]>
Commit | Line | Data |
---|---|---|
ac27a0ec | 1 | /* |
617ba13b | 2 | * linux/fs/ext4/super.c |
ac27a0ec DK |
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> | |
c5ca7c76 | 23 | #include <linux/vmalloc.h> |
dab291af | 24 | #include <linux/jbd2.h> |
ac27a0ec DK |
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> | |
a5694255 | 31 | #include <linux/exportfs.h> |
ac27a0ec DK |
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> | |
9f6200bb | 38 | #include <linux/proc_fs.h> |
3197ebdb | 39 | #include <linux/ctype.h> |
1330593e | 40 | #include <linux/log2.h> |
717d50e4 | 41 | #include <linux/crc16.h> |
ac27a0ec DK |
42 | #include <asm/uaccess.h> |
43 | ||
3dcf5451 CH |
44 | #include "ext4.h" |
45 | #include "ext4_jbd2.h" | |
ac27a0ec DK |
46 | #include "xattr.h" |
47 | #include "acl.h" | |
3661d286 | 48 | #include "mballoc.h" |
ac27a0ec | 49 | |
9bffad1e TT |
50 | #define CREATE_TRACE_POINTS |
51 | #include <trace/events/ext4.h> | |
52 | ||
9f6200bb | 53 | struct proc_dir_entry *ext4_proc_root; |
3197ebdb | 54 | static struct kset *ext4_kset; |
9f6200bb | 55 | |
617ba13b | 56 | static int ext4_load_journal(struct super_block *, struct ext4_super_block *, |
ac27a0ec | 57 | unsigned long journal_devnum); |
e2d67052 | 58 | static int ext4_commit_super(struct super_block *sb, int sync); |
2b2d6d01 TT |
59 | static void ext4_mark_recovery_complete(struct super_block *sb, |
60 | struct ext4_super_block *es); | |
61 | static void ext4_clear_journal_err(struct super_block *sb, | |
62 | struct ext4_super_block *es); | |
617ba13b | 63 | static int ext4_sync_fs(struct super_block *sb, int wait); |
2b2d6d01 | 64 | static const char *ext4_decode_error(struct super_block *sb, int errno, |
ac27a0ec | 65 | char nbuf[16]); |
2b2d6d01 TT |
66 | static int ext4_remount(struct super_block *sb, int *flags, char *data); |
67 | static int ext4_statfs(struct dentry *dentry, struct kstatfs *buf); | |
c4be0c1d | 68 | static int ext4_unfreeze(struct super_block *sb); |
2b2d6d01 | 69 | static void ext4_write_super(struct super_block *sb); |
c4be0c1d | 70 | static int ext4_freeze(struct super_block *sb); |
ba69f9ab JK |
71 | static int ext4_get_sb(struct file_system_type *fs_type, int flags, |
72 | const char *dev_name, void *data, struct vfsmount *mnt); | |
ac27a0ec | 73 | |
ba69f9ab JK |
74 | #if !defined(CONFIG_EXT3_FS) && !defined(CONFIG_EXT3_FS_MODULE) && defined(CONFIG_EXT4_USE_FOR_EXT23) |
75 | static struct file_system_type ext3_fs_type = { | |
76 | .owner = THIS_MODULE, | |
77 | .name = "ext3", | |
78 | .get_sb = ext4_get_sb, | |
79 | .kill_sb = kill_block_super, | |
80 | .fs_flags = FS_REQUIRES_DEV, | |
81 | }; | |
82 | #define IS_EXT3_SB(sb) ((sb)->s_bdev->bd_holder == &ext3_fs_type) | |
83 | #else | |
84 | #define IS_EXT3_SB(sb) (0) | |
85 | #endif | |
bd81d8ee | 86 | |
8fadc143 AR |
87 | ext4_fsblk_t ext4_block_bitmap(struct super_block *sb, |
88 | struct ext4_group_desc *bg) | |
bd81d8ee | 89 | { |
3a14589c | 90 | return le32_to_cpu(bg->bg_block_bitmap_lo) | |
8fadc143 | 91 | (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ? |
0b8e58a1 | 92 | (ext4_fsblk_t)le32_to_cpu(bg->bg_block_bitmap_hi) << 32 : 0); |
bd81d8ee LV |
93 | } |
94 | ||
8fadc143 AR |
95 | ext4_fsblk_t ext4_inode_bitmap(struct super_block *sb, |
96 | struct ext4_group_desc *bg) | |
bd81d8ee | 97 | { |
5272f837 | 98 | return le32_to_cpu(bg->bg_inode_bitmap_lo) | |
8fadc143 | 99 | (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ? |
0b8e58a1 | 100 | (ext4_fsblk_t)le32_to_cpu(bg->bg_inode_bitmap_hi) << 32 : 0); |
bd81d8ee LV |
101 | } |
102 | ||
8fadc143 AR |
103 | ext4_fsblk_t ext4_inode_table(struct super_block *sb, |
104 | struct ext4_group_desc *bg) | |
bd81d8ee | 105 | { |
5272f837 | 106 | return le32_to_cpu(bg->bg_inode_table_lo) | |
8fadc143 | 107 | (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ? |
0b8e58a1 | 108 | (ext4_fsblk_t)le32_to_cpu(bg->bg_inode_table_hi) << 32 : 0); |
bd81d8ee LV |
109 | } |
110 | ||
560671a0 AK |
111 | __u32 ext4_free_blks_count(struct super_block *sb, |
112 | struct ext4_group_desc *bg) | |
113 | { | |
114 | return le16_to_cpu(bg->bg_free_blocks_count_lo) | | |
115 | (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ? | |
0b8e58a1 | 116 | (__u32)le16_to_cpu(bg->bg_free_blocks_count_hi) << 16 : 0); |
560671a0 AK |
117 | } |
118 | ||
119 | __u32 ext4_free_inodes_count(struct super_block *sb, | |
120 | struct ext4_group_desc *bg) | |
121 | { | |
122 | return le16_to_cpu(bg->bg_free_inodes_count_lo) | | |
123 | (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ? | |
0b8e58a1 | 124 | (__u32)le16_to_cpu(bg->bg_free_inodes_count_hi) << 16 : 0); |
560671a0 AK |
125 | } |
126 | ||
127 | __u32 ext4_used_dirs_count(struct super_block *sb, | |
128 | struct ext4_group_desc *bg) | |
129 | { | |
130 | return le16_to_cpu(bg->bg_used_dirs_count_lo) | | |
131 | (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ? | |
0b8e58a1 | 132 | (__u32)le16_to_cpu(bg->bg_used_dirs_count_hi) << 16 : 0); |
560671a0 AK |
133 | } |
134 | ||
135 | __u32 ext4_itable_unused_count(struct super_block *sb, | |
136 | struct ext4_group_desc *bg) | |
137 | { | |
138 | return le16_to_cpu(bg->bg_itable_unused_lo) | | |
139 | (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT ? | |
0b8e58a1 | 140 | (__u32)le16_to_cpu(bg->bg_itable_unused_hi) << 16 : 0); |
560671a0 AK |
141 | } |
142 | ||
8fadc143 AR |
143 | void ext4_block_bitmap_set(struct super_block *sb, |
144 | struct ext4_group_desc *bg, ext4_fsblk_t blk) | |
bd81d8ee | 145 | { |
3a14589c | 146 | bg->bg_block_bitmap_lo = cpu_to_le32((u32)blk); |
8fadc143 AR |
147 | if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT) |
148 | bg->bg_block_bitmap_hi = cpu_to_le32(blk >> 32); | |
bd81d8ee LV |
149 | } |
150 | ||
8fadc143 AR |
151 | void ext4_inode_bitmap_set(struct super_block *sb, |
152 | struct ext4_group_desc *bg, ext4_fsblk_t blk) | |
bd81d8ee | 153 | { |
5272f837 | 154 | bg->bg_inode_bitmap_lo = cpu_to_le32((u32)blk); |
8fadc143 AR |
155 | if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT) |
156 | bg->bg_inode_bitmap_hi = cpu_to_le32(blk >> 32); | |
bd81d8ee LV |
157 | } |
158 | ||
8fadc143 AR |
159 | void ext4_inode_table_set(struct super_block *sb, |
160 | struct ext4_group_desc *bg, ext4_fsblk_t blk) | |
bd81d8ee | 161 | { |
5272f837 | 162 | bg->bg_inode_table_lo = cpu_to_le32((u32)blk); |
8fadc143 AR |
163 | if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT) |
164 | bg->bg_inode_table_hi = cpu_to_le32(blk >> 32); | |
bd81d8ee LV |
165 | } |
166 | ||
560671a0 AK |
167 | void ext4_free_blks_set(struct super_block *sb, |
168 | struct ext4_group_desc *bg, __u32 count) | |
169 | { | |
170 | bg->bg_free_blocks_count_lo = cpu_to_le16((__u16)count); | |
171 | if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT) | |
172 | bg->bg_free_blocks_count_hi = cpu_to_le16(count >> 16); | |
173 | } | |
174 | ||
175 | void ext4_free_inodes_set(struct super_block *sb, | |
176 | struct ext4_group_desc *bg, __u32 count) | |
177 | { | |
178 | bg->bg_free_inodes_count_lo = cpu_to_le16((__u16)count); | |
179 | if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT) | |
180 | bg->bg_free_inodes_count_hi = cpu_to_le16(count >> 16); | |
181 | } | |
182 | ||
183 | void ext4_used_dirs_set(struct super_block *sb, | |
184 | struct ext4_group_desc *bg, __u32 count) | |
185 | { | |
186 | bg->bg_used_dirs_count_lo = cpu_to_le16((__u16)count); | |
187 | if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT) | |
188 | bg->bg_used_dirs_count_hi = cpu_to_le16(count >> 16); | |
189 | } | |
190 | ||
191 | void ext4_itable_unused_set(struct super_block *sb, | |
192 | struct ext4_group_desc *bg, __u32 count) | |
193 | { | |
194 | bg->bg_itable_unused_lo = cpu_to_le16((__u16)count); | |
195 | if (EXT4_DESC_SIZE(sb) >= EXT4_MIN_DESC_SIZE_64BIT) | |
196 | bg->bg_itable_unused_hi = cpu_to_le16(count >> 16); | |
197 | } | |
198 | ||
d3d1faf6 CW |
199 | |
200 | /* Just increment the non-pointer handle value */ | |
201 | static handle_t *ext4_get_nojournal(void) | |
202 | { | |
203 | handle_t *handle = current->journal_info; | |
204 | unsigned long ref_cnt = (unsigned long)handle; | |
205 | ||
206 | BUG_ON(ref_cnt >= EXT4_NOJOURNAL_MAX_REF_COUNT); | |
207 | ||
208 | ref_cnt++; | |
209 | handle = (handle_t *)ref_cnt; | |
210 | ||
211 | current->journal_info = handle; | |
212 | return handle; | |
213 | } | |
214 | ||
215 | ||
216 | /* Decrement the non-pointer handle value */ | |
217 | static void ext4_put_nojournal(handle_t *handle) | |
218 | { | |
219 | unsigned long ref_cnt = (unsigned long)handle; | |
220 | ||
221 | BUG_ON(ref_cnt == 0); | |
222 | ||
223 | ref_cnt--; | |
224 | handle = (handle_t *)ref_cnt; | |
225 | ||
226 | current->journal_info = handle; | |
227 | } | |
228 | ||
ac27a0ec | 229 | /* |
dab291af | 230 | * Wrappers for jbd2_journal_start/end. |
ac27a0ec DK |
231 | * |
232 | * The only special thing we need to do here is to make sure that all | |
233 | * journal_end calls result in the superblock being marked dirty, so | |
234 | * that sync() will call the filesystem's write_super callback if | |
235 | * appropriate. | |
236 | */ | |
617ba13b | 237 | handle_t *ext4_journal_start_sb(struct super_block *sb, int nblocks) |
ac27a0ec DK |
238 | { |
239 | journal_t *journal; | |
240 | ||
241 | if (sb->s_flags & MS_RDONLY) | |
242 | return ERR_PTR(-EROFS); | |
243 | ||
6b0310fb | 244 | vfs_check_frozen(sb, SB_FREEZE_WRITE); |
ac27a0ec DK |
245 | /* Special case here: if the journal has aborted behind our |
246 | * backs (eg. EIO in the commit thread), then we still need to | |
247 | * take the FS itself readonly cleanly. */ | |
617ba13b | 248 | journal = EXT4_SB(sb)->s_journal; |
0390131b FM |
249 | if (journal) { |
250 | if (is_journal_aborted(journal)) { | |
c67d859e | 251 | ext4_abort(sb, "Detected aborted journal"); |
0390131b FM |
252 | return ERR_PTR(-EROFS); |
253 | } | |
254 | return jbd2_journal_start(journal, nblocks); | |
ac27a0ec | 255 | } |
d3d1faf6 | 256 | return ext4_get_nojournal(); |
ac27a0ec DK |
257 | } |
258 | ||
259 | /* | |
260 | * The only special thing we need to do here is to make sure that all | |
dab291af | 261 | * jbd2_journal_stop calls result in the superblock being marked dirty, so |
ac27a0ec DK |
262 | * that sync() will call the filesystem's write_super callback if |
263 | * appropriate. | |
264 | */ | |
c398eda0 | 265 | int __ext4_journal_stop(const char *where, unsigned int line, handle_t *handle) |
ac27a0ec DK |
266 | { |
267 | struct super_block *sb; | |
268 | int err; | |
269 | int rc; | |
270 | ||
0390131b | 271 | if (!ext4_handle_valid(handle)) { |
d3d1faf6 | 272 | ext4_put_nojournal(handle); |
0390131b FM |
273 | return 0; |
274 | } | |
ac27a0ec DK |
275 | sb = handle->h_transaction->t_journal->j_private; |
276 | err = handle->h_err; | |
dab291af | 277 | rc = jbd2_journal_stop(handle); |
ac27a0ec DK |
278 | |
279 | if (!err) | |
280 | err = rc; | |
281 | if (err) | |
c398eda0 | 282 | __ext4_std_error(sb, where, line, err); |
ac27a0ec DK |
283 | return err; |
284 | } | |
285 | ||
90c7201b TT |
286 | void ext4_journal_abort_handle(const char *caller, unsigned int line, |
287 | const char *err_fn, struct buffer_head *bh, | |
288 | handle_t *handle, int err) | |
ac27a0ec DK |
289 | { |
290 | char nbuf[16]; | |
617ba13b | 291 | const char *errstr = ext4_decode_error(NULL, err, nbuf); |
ac27a0ec | 292 | |
0390131b FM |
293 | BUG_ON(!ext4_handle_valid(handle)); |
294 | ||
ac27a0ec DK |
295 | if (bh) |
296 | BUFFER_TRACE(bh, "abort"); | |
297 | ||
298 | if (!handle->h_err) | |
299 | handle->h_err = err; | |
300 | ||
301 | if (is_handle_aborted(handle)) | |
302 | return; | |
303 | ||
90c7201b TT |
304 | printk(KERN_ERR "%s:%d: aborting transaction: %s in %s\n", |
305 | caller, line, errstr, err_fn); | |
ac27a0ec | 306 | |
dab291af | 307 | jbd2_journal_abort_handle(handle); |
ac27a0ec DK |
308 | } |
309 | ||
1c13d5c0 TT |
310 | static void __save_error_info(struct super_block *sb, const char *func, |
311 | unsigned int line) | |
312 | { | |
313 | struct ext4_super_block *es = EXT4_SB(sb)->s_es; | |
314 | ||
315 | EXT4_SB(sb)->s_mount_state |= EXT4_ERROR_FS; | |
316 | es->s_state |= cpu_to_le16(EXT4_ERROR_FS); | |
317 | es->s_last_error_time = cpu_to_le32(get_seconds()); | |
318 | strncpy(es->s_last_error_func, func, sizeof(es->s_last_error_func)); | |
319 | es->s_last_error_line = cpu_to_le32(line); | |
320 | if (!es->s_first_error_time) { | |
321 | es->s_first_error_time = es->s_last_error_time; | |
322 | strncpy(es->s_first_error_func, func, | |
323 | sizeof(es->s_first_error_func)); | |
324 | es->s_first_error_line = cpu_to_le32(line); | |
325 | es->s_first_error_ino = es->s_last_error_ino; | |
326 | es->s_first_error_block = es->s_last_error_block; | |
327 | } | |
66e61a9e TT |
328 | /* |
329 | * Start the daily error reporting function if it hasn't been | |
330 | * started already | |
331 | */ | |
332 | if (!es->s_error_count) | |
333 | mod_timer(&EXT4_SB(sb)->s_err_report, jiffies + 24*60*60*HZ); | |
1c13d5c0 TT |
334 | es->s_error_count = cpu_to_le32(le32_to_cpu(es->s_error_count) + 1); |
335 | } | |
336 | ||
337 | static void save_error_info(struct super_block *sb, const char *func, | |
338 | unsigned int line) | |
339 | { | |
340 | __save_error_info(sb, func, line); | |
341 | ext4_commit_super(sb, 1); | |
342 | } | |
343 | ||
344 | ||
ac27a0ec DK |
345 | /* Deal with the reporting of failure conditions on a filesystem such as |
346 | * inconsistencies detected or read IO failures. | |
347 | * | |
348 | * On ext2, we can store the error state of the filesystem in the | |
617ba13b | 349 | * superblock. That is not possible on ext4, because we may have other |
ac27a0ec DK |
350 | * write ordering constraints on the superblock which prevent us from |
351 | * writing it out straight away; and given that the journal is about to | |
352 | * be aborted, we can't rely on the current, or future, transactions to | |
353 | * write out the superblock safely. | |
354 | * | |
dab291af | 355 | * We'll just use the jbd2_journal_abort() error code to record an error in |
d6b198bc | 356 | * the journal instead. On recovery, the journal will complain about |
ac27a0ec DK |
357 | * that error until we've noted it down and cleared it. |
358 | */ | |
359 | ||
617ba13b | 360 | static void ext4_handle_error(struct super_block *sb) |
ac27a0ec | 361 | { |
ac27a0ec DK |
362 | if (sb->s_flags & MS_RDONLY) |
363 | return; | |
364 | ||
2b2d6d01 | 365 | if (!test_opt(sb, ERRORS_CONT)) { |
617ba13b | 366 | journal_t *journal = EXT4_SB(sb)->s_journal; |
ac27a0ec | 367 | |
4ab2f15b | 368 | EXT4_SB(sb)->s_mount_flags |= EXT4_MF_FS_ABORTED; |
ac27a0ec | 369 | if (journal) |
dab291af | 370 | jbd2_journal_abort(journal, -EIO); |
ac27a0ec | 371 | } |
2b2d6d01 | 372 | if (test_opt(sb, ERRORS_RO)) { |
b31e1552 | 373 | ext4_msg(sb, KERN_CRIT, "Remounting filesystem read-only"); |
ac27a0ec DK |
374 | sb->s_flags |= MS_RDONLY; |
375 | } | |
ac27a0ec | 376 | if (test_opt(sb, ERRORS_PANIC)) |
617ba13b | 377 | panic("EXT4-fs (device %s): panic forced after error\n", |
ac27a0ec DK |
378 | sb->s_id); |
379 | } | |
380 | ||
12062ddd | 381 | void __ext4_error(struct super_block *sb, const char *function, |
c398eda0 | 382 | unsigned int line, const char *fmt, ...) |
ac27a0ec DK |
383 | { |
384 | va_list args; | |
385 | ||
386 | va_start(args, fmt); | |
c398eda0 TT |
387 | printk(KERN_CRIT "EXT4-fs error (device %s): %s:%d: comm %s: ", |
388 | sb->s_id, function, line, current->comm); | |
ac27a0ec DK |
389 | vprintk(fmt, args); |
390 | printk("\n"); | |
391 | va_end(args); | |
392 | ||
617ba13b | 393 | ext4_handle_error(sb); |
ac27a0ec DK |
394 | } |
395 | ||
c398eda0 TT |
396 | void ext4_error_inode(struct inode *inode, const char *function, |
397 | unsigned int line, ext4_fsblk_t block, | |
273df556 FM |
398 | const char *fmt, ...) |
399 | { | |
400 | va_list args; | |
1c13d5c0 | 401 | struct ext4_super_block *es = EXT4_SB(inode->i_sb)->s_es; |
273df556 | 402 | |
1c13d5c0 TT |
403 | es->s_last_error_ino = cpu_to_le32(inode->i_ino); |
404 | es->s_last_error_block = cpu_to_le64(block); | |
405 | save_error_info(inode->i_sb, function, line); | |
273df556 | 406 | va_start(args, fmt); |
c398eda0 TT |
407 | printk(KERN_CRIT "EXT4-fs error (device %s): %s:%d: inode #%lu: ", |
408 | inode->i_sb->s_id, function, line, inode->i_ino); | |
409 | if (block) | |
410 | printk("block %llu: ", block); | |
411 | printk("comm %s: ", current->comm); | |
273df556 FM |
412 | vprintk(fmt, args); |
413 | printk("\n"); | |
414 | va_end(args); | |
415 | ||
416 | ext4_handle_error(inode->i_sb); | |
417 | } | |
418 | ||
c398eda0 TT |
419 | void ext4_error_file(struct file *file, const char *function, |
420 | unsigned int line, const char *fmt, ...) | |
273df556 FM |
421 | { |
422 | va_list args; | |
1c13d5c0 | 423 | struct ext4_super_block *es; |
273df556 FM |
424 | struct inode *inode = file->f_dentry->d_inode; |
425 | char pathname[80], *path; | |
426 | ||
1c13d5c0 TT |
427 | es = EXT4_SB(inode->i_sb)->s_es; |
428 | es->s_last_error_ino = cpu_to_le32(inode->i_ino); | |
429 | save_error_info(inode->i_sb, function, line); | |
273df556 FM |
430 | va_start(args, fmt); |
431 | path = d_path(&(file->f_path), pathname, sizeof(pathname)); | |
432 | if (!path) | |
433 | path = "(unknown)"; | |
434 | printk(KERN_CRIT | |
c398eda0 TT |
435 | "EXT4-fs error (device %s): %s:%d: inode #%lu " |
436 | "(comm %s path %s): ", | |
437 | inode->i_sb->s_id, function, line, inode->i_ino, | |
438 | current->comm, path); | |
273df556 FM |
439 | vprintk(fmt, args); |
440 | printk("\n"); | |
441 | va_end(args); | |
442 | ||
443 | ext4_handle_error(inode->i_sb); | |
444 | } | |
445 | ||
2b2d6d01 | 446 | static const char *ext4_decode_error(struct super_block *sb, int errno, |
ac27a0ec DK |
447 | char nbuf[16]) |
448 | { | |
449 | char *errstr = NULL; | |
450 | ||
451 | switch (errno) { | |
452 | case -EIO: | |
453 | errstr = "IO failure"; | |
454 | break; | |
455 | case -ENOMEM: | |
456 | errstr = "Out of memory"; | |
457 | break; | |
458 | case -EROFS: | |
78f1ddbb TT |
459 | if (!sb || (EXT4_SB(sb)->s_journal && |
460 | EXT4_SB(sb)->s_journal->j_flags & JBD2_ABORT)) | |
ac27a0ec DK |
461 | errstr = "Journal has aborted"; |
462 | else | |
463 | errstr = "Readonly filesystem"; | |
464 | break; | |
465 | default: | |
466 | /* If the caller passed in an extra buffer for unknown | |
467 | * errors, textualise them now. Else we just return | |
468 | * NULL. */ | |
469 | if (nbuf) { | |
470 | /* Check for truncated error codes... */ | |
471 | if (snprintf(nbuf, 16, "error %d", -errno) >= 0) | |
472 | errstr = nbuf; | |
473 | } | |
474 | break; | |
475 | } | |
476 | ||
477 | return errstr; | |
478 | } | |
479 | ||
617ba13b | 480 | /* __ext4_std_error decodes expected errors from journaling functions |
ac27a0ec DK |
481 | * automatically and invokes the appropriate error response. */ |
482 | ||
c398eda0 TT |
483 | void __ext4_std_error(struct super_block *sb, const char *function, |
484 | unsigned int line, int errno) | |
ac27a0ec DK |
485 | { |
486 | char nbuf[16]; | |
487 | const char *errstr; | |
488 | ||
489 | /* Special case: if the error is EROFS, and we're not already | |
490 | * inside a transaction, then there's really no point in logging | |
491 | * an error. */ | |
492 | if (errno == -EROFS && journal_current_handle() == NULL && | |
493 | (sb->s_flags & MS_RDONLY)) | |
494 | return; | |
495 | ||
617ba13b | 496 | errstr = ext4_decode_error(sb, errno, nbuf); |
c398eda0 TT |
497 | printk(KERN_CRIT "EXT4-fs error (device %s) in %s:%d: %s\n", |
498 | sb->s_id, function, line, errstr); | |
1c13d5c0 | 499 | save_error_info(sb, function, line); |
ac27a0ec | 500 | |
617ba13b | 501 | ext4_handle_error(sb); |
ac27a0ec DK |
502 | } |
503 | ||
504 | /* | |
617ba13b | 505 | * ext4_abort is a much stronger failure handler than ext4_error. The |
ac27a0ec DK |
506 | * abort function may be used to deal with unrecoverable failures such |
507 | * as journal IO errors or ENOMEM at a critical moment in log management. | |
508 | * | |
509 | * We unconditionally force the filesystem into an ABORT|READONLY state, | |
510 | * unless the error response on the fs has been set to panic in which | |
511 | * case we take the easy way out and panic immediately. | |
512 | */ | |
513 | ||
c67d859e | 514 | void __ext4_abort(struct super_block *sb, const char *function, |
c398eda0 | 515 | unsigned int line, const char *fmt, ...) |
ac27a0ec DK |
516 | { |
517 | va_list args; | |
518 | ||
1c13d5c0 | 519 | save_error_info(sb, function, line); |
ac27a0ec | 520 | va_start(args, fmt); |
c398eda0 TT |
521 | printk(KERN_CRIT "EXT4-fs error (device %s): %s:%d: ", sb->s_id, |
522 | function, line); | |
ac27a0ec DK |
523 | vprintk(fmt, args); |
524 | printk("\n"); | |
525 | va_end(args); | |
526 | ||
1c13d5c0 TT |
527 | if ((sb->s_flags & MS_RDONLY) == 0) { |
528 | ext4_msg(sb, KERN_CRIT, "Remounting filesystem read-only"); | |
529 | sb->s_flags |= MS_RDONLY; | |
530 | EXT4_SB(sb)->s_mount_flags |= EXT4_MF_FS_ABORTED; | |
531 | if (EXT4_SB(sb)->s_journal) | |
532 | jbd2_journal_abort(EXT4_SB(sb)->s_journal, -EIO); | |
533 | save_error_info(sb, function, line); | |
534 | } | |
ac27a0ec | 535 | if (test_opt(sb, ERRORS_PANIC)) |
617ba13b | 536 | panic("EXT4-fs panic from previous error\n"); |
ac27a0ec DK |
537 | } |
538 | ||
b31e1552 ES |
539 | void ext4_msg (struct super_block * sb, const char *prefix, |
540 | const char *fmt, ...) | |
541 | { | |
542 | va_list args; | |
543 | ||
544 | va_start(args, fmt); | |
545 | printk("%sEXT4-fs (%s): ", prefix, sb->s_id); | |
546 | vprintk(fmt, args); | |
547 | printk("\n"); | |
548 | va_end(args); | |
549 | } | |
550 | ||
12062ddd | 551 | void __ext4_warning(struct super_block *sb, const char *function, |
c398eda0 | 552 | unsigned int line, const char *fmt, ...) |
ac27a0ec DK |
553 | { |
554 | va_list args; | |
555 | ||
556 | va_start(args, fmt); | |
c398eda0 TT |
557 | printk(KERN_WARNING "EXT4-fs warning (device %s): %s:%d: ", |
558 | sb->s_id, function, line); | |
ac27a0ec DK |
559 | vprintk(fmt, args); |
560 | printk("\n"); | |
561 | va_end(args); | |
562 | } | |
563 | ||
e29136f8 TT |
564 | void __ext4_grp_locked_error(const char *function, unsigned int line, |
565 | struct super_block *sb, ext4_group_t grp, | |
566 | unsigned long ino, ext4_fsblk_t block, | |
567 | const char *fmt, ...) | |
5d1b1b3f AK |
568 | __releases(bitlock) |
569 | __acquires(bitlock) | |
570 | { | |
571 | va_list args; | |
572 | struct ext4_super_block *es = EXT4_SB(sb)->s_es; | |
573 | ||
1c13d5c0 TT |
574 | es->s_last_error_ino = cpu_to_le32(ino); |
575 | es->s_last_error_block = cpu_to_le64(block); | |
576 | __save_error_info(sb, function, line); | |
5d1b1b3f | 577 | va_start(args, fmt); |
e29136f8 TT |
578 | printk(KERN_CRIT "EXT4-fs error (device %s): %s:%d: group %u", |
579 | sb->s_id, function, line, grp); | |
580 | if (ino) | |
581 | printk("inode %lu: ", ino); | |
582 | if (block) | |
583 | printk("block %llu:", (unsigned long long) block); | |
5d1b1b3f AK |
584 | vprintk(fmt, args); |
585 | printk("\n"); | |
586 | va_end(args); | |
587 | ||
588 | if (test_opt(sb, ERRORS_CONT)) { | |
e2d67052 | 589 | ext4_commit_super(sb, 0); |
5d1b1b3f AK |
590 | return; |
591 | } | |
1c13d5c0 | 592 | |
5d1b1b3f AK |
593 | ext4_unlock_group(sb, grp); |
594 | ext4_handle_error(sb); | |
595 | /* | |
596 | * We only get here in the ERRORS_RO case; relocking the group | |
597 | * may be dangerous, but nothing bad will happen since the | |
598 | * filesystem will have already been marked read/only and the | |
599 | * journal has been aborted. We return 1 as a hint to callers | |
600 | * who might what to use the return value from | |
601 | * ext4_grp_locked_error() to distinguish beween the | |
602 | * ERRORS_CONT and ERRORS_RO case, and perhaps return more | |
603 | * aggressively from the ext4 function in question, with a | |
604 | * more appropriate error code. | |
605 | */ | |
606 | ext4_lock_group(sb, grp); | |
607 | return; | |
608 | } | |
609 | ||
617ba13b | 610 | void ext4_update_dynamic_rev(struct super_block *sb) |
ac27a0ec | 611 | { |
617ba13b | 612 | struct ext4_super_block *es = EXT4_SB(sb)->s_es; |
ac27a0ec | 613 | |
617ba13b | 614 | if (le32_to_cpu(es->s_rev_level) > EXT4_GOOD_OLD_REV) |
ac27a0ec DK |
615 | return; |
616 | ||
12062ddd | 617 | ext4_warning(sb, |
ac27a0ec DK |
618 | "updating to rev %d because of new feature flag, " |
619 | "running e2fsck is recommended", | |
617ba13b | 620 | EXT4_DYNAMIC_REV); |
ac27a0ec | 621 | |
617ba13b MC |
622 | es->s_first_ino = cpu_to_le32(EXT4_GOOD_OLD_FIRST_INO); |
623 | es->s_inode_size = cpu_to_le16(EXT4_GOOD_OLD_INODE_SIZE); | |
624 | es->s_rev_level = cpu_to_le32(EXT4_DYNAMIC_REV); | |
ac27a0ec DK |
625 | /* leave es->s_feature_*compat flags alone */ |
626 | /* es->s_uuid will be set by e2fsck if empty */ | |
627 | ||
628 | /* | |
629 | * The rest of the superblock fields should be zero, and if not it | |
630 | * means they are likely already in use, so leave them alone. We | |
631 | * can leave it up to e2fsck to clean up any inconsistencies there. | |
632 | */ | |
633 | } | |
634 | ||
635 | /* | |
636 | * Open the external journal device | |
637 | */ | |
b31e1552 | 638 | static struct block_device *ext4_blkdev_get(dev_t dev, struct super_block *sb) |
ac27a0ec DK |
639 | { |
640 | struct block_device *bdev; | |
641 | char b[BDEVNAME_SIZE]; | |
642 | ||
643 | bdev = open_by_devnum(dev, FMODE_READ|FMODE_WRITE); | |
644 | if (IS_ERR(bdev)) | |
645 | goto fail; | |
646 | return bdev; | |
647 | ||
648 | fail: | |
b31e1552 | 649 | ext4_msg(sb, KERN_ERR, "failed to open journal device %s: %ld", |
ac27a0ec DK |
650 | __bdevname(dev, b), PTR_ERR(bdev)); |
651 | return NULL; | |
652 | } | |
653 | ||
654 | /* | |
655 | * Release the journal device | |
656 | */ | |
617ba13b | 657 | static int ext4_blkdev_put(struct block_device *bdev) |
ac27a0ec DK |
658 | { |
659 | bd_release(bdev); | |
9a1c3542 | 660 | return blkdev_put(bdev, FMODE_READ|FMODE_WRITE); |
ac27a0ec DK |
661 | } |
662 | ||
617ba13b | 663 | static int ext4_blkdev_remove(struct ext4_sb_info *sbi) |
ac27a0ec DK |
664 | { |
665 | struct block_device *bdev; | |
666 | int ret = -ENODEV; | |
667 | ||
668 | bdev = sbi->journal_bdev; | |
669 | if (bdev) { | |
617ba13b | 670 | ret = ext4_blkdev_put(bdev); |
ac27a0ec DK |
671 | sbi->journal_bdev = NULL; |
672 | } | |
673 | return ret; | |
674 | } | |
675 | ||
676 | static inline struct inode *orphan_list_entry(struct list_head *l) | |
677 | { | |
617ba13b | 678 | return &list_entry(l, struct ext4_inode_info, i_orphan)->vfs_inode; |
ac27a0ec DK |
679 | } |
680 | ||
617ba13b | 681 | static void dump_orphan_list(struct super_block *sb, struct ext4_sb_info *sbi) |
ac27a0ec DK |
682 | { |
683 | struct list_head *l; | |
684 | ||
b31e1552 ES |
685 | ext4_msg(sb, KERN_ERR, "sb orphan head is %d", |
686 | le32_to_cpu(sbi->s_es->s_last_orphan)); | |
ac27a0ec DK |
687 | |
688 | printk(KERN_ERR "sb_info orphan list:\n"); | |
689 | list_for_each(l, &sbi->s_orphan) { | |
690 | struct inode *inode = orphan_list_entry(l); | |
691 | printk(KERN_ERR " " | |
692 | "inode %s:%lu at %p: mode %o, nlink %d, next %d\n", | |
693 | inode->i_sb->s_id, inode->i_ino, inode, | |
694 | inode->i_mode, inode->i_nlink, | |
695 | NEXT_ORPHAN(inode)); | |
696 | } | |
697 | } | |
698 | ||
2b2d6d01 | 699 | static void ext4_put_super(struct super_block *sb) |
ac27a0ec | 700 | { |
617ba13b MC |
701 | struct ext4_sb_info *sbi = EXT4_SB(sb); |
702 | struct ext4_super_block *es = sbi->s_es; | |
ef2cabf7 | 703 | int i, err; |
ac27a0ec | 704 | |
e0ccfd95 CH |
705 | dquot_disable(sb, -1, DQUOT_USAGE_ENABLED | DQUOT_LIMITS_ENABLED); |
706 | ||
4c0425ff MC |
707 | flush_workqueue(sbi->dio_unwritten_wq); |
708 | destroy_workqueue(sbi->dio_unwritten_wq); | |
709 | ||
a9e220f8 | 710 | lock_super(sb); |
6cfd0148 | 711 | lock_kernel(); |
8c85e125 | 712 | if (sb->s_dirt) |
ebc1ac16 | 713 | ext4_commit_super(sb, 1); |
8c85e125 | 714 | |
0390131b FM |
715 | if (sbi->s_journal) { |
716 | err = jbd2_journal_destroy(sbi->s_journal); | |
717 | sbi->s_journal = NULL; | |
718 | if (err < 0) | |
c67d859e | 719 | ext4_abort(sb, "Couldn't clean up the journal"); |
0390131b | 720 | } |
d4edac31 JB |
721 | |
722 | ext4_release_system_zone(sb); | |
723 | ext4_mb_release(sb); | |
724 | ext4_ext_release(sb); | |
725 | ext4_xattr_put_super(sb); | |
726 | ||
ac27a0ec | 727 | if (!(sb->s_flags & MS_RDONLY)) { |
617ba13b | 728 | EXT4_CLEAR_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER); |
ac27a0ec | 729 | es->s_state = cpu_to_le16(sbi->s_mount_state); |
e2d67052 | 730 | ext4_commit_super(sb, 1); |
ac27a0ec | 731 | } |
240799cd | 732 | if (sbi->s_proc) { |
9f6200bb | 733 | remove_proc_entry(sb->s_id, ext4_proc_root); |
240799cd | 734 | } |
3197ebdb | 735 | kobject_del(&sbi->s_kobj); |
ac27a0ec DK |
736 | |
737 | for (i = 0; i < sbi->s_gdb_count; i++) | |
738 | brelse(sbi->s_group_desc[i]); | |
739 | kfree(sbi->s_group_desc); | |
c5ca7c76 TT |
740 | if (is_vmalloc_addr(sbi->s_flex_groups)) |
741 | vfree(sbi->s_flex_groups); | |
742 | else | |
743 | kfree(sbi->s_flex_groups); | |
ac27a0ec DK |
744 | percpu_counter_destroy(&sbi->s_freeblocks_counter); |
745 | percpu_counter_destroy(&sbi->s_freeinodes_counter); | |
746 | percpu_counter_destroy(&sbi->s_dirs_counter); | |
6bc6e63f | 747 | percpu_counter_destroy(&sbi->s_dirtyblocks_counter); |
ac27a0ec DK |
748 | brelse(sbi->s_sbh); |
749 | #ifdef CONFIG_QUOTA | |
750 | for (i = 0; i < MAXQUOTAS; i++) | |
751 | kfree(sbi->s_qf_names[i]); | |
752 | #endif | |
753 | ||
754 | /* Debugging code just in case the in-memory inode orphan list | |
755 | * isn't empty. The on-disk one can be non-empty if we've | |
756 | * detected an error and taken the fs readonly, but the | |
757 | * in-memory list had better be clean by this point. */ | |
758 | if (!list_empty(&sbi->s_orphan)) | |
759 | dump_orphan_list(sb, sbi); | |
760 | J_ASSERT(list_empty(&sbi->s_orphan)); | |
761 | ||
f98393a6 | 762 | invalidate_bdev(sb->s_bdev); |
ac27a0ec DK |
763 | if (sbi->journal_bdev && sbi->journal_bdev != sb->s_bdev) { |
764 | /* | |
765 | * Invalidate the journal device's buffers. We don't want them | |
766 | * floating about in memory - the physical journal device may | |
767 | * hotswapped, and it breaks the `ro-after' testing code. | |
768 | */ | |
769 | sync_blockdev(sbi->journal_bdev); | |
f98393a6 | 770 | invalidate_bdev(sbi->journal_bdev); |
617ba13b | 771 | ext4_blkdev_remove(sbi); |
ac27a0ec DK |
772 | } |
773 | sb->s_fs_info = NULL; | |
3197ebdb TT |
774 | /* |
775 | * Now that we are completely done shutting down the | |
776 | * superblock, we need to actually destroy the kobject. | |
777 | */ | |
778 | unlock_kernel(); | |
779 | unlock_super(sb); | |
780 | kobject_put(&sbi->s_kobj); | |
781 | wait_for_completion(&sbi->s_kobj_unregister); | |
705895b6 | 782 | kfree(sbi->s_blockgroup_lock); |
ac27a0ec | 783 | kfree(sbi); |
ac27a0ec DK |
784 | } |
785 | ||
e18b890b | 786 | static struct kmem_cache *ext4_inode_cachep; |
ac27a0ec DK |
787 | |
788 | /* | |
789 | * Called inside transaction, so use GFP_NOFS | |
790 | */ | |
617ba13b | 791 | static struct inode *ext4_alloc_inode(struct super_block *sb) |
ac27a0ec | 792 | { |
617ba13b | 793 | struct ext4_inode_info *ei; |
ac27a0ec | 794 | |
e6b4f8da | 795 | ei = kmem_cache_alloc(ext4_inode_cachep, GFP_NOFS); |
ac27a0ec DK |
796 | if (!ei) |
797 | return NULL; | |
0b8e58a1 | 798 | |
ac27a0ec | 799 | ei->vfs_inode.i_version = 1; |
91246c00 | 800 | ei->vfs_inode.i_data.writeback_index = 0; |
a86c6181 | 801 | memset(&ei->i_cached_extent, 0, sizeof(struct ext4_ext_cache)); |
c9de560d AT |
802 | INIT_LIST_HEAD(&ei->i_prealloc_list); |
803 | spin_lock_init(&ei->i_prealloc_lock); | |
0390131b FM |
804 | /* |
805 | * Note: We can be called before EXT4_SB(sb)->s_journal is set, | |
806 | * therefore it can be null here. Don't check it, just initialize | |
807 | * jinode. | |
808 | */ | |
678aaf48 | 809 | jbd2_journal_init_jbd_inode(&ei->jinode, &ei->vfs_inode); |
d2a17637 MC |
810 | ei->i_reserved_data_blocks = 0; |
811 | ei->i_reserved_meta_blocks = 0; | |
812 | ei->i_allocated_meta_blocks = 0; | |
9d0be502 | 813 | ei->i_da_metadata_calc_len = 0; |
d2a17637 MC |
814 | ei->i_delalloc_reserved_flag = 0; |
815 | spin_lock_init(&(ei->i_block_reservation_lock)); | |
a9e7f447 DM |
816 | #ifdef CONFIG_QUOTA |
817 | ei->i_reserved_quota = 0; | |
818 | #endif | |
c7064ef1 | 819 | INIT_LIST_HEAD(&ei->i_completed_io_list); |
744692dc | 820 | spin_lock_init(&ei->i_completed_io_lock); |
8d5d02e6 | 821 | ei->cur_aio_dio = NULL; |
b436b9be JK |
822 | ei->i_sync_tid = 0; |
823 | ei->i_datasync_tid = 0; | |
0b8e58a1 | 824 | |
ac27a0ec DK |
825 | return &ei->vfs_inode; |
826 | } | |
827 | ||
617ba13b | 828 | static void ext4_destroy_inode(struct inode *inode) |
ac27a0ec | 829 | { |
9f7dd93d | 830 | if (!list_empty(&(EXT4_I(inode)->i_orphan))) { |
b31e1552 ES |
831 | ext4_msg(inode->i_sb, KERN_ERR, |
832 | "Inode %lu (%p): orphan list check failed!", | |
833 | inode->i_ino, EXT4_I(inode)); | |
9f7dd93d VA |
834 | print_hex_dump(KERN_INFO, "", DUMP_PREFIX_ADDRESS, 16, 4, |
835 | EXT4_I(inode), sizeof(struct ext4_inode_info), | |
836 | true); | |
837 | dump_stack(); | |
838 | } | |
617ba13b | 839 | kmem_cache_free(ext4_inode_cachep, EXT4_I(inode)); |
ac27a0ec DK |
840 | } |
841 | ||
51cc5068 | 842 | static void init_once(void *foo) |
ac27a0ec | 843 | { |
617ba13b | 844 | struct ext4_inode_info *ei = (struct ext4_inode_info *) foo; |
ac27a0ec | 845 | |
a35afb83 | 846 | INIT_LIST_HEAD(&ei->i_orphan); |
03010a33 | 847 | #ifdef CONFIG_EXT4_FS_XATTR |
a35afb83 | 848 | init_rwsem(&ei->xattr_sem); |
ac27a0ec | 849 | #endif |
0e855ac8 | 850 | init_rwsem(&ei->i_data_sem); |
a35afb83 | 851 | inode_init_once(&ei->vfs_inode); |
ac27a0ec DK |
852 | } |
853 | ||
854 | static int init_inodecache(void) | |
855 | { | |
617ba13b MC |
856 | ext4_inode_cachep = kmem_cache_create("ext4_inode_cache", |
857 | sizeof(struct ext4_inode_info), | |
ac27a0ec DK |
858 | 0, (SLAB_RECLAIM_ACCOUNT| |
859 | SLAB_MEM_SPREAD), | |
20c2df83 | 860 | init_once); |
617ba13b | 861 | if (ext4_inode_cachep == NULL) |
ac27a0ec DK |
862 | return -ENOMEM; |
863 | return 0; | |
864 | } | |
865 | ||
866 | static void destroy_inodecache(void) | |
867 | { | |
617ba13b | 868 | kmem_cache_destroy(ext4_inode_cachep); |
ac27a0ec DK |
869 | } |
870 | ||
617ba13b | 871 | static void ext4_clear_inode(struct inode *inode) |
ac27a0ec | 872 | { |
9f754758 | 873 | dquot_drop(inode); |
c2ea3fde | 874 | ext4_discard_preallocations(inode); |
0390131b FM |
875 | if (EXT4_JOURNAL(inode)) |
876 | jbd2_journal_release_jbd_inode(EXT4_SB(inode->i_sb)->s_journal, | |
678aaf48 | 877 | &EXT4_I(inode)->jinode); |
ac27a0ec DK |
878 | } |
879 | ||
2b2d6d01 TT |
880 | static inline void ext4_show_quota_options(struct seq_file *seq, |
881 | struct super_block *sb) | |
ac27a0ec DK |
882 | { |
883 | #if defined(CONFIG_QUOTA) | |
617ba13b | 884 | struct ext4_sb_info *sbi = EXT4_SB(sb); |
ac27a0ec | 885 | |
5a20bdfc JK |
886 | if (sbi->s_jquota_fmt) { |
887 | char *fmtname = ""; | |
888 | ||
889 | switch (sbi->s_jquota_fmt) { | |
890 | case QFMT_VFS_OLD: | |
891 | fmtname = "vfsold"; | |
892 | break; | |
893 | case QFMT_VFS_V0: | |
894 | fmtname = "vfsv0"; | |
895 | break; | |
896 | case QFMT_VFS_V1: | |
897 | fmtname = "vfsv1"; | |
898 | break; | |
899 | } | |
900 | seq_printf(seq, ",jqfmt=%s", fmtname); | |
901 | } | |
ac27a0ec DK |
902 | |
903 | if (sbi->s_qf_names[USRQUOTA]) | |
904 | seq_printf(seq, ",usrjquota=%s", sbi->s_qf_names[USRQUOTA]); | |
905 | ||
906 | if (sbi->s_qf_names[GRPQUOTA]) | |
907 | seq_printf(seq, ",grpjquota=%s", sbi->s_qf_names[GRPQUOTA]); | |
908 | ||
482a7425 | 909 | if (test_opt(sb, USRQUOTA)) |
ac27a0ec DK |
910 | seq_puts(seq, ",usrquota"); |
911 | ||
482a7425 | 912 | if (test_opt(sb, GRPQUOTA)) |
ac27a0ec DK |
913 | seq_puts(seq, ",grpquota"); |
914 | #endif | |
915 | } | |
916 | ||
d9c9bef1 MS |
917 | /* |
918 | * Show an option if | |
919 | * - it's set to a non-default value OR | |
920 | * - if the per-sb default is different from the global default | |
921 | */ | |
617ba13b | 922 | static int ext4_show_options(struct seq_file *seq, struct vfsmount *vfs) |
ac27a0ec | 923 | { |
aa22df2c AK |
924 | int def_errors; |
925 | unsigned long def_mount_opts; | |
ac27a0ec | 926 | struct super_block *sb = vfs->mnt_sb; |
d9c9bef1 MS |
927 | struct ext4_sb_info *sbi = EXT4_SB(sb); |
928 | struct ext4_super_block *es = sbi->s_es; | |
d9c9bef1 MS |
929 | |
930 | def_mount_opts = le32_to_cpu(es->s_default_mount_opts); | |
aa22df2c | 931 | def_errors = le16_to_cpu(es->s_errors); |
d9c9bef1 MS |
932 | |
933 | if (sbi->s_sb_block != 1) | |
934 | seq_printf(seq, ",sb=%llu", sbi->s_sb_block); | |
935 | if (test_opt(sb, MINIX_DF)) | |
936 | seq_puts(seq, ",minixdf"); | |
aa22df2c | 937 | if (test_opt(sb, GRPID) && !(def_mount_opts & EXT4_DEFM_BSDGROUPS)) |
d9c9bef1 MS |
938 | seq_puts(seq, ",grpid"); |
939 | if (!test_opt(sb, GRPID) && (def_mount_opts & EXT4_DEFM_BSDGROUPS)) | |
940 | seq_puts(seq, ",nogrpid"); | |
941 | if (sbi->s_resuid != EXT4_DEF_RESUID || | |
942 | le16_to_cpu(es->s_def_resuid) != EXT4_DEF_RESUID) { | |
943 | seq_printf(seq, ",resuid=%u", sbi->s_resuid); | |
944 | } | |
945 | if (sbi->s_resgid != EXT4_DEF_RESGID || | |
946 | le16_to_cpu(es->s_def_resgid) != EXT4_DEF_RESGID) { | |
947 | seq_printf(seq, ",resgid=%u", sbi->s_resgid); | |
948 | } | |
bb4f397a | 949 | if (test_opt(sb, ERRORS_RO)) { |
d9c9bef1 | 950 | if (def_errors == EXT4_ERRORS_PANIC || |
bb4f397a AK |
951 | def_errors == EXT4_ERRORS_CONTINUE) { |
952 | seq_puts(seq, ",errors=remount-ro"); | |
d9c9bef1 MS |
953 | } |
954 | } | |
aa22df2c | 955 | if (test_opt(sb, ERRORS_CONT) && def_errors != EXT4_ERRORS_CONTINUE) |
bb4f397a | 956 | seq_puts(seq, ",errors=continue"); |
aa22df2c | 957 | if (test_opt(sb, ERRORS_PANIC) && def_errors != EXT4_ERRORS_PANIC) |
d9c9bef1 | 958 | seq_puts(seq, ",errors=panic"); |
aa22df2c | 959 | if (test_opt(sb, NO_UID32) && !(def_mount_opts & EXT4_DEFM_UID16)) |
d9c9bef1 | 960 | seq_puts(seq, ",nouid32"); |
aa22df2c | 961 | if (test_opt(sb, DEBUG) && !(def_mount_opts & EXT4_DEFM_DEBUG)) |
d9c9bef1 MS |
962 | seq_puts(seq, ",debug"); |
963 | if (test_opt(sb, OLDALLOC)) | |
964 | seq_puts(seq, ",oldalloc"); | |
03010a33 | 965 | #ifdef CONFIG_EXT4_FS_XATTR |
aa22df2c AK |
966 | if (test_opt(sb, XATTR_USER) && |
967 | !(def_mount_opts & EXT4_DEFM_XATTR_USER)) | |
d9c9bef1 MS |
968 | seq_puts(seq, ",user_xattr"); |
969 | if (!test_opt(sb, XATTR_USER) && | |
970 | (def_mount_opts & EXT4_DEFM_XATTR_USER)) { | |
971 | seq_puts(seq, ",nouser_xattr"); | |
972 | } | |
973 | #endif | |
03010a33 | 974 | #ifdef CONFIG_EXT4_FS_POSIX_ACL |
aa22df2c | 975 | if (test_opt(sb, POSIX_ACL) && !(def_mount_opts & EXT4_DEFM_ACL)) |
d9c9bef1 MS |
976 | seq_puts(seq, ",acl"); |
977 | if (!test_opt(sb, POSIX_ACL) && (def_mount_opts & EXT4_DEFM_ACL)) | |
978 | seq_puts(seq, ",noacl"); | |
979 | #endif | |
30773840 | 980 | if (sbi->s_commit_interval != JBD2_DEFAULT_MAX_COMMIT_AGE*HZ) { |
d9c9bef1 MS |
981 | seq_printf(seq, ",commit=%u", |
982 | (unsigned) (sbi->s_commit_interval / HZ)); | |
983 | } | |
30773840 TT |
984 | if (sbi->s_min_batch_time != EXT4_DEF_MIN_BATCH_TIME) { |
985 | seq_printf(seq, ",min_batch_time=%u", | |
986 | (unsigned) sbi->s_min_batch_time); | |
987 | } | |
988 | if (sbi->s_max_batch_time != EXT4_DEF_MAX_BATCH_TIME) { | |
989 | seq_printf(seq, ",max_batch_time=%u", | |
990 | (unsigned) sbi->s_min_batch_time); | |
991 | } | |
992 | ||
571640ca ES |
993 | /* |
994 | * We're changing the default of barrier mount option, so | |
995 | * let's always display its mount state so it's clear what its | |
996 | * status is. | |
997 | */ | |
998 | seq_puts(seq, ",barrier="); | |
999 | seq_puts(seq, test_opt(sb, BARRIER) ? "1" : "0"); | |
cd0b6a39 TT |
1000 | if (test_opt(sb, JOURNAL_ASYNC_COMMIT)) |
1001 | seq_puts(seq, ",journal_async_commit"); | |
39a4bade JK |
1002 | else if (test_opt(sb, JOURNAL_CHECKSUM)) |
1003 | seq_puts(seq, ",journal_checksum"); | |
25ec56b5 JNC |
1004 | if (test_opt(sb, I_VERSION)) |
1005 | seq_puts(seq, ",i_version"); | |
dd919b98 AK |
1006 | if (!test_opt(sb, DELALLOC)) |
1007 | seq_puts(seq, ",nodelalloc"); | |
1008 | ||
ac27a0ec | 1009 | |
cb45bbe4 MS |
1010 | if (sbi->s_stripe) |
1011 | seq_printf(seq, ",stripe=%lu", sbi->s_stripe); | |
aa22df2c AK |
1012 | /* |
1013 | * journal mode get enabled in different ways | |
1014 | * So just print the value even if we didn't specify it | |
1015 | */ | |
617ba13b | 1016 | if (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_JOURNAL_DATA) |
ac27a0ec | 1017 | seq_puts(seq, ",data=journal"); |
617ba13b | 1018 | else if (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_ORDERED_DATA) |
ac27a0ec | 1019 | seq_puts(seq, ",data=ordered"); |
617ba13b | 1020 | else if (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_WRITEBACK_DATA) |
ac27a0ec DK |
1021 | seq_puts(seq, ",data=writeback"); |
1022 | ||
240799cd TT |
1023 | if (sbi->s_inode_readahead_blks != EXT4_DEF_INODE_READAHEAD_BLKS) |
1024 | seq_printf(seq, ",inode_readahead_blks=%u", | |
1025 | sbi->s_inode_readahead_blks); | |
1026 | ||
5bf5683a HK |
1027 | if (test_opt(sb, DATA_ERR_ABORT)) |
1028 | seq_puts(seq, ",data_err=abort"); | |
1029 | ||
afd4672d | 1030 | if (test_opt(sb, NO_AUTO_DA_ALLOC)) |
06705bff | 1031 | seq_puts(seq, ",noauto_da_alloc"); |
afd4672d | 1032 | |
5328e635 ES |
1033 | if (test_opt(sb, DISCARD)) |
1034 | seq_puts(seq, ",discard"); | |
1035 | ||
e3bb52ae ES |
1036 | if (test_opt(sb, NOLOAD)) |
1037 | seq_puts(seq, ",norecovery"); | |
1038 | ||
744692dc JZ |
1039 | if (test_opt(sb, DIOREAD_NOLOCK)) |
1040 | seq_puts(seq, ",dioread_nolock"); | |
1041 | ||
617ba13b | 1042 | ext4_show_quota_options(seq, sb); |
0b8e58a1 | 1043 | |
ac27a0ec DK |
1044 | return 0; |
1045 | } | |
1046 | ||
1b961ac0 | 1047 | static struct inode *ext4_nfs_get_inode(struct super_block *sb, |
0b8e58a1 | 1048 | u64 ino, u32 generation) |
ac27a0ec | 1049 | { |
ac27a0ec | 1050 | struct inode *inode; |
ac27a0ec | 1051 | |
617ba13b | 1052 | if (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO) |
ac27a0ec | 1053 | return ERR_PTR(-ESTALE); |
617ba13b | 1054 | if (ino > le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count)) |
ac27a0ec DK |
1055 | return ERR_PTR(-ESTALE); |
1056 | ||
1057 | /* iget isn't really right if the inode is currently unallocated!! | |
1058 | * | |
617ba13b | 1059 | * ext4_read_inode will return a bad_inode if the inode had been |
ac27a0ec DK |
1060 | * deleted, so we should be safe. |
1061 | * | |
1062 | * Currently we don't know the generation for parent directory, so | |
1063 | * a generation of 0 means "accept any" | |
1064 | */ | |
1d1fe1ee DH |
1065 | inode = ext4_iget(sb, ino); |
1066 | if (IS_ERR(inode)) | |
1067 | return ERR_CAST(inode); | |
1068 | if (generation && inode->i_generation != generation) { | |
ac27a0ec DK |
1069 | iput(inode); |
1070 | return ERR_PTR(-ESTALE); | |
1071 | } | |
1b961ac0 CH |
1072 | |
1073 | return inode; | |
1074 | } | |
1075 | ||
1076 | static struct dentry *ext4_fh_to_dentry(struct super_block *sb, struct fid *fid, | |
0b8e58a1 | 1077 | int fh_len, int fh_type) |
1b961ac0 CH |
1078 | { |
1079 | return generic_fh_to_dentry(sb, fid, fh_len, fh_type, | |
1080 | ext4_nfs_get_inode); | |
1081 | } | |
1082 | ||
1083 | static struct dentry *ext4_fh_to_parent(struct super_block *sb, struct fid *fid, | |
0b8e58a1 | 1084 | int fh_len, int fh_type) |
1b961ac0 CH |
1085 | { |
1086 | return generic_fh_to_parent(sb, fid, fh_len, fh_type, | |
1087 | ext4_nfs_get_inode); | |
ac27a0ec DK |
1088 | } |
1089 | ||
c39a7f84 TO |
1090 | /* |
1091 | * Try to release metadata pages (indirect blocks, directories) which are | |
1092 | * mapped via the block device. Since these pages could have journal heads | |
1093 | * which would prevent try_to_free_buffers() from freeing them, we must use | |
1094 | * jbd2 layer's try_to_free_buffers() function to release them. | |
1095 | */ | |
0b8e58a1 AD |
1096 | static int bdev_try_to_free_page(struct super_block *sb, struct page *page, |
1097 | gfp_t wait) | |
c39a7f84 TO |
1098 | { |
1099 | journal_t *journal = EXT4_SB(sb)->s_journal; | |
1100 | ||
1101 | WARN_ON(PageChecked(page)); | |
1102 | if (!page_has_buffers(page)) | |
1103 | return 0; | |
1104 | if (journal) | |
1105 | return jbd2_journal_try_to_free_buffers(journal, page, | |
1106 | wait & ~__GFP_WAIT); | |
1107 | return try_to_free_buffers(page); | |
1108 | } | |
1109 | ||
ac27a0ec | 1110 | #ifdef CONFIG_QUOTA |
af5bc92d | 1111 | #define QTYPE2NAME(t) ((t) == USRQUOTA ? "user" : "group") |
2b2d6d01 | 1112 | #define QTYPE2MOPT(on, t) ((t) == USRQUOTA?((on)##USRJQUOTA):((on)##GRPJQUOTA)) |
ac27a0ec | 1113 | |
617ba13b MC |
1114 | static int ext4_write_dquot(struct dquot *dquot); |
1115 | static int ext4_acquire_dquot(struct dquot *dquot); | |
1116 | static int ext4_release_dquot(struct dquot *dquot); | |
1117 | static int ext4_mark_dquot_dirty(struct dquot *dquot); | |
1118 | static int ext4_write_info(struct super_block *sb, int type); | |
6f28e087 | 1119 | static int ext4_quota_on(struct super_block *sb, int type, int format_id, |
307ae18a | 1120 | char *path); |
617ba13b MC |
1121 | static int ext4_quota_on_mount(struct super_block *sb, int type); |
1122 | static ssize_t ext4_quota_read(struct super_block *sb, int type, char *data, | |
ac27a0ec | 1123 | size_t len, loff_t off); |
617ba13b | 1124 | static ssize_t ext4_quota_write(struct super_block *sb, int type, |
ac27a0ec DK |
1125 | const char *data, size_t len, loff_t off); |
1126 | ||
61e225dc | 1127 | static const struct dquot_operations ext4_quota_operations = { |
a9e7f447 | 1128 | #ifdef CONFIG_QUOTA |
60e58e0f | 1129 | .get_reserved_space = ext4_get_reserved_space, |
a9e7f447 | 1130 | #endif |
617ba13b MC |
1131 | .write_dquot = ext4_write_dquot, |
1132 | .acquire_dquot = ext4_acquire_dquot, | |
1133 | .release_dquot = ext4_release_dquot, | |
1134 | .mark_dirty = ext4_mark_dquot_dirty, | |
a5b5ee32 JK |
1135 | .write_info = ext4_write_info, |
1136 | .alloc_dquot = dquot_alloc, | |
1137 | .destroy_dquot = dquot_destroy, | |
ac27a0ec DK |
1138 | }; |
1139 | ||
0d54b217 | 1140 | static const struct quotactl_ops ext4_qctl_operations = { |
617ba13b | 1141 | .quota_on = ext4_quota_on, |
287a8095 CH |
1142 | .quota_off = dquot_quota_off, |
1143 | .quota_sync = dquot_quota_sync, | |
1144 | .get_info = dquot_get_dqinfo, | |
1145 | .set_info = dquot_set_dqinfo, | |
1146 | .get_dqblk = dquot_get_dqblk, | |
1147 | .set_dqblk = dquot_set_dqblk | |
ac27a0ec DK |
1148 | }; |
1149 | #endif | |
1150 | ||
ee9b6d61 | 1151 | static const struct super_operations ext4_sops = { |
617ba13b MC |
1152 | .alloc_inode = ext4_alloc_inode, |
1153 | .destroy_inode = ext4_destroy_inode, | |
617ba13b MC |
1154 | .write_inode = ext4_write_inode, |
1155 | .dirty_inode = ext4_dirty_inode, | |
1156 | .delete_inode = ext4_delete_inode, | |
1157 | .put_super = ext4_put_super, | |
617ba13b | 1158 | .sync_fs = ext4_sync_fs, |
c4be0c1d TS |
1159 | .freeze_fs = ext4_freeze, |
1160 | .unfreeze_fs = ext4_unfreeze, | |
617ba13b MC |
1161 | .statfs = ext4_statfs, |
1162 | .remount_fs = ext4_remount, | |
1163 | .clear_inode = ext4_clear_inode, | |
1164 | .show_options = ext4_show_options, | |
ac27a0ec | 1165 | #ifdef CONFIG_QUOTA |
617ba13b MC |
1166 | .quota_read = ext4_quota_read, |
1167 | .quota_write = ext4_quota_write, | |
ac27a0ec | 1168 | #endif |
c39a7f84 | 1169 | .bdev_try_to_free_page = bdev_try_to_free_page, |
ac27a0ec DK |
1170 | }; |
1171 | ||
9ca92389 TT |
1172 | static const struct super_operations ext4_nojournal_sops = { |
1173 | .alloc_inode = ext4_alloc_inode, | |
1174 | .destroy_inode = ext4_destroy_inode, | |
1175 | .write_inode = ext4_write_inode, | |
1176 | .dirty_inode = ext4_dirty_inode, | |
1177 | .delete_inode = ext4_delete_inode, | |
1178 | .write_super = ext4_write_super, | |
1179 | .put_super = ext4_put_super, | |
1180 | .statfs = ext4_statfs, | |
1181 | .remount_fs = ext4_remount, | |
1182 | .clear_inode = ext4_clear_inode, | |
1183 | .show_options = ext4_show_options, | |
1184 | #ifdef CONFIG_QUOTA | |
1185 | .quota_read = ext4_quota_read, | |
1186 | .quota_write = ext4_quota_write, | |
1187 | #endif | |
1188 | .bdev_try_to_free_page = bdev_try_to_free_page, | |
1189 | }; | |
1190 | ||
39655164 | 1191 | static const struct export_operations ext4_export_ops = { |
1b961ac0 CH |
1192 | .fh_to_dentry = ext4_fh_to_dentry, |
1193 | .fh_to_parent = ext4_fh_to_parent, | |
617ba13b | 1194 | .get_parent = ext4_get_parent, |
ac27a0ec DK |
1195 | }; |
1196 | ||
1197 | enum { | |
1198 | Opt_bsd_df, Opt_minix_df, Opt_grpid, Opt_nogrpid, | |
1199 | Opt_resgid, Opt_resuid, Opt_sb, Opt_err_cont, Opt_err_panic, Opt_err_ro, | |
01436ef2 | 1200 | Opt_nouid32, Opt_debug, Opt_oldalloc, Opt_orlov, |
ac27a0ec | 1201 | Opt_user_xattr, Opt_nouser_xattr, Opt_acl, Opt_noacl, |
06705bff | 1202 | Opt_auto_da_alloc, Opt_noauto_da_alloc, Opt_noload, Opt_nobh, Opt_bh, |
30773840 | 1203 | Opt_commit, Opt_min_batch_time, Opt_max_batch_time, |
c3191067 | 1204 | Opt_journal_update, Opt_journal_dev, |
818d276c | 1205 | Opt_journal_checksum, Opt_journal_async_commit, |
ac27a0ec | 1206 | Opt_abort, Opt_data_journal, Opt_data_ordered, Opt_data_writeback, |
296c355c | 1207 | Opt_data_err_abort, Opt_data_err_ignore, |
ac27a0ec | 1208 | Opt_usrjquota, Opt_grpjquota, Opt_offusrjquota, Opt_offgrpjquota, |
5a20bdfc JK |
1209 | Opt_jqfmt_vfsold, Opt_jqfmt_vfsv0, Opt_jqfmt_vfsv1, Opt_quota, |
1210 | Opt_noquota, Opt_ignore, Opt_barrier, Opt_nobarrier, Opt_err, | |
1211 | Opt_resize, Opt_usrquota, Opt_grpquota, Opt_i_version, | |
01436ef2 | 1212 | Opt_stripe, Opt_delalloc, Opt_nodelalloc, |
6fd058f7 | 1213 | Opt_block_validity, Opt_noblock_validity, |
5328e635 | 1214 | Opt_inode_readahead_blks, Opt_journal_ioprio, |
744692dc | 1215 | Opt_dioread_nolock, Opt_dioread_lock, |
5328e635 | 1216 | Opt_discard, Opt_nodiscard, |
ac27a0ec DK |
1217 | }; |
1218 | ||
a447c093 | 1219 | static const match_table_t tokens = { |
ac27a0ec DK |
1220 | {Opt_bsd_df, "bsddf"}, |
1221 | {Opt_minix_df, "minixdf"}, | |
1222 | {Opt_grpid, "grpid"}, | |
1223 | {Opt_grpid, "bsdgroups"}, | |
1224 | {Opt_nogrpid, "nogrpid"}, | |
1225 | {Opt_nogrpid, "sysvgroups"}, | |
1226 | {Opt_resgid, "resgid=%u"}, | |
1227 | {Opt_resuid, "resuid=%u"}, | |
1228 | {Opt_sb, "sb=%u"}, | |
1229 | {Opt_err_cont, "errors=continue"}, | |
1230 | {Opt_err_panic, "errors=panic"}, | |
1231 | {Opt_err_ro, "errors=remount-ro"}, | |
1232 | {Opt_nouid32, "nouid32"}, | |
ac27a0ec DK |
1233 | {Opt_debug, "debug"}, |
1234 | {Opt_oldalloc, "oldalloc"}, | |
1235 | {Opt_orlov, "orlov"}, | |
1236 | {Opt_user_xattr, "user_xattr"}, | |
1237 | {Opt_nouser_xattr, "nouser_xattr"}, | |
1238 | {Opt_acl, "acl"}, | |
1239 | {Opt_noacl, "noacl"}, | |
ac27a0ec | 1240 | {Opt_noload, "noload"}, |
e3bb52ae | 1241 | {Opt_noload, "norecovery"}, |
ac27a0ec DK |
1242 | {Opt_nobh, "nobh"}, |
1243 | {Opt_bh, "bh"}, | |
1244 | {Opt_commit, "commit=%u"}, | |
30773840 TT |
1245 | {Opt_min_batch_time, "min_batch_time=%u"}, |
1246 | {Opt_max_batch_time, "max_batch_time=%u"}, | |
ac27a0ec | 1247 | {Opt_journal_update, "journal=update"}, |
ac27a0ec | 1248 | {Opt_journal_dev, "journal_dev=%u"}, |
818d276c GS |
1249 | {Opt_journal_checksum, "journal_checksum"}, |
1250 | {Opt_journal_async_commit, "journal_async_commit"}, | |
ac27a0ec DK |
1251 | {Opt_abort, "abort"}, |
1252 | {Opt_data_journal, "data=journal"}, | |
1253 | {Opt_data_ordered, "data=ordered"}, | |
1254 | {Opt_data_writeback, "data=writeback"}, | |
5bf5683a HK |
1255 | {Opt_data_err_abort, "data_err=abort"}, |
1256 | {Opt_data_err_ignore, "data_err=ignore"}, | |
ac27a0ec DK |
1257 | {Opt_offusrjquota, "usrjquota="}, |
1258 | {Opt_usrjquota, "usrjquota=%s"}, | |
1259 | {Opt_offgrpjquota, "grpjquota="}, | |
1260 | {Opt_grpjquota, "grpjquota=%s"}, | |
1261 | {Opt_jqfmt_vfsold, "jqfmt=vfsold"}, | |
1262 | {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"}, | |
5a20bdfc | 1263 | {Opt_jqfmt_vfsv1, "jqfmt=vfsv1"}, |
ac27a0ec DK |
1264 | {Opt_grpquota, "grpquota"}, |
1265 | {Opt_noquota, "noquota"}, | |
1266 | {Opt_quota, "quota"}, | |
1267 | {Opt_usrquota, "usrquota"}, | |
1268 | {Opt_barrier, "barrier=%u"}, | |
06705bff TT |
1269 | {Opt_barrier, "barrier"}, |
1270 | {Opt_nobarrier, "nobarrier"}, | |
25ec56b5 | 1271 | {Opt_i_version, "i_version"}, |
c9de560d | 1272 | {Opt_stripe, "stripe=%u"}, |
ac27a0ec | 1273 | {Opt_resize, "resize"}, |
64769240 | 1274 | {Opt_delalloc, "delalloc"}, |
dd919b98 | 1275 | {Opt_nodelalloc, "nodelalloc"}, |
6fd058f7 TT |
1276 | {Opt_block_validity, "block_validity"}, |
1277 | {Opt_noblock_validity, "noblock_validity"}, | |
240799cd | 1278 | {Opt_inode_readahead_blks, "inode_readahead_blks=%u"}, |
b3881f74 | 1279 | {Opt_journal_ioprio, "journal_ioprio=%u"}, |
afd4672d | 1280 | {Opt_auto_da_alloc, "auto_da_alloc=%u"}, |
06705bff TT |
1281 | {Opt_auto_da_alloc, "auto_da_alloc"}, |
1282 | {Opt_noauto_da_alloc, "noauto_da_alloc"}, | |
744692dc JZ |
1283 | {Opt_dioread_nolock, "dioread_nolock"}, |
1284 | {Opt_dioread_lock, "dioread_lock"}, | |
5328e635 ES |
1285 | {Opt_discard, "discard"}, |
1286 | {Opt_nodiscard, "nodiscard"}, | |
f3f12faa | 1287 | {Opt_err, NULL}, |
ac27a0ec DK |
1288 | }; |
1289 | ||
617ba13b | 1290 | static ext4_fsblk_t get_sb_block(void **data) |
ac27a0ec | 1291 | { |
617ba13b | 1292 | ext4_fsblk_t sb_block; |
ac27a0ec DK |
1293 | char *options = (char *) *data; |
1294 | ||
1295 | if (!options || strncmp(options, "sb=", 3) != 0) | |
1296 | return 1; /* Default location */ | |
0b8e58a1 | 1297 | |
ac27a0ec | 1298 | options += 3; |
0b8e58a1 | 1299 | /* TODO: use simple_strtoll with >32bit ext4 */ |
ac27a0ec DK |
1300 | sb_block = simple_strtoul(options, &options, 0); |
1301 | if (*options && *options != ',') { | |
4776004f | 1302 | printk(KERN_ERR "EXT4-fs: Invalid sb specification: %s\n", |
ac27a0ec DK |
1303 | (char *) *data); |
1304 | return 1; | |
1305 | } | |
1306 | if (*options == ',') | |
1307 | options++; | |
1308 | *data = (void *) options; | |
0b8e58a1 | 1309 | |
ac27a0ec DK |
1310 | return sb_block; |
1311 | } | |
1312 | ||
b3881f74 | 1313 | #define DEFAULT_JOURNAL_IOPRIO (IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, 3)) |
437ca0fd DM |
1314 | static char deprecated_msg[] = "Mount option \"%s\" will be removed by %s\n" |
1315 | "Contact linux-ext4@vger.kernel.org if you think we should keep it.\n"; | |
b3881f74 | 1316 | |
56c50f11 DM |
1317 | #ifdef CONFIG_QUOTA |
1318 | static int set_qf_name(struct super_block *sb, int qtype, substring_t *args) | |
1319 | { | |
1320 | struct ext4_sb_info *sbi = EXT4_SB(sb); | |
1321 | char *qname; | |
1322 | ||
1323 | if (sb_any_quota_loaded(sb) && | |
1324 | !sbi->s_qf_names[qtype]) { | |
1325 | ext4_msg(sb, KERN_ERR, | |
1326 | "Cannot change journaled " | |
1327 | "quota options when quota turned on"); | |
1328 | return 0; | |
1329 | } | |
1330 | qname = match_strdup(args); | |
1331 | if (!qname) { | |
1332 | ext4_msg(sb, KERN_ERR, | |
1333 | "Not enough memory for storing quotafile name"); | |
1334 | return 0; | |
1335 | } | |
1336 | if (sbi->s_qf_names[qtype] && | |
1337 | strcmp(sbi->s_qf_names[qtype], qname)) { | |
1338 | ext4_msg(sb, KERN_ERR, | |
1339 | "%s quota file already specified", QTYPE2NAME(qtype)); | |
1340 | kfree(qname); | |
1341 | return 0; | |
1342 | } | |
1343 | sbi->s_qf_names[qtype] = qname; | |
1344 | if (strchr(sbi->s_qf_names[qtype], '/')) { | |
1345 | ext4_msg(sb, KERN_ERR, | |
1346 | "quotafile must be on filesystem root"); | |
1347 | kfree(sbi->s_qf_names[qtype]); | |
1348 | sbi->s_qf_names[qtype] = NULL; | |
1349 | return 0; | |
1350 | } | |
1351 | set_opt(sbi->s_mount_opt, QUOTA); | |
1352 | return 1; | |
1353 | } | |
1354 | ||
1355 | static int clear_qf_name(struct super_block *sb, int qtype) | |
1356 | { | |
1357 | ||
1358 | struct ext4_sb_info *sbi = EXT4_SB(sb); | |
1359 | ||
1360 | if (sb_any_quota_loaded(sb) && | |
1361 | sbi->s_qf_names[qtype]) { | |
1362 | ext4_msg(sb, KERN_ERR, "Cannot change journaled quota options" | |
1363 | " when quota turned on"); | |
1364 | return 0; | |
1365 | } | |
1366 | /* | |
1367 | * The space will be released later when all options are confirmed | |
1368 | * to be correct | |
1369 | */ | |
1370 | sbi->s_qf_names[qtype] = NULL; | |
1371 | return 1; | |
1372 | } | |
1373 | #endif | |
1374 | ||
2b2d6d01 | 1375 | static int parse_options(char *options, struct super_block *sb, |
c3191067 | 1376 | unsigned long *journal_devnum, |
b3881f74 | 1377 | unsigned int *journal_ioprio, |
2b2d6d01 | 1378 | ext4_fsblk_t *n_blocks_count, int is_remount) |
ac27a0ec | 1379 | { |
617ba13b | 1380 | struct ext4_sb_info *sbi = EXT4_SB(sb); |
2b2d6d01 | 1381 | char *p; |
ac27a0ec DK |
1382 | substring_t args[MAX_OPT_ARGS]; |
1383 | int data_opt = 0; | |
1384 | int option; | |
1385 | #ifdef CONFIG_QUOTA | |
56c50f11 | 1386 | int qfmt; |
ac27a0ec DK |
1387 | #endif |
1388 | ||
1389 | if (!options) | |
1390 | return 1; | |
1391 | ||
2b2d6d01 | 1392 | while ((p = strsep(&options, ",")) != NULL) { |
ac27a0ec DK |
1393 | int token; |
1394 | if (!*p) | |
1395 | continue; | |
1396 | ||
15121c18 ES |
1397 | /* |
1398 | * Initialize args struct so we know whether arg was | |
1399 | * found; some options take optional arguments. | |
1400 | */ | |
1401 | args[0].to = args[0].from = 0; | |
ac27a0ec DK |
1402 | token = match_token(p, tokens, args); |
1403 | switch (token) { | |
1404 | case Opt_bsd_df: | |
437ca0fd | 1405 | ext4_msg(sb, KERN_WARNING, deprecated_msg, p, "2.6.38"); |
2b2d6d01 | 1406 | clear_opt(sbi->s_mount_opt, MINIX_DF); |
ac27a0ec DK |
1407 | break; |
1408 | case Opt_minix_df: | |
437ca0fd | 1409 | ext4_msg(sb, KERN_WARNING, deprecated_msg, p, "2.6.38"); |
2b2d6d01 | 1410 | set_opt(sbi->s_mount_opt, MINIX_DF); |
437ca0fd | 1411 | |
ac27a0ec DK |
1412 | break; |
1413 | case Opt_grpid: | |
437ca0fd | 1414 | ext4_msg(sb, KERN_WARNING, deprecated_msg, p, "2.6.38"); |
2b2d6d01 | 1415 | set_opt(sbi->s_mount_opt, GRPID); |
437ca0fd | 1416 | |
ac27a0ec DK |
1417 | break; |
1418 | case Opt_nogrpid: | |
437ca0fd | 1419 | ext4_msg(sb, KERN_WARNING, deprecated_msg, p, "2.6.38"); |
2b2d6d01 | 1420 | clear_opt(sbi->s_mount_opt, GRPID); |
437ca0fd | 1421 | |
ac27a0ec DK |
1422 | break; |
1423 | case Opt_resuid: | |
1424 | if (match_int(&args[0], &option)) | |
1425 | return 0; | |
1426 | sbi->s_resuid = option; | |
1427 | break; | |
1428 | case Opt_resgid: | |
1429 | if (match_int(&args[0], &option)) | |
1430 | return 0; | |
1431 | sbi->s_resgid = option; | |
1432 | break; | |
1433 | case Opt_sb: | |
1434 | /* handled by get_sb_block() instead of here */ | |
1435 | /* *sb_block = match_int(&args[0]); */ | |
1436 | break; | |
1437 | case Opt_err_panic: | |
2b2d6d01 TT |
1438 | clear_opt(sbi->s_mount_opt, ERRORS_CONT); |
1439 | clear_opt(sbi->s_mount_opt, ERRORS_RO); | |
1440 | set_opt(sbi->s_mount_opt, ERRORS_PANIC); | |
ac27a0ec DK |
1441 | break; |
1442 | case Opt_err_ro: | |
2b2d6d01 TT |
1443 | clear_opt(sbi->s_mount_opt, ERRORS_CONT); |
1444 | clear_opt(sbi->s_mount_opt, ERRORS_PANIC); | |
1445 | set_opt(sbi->s_mount_opt, ERRORS_RO); | |
ac27a0ec DK |
1446 | break; |
1447 | case Opt_err_cont: | |
2b2d6d01 TT |
1448 | clear_opt(sbi->s_mount_opt, ERRORS_RO); |
1449 | clear_opt(sbi->s_mount_opt, ERRORS_PANIC); | |
1450 | set_opt(sbi->s_mount_opt, ERRORS_CONT); | |
ac27a0ec DK |
1451 | break; |
1452 | case Opt_nouid32: | |
2b2d6d01 | 1453 | set_opt(sbi->s_mount_opt, NO_UID32); |
ac27a0ec | 1454 | break; |
ac27a0ec | 1455 | case Opt_debug: |
2b2d6d01 | 1456 | set_opt(sbi->s_mount_opt, DEBUG); |
ac27a0ec DK |
1457 | break; |
1458 | case Opt_oldalloc: | |
2b2d6d01 | 1459 | set_opt(sbi->s_mount_opt, OLDALLOC); |
ac27a0ec DK |
1460 | break; |
1461 | case Opt_orlov: | |
2b2d6d01 | 1462 | clear_opt(sbi->s_mount_opt, OLDALLOC); |
ac27a0ec | 1463 | break; |
03010a33 | 1464 | #ifdef CONFIG_EXT4_FS_XATTR |
ac27a0ec | 1465 | case Opt_user_xattr: |
2b2d6d01 | 1466 | set_opt(sbi->s_mount_opt, XATTR_USER); |
ac27a0ec DK |
1467 | break; |
1468 | case Opt_nouser_xattr: | |
2b2d6d01 | 1469 | clear_opt(sbi->s_mount_opt, XATTR_USER); |
ac27a0ec DK |
1470 | break; |
1471 | #else | |
1472 | case Opt_user_xattr: | |
1473 | case Opt_nouser_xattr: | |
b31e1552 | 1474 | ext4_msg(sb, KERN_ERR, "(no)user_xattr options not supported"); |
ac27a0ec DK |
1475 | break; |
1476 | #endif | |
03010a33 | 1477 | #ifdef CONFIG_EXT4_FS_POSIX_ACL |
ac27a0ec DK |
1478 | case Opt_acl: |
1479 | set_opt(sbi->s_mount_opt, POSIX_ACL); | |
1480 | break; | |
1481 | case Opt_noacl: | |
1482 | clear_opt(sbi->s_mount_opt, POSIX_ACL); | |
1483 | break; | |
1484 | #else | |
1485 | case Opt_acl: | |
1486 | case Opt_noacl: | |
b31e1552 | 1487 | ext4_msg(sb, KERN_ERR, "(no)acl options not supported"); |
ac27a0ec DK |
1488 | break; |
1489 | #endif | |
ac27a0ec DK |
1490 | case Opt_journal_update: |
1491 | /* @@@ FIXME */ | |
1492 | /* Eventually we will want to be able to create | |
1493 | a journal file here. For now, only allow the | |
1494 | user to specify an existing inode to be the | |
1495 | journal file. */ | |
1496 | if (is_remount) { | |
b31e1552 ES |
1497 | ext4_msg(sb, KERN_ERR, |
1498 | "Cannot specify journal on remount"); | |
ac27a0ec DK |
1499 | return 0; |
1500 | } | |
2b2d6d01 | 1501 | set_opt(sbi->s_mount_opt, UPDATE_JOURNAL); |
ac27a0ec | 1502 | break; |
ac27a0ec DK |
1503 | case Opt_journal_dev: |
1504 | if (is_remount) { | |
b31e1552 ES |
1505 | ext4_msg(sb, KERN_ERR, |
1506 | "Cannot specify journal on remount"); | |
ac27a0ec DK |
1507 | return 0; |
1508 | } | |
1509 | if (match_int(&args[0], &option)) | |
1510 | return 0; | |
1511 | *journal_devnum = option; | |
1512 | break; | |
818d276c | 1513 | case Opt_journal_checksum: |
d4da6c9c LT |
1514 | set_opt(sbi->s_mount_opt, JOURNAL_CHECKSUM); |
1515 | break; | |
818d276c GS |
1516 | case Opt_journal_async_commit: |
1517 | set_opt(sbi->s_mount_opt, JOURNAL_ASYNC_COMMIT); | |
d4da6c9c | 1518 | set_opt(sbi->s_mount_opt, JOURNAL_CHECKSUM); |
818d276c | 1519 | break; |
ac27a0ec | 1520 | case Opt_noload: |
2b2d6d01 | 1521 | set_opt(sbi->s_mount_opt, NOLOAD); |
ac27a0ec DK |
1522 | break; |
1523 | case Opt_commit: | |
1524 | if (match_int(&args[0], &option)) | |
1525 | return 0; | |
1526 | if (option < 0) | |
1527 | return 0; | |
1528 | if (option == 0) | |
cd02ff0b | 1529 | option = JBD2_DEFAULT_MAX_COMMIT_AGE; |
ac27a0ec DK |
1530 | sbi->s_commit_interval = HZ * option; |
1531 | break; | |
30773840 TT |
1532 | case Opt_max_batch_time: |
1533 | if (match_int(&args[0], &option)) | |
1534 | return 0; | |
1535 | if (option < 0) | |
1536 | return 0; | |
1537 | if (option == 0) | |
1538 | option = EXT4_DEF_MAX_BATCH_TIME; | |
1539 | sbi->s_max_batch_time = option; | |
1540 | break; | |
1541 | case Opt_min_batch_time: | |
1542 | if (match_int(&args[0], &option)) | |
1543 | return 0; | |
1544 | if (option < 0) | |
1545 | return 0; | |
1546 | sbi->s_min_batch_time = option; | |
1547 | break; | |
ac27a0ec | 1548 | case Opt_data_journal: |
617ba13b | 1549 | data_opt = EXT4_MOUNT_JOURNAL_DATA; |
ac27a0ec DK |
1550 | goto datacheck; |
1551 | case Opt_data_ordered: | |
617ba13b | 1552 | data_opt = EXT4_MOUNT_ORDERED_DATA; |
ac27a0ec DK |
1553 | goto datacheck; |
1554 | case Opt_data_writeback: | |
617ba13b | 1555 | data_opt = EXT4_MOUNT_WRITEBACK_DATA; |
ac27a0ec DK |
1556 | datacheck: |
1557 | if (is_remount) { | |
482a7425 | 1558 | if (test_opt(sb, DATA_FLAGS) != data_opt) { |
b31e1552 ES |
1559 | ext4_msg(sb, KERN_ERR, |
1560 | "Cannot change data mode on remount"); | |
ac27a0ec DK |
1561 | return 0; |
1562 | } | |
1563 | } else { | |
482a7425 | 1564 | clear_opt(sbi->s_mount_opt, DATA_FLAGS); |
ac27a0ec DK |
1565 | sbi->s_mount_opt |= data_opt; |
1566 | } | |
1567 | break; | |
5bf5683a HK |
1568 | case Opt_data_err_abort: |
1569 | set_opt(sbi->s_mount_opt, DATA_ERR_ABORT); | |
1570 | break; | |
1571 | case Opt_data_err_ignore: | |
1572 | clear_opt(sbi->s_mount_opt, DATA_ERR_ABORT); | |
1573 | break; | |
ac27a0ec DK |
1574 | #ifdef CONFIG_QUOTA |
1575 | case Opt_usrjquota: | |
56c50f11 | 1576 | if (!set_qf_name(sb, USRQUOTA, &args[0])) |
ac27a0ec | 1577 | return 0; |
56c50f11 DM |
1578 | break; |
1579 | case Opt_grpjquota: | |
1580 | if (!set_qf_name(sb, GRPQUOTA, &args[0])) | |
ac27a0ec | 1581 | return 0; |
ac27a0ec DK |
1582 | break; |
1583 | case Opt_offusrjquota: | |
56c50f11 DM |
1584 | if (!clear_qf_name(sb, USRQUOTA)) |
1585 | return 0; | |
1586 | break; | |
ac27a0ec | 1587 | case Opt_offgrpjquota: |
56c50f11 | 1588 | if (!clear_qf_name(sb, GRPQUOTA)) |
ac27a0ec | 1589 | return 0; |
ac27a0ec | 1590 | break; |
56c50f11 | 1591 | |
ac27a0ec | 1592 | case Opt_jqfmt_vfsold: |
dfc5d03f JK |
1593 | qfmt = QFMT_VFS_OLD; |
1594 | goto set_qf_format; | |
ac27a0ec | 1595 | case Opt_jqfmt_vfsv0: |
dfc5d03f | 1596 | qfmt = QFMT_VFS_V0; |
5a20bdfc JK |
1597 | goto set_qf_format; |
1598 | case Opt_jqfmt_vfsv1: | |
1599 | qfmt = QFMT_VFS_V1; | |
dfc5d03f | 1600 | set_qf_format: |
17bd13b3 | 1601 | if (sb_any_quota_loaded(sb) && |
dfc5d03f | 1602 | sbi->s_jquota_fmt != qfmt) { |
b31e1552 | 1603 | ext4_msg(sb, KERN_ERR, "Cannot change " |
dfc5d03f | 1604 | "journaled quota options when " |
b31e1552 | 1605 | "quota turned on"); |
dfc5d03f JK |
1606 | return 0; |
1607 | } | |
1608 | sbi->s_jquota_fmt = qfmt; | |
ac27a0ec DK |
1609 | break; |
1610 | case Opt_quota: | |
1611 | case Opt_usrquota: | |
1612 | set_opt(sbi->s_mount_opt, QUOTA); | |
1613 | set_opt(sbi->s_mount_opt, USRQUOTA); | |
1614 | break; | |
1615 | case Opt_grpquota: | |
1616 | set_opt(sbi->s_mount_opt, QUOTA); | |
1617 | set_opt(sbi->s_mount_opt, GRPQUOTA); | |
1618 | break; | |
1619 | case Opt_noquota: | |
17bd13b3 | 1620 | if (sb_any_quota_loaded(sb)) { |
b31e1552 ES |
1621 | ext4_msg(sb, KERN_ERR, "Cannot change quota " |
1622 | "options when quota turned on"); | |
ac27a0ec DK |
1623 | return 0; |
1624 | } | |
1625 | clear_opt(sbi->s_mount_opt, QUOTA); | |
1626 | clear_opt(sbi->s_mount_opt, USRQUOTA); | |
1627 | clear_opt(sbi->s_mount_opt, GRPQUOTA); | |
1628 | break; | |
1629 | #else | |
1630 | case Opt_quota: | |
1631 | case Opt_usrquota: | |
1632 | case Opt_grpquota: | |
b31e1552 ES |
1633 | ext4_msg(sb, KERN_ERR, |
1634 | "quota options not supported"); | |
cd59e7b9 | 1635 | break; |
ac27a0ec DK |
1636 | case Opt_usrjquota: |
1637 | case Opt_grpjquota: | |
1638 | case Opt_offusrjquota: | |
1639 | case Opt_offgrpjquota: | |
1640 | case Opt_jqfmt_vfsold: | |
1641 | case Opt_jqfmt_vfsv0: | |
5a20bdfc | 1642 | case Opt_jqfmt_vfsv1: |
b31e1552 ES |
1643 | ext4_msg(sb, KERN_ERR, |
1644 | "journaled quota options not supported"); | |
ac27a0ec DK |
1645 | break; |
1646 | case Opt_noquota: | |
1647 | break; | |
1648 | #endif | |
1649 | case Opt_abort: | |
4ab2f15b | 1650 | sbi->s_mount_flags |= EXT4_MF_FS_ABORTED; |
ac27a0ec | 1651 | break; |
06705bff TT |
1652 | case Opt_nobarrier: |
1653 | clear_opt(sbi->s_mount_opt, BARRIER); | |
1654 | break; | |
ac27a0ec | 1655 | case Opt_barrier: |
15121c18 ES |
1656 | if (args[0].from) { |
1657 | if (match_int(&args[0], &option)) | |
1658 | return 0; | |
1659 | } else | |
1660 | option = 1; /* No argument, default to 1 */ | |
ac27a0ec DK |
1661 | if (option) |
1662 | set_opt(sbi->s_mount_opt, BARRIER); | |
1663 | else | |
1664 | clear_opt(sbi->s_mount_opt, BARRIER); | |
1665 | break; | |
1666 | case Opt_ignore: | |
1667 | break; | |
1668 | case Opt_resize: | |
1669 | if (!is_remount) { | |
b31e1552 ES |
1670 | ext4_msg(sb, KERN_ERR, |
1671 | "resize option only available " | |
1672 | "for remount"); | |
ac27a0ec DK |
1673 | return 0; |
1674 | } | |
1675 | if (match_int(&args[0], &option) != 0) | |
1676 | return 0; | |
1677 | *n_blocks_count = option; | |
1678 | break; | |
1679 | case Opt_nobh: | |
206f7ab4 CH |
1680 | ext4_msg(sb, KERN_WARNING, |
1681 | "Ignoring deprecated nobh option"); | |
ac27a0ec DK |
1682 | break; |
1683 | case Opt_bh: | |
206f7ab4 CH |
1684 | ext4_msg(sb, KERN_WARNING, |
1685 | "Ignoring deprecated bh option"); | |
ac27a0ec | 1686 | break; |
25ec56b5 JNC |
1687 | case Opt_i_version: |
1688 | set_opt(sbi->s_mount_opt, I_VERSION); | |
1689 | sb->s_flags |= MS_I_VERSION; | |
1690 | break; | |
dd919b98 AK |
1691 | case Opt_nodelalloc: |
1692 | clear_opt(sbi->s_mount_opt, DELALLOC); | |
1693 | break; | |
c9de560d AT |
1694 | case Opt_stripe: |
1695 | if (match_int(&args[0], &option)) | |
1696 | return 0; | |
1697 | if (option < 0) | |
1698 | return 0; | |
1699 | sbi->s_stripe = option; | |
1700 | break; | |
64769240 AT |
1701 | case Opt_delalloc: |
1702 | set_opt(sbi->s_mount_opt, DELALLOC); | |
1703 | break; | |
6fd058f7 TT |
1704 | case Opt_block_validity: |
1705 | set_opt(sbi->s_mount_opt, BLOCK_VALIDITY); | |
1706 | break; | |
1707 | case Opt_noblock_validity: | |
1708 | clear_opt(sbi->s_mount_opt, BLOCK_VALIDITY); | |
1709 | break; | |
240799cd TT |
1710 | case Opt_inode_readahead_blks: |
1711 | if (match_int(&args[0], &option)) | |
1712 | return 0; | |
1713 | if (option < 0 || option > (1 << 30)) | |
1714 | return 0; | |
f7c43950 | 1715 | if (!is_power_of_2(option)) { |
b31e1552 ES |
1716 | ext4_msg(sb, KERN_ERR, |
1717 | "EXT4-fs: inode_readahead_blks" | |
1718 | " must be a power of 2"); | |
3197ebdb TT |
1719 | return 0; |
1720 | } | |
240799cd TT |
1721 | sbi->s_inode_readahead_blks = option; |
1722 | break; | |
b3881f74 TT |
1723 | case Opt_journal_ioprio: |
1724 | if (match_int(&args[0], &option)) | |
1725 | return 0; | |
1726 | if (option < 0 || option > 7) | |
1727 | break; | |
1728 | *journal_ioprio = IOPRIO_PRIO_VALUE(IOPRIO_CLASS_BE, | |
1729 | option); | |
1730 | break; | |
06705bff TT |
1731 | case Opt_noauto_da_alloc: |
1732 | set_opt(sbi->s_mount_opt,NO_AUTO_DA_ALLOC); | |
1733 | break; | |
afd4672d | 1734 | case Opt_auto_da_alloc: |
15121c18 ES |
1735 | if (args[0].from) { |
1736 | if (match_int(&args[0], &option)) | |
1737 | return 0; | |
1738 | } else | |
1739 | option = 1; /* No argument, default to 1 */ | |
afd4672d TT |
1740 | if (option) |
1741 | clear_opt(sbi->s_mount_opt, NO_AUTO_DA_ALLOC); | |
1742 | else | |
1743 | set_opt(sbi->s_mount_opt,NO_AUTO_DA_ALLOC); | |
1744 | break; | |
5328e635 ES |
1745 | case Opt_discard: |
1746 | set_opt(sbi->s_mount_opt, DISCARD); | |
1747 | break; | |
1748 | case Opt_nodiscard: | |
1749 | clear_opt(sbi->s_mount_opt, DISCARD); | |
1750 | break; | |
744692dc JZ |
1751 | case Opt_dioread_nolock: |
1752 | set_opt(sbi->s_mount_opt, DIOREAD_NOLOCK); | |
1753 | break; | |
1754 | case Opt_dioread_lock: | |
1755 | clear_opt(sbi->s_mount_opt, DIOREAD_NOLOCK); | |
1756 | break; | |
ac27a0ec | 1757 | default: |
b31e1552 ES |
1758 | ext4_msg(sb, KERN_ERR, |
1759 | "Unrecognized mount option \"%s\" " | |
1760 | "or missing value", p); | |
ac27a0ec DK |
1761 | return 0; |
1762 | } | |
1763 | } | |
1764 | #ifdef CONFIG_QUOTA | |
1765 | if (sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA]) { | |
482a7425 | 1766 | if (test_opt(sb, USRQUOTA) && sbi->s_qf_names[USRQUOTA]) |
ac27a0ec DK |
1767 | clear_opt(sbi->s_mount_opt, USRQUOTA); |
1768 | ||
482a7425 | 1769 | if (test_opt(sb, GRPQUOTA) && sbi->s_qf_names[GRPQUOTA]) |
ac27a0ec DK |
1770 | clear_opt(sbi->s_mount_opt, GRPQUOTA); |
1771 | ||
56c50f11 | 1772 | if (test_opt(sb, GRPQUOTA) || test_opt(sb, USRQUOTA)) { |
b31e1552 ES |
1773 | ext4_msg(sb, KERN_ERR, "old and new quota " |
1774 | "format mixing"); | |
ac27a0ec DK |
1775 | return 0; |
1776 | } | |
1777 | ||
1778 | if (!sbi->s_jquota_fmt) { | |
b31e1552 ES |
1779 | ext4_msg(sb, KERN_ERR, "journaled quota format " |
1780 | "not specified"); | |
ac27a0ec DK |
1781 | return 0; |
1782 | } | |
1783 | } else { | |
1784 | if (sbi->s_jquota_fmt) { | |
b31e1552 | 1785 | ext4_msg(sb, KERN_ERR, "journaled quota format " |
2c8be6b2 | 1786 | "specified with no journaling " |
b31e1552 | 1787 | "enabled"); |
ac27a0ec DK |
1788 | return 0; |
1789 | } | |
1790 | } | |
1791 | #endif | |
1792 | return 1; | |
1793 | } | |
1794 | ||
617ba13b | 1795 | static int ext4_setup_super(struct super_block *sb, struct ext4_super_block *es, |
ac27a0ec DK |
1796 | int read_only) |
1797 | { | |
617ba13b | 1798 | struct ext4_sb_info *sbi = EXT4_SB(sb); |
ac27a0ec DK |
1799 | int res = 0; |
1800 | ||
617ba13b | 1801 | if (le32_to_cpu(es->s_rev_level) > EXT4_MAX_SUPP_REV) { |
b31e1552 ES |
1802 | ext4_msg(sb, KERN_ERR, "revision level too high, " |
1803 | "forcing read-only mode"); | |
ac27a0ec DK |
1804 | res = MS_RDONLY; |
1805 | } | |
1806 | if (read_only) | |
1807 | return res; | |
617ba13b | 1808 | if (!(sbi->s_mount_state & EXT4_VALID_FS)) |
b31e1552 ES |
1809 | ext4_msg(sb, KERN_WARNING, "warning: mounting unchecked fs, " |
1810 | "running e2fsck is recommended"); | |
617ba13b | 1811 | else if ((sbi->s_mount_state & EXT4_ERROR_FS)) |
b31e1552 ES |
1812 | ext4_msg(sb, KERN_WARNING, |
1813 | "warning: mounting fs with errors, " | |
1814 | "running e2fsck is recommended"); | |
ac27a0ec DK |
1815 | else if ((__s16) le16_to_cpu(es->s_max_mnt_count) >= 0 && |
1816 | le16_to_cpu(es->s_mnt_count) >= | |
1817 | (unsigned short) (__s16) le16_to_cpu(es->s_max_mnt_count)) | |
b31e1552 ES |
1818 | ext4_msg(sb, KERN_WARNING, |
1819 | "warning: maximal mount count reached, " | |
1820 | "running e2fsck is recommended"); | |
ac27a0ec DK |
1821 | else if (le32_to_cpu(es->s_checkinterval) && |
1822 | (le32_to_cpu(es->s_lastcheck) + | |
1823 | le32_to_cpu(es->s_checkinterval) <= get_seconds())) | |
b31e1552 ES |
1824 | ext4_msg(sb, KERN_WARNING, |
1825 | "warning: checktime reached, " | |
1826 | "running e2fsck is recommended"); | |
0b8e58a1 | 1827 | if (!sbi->s_journal) |
0390131b | 1828 | es->s_state &= cpu_to_le16(~EXT4_VALID_FS); |
ac27a0ec | 1829 | if (!(__s16) le16_to_cpu(es->s_max_mnt_count)) |
617ba13b | 1830 | es->s_max_mnt_count = cpu_to_le16(EXT4_DFL_MAX_MNT_COUNT); |
e8546d06 | 1831 | le16_add_cpu(&es->s_mnt_count, 1); |
ac27a0ec | 1832 | es->s_mtime = cpu_to_le32(get_seconds()); |
617ba13b | 1833 | ext4_update_dynamic_rev(sb); |
0390131b FM |
1834 | if (sbi->s_journal) |
1835 | EXT4_SET_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER); | |
ac27a0ec | 1836 | |
e2d67052 | 1837 | ext4_commit_super(sb, 1); |
ac27a0ec | 1838 | if (test_opt(sb, DEBUG)) |
a9df9a49 | 1839 | printk(KERN_INFO "[EXT4 FS bs=%lu, gc=%u, " |
7f4520cc | 1840 | "bpg=%lu, ipg=%lu, mo=%04x]\n", |
ac27a0ec DK |
1841 | sb->s_blocksize, |
1842 | sbi->s_groups_count, | |
617ba13b MC |
1843 | EXT4_BLOCKS_PER_GROUP(sb), |
1844 | EXT4_INODES_PER_GROUP(sb), | |
ac27a0ec DK |
1845 | sbi->s_mount_opt); |
1846 | ||
ac27a0ec DK |
1847 | return res; |
1848 | } | |
1849 | ||
772cb7c8 JS |
1850 | static int ext4_fill_flex_info(struct super_block *sb) |
1851 | { | |
1852 | struct ext4_sb_info *sbi = EXT4_SB(sb); | |
1853 | struct ext4_group_desc *gdp = NULL; | |
772cb7c8 JS |
1854 | ext4_group_t flex_group_count; |
1855 | ext4_group_t flex_group; | |
1856 | int groups_per_flex = 0; | |
c5ca7c76 | 1857 | size_t size; |
772cb7c8 JS |
1858 | int i; |
1859 | ||
503358ae TT |
1860 | sbi->s_log_groups_per_flex = sbi->s_es->s_log_groups_per_flex; |
1861 | groups_per_flex = 1 << sbi->s_log_groups_per_flex; | |
1862 | ||
1863 | if (groups_per_flex < 2) { | |
772cb7c8 JS |
1864 | sbi->s_log_groups_per_flex = 0; |
1865 | return 1; | |
1866 | } | |
1867 | ||
c62a11fd FB |
1868 | /* We allocate both existing and potentially added groups */ |
1869 | flex_group_count = ((sbi->s_groups_count + groups_per_flex - 1) + | |
d94e99a6 AK |
1870 | ((le16_to_cpu(sbi->s_es->s_reserved_gdt_blocks) + 1) << |
1871 | EXT4_DESC_PER_BLOCK_BITS(sb))) / groups_per_flex; | |
c5ca7c76 TT |
1872 | size = flex_group_count * sizeof(struct flex_groups); |
1873 | sbi->s_flex_groups = kzalloc(size, GFP_KERNEL); | |
1874 | if (sbi->s_flex_groups == NULL) { | |
1875 | sbi->s_flex_groups = vmalloc(size); | |
1876 | if (sbi->s_flex_groups) | |
1877 | memset(sbi->s_flex_groups, 0, size); | |
1878 | } | |
772cb7c8 | 1879 | if (sbi->s_flex_groups == NULL) { |
b31e1552 ES |
1880 | ext4_msg(sb, KERN_ERR, "not enough memory for " |
1881 | "%u flex groups", flex_group_count); | |
772cb7c8 JS |
1882 | goto failed; |
1883 | } | |
772cb7c8 | 1884 | |
772cb7c8 | 1885 | for (i = 0; i < sbi->s_groups_count; i++) { |
88b6edd1 | 1886 | gdp = ext4_get_group_desc(sb, i, NULL); |
772cb7c8 JS |
1887 | |
1888 | flex_group = ext4_flex_group(sbi, i); | |
7ad9bb65 TT |
1889 | atomic_add(ext4_free_inodes_count(sb, gdp), |
1890 | &sbi->s_flex_groups[flex_group].free_inodes); | |
1891 | atomic_add(ext4_free_blks_count(sb, gdp), | |
1892 | &sbi->s_flex_groups[flex_group].free_blocks); | |
1893 | atomic_add(ext4_used_dirs_count(sb, gdp), | |
1894 | &sbi->s_flex_groups[flex_group].used_dirs); | |
772cb7c8 JS |
1895 | } |
1896 | ||
1897 | return 1; | |
1898 | failed: | |
1899 | return 0; | |
1900 | } | |
1901 | ||
717d50e4 AD |
1902 | __le16 ext4_group_desc_csum(struct ext4_sb_info *sbi, __u32 block_group, |
1903 | struct ext4_group_desc *gdp) | |
1904 | { | |
1905 | __u16 crc = 0; | |
1906 | ||
1907 | if (sbi->s_es->s_feature_ro_compat & | |
1908 | cpu_to_le32(EXT4_FEATURE_RO_COMPAT_GDT_CSUM)) { | |
1909 | int offset = offsetof(struct ext4_group_desc, bg_checksum); | |
1910 | __le32 le_group = cpu_to_le32(block_group); | |
1911 | ||
1912 | crc = crc16(~0, sbi->s_es->s_uuid, sizeof(sbi->s_es->s_uuid)); | |
1913 | crc = crc16(crc, (__u8 *)&le_group, sizeof(le_group)); | |
1914 | crc = crc16(crc, (__u8 *)gdp, offset); | |
1915 | offset += sizeof(gdp->bg_checksum); /* skip checksum */ | |
1916 | /* for checksum of struct ext4_group_desc do the rest...*/ | |
1917 | if ((sbi->s_es->s_feature_incompat & | |
1918 | cpu_to_le32(EXT4_FEATURE_INCOMPAT_64BIT)) && | |
1919 | offset < le16_to_cpu(sbi->s_es->s_desc_size)) | |
1920 | crc = crc16(crc, (__u8 *)gdp + offset, | |
1921 | le16_to_cpu(sbi->s_es->s_desc_size) - | |
1922 | offset); | |
1923 | } | |
1924 | ||
1925 | return cpu_to_le16(crc); | |
1926 | } | |
1927 | ||
1928 | int ext4_group_desc_csum_verify(struct ext4_sb_info *sbi, __u32 block_group, | |
1929 | struct ext4_group_desc *gdp) | |
1930 | { | |
1931 | if ((sbi->s_es->s_feature_ro_compat & | |
1932 | cpu_to_le32(EXT4_FEATURE_RO_COMPAT_GDT_CSUM)) && | |
1933 | (gdp->bg_checksum != ext4_group_desc_csum(sbi, block_group, gdp))) | |
1934 | return 0; | |
1935 | ||
1936 | return 1; | |
1937 | } | |
1938 | ||
ac27a0ec | 1939 | /* Called at mount-time, super-block is locked */ |
197cd65a | 1940 | static int ext4_check_descriptors(struct super_block *sb) |
ac27a0ec | 1941 | { |
617ba13b MC |
1942 | struct ext4_sb_info *sbi = EXT4_SB(sb); |
1943 | ext4_fsblk_t first_block = le32_to_cpu(sbi->s_es->s_first_data_block); | |
1944 | ext4_fsblk_t last_block; | |
bd81d8ee LV |
1945 | ext4_fsblk_t block_bitmap; |
1946 | ext4_fsblk_t inode_bitmap; | |
1947 | ext4_fsblk_t inode_table; | |
ce421581 | 1948 | int flexbg_flag = 0; |
fd2d4291 | 1949 | ext4_group_t i; |
ac27a0ec | 1950 | |
ce421581 JS |
1951 | if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FLEX_BG)) |
1952 | flexbg_flag = 1; | |
1953 | ||
af5bc92d | 1954 | ext4_debug("Checking group descriptors"); |
ac27a0ec | 1955 | |
197cd65a AM |
1956 | for (i = 0; i < sbi->s_groups_count; i++) { |
1957 | struct ext4_group_desc *gdp = ext4_get_group_desc(sb, i, NULL); | |
1958 | ||
ce421581 | 1959 | if (i == sbi->s_groups_count - 1 || flexbg_flag) |
bd81d8ee | 1960 | last_block = ext4_blocks_count(sbi->s_es) - 1; |
ac27a0ec DK |
1961 | else |
1962 | last_block = first_block + | |
617ba13b | 1963 | (EXT4_BLOCKS_PER_GROUP(sb) - 1); |
ac27a0ec | 1964 | |
8fadc143 | 1965 | block_bitmap = ext4_block_bitmap(sb, gdp); |
2b2d6d01 | 1966 | if (block_bitmap < first_block || block_bitmap > last_block) { |
b31e1552 | 1967 | ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: " |
a9df9a49 | 1968 | "Block bitmap for group %u not in group " |
b31e1552 | 1969 | "(block %llu)!", i, block_bitmap); |
ac27a0ec DK |
1970 | return 0; |
1971 | } | |
8fadc143 | 1972 | inode_bitmap = ext4_inode_bitmap(sb, gdp); |
2b2d6d01 | 1973 | if (inode_bitmap < first_block || inode_bitmap > last_block) { |
b31e1552 | 1974 | ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: " |
a9df9a49 | 1975 | "Inode bitmap for group %u not in group " |
b31e1552 | 1976 | "(block %llu)!", i, inode_bitmap); |
ac27a0ec DK |
1977 | return 0; |
1978 | } | |
8fadc143 | 1979 | inode_table = ext4_inode_table(sb, gdp); |
bd81d8ee | 1980 | if (inode_table < first_block || |
2b2d6d01 | 1981 | inode_table + sbi->s_itb_per_group - 1 > last_block) { |
b31e1552 | 1982 | ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: " |
a9df9a49 | 1983 | "Inode table for group %u not in group " |
b31e1552 | 1984 | "(block %llu)!", i, inode_table); |
ac27a0ec DK |
1985 | return 0; |
1986 | } | |
955ce5f5 | 1987 | ext4_lock_group(sb, i); |
717d50e4 | 1988 | if (!ext4_group_desc_csum_verify(sbi, i, gdp)) { |
b31e1552 ES |
1989 | ext4_msg(sb, KERN_ERR, "ext4_check_descriptors: " |
1990 | "Checksum for group %u failed (%u!=%u)", | |
1991 | i, le16_to_cpu(ext4_group_desc_csum(sbi, i, | |
1992 | gdp)), le16_to_cpu(gdp->bg_checksum)); | |
7ee1ec4c | 1993 | if (!(sb->s_flags & MS_RDONLY)) { |
955ce5f5 | 1994 | ext4_unlock_group(sb, i); |
8a266467 | 1995 | return 0; |
7ee1ec4c | 1996 | } |
717d50e4 | 1997 | } |
955ce5f5 | 1998 | ext4_unlock_group(sb, i); |
ce421581 JS |
1999 | if (!flexbg_flag) |
2000 | first_block += EXT4_BLOCKS_PER_GROUP(sb); | |
ac27a0ec DK |
2001 | } |
2002 | ||
bd81d8ee | 2003 | ext4_free_blocks_count_set(sbi->s_es, ext4_count_free_blocks(sb)); |
0b8e58a1 | 2004 | sbi->s_es->s_free_inodes_count =cpu_to_le32(ext4_count_free_inodes(sb)); |
ac27a0ec DK |
2005 | return 1; |
2006 | } | |
2007 | ||
617ba13b | 2008 | /* ext4_orphan_cleanup() walks a singly-linked list of inodes (starting at |
ac27a0ec DK |
2009 | * the superblock) which were deleted from all directories, but held open by |
2010 | * a process at the time of a crash. We walk the list and try to delete these | |
2011 | * inodes at recovery time (only with a read-write filesystem). | |
2012 | * | |
2013 | * In order to keep the orphan inode chain consistent during traversal (in | |
2014 | * case of crash during recovery), we link each inode into the superblock | |
2015 | * orphan list_head and handle it the same way as an inode deletion during | |
2016 | * normal operation (which journals the operations for us). | |
2017 | * | |
2018 | * We only do an iget() and an iput() on each inode, which is very safe if we | |
2019 | * accidentally point at an in-use or already deleted inode. The worst that | |
2020 | * can happen in this case is that we get a "bit already cleared" message from | |
617ba13b | 2021 | * ext4_free_inode(). The only reason we would point at a wrong inode is if |
ac27a0ec DK |
2022 | * e2fsck was run on this filesystem, and it must have already done the orphan |
2023 | * inode cleanup for us, so we can safely abort without any further action. | |
2024 | */ | |
2b2d6d01 TT |
2025 | static void ext4_orphan_cleanup(struct super_block *sb, |
2026 | struct ext4_super_block *es) | |
ac27a0ec DK |
2027 | { |
2028 | unsigned int s_flags = sb->s_flags; | |
2029 | int nr_orphans = 0, nr_truncates = 0; | |
2030 | #ifdef CONFIG_QUOTA | |
2031 | int i; | |
2032 | #endif | |
2033 | if (!es->s_last_orphan) { | |
2034 | jbd_debug(4, "no orphan inodes to clean up\n"); | |
2035 | return; | |
2036 | } | |
2037 | ||
a8f48a95 | 2038 | if (bdev_read_only(sb->s_bdev)) { |
b31e1552 ES |
2039 | ext4_msg(sb, KERN_ERR, "write access " |
2040 | "unavailable, skipping orphan cleanup"); | |
a8f48a95 ES |
2041 | return; |
2042 | } | |
2043 | ||
617ba13b | 2044 | if (EXT4_SB(sb)->s_mount_state & EXT4_ERROR_FS) { |
ac27a0ec DK |
2045 | if (es->s_last_orphan) |
2046 | jbd_debug(1, "Errors on filesystem, " | |
2047 | "clearing orphan list.\n"); | |
2048 | es->s_last_orphan = 0; | |
2049 | jbd_debug(1, "Skipping orphan recovery on fs with errors.\n"); | |
2050 | return; | |
2051 | } | |
2052 | ||
2053 | if (s_flags & MS_RDONLY) { | |
b31e1552 | 2054 | ext4_msg(sb, KERN_INFO, "orphan cleanup on readonly fs"); |
ac27a0ec DK |
2055 | sb->s_flags &= ~MS_RDONLY; |
2056 | } | |
2057 | #ifdef CONFIG_QUOTA | |
2058 | /* Needed for iput() to work correctly and not trash data */ | |
2059 | sb->s_flags |= MS_ACTIVE; | |
2060 | /* Turn on quotas so that they are updated correctly */ | |
2061 | for (i = 0; i < MAXQUOTAS; i++) { | |
617ba13b MC |
2062 | if (EXT4_SB(sb)->s_qf_names[i]) { |
2063 | int ret = ext4_quota_on_mount(sb, i); | |
ac27a0ec | 2064 | if (ret < 0) |
b31e1552 ES |
2065 | ext4_msg(sb, KERN_ERR, |
2066 | "Cannot turn on journaled " | |
2067 | "quota: error %d", ret); | |
ac27a0ec DK |
2068 | } |
2069 | } | |
2070 | #endif | |
2071 | ||
2072 | while (es->s_last_orphan) { | |
2073 | struct inode *inode; | |
2074 | ||
97bd42b9 JB |
2075 | inode = ext4_orphan_get(sb, le32_to_cpu(es->s_last_orphan)); |
2076 | if (IS_ERR(inode)) { | |
ac27a0ec DK |
2077 | es->s_last_orphan = 0; |
2078 | break; | |
2079 | } | |
2080 | ||
617ba13b | 2081 | list_add(&EXT4_I(inode)->i_orphan, &EXT4_SB(sb)->s_orphan); |
871a2931 | 2082 | dquot_initialize(inode); |
ac27a0ec | 2083 | if (inode->i_nlink) { |
b31e1552 ES |
2084 | ext4_msg(sb, KERN_DEBUG, |
2085 | "%s: truncating inode %lu to %lld bytes", | |
46e665e9 | 2086 | __func__, inode->i_ino, inode->i_size); |
e5f8eab8 | 2087 | jbd_debug(2, "truncating inode %lu to %lld bytes\n", |
ac27a0ec | 2088 | inode->i_ino, inode->i_size); |
617ba13b | 2089 | ext4_truncate(inode); |
ac27a0ec DK |
2090 | nr_truncates++; |
2091 | } else { | |
b31e1552 ES |
2092 | ext4_msg(sb, KERN_DEBUG, |
2093 | "%s: deleting unreferenced inode %lu", | |
46e665e9 | 2094 | __func__, inode->i_ino); |
ac27a0ec DK |
2095 | jbd_debug(2, "deleting unreferenced inode %lu\n", |
2096 | inode->i_ino); | |
2097 | nr_orphans++; | |
2098 | } | |
2099 | iput(inode); /* The delete magic happens here! */ | |
2100 | } | |
2101 | ||
2b2d6d01 | 2102 | #define PLURAL(x) (x), ((x) == 1) ? "" : "s" |
ac27a0ec DK |
2103 | |
2104 | if (nr_orphans) | |
b31e1552 ES |
2105 | ext4_msg(sb, KERN_INFO, "%d orphan inode%s deleted", |
2106 | PLURAL(nr_orphans)); | |
ac27a0ec | 2107 | if (nr_truncates) |
b31e1552 ES |
2108 | ext4_msg(sb, KERN_INFO, "%d truncate%s cleaned up", |
2109 | PLURAL(nr_truncates)); | |
ac27a0ec DK |
2110 | #ifdef CONFIG_QUOTA |
2111 | /* Turn quotas off */ | |
2112 | for (i = 0; i < MAXQUOTAS; i++) { | |
2113 | if (sb_dqopt(sb)->files[i]) | |
287a8095 | 2114 | dquot_quota_off(sb, i); |
ac27a0ec DK |
2115 | } |
2116 | #endif | |
2117 | sb->s_flags = s_flags; /* Restore MS_RDONLY status */ | |
2118 | } | |
0b8e58a1 | 2119 | |
cd2291a4 ES |
2120 | /* |
2121 | * Maximal extent format file size. | |
2122 | * Resulting logical blkno at s_maxbytes must fit in our on-disk | |
2123 | * extent format containers, within a sector_t, and within i_blocks | |
2124 | * in the vfs. ext4 inode has 48 bits of i_block in fsblock units, | |
2125 | * so that won't be a limiting factor. | |
2126 | * | |
2127 | * Note, this does *not* consider any metadata overhead for vfs i_blocks. | |
2128 | */ | |
f287a1a5 | 2129 | static loff_t ext4_max_size(int blkbits, int has_huge_files) |
cd2291a4 ES |
2130 | { |
2131 | loff_t res; | |
2132 | loff_t upper_limit = MAX_LFS_FILESIZE; | |
2133 | ||
2134 | /* small i_blocks in vfs inode? */ | |
f287a1a5 | 2135 | if (!has_huge_files || sizeof(blkcnt_t) < sizeof(u64)) { |
cd2291a4 | 2136 | /* |
90c699a9 | 2137 | * CONFIG_LBDAF is not enabled implies the inode |
cd2291a4 ES |
2138 | * i_block represent total blocks in 512 bytes |
2139 | * 32 == size of vfs inode i_blocks * 8 | |
2140 | */ | |
2141 | upper_limit = (1LL << 32) - 1; | |
2142 | ||
2143 | /* total blocks in file system block size */ | |
2144 | upper_limit >>= (blkbits - 9); | |
2145 | upper_limit <<= blkbits; | |
2146 | } | |
2147 | ||
2148 | /* 32-bit extent-start container, ee_block */ | |
2149 | res = 1LL << 32; | |
2150 | res <<= blkbits; | |
2151 | res -= 1; | |
2152 | ||
2153 | /* Sanity check against vm- & vfs- imposed limits */ | |
2154 | if (res > upper_limit) | |
2155 | res = upper_limit; | |
2156 | ||
2157 | return res; | |
2158 | } | |
ac27a0ec | 2159 | |
ac27a0ec | 2160 | /* |
cd2291a4 | 2161 | * Maximal bitmap file size. There is a direct, and {,double-,triple-}indirect |
0fc1b451 AK |
2162 | * block limit, and also a limit of (2^48 - 1) 512-byte sectors in i_blocks. |
2163 | * We need to be 1 filesystem block less than the 2^48 sector limit. | |
ac27a0ec | 2164 | */ |
f287a1a5 | 2165 | static loff_t ext4_max_bitmap_size(int bits, int has_huge_files) |
ac27a0ec | 2166 | { |
617ba13b | 2167 | loff_t res = EXT4_NDIR_BLOCKS; |
0fc1b451 AK |
2168 | int meta_blocks; |
2169 | loff_t upper_limit; | |
0b8e58a1 AD |
2170 | /* This is calculated to be the largest file size for a dense, block |
2171 | * mapped file such that the file's total number of 512-byte sectors, | |
2172 | * including data and all indirect blocks, does not exceed (2^48 - 1). | |
2173 | * | |
2174 | * __u32 i_blocks_lo and _u16 i_blocks_high represent the total | |
2175 | * number of 512-byte sectors of the file. | |
0fc1b451 AK |
2176 | */ |
2177 | ||
f287a1a5 | 2178 | if (!has_huge_files || sizeof(blkcnt_t) < sizeof(u64)) { |
0fc1b451 | 2179 | /* |
90c699a9 | 2180 | * !has_huge_files or CONFIG_LBDAF not enabled implies that |
0b8e58a1 AD |
2181 | * the inode i_block field represents total file blocks in |
2182 | * 2^32 512-byte sectors == size of vfs inode i_blocks * 8 | |
0fc1b451 AK |
2183 | */ |
2184 | upper_limit = (1LL << 32) - 1; | |
2185 | ||
2186 | /* total blocks in file system block size */ | |
2187 | upper_limit >>= (bits - 9); | |
2188 | ||
2189 | } else { | |
8180a562 AK |
2190 | /* |
2191 | * We use 48 bit ext4_inode i_blocks | |
2192 | * With EXT4_HUGE_FILE_FL set the i_blocks | |
2193 | * represent total number of blocks in | |
2194 | * file system block size | |
2195 | */ | |
0fc1b451 AK |
2196 | upper_limit = (1LL << 48) - 1; |
2197 | ||
0fc1b451 AK |
2198 | } |
2199 | ||
2200 | /* indirect blocks */ | |
2201 | meta_blocks = 1; | |
2202 | /* double indirect blocks */ | |
2203 | meta_blocks += 1 + (1LL << (bits-2)); | |
2204 | /* tripple indirect blocks */ | |
2205 | meta_blocks += 1 + (1LL << (bits-2)) + (1LL << (2*(bits-2))); | |
2206 | ||
2207 | upper_limit -= meta_blocks; | |
2208 | upper_limit <<= bits; | |
ac27a0ec DK |
2209 | |
2210 | res += 1LL << (bits-2); | |
2211 | res += 1LL << (2*(bits-2)); | |
2212 | res += 1LL << (3*(bits-2)); | |
2213 | res <<= bits; | |
2214 | if (res > upper_limit) | |
2215 | res = upper_limit; | |
0fc1b451 AK |
2216 | |
2217 | if (res > MAX_LFS_FILESIZE) | |
2218 | res = MAX_LFS_FILESIZE; | |
2219 | ||
ac27a0ec DK |
2220 | return res; |
2221 | } | |
2222 | ||
617ba13b | 2223 | static ext4_fsblk_t descriptor_loc(struct super_block *sb, |
0b8e58a1 | 2224 | ext4_fsblk_t logical_sb_block, int nr) |
ac27a0ec | 2225 | { |
617ba13b | 2226 | struct ext4_sb_info *sbi = EXT4_SB(sb); |
fd2d4291 | 2227 | ext4_group_t bg, first_meta_bg; |
ac27a0ec DK |
2228 | int has_super = 0; |
2229 | ||
2230 | first_meta_bg = le32_to_cpu(sbi->s_es->s_first_meta_bg); | |
2231 | ||
617ba13b | 2232 | if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_META_BG) || |
ac27a0ec | 2233 | nr < first_meta_bg) |
70bbb3e0 | 2234 | return logical_sb_block + nr + 1; |
ac27a0ec | 2235 | bg = sbi->s_desc_per_block * nr; |
617ba13b | 2236 | if (ext4_bg_has_super(sb, bg)) |
ac27a0ec | 2237 | has_super = 1; |
0b8e58a1 | 2238 | |
617ba13b | 2239 | return (has_super + ext4_group_first_block_no(sb, bg)); |
ac27a0ec DK |
2240 | } |
2241 | ||
c9de560d AT |
2242 | /** |
2243 | * ext4_get_stripe_size: Get the stripe size. | |
2244 | * @sbi: In memory super block info | |
2245 | * | |
2246 | * If we have specified it via mount option, then | |
2247 | * use the mount option value. If the value specified at mount time is | |
2248 | * greater than the blocks per group use the super block value. | |
2249 | * If the super block value is greater than blocks per group return 0. | |
2250 | * Allocator needs it be less than blocks per group. | |
2251 | * | |
2252 | */ | |
2253 | static unsigned long ext4_get_stripe_size(struct ext4_sb_info *sbi) | |
2254 | { | |
2255 | unsigned long stride = le16_to_cpu(sbi->s_es->s_raid_stride); | |
2256 | unsigned long stripe_width = | |
2257 | le32_to_cpu(sbi->s_es->s_raid_stripe_width); | |
2258 | ||
2259 | if (sbi->s_stripe && sbi->s_stripe <= sbi->s_blocks_per_group) | |
2260 | return sbi->s_stripe; | |
2261 | ||
2262 | if (stripe_width <= sbi->s_blocks_per_group) | |
2263 | return stripe_width; | |
2264 | ||
2265 | if (stride <= sbi->s_blocks_per_group) | |
2266 | return stride; | |
2267 | ||
2268 | return 0; | |
2269 | } | |
ac27a0ec | 2270 | |
3197ebdb TT |
2271 | /* sysfs supprt */ |
2272 | ||
2273 | struct ext4_attr { | |
2274 | struct attribute attr; | |
2275 | ssize_t (*show)(struct ext4_attr *, struct ext4_sb_info *, char *); | |
60e6679e | 2276 | ssize_t (*store)(struct ext4_attr *, struct ext4_sb_info *, |
3197ebdb TT |
2277 | const char *, size_t); |
2278 | int offset; | |
2279 | }; | |
2280 | ||
2281 | static int parse_strtoul(const char *buf, | |
2282 | unsigned long max, unsigned long *value) | |
2283 | { | |
2284 | char *endp; | |
2285 | ||
e7d2860b AGR |
2286 | *value = simple_strtoul(skip_spaces(buf), &endp, 0); |
2287 | endp = skip_spaces(endp); | |
3197ebdb TT |
2288 | if (*endp || *value > max) |
2289 | return -EINVAL; | |
2290 | ||
2291 | return 0; | |
2292 | } | |
2293 | ||
2294 | static ssize_t delayed_allocation_blocks_show(struct ext4_attr *a, | |
2295 | struct ext4_sb_info *sbi, | |
2296 | char *buf) | |
2297 | { | |
2298 | return snprintf(buf, PAGE_SIZE, "%llu\n", | |
2299 | (s64) percpu_counter_sum(&sbi->s_dirtyblocks_counter)); | |
2300 | } | |
2301 | ||
2302 | static ssize_t session_write_kbytes_show(struct ext4_attr *a, | |
2303 | struct ext4_sb_info *sbi, char *buf) | |
2304 | { | |
2305 | struct super_block *sb = sbi->s_buddy_cache->i_sb; | |
2306 | ||
2307 | return snprintf(buf, PAGE_SIZE, "%lu\n", | |
2308 | (part_stat_read(sb->s_bdev->bd_part, sectors[1]) - | |
2309 | sbi->s_sectors_written_start) >> 1); | |
2310 | } | |
2311 | ||
2312 | static ssize_t lifetime_write_kbytes_show(struct ext4_attr *a, | |
2313 | struct ext4_sb_info *sbi, char *buf) | |
2314 | { | |
2315 | struct super_block *sb = sbi->s_buddy_cache->i_sb; | |
2316 | ||
2317 | return snprintf(buf, PAGE_SIZE, "%llu\n", | |
a6b43e38 | 2318 | (unsigned long long)(sbi->s_kbytes_written + |
3197ebdb | 2319 | ((part_stat_read(sb->s_bdev->bd_part, sectors[1]) - |
a6b43e38 | 2320 | EXT4_SB(sb)->s_sectors_written_start) >> 1))); |
3197ebdb TT |
2321 | } |
2322 | ||
2323 | static ssize_t inode_readahead_blks_store(struct ext4_attr *a, | |
2324 | struct ext4_sb_info *sbi, | |
2325 | const char *buf, size_t count) | |
2326 | { | |
2327 | unsigned long t; | |
2328 | ||
2329 | if (parse_strtoul(buf, 0x40000000, &t)) | |
2330 | return -EINVAL; | |
2331 | ||
f7c43950 | 2332 | if (!is_power_of_2(t)) |
3197ebdb TT |
2333 | return -EINVAL; |
2334 | ||
2335 | sbi->s_inode_readahead_blks = t; | |
2336 | return count; | |
2337 | } | |
2338 | ||
2339 | static ssize_t sbi_ui_show(struct ext4_attr *a, | |
0b8e58a1 | 2340 | struct ext4_sb_info *sbi, char *buf) |
3197ebdb TT |
2341 | { |
2342 | unsigned int *ui = (unsigned int *) (((char *) sbi) + a->offset); | |
2343 | ||
2344 | return snprintf(buf, PAGE_SIZE, "%u\n", *ui); | |
2345 | } | |
2346 | ||
2347 | static ssize_t sbi_ui_store(struct ext4_attr *a, | |
2348 | struct ext4_sb_info *sbi, | |
2349 | const char *buf, size_t count) | |
2350 | { | |
2351 | unsigned int *ui = (unsigned int *) (((char *) sbi) + a->offset); | |
2352 | unsigned long t; | |
2353 | ||
2354 | if (parse_strtoul(buf, 0xffffffff, &t)) | |
2355 | return -EINVAL; | |
2356 | *ui = t; | |
2357 | return count; | |
2358 | } | |
2359 | ||
2360 | #define EXT4_ATTR_OFFSET(_name,_mode,_show,_store,_elname) \ | |
2361 | static struct ext4_attr ext4_attr_##_name = { \ | |
2362 | .attr = {.name = __stringify(_name), .mode = _mode }, \ | |
2363 | .show = _show, \ | |
2364 | .store = _store, \ | |
2365 | .offset = offsetof(struct ext4_sb_info, _elname), \ | |
2366 | } | |
2367 | #define EXT4_ATTR(name, mode, show, store) \ | |
2368 | static struct ext4_attr ext4_attr_##name = __ATTR(name, mode, show, store) | |
2369 | ||
2370 | #define EXT4_RO_ATTR(name) EXT4_ATTR(name, 0444, name##_show, NULL) | |
2371 | #define EXT4_RW_ATTR(name) EXT4_ATTR(name, 0644, name##_show, name##_store) | |
2372 | #define EXT4_RW_ATTR_SBI_UI(name, elname) \ | |
2373 | EXT4_ATTR_OFFSET(name, 0644, sbi_ui_show, sbi_ui_store, elname) | |
2374 | #define ATTR_LIST(name) &ext4_attr_##name.attr | |
2375 | ||
2376 | EXT4_RO_ATTR(delayed_allocation_blocks); | |
2377 | EXT4_RO_ATTR(session_write_kbytes); | |
2378 | EXT4_RO_ATTR(lifetime_write_kbytes); | |
2379 | EXT4_ATTR_OFFSET(inode_readahead_blks, 0644, sbi_ui_show, | |
2380 | inode_readahead_blks_store, s_inode_readahead_blks); | |
11013911 | 2381 | EXT4_RW_ATTR_SBI_UI(inode_goal, s_inode_goal); |
3197ebdb TT |
2382 | EXT4_RW_ATTR_SBI_UI(mb_stats, s_mb_stats); |
2383 | EXT4_RW_ATTR_SBI_UI(mb_max_to_scan, s_mb_max_to_scan); | |
2384 | EXT4_RW_ATTR_SBI_UI(mb_min_to_scan, s_mb_min_to_scan); | |
2385 | EXT4_RW_ATTR_SBI_UI(mb_order2_req, s_mb_order2_reqs); | |
2386 | EXT4_RW_ATTR_SBI_UI(mb_stream_req, s_mb_stream_request); | |
2387 | EXT4_RW_ATTR_SBI_UI(mb_group_prealloc, s_mb_group_prealloc); | |
55138e0b | 2388 | EXT4_RW_ATTR_SBI_UI(max_writeback_mb_bump, s_max_writeback_mb_bump); |
3197ebdb TT |
2389 | |
2390 | static struct attribute *ext4_attrs[] = { | |
2391 | ATTR_LIST(delayed_allocation_blocks), | |
2392 | ATTR_LIST(session_write_kbytes), | |
2393 | ATTR_LIST(lifetime_write_kbytes), | |
2394 | ATTR_LIST(inode_readahead_blks), | |
11013911 | 2395 | ATTR_LIST(inode_goal), |
3197ebdb TT |
2396 | ATTR_LIST(mb_stats), |
2397 | ATTR_LIST(mb_max_to_scan), | |
2398 | ATTR_LIST(mb_min_to_scan), | |
2399 | ATTR_LIST(mb_order2_req), | |
2400 | ATTR_LIST(mb_stream_req), | |
2401 | ATTR_LIST(mb_group_prealloc), | |
55138e0b | 2402 | ATTR_LIST(max_writeback_mb_bump), |
3197ebdb TT |
2403 | NULL, |
2404 | }; | |
2405 | ||
2406 | static ssize_t ext4_attr_show(struct kobject *kobj, | |
2407 | struct attribute *attr, char *buf) | |
2408 | { | |
2409 | struct ext4_sb_info *sbi = container_of(kobj, struct ext4_sb_info, | |
2410 | s_kobj); | |
2411 | struct ext4_attr *a = container_of(attr, struct ext4_attr, attr); | |
2412 | ||
2413 | return a->show ? a->show(a, sbi, buf) : 0; | |
2414 | } | |
2415 | ||
2416 | static ssize_t ext4_attr_store(struct kobject *kobj, | |
2417 | struct attribute *attr, | |
2418 | const char *buf, size_t len) | |
2419 | { | |
2420 | struct ext4_sb_info *sbi = container_of(kobj, struct ext4_sb_info, | |
2421 | s_kobj); | |
2422 | struct ext4_attr *a = container_of(attr, struct ext4_attr, attr); | |
2423 | ||
2424 | return a->store ? a->store(a, sbi, buf, len) : 0; | |
2425 | } | |
2426 | ||
2427 | static void ext4_sb_release(struct kobject *kobj) | |
2428 | { | |
2429 | struct ext4_sb_info *sbi = container_of(kobj, struct ext4_sb_info, | |
2430 | s_kobj); | |
2431 | complete(&sbi->s_kobj_unregister); | |
2432 | } | |
2433 | ||
2434 | ||
52cf25d0 | 2435 | static const struct sysfs_ops ext4_attr_ops = { |
3197ebdb TT |
2436 | .show = ext4_attr_show, |
2437 | .store = ext4_attr_store, | |
2438 | }; | |
2439 | ||
2440 | static struct kobj_type ext4_ktype = { | |
2441 | .default_attrs = ext4_attrs, | |
2442 | .sysfs_ops = &ext4_attr_ops, | |
2443 | .release = ext4_sb_release, | |
2444 | }; | |
2445 | ||
a13fb1a4 ES |
2446 | /* |
2447 | * Check whether this filesystem can be mounted based on | |
2448 | * the features present and the RDONLY/RDWR mount requested. | |
2449 | * Returns 1 if this filesystem can be mounted as requested, | |
2450 | * 0 if it cannot be. | |
2451 | */ | |
2452 | static int ext4_feature_set_ok(struct super_block *sb, int readonly) | |
2453 | { | |
2454 | if (EXT4_HAS_INCOMPAT_FEATURE(sb, ~EXT4_FEATURE_INCOMPAT_SUPP)) { | |
2455 | ext4_msg(sb, KERN_ERR, | |
2456 | "Couldn't mount because of " | |
2457 | "unsupported optional features (%x)", | |
2458 | (le32_to_cpu(EXT4_SB(sb)->s_es->s_feature_incompat) & | |
2459 | ~EXT4_FEATURE_INCOMPAT_SUPP)); | |
2460 | return 0; | |
2461 | } | |
2462 | ||
2463 | if (readonly) | |
2464 | return 1; | |
2465 | ||
2466 | /* Check that feature set is OK for a read-write mount */ | |
2467 | if (EXT4_HAS_RO_COMPAT_FEATURE(sb, ~EXT4_FEATURE_RO_COMPAT_SUPP)) { | |
2468 | ext4_msg(sb, KERN_ERR, "couldn't mount RDWR because of " | |
2469 | "unsupported optional features (%x)", | |
2470 | (le32_to_cpu(EXT4_SB(sb)->s_es->s_feature_ro_compat) & | |
2471 | ~EXT4_FEATURE_RO_COMPAT_SUPP)); | |
2472 | return 0; | |
2473 | } | |
2474 | /* | |
2475 | * Large file size enabled file system can only be mounted | |
2476 | * read-write on 32-bit systems if kernel is built with CONFIG_LBDAF | |
2477 | */ | |
2478 | if (EXT4_HAS_RO_COMPAT_FEATURE(sb, EXT4_FEATURE_RO_COMPAT_HUGE_FILE)) { | |
2479 | if (sizeof(blkcnt_t) < sizeof(u64)) { | |
2480 | ext4_msg(sb, KERN_ERR, "Filesystem with huge files " | |
2481 | "cannot be mounted RDWR without " | |
2482 | "CONFIG_LBDAF"); | |
2483 | return 0; | |
2484 | } | |
2485 | } | |
2486 | return 1; | |
2487 | } | |
2488 | ||
66e61a9e TT |
2489 | /* |
2490 | * This function is called once a day if we have errors logged | |
2491 | * on the file system | |
2492 | */ | |
2493 | static void print_daily_error_info(unsigned long arg) | |
2494 | { | |
2495 | struct super_block *sb = (struct super_block *) arg; | |
2496 | struct ext4_sb_info *sbi; | |
2497 | struct ext4_super_block *es; | |
2498 | ||
2499 | sbi = EXT4_SB(sb); | |
2500 | es = sbi->s_es; | |
2501 | ||
2502 | if (es->s_error_count) | |
2503 | ext4_msg(sb, KERN_NOTICE, "error count: %u", | |
2504 | le32_to_cpu(es->s_error_count)); | |
2505 | if (es->s_first_error_time) { | |
2506 | printk(KERN_NOTICE "EXT4-fs (%s): initial error at %u: %.*s:%d", | |
2507 | sb->s_id, le32_to_cpu(es->s_first_error_time), | |
2508 | (int) sizeof(es->s_first_error_func), | |
2509 | es->s_first_error_func, | |
2510 | le32_to_cpu(es->s_first_error_line)); | |
2511 | if (es->s_first_error_ino) | |
2512 | printk(": inode %u", | |
2513 | le32_to_cpu(es->s_first_error_ino)); | |
2514 | if (es->s_first_error_block) | |
2515 | printk(": block %llu", (unsigned long long) | |
2516 | le64_to_cpu(es->s_first_error_block)); | |
2517 | printk("\n"); | |
2518 | } | |
2519 | if (es->s_last_error_time) { | |
2520 | printk(KERN_NOTICE "EXT4-fs (%s): last error at %u: %.*s:%d", | |
2521 | sb->s_id, le32_to_cpu(es->s_last_error_time), | |
2522 | (int) sizeof(es->s_last_error_func), | |
2523 | es->s_last_error_func, | |
2524 | le32_to_cpu(es->s_last_error_line)); | |
2525 | if (es->s_last_error_ino) | |
2526 | printk(": inode %u", | |
2527 | le32_to_cpu(es->s_last_error_ino)); | |
2528 | if (es->s_last_error_block) | |
2529 | printk(": block %llu", (unsigned long long) | |
2530 | le64_to_cpu(es->s_last_error_block)); | |
2531 | printk("\n"); | |
2532 | } | |
2533 | mod_timer(&sbi->s_err_report, jiffies + 24*60*60*HZ); /* Once a day */ | |
2534 | } | |
2535 | ||
2b2d6d01 | 2536 | static int ext4_fill_super(struct super_block *sb, void *data, int silent) |
7477827f AK |
2537 | __releases(kernel_lock) |
2538 | __acquires(kernel_lock) | |
ac27a0ec | 2539 | { |
d4c402d9 | 2540 | char *orig_data = kstrdup(data, GFP_KERNEL); |
2b2d6d01 | 2541 | struct buffer_head *bh; |
617ba13b MC |
2542 | struct ext4_super_block *es = NULL; |
2543 | struct ext4_sb_info *sbi; | |
2544 | ext4_fsblk_t block; | |
2545 | ext4_fsblk_t sb_block = get_sb_block(&data); | |
70bbb3e0 | 2546 | ext4_fsblk_t logical_sb_block; |
ac27a0ec | 2547 | unsigned long offset = 0; |
ac27a0ec DK |
2548 | unsigned long journal_devnum = 0; |
2549 | unsigned long def_mount_opts; | |
2550 | struct inode *root; | |
9f6200bb | 2551 | char *cp; |
0390131b | 2552 | const char *descr; |
dcc7dae3 | 2553 | int ret = -ENOMEM; |
ac27a0ec | 2554 | int blocksize; |
4ec11028 TT |
2555 | unsigned int db_count; |
2556 | unsigned int i; | |
f287a1a5 | 2557 | int needs_recovery, has_huge_files; |
bd81d8ee | 2558 | __u64 blocks_count; |
833f4077 | 2559 | int err; |
b3881f74 | 2560 | unsigned int journal_ioprio = DEFAULT_JOURNAL_IOPRIO; |
ac27a0ec DK |
2561 | |
2562 | sbi = kzalloc(sizeof(*sbi), GFP_KERNEL); | |
2563 | if (!sbi) | |
dcc7dae3 | 2564 | goto out_free_orig; |
705895b6 PE |
2565 | |
2566 | sbi->s_blockgroup_lock = | |
2567 | kzalloc(sizeof(struct blockgroup_lock), GFP_KERNEL); | |
2568 | if (!sbi->s_blockgroup_lock) { | |
2569 | kfree(sbi); | |
dcc7dae3 | 2570 | goto out_free_orig; |
705895b6 | 2571 | } |
ac27a0ec DK |
2572 | sb->s_fs_info = sbi; |
2573 | sbi->s_mount_opt = 0; | |
617ba13b MC |
2574 | sbi->s_resuid = EXT4_DEF_RESUID; |
2575 | sbi->s_resgid = EXT4_DEF_RESGID; | |
240799cd | 2576 | sbi->s_inode_readahead_blks = EXT4_DEF_INODE_READAHEAD_BLKS; |
d9c9bef1 | 2577 | sbi->s_sb_block = sb_block; |
afc32f7e TT |
2578 | sbi->s_sectors_written_start = part_stat_read(sb->s_bdev->bd_part, |
2579 | sectors[1]); | |
ac27a0ec DK |
2580 | |
2581 | unlock_kernel(); | |
2582 | ||
9f6200bb TT |
2583 | /* Cleanup superblock name */ |
2584 | for (cp = sb->s_id; (cp = strchr(cp, '/'));) | |
2585 | *cp = '!'; | |
2586 | ||
dcc7dae3 | 2587 | ret = -EINVAL; |
617ba13b | 2588 | blocksize = sb_min_blocksize(sb, EXT4_MIN_BLOCK_SIZE); |
ac27a0ec | 2589 | if (!blocksize) { |
b31e1552 | 2590 | ext4_msg(sb, KERN_ERR, "unable to set blocksize"); |
ac27a0ec DK |
2591 | goto out_fail; |
2592 | } | |
2593 | ||
2594 | /* | |
617ba13b | 2595 | * The ext4 superblock will not be buffer aligned for other than 1kB |
ac27a0ec DK |
2596 | * block sizes. We need to calculate the offset from buffer start. |
2597 | */ | |
617ba13b | 2598 | if (blocksize != EXT4_MIN_BLOCK_SIZE) { |
70bbb3e0 AM |
2599 | logical_sb_block = sb_block * EXT4_MIN_BLOCK_SIZE; |
2600 | offset = do_div(logical_sb_block, blocksize); | |
ac27a0ec | 2601 | } else { |
70bbb3e0 | 2602 | logical_sb_block = sb_block; |
ac27a0ec DK |
2603 | } |
2604 | ||
70bbb3e0 | 2605 | if (!(bh = sb_bread(sb, logical_sb_block))) { |
b31e1552 | 2606 | ext4_msg(sb, KERN_ERR, "unable to read superblock"); |
ac27a0ec DK |
2607 | goto out_fail; |
2608 | } | |
2609 | /* | |
2610 | * Note: s_es must be initialized as soon as possible because | |
617ba13b | 2611 | * some ext4 macro-instructions depend on its value |
ac27a0ec | 2612 | */ |
617ba13b | 2613 | es = (struct ext4_super_block *) (((char *)bh->b_data) + offset); |
ac27a0ec DK |
2614 | sbi->s_es = es; |
2615 | sb->s_magic = le16_to_cpu(es->s_magic); | |
617ba13b MC |
2616 | if (sb->s_magic != EXT4_SUPER_MAGIC) |
2617 | goto cantfind_ext4; | |
afc32f7e | 2618 | sbi->s_kbytes_written = le64_to_cpu(es->s_kbytes_written); |
ac27a0ec DK |
2619 | |
2620 | /* Set defaults before we parse the mount options */ | |
2621 | def_mount_opts = le32_to_cpu(es->s_default_mount_opts); | |
617ba13b | 2622 | if (def_mount_opts & EXT4_DEFM_DEBUG) |
ac27a0ec | 2623 | set_opt(sbi->s_mount_opt, DEBUG); |
437ca0fd DM |
2624 | if (def_mount_opts & EXT4_DEFM_BSDGROUPS) { |
2625 | ext4_msg(sb, KERN_WARNING, deprecated_msg, "bsdgroups", | |
2626 | "2.6.38"); | |
ac27a0ec | 2627 | set_opt(sbi->s_mount_opt, GRPID); |
437ca0fd | 2628 | } |
617ba13b | 2629 | if (def_mount_opts & EXT4_DEFM_UID16) |
ac27a0ec | 2630 | set_opt(sbi->s_mount_opt, NO_UID32); |
03010a33 | 2631 | #ifdef CONFIG_EXT4_FS_XATTR |
617ba13b | 2632 | if (def_mount_opts & EXT4_DEFM_XATTR_USER) |
ac27a0ec | 2633 | set_opt(sbi->s_mount_opt, XATTR_USER); |
2e7842b8 | 2634 | #endif |
03010a33 | 2635 | #ifdef CONFIG_EXT4_FS_POSIX_ACL |
617ba13b | 2636 | if (def_mount_opts & EXT4_DEFM_ACL) |
ac27a0ec | 2637 | set_opt(sbi->s_mount_opt, POSIX_ACL); |
2e7842b8 | 2638 | #endif |
617ba13b | 2639 | if ((def_mount_opts & EXT4_DEFM_JMODE) == EXT4_DEFM_JMODE_DATA) |
482a7425 | 2640 | set_opt(sbi->s_mount_opt, JOURNAL_DATA); |
617ba13b | 2641 | else if ((def_mount_opts & EXT4_DEFM_JMODE) == EXT4_DEFM_JMODE_ORDERED) |
482a7425 | 2642 | set_opt(sbi->s_mount_opt, ORDERED_DATA); |
617ba13b | 2643 | else if ((def_mount_opts & EXT4_DEFM_JMODE) == EXT4_DEFM_JMODE_WBACK) |
482a7425 | 2644 | set_opt(sbi->s_mount_opt, WRITEBACK_DATA); |
617ba13b MC |
2645 | |
2646 | if (le16_to_cpu(sbi->s_es->s_errors) == EXT4_ERRORS_PANIC) | |
ac27a0ec | 2647 | set_opt(sbi->s_mount_opt, ERRORS_PANIC); |
bb4f397a | 2648 | else if (le16_to_cpu(sbi->s_es->s_errors) == EXT4_ERRORS_CONTINUE) |
ceea16bf | 2649 | set_opt(sbi->s_mount_opt, ERRORS_CONT); |
bb4f397a AK |
2650 | else |
2651 | set_opt(sbi->s_mount_opt, ERRORS_RO); | |
ac27a0ec DK |
2652 | |
2653 | sbi->s_resuid = le16_to_cpu(es->s_def_resuid); | |
2654 | sbi->s_resgid = le16_to_cpu(es->s_def_resgid); | |
30773840 TT |
2655 | sbi->s_commit_interval = JBD2_DEFAULT_MAX_COMMIT_AGE * HZ; |
2656 | sbi->s_min_batch_time = EXT4_DEF_MIN_BATCH_TIME; | |
2657 | sbi->s_max_batch_time = EXT4_DEF_MAX_BATCH_TIME; | |
ac27a0ec | 2658 | |
571640ca | 2659 | set_opt(sbi->s_mount_opt, BARRIER); |
ac27a0ec | 2660 | |
dd919b98 AK |
2661 | /* |
2662 | * enable delayed allocation by default | |
2663 | * Use -o nodelalloc to turn it off | |
2664 | */ | |
ba69f9ab JK |
2665 | if (!IS_EXT3_SB(sb)) |
2666 | set_opt(sbi->s_mount_opt, DELALLOC); | |
dd919b98 | 2667 | |
b3881f74 TT |
2668 | if (!parse_options((char *) data, sb, &journal_devnum, |
2669 | &journal_ioprio, NULL, 0)) | |
ac27a0ec DK |
2670 | goto failed_mount; |
2671 | ||
2672 | sb->s_flags = (sb->s_flags & ~MS_POSIXACL) | | |
482a7425 | 2673 | (test_opt(sb, POSIX_ACL) ? MS_POSIXACL : 0); |
ac27a0ec | 2674 | |
617ba13b MC |
2675 | if (le32_to_cpu(es->s_rev_level) == EXT4_GOOD_OLD_REV && |
2676 | (EXT4_HAS_COMPAT_FEATURE(sb, ~0U) || | |
2677 | EXT4_HAS_RO_COMPAT_FEATURE(sb, ~0U) || | |
2678 | EXT4_HAS_INCOMPAT_FEATURE(sb, ~0U))) | |
b31e1552 ES |
2679 | ext4_msg(sb, KERN_WARNING, |
2680 | "feature flags set on rev 0 fs, " | |
2681 | "running e2fsck is recommended"); | |
469108ff | 2682 | |
ac27a0ec DK |
2683 | /* |
2684 | * Check feature flags regardless of the revision level, since we | |
2685 | * previously didn't change the revision level when setting the flags, | |
2686 | * so there is a chance incompat flags are set on a rev 0 filesystem. | |
2687 | */ | |
a13fb1a4 | 2688 | if (!ext4_feature_set_ok(sb, (sb->s_flags & MS_RDONLY))) |
ac27a0ec | 2689 | goto failed_mount; |
a13fb1a4 | 2690 | |
ac27a0ec DK |
2691 | blocksize = BLOCK_SIZE << le32_to_cpu(es->s_log_block_size); |
2692 | ||
617ba13b MC |
2693 | if (blocksize < EXT4_MIN_BLOCK_SIZE || |
2694 | blocksize > EXT4_MAX_BLOCK_SIZE) { | |
b31e1552 ES |
2695 | ext4_msg(sb, KERN_ERR, |
2696 | "Unsupported filesystem blocksize %d", blocksize); | |
ac27a0ec DK |
2697 | goto failed_mount; |
2698 | } | |
2699 | ||
ac27a0ec | 2700 | if (sb->s_blocksize != blocksize) { |
ce40733c AK |
2701 | /* Validate the filesystem blocksize */ |
2702 | if (!sb_set_blocksize(sb, blocksize)) { | |
b31e1552 | 2703 | ext4_msg(sb, KERN_ERR, "bad block size %d", |
ce40733c | 2704 | blocksize); |
ac27a0ec DK |
2705 | goto failed_mount; |
2706 | } | |
2707 | ||
2b2d6d01 | 2708 | brelse(bh); |
70bbb3e0 AM |
2709 | logical_sb_block = sb_block * EXT4_MIN_BLOCK_SIZE; |
2710 | offset = do_div(logical_sb_block, blocksize); | |
2711 | bh = sb_bread(sb, logical_sb_block); | |
ac27a0ec | 2712 | if (!bh) { |
b31e1552 ES |
2713 | ext4_msg(sb, KERN_ERR, |
2714 | "Can't read superblock on 2nd try"); | |
ac27a0ec DK |
2715 | goto failed_mount; |
2716 | } | |
617ba13b | 2717 | es = (struct ext4_super_block *)(((char *)bh->b_data) + offset); |
ac27a0ec | 2718 | sbi->s_es = es; |
617ba13b | 2719 | if (es->s_magic != cpu_to_le16(EXT4_SUPER_MAGIC)) { |
b31e1552 ES |
2720 | ext4_msg(sb, KERN_ERR, |
2721 | "Magic mismatch, very weird!"); | |
ac27a0ec DK |
2722 | goto failed_mount; |
2723 | } | |
2724 | } | |
2725 | ||
a13fb1a4 ES |
2726 | has_huge_files = EXT4_HAS_RO_COMPAT_FEATURE(sb, |
2727 | EXT4_FEATURE_RO_COMPAT_HUGE_FILE); | |
f287a1a5 TT |
2728 | sbi->s_bitmap_maxbytes = ext4_max_bitmap_size(sb->s_blocksize_bits, |
2729 | has_huge_files); | |
2730 | sb->s_maxbytes = ext4_max_size(sb->s_blocksize_bits, has_huge_files); | |
ac27a0ec | 2731 | |
617ba13b MC |
2732 | if (le32_to_cpu(es->s_rev_level) == EXT4_GOOD_OLD_REV) { |
2733 | sbi->s_inode_size = EXT4_GOOD_OLD_INODE_SIZE; | |
2734 | sbi->s_first_ino = EXT4_GOOD_OLD_FIRST_INO; | |
ac27a0ec DK |
2735 | } else { |
2736 | sbi->s_inode_size = le16_to_cpu(es->s_inode_size); | |
2737 | sbi->s_first_ino = le32_to_cpu(es->s_first_ino); | |
617ba13b | 2738 | if ((sbi->s_inode_size < EXT4_GOOD_OLD_INODE_SIZE) || |
1330593e | 2739 | (!is_power_of_2(sbi->s_inode_size)) || |
ac27a0ec | 2740 | (sbi->s_inode_size > blocksize)) { |
b31e1552 ES |
2741 | ext4_msg(sb, KERN_ERR, |
2742 | "unsupported inode size: %d", | |
2b2d6d01 | 2743 | sbi->s_inode_size); |
ac27a0ec DK |
2744 | goto failed_mount; |
2745 | } | |
ef7f3835 KS |
2746 | if (sbi->s_inode_size > EXT4_GOOD_OLD_INODE_SIZE) |
2747 | sb->s_time_gran = 1 << (EXT4_EPOCH_BITS - 2); | |
ac27a0ec | 2748 | } |
0b8e58a1 | 2749 | |
0d1ee42f AR |
2750 | sbi->s_desc_size = le16_to_cpu(es->s_desc_size); |
2751 | if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_64BIT)) { | |
8fadc143 | 2752 | if (sbi->s_desc_size < EXT4_MIN_DESC_SIZE_64BIT || |
0d1ee42f | 2753 | sbi->s_desc_size > EXT4_MAX_DESC_SIZE || |
d8ea6cf8 | 2754 | !is_power_of_2(sbi->s_desc_size)) { |
b31e1552 ES |
2755 | ext4_msg(sb, KERN_ERR, |
2756 | "unsupported descriptor size %lu", | |
0d1ee42f AR |
2757 | sbi->s_desc_size); |
2758 | goto failed_mount; | |
2759 | } | |
2760 | } else | |
2761 | sbi->s_desc_size = EXT4_MIN_DESC_SIZE; | |
0b8e58a1 | 2762 | |
ac27a0ec | 2763 | sbi->s_blocks_per_group = le32_to_cpu(es->s_blocks_per_group); |
ac27a0ec | 2764 | sbi->s_inodes_per_group = le32_to_cpu(es->s_inodes_per_group); |
b47b6f38 | 2765 | if (EXT4_INODE_SIZE(sb) == 0 || EXT4_INODES_PER_GROUP(sb) == 0) |
617ba13b | 2766 | goto cantfind_ext4; |
0b8e58a1 | 2767 | |
617ba13b | 2768 | sbi->s_inodes_per_block = blocksize / EXT4_INODE_SIZE(sb); |
ac27a0ec | 2769 | if (sbi->s_inodes_per_block == 0) |
617ba13b | 2770 | goto cantfind_ext4; |
ac27a0ec DK |
2771 | sbi->s_itb_per_group = sbi->s_inodes_per_group / |
2772 | sbi->s_inodes_per_block; | |
0d1ee42f | 2773 | sbi->s_desc_per_block = blocksize / EXT4_DESC_SIZE(sb); |
ac27a0ec DK |
2774 | sbi->s_sbh = bh; |
2775 | sbi->s_mount_state = le16_to_cpu(es->s_state); | |
e57aa839 FW |
2776 | sbi->s_addr_per_block_bits = ilog2(EXT4_ADDR_PER_BLOCK(sb)); |
2777 | sbi->s_desc_per_block_bits = ilog2(EXT4_DESC_PER_BLOCK(sb)); | |
0b8e58a1 | 2778 | |
2b2d6d01 | 2779 | for (i = 0; i < 4; i++) |
ac27a0ec DK |
2780 | sbi->s_hash_seed[i] = le32_to_cpu(es->s_hash_seed[i]); |
2781 | sbi->s_def_hash_version = es->s_def_hash_version; | |
f99b2589 TT |
2782 | i = le32_to_cpu(es->s_flags); |
2783 | if (i & EXT2_FLAGS_UNSIGNED_HASH) | |
2784 | sbi->s_hash_unsigned = 3; | |
2785 | else if ((i & EXT2_FLAGS_SIGNED_HASH) == 0) { | |
2786 | #ifdef __CHAR_UNSIGNED__ | |
2787 | es->s_flags |= cpu_to_le32(EXT2_FLAGS_UNSIGNED_HASH); | |
2788 | sbi->s_hash_unsigned = 3; | |
2789 | #else | |
2790 | es->s_flags |= cpu_to_le32(EXT2_FLAGS_SIGNED_HASH); | |
2791 | #endif | |
2792 | sb->s_dirt = 1; | |
2793 | } | |
ac27a0ec DK |
2794 | |
2795 | if (sbi->s_blocks_per_group > blocksize * 8) { | |
b31e1552 ES |
2796 | ext4_msg(sb, KERN_ERR, |
2797 | "#blocks per group too big: %lu", | |
2b2d6d01 | 2798 | sbi->s_blocks_per_group); |
ac27a0ec DK |
2799 | goto failed_mount; |
2800 | } | |
ac27a0ec | 2801 | if (sbi->s_inodes_per_group > blocksize * 8) { |
b31e1552 ES |
2802 | ext4_msg(sb, KERN_ERR, |
2803 | "#inodes per group too big: %lu", | |
2b2d6d01 | 2804 | sbi->s_inodes_per_group); |
ac27a0ec DK |
2805 | goto failed_mount; |
2806 | } | |
2807 | ||
bf43d84b ES |
2808 | /* |
2809 | * Test whether we have more sectors than will fit in sector_t, | |
2810 | * and whether the max offset is addressable by the page cache. | |
2811 | */ | |
2812 | if ((ext4_blocks_count(es) > | |
2813 | (sector_t)(~0ULL) >> (sb->s_blocksize_bits - 9)) || | |
2814 | (ext4_blocks_count(es) > | |
2815 | (pgoff_t)(~0ULL) >> (PAGE_CACHE_SHIFT - sb->s_blocksize_bits))) { | |
b31e1552 | 2816 | ext4_msg(sb, KERN_ERR, "filesystem" |
bf43d84b | 2817 | " too large to mount safely on this system"); |
ac27a0ec | 2818 | if (sizeof(sector_t) < 8) |
90c699a9 | 2819 | ext4_msg(sb, KERN_WARNING, "CONFIG_LBDAF not enabled"); |
bf43d84b | 2820 | ret = -EFBIG; |
ac27a0ec DK |
2821 | goto failed_mount; |
2822 | } | |
2823 | ||
617ba13b MC |
2824 | if (EXT4_BLOCKS_PER_GROUP(sb) == 0) |
2825 | goto cantfind_ext4; | |
e7c95593 | 2826 | |
0f2ddca6 FTN |
2827 | /* check blocks count against device size */ |
2828 | blocks_count = sb->s_bdev->bd_inode->i_size >> sb->s_blocksize_bits; | |
2829 | if (blocks_count && ext4_blocks_count(es) > blocks_count) { | |
b31e1552 ES |
2830 | ext4_msg(sb, KERN_WARNING, "bad geometry: block count %llu " |
2831 | "exceeds size of device (%llu blocks)", | |
0f2ddca6 FTN |
2832 | ext4_blocks_count(es), blocks_count); |
2833 | goto failed_mount; | |
2834 | } | |
2835 | ||
0b8e58a1 AD |
2836 | /* |
2837 | * It makes no sense for the first data block to be beyond the end | |
2838 | * of the filesystem. | |
2839 | */ | |
2840 | if (le32_to_cpu(es->s_first_data_block) >= ext4_blocks_count(es)) { | |
b31e1552 ES |
2841 | ext4_msg(sb, KERN_WARNING, "bad geometry: first data" |
2842 | "block %u is beyond end of filesystem (%llu)", | |
2843 | le32_to_cpu(es->s_first_data_block), | |
2844 | ext4_blocks_count(es)); | |
e7c95593 ES |
2845 | goto failed_mount; |
2846 | } | |
bd81d8ee LV |
2847 | blocks_count = (ext4_blocks_count(es) - |
2848 | le32_to_cpu(es->s_first_data_block) + | |
2849 | EXT4_BLOCKS_PER_GROUP(sb) - 1); | |
2850 | do_div(blocks_count, EXT4_BLOCKS_PER_GROUP(sb)); | |
4ec11028 | 2851 | if (blocks_count > ((uint64_t)1<<32) - EXT4_DESC_PER_BLOCK(sb)) { |
b31e1552 | 2852 | ext4_msg(sb, KERN_WARNING, "groups count too large: %u " |
4ec11028 | 2853 | "(block count %llu, first data block %u, " |
b31e1552 | 2854 | "blocks per group %lu)", sbi->s_groups_count, |
4ec11028 TT |
2855 | ext4_blocks_count(es), |
2856 | le32_to_cpu(es->s_first_data_block), | |
2857 | EXT4_BLOCKS_PER_GROUP(sb)); | |
2858 | goto failed_mount; | |
2859 | } | |
bd81d8ee | 2860 | sbi->s_groups_count = blocks_count; |
fb0a387d ES |
2861 | sbi->s_blockfile_groups = min_t(ext4_group_t, sbi->s_groups_count, |
2862 | (EXT4_MAX_BLOCK_FILE_PHYS / EXT4_BLOCKS_PER_GROUP(sb))); | |
617ba13b MC |
2863 | db_count = (sbi->s_groups_count + EXT4_DESC_PER_BLOCK(sb) - 1) / |
2864 | EXT4_DESC_PER_BLOCK(sb); | |
2b2d6d01 | 2865 | sbi->s_group_desc = kmalloc(db_count * sizeof(struct buffer_head *), |
ac27a0ec DK |
2866 | GFP_KERNEL); |
2867 | if (sbi->s_group_desc == NULL) { | |
b31e1552 | 2868 | ext4_msg(sb, KERN_ERR, "not enough memory"); |
ac27a0ec DK |
2869 | goto failed_mount; |
2870 | } | |
2871 | ||
3244fcb1 | 2872 | #ifdef CONFIG_PROC_FS |
9f6200bb TT |
2873 | if (ext4_proc_root) |
2874 | sbi->s_proc = proc_mkdir(sb->s_id, ext4_proc_root); | |
3244fcb1 | 2875 | #endif |
240799cd | 2876 | |
705895b6 | 2877 | bgl_lock_init(sbi->s_blockgroup_lock); |
ac27a0ec DK |
2878 | |
2879 | for (i = 0; i < db_count; i++) { | |
70bbb3e0 | 2880 | block = descriptor_loc(sb, logical_sb_block, i); |
ac27a0ec DK |
2881 | sbi->s_group_desc[i] = sb_bread(sb, block); |
2882 | if (!sbi->s_group_desc[i]) { | |
b31e1552 ES |
2883 | ext4_msg(sb, KERN_ERR, |
2884 | "can't read group descriptor %d", i); | |
ac27a0ec DK |
2885 | db_count = i; |
2886 | goto failed_mount2; | |
2887 | } | |
2888 | } | |
2b2d6d01 | 2889 | if (!ext4_check_descriptors(sb)) { |
b31e1552 | 2890 | ext4_msg(sb, KERN_ERR, "group descriptors corrupted!"); |
ac27a0ec DK |
2891 | goto failed_mount2; |
2892 | } | |
772cb7c8 JS |
2893 | if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FLEX_BG)) |
2894 | if (!ext4_fill_flex_info(sb)) { | |
b31e1552 ES |
2895 | ext4_msg(sb, KERN_ERR, |
2896 | "unable to initialize " | |
2897 | "flex_bg meta info!"); | |
772cb7c8 JS |
2898 | goto failed_mount2; |
2899 | } | |
2900 | ||
ac27a0ec DK |
2901 | sbi->s_gdb_count = db_count; |
2902 | get_random_bytes(&sbi->s_next_generation, sizeof(u32)); | |
2903 | spin_lock_init(&sbi->s_next_gen_lock); | |
2904 | ||
c9de560d | 2905 | sbi->s_stripe = ext4_get_stripe_size(sbi); |
55138e0b | 2906 | sbi->s_max_writeback_mb_bump = 128; |
c9de560d | 2907 | |
ac27a0ec DK |
2908 | /* |
2909 | * set up enough so that it can read an inode | |
2910 | */ | |
9ca92389 TT |
2911 | if (!test_opt(sb, NOLOAD) && |
2912 | EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL)) | |
2913 | sb->s_op = &ext4_sops; | |
2914 | else | |
2915 | sb->s_op = &ext4_nojournal_sops; | |
617ba13b MC |
2916 | sb->s_export_op = &ext4_export_ops; |
2917 | sb->s_xattr = ext4_xattr_handlers; | |
ac27a0ec | 2918 | #ifdef CONFIG_QUOTA |
617ba13b MC |
2919 | sb->s_qcop = &ext4_qctl_operations; |
2920 | sb->dq_op = &ext4_quota_operations; | |
ac27a0ec DK |
2921 | #endif |
2922 | INIT_LIST_HEAD(&sbi->s_orphan); /* unlinked but open files */ | |
3b9d4ed2 | 2923 | mutex_init(&sbi->s_orphan_lock); |
32ed5058 | 2924 | mutex_init(&sbi->s_resize_lock); |
ac27a0ec DK |
2925 | |
2926 | sb->s_root = NULL; | |
2927 | ||
2928 | needs_recovery = (es->s_last_orphan != 0 || | |
617ba13b MC |
2929 | EXT4_HAS_INCOMPAT_FEATURE(sb, |
2930 | EXT4_FEATURE_INCOMPAT_RECOVER)); | |
ac27a0ec DK |
2931 | |
2932 | /* | |
2933 | * The first inode we look at is the journal inode. Don't try | |
2934 | * root first: it may be modified in the journal! | |
2935 | */ | |
2936 | if (!test_opt(sb, NOLOAD) && | |
617ba13b MC |
2937 | EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL)) { |
2938 | if (ext4_load_journal(sb, es, journal_devnum)) | |
ac27a0ec | 2939 | goto failed_mount3; |
0390131b FM |
2940 | } else if (test_opt(sb, NOLOAD) && !(sb->s_flags & MS_RDONLY) && |
2941 | EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER)) { | |
b31e1552 ES |
2942 | ext4_msg(sb, KERN_ERR, "required journal recovery " |
2943 | "suppressed and not mounted read-only"); | |
744692dc | 2944 | goto failed_mount_wq; |
ac27a0ec | 2945 | } else { |
0390131b FM |
2946 | clear_opt(sbi->s_mount_opt, DATA_FLAGS); |
2947 | set_opt(sbi->s_mount_opt, WRITEBACK_DATA); | |
2948 | sbi->s_journal = NULL; | |
2949 | needs_recovery = 0; | |
2950 | goto no_journal; | |
ac27a0ec DK |
2951 | } |
2952 | ||
eb40a09c JS |
2953 | if (ext4_blocks_count(es) > 0xffffffffULL && |
2954 | !jbd2_journal_set_features(EXT4_SB(sb)->s_journal, 0, 0, | |
2955 | JBD2_FEATURE_INCOMPAT_64BIT)) { | |
b31e1552 | 2956 | ext4_msg(sb, KERN_ERR, "Failed to set 64-bit journal feature"); |
744692dc | 2957 | goto failed_mount_wq; |
eb40a09c JS |
2958 | } |
2959 | ||
d4da6c9c LT |
2960 | if (test_opt(sb, JOURNAL_ASYNC_COMMIT)) { |
2961 | jbd2_journal_set_features(sbi->s_journal, | |
2962 | JBD2_FEATURE_COMPAT_CHECKSUM, 0, | |
818d276c | 2963 | JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT); |
d4da6c9c LT |
2964 | } else if (test_opt(sb, JOURNAL_CHECKSUM)) { |
2965 | jbd2_journal_set_features(sbi->s_journal, | |
2966 | JBD2_FEATURE_COMPAT_CHECKSUM, 0, 0); | |
818d276c GS |
2967 | jbd2_journal_clear_features(sbi->s_journal, 0, 0, |
2968 | JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT); | |
d4da6c9c LT |
2969 | } else { |
2970 | jbd2_journal_clear_features(sbi->s_journal, | |
2971 | JBD2_FEATURE_COMPAT_CHECKSUM, 0, | |
2972 | JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT); | |
2973 | } | |
818d276c | 2974 | |
ac27a0ec DK |
2975 | /* We have now updated the journal if required, so we can |
2976 | * validate the data journaling mode. */ | |
2977 | switch (test_opt(sb, DATA_FLAGS)) { | |
2978 | case 0: | |
2979 | /* No mode set, assume a default based on the journal | |
63f57933 AM |
2980 | * capabilities: ORDERED_DATA if the journal can |
2981 | * cope, else JOURNAL_DATA | |
2982 | */ | |
dab291af MC |
2983 | if (jbd2_journal_check_available_features |
2984 | (sbi->s_journal, 0, 0, JBD2_FEATURE_INCOMPAT_REVOKE)) | |
ac27a0ec DK |
2985 | set_opt(sbi->s_mount_opt, ORDERED_DATA); |
2986 | else | |
2987 | set_opt(sbi->s_mount_opt, JOURNAL_DATA); | |
2988 | break; | |
2989 | ||
617ba13b MC |
2990 | case EXT4_MOUNT_ORDERED_DATA: |
2991 | case EXT4_MOUNT_WRITEBACK_DATA: | |
dab291af MC |
2992 | if (!jbd2_journal_check_available_features |
2993 | (sbi->s_journal, 0, 0, JBD2_FEATURE_INCOMPAT_REVOKE)) { | |
b31e1552 ES |
2994 | ext4_msg(sb, KERN_ERR, "Journal does not support " |
2995 | "requested data journaling mode"); | |
744692dc | 2996 | goto failed_mount_wq; |
ac27a0ec DK |
2997 | } |
2998 | default: | |
2999 | break; | |
3000 | } | |
b3881f74 | 3001 | set_task_ioprio(sbi->s_journal->j_task, journal_ioprio); |
ac27a0ec | 3002 | |
0390131b | 3003 | no_journal: |
84061e07 DM |
3004 | err = percpu_counter_init(&sbi->s_freeblocks_counter, |
3005 | ext4_count_free_blocks(sb)); | |
3006 | if (!err) | |
3007 | err = percpu_counter_init(&sbi->s_freeinodes_counter, | |
3008 | ext4_count_free_inodes(sb)); | |
3009 | if (!err) | |
3010 | err = percpu_counter_init(&sbi->s_dirs_counter, | |
3011 | ext4_count_dirs(sb)); | |
3012 | if (!err) | |
3013 | err = percpu_counter_init(&sbi->s_dirtyblocks_counter, 0); | |
3014 | if (err) { | |
3015 | ext4_msg(sb, KERN_ERR, "insufficient memory"); | |
3016 | goto failed_mount_wq; | |
3017 | } | |
206f7ab4 | 3018 | |
4c0425ff MC |
3019 | EXT4_SB(sb)->dio_unwritten_wq = create_workqueue("ext4-dio-unwritten"); |
3020 | if (!EXT4_SB(sb)->dio_unwritten_wq) { | |
3021 | printk(KERN_ERR "EXT4-fs: failed to create DIO workqueue\n"); | |
3022 | goto failed_mount_wq; | |
3023 | } | |
3024 | ||
ac27a0ec | 3025 | /* |
dab291af | 3026 | * The jbd2_journal_load will have done any necessary log recovery, |
ac27a0ec DK |
3027 | * so we can safely mount the rest of the filesystem now. |
3028 | */ | |
3029 | ||
1d1fe1ee DH |
3030 | root = ext4_iget(sb, EXT4_ROOT_INO); |
3031 | if (IS_ERR(root)) { | |
b31e1552 | 3032 | ext4_msg(sb, KERN_ERR, "get root inode failed"); |
1d1fe1ee | 3033 | ret = PTR_ERR(root); |
ac27a0ec DK |
3034 | goto failed_mount4; |
3035 | } | |
3036 | if (!S_ISDIR(root->i_mode) || !root->i_blocks || !root->i_size) { | |
1d1fe1ee | 3037 | iput(root); |
b31e1552 | 3038 | ext4_msg(sb, KERN_ERR, "corrupt root inode, run e2fsck"); |
ac27a0ec DK |
3039 | goto failed_mount4; |
3040 | } | |
1d1fe1ee DH |
3041 | sb->s_root = d_alloc_root(root); |
3042 | if (!sb->s_root) { | |
b31e1552 | 3043 | ext4_msg(sb, KERN_ERR, "get root dentry failed"); |
1d1fe1ee DH |
3044 | iput(root); |
3045 | ret = -ENOMEM; | |
3046 | goto failed_mount4; | |
3047 | } | |
ac27a0ec | 3048 | |
2b2d6d01 | 3049 | ext4_setup_super(sb, es, sb->s_flags & MS_RDONLY); |
ef7f3835 KS |
3050 | |
3051 | /* determine the minimum size of new large inodes, if present */ | |
3052 | if (sbi->s_inode_size > EXT4_GOOD_OLD_INODE_SIZE) { | |
3053 | sbi->s_want_extra_isize = sizeof(struct ext4_inode) - | |
3054 | EXT4_GOOD_OLD_INODE_SIZE; | |
3055 | if (EXT4_HAS_RO_COMPAT_FEATURE(sb, | |
3056 | EXT4_FEATURE_RO_COMPAT_EXTRA_ISIZE)) { | |
3057 | if (sbi->s_want_extra_isize < | |
3058 | le16_to_cpu(es->s_want_extra_isize)) | |
3059 | sbi->s_want_extra_isize = | |
3060 | le16_to_cpu(es->s_want_extra_isize); | |
3061 | if (sbi->s_want_extra_isize < | |
3062 | le16_to_cpu(es->s_min_extra_isize)) | |
3063 | sbi->s_want_extra_isize = | |
3064 | le16_to_cpu(es->s_min_extra_isize); | |
3065 | } | |
3066 | } | |
3067 | /* Check if enough inode space is available */ | |
3068 | if (EXT4_GOOD_OLD_INODE_SIZE + sbi->s_want_extra_isize > | |
3069 | sbi->s_inode_size) { | |
3070 | sbi->s_want_extra_isize = sizeof(struct ext4_inode) - | |
3071 | EXT4_GOOD_OLD_INODE_SIZE; | |
b31e1552 ES |
3072 | ext4_msg(sb, KERN_INFO, "required extra inode space not" |
3073 | "available"); | |
ef7f3835 KS |
3074 | } |
3075 | ||
90576c0b TT |
3076 | if (test_opt(sb, DELALLOC) && |
3077 | (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_JOURNAL_DATA)) { | |
b31e1552 ES |
3078 | ext4_msg(sb, KERN_WARNING, "Ignoring delalloc option - " |
3079 | "requested data journaling mode"); | |
c2774d84 | 3080 | clear_opt(sbi->s_mount_opt, DELALLOC); |
90576c0b | 3081 | } |
744692dc JZ |
3082 | if (test_opt(sb, DIOREAD_NOLOCK)) { |
3083 | if (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_JOURNAL_DATA) { | |
3084 | ext4_msg(sb, KERN_WARNING, "Ignoring dioread_nolock " | |
3085 | "option - requested data journaling mode"); | |
3086 | clear_opt(sbi->s_mount_opt, DIOREAD_NOLOCK); | |
3087 | } | |
3088 | if (sb->s_blocksize < PAGE_SIZE) { | |
3089 | ext4_msg(sb, KERN_WARNING, "Ignoring dioread_nolock " | |
3090 | "option - block size is too small"); | |
3091 | clear_opt(sbi->s_mount_opt, DIOREAD_NOLOCK); | |
3092 | } | |
3093 | } | |
c2774d84 | 3094 | |
6fd058f7 TT |
3095 | err = ext4_setup_system_zone(sb); |
3096 | if (err) { | |
b31e1552 | 3097 | ext4_msg(sb, KERN_ERR, "failed to initialize system " |
fbe845dd | 3098 | "zone (%d)", err); |
6fd058f7 TT |
3099 | goto failed_mount4; |
3100 | } | |
3101 | ||
c2774d84 AK |
3102 | ext4_ext_init(sb); |
3103 | err = ext4_mb_init(sb, needs_recovery); | |
3104 | if (err) { | |
b31e1552 ES |
3105 | ext4_msg(sb, KERN_ERR, "failed to initalize mballoc (%d)", |
3106 | err); | |
c2774d84 AK |
3107 | goto failed_mount4; |
3108 | } | |
3109 | ||
3197ebdb TT |
3110 | sbi->s_kobj.kset = ext4_kset; |
3111 | init_completion(&sbi->s_kobj_unregister); | |
3112 | err = kobject_init_and_add(&sbi->s_kobj, &ext4_ktype, NULL, | |
3113 | "%s", sb->s_id); | |
3114 | if (err) { | |
3115 | ext4_mb_release(sb); | |
3116 | ext4_ext_release(sb); | |
3117 | goto failed_mount4; | |
3118 | }; | |
3119 | ||
617ba13b MC |
3120 | EXT4_SB(sb)->s_mount_state |= EXT4_ORPHAN_FS; |
3121 | ext4_orphan_cleanup(sb, es); | |
3122 | EXT4_SB(sb)->s_mount_state &= ~EXT4_ORPHAN_FS; | |
0390131b | 3123 | if (needs_recovery) { |
b31e1552 | 3124 | ext4_msg(sb, KERN_INFO, "recovery complete"); |
0390131b FM |
3125 | ext4_mark_recovery_complete(sb, es); |
3126 | } | |
3127 | if (EXT4_SB(sb)->s_journal) { | |
3128 | if (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_JOURNAL_DATA) | |
3129 | descr = " journalled data mode"; | |
3130 | else if (test_opt(sb, DATA_FLAGS) == EXT4_MOUNT_ORDERED_DATA) | |
3131 | descr = " ordered data mode"; | |
3132 | else | |
3133 | descr = " writeback data mode"; | |
3134 | } else | |
3135 | descr = "out journal"; | |
3136 | ||
d4c402d9 CW |
3137 | ext4_msg(sb, KERN_INFO, "mounted filesystem with%s. " |
3138 | "Opts: %s", descr, orig_data); | |
ac27a0ec | 3139 | |
66e61a9e TT |
3140 | init_timer(&sbi->s_err_report); |
3141 | sbi->s_err_report.function = print_daily_error_info; | |
3142 | sbi->s_err_report.data = (unsigned long) sb; | |
3143 | if (es->s_error_count) | |
3144 | mod_timer(&sbi->s_err_report, jiffies + 300*HZ); /* 5 minutes */ | |
3145 | ||
ac27a0ec | 3146 | lock_kernel(); |
d4c402d9 | 3147 | kfree(orig_data); |
ac27a0ec DK |
3148 | return 0; |
3149 | ||
617ba13b | 3150 | cantfind_ext4: |
ac27a0ec | 3151 | if (!silent) |
b31e1552 | 3152 | ext4_msg(sb, KERN_ERR, "VFS: Can't find ext4 filesystem"); |
ac27a0ec DK |
3153 | goto failed_mount; |
3154 | ||
3155 | failed_mount4: | |
b31e1552 | 3156 | ext4_msg(sb, KERN_ERR, "mount failed"); |
4c0425ff MC |
3157 | destroy_workqueue(EXT4_SB(sb)->dio_unwritten_wq); |
3158 | failed_mount_wq: | |
6fd058f7 | 3159 | ext4_release_system_zone(sb); |
0390131b FM |
3160 | if (sbi->s_journal) { |
3161 | jbd2_journal_destroy(sbi->s_journal); | |
3162 | sbi->s_journal = NULL; | |
3163 | } | |
84061e07 DM |
3164 | percpu_counter_destroy(&sbi->s_freeblocks_counter); |
3165 | percpu_counter_destroy(&sbi->s_freeinodes_counter); | |
3166 | percpu_counter_destroy(&sbi->s_dirs_counter); | |
3167 | percpu_counter_destroy(&sbi->s_dirtyblocks_counter); | |
ac27a0ec | 3168 | failed_mount3: |
c5ca7c76 TT |
3169 | if (sbi->s_flex_groups) { |
3170 | if (is_vmalloc_addr(sbi->s_flex_groups)) | |
3171 | vfree(sbi->s_flex_groups); | |
3172 | else | |
3173 | kfree(sbi->s_flex_groups); | |
3174 | } | |
ac27a0ec DK |
3175 | failed_mount2: |
3176 | for (i = 0; i < db_count; i++) | |
3177 | brelse(sbi->s_group_desc[i]); | |
3178 | kfree(sbi->s_group_desc); | |
3179 | failed_mount: | |
240799cd | 3180 | if (sbi->s_proc) { |
9f6200bb | 3181 | remove_proc_entry(sb->s_id, ext4_proc_root); |
240799cd | 3182 | } |
ac27a0ec DK |
3183 | #ifdef CONFIG_QUOTA |
3184 | for (i = 0; i < MAXQUOTAS; i++) | |
3185 | kfree(sbi->s_qf_names[i]); | |
3186 | #endif | |
617ba13b | 3187 | ext4_blkdev_remove(sbi); |
ac27a0ec DK |
3188 | brelse(bh); |
3189 | out_fail: | |
3190 | sb->s_fs_info = NULL; | |
f6830165 | 3191 | kfree(sbi->s_blockgroup_lock); |
ac27a0ec DK |
3192 | kfree(sbi); |
3193 | lock_kernel(); | |
dcc7dae3 | 3194 | out_free_orig: |
d4c402d9 | 3195 | kfree(orig_data); |
1d1fe1ee | 3196 | return ret; |
ac27a0ec DK |
3197 | } |
3198 | ||
3199 | /* | |
3200 | * Setup any per-fs journal parameters now. We'll do this both on | |
3201 | * initial mount, once the journal has been initialised but before we've | |
3202 | * done any recovery; and again on any subsequent remount. | |
3203 | */ | |
617ba13b | 3204 | static void ext4_init_journal_params(struct super_block *sb, journal_t *journal) |
ac27a0ec | 3205 | { |
617ba13b | 3206 | struct ext4_sb_info *sbi = EXT4_SB(sb); |
ac27a0ec | 3207 | |
30773840 TT |
3208 | journal->j_commit_interval = sbi->s_commit_interval; |
3209 | journal->j_min_batch_time = sbi->s_min_batch_time; | |
3210 | journal->j_max_batch_time = sbi->s_max_batch_time; | |
ac27a0ec DK |
3211 | |
3212 | spin_lock(&journal->j_state_lock); | |
3213 | if (test_opt(sb, BARRIER)) | |
dab291af | 3214 | journal->j_flags |= JBD2_BARRIER; |
ac27a0ec | 3215 | else |
dab291af | 3216 | journal->j_flags &= ~JBD2_BARRIER; |
5bf5683a HK |
3217 | if (test_opt(sb, DATA_ERR_ABORT)) |
3218 | journal->j_flags |= JBD2_ABORT_ON_SYNCDATA_ERR; | |
3219 | else | |
3220 | journal->j_flags &= ~JBD2_ABORT_ON_SYNCDATA_ERR; | |
ac27a0ec DK |
3221 | spin_unlock(&journal->j_state_lock); |
3222 | } | |
3223 | ||
617ba13b | 3224 | static journal_t *ext4_get_journal(struct super_block *sb, |
ac27a0ec DK |
3225 | unsigned int journal_inum) |
3226 | { | |
3227 | struct inode *journal_inode; | |
3228 | journal_t *journal; | |
3229 | ||
0390131b FM |
3230 | BUG_ON(!EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL)); |
3231 | ||
ac27a0ec DK |
3232 | /* First, test for the existence of a valid inode on disk. Bad |
3233 | * things happen if we iget() an unused inode, as the subsequent | |
3234 | * iput() will try to delete it. */ | |
3235 | ||
1d1fe1ee DH |
3236 | journal_inode = ext4_iget(sb, journal_inum); |
3237 | if (IS_ERR(journal_inode)) { | |
b31e1552 | 3238 | ext4_msg(sb, KERN_ERR, "no journal found"); |
ac27a0ec DK |
3239 | return NULL; |
3240 | } | |
3241 | if (!journal_inode->i_nlink) { | |
3242 | make_bad_inode(journal_inode); | |
3243 | iput(journal_inode); | |
b31e1552 | 3244 | ext4_msg(sb, KERN_ERR, "journal inode is deleted"); |
ac27a0ec DK |
3245 | return NULL; |
3246 | } | |
3247 | ||
e5f8eab8 | 3248 | jbd_debug(2, "Journal inode found at %p: %lld bytes\n", |
ac27a0ec | 3249 | journal_inode, journal_inode->i_size); |
1d1fe1ee | 3250 | if (!S_ISREG(journal_inode->i_mode)) { |
b31e1552 | 3251 | ext4_msg(sb, KERN_ERR, "invalid journal inode"); |
ac27a0ec DK |
3252 | iput(journal_inode); |
3253 | return NULL; | |
3254 | } | |
3255 | ||
dab291af | 3256 | journal = jbd2_journal_init_inode(journal_inode); |
ac27a0ec | 3257 | if (!journal) { |
b31e1552 | 3258 | ext4_msg(sb, KERN_ERR, "Could not load journal inode"); |
ac27a0ec DK |
3259 | iput(journal_inode); |
3260 | return NULL; | |
3261 | } | |
3262 | journal->j_private = sb; | |
617ba13b | 3263 | ext4_init_journal_params(sb, journal); |
ac27a0ec DK |
3264 | return journal; |
3265 | } | |
3266 | ||
617ba13b | 3267 | static journal_t *ext4_get_dev_journal(struct super_block *sb, |
ac27a0ec DK |
3268 | dev_t j_dev) |
3269 | { | |
2b2d6d01 | 3270 | struct buffer_head *bh; |
ac27a0ec | 3271 | journal_t *journal; |
617ba13b MC |
3272 | ext4_fsblk_t start; |
3273 | ext4_fsblk_t len; | |
ac27a0ec | 3274 | int hblock, blocksize; |
617ba13b | 3275 | ext4_fsblk_t sb_block; |
ac27a0ec | 3276 | unsigned long offset; |
2b2d6d01 | 3277 | struct ext4_super_block *es; |
ac27a0ec DK |
3278 | struct block_device *bdev; |
3279 | ||
0390131b FM |
3280 | BUG_ON(!EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL)); |
3281 | ||
b31e1552 | 3282 | bdev = ext4_blkdev_get(j_dev, sb); |
ac27a0ec DK |
3283 | if (bdev == NULL) |
3284 | return NULL; | |
3285 | ||
3286 | if (bd_claim(bdev, sb)) { | |
b31e1552 ES |
3287 | ext4_msg(sb, KERN_ERR, |
3288 | "failed to claim external journal device"); | |
9a1c3542 | 3289 | blkdev_put(bdev, FMODE_READ|FMODE_WRITE); |
ac27a0ec DK |
3290 | return NULL; |
3291 | } | |
3292 | ||
3293 | blocksize = sb->s_blocksize; | |
e1defc4f | 3294 | hblock = bdev_logical_block_size(bdev); |
ac27a0ec | 3295 | if (blocksize < hblock) { |
b31e1552 ES |
3296 | ext4_msg(sb, KERN_ERR, |
3297 | "blocksize too small for journal device"); | |
ac27a0ec DK |
3298 | goto out_bdev; |
3299 | } | |
3300 | ||
617ba13b MC |
3301 | sb_block = EXT4_MIN_BLOCK_SIZE / blocksize; |
3302 | offset = EXT4_MIN_BLOCK_SIZE % blocksize; | |
ac27a0ec DK |
3303 | set_blocksize(bdev, blocksize); |
3304 | if (!(bh = __bread(bdev, sb_block, blocksize))) { | |
b31e1552 ES |
3305 | ext4_msg(sb, KERN_ERR, "couldn't read superblock of " |
3306 | "external journal"); | |
ac27a0ec DK |
3307 | goto out_bdev; |
3308 | } | |
3309 | ||
617ba13b MC |
3310 | es = (struct ext4_super_block *) (((char *)bh->b_data) + offset); |
3311 | if ((le16_to_cpu(es->s_magic) != EXT4_SUPER_MAGIC) || | |
ac27a0ec | 3312 | !(le32_to_cpu(es->s_feature_incompat) & |
617ba13b | 3313 | EXT4_FEATURE_INCOMPAT_JOURNAL_DEV)) { |
b31e1552 ES |
3314 | ext4_msg(sb, KERN_ERR, "external journal has " |
3315 | "bad superblock"); | |
ac27a0ec DK |
3316 | brelse(bh); |
3317 | goto out_bdev; | |
3318 | } | |
3319 | ||
617ba13b | 3320 | if (memcmp(EXT4_SB(sb)->s_es->s_journal_uuid, es->s_uuid, 16)) { |
b31e1552 | 3321 | ext4_msg(sb, KERN_ERR, "journal UUID does not match"); |
ac27a0ec DK |
3322 | brelse(bh); |
3323 | goto out_bdev; | |
3324 | } | |
3325 | ||
bd81d8ee | 3326 | len = ext4_blocks_count(es); |
ac27a0ec DK |
3327 | start = sb_block + 1; |
3328 | brelse(bh); /* we're done with the superblock */ | |
3329 | ||
dab291af | 3330 | journal = jbd2_journal_init_dev(bdev, sb->s_bdev, |
ac27a0ec DK |
3331 | start, len, blocksize); |
3332 | if (!journal) { | |
b31e1552 | 3333 | ext4_msg(sb, KERN_ERR, "failed to create device journal"); |
ac27a0ec DK |
3334 | goto out_bdev; |
3335 | } | |
3336 | journal->j_private = sb; | |
3337 | ll_rw_block(READ, 1, &journal->j_sb_buffer); | |
3338 | wait_on_buffer(journal->j_sb_buffer); | |
3339 | if (!buffer_uptodate(journal->j_sb_buffer)) { | |
b31e1552 | 3340 | ext4_msg(sb, KERN_ERR, "I/O error on journal device"); |
ac27a0ec DK |
3341 | goto out_journal; |
3342 | } | |
3343 | if (be32_to_cpu(journal->j_superblock->s_nr_users) != 1) { | |
b31e1552 ES |
3344 | ext4_msg(sb, KERN_ERR, "External journal has more than one " |
3345 | "user (unsupported) - %d", | |
ac27a0ec DK |
3346 | be32_to_cpu(journal->j_superblock->s_nr_users)); |
3347 | goto out_journal; | |
3348 | } | |
617ba13b MC |
3349 | EXT4_SB(sb)->journal_bdev = bdev; |
3350 | ext4_init_journal_params(sb, journal); | |
ac27a0ec | 3351 | return journal; |
0b8e58a1 | 3352 | |
ac27a0ec | 3353 | out_journal: |
dab291af | 3354 | jbd2_journal_destroy(journal); |
ac27a0ec | 3355 | out_bdev: |
617ba13b | 3356 | ext4_blkdev_put(bdev); |
ac27a0ec DK |
3357 | return NULL; |
3358 | } | |
3359 | ||
617ba13b MC |
3360 | static int ext4_load_journal(struct super_block *sb, |
3361 | struct ext4_super_block *es, | |
ac27a0ec DK |
3362 | unsigned long journal_devnum) |
3363 | { | |
3364 | journal_t *journal; | |
3365 | unsigned int journal_inum = le32_to_cpu(es->s_journal_inum); | |
3366 | dev_t journal_dev; | |
3367 | int err = 0; | |
3368 | int really_read_only; | |
3369 | ||
0390131b FM |
3370 | BUG_ON(!EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL)); |
3371 | ||
ac27a0ec DK |
3372 | if (journal_devnum && |
3373 | journal_devnum != le32_to_cpu(es->s_journal_dev)) { | |
b31e1552 ES |
3374 | ext4_msg(sb, KERN_INFO, "external journal device major/minor " |
3375 | "numbers have changed"); | |
ac27a0ec DK |
3376 | journal_dev = new_decode_dev(journal_devnum); |
3377 | } else | |
3378 | journal_dev = new_decode_dev(le32_to_cpu(es->s_journal_dev)); | |
3379 | ||
3380 | really_read_only = bdev_read_only(sb->s_bdev); | |
3381 | ||
3382 | /* | |
3383 | * Are we loading a blank journal or performing recovery after a | |
3384 | * crash? For recovery, we need to check in advance whether we | |
3385 | * can get read-write access to the device. | |
3386 | */ | |
617ba13b | 3387 | if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER)) { |
ac27a0ec | 3388 | if (sb->s_flags & MS_RDONLY) { |
b31e1552 ES |
3389 | ext4_msg(sb, KERN_INFO, "INFO: recovery " |
3390 | "required on readonly filesystem"); | |
ac27a0ec | 3391 | if (really_read_only) { |
b31e1552 ES |
3392 | ext4_msg(sb, KERN_ERR, "write access " |
3393 | "unavailable, cannot proceed"); | |
ac27a0ec DK |
3394 | return -EROFS; |
3395 | } | |
b31e1552 ES |
3396 | ext4_msg(sb, KERN_INFO, "write access will " |
3397 | "be enabled during recovery"); | |
ac27a0ec DK |
3398 | } |
3399 | } | |
3400 | ||
3401 | if (journal_inum && journal_dev) { | |
b31e1552 ES |
3402 | ext4_msg(sb, KERN_ERR, "filesystem has both journal " |
3403 | "and inode journals!"); | |
ac27a0ec DK |
3404 | return -EINVAL; |
3405 | } | |
3406 | ||
3407 | if (journal_inum) { | |
617ba13b | 3408 | if (!(journal = ext4_get_journal(sb, journal_inum))) |
ac27a0ec DK |
3409 | return -EINVAL; |
3410 | } else { | |
617ba13b | 3411 | if (!(journal = ext4_get_dev_journal(sb, journal_dev))) |
ac27a0ec DK |
3412 | return -EINVAL; |
3413 | } | |
3414 | ||
90576c0b | 3415 | if (!(journal->j_flags & JBD2_BARRIER)) |
b31e1552 | 3416 | ext4_msg(sb, KERN_INFO, "barriers disabled"); |
4776004f | 3417 | |
ac27a0ec | 3418 | if (!really_read_only && test_opt(sb, UPDATE_JOURNAL)) { |
dab291af | 3419 | err = jbd2_journal_update_format(journal); |
ac27a0ec | 3420 | if (err) { |
b31e1552 | 3421 | ext4_msg(sb, KERN_ERR, "error updating journal"); |
dab291af | 3422 | jbd2_journal_destroy(journal); |
ac27a0ec DK |
3423 | return err; |
3424 | } | |
3425 | } | |
3426 | ||
617ba13b | 3427 | if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER)) |
dab291af | 3428 | err = jbd2_journal_wipe(journal, !really_read_only); |
1c13d5c0 TT |
3429 | if (!err) { |
3430 | char *save = kmalloc(EXT4_S_ERR_LEN, GFP_KERNEL); | |
3431 | if (save) | |
3432 | memcpy(save, ((char *) es) + | |
3433 | EXT4_S_ERR_START, EXT4_S_ERR_LEN); | |
dab291af | 3434 | err = jbd2_journal_load(journal); |
1c13d5c0 TT |
3435 | if (save) |
3436 | memcpy(((char *) es) + EXT4_S_ERR_START, | |
3437 | save, EXT4_S_ERR_LEN); | |
3438 | kfree(save); | |
3439 | } | |
ac27a0ec DK |
3440 | |
3441 | if (err) { | |
b31e1552 | 3442 | ext4_msg(sb, KERN_ERR, "error loading journal"); |
dab291af | 3443 | jbd2_journal_destroy(journal); |
ac27a0ec DK |
3444 | return err; |
3445 | } | |
3446 | ||
617ba13b MC |
3447 | EXT4_SB(sb)->s_journal = journal; |
3448 | ext4_clear_journal_err(sb, es); | |
ac27a0ec DK |
3449 | |
3450 | if (journal_devnum && | |
3451 | journal_devnum != le32_to_cpu(es->s_journal_dev)) { | |
3452 | es->s_journal_dev = cpu_to_le32(journal_devnum); | |
ac27a0ec DK |
3453 | |
3454 | /* Make sure we flush the recovery flag to disk. */ | |
e2d67052 | 3455 | ext4_commit_super(sb, 1); |
ac27a0ec DK |
3456 | } |
3457 | ||
3458 | return 0; | |
3459 | } | |
3460 | ||
e2d67052 | 3461 | static int ext4_commit_super(struct super_block *sb, int sync) |
ac27a0ec | 3462 | { |
e2d67052 | 3463 | struct ext4_super_block *es = EXT4_SB(sb)->s_es; |
617ba13b | 3464 | struct buffer_head *sbh = EXT4_SB(sb)->s_sbh; |
c4be0c1d | 3465 | int error = 0; |
ac27a0ec DK |
3466 | |
3467 | if (!sbh) | |
c4be0c1d | 3468 | return error; |
914258bf TT |
3469 | if (buffer_write_io_error(sbh)) { |
3470 | /* | |
3471 | * Oh, dear. A previous attempt to write the | |
3472 | * superblock failed. This could happen because the | |
3473 | * USB device was yanked out. Or it could happen to | |
3474 | * be a transient write error and maybe the block will | |
3475 | * be remapped. Nothing we can do but to retry the | |
3476 | * write and hope for the best. | |
3477 | */ | |
b31e1552 ES |
3478 | ext4_msg(sb, KERN_ERR, "previous I/O error to " |
3479 | "superblock detected"); | |
914258bf TT |
3480 | clear_buffer_write_io_error(sbh); |
3481 | set_buffer_uptodate(sbh); | |
3482 | } | |
71290b36 TT |
3483 | /* |
3484 | * If the file system is mounted read-only, don't update the | |
3485 | * superblock write time. This avoids updating the superblock | |
3486 | * write time when we are mounting the root file system | |
3487 | * read/only but we need to replay the journal; at that point, | |
3488 | * for people who are east of GMT and who make their clock | |
3489 | * tick in localtime for Windows bug-for-bug compatibility, | |
3490 | * the clock is set in the future, and this will cause e2fsck | |
3491 | * to complain and force a full file system check. | |
3492 | */ | |
3493 | if (!(sb->s_flags & MS_RDONLY)) | |
3494 | es->s_wtime = cpu_to_le32(get_seconds()); | |
afc32f7e | 3495 | es->s_kbytes_written = |
60e6679e | 3496 | cpu_to_le64(EXT4_SB(sb)->s_kbytes_written + |
afc32f7e TT |
3497 | ((part_stat_read(sb->s_bdev->bd_part, sectors[1]) - |
3498 | EXT4_SB(sb)->s_sectors_written_start) >> 1)); | |
5d1b1b3f AK |
3499 | ext4_free_blocks_count_set(es, percpu_counter_sum_positive( |
3500 | &EXT4_SB(sb)->s_freeblocks_counter)); | |
3501 | es->s_free_inodes_count = cpu_to_le32(percpu_counter_sum_positive( | |
3502 | &EXT4_SB(sb)->s_freeinodes_counter)); | |
7234ab2a | 3503 | sb->s_dirt = 0; |
ac27a0ec DK |
3504 | BUFFER_TRACE(sbh, "marking dirty"); |
3505 | mark_buffer_dirty(sbh); | |
914258bf | 3506 | if (sync) { |
c4be0c1d TS |
3507 | error = sync_dirty_buffer(sbh); |
3508 | if (error) | |
3509 | return error; | |
3510 | ||
3511 | error = buffer_write_io_error(sbh); | |
3512 | if (error) { | |
b31e1552 ES |
3513 | ext4_msg(sb, KERN_ERR, "I/O error while writing " |
3514 | "superblock"); | |
914258bf TT |
3515 | clear_buffer_write_io_error(sbh); |
3516 | set_buffer_uptodate(sbh); | |
3517 | } | |
3518 | } | |
c4be0c1d | 3519 | return error; |
ac27a0ec DK |
3520 | } |
3521 | ||
ac27a0ec DK |
3522 | /* |
3523 | * Have we just finished recovery? If so, and if we are mounting (or | |
3524 | * remounting) the filesystem readonly, then we will end up with a | |
3525 | * consistent fs on disk. Record that fact. | |
3526 | */ | |
2b2d6d01 TT |
3527 | static void ext4_mark_recovery_complete(struct super_block *sb, |
3528 | struct ext4_super_block *es) | |
ac27a0ec | 3529 | { |
617ba13b | 3530 | journal_t *journal = EXT4_SB(sb)->s_journal; |
ac27a0ec | 3531 | |
0390131b FM |
3532 | if (!EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL)) { |
3533 | BUG_ON(journal != NULL); | |
3534 | return; | |
3535 | } | |
dab291af | 3536 | jbd2_journal_lock_updates(journal); |
7ffe1ea8 HK |
3537 | if (jbd2_journal_flush(journal) < 0) |
3538 | goto out; | |
3539 | ||
617ba13b | 3540 | if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER) && |
ac27a0ec | 3541 | sb->s_flags & MS_RDONLY) { |
617ba13b | 3542 | EXT4_CLEAR_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER); |
e2d67052 | 3543 | ext4_commit_super(sb, 1); |
ac27a0ec | 3544 | } |
7ffe1ea8 HK |
3545 | |
3546 | out: | |
dab291af | 3547 | jbd2_journal_unlock_updates(journal); |
ac27a0ec DK |
3548 | } |
3549 | ||
3550 | /* | |
3551 | * If we are mounting (or read-write remounting) a filesystem whose journal | |
3552 | * has recorded an error from a previous lifetime, move that error to the | |
3553 | * main filesystem now. | |
3554 | */ | |
2b2d6d01 TT |
3555 | static void ext4_clear_journal_err(struct super_block *sb, |
3556 | struct ext4_super_block *es) | |
ac27a0ec DK |
3557 | { |
3558 | journal_t *journal; | |
3559 | int j_errno; | |
3560 | const char *errstr; | |
3561 | ||
0390131b FM |
3562 | BUG_ON(!EXT4_HAS_COMPAT_FEATURE(sb, EXT4_FEATURE_COMPAT_HAS_JOURNAL)); |
3563 | ||
617ba13b | 3564 | journal = EXT4_SB(sb)->s_journal; |
ac27a0ec DK |
3565 | |
3566 | /* | |
3567 | * Now check for any error status which may have been recorded in the | |
617ba13b | 3568 | * journal by a prior ext4_error() or ext4_abort() |
ac27a0ec DK |
3569 | */ |
3570 | ||
dab291af | 3571 | j_errno = jbd2_journal_errno(journal); |
ac27a0ec DK |
3572 | if (j_errno) { |
3573 | char nbuf[16]; | |
3574 | ||
617ba13b | 3575 | errstr = ext4_decode_error(sb, j_errno, nbuf); |
12062ddd | 3576 | ext4_warning(sb, "Filesystem error recorded " |
ac27a0ec | 3577 | "from previous mount: %s", errstr); |
12062ddd | 3578 | ext4_warning(sb, "Marking fs in need of filesystem check."); |
ac27a0ec | 3579 | |
617ba13b MC |
3580 | EXT4_SB(sb)->s_mount_state |= EXT4_ERROR_FS; |
3581 | es->s_state |= cpu_to_le16(EXT4_ERROR_FS); | |
e2d67052 | 3582 | ext4_commit_super(sb, 1); |
ac27a0ec | 3583 | |
dab291af | 3584 | jbd2_journal_clear_err(journal); |
ac27a0ec DK |
3585 | } |
3586 | } | |
3587 | ||
3588 | /* | |
3589 | * Force the running and committing transactions to commit, | |
3590 | * and wait on the commit. | |
3591 | */ | |
617ba13b | 3592 | int ext4_force_commit(struct super_block *sb) |
ac27a0ec DK |
3593 | { |
3594 | journal_t *journal; | |
0390131b | 3595 | int ret = 0; |
ac27a0ec DK |
3596 | |
3597 | if (sb->s_flags & MS_RDONLY) | |
3598 | return 0; | |
3599 | ||
617ba13b | 3600 | journal = EXT4_SB(sb)->s_journal; |
6b0310fb ES |
3601 | if (journal) { |
3602 | vfs_check_frozen(sb, SB_FREEZE_WRITE); | |
0390131b | 3603 | ret = ext4_journal_force_commit(journal); |
6b0310fb | 3604 | } |
0390131b | 3605 | |
ac27a0ec DK |
3606 | return ret; |
3607 | } | |
3608 | ||
2b2d6d01 | 3609 | static void ext4_write_super(struct super_block *sb) |
ac27a0ec | 3610 | { |
ebc1ac16 | 3611 | lock_super(sb); |
9ca92389 | 3612 | ext4_commit_super(sb, 1); |
ebc1ac16 | 3613 | unlock_super(sb); |
ac27a0ec DK |
3614 | } |
3615 | ||
617ba13b | 3616 | static int ext4_sync_fs(struct super_block *sb, int wait) |
ac27a0ec | 3617 | { |
14ce0cb4 | 3618 | int ret = 0; |
9eddacf9 | 3619 | tid_t target; |
8d5d02e6 | 3620 | struct ext4_sb_info *sbi = EXT4_SB(sb); |
ac27a0ec | 3621 | |
9bffad1e | 3622 | trace_ext4_sync_fs(sb, wait); |
8d5d02e6 MC |
3623 | flush_workqueue(sbi->dio_unwritten_wq); |
3624 | if (jbd2_journal_start_commit(sbi->s_journal, &target)) { | |
9ca92389 | 3625 | if (wait) |
8d5d02e6 | 3626 | jbd2_log_wait_commit(sbi->s_journal, target); |
0390131b | 3627 | } |
14ce0cb4 | 3628 | return ret; |
ac27a0ec DK |
3629 | } |
3630 | ||
3631 | /* | |
3632 | * LVM calls this function before a (read-only) snapshot is created. This | |
3633 | * gives us a chance to flush the journal completely and mark the fs clean. | |
3634 | */ | |
c4be0c1d | 3635 | static int ext4_freeze(struct super_block *sb) |
ac27a0ec | 3636 | { |
c4be0c1d TS |
3637 | int error = 0; |
3638 | journal_t *journal; | |
ac27a0ec | 3639 | |
9ca92389 TT |
3640 | if (sb->s_flags & MS_RDONLY) |
3641 | return 0; | |
ac27a0ec | 3642 | |
9ca92389 | 3643 | journal = EXT4_SB(sb)->s_journal; |
7ffe1ea8 | 3644 | |
9ca92389 TT |
3645 | /* Now we set up the journal barrier. */ |
3646 | jbd2_journal_lock_updates(journal); | |
ac27a0ec | 3647 | |
9ca92389 TT |
3648 | /* |
3649 | * Don't clear the needs_recovery flag if we failed to flush | |
3650 | * the journal. | |
3651 | */ | |
3652 | error = jbd2_journal_flush(journal); | |
6b0310fb ES |
3653 | if (error < 0) |
3654 | goto out; | |
9ca92389 TT |
3655 | |
3656 | /* Journal blocked and flushed, clear needs_recovery flag. */ | |
3657 | EXT4_CLEAR_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER); | |
3658 | error = ext4_commit_super(sb, 1); | |
6b0310fb ES |
3659 | out: |
3660 | /* we rely on s_frozen to stop further updates */ | |
3661 | jbd2_journal_unlock_updates(EXT4_SB(sb)->s_journal); | |
3662 | return error; | |
ac27a0ec DK |
3663 | } |
3664 | ||
3665 | /* | |
3666 | * Called by LVM after the snapshot is done. We need to reset the RECOVER | |
3667 | * flag here, even though the filesystem is not technically dirty yet. | |
3668 | */ | |
c4be0c1d | 3669 | static int ext4_unfreeze(struct super_block *sb) |
ac27a0ec | 3670 | { |
9ca92389 TT |
3671 | if (sb->s_flags & MS_RDONLY) |
3672 | return 0; | |
3673 | ||
3674 | lock_super(sb); | |
3675 | /* Reset the needs_recovery flag before the fs is unlocked. */ | |
3676 | EXT4_SET_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_RECOVER); | |
3677 | ext4_commit_super(sb, 1); | |
3678 | unlock_super(sb); | |
c4be0c1d | 3679 | return 0; |
ac27a0ec DK |
3680 | } |
3681 | ||
2b2d6d01 | 3682 | static int ext4_remount(struct super_block *sb, int *flags, char *data) |
ac27a0ec | 3683 | { |
2b2d6d01 | 3684 | struct ext4_super_block *es; |
617ba13b MC |
3685 | struct ext4_sb_info *sbi = EXT4_SB(sb); |
3686 | ext4_fsblk_t n_blocks_count = 0; | |
ac27a0ec | 3687 | unsigned long old_sb_flags; |
617ba13b | 3688 | struct ext4_mount_options old_opts; |
c79d967d | 3689 | int enable_quota = 0; |
8a266467 | 3690 | ext4_group_t g; |
b3881f74 | 3691 | unsigned int journal_ioprio = DEFAULT_JOURNAL_IOPRIO; |
ac27a0ec DK |
3692 | int err; |
3693 | #ifdef CONFIG_QUOTA | |
3694 | int i; | |
3695 | #endif | |
d4c402d9 | 3696 | char *orig_data = kstrdup(data, GFP_KERNEL); |
ac27a0ec | 3697 | |
337eb00a AIB |
3698 | lock_kernel(); |
3699 | ||
ac27a0ec | 3700 | /* Store the original options */ |
bbd6851a | 3701 | lock_super(sb); |
ac27a0ec DK |
3702 | old_sb_flags = sb->s_flags; |
3703 | old_opts.s_mount_opt = sbi->s_mount_opt; | |
3704 | old_opts.s_resuid = sbi->s_resuid; | |
3705 | old_opts.s_resgid = sbi->s_resgid; | |
3706 | old_opts.s_commit_interval = sbi->s_commit_interval; | |
30773840 TT |
3707 | old_opts.s_min_batch_time = sbi->s_min_batch_time; |
3708 | old_opts.s_max_batch_time = sbi->s_max_batch_time; | |
ac27a0ec DK |
3709 | #ifdef CONFIG_QUOTA |
3710 | old_opts.s_jquota_fmt = sbi->s_jquota_fmt; | |
3711 | for (i = 0; i < MAXQUOTAS; i++) | |
3712 | old_opts.s_qf_names[i] = sbi->s_qf_names[i]; | |
3713 | #endif | |
b3881f74 TT |
3714 | if (sbi->s_journal && sbi->s_journal->j_task->io_context) |
3715 | journal_ioprio = sbi->s_journal->j_task->io_context->ioprio; | |
ac27a0ec DK |
3716 | |
3717 | /* | |
3718 | * Allow the "check" option to be passed as a remount option. | |
3719 | */ | |
b3881f74 TT |
3720 | if (!parse_options(data, sb, NULL, &journal_ioprio, |
3721 | &n_blocks_count, 1)) { | |
ac27a0ec DK |
3722 | err = -EINVAL; |
3723 | goto restore_opts; | |
3724 | } | |
3725 | ||
4ab2f15b | 3726 | if (sbi->s_mount_flags & EXT4_MF_FS_ABORTED) |
c67d859e | 3727 | ext4_abort(sb, "Abort forced by user"); |
ac27a0ec DK |
3728 | |
3729 | sb->s_flags = (sb->s_flags & ~MS_POSIXACL) | | |
482a7425 | 3730 | (test_opt(sb, POSIX_ACL) ? MS_POSIXACL : 0); |
ac27a0ec DK |
3731 | |
3732 | es = sbi->s_es; | |
3733 | ||
b3881f74 | 3734 | if (sbi->s_journal) { |
0390131b | 3735 | ext4_init_journal_params(sb, sbi->s_journal); |
b3881f74 TT |
3736 | set_task_ioprio(sbi->s_journal->j_task, journal_ioprio); |
3737 | } | |
ac27a0ec DK |
3738 | |
3739 | if ((*flags & MS_RDONLY) != (sb->s_flags & MS_RDONLY) || | |
bd81d8ee | 3740 | n_blocks_count > ext4_blocks_count(es)) { |
4ab2f15b | 3741 | if (sbi->s_mount_flags & EXT4_MF_FS_ABORTED) { |
ac27a0ec DK |
3742 | err = -EROFS; |
3743 | goto restore_opts; | |
3744 | } | |
3745 | ||
3746 | if (*flags & MS_RDONLY) { | |
0f0dd62f CH |
3747 | err = dquot_suspend(sb, -1); |
3748 | if (err < 0) | |
c79d967d | 3749 | goto restore_opts; |
c79d967d | 3750 | |
ac27a0ec DK |
3751 | /* |
3752 | * First of all, the unconditional stuff we have to do | |
3753 | * to disable replay of the journal when we next remount | |
3754 | */ | |
3755 | sb->s_flags |= MS_RDONLY; | |
3756 | ||
3757 | /* | |
3758 | * OK, test if we are remounting a valid rw partition | |
3759 | * readonly, and if so set the rdonly flag and then | |
3760 | * mark the partition as valid again. | |
3761 | */ | |
617ba13b MC |
3762 | if (!(es->s_state & cpu_to_le16(EXT4_VALID_FS)) && |
3763 | (sbi->s_mount_state & EXT4_VALID_FS)) | |
ac27a0ec DK |
3764 | es->s_state = cpu_to_le16(sbi->s_mount_state); |
3765 | ||
a63c9eb2 | 3766 | if (sbi->s_journal) |
0390131b | 3767 | ext4_mark_recovery_complete(sb, es); |
ac27a0ec | 3768 | } else { |
a13fb1a4 ES |
3769 | /* Make sure we can mount this feature set readwrite */ |
3770 | if (!ext4_feature_set_ok(sb, 0)) { | |
ac27a0ec DK |
3771 | err = -EROFS; |
3772 | goto restore_opts; | |
3773 | } | |
8a266467 TT |
3774 | /* |
3775 | * Make sure the group descriptor checksums | |
0b8e58a1 | 3776 | * are sane. If they aren't, refuse to remount r/w. |
8a266467 TT |
3777 | */ |
3778 | for (g = 0; g < sbi->s_groups_count; g++) { | |
3779 | struct ext4_group_desc *gdp = | |
3780 | ext4_get_group_desc(sb, g, NULL); | |
3781 | ||
3782 | if (!ext4_group_desc_csum_verify(sbi, g, gdp)) { | |
b31e1552 ES |
3783 | ext4_msg(sb, KERN_ERR, |
3784 | "ext4_remount: Checksum for group %u failed (%u!=%u)", | |
8a266467 TT |
3785 | g, le16_to_cpu(ext4_group_desc_csum(sbi, g, gdp)), |
3786 | le16_to_cpu(gdp->bg_checksum)); | |
3787 | err = -EINVAL; | |
3788 | goto restore_opts; | |
3789 | } | |
3790 | } | |
3791 | ||
ead6596b ES |
3792 | /* |
3793 | * If we have an unprocessed orphan list hanging | |
3794 | * around from a previously readonly bdev mount, | |
3795 | * require a full umount/remount for now. | |
3796 | */ | |
3797 | if (es->s_last_orphan) { | |
b31e1552 | 3798 | ext4_msg(sb, KERN_WARNING, "Couldn't " |
ead6596b ES |
3799 | "remount RDWR because of unprocessed " |
3800 | "orphan inode list. Please " | |
b31e1552 | 3801 | "umount/remount instead"); |
ead6596b ES |
3802 | err = -EINVAL; |
3803 | goto restore_opts; | |
3804 | } | |
3805 | ||
ac27a0ec DK |
3806 | /* |
3807 | * Mounting a RDONLY partition read-write, so reread | |
3808 | * and store the current valid flag. (It may have | |
3809 | * been changed by e2fsck since we originally mounted | |
3810 | * the partition.) | |
3811 | */ | |
0390131b FM |
3812 | if (sbi->s_journal) |
3813 | ext4_clear_journal_err(sb, es); | |
ac27a0ec | 3814 | sbi->s_mount_state = le16_to_cpu(es->s_state); |
617ba13b | 3815 | if ((err = ext4_group_extend(sb, es, n_blocks_count))) |
ac27a0ec | 3816 | goto restore_opts; |
2b2d6d01 | 3817 | if (!ext4_setup_super(sb, es, 0)) |
ac27a0ec | 3818 | sb->s_flags &= ~MS_RDONLY; |
c79d967d | 3819 | enable_quota = 1; |
ac27a0ec DK |
3820 | } |
3821 | } | |
6fd058f7 | 3822 | ext4_setup_system_zone(sb); |
0390131b | 3823 | if (sbi->s_journal == NULL) |
e2d67052 | 3824 | ext4_commit_super(sb, 1); |
0390131b | 3825 | |
ac27a0ec DK |
3826 | #ifdef CONFIG_QUOTA |
3827 | /* Release old quota file names */ | |
3828 | for (i = 0; i < MAXQUOTAS; i++) | |
3829 | if (old_opts.s_qf_names[i] && | |
3830 | old_opts.s_qf_names[i] != sbi->s_qf_names[i]) | |
3831 | kfree(old_opts.s_qf_names[i]); | |
3832 | #endif | |
bbd6851a | 3833 | unlock_super(sb); |
337eb00a | 3834 | unlock_kernel(); |
c79d967d | 3835 | if (enable_quota) |
0f0dd62f | 3836 | dquot_resume(sb, -1); |
d4c402d9 CW |
3837 | |
3838 | ext4_msg(sb, KERN_INFO, "re-mounted. Opts: %s", orig_data); | |
3839 | kfree(orig_data); | |
ac27a0ec | 3840 | return 0; |
0b8e58a1 | 3841 | |
ac27a0ec DK |
3842 | restore_opts: |
3843 | sb->s_flags = old_sb_flags; | |
3844 | sbi->s_mount_opt = old_opts.s_mount_opt; | |
3845 | sbi->s_resuid = old_opts.s_resuid; | |
3846 | sbi->s_resgid = old_opts.s_resgid; | |
3847 | sbi->s_commit_interval = old_opts.s_commit_interval; | |
30773840 TT |
3848 | sbi->s_min_batch_time = old_opts.s_min_batch_time; |
3849 | sbi->s_max_batch_time = old_opts.s_max_batch_time; | |
ac27a0ec DK |
3850 | #ifdef CONFIG_QUOTA |
3851 | sbi->s_jquota_fmt = old_opts.s_jquota_fmt; | |
3852 | for (i = 0; i < MAXQUOTAS; i++) { | |
3853 | if (sbi->s_qf_names[i] && | |
3854 | old_opts.s_qf_names[i] != sbi->s_qf_names[i]) | |
3855 | kfree(sbi->s_qf_names[i]); | |
3856 | sbi->s_qf_names[i] = old_opts.s_qf_names[i]; | |
3857 | } | |
3858 | #endif | |
bbd6851a | 3859 | unlock_super(sb); |
337eb00a | 3860 | unlock_kernel(); |
d4c402d9 | 3861 | kfree(orig_data); |
ac27a0ec DK |
3862 | return err; |
3863 | } | |
3864 | ||
2b2d6d01 | 3865 | static int ext4_statfs(struct dentry *dentry, struct kstatfs *buf) |
ac27a0ec DK |
3866 | { |
3867 | struct super_block *sb = dentry->d_sb; | |
617ba13b MC |
3868 | struct ext4_sb_info *sbi = EXT4_SB(sb); |
3869 | struct ext4_super_block *es = sbi->s_es; | |
960cc398 | 3870 | u64 fsid; |
ac27a0ec | 3871 | |
5e70030d BP |
3872 | if (test_opt(sb, MINIX_DF)) { |
3873 | sbi->s_overhead_last = 0; | |
6bc9feff | 3874 | } else if (sbi->s_blocks_last != ext4_blocks_count(es)) { |
8df9675f | 3875 | ext4_group_t i, ngroups = ext4_get_groups_count(sb); |
5e70030d | 3876 | ext4_fsblk_t overhead = 0; |
ac27a0ec DK |
3877 | |
3878 | /* | |
5e70030d BP |
3879 | * Compute the overhead (FS structures). This is constant |
3880 | * for a given filesystem unless the number of block groups | |
3881 | * changes so we cache the previous value until it does. | |
ac27a0ec DK |
3882 | */ |
3883 | ||
3884 | /* | |
3885 | * All of the blocks before first_data_block are | |
3886 | * overhead | |
3887 | */ | |
3888 | overhead = le32_to_cpu(es->s_first_data_block); | |
3889 | ||
3890 | /* | |
3891 | * Add the overhead attributed to the superblock and | |
3892 | * block group descriptors. If the sparse superblocks | |
3893 | * feature is turned on, then not all groups have this. | |
3894 | */ | |
3895 | for (i = 0; i < ngroups; i++) { | |
617ba13b MC |
3896 | overhead += ext4_bg_has_super(sb, i) + |
3897 | ext4_bg_num_gdb(sb, i); | |
ac27a0ec DK |
3898 | cond_resched(); |
3899 | } | |
3900 | ||
3901 | /* | |
3902 | * Every block group has an inode bitmap, a block | |
3903 | * bitmap, and an inode table. | |
3904 | */ | |
5e70030d BP |
3905 | overhead += ngroups * (2 + sbi->s_itb_per_group); |
3906 | sbi->s_overhead_last = overhead; | |
3907 | smp_wmb(); | |
6bc9feff | 3908 | sbi->s_blocks_last = ext4_blocks_count(es); |
ac27a0ec DK |
3909 | } |
3910 | ||
617ba13b | 3911 | buf->f_type = EXT4_SUPER_MAGIC; |
ac27a0ec | 3912 | buf->f_bsize = sb->s_blocksize; |
5e70030d | 3913 | buf->f_blocks = ext4_blocks_count(es) - sbi->s_overhead_last; |
6bc6e63f AK |
3914 | buf->f_bfree = percpu_counter_sum_positive(&sbi->s_freeblocks_counter) - |
3915 | percpu_counter_sum_positive(&sbi->s_dirtyblocks_counter); | |
bd81d8ee LV |
3916 | buf->f_bavail = buf->f_bfree - ext4_r_blocks_count(es); |
3917 | if (buf->f_bfree < ext4_r_blocks_count(es)) | |
ac27a0ec DK |
3918 | buf->f_bavail = 0; |
3919 | buf->f_files = le32_to_cpu(es->s_inodes_count); | |
52d9f3b4 | 3920 | buf->f_ffree = percpu_counter_sum_positive(&sbi->s_freeinodes_counter); |
617ba13b | 3921 | buf->f_namelen = EXT4_NAME_LEN; |
960cc398 PE |
3922 | fsid = le64_to_cpup((void *)es->s_uuid) ^ |
3923 | le64_to_cpup((void *)es->s_uuid + sizeof(u64)); | |
3924 | buf->f_fsid.val[0] = fsid & 0xFFFFFFFFUL; | |
3925 | buf->f_fsid.val[1] = (fsid >> 32) & 0xFFFFFFFFUL; | |
0b8e58a1 | 3926 | |
ac27a0ec DK |
3927 | return 0; |
3928 | } | |
3929 | ||
0b8e58a1 AD |
3930 | /* Helper function for writing quotas on sync - we need to start transaction |
3931 | * before quota file is locked for write. Otherwise the are possible deadlocks: | |
ac27a0ec | 3932 | * Process 1 Process 2 |
617ba13b | 3933 | * ext4_create() quota_sync() |
a269eb18 | 3934 | * jbd2_journal_start() write_dquot() |
871a2931 | 3935 | * dquot_initialize() down(dqio_mutex) |
dab291af | 3936 | * down(dqio_mutex) jbd2_journal_start() |
ac27a0ec DK |
3937 | * |
3938 | */ | |
3939 | ||
3940 | #ifdef CONFIG_QUOTA | |
3941 | ||
3942 | static inline struct inode *dquot_to_inode(struct dquot *dquot) | |
3943 | { | |
3944 | return sb_dqopt(dquot->dq_sb)->files[dquot->dq_type]; | |
3945 | } | |
3946 | ||
617ba13b | 3947 | static int ext4_write_dquot(struct dquot *dquot) |
ac27a0ec DK |
3948 | { |
3949 | int ret, err; | |
3950 | handle_t *handle; | |
3951 | struct inode *inode; | |
3952 | ||
3953 | inode = dquot_to_inode(dquot); | |
617ba13b | 3954 | handle = ext4_journal_start(inode, |
0b8e58a1 | 3955 | EXT4_QUOTA_TRANS_BLOCKS(dquot->dq_sb)); |
ac27a0ec DK |
3956 | if (IS_ERR(handle)) |
3957 | return PTR_ERR(handle); | |
3958 | ret = dquot_commit(dquot); | |
617ba13b | 3959 | err = ext4_journal_stop(handle); |
ac27a0ec DK |
3960 | if (!ret) |
3961 | ret = err; | |
3962 | return ret; | |
3963 | } | |
3964 | ||
617ba13b | 3965 | static int ext4_acquire_dquot(struct dquot *dquot) |
ac27a0ec DK |
3966 | { |
3967 | int ret, err; | |
3968 | handle_t *handle; | |
3969 | ||
617ba13b | 3970 | handle = ext4_journal_start(dquot_to_inode(dquot), |
0b8e58a1 | 3971 | EXT4_QUOTA_INIT_BLOCKS(dquot->dq_sb)); |
ac27a0ec DK |
3972 | if (IS_ERR(handle)) |
3973 | return PTR_ERR(handle); | |
3974 | ret = dquot_acquire(dquot); | |
617ba13b | 3975 | err = ext4_journal_stop(handle); |
ac27a0ec DK |
3976 | if (!ret) |
3977 | ret = err; | |
3978 | return ret; | |
3979 | } | |
3980 | ||
617ba13b | 3981 | static int ext4_release_dquot(struct dquot *dquot) |
ac27a0ec DK |
3982 | { |
3983 | int ret, err; | |
3984 | handle_t *handle; | |
3985 | ||
617ba13b | 3986 | handle = ext4_journal_start(dquot_to_inode(dquot), |
0b8e58a1 | 3987 | EXT4_QUOTA_DEL_BLOCKS(dquot->dq_sb)); |
9c3013e9 JK |
3988 | if (IS_ERR(handle)) { |
3989 | /* Release dquot anyway to avoid endless cycle in dqput() */ | |
3990 | dquot_release(dquot); | |
ac27a0ec | 3991 | return PTR_ERR(handle); |
9c3013e9 | 3992 | } |
ac27a0ec | 3993 | ret = dquot_release(dquot); |
617ba13b | 3994 | err = ext4_journal_stop(handle); |
ac27a0ec DK |
3995 | if (!ret) |
3996 | ret = err; | |
3997 | return ret; | |
3998 | } | |
3999 | ||
617ba13b | 4000 | static int ext4_mark_dquot_dirty(struct dquot *dquot) |
ac27a0ec | 4001 | { |
2c8be6b2 | 4002 | /* Are we journaling quotas? */ |
617ba13b MC |
4003 | if (EXT4_SB(dquot->dq_sb)->s_qf_names[USRQUOTA] || |
4004 | EXT4_SB(dquot->dq_sb)->s_qf_names[GRPQUOTA]) { | |
ac27a0ec | 4005 | dquot_mark_dquot_dirty(dquot); |
617ba13b | 4006 | return ext4_write_dquot(dquot); |
ac27a0ec DK |
4007 | } else { |
4008 | return dquot_mark_dquot_dirty(dquot); | |
4009 | } | |
4010 | } | |
4011 | ||
617ba13b | 4012 | static int ext4_write_info(struct super_block *sb, int type) |
ac27a0ec DK |
4013 | { |
4014 | int ret, err; | |
4015 | handle_t *handle; | |
4016 | ||
4017 | /* Data block + inode block */ | |
617ba13b | 4018 | handle = ext4_journal_start(sb->s_root->d_inode, 2); |
ac27a0ec DK |
4019 | if (IS_ERR(handle)) |
4020 | return PTR_ERR(handle); | |
4021 | ret = dquot_commit_info(sb, type); | |
617ba13b | 4022 | err = ext4_journal_stop(handle); |
ac27a0ec DK |
4023 | if (!ret) |
4024 | ret = err; | |
4025 | return ret; | |
4026 | } | |
4027 | ||
4028 | /* | |
4029 | * Turn on quotas during mount time - we need to find | |
4030 | * the quota file and such... | |
4031 | */ | |
617ba13b | 4032 | static int ext4_quota_on_mount(struct super_block *sb, int type) |
ac27a0ec | 4033 | { |
287a8095 CH |
4034 | return dquot_quota_on_mount(sb, EXT4_SB(sb)->s_qf_names[type], |
4035 | EXT4_SB(sb)->s_jquota_fmt, type); | |
ac27a0ec DK |
4036 | } |
4037 | ||
4038 | /* | |
4039 | * Standard function to be called on quota_on | |
4040 | */ | |
617ba13b | 4041 | static int ext4_quota_on(struct super_block *sb, int type, int format_id, |
307ae18a | 4042 | char *name) |
ac27a0ec DK |
4043 | { |
4044 | int err; | |
8264613d | 4045 | struct path path; |
ac27a0ec DK |
4046 | |
4047 | if (!test_opt(sb, QUOTA)) | |
4048 | return -EINVAL; | |
0623543b | 4049 | |
8264613d | 4050 | err = kern_path(name, LOOKUP_FOLLOW, &path); |
ac27a0ec DK |
4051 | if (err) |
4052 | return err; | |
0623543b | 4053 | |
ac27a0ec | 4054 | /* Quotafile not on the same filesystem? */ |
8264613d AV |
4055 | if (path.mnt->mnt_sb != sb) { |
4056 | path_put(&path); | |
ac27a0ec DK |
4057 | return -EXDEV; |
4058 | } | |
0623543b JK |
4059 | /* Journaling quota? */ |
4060 | if (EXT4_SB(sb)->s_qf_names[type]) { | |
2b2d6d01 | 4061 | /* Quotafile not in fs root? */ |
8264613d | 4062 | if (path.dentry->d_parent != sb->s_root) |
b31e1552 ES |
4063 | ext4_msg(sb, KERN_WARNING, |
4064 | "Quota file not on filesystem root. " | |
4065 | "Journaled quota will not work"); | |
2b2d6d01 | 4066 | } |
0623543b JK |
4067 | |
4068 | /* | |
4069 | * When we journal data on quota file, we have to flush journal to see | |
4070 | * all updates to the file when we bypass pagecache... | |
4071 | */ | |
0390131b FM |
4072 | if (EXT4_SB(sb)->s_journal && |
4073 | ext4_should_journal_data(path.dentry->d_inode)) { | |
0623543b JK |
4074 | /* |
4075 | * We don't need to lock updates but journal_flush() could | |
4076 | * otherwise be livelocked... | |
4077 | */ | |
4078 | jbd2_journal_lock_updates(EXT4_SB(sb)->s_journal); | |
7ffe1ea8 | 4079 | err = jbd2_journal_flush(EXT4_SB(sb)->s_journal); |
0623543b | 4080 | jbd2_journal_unlock_updates(EXT4_SB(sb)->s_journal); |
7ffe1ea8 | 4081 | if (err) { |
8264613d | 4082 | path_put(&path); |
7ffe1ea8 HK |
4083 | return err; |
4084 | } | |
0623543b JK |
4085 | } |
4086 | ||
287a8095 | 4087 | err = dquot_quota_on_path(sb, type, format_id, &path); |
8264613d | 4088 | path_put(&path); |
77e69dac | 4089 | return err; |
ac27a0ec DK |
4090 | } |
4091 | ||
4092 | /* Read data from quotafile - avoid pagecache and such because we cannot afford | |
4093 | * acquiring the locks... As quota files are never truncated and quota code | |
4094 | * itself serializes the operations (and noone else should touch the files) | |
4095 | * we don't have to be afraid of races */ | |
617ba13b | 4096 | static ssize_t ext4_quota_read(struct super_block *sb, int type, char *data, |
ac27a0ec DK |
4097 | size_t len, loff_t off) |
4098 | { | |
4099 | struct inode *inode = sb_dqopt(sb)->files[type]; | |
725d26d3 | 4100 | ext4_lblk_t blk = off >> EXT4_BLOCK_SIZE_BITS(sb); |
ac27a0ec DK |
4101 | int err = 0; |
4102 | int offset = off & (sb->s_blocksize - 1); | |
4103 | int tocopy; | |
4104 | size_t toread; | |
4105 | struct buffer_head *bh; | |
4106 | loff_t i_size = i_size_read(inode); | |
4107 | ||
4108 | if (off > i_size) | |
4109 | return 0; | |
4110 | if (off+len > i_size) | |
4111 | len = i_size-off; | |
4112 | toread = len; | |
4113 | while (toread > 0) { | |
4114 | tocopy = sb->s_blocksize - offset < toread ? | |
4115 | sb->s_blocksize - offset : toread; | |
617ba13b | 4116 | bh = ext4_bread(NULL, inode, blk, 0, &err); |
ac27a0ec DK |
4117 | if (err) |
4118 | return err; | |
4119 | if (!bh) /* A hole? */ | |
4120 | memset(data, 0, tocopy); | |
4121 | else | |
4122 | memcpy(data, bh->b_data+offset, tocopy); | |
4123 | brelse(bh); | |
4124 | offset = 0; | |
4125 | toread -= tocopy; | |
4126 | data += tocopy; | |
4127 | blk++; | |
4128 | } | |
4129 | return len; | |
4130 | } | |
4131 | ||
4132 | /* Write to quotafile (we know the transaction is already started and has | |
4133 | * enough credits) */ | |
617ba13b | 4134 | static ssize_t ext4_quota_write(struct super_block *sb, int type, |
ac27a0ec DK |
4135 | const char *data, size_t len, loff_t off) |
4136 | { | |
4137 | struct inode *inode = sb_dqopt(sb)->files[type]; | |
725d26d3 | 4138 | ext4_lblk_t blk = off >> EXT4_BLOCK_SIZE_BITS(sb); |
ac27a0ec DK |
4139 | int err = 0; |
4140 | int offset = off & (sb->s_blocksize - 1); | |
ac27a0ec DK |
4141 | struct buffer_head *bh; |
4142 | handle_t *handle = journal_current_handle(); | |
4143 | ||
0390131b | 4144 | if (EXT4_SB(sb)->s_journal && !handle) { |
b31e1552 ES |
4145 | ext4_msg(sb, KERN_WARNING, "Quota write (off=%llu, len=%llu)" |
4146 | " cancelled because transaction is not started", | |
9c3013e9 JK |
4147 | (unsigned long long)off, (unsigned long long)len); |
4148 | return -EIO; | |
4149 | } | |
67eeb568 DM |
4150 | /* |
4151 | * Since we account only one data block in transaction credits, | |
4152 | * then it is impossible to cross a block boundary. | |
4153 | */ | |
4154 | if (sb->s_blocksize - offset < len) { | |
4155 | ext4_msg(sb, KERN_WARNING, "Quota write (off=%llu, len=%llu)" | |
4156 | " cancelled because not block aligned", | |
4157 | (unsigned long long)off, (unsigned long long)len); | |
4158 | return -EIO; | |
4159 | } | |
4160 | ||
ac27a0ec | 4161 | mutex_lock_nested(&inode->i_mutex, I_MUTEX_QUOTA); |
67eeb568 DM |
4162 | bh = ext4_bread(handle, inode, blk, 1, &err); |
4163 | if (!bh) | |
4164 | goto out; | |
62d2b5f2 JK |
4165 | err = ext4_journal_get_write_access(handle, bh); |
4166 | if (err) { | |
4167 | brelse(bh); | |
4168 | goto out; | |
ac27a0ec | 4169 | } |
67eeb568 DM |
4170 | lock_buffer(bh); |
4171 | memcpy(bh->b_data+offset, data, len); | |
4172 | flush_dcache_page(bh->b_page); | |
4173 | unlock_buffer(bh); | |
62d2b5f2 | 4174 | err = ext4_handle_dirty_metadata(handle, NULL, bh); |
67eeb568 | 4175 | brelse(bh); |
ac27a0ec | 4176 | out: |
67eeb568 | 4177 | if (err) { |
4d04e4fb | 4178 | mutex_unlock(&inode->i_mutex); |
ac27a0ec | 4179 | return err; |
4d04e4fb | 4180 | } |
67eeb568 DM |
4181 | if (inode->i_size < off + len) { |
4182 | i_size_write(inode, off + len); | |
617ba13b | 4183 | EXT4_I(inode)->i_disksize = inode->i_size; |
ac27a0ec | 4184 | } |
ac27a0ec | 4185 | inode->i_mtime = inode->i_ctime = CURRENT_TIME; |
617ba13b | 4186 | ext4_mark_inode_dirty(handle, inode); |
ac27a0ec | 4187 | mutex_unlock(&inode->i_mutex); |
67eeb568 | 4188 | return len; |
ac27a0ec DK |
4189 | } |
4190 | ||
4191 | #endif | |
4192 | ||
0b8e58a1 AD |
4193 | static int ext4_get_sb(struct file_system_type *fs_type, int flags, |
4194 | const char *dev_name, void *data, struct vfsmount *mnt) | |
ac27a0ec | 4195 | { |
0b8e58a1 | 4196 | return get_sb_bdev(fs_type, flags, dev_name, data, ext4_fill_super,mnt); |
ac27a0ec DK |
4197 | } |
4198 | ||
37f328eb | 4199 | #if !defined(CONFIG_EXT2_FS) && !defined(CONFIG_EXT2_FS_MODULE) && defined(CONFIG_EXT4_USE_FOR_EXT23) |
24b58424 TT |
4200 | static struct file_system_type ext2_fs_type = { |
4201 | .owner = THIS_MODULE, | |
4202 | .name = "ext2", | |
4203 | .get_sb = ext4_get_sb, | |
4204 | .kill_sb = kill_block_super, | |
4205 | .fs_flags = FS_REQUIRES_DEV, | |
4206 | }; | |
4207 | ||
4208 | static inline void register_as_ext2(void) | |
4209 | { | |
4210 | int err = register_filesystem(&ext2_fs_type); | |
4211 | if (err) | |
4212 | printk(KERN_WARNING | |
4213 | "EXT4-fs: Unable to register as ext2 (%d)\n", err); | |
4214 | } | |
4215 | ||
4216 | static inline void unregister_as_ext2(void) | |
4217 | { | |
4218 | unregister_filesystem(&ext2_fs_type); | |
4219 | } | |
51b7e3c9 | 4220 | MODULE_ALIAS("ext2"); |
24b58424 TT |
4221 | #else |
4222 | static inline void register_as_ext2(void) { } | |
4223 | static inline void unregister_as_ext2(void) { } | |
4224 | #endif | |
4225 | ||
37f328eb | 4226 | #if !defined(CONFIG_EXT3_FS) && !defined(CONFIG_EXT3_FS_MODULE) && defined(CONFIG_EXT4_USE_FOR_EXT23) |
24b58424 TT |
4227 | static inline void register_as_ext3(void) |
4228 | { | |
4229 | int err = register_filesystem(&ext3_fs_type); | |
4230 | if (err) | |
4231 | printk(KERN_WARNING | |
4232 | "EXT4-fs: Unable to register as ext3 (%d)\n", err); | |
4233 | } | |
4234 | ||
4235 | static inline void unregister_as_ext3(void) | |
4236 | { | |
4237 | unregister_filesystem(&ext3_fs_type); | |
4238 | } | |
51b7e3c9 | 4239 | MODULE_ALIAS("ext3"); |
24b58424 TT |
4240 | #else |
4241 | static inline void register_as_ext3(void) { } | |
4242 | static inline void unregister_as_ext3(void) { } | |
4243 | #endif | |
4244 | ||
03010a33 TT |
4245 | static struct file_system_type ext4_fs_type = { |
4246 | .owner = THIS_MODULE, | |
4247 | .name = "ext4", | |
4248 | .get_sb = ext4_get_sb, | |
4249 | .kill_sb = kill_block_super, | |
4250 | .fs_flags = FS_REQUIRES_DEV, | |
4251 | }; | |
4252 | ||
617ba13b | 4253 | static int __init init_ext4_fs(void) |
ac27a0ec | 4254 | { |
c9de560d AT |
4255 | int err; |
4256 | ||
12e9b892 | 4257 | ext4_check_flag_values(); |
6fd058f7 TT |
4258 | err = init_ext4_system_zone(); |
4259 | if (err) | |
4260 | return err; | |
3197ebdb TT |
4261 | ext4_kset = kset_create_and_add("ext4", NULL, fs_kobj); |
4262 | if (!ext4_kset) | |
6fd058f7 | 4263 | goto out4; |
9f6200bb | 4264 | ext4_proc_root = proc_mkdir("fs/ext4", NULL); |
c9de560d | 4265 | err = init_ext4_mballoc(); |
ac27a0ec | 4266 | if (err) |
6fd058f7 | 4267 | goto out3; |
c9de560d AT |
4268 | |
4269 | err = init_ext4_xattr(); | |
4270 | if (err) | |
4271 | goto out2; | |
ac27a0ec DK |
4272 | err = init_inodecache(); |
4273 | if (err) | |
4274 | goto out1; | |
24b58424 TT |
4275 | register_as_ext2(); |
4276 | register_as_ext3(); | |
03010a33 | 4277 | err = register_filesystem(&ext4_fs_type); |
ac27a0ec DK |
4278 | if (err) |
4279 | goto out; | |
4280 | return 0; | |
4281 | out: | |
24b58424 TT |
4282 | unregister_as_ext2(); |
4283 | unregister_as_ext3(); | |
ac27a0ec DK |
4284 | destroy_inodecache(); |
4285 | out1: | |
617ba13b | 4286 | exit_ext4_xattr(); |
c9de560d AT |
4287 | out2: |
4288 | exit_ext4_mballoc(); | |
6fd058f7 TT |
4289 | out3: |
4290 | remove_proc_entry("fs/ext4", NULL); | |
4291 | kset_unregister(ext4_kset); | |
4292 | out4: | |
4293 | exit_ext4_system_zone(); | |
ac27a0ec DK |
4294 | return err; |
4295 | } | |
4296 | ||
617ba13b | 4297 | static void __exit exit_ext4_fs(void) |
ac27a0ec | 4298 | { |
24b58424 TT |
4299 | unregister_as_ext2(); |
4300 | unregister_as_ext3(); | |
03010a33 | 4301 | unregister_filesystem(&ext4_fs_type); |
ac27a0ec | 4302 | destroy_inodecache(); |
617ba13b | 4303 | exit_ext4_xattr(); |
c9de560d | 4304 | exit_ext4_mballoc(); |
9f6200bb | 4305 | remove_proc_entry("fs/ext4", NULL); |
3197ebdb | 4306 | kset_unregister(ext4_kset); |
6fd058f7 | 4307 | exit_ext4_system_zone(); |
ac27a0ec DK |
4308 | } |
4309 | ||
4310 | MODULE_AUTHOR("Remy Card, Stephen Tweedie, Andrew Morton, Andreas Dilger, Theodore Ts'o and others"); | |
83982b6f | 4311 | MODULE_DESCRIPTION("Fourth Extended Filesystem"); |
ac27a0ec | 4312 | MODULE_LICENSE("GPL"); |
617ba13b MC |
4313 | module_init(init_ext4_fs) |
4314 | module_exit(exit_ext4_fs) |