nfs4proc.c 150.1 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
#include <linux/slab.h>
L
Linus Torvalds 已提交
43 44 45 46 47 48
#include <linux/sunrpc/clnt.h>
#include <linux/nfs.h>
#include <linux/nfs4.h>
#include <linux/nfs_fs.h>
#include <linux/nfs_page.h>
#include <linux/namei.h>
49
#include <linux/mount.h>
B
Benny Halevy 已提交
50
#include <linux/module.h>
51
#include <linux/sunrpc/bc_xprt.h>
52
#include <linux/xattr.h>
53
#include <linux/utsname.h>
54
#include <linux/mm.h>
L
Linus Torvalds 已提交
55

56
#include "nfs4_fs.h"
L
Linus Torvalds 已提交
57
#include "delegation.h"
58
#include "internal.h"
59
#include "iostat.h"
A
Andy Adamson 已提交
60
#include "callback.h"
61
#include "pnfs.h"
L
Linus Torvalds 已提交
62 63 64

#define NFSDBG_FACILITY		NFSDBG_PROC

65
#define NFS4_POLL_RETRY_MIN	(HZ/10)
L
Linus Torvalds 已提交
66 67
#define NFS4_POLL_RETRY_MAX	(15*HZ)

68 69
#define NFS4_MAX_LOOP_ON_RECOVER (10)

70
struct nfs4_opendata;
71
static int _nfs4_proc_open(struct nfs4_opendata *data);
72
static int _nfs4_recover_proc_open(struct nfs4_opendata *data);
L
Linus Torvalds 已提交
73
static int nfs4_do_fsinfo(struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *);
74
static int nfs4_async_handle_error(struct rpc_task *, const struct nfs_server *, struct nfs4_state *);
75 76
static int _nfs4_proc_lookup(struct inode *dir, const struct qstr *name, struct nfs_fh *fhandle, struct nfs_fattr *fattr);
static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fattr *fattr);
77 78 79
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
80 81

/* Prevent leaks of NFSv4 errors into userland */
82
static int nfs4_map_errors(int err)
L
Linus Torvalds 已提交
83
{
84 85 86 87 88 89
	if (err >= -1000)
		return err;
	switch (err) {
	case -NFS4ERR_RESOURCE:
		return -EREMOTEIO;
	default:
L
Linus Torvalds 已提交
90
		dprintk("%s could not handle NFSv4 error %d\n",
91
				__func__, -err);
92
		break;
L
Linus Torvalds 已提交
93
	}
94
	return -EIO;
L
Linus Torvalds 已提交
95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125
}

/*
 * 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
};

126
const u32 nfs4_pathconf_bitmap[2] = {
L
Linus Torvalds 已提交
127 128 129 130 131 132 133 134 135
	FATTR4_WORD0_MAXLINK
	| FATTR4_WORD0_MAXNAME,
	0
};

const u32 nfs4_fsinfo_bitmap[2] = { FATTR4_WORD0_MAXFILESIZE
			| FATTR4_WORD0_MAXREAD
			| FATTR4_WORD0_MAXWRITE
			| FATTR4_WORD0_LEASE_TIME,
136
			FATTR4_WORD1_TIME_DELTA
137
			| FATTR4_WORD1_FS_LAYOUT_TYPES
L
Linus Torvalds 已提交
138 139
};

140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158
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
159
static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dentry,
L
Linus Torvalds 已提交
160 161
		struct nfs4_readdir_arg *readdir)
{
162
	__be32 *start, *p;
L
Linus Torvalds 已提交
163 164 165

	BUG_ON(readdir->count < 80);
	if (cookie > 2) {
166
		readdir->cookie = cookie;
L
Linus Torvalds 已提交
167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182
		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.
	 */
183
	start = p = kmap_atomic(*readdir->pages, KM_USER0);
L
Linus Torvalds 已提交
184 185 186 187 188 189 190 191 192 193 194
	
	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
195
		p = xdr_encode_hyper(p, NFS_FILEID(dentry->d_inode));
L
Linus Torvalds 已提交
196 197 198 199 200 201 202 203 204 205 206
	}
	
	*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
207
	p = xdr_encode_hyper(p, NFS_FILEID(dentry->d_parent->d_inode));
L
Linus Torvalds 已提交
208 209 210 211 212 213

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

214 215 216 217 218 219
static int nfs4_wait_clnt_recover(struct nfs_client *clp)
{
	int res;

	might_sleep();

220
	res = wait_on_bit(&clp->cl_state, NFS4CLNT_MANAGER_RUNNING,
221
			nfs_wait_bit_killable, TASK_KILLABLE);
222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247
	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;
	schedule_timeout_killable(*timeout);
	if (fatal_signal_pending(current))
		res = -ERESTARTSYS;
	*timeout <<= 1;
	return res;
}

/* This is the error handling routine for processes that are allowed
 * to sleep.
 */
static int nfs4_handle_exception(const struct nfs_server *server, int errorcode, struct nfs4_exception *exception)
{
	struct nfs_client *clp = server->nfs_client;
248
	struct nfs4_state *state = exception->state;
249 250 251 252 253 254
	int ret = errorcode;

	exception->retry = 0;
	switch(errorcode) {
		case 0:
			return 0;
255 256 257 258 259
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OPENMODE:
			if (state == NULL)
				break;
260 261
			nfs4_schedule_stateid_recovery(server, state);
			goto wait_on_recovery;
262
		case -NFS4ERR_STALE_STATEID:
263
		case -NFS4ERR_STALE_CLIENTID:
264
		case -NFS4ERR_EXPIRED:
265 266
			nfs4_schedule_lease_recovery(clp);
			goto wait_on_recovery;
267
#if defined(CONFIG_NFS_V4_1)
268 269 270 271 272 273 274 275 276
		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);
277
			nfs4_schedule_session_recovery(clp->cl_session);
278
			exception->retry = 1;
279
			break;
280
#endif /* defined(CONFIG_NFS_V4_1) */
281
		case -NFS4ERR_FILE_OPEN:
282 283 284 285 286 287 288
			if (exception->timeout > HZ) {
				/* We have retried a decent amount, time to
				 * fail
				 */
				ret = -EBUSY;
				break;
			}
289 290
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
291
		case -EKEYEXPIRED:
292 293 294 295 296 297 298 299
			ret = nfs4_delay(server->client, &exception->timeout);
			if (ret != 0)
				break;
		case -NFS4ERR_OLD_STATEID:
			exception->retry = 1;
	}
	/* We failed to handle the error */
	return nfs4_map_errors(ret);
300
wait_on_recovery:
301 302 303 304
	ret = nfs4_wait_clnt_recover(clp);
	if (ret == 0)
		exception->retry = 1;
	return ret;
305 306 307
}


308
static void do_renew_lease(struct nfs_client *clp, unsigned long timestamp)
L
Linus Torvalds 已提交
309 310 311 312 313 314 315
{
	spin_lock(&clp->cl_lock);
	if (time_before(clp->cl_last_renewal,timestamp))
		clp->cl_last_renewal = timestamp;
	spin_unlock(&clp->cl_lock);
}

316 317 318 319 320
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
321 322
#if defined(CONFIG_NFS_V4_1)

A
Andy Adamson 已提交
323 324 325 326 327 328 329 330 331 332 333 334
/*
 * 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.
 * If none found, highest_used_slotid is set to -1.
335 336
 *
 * Must be called while holding tbl->slot_tbl_lock
A
Andy Adamson 已提交
337 338
 */
static void
339
nfs4_free_slot(struct nfs4_slot_table *tbl, struct nfs4_slot *free_slot)
A
Andy Adamson 已提交
340
{
341
	int free_slotid = free_slot - tbl->slots;
A
Andy Adamson 已提交
342 343
	int slotid = free_slotid;

344
	BUG_ON(slotid < 0 || slotid >= NFS4_MAX_SLOT_TABLE);
A
Andy Adamson 已提交
345 346 347 348 349 350
	/* 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
351
		if (slotid < tbl->max_slots)
A
Andy Adamson 已提交
352 353 354 355 356 357 358 359
			tbl->highest_used_slotid = slotid;
		else
			tbl->highest_used_slotid = -1;
	}
	dprintk("%s: free_slotid %u highest_used_slotid %d\n", __func__,
		free_slotid, tbl->highest_used_slotid);
}

360
/*
361
 * Signal state manager thread if session fore channel is drained
362
 */
363
static void nfs4_check_drain_fc_complete(struct nfs4_session *ses)
364
{
365 366
	struct rpc_task *task;

367
	if (!test_bit(NFS4_SESSION_DRAINING, &ses->session_state)) {
368 369 370
		task = rpc_wake_up_next(&ses->fc_slot_table.slot_tbl_waitq);
		if (task)
			rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
371 372 373 374 375 376
		return;
	}

	if (ses->fc_slot_table.highest_used_slotid != -1)
		return;

377 378 379 380 381 382 383 384 385 386 387 388 389 390
	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) ||
	    ses->bc_slot_table.highest_used_slotid != -1)
		return;
	dprintk("%s COMPLETE: Session Back Channel Drained\n", __func__);
	complete(&ses->bc_slot_table.complete);
391 392
}

393
static void nfs41_sequence_free_slot(struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
394 395 396
{
	struct nfs4_slot_table *tbl;

397
	tbl = &res->sr_session->fc_slot_table;
398
	if (!res->sr_slot) {
A
Andy Adamson 已提交
399 400
		/* just wake up the next guy waiting since
		 * we may have not consumed a slot after all */
A
Andy Adamson 已提交
401
		dprintk("%s: No slot\n", __func__);
402
		return;
A
Andy Adamson 已提交
403
	}
404

405
	spin_lock(&tbl->slot_tbl_lock);
406
	nfs4_free_slot(tbl, res->sr_slot);
407
	nfs4_check_drain_fc_complete(res->sr_session);
408
	spin_unlock(&tbl->slot_tbl_lock);
409
	res->sr_slot = NULL;
A
Andy Adamson 已提交
410 411
}

412
static int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
413 414
{
	unsigned long timestamp;
415
	struct nfs_client *clp;
A
Andy Adamson 已提交
416 417 418 419 420 421 422 423 424 425 426

	/*
	 * 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;

	/* -ERESTARTSYS can result in skipping nfs41_sequence_setup */
427
	if (!res->sr_slot)
A
Andy Adamson 已提交
428 429
		goto out;

A
Andy Adamson 已提交
430
	/* Check the SEQUENCE operation status */
431 432
	switch (res->sr_status) {
	case 0:
A
Andy Adamson 已提交
433
		/* Update the slot's sequence and clientid lease timer */
434
		++res->sr_slot->seq_nr;
A
Andy Adamson 已提交
435
		timestamp = res->sr_renewal_time;
436
		clp = res->sr_session->clp;
437
		do_renew_lease(clp, timestamp);
438
		/* Check sequence flags */
439 440
		if (res->sr_status_flags != 0)
			nfs4_schedule_lease_recovery(clp);
441 442 443 444 445 446
		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.
		 */
447
		dprintk("%s: slot=%td seq=%d: Operation in progress\n",
448 449 450
			__func__,
			res->sr_slot - res->sr_session->fc_slot_table.slots,
			res->sr_slot->seq_nr);
451 452 453
		goto out_retry;
	default:
		/* Just update the slot sequence no. */
454
		++res->sr_slot->seq_nr;
A
Andy Adamson 已提交
455 456 457 458
	}
out:
	/* The session may be reset by one of the error handlers. */
	dprintk("%s: Error %d free the slot \n", __func__, res->sr_status);
459
	nfs41_sequence_free_slot(res);
460 461
	return 1;
out_retry:
462
	if (!rpc_restart_call(task))
463 464 465
		goto out;
	rpc_delay(task, NFS4_POLL_RETRY_MAX);
	return 0;
A
Andy Adamson 已提交
466 467
}

468 469
static int nfs4_sequence_done(struct rpc_task *task,
			       struct nfs4_sequence_res *res)
470
{
471 472 473
	if (res->sr_session == NULL)
		return 1;
	return nfs41_sequence_done(task, res);
474 475
}

B
Benny Halevy 已提交
476 477 478 479 480 481 482
/*
 * 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.
 * The slot number is returned if found, or NFS4_MAX_SLOT_TABLE otherwise.
A
Andy Adamson 已提交
483 484
 *
 * Note: must be called with under the slot_tbl_lock.
B
Benny Halevy 已提交
485 486
 */
static u8
487
nfs4_find_slot(struct nfs4_slot_table *tbl)
B
Benny Halevy 已提交
488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508
{
	int slotid;
	u8 ret_id = NFS4_MAX_SLOT_TABLE;
	BUILD_BUG_ON((u8)NFS4_MAX_SLOT_TABLE != (int)NFS4_MAX_SLOT_TABLE);

	dprintk("--> %s used_slots=%04lx highest_used=%d max_slots=%d\n",
		__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);
	if (slotid > tbl->highest_used_slotid)
		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;
}

A
Andy Adamson 已提交
509 510 511 512 513 514
static int nfs41_setup_sequence(struct nfs4_session *session,
				struct nfs4_sequence_args *args,
				struct nfs4_sequence_res *res,
				int cache_reply,
				struct rpc_task *task)
{
A
Andy Adamson 已提交
515 516 517 518 519
	struct nfs4_slot *slot;
	struct nfs4_slot_table *tbl;
	u8 slotid;

	dprintk("--> %s\n", __func__);
A
Andy Adamson 已提交
520
	/* slot already allocated? */
521
	if (res->sr_slot != NULL)
A
Andy Adamson 已提交
522 523
		return 0;

A
Andy Adamson 已提交
524 525 526
	tbl = &session->fc_slot_table;

	spin_lock(&tbl->slot_tbl_lock);
527
	if (test_bit(NFS4_SESSION_DRAINING, &session->session_state) &&
528
	    !rpc_task_has_priority(task, RPC_PRIORITY_PRIVILEGED)) {
529 530 531 532
		/*
		 * The state manager will wait until the slot table is empty.
		 * Schedule the reset thread
		 */
A
Andy Adamson 已提交
533
		rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
534
		spin_unlock(&tbl->slot_tbl_lock);
T
Trond Myklebust 已提交
535
		dprintk("%s Schedule Session Reset\n", __func__);
A
Andy Adamson 已提交
536
		return -EAGAIN;
537 538
	}

539 540 541 542 543 544 545 546
	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;
	}

547
	slotid = nfs4_find_slot(tbl);
A
Andy Adamson 已提交
548 549 550 551 552 553 554 555
	if (slotid == NFS4_MAX_SLOT_TABLE) {
		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);

556
	rpc_task_set_priority(task, RPC_PRIORITY_NORMAL);
A
Andy Adamson 已提交
557
	slot = tbl->slots + slotid;
B
Benny Halevy 已提交
558
	args->sa_session = session;
A
Andy Adamson 已提交
559 560 561 562 563
	args->sa_slotid = slotid;
	args->sa_cache_this = cache_reply;

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

B
Benny Halevy 已提交
564
	res->sr_session = session;
565
	res->sr_slot = slot;
A
Andy Adamson 已提交
566
	res->sr_renewal_time = jiffies;
567
	res->sr_status_flags = 0;
A
Andy Adamson 已提交
568 569 570 571 572
	/*
	 * 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
573 574 575
	return 0;
}

576
int nfs4_setup_sequence(const struct nfs_server *server,
A
Andy Adamson 已提交
577 578 579 580 581
			struct nfs4_sequence_args *args,
			struct nfs4_sequence_res *res,
			int cache_reply,
			struct rpc_task *task)
{
582
	struct nfs4_session *session = nfs4_get_session(server);
A
Andy Adamson 已提交
583 584
	int ret = 0;

585 586 587 588 589 590
	if (session == NULL) {
		args->sa_session = NULL;
		res->sr_session = NULL;
		goto out;
	}

591
	dprintk("--> %s clp %p session %p sr_slot %td\n",
592 593
		__func__, session->clp, session, res->sr_slot ?
			res->sr_slot - session->fc_slot_table.slots : -1);
A
Andy Adamson 已提交
594

595
	ret = nfs41_setup_sequence(session, args, res, cache_reply,
A
Andy Adamson 已提交
596 597 598 599 600 601 602
				   task);
out:
	dprintk("<-- %s status=%d\n", __func__, ret);
	return ret;
}

struct nfs41_call_sync_data {
603
	const struct nfs_server *seq_server;
A
Andy Adamson 已提交
604 605 606 607 608 609 610 611 612
	struct nfs4_sequence_args *seq_args;
	struct nfs4_sequence_res *seq_res;
	int cache_reply;
};

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

613 614 615
	dprintk("--> %s data->seq_server %p\n", __func__, data->seq_server);

	if (nfs4_setup_sequence(data->seq_server, data->seq_args,
A
Andy Adamson 已提交
616 617 618 619 620
				data->seq_res, data->cache_reply, task))
		return;
	rpc_call_start(task);
}

621 622 623 624 625 626
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
627 628 629 630
static void nfs41_call_sync_done(struct rpc_task *task, void *calldata)
{
	struct nfs41_call_sync_data *data = calldata;

631
	nfs41_sequence_done(task, data->seq_res);
A
Andy Adamson 已提交
632 633
}

A
Andy Adamson 已提交
634 635
struct rpc_call_ops nfs41_call_sync_ops = {
	.rpc_call_prepare = nfs41_call_sync_prepare,
A
Andy Adamson 已提交
636
	.rpc_call_done = nfs41_call_sync_done,
A
Andy Adamson 已提交
637 638
};

639 640 641 642 643
struct rpc_call_ops nfs41_call_priv_sync_ops = {
	.rpc_call_prepare = nfs41_call_priv_sync_prepare,
	.rpc_call_done = nfs41_call_sync_done,
};

644
static int nfs4_call_sync_sequence(struct nfs_server *server,
A
Andy Adamson 已提交
645 646 647
				   struct rpc_message *msg,
				   struct nfs4_sequence_args *args,
				   struct nfs4_sequence_res *res,
648 649
				   int cache_reply,
				   int privileged)
A
Andy Adamson 已提交
650 651 652 653
{
	int ret;
	struct rpc_task *task;
	struct nfs41_call_sync_data data = {
654
		.seq_server = server,
A
Andy Adamson 已提交
655 656 657 658 659
		.seq_args = args,
		.seq_res = res,
		.cache_reply = cache_reply,
	};
	struct rpc_task_setup task_setup = {
660
		.rpc_client = server->client,
A
Andy Adamson 已提交
661 662 663 664 665
		.rpc_message = msg,
		.callback_ops = &nfs41_call_sync_ops,
		.callback_data = &data
	};

666
	res->sr_slot = NULL;
667 668
	if (privileged)
		task_setup.callback_ops = &nfs41_call_priv_sync_ops;
A
Andy Adamson 已提交
669 670 671 672 673 674 675 676 677 678
	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;
}

A
Andy Adamson 已提交
679 680 681 682 683 684
int _nfs4_call_sync_session(struct nfs_server *server,
			    struct rpc_message *msg,
			    struct nfs4_sequence_args *args,
			    struct nfs4_sequence_res *res,
			    int cache_reply)
{
685
	return nfs4_call_sync_sequence(server, msg, args, res, cache_reply, 0);
A
Andy Adamson 已提交
686 687
}

688
#else
689 690
static int nfs4_sequence_done(struct rpc_task *task,
			       struct nfs4_sequence_res *res)
691
{
692
	return 1;
693
}
A
Andy Adamson 已提交
694 695 696 697 698 699 700 701
#endif /* CONFIG_NFS_V4_1 */

int _nfs4_call_sync(struct nfs_server *server,
		    struct rpc_message *msg,
		    struct nfs4_sequence_args *args,
		    struct nfs4_sequence_res *res,
		    int cache_reply)
{
702
	args->sa_session = res->sr_session = NULL;
A
Andy Adamson 已提交
703 704 705 706
	return rpc_call_sync(server->client, msg, 0);
}

#define nfs4_call_sync(server, msg, args, res, cache_reply) \
707
	(server)->nfs_client->cl_mvops->call_sync((server), (msg), &(args)->seq_args, \
A
Andy Adamson 已提交
708 709
			&(res)->seq_res, (cache_reply))

710
static void update_changeattr(struct inode *dir, struct nfs4_change_info *cinfo)
L
Linus Torvalds 已提交
711
{
712
	struct nfs_inode *nfsi = NFS_I(dir);
L
Linus Torvalds 已提交
713

714
	spin_lock(&dir->i_lock);
715 716
	nfsi->cache_validity |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE|NFS_INO_INVALID_DATA;
	if (!cinfo->atomic || cinfo->before != nfsi->change_attr)
717
		nfs_force_lookup_revalidate(dir);
718
	nfsi->change_attr = cinfo->after;
719
	spin_unlock(&dir->i_lock);
L
Linus Torvalds 已提交
720 721
}

722
struct nfs4_opendata {
723
	struct kref kref;
724 725
	struct nfs_openargs o_arg;
	struct nfs_openres o_res;
726 727
	struct nfs_open_confirmargs c_arg;
	struct nfs_open_confirmres c_res;
728 729
	struct nfs_fattr f_attr;
	struct nfs_fattr dir_attr;
730
	struct path path;
731 732
	struct dentry *dir;
	struct nfs4_state_owner *owner;
733
	struct nfs4_state *state;
734
	struct iattr attrs;
735
	unsigned long timestamp;
736
	unsigned int rpc_done : 1;
737 738
	int rpc_status;
	int cancelled;
739 740
};

741 742 743 744 745

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;
746 747
	p->o_res.seqid = p->o_arg.seqid;
	p->c_res.seqid = p->c_arg.seqid;
748 749 750 751 752
	p->o_res.server = p->o_arg.server;
	nfs_fattr_init(&p->f_attr);
	nfs_fattr_init(&p->dir_attr);
}

753
static struct nfs4_opendata *nfs4_opendata_alloc(struct path *path,
754
		struct nfs4_state_owner *sp, fmode_t fmode, int flags,
755 756
		const struct iattr *attrs,
		gfp_t gfp_mask)
757
{
758
	struct dentry *parent = dget_parent(path->dentry);
759 760 761 762
	struct inode *dir = parent->d_inode;
	struct nfs_server *server = NFS_SERVER(dir);
	struct nfs4_opendata *p;

763
	p = kzalloc(sizeof(*p), gfp_mask);
764 765
	if (p == NULL)
		goto err;
766
	p->o_arg.seqid = nfs_alloc_seqid(&sp->so_seqid, gfp_mask);
767 768
	if (p->o_arg.seqid == NULL)
		goto err_free;
769 770
	path_get(path);
	p->path = *path;
771 772 773 774
	p->dir = parent;
	p->owner = sp;
	atomic_inc(&sp->so_count);
	p->o_arg.fh = NFS_FH(dir);
775 776
	p->o_arg.open_flags = flags;
	p->o_arg.fmode = fmode & (FMODE_READ|FMODE_WRITE);
777
	p->o_arg.clientid = server->nfs_client->cl_clientid;
778
	p->o_arg.id = sp->so_owner_id.id;
779
	p->o_arg.name = &p->path.dentry->d_name;
780 781 782
	p->o_arg.server = server;
	p->o_arg.bitmask = server->attr_bitmask;
	p->o_arg.claim = NFS4_OPEN_CLAIM_NULL;
783 784 785
	if (flags & O_CREAT) {
		u32 *s;

786 787
		p->o_arg.u.attrs = &p->attrs;
		memcpy(&p->attrs, attrs, sizeof(p->attrs));
788 789 790
		s = (u32 *) p->o_arg.u.verifier.data;
		s[0] = jiffies;
		s[1] = current->pid;
791
	}
792 793 794
	p->c_arg.fh = &p->o_res.fh;
	p->c_arg.stateid = &p->o_res.stateid;
	p->c_arg.seqid = p->o_arg.seqid;
795
	nfs4_init_opendata_res(p);
796
	kref_init(&p->kref);
797 798 799 800 801 802 803 804
	return p;
err_free:
	kfree(p);
err:
	dput(parent);
	return NULL;
}

805
static void nfs4_opendata_free(struct kref *kref)
806
{
807 808 809 810
	struct nfs4_opendata *p = container_of(kref,
			struct nfs4_opendata, kref);

	nfs_free_seqid(p->o_arg.seqid);
811 812
	if (p->state != NULL)
		nfs4_put_open_state(p->state);
813 814
	nfs4_put_state_owner(p->owner);
	dput(p->dir);
J
Jan Blunck 已提交
815
	path_put(&p->path);
816 817 818 819 820 821 822
	kfree(p);
}

static void nfs4_opendata_put(struct nfs4_opendata *p)
{
	if (p != NULL)
		kref_put(&p->kref, nfs4_opendata_free);
823 824
}

825 826 827 828 829 830 831 832
static int nfs4_wait_for_completion_rpc_task(struct rpc_task *task)
{
	int ret;

	ret = rpc_wait_for_completion_task(task);
	return ret;
}

833
static int can_open_cached(struct nfs4_state *state, fmode_t mode, int open_mode)
834 835
{
	int ret = 0;
836 837 838 839

	if (open_mode & O_EXCL)
		goto out;
	switch (mode & (FMODE_READ|FMODE_WRITE)) {
840
		case FMODE_READ:
841 842
			ret |= test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0
				&& state->n_rdonly != 0;
843 844
			break;
		case FMODE_WRITE:
845 846
			ret |= test_bit(NFS_O_WRONLY_STATE, &state->flags) != 0
				&& state->n_wronly != 0;
847 848
			break;
		case FMODE_READ|FMODE_WRITE:
849 850
			ret |= test_bit(NFS_O_RDWR_STATE, &state->flags) != 0
				&& state->n_rdwr != 0;
851
	}
852
out:
853 854 855
	return ret;
}

856
static int can_open_delegated(struct nfs_delegation *delegation, fmode_t fmode)
857
{
858
	if ((delegation->type & fmode) != fmode)
859
		return 0;
860
	if (test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags))
861
		return 0;
862
	nfs_mark_delegation_referenced(delegation);
863 864 865
	return 1;
}

866
static void update_open_stateflags(struct nfs4_state *state, fmode_t fmode)
867
{
868
	switch (fmode) {
869 870 871 872 873 874 875 876 877
		case FMODE_WRITE:
			state->n_wronly++;
			break;
		case FMODE_READ:
			state->n_rdonly++;
			break;
		case FMODE_READ|FMODE_WRITE:
			state->n_rdwr++;
	}
878
	nfs4_state_set_mode_locked(state, state->state | fmode);
879 880
}

881
static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
882 883 884 885
{
	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
		memcpy(state->stateid.data, stateid->data, sizeof(state->stateid.data));
	memcpy(state->open_stateid.data, stateid->data, sizeof(state->open_stateid.data));
886
	switch (fmode) {
887 888 889 890 891 892 893 894 895
		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);
	}
896 897
}

898
static void nfs_set_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
899
{
900
	write_seqlock(&state->seqlock);
901
	nfs_set_open_stateid_locked(state, stateid, fmode);
902
	write_sequnlock(&state->seqlock);
903 904
}

905
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
906
{
907 908 909 910 911
	/*
	 * Protect the call to nfs4_state_set_mode_locked and
	 * serialise the stateid update
	 */
	write_seqlock(&state->seqlock);
912 913 914 915 916
	if (deleg_stateid != NULL) {
		memcpy(state->stateid.data, deleg_stateid->data, sizeof(state->stateid.data));
		set_bit(NFS_DELEGATED_STATE, &state->flags);
	}
	if (open_stateid != NULL)
917
		nfs_set_open_stateid_locked(state, open_stateid, fmode);
918 919
	write_sequnlock(&state->seqlock);
	spin_lock(&state->owner->so_lock);
920
	update_open_stateflags(state, fmode);
921
	spin_unlock(&state->owner->so_lock);
L
Linus Torvalds 已提交
922 923
}

924
static int update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *delegation, fmode_t fmode)
925 926 927 928 929
{
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *deleg_cur;
	int ret = 0;

930
	fmode &= (FMODE_READ|FMODE_WRITE);
931 932 933 934 935 936 937 938

	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 ||
939
	    (deleg_cur->type & fmode) != fmode)
940 941 942 943 944 945 946
		goto no_delegation_unlock;

	if (delegation == NULL)
		delegation = &deleg_cur->stateid;
	else if (memcmp(deleg_cur->stateid.data, delegation->data, NFS4_STATEID_SIZE) != 0)
		goto no_delegation_unlock;

947
	nfs_mark_delegation_referenced(deleg_cur);
948
	__update_open_stateid(state, open_stateid, &deleg_cur->stateid, fmode);
949 950 951 952 953 954 955
	ret = 1;
no_delegation_unlock:
	spin_unlock(&deleg_cur->lock);
no_delegation:
	rcu_read_unlock();

	if (!ret && open_stateid != NULL) {
956
		__update_open_stateid(state, open_stateid, NULL, fmode);
957 958 959 960 961 962 963
		ret = 1;
	}

	return ret;
}


964
static void nfs4_return_incompatible_delegation(struct inode *inode, fmode_t fmode)
965 966 967 968 969
{
	struct nfs_delegation *delegation;

	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(inode)->delegation);
970
	if (delegation == NULL || (delegation->type & fmode) == fmode) {
971 972 973 974 975 976 977
		rcu_read_unlock();
		return;
	}
	rcu_read_unlock();
	nfs_inode_return_delegation(inode);
}

978
static struct nfs4_state *nfs4_try_open_cached(struct nfs4_opendata *opendata)
979 980 981 982
{
	struct nfs4_state *state = opendata->state;
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *delegation;
983 984
	int open_mode = opendata->o_arg.open_flags & O_EXCL;
	fmode_t fmode = opendata->o_arg.fmode;
985 986 987 988
	nfs4_stateid stateid;
	int ret = -EAGAIN;

	for (;;) {
989
		if (can_open_cached(state, fmode, open_mode)) {
990
			spin_lock(&state->owner->so_lock);
991 992
			if (can_open_cached(state, fmode, open_mode)) {
				update_open_stateflags(state, fmode);
993 994 995 996 997
				spin_unlock(&state->owner->so_lock);
				goto out_return_state;
			}
			spin_unlock(&state->owner->so_lock);
		}
998 999 1000
		rcu_read_lock();
		delegation = rcu_dereference(nfsi->delegation);
		if (delegation == NULL ||
1001
		    !can_open_delegated(delegation, fmode)) {
1002
			rcu_read_unlock();
1003
			break;
1004
		}
1005 1006 1007
		/* Save the delegation */
		memcpy(stateid.data, delegation->stateid.data, sizeof(stateid.data));
		rcu_read_unlock();
1008
		ret = nfs_may_open(state->inode, state->owner->so_cred, open_mode);
1009 1010 1011
		if (ret != 0)
			goto out;
		ret = -EAGAIN;
1012 1013

		/* Try to update the stateid using the delegation */
1014
		if (update_open_stateid(state, NULL, &stateid, fmode))
1015
			goto out_return_state;
1016 1017 1018 1019 1020 1021 1022 1023
	}
out:
	return ERR_PTR(ret);
out_return_state:
	atomic_inc(&state->count);
	return state;
}

1024 1025 1026 1027
static struct nfs4_state *nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data)
{
	struct inode *inode;
	struct nfs4_state *state = NULL;
1028
	struct nfs_delegation *delegation;
1029
	int ret;
1030

1031
	if (!data->rpc_done) {
1032
		state = nfs4_try_open_cached(data);
1033 1034 1035
		goto out;
	}

1036
	ret = -EAGAIN;
1037
	if (!(data->f_attr.valid & NFS_ATTR_FATTR))
1038
		goto err;
1039
	inode = nfs_fhget(data->dir->d_sb, &data->o_res.fh, &data->f_attr);
1040
	ret = PTR_ERR(inode);
1041
	if (IS_ERR(inode))
1042 1043
		goto err;
	ret = -ENOMEM;
1044 1045
	state = nfs4_get_open_state(inode, data->owner);
	if (state == NULL)
1046
		goto err_put_inode;
1047 1048 1049
	if (data->o_res.delegation_type != 0) {
		int delegation_flags = 0;

1050 1051 1052 1053 1054
		rcu_read_lock();
		delegation = rcu_dereference(NFS_I(inode)->delegation);
		if (delegation)
			delegation_flags = delegation->flags;
		rcu_read_unlock();
1055
		if ((delegation_flags & 1UL<<NFS_DELEGATION_NEED_RECLAIM) == 0)
1056 1057 1058 1059 1060 1061 1062 1063
			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);
	}
1064 1065

	update_open_stateid(state, &data->o_res.stateid, NULL,
1066
			data->o_arg.fmode);
1067
	iput(inode);
1068
out:
1069
	return state;
1070 1071 1072 1073
err_put_inode:
	iput(inode);
err:
	return ERR_PTR(ret);
1074 1075
}

1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092
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
1093 1094 1095 1096
static struct nfs4_opendata *nfs4_open_recoverdata_alloc(struct nfs_open_context *ctx, struct nfs4_state *state)
{
	struct nfs4_opendata *opendata;

1097
	opendata = nfs4_opendata_alloc(&ctx->path, state->owner, 0, 0, NULL, GFP_NOFS);
T
Trond Myklebust 已提交
1098 1099 1100 1101 1102 1103 1104
	if (opendata == NULL)
		return ERR_PTR(-ENOMEM);
	opendata->state = state;
	atomic_inc(&state->count);
	return opendata;
}

1105
static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, fmode_t fmode, struct nfs4_state **res)
1106
{
1107
	struct nfs4_state *newstate;
1108 1109
	int ret;

1110 1111
	opendata->o_arg.open_flags = 0;
	opendata->o_arg.fmode = fmode;
1112 1113 1114
	memset(&opendata->o_res, 0, sizeof(opendata->o_res));
	memset(&opendata->c_res, 0, sizeof(opendata->c_res));
	nfs4_init_opendata_res(opendata);
1115
	ret = _nfs4_recover_proc_open(opendata);
1116 1117
	if (ret != 0)
		return ret; 
1118
	newstate = nfs4_opendata_to_nfs4_state(opendata);
1119 1120
	if (IS_ERR(newstate))
		return PTR_ERR(newstate);
1121
	nfs4_close_state(&opendata->path, newstate, fmode);
1122
	*res = newstate;
1123 1124 1125 1126 1127 1128 1129 1130 1131
	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_* */
1132
	clear_bit(NFS_DELEGATED_STATE, &state->flags);
1133 1134
	smp_rmb();
	if (state->n_rdwr != 0) {
T
Trond Myklebust 已提交
1135
		clear_bit(NFS_O_RDWR_STATE, &state->flags);
1136
		ret = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE, &newstate);
1137 1138
		if (ret != 0)
			return ret;
1139 1140
		if (newstate != state)
			return -ESTALE;
1141 1142
	}
	if (state->n_wronly != 0) {
T
Trond Myklebust 已提交
1143
		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
1144
		ret = nfs4_open_recover_helper(opendata, FMODE_WRITE, &newstate);
1145 1146
		if (ret != 0)
			return ret;
1147 1148
		if (newstate != state)
			return -ESTALE;
1149 1150
	}
	if (state->n_rdonly != 0) {
T
Trond Myklebust 已提交
1151
		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
1152
		ret = nfs4_open_recover_helper(opendata, FMODE_READ, &newstate);
1153 1154
		if (ret != 0)
			return ret;
1155 1156
		if (newstate != state)
			return -ESTALE;
1157
	}
1158 1159 1160 1161 1162 1163
	/*
	 * 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 &&
	    memcmp(state->stateid.data, state->open_stateid.data, sizeof(state->stateid.data)) != 0) {
1164
		write_seqlock(&state->seqlock);
1165 1166
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
			memcpy(state->stateid.data, state->open_stateid.data, sizeof(state->stateid.data));
1167
		write_sequnlock(&state->seqlock);
1168
	}
1169 1170 1171
	return 0;
}

L
Linus Torvalds 已提交
1172 1173 1174 1175
/*
 * OPEN_RECLAIM:
 * 	reclaim state on the server after a reboot.
 */
1176
static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1177
{
1178
	struct nfs_delegation *delegation;
1179
	struct nfs4_opendata *opendata;
1180
	fmode_t delegation_type = 0;
L
Linus Torvalds 已提交
1181 1182
	int status;

T
Trond Myklebust 已提交
1183 1184 1185
	opendata = nfs4_open_recoverdata_alloc(ctx, state);
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1186 1187
	opendata->o_arg.claim = NFS4_OPEN_CLAIM_PREVIOUS;
	opendata->o_arg.fh = NFS_FH(state->inode);
1188 1189
	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
1190
	if (delegation != NULL && test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) != 0)
1191
		delegation_type = delegation->type;
1192
	rcu_read_unlock();
1193 1194
	opendata->o_arg.u.delegation_type = delegation_type;
	status = nfs4_open_recover(opendata, state);
1195
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
1196 1197 1198
	return status;
}

1199
static int nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1200 1201 1202 1203 1204
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;
	do {
1205
		err = _nfs4_do_open_reclaim(ctx, state);
1206
		if (err != -NFS4ERR_DELAY)
1207 1208
			break;
		nfs4_handle_exception(server, err, &exception);
L
Linus Torvalds 已提交
1209 1210 1211 1212
	} while (exception.retry);
	return err;
}

1213 1214 1215 1216 1217 1218 1219 1220
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);
1221
	ret = nfs4_do_open_reclaim(ctx, state);
1222 1223 1224 1225
	put_nfs_open_context(ctx);
	return ret;
}

1226
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
1227
{
1228
	struct nfs4_opendata *opendata;
1229
	int ret;
L
Linus Torvalds 已提交
1230

T
Trond Myklebust 已提交
1231 1232 1233
	opendata = nfs4_open_recoverdata_alloc(ctx, state);
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1234
	opendata->o_arg.claim = NFS4_OPEN_CLAIM_DELEGATE_CUR;
1235
	memcpy(opendata->o_arg.u.delegation.data, stateid->data,
1236
			sizeof(opendata->o_arg.u.delegation.data));
1237
	ret = nfs4_open_recover(opendata, state);
1238
	nfs4_opendata_put(opendata);
1239
	return ret;
L
Linus Torvalds 已提交
1240 1241
}

1242
int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid)
L
Linus Torvalds 已提交
1243 1244
{
	struct nfs4_exception exception = { };
1245
	struct nfs_server *server = NFS_SERVER(state->inode);
L
Linus Torvalds 已提交
1246 1247
	int err;
	do {
1248
		err = _nfs4_open_delegation_recall(ctx, state, stateid);
L
Linus Torvalds 已提交
1249 1250
		switch (err) {
			case 0:
1251 1252 1253
			case -ENOENT:
			case -ESTALE:
				goto out;
1254 1255 1256 1257 1258
			case -NFS4ERR_BADSESSION:
			case -NFS4ERR_BADSLOT:
			case -NFS4ERR_BAD_HIGH_SLOT:
			case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
			case -NFS4ERR_DEADSESSION:
1259
				nfs4_schedule_session_recovery(server->nfs_client->cl_session);
1260
				goto out;
L
Linus Torvalds 已提交
1261 1262 1263 1264
			case -NFS4ERR_STALE_CLIENTID:
			case -NFS4ERR_STALE_STATEID:
			case -NFS4ERR_EXPIRED:
				/* Don't recall a delegation if it was lost */
1265
				nfs4_schedule_lease_recovery(server->nfs_client);
1266 1267 1268 1269 1270 1271 1272 1273
				goto out;
			case -ERESTARTSYS:
				/*
				 * The show must go on: exit, but mark the
				 * stateid as needing recovery.
				 */
			case -NFS4ERR_ADMIN_REVOKED:
			case -NFS4ERR_BAD_STATEID:
1274
				nfs4_schedule_stateid_recovery(server, state);
1275 1276 1277 1278 1279 1280 1281
			case -EKEYEXPIRED:
				/*
				 * User RPCSEC_GSS context has expired.
				 * We cannot recover this stateid now, so
				 * skip it and allow recovery thread to
				 * proceed.
				 */
1282 1283 1284
			case -ENOMEM:
				err = 0;
				goto out;
L
Linus Torvalds 已提交
1285 1286 1287
		}
		err = nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
1288
out:
L
Linus Torvalds 已提交
1289 1290 1291
	return err;
}

1292 1293 1294 1295 1296
static void nfs4_open_confirm_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;

	data->rpc_status = task->tk_status;
1297
	if (data->rpc_status == 0) {
1298 1299
		memcpy(data->o_res.stateid.data, data->c_res.stateid.data,
				sizeof(data->o_res.stateid.data));
1300
		nfs_confirm_seqid(&data->owner->so_seqid, 0);
1301
		renew_lease(data->o_res.server, data->timestamp);
1302
		data->rpc_done = 1;
1303
	}
1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314
}

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! */
1315
	if (!data->rpc_done)
1316 1317
		goto out_free;
	state = nfs4_opendata_to_nfs4_state(data);
1318
	if (!IS_ERR(state))
1319
		nfs4_close_state(&data->path, state, data->o_arg.fmode);
1320
out_free:
1321
	nfs4_opendata_put(data);
1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335
}

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;
1336 1337 1338 1339 1340 1341
	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
1342 1343
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1344
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1345 1346
		.callback_ops = &nfs4_open_confirm_ops,
		.callback_data = data,
1347
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1348 1349
		.flags = RPC_TASK_ASYNC,
	};
L
Linus Torvalds 已提交
1350 1351
	int status;

1352
	kref_get(&data->kref);
1353 1354
	data->rpc_done = 0;
	data->rpc_status = 0;
1355
	data->timestamp = jiffies;
T
Trond Myklebust 已提交
1356
	task = rpc_run_task(&task_setup_data);
1357
	if (IS_ERR(task))
1358 1359 1360 1361 1362 1363 1364
		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;
1365
	rpc_put_task(task);
L
Linus Torvalds 已提交
1366 1367 1368
	return status;
}

1369
static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
L
Linus Torvalds 已提交
1370
{
1371 1372
	struct nfs4_opendata *data = calldata;
	struct nfs4_state_owner *sp = data->owner;
1373

1374 1375
	if (nfs_wait_on_sequence(data->o_arg.seqid, task) != 0)
		return;
1376 1377 1378 1379 1380 1381 1382
	/*
	 * 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;

1383
		if (can_open_cached(data->state, data->o_arg.fmode, data->o_arg.open_flags))
1384
			goto out_no_action;
1385 1386 1387
		rcu_read_lock();
		delegation = rcu_dereference(NFS_I(data->state->inode)->delegation);
		if (delegation != NULL &&
1388
		    test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) == 0) {
1389
			rcu_read_unlock();
1390
			goto out_no_action;
1391 1392 1393
		}
		rcu_read_unlock();
	}
1394
	/* Update sequence id. */
1395
	data->o_arg.id = sp->so_owner_id.id;
1396
	data->o_arg.clientid = sp->so_server->nfs_client->cl_clientid;
T
Trond Myklebust 已提交
1397
	if (data->o_arg.claim == NFS4_OPEN_CLAIM_PREVIOUS) {
1398
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR];
T
Trond Myklebust 已提交
1399 1400
		nfs_copy_fh(&data->o_res.fh, data->o_arg.fh);
	}
1401
	data->timestamp = jiffies;
1402
	if (nfs4_setup_sequence(data->o_arg.server,
1403 1404 1405
				&data->o_arg.seq_args,
				&data->o_res.seq_res, 1, task))
		return;
1406
	rpc_call_start(task);
1407 1408 1409 1410
	return;
out_no_action:
	task->tk_action = NULL;

1411
}
L
Linus Torvalds 已提交
1412

1413 1414 1415 1416 1417 1418
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);
}

1419 1420 1421
static void nfs4_open_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;
L
Linus Torvalds 已提交
1422

1423
	data->rpc_status = task->tk_status;
1424

1425 1426
	if (!nfs4_sequence_done(task, &data->o_res.seq_res))
		return;
1427

1428 1429
	if (task->tk_status == 0) {
		switch (data->o_res.f_attr->mode & S_IFMT) {
1430 1431 1432
			case S_IFREG:
				break;
			case S_IFLNK:
1433
				data->rpc_status = -ELOOP;
1434 1435
				break;
			case S_IFDIR:
1436
				data->rpc_status = -EISDIR;
1437 1438
				break;
			default:
1439
				data->rpc_status = -ENOTDIR;
1440
		}
1441
		renew_lease(data->o_res.server, data->timestamp);
1442 1443
		if (!(data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM))
			nfs_confirm_seqid(&data->owner->so_seqid, 0);
1444
	}
1445
	data->rpc_done = 1;
1446
}
1447

1448 1449 1450 1451 1452 1453 1454 1455 1456
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! */
1457
	if (data->rpc_status != 0 || !data->rpc_done)
1458 1459 1460 1461 1462
		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);
1463
	if (!IS_ERR(state))
1464
		nfs4_close_state(&data->path, state, data->o_arg.fmode);
1465
out_free:
1466
	nfs4_opendata_put(data);
1467 1468 1469 1470 1471 1472 1473 1474
}

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,
};

1475 1476 1477 1478 1479 1480 1481
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)
1482 1483 1484 1485 1486 1487
{
	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;
1488 1489 1490 1491 1492 1493
	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
1494 1495
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1496
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1497 1498
		.callback_ops = &nfs4_open_ops,
		.callback_data = data,
1499
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1500 1501
		.flags = RPC_TASK_ASYNC,
	};
1502 1503
	int status;

1504
	kref_get(&data->kref);
1505 1506
	data->rpc_done = 0;
	data->rpc_status = 0;
1507
	data->cancelled = 0;
1508 1509
	if (isrecover)
		task_setup_data.callback_ops = &nfs4_recover_open_ops;
T
Trond Myklebust 已提交
1510
	task = rpc_run_task(&task_setup_data);
1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556
        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;

	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);
1557
	if (status != 0 || !data->rpc_done)
1558 1559
		return status;

1560 1561 1562 1563 1564
	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
1565 1566
	if ((o_res->rflags & NFS4_OPEN_RESULT_LOCKTYPE_POSIX) == 0)
		server->caps &= ~NFS_CAP_POSIX_LOCK;
L
Linus Torvalds 已提交
1567
	if(o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
1568
		status = _nfs4_proc_open_confirm(data);
L
Linus Torvalds 已提交
1569
		if (status != 0)
1570
			return status;
L
Linus Torvalds 已提交
1571 1572
	}
	if (!(o_res->f_attr->valid & NFS_ATTR_FATTR))
1573
		_nfs4_proc_getattr(server, &o_res->fh, o_res->f_attr);
1574
	return 0;
L
Linus Torvalds 已提交
1575 1576
}

1577
static int nfs4_recover_expired_lease(struct nfs_server *server)
1578
{
1579
	struct nfs_client *clp = server->nfs_client;
1580
	unsigned int loop;
1581
	int ret;
1582

1583
	for (loop = NFS4_MAX_LOOP_ON_RECOVER; loop != 0; loop--) {
1584
		ret = nfs4_wait_clnt_recover(clp);
1585
		if (ret != 0)
1586
			break;
1587 1588
		if (!test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state) &&
		    !test_bit(NFS4CLNT_CHECK_LEASE,&clp->cl_state))
1589
			break;
1590
		nfs4_schedule_state_manager(clp);
1591
		ret = -EIO;
1592
	}
1593
	return ret;
1594 1595
}

L
Linus Torvalds 已提交
1596 1597 1598 1599 1600
/*
 * OPEN_EXPIRED:
 * 	reclaim state on the server after a network partition.
 * 	Assumes caller holds the appropriate lock
 */
1601
static int _nfs4_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1602
{
1603
	struct nfs4_opendata *opendata;
1604
	int ret;
L
Linus Torvalds 已提交
1605

T
Trond Myklebust 已提交
1606 1607 1608
	opendata = nfs4_open_recoverdata_alloc(ctx, state);
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1609
	ret = nfs4_open_recover(opendata, state);
1610
	if (ret == -ESTALE)
1611
		d_drop(ctx->path.dentry);
1612
	nfs4_opendata_put(opendata);
1613
	return ret;
L
Linus Torvalds 已提交
1614 1615
}

1616
static int nfs4_do_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
1617
{
1618
	struct nfs_server *server = NFS_SERVER(state->inode);
1619 1620 1621 1622
	struct nfs4_exception exception = { };
	int err;

	do {
1623
		err = _nfs4_open_expired(ctx, state);
1624 1625 1626 1627 1628 1629 1630 1631
		switch (err) {
		default:
			goto out;
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			nfs4_handle_exception(server, err, &exception);
			err = 0;
		}
1632
	} while (exception.retry);
1633
out:
1634 1635 1636
	return err;
}

L
Linus Torvalds 已提交
1637 1638 1639
static int nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
	struct nfs_open_context *ctx;
1640
	int ret;
L
Linus Torvalds 已提交
1641

1642 1643 1644
	ctx = nfs4_state_find_open_context(state);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);
1645
	ret = nfs4_do_open_expired(ctx, state);
1646 1647
	put_nfs_open_context(ctx);
	return ret;
L
Linus Torvalds 已提交
1648 1649
}

1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665
/*
 * 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
1666
/*
1667
 * Returns a referenced nfs4_state
L
Linus Torvalds 已提交
1668
 */
1669
static int _nfs4_do_open(struct inode *dir, struct path *path, fmode_t fmode, int flags, struct iattr *sattr, struct rpc_cred *cred, struct nfs4_state **res)
L
Linus Torvalds 已提交
1670 1671 1672 1673
{
	struct nfs4_state_owner  *sp;
	struct nfs4_state     *state = NULL;
	struct nfs_server       *server = NFS_SERVER(dir);
1674
	struct nfs4_opendata *opendata;
1675
	int status;
L
Linus Torvalds 已提交
1676 1677 1678 1679 1680 1681 1682

	/* Protect against reboot recovery conflicts */
	status = -ENOMEM;
	if (!(sp = nfs4_get_state_owner(server, cred))) {
		dprintk("nfs4_do_open: nfs4_get_state_owner failed!\n");
		goto out_err;
	}
1683 1684
	status = nfs4_recover_expired_lease(server);
	if (status != 0)
1685
		goto err_put_state_owner;
1686
	if (path->dentry->d_inode != NULL)
1687
		nfs4_return_incompatible_delegation(path->dentry->d_inode, fmode);
1688
	status = -ENOMEM;
1689
	opendata = nfs4_opendata_alloc(path, sp, fmode, flags, sattr, GFP_KERNEL);
1690
	if (opendata == NULL)
T
Trond Myklebust 已提交
1691
		goto err_put_state_owner;
L
Linus Torvalds 已提交
1692

1693 1694 1695
	if (path->dentry->d_inode != NULL)
		opendata->state = nfs4_get_open_state(path->dentry->d_inode, sp);

1696
	status = _nfs4_proc_open(opendata);
L
Linus Torvalds 已提交
1697
	if (status != 0)
1698
		goto err_opendata_put;
L
Linus Torvalds 已提交
1699

1700
	state = nfs4_opendata_to_nfs4_state(opendata);
1701 1702
	status = PTR_ERR(state);
	if (IS_ERR(state))
1703
		goto err_opendata_put;
T
Trond Myklebust 已提交
1704
	if (server->caps & NFS_CAP_POSIX_LOCK)
1705
		set_bit(NFS_STATE_POSIX_LOCKS, &state->flags);
1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717

	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);
	}
1718
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
1719 1720 1721
	nfs4_put_state_owner(sp);
	*res = state;
	return 0;
1722 1723
err_opendata_put:
	nfs4_opendata_put(opendata);
1724 1725
err_put_state_owner:
	nfs4_put_state_owner(sp);
L
Linus Torvalds 已提交
1726 1727 1728 1729 1730 1731
out_err:
	*res = NULL;
	return status;
}


1732
static struct nfs4_state *nfs4_do_open(struct inode *dir, struct path *path, fmode_t fmode, int flags, struct iattr *sattr, struct rpc_cred *cred)
L
Linus Torvalds 已提交
1733 1734 1735 1736 1737 1738
{
	struct nfs4_exception exception = { };
	struct nfs4_state *res;
	int status;

	do {
1739
		status = _nfs4_do_open(dir, path, fmode, flags, sattr, cred, &res);
L
Linus Torvalds 已提交
1740 1741 1742 1743 1744 1745 1746 1747
		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
1748
		 * doing an OPEN, we assume that nfs_increment_open_seqid() will
L
Linus Torvalds 已提交
1749 1750 1751 1752 1753
		 * 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) {
1754 1755 1756
			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
1757 1758 1759
			exception.retry = 1;
			continue;
		}
1760 1761 1762 1763 1764 1765 1766 1767 1768 1769
		/*
		 * 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;
		}
1770 1771 1772 1773 1774
		if (status == -EAGAIN) {
			/* We must have found a delegation */
			exception.retry = 1;
			continue;
		}
L
Linus Torvalds 已提交
1775 1776 1777 1778 1779 1780
		res = ERR_PTR(nfs4_handle_exception(NFS_SERVER(dir),
					status, &exception));
	} while (exception.retry);
	return res;
}

1781 1782 1783
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
1784
{
1785
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
1786
        struct nfs_setattrargs  arg = {
1787
                .fh             = NFS_FH(inode),
L
Linus Torvalds 已提交
1788 1789 1790 1791 1792 1793 1794 1795 1796
                .iap            = sattr,
		.server		= server,
		.bitmask = server->attr_bitmask,
        };
        struct nfs_setattrres  res = {
		.fattr		= fattr,
		.server		= server,
        };
        struct rpc_message msg = {
1797 1798 1799 1800
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETATTR],
		.rpc_argp	= &arg,
		.rpc_resp	= &res,
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
1801
        };
1802
	unsigned long timestamp = jiffies;
1803
	int status;
L
Linus Torvalds 已提交
1804

1805
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
1806

1807 1808 1809
	if (nfs4_copy_delegation_stateid(&arg.stateid, inode)) {
		/* Use that stateid */
	} else if (state != NULL) {
1810
		nfs4_copy_stateid(&arg.stateid, state, current->files, current->tgid);
1811
	} else
L
Linus Torvalds 已提交
1812 1813
		memcpy(&arg.stateid, &zero_stateid, sizeof(arg.stateid));

A
Andy Adamson 已提交
1814
	status = nfs4_call_sync(server, &msg, &arg, &res, 1);
1815 1816
	if (status == 0 && state != NULL)
		renew_lease(server, timestamp);
1817
	return status;
L
Linus Torvalds 已提交
1818 1819
}

1820 1821 1822
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
1823
{
1824
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
1825 1826 1827 1828
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(server,
1829
				_nfs4_do_setattr(inode, cred, fattr, sattr, state),
L
Linus Torvalds 已提交
1830 1831 1832 1833 1834 1835
				&exception);
	} while (exception.retry);
	return err;
}

struct nfs4_closedata {
1836
	struct path path;
L
Linus Torvalds 已提交
1837 1838 1839 1840
	struct inode *inode;
	struct nfs4_state *state;
	struct nfs_closeargs arg;
	struct nfs_closeres res;
1841
	struct nfs_fattr fattr;
1842
	unsigned long timestamp;
F
Fred Isaman 已提交
1843 1844
	bool roc;
	u32 roc_barrier;
L
Linus Torvalds 已提交
1845 1846
};

1847
static void nfs4_free_closedata(void *data)
1848
{
1849 1850
	struct nfs4_closedata *calldata = data;
	struct nfs4_state_owner *sp = calldata->state->owner;
1851

F
Fred Isaman 已提交
1852 1853
	if (calldata->roc)
		pnfs_roc_release(calldata->state->inode);
1854 1855 1856
	nfs4_put_open_state(calldata->state);
	nfs_free_seqid(calldata->arg.seqid);
	nfs4_put_state_owner(sp);
J
Jan Blunck 已提交
1857
	path_put(&calldata->path);
1858 1859 1860
	kfree(calldata);
}

1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872
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);
}

1873
static void nfs4_close_done(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
1874
{
1875
	struct nfs4_closedata *calldata = data;
L
Linus Torvalds 已提交
1876 1877 1878
	struct nfs4_state *state = calldata->state;
	struct nfs_server *server = NFS_SERVER(calldata->inode);

1879 1880
	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
		return;
L
Linus Torvalds 已提交
1881 1882 1883 1884 1885
        /* 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
1886 1887 1888
			if (calldata->roc)
				pnfs_roc_set_barrier(state->inode,
						     calldata->roc_barrier);
1889
			nfs_set_open_stateid(state, &calldata->res.stateid, 0);
1890
			renew_lease(server, calldata->timestamp);
1891 1892
			nfs4_close_clear_stateid_flags(state,
					calldata->arg.fmode);
L
Linus Torvalds 已提交
1893 1894
			break;
		case -NFS4ERR_STALE_STATEID:
1895 1896
		case -NFS4ERR_OLD_STATEID:
		case -NFS4ERR_BAD_STATEID:
L
Linus Torvalds 已提交
1897
		case -NFS4ERR_EXPIRED:
1898
			if (calldata->arg.fmode == 0)
1899
				break;
L
Linus Torvalds 已提交
1900
		default:
1901 1902
			if (nfs4_async_handle_error(task, server, state) == -EAGAIN)
				rpc_restart_call_prepare(task);
L
Linus Torvalds 已提交
1903
	}
1904
	nfs_release_seqid(calldata->arg.seqid);
1905
	nfs_refresh_inode(calldata->inode, calldata->res.fattr);
L
Linus Torvalds 已提交
1906 1907
}

T
Trond Myklebust 已提交
1908
static void nfs4_close_prepare(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
1909
{
T
Trond Myklebust 已提交
1910
	struct nfs4_closedata *calldata = data;
1911
	struct nfs4_state *state = calldata->state;
1912
	int call_close = 0;
1913

1914
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
1915
		return;
1916

1917 1918
	task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_DOWNGRADE];
	calldata->arg.fmode = FMODE_READ|FMODE_WRITE;
1919
	spin_lock(&state->owner->so_lock);
1920
	/* Calculate the change in open mode */
1921
	if (state->n_rdwr == 0) {
1922
		if (state->n_rdonly == 0) {
1923 1924 1925
			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;
1926 1927
		}
		if (state->n_wronly == 0) {
1928 1929 1930
			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;
1931
		}
1932
	}
1933
	spin_unlock(&state->owner->so_lock);
1934 1935

	if (!call_close) {
1936 1937
		/* Note: exit _without_ calling nfs4_close_done */
		task->tk_action = NULL;
1938 1939
		return;
	}
1940

F
Fred Isaman 已提交
1941
	if (calldata->arg.fmode == 0) {
1942
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE];
F
Fred Isaman 已提交
1943 1944 1945 1946 1947 1948 1949
		if (calldata->roc &&
		    pnfs_roc_drain(calldata->inode, &calldata->roc_barrier)) {
			rpc_sleep_on(&NFS_SERVER(calldata->inode)->roc_rpcwaitq,
				     task, NULL);
			return;
		}
	}
1950

1951
	nfs_fattr_init(calldata->res.fattr);
1952
	calldata->timestamp = jiffies;
1953
	if (nfs4_setup_sequence(NFS_SERVER(calldata->inode),
1954 1955 1956
				&calldata->arg.seq_args, &calldata->res.seq_res,
				1, task))
		return;
1957
	rpc_call_start(task);
L
Linus Torvalds 已提交
1958 1959
}

1960
static const struct rpc_call_ops nfs4_close_ops = {
T
Trond Myklebust 已提交
1961
	.rpc_call_prepare = nfs4_close_prepare,
1962 1963 1964 1965
	.rpc_call_done = nfs4_close_done,
	.rpc_release = nfs4_free_closedata,
};

L
Linus Torvalds 已提交
1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976
/* 
 * 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!
 */
F
Fred Isaman 已提交
1977
int nfs4_do_close(struct path *path, struct nfs4_state *state, gfp_t gfp_mask, int wait, bool roc)
L
Linus Torvalds 已提交
1978
{
1979
	struct nfs_server *server = NFS_SERVER(state->inode);
L
Linus Torvalds 已提交
1980
	struct nfs4_closedata *calldata;
1981 1982
	struct nfs4_state_owner *sp = state->owner;
	struct rpc_task *task;
1983 1984 1985 1986
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
1987 1988
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1989
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1990
		.callback_ops = &nfs4_close_ops,
1991
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1992 1993
		.flags = RPC_TASK_ASYNC,
	};
1994
	int status = -ENOMEM;
L
Linus Torvalds 已提交
1995

1996
	calldata = kzalloc(sizeof(*calldata), gfp_mask);
L
Linus Torvalds 已提交
1997
	if (calldata == NULL)
1998
		goto out;
1999
	calldata->inode = state->inode;
L
Linus Torvalds 已提交
2000
	calldata->state = state;
2001
	calldata->arg.fh = NFS_FH(state->inode);
2002
	calldata->arg.stateid = &state->open_stateid;
L
Linus Torvalds 已提交
2003
	/* Serialization for the sequence id */
2004
	calldata->arg.seqid = nfs_alloc_seqid(&state->owner->so_seqid, gfp_mask);
2005 2006
	if (calldata->arg.seqid == NULL)
		goto out_free_calldata;
2007
	calldata->arg.fmode = 0;
2008
	calldata->arg.bitmask = server->cache_consistency_bitmask;
2009
	calldata->res.fattr = &calldata->fattr;
2010
	calldata->res.seqid = calldata->arg.seqid;
2011
	calldata->res.server = server;
F
Fred Isaman 已提交
2012
	calldata->roc = roc;
2013 2014
	path_get(path);
	calldata->path = *path;
2015

2016 2017
	msg.rpc_argp = &calldata->arg;
	msg.rpc_resp = &calldata->res;
T
Trond Myklebust 已提交
2018 2019
	task_setup_data.callback_data = calldata;
	task = rpc_run_task(&task_setup_data);
2020 2021
	if (IS_ERR(task))
		return PTR_ERR(task);
2022 2023 2024
	status = 0;
	if (wait)
		status = rpc_wait_for_completion_task(task);
2025
	rpc_put_task(task);
2026
	return status;
2027 2028 2029
out_free_calldata:
	kfree(calldata);
out:
F
Fred Isaman 已提交
2030 2031
	if (roc)
		pnfs_roc_release(state->inode);
2032 2033
	nfs4_put_open_state(state);
	nfs4_put_state_owner(sp);
2034
	return status;
L
Linus Torvalds 已提交
2035 2036
}

2037
static struct inode *
2038
nfs4_atomic_open(struct inode *dir, struct nfs_open_context *ctx, int open_flags, struct iattr *attr)
L
Linus Torvalds 已提交
2039 2040 2041
{
	struct nfs4_state *state;

T
Trond Myklebust 已提交
2042
	/* Protect against concurrent sillydeletes */
2043
	state = nfs4_do_open(dir, &ctx->path, ctx->mode, open_flags, attr, ctx->cred);
2044 2045
	if (IS_ERR(state))
		return ERR_CAST(state);
2046
	ctx->state = state;
2047
	return igrab(state->inode);
L
Linus Torvalds 已提交
2048 2049
}

2050
static void nfs4_close_context(struct nfs_open_context *ctx, int is_sync)
T
Trond Myklebust 已提交
2051 2052 2053 2054 2055 2056 2057 2058
{
	if (ctx->state == NULL)
		return;
	if (is_sync)
		nfs4_close_sync(&ctx->path, ctx->state, ctx->mode);
	else
		nfs4_close_state(&ctx->path, ctx->state, ctx->mode);
}
L
Linus Torvalds 已提交
2059 2060 2061

static int _nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
{
B
Benny Halevy 已提交
2062 2063 2064
	struct nfs4_server_caps_arg args = {
		.fhandle = fhandle,
	};
L
Linus Torvalds 已提交
2065 2066 2067
	struct nfs4_server_caps_res res = {};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SERVER_CAPS],
B
Benny Halevy 已提交
2068
		.rpc_argp = &args,
L
Linus Torvalds 已提交
2069 2070 2071 2072
		.rpc_resp = &res,
	};
	int status;

A
Andy Adamson 已提交
2073
	status = nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
2074 2075
	if (status == 0) {
		memcpy(server->attr_bitmask, res.attr_bitmask, sizeof(server->attr_bitmask));
2076 2077 2078 2079 2080
		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
2081 2082 2083 2084 2085 2086
		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;
2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103
		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;

2104 2105 2106
		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
2107 2108
		server->acl_bitmask = res.acl_bitmask;
	}
A
Andy Adamson 已提交
2109

L
Linus Torvalds 已提交
2110 2111 2112
	return status;
}

2113
int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
L
Linus Torvalds 已提交
2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132
{
	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,
2133
		.fattr = info->fattr,
L
Linus Torvalds 已提交
2134 2135 2136 2137 2138 2139 2140
		.fh = fhandle,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP_ROOT],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
2141

2142
	nfs_fattr_init(info->fattr);
2143
	return nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158
}

static int nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle,
		struct nfs_fsinfo *info)
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(server,
				_nfs4_lookup_root(server, fhandle, info),
				&exception);
	} while (exception.retry);
	return err;
}

2159 2160 2161
/*
 * get the file handle for the "/" directory on the server
 */
L
Linus Torvalds 已提交
2162
static int nfs4_proc_get_root(struct nfs_server *server, struct nfs_fh *fhandle,
2163
			      struct nfs_fsinfo *info)
L
Linus Torvalds 已提交
2164 2165 2166 2167 2168 2169 2170 2171
{
	int status;

	status = nfs4_lookup_root(server, fhandle, info);
	if (status == 0)
		status = nfs4_server_capabilities(server, fhandle);
	if (status == 0)
		status = nfs4_do_fsinfo(server, fhandle, info);
2172
	return nfs4_map_errors(status);
L
Linus Torvalds 已提交
2173 2174
}

M
Manoj Naik 已提交
2175 2176 2177 2178 2179
/*
 * 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
 */
2180
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
2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192
{
	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;

2193
	status = nfs4_proc_fs_locations(dir, name, locations, page);
M
Manoj Naik 已提交
2194 2195 2196 2197
	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)) {
2198
		dprintk("%s: server did not return a different fsid for a referral at %s\n", __func__, name->name);
M
Manoj Naik 已提交
2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210
		status = -EIO;
		goto out;
	}

	memcpy(fattr, &locations->fattr, sizeof(struct nfs_fattr));
	fattr->valid |= NFS_ATTR_FATTR_V4_REFERRAL;
	if (!fattr->mode)
		fattr->mode = S_IFDIR;
	memset(fhandle, 0, sizeof(struct nfs_fh));
out:
	if (page)
		__free_page(page);
2211
	kfree(locations);
M
Manoj Naik 已提交
2212 2213 2214
	return status;
}

L
Linus Torvalds 已提交
2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230
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,
	};
	
2231
	nfs_fattr_init(fattr);
A
Andy Adamson 已提交
2232
	return nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267
}

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)
{
2268
	struct inode *inode = dentry->d_inode;
2269
	struct rpc_cred *cred = NULL;
2270
	struct nfs4_state *state = NULL;
L
Linus Torvalds 已提交
2271 2272
	int status;

2273
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
2274
	
2275
	/* Search for an existing open(O_WRITE) file */
2276 2277 2278 2279
	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
2280 2281 2282 2283
		if (ctx) {
			cred = ctx->cred;
			state = ctx->state;
		}
2284
	}
2285

2286
	status = nfs4_do_setattr(inode, cred, fattr, sattr, state);
2287 2288
	if (status == 0)
		nfs_setattr_update_inode(inode, sattr);
L
Linus Torvalds 已提交
2289 2290 2291
	return status;
}

2292 2293
static int _nfs4_proc_lookupfh(struct nfs_server *server, const struct nfs_fh *dirfh,
		const struct qstr *name, struct nfs_fh *fhandle,
D
David Howells 已提交
2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315
		struct nfs_fattr *fattr)
{
	int		       status;
	struct nfs4_lookup_arg args = {
		.bitmask = server->attr_bitmask,
		.dir_fh = dirfh,
		.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);

	dprintk("NFS call  lookupfh %s\n", name->name);
A
Andy Adamson 已提交
2316
	status = nfs4_call_sync(server, &msg, &args, &res, 0);
D
David Howells 已提交
2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327
	dprintk("NFS reply lookupfh: %d\n", status);
	return status;
}

static int nfs4_proc_lookupfh(struct nfs_server *server, struct nfs_fh *dirfh,
			      struct qstr *name, struct nfs_fh *fhandle,
			      struct nfs_fattr *fattr)
{
	struct nfs4_exception exception = { };
	int err;
	do {
2328 2329 2330 2331 2332 2333 2334
		err = _nfs4_proc_lookupfh(server, dirfh, name, fhandle, fattr);
		/* FIXME: !!!! */
		if (err == -NFS4ERR_MOVED) {
			err = -EREMOTE;
			break;
		}
		err = nfs4_handle_exception(server, err, &exception);
D
David Howells 已提交
2335 2336 2337 2338
	} while (exception.retry);
	return err;
}

2339
static int _nfs4_proc_lookup(struct inode *dir, const struct qstr *name,
L
Linus Torvalds 已提交
2340 2341
		struct nfs_fh *fhandle, struct nfs_fattr *fattr)
{
2342
	int status;
L
Linus Torvalds 已提交
2343 2344
	
	dprintk("NFS call  lookup %s\n", name->name);
2345
	status = _nfs4_proc_lookupfh(NFS_SERVER(dir), NFS_FH(dir), name, fhandle, fattr);
M
Manoj Naik 已提交
2346 2347
	if (status == -NFS4ERR_MOVED)
		status = nfs4_get_referral(dir, name, fattr, fhandle);
L
Linus Torvalds 已提交
2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365
	dprintk("NFS reply lookup: %d\n", status);
	return status;
}

static int nfs4_proc_lookup(struct inode *dir, struct qstr *name, struct nfs_fh *fhandle, struct nfs_fattr *fattr)
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(NFS_SERVER(dir),
				_nfs4_proc_lookup(dir, name, fhandle, fattr),
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
{
2366
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
2367 2368
	struct nfs4_accessargs args = {
		.fh = NFS_FH(inode),
2369 2370 2371 2372
		.bitmask = server->attr_bitmask,
	};
	struct nfs4_accessres res = {
		.server = server,
L
Linus Torvalds 已提交
2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398
	};
	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;
	}
2399 2400 2401 2402 2403

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

A
Andy Adamson 已提交
2404
	status = nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
2405 2406 2407 2408 2409 2410 2411 2412
	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;
2413
		nfs_refresh_inode(inode, res.fattr);
L
Linus Torvalds 已提交
2414
	}
2415
	nfs_free_fattr(res.fattr);
L
Linus Torvalds 已提交
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 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463
	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
2464
	struct nfs4_readlink_res res;
L
Linus Torvalds 已提交
2465 2466 2467
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READLINK],
		.rpc_argp = &args,
B
Benny Halevy 已提交
2468
		.rpc_resp = &res,
L
Linus Torvalds 已提交
2469 2470
	};

A
Andy Adamson 已提交
2471
	return nfs4_call_sync(NFS_SERVER(inode), &msg, &args, &res, 0);
L
Linus Torvalds 已提交
2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503
}

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,
2504
                 int flags, struct nfs_open_context *ctx)
L
Linus Torvalds 已提交
2505
{
2506
	struct path my_path = {
2507 2508
		.dentry = dentry,
	};
2509
	struct path *path = &my_path;
L
Linus Torvalds 已提交
2510
	struct nfs4_state *state;
2511 2512
	struct rpc_cred *cred = NULL;
	fmode_t fmode = 0;
L
Linus Torvalds 已提交
2513 2514
	int status = 0;

2515 2516 2517 2518
	if (ctx != NULL) {
		cred = ctx->cred;
		path = &ctx->path;
		fmode = ctx->mode;
L
Linus Torvalds 已提交
2519
	}
A
Aneesh Kumar K.V 已提交
2520
	sattr->ia_mode &= ~current_umask();
2521
	state = nfs4_do_open(dir, path, fmode, flags, sattr, cred);
2522
	d_drop(dentry);
L
Linus Torvalds 已提交
2523 2524
	if (IS_ERR(state)) {
		status = PTR_ERR(state);
2525
		goto out;
L
Linus Torvalds 已提交
2526
	}
2527
	d_add(dentry, igrab(state->inode));
2528
	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
2529 2530
	if (ctx != NULL)
		ctx->state = state;
2531
	else
2532
		nfs4_close_sync(path, state, fmode);
L
Linus Torvalds 已提交
2533 2534 2535 2536 2537 2538
out:
	return status;
}

static int _nfs4_proc_remove(struct inode *dir, struct qstr *name)
{
2539
	struct nfs_server *server = NFS_SERVER(dir);
2540
	struct nfs_removeargs args = {
L
Linus Torvalds 已提交
2541
		.fh = NFS_FH(dir),
2542 2543
		.name.len = name->len,
		.name.name = name->name,
2544 2545
		.bitmask = server->attr_bitmask,
	};
2546
	struct nfs_removeres res = {
2547
		.server = server,
L
Linus Torvalds 已提交
2548 2549
	};
	struct rpc_message msg = {
2550 2551 2552
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE],
		.rpc_argp = &args,
		.rpc_resp = &res,
L
Linus Torvalds 已提交
2553
	};
2554 2555 2556 2557 2558
	int status = -ENOMEM;

	res.dir_attr = nfs_alloc_fattr();
	if (res.dir_attr == NULL)
		goto out;
L
Linus Torvalds 已提交
2559

A
Andy Adamson 已提交
2560
	status = nfs4_call_sync(server, &msg, &args, &res, 1);
2561 2562
	if (status == 0) {
		update_changeattr(dir, &res.cinfo);
2563
		nfs_post_op_update_inode(dir, res.dir_attr);
2564
	}
2565 2566
	nfs_free_fattr(res.dir_attr);
out:
L
Linus Torvalds 已提交
2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581
	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;
}

2582
static void nfs4_proc_unlink_setup(struct rpc_message *msg, struct inode *dir)
L
Linus Torvalds 已提交
2583
{
2584 2585 2586
	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
2587

2588
	args->bitmask = server->cache_consistency_bitmask;
2589
	res->server = server;
2590
	res->seq_res.sr_slot = NULL;
L
Linus Torvalds 已提交
2591 2592 2593
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE];
}

2594
static int nfs4_proc_unlink_done(struct rpc_task *task, struct inode *dir)
L
Linus Torvalds 已提交
2595
{
2596 2597
	struct nfs_removeres *res = task->tk_msg.rpc_resp;

2598 2599
	if (!nfs4_sequence_done(task, &res->seq_res))
		return 0;
2600
	if (nfs4_async_handle_error(task, res->server, NULL) == -EAGAIN)
2601 2602
		return 0;
	update_changeattr(dir, &res->cinfo);
2603
	nfs_post_op_update_inode(dir, res->dir_attr);
2604
	return 1;
L
Linus Torvalds 已提交
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
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;
}

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
2635 2636 2637
static int _nfs4_proc_rename(struct inode *old_dir, struct qstr *old_name,
		struct inode *new_dir, struct qstr *new_name)
{
2638
	struct nfs_server *server = NFS_SERVER(old_dir);
2639
	struct nfs_renameargs arg = {
L
Linus Torvalds 已提交
2640 2641 2642 2643
		.old_dir = NFS_FH(old_dir),
		.new_dir = NFS_FH(new_dir),
		.old_name = old_name,
		.new_name = new_name,
2644 2645
		.bitmask = server->attr_bitmask,
	};
2646
	struct nfs_renameres res = {
2647
		.server = server,
L
Linus Torvalds 已提交
2648 2649 2650 2651 2652 2653
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RENAME],
		.rpc_argp = &arg,
		.rpc_resp = &res,
	};
2654
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2655
	
2656 2657 2658 2659
	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
2660

2661
	status = nfs4_call_sync(server, &msg, &arg, &res, 1);
L
Linus Torvalds 已提交
2662 2663
	if (!status) {
		update_changeattr(old_dir, &res.old_cinfo);
2664
		nfs_post_op_update_inode(old_dir, res.old_fattr);
L
Linus Torvalds 已提交
2665
		update_changeattr(new_dir, &res.new_cinfo);
2666
		nfs_post_op_update_inode(new_dir, res.new_fattr);
L
Linus Torvalds 已提交
2667
	}
2668 2669 2670
out:
	nfs_free_fattr(res.new_fattr);
	nfs_free_fattr(res.old_fattr);
L
Linus Torvalds 已提交
2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689
	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)
{
2690
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
2691 2692 2693 2694
	struct nfs4_link_arg arg = {
		.fh     = NFS_FH(inode),
		.dir_fh = NFS_FH(dir),
		.name   = name,
2695 2696 2697 2698
		.bitmask = server->attr_bitmask,
	};
	struct nfs4_link_res res = {
		.server = server,
L
Linus Torvalds 已提交
2699 2700 2701 2702
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LINK],
		.rpc_argp = &arg,
2703
		.rpc_resp = &res,
L
Linus Torvalds 已提交
2704
	};
2705 2706 2707 2708 2709 2710
	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
2711

A
Andy Adamson 已提交
2712
	status = nfs4_call_sync(server, &msg, &arg, &res, 1);
2713 2714 2715
	if (!status) {
		update_changeattr(dir, &res.cinfo);
		nfs_post_op_update_inode(dir, res.dir_attr);
2716
		nfs_post_op_update_inode(inode, res.fattr);
2717
	}
2718 2719 2720
out:
	nfs_free_fattr(res.dir_attr);
	nfs_free_fattr(res.fattr);
L
Linus Torvalds 已提交
2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735
	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;
}

2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774
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)
{
A
Andy Adamson 已提交
2775 2776
	int status = nfs4_call_sync(NFS_SERVER(dir), &data->msg,
				    &data->arg, &data->res, 1);
2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789
	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);
}

2790
static int _nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
2791
		struct page *page, unsigned int len, struct iattr *sattr)
L
Linus Torvalds 已提交
2792
{
2793 2794
	struct nfs4_createdata *data;
	int status = -ENAMETOOLONG;
L
Linus Torvalds 已提交
2795

2796
	if (len > NFS4_MAXPATHLEN)
2797
		goto out;
2798

2799 2800 2801 2802 2803 2804 2805 2806
	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
2807
	
2808 2809 2810 2811
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
2812 2813 2814
	return status;
}

2815
static int nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
2816
		struct page *page, unsigned int len, struct iattr *sattr)
L
Linus Torvalds 已提交
2817 2818 2819 2820 2821
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(NFS_SERVER(dir),
2822 2823
				_nfs4_proc_symlink(dir, dentry, page,
							len, sattr),
L
Linus Torvalds 已提交
2824 2825 2826 2827 2828 2829 2830 2831
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr)
{
2832 2833
	struct nfs4_createdata *data;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2834

2835 2836 2837 2838 2839 2840 2841 2842
	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
2843 2844 2845 2846 2847 2848 2849 2850
	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
2851 2852

	sattr->ia_mode &= ~current_umask();
L
Linus Torvalds 已提交
2853 2854 2855 2856 2857 2858 2859 2860 2861
	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
2862
		u64 cookie, struct page **pages, unsigned int count, int plus)
L
Linus Torvalds 已提交
2863 2864 2865 2866
{
	struct inode		*dir = dentry->d_inode;
	struct nfs4_readdir_arg args = {
		.fh = NFS_FH(dir),
B
Bryan Schumaker 已提交
2867
		.pages = pages,
L
Linus Torvalds 已提交
2868 2869
		.pgbase = 0,
		.count = count,
2870
		.bitmask = NFS_SERVER(dentry->d_inode)->attr_bitmask,
B
Bryan Schumaker 已提交
2871
		.plus = plus,
L
Linus Torvalds 已提交
2872 2873 2874 2875 2876 2877 2878 2879 2880 2881
	};
	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;

2882
	dprintk("%s: dentry = %s/%s, cookie = %Lu\n", __func__,
2883 2884 2885
			dentry->d_parent->d_name.name,
			dentry->d_name.name,
			(unsigned long long)cookie);
L
Linus Torvalds 已提交
2886 2887
	nfs4_setup_readdir(cookie, NFS_COOKIEVERF(dir), dentry, &args);
	res.pgbase = args.pgbase;
A
Andy Adamson 已提交
2888
	status = nfs4_call_sync(NFS_SERVER(dir), &msg, &args, &res, 0);
2889
	if (status >= 0) {
L
Linus Torvalds 已提交
2890
		memcpy(NFS_COOKIEVERF(dir), res.verifier.data, NFS4_VERIFIER_SIZE);
2891 2892
		status += args.pgbase;
	}
2893 2894 2895

	nfs_invalidate_atime(dir);

2896
	dprintk("%s: returns %d\n", __func__, status);
L
Linus Torvalds 已提交
2897 2898 2899 2900
	return status;
}

static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
B
Bryan Schumaker 已提交
2901
		u64 cookie, struct page **pages, unsigned int count, int plus)
L
Linus Torvalds 已提交
2902 2903 2904 2905 2906 2907
{
	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
2908
					pages, count, plus),
L
Linus Torvalds 已提交
2909 2910 2911 2912 2913 2914 2915 2916
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr, dev_t rdev)
{
2917 2918 2919
	struct nfs4_createdata *data;
	int mode = sattr->ia_mode;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2920 2921 2922

	BUG_ON(!(sattr->ia_valid & ATTR_MODE));
	BUG_ON(!S_ISFIFO(mode) && !S_ISBLK(mode) && !S_ISCHR(mode) && !S_ISSOCK(mode));
2923 2924 2925 2926 2927

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

L
Linus Torvalds 已提交
2928
	if (S_ISFIFO(mode))
2929
		data->arg.ftype = NF4FIFO;
L
Linus Torvalds 已提交
2930
	else if (S_ISBLK(mode)) {
2931 2932 2933
		data->arg.ftype = NF4BLK;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
L
Linus Torvalds 已提交
2934 2935
	}
	else if (S_ISCHR(mode)) {
2936 2937 2938
		data->arg.ftype = NF4CHR;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
L
Linus Torvalds 已提交
2939 2940
	}
	
2941 2942 2943 2944
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
2945 2946 2947 2948 2949 2950 2951 2952
	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
2953 2954

	sattr->ia_mode &= ~current_umask();
L
Linus Torvalds 已提交
2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969
	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
2970 2971 2972
	struct nfs4_statfs_res res = {
		.fsstat = fsstat,
	};
L
Linus Torvalds 已提交
2973 2974 2975
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_STATFS],
		.rpc_argp = &args,
B
Benny Halevy 已提交
2976
		.rpc_resp = &res,
L
Linus Torvalds 已提交
2977 2978
	};

2979
	nfs_fattr_init(fsstat->fattr);
A
Andy Adamson 已提交
2980
	return  nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001
}

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
3002 3003 3004
	struct nfs4_fsinfo_res res = {
		.fsinfo = fsinfo,
	};
L
Linus Torvalds 已提交
3005 3006 3007
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FSINFO],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3008
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3009 3010
	};

A
Andy Adamson 已提交
3011
	return nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028
}

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)
{
3029
	nfs_fattr_init(fsinfo->fattr);
L
Linus Torvalds 已提交
3030 3031 3032 3033 3034 3035 3036 3037 3038 3039
	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
3040 3041 3042
	struct nfs4_pathconf_res res = {
		.pathconf = pathconf,
	};
L
Linus Torvalds 已提交
3043 3044 3045
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_PATHCONF],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3046
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3047 3048 3049 3050 3051 3052 3053 3054
	};

	/* 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;
	}

3055
	nfs_fattr_init(pathconf->fattr);
A
Andy Adamson 已提交
3056
	return nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072
}

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;
}

T
Trond Myklebust 已提交
3073
static int nfs4_read_done(struct rpc_task *task, struct nfs_read_data *data)
L
Linus Torvalds 已提交
3074
{
T
Trond Myklebust 已提交
3075
	struct nfs_server *server = NFS_SERVER(data->inode);
L
Linus Torvalds 已提交
3076

3077 3078
	dprintk("--> %s\n", __func__);

3079 3080
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return -EAGAIN;
3081

3082
	if (nfs4_async_handle_error(task, server, data->args.context->state) == -EAGAIN) {
3083
		nfs_restart_rpc(task, server->nfs_client);
T
Trond Myklebust 已提交
3084
		return -EAGAIN;
L
Linus Torvalds 已提交
3085
	}
3086 3087

	nfs_invalidate_atime(data->inode);
L
Linus Torvalds 已提交
3088
	if (task->tk_status > 0)
T
Trond Myklebust 已提交
3089 3090
		renew_lease(server, data->timestamp);
	return 0;
L
Linus Torvalds 已提交
3091 3092
}

3093
static void nfs4_proc_read_setup(struct nfs_read_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
3094 3095
{
	data->timestamp   = jiffies;
3096
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READ];
L
Linus Torvalds 已提交
3097 3098
}

3099
static int nfs4_write_done(struct rpc_task *task, struct nfs_write_data *data)
L
Linus Torvalds 已提交
3100 3101 3102
{
	struct inode *inode = data->inode;
	
3103 3104
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return -EAGAIN;
3105

3106
	if (nfs4_async_handle_error(task, NFS_SERVER(inode), data->args.context->state) == -EAGAIN) {
3107
		nfs_restart_rpc(task, NFS_SERVER(inode)->nfs_client);
3108
		return -EAGAIN;
L
Linus Torvalds 已提交
3109
	}
3110
	if (task->tk_status >= 0) {
L
Linus Torvalds 已提交
3111
		renew_lease(NFS_SERVER(inode), data->timestamp);
3112
		nfs_post_op_update_inode_force_wcc(inode, data->res.fattr);
3113
	}
3114
	return 0;
L
Linus Torvalds 已提交
3115 3116
}

3117
static void nfs4_proc_write_setup(struct nfs_write_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
3118
{
3119 3120
	struct nfs_server *server = NFS_SERVER(data->inode);

3121
	data->args.bitmask = server->cache_consistency_bitmask;
3122
	data->res.server = server;
L
Linus Torvalds 已提交
3123 3124
	data->timestamp   = jiffies;

3125
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_WRITE];
L
Linus Torvalds 已提交
3126 3127
}

3128
static int nfs4_commit_done(struct rpc_task *task, struct nfs_write_data *data)
L
Linus Torvalds 已提交
3129 3130 3131
{
	struct inode *inode = data->inode;
	
3132 3133 3134
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return -EAGAIN;

3135
	if (nfs4_async_handle_error(task, NFS_SERVER(inode), NULL) == -EAGAIN) {
3136
		nfs_restart_rpc(task, NFS_SERVER(inode)->nfs_client);
3137
		return -EAGAIN;
L
Linus Torvalds 已提交
3138
	}
3139
	nfs_refresh_inode(inode, data->res.fattr);
3140
	return 0;
L
Linus Torvalds 已提交
3141 3142
}

3143
static void nfs4_proc_commit_setup(struct nfs_write_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
3144
{
3145
	struct nfs_server *server = NFS_SERVER(data->inode);
L
Linus Torvalds 已提交
3146
	
3147
	data->args.bitmask = server->cache_consistency_bitmask;
3148
	data->res.server = server;
3149
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_COMMIT];
L
Linus Torvalds 已提交
3150 3151
}

3152 3153 3154 3155 3156
struct nfs4_renewdata {
	struct nfs_client	*client;
	unsigned long		timestamp;
};

L
Linus Torvalds 已提交
3157 3158 3159 3160
/*
 * nfs4_proc_async_renew(): This is not one of the nfs_rpc_ops; it is a special
 * standalone procedure for queueing an asynchronous RENEW.
 */
3161
static void nfs4_renew_release(void *calldata)
3162
{
3163 3164
	struct nfs4_renewdata *data = calldata;
	struct nfs_client *clp = data->client;
3165

3166 3167 3168
	if (atomic_read(&clp->cl_count) > 1)
		nfs4_schedule_state_renewal(clp);
	nfs_put_client(clp);
3169
	kfree(data);
3170 3171
}

3172
static void nfs4_renew_done(struct rpc_task *task, void *calldata)
L
Linus Torvalds 已提交
3173
{
3174 3175 3176
	struct nfs4_renewdata *data = calldata;
	struct nfs_client *clp = data->client;
	unsigned long timestamp = data->timestamp;
L
Linus Torvalds 已提交
3177 3178

	if (task->tk_status < 0) {
3179 3180
		/* Unless we're shutting down, schedule state recovery! */
		if (test_bit(NFS_CS_RENEWD, &clp->cl_res_state) != 0)
3181
			nfs4_schedule_lease_recovery(clp);
L
Linus Torvalds 已提交
3182 3183
		return;
	}
3184
	do_renew_lease(clp, timestamp);
L
Linus Torvalds 已提交
3185 3186
}

3187 3188
static const struct rpc_call_ops nfs4_renew_ops = {
	.rpc_call_done = nfs4_renew_done,
3189
	.rpc_release = nfs4_renew_release,
3190 3191
};

3192
int nfs4_proc_async_renew(struct nfs_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
3193 3194 3195 3196
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
3197
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
3198
	};
3199
	struct nfs4_renewdata *data;
L
Linus Torvalds 已提交
3200

3201 3202
	if (!atomic_inc_not_zero(&clp->cl_count))
		return -EIO;
3203 3204 3205 3206 3207
	data = kmalloc(sizeof(*data), GFP_KERNEL);
	if (data == NULL)
		return -ENOMEM;
	data->client = clp;
	data->timestamp = jiffies;
L
Linus Torvalds 已提交
3208
	return rpc_call_async(clp->cl_rpcclient, &msg, RPC_TASK_SOFT,
3209
			&nfs4_renew_ops, data);
L
Linus Torvalds 已提交
3210 3211
}

3212
int nfs4_proc_renew(struct nfs_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
3213 3214 3215 3216
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
3217
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
3218 3219 3220 3221 3222 3223 3224
	};
	unsigned long now = jiffies;
	int status;

	status = rpc_call_sync(clp->cl_rpcclient, &msg, 0);
	if (status < 0)
		return status;
3225
	do_renew_lease(clp, now);
L
Linus Torvalds 已提交
3226 3227 3228
	return 0;
}

3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254
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)

static void buf_to_pages(const void *buf, size_t buflen,
		struct page **pages, unsigned int *pgbase)
{
	const void *p = buf;

	*pgbase = offset_in_page(buf);
	p -= *pgbase;
	while (p < buf + buflen) {
		*(pages++) = virt_to_page(p);
		p += PAGE_CACHE_SIZE;
	}
}

3255 3256 3257 3258 3259 3260 3261 3262 3263
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
3264
		len = min_t(size_t, PAGE_CACHE_SIZE, buflen);
3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283
		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;
}

3284 3285 3286
struct nfs4_cached_acl {
	int cached;
	size_t len;
A
Andrew Morton 已提交
3287
	char data[0];
3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350
};

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);
}

3351
static ssize_t __nfs4_get_acl_uncached(struct inode *inode, void *buf, size_t buflen)
3352 3353 3354 3355 3356 3357 3358
{
	struct page *pages[NFS4ACL_MAXPAGES];
	struct nfs_getaclargs args = {
		.fh = NFS_FH(inode),
		.acl_pages = pages,
		.acl_len = buflen,
	};
B
Benny Halevy 已提交
3359 3360 3361
	struct nfs_getaclres res = {
		.acl_len = buflen,
	};
3362
	void *resp_buf;
3363 3364 3365
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETACL],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3366
		.rpc_resp = &res,
3367
	};
3368
	struct page *localpage = NULL;
3369 3370
	int ret;

3371 3372 3373 3374 3375 3376 3377 3378 3379
	if (buflen < PAGE_SIZE) {
		/* As long as we're doing a round trip to the server anyway,
		 * let's be prepared for a page of acl data. */
		localpage = alloc_page(GFP_KERNEL);
		resp_buf = page_address(localpage);
		if (localpage == NULL)
			return -ENOMEM;
		args.acl_pages[0] = localpage;
		args.acl_pgbase = 0;
B
Benny Halevy 已提交
3380
		args.acl_len = PAGE_SIZE;
3381 3382 3383 3384
	} else {
		resp_buf = buf;
		buf_to_pages(buf, buflen, args.acl_pages, &args.acl_pgbase);
	}
A
Andy Adamson 已提交
3385
	ret = nfs4_call_sync(NFS_SERVER(inode), &msg, &args, &res, 0);
3386 3387
	if (ret)
		goto out_free;
B
Benny Halevy 已提交
3388 3389
	if (res.acl_len > args.acl_len)
		nfs4_write_cached_acl(inode, NULL, res.acl_len);
3390
	else
B
Benny Halevy 已提交
3391
		nfs4_write_cached_acl(inode, resp_buf, res.acl_len);
3392 3393
	if (buf) {
		ret = -ERANGE;
B
Benny Halevy 已提交
3394
		if (res.acl_len > buflen)
3395 3396
			goto out_free;
		if (localpage)
B
Benny Halevy 已提交
3397
			memcpy(buf, resp_buf, res.acl_len);
3398
	}
B
Benny Halevy 已提交
3399
	ret = res.acl_len;
3400 3401 3402
out_free:
	if (localpage)
		__free_page(localpage);
3403 3404 3405
	return ret;
}

3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418
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;
}

3419 3420 3421 3422 3423 3424 3425 3426 3427 3428
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;
3429 3430
	if (NFS_I(inode)->cache_validity & NFS_INO_INVALID_ACL)
		nfs_zap_acl_cache(inode);
3431 3432 3433 3434 3435 3436
	ret = nfs4_read_cached_acl(inode, buf, buflen);
	if (ret != -ENOENT)
		return ret;
	return nfs4_get_acl_uncached(inode, buf, buflen);
}

3437
static int __nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen)
3438 3439 3440 3441 3442 3443 3444 3445
{
	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
3446
	struct nfs_setaclres res;
3447 3448 3449
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETACL],
		.rpc_argp	= &arg,
B
Benny Halevy 已提交
3450
		.rpc_resp	= &res,
3451
	};
3452
	int ret, i;
3453 3454 3455

	if (!nfs4_server_supports_acls(server))
		return -EOPNOTSUPP;
3456 3457 3458
	i = buf_to_pages_noslab(buf, buflen, arg.acl_pages, &arg.acl_pgbase);
	if (i < 0)
		return i;
3459
	nfs_inode_return_delegation(inode);
A
Andy Adamson 已提交
3460
	ret = nfs4_call_sync(server, &msg, &arg, &res, 1);
3461 3462 3463 3464 3465 3466 3467 3468

	/*
	 * 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]);

3469 3470 3471 3472 3473 3474 3475
	/*
	 * 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);
3476 3477
	nfs_access_zap_cache(inode);
	nfs_zap_acl_cache(inode);
3478 3479 3480
	return ret;
}

3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492
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
3493
static int
3494
nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server, struct nfs4_state *state)
L
Linus Torvalds 已提交
3495
{
3496 3497 3498
	struct nfs_client *clp = server->nfs_client;

	if (task->tk_status >= 0)
L
Linus Torvalds 已提交
3499 3500
		return 0;
	switch(task->tk_status) {
3501 3502 3503 3504 3505
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OPENMODE:
			if (state == NULL)
				break;
3506 3507
			nfs4_schedule_stateid_recovery(server, state);
			goto wait_on_recovery;
L
Linus Torvalds 已提交
3508
		case -NFS4ERR_STALE_STATEID:
3509
		case -NFS4ERR_STALE_CLIENTID:
L
Linus Torvalds 已提交
3510
		case -NFS4ERR_EXPIRED:
3511 3512
			nfs4_schedule_lease_recovery(clp);
			goto wait_on_recovery;
3513 3514 3515 3516 3517 3518 3519 3520 3521 3522
#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);
3523
			nfs4_schedule_session_recovery(clp->cl_session);
3524 3525 3526
			task->tk_status = 0;
			return -EAGAIN;
#endif /* CONFIG_NFS_V4_1 */
L
Linus Torvalds 已提交
3527
		case -NFS4ERR_DELAY:
3528
			nfs_inc_server_stats(server, NFSIOS_DELAY);
3529
		case -NFS4ERR_GRACE:
3530
		case -EKEYEXPIRED:
L
Linus Torvalds 已提交
3531 3532 3533 3534 3535 3536 3537 3538 3539
			rpc_delay(task, NFS4_POLL_RETRY_MAX);
			task->tk_status = 0;
			return -EAGAIN;
		case -NFS4ERR_OLD_STATEID:
			task->tk_status = 0;
			return -EAGAIN;
	}
	task->tk_status = nfs4_map_errors(task->tk_status);
	return 0;
3540
wait_on_recovery:
3541 3542 3543 3544 3545
	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
3546 3547
}

3548 3549 3550
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
3551 3552 3553 3554 3555
{
	nfs4_verifier sc_verifier;
	struct nfs4_setclientid setclientid = {
		.sc_verifier = &sc_verifier,
		.sc_prog = program,
A
Andy Adamson 已提交
3556
		.sc_cb_ident = clp->cl_cb_ident,
L
Linus Torvalds 已提交
3557 3558 3559 3560
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID],
		.rpc_argp = &setclientid,
3561
		.rpc_resp = res,
3562
		.rpc_cred = cred,
L
Linus Torvalds 已提交
3563
	};
A
Al Viro 已提交
3564
	__be32 *p;
L
Linus Torvalds 已提交
3565 3566 3567
	int loop = 0;
	int status;

A
Al Viro 已提交
3568
	p = (__be32*)sc_verifier.data;
L
Linus Torvalds 已提交
3569 3570 3571 3572 3573
	*p++ = htonl((u32)clp->cl_boot_time.tv_sec);
	*p = htonl((u32)clp->cl_boot_time.tv_nsec);

	for(;;) {
		setclientid.sc_name_len = scnprintf(setclientid.sc_name,
3574
				sizeof(setclientid.sc_name), "%s/%s %s %s %u",
3575 3576 3577
				clp->cl_ipaddr,
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_ADDR),
3578 3579
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_PROTO),
3580
				clp->cl_rpcclient->cl_auth->au_ops->au_name,
L
Linus Torvalds 已提交
3581 3582
				clp->cl_id_uniquifier);
		setclientid.sc_netid_len = scnprintf(setclientid.sc_netid,
3583 3584 3585
				sizeof(setclientid.sc_netid),
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_NETID));
L
Linus Torvalds 已提交
3586
		setclientid.sc_uaddr_len = scnprintf(setclientid.sc_uaddr,
3587
				sizeof(setclientid.sc_uaddr), "%s.%u.%u",
L
Linus Torvalds 已提交
3588 3589 3590 3591 3592 3593 3594 3595
				clp->cl_ipaddr, port >> 8, port & 255);

		status = rpc_call_sync(clp->cl_rpcclient, &msg, 0);
		if (status != -NFS4ERR_CLID_INUSE)
			break;
		if (signalled())
			break;
		if (loop++ & 1)
3596
			ssleep(clp->cl_lease_time / HZ + 1);
L
Linus Torvalds 已提交
3597 3598 3599 3600 3601 3602 3603
		else
			if (++clp->cl_id_uniquifier == 0)
				break;
	}
	return status;
}

3604 3605 3606
static int _nfs4_proc_setclientid_confirm(struct nfs_client *clp,
		struct nfs4_setclientid_res *arg,
		struct rpc_cred *cred)
L
Linus Torvalds 已提交
3607 3608 3609 3610
{
	struct nfs_fsinfo fsinfo;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID_CONFIRM],
3611
		.rpc_argp = arg,
L
Linus Torvalds 已提交
3612
		.rpc_resp = &fsinfo,
3613
		.rpc_cred = cred,
L
Linus Torvalds 已提交
3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628
	};
	unsigned long now;
	int status;

	now = jiffies;
	status = rpc_call_sync(clp->cl_rpcclient, &msg, 0);
	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;
}

3629 3630 3631
int nfs4_proc_setclientid_confirm(struct nfs_client *clp,
		struct nfs4_setclientid_res *arg,
		struct rpc_cred *cred)
3632
{
3633
	long timeout = 0;
3634 3635
	int err;
	do {
3636
		err = _nfs4_proc_setclientid_confirm(clp, arg, cred);
3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648
		switch (err) {
			case 0:
				return err;
			case -NFS4ERR_RESOURCE:
				/* The IBM lawyers misread another document! */
			case -NFS4ERR_DELAY:
				err = nfs4_delay(clp->cl_rpcclient, &timeout);
		}
	} while (err == 0);
	return err;
}

3649 3650
struct nfs4_delegreturndata {
	struct nfs4_delegreturnargs args;
3651
	struct nfs4_delegreturnres res;
3652 3653
	struct nfs_fh fh;
	nfs4_stateid stateid;
3654
	unsigned long timestamp;
3655
	struct nfs_fattr fattr;
3656 3657 3658 3659 3660 3661
	int rpc_status;
};

static void nfs4_delegreturn_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_delegreturndata *data = calldata;
3662

3663 3664
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return;
3665

3666 3667 3668 3669
	switch (task->tk_status) {
	case -NFS4ERR_STALE_STATEID:
	case -NFS4ERR_EXPIRED:
	case 0:
3670
		renew_lease(data->res.server, data->timestamp);
3671 3672 3673 3674 3675 3676 3677 3678 3679
		break;
	default:
		if (nfs4_async_handle_error(task, data->res.server, NULL) ==
				-EAGAIN) {
			nfs_restart_rpc(task, data->res.server->nfs_client);
			return;
		}
	}
	data->rpc_status = task->tk_status;
3680 3681 3682 3683 3684 3685 3686
}

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

3687 3688 3689 3690 3691 3692 3693
#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;

3694
	if (nfs4_setup_sequence(d_data->res.server,
3695 3696 3697 3698 3699 3700 3701
				&d_data->args.seq_args,
				&d_data->res.seq_res, 1, task))
		return;
	rpc_call_start(task);
}
#endif /* CONFIG_NFS_V4_1 */

3702
static const struct rpc_call_ops nfs4_delegreturn_ops = {
3703 3704 3705
#if defined(CONFIG_NFS_V4_1)
	.rpc_call_prepare = nfs4_delegreturn_prepare,
#endif /* CONFIG_NFS_V4_1 */
3706 3707 3708 3709
	.rpc_call_done = nfs4_delegreturn_done,
	.rpc_release = nfs4_delegreturn_release,
};

3710
static int _nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
3711 3712
{
	struct nfs4_delegreturndata *data;
3713
	struct nfs_server *server = NFS_SERVER(inode);
3714
	struct rpc_task *task;
3715 3716 3717 3718
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DELEGRETURN],
		.rpc_cred = cred,
	};
T
Trond Myklebust 已提交
3719 3720
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
3721
		.rpc_message = &msg,
T
Trond Myklebust 已提交
3722 3723 3724
		.callback_ops = &nfs4_delegreturn_ops,
		.flags = RPC_TASK_ASYNC,
	};
3725
	int status = 0;
3726

3727
	data = kzalloc(sizeof(*data), GFP_NOFS);
3728 3729 3730 3731
	if (data == NULL)
		return -ENOMEM;
	data->args.fhandle = &data->fh;
	data->args.stateid = &data->stateid;
3732
	data->args.bitmask = server->attr_bitmask;
3733 3734
	nfs_copy_fh(&data->fh, NFS_FH(inode));
	memcpy(&data->stateid, stateid, sizeof(data->stateid));
3735 3736
	data->res.fattr = &data->fattr;
	data->res.server = server;
3737
	nfs_fattr_init(data->res.fattr);
3738
	data->timestamp = jiffies;
3739 3740
	data->rpc_status = 0;

T
Trond Myklebust 已提交
3741
	task_setup_data.callback_data = data;
3742 3743
	msg.rpc_argp = &data->args;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
3744
	task = rpc_run_task(&task_setup_data);
3745
	if (IS_ERR(task))
3746
		return PTR_ERR(task);
3747 3748
	if (!issync)
		goto out;
3749
	status = nfs4_wait_for_completion_rpc_task(task);
3750 3751 3752 3753 3754 3755 3756
	if (status != 0)
		goto out;
	status = data->rpc_status;
	if (status != 0)
		goto out;
	nfs_refresh_inode(inode, &data->fattr);
out:
3757
	rpc_put_task(task);
3758
	return status;
L
Linus Torvalds 已提交
3759 3760
}

3761
int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
L
Linus Torvalds 已提交
3762 3763 3764 3765 3766
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs4_exception exception = { };
	int err;
	do {
3767
		err = _nfs4_proc_delegreturn(inode, cred, stateid, issync);
L
Linus Torvalds 已提交
3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787
		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)
{
3788
	schedule_timeout_killable(timeout);
L
Linus Torvalds 已提交
3789 3790 3791 3792 3793 3794 3795 3796 3797 3798
	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);
3799
	struct nfs_client *clp = server->nfs_client;
T
Trond Myklebust 已提交
3800
	struct nfs_lockt_args arg = {
L
Linus Torvalds 已提交
3801
		.fh = NFS_FH(inode),
T
Trond Myklebust 已提交
3802
		.fl = request,
L
Linus Torvalds 已提交
3803
	};
T
Trond Myklebust 已提交
3804 3805
	struct nfs_lockt_res res = {
		.denied = request,
L
Linus Torvalds 已提交
3806 3807 3808 3809 3810 3811 3812 3813 3814 3815
	};
	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
3816
	arg.lock_owner.clientid = clp->cl_clientid;
3817 3818 3819 3820
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
	lsp = request->fl_u.nfs4_fl.owner;
3821
	arg.lock_owner.id = lsp->ls_id.id;
3822
	arg.lock_owner.s_dev = server->s_dev;
A
Andy Adamson 已提交
3823
	status = nfs4_call_sync(server, &msg, &arg, &res, 1);
T
Trond Myklebust 已提交
3824 3825 3826 3827 3828 3829
	switch (status) {
		case 0:
			request->fl_type = F_UNLCK;
			break;
		case -NFS4ERR_DENIED:
			status = 0;
L
Linus Torvalds 已提交
3830
	}
3831
	request->fl_ops->fl_release_private(request);
3832
out:
L
Linus Torvalds 已提交
3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864
	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;
}

3865
struct nfs4_unlockdata {
T
Trond Myklebust 已提交
3866 3867
	struct nfs_locku_args arg;
	struct nfs_locku_res res;
3868 3869
	struct nfs4_lock_state *lsp;
	struct nfs_open_context *ctx;
T
Trond Myklebust 已提交
3870 3871
	struct file_lock fl;
	const struct nfs_server *server;
3872
	unsigned long timestamp;
3873 3874
};

T
Trond Myklebust 已提交
3875 3876 3877 3878 3879 3880 3881 3882
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;

3883
	p = kzalloc(sizeof(*p), GFP_NOFS);
T
Trond Myklebust 已提交
3884 3885 3886 3887 3888
	if (p == NULL)
		return NULL;
	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
	p->arg.seqid = seqid;
3889
	p->res.seqid = seqid;
T
Trond Myklebust 已提交
3890 3891 3892 3893 3894 3895 3896 3897 3898 3899
	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;
}

3900
static void nfs4_locku_release_calldata(void *data)
3901
{
3902
	struct nfs4_unlockdata *calldata = data;
T
Trond Myklebust 已提交
3903
	nfs_free_seqid(calldata->arg.seqid);
3904 3905 3906
	nfs4_put_lock_state(calldata->lsp);
	put_nfs_open_context(calldata->ctx);
	kfree(calldata);
3907 3908
}

3909
static void nfs4_locku_done(struct rpc_task *task, void *data)
3910
{
3911
	struct nfs4_unlockdata *calldata = data;
3912

3913 3914
	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
		return;
3915 3916 3917
	switch (task->tk_status) {
		case 0:
			memcpy(calldata->lsp->ls_stateid.data,
T
Trond Myklebust 已提交
3918
					calldata->res.stateid.data,
3919
					sizeof(calldata->lsp->ls_stateid.data));
3920
			renew_lease(calldata->server, calldata->timestamp);
3921
			break;
3922 3923
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OLD_STATEID:
3924 3925 3926 3927
		case -NFS4ERR_STALE_STATEID:
		case -NFS4ERR_EXPIRED:
			break;
		default:
3928
			if (nfs4_async_handle_error(task, calldata->server, NULL) == -EAGAIN)
3929
				nfs_restart_rpc(task,
3930
						 calldata->server->nfs_client);
3931 3932 3933
	}
}

T
Trond Myklebust 已提交
3934
static void nfs4_locku_prepare(struct rpc_task *task, void *data)
3935
{
T
Trond Myklebust 已提交
3936
	struct nfs4_unlockdata *calldata = data;
3937

T
Trond Myklebust 已提交
3938
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
3939 3940
		return;
	if ((calldata->lsp->ls_flags & NFS_LOCK_INITIALIZED) == 0) {
3941 3942
		/* Note: exit _without_ running nfs4_locku_done */
		task->tk_action = NULL;
3943 3944
		return;
	}
3945
	calldata->timestamp = jiffies;
3946
	if (nfs4_setup_sequence(calldata->server,
3947 3948 3949
				&calldata->arg.seq_args,
				&calldata->res.seq_res, 1, task))
		return;
3950
	rpc_call_start(task);
3951 3952
}

3953
static const struct rpc_call_ops nfs4_locku_ops = {
T
Trond Myklebust 已提交
3954
	.rpc_call_prepare = nfs4_locku_prepare,
3955
	.rpc_call_done = nfs4_locku_done,
3956
	.rpc_release = nfs4_locku_release_calldata,
3957 3958
};

3959 3960 3961 3962 3963 3964
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;
3965 3966 3967 3968
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCKU],
		.rpc_cred = ctx->cred,
	};
T
Trond Myklebust 已提交
3969 3970
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(lsp->ls_state->inode),
3971
		.rpc_message = &msg,
T
Trond Myklebust 已提交
3972
		.callback_ops = &nfs4_locku_ops,
3973
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
3974 3975
		.flags = RPC_TASK_ASYNC,
	};
3976

3977 3978 3979 3980 3981
	/* 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;

3982 3983 3984 3985 3986 3987
	data = nfs4_alloc_unlockdata(fl, ctx, lsp, seqid);
	if (data == NULL) {
		nfs_free_seqid(seqid);
		return ERR_PTR(-ENOMEM);
	}

3988 3989
	msg.rpc_argp = &data->arg;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
3990 3991
	task_setup_data.callback_data = data;
	return rpc_run_task(&task_setup_data);
3992 3993
}

3994 3995
static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
{
3996
	struct nfs_inode *nfsi = NFS_I(state->inode);
T
Trond Myklebust 已提交
3997
	struct nfs_seqid *seqid;
L
Linus Torvalds 已提交
3998
	struct nfs4_lock_state *lsp;
3999 4000
	struct rpc_task *task;
	int status = 0;
4001
	unsigned char fl_flags = request->fl_flags;
4002

4003
	status = nfs4_set_lock_state(state, request);
4004 4005
	/* Unlock _before_ we do the RPC call */
	request->fl_flags |= FL_EXISTS;
4006 4007 4008
	down_read(&nfsi->rwsem);
	if (do_vfs_lock(request->fl_file, request) == -ENOENT) {
		up_read(&nfsi->rwsem);
4009
		goto out;
4010 4011
	}
	up_read(&nfsi->rwsem);
4012
	if (status != 0)
4013 4014 4015 4016
		goto out;
	/* Is this a delegated lock? */
	if (test_bit(NFS_DELEGATED_STATE, &state->flags))
		goto out;
4017
	lsp = request->fl_u.nfs4_fl.owner;
4018
	seqid = nfs_alloc_seqid(&lsp->ls_seqid, GFP_KERNEL);
4019
	status = -ENOMEM;
T
Trond Myklebust 已提交
4020
	if (seqid == NULL)
4021
		goto out;
4022
	task = nfs4_do_unlck(request, nfs_file_open_context(request->fl_file), lsp, seqid);
4023 4024
	status = PTR_ERR(task);
	if (IS_ERR(task))
4025
		goto out;
4026
	status = nfs4_wait_for_completion_rpc_task(task);
4027
	rpc_put_task(task);
4028
out:
4029
	request->fl_flags = fl_flags;
L
Linus Torvalds 已提交
4030 4031 4032
	return status;
}

4033 4034 4035 4036 4037 4038
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;
4039
	unsigned long timestamp;
4040 4041
	int rpc_status;
	int cancelled;
4042
	struct nfs_server *server;
4043 4044 4045
};

static struct nfs4_lockdata *nfs4_alloc_lockdata(struct file_lock *fl,
4046 4047
		struct nfs_open_context *ctx, struct nfs4_lock_state *lsp,
		gfp_t gfp_mask)
L
Linus Torvalds 已提交
4048
{
4049 4050
	struct nfs4_lockdata *p;
	struct inode *inode = lsp->ls_state->inode;
L
Linus Torvalds 已提交
4051
	struct nfs_server *server = NFS_SERVER(inode);
4052

4053
	p = kzalloc(sizeof(*p), gfp_mask);
4054 4055 4056 4057 4058
	if (p == NULL)
		return NULL;

	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
4059
	p->arg.open_seqid = nfs_alloc_seqid(&lsp->ls_state->owner->so_seqid, gfp_mask);
4060 4061
	if (p->arg.open_seqid == NULL)
		goto out_free;
4062
	p->arg.lock_seqid = nfs_alloc_seqid(&lsp->ls_seqid, gfp_mask);
4063
	if (p->arg.lock_seqid == NULL)
4064
		goto out_free_seqid;
4065
	p->arg.lock_stateid = &lsp->ls_stateid;
4066
	p->arg.lock_owner.clientid = server->nfs_client->cl_clientid;
4067
	p->arg.lock_owner.id = lsp->ls_id.id;
4068
	p->arg.lock_owner.s_dev = server->s_dev;
4069
	p->res.lock_seqid = p->arg.lock_seqid;
4070
	p->lsp = lsp;
4071
	p->server = server;
4072 4073 4074 4075
	atomic_inc(&lsp->ls_count);
	p->ctx = get_nfs_open_context(ctx);
	memcpy(&p->fl, fl, sizeof(p->fl));
	return p;
4076 4077
out_free_seqid:
	nfs_free_seqid(p->arg.open_seqid);
4078 4079 4080 4081 4082 4083 4084 4085 4086
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;
4087

4088
	dprintk("%s: begin!\n", __func__);
4089 4090
	if (nfs_wait_on_sequence(data->arg.lock_seqid, task) != 0)
		return;
4091 4092
	/* Do we need to do an open_to_lock_owner? */
	if (!(data->arg.lock_seqid->sequence->flags & NFS_SEQID_CONFIRMED)) {
4093 4094
		if (nfs_wait_on_sequence(data->arg.open_seqid, task) != 0)
			return;
4095 4096
		data->arg.open_stateid = &state->stateid;
		data->arg.new_lock_owner = 1;
4097
		data->res.open_seqid = data->arg.open_seqid;
4098 4099
	} else
		data->arg.new_lock_owner = 0;
4100
	data->timestamp = jiffies;
4101 4102
	if (nfs4_setup_sequence(data->server,
				&data->arg.seq_args,
4103 4104
				&data->res.seq_res, 1, task))
		return;
4105
	rpc_call_start(task);
4106
	dprintk("%s: done!, ret = %d\n", __func__, data->rpc_status);
4107 4108
}

4109 4110 4111 4112 4113 4114
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);
}

4115 4116 4117 4118
static void nfs4_lock_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_lockdata *data = calldata;

4119
	dprintk("%s: begin!\n", __func__);
4120

4121 4122
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return;
4123

4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134
	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) {
		memcpy(data->lsp->ls_stateid.data, data->res.stateid.data,
					sizeof(data->lsp->ls_stateid.data));
		data->lsp->ls_flags |= NFS_LOCK_INITIALIZED;
4135
		renew_lease(NFS_SERVER(data->ctx->path.dentry->d_inode), data->timestamp);
4136 4137
	}
out:
4138
	dprintk("%s: done, ret = %d!\n", __func__, data->rpc_status);
4139 4140 4141 4142 4143 4144
}

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

4145
	dprintk("%s: begin!\n", __func__);
4146
	nfs_free_seqid(data->arg.open_seqid);
4147 4148 4149 4150 4151
	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))
4152
			rpc_put_task_async(task);
4153
		dprintk("%s: cancelling lock!\n", __func__);
4154 4155 4156 4157 4158
	} else
		nfs_free_seqid(data->arg.lock_seqid);
	nfs4_put_lock_state(data->lsp);
	put_nfs_open_context(data->ctx);
	kfree(data);
4159
	dprintk("%s: done!\n", __func__);
4160 4161 4162 4163 4164 4165 4166 4167
}

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,
};

4168 4169 4170 4171 4172 4173
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,
};

4174 4175 4176 4177 4178
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:
4179
		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
4180 4181
		if (new_lock_owner != 0 ||
		   (lsp->ls_flags & NFS_LOCK_INITIALIZED) != 0)
4182
			nfs4_schedule_stateid_recovery(server, lsp->ls_state);
4183 4184 4185
		break;
	case -NFS4ERR_STALE_STATEID:
		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
4186 4187
	case -NFS4ERR_EXPIRED:
		nfs4_schedule_lease_recovery(server->nfs_client);
4188 4189 4190
	};
}

4191
static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *fl, int recovery_type)
4192 4193 4194
{
	struct nfs4_lockdata *data;
	struct rpc_task *task;
4195 4196 4197 4198
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCK],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
4199 4200
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(state->inode),
4201
		.rpc_message = &msg,
T
Trond Myklebust 已提交
4202
		.callback_ops = &nfs4_lock_ops,
4203
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
4204 4205
		.flags = RPC_TASK_ASYNC,
	};
4206 4207
	int ret;

4208
	dprintk("%s: begin!\n", __func__);
4209
	data = nfs4_alloc_lockdata(fl, nfs_file_open_context(fl->fl_file),
4210 4211
			fl->fl_u.nfs4_fl.owner,
			recovery_type == NFS_LOCK_NEW ? GFP_KERNEL : GFP_NOFS);
4212 4213 4214 4215
	if (data == NULL)
		return -ENOMEM;
	if (IS_SETLKW(cmd))
		data->arg.block = 1;
4216 4217
	if (recovery_type > NFS_LOCK_NEW) {
		if (recovery_type == NFS_LOCK_RECLAIM)
4218
			data->arg.reclaim = NFS_LOCK_RECLAIM;
4219 4220
		task_setup_data.callback_ops = &nfs4_recover_lock_ops;
	}
4221 4222
	msg.rpc_argp = &data->arg;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
4223 4224
	task_setup_data.callback_data = data;
	task = rpc_run_task(&task_setup_data);
4225
	if (IS_ERR(task))
4226 4227 4228 4229
		return PTR_ERR(task);
	ret = nfs4_wait_for_completion_rpc_task(task);
	if (ret == 0) {
		ret = data->rpc_status;
4230 4231 4232
		if (ret)
			nfs4_handle_setlk_error(data->server, data->lsp,
					data->arg.new_lock_owner, ret);
4233 4234
	} else
		data->cancelled = 1;
4235
	rpc_put_task(task);
4236
	dprintk("%s: done, ret = %d!\n", __func__, ret);
4237
	return ret;
L
Linus Torvalds 已提交
4238 4239 4240 4241
}

static int nfs4_lock_reclaim(struct nfs4_state *state, struct file_lock *request)
{
4242 4243 4244 4245 4246
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;

	do {
4247 4248 4249
		/* Cache the lock if possible... */
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
4250
		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_RECLAIM);
4251
		if (err != -NFS4ERR_DELAY)
4252 4253 4254 4255
			break;
		nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
	return err;
L
Linus Torvalds 已提交
4256 4257 4258 4259
}

static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request)
{
4260 4261 4262 4263
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;

4264 4265 4266
	err = nfs4_set_lock_state(state, request);
	if (err != 0)
		return err;
4267
	do {
4268 4269
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
4270
		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_EXPIRED);
4271 4272 4273 4274 4275 4276 4277 4278
		switch (err) {
		default:
			goto out;
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			nfs4_handle_exception(server, err, &exception);
			err = 0;
		}
4279
	} while (exception.retry);
4280
out:
4281
	return err;
L
Linus Torvalds 已提交
4282 4283 4284 4285
}

static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
4286
	struct nfs_inode *nfsi = NFS_I(state->inode);
4287
	unsigned char fl_flags = request->fl_flags;
4288
	int status = -ENOLCK;
L
Linus Torvalds 已提交
4289

4290 4291 4292
	if ((fl_flags & FL_POSIX) &&
			!test_bit(NFS_STATE_POSIX_LOCKS, &state->flags))
		goto out;
4293 4294 4295 4296
	/* Is this a delegated open? */
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
4297 4298 4299 4300
	request->fl_flags |= FL_ACCESS;
	status = do_vfs_lock(request->fl_file, request);
	if (status < 0)
		goto out;
4301
	down_read(&nfsi->rwsem);
4302 4303 4304
	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
		/* Yes: cache locks! */
		/* ...but avoid races with delegation recall... */
4305 4306 4307
		request->fl_flags = fl_flags & ~FL_SLEEP;
		status = do_vfs_lock(request->fl_file, request);
		goto out_unlock;
4308
	}
4309
	status = _nfs4_do_setlk(state, cmd, request, NFS_LOCK_NEW);
4310
	if (status != 0)
4311
		goto out_unlock;
4312
	/* Note: we always want to sleep here! */
4313
	request->fl_flags = fl_flags | FL_SLEEP;
4314
	if (do_vfs_lock(request->fl_file, request) < 0)
4315
		printk(KERN_WARNING "%s: VFS is out of sync with lock manager!\n", __func__);
4316
out_unlock:
4317
	up_read(&nfsi->rwsem);
4318 4319
out:
	request->fl_flags = fl_flags;
L
Linus Torvalds 已提交
4320 4321 4322 4323 4324 4325 4326 4327 4328
	return status;
}

static int nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
	struct nfs4_exception exception = { };
	int err;

	do {
4329 4330 4331
		err = _nfs4_proc_setlk(state, cmd, request);
		if (err == -NFS4ERR_DENIED)
			err = -EAGAIN;
L
Linus Torvalds 已提交
4332
		err = nfs4_handle_exception(NFS_SERVER(state->inode),
4333
				err, &exception);
L
Linus Torvalds 已提交
4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345 4346
	} 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 */
4347
	ctx = nfs_file_open_context(filp);
L
Linus Torvalds 已提交
4348 4349 4350 4351 4352
	state = ctx->state;

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

4353 4354 4355 4356 4357
	if (IS_GETLK(cmd)) {
		if (state != NULL)
			return nfs4_proc_getlk(state, F_GETLK, request);
		return 0;
	}
L
Linus Torvalds 已提交
4358 4359 4360 4361

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

4362 4363 4364 4365 4366
	if (request->fl_type == F_UNLCK) {
		if (state != NULL)
			return nfs4_proc_unlck(state, cmd, request);
		return 0;
	}
L
Linus Torvalds 已提交
4367

4368 4369
	if (state == NULL)
		return -ENOLCK;
L
Linus Torvalds 已提交
4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381
	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;
}

4382 4383 4384 4385 4386 4387 4388 4389 4390 4391
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 {
4392
		err = _nfs4_do_setlk(state, F_SETLK, fl, NFS_LOCK_NEW);
4393 4394 4395 4396 4397
		switch (err) {
			default:
				printk(KERN_ERR "%s: unhandled error %d.\n",
						__func__, err);
			case 0:
4398
			case -ESTALE:
4399 4400 4401
				goto out;
			case -NFS4ERR_EXPIRED:
			case -NFS4ERR_STALE_CLIENTID:
4402
			case -NFS4ERR_STALE_STATEID:
4403 4404
				nfs4_schedule_lease_recovery(server->nfs_client);
				goto out;
4405 4406 4407 4408 4409
			case -NFS4ERR_BADSESSION:
			case -NFS4ERR_BADSLOT:
			case -NFS4ERR_BAD_HIGH_SLOT:
			case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
			case -NFS4ERR_DEADSESSION:
4410
				nfs4_schedule_session_recovery(server->nfs_client->cl_session);
4411
				goto out;
4412 4413 4414 4415 4416 4417 4418 4419
			case -ERESTARTSYS:
				/*
				 * The show must go on: exit, but mark the
				 * stateid as needing recovery.
				 */
			case -NFS4ERR_ADMIN_REVOKED:
			case -NFS4ERR_BAD_STATEID:
			case -NFS4ERR_OPENMODE:
4420
				nfs4_schedule_stateid_recovery(server, state);
4421 4422
				err = 0;
				goto out;
4423 4424 4425 4426 4427 4428 4429 4430 4431
			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;
4432 4433 4434 4435 4436
			case -ENOMEM:
			case -NFS4ERR_DENIED:
				/* kill_proc(fl->fl_pid, SIGLOST, 1); */
				err = 0;
				goto out;
4437 4438 4439
			case -NFS4ERR_DELAY:
				break;
		}
4440 4441 4442 4443 4444
		err = nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
out:
	return err;
}
4445

4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469
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;
	args->lock_owner.id = lsp->ls_id.id;
4470
	args->lock_owner.s_dev = server->s_dev;
4471 4472 4473 4474
	msg.rpc_argp = args;
	rpc_call_async(server->client, &msg, 0, &nfs4_release_lockowner_ops, args);
}

4475 4476
#define XATTR_NAME_NFSV4_ACL "system.nfs4_acl"

4477 4478 4479
static int nfs4_xattr_set_nfs4_acl(struct dentry *dentry, const char *key,
				   const void *buf, size_t buflen,
				   int flags, int type)
4480
{
4481 4482
	if (strcmp(key, "") != 0)
		return -EINVAL;
4483

4484
	return nfs4_proc_set_acl(dentry->d_inode, buf, buflen);
4485 4486
}

4487 4488
static int nfs4_xattr_get_nfs4_acl(struct dentry *dentry, const char *key,
				   void *buf, size_t buflen, int type)
4489
{
4490 4491
	if (strcmp(key, "") != 0)
		return -EINVAL;
4492

4493
	return nfs4_proc_get_acl(dentry->d_inode, buf, buflen);
4494 4495
}

4496 4497 4498
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)
4499
{
4500
	size_t len = sizeof(XATTR_NAME_NFSV4_ACL);
4501

4502 4503
	if (!nfs4_server_supports_acls(NFS_SERVER(dentry->d_inode)))
		return 0;
4504 4505 4506

	if (list && len <= list_len)
		memcpy(list, XATTR_NAME_NFSV4_ACL, len);
4507
	return len;
4508 4509
}

4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522
static void nfs_fixup_referral_attributes(struct nfs_fattr *fattr)
{
	if (!((fattr->valid & NFS_ATTR_FATTR_FILEID) &&
		(fattr->valid & NFS_ATTR_FATTR_FSID) &&
		(fattr->valid & NFS_ATTR_FATTR_V4_REFERRAL)))
		return;

	fattr->valid |= NFS_ATTR_FATTR_TYPE | NFS_ATTR_FATTR_MODE |
		NFS_ATTR_FATTR_NLINK;
	fattr->mode = S_IFDIR | S_IRUGO | S_IXUGO;
	fattr->nlink = 2;
}

4523
int nfs4_proc_fs_locations(struct inode *dir, const struct qstr *name,
4524
		struct nfs4_fs_locations *fs_locations, struct page *page)
4525 4526 4527
{
	struct nfs_server *server = NFS_SERVER(dir);
	u32 bitmask[2] = {
4528 4529
		[0] = FATTR4_WORD0_FSID | FATTR4_WORD0_FS_LOCATIONS,
		[1] = FATTR4_WORD1_MOUNTED_ON_FILEID,
4530 4531 4532
	};
	struct nfs4_fs_locations_arg args = {
		.dir_fh = NFS_FH(dir),
4533
		.name = name,
4534 4535 4536
		.page = page,
		.bitmask = bitmask,
	};
B
Benny Halevy 已提交
4537 4538 4539
	struct nfs4_fs_locations_res res = {
		.fs_locations = fs_locations,
	};
4540 4541 4542
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FS_LOCATIONS],
		.rpc_argp = &args,
B
Benny Halevy 已提交
4543
		.rpc_resp = &res,
4544 4545 4546
	};
	int status;

4547
	dprintk("%s: start\n", __func__);
4548
	nfs_fattr_init(&fs_locations->fattr);
4549
	fs_locations->server = server;
4550
	fs_locations->nlocations = 0;
A
Andy Adamson 已提交
4551
	status = nfs4_call_sync(server, &msg, &args, &res, 0);
4552
	nfs_fixup_referral_attributes(&fs_locations->fattr);
4553
	dprintk("%s: returned status = %d\n", __func__, status);
4554 4555 4556
	return status;
}

4557
#ifdef CONFIG_NFS_V4_1
4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576
/*
 * 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;
}

B
Benny Halevy 已提交
4577 4578 4579 4580 4581 4582 4583 4584
/*
 * 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.
 */
4585
int nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred)
B
Benny Halevy 已提交
4586 4587 4588 4589
{
	nfs4_verifier verifier;
	struct nfs41_exchange_id_args args = {
		.client = clp,
4590
		.flags = EXCHGID4_FLAG_SUPP_MOVED_REFER,
B
Benny Halevy 已提交
4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605
	};
	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);
4606

B
Benny Halevy 已提交
4607 4608 4609 4610 4611
	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;

4612 4613 4614 4615 4616 4617
	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
4618

4619 4620 4621
	status = rpc_call_sync(clp->cl_rpcclient, &msg, 0);
	if (!status)
		status = nfs4_check_cl_exchange_flags(clp->cl_exchange_flags);
B
Benny Halevy 已提交
4622 4623 4624 4625
	dprintk("<-- %s status= %d\n", __func__, status);
	return status;
}

A
Andy Adamson 已提交
4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639
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__);
4640
	rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
A
Andy Adamson 已提交
4641 4642 4643
	/* just setup sequence, do not trigger session recovery
	   since we're invoked within one */
	ret = nfs41_setup_sequence(data->clp->cl_session,
4644 4645
				   &data->args->la_seq_args,
				   &data->res->lr_seq_res, 0, task);
A
Andy Adamson 已提交
4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661

	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__);
4662 4663
	if (!nfs41_sequence_done(task, &data->res->lr_seq_res))
		return;
A
Andy Adamson 已提交
4664 4665 4666 4667 4668 4669
	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;
4670
		nfs_restart_rpc(task, data->clp);
A
Andy Adamson 已提交
4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719
		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,
		.callback_data = &data
	};
	int status;

	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;
}

4720 4721 4722
/*
 * Reset a slot table
 */
A
Andy Adamson 已提交
4723 4724
static int nfs4_reset_slot_table(struct nfs4_slot_table *tbl, u32 max_reqs,
				 int ivalue)
4725
{
A
Andy Adamson 已提交
4726
	struct nfs4_slot *new = NULL;
4727
	int i;
4728 4729
	int ret = 0;

A
Andy Adamson 已提交
4730 4731
	dprintk("--> %s: max_reqs=%u, tbl->max_slots %d\n", __func__,
		max_reqs, tbl->max_slots);
4732

A
Andy Adamson 已提交
4733 4734 4735 4736
	/* Does the newly negotiated max_reqs match the existing slot table? */
	if (max_reqs != tbl->max_slots) {
		ret = -ENOMEM;
		new = kmalloc(max_reqs * sizeof(struct nfs4_slot),
4737
			      GFP_NOFS);
A
Andy Adamson 已提交
4738 4739 4740 4741
		if (!new)
			goto out;
		ret = 0;
		kfree(tbl->slots);
4742 4743
	}
	spin_lock(&tbl->slot_tbl_lock);
A
Andy Adamson 已提交
4744 4745 4746 4747 4748
	if (new) {
		tbl->slots = new;
		tbl->max_slots = max_reqs;
	}
	for (i = 0; i < tbl->max_slots; ++i)
4749
		tbl->slots[i].seq_nr = ivalue;
4750 4751 4752 4753 4754 4755 4756 4757
	spin_unlock(&tbl->slot_tbl_lock);
	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;
}

4758 4759 4760 4761 4762 4763 4764 4765
/*
 * Reset the forechannel and backchannel slot tables
 */
static int nfs4_reset_slot_tables(struct nfs4_session *session)
{
	int status;

	status = nfs4_reset_slot_table(&session->fc_slot_table,
A
Andy Adamson 已提交
4766
			session->fc_attrs.max_reqs, 1);
4767 4768 4769 4770
	if (status)
		return status;

	status = nfs4_reset_slot_table(&session->bc_slot_table,
A
Andy Adamson 已提交
4771
			session->bc_attrs.max_reqs, 0);
4772 4773 4774
	return status;
}

4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788
/* 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;
}

A
Andy Adamson 已提交
4789 4790 4791
/*
 * Initialize slot table
 */
4792 4793
static int nfs4_init_slot_table(struct nfs4_slot_table *tbl,
		int max_slots, int ivalue)
A
Andy Adamson 已提交
4794 4795 4796 4797 4798 4799
{
	struct nfs4_slot *slot;
	int ret = -ENOMEM;

	BUG_ON(max_slots > NFS4_MAX_SLOT_TABLE);

4800
	dprintk("--> %s: max_reqs=%u\n", __func__, max_slots);
A
Andy Adamson 已提交
4801

4802
	slot = kcalloc(max_slots, sizeof(struct nfs4_slot), GFP_NOFS);
A
Andy Adamson 已提交
4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818
	if (!slot)
		goto out;
	ret = 0;

	spin_lock(&tbl->slot_tbl_lock);
	tbl->max_slots = max_slots;
	tbl->slots = slot;
	tbl->highest_used_slotid = -1;  /* no slot is currently used */
	spin_unlock(&tbl->slot_tbl_lock);
	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;
}

4819 4820 4821 4822 4823
/*
 * Initialize the forechannel and backchannel tables
 */
static int nfs4_init_slot_tables(struct nfs4_session *session)
{
4824 4825
	struct nfs4_slot_table *tbl;
	int status = 0;
4826

4827 4828 4829 4830 4831 4832 4833
	tbl = &session->fc_slot_table;
	if (tbl->slots == NULL) {
		status = nfs4_init_slot_table(tbl,
				session->fc_attrs.max_reqs, 1);
		if (status)
			return status;
	}
4834

4835 4836 4837 4838 4839 4840 4841
	tbl = &session->bc_slot_table;
	if (tbl->slots == NULL) {
		status = nfs4_init_slot_table(tbl,
				session->bc_attrs.max_reqs, 0);
		if (status)
			nfs4_destroy_slot_tables(session);
	}
4842 4843

	return status;
4844 4845 4846 4847 4848 4849 4850
}

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

4851
	session = kzalloc(sizeof(struct nfs4_session), GFP_NOFS);
4852 4853
	if (!session)
		return NULL;
4854

4855
	tbl = &session->fc_slot_table;
4856
	tbl->highest_used_slotid = -1;
4857
	spin_lock_init(&tbl->slot_tbl_lock);
4858
	rpc_init_priority_wait_queue(&tbl->slot_tbl_waitq, "ForeChannel Slot table");
4859
	init_completion(&tbl->complete);
4860 4861

	tbl = &session->bc_slot_table;
4862
	tbl->highest_used_slotid = -1;
4863 4864
	spin_lock_init(&tbl->slot_tbl_lock);
	rpc_init_wait_queue(&tbl->slot_tbl_waitq, "BackChannel Slot table");
4865
	init_completion(&tbl->complete);
4866

4867 4868
	session->session_state = 1<<NFS4_SESSION_INITING;

4869 4870 4871 4872 4873 4874
	session->clp = clp;
	return session;
}

void nfs4_destroy_session(struct nfs4_session *session)
{
4875 4876 4877 4878 4879
	nfs4_proc_destroy_session(session);
	dprintk("%s Destroy backchannel for xprt %p\n",
		__func__, session->clp->cl_rpcclient->cl_xprt);
	xprt_destroy_backchannel(session->clp->cl_rpcclient->cl_xprt,
				NFS41_BC_MIN_CALLBACKS);
4880
	nfs4_destroy_slot_tables(session);
4881 4882 4883
	kfree(session);
}

A
Andy Adamson 已提交
4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910
/*
 * 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.headerpadsz = 0;
	args->fc_attrs.max_rqst_sz = mxrqst_sz;
	args->fc_attrs.max_resp_sz = mxresp_sz;
	args->fc_attrs.max_ops = NFS4_MAX_OPS;
	args->fc_attrs.max_reqs = session->clp->cl_rpcclient->cl_xprt->max_reqs;

	dprintk("%s: Fore Channel : max_rqst_sz=%u max_resp_sz=%u "
4911
		"max_ops=%u max_reqs=%u\n",
A
Andy Adamson 已提交
4912 4913
		__func__,
		args->fc_attrs.max_rqst_sz, args->fc_attrs.max_resp_sz,
4914
		args->fc_attrs.max_ops, args->fc_attrs.max_reqs);
A
Andy Adamson 已提交
4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931

	/* Back channel attributes */
	args->bc_attrs.headerpadsz = 0;
	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);
}

4932
static int nfs4_verify_fore_channel_attrs(struct nfs41_create_session_args *args, struct nfs4_session *session)
4933
{
4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951
	struct nfs4_channel_attrs *sent = &args->fc_attrs;
	struct nfs4_channel_attrs *rcvd = &session->fc_attrs;

	if (rcvd->headerpadsz > sent->headerpadsz)
		return -EINVAL;
	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;
	return 0;
4952 4953
}

4954 4955 4956 4957
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;
4958

4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971
	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: */
	if (rcvd->max_ops  == 0)
		return -EINVAL;
	if (rcvd->max_reqs == 0)
		return -EINVAL;
	return 0;
}
4972 4973 4974 4975

static int nfs4_verify_channel_attrs(struct nfs41_create_session_args *args,
				     struct nfs4_session *session)
{
4976
	int ret;
4977

4978 4979 4980 4981
	ret = nfs4_verify_fore_channel_attrs(args, session);
	if (ret)
		return ret;
	return nfs4_verify_back_channel_attrs(args, session);
4982 4983
}

A
Andy Adamson 已提交
4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001
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);
5002
	args.flags = (SESSION4_PERSIST | SESSION4_BACK_CHAN);
A
Andy Adamson 已提交
5003 5004 5005

	status = rpc_call_sync(session->clp->cl_rpcclient, &msg, 0);

5006 5007 5008
	if (!status)
		/* Verify the session's negotiated channel_attrs values */
		status = nfs4_verify_channel_attrs(&args, session);
A
Andy Adamson 已提交
5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021
	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.
 */
5022
int nfs4_proc_create_session(struct nfs_client *clp)
A
Andy Adamson 已提交
5023 5024 5025 5026
{
	int status;
	unsigned *ptr;
	struct nfs4_session *session = clp->cl_session;
5027 5028
	long timeout = 0;
	int err;
A
Andy Adamson 已提交
5029 5030 5031

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

5032 5033 5034 5035 5036 5037 5038 5039 5040
	do {
		status = _nfs4_proc_create_session(clp);
		if (status == -NFS4ERR_DELAY) {
			err = nfs4_delay(clp->cl_rpcclient, &timeout);
			if (err)
				status = err;
		}
	} while (status == -NFS4ERR_DELAY);

A
Andy Adamson 已提交
5041 5042 5043
	if (status)
		goto out;

5044 5045 5046 5047 5048 5049 5050
	/* Init and reset the fore channel */
	status = nfs4_init_slot_tables(session);
	dprintk("slot table initialization returned %d\n", status);
	if (status)
		goto out;
	status = nfs4_reset_slot_tables(session);
	dprintk("slot table reset returned %d\n", status);
A
Andy Adamson 已提交
5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061
	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
5062 5063 5064 5065 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
/*
 * 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;
	status = rpc_call_sync(session->clp->cl_rpcclient, &msg, 0);

	if (status)
		printk(KERN_WARNING
			"Got error %d from the server on DESTROY_SESSION. "
			"Session has been destroyed regardless...\n", status);

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

5092 5093 5094
int nfs4_init_session(struct nfs_server *server)
{
	struct nfs_client *clp = server->nfs_client;
5095
	struct nfs4_session *session;
5096
	unsigned int rsize, wsize;
5097 5098 5099 5100 5101
	int ret;

	if (!nfs4_has_session(clp))
		return 0;

5102 5103 5104 5105
	session = clp->cl_session;
	if (!test_and_clear_bit(NFS4_SESSION_INITING, &session->session_state))
		return 0;

5106 5107 5108 5109 5110 5111 5112 5113 5114
	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;
5115

5116 5117 5118 5119 5120 5121
	ret = nfs4_recover_expired_lease(server);
	if (!ret)
		ret = nfs4_check_client_ready(clp);
	return ret;
}

A
Andy Adamson 已提交
5122 5123 5124
/*
 * Renew the cl_session lease.
 */
5125 5126 5127 5128 5129 5130
struct nfs4_sequence_data {
	struct nfs_client *clp;
	struct nfs4_sequence_args args;
	struct nfs4_sequence_res res;
};

5131 5132
static void nfs41_sequence_release(void *data)
{
5133 5134
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
5135

5136 5137 5138
	if (atomic_read(&clp->cl_count) > 1)
		nfs4_schedule_state_renewal(clp);
	nfs_put_client(clp);
5139
	kfree(calldata);
5140 5141
}

5142 5143 5144 5145 5146 5147 5148
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:
5149
		nfs4_schedule_lease_recovery(clp);
5150 5151 5152 5153
	}
	return 0;
}

5154
static void nfs41_sequence_call_done(struct rpc_task *task, void *data)
A
Andy Adamson 已提交
5155
{
5156 5157
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
A
Andy Adamson 已提交
5158

5159 5160
	if (!nfs41_sequence_done(task, task->tk_msg.rpc_resp))
		return;
A
Andy Adamson 已提交
5161 5162 5163

	if (task->tk_status < 0) {
		dprintk("%s ERROR %d\n", __func__, task->tk_status);
5164 5165
		if (atomic_read(&clp->cl_count) == 1)
			goto out;
A
Andy Adamson 已提交
5166

5167 5168
		if (nfs41_sequence_handle_errors(task, clp) == -EAGAIN) {
			rpc_restart_call_prepare(task);
A
Andy Adamson 已提交
5169 5170 5171 5172
			return;
		}
	}
	dprintk("%s rpc_cred %p\n", __func__, task->tk_msg.rpc_cred);
5173
out:
A
Andy Adamson 已提交
5174 5175 5176 5177 5178
	dprintk("<-- %s\n", __func__);
}

static void nfs41_sequence_prepare(struct rpc_task *task, void *data)
{
5179 5180
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
A
Andy Adamson 已提交
5181 5182 5183 5184 5185 5186
	struct nfs4_sequence_args *args;
	struct nfs4_sequence_res *res;

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

5187
	if (nfs41_setup_sequence(clp->cl_session, args, res, 0, task))
A
Andy Adamson 已提交
5188 5189 5190 5191 5192 5193 5194
		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,
5195
	.rpc_release = nfs41_sequence_release,
A
Andy Adamson 已提交
5196 5197
};

5198
static struct rpc_task *_nfs41_proc_sequence(struct nfs_client *clp, struct rpc_cred *cred)
A
Andy Adamson 已提交
5199
{
5200
	struct nfs4_sequence_data *calldata;
A
Andy Adamson 已提交
5201 5202 5203 5204
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SEQUENCE],
		.rpc_cred = cred,
	};
5205 5206 5207 5208 5209 5210
	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
5211

5212
	if (!atomic_inc_not_zero(&clp->cl_count))
5213
		return ERR_PTR(-EIO);
5214
	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
5215
	if (calldata == NULL) {
5216
		nfs_put_client(clp);
5217
		return ERR_PTR(-ENOMEM);
A
Andy Adamson 已提交
5218
	}
5219 5220 5221
	msg.rpc_argp = &calldata->args;
	msg.rpc_resp = &calldata->res;
	calldata->clp = clp;
5222
	task_setup_data.callback_data = calldata;
A
Andy Adamson 已提交
5223

5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235
	return rpc_run_task(&task_setup_data);
}

static int nfs41_proc_async_sequence(struct nfs_client *clp, struct rpc_cred *cred)
{
	struct rpc_task *task;
	int ret = 0;

	task = _nfs41_proc_sequence(clp, cred);
	if (IS_ERR(task))
		ret = PTR_ERR(task);
	else
5236
		rpc_put_task_async(task);
5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251
	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);
5252 5253 5254 5255 5256
	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);
5257
		ret = task->tk_status;
5258
	}
5259 5260 5261 5262
	rpc_put_task(task);
out:
	dprintk("<-- %s status=%d\n", __func__, ret);
	return ret;
A
Andy Adamson 已提交
5263 5264
}

5265 5266 5267 5268 5269 5270 5271 5272 5273 5274
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;

5275
	rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
5276 5277
	if (nfs41_setup_sequence(calldata->clp->cl_session,
				&calldata->arg.seq_args,
5278 5279 5280 5281 5282 5283
				&calldata->res.seq_res, 0, task))
		return;

	rpc_call_start(task);
}

5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294
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);
		return -EAGAIN;
	default:
5295
		nfs4_schedule_lease_recovery(clp);
5296 5297 5298 5299
	}
	return 0;
}

5300 5301 5302 5303 5304 5305 5306
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__);
5307 5308
	if (!nfs41_sequence_done(task, res))
		return;
5309

5310 5311 5312 5313
	if (nfs41_reclaim_complete_handle_errors(task, clp) == -EAGAIN) {
		rpc_restart_call_prepare(task);
		return;
	}
5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348
	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__);
5349
	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
5350 5351 5352 5353 5354 5355 5356 5357 5358
	if (calldata == NULL)
		goto out;
	calldata->clp = clp;
	calldata->arg.one_fs = 0;

	msg.rpc_argp = &calldata->arg;
	msg.rpc_resp = &calldata->res;
	task_setup_data.callback_data = calldata;
	task = rpc_run_task(&task_setup_data);
5359
	if (IS_ERR(task)) {
5360
		status = PTR_ERR(task);
5361 5362
		goto out;
	}
5363 5364 5365
	status = nfs4_wait_for_completion_rpc_task(task);
	if (status == 0)
		status = task->tk_status;
5366
	rpc_put_task(task);
5367
	return 0;
5368 5369 5370 5371
out:
	dprintk("<-- %s status=%d\n", __func__, status);
	return status;
}
5372 5373 5374 5375 5376

static void
nfs4_layoutget_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;
F
Fred Isaman 已提交
5377
	struct nfs_server *server = NFS_SERVER(lgp->args.inode);
5378 5379

	dprintk("--> %s\n", __func__);
F
Fred Isaman 已提交
5380 5381 5382 5383 5384
	/* 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.
	 */
5385 5386 5387
	if (nfs4_setup_sequence(server, &lgp->args.seq_args,
				&lgp->res.seq_res, 0, task))
		return;
5388 5389 5390 5391 5392 5393
	if (pnfs_choose_layoutget_stateid(&lgp->args.stateid,
					  NFS_I(lgp->args.inode)->layout,
					  lgp->args.ctx->state)) {
		rpc_exit(task, NFS4_OK);
		return;
	}
5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453 5454 5455 5456 5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471
	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__);
	if (lgp->res.layout.buf != NULL)
		free_page((unsigned long) lgp->res.layout.buf);
	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__);

	lgp->res.layout.buf = (void *)__get_free_page(GFP_NOFS);
	if (lgp->res.layout.buf == NULL) {
		nfs4_layoutget_release(lgp);
		return -ENOMEM;
	}

	lgp->res.seq_res.sr_slot = NULL;
	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
5472 5473 5474 5475
	if (status == 0)
		status = task->tk_status;
	if (status == 0)
		status = pnfs_layout_process(lgp);
5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492 5493 5494 5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517
	rpc_put_task(task);
	dprintk("<-- %s status=%d\n", __func__, status);
	return status;
}

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__);
	status = nfs4_call_sync(server, &msg, &args, &res, 0);
	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);

5518 5519
#endif /* CONFIG_NFS_V4_1 */

5520
struct nfs4_state_recovery_ops nfs40_reboot_recovery_ops = {
5521
	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
5522
	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
L
Linus Torvalds 已提交
5523 5524
	.recover_open	= nfs4_open_reclaim,
	.recover_lock	= nfs4_lock_reclaim,
5525
	.establish_clid = nfs4_init_clientid,
5526
	.get_clid_cred	= nfs4_get_setclientid_cred,
L
Linus Torvalds 已提交
5527 5528
};

5529 5530 5531 5532 5533 5534
#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,
5535
	.establish_clid = nfs41_init_clientid,
5536
	.get_clid_cred	= nfs4_get_exchange_id_cred,
5537
	.reclaim_complete = nfs41_proc_reclaim_complete,
5538 5539 5540 5541 5542 5543 5544 5545 5546
};
#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,
5547
	.get_clid_cred	= nfs4_get_setclientid_cred,
5548 5549 5550 5551
};

#if defined(CONFIG_NFS_V4_1)
struct nfs4_state_recovery_ops nfs41_nograce_recovery_ops = {
5552
	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
5553
	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
L
Linus Torvalds 已提交
5554 5555
	.recover_open	= nfs4_open_expired,
	.recover_lock	= nfs4_lock_expired,
5556
	.establish_clid = nfs41_init_clientid,
5557
	.get_clid_cred	= nfs4_get_exchange_id_cred,
L
Linus Torvalds 已提交
5558
};
5559
#endif /* CONFIG_NFS_V4_1 */
L
Linus Torvalds 已提交
5560

B
Benny Halevy 已提交
5561 5562
struct nfs4_state_maintenance_ops nfs40_state_renewal_ops = {
	.sched_state_renewal = nfs4_proc_async_renew,
5563
	.get_state_renewal_cred_locked = nfs4_get_renew_cred_locked,
5564
	.renew_lease = nfs4_proc_renew,
B
Benny Halevy 已提交
5565 5566 5567 5568 5569
};

#if defined(CONFIG_NFS_V4_1)
struct nfs4_state_maintenance_ops nfs41_state_renewal_ops = {
	.sched_state_renewal = nfs41_proc_async_sequence,
5570
	.get_state_renewal_cred_locked = nfs4_get_machine_cred_locked,
5571
	.renew_lease = nfs4_proc_sequence,
B
Benny Halevy 已提交
5572 5573 5574
};
#endif

5575 5576 5577
static const struct nfs4_minor_version_ops nfs_v4_0_minor_ops = {
	.minor_version = 0,
	.call_sync = _nfs4_call_sync,
5578
	.validate_stateid = nfs4_validate_delegation_stateid,
5579 5580 5581
	.reboot_recovery_ops = &nfs40_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs40_nograce_recovery_ops,
	.state_renewal_ops = &nfs40_state_renewal_ops,
5582 5583 5584 5585 5586 5587
};

#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,
5588
	.validate_stateid = nfs41_validate_delegation_stateid,
5589 5590 5591
	.reboot_recovery_ops = &nfs41_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs41_nograce_recovery_ops,
	.state_renewal_ops = &nfs41_state_renewal_ops,
5592 5593 5594 5595 5596 5597 5598 5599 5600 5601
};
#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
};

5602
static const struct inode_operations nfs4_file_inode_operations = {
5603 5604 5605
	.permission	= nfs_permission,
	.getattr	= nfs_getattr,
	.setattr	= nfs_setattr,
5606 5607 5608 5609
	.getxattr	= generic_getxattr,
	.setxattr	= generic_setxattr,
	.listxattr	= generic_listxattr,
	.removexattr	= generic_removexattr,
5610 5611
};

D
David Howells 已提交
5612
const struct nfs_rpc_ops nfs_v4_clientops = {
L
Linus Torvalds 已提交
5613 5614 5615
	.version	= 4,			/* protocol version */
	.dentry_ops	= &nfs4_dentry_operations,
	.dir_inode_ops	= &nfs4_dir_inode_operations,
5616
	.file_inode_ops	= &nfs4_file_inode_operations,
L
Linus Torvalds 已提交
5617 5618 5619
	.getroot	= nfs4_proc_get_root,
	.getattr	= nfs4_proc_getattr,
	.setattr	= nfs4_proc_setattr,
D
David Howells 已提交
5620
	.lookupfh	= nfs4_proc_lookupfh,
L
Linus Torvalds 已提交
5621 5622 5623 5624 5625 5626 5627 5628
	.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,
5629 5630
	.rename_setup	= nfs4_proc_rename_setup,
	.rename_done	= nfs4_proc_rename_done,
L
Linus Torvalds 已提交
5631 5632 5633 5634 5635 5636 5637 5638 5639
	.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,
5640
	.set_capabilities = nfs4_server_capabilities,
L
Linus Torvalds 已提交
5641 5642
	.decode_dirent	= nfs4_decode_dirent,
	.read_setup	= nfs4_proc_read_setup,
T
Trond Myklebust 已提交
5643
	.read_done	= nfs4_read_done,
L
Linus Torvalds 已提交
5644
	.write_setup	= nfs4_proc_write_setup,
5645
	.write_done	= nfs4_write_done,
L
Linus Torvalds 已提交
5646
	.commit_setup	= nfs4_proc_commit_setup,
5647
	.commit_done	= nfs4_commit_done,
L
Linus Torvalds 已提交
5648
	.lock		= nfs4_proc_lock,
5649
	.clear_acl_cache = nfs4_zap_acl_attr,
T
Trond Myklebust 已提交
5650
	.close_context  = nfs4_close_context,
5651
	.open_context	= nfs4_atomic_open,
L
Linus Torvalds 已提交
5652 5653
};

5654 5655 5656 5657 5658 5659 5660 5661 5662 5663 5664 5665
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
};

L
Linus Torvalds 已提交
5666 5667 5668 5669 5670
/*
 * Local variables:
 *  c-basic-offset: 8
 * End:
 */