nfs4proc.c 153.0 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41
/*
 *  fs/nfs/nfs4proc.c
 *
 *  Client-side procedure declarations for NFSv4.
 *
 *  Copyright (c) 2002 The Regents of the University of Michigan.
 *  All rights reserved.
 *
 *  Kendrick Smith <kmsmith@umich.edu>
 *  Andy Adamson   <andros@umich.edu>
 *
 *  Redistribution and use in source and binary forms, with or without
 *  modification, are permitted provided that the following conditions
 *  are met:
 *
 *  1. Redistributions of source code must retain the above copyright
 *     notice, this list of conditions and the following disclaimer.
 *  2. Redistributions in binary form must reproduce the above copyright
 *     notice, this list of conditions and the following disclaimer in the
 *     documentation and/or other materials provided with the distribution.
 *  3. Neither the name of the University nor the names of its
 *     contributors may be used to endorse or promote products derived
 *     from this software without specific prior written permission.
 *
 *  THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
 *  WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
 *  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 *  DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 *  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
 *  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
 *  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
 *  BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
 *  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
 *  NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 *  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */

#include <linux/mm.h>
#include <linux/delay.h>
#include <linux/errno.h>
#include <linux/string.h>
42
#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>
L
Linus Torvalds 已提交
54

55
#include "nfs4_fs.h"
L
Linus Torvalds 已提交
56
#include "delegation.h"
57
#include "internal.h"
58
#include "iostat.h"
A
Andy Adamson 已提交
59
#include "callback.h"
60
#include "pnfs.h"
L
Linus Torvalds 已提交
61 62 63

#define NFSDBG_FACILITY		NFSDBG_PROC

64
#define NFS4_POLL_RETRY_MIN	(HZ/10)
L
Linus Torvalds 已提交
65 66
#define NFS4_POLL_RETRY_MAX	(15*HZ)

67 68
#define NFS4_MAX_LOOP_ON_RECOVER (10)

69
struct nfs4_opendata;
70
static int _nfs4_proc_open(struct nfs4_opendata *data);
71
static int _nfs4_recover_proc_open(struct nfs4_opendata *data);
L
Linus Torvalds 已提交
72
static int nfs4_do_fsinfo(struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *);
73
static int nfs4_async_handle_error(struct rpc_task *, const struct nfs_server *, struct nfs4_state *);
74 75
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);
76 77 78
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
79 80

/* Prevent leaks of NFSv4 errors into userland */
81
static int nfs4_map_errors(int err)
L
Linus Torvalds 已提交
82
{
83 84 85 86 87
	if (err >= -1000)
		return err;
	switch (err) {
	case -NFS4ERR_RESOURCE:
		return -EREMOTEIO;
88 89 90
	case -NFS4ERR_BADOWNER:
	case -NFS4ERR_BADNAME:
		return -EINVAL;
91
	default:
L
Linus Torvalds 已提交
92
		dprintk("%s could not handle NFSv4 error %d\n",
93
				__func__, -err);
94
		break;
L
Linus Torvalds 已提交
95
	}
96
	return -EIO;
L
Linus Torvalds 已提交
97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127
}

/*
 * 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
};

128
const u32 nfs4_pathconf_bitmap[2] = {
L
Linus Torvalds 已提交
129 130 131 132 133 134 135 136 137
	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,
138
			FATTR4_WORD1_TIME_DELTA
139
			| FATTR4_WORD1_FS_LAYOUT_TYPES
L
Linus Torvalds 已提交
140 141
};

142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160
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
161
static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dentry,
L
Linus Torvalds 已提交
162 163
		struct nfs4_readdir_arg *readdir)
{
164
	__be32 *start, *p;
L
Linus Torvalds 已提交
165 166 167

	BUG_ON(readdir->count < 80);
	if (cookie > 2) {
168
		readdir->cookie = cookie;
L
Linus Torvalds 已提交
169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184
		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.
	 */
185
	start = p = kmap_atomic(*readdir->pages, KM_USER0);
L
Linus Torvalds 已提交
186 187 188 189 190 191 192 193 194 195 196
	
	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
197
		p = xdr_encode_hyper(p, NFS_FILEID(dentry->d_inode));
L
Linus Torvalds 已提交
198 199 200 201 202 203 204 205 206 207 208
	}
	
	*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
209
	p = xdr_encode_hyper(p, NFS_FILEID(dentry->d_parent->d_inode));
L
Linus Torvalds 已提交
210 211 212 213 214 215

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

216 217 218 219 220 221
static int nfs4_wait_clnt_recover(struct nfs_client *clp)
{
	int res;

	might_sleep();

222
	res = wait_on_bit(&clp->cl_state, NFS4CLNT_MANAGER_RUNNING,
223
			nfs_wait_bit_killable, TASK_KILLABLE);
224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246
	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.
 */
247
static int nfs4_handle_exception(struct nfs_server *server, int errorcode, struct nfs4_exception *exception)
248 249
{
	struct nfs_client *clp = server->nfs_client;
250
	struct nfs4_state *state = exception->state;
251 252 253 254 255 256
	int ret = errorcode;

	exception->retry = 0;
	switch(errorcode) {
		case 0:
			return 0;
257 258 259 260 261
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OPENMODE:
			if (state == NULL)
				break;
262 263
			nfs4_schedule_stateid_recovery(server, state);
			goto wait_on_recovery;
264
		case -NFS4ERR_STALE_STATEID:
265
		case -NFS4ERR_STALE_CLIENTID:
266
		case -NFS4ERR_EXPIRED:
267 268
			nfs4_schedule_lease_recovery(clp);
			goto wait_on_recovery;
269
#if defined(CONFIG_NFS_V4_1)
270 271 272 273 274 275 276 277 278
		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);
279
			nfs4_schedule_session_recovery(clp->cl_session);
280
			exception->retry = 1;
281
			break;
282
#endif /* defined(CONFIG_NFS_V4_1) */
283
		case -NFS4ERR_FILE_OPEN:
284 285 286 287 288 289 290
			if (exception->timeout > HZ) {
				/* We have retried a decent amount, time to
				 * fail
				 */
				ret = -EBUSY;
				break;
			}
291 292
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
293
		case -EKEYEXPIRED:
294 295 296 297 298
			ret = nfs4_delay(server->client, &exception->timeout);
			if (ret != 0)
				break;
		case -NFS4ERR_OLD_STATEID:
			exception->retry = 1;
299 300 301 302 303 304 305 306 307 308 309 310 311
			break;
		case -NFS4ERR_BADOWNER:
			/* The following works around a Linux server bug! */
		case -NFS4ERR_BADNAME:
			if (server->caps & NFS_CAP_UIDGID_NOMAP) {
				server->caps &= ~NFS_CAP_UIDGID_NOMAP;
				exception->retry = 1;
				printk(KERN_WARNING "NFS: v4 server %s "
						"does not accept raw "
						"uid/gids. "
						"Reenabling the idmapper.\n",
						server->nfs_client->cl_hostname);
			}
312 313 314
	}
	/* We failed to handle the error */
	return nfs4_map_errors(ret);
315
wait_on_recovery:
316 317 318 319
	ret = nfs4_wait_clnt_recover(clp);
	if (ret == 0)
		exception->retry = 1;
	return ret;
320 321 322
}


323
static void do_renew_lease(struct nfs_client *clp, unsigned long timestamp)
L
Linus Torvalds 已提交
324 325 326 327 328 329 330
{
	spin_lock(&clp->cl_lock);
	if (time_before(clp->cl_last_renewal,timestamp))
		clp->cl_last_renewal = timestamp;
	spin_unlock(&clp->cl_lock);
}

331 332 333 334 335
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
336 337
#if defined(CONFIG_NFS_V4_1)

A
Andy Adamson 已提交
338 339 340 341 342 343 344 345 346 347 348 349
/*
 * 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.
350 351
 *
 * Must be called while holding tbl->slot_tbl_lock
A
Andy Adamson 已提交
352 353
 */
static void
354
nfs4_free_slot(struct nfs4_slot_table *tbl, struct nfs4_slot *free_slot)
A
Andy Adamson 已提交
355
{
356
	int free_slotid = free_slot - tbl->slots;
A
Andy Adamson 已提交
357 358
	int slotid = free_slotid;

359
	BUG_ON(slotid < 0 || slotid >= NFS4_MAX_SLOT_TABLE);
A
Andy Adamson 已提交
360 361 362 363 364 365
	/* 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
366
		if (slotid < tbl->max_slots)
A
Andy Adamson 已提交
367 368 369 370 371 372 373 374
			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);
}

375
/*
376
 * Signal state manager thread if session fore channel is drained
377
 */
378
static void nfs4_check_drain_fc_complete(struct nfs4_session *ses)
379
{
380 381
	struct rpc_task *task;

382
	if (!test_bit(NFS4_SESSION_DRAINING, &ses->session_state)) {
383 384 385
		task = rpc_wake_up_next(&ses->fc_slot_table.slot_tbl_waitq);
		if (task)
			rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
386 387 388 389 390 391
		return;
	}

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

392 393 394 395 396 397 398 399 400 401 402 403 404 405
	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);
406 407
}

408
static void nfs41_sequence_free_slot(struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
409 410 411
{
	struct nfs4_slot_table *tbl;

412
	tbl = &res->sr_session->fc_slot_table;
413
	if (!res->sr_slot) {
A
Andy Adamson 已提交
414 415
		/* just wake up the next guy waiting since
		 * we may have not consumed a slot after all */
A
Andy Adamson 已提交
416
		dprintk("%s: No slot\n", __func__);
417
		return;
A
Andy Adamson 已提交
418
	}
419

420
	spin_lock(&tbl->slot_tbl_lock);
421
	nfs4_free_slot(tbl, res->sr_slot);
422
	nfs4_check_drain_fc_complete(res->sr_session);
423
	spin_unlock(&tbl->slot_tbl_lock);
424
	res->sr_slot = NULL;
A
Andy Adamson 已提交
425 426
}

427
static int nfs41_sequence_done(struct rpc_task *task, struct nfs4_sequence_res *res)
A
Andy Adamson 已提交
428 429
{
	unsigned long timestamp;
430
	struct nfs_client *clp;
A
Andy Adamson 已提交
431 432 433 434 435 436 437 438 439 440 441

	/*
	 * 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 */
442
	if (!res->sr_slot)
A
Andy Adamson 已提交
443 444
		goto out;

A
Andy Adamson 已提交
445
	/* Check the SEQUENCE operation status */
446 447
	switch (res->sr_status) {
	case 0:
A
Andy Adamson 已提交
448
		/* Update the slot's sequence and clientid lease timer */
449
		++res->sr_slot->seq_nr;
A
Andy Adamson 已提交
450
		timestamp = res->sr_renewal_time;
451
		clp = res->sr_session->clp;
452
		do_renew_lease(clp, timestamp);
453
		/* Check sequence flags */
454 455
		if (res->sr_status_flags != 0)
			nfs4_schedule_lease_recovery(clp);
456 457 458 459 460 461
		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.
		 */
462
		dprintk("%s: slot=%td seq=%d: Operation in progress\n",
463 464 465
			__func__,
			res->sr_slot - res->sr_session->fc_slot_table.slots,
			res->sr_slot->seq_nr);
466 467 468
		goto out_retry;
	default:
		/* Just update the slot sequence no. */
469
		++res->sr_slot->seq_nr;
A
Andy Adamson 已提交
470 471 472 473
	}
out:
	/* The session may be reset by one of the error handlers. */
	dprintk("%s: Error %d free the slot \n", __func__, res->sr_status);
474
	nfs41_sequence_free_slot(res);
475 476
	return 1;
out_retry:
477
	if (!rpc_restart_call(task))
478 479 480
		goto out;
	rpc_delay(task, NFS4_POLL_RETRY_MAX);
	return 0;
A
Andy Adamson 已提交
481 482
}

483 484
static int nfs4_sequence_done(struct rpc_task *task,
			       struct nfs4_sequence_res *res)
485
{
486 487 488
	if (res->sr_session == NULL)
		return 1;
	return nfs41_sequence_done(task, res);
489 490
}

B
Benny Halevy 已提交
491 492 493 494 495 496 497
/*
 * 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
498 499
 *
 * Note: must be called with under the slot_tbl_lock.
B
Benny Halevy 已提交
500 501
 */
static u8
502
nfs4_find_slot(struct nfs4_slot_table *tbl)
B
Benny Halevy 已提交
503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523
{
	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
524
int nfs41_setup_sequence(struct nfs4_session *session,
A
Andy Adamson 已提交
525 526 527 528 529
				struct nfs4_sequence_args *args,
				struct nfs4_sequence_res *res,
				int cache_reply,
				struct rpc_task *task)
{
A
Andy Adamson 已提交
530 531 532 533 534
	struct nfs4_slot *slot;
	struct nfs4_slot_table *tbl;
	u8 slotid;

	dprintk("--> %s\n", __func__);
A
Andy Adamson 已提交
535
	/* slot already allocated? */
536
	if (res->sr_slot != NULL)
A
Andy Adamson 已提交
537 538
		return 0;

A
Andy Adamson 已提交
539 540 541
	tbl = &session->fc_slot_table;

	spin_lock(&tbl->slot_tbl_lock);
542
	if (test_bit(NFS4_SESSION_DRAINING, &session->session_state) &&
543
	    !rpc_task_has_priority(task, RPC_PRIORITY_PRIVILEGED)) {
544 545 546 547
		/*
		 * The state manager will wait until the slot table is empty.
		 * Schedule the reset thread
		 */
A
Andy Adamson 已提交
548
		rpc_sleep_on(&tbl->slot_tbl_waitq, task, NULL);
549
		spin_unlock(&tbl->slot_tbl_lock);
T
Trond Myklebust 已提交
550
		dprintk("%s Schedule Session Reset\n", __func__);
A
Andy Adamson 已提交
551
		return -EAGAIN;
552 553
	}

554 555 556 557 558 559 560 561
	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;
	}

562
	slotid = nfs4_find_slot(tbl);
A
Andy Adamson 已提交
563 564 565 566 567 568 569 570
	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);

571
	rpc_task_set_priority(task, RPC_PRIORITY_NORMAL);
A
Andy Adamson 已提交
572
	slot = tbl->slots + slotid;
B
Benny Halevy 已提交
573
	args->sa_session = session;
A
Andy Adamson 已提交
574 575 576 577 578
	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
579
	res->sr_session = session;
580
	res->sr_slot = slot;
A
Andy Adamson 已提交
581
	res->sr_renewal_time = jiffies;
582
	res->sr_status_flags = 0;
A
Andy Adamson 已提交
583 584 585 586 587
	/*
	 * 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
588 589
	return 0;
}
A
Andy Adamson 已提交
590
EXPORT_SYMBOL_GPL(nfs41_setup_sequence);
A
Andy Adamson 已提交
591

592
int nfs4_setup_sequence(const struct nfs_server *server,
A
Andy Adamson 已提交
593 594 595 596 597
			struct nfs4_sequence_args *args,
			struct nfs4_sequence_res *res,
			int cache_reply,
			struct rpc_task *task)
{
598
	struct nfs4_session *session = nfs4_get_session(server);
A
Andy Adamson 已提交
599 600
	int ret = 0;

601 602 603 604 605 606
	if (session == NULL) {
		args->sa_session = NULL;
		res->sr_session = NULL;
		goto out;
	}

607
	dprintk("--> %s clp %p session %p sr_slot %td\n",
608 609
		__func__, session->clp, session, res->sr_slot ?
			res->sr_slot - session->fc_slot_table.slots : -1);
A
Andy Adamson 已提交
610

611
	ret = nfs41_setup_sequence(session, args, res, cache_reply,
A
Andy Adamson 已提交
612 613 614 615 616 617 618
				   task);
out:
	dprintk("<-- %s status=%d\n", __func__, ret);
	return ret;
}

struct nfs41_call_sync_data {
619
	const struct nfs_server *seq_server;
A
Andy Adamson 已提交
620 621 622 623 624 625 626 627 628
	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;

629 630 631
	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
632 633 634 635 636
				data->seq_res, data->cache_reply, task))
		return;
	rpc_call_start(task);
}

637 638 639 640 641 642
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
643 644 645 646
static void nfs41_call_sync_done(struct rpc_task *task, void *calldata)
{
	struct nfs41_call_sync_data *data = calldata;

647
	nfs41_sequence_done(task, data->seq_res);
A
Andy Adamson 已提交
648 649
}

A
Andy Adamson 已提交
650 651
struct rpc_call_ops nfs41_call_sync_ops = {
	.rpc_call_prepare = nfs41_call_sync_prepare,
A
Andy Adamson 已提交
652
	.rpc_call_done = nfs41_call_sync_done,
A
Andy Adamson 已提交
653 654
};

655 656 657 658 659
struct rpc_call_ops nfs41_call_priv_sync_ops = {
	.rpc_call_prepare = nfs41_call_priv_sync_prepare,
	.rpc_call_done = nfs41_call_sync_done,
};

660
static int nfs4_call_sync_sequence(struct nfs_server *server,
A
Andy Adamson 已提交
661 662 663
				   struct rpc_message *msg,
				   struct nfs4_sequence_args *args,
				   struct nfs4_sequence_res *res,
664 665
				   int cache_reply,
				   int privileged)
A
Andy Adamson 已提交
666 667 668 669
{
	int ret;
	struct rpc_task *task;
	struct nfs41_call_sync_data data = {
670
		.seq_server = server,
A
Andy Adamson 已提交
671 672 673 674 675
		.seq_args = args,
		.seq_res = res,
		.cache_reply = cache_reply,
	};
	struct rpc_task_setup task_setup = {
676
		.rpc_client = server->client,
A
Andy Adamson 已提交
677 678 679 680 681
		.rpc_message = msg,
		.callback_ops = &nfs41_call_sync_ops,
		.callback_data = &data
	};

682
	res->sr_slot = NULL;
683 684
	if (privileged)
		task_setup.callback_ops = &nfs41_call_priv_sync_ops;
A
Andy Adamson 已提交
685 686 687 688 689 690 691 692 693 694
	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
695 696 697 698 699 700
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)
{
701
	return nfs4_call_sync_sequence(server, msg, args, res, cache_reply, 0);
A
Andy Adamson 已提交
702 703
}

704
#else
705 706
static int nfs4_sequence_done(struct rpc_task *task,
			       struct nfs4_sequence_res *res)
707
{
708
	return 1;
709
}
A
Andy Adamson 已提交
710 711 712 713 714 715 716 717
#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)
{
718
	args->sa_session = res->sr_session = NULL;
A
Andy Adamson 已提交
719 720 721 722
	return rpc_call_sync(server->client, msg, 0);
}

#define nfs4_call_sync(server, msg, args, res, cache_reply) \
723
	(server)->nfs_client->cl_mvops->call_sync((server), (msg), &(args)->seq_args, \
A
Andy Adamson 已提交
724 725
			&(res)->seq_res, (cache_reply))

726
static void update_changeattr(struct inode *dir, struct nfs4_change_info *cinfo)
L
Linus Torvalds 已提交
727
{
728
	struct nfs_inode *nfsi = NFS_I(dir);
L
Linus Torvalds 已提交
729

730
	spin_lock(&dir->i_lock);
731 732
	nfsi->cache_validity |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE|NFS_INO_INVALID_DATA;
	if (!cinfo->atomic || cinfo->before != nfsi->change_attr)
733
		nfs_force_lookup_revalidate(dir);
734
	nfsi->change_attr = cinfo->after;
735
	spin_unlock(&dir->i_lock);
L
Linus Torvalds 已提交
736 737
}

738
struct nfs4_opendata {
739
	struct kref kref;
740 741
	struct nfs_openargs o_arg;
	struct nfs_openres o_res;
742 743
	struct nfs_open_confirmargs c_arg;
	struct nfs_open_confirmres c_res;
744 745
	struct nfs_fattr f_attr;
	struct nfs_fattr dir_attr;
746
	struct path path;
747 748
	struct dentry *dir;
	struct nfs4_state_owner *owner;
749
	struct nfs4_state *state;
750
	struct iattr attrs;
751
	unsigned long timestamp;
752
	unsigned int rpc_done : 1;
753 754
	int rpc_status;
	int cancelled;
755 756
};

757 758 759 760 761

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;
762 763
	p->o_res.seqid = p->o_arg.seqid;
	p->c_res.seqid = p->c_arg.seqid;
764 765 766 767 768
	p->o_res.server = p->o_arg.server;
	nfs_fattr_init(&p->f_attr);
	nfs_fattr_init(&p->dir_attr);
}

769
static struct nfs4_opendata *nfs4_opendata_alloc(struct path *path,
770
		struct nfs4_state_owner *sp, fmode_t fmode, int flags,
771 772
		const struct iattr *attrs,
		gfp_t gfp_mask)
773
{
774
	struct dentry *parent = dget_parent(path->dentry);
775 776 777 778
	struct inode *dir = parent->d_inode;
	struct nfs_server *server = NFS_SERVER(dir);
	struct nfs4_opendata *p;

779
	p = kzalloc(sizeof(*p), gfp_mask);
780 781
	if (p == NULL)
		goto err;
782
	p->o_arg.seqid = nfs_alloc_seqid(&sp->so_seqid, gfp_mask);
783 784
	if (p->o_arg.seqid == NULL)
		goto err_free;
785 786
	path_get(path);
	p->path = *path;
787 788 789 790
	p->dir = parent;
	p->owner = sp;
	atomic_inc(&sp->so_count);
	p->o_arg.fh = NFS_FH(dir);
791 792
	p->o_arg.open_flags = flags;
	p->o_arg.fmode = fmode & (FMODE_READ|FMODE_WRITE);
793
	p->o_arg.clientid = server->nfs_client->cl_clientid;
794
	p->o_arg.id = sp->so_owner_id.id;
795
	p->o_arg.name = &p->path.dentry->d_name;
796 797 798
	p->o_arg.server = server;
	p->o_arg.bitmask = server->attr_bitmask;
	p->o_arg.claim = NFS4_OPEN_CLAIM_NULL;
799 800 801
	if (flags & O_CREAT) {
		u32 *s;

802 803
		p->o_arg.u.attrs = &p->attrs;
		memcpy(&p->attrs, attrs, sizeof(p->attrs));
804 805 806
		s = (u32 *) p->o_arg.u.verifier.data;
		s[0] = jiffies;
		s[1] = current->pid;
807
	}
808 809 810
	p->c_arg.fh = &p->o_res.fh;
	p->c_arg.stateid = &p->o_res.stateid;
	p->c_arg.seqid = p->o_arg.seqid;
811
	nfs4_init_opendata_res(p);
812
	kref_init(&p->kref);
813 814 815 816 817 818 819 820
	return p;
err_free:
	kfree(p);
err:
	dput(parent);
	return NULL;
}

821
static void nfs4_opendata_free(struct kref *kref)
822
{
823 824 825 826
	struct nfs4_opendata *p = container_of(kref,
			struct nfs4_opendata, kref);

	nfs_free_seqid(p->o_arg.seqid);
827 828
	if (p->state != NULL)
		nfs4_put_open_state(p->state);
829 830
	nfs4_put_state_owner(p->owner);
	dput(p->dir);
J
Jan Blunck 已提交
831
	path_put(&p->path);
832 833 834 835 836 837 838
	kfree(p);
}

static void nfs4_opendata_put(struct nfs4_opendata *p)
{
	if (p != NULL)
		kref_put(&p->kref, nfs4_opendata_free);
839 840
}

841 842 843 844 845 846 847 848
static int nfs4_wait_for_completion_rpc_task(struct rpc_task *task)
{
	int ret;

	ret = rpc_wait_for_completion_task(task);
	return ret;
}

849
static int can_open_cached(struct nfs4_state *state, fmode_t mode, int open_mode)
850 851
{
	int ret = 0;
852 853 854 855

	if (open_mode & O_EXCL)
		goto out;
	switch (mode & (FMODE_READ|FMODE_WRITE)) {
856
		case FMODE_READ:
857 858
			ret |= test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0
				&& state->n_rdonly != 0;
859 860
			break;
		case FMODE_WRITE:
861 862
			ret |= test_bit(NFS_O_WRONLY_STATE, &state->flags) != 0
				&& state->n_wronly != 0;
863 864
			break;
		case FMODE_READ|FMODE_WRITE:
865 866
			ret |= test_bit(NFS_O_RDWR_STATE, &state->flags) != 0
				&& state->n_rdwr != 0;
867
	}
868
out:
869 870 871
	return ret;
}

872
static int can_open_delegated(struct nfs_delegation *delegation, fmode_t fmode)
873
{
874
	if ((delegation->type & fmode) != fmode)
875
		return 0;
876
	if (test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags))
877
		return 0;
878
	nfs_mark_delegation_referenced(delegation);
879 880 881
	return 1;
}

882
static void update_open_stateflags(struct nfs4_state *state, fmode_t fmode)
883
{
884
	switch (fmode) {
885 886 887 888 889 890 891 892 893
		case FMODE_WRITE:
			state->n_wronly++;
			break;
		case FMODE_READ:
			state->n_rdonly++;
			break;
		case FMODE_READ|FMODE_WRITE:
			state->n_rdwr++;
	}
894
	nfs4_state_set_mode_locked(state, state->state | fmode);
895 896
}

897
static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
898 899 900 901
{
	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));
902
	switch (fmode) {
903 904 905 906 907 908 909 910 911
		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);
	}
912 913
}

914
static void nfs_set_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
915
{
916
	write_seqlock(&state->seqlock);
917
	nfs_set_open_stateid_locked(state, stateid, fmode);
918
	write_sequnlock(&state->seqlock);
919 920
}

921
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
922
{
923 924 925 926 927
	/*
	 * Protect the call to nfs4_state_set_mode_locked and
	 * serialise the stateid update
	 */
	write_seqlock(&state->seqlock);
928 929 930 931 932
	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)
933
		nfs_set_open_stateid_locked(state, open_stateid, fmode);
934 935
	write_sequnlock(&state->seqlock);
	spin_lock(&state->owner->so_lock);
936
	update_open_stateflags(state, fmode);
937
	spin_unlock(&state->owner->so_lock);
L
Linus Torvalds 已提交
938 939
}

940
static int update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *delegation, fmode_t fmode)
941 942 943 944 945
{
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *deleg_cur;
	int ret = 0;

946
	fmode &= (FMODE_READ|FMODE_WRITE);
947 948 949 950 951 952 953 954

	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 ||
955
	    (deleg_cur->type & fmode) != fmode)
956 957 958 959 960 961 962
		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;

963
	nfs_mark_delegation_referenced(deleg_cur);
964
	__update_open_stateid(state, open_stateid, &deleg_cur->stateid, fmode);
965 966 967 968 969 970 971
	ret = 1;
no_delegation_unlock:
	spin_unlock(&deleg_cur->lock);
no_delegation:
	rcu_read_unlock();

	if (!ret && open_stateid != NULL) {
972
		__update_open_stateid(state, open_stateid, NULL, fmode);
973 974 975 976 977 978 979
		ret = 1;
	}

	return ret;
}


980
static void nfs4_return_incompatible_delegation(struct inode *inode, fmode_t fmode)
981 982 983 984 985
{
	struct nfs_delegation *delegation;

	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(inode)->delegation);
986
	if (delegation == NULL || (delegation->type & fmode) == fmode) {
987 988 989 990 991 992 993
		rcu_read_unlock();
		return;
	}
	rcu_read_unlock();
	nfs_inode_return_delegation(inode);
}

994
static struct nfs4_state *nfs4_try_open_cached(struct nfs4_opendata *opendata)
995 996 997 998
{
	struct nfs4_state *state = opendata->state;
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *delegation;
999 1000
	int open_mode = opendata->o_arg.open_flags & O_EXCL;
	fmode_t fmode = opendata->o_arg.fmode;
1001 1002 1003 1004
	nfs4_stateid stateid;
	int ret = -EAGAIN;

	for (;;) {
1005
		if (can_open_cached(state, fmode, open_mode)) {
1006
			spin_lock(&state->owner->so_lock);
1007 1008
			if (can_open_cached(state, fmode, open_mode)) {
				update_open_stateflags(state, fmode);
1009 1010 1011 1012 1013
				spin_unlock(&state->owner->so_lock);
				goto out_return_state;
			}
			spin_unlock(&state->owner->so_lock);
		}
1014 1015 1016
		rcu_read_lock();
		delegation = rcu_dereference(nfsi->delegation);
		if (delegation == NULL ||
1017
		    !can_open_delegated(delegation, fmode)) {
1018
			rcu_read_unlock();
1019
			break;
1020
		}
1021 1022 1023
		/* Save the delegation */
		memcpy(stateid.data, delegation->stateid.data, sizeof(stateid.data));
		rcu_read_unlock();
1024
		ret = nfs_may_open(state->inode, state->owner->so_cred, open_mode);
1025 1026 1027
		if (ret != 0)
			goto out;
		ret = -EAGAIN;
1028 1029

		/* Try to update the stateid using the delegation */
1030
		if (update_open_stateid(state, NULL, &stateid, fmode))
1031
			goto out_return_state;
1032 1033 1034 1035 1036 1037 1038 1039
	}
out:
	return ERR_PTR(ret);
out_return_state:
	atomic_inc(&state->count);
	return state;
}

1040 1041 1042 1043
static struct nfs4_state *nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data)
{
	struct inode *inode;
	struct nfs4_state *state = NULL;
1044
	struct nfs_delegation *delegation;
1045
	int ret;
1046

1047
	if (!data->rpc_done) {
1048
		state = nfs4_try_open_cached(data);
1049 1050 1051
		goto out;
	}

1052
	ret = -EAGAIN;
1053
	if (!(data->f_attr.valid & NFS_ATTR_FATTR))
1054
		goto err;
1055
	inode = nfs_fhget(data->dir->d_sb, &data->o_res.fh, &data->f_attr);
1056
	ret = PTR_ERR(inode);
1057
	if (IS_ERR(inode))
1058 1059
		goto err;
	ret = -ENOMEM;
1060 1061
	state = nfs4_get_open_state(inode, data->owner);
	if (state == NULL)
1062
		goto err_put_inode;
1063 1064 1065
	if (data->o_res.delegation_type != 0) {
		int delegation_flags = 0;

1066 1067 1068 1069 1070
		rcu_read_lock();
		delegation = rcu_dereference(NFS_I(inode)->delegation);
		if (delegation)
			delegation_flags = delegation->flags;
		rcu_read_unlock();
1071
		if ((delegation_flags & 1UL<<NFS_DELEGATION_NEED_RECLAIM) == 0)
1072 1073 1074 1075 1076 1077 1078 1079
			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);
	}
1080 1081

	update_open_stateid(state, &data->o_res.stateid, NULL,
1082
			data->o_arg.fmode);
1083
	iput(inode);
1084
out:
1085
	return state;
1086 1087 1088 1089
err_put_inode:
	iput(inode);
err:
	return ERR_PTR(ret);
1090 1091
}

1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108
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
1109 1110 1111 1112
static struct nfs4_opendata *nfs4_open_recoverdata_alloc(struct nfs_open_context *ctx, struct nfs4_state *state)
{
	struct nfs4_opendata *opendata;

1113
	opendata = nfs4_opendata_alloc(&ctx->path, state->owner, 0, 0, NULL, GFP_NOFS);
T
Trond Myklebust 已提交
1114 1115 1116 1117 1118 1119 1120
	if (opendata == NULL)
		return ERR_PTR(-ENOMEM);
	opendata->state = state;
	atomic_inc(&state->count);
	return opendata;
}

1121
static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, fmode_t fmode, struct nfs4_state **res)
1122
{
1123
	struct nfs4_state *newstate;
1124 1125
	int ret;

1126 1127
	opendata->o_arg.open_flags = 0;
	opendata->o_arg.fmode = fmode;
1128 1129 1130
	memset(&opendata->o_res, 0, sizeof(opendata->o_res));
	memset(&opendata->c_res, 0, sizeof(opendata->c_res));
	nfs4_init_opendata_res(opendata);
1131
	ret = _nfs4_recover_proc_open(opendata);
1132 1133
	if (ret != 0)
		return ret; 
1134
	newstate = nfs4_opendata_to_nfs4_state(opendata);
1135 1136
	if (IS_ERR(newstate))
		return PTR_ERR(newstate);
1137
	nfs4_close_state(&opendata->path, newstate, fmode);
1138
	*res = newstate;
1139 1140 1141 1142 1143 1144 1145 1146 1147
	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_* */
1148
	clear_bit(NFS_DELEGATED_STATE, &state->flags);
1149 1150
	smp_rmb();
	if (state->n_rdwr != 0) {
T
Trond Myklebust 已提交
1151
		clear_bit(NFS_O_RDWR_STATE, &state->flags);
1152
		ret = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE, &newstate);
1153 1154
		if (ret != 0)
			return ret;
1155 1156
		if (newstate != state)
			return -ESTALE;
1157 1158
	}
	if (state->n_wronly != 0) {
T
Trond Myklebust 已提交
1159
		clear_bit(NFS_O_WRONLY_STATE, &state->flags);
1160
		ret = nfs4_open_recover_helper(opendata, FMODE_WRITE, &newstate);
1161 1162
		if (ret != 0)
			return ret;
1163 1164
		if (newstate != state)
			return -ESTALE;
1165 1166
	}
	if (state->n_rdonly != 0) {
T
Trond Myklebust 已提交
1167
		clear_bit(NFS_O_RDONLY_STATE, &state->flags);
1168
		ret = nfs4_open_recover_helper(opendata, FMODE_READ, &newstate);
1169 1170
		if (ret != 0)
			return ret;
1171 1172
		if (newstate != state)
			return -ESTALE;
1173
	}
1174 1175 1176 1177 1178 1179
	/*
	 * 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) {
1180
		write_seqlock(&state->seqlock);
1181 1182
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
			memcpy(state->stateid.data, state->open_stateid.data, sizeof(state->stateid.data));
1183
		write_sequnlock(&state->seqlock);
1184
	}
1185 1186 1187
	return 0;
}

L
Linus Torvalds 已提交
1188 1189 1190 1191
/*
 * OPEN_RECLAIM:
 * 	reclaim state on the server after a reboot.
 */
1192
static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1193
{
1194
	struct nfs_delegation *delegation;
1195
	struct nfs4_opendata *opendata;
1196
	fmode_t delegation_type = 0;
L
Linus Torvalds 已提交
1197 1198
	int status;

T
Trond Myklebust 已提交
1199 1200 1201
	opendata = nfs4_open_recoverdata_alloc(ctx, state);
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1202 1203
	opendata->o_arg.claim = NFS4_OPEN_CLAIM_PREVIOUS;
	opendata->o_arg.fh = NFS_FH(state->inode);
1204 1205
	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
1206
	if (delegation != NULL && test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) != 0)
1207
		delegation_type = delegation->type;
1208
	rcu_read_unlock();
1209 1210
	opendata->o_arg.u.delegation_type = delegation_type;
	status = nfs4_open_recover(opendata, state);
1211
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
1212 1213 1214
	return status;
}

1215
static int nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1216 1217 1218 1219 1220
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;
	do {
1221
		err = _nfs4_do_open_reclaim(ctx, state);
1222
		if (err != -NFS4ERR_DELAY)
1223 1224
			break;
		nfs4_handle_exception(server, err, &exception);
L
Linus Torvalds 已提交
1225 1226 1227 1228
	} while (exception.retry);
	return err;
}

1229 1230 1231 1232 1233 1234 1235 1236
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);
1237
	ret = nfs4_do_open_reclaim(ctx, state);
1238 1239 1240 1241
	put_nfs_open_context(ctx);
	return ret;
}

1242
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
1243
{
1244
	struct nfs4_opendata *opendata;
1245
	int ret;
L
Linus Torvalds 已提交
1246

T
Trond Myklebust 已提交
1247 1248 1249
	opendata = nfs4_open_recoverdata_alloc(ctx, state);
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1250
	opendata->o_arg.claim = NFS4_OPEN_CLAIM_DELEGATE_CUR;
1251
	memcpy(opendata->o_arg.u.delegation.data, stateid->data,
1252
			sizeof(opendata->o_arg.u.delegation.data));
1253
	ret = nfs4_open_recover(opendata, state);
1254
	nfs4_opendata_put(opendata);
1255
	return ret;
L
Linus Torvalds 已提交
1256 1257
}

1258
int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid)
L
Linus Torvalds 已提交
1259 1260
{
	struct nfs4_exception exception = { };
1261
	struct nfs_server *server = NFS_SERVER(state->inode);
L
Linus Torvalds 已提交
1262 1263
	int err;
	do {
1264
		err = _nfs4_open_delegation_recall(ctx, state, stateid);
L
Linus Torvalds 已提交
1265 1266
		switch (err) {
			case 0:
1267 1268 1269
			case -ENOENT:
			case -ESTALE:
				goto out;
1270 1271 1272 1273 1274
			case -NFS4ERR_BADSESSION:
			case -NFS4ERR_BADSLOT:
			case -NFS4ERR_BAD_HIGH_SLOT:
			case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
			case -NFS4ERR_DEADSESSION:
1275
				nfs4_schedule_session_recovery(server->nfs_client->cl_session);
1276
				goto out;
L
Linus Torvalds 已提交
1277 1278 1279 1280
			case -NFS4ERR_STALE_CLIENTID:
			case -NFS4ERR_STALE_STATEID:
			case -NFS4ERR_EXPIRED:
				/* Don't recall a delegation if it was lost */
1281
				nfs4_schedule_lease_recovery(server->nfs_client);
1282 1283 1284 1285 1286 1287 1288 1289
				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:
1290
				nfs4_schedule_stateid_recovery(server, state);
1291 1292 1293 1294 1295 1296 1297
			case -EKEYEXPIRED:
				/*
				 * User RPCSEC_GSS context has expired.
				 * We cannot recover this stateid now, so
				 * skip it and allow recovery thread to
				 * proceed.
				 */
1298 1299 1300
			case -ENOMEM:
				err = 0;
				goto out;
L
Linus Torvalds 已提交
1301 1302 1303
		}
		err = nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
1304
out:
L
Linus Torvalds 已提交
1305 1306 1307
	return err;
}

1308 1309 1310 1311 1312
static void nfs4_open_confirm_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;

	data->rpc_status = task->tk_status;
1313
	if (data->rpc_status == 0) {
1314 1315
		memcpy(data->o_res.stateid.data, data->c_res.stateid.data,
				sizeof(data->o_res.stateid.data));
1316
		nfs_confirm_seqid(&data->owner->so_seqid, 0);
1317
		renew_lease(data->o_res.server, data->timestamp);
1318
		data->rpc_done = 1;
1319
	}
1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330
}

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! */
1331
	if (!data->rpc_done)
1332 1333
		goto out_free;
	state = nfs4_opendata_to_nfs4_state(data);
1334
	if (!IS_ERR(state))
1335
		nfs4_close_state(&data->path, state, data->o_arg.fmode);
1336
out_free:
1337
	nfs4_opendata_put(data);
1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351
}

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;
1352 1353 1354 1355 1356 1357
	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
1358 1359
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1360
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1361 1362
		.callback_ops = &nfs4_open_confirm_ops,
		.callback_data = data,
1363
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1364 1365
		.flags = RPC_TASK_ASYNC,
	};
L
Linus Torvalds 已提交
1366 1367
	int status;

1368
	kref_get(&data->kref);
1369 1370
	data->rpc_done = 0;
	data->rpc_status = 0;
1371
	data->timestamp = jiffies;
T
Trond Myklebust 已提交
1372
	task = rpc_run_task(&task_setup_data);
1373
	if (IS_ERR(task))
1374 1375 1376 1377 1378 1379 1380
		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;
1381
	rpc_put_task(task);
L
Linus Torvalds 已提交
1382 1383 1384
	return status;
}

1385
static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
L
Linus Torvalds 已提交
1386
{
1387 1388
	struct nfs4_opendata *data = calldata;
	struct nfs4_state_owner *sp = data->owner;
1389

1390 1391
	if (nfs_wait_on_sequence(data->o_arg.seqid, task) != 0)
		return;
1392 1393 1394 1395 1396 1397 1398
	/*
	 * 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;

1399
		if (can_open_cached(data->state, data->o_arg.fmode, data->o_arg.open_flags))
1400
			goto out_no_action;
1401 1402 1403
		rcu_read_lock();
		delegation = rcu_dereference(NFS_I(data->state->inode)->delegation);
		if (delegation != NULL &&
1404
		    test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) == 0) {
1405
			rcu_read_unlock();
1406
			goto out_no_action;
1407 1408 1409
		}
		rcu_read_unlock();
	}
1410
	/* Update sequence id. */
1411
	data->o_arg.id = sp->so_owner_id.id;
1412
	data->o_arg.clientid = sp->so_server->nfs_client->cl_clientid;
T
Trond Myklebust 已提交
1413
	if (data->o_arg.claim == NFS4_OPEN_CLAIM_PREVIOUS) {
1414
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR];
T
Trond Myklebust 已提交
1415 1416
		nfs_copy_fh(&data->o_res.fh, data->o_arg.fh);
	}
1417
	data->timestamp = jiffies;
1418
	if (nfs4_setup_sequence(data->o_arg.server,
1419 1420 1421
				&data->o_arg.seq_args,
				&data->o_res.seq_res, 1, task))
		return;
1422
	rpc_call_start(task);
1423 1424 1425 1426
	return;
out_no_action:
	task->tk_action = NULL;

1427
}
L
Linus Torvalds 已提交
1428

1429 1430 1431 1432 1433 1434
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);
}

1435 1436 1437
static void nfs4_open_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;
L
Linus Torvalds 已提交
1438

1439
	data->rpc_status = task->tk_status;
1440

1441 1442
	if (!nfs4_sequence_done(task, &data->o_res.seq_res))
		return;
1443

1444 1445
	if (task->tk_status == 0) {
		switch (data->o_res.f_attr->mode & S_IFMT) {
1446 1447 1448
			case S_IFREG:
				break;
			case S_IFLNK:
1449
				data->rpc_status = -ELOOP;
1450 1451
				break;
			case S_IFDIR:
1452
				data->rpc_status = -EISDIR;
1453 1454
				break;
			default:
1455
				data->rpc_status = -ENOTDIR;
1456
		}
1457
		renew_lease(data->o_res.server, data->timestamp);
1458 1459
		if (!(data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM))
			nfs_confirm_seqid(&data->owner->so_seqid, 0);
1460
	}
1461
	data->rpc_done = 1;
1462
}
1463

1464 1465 1466 1467 1468 1469 1470 1471 1472
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! */
1473
	if (data->rpc_status != 0 || !data->rpc_done)
1474 1475 1476 1477 1478
		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);
1479
	if (!IS_ERR(state))
1480
		nfs4_close_state(&data->path, state, data->o_arg.fmode);
1481
out_free:
1482
	nfs4_opendata_put(data);
1483 1484 1485 1486 1487 1488 1489 1490
}

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,
};

1491 1492 1493 1494 1495 1496 1497
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)
1498 1499 1500 1501 1502 1503
{
	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;
1504 1505 1506 1507 1508 1509
	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
1510 1511
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1512
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1513 1514
		.callback_ops = &nfs4_open_ops,
		.callback_data = data,
1515
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1516 1517
		.flags = RPC_TASK_ASYNC,
	};
1518 1519
	int status;

1520
	kref_get(&data->kref);
1521 1522
	data->rpc_done = 0;
	data->rpc_status = 0;
1523
	data->cancelled = 0;
1524 1525
	if (isrecover)
		task_setup_data.callback_ops = &nfs4_recover_open_ops;
T
Trond Myklebust 已提交
1526
	task = rpc_run_task(&task_setup_data);
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 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572
        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);
1573
	if (status != 0 || !data->rpc_done)
1574 1575
		return status;

1576 1577 1578 1579 1580
	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
1581 1582
	if ((o_res->rflags & NFS4_OPEN_RESULT_LOCKTYPE_POSIX) == 0)
		server->caps &= ~NFS_CAP_POSIX_LOCK;
L
Linus Torvalds 已提交
1583
	if(o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
1584
		status = _nfs4_proc_open_confirm(data);
L
Linus Torvalds 已提交
1585
		if (status != 0)
1586
			return status;
L
Linus Torvalds 已提交
1587 1588
	}
	if (!(o_res->f_attr->valid & NFS_ATTR_FATTR))
1589
		_nfs4_proc_getattr(server, &o_res->fh, o_res->f_attr);
1590
	return 0;
L
Linus Torvalds 已提交
1591 1592
}

A
Andy Adamson 已提交
1593
static int nfs4_client_recover_expired_lease(struct nfs_client *clp)
1594
{
1595
	unsigned int loop;
1596
	int ret;
1597

1598
	for (loop = NFS4_MAX_LOOP_ON_RECOVER; loop != 0; loop--) {
1599
		ret = nfs4_wait_clnt_recover(clp);
1600
		if (ret != 0)
1601
			break;
1602 1603
		if (!test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state) &&
		    !test_bit(NFS4CLNT_CHECK_LEASE,&clp->cl_state))
1604
			break;
1605
		nfs4_schedule_state_manager(clp);
1606
		ret = -EIO;
1607
	}
1608
	return ret;
1609 1610
}

A
Andy Adamson 已提交
1611 1612 1613 1614 1615
static int nfs4_recover_expired_lease(struct nfs_server *server)
{
	return nfs4_client_recover_expired_lease(server->nfs_client);
}

L
Linus Torvalds 已提交
1616 1617 1618 1619 1620
/*
 * OPEN_EXPIRED:
 * 	reclaim state on the server after a network partition.
 * 	Assumes caller holds the appropriate lock
 */
1621
static int _nfs4_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1622
{
1623
	struct nfs4_opendata *opendata;
1624
	int ret;
L
Linus Torvalds 已提交
1625

T
Trond Myklebust 已提交
1626 1627 1628
	opendata = nfs4_open_recoverdata_alloc(ctx, state);
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1629
	ret = nfs4_open_recover(opendata, state);
1630
	if (ret == -ESTALE)
1631
		d_drop(ctx->path.dentry);
1632
	nfs4_opendata_put(opendata);
1633
	return ret;
L
Linus Torvalds 已提交
1634 1635
}

1636
static int nfs4_do_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
1637
{
1638
	struct nfs_server *server = NFS_SERVER(state->inode);
1639 1640 1641 1642
	struct nfs4_exception exception = { };
	int err;

	do {
1643
		err = _nfs4_open_expired(ctx, state);
1644 1645 1646 1647 1648 1649 1650 1651
		switch (err) {
		default:
			goto out;
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			nfs4_handle_exception(server, err, &exception);
			err = 0;
		}
1652
	} while (exception.retry);
1653
out:
1654 1655 1656
	return err;
}

L
Linus Torvalds 已提交
1657 1658 1659
static int nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
	struct nfs_open_context *ctx;
1660
	int ret;
L
Linus Torvalds 已提交
1661

1662 1663 1664
	ctx = nfs4_state_find_open_context(state);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);
1665
	ret = nfs4_do_open_expired(ctx, state);
1666 1667
	put_nfs_open_context(ctx);
	return ret;
L
Linus Torvalds 已提交
1668 1669
}

1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685
/*
 * 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
1686
/*
1687
 * Returns a referenced nfs4_state
L
Linus Torvalds 已提交
1688
 */
1689
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
1690 1691 1692 1693
{
	struct nfs4_state_owner  *sp;
	struct nfs4_state     *state = NULL;
	struct nfs_server       *server = NFS_SERVER(dir);
1694
	struct nfs4_opendata *opendata;
1695
	int status;
L
Linus Torvalds 已提交
1696 1697 1698 1699 1700 1701 1702

	/* 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;
	}
1703 1704
	status = nfs4_recover_expired_lease(server);
	if (status != 0)
1705
		goto err_put_state_owner;
1706
	if (path->dentry->d_inode != NULL)
1707
		nfs4_return_incompatible_delegation(path->dentry->d_inode, fmode);
1708
	status = -ENOMEM;
1709
	opendata = nfs4_opendata_alloc(path, sp, fmode, flags, sattr, GFP_KERNEL);
1710
	if (opendata == NULL)
T
Trond Myklebust 已提交
1711
		goto err_put_state_owner;
L
Linus Torvalds 已提交
1712

1713 1714 1715
	if (path->dentry->d_inode != NULL)
		opendata->state = nfs4_get_open_state(path->dentry->d_inode, sp);

1716
	status = _nfs4_proc_open(opendata);
L
Linus Torvalds 已提交
1717
	if (status != 0)
1718
		goto err_opendata_put;
L
Linus Torvalds 已提交
1719

1720
	state = nfs4_opendata_to_nfs4_state(opendata);
1721 1722
	status = PTR_ERR(state);
	if (IS_ERR(state))
1723
		goto err_opendata_put;
T
Trond Myklebust 已提交
1724
	if (server->caps & NFS_CAP_POSIX_LOCK)
1725
		set_bit(NFS_STATE_POSIX_LOCKS, &state->flags);
1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737

	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);
	}
1738
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
1739 1740 1741
	nfs4_put_state_owner(sp);
	*res = state;
	return 0;
1742 1743
err_opendata_put:
	nfs4_opendata_put(opendata);
1744 1745
err_put_state_owner:
	nfs4_put_state_owner(sp);
L
Linus Torvalds 已提交
1746 1747 1748 1749 1750 1751
out_err:
	*res = NULL;
	return status;
}


1752
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
1753 1754 1755 1756 1757 1758
{
	struct nfs4_exception exception = { };
	struct nfs4_state *res;
	int status;

	do {
1759
		status = _nfs4_do_open(dir, path, fmode, flags, sattr, cred, &res);
L
Linus Torvalds 已提交
1760 1761 1762 1763 1764 1765 1766 1767
		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
1768
		 * doing an OPEN, we assume that nfs_increment_open_seqid() will
L
Linus Torvalds 已提交
1769 1770 1771 1772 1773
		 * 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) {
1774 1775 1776
			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
1777 1778 1779
			exception.retry = 1;
			continue;
		}
1780 1781 1782 1783 1784 1785 1786 1787 1788 1789
		/*
		 * 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;
		}
1790 1791 1792 1793 1794
		if (status == -EAGAIN) {
			/* We must have found a delegation */
			exception.retry = 1;
			continue;
		}
L
Linus Torvalds 已提交
1795 1796 1797 1798 1799 1800
		res = ERR_PTR(nfs4_handle_exception(NFS_SERVER(dir),
					status, &exception));
	} while (exception.retry);
	return res;
}

1801 1802 1803
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
1804
{
1805
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
1806
        struct nfs_setattrargs  arg = {
1807
                .fh             = NFS_FH(inode),
L
Linus Torvalds 已提交
1808 1809 1810 1811 1812 1813 1814 1815 1816
                .iap            = sattr,
		.server		= server,
		.bitmask = server->attr_bitmask,
        };
        struct nfs_setattrres  res = {
		.fattr		= fattr,
		.server		= server,
        };
        struct rpc_message msg = {
1817 1818 1819 1820
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETATTR],
		.rpc_argp	= &arg,
		.rpc_resp	= &res,
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
1821
        };
1822
	unsigned long timestamp = jiffies;
1823
	int status;
L
Linus Torvalds 已提交
1824

1825
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
1826

1827 1828 1829
	if (nfs4_copy_delegation_stateid(&arg.stateid, inode)) {
		/* Use that stateid */
	} else if (state != NULL) {
1830
		nfs4_copy_stateid(&arg.stateid, state, current->files, current->tgid);
1831
	} else
L
Linus Torvalds 已提交
1832 1833
		memcpy(&arg.stateid, &zero_stateid, sizeof(arg.stateid));

A
Andy Adamson 已提交
1834
	status = nfs4_call_sync(server, &msg, &arg, &res, 1);
1835 1836
	if (status == 0 && state != NULL)
		renew_lease(server, timestamp);
1837
	return status;
L
Linus Torvalds 已提交
1838 1839
}

1840 1841 1842
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
1843
{
1844
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
1845 1846 1847 1848
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(server,
1849
				_nfs4_do_setattr(inode, cred, fattr, sattr, state),
L
Linus Torvalds 已提交
1850 1851 1852 1853 1854 1855
				&exception);
	} while (exception.retry);
	return err;
}

struct nfs4_closedata {
1856
	struct path path;
L
Linus Torvalds 已提交
1857 1858 1859 1860
	struct inode *inode;
	struct nfs4_state *state;
	struct nfs_closeargs arg;
	struct nfs_closeres res;
1861
	struct nfs_fattr fattr;
1862
	unsigned long timestamp;
F
Fred Isaman 已提交
1863 1864
	bool roc;
	u32 roc_barrier;
L
Linus Torvalds 已提交
1865 1866
};

1867
static void nfs4_free_closedata(void *data)
1868
{
1869 1870
	struct nfs4_closedata *calldata = data;
	struct nfs4_state_owner *sp = calldata->state->owner;
1871

F
Fred Isaman 已提交
1872 1873
	if (calldata->roc)
		pnfs_roc_release(calldata->state->inode);
1874 1875 1876
	nfs4_put_open_state(calldata->state);
	nfs_free_seqid(calldata->arg.seqid);
	nfs4_put_state_owner(sp);
J
Jan Blunck 已提交
1877
	path_put(&calldata->path);
1878 1879 1880
	kfree(calldata);
}

1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892
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);
}

1893
static void nfs4_close_done(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
1894
{
1895
	struct nfs4_closedata *calldata = data;
L
Linus Torvalds 已提交
1896 1897 1898
	struct nfs4_state *state = calldata->state;
	struct nfs_server *server = NFS_SERVER(calldata->inode);

1899 1900
	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
		return;
L
Linus Torvalds 已提交
1901 1902 1903 1904 1905
        /* 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
1906 1907 1908
			if (calldata->roc)
				pnfs_roc_set_barrier(state->inode,
						     calldata->roc_barrier);
1909
			nfs_set_open_stateid(state, &calldata->res.stateid, 0);
1910
			renew_lease(server, calldata->timestamp);
1911 1912
			nfs4_close_clear_stateid_flags(state,
					calldata->arg.fmode);
L
Linus Torvalds 已提交
1913 1914
			break;
		case -NFS4ERR_STALE_STATEID:
1915 1916
		case -NFS4ERR_OLD_STATEID:
		case -NFS4ERR_BAD_STATEID:
L
Linus Torvalds 已提交
1917
		case -NFS4ERR_EXPIRED:
1918
			if (calldata->arg.fmode == 0)
1919
				break;
L
Linus Torvalds 已提交
1920
		default:
1921 1922
			if (nfs4_async_handle_error(task, server, state) == -EAGAIN)
				rpc_restart_call_prepare(task);
L
Linus Torvalds 已提交
1923
	}
1924
	nfs_release_seqid(calldata->arg.seqid);
1925
	nfs_refresh_inode(calldata->inode, calldata->res.fattr);
L
Linus Torvalds 已提交
1926 1927
}

T
Trond Myklebust 已提交
1928
static void nfs4_close_prepare(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
1929
{
T
Trond Myklebust 已提交
1930
	struct nfs4_closedata *calldata = data;
1931
	struct nfs4_state *state = calldata->state;
1932
	int call_close = 0;
1933

1934
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
1935
		return;
1936

1937 1938
	task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_DOWNGRADE];
	calldata->arg.fmode = FMODE_READ|FMODE_WRITE;
1939
	spin_lock(&state->owner->so_lock);
1940
	/* Calculate the change in open mode */
1941
	if (state->n_rdwr == 0) {
1942
		if (state->n_rdonly == 0) {
1943 1944 1945
			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;
1946 1947
		}
		if (state->n_wronly == 0) {
1948 1949 1950
			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;
1951
		}
1952
	}
1953
	spin_unlock(&state->owner->so_lock);
1954 1955

	if (!call_close) {
1956 1957
		/* Note: exit _without_ calling nfs4_close_done */
		task->tk_action = NULL;
1958 1959
		return;
	}
1960

F
Fred Isaman 已提交
1961
	if (calldata->arg.fmode == 0) {
1962
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE];
F
Fred Isaman 已提交
1963 1964 1965 1966 1967 1968 1969
		if (calldata->roc &&
		    pnfs_roc_drain(calldata->inode, &calldata->roc_barrier)) {
			rpc_sleep_on(&NFS_SERVER(calldata->inode)->roc_rpcwaitq,
				     task, NULL);
			return;
		}
	}
1970

1971
	nfs_fattr_init(calldata->res.fattr);
1972
	calldata->timestamp = jiffies;
1973
	if (nfs4_setup_sequence(NFS_SERVER(calldata->inode),
1974 1975 1976
				&calldata->arg.seq_args, &calldata->res.seq_res,
				1, task))
		return;
1977
	rpc_call_start(task);
L
Linus Torvalds 已提交
1978 1979
}

1980
static const struct rpc_call_ops nfs4_close_ops = {
T
Trond Myklebust 已提交
1981
	.rpc_call_prepare = nfs4_close_prepare,
1982 1983 1984 1985
	.rpc_call_done = nfs4_close_done,
	.rpc_release = nfs4_free_closedata,
};

L
Linus Torvalds 已提交
1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
/* 
 * 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
1997
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
1998
{
1999
	struct nfs_server *server = NFS_SERVER(state->inode);
L
Linus Torvalds 已提交
2000
	struct nfs4_closedata *calldata;
2001 2002
	struct nfs4_state_owner *sp = state->owner;
	struct rpc_task *task;
2003 2004 2005 2006
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
2007 2008
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
2009
		.rpc_message = &msg,
T
Trond Myklebust 已提交
2010
		.callback_ops = &nfs4_close_ops,
2011
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
2012 2013
		.flags = RPC_TASK_ASYNC,
	};
2014
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2015

2016
	calldata = kzalloc(sizeof(*calldata), gfp_mask);
L
Linus Torvalds 已提交
2017
	if (calldata == NULL)
2018
		goto out;
2019
	calldata->inode = state->inode;
L
Linus Torvalds 已提交
2020
	calldata->state = state;
2021
	calldata->arg.fh = NFS_FH(state->inode);
2022
	calldata->arg.stateid = &state->open_stateid;
L
Linus Torvalds 已提交
2023
	/* Serialization for the sequence id */
2024
	calldata->arg.seqid = nfs_alloc_seqid(&state->owner->so_seqid, gfp_mask);
2025 2026
	if (calldata->arg.seqid == NULL)
		goto out_free_calldata;
2027
	calldata->arg.fmode = 0;
2028
	calldata->arg.bitmask = server->cache_consistency_bitmask;
2029
	calldata->res.fattr = &calldata->fattr;
2030
	calldata->res.seqid = calldata->arg.seqid;
2031
	calldata->res.server = server;
F
Fred Isaman 已提交
2032
	calldata->roc = roc;
2033 2034
	path_get(path);
	calldata->path = *path;
2035

2036 2037
	msg.rpc_argp = &calldata->arg;
	msg.rpc_resp = &calldata->res;
T
Trond Myklebust 已提交
2038 2039
	task_setup_data.callback_data = calldata;
	task = rpc_run_task(&task_setup_data);
2040 2041
	if (IS_ERR(task))
		return PTR_ERR(task);
2042 2043 2044
	status = 0;
	if (wait)
		status = rpc_wait_for_completion_task(task);
2045
	rpc_put_task(task);
2046
	return status;
2047 2048 2049
out_free_calldata:
	kfree(calldata);
out:
F
Fred Isaman 已提交
2050 2051
	if (roc)
		pnfs_roc_release(state->inode);
2052 2053
	nfs4_put_open_state(state);
	nfs4_put_state_owner(sp);
2054
	return status;
L
Linus Torvalds 已提交
2055 2056
}

2057
static struct inode *
2058
nfs4_atomic_open(struct inode *dir, struct nfs_open_context *ctx, int open_flags, struct iattr *attr)
L
Linus Torvalds 已提交
2059 2060 2061
{
	struct nfs4_state *state;

T
Trond Myklebust 已提交
2062
	/* Protect against concurrent sillydeletes */
2063
	state = nfs4_do_open(dir, &ctx->path, ctx->mode, open_flags, attr, ctx->cred);
2064 2065
	if (IS_ERR(state))
		return ERR_CAST(state);
2066
	ctx->state = state;
2067
	return igrab(state->inode);
L
Linus Torvalds 已提交
2068 2069
}

2070
static void nfs4_close_context(struct nfs_open_context *ctx, int is_sync)
T
Trond Myklebust 已提交
2071 2072 2073 2074 2075 2076 2077 2078
{
	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
2079 2080 2081

static int _nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
{
B
Benny Halevy 已提交
2082 2083 2084
	struct nfs4_server_caps_arg args = {
		.fhandle = fhandle,
	};
L
Linus Torvalds 已提交
2085 2086 2087
	struct nfs4_server_caps_res res = {};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SERVER_CAPS],
B
Benny Halevy 已提交
2088
		.rpc_argp = &args,
L
Linus Torvalds 已提交
2089 2090 2091 2092
		.rpc_resp = &res,
	};
	int status;

A
Andy Adamson 已提交
2093
	status = nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
2094 2095
	if (status == 0) {
		memcpy(server->attr_bitmask, res.attr_bitmask, sizeof(server->attr_bitmask));
2096 2097 2098 2099 2100
		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
2101 2102 2103 2104 2105 2106
		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;
2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123
		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;

2124 2125 2126
		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
2127 2128
		server->acl_bitmask = res.acl_bitmask;
	}
A
Andy Adamson 已提交
2129

L
Linus Torvalds 已提交
2130 2131 2132
	return status;
}

2133
int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
L
Linus Torvalds 已提交
2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152
{
	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,
2153
		.fattr = info->fattr,
L
Linus Torvalds 已提交
2154 2155 2156 2157 2158 2159 2160
		.fh = fhandle,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP_ROOT],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
2161

2162
	nfs_fattr_init(info->fattr);
2163
	return nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178
}

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;
}

2179 2180 2181
/*
 * get the file handle for the "/" directory on the server
 */
L
Linus Torvalds 已提交
2182
static int nfs4_proc_get_root(struct nfs_server *server, struct nfs_fh *fhandle,
2183
			      struct nfs_fsinfo *info)
L
Linus Torvalds 已提交
2184 2185 2186 2187 2188 2189 2190 2191
{
	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);
2192
	return nfs4_map_errors(status);
L
Linus Torvalds 已提交
2193 2194
}

M
Manoj Naik 已提交
2195 2196 2197 2198 2199
/*
 * 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
 */
2200
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
2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212
{
	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;

2213
	status = nfs4_proc_fs_locations(dir, name, locations, page);
M
Manoj Naik 已提交
2214 2215 2216 2217
	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)) {
2218
		dprintk("%s: server did not return a different fsid for a referral at %s\n", __func__, name->name);
M
Manoj Naik 已提交
2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230
		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);
2231
	kfree(locations);
M
Manoj Naik 已提交
2232 2233 2234
	return status;
}

L
Linus Torvalds 已提交
2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250
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,
	};
	
2251
	nfs_fattr_init(fattr);
A
Andy Adamson 已提交
2252
	return nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287
}

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)
{
2288
	struct inode *inode = dentry->d_inode;
2289
	struct rpc_cred *cred = NULL;
2290
	struct nfs4_state *state = NULL;
L
Linus Torvalds 已提交
2291 2292
	int status;

2293
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
2294
	
2295
	/* Search for an existing open(O_WRITE) file */
2296 2297 2298 2299
	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
2300 2301 2302 2303
		if (ctx) {
			cred = ctx->cred;
			state = ctx->state;
		}
2304
	}
2305

2306
	status = nfs4_do_setattr(inode, cred, fattr, sattr, state);
2307 2308
	if (status == 0)
		nfs_setattr_update_inode(inode, sattr);
L
Linus Torvalds 已提交
2309 2310 2311
	return status;
}

2312 2313
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
2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335
		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
2336
	status = nfs4_call_sync(server, &msg, &args, &res, 0);
D
David Howells 已提交
2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347
	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 {
2348 2349 2350 2351 2352 2353 2354
		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
2355 2356 2357 2358
	} while (exception.retry);
	return err;
}

2359
static int _nfs4_proc_lookup(struct inode *dir, const struct qstr *name,
L
Linus Torvalds 已提交
2360 2361
		struct nfs_fh *fhandle, struct nfs_fattr *fattr)
{
2362
	int status;
L
Linus Torvalds 已提交
2363 2364
	
	dprintk("NFS call  lookup %s\n", name->name);
2365
	status = _nfs4_proc_lookupfh(NFS_SERVER(dir), NFS_FH(dir), name, fhandle, fattr);
M
Manoj Naik 已提交
2366 2367
	if (status == -NFS4ERR_MOVED)
		status = nfs4_get_referral(dir, name, fattr, fhandle);
L
Linus Torvalds 已提交
2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385
	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)
{
2386
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
2387 2388
	struct nfs4_accessargs args = {
		.fh = NFS_FH(inode),
2389 2390 2391 2392
		.bitmask = server->attr_bitmask,
	};
	struct nfs4_accessres res = {
		.server = server,
L
Linus Torvalds 已提交
2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418
	};
	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;
	}
2419 2420 2421 2422 2423

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

A
Andy Adamson 已提交
2424
	status = nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
2425 2426 2427 2428 2429 2430 2431 2432
	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;
2433
		nfs_refresh_inode(inode, res.fattr);
L
Linus Torvalds 已提交
2434
	}
2435
	nfs_free_fattr(res.fattr);
L
Linus Torvalds 已提交
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 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483
	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
2484
	struct nfs4_readlink_res res;
L
Linus Torvalds 已提交
2485 2486 2487
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READLINK],
		.rpc_argp = &args,
B
Benny Halevy 已提交
2488
		.rpc_resp = &res,
L
Linus Torvalds 已提交
2489 2490
	};

A
Andy Adamson 已提交
2491
	return nfs4_call_sync(NFS_SERVER(inode), &msg, &args, &res, 0);
L
Linus Torvalds 已提交
2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523
}

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,
2524
                 int flags, struct nfs_open_context *ctx)
L
Linus Torvalds 已提交
2525
{
2526
	struct path my_path = {
2527 2528
		.dentry = dentry,
	};
2529
	struct path *path = &my_path;
L
Linus Torvalds 已提交
2530
	struct nfs4_state *state;
2531 2532
	struct rpc_cred *cred = NULL;
	fmode_t fmode = 0;
L
Linus Torvalds 已提交
2533 2534
	int status = 0;

2535 2536 2537 2538
	if (ctx != NULL) {
		cred = ctx->cred;
		path = &ctx->path;
		fmode = ctx->mode;
L
Linus Torvalds 已提交
2539
	}
A
Aneesh Kumar K.V 已提交
2540
	sattr->ia_mode &= ~current_umask();
2541
	state = nfs4_do_open(dir, path, fmode, flags, sattr, cred);
2542
	d_drop(dentry);
L
Linus Torvalds 已提交
2543 2544
	if (IS_ERR(state)) {
		status = PTR_ERR(state);
2545
		goto out;
L
Linus Torvalds 已提交
2546
	}
2547
	d_add(dentry, igrab(state->inode));
2548
	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
2549 2550
	if (ctx != NULL)
		ctx->state = state;
2551
	else
2552
		nfs4_close_sync(path, state, fmode);
L
Linus Torvalds 已提交
2553 2554 2555 2556 2557 2558
out:
	return status;
}

static int _nfs4_proc_remove(struct inode *dir, struct qstr *name)
{
2559
	struct nfs_server *server = NFS_SERVER(dir);
2560
	struct nfs_removeargs args = {
L
Linus Torvalds 已提交
2561
		.fh = NFS_FH(dir),
2562 2563
		.name.len = name->len,
		.name.name = name->name,
2564 2565
		.bitmask = server->attr_bitmask,
	};
2566
	struct nfs_removeres res = {
2567
		.server = server,
L
Linus Torvalds 已提交
2568 2569
	};
	struct rpc_message msg = {
2570 2571 2572
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE],
		.rpc_argp = &args,
		.rpc_resp = &res,
L
Linus Torvalds 已提交
2573
	};
2574 2575 2576 2577 2578
	int status = -ENOMEM;

	res.dir_attr = nfs_alloc_fattr();
	if (res.dir_attr == NULL)
		goto out;
L
Linus Torvalds 已提交
2579

A
Andy Adamson 已提交
2580
	status = nfs4_call_sync(server, &msg, &args, &res, 1);
2581 2582
	if (status == 0) {
		update_changeattr(dir, &res.cinfo);
2583
		nfs_post_op_update_inode(dir, res.dir_attr);
2584
	}
2585 2586
	nfs_free_fattr(res.dir_attr);
out:
L
Linus Torvalds 已提交
2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601
	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;
}

2602
static void nfs4_proc_unlink_setup(struct rpc_message *msg, struct inode *dir)
L
Linus Torvalds 已提交
2603
{
2604 2605 2606
	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
2607

2608
	args->bitmask = server->cache_consistency_bitmask;
2609
	res->server = server;
2610
	res->seq_res.sr_slot = NULL;
L
Linus Torvalds 已提交
2611 2612 2613
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE];
}

2614
static int nfs4_proc_unlink_done(struct rpc_task *task, struct inode *dir)
L
Linus Torvalds 已提交
2615
{
2616 2617
	struct nfs_removeres *res = task->tk_msg.rpc_resp;

2618 2619
	if (!nfs4_sequence_done(task, &res->seq_res))
		return 0;
2620
	if (nfs4_async_handle_error(task, res->server, NULL) == -EAGAIN)
2621 2622
		return 0;
	update_changeattr(dir, &res->cinfo);
2623
	nfs_post_op_update_inode(dir, res->dir_attr);
2624
	return 1;
L
Linus Torvalds 已提交
2625 2626
}

2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654
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
2655 2656 2657
static int _nfs4_proc_rename(struct inode *old_dir, struct qstr *old_name,
		struct inode *new_dir, struct qstr *new_name)
{
2658
	struct nfs_server *server = NFS_SERVER(old_dir);
2659
	struct nfs_renameargs arg = {
L
Linus Torvalds 已提交
2660 2661 2662 2663
		.old_dir = NFS_FH(old_dir),
		.new_dir = NFS_FH(new_dir),
		.old_name = old_name,
		.new_name = new_name,
2664 2665
		.bitmask = server->attr_bitmask,
	};
2666
	struct nfs_renameres res = {
2667
		.server = server,
L
Linus Torvalds 已提交
2668 2669 2670 2671 2672 2673
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RENAME],
		.rpc_argp = &arg,
		.rpc_resp = &res,
	};
2674
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2675
	
2676 2677 2678 2679
	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
2680

2681
	status = nfs4_call_sync(server, &msg, &arg, &res, 1);
L
Linus Torvalds 已提交
2682 2683
	if (!status) {
		update_changeattr(old_dir, &res.old_cinfo);
2684
		nfs_post_op_update_inode(old_dir, res.old_fattr);
L
Linus Torvalds 已提交
2685
		update_changeattr(new_dir, &res.new_cinfo);
2686
		nfs_post_op_update_inode(new_dir, res.new_fattr);
L
Linus Torvalds 已提交
2687
	}
2688 2689 2690
out:
	nfs_free_fattr(res.new_fattr);
	nfs_free_fattr(res.old_fattr);
L
Linus Torvalds 已提交
2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709
	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)
{
2710
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
2711 2712 2713 2714
	struct nfs4_link_arg arg = {
		.fh     = NFS_FH(inode),
		.dir_fh = NFS_FH(dir),
		.name   = name,
2715 2716 2717 2718
		.bitmask = server->attr_bitmask,
	};
	struct nfs4_link_res res = {
		.server = server,
L
Linus Torvalds 已提交
2719 2720 2721 2722
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LINK],
		.rpc_argp = &arg,
2723
		.rpc_resp = &res,
L
Linus Torvalds 已提交
2724
	};
2725 2726 2727 2728 2729 2730
	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
2731

A
Andy Adamson 已提交
2732
	status = nfs4_call_sync(server, &msg, &arg, &res, 1);
2733 2734 2735
	if (!status) {
		update_changeattr(dir, &res.cinfo);
		nfs_post_op_update_inode(dir, res.dir_attr);
2736
		nfs_post_op_update_inode(inode, res.fattr);
2737
	}
2738 2739 2740
out:
	nfs_free_fattr(res.dir_attr);
	nfs_free_fattr(res.fattr);
L
Linus Torvalds 已提交
2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755
	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;
}

2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794
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
2795 2796
	int status = nfs4_call_sync(NFS_SERVER(dir), &data->msg,
				    &data->arg, &data->res, 1);
2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809
	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);
}

2810
static int _nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
2811
		struct page *page, unsigned int len, struct iattr *sattr)
L
Linus Torvalds 已提交
2812
{
2813 2814
	struct nfs4_createdata *data;
	int status = -ENAMETOOLONG;
L
Linus Torvalds 已提交
2815

2816
	if (len > NFS4_MAXPATHLEN)
2817
		goto out;
2818

2819 2820 2821 2822 2823 2824 2825 2826
	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
2827
	
2828 2829 2830 2831
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
2832 2833 2834
	return status;
}

2835
static int nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
2836
		struct page *page, unsigned int len, struct iattr *sattr)
L
Linus Torvalds 已提交
2837 2838 2839 2840 2841
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(NFS_SERVER(dir),
2842 2843
				_nfs4_proc_symlink(dir, dentry, page,
							len, sattr),
L
Linus Torvalds 已提交
2844 2845 2846 2847 2848 2849 2850 2851
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr)
{
2852 2853
	struct nfs4_createdata *data;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2854

2855 2856 2857 2858 2859 2860 2861 2862
	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
2863 2864 2865 2866 2867 2868 2869 2870
	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
2871 2872

	sattr->ia_mode &= ~current_umask();
L
Linus Torvalds 已提交
2873 2874 2875 2876 2877 2878 2879 2880 2881
	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
2882
		u64 cookie, struct page **pages, unsigned int count, int plus)
L
Linus Torvalds 已提交
2883 2884 2885 2886
{
	struct inode		*dir = dentry->d_inode;
	struct nfs4_readdir_arg args = {
		.fh = NFS_FH(dir),
B
Bryan Schumaker 已提交
2887
		.pages = pages,
L
Linus Torvalds 已提交
2888 2889
		.pgbase = 0,
		.count = count,
2890
		.bitmask = NFS_SERVER(dentry->d_inode)->attr_bitmask,
B
Bryan Schumaker 已提交
2891
		.plus = plus,
L
Linus Torvalds 已提交
2892 2893 2894 2895 2896 2897 2898 2899 2900 2901
	};
	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;

2902
	dprintk("%s: dentry = %s/%s, cookie = %Lu\n", __func__,
2903 2904 2905
			dentry->d_parent->d_name.name,
			dentry->d_name.name,
			(unsigned long long)cookie);
L
Linus Torvalds 已提交
2906 2907
	nfs4_setup_readdir(cookie, NFS_COOKIEVERF(dir), dentry, &args);
	res.pgbase = args.pgbase;
A
Andy Adamson 已提交
2908
	status = nfs4_call_sync(NFS_SERVER(dir), &msg, &args, &res, 0);
2909
	if (status >= 0) {
L
Linus Torvalds 已提交
2910
		memcpy(NFS_COOKIEVERF(dir), res.verifier.data, NFS4_VERIFIER_SIZE);
2911 2912
		status += args.pgbase;
	}
2913 2914 2915

	nfs_invalidate_atime(dir);

2916
	dprintk("%s: returns %d\n", __func__, status);
L
Linus Torvalds 已提交
2917 2918 2919 2920
	return status;
}

static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
B
Bryan Schumaker 已提交
2921
		u64 cookie, struct page **pages, unsigned int count, int plus)
L
Linus Torvalds 已提交
2922 2923 2924 2925 2926 2927
{
	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
2928
					pages, count, plus),
L
Linus Torvalds 已提交
2929 2930 2931 2932 2933 2934 2935 2936
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr, dev_t rdev)
{
2937 2938 2939
	struct nfs4_createdata *data;
	int mode = sattr->ia_mode;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2940 2941 2942

	BUG_ON(!(sattr->ia_valid & ATTR_MODE));
	BUG_ON(!S_ISFIFO(mode) && !S_ISBLK(mode) && !S_ISCHR(mode) && !S_ISSOCK(mode));
2943 2944 2945 2946 2947

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

L
Linus Torvalds 已提交
2948
	if (S_ISFIFO(mode))
2949
		data->arg.ftype = NF4FIFO;
L
Linus Torvalds 已提交
2950
	else if (S_ISBLK(mode)) {
2951 2952 2953
		data->arg.ftype = NF4BLK;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
L
Linus Torvalds 已提交
2954 2955
	}
	else if (S_ISCHR(mode)) {
2956 2957 2958
		data->arg.ftype = NF4CHR;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
L
Linus Torvalds 已提交
2959 2960
	}
	
2961 2962 2963 2964
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
2965 2966 2967 2968 2969 2970 2971 2972
	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
2973 2974

	sattr->ia_mode &= ~current_umask();
L
Linus Torvalds 已提交
2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989
	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
2990 2991 2992
	struct nfs4_statfs_res res = {
		.fsstat = fsstat,
	};
L
Linus Torvalds 已提交
2993 2994 2995
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_STATFS],
		.rpc_argp = &args,
B
Benny Halevy 已提交
2996
		.rpc_resp = &res,
L
Linus Torvalds 已提交
2997 2998
	};

2999
	nfs_fattr_init(fsstat->fattr);
A
Andy Adamson 已提交
3000
	return  nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021
}

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
3022 3023 3024
	struct nfs4_fsinfo_res res = {
		.fsinfo = fsinfo,
	};
L
Linus Torvalds 已提交
3025 3026 3027
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FSINFO],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3028
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3029 3030
	};

A
Andy Adamson 已提交
3031
	return nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048
}

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)
{
3049
	nfs_fattr_init(fsinfo->fattr);
L
Linus Torvalds 已提交
3050 3051 3052 3053 3054 3055 3056 3057 3058 3059
	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
3060 3061 3062
	struct nfs4_pathconf_res res = {
		.pathconf = pathconf,
	};
L
Linus Torvalds 已提交
3063 3064 3065
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_PATHCONF],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3066
		.rpc_resp = &res,
L
Linus Torvalds 已提交
3067 3068 3069 3070 3071 3072 3073 3074
	};

	/* 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;
	}

3075
	nfs_fattr_init(pathconf->fattr);
A
Andy Adamson 已提交
3076
	return nfs4_call_sync(server, &msg, &args, &res, 0);
L
Linus Torvalds 已提交
3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092
}

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;
}

3093
static int nfs4_read_done_cb(struct rpc_task *task, struct nfs_read_data *data)
L
Linus Torvalds 已提交
3094
{
T
Trond Myklebust 已提交
3095
	struct nfs_server *server = NFS_SERVER(data->inode);
L
Linus Torvalds 已提交
3096

3097
	if (nfs4_async_handle_error(task, server, data->args.context->state) == -EAGAIN) {
3098
		nfs_restart_rpc(task, server->nfs_client);
T
Trond Myklebust 已提交
3099
		return -EAGAIN;
L
Linus Torvalds 已提交
3100
	}
3101 3102

	nfs_invalidate_atime(data->inode);
L
Linus Torvalds 已提交
3103
	if (task->tk_status > 0)
T
Trond Myklebust 已提交
3104 3105
		renew_lease(server, data->timestamp);
	return 0;
L
Linus Torvalds 已提交
3106 3107
}

3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118
static int nfs4_read_done(struct rpc_task *task, struct nfs_read_data *data)
{

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

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

	return data->read_done_cb(task, data);
}

3119
static void nfs4_proc_read_setup(struct nfs_read_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
3120 3121
{
	data->timestamp   = jiffies;
3122
	data->read_done_cb = nfs4_read_done_cb;
3123
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READ];
L
Linus Torvalds 已提交
3124 3125
}

3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141
/* Reset the the nfs_read_data to send the read to the MDS. */
void nfs4_reset_read(struct rpc_task *task, struct nfs_read_data *data)
{
	dprintk("%s Reset task for i/o through\n", __func__);
	put_lseg(data->lseg);
	data->lseg = NULL;
	/* offsets will differ in the dense stripe case */
	data->args.offset = data->mds_offset;
	data->ds_clp = NULL;
	data->args.fh     = NFS_FH(data->inode);
	data->read_done_cb = nfs4_read_done_cb;
	task->tk_ops = data->mds_ops;
	rpc_task_reset_client(task, NFS_CLIENT(data->inode));
}
EXPORT_SYMBOL_GPL(nfs4_reset_read);

3142
static int nfs4_write_done_cb(struct rpc_task *task, struct nfs_write_data *data)
L
Linus Torvalds 已提交
3143 3144 3145
{
	struct inode *inode = data->inode;
	
3146
	if (nfs4_async_handle_error(task, NFS_SERVER(inode), data->args.context->state) == -EAGAIN) {
3147
		nfs_restart_rpc(task, NFS_SERVER(inode)->nfs_client);
3148
		return -EAGAIN;
L
Linus Torvalds 已提交
3149
	}
3150
	if (task->tk_status >= 0) {
L
Linus Torvalds 已提交
3151
		renew_lease(NFS_SERVER(inode), data->timestamp);
3152
		nfs_post_op_update_inode_force_wcc(inode, data->res.fattr);
3153
	}
3154
	return 0;
L
Linus Torvalds 已提交
3155 3156
}

3157 3158 3159 3160 3161 3162 3163
static int nfs4_write_done(struct rpc_task *task, struct nfs_write_data *data)
{
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return -EAGAIN;
	return data->write_done_cb(task, data);
}

3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180
/* Reset the the nfs_write_data to send the write to the MDS. */
void nfs4_reset_write(struct rpc_task *task, struct nfs_write_data *data)
{
	dprintk("%s Reset task for i/o through\n", __func__);
	put_lseg(data->lseg);
	data->lseg          = NULL;
	data->ds_clp        = NULL;
	data->write_done_cb = nfs4_write_done_cb;
	data->args.fh       = NFS_FH(data->inode);
	data->args.bitmask  = data->res.server->cache_consistency_bitmask;
	data->args.offset   = data->mds_offset;
	data->res.fattr     = &data->fattr;
	task->tk_ops        = data->mds_ops;
	rpc_task_reset_client(task, NFS_CLIENT(data->inode));
}
EXPORT_SYMBOL_GPL(nfs4_reset_write);

3181
static void nfs4_proc_write_setup(struct nfs_write_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
3182
{
3183 3184
	struct nfs_server *server = NFS_SERVER(data->inode);

3185 3186 3187 3188 3189
	if (data->lseg) {
		data->args.bitmask = NULL;
		data->res.fattr = NULL;
	} else
		data->args.bitmask = server->cache_consistency_bitmask;
3190 3191
	if (!data->write_done_cb)
		data->write_done_cb = nfs4_write_done_cb;
3192
	data->res.server = server;
L
Linus Torvalds 已提交
3193 3194
	data->timestamp   = jiffies;

3195
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_WRITE];
L
Linus Torvalds 已提交
3196 3197
}

3198
static int nfs4_commit_done_cb(struct rpc_task *task, struct nfs_write_data *data)
L
Linus Torvalds 已提交
3199 3200
{
	struct inode *inode = data->inode;
3201

3202
	if (nfs4_async_handle_error(task, NFS_SERVER(inode), NULL) == -EAGAIN) {
3203
		nfs_restart_rpc(task, NFS_SERVER(inode)->nfs_client);
3204
		return -EAGAIN;
L
Linus Torvalds 已提交
3205
	}
3206
	nfs_refresh_inode(inode, data->res.fattr);
3207
	return 0;
L
Linus Torvalds 已提交
3208 3209
}

3210 3211 3212 3213 3214 3215 3216
static int nfs4_commit_done(struct rpc_task *task, struct nfs_write_data *data)
{
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return -EAGAIN;
	return data->write_done_cb(task, data);
}

3217
static void nfs4_proc_commit_setup(struct nfs_write_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
3218
{
3219
	struct nfs_server *server = NFS_SERVER(data->inode);
3220 3221 3222 3223 3224 3225

	if (data->lseg) {
		data->args.bitmask = NULL;
		data->res.fattr = NULL;
	} else
		data->args.bitmask = server->cache_consistency_bitmask;
3226 3227
	if (!data->write_done_cb)
		data->write_done_cb = nfs4_commit_done_cb;
3228
	data->res.server = server;
3229
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_COMMIT];
L
Linus Torvalds 已提交
3230 3231
}

3232 3233 3234 3235 3236
struct nfs4_renewdata {
	struct nfs_client	*client;
	unsigned long		timestamp;
};

L
Linus Torvalds 已提交
3237 3238 3239 3240
/*
 * nfs4_proc_async_renew(): This is not one of the nfs_rpc_ops; it is a special
 * standalone procedure for queueing an asynchronous RENEW.
 */
3241
static void nfs4_renew_release(void *calldata)
3242
{
3243 3244
	struct nfs4_renewdata *data = calldata;
	struct nfs_client *clp = data->client;
3245

3246 3247 3248
	if (atomic_read(&clp->cl_count) > 1)
		nfs4_schedule_state_renewal(clp);
	nfs_put_client(clp);
3249
	kfree(data);
3250 3251
}

3252
static void nfs4_renew_done(struct rpc_task *task, void *calldata)
L
Linus Torvalds 已提交
3253
{
3254 3255 3256
	struct nfs4_renewdata *data = calldata;
	struct nfs_client *clp = data->client;
	unsigned long timestamp = data->timestamp;
L
Linus Torvalds 已提交
3257 3258

	if (task->tk_status < 0) {
3259 3260
		/* Unless we're shutting down, schedule state recovery! */
		if (test_bit(NFS_CS_RENEWD, &clp->cl_res_state) != 0)
3261
			nfs4_schedule_lease_recovery(clp);
L
Linus Torvalds 已提交
3262 3263
		return;
	}
3264
	do_renew_lease(clp, timestamp);
L
Linus Torvalds 已提交
3265 3266
}

3267 3268
static const struct rpc_call_ops nfs4_renew_ops = {
	.rpc_call_done = nfs4_renew_done,
3269
	.rpc_release = nfs4_renew_release,
3270 3271
};

3272
int nfs4_proc_async_renew(struct nfs_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
3273 3274 3275 3276
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
3277
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
3278
	};
3279
	struct nfs4_renewdata *data;
L
Linus Torvalds 已提交
3280

3281 3282
	if (!atomic_inc_not_zero(&clp->cl_count))
		return -EIO;
3283 3284 3285 3286 3287
	data = kmalloc(sizeof(*data), GFP_KERNEL);
	if (data == NULL)
		return -ENOMEM;
	data->client = clp;
	data->timestamp = jiffies;
L
Linus Torvalds 已提交
3288
	return rpc_call_async(clp->cl_rpcclient, &msg, RPC_TASK_SOFT,
3289
			&nfs4_renew_ops, data);
L
Linus Torvalds 已提交
3290 3291
}

3292
int nfs4_proc_renew(struct nfs_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
3293 3294 3295 3296
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
3297
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
3298 3299 3300 3301 3302 3303 3304
	};
	unsigned long now = jiffies;
	int status;

	status = rpc_call_sync(clp->cl_rpcclient, &msg, 0);
	if (status < 0)
		return status;
3305
	do_renew_lease(clp, now);
L
Linus Torvalds 已提交
3306 3307 3308
	return 0;
}

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
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;
	}
}

3335 3336 3337 3338 3339 3340 3341 3342 3343
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
3344
		len = min_t(size_t, PAGE_CACHE_SIZE, buflen);
3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363
		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;
}

3364 3365 3366
struct nfs4_cached_acl {
	int cached;
	size_t len;
A
Andrew Morton 已提交
3367
	char data[0];
3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430
};

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);
}

3431
static ssize_t __nfs4_get_acl_uncached(struct inode *inode, void *buf, size_t buflen)
3432 3433 3434 3435 3436 3437 3438
{
	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
3439 3440 3441
	struct nfs_getaclres res = {
		.acl_len = buflen,
	};
3442
	void *resp_buf;
3443 3444 3445
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETACL],
		.rpc_argp = &args,
B
Benny Halevy 已提交
3446
		.rpc_resp = &res,
3447
	};
3448
	struct page *localpage = NULL;
3449 3450
	int ret;

3451 3452 3453 3454 3455 3456 3457 3458 3459
	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
3460
		args.acl_len = PAGE_SIZE;
3461 3462 3463 3464
	} else {
		resp_buf = buf;
		buf_to_pages(buf, buflen, args.acl_pages, &args.acl_pgbase);
	}
A
Andy Adamson 已提交
3465
	ret = nfs4_call_sync(NFS_SERVER(inode), &msg, &args, &res, 0);
3466 3467
	if (ret)
		goto out_free;
B
Benny Halevy 已提交
3468 3469
	if (res.acl_len > args.acl_len)
		nfs4_write_cached_acl(inode, NULL, res.acl_len);
3470
	else
B
Benny Halevy 已提交
3471
		nfs4_write_cached_acl(inode, resp_buf, res.acl_len);
3472 3473
	if (buf) {
		ret = -ERANGE;
B
Benny Halevy 已提交
3474
		if (res.acl_len > buflen)
3475 3476
			goto out_free;
		if (localpage)
B
Benny Halevy 已提交
3477
			memcpy(buf, resp_buf, res.acl_len);
3478
	}
B
Benny Halevy 已提交
3479
	ret = res.acl_len;
3480 3481 3482
out_free:
	if (localpage)
		__free_page(localpage);
3483 3484 3485
	return ret;
}

3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498
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;
}

3499 3500 3501 3502 3503 3504 3505 3506 3507 3508
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;
3509 3510
	if (NFS_I(inode)->cache_validity & NFS_INO_INVALID_ACL)
		nfs_zap_acl_cache(inode);
3511 3512 3513 3514 3515 3516
	ret = nfs4_read_cached_acl(inode, buf, buflen);
	if (ret != -ENOENT)
		return ret;
	return nfs4_get_acl_uncached(inode, buf, buflen);
}

3517
static int __nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen)
3518 3519 3520 3521 3522 3523 3524 3525
{
	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
3526
	struct nfs_setaclres res;
3527 3528 3529
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETACL],
		.rpc_argp	= &arg,
B
Benny Halevy 已提交
3530
		.rpc_resp	= &res,
3531
	};
3532
	int ret, i;
3533 3534 3535

	if (!nfs4_server_supports_acls(server))
		return -EOPNOTSUPP;
3536 3537 3538
	i = buf_to_pages_noslab(buf, buflen, arg.acl_pages, &arg.acl_pgbase);
	if (i < 0)
		return i;
3539
	nfs_inode_return_delegation(inode);
A
Andy Adamson 已提交
3540
	ret = nfs4_call_sync(server, &msg, &arg, &res, 1);
3541 3542 3543 3544 3545 3546 3547 3548

	/*
	 * 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]);

3549 3550 3551 3552 3553 3554 3555
	/*
	 * 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);
3556 3557
	nfs_access_zap_cache(inode);
	nfs_zap_acl_cache(inode);
3558 3559 3560
	return ret;
}

3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572
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
3573
static int
3574
nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server, struct nfs4_state *state)
L
Linus Torvalds 已提交
3575
{
3576 3577 3578
	struct nfs_client *clp = server->nfs_client;

	if (task->tk_status >= 0)
L
Linus Torvalds 已提交
3579 3580
		return 0;
	switch(task->tk_status) {
3581 3582 3583 3584 3585
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OPENMODE:
			if (state == NULL)
				break;
3586 3587
			nfs4_schedule_stateid_recovery(server, state);
			goto wait_on_recovery;
L
Linus Torvalds 已提交
3588
		case -NFS4ERR_STALE_STATEID:
3589
		case -NFS4ERR_STALE_CLIENTID:
L
Linus Torvalds 已提交
3590
		case -NFS4ERR_EXPIRED:
3591 3592
			nfs4_schedule_lease_recovery(clp);
			goto wait_on_recovery;
3593 3594 3595 3596 3597 3598 3599 3600 3601 3602
#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);
3603
			nfs4_schedule_session_recovery(clp->cl_session);
3604 3605 3606
			task->tk_status = 0;
			return -EAGAIN;
#endif /* CONFIG_NFS_V4_1 */
L
Linus Torvalds 已提交
3607
		case -NFS4ERR_DELAY:
3608
			nfs_inc_server_stats(server, NFSIOS_DELAY);
3609
		case -NFS4ERR_GRACE:
3610
		case -EKEYEXPIRED:
L
Linus Torvalds 已提交
3611 3612 3613 3614 3615 3616 3617 3618 3619
			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;
3620
wait_on_recovery:
3621 3622 3623 3624 3625
	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
3626 3627
}

3628 3629 3630
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
3631 3632 3633 3634 3635
{
	nfs4_verifier sc_verifier;
	struct nfs4_setclientid setclientid = {
		.sc_verifier = &sc_verifier,
		.sc_prog = program,
A
Andy Adamson 已提交
3636
		.sc_cb_ident = clp->cl_cb_ident,
L
Linus Torvalds 已提交
3637 3638 3639 3640
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID],
		.rpc_argp = &setclientid,
3641
		.rpc_resp = res,
3642
		.rpc_cred = cred,
L
Linus Torvalds 已提交
3643
	};
A
Al Viro 已提交
3644
	__be32 *p;
L
Linus Torvalds 已提交
3645 3646 3647
	int loop = 0;
	int status;

A
Al Viro 已提交
3648
	p = (__be32*)sc_verifier.data;
L
Linus Torvalds 已提交
3649 3650 3651 3652 3653
	*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,
3654
				sizeof(setclientid.sc_name), "%s/%s %s %s %u",
3655 3656 3657
				clp->cl_ipaddr,
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_ADDR),
3658 3659
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_PROTO),
3660
				clp->cl_rpcclient->cl_auth->au_ops->au_name,
L
Linus Torvalds 已提交
3661 3662
				clp->cl_id_uniquifier);
		setclientid.sc_netid_len = scnprintf(setclientid.sc_netid,
3663 3664 3665
				sizeof(setclientid.sc_netid),
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_NETID));
L
Linus Torvalds 已提交
3666
		setclientid.sc_uaddr_len = scnprintf(setclientid.sc_uaddr,
3667
				sizeof(setclientid.sc_uaddr), "%s.%u.%u",
L
Linus Torvalds 已提交
3668 3669 3670 3671 3672 3673 3674 3675
				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)
3676
			ssleep(clp->cl_lease_time / HZ + 1);
L
Linus Torvalds 已提交
3677 3678 3679 3680 3681 3682 3683
		else
			if (++clp->cl_id_uniquifier == 0)
				break;
	}
	return status;
}

3684 3685 3686
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
3687 3688 3689 3690
{
	struct nfs_fsinfo fsinfo;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID_CONFIRM],
3691
		.rpc_argp = arg,
L
Linus Torvalds 已提交
3692
		.rpc_resp = &fsinfo,
3693
		.rpc_cred = cred,
L
Linus Torvalds 已提交
3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708
	};
	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;
}

3709 3710 3711
int nfs4_proc_setclientid_confirm(struct nfs_client *clp,
		struct nfs4_setclientid_res *arg,
		struct rpc_cred *cred)
3712
{
3713
	long timeout = 0;
3714 3715
	int err;
	do {
3716
		err = _nfs4_proc_setclientid_confirm(clp, arg, cred);
3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728
		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;
}

3729 3730
struct nfs4_delegreturndata {
	struct nfs4_delegreturnargs args;
3731
	struct nfs4_delegreturnres res;
3732 3733
	struct nfs_fh fh;
	nfs4_stateid stateid;
3734
	unsigned long timestamp;
3735
	struct nfs_fattr fattr;
3736 3737 3738 3739 3740 3741
	int rpc_status;
};

static void nfs4_delegreturn_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_delegreturndata *data = calldata;
3742

3743 3744
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return;
3745

3746 3747 3748 3749
	switch (task->tk_status) {
	case -NFS4ERR_STALE_STATEID:
	case -NFS4ERR_EXPIRED:
	case 0:
3750
		renew_lease(data->res.server, data->timestamp);
3751 3752 3753 3754 3755 3756 3757 3758 3759
		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;
3760 3761 3762 3763 3764 3765 3766
}

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

3767 3768 3769 3770 3771 3772 3773
#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;

3774
	if (nfs4_setup_sequence(d_data->res.server,
3775 3776 3777 3778 3779 3780 3781
				&d_data->args.seq_args,
				&d_data->res.seq_res, 1, task))
		return;
	rpc_call_start(task);
}
#endif /* CONFIG_NFS_V4_1 */

3782
static const struct rpc_call_ops nfs4_delegreturn_ops = {
3783 3784 3785
#if defined(CONFIG_NFS_V4_1)
	.rpc_call_prepare = nfs4_delegreturn_prepare,
#endif /* CONFIG_NFS_V4_1 */
3786 3787 3788 3789
	.rpc_call_done = nfs4_delegreturn_done,
	.rpc_release = nfs4_delegreturn_release,
};

3790
static int _nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
3791 3792
{
	struct nfs4_delegreturndata *data;
3793
	struct nfs_server *server = NFS_SERVER(inode);
3794
	struct rpc_task *task;
3795 3796 3797 3798
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DELEGRETURN],
		.rpc_cred = cred,
	};
T
Trond Myklebust 已提交
3799 3800
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
3801
		.rpc_message = &msg,
T
Trond Myklebust 已提交
3802 3803 3804
		.callback_ops = &nfs4_delegreturn_ops,
		.flags = RPC_TASK_ASYNC,
	};
3805
	int status = 0;
3806

3807
	data = kzalloc(sizeof(*data), GFP_NOFS);
3808 3809 3810 3811
	if (data == NULL)
		return -ENOMEM;
	data->args.fhandle = &data->fh;
	data->args.stateid = &data->stateid;
3812
	data->args.bitmask = server->attr_bitmask;
3813 3814
	nfs_copy_fh(&data->fh, NFS_FH(inode));
	memcpy(&data->stateid, stateid, sizeof(data->stateid));
3815 3816
	data->res.fattr = &data->fattr;
	data->res.server = server;
3817
	nfs_fattr_init(data->res.fattr);
3818
	data->timestamp = jiffies;
3819 3820
	data->rpc_status = 0;

T
Trond Myklebust 已提交
3821
	task_setup_data.callback_data = data;
3822 3823
	msg.rpc_argp = &data->args;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
3824
	task = rpc_run_task(&task_setup_data);
3825
	if (IS_ERR(task))
3826
		return PTR_ERR(task);
3827 3828
	if (!issync)
		goto out;
3829
	status = nfs4_wait_for_completion_rpc_task(task);
3830 3831 3832 3833 3834 3835 3836
	if (status != 0)
		goto out;
	status = data->rpc_status;
	if (status != 0)
		goto out;
	nfs_refresh_inode(inode, &data->fattr);
out:
3837
	rpc_put_task(task);
3838
	return status;
L
Linus Torvalds 已提交
3839 3840
}

3841
int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
L
Linus Torvalds 已提交
3842 3843 3844 3845 3846
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs4_exception exception = { };
	int err;
	do {
3847
		err = _nfs4_proc_delegreturn(inode, cred, stateid, issync);
L
Linus Torvalds 已提交
3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867
		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)
{
3868
	schedule_timeout_killable(timeout);
L
Linus Torvalds 已提交
3869 3870 3871 3872 3873 3874 3875 3876 3877 3878
	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);
3879
	struct nfs_client *clp = server->nfs_client;
T
Trond Myklebust 已提交
3880
	struct nfs_lockt_args arg = {
L
Linus Torvalds 已提交
3881
		.fh = NFS_FH(inode),
T
Trond Myklebust 已提交
3882
		.fl = request,
L
Linus Torvalds 已提交
3883
	};
T
Trond Myklebust 已提交
3884 3885
	struct nfs_lockt_res res = {
		.denied = request,
L
Linus Torvalds 已提交
3886 3887 3888 3889 3890 3891 3892 3893 3894 3895
	};
	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
3896
	arg.lock_owner.clientid = clp->cl_clientid;
3897 3898 3899 3900
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
	lsp = request->fl_u.nfs4_fl.owner;
3901
	arg.lock_owner.id = lsp->ls_id.id;
3902
	arg.lock_owner.s_dev = server->s_dev;
A
Andy Adamson 已提交
3903
	status = nfs4_call_sync(server, &msg, &arg, &res, 1);
T
Trond Myklebust 已提交
3904 3905 3906 3907 3908 3909
	switch (status) {
		case 0:
			request->fl_type = F_UNLCK;
			break;
		case -NFS4ERR_DENIED:
			status = 0;
L
Linus Torvalds 已提交
3910
	}
3911
	request->fl_ops->fl_release_private(request);
3912
out:
L
Linus Torvalds 已提交
3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944
	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;
}

3945
struct nfs4_unlockdata {
T
Trond Myklebust 已提交
3946 3947
	struct nfs_locku_args arg;
	struct nfs_locku_res res;
3948 3949
	struct nfs4_lock_state *lsp;
	struct nfs_open_context *ctx;
T
Trond Myklebust 已提交
3950 3951
	struct file_lock fl;
	const struct nfs_server *server;
3952
	unsigned long timestamp;
3953 3954
};

T
Trond Myklebust 已提交
3955 3956 3957 3958 3959 3960 3961 3962
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;

3963
	p = kzalloc(sizeof(*p), GFP_NOFS);
T
Trond Myklebust 已提交
3964 3965 3966 3967 3968
	if (p == NULL)
		return NULL;
	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
	p->arg.seqid = seqid;
3969
	p->res.seqid = seqid;
T
Trond Myklebust 已提交
3970 3971 3972 3973 3974 3975 3976 3977 3978 3979
	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;
}

3980
static void nfs4_locku_release_calldata(void *data)
3981
{
3982
	struct nfs4_unlockdata *calldata = data;
T
Trond Myklebust 已提交
3983
	nfs_free_seqid(calldata->arg.seqid);
3984 3985 3986
	nfs4_put_lock_state(calldata->lsp);
	put_nfs_open_context(calldata->ctx);
	kfree(calldata);
3987 3988
}

3989
static void nfs4_locku_done(struct rpc_task *task, void *data)
3990
{
3991
	struct nfs4_unlockdata *calldata = data;
3992

3993 3994
	if (!nfs4_sequence_done(task, &calldata->res.seq_res))
		return;
3995 3996 3997
	switch (task->tk_status) {
		case 0:
			memcpy(calldata->lsp->ls_stateid.data,
T
Trond Myklebust 已提交
3998
					calldata->res.stateid.data,
3999
					sizeof(calldata->lsp->ls_stateid.data));
4000
			renew_lease(calldata->server, calldata->timestamp);
4001
			break;
4002 4003
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OLD_STATEID:
4004 4005 4006 4007
		case -NFS4ERR_STALE_STATEID:
		case -NFS4ERR_EXPIRED:
			break;
		default:
4008
			if (nfs4_async_handle_error(task, calldata->server, NULL) == -EAGAIN)
4009
				nfs_restart_rpc(task,
4010
						 calldata->server->nfs_client);
4011 4012 4013
	}
}

T
Trond Myklebust 已提交
4014
static void nfs4_locku_prepare(struct rpc_task *task, void *data)
4015
{
T
Trond Myklebust 已提交
4016
	struct nfs4_unlockdata *calldata = data;
4017

T
Trond Myklebust 已提交
4018
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
4019 4020
		return;
	if ((calldata->lsp->ls_flags & NFS_LOCK_INITIALIZED) == 0) {
4021 4022
		/* Note: exit _without_ running nfs4_locku_done */
		task->tk_action = NULL;
4023 4024
		return;
	}
4025
	calldata->timestamp = jiffies;
4026
	if (nfs4_setup_sequence(calldata->server,
4027 4028 4029
				&calldata->arg.seq_args,
				&calldata->res.seq_res, 1, task))
		return;
4030
	rpc_call_start(task);
4031 4032
}

4033
static const struct rpc_call_ops nfs4_locku_ops = {
T
Trond Myklebust 已提交
4034
	.rpc_call_prepare = nfs4_locku_prepare,
4035
	.rpc_call_done = nfs4_locku_done,
4036
	.rpc_release = nfs4_locku_release_calldata,
4037 4038
};

4039 4040 4041 4042 4043 4044
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;
4045 4046 4047 4048
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCKU],
		.rpc_cred = ctx->cred,
	};
T
Trond Myklebust 已提交
4049 4050
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(lsp->ls_state->inode),
4051
		.rpc_message = &msg,
T
Trond Myklebust 已提交
4052
		.callback_ops = &nfs4_locku_ops,
4053
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
4054 4055
		.flags = RPC_TASK_ASYNC,
	};
4056

4057 4058 4059 4060 4061
	/* 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;

4062 4063 4064 4065 4066 4067
	data = nfs4_alloc_unlockdata(fl, ctx, lsp, seqid);
	if (data == NULL) {
		nfs_free_seqid(seqid);
		return ERR_PTR(-ENOMEM);
	}

4068 4069
	msg.rpc_argp = &data->arg;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
4070 4071
	task_setup_data.callback_data = data;
	return rpc_run_task(&task_setup_data);
4072 4073
}

4074 4075
static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
{
4076
	struct nfs_inode *nfsi = NFS_I(state->inode);
T
Trond Myklebust 已提交
4077
	struct nfs_seqid *seqid;
L
Linus Torvalds 已提交
4078
	struct nfs4_lock_state *lsp;
4079 4080
	struct rpc_task *task;
	int status = 0;
4081
	unsigned char fl_flags = request->fl_flags;
4082

4083
	status = nfs4_set_lock_state(state, request);
4084 4085
	/* Unlock _before_ we do the RPC call */
	request->fl_flags |= FL_EXISTS;
4086 4087 4088
	down_read(&nfsi->rwsem);
	if (do_vfs_lock(request->fl_file, request) == -ENOENT) {
		up_read(&nfsi->rwsem);
4089
		goto out;
4090 4091
	}
	up_read(&nfsi->rwsem);
4092
	if (status != 0)
4093 4094 4095 4096
		goto out;
	/* Is this a delegated lock? */
	if (test_bit(NFS_DELEGATED_STATE, &state->flags))
		goto out;
4097
	lsp = request->fl_u.nfs4_fl.owner;
4098
	seqid = nfs_alloc_seqid(&lsp->ls_seqid, GFP_KERNEL);
4099
	status = -ENOMEM;
T
Trond Myklebust 已提交
4100
	if (seqid == NULL)
4101
		goto out;
4102
	task = nfs4_do_unlck(request, nfs_file_open_context(request->fl_file), lsp, seqid);
4103 4104
	status = PTR_ERR(task);
	if (IS_ERR(task))
4105
		goto out;
4106
	status = nfs4_wait_for_completion_rpc_task(task);
4107
	rpc_put_task(task);
4108
out:
4109
	request->fl_flags = fl_flags;
L
Linus Torvalds 已提交
4110 4111 4112
	return status;
}

4113 4114 4115 4116 4117 4118
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;
4119
	unsigned long timestamp;
4120 4121
	int rpc_status;
	int cancelled;
4122
	struct nfs_server *server;
4123 4124 4125
};

static struct nfs4_lockdata *nfs4_alloc_lockdata(struct file_lock *fl,
4126 4127
		struct nfs_open_context *ctx, struct nfs4_lock_state *lsp,
		gfp_t gfp_mask)
L
Linus Torvalds 已提交
4128
{
4129 4130
	struct nfs4_lockdata *p;
	struct inode *inode = lsp->ls_state->inode;
L
Linus Torvalds 已提交
4131
	struct nfs_server *server = NFS_SERVER(inode);
4132

4133
	p = kzalloc(sizeof(*p), gfp_mask);
4134 4135 4136 4137 4138
	if (p == NULL)
		return NULL;

	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
4139
	p->arg.open_seqid = nfs_alloc_seqid(&lsp->ls_state->owner->so_seqid, gfp_mask);
4140 4141
	if (p->arg.open_seqid == NULL)
		goto out_free;
4142
	p->arg.lock_seqid = nfs_alloc_seqid(&lsp->ls_seqid, gfp_mask);
4143
	if (p->arg.lock_seqid == NULL)
4144
		goto out_free_seqid;
4145
	p->arg.lock_stateid = &lsp->ls_stateid;
4146
	p->arg.lock_owner.clientid = server->nfs_client->cl_clientid;
4147
	p->arg.lock_owner.id = lsp->ls_id.id;
4148
	p->arg.lock_owner.s_dev = server->s_dev;
4149
	p->res.lock_seqid = p->arg.lock_seqid;
4150
	p->lsp = lsp;
4151
	p->server = server;
4152 4153 4154 4155
	atomic_inc(&lsp->ls_count);
	p->ctx = get_nfs_open_context(ctx);
	memcpy(&p->fl, fl, sizeof(p->fl));
	return p;
4156 4157
out_free_seqid:
	nfs_free_seqid(p->arg.open_seqid);
4158 4159 4160 4161 4162 4163 4164 4165 4166
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;
4167

4168
	dprintk("%s: begin!\n", __func__);
4169 4170
	if (nfs_wait_on_sequence(data->arg.lock_seqid, task) != 0)
		return;
4171 4172
	/* Do we need to do an open_to_lock_owner? */
	if (!(data->arg.lock_seqid->sequence->flags & NFS_SEQID_CONFIRMED)) {
4173 4174
		if (nfs_wait_on_sequence(data->arg.open_seqid, task) != 0)
			return;
4175 4176
		data->arg.open_stateid = &state->stateid;
		data->arg.new_lock_owner = 1;
4177
		data->res.open_seqid = data->arg.open_seqid;
4178 4179
	} else
		data->arg.new_lock_owner = 0;
4180
	data->timestamp = jiffies;
4181 4182
	if (nfs4_setup_sequence(data->server,
				&data->arg.seq_args,
4183 4184
				&data->res.seq_res, 1, task))
		return;
4185
	rpc_call_start(task);
4186
	dprintk("%s: done!, ret = %d\n", __func__, data->rpc_status);
4187 4188
}

4189 4190 4191 4192 4193 4194
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);
}

4195 4196 4197 4198
static void nfs4_lock_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_lockdata *data = calldata;

4199
	dprintk("%s: begin!\n", __func__);
4200

4201 4202
	if (!nfs4_sequence_done(task, &data->res.seq_res))
		return;
4203

4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214
	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;
4215
		renew_lease(NFS_SERVER(data->ctx->path.dentry->d_inode), data->timestamp);
4216 4217
	}
out:
4218
	dprintk("%s: done, ret = %d!\n", __func__, data->rpc_status);
4219 4220 4221 4222 4223 4224
}

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

4225
	dprintk("%s: begin!\n", __func__);
4226
	nfs_free_seqid(data->arg.open_seqid);
4227 4228 4229 4230 4231
	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))
4232
			rpc_put_task_async(task);
4233
		dprintk("%s: cancelling lock!\n", __func__);
4234 4235 4236 4237 4238
	} else
		nfs_free_seqid(data->arg.lock_seqid);
	nfs4_put_lock_state(data->lsp);
	put_nfs_open_context(data->ctx);
	kfree(data);
4239
	dprintk("%s: done!\n", __func__);
4240 4241 4242 4243 4244 4245 4246 4247
}

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,
};

4248 4249 4250 4251 4252 4253
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,
};

4254 4255 4256 4257 4258
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:
4259
		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
4260 4261
		if (new_lock_owner != 0 ||
		   (lsp->ls_flags & NFS_LOCK_INITIALIZED) != 0)
4262
			nfs4_schedule_stateid_recovery(server, lsp->ls_state);
4263 4264 4265
		break;
	case -NFS4ERR_STALE_STATEID:
		lsp->ls_seqid.flags &= ~NFS_SEQID_CONFIRMED;
4266 4267
	case -NFS4ERR_EXPIRED:
		nfs4_schedule_lease_recovery(server->nfs_client);
4268 4269 4270
	};
}

4271
static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *fl, int recovery_type)
4272 4273 4274
{
	struct nfs4_lockdata *data;
	struct rpc_task *task;
4275 4276 4277 4278
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCK],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
4279 4280
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(state->inode),
4281
		.rpc_message = &msg,
T
Trond Myklebust 已提交
4282
		.callback_ops = &nfs4_lock_ops,
4283
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
4284 4285
		.flags = RPC_TASK_ASYNC,
	};
4286 4287
	int ret;

4288
	dprintk("%s: begin!\n", __func__);
4289
	data = nfs4_alloc_lockdata(fl, nfs_file_open_context(fl->fl_file),
4290 4291
			fl->fl_u.nfs4_fl.owner,
			recovery_type == NFS_LOCK_NEW ? GFP_KERNEL : GFP_NOFS);
4292 4293 4294 4295
	if (data == NULL)
		return -ENOMEM;
	if (IS_SETLKW(cmd))
		data->arg.block = 1;
4296 4297
	if (recovery_type > NFS_LOCK_NEW) {
		if (recovery_type == NFS_LOCK_RECLAIM)
4298
			data->arg.reclaim = NFS_LOCK_RECLAIM;
4299 4300
		task_setup_data.callback_ops = &nfs4_recover_lock_ops;
	}
4301 4302
	msg.rpc_argp = &data->arg;
	msg.rpc_resp = &data->res;
T
Trond Myklebust 已提交
4303 4304
	task_setup_data.callback_data = data;
	task = rpc_run_task(&task_setup_data);
4305
	if (IS_ERR(task))
4306 4307 4308 4309
		return PTR_ERR(task);
	ret = nfs4_wait_for_completion_rpc_task(task);
	if (ret == 0) {
		ret = data->rpc_status;
4310 4311 4312
		if (ret)
			nfs4_handle_setlk_error(data->server, data->lsp,
					data->arg.new_lock_owner, ret);
4313 4314
	} else
		data->cancelled = 1;
4315
	rpc_put_task(task);
4316
	dprintk("%s: done, ret = %d!\n", __func__, ret);
4317
	return ret;
L
Linus Torvalds 已提交
4318 4319 4320 4321
}

static int nfs4_lock_reclaim(struct nfs4_state *state, struct file_lock *request)
{
4322 4323 4324 4325 4326
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;

	do {
4327 4328 4329
		/* Cache the lock if possible... */
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
4330
		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_RECLAIM);
4331
		if (err != -NFS4ERR_DELAY)
4332 4333 4334 4335
			break;
		nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
	return err;
L
Linus Torvalds 已提交
4336 4337 4338 4339
}

static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request)
{
4340 4341 4342 4343
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;

4344 4345 4346
	err = nfs4_set_lock_state(state, request);
	if (err != 0)
		return err;
4347
	do {
4348 4349
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
4350
		err = _nfs4_do_setlk(state, F_SETLK, request, NFS_LOCK_EXPIRED);
4351 4352 4353 4354 4355 4356 4357 4358
		switch (err) {
		default:
			goto out;
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			nfs4_handle_exception(server, err, &exception);
			err = 0;
		}
4359
	} while (exception.retry);
4360
out:
4361
	return err;
L
Linus Torvalds 已提交
4362 4363 4364 4365
}

static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
4366
	struct nfs_inode *nfsi = NFS_I(state->inode);
4367
	unsigned char fl_flags = request->fl_flags;
4368
	int status = -ENOLCK;
L
Linus Torvalds 已提交
4369

4370 4371 4372
	if ((fl_flags & FL_POSIX) &&
			!test_bit(NFS_STATE_POSIX_LOCKS, &state->flags))
		goto out;
4373 4374 4375 4376
	/* Is this a delegated open? */
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
4377 4378 4379 4380
	request->fl_flags |= FL_ACCESS;
	status = do_vfs_lock(request->fl_file, request);
	if (status < 0)
		goto out;
4381
	down_read(&nfsi->rwsem);
4382 4383 4384
	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
		/* Yes: cache locks! */
		/* ...but avoid races with delegation recall... */
4385 4386 4387
		request->fl_flags = fl_flags & ~FL_SLEEP;
		status = do_vfs_lock(request->fl_file, request);
		goto out_unlock;
4388
	}
4389
	status = _nfs4_do_setlk(state, cmd, request, NFS_LOCK_NEW);
4390
	if (status != 0)
4391
		goto out_unlock;
4392
	/* Note: we always want to sleep here! */
4393
	request->fl_flags = fl_flags | FL_SLEEP;
4394
	if (do_vfs_lock(request->fl_file, request) < 0)
4395
		printk(KERN_WARNING "%s: VFS is out of sync with lock manager!\n", __func__);
4396
out_unlock:
4397
	up_read(&nfsi->rwsem);
4398 4399
out:
	request->fl_flags = fl_flags;
L
Linus Torvalds 已提交
4400 4401 4402 4403 4404 4405 4406 4407 4408
	return status;
}

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

	do {
4409 4410 4411
		err = _nfs4_proc_setlk(state, cmd, request);
		if (err == -NFS4ERR_DENIED)
			err = -EAGAIN;
L
Linus Torvalds 已提交
4412
		err = nfs4_handle_exception(NFS_SERVER(state->inode),
4413
				err, &exception);
L
Linus Torvalds 已提交
4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426
	} 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 */
4427
	ctx = nfs_file_open_context(filp);
L
Linus Torvalds 已提交
4428 4429 4430 4431 4432
	state = ctx->state;

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

4433 4434 4435 4436 4437
	if (IS_GETLK(cmd)) {
		if (state != NULL)
			return nfs4_proc_getlk(state, F_GETLK, request);
		return 0;
	}
L
Linus Torvalds 已提交
4438 4439 4440 4441

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

4442 4443 4444 4445 4446
	if (request->fl_type == F_UNLCK) {
		if (state != NULL)
			return nfs4_proc_unlck(state, cmd, request);
		return 0;
	}
L
Linus Torvalds 已提交
4447

4448 4449
	if (state == NULL)
		return -ENOLCK;
L
Linus Torvalds 已提交
4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461
	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;
}

4462 4463 4464 4465 4466 4467 4468 4469 4470 4471
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 {
4472
		err = _nfs4_do_setlk(state, F_SETLK, fl, NFS_LOCK_NEW);
4473 4474 4475 4476 4477
		switch (err) {
			default:
				printk(KERN_ERR "%s: unhandled error %d.\n",
						__func__, err);
			case 0:
4478
			case -ESTALE:
4479 4480 4481
				goto out;
			case -NFS4ERR_EXPIRED:
			case -NFS4ERR_STALE_CLIENTID:
4482
			case -NFS4ERR_STALE_STATEID:
4483 4484
				nfs4_schedule_lease_recovery(server->nfs_client);
				goto out;
4485 4486 4487 4488 4489
			case -NFS4ERR_BADSESSION:
			case -NFS4ERR_BADSLOT:
			case -NFS4ERR_BAD_HIGH_SLOT:
			case -NFS4ERR_CONN_NOT_BOUND_TO_SESSION:
			case -NFS4ERR_DEADSESSION:
4490
				nfs4_schedule_session_recovery(server->nfs_client->cl_session);
4491
				goto out;
4492 4493 4494 4495 4496 4497 4498 4499
			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:
4500
				nfs4_schedule_stateid_recovery(server, state);
4501 4502
				err = 0;
				goto out;
4503 4504 4505 4506 4507 4508 4509 4510 4511
			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;
4512 4513 4514 4515 4516
			case -ENOMEM:
			case -NFS4ERR_DENIED:
				/* kill_proc(fl->fl_pid, SIGLOST, 1); */
				err = 0;
				goto out;
4517 4518 4519
			case -NFS4ERR_DELAY:
				break;
		}
4520 4521 4522 4523 4524
		err = nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
out:
	return err;
}
4525

4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549
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;
4550
	args->lock_owner.s_dev = server->s_dev;
4551 4552 4553 4554
	msg.rpc_argp = args;
	rpc_call_async(server->client, &msg, 0, &nfs4_release_lockowner_ops, args);
}

4555 4556
#define XATTR_NAME_NFSV4_ACL "system.nfs4_acl"

4557 4558 4559
static int nfs4_xattr_set_nfs4_acl(struct dentry *dentry, const char *key,
				   const void *buf, size_t buflen,
				   int flags, int type)
4560
{
4561 4562
	if (strcmp(key, "") != 0)
		return -EINVAL;
4563

4564
	return nfs4_proc_set_acl(dentry->d_inode, buf, buflen);
4565 4566
}

4567 4568
static int nfs4_xattr_get_nfs4_acl(struct dentry *dentry, const char *key,
				   void *buf, size_t buflen, int type)
4569
{
4570 4571
	if (strcmp(key, "") != 0)
		return -EINVAL;
4572

4573
	return nfs4_proc_get_acl(dentry->d_inode, buf, buflen);
4574 4575
}

4576 4577 4578
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)
4579
{
4580
	size_t len = sizeof(XATTR_NAME_NFSV4_ACL);
4581

4582 4583
	if (!nfs4_server_supports_acls(NFS_SERVER(dentry->d_inode)))
		return 0;
4584 4585 4586

	if (list && len <= list_len)
		memcpy(list, XATTR_NAME_NFSV4_ACL, len);
4587
	return len;
4588 4589
}

4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602
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;
}

4603
int nfs4_proc_fs_locations(struct inode *dir, const struct qstr *name,
4604
		struct nfs4_fs_locations *fs_locations, struct page *page)
4605 4606 4607
{
	struct nfs_server *server = NFS_SERVER(dir);
	u32 bitmask[2] = {
4608 4609
		[0] = FATTR4_WORD0_FSID | FATTR4_WORD0_FS_LOCATIONS,
		[1] = FATTR4_WORD1_MOUNTED_ON_FILEID,
4610 4611 4612
	};
	struct nfs4_fs_locations_arg args = {
		.dir_fh = NFS_FH(dir),
4613
		.name = name,
4614 4615 4616
		.page = page,
		.bitmask = bitmask,
	};
B
Benny Halevy 已提交
4617 4618 4619
	struct nfs4_fs_locations_res res = {
		.fs_locations = fs_locations,
	};
4620 4621 4622
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FS_LOCATIONS],
		.rpc_argp = &args,
B
Benny Halevy 已提交
4623
		.rpc_resp = &res,
4624 4625 4626
	};
	int status;

4627
	dprintk("%s: start\n", __func__);
4628
	nfs_fattr_init(&fs_locations->fattr);
4629
	fs_locations->server = server;
4630
	fs_locations->nlocations = 0;
A
Andy Adamson 已提交
4631
	status = nfs4_call_sync(server, &msg, &args, &res, 0);
4632
	nfs_fixup_referral_attributes(&fs_locations->fattr);
4633
	dprintk("%s: returned status = %d\n", __func__, status);
4634 4635 4636
	return status;
}

4637
#ifdef CONFIG_NFS_V4_1
4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656
/*
 * 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
4657 4658 4659 4660 4661 4662 4663 4664
/*
 * 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.
 */
4665
int nfs4_proc_exchange_id(struct nfs_client *clp, struct rpc_cred *cred)
B
Benny Halevy 已提交
4666 4667 4668 4669
{
	nfs4_verifier verifier;
	struct nfs41_exchange_id_args args = {
		.client = clp,
4670
		.flags = EXCHGID4_FLAG_SUPP_MOVED_REFER,
B
Benny Halevy 已提交
4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685
	};
	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);
4686

B
Benny Halevy 已提交
4687 4688 4689 4690 4691
	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;

4692 4693 4694 4695 4696 4697
	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
4698

4699 4700 4701
	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
4702 4703 4704 4705
	dprintk("<-- %s status= %d\n", __func__, status);
	return status;
}

A
Andy Adamson 已提交
4706 4707 4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719
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__);
4720
	rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
A
Andy Adamson 已提交
4721 4722 4723
	/* just setup sequence, do not trigger session recovery
	   since we're invoked within one */
	ret = nfs41_setup_sequence(data->clp->cl_session,
4724 4725
				   &data->args->la_seq_args,
				   &data->res->lr_seq_res, 0, task);
A
Andy Adamson 已提交
4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741

	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__);
4742 4743
	if (!nfs41_sequence_done(task, &data->res->lr_seq_res))
		return;
A
Andy Adamson 已提交
4744 4745 4746 4747 4748 4749
	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;
4750
		nfs_restart_rpc(task, data->clp);
A
Andy Adamson 已提交
4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799
		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;
}

4800 4801 4802
/*
 * Reset a slot table
 */
A
Andy Adamson 已提交
4803 4804
static int nfs4_reset_slot_table(struct nfs4_slot_table *tbl, u32 max_reqs,
				 int ivalue)
4805
{
A
Andy Adamson 已提交
4806
	struct nfs4_slot *new = NULL;
4807
	int i;
4808 4809
	int ret = 0;

A
Andy Adamson 已提交
4810 4811
	dprintk("--> %s: max_reqs=%u, tbl->max_slots %d\n", __func__,
		max_reqs, tbl->max_slots);
4812

A
Andy Adamson 已提交
4813 4814 4815 4816
	/* 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),
4817
			      GFP_NOFS);
A
Andy Adamson 已提交
4818 4819 4820 4821
		if (!new)
			goto out;
		ret = 0;
		kfree(tbl->slots);
4822 4823
	}
	spin_lock(&tbl->slot_tbl_lock);
A
Andy Adamson 已提交
4824 4825 4826 4827 4828
	if (new) {
		tbl->slots = new;
		tbl->max_slots = max_reqs;
	}
	for (i = 0; i < tbl->max_slots; ++i)
4829
		tbl->slots[i].seq_nr = ivalue;
4830 4831 4832 4833 4834 4835 4836 4837
	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;
}

4838 4839 4840 4841 4842 4843 4844 4845
/*
 * 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
4846
			session->fc_attrs.max_reqs, 1);
4847 4848 4849 4850
	if (status)
		return status;

	status = nfs4_reset_slot_table(&session->bc_slot_table,
A
Andy Adamson 已提交
4851
			session->bc_attrs.max_reqs, 0);
4852 4853 4854
	return status;
}

4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868
/* 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
4869 4870 4871
/*
 * Initialize slot table
 */
4872 4873
static int nfs4_init_slot_table(struct nfs4_slot_table *tbl,
		int max_slots, int ivalue)
A
Andy Adamson 已提交
4874 4875 4876 4877 4878 4879
{
	struct nfs4_slot *slot;
	int ret = -ENOMEM;

	BUG_ON(max_slots > NFS4_MAX_SLOT_TABLE);

4880
	dprintk("--> %s: max_reqs=%u\n", __func__, max_slots);
A
Andy Adamson 已提交
4881

4882
	slot = kcalloc(max_slots, sizeof(struct nfs4_slot), GFP_NOFS);
A
Andy Adamson 已提交
4883 4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898
	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;
}

4899 4900 4901 4902 4903
/*
 * Initialize the forechannel and backchannel tables
 */
static int nfs4_init_slot_tables(struct nfs4_session *session)
{
4904 4905
	struct nfs4_slot_table *tbl;
	int status = 0;
4906

4907 4908 4909 4910 4911 4912 4913
	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;
	}
4914

4915 4916 4917 4918 4919 4920 4921
	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);
	}
4922 4923

	return status;
4924 4925 4926 4927 4928 4929 4930
}

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

4931
	session = kzalloc(sizeof(struct nfs4_session), GFP_NOFS);
4932 4933
	if (!session)
		return NULL;
4934

4935
	tbl = &session->fc_slot_table;
4936
	tbl->highest_used_slotid = -1;
4937
	spin_lock_init(&tbl->slot_tbl_lock);
4938
	rpc_init_priority_wait_queue(&tbl->slot_tbl_waitq, "ForeChannel Slot table");
4939
	init_completion(&tbl->complete);
4940 4941

	tbl = &session->bc_slot_table;
4942
	tbl->highest_used_slotid = -1;
4943 4944
	spin_lock_init(&tbl->slot_tbl_lock);
	rpc_init_wait_queue(&tbl->slot_tbl_waitq, "BackChannel Slot table");
4945
	init_completion(&tbl->complete);
4946

4947 4948
	session->session_state = 1<<NFS4_SESSION_INITING;

4949 4950 4951 4952 4953 4954
	session->clp = clp;
	return session;
}

void nfs4_destroy_session(struct nfs4_session *session)
{
4955 4956 4957 4958 4959
	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);
4960
	nfs4_destroy_slot_tables(session);
4961 4962 4963
	kfree(session);
}

A
Andy Adamson 已提交
4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990
/*
 * 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 "
4991
		"max_ops=%u max_reqs=%u\n",
A
Andy Adamson 已提交
4992 4993
		__func__,
		args->fc_attrs.max_rqst_sz, args->fc_attrs.max_resp_sz,
4994
		args->fc_attrs.max_ops, args->fc_attrs.max_reqs);
A
Andy Adamson 已提交
4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011

	/* 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);
}

5012
static int nfs4_verify_fore_channel_attrs(struct nfs41_create_session_args *args, struct nfs4_session *session)
5013
{
5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031
	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;
5032 5033
}

5034 5035 5036 5037
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;
5038

5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051
	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;
}
5052 5053 5054 5055

static int nfs4_verify_channel_attrs(struct nfs41_create_session_args *args,
				     struct nfs4_session *session)
{
5056
	int ret;
5057

5058 5059 5060 5061
	ret = nfs4_verify_fore_channel_attrs(args, session);
	if (ret)
		return ret;
	return nfs4_verify_back_channel_attrs(args, session);
5062 5063
}

A
Andy Adamson 已提交
5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081
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);
5082
	args.flags = (SESSION4_PERSIST | SESSION4_BACK_CHAN);
A
Andy Adamson 已提交
5083 5084 5085

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

5086 5087 5088
	if (!status)
		/* Verify the session's negotiated channel_attrs values */
		status = nfs4_verify_channel_attrs(&args, session);
A
Andy Adamson 已提交
5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101
	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.
 */
5102
int nfs4_proc_create_session(struct nfs_client *clp)
A
Andy Adamson 已提交
5103 5104 5105 5106
{
	int status;
	unsigned *ptr;
	struct nfs4_session *session = clp->cl_session;
5107 5108
	long timeout = 0;
	int err;
A
Andy Adamson 已提交
5109 5110 5111

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

5112 5113 5114 5115 5116 5117 5118 5119 5120
	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
5121 5122 5123
	if (status)
		goto out;

5124 5125 5126 5127 5128 5129 5130
	/* 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
5131 5132 5133 5134 5135 5136 5137 5138 5139 5140 5141
	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
5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171
/*
 * 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;
}

5172 5173 5174
int nfs4_init_session(struct nfs_server *server)
{
	struct nfs_client *clp = server->nfs_client;
5175
	struct nfs4_session *session;
5176
	unsigned int rsize, wsize;
5177 5178 5179 5180 5181
	int ret;

	if (!nfs4_has_session(clp))
		return 0;

5182 5183 5184 5185
	session = clp->cl_session;
	if (!test_and_clear_bit(NFS4_SESSION_INITING, &session->session_state))
		return 0;

5186 5187 5188 5189 5190 5191 5192 5193 5194
	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;
5195

5196 5197 5198 5199 5200 5201
	ret = nfs4_recover_expired_lease(server);
	if (!ret)
		ret = nfs4_check_client_ready(clp);
	return ret;
}

A
Andy Adamson 已提交
5202 5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215 5216 5217 5218 5219 5220 5221 5222
int nfs4_init_ds_session(struct nfs_client *clp)
{
	struct nfs4_session *session = clp->cl_session;
	int ret;

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

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

}
EXPORT_SYMBOL_GPL(nfs4_init_ds_session);


A
Andy Adamson 已提交
5223 5224 5225
/*
 * Renew the cl_session lease.
 */
5226 5227 5228 5229 5230 5231
struct nfs4_sequence_data {
	struct nfs_client *clp;
	struct nfs4_sequence_args args;
	struct nfs4_sequence_res res;
};

5232 5233
static void nfs41_sequence_release(void *data)
{
5234 5235
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
5236

5237 5238 5239
	if (atomic_read(&clp->cl_count) > 1)
		nfs4_schedule_state_renewal(clp);
	nfs_put_client(clp);
5240
	kfree(calldata);
5241 5242
}

5243 5244 5245 5246 5247 5248 5249
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:
5250
		nfs4_schedule_lease_recovery(clp);
5251 5252 5253 5254
	}
	return 0;
}

5255
static void nfs41_sequence_call_done(struct rpc_task *task, void *data)
A
Andy Adamson 已提交
5256
{
5257 5258
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
A
Andy Adamson 已提交
5259

5260 5261
	if (!nfs41_sequence_done(task, task->tk_msg.rpc_resp))
		return;
A
Andy Adamson 已提交
5262 5263 5264

	if (task->tk_status < 0) {
		dprintk("%s ERROR %d\n", __func__, task->tk_status);
5265 5266
		if (atomic_read(&clp->cl_count) == 1)
			goto out;
A
Andy Adamson 已提交
5267

5268 5269
		if (nfs41_sequence_handle_errors(task, clp) == -EAGAIN) {
			rpc_restart_call_prepare(task);
A
Andy Adamson 已提交
5270 5271 5272 5273
			return;
		}
	}
	dprintk("%s rpc_cred %p\n", __func__, task->tk_msg.rpc_cred);
5274
out:
A
Andy Adamson 已提交
5275 5276 5277 5278 5279
	dprintk("<-- %s\n", __func__);
}

static void nfs41_sequence_prepare(struct rpc_task *task, void *data)
{
5280 5281
	struct nfs4_sequence_data *calldata = data;
	struct nfs_client *clp = calldata->clp;
A
Andy Adamson 已提交
5282 5283 5284 5285 5286 5287
	struct nfs4_sequence_args *args;
	struct nfs4_sequence_res *res;

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

5288
	if (nfs41_setup_sequence(clp->cl_session, args, res, 0, task))
A
Andy Adamson 已提交
5289 5290 5291 5292 5293 5294 5295
		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,
5296
	.rpc_release = nfs41_sequence_release,
A
Andy Adamson 已提交
5297 5298
};

5299
static struct rpc_task *_nfs41_proc_sequence(struct nfs_client *clp, struct rpc_cred *cred)
A
Andy Adamson 已提交
5300
{
5301
	struct nfs4_sequence_data *calldata;
A
Andy Adamson 已提交
5302 5303 5304 5305
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SEQUENCE],
		.rpc_cred = cred,
	};
5306 5307 5308 5309 5310 5311
	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
5312

5313
	if (!atomic_inc_not_zero(&clp->cl_count))
5314
		return ERR_PTR(-EIO);
5315
	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
5316
	if (calldata == NULL) {
5317
		nfs_put_client(clp);
5318
		return ERR_PTR(-ENOMEM);
A
Andy Adamson 已提交
5319
	}
5320 5321 5322
	msg.rpc_argp = &calldata->args;
	msg.rpc_resp = &calldata->res;
	calldata->clp = clp;
5323
	task_setup_data.callback_data = calldata;
A
Andy Adamson 已提交
5324

5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336
	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
5337
		rpc_put_task_async(task);
5338 5339 5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352
	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);
5353 5354 5355 5356 5357
	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);
5358
		ret = task->tk_status;
5359
	}
5360 5361 5362 5363
	rpc_put_task(task);
out:
	dprintk("<-- %s status=%d\n", __func__, ret);
	return ret;
A
Andy Adamson 已提交
5364 5365
}

5366 5367 5368 5369 5370 5371 5372 5373 5374 5375
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;

5376
	rpc_task_set_priority(task, RPC_PRIORITY_PRIVILEGED);
5377 5378
	if (nfs41_setup_sequence(calldata->clp->cl_session,
				&calldata->arg.seq_args,
5379 5380 5381 5382 5383 5384
				&calldata->res.seq_res, 0, task))
		return;

	rpc_call_start(task);
}

5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395
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:
5396
		nfs4_schedule_lease_recovery(clp);
5397 5398 5399 5400
	}
	return 0;
}

5401 5402 5403 5404 5405 5406 5407
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__);
5408 5409
	if (!nfs41_sequence_done(task, res))
		return;
5410

5411 5412 5413 5414
	if (nfs41_reclaim_complete_handle_errors(task, clp) == -EAGAIN) {
		rpc_restart_call_prepare(task);
		return;
	}
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
	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__);
5450
	calldata = kzalloc(sizeof(*calldata), GFP_NOFS);
5451 5452 5453 5454 5455 5456 5457 5458 5459
	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);
5460
	if (IS_ERR(task)) {
5461
		status = PTR_ERR(task);
5462 5463
		goto out;
	}
5464 5465 5466
	status = nfs4_wait_for_completion_rpc_task(task);
	if (status == 0)
		status = task->tk_status;
5467
	rpc_put_task(task);
5468
	return 0;
5469 5470 5471 5472
out:
	dprintk("<-- %s status=%d\n", __func__, status);
	return status;
}
5473 5474 5475 5476 5477

static void
nfs4_layoutget_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_layoutget *lgp = calldata;
F
Fred Isaman 已提交
5478
	struct nfs_server *server = NFS_SERVER(lgp->args.inode);
5479 5480

	dprintk("--> %s\n", __func__);
F
Fred Isaman 已提交
5481 5482 5483 5484 5485
	/* 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.
	 */
5486 5487 5488
	if (nfs4_setup_sequence(server, &lgp->args.seq_args,
				&lgp->res.seq_res, 0, task))
		return;
5489 5490 5491 5492 5493 5494
	if (pnfs_choose_layoutget_stateid(&lgp->args.stateid,
					  NFS_I(lgp->args.inode)->layout,
					  lgp->args.ctx->state)) {
		rpc_exit(task, NFS4_OK);
		return;
	}
5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572
	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
5573 5574 5575 5576
	if (status == 0)
		status = task->tk_status;
	if (status == 0)
		status = pnfs_layout_process(lgp);
5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618
	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);

5619 5620
#endif /* CONFIG_NFS_V4_1 */

5621
struct nfs4_state_recovery_ops nfs40_reboot_recovery_ops = {
5622
	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
5623
	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
L
Linus Torvalds 已提交
5624 5625
	.recover_open	= nfs4_open_reclaim,
	.recover_lock	= nfs4_lock_reclaim,
5626
	.establish_clid = nfs4_init_clientid,
5627
	.get_clid_cred	= nfs4_get_setclientid_cred,
L
Linus Torvalds 已提交
5628 5629
};

5630 5631 5632 5633 5634 5635
#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,
5636
	.establish_clid = nfs41_init_clientid,
5637
	.get_clid_cred	= nfs4_get_exchange_id_cred,
5638
	.reclaim_complete = nfs41_proc_reclaim_complete,
5639 5640 5641 5642 5643 5644 5645 5646 5647
};
#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,
5648
	.get_clid_cred	= nfs4_get_setclientid_cred,
5649 5650 5651 5652
};

#if defined(CONFIG_NFS_V4_1)
struct nfs4_state_recovery_ops nfs41_nograce_recovery_ops = {
5653
	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
5654
	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
L
Linus Torvalds 已提交
5655 5656
	.recover_open	= nfs4_open_expired,
	.recover_lock	= nfs4_lock_expired,
5657
	.establish_clid = nfs41_init_clientid,
5658
	.get_clid_cred	= nfs4_get_exchange_id_cred,
L
Linus Torvalds 已提交
5659
};
5660
#endif /* CONFIG_NFS_V4_1 */
L
Linus Torvalds 已提交
5661

B
Benny Halevy 已提交
5662 5663
struct nfs4_state_maintenance_ops nfs40_state_renewal_ops = {
	.sched_state_renewal = nfs4_proc_async_renew,
5664
	.get_state_renewal_cred_locked = nfs4_get_renew_cred_locked,
5665
	.renew_lease = nfs4_proc_renew,
B
Benny Halevy 已提交
5666 5667 5668 5669 5670
};

#if defined(CONFIG_NFS_V4_1)
struct nfs4_state_maintenance_ops nfs41_state_renewal_ops = {
	.sched_state_renewal = nfs41_proc_async_sequence,
5671
	.get_state_renewal_cred_locked = nfs4_get_machine_cred_locked,
5672
	.renew_lease = nfs4_proc_sequence,
B
Benny Halevy 已提交
5673 5674 5675
};
#endif

5676 5677 5678
static const struct nfs4_minor_version_ops nfs_v4_0_minor_ops = {
	.minor_version = 0,
	.call_sync = _nfs4_call_sync,
5679
	.validate_stateid = nfs4_validate_delegation_stateid,
5680 5681 5682
	.reboot_recovery_ops = &nfs40_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs40_nograce_recovery_ops,
	.state_renewal_ops = &nfs40_state_renewal_ops,
5683 5684 5685 5686 5687 5688
};

#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,
5689
	.validate_stateid = nfs41_validate_delegation_stateid,
5690 5691 5692
	.reboot_recovery_ops = &nfs41_reboot_recovery_ops,
	.nograce_recovery_ops = &nfs41_nograce_recovery_ops,
	.state_renewal_ops = &nfs41_state_renewal_ops,
5693 5694 5695 5696 5697 5698 5699 5700 5701 5702
};
#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
};

5703
static const struct inode_operations nfs4_file_inode_operations = {
5704 5705 5706
	.permission	= nfs_permission,
	.getattr	= nfs_getattr,
	.setattr	= nfs_setattr,
5707 5708 5709 5710
	.getxattr	= generic_getxattr,
	.setxattr	= generic_setxattr,
	.listxattr	= generic_listxattr,
	.removexattr	= generic_removexattr,
5711 5712
};

D
David Howells 已提交
5713
const struct nfs_rpc_ops nfs_v4_clientops = {
L
Linus Torvalds 已提交
5714 5715 5716
	.version	= 4,			/* protocol version */
	.dentry_ops	= &nfs4_dentry_operations,
	.dir_inode_ops	= &nfs4_dir_inode_operations,
5717
	.file_inode_ops	= &nfs4_file_inode_operations,
L
Linus Torvalds 已提交
5718 5719 5720
	.getroot	= nfs4_proc_get_root,
	.getattr	= nfs4_proc_getattr,
	.setattr	= nfs4_proc_setattr,
D
David Howells 已提交
5721
	.lookupfh	= nfs4_proc_lookupfh,
L
Linus Torvalds 已提交
5722 5723 5724 5725 5726 5727 5728 5729
	.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,
5730 5731
	.rename_setup	= nfs4_proc_rename_setup,
	.rename_done	= nfs4_proc_rename_done,
L
Linus Torvalds 已提交
5732 5733 5734 5735 5736 5737 5738 5739 5740
	.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,
5741
	.set_capabilities = nfs4_server_capabilities,
L
Linus Torvalds 已提交
5742 5743
	.decode_dirent	= nfs4_decode_dirent,
	.read_setup	= nfs4_proc_read_setup,
T
Trond Myklebust 已提交
5744
	.read_done	= nfs4_read_done,
L
Linus Torvalds 已提交
5745
	.write_setup	= nfs4_proc_write_setup,
5746
	.write_done	= nfs4_write_done,
L
Linus Torvalds 已提交
5747
	.commit_setup	= nfs4_proc_commit_setup,
5748
	.commit_done	= nfs4_commit_done,
L
Linus Torvalds 已提交
5749
	.lock		= nfs4_proc_lock,
5750
	.clear_acl_cache = nfs4_zap_acl_attr,
T
Trond Myklebust 已提交
5751
	.close_context  = nfs4_close_context,
5752
	.open_context	= nfs4_atomic_open,
5753
	.init_client	= nfs4_init_client,
L
Linus Torvalds 已提交
5754 5755
};

5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766 5767
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
5768 5769 5770 5771 5772
/*
 * Local variables:
 *  c-basic-offset: 8
 * End:
 */