xfs_attr.c 42.6 KB
Newer Older
D
Dave Chinner 已提交
1
// SPDX-License-Identifier: GPL-2.0
L
Linus Torvalds 已提交
2
/*
3 4
 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
 * All Rights Reserved.
L
Linus Torvalds 已提交
5 6
 */
#include "xfs.h"
7
#include "xfs_fs.h"
8
#include "xfs_shared.h"
9 10 11
#include "xfs_format.h"
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
L
Linus Torvalds 已提交
12
#include "xfs_mount.h"
13
#include "xfs_defer.h"
14
#include "xfs_da_format.h"
15 16
#include "xfs_da_btree.h"
#include "xfs_attr_sf.h"
L
Linus Torvalds 已提交
17
#include "xfs_inode.h"
18
#include "xfs_trans.h"
L
Linus Torvalds 已提交
19
#include "xfs_bmap.h"
20
#include "xfs_bmap_btree.h"
L
Linus Torvalds 已提交
21 22
#include "xfs_attr.h"
#include "xfs_attr_leaf.h"
23
#include "xfs_attr_remote.h"
L
Linus Torvalds 已提交
24 25
#include "xfs_quota.h"
#include "xfs_trans_space.h"
C
Christoph Hellwig 已提交
26
#include "xfs_trace.h"
27
#include "xfs_attr_item.h"
28
#include "xfs_log.h"
29 30 31

struct kmem_cache		*xfs_attri_cache;
struct kmem_cache		*xfs_attrd_cache;
L
Linus Torvalds 已提交
32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50

/*
 * xfs_attr.c
 *
 * Provide the external interfaces to manage attribute lists.
 */

/*========================================================================
 * Function prototypes for the kernel.
 *========================================================================*/

/*
 * Internal routines when attribute list fits inside the inode.
 */
STATIC int xfs_attr_shortform_addname(xfs_da_args_t *args);

/*
 * Internal routines when attribute list is one block.
 */
51
STATIC int xfs_attr_leaf_get(xfs_da_args_t *args);
L
Linus Torvalds 已提交
52
STATIC int xfs_attr_leaf_removename(xfs_da_args_t *args);
53
STATIC int xfs_attr_leaf_hasname(struct xfs_da_args *args, struct xfs_buf **bp);
54
STATIC int xfs_attr_leaf_try_add(struct xfs_da_args *args, struct xfs_buf *bp);
L
Linus Torvalds 已提交
55 56 57 58

/*
 * Internal routines when attribute list is more than one block.
 */
59
STATIC int xfs_attr_node_get(xfs_da_args_t *args);
60
STATIC void xfs_attr_restore_rmt_blk(struct xfs_da_args *args);
61 62 63
STATIC int xfs_attr_node_addname(struct xfs_attr_item *attr);
STATIC int xfs_attr_node_addname_find_attr(struct xfs_attr_item *attr);
STATIC int xfs_attr_node_addname_clear_incomplete(struct xfs_attr_item *attr);
64 65
STATIC int xfs_attr_node_hasname(xfs_da_args_t *args,
				 struct xfs_da_state **state);
L
Linus Torvalds 已提交
66 67
STATIC int xfs_attr_fillstate(xfs_da_state_t *state);
STATIC int xfs_attr_refillstate(xfs_da_state_t *state);
68 69
STATIC int xfs_attr_node_removename(struct xfs_da_args *args,
				    struct xfs_da_state *state);
L
Linus Torvalds 已提交
70

71
int
72 73 74 75
xfs_inode_hasattr(
	struct xfs_inode	*ip)
{
	if (!XFS_IFORK_Q(ip) ||
76
	    (ip->i_afp->if_format == XFS_DINODE_FMT_EXTENTS &&
77
	     ip->i_afp->if_nextents == 0))
78 79 80 81
		return 0;
	return 1;
}

82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101
/*
 * Returns true if the there is exactly only block in the attr fork, in which
 * case the attribute fork consists of a single leaf block entry.
 */
bool
xfs_attr_is_leaf(
	struct xfs_inode	*ip)
{
	struct xfs_ifork	*ifp = ip->i_afp;
	struct xfs_iext_cursor	icur;
	struct xfs_bmbt_irec	imap;

	if (ifp->if_nextents != 1 || ifp->if_format != XFS_DINODE_FMT_EXTENTS)
		return false;

	xfs_iext_first(ifp, &icur);
	xfs_iext_get_extent(ifp, &icur, &imap);
	return imap.br_startoff == 0 && imap.br_blockcount == 1;
}

L
Linus Torvalds 已提交
102 103 104 105
/*========================================================================
 * Overall external interface routines.
 *========================================================================*/

106 107 108 109
/*
 * Retrieve an extended attribute and its value.  Must have ilock.
 * Returns 0 on successful retrieval, otherwise an error.
 */
110 111 112 113
int
xfs_attr_get_ilocked(
	struct xfs_da_args	*args)
{
114
	ASSERT(xfs_isilocked(args->dp, XFS_ILOCK_SHARED | XFS_ILOCK_EXCL));
115

116
	if (!xfs_inode_hasattr(args->dp))
117
		return -ENOATTR;
118

119
	if (args->dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL)
120
		return xfs_attr_shortform_getvalue(args);
121
	if (xfs_attr_is_leaf(args->dp))
122
		return xfs_attr_leaf_get(args);
123
	return xfs_attr_node_get(args);
124 125
}

126 127 128
/*
 * Retrieve an extended attribute by name, and its value if requested.
 *
C
Christoph Hellwig 已提交
129 130 131
 * If args->valuelen is zero, then the caller does not want the value, just an
 * indication whether the attribute exists and the size of the value if it
 * exists. The size is returned in args.valuelen.
132
 *
133 134 135 136 137
 * If args->value is NULL but args->valuelen is non-zero, allocate the buffer
 * for the value after existence of the attribute has been determined. The
 * caller always has to free args->value if it is set, no matter if this
 * function was successful or not.
 *
138
 * If the attribute is found, but exceeds the size limit set by the caller in
139 140
 * args->valuelen, return -ERANGE with the size of the attribute that was found
 * in args->valuelen.
141
 */
142 143
int
xfs_attr_get(
144
	struct xfs_da_args	*args)
L
Linus Torvalds 已提交
145
{
146 147 148
	uint			lock_mode;
	int			error;

149
	XFS_STATS_INC(args->dp->i_mount, xs_attr_get);
150

151
	if (xfs_is_shutdown(args->dp->i_mount))
D
Dave Chinner 已提交
152
		return -EIO;
L
Linus Torvalds 已提交
153

154 155 156
	args->geo = args->dp->i_mount->m_attr_geo;
	args->whichfork = XFS_ATTR_FORK;
	args->hashval = xfs_da_hashname(args->name, args->namelen);
157

158
	/* Entirely possible to look up a name which doesn't exist */
159
	args->op_flags = XFS_DA_OP_OKNOENT;
L
Linus Torvalds 已提交
160

161
	lock_mode = xfs_ilock_attr_map_shared(args->dp);
162
	error = xfs_attr_get_ilocked(args);
163
	xfs_iunlock(args->dp, lock_mode);
164

165
	return error;
L
Linus Torvalds 已提交
166 167
}

168 169 170
/*
 * Calculate how many blocks we need for the new attribute,
 */
171
int
172
xfs_attr_calc_size(
173
	struct xfs_da_args	*args,
174 175
	int			*local)
{
176
	struct xfs_mount	*mp = args->dp->i_mount;
177 178 179 180 181 182 183
	int			size;
	int			nblks;

	/*
	 * Determine space new attribute will use, and if it would be
	 * "local" or "remote" (note: local != inline).
	 */
184
	size = xfs_attr_leaf_newentsize(args, local);
185 186
	nblks = XFS_DAENTER_SPACE_RES(mp, XFS_ATTR_FORK);
	if (*local) {
187
		if (size > (args->geo->blksize / 2)) {
188 189 190 191 192 193 194 195
			/* Double split possible */
			nblks *= 2;
		}
	} else {
		/*
		 * Out of line attribute, cannot double split, but
		 * make room for the attribute value itself.
		 */
196
		uint	dblocks = xfs_attr3_rmt_blocks(mp, args->valuelen);
197 198 199 200 201 202 203
		nblks += dblocks;
		nblks += XFS_NEXTENTADD_SPACE_RES(mp, dblocks, XFS_ATTR_FORK);
	}

	return nblks;
}

204 205 206 207 208 209
STATIC int
xfs_attr_try_sf_addname(
	struct xfs_inode	*dp,
	struct xfs_da_args	*args)
{

210 211 212 213 214 215 216
	int			error;

	/*
	 * Build initial attribute list (if required).
	 */
	if (dp->i_afp->if_format == XFS_DINODE_FMT_EXTENTS)
		xfs_attr_shortform_create(args);
217 218 219 220 221 222 223 224 225

	error = xfs_attr_shortform_addname(args);
	if (error == -ENOSPC)
		return error;

	/*
	 * Commit the shortform mods, and we're done.
	 * NOTE: this is also the error path (EEXIST, etc).
	 */
226
	if (!error && !(args->op_flags & XFS_DA_OP_NOTIME))
227 228
		xfs_trans_ichgtime(args->trans, dp, XFS_ICHGTIME_CHG);

229
	if (xfs_has_wsync(dp->i_mount))
230 231
		xfs_trans_set_sync(args->trans);

232
	return error;
233 234
}

235 236 237 238 239 240 241 242 243 244 245 246 247 248
/*
 * Check to see if the attr should be upgraded from non-existent or shortform to
 * single-leaf-block attribute list.
 */
static inline bool
xfs_attr_is_shortform(
	struct xfs_inode    *ip)
{
	return ip->i_afp->if_format == XFS_DINODE_FMT_LOCAL ||
	       (ip->i_afp->if_format == XFS_DINODE_FMT_EXTENTS &&
		ip->i_afp->if_nextents == 0);
}

STATIC int
249
xfs_attr_sf_addname(
250
	struct xfs_attr_item		*attr)
251
{
252
	struct xfs_da_args		*args = attr->xattri_da_args;
253 254
	struct xfs_inode		*dp = args->dp;
	int				error = 0;
255 256 257 258 259

	/*
	 * Try to add the attr to the attribute list in the inode.
	 */
	error = xfs_attr_try_sf_addname(dp, args);
260 261 262 263

	/* Should only be 0, -EEXIST or -ENOSPC */
	if (error != -ENOSPC)
		return error;
264

265 266 267 268
	/*
	 * It won't fit in the shortform, transform to a leaf block.  GROT:
	 * another possible req'mt for a double-split btree op.
	 */
269
	error = xfs_attr_shortform_to_leaf(args, &attr->xattri_leaf_bp);
270 271 272 273 274 275
	if (error)
		return error;

	/*
	 * Prevent the leaf buffer from being unlocked so that a concurrent AIL
	 * push cannot grab the half-baked leaf buffer and run into problems
276
	 * with the write verifier.
277
	 */
278
	xfs_trans_bhold(args->trans, attr->xattri_leaf_bp);
279

280 281 282 283 284
	/*
	 * We're still in XFS_DAS_UNINIT state here.  We've converted
	 * the attr fork to leaf format and will restart with the leaf
	 * add.
	 */
285
	trace_xfs_attr_sf_addname_return(XFS_DAS_UNINIT, args->dp);
286
	return -EAGAIN;
287 288
}

289 290
/*
 * Set the attribute specified in @args.
291 292 293 294
 * This routine is meant to function as a delayed operation, and may return
 * -EAGAIN when the transaction needs to be rolled.  Calling functions will need
 * to handle this, and recall the function until a successful error code is
 * returned.
295 296
 */
int
297
xfs_attr_set_iter(
298
	struct xfs_attr_item		*attr)
299
{
300
	struct xfs_da_args              *args = attr->xattri_da_args;
301 302 303
	struct xfs_inode		*dp = args->dp;
	struct xfs_buf			*bp = NULL;
	int				forkoff, error = 0;
304
	struct xfs_mount		*mp = args->dp->i_mount;
305 306

	/* State machine switch */
307
	switch (attr->xattri_dela_state) {
308 309 310 311 312 313 314 315 316
	case XFS_DAS_UNINIT:
		/*
		 * If the fork is shortform, attempt to add the attr. If there
		 * is no space, this converts to leaf format and returns
		 * -EAGAIN with the leaf buffer held across the roll. The caller
		 * will deal with a transaction roll error, but otherwise
		 * release the hold once we return with a clean transaction.
		 */
		if (xfs_attr_is_shortform(dp))
317 318 319 320 321
			return xfs_attr_sf_addname(attr);
		if (attr->xattri_leaf_bp != NULL) {
			xfs_trans_bhold_release(args->trans,
						attr->xattri_leaf_bp);
			attr->xattri_leaf_bp = NULL;
322
		}
323

324
		if (xfs_attr_is_leaf(dp)) {
325 326
			error = xfs_attr_leaf_try_add(args,
						      attr->xattri_leaf_bp);
327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344
			if (error == -ENOSPC) {
				error = xfs_attr3_leaf_to_node(args);
				if (error)
					return error;

				/*
				 * Finish any deferred work items and roll the
				 * transaction once more.  The goal here is to
				 * call node_addname with the inode and
				 * transaction in the same state (inode locked
				 * and joined, transaction clean) no matter how
				 * we got to this step.
				 *
				 * At this point, we are still in
				 * XFS_DAS_UNINIT, but when we come back, we'll
				 * be a node, so we'll fall down into the node
				 * handling code below
				 */
345
				trace_xfs_attr_set_iter_return(
346
					attr->xattri_dela_state, args->dp);
347 348
				return -EAGAIN;
			} else if (error) {
349
				return error;
350
			}
351

352
			attr->xattri_dela_state = XFS_DAS_FOUND_LBLK;
353
		} else {
354
			error = xfs_attr_node_addname_find_attr(attr);
355 356 357
			if (error)
				return error;

358
			error = xfs_attr_node_addname(attr);
359 360 361
			if (error)
				return error;

362 363 364 365 366 367 368 369
			/*
			 * If addname was successful, and we dont need to alloc
			 * or remove anymore blks, we're done.
			 */
			if (!args->rmtblkno &&
			    !(args->op_flags & XFS_DA_OP_RENAME))
				return 0;

370
			attr->xattri_dela_state = XFS_DAS_FOUND_NBLK;
371
		}
372 373
		trace_xfs_attr_set_iter_return(attr->xattri_dela_state,
					       args->dp);
374 375
		return -EAGAIN;
	case XFS_DAS_FOUND_LBLK:
376 377 378 379 380 381
		/*
		 * If there was an out-of-line value, allocate the blocks we
		 * identified for its storage and copy the value.  This is done
		 * after we create the attribute so that we don't overflow the
		 * maximum size of a transaction and/or hit a deadlock.
		 */
382 383

		/* Open coded xfs_attr_rmtval_set without trans handling */
384 385
		if ((attr->xattri_flags & XFS_DAC_LEAF_ADDNAME_INIT) == 0) {
			attr->xattri_flags |= XFS_DAC_LEAF_ADDNAME_INIT;
386
			if (args->rmtblkno > 0) {
387
				error = xfs_attr_rmtval_find_space(attr);
388 389 390 391 392 393 394 395 396
				if (error)
					return error;
			}
		}

		/*
		 * Repeat allocating remote blocks for the attr value until
		 * blkcnt drops to zero.
		 */
397 398
		if (attr->xattri_blkcnt > 0) {
			error = xfs_attr_rmtval_set_blk(attr);
399 400
			if (error)
				return error;
401
			trace_xfs_attr_set_iter_return(attr->xattri_dela_state,
402
						       args->dp);
403
			return -EAGAIN;
404 405
		}

406 407 408 409 410 411 412 413
		error = xfs_attr_rmtval_set_value(args);
		if (error)
			return error;

		/*
		 * If this is not a rename, clear the incomplete flag and we're
		 * done.
		 */
414 415 416
		if (!(args->op_flags & XFS_DA_OP_RENAME)) {
			if (args->rmtblkno > 0)
				error = xfs_attr3_leaf_clearflag(args);
417
			return error;
418
		}
419

420 421 422 423 424 425 426 427 428
		/*
		 * If this is an atomic rename operation, we must "flip" the
		 * incomplete flags on the "new" and "old" attribute/value pairs
		 * so that one disappears and one appears atomically.  Then we
		 * must remove the "old" attribute/value pair.
		 *
		 * In a separate transaction, set the incomplete flag on the
		 * "old" attr and clear the incomplete flag on the "new" attr.
		 */
429 430 431 432 433 434 435 436
		if (!xfs_has_larp(mp)) {
			error = xfs_attr3_leaf_flipflags(args);
			if (error)
				return error;
			/*
			 * Commit the flag value change and start the next trans
			 * in series.
			 */
437 438
			attr->xattri_dela_state = XFS_DAS_FLIP_LFLAG;
			trace_xfs_attr_set_iter_return(attr->xattri_dela_state,
439 440 441 442 443
						       args->dp);
			return -EAGAIN;
		}

		fallthrough;
444
	case XFS_DAS_FLIP_LFLAG:
445 446 447 448 449
		/*
		 * Dismantle the "old" attribute/value pair by removing a
		 * "remote" value (if it exists).
		 */
		xfs_attr_restore_rmt_blk(args);
450 451 452
		error = xfs_attr_rmtval_invalidate(args);
		if (error)
			return error;
453

454
		fallthrough;
455 456
	case XFS_DAS_RM_LBLK:
		/* Set state in case xfs_attr_rmtval_remove returns -EAGAIN */
457
		attr->xattri_dela_state = XFS_DAS_RM_LBLK;
458
		if (args->rmtblkno) {
459
			error = xfs_attr_rmtval_remove(attr);
460 461
			if (error == -EAGAIN)
				trace_xfs_attr_set_iter_return(
462
					attr->xattri_dela_state, args->dp);
463 464 465
			if (error)
				return error;

466 467 468
			attr->xattri_dela_state = XFS_DAS_RD_LEAF;
			trace_xfs_attr_set_iter_return(attr->xattri_dela_state,
						       args->dp);
469
			return -EAGAIN;
470 471
		}

472
		fallthrough;
473
	case XFS_DAS_RD_LEAF:
474
		/*
475 476 477
		 * This is the last step for leaf format. Read the block with
		 * the old attr, remove the old attr, check for shortform
		 * conversion and return.
478 479 480 481 482 483 484 485 486 487 488 489 490 491
		 */
		error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno,
					   &bp);
		if (error)
			return error;

		xfs_attr3_leaf_remove(bp, args);

		forkoff = xfs_attr_shortform_allfit(bp, dp);
		if (forkoff)
			error = xfs_attr3_leaf_to_shortform(bp, args, forkoff);
			/* bp is gone due to xfs_da_shrink_inode */

		return error;
492

493 494 495 496 497 498
	case XFS_DAS_FOUND_NBLK:
		/*
		 * Find space for remote blocks and fall into the allocation
		 * state.
		 */
		if (args->rmtblkno > 0) {
499
			error = xfs_attr_rmtval_find_space(attr);
500 501 502
			if (error)
				return error;
		}
503

504
		fallthrough;
505 506 507 508 509 510 511
	case XFS_DAS_ALLOC_NODE:
		/*
		 * If there was an out-of-line value, allocate the blocks we
		 * identified for its storage and copy the value.  This is done
		 * after we create the attribute so that we don't overflow the
		 * maximum size of a transaction and/or hit a deadlock.
		 */
512
		attr->xattri_dela_state = XFS_DAS_ALLOC_NODE;
513
		if (args->rmtblkno > 0) {
514 515
			if (attr->xattri_blkcnt > 0) {
				error = xfs_attr_rmtval_set_blk(attr);
516 517
				if (error)
					return error;
518
				trace_xfs_attr_set_iter_return(
519
					attr->xattri_dela_state, args->dp);
520 521 522 523 524 525 526
				return -EAGAIN;
			}

			error = xfs_attr_rmtval_set_value(args);
			if (error)
				return error;
		}
527

528 529 530 531 532 533 534 535 536
		/*
		 * If this was not a rename, clear the incomplete flag and we're
		 * done.
		 */
		if (!(args->op_flags & XFS_DA_OP_RENAME)) {
			if (args->rmtblkno > 0)
				error = xfs_attr3_leaf_clearflag(args);
			goto out;
		}
537

538 539 540 541 542 543 544 545 546
		/*
		 * If this is an atomic rename operation, we must "flip" the
		 * incomplete flags on the "new" and "old" attribute/value pairs
		 * so that one disappears and one appears atomically.  Then we
		 * must remove the "old" attribute/value pair.
		 *
		 * In a separate transaction, set the incomplete flag on the
		 * "old" attr and clear the incomplete flag on the "new" attr.
		 */
547 548 549 550 551 552 553 554
		if (!xfs_has_larp(mp)) {
			error = xfs_attr3_leaf_flipflags(args);
			if (error)
				goto out;
			/*
			 * Commit the flag value change and start the next trans
			 * in series
			 */
555 556
			attr->xattri_dela_state = XFS_DAS_FLIP_NFLAG;
			trace_xfs_attr_set_iter_return(attr->xattri_dela_state,
557 558 559
						       args->dp);
			return -EAGAIN;
		}
560

561
		fallthrough;
562 563 564 565 566 567
	case XFS_DAS_FLIP_NFLAG:
		/*
		 * Dismantle the "old" attribute/value pair by removing a
		 * "remote" value (if it exists).
		 */
		xfs_attr_restore_rmt_blk(args);
568 569 570 571 572

		error = xfs_attr_rmtval_invalidate(args);
		if (error)
			return error;

573
		fallthrough;
574 575
	case XFS_DAS_RM_NBLK:
		/* Set state in case xfs_attr_rmtval_remove returns -EAGAIN */
576
		attr->xattri_dela_state = XFS_DAS_RM_NBLK;
577
		if (args->rmtblkno) {
578
			error = xfs_attr_rmtval_remove(attr);
579 580
			if (error == -EAGAIN)
				trace_xfs_attr_set_iter_return(
581
					attr->xattri_dela_state, args->dp);
582

583 584 585
			if (error)
				return error;

586 587 588
			attr->xattri_dela_state = XFS_DAS_CLR_FLAG;
			trace_xfs_attr_set_iter_return(attr->xattri_dela_state,
						       args->dp);
589 590
			return -EAGAIN;
		}
591

592
		fallthrough;
593 594 595 596 597
	case XFS_DAS_CLR_FLAG:
		/*
		 * The last state for node format. Look up the old attr and
		 * remove it.
		 */
598
		error = xfs_attr_node_addname_clear_incomplete(attr);
599 600
		break;
	default:
601
		ASSERT(0);
602 603
		break;
	}
604
out:
605
	return error;
606 607
}

608

609 610 611
/*
 * Return EEXIST if attr is found, or ENOATTR if not
 */
D
Dave Chinner 已提交
612 613
static int
xfs_attr_lookup(
614 615 616 617 618 619 620 621 622
	struct xfs_da_args	*args)
{
	struct xfs_inode	*dp = args->dp;
	struct xfs_buf		*bp = NULL;
	int			error;

	if (!xfs_inode_hasattr(dp))
		return -ENOATTR;

C
Christoph Hellwig 已提交
623
	if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL)
624 625
		return xfs_attr_sf_findname(args, NULL, NULL);

626
	if (xfs_attr_is_leaf(dp)) {
627 628 629 630 631 632 633 634 635 636 637
		error = xfs_attr_leaf_hasname(args, &bp);

		if (bp)
			xfs_trans_brelse(args->trans, bp);

		return error;
	}

	return xfs_attr_node_hasname(args, NULL);
}

638
/*
639
 * Note: If args->value is NULL the attribute will be removed, just like the
640 641
 * Linux ->setattr API.
 */
642 643
int
xfs_attr_set(
644
	struct xfs_da_args	*args)
L
Linus Torvalds 已提交
645
{
646
	struct xfs_inode	*dp = args->dp;
647 648
	struct xfs_mount	*mp = dp->i_mount;
	struct xfs_trans_res	tres;
649
	bool			rsvd = (args->attr_filter & XFS_ATTR_ROOT);
650
	int			error, local;
651
	int			rmt_blks = 0;
652
	unsigned int		total;
653
	int			delayed = xfs_has_larp(mp);
654

655
	if (xfs_is_shutdown(dp->i_mount))
D
Dave Chinner 已提交
656
		return -EIO;
657

658 659 660 661
	error = xfs_qm_dqattach(dp);
	if (error)
		return error;

662 663 664
	args->geo = mp->m_attr_geo;
	args->whichfork = XFS_ATTR_FORK;
	args->hashval = xfs_da_hashname(args->name, args->namelen);
L
Linus Torvalds 已提交
665 666

	/*
667 668 669
	 * We have no control over the attribute names that userspace passes us
	 * to remove, so we have to allow the name lookup prior to attribute
	 * removal to fail as well.
L
Linus Torvalds 已提交
670
	 */
671
	args->op_flags = XFS_DA_OP_OKNOENT;
672

673
	if (args->value) {
674 675
		XFS_STATS_INC(mp, xs_attr_set);

676 677
		args->op_flags |= XFS_DA_OP_ADDNAME;
		args->total = xfs_attr_calc_size(args, &local);
678 679 680 681 682 683 684

		/*
		 * If the inode doesn't have an attribute fork, add one.
		 * (inode must not be locked when we call this routine)
		 */
		if (XFS_IFORK_Q(dp) == 0) {
			int sf_size = sizeof(struct xfs_attr_sf_hdr) +
685
				xfs_attr_sf_entsize_byname(args->namelen,
686
						args->valuelen);
687 688 689 690 691 692 693

			error = xfs_bmap_add_attrfork(dp, sf_size, rsvd);
			if (error)
				return error;
		}

		tres.tr_logres = M_RES(mp)->tr_attrsetm.tr_logres +
694 695
				 M_RES(mp)->tr_attrsetrt.tr_logres *
					args->total;
696 697
		tres.tr_logcount = XFS_ATTRSET_LOG_COUNT;
		tres.tr_logflags = XFS_TRANS_PERM_LOG_RES;
698
		total = args->total;
699 700 701

		if (!local)
			rmt_blks = xfs_attr3_rmt_blocks(mp, args->valuelen);
702 703
	} else {
		XFS_STATS_INC(mp, xs_attr_remove);
L
Linus Torvalds 已提交
704

705 706
		tres = M_RES(mp)->tr_attrrm;
		total = XFS_ATTRRM_SPACE_RES(mp);
707
		rmt_blks = xfs_attr3_rmt_blocks(mp, XFS_XATTR_SIZE_MAX);
708
	}
L
Linus Torvalds 已提交
709

710 711 712 713 714 715
	if (delayed) {
		error = xfs_attr_use_log_assist(mp);
		if (error)
			return error;
	}

L
Linus Torvalds 已提交
716 717 718 719
	/*
	 * Root fork attributes can use reserved data blocks for this
	 * operation if necessary
	 */
720
	error = xfs_trans_alloc_inode(dp, &tres, total, 0, rsvd, &args->trans);
721
	if (error)
722
		goto drop_incompat;
L
Linus Torvalds 已提交
723

724 725 726
	if (args->value || xfs_inode_hasattr(dp)) {
		error = xfs_iext_count_may_overflow(dp, XFS_ATTR_FORK,
				XFS_IEXT_ATTR_MANIP_CNT(rmt_blks));
727 728 729
		if (error == -EFBIG)
			error = xfs_iext_count_upgrade(args->trans, dp,
					XFS_IEXT_ATTR_MANIP_CNT(rmt_blks));
730 731 732 733
		if (error)
			goto out_trans_cancel;
	}

D
Dave Chinner 已提交
734
	error = xfs_attr_lookup(args);
735
	if (args->value) {
736 737 738 739 740 741 742
		if (error == -EEXIST && (args->attr_flags & XATTR_CREATE))
			goto out_trans_cancel;
		if (error == -ENOATTR && (args->attr_flags & XATTR_REPLACE))
			goto out_trans_cancel;
		if (error != -ENOATTR && error != -EEXIST)
			goto out_trans_cancel;

743
		error = xfs_attr_set_deferred(args);
744 745
		if (error)
			goto out_trans_cancel;
746

747
		/* shortform attribute has already been committed */
748
		if (!args->trans)
749 750
			goto out_unlock;
	} else {
751 752 753
		if (error != -EEXIST)
			goto out_trans_cancel;

754
		error = xfs_attr_remove_deferred(args);
755 756
		if (error)
			goto out_trans_cancel;
757
	}
L
Linus Torvalds 已提交
758 759 760 761 762

	/*
	 * If this is a synchronous mount, make sure that the
	 * transaction goes to disk before returning to the user.
	 */
763
	if (xfs_has_wsync(mp))
764
		xfs_trans_set_sync(args->trans);
L
Linus Torvalds 已提交
765

766
	if (!(args->op_flags & XFS_DA_OP_NOTIME))
767
		xfs_trans_ichgtime(args->trans, dp, XFS_ICHGTIME_CHG);
768

L
Linus Torvalds 已提交
769 770 771
	/*
	 * Commit the last in the sequence of transactions.
	 */
772 773
	xfs_trans_log_inode(args->trans, dp, XFS_ILOG_CORE);
	error = xfs_trans_commit(args->trans);
774
out_unlock:
L
Linus Torvalds 已提交
775
	xfs_iunlock(dp, XFS_ILOCK_EXCL);
776 777 778
drop_incompat:
	if (delayed)
		xlog_drop_incompat_feat(mp->m_log);
779
	return error;
L
Linus Torvalds 已提交
780

781
out_trans_cancel:
782 783
	if (args->trans)
		xfs_trans_cancel(args->trans);
784
	goto out_unlock;
L
Linus Torvalds 已提交
785 786
}

787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820
int __init
xfs_attri_init_cache(void)
{
	xfs_attri_cache = kmem_cache_create("xfs_attri",
					    sizeof(struct xfs_attri_log_item),
					    0, 0, NULL);

	return xfs_attri_cache != NULL ? 0 : -ENOMEM;
}

void
xfs_attri_destroy_cache(void)
{
	kmem_cache_destroy(xfs_attri_cache);
	xfs_attri_cache = NULL;
}

int __init
xfs_attrd_init_cache(void)
{
	xfs_attrd_cache = kmem_cache_create("xfs_attrd",
					    sizeof(struct xfs_attrd_log_item),
					    0, 0, NULL);

	return xfs_attrd_cache != NULL ? 0 : -ENOMEM;
}

void
xfs_attrd_destroy_cache(void)
{
	kmem_cache_destroy(xfs_attrd_cache);
	xfs_attrd_cache = NULL;
}

821 822 823 824 825 826 827 828 829 830 831
STATIC int
xfs_attr_item_init(
	struct xfs_da_args	*args,
	unsigned int		op_flags,	/* op flag (set or remove) */
	struct xfs_attr_item	**attr)		/* new xfs_attr_item */
{

	struct xfs_attr_item	*new;

	new = kmem_zalloc(sizeof(struct xfs_attr_item), KM_NOFS);
	new->xattri_op_flags = op_flags;
832
	new->xattri_da_args = args;
833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872

	*attr = new;
	return 0;
}

/* Sets an attribute for an inode as a deferred operation */
int
xfs_attr_set_deferred(
	struct xfs_da_args	*args)
{
	struct xfs_attr_item	*new;
	int			error = 0;

	error = xfs_attr_item_init(args, XFS_ATTR_OP_FLAGS_SET, &new);
	if (error)
		return error;

	xfs_defer_add(args->trans, XFS_DEFER_OPS_TYPE_ATTR, &new->xattri_list);

	return 0;
}

/* Removes an attribute for an inode as a deferred operation */
int
xfs_attr_remove_deferred(
	struct xfs_da_args	*args)
{

	struct xfs_attr_item	*new;
	int			error;

	error  = xfs_attr_item_init(args, XFS_ATTR_OP_FLAGS_REMOVE, &new);
	if (error)
		return error;

	xfs_defer_add(args->trans, XFS_DEFER_OPS_TYPE_ATTR, &new->xattri_list);

	return 0;
}

L
Linus Torvalds 已提交
873 874 875 876
/*========================================================================
 * External routines when attribute list is inside the inode
 *========================================================================*/

877 878 879 880 881 882 883 884
static inline int xfs_attr_sf_totsize(struct xfs_inode *dp)
{
	struct xfs_attr_shortform *sf;

	sf = (struct xfs_attr_shortform *)dp->i_afp->if_u1.if_data;
	return be16_to_cpu(sf->hdr.totsize);
}

L
Linus Torvalds 已提交
885 886 887 888 889 890 891
/*
 * Add a name to the shortform attribute list structure
 * This is the external routine.
 */
STATIC int
xfs_attr_shortform_addname(xfs_da_args_t *args)
{
892
	int newsize, forkoff, retval;
L
Linus Torvalds 已提交
893

894 895
	trace_xfs_attr_sf_addname(args);

L
Linus Torvalds 已提交
896
	retval = xfs_attr_shortform_lookup(args);
897
	if (retval == -ENOATTR && (args->attr_flags & XATTR_REPLACE))
E
Eric Sandeen 已提交
898
		return retval;
899
	if (retval == -EEXIST) {
900
		if (args->attr_flags & XATTR_CREATE)
E
Eric Sandeen 已提交
901
			return retval;
902
		retval = xfs_attr_sf_removename(args);
903 904 905 906 907 908 909
		if (retval)
			return retval;
		/*
		 * Since we have removed the old attr, clear ATTR_REPLACE so
		 * that the leaf format add routine won't trip over the attr
		 * not being around.
		 */
910
		args->attr_flags &= ~XATTR_REPLACE;
L
Linus Torvalds 已提交
911 912
	}

913 914
	if (args->namelen >= XFS_ATTR_SF_ENTSIZE_MAX ||
	    args->valuelen >= XFS_ATTR_SF_ENTSIZE_MAX)
D
Dave Chinner 已提交
915
		return -ENOSPC;
916

917 918
	newsize = xfs_attr_sf_totsize(args->dp);
	newsize += xfs_attr_sf_entsize_byname(args->namelen, args->valuelen);
919 920 921

	forkoff = xfs_attr_shortform_bytesfit(args->dp, newsize);
	if (!forkoff)
D
Dave Chinner 已提交
922
		return -ENOSPC;
923 924

	xfs_attr_shortform_add(args, forkoff);
E
Eric Sandeen 已提交
925
	return 0;
L
Linus Torvalds 已提交
926 927 928 929 930 931 932
}


/*========================================================================
 * External routines when attribute list is one block
 *========================================================================*/

933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956
/* Store info about a remote block */
STATIC void
xfs_attr_save_rmt_blk(
	struct xfs_da_args	*args)
{
	args->blkno2 = args->blkno;
	args->index2 = args->index;
	args->rmtblkno2 = args->rmtblkno;
	args->rmtblkcnt2 = args->rmtblkcnt;
	args->rmtvaluelen2 = args->rmtvaluelen;
}

/* Set stored info about a remote block */
STATIC void
xfs_attr_restore_rmt_blk(
	struct xfs_da_args	*args)
{
	args->blkno = args->blkno2;
	args->index = args->index2;
	args->rmtblkno = args->rmtblkno2;
	args->rmtblkcnt = args->rmtblkcnt2;
	args->rmtvaluelen = args->rmtvaluelen2;
}

L
Linus Torvalds 已提交
957
/*
958
 * Tries to add an attribute to an inode in leaf form
L
Linus Torvalds 已提交
959
 *
960 961 962 963 964 965
 * This function is meant to execute as part of a delayed operation and leaves
 * the transaction handling to the caller.  On success the attribute is added
 * and the inode and transaction are left dirty.  If there is not enough space,
 * the attr data is converted to node format and -ENOSPC is returned. Caller is
 * responsible for handling the dirty inode and transaction or adding the attr
 * in node format.
L
Linus Torvalds 已提交
966
 */
D
David Chinner 已提交
967
STATIC int
968 969 970
xfs_attr_leaf_try_add(
	struct xfs_da_args	*args,
	struct xfs_buf		*bp)
L
Linus Torvalds 已提交
971
{
972
	int			retval;
973

L
Linus Torvalds 已提交
974 975 976 977
	/*
	 * Look up the given attribute in the leaf block.  Figure out if
	 * the given flags produce an error or call for an atomic rename.
	 */
978 979 980
	retval = xfs_attr_leaf_hasname(args, &bp);
	if (retval != -ENOATTR && retval != -EEXIST)
		return retval;
981
	if (retval == -ENOATTR && (args->attr_flags & XATTR_REPLACE))
982 983
		goto out_brelse;
	if (retval == -EEXIST) {
984
		if (args->attr_flags & XATTR_CREATE)
985
			goto out_brelse;
986 987 988

		trace_xfs_attr_leaf_replace(args);

989
		/* save the attribute state for later removal*/
990
		args->op_flags |= XFS_DA_OP_RENAME;	/* an atomic rename */
991
		xfs_attr_save_rmt_blk(args);
992 993 994 995 996 997 998 999 1000

		/*
		 * clear the remote attr state now that it is saved so that the
		 * values reflect the state of the attribute we are about to
		 * add, not the attribute we just found and will remove later.
		 */
		args->rmtblkno = 0;
		args->rmtblkcnt = 0;
		args->rmtvaluelen = 0;
L
Linus Torvalds 已提交
1001 1002 1003
	}

	/*
1004
	 * Add the attribute to the leaf block
L
Linus Torvalds 已提交
1005
	 */
1006 1007
	return xfs_attr3_leaf_add(bp, args);

1008 1009 1010 1011
out_brelse:
	xfs_trans_brelse(args->trans, bp);
	return retval;
}
L
Linus Torvalds 已提交
1012

1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033
/*
 * Return EEXIST if attr is found, or ENOATTR if not
 */
STATIC int
xfs_attr_leaf_hasname(
	struct xfs_da_args	*args,
	struct xfs_buf		**bp)
{
	int                     error = 0;

	error = xfs_attr3_leaf_read(args->trans, args->dp, 0, bp);
	if (error)
		return error;

	error = xfs_attr3_leaf_lookup_int(*bp, args);
	if (error != -ENOATTR && error != -EEXIST)
		xfs_trans_brelse(args->trans, *bp);

	return error;
}

L
Linus Torvalds 已提交
1034 1035 1036 1037 1038 1039 1040
/*
 * Remove a name from the leaf attribute list structure
 *
 * This leaf block cannot have a "remote" value, we only call this routine
 * if bmap_one_block() says there is only one block (ie: no remote blks).
 */
STATIC int
1041 1042
xfs_attr_leaf_removename(
	struct xfs_da_args	*args)
L
Linus Torvalds 已提交
1043
{
1044 1045 1046
	struct xfs_inode	*dp;
	struct xfs_buf		*bp;
	int			error, forkoff;
L
Linus Torvalds 已提交
1047

1048 1049
	trace_xfs_attr_leaf_removename(args);

L
Linus Torvalds 已提交
1050 1051 1052 1053 1054
	/*
	 * Remove the attribute.
	 */
	dp = args->dp;

1055 1056
	error = xfs_attr_leaf_hasname(args, &bp);

D
Dave Chinner 已提交
1057
	if (error == -ENOATTR) {
1058
		xfs_trans_brelse(args->trans, bp);
D
Dave Chinner 已提交
1059
		return error;
1060 1061
	} else if (error != -EEXIST)
		return error;
L
Linus Torvalds 已提交
1062

D
Dave Chinner 已提交
1063
	xfs_attr3_leaf_remove(bp, args);
L
Linus Torvalds 已提交
1064 1065 1066 1067

	/*
	 * If the result is small enough, shrink it all into the inode.
	 */
1068 1069 1070
	forkoff = xfs_attr_shortform_allfit(bp, dp);
	if (forkoff)
		return xfs_attr3_leaf_to_shortform(bp, args, forkoff);
L
Linus Torvalds 已提交
1071
		/* bp is gone due to xfs_da_shrink_inode */
1072

D
Dave Chinner 已提交
1073
	return 0;
L
Linus Torvalds 已提交
1074 1075 1076 1077 1078 1079 1080
}

/*
 * Look up a name in a leaf attribute list structure.
 *
 * This leaf block cannot have a "remote" value, we only call this routine
 * if bmap_one_block() says there is only one block (ie: no remote blks).
1081 1082
 *
 * Returns 0 on successful retrieval, otherwise an error.
L
Linus Torvalds 已提交
1083
 */
1084
STATIC int
L
Linus Torvalds 已提交
1085 1086
xfs_attr_leaf_get(xfs_da_args_t *args)
{
1087
	struct xfs_buf *bp;
L
Linus Torvalds 已提交
1088 1089
	int error;

1090 1091
	trace_xfs_attr_leaf_get(args);

1092
	error = xfs_attr_leaf_hasname(args, &bp);
L
Linus Torvalds 已提交
1093

1094
	if (error == -ENOATTR)  {
1095
		xfs_trans_brelse(args->trans, bp);
D
Dave Chinner 已提交
1096
		return error;
1097 1098 1099 1100
	} else if (error != -EEXIST)
		return error;


D
Dave Chinner 已提交
1101
	error = xfs_attr3_leaf_getvalue(bp, args);
1102
	xfs_trans_brelse(args->trans, bp);
1103
	return error;
L
Linus Torvalds 已提交
1104 1105
}

1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120
/*
 * Return EEXIST if attr is found, or ENOATTR if not
 * statep: If not null is set to point at the found state.  Caller will
 *         be responsible for freeing the state in this case.
 */
STATIC int
xfs_attr_node_hasname(
	struct xfs_da_args	*args,
	struct xfs_da_state	**statep)
{
	struct xfs_da_state	*state;
	int			retval, error;

	state = xfs_da_state_alloc(args);
	if (statep != NULL)
1121
		*statep = state;
1122 1123 1124 1125 1126

	/*
	 * Search to see if name exists, and get back a pointer to it.
	 */
	error = xfs_da3_node_lookup_int(state, &retval);
1127 1128
	if (error)
		retval = error;
1129

1130
	if (!statep)
1131
		xfs_da_state_free(state);
1132

1133 1134 1135
	return retval;
}

L
Linus Torvalds 已提交
1136
/*========================================================================
1137
 * External routines when attribute list size > geo->blksize
L
Linus Torvalds 已提交
1138 1139 1140
 *========================================================================*/

STATIC int
1141
xfs_attr_node_addname_find_attr(
1142
	 struct xfs_attr_item		*attr)
L
Linus Torvalds 已提交
1143
{
1144
	struct xfs_da_args		*args = attr->xattri_da_args;
1145
	int				retval;
1146

L
Linus Torvalds 已提交
1147 1148 1149 1150
	/*
	 * Search to see if name already exists, and get back a pointer
	 * to where it should go.
	 */
1151
	retval = xfs_attr_node_hasname(args, &attr->xattri_da_state);
1152
	if (retval != -ENOATTR && retval != -EEXIST)
1153
		goto error;
1154

1155
	if (retval == -ENOATTR && (args->attr_flags & XATTR_REPLACE))
1156
		goto error;
1157
	if (retval == -EEXIST) {
1158
		if (args->attr_flags & XATTR_CREATE)
1159
			goto error;
1160 1161 1162

		trace_xfs_attr_node_replace(args);

1163
		/* save the attribute state for later removal*/
1164
		args->op_flags |= XFS_DA_OP_RENAME;	/* atomic rename op */
1165
		xfs_attr_save_rmt_blk(args);
1166 1167 1168 1169 1170 1171

		/*
		 * clear the remote attr state now that it is saved so that the
		 * values reflect the state of the attribute we are about to
		 * add, not the attribute we just found and will remove later.
		 */
L
Linus Torvalds 已提交
1172 1173
		args->rmtblkno = 0;
		args->rmtblkcnt = 0;
1174
		args->rmtvaluelen = 0;
L
Linus Torvalds 已提交
1175 1176
	}

1177 1178
	return 0;
error:
1179 1180
	if (attr->xattri_da_state)
		xfs_da_state_free(attr->xattri_da_state);
1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192
	return retval;
}

/*
 * Add a name to a Btree-format attribute list.
 *
 * This will involve walking down the Btree, and may involve splitting
 * leaf nodes and even splitting intermediate nodes up to and including
 * the root node (a special case of an intermediate node).
 *
 * "Remote" attribute values confuse the issue and atomic rename operations
 * add a whole extra layer of confusion on top of that.
1193 1194 1195 1196 1197
 *
 * This routine is meant to function as a delayed operation, and may return
 * -EAGAIN when the transaction needs to be rolled.  Calling functions will need
 * to handle this, and recall the function until a successful error code is
 *returned.
1198 1199 1200
 */
STATIC int
xfs_attr_node_addname(
1201
	struct xfs_attr_item		*attr)
1202
{
1203 1204
	struct xfs_da_args		*args = attr->xattri_da_args;
	struct xfs_da_state		*state = attr->xattri_da_state;
1205 1206
	struct xfs_da_state_blk		*blk;
	int				error;
1207 1208 1209 1210 1211 1212

	trace_xfs_attr_node_addname(args);

	blk = &state->path.blk[state->path.active-1];
	ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);

1213 1214
	error = xfs_attr3_leaf_add(blk->bp, state->args);
	if (error == -ENOSPC) {
L
Linus Torvalds 已提交
1215 1216 1217 1218 1219 1220 1221
		if (state->path.active == 1) {
			/*
			 * Its really a single leaf node, but it had
			 * out-of-line values so it looked like it *might*
			 * have been a b-tree.
			 */
			xfs_da_state_free(state);
1222
			state = NULL;
D
Dave Chinner 已提交
1223
			error = xfs_attr3_leaf_to_node(args);
1224
			if (error)
1225
				goto out;
L
Linus Torvalds 已提交
1226 1227

			/*
1228 1229 1230 1231 1232
			 * Now that we have converted the leaf to a node, we can
			 * roll the transaction, and try xfs_attr3_leaf_add
			 * again on re-entry.  No need to set dela_state to do
			 * this. dela_state is still unset by this function at
			 * this point.
L
Linus Torvalds 已提交
1233
			 */
1234
			trace_xfs_attr_node_addname_return(
1235
					attr->xattri_dela_state, args->dp);
1236
			return -EAGAIN;
L
Linus Torvalds 已提交
1237 1238 1239 1240 1241 1242 1243 1244
		}

		/*
		 * Split as many Btree elements as required.
		 * This code tracks the new and old attr's location
		 * in the index/blkno/rmtblkno/rmtblkcnt fields and
		 * in the index2/blkno2/rmtblkno2/rmtblkcnt2 fields.
		 */
1245
		error = xfs_da3_split(state);
1246
		if (error)
1247
			goto out;
L
Linus Torvalds 已提交
1248 1249 1250 1251
	} else {
		/*
		 * Addition succeeded, update Btree hashvals.
		 */
1252
		xfs_da3_fixhashpath(state, &state->path);
L
Linus Torvalds 已提交
1253 1254
	}

1255 1256 1257
out:
	if (state)
		xfs_da_state_free(state);
1258
	return error;
1259 1260 1261 1262 1263
}


STATIC int
xfs_attr_node_addname_clear_incomplete(
1264
	struct xfs_attr_item		*attr)
1265
{
1266
	struct xfs_da_args		*args = attr->xattri_da_args;
1267
	struct xfs_da_state		*state = NULL;
1268
	struct xfs_mount		*mp = args->dp->i_mount;
1269 1270 1271
	int				retval = 0;
	int				error = 0;

1272 1273 1274 1275
	/*
	 * Re-find the "old" attribute entry after any split ops. The INCOMPLETE
	 * flag means that we will find the "old" attr, not the "new" one.
	 */
1276 1277
	if (!xfs_has_larp(mp))
		args->attr_filter |= XFS_ATTR_INCOMPLETE;
1278 1279 1280 1281 1282
	state = xfs_da_state_alloc(args);
	state->inleaf = 0;
	error = xfs_da3_node_lookup_int(state, &retval);
	if (error)
		goto out;
L
Linus Torvalds 已提交
1283

1284
	error = xfs_attr_node_removename(args, state);
L
Linus Torvalds 已提交
1285

1286 1287 1288 1289 1290
	/*
	 * Check to see if the tree needs to be collapsed.
	 */
	if (retval && (state->path.active > 1)) {
		error = xfs_da3_join(state);
L
Linus Torvalds 已提交
1291 1292 1293 1294 1295 1296 1297 1298 1299
		if (error)
			goto out;
	}
	retval = error = 0;

out:
	if (state)
		xfs_da_state_free(state);
	if (error)
E
Eric Sandeen 已提交
1300 1301
		return error;
	return retval;
L
Linus Torvalds 已提交
1302 1303
}

1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333
/*
 * Shrink an attribute from leaf to shortform
 */
STATIC int
xfs_attr_node_shrink(
	struct xfs_da_args	*args,
	struct xfs_da_state     *state)
{
	struct xfs_inode	*dp = args->dp;
	int			error, forkoff;
	struct xfs_buf		*bp;

	/*
	 * Have to get rid of the copy of this dabuf in the state.
	 */
	ASSERT(state->path.active == 1);
	ASSERT(state->path.blk[0].bp);
	state->path.blk[0].bp = NULL;

	error = xfs_attr3_leaf_read(args->trans, args->dp, 0, &bp);
	if (error)
		return error;

	forkoff = xfs_attr_shortform_allfit(bp, dp);
	if (forkoff) {
		error = xfs_attr3_leaf_to_shortform(bp, args, forkoff);
		/* bp is gone due to xfs_da_shrink_inode */
	} else
		xfs_trans_brelse(args->trans, bp);

1334
	return error;
1335 1336
}

1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362
/*
 * Mark an attribute entry INCOMPLETE and save pointers to the relevant buffers
 * for later deletion of the entry.
 */
STATIC int
xfs_attr_leaf_mark_incomplete(
	struct xfs_da_args	*args,
	struct xfs_da_state	*state)
{
	int			error;

	/*
	 * Fill in disk block numbers in the state structure
	 * so that we can get the buffers back after we commit
	 * several transactions in the following calls.
	 */
	error = xfs_attr_fillstate(state);
	if (error)
		return error;

	/*
	 * Mark the attribute as INCOMPLETE
	 */
	return xfs_attr3_leaf_setflag(args);
}

1363 1364 1365 1366 1367 1368 1369
/*
 * Initial setup for xfs_attr_node_removename.  Make sure the attr is there and
 * the blocks are valid.  Attr keys with remote blocks will be marked
 * incomplete.
 */
STATIC
int xfs_attr_node_removename_setup(
1370
	struct xfs_attr_item		*attr)
1371
{
1372 1373
	struct xfs_da_args		*args = attr->xattri_da_args;
	struct xfs_da_state		**state = &attr->xattri_da_state;
1374
	int				error;
1375 1376 1377

	error = xfs_attr_node_hasname(args, state);
	if (error != -EEXIST)
1378
		goto out;
1379
	error = 0;
1380 1381 1382 1383 1384 1385 1386 1387

	ASSERT((*state)->path.blk[(*state)->path.active - 1].bp != NULL);
	ASSERT((*state)->path.blk[(*state)->path.active - 1].magic ==
		XFS_ATTR_LEAF_MAGIC);

	if (args->rmtblkno > 0) {
		error = xfs_attr_leaf_mark_incomplete(args, *state);
		if (error)
1388
			goto out;
1389

1390
		error = xfs_attr_rmtval_invalidate(args);
1391
	}
1392 1393 1394
out:
	if (error)
		xfs_da_state_free(*state);
1395

1396
	return error;
1397 1398
}

1399
STATIC int
1400
xfs_attr_node_removename(
1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417
	struct xfs_da_args	*args,
	struct xfs_da_state	*state)
{
	struct xfs_da_state_blk	*blk;
	int			retval;

	/*
	 * Remove the name and update the hashvals in the tree.
	 */
	blk = &state->path.blk[state->path.active-1];
	ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
	retval = xfs_attr3_leaf_remove(blk->bp, args);
	xfs_da3_fixhashpath(state, &state->path);

	return retval;
}

L
Linus Torvalds 已提交
1418
/*
1419
 * Remove the attribute specified in @args.
L
Linus Torvalds 已提交
1420 1421 1422 1423
 *
 * This will involve walking down the Btree, and may involve joining
 * leaf nodes and even joining intermediate nodes up to and including
 * the root node (a special case of an intermediate node).
1424 1425 1426 1427 1428
 *
 * This routine is meant to function as either an in-line or delayed operation,
 * and may return -EAGAIN when the transaction needs to be rolled.  Calling
 * functions will need to handle this, and call the function until a
 * successful error code is returned.
L
Linus Torvalds 已提交
1429
 */
1430 1431
int
xfs_attr_remove_iter(
1432
	struct xfs_attr_item		*attr)
L
Linus Torvalds 已提交
1433
{
1434 1435
	struct xfs_da_args		*args = attr->xattri_da_args;
	struct xfs_da_state		*state = attr->xattri_da_state;
1436
	int				retval, error = 0;
1437
	struct xfs_inode		*dp = args->dp;
L
Linus Torvalds 已提交
1438

1439 1440
	trace_xfs_attr_node_removename(args);

1441
	switch (attr->xattri_dela_state) {
1442 1443 1444
	case XFS_DAS_UNINIT:
		if (!xfs_inode_hasattr(dp))
			return -ENOATTR;
L
Linus Torvalds 已提交
1445

1446 1447 1448 1449 1450
		/*
		 * Shortform or leaf formats don't require transaction rolls and
		 * thus state transitions. Call the right helper and return.
		 */
		if (dp->i_afp->if_format == XFS_DINODE_FMT_LOCAL)
1451
			return xfs_attr_sf_removename(args);
1452 1453 1454

		if (xfs_attr_is_leaf(dp))
			return xfs_attr_leaf_removename(args);
1455 1456

		/*
1457 1458
		 * Node format may require transaction rolls. Set up the
		 * state context and fall into the state machine.
1459
		 */
1460 1461
		if (!attr->xattri_da_state) {
			error = xfs_attr_node_removename_setup(attr);
1462 1463
			if (error)
				return error;
1464
			state = attr->xattri_da_state;
1465 1466
		}

1467
		fallthrough;
1468
	case XFS_DAS_RMTBLK:
1469
		attr->xattri_dela_state = XFS_DAS_RMTBLK;
L
Linus Torvalds 已提交
1470 1471

		/*
1472 1473 1474 1475
		 * If there is an out-of-line value, de-allocate the blocks.
		 * This is done before we remove the attribute so that we don't
		 * overflow the maximum size of a transaction and/or hit a
		 * deadlock.
L
Linus Torvalds 已提交
1476
		 */
1477 1478 1479 1480 1481
		if (args->rmtblkno > 0) {
			/*
			 * May return -EAGAIN. Roll and repeat until all remote
			 * blocks are removed.
			 */
1482
			error = xfs_attr_rmtval_remove(attr);
1483 1484
			if (error == -EAGAIN) {
				trace_xfs_attr_remove_iter_return(
1485
					attr->xattri_dela_state, args->dp);
1486
				return error;
1487
			} else if (error) {
1488
				goto out;
1489
			}
L
Linus Torvalds 已提交
1490

1491 1492 1493 1494 1495 1496 1497 1498 1499
			/*
			 * Refill the state structure with buffers (the prior
			 * calls released our buffers) and close out this
			 * transaction before proceeding.
			 */
			ASSERT(args->rmtblkno == 0);
			error = xfs_attr_refillstate(state);
			if (error)
				goto out;
1500 1501 1502 1503

			attr->xattri_dela_state = XFS_DAS_RM_NAME;
			trace_xfs_attr_remove_iter_return(
					attr->xattri_dela_state, args->dp);
1504 1505 1506
			return -EAGAIN;
		}

1507
		fallthrough;
1508 1509 1510 1511 1512
	case XFS_DAS_RM_NAME:
		/*
		 * If we came here fresh from a transaction roll, reattach all
		 * the buffers to the current transaction.
		 */
1513
		if (attr->xattri_dela_state == XFS_DAS_RM_NAME) {
1514 1515 1516 1517
			error = xfs_attr_refillstate(state);
			if (error)
				goto out;
		}
L
Linus Torvalds 已提交
1518

1519
		retval = xfs_attr_node_removename(args, state);
1520 1521 1522 1523 1524 1525 1526 1527 1528 1529

		/*
		 * Check to see if the tree needs to be collapsed. If so, roll
		 * the transacton and fall into the shrink state.
		 */
		if (retval && (state->path.active > 1)) {
			error = xfs_da3_join(state);
			if (error)
				goto out;

1530
			attr->xattri_dela_state = XFS_DAS_RM_SHRINK;
1531
			trace_xfs_attr_remove_iter_return(
1532
					attr->xattri_dela_state, args->dp);
1533 1534 1535
			return -EAGAIN;
		}

1536
		fallthrough;
1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551
	case XFS_DAS_RM_SHRINK:
		/*
		 * If the result is small enough, push it all into the inode.
		 * This is our final state so it's safe to return a dirty
		 * transaction.
		 */
		if (xfs_attr_is_leaf(dp))
			error = xfs_attr_node_shrink(args, state);
		ASSERT(error != -EAGAIN);
		break;
	default:
		ASSERT(0);
		error = -EINVAL;
		goto out;
	}
L
Linus Torvalds 已提交
1552
out:
1553 1554
	if (state)
		xfs_da_state_free(state);
E
Eric Sandeen 已提交
1555
	return error;
L
Linus Torvalds 已提交
1556 1557 1558 1559 1560 1561
}

/*
 * Fill in the disk block numbers in the state structure for the buffers
 * that are attached to the state structure.
 * This is done so that we can quickly reattach ourselves to those buffers
1562
 * after some set of transaction commits have released these buffers.
L
Linus Torvalds 已提交
1563 1564 1565 1566 1567 1568 1569 1570
 */
STATIC int
xfs_attr_fillstate(xfs_da_state_t *state)
{
	xfs_da_state_path_t *path;
	xfs_da_state_blk_t *blk;
	int level;

1571 1572
	trace_xfs_attr_fillstate(state->args);

L
Linus Torvalds 已提交
1573 1574 1575 1576 1577 1578 1579 1580
	/*
	 * Roll down the "path" in the state structure, storing the on-disk
	 * block number for those buffers in the "path".
	 */
	path = &state->path;
	ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
	for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
		if (blk->bp) {
D
Dave Chinner 已提交
1581
			blk->disk_blkno = xfs_buf_daddr(blk->bp);
L
Linus Torvalds 已提交
1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595
			blk->bp = NULL;
		} else {
			blk->disk_blkno = 0;
		}
	}

	/*
	 * Roll down the "altpath" in the state structure, storing the on-disk
	 * block number for those buffers in the "altpath".
	 */
	path = &state->altpath;
	ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
	for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
		if (blk->bp) {
D
Dave Chinner 已提交
1596
			blk->disk_blkno = xfs_buf_daddr(blk->bp);
L
Linus Torvalds 已提交
1597 1598 1599 1600 1601 1602
			blk->bp = NULL;
		} else {
			blk->disk_blkno = 0;
		}
	}

E
Eric Sandeen 已提交
1603
	return 0;
L
Linus Torvalds 已提交
1604 1605 1606 1607 1608
}

/*
 * Reattach the buffers to the state structure based on the disk block
 * numbers stored in the state structure.
1609
 * This is done after some set of transaction commits have released those
L
Linus Torvalds 已提交
1610 1611 1612 1613 1614 1615 1616 1617 1618
 * buffers from our grip.
 */
STATIC int
xfs_attr_refillstate(xfs_da_state_t *state)
{
	xfs_da_state_path_t *path;
	xfs_da_state_blk_t *blk;
	int level, error;

1619 1620
	trace_xfs_attr_refillstate(state->args);

L
Linus Torvalds 已提交
1621 1622 1623 1624 1625 1626 1627 1628
	/*
	 * Roll down the "path" in the state structure, storing the on-disk
	 * block number for those buffers in the "path".
	 */
	path = &state->path;
	ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
	for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
		if (blk->disk_blkno) {
C
Christoph Hellwig 已提交
1629 1630 1631
			error = xfs_da3_node_read_mapped(state->args->trans,
					state->args->dp, blk->disk_blkno,
					&blk->bp, XFS_ATTR_FORK);
L
Linus Torvalds 已提交
1632
			if (error)
E
Eric Sandeen 已提交
1633
				return error;
L
Linus Torvalds 已提交
1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646
		} else {
			blk->bp = NULL;
		}
	}

	/*
	 * Roll down the "altpath" in the state structure, storing the on-disk
	 * block number for those buffers in the "altpath".
	 */
	path = &state->altpath;
	ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
	for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
		if (blk->disk_blkno) {
C
Christoph Hellwig 已提交
1647 1648 1649
			error = xfs_da3_node_read_mapped(state->args->trans,
					state->args->dp, blk->disk_blkno,
					&blk->bp, XFS_ATTR_FORK);
L
Linus Torvalds 已提交
1650
			if (error)
E
Eric Sandeen 已提交
1651
				return error;
L
Linus Torvalds 已提交
1652 1653 1654 1655 1656
		} else {
			blk->bp = NULL;
		}
	}

E
Eric Sandeen 已提交
1657
	return 0;
L
Linus Torvalds 已提交
1658 1659 1660
}

/*
1661
 * Retrieve the attribute data from a node attribute list.
L
Linus Torvalds 已提交
1662 1663 1664 1665
 *
 * This routine gets called for any attribute fork that has more than one
 * block, ie: both true Btree attr lists and for single-leaf-blocks with
 * "remote" values taking up more blocks.
1666 1667
 *
 * Returns 0 on successful retrieval, otherwise an error.
L
Linus Torvalds 已提交
1668
 */
1669
STATIC int
1670 1671
xfs_attr_node_get(
	struct xfs_da_args	*args)
L
Linus Torvalds 已提交
1672
{
1673 1674 1675 1676
	struct xfs_da_state	*state;
	struct xfs_da_state_blk	*blk;
	int			i;
	int			error;
L
Linus Torvalds 已提交
1677

1678 1679
	trace_xfs_attr_node_get(args);

L
Linus Torvalds 已提交
1680 1681 1682
	/*
	 * Search to see if name exists, and get back a pointer to it.
	 */
1683 1684
	error = xfs_attr_node_hasname(args, &state);
	if (error != -EEXIST)
1685 1686 1687 1688 1689 1690
		goto out_release;

	/*
	 * Get the value, local or "remote"
	 */
	blk = &state->path.blk[state->path.active - 1];
1691
	error = xfs_attr3_leaf_getvalue(blk->bp, args);
L
Linus Torvalds 已提交
1692 1693 1694 1695

	/*
	 * If not in a transaction, we have to release all the buffers.
	 */
1696
out_release:
1697
	for (i = 0; state != NULL && i < state->path.active; i++) {
1698
		xfs_trans_brelse(args->trans, state->path.blk[i].bp);
L
Linus Torvalds 已提交
1699 1700 1701
		state->path.blk[i].bp = NULL;
	}

1702 1703 1704
	if (state)
		xfs_da_state_free(state);
	return error;
L
Linus Torvalds 已提交
1705
}
1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722

/* Returns true if the attribute entry name is valid. */
bool
xfs_attr_namecheck(
	const void	*name,
	size_t		length)
{
	/*
	 * MAXNAMELEN includes the trailing null, but (name/length) leave it
	 * out, so use >= for the length check.
	 */
	if (length >= MAXNAMELEN)
		return false;

	/* There shouldn't be any nulls here */
	return !memchr(name, 0, length);
}