nfs4proc.c 215.0 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 41
/*
 *  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>
#include <linux/string.h>
42 43
#include <linux/ratelimit.h>
#include <linux/printk.h>
44
#include <linux/slab.h>
L
Linus Torvalds 已提交
45 46 47 48 49
#include <linux/sunrpc/clnt.h>
#include <linux/nfs.h>
#include <linux/nfs4.h>
#include <linux/nfs_fs.h>
#include <linux/nfs_page.h>
50
#include <linux/nfs_mount.h>
L
Linus Torvalds 已提交
51
#include <linux/namei.h>
52
#include <linux/mount.h>
B
Benny Halevy 已提交
53
#include <linux/module.h>
54
#include <linux/nfs_idmap.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 "nfs4session.h"
D
David Howells 已提交
67
#include "fscache.h"
L
Linus Torvalds 已提交
68

69 70
#include "nfs4trace.h"

L
Linus Torvalds 已提交
71 72
#define NFSDBG_FACILITY		NFSDBG_PROC

73
#define NFS4_POLL_RETRY_MIN	(HZ/10)
L
Linus Torvalds 已提交
74 75
#define NFS4_POLL_RETRY_MAX	(15*HZ)

76
struct nfs4_opendata;
77
static int _nfs4_proc_open(struct nfs4_opendata *data);
78
static int _nfs4_recover_proc_open(struct nfs4_opendata *data);
L
Linus Torvalds 已提交
79
static int nfs4_do_fsinfo(struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *);
80
static int nfs4_async_handle_error(struct rpc_task *, const struct nfs_server *, struct nfs4_state *);
81
static void nfs_fixup_referral_attributes(struct nfs_fattr *fattr);
82 83
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);
84 85
static int nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
			    struct nfs_fattr *fattr, struct iattr *sattr,
86 87
			    struct nfs4_state *state, struct nfs4_label *ilabel,
			    struct nfs4_label *olabel);
88
#ifdef CONFIG_NFS_V4_1
89 90 91 92
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 *);
93
#endif
94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143

#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;

	if (NFS_SERVER(dir)->nfs_client->cl_minorversion < 2)
		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
144
/* Prevent leaks of NFSv4 errors into userland */
145
static int nfs4_map_errors(int err)
L
Linus Torvalds 已提交
146
{
147 148 149 150
	if (err >= -1000)
		return err;
	switch (err) {
	case -NFS4ERR_RESOURCE:
151 152
	case -NFS4ERR_LAYOUTTRYLATER:
	case -NFS4ERR_RECALLCONFLICT:
153
		return -EREMOTEIO;
154
	case -NFS4ERR_WRONGSEC:
155
	case -NFS4ERR_WRONG_CRED:
156
		return -EPERM;
157 158 159
	case -NFS4ERR_BADOWNER:
	case -NFS4ERR_BADNAME:
		return -EINVAL;
160 161
	case -NFS4ERR_SHARE_DENIED:
		return -EACCES;
162 163
	case -NFS4ERR_MINOR_VERS_MISMATCH:
		return -EPROTONOSUPPORT;
164 165
	case -NFS4ERR_ACCESS:
		return -EACCES;
166 167
	case -NFS4ERR_FILE_OPEN:
		return -EBUSY;
168
	default:
L
Linus Torvalds 已提交
169
		dprintk("%s could not handle NFSv4 error %d\n",
170
				__func__, -err);
171
		break;
L
Linus Torvalds 已提交
172
	}
173
	return -EIO;
L
Linus Torvalds 已提交
174 175 176 177 178
}

/*
 * This is our standard bitmap for GETATTR requests.
 */
179
const u32 nfs4_fattr_bitmap[3] = {
L
Linus Torvalds 已提交
180 181 182 183 184 185 186 187 188 189 190 191 192
	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
193 194 195 196
	| FATTR4_WORD1_TIME_MODIFY,
#ifdef CONFIG_NFS_V4_SECURITY_LABEL
	FATTR4_WORD2_SECURITY_LABEL
#endif
L
Linus Torvalds 已提交
197 198
};

199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216
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
};

217 218 219 220 221 222
static const u32 nfs4_open_noattr_bitmap[3] = {
	FATTR4_WORD0_TYPE
	| FATTR4_WORD0_CHANGE
	| FATTR4_WORD0_FILEID,
};

223
const u32 nfs4_statfs_bitmap[3] = {
L
Linus Torvalds 已提交
224 225 226 227 228 229 230 231
	FATTR4_WORD0_FILES_AVAIL
	| FATTR4_WORD0_FILES_FREE
	| FATTR4_WORD0_FILES_TOTAL,
	FATTR4_WORD1_SPACE_AVAIL
	| FATTR4_WORD1_SPACE_FREE
	| FATTR4_WORD1_SPACE_TOTAL
};

232
const u32 nfs4_pathconf_bitmap[3] = {
L
Linus Torvalds 已提交
233 234 235 236 237
	FATTR4_WORD0_MAXLINK
	| FATTR4_WORD0_MAXNAME,
	0
};

238
const u32 nfs4_fsinfo_bitmap[3] = { FATTR4_WORD0_MAXFILESIZE
L
Linus Torvalds 已提交
239 240 241
			| FATTR4_WORD0_MAXREAD
			| FATTR4_WORD0_MAXWRITE
			| FATTR4_WORD0_LEASE_TIME,
242
			FATTR4_WORD1_TIME_DELTA
243 244
			| FATTR4_WORD1_FS_LAYOUT_TYPES,
			FATTR4_WORD2_LAYOUT_BLKSIZE
L
Linus Torvalds 已提交
245 246
};

247
const u32 nfs4_fs_locations_bitmap[3] = {
248 249 250 251 252 253 254 255 256 257 258 259 260 261 262
	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
263
	| FATTR4_WORD1_MOUNTED_ON_FILEID,
264 265
};

A
Al Viro 已提交
266
static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dentry,
L
Linus Torvalds 已提交
267 268
		struct nfs4_readdir_arg *readdir)
{
269
	__be32 *start, *p;
L
Linus Torvalds 已提交
270 271

	if (cookie > 2) {
272
		readdir->cookie = cookie;
L
Linus Torvalds 已提交
273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288
		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.
	 */
289
	start = p = kmap_atomic(*readdir->pages);
L
Linus Torvalds 已提交
290 291 292 293 294 295 296 297 298 299 300
	
	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 */
P
Peter Staubach 已提交
301
		p = xdr_encode_hyper(p, NFS_FILEID(dentry->d_inode));
L
Linus Torvalds 已提交
302 303 304 305 306 307 308 309 310 311 312
	}
	
	*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 */
P
Peter Staubach 已提交
313
	p = xdr_encode_hyper(p, NFS_FILEID(dentry->d_parent->d_inode));
L
Linus Torvalds 已提交
314 315 316

	readdir->pgbase = (char *)p - (char *)start;
	readdir->count -= readdir->pgbase;
317
	kunmap_atomic(start);
L
Linus Torvalds 已提交
318 319
}

320 321 322 323 324 325 326 327 328 329
static int nfs4_delay(struct rpc_clnt *clnt, long *timeout)
{
	int res = 0;

	might_sleep();

	if (*timeout <= 0)
		*timeout = NFS4_POLL_RETRY_MIN;
	if (*timeout > NFS4_POLL_RETRY_MAX)
		*timeout = NFS4_POLL_RETRY_MAX;
330
	freezable_schedule_timeout_killable_unsafe(*timeout);
331 332 333 334 335 336 337 338 339
	if (fatal_signal_pending(current))
		res = -ERESTARTSYS;
	*timeout <<= 1;
	return res;
}

/* This is the error handling routine for processes that are allowed
 * to sleep.
 */
340
static int nfs4_handle_exception(struct nfs_server *server, int errorcode, struct nfs4_exception *exception)
341 342
{
	struct nfs_client *clp = server->nfs_client;
343
	struct nfs4_state *state = exception->state;
344
	struct inode *inode = exception->inode;
345 346 347 348 349 350
	int ret = errorcode;

	exception->retry = 0;
	switch(errorcode) {
		case 0:
			return 0;
351
		case -NFS4ERR_OPENMODE:
352
			if (inode && nfs4_have_delegation(inode, FMODE_READ)) {
353
				nfs4_inode_return_delegation(inode);
354 355 356 357 358
				exception->retry = 1;
				return 0;
			}
			if (state == NULL)
				break;
359 360 361
			ret = nfs4_schedule_stateid_recovery(server, state);
			if (ret < 0)
				break;
362
			goto wait_on_recovery;
363
		case -NFS4ERR_DELEG_REVOKED:
364 365
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
366 367 368 369 370
			if (inode != NULL && nfs4_have_delegation(inode, FMODE_READ)) {
				nfs_remove_bad_delegation(inode);
				exception->retry = 1;
				break;
			}
371 372
			if (state == NULL)
				break;
373 374 375
			ret = nfs4_schedule_stateid_recovery(server, state);
			if (ret < 0)
				break;
376
			goto wait_on_recovery;
377
		case -NFS4ERR_EXPIRED:
378 379 380 381 382
			if (state != NULL) {
				ret = nfs4_schedule_stateid_recovery(server, state);
				if (ret < 0)
					break;
			}
383
		case -NFS4ERR_STALE_STATEID:
384
		case -NFS4ERR_STALE_CLIENTID:
385 386
			nfs4_schedule_lease_recovery(clp);
			goto wait_on_recovery;
387
#if defined(CONFIG_NFS_V4_1)
388 389 390 391 392 393 394 395 396
		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);
397
			nfs4_schedule_session_recovery(clp->cl_session, errorcode);
398
			goto wait_on_recovery;
399
#endif /* defined(CONFIG_NFS_V4_1) */
400
		case -NFS4ERR_FILE_OPEN:
401 402 403 404 405 406 407
			if (exception->timeout > HZ) {
				/* We have retried a decent amount, time to
				 * fail
				 */
				ret = -EBUSY;
				break;
			}
408 409 410 411 412
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			ret = nfs4_delay(server->client, &exception->timeout);
			if (ret != 0)
				break;
413
		case -NFS4ERR_RETRY_UNCACHED_REP:
414 415
		case -NFS4ERR_OLD_STATEID:
			exception->retry = 1;
416 417 418 419 420 421 422 423 424 425 426 427 428
			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);
			}
429 430 431
	}
	/* We failed to handle the error */
	return nfs4_map_errors(ret);
432
wait_on_recovery:
433 434 435 436
	ret = nfs4_wait_clnt_recover(clp);
	if (ret == 0)
		exception->retry = 1;
	return ret;
437 438
}

439 440 441 442 443 444 445 446 447 448 449 450 451 452
/*
 * 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;
}
453

454
static void do_renew_lease(struct nfs_client *clp, unsigned long timestamp)
L
Linus Torvalds 已提交
455 456 457 458 459 460 461
{
	spin_lock(&clp->cl_lock);
	if (time_before(clp->cl_last_renewal,timestamp))
		clp->cl_last_renewal = timestamp;
	spin_unlock(&clp->cl_lock);
}

462 463 464 465 466
static void renew_lease(const struct nfs_server *server, unsigned long timestamp)
{
	do_renew_lease(server->nfs_client, timestamp);
}

467 468 469 470 471 472
struct nfs4_call_sync_data {
	const struct nfs_server *seq_server;
	struct nfs4_sequence_args *seq_args;
	struct nfs4_sequence_res *seq_res;
};

473 474 475 476 477 478 479 480 481 482 483 484 485 486 487
static void nfs4_init_sequence(struct nfs4_sequence_args *args,
			       struct nfs4_sequence_res *res, int cache_reply)
{
	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;
}

C
Chuck Lever 已提交
488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548
static int nfs40_setup_sequence(const struct nfs_server *server,
				struct nfs4_sequence_args *args,
				struct nfs4_sequence_res *res,
				struct rpc_task *task)
{
	struct nfs4_slot_table *tbl = server->nfs_client->cl_slot_tbl;
	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;
}

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;

	if (!RPC_WAS_SENT(task))
		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
549 550
#if defined(CONFIG_NFS_V4_1)

551
static void nfs41_sequence_free_slot(struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
552
{
553
	struct nfs4_session *session;
A
Andy Adamson 已提交
554
	struct nfs4_slot_table *tbl;
555
	bool send_new_highest_used_slotid = false;
A
Andy Adamson 已提交
556

557
	if (!res->sr_slot) {
A
Andy Adamson 已提交
558 559
		/* just wake up the next guy waiting since
		 * we may have not consumed a slot after all */
A
Andy Adamson 已提交
560
		dprintk("%s: No slot\n", __func__);
561
		return;
A
Andy Adamson 已提交
562
	}
563 564
	tbl = res->sr_slot->table;
	session = tbl->session;
565

566
	spin_lock(&tbl->slot_tbl_lock);
567 568 569 570 571 572
	/* 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;

573 574 575 576
	if (nfs41_wake_and_assign_slot(tbl, res->sr_slot)) {
		send_new_highest_used_slotid = false;
		goto out_unlock;
	}
T
Trond Myklebust 已提交
577
	nfs4_free_slot(tbl, res->sr_slot);
578 579 580

	if (tbl->highest_used_slotid != NFS4_NO_SLOT)
		send_new_highest_used_slotid = false;
581
out_unlock:
582
	spin_unlock(&tbl->slot_tbl_lock);
583
	res->sr_slot = NULL;
584 585
	if (send_new_highest_used_slotid)
		nfs41_server_notify_highest_slotid_update(session->clp);
A
Andy Adamson 已提交
586 587
}

588
static int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
589
{
590
	struct nfs4_session *session;
591
	struct nfs4_slot *slot;
592
	struct nfs_client *clp;
593
	bool interrupted = false;
594
	int ret = 1;
A
Andy Adamson 已提交
595

596 597
	/* don't increment the sequence number if the task wasn't sent */
	if (!RPC_WAS_SENT(task))
A
Andy Adamson 已提交
598 599
		goto out;

600
	slot = res->sr_slot;
601
	session = slot->table->session;
602

603 604 605 606 607
	if (slot->interrupted) {
		slot->interrupted = 0;
		interrupted = true;
	}

608
	trace_nfs4_sequence_done(session, res);
A
Andy Adamson 已提交
609
	/* Check the SEQUENCE operation status */
610 611
	switch (res->sr_status) {
	case 0:
A
Andy Adamson 已提交
612
		/* Update the slot's sequence and clientid lease timer */
613
		++slot->seq_nr;
614
		clp = session->clp;
615
		do_renew_lease(clp, res->sr_timestamp);
616
		/* Check sequence flags */
617 618
		if (res->sr_status_flags != 0)
			nfs4_schedule_lease_recovery(clp);
619
		nfs41_update_target_slotid(slot->table, slot, res);
620
		break;
621 622 623 624 625 626 627 628 629
	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;
630 631 632 633 634
	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.
		 */
635
		dprintk("%s: slot=%u seq=%u: Operation in progress\n",
636
			__func__,
637
			slot->slot_nr,
638
			slot->seq_nr);
639
		goto out_retry;
640 641 642 643 644
	case -NFS4ERR_BADSLOT:
		/*
		 * The slot id we used was probably retired. Try again
		 * using a different slot id.
		 */
645 646
		goto retry_nowait;
	case -NFS4ERR_SEQ_MISORDERED:
647 648 649 650 651 652 653 654
		/*
		 * Was the last operation on this sequence interrupted?
		 * If so, retry after bumping the sequence number.
		 */
		if (interrupted) {
			++slot->seq_nr;
			goto retry_nowait;
		}
655 656 657 658
		/*
		 * Could this slot have been previously retired?
		 * If so, then the server may be expecting seq_nr = 1!
		 */
659 660 661 662 663
		if (slot->seq_nr != 1) {
			slot->seq_nr = 1;
			goto retry_nowait;
		}
		break;
664 665 666
	case -NFS4ERR_SEQ_FALSE_RETRY:
		++slot->seq_nr;
		goto retry_nowait;
667 668
	default:
		/* Just update the slot sequence no. */
669
		++slot->seq_nr;
A
Andy Adamson 已提交
670 671 672 673
	}
out:
	/* The session may be reset by one of the error handlers. */
	dprintk("%s: Error %d free the slot \n", __func__, res->sr_status);
674
	nfs41_sequence_free_slot(res);
675
	return ret;
676 677 678 679 680 681
retry_nowait:
	if (rpc_restart_call_prepare(task)) {
		task->tk_status = 0;
		ret = 0;
	}
	goto out;
682
out_retry:
683
	if (!rpc_restart_call(task))
684 685 686
		goto out;
	rpc_delay(task, NFS4_POLL_RETRY_MAX);
	return 0;
A
Andy Adamson 已提交
687 688
}

689 690
static int nfs4_sequence_done(struct rpc_task *task,
			       struct nfs4_sequence_res *res)
691
{
692
	if (res->sr_slot == NULL)
693
		return 1;
C
Chuck Lever 已提交
694 695
	if (!res->sr_slot->table->session)
		return nfs40_sequence_done(task, res);
696
	return nfs41_sequence_done(task, res);
697 698
}

A
Andy Adamson 已提交
699
int nfs41_setup_sequence(struct nfs4_session *session,
A
Andy Adamson 已提交
700 701 702 703
				struct nfs4_sequence_args *args,
				struct nfs4_sequence_res *res,
				struct rpc_task *task)
{
A
Andy Adamson 已提交
704 705 706 707
	struct nfs4_slot *slot;
	struct nfs4_slot_table *tbl;

	dprintk("--> %s\n", __func__);
A
Andy Adamson 已提交
708
	/* slot already allocated? */
709
	if (res->sr_slot != NULL)
710
		goto out_success;
A
Andy Adamson 已提交
711

A
Andy Adamson 已提交
712 713
	tbl = &session->fc_slot_table;

714 715
	task->tk_timeout = 0;

A
Andy Adamson 已提交
716
	spin_lock(&tbl->slot_tbl_lock);
717
	if (test_bit(NFS4_SLOT_TBL_DRAINING, &tbl->slot_tbl_state) &&
718
	    !args->sa_privileged) {
719 720
		/* The state manager will wait until the slot table is empty */
		dprintk("%s session is draining\n", __func__);
721
		goto out_sleep;
722 723
	}

724
	slot = nfs4_alloc_slot(tbl);
725 726 727 728
	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
729
		dprintk("<-- %s: no free slots\n", __func__);
730
		goto out_sleep;
A
Andy Adamson 已提交
731 732 733
	}
	spin_unlock(&tbl->slot_tbl_lock);

734
	args->sa_slot = slot;
A
Andy Adamson 已提交
735

736
	dprintk("<-- %s slotid=%u seqid=%u\n", __func__,
737
			slot->slot_nr, slot->seq_nr);
A
Andy Adamson 已提交
738

739
	res->sr_slot = slot;
740
	res->sr_timestamp = jiffies;
741
	res->sr_status_flags = 0;
A
Andy Adamson 已提交
742 743 744 745 746
	/*
	 * 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;
747
	trace_nfs4_setup_sequence(session, args);
748 749
out_success:
	rpc_call_start(task);
A
Andy Adamson 已提交
750
	return 0;
751
out_sleep:
752 753
	/* Privileged tasks are queued with top priority */
	if (args->sa_privileged)
754 755 756 757
		rpc_sleep_on_priority(&tbl->slot_tbl_waitq, task,
				NULL, RPC_PRIORITY_PRIVILEGED);
	else
		rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
758 759
	spin_unlock(&tbl->slot_tbl_lock);
	return -EAGAIN;
A
Andy Adamson 已提交
760
}
A
Andy Adamson 已提交
761
EXPORT_SYMBOL_GPL(nfs41_setup_sequence);
A
Andy Adamson 已提交
762

763 764 765 766
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
767
{
768
	struct nfs4_session *session = nfs4_get_session(server);
A
Andy Adamson 已提交
769 770
	int ret = 0;

C
Chuck Lever 已提交
771 772
	if (!session)
		return nfs40_setup_sequence(server, args, res, task);
773

774
	dprintk("--> %s clp %p session %p sr_slot %u\n",
775
		__func__, session->clp, session, res->sr_slot ?
776
			res->sr_slot->slot_nr : NFS4_NO_SLOT);
A
Andy Adamson 已提交
777

778
	ret = nfs41_setup_sequence(session, args, res, task);
C
Chuck Lever 已提交
779

A
Andy Adamson 已提交
780 781 782 783 784 785
	dprintk("<-- %s status=%d\n", __func__, ret);
	return ret;
}

static void nfs41_call_sync_prepare(struct rpc_task *task, void *calldata)
{
786
	struct nfs4_call_sync_data *data = calldata;
787
	struct nfs4_session *session = nfs4_get_session(data->seq_server);
A
Andy Adamson 已提交
788

789 790
	dprintk("--> %s data->seq_server %p\n", __func__, data->seq_server);

791
	nfs41_setup_sequence(session, data->seq_args, data->seq_res, task);
A
Andy Adamson 已提交
792 793
}

A
Andy Adamson 已提交
794 795
static void nfs41_call_sync_done(struct rpc_task *task, void *calldata)
{
796
	struct nfs4_call_sync_data *data = calldata;
A
Andy Adamson 已提交
797

798
	nfs41_sequence_done(task, data->seq_res);
A
Andy Adamson 已提交
799 800
}

801
static const struct rpc_call_ops nfs41_call_sync_ops = {
A
Andy Adamson 已提交
802
	.rpc_call_prepare = nfs41_call_sync_prepare,
A
Andy Adamson 已提交
803
	.rpc_call_done = nfs41_call_sync_done,
A
Andy Adamson 已提交
804 805
};

C
Chuck Lever 已提交
806 807
#else	/* !CONFIG_NFS_V4_1 */

808 809 810 811 812
static int nfs4_setup_sequence(const struct nfs_server *server,
			       struct nfs4_sequence_args *args,
			       struct nfs4_sequence_res *res,
			       struct rpc_task *task)
{
C
Chuck Lever 已提交
813
	return nfs40_setup_sequence(server, args, res, task);
814 815 816 817 818
}

static int nfs4_sequence_done(struct rpc_task *task,
			       struct nfs4_sequence_res *res)
{
C
Chuck Lever 已提交
819
	return nfs40_sequence_done(task, res);
820
}
C
Chuck Lever 已提交
821 822

#endif	/* !CONFIG_NFS_V4_1 */
823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841

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,
};

842 843
static int nfs4_call_sync_sequence(struct rpc_clnt *clnt,
				   struct nfs_server *server,
A
Andy Adamson 已提交
844 845
				   struct rpc_message *msg,
				   struct nfs4_sequence_args *args,
846
				   struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
847 848 849
{
	int ret;
	struct rpc_task *task;
850
	struct nfs_client *clp = server->nfs_client;
851
	struct nfs4_call_sync_data data = {
852
		.seq_server = server,
A
Andy Adamson 已提交
853 854 855 856
		.seq_args = args,
		.seq_res = res,
	};
	struct rpc_task_setup task_setup = {
857
		.rpc_client = clnt,
A
Andy Adamson 已提交
858
		.rpc_message = msg,
859
		.callback_ops = clp->cl_mvops->call_sync_ops,
A
Andy Adamson 已提交
860 861 862 863 864 865 866 867 868 869 870 871 872
		.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;
}

873
static
874 875
int nfs4_call_sync(struct rpc_clnt *clnt,
		   struct nfs_server *server,
876 877 878 879 880
		   struct rpc_message *msg,
		   struct nfs4_sequence_args *args,
		   struct nfs4_sequence_res *res,
		   int cache_reply)
{
881
	nfs4_init_sequence(args, res, cache_reply);
882
	return nfs4_call_sync_sequence(clnt, server, msg, args, res);
883
}
A
Andy Adamson 已提交
884

885
static void update_changeattr(struct inode *dir, struct nfs4_change_info *cinfo)
L
Linus Torvalds 已提交
886
{
887
	struct nfs_inode *nfsi = NFS_I(dir);
L
Linus Torvalds 已提交
888

889
	spin_lock(&dir->i_lock);
890
	nfsi->cache_validity |= NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA;
891
	if (!cinfo->atomic || cinfo->before != dir->i_version)
892
		nfs_force_lookup_revalidate(dir);
893
	dir->i_version = cinfo->after;
D
David Howells 已提交
894
	nfs_fscache_invalidate(dir);
895
	spin_unlock(&dir->i_lock);
L
Linus Torvalds 已提交
896 897
}

898
struct nfs4_opendata {
899
	struct kref kref;
900 901
	struct nfs_openargs o_arg;
	struct nfs_openres o_res;
902 903
	struct nfs_open_confirmargs c_arg;
	struct nfs_open_confirmres c_res;
904 905
	struct nfs4_string owner_name;
	struct nfs4_string group_name;
906
	struct nfs_fattr f_attr;
907
	struct nfs4_label *f_label;
908
	struct dentry *dir;
909
	struct dentry *dentry;
910
	struct nfs4_state_owner *owner;
911
	struct nfs4_state *state;
912
	struct iattr attrs;
913
	unsigned long timestamp;
914
	unsigned int rpc_done : 1;
915
	unsigned int is_recover : 1;
916 917
	int rpc_status;
	int cancelled;
918 919
};

920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948
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;
}

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;
	}
}
949 950 951 952

static void nfs4_init_opendata_res(struct nfs4_opendata *p)
{
	p->o_res.f_attr = &p->f_attr;
953
	p->o_res.f_label = p->f_label;
954 955
	p->o_res.seqid = p->o_arg.seqid;
	p->c_res.seqid = p->c_arg.seqid;
956
	p->o_res.server = p->o_arg.server;
957
	p->o_res.access_request = p->o_arg.access;
958
	nfs_fattr_init(&p->f_attr);
959
	nfs_fattr_init_names(&p->f_attr, &p->owner_name, &p->group_name);
960 961
}

962
static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry,
963
		struct nfs4_state_owner *sp, fmode_t fmode, int flags,
964
		const struct iattr *attrs,
965
		struct nfs4_label *label,
966
		enum open_claim_type4 claim,
967
		gfp_t gfp_mask)
968
{
969
	struct dentry *parent = dget_parent(dentry);
970 971 972 973
	struct inode *dir = parent->d_inode;
	struct nfs_server *server = NFS_SERVER(dir);
	struct nfs4_opendata *p;

974
	p = kzalloc(sizeof(*p), gfp_mask);
975 976
	if (p == NULL)
		goto err;
977 978 979 980 981

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

982
	p->o_arg.seqid = nfs_alloc_seqid(&sp->so_seqid, gfp_mask);
983
	if (p->o_arg.seqid == NULL)
984
		goto err_free_label;
985 986
	nfs_sb_active(dentry->d_sb);
	p->dentry = dget(dentry);
987 988 989
	p->dir = parent;
	p->owner = sp;
	atomic_inc(&sp->so_count);
990 991
	p->o_arg.open_flags = flags;
	p->o_arg.fmode = fmode & (FMODE_READ|FMODE_WRITE);
992 993 994 995 996 997 998 999
	/* 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;
	}
1000
	p->o_arg.clientid = server->nfs_client->cl_clientid;
1001 1002
	p->o_arg.id.create_time = ktime_to_ns(sp->so_seqid.create_time);
	p->o_arg.id.uniquifier = sp->so_seqid.owner_id;
1003
	p->o_arg.name = &dentry->d_name;
1004
	p->o_arg.server = server;
1005
	p->o_arg.bitmask = nfs4_bitmask(server, label);
1006
	p->o_arg.open_bitmap = &nfs4_fattr_bitmap[0];
1007
	p->o_arg.label = label;
1008 1009
	p->o_arg.claim = nfs4_map_atomic_open_claim(server, claim);
	switch (p->o_arg.claim) {
1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020
	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:
		p->o_arg.fh = NFS_FH(dentry->d_inode);
	}
1021
	if (attrs != NULL && attrs->ia_valid != 0) {
1022
		__u32 verf[2];
1023

1024 1025
		p->o_arg.u.attrs = &p->attrs;
		memcpy(&p->attrs, attrs, sizeof(p->attrs));
1026 1027 1028 1029 1030

		verf[0] = jiffies;
		verf[1] = current->pid;
		memcpy(p->o_arg.u.verifier.data, verf,
				sizeof(p->o_arg.u.verifier.data));
1031
	}
1032 1033 1034
	p->c_arg.fh = &p->o_res.fh;
	p->c_arg.stateid = &p->o_res.stateid;
	p->c_arg.seqid = p->o_arg.seqid;
1035
	nfs4_init_opendata_res(p);
1036
	kref_init(&p->kref);
1037
	return p;
1038 1039 1040 1041

err_free_label:
	nfs4_label_free(p->f_label);
err_free_p:
1042 1043 1044 1045 1046 1047
	kfree(p);
err:
	dput(parent);
	return NULL;
}

1048
static void nfs4_opendata_free(struct kref *kref)
1049
{
1050 1051
	struct nfs4_opendata *p = container_of(kref,
			struct nfs4_opendata, kref);
1052
	struct super_block *sb = p->dentry->d_sb;
1053 1054

	nfs_free_seqid(p->o_arg.seqid);
1055 1056
	if (p->state != NULL)
		nfs4_put_open_state(p->state);
1057
	nfs4_put_state_owner(p->owner);
1058 1059 1060

	nfs4_label_free(p->f_label);

1061
	dput(p->dir);
1062 1063
	dput(p->dentry);
	nfs_sb_deactive(sb);
1064
	nfs_fattr_free_names(&p->f_attr);
1065 1066 1067 1068 1069 1070 1071
	kfree(p);
}

static void nfs4_opendata_put(struct nfs4_opendata *p)
{
	if (p != NULL)
		kref_put(&p->kref, nfs4_opendata_free);
1072 1073
}

1074 1075 1076 1077 1078 1079 1080 1081
static int nfs4_wait_for_completion_rpc_task(struct rpc_task *task)
{
	int ret;

	ret = rpc_wait_for_completion_task(task);
	return ret;
}

1082
static int can_open_cached(struct nfs4_state *state, fmode_t mode, int open_mode)
1083 1084
{
	int ret = 0;
1085

1086
	if (open_mode & (O_EXCL|O_TRUNC))
1087 1088
		goto out;
	switch (mode & (FMODE_READ|FMODE_WRITE)) {
1089
		case FMODE_READ:
1090 1091
			ret |= test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0
				&& state->n_rdonly != 0;
1092 1093
			break;
		case FMODE_WRITE:
1094 1095
			ret |= test_bit(NFS_O_WRONLY_STATE, &state->flags) != 0
				&& state->n_wronly != 0;
1096 1097
			break;
		case FMODE_READ|FMODE_WRITE:
1098 1099
			ret |= test_bit(NFS_O_RDWR_STATE, &state->flags) != 0
				&& state->n_rdwr != 0;
1100
	}
1101
out:
1102 1103 1104
	return ret;
}

1105
static int can_open_delegated(struct nfs_delegation *delegation, fmode_t fmode)
1106
{
1107 1108
	if (delegation == NULL)
		return 0;
1109
	if ((delegation->type & fmode) != fmode)
1110
		return 0;
1111
	if (test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags))
1112
		return 0;
1113 1114
	if (test_bit(NFS_DELEGATION_RETURNING, &delegation->flags))
		return 0;
1115
	nfs_mark_delegation_referenced(delegation);
1116 1117 1118
	return 1;
}

1119
static void update_open_stateflags(struct nfs4_state *state, fmode_t fmode)
1120
{
1121
	switch (fmode) {
1122 1123 1124 1125 1126 1127 1128 1129 1130
		case FMODE_WRITE:
			state->n_wronly++;
			break;
		case FMODE_READ:
			state->n_rdonly++;
			break;
		case FMODE_READ|FMODE_WRITE:
			state->n_rdwr++;
	}
1131
	nfs4_state_set_mode_locked(state, state->state | fmode);
1132 1133
}

1134
static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
1135 1136
{
	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
1137 1138
		nfs4_stateid_copy(&state->stateid, stateid);
	nfs4_stateid_copy(&state->open_stateid, stateid);
1139
	set_bit(NFS_OPEN_STATE, &state->flags);
1140
	switch (fmode) {
1141 1142 1143 1144 1145 1146 1147 1148 1149
		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);
	}
1150 1151
}

1152
static void nfs_set_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
1153
{
1154
	write_seqlock(&state->seqlock);
1155
	nfs_set_open_stateid_locked(state, stateid, fmode);
1156
	write_sequnlock(&state->seqlock);
1157 1158
}

1159
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
1160
{
1161 1162 1163 1164 1165
	/*
	 * Protect the call to nfs4_state_set_mode_locked and
	 * serialise the stateid update
	 */
	write_seqlock(&state->seqlock);
1166
	if (deleg_stateid != NULL) {
1167
		nfs4_stateid_copy(&state->stateid, deleg_stateid);
1168 1169 1170
		set_bit(NFS_DELEGATED_STATE, &state->flags);
	}
	if (open_stateid != NULL)
1171
		nfs_set_open_stateid_locked(state, open_stateid, fmode);
1172 1173
	write_sequnlock(&state->seqlock);
	spin_lock(&state->owner->so_lock);
1174
	update_open_stateflags(state, fmode);
1175
	spin_unlock(&state->owner->so_lock);
L
Linus Torvalds 已提交
1176 1177
}

1178
static int update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *delegation, fmode_t fmode)
1179 1180 1181 1182 1183
{
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *deleg_cur;
	int ret = 0;

1184
	fmode &= (FMODE_READ|FMODE_WRITE);
1185 1186 1187 1188 1189 1190 1191

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

	spin_lock(&deleg_cur->lock);
1192
	if (rcu_dereference(nfsi->delegation) != deleg_cur ||
1193
	   test_bit(NFS_DELEGATION_RETURNING, &deleg_cur->flags) ||
1194
	    (deleg_cur->type & fmode) != fmode)
1195 1196 1197 1198
		goto no_delegation_unlock;

	if (delegation == NULL)
		delegation = &deleg_cur->stateid;
1199
	else if (!nfs4_stateid_match(&deleg_cur->stateid, delegation))
1200 1201
		goto no_delegation_unlock;

1202
	nfs_mark_delegation_referenced(deleg_cur);
1203
	__update_open_stateid(state, open_stateid, &deleg_cur->stateid, fmode);
1204 1205 1206 1207 1208 1209 1210
	ret = 1;
no_delegation_unlock:
	spin_unlock(&deleg_cur->lock);
no_delegation:
	rcu_read_unlock();

	if (!ret && open_stateid != NULL) {
1211
		__update_open_stateid(state, open_stateid, NULL, fmode);
1212 1213 1214 1215 1216 1217 1218
		ret = 1;
	}

	return ret;
}


1219
static void nfs4_return_incompatible_delegation(struct inode *inode, fmode_t fmode)
1220 1221 1222 1223 1224
{
	struct nfs_delegation *delegation;

	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(inode)->delegation);
1225
	if (delegation == NULL || (delegation->type & fmode) == fmode) {
1226 1227 1228 1229
		rcu_read_unlock();
		return;
	}
	rcu_read_unlock();
1230
	nfs4_inode_return_delegation(inode);
1231 1232
}

1233
static struct nfs4_state *nfs4_try_open_cached(struct nfs4_opendata *opendata)
1234 1235 1236 1237
{
	struct nfs4_state *state = opendata->state;
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *delegation;
1238
	int open_mode = opendata->o_arg.open_flags;
1239
	fmode_t fmode = opendata->o_arg.fmode;
1240 1241 1242 1243
	nfs4_stateid stateid;
	int ret = -EAGAIN;

	for (;;) {
1244
		if (can_open_cached(state, fmode, open_mode)) {
1245
			spin_lock(&state->owner->so_lock);
1246 1247
			if (can_open_cached(state, fmode, open_mode)) {
				update_open_stateflags(state, fmode);
1248 1249 1250 1251 1252
				spin_unlock(&state->owner->so_lock);
				goto out_return_state;
			}
			spin_unlock(&state->owner->so_lock);
		}
1253 1254
		rcu_read_lock();
		delegation = rcu_dereference(nfsi->delegation);
1255
		if (!can_open_delegated(delegation, fmode)) {
1256
			rcu_read_unlock();
1257
			break;
1258
		}
1259
		/* Save the delegation */
1260
		nfs4_stateid_copy(&stateid, &delegation->stateid);
1261
		rcu_read_unlock();
1262
		nfs_release_seqid(opendata->o_arg.seqid);
1263 1264 1265 1266 1267
		if (!opendata->is_recover) {
			ret = nfs_may_open(state->inode, state->owner->so_cred, open_mode);
			if (ret != 0)
				goto out;
		}
1268
		ret = -EAGAIN;
1269 1270

		/* Try to update the stateid using the delegation */
1271
		if (update_open_stateid(state, NULL, &stateid, fmode))
1272
			goto out_return_state;
1273 1274 1275 1276 1277 1278 1279 1280
	}
out:
	return ERR_PTR(ret);
out_return_state:
	atomic_inc(&state->count);
	return state;
}

1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338
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;

	if (!data->rpc_done) {
		ret = data->rpc_status;
		goto err;
	}

	ret = -ESTALE;
	if (!(data->f_attr.valid & NFS_ATTR_FATTR_TYPE) ||
	    !(data->f_attr.valid & NFS_ATTR_FATTR_FILEID) ||
	    !(data->f_attr.valid & NFS_ATTR_FATTR_CHANGE))
		goto err;

	ret = -ENOMEM;
	state = nfs4_get_open_state(inode, data->owner);
	if (state == NULL)
		goto err;

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

1339 1340
	nfs_setsecurity(inode, &data->f_attr, data->f_label);

1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353
	if (data->o_res.delegation_type != 0)
		nfs4_opendata_check_deleg(data, state);
	update_open_stateid(state, &data->o_res.stateid, NULL,
			    data->o_arg.fmode);

	return state;
err:
	return ERR_PTR(ret);

}

static struct nfs4_state *
_nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data)
1354 1355 1356
{
	struct inode *inode;
	struct nfs4_state *state = NULL;
1357
	int ret;
1358

1359
	if (!data->rpc_done) {
1360
		state = nfs4_try_open_cached(data);
1361 1362 1363
		goto out;
	}

1364
	ret = -EAGAIN;
1365
	if (!(data->f_attr.valid & NFS_ATTR_FATTR))
1366
		goto err;
1367
	inode = nfs_fhget(data->dir->d_sb, &data->o_res.fh, &data->f_attr, data->f_label);
1368
	ret = PTR_ERR(inode);
1369
	if (IS_ERR(inode))
1370 1371
		goto err;
	ret = -ENOMEM;
1372 1373
	state = nfs4_get_open_state(inode, data->owner);
	if (state == NULL)
1374
		goto err_put_inode;
1375 1376
	if (data->o_res.delegation_type != 0)
		nfs4_opendata_check_deleg(data, state);
1377
	update_open_stateid(state, &data->o_res.stateid, NULL,
1378
			data->o_arg.fmode);
1379
	iput(inode);
1380
out:
1381
	nfs_release_seqid(data->o_arg.seqid);
1382
	return state;
1383 1384 1385 1386
err_put_inode:
	iput(inode);
err:
	return ERR_PTR(ret);
1387 1388
}

1389 1390 1391 1392 1393 1394 1395 1396
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);
}

1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413
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);
}

1414 1415
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
1416 1417 1418
{
	struct nfs4_opendata *opendata;

1419
	opendata = nfs4_opendata_alloc(ctx->dentry, state->owner, 0, 0,
1420
			NULL, NULL, claim, GFP_NOFS);
T
Trond Myklebust 已提交
1421 1422 1423 1424 1425 1426 1427
	if (opendata == NULL)
		return ERR_PTR(-ENOMEM);
	opendata->state = state;
	atomic_inc(&state->count);
	return opendata;
}

1428
static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, fmode_t fmode, struct nfs4_state **res)
1429
{
1430
	struct nfs4_state *newstate;
1431 1432
	int ret;

1433 1434
	opendata->o_arg.open_flags = 0;
	opendata->o_arg.fmode = fmode;
1435 1436 1437
	memset(&opendata->o_res, 0, sizeof(opendata->o_res));
	memset(&opendata->c_res, 0, sizeof(opendata->c_res));
	nfs4_init_opendata_res(opendata);
1438
	ret = _nfs4_recover_proc_open(opendata);
1439 1440
	if (ret != 0)
		return ret; 
1441
	newstate = nfs4_opendata_to_nfs4_state(opendata);
1442 1443
	if (IS_ERR(newstate))
		return PTR_ERR(newstate);
1444
	nfs4_close_state(newstate, fmode);
1445
	*res = newstate;
1446 1447 1448 1449 1450 1451 1452 1453 1454
	return 0;
}

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

	/* memory barrier prior to reading state->n_* */
1455
	clear_bit(NFS_DELEGATED_STATE, &state->flags);
1456
	clear_bit(NFS_OPEN_STATE, &state->flags);
1457 1458
	smp_rmb();
	if (state->n_rdwr != 0) {
T
Trond Myklebust 已提交
1459
		clear_bit(NFS_O_RDWR_STATE, &state->flags);
1460
		ret = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE, &newstate);
1461 1462
		if (ret != 0)
			return ret;
1463 1464
		if (newstate != state)
			return -ESTALE;
1465 1466
	}
	if (state->n_wronly != 0) {
T
Trond Myklebust 已提交
1467
		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
1468
		ret = nfs4_open_recover_helper(opendata, FMODE_WRITE, &newstate);
1469 1470
		if (ret != 0)
			return ret;
1471 1472
		if (newstate != state)
			return -ESTALE;
1473 1474
	}
	if (state->n_rdonly != 0) {
T
Trond Myklebust 已提交
1475
		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
1476
		ret = nfs4_open_recover_helper(opendata, FMODE_READ, &newstate);
1477 1478
		if (ret != 0)
			return ret;
1479 1480
		if (newstate != state)
			return -ESTALE;
1481
	}
1482 1483 1484 1485 1486
	/*
	 * 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 &&
1487
	    !nfs4_stateid_match(&state->stateid, &state->open_stateid)) {
1488
		write_seqlock(&state->seqlock);
1489
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
1490
			nfs4_stateid_copy(&state->stateid, &state->open_stateid);
1491
		write_sequnlock(&state->seqlock);
1492
	}
1493 1494 1495
	return 0;
}

L
Linus Torvalds 已提交
1496 1497 1498 1499
/*
 * OPEN_RECLAIM:
 * 	reclaim state on the server after a reboot.
 */
1500
static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1501
{
1502
	struct nfs_delegation *delegation;
1503
	struct nfs4_opendata *opendata;
1504
	fmode_t delegation_type = 0;
L
Linus Torvalds 已提交
1505 1506
	int status;

1507 1508
	opendata = nfs4_open_recoverdata_alloc(ctx, state,
			NFS4_OPEN_CLAIM_PREVIOUS);
T
Trond Myklebust 已提交
1509 1510
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1511 1512
	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
1513
	if (delegation != NULL && test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) != 0)
1514
		delegation_type = delegation->type;
1515
	rcu_read_unlock();
1516 1517
	opendata->o_arg.u.delegation_type = delegation_type;
	status = nfs4_open_recover(opendata, state);
1518
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
1519 1520 1521
	return status;
}

1522
static int nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1523 1524 1525 1526 1527
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;
	do {
1528
		err = _nfs4_do_open_reclaim(ctx, state);
1529
		trace_nfs4_open_reclaim(ctx, 0, err);
1530 1531
		if (nfs4_clear_cap_atomic_open_v1(server, err, &exception))
			continue;
1532
		if (err != -NFS4ERR_DELAY)
1533 1534
			break;
		nfs4_handle_exception(server, err, &exception);
L
Linus Torvalds 已提交
1535 1536 1537 1538
	} while (exception.retry);
	return err;
}

1539 1540 1541 1542 1543 1544 1545
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))
1546
		return -EAGAIN;
1547
	ret = nfs4_do_open_reclaim(ctx, state);
1548 1549 1550 1551
	put_nfs_open_context(ctx);
	return ret;
}

1552
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
1553
{
1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579
	switch (err) {
		default:
			printk(KERN_ERR "NFS: %s: unhandled error "
					"%d.\n", __func__, err);
		case 0:
		case -ENOENT:
		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;
		case -NFS4ERR_DELEG_REVOKED:
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
1580
		case -NFS4ERR_OPENMODE:
1581 1582 1583 1584 1585 1586 1587 1588 1589
			nfs_inode_find_state_and_recover(state->inode,
					stateid);
			nfs4_schedule_stateid_recovery(server, state);
			return 0;
		case -NFS4ERR_DELAY:
		case -NFS4ERR_GRACE:
			set_bit(NFS_DELEGATED_STATE, &state->flags);
			ssleep(1);
			return -EAGAIN;
1590 1591 1592 1593
		case -ENOMEM:
		case -NFS4ERR_DENIED:
			/* kill_proc(fl->fl_pid, SIGLOST, 1); */
			return 0;
1594
	}
L
Linus Torvalds 已提交
1595 1596 1597
	return err;
}

1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613
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);
}

1614 1615 1616 1617 1618 1619 1620 1621
static void nfs4_open_confirm_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;

	nfs40_setup_sequence(data->o_arg.server, &data->o_arg.seq_args,
				&data->o_res.seq_res, task);
}

1622 1623 1624 1625
static void nfs4_open_confirm_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;

1626 1627
	nfs40_sequence_done(task, &data->o_res.seq_res);

1628
	data->rpc_status = task->tk_status;
1629
	if (data->rpc_status == 0) {
1630
		nfs4_stateid_copy(&data->o_res.stateid, &data->c_res.stateid);
1631
		nfs_confirm_seqid(&data->owner->so_seqid, 0);
1632
		renew_lease(data->o_res.server, data->timestamp);
1633
		data->rpc_done = 1;
1634
	}
1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645
}

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! */
1646
	if (!data->rpc_done)
1647 1648
		goto out_free;
	state = nfs4_opendata_to_nfs4_state(data);
1649
	if (!IS_ERR(state))
1650
		nfs4_close_state(state, data->o_arg.fmode);
1651
out_free:
1652
	nfs4_opendata_put(data);
1653 1654 1655
}

static const struct rpc_call_ops nfs4_open_confirm_ops = {
1656
	.rpc_call_prepare = nfs4_open_confirm_prepare,
1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667
	.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)
{
	struct nfs_server *server = NFS_SERVER(data->dir->d_inode);
	struct rpc_task *task;
1668 1669 1670 1671 1672 1673
	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
1674 1675
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1676
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1677 1678
		.callback_ops = &nfs4_open_confirm_ops,
		.callback_data = data,
1679
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1680 1681
		.flags = RPC_TASK_ASYNC,
	};
L
Linus Torvalds 已提交
1682 1683
	int status;

1684
	nfs4_init_sequence(&data->o_arg.seq_args, &data->o_res.seq_res, 1);
1685
	kref_get(&data->kref);
1686 1687
	data->rpc_done = 0;
	data->rpc_status = 0;
1688
	data->timestamp = jiffies;
T
Trond Myklebust 已提交
1689
	task = rpc_run_task(&task_setup_data);
1690
	if (IS_ERR(task))
1691 1692 1693 1694 1695 1696 1697
		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;
1698
	rpc_put_task(task);
L
Linus Torvalds 已提交
1699 1700 1701
	return status;
}

1702
static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
L
Linus Torvalds 已提交
1703
{
1704 1705
	struct nfs4_opendata *data = calldata;
	struct nfs4_state_owner *sp = data->owner;
1706
	struct nfs_client *clp = sp->so_server->nfs_client;
1707

1708
	if (nfs_wait_on_sequence(data->o_arg.seqid, task) != 0)
1709
		goto out_wait;
1710 1711 1712 1713 1714 1715 1716
	/*
	 * 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;

1717
		if (can_open_cached(data->state, data->o_arg.fmode, data->o_arg.open_flags))
1718
			goto out_no_action;
1719 1720
		rcu_read_lock();
		delegation = rcu_dereference(NFS_I(data->state->inode)->delegation);
1721
		if (data->o_arg.claim != NFS4_OPEN_CLAIM_DELEGATE_CUR &&
1722
		    data->o_arg.claim != NFS4_OPEN_CLAIM_DELEG_CUR_FH &&
1723 1724
		    can_open_delegated(delegation, data->o_arg.fmode))
			goto unlock_no_action;
1725 1726
		rcu_read_unlock();
	}
1727
	/* Update client id. */
1728
	data->o_arg.clientid = clp->cl_clientid;
1729 1730 1731 1732
	switch (data->o_arg.claim) {
	case NFS4_OPEN_CLAIM_PREVIOUS:
	case NFS4_OPEN_CLAIM_DELEG_CUR_FH:
	case NFS4_OPEN_CLAIM_DELEG_PREV_FH:
1733
		data->o_arg.open_bitmap = &nfs4_open_noattr_bitmap[0];
1734 1735
	case NFS4_OPEN_CLAIM_FH:
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR];
T
Trond Myklebust 已提交
1736 1737
		nfs_copy_fh(&data->o_res.fh, data->o_arg.fh);
	}
1738
	data->timestamp = jiffies;
1739
	if (nfs4_setup_sequence(data->o_arg.server,
1740
				&data->o_arg.seq_args,
1741 1742 1743
				&data->o_res.seq_res,
				task) != 0)
		nfs_release_seqid(data->o_arg.seqid);
1744 1745 1746 1747 1748 1749 1750 1751 1752 1753

	/* 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;
	}
1754
	return;
1755 1756
unlock_no_action:
	rcu_read_unlock();
1757 1758
out_no_action:
	task->tk_action = NULL;
1759
out_wait:
1760
	nfs4_sequence_done(task, &data->o_res.seq_res);
1761 1762
}

1763 1764 1765
static void nfs4_open_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;
L
Linus Torvalds 已提交
1766

1767
	data->rpc_status = task->tk_status;
1768

1769 1770
	if (!nfs4_sequence_done(task, &data->o_res.seq_res))
		return;
1771

1772
	if (task->tk_status == 0) {
1773 1774
		if (data->o_res.f_attr->valid & NFS_ATTR_FATTR_TYPE) {
			switch (data->o_res.f_attr->mode & S_IFMT) {
1775 1776 1777
			case S_IFREG:
				break;
			case S_IFLNK:
1778
				data->rpc_status = -ELOOP;
1779 1780
				break;
			case S_IFDIR:
1781
				data->rpc_status = -EISDIR;
1782 1783
				break;
			default:
1784
				data->rpc_status = -ENOTDIR;
1785
			}
1786
		}
1787
		renew_lease(data->o_res.server, data->timestamp);
1788 1789
		if (!(data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM))
			nfs_confirm_seqid(&data->owner->so_seqid, 0);
1790
	}
1791
	data->rpc_done = 1;
1792
}
1793

1794 1795 1796 1797 1798 1799 1800 1801 1802
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! */
1803
	if (data->rpc_status != 0 || !data->rpc_done)
1804 1805 1806 1807 1808
		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);
1809
	if (!IS_ERR(state))
1810
		nfs4_close_state(state, data->o_arg.fmode);
1811
out_free:
1812
	nfs4_opendata_put(data);
1813 1814 1815 1816 1817 1818 1819 1820
}

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,
};

1821
static int nfs4_run_open_task(struct nfs4_opendata *data, int isrecover)
1822 1823 1824 1825 1826 1827
{
	struct inode *dir = data->dir->d_inode;
	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;
1828 1829 1830 1831 1832 1833
	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
1834 1835
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1836
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1837 1838
		.callback_ops = &nfs4_open_ops,
		.callback_data = data,
1839
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1840 1841
		.flags = RPC_TASK_ASYNC,
	};
1842 1843
	int status;

1844
	nfs4_init_sequence(&o_arg->seq_args, &o_res->seq_res, 1);
1845
	kref_get(&data->kref);
1846 1847
	data->rpc_done = 0;
	data->rpc_status = 0;
1848
	data->cancelled = 0;
1849 1850
	data->is_recover = 0;
	if (isrecover) {
1851
		nfs4_set_sequence_privileged(&o_arg->seq_args);
1852 1853
		data->is_recover = 1;
	}
T
Trond Myklebust 已提交
1854
	task = rpc_run_task(&task_setup_data);
1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877
        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)
{
	struct inode *dir = data->dir->d_inode;
	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;

1878 1879
	nfs_fattr_map_and_free_names(NFS_SERVER(dir), &data->f_attr);

1880 1881 1882 1883 1884 1885 1886 1887 1888
	if (o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
		status = _nfs4_proc_open_confirm(data);
		if (status != 0)
			return status;
	}

	return status;
}

1889 1890
static int nfs4_opendata_access(struct rpc_cred *cred,
				struct nfs4_opendata *opendata,
1891 1892
				struct nfs4_state *state, fmode_t fmode,
				int openflags)
1893 1894 1895 1896 1897 1898 1899 1900 1901 1902
{
	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;
1903
	/* don't check MAY_WRITE - a newly created file may not have
1904 1905 1906 1907 1908 1909 1910 1911
	 * write mode bits, but POSIX allows the creating process to write.
	 * use openflags to check for exec, because fmode won't
	 * always have FMODE_EXEC set when file open for exec. */
	if (openflags & __FMODE_EXEC) {
		/* ONLY check for exec rights */
		mask = MAY_EXEC;
	} else if (fmode & FMODE_READ)
		mask = MAY_READ;
1912 1913 1914 1915 1916 1917

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

1918
	if ((mask & ~cache.mask & (MAY_READ | MAY_EXEC)) == 0)
1919 1920 1921 1922
		return 0;

	/* even though OPEN succeeded, access is denied. Close the file */
	nfs4_close_state(state, fmode);
1923
	return -EACCES;
1924 1925
}

1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937
/*
 * Note: On error, nfs4_proc_open will free the struct nfs4_opendata
 */
static int _nfs4_proc_open(struct nfs4_opendata *data)
{
	struct inode *dir = data->dir->d_inode;
	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);
1938 1939 1940 1941 1942 1943
	if (!data->rpc_done)
		return status;
	if (status != 0) {
		if (status == -NFS4ERR_BADNAME &&
				!(o_arg->open_flags & O_CREAT))
			return -ENOENT;
1944
		return status;
1945
	}
1946

1947 1948
	nfs_fattr_map_and_free_names(server, &data->f_attr);

1949
	if (o_arg->open_flags & O_CREAT)
1950
		update_changeattr(dir, &o_res->cinfo);
T
Trond Myklebust 已提交
1951 1952
	if ((o_res->rflags & NFS4_OPEN_RESULT_LOCKTYPE_POSIX) == 0)
		server->caps &= ~NFS_CAP_POSIX_LOCK;
L
Linus Torvalds 已提交
1953
	if(o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
1954
		status = _nfs4_proc_open_confirm(data);
L
Linus Torvalds 已提交
1955
		if (status != 0)
1956
			return status;
L
Linus Torvalds 已提交
1957 1958
	}
	if (!(o_res->f_attr->valid & NFS_ATTR_FATTR))
1959
		_nfs4_proc_getattr(server, &o_res->fh, o_res->f_attr, o_res->f_label);
1960
	return 0;
L
Linus Torvalds 已提交
1961 1962
}

A
Andy Adamson 已提交
1963 1964 1965 1966 1967
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
1968 1969 1970 1971 1972
/*
 * OPEN_EXPIRED:
 * 	reclaim state on the server after a network partition.
 * 	Assumes caller holds the appropriate lock
 */
1973
static int _nfs4_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1974
{
1975
	struct nfs4_opendata *opendata;
1976
	int ret;
L
Linus Torvalds 已提交
1977

1978
	opendata = nfs4_open_recoverdata_alloc(ctx, state,
1979
			NFS4_OPEN_CLAIM_FH);
T
Trond Myklebust 已提交
1980 1981
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1982
	ret = nfs4_open_recover(opendata, state);
1983
	if (ret == -ESTALE)
1984
		d_drop(ctx->dentry);
1985
	nfs4_opendata_put(opendata);
1986
	return ret;
L
Linus Torvalds 已提交
1987 1988
}

1989
static int nfs4_do_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
1990
{
1991
	struct nfs_server *server = NFS_SERVER(state->inode);
1992 1993 1994 1995
	struct nfs4_exception exception = { };
	int err;

	do {
1996
		err = _nfs4_open_expired(ctx, state);
1997
		trace_nfs4_open_expired(ctx, 0, err);
1998 1999
		if (nfs4_clear_cap_atomic_open_v1(server, err, &exception))
			continue;
2000 2001 2002 2003 2004 2005 2006 2007
		switch (err) {
		default:
			goto out;
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			nfs4_handle_exception(server, err, &exception);
			err = 0;
		}
2008
	} while (exception.retry);
2009
out:
2010 2011 2012
	return err;
}

L
Linus Torvalds 已提交
2013 2014 2015
static int nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
	struct nfs_open_context *ctx;
2016
	int ret;
L
Linus Torvalds 已提交
2017

2018 2019
	ctx = nfs4_state_find_open_context(state);
	if (IS_ERR(ctx))
2020
		return -EAGAIN;
2021
	ret = nfs4_do_open_expired(ctx, state);
2022 2023
	put_nfs_open_context(ctx);
	return ret;
L
Linus Torvalds 已提交
2024 2025
}

2026
#if defined(CONFIG_NFS_V4_1)
2027
static void nfs41_clear_delegation_stateid(struct nfs4_state *state)
2028 2029
{
	struct nfs_server *server = NFS_SERVER(state->inode);
2030
	nfs4_stateid *stateid = &state->stateid;
2031 2032 2033
	struct nfs_delegation *delegation;
	struct rpc_cred *cred = NULL;
	int status = -NFS4ERR_BAD_STATEID;
2034 2035 2036 2037 2038

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

2039 2040 2041 2042 2043 2044 2045 2046
	/* Get the delegation credential for use by test/free_stateid */
	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
	if (delegation != NULL &&
	    nfs4_stateid_match(&delegation->stateid, stateid)) {
		cred = get_rpccred(delegation->cred);
		rcu_read_unlock();
		status = nfs41_test_stateid(server, stateid, cred);
2047
		trace_nfs4_test_delegation_stateid(state, NULL, status);
2048 2049 2050
	} else
		rcu_read_unlock();

2051 2052 2053 2054
	if (status != NFS_OK) {
		/* Free the stateid unless the server explicitly
		 * informs us the stateid is unrecognized. */
		if (status != -NFS4ERR_BAD_STATEID)
2055
			nfs41_free_stateid(server, stateid, cred);
2056
		nfs_remove_bad_delegation(state->inode);
2057

2058 2059 2060
		write_seqlock(&state->seqlock);
		nfs4_stateid_copy(&state->stateid, &state->open_stateid);
		write_sequnlock(&state->seqlock);
2061 2062
		clear_bit(NFS_DELEGATED_STATE, &state->flags);
	}
2063 2064 2065

	if (cred != NULL)
		put_rpccred(cred);
2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078
}

/**
 * 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);
2079
	nfs4_stateid *stateid = &state->open_stateid;
2080
	struct rpc_cred *cred = state->owner->so_cred;
2081 2082 2083 2084 2085 2086 2087 2088
	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;

2089
	status = nfs41_test_stateid(server, stateid, cred);
2090
	trace_nfs4_test_open_stateid(state, NULL, status);
2091 2092 2093 2094
	if (status != NFS_OK) {
		/* Free the stateid unless the server explicitly
		 * informs us the stateid is unrecognized. */
		if (status != -NFS4ERR_BAD_STATEID)
2095
			nfs41_free_stateid(server, stateid, cred);
2096 2097 2098 2099

		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
		clear_bit(NFS_O_RDWR_STATE, &state->flags);
2100
		clear_bit(NFS_OPEN_STATE, &state->flags);
2101 2102 2103 2104 2105 2106
	}
	return status;
}

static int nfs41_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
2107
	int status;
2108

2109 2110
	nfs41_clear_delegation_stateid(state);
	status = nfs41_check_open_stateid(state);
2111 2112 2113
	if (status != NFS_OK)
		status = nfs4_open_expired(sp, state);
	return status;
2114 2115 2116
}
#endif

2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132
/*
 * 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;
}

2133 2134 2135
static int _nfs4_open_and_get_state(struct nfs4_opendata *opendata,
		fmode_t fmode,
		int flags,
2136
		struct nfs_open_context *ctx)
2137 2138 2139
{
	struct nfs4_state_owner *sp = opendata->owner;
	struct nfs_server *server = sp->so_server;
2140
	struct dentry *dentry;
2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157
	struct nfs4_state *state;
	unsigned int seq;
	int ret;

	seq = raw_seqcount_begin(&sp->so_reclaim_seqcount);

	ret = _nfs4_proc_open(opendata);
	if (ret != 0)
		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);

2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172
	dentry = opendata->dentry;
	if (dentry->d_inode == NULL) {
		/* 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,
				nfs_save_change_attribute(opendata->dir->d_inode));
	}

2173 2174 2175 2176
	ret = nfs4_opendata_access(sp->so_cred, opendata, state, fmode, flags);
	if (ret != 0)
		goto out;

2177
	ctx->state = state;
2178 2179 2180 2181 2182
	if (dentry->d_inode == state->inode) {
		nfs_inode_attach_open_context(ctx);
		if (read_seqcount_retry(&sp->so_reclaim_seqcount, seq))
			nfs4_schedule_stateid_recovery(server, state);
	}
2183 2184 2185 2186
out:
	return ret;
}

L
Linus Torvalds 已提交
2187
/*
2188
 * Returns a referenced nfs4_state
L
Linus Torvalds 已提交
2189
 */
2190
static int _nfs4_do_open(struct inode *dir,
2191
			struct nfs_open_context *ctx,
2192 2193
			int flags,
			struct iattr *sattr,
2194
			struct nfs4_label *label)
L
Linus Torvalds 已提交
2195 2196 2197 2198
{
	struct nfs4_state_owner  *sp;
	struct nfs4_state     *state = NULL;
	struct nfs_server       *server = NFS_SERVER(dir);
2199
	struct nfs4_opendata *opendata;
2200 2201 2202 2203
	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);
2204
	enum open_claim_type4 claim = NFS4_OPEN_CLAIM_NULL;
2205
	struct nfs4_label *olabel = NULL;
2206
	int status;
L
Linus Torvalds 已提交
2207 2208 2209

	/* Protect against reboot recovery conflicts */
	status = -ENOMEM;
2210 2211
	sp = nfs4_get_state_owner(server, cred, GFP_KERNEL);
	if (sp == NULL) {
L
Linus Torvalds 已提交
2212 2213 2214
		dprintk("nfs4_do_open: nfs4_get_state_owner failed!\n");
		goto out_err;
	}
2215 2216
	status = nfs4_recover_expired_lease(server);
	if (status != 0)
2217
		goto err_put_state_owner;
2218 2219
	if (dentry->d_inode != NULL)
		nfs4_return_incompatible_delegation(dentry->d_inode, fmode);
2220
	status = -ENOMEM;
2221 2222
	if (dentry->d_inode)
		claim = NFS4_OPEN_CLAIM_FH;
2223
	opendata = nfs4_opendata_alloc(dentry, sp, fmode, flags, sattr,
2224
			label, claim, GFP_KERNEL);
2225
	if (opendata == NULL)
T
Trond Myklebust 已提交
2226
		goto err_put_state_owner;
L
Linus Torvalds 已提交
2227

2228 2229 2230 2231 2232 2233 2234 2235
	if (label) {
		olabel = nfs4_label_alloc(server, GFP_KERNEL);
		if (IS_ERR(olabel)) {
			status = PTR_ERR(olabel);
			goto err_opendata_put;
		}
	}

2236 2237 2238
	if (ctx_th && server->attr_bitmask[2] & FATTR4_WORD2_MDSTHRESHOLD) {
		opendata->f_attr.mdsthreshold = pnfs_mdsthreshold_alloc();
		if (!opendata->f_attr.mdsthreshold)
2239
			goto err_free_label;
2240
		opendata->o_arg.open_bitmap = &nfs4_pnfs_open_bitmap[0];
2241
	}
2242 2243
	if (dentry->d_inode != NULL)
		opendata->state = nfs4_get_open_state(dentry->d_inode, sp);
2244

2245
	status = _nfs4_open_and_get_state(opendata, fmode, flags, ctx);
2246
	if (status != 0)
2247
		goto err_free_label;
2248
	state = ctx->state;
2249

2250 2251
	if ((opendata->o_arg.open_flags & O_EXCL) &&
	    (opendata->o_arg.createmode != NFS4_CREATE_GUARDED)) {
2252 2253 2254 2255 2256
		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,
2257 2258
				state, label, olabel);
		if (status == 0) {
2259
			nfs_setattr_update_inode(state->inode, sattr);
2260
			nfs_post_op_update_inode(state->inode, opendata->o_res.f_attr);
2261
			nfs_setsecurity(state->inode, opendata->o_res.f_attr, olabel);
2262
		}
2263
	}
2264 2265 2266 2267 2268 2269 2270

	if (pnfs_use_threshold(ctx_th, opendata->f_attr.mdsthreshold, server))
		*ctx_th = opendata->f_attr.mdsthreshold;
	else
		kfree(opendata->f_attr.mdsthreshold);
	opendata->f_attr.mdsthreshold = NULL;

2271 2272
	nfs4_label_free(olabel);

2273
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
2274 2275
	nfs4_put_state_owner(sp);
	return 0;
2276 2277
err_free_label:
	nfs4_label_free(olabel);
2278
err_opendata_put:
2279
	kfree(opendata->f_attr.mdsthreshold);
2280
	nfs4_opendata_put(opendata);
2281 2282
err_put_state_owner:
	nfs4_put_state_owner(sp);
L
Linus Torvalds 已提交
2283 2284 2285 2286 2287
out_err:
	return status;
}


2288
static struct nfs4_state *nfs4_do_open(struct inode *dir,
2289
					struct nfs_open_context *ctx,
2290 2291
					int flags,
					struct iattr *sattr,
2292
					struct nfs4_label *label)
L
Linus Torvalds 已提交
2293
{
2294
	struct nfs_server *server = NFS_SERVER(dir);
L
Linus Torvalds 已提交
2295 2296 2297 2298 2299
	struct nfs4_exception exception = { };
	struct nfs4_state *res;
	int status;

	do {
2300
		status = _nfs4_do_open(dir, ctx, flags, sattr, label);
2301
		res = ctx->state;
2302
		trace_nfs4_open_file(ctx, flags, status);
L
Linus Torvalds 已提交
2303 2304 2305 2306 2307 2308 2309 2310
		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
2311
		 * doing an OPEN, we assume that nfs_increment_open_seqid() will
L
Linus Torvalds 已提交
2312 2313 2314 2315 2316
		 * 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) {
2317
			pr_warn_ratelimited("NFS: v4 server %s "
2318 2319
					" returned a bad sequence-id error!\n",
					NFS_SERVER(dir)->nfs_client->cl_hostname);
L
Linus Torvalds 已提交
2320 2321 2322
			exception.retry = 1;
			continue;
		}
2323 2324 2325 2326 2327 2328 2329 2330 2331 2332
		/*
		 * 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;
		}
2333 2334 2335 2336 2337
		if (status == -EAGAIN) {
			/* We must have found a delegation */
			exception.retry = 1;
			continue;
		}
2338 2339 2340
		if (nfs4_clear_cap_atomic_open_v1(server, status, &exception))
			continue;
		res = ERR_PTR(nfs4_handle_exception(server,
L
Linus Torvalds 已提交
2341 2342 2343 2344 2345
					status, &exception));
	} while (exception.retry);
	return res;
}

2346 2347
static int _nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
			    struct nfs_fattr *fattr, struct iattr *sattr,
2348 2349
			    struct nfs4_state *state, struct nfs4_label *ilabel,
			    struct nfs4_label *olabel)
L
Linus Torvalds 已提交
2350
{
2351
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
2352
        struct nfs_setattrargs  arg = {
2353
                .fh             = NFS_FH(inode),
L
Linus Torvalds 已提交
2354 2355 2356
                .iap            = sattr,
		.server		= server,
		.bitmask = server->attr_bitmask,
2357
		.label		= ilabel,
L
Linus Torvalds 已提交
2358 2359 2360
        };
        struct nfs_setattrres  res = {
		.fattr		= fattr,
2361
		.label		= olabel,
L
Linus Torvalds 已提交
2362 2363 2364
		.server		= server,
        };
        struct rpc_message msg = {
2365 2366 2367 2368
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETATTR],
		.rpc_argp	= &arg,
		.rpc_resp	= &res,
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
2369
        };
2370
	unsigned long timestamp = jiffies;
2371 2372
	fmode_t fmode;
	bool truncate;
2373
	int status;
L
Linus Torvalds 已提交
2374

2375 2376 2377 2378
	arg.bitmask = nfs4_bitmask(server, ilabel);
	if (ilabel)
		arg.bitmask = nfs4_bitmask(server, olabel);

2379
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
2380

2381 2382 2383 2384 2385 2386 2387
	/* 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 */
	} else if (truncate && state != NULL && nfs4_valid_open_stateid(state)) {
2388 2389 2390 2391
		struct nfs_lockowner lockowner = {
			.l_owner = current->files,
			.l_pid = current->tgid,
		};
2392
		nfs4_select_rw_stateid(&arg.stateid, state, FMODE_WRITE,
2393
				&lockowner);
2394
	} else
2395
		nfs4_stateid_copy(&arg.stateid, &zero_stateid);
L
Linus Torvalds 已提交
2396

2397
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
2398 2399
	if (status == 0 && state != NULL)
		renew_lease(server, timestamp);
2400
	return status;
L
Linus Torvalds 已提交
2401 2402
}

2403 2404
static int nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
			   struct nfs_fattr *fattr, struct iattr *sattr,
2405 2406
			   struct nfs4_state *state, struct nfs4_label *ilabel,
			   struct nfs4_label *olabel)
L
Linus Torvalds 已提交
2407
{
2408
	struct nfs_server *server = NFS_SERVER(inode);
2409 2410
	struct nfs4_exception exception = {
		.state = state,
2411
		.inode = inode,
2412
	};
L
Linus Torvalds 已提交
2413 2414
	int err;
	do {
2415
		err = _nfs4_do_setattr(inode, cred, fattr, sattr, state, ilabel, olabel);
2416
		trace_nfs4_setattr(inode, err);
2417 2418
		switch (err) {
		case -NFS4ERR_OPENMODE:
2419 2420 2421 2422 2423 2424 2425
			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);
			}
2426 2427 2428 2429 2430 2431 2432 2433
			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
2434
	} while (exception.retry);
2435
out:
L
Linus Torvalds 已提交
2436 2437 2438 2439 2440 2441 2442 2443
	return err;
}

struct nfs4_closedata {
	struct inode *inode;
	struct nfs4_state *state;
	struct nfs_closeargs arg;
	struct nfs_closeres res;
2444
	struct nfs_fattr fattr;
2445
	unsigned long timestamp;
F
Fred Isaman 已提交
2446 2447
	bool roc;
	u32 roc_barrier;
L
Linus Torvalds 已提交
2448 2449
};

2450
static void nfs4_free_closedata(void *data)
2451
{
2452 2453
	struct nfs4_closedata *calldata = data;
	struct nfs4_state_owner *sp = calldata->state->owner;
2454
	struct super_block *sb = calldata->state->inode->i_sb;
2455

F
Fred Isaman 已提交
2456 2457
	if (calldata->roc)
		pnfs_roc_release(calldata->state->inode);
2458 2459 2460
	nfs4_put_open_state(calldata->state);
	nfs_free_seqid(calldata->arg.seqid);
	nfs4_put_state_owner(sp);
2461
	nfs_sb_deactive(sb);
2462 2463 2464
	kfree(calldata);
}

2465 2466 2467 2468
static void nfs4_close_clear_stateid_flags(struct nfs4_state *state,
		fmode_t fmode)
{
	spin_lock(&state->owner->so_lock);
2469 2470 2471
	clear_bit(NFS_O_RDWR_STATE, &state->flags);
	switch (fmode & (FMODE_READ|FMODE_WRITE)) {
	case FMODE_WRITE:
2472
		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
2473 2474
		break;
	case FMODE_READ:
2475
		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
2476 2477 2478 2479 2480 2481
		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);
	}
2482 2483 2484
	spin_unlock(&state->owner->so_lock);
}

2485
static void nfs4_close_done(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
2486
{
2487
	struct nfs4_closedata *calldata = data;
L
Linus Torvalds 已提交
2488 2489 2490
	struct nfs4_state *state = calldata->state;
	struct nfs_server *server = NFS_SERVER(calldata->inode);

2491
	dprintk("%s: begin!\n", __func__);
2492 2493
	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
		return;
2494
	trace_nfs4_close(state, &calldata->arg, &calldata->res, task->tk_status);
L
Linus Torvalds 已提交
2495 2496 2497 2498 2499
        /* 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:
F
Fred Isaman 已提交
2500 2501 2502
			if (calldata->roc)
				pnfs_roc_set_barrier(state->inode,
						     calldata->roc_barrier);
2503
			nfs_set_open_stateid(state, &calldata->res.stateid, 0);
2504
			renew_lease(server, calldata->timestamp);
2505 2506
			nfs4_close_clear_stateid_flags(state,
					calldata->arg.fmode);
L
Linus Torvalds 已提交
2507 2508
			break;
		case -NFS4ERR_STALE_STATEID:
2509 2510
		case -NFS4ERR_OLD_STATEID:
		case -NFS4ERR_BAD_STATEID:
L
Linus Torvalds 已提交
2511
		case -NFS4ERR_EXPIRED:
2512
			if (calldata->arg.fmode == 0)
2513
				break;
L
Linus Torvalds 已提交
2514
		default:
2515 2516
			if (nfs4_async_handle_error(task, server, state) == -EAGAIN)
				rpc_restart_call_prepare(task);
L
Linus Torvalds 已提交
2517
	}
2518
	nfs_release_seqid(calldata->arg.seqid);
2519
	nfs_refresh_inode(calldata->inode, calldata->res.fattr);
2520
	dprintk("%s: done, ret = %d!\n", __func__, task->tk_status);
L
Linus Torvalds 已提交
2521 2522
}

T
Trond Myklebust 已提交
2523
static void nfs4_close_prepare(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
2524
{
T
Trond Myklebust 已提交
2525
	struct nfs4_closedata *calldata = data;
2526
	struct nfs4_state *state = calldata->state;
2527
	struct inode *inode = calldata->inode;
2528
	int call_close = 0;
2529

2530
	dprintk("%s: begin!\n", __func__);
2531
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
2532
		goto out_wait;
2533

2534 2535
	task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_DOWNGRADE];
	calldata->arg.fmode = FMODE_READ|FMODE_WRITE;
2536
	spin_lock(&state->owner->so_lock);
2537
	/* Calculate the change in open mode */
2538
	if (state->n_rdwr == 0) {
2539
		if (state->n_rdonly == 0) {
2540 2541 2542
			call_close |= test_bit(NFS_O_RDONLY_STATE, &state->flags);
			call_close |= test_bit(NFS_O_RDWR_STATE, &state->flags);
			calldata->arg.fmode &= ~FMODE_READ;
2543 2544
		}
		if (state->n_wronly == 0) {
2545 2546 2547
			call_close |= test_bit(NFS_O_WRONLY_STATE, &state->flags);
			call_close |= test_bit(NFS_O_RDWR_STATE, &state->flags);
			calldata->arg.fmode &= ~FMODE_WRITE;
2548
		}
2549
	}
2550 2551
	if (!nfs4_valid_open_stateid(state))
		call_close = 0;
2552
	spin_unlock(&state->owner->so_lock);
2553 2554

	if (!call_close) {
2555
		/* Note: exit _without_ calling nfs4_close_done */
2556
		goto out_no_action;
2557
	}
2558

F
Fred Isaman 已提交
2559
	if (calldata->arg.fmode == 0) {
2560
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE];
F
Fred Isaman 已提交
2561
		if (calldata->roc &&
2562 2563
		    pnfs_roc_drain(inode, &calldata->roc_barrier, task)) {
			nfs_release_seqid(calldata->arg.seqid);
2564
			goto out_wait;
2565
		    }
F
Fred Isaman 已提交
2566
	}
2567

2568
	nfs_fattr_init(calldata->res.fattr);
2569
	calldata->timestamp = jiffies;
2570
	if (nfs4_setup_sequence(NFS_SERVER(inode),
2571 2572
				&calldata->arg.seq_args,
				&calldata->res.seq_res,
2573 2574
				task) != 0)
		nfs_release_seqid(calldata->arg.seqid);
2575
	dprintk("%s: done!\n", __func__);
2576 2577 2578 2579 2580
	return;
out_no_action:
	task->tk_action = NULL;
out_wait:
	nfs4_sequence_done(task, &calldata->res.seq_res);
L
Linus Torvalds 已提交
2581 2582
}

2583
static const struct rpc_call_ops nfs4_close_ops = {
T
Trond Myklebust 已提交
2584
	.rpc_call_prepare = nfs4_close_prepare,
2585 2586 2587 2588
	.rpc_call_done = nfs4_close_done,
	.rpc_release = nfs4_free_closedata,
};

L
Linus Torvalds 已提交
2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599
/* 
 * 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!
 */
2600
int nfs4_do_close(struct nfs4_state *state, gfp_t gfp_mask, int wait)
L
Linus Torvalds 已提交
2601
{
2602
	struct nfs_server *server = NFS_SERVER(state->inode);
L
Linus Torvalds 已提交
2603
	struct nfs4_closedata *calldata;
2604 2605
	struct nfs4_state_owner *sp = state->owner;
	struct rpc_task *task;
2606 2607 2608 2609
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
2610 2611
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
2612
		.rpc_message = &msg,
T
Trond Myklebust 已提交
2613
		.callback_ops = &nfs4_close_ops,
2614
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
2615 2616
		.flags = RPC_TASK_ASYNC,
	};
2617
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2618

2619 2620 2621
	nfs4_state_protect(server->nfs_client, NFS_SP4_MACH_CRED_CLEANUP,
		&task_setup_data.rpc_client, &msg);

2622
	calldata = kzalloc(sizeof(*calldata), gfp_mask);
L
Linus Torvalds 已提交
2623
	if (calldata == NULL)
2624
		goto out;
2625
	nfs4_init_sequence(&calldata->arg.seq_args, &calldata->res.seq_res, 1);
2626
	calldata->inode = state->inode;
L
Linus Torvalds 已提交
2627
	calldata->state = state;
2628
	calldata->arg.fh = NFS_FH(state->inode);
2629
	calldata->arg.stateid = &state->open_stateid;
L
Linus Torvalds 已提交
2630
	/* Serialization for the sequence id */
2631
	calldata->arg.seqid = nfs_alloc_seqid(&state->owner->so_seqid, gfp_mask);
2632 2633
	if (calldata->arg.seqid == NULL)
		goto out_free_calldata;
2634
	calldata->arg.fmode = 0;
2635
	calldata->arg.bitmask = server->cache_consistency_bitmask;
2636
	calldata->res.fattr = &calldata->fattr;
2637
	calldata->res.seqid = calldata->arg.seqid;
2638
	calldata->res.server = server;
2639
	calldata->roc = pnfs_roc(state->inode);
2640
	nfs_sb_active(calldata->inode->i_sb);
2641

2642 2643
	msg.rpc_argp = &calldata->arg;
	msg.rpc_resp = &calldata->res;
T
Trond Myklebust 已提交
2644 2645
	task_setup_data.callback_data = calldata;
	task = rpc_run_task(&task_setup_data);
2646 2647
	if (IS_ERR(task))
		return PTR_ERR(task);
2648 2649 2650
	status = 0;
	if (wait)
		status = rpc_wait_for_completion_task(task);
2651
	rpc_put_task(task);
2652
	return status;
2653 2654 2655
out_free_calldata:
	kfree(calldata);
out:
2656 2657
	nfs4_put_open_state(state);
	nfs4_put_state_owner(sp);
2658
	return status;
L
Linus Torvalds 已提交
2659 2660
}

2661
static struct inode *
2662
nfs4_atomic_open(struct inode *dir, struct nfs_open_context *ctx, int open_flags, struct iattr *attr)
L
Linus Torvalds 已提交
2663 2664
{
	struct nfs4_state *state;
2665 2666 2667
	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
2668

T
Trond Myklebust 已提交
2669
	/* Protect against concurrent sillydeletes */
2670
	state = nfs4_do_open(dir, ctx, open_flags, attr, label);
2671 2672 2673

	nfs4_label_release_security(label);

2674 2675
	if (IS_ERR(state))
		return ERR_CAST(state);
2676
	return state->inode;
L
Linus Torvalds 已提交
2677 2678
}

2679
static void nfs4_close_context(struct nfs_open_context *ctx, int is_sync)
T
Trond Myklebust 已提交
2680 2681 2682 2683
{
	if (ctx->state == NULL)
		return;
	if (is_sync)
2684
		nfs4_close_sync(ctx->state, ctx->mode);
T
Trond Myklebust 已提交
2685
	else
2686
		nfs4_close_state(ctx->state, ctx->mode);
T
Trond Myklebust 已提交
2687
}
L
Linus Torvalds 已提交
2688 2689 2690

static int _nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
{
B
Benny Halevy 已提交
2691 2692 2693
	struct nfs4_server_caps_arg args = {
		.fhandle = fhandle,
	};
L
Linus Torvalds 已提交
2694 2695 2696
	struct nfs4_server_caps_res res = {};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SERVER_CAPS],
B
Benny Halevy 已提交
2697
		.rpc_argp = &args,
L
Linus Torvalds 已提交
2698 2699 2700 2701
		.rpc_resp = &res,
	};
	int status;

2702
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
2703 2704
	if (status == 0) {
		memcpy(server->attr_bitmask, res.attr_bitmask, sizeof(server->attr_bitmask));
2705 2706 2707 2708 2709
		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|
				NFS_CAP_CTIME|NFS_CAP_MTIME);
L
Linus Torvalds 已提交
2710 2711 2712 2713 2714 2715
		if (res.attr_bitmask[0] & FATTR4_WORD0_ACL)
			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;
2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731
		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;
2732 2733 2734 2735 2736 2737
#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));
2738

2739 2740 2741 2742
		if (server->caps & NFS_CAP_SECURITY_LABEL) {
			server->attr_bitmask_nl[2] &= ~FATTR4_WORD2_SECURITY_LABEL;
			res.attr_bitmask[2] &= ~FATTR4_WORD2_SECURITY_LABEL;
		}
2743 2744 2745
		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;
L
Linus Torvalds 已提交
2746
		server->acl_bitmask = res.acl_bitmask;
2747
		server->fh_expire_type = res.fh_expire_type;
L
Linus Torvalds 已提交
2748
	}
A
Andy Adamson 已提交
2749

L
Linus Torvalds 已提交
2750 2751 2752
	return status;
}

2753
int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
L
Linus Torvalds 已提交
2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767
{
	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)
{
2768
	u32 bitmask[3];
L
Linus Torvalds 已提交
2769
	struct nfs4_lookup_root_arg args = {
2770
		.bitmask = bitmask,
L
Linus Torvalds 已提交
2771 2772 2773
	};
	struct nfs4_lookup_res res = {
		.server = server,
2774
		.fattr = info->fattr,
L
Linus Torvalds 已提交
2775 2776 2777 2778 2779 2780 2781
		.fh = fhandle,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP_ROOT],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
2782

2783 2784 2785 2786 2787 2788 2789
	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;

2790
	nfs_fattr_init(info->fattr);
2791
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
2792 2793 2794 2795 2796 2797 2798 2799
}

static int nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle,
		struct nfs_fsinfo *info)
{
	struct nfs4_exception exception = { };
	int err;
	do {
2800
		err = _nfs4_lookup_root(server, fhandle, info);
2801
		trace_nfs4_lookup_root(server, fhandle, info->fattr, err);
2802 2803 2804
		switch (err) {
		case 0:
		case -NFS4ERR_WRONGSEC:
2805
			goto out;
2806 2807 2808
		default:
			err = nfs4_handle_exception(server, err, &exception);
		}
L
Linus Torvalds 已提交
2809
	} while (exception.retry);
2810
out:
L
Linus Torvalds 已提交
2811 2812 2813
	return err;
}

2814 2815 2816
static int nfs4_lookup_root_sec(struct nfs_server *server, struct nfs_fh *fhandle,
				struct nfs_fsinfo *info, rpc_authflavor_t flavor)
{
2817 2818 2819
	struct rpc_auth_create_args auth_args = {
		.pseudoflavor = flavor,
	};
2820 2821 2822
	struct rpc_auth *auth;
	int ret;

2823
	auth = rpcauth_create(&auth_args, server->client);
2824
	if (IS_ERR(auth)) {
2825
		ret = -EACCES;
2826 2827 2828 2829 2830 2831 2832
		goto out;
	}
	ret = nfs4_lookup_root(server, fhandle, info);
out:
	return ret;
}

2833 2834 2835 2836 2837 2838 2839 2840 2841
/*
 * 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.
 */
2842
static int nfs4_find_root_sec(struct nfs_server *server, struct nfs_fh *fhandle,
2843
			      struct nfs_fsinfo *info)
L
Linus Torvalds 已提交
2844
{
2845 2846 2847 2848 2849
	/* Per 3530bis 15.33.5 */
	static const rpc_authflavor_t flav_array[] = {
		RPC_AUTH_GSS_KRB5P,
		RPC_AUTH_GSS_KRB5I,
		RPC_AUTH_GSS_KRB5,
2850
		RPC_AUTH_UNIX,			/* courtesy */
2851 2852 2853 2854
		RPC_AUTH_NULL,
	};
	int status = -EPERM;
	size_t i;
C
Chuck Lever 已提交
2855

2856
	for (i = 0; i < ARRAY_SIZE(flav_array); i++) {
2857
		status = nfs4_lookup_root_sec(server, fhandle, info, flav_array[i]);
2858
		if (status == -NFS4ERR_WRONGSEC || status == -EACCES)
2859 2860
			continue;
		break;
2861
	}
2862

2863 2864 2865 2866 2867 2868 2869 2870 2871
	/*
	 * -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;
2872 2873 2874
	return status;
}

C
Chuck Lever 已提交
2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886
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
2887
 * @auth_probe: probe the auth flavours
C
Chuck Lever 已提交
2888 2889
 *
 * Returns zero on success, or a negative errno.
2890
 */
2891
int nfs4_proc_get_rootfh(struct nfs_server *server, struct nfs_fh *fhandle,
2892 2893
			 struct nfs_fsinfo *info,
			 bool auth_probe)
2894
{
C
Chuck Lever 已提交
2895 2896
	int status;

2897 2898 2899 2900 2901 2902 2903 2904 2905
	switch (auth_probe) {
	case false:
		status = nfs4_lookup_root(server, fhandle, info);
		if (status != -NFS4ERR_WRONGSEC)
			break;
		/* Did user force a 'sec=' mount option? */
		if (server->flags & NFS_MOUNT_SECFLAVOUR)
			break;
	default:
C
Chuck Lever 已提交
2906
		status = nfs4_do_find_root_sec(server, fhandle, info);
2907
	}
C
Chuck Lever 已提交
2908

L
Linus Torvalds 已提交
2909 2910 2911 2912
	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
2913

2914
	return nfs4_map_errors(status);
L
Linus Torvalds 已提交
2915 2916
}

2917 2918 2919 2920 2921
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;
2922
	struct nfs4_label *label = NULL;
2923 2924 2925 2926 2927 2928 2929

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

2930 2931 2932 2933
	label = nfs4_label_alloc(server, GFP_KERNEL);
	if (IS_ERR(label))
		return PTR_ERR(label);

2934
	error = nfs4_proc_getattr(server, mntfh, fattr, label);
2935 2936
	if (error < 0) {
		dprintk("nfs4_get_root: getattr error = %d\n", -error);
2937
		goto err_free_label;
2938 2939 2940 2941 2942 2943
	}

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

2944 2945 2946
err_free_label:
	nfs4_label_free(label);

2947 2948 2949
	return error;
}

M
Manoj Naik 已提交
2950 2951 2952 2953 2954
/*
 * 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
 */
2955 2956 2957
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
2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969
{
	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;

2970
	status = nfs4_proc_fs_locations(client, dir, name, locations, page);
M
Manoj Naik 已提交
2971 2972 2973 2974
	if (status != 0)
		goto out;
	/* Make sure server returned a different fsid for the referral */
	if (nfs_fsid_equal(&NFS_SERVER(dir)->fsid, &locations->fattr.fsid)) {
2975 2976
		dprintk("%s: server did not return a different fsid for"
			" a referral at %s\n", __func__, name->name);
M
Manoj Naik 已提交
2977 2978 2979
		status = -EIO;
		goto out;
	}
2980 2981
	/* Fixup attributes for the nfs_lookup() call to nfs_fhget() */
	nfs_fixup_referral_attributes(&locations->fattr);
M
Manoj Naik 已提交
2982

2983
	/* replace the lookup nfs_fattr with the locations nfs_fattr */
M
Manoj Naik 已提交
2984 2985 2986 2987 2988
	memcpy(fattr, &locations->fattr, sizeof(struct nfs_fattr));
	memset(fhandle, 0, sizeof(struct nfs_fh));
out:
	if (page)
		__free_page(page);
2989
	kfree(locations);
M
Manoj Naik 已提交
2990 2991 2992
	return status;
}

2993 2994
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
2995 2996 2997 2998 2999 3000 3001
{
	struct nfs4_getattr_arg args = {
		.fh = fhandle,
		.bitmask = server->attr_bitmask,
	};
	struct nfs4_getattr_res res = {
		.fattr = fattr,
3002
		.label = label,
L
Linus Torvalds 已提交
3003 3004 3005 3006 3007 3008 3009
		.server = server,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETATTR],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
3010 3011 3012

	args.bitmask = nfs4_bitmask(server, label);

3013
	nfs_fattr_init(fattr);
3014
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3015 3016
}

3017 3018
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
3019 3020 3021 3022
{
	struct nfs4_exception exception = { };
	int err;
	do {
3023 3024 3025
		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
3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051
				&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)
{
3052
	struct inode *inode = dentry->d_inode;
3053
	struct rpc_cred *cred = NULL;
3054
	struct nfs4_state *state = NULL;
3055
	struct nfs4_label *label = NULL;
L
Linus Torvalds 已提交
3056 3057
	int status;

B
Benny Halevy 已提交
3058
	if (pnfs_ld_layoutret_on_setattr(inode))
3059
		pnfs_commit_and_return_layout(inode);
B
Benny Halevy 已提交
3060

3061
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
3062
	
3063 3064
	/* Deal with open(O_TRUNC) */
	if (sattr->ia_valid & ATTR_OPEN)
3065
		sattr->ia_valid &= ~(ATTR_MTIME|ATTR_CTIME);
3066 3067

	/* Optimization: if the end result is no change, don't RPC */
3068
	if ((sattr->ia_valid & ~(ATTR_FILE|ATTR_OPEN)) == 0)
3069 3070
		return 0;

3071
	/* Search for an existing open(O_WRITE) file */
3072 3073 3074 3075
	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
3076 3077 3078 3079
		if (ctx) {
			cred = ctx->cred;
			state = ctx->state;
		}
3080
	}
3081

3082 3083 3084 3085 3086
	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);
3087
	if (status == 0) {
3088
		nfs_setattr_update_inode(inode, sattr);
3089 3090
		nfs_setsecurity(inode, fattr, label);
	}
3091
	nfs4_label_free(label);
L
Linus Torvalds 已提交
3092 3093 3094
	return status;
}

3095 3096
static int _nfs4_proc_lookup(struct rpc_clnt *clnt, struct inode *dir,
		const struct qstr *name, struct nfs_fh *fhandle,
3097
		struct nfs_fattr *fattr, struct nfs4_label *label)
D
David Howells 已提交
3098
{
3099
	struct nfs_server *server = NFS_SERVER(dir);
D
David Howells 已提交
3100 3101 3102
	int		       status;
	struct nfs4_lookup_arg args = {
		.bitmask = server->attr_bitmask,
3103
		.dir_fh = NFS_FH(dir),
D
David Howells 已提交
3104 3105 3106 3107 3108
		.name = name,
	};
	struct nfs4_lookup_res res = {
		.server = server,
		.fattr = fattr,
3109
		.label = label,
D
David Howells 已提交
3110 3111 3112 3113 3114 3115 3116 3117
		.fh = fhandle,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};

3118 3119
	args.bitmask = nfs4_bitmask(server, label);

D
David Howells 已提交
3120 3121
	nfs_fattr_init(fattr);

L
Linus Torvalds 已提交
3122
	dprintk("NFS call  lookup %s\n", name->name);
3123
	status = nfs4_call_sync(clnt, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3124 3125 3126 3127
	dprintk("NFS reply lookup: %d\n", status);
	return status;
}

3128
static void nfs_fixup_secinfo_attributes(struct nfs_fattr *fattr)
3129 3130
{
	fattr->valid |= NFS_ATTR_FATTR_TYPE | NFS_ATTR_FATTR_MODE |
3131
		NFS_ATTR_FATTR_NLINK | NFS_ATTR_FATTR_MOUNTPOINT;
3132 3133 3134 3135
	fattr->mode = S_IFDIR | S_IRUGO | S_IXUGO;
	fattr->nlink = 2;
}

3136 3137
static int nfs4_proc_lookup_common(struct rpc_clnt **clnt, struct inode *dir,
				   struct qstr *name, struct nfs_fh *fhandle,
3138
				   struct nfs_fattr *fattr, struct nfs4_label *label)
L
Linus Torvalds 已提交
3139 3140
{
	struct nfs4_exception exception = { };
3141
	struct rpc_clnt *client = *clnt;
L
Linus Torvalds 已提交
3142 3143
	int err;
	do {
3144
		err = _nfs4_proc_lookup(client, dir, name, fhandle, fattr, label);
3145
		trace_nfs4_lookup(dir, name, err);
3146
		switch (err) {
3147
		case -NFS4ERR_BADNAME:
3148 3149
			err = -ENOENT;
			goto out;
3150
		case -NFS4ERR_MOVED:
3151
			err = nfs4_get_referral(client, dir, name, fattr, fhandle);
3152
			goto out;
3153
		case -NFS4ERR_WRONGSEC:
3154 3155 3156
			err = -EPERM;
			if (client != *clnt)
				goto out;
3157 3158 3159
			/* No security negotiation if the user specified 'sec=' */
			if (NFS_SERVER(dir)->flags & NFS_MOUNT_SECFLAVOUR)
				goto out;
3160 3161 3162 3163 3164 3165 3166 3167
			client = nfs4_create_sec_client(client, dir, name);
			if (IS_ERR(client))
				return PTR_ERR(client);

			exception.retry = 1;
			break;
		default:
			err = nfs4_handle_exception(NFS_SERVER(dir), err, &exception);
3168
		}
L
Linus Torvalds 已提交
3169
	} while (exception.retry);
3170 3171 3172 3173 3174 3175 3176

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

L
Linus Torvalds 已提交
3177 3178 3179
	return err;
}

3180
static int nfs4_proc_lookup(struct inode *dir, struct qstr *name,
3181 3182
			    struct nfs_fh *fhandle, struct nfs_fattr *fattr,
			    struct nfs4_label *label)
3183 3184 3185 3186
{
	int status;
	struct rpc_clnt *client = NFS_CLIENT(dir);

3187
	status = nfs4_proc_lookup_common(&client, dir, name, fhandle, fattr, label);
3188 3189 3190 3191 3192 3193 3194
	if (client != NFS_CLIENT(dir)) {
		rpc_shutdown_client(client);
		nfs_fixup_secinfo_attributes(fattr);
	}
	return status;
}

3195 3196 3197 3198 3199 3200 3201
struct rpc_clnt *
nfs4_proc_lookup_mountpoint(struct inode *dir, struct qstr *name,
			    struct nfs_fh *fhandle, struct nfs_fattr *fattr)
{
	int status;
	struct rpc_clnt *client = rpc_clone_client(NFS_CLIENT(dir));

3202
	status = nfs4_proc_lookup_common(&client, dir, name, fhandle, fattr, NULL);
3203 3204 3205 3206 3207 3208 3209
	if (status < 0) {
		rpc_shutdown_client(client);
		return ERR_PTR(status);
	}
	return client;
}

L
Linus Torvalds 已提交
3210 3211
static int _nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
{
3212
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
3213 3214
	struct nfs4_accessargs args = {
		.fh = NFS_FH(inode),
3215
		.bitmask = server->cache_consistency_bitmask,
3216 3217 3218
	};
	struct nfs4_accessres res = {
		.server = server,
L
Linus Torvalds 已提交
3219 3220 3221 3222 3223 3224 3225 3226
	};
	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;
3227
	int status = 0;
L
Linus Torvalds 已提交
3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244

	/*
	 * 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;
	}
3245 3246 3247 3248 3249

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

3250
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3251
	if (!status) {
3252
		nfs_access_set_mask(entry, res.access);
3253
		nfs_refresh_inode(inode, res.fattr);
L
Linus Torvalds 已提交
3254
	}
3255
	nfs_free_fattr(res.fattr);
L
Linus Torvalds 已提交
3256 3257 3258 3259 3260 3261 3262 3263
	return status;
}

static int nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
{
	struct nfs4_exception exception = { };
	int err;
	do {
3264 3265 3266
		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
3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288
				&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
3289
 * we get the post-operation mtime and size.
L
Linus Torvalds 已提交
3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302
 *
 * 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
3303
	struct nfs4_readlink_res res;
L
Linus Torvalds 已提交
3304 3305 3306
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READLINK],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3307
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3308 3309
	};

3310
	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
3311 3312 3313 3314 3315 3316 3317 3318
}

static int nfs4_proc_readlink(struct inode *inode, struct page *page,
		unsigned int pgbase, unsigned int pglen)
{
	struct nfs4_exception exception = { };
	int err;
	do {
3319 3320 3321
		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
3322 3323 3324 3325 3326 3327
				&exception);
	} while (exception.retry);
	return err;
}

/*
3328
 * This is just for mknod.  open(O_CREAT) will always do ->open_context().
L
Linus Torvalds 已提交
3329 3330 3331
 */
static int
nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
3332
		 int flags)
L
Linus Torvalds 已提交
3333
{
3334
	struct nfs4_label l, *ilabel = NULL;
3335
	struct nfs_open_context *ctx;
L
Linus Torvalds 已提交
3336 3337 3338
	struct nfs4_state *state;
	int status = 0;

3339 3340 3341 3342
	ctx = alloc_nfs_open_context(dentry, FMODE_READ);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);

3343 3344
	ilabel = nfs4_label_init_security(dir, dentry, sattr, &l);

A
Aneesh Kumar K.V 已提交
3345
	sattr->ia_mode &= ~current_umask();
3346
	state = nfs4_do_open(dir, ctx, flags, sattr, ilabel);
L
Linus Torvalds 已提交
3347 3348
	if (IS_ERR(state)) {
		status = PTR_ERR(state);
3349
		goto out;
L
Linus Torvalds 已提交
3350 3351
	}
out:
3352
	nfs4_label_release_security(ilabel);
3353
	put_nfs_open_context(ctx);
L
Linus Torvalds 已提交
3354 3355 3356 3357 3358
	return status;
}

static int _nfs4_proc_remove(struct inode *dir, struct qstr *name)
{
3359
	struct nfs_server *server = NFS_SERVER(dir);
3360
	struct nfs_removeargs args = {
L
Linus Torvalds 已提交
3361
		.fh = NFS_FH(dir),
3362
		.name = *name,
3363
	};
3364
	struct nfs_removeres res = {
3365
		.server = server,
L
Linus Torvalds 已提交
3366 3367
	};
	struct rpc_message msg = {
3368 3369 3370
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE],
		.rpc_argp = &args,
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3371
	};
3372
	int status;
L
Linus Torvalds 已提交
3373

3374
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 1);
3375
	if (status == 0)
3376
		update_changeattr(dir, &res.cinfo);
L
Linus Torvalds 已提交
3377 3378 3379 3380 3381 3382 3383 3384
	return status;
}

static int nfs4_proc_remove(struct inode *dir, struct qstr *name)
{
	struct nfs4_exception exception = { };
	int err;
	do {
3385 3386 3387
		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
3388 3389 3390 3391 3392
				&exception);
	} while (exception.retry);
	return err;
}

3393
static void nfs4_proc_unlink_setup(struct rpc_message *msg, struct inode *dir)
L
Linus Torvalds 已提交
3394
{
3395 3396 3397
	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
3398

3399
	res->server = server;
L
Linus Torvalds 已提交
3400
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE];
3401
	nfs4_init_sequence(&args->seq_args, &res->seq_res, 1);
3402 3403

	nfs_fattr_init(res->dir_attr);
L
Linus Torvalds 已提交
3404 3405
}

3406 3407
static void nfs4_proc_unlink_rpc_prepare(struct rpc_task *task, struct nfs_unlinkdata *data)
{
3408 3409 3410 3411
	nfs4_setup_sequence(NFS_SERVER(data->dir),
			&data->args.seq_args,
			&data->res.seq_res,
			task);
L
Linus Torvalds 已提交
3412 3413
}

3414
static int nfs4_proc_unlink_done(struct rpc_task *task, struct inode *dir)
L
Linus Torvalds 已提交
3415
{
3416 3417
	struct nfs_unlinkdata *data = task->tk_calldata;
	struct nfs_removeres *res = &data->res;
3418

3419 3420
	if (!nfs4_sequence_done(task, &res->seq_res))
		return 0;
3421
	if (nfs4_async_handle_error(task, res->server, NULL) == -EAGAIN)
3422 3423 3424
		return 0;
	update_changeattr(dir, &res->cinfo);
	return 1;
L
Linus Torvalds 已提交
3425 3426
}

3427 3428 3429 3430 3431 3432 3433 3434
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;
3435
	nfs4_init_sequence(&arg->seq_args, &res->seq_res, 1);
3436 3437
}

3438 3439
static void nfs4_proc_rename_rpc_prepare(struct rpc_task *task, struct nfs_renamedata *data)
{
3440 3441 3442 3443
	nfs4_setup_sequence(NFS_SERVER(data->old_dir),
			&data->args.seq_args,
			&data->res.seq_res,
			task);
3444 3445 3446 3447 3448
}

static int nfs4_proc_rename_done(struct rpc_task *task, struct inode *old_dir,
				 struct inode *new_dir)
{
3449 3450
	struct nfs_renamedata *data = task->tk_calldata;
	struct nfs_renameres *res = &data->res;
3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461

	if (!nfs4_sequence_done(task, &res->seq_res))
		return 0;
	if (nfs4_async_handle_error(task, res->server, NULL) == -EAGAIN)
		return 0;

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

L
Linus Torvalds 已提交
3462 3463 3464
static int _nfs4_proc_rename(struct inode *old_dir, struct qstr *old_name,
		struct inode *new_dir, struct qstr *new_name)
{
3465
	struct nfs_server *server = NFS_SERVER(old_dir);
3466
	struct nfs_renameargs arg = {
L
Linus Torvalds 已提交
3467 3468 3469 3470
		.old_dir = NFS_FH(old_dir),
		.new_dir = NFS_FH(new_dir),
		.old_name = old_name,
		.new_name = new_name,
3471
	};
3472
	struct nfs_renameres res = {
3473
		.server = server,
L
Linus Torvalds 已提交
3474 3475 3476 3477 3478 3479
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RENAME],
		.rpc_argp = &arg,
		.rpc_resp = &res,
	};
3480
	int status = -ENOMEM;
3481

3482
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
L
Linus Torvalds 已提交
3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495
	if (!status) {
		update_changeattr(old_dir, &res.old_cinfo);
		update_changeattr(new_dir, &res.new_cinfo);
	}
	return status;
}

static int nfs4_proc_rename(struct inode *old_dir, struct qstr *old_name,
		struct inode *new_dir, struct qstr *new_name)
{
	struct nfs4_exception exception = { };
	int err;
	do {
3496 3497 3498 3499
		err = _nfs4_proc_rename(old_dir, old_name,
					new_dir, new_name);
		trace_nfs4_rename(old_dir, old_name, new_dir, new_name, err);
		err = nfs4_handle_exception(NFS_SERVER(old_dir), err,
L
Linus Torvalds 已提交
3500 3501 3502 3503 3504 3505 3506
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_link(struct inode *inode, struct inode *dir, struct qstr *name)
{
3507
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
3508 3509 3510 3511
	struct nfs4_link_arg arg = {
		.fh     = NFS_FH(inode),
		.dir_fh = NFS_FH(dir),
		.name   = name,
3512 3513 3514 3515
		.bitmask = server->attr_bitmask,
	};
	struct nfs4_link_res res = {
		.server = server,
3516
		.label = NULL,
L
Linus Torvalds 已提交
3517 3518 3519 3520
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LINK],
		.rpc_argp = &arg,
3521
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3522
	};
3523 3524 3525
	int status = -ENOMEM;

	res.fattr = nfs_alloc_fattr();
3526
	if (res.fattr == NULL)
3527
		goto out;
L
Linus Torvalds 已提交
3528

3529 3530 3531 3532 3533
	res.label = nfs4_label_alloc(server, GFP_KERNEL);
	if (IS_ERR(res.label)) {
		status = PTR_ERR(res.label);
		goto out;
	}
3534
	arg.bitmask = nfs4_bitmask(server, res.label);
3535

3536
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
3537 3538
	if (!status) {
		update_changeattr(dir, &res.cinfo);
3539 3540 3541
		status = nfs_post_op_update_inode(inode, res.fattr);
		if (!status)
			nfs_setsecurity(inode, res.fattr, res.label);
3542
	}
3543 3544 3545 3546


	nfs4_label_free(res.label);

3547 3548
out:
	nfs_free_fattr(res.fattr);
L
Linus Torvalds 已提交
3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563
	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;
}

3564 3565 3566 3567 3568 3569
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;
3570
	struct nfs4_label *label;
3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581
};

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);

3582 3583 3584 3585
		data->label = nfs4_label_alloc(server, GFP_KERNEL);
		if (IS_ERR(data->label))
			goto out_free;

3586 3587 3588 3589 3590 3591 3592 3593
		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;
3594
		data->arg.bitmask = nfs4_bitmask(server, data->label);
3595 3596 3597
		data->res.server = server;
		data->res.fh = &data->fh;
		data->res.fattr = &data->fattr;
3598
		data->res.label = data->label;
3599 3600 3601
		nfs_fattr_init(data->res.fattr);
	}
	return data;
3602 3603 3604
out_free:
	kfree(data);
	return NULL;
3605 3606 3607 3608
}

static int nfs4_do_create(struct inode *dir, struct dentry *dentry, struct nfs4_createdata *data)
{
3609
	int status = nfs4_call_sync(NFS_SERVER(dir)->client, NFS_SERVER(dir), &data->msg,
3610
				    &data->arg.seq_args, &data->res.seq_res, 1);
3611 3612
	if (status == 0) {
		update_changeattr(dir, &data->res.dir_cinfo);
3613
		status = nfs_instantiate(dentry, data->res.fh, data->res.fattr, data->res.label);
3614 3615 3616 3617 3618 3619
	}
	return status;
}

static void nfs4_free_createdata(struct nfs4_createdata *data)
{
3620
	nfs4_label_free(data->label);
3621 3622 3623
	kfree(data);
}

3624
static int _nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
3625 3626
		struct page *page, unsigned int len, struct iattr *sattr,
		struct nfs4_label *label)
L
Linus Torvalds 已提交
3627
{
3628 3629
	struct nfs4_createdata *data;
	int status = -ENAMETOOLONG;
L
Linus Torvalds 已提交
3630

3631
	if (len > NFS4_MAXPATHLEN)
3632
		goto out;
3633

3634 3635 3636 3637 3638 3639 3640 3641
	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;
3642
	data->arg.label = label;
L
Linus Torvalds 已提交
3643
	
3644 3645 3646 3647
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
3648 3649 3650
	return status;
}

3651
static int nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
3652
		struct page *page, unsigned int len, struct iattr *sattr)
L
Linus Torvalds 已提交
3653 3654
{
	struct nfs4_exception exception = { };
3655
	struct nfs4_label l, *label = NULL;
L
Linus Torvalds 已提交
3656
	int err;
3657 3658 3659

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

L
Linus Torvalds 已提交
3660
	do {
3661 3662 3663
		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
3664 3665
				&exception);
	} while (exception.retry);
3666 3667

	nfs4_label_release_security(label);
L
Linus Torvalds 已提交
3668 3669 3670 3671
	return err;
}

static int _nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
3672
		struct iattr *sattr, struct nfs4_label *label)
L
Linus Torvalds 已提交
3673
{
3674 3675
	struct nfs4_createdata *data;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
3676

3677 3678 3679 3680
	data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4DIR);
	if (data == NULL)
		goto out;

3681
	data->arg.label = label;
3682 3683 3684 3685
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
3686 3687 3688 3689 3690 3691 3692
	return status;
}

static int nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr)
{
	struct nfs4_exception exception = { };
3693
	struct nfs4_label l, *label = NULL;
L
Linus Torvalds 已提交
3694
	int err;
A
Aneesh Kumar K.V 已提交
3695

3696 3697
	label = nfs4_label_init_security(dir, dentry, sattr, &l);

A
Aneesh Kumar K.V 已提交
3698
	sattr->ia_mode &= ~current_umask();
L
Linus Torvalds 已提交
3699
	do {
3700 3701 3702
		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
3703 3704
				&exception);
	} while (exception.retry);
3705 3706
	nfs4_label_release_security(label);

L
Linus Torvalds 已提交
3707 3708 3709 3710
	return err;
}

static int _nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
B
Bryan Schumaker 已提交
3711
		u64 cookie, struct page **pages, unsigned int count, int plus)
L
Linus Torvalds 已提交
3712 3713 3714 3715
{
	struct inode		*dir = dentry->d_inode;
	struct nfs4_readdir_arg args = {
		.fh = NFS_FH(dir),
B
Bryan Schumaker 已提交
3716
		.pages = pages,
L
Linus Torvalds 已提交
3717 3718
		.pgbase = 0,
		.count = count,
3719
		.bitmask = NFS_SERVER(dentry->d_inode)->attr_bitmask,
B
Bryan Schumaker 已提交
3720
		.plus = plus,
L
Linus Torvalds 已提交
3721 3722 3723 3724 3725 3726 3727 3728 3729 3730
	};
	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;

3731
	dprintk("%s: dentry = %s/%s, cookie = %Lu\n", __func__,
3732 3733 3734
			dentry->d_parent->d_name.name,
			dentry->d_name.name,
			(unsigned long long)cookie);
3735
	nfs4_setup_readdir(cookie, NFS_I(dir)->cookieverf, dentry, &args);
L
Linus Torvalds 已提交
3736
	res.pgbase = args.pgbase;
3737
	status = nfs4_call_sync(NFS_SERVER(dir)->client, NFS_SERVER(dir), &msg, &args.seq_args, &res.seq_res, 0);
3738
	if (status >= 0) {
3739
		memcpy(NFS_I(dir)->cookieverf, res.verifier.data, NFS4_VERIFIER_SIZE);
3740 3741
		status += args.pgbase;
	}
3742 3743 3744

	nfs_invalidate_atime(dir);

3745
	dprintk("%s: returns %d\n", __func__, status);
L
Linus Torvalds 已提交
3746 3747 3748 3749
	return status;
}

static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
B
Bryan Schumaker 已提交
3750
		u64 cookie, struct page **pages, unsigned int count, int plus)
L
Linus Torvalds 已提交
3751 3752 3753 3754
{
	struct nfs4_exception exception = { };
	int err;
	do {
3755 3756 3757 3758
		err = _nfs4_proc_readdir(dentry, cred, cookie,
				pages, count, plus);
		trace_nfs4_readdir(dentry->d_inode, err);
		err = nfs4_handle_exception(NFS_SERVER(dentry->d_inode), err,
L
Linus Torvalds 已提交
3759 3760 3761 3762 3763 3764
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
3765
		struct iattr *sattr, struct nfs4_label *label, dev_t rdev)
L
Linus Torvalds 已提交
3766
{
3767 3768 3769
	struct nfs4_createdata *data;
	int mode = sattr->ia_mode;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
3770

3771 3772 3773 3774
	data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4SOCK);
	if (data == NULL)
		goto out;

L
Linus Torvalds 已提交
3775
	if (S_ISFIFO(mode))
3776
		data->arg.ftype = NF4FIFO;
L
Linus Torvalds 已提交
3777
	else if (S_ISBLK(mode)) {
3778 3779 3780
		data->arg.ftype = NF4BLK;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
L
Linus Torvalds 已提交
3781 3782
	}
	else if (S_ISCHR(mode)) {
3783 3784 3785
		data->arg.ftype = NF4CHR;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
3786 3787 3788
	} else if (!S_ISSOCK(mode)) {
		status = -EINVAL;
		goto out_free;
L
Linus Torvalds 已提交
3789
	}
3790

3791
	data->arg.label = label;
3792
	status = nfs4_do_create(dir, dentry, data);
3793
out_free:
3794 3795
	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
3796 3797 3798 3799 3800 3801 3802
	return status;
}

static int nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr, dev_t rdev)
{
	struct nfs4_exception exception = { };
3803
	struct nfs4_label l, *label = NULL;
L
Linus Torvalds 已提交
3804
	int err;
A
Aneesh Kumar K.V 已提交
3805

3806 3807
	label = nfs4_label_init_security(dir, dentry, sattr, &l);

A
Aneesh Kumar K.V 已提交
3808
	sattr->ia_mode &= ~current_umask();
L
Linus Torvalds 已提交
3809
	do {
3810 3811 3812
		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
3813 3814
				&exception);
	} while (exception.retry);
3815 3816 3817

	nfs4_label_release_security(label);

L
Linus Torvalds 已提交
3818 3819 3820 3821 3822 3823 3824 3825 3826 3827
	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
3828 3829 3830
	struct nfs4_statfs_res res = {
		.fsstat = fsstat,
	};
L
Linus Torvalds 已提交
3831 3832 3833
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_STATFS],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3834
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3835 3836
	};

3837
	nfs_fattr_init(fsstat->fattr);
3838
	return  nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859
}

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
3860 3861 3862
	struct nfs4_fsinfo_res res = {
		.fsinfo = fsinfo,
	};
L
Linus Torvalds 已提交
3863 3864 3865
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FSINFO],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3866
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3867 3868
	};

3869
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3870 3871 3872 3873 3874
}

static int nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo)
{
	struct nfs4_exception exception = { };
3875
	unsigned long now = jiffies;
L
Linus Torvalds 已提交
3876 3877 3878
	int err;

	do {
3879
		err = _nfs4_do_fsinfo(server, fhandle, fsinfo);
3880
		trace_nfs4_fsinfo(server, fhandle, fsinfo->fattr, err);
3881 3882 3883 3884 3885 3886 3887 3888 3889 3890
		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
3891 3892 3893 3894 3895 3896
	} while (exception.retry);
	return err;
}

static int nfs4_proc_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo)
{
3897 3898
	int error;

3899
	nfs_fattr_init(fsinfo->fattr);
3900
	error = nfs4_do_fsinfo(server, fhandle, fsinfo);
3901 3902 3903
	if (error == 0) {
		/* block layout checks this! */
		server->pnfs_blksize = fsinfo->blksize;
3904
		set_pnfs_layoutdriver(server, fhandle, fsinfo->layouttype);
3905
	}
3906 3907

	return error;
L
Linus Torvalds 已提交
3908 3909 3910 3911 3912 3913 3914 3915 3916
}

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
3917 3918 3919
	struct nfs4_pathconf_res res = {
		.pathconf = pathconf,
	};
L
Linus Torvalds 已提交
3920 3921 3922
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_PATHCONF],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3923
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3924 3925 3926 3927 3928 3929 3930 3931
	};

	/* 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;
	}

3932
	nfs_fattr_init(pathconf->fattr);
3933
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949
}

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;
}

3950
int nfs4_set_rw_stateid(nfs4_stateid *stateid,
3951 3952 3953 3954 3955 3956 3957 3958
		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;
3959
	return nfs4_select_rw_stateid(stateid, ctx->state, fmode, lockowner);
3960 3961 3962
}
EXPORT_SYMBOL_GPL(nfs4_set_rw_stateid);

3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989
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;

	if (nfs4_set_rw_stateid(&current_stateid, ctx, l_ctx, fmode))
		return false;
	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;
}

3990 3991
void __nfs4_read_done_cb(struct nfs_read_data *data)
{
3992
	nfs_invalidate_atime(data->header->inode);
3993 3994
}

3995
static int nfs4_read_done_cb(struct rpc_task *task, struct nfs_read_data *data)
L
Linus Torvalds 已提交
3996
{
3997
	struct nfs_server *server = NFS_SERVER(data->header->inode);
L
Linus Torvalds 已提交
3998

3999
	trace_nfs4_read(data, task->tk_status);
4000
	if (nfs4_async_handle_error(task, server, data->args.context->state) == -EAGAIN) {
4001
		rpc_restart_call_prepare(task);
T
Trond Myklebust 已提交
4002
		return -EAGAIN;
L
Linus Torvalds 已提交
4003
	}
4004

4005
	__nfs4_read_done_cb(data);
L
Linus Torvalds 已提交
4006
	if (task->tk_status > 0)
T
Trond Myklebust 已提交
4007 4008
		renew_lease(server, data->timestamp);
	return 0;
L
Linus Torvalds 已提交
4009 4010
}

4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024
static bool nfs4_read_stateid_changed(struct rpc_task *task,
		struct nfs_readargs *args)
{

	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;
}

4025 4026 4027 4028 4029 4030 4031
static int nfs4_read_done(struct rpc_task *task, struct nfs_read_data *data)
{

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

	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return -EAGAIN;
4032 4033
	if (nfs4_read_stateid_changed(task, &data->args))
		return -EAGAIN;
4034 4035
	return data->read_done_cb ? data->read_done_cb(task, data) :
				    nfs4_read_done_cb(task, data);
4036 4037
}

4038
static void nfs4_proc_read_setup(struct nfs_read_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
4039 4040
{
	data->timestamp   = jiffies;
4041
	data->read_done_cb = nfs4_read_done_cb;
4042
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READ];
4043
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 0);
L
Linus Torvalds 已提交
4044 4045
}

4046
static int nfs4_proc_read_rpc_prepare(struct rpc_task *task, struct nfs_read_data *data)
4047
{
4048
	if (nfs4_setup_sequence(NFS_SERVER(data->header->inode),
4049 4050
			&data->args.seq_args,
			&data->res.seq_res,
4051
			task))
4052 4053 4054 4055 4056 4057 4058
		return 0;
	if (nfs4_set_rw_stateid(&data->args.stateid, data->args.context,
				data->args.lock_context, FMODE_READ) == -EIO)
		return -EIO;
	if (unlikely(test_bit(NFS_CONTEXT_BAD, &data->args.context->flags)))
		return -EIO;
	return 0;
L
Linus Torvalds 已提交
4059 4060
}

4061
static int nfs4_write_done_cb(struct rpc_task *task, struct nfs_write_data *data)
L
Linus Torvalds 已提交
4062
{
4063
	struct inode *inode = data->header->inode;
L
Linus Torvalds 已提交
4064
	
4065
	trace_nfs4_write(data, task->tk_status);
4066
	if (nfs4_async_handle_error(task, NFS_SERVER(inode), data->args.context->state) == -EAGAIN) {
4067
		rpc_restart_call_prepare(task);
4068
		return -EAGAIN;
L
Linus Torvalds 已提交
4069
	}
4070
	if (task->tk_status >= 0) {
L
Linus Torvalds 已提交
4071
		renew_lease(NFS_SERVER(inode), data->timestamp);
4072
		nfs_post_op_update_inode_force_wcc(inode, &data->fattr);
4073
	}
4074
	return 0;
L
Linus Torvalds 已提交
4075 4076
}

4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090
static bool nfs4_write_stateid_changed(struct rpc_task *task,
		struct nfs_writeargs *args)
{

	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;
}

4091 4092 4093 4094
static int nfs4_write_done(struct rpc_task *task, struct nfs_write_data *data)
{
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return -EAGAIN;
4095 4096
	if (nfs4_write_stateid_changed(task, &data->args))
		return -EAGAIN;
4097 4098
	return data->write_done_cb ? data->write_done_cb(task, data) :
		nfs4_write_done_cb(task, data);
4099 4100
}

4101 4102
static
bool nfs4_write_need_cache_consistency_data(const struct nfs_write_data *data)
4103
{
4104 4105 4106 4107 4108 4109 4110 4111
	const struct nfs_pgio_header *hdr = data->header;

	/* Don't request attributes for pNFS or O_DIRECT writes */
	if (data->ds_clp != NULL || hdr->dreq != NULL)
		return false;
	/* Otherwise, request attributes if and only if we don't hold
	 * a delegation
	 */
4112
	return nfs4_have_delegation(hdr->inode, FMODE_READ) == 0;
4113 4114
}

4115
static void nfs4_proc_write_setup(struct nfs_write_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
4116
{
4117
	struct nfs_server *server = NFS_SERVER(data->header->inode);
4118

4119
	if (!nfs4_write_need_cache_consistency_data(data)) {
4120 4121 4122 4123
		data->args.bitmask = NULL;
		data->res.fattr = NULL;
	} else
		data->args.bitmask = server->cache_consistency_bitmask;
4124

4125 4126
	if (!data->write_done_cb)
		data->write_done_cb = nfs4_write_done_cb;
4127
	data->res.server = server;
L
Linus Torvalds 已提交
4128 4129
	data->timestamp   = jiffies;

4130
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_WRITE];
4131
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
L
Linus Torvalds 已提交
4132 4133
}

4134
static int nfs4_proc_write_rpc_prepare(struct rpc_task *task, struct nfs_write_data *data)
4135
{
4136
	if (nfs4_setup_sequence(NFS_SERVER(data->header->inode),
4137 4138
			&data->args.seq_args,
			&data->res.seq_res,
4139
			task))
4140 4141 4142 4143 4144 4145 4146
		return 0;
	if (nfs4_set_rw_stateid(&data->args.stateid, data->args.context,
				data->args.lock_context, FMODE_WRITE) == -EIO)
		return -EIO;
	if (unlikely(test_bit(NFS_CONTEXT_BAD, &data->args.context->flags)))
		return -EIO;
	return 0;
L
Linus Torvalds 已提交
4147 4148
}

4149
static void nfs4_proc_commit_rpc_prepare(struct rpc_task *task, struct nfs_commit_data *data)
4150
{
4151 4152 4153 4154
	nfs4_setup_sequence(NFS_SERVER(data->inode),
			&data->args.seq_args,
			&data->res.seq_res,
			task);
L
Linus Torvalds 已提交
4155 4156
}

4157
static int nfs4_commit_done_cb(struct rpc_task *task, struct nfs_commit_data *data)
L
Linus Torvalds 已提交
4158 4159
{
	struct inode *inode = data->inode;
4160

4161
	trace_nfs4_commit(data, task->tk_status);
4162
	if (nfs4_async_handle_error(task, NFS_SERVER(inode), NULL) == -EAGAIN) {
4163
		rpc_restart_call_prepare(task);
4164
		return -EAGAIN;
L
Linus Torvalds 已提交
4165
	}
4166
	return 0;
L
Linus Torvalds 已提交
4167 4168
}

4169
static int nfs4_commit_done(struct rpc_task *task, struct nfs_commit_data *data)
4170 4171 4172
{
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return -EAGAIN;
4173
	return data->commit_done_cb(task, data);
4174 4175
}

4176
static void nfs4_proc_commit_setup(struct nfs_commit_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
4177
{
4178
	struct nfs_server *server = NFS_SERVER(data->inode);
4179

4180 4181
	if (data->commit_done_cb == NULL)
		data->commit_done_cb = nfs4_commit_done_cb;
4182
	data->res.server = server;
4183
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_COMMIT];
4184
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
L
Linus Torvalds 已提交
4185 4186
}

4187 4188 4189 4190 4191
struct nfs4_renewdata {
	struct nfs_client	*client;
	unsigned long		timestamp;
};

L
Linus Torvalds 已提交
4192 4193 4194 4195
/*
 * nfs4_proc_async_renew(): This is not one of the nfs_rpc_ops; it is a special
 * standalone procedure for queueing an asynchronous RENEW.
 */
4196
static void nfs4_renew_release(void *calldata)
4197
{
4198 4199
	struct nfs4_renewdata *data = calldata;
	struct nfs_client *clp = data->client;
4200

4201 4202 4203
	if (atomic_read(&clp->cl_count) > 1)
		nfs4_schedule_state_renewal(clp);
	nfs_put_client(clp);
4204
	kfree(data);
4205 4206
}

4207
static void nfs4_renew_done(struct rpc_task *task, void *calldata)
L
Linus Torvalds 已提交
4208
{
4209 4210 4211
	struct nfs4_renewdata *data = calldata;
	struct nfs_client *clp = data->client;
	unsigned long timestamp = data->timestamp;
L
Linus Torvalds 已提交
4212

4213
	trace_nfs4_renew_async(clp, task->tk_status);
L
Linus Torvalds 已提交
4214
	if (task->tk_status < 0) {
4215
		/* Unless we're shutting down, schedule state recovery! */
4216 4217 4218
		if (test_bit(NFS_CS_RENEWD, &clp->cl_res_state) == 0)
			return;
		if (task->tk_status != NFS4ERR_CB_PATH_DOWN) {
4219
			nfs4_schedule_lease_recovery(clp);
4220 4221 4222
			return;
		}
		nfs4_schedule_path_down_recovery(clp);
L
Linus Torvalds 已提交
4223
	}
4224
	do_renew_lease(clp, timestamp);
L
Linus Torvalds 已提交
4225 4226
}

4227 4228
static const struct rpc_call_ops nfs4_renew_ops = {
	.rpc_call_done = nfs4_renew_done,
4229
	.rpc_release = nfs4_renew_release,
4230 4231
};

4232
static int nfs4_proc_async_renew(struct nfs_client *clp, struct rpc_cred *cred, unsigned renew_flags)
L
Linus Torvalds 已提交
4233 4234 4235 4236
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
4237
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
4238
	};
4239
	struct nfs4_renewdata *data;
L
Linus Torvalds 已提交
4240

4241 4242
	if (renew_flags == 0)
		return 0;
4243 4244
	if (!atomic_inc_not_zero(&clp->cl_count))
		return -EIO;
4245
	data = kmalloc(sizeof(*data), GFP_NOFS);
4246 4247 4248 4249
	if (data == NULL)
		return -ENOMEM;
	data->client = clp;
	data->timestamp = jiffies;
4250
	return rpc_call_async(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT,
4251
			&nfs4_renew_ops, data);
L
Linus Torvalds 已提交
4252 4253
}

4254
static int nfs4_proc_renew(struct nfs_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
4255 4256 4257 4258
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
4259
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
4260 4261 4262 4263
	};
	unsigned long now = jiffies;
	int status;

4264
	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
L
Linus Torvalds 已提交
4265 4266
	if (status < 0)
		return status;
4267
	do_renew_lease(clp, now);
L
Linus Torvalds 已提交
4268 4269 4270
	return 0;
}

4271 4272 4273 4274 4275 4276 4277
static inline int nfs4_server_supports_acls(struct nfs_server *server)
{
	return (server->caps & NFS_CAP_ACLS)
		&& (server->acl_bitmask & ACL4_SUPPORT_ALLOW_ACL)
		&& (server->acl_bitmask & ACL4_SUPPORT_DENY_ACL);
}

4278 4279
/* 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
4280 4281
 * the stack.
 */
4282
#define NFS4ACL_MAXPAGES DIV_ROUND_UP(XATTR_SIZE_MAX, PAGE_SIZE)
4283

4284 4285 4286 4287 4288 4289 4290 4291 4292
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 {
4293
		len = min_t(size_t, PAGE_SIZE, buflen);
4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312
		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;
}

4313 4314 4315
struct nfs4_cached_acl {
	int cached;
	size_t len;
A
Andrew Morton 已提交
4316
	char data[0];
4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358
};

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;
}

4359
static void nfs4_write_cached_acl(struct inode *inode, struct page **pages, size_t pgbase, size_t acl_len)
4360 4361
{
	struct nfs4_cached_acl *acl;
4362
	size_t buflen = sizeof(*acl) + acl_len;
4363

4364
	if (buflen <= PAGE_SIZE) {
4365
		acl = kmalloc(buflen, GFP_KERNEL);
4366 4367 4368
		if (acl == NULL)
			goto out;
		acl->cached = 1;
4369
		_copy_from_pages(acl->data, pages, pgbase, acl_len);
4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380
	} 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);
}

4381 4382 4383 4384 4385 4386 4387 4388 4389 4390
/*
 * 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.
 */
4391
static ssize_t __nfs4_get_acl_uncached(struct inode *inode, void *buf, size_t buflen)
4392
{
4393
	struct page *pages[NFS4ACL_MAXPAGES] = {NULL, };
4394 4395 4396 4397 4398
	struct nfs_getaclargs args = {
		.fh = NFS_FH(inode),
		.acl_pages = pages,
		.acl_len = buflen,
	};
B
Benny Halevy 已提交
4399 4400 4401
	struct nfs_getaclres res = {
		.acl_len = buflen,
	};
4402 4403 4404
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETACL],
		.rpc_argp = &args,
B
Benny Halevy 已提交
4405
		.rpc_resp = &res,
4406
	};
4407 4408
	unsigned int npages = DIV_ROUND_UP(buflen, PAGE_SIZE);
	int ret = -ENOMEM, i;
4409

4410 4411 4412 4413
	/* 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;
4414 4415
	if (npages > ARRAY_SIZE(pages))
		return -ERANGE;
4416

4417 4418 4419 4420
	for (i = 0; i < npages; i++) {
		pages[i] = alloc_page(GFP_KERNEL);
		if (!pages[i])
			goto out_free;
4421
	}
4422 4423 4424 4425 4426 4427

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

4428 4429
	args.acl_len = npages * PAGE_SIZE;
	args.acl_pgbase = 0;
4430

4431
	dprintk("%s  buf %p buflen %zu npages %d args.acl_len %zu\n",
4432 4433 4434
		__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);
4435 4436
	if (ret)
		goto out_free;
4437

4438 4439 4440 4441 4442
	/* 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;
4443
		ret = -ERANGE;
4444
		goto out_free;
4445
	}
4446
	nfs4_write_cached_acl(inode, pages, res.acl_data_offset, res.acl_len);
4447 4448 4449 4450 4451
	if (buf) {
		if (res.acl_len > buflen) {
			ret = -ERANGE;
			goto out_free;
		}
4452
		_copy_from_pages(buf, pages, res.acl_data_offset, res.acl_len);
4453
	}
4454 4455
out_ok:
	ret = res.acl_len;
4456
out_free:
4457 4458 4459
	for (i = 0; i < npages; i++)
		if (pages[i])
			__free_page(pages[i]);
4460 4461
	if (res.acl_scratch)
		__free_page(res.acl_scratch);
4462 4463 4464
	return ret;
}

4465 4466 4467 4468 4469 4470
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);
4471
		trace_nfs4_get_acl(inode, ret);
4472 4473 4474 4475 4476 4477 4478
		if (ret >= 0)
			break;
		ret = nfs4_handle_exception(NFS_SERVER(inode), ret, &exception);
	} while (exception.retry);
	return ret;
}

4479 4480 4481 4482 4483 4484 4485 4486 4487 4488
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;
4489 4490
	if (NFS_I(inode)->cache_validity & NFS_INO_INVALID_ACL)
		nfs_zap_acl_cache(inode);
4491 4492
	ret = nfs4_read_cached_acl(inode, buf, buflen);
	if (ret != -ENOENT)
4493 4494
		/* -ENOENT is returned if there is no ACL or if there is an ACL
		 * but no cached acl data, just the acl length */
4495 4496 4497 4498
		return ret;
	return nfs4_get_acl_uncached(inode, buf, buflen);
}

4499
static int __nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen)
4500 4501 4502 4503 4504 4505 4506 4507
{
	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
4508
	struct nfs_setaclres res;
4509 4510 4511
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETACL],
		.rpc_argp	= &arg,
B
Benny Halevy 已提交
4512
		.rpc_resp	= &res,
4513
	};
4514
	unsigned int npages = DIV_ROUND_UP(buflen, PAGE_SIZE);
4515
	int ret, i;
4516 4517 4518

	if (!nfs4_server_supports_acls(server))
		return -EOPNOTSUPP;
4519 4520
	if (npages > ARRAY_SIZE(pages))
		return -ERANGE;
4521 4522 4523
	i = buf_to_pages_noslab(buf, buflen, arg.acl_pages, &arg.acl_pgbase);
	if (i < 0)
		return i;
4524
	nfs4_inode_return_delegation(inode);
4525
	ret = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
4526 4527 4528 4529 4530 4531 4532 4533

	/*
	 * 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]);

4534 4535 4536 4537 4538 4539 4540
	/*
	 * 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);
4541 4542
	nfs_access_zap_cache(inode);
	nfs_zap_acl_cache(inode);
4543 4544 4545
	return ret;
}

4546 4547 4548 4549 4550
static int nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen)
{
	struct nfs4_exception exception = { };
	int err;
	do {
4551 4552 4553
		err = __nfs4_proc_set_acl(inode, buf, buflen);
		trace_nfs4_set_acl(inode, err);
		err = nfs4_handle_exception(NFS_SERVER(inode), err,
4554 4555 4556 4557 4558
				&exception);
	} while (exception.retry);
	return err;
}

4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605
#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 };
	struct nfs4_getattr_arg args = {
		.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],
		.rpc_argp	= &args,
		.rpc_resp	= &res,
	};
	int ret;

	nfs_fattr_init(&fattr);

	ret = rpc_call_sync(server->client, &msg, 0);
	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 {
4606 4607 4608
		err = _nfs4_get_security_label(inode, buf, buflen);
		trace_nfs4_get_security_label(inode, err);
		err = nfs4_handle_exception(NFS_SERVER(inode), err,
4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659
				&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 };
	struct nfs_setattrargs args = {
		.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],
		.rpc_argp       = &args,
		.rpc_resp       = &res,
	};
	int status;

	nfs4_stateid_copy(&args.stateid, &zero_stateid);

	status = rpc_call_sync(server->client, &msg, 0);
	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 {
4660 4661 4662 4663
		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,
4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709
				&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;
	struct inode *inode = dentry->d_inode;
	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
4710
static int
4711
nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server, struct nfs4_state *state)
L
Linus Torvalds 已提交
4712
{
4713 4714 4715
	struct nfs_client *clp = server->nfs_client;

	if (task->tk_status >= 0)
L
Linus Torvalds 已提交
4716 4717
		return 0;
	switch(task->tk_status) {
4718
		case -NFS4ERR_DELEG_REVOKED:
4719 4720
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
4721 4722 4723
			if (state == NULL)
				break;
			nfs_remove_bad_delegation(state->inode);
4724 4725 4726
		case -NFS4ERR_OPENMODE:
			if (state == NULL)
				break;
4727 4728
			if (nfs4_schedule_stateid_recovery(server, state) < 0)
				goto stateid_invalid;
4729
			goto wait_on_recovery;
4730
		case -NFS4ERR_EXPIRED:
4731 4732 4733 4734
			if (state != NULL) {
				if (nfs4_schedule_stateid_recovery(server, state) < 0)
					goto stateid_invalid;
			}
L
Linus Torvalds 已提交
4735
		case -NFS4ERR_STALE_STATEID:
4736
		case -NFS4ERR_STALE_CLIENTID:
4737 4738
			nfs4_schedule_lease_recovery(clp);
			goto wait_on_recovery;
4739 4740 4741 4742 4743 4744 4745 4746 4747 4748
#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);
4749
			nfs4_schedule_session_recovery(clp->cl_session, task->tk_status);
4750 4751 4752
			task->tk_status = 0;
			return -EAGAIN;
#endif /* CONFIG_NFS_V4_1 */
L
Linus Torvalds 已提交
4753
		case -NFS4ERR_DELAY:
4754
			nfs_inc_server_stats(server, NFSIOS_DELAY);
4755
		case -NFS4ERR_GRACE:
L
Linus Torvalds 已提交
4756 4757 4758
			rpc_delay(task, NFS4_POLL_RETRY_MAX);
			task->tk_status = 0;
			return -EAGAIN;
4759
		case -NFS4ERR_RETRY_UNCACHED_REP:
L
Linus Torvalds 已提交
4760 4761 4762 4763 4764 4765
		case -NFS4ERR_OLD_STATEID:
			task->tk_status = 0;
			return -EAGAIN;
	}
	task->tk_status = nfs4_map_errors(task->tk_status);
	return 0;
4766 4767 4768
stateid_invalid:
	task->tk_status = -EIO;
	return 0;
4769
wait_on_recovery:
4770 4771 4772 4773 4774
	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);
	task->tk_status = 0;
	return -EAGAIN;
L
Linus Torvalds 已提交
4775 4776
}

4777 4778
static void nfs4_init_boot_verifier(const struct nfs_client *clp,
				    nfs4_verifier *bootverf)
4779 4780 4781
{
	__be32 verf[2];

4782 4783 4784 4785
	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;
4786
		verf[1] = cpu_to_be32(NSEC_PER_SEC + 1);
4787
	} else {
4788
		struct nfs_net *nn = net_generic(clp->cl_net, nfs_net_id);
4789 4790
		verf[0] = cpu_to_be32(nn->boot_time.tv_sec);
		verf[1] = cpu_to_be32(nn->boot_time.tv_nsec);
4791
	}
4792 4793 4794
	memcpy(bootverf->data, verf, sizeof(bootverf->data));
}

4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815
static unsigned int
nfs4_init_nonuniform_client_string(const struct nfs_client *clp,
				   char *buf, size_t len)
{
	unsigned int result;

	rcu_read_lock();
	result = scnprintf(buf, 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));
	rcu_read_unlock();
	return result;
}

static unsigned int
nfs4_init_uniform_client_string(const struct nfs_client *clp,
				char *buf, size_t len)
{
4816
	const char *nodename = clp->cl_rpcclient->cl_nodename;
4817 4818

	if (nfs4_client_id_uniquifier[0] != '\0')
4819 4820 4821 4822 4823
		return scnprintf(buf, len, "Linux NFSv%u.%u %s/%s",
				clp->rpc_ops->version,
				clp->cl_minorversion,
				nfs4_client_id_uniquifier,
				nodename);
4824 4825
	return scnprintf(buf, len, "Linux NFSv%u.%u %s",
				clp->rpc_ops->version, clp->cl_minorversion,
4826
				nodename);
4827 4828
}

4829 4830 4831 4832 4833 4834 4835 4836 4837 4838
/**
 * 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.
 */
4839 4840 4841
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
4842 4843 4844 4845 4846
{
	nfs4_verifier sc_verifier;
	struct nfs4_setclientid setclientid = {
		.sc_verifier = &sc_verifier,
		.sc_prog = program,
A
Andy Adamson 已提交
4847
		.sc_cb_ident = clp->cl_cb_ident,
L
Linus Torvalds 已提交
4848 4849 4850 4851
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID],
		.rpc_argp = &setclientid,
4852
		.rpc_resp = res,
4853
		.rpc_cred = cred,
L
Linus Torvalds 已提交
4854
	};
4855
	int status;
L
Linus Torvalds 已提交
4856

4857
	/* nfs_client_id4 */
4858
	nfs4_init_boot_verifier(clp, &sc_verifier);
4859 4860 4861 4862 4863 4864 4865 4866 4867 4868
	if (test_bit(NFS_CS_MIGRATION, &clp->cl_flags))
		setclientid.sc_name_len =
				nfs4_init_uniform_client_string(clp,
						setclientid.sc_name,
						sizeof(setclientid.sc_name));
	else
		setclientid.sc_name_len =
				nfs4_init_nonuniform_client_string(clp,
						setclientid.sc_name,
						sizeof(setclientid.sc_name));
4869
	/* cb_client4 */
4870
	rcu_read_lock();
4871
	setclientid.sc_netid_len = scnprintf(setclientid.sc_netid,
4872
				sizeof(setclientid.sc_netid), "%s",
4873 4874
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_NETID));
4875 4876
	rcu_read_unlock();
	setclientid.sc_uaddr_len = scnprintf(setclientid.sc_uaddr,
4877
				sizeof(setclientid.sc_uaddr), "%s.%u.%u",
L
Linus Torvalds 已提交
4878 4879
				clp->cl_ipaddr, port >> 8, port & 255);

4880 4881 4882 4883
	dprintk("NFS call  setclientid auth=%s, '%.*s'\n",
		clp->cl_rpcclient->cl_auth->au_ops->au_name,
		setclientid.sc_name_len, setclientid.sc_name);
	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
4884
	trace_nfs4_setclientid(clp, status);
4885 4886
	dprintk("NFS reply setclientid: %d\n", status);
	return status;
L
Linus Torvalds 已提交
4887 4888
}

4889 4890 4891 4892 4893 4894 4895 4896
/**
 * 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.
 */
4897
int nfs4_proc_setclientid_confirm(struct nfs_client *clp,
4898 4899
		struct nfs4_setclientid_res *arg,
		struct rpc_cred *cred)
L
Linus Torvalds 已提交
4900 4901 4902
{
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID_CONFIRM],
4903
		.rpc_argp = arg,
4904
		.rpc_cred = cred,
L
Linus Torvalds 已提交
4905 4906 4907
	};
	int status;

4908 4909 4910
	dprintk("NFS call  setclientid_confirm auth=%s, (client ID %llx)\n",
		clp->cl_rpcclient->cl_auth->au_ops->au_name,
		clp->cl_clientid);
4911
	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
4912
	trace_nfs4_setclientid_confirm(clp, status);
4913
	dprintk("NFS reply setclientid_confirm: %d\n", status);
L
Linus Torvalds 已提交
4914 4915 4916
	return status;
}

4917 4918
struct nfs4_delegreturndata {
	struct nfs4_delegreturnargs args;
4919
	struct nfs4_delegreturnres res;
4920 4921
	struct nfs_fh fh;
	nfs4_stateid stateid;
4922
	unsigned long timestamp;
4923
	struct nfs_fattr fattr;
4924 4925 4926 4927 4928 4929
	int rpc_status;
};

static void nfs4_delegreturn_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_delegreturndata *data = calldata;
4930

4931 4932
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return;
4933

4934
	trace_nfs4_delegreturn_exit(&data->args, &data->res, task->tk_status);
4935 4936 4937 4938
	switch (task->tk_status) {
	case -NFS4ERR_STALE_STATEID:
	case -NFS4ERR_EXPIRED:
	case 0:
4939
		renew_lease(data->res.server, data->timestamp);
4940 4941 4942 4943
		break;
	default:
		if (nfs4_async_handle_error(task, data->res.server, NULL) ==
				-EAGAIN) {
4944
			rpc_restart_call_prepare(task);
4945 4946 4947 4948
			return;
		}
	}
	data->rpc_status = task->tk_status;
4949 4950 4951 4952 4953 4954 4955
}

static void nfs4_delegreturn_release(void *calldata)
{
	kfree(calldata);
}

4956 4957 4958 4959 4960 4961
static void nfs4_delegreturn_prepare(struct rpc_task *task, void *data)
{
	struct nfs4_delegreturndata *d_data;

	d_data = (struct nfs4_delegreturndata *)data;

4962 4963 4964 4965
	nfs4_setup_sequence(d_data->res.server,
			&d_data->args.seq_args,
			&d_data->res.seq_res,
			task);
4966 4967
}

4968
static const struct rpc_call_ops nfs4_delegreturn_ops = {
4969
	.rpc_call_prepare = nfs4_delegreturn_prepare,
4970 4971 4972 4973
	.rpc_call_done = nfs4_delegreturn_done,
	.rpc_release = nfs4_delegreturn_release,
};

4974
static int _nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
4975 4976
{
	struct nfs4_delegreturndata *data;
4977
	struct nfs_server *server = NFS_SERVER(inode);
4978
	struct rpc_task *task;
4979 4980 4981 4982
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DELEGRETURN],
		.rpc_cred = cred,
	};
T
Trond Myklebust 已提交
4983 4984
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
4985
		.rpc_message = &msg,
T
Trond Myklebust 已提交
4986 4987 4988
		.callback_ops = &nfs4_delegreturn_ops,
		.flags = RPC_TASK_ASYNC,
	};
4989
	int status = 0;
4990

4991
	data = kzalloc(sizeof(*data), GFP_NOFS);
4992 4993
	if (data == NULL)
		return -ENOMEM;
4994
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
4995 4996
	data->args.fhandle = &data->fh;
	data->args.stateid = &data->stateid;
4997
	data->args.bitmask = server->cache_consistency_bitmask;
4998
	nfs_copy_fh(&data->fh, NFS_FH(inode));
4999
	nfs4_stateid_copy(&data->stateid, stateid);
5000 5001
	data->res.fattr = &data->fattr;
	data->res.server = server;
5002
	nfs_fattr_init(data->res.fattr);
5003
	data->timestamp = jiffies;
5004 5005
	data->rpc_status = 0;

T
Trond Myklebust 已提交
5006
	task_setup_data.callback_data = data;
5007 5008
	msg.rpc_argp = &data->args;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
5009
	task = rpc_run_task(&task_setup_data);
5010
	if (IS_ERR(task))
5011
		return PTR_ERR(task);
5012 5013
	if (!issync)
		goto out;
5014
	status = nfs4_wait_for_completion_rpc_task(task);
5015 5016 5017
	if (status != 0)
		goto out;
	status = data->rpc_status;
5018 5019 5020 5021
	if (status == 0)
		nfs_post_op_update_inode_force_wcc(inode, &data->fattr);
	else
		nfs_refresh_inode(inode, &data->fattr);
5022
out:
5023
	rpc_put_task(task);
5024
	return status;
L
Linus Torvalds 已提交
5025 5026
}

5027
int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
L
Linus Torvalds 已提交
5028 5029 5030 5031 5032
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs4_exception exception = { };
	int err;
	do {
5033
		err = _nfs4_proc_delegreturn(inode, cred, stateid, issync);
5034
		trace_nfs4_delegreturn(inode, err);
L
Linus Torvalds 已提交
5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054
		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)
{
5055
	freezable_schedule_timeout_killable_unsafe(timeout);
L
Linus Torvalds 已提交
5056 5057 5058 5059 5060 5061 5062 5063 5064 5065
	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);
5066
	struct nfs_client *clp = server->nfs_client;
T
Trond Myklebust 已提交
5067
	struct nfs_lockt_args arg = {
L
Linus Torvalds 已提交
5068
		.fh = NFS_FH(inode),
T
Trond Myklebust 已提交
5069
		.fl = request,
L
Linus Torvalds 已提交
5070
	};
T
Trond Myklebust 已提交
5071 5072
	struct nfs_lockt_res res = {
		.denied = request,
L
Linus Torvalds 已提交
5073 5074 5075 5076 5077 5078 5079 5080 5081 5082
	};
	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
5083
	arg.lock_owner.clientid = clp->cl_clientid;
5084 5085 5086 5087
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
	lsp = request->fl_u.nfs4_fl.owner;
5088
	arg.lock_owner.id = lsp->ls_seqid.owner_id;
5089
	arg.lock_owner.s_dev = server->s_dev;
5090
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
T
Trond Myklebust 已提交
5091 5092 5093 5094 5095 5096
	switch (status) {
		case 0:
			request->fl_type = F_UNLCK;
			break;
		case -NFS4ERR_DENIED:
			status = 0;
L
Linus Torvalds 已提交
5097
	}
5098
	request->fl_ops->fl_release_private(request);
5099
out:
L
Linus Torvalds 已提交
5100 5101 5102 5103 5104 5105 5106 5107 5108
	return status;
}

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

	do {
5109 5110 5111
		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
5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132
				&exception);
	} while (exception.retry);
	return err;
}

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

5133
struct nfs4_unlockdata {
T
Trond Myklebust 已提交
5134 5135
	struct nfs_locku_args arg;
	struct nfs_locku_res res;
5136 5137
	struct nfs4_lock_state *lsp;
	struct nfs_open_context *ctx;
T
Trond Myklebust 已提交
5138 5139
	struct file_lock fl;
	const struct nfs_server *server;
5140
	unsigned long timestamp;
5141 5142
};

T
Trond Myklebust 已提交
5143 5144 5145 5146 5147 5148 5149 5150
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;

5151
	p = kzalloc(sizeof(*p), GFP_NOFS);
T
Trond Myklebust 已提交
5152 5153 5154 5155 5156
	if (p == NULL)
		return NULL;
	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
	p->arg.seqid = seqid;
5157
	p->res.seqid = seqid;
T
Trond Myklebust 已提交
5158 5159 5160 5161 5162 5163 5164 5165 5166 5167
	p->arg.stateid = &lsp->ls_stateid;
	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;
}

5168
static void nfs4_locku_release_calldata(void *data)
5169
{
5170
	struct nfs4_unlockdata *calldata = data;
T
Trond Myklebust 已提交
5171
	nfs_free_seqid(calldata->arg.seqid);
5172 5173 5174
	nfs4_put_lock_state(calldata->lsp);
	put_nfs_open_context(calldata->ctx);
	kfree(calldata);
5175 5176
}

5177
static void nfs4_locku_done(struct rpc_task *task, void *data)
5178
{
5179
	struct nfs4_unlockdata *calldata = data;
5180

5181 5182
	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
		return;
5183 5184
	switch (task->tk_status) {
		case 0:
5185 5186
			nfs4_stateid_copy(&calldata->lsp->ls_stateid,
					&calldata->res.stateid);
5187
			renew_lease(calldata->server, calldata->timestamp);
5188
			break;
5189 5190
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OLD_STATEID:
5191 5192 5193 5194
		case -NFS4ERR_STALE_STATEID:
		case -NFS4ERR_EXPIRED:
			break;
		default:
5195
			if (nfs4_async_handle_error(task, calldata->server, NULL) == -EAGAIN)
5196
				rpc_restart_call_prepare(task);
5197
	}
5198
	nfs_release_seqid(calldata->arg.seqid);
5199 5200
}

T
Trond Myklebust 已提交
5201
static void nfs4_locku_prepare(struct rpc_task *task, void *data)
5202
{
T
Trond Myklebust 已提交
5203
	struct nfs4_unlockdata *calldata = data;
5204

T
Trond Myklebust 已提交
5205
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
5206
		goto out_wait;
5207
	if (test_bit(NFS_LOCK_INITIALIZED, &calldata->lsp->ls_flags) == 0) {
5208
		/* Note: exit _without_ running nfs4_locku_done */
5209
		goto out_no_action;
5210
	}
5211
	calldata->timestamp = jiffies;
5212
	if (nfs4_setup_sequence(calldata->server,
5213
				&calldata->arg.seq_args,
5214 5215 5216
				&calldata->res.seq_res,
				task) != 0)
		nfs_release_seqid(calldata->arg.seqid);
5217 5218 5219 5220 5221
	return;
out_no_action:
	task->tk_action = NULL;
out_wait:
	nfs4_sequence_done(task, &calldata->res.seq_res);
5222 5223
}

5224
static const struct rpc_call_ops nfs4_locku_ops = {
T
Trond Myklebust 已提交
5225
	.rpc_call_prepare = nfs4_locku_prepare,
5226
	.rpc_call_done = nfs4_locku_done,
5227
	.rpc_release = nfs4_locku_release_calldata,
5228 5229
};

5230 5231 5232 5233 5234 5235
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;
5236 5237 5238 5239
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCKU],
		.rpc_cred = ctx->cred,
	};
T
Trond Myklebust 已提交
5240 5241
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(lsp->ls_state->inode),
5242
		.rpc_message = &msg,
T
Trond Myklebust 已提交
5243
		.callback_ops = &nfs4_locku_ops,
5244
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
5245 5246
		.flags = RPC_TASK_ASYNC,
	};
5247

5248 5249 5250
	nfs4_state_protect(NFS_SERVER(lsp->ls_state->inode)->nfs_client,
		NFS_SP4_MACH_CRED_CLEANUP, &task_setup_data.rpc_client, &msg);

5251 5252 5253 5254 5255
	/* 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;

5256 5257 5258 5259 5260 5261
	data = nfs4_alloc_unlockdata(fl, ctx, lsp, seqid);
	if (data == NULL) {
		nfs_free_seqid(seqid);
		return ERR_PTR(-ENOMEM);
	}

5262
	nfs4_init_sequence(&data->arg.seq_args, &data->res.seq_res, 1);
5263 5264
	msg.rpc_argp = &data->arg;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
5265 5266
	task_setup_data.callback_data = data;
	return rpc_run_task(&task_setup_data);
5267 5268
}

5269 5270
static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
{
5271 5272 5273
	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
5274
	struct nfs_seqid *seqid;
L
Linus Torvalds 已提交
5275
	struct nfs4_lock_state *lsp;
5276 5277
	struct rpc_task *task;
	int status = 0;
5278
	unsigned char fl_flags = request->fl_flags;
5279

5280
	status = nfs4_set_lock_state(state, request);
5281 5282
	/* Unlock _before_ we do the RPC call */
	request->fl_flags |= FL_EXISTS;
5283 5284 5285
	/* Exclude nfs_delegation_claim_locks() */
	mutex_lock(&sp->so_delegreturn_mutex);
	/* Exclude nfs4_reclaim_open_stateid() - note nesting! */
5286 5287 5288
	down_read(&nfsi->rwsem);
	if (do_vfs_lock(request->fl_file, request) == -ENOENT) {
		up_read(&nfsi->rwsem);
5289
		mutex_unlock(&sp->so_delegreturn_mutex);
5290
		goto out;
5291 5292
	}
	up_read(&nfsi->rwsem);
5293
	mutex_unlock(&sp->so_delegreturn_mutex);
5294
	if (status != 0)
5295 5296
		goto out;
	/* Is this a delegated lock? */
5297
	lsp = request->fl_u.nfs4_fl.owner;
5298 5299
	if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) == 0)
		goto out;
5300
	seqid = nfs_alloc_seqid(&lsp->ls_seqid, GFP_KERNEL);
5301
	status = -ENOMEM;
T
Trond Myklebust 已提交
5302
	if (seqid == NULL)
5303
		goto out;
5304
	task = nfs4_do_unlck(request, nfs_file_open_context(request->fl_file), lsp, seqid);
5305 5306
	status = PTR_ERR(task);
	if (IS_ERR(task))
5307
		goto out;
5308
	status = nfs4_wait_for_completion_rpc_task(task);
5309
	rpc_put_task(task);
5310
out:
5311
	request->fl_flags = fl_flags;
5312
	trace_nfs4_unlock(request, state, F_SETLK, status);
L
Linus Torvalds 已提交
5313 5314 5315
	return status;
}

5316 5317 5318 5319 5320 5321
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;
5322
	unsigned long timestamp;
5323 5324
	int rpc_status;
	int cancelled;
5325
	struct nfs_server *server;
5326 5327 5328
};

static struct nfs4_lockdata *nfs4_alloc_lockdata(struct file_lock *fl,
5329 5330
		struct nfs_open_context *ctx, struct nfs4_lock_state *lsp,
		gfp_t gfp_mask)
L
Linus Torvalds 已提交
5331
{
5332 5333
	struct nfs4_lockdata *p;
	struct inode *inode = lsp->ls_state->inode;
L
Linus Torvalds 已提交
5334
	struct nfs_server *server = NFS_SERVER(inode);
5335

5336
	p = kzalloc(sizeof(*p), gfp_mask);
5337 5338 5339 5340 5341
	if (p == NULL)
		return NULL;

	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
5342
	p->arg.open_seqid = nfs_alloc_seqid(&lsp->ls_state->owner->so_seqid, gfp_mask);
5343 5344
	if (p->arg.open_seqid == NULL)
		goto out_free;
5345
	p->arg.lock_seqid = nfs_alloc_seqid(&lsp->ls_seqid, gfp_mask);
5346
	if (p->arg.lock_seqid == NULL)
5347
		goto out_free_seqid;
5348
	p->arg.lock_stateid = &lsp->ls_stateid;
5349
	p->arg.lock_owner.clientid = server->nfs_client->cl_clientid;
5350
	p->arg.lock_owner.id = lsp->ls_seqid.owner_id;
5351
	p->arg.lock_owner.s_dev = server->s_dev;
5352
	p->res.lock_seqid = p->arg.lock_seqid;
5353
	p->lsp = lsp;
5354
	p->server = server;
5355 5356 5357 5358
	atomic_inc(&lsp->ls_count);
	p->ctx = get_nfs_open_context(ctx);
	memcpy(&p->fl, fl, sizeof(p->fl));
	return p;
5359 5360
out_free_seqid:
	nfs_free_seqid(p->arg.open_seqid);
5361 5362 5363 5364 5365 5366 5367 5368 5369
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;
5370

5371
	dprintk("%s: begin!\n", __func__);
5372
	if (nfs_wait_on_sequence(data->arg.lock_seqid, task) != 0)
5373
		goto out_wait;
5374 5375
	/* Do we need to do an open_to_lock_owner? */
	if (!(data->arg.lock_seqid->sequence->flags & NFS_SEQID_CONFIRMED)) {
5376
		if (nfs_wait_on_sequence(data->arg.open_seqid, task) != 0) {
5377
			goto out_release_lock_seqid;
5378
		}
5379
		data->arg.open_stateid = &state->open_stateid;
5380
		data->arg.new_lock_owner = 1;
5381
		data->res.open_seqid = data->arg.open_seqid;
5382 5383
	} else
		data->arg.new_lock_owner = 0;
5384 5385 5386 5387 5388
	if (!nfs4_valid_open_stateid(state)) {
		data->rpc_status = -EBADF;
		task->tk_action = NULL;
		goto out_release_open_seqid;
	}
5389
	data->timestamp = jiffies;
5390 5391
	if (nfs4_setup_sequence(data->server,
				&data->arg.seq_args,
5392
				&data->res.seq_res,
5393
				task) == 0)
5394
		return;
5395
out_release_open_seqid:
5396 5397 5398
	nfs_release_seqid(data->arg.open_seqid);
out_release_lock_seqid:
	nfs_release_seqid(data->arg.lock_seqid);
5399 5400
out_wait:
	nfs4_sequence_done(task, &data->res.seq_res);
5401
	dprintk("%s: done!, ret = %d\n", __func__, data->rpc_status);
5402 5403
}

5404 5405 5406 5407
static void nfs4_lock_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_lockdata *data = calldata;

5408
	dprintk("%s: begin!\n", __func__);
5409

5410 5411
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return;
5412

5413 5414 5415 5416 5417 5418 5419 5420
	data->rpc_status = task->tk_status;
	if (data->arg.new_lock_owner != 0) {
		if (data->rpc_status == 0)
			nfs_confirm_seqid(&data->lsp->ls_seqid, 0);
		else
			goto out;
	}
	if (data->rpc_status == 0) {
5421
		nfs4_stateid_copy(&data->lsp->ls_stateid, &data->res.stateid);
5422
		set_bit(NFS_LOCK_INITIALIZED, &data->lsp->ls_flags);
5423
		renew_lease(NFS_SERVER(data->ctx->dentry->d_inode), data->timestamp);
5424 5425
	}
out:
5426
	dprintk("%s: done, ret = %d!\n", __func__, data->rpc_status);
5427 5428 5429 5430 5431 5432
}

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

5433
	dprintk("%s: begin!\n", __func__);
5434
	nfs_free_seqid(data->arg.open_seqid);
5435 5436 5437 5438 5439
	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))
5440
			rpc_put_task_async(task);
5441
		dprintk("%s: cancelling lock!\n", __func__);
5442 5443 5444 5445 5446
	} else
		nfs_free_seqid(data->arg.lock_seqid);
	nfs4_put_lock_state(data->lsp);
	put_nfs_open_context(data->ctx);
	kfree(data);
5447
	dprintk("%s: done!\n", __func__);
5448 5449 5450 5451 5452 5453 5454 5455
}

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,
};

5456 5457 5458 5459 5460
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:
5461
		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
5462
		if (new_lock_owner != 0 ||
5463
		   test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags) != 0)
5464
			nfs4_schedule_stateid_recovery(server, lsp->ls_state);
5465 5466 5467
		break;
	case -NFS4ERR_STALE_STATEID:
		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
5468 5469
	case -NFS4ERR_EXPIRED:
		nfs4_schedule_lease_recovery(server->nfs_client);
5470 5471 5472
	};
}

5473
static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *fl, int recovery_type)
5474 5475 5476
{
	struct nfs4_lockdata *data;
	struct rpc_task *task;
5477 5478 5479 5480
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCK],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
5481 5482
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(state->inode),
5483
		.rpc_message = &msg,
T
Trond Myklebust 已提交
5484
		.callback_ops = &nfs4_lock_ops,
5485
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
5486 5487
		.flags = RPC_TASK_ASYNC,
	};
5488 5489
	int ret;

5490
	dprintk("%s: begin!\n", __func__);
5491
	data = nfs4_alloc_lockdata(fl, nfs_file_open_context(fl->fl_file),
5492 5493
			fl->fl_u.nfs4_fl.owner,
			recovery_type == NFS_LOCK_NEW ? GFP_KERNEL : GFP_NOFS);
5494 5495 5496 5497
	if (data == NULL)
		return -ENOMEM;
	if (IS_SETLKW(cmd))
		data->arg.block = 1;
5498
	nfs4_init_sequence(&data->arg.seq_args, &data->res.seq_res, 1);
5499 5500
	msg.rpc_argp = &data->arg;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
5501
	task_setup_data.callback_data = data;
5502 5503 5504 5505 5506
	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);
	}
T
Trond Myklebust 已提交
5507
	task = rpc_run_task(&task_setup_data);
5508
	if (IS_ERR(task))
5509 5510 5511 5512
		return PTR_ERR(task);
	ret = nfs4_wait_for_completion_rpc_task(task);
	if (ret == 0) {
		ret = data->rpc_status;
5513 5514 5515
		if (ret)
			nfs4_handle_setlk_error(data->server, data->lsp,
					data->arg.new_lock_owner, ret);
5516 5517
	} else
		data->cancelled = 1;
5518
	rpc_put_task(task);
5519
	dprintk("%s: done, ret = %d!\n", __func__, ret);
5520
	return ret;
L
Linus Torvalds 已提交
5521 5522 5523 5524
}

static int nfs4_lock_reclaim(struct nfs4_state *state, struct file_lock *request)
{
5525
	struct nfs_server *server = NFS_SERVER(state->inode);
5526 5527 5528
	struct nfs4_exception exception = {
		.inode = state->inode,
	};
5529 5530 5531
	int err;

	do {
5532 5533 5534
		/* Cache the lock if possible... */
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
5535
		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_RECLAIM);
5536
		trace_nfs4_lock_reclaim(request, state, F_SETLK, err);
5537
		if (err != -NFS4ERR_DELAY)
5538 5539 5540 5541
			break;
		nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
	return err;
L
Linus Torvalds 已提交
5542 5543 5544 5545
}

static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request)
{
5546
	struct nfs_server *server = NFS_SERVER(state->inode);
5547 5548 5549
	struct nfs4_exception exception = {
		.inode = state->inode,
	};
5550 5551
	int err;

5552 5553 5554
	err = nfs4_set_lock_state(state, request);
	if (err != 0)
		return err;
5555
	if (!recover_lost_locks) {
5556 5557 5558
		set_bit(NFS_LOCK_LOST, &request->fl_u.nfs4_fl.owner->ls_flags);
		return 0;
	}
5559
	do {
5560 5561
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
5562
		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_EXPIRED);
5563
		trace_nfs4_lock_expired(request, state, F_SETLK, err);
5564 5565 5566 5567 5568 5569 5570 5571
		switch (err) {
		default:
			goto out;
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			nfs4_handle_exception(server, err, &exception);
			err = 0;
		}
5572
	} while (exception.retry);
5573
out:
5574
	return err;
L
Linus Torvalds 已提交
5575 5576
}

5577
#if defined(CONFIG_NFS_V4_1)
5578 5579 5580 5581 5582 5583 5584 5585
/**
 * 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.
 */
5586
static int nfs41_check_expired_locks(struct nfs4_state *state)
5587
{
5588
	int status, ret = -NFS4ERR_BAD_STATEID;
5589
	struct nfs4_lock_state *lsp;
5590 5591
	struct nfs_server *server = NFS_SERVER(state->inode);

5592
	list_for_each_entry(lsp, &state->lock_states, ls_locks) {
5593
		if (test_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags)) {
5594 5595 5596 5597 5598
			struct rpc_cred *cred = lsp->ls_state->owner->so_cred;

			status = nfs41_test_stateid(server,
					&lsp->ls_stateid,
					cred);
5599
			trace_nfs4_test_lock_stateid(state, lsp, status);
5600
			if (status != NFS_OK) {
5601 5602
				/* Free the stateid unless the server
				 * informs us the stateid is unrecognized. */
5603 5604
				if (status != -NFS4ERR_BAD_STATEID)
					nfs41_free_stateid(server,
5605 5606
							&lsp->ls_stateid,
							cred);
5607
				clear_bit(NFS_LOCK_INITIALIZED, &lsp->ls_flags);
5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621
				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);
5622 5623 5624
	if (status != NFS_OK)
		status = nfs4_lock_expired(state, request);
	return status;
5625 5626 5627
}
#endif

L
Linus Torvalds 已提交
5628 5629
static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
5630
	struct nfs4_state_owner *sp = state->owner;
5631
	struct nfs_inode *nfsi = NFS_I(state->inode);
5632
	unsigned char fl_flags = request->fl_flags;
5633
	unsigned int seq;
5634
	int status = -ENOLCK;
L
Linus Torvalds 已提交
5635

5636 5637 5638
	if ((fl_flags & FL_POSIX) &&
			!test_bit(NFS_STATE_POSIX_LOCKS, &state->flags))
		goto out;
5639 5640 5641 5642
	/* Is this a delegated open? */
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
5643 5644 5645 5646
	request->fl_flags |= FL_ACCESS;
	status = do_vfs_lock(request->fl_file, request);
	if (status < 0)
		goto out;
5647
	down_read(&nfsi->rwsem);
5648 5649 5650
	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
		/* Yes: cache locks! */
		/* ...but avoid races with delegation recall... */
5651 5652 5653
		request->fl_flags = fl_flags & ~FL_SLEEP;
		status = do_vfs_lock(request->fl_file, request);
		goto out_unlock;
5654
	}
5655 5656
	seq = raw_seqcount_begin(&sp->so_reclaim_seqcount);
	up_read(&nfsi->rwsem);
5657
	status = _nfs4_do_setlk(state, cmd, request, NFS_LOCK_NEW);
5658
	if (status != 0)
5659 5660 5661 5662
		goto out;
	down_read(&nfsi->rwsem);
	if (read_seqcount_retry(&sp->so_reclaim_seqcount, seq)) {
		status = -NFS4ERR_DELAY;
5663
		goto out_unlock;
5664
	}
5665
	/* Note: we always want to sleep here! */
5666
	request->fl_flags = fl_flags | FL_SLEEP;
5667
	if (do_vfs_lock(request->fl_file, request) < 0)
5668 5669
		printk(KERN_WARNING "NFS: %s: VFS is out of sync with lock "
			"manager!\n", __func__);
5670
out_unlock:
5671
	up_read(&nfsi->rwsem);
5672 5673
out:
	request->fl_flags = fl_flags;
L
Linus Torvalds 已提交
5674 5675 5676 5677 5678
	return status;
}

static int nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
5679 5680
	struct nfs4_exception exception = {
		.state = state,
5681
		.inode = state->inode,
5682
	};
L
Linus Torvalds 已提交
5683 5684 5685
	int err;

	do {
5686
		err = _nfs4_proc_setlk(state, cmd, request);
5687
		trace_nfs4_set_lock(request, state, cmd, err);
5688 5689
		if (err == -NFS4ERR_DENIED)
			err = -EAGAIN;
L
Linus Torvalds 已提交
5690
		err = nfs4_handle_exception(NFS_SERVER(state->inode),
5691
				err, &exception);
L
Linus Torvalds 已提交
5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704
	} 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 */
5705
	ctx = nfs_file_open_context(filp);
L
Linus Torvalds 已提交
5706 5707 5708 5709 5710
	state = ctx->state;

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

5711 5712 5713 5714 5715
	if (IS_GETLK(cmd)) {
		if (state != NULL)
			return nfs4_proc_getlk(state, F_GETLK, request);
		return 0;
	}
L
Linus Torvalds 已提交
5716 5717 5718 5719

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

5720 5721 5722 5723 5724
	if (request->fl_type == F_UNLCK) {
		if (state != NULL)
			return nfs4_proc_unlck(state, cmd, request);
		return 0;
	}
L
Linus Torvalds 已提交
5725

5726 5727
	if (state == NULL)
		return -ENOLCK;
5728 5729 5730 5731
	/*
	 * Don't rely on the VFS having checked the file open mode,
	 * since it won't do this for flock() locks.
	 */
5732
	switch (request->fl_type) {
5733 5734 5735 5736 5737 5738 5739 5740 5741
	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
5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753
	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;
}

5754
int nfs4_lock_delegation_recall(struct file_lock *fl, struct nfs4_state *state, const nfs4_stateid *stateid)
5755 5756 5757 5758 5759 5760
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	int err;

	err = nfs4_set_lock_state(state, fl);
	if (err != 0)
5761
		return err;
5762
	err = _nfs4_do_setlk(state, F_SETLK, fl, NFS_LOCK_NEW);
5763
	return nfs4_handle_delegation_recall_error(server, state, stateid, err);
5764
}
5765

5766 5767
struct nfs_release_lockowner_data {
	struct nfs4_lock_state *lsp;
5768
	struct nfs_server *server;
5769
	struct nfs_release_lockowner_args args;
5770 5771
	struct nfs4_sequence_args seq_args;
	struct nfs4_sequence_res seq_res;
5772 5773
};

5774 5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786
static void nfs4_release_lockowner_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs_release_lockowner_data *data = calldata;
	nfs40_setup_sequence(data->server,
				&data->seq_args, &data->seq_res, task);
}

static void nfs4_release_lockowner_done(struct rpc_task *task, void *calldata)
{
	struct nfs_release_lockowner_data *data = calldata;
	nfs40_sequence_done(task, &data->seq_res);
}

5787 5788
static void nfs4_release_lockowner_release(void *calldata)
{
5789
	struct nfs_release_lockowner_data *data = calldata;
5790
	nfs4_free_lock_state(data->server, data->lsp);
5791 5792 5793
	kfree(calldata);
}

5794
static const struct rpc_call_ops nfs4_release_lockowner_ops = {
5795 5796
	.rpc_call_prepare = nfs4_release_lockowner_prepare,
	.rpc_call_done = nfs4_release_lockowner_done,
5797 5798 5799
	.rpc_release = nfs4_release_lockowner_release,
};

5800
static int nfs4_release_lockowner(struct nfs_server *server, struct nfs4_lock_state *lsp)
5801
{
5802
	struct nfs_release_lockowner_data *data;
5803 5804 5805 5806 5807
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RELEASE_LOCKOWNER],
	};

	if (server->nfs_client->cl_mvops->minor_version != 0)
5808
		return -EINVAL;
5809

5810 5811 5812
	data = kmalloc(sizeof(*data), GFP_NOFS);
	if (!data)
		return -ENOMEM;
5813
	nfs4_init_sequence(&data->seq_args, &data->seq_res, 0);
5814
	data->lsp = lsp;
5815
	data->server = server;
5816 5817 5818
	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;
5819

5820 5821 5822
	msg.rpc_argp = &data->args;
	rpc_call_async(server->client, &msg, 0, &nfs4_release_lockowner_ops, data);
	return 0;
5823 5824
}

5825 5826
#define XATTR_NAME_NFSV4_ACL "system.nfs4_acl"

5827 5828 5829
static int nfs4_xattr_set_nfs4_acl(struct dentry *dentry, const char *key,
				   const void *buf, size_t buflen,
				   int flags, int type)
5830
{
5831 5832
	if (strcmp(key, "") != 0)
		return -EINVAL;
5833

5834
	return nfs4_proc_set_acl(dentry->d_inode, buf, buflen);
5835 5836
}

5837 5838
static int nfs4_xattr_get_nfs4_acl(struct dentry *dentry, const char *key,
				   void *buf, size_t buflen, int type)
5839
{
5840 5841
	if (strcmp(key, "") != 0)
		return -EINVAL;
5842

5843
	return nfs4_proc_get_acl(dentry->d_inode, buf, buflen);
5844 5845
}

5846 5847 5848
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)
5849
{
5850
	size_t len = sizeof(XATTR_NAME_NFSV4_ACL);
5851

5852 5853
	if (!nfs4_server_supports_acls(NFS_SERVER(dentry->d_inode)))
		return 0;
5854 5855 5856

	if (list && len <= list_len)
		memcpy(list, XATTR_NAME_NFSV4_ACL, len);
5857
	return len;
5858 5859
}

5860 5861 5862 5863 5864 5865 5866 5867 5868 5869 5870 5871 5872 5873 5874 5875 5876 5877 5878 5879 5880 5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895 5896 5897 5898 5899 5900 5901 5902 5903 5904 5905 5906
#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))
		return nfs4_get_security_label(dentry->d_inode, buf, buflen);
	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;

	if (nfs_server_capable(dentry->d_inode, NFS_CAP_SECURITY_LABEL)) {
		len = security_inode_listsecurity(dentry->d_inode, NULL, 0);
		if (list && len <= list_len)
			security_inode_listsecurity(dentry->d_inode, list, len);
	}
	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


5907 5908 5909
/*
 * nfs_fhget will use either the mounted_on_fileid or the fileid
 */
5910 5911
static void nfs_fixup_referral_attributes(struct nfs_fattr *fattr)
{
5912 5913 5914
	if (!(((fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID) ||
	       (fattr->valid & NFS_ATTR_FATTR_FILEID)) &&
	      (fattr->valid & NFS_ATTR_FATTR_FSID) &&
5915
	      (fattr->valid & NFS_ATTR_FATTR_V4_LOCATIONS)))
5916 5917 5918
		return;

	fattr->valid |= NFS_ATTR_FATTR_TYPE | NFS_ATTR_FATTR_MODE |
5919
		NFS_ATTR_FATTR_NLINK | NFS_ATTR_FATTR_V4_REFERRAL;
5920 5921 5922 5923
	fattr->mode = S_IFDIR | S_IRUGO | S_IXUGO;
	fattr->nlink = 2;
}

5924 5925 5926 5927
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)
5928 5929
{
	struct nfs_server *server = NFS_SERVER(dir);
5930
	u32 bitmask[3] = {
5931
		[0] = FATTR4_WORD0_FSID | FATTR4_WORD0_FS_LOCATIONS,
5932 5933 5934
	};
	struct nfs4_fs_locations_arg args = {
		.dir_fh = NFS_FH(dir),
5935
		.name = name,
5936 5937 5938
		.page = page,
		.bitmask = bitmask,
	};
B
Benny Halevy 已提交
5939 5940 5941
	struct nfs4_fs_locations_res res = {
		.fs_locations = fs_locations,
	};
5942 5943 5944
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FS_LOCATIONS],
		.rpc_argp = &args,
B
Benny Halevy 已提交
5945
		.rpc_resp = &res,
5946 5947 5948
	};
	int status;

5949
	dprintk("%s: start\n", __func__);
5950 5951 5952 5953 5954 5955 5956 5957

	/* 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;

5958
	nfs_fattr_init(&fs_locations->fattr);
5959
	fs_locations->server = server;
5960
	fs_locations->nlocations = 0;
5961
	status = nfs4_call_sync(client, server, &msg, &args.seq_args, &res.seq_res, 0);
5962
	dprintk("%s: returned status = %d\n", __func__, status);
5963 5964 5965
	return status;
}

5966 5967 5968 5969
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)
5970 5971 5972 5973
{
	struct nfs4_exception exception = { };
	int err;
	do {
5974 5975 5976 5977
		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,
5978 5979 5980 5981 5982
				&exception);
	} while (exception.retry);
	return err;
}

5983
/**
5984 5985 5986 5987 5988
 * 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.
5989
 */
5990
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
5991 5992 5993 5994 5995 5996 5997 5998 5999 6000 6001 6002 6003 6004
{
	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,
	};
6005 6006 6007 6008 6009 6010
	struct rpc_clnt *clnt = NFS_SERVER(dir)->client;

	if (use_integrity) {
		clnt = NFS_SERVER(dir)->nfs_client->cl_rpcclient;
		msg.rpc_cred = nfs4_get_clid_cred(NFS_SERVER(dir)->nfs_client);
	}
B
Bryan Schumaker 已提交
6011 6012

	dprintk("NFS call  secinfo %s\n", name->name);
6013 6014 6015 6016

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

6017 6018
	status = nfs4_call_sync(clnt, NFS_SERVER(dir), &msg, &args.seq_args,
				&res.seq_res, 0);
B
Bryan Schumaker 已提交
6019
	dprintk("NFS reply  secinfo: %d\n", status);
6020 6021 6022 6023

	if (msg.rpc_cred)
		put_rpccred(msg.rpc_cred);

B
Bryan Schumaker 已提交
6024 6025 6026
	return status;
}

6027 6028
int nfs4_proc_secinfo(struct inode *dir, const struct qstr *name,
		      struct nfs4_secinfo_flavors *flavors)
B
Bryan Schumaker 已提交
6029 6030 6031 6032
{
	struct nfs4_exception exception = { };
	int err;
	do {
6033 6034 6035 6036 6037 6038 6039 6040 6041 6042 6043 6044 6045 6046 6047
		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);

6048 6049
		trace_nfs4_secinfo(dir, name, err);
		err = nfs4_handle_exception(NFS_SERVER(dir), err,
B
Bryan Schumaker 已提交
6050 6051 6052 6053 6054
				&exception);
	} while (exception.retry);
	return err;
}

6055
#ifdef CONFIG_NFS_V4_1
6056 6057 6058 6059 6060 6061 6062 6063 6064 6065 6066 6067 6068 6069 6070 6071 6072 6073 6074
/*
 * 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;
}

6075
static bool
6076 6077
nfs41_same_server_scope(struct nfs41_server_scope *a,
			struct nfs41_server_scope *b)
6078 6079 6080 6081 6082 6083 6084 6085
{
	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;
}

6086 6087 6088 6089 6090 6091
/*
 * nfs4_proc_bind_conn_to_session()
 *
 * The 4.1 client currently uses the same TCP connection for the
 * fore and backchannel.
 */
6092
int nfs4_proc_bind_conn_to_session(struct nfs_client *clp, struct rpc_cred *cred)
6093 6094 6095 6096 6097 6098 6099 6100
{
	int status;
	struct nfs41_bind_conn_to_session_res res;
	struct rpc_message msg = {
		.rpc_proc =
			&nfs4_procedures[NFSPROC4_CLNT_BIND_CONN_TO_SESSION],
		.rpc_argp = clp,
		.rpc_resp = &res,
6101
		.rpc_cred = cred,
6102 6103 6104 6105 6106 6107 6108 6109 6110 6111 6112
	};

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

	res.session = kzalloc(sizeof(struct nfs4_session), GFP_NOFS);
	if (unlikely(res.session == NULL)) {
		status = -ENOMEM;
		goto out;
	}

	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
6113
	trace_nfs4_bind_conn_to_session(clp, status);
6114 6115 6116 6117 6118 6119 6120 6121 6122 6123 6124 6125 6126 6127 6128 6129 6130 6131 6132 6133 6134 6135 6136 6137 6138 6139 6140
	if (status == 0) {
		if (memcmp(res.session->sess_id.data,
		    clp->cl_session->sess_id.data, NFS4_MAX_SESSIONID_LEN)) {
			dprintk("NFS: %s: Session ID mismatch\n", __func__);
			status = -EIO;
			goto out_session;
		}
		if (res.dir != NFS4_CDFS4_BOTH) {
			dprintk("NFS: %s: Unexpected direction from server\n",
				__func__);
			status = -EIO;
			goto out_session;
		}
		if (res.use_conn_in_rdma_mode) {
			dprintk("NFS: %s: Server returned RDMA mode = true\n",
				__func__);
			status = -EIO;
			goto out_session;
		}
	}
out_session:
	kfree(res.session);
out:
	dprintk("<-- %s status= %d\n", __func__, status);
	return status;
}

B
Benny Halevy 已提交
6141
/*
6142 6143
 * 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
6144 6145 6146 6147 6148 6149 6150 6151 6152
 */
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)
6153 6154 6155
	},
	.allow.u.words = {
		[0] = 1 << (OP_CLOSE) |
6156 6157
		      1 << (OP_LOCKU) |
		      1 << (OP_COMMIT),
6158
		[1] = 1 << (OP_SECINFO - 32) |
6159 6160
		      1 << (OP_SECINFO_NO_NAME - 32) |
		      1 << (OP_TEST_STATEID - 32) |
6161 6162
		      1 << (OP_FREE_STATEID - 32) |
		      1 << (OP_WRITE - 32)
6163 6164 6165 6166 6167 6168
	}
};

/*
 * Select the state protection mode for client `clp' given the server results
 * from exchange_id in `sp'.
B
Benny Halevy 已提交
6169
 *
6170 6171 6172 6173 6174 6175 6176 6177 6178 6179 6180 6181 6182 6183 6184 6185 6186 6187 6188 6189 6190 6191 6192 6193 6194 6195 6196 6197 6198 6199 6200 6201 6202 6203 6204 6205 6206 6207 6208 6209 6210 6211 6212 6213 6214 6215 6216 6217 6218 6219 6220 6221
 * 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;
		}
6222 6223 6224 6225 6226 6227

		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);
		}
6228 6229 6230 6231 6232 6233

		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);
		}
6234 6235 6236 6237 6238 6239

		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);
		}
6240 6241 6242 6243 6244 6245 6246 6247 6248 6249

		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);
		}
6250 6251 6252 6253 6254 6255 6256
	}

	return 0;
}

/*
 * _nfs4_proc_exchange_id()
6257
 *
6258
 * Wrapper for EXCHANGE_ID operation.
B
Benny Halevy 已提交
6259
 */
6260 6261
static int _nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred,
	u32 sp4_how)
B
Benny Halevy 已提交
6262 6263 6264
{
	nfs4_verifier verifier;
	struct nfs41_exchange_id_args args = {
6265
		.verifier = &verifier,
B
Benny Halevy 已提交
6266
		.client = clp,
6267 6268
		.flags = EXCHGID4_FLAG_SUPP_MOVED_REFER |
			EXCHGID4_FLAG_BIND_PRINC_STATEID,
B
Benny Halevy 已提交
6269 6270
	};
	struct nfs41_exchange_id_res res = {
6271
		0
B
Benny Halevy 已提交
6272 6273 6274 6275 6276 6277 6278 6279 6280
	};
	int status;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_EXCHANGE_ID],
		.rpc_argp = &args,
		.rpc_resp = &res,
		.rpc_cred = cred,
	};

6281
	nfs4_init_boot_verifier(clp, &verifier);
6282 6283
	args.id_len = nfs4_init_uniform_client_string(clp, args.id,
							sizeof(args.id));
6284 6285 6286
	dprintk("NFS call  exchange_id auth=%s, '%.*s'\n",
		clp->cl_rpcclient->cl_auth->au_ops->au_name,
		args.id_len, args.id);
B
Benny Halevy 已提交
6287

6288
	res.server_owner = kzalloc(sizeof(struct nfs41_server_owner),
6289
					GFP_NOFS);
6290
	if (unlikely(res.server_owner == NULL)) {
6291 6292 6293
		status = -ENOMEM;
		goto out;
	}
6294

6295
	res.server_scope = kzalloc(sizeof(struct nfs41_server_scope),
6296
					GFP_NOFS);
6297
	if (unlikely(res.server_scope == NULL)) {
6298
		status = -ENOMEM;
6299
		goto out_server_owner;
6300
	}
6301

6302
	res.impl_id = kzalloc(sizeof(struct nfs41_impl_id), GFP_NOFS);
6303
	if (unlikely(res.impl_id == NULL)) {
6304 6305 6306 6307
		status = -ENOMEM;
		goto out_server_scope;
	}

6308 6309 6310 6311 6312 6313 6314 6315 6316 6317 6318 6319 6320 6321 6322 6323
	switch (sp4_how) {
	case SP4_NONE:
		args.state_protect.how = SP4_NONE;
		break;

	case SP4_MACH_CRED:
		args.state_protect = nfs4_sp4_mach_cred_request;
		break;

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

6324
	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
6325
	trace_nfs4_exchange_id(clp, status);
6326
	if (status == 0)
6327
		status = nfs4_check_cl_exchange_flags(res.flags);
6328

6329 6330 6331
	if (status == 0)
		status = nfs4_sp4_select_mode(clp, &res.state_protect);

6332
	if (status == 0) {
6333 6334 6335 6336 6337
		clp->cl_clientid = res.clientid;
		clp->cl_exchange_flags = (res.flags & ~EXCHGID4_FLAG_CONFIRMED_R);
		if (!(res.flags & EXCHGID4_FLAG_CONFIRMED_R))
			clp->cl_seqid = res.seqid;

6338 6339 6340
		kfree(clp->cl_serverowner);
		clp->cl_serverowner = res.server_owner;
		res.server_owner = NULL;
6341

6342
		/* use the most recent implementation id */
6343 6344
		kfree(clp->cl_implid);
		clp->cl_implid = res.impl_id;
6345

6346
		if (clp->cl_serverscope != NULL &&
6347
		    !nfs41_same_server_scope(clp->cl_serverscope,
6348 6349 6350 6351
					     res.server_scope)) {
			dprintk("%s: server_scope mismatch detected\n",
				__func__);
			set_bit(NFS4CLNT_SERVER_SCOPE_MISMATCH, &clp->cl_state);
6352 6353
			kfree(clp->cl_serverscope);
			clp->cl_serverscope = NULL;
6354 6355
		}

6356
		if (clp->cl_serverscope == NULL) {
6357
			clp->cl_serverscope = res.server_scope;
6358 6359
			goto out;
		}
6360 6361
	} else
		kfree(res.impl_id);
6362

6363 6364
out_server_owner:
	kfree(res.server_owner);
6365
out_server_scope:
6366 6367
	kfree(res.server_scope);
out:
6368
	if (clp->cl_implid != NULL)
6369
		dprintk("NFS reply exchange_id: Server Implementation ID: "
6370
			"domain: %s, name: %s, date: %llu,%u\n",
6371
			clp->cl_implid->domain, clp->cl_implid->name,
6372 6373
			clp->cl_implid->date.seconds,
			clp->cl_implid->date.nseconds);
6374
	dprintk("NFS reply exchange_id: %d\n", status);
B
Benny Halevy 已提交
6375 6376 6377
	return status;
}

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
/*
 * 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
6407 6408 6409 6410 6411 6412 6413 6414 6415 6416 6417
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);
6418
	trace_nfs4_destroy_clientid(clp, status);
T
Trond Myklebust 已提交
6419
	if (status)
6420
		dprintk("NFS: Got error %d from the server %s on "
T
Trond Myklebust 已提交
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
			"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;
6454 6455
	if (clp->cl_preserve_clid)
		goto out;
6456
	cred = nfs4_get_clid_cred(clp);
T
Trond Myklebust 已提交
6457 6458 6459 6460 6461 6462 6463 6464 6465 6466 6467 6468
	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
6469 6470 6471 6472 6473 6474 6475 6476 6477 6478 6479 6480 6481 6482 6483
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 */
6484 6485 6486 6487
	nfs41_setup_sequence(data->clp->cl_session,
			&data->args->la_seq_args,
			&data->res->lr_seq_res,
			task);
A
Andy Adamson 已提交
6488 6489 6490 6491 6492 6493 6494 6495 6496 6497 6498 6499 6500
	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__);
6501 6502
	if (!nfs41_sequence_done(task, &data->res->lr_seq_res))
		return;
A
Andy Adamson 已提交
6503 6504 6505 6506 6507 6508
	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;
6509 6510
		/* fall through */
	case -NFS4ERR_RETRY_UNCACHED_REP:
6511
		rpc_restart_call_prepare(task);
A
Andy Adamson 已提交
6512 6513 6514 6515 6516
		return;
	}
	dprintk("<-- %s\n", __func__);
}

6517
static const struct rpc_call_ops nfs4_get_lease_time_ops = {
A
Andy Adamson 已提交
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
	.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,
6543 6544
		.callback_data = &data,
		.flags = RPC_TASK_TIMEOUT,
A
Andy Adamson 已提交
6545 6546 6547
	};
	int status;

6548
	nfs4_init_sequence(&args.la_seq_args, &res.lr_seq_res, 0);
6549
	nfs4_set_sequence_privileged(&args.la_seq_args);
A
Andy Adamson 已提交
6550 6551 6552 6553 6554 6555 6556 6557 6558 6559 6560 6561 6562 6563
	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
6564 6565 6566 6567 6568 6569 6570 6571 6572 6573 6574
/*
 * 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)
{
6575 6576 6577 6578
	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
6579 6580

	/* Fore channel attributes */
6581 6582
	args->fc_attrs.max_rqst_sz = max_rqst_sz;
	args->fc_attrs.max_resp_sz = max_resp_sz;
A
Andy Adamson 已提交
6583
	args->fc_attrs.max_ops = NFS4_MAX_OPS;
6584
	args->fc_attrs.max_reqs = max_session_slots;
A
Andy Adamson 已提交
6585 6586

	dprintk("%s: Fore Channel : max_rqst_sz=%u max_resp_sz=%u "
6587
		"max_ops=%u max_reqs=%u\n",
A
Andy Adamson 已提交
6588 6589
		__func__,
		args->fc_attrs.max_rqst_sz, args->fc_attrs.max_resp_sz,
6590
		args->fc_attrs.max_ops, args->fc_attrs.max_reqs);
A
Andy Adamson 已提交
6591 6592 6593 6594 6595 6596 6597 6598 6599 6600 6601 6602 6603 6604 6605 6606

	/* 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);
}

6607
static int nfs4_verify_fore_channel_attrs(struct nfs41_create_session_args *args, struct nfs4_session *session)
6608
{
6609 6610 6611 6612 6613 6614 6615 6616 6617 6618 6619 6620 6621 6622 6623
	struct nfs4_channel_attrs *sent = &args->fc_attrs;
	struct nfs4_channel_attrs *rcvd = &session->fc_attrs;

	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;
6624 6625
	if (rcvd->max_reqs > NFS4_MAX_SLOT_TABLE)
		rcvd->max_reqs = NFS4_MAX_SLOT_TABLE;
6626
	return 0;
6627 6628
}

6629 6630 6631 6632
static int nfs4_verify_back_channel_attrs(struct nfs41_create_session_args *args, struct nfs4_session *session)
{
	struct nfs4_channel_attrs *sent = &args->bc_attrs;
	struct nfs4_channel_attrs *rcvd = &session->bc_attrs;
6633

6634 6635 6636 6637 6638 6639 6640
	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: */
6641
	if (rcvd->max_ops != sent->max_ops)
6642
		return -EINVAL;
6643
	if (rcvd->max_reqs != sent->max_reqs)
6644 6645 6646
		return -EINVAL;
	return 0;
}
6647 6648 6649 6650

static int nfs4_verify_channel_attrs(struct nfs41_create_session_args *args,
				     struct nfs4_session *session)
{
6651
	int ret;
6652

6653 6654 6655 6656
	ret = nfs4_verify_fore_channel_attrs(args, session);
	if (ret)
		return ret;
	return nfs4_verify_back_channel_attrs(args, session);
6657 6658
}

6659 6660
static int _nfs4_proc_create_session(struct nfs_client *clp,
		struct rpc_cred *cred)
A
Andy Adamson 已提交
6661 6662 6663 6664 6665 6666 6667 6668 6669 6670 6671 6672 6673
{
	struct nfs4_session *session = clp->cl_session;
	struct nfs41_create_session_args args = {
		.client = clp,
		.cb_program = NFS4_CALLBACK,
	};
	struct nfs41_create_session_res res = {
		.client = clp,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CREATE_SESSION],
		.rpc_argp = &args,
		.rpc_resp = &res,
6674
		.rpc_cred = cred,
A
Andy Adamson 已提交
6675 6676 6677 6678
	};
	int status;

	nfs4_init_channel_attrs(&args);
6679
	args.flags = (SESSION4_PERSIST | SESSION4_BACK_CHAN);
A
Andy Adamson 已提交
6680

6681
	status = rpc_call_sync(session->clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
6682
	trace_nfs4_create_session(clp, status);
A
Andy Adamson 已提交
6683

6684
	if (!status) {
6685 6686
		/* Verify the session's negotiated channel_attrs values */
		status = nfs4_verify_channel_attrs(&args, session);
A
Andy Adamson 已提交
6687 6688 6689 6690 6691 6692 6693 6694 6695 6696 6697 6698
		/* Increment the clientid slot sequence id */
		clp->cl_seqid++;
	}

	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.
 */
6699
int nfs4_proc_create_session(struct nfs_client *clp, struct rpc_cred *cred)
A
Andy Adamson 已提交
6700 6701 6702 6703 6704 6705 6706
{
	int status;
	unsigned *ptr;
	struct nfs4_session *session = clp->cl_session;

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

6707
	status = _nfs4_proc_create_session(clp, cred);
A
Andy Adamson 已提交
6708 6709 6710
	if (status)
		goto out;

6711 6712 6713
	/* 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
6714 6715 6716 6717 6718 6719 6720 6721 6722 6723 6724
	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
6725 6726 6727 6728
/*
 * Issue the over-the-wire RPC DESTROY_SESSION.
 * The caller must serialize access to this routine.
 */
6729 6730
int nfs4_proc_destroy_session(struct nfs4_session *session,
		struct rpc_cred *cred)
A
Andy Adamson 已提交
6731
{
6732 6733 6734 6735 6736
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DESTROY_SESSION],
		.rpc_argp = session,
		.rpc_cred = cred,
	};
A
Andy Adamson 已提交
6737 6738 6739 6740 6741 6742 6743 6744
	int status = 0;

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

	/* session is still being setup */
	if (session->clp->cl_cons_state != NFS_CS_READY)
		return status;

6745
	status = rpc_call_sync(session->clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
6746
	trace_nfs4_destroy_session(session->clp, status);
A
Andy Adamson 已提交
6747 6748

	if (status)
6749
		dprintk("NFS: Got error %d from the server on DESTROY_SESSION. "
A
Andy Adamson 已提交
6750 6751 6752 6753 6754 6755
			"Session has been destroyed regardless...\n", status);

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

A
Andy Adamson 已提交
6756 6757 6758
/*
 * Renew the cl_session lease.
 */
6759 6760 6761 6762 6763 6764
struct nfs4_sequence_data {
	struct nfs_client *clp;
	struct nfs4_sequence_args args;
	struct nfs4_sequence_res res;
};

6765 6766
static void nfs41_sequence_release(void *data)
{
6767 6768
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
6769

6770 6771 6772
	if (atomic_read(&clp->cl_count) > 1)
		nfs4_schedule_state_renewal(clp);
	nfs_put_client(clp);
6773
	kfree(calldata);
6774 6775
}

6776 6777 6778 6779 6780 6781 6782
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:
6783
		nfs4_schedule_lease_recovery(clp);
6784 6785 6786 6787
	}
	return 0;
}

6788
static void nfs41_sequence_call_done(struct rpc_task *task, void *data)
A
Andy Adamson 已提交
6789
{
6790 6791
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
A
Andy Adamson 已提交
6792

6793 6794
	if (!nfs41_sequence_done(task, task->tk_msg.rpc_resp))
		return;
A
Andy Adamson 已提交
6795

6796
	trace_nfs4_sequence(clp, task->tk_status);
A
Andy Adamson 已提交
6797 6798
	if (task->tk_status < 0) {
		dprintk("%s ERROR %d\n", __func__, task->tk_status);
6799 6800
		if (atomic_read(&clp->cl_count) == 1)
			goto out;
A
Andy Adamson 已提交
6801

6802 6803
		if (nfs41_sequence_handle_errors(task, clp) == -EAGAIN) {
			rpc_restart_call_prepare(task);
A
Andy Adamson 已提交
6804 6805 6806 6807
			return;
		}
	}
	dprintk("%s rpc_cred %p\n", __func__, task->tk_msg.rpc_cred);
6808
out:
A
Andy Adamson 已提交
6809 6810 6811 6812 6813
	dprintk("<-- %s\n", __func__);
}

static void nfs41_sequence_prepare(struct rpc_task *task, void *data)
{
6814 6815
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
A
Andy Adamson 已提交
6816 6817 6818 6819 6820 6821
	struct nfs4_sequence_args *args;
	struct nfs4_sequence_res *res;

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

6822
	nfs41_setup_sequence(clp->cl_session, args, res, task);
A
Andy Adamson 已提交
6823 6824 6825 6826 6827
}

static const struct rpc_call_ops nfs41_sequence_ops = {
	.rpc_call_done = nfs41_sequence_call_done,
	.rpc_call_prepare = nfs41_sequence_prepare,
6828
	.rpc_release = nfs41_sequence_release,
A
Andy Adamson 已提交
6829 6830
};

6831 6832 6833
static struct rpc_task *_nfs41_proc_sequence(struct nfs_client *clp,
		struct rpc_cred *cred,
		bool is_privileged)
A
Andy Adamson 已提交
6834
{
6835
	struct nfs4_sequence_data *calldata;
A
Andy Adamson 已提交
6836 6837 6838 6839
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SEQUENCE],
		.rpc_cred = cred,
	};
6840 6841 6842
	struct rpc_task_setup task_setup_data = {
		.rpc_client = clp->cl_rpcclient,
		.rpc_message = &msg,
6843
		.callback_ops = &nfs41_sequence_ops,
6844
		.flags = RPC_TASK_ASYNC | RPC_TASK_TIMEOUT,
6845
	};
A
Andy Adamson 已提交
6846

6847
	if (!atomic_inc_not_zero(&clp->cl_count))
6848
		return ERR_PTR(-EIO);
6849
	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
6850
	if (calldata == NULL) {
6851
		nfs_put_client(clp);
6852
		return ERR_PTR(-ENOMEM);
A
Andy Adamson 已提交
6853
	}
6854
	nfs4_init_sequence(&calldata->args, &calldata->res, 0);
6855 6856
	if (is_privileged)
		nfs4_set_sequence_privileged(&calldata->args);
6857 6858 6859
	msg.rpc_argp = &calldata->args;
	msg.rpc_resp = &calldata->res;
	calldata->clp = clp;
6860
	task_setup_data.callback_data = calldata;
A
Andy Adamson 已提交
6861

6862 6863 6864
	return rpc_run_task(&task_setup_data);
}

6865
static int nfs41_proc_async_sequence(struct nfs_client *clp, struct rpc_cred *cred, unsigned renew_flags)
6866 6867 6868 6869
{
	struct rpc_task *task;
	int ret = 0;

6870 6871
	if ((renew_flags & NFS4_RENEW_TIMEOUT) == 0)
		return 0;
6872
	task = _nfs41_proc_sequence(clp, cred, false);
6873 6874 6875
	if (IS_ERR(task))
		ret = PTR_ERR(task);
	else
6876
		rpc_put_task_async(task);
6877 6878 6879 6880 6881 6882 6883 6884 6885
	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;

6886
	task = _nfs41_proc_sequence(clp, cred, true);
6887 6888 6889 6890 6891
	if (IS_ERR(task)) {
		ret = PTR_ERR(task);
		goto out;
	}
	ret = rpc_wait_for_completion_task(task);
6892 6893 6894 6895 6896
	if (!ret) {
		struct nfs4_sequence_res *res = task->tk_msg.rpc_resp;

		if (task->tk_status == 0)
			nfs41_handle_sequence_flag_errors(clp, res->sr_status_flags);
6897
		ret = task->tk_status;
6898
	}
6899 6900 6901 6902
	rpc_put_task(task);
out:
	dprintk("<-- %s status=%d\n", __func__, ret);
	return ret;
A
Andy Adamson 已提交
6903 6904
}

6905 6906 6907 6908 6909 6910 6911 6912 6913 6914
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;

6915 6916 6917 6918
	nfs41_setup_sequence(calldata->clp->cl_session,
			&calldata->arg.seq_args,
			&calldata->res.seq_res,
			task);
6919 6920
}

6921 6922 6923 6924 6925 6926 6927 6928 6929
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);
6930 6931
		/* fall through */
	case -NFS4ERR_RETRY_UNCACHED_REP:
6932 6933
		return -EAGAIN;
	default:
6934
		nfs4_schedule_lease_recovery(clp);
6935 6936 6937 6938
	}
	return 0;
}

6939 6940 6941 6942 6943 6944 6945
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__);
6946 6947
	if (!nfs41_sequence_done(task, res))
		return;
6948

6949
	trace_nfs4_reclaim_complete(clp, task->tk_status);
6950 6951 6952 6953
	if (nfs41_reclaim_complete_handle_errors(task, clp) == -EAGAIN) {
		rpc_restart_call_prepare(task);
		return;
	}
6954 6955 6956 6957 6958 6959 6960 6961 6962 6963 6964 6965 6966 6967 6968 6969 6970 6971 6972
	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.
 */
6973 6974
static int nfs41_proc_reclaim_complete(struct nfs_client *clp,
		struct rpc_cred *cred)
6975 6976 6977 6978 6979
{
	struct nfs4_reclaim_complete_data *calldata;
	struct rpc_task *task;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RECLAIM_COMPLETE],
6980
		.rpc_cred = cred,
6981 6982 6983 6984 6985 6986 6987 6988 6989 6990
	};
	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__);
6991
	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
6992 6993 6994 6995 6996
	if (calldata == NULL)
		goto out;
	calldata->clp = clp;
	calldata->arg.one_fs = 0;

6997
	nfs4_init_sequence(&calldata->arg.seq_args, &calldata->res.seq_res, 0);
6998
	nfs4_set_sequence_privileged(&calldata->arg.seq_args);
6999 7000 7001 7002
	msg.rpc_argp = &calldata->arg;
	msg.rpc_resp = &calldata->res;
	task_setup_data.callback_data = calldata;
	task = rpc_run_task(&task_setup_data);
7003
	if (IS_ERR(task)) {
7004
		status = PTR_ERR(task);
7005 7006
		goto out;
	}
7007 7008 7009
	status = nfs4_wait_for_completion_rpc_task(task);
	if (status == 0)
		status = task->tk_status;
7010
	rpc_put_task(task);
7011
	return 0;
7012 7013 7014 7015
out:
	dprintk("<-- %s status=%d\n", __func__, status);
	return status;
}
7016 7017 7018 7019 7020

static void
nfs4_layoutget_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;
F
Fred Isaman 已提交
7021
	struct nfs_server *server = NFS_SERVER(lgp->args.inode);
7022
	struct nfs4_session *session = nfs4_get_session(server);
7023 7024

	dprintk("--> %s\n", __func__);
F
Fred Isaman 已提交
7025 7026 7027 7028 7029
	/* 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.
	 */
7030
	if (nfs41_setup_sequence(session, &lgp->args.seq_args,
7031
				&lgp->res.seq_res, task))
7032
		return;
7033 7034 7035 7036 7037
	if (pnfs_choose_layoutget_stateid(&lgp->args.stateid,
					  NFS_I(lgp->args.inode)->layout,
					  lgp->args.ctx->state)) {
		rpc_exit(task, NFS4_OK);
	}
7038 7039 7040 7041 7042
}

static void nfs4_layoutget_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;
7043 7044 7045 7046
	struct inode *inode = lgp->args.inode;
	struct nfs_server *server = NFS_SERVER(inode);
	struct pnfs_layout_hdr *lo;
	struct nfs4_state *state = NULL;
7047
	unsigned long timeo, giveup;
7048 7049 7050

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

7051
	if (!nfs41_sequence_done(task, &lgp->res.seq_res))
7052
		goto out;
7053 7054 7055

	switch (task->tk_status) {
	case 0:
7056
		goto out;
7057 7058
	case -NFS4ERR_LAYOUTTRYLATER:
	case -NFS4ERR_RECALLCONFLICT:
7059 7060 7061 7062
		timeo = rpc_get_timeout(task->tk_client);
		giveup = lgp->args.timestamp + timeo;
		if (time_after(giveup, jiffies))
			task->tk_status = -NFS4ERR_DELAY;
7063 7064 7065 7066 7067 7068 7069 7070 7071 7072 7073 7074 7075 7076 7077 7078 7079
		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);

			pnfs_mark_matching_lsegs_invalid(lo, &head, NULL);
			spin_unlock(&inode->i_lock);
			/* Mark the bad layout state as invalid, then
			 * retry using the open stateid. */
			pnfs_free_lseg_list(&head);
7080 7081
		}
	}
7082 7083 7084
	if (nfs4_async_handle_error(task, server, state) == -EAGAIN)
		rpc_restart_call_prepare(task);
out:
7085 7086 7087
	dprintk("<-- %s\n", __func__);
}

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 7121 7122 7123 7124 7125 7126 7127 7128 7129 7130 7131
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;
}

7132 7133 7134
static void nfs4_layoutget_release(void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;
7135 7136
	struct inode *inode = lgp->args.inode;
	struct nfs_server *server = NFS_SERVER(inode);
7137
	size_t max_pages = max_response_pages(server);
7138 7139

	dprintk("--> %s\n", __func__);
7140
	nfs4_free_pages(lgp->args.layout.pages, max_pages);
7141
	pnfs_put_layout_hdr(NFS_I(inode)->layout);
7142 7143 7144 7145 7146 7147 7148 7149 7150 7151 7152
	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,
};

7153 7154
struct pnfs_layout_segment *
nfs4_proc_layoutget(struct nfs4_layoutget *lgp, gfp_t gfp_flags)
7155
{
7156 7157
	struct inode *inode = lgp->args.inode;
	struct nfs_server *server = NFS_SERVER(inode);
7158
	size_t max_pages = max_response_pages(server);
7159 7160 7161 7162 7163
	struct rpc_task *task;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LAYOUTGET],
		.rpc_argp = &lgp->args,
		.rpc_resp = &lgp->res,
7164
		.rpc_cred = lgp->cred,
7165 7166 7167 7168 7169 7170 7171 7172
	};
	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,
	};
7173
	struct pnfs_layout_segment *lseg = NULL;
7174 7175 7176 7177
	int status = 0;

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

7178 7179 7180
	lgp->args.layout.pages = nfs4_alloc_pages(max_pages, gfp_flags);
	if (!lgp->args.layout.pages) {
		nfs4_layoutget_release(lgp);
7181
		return ERR_PTR(-ENOMEM);
7182 7183
	}
	lgp->args.layout.pglen = max_pages * PAGE_SIZE;
7184
	lgp->args.timestamp = jiffies;
7185

7186
	lgp->res.layoutp = &lgp->args.layout;
7187
	lgp->res.seq_res.sr_slot = NULL;
7188
	nfs4_init_sequence(&lgp->args.seq_args, &lgp->res.seq_res, 0);
7189 7190 7191 7192

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

7193 7194
	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task))
7195
		return ERR_CAST(task);
7196
	status = nfs4_wait_for_completion_rpc_task(task);
F
Fred Isaman 已提交
7197 7198
	if (status == 0)
		status = task->tk_status;
7199 7200 7201 7202
	trace_nfs4_layoutget(lgp->args.ctx,
			&lgp->args.range,
			&lgp->res.range,
			status);
7203 7204
	/* if layoutp->len is 0, nfs4_layoutget_prepare called rpc_exit */
	if (status == 0 && lgp->res.layoutp->len)
7205
		lseg = pnfs_layout_process(lgp);
7206 7207
	rpc_put_task(task);
	dprintk("<-- %s status=%d\n", __func__, status);
7208 7209 7210
	if (status)
		return ERR_PTR(status);
	return lseg;
7211 7212
}

B
Benny Halevy 已提交
7213 7214 7215 7216 7217 7218
static void
nfs4_layoutreturn_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutreturn *lrp = calldata;

	dprintk("--> %s\n", __func__);
7219 7220 7221 7222
	nfs41_setup_sequence(lrp->clp->cl_session,
			&lrp->args.seq_args,
			&lrp->res.seq_res,
			task);
B
Benny Halevy 已提交
7223 7224 7225 7226 7227 7228 7229 7230 7231
}

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

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

7232
	if (!nfs41_sequence_done(task, &lrp->res.seq_res))
B
Benny Halevy 已提交
7233 7234 7235 7236
		return;

	server = NFS_SERVER(lrp->args.inode);
	if (nfs4_async_handle_error(task, server, NULL) == -EAGAIN) {
7237
		rpc_restart_call_prepare(task);
B
Benny Halevy 已提交
7238 7239 7240 7241 7242 7243 7244 7245
		return;
	}
	dprintk("<-- %s\n", __func__);
}

static void nfs4_layoutreturn_release(void *calldata)
{
	struct nfs4_layoutreturn *lrp = calldata;
7246
	struct pnfs_layout_hdr *lo = lrp->args.layout;
B
Benny Halevy 已提交
7247 7248

	dprintk("--> %s\n", __func__);
7249 7250 7251 7252 7253
	spin_lock(&lo->plh_inode->i_lock);
	if (lrp->res.lrs_present)
		pnfs_set_layout_stateid(lo, &lrp->res.stateid, true);
	lo->plh_block_lgets--;
	spin_unlock(&lo->plh_inode->i_lock);
7254
	pnfs_put_layout_hdr(lrp->args.layout);
B
Benny Halevy 已提交
7255 7256 7257 7258 7259 7260 7261 7262 7263 7264 7265 7266 7267 7268 7269 7270 7271
	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,
};

int nfs4_proc_layoutreturn(struct nfs4_layoutreturn *lrp)
{
	struct rpc_task *task;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LAYOUTRETURN],
		.rpc_argp = &lrp->args,
		.rpc_resp = &lrp->res,
7272
		.rpc_cred = lrp->cred,
B
Benny Halevy 已提交
7273 7274
	};
	struct rpc_task_setup task_setup_data = {
7275
		.rpc_client = NFS_SERVER(lrp->args.inode)->client,
B
Benny Halevy 已提交
7276 7277 7278 7279 7280 7281 7282
		.rpc_message = &msg,
		.callback_ops = &nfs4_layoutreturn_call_ops,
		.callback_data = lrp,
	};
	int status;

	dprintk("--> %s\n", __func__);
7283
	nfs4_init_sequence(&lrp->args.seq_args, &lrp->res.seq_res, 1);
B
Benny Halevy 已提交
7284 7285 7286 7287
	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task))
		return PTR_ERR(task);
	status = task->tk_status;
7288
	trace_nfs4_layoutreturn(lrp->args.inode, status);
B
Benny Halevy 已提交
7289 7290 7291 7292 7293
	dprintk("<-- %s status=%d\n", __func__, status);
	rpc_put_task(task);
	return status;
}

A
Andy Adamson 已提交
7294 7295 7296 7297 7298 7299 7300 7301 7302 7303 7304 7305 7306 7307 7308 7309 7310 7311 7312 7313 7314 7315 7316 7317 7318 7319 7320 7321 7322 7323 7324 7325 7326 7327 7328 7329 7330 7331 7332 7333 7334 7335 7336 7337 7338 7339 7340 7341
/*
 * Retrieve the list of Data Server devices from the MDS.
 */
static int _nfs4_getdevicelist(struct nfs_server *server,
				    const struct nfs_fh *fh,
				    struct pnfs_devicelist *devlist)
{
	struct nfs4_getdevicelist_args args = {
		.fh = fh,
		.layoutclass = server->pnfs_curr_ld->id,
	};
	struct nfs4_getdevicelist_res res = {
		.devlist = devlist,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETDEVICELIST],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
	int status;

	dprintk("--> %s\n", __func__);
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args,
				&res.seq_res, 0);
	dprintk("<-- %s status=%d\n", __func__, status);
	return status;
}

int nfs4_proc_getdevicelist(struct nfs_server *server,
			    const struct nfs_fh *fh,
			    struct pnfs_devicelist *devlist)
{
	struct nfs4_exception exception = { };
	int err;

	do {
		err = nfs4_handle_exception(server,
				_nfs4_getdevicelist(server, fh, devlist),
				&exception);
	} while (exception.retry);

	dprintk("%s: err=%d, num_devs=%u\n", __func__,
		err, devlist->num_devs);

	return err;
}
EXPORT_SYMBOL_GPL(nfs4_proc_getdevicelist);

7342
static int
7343 7344 7345
_nfs4_proc_getdeviceinfo(struct nfs_server *server,
		struct pnfs_device *pdev,
		struct rpc_cred *cred)
7346 7347 7348 7349 7350 7351 7352 7353 7354 7355 7356
{
	struct nfs4_getdeviceinfo_args args = {
		.pdev = pdev,
	};
	struct nfs4_getdeviceinfo_res res = {
		.pdev = pdev,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETDEVICEINFO],
		.rpc_argp = &args,
		.rpc_resp = &res,
7357
		.rpc_cred = cred,
7358 7359 7360 7361
	};
	int status;

	dprintk("--> %s\n", __func__);
7362
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
7363 7364 7365 7366 7367
	dprintk("<-- %s status=%d\n", __func__, status);

	return status;
}

7368 7369 7370
int nfs4_proc_getdeviceinfo(struct nfs_server *server,
		struct pnfs_device *pdev,
		struct rpc_cred *cred)
7371 7372 7373 7374 7375 7376
{
	struct nfs4_exception exception = { };
	int err;

	do {
		err = nfs4_handle_exception(server,
7377
					_nfs4_proc_getdeviceinfo(server, pdev, cred),
7378 7379 7380 7381 7382 7383
					&exception);
	} while (exception.retry);
	return err;
}
EXPORT_SYMBOL_GPL(nfs4_proc_getdeviceinfo);

A
Andy Adamson 已提交
7384 7385 7386 7387
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);
7388
	struct nfs4_session *session = nfs4_get_session(server);
A
Andy Adamson 已提交
7389

7390 7391 7392 7393
	nfs41_setup_sequence(session,
			&data->args.seq_args,
			&data->res.seq_res,
			task);
A
Andy Adamson 已提交
7394 7395 7396 7397 7398 7399 7400 7401
}

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);

7402
	if (!nfs41_sequence_done(task, &data->res.seq_res))
A
Andy Adamson 已提交
7403 7404 7405
		return;

	switch (task->tk_status) { /* Just ignore these failures */
7406 7407 7408 7409
	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
7410
		task->tk_status = 0;
7411 7412
		break;
	case 0:
A
Andy Adamson 已提交
7413 7414
		nfs_post_op_update_inode_force_wcc(data->args.inode,
						   data->res.fattr);
7415 7416 7417 7418 7419 7420 7421
		break;
	default:
		if (nfs4_async_handle_error(task, server, NULL) == -EAGAIN) {
			rpc_restart_call_prepare(task);
			return;
		}
	}
A
Andy Adamson 已提交
7422 7423 7424 7425 7426 7427
}

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

A
Andy Adamson 已提交
7428
	pnfs_cleanup_layoutcommit(data);
A
Andy Adamson 已提交
7429 7430 7431 7432 7433 7434 7435 7436 7437 7438 7439
	put_rpccred(data->cred);
	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
7440
nfs4_proc_layoutcommit(struct nfs4_layoutcommit_data *data, bool sync)
A
Andy Adamson 已提交
7441 7442 7443 7444 7445 7446 7447 7448 7449 7450 7451 7452 7453 7454 7455 7456 7457 7458 7459 7460 7461 7462 7463 7464
{
	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,
		.flags = RPC_TASK_ASYNC,
	};
	struct rpc_task *task;
	int status = 0;

	dprintk("NFS: %4d initiating layoutcommit call. sync %d "
		"lbw: %llu inode %lu\n",
		data->task.tk_pid, sync,
		data->args.lastbytewritten,
		data->args.inode->i_ino);

7465
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
A
Andy Adamson 已提交
7466 7467 7468
	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task))
		return PTR_ERR(task);
7469
	if (sync == false)
A
Andy Adamson 已提交
7470 7471 7472 7473 7474
		goto out;
	status = nfs4_wait_for_completion_rpc_task(task);
	if (status != 0)
		goto out;
	status = task->tk_status;
7475
	trace_nfs4_layoutcommit(data->args.inode, status);
A
Andy Adamson 已提交
7476 7477 7478 7479 7480
out:
	dprintk("%s: status %d\n", __func__, status);
	rpc_put_task(task);
	return status;
}
7481

7482 7483 7484 7485
/**
 * Use the state managment nfs_client cl_rpcclient, which uses krb5i (if
 * possible) as per RFC3530bis and RFC5661 Security Considerations sections
 */
7486 7487
static int
_nfs41_proc_secinfo_no_name(struct nfs_server *server, struct nfs_fh *fhandle,
7488 7489
		    struct nfs_fsinfo *info,
		    struct nfs4_secinfo_flavors *flavors, bool use_integrity)
7490 7491 7492 7493 7494 7495 7496 7497 7498 7499 7500 7501
{
	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,
	};
7502 7503 7504 7505 7506 7507 7508 7509 7510 7511 7512 7513 7514 7515 7516 7517 7518
	struct rpc_clnt *clnt = server->client;
	int status;

	if (use_integrity) {
		clnt = server->nfs_client->cl_rpcclient;
		msg.rpc_cred = nfs4_get_clid_cred(server->nfs_client);
	}

	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);

	if (msg.rpc_cred)
		put_rpccred(msg.rpc_cred);

	return status;
7519 7520 7521 7522 7523 7524 7525 7526 7527
}

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 {
7528 7529 7530 7531 7532 7533 7534 7535 7536 7537 7538 7539 7540 7541 7542 7543 7544 7545
		/* 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);

7546 7547 7548 7549
		switch (err) {
		case 0:
		case -NFS4ERR_WRONGSEC:
		case -NFS4ERR_NOTSUPP:
7550
			goto out;
7551 7552 7553 7554
		default:
			err = nfs4_handle_exception(server, err, &exception);
		}
	} while (exception.retry);
7555
out:
7556 7557 7558 7559 7560 7561 7562 7563 7564 7565 7566 7567 7568 7569 7570 7571 7572 7573 7574 7575 7576 7577 7578 7579 7580 7581 7582 7583 7584 7585 7586 7587 7588 7589 7590 7591 7592 7593 7594 7595 7596 7597 7598
	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;
	rpc_authflavor_t flavor;
	struct nfs4_secinfo_flavors *flavors;

	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
	 */
	if (err == -NFS4ERR_WRONGSEC || err == -NFS4ERR_NOTSUPP) {
		err = nfs4_find_root_sec(server, fhandle, info);
		goto out_freepage;
	}
	if (err)
		goto out_freepage;

	flavor = nfs_find_best_sec(flavors);
	if (err == 0)
		err = nfs4_lookup_root_sec(server, fhandle, info, flavor);

out_freepage:
	put_page(page);
	if (err == -EACCES)
		return -EPERM;
out:
	return err;
}
7599

7600 7601 7602
static int _nfs41_test_stateid(struct nfs_server *server,
		nfs4_stateid *stateid,
		struct rpc_cred *cred)
B
Bryan Schumaker 已提交
7603 7604 7605
{
	int status;
	struct nfs41_test_stateid_args args = {
7606
		.stateid = stateid,
B
Bryan Schumaker 已提交
7607 7608 7609 7610 7611 7612
	};
	struct nfs41_test_stateid_res res;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_TEST_STATEID],
		.rpc_argp = &args,
		.rpc_resp = &res,
7613
		.rpc_cred = cred,
B
Bryan Schumaker 已提交
7614
	};
7615 7616 7617 7618
	struct rpc_clnt *rpc_client = server->client;

	nfs4_state_protect(server->nfs_client, NFS_SP4_MACH_CRED_STATEID,
		&rpc_client, &msg);
7619

7620
	dprintk("NFS call  test_stateid %p\n", stateid);
7621
	nfs4_init_sequence(&args.seq_args, &res.seq_res, 0);
7622
	nfs4_set_sequence_privileged(&args.seq_args);
7623
	status = nfs4_call_sync_sequence(rpc_client, server, &msg,
7624
			&args.seq_args, &res.seq_res);
7625 7626
	if (status != NFS_OK) {
		dprintk("NFS reply test_stateid: failed, %d\n", status);
7627
		return status;
7628 7629
	}
	dprintk("NFS reply test_stateid: succeeded, %d\n", -res.status);
7630
	return -res.status;
B
Bryan Schumaker 已提交
7631 7632
}

7633 7634 7635 7636 7637
/**
 * nfs41_test_stateid - perform a TEST_STATEID operation
 *
 * @server: server / transport on which to perform the operation
 * @stateid: state ID to test
7638
 * @cred: credential
7639 7640 7641 7642 7643
 *
 * 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.
 */
7644 7645 7646
static int nfs41_test_stateid(struct nfs_server *server,
		nfs4_stateid *stateid,
		struct rpc_cred *cred)
B
Bryan Schumaker 已提交
7647 7648 7649 7650
{
	struct nfs4_exception exception = { };
	int err;
	do {
7651
		err = _nfs41_test_stateid(server, stateid, cred);
7652 7653 7654
		if (err != -NFS4ERR_DELAY)
			break;
		nfs4_handle_exception(server, err, &exception);
B
Bryan Schumaker 已提交
7655 7656 7657
	} while (exception.retry);
	return err;
}
B
Bryan Schumaker 已提交
7658

7659 7660 7661
struct nfs_free_stateid_data {
	struct nfs_server *server;
	struct nfs41_free_stateid_args args;
B
Bryan Schumaker 已提交
7662
	struct nfs41_free_stateid_res res;
7663 7664 7665 7666 7667 7668 7669 7670 7671 7672 7673 7674 7675 7676 7677 7678 7679 7680 7681 7682 7683 7684 7685 7686 7687 7688 7689 7690 7691
};

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:
		if (nfs4_async_handle_error(task, data->server, NULL) == -EAGAIN)
			rpc_restart_call_prepare(task);
	}
}

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

7692
static const struct rpc_call_ops nfs41_free_stateid_ops = {
7693 7694 7695 7696 7697 7698 7699
	.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,
7700
		struct rpc_cred *cred,
7701 7702
		bool privileged)
{
B
Bryan Schumaker 已提交
7703 7704
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FREE_STATEID],
7705
		.rpc_cred = cred,
B
Bryan Schumaker 已提交
7706
	};
7707 7708 7709 7710 7711 7712 7713
	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
7714

7715 7716 7717
	nfs4_state_protect(server->nfs_client, NFS_SP4_MACH_CRED_STATEID,
		&task_setup.rpc_client, &msg);

7718
	dprintk("NFS call  free_stateid %p\n", stateid);
7719 7720 7721 7722 7723 7724 7725 7726 7727 7728
	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;
7729
	nfs4_init_sequence(&data->args.seq_args, &data->res.seq_res, 0);
7730 7731 7732 7733
	if (privileged)
		nfs4_set_sequence_privileged(&data->args.seq_args);

	return rpc_run_task(&task_setup);
B
Bryan Schumaker 已提交
7734 7735
}

7736 7737 7738 7739 7740
/**
 * nfs41_free_stateid - perform a FREE_STATEID operation
 *
 * @server: server / transport on which to perform the operation
 * @stateid: state ID to release
7741
 * @cred: credential
7742 7743 7744 7745
 *
 * Returns NFS_OK if the server freed "stateid".  Otherwise a
 * negative NFS4ERR value is returned.
 */
7746 7747 7748
static int nfs41_free_stateid(struct nfs_server *server,
		nfs4_stateid *stateid,
		struct rpc_cred *cred)
B
Bryan Schumaker 已提交
7749
{
7750 7751 7752
	struct rpc_task *task;
	int ret;

7753
	task = _nfs41_free_stateid(server, stateid, cred, true);
7754 7755 7756 7757 7758 7759 7760
	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
7761
}
7762

7763 7764 7765
static int nfs41_free_lock_state(struct nfs_server *server, struct nfs4_lock_state *lsp)
{
	struct rpc_task *task;
7766
	struct rpc_cred *cred = lsp->ls_state->owner->so_cred;
7767

7768
	task = _nfs41_free_stateid(server, &lsp->ls_stateid, cred, false);
7769 7770 7771 7772 7773 7774 7775
	nfs4_free_lock_state(server, lsp);
	if (IS_ERR(task))
		return PTR_ERR(task);
	rpc_put_task(task);
	return 0;
}

7776 7777 7778
static bool nfs41_match_stateid(const nfs4_stateid *s1,
		const nfs4_stateid *s2)
{
7779
	if (memcmp(s1->other, s2->other, sizeof(s1->other)) != 0)
7780 7781
		return false;

7782
	if (s1->seqid == s2->seqid)
7783
		return true;
7784
	if (s1->seqid == 0 || s2->seqid == 0)
7785 7786 7787 7788 7789
		return true;

	return false;
}

7790 7791
#endif /* CONFIG_NFS_V4_1 */

7792 7793 7794
static bool nfs4_match_stateid(const nfs4_stateid *s1,
		const nfs4_stateid *s2)
{
7795
	return nfs4_stateid_match(s1, s2);
7796 7797 7798
}


7799
static const struct nfs4_state_recovery_ops nfs40_reboot_recovery_ops = {
7800
	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
7801
	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
L
Linus Torvalds 已提交
7802 7803
	.recover_open	= nfs4_open_reclaim,
	.recover_lock	= nfs4_lock_reclaim,
7804
	.establish_clid = nfs4_init_clientid,
7805
	.detect_trunking = nfs40_discover_server_trunking,
L
Linus Torvalds 已提交
7806 7807
};

7808
#if defined(CONFIG_NFS_V4_1)
7809
static const struct nfs4_state_recovery_ops nfs41_reboot_recovery_ops = {
7810 7811 7812 7813
	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
	.recover_open	= nfs4_open_reclaim,
	.recover_lock	= nfs4_lock_reclaim,
7814
	.establish_clid = nfs41_init_clientid,
7815
	.reclaim_complete = nfs41_proc_reclaim_complete,
7816
	.detect_trunking = nfs41_discover_server_trunking,
7817 7818 7819
};
#endif /* CONFIG_NFS_V4_1 */

7820
static const struct nfs4_state_recovery_ops nfs40_nograce_recovery_ops = {
7821 7822 7823 7824 7825 7826 7827 7828
	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
	.recover_open	= nfs4_open_expired,
	.recover_lock	= nfs4_lock_expired,
	.establish_clid = nfs4_init_clientid,
};

#if defined(CONFIG_NFS_V4_1)
7829
static const struct nfs4_state_recovery_ops nfs41_nograce_recovery_ops = {
7830
	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
7831
	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
7832 7833
	.recover_open	= nfs41_open_expired,
	.recover_lock	= nfs41_lock_expired,
7834
	.establish_clid = nfs41_init_clientid,
L
Linus Torvalds 已提交
7835
};
7836
#endif /* CONFIG_NFS_V4_1 */
L
Linus Torvalds 已提交
7837

7838
static const struct nfs4_state_maintenance_ops nfs40_state_renewal_ops = {
B
Benny Halevy 已提交
7839
	.sched_state_renewal = nfs4_proc_async_renew,
7840
	.get_state_renewal_cred_locked = nfs4_get_renew_cred_locked,
7841
	.renew_lease = nfs4_proc_renew,
B
Benny Halevy 已提交
7842 7843 7844
};

#if defined(CONFIG_NFS_V4_1)
7845
static const struct nfs4_state_maintenance_ops nfs41_state_renewal_ops = {
B
Benny Halevy 已提交
7846
	.sched_state_renewal = nfs41_proc_async_sequence,
7847
	.get_state_renewal_cred_locked = nfs4_get_machine_cred_locked,
7848
	.renew_lease = nfs4_proc_sequence,
B
Benny Halevy 已提交
7849 7850 7851
};
#endif

7852 7853
static const struct nfs4_minor_version_ops nfs_v4_0_minor_ops = {
	.minor_version = 0,
7854 7855 7856 7857
	.init_caps = NFS_CAP_READDIRPLUS
		| NFS_CAP_ATOMIC_OPEN
		| NFS_CAP_CHANGE_ATTR
		| NFS_CAP_POSIX_LOCK,
7858 7859
	.init_client = nfs40_init_client,
	.shutdown_client = nfs40_shutdown_client,
7860
	.match_stateid = nfs4_match_stateid,
7861
	.find_root_sec = nfs4_find_root_sec,
7862
	.free_lock_state = nfs4_release_lockowner,
7863
	.call_sync_ops = &nfs40_call_sync_ops,
7864 7865 7866
	.reboot_recovery_ops = &nfs40_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs40_nograce_recovery_ops,
	.state_renewal_ops = &nfs40_state_renewal_ops,
7867 7868 7869 7870 7871
};

#if defined(CONFIG_NFS_V4_1)
static const struct nfs4_minor_version_ops nfs_v4_1_minor_ops = {
	.minor_version = 1,
7872 7873 7874
	.init_caps = NFS_CAP_READDIRPLUS
		| NFS_CAP_ATOMIC_OPEN
		| NFS_CAP_CHANGE_ATTR
7875
		| NFS_CAP_POSIX_LOCK
7876 7877
		| NFS_CAP_STATEID_NFSV41
		| NFS_CAP_ATOMIC_OPEN_V1,
7878 7879
	.init_client = nfs41_init_client,
	.shutdown_client = nfs41_shutdown_client,
7880
	.match_stateid = nfs41_match_stateid,
7881
	.find_root_sec = nfs41_find_root_sec,
7882
	.free_lock_state = nfs41_free_lock_state,
7883
	.call_sync_ops = &nfs41_call_sync_ops,
7884 7885 7886
	.reboot_recovery_ops = &nfs41_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs41_nograce_recovery_ops,
	.state_renewal_ops = &nfs41_state_renewal_ops,
7887 7888 7889
};
#endif

7890 7891 7892
#if defined(CONFIG_NFS_V4_2)
static const struct nfs4_minor_version_ops nfs_v4_2_minor_ops = {
	.minor_version = 2,
7893 7894 7895 7896 7897 7898
	.init_caps = NFS_CAP_READDIRPLUS
		| NFS_CAP_ATOMIC_OPEN
		| NFS_CAP_CHANGE_ATTR
		| NFS_CAP_POSIX_LOCK
		| NFS_CAP_STATEID_NFSV41
		| NFS_CAP_ATOMIC_OPEN_V1,
7899 7900
	.init_client = nfs41_init_client,
	.shutdown_client = nfs41_shutdown_client,
7901 7902
	.match_stateid = nfs41_match_stateid,
	.find_root_sec = nfs41_find_root_sec,
7903
	.free_lock_state = nfs41_free_lock_state,
7904
	.call_sync_ops = &nfs41_call_sync_ops,
7905 7906 7907 7908 7909 7910
	.reboot_recovery_ops = &nfs41_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs41_nograce_recovery_ops,
	.state_renewal_ops = &nfs41_state_renewal_ops,
};
#endif

7911 7912 7913 7914 7915
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
7916 7917 7918
#if defined(CONFIG_NFS_V4_2)
	[2] = &nfs_v4_2_minor_ops,
#endif
7919 7920
};

7921
static const struct inode_operations nfs4_dir_inode_operations = {
7922 7923 7924 7925 7926 7927 7928 7929 7930 7931 7932 7933 7934 7935 7936 7937 7938 7939 7940
	.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,
};

7941
static const struct inode_operations nfs4_file_inode_operations = {
7942 7943 7944
	.permission	= nfs_permission,
	.getattr	= nfs_getattr,
	.setattr	= nfs_setattr,
7945 7946 7947 7948
	.getxattr	= generic_getxattr,
	.setxattr	= generic_setxattr,
	.listxattr	= generic_listxattr,
	.removexattr	= generic_removexattr,
7949 7950
};

D
David Howells 已提交
7951
const struct nfs_rpc_ops nfs_v4_clientops = {
L
Linus Torvalds 已提交
7952 7953 7954
	.version	= 4,			/* protocol version */
	.dentry_ops	= &nfs4_dentry_operations,
	.dir_inode_ops	= &nfs4_dir_inode_operations,
7955
	.file_inode_ops	= &nfs4_file_inode_operations,
7956
	.file_ops	= &nfs4_file_operations,
L
Linus Torvalds 已提交
7957
	.getroot	= nfs4_proc_get_root,
B
Bryan Schumaker 已提交
7958
	.submount	= nfs4_submount,
B
Bryan Schumaker 已提交
7959
	.try_mount	= nfs4_try_mount,
L
Linus Torvalds 已提交
7960 7961 7962 7963 7964 7965 7966 7967
	.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,
7968
	.unlink_rpc_prepare = nfs4_proc_unlink_rpc_prepare,
L
Linus Torvalds 已提交
7969 7970
	.unlink_done	= nfs4_proc_unlink_done,
	.rename		= nfs4_proc_rename,
7971
	.rename_setup	= nfs4_proc_rename_setup,
7972
	.rename_rpc_prepare = nfs4_proc_rename_rpc_prepare,
7973
	.rename_done	= nfs4_proc_rename_done,
L
Linus Torvalds 已提交
7974 7975 7976 7977 7978 7979 7980 7981 7982
	.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,
7983
	.set_capabilities = nfs4_server_capabilities,
L
Linus Torvalds 已提交
7984 7985
	.decode_dirent	= nfs4_decode_dirent,
	.read_setup	= nfs4_proc_read_setup,
7986
	.read_pageio_init = pnfs_pageio_init_read,
7987
	.read_rpc_prepare = nfs4_proc_read_rpc_prepare,
T
Trond Myklebust 已提交
7988
	.read_done	= nfs4_read_done,
L
Linus Torvalds 已提交
7989
	.write_setup	= nfs4_proc_write_setup,
7990
	.write_pageio_init = pnfs_pageio_init_write,
7991
	.write_rpc_prepare = nfs4_proc_write_rpc_prepare,
7992
	.write_done	= nfs4_write_done,
L
Linus Torvalds 已提交
7993
	.commit_setup	= nfs4_proc_commit_setup,
7994
	.commit_rpc_prepare = nfs4_proc_commit_rpc_prepare,
7995
	.commit_done	= nfs4_commit_done,
L
Linus Torvalds 已提交
7996
	.lock		= nfs4_proc_lock,
7997
	.clear_acl_cache = nfs4_zap_acl_attr,
T
Trond Myklebust 已提交
7998
	.close_context  = nfs4_close_context,
7999
	.open_context	= nfs4_atomic_open,
8000
	.have_delegation = nfs4_have_delegation,
8001
	.return_delegation = nfs4_inode_return_delegation,
8002
	.alloc_client	= nfs4_alloc_client,
8003
	.init_client	= nfs4_init_client,
8004
	.free_client	= nfs4_free_client,
8005 8006
	.create_server	= nfs4_create_server,
	.clone_server	= nfs_clone_server,
L
Linus Torvalds 已提交
8007 8008
};

8009 8010 8011 8012 8013 8014 8015 8016 8017
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,
8018 8019 8020
#ifdef CONFIG_NFS_V4_SECURITY_LABEL
	&nfs4_xattr_nfs4_label_handler,
#endif
8021 8022 8023
	NULL
};

L
Linus Torvalds 已提交
8024 8025 8026 8027 8028
/*
 * Local variables:
 *  c-basic-offset: 8
 * End:
 */