1/*
2 * Copyright (C) Sistina Software, Inc.  1997-2003 All rights reserved.
3 * Copyright (C) 2004-2006 Red Hat, Inc.  All rights reserved.
4 *
5 * This copyrighted material is made available to anyone wishing to use,
6 * modify, copy, or redistribute it subject to the terms and conditions
7 * of the GNU General Public License version 2.
8 */
9
10/*
11 * Implements Extendible Hashing as described in:
12 *   "Extendible Hashing" by Fagin, et al in
13 *     __ACM Trans. on Database Systems__, Sept 1979.
14 *
15 *
16 * Here's the layout of dirents which is essentially the same as that of ext2
17 * within a single block. The field de_name_len is the number of bytes
18 * actually required for the name (no null terminator). The field de_rec_len
19 * is the number of bytes allocated to the dirent. The offset of the next
20 * dirent in the block is (dirent + dirent->de_rec_len). When a dirent is
21 * deleted, the preceding dirent inherits its allocated space, ie
22 * prev->de_rec_len += deleted->de_rec_len. Since the next dirent is obtained
23 * by adding de_rec_len to the current dirent, this essentially causes the
24 * deleted dirent to get jumped over when iterating through all the dirents.
25 *
26 * When deleting the first dirent in a block, there is no previous dirent so
27 * the field de_ino is set to zero to designate it as deleted. When allocating
28 * a dirent, gfs2_dirent_alloc iterates through the dirents in a block. If the
29 * first dirent has (de_ino == 0) and de_rec_len is large enough, this first
30 * dirent is allocated. Otherwise it must go through all the 'used' dirents
31 * searching for one in which the amount of total space minus the amount of
32 * used space will provide enough space for the new dirent.
33 *
34 * There are two types of blocks in which dirents reside. In a stuffed dinode,
35 * the dirents begin at offset sizeof(struct gfs2_dinode) from the beginning of
36 * the block.  In leaves, they begin at offset sizeof(struct gfs2_leaf) from the
37 * beginning of the leaf block. The dirents reside in leaves when
38 *
39 * dip->i_diskflags & GFS2_DIF_EXHASH is true
40 *
41 * Otherwise, the dirents are "linear", within a single stuffed dinode block.
42 *
43 * When the dirents are in leaves, the actual contents of the directory file are
44 * used as an array of 64-bit block pointers pointing to the leaf blocks. The
45 * dirents are NOT in the directory file itself. There can be more than one
46 * block pointer in the array that points to the same leaf. In fact, when a
47 * directory is first converted from linear to exhash, all of the pointers
48 * point to the same leaf.
49 *
50 * When a leaf is completely full, the size of the hash table can be
51 * doubled unless it is already at the maximum size which is hard coded into
52 * GFS2_DIR_MAX_DEPTH. After that, leaves are chained together in a linked list,
53 * but never before the maximum hash table size has been reached.
54 */
55
56#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
57
58#include <linux/slab.h>
59#include <linux/spinlock.h>
60#include <linux/buffer_head.h>
61#include <linux/sort.h>
62#include <linux/gfs2_ondisk.h>
63#include <linux/crc32.h>
64#include <linux/vmalloc.h>
65
66#include "gfs2.h"
67#include "incore.h"
68#include "dir.h"
69#include "glock.h"
70#include "inode.h"
71#include "meta_io.h"
72#include "quota.h"
73#include "rgrp.h"
74#include "trans.h"
75#include "bmap.h"
76#include "util.h"
77
78#define IS_LEAF     1 /* Hashed (leaf) directory */
79#define IS_DINODE   2 /* Linear (stuffed dinode block) directory */
80
81#define MAX_RA_BLOCKS 32 /* max read-ahead blocks */
82
83#define gfs2_disk_hash2offset(h) (((u64)(h)) >> 1)
84#define gfs2_dir_offset2hash(p) ((u32)(((u64)(p)) << 1))
85
86struct qstr gfs2_qdot __read_mostly;
87struct qstr gfs2_qdotdot __read_mostly;
88
89typedef int (*gfs2_dscan_t)(const struct gfs2_dirent *dent,
90			    const struct qstr *name, void *opaque);
91
92int gfs2_dir_get_new_buffer(struct gfs2_inode *ip, u64 block,
93			    struct buffer_head **bhp)
94{
95	struct buffer_head *bh;
96
97	bh = gfs2_meta_new(ip->i_gl, block);
98	gfs2_trans_add_meta(ip->i_gl, bh);
99	gfs2_metatype_set(bh, GFS2_METATYPE_JD, GFS2_FORMAT_JD);
100	gfs2_buffer_clear_tail(bh, sizeof(struct gfs2_meta_header));
101	*bhp = bh;
102	return 0;
103}
104
105static int gfs2_dir_get_existing_buffer(struct gfs2_inode *ip, u64 block,
106					struct buffer_head **bhp)
107{
108	struct buffer_head *bh;
109	int error;
110
111	error = gfs2_meta_read(ip->i_gl, block, DIO_WAIT, &bh);
112	if (error)
113		return error;
114	if (gfs2_metatype_check(GFS2_SB(&ip->i_inode), bh, GFS2_METATYPE_JD)) {
115		brelse(bh);
116		return -EIO;
117	}
118	*bhp = bh;
119	return 0;
120}
121
122static int gfs2_dir_write_stuffed(struct gfs2_inode *ip, const char *buf,
123				  unsigned int offset, unsigned int size)
124{
125	struct buffer_head *dibh;
126	int error;
127
128	error = gfs2_meta_inode_buffer(ip, &dibh);
129	if (error)
130		return error;
131
132	gfs2_trans_add_meta(ip->i_gl, dibh);
133	memcpy(dibh->b_data + offset + sizeof(struct gfs2_dinode), buf, size);
134	if (ip->i_inode.i_size < offset + size)
135		i_size_write(&ip->i_inode, offset + size);
136	ip->i_inode.i_mtime = ip->i_inode.i_ctime = CURRENT_TIME;
137	gfs2_dinode_out(ip, dibh->b_data);
138
139	brelse(dibh);
140
141	return size;
142}
143
144
145
146/**
147 * gfs2_dir_write_data - Write directory information to the inode
148 * @ip: The GFS2 inode
149 * @buf: The buffer containing information to be written
150 * @offset: The file offset to start writing at
151 * @size: The amount of data to write
152 *
153 * Returns: The number of bytes correctly written or error code
154 */
155static int gfs2_dir_write_data(struct gfs2_inode *ip, const char *buf,
156			       u64 offset, unsigned int size)
157{
158	struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
159	struct buffer_head *dibh;
160	u64 lblock, dblock;
161	u32 extlen = 0;
162	unsigned int o;
163	int copied = 0;
164	int error = 0;
165	int new = 0;
166
167	if (!size)
168		return 0;
169
170	if (gfs2_is_stuffed(ip) &&
171	    offset + size <= sdp->sd_sb.sb_bsize - sizeof(struct gfs2_dinode))
172		return gfs2_dir_write_stuffed(ip, buf, (unsigned int)offset,
173					      size);
174
175	if (gfs2_assert_warn(sdp, gfs2_is_jdata(ip)))
176		return -EINVAL;
177
178	if (gfs2_is_stuffed(ip)) {
179		error = gfs2_unstuff_dinode(ip, NULL);
180		if (error)
181			return error;
182	}
183
184	lblock = offset;
185	o = do_div(lblock, sdp->sd_jbsize) + sizeof(struct gfs2_meta_header);
186
187	while (copied < size) {
188		unsigned int amount;
189		struct buffer_head *bh;
190
191		amount = size - copied;
192		if (amount > sdp->sd_sb.sb_bsize - o)
193			amount = sdp->sd_sb.sb_bsize - o;
194
195		if (!extlen) {
196			new = 1;
197			error = gfs2_extent_map(&ip->i_inode, lblock, &new,
198						&dblock, &extlen);
199			if (error)
200				goto fail;
201			error = -EIO;
202			if (gfs2_assert_withdraw(sdp, dblock))
203				goto fail;
204		}
205
206		if (amount == sdp->sd_jbsize || new)
207			error = gfs2_dir_get_new_buffer(ip, dblock, &bh);
208		else
209			error = gfs2_dir_get_existing_buffer(ip, dblock, &bh);
210
211		if (error)
212			goto fail;
213
214		gfs2_trans_add_meta(ip->i_gl, bh);
215		memcpy(bh->b_data + o, buf, amount);
216		brelse(bh);
217
218		buf += amount;
219		copied += amount;
220		lblock++;
221		dblock++;
222		extlen--;
223
224		o = sizeof(struct gfs2_meta_header);
225	}
226
227out:
228	error = gfs2_meta_inode_buffer(ip, &dibh);
229	if (error)
230		return error;
231
232	if (ip->i_inode.i_size < offset + copied)
233		i_size_write(&ip->i_inode, offset + copied);
234	ip->i_inode.i_mtime = ip->i_inode.i_ctime = CURRENT_TIME;
235
236	gfs2_trans_add_meta(ip->i_gl, dibh);
237	gfs2_dinode_out(ip, dibh->b_data);
238	brelse(dibh);
239
240	return copied;
241fail:
242	if (copied)
243		goto out;
244	return error;
245}
246
247static int gfs2_dir_read_stuffed(struct gfs2_inode *ip, __be64 *buf,
248				 unsigned int size)
249{
250	struct buffer_head *dibh;
251	int error;
252
253	error = gfs2_meta_inode_buffer(ip, &dibh);
254	if (!error) {
255		memcpy(buf, dibh->b_data + sizeof(struct gfs2_dinode), size);
256		brelse(dibh);
257	}
258
259	return (error) ? error : size;
260}
261
262
263/**
264 * gfs2_dir_read_data - Read a data from a directory inode
265 * @ip: The GFS2 Inode
266 * @buf: The buffer to place result into
267 * @size: Amount of data to transfer
268 *
269 * Returns: The amount of data actually copied or the error
270 */
271static int gfs2_dir_read_data(struct gfs2_inode *ip, __be64 *buf,
272			      unsigned int size)
273{
274	struct gfs2_sbd *sdp = GFS2_SB(&ip->i_inode);
275	u64 lblock, dblock;
276	u32 extlen = 0;
277	unsigned int o;
278	int copied = 0;
279	int error = 0;
280
281	if (gfs2_is_stuffed(ip))
282		return gfs2_dir_read_stuffed(ip, buf, size);
283
284	if (gfs2_assert_warn(sdp, gfs2_is_jdata(ip)))
285		return -EINVAL;
286
287	lblock = 0;
288	o = do_div(lblock, sdp->sd_jbsize) + sizeof(struct gfs2_meta_header);
289
290	while (copied < size) {
291		unsigned int amount;
292		struct buffer_head *bh;
293		int new;
294
295		amount = size - copied;
296		if (amount > sdp->sd_sb.sb_bsize - o)
297			amount = sdp->sd_sb.sb_bsize - o;
298
299		if (!extlen) {
300			new = 0;
301			error = gfs2_extent_map(&ip->i_inode, lblock, &new,
302						&dblock, &extlen);
303			if (error || !dblock)
304				goto fail;
305			BUG_ON(extlen < 1);
306			bh = gfs2_meta_ra(ip->i_gl, dblock, extlen);
307		} else {
308			error = gfs2_meta_read(ip->i_gl, dblock, DIO_WAIT, &bh);
309			if (error)
310				goto fail;
311		}
312		error = gfs2_metatype_check(sdp, bh, GFS2_METATYPE_JD);
313		if (error) {
314			brelse(bh);
315			goto fail;
316		}
317		dblock++;
318		extlen--;
319		memcpy(buf, bh->b_data + o, amount);
320		brelse(bh);
321		buf += (amount/sizeof(__be64));
322		copied += amount;
323		lblock++;
324		o = sizeof(struct gfs2_meta_header);
325	}
326
327	return copied;
328fail:
329	return (copied) ? copied : error;
330}
331
332/**
333 * gfs2_dir_get_hash_table - Get pointer to the dir hash table
334 * @ip: The inode in question
335 *
336 * Returns: The hash table or an error
337 */
338
339static __be64 *gfs2_dir_get_hash_table(struct gfs2_inode *ip)
340{
341	struct inode *inode = &ip->i_inode;
342	int ret;
343	u32 hsize;
344	__be64 *hc;
345
346	BUG_ON(!(ip->i_diskflags & GFS2_DIF_EXHASH));
347
348	hc = ip->i_hash_cache;
349	if (hc)
350		return hc;
351
352	hsize = 1 << ip->i_depth;
353	hsize *= sizeof(__be64);
354	if (hsize != i_size_read(&ip->i_inode)) {
355		gfs2_consist_inode(ip);
356		return ERR_PTR(-EIO);
357	}
358
359	hc = kmalloc(hsize, GFP_NOFS | __GFP_NOWARN);
360	if (hc == NULL)
361		hc = __vmalloc(hsize, GFP_NOFS, PAGE_KERNEL);
362
363	if (hc == NULL)
364		return ERR_PTR(-ENOMEM);
365
366	ret = gfs2_dir_read_data(ip, hc, hsize);
367	if (ret < 0) {
368		kvfree(hc);
369		return ERR_PTR(ret);
370	}
371
372	spin_lock(&inode->i_lock);
373	if (likely(!ip->i_hash_cache)) {
374		ip->i_hash_cache = hc;
375		hc = NULL;
376	}
377	spin_unlock(&inode->i_lock);
378	kvfree(hc);
379
380	return ip->i_hash_cache;
381}
382
383/**
384 * gfs2_dir_hash_inval - Invalidate dir hash
385 * @ip: The directory inode
386 *
387 * Must be called with an exclusive glock, or during glock invalidation.
388 */
389void gfs2_dir_hash_inval(struct gfs2_inode *ip)
390{
391	__be64 *hc;
392
393	spin_lock(&ip->i_inode.i_lock);
394	hc = ip->i_hash_cache;
395	ip->i_hash_cache = NULL;
396	spin_unlock(&ip->i_inode.i_lock);
397
398	kvfree(hc);
399}
400
401static inline int gfs2_dirent_sentinel(const struct gfs2_dirent *dent)
402{
403	return dent->de_inum.no_addr == 0 || dent->de_inum.no_formal_ino == 0;
404}
405
406static inline int __gfs2_dirent_find(const struct gfs2_dirent *dent,
407				     const struct qstr *name, int ret)
408{
409	if (!gfs2_dirent_sentinel(dent) &&
410	    be32_to_cpu(dent->de_hash) == name->hash &&
411	    be16_to_cpu(dent->de_name_len) == name->len &&
412	    memcmp(dent+1, name->name, name->len) == 0)
413		return ret;
414	return 0;
415}
416
417static int gfs2_dirent_find(const struct gfs2_dirent *dent,
418			    const struct qstr *name,
419			    void *opaque)
420{
421	return __gfs2_dirent_find(dent, name, 1);
422}
423
424static int gfs2_dirent_prev(const struct gfs2_dirent *dent,
425			    const struct qstr *name,
426			    void *opaque)
427{
428	return __gfs2_dirent_find(dent, name, 2);
429}
430
431/*
432 * name->name holds ptr to start of block.
433 * name->len holds size of block.
434 */
435static int gfs2_dirent_last(const struct gfs2_dirent *dent,
436			    const struct qstr *name,
437			    void *opaque)
438{
439	const char *start = name->name;
440	const char *end = (const char *)dent + be16_to_cpu(dent->de_rec_len);
441	if (name->len == (end - start))
442		return 1;
443	return 0;
444}
445
446static int gfs2_dirent_find_space(const struct gfs2_dirent *dent,
447				  const struct qstr *name,
448				  void *opaque)
449{
450	unsigned required = GFS2_DIRENT_SIZE(name->len);
451	unsigned actual = GFS2_DIRENT_SIZE(be16_to_cpu(dent->de_name_len));
452	unsigned totlen = be16_to_cpu(dent->de_rec_len);
453
454	if (gfs2_dirent_sentinel(dent))
455		actual = 0;
456	if (totlen - actual >= required)
457		return 1;
458	return 0;
459}
460
461struct dirent_gather {
462	const struct gfs2_dirent **pdent;
463	unsigned offset;
464};
465
466static int gfs2_dirent_gather(const struct gfs2_dirent *dent,
467			      const struct qstr *name,
468			      void *opaque)
469{
470	struct dirent_gather *g = opaque;
471	if (!gfs2_dirent_sentinel(dent)) {
472		g->pdent[g->offset++] = dent;
473	}
474	return 0;
475}
476
477/*
478 * Other possible things to check:
479 * - Inode located within filesystem size (and on valid block)
480 * - Valid directory entry type
481 * Not sure how heavy-weight we want to make this... could also check
482 * hash is correct for example, but that would take a lot of extra time.
483 * For now the most important thing is to check that the various sizes
484 * are correct.
485 */
486static int gfs2_check_dirent(struct gfs2_dirent *dent, unsigned int offset,
487			     unsigned int size, unsigned int len, int first)
488{
489	const char *msg = "gfs2_dirent too small";
490	if (unlikely(size < sizeof(struct gfs2_dirent)))
491		goto error;
492	msg = "gfs2_dirent misaligned";
493	if (unlikely(offset & 0x7))
494		goto error;
495	msg = "gfs2_dirent points beyond end of block";
496	if (unlikely(offset + size > len))
497		goto error;
498	msg = "zero inode number";
499	if (unlikely(!first && gfs2_dirent_sentinel(dent)))
500		goto error;
501	msg = "name length is greater than space in dirent";
502	if (!gfs2_dirent_sentinel(dent) &&
503	    unlikely(sizeof(struct gfs2_dirent)+be16_to_cpu(dent->de_name_len) >
504		     size))
505		goto error;
506	return 0;
507error:
508	pr_warn("%s: %s (%s)\n",
509		__func__, msg, first ? "first in block" : "not first in block");
510	return -EIO;
511}
512
513static int gfs2_dirent_offset(const void *buf)
514{
515	const struct gfs2_meta_header *h = buf;
516	int offset;
517
518	BUG_ON(buf == NULL);
519
520	switch(be32_to_cpu(h->mh_type)) {
521	case GFS2_METATYPE_LF:
522		offset = sizeof(struct gfs2_leaf);
523		break;
524	case GFS2_METATYPE_DI:
525		offset = sizeof(struct gfs2_dinode);
526		break;
527	default:
528		goto wrong_type;
529	}
530	return offset;
531wrong_type:
532	pr_warn("%s: wrong block type %u\n", __func__, be32_to_cpu(h->mh_type));
533	return -1;
534}
535
536static struct gfs2_dirent *gfs2_dirent_scan(struct inode *inode, void *buf,
537					    unsigned int len, gfs2_dscan_t scan,
538					    const struct qstr *name,
539					    void *opaque)
540{
541	struct gfs2_dirent *dent, *prev;
542	unsigned offset;
543	unsigned size;
544	int ret = 0;
545
546	ret = gfs2_dirent_offset(buf);
547	if (ret < 0)
548		goto consist_inode;
549
550	offset = ret;
551	prev = NULL;
552	dent = buf + offset;
553	size = be16_to_cpu(dent->de_rec_len);
554	if (gfs2_check_dirent(dent, offset, size, len, 1))
555		goto consist_inode;
556	do {
557		ret = scan(dent, name, opaque);
558		if (ret)
559			break;
560		offset += size;
561		if (offset == len)
562			break;
563		prev = dent;
564		dent = buf + offset;
565		size = be16_to_cpu(dent->de_rec_len);
566		if (gfs2_check_dirent(dent, offset, size, len, 0))
567			goto consist_inode;
568	} while(1);
569
570	switch(ret) {
571	case 0:
572		return NULL;
573	case 1:
574		return dent;
575	case 2:
576		return prev ? prev : dent;
577	default:
578		BUG_ON(ret > 0);
579		return ERR_PTR(ret);
580	}
581
582consist_inode:
583	gfs2_consist_inode(GFS2_I(inode));
584	return ERR_PTR(-EIO);
585}
586
587static int dirent_check_reclen(struct gfs2_inode *dip,
588			       const struct gfs2_dirent *d, const void *end_p)
589{
590	const void *ptr = d;
591	u16 rec_len = be16_to_cpu(d->de_rec_len);
592
593	if (unlikely(rec_len < sizeof(struct gfs2_dirent)))
594		goto broken;
595	ptr += rec_len;
596	if (ptr < end_p)
597		return rec_len;
598	if (ptr == end_p)
599		return -ENOENT;
600broken:
601	gfs2_consist_inode(dip);
602	return -EIO;
603}
604
605/**
606 * dirent_next - Next dirent
607 * @dip: the directory
608 * @bh: The buffer
609 * @dent: Pointer to list of dirents
610 *
611 * Returns: 0 on success, error code otherwise
612 */
613
614static int dirent_next(struct gfs2_inode *dip, struct buffer_head *bh,
615		       struct gfs2_dirent **dent)
616{
617	struct gfs2_dirent *cur = *dent, *tmp;
618	char *bh_end = bh->b_data + bh->b_size;
619	int ret;
620
621	ret = dirent_check_reclen(dip, cur, bh_end);
622	if (ret < 0)
623		return ret;
624
625	tmp = (void *)cur + ret;
626	ret = dirent_check_reclen(dip, tmp, bh_end);
627	if (ret == -EIO)
628		return ret;
629
630        /* Only the first dent could ever have de_inum.no_addr == 0 */
631	if (gfs2_dirent_sentinel(tmp)) {
632		gfs2_consist_inode(dip);
633		return -EIO;
634	}
635
636	*dent = tmp;
637	return 0;
638}
639
640/**
641 * dirent_del - Delete a dirent
642 * @dip: The GFS2 inode
643 * @bh: The buffer
644 * @prev: The previous dirent
645 * @cur: The current dirent
646 *
647 */
648
649static void dirent_del(struct gfs2_inode *dip, struct buffer_head *bh,
650		       struct gfs2_dirent *prev, struct gfs2_dirent *cur)
651{
652	u16 cur_rec_len, prev_rec_len;
653
654	if (gfs2_dirent_sentinel(cur)) {
655		gfs2_consist_inode(dip);
656		return;
657	}
658
659	gfs2_trans_add_meta(dip->i_gl, bh);
660
661	/* If there is no prev entry, this is the first entry in the block.
662	   The de_rec_len is already as big as it needs to be.  Just zero
663	   out the inode number and return.  */
664
665	if (!prev) {
666		cur->de_inum.no_addr = 0;
667		cur->de_inum.no_formal_ino = 0;
668		return;
669	}
670
671	/*  Combine this dentry with the previous one.  */
672
673	prev_rec_len = be16_to_cpu(prev->de_rec_len);
674	cur_rec_len = be16_to_cpu(cur->de_rec_len);
675
676	if ((char *)prev + prev_rec_len != (char *)cur)
677		gfs2_consist_inode(dip);
678	if ((char *)cur + cur_rec_len > bh->b_data + bh->b_size)
679		gfs2_consist_inode(dip);
680
681	prev_rec_len += cur_rec_len;
682	prev->de_rec_len = cpu_to_be16(prev_rec_len);
683}
684
685/*
686 * Takes a dent from which to grab space as an argument. Returns the
687 * newly created dent.
688 */
689static struct gfs2_dirent *gfs2_init_dirent(struct inode *inode,
690					    struct gfs2_dirent *dent,
691					    const struct qstr *name,
692					    struct buffer_head *bh)
693{
694	struct gfs2_inode *ip = GFS2_I(inode);
695	struct gfs2_dirent *ndent;
696	unsigned offset = 0, totlen;
697
698	if (!gfs2_dirent_sentinel(dent))
699		offset = GFS2_DIRENT_SIZE(be16_to_cpu(dent->de_name_len));
700	totlen = be16_to_cpu(dent->de_rec_len);
701	BUG_ON(offset + name->len > totlen);
702	gfs2_trans_add_meta(ip->i_gl, bh);
703	ndent = (struct gfs2_dirent *)((char *)dent + offset);
704	dent->de_rec_len = cpu_to_be16(offset);
705	gfs2_qstr2dirent(name, totlen - offset, ndent);
706	return ndent;
707}
708
709static struct gfs2_dirent *gfs2_dirent_alloc(struct inode *inode,
710					     struct buffer_head *bh,
711					     const struct qstr *name)
712{
713	struct gfs2_dirent *dent;
714	dent = gfs2_dirent_scan(inode, bh->b_data, bh->b_size,
715				gfs2_dirent_find_space, name, NULL);
716	if (!dent || IS_ERR(dent))
717		return dent;
718	return gfs2_init_dirent(inode, dent, name, bh);
719}
720
721static int get_leaf(struct gfs2_inode *dip, u64 leaf_no,
722		    struct buffer_head **bhp)
723{
724	int error;
725
726	error = gfs2_meta_read(dip->i_gl, leaf_no, DIO_WAIT, bhp);
727	if (!error && gfs2_metatype_check(GFS2_SB(&dip->i_inode), *bhp, GFS2_METATYPE_LF)) {
728		/* pr_info("block num=%llu\n", leaf_no); */
729		error = -EIO;
730	}
731
732	return error;
733}
734
735/**
736 * get_leaf_nr - Get a leaf number associated with the index
737 * @dip: The GFS2 inode
738 * @index:
739 * @leaf_out:
740 *
741 * Returns: 0 on success, error code otherwise
742 */
743
744static int get_leaf_nr(struct gfs2_inode *dip, u32 index,
745		       u64 *leaf_out)
746{
747	__be64 *hash;
748
749	hash = gfs2_dir_get_hash_table(dip);
750	if (IS_ERR(hash))
751		return PTR_ERR(hash);
752	*leaf_out = be64_to_cpu(*(hash + index));
753	return 0;
754}
755
756static int get_first_leaf(struct gfs2_inode *dip, u32 index,
757			  struct buffer_head **bh_out)
758{
759	u64 leaf_no;
760	int error;
761
762	error = get_leaf_nr(dip, index, &leaf_no);
763	if (!error)
764		error = get_leaf(dip, leaf_no, bh_out);
765
766	return error;
767}
768
769static struct gfs2_dirent *gfs2_dirent_search(struct inode *inode,
770					      const struct qstr *name,
771					      gfs2_dscan_t scan,
772					      struct buffer_head **pbh)
773{
774	struct buffer_head *bh;
775	struct gfs2_dirent *dent;
776	struct gfs2_inode *ip = GFS2_I(inode);
777	int error;
778
779	if (ip->i_diskflags & GFS2_DIF_EXHASH) {
780		struct gfs2_leaf *leaf;
781		unsigned hsize = 1 << ip->i_depth;
782		unsigned index;
783		u64 ln;
784		if (hsize * sizeof(u64) != i_size_read(inode)) {
785			gfs2_consist_inode(ip);
786			return ERR_PTR(-EIO);
787		}
788
789		index = name->hash >> (32 - ip->i_depth);
790		error = get_first_leaf(ip, index, &bh);
791		if (error)
792			return ERR_PTR(error);
793		do {
794			dent = gfs2_dirent_scan(inode, bh->b_data, bh->b_size,
795						scan, name, NULL);
796			if (dent)
797				goto got_dent;
798			leaf = (struct gfs2_leaf *)bh->b_data;
799			ln = be64_to_cpu(leaf->lf_next);
800			brelse(bh);
801			if (!ln)
802				break;
803
804			error = get_leaf(ip, ln, &bh);
805		} while(!error);
806
807		return error ? ERR_PTR(error) : NULL;
808	}
809
810
811	error = gfs2_meta_inode_buffer(ip, &bh);
812	if (error)
813		return ERR_PTR(error);
814	dent = gfs2_dirent_scan(inode, bh->b_data, bh->b_size, scan, name, NULL);
815got_dent:
816	if (unlikely(dent == NULL || IS_ERR(dent))) {
817		brelse(bh);
818		bh = NULL;
819	}
820	*pbh = bh;
821	return dent;
822}
823
824static struct gfs2_leaf *new_leaf(struct inode *inode, struct buffer_head **pbh, u16 depth)
825{
826	struct gfs2_inode *ip = GFS2_I(inode);
827	unsigned int n = 1;
828	u64 bn;
829	int error;
830	struct buffer_head *bh;
831	struct gfs2_leaf *leaf;
832	struct gfs2_dirent *dent;
833	struct qstr name = { .name = "" };
834	struct timespec tv = CURRENT_TIME;
835
836	error = gfs2_alloc_blocks(ip, &bn, &n, 0, NULL);
837	if (error)
838		return NULL;
839	bh = gfs2_meta_new(ip->i_gl, bn);
840	if (!bh)
841		return NULL;
842
843	gfs2_trans_add_unrevoke(GFS2_SB(inode), bn, 1);
844	gfs2_trans_add_meta(ip->i_gl, bh);
845	gfs2_metatype_set(bh, GFS2_METATYPE_LF, GFS2_FORMAT_LF);
846	leaf = (struct gfs2_leaf *)bh->b_data;
847	leaf->lf_depth = cpu_to_be16(depth);
848	leaf->lf_entries = 0;
849	leaf->lf_dirent_format = cpu_to_be32(GFS2_FORMAT_DE);
850	leaf->lf_next = 0;
851	leaf->lf_inode = cpu_to_be64(ip->i_no_addr);
852	leaf->lf_dist = cpu_to_be32(1);
853	leaf->lf_nsec = cpu_to_be32(tv.tv_nsec);
854	leaf->lf_sec = cpu_to_be64(tv.tv_sec);
855	memset(leaf->lf_reserved2, 0, sizeof(leaf->lf_reserved2));
856	dent = (struct gfs2_dirent *)(leaf+1);
857	gfs2_qstr2dirent(&name, bh->b_size - sizeof(struct gfs2_leaf), dent);
858	*pbh = bh;
859	return leaf;
860}
861
862/**
863 * dir_make_exhash - Convert a stuffed directory into an ExHash directory
864 * @dip: The GFS2 inode
865 *
866 * Returns: 0 on success, error code otherwise
867 */
868
869static int dir_make_exhash(struct inode *inode)
870{
871	struct gfs2_inode *dip = GFS2_I(inode);
872	struct gfs2_sbd *sdp = GFS2_SB(inode);
873	struct gfs2_dirent *dent;
874	struct qstr args;
875	struct buffer_head *bh, *dibh;
876	struct gfs2_leaf *leaf;
877	int y;
878	u32 x;
879	__be64 *lp;
880	u64 bn;
881	int error;
882
883	error = gfs2_meta_inode_buffer(dip, &dibh);
884	if (error)
885		return error;
886
887	/*  Turn over a new leaf  */
888
889	leaf = new_leaf(inode, &bh, 0);
890	if (!leaf)
891		return -ENOSPC;
892	bn = bh->b_blocknr;
893
894	gfs2_assert(sdp, dip->i_entries < (1 << 16));
895	leaf->lf_entries = cpu_to_be16(dip->i_entries);
896
897	/*  Copy dirents  */
898
899	gfs2_buffer_copy_tail(bh, sizeof(struct gfs2_leaf), dibh,
900			     sizeof(struct gfs2_dinode));
901
902	/*  Find last entry  */
903
904	x = 0;
905	args.len = bh->b_size - sizeof(struct gfs2_dinode) +
906		   sizeof(struct gfs2_leaf);
907	args.name = bh->b_data;
908	dent = gfs2_dirent_scan(&dip->i_inode, bh->b_data, bh->b_size,
909				gfs2_dirent_last, &args, NULL);
910	if (!dent) {
911		brelse(bh);
912		brelse(dibh);
913		return -EIO;
914	}
915	if (IS_ERR(dent)) {
916		brelse(bh);
917		brelse(dibh);
918		return PTR_ERR(dent);
919	}
920
921	/*  Adjust the last dirent's record length
922	   (Remember that dent still points to the last entry.)  */
923
924	dent->de_rec_len = cpu_to_be16(be16_to_cpu(dent->de_rec_len) +
925		sizeof(struct gfs2_dinode) -
926		sizeof(struct gfs2_leaf));
927
928	brelse(bh);
929
930	/*  We're done with the new leaf block, now setup the new
931	    hash table.  */
932
933	gfs2_trans_add_meta(dip->i_gl, dibh);
934	gfs2_buffer_clear_tail(dibh, sizeof(struct gfs2_dinode));
935
936	lp = (__be64 *)(dibh->b_data + sizeof(struct gfs2_dinode));
937
938	for (x = sdp->sd_hash_ptrs; x--; lp++)
939		*lp = cpu_to_be64(bn);
940
941	i_size_write(inode, sdp->sd_sb.sb_bsize / 2);
942	gfs2_add_inode_blocks(&dip->i_inode, 1);
943	dip->i_diskflags |= GFS2_DIF_EXHASH;
944
945	for (x = sdp->sd_hash_ptrs, y = -1; x; x >>= 1, y++) ;
946	dip->i_depth = y;
947
948	gfs2_dinode_out(dip, dibh->b_data);
949
950	brelse(dibh);
951
952	return 0;
953}
954
955/**
956 * dir_split_leaf - Split a leaf block into two
957 * @dip: The GFS2 inode
958 * @index:
959 * @leaf_no:
960 *
961 * Returns: 0 on success, error code on failure
962 */
963
964static int dir_split_leaf(struct inode *inode, const struct qstr *name)
965{
966	struct gfs2_inode *dip = GFS2_I(inode);
967	struct buffer_head *nbh, *obh, *dibh;
968	struct gfs2_leaf *nleaf, *oleaf;
969	struct gfs2_dirent *dent = NULL, *prev = NULL, *next = NULL, *new;
970	u32 start, len, half_len, divider;
971	u64 bn, leaf_no;
972	__be64 *lp;
973	u32 index;
974	int x, moved = 0;
975	int error;
976
977	index = name->hash >> (32 - dip->i_depth);
978	error = get_leaf_nr(dip, index, &leaf_no);
979	if (error)
980		return error;
981
982	/*  Get the old leaf block  */
983	error = get_leaf(dip, leaf_no, &obh);
984	if (error)
985		return error;
986
987	oleaf = (struct gfs2_leaf *)obh->b_data;
988	if (dip->i_depth == be16_to_cpu(oleaf->lf_depth)) {
989		brelse(obh);
990		return 1; /* can't split */
991	}
992
993	gfs2_trans_add_meta(dip->i_gl, obh);
994
995	nleaf = new_leaf(inode, &nbh, be16_to_cpu(oleaf->lf_depth) + 1);
996	if (!nleaf) {
997		brelse(obh);
998		return -ENOSPC;
999	}
1000	bn = nbh->b_blocknr;
1001
1002	/*  Compute the start and len of leaf pointers in the hash table.  */
1003	len = 1 << (dip->i_depth - be16_to_cpu(oleaf->lf_depth));
1004	half_len = len >> 1;
1005	if (!half_len) {
1006		pr_warn("i_depth %u lf_depth %u index %u\n",
1007			dip->i_depth, be16_to_cpu(oleaf->lf_depth), index);
1008		gfs2_consist_inode(dip);
1009		error = -EIO;
1010		goto fail_brelse;
1011	}
1012
1013	start = (index & ~(len - 1));
1014
1015	/* Change the pointers.
1016	   Don't bother distinguishing stuffed from non-stuffed.
1017	   This code is complicated enough already. */
1018	lp = kmalloc(half_len * sizeof(__be64), GFP_NOFS);
1019	if (!lp) {
1020		error = -ENOMEM;
1021		goto fail_brelse;
1022	}
1023
1024	/*  Change the pointers  */
1025	for (x = 0; x < half_len; x++)
1026		lp[x] = cpu_to_be64(bn);
1027
1028	gfs2_dir_hash_inval(dip);
1029
1030	error = gfs2_dir_write_data(dip, (char *)lp, start * sizeof(u64),
1031				    half_len * sizeof(u64));
1032	if (error != half_len * sizeof(u64)) {
1033		if (error >= 0)
1034			error = -EIO;
1035		goto fail_lpfree;
1036	}
1037
1038	kfree(lp);
1039
1040	/*  Compute the divider  */
1041	divider = (start + half_len) << (32 - dip->i_depth);
1042
1043	/*  Copy the entries  */
1044	dent = (struct gfs2_dirent *)(obh->b_data + sizeof(struct gfs2_leaf));
1045
1046	do {
1047		next = dent;
1048		if (dirent_next(dip, obh, &next))
1049			next = NULL;
1050
1051		if (!gfs2_dirent_sentinel(dent) &&
1052		    be32_to_cpu(dent->de_hash) < divider) {
1053			struct qstr str;
1054			str.name = (char*)(dent+1);
1055			str.len = be16_to_cpu(dent->de_name_len);
1056			str.hash = be32_to_cpu(dent->de_hash);
1057			new = gfs2_dirent_alloc(inode, nbh, &str);
1058			if (IS_ERR(new)) {
1059				error = PTR_ERR(new);
1060				break;
1061			}
1062
1063			new->de_inum = dent->de_inum; /* No endian worries */
1064			new->de_type = dent->de_type; /* No endian worries */
1065			be16_add_cpu(&nleaf->lf_entries, 1);
1066
1067			dirent_del(dip, obh, prev, dent);
1068
1069			if (!oleaf->lf_entries)
1070				gfs2_consist_inode(dip);
1071			be16_add_cpu(&oleaf->lf_entries, -1);
1072
1073			if (!prev)
1074				prev = dent;
1075
1076			moved = 1;
1077		} else {
1078			prev = dent;
1079		}
1080		dent = next;
1081	} while (dent);
1082
1083	oleaf->lf_depth = nleaf->lf_depth;
1084
1085	error = gfs2_meta_inode_buffer(dip, &dibh);
1086	if (!gfs2_assert_withdraw(GFS2_SB(&dip->i_inode), !error)) {
1087		gfs2_trans_add_meta(dip->i_gl, dibh);
1088		gfs2_add_inode_blocks(&dip->i_inode, 1);
1089		gfs2_dinode_out(dip, dibh->b_data);
1090		brelse(dibh);
1091	}
1092
1093	brelse(obh);
1094	brelse(nbh);
1095
1096	return error;
1097
1098fail_lpfree:
1099	kfree(lp);
1100
1101fail_brelse:
1102	brelse(obh);
1103	brelse(nbh);
1104	return error;
1105}
1106
1107/**
1108 * dir_double_exhash - Double size of ExHash table
1109 * @dip: The GFS2 dinode
1110 *
1111 * Returns: 0 on success, error code on failure
1112 */
1113
1114static int dir_double_exhash(struct gfs2_inode *dip)
1115{
1116	struct buffer_head *dibh;
1117	u32 hsize;
1118	u32 hsize_bytes;
1119	__be64 *hc;
1120	__be64 *hc2, *h;
1121	int x;
1122	int error = 0;
1123
1124	hsize = 1 << dip->i_depth;
1125	hsize_bytes = hsize * sizeof(__be64);
1126
1127	hc = gfs2_dir_get_hash_table(dip);
1128	if (IS_ERR(hc))
1129		return PTR_ERR(hc);
1130
1131	hc2 = kmalloc(hsize_bytes * 2, GFP_NOFS | __GFP_NOWARN);
1132	if (hc2 == NULL)
1133		hc2 = __vmalloc(hsize_bytes * 2, GFP_NOFS, PAGE_KERNEL);
1134
1135	if (!hc2)
1136		return -ENOMEM;
1137
1138	h = hc2;
1139	error = gfs2_meta_inode_buffer(dip, &dibh);
1140	if (error)
1141		goto out_kfree;
1142
1143	for (x = 0; x < hsize; x++) {
1144		*h++ = *hc;
1145		*h++ = *hc;
1146		hc++;
1147	}
1148
1149	error = gfs2_dir_write_data(dip, (char *)hc2, 0, hsize_bytes * 2);
1150	if (error != (hsize_bytes * 2))
1151		goto fail;
1152
1153	gfs2_dir_hash_inval(dip);
1154	dip->i_hash_cache = hc2;
1155	dip->i_depth++;
1156	gfs2_dinode_out(dip, dibh->b_data);
1157	brelse(dibh);
1158	return 0;
1159
1160fail:
1161	/* Replace original hash table & size */
1162	gfs2_dir_write_data(dip, (char *)hc, 0, hsize_bytes);
1163	i_size_write(&dip->i_inode, hsize_bytes);
1164	gfs2_dinode_out(dip, dibh->b_data);
1165	brelse(dibh);
1166out_kfree:
1167	kvfree(hc2);
1168	return error;
1169}
1170
1171/**
1172 * compare_dents - compare directory entries by hash value
1173 * @a: first dent
1174 * @b: second dent
1175 *
1176 * When comparing the hash entries of @a to @b:
1177 *   gt: returns 1
1178 *   lt: returns -1
1179 *   eq: returns 0
1180 */
1181
1182static int compare_dents(const void *a, const void *b)
1183{
1184	const struct gfs2_dirent *dent_a, *dent_b;
1185	u32 hash_a, hash_b;
1186	int ret = 0;
1187
1188	dent_a = *(const struct gfs2_dirent **)a;
1189	hash_a = be32_to_cpu(dent_a->de_hash);
1190
1191	dent_b = *(const struct gfs2_dirent **)b;
1192	hash_b = be32_to_cpu(dent_b->de_hash);
1193
1194	if (hash_a > hash_b)
1195		ret = 1;
1196	else if (hash_a < hash_b)
1197		ret = -1;
1198	else {
1199		unsigned int len_a = be16_to_cpu(dent_a->de_name_len);
1200		unsigned int len_b = be16_to_cpu(dent_b->de_name_len);
1201
1202		if (len_a > len_b)
1203			ret = 1;
1204		else if (len_a < len_b)
1205			ret = -1;
1206		else
1207			ret = memcmp(dent_a + 1, dent_b + 1, len_a);
1208	}
1209
1210	return ret;
1211}
1212
1213/**
1214 * do_filldir_main - read out directory entries
1215 * @dip: The GFS2 inode
1216 * @ctx: what to feed the entries to
1217 * @darr: an array of struct gfs2_dirent pointers to read
1218 * @entries: the number of entries in darr
1219 * @copied: pointer to int that's non-zero if a entry has been copied out
1220 *
1221 * Jump through some hoops to make sure that if there are hash collsions,
1222 * they are read out at the beginning of a buffer.  We want to minimize
1223 * the possibility that they will fall into different readdir buffers or
1224 * that someone will want to seek to that location.
1225 *
1226 * Returns: errno, >0 if the actor tells you to stop
1227 */
1228
1229static int do_filldir_main(struct gfs2_inode *dip, struct dir_context *ctx,
1230			   const struct gfs2_dirent **darr, u32 entries,
1231			   int *copied)
1232{
1233	const struct gfs2_dirent *dent, *dent_next;
1234	u64 off, off_next;
1235	unsigned int x, y;
1236	int run = 0;
1237
1238	sort(darr, entries, sizeof(struct gfs2_dirent *), compare_dents, NULL);
1239
1240	dent_next = darr[0];
1241	off_next = be32_to_cpu(dent_next->de_hash);
1242	off_next = gfs2_disk_hash2offset(off_next);
1243
1244	for (x = 0, y = 1; x < entries; x++, y++) {
1245		dent = dent_next;
1246		off = off_next;
1247
1248		if (y < entries) {
1249			dent_next = darr[y];
1250			off_next = be32_to_cpu(dent_next->de_hash);
1251			off_next = gfs2_disk_hash2offset(off_next);
1252
1253			if (off < ctx->pos)
1254				continue;
1255			ctx->pos = off;
1256
1257			if (off_next == off) {
1258				if (*copied && !run)
1259					return 1;
1260				run = 1;
1261			} else
1262				run = 0;
1263		} else {
1264			if (off < ctx->pos)
1265				continue;
1266			ctx->pos = off;
1267		}
1268
1269		if (!dir_emit(ctx, (const char *)(dent + 1),
1270				be16_to_cpu(dent->de_name_len),
1271				be64_to_cpu(dent->de_inum.no_addr),
1272				be16_to_cpu(dent->de_type)))
1273			return 1;
1274
1275		*copied = 1;
1276	}
1277
1278	/* Increment the ctx->pos by one, so the next time we come into the
1279	   do_filldir fxn, we get the next entry instead of the last one in the
1280	   current leaf */
1281
1282	ctx->pos++;
1283
1284	return 0;
1285}
1286
1287static void *gfs2_alloc_sort_buffer(unsigned size)
1288{
1289	void *ptr = NULL;
1290
1291	if (size < KMALLOC_MAX_SIZE)
1292		ptr = kmalloc(size, GFP_NOFS | __GFP_NOWARN);
1293	if (!ptr)
1294		ptr = __vmalloc(size, GFP_NOFS, PAGE_KERNEL);
1295	return ptr;
1296}
1297
1298static int gfs2_dir_read_leaf(struct inode *inode, struct dir_context *ctx,
1299			      int *copied, unsigned *depth,
1300			      u64 leaf_no)
1301{
1302	struct gfs2_inode *ip = GFS2_I(inode);
1303	struct gfs2_sbd *sdp = GFS2_SB(inode);
1304	struct buffer_head *bh;
1305	struct gfs2_leaf *lf;
1306	unsigned entries = 0, entries2 = 0;
1307	unsigned leaves = 0;
1308	const struct gfs2_dirent **darr, *dent;
1309	struct dirent_gather g;
1310	struct buffer_head **larr;
1311	int leaf = 0;
1312	int error, i;
1313	u64 lfn = leaf_no;
1314
1315	do {
1316		error = get_leaf(ip, lfn, &bh);
1317		if (error)
1318			goto out;
1319		lf = (struct gfs2_leaf *)bh->b_data;
1320		if (leaves == 0)
1321			*depth = be16_to_cpu(lf->lf_depth);
1322		entries += be16_to_cpu(lf->lf_entries);
1323		leaves++;
1324		lfn = be64_to_cpu(lf->lf_next);
1325		brelse(bh);
1326	} while(lfn);
1327
1328	if (!entries)
1329		return 0;
1330
1331	error = -ENOMEM;
1332	/*
1333	 * The extra 99 entries are not normally used, but are a buffer
1334	 * zone in case the number of entries in the leaf is corrupt.
1335	 * 99 is the maximum number of entries that can fit in a single
1336	 * leaf block.
1337	 */
1338	larr = gfs2_alloc_sort_buffer((leaves + entries + 99) * sizeof(void *));
1339	if (!larr)
1340		goto out;
1341	darr = (const struct gfs2_dirent **)(larr + leaves);
1342	g.pdent = darr;
1343	g.offset = 0;
1344	lfn = leaf_no;
1345
1346	do {
1347		error = get_leaf(ip, lfn, &bh);
1348		if (error)
1349			goto out_free;
1350		lf = (struct gfs2_leaf *)bh->b_data;
1351		lfn = be64_to_cpu(lf->lf_next);
1352		if (lf->lf_entries) {
1353			entries2 += be16_to_cpu(lf->lf_entries);
1354			dent = gfs2_dirent_scan(inode, bh->b_data, bh->b_size,
1355						gfs2_dirent_gather, NULL, &g);
1356			error = PTR_ERR(dent);
1357			if (IS_ERR(dent))
1358				goto out_free;
1359			if (entries2 != g.offset) {
1360				fs_warn(sdp, "Number of entries corrupt in dir "
1361						"leaf %llu, entries2 (%u) != "
1362						"g.offset (%u)\n",
1363					(unsigned long long)bh->b_blocknr,
1364					entries2, g.offset);
1365
1366				error = -EIO;
1367				goto out_free;
1368			}
1369			error = 0;
1370			larr[leaf++] = bh;
1371		} else {
1372			brelse(bh);
1373		}
1374	} while(lfn);
1375
1376	BUG_ON(entries2 != entries);
1377	error = do_filldir_main(ip, ctx, darr, entries, copied);
1378out_free:
1379	for(i = 0; i < leaf; i++)
1380		brelse(larr[i]);
1381	kvfree(larr);
1382out:
1383	return error;
1384}
1385
1386/**
1387 * gfs2_dir_readahead - Issue read-ahead requests for leaf blocks.
1388 *
1389 * Note: we can't calculate each index like dir_e_read can because we don't
1390 * have the leaf, and therefore we don't have the depth, and therefore we
1391 * don't have the length. So we have to just read enough ahead to make up
1392 * for the loss of information.
1393 */
1394static void gfs2_dir_readahead(struct inode *inode, unsigned hsize, u32 index,
1395			       struct file_ra_state *f_ra)
1396{
1397	struct gfs2_inode *ip = GFS2_I(inode);
1398	struct gfs2_glock *gl = ip->i_gl;
1399	struct buffer_head *bh;
1400	u64 blocknr = 0, last;
1401	unsigned count;
1402
1403	/* First check if we've already read-ahead for the whole range. */
1404	if (index + MAX_RA_BLOCKS < f_ra->start)
1405		return;
1406
1407	f_ra->start = max((pgoff_t)index, f_ra->start);
1408	for (count = 0; count < MAX_RA_BLOCKS; count++) {
1409		if (f_ra->start >= hsize) /* if exceeded the hash table */
1410			break;
1411
1412		last = blocknr;
1413		blocknr = be64_to_cpu(ip->i_hash_cache[f_ra->start]);
1414		f_ra->start++;
1415		if (blocknr == last)
1416			continue;
1417
1418		bh = gfs2_getbuf(gl, blocknr, 1);
1419		if (trylock_buffer(bh)) {
1420			if (buffer_uptodate(bh)) {
1421				unlock_buffer(bh);
1422				brelse(bh);
1423				continue;
1424			}
1425			bh->b_end_io = end_buffer_read_sync;
1426			submit_bh(READA | REQ_META, bh);
1427			continue;
1428		}
1429		brelse(bh);
1430	}
1431}
1432
1433/**
1434 * dir_e_read - Reads the entries from a directory into a filldir buffer
1435 * @dip: dinode pointer
1436 * @ctx: actor to feed the entries to
1437 *
1438 * Returns: errno
1439 */
1440
1441static int dir_e_read(struct inode *inode, struct dir_context *ctx,
1442		      struct file_ra_state *f_ra)
1443{
1444	struct gfs2_inode *dip = GFS2_I(inode);
1445	u32 hsize, len = 0;
1446	u32 hash, index;
1447	__be64 *lp;
1448	int copied = 0;
1449	int error = 0;
1450	unsigned depth = 0;
1451
1452	hsize = 1 << dip->i_depth;
1453	hash = gfs2_dir_offset2hash(ctx->pos);
1454	index = hash >> (32 - dip->i_depth);
1455
1456	if (dip->i_hash_cache == NULL)
1457		f_ra->start = 0;
1458	lp = gfs2_dir_get_hash_table(dip);
1459	if (IS_ERR(lp))
1460		return PTR_ERR(lp);
1461
1462	gfs2_dir_readahead(inode, hsize, index, f_ra);
1463
1464	while (index < hsize) {
1465		error = gfs2_dir_read_leaf(inode, ctx,
1466					   &copied, &depth,
1467					   be64_to_cpu(lp[index]));
1468		if (error)
1469			break;
1470
1471		len = 1 << (dip->i_depth - depth);
1472		index = (index & ~(len - 1)) + len;
1473	}
1474
1475	if (error > 0)
1476		error = 0;
1477	return error;
1478}
1479
1480int gfs2_dir_read(struct inode *inode, struct dir_context *ctx,
1481		  struct file_ra_state *f_ra)
1482{
1483	struct gfs2_inode *dip = GFS2_I(inode);
1484	struct gfs2_sbd *sdp = GFS2_SB(inode);
1485	struct dirent_gather g;
1486	const struct gfs2_dirent **darr, *dent;
1487	struct buffer_head *dibh;
1488	int copied = 0;
1489	int error;
1490
1491	if (!dip->i_entries)
1492		return 0;
1493
1494	if (dip->i_diskflags & GFS2_DIF_EXHASH)
1495		return dir_e_read(inode, ctx, f_ra);
1496
1497	if (!gfs2_is_stuffed(dip)) {
1498		gfs2_consist_inode(dip);
1499		return -EIO;
1500	}
1501
1502	error = gfs2_meta_inode_buffer(dip, &dibh);
1503	if (error)
1504		return error;
1505
1506	error = -ENOMEM;
1507	/* 96 is max number of dirents which can be stuffed into an inode */
1508	darr = kmalloc(96 * sizeof(struct gfs2_dirent *), GFP_NOFS);
1509	if (darr) {
1510		g.pdent = darr;
1511		g.offset = 0;
1512		dent = gfs2_dirent_scan(inode, dibh->b_data, dibh->b_size,
1513					gfs2_dirent_gather, NULL, &g);
1514		if (IS_ERR(dent)) {
1515			error = PTR_ERR(dent);
1516			goto out;
1517		}
1518		if (dip->i_entries != g.offset) {
1519			fs_warn(sdp, "Number of entries corrupt in dir %llu, "
1520				"ip->i_entries (%u) != g.offset (%u)\n",
1521				(unsigned long long)dip->i_no_addr,
1522				dip->i_entries,
1523				g.offset);
1524			error = -EIO;
1525			goto out;
1526		}
1527		error = do_filldir_main(dip, ctx, darr,
1528					dip->i_entries, &copied);
1529out:
1530		kfree(darr);
1531	}
1532
1533	if (error > 0)
1534		error = 0;
1535
1536	brelse(dibh);
1537
1538	return error;
1539}
1540
1541/**
1542 * gfs2_dir_search - Search a directory
1543 * @dip: The GFS2 dir inode
1544 * @name: The name we are looking up
1545 * @fail_on_exist: Fail if the name exists rather than looking it up
1546 *
1547 * This routine searches a directory for a file or another directory.
1548 * Assumes a glock is held on dip.
1549 *
1550 * Returns: errno
1551 */
1552
1553struct inode *gfs2_dir_search(struct inode *dir, const struct qstr *name,
1554			      bool fail_on_exist)
1555{
1556	struct buffer_head *bh;
1557	struct gfs2_dirent *dent;
1558	u64 addr, formal_ino;
1559	u16 dtype;
1560
1561	dent = gfs2_dirent_search(dir, name, gfs2_dirent_find, &bh);
1562	if (dent) {
1563		if (IS_ERR(dent))
1564			return ERR_CAST(dent);
1565		dtype = be16_to_cpu(dent->de_type);
1566		addr = be64_to_cpu(dent->de_inum.no_addr);
1567		formal_ino = be64_to_cpu(dent->de_inum.no_formal_ino);
1568		brelse(bh);
1569		if (fail_on_exist)
1570			return ERR_PTR(-EEXIST);
1571		return gfs2_inode_lookup(dir->i_sb, dtype, addr, formal_ino, 0);
1572	}
1573	return ERR_PTR(-ENOENT);
1574}
1575
1576int gfs2_dir_check(struct inode *dir, const struct qstr *name,
1577		   const struct gfs2_inode *ip)
1578{
1579	struct buffer_head *bh;
1580	struct gfs2_dirent *dent;
1581	int ret = -ENOENT;
1582
1583	dent = gfs2_dirent_search(dir, name, gfs2_dirent_find, &bh);
1584	if (dent) {
1585		if (IS_ERR(dent))
1586			return PTR_ERR(dent);
1587		if (ip) {
1588			if (be64_to_cpu(dent->de_inum.no_addr) != ip->i_no_addr)
1589				goto out;
1590			if (be64_to_cpu(dent->de_inum.no_formal_ino) !=
1591			    ip->i_no_formal_ino)
1592				goto out;
1593			if (unlikely(IF2DT(ip->i_inode.i_mode) !=
1594			    be16_to_cpu(dent->de_type))) {
1595				gfs2_consist_inode(GFS2_I(dir));
1596				ret = -EIO;
1597				goto out;
1598			}
1599		}
1600		ret = 0;
1601out:
1602		brelse(bh);
1603	}
1604	return ret;
1605}
1606
1607/**
1608 * dir_new_leaf - Add a new leaf onto hash chain
1609 * @inode: The directory
1610 * @name: The name we are adding
1611 *
1612 * This adds a new dir leaf onto an existing leaf when there is not
1613 * enough space to add a new dir entry. This is a last resort after
1614 * we've expanded the hash table to max size and also split existing
1615 * leaf blocks, so it will only occur for very large directories.
1616 *
1617 * The dist parameter is set to 1 for leaf blocks directly attached
1618 * to the hash table, 2 for one layer of indirection, 3 for two layers
1619 * etc. We are thus able to tell the difference between an old leaf
1620 * with dist set to zero (i.e. "don't know") and a new one where we
1621 * set this information for debug/fsck purposes.
1622 *
1623 * Returns: 0 on success, or -ve on error
1624 */
1625
1626static int dir_new_leaf(struct inode *inode, const struct qstr *name)
1627{
1628	struct buffer_head *bh, *obh;
1629	struct gfs2_inode *ip = GFS2_I(inode);
1630	struct gfs2_leaf *leaf, *oleaf;
1631	u32 dist = 1;
1632	int error;
1633	u32 index;
1634	u64 bn;
1635
1636	index = name->hash >> (32 - ip->i_depth);
1637	error = get_first_leaf(ip, index, &obh);
1638	if (error)
1639		return error;
1640	do {
1641		dist++;
1642		oleaf = (struct gfs2_leaf *)obh->b_data;
1643		bn = be64_to_cpu(oleaf->lf_next);
1644		if (!bn)
1645			break;
1646		brelse(obh);
1647		error = get_leaf(ip, bn, &obh);
1648		if (error)
1649			return error;
1650	} while(1);
1651
1652	gfs2_trans_add_meta(ip->i_gl, obh);
1653
1654	leaf = new_leaf(inode, &bh, be16_to_cpu(oleaf->lf_depth));
1655	if (!leaf) {
1656		brelse(obh);
1657		return -ENOSPC;
1658	}
1659	leaf->lf_dist = cpu_to_be32(dist);
1660	oleaf->lf_next = cpu_to_be64(bh->b_blocknr);
1661	brelse(bh);
1662	brelse(obh);
1663
1664	error = gfs2_meta_inode_buffer(ip, &bh);
1665	if (error)
1666		return error;
1667	gfs2_trans_add_meta(ip->i_gl, bh);
1668	gfs2_add_inode_blocks(&ip->i_inode, 1);
1669	gfs2_dinode_out(ip, bh->b_data);
1670	brelse(bh);
1671	return 0;
1672}
1673
1674static u16 gfs2_inode_ra_len(const struct gfs2_inode *ip)
1675{
1676	u64 where = ip->i_no_addr + 1;
1677	if (ip->i_eattr == where)
1678		return 1;
1679	return 0;
1680}
1681
1682/**
1683 * gfs2_dir_add - Add new filename into directory
1684 * @inode: The directory inode
1685 * @name: The new name
1686 * @nip: The GFS2 inode to be linked in to the directory
1687 * @da: The directory addition info
1688 *
1689 * If the call to gfs2_diradd_alloc_required resulted in there being
1690 * no need to allocate any new directory blocks, then it will contain
1691 * a pointer to the directory entry and the bh in which it resides. We
1692 * can use that without having to repeat the search. If there was no
1693 * free space, then we must now create more space.
1694 *
1695 * Returns: 0 on success, error code on failure
1696 */
1697
1698int gfs2_dir_add(struct inode *inode, const struct qstr *name,
1699		 const struct gfs2_inode *nip, struct gfs2_diradd *da)
1700{
1701	struct gfs2_inode *ip = GFS2_I(inode);
1702	struct buffer_head *bh = da->bh;
1703	struct gfs2_dirent *dent = da->dent;
1704	struct timespec tv;
1705	struct gfs2_leaf *leaf;
1706	int error;
1707
1708	while(1) {
1709		if (da->bh == NULL) {
1710			dent = gfs2_dirent_search(inode, name,
1711						  gfs2_dirent_find_space, &bh);
1712		}
1713		if (dent) {
1714			if (IS_ERR(dent))
1715				return PTR_ERR(dent);
1716			dent = gfs2_init_dirent(inode, dent, name, bh);
1717			gfs2_inum_out(nip, dent);
1718			dent->de_type = cpu_to_be16(IF2DT(nip->i_inode.i_mode));
1719			dent->de_rahead = cpu_to_be16(gfs2_inode_ra_len(nip));
1720			tv = CURRENT_TIME;
1721			if (ip->i_diskflags & GFS2_DIF_EXHASH) {
1722				leaf = (struct gfs2_leaf *)bh->b_data;
1723				be16_add_cpu(&leaf->lf_entries, 1);
1724				leaf->lf_nsec = cpu_to_be32(tv.tv_nsec);
1725				leaf->lf_sec = cpu_to_be64(tv.tv_sec);
1726			}
1727			da->dent = NULL;
1728			da->bh = NULL;
1729			brelse(bh);
1730			ip->i_entries++;
1731			ip->i_inode.i_mtime = ip->i_inode.i_ctime = tv;
1732			if (S_ISDIR(nip->i_inode.i_mode))
1733				inc_nlink(&ip->i_inode);
1734			mark_inode_dirty(inode);
1735			error = 0;
1736			break;
1737		}
1738		if (!(ip->i_diskflags & GFS2_DIF_EXHASH)) {
1739			error = dir_make_exhash(inode);
1740			if (error)
1741				break;
1742			continue;
1743		}
1744		error = dir_split_leaf(inode, name);
1745		if (error == 0)
1746			continue;
1747		if (error < 0)
1748			break;
1749		if (ip->i_depth < GFS2_DIR_MAX_DEPTH) {
1750			error = dir_double_exhash(ip);
1751			if (error)
1752				break;
1753			error = dir_split_leaf(inode, name);
1754			if (error < 0)
1755				break;
1756			if (error == 0)
1757				continue;
1758		}
1759		error = dir_new_leaf(inode, name);
1760		if (!error)
1761			continue;
1762		error = -ENOSPC;
1763		break;
1764	}
1765	return error;
1766}
1767
1768
1769/**
1770 * gfs2_dir_del - Delete a directory entry
1771 * @dip: The GFS2 inode
1772 * @filename: The filename
1773 *
1774 * Returns: 0 on success, error code on failure
1775 */
1776
1777int gfs2_dir_del(struct gfs2_inode *dip, const struct dentry *dentry)
1778{
1779	const struct qstr *name = &dentry->d_name;
1780	struct gfs2_dirent *dent, *prev = NULL;
1781	struct buffer_head *bh;
1782	struct timespec tv = CURRENT_TIME;
1783
1784	/* Returns _either_ the entry (if its first in block) or the
1785	   previous entry otherwise */
1786	dent = gfs2_dirent_search(&dip->i_inode, name, gfs2_dirent_prev, &bh);
1787	if (!dent) {
1788		gfs2_consist_inode(dip);
1789		return -EIO;
1790	}
1791	if (IS_ERR(dent)) {
1792		gfs2_consist_inode(dip);
1793		return PTR_ERR(dent);
1794	}
1795	/* If not first in block, adjust pointers accordingly */
1796	if (gfs2_dirent_find(dent, name, NULL) == 0) {
1797		prev = dent;
1798		dent = (struct gfs2_dirent *)((char *)dent + be16_to_cpu(prev->de_rec_len));
1799	}
1800
1801	dirent_del(dip, bh, prev, dent);
1802	if (dip->i_diskflags & GFS2_DIF_EXHASH) {
1803		struct gfs2_leaf *leaf = (struct gfs2_leaf *)bh->b_data;
1804		u16 entries = be16_to_cpu(leaf->lf_entries);
1805		if (!entries)
1806			gfs2_consist_inode(dip);
1807		leaf->lf_entries = cpu_to_be16(--entries);
1808		leaf->lf_nsec = cpu_to_be32(tv.tv_nsec);
1809		leaf->lf_sec = cpu_to_be64(tv.tv_sec);
1810	}
1811	brelse(bh);
1812
1813	if (!dip->i_entries)
1814		gfs2_consist_inode(dip);
1815	dip->i_entries--;
1816	dip->i_inode.i_mtime = dip->i_inode.i_ctime = tv;
1817	if (d_is_dir(dentry))
1818		drop_nlink(&dip->i_inode);
1819	mark_inode_dirty(&dip->i_inode);
1820
1821	return 0;
1822}
1823
1824/**
1825 * gfs2_dir_mvino - Change inode number of directory entry
1826 * @dip: The GFS2 inode
1827 * @filename:
1828 * @new_inode:
1829 *
1830 * This routine changes the inode number of a directory entry.  It's used
1831 * by rename to change ".." when a directory is moved.
1832 * Assumes a glock is held on dvp.
1833 *
1834 * Returns: errno
1835 */
1836
1837int gfs2_dir_mvino(struct gfs2_inode *dip, const struct qstr *filename,
1838		   const struct gfs2_inode *nip, unsigned int new_type)
1839{
1840	struct buffer_head *bh;
1841	struct gfs2_dirent *dent;
1842	int error;
1843
1844	dent = gfs2_dirent_search(&dip->i_inode, filename, gfs2_dirent_find, &bh);
1845	if (!dent) {
1846		gfs2_consist_inode(dip);
1847		return -EIO;
1848	}
1849	if (IS_ERR(dent))
1850		return PTR_ERR(dent);
1851
1852	gfs2_trans_add_meta(dip->i_gl, bh);
1853	gfs2_inum_out(nip, dent);
1854	dent->de_type = cpu_to_be16(new_type);
1855
1856	if (dip->i_diskflags & GFS2_DIF_EXHASH) {
1857		brelse(bh);
1858		error = gfs2_meta_inode_buffer(dip, &bh);
1859		if (error)
1860			return error;
1861		gfs2_trans_add_meta(dip->i_gl, bh);
1862	}
1863
1864	dip->i_inode.i_mtime = dip->i_inode.i_ctime = CURRENT_TIME;
1865	gfs2_dinode_out(dip, bh->b_data);
1866	brelse(bh);
1867	return 0;
1868}
1869
1870/**
1871 * leaf_dealloc - Deallocate a directory leaf
1872 * @dip: the directory
1873 * @index: the hash table offset in the directory
1874 * @len: the number of pointers to this leaf
1875 * @leaf_no: the leaf number
1876 * @leaf_bh: buffer_head for the starting leaf
1877 * last_dealloc: 1 if this is the final dealloc for the leaf, else 0
1878 *
1879 * Returns: errno
1880 */
1881
1882static int leaf_dealloc(struct gfs2_inode *dip, u32 index, u32 len,
1883			u64 leaf_no, struct buffer_head *leaf_bh,
1884			int last_dealloc)
1885{
1886	struct gfs2_sbd *sdp = GFS2_SB(&dip->i_inode);
1887	struct gfs2_leaf *tmp_leaf;
1888	struct gfs2_rgrp_list rlist;
1889	struct buffer_head *bh, *dibh;
1890	u64 blk, nblk;
1891	unsigned int rg_blocks = 0, l_blocks = 0;
1892	char *ht;
1893	unsigned int x, size = len * sizeof(u64);
1894	int error;
1895
1896	error = gfs2_rindex_update(sdp);
1897	if (error)
1898		return error;
1899
1900	memset(&rlist, 0, sizeof(struct gfs2_rgrp_list));
1901
1902	ht = kzalloc(size, GFP_NOFS | __GFP_NOWARN);
1903	if (ht == NULL)
1904		ht = __vmalloc(size, GFP_NOFS | __GFP_NOWARN | __GFP_ZERO,
1905			       PAGE_KERNEL);
1906	if (!ht)
1907		return -ENOMEM;
1908
1909	error = gfs2_quota_hold(dip, NO_UID_QUOTA_CHANGE, NO_GID_QUOTA_CHANGE);
1910	if (error)
1911		goto out;
1912
1913	/*  Count the number of leaves  */
1914	bh = leaf_bh;
1915
1916	for (blk = leaf_no; blk; blk = nblk) {
1917		if (blk != leaf_no) {
1918			error = get_leaf(dip, blk, &bh);
1919			if (error)
1920				goto out_rlist;
1921		}
1922		tmp_leaf = (struct gfs2_leaf *)bh->b_data;
1923		nblk = be64_to_cpu(tmp_leaf->lf_next);
1924		if (blk != leaf_no)
1925			brelse(bh);
1926
1927		gfs2_rlist_add(dip, &rlist, blk);
1928		l_blocks++;
1929	}
1930
1931	gfs2_rlist_alloc(&rlist, LM_ST_EXCLUSIVE);
1932
1933	for (x = 0; x < rlist.rl_rgrps; x++) {
1934		struct gfs2_rgrpd *rgd;
1935		rgd = rlist.rl_ghs[x].gh_gl->gl_object;
1936		rg_blocks += rgd->rd_length;
1937	}
1938
1939	error = gfs2_glock_nq_m(rlist.rl_rgrps, rlist.rl_ghs);
1940	if (error)
1941		goto out_rlist;
1942
1943	error = gfs2_trans_begin(sdp,
1944			rg_blocks + (DIV_ROUND_UP(size, sdp->sd_jbsize) + 1) +
1945			RES_DINODE + RES_STATFS + RES_QUOTA, l_blocks);
1946	if (error)
1947		goto out_rg_gunlock;
1948
1949	bh = leaf_bh;
1950
1951	for (blk = leaf_no; blk; blk = nblk) {
1952		if (blk != leaf_no) {
1953			error = get_leaf(dip, blk, &bh);
1954			if (error)
1955				goto out_end_trans;
1956		}
1957		tmp_leaf = (struct gfs2_leaf *)bh->b_data;
1958		nblk = be64_to_cpu(tmp_leaf->lf_next);
1959		if (blk != leaf_no)
1960			brelse(bh);
1961
1962		gfs2_free_meta(dip, blk, 1);
1963		gfs2_add_inode_blocks(&dip->i_inode, -1);
1964	}
1965
1966	error = gfs2_dir_write_data(dip, ht, index * sizeof(u64), size);
1967	if (error != size) {
1968		if (error >= 0)
1969			error = -EIO;
1970		goto out_end_trans;
1971	}
1972
1973	error = gfs2_meta_inode_buffer(dip, &dibh);
1974	if (error)
1975		goto out_end_trans;
1976
1977	gfs2_trans_add_meta(dip->i_gl, dibh);
1978	/* On the last dealloc, make this a regular file in case we crash.
1979	   (We don't want to free these blocks a second time.)  */
1980	if (last_dealloc)
1981		dip->i_inode.i_mode = S_IFREG;
1982	gfs2_dinode_out(dip, dibh->b_data);
1983	brelse(dibh);
1984
1985out_end_trans:
1986	gfs2_trans_end(sdp);
1987out_rg_gunlock:
1988	gfs2_glock_dq_m(rlist.rl_rgrps, rlist.rl_ghs);
1989out_rlist:
1990	gfs2_rlist_free(&rlist);
1991	gfs2_quota_unhold(dip);
1992out:
1993	kvfree(ht);
1994	return error;
1995}
1996
1997/**
1998 * gfs2_dir_exhash_dealloc - free all the leaf blocks in a directory
1999 * @dip: the directory
2000 *
2001 * Dealloc all on-disk directory leaves to FREEMETA state
2002 * Change on-disk inode type to "regular file"
2003 *
2004 * Returns: errno
2005 */
2006
2007int gfs2_dir_exhash_dealloc(struct gfs2_inode *dip)
2008{
2009	struct buffer_head *bh;
2010	struct gfs2_leaf *leaf;
2011	u32 hsize, len;
2012	u32 index = 0, next_index;
2013	__be64 *lp;
2014	u64 leaf_no;
2015	int error = 0, last;
2016
2017	hsize = 1 << dip->i_depth;
2018
2019	lp = gfs2_dir_get_hash_table(dip);
2020	if (IS_ERR(lp))
2021		return PTR_ERR(lp);
2022
2023	while (index < hsize) {
2024		leaf_no = be64_to_cpu(lp[index]);
2025		if (leaf_no) {
2026			error = get_leaf(dip, leaf_no, &bh);
2027			if (error)
2028				goto out;
2029			leaf = (struct gfs2_leaf *)bh->b_data;
2030			len = 1 << (dip->i_depth - be16_to_cpu(leaf->lf_depth));
2031
2032			next_index = (index & ~(len - 1)) + len;
2033			last = ((next_index >= hsize) ? 1 : 0);
2034			error = leaf_dealloc(dip, index, len, leaf_no, bh,
2035					     last);
2036			brelse(bh);
2037			if (error)
2038				goto out;
2039			index = next_index;
2040		} else
2041			index++;
2042	}
2043
2044	if (index != hsize) {
2045		gfs2_consist_inode(dip);
2046		error = -EIO;
2047	}
2048
2049out:
2050
2051	return error;
2052}
2053
2054/**
2055 * gfs2_diradd_alloc_required - find if adding entry will require an allocation
2056 * @ip: the file being written to
2057 * @filname: the filename that's going to be added
2058 * @da: The structure to return dir alloc info
2059 *
2060 * Returns: 0 if ok, -ve on error
2061 */
2062
2063int gfs2_diradd_alloc_required(struct inode *inode, const struct qstr *name,
2064			       struct gfs2_diradd *da)
2065{
2066	struct gfs2_inode *ip = GFS2_I(inode);
2067	struct gfs2_sbd *sdp = GFS2_SB(inode);
2068	const unsigned int extra = sizeof(struct gfs2_dinode) - sizeof(struct gfs2_leaf);
2069	struct gfs2_dirent *dent;
2070	struct buffer_head *bh;
2071
2072	da->nr_blocks = 0;
2073	da->bh = NULL;
2074	da->dent = NULL;
2075
2076	dent = gfs2_dirent_search(inode, name, gfs2_dirent_find_space, &bh);
2077	if (!dent) {
2078		da->nr_blocks = sdp->sd_max_dirres;
2079		if (!(ip->i_diskflags & GFS2_DIF_EXHASH) &&
2080		    (GFS2_DIRENT_SIZE(name->len) < extra))
2081			da->nr_blocks = 1;
2082		return 0;
2083	}
2084	if (IS_ERR(dent))
2085		return PTR_ERR(dent);
2086
2087	if (da->save_loc) {
2088		da->bh = bh;
2089		da->dent = dent;
2090	} else {
2091		brelse(bh);
2092	}
2093	return 0;
2094}
2095
2096