nfs4proc.c 101.1 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49
/*
 *  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/utsname.h>
#include <linux/delay.h>
#include <linux/errno.h>
#include <linux/string.h>
#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/smp_lock.h>
#include <linux/namei.h>
50
#include <linux/mount.h>
L
Linus Torvalds 已提交
51

52
#include "nfs4_fs.h"
L
Linus Torvalds 已提交
53
#include "delegation.h"
54
#include "internal.h"
55
#include "iostat.h"
L
Linus Torvalds 已提交
56 57 58

#define NFSDBG_FACILITY		NFSDBG_PROC

59
#define NFS4_POLL_RETRY_MIN	(HZ/10)
L
Linus Torvalds 已提交
60 61
#define NFS4_POLL_RETRY_MAX	(15*HZ)

62
struct nfs4_opendata;
63
static int _nfs4_proc_open(struct nfs4_opendata *data);
L
Linus Torvalds 已提交
64
static int nfs4_do_fsinfo(struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *);
65
static int nfs4_async_handle_error(struct rpc_task *, const struct nfs_server *, struct nfs4_state *);
66 67
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);
L
Linus Torvalds 已提交
68 69

/* Prevent leaks of NFSv4 errors into userland */
70
static int nfs4_map_errors(int err)
L
Linus Torvalds 已提交
71 72 73
{
	if (err < -1000) {
		dprintk("%s could not handle NFSv4 error %d\n",
74
				__func__, -err);
L
Linus Torvalds 已提交
75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108
		return -EIO;
	}
	return err;
}

/*
 * 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
};

109
const u32 nfs4_pathconf_bitmap[2] = {
L
Linus Torvalds 已提交
110 111 112 113 114 115 116 117 118 119 120 121
	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,
			0
};

122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140
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
141
static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dentry,
L
Linus Torvalds 已提交
142 143
		struct nfs4_readdir_arg *readdir)
{
144
	__be32 *start, *p;
L
Linus Torvalds 已提交
145 146 147

	BUG_ON(readdir->count < 80);
	if (cookie > 2) {
148
		readdir->cookie = cookie;
L
Linus Torvalds 已提交
149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164
		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.
	 */
165
	start = p = kmap_atomic(*readdir->pages, KM_USER0);
L
Linus Torvalds 已提交
166 167 168 169 170 171 172 173 174 175 176
	
	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
177
		p = xdr_encode_hyper(p, NFS_FILEID(dentry->d_inode));
L
Linus Torvalds 已提交
178 179 180 181 182 183 184 185 186 187 188
	}
	
	*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
189
	p = xdr_encode_hyper(p, NFS_FILEID(dentry->d_parent->d_inode));
L
Linus Torvalds 已提交
190 191 192 193 194 195

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

196 197 198 199 200 201
static int nfs4_wait_clnt_recover(struct nfs_client *clp)
{
	int res;

	might_sleep();

202
	res = wait_on_bit(&clp->cl_state, NFS4CLNT_MANAGER_RUNNING,
203
			nfs_wait_bit_killable, TASK_KILLABLE);
204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229
	return res;
}

static int nfs4_delay(struct rpc_clnt *clnt, long *timeout)
{
	int res = 0;

	might_sleep();

	if (*timeout <= 0)
		*timeout = NFS4_POLL_RETRY_MIN;
	if (*timeout > NFS4_POLL_RETRY_MAX)
		*timeout = NFS4_POLL_RETRY_MAX;
	schedule_timeout_killable(*timeout);
	if (fatal_signal_pending(current))
		res = -ERESTARTSYS;
	*timeout <<= 1;
	return res;
}

/* This is the error handling routine for processes that are allowed
 * to sleep.
 */
static int nfs4_handle_exception(const struct nfs_server *server, int errorcode, struct nfs4_exception *exception)
{
	struct nfs_client *clp = server->nfs_client;
230
	struct nfs4_state *state = exception->state;
231 232 233 234 235 236
	int ret = errorcode;

	exception->retry = 0;
	switch(errorcode) {
		case 0:
			return 0;
237 238 239 240 241 242
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OPENMODE:
			if (state == NULL)
				break;
			nfs4_state_mark_reclaim_nograce(clp, state);
243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264
		case -NFS4ERR_STALE_CLIENTID:
		case -NFS4ERR_STALE_STATEID:
		case -NFS4ERR_EXPIRED:
			nfs4_schedule_state_recovery(clp);
			ret = nfs4_wait_clnt_recover(clp);
			if (ret == 0)
				exception->retry = 1;
			break;
		case -NFS4ERR_FILE_OPEN:
		case -NFS4ERR_GRACE:
		case -NFS4ERR_DELAY:
			ret = nfs4_delay(server->client, &exception->timeout);
			if (ret != 0)
				break;
		case -NFS4ERR_OLD_STATEID:
			exception->retry = 1;
	}
	/* We failed to handle the error */
	return nfs4_map_errors(ret);
}


265
static void renew_lease(const struct nfs_server *server, unsigned long timestamp)
L
Linus Torvalds 已提交
266
{
267
	struct nfs_client *clp = server->nfs_client;
L
Linus Torvalds 已提交
268 269 270 271 272 273
	spin_lock(&clp->cl_lock);
	if (time_before(clp->cl_last_renewal,timestamp))
		clp->cl_last_renewal = timestamp;
	spin_unlock(&clp->cl_lock);
}

274
static void update_changeattr(struct inode *dir, struct nfs4_change_info *cinfo)
L
Linus Torvalds 已提交
275
{
276
	struct nfs_inode *nfsi = NFS_I(dir);
L
Linus Torvalds 已提交
277

278
	spin_lock(&dir->i_lock);
279 280
	nfsi->cache_validity |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE|NFS_INO_INVALID_DATA;
	if (!cinfo->atomic || cinfo->before != nfsi->change_attr)
281
		nfs_force_lookup_revalidate(dir);
282
	nfsi->change_attr = cinfo->after;
283
	spin_unlock(&dir->i_lock);
L
Linus Torvalds 已提交
284 285
}

286
struct nfs4_opendata {
287
	struct kref kref;
288 289
	struct nfs_openargs o_arg;
	struct nfs_openres o_res;
290 291
	struct nfs_open_confirmargs c_arg;
	struct nfs_open_confirmres c_res;
292 293
	struct nfs_fattr f_attr;
	struct nfs_fattr dir_attr;
294
	struct path path;
295 296
	struct dentry *dir;
	struct nfs4_state_owner *owner;
297
	struct nfs4_state *state;
298
	struct iattr attrs;
299
	unsigned long timestamp;
300
	unsigned int rpc_done : 1;
301 302
	int rpc_status;
	int cancelled;
303 304
};

305 306 307 308 309

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;
310 311
	p->o_res.seqid = p->o_arg.seqid;
	p->c_res.seqid = p->c_arg.seqid;
312 313 314 315 316
	p->o_res.server = p->o_arg.server;
	nfs_fattr_init(&p->f_attr);
	nfs_fattr_init(&p->dir_attr);
}

317
static struct nfs4_opendata *nfs4_opendata_alloc(struct path *path,
318
		struct nfs4_state_owner *sp, fmode_t fmode, int flags,
319 320
		const struct iattr *attrs)
{
321
	struct dentry *parent = dget_parent(path->dentry);
322 323 324 325 326 327 328 329 330 331
	struct inode *dir = parent->d_inode;
	struct nfs_server *server = NFS_SERVER(dir);
	struct nfs4_opendata *p;

	p = kzalloc(sizeof(*p), GFP_KERNEL);
	if (p == NULL)
		goto err;
	p->o_arg.seqid = nfs_alloc_seqid(&sp->so_seqid);
	if (p->o_arg.seqid == NULL)
		goto err_free;
332 333
	p->path.mnt = mntget(path->mnt);
	p->path.dentry = dget(path->dentry);
334 335 336 337
	p->dir = parent;
	p->owner = sp;
	atomic_inc(&sp->so_count);
	p->o_arg.fh = NFS_FH(dir);
338 339
	p->o_arg.open_flags = flags;
	p->o_arg.fmode = fmode & (FMODE_READ|FMODE_WRITE);
340
	p->o_arg.clientid = server->nfs_client->cl_clientid;
341
	p->o_arg.id = sp->so_owner_id.id;
342
	p->o_arg.name = &p->path.dentry->d_name;
343 344 345 346 347 348 349 350 351 352 353
	p->o_arg.server = server;
	p->o_arg.bitmask = server->attr_bitmask;
	p->o_arg.claim = NFS4_OPEN_CLAIM_NULL;
	if (flags & O_EXCL) {
		u32 *s = (u32 *) p->o_arg.u.verifier.data;
		s[0] = jiffies;
		s[1] = current->pid;
	} else if (flags & O_CREAT) {
		p->o_arg.u.attrs = &p->attrs;
		memcpy(&p->attrs, attrs, sizeof(p->attrs));
	}
354 355 356
	p->c_arg.fh = &p->o_res.fh;
	p->c_arg.stateid = &p->o_res.stateid;
	p->c_arg.seqid = p->o_arg.seqid;
357
	nfs4_init_opendata_res(p);
358
	kref_init(&p->kref);
359 360 361 362 363 364 365 366
	return p;
err_free:
	kfree(p);
err:
	dput(parent);
	return NULL;
}

367
static void nfs4_opendata_free(struct kref *kref)
368
{
369 370 371 372
	struct nfs4_opendata *p = container_of(kref,
			struct nfs4_opendata, kref);

	nfs_free_seqid(p->o_arg.seqid);
373 374
	if (p->state != NULL)
		nfs4_put_open_state(p->state);
375 376
	nfs4_put_state_owner(p->owner);
	dput(p->dir);
J
Jan Blunck 已提交
377
	path_put(&p->path);
378 379 380 381 382 383 384
	kfree(p);
}

static void nfs4_opendata_put(struct nfs4_opendata *p)
{
	if (p != NULL)
		kref_put(&p->kref, nfs4_opendata_free);
385 386
}

387 388 389 390 391 392 393 394
static int nfs4_wait_for_completion_rpc_task(struct rpc_task *task)
{
	int ret;

	ret = rpc_wait_for_completion_task(task);
	return ret;
}

395
static int can_open_cached(struct nfs4_state *state, fmode_t mode, int open_mode)
396 397
{
	int ret = 0;
398 399 400 401

	if (open_mode & O_EXCL)
		goto out;
	switch (mode & (FMODE_READ|FMODE_WRITE)) {
402 403 404 405 406 407 408 409 410
		case FMODE_READ:
			ret |= test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0;
			break;
		case FMODE_WRITE:
			ret |= test_bit(NFS_O_WRONLY_STATE, &state->flags) != 0;
			break;
		case FMODE_READ|FMODE_WRITE:
			ret |= test_bit(NFS_O_RDWR_STATE, &state->flags) != 0;
	}
411
out:
412 413 414
	return ret;
}

415
static int can_open_delegated(struct nfs_delegation *delegation, fmode_t fmode)
416
{
417
	if ((delegation->type & fmode) != fmode)
418
		return 0;
419
	if (test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags))
420
		return 0;
421
	nfs_mark_delegation_referenced(delegation);
422 423 424
	return 1;
}

425
static void update_open_stateflags(struct nfs4_state *state, fmode_t fmode)
426
{
427
	switch (fmode) {
428 429 430 431 432 433 434 435 436
		case FMODE_WRITE:
			state->n_wronly++;
			break;
		case FMODE_READ:
			state->n_rdonly++;
			break;
		case FMODE_READ|FMODE_WRITE:
			state->n_rdwr++;
	}
437
	nfs4_state_set_mode_locked(state, state->state | fmode);
438 439
}

440
static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
441 442 443 444
{
	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));
445
	switch (fmode) {
446 447 448 449 450 451 452 453 454
		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);
	}
455 456
}

457
static void nfs_set_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid, fmode_t fmode)
458
{
459
	write_seqlock(&state->seqlock);
460
	nfs_set_open_stateid_locked(state, stateid, fmode);
461
	write_sequnlock(&state->seqlock);
462 463
}

464
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
465
{
466 467 468 469 470
	/*
	 * Protect the call to nfs4_state_set_mode_locked and
	 * serialise the stateid update
	 */
	write_seqlock(&state->seqlock);
471 472 473 474 475
	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)
476
		nfs_set_open_stateid_locked(state, open_stateid, fmode);
477 478
	write_sequnlock(&state->seqlock);
	spin_lock(&state->owner->so_lock);
479
	update_open_stateflags(state, fmode);
480
	spin_unlock(&state->owner->so_lock);
L
Linus Torvalds 已提交
481 482
}

483
static int update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *delegation, fmode_t fmode)
484 485 486 487 488
{
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *deleg_cur;
	int ret = 0;

489
	fmode &= (FMODE_READ|FMODE_WRITE);
490 491 492 493 494 495 496 497

	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 ||
498
	    (deleg_cur->type & fmode) != fmode)
499 500 501 502 503 504 505
		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;

506
	nfs_mark_delegation_referenced(deleg_cur);
507
	__update_open_stateid(state, open_stateid, &deleg_cur->stateid, fmode);
508 509 510 511 512 513 514
	ret = 1;
no_delegation_unlock:
	spin_unlock(&deleg_cur->lock);
no_delegation:
	rcu_read_unlock();

	if (!ret && open_stateid != NULL) {
515
		__update_open_stateid(state, open_stateid, NULL, fmode);
516 517 518 519 520 521 522
		ret = 1;
	}

	return ret;
}


523
static void nfs4_return_incompatible_delegation(struct inode *inode, fmode_t fmode)
524 525 526 527 528
{
	struct nfs_delegation *delegation;

	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(inode)->delegation);
529
	if (delegation == NULL || (delegation->type & fmode) == fmode) {
530 531 532 533 534 535 536
		rcu_read_unlock();
		return;
	}
	rcu_read_unlock();
	nfs_inode_return_delegation(inode);
}

537
static struct nfs4_state *nfs4_try_open_cached(struct nfs4_opendata *opendata)
538 539 540 541
{
	struct nfs4_state *state = opendata->state;
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_delegation *delegation;
542 543
	int open_mode = opendata->o_arg.open_flags & O_EXCL;
	fmode_t fmode = opendata->o_arg.fmode;
544 545 546 547
	nfs4_stateid stateid;
	int ret = -EAGAIN;

	for (;;) {
548
		if (can_open_cached(state, fmode, open_mode)) {
549
			spin_lock(&state->owner->so_lock);
550 551
			if (can_open_cached(state, fmode, open_mode)) {
				update_open_stateflags(state, fmode);
552 553 554 555 556
				spin_unlock(&state->owner->so_lock);
				goto out_return_state;
			}
			spin_unlock(&state->owner->so_lock);
		}
557 558 559
		rcu_read_lock();
		delegation = rcu_dereference(nfsi->delegation);
		if (delegation == NULL ||
560
		    !can_open_delegated(delegation, fmode)) {
561
			rcu_read_unlock();
562
			break;
563
		}
564 565 566
		/* Save the delegation */
		memcpy(stateid.data, delegation->stateid.data, sizeof(stateid.data));
		rcu_read_unlock();
567
		ret = nfs_may_open(state->inode, state->owner->so_cred, open_mode);
568 569 570
		if (ret != 0)
			goto out;
		ret = -EAGAIN;
571 572

		/* Try to update the stateid using the delegation */
573
		if (update_open_stateid(state, NULL, &stateid, fmode))
574
			goto out_return_state;
575 576 577 578 579 580 581 582
	}
out:
	return ERR_PTR(ret);
out_return_state:
	atomic_inc(&state->count);
	return state;
}

583 584 585 586
static struct nfs4_state *nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data)
{
	struct inode *inode;
	struct nfs4_state *state = NULL;
587
	struct nfs_delegation *delegation;
588
	int ret;
589

590
	if (!data->rpc_done) {
591
		state = nfs4_try_open_cached(data);
592 593 594
		goto out;
	}

595
	ret = -EAGAIN;
596
	if (!(data->f_attr.valid & NFS_ATTR_FATTR))
597
		goto err;
598
	inode = nfs_fhget(data->dir->d_sb, &data->o_res.fh, &data->f_attr);
599
	ret = PTR_ERR(inode);
600
	if (IS_ERR(inode))
601 602
		goto err;
	ret = -ENOMEM;
603 604
	state = nfs4_get_open_state(inode, data->owner);
	if (state == NULL)
605
		goto err_put_inode;
606 607 608
	if (data->o_res.delegation_type != 0) {
		int delegation_flags = 0;

609 610 611 612 613
		rcu_read_lock();
		delegation = rcu_dereference(NFS_I(inode)->delegation);
		if (delegation)
			delegation_flags = delegation->flags;
		rcu_read_unlock();
614
		if ((delegation_flags & 1UL<<NFS_DELEGATION_NEED_RECLAIM) == 0)
615 616 617 618 619 620 621 622
			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);
	}
623 624

	update_open_stateid(state, &data->o_res.stateid, NULL,
625
			data->o_arg.fmode);
626
	iput(inode);
627
out:
628
	return state;
629 630 631 632
err_put_inode:
	iput(inode);
err:
	return ERR_PTR(ret);
633 634
}

635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651
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
652 653 654 655
static struct nfs4_opendata *nfs4_open_recoverdata_alloc(struct nfs_open_context *ctx, struct nfs4_state *state)
{
	struct nfs4_opendata *opendata;

656
	opendata = nfs4_opendata_alloc(&ctx->path, state->owner, 0, 0, NULL);
T
Trond Myklebust 已提交
657 658 659 660 661 662 663
	if (opendata == NULL)
		return ERR_PTR(-ENOMEM);
	opendata->state = state;
	atomic_inc(&state->count);
	return opendata;
}

664
static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, fmode_t fmode, struct nfs4_state **res)
665
{
666
	struct nfs4_state *newstate;
667 668
	int ret;

669 670
	opendata->o_arg.open_flags = 0;
	opendata->o_arg.fmode = fmode;
671 672 673
	memset(&opendata->o_res, 0, sizeof(opendata->o_res));
	memset(&opendata->c_res, 0, sizeof(opendata->c_res));
	nfs4_init_opendata_res(opendata);
674 675 676
	ret = _nfs4_proc_open(opendata);
	if (ret != 0)
		return ret; 
677
	newstate = nfs4_opendata_to_nfs4_state(opendata);
678 679
	if (IS_ERR(newstate))
		return PTR_ERR(newstate);
680
	nfs4_close_state(&opendata->path, newstate, fmode);
681
	*res = newstate;
682 683 684 685 686 687 688 689 690
	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_* */
691
	clear_bit(NFS_DELEGATED_STATE, &state->flags);
692 693
	smp_rmb();
	if (state->n_rdwr != 0) {
694
		ret = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE, &newstate);
695 696
		if (ret != 0)
			return ret;
697 698
		if (newstate != state)
			return -ESTALE;
699 700
	}
	if (state->n_wronly != 0) {
701
		ret = nfs4_open_recover_helper(opendata, FMODE_WRITE, &newstate);
702 703
		if (ret != 0)
			return ret;
704 705
		if (newstate != state)
			return -ESTALE;
706 707
	}
	if (state->n_rdonly != 0) {
708
		ret = nfs4_open_recover_helper(opendata, FMODE_READ, &newstate);
709 710
		if (ret != 0)
			return ret;
711 712
		if (newstate != state)
			return -ESTALE;
713
	}
714 715 716 717 718 719
	/*
	 * 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) {
720
		write_seqlock(&state->seqlock);
721 722
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
			memcpy(state->stateid.data, state->open_stateid.data, sizeof(state->stateid.data));
723
		write_sequnlock(&state->seqlock);
724
	}
725 726 727
	return 0;
}

L
Linus Torvalds 已提交
728 729 730 731
/*
 * OPEN_RECLAIM:
 * 	reclaim state on the server after a reboot.
 */
732
static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
733
{
734
	struct nfs_delegation *delegation;
735
	struct nfs4_opendata *opendata;
736
	fmode_t delegation_type = 0;
L
Linus Torvalds 已提交
737 738
	int status;

T
Trond Myklebust 已提交
739 740 741
	opendata = nfs4_open_recoverdata_alloc(ctx, state);
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
742 743
	opendata->o_arg.claim = NFS4_OPEN_CLAIM_PREVIOUS;
	opendata->o_arg.fh = NFS_FH(state->inode);
744 745
	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(state->inode)->delegation);
746
	if (delegation != NULL && test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) != 0)
747
		delegation_type = delegation->type;
748
	rcu_read_unlock();
749 750
	opendata->o_arg.u.delegation_type = delegation_type;
	status = nfs4_open_recover(opendata, state);
751
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
752 753 754
	return status;
}

755
static int nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
756 757 758 759 760
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;
	do {
761
		err = _nfs4_do_open_reclaim(ctx, state);
762 763 764
		if (err != -NFS4ERR_DELAY)
			break;
		nfs4_handle_exception(server, err, &exception);
L
Linus Torvalds 已提交
765 766 767 768
	} while (exception.retry);
	return err;
}

769 770 771 772 773 774 775 776
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);
777
	ret = nfs4_do_open_reclaim(ctx, state);
778 779 780 781
	put_nfs_open_context(ctx);
	return ret;
}

782
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
783
{
784
	struct nfs4_opendata *opendata;
785
	int ret;
L
Linus Torvalds 已提交
786

T
Trond Myklebust 已提交
787 788 789
	opendata = nfs4_open_recoverdata_alloc(ctx, state);
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
790
	opendata->o_arg.claim = NFS4_OPEN_CLAIM_DELEGATE_CUR;
791
	memcpy(opendata->o_arg.u.delegation.data, stateid->data,
792
			sizeof(opendata->o_arg.u.delegation.data));
793
	ret = nfs4_open_recover(opendata, state);
794
	nfs4_opendata_put(opendata);
795
	return ret;
L
Linus Torvalds 已提交
796 797
}

798
int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid)
L
Linus Torvalds 已提交
799 800
{
	struct nfs4_exception exception = { };
801
	struct nfs_server *server = NFS_SERVER(state->inode);
L
Linus Torvalds 已提交
802 803
	int err;
	do {
804
		err = _nfs4_open_delegation_recall(ctx, state, stateid);
L
Linus Torvalds 已提交
805 806 807 808 809 810 811
		switch (err) {
			case 0:
				return err;
			case -NFS4ERR_STALE_CLIENTID:
			case -NFS4ERR_STALE_STATEID:
			case -NFS4ERR_EXPIRED:
				/* Don't recall a delegation if it was lost */
812
				nfs4_schedule_state_recovery(server->nfs_client);
L
Linus Torvalds 已提交
813 814 815 816 817 818 819
				return err;
		}
		err = nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
	return err;
}

820 821 822 823 824 825 826
static void nfs4_open_confirm_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;

	data->rpc_status = task->tk_status;
	if (RPC_ASSASSINATED(task))
		return;
827
	if (data->rpc_status == 0) {
828 829
		memcpy(data->o_res.stateid.data, data->c_res.stateid.data,
				sizeof(data->o_res.stateid.data));
830
		nfs_confirm_seqid(&data->owner->so_seqid, 0);
831
		renew_lease(data->o_res.server, data->timestamp);
832
		data->rpc_done = 1;
833
	}
834 835 836 837 838 839 840 841 842 843 844
}

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! */
845
	if (!data->rpc_done)
846 847
		goto out_free;
	state = nfs4_opendata_to_nfs4_state(data);
848
	if (!IS_ERR(state))
849
		nfs4_close_state(&data->path, state, data->o_arg.fmode);
850
out_free:
851
	nfs4_opendata_put(data);
852 853 854 855 856 857 858 859 860 861 862 863 864 865
}

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;
866 867 868 869 870 871
	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
872 873
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
874
		.rpc_message = &msg,
T
Trond Myklebust 已提交
875 876
		.callback_ops = &nfs4_open_confirm_ops,
		.callback_data = data,
877
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
878 879
		.flags = RPC_TASK_ASYNC,
	};
L
Linus Torvalds 已提交
880 881
	int status;

882
	kref_get(&data->kref);
883 884
	data->rpc_done = 0;
	data->rpc_status = 0;
885
	data->timestamp = jiffies;
T
Trond Myklebust 已提交
886
	task = rpc_run_task(&task_setup_data);
887
	if (IS_ERR(task))
888 889 890 891 892 893 894
		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;
895
	rpc_put_task(task);
L
Linus Torvalds 已提交
896 897 898
	return status;
}

899
static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
L
Linus Torvalds 已提交
900
{
901 902
	struct nfs4_opendata *data = calldata;
	struct nfs4_state_owner *sp = data->owner;
903

904 905
	if (nfs_wait_on_sequence(data->o_arg.seqid, task) != 0)
		return;
906 907 908 909 910 911 912
	/*
	 * 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;

913
		if (can_open_cached(data->state, data->o_arg.fmode, data->o_arg.open_flags))
914
			goto out_no_action;
915 916 917
		rcu_read_lock();
		delegation = rcu_dereference(NFS_I(data->state->inode)->delegation);
		if (delegation != NULL &&
918
		    test_bit(NFS_DELEGATION_NEED_RECLAIM, &delegation->flags) == 0) {
919
			rcu_read_unlock();
920
			goto out_no_action;
921 922 923
		}
		rcu_read_unlock();
	}
924
	/* Update sequence id. */
925
	data->o_arg.id = sp->so_owner_id.id;
926
	data->o_arg.clientid = sp->so_client->cl_clientid;
T
Trond Myklebust 已提交
927
	if (data->o_arg.claim == NFS4_OPEN_CLAIM_PREVIOUS) {
928
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR];
T
Trond Myklebust 已提交
929 930
		nfs_copy_fh(&data->o_res.fh, data->o_arg.fh);
	}
931
	data->timestamp = jiffies;
932
	rpc_call_start(task);
933 934 935 936
	return;
out_no_action:
	task->tk_action = NULL;

937
}
L
Linus Torvalds 已提交
938

939 940 941
static void nfs4_open_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;
L
Linus Torvalds 已提交
942

943 944 945 946 947
	data->rpc_status = task->tk_status;
	if (RPC_ASSASSINATED(task))
		return;
	if (task->tk_status == 0) {
		switch (data->o_res.f_attr->mode & S_IFMT) {
948 949 950
			case S_IFREG:
				break;
			case S_IFLNK:
951
				data->rpc_status = -ELOOP;
952 953
				break;
			case S_IFDIR:
954
				data->rpc_status = -EISDIR;
955 956
				break;
			default:
957
				data->rpc_status = -ENOTDIR;
958
		}
959
		renew_lease(data->o_res.server, data->timestamp);
960 961
		if (!(data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM))
			nfs_confirm_seqid(&data->owner->so_seqid, 0);
962
	}
963
	data->rpc_done = 1;
964
}
965

966 967 968 969 970 971 972 973 974
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! */
975
	if (data->rpc_status != 0 || !data->rpc_done)
976 977 978 979 980
		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);
981
	if (!IS_ERR(state))
982
		nfs4_close_state(&data->path, state, data->o_arg.fmode);
983
out_free:
984
	nfs4_opendata_put(data);
985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002
}

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,
};

/*
 * 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;
	struct rpc_task *task;
1003 1004 1005 1006 1007 1008
	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
1009 1010
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1011
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1012 1013
		.callback_ops = &nfs4_open_ops,
		.callback_data = data,
1014
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1015 1016
		.flags = RPC_TASK_ASYNC,
	};
1017 1018
	int status;

1019
	kref_get(&data->kref);
1020 1021
	data->rpc_done = 0;
	data->rpc_status = 0;
1022
	data->cancelled = 0;
T
Trond Myklebust 已提交
1023
	task = rpc_run_task(&task_setup_data);
1024
	if (IS_ERR(task))
1025 1026 1027 1028 1029 1030 1031
		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;
1032
	rpc_put_task(task);
1033
	if (status != 0 || !data->rpc_done)
1034 1035
		return status;

1036 1037 1038
	if (o_res->fh.size == 0)
		_nfs4_proc_lookup(dir, o_arg->name, &o_res->fh, o_res->f_attr);

1039 1040 1041 1042 1043
	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);
L
Linus Torvalds 已提交
1044
	if(o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
1045
		status = _nfs4_proc_open_confirm(data);
L
Linus Torvalds 已提交
1046
		if (status != 0)
1047
			return status;
L
Linus Torvalds 已提交
1048 1049
	}
	if (!(o_res->f_attr->valid & NFS_ATTR_FATTR))
1050
		_nfs4_proc_getattr(server, &o_res->fh, o_res->f_attr);
1051
	return 0;
L
Linus Torvalds 已提交
1052 1053
}

1054
static int nfs4_recover_expired_lease(struct nfs_server *server)
1055
{
1056
	struct nfs_client *clp = server->nfs_client;
1057
	int ret;
1058

1059
	for (;;) {
1060
		ret = nfs4_wait_clnt_recover(clp);
1061 1062
		if (ret != 0)
			return ret;
1063 1064
		if (!test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state) &&
		    !test_bit(NFS4CLNT_CHECK_LEASE,&clp->cl_state))
1065
			break;
1066
		nfs4_schedule_state_recovery(clp);
1067 1068
	}
	return 0;
1069 1070
}

L
Linus Torvalds 已提交
1071 1072 1073 1074 1075
/*
 * OPEN_EXPIRED:
 * 	reclaim state on the server after a network partition.
 * 	Assumes caller holds the appropriate lock
 */
1076
static int _nfs4_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
L
Linus Torvalds 已提交
1077
{
1078
	struct nfs4_opendata *opendata;
1079
	int ret;
L
Linus Torvalds 已提交
1080

T
Trond Myklebust 已提交
1081 1082 1083
	opendata = nfs4_open_recoverdata_alloc(ctx, state);
	if (IS_ERR(opendata))
		return PTR_ERR(opendata);
1084
	ret = nfs4_open_recover(opendata, state);
1085
	if (ret == -ESTALE)
1086
		d_drop(ctx->path.dentry);
1087
	nfs4_opendata_put(opendata);
1088
	return ret;
L
Linus Torvalds 已提交
1089 1090
}

1091
static inline int nfs4_do_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
1092
{
1093
	struct nfs_server *server = NFS_SERVER(state->inode);
1094 1095 1096 1097
	struct nfs4_exception exception = { };
	int err;

	do {
1098
		err = _nfs4_open_expired(ctx, state);
1099 1100 1101
		if (err != -NFS4ERR_DELAY)
			break;
		nfs4_handle_exception(server, err, &exception);
1102 1103 1104 1105
	} while (exception.retry);
	return err;
}

L
Linus Torvalds 已提交
1106 1107 1108
static int nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
	struct nfs_open_context *ctx;
1109
	int ret;
L
Linus Torvalds 已提交
1110

1111 1112 1113
	ctx = nfs4_state_find_open_context(state);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);
1114
	ret = nfs4_do_open_expired(ctx, state);
1115 1116
	put_nfs_open_context(ctx);
	return ret;
L
Linus Torvalds 已提交
1117 1118
}

1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134
/*
 * 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
1135
/*
1136
 * Returns a referenced nfs4_state
L
Linus Torvalds 已提交
1137
 */
1138
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
1139 1140 1141 1142
{
	struct nfs4_state_owner  *sp;
	struct nfs4_state     *state = NULL;
	struct nfs_server       *server = NFS_SERVER(dir);
1143
	struct nfs4_opendata *opendata;
1144
	int status;
L
Linus Torvalds 已提交
1145 1146 1147 1148 1149 1150 1151

	/* 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;
	}
1152 1153
	status = nfs4_recover_expired_lease(server);
	if (status != 0)
1154
		goto err_put_state_owner;
1155
	if (path->dentry->d_inode != NULL)
1156
		nfs4_return_incompatible_delegation(path->dentry->d_inode, fmode);
1157
	status = -ENOMEM;
1158
	opendata = nfs4_opendata_alloc(path, sp, fmode, flags, sattr);
1159
	if (opendata == NULL)
T
Trond Myklebust 已提交
1160
		goto err_put_state_owner;
L
Linus Torvalds 已提交
1161

1162 1163 1164
	if (path->dentry->d_inode != NULL)
		opendata->state = nfs4_get_open_state(path->dentry->d_inode, sp);

1165
	status = _nfs4_proc_open(opendata);
L
Linus Torvalds 已提交
1166
	if (status != 0)
1167
		goto err_opendata_put;
L
Linus Torvalds 已提交
1168

1169 1170 1171
	if (opendata->o_arg.open_flags & O_EXCL)
		nfs4_exclusive_attrset(opendata, sattr);

1172
	state = nfs4_opendata_to_nfs4_state(opendata);
1173 1174
	status = PTR_ERR(state);
	if (IS_ERR(state))
1175 1176
		goto err_opendata_put;
	nfs4_opendata_put(opendata);
L
Linus Torvalds 已提交
1177 1178 1179
	nfs4_put_state_owner(sp);
	*res = state;
	return 0;
1180 1181
err_opendata_put:
	nfs4_opendata_put(opendata);
1182 1183
err_put_state_owner:
	nfs4_put_state_owner(sp);
L
Linus Torvalds 已提交
1184 1185 1186 1187 1188 1189
out_err:
	*res = NULL;
	return status;
}


1190
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
1191 1192 1193 1194 1195 1196
{
	struct nfs4_exception exception = { };
	struct nfs4_state *res;
	int status;

	do {
1197
		status = _nfs4_do_open(dir, path, fmode, flags, sattr, cred, &res);
L
Linus Torvalds 已提交
1198 1199 1200 1201 1202 1203 1204 1205
		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
1206
		 * doing an OPEN, we assume that nfs_increment_open_seqid() will
L
Linus Torvalds 已提交
1207 1208 1209 1210 1211
		 * 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) {
1212 1213 1214
			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
1215 1216 1217
			exception.retry = 1;
			continue;
		}
1218 1219 1220 1221 1222 1223 1224 1225 1226 1227
		/*
		 * 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;
		}
1228 1229 1230 1231 1232
		if (status == -EAGAIN) {
			/* We must have found a delegation */
			exception.retry = 1;
			continue;
		}
L
Linus Torvalds 已提交
1233 1234 1235 1236 1237 1238
		res = ERR_PTR(nfs4_handle_exception(NFS_SERVER(dir),
					status, &exception));
	} while (exception.retry);
	return res;
}

1239 1240 1241
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
1242
{
1243
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
1244
        struct nfs_setattrargs  arg = {
1245
                .fh             = NFS_FH(inode),
L
Linus Torvalds 已提交
1246 1247 1248 1249 1250 1251 1252 1253 1254
                .iap            = sattr,
		.server		= server,
		.bitmask = server->attr_bitmask,
        };
        struct nfs_setattrres  res = {
		.fattr		= fattr,
		.server		= server,
        };
        struct rpc_message msg = {
1255 1256 1257 1258
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETATTR],
		.rpc_argp	= &arg,
		.rpc_resp	= &res,
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
1259
        };
1260
	unsigned long timestamp = jiffies;
1261
	int status;
L
Linus Torvalds 已提交
1262

1263
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
1264

1265 1266 1267
	if (nfs4_copy_delegation_stateid(&arg.stateid, inode)) {
		/* Use that stateid */
	} else if (state != NULL) {
1268 1269
		nfs4_copy_stateid(&arg.stateid, state, current->files);
	} else
L
Linus Torvalds 已提交
1270 1271
		memcpy(&arg.stateid, &zero_stateid, sizeof(arg.stateid));

1272
	status = rpc_call_sync(server->client, &msg, 0);
1273 1274
	if (status == 0 && state != NULL)
		renew_lease(server, timestamp);
1275
	return status;
L
Linus Torvalds 已提交
1276 1277
}

1278 1279 1280
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
1281
{
1282
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
1283 1284 1285 1286
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(server,
1287
				_nfs4_do_setattr(inode, cred, fattr, sattr, state),
L
Linus Torvalds 已提交
1288 1289 1290 1291 1292 1293
				&exception);
	} while (exception.retry);
	return err;
}

struct nfs4_closedata {
1294
	struct path path;
L
Linus Torvalds 已提交
1295 1296 1297 1298
	struct inode *inode;
	struct nfs4_state *state;
	struct nfs_closeargs arg;
	struct nfs_closeres res;
1299
	struct nfs_fattr fattr;
1300
	unsigned long timestamp;
L
Linus Torvalds 已提交
1301 1302
};

1303
static void nfs4_free_closedata(void *data)
1304
{
1305 1306
	struct nfs4_closedata *calldata = data;
	struct nfs4_state_owner *sp = calldata->state->owner;
1307 1308 1309 1310

	nfs4_put_open_state(calldata->state);
	nfs_free_seqid(calldata->arg.seqid);
	nfs4_put_state_owner(sp);
J
Jan Blunck 已提交
1311
	path_put(&calldata->path);
1312 1313 1314
	kfree(calldata);
}

1315
static void nfs4_close_done(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
1316
{
1317
	struct nfs4_closedata *calldata = data;
L
Linus Torvalds 已提交
1318 1319 1320
	struct nfs4_state *state = calldata->state;
	struct nfs_server *server = NFS_SERVER(calldata->inode);

T
Trond Myklebust 已提交
1321 1322
	if (RPC_ASSASSINATED(task))
		return;
L
Linus Torvalds 已提交
1323 1324 1325 1326 1327
        /* 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:
1328
			nfs_set_open_stateid(state, &calldata->res.stateid, 0);
1329
			renew_lease(server, calldata->timestamp);
L
Linus Torvalds 已提交
1330 1331
			break;
		case -NFS4ERR_STALE_STATEID:
1332 1333
		case -NFS4ERR_OLD_STATEID:
		case -NFS4ERR_BAD_STATEID:
L
Linus Torvalds 已提交
1334
		case -NFS4ERR_EXPIRED:
1335
			if (calldata->arg.fmode == 0)
1336
				break;
L
Linus Torvalds 已提交
1337
		default:
1338
			if (nfs4_async_handle_error(task, server, state) == -EAGAIN) {
L
Linus Torvalds 已提交
1339 1340 1341 1342
				rpc_restart_call(task);
				return;
			}
	}
1343
	nfs_refresh_inode(calldata->inode, calldata->res.fattr);
L
Linus Torvalds 已提交
1344 1345
}

T
Trond Myklebust 已提交
1346
static void nfs4_close_prepare(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
1347
{
T
Trond Myklebust 已提交
1348
	struct nfs4_closedata *calldata = data;
1349
	struct nfs4_state *state = calldata->state;
1350
	int clear_rd, clear_wr, clear_rdwr;
1351

1352
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
1353
		return;
1354 1355

	clear_rd = clear_wr = clear_rdwr = 0;
1356
	spin_lock(&state->owner->so_lock);
1357
	/* Calculate the change in open mode */
1358
	if (state->n_rdwr == 0) {
1359 1360 1361 1362 1363 1364 1365 1366
		if (state->n_rdonly == 0) {
			clear_rd |= test_and_clear_bit(NFS_O_RDONLY_STATE, &state->flags);
			clear_rdwr |= test_and_clear_bit(NFS_O_RDWR_STATE, &state->flags);
		}
		if (state->n_wronly == 0) {
			clear_wr |= test_and_clear_bit(NFS_O_WRONLY_STATE, &state->flags);
			clear_rdwr |= test_and_clear_bit(NFS_O_RDWR_STATE, &state->flags);
		}
1367
	}
1368
	spin_unlock(&state->owner->so_lock);
1369
	if (!clear_rd && !clear_wr && !clear_rdwr) {
1370 1371
		/* Note: exit _without_ calling nfs4_close_done */
		task->tk_action = NULL;
1372 1373
		return;
	}
1374
	nfs_fattr_init(calldata->res.fattr);
1375
	if (test_bit(NFS_O_RDONLY_STATE, &state->flags) != 0) {
1376
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_DOWNGRADE];
1377
		calldata->arg.fmode = FMODE_READ;
1378
	} else if (test_bit(NFS_O_WRONLY_STATE, &state->flags) != 0) {
1379
		task->tk_msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_DOWNGRADE];
1380
		calldata->arg.fmode = FMODE_WRITE;
1381
	}
1382
	calldata->timestamp = jiffies;
1383
	rpc_call_start(task);
L
Linus Torvalds 已提交
1384 1385
}

1386
static const struct rpc_call_ops nfs4_close_ops = {
T
Trond Myklebust 已提交
1387
	.rpc_call_prepare = nfs4_close_prepare,
1388 1389 1390 1391
	.rpc_call_done = nfs4_close_done,
	.rpc_release = nfs4_free_closedata,
};

L
Linus Torvalds 已提交
1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402
/* 
 * 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!
 */
1403
int nfs4_do_close(struct path *path, struct nfs4_state *state, int wait)
L
Linus Torvalds 已提交
1404
{
1405
	struct nfs_server *server = NFS_SERVER(state->inode);
L
Linus Torvalds 已提交
1406
	struct nfs4_closedata *calldata;
1407 1408
	struct nfs4_state_owner *sp = state->owner;
	struct rpc_task *task;
1409 1410 1411 1412
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CLOSE],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
1413 1414
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
1415
		.rpc_message = &msg,
T
Trond Myklebust 已提交
1416
		.callback_ops = &nfs4_close_ops,
1417
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
1418 1419
		.flags = RPC_TASK_ASYNC,
	};
1420
	int status = -ENOMEM;
L
Linus Torvalds 已提交
1421

1422
	calldata = kmalloc(sizeof(*calldata), GFP_KERNEL);
L
Linus Torvalds 已提交
1423
	if (calldata == NULL)
1424
		goto out;
1425
	calldata->inode = state->inode;
L
Linus Torvalds 已提交
1426
	calldata->state = state;
1427
	calldata->arg.fh = NFS_FH(state->inode);
1428
	calldata->arg.stateid = &state->open_stateid;
L
Linus Torvalds 已提交
1429
	/* Serialization for the sequence id */
1430
	calldata->arg.seqid = nfs_alloc_seqid(&state->owner->so_seqid);
1431 1432
	if (calldata->arg.seqid == NULL)
		goto out_free_calldata;
1433
	calldata->arg.fmode = 0;
1434
	calldata->arg.bitmask = server->cache_consistency_bitmask;
1435
	calldata->res.fattr = &calldata->fattr;
1436
	calldata->res.seqid = calldata->arg.seqid;
1437
	calldata->res.server = server;
1438 1439
	calldata->path.mnt = mntget(path->mnt);
	calldata->path.dentry = dget(path->dentry);
1440

1441 1442
	msg.rpc_argp = &calldata->arg,
	msg.rpc_resp = &calldata->res,
T
Trond Myklebust 已提交
1443 1444
	task_setup_data.callback_data = calldata;
	task = rpc_run_task(&task_setup_data);
1445 1446
	if (IS_ERR(task))
		return PTR_ERR(task);
1447 1448 1449
	status = 0;
	if (wait)
		status = rpc_wait_for_completion_task(task);
1450
	rpc_put_task(task);
1451
	return status;
1452 1453 1454
out_free_calldata:
	kfree(calldata);
out:
1455 1456
	nfs4_put_open_state(state);
	nfs4_put_state_owner(sp);
1457
	return status;
L
Linus Torvalds 已提交
1458 1459
}

1460
static int nfs4_intent_set_file(struct nameidata *nd, struct path *path, struct nfs4_state *state, fmode_t fmode)
1461 1462
{
	struct file *filp;
1463
	int ret;
1464

1465
	/* If the open_intent is for execute, we have an extra check to make */
1466
	if (fmode & FMODE_EXEC) {
1467
		ret = nfs_may_open(state->inode,
1468 1469 1470 1471 1472
				state->owner->so_cred,
				nd->intent.open.flags);
		if (ret < 0)
			goto out_close;
	}
1473
	filp = lookup_instantiate_filp(nd, path->dentry, NULL);
1474 1475
	if (!IS_ERR(filp)) {
		struct nfs_open_context *ctx;
1476
		ctx = nfs_file_open_context(filp);
1477
		ctx->state = state;
1478 1479
		return 0;
	}
1480 1481
	ret = PTR_ERR(filp);
out_close:
1482
	nfs4_close_sync(path, state, fmode & (FMODE_READ|FMODE_WRITE));
1483
	return ret;
1484 1485 1486
}

struct dentry *
L
Linus Torvalds 已提交
1487 1488
nfs4_atomic_open(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
{
1489
	struct path path = {
1490
		.mnt = nd->path.mnt,
1491 1492
		.dentry = dentry,
	};
1493
	struct dentry *parent;
L
Linus Torvalds 已提交
1494 1495 1496
	struct iattr attr;
	struct rpc_cred *cred;
	struct nfs4_state *state;
1497
	struct dentry *res;
1498
	fmode_t fmode = nd->intent.open.flags & (FMODE_READ | FMODE_WRITE | FMODE_EXEC);
L
Linus Torvalds 已提交
1499 1500 1501 1502 1503

	if (nd->flags & LOOKUP_CREATE) {
		attr.ia_mode = nd->intent.open.create_mode;
		attr.ia_valid = ATTR_MODE;
		if (!IS_POSIXACL(dir))
A
Al Viro 已提交
1504
			attr.ia_mode &= ~current_umask();
L
Linus Torvalds 已提交
1505 1506 1507 1508 1509
	} else {
		attr.ia_valid = 0;
		BUG_ON(nd->intent.open.flags & O_CREAT);
	}

1510
	cred = rpc_lookup_cred();
L
Linus Torvalds 已提交
1511
	if (IS_ERR(cred))
1512
		return (struct dentry *)cred;
T
Trond Myklebust 已提交
1513 1514 1515
	parent = dentry->d_parent;
	/* Protect against concurrent sillydeletes */
	nfs_block_sillyrename(parent);
1516
	state = nfs4_do_open(dir, &path, fmode, nd->intent.open.flags, &attr, cred);
L
Linus Torvalds 已提交
1517
	put_rpccred(cred);
1518
	if (IS_ERR(state)) {
1519
		if (PTR_ERR(state) == -ENOENT) {
1520
			d_add(dentry, NULL);
1521 1522
			nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
		}
T
Trond Myklebust 已提交
1523
		nfs_unblock_sillyrename(parent);
1524 1525
		return (struct dentry *)state;
	}
1526
	res = d_add_unique(dentry, igrab(state->inode));
1527
	if (res != NULL)
1528
		path.dentry = res;
1529
	nfs_set_verifier(path.dentry, nfs_save_change_attribute(dir));
T
Trond Myklebust 已提交
1530
	nfs_unblock_sillyrename(parent);
1531
	nfs4_intent_set_file(nd, &path, state, fmode);
1532
	return res;
L
Linus Torvalds 已提交
1533 1534 1535
}

int
1536
nfs4_open_revalidate(struct inode *dir, struct dentry *dentry, int openflags, struct nameidata *nd)
L
Linus Torvalds 已提交
1537
{
1538
	struct path path = {
1539
		.mnt = nd->path.mnt,
1540 1541
		.dentry = dentry,
	};
L
Linus Torvalds 已提交
1542 1543
	struct rpc_cred *cred;
	struct nfs4_state *state;
1544
	fmode_t fmode = openflags & (FMODE_READ | FMODE_WRITE);
L
Linus Torvalds 已提交
1545

1546
	cred = rpc_lookup_cred();
L
Linus Torvalds 已提交
1547 1548
	if (IS_ERR(cred))
		return PTR_ERR(cred);
1549
	state = nfs4_do_open(dir, &path, fmode, openflags, NULL, cred);
L
Linus Torvalds 已提交
1550
	put_rpccred(cred);
1551 1552 1553 1554 1555 1556 1557 1558 1559
	if (IS_ERR(state)) {
		switch (PTR_ERR(state)) {
			case -EPERM:
			case -EACCES:
			case -EDQUOT:
			case -ENOSPC:
			case -EROFS:
				lookup_instantiate_filp(nd, (struct dentry *)state, NULL);
				return 1;
1560 1561
			default:
				goto out_drop;
1562 1563
		}
	}
1564
	if (state->inode == dentry->d_inode) {
1565
		nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
1566
		nfs4_intent_set_file(nd, &path, state, fmode);
L
Linus Torvalds 已提交
1567 1568
		return 1;
	}
1569
	nfs4_close_sync(&path, state, fmode);
1570 1571
out_drop:
	d_drop(dentry);
L
Linus Torvalds 已提交
1572 1573 1574
	return 0;
}

T
Trond Myklebust 已提交
1575 1576 1577 1578 1579 1580 1581 1582 1583
void nfs4_close_context(struct nfs_open_context *ctx, int is_sync)
{
	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
1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603

static int _nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
{
	struct nfs4_server_caps_res res = {};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SERVER_CAPS],
		.rpc_argp = fhandle,
		.rpc_resp = &res,
	};
	int status;

	status = rpc_call_sync(server->client, &msg, 0);
	if (status == 0) {
		memcpy(server->attr_bitmask, res.attr_bitmask, sizeof(server->attr_bitmask));
		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;
1604 1605 1606
		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
1607 1608 1609 1610 1611
		server->acl_bitmask = res.acl_bitmask;
	}
	return status;
}

1612
int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
L
Linus Torvalds 已提交
1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631
{
	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,
1632
		.fattr = info->fattr,
L
Linus Torvalds 已提交
1633 1634 1635 1636 1637 1638 1639
		.fh = fhandle,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOOKUP_ROOT],
		.rpc_argp = &args,
		.rpc_resp = &res,
	};
1640
	nfs_fattr_init(info->fattr);
L
Linus Torvalds 已提交
1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656
	return rpc_call_sync(server->client, &msg, 0);
}

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;
}

1657 1658 1659
/*
 * get the file handle for the "/" directory on the server
 */
L
Linus Torvalds 已提交
1660
static int nfs4_proc_get_root(struct nfs_server *server, struct nfs_fh *fhandle,
1661
			      struct nfs_fsinfo *info)
L
Linus Torvalds 已提交
1662 1663 1664 1665 1666 1667 1668 1669
{
	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);
1670
	return nfs4_map_errors(status);
L
Linus Torvalds 已提交
1671 1672
}

M
Manoj Naik 已提交
1673 1674 1675 1676 1677
/*
 * 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
 */
1678
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
1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690
{
	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;

1691
	status = nfs4_proc_fs_locations(dir, name, locations, page);
M
Manoj Naik 已提交
1692 1693 1694 1695
	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)) {
1696
		dprintk("%s: server did not return a different fsid for a referral at %s\n", __func__, name->name);
M
Manoj Naik 已提交
1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713
		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);
	if (locations)
		kfree(locations);
	return status;
}

L
Linus Torvalds 已提交
1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729
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,
	};
	
1730
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766
	return rpc_call_sync(server->client, &msg, 0);
}

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)
{
1767
	struct inode *inode = dentry->d_inode;
1768
	struct rpc_cred *cred = NULL;
1769
	struct nfs4_state *state = NULL;
L
Linus Torvalds 已提交
1770 1771
	int status;

1772
	nfs_fattr_init(fattr);
L
Linus Torvalds 已提交
1773
	
1774
	/* Search for an existing open(O_WRITE) file */
1775 1776 1777 1778
	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
1779 1780 1781 1782
		if (ctx) {
			cred = ctx->cred;
			state = ctx->state;
		}
1783
	}
1784

1785
	status = nfs4_do_setattr(inode, cred, fattr, sattr, state);
1786 1787
	if (status == 0)
		nfs_setattr_update_inode(inode, sattr);
L
Linus Torvalds 已提交
1788 1789 1790
	return status;
}

1791 1792
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
1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826
		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);
	status = rpc_call_sync(server->client, &msg, 0);
	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 {
1827 1828 1829 1830 1831 1832 1833
		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
1834 1835 1836 1837
	} while (exception.retry);
	return err;
}

1838
static int _nfs4_proc_lookup(struct inode *dir, const struct qstr *name,
L
Linus Torvalds 已提交
1839 1840
		struct nfs_fh *fhandle, struct nfs_fattr *fattr)
{
1841
	int status;
L
Linus Torvalds 已提交
1842 1843
	
	dprintk("NFS call  lookup %s\n", name->name);
1844
	status = _nfs4_proc_lookupfh(NFS_SERVER(dir), NFS_FH(dir), name, fhandle, fattr);
M
Manoj Naik 已提交
1845 1846
	if (status == -NFS4ERR_MOVED)
		status = nfs4_get_referral(dir, name, fattr, fhandle);
L
Linus Torvalds 已提交
1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864
	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)
{
1865 1866
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs_fattr fattr;
L
Linus Torvalds 已提交
1867 1868
	struct nfs4_accessargs args = {
		.fh = NFS_FH(inode),
1869 1870 1871 1872 1873
		.bitmask = server->attr_bitmask,
	};
	struct nfs4_accessres res = {
		.server = server,
		.fattr = &fattr,
L
Linus Torvalds 已提交
1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899
	};
	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;
	}
1900
	nfs_fattr_init(&fattr);
L
Linus Torvalds 已提交
1901 1902 1903 1904 1905 1906 1907 1908 1909
	status = rpc_call_sync(NFS_CLIENT(inode), &msg, 0);
	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;
1910
		nfs_refresh_inode(inode, &fattr);
L
Linus Torvalds 已提交
1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998
	}
	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,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READLINK],
		.rpc_argp = &args,
		.rpc_resp = NULL,
	};

	return rpc_call_sync(NFS_CLIENT(inode), &msg, 0);
}

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,
1999
                 int flags, struct nameidata *nd)
L
Linus Torvalds 已提交
2000
{
2001
	struct path path = {
2002
		.mnt = nd->path.mnt,
2003 2004
		.dentry = dentry,
	};
L
Linus Torvalds 已提交
2005 2006
	struct nfs4_state *state;
	struct rpc_cred *cred;
2007
	fmode_t fmode = flags & (FMODE_READ | FMODE_WRITE);
L
Linus Torvalds 已提交
2008 2009
	int status = 0;

2010
	cred = rpc_lookup_cred();
L
Linus Torvalds 已提交
2011 2012 2013 2014
	if (IS_ERR(cred)) {
		status = PTR_ERR(cred);
		goto out;
	}
2015
	state = nfs4_do_open(dir, &path, fmode, flags, sattr, cred);
2016
	d_drop(dentry);
L
Linus Torvalds 已提交
2017 2018
	if (IS_ERR(state)) {
		status = PTR_ERR(state);
2019
		goto out_putcred;
L
Linus Torvalds 已提交
2020
	}
2021
	d_add(dentry, igrab(state->inode));
2022
	nfs_set_verifier(dentry, nfs_save_change_attribute(dir));
L
Linus Torvalds 已提交
2023 2024
	if (flags & O_EXCL) {
		struct nfs_fattr fattr;
2025
		status = nfs4_do_setattr(state->inode, cred, &fattr, sattr, state);
2026
		if (status == 0)
2027
			nfs_setattr_update_inode(state->inode, sattr);
2028
		nfs_post_op_update_inode(state->inode, &fattr);
2029
	}
2030
	if (status == 0 && (nd->flags & LOOKUP_OPEN) != 0)
2031
		status = nfs4_intent_set_file(nd, &path, state, fmode);
2032
	else
2033
		nfs4_close_sync(&path, state, fmode);
2034 2035
out_putcred:
	put_rpccred(cred);
L
Linus Torvalds 已提交
2036 2037 2038 2039 2040 2041
out:
	return status;
}

static int _nfs4_proc_remove(struct inode *dir, struct qstr *name)
{
2042
	struct nfs_server *server = NFS_SERVER(dir);
2043
	struct nfs_removeargs args = {
L
Linus Torvalds 已提交
2044
		.fh = NFS_FH(dir),
2045 2046
		.name.len = name->len,
		.name.name = name->name,
2047 2048
		.bitmask = server->attr_bitmask,
	};
2049
	struct nfs_removeres res = {
2050
		.server = server,
L
Linus Torvalds 已提交
2051 2052
	};
	struct rpc_message msg = {
2053 2054 2055
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE],
		.rpc_argp = &args,
		.rpc_resp = &res,
L
Linus Torvalds 已提交
2056 2057 2058
	};
	int			status;

2059
	nfs_fattr_init(&res.dir_attr);
2060 2061 2062
	status = rpc_call_sync(server->client, &msg, 0);
	if (status == 0) {
		update_changeattr(dir, &res.cinfo);
2063
		nfs_post_op_update_inode(dir, &res.dir_attr);
2064
	}
L
Linus Torvalds 已提交
2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079
	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;
}

2080
static void nfs4_proc_unlink_setup(struct rpc_message *msg, struct inode *dir)
L
Linus Torvalds 已提交
2081
{
2082 2083 2084
	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
2085

2086
	args->bitmask = server->cache_consistency_bitmask;
2087
	res->server = server;
L
Linus Torvalds 已提交
2088 2089 2090
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_REMOVE];
}

2091
static int nfs4_proc_unlink_done(struct rpc_task *task, struct inode *dir)
L
Linus Torvalds 已提交
2092
{
2093 2094
	struct nfs_removeres *res = task->tk_msg.rpc_resp;

2095
	if (nfs4_async_handle_error(task, res->server, NULL) == -EAGAIN)
2096 2097 2098 2099
		return 0;
	update_changeattr(dir, &res->cinfo);
	nfs_post_op_update_inode(dir, &res->dir_attr);
	return 1;
L
Linus Torvalds 已提交
2100 2101 2102 2103 2104
}

static int _nfs4_proc_rename(struct inode *old_dir, struct qstr *old_name,
		struct inode *new_dir, struct qstr *new_name)
{
2105
	struct nfs_server *server = NFS_SERVER(old_dir);
L
Linus Torvalds 已提交
2106 2107 2108 2109 2110
	struct nfs4_rename_arg arg = {
		.old_dir = NFS_FH(old_dir),
		.new_dir = NFS_FH(new_dir),
		.old_name = old_name,
		.new_name = new_name,
2111 2112 2113 2114 2115 2116 2117
		.bitmask = server->attr_bitmask,
	};
	struct nfs_fattr old_fattr, new_fattr;
	struct nfs4_rename_res res = {
		.server = server,
		.old_fattr = &old_fattr,
		.new_fattr = &new_fattr,
L
Linus Torvalds 已提交
2118 2119 2120 2121 2122 2123 2124 2125
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_RENAME],
		.rpc_argp = &arg,
		.rpc_resp = &res,
	};
	int			status;
	
2126 2127 2128
	nfs_fattr_init(res.old_fattr);
	nfs_fattr_init(res.new_fattr);
	status = rpc_call_sync(server->client, &msg, 0);
L
Linus Torvalds 已提交
2129 2130 2131

	if (!status) {
		update_changeattr(old_dir, &res.old_cinfo);
2132
		nfs_post_op_update_inode(old_dir, res.old_fattr);
L
Linus Torvalds 已提交
2133
		update_changeattr(new_dir, &res.new_cinfo);
2134
		nfs_post_op_update_inode(new_dir, res.new_fattr);
L
Linus Torvalds 已提交
2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154
	}
	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)
{
2155
	struct nfs_server *server = NFS_SERVER(inode);
L
Linus Torvalds 已提交
2156 2157 2158 2159
	struct nfs4_link_arg arg = {
		.fh     = NFS_FH(inode),
		.dir_fh = NFS_FH(dir),
		.name   = name,
2160 2161 2162 2163 2164 2165 2166
		.bitmask = server->attr_bitmask,
	};
	struct nfs_fattr fattr, dir_attr;
	struct nfs4_link_res res = {
		.server = server,
		.fattr = &fattr,
		.dir_attr = &dir_attr,
L
Linus Torvalds 已提交
2167 2168 2169 2170
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LINK],
		.rpc_argp = &arg,
2171
		.rpc_resp = &res,
L
Linus Torvalds 已提交
2172 2173 2174
	};
	int			status;

2175 2176 2177 2178 2179 2180
	nfs_fattr_init(res.fattr);
	nfs_fattr_init(res.dir_attr);
	status = rpc_call_sync(server->client, &msg, 0);
	if (!status) {
		update_changeattr(dir, &res.cinfo);
		nfs_post_op_update_inode(dir, res.dir_attr);
2181
		nfs_post_op_update_inode(inode, res.fattr);
2182
	}
L
Linus Torvalds 已提交
2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198

	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;
}

2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251
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)
{
	int status = rpc_call_sync(NFS_CLIENT(dir), &data->msg, 0);
	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);
}

2252
static int _nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
2253
		struct page *page, unsigned int len, struct iattr *sattr)
L
Linus Torvalds 已提交
2254
{
2255 2256
	struct nfs4_createdata *data;
	int status = -ENAMETOOLONG;
L
Linus Torvalds 已提交
2257

2258
	if (len > NFS4_MAXPATHLEN)
2259
		goto out;
2260

2261 2262 2263 2264 2265 2266 2267 2268
	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
2269
	
2270 2271 2272 2273
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
2274 2275 2276
	return status;
}

2277
static int nfs4_proc_symlink(struct inode *dir, struct dentry *dentry,
2278
		struct page *page, unsigned int len, struct iattr *sattr)
L
Linus Torvalds 已提交
2279 2280 2281 2282 2283
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(NFS_SERVER(dir),
2284 2285
				_nfs4_proc_symlink(dir, dentry, page,
							len, sattr),
L
Linus Torvalds 已提交
2286 2287 2288 2289 2290 2291 2292 2293
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr)
{
2294 2295
	struct nfs4_createdata *data;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2296

2297 2298 2299 2300 2301 2302 2303 2304
	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
2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329
	return status;
}

static int nfs4_proc_mkdir(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr)
{
	struct nfs4_exception exception = { };
	int err;
	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,
                  u64 cookie, struct page *page, unsigned int count, int plus)
{
	struct inode		*dir = dentry->d_inode;
	struct nfs4_readdir_arg args = {
		.fh = NFS_FH(dir),
		.pages = &page,
		.pgbase = 0,
		.count = count,
2330
		.bitmask = NFS_SERVER(dentry->d_inode)->cache_consistency_bitmask,
L
Linus Torvalds 已提交
2331 2332 2333 2334 2335 2336 2337 2338 2339 2340
	};
	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;

2341
	dprintk("%s: dentry = %s/%s, cookie = %Lu\n", __func__,
2342 2343 2344
			dentry->d_parent->d_name.name,
			dentry->d_name.name,
			(unsigned long long)cookie);
L
Linus Torvalds 已提交
2345 2346 2347 2348 2349
	nfs4_setup_readdir(cookie, NFS_COOKIEVERF(dir), dentry, &args);
	res.pgbase = args.pgbase;
	status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);
	if (status == 0)
		memcpy(NFS_COOKIEVERF(dir), res.verifier.data, NFS4_VERIFIER_SIZE);
2350 2351 2352

	nfs_invalidate_atime(dir);

2353
	dprintk("%s: returns %d\n", __func__, status);
L
Linus Torvalds 已提交
2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373
	return status;
}

static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
                  u64 cookie, struct page *page, unsigned int count, int plus)
{
	struct nfs4_exception exception = { };
	int err;
	do {
		err = nfs4_handle_exception(NFS_SERVER(dentry->d_inode),
				_nfs4_proc_readdir(dentry, cred, cookie,
					page, count, plus),
				&exception);
	} while (exception.retry);
	return err;
}

static int _nfs4_proc_mknod(struct inode *dir, struct dentry *dentry,
		struct iattr *sattr, dev_t rdev)
{
2374 2375 2376
	struct nfs4_createdata *data;
	int mode = sattr->ia_mode;
	int status = -ENOMEM;
L
Linus Torvalds 已提交
2377 2378 2379

	BUG_ON(!(sattr->ia_valid & ATTR_MODE));
	BUG_ON(!S_ISFIFO(mode) && !S_ISBLK(mode) && !S_ISCHR(mode) && !S_ISSOCK(mode));
2380 2381 2382 2383 2384

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

L
Linus Torvalds 已提交
2385
	if (S_ISFIFO(mode))
2386
		data->arg.ftype = NF4FIFO;
L
Linus Torvalds 已提交
2387
	else if (S_ISBLK(mode)) {
2388 2389 2390
		data->arg.ftype = NF4BLK;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
L
Linus Torvalds 已提交
2391 2392
	}
	else if (S_ISCHR(mode)) {
2393 2394 2395
		data->arg.ftype = NF4CHR;
		data->arg.u.device.specdata1 = MAJOR(rdev);
		data->arg.u.device.specdata2 = MINOR(rdev);
L
Linus Torvalds 已提交
2396 2397
	}
	
2398 2399 2400 2401
	status = nfs4_do_create(dir, dentry, data);

	nfs4_free_createdata(data);
out:
L
Linus Torvalds 已提交
2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430
	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;
	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,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_STATFS],
		.rpc_argp = &args,
		.rpc_resp = fsstat,
	};

2431
	nfs_fattr_init(fsstat->fattr);
L
Linus Torvalds 已提交
2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477
	return rpc_call_sync(server->client, &msg, 0);
}

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,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FSINFO],
		.rpc_argp = &args,
		.rpc_resp = fsinfo,
	};

	return rpc_call_sync(server->client, &msg, 0);
}

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)
{
2478
	nfs_fattr_init(fsinfo->fattr);
L
Linus Torvalds 已提交
2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500
	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,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_PATHCONF],
		.rpc_argp = &args,
		.rpc_resp = pathconf,
	};

	/* 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;
	}

2501
	nfs_fattr_init(pathconf->fattr);
L
Linus Torvalds 已提交
2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518
	return rpc_call_sync(server->client, &msg, 0);
}

static int nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle,
		struct nfs_pathconf *pathconf)
{
	struct nfs4_exception exception = { };
	int err;

	do {
		err = nfs4_handle_exception(server,
				_nfs4_proc_pathconf(server, fhandle, pathconf),
				&exception);
	} while (exception.retry);
	return err;
}

T
Trond Myklebust 已提交
2519
static int nfs4_read_done(struct rpc_task *task, struct nfs_read_data *data)
L
Linus Torvalds 已提交
2520
{
T
Trond Myklebust 已提交
2521
	struct nfs_server *server = NFS_SERVER(data->inode);
L
Linus Torvalds 已提交
2522

2523
	if (nfs4_async_handle_error(task, server, data->args.context->state) == -EAGAIN) {
L
Linus Torvalds 已提交
2524
		rpc_restart_call(task);
T
Trond Myklebust 已提交
2525
		return -EAGAIN;
L
Linus Torvalds 已提交
2526
	}
2527 2528

	nfs_invalidate_atime(data->inode);
L
Linus Torvalds 已提交
2529
	if (task->tk_status > 0)
T
Trond Myklebust 已提交
2530 2531
		renew_lease(server, data->timestamp);
	return 0;
L
Linus Torvalds 已提交
2532 2533
}

2534
static void nfs4_proc_read_setup(struct nfs_read_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
2535 2536
{
	data->timestamp   = jiffies;
2537
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_READ];
L
Linus Torvalds 已提交
2538 2539
}

2540
static int nfs4_write_done(struct rpc_task *task, struct nfs_write_data *data)
L
Linus Torvalds 已提交
2541 2542 2543
{
	struct inode *inode = data->inode;
	
2544
	if (nfs4_async_handle_error(task, NFS_SERVER(inode), data->args.context->state) == -EAGAIN) {
L
Linus Torvalds 已提交
2545
		rpc_restart_call(task);
2546
		return -EAGAIN;
L
Linus Torvalds 已提交
2547
	}
2548
	if (task->tk_status >= 0) {
L
Linus Torvalds 已提交
2549
		renew_lease(NFS_SERVER(inode), data->timestamp);
2550
		nfs_post_op_update_inode_force_wcc(inode, data->res.fattr);
2551
	}
2552
	return 0;
L
Linus Torvalds 已提交
2553 2554
}

2555
static void nfs4_proc_write_setup(struct nfs_write_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
2556
{
2557 2558
	struct nfs_server *server = NFS_SERVER(data->inode);

2559
	data->args.bitmask = server->cache_consistency_bitmask;
2560
	data->res.server = server;
L
Linus Torvalds 已提交
2561 2562
	data->timestamp   = jiffies;

2563
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_WRITE];
L
Linus Torvalds 已提交
2564 2565
}

2566
static int nfs4_commit_done(struct rpc_task *task, struct nfs_write_data *data)
L
Linus Torvalds 已提交
2567 2568 2569
{
	struct inode *inode = data->inode;
	
2570
	if (nfs4_async_handle_error(task, NFS_SERVER(inode), NULL) == -EAGAIN) {
L
Linus Torvalds 已提交
2571
		rpc_restart_call(task);
2572
		return -EAGAIN;
L
Linus Torvalds 已提交
2573
	}
2574
	nfs_refresh_inode(inode, data->res.fattr);
2575
	return 0;
L
Linus Torvalds 已提交
2576 2577
}

2578
static void nfs4_proc_commit_setup(struct nfs_write_data *data, struct rpc_message *msg)
L
Linus Torvalds 已提交
2579
{
2580
	struct nfs_server *server = NFS_SERVER(data->inode);
L
Linus Torvalds 已提交
2581
	
2582
	data->args.bitmask = server->cache_consistency_bitmask;
2583
	data->res.server = server;
2584
	msg->rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_COMMIT];
L
Linus Torvalds 已提交
2585 2586 2587 2588 2589 2590
}

/*
 * nfs4_proc_async_renew(): This is not one of the nfs_rpc_ops; it is a special
 * standalone procedure for queueing an asynchronous RENEW.
 */
2591
static void nfs4_renew_done(struct rpc_task *task, void *data)
L
Linus Torvalds 已提交
2592
{
2593
	struct nfs_client *clp = (struct nfs_client *)task->tk_msg.rpc_argp;
2594
	unsigned long timestamp = (unsigned long)data;
L
Linus Torvalds 已提交
2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610

	if (task->tk_status < 0) {
		switch (task->tk_status) {
			case -NFS4ERR_STALE_CLIENTID:
			case -NFS4ERR_EXPIRED:
			case -NFS4ERR_CB_PATH_DOWN:
				nfs4_schedule_state_recovery(clp);
		}
		return;
	}
	spin_lock(&clp->cl_lock);
	if (time_before(clp->cl_last_renewal,timestamp))
		clp->cl_last_renewal = timestamp;
	spin_unlock(&clp->cl_lock);
}

2611 2612 2613 2614
static const struct rpc_call_ops nfs4_renew_ops = {
	.rpc_call_done = nfs4_renew_done,
};

2615
int nfs4_proc_async_renew(struct nfs_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
2616 2617 2618 2619
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
2620
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
2621 2622 2623
	};

	return rpc_call_async(clp->cl_rpcclient, &msg, RPC_TASK_SOFT,
2624
			&nfs4_renew_ops, (void *)jiffies);
L
Linus Torvalds 已提交
2625 2626
}

2627
int nfs4_proc_renew(struct nfs_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
2628 2629 2630 2631
{
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_RENEW],
		.rpc_argp	= clp,
2632
		.rpc_cred	= cred,
L
Linus Torvalds 已提交
2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646
	};
	unsigned long now = jiffies;
	int status;

	status = rpc_call_sync(clp->cl_rpcclient, &msg, 0);
	if (status < 0)
		return status;
	spin_lock(&clp->cl_lock);
	if (time_before(clp->cl_last_renewal,now))
		clp->cl_last_renewal = now;
	spin_unlock(&clp->cl_lock);
	return 0;
}

2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672
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;
	}
}

2673 2674 2675
struct nfs4_cached_acl {
	int cached;
	size_t len;
A
Andrew Morton 已提交
2676
	char data[0];
2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739
};

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);
}

2740
static ssize_t __nfs4_get_acl_uncached(struct inode *inode, void *buf, size_t buflen)
2741 2742 2743 2744 2745 2746 2747 2748
{
	struct page *pages[NFS4ACL_MAXPAGES];
	struct nfs_getaclargs args = {
		.fh = NFS_FH(inode),
		.acl_pages = pages,
		.acl_len = buflen,
	};
	size_t resp_len = buflen;
2749
	void *resp_buf;
2750 2751 2752 2753 2754
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_GETACL],
		.rpc_argp = &args,
		.rpc_resp = &resp_len,
	};
2755
	struct page *localpage = NULL;
2756 2757
	int ret;

2758 2759 2760 2761 2762 2763 2764 2765 2766
	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;
J
J. Bruce Fields 已提交
2767
		resp_len = args.acl_len = PAGE_SIZE;
2768 2769 2770 2771
	} else {
		resp_buf = buf;
		buf_to_pages(buf, buflen, args.acl_pages, &args.acl_pgbase);
	}
2772
	ret = rpc_call_sync(NFS_CLIENT(inode), &msg, 0);
2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789
	if (ret)
		goto out_free;
	if (resp_len > args.acl_len)
		nfs4_write_cached_acl(inode, NULL, resp_len);
	else
		nfs4_write_cached_acl(inode, resp_buf, resp_len);
	if (buf) {
		ret = -ERANGE;
		if (resp_len > buflen)
			goto out_free;
		if (localpage)
			memcpy(buf, resp_buf, resp_len);
	}
	ret = resp_len;
out_free:
	if (localpage)
		__free_page(localpage);
2790 2791 2792
	return ret;
}

2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805
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;
}

2806 2807 2808 2809 2810 2811 2812 2813 2814 2815
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;
2816 2817
	if (NFS_I(inode)->cache_validity & NFS_INO_INVALID_ACL)
		nfs_zap_acl_cache(inode);
2818 2819 2820 2821 2822 2823
	ret = nfs4_read_cached_acl(inode, buf, buflen);
	if (ret != -ENOENT)
		return ret;
	return nfs4_get_acl_uncached(inode, buf, buflen);
}

2824
static int __nfs4_proc_set_acl(struct inode *inode, const void *buf, size_t buflen)
2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841
{
	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,
	};
	struct rpc_message msg = {
		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_SETACL],
		.rpc_argp	= &arg,
		.rpc_resp	= NULL,
	};
	int ret;

	if (!nfs4_server_supports_acls(server))
		return -EOPNOTSUPP;
2842
	nfs_inode_return_delegation(inode);
2843
	buf_to_pages(buf, buflen, arg.acl_pages, &arg.acl_pgbase);
2844
	ret = rpc_call_sync(NFS_CLIENT(inode), &msg, 0);
2845 2846
	nfs_access_zap_cache(inode);
	nfs_zap_acl_cache(inode);
2847 2848 2849
	return ret;
}

2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861
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
2862
static int
2863
nfs4_async_handle_error(struct rpc_task *task, const struct nfs_server *server, struct nfs4_state *state)
L
Linus Torvalds 已提交
2864
{
2865
	struct nfs_client *clp = server->nfs_client;
L
Linus Torvalds 已提交
2866 2867 2868 2869

	if (!clp || task->tk_status >= 0)
		return 0;
	switch(task->tk_status) {
2870 2871 2872 2873 2874 2875
		case -NFS4ERR_ADMIN_REVOKED:
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OPENMODE:
			if (state == NULL)
				break;
			nfs4_state_mark_reclaim_nograce(clp, state);
L
Linus Torvalds 已提交
2876 2877 2878
		case -NFS4ERR_STALE_CLIENTID:
		case -NFS4ERR_STALE_STATEID:
		case -NFS4ERR_EXPIRED:
2879
			rpc_sleep_on(&clp->cl_rpcwaitq, task, NULL);
L
Linus Torvalds 已提交
2880
			nfs4_schedule_state_recovery(clp);
2881
			if (test_bit(NFS4CLNT_MANAGER_RUNNING, &clp->cl_state) == 0)
2882
				rpc_wake_up_queued_task(&clp->cl_rpcwaitq, task);
L
Linus Torvalds 已提交
2883 2884 2885
			task->tk_status = 0;
			return -EAGAIN;
		case -NFS4ERR_DELAY:
2886
			nfs_inc_server_stats(server, NFSIOS_DELAY);
2887
		case -NFS4ERR_GRACE:
L
Linus Torvalds 已提交
2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898
			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;
}

2899
int nfs4_proc_setclientid(struct nfs_client *clp, u32 program, unsigned short port, struct rpc_cred *cred)
L
Linus Torvalds 已提交
2900 2901 2902 2903 2904 2905 2906 2907 2908 2909
{
	nfs4_verifier sc_verifier;
	struct nfs4_setclientid setclientid = {
		.sc_verifier = &sc_verifier,
		.sc_prog = program,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID],
		.rpc_argp = &setclientid,
		.rpc_resp = clp,
2910
		.rpc_cred = cred,
L
Linus Torvalds 已提交
2911
	};
A
Al Viro 已提交
2912
	__be32 *p;
L
Linus Torvalds 已提交
2913 2914 2915
	int loop = 0;
	int status;

A
Al Viro 已提交
2916
	p = (__be32*)sc_verifier.data;
L
Linus Torvalds 已提交
2917 2918 2919 2920 2921
	*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,
2922
				sizeof(setclientid.sc_name), "%s/%s %s %s %u",
2923 2924 2925
				clp->cl_ipaddr,
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_ADDR),
2926 2927
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_PROTO),
2928
				clp->cl_rpcclient->cl_auth->au_ops->au_name,
L
Linus Torvalds 已提交
2929 2930
				clp->cl_id_uniquifier);
		setclientid.sc_netid_len = scnprintf(setclientid.sc_netid,
2931 2932 2933
				sizeof(setclientid.sc_netid),
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_NETID));
L
Linus Torvalds 已提交
2934
		setclientid.sc_uaddr_len = scnprintf(setclientid.sc_uaddr,
2935
				sizeof(setclientid.sc_uaddr), "%s.%u.%u",
L
Linus Torvalds 已提交
2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951
				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)
			ssleep(clp->cl_lease_time + 1);
		else
			if (++clp->cl_id_uniquifier == 0)
				break;
	}
	return status;
}

2952
static int _nfs4_proc_setclientid_confirm(struct nfs_client *clp, struct rpc_cred *cred)
L
Linus Torvalds 已提交
2953 2954 2955 2956 2957 2958
{
	struct nfs_fsinfo fsinfo;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID_CONFIRM],
		.rpc_argp = clp,
		.rpc_resp = &fsinfo,
2959
		.rpc_cred = cred,
L
Linus Torvalds 已提交
2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974
	};
	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;
}

2975
int nfs4_proc_setclientid_confirm(struct nfs_client *clp, struct rpc_cred *cred)
2976
{
2977
	long timeout = 0;
2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992
	int err;
	do {
		err = _nfs4_proc_setclientid_confirm(clp, cred);
		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;
}

2993 2994
struct nfs4_delegreturndata {
	struct nfs4_delegreturnargs args;
2995
	struct nfs4_delegreturnres res;
2996 2997
	struct nfs_fh fh;
	nfs4_stateid stateid;
2998
	unsigned long timestamp;
2999
	struct nfs_fattr fattr;
3000 3001 3002 3003 3004 3005 3006
	int rpc_status;
};

static void nfs4_delegreturn_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_delegreturndata *data = calldata;
	data->rpc_status = task->tk_status;
3007
	if (data->rpc_status == 0)
3008
		renew_lease(data->res.server, data->timestamp);
3009 3010 3011 3012 3013 3014 3015
}

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

3016
static const struct rpc_call_ops nfs4_delegreturn_ops = {
3017 3018 3019 3020
	.rpc_call_done = nfs4_delegreturn_done,
	.rpc_release = nfs4_delegreturn_release,
};

3021
static int _nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
3022 3023
{
	struct nfs4_delegreturndata *data;
3024
	struct nfs_server *server = NFS_SERVER(inode);
3025
	struct rpc_task *task;
3026 3027 3028 3029
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DELEGRETURN],
		.rpc_cred = cred,
	};
T
Trond Myklebust 已提交
3030 3031
	struct rpc_task_setup task_setup_data = {
		.rpc_client = server->client,
3032
		.rpc_message = &msg,
T
Trond Myklebust 已提交
3033 3034 3035
		.callback_ops = &nfs4_delegreturn_ops,
		.flags = RPC_TASK_ASYNC,
	};
3036
	int status = 0;
3037 3038 3039 3040 3041 3042

	data = kmalloc(sizeof(*data), GFP_KERNEL);
	if (data == NULL)
		return -ENOMEM;
	data->args.fhandle = &data->fh;
	data->args.stateid = &data->stateid;
3043
	data->args.bitmask = server->attr_bitmask;
3044 3045
	nfs_copy_fh(&data->fh, NFS_FH(inode));
	memcpy(&data->stateid, stateid, sizeof(data->stateid));
3046 3047
	data->res.fattr = &data->fattr;
	data->res.server = server;
3048
	nfs_fattr_init(data->res.fattr);
3049
	data->timestamp = jiffies;
3050 3051
	data->rpc_status = 0;

T
Trond Myklebust 已提交
3052
	task_setup_data.callback_data = data;
3053 3054
	msg.rpc_argp = &data->args,
	msg.rpc_resp = &data->res,
T
Trond Myklebust 已提交
3055
	task = rpc_run_task(&task_setup_data);
3056
	if (IS_ERR(task))
3057
		return PTR_ERR(task);
3058 3059
	if (!issync)
		goto out;
3060
	status = nfs4_wait_for_completion_rpc_task(task);
3061 3062 3063 3064 3065 3066 3067
	if (status != 0)
		goto out;
	status = data->rpc_status;
	if (status != 0)
		goto out;
	nfs_refresh_inode(inode, &data->fattr);
out:
3068
	rpc_put_task(task);
3069
	return status;
L
Linus Torvalds 已提交
3070 3071
}

3072
int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid, int issync)
L
Linus Torvalds 已提交
3073 3074 3075 3076 3077
{
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs4_exception exception = { };
	int err;
	do {
3078
		err = _nfs4_proc_delegreturn(inode, cred, stateid, issync);
L
Linus Torvalds 已提交
3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098
		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)
{
3099
	schedule_timeout_killable(timeout);
L
Linus Torvalds 已提交
3100 3101 3102 3103 3104 3105 3106 3107 3108 3109
	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);
3110
	struct nfs_client *clp = server->nfs_client;
T
Trond Myklebust 已提交
3111
	struct nfs_lockt_args arg = {
L
Linus Torvalds 已提交
3112
		.fh = NFS_FH(inode),
T
Trond Myklebust 已提交
3113
		.fl = request,
L
Linus Torvalds 已提交
3114
	};
T
Trond Myklebust 已提交
3115 3116
	struct nfs_lockt_res res = {
		.denied = request,
L
Linus Torvalds 已提交
3117 3118 3119 3120 3121 3122 3123 3124 3125 3126
	};
	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
3127
	arg.lock_owner.clientid = clp->cl_clientid;
3128 3129 3130 3131
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
	lsp = request->fl_u.nfs4_fl.owner;
3132
	arg.lock_owner.id = lsp->ls_id.id;
L
Linus Torvalds 已提交
3133
	status = rpc_call_sync(server->client, &msg, 0);
T
Trond Myklebust 已提交
3134 3135 3136 3137 3138 3139
	switch (status) {
		case 0:
			request->fl_type = F_UNLCK;
			break;
		case -NFS4ERR_DENIED:
			status = 0;
L
Linus Torvalds 已提交
3140
	}
3141
	request->fl_ops->fl_release_private(request);
3142
out:
L
Linus Torvalds 已提交
3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174
	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;
}

3175
struct nfs4_unlockdata {
T
Trond Myklebust 已提交
3176 3177
	struct nfs_locku_args arg;
	struct nfs_locku_res res;
3178 3179
	struct nfs4_lock_state *lsp;
	struct nfs_open_context *ctx;
T
Trond Myklebust 已提交
3180 3181
	struct file_lock fl;
	const struct nfs_server *server;
3182
	unsigned long timestamp;
3183 3184
};

T
Trond Myklebust 已提交
3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198
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;

	p = kmalloc(sizeof(*p), GFP_KERNEL);
	if (p == NULL)
		return NULL;
	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
	p->arg.seqid = seqid;
3199
	p->res.seqid = seqid;
T
Trond Myklebust 已提交
3200 3201 3202 3203 3204 3205 3206 3207 3208 3209
	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;
}

3210
static void nfs4_locku_release_calldata(void *data)
3211
{
3212
	struct nfs4_unlockdata *calldata = data;
T
Trond Myklebust 已提交
3213
	nfs_free_seqid(calldata->arg.seqid);
3214 3215 3216
	nfs4_put_lock_state(calldata->lsp);
	put_nfs_open_context(calldata->ctx);
	kfree(calldata);
3217 3218
}

3219
static void nfs4_locku_done(struct rpc_task *task, void *data)
3220
{
3221
	struct nfs4_unlockdata *calldata = data;
3222

3223 3224
	if (RPC_ASSASSINATED(task))
		return;
3225 3226 3227
	switch (task->tk_status) {
		case 0:
			memcpy(calldata->lsp->ls_stateid.data,
T
Trond Myklebust 已提交
3228
					calldata->res.stateid.data,
3229
					sizeof(calldata->lsp->ls_stateid.data));
3230
			renew_lease(calldata->server, calldata->timestamp);
3231
			break;
3232 3233
		case -NFS4ERR_BAD_STATEID:
		case -NFS4ERR_OLD_STATEID:
3234 3235 3236 3237
		case -NFS4ERR_STALE_STATEID:
		case -NFS4ERR_EXPIRED:
			break;
		default:
3238
			if (nfs4_async_handle_error(task, calldata->server, NULL) == -EAGAIN)
3239 3240 3241 3242
				rpc_restart_call(task);
	}
}

T
Trond Myklebust 已提交
3243
static void nfs4_locku_prepare(struct rpc_task *task, void *data)
3244
{
T
Trond Myklebust 已提交
3245
	struct nfs4_unlockdata *calldata = data;
3246

T
Trond Myklebust 已提交
3247
	if (nfs_wait_on_sequence(calldata->arg.seqid, task) != 0)
3248 3249
		return;
	if ((calldata->lsp->ls_flags & NFS_LOCK_INITIALIZED) == 0) {
3250 3251
		/* Note: exit _without_ running nfs4_locku_done */
		task->tk_action = NULL;
3252 3253
		return;
	}
3254
	calldata->timestamp = jiffies;
3255
	rpc_call_start(task);
3256 3257
}

3258
static const struct rpc_call_ops nfs4_locku_ops = {
T
Trond Myklebust 已提交
3259
	.rpc_call_prepare = nfs4_locku_prepare,
3260
	.rpc_call_done = nfs4_locku_done,
3261
	.rpc_release = nfs4_locku_release_calldata,
3262 3263
};

3264 3265 3266 3267 3268 3269
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;
3270 3271 3272 3273
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCKU],
		.rpc_cred = ctx->cred,
	};
T
Trond Myklebust 已提交
3274 3275
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(lsp->ls_state->inode),
3276
		.rpc_message = &msg,
T
Trond Myklebust 已提交
3277
		.callback_ops = &nfs4_locku_ops,
3278
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
3279 3280
		.flags = RPC_TASK_ASYNC,
	};
3281

3282 3283 3284 3285 3286
	/* 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;

3287 3288 3289 3290 3291 3292
	data = nfs4_alloc_unlockdata(fl, ctx, lsp, seqid);
	if (data == NULL) {
		nfs_free_seqid(seqid);
		return ERR_PTR(-ENOMEM);
	}

3293 3294
	msg.rpc_argp = &data->arg,
	msg.rpc_resp = &data->res,
T
Trond Myklebust 已提交
3295 3296
	task_setup_data.callback_data = data;
	return rpc_run_task(&task_setup_data);
3297 3298
}

3299 3300
static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
{
3301
	struct nfs_inode *nfsi = NFS_I(state->inode);
T
Trond Myklebust 已提交
3302
	struct nfs_seqid *seqid;
L
Linus Torvalds 已提交
3303
	struct nfs4_lock_state *lsp;
3304 3305
	struct rpc_task *task;
	int status = 0;
3306
	unsigned char fl_flags = request->fl_flags;
3307

3308
	status = nfs4_set_lock_state(state, request);
3309 3310
	/* Unlock _before_ we do the RPC call */
	request->fl_flags |= FL_EXISTS;
3311 3312 3313
	down_read(&nfsi->rwsem);
	if (do_vfs_lock(request->fl_file, request) == -ENOENT) {
		up_read(&nfsi->rwsem);
3314
		goto out;
3315 3316
	}
	up_read(&nfsi->rwsem);
3317
	if (status != 0)
3318 3319 3320 3321
		goto out;
	/* Is this a delegated lock? */
	if (test_bit(NFS_DELEGATED_STATE, &state->flags))
		goto out;
3322
	lsp = request->fl_u.nfs4_fl.owner;
T
Trond Myklebust 已提交
3323
	seqid = nfs_alloc_seqid(&lsp->ls_seqid);
3324
	status = -ENOMEM;
T
Trond Myklebust 已提交
3325
	if (seqid == NULL)
3326
		goto out;
3327
	task = nfs4_do_unlck(request, nfs_file_open_context(request->fl_file), lsp, seqid);
3328 3329
	status = PTR_ERR(task);
	if (IS_ERR(task))
3330
		goto out;
3331
	status = nfs4_wait_for_completion_rpc_task(task);
3332
	rpc_put_task(task);
3333
out:
3334
	request->fl_flags = fl_flags;
L
Linus Torvalds 已提交
3335 3336 3337
	return status;
}

3338 3339 3340 3341 3342 3343
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;
3344
	unsigned long timestamp;
3345 3346 3347 3348 3349 3350
	int rpc_status;
	int cancelled;
};

static struct nfs4_lockdata *nfs4_alloc_lockdata(struct file_lock *fl,
		struct nfs_open_context *ctx, struct nfs4_lock_state *lsp)
L
Linus Torvalds 已提交
3351
{
3352 3353
	struct nfs4_lockdata *p;
	struct inode *inode = lsp->ls_state->inode;
L
Linus Torvalds 已提交
3354
	struct nfs_server *server = NFS_SERVER(inode);
3355 3356 3357 3358 3359 3360 3361

	p = kzalloc(sizeof(*p), GFP_KERNEL);
	if (p == NULL)
		return NULL;

	p->arg.fh = NFS_FH(inode);
	p->arg.fl = &p->fl;
3362 3363 3364
	p->arg.open_seqid = nfs_alloc_seqid(&lsp->ls_state->owner->so_seqid);
	if (p->arg.open_seqid == NULL)
		goto out_free;
3365 3366
	p->arg.lock_seqid = nfs_alloc_seqid(&lsp->ls_seqid);
	if (p->arg.lock_seqid == NULL)
3367
		goto out_free_seqid;
3368
	p->arg.lock_stateid = &lsp->ls_stateid;
3369
	p->arg.lock_owner.clientid = server->nfs_client->cl_clientid;
3370
	p->arg.lock_owner.id = lsp->ls_id.id;
3371
	p->res.lock_seqid = p->arg.lock_seqid;
3372 3373 3374 3375 3376
	p->lsp = lsp;
	atomic_inc(&lsp->ls_count);
	p->ctx = get_nfs_open_context(ctx);
	memcpy(&p->fl, fl, sizeof(p->fl));
	return p;
3377 3378
out_free_seqid:
	nfs_free_seqid(p->arg.open_seqid);
3379 3380 3381 3382 3383 3384 3385 3386 3387
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;
3388

3389
	dprintk("%s: begin!\n", __func__);
3390 3391
	if (nfs_wait_on_sequence(data->arg.lock_seqid, task) != 0)
		return;
3392 3393
	/* Do we need to do an open_to_lock_owner? */
	if (!(data->arg.lock_seqid->sequence->flags & NFS_SEQID_CONFIRMED)) {
3394 3395
		if (nfs_wait_on_sequence(data->arg.open_seqid, task) != 0)
			return;
3396 3397
		data->arg.open_stateid = &state->stateid;
		data->arg.new_lock_owner = 1;
3398
		data->res.open_seqid = data->arg.open_seqid;
3399 3400
	} else
		data->arg.new_lock_owner = 0;
3401
	data->timestamp = jiffies;
3402
	rpc_call_start(task);
3403
	dprintk("%s: done!, ret = %d\n", __func__, data->rpc_status);
3404 3405 3406 3407 3408 3409
}

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

3410
	dprintk("%s: begin!\n", __func__);
3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424

	data->rpc_status = task->tk_status;
	if (RPC_ASSASSINATED(task))
		goto out;
	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;
3425
		renew_lease(NFS_SERVER(data->ctx->path.dentry->d_inode), data->timestamp);
3426 3427
	}
out:
3428
	dprintk("%s: done, ret = %d!\n", __func__, data->rpc_status);
3429 3430 3431 3432 3433 3434
}

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

3435
	dprintk("%s: begin!\n", __func__);
3436
	nfs_free_seqid(data->arg.open_seqid);
3437 3438 3439 3440 3441
	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))
3442
			rpc_put_task(task);
3443
		dprintk("%s: cancelling lock!\n", __func__);
3444 3445 3446 3447 3448
	} else
		nfs_free_seqid(data->arg.lock_seqid);
	nfs4_put_lock_state(data->lsp);
	put_nfs_open_context(data->ctx);
	kfree(data);
3449
	dprintk("%s: done!\n", __func__);
3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461
}

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,
};

static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *fl, int reclaim)
{
	struct nfs4_lockdata *data;
	struct rpc_task *task;
3462 3463 3464 3465
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_LOCK],
		.rpc_cred = state->owner->so_cred,
	};
T
Trond Myklebust 已提交
3466 3467
	struct rpc_task_setup task_setup_data = {
		.rpc_client = NFS_CLIENT(state->inode),
3468
		.rpc_message = &msg,
T
Trond Myklebust 已提交
3469
		.callback_ops = &nfs4_lock_ops,
3470
		.workqueue = nfsiod_workqueue,
T
Trond Myklebust 已提交
3471 3472
		.flags = RPC_TASK_ASYNC,
	};
3473 3474
	int ret;

3475
	dprintk("%s: begin!\n", __func__);
3476
	data = nfs4_alloc_lockdata(fl, nfs_file_open_context(fl->fl_file),
3477 3478 3479 3480 3481 3482 3483
			fl->fl_u.nfs4_fl.owner);
	if (data == NULL)
		return -ENOMEM;
	if (IS_SETLKW(cmd))
		data->arg.block = 1;
	if (reclaim != 0)
		data->arg.reclaim = 1;
3484 3485
	msg.rpc_argp = &data->arg,
	msg.rpc_resp = &data->res,
T
Trond Myklebust 已提交
3486 3487
	task_setup_data.callback_data = data;
	task = rpc_run_task(&task_setup_data);
3488
	if (IS_ERR(task))
3489 3490 3491 3492 3493 3494 3495 3496
		return PTR_ERR(task);
	ret = nfs4_wait_for_completion_rpc_task(task);
	if (ret == 0) {
		ret = data->rpc_status;
		if (ret == -NFS4ERR_DENIED)
			ret = -EAGAIN;
	} else
		data->cancelled = 1;
3497
	rpc_put_task(task);
3498
	dprintk("%s: done, ret = %d!\n", __func__, ret);
3499
	return ret;
L
Linus Torvalds 已提交
3500 3501 3502 3503
}

static int nfs4_lock_reclaim(struct nfs4_state *state, struct file_lock *request)
{
3504 3505 3506 3507 3508
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;

	do {
3509 3510 3511
		/* Cache the lock if possible... */
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
3512 3513 3514 3515 3516 3517
		err = _nfs4_do_setlk(state, F_SETLK, request, 1);
		if (err != -NFS4ERR_DELAY)
			break;
		nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
	return err;
L
Linus Torvalds 已提交
3518 3519 3520 3521
}

static int nfs4_lock_expired(struct nfs4_state *state, struct file_lock *request)
{
3522 3523 3524 3525
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;

3526 3527 3528
	err = nfs4_set_lock_state(state, request);
	if (err != 0)
		return err;
3529
	do {
3530 3531
		if (test_bit(NFS_DELEGATED_STATE, &state->flags) != 0)
			return 0;
3532 3533 3534 3535 3536 3537
		err = _nfs4_do_setlk(state, F_SETLK, request, 0);
		if (err != -NFS4ERR_DELAY)
			break;
		nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
	return err;
L
Linus Torvalds 已提交
3538 3539 3540 3541
}

static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
{
3542
	struct nfs_inode *nfsi = NFS_I(state->inode);
3543
	unsigned char fl_flags = request->fl_flags;
L
Linus Torvalds 已提交
3544 3545
	int status;

3546 3547 3548 3549
	/* Is this a delegated open? */
	status = nfs4_set_lock_state(state, request);
	if (status != 0)
		goto out;
3550 3551 3552 3553
	request->fl_flags |= FL_ACCESS;
	status = do_vfs_lock(request->fl_file, request);
	if (status < 0)
		goto out;
3554
	down_read(&nfsi->rwsem);
3555 3556 3557
	if (test_bit(NFS_DELEGATED_STATE, &state->flags)) {
		/* Yes: cache locks! */
		/* ...but avoid races with delegation recall... */
3558 3559 3560
		request->fl_flags = fl_flags & ~FL_SLEEP;
		status = do_vfs_lock(request->fl_file, request);
		goto out_unlock;
3561
	}
3562 3563
	status = _nfs4_do_setlk(state, cmd, request, 0);
	if (status != 0)
3564
		goto out_unlock;
3565
	/* Note: we always want to sleep here! */
3566
	request->fl_flags = fl_flags | FL_SLEEP;
3567
	if (do_vfs_lock(request->fl_file, request) < 0)
3568
		printk(KERN_WARNING "%s: VFS is out of sync with lock manager!\n", __func__);
3569
out_unlock:
3570
	up_read(&nfsi->rwsem);
3571 3572
out:
	request->fl_flags = fl_flags;
L
Linus Torvalds 已提交
3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597
	return status;
}

static int nfs4_proc_setlk(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_setlk(state, cmd, request),
				&exception);
	} 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 */
3598
	ctx = nfs_file_open_context(filp);
L
Linus Torvalds 已提交
3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624
	state = ctx->state;

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

	if (IS_GETLK(cmd))
		return nfs4_proc_getlk(state, F_GETLK, request);

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

	if (request->fl_type == F_UNLCK)
		return nfs4_proc_unlck(state, cmd, request);

	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;
}

3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642
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 {
		err = _nfs4_do_setlk(state, F_SETLK, fl, 0);
		if (err != -NFS4ERR_DELAY)
			break;
		err = nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
out:
	return err;
}
3643

3644 3645
#define XATTR_NAME_NFSV4_ACL "system.nfs4_acl"

3646 3647 3648
int nfs4_setxattr(struct dentry *dentry, const char *key, const void *buf,
		size_t buflen, int flags)
{
3649 3650 3651 3652 3653 3654
	struct inode *inode = dentry->d_inode;

	if (strcmp(key, XATTR_NAME_NFSV4_ACL) != 0)
		return -EOPNOTSUPP;

	return nfs4_proc_set_acl(inode, buf, buflen);
3655 3656 3657 3658 3659 3660 3661 3662 3663
}

/* The getxattr man page suggests returning -ENODATA for unknown attributes,
 * and that's what we'll do for e.g. user attributes that haven't been set.
 * But we'll follow ext2/ext3's lead by returning -EOPNOTSUPP for unsupported
 * attributes in kernel-managed attribute namespaces. */
ssize_t nfs4_getxattr(struct dentry *dentry, const char *key, void *buf,
		size_t buflen)
{
3664 3665 3666 3667 3668 3669
	struct inode *inode = dentry->d_inode;

	if (strcmp(key, XATTR_NAME_NFSV4_ACL) != 0)
		return -EOPNOTSUPP;

	return nfs4_proc_get_acl(inode, buf, buflen);
3670 3671 3672 3673
}

ssize_t nfs4_listxattr(struct dentry *dentry, char *buf, size_t buflen)
{
3674
	size_t len = strlen(XATTR_NAME_NFSV4_ACL) + 1;
3675

3676 3677
	if (!nfs4_server_supports_acls(NFS_SERVER(dentry->d_inode)))
		return 0;
3678 3679 3680
	if (buf && buflen < len)
		return -ERANGE;
	if (buf)
3681 3682
		memcpy(buf, XATTR_NAME_NFSV4_ACL, len);
	return len;
3683 3684
}

3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697
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;
}

3698
int nfs4_proc_fs_locations(struct inode *dir, const struct qstr *name,
3699
		struct nfs4_fs_locations *fs_locations, struct page *page)
3700 3701 3702
{
	struct nfs_server *server = NFS_SERVER(dir);
	u32 bitmask[2] = {
3703 3704
		[0] = FATTR4_WORD0_FSID | FATTR4_WORD0_FS_LOCATIONS,
		[1] = FATTR4_WORD1_MOUNTED_ON_FILEID,
3705 3706 3707
	};
	struct nfs4_fs_locations_arg args = {
		.dir_fh = NFS_FH(dir),
3708
		.name = name,
3709 3710 3711 3712 3713 3714
		.page = page,
		.bitmask = bitmask,
	};
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_FS_LOCATIONS],
		.rpc_argp = &args,
3715
		.rpc_resp = fs_locations,
3716 3717 3718
	};
	int status;

3719
	dprintk("%s: start\n", __func__);
3720
	nfs_fattr_init(&fs_locations->fattr);
3721
	fs_locations->server = server;
3722
	fs_locations->nlocations = 0;
3723
	status = rpc_call_sync(server->client, &msg, 0);
3724
	nfs_fixup_referral_attributes(&fs_locations->fattr);
3725
	dprintk("%s: returned status = %d\n", __func__, status);
3726 3727 3728
	return status;
}

L
Linus Torvalds 已提交
3729
struct nfs4_state_recovery_ops nfs4_reboot_recovery_ops = {
3730
	.owner_flag_bit = NFS_OWNER_RECLAIM_REBOOT,
3731
	.state_flag_bit	= NFS_STATE_RECLAIM_REBOOT,
L
Linus Torvalds 已提交
3732 3733 3734 3735
	.recover_open	= nfs4_open_reclaim,
	.recover_lock	= nfs4_lock_reclaim,
};

3736
struct nfs4_state_recovery_ops nfs4_nograce_recovery_ops = {
3737
	.owner_flag_bit = NFS_OWNER_RECLAIM_NOGRACE,
3738
	.state_flag_bit	= NFS_STATE_RECLAIM_NOGRACE,
L
Linus Torvalds 已提交
3739 3740 3741 3742
	.recover_open	= nfs4_open_expired,
	.recover_lock	= nfs4_lock_expired,
};

3743
static const struct inode_operations nfs4_file_inode_operations = {
3744 3745 3746 3747 3748 3749 3750 3751
	.permission	= nfs_permission,
	.getattr	= nfs_getattr,
	.setattr	= nfs_setattr,
	.getxattr	= nfs4_getxattr,
	.setxattr	= nfs4_setxattr,
	.listxattr	= nfs4_listxattr,
};

D
David Howells 已提交
3752
const struct nfs_rpc_ops nfs_v4_clientops = {
L
Linus Torvalds 已提交
3753 3754 3755
	.version	= 4,			/* protocol version */
	.dentry_ops	= &nfs4_dentry_operations,
	.dir_inode_ops	= &nfs4_dir_inode_operations,
3756
	.file_inode_ops	= &nfs4_file_inode_operations,
L
Linus Torvalds 已提交
3757 3758 3759
	.getroot	= nfs4_proc_get_root,
	.getattr	= nfs4_proc_getattr,
	.setattr	= nfs4_proc_setattr,
D
David Howells 已提交
3760
	.lookupfh	= nfs4_proc_lookupfh,
L
Linus Torvalds 已提交
3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777
	.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,
	.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,
3778
	.set_capabilities = nfs4_server_capabilities,
L
Linus Torvalds 已提交
3779 3780
	.decode_dirent	= nfs4_decode_dirent,
	.read_setup	= nfs4_proc_read_setup,
T
Trond Myklebust 已提交
3781
	.read_done	= nfs4_read_done,
L
Linus Torvalds 已提交
3782
	.write_setup	= nfs4_proc_write_setup,
3783
	.write_done	= nfs4_write_done,
L
Linus Torvalds 已提交
3784
	.commit_setup	= nfs4_proc_commit_setup,
3785
	.commit_done	= nfs4_commit_done,
L
Linus Torvalds 已提交
3786
	.lock		= nfs4_proc_lock,
3787
	.clear_acl_cache = nfs4_zap_acl_attr,
T
Trond Myklebust 已提交
3788
	.close_context  = nfs4_close_context,
L
Linus Torvalds 已提交
3789 3790 3791 3792 3793 3794 3795
};

/*
 * Local variables:
 *  c-basic-offset: 8
 * End:
 */