dir.c 46.1 KB
Newer Older
D
David Teigland 已提交
1 2
/*
 * Copyright (C) Sistina Software, Inc.  1997-2003 All rights reserved.
3
 * Copyright (C) 2004-2006 Red Hat, Inc.  All rights reserved.
D
David Teigland 已提交
4 5 6 7 8 9 10
 *
 * This copyrighted material is made available to anyone wishing to use,
 * modify, copy, or redistribute it subject to the terms and conditions
 * of the GNU General Public License v.2.
 */

/*
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 50 51 52 53 54
 * Implements Extendible Hashing as described in:
 *   "Extendible Hashing" by Fagin, et al in
 *     __ACM Trans. on Database Systems__, Sept 1979.
 *
 *
 * Here's the layout of dirents which is essentially the same as that of ext2
 * within a single block. The field de_name_len is the number of bytes
 * actually required for the name (no null terminator). The field de_rec_len
 * is the number of bytes allocated to the dirent. The offset of the next
 * dirent in the block is (dirent + dirent->de_rec_len). When a dirent is
 * deleted, the preceding dirent inherits its allocated space, ie
 * prev->de_rec_len += deleted->de_rec_len. Since the next dirent is obtained
 * by adding de_rec_len to the current dirent, this essentially causes the
 * deleted dirent to get jumped over when iterating through all the dirents.
 *
 * When deleting the first dirent in a block, there is no previous dirent so
 * the field de_ino is set to zero to designate it as deleted. When allocating
 * a dirent, gfs2_dirent_alloc iterates through the dirents in a block. If the
 * first dirent has (de_ino == 0) and de_rec_len is large enough, this first
 * dirent is allocated. Otherwise it must go through all the 'used' dirents
 * searching for one in which the amount of total space minus the amount of
 * used space will provide enough space for the new dirent.
 *
 * There are two types of blocks in which dirents reside. In a stuffed dinode,
 * the dirents begin at offset sizeof(struct gfs2_dinode) from the beginning of
 * the block.  In leaves, they begin at offset sizeof(struct gfs2_leaf) from the
 * beginning of the leaf block. The dirents reside in leaves when
 *
 * dip->i_di.di_flags & GFS2_DIF_EXHASH is true
 *
 * Otherwise, the dirents are "linear", within a single stuffed dinode block.
 *
 * When the dirents are in leaves, the actual contents of the directory file are
 * used as an array of 64-bit block pointers pointing to the leaf blocks. The
 * dirents are NOT in the directory file itself. There can be more than one
 * block pointer in the array that points to the same leaf. In fact, when a
 * directory is first converted from linear to exhash, all of the pointers
 * point to the same leaf.
 *
 * When a leaf is completely full, the size of the hash table can be
 * doubled unless it is already at the maximum size which is hard coded into
 * GFS2_DIR_MAX_DEPTH. After that, leaves are chained together in a linked list,
 * but never before the maximum hash table size has been reached.
 */
D
David Teigland 已提交
55 56 57 58 59 60

#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <linux/buffer_head.h>
#include <linux/sort.h>
61
#include <linux/gfs2_ondisk.h>
62
#include <linux/crc32.h>
63
#include <linux/vmalloc.h>
D
David Teigland 已提交
64 65

#include "gfs2.h"
66 67
#include "lm_interface.h"
#include "incore.h"
D
David Teigland 已提交
68 69 70 71 72 73 74
#include "dir.h"
#include "glock.h"
#include "inode.h"
#include "meta_io.h"
#include "quota.h"
#include "rgrp.h"
#include "trans.h"
75
#include "bmap.h"
76
#include "util.h"
D
David Teigland 已提交
77 78 79 80 81 82 83 84 85 86 87

#define IS_LEAF     1 /* Hashed (leaf) directory */
#define IS_DINODE   2 /* Linear (stuffed dinode block) directory */

#define gfs2_disk_hash2offset(h) (((uint64_t)(h)) >> 1)
#define gfs2_dir_offset2hash(p) ((uint32_t)(((uint64_t)(p)) << 1))

typedef int (*leaf_call_t) (struct gfs2_inode *dip,
			    uint32_t index, uint32_t len, uint64_t leaf_no,
			    void *data);

88 89 90

int gfs2_dir_get_new_buffer(struct gfs2_inode *ip, uint64_t block,
			    struct buffer_head **bhp)
91 92 93
{
	struct buffer_head *bh;

94 95 96 97
	bh = gfs2_meta_new(ip->i_gl, block);
	gfs2_trans_add_bh(ip->i_gl, bh, 1);
	gfs2_metatype_set(bh, GFS2_METATYPE_JD, GFS2_FORMAT_JD);
	gfs2_buffer_clear_tail(bh, sizeof(struct gfs2_meta_header));
98 99 100 101
	*bhp = bh;
	return 0;
}

102 103 104 105 106
static int gfs2_dir_get_existing_buffer(struct gfs2_inode *ip, uint64_t block,
					struct buffer_head **bhp)
{
	struct buffer_head *bh;
	int error;
107

108 109 110
	error = gfs2_meta_read(ip->i_gl, block, DIO_START | DIO_WAIT, &bh);
	if (error)
		return error;
111
	if (gfs2_metatype_check(GFS2_SB(&ip->i_inode), bh, GFS2_METATYPE_JD)) {
112 113 114 115 116 117
		brelse(bh);
		return -EIO;
	}
	*bhp = bh;
	return 0;
}
118 119 120 121 122 123 124 125 126 127 128 129 130

static int gfs2_dir_write_stuffed(struct gfs2_inode *ip, const char *buf,
				  unsigned int offset, unsigned int size)
                               
{
	struct buffer_head *dibh;
	int error;

	error = gfs2_meta_inode_buffer(ip, &dibh);
	if (error)
		return error;

	gfs2_trans_add_bh(ip->i_gl, dibh, 1);
131
	memcpy(dibh->b_data + offset + sizeof(struct gfs2_dinode), buf, size);
132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155
	if (ip->i_di.di_size < offset + size)
		ip->i_di.di_size = offset + size;
	ip->i_di.di_mtime = ip->i_di.di_ctime = get_seconds();
	gfs2_dinode_out(&ip->i_di, dibh->b_data);

	brelse(dibh);

	return size;
}



/**
 * gfs2_dir_write_data - Write directory information to the inode
 * @ip: The GFS2 inode
 * @buf: The buffer containing information to be written
 * @offset: The file offset to start writing at
 * @size: The amount of data to write
 *
 * Returns: The number of bytes correctly written or error code
 */
static int gfs2_dir_write_data(struct gfs2_inode *ip, const char *buf,
			       uint64_t offset, unsigned int size)
{
156
	struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
157 158 159 160 161 162 163 164 165 166 167 168
	struct buffer_head *dibh;
	uint64_t lblock, dblock;
	uint32_t extlen = 0;
	unsigned int o;
	int copied = 0;
	int error = 0;

	if (!size)
		return 0;

	if (gfs2_is_stuffed(ip) &&
	    offset + size <= sdp->sd_sb.sb_bsize - sizeof(struct gfs2_dinode))
169 170
		return gfs2_dir_write_stuffed(ip, buf, (unsigned int)offset,
					      size);
171 172 173 174 175 176 177

	if (gfs2_assert_warn(sdp, gfs2_is_jdata(ip)))
		return -EINVAL;

	if (gfs2_is_stuffed(ip)) {
		error = gfs2_unstuff_dinode(ip, NULL, NULL);
		if (error)
178
			return error;
179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194
	}

	lblock = offset;
	o = do_div(lblock, sdp->sd_jbsize) + sizeof(struct gfs2_meta_header);

	while (copied < size) {
		unsigned int amount;
		struct buffer_head *bh;
		int new;

		amount = size - copied;
		if (amount > sdp->sd_sb.sb_bsize - o)
			amount = sdp->sd_sb.sb_bsize - o;

		if (!extlen) {
			new = 1;
195
			error = gfs2_extent_map(&ip->i_inode, lblock, &new,
S
Steven Whitehouse 已提交
196
						&dblock, &extlen);
197 198 199 200 201 202 203
			if (error)
				goto fail;
			error = -EIO;
			if (gfs2_assert_withdraw(sdp, dblock))
				goto fail;
		}

204 205 206 207 208
		if (amount == sdp->sd_jbsize || new)
			error = gfs2_dir_get_new_buffer(ip, dblock, &bh);
		else
			error = gfs2_dir_get_existing_buffer(ip, dblock, &bh);

209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246
		if (error)
			goto fail;

		gfs2_trans_add_bh(ip->i_gl, bh, 1);
		memcpy(bh->b_data + o, buf, amount);
		brelse(bh);
		if (error)
			goto fail;

		copied += amount;
		lblock++;
		dblock++;
		extlen--;

		o = sizeof(struct gfs2_meta_header);
	}

out:
	error = gfs2_meta_inode_buffer(ip, &dibh);
	if (error)
		return error;

	if (ip->i_di.di_size < offset + copied)
		ip->i_di.di_size = offset + copied;
	ip->i_di.di_mtime = ip->i_di.di_ctime = get_seconds();

	gfs2_trans_add_bh(ip->i_gl, dibh, 1);
	gfs2_dinode_out(&ip->i_di, dibh->b_data);
	brelse(dibh);

	return copied;
fail:
	if (copied)
		goto out;
	return error;
}

static int gfs2_dir_read_stuffed(struct gfs2_inode *ip, char *buf,
247
				 unsigned int offset, unsigned int size)
248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274
{
	struct buffer_head *dibh;
	int error;

	error = gfs2_meta_inode_buffer(ip, &dibh);
	if (!error) {
		offset += sizeof(struct gfs2_dinode);
		memcpy(buf, dibh->b_data + offset, size);
		brelse(dibh);
	}

	return (error) ? error : size;
}


/**
 * gfs2_dir_read_data - Read a data from a directory inode
 * @ip: The GFS2 Inode
 * @buf: The buffer to place result into
 * @offset: File offset to begin jdata_readng from
 * @size: Amount of data to transfer
 *
 * Returns: The amount of data actually copied or the error
 */
static int gfs2_dir_read_data(struct gfs2_inode *ip, char *buf,
			      uint64_t offset, unsigned int size)
{
275
	struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291
	uint64_t lblock, dblock;
	uint32_t extlen = 0;
	unsigned int o;
	int copied = 0;
	int error = 0;

	if (offset >= ip->i_di.di_size)
		return 0;

	if ((offset + size) > ip->i_di.di_size)
		size = ip->i_di.di_size - offset;

	if (!size)
		return 0;

	if (gfs2_is_stuffed(ip))
292 293
		return gfs2_dir_read_stuffed(ip, buf, (unsigned int)offset,
					     size);
294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311

	if (gfs2_assert_warn(sdp, gfs2_is_jdata(ip)))
		return -EINVAL;

	lblock = offset;
	o = do_div(lblock, sdp->sd_jbsize) + sizeof(struct gfs2_meta_header);

	while (copied < size) {
		unsigned int amount;
		struct buffer_head *bh;
		int new;

		amount = size - copied;
		if (amount > sdp->sd_sb.sb_bsize - o)
			amount = sdp->sd_sb.sb_bsize - o;

		if (!extlen) {
			new = 0;
312
			error = gfs2_extent_map(&ip->i_inode, lblock, &new,
S
Steven Whitehouse 已提交
313
						&dblock, &extlen);
314 315 316 317 318 319 320 321
			if (error)
				goto fail;
		}

		if (extlen > 1)
			gfs2_meta_ra(ip->i_gl, dblock, extlen);

		if (dblock) {
322 323 324 325
			if (new)
				error = gfs2_dir_get_new_buffer(ip, dblock, &bh);
			else
				error = gfs2_dir_get_existing_buffer(ip, dblock, &bh);
326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348
			if (error)
				goto fail;
			dblock++;
			extlen--;
		} else
			bh = NULL;

		memcpy(buf, bh->b_data + o, amount);
		brelse(bh);
		if (error)
			goto fail;

		copied += amount;
		lblock++;

		o = sizeof(struct gfs2_meta_header);
	}

	return copied;
fail:
	return (copied) ? copied : error;
}

349
typedef int (*gfs2_dscan_t)(const struct gfs2_dirent *dent,
350 351
			    const struct qstr *name,
			    void *opaque);
352 353 354 355 356 357 358 359 360 361 362 363 364

static inline int __gfs2_dirent_find(const struct gfs2_dirent *dent,
				     const struct qstr *name, int ret)
{
	if (dent->de_inum.no_addr != 0 &&
	    be32_to_cpu(dent->de_hash) == name->hash &&
	    be16_to_cpu(dent->de_name_len) == name->len &&
	    memcmp((char *)(dent+1), name->name, name->len) == 0)
		return ret;
	return 0;
}

static int gfs2_dirent_find(const struct gfs2_dirent *dent,
365 366
			    const struct qstr *name,
			    void *opaque)
367 368 369 370 371
{
	return __gfs2_dirent_find(dent, name, 1);
}

static int gfs2_dirent_prev(const struct gfs2_dirent *dent,
372 373
			    const struct qstr *name,
			    void *opaque)
374 375 376 377 378 379 380
{
	return __gfs2_dirent_find(dent, name, 2);
}

/*
 * name->name holds ptr to start of block.
 * name->len holds size of block.
D
David Teigland 已提交
381
 */
382
static int gfs2_dirent_last(const struct gfs2_dirent *dent,
383 384
			    const struct qstr *name,
			    void *opaque)
385 386 387 388 389 390 391
{
	const char *start = name->name;
	const char *end = (const char *)dent + be16_to_cpu(dent->de_rec_len);
	if (name->len == (end - start))
		return 1;
	return 0;
}
D
David Teigland 已提交
392

393
static int gfs2_dirent_find_space(const struct gfs2_dirent *dent,
394 395
				  const struct qstr *name,
				  void *opaque)
D
David Teigland 已提交
396
{
397 398 399 400
	unsigned required = GFS2_DIRENT_SIZE(name->len);
	unsigned actual = GFS2_DIRENT_SIZE(be16_to_cpu(dent->de_name_len));
	unsigned totlen = be16_to_cpu(dent->de_rec_len);

401 402
	if (!dent->de_inum.no_addr)
		actual = GFS2_DIRENT_SIZE(0);
403 404 405 406 407
	if ((totlen - actual) >= required)
		return 1;
	return 0;
}

408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423
struct dirent_gather {
	const struct gfs2_dirent **pdent;
	unsigned offset;
};

static int gfs2_dirent_gather(const struct gfs2_dirent *dent,
			      const struct qstr *name,
			      void *opaque)
{
	struct dirent_gather *g = opaque;
	if (dent->de_inum.no_addr) {
		g->pdent[g->offset++] = dent;
	}
	return 0;
}

424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457
/*
 * Other possible things to check:
 * - Inode located within filesystem size (and on valid block)
 * - Valid directory entry type
 * Not sure how heavy-weight we want to make this... could also check
 * hash is correct for example, but that would take a lot of extra time.
 * For now the most important thing is to check that the various sizes
 * are correct.
 */
static int gfs2_check_dirent(struct gfs2_dirent *dent, unsigned int offset,
			     unsigned int size, unsigned int len, int first)
{
	const char *msg = "gfs2_dirent too small";
	if (unlikely(size < sizeof(struct gfs2_dirent)))
		goto error;
	msg = "gfs2_dirent misaligned";
	if (unlikely(offset & 0x7))
		goto error;
	msg = "gfs2_dirent points beyond end of block";
	if (unlikely(offset + size > len))
		goto error;
	msg = "zero inode number";
	if (unlikely(!first && !dent->de_inum.no_addr))
		goto error;
	msg = "name length is greater than space in dirent";
	if (dent->de_inum.no_addr &&
	    unlikely(sizeof(struct gfs2_dirent)+be16_to_cpu(dent->de_name_len) >
		     size))
		goto error;
	return 0;
error:
	printk(KERN_WARNING "gfs2_check_dirent: %s (%s)\n", msg,
	       first ? "first in block" : "not first in block");
	return -EIO;
D
David Teigland 已提交
458 459
}

460
static int gfs2_dirent_offset(const void *buf)
461
{
462 463
	const struct gfs2_meta_header *h = buf;
	int offset;
464 465 466

	BUG_ON(buf == NULL);

467
	switch(be32_to_cpu(h->mh_type)) {
468 469 470 471 472 473 474 475 476
	case GFS2_METATYPE_LF:
		offset = sizeof(struct gfs2_leaf);
		break;
	case GFS2_METATYPE_DI:
		offset = sizeof(struct gfs2_dinode);
		break;
	default:
		goto wrong_type;
	}
477 478 479
	return offset;
wrong_type:
	printk(KERN_WARNING "gfs2_scan_dirent: wrong block type %u\n",
480
	       be32_to_cpu(h->mh_type));
481 482 483 484 485 486 487 488 489 490 491 492 493
	return -1;
}

static struct gfs2_dirent *gfs2_dirent_scan(struct inode *inode,
					    void *buf,
					    unsigned int len, gfs2_dscan_t scan,
					    const struct qstr *name,
					    void *opaque)
{
	struct gfs2_dirent *dent, *prev;
	unsigned offset;
	unsigned size;
	int ret = 0;
494

495 496 497 498 499
	ret = gfs2_dirent_offset(buf);
	if (ret < 0)
		goto consist_inode;

	offset = ret;
500 501 502 503 504 505
	prev = NULL;
	dent = (struct gfs2_dirent *)(buf + offset);
	size = be16_to_cpu(dent->de_rec_len);
	if (gfs2_check_dirent(dent, offset, size, len, 1))
		goto consist_inode;
	do {
506
		ret = scan(dent, name, opaque);
507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531
		if (ret)
			break;
		offset += size;
		if (offset == len)
			break;
		prev = dent;
		dent = (struct gfs2_dirent *)(buf + offset);
		size = be16_to_cpu(dent->de_rec_len);
		if (gfs2_check_dirent(dent, offset, size, len, 0))
			goto consist_inode;
	} while(1);

	switch(ret) {
	case 0:
		return NULL;
	case 1:
		return dent;
	case 2:
		return prev ? prev : dent;
	default:
		BUG_ON(ret > 0);
		return ERR_PTR(ret);
	}

consist_inode:
532
	gfs2_consist_inode(GFS2_I(inode));
533 534 535 536
	return ERR_PTR(-EIO);
}


D
David Teigland 已提交
537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552
/**
 * dirent_first - Return the first dirent
 * @dip: the directory
 * @bh: The buffer
 * @dent: Pointer to list of dirents
 *
 * return first dirent whether bh points to leaf or stuffed dinode
 *
 * Returns: IS_LEAF, IS_DINODE, or -errno
 */

static int dirent_first(struct gfs2_inode *dip, struct buffer_head *bh,
			struct gfs2_dirent **dent)
{
	struct gfs2_meta_header *h = (struct gfs2_meta_header *)bh->b_data;

553
	if (be32_to_cpu(h->mh_type) == GFS2_METATYPE_LF) {
554
		if (gfs2_meta_check(GFS2_SB(&dip->i_inode), bh))
D
David Teigland 已提交
555 556 557 558 559
			return -EIO;
		*dent = (struct gfs2_dirent *)(bh->b_data +
					       sizeof(struct gfs2_leaf));
		return IS_LEAF;
	} else {
560
		if (gfs2_metatype_check(GFS2_SB(&dip->i_inode), bh, GFS2_METATYPE_DI))
D
David Teigland 已提交
561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581
			return -EIO;
		*dent = (struct gfs2_dirent *)(bh->b_data +
					       sizeof(struct gfs2_dinode));
		return IS_DINODE;
	}
}

/**
 * dirent_next - Next dirent
 * @dip: the directory
 * @bh: The buffer
 * @dent: Pointer to list of dirents
 *
 * Returns: 0 on success, error code otherwise
 */

static int dirent_next(struct gfs2_inode *dip, struct buffer_head *bh,
		       struct gfs2_dirent **dent)
{
	struct gfs2_dirent *tmp, *cur;
	char *bh_end;
582
	uint16_t cur_rec_len;
D
David Teigland 已提交
583 584 585

	cur = *dent;
	bh_end = bh->b_data + bh->b_size;
586
	cur_rec_len = be16_to_cpu(cur->de_rec_len);
D
David Teigland 已提交
587 588 589 590 591 592 593 594 595 596 597

	if ((char *)cur + cur_rec_len >= bh_end) {
		if ((char *)cur + cur_rec_len > bh_end) {
			gfs2_consist_inode(dip);
			return -EIO;
		}
		return -ENOENT;
	}

	tmp = (struct gfs2_dirent *)((char *)cur + cur_rec_len);

598
	if ((char *)tmp + be16_to_cpu(tmp->de_rec_len) > bh_end) {
D
David Teigland 已提交
599 600 601
		gfs2_consist_inode(dip);
		return -EIO;
	}
602 603 604 605 606 607

	if (cur_rec_len == 0) {
		gfs2_consist_inode(dip);
		return -EIO;
	}

D
David Teigland 已提交
608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630
        /* Only the first dent could ever have de_inum.no_addr == 0 */
	if (!tmp->de_inum.no_addr) {
		gfs2_consist_inode(dip);
		return -EIO;
	}

	*dent = tmp;

	return 0;
}

/**
 * dirent_del - Delete a dirent
 * @dip: The GFS2 inode
 * @bh: The buffer
 * @prev: The previous dirent
 * @cur: The current dirent
 *
 */

static void dirent_del(struct gfs2_inode *dip, struct buffer_head *bh,
		       struct gfs2_dirent *prev, struct gfs2_dirent *cur)
{
631
	uint16_t cur_rec_len, prev_rec_len;
D
David Teigland 已提交
632 633 634 635 636 637

	if (!cur->de_inum.no_addr) {
		gfs2_consist_inode(dip);
		return;
	}

638
	gfs2_trans_add_bh(dip->i_gl, bh, 1);
D
David Teigland 已提交
639 640 641 642 643 644 645 646 647 648 649 650

	/* If there is no prev entry, this is the first entry in the block.
	   The de_rec_len is already as big as it needs to be.  Just zero
	   out the inode number and return.  */

	if (!prev) {
		cur->de_inum.no_addr = 0;	/* No endianess worries */
		return;
	}

	/*  Combine this dentry with the previous one.  */

651 652
	prev_rec_len = be16_to_cpu(prev->de_rec_len);
	cur_rec_len = be16_to_cpu(cur->de_rec_len);
D
David Teigland 已提交
653 654 655 656 657 658 659

	if ((char *)prev + prev_rec_len != (char *)cur)
		gfs2_consist_inode(dip);
	if ((char *)cur + cur_rec_len > bh->b_data + bh->b_size)
		gfs2_consist_inode(dip);

	prev_rec_len += cur_rec_len;
660
	prev->de_rec_len = cpu_to_be16(prev_rec_len);
D
David Teigland 已提交
661 662
}

663 664 665
/*
 * Takes a dent from which to grab space as an argument. Returns the
 * newly created dent.
D
David Teigland 已提交
666
 */
667 668 669 670
static struct gfs2_dirent *gfs2_init_dirent(struct inode *inode,
					    struct gfs2_dirent *dent,
					    const struct qstr *name,
					    struct buffer_head *bh)
D
David Teigland 已提交
671
{
672
	struct gfs2_inode *ip = GFS2_I(inode);
673 674 675 676 677 678 679 680 681 682 683 684
	struct gfs2_dirent *ndent;
	unsigned offset = 0, totlen;

	if (dent->de_inum.no_addr)
		offset = GFS2_DIRENT_SIZE(be16_to_cpu(dent->de_name_len));
	totlen = be16_to_cpu(dent->de_rec_len);
	BUG_ON(offset + name->len > totlen);
	gfs2_trans_add_bh(ip->i_gl, bh, 1);
	ndent = (struct gfs2_dirent *)((char *)dent + offset);
	dent->de_rec_len = cpu_to_be16(offset);
	gfs2_qstr2dirent(name, totlen - offset, ndent);
	return ndent;
D
David Teigland 已提交
685 686
}

687 688 689
static struct gfs2_dirent *gfs2_dirent_alloc(struct inode *inode,
					     struct buffer_head *bh,
					     const struct qstr *name)
D
David Teigland 已提交
690 691
{
	struct gfs2_dirent *dent;
692 693
	dent = gfs2_dirent_scan(inode, bh->b_data, bh->b_size, 
				gfs2_dirent_find_space, name, NULL);
694 695 696
	if (!dent || IS_ERR(dent))
		return dent;
	return gfs2_init_dirent(inode, dent, name, bh);
D
David Teigland 已提交
697 698 699 700 701 702 703 704
}

static int get_leaf(struct gfs2_inode *dip, uint64_t leaf_no,
		    struct buffer_head **bhp)
{
	int error;

	error = gfs2_meta_read(dip->i_gl, leaf_no, DIO_START | DIO_WAIT, bhp);
705 706
	if (!error && gfs2_metatype_check(GFS2_SB(&dip->i_inode), *bhp, GFS2_METATYPE_LF)) {
		/* printk(KERN_INFO "block num=%llu\n", leaf_no); */
D
David Teigland 已提交
707
		error = -EIO;
708
	}
D
David Teigland 已提交
709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727

	return error;
}

/**
 * get_leaf_nr - Get a leaf number associated with the index
 * @dip: The GFS2 inode
 * @index:
 * @leaf_out:
 *
 * Returns: 0 on success, error code otherwise
 */

static int get_leaf_nr(struct gfs2_inode *dip, uint32_t index,
		       uint64_t *leaf_out)
{
	uint64_t leaf_no;
	int error;

728
	error = gfs2_dir_read_data(dip, (char *)&leaf_no,
D
David Teigland 已提交
729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751
				    index * sizeof(uint64_t),
				    sizeof(uint64_t));
	if (error != sizeof(uint64_t))
		return (error < 0) ? error : -EIO;

	*leaf_out = be64_to_cpu(leaf_no);

	return 0;
}

static int get_first_leaf(struct gfs2_inode *dip, uint32_t index,
			  struct buffer_head **bh_out)
{
	uint64_t leaf_no;
	int error;

	error = get_leaf_nr(dip, index, &leaf_no);
	if (!error)
		error = get_leaf(dip, leaf_no, bh_out);

	return error;
}

752 753 754 755
static struct gfs2_dirent *gfs2_dirent_search(struct inode *inode,
					      const struct qstr *name,
					      gfs2_dscan_t scan,
					      struct buffer_head **pbh)
D
David Teigland 已提交
756
{
757 758
	struct buffer_head *bh;
	struct gfs2_dirent *dent;
759
	struct gfs2_inode *ip = GFS2_I(inode);
D
David Teigland 已提交
760 761
	int error;

762 763 764 765 766 767 768 769 770
	if (ip->i_di.di_flags & GFS2_DIF_EXHASH) {
		struct gfs2_leaf *leaf;
		unsigned hsize = 1 << ip->i_di.di_depth;
		unsigned index;
		u64 ln;
		if (hsize * sizeof(u64) != ip->i_di.di_size) {
			gfs2_consist_inode(ip);
			return ERR_PTR(-EIO);
		}
771
		
772 773 774 775 776 777
		index = name->hash >> (32 - ip->i_di.di_depth);
		error = get_first_leaf(ip, index, &bh);
		if (error)
			return ERR_PTR(error);
		do {
			dent = gfs2_dirent_scan(inode, bh->b_data, bh->b_size,
778
						scan, name, NULL);
779 780 781 782
			if (dent)
				goto got_dent;
			leaf = (struct gfs2_leaf *)bh->b_data;
			ln = be64_to_cpu(leaf->lf_next);
783
			brelse(bh);
784 785
			if (!ln)
				break;
786
			
787 788
			error = get_leaf(ip, ln, &bh);
		} while(!error);
D
David Teigland 已提交
789

790
		return error ? ERR_PTR(error) : NULL;
D
David Teigland 已提交
791 792
	}

793
	
794 795 796
	error = gfs2_meta_inode_buffer(ip, &bh);
	if (error)
		return ERR_PTR(error);
797
	dent = gfs2_dirent_scan(inode, bh->b_data, bh->b_size, scan, name, NULL);
798
got_dent:
799
	if (unlikely(dent == NULL || IS_ERR(dent))) {
800 801 802
		brelse(bh);
		bh = NULL;
	}
803 804 805
	*pbh = bh;
	return dent;
}
D
David Teigland 已提交
806

807 808
static struct gfs2_leaf *new_leaf(struct inode *inode, struct buffer_head **pbh, u16 depth)
{
809
	struct gfs2_inode *ip = GFS2_I(inode);
810 811 812 813
	u64 bn = gfs2_alloc_meta(ip);
	struct buffer_head *bh = gfs2_meta_new(ip->i_gl, bn);
	struct gfs2_leaf *leaf;
	struct gfs2_dirent *dent;
814
	struct qstr name = { .name = "", .len = 0, .hash = 0 };
815 816
	if (!bh)
		return NULL;
817
	
818 819 820 821 822 823 824 825 826
	gfs2_trans_add_bh(ip->i_gl, bh, 1);
	gfs2_metatype_set(bh, GFS2_METATYPE_LF, GFS2_FORMAT_LF);
	leaf = (struct gfs2_leaf *)bh->b_data;
	leaf->lf_depth = cpu_to_be16(depth);
	leaf->lf_entries = cpu_to_be16(0);
	leaf->lf_dirent_format = cpu_to_be16(GFS2_FORMAT_DE);
	leaf->lf_next = cpu_to_be64(0);
	memset(leaf->lf_reserved, 0, sizeof(leaf->lf_reserved));
	dent = (struct gfs2_dirent *)(leaf+1);
827
	gfs2_qstr2dirent(&name, bh->b_size - sizeof(struct gfs2_leaf), dent);
828 829
	*pbh = bh;
	return leaf;
D
David Teigland 已提交
830 831 832 833 834 835 836 837 838
}

/**
 * dir_make_exhash - Convert a stuffed directory into an ExHash directory
 * @dip: The GFS2 inode
 *
 * Returns: 0 on success, error code otherwise
 */

839
static int dir_make_exhash(struct inode *inode)
D
David Teigland 已提交
840
{
841 842
	struct gfs2_inode *dip = GFS2_I(inode);
	struct gfs2_sbd *sdp = GFS2_SB(inode);
D
David Teigland 已提交
843
	struct gfs2_dirent *dent;
844
	struct qstr args;
D
David Teigland 已提交
845 846 847 848 849 850 851 852 853 854 855 856 857
	struct buffer_head *bh, *dibh;
	struct gfs2_leaf *leaf;
	int y;
	uint32_t x;
	uint64_t *lp, bn;
	int error;

	error = gfs2_meta_inode_buffer(dip, &dibh);
	if (error)
		return error;

	/*  Turn over a new leaf  */

858 859 860 861
	leaf = new_leaf(inode, &bh, 0);
	if (!leaf)
		return -ENOSPC;
	bn = bh->b_blocknr;
D
David Teigland 已提交
862 863 864 865 866 867 868 869 870 871 872 873

	gfs2_assert(sdp, dip->i_di.di_entries < (1 << 16));
	leaf->lf_entries = cpu_to_be16(dip->i_di.di_entries);

	/*  Copy dirents  */

	gfs2_buffer_copy_tail(bh, sizeof(struct gfs2_leaf), dibh,
			     sizeof(struct gfs2_dinode));

	/*  Find last entry  */

	x = 0;
874 875 876
	args.len = bh->b_size - sizeof(struct gfs2_dinode) +
		   sizeof(struct gfs2_leaf);
	args.name = bh->b_data;
877
	dent = gfs2_dirent_scan(&dip->i_inode, bh->b_data, bh->b_size,
878
				gfs2_dirent_last, &args, NULL);
879 880 881 882 883 884 885 886 887
	if (!dent) {
		brelse(bh);
		brelse(dibh);
		return -EIO;
	}
	if (IS_ERR(dent)) {
		brelse(bh);
		brelse(dibh);
		return PTR_ERR(dent);
D
David Teigland 已提交
888 889 890 891 892
	}

	/*  Adjust the last dirent's record length
	   (Remember that dent still points to the last entry.)  */

893
	dent->de_rec_len = cpu_to_be16(be16_to_cpu(dent->de_rec_len) +
D
David Teigland 已提交
894
		sizeof(struct gfs2_dinode) -
895
		sizeof(struct gfs2_leaf));
D
David Teigland 已提交
896 897 898 899 900 901

	brelse(bh);

	/*  We're done with the new leaf block, now setup the new
	    hash table.  */

902
	gfs2_trans_add_bh(dip->i_gl, dibh, 1);
D
David Teigland 已提交
903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933
	gfs2_buffer_clear_tail(dibh, sizeof(struct gfs2_dinode));

	lp = (uint64_t *)(dibh->b_data + sizeof(struct gfs2_dinode));

	for (x = sdp->sd_hash_ptrs; x--; lp++)
		*lp = cpu_to_be64(bn);

	dip->i_di.di_size = sdp->sd_sb.sb_bsize / 2;
	dip->i_di.di_blocks++;
	dip->i_di.di_flags |= GFS2_DIF_EXHASH;
	dip->i_di.di_payload_format = 0;

	for (x = sdp->sd_hash_ptrs, y = -1; x; x >>= 1, y++) ;
	dip->i_di.di_depth = y;

	gfs2_dinode_out(&dip->i_di, dibh->b_data);

	brelse(dibh);

	return 0;
}

/**
 * dir_split_leaf - Split a leaf block into two
 * @dip: The GFS2 inode
 * @index:
 * @leaf_no:
 *
 * Returns: 0 on success, error code on failure
 */

934
static int dir_split_leaf(struct inode *inode, const struct qstr *name)
D
David Teigland 已提交
935
{
936
	struct gfs2_inode *dip = GFS2_I(inode);
D
David Teigland 已提交
937 938 939 940
	struct buffer_head *nbh, *obh, *dibh;
	struct gfs2_leaf *nleaf, *oleaf;
	struct gfs2_dirent *dent, *prev = NULL, *next = NULL, *new;
	uint32_t start, len, half_len, divider;
941 942
	uint64_t bn, *lp, leaf_no;
	uint32_t index;
D
David Teigland 已提交
943 944 945
	int x, moved = 0;
	int error;

946 947 948 949
	index = name->hash >> (32 - dip->i_di.di_depth);
	error = get_leaf_nr(dip, index, &leaf_no);
	if (error)
		return error;
D
David Teigland 已提交
950 951 952 953

	/*  Get the old leaf block  */
	error = get_leaf(dip, leaf_no, &obh);
	if (error)
954
		return error;
D
David Teigland 已提交
955 956

	oleaf = (struct gfs2_leaf *)obh->b_data;
957 958 959 960 961 962
	if (dip->i_di.di_depth == be16_to_cpu(oleaf->lf_depth)) {
		brelse(obh);
		return 1; /* can't split */
	}

	gfs2_trans_add_bh(dip->i_gl, obh, 1);
D
David Teigland 已提交
963

964 965 966 967 968 969
	nleaf = new_leaf(inode, &nbh, be16_to_cpu(oleaf->lf_depth) + 1);
	if (!nleaf) {
		brelse(obh);
		return -ENOSPC;
	}
	bn = nbh->b_blocknr;
D
David Teigland 已提交
970

971
	/*  Compute the start and len of leaf pointers in the hash table.  */
D
David Teigland 已提交
972 973 974
	len = 1 << (dip->i_di.di_depth - be16_to_cpu(oleaf->lf_depth));
	half_len = len >> 1;
	if (!half_len) {
975
		printk(KERN_WARNING "di_depth %u lf_depth %u index %u\n", dip->i_di.di_depth, be16_to_cpu(oleaf->lf_depth), index);
D
David Teigland 已提交
976 977 978 979 980 981 982 983 984 985
		gfs2_consist_inode(dip);
		error = -EIO;
		goto fail_brelse;
	}

	start = (index & ~(len - 1));

	/* Change the pointers.
	   Don't bother distinguishing stuffed from non-stuffed.
	   This code is complicated enough already. */
986
	lp = kmalloc(half_len * sizeof(uint64_t), GFP_NOFS | __GFP_NOFAIL);
D
David Teigland 已提交
987 988 989 990
	/*  Change the pointers  */
	for (x = 0; x < half_len; x++)
		lp[x] = cpu_to_be64(bn);

991
	error = gfs2_dir_write_data(dip, (char *)lp, start * sizeof(uint64_t),
992
				    half_len * sizeof(uint64_t));
D
David Teigland 已提交
993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013
	if (error != half_len * sizeof(uint64_t)) {
		if (error >= 0)
			error = -EIO;
		goto fail_lpfree;
	}

	kfree(lp);

	/*  Compute the divider  */
	divider = (start + half_len) << (32 - dip->i_di.di_depth);

	/*  Copy the entries  */
	dirent_first(dip, obh, &dent);

	do {
		next = dent;
		if (dirent_next(dip, obh, &next))
			next = NULL;

		if (dent->de_inum.no_addr &&
		    be32_to_cpu(dent->de_hash) < divider) {
1014 1015 1016 1017
			struct qstr str;
			str.name = (char*)(dent+1);
			str.len = be16_to_cpu(dent->de_name_len);
			str.hash = be32_to_cpu(dent->de_hash);
1018
			new = gfs2_dirent_alloc(inode, nbh, &str);
1019 1020 1021 1022
			if (IS_ERR(new)) {
				error = PTR_ERR(new);
				break;
			}
D
David Teigland 已提交
1023 1024 1025

			new->de_inum = dent->de_inum; /* No endian worries */
			new->de_type = dent->de_type; /* No endian worries */
1026
			nleaf->lf_entries = cpu_to_be16(be16_to_cpu(nleaf->lf_entries)+1);
D
David Teigland 已提交
1027 1028 1029 1030 1031

			dirent_del(dip, obh, prev, dent);

			if (!oleaf->lf_entries)
				gfs2_consist_inode(dip);
1032
			oleaf->lf_entries = cpu_to_be16(be16_to_cpu(oleaf->lf_entries)-1);
D
David Teigland 已提交
1033 1034 1035 1036 1037

			if (!prev)
				prev = dent;

			moved = 1;
1038
		} else {
D
David Teigland 已提交
1039
			prev = dent;
1040
		}
D
David Teigland 已提交
1041
		dent = next;
1042
	} while (dent);
D
David Teigland 已提交
1043

1044
	oleaf->lf_depth = nleaf->lf_depth;
D
David Teigland 已提交
1045 1046

	error = gfs2_meta_inode_buffer(dip, &dibh);
1047
	if (!gfs2_assert_withdraw(GFS2_SB(&dip->i_inode), !error)) {
D
David Teigland 已提交
1048 1049 1050 1051 1052 1053 1054 1055 1056 1057
		dip->i_di.di_blocks++;
		gfs2_dinode_out(&dip->i_di, dibh->b_data);
		brelse(dibh);
	}

	brelse(obh);
	brelse(nbh);

	return error;

1058
fail_lpfree:
D
David Teigland 已提交
1059 1060
	kfree(lp);

1061
fail_brelse:
D
David Teigland 已提交
1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075
	brelse(obh);
	brelse(nbh);
	return error;
}

/**
 * dir_double_exhash - Double size of ExHash table
 * @dip: The GFS2 dinode
 *
 * Returns: 0 on success, error code on failure
 */

static int dir_double_exhash(struct gfs2_inode *dip)
{
1076
	struct gfs2_sbd *sdp = GFS2_SB(&dip->i_inode);
D
David Teigland 已提交
1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095
	struct buffer_head *dibh;
	uint32_t hsize;
	uint64_t *buf;
	uint64_t *from, *to;
	uint64_t block;
	int x;
	int error = 0;

	hsize = 1 << dip->i_di.di_depth;
	if (hsize * sizeof(uint64_t) != dip->i_di.di_size) {
		gfs2_consist_inode(dip);
		return -EIO;
	}

	/*  Allocate both the "from" and "to" buffers in one big chunk  */

	buf = kcalloc(3, sdp->sd_hash_bsize, GFP_KERNEL | __GFP_NOFAIL);

	for (block = dip->i_di.di_size >> sdp->sd_hash_bsize_shift; block--;) {
1096
		error = gfs2_dir_read_data(dip, (char *)buf,
D
David Teigland 已提交
1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112
					    block * sdp->sd_hash_bsize,
					    sdp->sd_hash_bsize);
		if (error != sdp->sd_hash_bsize) {
			if (error >= 0)
				error = -EIO;
			goto fail;
		}

		from = buf;
		to = (uint64_t *)((char *)buf + sdp->sd_hash_bsize);

		for (x = sdp->sd_hash_ptrs; x--; from++) {
			*to++ = *from;	/*  No endianess worries  */
			*to++ = *from;
		}

1113
		error = gfs2_dir_write_data(dip,
D
David Teigland 已提交
1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158
					     (char *)buf + sdp->sd_hash_bsize,
					     block * sdp->sd_sb.sb_bsize,
					     sdp->sd_sb.sb_bsize);
		if (error != sdp->sd_sb.sb_bsize) {
			if (error >= 0)
				error = -EIO;
			goto fail;
		}
	}

	kfree(buf);

	error = gfs2_meta_inode_buffer(dip, &dibh);
	if (!gfs2_assert_withdraw(sdp, !error)) {
		dip->i_di.di_depth++;
		gfs2_dinode_out(&dip->i_di, dibh->b_data);
		brelse(dibh);
	}

	return error;

 fail:
	kfree(buf);

	return error;
}

/**
 * compare_dents - compare directory entries by hash value
 * @a: first dent
 * @b: second dent
 *
 * When comparing the hash entries of @a to @b:
 *   gt: returns 1
 *   lt: returns -1
 *   eq: returns 0
 */

static int compare_dents(const void *a, const void *b)
{
	struct gfs2_dirent *dent_a, *dent_b;
	uint32_t hash_a, hash_b;
	int ret = 0;

	dent_a = *(struct gfs2_dirent **)a;
1159
	hash_a = be32_to_cpu(dent_a->de_hash);
D
David Teigland 已提交
1160 1161

	dent_b = *(struct gfs2_dirent **)b;
1162
	hash_b = be32_to_cpu(dent_b->de_hash);
D
David Teigland 已提交
1163 1164 1165 1166 1167 1168

	if (hash_a > hash_b)
		ret = 1;
	else if (hash_a < hash_b)
		ret = -1;
	else {
1169 1170
		unsigned int len_a = be16_to_cpu(dent_a->de_name_len);
		unsigned int len_b = be16_to_cpu(dent_b->de_name_len);
D
David Teigland 已提交
1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204

		if (len_a > len_b)
			ret = 1;
		else if (len_a < len_b)
			ret = -1;
		else
			ret = memcmp((char *)(dent_a + 1),
				     (char *)(dent_b + 1),
				     len_a);
	}

	return ret;
}

/**
 * do_filldir_main - read out directory entries
 * @dip: The GFS2 inode
 * @offset: The offset in the file to read from
 * @opaque: opaque data to pass to filldir
 * @filldir: The function to pass entries to
 * @darr: an array of struct gfs2_dirent pointers to read
 * @entries: the number of entries in darr
 * @copied: pointer to int that's non-zero if a entry has been copied out
 *
 * Jump through some hoops to make sure that if there are hash collsions,
 * they are read out at the beginning of a buffer.  We want to minimize
 * the possibility that they will fall into different readdir buffers or
 * that someone will want to seek to that location.
 *
 * Returns: errno, >0 on exception from filldir
 */

static int do_filldir_main(struct gfs2_inode *dip, uint64_t *offset,
			   void *opaque, gfs2_filldir_t filldir,
1205
			   const struct gfs2_dirent **darr, uint32_t entries,
D
David Teigland 已提交
1206 1207
			   int *copied)
{
1208
	const struct gfs2_dirent *dent, *dent_next;
D
David Teigland 已提交
1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248
	struct gfs2_inum inum;
	uint64_t off, off_next;
	unsigned int x, y;
	int run = 0;
	int error = 0;

	sort(darr, entries, sizeof(struct gfs2_dirent *), compare_dents, NULL);

	dent_next = darr[0];
	off_next = be32_to_cpu(dent_next->de_hash);
	off_next = gfs2_disk_hash2offset(off_next);

	for (x = 0, y = 1; x < entries; x++, y++) {
		dent = dent_next;
		off = off_next;

		if (y < entries) {
			dent_next = darr[y];
			off_next = be32_to_cpu(dent_next->de_hash);
			off_next = gfs2_disk_hash2offset(off_next);

			if (off < *offset)
				continue;
			*offset = off;

			if (off_next == off) {
				if (*copied && !run)
					return 1;
				run = 1;
			} else
				run = 0;
		} else {
			if (off < *offset)
				continue;
			*offset = off;
		}

		gfs2_inum_in(&inum, (char *)&dent->de_inum);

		error = filldir(opaque, (char *)(dent + 1),
1249
				be16_to_cpu(dent->de_name_len),
D
David Teigland 已提交
1250
				off, &inum,
1251
				be16_to_cpu(dent->de_type));
D
David Teigland 已提交
1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266
		if (error)
			return 1;

		*copied = 1;
	}

	/* Increment the *offset by one, so the next time we come into the
	   do_filldir fxn, we get the next entry instead of the last one in the
	   current leaf */

	(*offset)++;

	return 0;
}

1267 1268 1269
static int gfs2_dir_read_leaf(struct inode *inode, u64 *offset, void *opaque,
			      gfs2_filldir_t filldir, int *copied,
			      unsigned *depth, u64 leaf_no)
D
David Teigland 已提交
1270
{
1271
	struct gfs2_inode *ip = GFS2_I(inode);
1272 1273 1274 1275 1276 1277 1278 1279 1280 1281
	struct buffer_head *bh;
	struct gfs2_leaf *lf;
	unsigned entries = 0;
	unsigned leaves = 0;
	const struct gfs2_dirent **darr, *dent;
	struct dirent_gather g;
	struct buffer_head **larr;
	int leaf = 0;
	int error, i;
	u64 lfn = leaf_no;
D
David Teigland 已提交
1282 1283

	do {
1284
		error = get_leaf(ip, lfn, &bh);
D
David Teigland 已提交
1285
		if (error)
1286 1287 1288 1289 1290 1291 1292 1293 1294
			goto out;
		lf = (struct gfs2_leaf *)bh->b_data;
		if (leaves == 0)
			*depth = be16_to_cpu(lf->lf_depth);
		entries += be16_to_cpu(lf->lf_entries);
		leaves++;
		lfn = be64_to_cpu(lf->lf_next);
		brelse(bh);
	} while(lfn);
D
David Teigland 已提交
1295 1296 1297 1298

	if (!entries)
		return 0;

1299
	error = -ENOMEM;
1300
	larr = vmalloc((leaves + entries) * sizeof(void*));
1301 1302 1303 1304 1305 1306
	if (!larr)
		goto out;
	darr = (const struct gfs2_dirent **)(larr + leaves);
	g.pdent = darr;
	g.offset = 0;
	lfn = leaf_no;
D
David Teigland 已提交
1307

1308 1309
	do {
		error = get_leaf(ip, lfn, &bh);
D
David Teigland 已提交
1310
		if (error)
1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322
			goto out_kfree;
		lf = (struct gfs2_leaf *)bh->b_data;
		lfn = be64_to_cpu(lf->lf_next);
		if (lf->lf_entries) {
			dent = gfs2_dirent_scan(inode, bh->b_data, bh->b_size,
						gfs2_dirent_gather, NULL, &g);
			error = PTR_ERR(dent);
			if (IS_ERR(dent)) {
				goto out_kfree;
			}
			error = 0;
			larr[leaf++] = bh;
D
David Teigland 已提交
1323
		} else {
1324
			brelse(bh);
D
David Teigland 已提交
1325
		}
1326
	} while(lfn);
D
David Teigland 已提交
1327

1328
	error = do_filldir_main(ip, offset, opaque, filldir, darr,
D
David Teigland 已提交
1329
				entries, copied);
1330 1331 1332
out_kfree:
	for(i = 0; i < leaf; i++)
		brelse(larr[i]);
1333
	vfree(larr);
1334
out:
D
David Teigland 已提交
1335 1336 1337 1338
	return error;
}

/**
1339 1340 1341 1342 1343
 * dir_e_read - Reads the entries from a directory into a filldir buffer
 * @dip: dinode pointer
 * @offset: the hash of the last entry read shifted to the right once
 * @opaque: buffer for the filldir function to fill
 * @filldir: points to the filldir function to use
D
David Teigland 已提交
1344
 *
1345
 * Returns: errno
D
David Teigland 已提交
1346 1347
 */

1348
static int dir_e_read(struct inode *inode, uint64_t *offset, void *opaque,
1349
		      gfs2_filldir_t filldir)
D
David Teigland 已提交
1350
{
1351 1352
	struct gfs2_inode *dip = GFS2_I(inode);
	struct gfs2_sbd *sdp = GFS2_SB(inode);
1353
	uint32_t hsize, len = 0;
1354 1355 1356 1357 1358
	uint32_t ht_offset, lp_offset, ht_offset_cur = -1;
	uint32_t hash, index;
	uint64_t *lp;
	int copied = 0;
	int error = 0;
1359
	unsigned depth;
D
David Teigland 已提交
1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378

	hsize = 1 << dip->i_di.di_depth;
	if (hsize * sizeof(uint64_t) != dip->i_di.di_size) {
		gfs2_consist_inode(dip);
		return -EIO;
	}

	hash = gfs2_dir_offset2hash(*offset);
	index = hash >> (32 - dip->i_di.di_depth);

	lp = kmalloc(sdp->sd_hash_bsize, GFP_KERNEL);
	if (!lp)
		return -ENOMEM;

	while (index < hsize) {
		lp_offset = index & (sdp->sd_hash_ptrs - 1);
		ht_offset = index - lp_offset;

		if (ht_offset_cur != ht_offset) {
1379
			error = gfs2_dir_read_data(dip, (char *)lp,
D
David Teigland 已提交
1380 1381 1382 1383 1384 1385 1386 1387 1388 1389
						ht_offset * sizeof(uint64_t),
						sdp->sd_hash_bsize);
			if (error != sdp->sd_hash_bsize) {
				if (error >= 0)
					error = -EIO;
				goto out;
			}
			ht_offset_cur = ht_offset;
		}

1390 1391 1392
		error = gfs2_dir_read_leaf(inode, offset, opaque, filldir,
					   &copied, &depth,
					   be64_to_cpu(lp[lp_offset]));
D
David Teigland 已提交
1393
		if (error)
1394
			break;
D
David Teigland 已提交
1395

1396
		len = 1 << (dip->i_di.di_depth - depth);
D
David Teigland 已提交
1397 1398 1399
		index = (index & ~(len - 1)) + len;
	}

1400
out:
D
David Teigland 已提交
1401
	kfree(lp);
1402 1403
	if (error > 0)
		error = 0;
D
David Teigland 已提交
1404 1405 1406
	return error;
}

1407 1408
int gfs2_dir_read(struct inode *inode, uint64_t *offset, void *opaque,
		  gfs2_filldir_t filldir)
D
David Teigland 已提交
1409
{
1410
	struct gfs2_inode *dip = GFS2_I(inode);
1411 1412
	struct dirent_gather g;
	const struct gfs2_dirent **darr, *dent;
D
David Teigland 已提交
1413 1414 1415 1416
	struct buffer_head *dibh;
	int copied = 0;
	int error;

1417 1418 1419 1420 1421 1422
	if (!dip->i_di.di_entries)
		return 0;

	if (dip->i_di.di_flags & GFS2_DIF_EXHASH)
		return dir_e_read(inode, offset, opaque, filldir);

D
David Teigland 已提交
1423 1424 1425 1426 1427 1428 1429 1430 1431
	if (!gfs2_is_stuffed(dip)) {
		gfs2_consist_inode(dip);
		return -EIO;
	}

	error = gfs2_meta_inode_buffer(dip, &dibh);
	if (error)
		return error;

1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449
	error = -ENOMEM;
	darr = kmalloc(dip->i_di.di_entries * sizeof(struct gfs2_dirent *),
		       GFP_KERNEL);
	if (darr) {
		g.pdent = darr;
		g.offset = 0;
		dent = gfs2_dirent_scan(inode, dibh->b_data, dibh->b_size,
					gfs2_dirent_gather, NULL, &g);
		if (IS_ERR(dent)) {
			error = PTR_ERR(dent);
			goto out;
		}
		error = do_filldir_main(dip, offset, opaque, filldir, darr,
					dip->i_di.di_entries, &copied);
out:
		kfree(darr);
	}

D
David Teigland 已提交
1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469
	if (error > 0)
		error = 0;

	brelse(dibh);

	return error;
}

/**
 * gfs2_dir_search - Search a directory
 * @dip: The GFS2 inode
 * @filename:
 * @inode:
 *
 * This routine searches a directory for a file or another directory.
 * Assumes a glock is held on dip.
 *
 * Returns: errno
 */

1470
int gfs2_dir_search(struct inode *dir, const struct qstr *name,
D
David Teigland 已提交
1471 1472
		    struct gfs2_inum *inum, unsigned int *type)
{
1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492
	struct buffer_head *bh;
	struct gfs2_dirent *dent;

	dent = gfs2_dirent_search(dir, name, gfs2_dirent_find, &bh);
	if (dent) {
		if (IS_ERR(dent))
			return PTR_ERR(dent);
		if (inum)
			gfs2_inum_in(inum, (char *)&dent->de_inum);
		if (type)
			*type = be16_to_cpu(dent->de_type);
		brelse(bh);
		return 0;
	}
	return -ENOENT;
}

static int dir_new_leaf(struct inode *inode, const struct qstr *name)
{
	struct buffer_head *bh, *obh;
1493
	struct gfs2_inode *ip = GFS2_I(inode);
1494
	struct gfs2_leaf *leaf, *oleaf;
D
David Teigland 已提交
1495
	int error;
1496 1497
	u32 index;
	u64 bn;
D
David Teigland 已提交
1498

1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512
	index = name->hash >> (32 - ip->i_di.di_depth);
	error = get_first_leaf(ip, index, &obh);
	if (error)
		return error;
	do {
		oleaf = (struct gfs2_leaf *)obh->b_data;
		bn = be64_to_cpu(oleaf->lf_next);
		if (!bn)
			break;
		brelse(obh);
		error = get_leaf(ip, bn, &obh);
		if (error)
			return error;
	} while(1);
D
David Teigland 已提交
1513

1514 1515 1516 1517 1518 1519 1520
	gfs2_trans_add_bh(ip->i_gl, obh, 1);

	leaf = new_leaf(inode, &bh, be16_to_cpu(oleaf->lf_depth));
	if (!leaf) {
		brelse(obh);
		return -ENOSPC;
	}
1521
	oleaf->lf_next = cpu_to_be64(bh->b_blocknr);
1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532
	brelse(bh);
	brelse(obh);

	error = gfs2_meta_inode_buffer(ip, &bh);
	if (error)
		return error;
	gfs2_trans_add_bh(ip->i_gl, bh, 1);
	ip->i_di.di_blocks++;
	gfs2_dinode_out(&ip->i_di, bh->b_data);
	brelse(bh);
	return 0;
D
David Teigland 已提交
1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544
}

/**
 * gfs2_dir_add - Add new filename into directory
 * @dip: The GFS2 inode
 * @filename: The new name
 * @inode: The inode number of the entry
 * @type: The type of the entry
 *
 * Returns: 0 on success, error code on failure
 */

1545 1546
int gfs2_dir_add(struct inode *inode, const struct qstr *name,
		 const struct gfs2_inum *inum, unsigned type)
D
David Teigland 已提交
1547
{
1548
	struct gfs2_inode *ip = GFS2_I(inode);
1549 1550 1551
	struct buffer_head *bh;
	struct gfs2_dirent *dent;
	struct gfs2_leaf *leaf;
D
David Teigland 已提交
1552 1553
	int error;

1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587
	while(1) {
		dent = gfs2_dirent_search(inode, name, gfs2_dirent_find_space,
					  &bh);
		if (dent) {
			if (IS_ERR(dent))
				return PTR_ERR(dent);
			dent = gfs2_init_dirent(inode, dent, name, bh);
			gfs2_inum_out(inum, (char *)&dent->de_inum);
			dent->de_type = cpu_to_be16(type);
			if (ip->i_di.di_flags & GFS2_DIF_EXHASH) {
				leaf = (struct gfs2_leaf *)bh->b_data;
				leaf->lf_entries = cpu_to_be16(be16_to_cpu(leaf->lf_entries) + 1);
			}
			brelse(bh);
			error = gfs2_meta_inode_buffer(ip, &bh);
			if (error)
				break;
			gfs2_trans_add_bh(ip->i_gl, bh, 1);
			ip->i_di.di_entries++;
			ip->i_di.di_mtime = ip->i_di.di_ctime = get_seconds();
			gfs2_dinode_out(&ip->i_di, bh->b_data);
			brelse(bh);
			error = 0;
			break;
		}
		if (!(ip->i_di.di_flags & GFS2_DIF_EXHASH)) {
			error = dir_make_exhash(inode);
			if (error)
				break;
			continue;
		}
		error = dir_split_leaf(inode, name);
		if (error == 0)
			continue;
1588
		if (error < 0)
1589 1590 1591 1592 1593 1594
			break;
		if (ip->i_di.di_depth < GFS2_DIR_MAX_DEPTH) {
			error = dir_double_exhash(ip);
			if (error)
				break;
			error = dir_split_leaf(inode, name);
1595
			if (error < 0)
1596
				break;
1597 1598
			if (error == 0)
				continue;
1599 1600 1601 1602 1603 1604 1605
		}
		error = dir_new_leaf(inode, name);
		if (!error)
			continue;
		error = -ENOSPC;
		break;
	}
D
David Teigland 已提交
1606 1607 1608
	return error;
}

1609

D
David Teigland 已提交
1610 1611 1612 1613 1614 1615 1616 1617
/**
 * gfs2_dir_del - Delete a directory entry
 * @dip: The GFS2 inode
 * @filename: The filename
 *
 * Returns: 0 on success, error code on failure
 */

1618
int gfs2_dir_del(struct gfs2_inode *dip, const struct qstr *name)
D
David Teigland 已提交
1619
{
1620 1621
	struct gfs2_dirent *dent, *prev = NULL;
	struct buffer_head *bh;
D
David Teigland 已提交
1622 1623
	int error;

1624 1625
	/* Returns _either_ the entry (if its first in block) or the
	   previous entry otherwise */
1626
	dent = gfs2_dirent_search(&dip->i_inode, name, gfs2_dirent_prev, &bh);
1627 1628 1629 1630 1631 1632 1633 1634 1635
	if (!dent) {
		gfs2_consist_inode(dip);
		return -EIO;
	}
	if (IS_ERR(dent)) {
		gfs2_consist_inode(dip);
		return PTR_ERR(dent);
	}
	/* If not first in block, adjust pointers accordingly */
1636
	if (gfs2_dirent_find(dent, name, NULL) == 0) {
1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648
		prev = dent;
		dent = (struct gfs2_dirent *)((char *)dent + be16_to_cpu(prev->de_rec_len));
	}

	dirent_del(dip, bh, prev, dent);
	if (dip->i_di.di_flags & GFS2_DIF_EXHASH) {
		struct gfs2_leaf *leaf = (struct gfs2_leaf *)bh->b_data;
		u16 entries = be16_to_cpu(leaf->lf_entries);
		if (!entries)
			gfs2_consist_inode(dip);
		leaf->lf_entries = cpu_to_be16(--entries);
	}
1649
	brelse(bh);
1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661

	error = gfs2_meta_inode_buffer(dip, &bh);
	if (error)
		return error;

	if (!dip->i_di.di_entries)
		gfs2_consist_inode(dip);
	gfs2_trans_add_bh(dip->i_gl, bh, 1);
	dip->i_di.di_entries--;
	dip->i_di.di_mtime = dip->i_di.di_ctime = get_seconds();
	gfs2_dinode_out(&dip->i_di, bh->b_data);
	brelse(bh);
1662
	mark_inode_dirty(&dip->i_inode);
D
David Teigland 已提交
1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679

	return error;
}

/**
 * gfs2_dir_mvino - Change inode number of directory entry
 * @dip: The GFS2 inode
 * @filename:
 * @new_inode:
 *
 * This routine changes the inode number of a directory entry.  It's used
 * by rename to change ".." when a directory is moved.
 * Assumes a glock is held on dvp.
 *
 * Returns: errno
 */

1680
int gfs2_dir_mvino(struct gfs2_inode *dip, const struct qstr *filename,
D
David Teigland 已提交
1681 1682
		   struct gfs2_inum *inum, unsigned int new_type)
{
1683 1684
	struct buffer_head *bh;
	struct gfs2_dirent *dent;
D
David Teigland 已提交
1685 1686
	int error;

1687
	dent = gfs2_dirent_search(&dip->i_inode, filename, gfs2_dirent_find, &bh);
1688 1689 1690 1691 1692 1693
	if (!dent) {
		gfs2_consist_inode(dip);
		return -EIO;
	}
	if (IS_ERR(dent))
		return PTR_ERR(dent);
D
David Teigland 已提交
1694

1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710
	gfs2_trans_add_bh(dip->i_gl, bh, 1);
	gfs2_inum_out(inum, (char *)&dent->de_inum);
	dent->de_type = cpu_to_be16(new_type);

	if (dip->i_di.di_flags & GFS2_DIF_EXHASH) {
		brelse(bh);
		error = gfs2_meta_inode_buffer(dip, &bh);
		if (error)
			return error;
		gfs2_trans_add_bh(dip->i_gl, bh, 1);
	}

	dip->i_di.di_mtime = dip->i_di.di_ctime = get_seconds();
	gfs2_dinode_out(&dip->i_di, bh->b_data);
	brelse(bh);
	return 0;
D
David Teigland 已提交
1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723
}

/**
 * foreach_leaf - call a function for each leaf in a directory
 * @dip: the directory
 * @lc: the function to call for each each
 * @data: private data to pass to it
 *
 * Returns: errno
 */

static int foreach_leaf(struct gfs2_inode *dip, leaf_call_t lc, void *data)
{
1724
	struct gfs2_sbd *sdp = GFS2_SB(&dip->i_inode);
D
David Teigland 已提交
1725
	struct buffer_head *bh;
1726
	struct gfs2_leaf *leaf;
D
David Teigland 已提交
1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748
	uint32_t hsize, len;
	uint32_t ht_offset, lp_offset, ht_offset_cur = -1;
	uint32_t index = 0;
	uint64_t *lp;
	uint64_t leaf_no;
	int error = 0;

	hsize = 1 << dip->i_di.di_depth;
	if (hsize * sizeof(uint64_t) != dip->i_di.di_size) {
		gfs2_consist_inode(dip);
		return -EIO;
	}

	lp = kmalloc(sdp->sd_hash_bsize, GFP_KERNEL);
	if (!lp)
		return -ENOMEM;

	while (index < hsize) {
		lp_offset = index & (sdp->sd_hash_ptrs - 1);
		ht_offset = index - lp_offset;

		if (ht_offset_cur != ht_offset) {
1749
			error = gfs2_dir_read_data(dip, (char *)lp,
D
David Teigland 已提交
1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764
						ht_offset * sizeof(uint64_t),
						sdp->sd_hash_bsize);
			if (error != sdp->sd_hash_bsize) {
				if (error >= 0)
					error = -EIO;
				goto out;
			}
			ht_offset_cur = ht_offset;
		}

		leaf_no = be64_to_cpu(lp[lp_offset]);
		if (leaf_no) {
			error = get_leaf(dip, leaf_no, &bh);
			if (error)
				goto out;
1765
			leaf = (struct gfs2_leaf *)bh->b_data;
D
David Teigland 已提交
1766 1767
			brelse(bh);

1768
			len = 1 << (dip->i_di.di_depth - be16_to_cpu(leaf->lf_depth));
D
David Teigland 已提交
1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803

			error = lc(dip, index, len, leaf_no, data);
			if (error)
				goto out;

			index = (index & ~(len - 1)) + len;
		} else
			index++;
	}

	if (index != hsize) {
		gfs2_consist_inode(dip);
		error = -EIO;
	}

 out:
	kfree(lp);

	return error;
}

/**
 * leaf_dealloc - Deallocate a directory leaf
 * @dip: the directory
 * @index: the hash table offset in the directory
 * @len: the number of pointers to this leaf
 * @leaf_no: the leaf number
 * @data: not used
 *
 * Returns: errno
 */

static int leaf_dealloc(struct gfs2_inode *dip, uint32_t index, uint32_t len,
			uint64_t leaf_no, void *data)
{
1804
	struct gfs2_sbd *sdp = GFS2_SB(&dip->i_inode);
1805
	struct gfs2_leaf *tmp_leaf;
D
David Teigland 已提交
1806 1807
	struct gfs2_rgrp_list rlist;
	struct buffer_head *bh, *dibh;
1808
	uint64_t blk, nblk;
D
David Teigland 已提交
1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831
	unsigned int rg_blocks = 0, l_blocks = 0;
	char *ht;
	unsigned int x, size = len * sizeof(uint64_t);
	int error;

	memset(&rlist, 0, sizeof(struct gfs2_rgrp_list));

	ht = kzalloc(size, GFP_KERNEL);
	if (!ht)
		return -ENOMEM;

	gfs2_alloc_get(dip);

	error = gfs2_quota_hold(dip, NO_QUOTA_CHANGE, NO_QUOTA_CHANGE);
	if (error)
		goto out;

	error = gfs2_rindex_hold(sdp, &dip->i_alloc.al_ri_gh);
	if (error)
		goto out_qs;

	/*  Count the number of leaves  */

1832
	for (blk = leaf_no; blk; blk = nblk) {
D
David Teigland 已提交
1833 1834 1835
		error = get_leaf(dip, blk, &bh);
		if (error)
			goto out_rlist;
1836 1837
		tmp_leaf = (struct gfs2_leaf *)bh->b_data;
		nblk = be64_to_cpu(tmp_leaf->lf_next);
D
David Teigland 已提交
1838 1839 1840 1841 1842 1843 1844 1845 1846 1847
		brelse(bh);

		gfs2_rlist_add(sdp, &rlist, blk);
		l_blocks++;
	}

	gfs2_rlist_alloc(&rlist, LM_ST_EXCLUSIVE, 0);

	for (x = 0; x < rlist.rl_rgrps; x++) {
		struct gfs2_rgrpd *rgd;
1848
		rgd = rlist.rl_ghs[x].gh_gl->gl_object;
D
David Teigland 已提交
1849 1850 1851 1852 1853 1854 1855 1856
		rg_blocks += rgd->rd_ri.ri_length;
	}

	error = gfs2_glock_nq_m(rlist.rl_rgrps, rlist.rl_ghs);
	if (error)
		goto out_rlist;

	error = gfs2_trans_begin(sdp,
1857
			rg_blocks + (DIV_ROUND_UP(size, sdp->sd_jbsize) + 1) +
D
David Teigland 已提交
1858 1859 1860 1861
			RES_DINODE + RES_STATFS + RES_QUOTA, l_blocks);
	if (error)
		goto out_rg_gunlock;

1862
	for (blk = leaf_no; blk; blk = nblk) {
D
David Teigland 已提交
1863 1864 1865
		error = get_leaf(dip, blk, &bh);
		if (error)
			goto out_end_trans;
1866 1867
		tmp_leaf = (struct gfs2_leaf *)bh->b_data;
		nblk = be64_to_cpu(tmp_leaf->lf_next);
D
David Teigland 已提交
1868 1869 1870 1871 1872 1873 1874 1875 1876
		brelse(bh);

		gfs2_free_meta(dip, blk, 1);

		if (!dip->i_di.di_blocks)
			gfs2_consist_inode(dip);
		dip->i_di.di_blocks--;
	}

1877
	error = gfs2_dir_write_data(dip, ht, index * sizeof(uint64_t), size);
D
David Teigland 已提交
1878 1879 1880 1881 1882 1883 1884 1885 1886 1887
	if (error != size) {
		if (error >= 0)
			error = -EIO;
		goto out_end_trans;
	}

	error = gfs2_meta_inode_buffer(dip, &dibh);
	if (error)
		goto out_end_trans;

1888
	gfs2_trans_add_bh(dip->i_gl, dibh, 1);
D
David Teigland 已提交
1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923
	gfs2_dinode_out(&dip->i_di, dibh->b_data);
	brelse(dibh);

 out_end_trans:
	gfs2_trans_end(sdp);

 out_rg_gunlock:
	gfs2_glock_dq_m(rlist.rl_rgrps, rlist.rl_ghs);

 out_rlist:
	gfs2_rlist_free(&rlist);
	gfs2_glock_dq_uninit(&dip->i_alloc.al_ri_gh);

 out_qs:
	gfs2_quota_unhold(dip);

 out:
	gfs2_alloc_put(dip);
	kfree(ht);

	return error;
}

/**
 * gfs2_dir_exhash_dealloc - free all the leaf blocks in a directory
 * @dip: the directory
 *
 * Dealloc all on-disk directory leaves to FREEMETA state
 * Change on-disk inode type to "regular file"
 *
 * Returns: errno
 */

int gfs2_dir_exhash_dealloc(struct gfs2_inode *dip)
{
1924
	struct gfs2_sbd *sdp = GFS2_SB(&dip->i_inode);
D
David Teigland 已提交
1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941
	struct buffer_head *bh;
	int error;

	/* Dealloc on-disk leaves to FREEMETA state */
	error = foreach_leaf(dip, leaf_dealloc, NULL);
	if (error)
		return error;

	/* Make this a regular file in case we crash.
	   (We don't want to free these blocks a second time.)  */

	error = gfs2_trans_begin(sdp, RES_DINODE, 0);
	if (error)
		return error;

	error = gfs2_meta_inode_buffer(dip, &bh);
	if (!error) {
1942
		gfs2_trans_add_bh(dip->i_gl, bh, 1);
1943 1944
		((struct gfs2_dinode *)bh->b_data)->di_mode =
						cpu_to_be32(S_IFREG);
D
David Teigland 已提交
1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957
		brelse(bh);
	}

	gfs2_trans_end(sdp);

	return error;
}

/**
 * gfs2_diradd_alloc_required - find if adding entry will require an allocation
 * @ip: the file being written to
 * @filname: the filename that's going to be added
 *
1958
 * Returns: 1 if alloc required, 0 if not, -ve on error
D
David Teigland 已提交
1959 1960
 */

1961
int gfs2_diradd_alloc_required(struct inode *inode, const struct qstr *name)
D
David Teigland 已提交
1962
{
1963 1964
	struct gfs2_dirent *dent;
	struct buffer_head *bh;
D
David Teigland 已提交
1965

1966
	dent = gfs2_dirent_search(inode, name, gfs2_dirent_find_space, &bh);
1967
	if (!dent) {
1968
		return 1;
1969
	}
1970 1971 1972 1973
	if (IS_ERR(dent))
		return PTR_ERR(dent);
	brelse(bh);
	return 0;
D
David Teigland 已提交
1974 1975
}