]> git.proxmox.com Git - mirror_ubuntu-hirsute-kernel.git/blame - fs/f2fs/super.c
f2fs: Don't overwrite all types of node to keep node chain
[mirror_ubuntu-hirsute-kernel.git] / fs / f2fs / super.c
CommitLineData
0a8165d7 1/*
aff063e2
JK
2 * fs/f2fs/super.c
3 *
4 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
5 * http://www.samsung.com/
6 *
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License version 2 as
9 * published by the Free Software Foundation.
10 */
11#include <linux/module.h>
12#include <linux/init.h>
13#include <linux/fs.h>
14#include <linux/statfs.h>
aff063e2
JK
15#include <linux/buffer_head.h>
16#include <linux/backing-dev.h>
17#include <linux/kthread.h>
18#include <linux/parser.h>
19#include <linux/mount.h>
20#include <linux/seq_file.h>
5e176d54 21#include <linux/proc_fs.h>
aff063e2
JK
22#include <linux/random.h>
23#include <linux/exportfs.h>
d3ee456d 24#include <linux/blkdev.h>
0abd675e 25#include <linux/quotaops.h>
aff063e2 26#include <linux/f2fs_fs.h>
b59d0bae 27#include <linux/sysfs.h>
4b2414d0 28#include <linux/quota.h>
aff063e2
JK
29
30#include "f2fs.h"
31#include "node.h"
5ec4e49f 32#include "segment.h"
aff063e2 33#include "xattr.h"
b59d0bae 34#include "gc.h"
db9f7c1a 35#include "trace.h"
aff063e2 36
a2a4a7e4
NJ
37#define CREATE_TRACE_POINTS
38#include <trace/events/f2fs.h>
39
aff063e2
JK
40static struct kmem_cache *f2fs_inode_cachep;
41
73faec4d 42#ifdef CONFIG_F2FS_FAULT_INJECTION
2c63fead
JK
43
44char *fault_name[FAULT_MAX] = {
45 [FAULT_KMALLOC] = "kmalloc",
628b3d14 46 [FAULT_KVMALLOC] = "kvmalloc",
c41f3cc3 47 [FAULT_PAGE_ALLOC] = "page alloc",
01eccef7 48 [FAULT_PAGE_GET] = "page get",
d62fe971 49 [FAULT_ALLOC_BIO] = "alloc bio",
cb78942b
JK
50 [FAULT_ALLOC_NID] = "alloc nid",
51 [FAULT_ORPHAN] = "orphan",
52 [FAULT_BLOCK] = "no more block",
53 [FAULT_DIR_DEPTH] = "too big dir depth",
53aa6bbf 54 [FAULT_EVICT_INODE] = "evict_inode fail",
14b44d23 55 [FAULT_TRUNCATE] = "truncate fail",
8b038c70 56 [FAULT_IO] = "IO error",
0f348028 57 [FAULT_CHECKPOINT] = "checkpoint error",
2c63fead 58};
08796897 59
1ecc0c5c
CY
60static void f2fs_build_fault_attr(struct f2fs_sb_info *sbi,
61 unsigned int rate)
08796897 62{
1ecc0c5c
CY
63 struct f2fs_fault_info *ffi = &sbi->fault_info;
64
08796897 65 if (rate) {
1ecc0c5c
CY
66 atomic_set(&ffi->inject_ops, 0);
67 ffi->inject_rate = rate;
68 ffi->inject_type = (1 << FAULT_MAX) - 1;
08796897 69 } else {
1ecc0c5c 70 memset(ffi, 0, sizeof(struct f2fs_fault_info));
08796897
SY
71 }
72}
73faec4d
JK
73#endif
74
2658e50d
JK
75/* f2fs-wide shrinker description */
76static struct shrinker f2fs_shrinker_info = {
77 .scan_objects = f2fs_shrink_scan,
78 .count_objects = f2fs_shrink_count,
79 .seeks = DEFAULT_SEEKS,
80};
81
aff063e2 82enum {
696c018c 83 Opt_gc_background,
aff063e2 84 Opt_disable_roll_forward,
2d834bf9 85 Opt_norecovery,
aff063e2 86 Opt_discard,
64058be9 87 Opt_nodiscard,
aff063e2 88 Opt_noheap,
7a20b8a6 89 Opt_heap,
4058c511 90 Opt_user_xattr,
aff063e2 91 Opt_nouser_xattr,
4058c511 92 Opt_acl,
aff063e2
JK
93 Opt_noacl,
94 Opt_active_logs,
95 Opt_disable_ext_identify,
444c580f 96 Opt_inline_xattr,
23cf7212 97 Opt_noinline_xattr,
6afc662e 98 Opt_inline_xattr_size,
8274de77 99 Opt_inline_data,
5efd3c6f 100 Opt_inline_dentry,
97c1794a 101 Opt_noinline_dentry,
6b4afdd7 102 Opt_flush_merge,
69e9e427 103 Opt_noflush_merge,
0f7b2abd 104 Opt_nobarrier,
d5053a34 105 Opt_fastboot,
89672159 106 Opt_extent_cache,
7daaea25 107 Opt_noextent_cache,
75342797 108 Opt_noinline_data,
343f40f0 109 Opt_data_flush,
7e65be49 110 Opt_reserve_root,
7c2e5963
JK
111 Opt_resgid,
112 Opt_resuid,
36abef4e 113 Opt_mode,
ec91538d 114 Opt_io_size_bits,
73faec4d 115 Opt_fault_injection,
6d94c74a
JK
116 Opt_lazytime,
117 Opt_nolazytime,
4b2414d0
CY
118 Opt_quota,
119 Opt_noquota,
0abd675e
CY
120 Opt_usrquota,
121 Opt_grpquota,
5c57132e 122 Opt_prjquota,
4b2414d0
CY
123 Opt_usrjquota,
124 Opt_grpjquota,
125 Opt_prjjquota,
126 Opt_offusrjquota,
127 Opt_offgrpjquota,
128 Opt_offprjjquota,
129 Opt_jqfmt_vfsold,
130 Opt_jqfmt_vfsv0,
131 Opt_jqfmt_vfsv1,
0cdd3195 132 Opt_whint,
07939627 133 Opt_alloc,
93cf93f1 134 Opt_fsync,
aff063e2
JK
135 Opt_err,
136};
137
138static match_table_t f2fs_tokens = {
696c018c 139 {Opt_gc_background, "background_gc=%s"},
aff063e2 140 {Opt_disable_roll_forward, "disable_roll_forward"},
2d834bf9 141 {Opt_norecovery, "norecovery"},
aff063e2 142 {Opt_discard, "discard"},
64058be9 143 {Opt_nodiscard, "nodiscard"},
aff063e2 144 {Opt_noheap, "no_heap"},
7a20b8a6 145 {Opt_heap, "heap"},
4058c511 146 {Opt_user_xattr, "user_xattr"},
aff063e2 147 {Opt_nouser_xattr, "nouser_xattr"},
4058c511 148 {Opt_acl, "acl"},
aff063e2
JK
149 {Opt_noacl, "noacl"},
150 {Opt_active_logs, "active_logs=%u"},
151 {Opt_disable_ext_identify, "disable_ext_identify"},
444c580f 152 {Opt_inline_xattr, "inline_xattr"},
23cf7212 153 {Opt_noinline_xattr, "noinline_xattr"},
6afc662e 154 {Opt_inline_xattr_size, "inline_xattr_size=%u"},
8274de77 155 {Opt_inline_data, "inline_data"},
5efd3c6f 156 {Opt_inline_dentry, "inline_dentry"},
97c1794a 157 {Opt_noinline_dentry, "noinline_dentry"},
6b4afdd7 158 {Opt_flush_merge, "flush_merge"},
69e9e427 159 {Opt_noflush_merge, "noflush_merge"},
0f7b2abd 160 {Opt_nobarrier, "nobarrier"},
d5053a34 161 {Opt_fastboot, "fastboot"},
89672159 162 {Opt_extent_cache, "extent_cache"},
7daaea25 163 {Opt_noextent_cache, "noextent_cache"},
75342797 164 {Opt_noinline_data, "noinline_data"},
343f40f0 165 {Opt_data_flush, "data_flush"},
7e65be49 166 {Opt_reserve_root, "reserve_root=%u"},
7c2e5963
JK
167 {Opt_resgid, "resgid=%u"},
168 {Opt_resuid, "resuid=%u"},
36abef4e 169 {Opt_mode, "mode=%s"},
ec91538d 170 {Opt_io_size_bits, "io_bits=%u"},
73faec4d 171 {Opt_fault_injection, "fault_injection=%u"},
6d94c74a
JK
172 {Opt_lazytime, "lazytime"},
173 {Opt_nolazytime, "nolazytime"},
4b2414d0
CY
174 {Opt_quota, "quota"},
175 {Opt_noquota, "noquota"},
0abd675e
CY
176 {Opt_usrquota, "usrquota"},
177 {Opt_grpquota, "grpquota"},
5c57132e 178 {Opt_prjquota, "prjquota"},
4b2414d0
CY
179 {Opt_usrjquota, "usrjquota=%s"},
180 {Opt_grpjquota, "grpjquota=%s"},
181 {Opt_prjjquota, "prjjquota=%s"},
182 {Opt_offusrjquota, "usrjquota="},
183 {Opt_offgrpjquota, "grpjquota="},
184 {Opt_offprjjquota, "prjjquota="},
185 {Opt_jqfmt_vfsold, "jqfmt=vfsold"},
186 {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
187 {Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
0cdd3195 188 {Opt_whint, "whint_mode=%s"},
07939627 189 {Opt_alloc, "alloc_mode=%s"},
93cf93f1 190 {Opt_fsync, "fsync_mode=%s"},
aff063e2
JK
191 {Opt_err, NULL},
192};
193
a07ef784
NJ
194void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...)
195{
196 struct va_format vaf;
197 va_list args;
198
199 va_start(args, fmt);
200 vaf.fmt = fmt;
201 vaf.va = &args;
a36c106d 202 printk_ratelimited("%sF2FS-fs (%s): %pV\n", level, sb->s_id, &vaf);
a07ef784
NJ
203 va_end(args);
204}
205
7e65be49
JK
206static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
207{
208 block_t limit = (sbi->user_block_count << 1) / 1000;
209
210 /* limit is 0.2% */
211 if (test_opt(sbi, RESERVE_ROOT) && sbi->root_reserved_blocks > limit) {
212 sbi->root_reserved_blocks = limit;
213 f2fs_msg(sbi->sb, KERN_INFO,
214 "Reduce reserved blocks for root = %u",
215 sbi->root_reserved_blocks);
216 }
7c2e5963
JK
217 if (!test_opt(sbi, RESERVE_ROOT) &&
218 (!uid_eq(sbi->s_resuid,
219 make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
220 !gid_eq(sbi->s_resgid,
221 make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
222 f2fs_msg(sbi->sb, KERN_INFO,
223 "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
224 from_kuid_munged(&init_user_ns, sbi->s_resuid),
225 from_kgid_munged(&init_user_ns, sbi->s_resgid));
7e65be49
JK
226}
227
aff063e2
JK
228static void init_once(void *foo)
229{
230 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
231
aff063e2
JK
232 inode_init_once(&fi->vfs_inode);
233}
234
4b2414d0
CY
235#ifdef CONFIG_QUOTA
236static const char * const quotatypes[] = INITQFNAMES;
237#define QTYPE2NAME(t) (quotatypes[t])
238static int f2fs_set_qf_name(struct super_block *sb, int qtype,
239 substring_t *args)
240{
241 struct f2fs_sb_info *sbi = F2FS_SB(sb);
242 char *qname;
243 int ret = -EINVAL;
244
245 if (sb_any_quota_loaded(sb) && !sbi->s_qf_names[qtype]) {
246 f2fs_msg(sb, KERN_ERR,
247 "Cannot change journaled "
248 "quota options when quota turned on");
249 return -EINVAL;
250 }
ea676733
JK
251 if (f2fs_sb_has_quota_ino(sb)) {
252 f2fs_msg(sb, KERN_INFO,
253 "QUOTA feature is enabled, so ignore qf_name");
254 return 0;
255 }
256
4b2414d0
CY
257 qname = match_strdup(args);
258 if (!qname) {
259 f2fs_msg(sb, KERN_ERR,
260 "Not enough memory for storing quotafile name");
261 return -EINVAL;
262 }
263 if (sbi->s_qf_names[qtype]) {
264 if (strcmp(sbi->s_qf_names[qtype], qname) == 0)
265 ret = 0;
266 else
267 f2fs_msg(sb, KERN_ERR,
268 "%s quota file already specified",
269 QTYPE2NAME(qtype));
270 goto errout;
271 }
272 if (strchr(qname, '/')) {
273 f2fs_msg(sb, KERN_ERR,
274 "quotafile must be on filesystem root");
275 goto errout;
276 }
277 sbi->s_qf_names[qtype] = qname;
278 set_opt(sbi, QUOTA);
279 return 0;
280errout:
281 kfree(qname);
282 return ret;
283}
284
285static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
286{
287 struct f2fs_sb_info *sbi = F2FS_SB(sb);
288
289 if (sb_any_quota_loaded(sb) && sbi->s_qf_names[qtype]) {
290 f2fs_msg(sb, KERN_ERR, "Cannot change journaled quota options"
291 " when quota turned on");
292 return -EINVAL;
293 }
294 kfree(sbi->s_qf_names[qtype]);
295 sbi->s_qf_names[qtype] = NULL;
296 return 0;
297}
298
299static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
300{
301 /*
302 * We do the test below only for project quotas. 'usrquota' and
303 * 'grpquota' mount options are allowed even without quota feature
304 * to support legacy quotas in quota files.
305 */
306 if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi->sb)) {
307 f2fs_msg(sbi->sb, KERN_ERR, "Project quota feature not enabled. "
308 "Cannot enable project quota enforcement.");
309 return -1;
310 }
311 if (sbi->s_qf_names[USRQUOTA] || sbi->s_qf_names[GRPQUOTA] ||
312 sbi->s_qf_names[PRJQUOTA]) {
313 if (test_opt(sbi, USRQUOTA) && sbi->s_qf_names[USRQUOTA])
314 clear_opt(sbi, USRQUOTA);
315
316 if (test_opt(sbi, GRPQUOTA) && sbi->s_qf_names[GRPQUOTA])
317 clear_opt(sbi, GRPQUOTA);
318
319 if (test_opt(sbi, PRJQUOTA) && sbi->s_qf_names[PRJQUOTA])
320 clear_opt(sbi, PRJQUOTA);
321
322 if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
323 test_opt(sbi, PRJQUOTA)) {
324 f2fs_msg(sbi->sb, KERN_ERR, "old and new quota "
325 "format mixing");
326 return -1;
327 }
328
329 if (!sbi->s_jquota_fmt) {
330 f2fs_msg(sbi->sb, KERN_ERR, "journaled quota format "
331 "not specified");
332 return -1;
333 }
334 }
ea676733
JK
335
336 if (f2fs_sb_has_quota_ino(sbi->sb) && sbi->s_jquota_fmt) {
337 f2fs_msg(sbi->sb, KERN_INFO,
338 "QUOTA feature is enabled, so ignore jquota_fmt");
339 sbi->s_jquota_fmt = 0;
340 }
deeedd71 341 if (f2fs_sb_has_quota_ino(sbi->sb) && f2fs_readonly(sbi->sb)) {
ea676733
JK
342 f2fs_msg(sbi->sb, KERN_INFO,
343 "Filesystem with quota feature cannot be mounted RDWR "
344 "without CONFIG_QUOTA");
345 return -1;
346 }
4b2414d0
CY
347 return 0;
348}
349#endif
350
696c018c
NJ
351static int parse_options(struct super_block *sb, char *options)
352{
353 struct f2fs_sb_info *sbi = F2FS_SB(sb);
09d54cdd 354 struct request_queue *q;
696c018c
NJ
355 substring_t args[MAX_OPT_ARGS];
356 char *p, *name;
357 int arg = 0;
7c2e5963
JK
358 kuid_t uid;
359 kgid_t gid;
4b2414d0
CY
360#ifdef CONFIG_QUOTA
361 int ret;
362#endif
696c018c
NJ
363
364 if (!options)
365 return 0;
366
367 while ((p = strsep(&options, ",")) != NULL) {
368 int token;
369 if (!*p)
370 continue;
371 /*
372 * Initialize args struct so we know whether arg was
373 * found; some options take optional arguments.
374 */
375 args[0].to = args[0].from = NULL;
376 token = match_token(p, f2fs_tokens, args);
377
378 switch (token) {
379 case Opt_gc_background:
380 name = match_strdup(&args[0]);
381
382 if (!name)
383 return -ENOMEM;
6aefd93b 384 if (strlen(name) == 2 && !strncmp(name, "on", 2)) {
696c018c 385 set_opt(sbi, BG_GC);
6aefd93b
JK
386 clear_opt(sbi, FORCE_FG_GC);
387 } else if (strlen(name) == 3 && !strncmp(name, "off", 3)) {
696c018c 388 clear_opt(sbi, BG_GC);
6aefd93b
JK
389 clear_opt(sbi, FORCE_FG_GC);
390 } else if (strlen(name) == 4 && !strncmp(name, "sync", 4)) {
391 set_opt(sbi, BG_GC);
392 set_opt(sbi, FORCE_FG_GC);
393 } else {
696c018c
NJ
394 kfree(name);
395 return -EINVAL;
396 }
397 kfree(name);
398 break;
399 case Opt_disable_roll_forward:
400 set_opt(sbi, DISABLE_ROLL_FORWARD);
401 break;
2d834bf9
JK
402 case Opt_norecovery:
403 /* this option mounts f2fs with ro */
404 set_opt(sbi, DISABLE_ROLL_FORWARD);
405 if (!f2fs_readonly(sb))
406 return -EINVAL;
407 break;
696c018c 408 case Opt_discard:
09d54cdd
CY
409 q = bdev_get_queue(sb->s_bdev);
410 if (blk_queue_discard(q)) {
411 set_opt(sbi, DISCARD);
ccd31cb2 412 } else if (!f2fs_sb_has_blkzoned(sb)) {
09d54cdd
CY
413 f2fs_msg(sb, KERN_WARNING,
414 "mounting with \"discard\" option, but "
415 "the device does not support discard");
416 }
696c018c 417 break;
64058be9 418 case Opt_nodiscard:
ccd31cb2 419 if (f2fs_sb_has_blkzoned(sb)) {
96ba2dec
DLM
420 f2fs_msg(sb, KERN_WARNING,
421 "discard is required for zoned block devices");
422 return -EINVAL;
423 }
64058be9 424 clear_opt(sbi, DISCARD);
487df616 425 break;
696c018c
NJ
426 case Opt_noheap:
427 set_opt(sbi, NOHEAP);
428 break;
7a20b8a6
JK
429 case Opt_heap:
430 clear_opt(sbi, NOHEAP);
431 break;
696c018c 432#ifdef CONFIG_F2FS_FS_XATTR
4058c511
KA
433 case Opt_user_xattr:
434 set_opt(sbi, XATTR_USER);
435 break;
696c018c
NJ
436 case Opt_nouser_xattr:
437 clear_opt(sbi, XATTR_USER);
438 break;
444c580f
JK
439 case Opt_inline_xattr:
440 set_opt(sbi, INLINE_XATTR);
441 break;
23cf7212
CY
442 case Opt_noinline_xattr:
443 clear_opt(sbi, INLINE_XATTR);
444 break;
6afc662e
CY
445 case Opt_inline_xattr_size:
446 if (args->from && match_int(args, &arg))
447 return -EINVAL;
448 set_opt(sbi, INLINE_XATTR_SIZE);
449 sbi->inline_xattr_size = arg;
450 break;
696c018c 451#else
4058c511
KA
452 case Opt_user_xattr:
453 f2fs_msg(sb, KERN_INFO,
454 "user_xattr options not supported");
455 break;
696c018c
NJ
456 case Opt_nouser_xattr:
457 f2fs_msg(sb, KERN_INFO,
458 "nouser_xattr options not supported");
459 break;
444c580f
JK
460 case Opt_inline_xattr:
461 f2fs_msg(sb, KERN_INFO,
462 "inline_xattr options not supported");
463 break;
23cf7212
CY
464 case Opt_noinline_xattr:
465 f2fs_msg(sb, KERN_INFO,
466 "noinline_xattr options not supported");
467 break;
696c018c
NJ
468#endif
469#ifdef CONFIG_F2FS_FS_POSIX_ACL
4058c511
KA
470 case Opt_acl:
471 set_opt(sbi, POSIX_ACL);
472 break;
696c018c
NJ
473 case Opt_noacl:
474 clear_opt(sbi, POSIX_ACL);
475 break;
476#else
4058c511
KA
477 case Opt_acl:
478 f2fs_msg(sb, KERN_INFO, "acl options not supported");
479 break;
696c018c
NJ
480 case Opt_noacl:
481 f2fs_msg(sb, KERN_INFO, "noacl options not supported");
482 break;
483#endif
484 case Opt_active_logs:
485 if (args->from && match_int(args, &arg))
486 return -EINVAL;
487 if (arg != 2 && arg != 4 && arg != NR_CURSEG_TYPE)
488 return -EINVAL;
489 sbi->active_logs = arg;
490 break;
491 case Opt_disable_ext_identify:
492 set_opt(sbi, DISABLE_EXT_IDENTIFY);
493 break;
8274de77
HL
494 case Opt_inline_data:
495 set_opt(sbi, INLINE_DATA);
496 break;
5efd3c6f
CY
497 case Opt_inline_dentry:
498 set_opt(sbi, INLINE_DENTRY);
499 break;
97c1794a
CY
500 case Opt_noinline_dentry:
501 clear_opt(sbi, INLINE_DENTRY);
502 break;
6b4afdd7
JK
503 case Opt_flush_merge:
504 set_opt(sbi, FLUSH_MERGE);
505 break;
69e9e427
JK
506 case Opt_noflush_merge:
507 clear_opt(sbi, FLUSH_MERGE);
508 break;
0f7b2abd
JK
509 case Opt_nobarrier:
510 set_opt(sbi, NOBARRIER);
511 break;
d5053a34
JK
512 case Opt_fastboot:
513 set_opt(sbi, FASTBOOT);
514 break;
89672159
CY
515 case Opt_extent_cache:
516 set_opt(sbi, EXTENT_CACHE);
517 break;
7daaea25
JK
518 case Opt_noextent_cache:
519 clear_opt(sbi, EXTENT_CACHE);
520 break;
75342797
WL
521 case Opt_noinline_data:
522 clear_opt(sbi, INLINE_DATA);
523 break;
343f40f0
CY
524 case Opt_data_flush:
525 set_opt(sbi, DATA_FLUSH);
526 break;
7e65be49
JK
527 case Opt_reserve_root:
528 if (args->from && match_int(args, &arg))
529 return -EINVAL;
530 if (test_opt(sbi, RESERVE_ROOT)) {
531 f2fs_msg(sb, KERN_INFO,
532 "Preserve previous reserve_root=%u",
533 sbi->root_reserved_blocks);
534 } else {
535 sbi->root_reserved_blocks = arg;
536 set_opt(sbi, RESERVE_ROOT);
537 }
538 break;
7c2e5963
JK
539 case Opt_resuid:
540 if (args->from && match_int(args, &arg))
541 return -EINVAL;
542 uid = make_kuid(current_user_ns(), arg);
543 if (!uid_valid(uid)) {
544 f2fs_msg(sb, KERN_ERR,
545 "Invalid uid value %d", arg);
546 return -EINVAL;
547 }
548 sbi->s_resuid = uid;
549 break;
550 case Opt_resgid:
551 if (args->from && match_int(args, &arg))
552 return -EINVAL;
553 gid = make_kgid(current_user_ns(), arg);
554 if (!gid_valid(gid)) {
555 f2fs_msg(sb, KERN_ERR,
556 "Invalid gid value %d", arg);
557 return -EINVAL;
558 }
559 sbi->s_resgid = gid;
560 break;
36abef4e
JK
561 case Opt_mode:
562 name = match_strdup(&args[0]);
563
564 if (!name)
565 return -ENOMEM;
566 if (strlen(name) == 8 &&
567 !strncmp(name, "adaptive", 8)) {
ccd31cb2 568 if (f2fs_sb_has_blkzoned(sb)) {
3adc57e9
DLM
569 f2fs_msg(sb, KERN_WARNING,
570 "adaptive mode is not allowed with "
571 "zoned block device feature");
572 kfree(name);
573 return -EINVAL;
574 }
52763a4b 575 set_opt_mode(sbi, F2FS_MOUNT_ADAPTIVE);
36abef4e
JK
576 } else if (strlen(name) == 3 &&
577 !strncmp(name, "lfs", 3)) {
52763a4b 578 set_opt_mode(sbi, F2FS_MOUNT_LFS);
36abef4e
JK
579 } else {
580 kfree(name);
581 return -EINVAL;
582 }
583 kfree(name);
584 break;
ec91538d
JK
585 case Opt_io_size_bits:
586 if (args->from && match_int(args, &arg))
587 return -EINVAL;
588 if (arg > __ilog2_u32(BIO_MAX_PAGES)) {
589 f2fs_msg(sb, KERN_WARNING,
590 "Not support %d, larger than %d",
591 1 << arg, BIO_MAX_PAGES);
592 return -EINVAL;
593 }
594 sbi->write_io_size_bits = arg;
595 break;
73faec4d
JK
596 case Opt_fault_injection:
597 if (args->from && match_int(args, &arg))
598 return -EINVAL;
599#ifdef CONFIG_F2FS_FAULT_INJECTION
1ecc0c5c 600 f2fs_build_fault_attr(sbi, arg);
0cc0dec2 601 set_opt(sbi, FAULT_INJECTION);
73faec4d
JK
602#else
603 f2fs_msg(sb, KERN_INFO,
604 "FAULT_INJECTION was not selected");
605#endif
606 break;
6d94c74a 607 case Opt_lazytime:
1751e8a6 608 sb->s_flags |= SB_LAZYTIME;
6d94c74a
JK
609 break;
610 case Opt_nolazytime:
1751e8a6 611 sb->s_flags &= ~SB_LAZYTIME;
6d94c74a 612 break;
0abd675e 613#ifdef CONFIG_QUOTA
4b2414d0 614 case Opt_quota:
0abd675e
CY
615 case Opt_usrquota:
616 set_opt(sbi, USRQUOTA);
617 break;
618 case Opt_grpquota:
619 set_opt(sbi, GRPQUOTA);
620 break;
5c57132e
CY
621 case Opt_prjquota:
622 set_opt(sbi, PRJQUOTA);
623 break;
4b2414d0
CY
624 case Opt_usrjquota:
625 ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
626 if (ret)
627 return ret;
628 break;
629 case Opt_grpjquota:
630 ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
631 if (ret)
632 return ret;
633 break;
634 case Opt_prjjquota:
635 ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
636 if (ret)
637 return ret;
638 break;
639 case Opt_offusrjquota:
640 ret = f2fs_clear_qf_name(sb, USRQUOTA);
641 if (ret)
642 return ret;
643 break;
644 case Opt_offgrpjquota:
645 ret = f2fs_clear_qf_name(sb, GRPQUOTA);
646 if (ret)
647 return ret;
648 break;
649 case Opt_offprjjquota:
650 ret = f2fs_clear_qf_name(sb, PRJQUOTA);
651 if (ret)
652 return ret;
653 break;
654 case Opt_jqfmt_vfsold:
655 sbi->s_jquota_fmt = QFMT_VFS_OLD;
656 break;
657 case Opt_jqfmt_vfsv0:
658 sbi->s_jquota_fmt = QFMT_VFS_V0;
659 break;
660 case Opt_jqfmt_vfsv1:
661 sbi->s_jquota_fmt = QFMT_VFS_V1;
662 break;
663 case Opt_noquota:
664 clear_opt(sbi, QUOTA);
665 clear_opt(sbi, USRQUOTA);
666 clear_opt(sbi, GRPQUOTA);
667 clear_opt(sbi, PRJQUOTA);
668 break;
0abd675e 669#else
4b2414d0 670 case Opt_quota:
0abd675e
CY
671 case Opt_usrquota:
672 case Opt_grpquota:
5c57132e 673 case Opt_prjquota:
4b2414d0
CY
674 case Opt_usrjquota:
675 case Opt_grpjquota:
676 case Opt_prjjquota:
677 case Opt_offusrjquota:
678 case Opt_offgrpjquota:
679 case Opt_offprjjquota:
680 case Opt_jqfmt_vfsold:
681 case Opt_jqfmt_vfsv0:
682 case Opt_jqfmt_vfsv1:
683 case Opt_noquota:
0abd675e
CY
684 f2fs_msg(sb, KERN_INFO,
685 "quota operations not supported");
686 break;
687#endif
0cdd3195
HL
688 case Opt_whint:
689 name = match_strdup(&args[0]);
690 if (!name)
691 return -ENOMEM;
692 if (strlen(name) == 10 &&
693 !strncmp(name, "user-based", 10)) {
694 sbi->whint_mode = WHINT_MODE_USER;
695 } else if (strlen(name) == 3 &&
696 !strncmp(name, "off", 3)) {
697 sbi->whint_mode = WHINT_MODE_OFF;
f2e703f9
HL
698 } else if (strlen(name) == 8 &&
699 !strncmp(name, "fs-based", 8)) {
700 sbi->whint_mode = WHINT_MODE_FS;
0cdd3195
HL
701 } else {
702 kfree(name);
703 return -EINVAL;
704 }
705 kfree(name);
706 break;
07939627
JK
707 case Opt_alloc:
708 name = match_strdup(&args[0]);
709 if (!name)
710 return -ENOMEM;
711
712 if (strlen(name) == 7 &&
713 !strncmp(name, "default", 7)) {
714 sbi->alloc_mode = ALLOC_MODE_DEFAULT;
715 } else if (strlen(name) == 5 &&
716 !strncmp(name, "reuse", 5)) {
717 sbi->alloc_mode = ALLOC_MODE_REUSE;
718 } else {
719 kfree(name);
720 return -EINVAL;
721 }
722 kfree(name);
723 break;
93cf93f1
JZ
724 case Opt_fsync:
725 name = match_strdup(&args[0]);
726 if (!name)
727 return -ENOMEM;
728 if (strlen(name) == 5 &&
729 !strncmp(name, "posix", 5)) {
730 sbi->fsync_mode = FSYNC_MODE_POSIX;
731 } else if (strlen(name) == 6 &&
732 !strncmp(name, "strict", 6)) {
733 sbi->fsync_mode = FSYNC_MODE_STRICT;
734 } else {
735 kfree(name);
736 return -EINVAL;
737 }
738 kfree(name);
739 break;
696c018c
NJ
740 default:
741 f2fs_msg(sb, KERN_ERR,
742 "Unrecognized mount option \"%s\" or missing value",
743 p);
744 return -EINVAL;
745 }
746 }
4b2414d0
CY
747#ifdef CONFIG_QUOTA
748 if (f2fs_check_quota_options(sbi))
749 return -EINVAL;
750#endif
ec91538d
JK
751
752 if (F2FS_IO_SIZE_BITS(sbi) && !test_opt(sbi, LFS)) {
753 f2fs_msg(sb, KERN_ERR,
754 "Should set mode=lfs with %uKB-sized IO",
755 F2FS_IO_SIZE_KB(sbi));
756 return -EINVAL;
757 }
6afc662e
CY
758
759 if (test_opt(sbi, INLINE_XATTR_SIZE)) {
4d817ae0
CY
760 if (!f2fs_sb_has_extra_attr(sb) ||
761 !f2fs_sb_has_flexible_inline_xattr(sb)) {
762 f2fs_msg(sb, KERN_ERR,
763 "extra_attr or flexible_inline_xattr "
764 "feature is off");
765 return -EINVAL;
766 }
6afc662e
CY
767 if (!test_opt(sbi, INLINE_XATTR)) {
768 f2fs_msg(sb, KERN_ERR,
769 "inline_xattr_size option should be "
770 "set with inline_xattr option");
771 return -EINVAL;
772 }
773 if (!sbi->inline_xattr_size ||
774 sbi->inline_xattr_size >= DEF_ADDRS_PER_INODE -
775 F2FS_TOTAL_EXTRA_ATTR_SIZE -
776 DEF_INLINE_RESERVED_SIZE -
777 DEF_MIN_INLINE_SIZE) {
778 f2fs_msg(sb, KERN_ERR,
779 "inline xattr size is out of range");
780 return -EINVAL;
781 }
782 }
0cdd3195
HL
783
784 /* Not pass down write hints if the number of active logs is lesser
785 * than NR_CURSEG_TYPE.
786 */
787 if (sbi->active_logs != NR_CURSEG_TYPE)
788 sbi->whint_mode = WHINT_MODE_OFF;
696c018c
NJ
789 return 0;
790}
791
aff063e2
JK
792static struct inode *f2fs_alloc_inode(struct super_block *sb)
793{
794 struct f2fs_inode_info *fi;
795
a0acdfe0 796 fi = kmem_cache_alloc(f2fs_inode_cachep, GFP_F2FS_ZERO);
aff063e2
JK
797 if (!fi)
798 return NULL;
799
800 init_once((void *) fi);
801
434720fa 802 /* Initialize f2fs-specific inode info */
204706c7 803 atomic_set(&fi->dirty_pages, 0);
aff063e2
JK
804 fi->i_current_depth = 1;
805 fi->i_advise = 0;
d928bfbf 806 init_rwsem(&fi->i_sem);
2710fd7e 807 INIT_LIST_HEAD(&fi->dirty_list);
0f18b462 808 INIT_LIST_HEAD(&fi->gdirty_list);
57864ae5 809 INIT_LIST_HEAD(&fi->inmem_ilist);
88b88a66
JK
810 INIT_LIST_HEAD(&fi->inmem_pages);
811 mutex_init(&fi->inmem_lock);
82e0a5aa
CY
812 init_rwsem(&fi->dio_rwsem[READ]);
813 init_rwsem(&fi->dio_rwsem[WRITE]);
5a3a2d83 814 init_rwsem(&fi->i_mmap_sem);
27161f13 815 init_rwsem(&fi->i_xattr_sem);
aff063e2 816
0abd675e
CY
817#ifdef CONFIG_QUOTA
818 memset(&fi->i_dquot, 0, sizeof(fi->i_dquot));
819 fi->i_reserved_quota = 0;
820#endif
ab9fa662
JK
821 /* Will be used by directory only */
822 fi->i_dir_level = F2FS_SB(sb)->dir_level;
f2470371 823
aff063e2
JK
824 return &fi->vfs_inode;
825}
826
531ad7d5
JK
827static int f2fs_drop_inode(struct inode *inode)
828{
b8d96a30 829 int ret;
531ad7d5
JK
830 /*
831 * This is to avoid a deadlock condition like below.
832 * writeback_single_inode(inode)
833 * - f2fs_write_data_page
834 * - f2fs_gc -> iput -> evict
835 * - inode_wait_for_writeback(inode)
836 */
0f18b462 837 if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
06e1bc05 838 if (!inode->i_nlink && !is_bad_inode(inode)) {
3e72f721
JK
839 /* to avoid evict_inode call simultaneously */
840 atomic_inc(&inode->i_count);
06e1bc05
JK
841 spin_unlock(&inode->i_lock);
842
843 /* some remained atomic pages should discarded */
844 if (f2fs_is_atomic_file(inode))
29b96b54 845 drop_inmem_pages(inode);
06e1bc05 846
3e72f721
JK
847 /* should remain fi->extent_tree for writepage */
848 f2fs_destroy_extent_node(inode);
849
06e1bc05 850 sb_start_intwrite(inode->i_sb);
fc9581c8 851 f2fs_i_size_write(inode, 0);
06e1bc05
JK
852
853 if (F2FS_HAS_BLOCKS(inode))
9a449e9c 854 f2fs_truncate(inode);
06e1bc05
JK
855
856 sb_end_intwrite(inode->i_sb);
857
06e1bc05 858 spin_lock(&inode->i_lock);
3e72f721 859 atomic_dec(&inode->i_count);
06e1bc05 860 }
b8d96a30 861 trace_f2fs_drop_inode(inode, 0);
531ad7d5 862 return 0;
06e1bc05 863 }
b8d96a30
HP
864 ret = generic_drop_inode(inode);
865 trace_f2fs_drop_inode(inode, ret);
866 return ret;
531ad7d5
JK
867}
868
7c45729a 869int f2fs_inode_dirtied(struct inode *inode, bool sync)
b3783873 870{
0f18b462 871 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
7c45729a 872 int ret = 0;
0f18b462 873
0f18b462
JK
874 spin_lock(&sbi->inode_lock[DIRTY_META]);
875 if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
7c45729a
JK
876 ret = 1;
877 } else {
878 set_inode_flag(inode, FI_DIRTY_INODE);
879 stat_inc_dirty_inode(sbi, DIRTY_META);
0f18b462 880 }
7c45729a
JK
881 if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
882 list_add_tail(&F2FS_I(inode)->gdirty_list,
0f18b462 883 &sbi->inode_list[DIRTY_META]);
7c45729a
JK
884 inc_page_count(sbi, F2FS_DIRTY_IMETA);
885 }
338bbfa0 886 spin_unlock(&sbi->inode_lock[DIRTY_META]);
7c45729a 887 return ret;
0f18b462
JK
888}
889
890void f2fs_inode_synced(struct inode *inode)
891{
892 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
893
894 spin_lock(&sbi->inode_lock[DIRTY_META]);
895 if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
896 spin_unlock(&sbi->inode_lock[DIRTY_META]);
897 return;
898 }
7c45729a
JK
899 if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
900 list_del_init(&F2FS_I(inode)->gdirty_list);
901 dec_page_count(sbi, F2FS_DIRTY_IMETA);
902 }
0f18b462 903 clear_inode_flag(inode, FI_DIRTY_INODE);
26de9b11 904 clear_inode_flag(inode, FI_AUTO_RECOVER);
0f18b462 905 stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
338bbfa0 906 spin_unlock(&sbi->inode_lock[DIRTY_META]);
b3783873
JK
907}
908
b56ab837
JK
909/*
910 * f2fs_dirty_inode() is called from __mark_inode_dirty()
911 *
912 * We should call set_dirty_inode to write the dirty inode through write_inode.
913 */
914static void f2fs_dirty_inode(struct inode *inode, int flags)
915{
916 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
917
918 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
919 inode->i_ino == F2FS_META_INO(sbi))
920 return;
921
922 if (flags == I_DIRTY_TIME)
923 return;
924
925 if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
926 clear_inode_flag(inode, FI_AUTO_RECOVER);
927
7c45729a 928 f2fs_inode_dirtied(inode, false);
b56ab837
JK
929}
930
aff063e2
JK
931static void f2fs_i_callback(struct rcu_head *head)
932{
933 struct inode *inode = container_of(head, struct inode, i_rcu);
934 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
935}
936
25ca923b 937static void f2fs_destroy_inode(struct inode *inode)
aff063e2
JK
938{
939 call_rcu(&inode->i_rcu, f2fs_i_callback);
940}
941
523be8a6
JK
942static void destroy_percpu_info(struct f2fs_sb_info *sbi)
943{
41382ec4 944 percpu_counter_destroy(&sbi->alloc_valid_block_count);
513c5f37 945 percpu_counter_destroy(&sbi->total_valid_inode_count);
523be8a6
JK
946}
947
3c62be17
JK
948static void destroy_device_list(struct f2fs_sb_info *sbi)
949{
950 int i;
951
952 for (i = 0; i < sbi->s_ndevs; i++) {
953 blkdev_put(FDEV(i).bdev, FMODE_EXCL);
954#ifdef CONFIG_BLK_DEV_ZONED
955 kfree(FDEV(i).blkz_type);
956#endif
957 }
958 kfree(sbi->devs);
959}
960
aff063e2
JK
961static void f2fs_put_super(struct super_block *sb)
962{
963 struct f2fs_sb_info *sbi = F2FS_SB(sb);
a398101a 964 int i;
cf5c759f 965 bool dropped;
aff063e2 966
0abd675e 967 f2fs_quota_off_umount(sb);
aff063e2 968
2658e50d
JK
969 /* prevent remaining shrinker jobs */
970 mutex_lock(&sbi->umount_mutex);
971
85dc2f2c
JK
972 /*
973 * We don't need to do checkpoint when superblock is clean.
974 * But, the previous checkpoint was not done by umount, it needs to do
975 * clean checkpoint again.
976 */
caf0047e 977 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
aaec2b1d 978 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
75ab4cb8
JK
979 struct cp_control cpc = {
980 .reason = CP_UMOUNT,
981 };
982 write_checkpoint(sbi, &cpc);
983 }
aff063e2 984
4e6a8d9b 985 /* be sure to wait for any on-going discard commands */
cf5c759f 986 dropped = f2fs_wait_discard_bios(sbi);
4e6a8d9b 987
cf5c759f 988 if (f2fs_discard_en(sbi) && !sbi->discard_blks && !dropped) {
1f43e2ad
CY
989 struct cp_control cpc = {
990 .reason = CP_UMOUNT | CP_TRIMMED,
991 };
992 write_checkpoint(sbi, &cpc);
993 }
994
eca616f8
JK
995 /* write_checkpoint can update stat informaion */
996 f2fs_destroy_stats(sbi);
997
cf779cab
JK
998 /*
999 * normally superblock is clean, so we need to release this.
1000 * In addition, EIO will skip do checkpoint, we need this as well.
1001 */
74ef9241 1002 release_ino_entry(sbi, true);
6f12ac25 1003
2658e50d
JK
1004 f2fs_leave_shrinker(sbi);
1005 mutex_unlock(&sbi->umount_mutex);
1006
17c19120 1007 /* our cp_error case, we can wait for any writeback page */
b9109b0e 1008 f2fs_flush_merged_writes(sbi);
17c19120 1009
aff063e2
JK
1010 iput(sbi->node_inode);
1011 iput(sbi->meta_inode);
1012
1013 /* destroy f2fs internal modules */
1014 destroy_node_manager(sbi);
1015 destroy_segment_manager(sbi);
1016
1017 kfree(sbi->ckpt);
a398101a 1018
dc6b2055 1019 f2fs_unregister_sysfs(sbi);
aff063e2
JK
1020
1021 sb->s_fs_info = NULL;
43b6573b
KM
1022 if (sbi->s_chksum_driver)
1023 crypto_free_shash(sbi->s_chksum_driver);
b39f0de2 1024 kfree(sbi->raw_super);
523be8a6 1025
3c62be17 1026 destroy_device_list(sbi);
b6895e8f 1027 mempool_destroy(sbi->write_io_dummy);
4b2414d0
CY
1028#ifdef CONFIG_QUOTA
1029 for (i = 0; i < MAXQUOTAS; i++)
1030 kfree(sbi->s_qf_names[i]);
1031#endif
523be8a6 1032 destroy_percpu_info(sbi);
a912b54d
JK
1033 for (i = 0; i < NR_PAGE_TYPE; i++)
1034 kfree(sbi->write_io[i]);
aff063e2
JK
1035 kfree(sbi);
1036}
1037
1038int f2fs_sync_fs(struct super_block *sb, int sync)
1039{
1040 struct f2fs_sb_info *sbi = F2FS_SB(sb);
c34f42e2 1041 int err = 0;
aff063e2 1042
1f227a3e
JK
1043 if (unlikely(f2fs_cp_error(sbi)))
1044 return 0;
1045
a2a4a7e4
NJ
1046 trace_f2fs_sync_fs(sb, sync);
1047
4b2414d0
CY
1048 if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1049 return -EAGAIN;
1050
b7473754 1051 if (sync) {
d5053a34
JK
1052 struct cp_control cpc;
1053
119ee914
JK
1054 cpc.reason = __get_cp_reason(sbi);
1055
b7473754 1056 mutex_lock(&sbi->gc_mutex);
c34f42e2 1057 err = write_checkpoint(sbi, &cpc);
b7473754 1058 mutex_unlock(&sbi->gc_mutex);
b7473754 1059 }
05ca3632 1060 f2fs_trace_ios(NULL, 1);
aff063e2 1061
c34f42e2 1062 return err;
aff063e2
JK
1063}
1064
d6212a5f
CL
1065static int f2fs_freeze(struct super_block *sb)
1066{
77888c1e 1067 if (f2fs_readonly(sb))
d6212a5f
CL
1068 return 0;
1069
b4b9d34c
JK
1070 /* IO error happened before */
1071 if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1072 return -EIO;
1073
1074 /* must be clean, since sync_filesystem() was already called */
1075 if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1076 return -EINVAL;
1077 return 0;
d6212a5f
CL
1078}
1079
1080static int f2fs_unfreeze(struct super_block *sb)
1081{
1082 return 0;
1083}
1084
ddc34e32
CY
1085#ifdef CONFIG_QUOTA
1086static int f2fs_statfs_project(struct super_block *sb,
1087 kprojid_t projid, struct kstatfs *buf)
1088{
1089 struct kqid qid;
1090 struct dquot *dquot;
1091 u64 limit;
1092 u64 curblock;
1093
1094 qid = make_kqid_projid(projid);
1095 dquot = dqget(sb, qid);
1096 if (IS_ERR(dquot))
1097 return PTR_ERR(dquot);
1098 spin_lock(&dq_data_lock);
1099
1100 limit = (dquot->dq_dqb.dqb_bsoftlimit ?
1101 dquot->dq_dqb.dqb_bsoftlimit :
1102 dquot->dq_dqb.dqb_bhardlimit) >> sb->s_blocksize_bits;
1103 if (limit && buf->f_blocks > limit) {
1104 curblock = dquot->dq_dqb.dqb_curspace >> sb->s_blocksize_bits;
1105 buf->f_blocks = limit;
1106 buf->f_bfree = buf->f_bavail =
1107 (buf->f_blocks > curblock) ?
1108 (buf->f_blocks - curblock) : 0;
1109 }
1110
1111 limit = dquot->dq_dqb.dqb_isoftlimit ?
1112 dquot->dq_dqb.dqb_isoftlimit :
1113 dquot->dq_dqb.dqb_ihardlimit;
1114 if (limit && buf->f_files > limit) {
1115 buf->f_files = limit;
1116 buf->f_ffree =
1117 (buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1118 (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1119 }
1120
1121 spin_unlock(&dq_data_lock);
1122 dqput(dquot);
1123 return 0;
1124}
1125#endif
1126
aff063e2
JK
1127static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1128{
1129 struct super_block *sb = dentry->d_sb;
1130 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1131 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
f66c027e 1132 block_t total_count, user_block_count, start_count;
0cc091d0 1133 u64 avail_node_count;
aff063e2
JK
1134
1135 total_count = le64_to_cpu(sbi->raw_super->block_count);
1136 user_block_count = sbi->user_block_count;
1137 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
aff063e2
JK
1138 buf->f_type = F2FS_SUPER_MAGIC;
1139 buf->f_bsize = sbi->blocksize;
1140
1141 buf->f_blocks = total_count - start_count;
f66c027e 1142 buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
80d42145 1143 sbi->current_reserved_blocks;
7e65be49
JK
1144 if (buf->f_bfree > sbi->root_reserved_blocks)
1145 buf->f_bavail = buf->f_bfree - sbi->root_reserved_blocks;
1146 else
1147 buf->f_bavail = 0;
aff063e2 1148
292c196a
CY
1149 avail_node_count = sbi->total_node_count - sbi->nquota_files -
1150 F2FS_RESERVED_NODE_NUM;
0cc091d0
JK
1151
1152 if (avail_node_count > user_block_count) {
1153 buf->f_files = user_block_count;
1154 buf->f_ffree = buf->f_bavail;
1155 } else {
1156 buf->f_files = avail_node_count;
1157 buf->f_ffree = min(avail_node_count - valid_node_count(sbi),
1158 buf->f_bavail);
1159 }
aff063e2 1160
5a20d339 1161 buf->f_namelen = F2FS_NAME_LEN;
aff063e2
JK
1162 buf->f_fsid.val[0] = (u32)id;
1163 buf->f_fsid.val[1] = (u32)(id >> 32);
1164
ddc34e32
CY
1165#ifdef CONFIG_QUOTA
1166 if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1167 sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1168 f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1169 }
1170#endif
aff063e2
JK
1171 return 0;
1172}
1173
4b2414d0
CY
1174static inline void f2fs_show_quota_options(struct seq_file *seq,
1175 struct super_block *sb)
1176{
1177#ifdef CONFIG_QUOTA
1178 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1179
1180 if (sbi->s_jquota_fmt) {
1181 char *fmtname = "";
1182
1183 switch (sbi->s_jquota_fmt) {
1184 case QFMT_VFS_OLD:
1185 fmtname = "vfsold";
1186 break;
1187 case QFMT_VFS_V0:
1188 fmtname = "vfsv0";
1189 break;
1190 case QFMT_VFS_V1:
1191 fmtname = "vfsv1";
1192 break;
1193 }
1194 seq_printf(seq, ",jqfmt=%s", fmtname);
1195 }
1196
1197 if (sbi->s_qf_names[USRQUOTA])
1198 seq_show_option(seq, "usrjquota", sbi->s_qf_names[USRQUOTA]);
1199
1200 if (sbi->s_qf_names[GRPQUOTA])
1201 seq_show_option(seq, "grpjquota", sbi->s_qf_names[GRPQUOTA]);
1202
1203 if (sbi->s_qf_names[PRJQUOTA])
1204 seq_show_option(seq, "prjjquota", sbi->s_qf_names[PRJQUOTA]);
1205#endif
1206}
1207
aff063e2
JK
1208static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1209{
1210 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1211
6aefd93b
JK
1212 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, BG_GC)) {
1213 if (test_opt(sbi, FORCE_FG_GC))
1214 seq_printf(seq, ",background_gc=%s", "sync");
1215 else
1216 seq_printf(seq, ",background_gc=%s", "on");
1217 } else {
696c018c 1218 seq_printf(seq, ",background_gc=%s", "off");
6aefd93b 1219 }
aff063e2
JK
1220 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1221 seq_puts(seq, ",disable_roll_forward");
1222 if (test_opt(sbi, DISCARD))
1223 seq_puts(seq, ",discard");
1224 if (test_opt(sbi, NOHEAP))
7a20b8a6
JK
1225 seq_puts(seq, ",no_heap");
1226 else
1227 seq_puts(seq, ",heap");
aff063e2
JK
1228#ifdef CONFIG_F2FS_FS_XATTR
1229 if (test_opt(sbi, XATTR_USER))
1230 seq_puts(seq, ",user_xattr");
1231 else
1232 seq_puts(seq, ",nouser_xattr");
444c580f
JK
1233 if (test_opt(sbi, INLINE_XATTR))
1234 seq_puts(seq, ",inline_xattr");
23cf7212
CY
1235 else
1236 seq_puts(seq, ",noinline_xattr");
6afc662e
CY
1237 if (test_opt(sbi, INLINE_XATTR_SIZE))
1238 seq_printf(seq, ",inline_xattr_size=%u",
1239 sbi->inline_xattr_size);
aff063e2
JK
1240#endif
1241#ifdef CONFIG_F2FS_FS_POSIX_ACL
1242 if (test_opt(sbi, POSIX_ACL))
1243 seq_puts(seq, ",acl");
1244 else
1245 seq_puts(seq, ",noacl");
1246#endif
1247 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
aa43507f 1248 seq_puts(seq, ",disable_ext_identify");
8274de77
HL
1249 if (test_opt(sbi, INLINE_DATA))
1250 seq_puts(seq, ",inline_data");
75342797
WL
1251 else
1252 seq_puts(seq, ",noinline_data");
5efd3c6f
CY
1253 if (test_opt(sbi, INLINE_DENTRY))
1254 seq_puts(seq, ",inline_dentry");
97c1794a
CY
1255 else
1256 seq_puts(seq, ",noinline_dentry");
b270ad6f 1257 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
6b4afdd7 1258 seq_puts(seq, ",flush_merge");
0f7b2abd
JK
1259 if (test_opt(sbi, NOBARRIER))
1260 seq_puts(seq, ",nobarrier");
d5053a34
JK
1261 if (test_opt(sbi, FASTBOOT))
1262 seq_puts(seq, ",fastboot");
89672159
CY
1263 if (test_opt(sbi, EXTENT_CACHE))
1264 seq_puts(seq, ",extent_cache");
7daaea25
JK
1265 else
1266 seq_puts(seq, ",noextent_cache");
343f40f0
CY
1267 if (test_opt(sbi, DATA_FLUSH))
1268 seq_puts(seq, ",data_flush");
36abef4e
JK
1269
1270 seq_puts(seq, ",mode=");
1271 if (test_opt(sbi, ADAPTIVE))
1272 seq_puts(seq, "adaptive");
1273 else if (test_opt(sbi, LFS))
1274 seq_puts(seq, "lfs");
aff063e2 1275 seq_printf(seq, ",active_logs=%u", sbi->active_logs);
7e65be49 1276 if (test_opt(sbi, RESERVE_ROOT))
7c2e5963
JK
1277 seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
1278 sbi->root_reserved_blocks,
1279 from_kuid_munged(&init_user_ns, sbi->s_resuid),
1280 from_kgid_munged(&init_user_ns, sbi->s_resgid));
ec91538d
JK
1281 if (F2FS_IO_SIZE_BITS(sbi))
1282 seq_printf(seq, ",io_size=%uKB", F2FS_IO_SIZE_KB(sbi));
0cc0dec2
KX
1283#ifdef CONFIG_F2FS_FAULT_INJECTION
1284 if (test_opt(sbi, FAULT_INJECTION))
44529f89
CY
1285 seq_printf(seq, ",fault_injection=%u",
1286 sbi->fault_info.inject_rate);
0cc0dec2 1287#endif
0abd675e 1288#ifdef CONFIG_QUOTA
4b2414d0
CY
1289 if (test_opt(sbi, QUOTA))
1290 seq_puts(seq, ",quota");
0abd675e
CY
1291 if (test_opt(sbi, USRQUOTA))
1292 seq_puts(seq, ",usrquota");
1293 if (test_opt(sbi, GRPQUOTA))
1294 seq_puts(seq, ",grpquota");
5c57132e
CY
1295 if (test_opt(sbi, PRJQUOTA))
1296 seq_puts(seq, ",prjquota");
0cc0dec2 1297#endif
4b2414d0 1298 f2fs_show_quota_options(seq, sbi->sb);
0cdd3195
HL
1299 if (sbi->whint_mode == WHINT_MODE_USER)
1300 seq_printf(seq, ",whint_mode=%s", "user-based");
f2e703f9
HL
1301 else if (sbi->whint_mode == WHINT_MODE_FS)
1302 seq_printf(seq, ",whint_mode=%s", "fs-based");
aff063e2 1303
07939627
JK
1304 if (sbi->alloc_mode == ALLOC_MODE_DEFAULT)
1305 seq_printf(seq, ",alloc_mode=%s", "default");
1306 else if (sbi->alloc_mode == ALLOC_MODE_REUSE)
1307 seq_printf(seq, ",alloc_mode=%s", "reuse");
93cf93f1
JZ
1308
1309 if (sbi->fsync_mode == FSYNC_MODE_POSIX)
1310 seq_printf(seq, ",fsync_mode=%s", "posix");
1311 else if (sbi->fsync_mode == FSYNC_MODE_STRICT)
1312 seq_printf(seq, ",fsync_mode=%s", "strict");
aff063e2
JK
1313 return 0;
1314}
1315
498c5e9f
YH
1316static void default_options(struct f2fs_sb_info *sbi)
1317{
1318 /* init some FS parameters */
1319 sbi->active_logs = NR_CURSEG_TYPE;
6afc662e 1320 sbi->inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
0cdd3195 1321 sbi->whint_mode = WHINT_MODE_OFF;
07939627 1322 sbi->alloc_mode = ALLOC_MODE_DEFAULT;
93cf93f1 1323 sbi->fsync_mode = FSYNC_MODE_POSIX;
782911f4 1324 sbi->readdir_ra = 1;
498c5e9f
YH
1325
1326 set_opt(sbi, BG_GC);
39133a50 1327 set_opt(sbi, INLINE_XATTR);
498c5e9f 1328 set_opt(sbi, INLINE_DATA);
97c1794a 1329 set_opt(sbi, INLINE_DENTRY);
3e72f721 1330 set_opt(sbi, EXTENT_CACHE);
7a20b8a6 1331 set_opt(sbi, NOHEAP);
1751e8a6 1332 sbi->sb->s_flags |= SB_LAZYTIME;
69e9e427 1333 set_opt(sbi, FLUSH_MERGE);
ccd31cb2 1334 if (f2fs_sb_has_blkzoned(sbi->sb)) {
52763a4b
JK
1335 set_opt_mode(sbi, F2FS_MOUNT_LFS);
1336 set_opt(sbi, DISCARD);
1337 } else {
1338 set_opt_mode(sbi, F2FS_MOUNT_ADAPTIVE);
1339 }
498c5e9f
YH
1340
1341#ifdef CONFIG_F2FS_FS_XATTR
1342 set_opt(sbi, XATTR_USER);
1343#endif
1344#ifdef CONFIG_F2FS_FS_POSIX_ACL
1345 set_opt(sbi, POSIX_ACL);
1346#endif
36dbd328
CY
1347
1348#ifdef CONFIG_F2FS_FAULT_INJECTION
1349 f2fs_build_fault_attr(sbi, 0);
1350#endif
498c5e9f
YH
1351}
1352
ea676733
JK
1353#ifdef CONFIG_QUOTA
1354static int f2fs_enable_quotas(struct super_block *sb);
1355#endif
696c018c
NJ
1356static int f2fs_remount(struct super_block *sb, int *flags, char *data)
1357{
1358 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1359 struct f2fs_mount_info org_mount_opt;
0abd675e 1360 unsigned long old_sb_flags;
696c018c 1361 int err, active_logs;
876dc59e
GZ
1362 bool need_restart_gc = false;
1363 bool need_stop_gc = false;
9cd81ce3 1364 bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
0cdd3195 1365 int old_whint_mode = sbi->whint_mode;
07939627 1366 int old_alloc_mode = sbi->alloc_mode;
93cf93f1 1367 int old_fsync_mode = sbi->fsync_mode;
eea52882
CY
1368 int old_inline_xattr_size = sbi->inline_xattr_size;
1369 block_t old_root_reserved_blocks = sbi->root_reserved_blocks;
1370 kuid_t old_resuid = sbi->s_resuid;
1371 kgid_t old_resgid = sbi->s_resgid;
1372 int old_write_io_size_bits = sbi->write_io_size_bits;
2443b8b3
CY
1373#ifdef CONFIG_F2FS_FAULT_INJECTION
1374 struct f2fs_fault_info ffi = sbi->fault_info;
1375#endif
4b2414d0
CY
1376#ifdef CONFIG_QUOTA
1377 int s_jquota_fmt;
1378 char *s_qf_names[MAXQUOTAS];
1379 int i, j;
1380#endif
696c018c
NJ
1381
1382 /*
1383 * Save the old mount options in case we
1384 * need to restore them.
1385 */
1386 org_mount_opt = sbi->mount_opt;
0abd675e 1387 old_sb_flags = sb->s_flags;
696c018c
NJ
1388 active_logs = sbi->active_logs;
1389
4b2414d0
CY
1390#ifdef CONFIG_QUOTA
1391 s_jquota_fmt = sbi->s_jquota_fmt;
1392 for (i = 0; i < MAXQUOTAS; i++) {
1393 if (sbi->s_qf_names[i]) {
1394 s_qf_names[i] = kstrdup(sbi->s_qf_names[i],
1395 GFP_KERNEL);
1396 if (!s_qf_names[i]) {
1397 for (j = 0; j < i; j++)
1398 kfree(s_qf_names[j]);
1399 return -ENOMEM;
1400 }
1401 } else {
1402 s_qf_names[i] = NULL;
1403 }
1404 }
1405#endif
1406
df728b0f 1407 /* recover superblocks we couldn't write due to previous RO mount */
1751e8a6 1408 if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
df728b0f
JK
1409 err = f2fs_commit_super(sbi, false);
1410 f2fs_msg(sb, KERN_INFO,
1411 "Try to recover all the superblocks, ret: %d", err);
1412 if (!err)
1413 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
1414 }
1415
498c5e9f 1416 default_options(sbi);
26666c8a 1417
696c018c
NJ
1418 /* parse mount options */
1419 err = parse_options(sb, data);
1420 if (err)
1421 goto restore_opts;
1422
1423 /*
1424 * Previous and new state of filesystem is RO,
876dc59e 1425 * so skip checking GC and FLUSH_MERGE conditions.
696c018c 1426 */
1751e8a6 1427 if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
696c018c
NJ
1428 goto skip;
1429
ea676733 1430#ifdef CONFIG_QUOTA
1751e8a6 1431 if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
0abd675e
CY
1432 err = dquot_suspend(sb, -1);
1433 if (err < 0)
1434 goto restore_opts;
6279398d 1435 } else if (f2fs_readonly(sb) && !(*flags & MS_RDONLY)) {
0abd675e 1436 /* dquot_resume needs RW */
1751e8a6 1437 sb->s_flags &= ~SB_RDONLY;
ea676733
JK
1438 if (sb_any_quota_suspended(sb)) {
1439 dquot_resume(sb, -1);
1440 } else if (f2fs_sb_has_quota_ino(sb)) {
1441 err = f2fs_enable_quotas(sb);
1442 if (err)
1443 goto restore_opts;
1444 }
0abd675e 1445 }
ea676733 1446#endif
9cd81ce3
CY
1447 /* disallow enable/disable extent_cache dynamically */
1448 if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
1449 err = -EINVAL;
1450 f2fs_msg(sbi->sb, KERN_WARNING,
1451 "switch extent_cache option is not allowed");
1452 goto restore_opts;
1453 }
1454
696c018c
NJ
1455 /*
1456 * We stop the GC thread if FS is mounted as RO
1457 * or if background_gc = off is passed in mount
1458 * option. Also sync the filesystem.
1459 */
1751e8a6 1460 if ((*flags & SB_RDONLY) || !test_opt(sbi, BG_GC)) {
696c018c
NJ
1461 if (sbi->gc_thread) {
1462 stop_gc_thread(sbi);
876dc59e 1463 need_restart_gc = true;
696c018c 1464 }
aba291b3 1465 } else if (!sbi->gc_thread) {
696c018c
NJ
1466 err = start_gc_thread(sbi);
1467 if (err)
1468 goto restore_opts;
876dc59e
GZ
1469 need_stop_gc = true;
1470 }
1471
0cdd3195 1472 if (*flags & SB_RDONLY || sbi->whint_mode != old_whint_mode) {
faa0e55b
JK
1473 writeback_inodes_sb(sb, WB_REASON_SYNC);
1474 sync_inodes_sb(sb);
1475
1476 set_sbi_flag(sbi, SBI_IS_DIRTY);
1477 set_sbi_flag(sbi, SBI_IS_CLOSE);
1478 f2fs_sync_fs(sb, 1);
1479 clear_sbi_flag(sbi, SBI_IS_CLOSE);
1480 }
1481
876dc59e
GZ
1482 /*
1483 * We stop issue flush thread if FS is mounted as RO
1484 * or if flush_merge is not passed in mount option.
1485 */
1751e8a6 1486 if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
5eba8c5d
JK
1487 clear_opt(sbi, FLUSH_MERGE);
1488 destroy_flush_cmd_control(sbi, false);
1489 } else {
2163d198
GZ
1490 err = create_flush_cmd_control(sbi);
1491 if (err)
a688b9d9 1492 goto restore_gc;
696c018c
NJ
1493 }
1494skip:
4b2414d0
CY
1495#ifdef CONFIG_QUOTA
1496 /* Release old quota file names */
1497 for (i = 0; i < MAXQUOTAS; i++)
1498 kfree(s_qf_names[i]);
1499#endif
696c018c 1500 /* Update the POSIXACL Flag */
1751e8a6
LT
1501 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
1502 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
df728b0f 1503
7e65be49 1504 limit_reserve_root(sbi);
696c018c 1505 return 0;
876dc59e
GZ
1506restore_gc:
1507 if (need_restart_gc) {
1508 if (start_gc_thread(sbi))
1509 f2fs_msg(sbi->sb, KERN_WARNING,
e1c42045 1510 "background gc thread has stopped");
876dc59e
GZ
1511 } else if (need_stop_gc) {
1512 stop_gc_thread(sbi);
1513 }
696c018c 1514restore_opts:
4b2414d0
CY
1515#ifdef CONFIG_QUOTA
1516 sbi->s_jquota_fmt = s_jquota_fmt;
1517 for (i = 0; i < MAXQUOTAS; i++) {
1518 kfree(sbi->s_qf_names[i]);
1519 sbi->s_qf_names[i] = s_qf_names[i];
1520 }
1521#endif
eea52882
CY
1522 sbi->write_io_size_bits = old_write_io_size_bits;
1523 sbi->s_resgid = old_resgid;
1524 sbi->s_resuid = old_resuid;
1525 sbi->root_reserved_blocks = old_root_reserved_blocks;
1526 sbi->inline_xattr_size = old_inline_xattr_size;
07939627 1527 sbi->alloc_mode = old_alloc_mode;
93cf93f1 1528 sbi->fsync_mode = old_fsync_mode;
0cdd3195 1529 sbi->whint_mode = old_whint_mode;
696c018c
NJ
1530 sbi->mount_opt = org_mount_opt;
1531 sbi->active_logs = active_logs;
0abd675e 1532 sb->s_flags = old_sb_flags;
2443b8b3
CY
1533#ifdef CONFIG_F2FS_FAULT_INJECTION
1534 sbi->fault_info = ffi;
1535#endif
696c018c
NJ
1536 return err;
1537}
1538
0abd675e
CY
1539#ifdef CONFIG_QUOTA
1540/* Read data from quotafile */
1541static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
1542 size_t len, loff_t off)
1543{
1544 struct inode *inode = sb_dqopt(sb)->files[type];
1545 struct address_space *mapping = inode->i_mapping;
1546 block_t blkidx = F2FS_BYTES_TO_BLK(off);
1547 int offset = off & (sb->s_blocksize - 1);
1548 int tocopy;
1549 size_t toread;
1550 loff_t i_size = i_size_read(inode);
1551 struct page *page;
1552 char *kaddr;
1553
1554 if (off > i_size)
1555 return 0;
1556
1557 if (off + len > i_size)
1558 len = i_size - off;
1559 toread = len;
1560 while (toread > 0) {
1561 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
1562repeat:
1563 page = read_mapping_page(mapping, blkidx, NULL);
4e46a023
JK
1564 if (IS_ERR(page)) {
1565 if (PTR_ERR(page) == -ENOMEM) {
1566 congestion_wait(BLK_RW_ASYNC, HZ/50);
1567 goto repeat;
1568 }
0abd675e 1569 return PTR_ERR(page);
4e46a023 1570 }
0abd675e
CY
1571
1572 lock_page(page);
1573
1574 if (unlikely(page->mapping != mapping)) {
1575 f2fs_put_page(page, 1);
1576 goto repeat;
1577 }
1578 if (unlikely(!PageUptodate(page))) {
1579 f2fs_put_page(page, 1);
1580 return -EIO;
1581 }
1582
1583 kaddr = kmap_atomic(page);
1584 memcpy(data, kaddr + offset, tocopy);
1585 kunmap_atomic(kaddr);
1586 f2fs_put_page(page, 1);
1587
1588 offset = 0;
1589 toread -= tocopy;
1590 data += tocopy;
1591 blkidx++;
1592 }
1593 return len;
1594}
1595
1596/* Write to quotafile */
1597static ssize_t f2fs_quota_write(struct super_block *sb, int type,
1598 const char *data, size_t len, loff_t off)
1599{
1600 struct inode *inode = sb_dqopt(sb)->files[type];
1601 struct address_space *mapping = inode->i_mapping;
1602 const struct address_space_operations *a_ops = mapping->a_ops;
1603 int offset = off & (sb->s_blocksize - 1);
1604 size_t towrite = len;
1605 struct page *page;
1606 char *kaddr;
1607 int err = 0;
1608 int tocopy;
1609
1610 while (towrite > 0) {
1611 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
1612 towrite);
4e46a023 1613retry:
0abd675e
CY
1614 err = a_ops->write_begin(NULL, mapping, off, tocopy, 0,
1615 &page, NULL);
4e46a023
JK
1616 if (unlikely(err)) {
1617 if (err == -ENOMEM) {
1618 congestion_wait(BLK_RW_ASYNC, HZ/50);
1619 goto retry;
1620 }
0abd675e 1621 break;
4e46a023 1622 }
0abd675e
CY
1623
1624 kaddr = kmap_atomic(page);
1625 memcpy(kaddr + offset, data, tocopy);
1626 kunmap_atomic(kaddr);
1627 flush_dcache_page(page);
1628
1629 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
1630 page, NULL);
1631 offset = 0;
1632 towrite -= tocopy;
1633 off += tocopy;
1634 data += tocopy;
1635 cond_resched();
1636 }
1637
1638 if (len == towrite)
6e5b5d41 1639 return err;
0abd675e
CY
1640 inode->i_mtime = inode->i_ctime = current_time(inode);
1641 f2fs_mark_inode_dirty_sync(inode, false);
1642 return len - towrite;
1643}
1644
1645static struct dquot **f2fs_get_dquots(struct inode *inode)
1646{
1647 return F2FS_I(inode)->i_dquot;
1648}
1649
1650static qsize_t *f2fs_get_reserved_space(struct inode *inode)
1651{
1652 return &F2FS_I(inode)->i_reserved_quota;
1653}
1654
4b2414d0
CY
1655static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
1656{
1657 return dquot_quota_on_mount(sbi->sb, sbi->s_qf_names[type],
1658 sbi->s_jquota_fmt, type);
1659}
1660
ea676733 1661int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
4b2414d0 1662{
ea676733
JK
1663 int enabled = 0;
1664 int i, err;
1665
1666 if (f2fs_sb_has_quota_ino(sbi->sb) && rdonly) {
1667 err = f2fs_enable_quotas(sbi->sb);
1668 if (err) {
1669 f2fs_msg(sbi->sb, KERN_ERR,
1670 "Cannot turn on quota_ino: %d", err);
1671 return 0;
1672 }
1673 return 1;
1674 }
4b2414d0
CY
1675
1676 for (i = 0; i < MAXQUOTAS; i++) {
1677 if (sbi->s_qf_names[i]) {
ea676733
JK
1678 err = f2fs_quota_on_mount(sbi, i);
1679 if (!err) {
1680 enabled = 1;
1681 continue;
1682 }
1683 f2fs_msg(sbi->sb, KERN_ERR,
1684 "Cannot turn on quotas: %d on %d", err, i);
4b2414d0
CY
1685 }
1686 }
ea676733
JK
1687 return enabled;
1688}
1689
1690static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
1691 unsigned int flags)
1692{
1693 struct inode *qf_inode;
1694 unsigned long qf_inum;
1695 int err;
1696
1697 BUG_ON(!f2fs_sb_has_quota_ino(sb));
1698
1699 qf_inum = f2fs_qf_ino(sb, type);
1700 if (!qf_inum)
1701 return -EPERM;
1702
1703 qf_inode = f2fs_iget(sb, qf_inum);
1704 if (IS_ERR(qf_inode)) {
1705 f2fs_msg(sb, KERN_ERR,
1706 "Bad quota inode %u:%lu", type, qf_inum);
1707 return PTR_ERR(qf_inode);
1708 }
1709
1710 /* Don't account quota for quota files to avoid recursion */
1711 qf_inode->i_flags |= S_NOQUOTA;
1712 err = dquot_enable(qf_inode, type, format_id, flags);
1713 iput(qf_inode);
1714 return err;
1715}
1716
1717static int f2fs_enable_quotas(struct super_block *sb)
1718{
1719 int type, err = 0;
1720 unsigned long qf_inum;
1721 bool quota_mopt[MAXQUOTAS] = {
1722 test_opt(F2FS_SB(sb), USRQUOTA),
1723 test_opt(F2FS_SB(sb), GRPQUOTA),
1724 test_opt(F2FS_SB(sb), PRJQUOTA),
1725 };
1726
1727 sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE | DQUOT_NOLIST_DIRTY;
1728 for (type = 0; type < MAXQUOTAS; type++) {
1729 qf_inum = f2fs_qf_ino(sb, type);
1730 if (qf_inum) {
1731 err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
1732 DQUOT_USAGE_ENABLED |
1733 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
1734 if (err) {
1735 f2fs_msg(sb, KERN_ERR,
1736 "Failed to enable quota tracking "
1737 "(type=%d, err=%d). Please run "
1738 "fsck to fix.", type, err);
1739 for (type--; type >= 0; type--)
1740 dquot_quota_off(sb, type);
1741 return err;
1742 }
4b2414d0
CY
1743 }
1744 }
ea676733 1745 return 0;
4b2414d0
CY
1746}
1747
0abd675e
CY
1748static int f2fs_quota_sync(struct super_block *sb, int type)
1749{
1750 struct quota_info *dqopt = sb_dqopt(sb);
1751 int cnt;
1752 int ret;
1753
1754 ret = dquot_writeback_dquots(sb, type);
1755 if (ret)
1756 return ret;
1757
1758 /*
1759 * Now when everything is written we can discard the pagecache so
1760 * that userspace sees the changes.
1761 */
1762 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
1763 if (type != -1 && cnt != type)
1764 continue;
1765 if (!sb_has_quota_active(sb, cnt))
1766 continue;
1767
1768 ret = filemap_write_and_wait(dqopt->files[cnt]->i_mapping);
1769 if (ret)
1770 return ret;
1771
1772 inode_lock(dqopt->files[cnt]);
1773 truncate_inode_pages(&dqopt->files[cnt]->i_data, 0);
1774 inode_unlock(dqopt->files[cnt]);
1775 }
1776 return 0;
1777}
1778
1779static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
1780 const struct path *path)
1781{
1782 struct inode *inode;
1783 int err;
1784
9a20d391 1785 err = f2fs_quota_sync(sb, type);
0abd675e
CY
1786 if (err)
1787 return err;
1788
1789 err = dquot_quota_on(sb, type, format_id, path);
1790 if (err)
1791 return err;
1792
1793 inode = d_inode(path->dentry);
1794
1795 inode_lock(inode);
1796 F2FS_I(inode)->i_flags |= FS_NOATIME_FL | FS_IMMUTABLE_FL;
1797 inode_set_flags(inode, S_NOATIME | S_IMMUTABLE,
1798 S_NOATIME | S_IMMUTABLE);
1799 inode_unlock(inode);
1800 f2fs_mark_inode_dirty_sync(inode, false);
1801
1802 return 0;
1803}
1804
1805static int f2fs_quota_off(struct super_block *sb, int type)
1806{
1807 struct inode *inode = sb_dqopt(sb)->files[type];
1808 int err;
1809
1810 if (!inode || !igrab(inode))
1811 return dquot_quota_off(sb, type);
1812
9a20d391 1813 f2fs_quota_sync(sb, type);
0abd675e
CY
1814
1815 err = dquot_quota_off(sb, type);
ea676733 1816 if (err || f2fs_sb_has_quota_ino(sb))
0abd675e
CY
1817 goto out_put;
1818
1819 inode_lock(inode);
1820 F2FS_I(inode)->i_flags &= ~(FS_NOATIME_FL | FS_IMMUTABLE_FL);
1821 inode_set_flags(inode, 0, S_NOATIME | S_IMMUTABLE);
1822 inode_unlock(inode);
1823 f2fs_mark_inode_dirty_sync(inode, false);
1824out_put:
1825 iput(inode);
1826 return err;
1827}
1828
4b2414d0 1829void f2fs_quota_off_umount(struct super_block *sb)
0abd675e
CY
1830{
1831 int type;
1832
1833 for (type = 0; type < MAXQUOTAS; type++)
1834 f2fs_quota_off(sb, type);
1835}
1836
94b1e10e 1837static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
5c57132e
CY
1838{
1839 *projid = F2FS_I(inode)->i_projid;
1840 return 0;
1841}
1842
0abd675e
CY
1843static const struct dquot_operations f2fs_quota_operations = {
1844 .get_reserved_space = f2fs_get_reserved_space,
1845 .write_dquot = dquot_commit,
1846 .acquire_dquot = dquot_acquire,
1847 .release_dquot = dquot_release,
1848 .mark_dirty = dquot_mark_dquot_dirty,
1849 .write_info = dquot_commit_info,
1850 .alloc_dquot = dquot_alloc,
1851 .destroy_dquot = dquot_destroy,
5c57132e 1852 .get_projid = f2fs_get_projid,
0abd675e
CY
1853 .get_next_id = dquot_get_next_id,
1854};
1855
1856static const struct quotactl_ops f2fs_quotactl_ops = {
1857 .quota_on = f2fs_quota_on,
1858 .quota_off = f2fs_quota_off,
1859 .quota_sync = f2fs_quota_sync,
1860 .get_state = dquot_get_state,
1861 .set_info = dquot_set_dqinfo,
1862 .get_dqblk = dquot_get_dqblk,
1863 .set_dqblk = dquot_set_dqblk,
1864 .get_nextdqblk = dquot_get_next_dqblk,
1865};
1866#else
4b2414d0 1867void f2fs_quota_off_umount(struct super_block *sb)
0abd675e
CY
1868{
1869}
1870#endif
1871
f62fc9f9 1872static const struct super_operations f2fs_sops = {
aff063e2 1873 .alloc_inode = f2fs_alloc_inode,
531ad7d5 1874 .drop_inode = f2fs_drop_inode,
aff063e2
JK
1875 .destroy_inode = f2fs_destroy_inode,
1876 .write_inode = f2fs_write_inode,
b3783873 1877 .dirty_inode = f2fs_dirty_inode,
aff063e2 1878 .show_options = f2fs_show_options,
0abd675e
CY
1879#ifdef CONFIG_QUOTA
1880 .quota_read = f2fs_quota_read,
1881 .quota_write = f2fs_quota_write,
1882 .get_dquots = f2fs_get_dquots,
1883#endif
aff063e2
JK
1884 .evict_inode = f2fs_evict_inode,
1885 .put_super = f2fs_put_super,
1886 .sync_fs = f2fs_sync_fs,
d6212a5f
CL
1887 .freeze_fs = f2fs_freeze,
1888 .unfreeze_fs = f2fs_unfreeze,
aff063e2 1889 .statfs = f2fs_statfs,
696c018c 1890 .remount_fs = f2fs_remount,
aff063e2
JK
1891};
1892
0b81d077
JK
1893#ifdef CONFIG_F2FS_FS_ENCRYPTION
1894static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
1895{
1896 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
1897 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
1898 ctx, len, NULL);
1899}
1900
1901static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
1902 void *fs_data)
1903{
1904 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
1905 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
1906 ctx, len, fs_data, XATTR_CREATE);
1907}
1908
1909static unsigned f2fs_max_namelen(struct inode *inode)
1910{
1911 return S_ISLNK(inode->i_mode) ?
1912 inode->i_sb->s_blocksize : F2FS_NAME_LEN;
1913}
1914
6f69f0ed 1915static const struct fscrypt_operations f2fs_cryptops = {
a5d431ef 1916 .key_prefix = "f2fs:",
0b81d077
JK
1917 .get_context = f2fs_get_context,
1918 .set_context = f2fs_set_context,
0b81d077
JK
1919 .empty_dir = f2fs_empty_dir,
1920 .max_namelen = f2fs_max_namelen,
1921};
0b81d077
JK
1922#endif
1923
aff063e2
JK
1924static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
1925 u64 ino, u32 generation)
1926{
1927 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1928 struct inode *inode;
1929
d6b7d4b3 1930 if (check_nid_range(sbi, ino))
910bb12d 1931 return ERR_PTR(-ESTALE);
aff063e2
JK
1932
1933 /*
1934 * f2fs_iget isn't quite right if the inode is currently unallocated!
1935 * However f2fs_iget currently does appropriate checks to handle stale
1936 * inodes so everything is OK.
1937 */
1938 inode = f2fs_iget(sb, ino);
1939 if (IS_ERR(inode))
1940 return ERR_CAST(inode);
6bacf52f 1941 if (unlikely(generation && inode->i_generation != generation)) {
aff063e2
JK
1942 /* we didn't find the right inode.. */
1943 iput(inode);
1944 return ERR_PTR(-ESTALE);
1945 }
1946 return inode;
1947}
1948
1949static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
1950 int fh_len, int fh_type)
1951{
1952 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
1953 f2fs_nfs_get_inode);
1954}
1955
1956static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
1957 int fh_len, int fh_type)
1958{
1959 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
1960 f2fs_nfs_get_inode);
1961}
1962
1963static const struct export_operations f2fs_export_ops = {
1964 .fh_to_dentry = f2fs_fh_to_dentry,
1965 .fh_to_parent = f2fs_fh_to_parent,
1966 .get_parent = f2fs_get_parent,
1967};
1968
e0afc4d6 1969static loff_t max_file_blocks(void)
aff063e2 1970{
7a2af766 1971 loff_t result = 0;
aff063e2
JK
1972 loff_t leaf_count = ADDRS_PER_BLOCK;
1973
7a2af766
CY
1974 /*
1975 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
6afc662e 1976 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
7a2af766
CY
1977 * space in inode.i_addr, it will be more safe to reassign
1978 * result as zero.
1979 */
1980
aff063e2
JK
1981 /* two direct node blocks */
1982 result += (leaf_count * 2);
1983
1984 /* two indirect node blocks */
1985 leaf_count *= NIDS_PER_BLOCK;
1986 result += (leaf_count * 2);
1987
1988 /* one double indirect node block */
1989 leaf_count *= NIDS_PER_BLOCK;
1990 result += leaf_count;
1991
aff063e2
JK
1992 return result;
1993}
1994
fd694733
JK
1995static int __f2fs_commit_super(struct buffer_head *bh,
1996 struct f2fs_super_block *super)
1997{
1998 lock_buffer(bh);
1999 if (super)
2000 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
fd694733
JK
2001 set_buffer_dirty(bh);
2002 unlock_buffer(bh);
2003
2004 /* it's rare case, we can do fua all the time */
3adc5fcb 2005 return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
fd694733
JK
2006}
2007
df728b0f 2008static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
fd694733 2009 struct buffer_head *bh)
9a59b62f 2010{
fd694733
JK
2011 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2012 (bh->b_data + F2FS_SUPER_OFFSET);
df728b0f 2013 struct super_block *sb = sbi->sb;
9a59b62f
CY
2014 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
2015 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
2016 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
2017 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
2018 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
2019 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
2020 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
2021 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
2022 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
2023 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
2024 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2025 u32 segment_count = le32_to_cpu(raw_super->segment_count);
2026 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
fd694733
JK
2027 u64 main_end_blkaddr = main_blkaddr +
2028 (segment_count_main << log_blocks_per_seg);
2029 u64 seg_end_blkaddr = segment0_blkaddr +
2030 (segment_count << log_blocks_per_seg);
9a59b62f
CY
2031
2032 if (segment0_blkaddr != cp_blkaddr) {
2033 f2fs_msg(sb, KERN_INFO,
2034 "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
2035 segment0_blkaddr, cp_blkaddr);
2036 return true;
2037 }
2038
2039 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
2040 sit_blkaddr) {
2041 f2fs_msg(sb, KERN_INFO,
2042 "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
2043 cp_blkaddr, sit_blkaddr,
2044 segment_count_ckpt << log_blocks_per_seg);
2045 return true;
2046 }
2047
2048 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
2049 nat_blkaddr) {
2050 f2fs_msg(sb, KERN_INFO,
2051 "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
2052 sit_blkaddr, nat_blkaddr,
2053 segment_count_sit << log_blocks_per_seg);
2054 return true;
2055 }
2056
2057 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
2058 ssa_blkaddr) {
2059 f2fs_msg(sb, KERN_INFO,
2060 "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
2061 nat_blkaddr, ssa_blkaddr,
2062 segment_count_nat << log_blocks_per_seg);
2063 return true;
2064 }
2065
2066 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
2067 main_blkaddr) {
2068 f2fs_msg(sb, KERN_INFO,
2069 "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
2070 ssa_blkaddr, main_blkaddr,
2071 segment_count_ssa << log_blocks_per_seg);
2072 return true;
2073 }
2074
fd694733 2075 if (main_end_blkaddr > seg_end_blkaddr) {
9a59b62f 2076 f2fs_msg(sb, KERN_INFO,
fd694733 2077 "Wrong MAIN_AREA boundary, start(%u) end(%u) block(%u)",
9a59b62f 2078 main_blkaddr,
fd694733
JK
2079 segment0_blkaddr +
2080 (segment_count << log_blocks_per_seg),
9a59b62f
CY
2081 segment_count_main << log_blocks_per_seg);
2082 return true;
fd694733
JK
2083 } else if (main_end_blkaddr < seg_end_blkaddr) {
2084 int err = 0;
2085 char *res;
2086
2087 /* fix in-memory information all the time */
2088 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
2089 segment0_blkaddr) >> log_blocks_per_seg);
2090
2091 if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
df728b0f 2092 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
fd694733
JK
2093 res = "internally";
2094 } else {
2095 err = __f2fs_commit_super(bh, NULL);
2096 res = err ? "failed" : "done";
2097 }
2098 f2fs_msg(sb, KERN_INFO,
2099 "Fix alignment : %s, start(%u) end(%u) block(%u)",
2100 res, main_blkaddr,
2101 segment0_blkaddr +
2102 (segment_count << log_blocks_per_seg),
2103 segment_count_main << log_blocks_per_seg);
2104 if (err)
2105 return true;
9a59b62f 2106 }
9a59b62f
CY
2107 return false;
2108}
2109
df728b0f 2110static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
fd694733 2111 struct buffer_head *bh)
aff063e2 2112{
fd694733
JK
2113 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2114 (bh->b_data + F2FS_SUPER_OFFSET);
df728b0f 2115 struct super_block *sb = sbi->sb;
aff063e2
JK
2116 unsigned int blocksize;
2117
a07ef784
NJ
2118 if (F2FS_SUPER_MAGIC != le32_to_cpu(raw_super->magic)) {
2119 f2fs_msg(sb, KERN_INFO,
2120 "Magic Mismatch, valid(0x%x) - read(0x%x)",
2121 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
aff063e2 2122 return 1;
a07ef784 2123 }
aff063e2 2124
5c9b4692 2125 /* Currently, support only 4KB page cache size */
09cbfeaf 2126 if (F2FS_BLKSIZE != PAGE_SIZE) {
5c9b4692 2127 f2fs_msg(sb, KERN_INFO,
14d7e9de 2128 "Invalid page_cache_size (%lu), supports only 4KB\n",
09cbfeaf 2129 PAGE_SIZE);
5c9b4692 2130 return 1;
2131 }
2132
aff063e2
JK
2133 /* Currently, support only 4KB block size */
2134 blocksize = 1 << le32_to_cpu(raw_super->log_blocksize);
5c9b4692 2135 if (blocksize != F2FS_BLKSIZE) {
a07ef784
NJ
2136 f2fs_msg(sb, KERN_INFO,
2137 "Invalid blocksize (%u), supports only 4KB\n",
2138 blocksize);
aff063e2 2139 return 1;
a07ef784 2140 }
5c9b4692 2141
9a59b62f
CY
2142 /* check log blocks per segment */
2143 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
2144 f2fs_msg(sb, KERN_INFO,
2145 "Invalid log blocks per segment (%u)\n",
2146 le32_to_cpu(raw_super->log_blocks_per_seg));
2147 return 1;
2148 }
2149
55cf9cb6
CY
2150 /* Currently, support 512/1024/2048/4096 bytes sector size */
2151 if (le32_to_cpu(raw_super->log_sectorsize) >
2152 F2FS_MAX_LOG_SECTOR_SIZE ||
2153 le32_to_cpu(raw_super->log_sectorsize) <
2154 F2FS_MIN_LOG_SECTOR_SIZE) {
2155 f2fs_msg(sb, KERN_INFO, "Invalid log sectorsize (%u)",
2156 le32_to_cpu(raw_super->log_sectorsize));
aff063e2 2157 return 1;
a07ef784 2158 }
55cf9cb6
CY
2159 if (le32_to_cpu(raw_super->log_sectors_per_block) +
2160 le32_to_cpu(raw_super->log_sectorsize) !=
2161 F2FS_MAX_LOG_SECTOR_SIZE) {
2162 f2fs_msg(sb, KERN_INFO,
2163 "Invalid log sectors per block(%u) log sectorsize(%u)",
2164 le32_to_cpu(raw_super->log_sectors_per_block),
2165 le32_to_cpu(raw_super->log_sectorsize));
aff063e2 2166 return 1;
a07ef784 2167 }
9a59b62f
CY
2168
2169 /* check reserved ino info */
2170 if (le32_to_cpu(raw_super->node_ino) != 1 ||
2171 le32_to_cpu(raw_super->meta_ino) != 2 ||
2172 le32_to_cpu(raw_super->root_ino) != 3) {
2173 f2fs_msg(sb, KERN_INFO,
2174 "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
2175 le32_to_cpu(raw_super->node_ino),
2176 le32_to_cpu(raw_super->meta_ino),
2177 le32_to_cpu(raw_super->root_ino));
2178 return 1;
2179 }
2180
b9dd4618
JQ
2181 if (le32_to_cpu(raw_super->segment_count) > F2FS_MAX_SEGMENT) {
2182 f2fs_msg(sb, KERN_INFO,
2183 "Invalid segment count (%u)",
2184 le32_to_cpu(raw_super->segment_count));
2185 return 1;
2186 }
2187
9a59b62f 2188 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
df728b0f 2189 if (sanity_check_area_boundary(sbi, bh))
9a59b62f
CY
2190 return 1;
2191
aff063e2
JK
2192 return 0;
2193}
2194
984ec63c 2195int sanity_check_ckpt(struct f2fs_sb_info *sbi)
aff063e2
JK
2196{
2197 unsigned int total, fsmeta;
577e3495
JK
2198 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
2199 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2040fce8 2200 unsigned int ovp_segments, reserved_segments;
15d3042a
JQ
2201 unsigned int main_segs, blocks_per_seg;
2202 int i;
aff063e2
JK
2203
2204 total = le32_to_cpu(raw_super->segment_count);
2205 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
2206 fsmeta += le32_to_cpu(raw_super->segment_count_sit);
2207 fsmeta += le32_to_cpu(raw_super->segment_count_nat);
2208 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
2209 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
2210
6bacf52f 2211 if (unlikely(fsmeta >= total))
aff063e2 2212 return 1;
577e3495 2213
2040fce8
JK
2214 ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
2215 reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
2216
2217 if (unlikely(fsmeta < F2FS_MIN_SEGMENTS ||
2218 ovp_segments == 0 || reserved_segments == 0)) {
2219 f2fs_msg(sbi->sb, KERN_ERR,
2220 "Wrong layout: check mkfs.f2fs version");
2221 return 1;
2222 }
2223
15d3042a
JQ
2224 main_segs = le32_to_cpu(raw_super->segment_count_main);
2225 blocks_per_seg = sbi->blocks_per_seg;
2226
2227 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
2228 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
2229 le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
2230 return 1;
2231 }
2232 for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
2233 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
2234 le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
2235 return 1;
2236 }
2237
1e968fdf 2238 if (unlikely(f2fs_cp_error(sbi))) {
577e3495
JK
2239 f2fs_msg(sbi->sb, KERN_ERR, "A bug case: need to run fsck");
2240 return 1;
2241 }
aff063e2
JK
2242 return 0;
2243}
2244
2245static void init_sb_info(struct f2fs_sb_info *sbi)
2246{
2247 struct f2fs_super_block *raw_super = sbi->raw_super;
e41e6d75 2248 int i, j;
aff063e2
JK
2249
2250 sbi->log_sectors_per_block =
2251 le32_to_cpu(raw_super->log_sectors_per_block);
2252 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
2253 sbi->blocksize = 1 << sbi->log_blocksize;
2254 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2255 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
2256 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
2257 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
2258 sbi->total_sections = le32_to_cpu(raw_super->section_count);
2259 sbi->total_node_count =
2260 (le32_to_cpu(raw_super->segment_count_nat) / 2)
2261 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
2262 sbi->root_ino_num = le32_to_cpu(raw_super->root_ino);
2263 sbi->node_ino_num = le32_to_cpu(raw_super->node_ino);
2264 sbi->meta_ino_num = le32_to_cpu(raw_super->meta_ino);
5ec4e49f 2265 sbi->cur_victim_sec = NULL_SECNO;
b1c57c1c 2266 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
aff063e2 2267
ab9fa662 2268 sbi->dir_level = DEF_DIR_LEVEL;
6beceb54 2269 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
d0239e1b 2270 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
caf0047e 2271 clear_sbi_flag(sbi, SBI_NEED_FSCK);
2658e50d 2272
35782b23
JK
2273 for (i = 0; i < NR_COUNT_TYPE; i++)
2274 atomic_set(&sbi->nr_pages[i], 0);
2275
687de7f1
JK
2276 atomic_set(&sbi->wb_sync_req, 0);
2277
2658e50d
JK
2278 INIT_LIST_HEAD(&sbi->s_list);
2279 mutex_init(&sbi->umount_mutex);
e41e6d75
CY
2280 for (i = 0; i < NR_PAGE_TYPE - 1; i++)
2281 for (j = HOT; j < NR_TEMP_TYPE; j++)
2282 mutex_init(&sbi->wio_mutex[i][j]);
aaec2b1d 2283 spin_lock_init(&sbi->cp_lock);
1228b482
CY
2284
2285 sbi->dirty_device = 0;
2286 spin_lock_init(&sbi->dev_lock);
d0d3f1b3 2287
846ae671 2288 init_rwsem(&sbi->sb_lock);
aff063e2
JK
2289}
2290
523be8a6
JK
2291static int init_percpu_info(struct f2fs_sb_info *sbi)
2292{
35782b23 2293 int err;
41382ec4 2294
513c5f37
JK
2295 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
2296 if (err)
2297 return err;
2298
2299 return percpu_counter_init(&sbi->total_valid_inode_count, 0,
41382ec4 2300 GFP_KERNEL);
523be8a6
JK
2301}
2302
178053e2 2303#ifdef CONFIG_BLK_DEV_ZONED
3c62be17 2304static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
178053e2 2305{
3c62be17 2306 struct block_device *bdev = FDEV(devi).bdev;
178053e2
DLM
2307 sector_t nr_sectors = bdev->bd_part->nr_sects;
2308 sector_t sector = 0;
2309 struct blk_zone *zones;
2310 unsigned int i, nr_zones;
2311 unsigned int n = 0;
2312 int err = -EIO;
2313
ccd31cb2 2314 if (!f2fs_sb_has_blkzoned(sbi->sb))
178053e2
DLM
2315 return 0;
2316
3c62be17 2317 if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
f99e8648 2318 SECTOR_TO_BLOCK(bdev_zone_sectors(bdev)))
3c62be17 2319 return -EINVAL;
f99e8648 2320 sbi->blocks_per_blkz = SECTOR_TO_BLOCK(bdev_zone_sectors(bdev));
3c62be17
JK
2321 if (sbi->log_blocks_per_blkz && sbi->log_blocks_per_blkz !=
2322 __ilog2_u32(sbi->blocks_per_blkz))
2323 return -EINVAL;
178053e2 2324 sbi->log_blocks_per_blkz = __ilog2_u32(sbi->blocks_per_blkz);
3c62be17
JK
2325 FDEV(devi).nr_blkz = SECTOR_TO_BLOCK(nr_sectors) >>
2326 sbi->log_blocks_per_blkz;
f99e8648 2327 if (nr_sectors & (bdev_zone_sectors(bdev) - 1))
3c62be17 2328 FDEV(devi).nr_blkz++;
178053e2 2329
4e6aad29
CY
2330 FDEV(devi).blkz_type = f2fs_kmalloc(sbi, FDEV(devi).nr_blkz,
2331 GFP_KERNEL);
3c62be17 2332 if (!FDEV(devi).blkz_type)
178053e2
DLM
2333 return -ENOMEM;
2334
2335#define F2FS_REPORT_NR_ZONES 4096
2336
4e6aad29
CY
2337 zones = f2fs_kzalloc(sbi, sizeof(struct blk_zone) *
2338 F2FS_REPORT_NR_ZONES, GFP_KERNEL);
178053e2
DLM
2339 if (!zones)
2340 return -ENOMEM;
2341
2342 /* Get block zones type */
2343 while (zones && sector < nr_sectors) {
2344
2345 nr_zones = F2FS_REPORT_NR_ZONES;
2346 err = blkdev_report_zones(bdev, sector,
2347 zones, &nr_zones,
2348 GFP_KERNEL);
2349 if (err)
2350 break;
2351 if (!nr_zones) {
2352 err = -EIO;
2353 break;
2354 }
2355
2356 for (i = 0; i < nr_zones; i++) {
3c62be17 2357 FDEV(devi).blkz_type[n] = zones[i].type;
178053e2
DLM
2358 sector += zones[i].len;
2359 n++;
2360 }
2361 }
2362
2363 kfree(zones);
2364
2365 return err;
2366}
2367#endif
2368
9076a75f
GZ
2369/*
2370 * Read f2fs raw super block.
2b39e907
SL
2371 * Because we have two copies of super block, so read both of them
2372 * to get the first valid one. If any one of them is broken, we pass
2373 * them recovery flag back to the caller.
9076a75f 2374 */
df728b0f 2375static int read_raw_super_block(struct f2fs_sb_info *sbi,
9076a75f 2376 struct f2fs_super_block **raw_super,
e8240f65 2377 int *valid_super_block, int *recovery)
14d7e9de 2378{
df728b0f 2379 struct super_block *sb = sbi->sb;
2b39e907 2380 int block;
e8240f65 2381 struct buffer_head *bh;
fd694733 2382 struct f2fs_super_block *super;
da554e48 2383 int err = 0;
14d7e9de 2384
b39f0de2
YH
2385 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
2386 if (!super)
2387 return -ENOMEM;
2b39e907
SL
2388
2389 for (block = 0; block < 2; block++) {
2390 bh = sb_bread(sb, block);
2391 if (!bh) {
2392 f2fs_msg(sb, KERN_ERR, "Unable to read %dth superblock",
9076a75f 2393 block + 1);
2b39e907
SL
2394 err = -EIO;
2395 continue;
2396 }
14d7e9de 2397
2b39e907 2398 /* sanity checking of raw super */
df728b0f 2399 if (sanity_check_raw_super(sbi, bh)) {
2b39e907
SL
2400 f2fs_msg(sb, KERN_ERR,
2401 "Can't find valid F2FS filesystem in %dth superblock",
2402 block + 1);
2403 err = -EINVAL;
2404 brelse(bh);
2405 continue;
2406 }
14d7e9de 2407
2b39e907 2408 if (!*raw_super) {
fd694733
JK
2409 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
2410 sizeof(*super));
2b39e907
SL
2411 *valid_super_block = block;
2412 *raw_super = super;
2413 }
2414 brelse(bh);
da554e48 2415 }
2416
2b39e907
SL
2417 /* Fail to read any one of the superblocks*/
2418 if (err < 0)
2419 *recovery = 1;
da554e48 2420
da554e48 2421 /* No valid superblock */
2b39e907 2422 if (!*raw_super)
b39f0de2 2423 kfree(super);
2b39e907
SL
2424 else
2425 err = 0;
da554e48 2426
2b39e907 2427 return err;
14d7e9de 2428}
2429
fd694733 2430int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
26d815ad 2431{
5d909cdb 2432 struct buffer_head *bh;
26d815ad
JK
2433 int err;
2434
df728b0f
JK
2435 if ((recover && f2fs_readonly(sbi->sb)) ||
2436 bdev_read_only(sbi->sb->s_bdev)) {
2437 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
f2353d7b 2438 return -EROFS;
df728b0f 2439 }
f2353d7b 2440
fd694733 2441 /* write back-up superblock first */
0964fc1a 2442 bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
5d909cdb
JK
2443 if (!bh)
2444 return -EIO;
fd694733 2445 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
5d909cdb 2446 brelse(bh);
c5bda1c8
CY
2447
2448 /* if we are in recovery path, skip writing valid superblock */
2449 if (recover || err)
5d909cdb 2450 return err;
26d815ad
JK
2451
2452 /* write current valid superblock */
0964fc1a 2453 bh = sb_bread(sbi->sb, sbi->valid_super_block);
fd694733
JK
2454 if (!bh)
2455 return -EIO;
2456 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
2457 brelse(bh);
2458 return err;
26d815ad
JK
2459}
2460
3c62be17
JK
2461static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
2462{
2463 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
7bb3a371 2464 unsigned int max_devices = MAX_DEVICES;
3c62be17
JK
2465 int i;
2466
7bb3a371
MS
2467 /* Initialize single device information */
2468 if (!RDEV(0).path[0]) {
2469 if (!bdev_is_zoned(sbi->sb->s_bdev))
3c62be17 2470 return 0;
7bb3a371
MS
2471 max_devices = 1;
2472 }
3c62be17 2473
7bb3a371
MS
2474 /*
2475 * Initialize multiple devices information, or single
2476 * zoned block device information.
2477 */
4e6aad29
CY
2478 sbi->devs = f2fs_kzalloc(sbi, sizeof(struct f2fs_dev_info) *
2479 max_devices, GFP_KERNEL);
7bb3a371
MS
2480 if (!sbi->devs)
2481 return -ENOMEM;
3c62be17 2482
7bb3a371 2483 for (i = 0; i < max_devices; i++) {
3c62be17 2484
7bb3a371
MS
2485 if (i > 0 && !RDEV(i).path[0])
2486 break;
2487
2488 if (max_devices == 1) {
2489 /* Single zoned block device mount */
2490 FDEV(0).bdev =
2491 blkdev_get_by_dev(sbi->sb->s_bdev->bd_dev,
3c62be17 2492 sbi->sb->s_mode, sbi->sb->s_type);
7bb3a371
MS
2493 } else {
2494 /* Multi-device mount */
2495 memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
2496 FDEV(i).total_segments =
2497 le32_to_cpu(RDEV(i).total_segments);
2498 if (i == 0) {
2499 FDEV(i).start_blk = 0;
2500 FDEV(i).end_blk = FDEV(i).start_blk +
2501 (FDEV(i).total_segments <<
2502 sbi->log_blocks_per_seg) - 1 +
2503 le32_to_cpu(raw_super->segment0_blkaddr);
2504 } else {
2505 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
2506 FDEV(i).end_blk = FDEV(i).start_blk +
2507 (FDEV(i).total_segments <<
2508 sbi->log_blocks_per_seg) - 1;
2509 }
2510 FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
3c62be17 2511 sbi->sb->s_mode, sbi->sb->s_type);
7bb3a371 2512 }
3c62be17
JK
2513 if (IS_ERR(FDEV(i).bdev))
2514 return PTR_ERR(FDEV(i).bdev);
2515
2516 /* to release errored devices */
2517 sbi->s_ndevs = i + 1;
2518
2519#ifdef CONFIG_BLK_DEV_ZONED
2520 if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
ccd31cb2 2521 !f2fs_sb_has_blkzoned(sbi->sb)) {
3c62be17
JK
2522 f2fs_msg(sbi->sb, KERN_ERR,
2523 "Zoned block device feature not enabled\n");
2524 return -EINVAL;
2525 }
2526 if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
2527 if (init_blkz_info(sbi, i)) {
2528 f2fs_msg(sbi->sb, KERN_ERR,
2529 "Failed to initialize F2FS blkzone information");
2530 return -EINVAL;
2531 }
7bb3a371
MS
2532 if (max_devices == 1)
2533 break;
3c62be17
JK
2534 f2fs_msg(sbi->sb, KERN_INFO,
2535 "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
2536 i, FDEV(i).path,
2537 FDEV(i).total_segments,
2538 FDEV(i).start_blk, FDEV(i).end_blk,
2539 bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
2540 "Host-aware" : "Host-managed");
2541 continue;
2542 }
2543#endif
2544 f2fs_msg(sbi->sb, KERN_INFO,
2545 "Mount Device [%2d]: %20s, %8u, %8x - %8x",
2546 i, FDEV(i).path,
2547 FDEV(i).total_segments,
2548 FDEV(i).start_blk, FDEV(i).end_blk);
2549 }
0a595eba
JK
2550 f2fs_msg(sbi->sb, KERN_INFO,
2551 "IO Block Size: %8d KB", F2FS_IO_SIZE_KB(sbi));
3c62be17
JK
2552 return 0;
2553}
2554
84b89e5d
JK
2555static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
2556{
2557 struct f2fs_sm_info *sm_i = SM_I(sbi);
2558
2559 /* adjust parameters according to the volume size */
2560 if (sm_i->main_segments <= SMALL_VOLUME_SEGMENTS) {
2561 sbi->alloc_mode = ALLOC_MODE_REUSE;
2562 sm_i->dcc_info->discard_granularity = 1;
2563 sm_i->ipu_policy = 1 << F2FS_IPU_FORCE;
2564 }
2565}
2566
aff063e2
JK
2567static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
2568{
2569 struct f2fs_sb_info *sbi;
da554e48 2570 struct f2fs_super_block *raw_super;
aff063e2 2571 struct inode *root;
99e3e858 2572 int err;
2adc3505 2573 bool retry = true, need_fsck = false;
dabc4a5c 2574 char *options = NULL;
e8240f65 2575 int recovery, i, valid_super_block;
8f1dbbbb 2576 struct curseg_info *seg_i;
aff063e2 2577
ed2e621a 2578try_onemore:
da554e48 2579 err = -EINVAL;
2580 raw_super = NULL;
e8240f65 2581 valid_super_block = -1;
da554e48 2582 recovery = 0;
2583
aff063e2
JK
2584 /* allocate memory for f2fs-specific super block info */
2585 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
2586 if (!sbi)
2587 return -ENOMEM;
2588
df728b0f
JK
2589 sbi->sb = sb;
2590
43b6573b
KM
2591 /* Load the checksum driver */
2592 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
2593 if (IS_ERR(sbi->s_chksum_driver)) {
2594 f2fs_msg(sb, KERN_ERR, "Cannot load crc32 driver.");
2595 err = PTR_ERR(sbi->s_chksum_driver);
2596 sbi->s_chksum_driver = NULL;
2597 goto free_sbi;
2598 }
2599
ff9234ad 2600 /* set a block size */
6bacf52f 2601 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
a07ef784 2602 f2fs_msg(sb, KERN_ERR, "unable to set blocksize");
aff063e2 2603 goto free_sbi;
a07ef784 2604 }
aff063e2 2605
df728b0f 2606 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
e8240f65 2607 &recovery);
9076a75f
GZ
2608 if (err)
2609 goto free_sbi;
2610
5fb08372 2611 sb->s_fs_info = sbi;
52763a4b
JK
2612 sbi->raw_super = raw_super;
2613
7c2e5963
JK
2614 sbi->s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
2615 sbi->s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
2616
704956ec
CY
2617 /* precompute checksum seed for metadata */
2618 if (f2fs_sb_has_inode_chksum(sb))
2619 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
2620 sizeof(raw_super->uuid));
2621
d1b959c8
DLM
2622 /*
2623 * The BLKZONED feature indicates that the drive was formatted with
2624 * zone alignment optimization. This is optional for host-aware
2625 * devices, but mandatory for host-managed zoned block devices.
2626 */
2627#ifndef CONFIG_BLK_DEV_ZONED
ccd31cb2 2628 if (f2fs_sb_has_blkzoned(sb)) {
d1b959c8
DLM
2629 f2fs_msg(sb, KERN_ERR,
2630 "Zoned block device support is not enabled\n");
1727f317 2631 err = -EOPNOTSUPP;
d1b959c8
DLM
2632 goto free_sb_buf;
2633 }
d1b959c8 2634#endif
498c5e9f 2635 default_options(sbi);
aff063e2 2636 /* parse mount options */
dabc4a5c
JK
2637 options = kstrdup((const char *)data, GFP_KERNEL);
2638 if (data && !options) {
2639 err = -ENOMEM;
aff063e2 2640 goto free_sb_buf;
dabc4a5c
JK
2641 }
2642
2643 err = parse_options(sb, options);
2644 if (err)
2645 goto free_options;
aff063e2 2646
e0afc4d6
CY
2647 sbi->max_file_blocks = max_file_blocks();
2648 sb->s_maxbytes = sbi->max_file_blocks <<
2649 le32_to_cpu(raw_super->log_blocksize);
aff063e2
JK
2650 sb->s_max_links = F2FS_LINK_MAX;
2651 get_random_bytes(&sbi->s_next_generation, sizeof(u32));
2652
0abd675e
CY
2653#ifdef CONFIG_QUOTA
2654 sb->dq_op = &f2fs_quota_operations;
ea676733
JK
2655 if (f2fs_sb_has_quota_ino(sb))
2656 sb->s_qcop = &dquot_quotactl_sysfile_ops;
2657 else
2658 sb->s_qcop = &f2fs_quotactl_ops;
5c57132e 2659 sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
292c196a
CY
2660
2661 if (f2fs_sb_has_quota_ino(sbi->sb)) {
2662 for (i = 0; i < MAXQUOTAS; i++) {
2663 if (f2fs_qf_ino(sbi->sb, i))
2664 sbi->nquota_files++;
2665 }
2666 }
0abd675e
CY
2667#endif
2668
aff063e2 2669 sb->s_op = &f2fs_sops;
ffcc4182 2670#ifdef CONFIG_F2FS_FS_ENCRYPTION
0b81d077 2671 sb->s_cop = &f2fs_cryptops;
ffcc4182 2672#endif
aff063e2
JK
2673 sb->s_xattr = f2fs_xattr_handlers;
2674 sb->s_export_op = &f2fs_export_ops;
2675 sb->s_magic = F2FS_SUPER_MAGIC;
aff063e2 2676 sb->s_time_gran = 1;
1751e8a6
LT
2677 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
2678 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
85787090 2679 memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
578c6478 2680 sb->s_iflags |= SB_I_CGROUPWB;
aff063e2
JK
2681
2682 /* init f2fs-specific super block info */
e8240f65 2683 sbi->valid_super_block = valid_super_block;
aff063e2 2684 mutex_init(&sbi->gc_mutex);
aff063e2 2685 mutex_init(&sbi->cp_mutex);
b3582c68 2686 init_rwsem(&sbi->node_write);
59c9081b 2687 init_rwsem(&sbi->node_change);
315df839
JK
2688
2689 /* disallow all the data/node/meta page writes */
2690 set_sbi_flag(sbi, SBI_POR_DOING);
aff063e2 2691 spin_lock_init(&sbi->stat_lock);
971767ca 2692
b0af6d49
CY
2693 /* init iostat info */
2694 spin_lock_init(&sbi->iostat_lock);
2695 sbi->iostat_enable = false;
2696
458e6197 2697 for (i = 0; i < NR_PAGE_TYPE; i++) {
a912b54d
JK
2698 int n = (i == META) ? 1: NR_TEMP_TYPE;
2699 int j;
2700
4e6aad29
CY
2701 sbi->write_io[i] = f2fs_kmalloc(sbi,
2702 n * sizeof(struct f2fs_bio_info),
2703 GFP_KERNEL);
b63def91
CJ
2704 if (!sbi->write_io[i]) {
2705 err = -ENOMEM;
a912b54d 2706 goto free_options;
b63def91 2707 }
a912b54d
JK
2708
2709 for (j = HOT; j < n; j++) {
2710 init_rwsem(&sbi->write_io[i][j].io_rwsem);
2711 sbi->write_io[i][j].sbi = sbi;
2712 sbi->write_io[i][j].bio = NULL;
fb830fc5
CY
2713 spin_lock_init(&sbi->write_io[i][j].io_lock);
2714 INIT_LIST_HEAD(&sbi->write_io[i][j].io_list);
a912b54d 2715 }
458e6197 2716 }
971767ca 2717
b873b798 2718 init_rwsem(&sbi->cp_rwsem);
fb51b5ef 2719 init_waitqueue_head(&sbi->cp_wait);
aff063e2
JK
2720 init_sb_info(sbi);
2721
523be8a6
JK
2722 err = init_percpu_info(sbi);
2723 if (err)
d7997e63 2724 goto free_bio_info;
523be8a6 2725
0a595eba
JK
2726 if (F2FS_IO_SIZE(sbi) > 1) {
2727 sbi->write_io_dummy =
a3ebfe4f 2728 mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
1727f317
CY
2729 if (!sbi->write_io_dummy) {
2730 err = -ENOMEM;
d7997e63 2731 goto free_percpu;
1727f317 2732 }
0a595eba
JK
2733 }
2734
aff063e2
JK
2735 /* get an inode for meta space */
2736 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
2737 if (IS_ERR(sbi->meta_inode)) {
a07ef784 2738 f2fs_msg(sb, KERN_ERR, "Failed to read F2FS meta data inode");
aff063e2 2739 err = PTR_ERR(sbi->meta_inode);
0a595eba 2740 goto free_io_dummy;
aff063e2
JK
2741 }
2742
2743 err = get_valid_checkpoint(sbi);
a07ef784
NJ
2744 if (err) {
2745 f2fs_msg(sb, KERN_ERR, "Failed to get valid F2FS checkpoint");
aff063e2 2746 goto free_meta_inode;
a07ef784 2747 }
aff063e2 2748
3c62be17
JK
2749 /* Initialize device list */
2750 err = f2fs_scan_devices(sbi);
2751 if (err) {
2752 f2fs_msg(sb, KERN_ERR, "Failed to find devices");
2753 goto free_devices;
2754 }
2755
aff063e2
JK
2756 sbi->total_valid_node_count =
2757 le32_to_cpu(sbi->ckpt->valid_node_count);
513c5f37
JK
2758 percpu_counter_set(&sbi->total_valid_inode_count,
2759 le32_to_cpu(sbi->ckpt->valid_inode_count));
aff063e2
JK
2760 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
2761 sbi->total_valid_block_count =
2762 le64_to_cpu(sbi->ckpt->valid_block_count);
2763 sbi->last_valid_block_count = sbi->total_valid_block_count;
daeb433e 2764 sbi->reserved_blocks = 0;
80d42145 2765 sbi->current_reserved_blocks = 0;
7e65be49 2766 limit_reserve_root(sbi);
41382ec4 2767
c227f912
CY
2768 for (i = 0; i < NR_INODE_TYPE; i++) {
2769 INIT_LIST_HEAD(&sbi->inode_list[i]);
2770 spin_lock_init(&sbi->inode_lock[i]);
2771 }
aff063e2 2772
1dcc336b
CY
2773 init_extent_cache_info(sbi);
2774
6451e041 2775 init_ino_entry_info(sbi);
aff063e2
JK
2776
2777 /* setup f2fs internal modules */
2778 err = build_segment_manager(sbi);
a07ef784
NJ
2779 if (err) {
2780 f2fs_msg(sb, KERN_ERR,
2781 "Failed to initialize F2FS segment manager");
aff063e2 2782 goto free_sm;
a07ef784 2783 }
aff063e2 2784 err = build_node_manager(sbi);
a07ef784
NJ
2785 if (err) {
2786 f2fs_msg(sb, KERN_ERR,
2787 "Failed to initialize F2FS node manager");
aff063e2 2788 goto free_nm;
a07ef784 2789 }
aff063e2 2790
8f1dbbbb
SL
2791 /* For write statistics */
2792 if (sb->s_bdev->bd_part)
2793 sbi->sectors_written_start =
2794 (u64)part_stat_read(sb->s_bdev->bd_part, sectors[1]);
2795
2796 /* Read accumulated write IO statistics if exists */
2797 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
2798 if (__exist_node_summaries(sbi))
2799 sbi->kbytes_written =
b2dde6fc 2800 le64_to_cpu(seg_i->journal->info.kbytes_written);
8f1dbbbb 2801
aff063e2
JK
2802 build_gc_manager(sbi);
2803
2804 /* get an inode for node space */
2805 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
2806 if (IS_ERR(sbi->node_inode)) {
a07ef784 2807 f2fs_msg(sb, KERN_ERR, "Failed to read node inode");
aff063e2
JK
2808 err = PTR_ERR(sbi->node_inode);
2809 goto free_nm;
2810 }
2811
aa51d08a
JK
2812 err = f2fs_build_stats(sbi);
2813 if (err)
bae01eda 2814 goto free_node_inode;
aa51d08a 2815
aff063e2
JK
2816 /* read root inode and dentry */
2817 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
2818 if (IS_ERR(root)) {
a07ef784 2819 f2fs_msg(sb, KERN_ERR, "Failed to read root inode");
aff063e2 2820 err = PTR_ERR(root);
bae01eda 2821 goto free_stats;
aff063e2 2822 }
8f99a946 2823 if (!S_ISDIR(root->i_mode) || !root->i_blocks || !root->i_size) {
9d847950 2824 iput(root);
8f99a946 2825 err = -EINVAL;
9d847950 2826 goto free_node_inode;
8f99a946 2827 }
aff063e2
JK
2828
2829 sb->s_root = d_make_root(root); /* allocate root dentry */
2830 if (!sb->s_root) {
2831 err = -ENOMEM;
2832 goto free_root_inode;
2833 }
2834
dc6b2055 2835 err = f2fs_register_sysfs(sbi);
b59d0bae 2836 if (err)
a398101a 2837 goto free_root_inode;
b59d0bae 2838
ea676733
JK
2839#ifdef CONFIG_QUOTA
2840 /*
2841 * Turn on quotas which were not enabled for read-only mounts if
2842 * filesystem has quota feature, so that they are updated correctly.
2843 */
deeedd71 2844 if (f2fs_sb_has_quota_ino(sb) && !f2fs_readonly(sb)) {
ea676733
JK
2845 err = f2fs_enable_quotas(sb);
2846 if (err) {
2847 f2fs_msg(sb, KERN_ERR,
2848 "Cannot turn on quotas: error %d", err);
2849 goto free_sysfs;
2850 }
2851 }
2852#endif
4b2414d0
CY
2853 /* if there are nt orphan nodes free them */
2854 err = recover_orphan_inodes(sbi);
2855 if (err)
ea676733 2856 goto free_meta;
4b2414d0 2857
6437d1b0
JK
2858 /* recover fsynced data */
2859 if (!test_opt(sbi, DISABLE_ROLL_FORWARD)) {
081d78c2
JK
2860 /*
2861 * mount should be failed, when device has readonly mode, and
2862 * previous checkpoint was not done by clean system shutdown.
2863 */
2864 if (bdev_read_only(sb->s_bdev) &&
aaec2b1d 2865 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
081d78c2 2866 err = -EROFS;
4b2414d0 2867 goto free_meta;
081d78c2 2868 }
2adc3505
CY
2869
2870 if (need_fsck)
2871 set_sbi_flag(sbi, SBI_NEED_FSCK);
2872
a468f0ef
JK
2873 if (!retry)
2874 goto skip_recovery;
2875
6781eabb
JK
2876 err = recover_fsync_data(sbi, false);
2877 if (err < 0) {
2adc3505 2878 need_fsck = true;
6437d1b0 2879 f2fs_msg(sb, KERN_ERR,
99e3e858 2880 "Cannot recover all fsync data errno=%d", err);
4b2414d0 2881 goto free_meta;
ed2e621a 2882 }
6781eabb
JK
2883 } else {
2884 err = recover_fsync_data(sbi, true);
2885
2886 if (!f2fs_readonly(sb) && err > 0) {
2887 err = -EINVAL;
2888 f2fs_msg(sb, KERN_ERR,
2889 "Need to recover fsync data");
ea676733 2890 goto free_meta;
6781eabb 2891 }
6437d1b0 2892 }
a468f0ef 2893skip_recovery:
315df839
JK
2894 /* recover_fsync_data() cleared this already */
2895 clear_sbi_flag(sbi, SBI_POR_DOING);
b59d0bae 2896
6437d1b0
JK
2897 /*
2898 * If filesystem is not mounted as read-only then
2899 * do start the gc_thread.
2900 */
6c029932 2901 if (test_opt(sbi, BG_GC) && !f2fs_readonly(sb)) {
6437d1b0
JK
2902 /* After POR, we can run background GC thread.*/
2903 err = start_gc_thread(sbi);
2904 if (err)
4b2414d0 2905 goto free_meta;
6437d1b0 2906 }
dabc4a5c 2907 kfree(options);
da554e48 2908
2909 /* recover broken superblock */
f2353d7b 2910 if (recovery) {
41214b3c
CY
2911 err = f2fs_commit_super(sbi, true);
2912 f2fs_msg(sb, KERN_INFO,
99e3e858 2913 "Try to recover %dth superblock, ret: %d",
41214b3c 2914 sbi->valid_super_block ? 1 : 2, err);
da554e48 2915 }
2916
bae01eda
CY
2917 f2fs_join_shrinker(sbi);
2918
84b89e5d
JK
2919 f2fs_tuning_parameters(sbi);
2920
1200abb2
JK
2921 f2fs_msg(sbi->sb, KERN_NOTICE, "Mounted with checkpoint version = %llx",
2922 cur_cp_version(F2FS_CKPT(sbi)));
6beceb54 2923 f2fs_update_time(sbi, CP_TIME);
d0239e1b 2924 f2fs_update_time(sbi, REQ_TIME);
aff063e2 2925 return 0;
6437d1b0 2926
4b2414d0 2927free_meta:
ea676733 2928#ifdef CONFIG_QUOTA
deeedd71 2929 if (f2fs_sb_has_quota_ino(sb) && !f2fs_readonly(sb))
ea676733
JK
2930 f2fs_quota_off_umount(sbi->sb);
2931#endif
0f18b462 2932 f2fs_sync_inode_meta(sbi);
4b2414d0
CY
2933 /*
2934 * Some dirty meta pages can be produced by recover_orphan_inodes()
2935 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
2936 * followed by write_checkpoint() through f2fs_write_node_pages(), which
2937 * falls into an infinite loop in sync_meta_pages().
2938 */
2939 truncate_inode_pages_final(META_MAPPING(sbi));
ea676733 2940#ifdef CONFIG_QUOTA
4b2414d0 2941free_sysfs:
ea676733 2942#endif
dc6b2055 2943 f2fs_unregister_sysfs(sbi);
aff063e2
JK
2944free_root_inode:
2945 dput(sb->s_root);
2946 sb->s_root = NULL;
bae01eda
CY
2947free_stats:
2948 f2fs_destroy_stats(sbi);
aff063e2 2949free_node_inode:
d41065e2 2950 release_ino_entry(sbi, true);
bae01eda 2951 truncate_inode_pages_final(NODE_MAPPING(sbi));
aff063e2
JK
2952 iput(sbi->node_inode);
2953free_nm:
2954 destroy_node_manager(sbi);
2955free_sm:
2956 destroy_segment_manager(sbi);
3c62be17
JK
2957free_devices:
2958 destroy_device_list(sbi);
aff063e2
JK
2959 kfree(sbi->ckpt);
2960free_meta_inode:
2961 make_bad_inode(sbi->meta_inode);
2962 iput(sbi->meta_inode);
0a595eba
JK
2963free_io_dummy:
2964 mempool_destroy(sbi->write_io_dummy);
d7997e63
CY
2965free_percpu:
2966 destroy_percpu_info(sbi);
2967free_bio_info:
a912b54d
JK
2968 for (i = 0; i < NR_PAGE_TYPE; i++)
2969 kfree(sbi->write_io[i]);
d7997e63 2970free_options:
4b2414d0
CY
2971#ifdef CONFIG_QUOTA
2972 for (i = 0; i < MAXQUOTAS; i++)
2973 kfree(sbi->s_qf_names[i]);
2974#endif
dabc4a5c 2975 kfree(options);
aff063e2 2976free_sb_buf:
b39f0de2 2977 kfree(raw_super);
aff063e2 2978free_sbi:
43b6573b
KM
2979 if (sbi->s_chksum_driver)
2980 crypto_free_shash(sbi->s_chksum_driver);
aff063e2 2981 kfree(sbi);
ed2e621a
JK
2982
2983 /* give only one another chance */
2984 if (retry) {
9df47ba7 2985 retry = false;
ed2e621a
JK
2986 shrink_dcache_sb(sb);
2987 goto try_onemore;
2988 }
aff063e2
JK
2989 return err;
2990}
2991
2992static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
2993 const char *dev_name, void *data)
2994{
2995 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
2996}
2997
30a5537f
JK
2998static void kill_f2fs_super(struct super_block *sb)
2999{
cce13252 3000 if (sb->s_root) {
caf0047e 3001 set_sbi_flag(F2FS_SB(sb), SBI_IS_CLOSE);
cce13252
CY
3002 stop_gc_thread(F2FS_SB(sb));
3003 stop_discard_thread(F2FS_SB(sb));
3004 }
30a5537f
JK
3005 kill_block_super(sb);
3006}
3007
aff063e2
JK
3008static struct file_system_type f2fs_fs_type = {
3009 .owner = THIS_MODULE,
3010 .name = "f2fs",
3011 .mount = f2fs_mount,
30a5537f 3012 .kill_sb = kill_f2fs_super,
aff063e2
JK
3013 .fs_flags = FS_REQUIRES_DEV,
3014};
7f78e035 3015MODULE_ALIAS_FS("f2fs");
aff063e2 3016
6e6093a8 3017static int __init init_inodecache(void)
aff063e2 3018{
5d097056
VD
3019 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
3020 sizeof(struct f2fs_inode_info), 0,
3021 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
6bacf52f 3022 if (!f2fs_inode_cachep)
aff063e2
JK
3023 return -ENOMEM;
3024 return 0;
3025}
3026
3027static void destroy_inodecache(void)
3028{
3029 /*
3030 * Make sure all delayed rcu free inodes are flushed before we
3031 * destroy cache.
3032 */
3033 rcu_barrier();
3034 kmem_cache_destroy(f2fs_inode_cachep);
3035}
3036
3037static int __init init_f2fs_fs(void)
3038{
3039 int err;
3040
c0508650
JK
3041 f2fs_build_trace_ios();
3042
aff063e2
JK
3043 err = init_inodecache();
3044 if (err)
3045 goto fail;
3046 err = create_node_manager_caches();
3047 if (err)
9890ff3f 3048 goto free_inodecache;
7fd9e544 3049 err = create_segment_manager_caches();
aff063e2 3050 if (err)
9890ff3f 3051 goto free_node_manager_caches;
aff063e2
JK
3052 err = create_checkpoint_caches();
3053 if (err)
06292073 3054 goto free_segment_manager_caches;
1dcc336b
CY
3055 err = create_extent_cache();
3056 if (err)
3057 goto free_checkpoint_caches;
dc6b2055 3058 err = f2fs_init_sysfs();
a398101a 3059 if (err)
1dcc336b 3060 goto free_extent_cache;
2658e50d 3061 err = register_shrinker(&f2fs_shrinker_info);
cfc4d971 3062 if (err)
a398101a 3063 goto free_sysfs;
2658e50d
JK
3064 err = register_filesystem(&f2fs_fs_type);
3065 if (err)
3066 goto free_shrinker;
787c7b8c
CY
3067 err = f2fs_create_root_stats();
3068 if (err)
3069 goto free_filesystem;
9890ff3f
ZH
3070 return 0;
3071
787c7b8c
CY
3072free_filesystem:
3073 unregister_filesystem(&f2fs_fs_type);
2658e50d
JK
3074free_shrinker:
3075 unregister_shrinker(&f2fs_shrinker_info);
a398101a 3076free_sysfs:
dc6b2055 3077 f2fs_exit_sysfs();
1dcc336b
CY
3078free_extent_cache:
3079 destroy_extent_cache();
9890ff3f
ZH
3080free_checkpoint_caches:
3081 destroy_checkpoint_caches();
7fd9e544
JK
3082free_segment_manager_caches:
3083 destroy_segment_manager_caches();
9890ff3f
ZH
3084free_node_manager_caches:
3085 destroy_node_manager_caches();
3086free_inodecache:
3087 destroy_inodecache();
aff063e2
JK
3088fail:
3089 return err;
3090}
3091
3092static void __exit exit_f2fs_fs(void)
3093{
4589d25d 3094 f2fs_destroy_root_stats();
aff063e2 3095 unregister_filesystem(&f2fs_fs_type);
b8bef79d 3096 unregister_shrinker(&f2fs_shrinker_info);
dc6b2055 3097 f2fs_exit_sysfs();
fdf6c8be 3098 destroy_extent_cache();
aff063e2 3099 destroy_checkpoint_caches();
5dcd8a71 3100 destroy_segment_manager_caches();
aff063e2
JK
3101 destroy_node_manager_caches();
3102 destroy_inodecache();
351f4fba 3103 f2fs_destroy_trace_ios();
aff063e2
JK
3104}
3105
3106module_init(init_f2fs_fs)
3107module_exit(exit_f2fs_fs)
3108
3109MODULE_AUTHOR("Samsung Electronics's Praesto Team");
3110MODULE_DESCRIPTION("Flash Friendly File System");
3111MODULE_LICENSE("GPL");
b4b9d34c 3112