struct rpc_task *task;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LAYOUTGET],
		.rpc_argp = &lgp->args,
		.rpc_resp = &lgp->res,
7882
		.rpc_cred = lgp->cred,
7883 7884 7885 7886 7887 7888 7889 7890
	};
	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,
	};
7891
	struct pnfs_layout_segment *lseg = NULL;
7892 7893 7894 7895
	int status = 0;

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

7896 7897 7898
	/* nfs4_layoutget_release calls pnfs_put_layout_hdr */
	pnfs_get_layout_hdr(NFS_I(inode)->layout);

7899 7900 7901
	lgp->args.layout.pages = nfs4_alloc_pages(max_pages, gfp_flags);
	if (!lgp->args.layout.pages) {
		nfs4_layoutget_release(lgp);
7902
		return ERR_PTR(-ENOMEM);
7903 7904
	}
	lgp->args.layout.pglen = max_pages * PAGE_SIZE;
7905
	lgp->args.timestamp = jiffies;
7906

7907
	lgp->res.layoutp = &lgp->args.layout;
7908
	lgp->res.seq_res.sr_slot = NULL;
7909
	nfs4_init_sequence(&lgp->args.seq_args, &lgp->res.seq_res, 0);
7910

7911 7912
	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task))
7913
		return ERR_CAST(task);
7914
	status = nfs4_wait_for_completion_rpc_task(task);
F
Fred Isaman 已提交
7915 7916
	if (status == 0)
		status = task->tk_status;
7917 7918 7919 7920
	trace_nfs4_layoutget(lgp->args.ctx,
			&lgp->args.range,
			&lgp->res.range,
			status);
7921 7922
	/* if layoutp->len is 0, nfs4_layoutget_prepare called rpc_exit */
	if (status == 0 && lgp->res.layoutp->len)
7923
		lseg = pnfs_layout_process(lgp);
7924 7925
	rpc_put_task(task);
	dprintk("<-- %s status=%d\n", __func__, status);
7926 7927 7928
	if (status)
		return ERR_PTR(status);
	return lseg;
7929 7930
}

B
Benny Halevy 已提交
7931 7932 7933 7934 7935 7936
static void
nfs4_layoutreturn_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutreturn *lrp = calldata;

	dprintk("--> %s\n", __func__);
7937 7938 7939 7940
	nfs41_setup_sequence(lrp->clp->cl_session,
			&lrp->args.seq_args,
			&lrp->res.seq_res,
			task);
B
Benny Halevy 已提交
7941 7942 7943 7944 7945 7946 7947 7948 7949
}

static void nfs4_layoutreturn_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutreturn *lrp = calldata;
	struct nfs_server *server;

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

7950
	if (!nfs41_sequence_done(task, &lrp->res.seq_res))
B
Benny Halevy 已提交
7951 7952 7953
		return;

	server = NFS_SERVER(lrp->args.inode);
7954 7955 7956 7957 7958 7959
	switch (task->tk_status) {
	default:
		task->tk_status = 0;
	case 0:
		break;
	case -NFS4ERR_DELAY:
7960
		if (nfs4_async_handle_error(task, server, NULL, NULL) != -EAGAIN)
7961
			break;
7962
		rpc_restart_call_prepare(task);
B
Benny Halevy 已提交
7963 7964 7965 7966 7967 7968 7969 7970
		return;
	}
	dprintk("<-- %s\n", __func__);
}

static void nfs4_layoutreturn_release(void *calldata)
{
	struct nfs4_layoutreturn *lrp = calldata;
7971
	struct pnfs_layout_hdr *lo = lrp->args.layout;
7972
	LIST_HEAD(freeme);
B
Benny Halevy 已提交
7973 7974

	dprintk("--> %s\n", __func__);
7975 7976 7977
	spin_lock(&lo->plh_inode->i_lock);
	if (lrp->res.lrs_present)
		pnfs_set_layout_stateid(lo, &lrp->res.stateid, true);
7978
	pnfs_mark_matching_lsegs_invalid(lo, &freeme, &lrp->args.range);
7979
	pnfs_clear_layoutreturn_waitbit(lo);
7980 7981
	lo->plh_block_lgets--;
	spin_unlock(&lo->plh_inode->i_lock);
7982
	pnfs_free_lseg_list(&freeme);
7983
	pnfs_put_layout_hdr(lrp->args.layout);
7984
	nfs_iput_and_deactive(lrp->inode);
B
Benny Halevy 已提交
7985 7986 7987 7988 7989 7990 7991 7992 7993 7994
	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,
};

7995
int nfs4_proc_layoutreturn(struct nfs4_layoutreturn *lrp, bool sync)
B
Benny Halevy 已提交
7996 7997 7998 7999 8000 8001
{
	struct rpc_task *task;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LAYOUTRETURN],
		.rpc_argp = &lrp->args,
		.rpc_resp = &lrp->res,
8002
		.rpc_cred = lrp->cred,
B
Benny Halevy 已提交
8003 8004
	};
	struct rpc_task_setup task_setup_data = {
8005
		.rpc_client = NFS_SERVER(lrp->args.inode)->client,
B
Benny Halevy 已提交
8006 8007 8008 8009
		.rpc_message = &msg,
		.callback_ops = &nfs4_layoutreturn_call_ops,
		.callback_data = lrp,
	};
8010
	int status = 0;
B
Benny Halevy 已提交
8011 8012

	dprintk("--> %s\n", __func__);
8013 8014 8015 8016 8017 8018 8019 8020
	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;
	}
8021
	nfs4_init_sequence(&lrp->args.seq_args, &lrp->res.seq_res, 1);
B
Benny Halevy 已提交
8022 8023 8024
	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task))
		return PTR_ERR(task);
8025 8026
	if (sync)
		status = task->tk_status;
8027
	trace_nfs4_layoutreturn(lrp->args.inode, status);
B
Benny Halevy 已提交
8028 8029 8030 8031 8032
	dprintk("<-- %s status=%d\n", __func__, status);
	rpc_put_task(task);
	return status;
}

8033
static int
8034 8035 8036
_nfs4_proc_getdeviceinfo(struct nfs_server *server,
		struct pnfs_device *pdev,
		struct rpc_cred *cred)
8037 8038 8039
{
	struct nfs4_getdeviceinfo_args args = {
		.pdev = pdev,
8040 8041
		.notify_types = NOTIFY_DEVICEID4_CHANGE |
			NOTIFY_DEVICEID4_DELETE,
8042 8043 8044 8045 8046 8047 8048 8049
	};
	struct nfs4_getdeviceinfo_res res = {
		.pdev = pdev,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETDEVICEINFO],
		.rpc_argp = &args,
		.rpc_resp = &res,
8050
		.rpc_cred = cred,
8051 8052 8053 8054
	};
	int status;

	dprintk("--> %s\n", __func__);
8055
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
8056 8057
	if (res.notification & ~args.notify_types)
		dprintk("%s: unsupported notification\n", __func__);
8058 8059
	if (res.notification != args.notify_types)
		pdev->nocache = 1;
8060

8061 8062 8063 8064 8065
	dprintk("<-- %s status=%d\n", __func__, status);

	return status;
}

8066 8067 8068
int nfs4_proc_getdeviceinfo(struct nfs_server *server,
		struct pnfs_device *pdev,
		struct rpc_cred *cred)
8069 8070 8071 8072 8073 8074
{
	struct nfs4_exception exception = { };
	int err;

	do {
		err = nfs4_handle_exception(server,
8075
					_nfs4_proc_getdeviceinfo(server, pdev, cred),
8076 8077 8078 8079 8080 8081
					&exception);
	} while (exception.retry);
	return err;
}
EXPORT_SYMBOL_GPL(nfs4_proc_getdeviceinfo);

A
Andy Adamson 已提交
8082 8083 8084 8085
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);
8086
	struct nfs4_session *session = nfs4_get_session(server);
A
Andy Adamson 已提交
8087

8088 8089 8090 8091
	nfs41_setup_sequence(session,
			&data->args.seq_args,
			&data->res.seq_res,
			task);
A
Andy Adamson 已提交
8092 8093 8094 8095 8096 8097 8098 8099
}

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);

8100
	if (!nfs41_sequence_done(task, &data->res.seq_res))
A
Andy Adamson 已提交
8101 8102 8103
		return;

	switch (task->tk_status) { /* Just ignore these failures */
8104 8105 8106 8107
	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
8108
		task->tk_status = 0;
8109 8110 8111
	case 0:
		break;
	default:
8112
		if (nfs4_async_handle_error(task, server, NULL, NULL) == -EAGAIN) {
8113 8114 8115 8116
			rpc_restart_call_prepare(task);
			return;
		}
	}
A
Andy Adamson 已提交
8117 8118 8119 8120 8121 8122
}

static void nfs4_layoutcommit_release(void *calldata)
{
	struct nfs4_layoutcommit_data *data = calldata;

A
Andy Adamson 已提交
8123
	pnfs_cleanup_layoutcommit(data);
8124 8125
	nfs_post_op_update_inode_force_wcc(data->args.inode,
					   data->res.fattr);
A
Andy Adamson 已提交
8126
	put_rpccred(data->cred);
8127
	nfs_iput_and_deactive(data->inode);
A
Andy Adamson 已提交
8128 8129 8130 8131 8132 8133 8134 8135 8136 8137
	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
8138
nfs4_proc_layoutcommit(struct nfs4_layoutcommit_data *data, bool sync)
A
Andy Adamson 已提交
8139 8140 8141 8142 8143 8144 8145 8146 8147 8148 8149 8150 8151 8152 8153 8154 8155
{
	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;

8156 8157
	dprintk("NFS: initiating layoutcommit call. sync %d "
		"lbw: %llu inode %lu\n", sync,
A
Andy Adamson 已提交
8158 8159 8160
		data->args.lastbytewritten,
		data->args.inode->i_ino);

8161 8162 8163 8164 8165 8166 8167 8168
	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;
	}
8169
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
A
Andy Adamson 已提交
8170 8171 8172
	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task))
		return PTR_ERR(task);
8173 8174
	if (sync)
		status = task->tk_status;
8175
	trace_nfs4_layoutcommit(data->args.inode, status);
A
Andy Adamson 已提交
8176 8177 8178 8179
	dprintk("%s: status %d\n", __func__, status);
	rpc_put_task(task);
	return status;
}
8180

8181 8182 8183 8184
/**
 * Use the state managment nfs_client cl_rpcclient, which uses krb5i (if
 * possible) as per RFC3530bis and RFC5661 Security Considerations sections
 */
8185 8186
static int
_nfs41_proc_secinfo_no_name(struct nfs_server *server, struct nfs_fh *fhandle,
8187 8188
		    struct nfs_fsinfo *info,
		    struct nfs4_secinfo_flavors *flavors, bool use_integrity)
8189 8190 8191 8192 8193 8194 8195 8196 8197 8198 8199 8200
{
	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,
	};
8201
	struct rpc_clnt *clnt = server->client;
8202
	struct rpc_cred *cred = NULL;
8203 8204 8205 8206
	int status;

	if (use_integrity) {
		clnt = server->nfs_client->cl_rpcclient;
8207 8208
		cred = nfs4_get_clid_cred(server->nfs_client);
		msg.rpc_cred = cred;
8209 8210 8211 8212 8213 8214 8215
	}

	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);

8216 8217
	if (cred)
		put_rpccred(cred);
8218 8219

	return status;
8220 8221 8222 8223 8224 8225 8226 8227 8228
}

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 {
8229 8230 8231 8232 8233 8234 8235 8236 8237 8238 8239 8240 8241 8242 8243 8244 8245 8246
		/* 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);

8247 8248 8249
		switch (err) {
		case 0:
		case -NFS4ERR_WRONGSEC:
8250
		case -ENOTSUPP:
8251
			goto out;
8252 8253 8254 8255
		default:
			err = nfs4_handle_exception(server, err, &exception);
		}
	} while (exception.retry);
8256
out:
8257 8258 8259 8260 8261 8262 8263 8264 8265
	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;
8266
	rpc_authflavor_t flavor = RPC_AUTH_MAXFLAVOR;
8267
	struct nfs4_secinfo_flavors *flavors;
8268 8269
	struct nfs4_secinfo4 *secinfo;
	int i;
8270 8271 8272 8273 8274 8275 8276 8277 8278 8279 8280 8281 8282 8283

	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
	 */
8284
	if (err == -NFS4ERR_WRONGSEC || err == -ENOTSUPP) {
8285 8286 8287 8288 8289 8290
		err = nfs4_find_root_sec(server, fhandle, info);
		goto out_freepage;
	}
	if (err)
		goto out_freepage;

8291 8292 8293 8294 8295 8296 8297 8298 8299 8300 8301 8302 8303 8304 8305
	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;
		}

8306 8307 8308
		if (!nfs_auth_info_match(&server->auth_info, flavor))
			flavor = RPC_AUTH_MAXFLAVOR;

8309 8310 8311 8312 8313 8314 8315 8316 8317 8318
		if (flavor != RPC_AUTH_MAXFLAVOR) {
			err = nfs4_lookup_root_sec(server, fhandle,
						   info, flavor);
			if (!err)
				break;
		}
	}

	if (flavor == RPC_AUTH_MAXFLAVOR)
		err = -EPERM;
8319 8320 8321 8322 8323 8324 8325 8326

out_freepage:
	put_page(page);
	if (err == -EACCES)
		return -EPERM;
out:
	return err;
}
8327

8328 8329 8330
static int _nfs41_test_stateid(struct nfs_server *server,
		nfs4_stateid *stateid,
		struct rpc_cred *cred)
B
Bryan Schumaker 已提交
8331 8332 8333
{
	int status;
	struct nfs41_test_stateid_args args = {
8334
		.stateid = stateid,
B
Bryan Schumaker 已提交
8335 8336 8337 8338 8339 8340
	};
	struct nfs41_test_stateid_res res;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_TEST_STATEID],
		.rpc_argp = &args,
		.rpc_resp = &res,
8341
		.rpc_cred = cred,
B
Bryan Schumaker 已提交
8342
	};
8343 8344 8345 8346
	struct rpc_clnt *rpc_client = server->client;

	nfs4_state_protect(server->nfs_client, NFS_SP4_MACH_CRED_STATEID,
		&rpc_client, &msg);
8347

8348
	dprintk("NFS call  test_stateid %p\n", stateid);
8349
	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0);
8350
	nfs4_set_sequence_privileged(&args.seq_args);
8351
	status = nfs4_call_sync_sequence(rpc_client, server, &msg,
8352
			&args.seq_args, &res.seq_res);
8353 8354
	if (status != NFS_OK) {
		dprintk("NFS reply test_stateid: failed, %d\n", status);
8355
		return status;
8356 8357
	}
	dprintk("NFS reply test_stateid: succeeded, %d\n", -res.status);
8358
	return -res.status;
B
Bryan Schumaker 已提交
8359 8360
}

8361 8362 8363 8364 8365
/**
 * nfs41_test_stateid - perform a TEST_STATEID operation
 *
 * @server: server / transport on which to perform the operation
 * @stateid: state ID to test
8366
 * @cred: credential
8367 8368 8369 8370 8371
 *
 * 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.
 */
8372 8373 8374
static int nfs41_test_stateid(struct nfs_server *server,
		nfs4_stateid *stateid,
		struct rpc_cred *cred)
B
Bryan Schumaker 已提交
8375 8376 8377 8378
{
	struct nfs4_exception exception = { };
	int err;
	do {
8379
		err = _nfs41_test_stateid(server, stateid, cred);
8380 8381 8382
		if (err != -NFS4ERR_DELAY)
			break;
		nfs4_handle_exception(server, err, &exception);
B
Bryan Schumaker 已提交
8383 8384 8385
	} while (exception.retry);
	return err;
}
B
Bryan Schumaker 已提交
8386

8387 8388 8389
struct nfs_free_stateid_data {
	struct nfs_server *server;
	struct nfs41_free_stateid_args args;
B
Bryan Schumaker 已提交
8390
	struct nfs41_free_stateid_res res;
8391 8392 8393 8394 8395 8396 8397 8398 8399 8400 8401 8402 8403 8404 8405 8406 8407 8408 8409
};

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:
8410
		if (nfs4_async_handle_error(task, data->server, NULL, NULL) == -EAGAIN)
8411 8412 8413 8414 8415 8416 8417 8418 8419
			rpc_restart_call_prepare(task);
	}
}

static void nfs41_free_stateid_release(void *calldata)
{
	kfree(calldata);
}

8420
static const struct rpc_call_ops nfs41_free_stateid_ops = {
8421 8422 8423 8424 8425 8426 8427
	.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,
8428
		struct rpc_cred *cred,
8429 8430
		bool privileged)
{
B
Bryan Schumaker 已提交
8431 8432
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FREE_STATEID],
8433
		.rpc_cred = cred,
B
Bryan Schumaker 已提交
8434
	};
8435 8436 8437 8438 8439 8440 8441
	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
8442

8443 8444 8445
	nfs4_state_protect(server->nfs_client, NFS_SP4_MACH_CRED_STATEID,
		&task_setup.rpc_client, &msg);

8446
	dprintk("NFS call  free_stateid %p\n", stateid);
8447 8448 8449 8450 8451 8452 8453 8454 8455 8456
	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;
8457
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 0);
8458 8459 8460 8461
	if (privileged)
		nfs4_set_sequence_privileged(&data->args.seq_args);

	return rpc_run_task(&task_setup);
B
Bryan Schumaker 已提交
8462 8463
}

8464 8465 8466 8467 8468
/**
 * nfs41_free_stateid - perform a FREE_STATEID operation
 *
 * @server: server / transport on which to perform the operation
 * @stateid: state ID to release
8469
 * @cred: credential
8470 8471 8472 8473
 *
 * Returns NFS_OK if the server freed "stateid".  Otherwise a
 * negative NFS4ERR value is returned.
 */
8474 8475 8476
static int nfs41_free_stateid(struct nfs_server *server,
		nfs4_stateid *stateid,
		struct rpc_cred *cred)
B
Bryan Schumaker 已提交
8477
{
8478 8479 8480
	struct rpc_task *task;
	int ret;

8481
	task = _nfs41_free_stateid(server, stateid, cred, true);
8482 8483 8484 8485 8486 8487 8488
	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
8489
}
8490

8491 8492
static void
nfs41_free_lock_state(struct nfs_server *server, struct nfs4_lock_state *lsp)
8493 8494
{
	struct rpc_task *task;
8495
	struct rpc_cred *cred = lsp->ls_state->owner->so_cred;
8496

8497
	task = _nfs41_free_stateid(server, &lsp->ls_stateid, cred, false);
8498 8499
	nfs4_free_lock_state(server, lsp);
	if (IS_ERR(task))
8500
		return;
8501 8502 8503
	rpc_put_task(task);
}

8504 8505 8506
static bool nfs41_match_stateid(const nfs4_stateid *s1,
		const nfs4_stateid *s2)
{
8507
	if (memcmp(s1->other, s2->other, sizeof(s1->other)) != 0)
8508 8509
		return false;

8510
	if (s1->seqid == s2->seqid)
8511
		return true;
8512
	if (s1->seqid == 0 || s2->seqid == 0)
8513 8514 8515 8516 8517
		return true;

	return false;
}

8518 8519
#endif /* CONFIG_NFS_V4_1 */

8520 8521 8522
static bool nfs4_match_stateid(const nfs4_stateid *s1,
		const nfs4_stateid *s2)
{
8523
	return nfs4_stateid_match(s1, s2);
8524 8525 8526
}


8527
static const struct nfs4_state_recovery_ops nfs40_reboot_recovery_ops = {
8528
	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
8529
	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
L
Linus Torvalds 已提交
8530 8531
	.recover_open	= nfs4_open_reclaim,
	.recover_lock	= nfs4_lock_reclaim,
8532
	.establish_clid = nfs4_init_clientid,
8533
	.detect_trunking = nfs40_discover_server_trunking,
L
Linus Torvalds 已提交
8534 8535
};

8536
#if defined(CONFIG_NFS_V4_1)
8537
static const struct nfs4_state_recovery_ops nfs41_reboot_recovery_ops = {
8538 8539 8540 8541
	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
	.recover_open	= nfs4_open_reclaim,
	.recover_lock	= nfs4_lock_reclaim,
8542
	.establish_clid = nfs41_init_clientid,
8543
	.reclaim_complete = nfs41_proc_reclaim_complete,
8544
	.detect_trunking = nfs41_discover_server_trunking,
8545 8546 8547
};
#endif /* CONFIG_NFS_V4_1 */

8548
static const struct nfs4_state_recovery_ops nfs40_nograce_recovery_ops = {
8549 8550
	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
8551
	.recover_open	= nfs40_open_expired,
8552 8553 8554 8555 8556
	.recover_lock	= nfs4_lock_expired,
	.establish_clid = nfs4_init_clientid,
};

#if defined(CONFIG_NFS_V4_1)
8557
static const struct nfs4_state_recovery_ops nfs41_nograce_recovery_ops = {
8558
	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
8559
	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
8560 8561
	.recover_open	= nfs41_open_expired,
	.recover_lock	= nfs41_lock_expired,
8562
	.establish_clid = nfs41_init_clientid,
L
Linus Torvalds 已提交
8563
};
8564
#endif /* CONFIG_NFS_V4_1 */
L
Linus Torvalds 已提交
8565

8566
static const struct nfs4_state_maintenance_ops nfs40_state_renewal_ops = {
B
Benny Halevy 已提交
8567
	.sched_state_renewal = nfs4_proc_async_renew,
8568
	.get_state_renewal_cred_locked = nfs4_get_renew_cred_locked,
8569
	.renew_lease = nfs4_proc_renew,
B
Benny Halevy 已提交
8570 8571 8572
};

#if defined(CONFIG_NFS_V4_1)
8573
static const struct nfs4_state_maintenance_ops nfs41_state_renewal_ops = {
B
Benny Halevy 已提交
8574
	.sched_state_renewal = nfs41_proc_async_sequence,
8575
	.get_state_renewal_cred_locked = nfs4_get_machine_cred_locked,
8576
	.renew_lease = nfs4_proc_sequence,
B
Benny Halevy 已提交
8577 8578 8579
};
#endif

8580
static const struct nfs4_mig_recovery_ops nfs40_mig_recovery_ops = {
8581
	.get_locations = _nfs40_proc_get_locations,
8582
	.fsid_present = _nfs40_proc_fsid_present,
8583 8584 8585 8586
};

#if defined(CONFIG_NFS_V4_1)
static const struct nfs4_mig_recovery_ops nfs41_mig_recovery_ops = {
8587
	.get_locations = _nfs41_proc_get_locations,
8588
	.fsid_present = _nfs41_proc_fsid_present,
8589 8590 8591
};
#endif	/* CONFIG_NFS_V4_1 */

8592 8593
static const struct nfs4_minor_version_ops nfs_v4_0_minor_ops = {
	.minor_version = 0,
8594 8595 8596
	.init_caps = NFS_CAP_READDIRPLUS
		| NFS_CAP_ATOMIC_OPEN
		| NFS_CAP_POSIX_LOCK,
8597 8598
	.init_client = nfs40_init_client,
	.shutdown_client = nfs40_shutdown_client,
8599
	.match_stateid = nfs4_match_stateid,
8600
	.find_root_sec = nfs4_find_root_sec,
8601
	.free_lock_state = nfs4_release_lockowner,
8602
	.alloc_seqid = nfs_alloc_seqid,
8603
	.call_sync_ops = &nfs40_call_sync_ops,
8604 8605 8606
	.reboot_recovery_ops = &nfs40_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs40_nograce_recovery_ops,
	.state_renewal_ops = &nfs40_state_renewal_ops,
8607
	.mig_recovery_ops = &nfs40_mig_recovery_ops,
8608 8609 8610
};

#if defined(CONFIG_NFS_V4_1)
8611 8612 8613 8614 8615 8616
static struct nfs_seqid *
nfs_alloc_no_seqid(struct nfs_seqid_counter *arg1, gfp_t arg2)
{
	return NULL;
}

8617 8618
static const struct nfs4_minor_version_ops nfs_v4_1_minor_ops = {
	.minor_version = 1,
8619 8620
	.init_caps = NFS_CAP_READDIRPLUS
		| NFS_CAP_ATOMIC_OPEN
8621
		| NFS_CAP_POSIX_LOCK
8622
		| NFS_CAP_STATEID_NFSV41
A
Anna Schumaker 已提交
8623
		| NFS_CAP_ATOMIC_OPEN_V1,
8624 8625
	.init_client = nfs41_init_client,
	.shutdown_client = nfs41_shutdown_client,
8626
	.match_stateid = nfs41_match_stateid,
8627
	.find_root_sec = nfs41_find_root_sec,
8628
	.free_lock_state = nfs41_free_lock_state,
8629
	.alloc_seqid = nfs_alloc_no_seqid,
8630
	.call_sync_ops = &nfs41_call_sync_ops,
8631 8632 8633
	.reboot_recovery_ops = &nfs41_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs41_nograce_recovery_ops,
	.state_renewal_ops = &nfs41_state_renewal_ops,
8634
	.mig_recovery_ops = &nfs41_mig_recovery_ops,
8635 8636 8637
};
#endif

8638 8639 8640
#if defined(CONFIG_NFS_V4_2)
static const struct nfs4_minor_version_ops nfs_v4_2_minor_ops = {
	.minor_version = 2,
8641 8642 8643 8644
	.init_caps = NFS_CAP_READDIRPLUS
		| NFS_CAP_ATOMIC_OPEN
		| NFS_CAP_POSIX_LOCK
		| NFS_CAP_STATEID_NFSV41
A
Anna Schumaker 已提交
8645
		| NFS_CAP_ATOMIC_OPEN_V1
A
Anna Schumaker 已提交
8646
		| NFS_CAP_ALLOCATE
A
Anna Schumaker 已提交
8647
		| NFS_CAP_DEALLOCATE
8648 8649
		| NFS_CAP_SEEK
		| NFS_CAP_LAYOUTSTATS,
8650 8651
	.init_client = nfs41_init_client,
	.shutdown_client = nfs41_shutdown_client,
8652 8653
	.match_stateid = nfs41_match_stateid,
	.find_root_sec = nfs41_find_root_sec,
8654
	.free_lock_state = nfs41_free_lock_state,
8655
	.call_sync_ops = &nfs41_call_sync_ops,
8656
	.alloc_seqid = nfs_alloc_no_seqid,
8657 8658 8659
	.reboot_recovery_ops = &nfs41_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs41_nograce_recovery_ops,
	.state_renewal_ops = &nfs41_state_renewal_ops,
8660
	.mig_recovery_ops = &nfs41_mig_recovery_ops,
8661 8662 8663
};
#endif

8664 8665 8666 8667 8668
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
8669 8670 8671
#if defined(CONFIG_NFS_V4_2)
	[2] = &nfs_v4_2_minor_ops,
#endif
8672 8673
};

8674
static const struct inode_operations nfs4_dir_inode_operations = {
8675 8676 8677 8678 8679 8680 8681 8682 8683 8684 8685 8686 8687 8688 8689 8690 8691 8692 8693
	.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,
};

8694
static const struct inode_operations nfs4_file_inode_operations = {
8695 8696 8697
	.permission	= nfs_permission,
	.getattr	= nfs_getattr,
	.setattr	= nfs_setattr,
8698 8699 8700 8701
	.getxattr	= generic_getxattr,
	.setxattr	= generic_setxattr,
	.listxattr	= generic_listxattr,
	.removexattr	= generic_removexattr,
8702 8703
};

D
David Howells 已提交
8704
const struct nfs_rpc_ops nfs_v4_clientops = {
L
Linus Torvalds 已提交
8705 8706 8707
	.version	= 4,			/* protocol version */
	.dentry_ops	= &nfs4_dentry_operations,
	.dir_inode_ops	= &nfs4_dir_inode_operations,
8708
	.file_inode_ops	= &nfs4_file_inode_operations,
8709
	.file_ops	= &nfs4_file_operations,
L
Linus Torvalds 已提交
8710
	.getroot	= nfs4_proc_get_root,
B
Bryan Schumaker 已提交
8711
	.submount	= nfs4_submount,
B
Bryan Schumaker 已提交
8712
	.try_mount	= nfs4_try_mount,
L
Linus Torvalds 已提交
8713 8714 8715 8716 8717 8718 8719 8720
	.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,
8721
	.unlink_rpc_prepare = nfs4_proc_unlink_rpc_prepare,
L
Linus Torvalds 已提交
8722
	.unlink_done	= nfs4_proc_unlink_done,
8723
	.rename_setup	= nfs4_proc_rename_setup,
8724
	.rename_rpc_prepare = nfs4_proc_rename_rpc_prepare,
8725
	.rename_done	= nfs4_proc_rename_done,
L
Linus Torvalds 已提交
8726 8727 8728 8729 8730 8731 8732 8733 8734
	.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,
8735
	.set_capabilities = nfs4_server_capabilities,
L
Linus Torvalds 已提交
8736
	.decode_dirent	= nfs4_decode_dirent,
8737
	.pgio_rpc_prepare = nfs4_proc_pgio_rpc_prepare,
L
Linus Torvalds 已提交
8738
	.read_setup	= nfs4_proc_read_setup,
T
Trond Myklebust 已提交
8739
	.read_done	= nfs4_read_done,
L
Linus Torvalds 已提交
8740
	.write_setup	= nfs4_proc_write_setup,
8741
	.write_done	= nfs4_write_done,
L
Linus Torvalds 已提交
8742
	.commit_setup	= nfs4_proc_commit_setup,
8743
	.commit_rpc_prepare = nfs4_proc_commit_rpc_prepare,
8744
	.commit_done	= nfs4_commit_done,
L
Linus Torvalds 已提交
8745
	.lock		= nfs4_proc_lock,
8746
	.clear_acl_cache = nfs4_zap_acl_attr,
T
Trond Myklebust 已提交
8747
	.close_context  = nfs4_close_context,
8748
	.open_context	= nfs4_atomic_open,
8749
	.have_delegation = nfs4_have_delegation,
8750
	.return_delegation = nfs4_inode_return_delegation,
8751
	.alloc_client	= nfs4_alloc_client,
8752
	.init_client	= nfs4_init_client,
8753
	.free_client	= nfs4_free_client,
8754 8755
	.create_server	= nfs4_create_server,
	.clone_server	= nfs_clone_server,
L
Linus Torvalds 已提交
8756 8757
};

8758 8759 8760 8761 8762 8763 8764 8765 8766
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,
8767 8768 8769
#ifdef CONFIG_NFS_V4_SECURITY_LABEL
	&nfs4_xattr_nfs4_label_handler,
#endif
8770 8771 8772
	NULL
};

L
Linus Torvalds 已提交
8773 8774 8775 8776 8777
/*
 * Local variables:
 *  c-basic-offset: 8
 * End:
 */