dir.c 115.6 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
/* -*- mode: c; c-basic-offset: 8; -*-
 * vim: noexpandtab sw=8 ts=8 sts=0:
 *
 * dir.c
 *
 * Creates, reads, walks and deletes directory-nodes
 *
 * Copyright (C) 2002, 2004 Oracle.  All rights reserved.
 *
 *  Portions of this code from linux/fs/ext3/dir.c
 *
 *  Copyright (C) 1992, 1993, 1994, 1995
 *  Remy Card (card@masi.ibp.fr)
 *  Laboratoire MASI - Institut Blaise pascal
 *  Universite Pierre et Marie Curie (Paris VI)
 *
 *   from
 *
 *   linux/fs/minix/dir.c
 *
 *   Copyright (C) 1991, 1992 Linux Torvalds
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License as published by the Free Software Foundation; either
 * version 2 of the License, or (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

#include <linux/fs.h>
#include <linux/types.h>
#include <linux/slab.h>
#include <linux/highmem.h>
43
#include <linux/quotaops.h>
44
#include <linux/sort.h>
45 46 47 48 49 50

#include <cluster/masklog.h>

#include "ocfs2.h"

#include "alloc.h"
51
#include "blockcheck.h"
52 53 54 55 56 57 58 59
#include "dir.h"
#include "dlmglue.h"
#include "extent_map.h"
#include "file.h"
#include "inode.h"
#include "journal.h"
#include "namei.h"
#include "suballoc.h"
60
#include "super.h"
61
#include "sysfile.h"
62
#include "uptodate.h"
63
#include "ocfs2_trace.h"
64 65 66

#include "buffer_head_io.h"

67 68 69 70
#define NAMEI_RA_CHUNKS  2
#define NAMEI_RA_BLOCKS  4
#define NAMEI_RA_SIZE        (NAMEI_RA_CHUNKS * NAMEI_RA_BLOCKS)

71 72 73 74
static unsigned char ocfs2_filetype_table[] = {
	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
};

75 76 77 78 79 80 81
static int ocfs2_do_extend_dir(struct super_block *sb,
			       handle_t *handle,
			       struct inode *dir,
			       struct buffer_head *parent_fe_bh,
			       struct ocfs2_alloc_context *data_ac,
			       struct ocfs2_alloc_context *meta_ac,
			       struct buffer_head **new_bh);
82
static int ocfs2_dir_indexed(struct inode *inode);
83

84 85 86 87
/*
 * These are distinct checks because future versions of the file system will
 * want to have a trailing dirent structure independent of indexing.
 */
88
static int ocfs2_supports_dir_trailer(struct inode *dir)
89
{
90 91
	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);

92 93 94
	if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL)
		return 0;

95
	return ocfs2_meta_ecc(osb) || ocfs2_dir_indexed(dir);
96 97
}

98 99 100 101 102 103 104 105 106
/*
 * "new' here refers to the point at which we're creating a new
 * directory via "mkdir()", but also when we're expanding an inline
 * directory. In either case, we don't yet have the indexing bit set
 * on the directory, so the standard checks will fail in when metaecc
 * is turned off. Only directory-initialization type functions should
 * use this then. Everything else wants ocfs2_supports_dir_trailer()
 */
static int ocfs2_new_dir_wants_trailer(struct inode *dir)
107
{
108 109 110 111
	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);

	return ocfs2_meta_ecc(osb) ||
		ocfs2_supports_indexed_dirs(osb);
112 113 114 115 116 117 118 119 120
}

static inline unsigned int ocfs2_dir_trailer_blk_off(struct super_block *sb)
{
	return sb->s_blocksize - sizeof(struct ocfs2_dir_block_trailer);
}

#define ocfs2_trailer_from_bh(_bh, _sb) ((struct ocfs2_dir_block_trailer *) ((_bh)->b_data + ocfs2_dir_trailer_blk_off((_sb))))

121 122 123 124 125 126 127 128 129 130 131
/* XXX ocfs2_block_dqtrailer() is similar but not quite - can we make
 * them more consistent? */
struct ocfs2_dir_block_trailer *ocfs2_dir_trailer_from_size(int blocksize,
							    void *data)
{
	char *p = data;

	p += blocksize - sizeof(struct ocfs2_dir_block_trailer);
	return (struct ocfs2_dir_block_trailer *)p;
}

132 133 134 135 136 137 138 139 140 141 142
/*
 * XXX: This is executed once on every dirent. We should consider optimizing
 * it.
 */
static int ocfs2_skip_dir_trailer(struct inode *dir,
				  struct ocfs2_dir_entry *de,
				  unsigned long offset,
				  unsigned long blklen)
{
	unsigned long toff = blklen - sizeof(struct ocfs2_dir_block_trailer);

143
	if (!ocfs2_supports_dir_trailer(dir))
144 145 146 147 148 149 150 151 152
		return 0;

	if (offset != toff)
		return 0;

	return 1;
}

static void ocfs2_init_dir_trailer(struct inode *inode,
153
				   struct buffer_head *bh, u16 rec_len)
154 155 156 157 158 159 160 161 162
{
	struct ocfs2_dir_block_trailer *trailer;

	trailer = ocfs2_trailer_from_bh(bh, inode->i_sb);
	strcpy(trailer->db_signature, OCFS2_DIR_TRAILER_SIGNATURE);
	trailer->db_compat_rec_len =
			cpu_to_le16(sizeof(struct ocfs2_dir_block_trailer));
	trailer->db_parent_dinode = cpu_to_le64(OCFS2_I(inode)->ip_blkno);
	trailer->db_blkno = cpu_to_le64(bh->b_blocknr);
163 164 165 166 167 168 169 170 171 172 173 174 175 176 177
	trailer->db_free_rec_len = cpu_to_le16(rec_len);
}
/*
 * Link an unindexed block with a dir trailer structure into the index free
 * list. This function will modify dirdata_bh, but assumes you've already
 * passed it to the journal.
 */
static int ocfs2_dx_dir_link_trailer(struct inode *dir, handle_t *handle,
				     struct buffer_head *dx_root_bh,
				     struct buffer_head *dirdata_bh)
{
	int ret;
	struct ocfs2_dx_root_block *dx_root;
	struct ocfs2_dir_block_trailer *trailer;

178
	ret = ocfs2_journal_access_dr(handle, INODE_CACHE(dir), dx_root_bh,
179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}
	trailer = ocfs2_trailer_from_bh(dirdata_bh, dir->i_sb);
	dx_root = (struct ocfs2_dx_root_block *)dx_root_bh->b_data;

	trailer->db_free_next = dx_root->dr_free_blk;
	dx_root->dr_free_blk = cpu_to_le64(dirdata_bh->b_blocknr);

	ocfs2_journal_dirty(handle, dx_root_bh);

out:
	return ret;
}

static int ocfs2_free_list_at_root(struct ocfs2_dir_lookup_result *res)
{
	return res->dl_prev_leaf_bh == NULL;
199 200
}

201 202
void ocfs2_free_dir_lookup_result(struct ocfs2_dir_lookup_result *res)
{
203
	brelse(res->dl_dx_root_bh);
204
	brelse(res->dl_leaf_bh);
205
	brelse(res->dl_dx_leaf_bh);
206
	brelse(res->dl_prev_leaf_bh);
207 208 209 210 211 212 213 214 215
}

static int ocfs2_dir_indexed(struct inode *inode)
{
	if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INDEXED_DIR_FL)
		return 1;
	return 0;
}

216 217 218 219 220
static inline int ocfs2_dx_root_inline(struct ocfs2_dx_root_block *dx_root)
{
	return dx_root->dr_flags & OCFS2_DX_FLAG_INLINE;
}

221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309
/*
 * Hashing code adapted from ext3
 */
#define DELTA 0x9E3779B9

static void TEA_transform(__u32 buf[4], __u32 const in[])
{
	__u32	sum = 0;
	__u32	b0 = buf[0], b1 = buf[1];
	__u32	a = in[0], b = in[1], c = in[2], d = in[3];
	int	n = 16;

	do {
		sum += DELTA;
		b0 += ((b1 << 4)+a) ^ (b1+sum) ^ ((b1 >> 5)+b);
		b1 += ((b0 << 4)+c) ^ (b0+sum) ^ ((b0 >> 5)+d);
	} while (--n);

	buf[0] += b0;
	buf[1] += b1;
}

static void str2hashbuf(const char *msg, int len, __u32 *buf, int num)
{
	__u32	pad, val;
	int	i;

	pad = (__u32)len | ((__u32)len << 8);
	pad |= pad << 16;

	val = pad;
	if (len > num*4)
		len = num * 4;
	for (i = 0; i < len; i++) {
		if ((i % 4) == 0)
			val = pad;
		val = msg[i] + (val << 8);
		if ((i % 4) == 3) {
			*buf++ = val;
			val = pad;
			num--;
		}
	}
	if (--num >= 0)
		*buf++ = val;
	while (--num >= 0)
		*buf++ = pad;
}

static void ocfs2_dx_dir_name_hash(struct inode *dir, const char *name, int len,
				   struct ocfs2_dx_hinfo *hinfo)
{
	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
	const char	*p;
	__u32		in[8], buf[4];

	/*
	 * XXX: Is this really necessary, if the index is never looked
	 * at by readdir? Is a hash value of '0' a bad idea?
	 */
	if ((len == 1 && !strncmp(".", name, 1)) ||
	    (len == 2 && !strncmp("..", name, 2))) {
		buf[0] = buf[1] = 0;
		goto out;
	}

#ifdef OCFS2_DEBUG_DX_DIRS
	/*
	 * This makes it very easy to debug indexing problems. We
	 * should never allow this to be selected without hand editing
	 * this file though.
	 */
	buf[0] = buf[1] = len;
	goto out;
#endif

	memcpy(buf, osb->osb_dx_seed, sizeof(buf));

	p = name;
	while (len > 0) {
		str2hashbuf(p, len, in, 4);
		TEA_transform(buf, in);
		len -= 16;
		p += 16;
	}

out:
	hinfo->major_hash = buf[0];
	hinfo->minor_hash = buf[1];
310 311
}

312 313 314 315
/*
 * bh passed here can be an inode block or a dir data block, depending
 * on the inode inline data flag.
 */
M
Mark Fasheh 已提交
316 317 318 319
static int ocfs2_check_dir_entry(struct inode * dir,
				 struct ocfs2_dir_entry * de,
				 struct buffer_head * bh,
				 unsigned long offset)
320 321 322 323
{
	const char *error_msg = NULL;
	const int rlen = le16_to_cpu(de->rec_len);

324
	if (unlikely(rlen < OCFS2_DIR_REC_LEN(1)))
325
		error_msg = "rec_len is smaller than minimal";
326
	else if (unlikely(rlen % 4 != 0))
327
		error_msg = "rec_len % 4 != 0";
328
	else if (unlikely(rlen < OCFS2_DIR_REC_LEN(de->name_len)))
329
		error_msg = "rec_len is too small for name_len";
330 331
	else if (unlikely(
		 ((char *) de - bh->b_data) + rlen > dir->i_sb->s_blocksize))
332 333
		error_msg = "directory entry across blocks";

334
	if (unlikely(error_msg != NULL))
335 336 337 338 339
		mlog(ML_ERROR, "bad entry in directory #%llu: %s - "
		     "offset=%lu, inode=%llu, rec_len=%d, name_len=%d\n",
		     (unsigned long long)OCFS2_I(dir)->ip_blkno, error_msg,
		     offset, (unsigned long long)le64_to_cpu(de->inode), rlen,
		     de->name_len);
340

341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357
	return error_msg == NULL ? 1 : 0;
}

static inline int ocfs2_match(int len,
			      const char * const name,
			      struct ocfs2_dir_entry *de)
{
	if (len != de->name_len)
		return 0;
	if (!de->inode)
		return 0;
	return !memcmp(name, de->name, len);
}

/*
 * Returns 0 if not found, -1 on failure, and 1 on success
 */
358
static inline int ocfs2_search_dirblock(struct buffer_head *bh,
359 360 361
					struct inode *dir,
					const char *name, int namelen,
					unsigned long offset,
362 363
					char *first_de,
					unsigned int bytes,
364 365 366 367 368 369 370
					struct ocfs2_dir_entry **res_dir)
{
	struct ocfs2_dir_entry *de;
	char *dlimit, *de_buf;
	int de_len;
	int ret = 0;

371 372
	de_buf = first_de;
	dlimit = de_buf + bytes;
373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403

	while (de_buf < dlimit) {
		/* this code is executed quadratically often */
		/* do minimal checking `by hand' */

		de = (struct ocfs2_dir_entry *) de_buf;

		if (de_buf + namelen <= dlimit &&
		    ocfs2_match(namelen, name, de)) {
			/* found a match - just to be sure, do a full check */
			if (!ocfs2_check_dir_entry(dir, de, bh, offset)) {
				ret = -1;
				goto bail;
			}
			*res_dir = de;
			ret = 1;
			goto bail;
		}

		/* prevent looping on a bad block */
		de_len = le16_to_cpu(de->rec_len);
		if (de_len <= 0) {
			ret = -1;
			goto bail;
		}

		de_buf += de_len;
		offset += de_len;
	}

bail:
404
	trace_ocfs2_search_dirblock(ret);
405 406 407
	return ret;
}

408 409 410 411 412 413 414 415 416 417
static struct buffer_head *ocfs2_find_entry_id(const char *name,
					       int namelen,
					       struct inode *dir,
					       struct ocfs2_dir_entry **res_dir)
{
	int ret, found;
	struct buffer_head *di_bh = NULL;
	struct ocfs2_dinode *di;
	struct ocfs2_inline_data *data;

418
	ret = ocfs2_read_inode_block(dir, &di_bh);
419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	di = (struct ocfs2_dinode *)di_bh->b_data;
	data = &di->id2.i_data;

	found = ocfs2_search_dirblock(di_bh, dir, name, namelen, 0,
				      data->id_data, i_size_read(dir), res_dir);
	if (found == 1)
		return di_bh;

	brelse(di_bh);
out:
	return NULL;
}

437 438 439
static int ocfs2_validate_dir_block(struct super_block *sb,
				    struct buffer_head *bh)
{
440 441 442 443 444
	int rc;
	struct ocfs2_dir_block_trailer *trailer =
		ocfs2_trailer_from_bh(bh, sb);


445
	/*
446
	 * We don't validate dirents here, that's handled
447 448
	 * in-place when the code walks them.
	 */
449
	trace_ocfs2_validate_dir_block((unsigned long long)bh->b_blocknr);
450

451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467
	BUG_ON(!buffer_uptodate(bh));

	/*
	 * If the ecc fails, we return the error but otherwise
	 * leave the filesystem running.  We know any error is
	 * local to this block.
	 *
	 * Note that we are safe to call this even if the directory
	 * doesn't have a trailer.  Filesystems without metaecc will do
	 * nothing, and filesystems with it will have one.
	 */
	rc = ocfs2_validate_meta_ecc(sb, bh->b_data, &trailer->db_check);
	if (rc)
		mlog(ML_ERROR, "Checksum failed for dinode %llu\n",
		     (unsigned long long)bh->b_blocknr);

	return rc;
468 469
}

470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515
/*
 * Validate a directory trailer.
 *
 * We check the trailer here rather than in ocfs2_validate_dir_block()
 * because that function doesn't have the inode to test.
 */
static int ocfs2_check_dir_trailer(struct inode *dir, struct buffer_head *bh)
{
	int rc = 0;
	struct ocfs2_dir_block_trailer *trailer;

	trailer = ocfs2_trailer_from_bh(bh, dir->i_sb);
	if (!OCFS2_IS_VALID_DIR_TRAILER(trailer)) {
		rc = -EINVAL;
		ocfs2_error(dir->i_sb,
			    "Invalid dirblock #%llu: "
			    "signature = %.*s\n",
			    (unsigned long long)bh->b_blocknr, 7,
			    trailer->db_signature);
		goto out;
	}
	if (le64_to_cpu(trailer->db_blkno) != bh->b_blocknr) {
		rc = -EINVAL;
		ocfs2_error(dir->i_sb,
			    "Directory block #%llu has an invalid "
			    "db_blkno of %llu",
			    (unsigned long long)bh->b_blocknr,
			    (unsigned long long)le64_to_cpu(trailer->db_blkno));
		goto out;
	}
	if (le64_to_cpu(trailer->db_parent_dinode) !=
	    OCFS2_I(dir)->ip_blkno) {
		rc = -EINVAL;
		ocfs2_error(dir->i_sb,
			    "Directory block #%llu on dinode "
			    "#%llu has an invalid parent_dinode "
			    "of %llu",
			    (unsigned long long)bh->b_blocknr,
			    (unsigned long long)OCFS2_I(dir)->ip_blkno,
			    (unsigned long long)le64_to_cpu(trailer->db_blkno));
		goto out;
	}
out:
	return rc;
}

516 517 518 519 520 521 522 523 524 525 526 527
/*
 * This function forces all errors to -EIO for consistency with its
 * predecessor, ocfs2_bread().  We haven't audited what returning the
 * real error codes would do to callers.  We log the real codes with
 * mlog_errno() before we squash them.
 */
static int ocfs2_read_dir_block(struct inode *inode, u64 v_block,
				struct buffer_head **bh, int flags)
{
	int rc = 0;
	struct buffer_head *tmp = *bh;

528 529
	rc = ocfs2_read_virt_blocks(inode, v_block, 1, &tmp, flags,
				    ocfs2_validate_dir_block);
530
	if (rc) {
531
		mlog_errno(rc);
532 533 534 535
		goto out;
	}

	if (!(flags & OCFS2_BH_READAHEAD) &&
536
	    ocfs2_supports_dir_trailer(inode)) {
537 538 539 540 541
		rc = ocfs2_check_dir_trailer(inode, tmp);
		if (rc) {
			if (!*bh)
				brelse(tmp);
			mlog_errno(rc);
542 543 544
			goto out;
		}
	}
545

546
	/* If ocfs2_read_virt_blocks() got us a new bh, pass it up. */
547
	if (!*bh)
548 549
		*bh = tmp;

550
out:
551 552 553
	return rc ? -EIO : 0;
}

554 555 556 557 558 559 560 561 562 563 564
/*
 * Read the block at 'phys' which belongs to this directory
 * inode. This function does no virtual->physical block translation -
 * what's passed in is assumed to be a valid directory block.
 */
static int ocfs2_read_dir_block_direct(struct inode *dir, u64 phys,
				       struct buffer_head **bh)
{
	int ret;
	struct buffer_head *tmp = *bh;

565 566
	ret = ocfs2_read_block(INODE_CACHE(dir), phys, &tmp,
			       ocfs2_validate_dir_block);
567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	if (ocfs2_supports_dir_trailer(dir)) {
		ret = ocfs2_check_dir_trailer(dir, tmp);
		if (ret) {
			if (!*bh)
				brelse(tmp);
			mlog_errno(ret);
			goto out;
		}
	}

	if (!ret && !*bh)
		*bh = tmp;
out:
	return ret;
}

static int ocfs2_validate_dx_root(struct super_block *sb,
				  struct buffer_head *bh)
{
	int ret;
	struct ocfs2_dx_root_block *dx_root;

	BUG_ON(!buffer_uptodate(bh));

	dx_root = (struct ocfs2_dx_root_block *) bh->b_data;

	ret = ocfs2_validate_meta_ecc(sb, bh->b_data, &dx_root->dr_check);
	if (ret) {
		mlog(ML_ERROR,
		     "Checksum failed for dir index root block %llu\n",
		     (unsigned long long)bh->b_blocknr);
		return ret;
	}

	if (!OCFS2_IS_VALID_DX_ROOT(dx_root)) {
		ocfs2_error(sb,
			    "Dir Index Root # %llu has bad signature %.*s",
			    (unsigned long long)le64_to_cpu(dx_root->dr_blkno),
			    7, dx_root->dr_signature);
		return -EINVAL;
	}

	return 0;
}

static int ocfs2_read_dx_root(struct inode *dir, struct ocfs2_dinode *di,
			      struct buffer_head **dx_root_bh)
{
	int ret;
	u64 blkno = le64_to_cpu(di->i_dx_root);
	struct buffer_head *tmp = *dx_root_bh;

624 625
	ret = ocfs2_read_block(INODE_CACHE(dir), blkno, &tmp,
			       ocfs2_validate_dx_root);
626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664

	/* If ocfs2_read_block() got us a new bh, pass it up. */
	if (!ret && !*dx_root_bh)
		*dx_root_bh = tmp;

	return ret;
}

static int ocfs2_validate_dx_leaf(struct super_block *sb,
				  struct buffer_head *bh)
{
	int ret;
	struct ocfs2_dx_leaf *dx_leaf = (struct ocfs2_dx_leaf *)bh->b_data;

	BUG_ON(!buffer_uptodate(bh));

	ret = ocfs2_validate_meta_ecc(sb, bh->b_data, &dx_leaf->dl_check);
	if (ret) {
		mlog(ML_ERROR,
		     "Checksum failed for dir index leaf block %llu\n",
		     (unsigned long long)bh->b_blocknr);
		return ret;
	}

	if (!OCFS2_IS_VALID_DX_LEAF(dx_leaf)) {
		ocfs2_error(sb, "Dir Index Leaf has bad signature %.*s",
			    7, dx_leaf->dl_signature);
		return -EROFS;
	}

	return 0;
}

static int ocfs2_read_dx_leaf(struct inode *dir, u64 blkno,
			      struct buffer_head **dx_leaf_bh)
{
	int ret;
	struct buffer_head *tmp = *dx_leaf_bh;

665 666
	ret = ocfs2_read_block(INODE_CACHE(dir), blkno, &tmp,
			       ocfs2_validate_dx_leaf);
667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683

	/* If ocfs2_read_block() got us a new bh, pass it up. */
	if (!ret && !*dx_leaf_bh)
		*dx_leaf_bh = tmp;

	return ret;
}

/*
 * Read a series of dx_leaf blocks. This expects all buffer_head
 * pointers to be NULL on function entry.
 */
static int ocfs2_read_dx_leaves(struct inode *dir, u64 start, int num,
				struct buffer_head **dx_leaf_bhs)
{
	int ret;

684
	ret = ocfs2_read_blocks(INODE_CACHE(dir), start, num, dx_leaf_bhs, 0,
685 686 687 688 689 690 691
				ocfs2_validate_dx_leaf);
	if (ret)
		mlog_errno(ret);

	return ret;
}

692 693 694
static struct buffer_head *ocfs2_find_entry_el(const char *name, int namelen,
					       struct inode *dir,
					       struct ocfs2_dir_entry **res_dir)
695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735
{
	struct super_block *sb;
	struct buffer_head *bh_use[NAMEI_RA_SIZE];
	struct buffer_head *bh, *ret = NULL;
	unsigned long start, block, b;
	int ra_max = 0;		/* Number of bh's in the readahead
				   buffer, bh_use[] */
	int ra_ptr = 0;		/* Current index into readahead
				   buffer */
	int num = 0;
	int nblocks, i, err;

	sb = dir->i_sb;

	nblocks = i_size_read(dir) >> sb->s_blocksize_bits;
	start = OCFS2_I(dir)->ip_dir_start_lookup;
	if (start >= nblocks)
		start = 0;
	block = start;

restart:
	do {
		/*
		 * We deal with the read-ahead logic here.
		 */
		if (ra_ptr >= ra_max) {
			/* Refill the readahead buffer */
			ra_ptr = 0;
			b = block;
			for (ra_max = 0; ra_max < NAMEI_RA_SIZE; ra_max++) {
				/*
				 * Terminate if we reach the end of the
				 * directory and must wrap, or if our
				 * search has finished at this block.
				 */
				if (b >= nblocks || (num && block == start)) {
					bh_use[ra_max] = NULL;
					break;
				}
				num++;

736 737 738
				bh = NULL;
				err = ocfs2_read_dir_block(dir, b++, &bh,
							   OCFS2_BH_READAHEAD);
739 740 741 742 743
				bh_use[ra_max] = bh;
			}
		}
		if ((bh = bh_use[ra_ptr++]) == NULL)
			goto next;
744
		if (ocfs2_read_dir_block(dir, block, &bh, 0)) {
745
			/* read error, skip block & hope for the best.
746
			 * ocfs2_read_dir_block() has released the bh. */
747 748 749 750 751 752 753 754
			ocfs2_error(dir->i_sb, "reading directory %llu, "
				    "offset %lu\n",
				    (unsigned long long)OCFS2_I(dir)->ip_blkno,
				    block);
			goto next;
		}
		i = ocfs2_search_dirblock(bh, dir, name, namelen,
					  block << sb->s_blocksize_bits,
755
					  bh->b_data, sb->s_blocksize,
756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786
					  res_dir);
		if (i == 1) {
			OCFS2_I(dir)->ip_dir_start_lookup = block;
			ret = bh;
			goto cleanup_and_exit;
		} else {
			brelse(bh);
			if (i < 0)
				goto cleanup_and_exit;
		}
	next:
		if (++block >= nblocks)
			block = 0;
	} while (block != start);

	/*
	 * If the directory has grown while we were searching, then
	 * search the last part of the directory before giving up.
	 */
	block = nblocks;
	nblocks = i_size_read(dir) >> sb->s_blocksize_bits;
	if (block < nblocks) {
		start = 0;
		goto restart;
	}

cleanup_and_exit:
	/* Clean up the read-ahead blocks */
	for (; ra_ptr < ra_max; ra_ptr++)
		brelse(bh_use[ra_ptr]);

787
	trace_ocfs2_find_entry_el(ret);
788 789 790
	return ret;
}

791 792 793 794 795 796
static int ocfs2_dx_dir_lookup_rec(struct inode *inode,
				   struct ocfs2_extent_list *el,
				   u32 major_hash,
				   u32 *ret_cpos,
				   u64 *ret_phys_blkno,
				   unsigned int *ret_clen)
797
{
798 799 800 801
	int ret = 0, i, found;
	struct buffer_head *eb_bh = NULL;
	struct ocfs2_extent_block *eb;
	struct ocfs2_extent_rec *rec = NULL;
802

803
	if (el->l_tree_depth) {
804 805
		ret = ocfs2_find_leaf(INODE_CACHE(inode), el, major_hash,
				      &eb_bh);
806 807 808 809
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
810

811 812
		eb = (struct ocfs2_extent_block *) eb_bh->b_data;
		el = &eb->h_list;
813

814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852
		if (el->l_tree_depth) {
			ocfs2_error(inode->i_sb,
				    "Inode %lu has non zero tree depth in "
				    "btree tree block %llu\n", inode->i_ino,
				    (unsigned long long)eb_bh->b_blocknr);
			ret = -EROFS;
			goto out;
		}
	}

	found = 0;
	for (i = le16_to_cpu(el->l_next_free_rec) - 1; i >= 0; i--) {
		rec = &el->l_recs[i];

		if (le32_to_cpu(rec->e_cpos) <= major_hash) {
			found = 1;
			break;
		}
	}

	if (!found) {
		ocfs2_error(inode->i_sb, "Inode %lu has bad extent "
			    "record (%u, %u, 0) in btree", inode->i_ino,
			    le32_to_cpu(rec->e_cpos),
			    ocfs2_rec_clusters(el, rec));
		ret = -EROFS;
		goto out;
	}

	if (ret_phys_blkno)
		*ret_phys_blkno = le64_to_cpu(rec->e_blkno);
	if (ret_cpos)
		*ret_cpos = le32_to_cpu(rec->e_cpos);
	if (ret_clen)
		*ret_clen = le16_to_cpu(rec->e_leaf_clusters);

out:
	brelse(eb_bh);
	return ret;
853 854
}

855
/*
856 857
 * Returns the block index, from the start of the cluster which this
 * hash belongs too.
858
 */
859 860
static inline unsigned int __ocfs2_dx_dir_hash_idx(struct ocfs2_super *osb,
						   u32 minor_hash)
M
Mark Fasheh 已提交
861
{
862 863
	return minor_hash & osb->osb_dx_mask;
}
864

865 866 867 868 869 870
static inline unsigned int ocfs2_dx_dir_hash_idx(struct ocfs2_super *osb,
					  struct ocfs2_dx_hinfo *hinfo)
{
	return __ocfs2_dx_dir_hash_idx(osb, hinfo->minor_hash);
}

871 872 873 874 875 876 877 878 879 880 881
static int ocfs2_dx_dir_lookup(struct inode *inode,
			       struct ocfs2_extent_list *el,
			       struct ocfs2_dx_hinfo *hinfo,
			       u32 *ret_cpos,
			       u64 *ret_phys_blkno)
{
	int ret = 0;
	unsigned int cend, uninitialized_var(clen);
	u32 uninitialized_var(cpos);
	u64 uninitialized_var(blkno);
	u32 name_hash = hinfo->major_hash;
882

883 884
	ret = ocfs2_dx_dir_lookup_rec(inode, el, name_hash, &cpos, &blkno,
				      &clen);
M
Mark Fasheh 已提交
885 886 887 888 889
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

890 891 892 893 894 895 896 897 898 899
	cend = cpos + clen;
	if (name_hash >= cend) {
		/* We want the last cluster */
		blkno += ocfs2_clusters_to_blocks(inode->i_sb, clen - 1);
		cpos += clen - 1;
	} else {
		blkno += ocfs2_clusters_to_blocks(inode->i_sb,
						  name_hash - cpos);
		cpos = name_hash;
	}
M
Mark Fasheh 已提交
900

901 902 903 904 905 906 907 908 909 910 911
	/*
	 * We now have the cluster which should hold our entry. To
	 * find the exact block from the start of the cluster to
	 * search, we take the lower bits of the hash.
	 */
	blkno += ocfs2_dx_dir_hash_idx(OCFS2_SB(inode->i_sb), hinfo);

	if (ret_phys_blkno)
		*ret_phys_blkno = blkno;
	if (ret_cpos)
		*ret_cpos = cpos;
M
Mark Fasheh 已提交
912 913

out:
914

M
Mark Fasheh 已提交
915 916 917
	return ret;
}

918 919
static int ocfs2_dx_dir_search(const char *name, int namelen,
			       struct inode *dir,
920
			       struct ocfs2_dx_root_block *dx_root,
921
			       struct ocfs2_dir_lookup_result *res)
922
{
923 924 925 926 927 928 929 930
	int ret, i, found;
	u64 uninitialized_var(phys);
	struct buffer_head *dx_leaf_bh = NULL;
	struct ocfs2_dx_leaf *dx_leaf;
	struct ocfs2_dx_entry *dx_entry = NULL;
	struct buffer_head *dir_ent_bh = NULL;
	struct ocfs2_dir_entry *dir_ent = NULL;
	struct ocfs2_dx_hinfo *hinfo = &res->dl_hinfo;
931 932
	struct ocfs2_extent_list *dr_el;
	struct ocfs2_dx_entry_list *entry_list;
933 934 935

	ocfs2_dx_dir_name_hash(dir, name, namelen, &res->dl_hinfo);

936 937 938 939 940 941 942
	if (ocfs2_dx_root_inline(dx_root)) {
		entry_list = &dx_root->dr_entries;
		goto search;
	}

	dr_el = &dx_root->dr_list;

943 944 945 946 947
	ret = ocfs2_dx_dir_lookup(dir, dr_el, hinfo, NULL, &phys);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}
948

949 950 951
	trace_ocfs2_dx_dir_search((unsigned long long)OCFS2_I(dir)->ip_blkno,
				  namelen, name, hinfo->major_hash,
				  hinfo->minor_hash, (unsigned long long)phys);
952

953 954 955 956 957
	ret = ocfs2_read_dx_leaf(dir, phys, &dx_leaf_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}
958

959 960
	dx_leaf = (struct ocfs2_dx_leaf *) dx_leaf_bh->b_data;

961 962 963
	trace_ocfs2_dx_dir_search_leaf_info(
			le16_to_cpu(dx_leaf->dl_list.de_num_used),
			le16_to_cpu(dx_leaf->dl_list.de_count));
964

965 966 967
	entry_list = &dx_leaf->dl_list;

search:
968 969 970 971
	/*
	 * Empty leaf is legal, so no need to check for that.
	 */
	found = 0;
972 973
	for (i = 0; i < le16_to_cpu(entry_list->de_num_used); i++) {
		dx_entry = &entry_list->de_entries[i];
974 975 976 977 978 979 980 981 982 983 984 985 986 987 988

		if (hinfo->major_hash != le32_to_cpu(dx_entry->dx_major_hash)
		    || hinfo->minor_hash != le32_to_cpu(dx_entry->dx_minor_hash))
			continue;

		/*
		 * Search unindexed leaf block now. We're not
		 * guaranteed to find anything.
		 */
		ret = ocfs2_read_dir_block_direct(dir,
					  le64_to_cpu(dx_entry->dx_dirent_blk),
					  &dir_ent_bh);
		if (ret) {
			mlog_errno(ret);
			goto out;
989
		}
990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056

		/*
		 * XXX: We should check the unindexed block here,
		 * before using it.
		 */

		found = ocfs2_search_dirblock(dir_ent_bh, dir, name, namelen,
					      0, dir_ent_bh->b_data,
					      dir->i_sb->s_blocksize, &dir_ent);
		if (found == 1)
			break;

		if (found == -1) {
			/* This means we found a bad directory entry. */
			ret = -EIO;
			mlog_errno(ret);
			goto out;
		}

		brelse(dir_ent_bh);
		dir_ent_bh = NULL;
	}

	if (found <= 0) {
		ret = -ENOENT;
		goto out;
	}

	res->dl_leaf_bh = dir_ent_bh;
	res->dl_entry = dir_ent;
	res->dl_dx_leaf_bh = dx_leaf_bh;
	res->dl_dx_entry = dx_entry;

	ret = 0;
out:
	if (ret) {
		brelse(dx_leaf_bh);
		brelse(dir_ent_bh);
	}
	return ret;
}

static int ocfs2_find_entry_dx(const char *name, int namelen,
			       struct inode *dir,
			       struct ocfs2_dir_lookup_result *lookup)
{
	int ret;
	struct buffer_head *di_bh = NULL;
	struct ocfs2_dinode *di;
	struct buffer_head *dx_root_bh = NULL;
	struct ocfs2_dx_root_block *dx_root;

	ret = ocfs2_read_inode_block(dir, &di_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	di = (struct ocfs2_dinode *)di_bh->b_data;

	ret = ocfs2_read_dx_root(dir, di, &dx_root_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}
	dx_root = (struct ocfs2_dx_root_block *) dx_root_bh->b_data;

1057
	ret = ocfs2_dx_dir_search(name, namelen, dir, dx_root, lookup);
1058 1059 1060 1061 1062 1063
	if (ret) {
		if (ret != -ENOENT)
			mlog_errno(ret);
		goto out;
	}

1064 1065
	lookup->dl_dx_root_bh = dx_root_bh;
	dx_root_bh = NULL;
1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133
out:
	brelse(di_bh);
	brelse(dx_root_bh);
	return ret;
}

/*
 * Try to find an entry of the provided name within 'dir'.
 *
 * If nothing was found, -ENOENT is returned. Otherwise, zero is
 * returned and the struct 'res' will contain information useful to
 * other directory manipulation functions.
 *
 * Caller can NOT assume anything about the contents of the
 * buffer_heads - they are passed back only so that it can be passed
 * into any one of the manipulation functions (add entry, delete
 * entry, etc). As an example, bh in the extent directory case is a
 * data block, in the inline-data case it actually points to an inode,
 * in the indexed directory case, multiple buffers are involved.
 */
int ocfs2_find_entry(const char *name, int namelen,
		     struct inode *dir, struct ocfs2_dir_lookup_result *lookup)
{
	struct buffer_head *bh;
	struct ocfs2_dir_entry *res_dir = NULL;

	if (ocfs2_dir_indexed(dir))
		return ocfs2_find_entry_dx(name, namelen, dir, lookup);

	/*
	 * The unindexed dir code only uses part of the lookup
	 * structure, so there's no reason to push it down further
	 * than this.
	 */
	if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL)
		bh = ocfs2_find_entry_id(name, namelen, dir, &res_dir);
	else
		bh = ocfs2_find_entry_el(name, namelen, dir, &res_dir);

	if (bh == NULL)
		return -ENOENT;

	lookup->dl_leaf_bh = bh;
	lookup->dl_entry = res_dir;
	return 0;
}

/*
 * Update inode number and type of a previously found directory entry.
 */
int ocfs2_update_entry(struct inode *dir, handle_t *handle,
		       struct ocfs2_dir_lookup_result *res,
		       struct inode *new_entry_inode)
{
	int ret;
	ocfs2_journal_access_func access = ocfs2_journal_access_db;
	struct ocfs2_dir_entry *de = res->dl_entry;
	struct buffer_head *de_bh = res->dl_leaf_bh;

	/*
	 * The same code works fine for both inline-data and extent
	 * based directories, so no need to split this up.  The only
	 * difference is the journal_access function.
	 */

	if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL)
		access = ocfs2_journal_access_di;

1134 1135
	ret = access(handle, INODE_CACHE(dir), de_bh,
		     OCFS2_JOURNAL_ACCESS_WRITE);
1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	de->inode = cpu_to_le64(OCFS2_I(new_entry_inode)->ip_blkno);
	ocfs2_set_de_type(de, new_entry_inode->i_mode);

	ocfs2_journal_dirty(handle, de_bh);

out:
	return ret;
}

/*
 * __ocfs2_delete_entry deletes a directory entry by merging it with the
 * previous entry
 */
static int __ocfs2_delete_entry(handle_t *handle, struct inode *dir,
				struct ocfs2_dir_entry *de_del,
				struct buffer_head *bh, char *first_de,
				unsigned int bytes)
{
	struct ocfs2_dir_entry *de, *pde;
	int i, status = -ENOENT;
	ocfs2_journal_access_func access = ocfs2_journal_access_db;

	if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL)
		access = ocfs2_journal_access_di;

	i = 0;
	pde = NULL;
	de = (struct ocfs2_dir_entry *) first_de;
	while (i < bytes) {
		if (!ocfs2_check_dir_entry(dir, de, bh, i)) {
			status = -EIO;
			mlog_errno(status);
			goto bail;
		}
		if (de == de_del)  {
1176
			status = access(handle, INODE_CACHE(dir), bh,
1177
					OCFS2_JOURNAL_ACCESS_WRITE);
1178 1179 1180 1181 1182 1183
			if (status < 0) {
				status = -EIO;
				mlog_errno(status);
				goto bail;
			}
			if (pde)
1184 1185
				le16_add_cpu(&pde->rec_len,
						le16_to_cpu(de->rec_len));
1186
			de->inode = 0;
1187
			dir->i_version++;
1188
			ocfs2_journal_dirty(handle, bh);
1189 1190 1191 1192 1193 1194 1195 1196 1197 1198
			goto bail;
		}
		i += le16_to_cpu(de->rec_len);
		pde = de;
		de = (struct ocfs2_dir_entry *)((char *)de + le16_to_cpu(de->rec_len));
	}
bail:
	return status;
}

1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 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
static unsigned int ocfs2_figure_dirent_hole(struct ocfs2_dir_entry *de)
{
	unsigned int hole;

	if (le64_to_cpu(de->inode) == 0)
		hole = le16_to_cpu(de->rec_len);
	else
		hole = le16_to_cpu(de->rec_len) -
			OCFS2_DIR_REC_LEN(de->name_len);

	return hole;
}

static int ocfs2_find_max_rec_len(struct super_block *sb,
				  struct buffer_head *dirblock_bh)
{
	int size, this_hole, largest_hole = 0;
	char *trailer, *de_buf, *limit, *start = dirblock_bh->b_data;
	struct ocfs2_dir_entry *de;

	trailer = (char *)ocfs2_trailer_from_bh(dirblock_bh, sb);
	size = ocfs2_dir_trailer_blk_off(sb);
	limit = start + size;
	de_buf = start;
	de = (struct ocfs2_dir_entry *)de_buf;
	do {
		if (de_buf != trailer) {
			this_hole = ocfs2_figure_dirent_hole(de);
			if (this_hole > largest_hole)
				largest_hole = this_hole;
		}

		de_buf += le16_to_cpu(de->rec_len);
		de = (struct ocfs2_dir_entry *)de_buf;
	} while (de_buf < limit);

	if (largest_hole >= OCFS2_DIR_MIN_REC_LEN)
		return largest_hole;
	return 0;
}

1240 1241
static void ocfs2_dx_list_remove_entry(struct ocfs2_dx_entry_list *entry_list,
				       int index)
1242
{
1243
	int num_used = le16_to_cpu(entry_list->de_num_used);
1244 1245 1246 1247

	if (num_used == 1 || index == (num_used - 1))
		goto clear;

1248 1249
	memmove(&entry_list->de_entries[index],
		&entry_list->de_entries[index + 1],
1250 1251 1252
		(num_used - index - 1)*sizeof(struct ocfs2_dx_entry));
clear:
	num_used--;
1253
	memset(&entry_list->de_entries[num_used], 0,
1254
	       sizeof(struct ocfs2_dx_entry));
1255
	entry_list->de_num_used = cpu_to_le16(num_used);
1256 1257 1258 1259 1260
}

static int ocfs2_delete_entry_dx(handle_t *handle, struct inode *dir,
				 struct ocfs2_dir_lookup_result *lookup)
{
1261
	int ret, index, max_rec_len, add_to_free_list = 0;
1262
	struct buffer_head *dx_root_bh = lookup->dl_dx_root_bh;
1263 1264 1265
	struct buffer_head *leaf_bh = lookup->dl_leaf_bh;
	struct ocfs2_dx_leaf *dx_leaf;
	struct ocfs2_dx_entry *dx_entry = lookup->dl_dx_entry;
1266
	struct ocfs2_dir_block_trailer *trailer;
1267 1268 1269
	struct ocfs2_dx_root_block *dx_root;
	struct ocfs2_dx_entry_list *entry_list;

1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
	/*
	 * This function gets a bit messy because we might have to
	 * modify the root block, regardless of whether the indexed
	 * entries are stored inline.
	 */

	/*
	 * *Only* set 'entry_list' here, based on where we're looking
	 * for the indexed entries. Later, we might still want to
	 * journal both blocks, based on free list state.
	 */
1281 1282 1283 1284 1285 1286 1287
	dx_root = (struct ocfs2_dx_root_block *)dx_root_bh->b_data;
	if (ocfs2_dx_root_inline(dx_root)) {
		entry_list = &dx_root->dr_entries;
	} else {
		dx_leaf = (struct ocfs2_dx_leaf *) lookup->dl_dx_leaf_bh->b_data;
		entry_list = &dx_leaf->dl_list;
	}
1288 1289 1290

	/* Neither of these are a disk corruption - that should have
	 * been caught by lookup, before we got here. */
1291 1292
	BUG_ON(le16_to_cpu(entry_list->de_count) <= 0);
	BUG_ON(le16_to_cpu(entry_list->de_num_used) <= 0);
1293

1294
	index = (char *)dx_entry - (char *)entry_list->de_entries;
1295 1296
	index /= sizeof(*dx_entry);

1297
	if (index >= le16_to_cpu(entry_list->de_num_used)) {
1298
		mlog(ML_ERROR, "Dir %llu: Bad dx_entry ptr idx %d, (%p, %p)\n",
1299 1300
		     (unsigned long long)OCFS2_I(dir)->ip_blkno, index,
		     entry_list, dx_entry);
1301 1302 1303
		return -EIO;
	}

1304 1305 1306 1307 1308 1309 1310 1311 1312
	/*
	 * We know that removal of this dirent will leave enough room
	 * for a new one, so add this block to the free list if it
	 * isn't already there.
	 */
	trailer = ocfs2_trailer_from_bh(leaf_bh, dir->i_sb);
	if (trailer->db_free_rec_len == 0)
		add_to_free_list = 1;

1313
	/*
1314 1315 1316 1317 1318 1319
	 * Add the block holding our index into the journal before
	 * removing the unindexed entry. If we get an error return
	 * from __ocfs2_delete_entry(), then it hasn't removed the
	 * entry yet. Likewise, successful return means we *must*
	 * remove the indexed entry.
	 *
1320 1321 1322
	 * We're also careful to journal the root tree block here as
	 * the entry count needs to be updated. Also, we might be
	 * adding to the start of the free list.
1323
	 */
1324
	ret = ocfs2_journal_access_dr(handle, INODE_CACHE(dir), dx_root_bh,
1325 1326 1327 1328
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out;
1329 1330 1331
	}

	if (!ocfs2_dx_root_inline(dx_root)) {
1332
		ret = ocfs2_journal_access_dl(handle, INODE_CACHE(dir),
1333 1334 1335 1336 1337 1338
					      lookup->dl_dx_leaf_bh,
					      OCFS2_JOURNAL_ACCESS_WRITE);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
1339 1340
	}

1341 1342
	trace_ocfs2_delete_entry_dx((unsigned long long)OCFS2_I(dir)->ip_blkno,
				    index);
1343

1344 1345 1346 1347 1348 1349 1350
	ret = __ocfs2_delete_entry(handle, dir, lookup->dl_entry,
				   leaf_bh, leaf_bh->b_data, leaf_bh->b_size);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361
	max_rec_len = ocfs2_find_max_rec_len(dir->i_sb, leaf_bh);
	trailer->db_free_rec_len = cpu_to_le16(max_rec_len);
	if (add_to_free_list) {
		trailer->db_free_next = dx_root->dr_free_blk;
		dx_root->dr_free_blk = cpu_to_le64(leaf_bh->b_blocknr);
		ocfs2_journal_dirty(handle, dx_root_bh);
	}

	/* leaf_bh was journal_accessed for us in __ocfs2_delete_entry */
	ocfs2_journal_dirty(handle, leaf_bh);

1362 1363 1364
	le32_add_cpu(&dx_root->dr_num_entries, -1);
	ocfs2_journal_dirty(handle, dx_root_bh);

1365
	ocfs2_dx_list_remove_entry(entry_list, index);
1366

1367
	if (!ocfs2_dx_root_inline(dx_root))
1368
		ocfs2_journal_dirty(handle, lookup->dl_dx_leaf_bh);
1369 1370 1371 1372 1373

out:
	return ret;
}

1374 1375 1376 1377 1378 1379 1380 1381 1382 1383
static inline int ocfs2_delete_entry_id(handle_t *handle,
					struct inode *dir,
					struct ocfs2_dir_entry *de_del,
					struct buffer_head *bh)
{
	int ret;
	struct buffer_head *di_bh = NULL;
	struct ocfs2_dinode *di;
	struct ocfs2_inline_data *data;

1384
	ret = ocfs2_read_inode_block(dir, &di_bh);
1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	di = (struct ocfs2_dinode *)di_bh->b_data;
	data = &di->id2.i_data;

	ret = __ocfs2_delete_entry(handle, dir, de_del, bh, data->id_data,
				   i_size_read(dir));

	brelse(di_bh);
out:
	return ret;
}

static inline int ocfs2_delete_entry_el(handle_t *handle,
					struct inode *dir,
					struct ocfs2_dir_entry *de_del,
					struct buffer_head *bh)
{
	return __ocfs2_delete_entry(handle, dir, de_del, bh, bh->b_data,
				    bh->b_size);
}

/*
1411 1412
 * Delete a directory entry. Hide the details of directory
 * implementation from the caller.
1413 1414 1415
 */
int ocfs2_delete_entry(handle_t *handle,
		       struct inode *dir,
1416
		       struct ocfs2_dir_lookup_result *res)
1417
{
1418 1419 1420
	if (ocfs2_dir_indexed(dir))
		return ocfs2_delete_entry_dx(handle, dir, res);

1421
	if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL)
1422 1423
		return ocfs2_delete_entry_id(handle, dir, res->dl_entry,
					     res->dl_leaf_bh);
1424

1425 1426
	return ocfs2_delete_entry_el(handle, dir, res->dl_entry,
				     res->dl_leaf_bh);
1427 1428
}

M
Mark Fasheh 已提交
1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453
/*
 * Check whether 'de' has enough room to hold an entry of
 * 'new_rec_len' bytes.
 */
static inline int ocfs2_dirent_would_fit(struct ocfs2_dir_entry *de,
					 unsigned int new_rec_len)
{
	unsigned int de_really_used;

	/* Check whether this is an empty record with enough space */
	if (le64_to_cpu(de->inode) == 0 &&
	    le16_to_cpu(de->rec_len) >= new_rec_len)
		return 1;

	/*
	 * Record might have free space at the end which we can
	 * use.
	 */
	de_really_used = OCFS2_DIR_REC_LEN(de->name_len);
	if (le16_to_cpu(de->rec_len) >= (de_really_used + new_rec_len))
	    return 1;

	return 0;
}

1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464
static void ocfs2_dx_dir_leaf_insert_tail(struct ocfs2_dx_leaf *dx_leaf,
					  struct ocfs2_dx_entry *dx_new_entry)
{
	int i;

	i = le16_to_cpu(dx_leaf->dl_list.de_num_used);
	dx_leaf->dl_list.de_entries[i] = *dx_new_entry;

	le16_add_cpu(&dx_leaf->dl_list.de_num_used, 1);
}

1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482
static void ocfs2_dx_entry_list_insert(struct ocfs2_dx_entry_list *entry_list,
				       struct ocfs2_dx_hinfo *hinfo,
				       u64 dirent_blk)
{
	int i;
	struct ocfs2_dx_entry *dx_entry;

	i = le16_to_cpu(entry_list->de_num_used);
	dx_entry = &entry_list->de_entries[i];

	memset(dx_entry, 0, sizeof(*dx_entry));
	dx_entry->dx_major_hash = cpu_to_le32(hinfo->major_hash);
	dx_entry->dx_minor_hash = cpu_to_le32(hinfo->minor_hash);
	dx_entry->dx_dirent_blk = cpu_to_le64(dirent_blk);

	le16_add_cpu(&entry_list->de_num_used, 1);
}

1483 1484 1485 1486 1487
static int __ocfs2_dx_dir_leaf_insert(struct inode *dir, handle_t *handle,
				      struct ocfs2_dx_hinfo *hinfo,
				      u64 dirent_blk,
				      struct buffer_head *dx_leaf_bh)
{
1488
	int ret;
1489 1490
	struct ocfs2_dx_leaf *dx_leaf;

1491
	ret = ocfs2_journal_access_dl(handle, INODE_CACHE(dir), dx_leaf_bh,
1492 1493 1494 1495 1496 1497 1498
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	dx_leaf = (struct ocfs2_dx_leaf *)dx_leaf_bh->b_data;
1499 1500
	ocfs2_dx_entry_list_insert(&dx_leaf->dl_list, hinfo, dirent_blk);
	ocfs2_journal_dirty(handle, dx_leaf_bh);
1501

1502 1503 1504
out:
	return ret;
}
1505

1506 1507 1508 1509
static void ocfs2_dx_inline_root_insert(struct inode *dir, handle_t *handle,
					struct ocfs2_dx_hinfo *hinfo,
					u64 dirent_blk,
					struct ocfs2_dx_root_block *dx_root)
1510
{
1511 1512 1513 1514 1515 1516 1517
	ocfs2_dx_entry_list_insert(&dx_root->dr_entries, hinfo, dirent_blk);
}

static int ocfs2_dx_dir_insert(struct inode *dir, handle_t *handle,
			       struct ocfs2_dir_lookup_result *lookup)
{
	int ret = 0;
1518
	struct ocfs2_dx_root_block *dx_root;
1519
	struct buffer_head *dx_root_bh = lookup->dl_dx_root_bh;
1520

1521
	ret = ocfs2_journal_access_dr(handle, INODE_CACHE(dir), dx_root_bh,
1522 1523 1524 1525 1526 1527
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542
	dx_root = (struct ocfs2_dx_root_block *)lookup->dl_dx_root_bh->b_data;
	if (ocfs2_dx_root_inline(dx_root)) {
		ocfs2_dx_inline_root_insert(dir, handle,
					    &lookup->dl_hinfo,
					    lookup->dl_leaf_bh->b_blocknr,
					    dx_root);
	} else {
		ret = __ocfs2_dx_dir_leaf_insert(dir, handle, &lookup->dl_hinfo,
						 lookup->dl_leaf_bh->b_blocknr,
						 lookup->dl_dx_leaf_bh);
		if (ret)
			goto out;
	}

	le32_add_cpu(&dx_root->dr_num_entries, 1);
1543
	ocfs2_journal_dirty(handle, dx_root_bh);
1544 1545 1546 1547 1548

out:
	return ret;
}

1549 1550 1551 1552 1553 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 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601
static void ocfs2_remove_block_from_free_list(struct inode *dir,
				       handle_t *handle,
				       struct ocfs2_dir_lookup_result *lookup)
{
	struct ocfs2_dir_block_trailer *trailer, *prev;
	struct ocfs2_dx_root_block *dx_root;
	struct buffer_head *bh;

	trailer = ocfs2_trailer_from_bh(lookup->dl_leaf_bh, dir->i_sb);

	if (ocfs2_free_list_at_root(lookup)) {
		bh = lookup->dl_dx_root_bh;
		dx_root = (struct ocfs2_dx_root_block *)bh->b_data;
		dx_root->dr_free_blk = trailer->db_free_next;
	} else {
		bh = lookup->dl_prev_leaf_bh;
		prev = ocfs2_trailer_from_bh(bh, dir->i_sb);
		prev->db_free_next = trailer->db_free_next;
	}

	trailer->db_free_rec_len = cpu_to_le16(0);
	trailer->db_free_next = cpu_to_le64(0);

	ocfs2_journal_dirty(handle, bh);
	ocfs2_journal_dirty(handle, lookup->dl_leaf_bh);
}

/*
 * This expects that a journal write has been reserved on
 * lookup->dl_prev_leaf_bh or lookup->dl_dx_root_bh
 */
static void ocfs2_recalc_free_list(struct inode *dir, handle_t *handle,
				   struct ocfs2_dir_lookup_result *lookup)
{
	int max_rec_len;
	struct ocfs2_dir_block_trailer *trailer;

	/* Walk dl_leaf_bh to figure out what the new free rec_len is. */
	max_rec_len = ocfs2_find_max_rec_len(dir->i_sb, lookup->dl_leaf_bh);
	if (max_rec_len) {
		/*
		 * There's still room in this block, so no need to remove it
		 * from the free list. In this case, we just want to update
		 * the rec len accounting.
		 */
		trailer = ocfs2_trailer_from_bh(lookup->dl_leaf_bh, dir->i_sb);
		trailer->db_free_rec_len = cpu_to_le16(max_rec_len);
		ocfs2_journal_dirty(handle, lookup->dl_leaf_bh);
	} else {
		ocfs2_remove_block_from_free_list(dir, handle, lookup);
	}
}

1602 1603 1604
/* we don't always have a dentry for what we want to add, so people
 * like orphan dir can call this instead.
 *
1605 1606
 * The lookup context must have been filled from
 * ocfs2_prepare_dir_for_insert.
1607 1608 1609 1610 1611 1612
 */
int __ocfs2_add_entry(handle_t *handle,
		      struct inode *dir,
		      const char *name, int namelen,
		      struct inode *inode, u64 blkno,
		      struct buffer_head *parent_fe_bh,
1613
		      struct ocfs2_dir_lookup_result *lookup)
1614 1615 1616 1617
{
	unsigned long offset;
	unsigned short rec_len;
	struct ocfs2_dir_entry *de, *de1;
1618 1619
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)parent_fe_bh->b_data;
	struct super_block *sb = dir->i_sb;
1620
	int retval, status;
1621
	unsigned int size = sb->s_blocksize;
1622
	struct buffer_head *insert_bh = lookup->dl_leaf_bh;
1623
	char *data_start = insert_bh->b_data;
1624 1625 1626 1627

	if (!namelen)
		return -EINVAL;

1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640
	if (ocfs2_dir_indexed(dir)) {
		struct buffer_head *bh;

		/*
		 * An indexed dir may require that we update the free space
		 * list. Reserve a write to the previous node in the list so
		 * that we don't fail later.
		 *
		 * XXX: This can be either a dx_root_block, or an unindexed
		 * directory tree leaf block.
		 */
		if (ocfs2_free_list_at_root(lookup)) {
			bh = lookup->dl_dx_root_bh;
1641 1642
			retval = ocfs2_journal_access_dr(handle,
						 INODE_CACHE(dir), bh,
1643 1644 1645
						 OCFS2_JOURNAL_ACCESS_WRITE);
		} else {
			bh = lookup->dl_prev_leaf_bh;
1646 1647
			retval = ocfs2_journal_access_db(handle,
						 INODE_CACHE(dir), bh,
1648 1649 1650 1651 1652 1653 1654
						 OCFS2_JOURNAL_ACCESS_WRITE);
		}
		if (retval) {
			mlog_errno(retval);
			return retval;
		}
	} else if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL) {
1655 1656 1657 1658 1659 1660
		data_start = di->id2.i_data.id_data;
		size = i_size_read(dir);

		BUG_ON(insert_bh != parent_fe_bh);
	}

1661 1662
	rec_len = OCFS2_DIR_REC_LEN(namelen);
	offset = 0;
1663
	de = (struct ocfs2_dir_entry *) data_start;
1664
	while (1) {
1665 1666
		BUG_ON((char *)de >= (size + data_start));

1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677
		/* These checks should've already been passed by the
		 * prepare function, but I guess we can leave them
		 * here anyway. */
		if (!ocfs2_check_dir_entry(dir, de, insert_bh, offset)) {
			retval = -ENOENT;
			goto bail;
		}
		if (ocfs2_match(namelen, name, de)) {
			retval = -EEXIST;
			goto bail;
		}
M
Mark Fasheh 已提交
1678

1679 1680 1681 1682 1683 1684 1685 1686 1687 1688
		/* We're guaranteed that we should have space, so we
		 * can't possibly have hit the trailer...right? */
		mlog_bug_on_msg(ocfs2_skip_dir_trailer(dir, de, offset, size),
				"Hit dir trailer trying to insert %.*s "
			        "(namelen %d) into directory %llu.  "
				"offset is %lu, trailer offset is %d\n",
				namelen, name, namelen,
				(unsigned long long)parent_fe_bh->b_blocknr,
				offset, ocfs2_dir_trailer_blk_off(dir->i_sb));

M
Mark Fasheh 已提交
1689
		if (ocfs2_dirent_would_fit(de, rec_len)) {
1690 1691 1692 1693 1694 1695 1696
			dir->i_mtime = dir->i_ctime = CURRENT_TIME;
			retval = ocfs2_mark_inode_dirty(handle, dir, parent_fe_bh);
			if (retval < 0) {
				mlog_errno(retval);
				goto bail;
			}

1697
			if (insert_bh == parent_fe_bh)
1698 1699
				status = ocfs2_journal_access_di(handle,
								 INODE_CACHE(dir),
1700 1701
								 insert_bh,
								 OCFS2_JOURNAL_ACCESS_WRITE);
1702
			else {
1703 1704
				status = ocfs2_journal_access_db(handle,
								 INODE_CACHE(dir),
1705
								 insert_bh,
1706 1707
					      OCFS2_JOURNAL_ACCESS_WRITE);

1708
				if (ocfs2_dir_indexed(dir)) {
1709 1710 1711
					status = ocfs2_dx_dir_insert(dir,
								handle,
								lookup);
1712 1713 1714 1715 1716 1717 1718
					if (status) {
						mlog_errno(status);
						goto bail;
					}
				}
			}

1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738
			/* By now the buffer is marked for journaling */
			offset += le16_to_cpu(de->rec_len);
			if (le64_to_cpu(de->inode)) {
				de1 = (struct ocfs2_dir_entry *)((char *) de +
					OCFS2_DIR_REC_LEN(de->name_len));
				de1->rec_len =
					cpu_to_le16(le16_to_cpu(de->rec_len) -
					OCFS2_DIR_REC_LEN(de->name_len));
				de->rec_len = cpu_to_le16(OCFS2_DIR_REC_LEN(de->name_len));
				de = de1;
			}
			de->file_type = OCFS2_FT_UNKNOWN;
			if (blkno) {
				de->inode = cpu_to_le64(blkno);
				ocfs2_set_de_type(de, inode->i_mode);
			} else
				de->inode = 0;
			de->name_len = namelen;
			memcpy(de->name, name, namelen);

1739 1740 1741
			if (ocfs2_dir_indexed(dir))
				ocfs2_recalc_free_list(dir, handle, lookup);

1742
			dir->i_version++;
1743
			ocfs2_journal_dirty(handle, insert_bh);
1744 1745 1746
			retval = 0;
			goto bail;
		}
1747

1748 1749 1750 1751 1752 1753 1754 1755
		offset += le16_to_cpu(de->rec_len);
		de = (struct ocfs2_dir_entry *) ((char *) de + le16_to_cpu(de->rec_len));
	}

	/* when you think about it, the assert above should prevent us
	 * from ever getting here. */
	retval = -ENOSPC;
bail:
T
Tao Ma 已提交
1756 1757
	if (retval)
		mlog_errno(retval);
1758 1759 1760 1761

	return retval;
}

1762
static int ocfs2_dir_foreach_blk_id(struct inode *inode,
1763
				    u64 *f_version,
1764
				    loff_t *f_pos, void *priv,
1765
				    filldir_t filldir, int *filldir_err)
1766 1767 1768 1769 1770 1771 1772 1773
{
	int ret, i, filldir_ret;
	unsigned long offset = *f_pos;
	struct buffer_head *di_bh = NULL;
	struct ocfs2_dinode *di;
	struct ocfs2_inline_data *data;
	struct ocfs2_dir_entry *de;

1774
	ret = ocfs2_read_inode_block(inode, &di_bh);
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 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823
	if (ret) {
		mlog(ML_ERROR, "Unable to read inode block for dir %llu\n",
		     (unsigned long long)OCFS2_I(inode)->ip_blkno);
		goto out;
	}

	di = (struct ocfs2_dinode *)di_bh->b_data;
	data = &di->id2.i_data;

	while (*f_pos < i_size_read(inode)) {
revalidate:
		/* If the dir block has changed since the last call to
		 * readdir(2), then we might be pointing to an invalid
		 * dirent right now.  Scan from the start of the block
		 * to make sure. */
		if (*f_version != inode->i_version) {
			for (i = 0; i < i_size_read(inode) && i < offset; ) {
				de = (struct ocfs2_dir_entry *)
					(data->id_data + i);
				/* It's too expensive to do a full
				 * dirent test each time round this
				 * loop, but we do have to test at
				 * least that it is non-zero.  A
				 * failure will be detected in the
				 * dirent test below. */
				if (le16_to_cpu(de->rec_len) <
				    OCFS2_DIR_REC_LEN(1))
					break;
				i += le16_to_cpu(de->rec_len);
			}
			*f_pos = offset = i;
			*f_version = inode->i_version;
		}

		de = (struct ocfs2_dir_entry *) (data->id_data + *f_pos);
		if (!ocfs2_check_dir_entry(inode, de, di_bh, *f_pos)) {
			/* On error, skip the f_pos to the end. */
			*f_pos = i_size_read(inode);
			goto out;
		}
		offset += le16_to_cpu(de->rec_len);
		if (le64_to_cpu(de->inode)) {
			/* We might block in the next section
			 * if the data destination is
			 * currently swapped out.  So, use a
			 * version stamp to detect whether or
			 * not the directory has been modified
			 * during the copy operation.
			 */
1824
			u64 version = *f_version;
1825 1826 1827 1828 1829 1830 1831 1832 1833 1834
			unsigned char d_type = DT_UNKNOWN;

			if (de->file_type < OCFS2_FT_MAX)
				d_type = ocfs2_filetype_table[de->file_type];

			filldir_ret = filldir(priv, de->name,
					      de->name_len,
					      *f_pos,
					      le64_to_cpu(de->inode),
					      d_type);
1835 1836 1837
			if (filldir_ret) {
				if (filldir_err)
					*filldir_err = filldir_ret;
1838
				break;
1839
			}
1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851
			if (version != *f_version)
				goto revalidate;
		}
		*f_pos += le16_to_cpu(de->rec_len);
	}

out:
	brelse(di_bh);

	return 0;
}

1852 1853 1854 1855
/*
 * NOTE: This function can be called against unindexed directories,
 * and indexed ones.
 */
1856
static int ocfs2_dir_foreach_blk_el(struct inode *inode,
1857
				    u64 *f_version,
1858
				    loff_t *f_pos, void *priv,
1859
				    filldir_t filldir, int *filldir_err)
1860 1861
{
	int error = 0;
1862 1863
	unsigned long offset, blk, last_ra_blk = 0;
	int i, stored;
1864 1865 1866
	struct buffer_head * bh, * tmp;
	struct ocfs2_dir_entry * de;
	struct super_block * sb = inode->i_sb;
1867
	unsigned int ra_sectors = 16;
1868 1869 1870 1871

	stored = 0;
	bh = NULL;

1872
	offset = (*f_pos) & (sb->s_blocksize - 1);
1873

1874 1875
	while (!error && !stored && *f_pos < i_size_read(inode)) {
		blk = (*f_pos) >> sb->s_blocksize_bits;
1876 1877
		if (ocfs2_read_dir_block(inode, blk, &bh, 0)) {
			/* Skip the corrupt dirblock and keep trying */
1878
			*f_pos += sb->s_blocksize - offset;
1879 1880 1881
			continue;
		}

1882 1883 1884 1885 1886 1887 1888 1889
		/* The idea here is to begin with 8k read-ahead and to stay
		 * 4k ahead of our current position.
		 *
		 * TODO: Use the pagecache for this. We just need to
		 * make sure it's cluster-safe... */
		if (!last_ra_blk
		    || (((last_ra_blk - blk) << 9) <= (ra_sectors / 2))) {
			for (i = ra_sectors >> (sb->s_blocksize_bits - 9);
1890
			     i > 0; i--) {
1891 1892 1893 1894
				tmp = NULL;
				if (!ocfs2_read_dir_block(inode, ++blk, &tmp,
							  OCFS2_BH_READAHEAD))
					brelse(tmp);
1895
			}
1896 1897
			last_ra_blk = blk;
			ra_sectors = 8;
1898 1899 1900 1901 1902 1903 1904
		}

revalidate:
		/* If the dir block has changed since the last call to
		 * readdir(2), then we might be pointing to an invalid
		 * dirent right now.  Scan from the start of the block
		 * to make sure. */
1905
		if (*f_version != inode->i_version) {
1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919
			for (i = 0; i < sb->s_blocksize && i < offset; ) {
				de = (struct ocfs2_dir_entry *) (bh->b_data + i);
				/* It's too expensive to do a full
				 * dirent test each time round this
				 * loop, but we do have to test at
				 * least that it is non-zero.  A
				 * failure will be detected in the
				 * dirent test below. */
				if (le16_to_cpu(de->rec_len) <
				    OCFS2_DIR_REC_LEN(1))
					break;
				i += le16_to_cpu(de->rec_len);
			}
			offset = i;
1920
			*f_pos = ((*f_pos) & ~(sb->s_blocksize - 1))
1921
				| offset;
1922
			*f_version = inode->i_version;
1923 1924
		}

1925
		while (!error && *f_pos < i_size_read(inode)
1926 1927 1928 1929 1930
		       && offset < sb->s_blocksize) {
			de = (struct ocfs2_dir_entry *) (bh->b_data + offset);
			if (!ocfs2_check_dir_entry(inode, de, bh, offset)) {
				/* On error, skip the f_pos to the
				   next block. */
1931
				*f_pos = ((*f_pos) | (sb->s_blocksize - 1)) + 1;
1932
				brelse(bh);
1933
				goto out;
1934 1935 1936 1937 1938 1939 1940 1941 1942 1943
			}
			offset += le16_to_cpu(de->rec_len);
			if (le64_to_cpu(de->inode)) {
				/* We might block in the next section
				 * if the data destination is
				 * currently swapped out.  So, use a
				 * version stamp to detect whether or
				 * not the directory has been modified
				 * during the copy operation.
				 */
1944
				unsigned long version = *f_version;
1945 1946 1947 1948
				unsigned char d_type = DT_UNKNOWN;

				if (de->file_type < OCFS2_FT_MAX)
					d_type = ocfs2_filetype_table[de->file_type];
1949
				error = filldir(priv, de->name,
1950
						de->name_len,
1951
						*f_pos,
M
Mark Fasheh 已提交
1952
						le64_to_cpu(de->inode),
1953
						d_type);
1954 1955 1956
				if (error) {
					if (filldir_err)
						*filldir_err = error;
1957
					break;
1958
				}
1959
				if (version != *f_version)
1960 1961 1962
					goto revalidate;
				stored ++;
			}
1963
			*f_pos += le16_to_cpu(de->rec_len);
1964 1965 1966
		}
		offset = 0;
		brelse(bh);
1967
		bh = NULL;
1968 1969 1970
	}

	stored = 0;
1971 1972 1973 1974
out:
	return stored;
}

1975
static int ocfs2_dir_foreach_blk(struct inode *inode, u64 *f_version,
1976 1977
				 loff_t *f_pos, void *priv, filldir_t filldir,
				 int *filldir_err)
1978 1979 1980
{
	if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL)
		return ocfs2_dir_foreach_blk_id(inode, f_version, f_pos, priv,
1981
						filldir, filldir_err);
1982

1983 1984
	return ocfs2_dir_foreach_blk_el(inode, f_version, f_pos, priv, filldir,
					filldir_err);
1985 1986
}

M
Mark Fasheh 已提交
1987 1988 1989 1990 1991 1992 1993
/*
 * This is intended to be called from inside other kernel functions,
 * so we fake some arguments.
 */
int ocfs2_dir_foreach(struct inode *inode, loff_t *f_pos, void *priv,
		      filldir_t filldir)
{
1994
	int ret = 0, filldir_err = 0;
1995
	u64 version = inode->i_version;
M
Mark Fasheh 已提交
1996 1997 1998

	while (*f_pos < i_size_read(inode)) {
		ret = ocfs2_dir_foreach_blk(inode, &version, f_pos, priv,
1999 2000
					    filldir, &filldir_err);
		if (ret || filldir_err)
M
Mark Fasheh 已提交
2001 2002 2003
			break;
	}

2004 2005 2006
	if (ret > 0)
		ret = -EIO;

M
Mark Fasheh 已提交
2007 2008 2009
	return 0;
}

2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
/*
 * ocfs2_readdir()
 *
 */
int ocfs2_readdir(struct file * filp, void * dirent, filldir_t filldir)
{
	int error = 0;
	struct inode *inode = filp->f_path.dentry->d_inode;
	int lock_level = 0;

2020
	trace_ocfs2_readdir((unsigned long long)OCFS2_I(inode)->ip_blkno);
2021

M
Mark Fasheh 已提交
2022
	error = ocfs2_inode_lock_atime(inode, filp->f_vfsmnt, &lock_level);
2023 2024 2025 2026
	if (lock_level && error >= 0) {
		/* We release EX lock which used to update atime
		 * and get PR lock again to reduce contention
		 * on commonly accessed directories. */
M
Mark Fasheh 已提交
2027
		ocfs2_inode_unlock(inode, 1);
2028
		lock_level = 0;
M
Mark Fasheh 已提交
2029
		error = ocfs2_inode_lock(inode, NULL, 0);
2030 2031 2032 2033 2034 2035 2036 2037 2038
	}
	if (error < 0) {
		if (error != -ENOENT)
			mlog_errno(error);
		/* we haven't got any yet, so propagate the error. */
		goto bail_nolock;
	}

	error = ocfs2_dir_foreach_blk(inode, &filp->f_version, &filp->f_pos,
2039
				      dirent, filldir, NULL);
2040

M
Mark Fasheh 已提交
2041
	ocfs2_inode_unlock(inode, lock_level);
T
Tao Ma 已提交
2042 2043
	if (error)
		mlog_errno(error);
2044

2045
bail_nolock:
2046

2047
	return error;
2048 2049 2050
}

/*
2051
 * NOTE: this should always be called with parent dir i_mutex taken.
2052 2053 2054 2055 2056
 */
int ocfs2_find_files_on_disk(const char *name,
			     int namelen,
			     u64 *blkno,
			     struct inode *inode,
2057
			     struct ocfs2_dir_lookup_result *lookup)
2058 2059 2060
{
	int status = -ENOENT;

2061 2062
	trace_ocfs2_find_files_on_disk(namelen, name, blkno,
				(unsigned long long)OCFS2_I(inode)->ip_blkno);
2063

2064 2065
	status = ocfs2_find_entry(name, namelen, inode, lookup);
	if (status)
2066 2067
		goto leave;

2068
	*blkno = le64_to_cpu(lookup->dl_entry->inode);
2069 2070 2071 2072 2073 2074 2075

	status = 0;
leave:

	return status;
}

2076 2077 2078 2079 2080 2081 2082 2083
/*
 * Convenience function for callers which just want the block number
 * mapped to a name and don't require the full dirent info, etc.
 */
int ocfs2_lookup_ino_from_name(struct inode *dir, const char *name,
			       int namelen, u64 *blkno)
{
	int ret;
2084
	struct ocfs2_dir_lookup_result lookup = { NULL, };
2085

2086 2087
	ret = ocfs2_find_files_on_disk(name, namelen, blkno, dir, &lookup);
	ocfs2_free_dir_lookup_result(&lookup);
2088 2089 2090 2091

	return ret;
}

2092 2093 2094 2095 2096
/* Check for a name within a directory.
 *
 * Return 0 if the name does not exist
 * Return -EEXIST if the directory contains the name
 *
2097
 * Callers should have i_mutex + a cluster lock on dir
2098 2099 2100 2101 2102 2103
 */
int ocfs2_check_dir_for_entry(struct inode *dir,
			      const char *name,
			      int namelen)
{
	int ret;
2104
	struct ocfs2_dir_lookup_result lookup = { NULL, };
2105

2106 2107
	trace_ocfs2_check_dir_for_entry(
		(unsigned long long)OCFS2_I(dir)->ip_blkno, namelen, name);
2108 2109

	ret = -EEXIST;
2110
	if (ocfs2_find_entry(name, namelen, dir, &lookup) == 0)
2111 2112 2113 2114
		goto bail;

	ret = 0;
bail:
2115
	ocfs2_free_dir_lookup_result(&lookup);
2116

T
Tao Ma 已提交
2117 2118
	if (ret)
		mlog_errno(ret);
2119 2120 2121
	return ret;
}

2122 2123 2124 2125
struct ocfs2_empty_dir_priv {
	unsigned seen_dot;
	unsigned seen_dot_dot;
	unsigned seen_other;
2126
	unsigned dx_dir;
2127 2128 2129 2130 2131 2132 2133 2134 2135
};
static int ocfs2_empty_dir_filldir(void *priv, const char *name, int name_len,
				   loff_t pos, u64 ino, unsigned type)
{
	struct ocfs2_empty_dir_priv *p = priv;

	/*
	 * Check the positions of "." and ".." records to be sure
	 * they're in the correct place.
2136 2137 2138 2139 2140 2141 2142
	 *
	 * Indexed directories don't need to proceed past the first
	 * two entries, so we end the scan after seeing '..'. Despite
	 * that, we allow the scan to proceed In the event that we
	 * have a corrupted indexed directory (no dot or dot dot
	 * entries). This allows us to double check for existing
	 * entries which might not have been found in the index.
2143 2144 2145 2146 2147 2148 2149 2150 2151
	 */
	if (name_len == 1 && !strncmp(".", name, 1) && pos == 0) {
		p->seen_dot = 1;
		return 0;
	}

	if (name_len == 2 && !strncmp("..", name, 2) &&
	    pos == OCFS2_DIR_REC_LEN(1)) {
		p->seen_dot_dot = 1;
2152 2153 2154 2155

		if (p->dx_dir && p->seen_dot)
			return 1;

2156 2157 2158 2159 2160 2161
		return 0;
	}

	p->seen_other = 1;
	return 1;
}
2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196

static int ocfs2_empty_dir_dx(struct inode *inode,
			      struct ocfs2_empty_dir_priv *priv)
{
	int ret;
	struct buffer_head *di_bh = NULL;
	struct buffer_head *dx_root_bh = NULL;
	struct ocfs2_dinode *di;
	struct ocfs2_dx_root_block *dx_root;

	priv->dx_dir = 1;

	ret = ocfs2_read_inode_block(inode, &di_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}
	di = (struct ocfs2_dinode *)di_bh->b_data;

	ret = ocfs2_read_dx_root(inode, di, &dx_root_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}
	dx_root = (struct ocfs2_dx_root_block *)dx_root_bh->b_data;

	if (le32_to_cpu(dx_root->dr_num_entries) != 2)
		priv->seen_other = 1;

out:
	brelse(di_bh);
	brelse(dx_root_bh);
	return ret;
}

2197 2198
/*
 * routine to check that the specified directory is empty (for rmdir)
2199 2200
 *
 * Returns 1 if dir is empty, zero otherwise.
2201
 *
2202
 * XXX: This is a performance problem for unindexed directories.
2203 2204 2205
 */
int ocfs2_empty_dir(struct inode *inode)
{
2206 2207 2208
	int ret;
	loff_t start = 0;
	struct ocfs2_empty_dir_priv priv;
2209

2210
	memset(&priv, 0, sizeof(priv));
2211

2212 2213 2214 2215 2216 2217 2218 2219 2220 2221
	if (ocfs2_dir_indexed(inode)) {
		ret = ocfs2_empty_dir_dx(inode, &priv);
		if (ret)
			mlog_errno(ret);
		/*
		 * We still run ocfs2_dir_foreach to get the checks
		 * for "." and "..".
		 */
	}

2222 2223 2224 2225 2226 2227
	ret = ocfs2_dir_foreach(inode, &start, &priv, ocfs2_empty_dir_filldir);
	if (ret)
		mlog_errno(ret);

	if (!priv.seen_dot || !priv.seen_dot_dot) {
		mlog(ML_ERROR, "bad directory (dir #%llu) - no `.' or `..'\n",
2228
		     (unsigned long long)OCFS2_I(inode)->ip_blkno);
2229 2230 2231
		/*
		 * XXX: Is it really safe to allow an unlink to continue?
		 */
2232 2233
		return 1;
	}
2234 2235

	return !priv.seen_other;
2236 2237
}

2238 2239 2240 2241 2242 2243 2244 2245 2246
/*
 * Fills "." and ".." dirents in a new directory block. Returns dirent for
 * "..", which might be used during creation of a directory with a trailing
 * header. It is otherwise safe to ignore the return code.
 */
static struct ocfs2_dir_entry *ocfs2_fill_initial_dirents(struct inode *inode,
							  struct inode *parent,
							  char *start,
							  unsigned int size)
2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262
{
	struct ocfs2_dir_entry *de = (struct ocfs2_dir_entry *)start;

	de->inode = cpu_to_le64(OCFS2_I(inode)->ip_blkno);
	de->name_len = 1;
	de->rec_len =
		cpu_to_le16(OCFS2_DIR_REC_LEN(de->name_len));
	strcpy(de->name, ".");
	ocfs2_set_de_type(de, S_IFDIR);

	de = (struct ocfs2_dir_entry *) ((char *)de + le16_to_cpu(de->rec_len));
	de->inode = cpu_to_le64(OCFS2_I(parent)->ip_blkno);
	de->rec_len = cpu_to_le16(size - OCFS2_DIR_REC_LEN(1));
	de->name_len = 2;
	strcpy(de->name, "..");
	ocfs2_set_de_type(de, S_IFDIR);
2263 2264

	return de;
2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281
}

/*
 * This works together with code in ocfs2_mknod_locked() which sets
 * the inline-data flag and initializes the inline-data section.
 */
static int ocfs2_fill_new_dir_id(struct ocfs2_super *osb,
				 handle_t *handle,
				 struct inode *parent,
				 struct inode *inode,
				 struct buffer_head *di_bh)
{
	int ret;
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
	struct ocfs2_inline_data *data = &di->id2.i_data;
	unsigned int size = le16_to_cpu(data->id_count);

2282
	ret = ocfs2_journal_access_di(handle, INODE_CACHE(inode), di_bh,
2283
				      OCFS2_JOURNAL_ACCESS_WRITE);
2284 2285 2286 2287 2288 2289 2290 2291 2292
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ocfs2_fill_initial_dirents(inode, parent, data->id_data, size);
	ocfs2_journal_dirty(handle, di_bh);

	i_size_write(inode, size);
M
Miklos Szeredi 已提交
2293
	set_nlink(inode, 2);
2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308
	inode->i_blocks = ocfs2_inode_sector_count(inode);

	ret = ocfs2_mark_inode_dirty(handle, inode, di_bh);
	if (ret < 0)
		mlog_errno(ret);

out:
	return ret;
}

static int ocfs2_fill_new_dir_el(struct ocfs2_super *osb,
				 handle_t *handle,
				 struct inode *parent,
				 struct inode *inode,
				 struct buffer_head *fe_bh,
2309 2310
				 struct ocfs2_alloc_context *data_ac,
				 struct buffer_head **ret_new_bh)
2311 2312
{
	int status;
2313
	unsigned int size = osb->sb->s_blocksize;
2314
	struct buffer_head *new_bh = NULL;
2315
	struct ocfs2_dir_entry *de;
2316

2317
	if (ocfs2_new_dir_wants_trailer(inode))
2318 2319
		size = ocfs2_dir_trailer_blk_off(parent->i_sb);

2320 2321 2322 2323 2324 2325 2326
	status = ocfs2_do_extend_dir(osb->sb, handle, inode, fe_bh,
				     data_ac, NULL, &new_bh);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

2327
	ocfs2_set_new_buffer_uptodate(INODE_CACHE(inode), new_bh);
2328

2329
	status = ocfs2_journal_access_db(handle, INODE_CACHE(inode), new_bh,
2330
					 OCFS2_JOURNAL_ACCESS_CREATE);
2331 2332 2333 2334 2335 2336
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}
	memset(new_bh->b_data, 0, osb->sb->s_blocksize);

2337
	de = ocfs2_fill_initial_dirents(inode, parent, new_bh->b_data, size);
2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350
	if (ocfs2_new_dir_wants_trailer(inode)) {
		int size = le16_to_cpu(de->rec_len);

		/*
		 * Figure out the size of the hole left over after
		 * insertion of '.' and '..'. The trailer wants this
		 * information.
		 */
		size -= OCFS2_DIR_REC_LEN(2);
		size -= sizeof(struct ocfs2_dir_block_trailer);

		ocfs2_init_dir_trailer(inode, new_bh, size);
	}
2351

2352
	ocfs2_journal_dirty(handle, new_bh);
2353 2354

	i_size_write(inode, inode->i_sb->s_blocksize);
M
Miklos Szeredi 已提交
2355
	set_nlink(inode, 2);
2356 2357 2358 2359 2360 2361 2362 2363
	inode->i_blocks = ocfs2_inode_sector_count(inode);
	status = ocfs2_mark_inode_dirty(handle, inode, fe_bh);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	status = 0;
2364 2365 2366 2367
	if (ret_new_bh) {
		*ret_new_bh = new_bh;
		new_bh = NULL;
	}
2368
bail:
2369
	brelse(new_bh);
2370 2371 2372 2373

	return status;
}

2374 2375 2376
static int ocfs2_dx_dir_attach_index(struct ocfs2_super *osb,
				     handle_t *handle, struct inode *dir,
				     struct buffer_head *di_bh,
2377
				     struct buffer_head *dirdata_bh,
2378
				     struct ocfs2_alloc_context *meta_ac,
2379
				     int dx_inline, u32 num_entries,
2380
				     struct buffer_head **ret_dx_root_bh)
2381
{
2382 2383 2384
	int ret;
	struct ocfs2_dinode *di = (struct ocfs2_dinode *) di_bh->b_data;
	u16 dr_suballoc_bit;
2385
	u64 suballoc_loc, dr_blkno;
2386 2387 2388
	unsigned int num_bits;
	struct buffer_head *dx_root_bh = NULL;
	struct ocfs2_dx_root_block *dx_root;
2389 2390
	struct ocfs2_dir_block_trailer *trailer =
		ocfs2_trailer_from_bh(dirdata_bh, dir->i_sb);
2391

2392 2393
	ret = ocfs2_claim_metadata(handle, meta_ac, 1, &suballoc_loc,
				   &dr_suballoc_bit, &num_bits, &dr_blkno);
2394 2395 2396 2397
	if (ret) {
		mlog_errno(ret);
		goto out;
	}
2398

2399 2400 2401
	trace_ocfs2_dx_dir_attach_index(
				(unsigned long long)OCFS2_I(dir)->ip_blkno,
				(unsigned long long)dr_blkno);
2402

2403 2404 2405 2406 2407
	dx_root_bh = sb_getblk(osb->sb, dr_blkno);
	if (dx_root_bh == NULL) {
		ret = -EIO;
		goto out;
	}
2408
	ocfs2_set_new_buffer_uptodate(INODE_CACHE(dir), dx_root_bh);
2409

2410
	ret = ocfs2_journal_access_dr(handle, INODE_CACHE(dir), dx_root_bh,
2411 2412 2413 2414 2415
				      OCFS2_JOURNAL_ACCESS_CREATE);
	if (ret < 0) {
		mlog_errno(ret);
		goto out;
	}
2416

2417 2418 2419
	dx_root = (struct ocfs2_dx_root_block *)dx_root_bh->b_data;
	memset(dx_root, 0, osb->sb->s_blocksize);
	strcpy(dx_root->dr_signature, OCFS2_DX_ROOT_SIGNATURE);
2420
	dx_root->dr_suballoc_slot = cpu_to_le16(meta_ac->ac_alloc_slot);
2421
	dx_root->dr_suballoc_loc = cpu_to_le64(suballoc_loc);
2422 2423 2424 2425
	dx_root->dr_suballoc_bit = cpu_to_le16(dr_suballoc_bit);
	dx_root->dr_fs_generation = cpu_to_le32(osb->fs_generation);
	dx_root->dr_blkno = cpu_to_le64(dr_blkno);
	dx_root->dr_dir_blkno = cpu_to_le64(OCFS2_I(dir)->ip_blkno);
2426
	dx_root->dr_num_entries = cpu_to_le32(num_entries);
2427 2428 2429 2430
	if (le16_to_cpu(trailer->db_free_rec_len))
		dx_root->dr_free_blk = cpu_to_le64(dirdata_bh->b_blocknr);
	else
		dx_root->dr_free_blk = cpu_to_le64(0);
2431 2432 2433 2434 2435 2436 2437 2438 2439

	if (dx_inline) {
		dx_root->dr_flags |= OCFS2_DX_FLAG_INLINE;
		dx_root->dr_entries.de_count =
			cpu_to_le16(ocfs2_dx_entries_per_root(osb->sb));
	} else {
		dx_root->dr_list.l_count =
			cpu_to_le16(ocfs2_extent_recs_per_dx_root(osb->sb));
	}
2440
	ocfs2_journal_dirty(handle, dx_root_bh);
2441

2442
	ret = ocfs2_journal_access_di(handle, INODE_CACHE(dir), di_bh,
2443 2444 2445 2446 2447 2448 2449 2450
				      OCFS2_JOURNAL_ACCESS_CREATE);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	di->i_dx_root = cpu_to_le64(dr_blkno);

2451
	spin_lock(&OCFS2_I(dir)->ip_lock);
2452 2453
	OCFS2_I(dir)->ip_dyn_features |= OCFS2_INDEXED_DIR_FL;
	di->i_dyn_features = cpu_to_le16(OCFS2_I(dir)->ip_dyn_features);
2454
	spin_unlock(&OCFS2_I(dir)->ip_lock);
2455

2456
	ocfs2_journal_dirty(handle, di_bh);
2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482

	*ret_dx_root_bh = dx_root_bh;
	dx_root_bh = NULL;

out:
	brelse(dx_root_bh);
	return ret;
}

static int ocfs2_dx_dir_format_cluster(struct ocfs2_super *osb,
				       handle_t *handle, struct inode *dir,
				       struct buffer_head **dx_leaves,
				       int num_dx_leaves, u64 start_blk)
{
	int ret, i;
	struct ocfs2_dx_leaf *dx_leaf;
	struct buffer_head *bh;

	for (i = 0; i < num_dx_leaves; i++) {
		bh = sb_getblk(osb->sb, start_blk + i);
		if (bh == NULL) {
			ret = -EIO;
			goto out;
		}
		dx_leaves[i] = bh;

2483
		ocfs2_set_new_buffer_uptodate(INODE_CACHE(dir), bh);
2484

2485
		ret = ocfs2_journal_access_dl(handle, INODE_CACHE(dir), bh,
2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500
					      OCFS2_JOURNAL_ACCESS_CREATE);
		if (ret < 0) {
			mlog_errno(ret);
			goto out;
		}

		dx_leaf = (struct ocfs2_dx_leaf *) bh->b_data;

		memset(dx_leaf, 0, osb->sb->s_blocksize);
		strcpy(dx_leaf->dl_signature, OCFS2_DX_LEAF_SIGNATURE);
		dx_leaf->dl_fs_generation = cpu_to_le32(osb->fs_generation);
		dx_leaf->dl_blkno = cpu_to_le64(bh->b_blocknr);
		dx_leaf->dl_list.de_count =
			cpu_to_le16(ocfs2_dx_entries_per_leaf(osb->sb));

2501 2502 2503 2504
		trace_ocfs2_dx_dir_format_cluster(
				(unsigned long long)OCFS2_I(dir)->ip_blkno,
				(unsigned long long)bh->b_blocknr,
				le16_to_cpu(dx_leaf->dl_list.de_count));
2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535

		ocfs2_journal_dirty(handle, bh);
	}

	ret = 0;
out:
	return ret;
}

/*
 * Allocates and formats a new cluster for use in an indexed dir
 * leaf. This version will not do the extent insert, so that it can be
 * used by operations which need careful ordering.
 */
static int __ocfs2_dx_dir_new_cluster(struct inode *dir,
				      u32 cpos, handle_t *handle,
				      struct ocfs2_alloc_context *data_ac,
				      struct buffer_head **dx_leaves,
				      int num_dx_leaves, u64 *ret_phys_blkno)
{
	int ret;
	u32 phys, num;
	u64 phys_blkno;
	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);

	/*
	 * XXX: For create, this should claim cluster for the index
	 * *before* the unindexed insert so that we have a better
	 * chance of contiguousness as the directory grows in number
	 * of entries.
	 */
2536
	ret = __ocfs2_claim_clusters(handle, data_ac, 1, 1, &phys, &num);
2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	/*
	 * Format the new cluster first. That way, we're inserting
	 * valid data.
	 */
	phys_blkno = ocfs2_clusters_to_blocks(osb->sb, phys);
	ret = ocfs2_dx_dir_format_cluster(osb, handle, dir, dx_leaves,
					  num_dx_leaves, phys_blkno);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	*ret_phys_blkno = phys_blkno;
out:
	return ret;
}

static int ocfs2_dx_dir_new_cluster(struct inode *dir,
				    struct ocfs2_extent_tree *et,
				    u32 cpos, handle_t *handle,
				    struct ocfs2_alloc_context *data_ac,
				    struct ocfs2_alloc_context *meta_ac,
				    struct buffer_head **dx_leaves,
				    int num_dx_leaves)
{
	int ret;
	u64 phys_blkno;

	ret = __ocfs2_dx_dir_new_cluster(dir, cpos, handle, data_ac, dx_leaves,
					 num_dx_leaves, &phys_blkno);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

2577
	ret = ocfs2_insert_extent(handle, et, cpos, phys_blkno, 1, 0,
2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606
				  meta_ac);
	if (ret)
		mlog_errno(ret);
out:
	return ret;
}

static struct buffer_head **ocfs2_dx_dir_kmalloc_leaves(struct super_block *sb,
							int *ret_num_leaves)
{
	int num_dx_leaves = ocfs2_clusters_to_blocks(sb, 1);
	struct buffer_head **dx_leaves;

	dx_leaves = kcalloc(num_dx_leaves, sizeof(struct buffer_head *),
			    GFP_NOFS);
	if (dx_leaves && ret_num_leaves)
		*ret_num_leaves = num_dx_leaves;

	return dx_leaves;
}

static int ocfs2_fill_new_dir_dx(struct ocfs2_super *osb,
				 handle_t *handle,
				 struct inode *parent,
				 struct inode *inode,
				 struct buffer_head *di_bh,
				 struct ocfs2_alloc_context *data_ac,
				 struct ocfs2_alloc_context *meta_ac)
{
2607
	int ret;
2608 2609 2610
	struct buffer_head *leaf_bh = NULL;
	struct buffer_head *dx_root_bh = NULL;
	struct ocfs2_dx_hinfo hinfo;
2611 2612
	struct ocfs2_dx_root_block *dx_root;
	struct ocfs2_dx_entry_list *entry_list;
2613 2614 2615 2616 2617 2618 2619 2620

	/*
	 * Our strategy is to create the directory as though it were
	 * unindexed, then add the index block. This works with very
	 * little complication since the state of a new directory is a
	 * very well known quantity.
	 *
	 * Essentially, we have two dirents ("." and ".."), in the 1st
2621 2622
	 * block which need indexing. These are easily inserted into
	 * the index block.
2623 2624 2625 2626 2627 2628 2629 2630 2631
	 */

	ret = ocfs2_fill_new_dir_el(osb, handle, parent, inode, di_bh,
				    data_ac, &leaf_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

2632
	ret = ocfs2_dx_dir_attach_index(osb, handle, inode, di_bh, leaf_bh,
2633
					meta_ac, 1, 2, &dx_root_bh);
2634 2635 2636 2637
	if (ret) {
		mlog_errno(ret);
		goto out;
	}
2638 2639
	dx_root = (struct ocfs2_dx_root_block *)dx_root_bh->b_data;
	entry_list = &dx_root->dr_entries;
2640

2641
	/* Buffer has been journaled for us by ocfs2_dx_dir_attach_index */
2642
	ocfs2_dx_dir_name_hash(inode, ".", 1, &hinfo);
2643
	ocfs2_dx_entry_list_insert(entry_list, &hinfo, leaf_bh->b_blocknr);
2644 2645

	ocfs2_dx_dir_name_hash(inode, "..", 2, &hinfo);
2646
	ocfs2_dx_entry_list_insert(entry_list, &hinfo, leaf_bh->b_blocknr);
2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679

out:
	brelse(dx_root_bh);
	brelse(leaf_bh);
	return ret;
}

int ocfs2_fill_new_dir(struct ocfs2_super *osb,
		       handle_t *handle,
		       struct inode *parent,
		       struct inode *inode,
		       struct buffer_head *fe_bh,
		       struct ocfs2_alloc_context *data_ac,
		       struct ocfs2_alloc_context *meta_ac)

{
	BUG_ON(!ocfs2_supports_inline_data(osb) && data_ac == NULL);

	if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL)
		return ocfs2_fill_new_dir_id(osb, handle, parent, inode, fe_bh);

	if (ocfs2_supports_indexed_dirs(osb))
		return ocfs2_fill_new_dir_dx(osb, handle, parent, inode, fe_bh,
					     data_ac, meta_ac);

	return ocfs2_fill_new_dir_el(osb, handle, parent, inode, fe_bh,
				     data_ac, NULL);
}

static int ocfs2_dx_dir_index_block(struct inode *dir,
				    handle_t *handle,
				    struct buffer_head **dx_leaves,
				    int num_dx_leaves,
2680
				    u32 *num_dx_entries,
2681 2682
				    struct buffer_head *dirent_bh)
{
T
Tao Ma 已提交
2683
	int ret = 0, namelen, i;
2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711
	char *de_buf, *limit;
	struct ocfs2_dir_entry *de;
	struct buffer_head *dx_leaf_bh;
	struct ocfs2_dx_hinfo hinfo;
	u64 dirent_blk = dirent_bh->b_blocknr;

	de_buf = dirent_bh->b_data;
	limit = de_buf + dir->i_sb->s_blocksize;

	while (de_buf < limit) {
		de = (struct ocfs2_dir_entry *)de_buf;

		namelen = de->name_len;
		if (!namelen || !de->inode)
			goto inc;

		ocfs2_dx_dir_name_hash(dir, de->name, namelen, &hinfo);

		i = ocfs2_dx_dir_hash_idx(OCFS2_SB(dir->i_sb), &hinfo);
		dx_leaf_bh = dx_leaves[i];

		ret = __ocfs2_dx_dir_leaf_insert(dir, handle, &hinfo,
						 dirent_blk, dx_leaf_bh);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

2712 2713
		*num_dx_entries = *num_dx_entries + 1;

2714 2715 2716 2717 2718 2719 2720
inc:
		de_buf += le16_to_cpu(de->rec_len);
	}

out:
	return ret;
}
2721 2722

/*
2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747
 * XXX: This expects dx_root_bh to already be part of the transaction.
 */
static void ocfs2_dx_dir_index_root_block(struct inode *dir,
					 struct buffer_head *dx_root_bh,
					 struct buffer_head *dirent_bh)
{
	char *de_buf, *limit;
	struct ocfs2_dx_root_block *dx_root;
	struct ocfs2_dir_entry *de;
	struct ocfs2_dx_hinfo hinfo;
	u64 dirent_blk = dirent_bh->b_blocknr;

	dx_root = (struct ocfs2_dx_root_block *)dx_root_bh->b_data;

	de_buf = dirent_bh->b_data;
	limit = de_buf + dir->i_sb->s_blocksize;

	while (de_buf < limit) {
		de = (struct ocfs2_dir_entry *)de_buf;

		if (!de->name_len || !de->inode)
			goto inc;

		ocfs2_dx_dir_name_hash(dir, de->name, de->name_len, &hinfo);

2748 2749 2750 2751 2752
		trace_ocfs2_dx_dir_index_root_block(
				(unsigned long long)dir->i_ino,
				hinfo.major_hash, hinfo.minor_hash,
				de->name_len, de->name,
				le16_to_cpu(dx_root->dr_entries.de_num_used));
2753 2754 2755

		ocfs2_dx_entry_list_insert(&dx_root->dr_entries, &hinfo,
					   dirent_blk);
2756 2757

		le32_add_cpu(&dx_root->dr_num_entries, 1);
2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790
inc:
		de_buf += le16_to_cpu(de->rec_len);
	}
}

/*
 * Count the number of inline directory entries in di_bh and compare
 * them against the number of entries we can hold in an inline dx root
 * block.
 */
static int ocfs2_new_dx_should_be_inline(struct inode *dir,
					 struct buffer_head *di_bh)
{
	int dirent_count = 0;
	char *de_buf, *limit;
	struct ocfs2_dir_entry *de;
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;

	de_buf = di->id2.i_data.id_data;
	limit = de_buf + i_size_read(dir);

	while (de_buf < limit) {
		de = (struct ocfs2_dir_entry *)de_buf;

		if (de->name_len && de->inode)
			dirent_count++;

		de_buf += le16_to_cpu(de->rec_len);
	}

	/* We are careful to leave room for one extra record. */
	return dirent_count < ocfs2_dx_entries_per_root(dir->i_sb);
}
2791 2792 2793 2794 2795 2796 2797

/*
 * Expand rec_len of the rightmost dirent in a directory block so that it
 * contains the end of our valid space for dirents. We do this during
 * expansion from an inline directory to one with extents. The first dir block
 * in that case is taken from the inline data portion of the inode block.
 *
2798 2799 2800 2801 2802 2803
 * This will also return the largest amount of contiguous space for a dirent
 * in the block. That value is *not* necessarily the last dirent, even after
 * expansion. The directory indexing code wants this value for free space
 * accounting. We do this here since we're already walking the entire dir
 * block.
 *
2804 2805
 * We add the dir trailer if this filesystem wants it.
 */
2806 2807
static unsigned int ocfs2_expand_last_dirent(char *start, unsigned int old_size,
					     struct inode *dir)
2808
{
2809
	struct super_block *sb = dir->i_sb;
2810 2811 2812 2813
	struct ocfs2_dir_entry *de;
	struct ocfs2_dir_entry *prev_de;
	char *de_buf, *limit;
	unsigned int new_size = sb->s_blocksize;
2814 2815
	unsigned int bytes, this_hole;
	unsigned int largest_hole = 0;
2816

2817
	if (ocfs2_new_dir_wants_trailer(dir))
2818 2819 2820 2821 2822 2823 2824
		new_size = ocfs2_dir_trailer_blk_off(sb);

	bytes = new_size - old_size;

	limit = start + old_size;
	de_buf = start;
	de = (struct ocfs2_dir_entry *)de_buf;
2825
	do {
2826 2827 2828 2829
		this_hole = ocfs2_figure_dirent_hole(de);
		if (this_hole > largest_hole)
			largest_hole = this_hole;

2830 2831 2832 2833 2834 2835
		prev_de = de;
		de_buf += le16_to_cpu(de->rec_len);
		de = (struct ocfs2_dir_entry *)de_buf;
	} while (de_buf < limit);

	le16_add_cpu(&prev_de->rec_len, bytes);
2836 2837 2838 2839 2840 2841 2842 2843 2844 2845

	/* We need to double check this after modification of the final
	 * dirent. */
	this_hole = ocfs2_figure_dirent_hole(prev_de);
	if (this_hole > largest_hole)
		largest_hole = this_hole;

	if (largest_hole >= OCFS2_DIR_MIN_REC_LEN)
		return largest_hole;
	return 0;
2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857
}

/*
 * We allocate enough clusters to fulfill "blocks_wanted", but set
 * i_size to exactly one block. Ocfs2_extend_dir() will handle the
 * rest automatically for us.
 *
 * *first_block_bh is a pointer to the 1st data block allocated to the
 *  directory.
 */
static int ocfs2_expand_inline_dir(struct inode *dir, struct buffer_head *di_bh,
				   unsigned int blocks_wanted,
2858
				   struct ocfs2_dir_lookup_result *lookup,
2859 2860
				   struct buffer_head **first_block_bh)
{
2861
	u32 alloc, dx_alloc, bit_off, len, num_dx_entries = 0;
2862
	struct super_block *sb = dir->i_sb;
2863
	int ret, i, num_dx_leaves = 0, dx_inline = 0,
2864 2865 2866
		credits = ocfs2_inline_to_extents_credits(sb);
	u64 dx_insert_blkno, blkno,
		bytes = blocks_wanted << sb->s_blocksize_bits;
2867 2868
	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
	struct ocfs2_inode_info *oi = OCFS2_I(dir);
2869
	struct ocfs2_alloc_context *data_ac = NULL;
2870
	struct ocfs2_alloc_context *meta_ac = NULL;
2871
	struct buffer_head *dirdata_bh = NULL;
2872 2873
	struct buffer_head *dx_root_bh = NULL;
	struct buffer_head **dx_leaves = NULL;
2874 2875
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
	handle_t *handle;
2876
	struct ocfs2_extent_tree et;
2877 2878
	struct ocfs2_extent_tree dx_et;
	int did_quota = 0, bytes_allocated = 0;
2879

2880
	ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(dir), di_bh);
2881 2882

	alloc = ocfs2_clusters_for_bytes(sb, bytes);
2883 2884
	dx_alloc = 0;

2885 2886
	down_write(&oi->ip_alloc_sem);

2887 2888 2889
	if (ocfs2_supports_indexed_dirs(osb)) {
		credits += ocfs2_add_dir_index_credits(sb);

2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900
		dx_inline = ocfs2_new_dx_should_be_inline(dir, di_bh);
		if (!dx_inline) {
			/* Add one more cluster for an index leaf */
			dx_alloc++;
			dx_leaves = ocfs2_dx_dir_kmalloc_leaves(sb,
								&num_dx_leaves);
			if (!dx_leaves) {
				ret = -ENOMEM;
				mlog_errno(ret);
				goto out;
			}
2901 2902 2903 2904 2905 2906 2907 2908 2909
		}

		/* This gets us the dx_root */
		ret = ocfs2_reserve_new_metadata_blocks(osb, 1, &meta_ac);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
	}
2910 2911

	/*
2912 2913 2914
	 * We should never need more than 2 clusters for the unindexed
	 * tree - maximum dirent size is far less than one block. In
	 * fact, the only time we'd need more than one cluster is if
2915 2916 2917 2918 2919 2920
	 * blocksize == clustersize and the dirent won't fit in the
	 * extra space that the expansion to a single block gives. As
	 * of today, that only happens on 4k/4k file systems.
	 */
	BUG_ON(alloc > 2);

2921
	ret = ocfs2_reserve_clusters(osb, alloc + dx_alloc, &data_ac);
2922 2923 2924 2925 2926 2927
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	/*
J
Joe Perches 已提交
2928
	 * Prepare for worst case allocation scenario of two separate
2929
	 * extents in the unindexed tree.
2930 2931 2932 2933 2934 2935 2936 2937
	 */
	if (alloc == 2)
		credits += OCFS2_SUBALLOC_ALLOC;

	handle = ocfs2_start_trans(osb, credits);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		mlog_errno(ret);
2938
		goto out;
2939 2940
	}

2941 2942 2943
	ret = dquot_alloc_space_nodirty(dir,
		ocfs2_clusters_to_bytes(osb->sb, alloc + dx_alloc));
	if (ret)
2944 2945
		goto out_commit;
	did_quota = 1;
2946

2947
	if (ocfs2_supports_indexed_dirs(osb) && !dx_inline) {
2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962
		/*
		 * Allocate our index cluster first, to maximize the
		 * possibility that unindexed leaves grow
		 * contiguously.
		 */
		ret = __ocfs2_dx_dir_new_cluster(dir, 0, handle, data_ac,
						 dx_leaves, num_dx_leaves,
						 &dx_insert_blkno);
		if (ret) {
			mlog_errno(ret);
			goto out_commit;
		}
		bytes_allocated += ocfs2_clusters_to_bytes(dir->i_sb, 1);
	}

2963 2964 2965 2966 2967
	/*
	 * Try to claim as many clusters as the bitmap can give though
	 * if we only get one now, that's enough to continue. The rest
	 * will be claimed after the conversion to extents.
	 */
2968 2969
	if (ocfs2_dir_resv_allowed(osb))
		data_ac->ac_resv = &oi->ip_la_data_resv;
2970
	ret = ocfs2_claim_clusters(handle, data_ac, 1, &bit_off, &len);
2971 2972 2973 2974
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}
2975
	bytes_allocated += ocfs2_clusters_to_bytes(dir->i_sb, 1);
2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989

	/*
	 * Operations are carefully ordered so that we set up the new
	 * data block first. The conversion from inline data to
	 * extents follows.
	 */
	blkno = ocfs2_clusters_to_blocks(dir->i_sb, bit_off);
	dirdata_bh = sb_getblk(sb, blkno);
	if (!dirdata_bh) {
		ret = -EIO;
		mlog_errno(ret);
		goto out_commit;
	}

2990
	ocfs2_set_new_buffer_uptodate(INODE_CACHE(dir), dirdata_bh);
2991

2992
	ret = ocfs2_journal_access_db(handle, INODE_CACHE(dir), dirdata_bh,
2993
				      OCFS2_JOURNAL_ACCESS_CREATE);
2994 2995 2996 2997 2998 2999 3000 3001
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	memcpy(dirdata_bh->b_data, di->id2.i_data.id_data, i_size_read(dir));
	memset(dirdata_bh->b_data + i_size_read(dir), 0,
	       sb->s_blocksize - i_size_read(dir));
3002 3003 3004 3005 3006 3007 3008 3009 3010 3011
	i = ocfs2_expand_last_dirent(dirdata_bh->b_data, i_size_read(dir), dir);
	if (ocfs2_new_dir_wants_trailer(dir)) {
		/*
		 * Prepare the dir trailer up front. It will otherwise look
		 * like a valid dirent. Even if inserting the index fails
		 * (unlikely), then all we'll have done is given first dir
		 * block a small amount of fragmentation.
		 */
		ocfs2_init_dir_trailer(dir, dirdata_bh, i);
	}
3012

3013
	ocfs2_journal_dirty(handle, dirdata_bh);
3014

3015 3016 3017 3018
	if (ocfs2_supports_indexed_dirs(osb) && !dx_inline) {
		/*
		 * Dx dirs with an external cluster need to do this up
		 * front. Inline dx root's get handled later, after
3019 3020 3021 3022
		 * we've allocated our root block. We get passed back
		 * a total number of items so that dr_num_entries can
		 * be correctly set once the dx_root has been
		 * allocated.
3023
		 */
3024
		ret = ocfs2_dx_dir_index_block(dir, handle, dx_leaves,
3025 3026
					       num_dx_leaves, &num_dx_entries,
					       dirdata_bh);
3027 3028 3029 3030 3031 3032
		if (ret) {
			mlog_errno(ret);
			goto out_commit;
		}
	}

3033 3034 3035 3036 3037 3038 3039 3040
	/*
	 * Set extent, i_size, etc on the directory. After this, the
	 * inode should contain the same exact dirents as before and
	 * be fully accessible from system calls.
	 *
	 * We let the later dirent insert modify c/mtime - to the user
	 * the data hasn't changed.
	 */
3041
	ret = ocfs2_journal_access_di(handle, INODE_CACHE(dir), di_bh,
3042
				      OCFS2_JOURNAL_ACCESS_CREATE);
3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	spin_lock(&oi->ip_lock);
	oi->ip_dyn_features &= ~OCFS2_INLINE_DATA_FL;
	di->i_dyn_features = cpu_to_le16(oi->ip_dyn_features);
	spin_unlock(&oi->ip_lock);

	ocfs2_dinode_new_extent_list(dir, di);

	i_size_write(dir, sb->s_blocksize);
	dir->i_mtime = dir->i_ctime = CURRENT_TIME;

	di->i_size = cpu_to_le64(sb->s_blocksize);
	di->i_ctime = di->i_mtime = cpu_to_le64(dir->i_ctime.tv_sec);
	di->i_ctime_nsec = di->i_mtime_nsec = cpu_to_le32(dir->i_ctime.tv_nsec);

	/*
	 * This should never fail as our extent list is empty and all
	 * related blocks have been journaled already.
	 */
3066
	ret = ocfs2_insert_extent(handle, &et, 0, blkno, len,
3067
				  0, NULL);
3068 3069
	if (ret) {
		mlog_errno(ret);
3070
		goto out_commit;
3071 3072
	}

3073 3074 3075 3076 3077 3078
	/*
	 * Set i_blocks after the extent insert for the most up to
	 * date ip_clusters value.
	 */
	dir->i_blocks = ocfs2_inode_sector_count(dir);

3079
	ocfs2_journal_dirty(handle, di_bh);
3080

3081 3082
	if (ocfs2_supports_indexed_dirs(osb)) {
		ret = ocfs2_dx_dir_attach_index(osb, handle, dir, di_bh,
3083
						dirdata_bh, meta_ac, dx_inline,
3084
						num_dx_entries, &dx_root_bh);
3085 3086 3087 3088 3089
		if (ret) {
			mlog_errno(ret);
			goto out_commit;
		}

3090 3091 3092 3093
		if (dx_inline) {
			ocfs2_dx_dir_index_root_block(dir, dx_root_bh,
						      dirdata_bh);
		} else {
3094 3095 3096
			ocfs2_init_dx_root_extent_tree(&dx_et,
						       INODE_CACHE(dir),
						       dx_root_bh);
3097
			ret = ocfs2_insert_extent(handle, &dx_et, 0,
3098 3099 3100 3101
						  dx_insert_blkno, 1, 0, NULL);
			if (ret)
				mlog_errno(ret);
		}
3102 3103
	}

3104 3105 3106 3107 3108
	/*
	 * We asked for two clusters, but only got one in the 1st
	 * pass. Claim the 2nd cluster as a separate extent.
	 */
	if (alloc > len) {
3109
		ret = ocfs2_claim_clusters(handle, data_ac, 1, &bit_off,
3110 3111 3112 3113 3114 3115 3116
					   &len);
		if (ret) {
			mlog_errno(ret);
			goto out_commit;
		}
		blkno = ocfs2_clusters_to_blocks(dir->i_sb, bit_off);

3117
		ret = ocfs2_insert_extent(handle, &et, 1,
3118
					  blkno, len, 0, NULL);
3119 3120
		if (ret) {
			mlog_errno(ret);
3121
			goto out_commit;
3122
		}
3123
		bytes_allocated += ocfs2_clusters_to_bytes(dir->i_sb, 1);
3124 3125 3126 3127
	}

	*first_block_bh = dirdata_bh;
	dirdata_bh = NULL;
3128 3129 3130
	if (ocfs2_supports_indexed_dirs(osb)) {
		unsigned int off;

3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142
		if (!dx_inline) {
			/*
			 * We need to return the correct block within the
			 * cluster which should hold our entry.
			 */
			off = ocfs2_dx_dir_hash_idx(OCFS2_SB(dir->i_sb),
						    &lookup->dl_hinfo);
			get_bh(dx_leaves[off]);
			lookup->dl_dx_leaf_bh = dx_leaves[off];
		}
		lookup->dl_dx_root_bh = dx_root_bh;
		dx_root_bh = NULL;
3143
	}
3144 3145

out_commit:
3146
	if (ret < 0 && did_quota)
3147
		dquot_free_space_nodirty(dir, bytes_allocated);
3148

3149 3150 3151
	ocfs2_commit_trans(osb, handle);

out:
3152
	up_write(&oi->ip_alloc_sem);
3153 3154
	if (data_ac)
		ocfs2_free_alloc_context(data_ac);
3155 3156 3157 3158 3159 3160 3161 3162
	if (meta_ac)
		ocfs2_free_alloc_context(meta_ac);

	if (dx_leaves) {
		for (i = 0; i < num_dx_leaves; i++)
			brelse(dx_leaves[i]);
		kfree(dx_leaves);
	}
3163 3164

	brelse(dirdata_bh);
3165
	brelse(dx_root_bh);
3166 3167 3168 3169

	return ret;
}

3170
/* returns a bh of the 1st new block in the allocation. */
3171 3172 3173 3174 3175 3176 3177
static int ocfs2_do_extend_dir(struct super_block *sb,
			       handle_t *handle,
			       struct inode *dir,
			       struct buffer_head *parent_fe_bh,
			       struct ocfs2_alloc_context *data_ac,
			       struct ocfs2_alloc_context *meta_ac,
			       struct buffer_head **new_bh)
3178 3179
{
	int status;
3180
	int extend, did_quota = 0;
3181
	u64 p_blkno, v_blkno;
3182 3183 3184 3185 3186 3187

	spin_lock(&OCFS2_I(dir)->ip_lock);
	extend = (i_size_read(dir) == ocfs2_clusters_to_bytes(sb, OCFS2_I(dir)->ip_clusters));
	spin_unlock(&OCFS2_I(dir)->ip_lock);

	if (extend) {
M
Mark Fasheh 已提交
3188 3189
		u32 offset = OCFS2_I(dir)->ip_clusters;

3190 3191 3192
		status = dquot_alloc_space_nodirty(dir,
					ocfs2_clusters_to_bytes(sb, 1));
		if (status)
3193 3194 3195
			goto bail;
		did_quota = 1;

3196 3197 3198
		status = ocfs2_add_inode_data(OCFS2_SB(sb), dir, &offset,
					      1, 0, parent_fe_bh, handle,
					      data_ac, meta_ac, NULL);
3199 3200 3201 3202 3203 3204 3205
		BUG_ON(status == -EAGAIN);
		if (status < 0) {
			mlog_errno(status);
			goto bail;
		}
	}

3206 3207
	v_blkno = ocfs2_blocks_for_bytes(sb, i_size_read(dir));
	status = ocfs2_extent_map_get_blocks(dir, v_blkno, &p_blkno, NULL, NULL);
3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

	*new_bh = sb_getblk(sb, p_blkno);
	if (!*new_bh) {
		status = -EIO;
		mlog_errno(status);
		goto bail;
	}
	status = 0;
bail:
3221
	if (did_quota && status < 0)
3222
		dquot_free_space_nodirty(dir, ocfs2_clusters_to_bytes(sb, 1));
3223 3224 3225
	return status;
}

3226 3227 3228 3229 3230 3231 3232
/*
 * Assumes you already have a cluster lock on the directory.
 *
 * 'blocks_wanted' is only used if we have an inline directory which
 * is to be turned into an extent based one. The size of the dirent to
 * insert might be larger than the space gained by growing to just one
 * block, so we may have to grow the inode by two blocks in that case.
3233 3234
 *
 * If the directory is already indexed, dx_root_bh must be provided.
3235
 */
3236 3237 3238
static int ocfs2_extend_dir(struct ocfs2_super *osb,
			    struct inode *dir,
			    struct buffer_head *parent_fe_bh,
3239
			    unsigned int blocks_wanted,
3240
			    struct ocfs2_dir_lookup_result *lookup,
3241 3242 3243
			    struct buffer_head **new_de_bh)
{
	int status = 0;
3244
	int credits, num_free_extents, drop_alloc_sem = 0;
3245 3246
	loff_t dir_i_size;
	struct ocfs2_dinode *fe = (struct ocfs2_dinode *) parent_fe_bh->b_data;
3247
	struct ocfs2_extent_list *el = &fe->id2.i_list;
3248 3249
	struct ocfs2_alloc_context *data_ac = NULL;
	struct ocfs2_alloc_context *meta_ac = NULL;
3250
	handle_t *handle = NULL;
3251 3252 3253
	struct buffer_head *new_bh = NULL;
	struct ocfs2_dir_entry * de;
	struct super_block *sb = osb->sb;
3254
	struct ocfs2_extent_tree et;
3255
	struct buffer_head *dx_root_bh = lookup->dl_dx_root_bh;
3256

3257
	if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL) {
3258 3259 3260 3261 3262 3263
		/*
		 * This would be a code error as an inline directory should
		 * never have an index root.
		 */
		BUG_ON(dx_root_bh);

3264
		status = ocfs2_expand_inline_dir(dir, parent_fe_bh,
3265 3266
						 blocks_wanted, lookup,
						 &new_bh);
3267 3268 3269 3270 3271
		if (status) {
			mlog_errno(status);
			goto bail;
		}

3272 3273 3274 3275
		/* Expansion from inline to an indexed directory will
		 * have given us this. */
		dx_root_bh = lookup->dl_dx_root_bh;

3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294
		if (blocks_wanted == 1) {
			/*
			 * If the new dirent will fit inside the space
			 * created by pushing out to one block, then
			 * we can complete the operation
			 * here. Otherwise we have to expand i_size
			 * and format the 2nd block below.
			 */
			BUG_ON(new_bh == NULL);
			goto bail_bh;
		}

		/*
		 * Get rid of 'new_bh' - we want to format the 2nd
		 * data block and return that instead.
		 */
		brelse(new_bh);
		new_bh = NULL;

3295 3296
		down_write(&OCFS2_I(dir)->ip_alloc_sem);
		drop_alloc_sem = 1;
3297 3298 3299 3300 3301
		dir_i_size = i_size_read(dir);
		credits = OCFS2_SIMPLE_DIR_EXTEND_CREDITS;
		goto do_extend;
	}

3302 3303
	down_write(&OCFS2_I(dir)->ip_alloc_sem);
	drop_alloc_sem = 1;
3304
	dir_i_size = i_size_read(dir);
3305 3306
	trace_ocfs2_extend_dir((unsigned long long)OCFS2_I(dir)->ip_blkno,
			       dir_i_size);
3307 3308 3309 3310 3311

	/* dir->i_size is always block aligned. */
	spin_lock(&OCFS2_I(dir)->ip_lock);
	if (dir_i_size == ocfs2_clusters_to_bytes(sb, OCFS2_I(dir)->ip_clusters)) {
		spin_unlock(&OCFS2_I(dir)->ip_lock);
3312 3313
		ocfs2_init_dinode_extent_tree(&et, INODE_CACHE(dir),
					      parent_fe_bh);
3314
		num_free_extents = ocfs2_num_free_extents(osb, &et);
3315 3316 3317 3318 3319 3320 3321
		if (num_free_extents < 0) {
			status = num_free_extents;
			mlog_errno(status);
			goto bail;
		}

		if (!num_free_extents) {
3322
			status = ocfs2_reserve_new_metadata(osb, el, &meta_ac);
3323 3324 3325 3326 3327 3328 3329
			if (status < 0) {
				if (status != -ENOSPC)
					mlog_errno(status);
				goto bail;
			}
		}

3330
		status = ocfs2_reserve_clusters(osb, 1, &data_ac);
3331 3332 3333 3334 3335 3336
		if (status < 0) {
			if (status != -ENOSPC)
				mlog_errno(status);
			goto bail;
		}

3337 3338
		if (ocfs2_dir_resv_allowed(osb))
			data_ac->ac_resv = &OCFS2_I(dir)->ip_la_data_resv;
3339

3340
		credits = ocfs2_calc_extend_credits(sb, el, 1);
3341 3342 3343 3344 3345
	} else {
		spin_unlock(&OCFS2_I(dir)->ip_lock);
		credits = OCFS2_SIMPLE_DIR_EXTEND_CREDITS;
	}

3346
do_extend:
3347 3348 3349 3350
	if (ocfs2_dir_indexed(dir))
		credits++; /* For attaching the new dirent block to the
			    * dx_root */

3351
	handle = ocfs2_start_trans(osb, credits);
3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365
	if (IS_ERR(handle)) {
		status = PTR_ERR(handle);
		handle = NULL;
		mlog_errno(status);
		goto bail;
	}

	status = ocfs2_do_extend_dir(osb->sb, handle, dir, parent_fe_bh,
				     data_ac, meta_ac, &new_bh);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

3366
	ocfs2_set_new_buffer_uptodate(INODE_CACHE(dir), new_bh);
3367

3368
	status = ocfs2_journal_access_db(handle, INODE_CACHE(dir), new_bh,
3369
					 OCFS2_JOURNAL_ACCESS_CREATE);
3370 3371 3372 3373 3374
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}
	memset(new_bh->b_data, 0, sb->s_blocksize);
3375

3376 3377
	de = (struct ocfs2_dir_entry *) new_bh->b_data;
	de->inode = 0;
3378
	if (ocfs2_supports_dir_trailer(dir)) {
3379
		de->rec_len = cpu_to_le16(ocfs2_dir_trailer_blk_off(sb));
3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390

		ocfs2_init_dir_trailer(dir, new_bh, le16_to_cpu(de->rec_len));

		if (ocfs2_dir_indexed(dir)) {
			status = ocfs2_dx_dir_link_trailer(dir, handle,
							   dx_root_bh, new_bh);
			if (status) {
				mlog_errno(status);
				goto bail;
			}
		}
3391 3392 3393
	} else {
		de->rec_len = cpu_to_le16(sb->s_blocksize);
	}
3394
	ocfs2_journal_dirty(handle, new_bh);
3395 3396 3397

	dir_i_size += dir->i_sb->s_blocksize;
	i_size_write(dir, dir_i_size);
3398
	dir->i_blocks = ocfs2_inode_sector_count(dir);
3399 3400 3401 3402 3403 3404
	status = ocfs2_mark_inode_dirty(handle, dir, parent_fe_bh);
	if (status < 0) {
		mlog_errno(status);
		goto bail;
	}

3405
bail_bh:
3406 3407 3408 3409
	*new_de_bh = new_bh;
	get_bh(*new_de_bh);
bail:
	if (handle)
3410
		ocfs2_commit_trans(osb, handle);
3411 3412
	if (drop_alloc_sem)
		up_write(&OCFS2_I(dir)->ip_alloc_sem);
3413 3414 3415 3416 3417 3418

	if (data_ac)
		ocfs2_free_alloc_context(data_ac);
	if (meta_ac)
		ocfs2_free_alloc_context(meta_ac);

3419
	brelse(new_bh);
3420 3421 3422 3423

	return status;
}

3424 3425 3426 3427
static int ocfs2_find_dir_space_id(struct inode *dir, struct buffer_head *di_bh,
				   const char *name, int namelen,
				   struct buffer_head **ret_de_bh,
				   unsigned int *blocks_wanted)
3428
{
3429
	int ret;
3430
	struct super_block *sb = dir->i_sb;
3431 3432 3433 3434
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
	struct ocfs2_dir_entry *de, *last_de = NULL;
	char *de_buf, *limit;
	unsigned long offset = 0;
3435 3436 3437 3438 3439 3440
	unsigned int rec_len, new_rec_len, free_space = dir->i_sb->s_blocksize;

	/*
	 * This calculates how many free bytes we'd have in block zero, should
	 * this function force expansion to an extent tree.
	 */
3441
	if (ocfs2_new_dir_wants_trailer(dir))
3442 3443 3444
		free_space = ocfs2_dir_trailer_blk_off(sb) - i_size_read(dir);
	else
		free_space = dir->i_sb->s_blocksize - i_size_read(dir);
3445 3446 3447 3448

	de_buf = di->id2.i_data.id_data;
	limit = de_buf + i_size_read(dir);
	rec_len = OCFS2_DIR_REC_LEN(namelen);
3449

3450 3451
	while (de_buf < limit) {
		de = (struct ocfs2_dir_entry *)de_buf;
3452

3453 3454 3455 3456 3457 3458 3459 3460
		if (!ocfs2_check_dir_entry(dir, de, di_bh, offset)) {
			ret = -ENOENT;
			goto out;
		}
		if (ocfs2_match(namelen, name, de)) {
			ret = -EEXIST;
			goto out;
		}
3461 3462 3463 3464 3465
		/*
		 * No need to check for a trailing dirent record here as
		 * they're not used for inline dirs.
		 */

3466 3467 3468 3469 3470 3471 3472 3473
		if (ocfs2_dirent_would_fit(de, rec_len)) {
			/* Ok, we found a spot. Return this bh and let
			 * the caller actually fill it in. */
			*ret_de_bh = di_bh;
			get_bh(*ret_de_bh);
			ret = 0;
			goto out;
		}
3474

3475 3476 3477 3478
		last_de = de;
		de_buf += le16_to_cpu(de->rec_len);
		offset += le16_to_cpu(de->rec_len);
	}
3479

3480 3481 3482 3483 3484 3485
	/*
	 * We're going to require expansion of the directory - figure
	 * out how many blocks we'll need so that a place for the
	 * dirent can be found.
	 */
	*blocks_wanted = 1;
3486
	new_rec_len = le16_to_cpu(last_de->rec_len) + free_space;
3487 3488
	if (new_rec_len < (rec_len + OCFS2_DIR_REC_LEN(last_de->name_len)))
		*blocks_wanted = 2;
3489

3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503
	ret = -ENOSPC;
out:
	return ret;
}

static int ocfs2_find_dir_space_el(struct inode *dir, const char *name,
				   int namelen, struct buffer_head **ret_de_bh)
{
	unsigned long offset;
	struct buffer_head *bh = NULL;
	unsigned short rec_len;
	struct ocfs2_dir_entry *de;
	struct super_block *sb = dir->i_sb;
	int status;
3504
	int blocksize = dir->i_sb->s_blocksize;
3505

3506 3507
	status = ocfs2_read_dir_block(dir, 0, &bh, 0);
	if (status) {
3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520
		mlog_errno(status);
		goto bail;
	}

	rec_len = OCFS2_DIR_REC_LEN(namelen);
	offset = 0;
	de = (struct ocfs2_dir_entry *) bh->b_data;
	while (1) {
		if ((char *)de >= sb->s_blocksize + bh->b_data) {
			brelse(bh);
			bh = NULL;

			if (i_size_read(dir) <= offset) {
3521 3522 3523 3524 3525
				/*
				 * Caller will have to expand this
				 * directory.
				 */
				status = -ENOSPC;
3526 3527
				goto bail;
			}
3528 3529 3530 3531
			status = ocfs2_read_dir_block(dir,
					     offset >> sb->s_blocksize_bits,
					     &bh, 0);
			if (status) {
3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545
				mlog_errno(status);
				goto bail;
			}
			/* move to next block */
			de = (struct ocfs2_dir_entry *) bh->b_data;
		}
		if (!ocfs2_check_dir_entry(dir, de, bh, offset)) {
			status = -ENOENT;
			goto bail;
		}
		if (ocfs2_match(namelen, name, de)) {
			status = -EEXIST;
			goto bail;
		}
3546 3547 3548 3549 3550

		if (ocfs2_skip_dir_trailer(dir, de, offset % blocksize,
					   blocksize))
			goto next;

M
Mark Fasheh 已提交
3551
		if (ocfs2_dirent_would_fit(de, rec_len)) {
3552 3553 3554 3555 3556 3557 3558
			/* Ok, we found a spot. Return this bh and let
			 * the caller actually fill it in. */
			*ret_de_bh = bh;
			get_bh(*ret_de_bh);
			status = 0;
			goto bail;
		}
3559
next:
3560 3561 3562 3563 3564 3565
		offset += le16_to_cpu(de->rec_len);
		de = (struct ocfs2_dir_entry *)((char *) de + le16_to_cpu(de->rec_len));
	}

	status = 0;
bail:
3566
	brelse(bh);
T
Tao Ma 已提交
3567 3568
	if (status)
		mlog_errno(status);
3569 3570 3571

	return status;
}
3572

3573
static int dx_leaf_sort_cmp(const void *a, const void *b)
3574
{
3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799
	const struct ocfs2_dx_entry *entry1 = a;
	const struct ocfs2_dx_entry *entry2 = b;
	u32 major_hash1 = le32_to_cpu(entry1->dx_major_hash);
	u32 major_hash2 = le32_to_cpu(entry2->dx_major_hash);
	u32 minor_hash1 = le32_to_cpu(entry1->dx_minor_hash);
	u32 minor_hash2 = le32_to_cpu(entry2->dx_minor_hash);

	if (major_hash1 > major_hash2)
		return 1;
	if (major_hash1 < major_hash2)
		return -1;

	/*
	 * It is not strictly necessary to sort by minor
	 */
	if (minor_hash1 > minor_hash2)
		return 1;
	if (minor_hash1 < minor_hash2)
		return -1;
	return 0;
}

static void dx_leaf_sort_swap(void *a, void *b, int size)
{
	struct ocfs2_dx_entry *entry1 = a;
	struct ocfs2_dx_entry *entry2 = b;
	struct ocfs2_dx_entry tmp;

	BUG_ON(size != sizeof(*entry1));

	tmp = *entry1;
	*entry1 = *entry2;
	*entry2 = tmp;
}

static int ocfs2_dx_leaf_same_major(struct ocfs2_dx_leaf *dx_leaf)
{
	struct ocfs2_dx_entry_list *dl_list = &dx_leaf->dl_list;
	int i, num = le16_to_cpu(dl_list->de_num_used);

	for (i = 0; i < (num - 1); i++) {
		if (le32_to_cpu(dl_list->de_entries[i].dx_major_hash) !=
		    le32_to_cpu(dl_list->de_entries[i + 1].dx_major_hash))
			return 0;
	}

	return 1;
}

/*
 * Find the optimal value to split this leaf on. This expects the leaf
 * entries to be in sorted order.
 *
 * leaf_cpos is the cpos of the leaf we're splitting. insert_hash is
 * the hash we want to insert.
 *
 * This function is only concerned with the major hash - that which
 * determines which cluster an item belongs to.
 */
static int ocfs2_dx_dir_find_leaf_split(struct ocfs2_dx_leaf *dx_leaf,
					u32 leaf_cpos, u32 insert_hash,
					u32 *split_hash)
{
	struct ocfs2_dx_entry_list *dl_list = &dx_leaf->dl_list;
	int i, num_used = le16_to_cpu(dl_list->de_num_used);
	int allsame;

	/*
	 * There's a couple rare, but nasty corner cases we have to
	 * check for here. All of them involve a leaf where all value
	 * have the same hash, which is what we look for first.
	 *
	 * Most of the time, all of the above is false, and we simply
	 * pick the median value for a split.
	 */
	allsame = ocfs2_dx_leaf_same_major(dx_leaf);
	if (allsame) {
		u32 val = le32_to_cpu(dl_list->de_entries[0].dx_major_hash);

		if (val == insert_hash) {
			/*
			 * No matter where we would choose to split,
			 * the new entry would want to occupy the same
			 * block as these. Since there's no space left
			 * in their existing block, we know there
			 * won't be space after the split.
			 */
			return -ENOSPC;
		}

		if (val == leaf_cpos) {
			/*
			 * Because val is the same as leaf_cpos (which
			 * is the smallest value this leaf can have),
			 * yet is not equal to insert_hash, then we
			 * know that insert_hash *must* be larger than
			 * val (and leaf_cpos). At least cpos+1 in value.
			 *
			 * We also know then, that there cannot be an
			 * adjacent extent (otherwise we'd be looking
			 * at it). Choosing this value gives us a
			 * chance to get some contiguousness.
			 */
			*split_hash = leaf_cpos + 1;
			return 0;
		}

		if (val > insert_hash) {
			/*
			 * val can not be the same as insert hash, and
			 * also must be larger than leaf_cpos. Also,
			 * we know that there can't be a leaf between
			 * cpos and val, otherwise the entries with
			 * hash 'val' would be there.
			 */
			*split_hash = val;
			return 0;
		}

		*split_hash = insert_hash;
		return 0;
	}

	/*
	 * Since the records are sorted and the checks above
	 * guaranteed that not all records in this block are the same,
	 * we simple travel forward, from the median, and pick the 1st
	 * record whose value is larger than leaf_cpos.
	 */
	for (i = (num_used / 2); i < num_used; i++)
		if (le32_to_cpu(dl_list->de_entries[i].dx_major_hash) >
		    leaf_cpos)
			break;

	BUG_ON(i == num_used); /* Should be impossible */
	*split_hash = le32_to_cpu(dl_list->de_entries[i].dx_major_hash);
	return 0;
}

/*
 * Transfer all entries in orig_dx_leaves whose major hash is equal to or
 * larger than split_hash into new_dx_leaves. We use a temporary
 * buffer (tmp_dx_leaf) to make the changes to the original leaf blocks.
 *
 * Since the block offset inside a leaf (cluster) is a constant mask
 * of minor_hash, we can optimize - an item at block offset X within
 * the original cluster, will be at offset X within the new cluster.
 */
static void ocfs2_dx_dir_transfer_leaf(struct inode *dir, u32 split_hash,
				       handle_t *handle,
				       struct ocfs2_dx_leaf *tmp_dx_leaf,
				       struct buffer_head **orig_dx_leaves,
				       struct buffer_head **new_dx_leaves,
				       int num_dx_leaves)
{
	int i, j, num_used;
	u32 major_hash;
	struct ocfs2_dx_leaf *orig_dx_leaf, *new_dx_leaf;
	struct ocfs2_dx_entry_list *orig_list, *new_list, *tmp_list;
	struct ocfs2_dx_entry *dx_entry;

	tmp_list = &tmp_dx_leaf->dl_list;

	for (i = 0; i < num_dx_leaves; i++) {
		orig_dx_leaf = (struct ocfs2_dx_leaf *) orig_dx_leaves[i]->b_data;
		orig_list = &orig_dx_leaf->dl_list;
		new_dx_leaf = (struct ocfs2_dx_leaf *) new_dx_leaves[i]->b_data;
		new_list = &new_dx_leaf->dl_list;

		num_used = le16_to_cpu(orig_list->de_num_used);

		memcpy(tmp_dx_leaf, orig_dx_leaf, dir->i_sb->s_blocksize);
		tmp_list->de_num_used = cpu_to_le16(0);
		memset(&tmp_list->de_entries, 0, sizeof(*dx_entry)*num_used);

		for (j = 0; j < num_used; j++) {
			dx_entry = &orig_list->de_entries[j];
			major_hash = le32_to_cpu(dx_entry->dx_major_hash);
			if (major_hash >= split_hash)
				ocfs2_dx_dir_leaf_insert_tail(new_dx_leaf,
							      dx_entry);
			else
				ocfs2_dx_dir_leaf_insert_tail(tmp_dx_leaf,
							      dx_entry);
		}
		memcpy(orig_dx_leaf, tmp_dx_leaf, dir->i_sb->s_blocksize);

		ocfs2_journal_dirty(handle, orig_dx_leaves[i]);
		ocfs2_journal_dirty(handle, new_dx_leaves[i]);
	}
}

static int ocfs2_dx_dir_rebalance_credits(struct ocfs2_super *osb,
					  struct ocfs2_dx_root_block *dx_root)
{
	int credits = ocfs2_clusters_to_blocks(osb->sb, 2);

	credits += ocfs2_calc_extend_credits(osb->sb, &dx_root->dr_list, 1);
	credits += ocfs2_quota_trans_credits(osb->sb);
	return credits;
}

/*
 * Find the median value in dx_leaf_bh and allocate a new leaf to move
 * half our entries into.
 */
static int ocfs2_dx_dir_rebalance(struct ocfs2_super *osb, struct inode *dir,
				  struct buffer_head *dx_root_bh,
				  struct buffer_head *dx_leaf_bh,
				  struct ocfs2_dx_hinfo *hinfo, u32 leaf_cpos,
				  u64 leaf_blkno)
{
	struct ocfs2_dx_leaf *dx_leaf = (struct ocfs2_dx_leaf *)dx_leaf_bh->b_data;
	int credits, ret, i, num_used, did_quota = 0;
	u32 cpos, split_hash, insert_hash = hinfo->major_hash;
	u64 orig_leaves_start;
	int num_dx_leaves;
	struct buffer_head **orig_dx_leaves = NULL;
	struct buffer_head **new_dx_leaves = NULL;
	struct ocfs2_alloc_context *data_ac = NULL, *meta_ac = NULL;
	struct ocfs2_extent_tree et;
	handle_t *handle = NULL;
	struct ocfs2_dx_root_block *dx_root;
	struct ocfs2_dx_leaf *tmp_dx_leaf = NULL;

3800 3801 3802
	trace_ocfs2_dx_dir_rebalance((unsigned long long)OCFS2_I(dir)->ip_blkno,
				     (unsigned long long)leaf_blkno,
				     insert_hash);
3803

3804
	ocfs2_init_dx_root_extent_tree(&et, INODE_CACHE(dir), dx_root_bh);
3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852

	dx_root = (struct ocfs2_dx_root_block *)dx_root_bh->b_data;
	/*
	 * XXX: This is a rather large limit. We should use a more
	 * realistic value.
	 */
	if (le32_to_cpu(dx_root->dr_clusters) == UINT_MAX)
		return -ENOSPC;

	num_used = le16_to_cpu(dx_leaf->dl_list.de_num_used);
	if (num_used < le16_to_cpu(dx_leaf->dl_list.de_count)) {
		mlog(ML_ERROR, "DX Dir: %llu, Asked to rebalance empty leaf: "
		     "%llu, %d\n", (unsigned long long)OCFS2_I(dir)->ip_blkno,
		     (unsigned long long)leaf_blkno, num_used);
		ret = -EIO;
		goto out;
	}

	orig_dx_leaves = ocfs2_dx_dir_kmalloc_leaves(osb->sb, &num_dx_leaves);
	if (!orig_dx_leaves) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out;
	}

	new_dx_leaves = ocfs2_dx_dir_kmalloc_leaves(osb->sb, NULL);
	if (!new_dx_leaves) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_lock_allocators(dir, &et, 1, 0, &data_ac, &meta_ac);
	if (ret) {
		if (ret != -ENOSPC)
			mlog_errno(ret);
		goto out;
	}

	credits = ocfs2_dx_dir_rebalance_credits(osb, dx_root);
	handle = ocfs2_start_trans(osb, credits);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		handle = NULL;
		mlog_errno(ret);
		goto out;
	}

3853 3854 3855
	ret = dquot_alloc_space_nodirty(dir,
				       ocfs2_clusters_to_bytes(dir->i_sb, 1));
	if (ret)
3856 3857 3858
		goto out_commit;
	did_quota = 1;

3859
	ret = ocfs2_journal_access_dl(handle, INODE_CACHE(dir), dx_leaf_bh,
3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	/*
	 * This block is changing anyway, so we can sort it in place.
	 */
	sort(dx_leaf->dl_list.de_entries, num_used,
	     sizeof(struct ocfs2_dx_entry), dx_leaf_sort_cmp,
	     dx_leaf_sort_swap);

3873
	ocfs2_journal_dirty(handle, dx_leaf_bh);
3874 3875 3876 3877 3878 3879 3880 3881

	ret = ocfs2_dx_dir_find_leaf_split(dx_leaf, leaf_cpos, insert_hash,
					   &split_hash);
	if (ret) {
		mlog_errno(ret);
		goto  out_commit;
	}

3882
	trace_ocfs2_dx_dir_rebalance_split(leaf_cpos, split_hash, insert_hash);
3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908

	/*
	 * We have to carefully order operations here. There are items
	 * which want to be in the new cluster before insert, but in
	 * order to put those items in the new cluster, we alter the
	 * old cluster. A failure to insert gets nasty.
	 *
	 * So, start by reserving writes to the old
	 * cluster. ocfs2_dx_dir_new_cluster will reserve writes on
	 * the new cluster for us, before inserting it. The insert
	 * won't happen if there's an error before that. Once the
	 * insert is done then, we can transfer from one leaf into the
	 * other without fear of hitting any error.
	 */

	/*
	 * The leaf transfer wants some scratch space so that we don't
	 * wind up doing a bunch of expensive memmove().
	 */
	tmp_dx_leaf = kmalloc(osb->sb->s_blocksize, GFP_NOFS);
	if (!tmp_dx_leaf) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out_commit;
	}

3909
	orig_leaves_start = ocfs2_block_to_cluster_start(dir->i_sb, leaf_blkno);
3910 3911 3912 3913 3914 3915 3916
	ret = ocfs2_read_dx_leaves(dir, orig_leaves_start, num_dx_leaves,
				   orig_dx_leaves);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

3917 3918 3919 3920 3921 3922 3923 3924 3925
	cpos = split_hash;
	ret = ocfs2_dx_dir_new_cluster(dir, &et, cpos, handle,
				       data_ac, meta_ac, new_dx_leaves,
				       num_dx_leaves);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

3926
	for (i = 0; i < num_dx_leaves; i++) {
3927 3928
		ret = ocfs2_journal_access_dl(handle, INODE_CACHE(dir),
					      orig_dx_leaves[i],
3929 3930 3931 3932 3933 3934
					      OCFS2_JOURNAL_ACCESS_WRITE);
		if (ret) {
			mlog_errno(ret);
			goto out_commit;
		}

3935 3936 3937 3938 3939 3940 3941
		ret = ocfs2_journal_access_dl(handle, INODE_CACHE(dir),
					      new_dx_leaves[i],
					      OCFS2_JOURNAL_ACCESS_WRITE);
		if (ret) {
			mlog_errno(ret);
			goto out_commit;
		}
3942 3943 3944 3945 3946 3947 3948
	}

	ocfs2_dx_dir_transfer_leaf(dir, split_hash, handle, tmp_dx_leaf,
				   orig_dx_leaves, new_dx_leaves, num_dx_leaves);

out_commit:
	if (ret < 0 && did_quota)
3949
		dquot_free_space_nodirty(dir,
3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974
				ocfs2_clusters_to_bytes(dir->i_sb, 1));

	ocfs2_commit_trans(osb, handle);

out:
	if (orig_dx_leaves || new_dx_leaves) {
		for (i = 0; i < num_dx_leaves; i++) {
			if (orig_dx_leaves)
				brelse(orig_dx_leaves[i]);
			if (new_dx_leaves)
				brelse(new_dx_leaves[i]);
		}
		kfree(orig_dx_leaves);
		kfree(new_dx_leaves);
	}

	if (meta_ac)
		ocfs2_free_alloc_context(meta_ac);
	if (data_ac)
		ocfs2_free_alloc_context(data_ac);

	kfree(tmp_dx_leaf);
	return ret;
}

3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095
static int ocfs2_find_dir_space_dx(struct ocfs2_super *osb, struct inode *dir,
				   struct buffer_head *di_bh,
				   struct buffer_head *dx_root_bh,
				   const char *name, int namelen,
				   struct ocfs2_dir_lookup_result *lookup)
{
	int ret, rebalanced = 0;
	struct ocfs2_dx_root_block *dx_root;
	struct buffer_head *dx_leaf_bh = NULL;
	struct ocfs2_dx_leaf *dx_leaf;
	u64 blkno;
	u32 leaf_cpos;

	dx_root = (struct ocfs2_dx_root_block *)dx_root_bh->b_data;

restart_search:
	ret = ocfs2_dx_dir_lookup(dir, &dx_root->dr_list, &lookup->dl_hinfo,
				  &leaf_cpos, &blkno);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	ret = ocfs2_read_dx_leaf(dir, blkno, &dx_leaf_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	dx_leaf = (struct ocfs2_dx_leaf *)dx_leaf_bh->b_data;

	if (le16_to_cpu(dx_leaf->dl_list.de_num_used) >=
	    le16_to_cpu(dx_leaf->dl_list.de_count)) {
		if (rebalanced) {
			/*
			 * Rebalancing should have provided us with
			 * space in an appropriate leaf.
			 *
			 * XXX: Is this an abnormal condition then?
			 * Should we print a message here?
			 */
			ret = -ENOSPC;
			goto out;
		}

		ret = ocfs2_dx_dir_rebalance(osb, dir, dx_root_bh, dx_leaf_bh,
					     &lookup->dl_hinfo, leaf_cpos,
					     blkno);
		if (ret) {
			if (ret != -ENOSPC)
				mlog_errno(ret);
			goto out;
		}

		/*
		 * Restart the lookup. The rebalance might have
		 * changed which block our item fits into. Mark our
		 * progress, so we only execute this once.
		 */
		brelse(dx_leaf_bh);
		dx_leaf_bh = NULL;
		rebalanced = 1;
		goto restart_search;
	}

	lookup->dl_dx_leaf_bh = dx_leaf_bh;
	dx_leaf_bh = NULL;

out:
	brelse(dx_leaf_bh);
	return ret;
}

static int ocfs2_search_dx_free_list(struct inode *dir,
				     struct buffer_head *dx_root_bh,
				     int namelen,
				     struct ocfs2_dir_lookup_result *lookup)
{
	int ret = -ENOSPC;
	struct buffer_head *leaf_bh = NULL, *prev_leaf_bh = NULL;
	struct ocfs2_dir_block_trailer *db;
	u64 next_block;
	int rec_len = OCFS2_DIR_REC_LEN(namelen);
	struct ocfs2_dx_root_block *dx_root;

	dx_root = (struct ocfs2_dx_root_block *)dx_root_bh->b_data;
	next_block = le64_to_cpu(dx_root->dr_free_blk);

	while (next_block) {
		brelse(prev_leaf_bh);
		prev_leaf_bh = leaf_bh;
		leaf_bh = NULL;

		ret = ocfs2_read_dir_block_direct(dir, next_block, &leaf_bh);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		db = ocfs2_trailer_from_bh(leaf_bh, dir->i_sb);
		if (rec_len <= le16_to_cpu(db->db_free_rec_len)) {
			lookup->dl_leaf_bh = leaf_bh;
			lookup->dl_prev_leaf_bh = prev_leaf_bh;
			leaf_bh = NULL;
			prev_leaf_bh = NULL;
			break;
		}

		next_block = le64_to_cpu(db->db_free_next);
	}

	if (!next_block)
		ret = -ENOSPC;

out:

	brelse(leaf_bh);
	brelse(prev_leaf_bh);
	return ret;
}

4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130
static int ocfs2_expand_inline_dx_root(struct inode *dir,
				       struct buffer_head *dx_root_bh)
{
	int ret, num_dx_leaves, i, j, did_quota = 0;
	struct buffer_head **dx_leaves = NULL;
	struct ocfs2_extent_tree et;
	u64 insert_blkno;
	struct ocfs2_alloc_context *data_ac = NULL;
	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
	handle_t *handle = NULL;
	struct ocfs2_dx_root_block *dx_root;
	struct ocfs2_dx_entry_list *entry_list;
	struct ocfs2_dx_entry *dx_entry;
	struct ocfs2_dx_leaf *target_leaf;

	ret = ocfs2_reserve_clusters(osb, 1, &data_ac);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	dx_leaves = ocfs2_dx_dir_kmalloc_leaves(osb->sb, &num_dx_leaves);
	if (!dx_leaves) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out;
	}

	handle = ocfs2_start_trans(osb, ocfs2_calc_dxi_expand_credits(osb->sb));
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		mlog_errno(ret);
		goto out;
	}

4131 4132 4133
	ret = dquot_alloc_space_nodirty(dir,
				       ocfs2_clusters_to_bytes(osb->sb, 1));
	if (ret)
4134 4135 4136 4137 4138 4139 4140 4141
		goto out_commit;
	did_quota = 1;

	/*
	 * We do this up front, before the allocation, so that a
	 * failure to add the dx_root_bh to the journal won't result
	 * us losing clusters.
	 */
4142
	ret = ocfs2_journal_access_dr(handle, INODE_CACHE(dir), dx_root_bh,
4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	ret = __ocfs2_dx_dir_new_cluster(dir, 0, handle, data_ac, dx_leaves,
					 num_dx_leaves, &insert_blkno);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

	/*
	 * Transfer the entries from our dx_root into the appropriate
	 * block
	 */
	dx_root = (struct ocfs2_dx_root_block *) dx_root_bh->b_data;
	entry_list = &dx_root->dr_entries;

	for (i = 0; i < le16_to_cpu(entry_list->de_num_used); i++) {
		dx_entry = &entry_list->de_entries[i];

		j = __ocfs2_dx_dir_hash_idx(osb,
					    le32_to_cpu(dx_entry->dx_minor_hash));
		target_leaf = (struct ocfs2_dx_leaf *)dx_leaves[j]->b_data;

		ocfs2_dx_dir_leaf_insert_tail(target_leaf, dx_entry);

		/* Each leaf has been passed to the journal already
		 * via __ocfs2_dx_dir_new_cluster() */
	}

	dx_root->dr_flags &= ~OCFS2_DX_FLAG_INLINE;
	memset(&dx_root->dr_list, 0, osb->sb->s_blocksize -
	       offsetof(struct ocfs2_dx_root_block, dr_list));
	dx_root->dr_list.l_count =
		cpu_to_le16(ocfs2_extent_recs_per_dx_root(osb->sb));

	/* This should never fail considering we start with an empty
	 * dx_root. */
4184
	ocfs2_init_dx_root_extent_tree(&et, INODE_CACHE(dir), dx_root_bh);
4185
	ret = ocfs2_insert_extent(handle, &et, 0, insert_blkno, 1, 0, NULL);
4186 4187 4188 4189 4190 4191 4192 4193
	if (ret)
		mlog_errno(ret);
	did_quota = 0;

	ocfs2_journal_dirty(handle, dx_root_bh);

out_commit:
	if (ret < 0 && did_quota)
4194
		dquot_free_space_nodirty(dir,
4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225
					  ocfs2_clusters_to_bytes(dir->i_sb, 1));

	ocfs2_commit_trans(osb, handle);

out:
	if (data_ac)
		ocfs2_free_alloc_context(data_ac);

	if (dx_leaves) {
		for (i = 0; i < num_dx_leaves; i++)
			brelse(dx_leaves[i]);
		kfree(dx_leaves);
	}
	return ret;
}

static int ocfs2_inline_dx_has_space(struct buffer_head *dx_root_bh)
{
	struct ocfs2_dx_root_block *dx_root;
	struct ocfs2_dx_entry_list *entry_list;

	dx_root = (struct ocfs2_dx_root_block *) dx_root_bh->b_data;
	entry_list = &dx_root->dr_entries;

	if (le16_to_cpu(entry_list->de_num_used) >=
	    le16_to_cpu(entry_list->de_count))
		return -ENOSPC;

	return 0;
}

4226 4227 4228 4229 4230
static int ocfs2_prepare_dx_dir_for_insert(struct inode *dir,
					   struct buffer_head *di_bh,
					   const char *name,
					   int namelen,
					   struct ocfs2_dir_lookup_result *lookup)
4231
{
4232 4233
	int ret, free_dx_root = 1;
	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
4234
	struct buffer_head *dx_root_bh = NULL;
4235
	struct buffer_head *leaf_bh = NULL;
4236
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
4237
	struct ocfs2_dx_root_block *dx_root;
4238 4239 4240 4241 4242 4243 4244 4245

	ret = ocfs2_read_dx_root(dir, di, &dx_root_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

	dx_root = (struct ocfs2_dx_root_block *)dx_root_bh->b_data;
4246 4247 4248 4249 4250 4251
	if (le32_to_cpu(dx_root->dr_num_entries) == OCFS2_DX_ENTRIES_MAX) {
		ret = -ENOSPC;
		mlog_errno(ret);
		goto out;
	}

4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268
	if (ocfs2_dx_root_inline(dx_root)) {
		ret = ocfs2_inline_dx_has_space(dx_root_bh);

		if (ret == 0)
			goto search_el;

		/*
		 * We ran out of room in the root block. Expand it to
		 * an extent, then allow ocfs2_find_dir_space_dx to do
		 * the rest.
		 */
		ret = ocfs2_expand_inline_dx_root(dir, dx_root_bh);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}
	}
4269

4270 4271 4272 4273 4274 4275 4276 4277
	/*
	 * Insert preparation for an indexed directory is split into two
	 * steps. The call to find_dir_space_dx reserves room in the index for
	 * an additional item. If we run out of space there, it's a real error
	 * we can't continue on.
	 */
	ret = ocfs2_find_dir_space_dx(osb, dir, di_bh, dx_root_bh, name,
				      namelen, lookup);
4278 4279 4280 4281 4282
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

4283 4284 4285 4286 4287 4288 4289 4290 4291
search_el:
	/*
	 * Next, we need to find space in the unindexed tree. This call
	 * searches using the free space linked list. If the unindexed tree
	 * lacks sufficient space, we'll expand it below. The expansion code
	 * is smart enough to add any new blocks to the free space list.
	 */
	ret = ocfs2_search_dx_free_list(dir, dx_root_bh, namelen, lookup);
	if (ret && ret != -ENOSPC) {
4292 4293 4294 4295
		mlog_errno(ret);
		goto out;
	}

4296 4297 4298
	/* Do this up here - ocfs2_extend_dir might need the dx_root */
	lookup->dl_dx_root_bh = dx_root_bh;
	free_dx_root = 0;
4299

4300 4301
	if (ret == -ENOSPC) {
		ret = ocfs2_extend_dir(osb, dir, di_bh, 1, lookup, &leaf_bh);
4302 4303

		if (ret) {
4304
			mlog_errno(ret);
4305 4306 4307 4308
			goto out;
		}

		/*
4309 4310
		 * We make the assumption here that new leaf blocks are added
		 * to the front of our free list.
4311
		 */
4312 4313
		lookup->dl_prev_leaf_bh = NULL;
		lookup->dl_leaf_bh = leaf_bh;
4314 4315 4316
	}

out:
4317 4318
	if (free_dx_root)
		brelse(dx_root_bh);
4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335
	return ret;
}

/*
 * Get a directory ready for insert. Any directory allocation required
 * happens here. Success returns zero, and enough context in the dir
 * lookup result that ocfs2_add_entry() will be able complete the task
 * with minimal performance impact.
 */
int ocfs2_prepare_dir_for_insert(struct ocfs2_super *osb,
				 struct inode *dir,
				 struct buffer_head *parent_fe_bh,
				 const char *name,
				 int namelen,
				 struct ocfs2_dir_lookup_result *lookup)
{
	int ret;
4336 4337 4338
	unsigned int blocks_wanted = 1;
	struct buffer_head *bh = NULL;

4339 4340
	trace_ocfs2_prepare_dir_for_insert(
		(unsigned long long)OCFS2_I(dir)->ip_blkno, namelen);
4341 4342 4343 4344 4345 4346 4347

	if (!namelen) {
		ret = -EINVAL;
		mlog_errno(ret);
		goto out;
	}

4348 4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362
	/*
	 * Do this up front to reduce confusion.
	 *
	 * The directory might start inline, then be turned into an
	 * indexed one, in which case we'd need to hash deep inside
	 * ocfs2_find_dir_space_id(). Since
	 * ocfs2_prepare_dx_dir_for_insert() also needs this hash
	 * done, there seems no point in spreading out the calls. We
	 * can optimize away the case where the file system doesn't
	 * support indexing.
	 */
	if (ocfs2_supports_indexed_dirs(osb))
		ocfs2_dx_dir_name_hash(dir, name, namelen, &lookup->dl_hinfo);

	if (ocfs2_dir_indexed(dir)) {
4363 4364 4365
		ret = ocfs2_prepare_dx_dir_for_insert(dir, parent_fe_bh,
						      name, namelen, lookup);
		if (ret)
4366
			mlog_errno(ret);
4367
		goto out;
4368 4369
	}

4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387
	if (OCFS2_I(dir)->ip_dyn_features & OCFS2_INLINE_DATA_FL) {
		ret = ocfs2_find_dir_space_id(dir, parent_fe_bh, name,
					      namelen, &bh, &blocks_wanted);
	} else
		ret = ocfs2_find_dir_space_el(dir, name, namelen, &bh);

	if (ret && ret != -ENOSPC) {
		mlog_errno(ret);
		goto out;
	}

	if (ret == -ENOSPC) {
		/*
		 * We have to expand the directory to add this name.
		 */
		BUG_ON(bh);

		ret = ocfs2_extend_dir(osb, dir, parent_fe_bh, blocks_wanted,
4388
				       lookup, &bh);
4389 4390 4391 4392 4393 4394 4395 4396 4397
		if (ret) {
			if (ret != -ENOSPC)
				mlog_errno(ret);
			goto out;
		}

		BUG_ON(!bh);
	}

4398
	lookup->dl_leaf_bh = bh;
4399 4400
	bh = NULL;
out:
4401
	brelse(bh);
4402 4403
	return ret;
}
4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444

static int ocfs2_dx_dir_remove_index(struct inode *dir,
				     struct buffer_head *di_bh,
				     struct buffer_head *dx_root_bh)
{
	int ret;
	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
	struct ocfs2_dx_root_block *dx_root;
	struct inode *dx_alloc_inode = NULL;
	struct buffer_head *dx_alloc_bh = NULL;
	handle_t *handle;
	u64 blk;
	u16 bit;
	u64 bg_blkno;

	dx_root = (struct ocfs2_dx_root_block *) dx_root_bh->b_data;

	dx_alloc_inode = ocfs2_get_system_file_inode(osb,
					EXTENT_ALLOC_SYSTEM_INODE,
					le16_to_cpu(dx_root->dr_suballoc_slot));
	if (!dx_alloc_inode) {
		ret = -ENOMEM;
		mlog_errno(ret);
		goto out;
	}
	mutex_lock(&dx_alloc_inode->i_mutex);

	ret = ocfs2_inode_lock(dx_alloc_inode, &dx_alloc_bh, 1);
	if (ret) {
		mlog_errno(ret);
		goto out_mutex;
	}

	handle = ocfs2_start_trans(osb, OCFS2_DX_ROOT_REMOVE_CREDITS);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		mlog_errno(ret);
		goto out_unlock;
	}

4445
	ret = ocfs2_journal_access_di(handle, INODE_CACHE(dir), di_bh,
4446 4447 4448 4449 4450 4451
				      OCFS2_JOURNAL_ACCESS_WRITE);
	if (ret) {
		mlog_errno(ret);
		goto out_commit;
	}

4452
	spin_lock(&OCFS2_I(dir)->ip_lock);
4453 4454
	OCFS2_I(dir)->ip_dyn_features &= ~OCFS2_INDEXED_DIR_FL;
	di->i_dyn_features = cpu_to_le16(OCFS2_I(dir)->ip_dyn_features);
4455
	spin_unlock(&OCFS2_I(dir)->ip_lock);
4456 4457 4458 4459 4460 4461
	di->i_dx_root = cpu_to_le64(0ULL);

	ocfs2_journal_dirty(handle, di_bh);

	blk = le64_to_cpu(dx_root->dr_blkno);
	bit = le16_to_cpu(dx_root->dr_suballoc_bit);
4462 4463 4464 4465
	if (dx_root->dr_suballoc_loc)
		bg_blkno = le64_to_cpu(dx_root->dr_suballoc_loc);
	else
		bg_blkno = ocfs2_which_suballoc_group(blk, bit);
4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507
	ret = ocfs2_free_suballoc_bits(handle, dx_alloc_inode, dx_alloc_bh,
				       bit, bg_blkno, 1);
	if (ret)
		mlog_errno(ret);

out_commit:
	ocfs2_commit_trans(osb, handle);

out_unlock:
	ocfs2_inode_unlock(dx_alloc_inode, 1);

out_mutex:
	mutex_unlock(&dx_alloc_inode->i_mutex);
	brelse(dx_alloc_bh);
out:
	iput(dx_alloc_inode);
	return ret;
}

int ocfs2_dx_dir_truncate(struct inode *dir, struct buffer_head *di_bh)
{
	int ret;
	unsigned int uninitialized_var(clen);
	u32 major_hash = UINT_MAX, p_cpos, uninitialized_var(cpos);
	u64 uninitialized_var(blkno);
	struct ocfs2_super *osb = OCFS2_SB(dir->i_sb);
	struct buffer_head *dx_root_bh = NULL;
	struct ocfs2_dx_root_block *dx_root;
	struct ocfs2_dinode *di = (struct ocfs2_dinode *)di_bh->b_data;
	struct ocfs2_cached_dealloc_ctxt dealloc;
	struct ocfs2_extent_tree et;

	ocfs2_init_dealloc_ctxt(&dealloc);

	if (!ocfs2_dir_indexed(dir))
		return 0;

	ret = ocfs2_read_dx_root(dir, di, &dx_root_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}
4508
	dx_root = (struct ocfs2_dx_root_block *)dx_root_bh->b_data;
4509

4510 4511
	if (ocfs2_dx_root_inline(dx_root))
		goto remove_index;
4512

4513
	ocfs2_init_dx_root_extent_tree(&et, INODE_CACHE(dir), dx_root_bh);
4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525

	/* XXX: What if dr_clusters is too large? */
	while (le32_to_cpu(dx_root->dr_clusters)) {
		ret = ocfs2_dx_dir_lookup_rec(dir, &dx_root->dr_list,
					      major_hash, &cpos, &blkno, &clen);
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		p_cpos = ocfs2_blocks_to_clusters(dir->i_sb, blkno);

4526 4527
		ret = ocfs2_remove_btree_range(dir, &et, cpos, p_cpos, clen, 0,
					       &dealloc, 0);
4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538
		if (ret) {
			mlog_errno(ret);
			goto out;
		}

		if (cpos == 0)
			break;

		major_hash = cpos - 1;
	}

4539
remove_index:
4540 4541 4542 4543 4544 4545
	ret = ocfs2_dx_dir_remove_index(dir, di_bh, dx_root_bh);
	if (ret) {
		mlog_errno(ret);
		goto out;
	}

4546
	ocfs2_remove_from_cache(INODE_CACHE(dir), dx_root_bh);
4547 4548 4549 4550 4551 4552 4553
out:
	ocfs2_schedule_truncate_log_flush(osb, 1);
	ocfs2_run_deallocs(osb, &dealloc);

	brelse(dx_root_bh);
	return ret;
}