Root/fs/ext4/dir.c

1/*
2 * linux/fs/ext4/dir.c
3 *
4 * Copyright (C) 1992, 1993, 1994, 1995
5 * Remy Card (card@masi.ibp.fr)
6 * Laboratoire MASI - Institut Blaise Pascal
7 * Universite Pierre et Marie Curie (Paris VI)
8 *
9 * from
10 *
11 * linux/fs/minix/dir.c
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 *
15 * ext4 directory handling functions
16 *
17 * Big-endian to little-endian byte-swapping/bitmaps by
18 * David S. Miller (davem@caip.rutgers.edu), 1995
19 *
20 * Hash Tree Directory indexing (c) 2001 Daniel Phillips
21 *
22 */
23
24#include <linux/fs.h>
25#include <linux/jbd2.h>
26#include <linux/buffer_head.h>
27#include <linux/slab.h>
28#include <linux/rbtree.h>
29#include "ext4.h"
30
31static unsigned char ext4_filetype_table[] = {
32    DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
33};
34
35static int ext4_readdir(struct file *, void *, filldir_t);
36static int ext4_dx_readdir(struct file *filp,
37               void *dirent, filldir_t filldir);
38static int ext4_release_dir(struct inode *inode,
39                struct file *filp);
40
41const struct file_operations ext4_dir_operations = {
42    .llseek = generic_file_llseek,
43    .read = generic_read_dir,
44    .readdir = ext4_readdir, /* we take BKL. needed?*/
45    .unlocked_ioctl = ext4_ioctl,
46#ifdef CONFIG_COMPAT
47    .compat_ioctl = ext4_compat_ioctl,
48#endif
49    .fsync = ext4_sync_file,
50    .release = ext4_release_dir,
51};
52
53
54static unsigned char get_dtype(struct super_block *sb, int filetype)
55{
56    if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FILETYPE) ||
57        (filetype >= EXT4_FT_MAX))
58        return DT_UNKNOWN;
59
60    return (ext4_filetype_table[filetype]);
61}
62
63
64int ext4_check_dir_entry(const char *function, struct inode *dir,
65             struct ext4_dir_entry_2 *de,
66             struct buffer_head *bh,
67             unsigned int offset)
68{
69    const char *error_msg = NULL;
70    const int rlen = ext4_rec_len_from_disk(de->rec_len,
71                        dir->i_sb->s_blocksize);
72
73    if (rlen < EXT4_DIR_REC_LEN(1))
74        error_msg = "rec_len is smaller than minimal";
75    else if (rlen % 4 != 0)
76        error_msg = "rec_len % 4 != 0";
77    else if (rlen < EXT4_DIR_REC_LEN(de->name_len))
78        error_msg = "rec_len is too small for name_len";
79    else if (((char *) de - bh->b_data) + rlen > dir->i_sb->s_blocksize)
80        error_msg = "directory entry across blocks";
81    else if (le32_to_cpu(de->inode) >
82            le32_to_cpu(EXT4_SB(dir->i_sb)->s_es->s_inodes_count))
83        error_msg = "inode out of bounds";
84
85    if (error_msg != NULL)
86        ext4_error(dir->i_sb, function,
87            "bad entry in directory #%lu: %s - "
88            "offset=%u, inode=%u, rec_len=%d, name_len=%d",
89            dir->i_ino, error_msg, offset,
90            le32_to_cpu(de->inode),
91            rlen, de->name_len);
92    return error_msg == NULL ? 1 : 0;
93}
94
95static int ext4_readdir(struct file *filp,
96             void *dirent, filldir_t filldir)
97{
98    int error = 0;
99    unsigned int offset;
100    int i, stored;
101    struct ext4_dir_entry_2 *de;
102    struct super_block *sb;
103    int err;
104    struct inode *inode = filp->f_path.dentry->d_inode;
105    int ret = 0;
106    int dir_has_error = 0;
107
108    sb = inode->i_sb;
109
110    if (EXT4_HAS_COMPAT_FEATURE(inode->i_sb,
111                    EXT4_FEATURE_COMPAT_DIR_INDEX) &&
112        ((EXT4_I(inode)->i_flags & EXT4_INDEX_FL) ||
113         ((inode->i_size >> sb->s_blocksize_bits) == 1))) {
114        err = ext4_dx_readdir(filp, dirent, filldir);
115        if (err != ERR_BAD_DX_DIR) {
116            ret = err;
117            goto out;
118        }
119        /*
120         * We don't set the inode dirty flag since it's not
121         * critical that it get flushed back to the disk.
122         */
123        EXT4_I(filp->f_path.dentry->d_inode)->i_flags &= ~EXT4_INDEX_FL;
124    }
125    stored = 0;
126    offset = filp->f_pos & (sb->s_blocksize - 1);
127
128    while (!error && !stored && filp->f_pos < inode->i_size) {
129        ext4_lblk_t blk = filp->f_pos >> EXT4_BLOCK_SIZE_BITS(sb);
130        struct buffer_head map_bh;
131        struct buffer_head *bh = NULL;
132
133        map_bh.b_state = 0;
134        err = ext4_get_blocks(NULL, inode, blk, 1, &map_bh, 0);
135        if (err > 0) {
136            pgoff_t index = map_bh.b_blocknr >>
137                    (PAGE_CACHE_SHIFT - inode->i_blkbits);
138            if (!ra_has_index(&filp->f_ra, index))
139                page_cache_sync_readahead(
140                    sb->s_bdev->bd_inode->i_mapping,
141                    &filp->f_ra, filp,
142                    index, 1);
143            filp->f_ra.prev_pos = (loff_t)index << PAGE_CACHE_SHIFT;
144            bh = ext4_bread(NULL, inode, blk, 0, &err);
145        }
146
147        /*
148         * We ignore I/O errors on directories so users have a chance
149         * of recovering data when there's a bad sector
150         */
151        if (!bh) {
152            if (!dir_has_error) {
153                ext4_error(sb, __func__, "directory #%lu "
154                       "contains a hole at offset %Lu",
155                       inode->i_ino,
156                       (unsigned long long) filp->f_pos);
157                dir_has_error = 1;
158            }
159            /* corrupt size? Maybe no more blocks to read */
160            if (filp->f_pos > inode->i_blocks << 9)
161                break;
162            filp->f_pos += sb->s_blocksize - offset;
163            continue;
164        }
165
166revalidate:
167        /* If the dir block has changed since the last call to
168         * readdir(2), then we might be pointing to an invalid
169         * dirent right now. Scan from the start of the block
170         * to make sure. */
171        if (filp->f_version != inode->i_version) {
172            for (i = 0; i < sb->s_blocksize && i < offset; ) {
173                de = (struct ext4_dir_entry_2 *)
174                    (bh->b_data + i);
175                /* It's too expensive to do a full
176                 * dirent test each time round this
177                 * loop, but we do have to test at
178                 * least that it is non-zero. A
179                 * failure will be detected in the
180                 * dirent test below. */
181                if (ext4_rec_len_from_disk(de->rec_len,
182                    sb->s_blocksize) < EXT4_DIR_REC_LEN(1))
183                    break;
184                i += ext4_rec_len_from_disk(de->rec_len,
185                                sb->s_blocksize);
186            }
187            offset = i;
188            filp->f_pos = (filp->f_pos & ~(sb->s_blocksize - 1))
189                | offset;
190            filp->f_version = inode->i_version;
191        }
192
193        while (!error && filp->f_pos < inode->i_size
194               && offset < sb->s_blocksize) {
195            de = (struct ext4_dir_entry_2 *) (bh->b_data + offset);
196            if (!ext4_check_dir_entry("ext4_readdir", inode, de,
197                          bh, offset)) {
198                /*
199                 * On error, skip the f_pos to the next block
200                 */
201                filp->f_pos = (filp->f_pos |
202                        (sb->s_blocksize - 1)) + 1;
203                brelse(bh);
204                ret = stored;
205                goto out;
206            }
207            offset += ext4_rec_len_from_disk(de->rec_len,
208                    sb->s_blocksize);
209            if (le32_to_cpu(de->inode)) {
210                /* We might block in the next section
211                 * if the data destination is
212                 * currently swapped out. So, use a
213                 * version stamp to detect whether or
214                 * not the directory has been modified
215                 * during the copy operation.
216                 */
217                u64 version = filp->f_version;
218
219                error = filldir(dirent, de->name,
220                        de->name_len,
221                        filp->f_pos,
222                        le32_to_cpu(de->inode),
223                        get_dtype(sb, de->file_type));
224                if (error)
225                    break;
226                if (version != filp->f_version)
227                    goto revalidate;
228                stored++;
229            }
230            filp->f_pos += ext4_rec_len_from_disk(de->rec_len,
231                        sb->s_blocksize);
232        }
233        offset = 0;
234        brelse(bh);
235    }
236out:
237    return ret;
238}
239
240/*
241 * These functions convert from the major/minor hash to an f_pos
242 * value.
243 *
244 * Currently we only use major hash numer. This is unfortunate, but
245 * on 32-bit machines, the same VFS interface is used for lseek and
246 * llseek, so if we use the 64 bit offset, then the 32-bit versions of
247 * lseek/telldir/seekdir will blow out spectacularly, and from within
248 * the ext2 low-level routine, we don't know if we're being called by
249 * a 64-bit version of the system call or the 32-bit version of the
250 * system call. Worse yet, NFSv2 only allows for a 32-bit readdir
251 * cookie. Sigh.
252 */
253#define hash2pos(major, minor) (major >> 1)
254#define pos2maj_hash(pos) ((pos << 1) & 0xffffffff)
255#define pos2min_hash(pos) (0)
256
257/*
258 * This structure holds the nodes of the red-black tree used to store
259 * the directory entry in hash order.
260 */
261struct fname {
262    __u32 hash;
263    __u32 minor_hash;
264    struct rb_node rb_hash;
265    struct fname *next;
266    __u32 inode;
267    __u8 name_len;
268    __u8 file_type;
269    char name[0];
270};
271
272/*
273 * This functoin implements a non-recursive way of freeing all of the
274 * nodes in the red-black tree.
275 */
276static void free_rb_tree_fname(struct rb_root *root)
277{
278    struct rb_node *n = root->rb_node;
279    struct rb_node *parent;
280    struct fname *fname;
281
282    while (n) {
283        /* Do the node's children first */
284        if (n->rb_left) {
285            n = n->rb_left;
286            continue;
287        }
288        if (n->rb_right) {
289            n = n->rb_right;
290            continue;
291        }
292        /*
293         * The node has no children; free it, and then zero
294         * out parent's link to it. Finally go to the
295         * beginning of the loop and try to free the parent
296         * node.
297         */
298        parent = rb_parent(n);
299        fname = rb_entry(n, struct fname, rb_hash);
300        while (fname) {
301            struct fname *old = fname;
302            fname = fname->next;
303            kfree(old);
304        }
305        if (!parent)
306            root->rb_node = NULL;
307        else if (parent->rb_left == n)
308            parent->rb_left = NULL;
309        else if (parent->rb_right == n)
310            parent->rb_right = NULL;
311        n = parent;
312    }
313}
314
315
316static struct dir_private_info *ext4_htree_create_dir_info(loff_t pos)
317{
318    struct dir_private_info *p;
319
320    p = kzalloc(sizeof(struct dir_private_info), GFP_KERNEL);
321    if (!p)
322        return NULL;
323    p->curr_hash = pos2maj_hash(pos);
324    p->curr_minor_hash = pos2min_hash(pos);
325    return p;
326}
327
328void ext4_htree_free_dir_info(struct dir_private_info *p)
329{
330    free_rb_tree_fname(&p->root);
331    kfree(p);
332}
333
334/*
335 * Given a directory entry, enter it into the fname rb tree.
336 */
337int ext4_htree_store_dirent(struct file *dir_file, __u32 hash,
338                 __u32 minor_hash,
339                 struct ext4_dir_entry_2 *dirent)
340{
341    struct rb_node **p, *parent = NULL;
342    struct fname *fname, *new_fn;
343    struct dir_private_info *info;
344    int len;
345
346    info = (struct dir_private_info *) dir_file->private_data;
347    p = &info->root.rb_node;
348
349    /* Create and allocate the fname structure */
350    len = sizeof(struct fname) + dirent->name_len + 1;
351    new_fn = kzalloc(len, GFP_KERNEL);
352    if (!new_fn)
353        return -ENOMEM;
354    new_fn->hash = hash;
355    new_fn->minor_hash = minor_hash;
356    new_fn->inode = le32_to_cpu(dirent->inode);
357    new_fn->name_len = dirent->name_len;
358    new_fn->file_type = dirent->file_type;
359    memcpy(new_fn->name, dirent->name, dirent->name_len);
360    new_fn->name[dirent->name_len] = 0;
361
362    while (*p) {
363        parent = *p;
364        fname = rb_entry(parent, struct fname, rb_hash);
365
366        /*
367         * If the hash and minor hash match up, then we put
368         * them on a linked list. This rarely happens...
369         */
370        if ((new_fn->hash == fname->hash) &&
371            (new_fn->minor_hash == fname->minor_hash)) {
372            new_fn->next = fname->next;
373            fname->next = new_fn;
374            return 0;
375        }
376
377        if (new_fn->hash < fname->hash)
378            p = &(*p)->rb_left;
379        else if (new_fn->hash > fname->hash)
380            p = &(*p)->rb_right;
381        else if (new_fn->minor_hash < fname->minor_hash)
382            p = &(*p)->rb_left;
383        else /* if (new_fn->minor_hash > fname->minor_hash) */
384            p = &(*p)->rb_right;
385    }
386
387    rb_link_node(&new_fn->rb_hash, parent, p);
388    rb_insert_color(&new_fn->rb_hash, &info->root);
389    return 0;
390}
391
392
393
394/*
395 * This is a helper function for ext4_dx_readdir. It calls filldir
396 * for all entres on the fname linked list. (Normally there is only
397 * one entry on the linked list, unless there are 62 bit hash collisions.)
398 */
399static int call_filldir(struct file *filp, void *dirent,
400            filldir_t filldir, struct fname *fname)
401{
402    struct dir_private_info *info = filp->private_data;
403    loff_t curr_pos;
404    struct inode *inode = filp->f_path.dentry->d_inode;
405    struct super_block *sb;
406    int error;
407
408    sb = inode->i_sb;
409
410    if (!fname) {
411        printk(KERN_ERR "EXT4-fs: call_filldir: called with "
412               "null fname?!?\n");
413        return 0;
414    }
415    curr_pos = hash2pos(fname->hash, fname->minor_hash);
416    while (fname) {
417        error = filldir(dirent, fname->name,
418                fname->name_len, curr_pos,
419                fname->inode,
420                get_dtype(sb, fname->file_type));
421        if (error) {
422            filp->f_pos = curr_pos;
423            info->extra_fname = fname;
424            return error;
425        }
426        fname = fname->next;
427    }
428    return 0;
429}
430
431static int ext4_dx_readdir(struct file *filp,
432             void *dirent, filldir_t filldir)
433{
434    struct dir_private_info *info = filp->private_data;
435    struct inode *inode = filp->f_path.dentry->d_inode;
436    struct fname *fname;
437    int ret;
438
439    if (!info) {
440        info = ext4_htree_create_dir_info(filp->f_pos);
441        if (!info)
442            return -ENOMEM;
443        filp->private_data = info;
444    }
445
446    if (filp->f_pos == EXT4_HTREE_EOF)
447        return 0; /* EOF */
448
449    /* Some one has messed with f_pos; reset the world */
450    if (info->last_pos != filp->f_pos) {
451        free_rb_tree_fname(&info->root);
452        info->curr_node = NULL;
453        info->extra_fname = NULL;
454        info->curr_hash = pos2maj_hash(filp->f_pos);
455        info->curr_minor_hash = pos2min_hash(filp->f_pos);
456    }
457
458    /*
459     * If there are any leftover names on the hash collision
460     * chain, return them first.
461     */
462    if (info->extra_fname) {
463        if (call_filldir(filp, dirent, filldir, info->extra_fname))
464            goto finished;
465        info->extra_fname = NULL;
466        goto next_node;
467    } else if (!info->curr_node)
468        info->curr_node = rb_first(&info->root);
469
470    while (1) {
471        /*
472         * Fill the rbtree if we have no more entries,
473         * or the inode has changed since we last read in the
474         * cached entries.
475         */
476        if ((!info->curr_node) ||
477            (filp->f_version != inode->i_version)) {
478            info->curr_node = NULL;
479            free_rb_tree_fname(&info->root);
480            filp->f_version = inode->i_version;
481            ret = ext4_htree_fill_tree(filp, info->curr_hash,
482                           info->curr_minor_hash,
483                           &info->next_hash);
484            if (ret < 0)
485                return ret;
486            if (ret == 0) {
487                filp->f_pos = EXT4_HTREE_EOF;
488                break;
489            }
490            info->curr_node = rb_first(&info->root);
491        }
492
493        fname = rb_entry(info->curr_node, struct fname, rb_hash);
494        info->curr_hash = fname->hash;
495        info->curr_minor_hash = fname->minor_hash;
496        if (call_filldir(filp, dirent, filldir, fname))
497            break;
498    next_node:
499        info->curr_node = rb_next(info->curr_node);
500        if (info->curr_node) {
501            fname = rb_entry(info->curr_node, struct fname,
502                     rb_hash);
503            info->curr_hash = fname->hash;
504            info->curr_minor_hash = fname->minor_hash;
505        } else {
506            if (info->next_hash == ~0) {
507                filp->f_pos = EXT4_HTREE_EOF;
508                break;
509            }
510            info->curr_hash = info->next_hash;
511            info->curr_minor_hash = 0;
512        }
513    }
514finished:
515    info->last_pos = filp->f_pos;
516    return 0;
517}
518
519static int ext4_release_dir(struct inode *inode, struct file *filp)
520{
521    if (filp->private_data)
522        ext4_htree_free_dir_info(filp->private_data);
523
524    return 0;
525}
526

Archive Download this file



interactive