Root/fs/ext4/super.c

1/*
2 * linux/fs/ext4/super.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/inode.c
12 *
13 * Copyright (C) 1991, 1992 Linus Torvalds
14 *
15 * Big-endian to little-endian byte-swapping/bitmaps by
16 * David S. Miller (davem@caip.rutgers.edu), 1995
17 */
18
19#include <linux/module.h>
20#include <linux/string.h>
21#include <linux/fs.h>
22#include <linux/time.h>
23#include <linux/vmalloc.h>
24#include <linux/jbd2.h>
25#include <linux/slab.h>
26#include <linux/init.h>
27#include <linux/blkdev.h>
28#include <linux/parser.h>
29#include <linux/smp_lock.h>
30#include <linux/buffer_head.h>
31#include <linux/exportfs.h>
32#include <linux/vfs.h>
33#include <linux/random.h>
34#include <linux/mount.h>
35#include <linux/namei.h>
36#include <linux/quotaops.h>
37#include <linux/seq_file.h>
38#include <linux/proc_fs.h>
39#include <linux/ctype.h>
40#include <linux/log2.h>
41#include <linux/crc16.h>
42#include <asm/uaccess.h>
43
44#include "ext4.h"
45#include "ext4_jbd2.h"
46#include "xattr.h"
47#include "acl.h"
48
49#define CREATE_TRACE_POINTS
50#include <trace/events/ext4.h>
51
52static int default_mb_history_length = 1000;
53
54module_param_named(default_mb_history_length, default_mb_history_length,
55           int, 0644);
56MODULE_PARM_DESC(default_mb_history_length,
57         "Default number of entries saved for mb_history");
58
59struct proc_dir_entry *ext4_proc_root;
60static struct kset *ext4_kset;
61
62static int ext4_load_journal(struct super_block *, struct ext4_super_block *,
63                 unsigned long journal_devnum);
64static int ext4_commit_super(struct super_block *sb, int sync);
65static void ext4_mark_recovery_complete(struct super_block *sb,
66                    struct ext4_super_block *es);
67static void ext4_clear_journal_err(struct super_block *sb,
68                   struct ext4_super_block *es);
69static int ext4_sync_fs(struct super_block *sb, int wait);
70static const char *ext4_decode_error(struct super_block *sb, int errno,
71                     char nbuf[16]);
72static int ext4_remount(struct super_block *sb, int *flags, char *data);
73static int ext4_statfs(struct dentry *dentry, struct kstatfs *buf);
74static int ext4_unfreeze(struct super_block *sb);
75static void ext4_write_super(struct super_block *sb);
76static int ext4_freeze(struct super_block *sb);
77
78
79ext4_fsblk_t ext4_block_bitmap(struct super_block *sb,
80                   struct ext4_group_desc *bg)
81{
82    return le32_to_cpu(bg->bg_block_bitmap_lo) |
83        (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ?
84         (ext4_fsblk_t)le32_to_cpu(bg->bg_block_bitmap_hi) << 32 : 0);
85}
86
87ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb,
88                   struct ext4_group_desc *bg)
89{
90    return le32_to_cpu(bg->bg_inode_bitmap_lo) |
91        (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ?
92         (ext4_fsblk_t)le32_to_cpu(bg->bg_inode_bitmap_hi) << 32 : 0);
93}
94
95ext4_fsblk_t ext4_inode_table(struct super_block *sb,
96                  struct ext4_group_desc *bg)
97{
98    return le32_to_cpu(bg->bg_inode_table_lo) |
99        (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ?
100         (ext4_fsblk_t)le32_to_cpu(bg->bg_inode_table_hi) << 32 : 0);
101}
102
103__u32 ext4_free_blks_count(struct super_block *sb,
104                  struct ext4_group_desc *bg)
105{
106    return le16_to_cpu(bg->bg_free_blocks_count_lo) |
107        (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ?
108         (__u32)le16_to_cpu(bg->bg_free_blocks_count_hi) << 16 : 0);
109}
110
111__u32 ext4_free_inodes_count(struct super_block *sb,
112                  struct ext4_group_desc *bg)
113{
114    return le16_to_cpu(bg->bg_free_inodes_count_lo) |
115        (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ?
116         (__u32)le16_to_cpu(bg->bg_free_inodes_count_hi) << 16 : 0);
117}
118
119__u32 ext4_used_dirs_count(struct super_block *sb,
120                  struct ext4_group_desc *bg)
121{
122    return le16_to_cpu(bg->bg_used_dirs_count_lo) |
123        (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ?
124         (__u32)le16_to_cpu(bg->bg_used_dirs_count_hi) << 16 : 0);
125}
126
127__u32 ext4_itable_unused_count(struct super_block *sb,
128                  struct ext4_group_desc *bg)
129{
130    return le16_to_cpu(bg->bg_itable_unused_lo) |
131        (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ?
132         (__u32)le16_to_cpu(bg->bg_itable_unused_hi) << 16 : 0);
133}
134
135void ext4_block_bitmap_set(struct super_block *sb,
136               struct ext4_group_desc *bg, ext4_fsblk_t blk)
137{
138    bg->bg_block_bitmap_lo = cpu_to_le32((u32)blk);
139    if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT)
140        bg->bg_block_bitmap_hi = cpu_to_le32(blk >> 32);
141}
142
143void ext4_inode_bitmap_set(struct super_block *sb,
144               struct ext4_group_desc *bg, ext4_fsblk_t blk)
145{
146    bg->bg_inode_bitmap_lo = cpu_to_le32((u32)blk);
147    if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT)
148        bg->bg_inode_bitmap_hi = cpu_to_le32(blk >> 32);
149}
150
151void ext4_inode_table_set(struct super_block *sb,
152              struct ext4_group_desc *bg, ext4_fsblk_t blk)
153{
154    bg->bg_inode_table_lo = cpu_to_le32((u32)blk);
155    if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT)
156        bg->bg_inode_table_hi = cpu_to_le32(blk >> 32);
157}
158
159void ext4_free_blks_set(struct super_block *sb,
160              struct ext4_group_desc *bg, __u32 count)
161{
162    bg->bg_free_blocks_count_lo = cpu_to_le16((__u16)count);
163    if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT)
164        bg->bg_free_blocks_count_hi = cpu_to_le16(count >> 16);
165}
166
167void ext4_free_inodes_set(struct super_block *sb,
168              struct ext4_group_desc *bg, __u32 count)
169{
170    bg->bg_free_inodes_count_lo = cpu_to_le16((__u16)count);
171    if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT)
172        bg->bg_free_inodes_count_hi = cpu_to_le16(count >> 16);
173}
174
175void ext4_used_dirs_set(struct super_block *sb,
176              struct ext4_group_desc *bg, __u32 count)
177{
178    bg->bg_used_dirs_count_lo = cpu_to_le16((__u16)count);
179    if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT)
180        bg->bg_used_dirs_count_hi = cpu_to_le16(count >> 16);
181}
182
183void ext4_itable_unused_set(struct super_block *sb,
184              struct ext4_group_desc *bg, __u32 count)
185{
186    bg->bg_itable_unused_lo = cpu_to_le16((__u16)count);
187    if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT)
188        bg->bg_itable_unused_hi = cpu_to_le16(count >> 16);
189}
190
191/*
192 * Wrappers for jbd2_journal_start/end.
193 *
194 * The only special thing we need to do here is to make sure that all
195 * journal_end calls result in the superblock being marked dirty, so
196 * that sync() will call the filesystem's write_super callback if
197 * appropriate.
198 */
199handle_t *ext4_journal_start_sb(struct super_block *sb, int nblocks)
200{
201    journal_t *journal;
202
203    if (sb->s_flags & MS_RDONLY)
204        return ERR_PTR(-EROFS);
205
206    /* Special case here: if the journal has aborted behind our
207     * backs (eg. EIO in the commit thread), then we still need to
208     * take the FS itself readonly cleanly. */
209    journal = EXT4_SB(sb)->s_journal;
210    if (journal) {
211        if (is_journal_aborted(journal)) {
212            ext4_abort(sb, __func__, "Detected aborted journal");
213            return ERR_PTR(-EROFS);
214        }
215        return jbd2_journal_start(journal, nblocks);
216    }
217    /*
218     * We're not journaling, return the appropriate indication.
219     */
220    current->journal_info = EXT4_NOJOURNAL_HANDLE;
221    return current->journal_info;
222}
223
224/*
225 * The only special thing we need to do here is to make sure that all
226 * jbd2_journal_stop calls result in the superblock being marked dirty, so
227 * that sync() will call the filesystem's write_super callback if
228 * appropriate.
229 */
230int __ext4_journal_stop(const char *where, handle_t *handle)
231{
232    struct super_block *sb;
233    int err;
234    int rc;
235
236    if (!ext4_handle_valid(handle)) {
237        /*
238         * Do this here since we don't call jbd2_journal_stop() in
239         * no-journal mode.
240         */
241        current->journal_info = NULL;
242        return 0;
243    }
244    sb = handle->h_transaction->t_journal->j_private;
245    err = handle->h_err;
246    rc = jbd2_journal_stop(handle);
247
248    if (!err)
249        err = rc;
250    if (err)
251        __ext4_std_error(sb, where, err);
252    return err;
253}
254
255void ext4_journal_abort_handle(const char *caller, const char *err_fn,
256        struct buffer_head *bh, handle_t *handle, int err)
257{
258    char nbuf[16];
259    const char *errstr = ext4_decode_error(NULL, err, nbuf);
260
261    BUG_ON(!ext4_handle_valid(handle));
262
263    if (bh)
264        BUFFER_TRACE(bh, "abort");
265
266    if (!handle->h_err)
267        handle->h_err = err;
268
269    if (is_handle_aborted(handle))
270        return;
271
272    printk(KERN_ERR "%s: aborting transaction: %s in %s\n",
273           caller, errstr, err_fn);
274
275    jbd2_journal_abort_handle(handle);
276}
277
278/* Deal with the reporting of failure conditions on a filesystem such as
279 * inconsistencies detected or read IO failures.
280 *
281 * On ext2, we can store the error state of the filesystem in the
282 * superblock. That is not possible on ext4, because we may have other
283 * write ordering constraints on the superblock which prevent us from
284 * writing it out straight away; and given that the journal is about to
285 * be aborted, we can't rely on the current, or future, transactions to
286 * write out the superblock safely.
287 *
288 * We'll just use the jbd2_journal_abort() error code to record an error in
289 * the journal instead. On recovery, the journal will compain about
290 * that error until we've noted it down and cleared it.
291 */
292
293static void ext4_handle_error(struct super_block *sb)
294{
295    struct ext4_super_block *es = EXT4_SB(sb)->s_es;
296
297    EXT4_SB(sb)->s_mount_state |= EXT4_ERROR_FS;
298    es->s_state |= cpu_to_le16(EXT4_ERROR_FS);
299
300    if (sb->s_flags & MS_RDONLY)
301        return;
302
303    if (!test_opt(sb, ERRORS_CONT)) {
304        journal_t *journal = EXT4_SB(sb)->s_journal;
305
306        EXT4_SB(sb)->s_mount_flags |= EXT4_MF_FS_ABORTED;
307        if (journal)
308            jbd2_journal_abort(journal, -EIO);
309    }
310    if (test_opt(sb, ERRORS_RO)) {
311        ext4_msg(sb, KERN_CRIT, "Remounting filesystem read-only");
312        sb->s_flags |= MS_RDONLY;
313    }
314    ext4_commit_super(sb, 1);
315    if (test_opt(sb, ERRORS_PANIC))
316        panic("EXT4-fs (device %s): panic forced after error\n",
317            sb->s_id);
318}
319
320void ext4_error(struct super_block *sb, const char *function,
321        const char *fmt, ...)
322{
323    va_list args;
324
325    va_start(args, fmt);
326    printk(KERN_CRIT "EXT4-fs error (device %s): %s: ", sb->s_id, function);
327    vprintk(fmt, args);
328    printk("\n");
329    va_end(args);
330
331    ext4_handle_error(sb);
332}
333
334static const char *ext4_decode_error(struct super_block *sb, int errno,
335                     char nbuf[16])
336{
337    char *errstr = NULL;
338
339    switch (errno) {
340    case -EIO:
341        errstr = "IO failure";
342        break;
343    case -ENOMEM:
344        errstr = "Out of memory";
345        break;
346    case -EROFS:
347        if (!sb || EXT4_SB(sb)->s_journal->j_flags & JBD2_ABORT)
348            errstr = "Journal has aborted";
349        else
350            errstr = "Readonly filesystem";
351        break;
352    default:
353        /* If the caller passed in an extra buffer for unknown
354         * errors, textualise them now. Else we just return
355         * NULL. */
356        if (nbuf) {
357            /* Check for truncated error codes... */
358            if (snprintf(nbuf, 16, "error %d", -errno) >= 0)
359                errstr = nbuf;
360        }
361        break;
362    }
363
364    return errstr;
365}
366
367/* __ext4_std_error decodes expected errors from journaling functions
368 * automatically and invokes the appropriate error response. */
369
370void __ext4_std_error(struct super_block *sb, const char *function, int errno)
371{
372    char nbuf[16];
373    const char *errstr;
374
375    /* Special case: if the error is EROFS, and we're not already
376     * inside a transaction, then there's really no point in logging
377     * an error. */
378    if (errno == -EROFS && journal_current_handle() == NULL &&
379        (sb->s_flags & MS_RDONLY))
380        return;
381
382    errstr = ext4_decode_error(sb, errno, nbuf);
383    printk(KERN_CRIT "EXT4-fs error (device %s) in %s: %s\n",
384           sb->s_id, function, errstr);
385
386    ext4_handle_error(sb);
387}
388
389/*
390 * ext4_abort is a much stronger failure handler than ext4_error. The
391 * abort function may be used to deal with unrecoverable failures such
392 * as journal IO errors or ENOMEM at a critical moment in log management.
393 *
394 * We unconditionally force the filesystem into an ABORT|READONLY state,
395 * unless the error response on the fs has been set to panic in which
396 * case we take the easy way out and panic immediately.
397 */
398
399void ext4_abort(struct super_block *sb, const char *function,
400        const char *fmt, ...)
401{
402    va_list args;
403
404    va_start(args, fmt);
405    printk(KERN_CRIT "EXT4-fs error (device %s): %s: ", sb->s_id, function);
406    vprintk(fmt, args);
407    printk("\n");
408    va_end(args);
409
410    if (test_opt(sb, ERRORS_PANIC))
411        panic("EXT4-fs panic from previous error\n");
412
413    if (sb->s_flags & MS_RDONLY)
414        return;
415
416    ext4_msg(sb, KERN_CRIT, "Remounting filesystem read-only");
417    EXT4_SB(sb)->s_mount_state |= EXT4_ERROR_FS;
418    sb->s_flags |= MS_RDONLY;
419    EXT4_SB(sb)->s_mount_flags |= EXT4_MF_FS_ABORTED;
420    if (EXT4_SB(sb)->s_journal)
421        jbd2_journal_abort(EXT4_SB(sb)->s_journal, -EIO);
422}
423
424void ext4_msg (struct super_block * sb, const char *prefix,
425           const char *fmt, ...)
426{
427    va_list args;
428
429    va_start(args, fmt);
430    printk("%sEXT4-fs (%s): ", prefix, sb->s_id);
431    vprintk(fmt, args);
432    printk("\n");
433    va_end(args);
434}
435
436void ext4_warning(struct super_block *sb, const char *function,
437          const char *fmt, ...)
438{
439    va_list args;
440
441    va_start(args, fmt);
442    printk(KERN_WARNING "EXT4-fs warning (device %s): %s: ",
443           sb->s_id, function);
444    vprintk(fmt, args);
445    printk("\n");
446    va_end(args);
447}
448
449void ext4_grp_locked_error(struct super_block *sb, ext4_group_t grp,
450               const char *function, const char *fmt, ...)
451__releases(bitlock)
452__acquires(bitlock)
453{
454    va_list args;
455    struct ext4_super_block *es = EXT4_SB(sb)->s_es;
456
457    va_start(args, fmt);
458    printk(KERN_CRIT "EXT4-fs error (device %s): %s: ", sb->s_id, function);
459    vprintk(fmt, args);
460    printk("\n");
461    va_end(args);
462
463    if (test_opt(sb, ERRORS_CONT)) {
464        EXT4_SB(sb)->s_mount_state |= EXT4_ERROR_FS;
465        es->s_state |= cpu_to_le16(EXT4_ERROR_FS);
466        ext4_commit_super(sb, 0);
467        return;
468    }
469    ext4_unlock_group(sb, grp);
470    ext4_handle_error(sb);
471    /*
472     * We only get here in the ERRORS_RO case; relocking the group
473     * may be dangerous, but nothing bad will happen since the
474     * filesystem will have already been marked read/only and the
475     * journal has been aborted. We return 1 as a hint to callers
476     * who might what to use the return value from
477     * ext4_grp_locked_error() to distinguish beween the
478     * ERRORS_CONT and ERRORS_RO case, and perhaps return more
479     * aggressively from the ext4 function in question, with a
480     * more appropriate error code.
481     */
482    ext4_lock_group(sb, grp);
483    return;
484}
485
486void ext4_update_dynamic_rev(struct super_block *sb)
487{
488    struct ext4_super_block *es = EXT4_SB(sb)->s_es;
489
490    if (le32_to_cpu(es->s_rev_level) > EXT4_GOOD_OLD_REV)
491        return;
492
493    ext4_warning(sb, __func__,
494             "updating to rev %d because of new feature flag, "
495             "running e2fsck is recommended",
496             EXT4_DYNAMIC_REV);
497
498    es->s_first_ino = cpu_to_le32(EXT4_GOOD_OLD_FIRST_INO);
499    es->s_inode_size = cpu_to_le16(EXT4_GOOD_OLD_INODE_SIZE);
500    es->s_rev_level = cpu_to_le32(EXT4_DYNAMIC_REV);
501    /* leave es->s_feature_*compat flags alone */
502    /* es->s_uuid will be set by e2fsck if empty */
503
504    /*
505     * The rest of the superblock fields should be zero, and if not it
506     * means they are likely already in use, so leave them alone. We
507     * can leave it up to e2fsck to clean up any inconsistencies there.
508     */
509}
510
511/*
512 * Open the external journal device
513 */
514static struct block_device *ext4_blkdev_get(dev_t dev, struct super_block *sb)
515{
516    struct block_device *bdev;
517    char b[BDEVNAME_SIZE];
518
519    bdev = open_by_devnum(dev, FMODE_READ|FMODE_WRITE);
520    if (IS_ERR(bdev))
521        goto fail;
522    return bdev;
523
524fail:
525    ext4_msg(sb, KERN_ERR, "failed to open journal device %s: %ld",
526            __bdevname(dev, b), PTR_ERR(bdev));
527    return NULL;
528}
529
530/*
531 * Release the journal device
532 */
533static int ext4_blkdev_put(struct block_device *bdev)
534{
535    bd_release(bdev);
536    return blkdev_put(bdev, FMODE_READ|FMODE_WRITE);
537}
538
539static int ext4_blkdev_remove(struct ext4_sb_info *sbi)
540{
541    struct block_device *bdev;
542    int ret = -ENODEV;
543
544    bdev = sbi->journal_bdev;
545    if (bdev) {
546        ret = ext4_blkdev_put(bdev);
547        sbi->journal_bdev = NULL;
548    }
549    return ret;
550}
551
552static inline struct inode *orphan_list_entry(struct list_head *l)
553{
554    return &list_entry(l, struct ext4_inode_info, i_orphan)->vfs_inode;
555}
556
557static void dump_orphan_list(struct super_block *sb, struct ext4_sb_info *sbi)
558{
559    struct list_head *l;
560
561    ext4_msg(sb, KERN_ERR, "sb orphan head is %d",
562         le32_to_cpu(sbi->s_es->s_last_orphan));
563
564    printk(KERN_ERR "sb_info orphan list:\n");
565    list_for_each(l, &sbi->s_orphan) {
566        struct inode *inode = orphan_list_entry(l);
567        printk(KERN_ERR " "
568               "inode %s:%lu at %p: mode %o, nlink %d, next %d\n",
569               inode->i_sb->s_id, inode->i_ino, inode,
570               inode->i_mode, inode->i_nlink,
571               NEXT_ORPHAN(inode));
572    }
573}
574
575static void ext4_put_super(struct super_block *sb)
576{
577    struct ext4_sb_info *sbi = EXT4_SB(sb);
578    struct ext4_super_block *es = sbi->s_es;
579    int i, err;
580
581    lock_super(sb);
582    lock_kernel();
583    if (sb->s_dirt)
584        ext4_commit_super(sb, 1);
585
586    ext4_release_system_zone(sb);
587    ext4_mb_release(sb);
588    ext4_ext_release(sb);
589    ext4_xattr_put_super(sb);
590    if (sbi->s_journal) {
591        err = jbd2_journal_destroy(sbi->s_journal);
592        sbi->s_journal = NULL;
593        if (err < 0)
594            ext4_abort(sb, __func__,
595                   "Couldn't clean up the journal");
596    }
597    if (!(sb->s_flags & MS_RDONLY)) {
598        EXT4_CLEAR_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER);
599        es->s_state = cpu_to_le16(sbi->s_mount_state);
600        ext4_commit_super(sb, 1);
601    }
602    if (sbi->s_proc) {
603        remove_proc_entry(sb->s_id, ext4_proc_root);
604    }
605    kobject_del(&sbi->s_kobj);
606
607    for (i = 0; i < sbi->s_gdb_count; i++)
608        brelse(sbi->s_group_desc[i]);
609    kfree(sbi->s_group_desc);
610    if (is_vmalloc_addr(sbi->s_flex_groups))
611        vfree(sbi->s_flex_groups);
612    else
613        kfree(sbi->s_flex_groups);
614    percpu_counter_destroy(&sbi->s_freeblocks_counter);
615    percpu_counter_destroy(&sbi->s_freeinodes_counter);
616    percpu_counter_destroy(&sbi->s_dirs_counter);
617    percpu_counter_destroy(&sbi->s_dirtyblocks_counter);
618    brelse(sbi->s_sbh);
619#ifdef CONFIG_QUOTA
620    for (i = 0; i < MAXQUOTAS; i++)
621        kfree(sbi->s_qf_names[i]);
622#endif
623
624    /* Debugging code just in case the in-memory inode orphan list
625     * isn't empty. The on-disk one can be non-empty if we've
626     * detected an error and taken the fs readonly, but the
627     * in-memory list had better be clean by this point. */
628    if (!list_empty(&sbi->s_orphan))
629        dump_orphan_list(sb, sbi);
630    J_ASSERT(list_empty(&sbi->s_orphan));
631
632    invalidate_bdev(sb->s_bdev);
633    if (sbi->journal_bdev && sbi->journal_bdev != sb->s_bdev) {
634        /*
635         * Invalidate the journal device's buffers. We don't want them
636         * floating about in memory - the physical journal device may
637         * hotswapped, and it breaks the `ro-after' testing code.
638         */
639        sync_blockdev(sbi->journal_bdev);
640        invalidate_bdev(sbi->journal_bdev);
641        ext4_blkdev_remove(sbi);
642    }
643    sb->s_fs_info = NULL;
644    /*
645     * Now that we are completely done shutting down the
646     * superblock, we need to actually destroy the kobject.
647     */
648    unlock_kernel();
649    unlock_super(sb);
650    kobject_put(&sbi->s_kobj);
651    wait_for_completion(&sbi->s_kobj_unregister);
652    kfree(sbi->s_blockgroup_lock);
653    kfree(sbi);
654}
655
656static struct kmem_cache *ext4_inode_cachep;
657
658/*
659 * Called inside transaction, so use GFP_NOFS
660 */
661static struct inode *ext4_alloc_inode(struct super_block *sb)
662{
663    struct ext4_inode_info *ei;
664
665    ei = kmem_cache_alloc(ext4_inode_cachep, GFP_NOFS);
666    if (!ei)
667        return NULL;
668
669    ei->vfs_inode.i_version = 1;
670    ei->vfs_inode.i_data.writeback_index = 0;
671    memset(&ei->i_cached_extent, 0, sizeof(struct ext4_ext_cache));
672    INIT_LIST_HEAD(&ei->i_prealloc_list);
673    spin_lock_init(&ei->i_prealloc_lock);
674    /*
675     * Note: We can be called before EXT4_SB(sb)->s_journal is set,
676     * therefore it can be null here. Don't check it, just initialize
677     * jinode.
678     */
679    jbd2_journal_init_jbd_inode(&ei->jinode, &ei->vfs_inode);
680    ei->i_reserved_data_blocks = 0;
681    ei->i_reserved_meta_blocks = 0;
682    ei->i_allocated_meta_blocks = 0;
683    ei->i_delalloc_reserved_flag = 0;
684    spin_lock_init(&(ei->i_block_reservation_lock));
685
686    return &ei->vfs_inode;
687}
688
689static void ext4_destroy_inode(struct inode *inode)
690{
691    if (!list_empty(&(EXT4_I(inode)->i_orphan))) {
692        ext4_msg(inode->i_sb, KERN_ERR,
693             "Inode %lu (%p): orphan list check failed!",
694             inode->i_ino, EXT4_I(inode));
695        print_hex_dump(KERN_INFO, "", DUMP_PREFIX_ADDRESS, 16, 4,
696                EXT4_I(inode), sizeof(struct ext4_inode_info),
697                true);
698        dump_stack();
699    }
700    kmem_cache_free(ext4_inode_cachep, EXT4_I(inode));
701}
702
703static void init_once(void *foo)
704{
705    struct ext4_inode_info *ei = (struct ext4_inode_info *) foo;
706
707    INIT_LIST_HEAD(&ei->i_orphan);
708#ifdef CONFIG_EXT4_FS_XATTR
709    init_rwsem(&ei->xattr_sem);
710#endif
711    init_rwsem(&ei->i_data_sem);
712    inode_init_once(&ei->vfs_inode);
713}
714
715static int init_inodecache(void)
716{
717    ext4_inode_cachep = kmem_cache_create("ext4_inode_cache",
718                         sizeof(struct ext4_inode_info),
719                         0, (SLAB_RECLAIM_ACCOUNT|
720                        SLAB_MEM_SPREAD),
721                         init_once);
722    if (ext4_inode_cachep == NULL)
723        return -ENOMEM;
724    return 0;
725}
726
727static void destroy_inodecache(void)
728{
729    kmem_cache_destroy(ext4_inode_cachep);
730}
731
732static void ext4_clear_inode(struct inode *inode)
733{
734    ext4_discard_preallocations(inode);
735    if (EXT4_JOURNAL(inode))
736        jbd2_journal_release_jbd_inode(EXT4_SB(inode->i_sb)->s_journal,
737                       &EXT4_I(inode)->jinode);
738}
739
740static inline void ext4_show_quota_options(struct seq_file *seq,
741                       struct super_block *sb)
742{
743#if defined(CONFIG_QUOTA)
744    struct ext4_sb_info *sbi = EXT4_SB(sb);
745
746    if (sbi->s_jquota_fmt)
747        seq_printf(seq, ",jqfmt=%s",
748        (sbi->s_jquota_fmt == QFMT_VFS_OLD) ? "vfsold" : "vfsv0");
749
750    if (sbi->s_qf_names[USRQUOTA])
751        seq_printf(seq, ",usrjquota=%s", sbi->s_qf_names[USRQUOTA]);
752
753    if (sbi->s_qf_names[GRPQUOTA])
754        seq_printf(seq, ",grpjquota=%s", sbi->s_qf_names[GRPQUOTA]);
755
756    if (sbi->s_mount_opt & EXT4_MOUNT_USRQUOTA)
757        seq_puts(seq, ",usrquota");
758
759    if (sbi->s_mount_opt & EXT4_MOUNT_GRPQUOTA)
760        seq_puts(seq, ",grpquota");
761#endif
762}
763
764/*
765 * Show an option if
766 * - it's set to a non-default value OR
767 * - if the per-sb default is different from the global default
768 */
769static int ext4_show_options(struct seq_file *seq, struct vfsmount *vfs)
770{
771    int def_errors;
772    unsigned long def_mount_opts;
773    struct super_block *sb = vfs->mnt_sb;
774    struct ext4_sb_info *sbi = EXT4_SB(sb);
775    struct ext4_super_block *es = sbi->s_es;
776
777    def_mount_opts = le32_to_cpu(es->s_default_mount_opts);
778    def_errors = le16_to_cpu(es->s_errors);
779
780    if (sbi->s_sb_block != 1)
781        seq_printf(seq, ",sb=%llu", sbi->s_sb_block);
782    if (test_opt(sb, MINIX_DF))
783        seq_puts(seq, ",minixdf");
784    if (test_opt(sb, GRPID) && !(def_mount_opts & EXT4_DEFM_BSDGROUPS))
785        seq_puts(seq, ",grpid");
786    if (!test_opt(sb, GRPID) && (def_mount_opts & EXT4_DEFM_BSDGROUPS))
787        seq_puts(seq, ",nogrpid");
788    if (sbi->s_resuid != EXT4_DEF_RESUID ||
789        le16_to_cpu(es->s_def_resuid) != EXT4_DEF_RESUID) {
790        seq_printf(seq, ",resuid=%u", sbi->s_resuid);
791    }
792    if (sbi->s_resgid != EXT4_DEF_RESGID ||
793        le16_to_cpu(es->s_def_resgid) != EXT4_DEF_RESGID) {
794        seq_printf(seq, ",resgid=%u", sbi->s_resgid);
795    }
796    if (test_opt(sb, ERRORS_RO)) {
797        if (def_errors == EXT4_ERRORS_PANIC ||
798            def_errors == EXT4_ERRORS_CONTINUE) {
799            seq_puts(seq, ",errors=remount-ro");
800        }
801    }
802    if (test_opt(sb, ERRORS_CONT) && def_errors != EXT4_ERRORS_CONTINUE)
803        seq_puts(seq, ",errors=continue");
804    if (test_opt(sb, ERRORS_PANIC) && def_errors != EXT4_ERRORS_PANIC)
805        seq_puts(seq, ",errors=panic");
806    if (test_opt(sb, NO_UID32) && !(def_mount_opts & EXT4_DEFM_UID16))
807        seq_puts(seq, ",nouid32");
808    if (test_opt(sb, DEBUG) && !(def_mount_opts & EXT4_DEFM_DEBUG))
809        seq_puts(seq, ",debug");
810    if (test_opt(sb, OLDALLOC))
811        seq_puts(seq, ",oldalloc");
812#ifdef CONFIG_EXT4_FS_XATTR
813    if (test_opt(sb, XATTR_USER) &&
814        !(def_mount_opts & EXT4_DEFM_XATTR_USER))
815        seq_puts(seq, ",user_xattr");
816    if (!test_opt(sb, XATTR_USER) &&
817        (def_mount_opts & EXT4_DEFM_XATTR_USER)) {
818        seq_puts(seq, ",nouser_xattr");
819    }
820#endif
821#ifdef CONFIG_EXT4_FS_POSIX_ACL
822    if (test_opt(sb, POSIX_ACL) && !(def_mount_opts & EXT4_DEFM_ACL))
823        seq_puts(seq, ",acl");
824    if (!test_opt(sb, POSIX_ACL) && (def_mount_opts & EXT4_DEFM_ACL))
825        seq_puts(seq, ",noacl");
826#endif
827    if (sbi->s_commit_interval != JBD2_DEFAULT_MAX_COMMIT_AGE*HZ) {
828        seq_printf(seq, ",commit=%u",
829               (unsigned) (sbi->s_commit_interval / HZ));
830    }
831    if (sbi->s_min_batch_time != EXT4_DEF_MIN_BATCH_TIME) {
832        seq_printf(seq, ",min_batch_time=%u",
833               (unsigned) sbi->s_min_batch_time);
834    }
835    if (sbi->s_max_batch_time != EXT4_DEF_MAX_BATCH_TIME) {
836        seq_printf(seq, ",max_batch_time=%u",
837               (unsigned) sbi->s_min_batch_time);
838    }
839
840    /*
841     * We're changing the default of barrier mount option, so
842     * let's always display its mount state so it's clear what its
843     * status is.
844     */
845    seq_puts(seq, ",barrier=");
846    seq_puts(seq, test_opt(sb, BARRIER) ? "1" : "0");
847    if (test_opt(sb, JOURNAL_ASYNC_COMMIT))
848        seq_puts(seq, ",journal_async_commit");
849    if (test_opt(sb, NOBH))
850        seq_puts(seq, ",nobh");
851    if (test_opt(sb, I_VERSION))
852        seq_puts(seq, ",i_version");
853    if (!test_opt(sb, DELALLOC))
854        seq_puts(seq, ",nodelalloc");
855
856
857    if (sbi->s_stripe)
858        seq_printf(seq, ",stripe=%lu", sbi->s_stripe);
859    /*
860     * journal mode get enabled in different ways
861     * So just print the value even if we didn't specify it
862     */
863    if (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_JOURNAL_DATA)
864        seq_puts(seq, ",data=journal");
865    else if (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_ORDERED_DATA)
866        seq_puts(seq, ",data=ordered");
867    else if (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_WRITEBACK_DATA)
868        seq_puts(seq, ",data=writeback");
869
870    if (sbi->s_inode_readahead_blks != EXT4_DEF_INODE_READAHEAD_BLKS)
871        seq_printf(seq, ",inode_readahead_blks=%u",
872               sbi->s_inode_readahead_blks);
873
874    if (test_opt(sb, DATA_ERR_ABORT))
875        seq_puts(seq, ",data_err=abort");
876
877    if (test_opt(sb, NO_AUTO_DA_ALLOC))
878        seq_puts(seq, ",noauto_da_alloc");
879
880    ext4_show_quota_options(seq, sb);
881
882    return 0;
883}
884
885static struct inode *ext4_nfs_get_inode(struct super_block *sb,
886                    u64 ino, u32 generation)
887{
888    struct inode *inode;
889
890    if (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO)
891        return ERR_PTR(-ESTALE);
892    if (ino > le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count))
893        return ERR_PTR(-ESTALE);
894
895    /* iget isn't really right if the inode is currently unallocated!!
896     *
897     * ext4_read_inode will return a bad_inode if the inode had been
898     * deleted, so we should be safe.
899     *
900     * Currently we don't know the generation for parent directory, so
901     * a generation of 0 means "accept any"
902     */
903    inode = ext4_iget(sb, ino);
904    if (IS_ERR(inode))
905        return ERR_CAST(inode);
906    if (generation && inode->i_generation != generation) {
907        iput(inode);
908        return ERR_PTR(-ESTALE);
909    }
910
911    return inode;
912}
913
914static struct dentry *ext4_fh_to_dentry(struct super_block *sb, struct fid *fid,
915                    int fh_len, int fh_type)
916{
917    return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
918                    ext4_nfs_get_inode);
919}
920
921static struct dentry *ext4_fh_to_parent(struct super_block *sb, struct fid *fid,
922                    int fh_len, int fh_type)
923{
924    return generic_fh_to_parent(sb, fid, fh_len, fh_type,
925                    ext4_nfs_get_inode);
926}
927
928/*
929 * Try to release metadata pages (indirect blocks, directories) which are
930 * mapped via the block device. Since these pages could have journal heads
931 * which would prevent try_to_free_buffers() from freeing them, we must use
932 * jbd2 layer's try_to_free_buffers() function to release them.
933 */
934static int bdev_try_to_free_page(struct super_block *sb, struct page *page,
935                 gfp_t wait)
936{
937    journal_t *journal = EXT4_SB(sb)->s_journal;
938
939    WARN_ON(PageChecked(page));
940    if (!page_has_buffers(page))
941        return 0;
942    if (journal)
943        return jbd2_journal_try_to_free_buffers(journal, page,
944                            wait & ~__GFP_WAIT);
945    return try_to_free_buffers(page);
946}
947
948#ifdef CONFIG_QUOTA
949#define QTYPE2NAME(t) ((t) == USRQUOTA ? "user" : "group")
950#define QTYPE2MOPT(on, t) ((t) == USRQUOTA?((on)##USRJQUOTA):((on)##GRPJQUOTA))
951
952static int ext4_write_dquot(struct dquot *dquot);
953static int ext4_acquire_dquot(struct dquot *dquot);
954static int ext4_release_dquot(struct dquot *dquot);
955static int ext4_mark_dquot_dirty(struct dquot *dquot);
956static int ext4_write_info(struct super_block *sb, int type);
957static int ext4_quota_on(struct super_block *sb, int type, int format_id,
958                char *path, int remount);
959static int ext4_quota_on_mount(struct super_block *sb, int type);
960static ssize_t ext4_quota_read(struct super_block *sb, int type, char *data,
961                   size_t len, loff_t off);
962static ssize_t ext4_quota_write(struct super_block *sb, int type,
963                const char *data, size_t len, loff_t off);
964
965static struct dquot_operations ext4_quota_operations = {
966    .initialize = dquot_initialize,
967    .drop = dquot_drop,
968    .alloc_space = dquot_alloc_space,
969    .reserve_space = dquot_reserve_space,
970    .claim_space = dquot_claim_space,
971    .release_rsv = dquot_release_reserved_space,
972    .get_reserved_space = ext4_get_reserved_space,
973    .alloc_inode = dquot_alloc_inode,
974    .free_space = dquot_free_space,
975    .free_inode = dquot_free_inode,
976    .transfer = dquot_transfer,
977    .write_dquot = ext4_write_dquot,
978    .acquire_dquot = ext4_acquire_dquot,
979    .release_dquot = ext4_release_dquot,
980    .mark_dirty = ext4_mark_dquot_dirty,
981    .write_info = ext4_write_info,
982    .alloc_dquot = dquot_alloc,
983    .destroy_dquot = dquot_destroy,
984};
985
986static struct quotactl_ops ext4_qctl_operations = {
987    .quota_on = ext4_quota_on,
988    .quota_off = vfs_quota_off,
989    .quota_sync = vfs_quota_sync,
990    .get_info = vfs_get_dqinfo,
991    .set_info = vfs_set_dqinfo,
992    .get_dqblk = vfs_get_dqblk,
993    .set_dqblk = vfs_set_dqblk
994};
995#endif
996
997static const struct super_operations ext4_sops = {
998    .alloc_inode = ext4_alloc_inode,
999    .destroy_inode = ext4_destroy_inode,
1000    .write_inode = ext4_write_inode,
1001    .dirty_inode = ext4_dirty_inode,
1002    .delete_inode = ext4_delete_inode,
1003    .put_super = ext4_put_super,
1004    .sync_fs = ext4_sync_fs,
1005    .freeze_fs = ext4_freeze,
1006    .unfreeze_fs = ext4_unfreeze,
1007    .statfs = ext4_statfs,
1008    .remount_fs = ext4_remount,
1009    .clear_inode = ext4_clear_inode,
1010    .show_options = ext4_show_options,
1011#ifdef CONFIG_QUOTA
1012    .quota_read = ext4_quota_read,
1013    .quota_write = ext4_quota_write,
1014#endif
1015    .bdev_try_to_free_page = bdev_try_to_free_page,
1016};
1017
1018static const struct super_operations ext4_nojournal_sops = {
1019    .alloc_inode = ext4_alloc_inode,
1020    .destroy_inode = ext4_destroy_inode,
1021    .write_inode = ext4_write_inode,
1022    .dirty_inode = ext4_dirty_inode,
1023    .delete_inode = ext4_delete_inode,
1024    .write_super = ext4_write_super,
1025    .put_super = ext4_put_super,
1026    .statfs = ext4_statfs,
1027    .remount_fs = ext4_remount,
1028    .clear_inode = ext4_clear_inode,
1029    .show_options = ext4_show_options,
1030#ifdef CONFIG_QUOTA
1031    .quota_read = ext4_quota_read,
1032    .quota_write = ext4_quota_write,
1033#endif
1034    .bdev_try_to_free_page = bdev_try_to_free_page,
1035};
1036
1037static const struct export_operations ext4_export_ops = {
1038    .fh_to_dentry = ext4_fh_to_dentry,
1039    .fh_to_parent = ext4_fh_to_parent,
1040    .get_parent = ext4_get_parent,
1041};
1042
1043enum {
1044    Opt_bsd_df, Opt_minix_df, Opt_grpid, Opt_nogrpid,
1045    Opt_resgid, Opt_resuid, Opt_sb, Opt_err_cont, Opt_err_panic, Opt_err_ro,
1046    Opt_nouid32, Opt_debug, Opt_oldalloc, Opt_orlov,
1047    Opt_user_xattr, Opt_nouser_xattr, Opt_acl, Opt_noacl,
1048    Opt_auto_da_alloc, Opt_noauto_da_alloc, Opt_noload, Opt_nobh, Opt_bh,
1049    Opt_commit, Opt_min_batch_time, Opt_max_batch_time,
1050    Opt_journal_update, Opt_journal_dev,
1051    Opt_journal_checksum, Opt_journal_async_commit,
1052    Opt_abort, Opt_data_journal, Opt_data_ordered, Opt_data_writeback,
1053    Opt_data_err_abort, Opt_data_err_ignore, Opt_mb_history_length,
1054    Opt_usrjquota, Opt_grpjquota, Opt_offusrjquota, Opt_offgrpjquota,
1055    Opt_jqfmt_vfsold, Opt_jqfmt_vfsv0, Opt_quota, Opt_noquota,
1056    Opt_ignore, Opt_barrier, Opt_nobarrier, Opt_err, Opt_resize,
1057    Opt_usrquota, Opt_grpquota, Opt_i_version,
1058    Opt_stripe, Opt_delalloc, Opt_nodelalloc,
1059    Opt_block_validity, Opt_noblock_validity,
1060    Opt_inode_readahead_blks, Opt_journal_ioprio
1061};
1062
1063static const match_table_t tokens = {
1064    {Opt_bsd_df, "bsddf"},
1065    {Opt_minix_df, "minixdf"},
1066    {Opt_grpid, "grpid"},
1067    {Opt_grpid, "bsdgroups"},
1068    {Opt_nogrpid, "nogrpid"},
1069    {Opt_nogrpid, "sysvgroups"},
1070    {Opt_resgid, "resgid=%u"},
1071    {Opt_resuid, "resuid=%u"},
1072    {Opt_sb, "sb=%u"},
1073    {Opt_err_cont, "errors=continue"},
1074    {Opt_err_panic, "errors=panic"},
1075    {Opt_err_ro, "errors=remount-ro"},
1076    {Opt_nouid32, "nouid32"},
1077    {Opt_debug, "debug"},
1078    {Opt_oldalloc, "oldalloc"},
1079    {Opt_orlov, "orlov"},
1080    {Opt_user_xattr, "user_xattr"},
1081    {Opt_nouser_xattr, "nouser_xattr"},
1082    {Opt_acl, "acl"},
1083    {Opt_noacl, "noacl"},
1084    {Opt_noload, "noload"},
1085    {Opt_nobh, "nobh"},
1086    {Opt_bh, "bh"},
1087    {Opt_commit, "commit=%u"},
1088    {Opt_min_batch_time, "min_batch_time=%u"},
1089    {Opt_max_batch_time, "max_batch_time=%u"},
1090    {Opt_journal_update, "journal=update"},
1091    {Opt_journal_dev, "journal_dev=%u"},
1092    {Opt_journal_checksum, "journal_checksum"},
1093    {Opt_journal_async_commit, "journal_async_commit"},
1094    {Opt_abort, "abort"},
1095    {Opt_data_journal, "data=journal"},
1096    {Opt_data_ordered, "data=ordered"},
1097    {Opt_data_writeback, "data=writeback"},
1098    {Opt_data_err_abort, "data_err=abort"},
1099    {Opt_data_err_ignore, "data_err=ignore"},
1100    {Opt_mb_history_length, "mb_history_length=%u"},
1101    {Opt_offusrjquota, "usrjquota="},
1102    {Opt_usrjquota, "usrjquota=%s"},
1103    {Opt_offgrpjquota, "grpjquota="},
1104    {Opt_grpjquota, "grpjquota=%s"},
1105    {Opt_jqfmt_vfsold, "jqfmt=vfsold"},
1106    {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
1107    {Opt_grpquota, "grpquota"},
1108    {Opt_noquota, "noquota"},
1109    {Opt_quota, "quota"},
1110    {Opt_usrquota, "usrquota"},
1111    {Opt_barrier, "barrier=%u"},
1112    {Opt_barrier, "barrier"},
1113    {Opt_nobarrier, "nobarrier"},
1114    {Opt_i_version, "i_version"},
1115    {Opt_stripe, "stripe=%u"},
1116    {Opt_resize, "resize"},
1117    {Opt_delalloc, "delalloc"},
1118    {Opt_nodelalloc, "nodelalloc"},
1119    {Opt_block_validity, "block_validity"},
1120    {Opt_noblock_validity, "noblock_validity"},
1121    {Opt_inode_readahead_blks, "inode_readahead_blks=%u"},
1122    {Opt_journal_ioprio, "journal_ioprio=%u"},
1123    {Opt_auto_da_alloc, "auto_da_alloc=%u"},
1124    {Opt_auto_da_alloc, "auto_da_alloc"},
1125    {Opt_noauto_da_alloc, "noauto_da_alloc"},
1126    {Opt_err, NULL},
1127};
1128
1129static ext4_fsblk_t get_sb_block(void **data)
1130{
1131    ext4_fsblk_t sb_block;
1132    char *options = (char *) *data;
1133
1134    if (!options || strncmp(options, "sb=", 3) != 0)
1135        return 1; /* Default location */
1136
1137    options += 3;
1138    /* TODO: use simple_strtoll with >32bit ext4 */
1139    sb_block = simple_strtoul(options, &options, 0);
1140    if (*options && *options != ',') {
1141        printk(KERN_ERR "EXT4-fs: Invalid sb specification: %s\n",
1142               (char *) *data);
1143        return 1;
1144    }
1145    if (*options == ',')
1146        options++;
1147    *data = (void *) options;
1148
1149    return sb_block;
1150}
1151
1152#define DEFAULT_JOURNAL_IOPRIO (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 3))
1153
1154static int parse_options(char *options, struct super_block *sb,
1155             unsigned long *journal_devnum,
1156             unsigned int *journal_ioprio,
1157             ext4_fsblk_t *n_blocks_count, int is_remount)
1158{
1159    struct ext4_sb_info *sbi = EXT4_SB(sb);
1160    char *p;
1161    substring_t args[MAX_OPT_ARGS];
1162    int data_opt = 0;
1163    int option;
1164#ifdef CONFIG_QUOTA
1165    int qtype, qfmt;
1166    char *qname;
1167#endif
1168
1169    if (!options)
1170        return 1;
1171
1172    while ((p = strsep(&options, ",")) != NULL) {
1173        int token;
1174        if (!*p)
1175            continue;
1176
1177        token = match_token(p, tokens, args);
1178        switch (token) {
1179        case Opt_bsd_df:
1180            clear_opt(sbi->s_mount_opt, MINIX_DF);
1181            break;
1182        case Opt_minix_df:
1183            set_opt(sbi->s_mount_opt, MINIX_DF);
1184            break;
1185        case Opt_grpid:
1186            set_opt(sbi->s_mount_opt, GRPID);
1187            break;
1188        case Opt_nogrpid:
1189            clear_opt(sbi->s_mount_opt, GRPID);
1190            break;
1191        case Opt_resuid:
1192            if (match_int(&args[0], &option))
1193                return 0;
1194            sbi->s_resuid = option;
1195            break;
1196        case Opt_resgid:
1197            if (match_int(&args[0], &option))
1198                return 0;
1199            sbi->s_resgid = option;
1200            break;
1201        case Opt_sb:
1202            /* handled by get_sb_block() instead of here */
1203            /* *sb_block = match_int(&args[0]); */
1204            break;
1205        case Opt_err_panic:
1206            clear_opt(sbi->s_mount_opt, ERRORS_CONT);
1207            clear_opt(sbi->s_mount_opt, ERRORS_RO);
1208            set_opt(sbi->s_mount_opt, ERRORS_PANIC);
1209            break;
1210        case Opt_err_ro:
1211            clear_opt(sbi->s_mount_opt, ERRORS_CONT);
1212            clear_opt(sbi->s_mount_opt, ERRORS_PANIC);
1213            set_opt(sbi->s_mount_opt, ERRORS_RO);
1214            break;
1215        case Opt_err_cont:
1216            clear_opt(sbi->s_mount_opt, ERRORS_RO);
1217            clear_opt(sbi->s_mount_opt, ERRORS_PANIC);
1218            set_opt(sbi->s_mount_opt, ERRORS_CONT);
1219            break;
1220        case Opt_nouid32:
1221            set_opt(sbi->s_mount_opt, NO_UID32);
1222            break;
1223        case Opt_debug:
1224            set_opt(sbi->s_mount_opt, DEBUG);
1225            break;
1226        case Opt_oldalloc:
1227            set_opt(sbi->s_mount_opt, OLDALLOC);
1228            break;
1229        case Opt_orlov:
1230            clear_opt(sbi->s_mount_opt, OLDALLOC);
1231            break;
1232#ifdef CONFIG_EXT4_FS_XATTR
1233        case Opt_user_xattr:
1234            set_opt(sbi->s_mount_opt, XATTR_USER);
1235            break;
1236        case Opt_nouser_xattr:
1237            clear_opt(sbi->s_mount_opt, XATTR_USER);
1238            break;
1239#else
1240        case Opt_user_xattr:
1241        case Opt_nouser_xattr:
1242            ext4_msg(sb, KERN_ERR, "(no)user_xattr options not supported");
1243            break;
1244#endif
1245#ifdef CONFIG_EXT4_FS_POSIX_ACL
1246        case Opt_acl:
1247            set_opt(sbi->s_mount_opt, POSIX_ACL);
1248            break;
1249        case Opt_noacl:
1250            clear_opt(sbi->s_mount_opt, POSIX_ACL);
1251            break;
1252#else
1253        case Opt_acl:
1254        case Opt_noacl:
1255            ext4_msg(sb, KERN_ERR, "(no)acl options not supported");
1256            break;
1257#endif
1258        case Opt_journal_update:
1259            /* @@@ FIXME */
1260            /* Eventually we will want to be able to create
1261               a journal file here. For now, only allow the
1262               user to specify an existing inode to be the
1263               journal file. */
1264            if (is_remount) {
1265                ext4_msg(sb, KERN_ERR,
1266                     "Cannot specify journal on remount");
1267                return 0;
1268            }
1269            set_opt(sbi->s_mount_opt, UPDATE_JOURNAL);
1270            break;
1271        case Opt_journal_dev:
1272            if (is_remount) {
1273                ext4_msg(sb, KERN_ERR,
1274                    "Cannot specify journal on remount");
1275                return 0;
1276            }
1277            if (match_int(&args[0], &option))
1278                return 0;
1279            *journal_devnum = option;
1280            break;
1281        case Opt_journal_checksum:
1282            set_opt(sbi->s_mount_opt, JOURNAL_CHECKSUM);
1283            break;
1284        case Opt_journal_async_commit:
1285            set_opt(sbi->s_mount_opt, JOURNAL_ASYNC_COMMIT);
1286            set_opt(sbi->s_mount_opt, JOURNAL_CHECKSUM);
1287            break;
1288        case Opt_noload:
1289            set_opt(sbi->s_mount_opt, NOLOAD);
1290            break;
1291        case Opt_commit:
1292            if (match_int(&args[0], &option))
1293                return 0;
1294            if (option < 0)
1295                return 0;
1296            if (option == 0)
1297                option = JBD2_DEFAULT_MAX_COMMIT_AGE;
1298            sbi->s_commit_interval = HZ * option;
1299            break;
1300        case Opt_max_batch_time:
1301            if (match_int(&args[0], &option))
1302                return 0;
1303            if (option < 0)
1304                return 0;
1305            if (option == 0)
1306                option = EXT4_DEF_MAX_BATCH_TIME;
1307            sbi->s_max_batch_time = option;
1308            break;
1309        case Opt_min_batch_time:
1310            if (match_int(&args[0], &option))
1311                return 0;
1312            if (option < 0)
1313                return 0;
1314            sbi->s_min_batch_time = option;
1315            break;
1316        case Opt_data_journal:
1317            data_opt = EXT4_MOUNT_JOURNAL_DATA;
1318            goto datacheck;
1319        case Opt_data_ordered:
1320            data_opt = EXT4_MOUNT_ORDERED_DATA;
1321            goto datacheck;
1322        case Opt_data_writeback:
1323            data_opt = EXT4_MOUNT_WRITEBACK_DATA;
1324        datacheck:
1325            if (is_remount) {
1326                if ((sbi->s_mount_opt & EXT4_MOUNT_DATA_FLAGS)
1327                        != data_opt) {
1328                    ext4_msg(sb, KERN_ERR,
1329                        "Cannot change data mode on remount");
1330                    return 0;
1331                }
1332            } else {
1333                sbi->s_mount_opt &= ~EXT4_MOUNT_DATA_FLAGS;
1334                sbi->s_mount_opt |= data_opt;
1335            }
1336            break;
1337        case Opt_data_err_abort:
1338            set_opt(sbi->s_mount_opt, DATA_ERR_ABORT);
1339            break;
1340        case Opt_data_err_ignore:
1341            clear_opt(sbi->s_mount_opt, DATA_ERR_ABORT);
1342            break;
1343        case Opt_mb_history_length:
1344            if (match_int(&args[0], &option))
1345                return 0;
1346            if (option < 0)
1347                return 0;
1348            sbi->s_mb_history_max = option;
1349            break;
1350#ifdef CONFIG_QUOTA
1351        case Opt_usrjquota:
1352            qtype = USRQUOTA;
1353            goto set_qf_name;
1354        case Opt_grpjquota:
1355            qtype = GRPQUOTA;
1356set_qf_name:
1357            if (sb_any_quota_loaded(sb) &&
1358                !sbi->s_qf_names[qtype]) {
1359                ext4_msg(sb, KERN_ERR,
1360                       "Cannot change journaled "
1361                       "quota options when quota turned on");
1362                return 0;
1363            }
1364            qname = match_strdup(&args[0]);
1365            if (!qname) {
1366                ext4_msg(sb, KERN_ERR,
1367                    "Not enough memory for "
1368                    "storing quotafile name");
1369                return 0;
1370            }
1371            if (sbi->s_qf_names[qtype] &&
1372                strcmp(sbi->s_qf_names[qtype], qname)) {
1373                ext4_msg(sb, KERN_ERR,
1374                    "%s quota file already "
1375                    "specified", QTYPE2NAME(qtype));
1376                kfree(qname);
1377                return 0;
1378            }
1379            sbi->s_qf_names[qtype] = qname;
1380            if (strchr(sbi->s_qf_names[qtype], '/')) {
1381                ext4_msg(sb, KERN_ERR,
1382                    "quotafile must be on "
1383                    "filesystem root");
1384                kfree(sbi->s_qf_names[qtype]);
1385                sbi->s_qf_names[qtype] = NULL;
1386                return 0;
1387            }
1388            set_opt(sbi->s_mount_opt, QUOTA);
1389            break;
1390        case Opt_offusrjquota:
1391            qtype = USRQUOTA;
1392            goto clear_qf_name;
1393        case Opt_offgrpjquota:
1394            qtype = GRPQUOTA;
1395clear_qf_name:
1396            if (sb_any_quota_loaded(sb) &&
1397                sbi->s_qf_names[qtype]) {
1398                ext4_msg(sb, KERN_ERR, "Cannot change "
1399                    "journaled quota options when "
1400                    "quota turned on");
1401                return 0;
1402            }
1403            /*
1404             * The space will be released later when all options
1405             * are confirmed to be correct
1406             */
1407            sbi->s_qf_names[qtype] = NULL;
1408            break;
1409        case Opt_jqfmt_vfsold:
1410            qfmt = QFMT_VFS_OLD;
1411            goto set_qf_format;
1412        case Opt_jqfmt_vfsv0:
1413            qfmt = QFMT_VFS_V0;
1414set_qf_format:
1415            if (sb_any_quota_loaded(sb) &&
1416                sbi->s_jquota_fmt != qfmt) {
1417                ext4_msg(sb, KERN_ERR, "Cannot change "
1418                    "journaled quota options when "
1419                    "quota turned on");
1420                return 0;
1421            }
1422            sbi->s_jquota_fmt = qfmt;
1423            break;
1424        case Opt_quota:
1425        case Opt_usrquota:
1426            set_opt(sbi->s_mount_opt, QUOTA);
1427            set_opt(sbi->s_mount_opt, USRQUOTA);
1428            break;
1429        case Opt_grpquota:
1430            set_opt(sbi->s_mount_opt, QUOTA);
1431            set_opt(sbi->s_mount_opt, GRPQUOTA);
1432            break;
1433        case Opt_noquota:
1434            if (sb_any_quota_loaded(sb)) {
1435                ext4_msg(sb, KERN_ERR, "Cannot change quota "
1436                    "options when quota turned on");
1437                return 0;
1438            }
1439            clear_opt(sbi->s_mount_opt, QUOTA);
1440            clear_opt(sbi->s_mount_opt, USRQUOTA);
1441            clear_opt(sbi->s_mount_opt, GRPQUOTA);
1442            break;
1443#else
1444        case Opt_quota:
1445        case Opt_usrquota:
1446        case Opt_grpquota:
1447            ext4_msg(sb, KERN_ERR,
1448                "quota options not supported");
1449            break;
1450        case Opt_usrjquota:
1451        case Opt_grpjquota:
1452        case Opt_offusrjquota:
1453        case Opt_offgrpjquota:
1454        case Opt_jqfmt_vfsold:
1455        case Opt_jqfmt_vfsv0:
1456            ext4_msg(sb, KERN_ERR,
1457                "journaled quota options not supported");
1458            break;
1459        case Opt_noquota:
1460            break;
1461#endif
1462        case Opt_abort:
1463            sbi->s_mount_flags |= EXT4_MF_FS_ABORTED;
1464            break;
1465        case Opt_nobarrier:
1466            clear_opt(sbi->s_mount_opt, BARRIER);
1467            break;
1468        case Opt_barrier:
1469            if (match_int(&args[0], &option)) {
1470                set_opt(sbi->s_mount_opt, BARRIER);
1471                break;
1472            }
1473            if (option)
1474                set_opt(sbi->s_mount_opt, BARRIER);
1475            else
1476                clear_opt(sbi->s_mount_opt, BARRIER);
1477            break;
1478        case Opt_ignore:
1479            break;
1480        case Opt_resize:
1481            if (!is_remount) {
1482                ext4_msg(sb, KERN_ERR,
1483                    "resize option only available "
1484                    "for remount");
1485                return 0;
1486            }
1487            if (match_int(&args[0], &option) != 0)
1488                return 0;
1489            *n_blocks_count = option;
1490            break;
1491        case Opt_nobh:
1492            set_opt(sbi->s_mount_opt, NOBH);
1493            break;
1494        case Opt_bh:
1495            clear_opt(sbi->s_mount_opt, NOBH);
1496            break;
1497        case Opt_i_version:
1498            set_opt(sbi->s_mount_opt, I_VERSION);
1499            sb->s_flags |= MS_I_VERSION;
1500            break;
1501        case Opt_nodelalloc:
1502            clear_opt(sbi->s_mount_opt, DELALLOC);
1503            break;
1504        case Opt_stripe:
1505            if (match_int(&args[0], &option))
1506                return 0;
1507            if (option < 0)
1508                return 0;
1509            sbi->s_stripe = option;
1510            break;
1511        case Opt_delalloc:
1512            set_opt(sbi->s_mount_opt, DELALLOC);
1513            break;
1514        case Opt_block_validity:
1515            set_opt(sbi->s_mount_opt, BLOCK_VALIDITY);
1516            break;
1517        case Opt_noblock_validity:
1518            clear_opt(sbi->s_mount_opt, BLOCK_VALIDITY);
1519            break;
1520        case Opt_inode_readahead_blks:
1521            if (match_int(&args[0], &option))
1522                return 0;
1523            if (option < 0 || option > (1 << 30))
1524                return 0;
1525            if (!is_power_of_2(option)) {
1526                ext4_msg(sb, KERN_ERR,
1527                     "EXT4-fs: inode_readahead_blks"
1528                     " must be a power of 2");
1529                return 0;
1530            }
1531            sbi->s_inode_readahead_blks = option;
1532            break;
1533        case Opt_journal_ioprio:
1534            if (match_int(&args[0], &option))
1535                return 0;
1536            if (option < 0 || option > 7)
1537                break;
1538            *journal_ioprio = IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE,
1539                                option);
1540            break;
1541        case Opt_noauto_da_alloc:
1542            set_opt(sbi->s_mount_opt,NO_AUTO_DA_ALLOC);
1543            break;
1544        case Opt_auto_da_alloc:
1545            if (match_int(&args[0], &option)) {
1546                clear_opt(sbi->s_mount_opt, NO_AUTO_DA_ALLOC);
1547                break;
1548            }
1549            if (option)
1550                clear_opt(sbi->s_mount_opt, NO_AUTO_DA_ALLOC);
1551            else
1552                set_opt(sbi->s_mount_opt,NO_AUTO_DA_ALLOC);
1553            break;
1554        default:
1555            ext4_msg(sb, KERN_ERR,
1556                   "Unrecognized mount option \"%s\" "
1557                   "or missing value", p);
1558            return 0;
1559        }
1560    }
1561#ifdef CONFIG_QUOTA
1562    if (sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]) {
1563        if ((sbi->s_mount_opt & EXT4_MOUNT_USRQUOTA) &&
1564             sbi->s_qf_names[USRQUOTA])
1565            clear_opt(sbi->s_mount_opt, USRQUOTA);
1566
1567        if ((sbi->s_mount_opt & EXT4_MOUNT_GRPQUOTA) &&
1568             sbi->s_qf_names[GRPQUOTA])
1569            clear_opt(sbi->s_mount_opt, GRPQUOTA);
1570
1571        if ((sbi->s_qf_names[USRQUOTA] &&
1572                (sbi->s_mount_opt & EXT4_MOUNT_GRPQUOTA)) ||
1573            (sbi->s_qf_names[GRPQUOTA] &&
1574                (sbi->s_mount_opt & EXT4_MOUNT_USRQUOTA))) {
1575            ext4_msg(sb, KERN_ERR, "old and new quota "
1576                    "format mixing");
1577            return 0;
1578        }
1579
1580        if (!sbi->s_jquota_fmt) {
1581            ext4_msg(sb, KERN_ERR, "journaled quota format "
1582                    "not specified");
1583            return 0;
1584        }
1585    } else {
1586        if (sbi->s_jquota_fmt) {
1587            ext4_msg(sb, KERN_ERR, "journaled quota format "
1588                    "specified with no journaling "
1589                    "enabled");
1590            return 0;
1591        }
1592    }
1593#endif
1594    return 1;
1595}
1596
1597static int ext4_setup_super(struct super_block *sb, struct ext4_super_block *es,
1598                int read_only)
1599{
1600    struct ext4_sb_info *sbi = EXT4_SB(sb);
1601    int res = 0;
1602
1603    if (le32_to_cpu(es->s_rev_level) > EXT4_MAX_SUPP_REV) {
1604        ext4_msg(sb, KERN_ERR, "revision level too high, "
1605             "forcing read-only mode");
1606        res = MS_RDONLY;
1607    }
1608    if (read_only)
1609        return res;
1610    if (!(sbi->s_mount_state & EXT4_VALID_FS))
1611        ext4_msg(sb, KERN_WARNING, "warning: mounting unchecked fs, "
1612             "running e2fsck is recommended");
1613    else if ((sbi->s_mount_state & EXT4_ERROR_FS))
1614        ext4_msg(sb, KERN_WARNING,
1615             "warning: mounting fs with errors, "
1616             "running e2fsck is recommended");
1617    else if ((__s16) le16_to_cpu(es->s_max_mnt_count) >= 0 &&
1618         le16_to_cpu(es->s_mnt_count) >=
1619         (unsigned short) (__s16) le16_to_cpu(es->s_max_mnt_count))
1620        ext4_msg(sb, KERN_WARNING,
1621             "warning: maximal mount count reached, "
1622             "running e2fsck is recommended");
1623    else if (le32_to_cpu(es->s_checkinterval) &&
1624        (le32_to_cpu(es->s_lastcheck) +
1625            le32_to_cpu(es->s_checkinterval) <= get_seconds()))
1626        ext4_msg(sb, KERN_WARNING,
1627             "warning: checktime reached, "
1628             "running e2fsck is recommended");
1629    if (!sbi->s_journal)
1630        es->s_state &= cpu_to_le16(~EXT4_VALID_FS);
1631    if (!(__s16) le16_to_cpu(es->s_max_mnt_count))
1632        es->s_max_mnt_count = cpu_to_le16(EXT4_DFL_MAX_MNT_COUNT);
1633    le16_add_cpu(&es->s_mnt_count, 1);
1634    es->s_mtime = cpu_to_le32(get_seconds());
1635    ext4_update_dynamic_rev(sb);
1636    if (sbi->s_journal)
1637        EXT4_SET_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER);
1638
1639    ext4_commit_super(sb, 1);
1640    if (test_opt(sb, DEBUG))
1641        printk(KERN_INFO "[EXT4 FS bs=%lu, gc=%u, "
1642                "bpg=%lu, ipg=%lu, mo=%04x]\n",
1643            sb->s_blocksize,
1644            sbi->s_groups_count,
1645            EXT4_BLOCKS_PER_GROUP(sb),
1646            EXT4_INODES_PER_GROUP(sb),
1647            sbi->s_mount_opt);
1648
1649    if (EXT4_SB(sb)->s_journal) {
1650        ext4_msg(sb, KERN_INFO, "%s journal on %s",
1651               EXT4_SB(sb)->s_journal->j_inode ? "internal" :
1652               "external", EXT4_SB(sb)->s_journal->j_devname);
1653    } else {
1654        ext4_msg(sb, KERN_INFO, "no journal");
1655    }
1656    return res;
1657}
1658
1659static int ext4_fill_flex_info(struct super_block *sb)
1660{
1661    struct ext4_sb_info *sbi = EXT4_SB(sb);
1662    struct ext4_group_desc *gdp = NULL;
1663    ext4_group_t flex_group_count;
1664    ext4_group_t flex_group;
1665    int groups_per_flex = 0;
1666    size_t size;
1667    int i;
1668
1669    if (!sbi->s_es->s_log_groups_per_flex) {
1670        sbi->s_log_groups_per_flex = 0;
1671        return 1;
1672    }
1673
1674    sbi->s_log_groups_per_flex = sbi->s_es->s_log_groups_per_flex;
1675    groups_per_flex = 1 << sbi->s_log_groups_per_flex;
1676
1677    /* We allocate both existing and potentially added groups */
1678    flex_group_count = ((sbi->s_groups_count + groups_per_flex - 1) +
1679            ((le16_to_cpu(sbi->s_es->s_reserved_gdt_blocks) + 1) <<
1680                  EXT4_DESC_PER_BLOCK_BITS(sb))) / groups_per_flex;
1681    size = flex_group_count * sizeof(struct flex_groups);
1682    sbi->s_flex_groups = kzalloc(size, GFP_KERNEL);
1683    if (sbi->s_flex_groups == NULL) {
1684        sbi->s_flex_groups = vmalloc(size);
1685        if (sbi->s_flex_groups)
1686            memset(sbi->s_flex_groups, 0, size);
1687    }
1688    if (sbi->s_flex_groups == NULL) {
1689        ext4_msg(sb, KERN_ERR, "not enough memory for "
1690                "%u flex groups", flex_group_count);
1691        goto failed;
1692    }
1693
1694    for (i = 0; i < sbi->s_groups_count; i++) {
1695        gdp = ext4_get_group_desc(sb, i, NULL);
1696
1697        flex_group = ext4_flex_group(sbi, i);
1698        atomic_set(&sbi->s_flex_groups[flex_group].free_inodes,
1699               ext4_free_inodes_count(sb, gdp));
1700        atomic_set(&sbi->s_flex_groups[flex_group].free_blocks,
1701               ext4_free_blks_count(sb, gdp));
1702        atomic_set(&sbi->s_flex_groups[flex_group].used_dirs,
1703               ext4_used_dirs_count(sb, gdp));
1704    }
1705
1706    return 1;
1707failed:
1708    return 0;
1709}
1710
1711__le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 block_group,
1712                struct ext4_group_desc *gdp)
1713{
1714    __u16 crc = 0;
1715
1716    if (sbi->s_es->s_feature_ro_compat &
1717        cpu_to_le32(EXT4_FEATURE_RO_COMPAT_GDT_CSUM)) {
1718        int offset = offsetof(struct ext4_group_desc, bg_checksum);
1719        __le32 le_group = cpu_to_le32(block_group);
1720
1721        crc = crc16(~0, sbi->s_es->s_uuid, sizeof(sbi->s_es->s_uuid));
1722        crc = crc16(crc, (__u8 *)&le_group, sizeof(le_group));
1723        crc = crc16(crc, (__u8 *)gdp, offset);
1724        offset += sizeof(gdp->bg_checksum); /* skip checksum */
1725        /* for checksum of struct ext4_group_desc do the rest...*/
1726        if ((sbi->s_es->s_feature_incompat &
1727             cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT)) &&
1728            offset < le16_to_cpu(sbi->s_es->s_desc_size))
1729            crc = crc16(crc, (__u8 *)gdp + offset,
1730                    le16_to_cpu(sbi->s_es->s_desc_size) -
1731                    offset);
1732    }
1733
1734    return cpu_to_le16(crc);
1735}
1736
1737int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 block_group,
1738                struct ext4_group_desc *gdp)
1739{
1740    if ((sbi->s_es->s_feature_ro_compat &
1741         cpu_to_le32(EXT4_FEATURE_RO_COMPAT_GDT_CSUM)) &&
1742        (gdp->bg_checksum != ext4_group_desc_csum(sbi, block_group, gdp)))
1743        return 0;
1744
1745    return 1;
1746}
1747
1748/* Called at mount-time, super-block is locked */
1749static int ext4_check_descriptors(struct super_block *sb)
1750{
1751    struct ext4_sb_info *sbi = EXT4_SB(sb);
1752    ext4_fsblk_t first_block = le32_to_cpu(sbi->s_es->s_first_data_block);
1753    ext4_fsblk_t last_block;
1754    ext4_fsblk_t block_bitmap;
1755    ext4_fsblk_t inode_bitmap;
1756    ext4_fsblk_t inode_table;
1757    int flexbg_flag = 0;
1758    ext4_group_t i;
1759
1760    if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FLEX_BG))
1761        flexbg_flag = 1;
1762
1763    ext4_debug("Checking group descriptors");
1764
1765    for (i = 0; i < sbi->s_groups_count; i++) {
1766        struct ext4_group_desc *gdp = ext4_get_group_desc(sb, i, NULL);
1767
1768        if (i == sbi->s_groups_count - 1 || flexbg_flag)
1769            last_block = ext4_blocks_count(sbi->s_es) - 1;
1770        else
1771            last_block = first_block +
1772                (EXT4_BLOCKS_PER_GROUP(sb) - 1);
1773
1774        block_bitmap = ext4_block_bitmap(sb, gdp);
1775        if (block_bitmap < first_block || block_bitmap > last_block) {
1776            ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: "
1777                   "Block bitmap for group %u not in group "
1778                   "(block %llu)!", i, block_bitmap);
1779            return 0;
1780        }
1781        inode_bitmap = ext4_inode_bitmap(sb, gdp);
1782        if (inode_bitmap < first_block || inode_bitmap > last_block) {
1783            ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: "
1784                   "Inode bitmap for group %u not in group "
1785                   "(block %llu)!", i, inode_bitmap);
1786            return 0;
1787        }
1788        inode_table = ext4_inode_table(sb, gdp);
1789        if (inode_table < first_block ||
1790            inode_table + sbi->s_itb_per_group - 1 > last_block) {
1791            ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: "
1792                   "Inode table for group %u not in group "
1793                   "(block %llu)!", i, inode_table);
1794            return 0;
1795        }
1796        ext4_lock_group(sb, i);
1797        if (!ext4_group_desc_csum_verify(sbi, i, gdp)) {
1798            ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: "
1799                 "Checksum for group %u failed (%u!=%u)",
1800                 i, le16_to_cpu(ext4_group_desc_csum(sbi, i,
1801                     gdp)), le16_to_cpu(gdp->bg_checksum));
1802            if (!(sb->s_flags & MS_RDONLY)) {
1803                ext4_unlock_group(sb, i);
1804                return 0;
1805            }
1806        }
1807        ext4_unlock_group(sb, i);
1808        if (!flexbg_flag)
1809            first_block += EXT4_BLOCKS_PER_GROUP(sb);
1810    }
1811
1812    ext4_free_blocks_count_set(sbi->s_es, ext4_count_free_blocks(sb));
1813    sbi->s_es->s_free_inodes_count =cpu_to_le32(ext4_count_free_inodes(sb));
1814    return 1;
1815}
1816
1817/* ext4_orphan_cleanup() walks a singly-linked list of inodes (starting at
1818 * the superblock) which were deleted from all directories, but held open by
1819 * a process at the time of a crash. We walk the list and try to delete these
1820 * inodes at recovery time (only with a read-write filesystem).
1821 *
1822 * In order to keep the orphan inode chain consistent during traversal (in
1823 * case of crash during recovery), we link each inode into the superblock
1824 * orphan list_head and handle it the same way as an inode deletion during
1825 * normal operation (which journals the operations for us).
1826 *
1827 * We only do an iget() and an iput() on each inode, which is very safe if we
1828 * accidentally point at an in-use or already deleted inode. The worst that
1829 * can happen in this case is that we get a "bit already cleared" message from
1830 * ext4_free_inode(). The only reason we would point at a wrong inode is if
1831 * e2fsck was run on this filesystem, and it must have already done the orphan
1832 * inode cleanup for us, so we can safely abort without any further action.
1833 */
1834static void ext4_orphan_cleanup(struct super_block *sb,
1835                struct ext4_super_block *es)
1836{
1837    unsigned int s_flags = sb->s_flags;
1838    int nr_orphans = 0, nr_truncates = 0;
1839#ifdef CONFIG_QUOTA
1840    int i;
1841#endif
1842    if (!es->s_last_orphan) {
1843        jbd_debug(4, "no orphan inodes to clean up\n");
1844        return;
1845    }
1846
1847    if (bdev_read_only(sb->s_bdev)) {
1848        ext4_msg(sb, KERN_ERR, "write access "
1849            "unavailable, skipping orphan cleanup");
1850        return;
1851    }
1852
1853    if (EXT4_SB(sb)->s_mount_state & EXT4_ERROR_FS) {
1854        if (es->s_last_orphan)
1855            jbd_debug(1, "Errors on filesystem, "
1856                  "clearing orphan list.\n");
1857        es->s_last_orphan = 0;
1858        jbd_debug(1, "Skipping orphan recovery on fs with errors.\n");
1859        return;
1860    }
1861
1862    if (s_flags & MS_RDONLY) {
1863        ext4_msg(sb, KERN_INFO, "orphan cleanup on readonly fs");
1864        sb->s_flags &= ~MS_RDONLY;
1865    }
1866#ifdef CONFIG_QUOTA
1867    /* Needed for iput() to work correctly and not trash data */
1868    sb->s_flags |= MS_ACTIVE;
1869    /* Turn on quotas so that they are updated correctly */
1870    for (i = 0; i < MAXQUOTAS; i++) {
1871        if (EXT4_SB(sb)->s_qf_names[i]) {
1872            int ret = ext4_quota_on_mount(sb, i);
1873            if (ret < 0)
1874                ext4_msg(sb, KERN_ERR,
1875                    "Cannot turn on journaled "
1876                    "quota: error %d", ret);
1877        }
1878    }
1879#endif
1880
1881    while (es->s_last_orphan) {
1882        struct inode *inode;
1883
1884        inode = ext4_orphan_get(sb, le32_to_cpu(es->s_last_orphan));
1885        if (IS_ERR(inode)) {
1886            es->s_last_orphan = 0;
1887            break;
1888        }
1889
1890        list_add(&EXT4_I(inode)->i_orphan, &EXT4_SB(sb)->s_orphan);
1891        vfs_dq_init(inode);
1892        if (inode->i_nlink) {
1893            ext4_msg(sb, KERN_DEBUG,
1894                "%s: truncating inode %lu to %lld bytes",
1895                __func__, inode->i_ino, inode->i_size);
1896            jbd_debug(2, "truncating inode %lu to %lld bytes\n",
1897                  inode->i_ino, inode->i_size);
1898            ext4_truncate(inode);
1899            nr_truncates++;
1900        } else {
1901            ext4_msg(sb, KERN_DEBUG,
1902                "%s: deleting unreferenced inode %lu",
1903                __func__, inode->i_ino);
1904            jbd_debug(2, "deleting unreferenced inode %lu\n",
1905                  inode->i_ino);
1906            nr_orphans++;
1907        }
1908        iput(inode); /* The delete magic happens here! */
1909    }
1910
1911#define PLURAL(x) (x), ((x) == 1) ? "" : "s"
1912
1913    if (nr_orphans)
1914        ext4_msg(sb, KERN_INFO, "%d orphan inode%s deleted",
1915               PLURAL(nr_orphans));
1916    if (nr_truncates)
1917        ext4_msg(sb, KERN_INFO, "%d truncate%s cleaned up",
1918               PLURAL(nr_truncates));
1919#ifdef CONFIG_QUOTA
1920    /* Turn quotas off */
1921    for (i = 0; i < MAXQUOTAS; i++) {
1922        if (sb_dqopt(sb)->files[i])
1923            vfs_quota_off(sb, i, 0);
1924    }
1925#endif
1926    sb->s_flags = s_flags; /* Restore MS_RDONLY status */
1927}
1928
1929/*
1930 * Maximal extent format file size.
1931 * Resulting logical blkno at s_maxbytes must fit in our on-disk
1932 * extent format containers, within a sector_t, and within i_blocks
1933 * in the vfs. ext4 inode has 48 bits of i_block in fsblock units,
1934 * so that won't be a limiting factor.
1935 *
1936 * Note, this does *not* consider any metadata overhead for vfs i_blocks.
1937 */
1938static loff_t ext4_max_size(int blkbits, int has_huge_files)
1939{
1940    loff_t res;
1941    loff_t upper_limit = MAX_LFS_FILESIZE;
1942
1943    /* small i_blocks in vfs inode? */
1944    if (!has_huge_files || sizeof(blkcnt_t) < sizeof(u64)) {
1945        /*
1946         * CONFIG_LBDAF is not enabled implies the inode
1947         * i_block represent total blocks in 512 bytes
1948         * 32 == size of vfs inode i_blocks * 8
1949         */
1950        upper_limit = (1LL << 32) - 1;
1951
1952        /* total blocks in file system block size */
1953        upper_limit >>= (blkbits - 9);
1954        upper_limit <<= blkbits;
1955    }
1956
1957    /* 32-bit extent-start container, ee_block */
1958    res = 1LL << 32;
1959    res <<= blkbits;
1960    res -= 1;
1961
1962    /* Sanity check against vm- & vfs- imposed limits */
1963    if (res > upper_limit)
1964        res = upper_limit;
1965
1966    return res;
1967}
1968
1969/*
1970 * Maximal bitmap file size. There is a direct, and {,double-,triple-}indirect
1971 * block limit, and also a limit of (2^48 - 1) 512-byte sectors in i_blocks.
1972 * We need to be 1 filesystem block less than the 2^48 sector limit.
1973 */
1974static loff_t ext4_max_bitmap_size(int bits, int has_huge_files)
1975{
1976    loff_t res = EXT4_NDIR_BLOCKS;
1977    int meta_blocks;
1978    loff_t upper_limit;
1979    /* This is calculated to be the largest file size for a dense, block
1980     * mapped file such that the file's total number of 512-byte sectors,
1981     * including data and all indirect blocks, does not exceed (2^48 - 1).
1982     *
1983     * __u32 i_blocks_lo and _u16 i_blocks_high represent the total
1984     * number of 512-byte sectors of the file.
1985     */
1986
1987    if (!has_huge_files || sizeof(blkcnt_t) < sizeof(u64)) {
1988        /*
1989         * !has_huge_files or CONFIG_LBDAF not enabled implies that
1990         * the inode i_block field represents total file blocks in
1991         * 2^32 512-byte sectors == size of vfs inode i_blocks * 8
1992         */
1993        upper_limit = (1LL << 32) - 1;
1994
1995        /* total blocks in file system block size */
1996        upper_limit >>= (bits - 9);
1997
1998    } else {
1999        /*
2000         * We use 48 bit ext4_inode i_blocks
2001         * With EXT4_HUGE_FILE_FL set the i_blocks
2002         * represent total number of blocks in
2003         * file system block size
2004         */
2005        upper_limit = (1LL << 48) - 1;
2006
2007    }
2008
2009    /* indirect blocks */
2010    meta_blocks = 1;
2011    /* double indirect blocks */
2012    meta_blocks += 1 + (1LL << (bits-2));
2013    /* tripple indirect blocks */
2014    meta_blocks += 1 + (1LL << (bits-2)) + (1LL << (2*(bits-2)));
2015
2016    upper_limit -= meta_blocks;
2017    upper_limit <<= bits;
2018
2019    res += 1LL << (bits-2);
2020    res += 1LL << (2*(bits-2));
2021    res += 1LL << (3*(bits-2));
2022    res <<= bits;
2023    if (res > upper_limit)
2024        res = upper_limit;
2025
2026    if (res > MAX_LFS_FILESIZE)
2027        res = MAX_LFS_FILESIZE;
2028
2029    return res;
2030}
2031
2032static ext4_fsblk_t descriptor_loc(struct super_block *sb,
2033                   ext4_fsblk_t logical_sb_block, int nr)
2034{
2035    struct ext4_sb_info *sbi = EXT4_SB(sb);
2036    ext4_group_t bg, first_meta_bg;
2037    int has_super = 0;
2038
2039    first_meta_bg = le32_to_cpu(sbi->s_es->s_first_meta_bg);
2040
2041    if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_META_BG) ||
2042        nr < first_meta_bg)
2043        return logical_sb_block + nr + 1;
2044    bg = sbi->s_desc_per_block * nr;
2045    if (ext4_bg_has_super(sb, bg))
2046        has_super = 1;
2047
2048    return (has_super + ext4_group_first_block_no(sb, bg));
2049}
2050
2051/**
2052 * ext4_get_stripe_size: Get the stripe size.
2053 * @sbi: In memory super block info
2054 *
2055 * If we have specified it via mount option, then
2056 * use the mount option value. If the value specified at mount time is
2057 * greater than the blocks per group use the super block value.
2058 * If the super block value is greater than blocks per group return 0.
2059 * Allocator needs it be less than blocks per group.
2060 *
2061 */
2062static unsigned long ext4_get_stripe_size(struct ext4_sb_info *sbi)
2063{
2064    unsigned long stride = le16_to_cpu(sbi->s_es->s_raid_stride);
2065    unsigned long stripe_width =
2066            le32_to_cpu(sbi->s_es->s_raid_stripe_width);
2067
2068    if (sbi->s_stripe && sbi->s_stripe <= sbi->s_blocks_per_group)
2069        return sbi->s_stripe;
2070
2071    if (stripe_width <= sbi->s_blocks_per_group)
2072        return stripe_width;
2073
2074    if (stride <= sbi->s_blocks_per_group)
2075        return stride;
2076
2077    return 0;
2078}
2079
2080/* sysfs supprt */
2081
2082struct ext4_attr {
2083    struct attribute attr;
2084    ssize_t (*show)(struct ext4_attr *, struct ext4_sb_info *, char *);
2085    ssize_t (*store)(struct ext4_attr *, struct ext4_sb_info *,
2086             const char *, size_t);
2087    int offset;
2088};
2089
2090static int parse_strtoul(const char *buf,
2091        unsigned long max, unsigned long *value)
2092{
2093    char *endp;
2094
2095    while (*buf && isspace(*buf))
2096        buf++;
2097    *value = simple_strtoul(buf, &endp, 0);
2098    while (*endp && isspace(*endp))
2099        endp++;
2100    if (*endp || *value > max)
2101        return -EINVAL;
2102
2103    return 0;
2104}
2105
2106static ssize_t delayed_allocation_blocks_show(struct ext4_attr *a,
2107                          struct ext4_sb_info *sbi,
2108                          char *buf)
2109{
2110    return snprintf(buf, PAGE_SIZE, "%llu\n",
2111            (s64) percpu_counter_sum(&sbi->s_dirtyblocks_counter));
2112}
2113
2114static ssize_t session_write_kbytes_show(struct ext4_attr *a,
2115                     struct ext4_sb_info *sbi, char *buf)
2116{
2117    struct super_block *sb = sbi->s_buddy_cache->i_sb;
2118
2119    return snprintf(buf, PAGE_SIZE, "%lu\n",
2120            (part_stat_read(sb->s_bdev->bd_part, sectors[1]) -
2121             sbi->s_sectors_written_start) >> 1);
2122}
2123
2124static ssize_t lifetime_write_kbytes_show(struct ext4_attr *a,
2125                      struct ext4_sb_info *sbi, char *buf)
2126{
2127    struct super_block *sb = sbi->s_buddy_cache->i_sb;
2128
2129    return snprintf(buf, PAGE_SIZE, "%llu\n",
2130            sbi->s_kbytes_written +
2131            ((part_stat_read(sb->s_bdev->bd_part, sectors[1]) -
2132              EXT4_SB(sb)->s_sectors_written_start) >> 1));
2133}
2134
2135static ssize_t inode_readahead_blks_store(struct ext4_attr *a,
2136                      struct ext4_sb_info *sbi,
2137                      const char *buf, size_t count)
2138{
2139    unsigned long t;
2140
2141    if (parse_strtoul(buf, 0x40000000, &t))
2142        return -EINVAL;
2143
2144    if (!is_power_of_2(t))
2145        return -EINVAL;
2146
2147    sbi->s_inode_readahead_blks = t;
2148    return count;
2149}
2150
2151static ssize_t sbi_ui_show(struct ext4_attr *a,
2152               struct ext4_sb_info *sbi, char *buf)
2153{
2154    unsigned int *ui = (unsigned int *) (((char *) sbi) + a->offset);
2155
2156    return snprintf(buf, PAGE_SIZE, "%u\n", *ui);
2157}
2158
2159static ssize_t sbi_ui_store(struct ext4_attr *a,
2160                struct ext4_sb_info *sbi,
2161                const char *buf, size_t count)
2162{
2163    unsigned int *ui = (unsigned int *) (((char *) sbi) + a->offset);
2164    unsigned long t;
2165
2166    if (parse_strtoul(buf, 0xffffffff, &t))
2167        return -EINVAL;
2168    *ui = t;
2169    return count;
2170}
2171
2172#define EXT4_ATTR_OFFSET(_name,_mode,_show,_store,_elname) \
2173static struct ext4_attr ext4_attr_##_name = { \
2174    .attr = {.name = __stringify(_name), .mode = _mode }, \
2175    .show = _show, \
2176    .store = _store, \
2177    .offset = offsetof(struct ext4_sb_info, _elname), \
2178}
2179#define EXT4_ATTR(name, mode, show, store) \
2180static struct ext4_attr ext4_attr_##name = __ATTR(name, mode, show, store)
2181
2182#define EXT4_RO_ATTR(name) EXT4_ATTR(name, 0444, name##_show, NULL)
2183#define EXT4_RW_ATTR(name) EXT4_ATTR(name, 0644, name##_show, name##_store)
2184#define EXT4_RW_ATTR_SBI_UI(name, elname) \
2185    EXT4_ATTR_OFFSET(name, 0644, sbi_ui_show, sbi_ui_store, elname)
2186#define ATTR_LIST(name) &ext4_attr_##name.attr
2187
2188EXT4_RO_ATTR(delayed_allocation_blocks);
2189EXT4_RO_ATTR(session_write_kbytes);
2190EXT4_RO_ATTR(lifetime_write_kbytes);
2191EXT4_ATTR_OFFSET(inode_readahead_blks, 0644, sbi_ui_show,
2192         inode_readahead_blks_store, s_inode_readahead_blks);
2193EXT4_RW_ATTR_SBI_UI(inode_goal, s_inode_goal);
2194EXT4_RW_ATTR_SBI_UI(mb_stats, s_mb_stats);
2195EXT4_RW_ATTR_SBI_UI(mb_max_to_scan, s_mb_max_to_scan);
2196EXT4_RW_ATTR_SBI_UI(mb_min_to_scan, s_mb_min_to_scan);
2197EXT4_RW_ATTR_SBI_UI(mb_order2_req, s_mb_order2_reqs);
2198EXT4_RW_ATTR_SBI_UI(mb_stream_req, s_mb_stream_request);
2199EXT4_RW_ATTR_SBI_UI(mb_group_prealloc, s_mb_group_prealloc);
2200
2201static struct attribute *ext4_attrs[] = {
2202    ATTR_LIST(delayed_allocation_blocks),
2203    ATTR_LIST(session_write_kbytes),
2204    ATTR_LIST(lifetime_write_kbytes),
2205    ATTR_LIST(inode_readahead_blks),
2206    ATTR_LIST(inode_goal),
2207    ATTR_LIST(mb_stats),
2208    ATTR_LIST(mb_max_to_scan),
2209    ATTR_LIST(mb_min_to_scan),
2210    ATTR_LIST(mb_order2_req),
2211    ATTR_LIST(mb_stream_req),
2212    ATTR_LIST(mb_group_prealloc),
2213    NULL,
2214};
2215
2216static ssize_t ext4_attr_show(struct kobject *kobj,
2217                  struct attribute *attr, char *buf)
2218{
2219    struct ext4_sb_info *sbi = container_of(kobj, struct ext4_sb_info,
2220                        s_kobj);
2221    struct ext4_attr *a = container_of(attr, struct ext4_attr, attr);
2222
2223    return a->show ? a->show(a, sbi, buf) : 0;
2224}
2225
2226static ssize_t ext4_attr_store(struct kobject *kobj,
2227                   struct attribute *attr,
2228                   const char *buf, size_t len)
2229{
2230    struct ext4_sb_info *sbi = container_of(kobj, struct ext4_sb_info,
2231                        s_kobj);
2232    struct ext4_attr *a = container_of(attr, struct ext4_attr, attr);
2233
2234    return a->store ? a->store(a, sbi, buf, len) : 0;
2235}
2236
2237static void ext4_sb_release(struct kobject *kobj)
2238{
2239    struct ext4_sb_info *sbi = container_of(kobj, struct ext4_sb_info,
2240                        s_kobj);
2241    complete(&sbi->s_kobj_unregister);
2242}
2243
2244
2245static struct sysfs_ops ext4_attr_ops = {
2246    .show = ext4_attr_show,
2247    .store = ext4_attr_store,
2248};
2249
2250static struct kobj_type ext4_ktype = {
2251    .default_attrs = ext4_attrs,
2252    .sysfs_ops = &ext4_attr_ops,
2253    .release = ext4_sb_release,
2254};
2255
2256static int ext4_fill_super(struct super_block *sb, void *data, int silent)
2257                __releases(kernel_lock)
2258                __acquires(kernel_lock)
2259{
2260    struct buffer_head *bh;
2261    struct ext4_super_block *es = NULL;
2262    struct ext4_sb_info *sbi;
2263    ext4_fsblk_t block;
2264    ext4_fsblk_t sb_block = get_sb_block(&data);
2265    ext4_fsblk_t logical_sb_block;
2266    unsigned long offset = 0;
2267    unsigned long journal_devnum = 0;
2268    unsigned long def_mount_opts;
2269    struct inode *root;
2270    char *cp;
2271    const char *descr;
2272    int ret = -EINVAL;
2273    int blocksize;
2274    unsigned int db_count;
2275    unsigned int i;
2276    int needs_recovery, has_huge_files;
2277    int features;
2278    __u64 blocks_count;
2279    int err;
2280    unsigned int journal_ioprio = DEFAULT_JOURNAL_IOPRIO;
2281
2282    sbi = kzalloc(sizeof(*sbi), GFP_KERNEL);
2283    if (!sbi)
2284        return -ENOMEM;
2285
2286    sbi->s_blockgroup_lock =
2287        kzalloc(sizeof(struct blockgroup_lock), GFP_KERNEL);
2288    if (!sbi->s_blockgroup_lock) {
2289        kfree(sbi);
2290        return -ENOMEM;
2291    }
2292    sb->s_fs_info = sbi;
2293    sbi->s_mount_opt = 0;
2294    sbi->s_resuid = EXT4_DEF_RESUID;
2295    sbi->s_resgid = EXT4_DEF_RESGID;
2296    sbi->s_inode_readahead_blks = EXT4_DEF_INODE_READAHEAD_BLKS;
2297    sbi->s_sb_block = sb_block;
2298    sbi->s_sectors_written_start = part_stat_read(sb->s_bdev->bd_part,
2299                              sectors[1]);
2300
2301    unlock_kernel();
2302
2303    /* Cleanup superblock name */
2304    for (cp = sb->s_id; (cp = strchr(cp, '/'));)
2305        *cp = '!';
2306
2307    blocksize = sb_min_blocksize(sb, EXT4_MIN_BLOCK_SIZE);
2308    if (!blocksize) {
2309        ext4_msg(sb, KERN_ERR, "unable to set blocksize");
2310        goto out_fail;
2311    }
2312
2313    /*
2314     * The ext4 superblock will not be buffer aligned for other than 1kB
2315     * block sizes. We need to calculate the offset from buffer start.
2316     */
2317    if (blocksize != EXT4_MIN_BLOCK_SIZE) {
2318        logical_sb_block = sb_block * EXT4_MIN_BLOCK_SIZE;
2319        offset = do_div(logical_sb_block, blocksize);
2320    } else {
2321        logical_sb_block = sb_block;
2322    }
2323
2324    if (!(bh = sb_bread(sb, logical_sb_block))) {
2325        ext4_msg(sb, KERN_ERR, "unable to read superblock");
2326        goto out_fail;
2327    }
2328    /*
2329     * Note: s_es must be initialized as soon as possible because
2330     * some ext4 macro-instructions depend on its value
2331     */
2332    es = (struct ext4_super_block *) (((char *)bh->b_data) + offset);
2333    sbi->s_es = es;
2334    sb->s_magic = le16_to_cpu(es->s_magic);
2335    if (sb->s_magic != EXT4_SUPER_MAGIC)
2336        goto cantfind_ext4;
2337    sbi->s_kbytes_written = le64_to_cpu(es->s_kbytes_written);
2338
2339    /* Set defaults before we parse the mount options */
2340    def_mount_opts = le32_to_cpu(es->s_default_mount_opts);
2341    if (def_mount_opts & EXT4_DEFM_DEBUG)
2342        set_opt(sbi->s_mount_opt, DEBUG);
2343    if (def_mount_opts & EXT4_DEFM_BSDGROUPS)
2344        set_opt(sbi->s_mount_opt, GRPID);
2345    if (def_mount_opts & EXT4_DEFM_UID16)
2346        set_opt(sbi->s_mount_opt, NO_UID32);
2347#ifdef CONFIG_EXT4_FS_XATTR
2348    if (def_mount_opts & EXT4_DEFM_XATTR_USER)
2349        set_opt(sbi->s_mount_opt, XATTR_USER);
2350#endif
2351#ifdef CONFIG_EXT4_FS_POSIX_ACL
2352    if (def_mount_opts & EXT4_DEFM_ACL)
2353        set_opt(sbi->s_mount_opt, POSIX_ACL);
2354#endif
2355    if ((def_mount_opts & EXT4_DEFM_JMODE) == EXT4_DEFM_JMODE_DATA)
2356        sbi->s_mount_opt |= EXT4_MOUNT_JOURNAL_DATA;
2357    else if ((def_mount_opts & EXT4_DEFM_JMODE) == EXT4_DEFM_JMODE_ORDERED)
2358        sbi->s_mount_opt |= EXT4_MOUNT_ORDERED_DATA;
2359    else if ((def_mount_opts & EXT4_DEFM_JMODE) == EXT4_DEFM_JMODE_WBACK)
2360        sbi->s_mount_opt |= EXT4_MOUNT_WRITEBACK_DATA;
2361
2362    if (le16_to_cpu(sbi->s_es->s_errors) == EXT4_ERRORS_PANIC)
2363        set_opt(sbi->s_mount_opt, ERRORS_PANIC);
2364    else if (le16_to_cpu(sbi->s_es->s_errors) == EXT4_ERRORS_CONTINUE)
2365        set_opt(sbi->s_mount_opt, ERRORS_CONT);
2366    else
2367        set_opt(sbi->s_mount_opt, ERRORS_RO);
2368
2369    sbi->s_resuid = le16_to_cpu(es->s_def_resuid);
2370    sbi->s_resgid = le16_to_cpu(es->s_def_resgid);
2371    sbi->s_commit_interval = JBD2_DEFAULT_MAX_COMMIT_AGE * HZ;
2372    sbi->s_min_batch_time = EXT4_DEF_MIN_BATCH_TIME;
2373    sbi->s_max_batch_time = EXT4_DEF_MAX_BATCH_TIME;
2374    sbi->s_mb_history_max = default_mb_history_length;
2375
2376    set_opt(sbi->s_mount_opt, BARRIER);
2377
2378    /*
2379     * enable delayed allocation by default
2380     * Use -o nodelalloc to turn it off
2381     */
2382    set_opt(sbi->s_mount_opt, DELALLOC);
2383
2384    if (!parse_options((char *) data, sb, &journal_devnum,
2385               &journal_ioprio, NULL, 0))
2386        goto failed_mount;
2387
2388    sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
2389        ((sbi->s_mount_opt & EXT4_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
2390
2391    if (le32_to_cpu(es->s_rev_level) == EXT4_GOOD_OLD_REV &&
2392        (EXT4_HAS_COMPAT_FEATURE(sb, ~0U) ||
2393         EXT4_HAS_RO_COMPAT_FEATURE(sb, ~0U) ||
2394         EXT4_HAS_INCOMPAT_FEATURE(sb, ~0U)))
2395        ext4_msg(sb, KERN_WARNING,
2396               "feature flags set on rev 0 fs, "
2397               "running e2fsck is recommended");
2398
2399    /*
2400     * Check feature flags regardless of the revision level, since we
2401     * previously didn't change the revision level when setting the flags,
2402     * so there is a chance incompat flags are set on a rev 0 filesystem.
2403     */
2404    features = EXT4_HAS_INCOMPAT_FEATURE(sb, ~EXT4_FEATURE_INCOMPAT_SUPP);
2405    if (features) {
2406        ext4_msg(sb, KERN_ERR,
2407            "Couldn't mount because of "
2408            "unsupported optional features (%x)",
2409            (le32_to_cpu(EXT4_SB(sb)->s_es->s_feature_incompat) &
2410            ~EXT4_FEATURE_INCOMPAT_SUPP));
2411        goto failed_mount;
2412    }
2413    features = EXT4_HAS_RO_COMPAT_FEATURE(sb, ~EXT4_FEATURE_RO_COMPAT_SUPP);
2414    if (!(sb->s_flags & MS_RDONLY) && features) {
2415        ext4_msg(sb, KERN_ERR,
2416            "Couldn't mount RDWR because of "
2417            "unsupported optional features (%x)",
2418            (le32_to_cpu(EXT4_SB(sb)->s_es->s_feature_ro_compat) &
2419            ~EXT4_FEATURE_RO_COMPAT_SUPP));
2420        goto failed_mount;
2421    }
2422    has_huge_files = EXT4_HAS_RO_COMPAT_FEATURE(sb,
2423                    EXT4_FEATURE_RO_COMPAT_HUGE_FILE);
2424    if (has_huge_files) {
2425        /*
2426         * Large file size enabled file system can only be
2427         * mount if kernel is build with CONFIG_LBDAF
2428         */
2429        if (sizeof(root->i_blocks) < sizeof(u64) &&
2430                !(sb->s_flags & MS_RDONLY)) {
2431            ext4_msg(sb, KERN_ERR, "Filesystem with huge "
2432                    "files cannot be mounted read-write "
2433                    "without CONFIG_LBDAF");
2434            goto failed_mount;
2435        }
2436    }
2437    blocksize = BLOCK_SIZE << le32_to_cpu(es->s_log_block_size);
2438
2439    if (blocksize < EXT4_MIN_BLOCK_SIZE ||
2440        blocksize > EXT4_MAX_BLOCK_SIZE) {
2441        ext4_msg(sb, KERN_ERR,
2442               "Unsupported filesystem blocksize %d", blocksize);
2443        goto failed_mount;
2444    }
2445
2446    if (sb->s_blocksize != blocksize) {
2447        /* Validate the filesystem blocksize */
2448        if (!sb_set_blocksize(sb, blocksize)) {
2449            ext4_msg(sb, KERN_ERR, "bad block size %d",
2450                    blocksize);
2451            goto failed_mount;
2452        }
2453
2454        brelse(bh);
2455        logical_sb_block = sb_block * EXT4_MIN_BLOCK_SIZE;
2456        offset = do_div(logical_sb_block, blocksize);
2457        bh = sb_bread(sb, logical_sb_block);
2458        if (!bh) {
2459            ext4_msg(sb, KERN_ERR,
2460                   "Can't read superblock on 2nd try");
2461            goto failed_mount;
2462        }
2463        es = (struct ext4_super_block *)(((char *)bh->b_data) + offset);
2464        sbi->s_es = es;
2465        if (es->s_magic != cpu_to_le16(EXT4_SUPER_MAGIC)) {
2466            ext4_msg(sb, KERN_ERR,
2467                   "Magic mismatch, very weird!");
2468            goto failed_mount;
2469        }
2470    }
2471
2472    sbi->s_bitmap_maxbytes = ext4_max_bitmap_size(sb->s_blocksize_bits,
2473                              has_huge_files);
2474    sb->s_maxbytes = ext4_max_size(sb->s_blocksize_bits, has_huge_files);
2475
2476    if (le32_to_cpu(es->s_rev_level) == EXT4_GOOD_OLD_REV) {
2477        sbi->s_inode_size = EXT4_GOOD_OLD_INODE_SIZE;
2478        sbi->s_first_ino = EXT4_GOOD_OLD_FIRST_INO;
2479    } else {
2480        sbi->s_inode_size = le16_to_cpu(es->s_inode_size);
2481        sbi->s_first_ino = le32_to_cpu(es->s_first_ino);
2482        if ((sbi->s_inode_size < EXT4_GOOD_OLD_INODE_SIZE) ||
2483            (!is_power_of_2(sbi->s_inode_size)) ||
2484            (sbi->s_inode_size > blocksize)) {
2485            ext4_msg(sb, KERN_ERR,
2486                   "unsupported inode size: %d",
2487                   sbi->s_inode_size);
2488            goto failed_mount;
2489        }
2490        if (sbi->s_inode_size > EXT4_GOOD_OLD_INODE_SIZE)
2491            sb->s_time_gran = 1 << (EXT4_EPOCH_BITS - 2);
2492    }
2493
2494    sbi->s_desc_size = le16_to_cpu(es->s_desc_size);
2495    if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_64BIT)) {
2496        if (sbi->s_desc_size < EXT4_MIN_DESC_SIZE_64BIT ||
2497            sbi->s_desc_size > EXT4_MAX_DESC_SIZE ||
2498            !is_power_of_2(sbi->s_desc_size)) {
2499            ext4_msg(sb, KERN_ERR,
2500                   "unsupported descriptor size %lu",
2501                   sbi->s_desc_size);
2502            goto failed_mount;
2503        }
2504    } else
2505        sbi->s_desc_size = EXT4_MIN_DESC_SIZE;
2506
2507    sbi->s_blocks_per_group = le32_to_cpu(es->s_blocks_per_group);
2508    sbi->s_inodes_per_group = le32_to_cpu(es->s_inodes_per_group);
2509    if (EXT4_INODE_SIZE(sb) == 0 || EXT4_INODES_PER_GROUP(sb) == 0)
2510        goto cantfind_ext4;
2511
2512    sbi->s_inodes_per_block = blocksize / EXT4_INODE_SIZE(sb);
2513    if (sbi->s_inodes_per_block == 0)
2514        goto cantfind_ext4;
2515    sbi->s_itb_per_group = sbi->s_inodes_per_group /
2516                    sbi->s_inodes_per_block;
2517    sbi->s_desc_per_block = blocksize / EXT4_DESC_SIZE(sb);
2518    sbi->s_sbh = bh;
2519    sbi->s_mount_state = le16_to_cpu(es->s_state);
2520    sbi->s_addr_per_block_bits = ilog2(EXT4_ADDR_PER_BLOCK(sb));
2521    sbi->s_desc_per_block_bits = ilog2(EXT4_DESC_PER_BLOCK(sb));
2522
2523    for (i = 0; i < 4; i++)
2524        sbi->s_hash_seed[i] = le32_to_cpu(es->s_hash_seed[i]);
2525    sbi->s_def_hash_version = es->s_def_hash_version;
2526    i = le32_to_cpu(es->s_flags);
2527    if (i & EXT2_FLAGS_UNSIGNED_HASH)
2528        sbi->s_hash_unsigned = 3;
2529    else if ((i & EXT2_FLAGS_SIGNED_HASH) == 0) {
2530#ifdef __CHAR_UNSIGNED__
2531        es->s_flags |= cpu_to_le32(EXT2_FLAGS_UNSIGNED_HASH);
2532        sbi->s_hash_unsigned = 3;
2533#else
2534        es->s_flags |= cpu_to_le32(EXT2_FLAGS_SIGNED_HASH);
2535#endif
2536        sb->s_dirt = 1;
2537    }
2538
2539    if (sbi->s_blocks_per_group > blocksize * 8) {
2540        ext4_msg(sb, KERN_ERR,
2541               "#blocks per group too big: %lu",
2542               sbi->s_blocks_per_group);
2543        goto failed_mount;
2544    }
2545    if (sbi->s_inodes_per_group > blocksize * 8) {
2546        ext4_msg(sb, KERN_ERR,
2547               "#inodes per group too big: %lu",
2548               sbi->s_inodes_per_group);
2549        goto failed_mount;
2550    }
2551
2552    if (ext4_blocks_count(es) >
2553            (sector_t)(~0ULL) >> (sb->s_blocksize_bits - 9)) {
2554        ext4_msg(sb, KERN_ERR, "filesystem"
2555            " too large to mount safely");
2556        if (sizeof(sector_t) < 8)
2557            ext4_msg(sb, KERN_WARNING, "CONFIG_LBDAF not enabled");
2558        goto failed_mount;
2559    }
2560
2561    if (EXT4_BLOCKS_PER_GROUP(sb) == 0)
2562        goto cantfind_ext4;
2563
2564    /* check blocks count against device size */
2565    blocks_count = sb->s_bdev->bd_inode->i_size >> sb->s_blocksize_bits;
2566    if (blocks_count && ext4_blocks_count(es) > blocks_count) {
2567        ext4_msg(sb, KERN_WARNING, "bad geometry: block count %llu "
2568               "exceeds size of device (%llu blocks)",
2569               ext4_blocks_count(es), blocks_count);
2570        goto failed_mount;
2571    }
2572
2573    /*
2574     * It makes no sense for the first data block to be beyond the end
2575     * of the filesystem.
2576     */
2577    if (le32_to_cpu(es->s_first_data_block) >= ext4_blocks_count(es)) {
2578                ext4_msg(sb, KERN_WARNING, "bad geometry: first data"
2579             "block %u is beyond end of filesystem (%llu)",
2580             le32_to_cpu(es->s_first_data_block),
2581             ext4_blocks_count(es));
2582        goto failed_mount;
2583    }
2584    blocks_count = (ext4_blocks_count(es) -
2585            le32_to_cpu(es->s_first_data_block) +
2586            EXT4_BLOCKS_PER_GROUP(sb) - 1);
2587    do_div(blocks_count, EXT4_BLOCKS_PER_GROUP(sb));
2588    if (blocks_count > ((uint64_t)1<<32) - EXT4_DESC_PER_BLOCK(sb)) {
2589        ext4_msg(sb, KERN_WARNING, "groups count too large: %u "
2590               "(block count %llu, first data block %u, "
2591               "blocks per group %lu)", sbi->s_groups_count,
2592               ext4_blocks_count(es),
2593               le32_to_cpu(es->s_first_data_block),
2594               EXT4_BLOCKS_PER_GROUP(sb));
2595        goto failed_mount;
2596    }
2597    sbi->s_groups_count = blocks_count;
2598    db_count = (sbi->s_groups_count + EXT4_DESC_PER_BLOCK(sb) - 1) /
2599           EXT4_DESC_PER_BLOCK(sb);
2600    sbi->s_group_desc = kmalloc(db_count * sizeof(struct buffer_head *),
2601                    GFP_KERNEL);
2602    if (sbi->s_group_desc == NULL) {
2603        ext4_msg(sb, KERN_ERR, "not enough memory");
2604        goto failed_mount;
2605    }
2606
2607#ifdef CONFIG_PROC_FS
2608    if (ext4_proc_root)
2609        sbi->s_proc = proc_mkdir(sb->s_id, ext4_proc_root);
2610#endif
2611
2612    bgl_lock_init(sbi->s_blockgroup_lock);
2613
2614    for (i = 0; i < db_count; i++) {
2615        block = descriptor_loc(sb, logical_sb_block, i);
2616        sbi->s_group_desc[i] = sb_bread(sb, block);
2617        if (!sbi->s_group_desc[i]) {
2618            ext4_msg(sb, KERN_ERR,
2619                   "can't read group descriptor %d", i);
2620            db_count = i;
2621            goto failed_mount2;
2622        }
2623    }
2624    if (!ext4_check_descriptors(sb)) {
2625        ext4_msg(sb, KERN_ERR, "group descriptors corrupted!");
2626        goto failed_mount2;
2627    }
2628    if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FLEX_BG))
2629        if (!ext4_fill_flex_info(sb)) {
2630            ext4_msg(sb, KERN_ERR,
2631                   "unable to initialize "
2632                   "flex_bg meta info!");
2633            goto failed_mount2;
2634        }
2635
2636    sbi->s_gdb_count = db_count;
2637    get_random_bytes(&sbi->s_next_generation, sizeof(u32));
2638    spin_lock_init(&sbi->s_next_gen_lock);
2639
2640    err = percpu_counter_init(&sbi->s_freeblocks_counter,
2641            ext4_count_free_blocks(sb));
2642    if (!err) {
2643        err = percpu_counter_init(&sbi->s_freeinodes_counter,
2644                ext4_count_free_inodes(sb));
2645    }
2646    if (!err) {
2647        err = percpu_counter_init(&sbi->s_dirs_counter,
2648                ext4_count_dirs(sb));
2649    }
2650    if (!err) {
2651        err = percpu_counter_init(&sbi->s_dirtyblocks_counter, 0);
2652    }
2653    if (err) {
2654        ext4_msg(sb, KERN_ERR, "insufficient memory");
2655        goto failed_mount3;
2656    }
2657
2658    sbi->s_stripe = ext4_get_stripe_size(sbi);
2659
2660    /*
2661     * set up enough so that it can read an inode
2662     */
2663    if (!test_opt(sb, NOLOAD) &&
2664        EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL))
2665        sb->s_op = &ext4_sops;
2666    else
2667        sb->s_op = &ext4_nojournal_sops;
2668    sb->s_export_op = &ext4_export_ops;
2669    sb->s_xattr = ext4_xattr_handlers;
2670#ifdef CONFIG_QUOTA
2671    sb->s_qcop = &ext4_qctl_operations;
2672    sb->dq_op = &ext4_quota_operations;
2673#endif
2674    INIT_LIST_HEAD(&sbi->s_orphan); /* unlinked but open files */
2675    mutex_init(&sbi->s_orphan_lock);
2676    mutex_init(&sbi->s_resize_lock);
2677
2678    sb->s_root = NULL;
2679
2680    needs_recovery = (es->s_last_orphan != 0 ||
2681              EXT4_HAS_INCOMPAT_FEATURE(sb,
2682                    EXT4_FEATURE_INCOMPAT_RECOVER));
2683
2684    /*
2685     * The first inode we look at is the journal inode. Don't try
2686     * root first: it may be modified in the journal!
2687     */
2688    if (!test_opt(sb, NOLOAD) &&
2689        EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL)) {
2690        if (ext4_load_journal(sb, es, journal_devnum))
2691            goto failed_mount3;
2692        if (!(sb->s_flags & MS_RDONLY) &&
2693            EXT4_SB(sb)->s_journal->j_failed_commit) {
2694            ext4_msg(sb, KERN_CRIT, "error: "
2695                   "ext4_fill_super: Journal transaction "
2696                   "%u is corrupt",
2697                   EXT4_SB(sb)->s_journal->j_failed_commit);
2698            if (test_opt(sb, ERRORS_RO)) {
2699                ext4_msg(sb, KERN_CRIT,
2700                       "Mounting filesystem read-only");
2701                sb->s_flags |= MS_RDONLY;
2702                EXT4_SB(sb)->s_mount_state |= EXT4_ERROR_FS;
2703                es->s_state |= cpu_to_le16(EXT4_ERROR_FS);
2704            }
2705            if (test_opt(sb, ERRORS_PANIC)) {
2706                EXT4_SB(sb)->s_mount_state |= EXT4_ERROR_FS;
2707                es->s_state |= cpu_to_le16(EXT4_ERROR_FS);
2708                ext4_commit_super(sb, 1);
2709                goto failed_mount4;
2710            }
2711        }
2712    } else if (test_opt(sb, NOLOAD) && !(sb->s_flags & MS_RDONLY) &&
2713          EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER)) {
2714        ext4_msg(sb, KERN_ERR, "required journal recovery "
2715               "suppressed and not mounted read-only");
2716        goto failed_mount4;
2717    } else {
2718        clear_opt(sbi->s_mount_opt, DATA_FLAGS);
2719        set_opt(sbi->s_mount_opt, WRITEBACK_DATA);
2720        sbi->s_journal = NULL;
2721        needs_recovery = 0;
2722        goto no_journal;
2723    }
2724
2725    if (ext4_blocks_count(es) > 0xffffffffULL &&
2726        !jbd2_journal_set_features(EXT4_SB(sb)->s_journal, 0, 0,
2727                       JBD2_FEATURE_INCOMPAT_64BIT)) {
2728        ext4_msg(sb, KERN_ERR, "Failed to set 64-bit journal feature");
2729        goto failed_mount4;
2730    }
2731
2732    if (test_opt(sb, JOURNAL_ASYNC_COMMIT)) {
2733        jbd2_journal_set_features(sbi->s_journal,
2734                JBD2_FEATURE_COMPAT_CHECKSUM, 0,
2735                JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT);
2736    } else if (test_opt(sb, JOURNAL_CHECKSUM)) {
2737        jbd2_journal_set_features(sbi->s_journal,
2738                JBD2_FEATURE_COMPAT_CHECKSUM, 0, 0);
2739        jbd2_journal_clear_features(sbi->s_journal, 0, 0,
2740                JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT);
2741    } else {
2742        jbd2_journal_clear_features(sbi->s_journal,
2743                JBD2_FEATURE_COMPAT_CHECKSUM, 0,
2744                JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT);
2745    }
2746
2747    /* We have now updated the journal if required, so we can
2748     * validate the data journaling mode. */
2749    switch (test_opt(sb, DATA_FLAGS)) {
2750    case 0:
2751        /* No mode set, assume a default based on the journal
2752         * capabilities: ORDERED_DATA if the journal can
2753         * cope, else JOURNAL_DATA
2754         */
2755        if (jbd2_journal_check_available_features
2756            (sbi->s_journal, 0, 0, JBD2_FEATURE_INCOMPAT_REVOKE))
2757            set_opt(sbi->s_mount_opt, ORDERED_DATA);
2758        else
2759            set_opt(sbi->s_mount_opt, JOURNAL_DATA);
2760        break;
2761
2762    case EXT4_MOUNT_ORDERED_DATA:
2763    case EXT4_MOUNT_WRITEBACK_DATA:
2764        if (!jbd2_journal_check_available_features
2765            (sbi->s_journal, 0, 0, JBD2_FEATURE_INCOMPAT_REVOKE)) {
2766            ext4_msg(sb, KERN_ERR, "Journal does not support "
2767                   "requested data journaling mode");
2768            goto failed_mount4;
2769        }
2770    default:
2771        break;
2772    }
2773    set_task_ioprio(sbi->s_journal->j_task, journal_ioprio);
2774
2775no_journal:
2776
2777    if (test_opt(sb, NOBH)) {
2778        if (!(test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_WRITEBACK_DATA)) {
2779            ext4_msg(sb, KERN_WARNING, "Ignoring nobh option - "
2780                "its supported only with writeback mode");
2781            clear_opt(sbi->s_mount_opt, NOBH);
2782        }
2783    }
2784    /*
2785     * The jbd2_journal_load will have done any necessary log recovery,
2786     * so we can safely mount the rest of the filesystem now.
2787     */
2788
2789    root = ext4_iget(sb, EXT4_ROOT_INO);
2790    if (IS_ERR(root)) {
2791        ext4_msg(sb, KERN_ERR, "get root inode failed");
2792        ret = PTR_ERR(root);
2793        goto failed_mount4;
2794    }
2795    if (!S_ISDIR(root->i_mode) || !root->i_blocks || !root->i_size) {
2796        iput(root);
2797        ext4_msg(sb, KERN_ERR, "corrupt root inode, run e2fsck");
2798        goto failed_mount4;
2799    }
2800    sb->s_root = d_alloc_root(root);
2801    if (!sb->s_root) {
2802        ext4_msg(sb, KERN_ERR, "get root dentry failed");
2803        iput(root);
2804        ret = -ENOMEM;
2805        goto failed_mount4;
2806    }
2807
2808    ext4_setup_super(sb, es, sb->s_flags & MS_RDONLY);
2809
2810    /* determine the minimum size of new large inodes, if present */
2811    if (sbi->s_inode_size > EXT4_GOOD_OLD_INODE_SIZE) {
2812        sbi->s_want_extra_isize = sizeof(struct ext4_inode) -
2813                             EXT4_GOOD_OLD_INODE_SIZE;
2814        if (EXT4_HAS_RO_COMPAT_FEATURE(sb,
2815                       EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE)) {
2816            if (sbi->s_want_extra_isize <
2817                le16_to_cpu(es->s_want_extra_isize))
2818                sbi->s_want_extra_isize =
2819                    le16_to_cpu(es->s_want_extra_isize);
2820            if (sbi->s_want_extra_isize <
2821                le16_to_cpu(es->s_min_extra_isize))
2822                sbi->s_want_extra_isize =
2823                    le16_to_cpu(es->s_min_extra_isize);
2824        }
2825    }
2826    /* Check if enough inode space is available */
2827    if (EXT4_GOOD_OLD_INODE_SIZE + sbi->s_want_extra_isize >
2828                            sbi->s_inode_size) {
2829        sbi->s_want_extra_isize = sizeof(struct ext4_inode) -
2830                               EXT4_GOOD_OLD_INODE_SIZE;
2831        ext4_msg(sb, KERN_INFO, "required extra inode space not"
2832             "available");
2833    }
2834
2835    if (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_JOURNAL_DATA) {
2836        ext4_msg(sb, KERN_WARNING, "Ignoring delalloc option - "
2837             "requested data journaling mode");
2838        clear_opt(sbi->s_mount_opt, DELALLOC);
2839    } else if (test_opt(sb, DELALLOC))
2840        ext4_msg(sb, KERN_INFO, "delayed allocation enabled");
2841
2842    err = ext4_setup_system_zone(sb);
2843    if (err) {
2844        ext4_msg(sb, KERN_ERR, "failed to initialize system "
2845             "zone (%d)\n", err);
2846        goto failed_mount4;
2847    }
2848
2849    ext4_ext_init(sb);
2850    err = ext4_mb_init(sb, needs_recovery);
2851    if (err) {
2852        ext4_msg(sb, KERN_ERR, "failed to initalize mballoc (%d)",
2853             err);
2854        goto failed_mount4;
2855    }
2856
2857    sbi->s_kobj.kset = ext4_kset;
2858    init_completion(&sbi->s_kobj_unregister);
2859    err = kobject_init_and_add(&sbi->s_kobj, &ext4_ktype, NULL,
2860                   "%s", sb->s_id);
2861    if (err) {
2862        ext4_mb_release(sb);
2863        ext4_ext_release(sb);
2864        goto failed_mount4;
2865    };
2866
2867    EXT4_SB(sb)->s_mount_state |= EXT4_ORPHAN_FS;
2868    ext4_orphan_cleanup(sb, es);
2869    EXT4_SB(sb)->s_mount_state &= ~EXT4_ORPHAN_FS;
2870    if (needs_recovery) {
2871        ext4_msg(sb, KERN_INFO, "recovery complete");
2872        ext4_mark_recovery_complete(sb, es);
2873    }
2874    if (EXT4_SB(sb)->s_journal) {
2875        if (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_JOURNAL_DATA)
2876            descr = " journalled data mode";
2877        else if (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_ORDERED_DATA)
2878            descr = " ordered data mode";
2879        else
2880            descr = " writeback data mode";
2881    } else
2882        descr = "out journal";
2883
2884    ext4_msg(sb, KERN_INFO, "mounted filesystem with%s", descr);
2885
2886    lock_kernel();
2887    return 0;
2888
2889cantfind_ext4:
2890    if (!silent)
2891        ext4_msg(sb, KERN_ERR, "VFS: Can't find ext4 filesystem");
2892    goto failed_mount;
2893
2894failed_mount4:
2895    ext4_msg(sb, KERN_ERR, "mount failed");
2896    ext4_release_system_zone(sb);
2897    if (sbi->s_journal) {
2898        jbd2_journal_destroy(sbi->s_journal);
2899        sbi->s_journal = NULL;
2900    }
2901failed_mount3:
2902    if (sbi->s_flex_groups) {
2903        if (is_vmalloc_addr(sbi->s_flex_groups))
2904            vfree(sbi->s_flex_groups);
2905        else
2906            kfree(sbi->s_flex_groups);
2907    }
2908    percpu_counter_destroy(&sbi->s_freeblocks_counter);
2909    percpu_counter_destroy(&sbi->s_freeinodes_counter);
2910    percpu_counter_destroy(&sbi->s_dirs_counter);
2911    percpu_counter_destroy(&sbi->s_dirtyblocks_counter);
2912failed_mount2:
2913    for (i = 0; i < db_count; i++)
2914        brelse(sbi->s_group_desc[i]);
2915    kfree(sbi->s_group_desc);
2916failed_mount:
2917    if (sbi->s_proc) {
2918        remove_proc_entry(sb->s_id, ext4_proc_root);
2919    }
2920#ifdef CONFIG_QUOTA
2921    for (i = 0; i < MAXQUOTAS; i++)
2922        kfree(sbi->s_qf_names[i]);
2923#endif
2924    ext4_blkdev_remove(sbi);
2925    brelse(bh);
2926out_fail:
2927    sb->s_fs_info = NULL;
2928    kfree(sbi->s_blockgroup_lock);
2929    kfree(sbi);
2930    lock_kernel();
2931    return ret;
2932}
2933
2934/*
2935 * Setup any per-fs journal parameters now. We'll do this both on
2936 * initial mount, once the journal has been initialised but before we've
2937 * done any recovery; and again on any subsequent remount.
2938 */
2939static void ext4_init_journal_params(struct super_block *sb, journal_t *journal)
2940{
2941    struct ext4_sb_info *sbi = EXT4_SB(sb);
2942
2943    journal->j_commit_interval = sbi->s_commit_interval;
2944    journal->j_min_batch_time = sbi->s_min_batch_time;
2945    journal->j_max_batch_time = sbi->s_max_batch_time;
2946
2947    spin_lock(&journal->j_state_lock);
2948    if (test_opt(sb, BARRIER))
2949        journal->j_flags |= JBD2_BARRIER;
2950    else
2951        journal->j_flags &= ~JBD2_BARRIER;
2952    if (test_opt(sb, DATA_ERR_ABORT))
2953        journal->j_flags |= JBD2_ABORT_ON_SYNCDATA_ERR;
2954    else
2955        journal->j_flags &= ~JBD2_ABORT_ON_SYNCDATA_ERR;
2956    spin_unlock(&journal->j_state_lock);
2957}
2958
2959static journal_t *ext4_get_journal(struct super_block *sb,
2960                   unsigned int journal_inum)
2961{
2962    struct inode *journal_inode;
2963    journal_t *journal;
2964
2965    BUG_ON(!EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL));
2966
2967    /* First, test for the existence of a valid inode on disk. Bad
2968     * things happen if we iget() an unused inode, as the subsequent
2969     * iput() will try to delete it. */
2970
2971    journal_inode = ext4_iget(sb, journal_inum);
2972    if (IS_ERR(journal_inode)) {
2973        ext4_msg(sb, KERN_ERR, "no journal found");
2974        return NULL;
2975    }
2976    if (!journal_inode->i_nlink) {
2977        make_bad_inode(journal_inode);
2978        iput(journal_inode);
2979        ext4_msg(sb, KERN_ERR, "journal inode is deleted");
2980        return NULL;
2981    }
2982
2983    jbd_debug(2, "Journal inode found at %p: %lld bytes\n",
2984          journal_inode, journal_inode->i_size);
2985    if (!S_ISREG(journal_inode->i_mode)) {
2986        ext4_msg(sb, KERN_ERR, "invalid journal inode");
2987        iput(journal_inode);
2988        return NULL;
2989    }
2990
2991    journal = jbd2_journal_init_inode(journal_inode);
2992    if (!journal) {
2993        ext4_msg(sb, KERN_ERR, "Could not load journal inode");
2994        iput(journal_inode);
2995        return NULL;
2996    }
2997    journal->j_private = sb;
2998    ext4_init_journal_params(sb, journal);
2999    return journal;
3000}
3001
3002static journal_t *ext4_get_dev_journal(struct super_block *sb,
3003                       dev_t j_dev)
3004{
3005    struct buffer_head *bh;
3006    journal_t *journal;
3007    ext4_fsblk_t start;
3008    ext4_fsblk_t len;
3009    int hblock, blocksize;
3010    ext4_fsblk_t sb_block;
3011    unsigned long offset;
3012    struct ext4_super_block *es;
3013    struct block_device *bdev;
3014
3015    BUG_ON(!EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL));
3016
3017    bdev = ext4_blkdev_get(j_dev, sb);
3018    if (bdev == NULL)
3019        return NULL;
3020
3021    if (bd_claim(bdev, sb)) {
3022        ext4_msg(sb, KERN_ERR,
3023            "failed to claim external journal device");
3024        blkdev_put(bdev, FMODE_READ|FMODE_WRITE);
3025        return NULL;
3026    }
3027
3028    blocksize = sb->s_blocksize;
3029    hblock = bdev_logical_block_size(bdev);
3030    if (blocksize < hblock) {
3031        ext4_msg(sb, KERN_ERR,
3032            "blocksize too small for journal device");
3033        goto out_bdev;
3034    }
3035
3036    sb_block = EXT4_MIN_BLOCK_SIZE / blocksize;
3037    offset = EXT4_MIN_BLOCK_SIZE % blocksize;
3038    set_blocksize(bdev, blocksize);
3039    if (!(bh = __bread(bdev, sb_block, blocksize))) {
3040        ext4_msg(sb, KERN_ERR, "couldn't read superblock of "
3041               "external journal");
3042        goto out_bdev;
3043    }
3044
3045    es = (struct ext4_super_block *) (((char *)bh->b_data) + offset);
3046    if ((le16_to_cpu(es->s_magic) != EXT4_SUPER_MAGIC) ||
3047        !(le32_to_cpu(es->s_feature_incompat) &
3048          EXT4_FEATURE_INCOMPAT_JOURNAL_DEV)) {
3049        ext4_msg(sb, KERN_ERR, "external journal has "
3050                    "bad superblock");
3051        brelse(bh);
3052        goto out_bdev;
3053    }
3054
3055    if (memcmp(EXT4_SB(sb)->s_es->s_journal_uuid, es->s_uuid, 16)) {
3056        ext4_msg(sb, KERN_ERR, "journal UUID does not match");
3057        brelse(bh);
3058        goto out_bdev;
3059    }
3060
3061    len = ext4_blocks_count(es);
3062    start = sb_block + 1;
3063    brelse(bh); /* we're done with the superblock */
3064
3065    journal = jbd2_journal_init_dev(bdev, sb->s_bdev,
3066                    start, len, blocksize);
3067    if (!journal) {
3068        ext4_msg(sb, KERN_ERR, "failed to create device journal");
3069        goto out_bdev;
3070    }
3071    journal->j_private = sb;
3072    ll_rw_block(READ, 1, &journal->j_sb_buffer);
3073    wait_on_buffer(journal->j_sb_buffer);
3074    if (!buffer_uptodate(journal->j_sb_buffer)) {
3075        ext4_msg(sb, KERN_ERR, "I/O error on journal device");
3076        goto out_journal;
3077    }
3078    if (be32_to_cpu(journal->j_superblock->s_nr_users) != 1) {
3079        ext4_msg(sb, KERN_ERR, "External journal has more than one "
3080                    "user (unsupported) - %d",
3081            be32_to_cpu(journal->j_superblock->s_nr_users));
3082        goto out_journal;
3083    }
3084    EXT4_SB(sb)->journal_bdev = bdev;
3085    ext4_init_journal_params(sb, journal);
3086    return journal;
3087
3088out_journal:
3089    jbd2_journal_destroy(journal);
3090out_bdev:
3091    ext4_blkdev_put(bdev);
3092    return NULL;
3093}
3094
3095static int ext4_load_journal(struct super_block *sb,
3096                 struct ext4_super_block *es,
3097                 unsigned long journal_devnum)
3098{
3099    journal_t *journal;
3100    unsigned int journal_inum = le32_to_cpu(es->s_journal_inum);
3101    dev_t journal_dev;
3102    int err = 0;
3103    int really_read_only;
3104
3105    BUG_ON(!EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL));
3106
3107    if (journal_devnum &&
3108        journal_devnum != le32_to_cpu(es->s_journal_dev)) {
3109        ext4_msg(sb, KERN_INFO, "external journal device major/minor "
3110            "numbers have changed");
3111        journal_dev = new_decode_dev(journal_devnum);
3112    } else
3113        journal_dev = new_decode_dev(le32_to_cpu(es->s_journal_dev));
3114
3115    really_read_only = bdev_read_only(sb->s_bdev);
3116
3117    /*
3118     * Are we loading a blank journal or performing recovery after a
3119     * crash? For recovery, we need to check in advance whether we
3120     * can get read-write access to the device.
3121     */
3122    if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER)) {
3123        if (sb->s_flags & MS_RDONLY) {
3124            ext4_msg(sb, KERN_INFO, "INFO: recovery "
3125                    "required on readonly filesystem");
3126            if (really_read_only) {
3127                ext4_msg(sb, KERN_ERR, "write access "
3128                    "unavailable, cannot proceed");
3129                return -EROFS;
3130            }
3131            ext4_msg(sb, KERN_INFO, "write access will "
3132                   "be enabled during recovery");
3133        }
3134    }
3135
3136    if (journal_inum && journal_dev) {
3137        ext4_msg(sb, KERN_ERR, "filesystem has both journal "
3138               "and inode journals!");
3139        return -EINVAL;
3140    }
3141
3142    if (journal_inum) {
3143        if (!(journal = ext4_get_journal(sb, journal_inum)))
3144            return -EINVAL;
3145    } else {
3146        if (!(journal = ext4_get_dev_journal(sb, journal_dev)))
3147            return -EINVAL;
3148    }
3149
3150    if (journal->j_flags & JBD2_BARRIER)
3151        ext4_msg(sb, KERN_INFO, "barriers enabled");
3152    else
3153        ext4_msg(sb, KERN_INFO, "barriers disabled");
3154
3155    if (!really_read_only && test_opt(sb, UPDATE_JOURNAL)) {
3156        err = jbd2_journal_update_format(journal);
3157        if (err) {
3158            ext4_msg(sb, KERN_ERR, "error updating journal");
3159            jbd2_journal_destroy(journal);
3160            return err;
3161        }
3162    }
3163
3164    if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER))
3165        err = jbd2_journal_wipe(journal, !really_read_only);
3166    if (!err)
3167        err = jbd2_journal_load(journal);
3168
3169    if (err) {
3170        ext4_msg(sb, KERN_ERR, "error loading journal");
3171        jbd2_journal_destroy(journal);
3172        return err;
3173    }
3174
3175    EXT4_SB(sb)->s_journal = journal;
3176    ext4_clear_journal_err(sb, es);
3177
3178    if (journal_devnum &&
3179        journal_devnum != le32_to_cpu(es->s_journal_dev)) {
3180        es->s_journal_dev = cpu_to_le32(journal_devnum);
3181
3182        /* Make sure we flush the recovery flag to disk. */
3183        ext4_commit_super(sb, 1);
3184    }
3185
3186    return 0;
3187}
3188
3189static int ext4_commit_super(struct super_block *sb, int sync)
3190{
3191    struct ext4_super_block *es = EXT4_SB(sb)->s_es;
3192    struct buffer_head *sbh = EXT4_SB(sb)->s_sbh;
3193    int error = 0;
3194
3195    if (!sbh)
3196        return error;
3197    if (buffer_write_io_error(sbh)) {
3198        /*
3199         * Oh, dear. A previous attempt to write the
3200         * superblock failed. This could happen because the
3201         * USB device was yanked out. Or it could happen to
3202         * be a transient write error and maybe the block will
3203         * be remapped. Nothing we can do but to retry the
3204         * write and hope for the best.
3205         */
3206        ext4_msg(sb, KERN_ERR, "previous I/O error to "
3207               "superblock detected");
3208        clear_buffer_write_io_error(sbh);
3209        set_buffer_uptodate(sbh);
3210    }
3211    es->s_wtime = cpu_to_le32(get_seconds());
3212    es->s_kbytes_written =
3213        cpu_to_le64(EXT4_SB(sb)->s_kbytes_written +
3214                ((part_stat_read(sb->s_bdev->bd_part, sectors[1]) -
3215                  EXT4_SB(sb)->s_sectors_written_start) >> 1));
3216    ext4_free_blocks_count_set(es, percpu_counter_sum_positive(
3217                    &EXT4_SB(sb)->s_freeblocks_counter));
3218    es->s_free_inodes_count = cpu_to_le32(percpu_counter_sum_positive(
3219                    &EXT4_SB(sb)->s_freeinodes_counter));
3220    sb->s_dirt = 0;
3221    BUFFER_TRACE(sbh, "marking dirty");
3222    mark_buffer_dirty(sbh);
3223    if (sync) {
3224        error = sync_dirty_buffer(sbh);
3225        if (error)
3226            return error;
3227
3228        error = buffer_write_io_error(sbh);
3229        if (error) {
3230            ext4_msg(sb, KERN_ERR, "I/O error while writing "
3231                   "superblock");
3232            clear_buffer_write_io_error(sbh);
3233            set_buffer_uptodate(sbh);
3234        }
3235    }
3236    return error;
3237}
3238
3239/*
3240 * Have we just finished recovery? If so, and if we are mounting (or
3241 * remounting) the filesystem readonly, then we will end up with a
3242 * consistent fs on disk. Record that fact.
3243 */
3244static void ext4_mark_recovery_complete(struct super_block *sb,
3245                    struct ext4_super_block *es)
3246{
3247    journal_t *journal = EXT4_SB(sb)->s_journal;
3248
3249    if (!EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL)) {
3250        BUG_ON(journal != NULL);
3251        return;
3252    }
3253    jbd2_journal_lock_updates(journal);
3254    if (jbd2_journal_flush(journal) < 0)
3255        goto out;
3256
3257    if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER) &&
3258        sb->s_flags & MS_RDONLY) {
3259        EXT4_CLEAR_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER);
3260        ext4_commit_super(sb, 1);
3261    }
3262
3263out:
3264    jbd2_journal_unlock_updates(journal);
3265}
3266
3267/*
3268 * If we are mounting (or read-write remounting) a filesystem whose journal
3269 * has recorded an error from a previous lifetime, move that error to the
3270 * main filesystem now.
3271 */
3272static void ext4_clear_journal_err(struct super_block *sb,
3273                   struct ext4_super_block *es)
3274{
3275    journal_t *journal;
3276    int j_errno;
3277    const char *errstr;
3278
3279    BUG_ON(!EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL));
3280
3281    journal = EXT4_SB(sb)->s_journal;
3282
3283    /*
3284     * Now check for any error status which may have been recorded in the
3285     * journal by a prior ext4_error() or ext4_abort()
3286     */
3287
3288    j_errno = jbd2_journal_errno(journal);
3289    if (j_errno) {
3290        char nbuf[16];
3291
3292        errstr = ext4_decode_error(sb, j_errno, nbuf);
3293        ext4_warning(sb, __func__, "Filesystem error recorded "
3294                 "from previous mount: %s", errstr);
3295        ext4_warning(sb, __func__, "Marking fs in need of "
3296                 "filesystem check.");
3297
3298        EXT4_SB(sb)->s_mount_state |= EXT4_ERROR_FS;
3299        es->s_state |= cpu_to_le16(EXT4_ERROR_FS);
3300        ext4_commit_super(sb, 1);
3301
3302        jbd2_journal_clear_err(journal);
3303    }
3304}
3305
3306/*
3307 * Force the running and committing transactions to commit,
3308 * and wait on the commit.
3309 */
3310int ext4_force_commit(struct super_block *sb)
3311{
3312    journal_t *journal;
3313    int ret = 0;
3314
3315    if (sb->s_flags & MS_RDONLY)
3316        return 0;
3317
3318    journal = EXT4_SB(sb)->s_journal;
3319    if (journal)
3320        ret = ext4_journal_force_commit(journal);
3321
3322    return ret;
3323}
3324
3325static void ext4_write_super(struct super_block *sb)
3326{
3327    lock_super(sb);
3328    ext4_commit_super(sb, 1);
3329    unlock_super(sb);
3330}
3331
3332static int ext4_sync_fs(struct super_block *sb, int wait)
3333{
3334    int ret = 0;
3335    tid_t target;
3336
3337    trace_ext4_sync_fs(sb, wait);
3338    if (jbd2_journal_start_commit(EXT4_SB(sb)->s_journal, &target)) {
3339        if (wait)
3340            jbd2_log_wait_commit(EXT4_SB(sb)->s_journal, target);
3341    }
3342    return ret;
3343}
3344
3345/*
3346 * LVM calls this function before a (read-only) snapshot is created. This
3347 * gives us a chance to flush the journal completely and mark the fs clean.
3348 */
3349static int ext4_freeze(struct super_block *sb)
3350{
3351    int error = 0;
3352    journal_t *journal;
3353
3354    if (sb->s_flags & MS_RDONLY)
3355        return 0;
3356
3357    journal = EXT4_SB(sb)->s_journal;
3358
3359    /* Now we set up the journal barrier. */
3360    jbd2_journal_lock_updates(journal);
3361
3362    /*
3363     * Don't clear the needs_recovery flag if we failed to flush
3364     * the journal.
3365     */
3366    error = jbd2_journal_flush(journal);
3367    if (error < 0) {
3368    out:
3369        jbd2_journal_unlock_updates(journal);
3370        return error;
3371    }
3372
3373    /* Journal blocked and flushed, clear needs_recovery flag. */
3374    EXT4_CLEAR_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER);
3375    error = ext4_commit_super(sb, 1);
3376    if (error)
3377        goto out;
3378    return 0;
3379}
3380
3381/*
3382 * Called by LVM after the snapshot is done. We need to reset the RECOVER
3383 * flag here, even though the filesystem is not technically dirty yet.
3384 */
3385static int ext4_unfreeze(struct super_block *sb)
3386{
3387    if (sb->s_flags & MS_RDONLY)
3388        return 0;
3389
3390    lock_super(sb);
3391    /* Reset the needs_recovery flag before the fs is unlocked. */
3392    EXT4_SET_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER);
3393    ext4_commit_super(sb, 1);
3394    unlock_super(sb);
3395    jbd2_journal_unlock_updates(EXT4_SB(sb)->s_journal);
3396    return 0;
3397}
3398
3399static int ext4_remount(struct super_block *sb, int *flags, char *data)
3400{
3401    struct ext4_super_block *es;
3402    struct ext4_sb_info *sbi = EXT4_SB(sb);
3403    ext4_fsblk_t n_blocks_count = 0;
3404    unsigned long old_sb_flags;
3405    struct ext4_mount_options old_opts;
3406    ext4_group_t g;
3407    unsigned int journal_ioprio = DEFAULT_JOURNAL_IOPRIO;
3408    int err;
3409#ifdef CONFIG_QUOTA
3410    int i;
3411#endif
3412
3413    lock_kernel();
3414
3415    /* Store the original options */
3416    lock_super(sb);
3417    old_sb_flags = sb->s_flags;
3418    old_opts.s_mount_opt = sbi->s_mount_opt;
3419    old_opts.s_resuid = sbi->s_resuid;
3420    old_opts.s_resgid = sbi->s_resgid;
3421    old_opts.s_commit_interval = sbi->s_commit_interval;
3422    old_opts.s_min_batch_time = sbi->s_min_batch_time;
3423    old_opts.s_max_batch_time = sbi->s_max_batch_time;
3424#ifdef CONFIG_QUOTA
3425    old_opts.s_jquota_fmt = sbi->s_jquota_fmt;
3426    for (i = 0; i < MAXQUOTAS; i++)
3427        old_opts.s_qf_names[i] = sbi->s_qf_names[i];
3428#endif
3429    if (sbi->s_journal && sbi->s_journal->j_task->io_context)
3430        journal_ioprio = sbi->s_journal->j_task->io_context->ioprio;
3431
3432    /*
3433     * Allow the "check" option to be passed as a remount option.
3434     */
3435    if (!parse_options(data, sb, NULL, &journal_ioprio,
3436               &n_blocks_count, 1)) {
3437        err = -EINVAL;
3438        goto restore_opts;
3439    }
3440
3441    if (sbi->s_mount_flags & EXT4_MF_FS_ABORTED)
3442        ext4_abort(sb, __func__, "Abort forced by user");
3443
3444    sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
3445        ((sbi->s_mount_opt & EXT4_MOUNT_POSIX_ACL) ? MS_POSIXACL : 0);
3446
3447    es = sbi->s_es;
3448
3449    if (sbi->s_journal) {
3450        ext4_init_journal_params(sb, sbi->s_journal);
3451        set_task_ioprio(sbi->s_journal->j_task, journal_ioprio);
3452    }
3453
3454    if ((*flags & MS_RDONLY) != (sb->s_flags & MS_RDONLY) ||
3455        n_blocks_count > ext4_blocks_count(es)) {
3456        if (sbi->s_mount_flags & EXT4_MF_FS_ABORTED) {
3457            err = -EROFS;
3458            goto restore_opts;
3459        }
3460
3461        if (*flags & MS_RDONLY) {
3462            /*
3463             * First of all, the unconditional stuff we have to do
3464             * to disable replay of the journal when we next remount
3465             */
3466            sb->s_flags |= MS_RDONLY;
3467
3468            /*
3469             * OK, test if we are remounting a valid rw partition
3470             * readonly, and if so set the rdonly flag and then
3471             * mark the partition as valid again.
3472             */
3473            if (!(es->s_state & cpu_to_le16(EXT4_VALID_FS)) &&
3474                (sbi->s_mount_state & EXT4_VALID_FS))
3475                es->s_state = cpu_to_le16(sbi->s_mount_state);
3476
3477            if (sbi->s_journal)
3478                ext4_mark_recovery_complete(sb, es);
3479        } else {
3480            int ret;
3481            if ((ret = EXT4_HAS_RO_COMPAT_FEATURE(sb,
3482                    ~EXT4_FEATURE_RO_COMPAT_SUPP))) {
3483                ext4_msg(sb, KERN_WARNING, "couldn't "
3484                       "remount RDWR because of unsupported "
3485                       "optional features (%x)",
3486                (le32_to_cpu(sbi->s_es->s_feature_ro_compat) &
3487                    ~EXT4_FEATURE_RO_COMPAT_SUPP));
3488                err = -EROFS;
3489                goto restore_opts;
3490            }
3491
3492            /*
3493             * Make sure the group descriptor checksums
3494             * are sane. If they aren't, refuse to remount r/w.
3495             */
3496            for (g = 0; g < sbi->s_groups_count; g++) {
3497                struct ext4_group_desc *gdp =
3498                    ext4_get_group_desc(sb, g, NULL);
3499
3500                if (!ext4_group_desc_csum_verify(sbi, g, gdp)) {
3501                    ext4_msg(sb, KERN_ERR,
3502           "ext4_remount: Checksum for group %u failed (%u!=%u)",
3503        g, le16_to_cpu(ext4_group_desc_csum(sbi, g, gdp)),
3504                           le16_to_cpu(gdp->bg_checksum));
3505                    err = -EINVAL;
3506                    goto restore_opts;
3507                }
3508            }
3509
3510            /*
3511             * If we have an unprocessed orphan list hanging
3512             * around from a previously readonly bdev mount,
3513             * require a full umount/remount for now.
3514             */
3515            if (es->s_last_orphan) {
3516                ext4_msg(sb, KERN_WARNING, "Couldn't "
3517                       "remount RDWR because of unprocessed "
3518                       "orphan inode list. Please "
3519                       "umount/remount instead");
3520                err = -EINVAL;
3521                goto restore_opts;
3522            }
3523
3524            /*
3525             * Mounting a RDONLY partition read-write, so reread
3526             * and store the current valid flag. (It may have
3527             * been changed by e2fsck since we originally mounted
3528             * the partition.)
3529             */
3530            if (sbi->s_journal)
3531                ext4_clear_journal_err(sb, es);
3532            sbi->s_mount_state = le16_to_cpu(es->s_state);
3533            if ((err = ext4_group_extend(sb, es, n_blocks_count)))
3534                goto restore_opts;
3535            if (!ext4_setup_super(sb, es, 0))
3536                sb->s_flags &= ~MS_RDONLY;
3537        }
3538    }
3539    ext4_setup_system_zone(sb);
3540    if (sbi->s_journal == NULL)
3541        ext4_commit_super(sb, 1);
3542
3543#ifdef CONFIG_QUOTA
3544    /* Release old quota file names */
3545    for (i = 0; i < MAXQUOTAS; i++)
3546        if (old_opts.s_qf_names[i] &&
3547            old_opts.s_qf_names[i] != sbi->s_qf_names[i])
3548            kfree(old_opts.s_qf_names[i]);
3549#endif
3550    unlock_super(sb);
3551    unlock_kernel();
3552    return 0;
3553
3554restore_opts:
3555    sb->s_flags = old_sb_flags;
3556    sbi->s_mount_opt = old_opts.s_mount_opt;
3557    sbi->s_resuid = old_opts.s_resuid;
3558    sbi->s_resgid = old_opts.s_resgid;
3559    sbi->s_commit_interval = old_opts.s_commit_interval;
3560    sbi->s_min_batch_time = old_opts.s_min_batch_time;
3561    sbi->s_max_batch_time = old_opts.s_max_batch_time;
3562#ifdef CONFIG_QUOTA
3563    sbi->s_jquota_fmt = old_opts.s_jquota_fmt;
3564    for (i = 0; i < MAXQUOTAS; i++) {
3565        if (sbi->s_qf_names[i] &&
3566            old_opts.s_qf_names[i] != sbi->s_qf_names[i])
3567            kfree(sbi->s_qf_names[i]);
3568        sbi->s_qf_names[i] = old_opts.s_qf_names[i];
3569    }
3570#endif
3571    unlock_super(sb);
3572    unlock_kernel();
3573    return err;
3574}
3575
3576static int ext4_statfs(struct dentry *dentry, struct kstatfs *buf)
3577{
3578    struct super_block *sb = dentry->d_sb;
3579    struct ext4_sb_info *sbi = EXT4_SB(sb);
3580    struct ext4_super_block *es = sbi->s_es;
3581    u64 fsid;
3582
3583    if (test_opt(sb, MINIX_DF)) {
3584        sbi->s_overhead_last = 0;
3585    } else if (sbi->s_blocks_last != ext4_blocks_count(es)) {
3586        ext4_group_t i, ngroups = ext4_get_groups_count(sb);
3587        ext4_fsblk_t overhead = 0;
3588
3589        /*
3590         * Compute the overhead (FS structures). This is constant
3591         * for a given filesystem unless the number of block groups
3592         * changes so we cache the previous value until it does.
3593         */
3594
3595        /*
3596         * All of the blocks before first_data_block are
3597         * overhead
3598         */
3599        overhead = le32_to_cpu(es->s_first_data_block);
3600
3601        /*
3602         * Add the overhead attributed to the superblock and
3603         * block group descriptors. If the sparse superblocks
3604         * feature is turned on, then not all groups have this.
3605         */
3606        for (i = 0; i < ngroups; i++) {
3607            overhead += ext4_bg_has_super(sb, i) +
3608                ext4_bg_num_gdb(sb, i);
3609            cond_resched();
3610        }
3611
3612        /*
3613         * Every block group has an inode bitmap, a block
3614         * bitmap, and an inode table.
3615         */
3616        overhead += ngroups * (2 + sbi->s_itb_per_group);
3617        sbi->s_overhead_last = overhead;
3618        smp_wmb();
3619        sbi->s_blocks_last = ext4_blocks_count(es);
3620    }
3621
3622    buf->f_type = EXT4_SUPER_MAGIC;
3623    buf->f_bsize = sb->s_blocksize;
3624    buf->f_blocks = ext4_blocks_count(es) - sbi->s_overhead_last;
3625    buf->f_bfree = percpu_counter_sum_positive(&sbi->s_freeblocks_counter) -
3626               percpu_counter_sum_positive(&sbi->s_dirtyblocks_counter);
3627    ext4_free_blocks_count_set(es, buf->f_bfree);
3628    buf->f_bavail = buf->f_bfree - ext4_r_blocks_count(es);
3629    if (buf->f_bfree < ext4_r_blocks_count(es))
3630        buf->f_bavail = 0;
3631    buf->f_files = le32_to_cpu(es->s_inodes_count);
3632    buf->f_ffree = percpu_counter_sum_positive(&sbi->s_freeinodes_counter);
3633    es->s_free_inodes_count = cpu_to_le32(buf->f_ffree);
3634    buf->f_namelen = EXT4_NAME_LEN;
3635    fsid = le64_to_cpup((void *)es->s_uuid) ^
3636           le64_to_cpup((void *)es->s_uuid + sizeof(u64));
3637    buf->f_fsid.val[0] = fsid & 0xFFFFFFFFUL;
3638    buf->f_fsid.val[1] = (fsid >> 32) & 0xFFFFFFFFUL;
3639
3640    return 0;
3641}
3642
3643/* Helper function for writing quotas on sync - we need to start transaction
3644 * before quota file is locked for write. Otherwise the are possible deadlocks:
3645 * Process 1 Process 2
3646 * ext4_create() quota_sync()
3647 * jbd2_journal_start() write_dquot()
3648 * vfs_dq_init() down(dqio_mutex)
3649 * down(dqio_mutex) jbd2_journal_start()
3650 *
3651 */
3652
3653#ifdef CONFIG_QUOTA
3654
3655static inline struct inode *dquot_to_inode(struct dquot *dquot)
3656{
3657    return sb_dqopt(dquot->dq_sb)->files[dquot->dq_type];
3658}
3659
3660static int ext4_write_dquot(struct dquot *dquot)
3661{
3662    int ret, err;
3663    handle_t *handle;
3664    struct inode *inode;
3665
3666    inode = dquot_to_inode(dquot);
3667    handle = ext4_journal_start(inode,
3668                    EXT4_QUOTA_TRANS_BLOCKS(dquot->dq_sb));
3669    if (IS_ERR(handle))
3670        return PTR_ERR(handle);
3671    ret = dquot_commit(dquot);
3672    err = ext4_journal_stop(handle);
3673    if (!ret)
3674        ret = err;
3675    return ret;
3676}
3677
3678static int ext4_acquire_dquot(struct dquot *dquot)
3679{
3680    int ret, err;
3681    handle_t *handle;
3682
3683    handle = ext4_journal_start(dquot_to_inode(dquot),
3684                    EXT4_QUOTA_INIT_BLOCKS(dquot->dq_sb));
3685    if (IS_ERR(handle))
3686        return PTR_ERR(handle);
3687    ret = dquot_acquire(dquot);
3688    err = ext4_journal_stop(handle);
3689    if (!ret)
3690        ret = err;
3691    return ret;
3692}
3693
3694static int ext4_release_dquot(struct dquot *dquot)
3695{
3696    int ret, err;
3697    handle_t *handle;
3698
3699    handle = ext4_journal_start(dquot_to_inode(dquot),
3700                    EXT4_QUOTA_DEL_BLOCKS(dquot->dq_sb));
3701    if (IS_ERR(handle)) {
3702        /* Release dquot anyway to avoid endless cycle in dqput() */
3703        dquot_release(dquot);
3704        return PTR_ERR(handle);
3705    }
3706    ret = dquot_release(dquot);
3707    err = ext4_journal_stop(handle);
3708    if (!ret)
3709        ret = err;
3710    return ret;
3711}
3712
3713static int ext4_mark_dquot_dirty(struct dquot *dquot)
3714{
3715    /* Are we journaling quotas? */
3716    if (EXT4_SB(dquot->dq_sb)->s_qf_names[USRQUOTA] ||
3717        EXT4_SB(dquot->dq_sb)->s_qf_names[GRPQUOTA]) {
3718        dquot_mark_dquot_dirty(dquot);
3719        return ext4_write_dquot(dquot);
3720    } else {
3721        return dquot_mark_dquot_dirty(dquot);
3722    }
3723}
3724
3725static int ext4_write_info(struct super_block *sb, int type)
3726{
3727    int ret, err;
3728    handle_t *handle;
3729
3730    /* Data block + inode block */
3731    handle = ext4_journal_start(sb->s_root->d_inode, 2);
3732    if (IS_ERR(handle))
3733        return PTR_ERR(handle);
3734    ret = dquot_commit_info(sb, type);
3735    err = ext4_journal_stop(handle);
3736    if (!ret)
3737        ret = err;
3738    return ret;
3739}
3740
3741/*
3742 * Turn on quotas during mount time - we need to find
3743 * the quota file and such...
3744 */
3745static int ext4_quota_on_mount(struct super_block *sb, int type)
3746{
3747    return vfs_quota_on_mount(sb, EXT4_SB(sb)->s_qf_names[type],
3748                  EXT4_SB(sb)->s_jquota_fmt, type);
3749}
3750
3751/*
3752 * Standard function to be called on quota_on
3753 */
3754static int ext4_quota_on(struct super_block *sb, int type, int format_id,
3755             char *name, int remount)
3756{
3757    int err;
3758    struct path path;
3759
3760    if (!test_opt(sb, QUOTA))
3761        return -EINVAL;
3762    /* When remounting, no checks are needed and in fact, name is NULL */
3763    if (remount)
3764        return vfs_quota_on(sb, type, format_id, name, remount);
3765
3766    err = kern_path(name, LOOKUP_FOLLOW, &path);
3767    if (err)
3768        return err;
3769
3770    /* Quotafile not on the same filesystem? */
3771    if (path.mnt->mnt_sb != sb) {
3772        path_put(&path);
3773        return -EXDEV;
3774    }
3775    /* Journaling quota? */
3776    if (EXT4_SB(sb)->s_qf_names[type]) {
3777        /* Quotafile not in fs root? */
3778        if (path.dentry->d_parent != sb->s_root)
3779            ext4_msg(sb, KERN_WARNING,
3780                "Quota file not on filesystem root. "
3781                "Journaled quota will not work");
3782    }
3783
3784    /*
3785     * When we journal data on quota file, we have to flush journal to see
3786     * all updates to the file when we bypass pagecache...
3787     */
3788    if (EXT4_SB(sb)->s_journal &&
3789        ext4_should_journal_data(path.dentry->d_inode)) {
3790        /*
3791         * We don't need to lock updates but journal_flush() could
3792         * otherwise be livelocked...
3793         */
3794        jbd2_journal_lock_updates(EXT4_SB(sb)->s_journal);
3795        err = jbd2_journal_flush(EXT4_SB(sb)->s_journal);
3796        jbd2_journal_unlock_updates(EXT4_SB(sb)->s_journal);
3797        if (err) {
3798            path_put(&path);
3799            return err;
3800        }
3801    }
3802
3803    err = vfs_quota_on_path(sb, type, format_id, &path);
3804    path_put(&path);
3805    return err;
3806}
3807
3808/* Read data from quotafile - avoid pagecache and such because we cannot afford
3809 * acquiring the locks... As quota files are never truncated and quota code
3810 * itself serializes the operations (and noone else should touch the files)
3811 * we don't have to be afraid of races */
3812static ssize_t ext4_quota_read(struct super_block *sb, int type, char *data,
3813                   size_t len, loff_t off)
3814{
3815    struct inode *inode = sb_dqopt(sb)->files[type];
3816    ext4_lblk_t blk = off >> EXT4_BLOCK_SIZE_BITS(sb);
3817    int err = 0;
3818    int offset = off & (sb->s_blocksize - 1);
3819    int tocopy;
3820    size_t toread;
3821    struct buffer_head *bh;
3822    loff_t i_size = i_size_read(inode);
3823
3824    if (off > i_size)
3825        return 0;
3826    if (off+len > i_size)
3827        len = i_size-off;
3828    toread = len;
3829    while (toread > 0) {
3830        tocopy = sb->s_blocksize - offset < toread ?
3831                sb->s_blocksize - offset : toread;
3832        bh = ext4_bread(NULL, inode, blk, 0, &err);
3833        if (err)
3834            return err;
3835        if (!bh) /* A hole? */
3836            memset(data, 0, tocopy);
3837        else
3838            memcpy(data, bh->b_data+offset, tocopy);
3839        brelse(bh);
3840        offset = 0;
3841        toread -= tocopy;
3842        data += tocopy;
3843        blk++;
3844    }
3845    return len;
3846}
3847
3848/* Write to quotafile (we know the transaction is already started and has
3849 * enough credits) */
3850static ssize_t ext4_quota_write(struct super_block *sb, int type,
3851                const char *data, size_t len, loff_t off)
3852{
3853    struct inode *inode = sb_dqopt(sb)->files[type];
3854    ext4_lblk_t blk = off >> EXT4_BLOCK_SIZE_BITS(sb);
3855    int err = 0;
3856    int offset = off & (sb->s_blocksize - 1);
3857    int tocopy;
3858    int journal_quota = EXT4_SB(sb)->s_qf_names[type] != NULL;
3859    size_t towrite = len;
3860    struct buffer_head *bh;
3861    handle_t *handle = journal_current_handle();
3862
3863    if (EXT4_SB(sb)->s_journal && !handle) {
3864        ext4_msg(sb, KERN_WARNING, "Quota write (off=%llu, len=%llu)"
3865            " cancelled because transaction is not started",
3866            (unsigned long long)off, (unsigned long long)len);
3867        return -EIO;
3868    }
3869    mutex_lock_nested(&inode->i_mutex, I_MUTEX_QUOTA);
3870    while (towrite > 0) {
3871        tocopy = sb->s_blocksize - offset < towrite ?
3872                sb->s_blocksize - offset : towrite;
3873        bh = ext4_bread(handle, inode, blk, 1, &err);
3874        if (!bh)
3875            goto out;
3876        if (journal_quota) {
3877            err = ext4_journal_get_write_access(handle, bh);
3878            if (err) {
3879                brelse(bh);
3880                goto out;
3881            }
3882        }
3883        lock_buffer(bh);
3884        memcpy(bh->b_data+offset, data, tocopy);
3885        flush_dcache_page(bh->b_page);
3886        unlock_buffer(bh);
3887        if (journal_quota)
3888            err = ext4_handle_dirty_metadata(handle, NULL, bh);
3889        else {
3890            /* Always do at least ordered writes for quotas */
3891            err = ext4_jbd2_file_inode(handle, inode);
3892            mark_buffer_dirty(bh);
3893        }
3894        brelse(bh);
3895        if (err)
3896            goto out;
3897        offset = 0;
3898        towrite -= tocopy;
3899        data += tocopy;
3900        blk++;
3901    }
3902out:
3903    if (len == towrite) {
3904        mutex_unlock(&inode->i_mutex);
3905        return err;
3906    }
3907    if (inode->i_size < off+len-towrite) {
3908        i_size_write(inode, off+len-towrite);
3909        EXT4_I(inode)->i_disksize = inode->i_size;
3910    }
3911    inode->i_mtime = inode->i_ctime = CURRENT_TIME;
3912    ext4_mark_inode_dirty(handle, inode);
3913    mutex_unlock(&inode->i_mutex);
3914    return len - towrite;
3915}
3916
3917#endif
3918
3919static int ext4_get_sb(struct file_system_type *fs_type, int flags,
3920               const char *dev_name, void *data, struct vfsmount *mnt)
3921{
3922    return get_sb_bdev(fs_type, flags, dev_name, data, ext4_fill_super,mnt);
3923}
3924
3925static struct file_system_type ext4_fs_type = {
3926    .owner = THIS_MODULE,
3927    .name = "ext4",
3928    .get_sb = ext4_get_sb,
3929    .kill_sb = kill_block_super,
3930    .fs_flags = FS_REQUIRES_DEV,
3931};
3932
3933#ifdef CONFIG_EXT4DEV_COMPAT
3934static int ext4dev_get_sb(struct file_system_type *fs_type, int flags,
3935              const char *dev_name, void *data,struct vfsmount *mnt)
3936{
3937    printk(KERN_WARNING "EXT4-fs (%s): Update your userspace programs "
3938           "to mount using ext4\n", dev_name);
3939    printk(KERN_WARNING "EXT4-fs (%s): ext4dev backwards compatibility "
3940           "will go away by 2.6.31\n", dev_name);
3941    return get_sb_bdev(fs_type, flags, dev_name, data, ext4_fill_super,mnt);
3942}
3943
3944static struct file_system_type ext4dev_fs_type = {
3945    .owner = THIS_MODULE,
3946    .name = "ext4dev",
3947    .get_sb = ext4dev_get_sb,
3948    .kill_sb = kill_block_super,
3949    .fs_flags = FS_REQUIRES_DEV,
3950};
3951MODULE_ALIAS("ext4dev");
3952#endif
3953
3954static int __init init_ext4_fs(void)
3955{
3956    int err;
3957
3958    err = init_ext4_system_zone();
3959    if (err)
3960        return err;
3961    ext4_kset = kset_create_and_add("ext4", NULL, fs_kobj);
3962    if (!ext4_kset)
3963        goto out4;
3964    ext4_proc_root = proc_mkdir("fs/ext4", NULL);
3965    err = init_ext4_mballoc();
3966    if (err)
3967        goto out3;
3968
3969    err = init_ext4_xattr();
3970    if (err)
3971        goto out2;
3972    err = init_inodecache();
3973    if (err)
3974        goto out1;
3975    err = register_filesystem(&ext4_fs_type);
3976    if (err)
3977        goto out;
3978#ifdef CONFIG_EXT4DEV_COMPAT
3979    err = register_filesystem(&ext4dev_fs_type);
3980    if (err) {
3981        unregister_filesystem(&ext4_fs_type);
3982        goto out;
3983    }
3984#endif
3985    return 0;
3986out:
3987    destroy_inodecache();
3988out1:
3989    exit_ext4_xattr();
3990out2:
3991    exit_ext4_mballoc();
3992out3:
3993    remove_proc_entry("fs/ext4", NULL);
3994    kset_unregister(ext4_kset);
3995out4:
3996    exit_ext4_system_zone();
3997    return err;
3998}
3999
4000static void __exit exit_ext4_fs(void)
4001{
4002    unregister_filesystem(&ext4_fs_type);
4003#ifdef CONFIG_EXT4DEV_COMPAT
4004    unregister_filesystem(&ext4dev_fs_type);
4005#endif
4006    destroy_inodecache();
4007    exit_ext4_xattr();
4008    exit_ext4_mballoc();
4009    remove_proc_entry("fs/ext4", NULL);
4010    kset_unregister(ext4_kset);
4011    exit_ext4_system_zone();
4012}
4013
4014MODULE_AUTHOR("Remy Card, Stephen Tweedie, Andrew Morton, Andreas Dilger, Theodore Ts'o and others");
4015MODULE_DESCRIPTION("Fourth Extended Filesystem");
4016MODULE_LICENSE("GPL");
4017module_init(init_ext4_fs)
4018module_exit(exit_ext4_fs)
4019

Archive Download this file



interactive