nfs4proc.c 172.9 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
#include <linux/sunrpc/clnt.h>
46
#include <linux/sunrpc/gss_api.h>
L
Linus Torvalds 已提交
47 48 49 50
#include <linux/nfs.h>
#include <linux/nfs4.h>
#include <linux/nfs_fs.h>
#include <linux/nfs_page.h>
51
#include <linux/nfs_mount.h>
L
Linus Torvalds 已提交
52
#include <linux/namei.h>
53
#include <linux/mount.h>
B
Benny Halevy 已提交
54
#include <linux/module.h>
55
#include <linux/nfs_idmap.h>
56
#include <linux/sunrpc/bc_xprt.h>
57
#include <linux/xattr.h>
58
#include <linux/utsname.h>
59
#include <linux/freezer.h>
L
Linus Torvalds 已提交
60

61
#include "nfs4_fs.h"
L
Linus Torvalds 已提交
62
#include "delegation.h"
63
#include "internal.h"
64
#include "iostat.h"
A
Andy Adamson 已提交
65
#include "callback.h"
66
#include "pnfs.h"
L
Linus Torvalds 已提交
67 68 69

#define NFSDBG_FACILITY		NFSDBG_PROC

70
#define NFS4_POLL_RETRY_MIN	(HZ/10)
L
Linus Torvalds 已提交
71 72
#define NFS4_POLL_RETRY_MAX	(15*HZ)

73 74
#define NFS4_MAX_LOOP_ON_RECOVER (10)

75 76
static unsigned short max_session_slots = NFS4_DEF_SLOT_TABLE_SIZE;

77
struct nfs4_opendata;
78
static int _nfs4_proc_open(struct nfs4_opendata *data);
79
static int _nfs4_recover_proc_open(struct nfs4_opendata *data);
L
Linus Torvalds 已提交
80
static int nfs4_do_fsinfo(struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *);
81
static int nfs4_async_handle_error(struct rpc_task *, const struct nfs_server *, struct nfs4_state *);
82
static void nfs_fixup_referral_attributes(struct nfs_fattr *fattr);
83
static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fattr *fattr);
84 85 86
static int nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
			    struct nfs_fattr *fattr, struct iattr *sattr,
			    struct nfs4_state *state);
87
#ifdef CONFIG_NFS_V4_1
88 89
static int nfs41_test_stateid(struct nfs_server *, nfs4_stateid *);
static int nfs41_free_stateid(struct nfs_server *, nfs4_stateid *);
90
#endif
L
Linus Torvalds 已提交
91
/* Prevent leaks of NFSv4 errors into userland */
92
static int nfs4_map_errors(int err)
L
Linus Torvalds 已提交
93
{
94 95 96 97 98
	if (err >= -1000)
		return err;
	switch (err) {
	case -NFS4ERR_RESOURCE:
		return -EREMOTEIO;
99 100
	case -NFS4ERR_WRONGSEC:
		return -EPERM;
101 102 103
	case -NFS4ERR_BADOWNER:
	case -NFS4ERR_BADNAME:
		return -EINVAL;
104
	default:
L
Linus Torvalds 已提交
105
		dprintk("%s could not handle NFSv4 error %d\n",
106
				__func__, -err);
107
		break;
L
Linus Torvalds 已提交
108
	}
109
	return -EIO;
L
Linus Torvalds 已提交
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
}

/*
 * This is our standard bitmap for GETATTR requests.
 */
const u32 nfs4_fattr_bitmap[2] = {
	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
};

const u32 nfs4_statfs_bitmap[2] = {
	FATTR4_WORD0_FILES_AVAIL
	| FATTR4_WORD0_FILES_FREE
	| FATTR4_WORD0_FILES_TOTAL,
	FATTR4_WORD1_SPACE_AVAIL
	| FATTR4_WORD1_SPACE_FREE
	| FATTR4_WORD1_SPACE_TOTAL
};

141
const u32 nfs4_pathconf_bitmap[2] = {
L
Linus Torvalds 已提交
142 143 144 145 146
	FATTR4_WORD0_MAXLINK
	| FATTR4_WORD0_MAXNAME,
	0
};

147
const u32 nfs4_fsinfo_bitmap[3] = { FATTR4_WORD0_MAXFILESIZE
L
Linus Torvalds 已提交
148 149 150
			| FATTR4_WORD0_MAXREAD
			| FATTR4_WORD0_MAXWRITE
			| FATTR4_WORD0_LEASE_TIME,
151
			FATTR4_WORD1_TIME_DELTA
152 153
			| FATTR4_WORD1_FS_LAYOUT_TYPES,
			FATTR4_WORD2_LAYOUT_BLKSIZE
L
Linus Torvalds 已提交
154 155
};

156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
const u32 nfs4_fs_locations_bitmap[2] = {
	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
	| FATTR4_WORD1_MOUNTED_ON_FILEID
};

A
Al Viro 已提交
175
static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dentry,
L
Linus Torvalds 已提交
176 177
		struct nfs4_readdir_arg *readdir)
{
178
	__be32 *start, *p;
L
Linus Torvalds 已提交
179 180 181

	BUG_ON(readdir->count < 80);
	if (cookie > 2) {
182
		readdir->cookie = cookie;
L
Linus Torvalds 已提交
183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198
		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.
	 */
199
	start = p = kmap_atomic(*readdir->pages, KM_USER0);
L
Linus Torvalds 已提交
200 201 202 203 204 205 206 207 208 209 210
	
	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
211
		p = xdr_encode_hyper(p, NFS_FILEID(dentry->d_inode));
L
Linus Torvalds 已提交
212 213 214 215 216 217 218 219 220 221 222
	}
	
	*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
223
	p = xdr_encode_hyper(p, NFS_FILEID(dentry->d_parent->d_inode));
L
Linus Torvalds 已提交
224 225 226 227 228 229

	readdir->pgbase = (char *)p - (char *)start;
	readdir->count -= readdir->pgbase;
	kunmap_atomic(start, KM_USER0);
}

230 231 232 233 234 235
static int nfs4_wait_clnt_recover(struct nfs_client *clp)
{
	int res;

	might_sleep();

236
	res = wait_on_bit(&clp->cl_state, NFS4CLNT_MANAGER_RUNNING,
237
			nfs_wait_bit_killable, TASK_KILLABLE);
238 239 240 241 242 243 244 245 246 247 248 249 250
	return res;
}

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;
251
	freezable_schedule_timeout_killable(*timeout);
252 253 254 255 256 257 258 259 260
	if (fatal_signal_pending(current))
		res = -ERESTARTSYS;
	*timeout <<= 1;
	return res;
}

/* This is the error handling routine for processes that are allowed
 * to sleep.
 */
261
static int nfs4_handle_exception(struct nfs_server *server, int errorcode, struct nfs4_exception *exception)
262 263
{
	struct nfs_client *clp = server->nfs_client;
264
	struct nfs4_state *state = exception->state;
265 266 267 268 269 270
	int ret = errorcode;

	exception->retry = 0;
	switch(errorcode) {
		case 0:
			return 0;
271
		case -NFS4ERR_DELEG_REVOKED:
272 273
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
274 275
			if (state != NULL)
				nfs_remove_bad_delegation(state->inode);
276 277 278
		case -NFS4ERR_OPENMODE:
			if (state == NULL)
				break;
279 280
			nfs4_schedule_stateid_recovery(server, state);
			goto wait_on_recovery;
281 282 283
		case -NFS4ERR_EXPIRED:
			if (state != NULL)
				nfs4_schedule_stateid_recovery(server, state);
284
		case -NFS4ERR_STALE_STATEID:
285
		case -NFS4ERR_STALE_CLIENTID:
286 287
			nfs4_schedule_lease_recovery(clp);
			goto wait_on_recovery;
288
#if defined(CONFIG_NFS_V4_1)
289 290 291 292 293 294 295 296 297
		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);
298
			nfs4_schedule_session_recovery(clp->cl_session);
299
			exception->retry = 1;
300
			break;
301
#endif /* defined(CONFIG_NFS_V4_1) */
302
		case -NFS4ERR_FILE_OPEN:
303 304 305 306 307 308 309
			if (exception->timeout > HZ) {
				/* We have retried a decent amount, time to
				 * fail
				 */
				ret = -EBUSY;
				break;
			}
310 311
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
312
		case -EKEYEXPIRED:
313 314 315
			ret = nfs4_delay(server->client, &exception->timeout);
			if (ret != 0)
				break;
316
		case -NFS4ERR_RETRY_UNCACHED_REP:
317 318
		case -NFS4ERR_OLD_STATEID:
			exception->retry = 1;
319 320 321 322 323 324 325 326 327 328 329 330 331
			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);
			}
332 333 334
	}
	/* We failed to handle the error */
	return nfs4_map_errors(ret);
335
wait_on_recovery:
336 337 338 339
	ret = nfs4_wait_clnt_recover(clp);
	if (ret == 0)
		exception->retry = 1;
	return ret;
340 341 342
}


343
static void do_renew_lease(struct nfs_client *clp, unsigned long timestamp)
L
Linus Torvalds 已提交
344 345 346 347 348 349 350
{
	spin_lock(&clp->cl_lock);
	if (time_before(clp->cl_last_renewal,timestamp))
		clp->cl_last_renewal = timestamp;
	spin_unlock(&clp->cl_lock);
}

351 352 353 354 355
static void renew_lease(const struct nfs_server *server, unsigned long timestamp)
{
	do_renew_lease(server->nfs_client, timestamp);
}

A
Andy Adamson 已提交
356 357
#if defined(CONFIG_NFS_V4_1)

A
Andy Adamson 已提交
358 359 360 361 362 363 364 365 366 367 368
/*
 * nfs4_free_slot - free a slot and efficiently update slot table.
 *
 * freeing a slot is trivially done by clearing its respective bit
 * in the bitmap.
 * If the freed slotid equals highest_used_slotid we want to update it
 * so that the server would be able to size down the slot table if needed,
 * otherwise we know that the highest_used_slotid is still in use.
 * When updating highest_used_slotid there may be "holes" in the bitmap
 * so we need to scan down from highest_used_slotid to 0 looking for the now
 * highest slotid in use.
369
 * If none found, highest_used_slotid is set to NFS4_NO_SLOT.
370 371
 *
 * Must be called while holding tbl->slot_tbl_lock
A
Andy Adamson 已提交
372 373
 */
static void
374
nfs4_free_slot(struct nfs4_slot_table *tbl, u32 slotid)
A
Andy Adamson 已提交
375
{
376
	BUG_ON(slotid >= NFS4_MAX_SLOT_TABLE);
A
Andy Adamson 已提交
377 378 379 380 381 382
	/* clear used bit in bitmap */
	__clear_bit(slotid, tbl->used_slots);

	/* update highest_used_slotid when it is freed */
	if (slotid == tbl->highest_used_slotid) {
		slotid = find_last_bit(tbl->used_slots, tbl->max_slots);
T
Trond Myklebust 已提交
383
		if (slotid < tbl->max_slots)
A
Andy Adamson 已提交
384 385
			tbl->highest_used_slotid = slotid;
		else
386
			tbl->highest_used_slotid = NFS4_NO_SLOT;
A
Andy Adamson 已提交
387
	}
388 389
	dprintk("%s: slotid %u highest_used_slotid %d\n", __func__,
		slotid, tbl->highest_used_slotid);
A
Andy Adamson 已提交
390 391
}

392 393 394 395 396 397
bool nfs4_set_task_privileged(struct rpc_task *task, void *dummy)
{
	rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
	return true;
}

398
/*
399
 * Signal state manager thread if session fore channel is drained
400
 */
401
static void nfs4_check_drain_fc_complete(struct nfs4_session *ses)
402
{
403
	if (!test_bit(NFS4_SESSION_DRAINING, &ses->session_state)) {
404 405
		rpc_wake_up_first(&ses->fc_slot_table.slot_tbl_waitq,
				nfs4_set_task_privileged, NULL);
406 407 408
		return;
	}

409
	if (ses->fc_slot_table.highest_used_slotid != NFS4_NO_SLOT)
410 411
		return;

412 413 414 415 416 417 418 419 420 421
	dprintk("%s COMPLETE: Session Fore Channel Drained\n", __func__);
	complete(&ses->fc_slot_table.complete);
}

/*
 * Signal state manager thread if session back channel is drained
 */
void nfs4_check_drain_bc_complete(struct nfs4_session *ses)
{
	if (!test_bit(NFS4_SESSION_DRAINING, &ses->session_state) ||
422
	    ses->bc_slot_table.highest_used_slotid != NFS4_NO_SLOT)
423 424 425
		return;
	dprintk("%s COMPLETE: Session Back Channel Drained\n", __func__);
	complete(&ses->bc_slot_table.complete);
426 427
}

428
static void nfs41_sequence_free_slot(struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
429 430 431
{
	struct nfs4_slot_table *tbl;

432
	tbl = &res->sr_session->fc_slot_table;
433
	if (!res->sr_slot) {
A
Andy Adamson 已提交
434 435
		/* just wake up the next guy waiting since
		 * we may have not consumed a slot after all */
A
Andy Adamson 已提交
436
		dprintk("%s: No slot\n", __func__);
437
		return;
A
Andy Adamson 已提交
438
	}
439

440
	spin_lock(&tbl->slot_tbl_lock);
441
	nfs4_free_slot(tbl, res->sr_slot - tbl->slots);
442
	nfs4_check_drain_fc_complete(res->sr_session);
443
	spin_unlock(&tbl->slot_tbl_lock);
444
	res->sr_slot = NULL;
A
Andy Adamson 已提交
445 446
}

447
static int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
448 449
{
	unsigned long timestamp;
450
	struct nfs_client *clp;
A
Andy Adamson 已提交
451 452 453 454 455 456 457 458 459 460

	/*
	 * sr_status remains 1 if an RPC level error occurred. The server
	 * may or may not have processed the sequence operation..
	 * Proceed as if the server received and processed the sequence
	 * operation.
	 */
	if (res->sr_status == 1)
		res->sr_status = NFS_OK;

461 462
	/* don't increment the sequence number if the task wasn't sent */
	if (!RPC_WAS_SENT(task))
A
Andy Adamson 已提交
463 464
		goto out;

A
Andy Adamson 已提交
465
	/* Check the SEQUENCE operation status */
466 467
	switch (res->sr_status) {
	case 0:
A
Andy Adamson 已提交
468
		/* Update the slot's sequence and clientid lease timer */
469
		++res->sr_slot->seq_nr;
A
Andy Adamson 已提交
470
		timestamp = res->sr_renewal_time;
471
		clp = res->sr_session->clp;
472
		do_renew_lease(clp, timestamp);
473
		/* Check sequence flags */
474 475
		if (res->sr_status_flags != 0)
			nfs4_schedule_lease_recovery(clp);
476 477 478 479 480 481
		break;
	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.
		 */
482
		dprintk("%s: slot=%td seq=%d: Operation in progress\n",
483 484 485
			__func__,
			res->sr_slot - res->sr_session->fc_slot_table.slots,
			res->sr_slot->seq_nr);
486 487 488
		goto out_retry;
	default:
		/* Just update the slot sequence no. */
489
		++res->sr_slot->seq_nr;
A
Andy Adamson 已提交
490 491 492 493
	}
out:
	/* The session may be reset by one of the error handlers. */
	dprintk("%s: Error %d free the slot \n", __func__, res->sr_status);
494
	nfs41_sequence_free_slot(res);
495 496
	return 1;
out_retry:
497
	if (!rpc_restart_call(task))
498 499 500
		goto out;
	rpc_delay(task, NFS4_POLL_RETRY_MAX);
	return 0;
A
Andy Adamson 已提交
501 502
}

503 504
static int nfs4_sequence_done(struct rpc_task *task,
			       struct nfs4_sequence_res *res)
505
{
506 507 508
	if (res->sr_session == NULL)
		return 1;
	return nfs41_sequence_done(task, res);
509 510
}

B
Benny Halevy 已提交
511 512 513 514 515 516
/*
 * nfs4_find_slot - efficiently look for a free slot
 *
 * nfs4_find_slot looks for an unset bit in the used_slots bitmap.
 * If found, we mark the slot as used, update the highest_used_slotid,
 * and respectively set up the sequence operation args.
517
 * The slot number is returned if found, or NFS4_NO_SLOT otherwise.
A
Andy Adamson 已提交
518 519
 *
 * Note: must be called with under the slot_tbl_lock.
B
Benny Halevy 已提交
520
 */
521
static u32
522
nfs4_find_slot(struct nfs4_slot_table *tbl)
B
Benny Halevy 已提交
523
{
524 525
	u32 slotid;
	u32 ret_id = NFS4_NO_SLOT;
B
Benny Halevy 已提交
526

527
	dprintk("--> %s used_slots=%04lx highest_used=%u max_slots=%u\n",
B
Benny Halevy 已提交
528 529 530 531 532 533
		__func__, tbl->used_slots[0], tbl->highest_used_slotid,
		tbl->max_slots);
	slotid = find_first_zero_bit(tbl->used_slots, tbl->max_slots);
	if (slotid >= tbl->max_slots)
		goto out;
	__set_bit(slotid, tbl->used_slots);
534 535
	if (slotid > tbl->highest_used_slotid ||
			tbl->highest_used_slotid == NFS4_NO_SLOT)
B
Benny Halevy 已提交
536 537 538 539 540 541 542 543
		tbl->highest_used_slotid = slotid;
	ret_id = slotid;
out:
	dprintk("<-- %s used_slots=%04lx highest_used=%d slotid=%d \n",
		__func__, tbl->used_slots[0], tbl->highest_used_slotid, ret_id);
	return ret_id;
}

544 545 546 547 548 549 550 551 552 553 554
static void nfs41_init_sequence(struct nfs4_sequence_args *args,
		struct nfs4_sequence_res *res, int cache_reply)
{
	args->sa_session = NULL;
	args->sa_cache_this = 0;
	if (cache_reply)
		args->sa_cache_this = 1;
	res->sr_session = NULL;
	res->sr_slot = NULL;
}

A
Andy Adamson 已提交
555
int nfs41_setup_sequence(struct nfs4_session *session,
A
Andy Adamson 已提交
556 557 558 559
				struct nfs4_sequence_args *args,
				struct nfs4_sequence_res *res,
				struct rpc_task *task)
{
A
Andy Adamson 已提交
560 561
	struct nfs4_slot *slot;
	struct nfs4_slot_table *tbl;
562
	u32 slotid;
A
Andy Adamson 已提交
563 564

	dprintk("--> %s\n", __func__);
A
Andy Adamson 已提交
565
	/* slot already allocated? */
566
	if (res->sr_slot != NULL)
A
Andy Adamson 已提交
567 568
		return 0;

A
Andy Adamson 已提交
569 570 571
	tbl = &session->fc_slot_table;

	spin_lock(&tbl->slot_tbl_lock);
572
	if (test_bit(NFS4_SESSION_DRAINING, &session->session_state) &&
573
	    !rpc_task_has_priority(task, RPC_PRIORITY_PRIVILEGED)) {
574
		/* The state manager will wait until the slot table is empty */
A
Andy Adamson 已提交
575
		rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
576
		spin_unlock(&tbl->slot_tbl_lock);
577
		dprintk("%s session is draining\n", __func__);
A
Andy Adamson 已提交
578
		return -EAGAIN;
579 580
	}

581 582 583 584 585 586 587 588
	if (!rpc_queue_empty(&tbl->slot_tbl_waitq) &&
	    !rpc_task_has_priority(task, RPC_PRIORITY_PRIVILEGED)) {
		rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
		spin_unlock(&tbl->slot_tbl_lock);
		dprintk("%s enforce FIFO order\n", __func__);
		return -EAGAIN;
	}

589
	slotid = nfs4_find_slot(tbl);
590
	if (slotid == NFS4_NO_SLOT) {
A
Andy Adamson 已提交
591 592 593 594 595 596 597
		rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
		spin_unlock(&tbl->slot_tbl_lock);
		dprintk("<-- %s: no free slots\n", __func__);
		return -EAGAIN;
	}
	spin_unlock(&tbl->slot_tbl_lock);

598
	rpc_task_set_priority(task, RPC_PRIORITY_NORMAL);
A
Andy Adamson 已提交
599
	slot = tbl->slots + slotid;
B
Benny Halevy 已提交
600
	args->sa_session = session;
A
Andy Adamson 已提交
601 602 603 604
	args->sa_slotid = slotid;

	dprintk("<-- %s slotid=%d seqid=%d\n", __func__, slotid, slot->seq_nr);

B
Benny Halevy 已提交
605
	res->sr_session = session;
606
	res->sr_slot = slot;
A
Andy Adamson 已提交
607
	res->sr_renewal_time = jiffies;
608
	res->sr_status_flags = 0;
A
Andy Adamson 已提交
609 610 611 612 613
	/*
	 * 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;
A
Andy Adamson 已提交
614 615
	return 0;
}
A
Andy Adamson 已提交
616
EXPORT_SYMBOL_GPL(nfs41_setup_sequence);
A
Andy Adamson 已提交
617

618
int nfs4_setup_sequence(const struct nfs_server *server,
A
Andy Adamson 已提交
619 620 621 622
			struct nfs4_sequence_args *args,
			struct nfs4_sequence_res *res,
			struct rpc_task *task)
{
623
	struct nfs4_session *session = nfs4_get_session(server);
A
Andy Adamson 已提交
624 625
	int ret = 0;

626
	if (session == NULL)
627 628
		goto out;

629
	dprintk("--> %s clp %p session %p sr_slot %td\n",
630 631
		__func__, session->clp, session, res->sr_slot ?
			res->sr_slot - session->fc_slot_table.slots : -1);
A
Andy Adamson 已提交
632

633
	ret = nfs41_setup_sequence(session, args, res, task);
A
Andy Adamson 已提交
634 635 636 637 638 639
out:
	dprintk("<-- %s status=%d\n", __func__, ret);
	return ret;
}

struct nfs41_call_sync_data {
640
	const struct nfs_server *seq_server;
A
Andy Adamson 已提交
641 642 643 644 645 646 647 648
	struct nfs4_sequence_args *seq_args;
	struct nfs4_sequence_res *seq_res;
};

static void nfs41_call_sync_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs41_call_sync_data *data = calldata;

649 650 651
	dprintk("--> %s data->seq_server %p\n", __func__, data->seq_server);

	if (nfs4_setup_sequence(data->seq_server, data->seq_args,
652
				data->seq_res, task))
A
Andy Adamson 已提交
653 654 655 656
		return;
	rpc_call_start(task);
}

657 658 659 660 661 662
static void nfs41_call_priv_sync_prepare(struct rpc_task *task, void *calldata)
{
	rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
	nfs41_call_sync_prepare(task, calldata);
}

A
Andy Adamson 已提交
663 664 665 666
static void nfs41_call_sync_done(struct rpc_task *task, void *calldata)
{
	struct nfs41_call_sync_data *data = calldata;

667
	nfs41_sequence_done(task, data->seq_res);
A
Andy Adamson 已提交
668 669
}

A
Andy Adamson 已提交
670 671
struct rpc_call_ops nfs41_call_sync_ops = {
	.rpc_call_prepare = nfs41_call_sync_prepare,
A
Andy Adamson 已提交
672
	.rpc_call_done = nfs41_call_sync_done,
A
Andy Adamson 已提交
673 674
};

675 676 677 678 679
struct rpc_call_ops nfs41_call_priv_sync_ops = {
	.rpc_call_prepare = nfs41_call_priv_sync_prepare,
	.rpc_call_done = nfs41_call_sync_done,
};

680 681
static int nfs4_call_sync_sequence(struct rpc_clnt *clnt,
				   struct nfs_server *server,
A
Andy Adamson 已提交
682 683 684
				   struct rpc_message *msg,
				   struct nfs4_sequence_args *args,
				   struct nfs4_sequence_res *res,
685
				   int privileged)
A
Andy Adamson 已提交
686 687 688 689
{
	int ret;
	struct rpc_task *task;
	struct nfs41_call_sync_data data = {
690
		.seq_server = server,
A
Andy Adamson 已提交
691 692 693 694
		.seq_args = args,
		.seq_res = res,
	};
	struct rpc_task_setup task_setup = {
695
		.rpc_client = clnt,
A
Andy Adamson 已提交
696 697 698 699 700
		.rpc_message = msg,
		.callback_ops = &nfs41_call_sync_ops,
		.callback_data = &data
	};

701 702
	if (privileged)
		task_setup.callback_ops = &nfs41_call_priv_sync_ops;
A
Andy Adamson 已提交
703 704 705 706 707 708 709 710 711 712
	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;
}

713 714
int _nfs4_call_sync_session(struct rpc_clnt *clnt,
			    struct nfs_server *server,
A
Andy Adamson 已提交
715 716 717 718 719
			    struct rpc_message *msg,
			    struct nfs4_sequence_args *args,
			    struct nfs4_sequence_res *res,
			    int cache_reply)
{
720 721
	nfs41_init_sequence(args, res, cache_reply);
	return nfs4_call_sync_sequence(clnt, server, msg, args, res, 0);
A
Andy Adamson 已提交
722 723
}

724
#else
725 726 727 728 729 730
static inline
void nfs41_init_sequence(struct nfs4_sequence_args *args,
		struct nfs4_sequence_res *res, int cache_reply)
{
}

731 732
static int nfs4_sequence_done(struct rpc_task *task,
			       struct nfs4_sequence_res *res)
733
{
734
	return 1;
735
}
A
Andy Adamson 已提交
736 737
#endif /* CONFIG_NFS_V4_1 */

738 739
int _nfs4_call_sync(struct rpc_clnt *clnt,
		    struct nfs_server *server,
A
Andy Adamson 已提交
740 741 742 743 744
		    struct rpc_message *msg,
		    struct nfs4_sequence_args *args,
		    struct nfs4_sequence_res *res,
		    int cache_reply)
{
745
	nfs41_init_sequence(args, res, cache_reply);
746
	return rpc_call_sync(clnt, msg, 0);
A
Andy Adamson 已提交
747 748
}

749
static inline
750 751
int nfs4_call_sync(struct rpc_clnt *clnt,
		   struct nfs_server *server,
752 753 754 755 756
		   struct rpc_message *msg,
		   struct nfs4_sequence_args *args,
		   struct nfs4_sequence_res *res,
		   int cache_reply)
{
757 758
	return server->nfs_client->cl_mvops->call_sync(clnt, server, msg,
						args, res, cache_reply);
759
}
A
Andy Adamson 已提交
760

761
static void update_changeattr(struct inode *dir, struct nfs4_change_info *cinfo)
L
Linus Torvalds 已提交
762
{
763
	struct nfs_inode *nfsi = NFS_I(dir);
L
Linus Torvalds 已提交
764

765
	spin_lock(&dir->i_lock);
766
	nfsi->cache_validity |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE|NFS_INO_INVALID_DATA;
767
	if (!cinfo->atomic || cinfo->before != dir->i_version)
768
		nfs_force_lookup_revalidate(dir);
769
	dir->i_version = cinfo->after;
770
	spin_unlock(&dir->i_lock);
L
Linus Torvalds 已提交
771 772
}

773
struct nfs4_opendata {
774
	struct kref kref;
775 776
	struct nfs_openargs o_arg;
	struct nfs_openres o_res;
777 778
	struct nfs_open_confirmargs c_arg;
	struct nfs_open_confirmres c_res;
779 780
	struct nfs4_string owner_name;
	struct nfs4_string group_name;
781 782 783
	struct nfs_fattr f_attr;
	struct nfs_fattr dir_attr;
	struct dentry *dir;
784
	struct dentry *dentry;
785
	struct nfs4_state_owner *owner;
786
	struct nfs4_state *state;
787
	struct iattr attrs;
788
	unsigned long timestamp;
789
	unsigned int rpc_done : 1;
790 791
	int rpc_status;
	int cancelled;
792 793
};

794 795 796 797 798

static void nfs4_init_opendata_res(struct nfs4_opendata *p)
{
	p->o_res.f_attr = &p->f_attr;
	p->o_res.dir_attr = &p->dir_attr;
799 800
	p->o_res.seqid = p->o_arg.seqid;
	p->c_res.seqid = p->c_arg.seqid;
801 802 803
	p->o_res.server = p->o_arg.server;
	nfs_fattr_init(&p->f_attr);
	nfs_fattr_init(&p->dir_attr);
804
	nfs_fattr_init_names(&p->f_attr, &p->owner_name, &p->group_name);
805 806
}

807
static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry,
808
		struct nfs4_state_owner *sp, fmode_t fmode, int flags,
809 810
		const struct iattr *attrs,
		gfp_t gfp_mask)
811
{
812
	struct dentry *parent = dget_parent(dentry);
813 814 815 816
	struct inode *dir = parent->d_inode;
	struct nfs_server *server = NFS_SERVER(dir);
	struct nfs4_opendata *p;

817
	p = kzalloc(sizeof(*p), gfp_mask);
818 819
	if (p == NULL)
		goto err;
820
	p->o_arg.seqid = nfs_alloc_seqid(&sp->so_seqid, gfp_mask);
821 822
	if (p->o_arg.seqid == NULL)
		goto err_free;
823 824
	nfs_sb_active(dentry->d_sb);
	p->dentry = dget(dentry);
825 826 827 828
	p->dir = parent;
	p->owner = sp;
	atomic_inc(&sp->so_count);
	p->o_arg.fh = NFS_FH(dir);
829 830
	p->o_arg.open_flags = flags;
	p->o_arg.fmode = fmode & (FMODE_READ|FMODE_WRITE);
831
	p->o_arg.clientid = server->nfs_client->cl_clientid;
832
	p->o_arg.id = sp->so_seqid.owner_id;
833
	p->o_arg.name = &dentry->d_name;
834 835
	p->o_arg.server = server;
	p->o_arg.bitmask = server->attr_bitmask;
836
	p->o_arg.dir_bitmask = server->cache_consistency_bitmask;
837
	p->o_arg.claim = NFS4_OPEN_CLAIM_NULL;
838
	if (attrs != NULL && attrs->ia_valid != 0) {
839 840
		u32 *s;

841 842
		p->o_arg.u.attrs = &p->attrs;
		memcpy(&p->attrs, attrs, sizeof(p->attrs));
843 844 845
		s = (u32 *) p->o_arg.u.verifier.data;
		s[0] = jiffies;
		s[1] = current->pid;
846
	}
847 848 849
	p->c_arg.fh = &p->o_res.fh;
	p->c_arg.stateid = &p->o_res.stateid;
	p->c_arg.seqid = p->o_arg.seqid;
850
	nfs4_init_opendata_res(p);
851
	kref_init(&p->kref);
852 853 854 855 856 857 858 859
	return p;
err_free:
	kfree(p);
err:
	dput(parent);
	return NULL;
}

860
static void nfs4_opendata_free(struct kref *kref)
861
{
862 863
	struct nfs4_opendata *p = container_of(kref,
			struct nfs4_opendata, kref);
864
	struct super_block *sb = p->dentry->d_sb;
865 866

	nfs_free_seqid(p->o_arg.seqid);
867 868
	if (p->state != NULL)
		nfs4_put_open_state(p->state);
869 870
	nfs4_put_state_owner(p->owner);
	dput(p->dir);
871 872
	dput(p->dentry);
	nfs_sb_deactive(sb);
873
	nfs_fattr_free_names(&p->f_attr);
874 875 876 877 878 879 880
	kfree(p);
}

static void nfs4_opendata_put(struct nfs4_opendata *p)
{
	if (p != NULL)
		kref_put(&p->kref, nfs4_opendata_free);
881 882
}

883 884 885 886 887 888 889 890
static int nfs4_wait_for_completion_rpc_task(struct rpc_task *task)
{
	int ret;

	ret = rpc_wait_for_completion_task(task);
	return ret;
}

891
static int can_open_cached(struct nfs4_state *state, fmode_t mode, int open_mode)
892 893
{
	int ret = 0;
894

895
	if (open_mode & (O_EXCL|O_TRUNC))
896 897
		goto out;
	switch (mode & (FMODE_READ|FMODE_WRITE)) {
898
		case FMODE_READ:
899 900
			ret |= test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0
				&& state->n_rdonly != 0;
901 902
			break;
		case FMODE_WRITE:
903 904
			ret |= test_bit(NFS_O_WRONLY_STATE, &state->flags) != 0
				&& state->n_wronly != 0;
905 906
			break;
		case FMODE_READ|FMODE_WRITE:
907 908
			ret |= test_bit(NFS_O_RDWR_STATE, &state->flags) != 0
				&& state->n_rdwr != 0;
909
	}
910
out:
911 912 913
	return ret;
}

914
static int can_open_delegated(struct nfs_delegation *delegation, fmode_t fmode)
915
{
916 917
	if (delegation == NULL)
		return 0;
918
	if ((delegation->type & fmode) != fmode)
919
		return 0;
920
	if (test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags))
921
		return 0;
922
	nfs_mark_delegation_referenced(delegation);
923 924 925
	return 1;
}

926
static void update_open_stateflags(struct nfs4_state *state, fmode_t fmode)
927
{
928
	switch (fmode) {
929 930 931 932 933 934 935 936 937
		case FMODE_WRITE:
			state->n_wronly++;
			break;
		case FMODE_READ:
			state->n_rdonly++;
			break;
		case FMODE_READ|FMODE_WRITE:
			state->n_rdwr++;
	}
938
	nfs4_state_set_mode_locked(state, state->state | fmode);
939 940
}

941
static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
942 943
{
	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
944 945
		nfs4_stateid_copy(&state->stateid, stateid);
	nfs4_stateid_copy(&state->open_stateid, stateid);
946
	switch (fmode) {
947 948 949 950 951 952 953 954 955
		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);
	}
956 957
}

958
static void nfs_set_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
959
{
960
	write_seqlock(&state->seqlock);
961
	nfs_set_open_stateid_locked(state, stateid, fmode);
962
	write_sequnlock(&state->seqlock);
963 964
}

965
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
966
{
967 968 969 970 971
	/*
	 * Protect the call to nfs4_state_set_mode_locked and
	 * serialise the stateid update
	 */
	write_seqlock(&state->seqlock);
972
	if (deleg_stateid != NULL) {
973
		nfs4_stateid_copy(&state->stateid, deleg_stateid);
974 975 976
		set_bit(NFS_DELEGATED_STATE, &state->flags);
	}
	if (open_stateid != NULL)
977
		nfs_set_open_stateid_locked(state, open_stateid, fmode);
978 979
	write_sequnlock(&state->seqlock);
	spin_lock(&state->owner->so_lock);
980
	update_open_stateflags(state, fmode);
981
	spin_unlock(&state->owner->so_lock);
L
Linus Torvalds 已提交
982 983
}

984
static int update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *delegation, fmode_t fmode)
985 986 987 988 989
{
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *deleg_cur;
	int ret = 0;

990
	fmode &= (FMODE_READ|FMODE_WRITE);
991 992 993 994 995 996 997 998

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

	spin_lock(&deleg_cur->lock);
	if (nfsi->delegation != deleg_cur ||
999
	    (deleg_cur->type & fmode) != fmode)
1000 1001 1002 1003
		goto no_delegation_unlock;

	if (delegation == NULL)
		delegation = &deleg_cur->stateid;
1004
	else if (!nfs4_stateid_match(&deleg_cur->stateid, delegation))
1005 1006
		goto no_delegation_unlock;

1007
	nfs_mark_delegation_referenced(deleg_cur);
1008
	__update_open_stateid(state, open_stateid, &deleg_cur->stateid, fmode);
1009 1010 1011 1012 1013 1014 1015
	ret = 1;
no_delegation_unlock:
	spin_unlock(&deleg_cur->lock);
no_delegation:
	rcu_read_unlock();

	if (!ret && open_stateid != NULL) {
1016
		__update_open_stateid(state, open_stateid, NULL, fmode);
1017 1018 1019 1020 1021 1022 1023
		ret = 1;
	}

	return ret;
}


1024
static void nfs4_return_incompatible_delegation(struct inode *inode, fmode_t fmode)
1025 1026 1027 1028 1029
{
	struct nfs_delegation *delegation;

	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(inode)->delegation);
1030
	if (delegation == NULL || (delegation->type & fmode) == fmode) {
1031 1032 1033 1034 1035 1036 1037
		rcu_read_unlock();
		return;
	}
	rcu_read_unlock();
	nfs_inode_return_delegation(inode);
}

1038
static struct nfs4_state *nfs4_try_open_cached(struct nfs4_opendata *opendata)
1039 1040 1041 1042
{
	struct nfs4_state *state = opendata->state;
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *delegation;
1043
	int open_mode = opendata->o_arg.open_flags & (O_EXCL|O_TRUNC);
1044
	fmode_t fmode = opendata->o_arg.fmode;
1045 1046 1047 1048
	nfs4_stateid stateid;
	int ret = -EAGAIN;

	for (;;) {
1049
		if (can_open_cached(state, fmode, open_mode)) {
1050
			spin_lock(&state->owner->so_lock);
1051 1052
			if (can_open_cached(state, fmode, open_mode)) {
				update_open_stateflags(state, fmode);
1053 1054 1055 1056 1057
				spin_unlock(&state->owner->so_lock);
				goto out_return_state;
			}
			spin_unlock(&state->owner->so_lock);
		}
1058 1059
		rcu_read_lock();
		delegation = rcu_dereference(nfsi->delegation);
1060
		if (!can_open_delegated(delegation, fmode)) {
1061
			rcu_read_unlock();
1062
			break;
1063
		}
1064
		/* Save the delegation */
1065
		nfs4_stateid_copy(&stateid, &delegation->stateid);
1066
		rcu_read_unlock();
1067
		ret = nfs_may_open(state->inode, state->owner->so_cred, open_mode);
1068 1069 1070
		if (ret != 0)
			goto out;
		ret = -EAGAIN;
1071 1072

		/* Try to update the stateid using the delegation */
1073
		if (update_open_stateid(state, NULL, &stateid, fmode))
1074
			goto out_return_state;
1075 1076 1077 1078 1079 1080 1081 1082
	}
out:
	return ERR_PTR(ret);
out_return_state:
	atomic_inc(&state->count);
	return state;
}

1083 1084 1085 1086
static struct nfs4_state *nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data)
{
	struct inode *inode;
	struct nfs4_state *state = NULL;
1087
	struct nfs_delegation *delegation;
1088
	int ret;
1089

1090
	if (!data->rpc_done) {
1091
		state = nfs4_try_open_cached(data);
1092 1093 1094
		goto out;
	}

1095
	ret = -EAGAIN;
1096
	if (!(data->f_attr.valid & NFS_ATTR_FATTR))
1097
		goto err;
1098
	inode = nfs_fhget(data->dir->d_sb, &data->o_res.fh, &data->f_attr);
1099
	ret = PTR_ERR(inode);
1100
	if (IS_ERR(inode))
1101 1102
		goto err;
	ret = -ENOMEM;
1103 1104
	state = nfs4_get_open_state(inode, data->owner);
	if (state == NULL)
1105
		goto err_put_inode;
1106
	if (data->o_res.delegation_type != 0) {
1107
		struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
1108 1109
		int delegation_flags = 0;

1110 1111 1112 1113 1114
		rcu_read_lock();
		delegation = rcu_dereference(NFS_I(inode)->delegation);
		if (delegation)
			delegation_flags = delegation->flags;
		rcu_read_unlock();
1115 1116 1117 1118
		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",
1119
					clp->cl_hostname);
1120
		} else if ((delegation_flags & 1UL<<NFS_DELEGATION_NEED_RECLAIM) == 0)
1121 1122 1123 1124 1125 1126 1127 1128
			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);
	}
1129 1130

	update_open_stateid(state, &data->o_res.stateid, NULL,
1131
			data->o_arg.fmode);
1132
	iput(inode);
1133
out:
1134
	return state;
1135 1136 1137 1138
err_put_inode:
	iput(inode);
err:
	return ERR_PTR(ret);
1139 1140
}

1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157
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);
}

T
Trond Myklebust 已提交
1158 1159 1160 1161
static struct nfs4_opendata *nfs4_open_recoverdata_alloc(struct nfs_open_context *ctx, struct nfs4_state *state)
{
	struct nfs4_opendata *opendata;

1162
	opendata = nfs4_opendata_alloc(ctx->dentry, state->owner, 0, 0, NULL, GFP_NOFS);
T
Trond Myklebust 已提交
1163 1164 1165 1166 1167 1168 1169
	if (opendata == NULL)
		return ERR_PTR(-ENOMEM);
	opendata->state = state;
	atomic_inc(&state->count);
	return opendata;
}

1170
static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, fmode_t fmode, struct nfs4_state **res)
1171
{
1172
	struct nfs4_state *newstate;
1173 1174
	int ret;

1175 1176
	opendata->o_arg.open_flags = 0;
	opendata->o_arg.fmode = fmode;
1177 1178 1179
	memset(&opendata->o_res, 0, sizeof(opendata->o_res));
	memset(&opendata->c_res, 0, sizeof(opendata->c_res));
	nfs4_init_opendata_res(opendata);
1180
	ret = _nfs4_recover_proc_open(opendata);
1181 1182
	if (ret != 0)
		return ret; 
1183
	newstate = nfs4_opendata_to_nfs4_state(opendata);
1184 1185
	if (IS_ERR(newstate))
		return PTR_ERR(newstate);
1186
	nfs4_close_state(newstate, fmode);
1187
	*res = newstate;
1188 1189 1190 1191 1192 1193 1194 1195 1196
	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_* */
1197
	clear_bit(NFS_DELEGATED_STATE, &state->flags);
1198 1199
	smp_rmb();
	if (state->n_rdwr != 0) {
T
Trond Myklebust 已提交
1200
		clear_bit(NFS_O_RDWR_STATE, &state->flags);
1201
		ret = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE, &newstate);
1202 1203
		if (ret != 0)
			return ret;
1204 1205
		if (newstate != state)
			return -ESTALE;
1206 1207
	}
	if (state->n_wronly != 0) {
T
Trond Myklebust 已提交
1208
		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
1209
		ret = nfs4_open_recover_helper(opendata, FMODE_WRITE, &newstate);
1210 1211
		if (ret != 0)
			return ret;
1212 1213
		if (newstate != state)
			return -ESTALE;
1214 1215
	}
	if (state->n_rdonly != 0) {
T
Trond Myklebust 已提交
1216
		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
1217
		ret = nfs4_open_recover_helper(opendata, FMODE_READ, &newstate);
1218 1219
		if (ret != 0)
			return ret;
1220 1221
		if (newstate != state)
			return -ESTALE;
1222
	}
1223 1224 1225 1226 1227
	/*
	 * 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 &&
1228
	    !nfs4_stateid_match(&state->stateid, &state->open_stateid)) {
1229
		write_seqlock(&state->seqlock);
1230
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
1231
			nfs4_stateid_copy(&state->stateid, &state->open_stateid);
1232
		write_sequnlock(&state->seqlock);
1233
	}
1234 1235 1236
	return 0;
}

L
Linus Torvalds 已提交
1237 1238 1239 1240
/*
 * OPEN_RECLAIM:
 * 	reclaim state on the server after a reboot.
 */
1241
static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1242
{
1243
	struct nfs_delegation *delegation;
1244
	struct nfs4_opendata *opendata;
1245
	fmode_t delegation_type = 0;
L
Linus Torvalds 已提交
1246 1247
	int status;

T
Trond Myklebust 已提交
1248 1249 1250
	opendata = nfs4_open_recoverdata_alloc(ctx, state);
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1251 1252
	opendata->o_arg.claim = NFS4_OPEN_CLAIM_PREVIOUS;
	opendata->o_arg.fh = NFS_FH(state->inode);
1253 1254
	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
1255
	if (delegation != NULL && test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) != 0)
1256
		delegation_type = delegation->type;
1257
	rcu_read_unlock();
1258 1259
	opendata->o_arg.u.delegation_type = delegation_type;
	status = nfs4_open_recover(opendata, state);
1260
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
1261 1262 1263
	return status;
}

1264
static int nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1265 1266 1267 1268 1269
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;
	do {
1270
		err = _nfs4_do_open_reclaim(ctx, state);
1271
		if (err != -NFS4ERR_DELAY)
1272 1273
			break;
		nfs4_handle_exception(server, err, &exception);
L
Linus Torvalds 已提交
1274 1275 1276 1277
	} while (exception.retry);
	return err;
}

1278 1279 1280 1281 1282 1283 1284 1285
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))
		return PTR_ERR(ctx);
1286
	ret = nfs4_do_open_reclaim(ctx, state);
1287 1288 1289 1290
	put_nfs_open_context(ctx);
	return ret;
}

1291
static int _nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid)
L
Linus Torvalds 已提交
1292
{
1293
	struct nfs4_opendata *opendata;
1294
	int ret;
L
Linus Torvalds 已提交
1295

T
Trond Myklebust 已提交
1296 1297 1298
	opendata = nfs4_open_recoverdata_alloc(ctx, state);
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1299
	opendata->o_arg.claim = NFS4_OPEN_CLAIM_DELEGATE_CUR;
1300
	nfs4_stateid_copy(&opendata->o_arg.u.delegation, stateid);
1301
	ret = nfs4_open_recover(opendata, state);
1302
	nfs4_opendata_put(opendata);
1303
	return ret;
L
Linus Torvalds 已提交
1304 1305
}

1306
int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid)
L
Linus Torvalds 已提交
1307 1308
{
	struct nfs4_exception exception = { };
1309
	struct nfs_server *server = NFS_SERVER(state->inode);
L
Linus Torvalds 已提交
1310 1311
	int err;
	do {
1312
		err = _nfs4_open_delegation_recall(ctx, state, stateid);
L
Linus Torvalds 已提交
1313 1314
		switch (err) {
			case 0:
1315 1316 1317
			case -ENOENT:
			case -ESTALE:
				goto out;
1318 1319 1320 1321 1322
			case -NFS4ERR_BADSESSION:
			case -NFS4ERR_BADSLOT:
			case -NFS4ERR_BAD_HIGH_SLOT:
			case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
			case -NFS4ERR_DEADSESSION:
1323
				nfs4_schedule_session_recovery(server->nfs_client->cl_session);
1324
				goto out;
L
Linus Torvalds 已提交
1325 1326 1327 1328
			case -NFS4ERR_STALE_CLIENTID:
			case -NFS4ERR_STALE_STATEID:
			case -NFS4ERR_EXPIRED:
				/* Don't recall a delegation if it was lost */
1329
				nfs4_schedule_lease_recovery(server->nfs_client);
1330 1331 1332 1333 1334 1335
				goto out;
			case -ERESTARTSYS:
				/*
				 * The show must go on: exit, but mark the
				 * stateid as needing recovery.
				 */
1336
			case -NFS4ERR_DELEG_REVOKED:
1337 1338
			case -NFS4ERR_ADMIN_REVOKED:
			case -NFS4ERR_BAD_STATEID:
1339 1340
				nfs_inode_find_state_and_recover(state->inode,
						stateid);
1341
				nfs4_schedule_stateid_recovery(server, state);
1342 1343 1344 1345 1346 1347 1348
			case -EKEYEXPIRED:
				/*
				 * User RPCSEC_GSS context has expired.
				 * We cannot recover this stateid now, so
				 * skip it and allow recovery thread to
				 * proceed.
				 */
1349 1350 1351
			case -ENOMEM:
				err = 0;
				goto out;
L
Linus Torvalds 已提交
1352 1353 1354
		}
		err = nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
1355
out:
L
Linus Torvalds 已提交
1356 1357 1358
	return err;
}

1359 1360 1361 1362 1363
static void nfs4_open_confirm_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;

	data->rpc_status = task->tk_status;
1364
	if (data->rpc_status == 0) {
1365
		nfs4_stateid_copy(&data->o_res.stateid, &data->c_res.stateid);
1366
		nfs_confirm_seqid(&data->owner->so_seqid, 0);
1367
		renew_lease(data->o_res.server, data->timestamp);
1368
		data->rpc_done = 1;
1369
	}
1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380
}

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! */
1381
	if (!data->rpc_done)
1382 1383
		goto out_free;
	state = nfs4_opendata_to_nfs4_state(data);
1384
	if (!IS_ERR(state))
1385
		nfs4_close_state(state, data->o_arg.fmode);
1386
out_free:
1387
	nfs4_opendata_put(data);
1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401
}

static const struct rpc_call_ops nfs4_open_confirm_ops = {
	.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;
1402 1403 1404 1405 1406 1407
	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
1408 1409
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1410
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1411 1412
		.callback_ops = &nfs4_open_confirm_ops,
		.callback_data = data,
1413
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1414 1415
		.flags = RPC_TASK_ASYNC,
	};
L
Linus Torvalds 已提交
1416 1417
	int status;

1418
	kref_get(&data->kref);
1419 1420
	data->rpc_done = 0;
	data->rpc_status = 0;
1421
	data->timestamp = jiffies;
T
Trond Myklebust 已提交
1422
	task = rpc_run_task(&task_setup_data);
1423
	if (IS_ERR(task))
1424 1425 1426 1427 1428 1429 1430
		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;
1431
	rpc_put_task(task);
L
Linus Torvalds 已提交
1432 1433 1434
	return status;
}

1435
static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
L
Linus Torvalds 已提交
1436
{
1437 1438
	struct nfs4_opendata *data = calldata;
	struct nfs4_state_owner *sp = data->owner;
1439

1440 1441
	if (nfs_wait_on_sequence(data->o_arg.seqid, task) != 0)
		return;
1442 1443 1444 1445 1446 1447 1448
	/*
	 * 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;

1449
		if (can_open_cached(data->state, data->o_arg.fmode, data->o_arg.open_flags))
1450
			goto out_no_action;
1451 1452
		rcu_read_lock();
		delegation = rcu_dereference(NFS_I(data->state->inode)->delegation);
1453 1454 1455
		if (data->o_arg.claim != NFS4_OPEN_CLAIM_DELEGATE_CUR &&
		    can_open_delegated(delegation, data->o_arg.fmode))
			goto unlock_no_action;
1456 1457
		rcu_read_unlock();
	}
1458
	/* Update sequence id. */
1459
	data->o_arg.id = sp->so_seqid.owner_id;
1460
	data->o_arg.clientid = sp->so_server->nfs_client->cl_clientid;
T
Trond Myklebust 已提交
1461
	if (data->o_arg.claim == NFS4_OPEN_CLAIM_PREVIOUS) {
1462
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR];
T
Trond Myklebust 已提交
1463 1464
		nfs_copy_fh(&data->o_res.fh, data->o_arg.fh);
	}
1465
	data->timestamp = jiffies;
1466
	if (nfs4_setup_sequence(data->o_arg.server,
1467
				&data->o_arg.seq_args,
1468
				&data->o_res.seq_res, task))
1469
		return;
1470
	rpc_call_start(task);
1471
	return;
1472 1473
unlock_no_action:
	rcu_read_unlock();
1474 1475 1476
out_no_action:
	task->tk_action = NULL;

1477
}
L
Linus Torvalds 已提交
1478

1479 1480 1481 1482 1483 1484
static void nfs4_recover_open_prepare(struct rpc_task *task, void *calldata)
{
	rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
	nfs4_open_prepare(task, calldata);
}

1485 1486 1487
static void nfs4_open_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;
L
Linus Torvalds 已提交
1488

1489
	data->rpc_status = task->tk_status;
1490

1491 1492
	if (!nfs4_sequence_done(task, &data->o_res.seq_res))
		return;
1493

1494 1495
	if (task->tk_status == 0) {
		switch (data->o_res.f_attr->mode & S_IFMT) {
1496 1497 1498
			case S_IFREG:
				break;
			case S_IFLNK:
1499
				data->rpc_status = -ELOOP;
1500 1501
				break;
			case S_IFDIR:
1502
				data->rpc_status = -EISDIR;
1503 1504
				break;
			default:
1505
				data->rpc_status = -ENOTDIR;
1506
		}
1507
		renew_lease(data->o_res.server, data->timestamp);
1508 1509
		if (!(data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM))
			nfs_confirm_seqid(&data->owner->so_seqid, 0);
1510
	}
1511
	data->rpc_done = 1;
1512
}
1513

1514 1515 1516 1517 1518 1519 1520 1521 1522
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! */
1523
	if (data->rpc_status != 0 || !data->rpc_done)
1524 1525 1526 1527 1528
		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);
1529
	if (!IS_ERR(state))
1530
		nfs4_close_state(state, data->o_arg.fmode);
1531
out_free:
1532
	nfs4_opendata_put(data);
1533 1534 1535 1536 1537 1538 1539 1540
}

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,
};

1541 1542 1543 1544 1545 1546 1547
static const struct rpc_call_ops nfs4_recover_open_ops = {
	.rpc_call_prepare = nfs4_recover_open_prepare,
	.rpc_call_done = nfs4_open_done,
	.rpc_release = nfs4_open_release,
};

static int nfs4_run_open_task(struct nfs4_opendata *data, int isrecover)
1548 1549 1550 1551 1552 1553
{
	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;
1554 1555 1556 1557 1558 1559
	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
1560 1561
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1562
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1563 1564
		.callback_ops = &nfs4_open_ops,
		.callback_data = data,
1565
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1566 1567
		.flags = RPC_TASK_ASYNC,
	};
1568 1569
	int status;

1570
	nfs41_init_sequence(&o_arg->seq_args, &o_res->seq_res, 1);
1571
	kref_get(&data->kref);
1572 1573
	data->rpc_done = 0;
	data->rpc_status = 0;
1574
	data->cancelled = 0;
1575 1576
	if (isrecover)
		task_setup_data.callback_ops = &nfs4_recover_open_ops;
T
Trond Myklebust 已提交
1577
	task = rpc_run_task(&task_setup_data);
1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600
        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;

1601 1602
	nfs_fattr_map_and_free_names(NFS_SERVER(dir), &data->f_attr);

1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625
	nfs_refresh_inode(dir, o_res->dir_attr);

	if (o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
		status = _nfs4_proc_open_confirm(data);
		if (status != 0)
			return status;
	}

	return status;
}

/*
 * 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);
1626 1627 1628 1629 1630 1631
	if (!data->rpc_done)
		return status;
	if (status != 0) {
		if (status == -NFS4ERR_BADNAME &&
				!(o_arg->open_flags & O_CREAT))
			return -ENOENT;
1632
		return status;
1633
	}
1634

1635 1636
	nfs_fattr_map_and_free_names(server, &data->f_attr);

1637 1638 1639 1640 1641
	if (o_arg->open_flags & O_CREAT) {
		update_changeattr(dir, &o_res->cinfo);
		nfs_post_op_update_inode(dir, o_res->dir_attr);
	} else
		nfs_refresh_inode(dir, o_res->dir_attr);
T
Trond Myklebust 已提交
1642 1643
	if ((o_res->rflags & NFS4_OPEN_RESULT_LOCKTYPE_POSIX) == 0)
		server->caps &= ~NFS_CAP_POSIX_LOCK;
L
Linus Torvalds 已提交
1644
	if(o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
1645
		status = _nfs4_proc_open_confirm(data);
L
Linus Torvalds 已提交
1646
		if (status != 0)
1647
			return status;
L
Linus Torvalds 已提交
1648 1649
	}
	if (!(o_res->f_attr->valid & NFS_ATTR_FATTR))
1650
		_nfs4_proc_getattr(server, &o_res->fh, o_res->f_attr);
1651
	return 0;
L
Linus Torvalds 已提交
1652 1653
}

A
Andy Adamson 已提交
1654
static int nfs4_client_recover_expired_lease(struct nfs_client *clp)
1655
{
1656
	unsigned int loop;
1657
	int ret;
1658

1659
	for (loop = NFS4_MAX_LOOP_ON_RECOVER; loop != 0; loop--) {
1660
		ret = nfs4_wait_clnt_recover(clp);
1661
		if (ret != 0)
1662
			break;
1663 1664
		if (!test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state) &&
		    !test_bit(NFS4CLNT_CHECK_LEASE,&clp->cl_state))
1665
			break;
1666
		nfs4_schedule_state_manager(clp);
1667
		ret = -EIO;
1668
	}
1669
	return ret;
1670 1671
}

A
Andy Adamson 已提交
1672 1673 1674 1675 1676
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
1677 1678 1679 1680 1681
/*
 * OPEN_EXPIRED:
 * 	reclaim state on the server after a network partition.
 * 	Assumes caller holds the appropriate lock
 */
1682
static int _nfs4_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1683
{
1684
	struct nfs4_opendata *opendata;
1685
	int ret;
L
Linus Torvalds 已提交
1686

T
Trond Myklebust 已提交
1687 1688 1689
	opendata = nfs4_open_recoverdata_alloc(ctx, state);
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1690
	ret = nfs4_open_recover(opendata, state);
1691
	if (ret == -ESTALE)
1692
		d_drop(ctx->dentry);
1693
	nfs4_opendata_put(opendata);
1694
	return ret;
L
Linus Torvalds 已提交
1695 1696
}

1697
static int nfs4_do_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
1698
{
1699
	struct nfs_server *server = NFS_SERVER(state->inode);
1700 1701 1702 1703
	struct nfs4_exception exception = { };
	int err;

	do {
1704
		err = _nfs4_open_expired(ctx, state);
1705 1706 1707 1708 1709 1710 1711 1712
		switch (err) {
		default:
			goto out;
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			nfs4_handle_exception(server, err, &exception);
			err = 0;
		}
1713
	} while (exception.retry);
1714
out:
1715 1716 1717
	return err;
}

L
Linus Torvalds 已提交
1718 1719 1720
static int nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
	struct nfs_open_context *ctx;
1721
	int ret;
L
Linus Torvalds 已提交
1722

1723 1724 1725
	ctx = nfs4_state_find_open_context(state);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);
1726
	ret = nfs4_do_open_expired(ctx, state);
1727 1728
	put_nfs_open_context(ctx);
	return ret;
L
Linus Torvalds 已提交
1729 1730
}

1731
#if defined(CONFIG_NFS_V4_1)
1732
static int nfs41_check_expired_stateid(struct nfs4_state *state, nfs4_stateid *stateid, unsigned int flags)
1733
{
1734
	int status = NFS_OK;
1735 1736
	struct nfs_server *server = NFS_SERVER(state->inode);

1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757
	if (state->flags & flags) {
		status = nfs41_test_stateid(server, stateid);
		if (status != NFS_OK) {
			nfs41_free_stateid(server, stateid);
			state->flags &= ~flags;
		}
	}
	return status;
}

static int nfs41_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
	int deleg_status, open_status;
	int deleg_flags = 1 << NFS_DELEGATED_STATE;
	int open_flags = (1 << NFS_O_RDONLY_STATE) | (1 << NFS_O_WRONLY_STATE) | (1 << NFS_O_RDWR_STATE);

	deleg_status = nfs41_check_expired_stateid(state, &state->stateid, deleg_flags);
	open_status = nfs41_check_expired_stateid(state,  &state->open_stateid, open_flags);

	if ((deleg_status == NFS_OK) && (open_status == NFS_OK))
		return NFS_OK;
1758 1759 1760 1761
	return nfs4_open_expired(sp, state);
}
#endif

1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777
/*
 * 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;
}

L
Linus Torvalds 已提交
1778
/*
1779
 * Returns a referenced nfs4_state
L
Linus Torvalds 已提交
1780
 */
1781
static int _nfs4_do_open(struct inode *dir, struct dentry *dentry, fmode_t fmode, int flags, struct iattr *sattr, struct rpc_cred *cred, struct nfs4_state **res)
L
Linus Torvalds 已提交
1782 1783 1784 1785
{
	struct nfs4_state_owner  *sp;
	struct nfs4_state     *state = NULL;
	struct nfs_server       *server = NFS_SERVER(dir);
1786
	struct nfs4_opendata *opendata;
1787
	int status;
L
Linus Torvalds 已提交
1788 1789 1790

	/* Protect against reboot recovery conflicts */
	status = -ENOMEM;
1791 1792
	sp = nfs4_get_state_owner(server, cred, GFP_KERNEL);
	if (sp == NULL) {
L
Linus Torvalds 已提交
1793 1794 1795
		dprintk("nfs4_do_open: nfs4_get_state_owner failed!\n");
		goto out_err;
	}
1796 1797
	status = nfs4_recover_expired_lease(server);
	if (status != 0)
1798
		goto err_put_state_owner;
1799 1800
	if (dentry->d_inode != NULL)
		nfs4_return_incompatible_delegation(dentry->d_inode, fmode);
1801
	status = -ENOMEM;
1802
	opendata = nfs4_opendata_alloc(dentry, sp, fmode, flags, sattr, GFP_KERNEL);
1803
	if (opendata == NULL)
T
Trond Myklebust 已提交
1804
		goto err_put_state_owner;
L
Linus Torvalds 已提交
1805

1806 1807
	if (dentry->d_inode != NULL)
		opendata->state = nfs4_get_open_state(dentry->d_inode, sp);
1808

1809
	status = _nfs4_proc_open(opendata);
L
Linus Torvalds 已提交
1810
	if (status != 0)
1811
		goto err_opendata_put;
L
Linus Torvalds 已提交
1812

1813
	state = nfs4_opendata_to_nfs4_state(opendata);
1814 1815
	status = PTR_ERR(state);
	if (IS_ERR(state))
1816
		goto err_opendata_put;
T
Trond Myklebust 已提交
1817
	if (server->caps & NFS_CAP_POSIX_LOCK)
1818
		set_bit(NFS_STATE_POSIX_LOCKS, &state->flags);
1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830

	if (opendata->o_arg.open_flags & O_EXCL) {
		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,
				state);
		if (status == 0)
			nfs_setattr_update_inode(state->inode, sattr);
		nfs_post_op_update_inode(state->inode, opendata->o_res.f_attr);
	}
1831
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
1832 1833 1834
	nfs4_put_state_owner(sp);
	*res = state;
	return 0;
1835 1836
err_opendata_put:
	nfs4_opendata_put(opendata);
1837 1838
err_put_state_owner:
	nfs4_put_state_owner(sp);
L
Linus Torvalds 已提交
1839 1840 1841 1842 1843 1844
out_err:
	*res = NULL;
	return status;
}


1845
static struct nfs4_state *nfs4_do_open(struct inode *dir, struct dentry *dentry, fmode_t fmode, int flags, struct iattr *sattr, struct rpc_cred *cred)
L
Linus Torvalds 已提交
1846 1847 1848 1849 1850 1851
{
	struct nfs4_exception exception = { };
	struct nfs4_state *res;
	int status;

	do {
1852
		status = _nfs4_do_open(dir, dentry, fmode, flags, sattr, cred, &res);
L
Linus Torvalds 已提交
1853 1854 1855 1856 1857 1858 1859 1860
		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
1861
		 * doing an OPEN, we assume that nfs_increment_open_seqid() will
L
Linus Torvalds 已提交
1862 1863 1864 1865 1866
		 * 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) {
1867 1868 1869
			printk(KERN_WARNING "NFS: v4 server %s "
					" returned a bad sequence-id error!\n",
					NFS_SERVER(dir)->nfs_client->cl_hostname);
L
Linus Torvalds 已提交
1870 1871 1872
			exception.retry = 1;
			continue;
		}
1873 1874 1875 1876 1877 1878 1879 1880 1881 1882
		/*
		 * 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;
		}
1883 1884 1885 1886 1887
		if (status == -EAGAIN) {
			/* We must have found a delegation */
			exception.retry = 1;
			continue;
		}
L
Linus Torvalds 已提交
1888 1889 1890 1891 1892 1893
		res = ERR_PTR(nfs4_handle_exception(NFS_SERVER(dir),
					status, &exception));
	} while (exception.retry);
	return res;
}

1894 1895 1896
static int _nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
			    struct nfs_fattr *fattr, struct iattr *sattr,
			    struct nfs4_state *state)
L
Linus Torvalds 已提交
1897
{
1898
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
1899
        struct nfs_setattrargs  arg = {
1900
                .fh             = NFS_FH(inode),
L
Linus Torvalds 已提交
1901 1902 1903 1904 1905 1906 1907 1908 1909
                .iap            = sattr,
		.server		= server,
		.bitmask = server->attr_bitmask,
        };
        struct nfs_setattrres  res = {
		.fattr		= fattr,
		.server		= server,
        };
        struct rpc_message msg = {
1910 1911 1912 1913
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETATTR],
		.rpc_argp	= &arg,
		.rpc_resp	= &res,
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
1914
        };
1915
	unsigned long timestamp = jiffies;
1916
	int status;
L
Linus Torvalds 已提交
1917

1918
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
1919

1920 1921 1922
	if (nfs4_copy_delegation_stateid(&arg.stateid, inode)) {
		/* Use that stateid */
	} else if (state != NULL) {
1923
		nfs4_select_rw_stateid(&arg.stateid, state, current->files, current->tgid);
1924
	} else
1925
		nfs4_stateid_copy(&arg.stateid, &zero_stateid);
L
Linus Torvalds 已提交
1926

1927
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
1928 1929
	if (status == 0 && state != NULL)
		renew_lease(server, timestamp);
1930
	return status;
L
Linus Torvalds 已提交
1931 1932
}

1933 1934 1935
static int nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
			   struct nfs_fattr *fattr, struct iattr *sattr,
			   struct nfs4_state *state)
L
Linus Torvalds 已提交
1936
{
1937
	struct nfs_server *server = NFS_SERVER(inode);
1938 1939 1940
	struct nfs4_exception exception = {
		.state = state,
	};
L
Linus Torvalds 已提交
1941 1942 1943
	int err;
	do {
		err = nfs4_handle_exception(server,
1944
				_nfs4_do_setattr(inode, cred, fattr, sattr, state),
L
Linus Torvalds 已提交
1945 1946 1947 1948 1949 1950 1951 1952 1953 1954
				&exception);
	} while (exception.retry);
	return err;
}

struct nfs4_closedata {
	struct inode *inode;
	struct nfs4_state *state;
	struct nfs_closeargs arg;
	struct nfs_closeres res;
1955
	struct nfs_fattr fattr;
1956
	unsigned long timestamp;
F
Fred Isaman 已提交
1957 1958
	bool roc;
	u32 roc_barrier;
L
Linus Torvalds 已提交
1959 1960
};

1961
static void nfs4_free_closedata(void *data)
1962
{
1963 1964
	struct nfs4_closedata *calldata = data;
	struct nfs4_state_owner *sp = calldata->state->owner;
1965
	struct super_block *sb = calldata->state->inode->i_sb;
1966

F
Fred Isaman 已提交
1967 1968
	if (calldata->roc)
		pnfs_roc_release(calldata->state->inode);
1969 1970 1971
	nfs4_put_open_state(calldata->state);
	nfs_free_seqid(calldata->arg.seqid);
	nfs4_put_state_owner(sp);
1972
	nfs_sb_deactive(sb);
1973 1974 1975
	kfree(calldata);
}

1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987
static void nfs4_close_clear_stateid_flags(struct nfs4_state *state,
		fmode_t fmode)
{
	spin_lock(&state->owner->so_lock);
	if (!(fmode & FMODE_READ))
		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
	if (!(fmode & FMODE_WRITE))
		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
	clear_bit(NFS_O_RDWR_STATE, &state->flags);
	spin_unlock(&state->owner->so_lock);
}

1988
static void nfs4_close_done(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
1989
{
1990
	struct nfs4_closedata *calldata = data;
L
Linus Torvalds 已提交
1991 1992 1993
	struct nfs4_state *state = calldata->state;
	struct nfs_server *server = NFS_SERVER(calldata->inode);

1994
	dprintk("%s: begin!\n", __func__);
1995 1996
	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
		return;
L
Linus Torvalds 已提交
1997 1998 1999 2000 2001
        /* 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
2002 2003 2004
			if (calldata->roc)
				pnfs_roc_set_barrier(state->inode,
						     calldata->roc_barrier);
2005
			nfs_set_open_stateid(state, &calldata->res.stateid, 0);
2006
			renew_lease(server, calldata->timestamp);
2007 2008
			nfs4_close_clear_stateid_flags(state,
					calldata->arg.fmode);
L
Linus Torvalds 已提交
2009 2010
			break;
		case -NFS4ERR_STALE_STATEID:
2011 2012
		case -NFS4ERR_OLD_STATEID:
		case -NFS4ERR_BAD_STATEID:
L
Linus Torvalds 已提交
2013
		case -NFS4ERR_EXPIRED:
2014
			if (calldata->arg.fmode == 0)
2015
				break;
L
Linus Torvalds 已提交
2016
		default:
2017 2018
			if (nfs4_async_handle_error(task, server, state) == -EAGAIN)
				rpc_restart_call_prepare(task);
L
Linus Torvalds 已提交
2019
	}
2020
	nfs_release_seqid(calldata->arg.seqid);
2021
	nfs_refresh_inode(calldata->inode, calldata->res.fattr);
2022
	dprintk("%s: done, ret = %d!\n", __func__, task->tk_status);
L
Linus Torvalds 已提交
2023 2024
}

T
Trond Myklebust 已提交
2025
static void nfs4_close_prepare(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
2026
{
T
Trond Myklebust 已提交
2027
	struct nfs4_closedata *calldata = data;
2028
	struct nfs4_state *state = calldata->state;
2029
	int call_close = 0;
2030

2031
	dprintk("%s: begin!\n", __func__);
2032
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
2033
		return;
2034

2035 2036
	task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_DOWNGRADE];
	calldata->arg.fmode = FMODE_READ|FMODE_WRITE;
2037
	spin_lock(&state->owner->so_lock);
2038
	/* Calculate the change in open mode */
2039
	if (state->n_rdwr == 0) {
2040
		if (state->n_rdonly == 0) {
2041 2042 2043
			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;
2044 2045
		}
		if (state->n_wronly == 0) {
2046 2047 2048
			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;
2049
		}
2050
	}
2051
	spin_unlock(&state->owner->so_lock);
2052 2053

	if (!call_close) {
2054 2055
		/* Note: exit _without_ calling nfs4_close_done */
		task->tk_action = NULL;
2056
		goto out;
2057
	}
2058

F
Fred Isaman 已提交
2059
	if (calldata->arg.fmode == 0) {
2060
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE];
F
Fred Isaman 已提交
2061 2062 2063 2064
		if (calldata->roc &&
		    pnfs_roc_drain(calldata->inode, &calldata->roc_barrier)) {
			rpc_sleep_on(&NFS_SERVER(calldata->inode)->roc_rpcwaitq,
				     task, NULL);
2065
			goto out;
F
Fred Isaman 已提交
2066 2067
		}
	}
2068

2069
	nfs_fattr_init(calldata->res.fattr);
2070
	calldata->timestamp = jiffies;
2071
	if (nfs4_setup_sequence(NFS_SERVER(calldata->inode),
2072 2073 2074
				&calldata->arg.seq_args,
				&calldata->res.seq_res,
				task))
2075
		goto out;
2076
	rpc_call_start(task);
2077 2078
out:
	dprintk("%s: done!\n", __func__);
L
Linus Torvalds 已提交
2079 2080
}

2081
static const struct rpc_call_ops nfs4_close_ops = {
T
Trond Myklebust 已提交
2082
	.rpc_call_prepare = nfs4_close_prepare,
2083 2084 2085 2086
	.rpc_call_done = nfs4_close_done,
	.rpc_release = nfs4_free_closedata,
};

L
Linus Torvalds 已提交
2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097
/* 
 * 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!
 */
2098
int nfs4_do_close(struct nfs4_state *state, gfp_t gfp_mask, int wait, bool roc)
L
Linus Torvalds 已提交
2099
{
2100
	struct nfs_server *server = NFS_SERVER(state->inode);
L
Linus Torvalds 已提交
2101
	struct nfs4_closedata *calldata;
2102 2103
	struct nfs4_state_owner *sp = state->owner;
	struct rpc_task *task;
2104 2105 2106 2107
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
2108 2109
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
2110
		.rpc_message = &msg,
T
Trond Myklebust 已提交
2111
		.callback_ops = &nfs4_close_ops,
2112
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
2113 2114
		.flags = RPC_TASK_ASYNC,
	};
2115
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2116

2117
	calldata = kzalloc(sizeof(*calldata), gfp_mask);
L
Linus Torvalds 已提交
2118
	if (calldata == NULL)
2119
		goto out;
2120
	nfs41_init_sequence(&calldata->arg.seq_args, &calldata->res.seq_res, 1);
2121
	calldata->inode = state->inode;
L
Linus Torvalds 已提交
2122
	calldata->state = state;
2123
	calldata->arg.fh = NFS_FH(state->inode);
2124
	calldata->arg.stateid = &state->open_stateid;
L
Linus Torvalds 已提交
2125
	/* Serialization for the sequence id */
2126
	calldata->arg.seqid = nfs_alloc_seqid(&state->owner->so_seqid, gfp_mask);
2127 2128
	if (calldata->arg.seqid == NULL)
		goto out_free_calldata;
2129
	calldata->arg.fmode = 0;
2130
	calldata->arg.bitmask = server->cache_consistency_bitmask;
2131
	calldata->res.fattr = &calldata->fattr;
2132
	calldata->res.seqid = calldata->arg.seqid;
2133
	calldata->res.server = server;
F
Fred Isaman 已提交
2134
	calldata->roc = roc;
2135
	nfs_sb_active(calldata->inode->i_sb);
2136

2137 2138
	msg.rpc_argp = &calldata->arg;
	msg.rpc_resp = &calldata->res;
T
Trond Myklebust 已提交
2139 2140
	task_setup_data.callback_data = calldata;
	task = rpc_run_task(&task_setup_data);
2141 2142
	if (IS_ERR(task))
		return PTR_ERR(task);
2143 2144 2145
	status = 0;
	if (wait)
		status = rpc_wait_for_completion_task(task);
2146
	rpc_put_task(task);
2147
	return status;
2148 2149 2150
out_free_calldata:
	kfree(calldata);
out:
F
Fred Isaman 已提交
2151 2152
	if (roc)
		pnfs_roc_release(state->inode);
2153 2154
	nfs4_put_open_state(state);
	nfs4_put_state_owner(sp);
2155
	return status;
L
Linus Torvalds 已提交
2156 2157
}

2158
static struct inode *
2159
nfs4_atomic_open(struct inode *dir, struct nfs_open_context *ctx, int open_flags, struct iattr *attr)
L
Linus Torvalds 已提交
2160 2161 2162
{
	struct nfs4_state *state;

T
Trond Myklebust 已提交
2163
	/* Protect against concurrent sillydeletes */
2164
	state = nfs4_do_open(dir, ctx->dentry, ctx->mode, open_flags, attr, ctx->cred);
2165 2166
	if (IS_ERR(state))
		return ERR_CAST(state);
2167
	ctx->state = state;
2168
	return igrab(state->inode);
L
Linus Torvalds 已提交
2169 2170
}

2171
static void nfs4_close_context(struct nfs_open_context *ctx, int is_sync)
T
Trond Myklebust 已提交
2172 2173 2174 2175
{
	if (ctx->state == NULL)
		return;
	if (is_sync)
2176
		nfs4_close_sync(ctx->state, ctx->mode);
T
Trond Myklebust 已提交
2177
	else
2178
		nfs4_close_state(ctx->state, ctx->mode);
T
Trond Myklebust 已提交
2179
}
L
Linus Torvalds 已提交
2180 2181 2182

static int _nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
{
B
Benny Halevy 已提交
2183 2184 2185
	struct nfs4_server_caps_arg args = {
		.fhandle = fhandle,
	};
L
Linus Torvalds 已提交
2186 2187 2188
	struct nfs4_server_caps_res res = {};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SERVER_CAPS],
B
Benny Halevy 已提交
2189
		.rpc_argp = &args,
L
Linus Torvalds 已提交
2190 2191 2192 2193
		.rpc_resp = &res,
	};
	int status;

2194
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
2195 2196
	if (status == 0) {
		memcpy(server->attr_bitmask, res.attr_bitmask, sizeof(server->attr_bitmask));
2197 2198 2199 2200 2201
		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
2202 2203 2204 2205 2206 2207
		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;
2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224
		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;

2225 2226 2227
		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
2228
		server->acl_bitmask = res.acl_bitmask;
2229
		server->fh_expire_type = res.fh_expire_type;
L
Linus Torvalds 已提交
2230
	}
A
Andy Adamson 已提交
2231

L
Linus Torvalds 已提交
2232 2233 2234
	return status;
}

2235
int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
L
Linus Torvalds 已提交
2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254
{
	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)
{
	struct nfs4_lookup_root_arg args = {
		.bitmask = nfs4_fattr_bitmap,
	};
	struct nfs4_lookup_res res = {
		.server = server,
2255
		.fattr = info->fattr,
L
Linus Torvalds 已提交
2256 2257 2258 2259 2260 2261 2262
		.fh = fhandle,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP_ROOT],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
2263

2264
	nfs_fattr_init(info->fattr);
2265
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
2266 2267 2268 2269 2270 2271 2272 2273
}

static int nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle,
		struct nfs_fsinfo *info)
{
	struct nfs4_exception exception = { };
	int err;
	do {
2274 2275 2276 2277 2278 2279 2280 2281
		err = _nfs4_lookup_root(server, fhandle, info);
		switch (err) {
		case 0:
		case -NFS4ERR_WRONGSEC:
			break;
		default:
			err = nfs4_handle_exception(server, err, &exception);
		}
L
Linus Torvalds 已提交
2282 2283 2284 2285
	} while (exception.retry);
	return err;
}

2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301
static int nfs4_lookup_root_sec(struct nfs_server *server, struct nfs_fh *fhandle,
				struct nfs_fsinfo *info, rpc_authflavor_t flavor)
{
	struct rpc_auth *auth;
	int ret;

	auth = rpcauth_create(flavor, server->client);
	if (!auth) {
		ret = -EIO;
		goto out;
	}
	ret = nfs4_lookup_root(server, fhandle, info);
out:
	return ret;
}

2302
static int nfs4_find_root_sec(struct nfs_server *server, struct nfs_fh *fhandle,
2303
			      struct nfs_fsinfo *info)
L
Linus Torvalds 已提交
2304
{
2305
	int i, len, status = 0;
2306
	rpc_authflavor_t flav_array[NFS_MAX_SECFLAVORS];
L
Linus Torvalds 已提交
2307

2308 2309 2310
	len = gss_mech_list_pseudoflavors(&flav_array[0]);
	flav_array[len] = RPC_AUTH_NULL;
	len += 1;
2311 2312 2313

	for (i = 0; i < len; i++) {
		status = nfs4_lookup_root_sec(server, fhandle, info, flav_array[i]);
2314
		if (status == -NFS4ERR_WRONGSEC || status == -EACCES)
2315 2316
			continue;
		break;
2317
	}
2318 2319 2320 2321 2322 2323 2324 2325 2326
	/*
	 * -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;
2327 2328 2329 2330 2331 2332 2333 2334 2335
	return status;
}

/*
 * get the file handle for the "/" directory on the server
 */
static int nfs4_proc_get_root(struct nfs_server *server, struct nfs_fh *fhandle,
			      struct nfs_fsinfo *info)
{
2336
	int minor_version = server->nfs_client->cl_minorversion;
2337
	int status = nfs4_lookup_root(server, fhandle, info);
2338 2339 2340 2341 2342
	if ((status == -NFS4ERR_WRONGSEC) && !(server->flags & NFS_MOUNT_SECFLAVOUR))
		/*
		 * A status of -NFS4ERR_WRONGSEC will be mapped to -EPERM
		 * by nfs4_map_errors() as this function exits.
		 */
2343
		status = nfs_v4_minor_ops[minor_version]->find_root_sec(server, fhandle, info);
L
Linus Torvalds 已提交
2344 2345 2346 2347
	if (status == 0)
		status = nfs4_server_capabilities(server, fhandle);
	if (status == 0)
		status = nfs4_do_fsinfo(server, fhandle, info);
2348
	return nfs4_map_errors(status);
L
Linus Torvalds 已提交
2349 2350
}

M
Manoj Naik 已提交
2351 2352 2353 2354 2355
/*
 * 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
 */
2356 2357
static int nfs4_get_referral(struct inode *dir, const struct qstr *name,
			     struct nfs_fattr *fattr, struct nfs_fh *fhandle)
M
Manoj Naik 已提交
2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369
{
	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;

2370
	status = nfs4_proc_fs_locations(dir, name, locations, page);
M
Manoj Naik 已提交
2371 2372 2373 2374
	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)) {
2375 2376
		dprintk("%s: server did not return a different fsid for"
			" a referral at %s\n", __func__, name->name);
M
Manoj Naik 已提交
2377 2378 2379
		status = -EIO;
		goto out;
	}
2380 2381
	/* Fixup attributes for the nfs_lookup() call to nfs_fhget() */
	nfs_fixup_referral_attributes(&locations->fattr);
M
Manoj Naik 已提交
2382

2383
	/* replace the lookup nfs_fattr with the locations nfs_fattr */
M
Manoj Naik 已提交
2384 2385 2386 2387 2388
	memcpy(fattr, &locations->fattr, sizeof(struct nfs_fattr));
	memset(fhandle, 0, sizeof(struct nfs_fh));
out:
	if (page)
		__free_page(page);
2389
	kfree(locations);
M
Manoj Naik 已提交
2390 2391 2392
	return status;
}

L
Linus Torvalds 已提交
2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408
static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fattr *fattr)
{
	struct nfs4_getattr_arg args = {
		.fh = fhandle,
		.bitmask = server->attr_bitmask,
	};
	struct nfs4_getattr_res res = {
		.fattr = fattr,
		.server = server,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETATTR],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
	
2409
	nfs_fattr_init(fattr);
2410
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445
}

static int nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fattr *fattr)
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(server,
				_nfs4_proc_getattr(server, fhandle, fattr),
				&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)
{
2446
	struct inode *inode = dentry->d_inode;
2447
	struct rpc_cred *cred = NULL;
2448
	struct nfs4_state *state = NULL;
L
Linus Torvalds 已提交
2449 2450
	int status;

B
Benny Halevy 已提交
2451 2452 2453
	if (pnfs_ld_layoutret_on_setattr(inode))
		pnfs_return_layout(inode);

2454
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
2455
	
2456
	/* Search for an existing open(O_WRITE) file */
2457 2458 2459 2460
	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
2461 2462 2463 2464
		if (ctx) {
			cred = ctx->cred;
			state = ctx->state;
		}
2465
	}
2466

2467 2468 2469 2470
	/* Deal with open(O_TRUNC) */
	if (sattr->ia_valid & ATTR_OPEN)
		sattr->ia_valid &= ~(ATTR_MTIME|ATTR_CTIME|ATTR_OPEN);

2471
	status = nfs4_do_setattr(inode, cred, fattr, sattr, state);
2472 2473
	if (status == 0)
		nfs_setattr_update_inode(inode, sattr);
L
Linus Torvalds 已提交
2474 2475 2476
	return status;
}

2477 2478 2479
static int _nfs4_proc_lookup(struct rpc_clnt *clnt, struct inode *dir,
		const struct qstr *name, struct nfs_fh *fhandle,
		struct nfs_fattr *fattr)
D
David Howells 已提交
2480
{
2481
	struct nfs_server *server = NFS_SERVER(dir);
D
David Howells 已提交
2482 2483 2484
	int		       status;
	struct nfs4_lookup_arg args = {
		.bitmask = server->attr_bitmask,
2485
		.dir_fh = NFS_FH(dir),
D
David Howells 已提交
2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500
		.name = name,
	};
	struct nfs4_lookup_res res = {
		.server = server,
		.fattr = fattr,
		.fh = fhandle,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};

	nfs_fattr_init(fattr);

L
Linus Torvalds 已提交
2501
	dprintk("NFS call  lookup %s\n", name->name);
2502
	status = nfs4_call_sync(clnt, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
2503 2504 2505 2506
	dprintk("NFS reply lookup: %d\n", status);
	return status;
}

2507 2508 2509 2510 2511 2512 2513 2514 2515 2516
void nfs_fixup_secinfo_attributes(struct nfs_fattr *fattr, struct nfs_fh *fh)
{
	memset(fh, 0, sizeof(struct nfs_fh));
	fattr->fsid.major = 1;
	fattr->valid |= NFS_ATTR_FATTR_TYPE | NFS_ATTR_FATTR_MODE |
		NFS_ATTR_FATTR_NLINK | NFS_ATTR_FATTR_FSID | NFS_ATTR_FATTR_MOUNTPOINT;
	fattr->mode = S_IFDIR | S_IRUGO | S_IXUGO;
	fattr->nlink = 2;
}

2517 2518
static int nfs4_proc_lookup(struct rpc_clnt *clnt, struct inode *dir, struct qstr *name,
			    struct nfs_fh *fhandle, struct nfs_fattr *fattr)
L
Linus Torvalds 已提交
2519 2520 2521 2522
{
	struct nfs4_exception exception = { };
	int err;
	do {
2523 2524 2525 2526
		int status;

		status = _nfs4_proc_lookup(clnt, dir, name, fhandle, fattr);
		switch (status) {
2527 2528
		case -NFS4ERR_BADNAME:
			return -ENOENT;
2529
		case -NFS4ERR_MOVED:
2530
			return nfs4_get_referral(dir, name, fattr, fhandle);
2531
		case -NFS4ERR_WRONGSEC:
2532
			nfs_fixup_secinfo_attributes(fattr, fhandle);
2533 2534 2535
		}
		err = nfs4_handle_exception(NFS_SERVER(dir),
				status, &exception);
L
Linus Torvalds 已提交
2536 2537 2538 2539 2540 2541
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
{
2542
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
2543 2544
	struct nfs4_accessargs args = {
		.fh = NFS_FH(inode),
2545
		.bitmask = server->cache_consistency_bitmask,
2546 2547 2548
	};
	struct nfs4_accessres res = {
		.server = server,
L
Linus Torvalds 已提交
2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574
	};
	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;
	int status;

	/*
	 * 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;
	}
2575 2576 2577 2578 2579

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

2580
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
2581 2582 2583 2584 2585 2586 2587 2588
	if (!status) {
		entry->mask = 0;
		if (res.access & NFS4_ACCESS_READ)
			entry->mask |= MAY_READ;
		if (res.access & (NFS4_ACCESS_MODIFY | NFS4_ACCESS_EXTEND | NFS4_ACCESS_DELETE))
			entry->mask |= MAY_WRITE;
		if (res.access & (NFS4_ACCESS_LOOKUP|NFS4_ACCESS_EXECUTE))
			entry->mask |= MAY_EXEC;
2589
		nfs_refresh_inode(inode, res.fattr);
L
Linus Torvalds 已提交
2590
	}
2591
	nfs_free_fattr(res.fattr);
L
Linus Torvalds 已提交
2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639
	return status;
}

static int nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(NFS_SERVER(inode),
				_nfs4_proc_access(inode, entry),
				&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
 * we get the post-operation mtime and size.  This means that
 * we can't use xdr_encode_pages() as written: we need a variant
 * of it which would leave room in the 'tail' iovec.
 *
 * 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
2640
	struct nfs4_readlink_res res;
L
Linus Torvalds 已提交
2641 2642 2643
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READLINK],
		.rpc_argp = &args,
B
Benny Halevy 已提交
2644
		.rpc_resp = &res,
L
Linus Torvalds 已提交
2645 2646
	};

2647
	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
2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679
}

static int nfs4_proc_readlink(struct inode *inode, struct page *page,
		unsigned int pgbase, unsigned int pglen)
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(NFS_SERVER(inode),
				_nfs4_proc_readlink(inode, page, pgbase, pglen),
				&exception);
	} while (exception.retry);
	return err;
}

/*
 * Got race?
 * We will need to arrange for the VFS layer to provide an atomic open.
 * Until then, this create/open method is prone to inefficiency and race
 * conditions due to the lookup, create, and open VFS calls from sys_open()
 * placed on the wire.
 *
 * Given the above sorry state of affairs, I'm simply sending an OPEN.
 * The file will be opened again in the subsequent VFS open call
 * (nfs4_proc_file_open).
 *
 * The open for read will just hang around to be used by any process that
 * opens the file O_RDONLY. This will all be resolved with the VFS changes.
 */

static int
nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
2680
                 int flags, struct nfs_open_context *ctx)
L
Linus Torvalds 已提交
2681
{
2682
	struct dentry *de = dentry;
L
Linus Torvalds 已提交
2683
	struct nfs4_state *state;
2684 2685
	struct rpc_cred *cred = NULL;
	fmode_t fmode = 0;
L
Linus Torvalds 已提交
2686 2687
	int status = 0;

2688 2689
	if (ctx != NULL) {
		cred = ctx->cred;
2690
		de = ctx->dentry;
2691
		fmode = ctx->mode;
L
Linus Torvalds 已提交
2692
	}
A
Aneesh Kumar K.V 已提交
2693
	sattr->ia_mode &= ~current_umask();
2694
	state = nfs4_do_open(dir, de, fmode, flags, sattr, cred);
2695
	d_drop(dentry);
L
Linus Torvalds 已提交
2696 2697
	if (IS_ERR(state)) {
		status = PTR_ERR(state);
2698
		goto out;
L
Linus Torvalds 已提交
2699
	}
2700
	d_add(dentry, igrab(state->inode));
2701
	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
2702 2703
	if (ctx != NULL)
		ctx->state = state;
2704
	else
2705
		nfs4_close_sync(state, fmode);
L
Linus Torvalds 已提交
2706 2707 2708 2709 2710 2711
out:
	return status;
}

static int _nfs4_proc_remove(struct inode *dir, struct qstr *name)
{
2712
	struct nfs_server *server = NFS_SERVER(dir);
2713
	struct nfs_removeargs args = {
L
Linus Torvalds 已提交
2714
		.fh = NFS_FH(dir),
2715 2716
		.name.len = name->len,
		.name.name = name->name,
2717 2718
		.bitmask = server->attr_bitmask,
	};
2719
	struct nfs_removeres res = {
2720
		.server = server,
L
Linus Torvalds 已提交
2721 2722
	};
	struct rpc_message msg = {
2723 2724 2725
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE],
		.rpc_argp = &args,
		.rpc_resp = &res,
L
Linus Torvalds 已提交
2726
	};
2727 2728 2729 2730 2731
	int status = -ENOMEM;

	res.dir_attr = nfs_alloc_fattr();
	if (res.dir_attr == NULL)
		goto out;
L
Linus Torvalds 已提交
2732

2733
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 1);
2734 2735
	if (status == 0) {
		update_changeattr(dir, &res.cinfo);
2736
		nfs_post_op_update_inode(dir, res.dir_attr);
2737
	}
2738 2739
	nfs_free_fattr(res.dir_attr);
out:
L
Linus Torvalds 已提交
2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754
	return status;
}

static int nfs4_proc_remove(struct inode *dir, struct qstr *name)
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(NFS_SERVER(dir),
				_nfs4_proc_remove(dir, name),
				&exception);
	} while (exception.retry);
	return err;
}

2755
static void nfs4_proc_unlink_setup(struct rpc_message *msg, struct inode *dir)
L
Linus Torvalds 已提交
2756
{
2757 2758 2759
	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
2760

2761
	args->bitmask = server->cache_consistency_bitmask;
2762
	res->server = server;
L
Linus Torvalds 已提交
2763
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE];
2764
	nfs41_init_sequence(&args->seq_args, &res->seq_res, 1);
L
Linus Torvalds 已提交
2765 2766
}

2767
static int nfs4_proc_unlink_done(struct rpc_task *task, struct inode *dir)
L
Linus Torvalds 已提交
2768
{
2769 2770
	struct nfs_removeres *res = task->tk_msg.rpc_resp;

2771 2772
	if (!nfs4_sequence_done(task, &res->seq_res))
		return 0;
2773
	if (nfs4_async_handle_error(task, res->server, NULL) == -EAGAIN)
2774 2775
		return 0;
	update_changeattr(dir, &res->cinfo);
2776
	nfs_post_op_update_inode(dir, res->dir_attr);
2777
	return 1;
L
Linus Torvalds 已提交
2778 2779
}

2780 2781 2782 2783 2784 2785 2786 2787 2788
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];
	arg->bitmask = server->attr_bitmask;
	res->server = server;
2789
	nfs41_init_sequence(&arg->seq_args, &res->seq_res, 1);
2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808
}

static int nfs4_proc_rename_done(struct rpc_task *task, struct inode *old_dir,
				 struct inode *new_dir)
{
	struct nfs_renameres *res = task->tk_msg.rpc_resp;

	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);
	nfs_post_op_update_inode(old_dir, res->old_fattr);
	update_changeattr(new_dir, &res->new_cinfo);
	nfs_post_op_update_inode(new_dir, res->new_fattr);
	return 1;
}

L
Linus Torvalds 已提交
2809 2810 2811
static int _nfs4_proc_rename(struct inode *old_dir, struct qstr *old_name,
		struct inode *new_dir, struct qstr *new_name)
{
2812
	struct nfs_server *server = NFS_SERVER(old_dir);
2813
	struct nfs_renameargs arg = {
L
Linus Torvalds 已提交
2814 2815 2816 2817
		.old_dir = NFS_FH(old_dir),
		.new_dir = NFS_FH(new_dir),
		.old_name = old_name,
		.new_name = new_name,
2818 2819
		.bitmask = server->attr_bitmask,
	};
2820
	struct nfs_renameres res = {
2821
		.server = server,
L
Linus Torvalds 已提交
2822 2823 2824 2825 2826 2827
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RENAME],
		.rpc_argp = &arg,
		.rpc_resp = &res,
	};
2828
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2829
	
2830 2831 2832 2833
	res.old_fattr = nfs_alloc_fattr();
	res.new_fattr = nfs_alloc_fattr();
	if (res.old_fattr == NULL || res.new_fattr == NULL)
		goto out;
L
Linus Torvalds 已提交
2834

2835
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
L
Linus Torvalds 已提交
2836 2837
	if (!status) {
		update_changeattr(old_dir, &res.old_cinfo);
2838
		nfs_post_op_update_inode(old_dir, res.old_fattr);
L
Linus Torvalds 已提交
2839
		update_changeattr(new_dir, &res.new_cinfo);
2840
		nfs_post_op_update_inode(new_dir, res.new_fattr);
L
Linus Torvalds 已提交
2841
	}
2842 2843 2844
out:
	nfs_free_fattr(res.new_fattr);
	nfs_free_fattr(res.old_fattr);
L
Linus Torvalds 已提交
2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863
	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 {
		err = nfs4_handle_exception(NFS_SERVER(old_dir),
				_nfs4_proc_rename(old_dir, old_name,
					new_dir, new_name),
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_link(struct inode *inode, struct inode *dir, struct qstr *name)
{
2864
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
2865 2866 2867 2868
	struct nfs4_link_arg arg = {
		.fh     = NFS_FH(inode),
		.dir_fh = NFS_FH(dir),
		.name   = name,
2869 2870 2871 2872
		.bitmask = server->attr_bitmask,
	};
	struct nfs4_link_res res = {
		.server = server,
L
Linus Torvalds 已提交
2873 2874 2875 2876
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LINK],
		.rpc_argp = &arg,
2877
		.rpc_resp = &res,
L
Linus Torvalds 已提交
2878
	};
2879 2880 2881 2882 2883 2884
	int status = -ENOMEM;

	res.fattr = nfs_alloc_fattr();
	res.dir_attr = nfs_alloc_fattr();
	if (res.fattr == NULL || res.dir_attr == NULL)
		goto out;
L
Linus Torvalds 已提交
2885

2886
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
2887 2888 2889
	if (!status) {
		update_changeattr(dir, &res.cinfo);
		nfs_post_op_update_inode(dir, res.dir_attr);
2890
		nfs_post_op_update_inode(inode, res.fattr);
2891
	}
2892 2893 2894
out:
	nfs_free_fattr(res.dir_attr);
	nfs_free_fattr(res.fattr);
L
Linus Torvalds 已提交
2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909
	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;
}

2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948
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;
	struct nfs_fattr dir_fattr;
};

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);

		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;
		data->arg.bitmask = server->attr_bitmask;
		data->res.server = server;
		data->res.fh = &data->fh;
		data->res.fattr = &data->fattr;
		data->res.dir_fattr = &data->dir_fattr;
		nfs_fattr_init(data->res.fattr);
		nfs_fattr_init(data->res.dir_fattr);
	}
	return data;
}

static int nfs4_do_create(struct inode *dir, struct dentry *dentry, struct nfs4_createdata *data)
{
2949
	int status = nfs4_call_sync(NFS_SERVER(dir)->client, NFS_SERVER(dir), &data->msg,
2950
				    &data->arg.seq_args, &data->res.seq_res, 1);
2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963
	if (status == 0) {
		update_changeattr(dir, &data->res.dir_cinfo);
		nfs_post_op_update_inode(dir, data->res.dir_fattr);
		status = nfs_instantiate(dentry, data->res.fh, data->res.fattr);
	}
	return status;
}

static void nfs4_free_createdata(struct nfs4_createdata *data)
{
	kfree(data);
}

2964
static int _nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
2965
		struct page *page, unsigned int len, struct iattr *sattr)
L
Linus Torvalds 已提交
2966
{
2967 2968
	struct nfs4_createdata *data;
	int status = -ENAMETOOLONG;
L
Linus Torvalds 已提交
2969

2970
	if (len > NFS4_MAXPATHLEN)
2971
		goto out;
2972

2973 2974 2975 2976 2977 2978 2979 2980
	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;
L
Linus Torvalds 已提交
2981
	
2982 2983 2984 2985
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
2986 2987 2988
	return status;
}

2989
static int nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
2990
		struct page *page, unsigned int len, struct iattr *sattr)
L
Linus Torvalds 已提交
2991 2992 2993 2994 2995
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(NFS_SERVER(dir),
2996 2997
				_nfs4_proc_symlink(dir, dentry, page,
							len, sattr),
L
Linus Torvalds 已提交
2998 2999 3000 3001 3002 3003 3004 3005
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr)
{
3006 3007
	struct nfs4_createdata *data;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
3008

3009 3010 3011 3012 3013 3014 3015 3016
	data = nfs4_alloc_createdata(dir, &dentry->d_name, sattr, NF4DIR);
	if (data == NULL)
		goto out;

	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
3017 3018 3019 3020 3021 3022 3023 3024
	return status;
}

static int nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr)
{
	struct nfs4_exception exception = { };
	int err;
A
Aneesh Kumar K.V 已提交
3025 3026

	sattr->ia_mode &= ~current_umask();
L
Linus Torvalds 已提交
3027 3028 3029 3030 3031 3032 3033 3034 3035
	do {
		err = nfs4_handle_exception(NFS_SERVER(dir),
				_nfs4_proc_mkdir(dir, dentry, sattr),
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
B
Bryan Schumaker 已提交
3036
		u64 cookie, struct page **pages, unsigned int count, int plus)
L
Linus Torvalds 已提交
3037 3038 3039 3040
{
	struct inode		*dir = dentry->d_inode;
	struct nfs4_readdir_arg args = {
		.fh = NFS_FH(dir),
B
Bryan Schumaker 已提交
3041
		.pages = pages,
L
Linus Torvalds 已提交
3042 3043
		.pgbase = 0,
		.count = count,
3044
		.bitmask = NFS_SERVER(dentry->d_inode)->attr_bitmask,
B
Bryan Schumaker 已提交
3045
		.plus = plus,
L
Linus Torvalds 已提交
3046 3047 3048 3049 3050 3051 3052 3053 3054 3055
	};
	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;

3056
	dprintk("%s: dentry = %s/%s, cookie = %Lu\n", __func__,
3057 3058 3059
			dentry->d_parent->d_name.name,
			dentry->d_name.name,
			(unsigned long long)cookie);
L
Linus Torvalds 已提交
3060 3061
	nfs4_setup_readdir(cookie, NFS_COOKIEVERF(dir), dentry, &args);
	res.pgbase = args.pgbase;
3062
	status = nfs4_call_sync(NFS_SERVER(dir)->client, NFS_SERVER(dir), &msg, &args.seq_args, &res.seq_res, 0);
3063
	if (status >= 0) {
L
Linus Torvalds 已提交
3064
		memcpy(NFS_COOKIEVERF(dir), res.verifier.data, NFS4_VERIFIER_SIZE);
3065 3066
		status += args.pgbase;
	}
3067 3068 3069

	nfs_invalidate_atime(dir);

3070
	dprintk("%s: returns %d\n", __func__, status);
L
Linus Torvalds 已提交
3071 3072 3073 3074
	return status;
}

static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
B
Bryan Schumaker 已提交
3075
		u64 cookie, struct page **pages, unsigned int count, int plus)
L
Linus Torvalds 已提交
3076 3077 3078 3079 3080 3081
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(NFS_SERVER(dentry->d_inode),
				_nfs4_proc_readdir(dentry, cred, cookie,
B
Bryan Schumaker 已提交
3082
					pages, count, plus),
L
Linus Torvalds 已提交
3083 3084 3085 3086 3087 3088 3089 3090
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr, dev_t rdev)
{
3091 3092 3093
	struct nfs4_createdata *data;
	int mode = sattr->ia_mode;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
3094 3095 3096

	BUG_ON(!(sattr->ia_valid & ATTR_MODE));
	BUG_ON(!S_ISFIFO(mode) && !S_ISBLK(mode) && !S_ISCHR(mode) && !S_ISSOCK(mode));
3097 3098 3099 3100 3101

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

L
Linus Torvalds 已提交
3102
	if (S_ISFIFO(mode))
3103
		data->arg.ftype = NF4FIFO;
L
Linus Torvalds 已提交
3104
	else if (S_ISBLK(mode)) {
3105 3106 3107
		data->arg.ftype = NF4BLK;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
L
Linus Torvalds 已提交
3108 3109
	}
	else if (S_ISCHR(mode)) {
3110 3111 3112
		data->arg.ftype = NF4CHR;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
L
Linus Torvalds 已提交
3113 3114
	}
	
3115 3116 3117 3118
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
3119 3120 3121 3122 3123 3124 3125 3126
	return status;
}

static int nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr, dev_t rdev)
{
	struct nfs4_exception exception = { };
	int err;
A
Aneesh Kumar K.V 已提交
3127 3128

	sattr->ia_mode &= ~current_umask();
L
Linus Torvalds 已提交
3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143
	do {
		err = nfs4_handle_exception(NFS_SERVER(dir),
				_nfs4_proc_mknod(dir, dentry, sattr, rdev),
				&exception);
	} while (exception.retry);
	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
3144 3145 3146
	struct nfs4_statfs_res res = {
		.fsstat = fsstat,
	};
L
Linus Torvalds 已提交
3147 3148 3149
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_STATFS],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3150
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3151 3152
	};

3153
	nfs_fattr_init(fsstat->fattr);
3154
	return  nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175
}

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
3176 3177 3178
	struct nfs4_fsinfo_res res = {
		.fsinfo = fsinfo,
	};
L
Linus Torvalds 已提交
3179 3180 3181
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FSINFO],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3182
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3183 3184
	};

3185
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202
}

static int nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo)
{
	struct nfs4_exception exception = { };
	int err;

	do {
		err = nfs4_handle_exception(server,
				_nfs4_do_fsinfo(server, fhandle, fsinfo),
				&exception);
	} while (exception.retry);
	return err;
}

static int nfs4_proc_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo)
{
3203
	nfs_fattr_init(fsinfo->fattr);
L
Linus Torvalds 已提交
3204 3205 3206 3207 3208 3209 3210 3211 3212 3213
	return nfs4_do_fsinfo(server, fhandle, fsinfo);
}

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
3214 3215 3216
	struct nfs4_pathconf_res res = {
		.pathconf = pathconf,
	};
L
Linus Torvalds 已提交
3217 3218 3219
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_PATHCONF],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3220
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3221 3222 3223 3224 3225 3226 3227 3228
	};

	/* 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;
	}

3229
	nfs_fattr_init(pathconf->fattr);
3230
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
L
Linus Torvalds 已提交
3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246
}

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;
}

3247 3248 3249 3250 3251
void __nfs4_read_done_cb(struct nfs_read_data *data)
{
	nfs_invalidate_atime(data->inode);
}

3252
static int nfs4_read_done_cb(struct rpc_task *task, struct nfs_read_data *data)
L
Linus Torvalds 已提交
3253
{
T
Trond Myklebust 已提交
3254
	struct nfs_server *server = NFS_SERVER(data->inode);
L
Linus Torvalds 已提交
3255

3256
	if (nfs4_async_handle_error(task, server, data->args.context->state) == -EAGAIN) {
3257
		rpc_restart_call_prepare(task);
T
Trond Myklebust 已提交
3258
		return -EAGAIN;
L
Linus Torvalds 已提交
3259
	}
3260

3261
	__nfs4_read_done_cb(data);
L
Linus Torvalds 已提交
3262
	if (task->tk_status > 0)
T
Trond Myklebust 已提交
3263 3264
		renew_lease(server, data->timestamp);
	return 0;
L
Linus Torvalds 已提交
3265 3266
}

3267 3268 3269 3270 3271 3272 3273 3274
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;

3275 3276
	return data->read_done_cb ? data->read_done_cb(task, data) :
				    nfs4_read_done_cb(task, data);
3277 3278
}

3279
static void nfs4_proc_read_setup(struct nfs_read_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
3280 3281
{
	data->timestamp   = jiffies;
3282
	data->read_done_cb = nfs4_read_done_cb;
3283
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READ];
3284
	nfs41_init_sequence(&data->args.seq_args, &data->res.seq_res, 0);
L
Linus Torvalds 已提交
3285 3286
}

3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302
/* Reset the the nfs_read_data to send the read to the MDS. */
void nfs4_reset_read(struct rpc_task *task, struct nfs_read_data *data)
{
	dprintk("%s Reset task for i/o through\n", __func__);
	put_lseg(data->lseg);
	data->lseg = NULL;
	/* offsets will differ in the dense stripe case */
	data->args.offset = data->mds_offset;
	data->ds_clp = NULL;
	data->args.fh     = NFS_FH(data->inode);
	data->read_done_cb = nfs4_read_done_cb;
	task->tk_ops = data->mds_ops;
	rpc_task_reset_client(task, NFS_CLIENT(data->inode));
}
EXPORT_SYMBOL_GPL(nfs4_reset_read);

3303
static int nfs4_write_done_cb(struct rpc_task *task, struct nfs_write_data *data)
L
Linus Torvalds 已提交
3304 3305 3306
{
	struct inode *inode = data->inode;
	
3307
	if (nfs4_async_handle_error(task, NFS_SERVER(inode), data->args.context->state) == -EAGAIN) {
3308
		rpc_restart_call_prepare(task);
3309
		return -EAGAIN;
L
Linus Torvalds 已提交
3310
	}
3311
	if (task->tk_status >= 0) {
L
Linus Torvalds 已提交
3312
		renew_lease(NFS_SERVER(inode), data->timestamp);
3313
		nfs_post_op_update_inode_force_wcc(inode, data->res.fattr);
3314
	}
3315
	return 0;
L
Linus Torvalds 已提交
3316 3317
}

3318 3319 3320 3321
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;
3322 3323
	return data->write_done_cb ? data->write_done_cb(task, data) :
		nfs4_write_done_cb(task, data);
3324 3325
}

3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342
/* Reset the the nfs_write_data to send the write to the MDS. */
void nfs4_reset_write(struct rpc_task *task, struct nfs_write_data *data)
{
	dprintk("%s Reset task for i/o through\n", __func__);
	put_lseg(data->lseg);
	data->lseg          = NULL;
	data->ds_clp        = NULL;
	data->write_done_cb = nfs4_write_done_cb;
	data->args.fh       = NFS_FH(data->inode);
	data->args.bitmask  = data->res.server->cache_consistency_bitmask;
	data->args.offset   = data->mds_offset;
	data->res.fattr     = &data->fattr;
	task->tk_ops        = data->mds_ops;
	rpc_task_reset_client(task, NFS_CLIENT(data->inode));
}
EXPORT_SYMBOL_GPL(nfs4_reset_write);

3343
static void nfs4_proc_write_setup(struct nfs_write_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
3344
{
3345 3346
	struct nfs_server *server = NFS_SERVER(data->inode);

3347 3348 3349 3350 3351
	if (data->lseg) {
		data->args.bitmask = NULL;
		data->res.fattr = NULL;
	} else
		data->args.bitmask = server->cache_consistency_bitmask;
3352 3353
	if (!data->write_done_cb)
		data->write_done_cb = nfs4_write_done_cb;
3354
	data->res.server = server;
L
Linus Torvalds 已提交
3355 3356
	data->timestamp   = jiffies;

3357
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_WRITE];
3358
	nfs41_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
L
Linus Torvalds 已提交
3359 3360
}

3361
static int nfs4_commit_done_cb(struct rpc_task *task, struct nfs_write_data *data)
L
Linus Torvalds 已提交
3362 3363
{
	struct inode *inode = data->inode;
3364

3365
	if (nfs4_async_handle_error(task, NFS_SERVER(inode), NULL) == -EAGAIN) {
3366
		rpc_restart_call_prepare(task);
3367
		return -EAGAIN;
L
Linus Torvalds 已提交
3368
	}
3369
	nfs_refresh_inode(inode, data->res.fattr);
3370
	return 0;
L
Linus Torvalds 已提交
3371 3372
}

3373 3374 3375 3376 3377 3378 3379
static int nfs4_commit_done(struct rpc_task *task, struct nfs_write_data *data)
{
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return -EAGAIN;
	return data->write_done_cb(task, data);
}

3380
static void nfs4_proc_commit_setup(struct nfs_write_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
3381
{
3382
	struct nfs_server *server = NFS_SERVER(data->inode);
3383 3384 3385 3386 3387 3388

	if (data->lseg) {
		data->args.bitmask = NULL;
		data->res.fattr = NULL;
	} else
		data->args.bitmask = server->cache_consistency_bitmask;
3389 3390
	if (!data->write_done_cb)
		data->write_done_cb = nfs4_commit_done_cb;
3391
	data->res.server = server;
3392
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_COMMIT];
3393
	nfs41_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
L
Linus Torvalds 已提交
3394 3395
}

3396 3397 3398 3399 3400
struct nfs4_renewdata {
	struct nfs_client	*client;
	unsigned long		timestamp;
};

L
Linus Torvalds 已提交
3401 3402 3403 3404
/*
 * nfs4_proc_async_renew(): This is not one of the nfs_rpc_ops; it is a special
 * standalone procedure for queueing an asynchronous RENEW.
 */
3405
static void nfs4_renew_release(void *calldata)
3406
{
3407 3408
	struct nfs4_renewdata *data = calldata;
	struct nfs_client *clp = data->client;
3409

3410 3411 3412
	if (atomic_read(&clp->cl_count) > 1)
		nfs4_schedule_state_renewal(clp);
	nfs_put_client(clp);
3413
	kfree(data);
3414 3415
}

3416
static void nfs4_renew_done(struct rpc_task *task, void *calldata)
L
Linus Torvalds 已提交
3417
{
3418 3419 3420
	struct nfs4_renewdata *data = calldata;
	struct nfs_client *clp = data->client;
	unsigned long timestamp = data->timestamp;
L
Linus Torvalds 已提交
3421 3422

	if (task->tk_status < 0) {
3423
		/* Unless we're shutting down, schedule state recovery! */
3424 3425 3426
		if (test_bit(NFS_CS_RENEWD, &clp->cl_res_state) == 0)
			return;
		if (task->tk_status != NFS4ERR_CB_PATH_DOWN) {
3427
			nfs4_schedule_lease_recovery(clp);
3428 3429 3430
			return;
		}
		nfs4_schedule_path_down_recovery(clp);
L
Linus Torvalds 已提交
3431
	}
3432
	do_renew_lease(clp, timestamp);
L
Linus Torvalds 已提交
3433 3434
}

3435 3436
static const struct rpc_call_ops nfs4_renew_ops = {
	.rpc_call_done = nfs4_renew_done,
3437
	.rpc_release = nfs4_renew_release,
3438 3439
};

3440
static int nfs4_proc_async_renew(struct nfs_client *clp, struct rpc_cred *cred, unsigned renew_flags)
L
Linus Torvalds 已提交
3441 3442 3443 3444
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
3445
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
3446
	};
3447
	struct nfs4_renewdata *data;
L
Linus Torvalds 已提交
3448

3449 3450
	if (renew_flags == 0)
		return 0;
3451 3452
	if (!atomic_inc_not_zero(&clp->cl_count))
		return -EIO;
3453
	data = kmalloc(sizeof(*data), GFP_NOFS);
3454 3455 3456 3457
	if (data == NULL)
		return -ENOMEM;
	data->client = clp;
	data->timestamp = jiffies;
L
Linus Torvalds 已提交
3458
	return rpc_call_async(clp->cl_rpcclient, &msg, RPC_TASK_SOFT,
3459
			&nfs4_renew_ops, data);
L
Linus Torvalds 已提交
3460 3461
}

3462
static int nfs4_proc_renew(struct nfs_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
3463 3464 3465 3466
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
3467
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
3468 3469 3470 3471 3472 3473 3474
	};
	unsigned long now = jiffies;
	int status;

	status = rpc_call_sync(clp->cl_rpcclient, &msg, 0);
	if (status < 0)
		return status;
3475
	do_renew_lease(clp, now);
L
Linus Torvalds 已提交
3476 3477 3478
	return 0;
}

3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491
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);
}

/* Assuming that XATTR_SIZE_MAX is a multiple of PAGE_CACHE_SIZE, and that
 * it's OK to put sizeof(void) * (XATTR_SIZE_MAX/PAGE_CACHE_SIZE) bytes on
 * the stack.
 */
#define NFS4ACL_MAXPAGES (XATTR_SIZE_MAX >> PAGE_CACHE_SHIFT)

3492 3493 3494 3495 3496 3497 3498 3499 3500
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 {
J
Jovi Zhang 已提交
3501
		len = min_t(size_t, PAGE_CACHE_SIZE, buflen);
3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520
		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;
}

3521 3522 3523
struct nfs4_cached_acl {
	int cached;
	size_t len;
A
Andrew Morton 已提交
3524
	char data[0];
3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587
};

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;
}

static void nfs4_write_cached_acl(struct inode *inode, const char *buf, size_t acl_len)
{
	struct nfs4_cached_acl *acl;

	if (buf && acl_len <= PAGE_SIZE) {
		acl = kmalloc(sizeof(*acl) + acl_len, GFP_KERNEL);
		if (acl == NULL)
			goto out;
		acl->cached = 1;
		memcpy(acl->data, buf, acl_len);
	} 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);
}

3588 3589 3590 3591 3592 3593 3594 3595 3596 3597
/*
 * 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.
 */
3598
static ssize_t __nfs4_get_acl_uncached(struct inode *inode, void *buf, size_t buflen)
3599
{
3600
	struct page *pages[NFS4ACL_MAXPAGES] = {NULL, };
3601 3602 3603 3604 3605
	struct nfs_getaclargs args = {
		.fh = NFS_FH(inode),
		.acl_pages = pages,
		.acl_len = buflen,
	};
B
Benny Halevy 已提交
3606 3607 3608
	struct nfs_getaclres res = {
		.acl_len = buflen,
	};
3609
	void *resp_buf;
3610 3611 3612
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETACL],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3613
		.rpc_resp = &res,
3614
	};
3615
	int ret = -ENOMEM, npages, i, acl_len = 0;
3616

3617 3618 3619 3620 3621 3622 3623 3624 3625 3626
	npages = (buflen + PAGE_SIZE - 1) >> PAGE_SHIFT;
	/* 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;

	for (i = 0; i < npages; i++) {
		pages[i] = alloc_page(GFP_KERNEL);
		if (!pages[i])
			goto out_free;
3627
	}
3628 3629
	if (npages > 1) {
		/* for decoding across pages */
3630 3631
		res.acl_scratch = alloc_page(GFP_KERNEL);
		if (!res.acl_scratch)
3632
			goto out_free;
3633
	}
3634 3635 3636 3637 3638 3639 3640 3641
	args.acl_len = npages * PAGE_SIZE;
	args.acl_pgbase = 0;
	/* Let decode_getfacl know not to fail if the ACL data is larger than
	 * the page we send as a guess */
	if (buf == NULL)
		res.acl_flags |= NFS4_ACL_LEN_REQUEST;
	resp_buf = page_address(pages[0]);

3642
	dprintk("%s  buf %p buflen %zu npages %d args.acl_len %zu\n",
3643 3644 3645
		__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);
3646 3647
	if (ret)
		goto out_free;
3648 3649 3650 3651

	acl_len = res.acl_len - res.acl_data_offset;
	if (acl_len > args.acl_len)
		nfs4_write_cached_acl(inode, NULL, acl_len);
3652
	else
3653 3654
		nfs4_write_cached_acl(inode, resp_buf + res.acl_data_offset,
				      acl_len);
3655 3656
	if (buf) {
		ret = -ERANGE;
3657
		if (acl_len > buflen)
3658
			goto out_free;
3659 3660
		_copy_from_pages(buf, pages, res.acl_data_offset,
				res.acl_len);
3661
	}
3662
	ret = acl_len;
3663
out_free:
3664 3665 3666
	for (i = 0; i < npages; i++)
		if (pages[i])
			__free_page(pages[i]);
3667 3668
	if (res.acl_scratch)
		__free_page(res.acl_scratch);
3669 3670 3671
	return ret;
}

3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684
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);
		if (ret >= 0)
			break;
		ret = nfs4_handle_exception(NFS_SERVER(inode), ret, &exception);
	} while (exception.retry);
	return ret;
}

3685 3686 3687 3688 3689 3690 3691 3692 3693 3694
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;
3695 3696
	if (NFS_I(inode)->cache_validity & NFS_INO_INVALID_ACL)
		nfs_zap_acl_cache(inode);
3697 3698
	ret = nfs4_read_cached_acl(inode, buf, buflen);
	if (ret != -ENOENT)
3699 3700
		/* -ENOENT is returned if there is no ACL or if there is an ACL
		 * but no cached acl data, just the acl length */
3701 3702 3703 3704
		return ret;
	return nfs4_get_acl_uncached(inode, buf, buflen);
}

3705
static int __nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen)
3706 3707 3708 3709 3710 3711 3712 3713
{
	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
3714
	struct nfs_setaclres res;
3715 3716 3717
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETACL],
		.rpc_argp	= &arg,
B
Benny Halevy 已提交
3718
		.rpc_resp	= &res,
3719
	};
3720
	int ret, i;
3721 3722 3723

	if (!nfs4_server_supports_acls(server))
		return -EOPNOTSUPP;
3724 3725 3726
	i = buf_to_pages_noslab(buf, buflen, arg.acl_pages, &arg.acl_pgbase);
	if (i < 0)
		return i;
3727
	nfs_inode_return_delegation(inode);
3728
	ret = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
3729 3730 3731 3732 3733 3734 3735 3736

	/*
	 * 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]);

3737 3738 3739 3740 3741 3742 3743
	/*
	 * 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);
3744 3745
	nfs_access_zap_cache(inode);
	nfs_zap_acl_cache(inode);
3746 3747 3748
	return ret;
}

3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760
static int nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen)
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(NFS_SERVER(inode),
				__nfs4_proc_set_acl(inode, buf, buflen),
				&exception);
	} while (exception.retry);
	return err;
}

L
Linus Torvalds 已提交
3761
static int
3762
nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server, struct nfs4_state *state)
L
Linus Torvalds 已提交
3763
{
3764 3765 3766
	struct nfs_client *clp = server->nfs_client;

	if (task->tk_status >= 0)
L
Linus Torvalds 已提交
3767 3768
		return 0;
	switch(task->tk_status) {
3769
		case -NFS4ERR_DELEG_REVOKED:
3770 3771
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
3772 3773
			if (state != NULL)
				nfs_remove_bad_delegation(state->inode);
3774 3775 3776
		case -NFS4ERR_OPENMODE:
			if (state == NULL)
				break;
3777 3778
			nfs4_schedule_stateid_recovery(server, state);
			goto wait_on_recovery;
3779 3780 3781
		case -NFS4ERR_EXPIRED:
			if (state != NULL)
				nfs4_schedule_stateid_recovery(server, state);
L
Linus Torvalds 已提交
3782
		case -NFS4ERR_STALE_STATEID:
3783
		case -NFS4ERR_STALE_CLIENTID:
3784 3785
			nfs4_schedule_lease_recovery(clp);
			goto wait_on_recovery;
3786 3787 3788 3789 3790 3791 3792 3793 3794 3795
#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);
3796
			nfs4_schedule_session_recovery(clp->cl_session);
3797 3798 3799
			task->tk_status = 0;
			return -EAGAIN;
#endif /* CONFIG_NFS_V4_1 */
L
Linus Torvalds 已提交
3800
		case -NFS4ERR_DELAY:
3801
			nfs_inc_server_stats(server, NFSIOS_DELAY);
3802
		case -NFS4ERR_GRACE:
3803
		case -EKEYEXPIRED:
L
Linus Torvalds 已提交
3804 3805 3806
			rpc_delay(task, NFS4_POLL_RETRY_MAX);
			task->tk_status = 0;
			return -EAGAIN;
3807
		case -NFS4ERR_RETRY_UNCACHED_REP:
L
Linus Torvalds 已提交
3808 3809 3810 3811 3812 3813
		case -NFS4ERR_OLD_STATEID:
			task->tk_status = 0;
			return -EAGAIN;
	}
	task->tk_status = nfs4_map_errors(task->tk_status);
	return 0;
3814
wait_on_recovery:
3815 3816 3817 3818 3819
	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
3820 3821
}

3822 3823 3824
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
3825 3826 3827 3828 3829
{
	nfs4_verifier sc_verifier;
	struct nfs4_setclientid setclientid = {
		.sc_verifier = &sc_verifier,
		.sc_prog = program,
A
Andy Adamson 已提交
3830
		.sc_cb_ident = clp->cl_cb_ident,
L
Linus Torvalds 已提交
3831 3832 3833 3834
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID],
		.rpc_argp = &setclientid,
3835
		.rpc_resp = res,
3836
		.rpc_cred = cred,
L
Linus Torvalds 已提交
3837
	};
A
Al Viro 已提交
3838
	__be32 *p;
L
Linus Torvalds 已提交
3839 3840 3841
	int loop = 0;
	int status;

A
Al Viro 已提交
3842
	p = (__be32*)sc_verifier.data;
L
Linus Torvalds 已提交
3843 3844 3845 3846
	*p++ = htonl((u32)clp->cl_boot_time.tv_sec);
	*p = htonl((u32)clp->cl_boot_time.tv_nsec);

	for(;;) {
3847
		rcu_read_lock();
L
Linus Torvalds 已提交
3848
		setclientid.sc_name_len = scnprintf(setclientid.sc_name,
3849
				sizeof(setclientid.sc_name), "%s/%s %s %s %u",
3850 3851 3852
				clp->cl_ipaddr,
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_ADDR),
3853 3854
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_PROTO),
3855
				clp->cl_rpcclient->cl_auth->au_ops->au_name,
L
Linus Torvalds 已提交
3856 3857
				clp->cl_id_uniquifier);
		setclientid.sc_netid_len = scnprintf(setclientid.sc_netid,
3858 3859 3860
				sizeof(setclientid.sc_netid),
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_NETID));
L
Linus Torvalds 已提交
3861
		setclientid.sc_uaddr_len = scnprintf(setclientid.sc_uaddr,
3862
				sizeof(setclientid.sc_uaddr), "%s.%u.%u",
L
Linus Torvalds 已提交
3863
				clp->cl_ipaddr, port >> 8, port & 255);
3864
		rcu_read_unlock();
L
Linus Torvalds 已提交
3865

3866
		status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
L
Linus Torvalds 已提交
3867 3868
		if (status != -NFS4ERR_CLID_INUSE)
			break;
3869 3870
		if (loop != 0) {
			++clp->cl_id_uniquifier;
L
Linus Torvalds 已提交
3871
			break;
3872 3873 3874
		}
		++loop;
		ssleep(clp->cl_lease_time / HZ + 1);
L
Linus Torvalds 已提交
3875 3876 3877 3878
	}
	return status;
}

3879
int nfs4_proc_setclientid_confirm(struct nfs_client *clp,
3880 3881
		struct nfs4_setclientid_res *arg,
		struct rpc_cred *cred)
L
Linus Torvalds 已提交
3882 3883 3884 3885
{
	struct nfs_fsinfo fsinfo;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID_CONFIRM],
3886
		.rpc_argp = arg,
L
Linus Torvalds 已提交
3887
		.rpc_resp = &fsinfo,
3888
		.rpc_cred = cred,
L
Linus Torvalds 已提交
3889 3890 3891 3892 3893
	};
	unsigned long now;
	int status;

	now = jiffies;
3894
	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
L
Linus Torvalds 已提交
3895 3896 3897 3898 3899 3900 3901 3902 3903
	if (status == 0) {
		spin_lock(&clp->cl_lock);
		clp->cl_lease_time = fsinfo.lease_time * HZ;
		clp->cl_last_renewal = now;
		spin_unlock(&clp->cl_lock);
	}
	return status;
}

3904 3905
struct nfs4_delegreturndata {
	struct nfs4_delegreturnargs args;
3906
	struct nfs4_delegreturnres res;
3907 3908
	struct nfs_fh fh;
	nfs4_stateid stateid;
3909
	unsigned long timestamp;
3910
	struct nfs_fattr fattr;
3911 3912 3913 3914 3915 3916
	int rpc_status;
};

static void nfs4_delegreturn_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_delegreturndata *data = calldata;
3917

3918 3919
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return;
3920

3921 3922 3923 3924
	switch (task->tk_status) {
	case -NFS4ERR_STALE_STATEID:
	case -NFS4ERR_EXPIRED:
	case 0:
3925
		renew_lease(data->res.server, data->timestamp);
3926 3927 3928 3929
		break;
	default:
		if (nfs4_async_handle_error(task, data->res.server, NULL) ==
				-EAGAIN) {
3930
			rpc_restart_call_prepare(task);
3931 3932 3933 3934
			return;
		}
	}
	data->rpc_status = task->tk_status;
3935 3936 3937 3938 3939 3940 3941
}

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

3942 3943 3944 3945 3946 3947 3948
#if defined(CONFIG_NFS_V4_1)
static void nfs4_delegreturn_prepare(struct rpc_task *task, void *data)
{
	struct nfs4_delegreturndata *d_data;

	d_data = (struct nfs4_delegreturndata *)data;

3949
	if (nfs4_setup_sequence(d_data->res.server,
3950
				&d_data->args.seq_args,
3951
				&d_data->res.seq_res, task))
3952 3953 3954 3955 3956
		return;
	rpc_call_start(task);
}
#endif /* CONFIG_NFS_V4_1 */

3957
static const struct rpc_call_ops nfs4_delegreturn_ops = {
3958 3959 3960
#if defined(CONFIG_NFS_V4_1)
	.rpc_call_prepare = nfs4_delegreturn_prepare,
#endif /* CONFIG_NFS_V4_1 */
3961 3962 3963 3964
	.rpc_call_done = nfs4_delegreturn_done,
	.rpc_release = nfs4_delegreturn_release,
};

3965
static int _nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
3966 3967
{
	struct nfs4_delegreturndata *data;
3968
	struct nfs_server *server = NFS_SERVER(inode);
3969
	struct rpc_task *task;
3970 3971 3972 3973
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DELEGRETURN],
		.rpc_cred = cred,
	};
T
Trond Myklebust 已提交
3974 3975
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
3976
		.rpc_message = &msg,
T
Trond Myklebust 已提交
3977 3978 3979
		.callback_ops = &nfs4_delegreturn_ops,
		.flags = RPC_TASK_ASYNC,
	};
3980
	int status = 0;
3981

3982
	data = kzalloc(sizeof(*data), GFP_NOFS);
3983 3984
	if (data == NULL)
		return -ENOMEM;
3985
	nfs41_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
3986 3987
	data->args.fhandle = &data->fh;
	data->args.stateid = &data->stateid;
3988
	data->args.bitmask = server->attr_bitmask;
3989
	nfs_copy_fh(&data->fh, NFS_FH(inode));
3990
	nfs4_stateid_copy(&data->stateid, stateid);
3991 3992
	data->res.fattr = &data->fattr;
	data->res.server = server;
3993
	nfs_fattr_init(data->res.fattr);
3994
	data->timestamp = jiffies;
3995 3996
	data->rpc_status = 0;

T
Trond Myklebust 已提交
3997
	task_setup_data.callback_data = data;
3998 3999
	msg.rpc_argp = &data->args;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
4000
	task = rpc_run_task(&task_setup_data);
4001
	if (IS_ERR(task))
4002
		return PTR_ERR(task);
4003 4004
	if (!issync)
		goto out;
4005
	status = nfs4_wait_for_completion_rpc_task(task);
4006 4007 4008 4009 4010 4011 4012
	if (status != 0)
		goto out;
	status = data->rpc_status;
	if (status != 0)
		goto out;
	nfs_refresh_inode(inode, &data->fattr);
out:
4013
	rpc_put_task(task);
4014
	return status;
L
Linus Torvalds 已提交
4015 4016
}

4017
int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
L
Linus Torvalds 已提交
4018 4019 4020 4021 4022
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs4_exception exception = { };
	int err;
	do {
4023
		err = _nfs4_proc_delegreturn(inode, cred, stateid, issync);
L
Linus Torvalds 已提交
4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043
		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)
{
4044
	freezable_schedule_timeout_killable(timeout);
L
Linus Torvalds 已提交
4045 4046 4047 4048 4049 4050 4051 4052 4053 4054
	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);
4055
	struct nfs_client *clp = server->nfs_client;
T
Trond Myklebust 已提交
4056
	struct nfs_lockt_args arg = {
L
Linus Torvalds 已提交
4057
		.fh = NFS_FH(inode),
T
Trond Myklebust 已提交
4058
		.fl = request,
L
Linus Torvalds 已提交
4059
	};
T
Trond Myklebust 已提交
4060 4061
	struct nfs_lockt_res res = {
		.denied = request,
L
Linus Torvalds 已提交
4062 4063 4064 4065 4066 4067 4068 4069 4070 4071
	};
	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
4072
	arg.lock_owner.clientid = clp->cl_clientid;
4073 4074 4075 4076
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
	lsp = request->fl_u.nfs4_fl.owner;
4077
	arg.lock_owner.id = lsp->ls_seqid.owner_id;
4078
	arg.lock_owner.s_dev = server->s_dev;
4079
	status = nfs4_call_sync(server->client, server, &msg, &arg.seq_args, &res.seq_res, 1);
T
Trond Myklebust 已提交
4080 4081 4082 4083 4084 4085
	switch (status) {
		case 0:
			request->fl_type = F_UNLCK;
			break;
		case -NFS4ERR_DENIED:
			status = 0;
L
Linus Torvalds 已提交
4086
	}
4087
	request->fl_ops->fl_release_private(request);
4088
out:
L
Linus Torvalds 已提交
4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120
	return status;
}

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

	do {
		err = nfs4_handle_exception(NFS_SERVER(state->inode),
				_nfs4_proc_getlk(state, cmd, request),
				&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;
}

4121
struct nfs4_unlockdata {
T
Trond Myklebust 已提交
4122 4123
	struct nfs_locku_args arg;
	struct nfs_locku_res res;
4124 4125
	struct nfs4_lock_state *lsp;
	struct nfs_open_context *ctx;
T
Trond Myklebust 已提交
4126 4127
	struct file_lock fl;
	const struct nfs_server *server;
4128
	unsigned long timestamp;
4129 4130
};

T
Trond Myklebust 已提交
4131 4132 4133 4134 4135 4136 4137 4138
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;

4139
	p = kzalloc(sizeof(*p), GFP_NOFS);
T
Trond Myklebust 已提交
4140 4141 4142 4143 4144
	if (p == NULL)
		return NULL;
	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
	p->arg.seqid = seqid;
4145
	p->res.seqid = seqid;
T
Trond Myklebust 已提交
4146 4147 4148 4149 4150 4151 4152 4153 4154 4155
	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;
}

4156
static void nfs4_locku_release_calldata(void *data)
4157
{
4158
	struct nfs4_unlockdata *calldata = data;
T
Trond Myklebust 已提交
4159
	nfs_free_seqid(calldata->arg.seqid);
4160 4161 4162
	nfs4_put_lock_state(calldata->lsp);
	put_nfs_open_context(calldata->ctx);
	kfree(calldata);
4163 4164
}

4165
static void nfs4_locku_done(struct rpc_task *task, void *data)
4166
{
4167
	struct nfs4_unlockdata *calldata = data;
4168

4169 4170
	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
		return;
4171 4172
	switch (task->tk_status) {
		case 0:
4173 4174
			nfs4_stateid_copy(&calldata->lsp->ls_stateid,
					&calldata->res.stateid);
4175
			renew_lease(calldata->server, calldata->timestamp);
4176
			break;
4177 4178
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OLD_STATEID:
4179 4180 4181 4182
		case -NFS4ERR_STALE_STATEID:
		case -NFS4ERR_EXPIRED:
			break;
		default:
4183
			if (nfs4_async_handle_error(task, calldata->server, NULL) == -EAGAIN)
4184
				rpc_restart_call_prepare(task);
4185 4186 4187
	}
}

T
Trond Myklebust 已提交
4188
static void nfs4_locku_prepare(struct rpc_task *task, void *data)
4189
{
T
Trond Myklebust 已提交
4190
	struct nfs4_unlockdata *calldata = data;
4191

T
Trond Myklebust 已提交
4192
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
4193 4194
		return;
	if ((calldata->lsp->ls_flags & NFS_LOCK_INITIALIZED) == 0) {
4195 4196
		/* Note: exit _without_ running nfs4_locku_done */
		task->tk_action = NULL;
4197 4198
		return;
	}
4199
	calldata->timestamp = jiffies;
4200
	if (nfs4_setup_sequence(calldata->server,
4201
				&calldata->arg.seq_args,
4202
				&calldata->res.seq_res, task))
4203
		return;
4204
	rpc_call_start(task);
4205 4206
}

4207
static const struct rpc_call_ops nfs4_locku_ops = {
T
Trond Myklebust 已提交
4208
	.rpc_call_prepare = nfs4_locku_prepare,
4209
	.rpc_call_done = nfs4_locku_done,
4210
	.rpc_release = nfs4_locku_release_calldata,
4211 4212
};

4213 4214 4215 4216 4217 4218
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;
4219 4220 4221 4222
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCKU],
		.rpc_cred = ctx->cred,
	};
T
Trond Myklebust 已提交
4223 4224
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(lsp->ls_state->inode),
4225
		.rpc_message = &msg,
T
Trond Myklebust 已提交
4226
		.callback_ops = &nfs4_locku_ops,
4227
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
4228 4229
		.flags = RPC_TASK_ASYNC,
	};
4230

4231 4232 4233 4234 4235
	/* 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;

4236 4237 4238 4239 4240 4241
	data = nfs4_alloc_unlockdata(fl, ctx, lsp, seqid);
	if (data == NULL) {
		nfs_free_seqid(seqid);
		return ERR_PTR(-ENOMEM);
	}

4242
	nfs41_init_sequence(&data->arg.seq_args, &data->res.seq_res, 1);
4243 4244
	msg.rpc_argp = &data->arg;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
4245 4246
	task_setup_data.callback_data = data;
	return rpc_run_task(&task_setup_data);
4247 4248
}

4249 4250
static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
{
4251
	struct nfs_inode *nfsi = NFS_I(state->inode);
T
Trond Myklebust 已提交
4252
	struct nfs_seqid *seqid;
L
Linus Torvalds 已提交
4253
	struct nfs4_lock_state *lsp;
4254 4255
	struct rpc_task *task;
	int status = 0;
4256
	unsigned char fl_flags = request->fl_flags;
4257

4258
	status = nfs4_set_lock_state(state, request);
4259 4260
	/* Unlock _before_ we do the RPC call */
	request->fl_flags |= FL_EXISTS;
4261 4262 4263
	down_read(&nfsi->rwsem);
	if (do_vfs_lock(request->fl_file, request) == -ENOENT) {
		up_read(&nfsi->rwsem);
4264
		goto out;
4265 4266
	}
	up_read(&nfsi->rwsem);
4267
	if (status != 0)
4268 4269 4270 4271
		goto out;
	/* Is this a delegated lock? */
	if (test_bit(NFS_DELEGATED_STATE, &state->flags))
		goto out;
4272
	lsp = request->fl_u.nfs4_fl.owner;
4273
	seqid = nfs_alloc_seqid(&lsp->ls_seqid, GFP_KERNEL);
4274
	status = -ENOMEM;
T
Trond Myklebust 已提交
4275
	if (seqid == NULL)
4276
		goto out;
4277
	task = nfs4_do_unlck(request, nfs_file_open_context(request->fl_file), lsp, seqid);
4278 4279
	status = PTR_ERR(task);
	if (IS_ERR(task))
4280
		goto out;
4281
	status = nfs4_wait_for_completion_rpc_task(task);
4282
	rpc_put_task(task);
4283
out:
4284
	request->fl_flags = fl_flags;
L
Linus Torvalds 已提交
4285 4286 4287
	return status;
}

4288 4289 4290 4291 4292 4293
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;
4294
	unsigned long timestamp;
4295 4296
	int rpc_status;
	int cancelled;
4297
	struct nfs_server *server;
4298 4299 4300
};

static struct nfs4_lockdata *nfs4_alloc_lockdata(struct file_lock *fl,
4301 4302
		struct nfs_open_context *ctx, struct nfs4_lock_state *lsp,
		gfp_t gfp_mask)
L
Linus Torvalds 已提交
4303
{
4304 4305
	struct nfs4_lockdata *p;
	struct inode *inode = lsp->ls_state->inode;
L
Linus Torvalds 已提交
4306
	struct nfs_server *server = NFS_SERVER(inode);
4307

4308
	p = kzalloc(sizeof(*p), gfp_mask);
4309 4310 4311 4312 4313
	if (p == NULL)
		return NULL;

	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
4314
	p->arg.open_seqid = nfs_alloc_seqid(&lsp->ls_state->owner->so_seqid, gfp_mask);
4315 4316
	if (p->arg.open_seqid == NULL)
		goto out_free;
4317
	p->arg.lock_seqid = nfs_alloc_seqid(&lsp->ls_seqid, gfp_mask);
4318
	if (p->arg.lock_seqid == NULL)
4319
		goto out_free_seqid;
4320
	p->arg.lock_stateid = &lsp->ls_stateid;
4321
	p->arg.lock_owner.clientid = server->nfs_client->cl_clientid;
4322
	p->arg.lock_owner.id = lsp->ls_seqid.owner_id;
4323
	p->arg.lock_owner.s_dev = server->s_dev;
4324
	p->res.lock_seqid = p->arg.lock_seqid;
4325
	p->lsp = lsp;
4326
	p->server = server;
4327 4328 4329 4330
	atomic_inc(&lsp->ls_count);
	p->ctx = get_nfs_open_context(ctx);
	memcpy(&p->fl, fl, sizeof(p->fl));
	return p;
4331 4332
out_free_seqid:
	nfs_free_seqid(p->arg.open_seqid);
4333 4334 4335 4336 4337 4338 4339 4340 4341
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;
4342

4343
	dprintk("%s: begin!\n", __func__);
4344 4345
	if (nfs_wait_on_sequence(data->arg.lock_seqid, task) != 0)
		return;
4346 4347
	/* Do we need to do an open_to_lock_owner? */
	if (!(data->arg.lock_seqid->sequence->flags & NFS_SEQID_CONFIRMED)) {
4348 4349
		if (nfs_wait_on_sequence(data->arg.open_seqid, task) != 0)
			return;
4350 4351
		data->arg.open_stateid = &state->stateid;
		data->arg.new_lock_owner = 1;
4352
		data->res.open_seqid = data->arg.open_seqid;
4353 4354
	} else
		data->arg.new_lock_owner = 0;
4355
	data->timestamp = jiffies;
4356 4357
	if (nfs4_setup_sequence(data->server,
				&data->arg.seq_args,
4358
				&data->res.seq_res, task))
4359
		return;
4360
	rpc_call_start(task);
4361
	dprintk("%s: done!, ret = %d\n", __func__, data->rpc_status);
4362 4363
}

4364 4365 4366 4367 4368 4369
static void nfs4_recover_lock_prepare(struct rpc_task *task, void *calldata)
{
	rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
	nfs4_lock_prepare(task, calldata);
}

4370 4371 4372 4373
static void nfs4_lock_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_lockdata *data = calldata;

4374
	dprintk("%s: begin!\n", __func__);
4375

4376 4377
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return;
4378

4379 4380 4381 4382 4383 4384 4385 4386
	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) {
4387
		nfs4_stateid_copy(&data->lsp->ls_stateid, &data->res.stateid);
4388
		data->lsp->ls_flags |= NFS_LOCK_INITIALIZED;
4389
		renew_lease(NFS_SERVER(data->ctx->dentry->d_inode), data->timestamp);
4390 4391
	}
out:
4392
	dprintk("%s: done, ret = %d!\n", __func__, data->rpc_status);
4393 4394 4395 4396 4397 4398
}

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

4399
	dprintk("%s: begin!\n", __func__);
4400
	nfs_free_seqid(data->arg.open_seqid);
4401 4402 4403 4404 4405
	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))
4406
			rpc_put_task_async(task);
4407
		dprintk("%s: cancelling lock!\n", __func__);
4408 4409 4410 4411 4412
	} else
		nfs_free_seqid(data->arg.lock_seqid);
	nfs4_put_lock_state(data->lsp);
	put_nfs_open_context(data->ctx);
	kfree(data);
4413
	dprintk("%s: done!\n", __func__);
4414 4415 4416 4417 4418 4419 4420 4421
}

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,
};

4422 4423 4424 4425 4426 4427
static const struct rpc_call_ops nfs4_recover_lock_ops = {
	.rpc_call_prepare = nfs4_recover_lock_prepare,
	.rpc_call_done = nfs4_lock_done,
	.rpc_release = nfs4_lock_release,
};

4428 4429 4430 4431 4432
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:
4433
		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
4434 4435
		if (new_lock_owner != 0 ||
		   (lsp->ls_flags & NFS_LOCK_INITIALIZED) != 0)
4436
			nfs4_schedule_stateid_recovery(server, lsp->ls_state);
4437 4438 4439
		break;
	case -NFS4ERR_STALE_STATEID:
		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
4440 4441
	case -NFS4ERR_EXPIRED:
		nfs4_schedule_lease_recovery(server->nfs_client);
4442 4443 4444
	};
}

4445
static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *fl, int recovery_type)
4446 4447 4448
{
	struct nfs4_lockdata *data;
	struct rpc_task *task;
4449 4450 4451 4452
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCK],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
4453 4454
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(state->inode),
4455
		.rpc_message = &msg,
T
Trond Myklebust 已提交
4456
		.callback_ops = &nfs4_lock_ops,
4457
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
4458 4459
		.flags = RPC_TASK_ASYNC,
	};
4460 4461
	int ret;

4462
	dprintk("%s: begin!\n", __func__);
4463
	data = nfs4_alloc_lockdata(fl, nfs_file_open_context(fl->fl_file),
4464 4465
			fl->fl_u.nfs4_fl.owner,
			recovery_type == NFS_LOCK_NEW ? GFP_KERNEL : GFP_NOFS);
4466 4467 4468 4469
	if (data == NULL)
		return -ENOMEM;
	if (IS_SETLKW(cmd))
		data->arg.block = 1;
4470 4471
	if (recovery_type > NFS_LOCK_NEW) {
		if (recovery_type == NFS_LOCK_RECLAIM)
4472
			data->arg.reclaim = NFS_LOCK_RECLAIM;
4473 4474
		task_setup_data.callback_ops = &nfs4_recover_lock_ops;
	}
4475
	nfs41_init_sequence(&data->arg.seq_args, &data->res.seq_res, 1);
4476 4477
	msg.rpc_argp = &data->arg;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
4478 4479
	task_setup_data.callback_data = data;
	task = rpc_run_task(&task_setup_data);
4480
	if (IS_ERR(task))
4481 4482 4483 4484
		return PTR_ERR(task);
	ret = nfs4_wait_for_completion_rpc_task(task);
	if (ret == 0) {
		ret = data->rpc_status;
4485 4486 4487
		if (ret)
			nfs4_handle_setlk_error(data->server, data->lsp,
					data->arg.new_lock_owner, ret);
4488 4489
	} else
		data->cancelled = 1;
4490
	rpc_put_task(task);
4491
	dprintk("%s: done, ret = %d!\n", __func__, ret);
4492
	return ret;
L
Linus Torvalds 已提交
4493 4494 4495 4496
}

static int nfs4_lock_reclaim(struct nfs4_state *state, struct file_lock *request)
{
4497 4498 4499 4500 4501
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;

	do {
4502 4503 4504
		/* Cache the lock if possible... */
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
4505
		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_RECLAIM);
4506
		if (err != -NFS4ERR_DELAY)
4507 4508 4509 4510
			break;
		nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
	return err;
L
Linus Torvalds 已提交
4511 4512 4513 4514
}

static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request)
{
4515 4516 4517 4518
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;

4519 4520 4521
	err = nfs4_set_lock_state(state, request);
	if (err != 0)
		return err;
4522
	do {
4523 4524
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
4525
		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_EXPIRED);
4526 4527 4528 4529 4530 4531 4532 4533
		switch (err) {
		default:
			goto out;
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			nfs4_handle_exception(server, err, &exception);
			err = 0;
		}
4534
	} while (exception.retry);
4535
out:
4536
	return err;
L
Linus Torvalds 已提交
4537 4538
}

4539
#if defined(CONFIG_NFS_V4_1)
4540
static int nfs41_check_expired_locks(struct nfs4_state *state)
4541
{
4542 4543
	int status, ret = NFS_OK;
	struct nfs4_lock_state *lsp;
4544 4545
	struct nfs_server *server = NFS_SERVER(state->inode);

4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565
	list_for_each_entry(lsp, &state->lock_states, ls_locks) {
		if (lsp->ls_flags & NFS_LOCK_INITIALIZED) {
			status = nfs41_test_stateid(server, &lsp->ls_stateid);
			if (status != NFS_OK) {
				nfs41_free_stateid(server, &lsp->ls_stateid);
				lsp->ls_flags &= ~NFS_LOCK_INITIALIZED;
				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);
4566
	if (status == NFS_OK)
4567
		return status;
4568 4569 4570 4571
	return nfs4_lock_expired(state, request);
}
#endif

L
Linus Torvalds 已提交
4572 4573
static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
4574
	struct nfs_inode *nfsi = NFS_I(state->inode);
4575
	unsigned char fl_flags = request->fl_flags;
4576
	int status = -ENOLCK;
L
Linus Torvalds 已提交
4577

4578 4579 4580
	if ((fl_flags & FL_POSIX) &&
			!test_bit(NFS_STATE_POSIX_LOCKS, &state->flags))
		goto out;
4581 4582 4583 4584
	/* Is this a delegated open? */
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
4585 4586 4587 4588
	request->fl_flags |= FL_ACCESS;
	status = do_vfs_lock(request->fl_file, request);
	if (status < 0)
		goto out;
4589
	down_read(&nfsi->rwsem);
4590 4591 4592
	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
		/* Yes: cache locks! */
		/* ...but avoid races with delegation recall... */
4593 4594 4595
		request->fl_flags = fl_flags & ~FL_SLEEP;
		status = do_vfs_lock(request->fl_file, request);
		goto out_unlock;
4596
	}
4597
	status = _nfs4_do_setlk(state, cmd, request, NFS_LOCK_NEW);
4598
	if (status != 0)
4599
		goto out_unlock;
4600
	/* Note: we always want to sleep here! */
4601
	request->fl_flags = fl_flags | FL_SLEEP;
4602
	if (do_vfs_lock(request->fl_file, request) < 0)
4603 4604
		printk(KERN_WARNING "NFS: %s: VFS is out of sync with lock "
			"manager!\n", __func__);
4605
out_unlock:
4606
	up_read(&nfsi->rwsem);
4607 4608
out:
	request->fl_flags = fl_flags;
L
Linus Torvalds 已提交
4609 4610 4611 4612 4613
	return status;
}

static int nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
4614 4615 4616
	struct nfs4_exception exception = {
		.state = state,
	};
L
Linus Torvalds 已提交
4617 4618 4619
	int err;

	do {
4620 4621 4622
		err = _nfs4_proc_setlk(state, cmd, request);
		if (err == -NFS4ERR_DENIED)
			err = -EAGAIN;
L
Linus Torvalds 已提交
4623
		err = nfs4_handle_exception(NFS_SERVER(state->inode),
4624
				err, &exception);
L
Linus Torvalds 已提交
4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637
	} 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 */
4638
	ctx = nfs_file_open_context(filp);
L
Linus Torvalds 已提交
4639 4640 4641 4642 4643
	state = ctx->state;

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

4644 4645 4646 4647 4648
	if (IS_GETLK(cmd)) {
		if (state != NULL)
			return nfs4_proc_getlk(state, F_GETLK, request);
		return 0;
	}
L
Linus Torvalds 已提交
4649 4650 4651 4652

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

4653 4654 4655 4656 4657
	if (request->fl_type == F_UNLCK) {
		if (state != NULL)
			return nfs4_proc_unlck(state, cmd, request);
		return 0;
	}
L
Linus Torvalds 已提交
4658

4659 4660
	if (state == NULL)
		return -ENOLCK;
L
Linus Torvalds 已提交
4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672
	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;
}

4673 4674 4675 4676 4677 4678 4679 4680 4681 4682
int nfs4_lock_delegation_recall(struct nfs4_state *state, struct file_lock *fl)
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;

	err = nfs4_set_lock_state(state, fl);
	if (err != 0)
		goto out;
	do {
4683
		err = _nfs4_do_setlk(state, F_SETLK, fl, NFS_LOCK_NEW);
4684 4685
		switch (err) {
			default:
4686 4687
				printk(KERN_ERR "NFS: %s: unhandled error "
					"%d.\n", __func__, err);
4688
			case 0:
4689
			case -ESTALE:
4690 4691
				goto out;
			case -NFS4ERR_EXPIRED:
4692
				nfs4_schedule_stateid_recovery(server, state);
4693
			case -NFS4ERR_STALE_CLIENTID:
4694
			case -NFS4ERR_STALE_STATEID:
4695 4696
				nfs4_schedule_lease_recovery(server->nfs_client);
				goto out;
4697 4698 4699 4700 4701
			case -NFS4ERR_BADSESSION:
			case -NFS4ERR_BADSLOT:
			case -NFS4ERR_BAD_HIGH_SLOT:
			case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
			case -NFS4ERR_DEADSESSION:
4702
				nfs4_schedule_session_recovery(server->nfs_client->cl_session);
4703
				goto out;
4704 4705 4706 4707 4708
			case -ERESTARTSYS:
				/*
				 * The show must go on: exit, but mark the
				 * stateid as needing recovery.
				 */
4709
			case -NFS4ERR_DELEG_REVOKED:
4710 4711 4712
			case -NFS4ERR_ADMIN_REVOKED:
			case -NFS4ERR_BAD_STATEID:
			case -NFS4ERR_OPENMODE:
4713
				nfs4_schedule_stateid_recovery(server, state);
4714 4715
				err = 0;
				goto out;
4716 4717 4718 4719 4720 4721 4722 4723 4724
			case -EKEYEXPIRED:
				/*
				 * User RPCSEC_GSS context has expired.
				 * We cannot recover this stateid now, so
				 * skip it and allow recovery thread to
				 * proceed.
				 */
				err = 0;
				goto out;
4725 4726 4727 4728 4729
			case -ENOMEM:
			case -NFS4ERR_DENIED:
				/* kill_proc(fl->fl_pid, SIGLOST, 1); */
				err = 0;
				goto out;
4730 4731 4732
			case -NFS4ERR_DELAY:
				break;
		}
4733 4734 4735 4736 4737
		err = nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
out:
	return err;
}
4738

4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761
static void nfs4_release_lockowner_release(void *calldata)
{
	kfree(calldata);
}

const struct rpc_call_ops nfs4_release_lockowner_ops = {
	.rpc_release = nfs4_release_lockowner_release,
};

void nfs4_release_lockowner(const struct nfs4_lock_state *lsp)
{
	struct nfs_server *server = lsp->ls_state->owner->so_server;
	struct nfs_release_lockowner_args *args;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RELEASE_LOCKOWNER],
	};

	if (server->nfs_client->cl_mvops->minor_version != 0)
		return;
	args = kmalloc(sizeof(*args), GFP_NOFS);
	if (!args)
		return;
	args->lock_owner.clientid = server->nfs_client->cl_clientid;
4762
	args->lock_owner.id = lsp->ls_seqid.owner_id;
4763
	args->lock_owner.s_dev = server->s_dev;
4764 4765 4766 4767
	msg.rpc_argp = args;
	rpc_call_async(server->client, &msg, 0, &nfs4_release_lockowner_ops, args);
}

4768 4769
#define XATTR_NAME_NFSV4_ACL "system.nfs4_acl"

4770 4771 4772
static int nfs4_xattr_set_nfs4_acl(struct dentry *dentry, const char *key,
				   const void *buf, size_t buflen,
				   int flags, int type)
4773
{
4774 4775
	if (strcmp(key, "") != 0)
		return -EINVAL;
4776

4777
	return nfs4_proc_set_acl(dentry->d_inode, buf, buflen);
4778 4779
}

4780 4781
static int nfs4_xattr_get_nfs4_acl(struct dentry *dentry, const char *key,
				   void *buf, size_t buflen, int type)
4782
{
4783 4784
	if (strcmp(key, "") != 0)
		return -EINVAL;
4785

4786
	return nfs4_proc_get_acl(dentry->d_inode, buf, buflen);
4787 4788
}

4789 4790 4791
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)
4792
{
4793
	size_t len = sizeof(XATTR_NAME_NFSV4_ACL);
4794

4795 4796
	if (!nfs4_server_supports_acls(NFS_SERVER(dentry->d_inode)))
		return 0;
4797 4798 4799

	if (list && len <= list_len)
		memcpy(list, XATTR_NAME_NFSV4_ACL, len);
4800
	return len;
4801 4802
}

4803 4804 4805
/*
 * nfs_fhget will use either the mounted_on_fileid or the fileid
 */
4806 4807
static void nfs_fixup_referral_attributes(struct nfs_fattr *fattr)
{
4808 4809 4810
	if (!(((fattr->valid & NFS_ATTR_FATTR_MOUNTED_ON_FILEID) ||
	       (fattr->valid & NFS_ATTR_FATTR_FILEID)) &&
	      (fattr->valid & NFS_ATTR_FATTR_FSID) &&
4811
	      (fattr->valid & NFS_ATTR_FATTR_V4_LOCATIONS)))
4812 4813 4814
		return;

	fattr->valid |= NFS_ATTR_FATTR_TYPE | NFS_ATTR_FATTR_MODE |
4815
		NFS_ATTR_FATTR_NLINK | NFS_ATTR_FATTR_V4_REFERRAL;
4816 4817 4818 4819
	fattr->mode = S_IFDIR | S_IRUGO | S_IXUGO;
	fattr->nlink = 2;
}

4820
int nfs4_proc_fs_locations(struct inode *dir, const struct qstr *name,
4821
		struct nfs4_fs_locations *fs_locations, struct page *page)
4822 4823 4824
{
	struct nfs_server *server = NFS_SERVER(dir);
	u32 bitmask[2] = {
4825
		[0] = FATTR4_WORD0_FSID | FATTR4_WORD0_FS_LOCATIONS,
4826 4827 4828
	};
	struct nfs4_fs_locations_arg args = {
		.dir_fh = NFS_FH(dir),
4829
		.name = name,
4830 4831 4832
		.page = page,
		.bitmask = bitmask,
	};
B
Benny Halevy 已提交
4833 4834 4835
	struct nfs4_fs_locations_res res = {
		.fs_locations = fs_locations,
	};
4836 4837 4838
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FS_LOCATIONS],
		.rpc_argp = &args,
B
Benny Halevy 已提交
4839
		.rpc_resp = &res,
4840 4841 4842
	};
	int status;

4843
	dprintk("%s: start\n", __func__);
4844 4845 4846 4847 4848 4849 4850 4851

	/* 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;

4852
	nfs_fattr_init(&fs_locations->fattr);
4853
	fs_locations->server = server;
4854
	fs_locations->nlocations = 0;
4855
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
4856
	dprintk("%s: returned status = %d\n", __func__, status);
4857 4858 4859
	return status;
}

B
Bryan Schumaker 已提交
4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891 4892 4893
static int _nfs4_proc_secinfo(struct inode *dir, const struct qstr *name, struct nfs4_secinfo_flavors *flavors)
{
	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,
	};

	dprintk("NFS call  secinfo %s\n", name->name);
	status = nfs4_call_sync(NFS_SERVER(dir)->client, NFS_SERVER(dir), &msg, &args.seq_args, &res.seq_res, 0);
	dprintk("NFS reply  secinfo: %d\n", status);
	return status;
}

int nfs4_proc_secinfo(struct inode *dir, const struct qstr *name, struct nfs4_secinfo_flavors *flavors)
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(NFS_SERVER(dir),
				_nfs4_proc_secinfo(dir, name, flavors),
				&exception);
	} while (exception.retry);
	return err;
}

4894
#ifdef CONFIG_NFS_V4_1
4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913
/*
 * 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;
}

4914 4915 4916 4917 4918 4919 4920 4921 4922 4923
static bool
nfs41_same_server_scope(struct server_scope *a, struct server_scope *b)
{
	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;
}

B
Benny Halevy 已提交
4924 4925 4926 4927 4928 4929 4930 4931
/*
 * nfs4_proc_exchange_id()
 *
 * 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.
 */
4932
int nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred)
B
Benny Halevy 已提交
4933 4934 4935 4936
{
	nfs4_verifier verifier;
	struct nfs41_exchange_id_args args = {
		.client = clp,
4937
		.flags = EXCHGID4_FLAG_SUPP_MOVED_REFER,
B
Benny Halevy 已提交
4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951 4952
	};
	struct nfs41_exchange_id_res res = {
		.client = clp,
	};
	int status;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_EXCHANGE_ID],
		.rpc_argp = &args,
		.rpc_resp = &res,
		.rpc_cred = cred,
	};
	__be32 *p;

	dprintk("--> %s\n", __func__);
	BUG_ON(clp == NULL);
4953

B
Benny Halevy 已提交
4954 4955 4956 4957 4958
	p = (u32 *)verifier.data;
	*p++ = htonl((u32)clp->cl_boot_time.tv_sec);
	*p = htonl((u32)clp->cl_boot_time.tv_nsec);
	args.verifier = &verifier;

4959 4960 4961 4962 4963 4964
	args.id_len = scnprintf(args.id, sizeof(args.id),
				"%s/%s.%s/%u",
				clp->cl_ipaddr,
				init_utsname()->nodename,
				init_utsname()->domainname,
				clp->cl_rpcclient->cl_auth->au_flavor);
B
Benny Halevy 已提交
4965

4966
	res.server_scope = kzalloc(sizeof(struct server_scope), GFP_KERNEL);
4967 4968 4969 4970
	if (unlikely(!res.server_scope)) {
		status = -ENOMEM;
		goto out;
	}
4971

4972 4973 4974 4975 4976 4977
	res.impl_id = kzalloc(sizeof(struct nfs41_impl_id), GFP_KERNEL);
	if (unlikely(!res.impl_id)) {
		status = -ENOMEM;
		goto out_server_scope;
	}

4978
	status = rpc_call_sync(clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
4979 4980
	if (!status)
		status = nfs4_check_cl_exchange_flags(clp->cl_exchange_flags);
4981

4982 4983 4984 4985 4986 4987 4988
	if (!status) {
		/* use the most recent implementation id */
		kfree(clp->impl_id);
		clp->impl_id = res.impl_id;
	} else
		kfree(res.impl_id);

4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999
	if (!status) {
		if (clp->server_scope &&
		    !nfs41_same_server_scope(clp->server_scope,
					     res.server_scope)) {
			dprintk("%s: server_scope mismatch detected\n",
				__func__);
			set_bit(NFS4CLNT_SERVER_SCOPE_MISMATCH, &clp->cl_state);
			kfree(clp->server_scope);
			clp->server_scope = NULL;
		}

5000
		if (!clp->server_scope) {
5001
			clp->server_scope = res.server_scope;
5002 5003
			goto out;
		}
5004
	}
5005 5006

out_server_scope:
5007 5008
	kfree(res.server_scope);
out:
5009 5010 5011 5012 5013 5014
	if (clp->impl_id)
		dprintk("%s: Server Implementation ID: "
			"domain: %s, name: %s, date: %llu,%u\n",
			__func__, clp->impl_id->domain, clp->impl_id->name,
			clp->impl_id->date.seconds,
			clp->impl_id->date.nseconds);
B
Benny Halevy 已提交
5015 5016 5017 5018
	dprintk("<-- %s status= %d\n", __func__, status);
	return status;
}

A
Andy Adamson 已提交
5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032
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)
{
	int ret;
	struct nfs4_get_lease_time_data *data =
			(struct nfs4_get_lease_time_data *)calldata;

	dprintk("--> %s\n", __func__);
5033
	rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
A
Andy Adamson 已提交
5034 5035 5036
	/* just setup sequence, do not trigger session recovery
	   since we're invoked within one */
	ret = nfs41_setup_sequence(data->clp->cl_session,
5037
				   &data->args->la_seq_args,
5038
				   &data->res->lr_seq_res, task);
A
Andy Adamson 已提交
5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054

	BUG_ON(ret == -EAGAIN);
	rpc_call_start(task);
	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__);
5055 5056
	if (!nfs41_sequence_done(task, &data->res->lr_seq_res))
		return;
A
Andy Adamson 已提交
5057 5058 5059 5060 5061 5062
	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;
5063 5064
		/* fall through */
	case -NFS4ERR_RETRY_UNCACHED_REP:
5065
		rpc_restart_call_prepare(task);
A
Andy Adamson 已提交
5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096
		return;
	}
	dprintk("<-- %s\n", __func__);
}

struct rpc_call_ops nfs4_get_lease_time_ops = {
	.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,
5097 5098
		.callback_data = &data,
		.flags = RPC_TASK_TIMEOUT,
A
Andy Adamson 已提交
5099 5100 5101
	};
	int status;

5102
	nfs41_init_sequence(&args.la_seq_args, &res.lr_seq_res, 0);
A
Andy Adamson 已提交
5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116
	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;
}

5117 5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142
static struct nfs4_slot *nfs4_alloc_slots(u32 max_slots, gfp_t gfp_flags)
{
	return kcalloc(max_slots, sizeof(struct nfs4_slot), gfp_flags);
}

static void nfs4_add_and_init_slots(struct nfs4_slot_table *tbl,
		struct nfs4_slot *new,
		u32 max_slots,
		u32 ivalue)
{
	struct nfs4_slot *old = NULL;
	u32 i;

	spin_lock(&tbl->slot_tbl_lock);
	if (new) {
		old = tbl->slots;
		tbl->slots = new;
		tbl->max_slots = max_slots;
	}
	tbl->highest_used_slotid = -1;	/* no slot is currently used */
	for (i = 0; i < tbl->max_slots; i++)
		tbl->slots[i].seq_nr = ivalue;
	spin_unlock(&tbl->slot_tbl_lock);
	kfree(old);
}

5143
/*
5144
 * (re)Initialise a slot table
5145
 */
5146 5147
static int nfs4_realloc_slot_table(struct nfs4_slot_table *tbl, u32 max_reqs,
				 u32 ivalue)
5148
{
A
Andy Adamson 已提交
5149
	struct nfs4_slot *new = NULL;
5150
	int ret = -ENOMEM;
5151

A
Andy Adamson 已提交
5152 5153
	dprintk("--> %s: max_reqs=%u, tbl->max_slots %d\n", __func__,
		max_reqs, tbl->max_slots);
5154

A
Andy Adamson 已提交
5155 5156
	/* Does the newly negotiated max_reqs match the existing slot table? */
	if (max_reqs != tbl->max_slots) {
5157
		new = nfs4_alloc_slots(max_reqs, GFP_NOFS);
A
Andy Adamson 已提交
5158 5159 5160
		if (!new)
			goto out;
	}
5161 5162 5163
	ret = 0;

	nfs4_add_and_init_slots(tbl, new, max_reqs, ivalue);
5164 5165 5166 5167 5168 5169 5170
	dprintk("%s: tbl=%p slots=%p max_slots=%d\n", __func__,
		tbl, tbl->slots, tbl->max_slots);
out:
	dprintk("<-- %s: return %d\n", __func__, ret);
	return ret;
}

5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184
/* Destroy the slot table */
static void nfs4_destroy_slot_tables(struct nfs4_session *session)
{
	if (session->fc_slot_table.slots != NULL) {
		kfree(session->fc_slot_table.slots);
		session->fc_slot_table.slots = NULL;
	}
	if (session->bc_slot_table.slots != NULL) {
		kfree(session->bc_slot_table.slots);
		session->bc_slot_table.slots = NULL;
	}
	return;
}

5185
/*
5186
 * Initialize or reset the forechannel and backchannel tables
5187
 */
5188
static int nfs4_setup_session_slot_tables(struct nfs4_session *ses)
5189
{
5190
	struct nfs4_slot_table *tbl;
5191
	int status;
5192

5193 5194 5195
	dprintk("--> %s\n", __func__);
	/* Fore channel */
	tbl = &ses->fc_slot_table;
5196 5197 5198
	status = nfs4_realloc_slot_table(tbl, ses->fc_attrs.max_reqs, 1);
	if (status) /* -ENOMEM */
		return status;
5199 5200
	/* Back channel */
	tbl = &ses->bc_slot_table;
5201 5202 5203 5204 5205
	status = nfs4_realloc_slot_table(tbl, ses->bc_attrs.max_reqs, 0);
	if (status && tbl->slots == NULL)
		/* Fore and back channel share a connection so get
		 * both slot tables or neither */
		nfs4_destroy_slot_tables(ses);
5206
	return status;
5207 5208 5209 5210 5211 5212 5213
}

struct nfs4_session *nfs4_alloc_session(struct nfs_client *clp)
{
	struct nfs4_session *session;
	struct nfs4_slot_table *tbl;

5214
	session = kzalloc(sizeof(struct nfs4_session), GFP_NOFS);
5215 5216
	if (!session)
		return NULL;
5217

5218
	tbl = &session->fc_slot_table;
5219
	tbl->highest_used_slotid = NFS4_NO_SLOT;
5220
	spin_lock_init(&tbl->slot_tbl_lock);
5221
	rpc_init_priority_wait_queue(&tbl->slot_tbl_waitq, "ForeChannel Slot table");
5222
	init_completion(&tbl->complete);
5223 5224

	tbl = &session->bc_slot_table;
5225
	tbl->highest_used_slotid = NFS4_NO_SLOT;
5226 5227
	spin_lock_init(&tbl->slot_tbl_lock);
	rpc_init_wait_queue(&tbl->slot_tbl_waitq, "BackChannel Slot table");
5228
	init_completion(&tbl->complete);
5229

5230 5231
	session->session_state = 1<<NFS4_SESSION_INITING;

5232 5233 5234 5235 5236 5237
	session->clp = clp;
	return session;
}

void nfs4_destroy_session(struct nfs4_session *session)
{
5238 5239
	struct rpc_xprt *xprt;

5240
	nfs4_proc_destroy_session(session);
5241 5242 5243 5244

	rcu_read_lock();
	xprt = rcu_dereference(session->clp->cl_rpcclient->cl_xprt);
	rcu_read_unlock();
5245
	dprintk("%s Destroy backchannel for xprt %p\n",
5246 5247
		__func__, xprt);
	xprt_destroy_backchannel(xprt, NFS41_BC_MIN_CALLBACKS);
5248
	nfs4_destroy_slot_tables(session);
5249 5250 5251
	kfree(session);
}

A
Andy Adamson 已提交
5252 5253 5254 5255 5256 5257 5258 5259 5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273 5274
/*
 * 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)
{
	struct nfs4_session *session = args->client->cl_session;
	unsigned int mxrqst_sz = session->fc_attrs.max_rqst_sz,
		     mxresp_sz = session->fc_attrs.max_resp_sz;

	if (mxrqst_sz == 0)
		mxrqst_sz = NFS_MAX_FILE_IO_SIZE;
	if (mxresp_sz == 0)
		mxresp_sz = NFS_MAX_FILE_IO_SIZE;
	/* Fore channel attributes */
	args->fc_attrs.max_rqst_sz = mxrqst_sz;
	args->fc_attrs.max_resp_sz = mxresp_sz;
	args->fc_attrs.max_ops = NFS4_MAX_OPS;
5275
	args->fc_attrs.max_reqs = max_session_slots;
A
Andy Adamson 已提交
5276 5277

	dprintk("%s: Fore Channel : max_rqst_sz=%u max_resp_sz=%u "
5278
		"max_ops=%u max_reqs=%u\n",
A
Andy Adamson 已提交
5279 5280
		__func__,
		args->fc_attrs.max_rqst_sz, args->fc_attrs.max_resp_sz,
5281
		args->fc_attrs.max_ops, args->fc_attrs.max_reqs);
A
Andy Adamson 已提交
5282 5283 5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294 5295 5296 5297

	/* 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);
}

5298
static int nfs4_verify_fore_channel_attrs(struct nfs41_create_session_args *args, struct nfs4_session *session)
5299
{
5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314
	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;
5315 5316
	if (rcvd->max_reqs > NFS4_MAX_SLOT_TABLE)
		rcvd->max_reqs = NFS4_MAX_SLOT_TABLE;
5317
	return 0;
5318 5319
}

5320 5321 5322 5323
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;
5324

5325 5326 5327 5328 5329 5330 5331
	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: */
5332
	if (rcvd->max_ops != sent->max_ops)
5333
		return -EINVAL;
5334
	if (rcvd->max_reqs != sent->max_reqs)
5335 5336 5337
		return -EINVAL;
	return 0;
}
5338 5339 5340 5341

static int nfs4_verify_channel_attrs(struct nfs41_create_session_args *args,
				     struct nfs4_session *session)
{
5342
	int ret;
5343

5344 5345 5346 5347
	ret = nfs4_verify_fore_channel_attrs(args, session);
	if (ret)
		return ret;
	return nfs4_verify_back_channel_attrs(args, session);
5348 5349
}

A
Andy Adamson 已提交
5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367
static int _nfs4_proc_create_session(struct nfs_client *clp)
{
	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,
	};
	int status;

	nfs4_init_channel_attrs(&args);
5368
	args.flags = (SESSION4_PERSIST | SESSION4_BACK_CHAN);
A
Andy Adamson 已提交
5369

5370
	status = rpc_call_sync(session->clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
A
Andy Adamson 已提交
5371

5372 5373 5374
	if (!status)
		/* Verify the session's negotiated channel_attrs values */
		status = nfs4_verify_channel_attrs(&args, session);
A
Andy Adamson 已提交
5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387
	if (!status) {
		/* 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.
 */
5388
int nfs4_proc_create_session(struct nfs_client *clp)
A
Andy Adamson 已提交
5389 5390 5391 5392 5393 5394 5395
{
	int status;
	unsigned *ptr;
	struct nfs4_session *session = clp->cl_session;

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

5396
	status = _nfs4_proc_create_session(clp);
A
Andy Adamson 已提交
5397 5398 5399
	if (status)
		goto out;

5400 5401 5402
	/* 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
5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413
	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
5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432
/*
 * Issue the over-the-wire RPC DESTROY_SESSION.
 * The caller must serialize access to this routine.
 */
int nfs4_proc_destroy_session(struct nfs4_session *session)
{
	int status = 0;
	struct rpc_message msg;

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

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

	msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DESTROY_SESSION];
	msg.rpc_argp = session;
	msg.rpc_resp = NULL;
	msg.rpc_cred = NULL;
5433
	status = rpc_call_sync(session->clp->cl_rpcclient, &msg, RPC_TASK_TIMEOUT);
A
Andy Adamson 已提交
5434 5435 5436

	if (status)
		printk(KERN_WARNING
5437
			"NFS: Got error %d from the server on DESTROY_SESSION. "
A
Andy Adamson 已提交
5438 5439 5440 5441 5442 5443
			"Session has been destroyed regardless...\n", status);

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

5444 5445 5446
int nfs4_init_session(struct nfs_server *server)
{
	struct nfs_client *clp = server->nfs_client;
5447
	struct nfs4_session *session;
5448
	unsigned int rsize, wsize;
5449 5450 5451 5452 5453
	int ret;

	if (!nfs4_has_session(clp))
		return 0;

5454 5455 5456 5457
	session = clp->cl_session;
	if (!test_and_clear_bit(NFS4_SESSION_INITING, &session->session_state))
		return 0;

5458 5459 5460 5461 5462 5463 5464 5465 5466
	rsize = server->rsize;
	if (rsize == 0)
		rsize = NFS_MAX_FILE_IO_SIZE;
	wsize = server->wsize;
	if (wsize == 0)
		wsize = NFS_MAX_FILE_IO_SIZE;

	session->fc_attrs.max_rqst_sz = wsize + nfs41_maxwrite_overhead;
	session->fc_attrs.max_resp_sz = rsize + nfs41_maxread_overhead;
5467

5468 5469 5470 5471 5472 5473
	ret = nfs4_recover_expired_lease(server);
	if (!ret)
		ret = nfs4_check_client_ready(clp);
	return ret;
}

A
Andy Adamson 已提交
5474 5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492 5493 5494
int nfs4_init_ds_session(struct nfs_client *clp)
{
	struct nfs4_session *session = clp->cl_session;
	int ret;

	if (!test_and_clear_bit(NFS4_SESSION_INITING, &session->session_state))
		return 0;

	ret = nfs4_client_recover_expired_lease(clp);
	if (!ret)
		/* Test for the DS role */
		if (!is_ds_client(clp))
			ret = -ENODEV;
	if (!ret)
		ret = nfs4_check_client_ready(clp);
	return ret;

}
EXPORT_SYMBOL_GPL(nfs4_init_ds_session);


A
Andy Adamson 已提交
5495 5496 5497
/*
 * Renew the cl_session lease.
 */
5498 5499 5500 5501 5502 5503
struct nfs4_sequence_data {
	struct nfs_client *clp;
	struct nfs4_sequence_args args;
	struct nfs4_sequence_res res;
};

5504 5505
static void nfs41_sequence_release(void *data)
{
5506 5507
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
5508

5509 5510 5511
	if (atomic_read(&clp->cl_count) > 1)
		nfs4_schedule_state_renewal(clp);
	nfs_put_client(clp);
5512
	kfree(calldata);
5513 5514
}

5515 5516 5517 5518 5519 5520 5521
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:
5522
		nfs4_schedule_lease_recovery(clp);
5523 5524 5525 5526
	}
	return 0;
}

5527
static void nfs41_sequence_call_done(struct rpc_task *task, void *data)
A
Andy Adamson 已提交
5528
{
5529 5530
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
A
Andy Adamson 已提交
5531

5532 5533
	if (!nfs41_sequence_done(task, task->tk_msg.rpc_resp))
		return;
A
Andy Adamson 已提交
5534 5535 5536

	if (task->tk_status < 0) {
		dprintk("%s ERROR %d\n", __func__, task->tk_status);
5537 5538
		if (atomic_read(&clp->cl_count) == 1)
			goto out;
A
Andy Adamson 已提交
5539

5540 5541
		if (nfs41_sequence_handle_errors(task, clp) == -EAGAIN) {
			rpc_restart_call_prepare(task);
A
Andy Adamson 已提交
5542 5543 5544 5545
			return;
		}
	}
	dprintk("%s rpc_cred %p\n", __func__, task->tk_msg.rpc_cred);
5546
out:
A
Andy Adamson 已提交
5547 5548 5549 5550 5551
	dprintk("<-- %s\n", __func__);
}

static void nfs41_sequence_prepare(struct rpc_task *task, void *data)
{
5552 5553
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
A
Andy Adamson 已提交
5554 5555 5556 5557 5558 5559
	struct nfs4_sequence_args *args;
	struct nfs4_sequence_res *res;

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

5560
	if (nfs41_setup_sequence(clp->cl_session, args, res, task))
A
Andy Adamson 已提交
5561 5562 5563 5564 5565 5566 5567
		return;
	rpc_call_start(task);
}

static const struct rpc_call_ops nfs41_sequence_ops = {
	.rpc_call_done = nfs41_sequence_call_done,
	.rpc_call_prepare = nfs41_sequence_prepare,
5568
	.rpc_release = nfs41_sequence_release,
A
Andy Adamson 已提交
5569 5570
};

5571
static struct rpc_task *_nfs41_proc_sequence(struct nfs_client *clp, struct rpc_cred *cred)
A
Andy Adamson 已提交
5572
{
5573
	struct nfs4_sequence_data *calldata;
A
Andy Adamson 已提交
5574 5575 5576 5577
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SEQUENCE],
		.rpc_cred = cred,
	};
5578 5579 5580 5581 5582 5583
	struct rpc_task_setup task_setup_data = {
		.rpc_client = clp->cl_rpcclient,
		.rpc_message = &msg,
		.callback_ops = &nfs41_sequence_ops,
		.flags = RPC_TASK_ASYNC | RPC_TASK_SOFT,
	};
A
Andy Adamson 已提交
5584

5585
	if (!atomic_inc_not_zero(&clp->cl_count))
5586
		return ERR_PTR(-EIO);
5587
	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
5588
	if (calldata == NULL) {
5589
		nfs_put_client(clp);
5590
		return ERR_PTR(-ENOMEM);
A
Andy Adamson 已提交
5591
	}
5592
	nfs41_init_sequence(&calldata->args, &calldata->res, 0);
5593 5594 5595
	msg.rpc_argp = &calldata->args;
	msg.rpc_resp = &calldata->res;
	calldata->clp = clp;
5596
	task_setup_data.callback_data = calldata;
A
Andy Adamson 已提交
5597

5598 5599 5600
	return rpc_run_task(&task_setup_data);
}

5601
static int nfs41_proc_async_sequence(struct nfs_client *clp, struct rpc_cred *cred, unsigned renew_flags)
5602 5603 5604 5605
{
	struct rpc_task *task;
	int ret = 0;

5606 5607
	if ((renew_flags & NFS4_RENEW_TIMEOUT) == 0)
		return 0;
5608 5609 5610 5611
	task = _nfs41_proc_sequence(clp, cred);
	if (IS_ERR(task))
		ret = PTR_ERR(task);
	else
5612
		rpc_put_task_async(task);
5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627
	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;

	task = _nfs41_proc_sequence(clp, cred);
	if (IS_ERR(task)) {
		ret = PTR_ERR(task);
		goto out;
	}
	ret = rpc_wait_for_completion_task(task);
5628 5629 5630 5631 5632
	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);
5633
		ret = task->tk_status;
5634
	}
5635 5636 5637 5638
	rpc_put_task(task);
out:
	dprintk("<-- %s status=%d\n", __func__, ret);
	return ret;
A
Andy Adamson 已提交
5639 5640
}

5641 5642 5643 5644 5645 5646 5647 5648 5649 5650
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;

5651
	rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
5652 5653
	if (nfs41_setup_sequence(calldata->clp->cl_session,
				&calldata->arg.seq_args,
5654
				&calldata->res.seq_res, task))
5655 5656 5657 5658 5659
		return;

	rpc_call_start(task);
}

5660 5661 5662 5663 5664 5665 5666 5667 5668
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);
5669 5670
		/* fall through */
	case -NFS4ERR_RETRY_UNCACHED_REP:
5671 5672
		return -EAGAIN;
	default:
5673
		nfs4_schedule_lease_recovery(clp);
5674 5675 5676 5677
	}
	return 0;
}

5678 5679 5680 5681 5682 5683 5684
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__);
5685 5686
	if (!nfs41_sequence_done(task, res))
		return;
5687

5688 5689 5690 5691
	if (nfs41_reclaim_complete_handle_errors(task, clp) == -EAGAIN) {
		rpc_restart_call_prepare(task);
		return;
	}
5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726
	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.
 */
static int nfs41_proc_reclaim_complete(struct nfs_client *clp)
{
	struct nfs4_reclaim_complete_data *calldata;
	struct rpc_task *task;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RECLAIM_COMPLETE],
	};
	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__);
5727
	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
5728 5729 5730 5731 5732
	if (calldata == NULL)
		goto out;
	calldata->clp = clp;
	calldata->arg.one_fs = 0;

5733
	nfs41_init_sequence(&calldata->arg.seq_args, &calldata->res.seq_res, 0);
5734 5735 5736 5737
	msg.rpc_argp = &calldata->arg;
	msg.rpc_resp = &calldata->res;
	task_setup_data.callback_data = calldata;
	task = rpc_run_task(&task_setup_data);
5738
	if (IS_ERR(task)) {
5739
		status = PTR_ERR(task);
5740 5741
		goto out;
	}
5742 5743 5744
	status = nfs4_wait_for_completion_rpc_task(task);
	if (status == 0)
		status = task->tk_status;
5745
	rpc_put_task(task);
5746
	return 0;
5747 5748 5749 5750
out:
	dprintk("<-- %s status=%d\n", __func__, status);
	return status;
}
5751 5752 5753 5754 5755

static void
nfs4_layoutget_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;
F
Fred Isaman 已提交
5756
	struct nfs_server *server = NFS_SERVER(lgp->args.inode);
5757 5758

	dprintk("--> %s\n", __func__);
F
Fred Isaman 已提交
5759 5760 5761 5762 5763
	/* 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.
	 */
5764
	if (nfs4_setup_sequence(server, &lgp->args.seq_args,
5765
				&lgp->res.seq_res, task))
5766
		return;
5767 5768 5769 5770 5771 5772
	if (pnfs_choose_layoutget_stateid(&lgp->args.stateid,
					  NFS_I(lgp->args.inode)->layout,
					  lgp->args.ctx->state)) {
		rpc_exit(task, NFS4_OK);
		return;
	}
5773 5774 5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795 5796 5797 5798 5799 5800 5801 5802 5803 5804 5805 5806 5807 5808 5809 5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824 5825 5826 5827 5828 5829 5830 5831 5832 5833 5834 5835 5836 5837
	rpc_call_start(task);
}

static void nfs4_layoutget_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;
	struct nfs_server *server = NFS_SERVER(lgp->args.inode);

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

	if (!nfs4_sequence_done(task, &lgp->res.seq_res))
		return;

	switch (task->tk_status) {
	case 0:
		break;
	case -NFS4ERR_LAYOUTTRYLATER:
	case -NFS4ERR_RECALLCONFLICT:
		task->tk_status = -NFS4ERR_DELAY;
		/* Fall through */
	default:
		if (nfs4_async_handle_error(task, server, NULL) == -EAGAIN) {
			rpc_restart_call_prepare(task);
			return;
		}
	}
	dprintk("<-- %s\n", __func__);
}

static void nfs4_layoutget_release(void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;

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

int nfs4_proc_layoutget(struct nfs4_layoutget *lgp)
{
	struct nfs_server *server = NFS_SERVER(lgp->args.inode);
	struct rpc_task *task;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LAYOUTGET],
		.rpc_argp = &lgp->args,
		.rpc_resp = &lgp->res,
	};
	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,
	};
	int status = 0;

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

5838
	lgp->res.layoutp = &lgp->args.layout;
5839
	lgp->res.seq_res.sr_slot = NULL;
5840
	nfs41_init_sequence(&lgp->args.seq_args, &lgp->res.seq_res, 0);
5841 5842 5843 5844
	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task))
		return PTR_ERR(task);
	status = nfs4_wait_for_completion_rpc_task(task);
F
Fred Isaman 已提交
5845 5846 5847 5848
	if (status == 0)
		status = task->tk_status;
	if (status == 0)
		status = pnfs_layout_process(lgp);
5849 5850 5851 5852 5853
	rpc_put_task(task);
	dprintk("<-- %s status=%d\n", __func__, status);
	return status;
}

B
Benny Halevy 已提交
5854 5855 5856 5857 5858 5859 5860
static void
nfs4_layoutreturn_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutreturn *lrp = calldata;

	dprintk("--> %s\n", __func__);
	if (nfs41_setup_sequence(lrp->clp->cl_session, &lrp->args.seq_args,
5861
				&lrp->res.seq_res, task))
B
Benny Halevy 已提交
5862 5863 5864 5865 5866 5867 5868 5869
		return;
	rpc_call_start(task);
}

static void nfs4_layoutreturn_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutreturn *lrp = calldata;
	struct nfs_server *server;
T
Trond Myklebust 已提交
5870
	struct pnfs_layout_hdr *lo = lrp->args.layout;
B
Benny Halevy 已提交
5871 5872 5873 5874 5875 5876 5877 5878

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

	if (!nfs4_sequence_done(task, &lrp->res.seq_res))
		return;

	server = NFS_SERVER(lrp->args.inode);
	if (nfs4_async_handle_error(task, server, NULL) == -EAGAIN) {
5879
		rpc_restart_call_prepare(task);
B
Benny Halevy 已提交
5880 5881
		return;
	}
5882
	spin_lock(&lo->plh_inode->i_lock);
B
Benny Halevy 已提交
5883 5884 5885 5886 5887 5888
	if (task->tk_status == 0) {
		if (lrp->res.lrs_present) {
			pnfs_set_layout_stateid(lo, &lrp->res.stateid, true);
		} else
			BUG_ON(!list_empty(&lo->plh_segs));
	}
5889 5890
	lo->plh_block_lgets--;
	spin_unlock(&lo->plh_inode->i_lock);
B
Benny Halevy 已提交
5891 5892 5893 5894 5895 5896 5897 5898
	dprintk("<-- %s\n", __func__);
}

static void nfs4_layoutreturn_release(void *calldata)
{
	struct nfs4_layoutreturn *lrp = calldata;

	dprintk("--> %s\n", __func__);
T
Trond Myklebust 已提交
5899
	put_layout_hdr(lrp->args.layout);
B
Benny Halevy 已提交
5900 5901 5902 5903 5904 5905 5906 5907 5908 5909 5910 5911 5912 5913 5914 5915 5916 5917 5918 5919 5920 5921 5922 5923 5924 5925 5926
	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,
	};
	struct rpc_task_setup task_setup_data = {
		.rpc_client = lrp->clp->cl_rpcclient,
		.rpc_message = &msg,
		.callback_ops = &nfs4_layoutreturn_call_ops,
		.callback_data = lrp,
	};
	int status;

	dprintk("--> %s\n", __func__);
5927
	nfs41_init_sequence(&lrp->args.seq_args, &lrp->res.seq_res, 1);
B
Benny Halevy 已提交
5928 5929 5930 5931 5932 5933 5934 5935 5936
	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task))
		return PTR_ERR(task);
	status = task->tk_status;
	dprintk("<-- %s status=%d\n", __func__, status);
	rpc_put_task(task);
	return status;
}

A
Andy Adamson 已提交
5937 5938 5939 5940 5941 5942 5943 5944 5945 5946 5947 5948 5949 5950 5951 5952 5953 5954 5955 5956 5957 5958 5959 5960 5961 5962 5963 5964 5965 5966 5967 5968 5969 5970 5971 5972 5973 5974 5975 5976 5977 5978 5979 5980 5981 5982 5983 5984
/*
 * 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);

5985 5986 5987 5988 5989 5990 5991 5992 5993 5994 5995 5996 5997 5998 5999 6000 6001
static int
_nfs4_proc_getdeviceinfo(struct nfs_server *server, struct pnfs_device *pdev)
{
	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,
	};
	int status;

	dprintk("--> %s\n", __func__);
6002
	status = nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
6003 6004 6005 6006 6007 6008 6009 6010 6011 6012 6013 6014 6015 6016 6017 6018 6019 6020 6021
	dprintk("<-- %s status=%d\n", __func__, status);

	return status;
}

int nfs4_proc_getdeviceinfo(struct nfs_server *server, struct pnfs_device *pdev)
{
	struct nfs4_exception exception = { };
	int err;

	do {
		err = nfs4_handle_exception(server,
					_nfs4_proc_getdeviceinfo(server, pdev),
					&exception);
	} while (exception.retry);
	return err;
}
EXPORT_SYMBOL_GPL(nfs4_proc_getdeviceinfo);

A
Andy Adamson 已提交
6022 6023 6024 6025 6026 6027
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);

	if (nfs4_setup_sequence(server, &data->args.seq_args,
6028
				&data->res.seq_res, task))
A
Andy Adamson 已提交
6029 6030 6031 6032 6033 6034 6035 6036 6037 6038 6039 6040 6041 6042 6043 6044 6045 6046 6047 6048 6049 6050
		return;
	rpc_call_start(task);
}

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);

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

	switch (task->tk_status) { /* Just ignore these failures */
	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 */
		task->tk_status = 0;
	}

	if (nfs4_async_handle_error(task, server, NULL) == -EAGAIN) {
6051
		rpc_restart_call_prepare(task);
A
Andy Adamson 已提交
6052 6053 6054 6055 6056 6057 6058 6059 6060 6061 6062
		return;
	}

	if (task->tk_status == 0)
		nfs_post_op_update_inode_force_wcc(data->args.inode,
						   data->res.fattr);
}

static void nfs4_layoutcommit_release(void *calldata)
{
	struct nfs4_layoutcommit_data *data = calldata;
6063
	struct pnfs_layout_segment *lseg, *tmp;
P
Peng Tao 已提交
6064
	unsigned long *bitlock = &NFS_I(data->args.inode)->flags;
A
Andy Adamson 已提交
6065

A
Andy Adamson 已提交
6066
	pnfs_cleanup_layoutcommit(data);
A
Andy Adamson 已提交
6067
	/* Matched by references in pnfs_set_layoutcommit */
6068 6069 6070 6071 6072 6073
	list_for_each_entry_safe(lseg, tmp, &data->lseg_list, pls_lc_list) {
		list_del_init(&lseg->pls_lc_list);
		if (test_and_clear_bit(NFS_LSEG_LAYOUTCOMMIT,
				       &lseg->pls_flags))
			put_lseg(lseg);
	}
P
Peng Tao 已提交
6074 6075 6076 6077 6078

	clear_bit_unlock(NFS_INO_LAYOUTCOMMITTING, bitlock);
	smp_mb__after_clear_bit();
	wake_up_bit(bitlock, NFS_INO_LAYOUTCOMMITTING);

A
Andy Adamson 已提交
6079 6080 6081 6082 6083 6084 6085 6086 6087 6088 6089
	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
6090
nfs4_proc_layoutcommit(struct nfs4_layoutcommit_data *data, bool sync)
A
Andy Adamson 已提交
6091 6092 6093 6094 6095 6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106 6107 6108 6109 6110 6111 6112 6113 6114
{
	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);

6115
	nfs41_init_sequence(&data->args.seq_args, &data->res.seq_res, 1);
A
Andy Adamson 已提交
6116 6117 6118
	task = rpc_run_task(&task_setup_data);
	if (IS_ERR(task))
		return PTR_ERR(task);
6119
	if (sync == false)
A
Andy Adamson 已提交
6120 6121 6122 6123 6124 6125 6126 6127 6128 6129
		goto out;
	status = nfs4_wait_for_completion_rpc_task(task);
	if (status != 0)
		goto out;
	status = task->tk_status;
out:
	dprintk("%s: status %d\n", __func__, status);
	rpc_put_task(task);
	return status;
}
6130 6131 6132 6133 6134 6135 6136 6137 6138 6139 6140 6141 6142 6143 6144 6145 6146 6147 6148 6149 6150 6151 6152 6153 6154 6155 6156 6157 6158 6159 6160 6161 6162 6163 6164 6165 6166 6167 6168 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

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 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,
	};
	return nfs4_call_sync(server->client, server, &msg, &args.seq_args, &res.seq_res, 0);
}

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 {
		err = _nfs41_proc_secinfo_no_name(server, fhandle, info, flavors);
		switch (err) {
		case 0:
		case -NFS4ERR_WRONGSEC:
		case -NFS4ERR_NOTSUPP:
			break;
		default:
			err = nfs4_handle_exception(server, err, &exception);
		}
	} while (exception.retry);
	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;
}
6209 6210

static int _nfs41_test_stateid(struct nfs_server *server, nfs4_stateid *stateid)
B
Bryan Schumaker 已提交
6211
{
6212
	int status;
B
Bryan Schumaker 已提交
6213
	struct nfs41_test_stateid_args args = {
6214
		.stateid = stateid,
B
Bryan Schumaker 已提交
6215 6216 6217 6218 6219 6220 6221
	};
	struct nfs41_test_stateid_res res;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_TEST_STATEID],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
6222

6223
	nfs41_init_sequence(&args.seq_args, &res.seq_res, 0);
6224 6225 6226 6227 6228
	status = nfs4_call_sync_sequence(server->client, server, &msg, &args.seq_args, &res.seq_res, 1);

	if (status == NFS_OK)
		return res.status;
	return status;
B
Bryan Schumaker 已提交
6229 6230
}

6231
static int nfs41_test_stateid(struct nfs_server *server, nfs4_stateid *stateid)
B
Bryan Schumaker 已提交
6232 6233 6234 6235 6236
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(server,
6237
				_nfs41_test_stateid(server, stateid),
B
Bryan Schumaker 已提交
6238 6239 6240 6241
				&exception);
	} while (exception.retry);
	return err;
}
B
Bryan Schumaker 已提交
6242

6243
static int _nfs4_free_stateid(struct nfs_server *server, nfs4_stateid *stateid)
B
Bryan Schumaker 已提交
6244 6245
{
	struct nfs41_free_stateid_args args = {
6246
		.stateid = stateid,
B
Bryan Schumaker 已提交
6247 6248 6249 6250 6251 6252 6253 6254
	};
	struct nfs41_free_stateid_res res;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FREE_STATEID],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};

6255 6256
	nfs41_init_sequence(&args.seq_args, &res.seq_res, 0);
	return nfs4_call_sync_sequence(server->client, server, &msg, &args.seq_args, &res.seq_res, 1);
B
Bryan Schumaker 已提交
6257 6258
}

6259
static int nfs41_free_stateid(struct nfs_server *server, nfs4_stateid *stateid)
B
Bryan Schumaker 已提交
6260 6261 6262 6263 6264
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(server,
6265
				_nfs4_free_stateid(server, stateid),
B
Bryan Schumaker 已提交
6266 6267 6268 6269
				&exception);
	} while (exception.retry);
	return err;
}
6270 6271 6272 6273

static bool nfs41_match_stateid(const nfs4_stateid *s1,
		const nfs4_stateid *s2)
{
6274
	if (memcmp(s1->other, s2->other, sizeof(s1->other)) != 0)
6275 6276
		return false;

6277
	if (s1->seqid == s2->seqid)
6278
		return true;
6279
	if (s1->seqid == 0 || s2->seqid == 0)
6280 6281 6282 6283 6284
		return true;

	return false;
}

6285 6286
#endif /* CONFIG_NFS_V4_1 */

6287 6288 6289
static bool nfs4_match_stateid(const nfs4_stateid *s1,
		const nfs4_stateid *s2)
{
6290
	return nfs4_stateid_match(s1, s2);
6291 6292 6293
}


6294
struct nfs4_state_recovery_ops nfs40_reboot_recovery_ops = {
6295
	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
6296
	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
L
Linus Torvalds 已提交
6297 6298
	.recover_open	= nfs4_open_reclaim,
	.recover_lock	= nfs4_lock_reclaim,
6299
	.establish_clid = nfs4_init_clientid,
6300
	.get_clid_cred	= nfs4_get_setclientid_cred,
L
Linus Torvalds 已提交
6301 6302
};

6303 6304 6305 6306 6307 6308
#if defined(CONFIG_NFS_V4_1)
struct nfs4_state_recovery_ops nfs41_reboot_recovery_ops = {
	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
	.recover_open	= nfs4_open_reclaim,
	.recover_lock	= nfs4_lock_reclaim,
6309
	.establish_clid = nfs41_init_clientid,
6310
	.get_clid_cred	= nfs4_get_exchange_id_cred,
6311
	.reclaim_complete = nfs41_proc_reclaim_complete,
6312 6313 6314 6315 6316 6317 6318 6319 6320
};
#endif /* CONFIG_NFS_V4_1 */

struct nfs4_state_recovery_ops nfs40_nograce_recovery_ops = {
	.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,
6321
	.get_clid_cred	= nfs4_get_setclientid_cred,
6322 6323 6324 6325
};

#if defined(CONFIG_NFS_V4_1)
struct nfs4_state_recovery_ops nfs41_nograce_recovery_ops = {
6326
	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
6327
	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
6328 6329
	.recover_open	= nfs41_open_expired,
	.recover_lock	= nfs41_lock_expired,
6330
	.establish_clid = nfs41_init_clientid,
6331
	.get_clid_cred	= nfs4_get_exchange_id_cred,
L
Linus Torvalds 已提交
6332
};
6333
#endif /* CONFIG_NFS_V4_1 */
L
Linus Torvalds 已提交
6334

B
Benny Halevy 已提交
6335 6336
struct nfs4_state_maintenance_ops nfs40_state_renewal_ops = {
	.sched_state_renewal = nfs4_proc_async_renew,
6337
	.get_state_renewal_cred_locked = nfs4_get_renew_cred_locked,
6338
	.renew_lease = nfs4_proc_renew,
B
Benny Halevy 已提交
6339 6340 6341 6342 6343
};

#if defined(CONFIG_NFS_V4_1)
struct nfs4_state_maintenance_ops nfs41_state_renewal_ops = {
	.sched_state_renewal = nfs41_proc_async_sequence,
6344
	.get_state_renewal_cred_locked = nfs4_get_machine_cred_locked,
6345
	.renew_lease = nfs4_proc_sequence,
B
Benny Halevy 已提交
6346 6347 6348
};
#endif

6349 6350 6351
static const struct nfs4_minor_version_ops nfs_v4_0_minor_ops = {
	.minor_version = 0,
	.call_sync = _nfs4_call_sync,
6352
	.match_stateid = nfs4_match_stateid,
6353
	.find_root_sec = nfs4_find_root_sec,
6354 6355 6356
	.reboot_recovery_ops = &nfs40_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs40_nograce_recovery_ops,
	.state_renewal_ops = &nfs40_state_renewal_ops,
6357 6358 6359 6360 6361 6362
};

#if defined(CONFIG_NFS_V4_1)
static const struct nfs4_minor_version_ops nfs_v4_1_minor_ops = {
	.minor_version = 1,
	.call_sync = _nfs4_call_sync_session,
6363
	.match_stateid = nfs41_match_stateid,
6364
	.find_root_sec = nfs41_find_root_sec,
6365 6366 6367
	.reboot_recovery_ops = &nfs41_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs41_nograce_recovery_ops,
	.state_renewal_ops = &nfs41_state_renewal_ops,
6368 6369 6370 6371 6372 6373 6374 6375 6376 6377
};
#endif

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
};

6378
static const struct inode_operations nfs4_file_inode_operations = {
6379 6380 6381
	.permission	= nfs_permission,
	.getattr	= nfs_getattr,
	.setattr	= nfs_setattr,
6382 6383 6384 6385
	.getxattr	= generic_getxattr,
	.setxattr	= generic_setxattr,
	.listxattr	= generic_listxattr,
	.removexattr	= generic_removexattr,
6386 6387
};

D
David Howells 已提交
6388
const struct nfs_rpc_ops nfs_v4_clientops = {
L
Linus Torvalds 已提交
6389 6390 6391
	.version	= 4,			/* protocol version */
	.dentry_ops	= &nfs4_dentry_operations,
	.dir_inode_ops	= &nfs4_dir_inode_operations,
6392
	.file_inode_ops	= &nfs4_file_inode_operations,
6393
	.file_ops	= &nfs4_file_operations,
L
Linus Torvalds 已提交
6394 6395 6396 6397 6398 6399 6400 6401 6402 6403 6404
	.getroot	= nfs4_proc_get_root,
	.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,
	.unlink_done	= nfs4_proc_unlink_done,
	.rename		= nfs4_proc_rename,
6405 6406
	.rename_setup	= nfs4_proc_rename_setup,
	.rename_done	= nfs4_proc_rename_done,
L
Linus Torvalds 已提交
6407 6408 6409 6410 6411 6412 6413 6414 6415
	.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,
6416
	.set_capabilities = nfs4_server_capabilities,
L
Linus Torvalds 已提交
6417 6418
	.decode_dirent	= nfs4_decode_dirent,
	.read_setup	= nfs4_proc_read_setup,
T
Trond Myklebust 已提交
6419
	.read_done	= nfs4_read_done,
L
Linus Torvalds 已提交
6420
	.write_setup	= nfs4_proc_write_setup,
6421
	.write_done	= nfs4_write_done,
L
Linus Torvalds 已提交
6422
	.commit_setup	= nfs4_proc_commit_setup,
6423
	.commit_done	= nfs4_commit_done,
L
Linus Torvalds 已提交
6424
	.lock		= nfs4_proc_lock,
6425
	.clear_acl_cache = nfs4_zap_acl_attr,
T
Trond Myklebust 已提交
6426
	.close_context  = nfs4_close_context,
6427
	.open_context	= nfs4_atomic_open,
6428
	.init_client	= nfs4_init_client,
B
Bryan Schumaker 已提交
6429
	.secinfo	= nfs4_proc_secinfo,
L
Linus Torvalds 已提交
6430 6431
};

6432 6433 6434 6435 6436 6437 6438 6439 6440 6441 6442 6443
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,
	NULL
};

6444 6445 6446 6447
module_param(max_session_slots, ushort, 0644);
MODULE_PARM_DESC(max_session_slots, "Maximum number of outstanding NFSv4.1 "
		"requests the client will negotiate");

L
Linus Torvalds 已提交
6448 6449 6450 6451 6452
/*
 * Local variables:
 *  c-basic-offset: 8
 * End:
 */