f2fs: fix retry logic in f2fs_write_cache_pages()
[linux-block.git] / fs / f2fs / super.c
CommitLineData
7c1a000d 1// SPDX-License-Identifier: GPL-2.0
0a8165d7 2/*
aff063e2
JK
3 * fs/f2fs/super.c
4 *
5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6 * http://www.samsung.com/
aff063e2
JK
7 */
8#include <linux/module.h>
9#include <linux/init.h>
10#include <linux/fs.h>
11#include <linux/statfs.h>
aff063e2
JK
12#include <linux/buffer_head.h>
13#include <linux/backing-dev.h>
14#include <linux/kthread.h>
15#include <linux/parser.h>
16#include <linux/mount.h>
17#include <linux/seq_file.h>
5e176d54 18#include <linux/proc_fs.h>
aff063e2
JK
19#include <linux/random.h>
20#include <linux/exportfs.h>
d3ee456d 21#include <linux/blkdev.h>
0abd675e 22#include <linux/quotaops.h>
aff063e2 23#include <linux/f2fs_fs.h>
b59d0bae 24#include <linux/sysfs.h>
4b2414d0 25#include <linux/quota.h>
5aba5430 26#include <linux/unicode.h>
c6a564ff 27#include <linux/part_stat.h>
aff063e2
JK
28
29#include "f2fs.h"
30#include "node.h"
5ec4e49f 31#include "segment.h"
aff063e2 32#include "xattr.h"
b59d0bae 33#include "gc.h"
db9f7c1a 34#include "trace.h"
aff063e2 35
a2a4a7e4
NJ
36#define CREATE_TRACE_POINTS
37#include <trace/events/f2fs.h>
38
aff063e2
JK
39static struct kmem_cache *f2fs_inode_cachep;
40
73faec4d 41#ifdef CONFIG_F2FS_FAULT_INJECTION
2c63fead 42
19880e6e 43const char *f2fs_fault_name[FAULT_MAX] = {
2c63fead 44 [FAULT_KMALLOC] = "kmalloc",
628b3d14 45 [FAULT_KVMALLOC] = "kvmalloc",
c41f3cc3 46 [FAULT_PAGE_ALLOC] = "page alloc",
01eccef7 47 [FAULT_PAGE_GET] = "page get",
d62fe971 48 [FAULT_ALLOC_BIO] = "alloc bio",
cb78942b
JK
49 [FAULT_ALLOC_NID] = "alloc nid",
50 [FAULT_ORPHAN] = "orphan",
51 [FAULT_BLOCK] = "no more block",
52 [FAULT_DIR_DEPTH] = "too big dir depth",
53aa6bbf 53 [FAULT_EVICT_INODE] = "evict_inode fail",
14b44d23 54 [FAULT_TRUNCATE] = "truncate fail",
6f5c2ed0 55 [FAULT_READ_IO] = "read IO error",
0f348028 56 [FAULT_CHECKPOINT] = "checkpoint error",
b83dcfe6 57 [FAULT_DISCARD] = "discard error",
6f5c2ed0 58 [FAULT_WRITE_IO] = "write IO error",
2c63fead 59};
08796897 60
d494500a
CY
61void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
62 unsigned int type)
08796897 63{
63189b78 64 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
1ecc0c5c 65
08796897 66 if (rate) {
1ecc0c5c
CY
67 atomic_set(&ffi->inject_ops, 0);
68 ffi->inject_rate = rate;
08796897 69 }
d494500a
CY
70
71 if (type)
72 ffi->inject_type = type;
73
74 if (!rate && !type)
75 memset(ffi, 0, sizeof(struct f2fs_fault_info));
08796897 76}
73faec4d
JK
77#endif
78
2658e50d
JK
79/* f2fs-wide shrinker description */
80static struct shrinker f2fs_shrinker_info = {
81 .scan_objects = f2fs_shrink_scan,
82 .count_objects = f2fs_shrink_count,
83 .seeks = DEFAULT_SEEKS,
84};
85
aff063e2 86enum {
696c018c 87 Opt_gc_background,
aff063e2 88 Opt_disable_roll_forward,
2d834bf9 89 Opt_norecovery,
aff063e2 90 Opt_discard,
64058be9 91 Opt_nodiscard,
aff063e2 92 Opt_noheap,
7a20b8a6 93 Opt_heap,
4058c511 94 Opt_user_xattr,
aff063e2 95 Opt_nouser_xattr,
4058c511 96 Opt_acl,
aff063e2
JK
97 Opt_noacl,
98 Opt_active_logs,
99 Opt_disable_ext_identify,
444c580f 100 Opt_inline_xattr,
23cf7212 101 Opt_noinline_xattr,
6afc662e 102 Opt_inline_xattr_size,
8274de77 103 Opt_inline_data,
5efd3c6f 104 Opt_inline_dentry,
97c1794a 105 Opt_noinline_dentry,
6b4afdd7 106 Opt_flush_merge,
69e9e427 107 Opt_noflush_merge,
0f7b2abd 108 Opt_nobarrier,
d5053a34 109 Opt_fastboot,
89672159 110 Opt_extent_cache,
7daaea25 111 Opt_noextent_cache,
75342797 112 Opt_noinline_data,
343f40f0 113 Opt_data_flush,
7e65be49 114 Opt_reserve_root,
7c2e5963
JK
115 Opt_resgid,
116 Opt_resuid,
36abef4e 117 Opt_mode,
ec91538d 118 Opt_io_size_bits,
73faec4d 119 Opt_fault_injection,
d494500a 120 Opt_fault_type,
6d94c74a
JK
121 Opt_lazytime,
122 Opt_nolazytime,
4b2414d0
CY
123 Opt_quota,
124 Opt_noquota,
0abd675e
CY
125 Opt_usrquota,
126 Opt_grpquota,
5c57132e 127 Opt_prjquota,
4b2414d0
CY
128 Opt_usrjquota,
129 Opt_grpjquota,
130 Opt_prjjquota,
131 Opt_offusrjquota,
132 Opt_offgrpjquota,
133 Opt_offprjjquota,
134 Opt_jqfmt_vfsold,
135 Opt_jqfmt_vfsv0,
136 Opt_jqfmt_vfsv1,
0cdd3195 137 Opt_whint,
07939627 138 Opt_alloc,
93cf93f1 139 Opt_fsync,
ff62af20 140 Opt_test_dummy_encryption,
4d3aed70
DR
141 Opt_checkpoint_disable,
142 Opt_checkpoint_disable_cap,
143 Opt_checkpoint_disable_cap_perc,
144 Opt_checkpoint_enable,
4c8ff709
CY
145 Opt_compress_algorithm,
146 Opt_compress_log_size,
147 Opt_compress_extension,
aff063e2
JK
148 Opt_err,
149};
150
151static match_table_t f2fs_tokens = {
696c018c 152 {Opt_gc_background, "background_gc=%s"},
aff063e2 153 {Opt_disable_roll_forward, "disable_roll_forward"},
2d834bf9 154 {Opt_norecovery, "norecovery"},
aff063e2 155 {Opt_discard, "discard"},
64058be9 156 {Opt_nodiscard, "nodiscard"},
aff063e2 157 {Opt_noheap, "no_heap"},
7a20b8a6 158 {Opt_heap, "heap"},
4058c511 159 {Opt_user_xattr, "user_xattr"},
aff063e2 160 {Opt_nouser_xattr, "nouser_xattr"},
4058c511 161 {Opt_acl, "acl"},
aff063e2
JK
162 {Opt_noacl, "noacl"},
163 {Opt_active_logs, "active_logs=%u"},
164 {Opt_disable_ext_identify, "disable_ext_identify"},
444c580f 165 {Opt_inline_xattr, "inline_xattr"},
23cf7212 166 {Opt_noinline_xattr, "noinline_xattr"},
6afc662e 167 {Opt_inline_xattr_size, "inline_xattr_size=%u"},
8274de77 168 {Opt_inline_data, "inline_data"},
5efd3c6f 169 {Opt_inline_dentry, "inline_dentry"},
97c1794a 170 {Opt_noinline_dentry, "noinline_dentry"},
6b4afdd7 171 {Opt_flush_merge, "flush_merge"},
69e9e427 172 {Opt_noflush_merge, "noflush_merge"},
0f7b2abd 173 {Opt_nobarrier, "nobarrier"},
d5053a34 174 {Opt_fastboot, "fastboot"},
89672159 175 {Opt_extent_cache, "extent_cache"},
7daaea25 176 {Opt_noextent_cache, "noextent_cache"},
75342797 177 {Opt_noinline_data, "noinline_data"},
343f40f0 178 {Opt_data_flush, "data_flush"},
7e65be49 179 {Opt_reserve_root, "reserve_root=%u"},
7c2e5963
JK
180 {Opt_resgid, "resgid=%u"},
181 {Opt_resuid, "resuid=%u"},
36abef4e 182 {Opt_mode, "mode=%s"},
ec91538d 183 {Opt_io_size_bits, "io_bits=%u"},
73faec4d 184 {Opt_fault_injection, "fault_injection=%u"},
d494500a 185 {Opt_fault_type, "fault_type=%u"},
6d94c74a
JK
186 {Opt_lazytime, "lazytime"},
187 {Opt_nolazytime, "nolazytime"},
4b2414d0
CY
188 {Opt_quota, "quota"},
189 {Opt_noquota, "noquota"},
0abd675e
CY
190 {Opt_usrquota, "usrquota"},
191 {Opt_grpquota, "grpquota"},
5c57132e 192 {Opt_prjquota, "prjquota"},
4b2414d0
CY
193 {Opt_usrjquota, "usrjquota=%s"},
194 {Opt_grpjquota, "grpjquota=%s"},
195 {Opt_prjjquota, "prjjquota=%s"},
196 {Opt_offusrjquota, "usrjquota="},
197 {Opt_offgrpjquota, "grpjquota="},
198 {Opt_offprjjquota, "prjjquota="},
199 {Opt_jqfmt_vfsold, "jqfmt=vfsold"},
200 {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
201 {Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
0cdd3195 202 {Opt_whint, "whint_mode=%s"},
07939627 203 {Opt_alloc, "alloc_mode=%s"},
93cf93f1 204 {Opt_fsync, "fsync_mode=%s"},
ff62af20 205 {Opt_test_dummy_encryption, "test_dummy_encryption"},
4d3aed70
DR
206 {Opt_checkpoint_disable, "checkpoint=disable"},
207 {Opt_checkpoint_disable_cap, "checkpoint=disable:%u"},
208 {Opt_checkpoint_disable_cap_perc, "checkpoint=disable:%u%%"},
209 {Opt_checkpoint_enable, "checkpoint=enable"},
4c8ff709
CY
210 {Opt_compress_algorithm, "compress_algorithm=%s"},
211 {Opt_compress_log_size, "compress_log_size=%u"},
212 {Opt_compress_extension, "compress_extension=%s"},
aff063e2
JK
213 {Opt_err, NULL},
214};
215
dcbb4c10 216void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...)
a07ef784
NJ
217{
218 struct va_format vaf;
219 va_list args;
dcbb4c10 220 int level;
a07ef784
NJ
221
222 va_start(args, fmt);
dcbb4c10
JP
223
224 level = printk_get_level(fmt);
225 vaf.fmt = printk_skip_level(fmt);
a07ef784 226 vaf.va = &args;
dcbb4c10
JP
227 printk("%c%cF2FS-fs (%s): %pV\n",
228 KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
229
a07ef784
NJ
230 va_end(args);
231}
232
5aba5430
DR
233#ifdef CONFIG_UNICODE
234static const struct f2fs_sb_encodings {
235 __u16 magic;
236 char *name;
237 char *version;
238} f2fs_sb_encoding_map[] = {
239 {F2FS_ENC_UTF8_12_1, "utf8", "12.1.0"},
240};
241
242static int f2fs_sb_read_encoding(const struct f2fs_super_block *sb,
243 const struct f2fs_sb_encodings **encoding,
244 __u16 *flags)
245{
246 __u16 magic = le16_to_cpu(sb->s_encoding);
247 int i;
248
249 for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
250 if (magic == f2fs_sb_encoding_map[i].magic)
251 break;
252
253 if (i >= ARRAY_SIZE(f2fs_sb_encoding_map))
254 return -EINVAL;
255
256 *encoding = &f2fs_sb_encoding_map[i];
257 *flags = le16_to_cpu(sb->s_encoding_flags);
258
259 return 0;
260}
261#endif
262
7e65be49
JK
263static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
264{
9a9aecaa
DR
265 block_t limit = min((sbi->user_block_count << 1) / 1000,
266 sbi->user_block_count - sbi->reserved_blocks);
7e65be49
JK
267
268 /* limit is 0.2% */
63189b78
CY
269 if (test_opt(sbi, RESERVE_ROOT) &&
270 F2FS_OPTION(sbi).root_reserved_blocks > limit) {
271 F2FS_OPTION(sbi).root_reserved_blocks = limit;
dcbb4c10
JP
272 f2fs_info(sbi, "Reduce reserved blocks for root = %u",
273 F2FS_OPTION(sbi).root_reserved_blocks);
7e65be49 274 }
7c2e5963 275 if (!test_opt(sbi, RESERVE_ROOT) &&
63189b78 276 (!uid_eq(F2FS_OPTION(sbi).s_resuid,
7c2e5963 277 make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
63189b78 278 !gid_eq(F2FS_OPTION(sbi).s_resgid,
7c2e5963 279 make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
dcbb4c10
JP
280 f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
281 from_kuid_munged(&init_user_ns,
282 F2FS_OPTION(sbi).s_resuid),
283 from_kgid_munged(&init_user_ns,
284 F2FS_OPTION(sbi).s_resgid));
7e65be49
JK
285}
286
1ae18f71
JK
287static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
288{
289 if (!F2FS_OPTION(sbi).unusable_cap_perc)
290 return;
291
292 if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
293 F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
294 else
295 F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
296 F2FS_OPTION(sbi).unusable_cap_perc;
297
298 f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
299 F2FS_OPTION(sbi).unusable_cap,
300 F2FS_OPTION(sbi).unusable_cap_perc);
301}
302
aff063e2
JK
303static void init_once(void *foo)
304{
305 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
306
aff063e2
JK
307 inode_init_once(&fi->vfs_inode);
308}
309
4b2414d0
CY
310#ifdef CONFIG_QUOTA
311static const char * const quotatypes[] = INITQFNAMES;
312#define QTYPE2NAME(t) (quotatypes[t])
313static int f2fs_set_qf_name(struct super_block *sb, int qtype,
314 substring_t *args)
315{
316 struct f2fs_sb_info *sbi = F2FS_SB(sb);
317 char *qname;
318 int ret = -EINVAL;
319
63189b78 320 if (sb_any_quota_loaded(sb) && !F2FS_OPTION(sbi).s_qf_names[qtype]) {
dcbb4c10 321 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
4b2414d0
CY
322 return -EINVAL;
323 }
7beb01f7 324 if (f2fs_sb_has_quota_ino(sbi)) {
dcbb4c10 325 f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
ea676733
JK
326 return 0;
327 }
328
4b2414d0
CY
329 qname = match_strdup(args);
330 if (!qname) {
dcbb4c10 331 f2fs_err(sbi, "Not enough memory for storing quotafile name");
f365c6cc 332 return -ENOMEM;
4b2414d0 333 }
63189b78
CY
334 if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
335 if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
4b2414d0
CY
336 ret = 0;
337 else
dcbb4c10 338 f2fs_err(sbi, "%s quota file already specified",
4b2414d0
CY
339 QTYPE2NAME(qtype));
340 goto errout;
341 }
342 if (strchr(qname, '/')) {
dcbb4c10 343 f2fs_err(sbi, "quotafile must be on filesystem root");
4b2414d0
CY
344 goto errout;
345 }
63189b78 346 F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
4b2414d0
CY
347 set_opt(sbi, QUOTA);
348 return 0;
349errout:
5222595d 350 kvfree(qname);
4b2414d0
CY
351 return ret;
352}
353
354static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
355{
356 struct f2fs_sb_info *sbi = F2FS_SB(sb);
357
63189b78 358 if (sb_any_quota_loaded(sb) && F2FS_OPTION(sbi).s_qf_names[qtype]) {
dcbb4c10 359 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
4b2414d0
CY
360 return -EINVAL;
361 }
5222595d 362 kvfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
63189b78 363 F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
4b2414d0
CY
364 return 0;
365}
366
367static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
368{
369 /*
370 * We do the test below only for project quotas. 'usrquota' and
371 * 'grpquota' mount options are allowed even without quota feature
372 * to support legacy quotas in quota files.
373 */
7beb01f7 374 if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi)) {
dcbb4c10 375 f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
4b2414d0
CY
376 return -1;
377 }
63189b78
CY
378 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
379 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
380 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) {
381 if (test_opt(sbi, USRQUOTA) &&
382 F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
4b2414d0
CY
383 clear_opt(sbi, USRQUOTA);
384
63189b78
CY
385 if (test_opt(sbi, GRPQUOTA) &&
386 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
4b2414d0
CY
387 clear_opt(sbi, GRPQUOTA);
388
63189b78
CY
389 if (test_opt(sbi, PRJQUOTA) &&
390 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
4b2414d0
CY
391 clear_opt(sbi, PRJQUOTA);
392
393 if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
394 test_opt(sbi, PRJQUOTA)) {
dcbb4c10 395 f2fs_err(sbi, "old and new quota format mixing");
4b2414d0
CY
396 return -1;
397 }
398
63189b78 399 if (!F2FS_OPTION(sbi).s_jquota_fmt) {
dcbb4c10 400 f2fs_err(sbi, "journaled quota format not specified");
4b2414d0
CY
401 return -1;
402 }
403 }
ea676733 404
7beb01f7 405 if (f2fs_sb_has_quota_ino(sbi) && F2FS_OPTION(sbi).s_jquota_fmt) {
dcbb4c10 406 f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
63189b78 407 F2FS_OPTION(sbi).s_jquota_fmt = 0;
ea676733 408 }
4b2414d0
CY
409 return 0;
410}
411#endif
412
696c018c
NJ
413static int parse_options(struct super_block *sb, char *options)
414{
415 struct f2fs_sb_info *sbi = F2FS_SB(sb);
416 substring_t args[MAX_OPT_ARGS];
4c8ff709 417 unsigned char (*ext)[F2FS_EXTENSION_LEN];
696c018c 418 char *p, *name;
4c8ff709 419 int arg = 0, ext_cnt;
7c2e5963
JK
420 kuid_t uid;
421 kgid_t gid;
4b2414d0
CY
422#ifdef CONFIG_QUOTA
423 int ret;
424#endif
696c018c
NJ
425
426 if (!options)
427 return 0;
428
429 while ((p = strsep(&options, ",")) != NULL) {
430 int token;
431 if (!*p)
432 continue;
433 /*
434 * Initialize args struct so we know whether arg was
435 * found; some options take optional arguments.
436 */
437 args[0].to = args[0].from = NULL;
438 token = match_token(p, f2fs_tokens, args);
439
440 switch (token) {
441 case Opt_gc_background:
442 name = match_strdup(&args[0]);
443
444 if (!name)
445 return -ENOMEM;
3c57f751 446 if (!strcmp(name, "on")) {
bbbc34fd 447 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
3c57f751 448 } else if (!strcmp(name, "off")) {
bbbc34fd 449 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_OFF;
3c57f751 450 } else if (!strcmp(name, "sync")) {
bbbc34fd 451 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_SYNC;
6aefd93b 452 } else {
5222595d 453 kvfree(name);
696c018c
NJ
454 return -EINVAL;
455 }
5222595d 456 kvfree(name);
696c018c
NJ
457 break;
458 case Opt_disable_roll_forward:
459 set_opt(sbi, DISABLE_ROLL_FORWARD);
460 break;
2d834bf9
JK
461 case Opt_norecovery:
462 /* this option mounts f2fs with ro */
a9117eca 463 set_opt(sbi, NORECOVERY);
2d834bf9
JK
464 if (!f2fs_readonly(sb))
465 return -EINVAL;
466 break;
696c018c 467 case Opt_discard:
7d20c8ab 468 set_opt(sbi, DISCARD);
696c018c 469 break;
64058be9 470 case Opt_nodiscard:
7beb01f7 471 if (f2fs_sb_has_blkzoned(sbi)) {
dcbb4c10 472 f2fs_warn(sbi, "discard is required for zoned block devices");
96ba2dec
DLM
473 return -EINVAL;
474 }
64058be9 475 clear_opt(sbi, DISCARD);
487df616 476 break;
696c018c
NJ
477 case Opt_noheap:
478 set_opt(sbi, NOHEAP);
479 break;
7a20b8a6
JK
480 case Opt_heap:
481 clear_opt(sbi, NOHEAP);
482 break;
696c018c 483#ifdef CONFIG_F2FS_FS_XATTR
4058c511
KA
484 case Opt_user_xattr:
485 set_opt(sbi, XATTR_USER);
486 break;
696c018c
NJ
487 case Opt_nouser_xattr:
488 clear_opt(sbi, XATTR_USER);
489 break;
444c580f
JK
490 case Opt_inline_xattr:
491 set_opt(sbi, INLINE_XATTR);
492 break;
23cf7212
CY
493 case Opt_noinline_xattr:
494 clear_opt(sbi, INLINE_XATTR);
495 break;
6afc662e
CY
496 case Opt_inline_xattr_size:
497 if (args->from && match_int(args, &arg))
498 return -EINVAL;
499 set_opt(sbi, INLINE_XATTR_SIZE);
63189b78 500 F2FS_OPTION(sbi).inline_xattr_size = arg;
6afc662e 501 break;
696c018c 502#else
4058c511 503 case Opt_user_xattr:
dcbb4c10 504 f2fs_info(sbi, "user_xattr options not supported");
4058c511 505 break;
696c018c 506 case Opt_nouser_xattr:
dcbb4c10 507 f2fs_info(sbi, "nouser_xattr options not supported");
696c018c 508 break;
444c580f 509 case Opt_inline_xattr:
dcbb4c10 510 f2fs_info(sbi, "inline_xattr options not supported");
444c580f 511 break;
23cf7212 512 case Opt_noinline_xattr:
dcbb4c10 513 f2fs_info(sbi, "noinline_xattr options not supported");
23cf7212 514 break;
696c018c
NJ
515#endif
516#ifdef CONFIG_F2FS_FS_POSIX_ACL
4058c511
KA
517 case Opt_acl:
518 set_opt(sbi, POSIX_ACL);
519 break;
696c018c
NJ
520 case Opt_noacl:
521 clear_opt(sbi, POSIX_ACL);
522 break;
523#else
4058c511 524 case Opt_acl:
dcbb4c10 525 f2fs_info(sbi, "acl options not supported");
4058c511 526 break;
696c018c 527 case Opt_noacl:
dcbb4c10 528 f2fs_info(sbi, "noacl options not supported");
696c018c
NJ
529 break;
530#endif
531 case Opt_active_logs:
532 if (args->from && match_int(args, &arg))
533 return -EINVAL;
534 if (arg != 2 && arg != 4 && arg != NR_CURSEG_TYPE)
535 return -EINVAL;
63189b78 536 F2FS_OPTION(sbi).active_logs = arg;
696c018c
NJ
537 break;
538 case Opt_disable_ext_identify:
539 set_opt(sbi, DISABLE_EXT_IDENTIFY);
540 break;
8274de77
HL
541 case Opt_inline_data:
542 set_opt(sbi, INLINE_DATA);
543 break;
5efd3c6f
CY
544 case Opt_inline_dentry:
545 set_opt(sbi, INLINE_DENTRY);
546 break;
97c1794a
CY
547 case Opt_noinline_dentry:
548 clear_opt(sbi, INLINE_DENTRY);
549 break;
6b4afdd7
JK
550 case Opt_flush_merge:
551 set_opt(sbi, FLUSH_MERGE);
552 break;
69e9e427
JK
553 case Opt_noflush_merge:
554 clear_opt(sbi, FLUSH_MERGE);
555 break;
0f7b2abd
JK
556 case Opt_nobarrier:
557 set_opt(sbi, NOBARRIER);
558 break;
d5053a34
JK
559 case Opt_fastboot:
560 set_opt(sbi, FASTBOOT);
561 break;
89672159
CY
562 case Opt_extent_cache:
563 set_opt(sbi, EXTENT_CACHE);
564 break;
7daaea25
JK
565 case Opt_noextent_cache:
566 clear_opt(sbi, EXTENT_CACHE);
567 break;
75342797
WL
568 case Opt_noinline_data:
569 clear_opt(sbi, INLINE_DATA);
570 break;
343f40f0
CY
571 case Opt_data_flush:
572 set_opt(sbi, DATA_FLUSH);
573 break;
7e65be49
JK
574 case Opt_reserve_root:
575 if (args->from && match_int(args, &arg))
576 return -EINVAL;
577 if (test_opt(sbi, RESERVE_ROOT)) {
dcbb4c10
JP
578 f2fs_info(sbi, "Preserve previous reserve_root=%u",
579 F2FS_OPTION(sbi).root_reserved_blocks);
7e65be49 580 } else {
63189b78 581 F2FS_OPTION(sbi).root_reserved_blocks = arg;
7e65be49
JK
582 set_opt(sbi, RESERVE_ROOT);
583 }
584 break;
7c2e5963
JK
585 case Opt_resuid:
586 if (args->from && match_int(args, &arg))
587 return -EINVAL;
588 uid = make_kuid(current_user_ns(), arg);
589 if (!uid_valid(uid)) {
dcbb4c10 590 f2fs_err(sbi, "Invalid uid value %d", arg);
7c2e5963
JK
591 return -EINVAL;
592 }
63189b78 593 F2FS_OPTION(sbi).s_resuid = uid;
7c2e5963
JK
594 break;
595 case Opt_resgid:
596 if (args->from && match_int(args, &arg))
597 return -EINVAL;
598 gid = make_kgid(current_user_ns(), arg);
599 if (!gid_valid(gid)) {
dcbb4c10 600 f2fs_err(sbi, "Invalid gid value %d", arg);
7c2e5963
JK
601 return -EINVAL;
602 }
63189b78 603 F2FS_OPTION(sbi).s_resgid = gid;
7c2e5963 604 break;
36abef4e
JK
605 case Opt_mode:
606 name = match_strdup(&args[0]);
607
608 if (!name)
609 return -ENOMEM;
3c57f751 610 if (!strcmp(name, "adaptive")) {
7beb01f7 611 if (f2fs_sb_has_blkzoned(sbi)) {
dcbb4c10 612 f2fs_warn(sbi, "adaptive mode is not allowed with zoned block device feature");
5222595d 613 kvfree(name);
3adc57e9
DLM
614 return -EINVAL;
615 }
b0332a0f 616 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
3c57f751 617 } else if (!strcmp(name, "lfs")) {
b0332a0f 618 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
36abef4e 619 } else {
5222595d 620 kvfree(name);
36abef4e
JK
621 return -EINVAL;
622 }
5222595d 623 kvfree(name);
36abef4e 624 break;
ec91538d
JK
625 case Opt_io_size_bits:
626 if (args->from && match_int(args, &arg))
627 return -EINVAL;
6d52e135 628 if (arg <= 0 || arg > __ilog2_u32(BIO_MAX_PAGES)) {
dcbb4c10
JP
629 f2fs_warn(sbi, "Not support %d, larger than %d",
630 1 << arg, BIO_MAX_PAGES);
ec91538d
JK
631 return -EINVAL;
632 }
63189b78 633 F2FS_OPTION(sbi).write_io_size_bits = arg;
ec91538d 634 break;
4cb037ec 635#ifdef CONFIG_F2FS_FAULT_INJECTION
73faec4d
JK
636 case Opt_fault_injection:
637 if (args->from && match_int(args, &arg))
638 return -EINVAL;
d494500a
CY
639 f2fs_build_fault_attr(sbi, arg, F2FS_ALL_FAULT_TYPE);
640 set_opt(sbi, FAULT_INJECTION);
d494500a 641 break;
4cb037ec 642
d494500a
CY
643 case Opt_fault_type:
644 if (args->from && match_int(args, &arg))
645 return -EINVAL;
d494500a 646 f2fs_build_fault_attr(sbi, 0, arg);
0cc0dec2 647 set_opt(sbi, FAULT_INJECTION);
4cb037ec 648 break;
73faec4d 649#else
4cb037ec 650 case Opt_fault_injection:
dcbb4c10 651 f2fs_info(sbi, "fault_injection options not supported");
73faec4d 652 break;
4cb037ec
CX
653
654 case Opt_fault_type:
dcbb4c10 655 f2fs_info(sbi, "fault_type options not supported");
4cb037ec
CX
656 break;
657#endif
6d94c74a 658 case Opt_lazytime:
1751e8a6 659 sb->s_flags |= SB_LAZYTIME;
6d94c74a
JK
660 break;
661 case Opt_nolazytime:
1751e8a6 662 sb->s_flags &= ~SB_LAZYTIME;
6d94c74a 663 break;
0abd675e 664#ifdef CONFIG_QUOTA
4b2414d0 665 case Opt_quota:
0abd675e
CY
666 case Opt_usrquota:
667 set_opt(sbi, USRQUOTA);
668 break;
669 case Opt_grpquota:
670 set_opt(sbi, GRPQUOTA);
671 break;
5c57132e
CY
672 case Opt_prjquota:
673 set_opt(sbi, PRJQUOTA);
674 break;
4b2414d0
CY
675 case Opt_usrjquota:
676 ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
677 if (ret)
678 return ret;
679 break;
680 case Opt_grpjquota:
681 ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
682 if (ret)
683 return ret;
684 break;
685 case Opt_prjjquota:
686 ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
687 if (ret)
688 return ret;
689 break;
690 case Opt_offusrjquota:
691 ret = f2fs_clear_qf_name(sb, USRQUOTA);
692 if (ret)
693 return ret;
694 break;
695 case Opt_offgrpjquota:
696 ret = f2fs_clear_qf_name(sb, GRPQUOTA);
697 if (ret)
698 return ret;
699 break;
700 case Opt_offprjjquota:
701 ret = f2fs_clear_qf_name(sb, PRJQUOTA);
702 if (ret)
703 return ret;
704 break;
705 case Opt_jqfmt_vfsold:
63189b78 706 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
4b2414d0
CY
707 break;
708 case Opt_jqfmt_vfsv0:
63189b78 709 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
4b2414d0
CY
710 break;
711 case Opt_jqfmt_vfsv1:
63189b78 712 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
4b2414d0
CY
713 break;
714 case Opt_noquota:
715 clear_opt(sbi, QUOTA);
716 clear_opt(sbi, USRQUOTA);
717 clear_opt(sbi, GRPQUOTA);
718 clear_opt(sbi, PRJQUOTA);
719 break;
0abd675e 720#else
4b2414d0 721 case Opt_quota:
0abd675e
CY
722 case Opt_usrquota:
723 case Opt_grpquota:
5c57132e 724 case Opt_prjquota:
4b2414d0
CY
725 case Opt_usrjquota:
726 case Opt_grpjquota:
727 case Opt_prjjquota:
728 case Opt_offusrjquota:
729 case Opt_offgrpjquota:
730 case Opt_offprjjquota:
731 case Opt_jqfmt_vfsold:
732 case Opt_jqfmt_vfsv0:
733 case Opt_jqfmt_vfsv1:
734 case Opt_noquota:
dcbb4c10 735 f2fs_info(sbi, "quota operations not supported");
0abd675e
CY
736 break;
737#endif
0cdd3195
HL
738 case Opt_whint:
739 name = match_strdup(&args[0]);
740 if (!name)
741 return -ENOMEM;
3c57f751 742 if (!strcmp(name, "user-based")) {
63189b78 743 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_USER;
3c57f751 744 } else if (!strcmp(name, "off")) {
63189b78 745 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
3c57f751 746 } else if (!strcmp(name, "fs-based")) {
63189b78 747 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_FS;
0cdd3195 748 } else {
5222595d 749 kvfree(name);
0cdd3195
HL
750 return -EINVAL;
751 }
5222595d 752 kvfree(name);
0cdd3195 753 break;
07939627
JK
754 case Opt_alloc:
755 name = match_strdup(&args[0]);
756 if (!name)
757 return -ENOMEM;
758
3c57f751 759 if (!strcmp(name, "default")) {
63189b78 760 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
3c57f751 761 } else if (!strcmp(name, "reuse")) {
63189b78 762 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
07939627 763 } else {
5222595d 764 kvfree(name);
07939627
JK
765 return -EINVAL;
766 }
5222595d 767 kvfree(name);
07939627 768 break;
93cf93f1
JZ
769 case Opt_fsync:
770 name = match_strdup(&args[0]);
771 if (!name)
772 return -ENOMEM;
3c57f751 773 if (!strcmp(name, "posix")) {
63189b78 774 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
3c57f751 775 } else if (!strcmp(name, "strict")) {
63189b78 776 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_STRICT;
3c57f751 777 } else if (!strcmp(name, "nobarrier")) {
d6290814
JK
778 F2FS_OPTION(sbi).fsync_mode =
779 FSYNC_MODE_NOBARRIER;
93cf93f1 780 } else {
5222595d 781 kvfree(name);
93cf93f1
JZ
782 return -EINVAL;
783 }
5222595d 784 kvfree(name);
93cf93f1 785 break;
ff62af20 786 case Opt_test_dummy_encryption:
643fa961 787#ifdef CONFIG_FS_ENCRYPTION
7beb01f7 788 if (!f2fs_sb_has_encrypt(sbi)) {
dcbb4c10 789 f2fs_err(sbi, "Encrypt feature is off");
ff62af20
SY
790 return -EINVAL;
791 }
792
793 F2FS_OPTION(sbi).test_dummy_encryption = true;
dcbb4c10 794 f2fs_info(sbi, "Test dummy encryption mode enabled");
ff62af20 795#else
dcbb4c10 796 f2fs_info(sbi, "Test dummy encryption mount option ignored");
ff62af20
SY
797#endif
798 break;
4d3aed70
DR
799 case Opt_checkpoint_disable_cap_perc:
800 if (args->from && match_int(args, &arg))
4354994f 801 return -EINVAL;
4d3aed70
DR
802 if (arg < 0 || arg > 100)
803 return -EINVAL;
1ae18f71 804 F2FS_OPTION(sbi).unusable_cap_perc = arg;
4d3aed70
DR
805 set_opt(sbi, DISABLE_CHECKPOINT);
806 break;
807 case Opt_checkpoint_disable_cap:
808 if (args->from && match_int(args, &arg))
809 return -EINVAL;
810 F2FS_OPTION(sbi).unusable_cap = arg;
811 set_opt(sbi, DISABLE_CHECKPOINT);
812 break;
813 case Opt_checkpoint_disable:
814 set_opt(sbi, DISABLE_CHECKPOINT);
815 break;
816 case Opt_checkpoint_enable:
817 clear_opt(sbi, DISABLE_CHECKPOINT);
4354994f 818 break;
4c8ff709
CY
819 case Opt_compress_algorithm:
820 if (!f2fs_sb_has_compression(sbi)) {
821 f2fs_err(sbi, "Compression feature if off");
822 return -EINVAL;
823 }
824 name = match_strdup(&args[0]);
825 if (!name)
826 return -ENOMEM;
3c57f751 827 if (!strcmp(name, "lzo")) {
4c8ff709
CY
828 F2FS_OPTION(sbi).compress_algorithm =
829 COMPRESS_LZO;
3c57f751 830 } else if (!strcmp(name, "lz4")) {
4c8ff709
CY
831 F2FS_OPTION(sbi).compress_algorithm =
832 COMPRESS_LZ4;
3c57f751 833 } else if (!strcmp(name, "zstd")) {
50cfa66f
CY
834 F2FS_OPTION(sbi).compress_algorithm =
835 COMPRESS_ZSTD;
6d92b201
CY
836 } else if (!strcmp(name, "lzo-rle")) {
837 F2FS_OPTION(sbi).compress_algorithm =
838 COMPRESS_LZORLE;
4c8ff709
CY
839 } else {
840 kfree(name);
841 return -EINVAL;
842 }
843 kfree(name);
844 break;
845 case Opt_compress_log_size:
846 if (!f2fs_sb_has_compression(sbi)) {
847 f2fs_err(sbi, "Compression feature is off");
848 return -EINVAL;
849 }
850 if (args->from && match_int(args, &arg))
851 return -EINVAL;
852 if (arg < MIN_COMPRESS_LOG_SIZE ||
853 arg > MAX_COMPRESS_LOG_SIZE) {
854 f2fs_err(sbi,
855 "Compress cluster log size is out of range");
856 return -EINVAL;
857 }
858 F2FS_OPTION(sbi).compress_log_size = arg;
859 break;
860 case Opt_compress_extension:
861 if (!f2fs_sb_has_compression(sbi)) {
862 f2fs_err(sbi, "Compression feature is off");
863 return -EINVAL;
864 }
865 name = match_strdup(&args[0]);
866 if (!name)
867 return -ENOMEM;
868
869 ext = F2FS_OPTION(sbi).extensions;
870 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
871
872 if (strlen(name) >= F2FS_EXTENSION_LEN ||
873 ext_cnt >= COMPRESS_EXT_NUM) {
874 f2fs_err(sbi,
875 "invalid extension length/number");
876 kfree(name);
877 return -EINVAL;
878 }
879
880 strcpy(ext[ext_cnt], name);
881 F2FS_OPTION(sbi).compress_ext_cnt++;
882 kfree(name);
883 break;
696c018c 884 default:
dcbb4c10
JP
885 f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
886 p);
696c018c
NJ
887 return -EINVAL;
888 }
889 }
4b2414d0
CY
890#ifdef CONFIG_QUOTA
891 if (f2fs_check_quota_options(sbi))
892 return -EINVAL;
00960c2c 893#else
7beb01f7 894 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sbi->sb)) {
dcbb4c10 895 f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
00960c2c
SY
896 return -EINVAL;
897 }
7beb01f7 898 if (f2fs_sb_has_project_quota(sbi) && !f2fs_readonly(sbi->sb)) {
dcbb4c10 899 f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
4ddc1b28
CY
900 return -EINVAL;
901 }
4b2414d0 902#endif
5aba5430
DR
903#ifndef CONFIG_UNICODE
904 if (f2fs_sb_has_casefold(sbi)) {
905 f2fs_err(sbi,
906 "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
907 return -EINVAL;
908 }
909#endif
ec91538d 910
b0332a0f 911 if (F2FS_IO_SIZE_BITS(sbi) && !f2fs_lfs_mode(sbi)) {
dcbb4c10
JP
912 f2fs_err(sbi, "Should set mode=lfs with %uKB-sized IO",
913 F2FS_IO_SIZE_KB(sbi));
ec91538d
JK
914 return -EINVAL;
915 }
6afc662e
CY
916
917 if (test_opt(sbi, INLINE_XATTR_SIZE)) {
70db5b04
JK
918 int min_size, max_size;
919
7beb01f7
CY
920 if (!f2fs_sb_has_extra_attr(sbi) ||
921 !f2fs_sb_has_flexible_inline_xattr(sbi)) {
dcbb4c10 922 f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
4d817ae0
CY
923 return -EINVAL;
924 }
6afc662e 925 if (!test_opt(sbi, INLINE_XATTR)) {
dcbb4c10 926 f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
6afc662e
CY
927 return -EINVAL;
928 }
70db5b04
JK
929
930 min_size = sizeof(struct f2fs_xattr_header) / sizeof(__le32);
dd6c89b5 931 max_size = MAX_INLINE_XATTR_SIZE;
70db5b04
JK
932
933 if (F2FS_OPTION(sbi).inline_xattr_size < min_size ||
934 F2FS_OPTION(sbi).inline_xattr_size > max_size) {
dcbb4c10
JP
935 f2fs_err(sbi, "inline xattr size is out of range: %d ~ %d",
936 min_size, max_size);
6afc662e
CY
937 return -EINVAL;
938 }
939 }
0cdd3195 940
b0332a0f 941 if (test_opt(sbi, DISABLE_CHECKPOINT) && f2fs_lfs_mode(sbi)) {
dcbb4c10 942 f2fs_err(sbi, "LFS not compatible with checkpoint=disable\n");
4354994f
DR
943 return -EINVAL;
944 }
945
0cdd3195
HL
946 /* Not pass down write hints if the number of active logs is lesser
947 * than NR_CURSEG_TYPE.
948 */
63189b78
CY
949 if (F2FS_OPTION(sbi).active_logs != NR_CURSEG_TYPE)
950 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
696c018c
NJ
951 return 0;
952}
953
aff063e2
JK
954static struct inode *f2fs_alloc_inode(struct super_block *sb)
955{
956 struct f2fs_inode_info *fi;
957
a0acdfe0 958 fi = kmem_cache_alloc(f2fs_inode_cachep, GFP_F2FS_ZERO);
aff063e2
JK
959 if (!fi)
960 return NULL;
961
962 init_once((void *) fi);
963
434720fa 964 /* Initialize f2fs-specific inode info */
204706c7 965 atomic_set(&fi->dirty_pages, 0);
d928bfbf 966 init_rwsem(&fi->i_sem);
c10c9820 967 spin_lock_init(&fi->i_size_lock);
2710fd7e 968 INIT_LIST_HEAD(&fi->dirty_list);
0f18b462 969 INIT_LIST_HEAD(&fi->gdirty_list);
57864ae5 970 INIT_LIST_HEAD(&fi->inmem_ilist);
88b88a66
JK
971 INIT_LIST_HEAD(&fi->inmem_pages);
972 mutex_init(&fi->inmem_lock);
b2532c69
CY
973 init_rwsem(&fi->i_gc_rwsem[READ]);
974 init_rwsem(&fi->i_gc_rwsem[WRITE]);
5a3a2d83 975 init_rwsem(&fi->i_mmap_sem);
27161f13 976 init_rwsem(&fi->i_xattr_sem);
aff063e2 977
ab9fa662
JK
978 /* Will be used by directory only */
979 fi->i_dir_level = F2FS_SB(sb)->dir_level;
f2470371 980
aff063e2
JK
981 return &fi->vfs_inode;
982}
983
531ad7d5
JK
984static int f2fs_drop_inode(struct inode *inode)
985{
a8933b6b 986 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
b8d96a30 987 int ret;
a8933b6b
CY
988
989 /*
990 * during filesystem shutdown, if checkpoint is disabled,
991 * drop useless meta/node dirty pages.
992 */
993 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
994 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
995 inode->i_ino == F2FS_META_INO(sbi)) {
996 trace_f2fs_drop_inode(inode, 1);
997 return 1;
998 }
999 }
1000
531ad7d5
JK
1001 /*
1002 * This is to avoid a deadlock condition like below.
1003 * writeback_single_inode(inode)
1004 * - f2fs_write_data_page
1005 * - f2fs_gc -> iput -> evict
1006 * - inode_wait_for_writeback(inode)
1007 */
0f18b462 1008 if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
06e1bc05 1009 if (!inode->i_nlink && !is_bad_inode(inode)) {
3e72f721
JK
1010 /* to avoid evict_inode call simultaneously */
1011 atomic_inc(&inode->i_count);
06e1bc05
JK
1012 spin_unlock(&inode->i_lock);
1013
1014 /* some remained atomic pages should discarded */
1015 if (f2fs_is_atomic_file(inode))
4d57b86d 1016 f2fs_drop_inmem_pages(inode);
06e1bc05 1017
3e72f721
JK
1018 /* should remain fi->extent_tree for writepage */
1019 f2fs_destroy_extent_node(inode);
1020
06e1bc05 1021 sb_start_intwrite(inode->i_sb);
fc9581c8 1022 f2fs_i_size_write(inode, 0);
06e1bc05 1023
a0770e13 1024 f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1025 inode, NULL, 0, DATA);
1026 truncate_inode_pages_final(inode->i_mapping);
1027
06e1bc05 1028 if (F2FS_HAS_BLOCKS(inode))
9a449e9c 1029 f2fs_truncate(inode);
06e1bc05
JK
1030
1031 sb_end_intwrite(inode->i_sb);
1032
06e1bc05 1033 spin_lock(&inode->i_lock);
3e72f721 1034 atomic_dec(&inode->i_count);
06e1bc05 1035 }
b8d96a30 1036 trace_f2fs_drop_inode(inode, 0);
531ad7d5 1037 return 0;
06e1bc05 1038 }
b8d96a30 1039 ret = generic_drop_inode(inode);
8ce589c7
EB
1040 if (!ret)
1041 ret = fscrypt_drop_inode(inode);
b8d96a30
HP
1042 trace_f2fs_drop_inode(inode, ret);
1043 return ret;
531ad7d5
JK
1044}
1045
7c45729a 1046int f2fs_inode_dirtied(struct inode *inode, bool sync)
b3783873 1047{
0f18b462 1048 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
7c45729a 1049 int ret = 0;
0f18b462 1050
0f18b462
JK
1051 spin_lock(&sbi->inode_lock[DIRTY_META]);
1052 if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
7c45729a
JK
1053 ret = 1;
1054 } else {
1055 set_inode_flag(inode, FI_DIRTY_INODE);
1056 stat_inc_dirty_inode(sbi, DIRTY_META);
0f18b462 1057 }
7c45729a
JK
1058 if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1059 list_add_tail(&F2FS_I(inode)->gdirty_list,
0f18b462 1060 &sbi->inode_list[DIRTY_META]);
7c45729a
JK
1061 inc_page_count(sbi, F2FS_DIRTY_IMETA);
1062 }
338bbfa0 1063 spin_unlock(&sbi->inode_lock[DIRTY_META]);
7c45729a 1064 return ret;
0f18b462
JK
1065}
1066
1067void f2fs_inode_synced(struct inode *inode)
1068{
1069 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1070
1071 spin_lock(&sbi->inode_lock[DIRTY_META]);
1072 if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1073 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1074 return;
1075 }
7c45729a
JK
1076 if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1077 list_del_init(&F2FS_I(inode)->gdirty_list);
1078 dec_page_count(sbi, F2FS_DIRTY_IMETA);
1079 }
0f18b462 1080 clear_inode_flag(inode, FI_DIRTY_INODE);
26de9b11 1081 clear_inode_flag(inode, FI_AUTO_RECOVER);
0f18b462 1082 stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
338bbfa0 1083 spin_unlock(&sbi->inode_lock[DIRTY_META]);
b3783873
JK
1084}
1085
b56ab837
JK
1086/*
1087 * f2fs_dirty_inode() is called from __mark_inode_dirty()
1088 *
1089 * We should call set_dirty_inode to write the dirty inode through write_inode.
1090 */
1091static void f2fs_dirty_inode(struct inode *inode, int flags)
1092{
1093 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1094
1095 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1096 inode->i_ino == F2FS_META_INO(sbi))
1097 return;
1098
1099 if (flags == I_DIRTY_TIME)
1100 return;
1101
1102 if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1103 clear_inode_flag(inode, FI_AUTO_RECOVER);
1104
7c45729a 1105 f2fs_inode_dirtied(inode, false);
b56ab837
JK
1106}
1107
d01718a0 1108static void f2fs_free_inode(struct inode *inode)
aff063e2 1109{
2c58d548 1110 fscrypt_free_inode(inode);
aff063e2
JK
1111 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1112}
1113
523be8a6
JK
1114static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1115{
41382ec4 1116 percpu_counter_destroy(&sbi->alloc_valid_block_count);
513c5f37 1117 percpu_counter_destroy(&sbi->total_valid_inode_count);
523be8a6
JK
1118}
1119
3c62be17
JK
1120static void destroy_device_list(struct f2fs_sb_info *sbi)
1121{
1122 int i;
1123
1124 for (i = 0; i < sbi->s_ndevs; i++) {
1125 blkdev_put(FDEV(i).bdev, FMODE_EXCL);
1126#ifdef CONFIG_BLK_DEV_ZONED
95175daf 1127 kvfree(FDEV(i).blkz_seq);
3c62be17
JK
1128#endif
1129 }
5222595d 1130 kvfree(sbi->devs);
3c62be17
JK
1131}
1132
aff063e2
JK
1133static void f2fs_put_super(struct super_block *sb)
1134{
1135 struct f2fs_sb_info *sbi = F2FS_SB(sb);
a398101a 1136 int i;
cf5c759f 1137 bool dropped;
aff063e2 1138
0abd675e 1139 f2fs_quota_off_umount(sb);
aff063e2 1140
2658e50d
JK
1141 /* prevent remaining shrinker jobs */
1142 mutex_lock(&sbi->umount_mutex);
1143
85dc2f2c
JK
1144 /*
1145 * We don't need to do checkpoint when superblock is clean.
1146 * But, the previous checkpoint was not done by umount, it needs to do
1147 * clean checkpoint again.
1148 */
4354994f
DR
1149 if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1150 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
75ab4cb8
JK
1151 struct cp_control cpc = {
1152 .reason = CP_UMOUNT,
1153 };
4d57b86d 1154 f2fs_write_checkpoint(sbi, &cpc);
75ab4cb8 1155 }
aff063e2 1156
4e6a8d9b 1157 /* be sure to wait for any on-going discard commands */
03f2c02d 1158 dropped = f2fs_issue_discard_timeout(sbi);
4e6a8d9b 1159
7d20c8ab
CY
1160 if ((f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi)) &&
1161 !sbi->discard_blks && !dropped) {
1f43e2ad
CY
1162 struct cp_control cpc = {
1163 .reason = CP_UMOUNT | CP_TRIMMED,
1164 };
4d57b86d 1165 f2fs_write_checkpoint(sbi, &cpc);
1f43e2ad
CY
1166 }
1167
cf779cab
JK
1168 /*
1169 * normally superblock is clean, so we need to release this.
1170 * In addition, EIO will skip do checkpoint, we need this as well.
1171 */
4d57b86d 1172 f2fs_release_ino_entry(sbi, true);
6f12ac25 1173
2658e50d
JK
1174 f2fs_leave_shrinker(sbi);
1175 mutex_unlock(&sbi->umount_mutex);
1176
17c19120 1177 /* our cp_error case, we can wait for any writeback page */
b9109b0e 1178 f2fs_flush_merged_writes(sbi);
17c19120 1179
bf22c3cc 1180 f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
50fa53ec
CY
1181
1182 f2fs_bug_on(sbi, sbi->fsync_node_num);
1183
aff063e2 1184 iput(sbi->node_inode);
7c77bf7d
JK
1185 sbi->node_inode = NULL;
1186
aff063e2 1187 iput(sbi->meta_inode);
7c77bf7d 1188 sbi->meta_inode = NULL;
aff063e2 1189
60aa4d55
ST
1190 /*
1191 * iput() can update stat information, if f2fs_write_checkpoint()
1192 * above failed with error.
1193 */
1194 f2fs_destroy_stats(sbi);
1195
aff063e2 1196 /* destroy f2fs internal modules */
4d57b86d
CY
1197 f2fs_destroy_node_manager(sbi);
1198 f2fs_destroy_segment_manager(sbi);
aff063e2 1199
4c8ff709
CY
1200 f2fs_destroy_post_read_wq(sbi);
1201
5222595d 1202 kvfree(sbi->ckpt);
a398101a 1203
dc6b2055 1204 f2fs_unregister_sysfs(sbi);
aff063e2
JK
1205
1206 sb->s_fs_info = NULL;
43b6573b
KM
1207 if (sbi->s_chksum_driver)
1208 crypto_free_shash(sbi->s_chksum_driver);
5222595d 1209 kvfree(sbi->raw_super);
523be8a6 1210
3c62be17 1211 destroy_device_list(sbi);
a999150f 1212 f2fs_destroy_xattr_caches(sbi);
b6895e8f 1213 mempool_destroy(sbi->write_io_dummy);
4b2414d0
CY
1214#ifdef CONFIG_QUOTA
1215 for (i = 0; i < MAXQUOTAS; i++)
5222595d 1216 kvfree(F2FS_OPTION(sbi).s_qf_names[i]);
4b2414d0 1217#endif
523be8a6 1218 destroy_percpu_info(sbi);
a912b54d 1219 for (i = 0; i < NR_PAGE_TYPE; i++)
5222595d 1220 kvfree(sbi->write_io[i]);
5aba5430
DR
1221#ifdef CONFIG_UNICODE
1222 utf8_unload(sbi->s_encoding);
1223#endif
5222595d 1224 kvfree(sbi);
aff063e2
JK
1225}
1226
1227int f2fs_sync_fs(struct super_block *sb, int sync)
1228{
1229 struct f2fs_sb_info *sbi = F2FS_SB(sb);
c34f42e2 1230 int err = 0;
aff063e2 1231
1f227a3e
JK
1232 if (unlikely(f2fs_cp_error(sbi)))
1233 return 0;
4354994f
DR
1234 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1235 return 0;
1f227a3e 1236
a2a4a7e4
NJ
1237 trace_f2fs_sync_fs(sb, sync);
1238
4b2414d0
CY
1239 if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1240 return -EAGAIN;
1241
b7473754 1242 if (sync) {
d5053a34
JK
1243 struct cp_control cpc;
1244
119ee914
JK
1245 cpc.reason = __get_cp_reason(sbi);
1246
fb24fea7 1247 down_write(&sbi->gc_lock);
4d57b86d 1248 err = f2fs_write_checkpoint(sbi, &cpc);
fb24fea7 1249 up_write(&sbi->gc_lock);
b7473754 1250 }
05ca3632 1251 f2fs_trace_ios(NULL, 1);
aff063e2 1252
c34f42e2 1253 return err;
aff063e2
JK
1254}
1255
d6212a5f
CL
1256static int f2fs_freeze(struct super_block *sb)
1257{
77888c1e 1258 if (f2fs_readonly(sb))
d6212a5f
CL
1259 return 0;
1260
b4b9d34c
JK
1261 /* IO error happened before */
1262 if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1263 return -EIO;
1264
1265 /* must be clean, since sync_filesystem() was already called */
1266 if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1267 return -EINVAL;
1268 return 0;
d6212a5f
CL
1269}
1270
1271static int f2fs_unfreeze(struct super_block *sb)
1272{
1273 return 0;
1274}
1275
ddc34e32
CY
1276#ifdef CONFIG_QUOTA
1277static int f2fs_statfs_project(struct super_block *sb,
1278 kprojid_t projid, struct kstatfs *buf)
1279{
1280 struct kqid qid;
1281 struct dquot *dquot;
1282 u64 limit;
1283 u64 curblock;
1284
1285 qid = make_kqid_projid(projid);
1286 dquot = dqget(sb, qid);
1287 if (IS_ERR(dquot))
1288 return PTR_ERR(dquot);
955ac6e5 1289 spin_lock(&dquot->dq_dqb_lock);
ddc34e32 1290
bf2cbd3c
CX
1291 limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1292 dquot->dq_dqb.dqb_bhardlimit);
acdf2172
CX
1293 if (limit)
1294 limit >>= sb->s_blocksize_bits;
909110c0 1295
ddc34e32 1296 if (limit && buf->f_blocks > limit) {
baaa7ebf
KK
1297 curblock = (dquot->dq_dqb.dqb_curspace +
1298 dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
ddc34e32
CY
1299 buf->f_blocks = limit;
1300 buf->f_bfree = buf->f_bavail =
1301 (buf->f_blocks > curblock) ?
1302 (buf->f_blocks - curblock) : 0;
1303 }
1304
bf2cbd3c
CX
1305 limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1306 dquot->dq_dqb.dqb_ihardlimit);
909110c0 1307
ddc34e32
CY
1308 if (limit && buf->f_files > limit) {
1309 buf->f_files = limit;
1310 buf->f_ffree =
1311 (buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1312 (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1313 }
1314
955ac6e5 1315 spin_unlock(&dquot->dq_dqb_lock);
ddc34e32
CY
1316 dqput(dquot);
1317 return 0;
1318}
1319#endif
1320
aff063e2
JK
1321static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1322{
1323 struct super_block *sb = dentry->d_sb;
1324 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1325 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
f66c027e 1326 block_t total_count, user_block_count, start_count;
0cc091d0 1327 u64 avail_node_count;
aff063e2
JK
1328
1329 total_count = le64_to_cpu(sbi->raw_super->block_count);
1330 user_block_count = sbi->user_block_count;
1331 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
aff063e2
JK
1332 buf->f_type = F2FS_SUPER_MAGIC;
1333 buf->f_bsize = sbi->blocksize;
1334
1335 buf->f_blocks = total_count - start_count;
f66c027e 1336 buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
80d42145 1337 sbi->current_reserved_blocks;
c9c8ed50
CY
1338
1339 spin_lock(&sbi->stat_lock);
4354994f
DR
1340 if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1341 buf->f_bfree = 0;
1342 else
1343 buf->f_bfree -= sbi->unusable_block_count;
c9c8ed50 1344 spin_unlock(&sbi->stat_lock);
4354994f 1345
63189b78
CY
1346 if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1347 buf->f_bavail = buf->f_bfree -
1348 F2FS_OPTION(sbi).root_reserved_blocks;
7e65be49
JK
1349 else
1350 buf->f_bavail = 0;
aff063e2 1351
27cae0bc 1352 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
0cc091d0
JK
1353
1354 if (avail_node_count > user_block_count) {
1355 buf->f_files = user_block_count;
1356 buf->f_ffree = buf->f_bavail;
1357 } else {
1358 buf->f_files = avail_node_count;
1359 buf->f_ffree = min(avail_node_count - valid_node_count(sbi),
1360 buf->f_bavail);
1361 }
aff063e2 1362
5a20d339 1363 buf->f_namelen = F2FS_NAME_LEN;
aff063e2
JK
1364 buf->f_fsid.val[0] = (u32)id;
1365 buf->f_fsid.val[1] = (u32)(id >> 32);
1366
ddc34e32
CY
1367#ifdef CONFIG_QUOTA
1368 if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1369 sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1370 f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1371 }
1372#endif
aff063e2
JK
1373 return 0;
1374}
1375
4b2414d0
CY
1376static inline void f2fs_show_quota_options(struct seq_file *seq,
1377 struct super_block *sb)
1378{
1379#ifdef CONFIG_QUOTA
1380 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1381
63189b78 1382 if (F2FS_OPTION(sbi).s_jquota_fmt) {
4b2414d0
CY
1383 char *fmtname = "";
1384
63189b78 1385 switch (F2FS_OPTION(sbi).s_jquota_fmt) {
4b2414d0
CY
1386 case QFMT_VFS_OLD:
1387 fmtname = "vfsold";
1388 break;
1389 case QFMT_VFS_V0:
1390 fmtname = "vfsv0";
1391 break;
1392 case QFMT_VFS_V1:
1393 fmtname = "vfsv1";
1394 break;
1395 }
1396 seq_printf(seq, ",jqfmt=%s", fmtname);
1397 }
1398
63189b78
CY
1399 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1400 seq_show_option(seq, "usrjquota",
1401 F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
4b2414d0 1402
63189b78
CY
1403 if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1404 seq_show_option(seq, "grpjquota",
1405 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
4b2414d0 1406
63189b78
CY
1407 if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1408 seq_show_option(seq, "prjjquota",
1409 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
4b2414d0
CY
1410#endif
1411}
1412
4c8ff709
CY
1413static inline void f2fs_show_compress_options(struct seq_file *seq,
1414 struct super_block *sb)
1415{
1416 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1417 char *algtype = "";
1418 int i;
1419
1420 if (!f2fs_sb_has_compression(sbi))
1421 return;
1422
1423 switch (F2FS_OPTION(sbi).compress_algorithm) {
1424 case COMPRESS_LZO:
1425 algtype = "lzo";
1426 break;
1427 case COMPRESS_LZ4:
1428 algtype = "lz4";
1429 break;
50cfa66f
CY
1430 case COMPRESS_ZSTD:
1431 algtype = "zstd";
1432 break;
6d92b201
CY
1433 case COMPRESS_LZORLE:
1434 algtype = "lzo-rle";
1435 break;
4c8ff709
CY
1436 }
1437 seq_printf(seq, ",compress_algorithm=%s", algtype);
1438
1439 seq_printf(seq, ",compress_log_size=%u",
1440 F2FS_OPTION(sbi).compress_log_size);
1441
1442 for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1443 seq_printf(seq, ",compress_extension=%s",
1444 F2FS_OPTION(sbi).extensions[i]);
1445 }
1446}
1447
aff063e2
JK
1448static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1449{
1450 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1451
bbbc34fd
CY
1452 if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1453 seq_printf(seq, ",background_gc=%s", "sync");
1454 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1455 seq_printf(seq, ",background_gc=%s", "on");
1456 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
696c018c 1457 seq_printf(seq, ",background_gc=%s", "off");
bbbc34fd 1458
aff063e2
JK
1459 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1460 seq_puts(seq, ",disable_roll_forward");
a9117eca
CY
1461 if (test_opt(sbi, NORECOVERY))
1462 seq_puts(seq, ",norecovery");
aff063e2
JK
1463 if (test_opt(sbi, DISCARD))
1464 seq_puts(seq, ",discard");
81621f97
ST
1465 else
1466 seq_puts(seq, ",nodiscard");
aff063e2 1467 if (test_opt(sbi, NOHEAP))
7a20b8a6
JK
1468 seq_puts(seq, ",no_heap");
1469 else
1470 seq_puts(seq, ",heap");
aff063e2
JK
1471#ifdef CONFIG_F2FS_FS_XATTR
1472 if (test_opt(sbi, XATTR_USER))
1473 seq_puts(seq, ",user_xattr");
1474 else
1475 seq_puts(seq, ",nouser_xattr");
444c580f
JK
1476 if (test_opt(sbi, INLINE_XATTR))
1477 seq_puts(seq, ",inline_xattr");
23cf7212
CY
1478 else
1479 seq_puts(seq, ",noinline_xattr");
6afc662e
CY
1480 if (test_opt(sbi, INLINE_XATTR_SIZE))
1481 seq_printf(seq, ",inline_xattr_size=%u",
63189b78 1482 F2FS_OPTION(sbi).inline_xattr_size);
aff063e2
JK
1483#endif
1484#ifdef CONFIG_F2FS_FS_POSIX_ACL
1485 if (test_opt(sbi, POSIX_ACL))
1486 seq_puts(seq, ",acl");
1487 else
1488 seq_puts(seq, ",noacl");
1489#endif
1490 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
aa43507f 1491 seq_puts(seq, ",disable_ext_identify");
8274de77
HL
1492 if (test_opt(sbi, INLINE_DATA))
1493 seq_puts(seq, ",inline_data");
75342797
WL
1494 else
1495 seq_puts(seq, ",noinline_data");
5efd3c6f
CY
1496 if (test_opt(sbi, INLINE_DENTRY))
1497 seq_puts(seq, ",inline_dentry");
97c1794a
CY
1498 else
1499 seq_puts(seq, ",noinline_dentry");
b270ad6f 1500 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
6b4afdd7 1501 seq_puts(seq, ",flush_merge");
0f7b2abd
JK
1502 if (test_opt(sbi, NOBARRIER))
1503 seq_puts(seq, ",nobarrier");
d5053a34
JK
1504 if (test_opt(sbi, FASTBOOT))
1505 seq_puts(seq, ",fastboot");
89672159
CY
1506 if (test_opt(sbi, EXTENT_CACHE))
1507 seq_puts(seq, ",extent_cache");
7daaea25
JK
1508 else
1509 seq_puts(seq, ",noextent_cache");
343f40f0
CY
1510 if (test_opt(sbi, DATA_FLUSH))
1511 seq_puts(seq, ",data_flush");
36abef4e
JK
1512
1513 seq_puts(seq, ",mode=");
b0332a0f 1514 if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
36abef4e 1515 seq_puts(seq, "adaptive");
b0332a0f 1516 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
36abef4e 1517 seq_puts(seq, "lfs");
63189b78 1518 seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
7e65be49 1519 if (test_opt(sbi, RESERVE_ROOT))
7c2e5963 1520 seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
63189b78
CY
1521 F2FS_OPTION(sbi).root_reserved_blocks,
1522 from_kuid_munged(&init_user_ns,
1523 F2FS_OPTION(sbi).s_resuid),
1524 from_kgid_munged(&init_user_ns,
1525 F2FS_OPTION(sbi).s_resgid));
ec91538d 1526 if (F2FS_IO_SIZE_BITS(sbi))
c6b1867b
CX
1527 seq_printf(seq, ",io_bits=%u",
1528 F2FS_OPTION(sbi).write_io_size_bits);
0cc0dec2 1529#ifdef CONFIG_F2FS_FAULT_INJECTION
d494500a 1530 if (test_opt(sbi, FAULT_INJECTION)) {
44529f89 1531 seq_printf(seq, ",fault_injection=%u",
63189b78 1532 F2FS_OPTION(sbi).fault_info.inject_rate);
d494500a
CY
1533 seq_printf(seq, ",fault_type=%u",
1534 F2FS_OPTION(sbi).fault_info.inject_type);
1535 }
0cc0dec2 1536#endif
0abd675e 1537#ifdef CONFIG_QUOTA
4b2414d0
CY
1538 if (test_opt(sbi, QUOTA))
1539 seq_puts(seq, ",quota");
0abd675e
CY
1540 if (test_opt(sbi, USRQUOTA))
1541 seq_puts(seq, ",usrquota");
1542 if (test_opt(sbi, GRPQUOTA))
1543 seq_puts(seq, ",grpquota");
5c57132e
CY
1544 if (test_opt(sbi, PRJQUOTA))
1545 seq_puts(seq, ",prjquota");
0cc0dec2 1546#endif
4b2414d0 1547 f2fs_show_quota_options(seq, sbi->sb);
63189b78 1548 if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_USER)
0cdd3195 1549 seq_printf(seq, ",whint_mode=%s", "user-based");
63189b78 1550 else if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_FS)
f2e703f9 1551 seq_printf(seq, ",whint_mode=%s", "fs-based");
643fa961 1552#ifdef CONFIG_FS_ENCRYPTION
ff62af20
SY
1553 if (F2FS_OPTION(sbi).test_dummy_encryption)
1554 seq_puts(seq, ",test_dummy_encryption");
1555#endif
aff063e2 1556
63189b78 1557 if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
07939627 1558 seq_printf(seq, ",alloc_mode=%s", "default");
63189b78 1559 else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
07939627 1560 seq_printf(seq, ",alloc_mode=%s", "reuse");
93cf93f1 1561
4354994f 1562 if (test_opt(sbi, DISABLE_CHECKPOINT))
4d3aed70
DR
1563 seq_printf(seq, ",checkpoint=disable:%u",
1564 F2FS_OPTION(sbi).unusable_cap);
63189b78 1565 if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
93cf93f1 1566 seq_printf(seq, ",fsync_mode=%s", "posix");
63189b78 1567 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
93cf93f1 1568 seq_printf(seq, ",fsync_mode=%s", "strict");
dc132802
ST
1569 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
1570 seq_printf(seq, ",fsync_mode=%s", "nobarrier");
4c8ff709
CY
1571
1572 f2fs_show_compress_options(seq, sbi->sb);
aff063e2
JK
1573 return 0;
1574}
1575
498c5e9f
YH
1576static void default_options(struct f2fs_sb_info *sbi)
1577{
1578 /* init some FS parameters */
63189b78
CY
1579 F2FS_OPTION(sbi).active_logs = NR_CURSEG_TYPE;
1580 F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
1581 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1582 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
1583 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
ff62af20 1584 F2FS_OPTION(sbi).test_dummy_encryption = false;
0aa7e0f8
CY
1585 F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
1586 F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
91faa534 1587 F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
4c8ff709
CY
1588 F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
1589 F2FS_OPTION(sbi).compress_ext_cnt = 0;
bbbc34fd 1590 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
498c5e9f 1591
39133a50 1592 set_opt(sbi, INLINE_XATTR);
498c5e9f 1593 set_opt(sbi, INLINE_DATA);
97c1794a 1594 set_opt(sbi, INLINE_DENTRY);
3e72f721 1595 set_opt(sbi, EXTENT_CACHE);
7a20b8a6 1596 set_opt(sbi, NOHEAP);
4354994f 1597 clear_opt(sbi, DISABLE_CHECKPOINT);
4d3aed70 1598 F2FS_OPTION(sbi).unusable_cap = 0;
1751e8a6 1599 sbi->sb->s_flags |= SB_LAZYTIME;
69e9e427 1600 set_opt(sbi, FLUSH_MERGE);
7d20c8ab 1601 set_opt(sbi, DISCARD);
7beb01f7 1602 if (f2fs_sb_has_blkzoned(sbi))
b0332a0f 1603 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
a39e5365 1604 else
b0332a0f 1605 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
498c5e9f
YH
1606
1607#ifdef CONFIG_F2FS_FS_XATTR
1608 set_opt(sbi, XATTR_USER);
1609#endif
1610#ifdef CONFIG_F2FS_FS_POSIX_ACL
1611 set_opt(sbi, POSIX_ACL);
1612#endif
36dbd328 1613
d494500a 1614 f2fs_build_fault_attr(sbi, 0, 0);
498c5e9f
YH
1615}
1616
ea676733
JK
1617#ifdef CONFIG_QUOTA
1618static int f2fs_enable_quotas(struct super_block *sb);
1619#endif
4354994f
DR
1620
1621static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
1622{
812a9597 1623 unsigned int s_flags = sbi->sb->s_flags;
4354994f 1624 struct cp_control cpc;
812a9597
JK
1625 int err = 0;
1626 int ret;
4d3aed70 1627 block_t unusable;
4354994f 1628
812a9597 1629 if (s_flags & SB_RDONLY) {
dcbb4c10 1630 f2fs_err(sbi, "checkpoint=disable on readonly fs");
812a9597
JK
1631 return -EINVAL;
1632 }
4354994f
DR
1633 sbi->sb->s_flags |= SB_ACTIVE;
1634
4354994f
DR
1635 f2fs_update_time(sbi, DISABLE_TIME);
1636
1637 while (!f2fs_time_over(sbi, DISABLE_TIME)) {
fb24fea7 1638 down_write(&sbi->gc_lock);
4354994f 1639 err = f2fs_gc(sbi, true, false, NULL_SEGNO);
812a9597
JK
1640 if (err == -ENODATA) {
1641 err = 0;
4354994f 1642 break;
812a9597 1643 }
8f31b466 1644 if (err && err != -EAGAIN)
812a9597 1645 break;
4354994f 1646 }
4354994f 1647
812a9597
JK
1648 ret = sync_filesystem(sbi->sb);
1649 if (ret || err) {
1650 err = ret ? ret: err;
1651 goto restore_flag;
1652 }
4354994f 1653
4d3aed70
DR
1654 unusable = f2fs_get_unusable_blocks(sbi);
1655 if (f2fs_disable_cp_again(sbi, unusable)) {
812a9597
JK
1656 err = -EAGAIN;
1657 goto restore_flag;
1658 }
4354994f 1659
fb24fea7 1660 down_write(&sbi->gc_lock);
4354994f
DR
1661 cpc.reason = CP_PAUSE;
1662 set_sbi_flag(sbi, SBI_CP_DISABLED);
896285ad
CY
1663 err = f2fs_write_checkpoint(sbi, &cpc);
1664 if (err)
1665 goto out_unlock;
4354994f 1666
c9c8ed50 1667 spin_lock(&sbi->stat_lock);
4d3aed70 1668 sbi->unusable_block_count = unusable;
c9c8ed50
CY
1669 spin_unlock(&sbi->stat_lock);
1670
896285ad 1671out_unlock:
fb24fea7 1672 up_write(&sbi->gc_lock);
812a9597 1673restore_flag:
7a88ddb5 1674 sbi->sb->s_flags = s_flags; /* Restore SB_RDONLY status */
812a9597 1675 return err;
4354994f
DR
1676}
1677
1678static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
1679{
fb24fea7 1680 down_write(&sbi->gc_lock);
4354994f
DR
1681 f2fs_dirty_to_prefree(sbi);
1682
1683 clear_sbi_flag(sbi, SBI_CP_DISABLED);
1684 set_sbi_flag(sbi, SBI_IS_DIRTY);
fb24fea7 1685 up_write(&sbi->gc_lock);
4354994f
DR
1686
1687 f2fs_sync_fs(sbi->sb, 1);
1688}
1689
696c018c
NJ
1690static int f2fs_remount(struct super_block *sb, int *flags, char *data)
1691{
1692 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1693 struct f2fs_mount_info org_mount_opt;
0abd675e 1694 unsigned long old_sb_flags;
63189b78 1695 int err;
876dc59e
GZ
1696 bool need_restart_gc = false;
1697 bool need_stop_gc = false;
9cd81ce3 1698 bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
4354994f 1699 bool disable_checkpoint = test_opt(sbi, DISABLE_CHECKPOINT);
1f78adfa 1700 bool no_io_align = !F2FS_IO_ALIGNED(sbi);
4354994f 1701 bool checkpoint_changed;
4b2414d0 1702#ifdef CONFIG_QUOTA
4b2414d0
CY
1703 int i, j;
1704#endif
696c018c
NJ
1705
1706 /*
1707 * Save the old mount options in case we
1708 * need to restore them.
1709 */
1710 org_mount_opt = sbi->mount_opt;
0abd675e 1711 old_sb_flags = sb->s_flags;
696c018c 1712
4b2414d0 1713#ifdef CONFIG_QUOTA
63189b78 1714 org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
4b2414d0 1715 for (i = 0; i < MAXQUOTAS; i++) {
63189b78
CY
1716 if (F2FS_OPTION(sbi).s_qf_names[i]) {
1717 org_mount_opt.s_qf_names[i] =
1718 kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
1719 GFP_KERNEL);
1720 if (!org_mount_opt.s_qf_names[i]) {
4b2414d0 1721 for (j = 0; j < i; j++)
5222595d 1722 kvfree(org_mount_opt.s_qf_names[j]);
4b2414d0
CY
1723 return -ENOMEM;
1724 }
1725 } else {
63189b78 1726 org_mount_opt.s_qf_names[i] = NULL;
4b2414d0
CY
1727 }
1728 }
1729#endif
1730
df728b0f 1731 /* recover superblocks we couldn't write due to previous RO mount */
1751e8a6 1732 if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
df728b0f 1733 err = f2fs_commit_super(sbi, false);
dcbb4c10
JP
1734 f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
1735 err);
df728b0f
JK
1736 if (!err)
1737 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
1738 }
1739
498c5e9f 1740 default_options(sbi);
26666c8a 1741
696c018c
NJ
1742 /* parse mount options */
1743 err = parse_options(sb, data);
1744 if (err)
1745 goto restore_opts;
4354994f
DR
1746 checkpoint_changed =
1747 disable_checkpoint != test_opt(sbi, DISABLE_CHECKPOINT);
696c018c
NJ
1748
1749 /*
1750 * Previous and new state of filesystem is RO,
876dc59e 1751 * so skip checking GC and FLUSH_MERGE conditions.
696c018c 1752 */
1751e8a6 1753 if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
696c018c
NJ
1754 goto skip;
1755
ea676733 1756#ifdef CONFIG_QUOTA
1751e8a6 1757 if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
0abd675e
CY
1758 err = dquot_suspend(sb, -1);
1759 if (err < 0)
1760 goto restore_opts;
4354994f 1761 } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
0abd675e 1762 /* dquot_resume needs RW */
1751e8a6 1763 sb->s_flags &= ~SB_RDONLY;
ea676733
JK
1764 if (sb_any_quota_suspended(sb)) {
1765 dquot_resume(sb, -1);
7beb01f7 1766 } else if (f2fs_sb_has_quota_ino(sbi)) {
ea676733
JK
1767 err = f2fs_enable_quotas(sb);
1768 if (err)
1769 goto restore_opts;
1770 }
0abd675e 1771 }
ea676733 1772#endif
9cd81ce3
CY
1773 /* disallow enable/disable extent_cache dynamically */
1774 if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
1775 err = -EINVAL;
dcbb4c10 1776 f2fs_warn(sbi, "switch extent_cache option is not allowed");
9cd81ce3
CY
1777 goto restore_opts;
1778 }
1779
1f78adfa
CY
1780 if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
1781 err = -EINVAL;
1782 f2fs_warn(sbi, "switch io_bits option is not allowed");
1783 goto restore_opts;
1784 }
1785
4354994f
DR
1786 if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
1787 err = -EINVAL;
dcbb4c10 1788 f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
4354994f
DR
1789 goto restore_opts;
1790 }
1791
696c018c
NJ
1792 /*
1793 * We stop the GC thread if FS is mounted as RO
1794 * or if background_gc = off is passed in mount
1795 * option. Also sync the filesystem.
1796 */
bbbc34fd
CY
1797 if ((*flags & SB_RDONLY) ||
1798 F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF) {
696c018c 1799 if (sbi->gc_thread) {
4d57b86d 1800 f2fs_stop_gc_thread(sbi);
876dc59e 1801 need_restart_gc = true;
696c018c 1802 }
aba291b3 1803 } else if (!sbi->gc_thread) {
4d57b86d 1804 err = f2fs_start_gc_thread(sbi);
696c018c
NJ
1805 if (err)
1806 goto restore_opts;
876dc59e
GZ
1807 need_stop_gc = true;
1808 }
1809
63189b78
CY
1810 if (*flags & SB_RDONLY ||
1811 F2FS_OPTION(sbi).whint_mode != org_mount_opt.whint_mode) {
faa0e55b
JK
1812 writeback_inodes_sb(sb, WB_REASON_SYNC);
1813 sync_inodes_sb(sb);
1814
1815 set_sbi_flag(sbi, SBI_IS_DIRTY);
1816 set_sbi_flag(sbi, SBI_IS_CLOSE);
1817 f2fs_sync_fs(sb, 1);
1818 clear_sbi_flag(sbi, SBI_IS_CLOSE);
1819 }
1820
4354994f
DR
1821 if (checkpoint_changed) {
1822 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
1823 err = f2fs_disable_checkpoint(sbi);
1824 if (err)
1825 goto restore_gc;
1826 } else {
1827 f2fs_enable_checkpoint(sbi);
1828 }
1829 }
1830
876dc59e
GZ
1831 /*
1832 * We stop issue flush thread if FS is mounted as RO
1833 * or if flush_merge is not passed in mount option.
1834 */
1751e8a6 1835 if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
5eba8c5d 1836 clear_opt(sbi, FLUSH_MERGE);
4d57b86d 1837 f2fs_destroy_flush_cmd_control(sbi, false);
5eba8c5d 1838 } else {
4d57b86d 1839 err = f2fs_create_flush_cmd_control(sbi);
2163d198 1840 if (err)
a688b9d9 1841 goto restore_gc;
696c018c
NJ
1842 }
1843skip:
4b2414d0
CY
1844#ifdef CONFIG_QUOTA
1845 /* Release old quota file names */
1846 for (i = 0; i < MAXQUOTAS; i++)
5222595d 1847 kvfree(org_mount_opt.s_qf_names[i]);
4b2414d0 1848#endif
696c018c 1849 /* Update the POSIXACL Flag */
1751e8a6
LT
1850 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
1851 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
df728b0f 1852
7e65be49 1853 limit_reserve_root(sbi);
1ae18f71 1854 adjust_unusable_cap_perc(sbi);
095680f2 1855 *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
696c018c 1856 return 0;
876dc59e
GZ
1857restore_gc:
1858 if (need_restart_gc) {
4d57b86d 1859 if (f2fs_start_gc_thread(sbi))
dcbb4c10 1860 f2fs_warn(sbi, "background gc thread has stopped");
876dc59e 1861 } else if (need_stop_gc) {
4d57b86d 1862 f2fs_stop_gc_thread(sbi);
876dc59e 1863 }
696c018c 1864restore_opts:
4b2414d0 1865#ifdef CONFIG_QUOTA
63189b78 1866 F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
4b2414d0 1867 for (i = 0; i < MAXQUOTAS; i++) {
5222595d 1868 kvfree(F2FS_OPTION(sbi).s_qf_names[i]);
63189b78 1869 F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
4b2414d0
CY
1870 }
1871#endif
696c018c 1872 sbi->mount_opt = org_mount_opt;
0abd675e 1873 sb->s_flags = old_sb_flags;
696c018c
NJ
1874 return err;
1875}
1876
0abd675e
CY
1877#ifdef CONFIG_QUOTA
1878/* Read data from quotafile */
1879static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
1880 size_t len, loff_t off)
1881{
1882 struct inode *inode = sb_dqopt(sb)->files[type];
1883 struct address_space *mapping = inode->i_mapping;
1884 block_t blkidx = F2FS_BYTES_TO_BLK(off);
1885 int offset = off & (sb->s_blocksize - 1);
1886 int tocopy;
1887 size_t toread;
1888 loff_t i_size = i_size_read(inode);
1889 struct page *page;
1890 char *kaddr;
1891
1892 if (off > i_size)
1893 return 0;
1894
1895 if (off + len > i_size)
1896 len = i_size - off;
1897 toread = len;
1898 while (toread > 0) {
1899 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
1900repeat:
02117b8a 1901 page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
4e46a023
JK
1902 if (IS_ERR(page)) {
1903 if (PTR_ERR(page) == -ENOMEM) {
5df7731f
CY
1904 congestion_wait(BLK_RW_ASYNC,
1905 DEFAULT_IO_TIMEOUT);
4e46a023
JK
1906 goto repeat;
1907 }
af033b2a 1908 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
0abd675e 1909 return PTR_ERR(page);
4e46a023 1910 }
0abd675e
CY
1911
1912 lock_page(page);
1913
1914 if (unlikely(page->mapping != mapping)) {
1915 f2fs_put_page(page, 1);
1916 goto repeat;
1917 }
1918 if (unlikely(!PageUptodate(page))) {
1919 f2fs_put_page(page, 1);
af033b2a 1920 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
0abd675e
CY
1921 return -EIO;
1922 }
1923
1924 kaddr = kmap_atomic(page);
1925 memcpy(data, kaddr + offset, tocopy);
1926 kunmap_atomic(kaddr);
1927 f2fs_put_page(page, 1);
1928
1929 offset = 0;
1930 toread -= tocopy;
1931 data += tocopy;
1932 blkidx++;
1933 }
1934 return len;
1935}
1936
1937/* Write to quotafile */
1938static ssize_t f2fs_quota_write(struct super_block *sb, int type,
1939 const char *data, size_t len, loff_t off)
1940{
1941 struct inode *inode = sb_dqopt(sb)->files[type];
1942 struct address_space *mapping = inode->i_mapping;
1943 const struct address_space_operations *a_ops = mapping->a_ops;
1944 int offset = off & (sb->s_blocksize - 1);
1945 size_t towrite = len;
1946 struct page *page;
62f63eea 1947 void *fsdata = NULL;
0abd675e
CY
1948 char *kaddr;
1949 int err = 0;
1950 int tocopy;
1951
1952 while (towrite > 0) {
1953 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
1954 towrite);
4e46a023 1955retry:
0abd675e 1956 err = a_ops->write_begin(NULL, mapping, off, tocopy, 0,
62f63eea 1957 &page, &fsdata);
4e46a023
JK
1958 if (unlikely(err)) {
1959 if (err == -ENOMEM) {
5df7731f
CY
1960 congestion_wait(BLK_RW_ASYNC,
1961 DEFAULT_IO_TIMEOUT);
4e46a023
JK
1962 goto retry;
1963 }
af033b2a 1964 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
0abd675e 1965 break;
4e46a023 1966 }
0abd675e
CY
1967
1968 kaddr = kmap_atomic(page);
1969 memcpy(kaddr + offset, data, tocopy);
1970 kunmap_atomic(kaddr);
1971 flush_dcache_page(page);
1972
1973 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
62f63eea 1974 page, fsdata);
0abd675e
CY
1975 offset = 0;
1976 towrite -= tocopy;
1977 off += tocopy;
1978 data += tocopy;
1979 cond_resched();
1980 }
1981
1982 if (len == towrite)
6e5b5d41 1983 return err;
0abd675e
CY
1984 inode->i_mtime = inode->i_ctime = current_time(inode);
1985 f2fs_mark_inode_dirty_sync(inode, false);
1986 return len - towrite;
1987}
1988
1989static struct dquot **f2fs_get_dquots(struct inode *inode)
1990{
1991 return F2FS_I(inode)->i_dquot;
1992}
1993
1994static qsize_t *f2fs_get_reserved_space(struct inode *inode)
1995{
1996 return &F2FS_I(inode)->i_reserved_quota;
1997}
1998
4b2414d0
CY
1999static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2000{
af033b2a 2001 if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
dcbb4c10 2002 f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
af033b2a
CY
2003 return 0;
2004 }
2005
63189b78
CY
2006 return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2007 F2FS_OPTION(sbi).s_jquota_fmt, type);
4b2414d0
CY
2008}
2009
ea676733 2010int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
4b2414d0 2011{
ea676733
JK
2012 int enabled = 0;
2013 int i, err;
2014
7beb01f7 2015 if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
ea676733
JK
2016 err = f2fs_enable_quotas(sbi->sb);
2017 if (err) {
dcbb4c10 2018 f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
ea676733
JK
2019 return 0;
2020 }
2021 return 1;
2022 }
4b2414d0
CY
2023
2024 for (i = 0; i < MAXQUOTAS; i++) {
63189b78 2025 if (F2FS_OPTION(sbi).s_qf_names[i]) {
ea676733
JK
2026 err = f2fs_quota_on_mount(sbi, i);
2027 if (!err) {
2028 enabled = 1;
2029 continue;
2030 }
dcbb4c10
JP
2031 f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2032 err, i);
4b2414d0
CY
2033 }
2034 }
ea676733
JK
2035 return enabled;
2036}
2037
2038static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2039 unsigned int flags)
2040{
2041 struct inode *qf_inode;
2042 unsigned long qf_inum;
2043 int err;
2044
7beb01f7 2045 BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
ea676733
JK
2046
2047 qf_inum = f2fs_qf_ino(sb, type);
2048 if (!qf_inum)
2049 return -EPERM;
2050
2051 qf_inode = f2fs_iget(sb, qf_inum);
2052 if (IS_ERR(qf_inode)) {
dcbb4c10 2053 f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
ea676733
JK
2054 return PTR_ERR(qf_inode);
2055 }
2056
2057 /* Don't account quota for quota files to avoid recursion */
2058 qf_inode->i_flags |= S_NOQUOTA;
7212b95e 2059 err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
ea676733
JK
2060 iput(qf_inode);
2061 return err;
2062}
2063
2064static int f2fs_enable_quotas(struct super_block *sb)
2065{
dcbb4c10 2066 struct f2fs_sb_info *sbi = F2FS_SB(sb);
ea676733
JK
2067 int type, err = 0;
2068 unsigned long qf_inum;
2069 bool quota_mopt[MAXQUOTAS] = {
dcbb4c10
JP
2070 test_opt(sbi, USRQUOTA),
2071 test_opt(sbi, GRPQUOTA),
2072 test_opt(sbi, PRJQUOTA),
ea676733
JK
2073 };
2074
af033b2a 2075 if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
dcbb4c10 2076 f2fs_err(sbi, "quota file may be corrupted, skip loading it");
af033b2a
CY
2077 return 0;
2078 }
2079
2080 sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2081
ea676733
JK
2082 for (type = 0; type < MAXQUOTAS; type++) {
2083 qf_inum = f2fs_qf_ino(sb, type);
2084 if (qf_inum) {
2085 err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2086 DQUOT_USAGE_ENABLED |
2087 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2088 if (err) {
dcbb4c10
JP
2089 f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2090 type, err);
ea676733
JK
2091 for (type--; type >= 0; type--)
2092 dquot_quota_off(sb, type);
af033b2a
CY
2093 set_sbi_flag(F2FS_SB(sb),
2094 SBI_QUOTA_NEED_REPAIR);
ea676733
JK
2095 return err;
2096 }
4b2414d0
CY
2097 }
2098 }
ea676733 2099 return 0;
4b2414d0
CY
2100}
2101
af033b2a 2102int f2fs_quota_sync(struct super_block *sb, int type)
0abd675e 2103{
af033b2a 2104 struct f2fs_sb_info *sbi = F2FS_SB(sb);
0abd675e
CY
2105 struct quota_info *dqopt = sb_dqopt(sb);
2106 int cnt;
2107 int ret;
2108
db6ec53b
JK
2109 /*
2110 * do_quotactl
2111 * f2fs_quota_sync
2112 * down_read(quota_sem)
2113 * dquot_writeback_dquots()
2114 * f2fs_dquot_commit
2115 * block_operation
2116 * down_read(quota_sem)
2117 */
2118 f2fs_lock_op(sbi);
2119
2120 down_read(&sbi->quota_sem);
0abd675e
CY
2121 ret = dquot_writeback_dquots(sb, type);
2122 if (ret)
af033b2a 2123 goto out;
0abd675e
CY
2124
2125 /*
2126 * Now when everything is written we can discard the pagecache so
2127 * that userspace sees the changes.
2128 */
2129 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
af033b2a
CY
2130 struct address_space *mapping;
2131
0abd675e
CY
2132 if (type != -1 && cnt != type)
2133 continue;
2134 if (!sb_has_quota_active(sb, cnt))
2135 continue;
2136
af033b2a
CY
2137 mapping = dqopt->files[cnt]->i_mapping;
2138
2139 ret = filemap_fdatawrite(mapping);
0abd675e 2140 if (ret)
af033b2a
CY
2141 goto out;
2142
2143 /* if we are using journalled quota */
2144 if (is_journalled_quota(sbi))
2145 continue;
2146
2147 ret = filemap_fdatawait(mapping);
2148 if (ret)
2149 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
0abd675e
CY
2150
2151 inode_lock(dqopt->files[cnt]);
2152 truncate_inode_pages(&dqopt->files[cnt]->i_data, 0);
2153 inode_unlock(dqopt->files[cnt]);
2154 }
af033b2a
CY
2155out:
2156 if (ret)
2157 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
db6ec53b
JK
2158 up_read(&sbi->quota_sem);
2159 f2fs_unlock_op(sbi);
af033b2a 2160 return ret;
0abd675e
CY
2161}
2162
2163static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2164 const struct path *path)
2165{
2166 struct inode *inode;
2167 int err;
2168
fe973b06
CY
2169 /* if quota sysfile exists, deny enabling quota with specific file */
2170 if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2171 f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2172 return -EBUSY;
2173 }
2174
9a20d391 2175 err = f2fs_quota_sync(sb, type);
0abd675e
CY
2176 if (err)
2177 return err;
2178
2179 err = dquot_quota_on(sb, type, format_id, path);
2180 if (err)
2181 return err;
2182
2183 inode = d_inode(path->dentry);
2184
2185 inode_lock(inode);
59c84408 2186 F2FS_I(inode)->i_flags |= F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL;
9149a5eb 2187 f2fs_set_inode_flags(inode);
0abd675e
CY
2188 inode_unlock(inode);
2189 f2fs_mark_inode_dirty_sync(inode, false);
2190
2191 return 0;
2192}
2193
fe973b06 2194static int __f2fs_quota_off(struct super_block *sb, int type)
0abd675e
CY
2195{
2196 struct inode *inode = sb_dqopt(sb)->files[type];
2197 int err;
2198
2199 if (!inode || !igrab(inode))
2200 return dquot_quota_off(sb, type);
2201
cda9cc59
YH
2202 err = f2fs_quota_sync(sb, type);
2203 if (err)
2204 goto out_put;
0abd675e
CY
2205
2206 err = dquot_quota_off(sb, type);
7beb01f7 2207 if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
0abd675e
CY
2208 goto out_put;
2209
2210 inode_lock(inode);
59c84408 2211 F2FS_I(inode)->i_flags &= ~(F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL);
9149a5eb 2212 f2fs_set_inode_flags(inode);
0abd675e
CY
2213 inode_unlock(inode);
2214 f2fs_mark_inode_dirty_sync(inode, false);
2215out_put:
2216 iput(inode);
2217 return err;
2218}
2219
fe973b06
CY
2220static int f2fs_quota_off(struct super_block *sb, int type)
2221{
2222 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2223 int err;
2224
2225 err = __f2fs_quota_off(sb, type);
2226
2227 /*
2228 * quotactl can shutdown journalled quota, result in inconsistence
2229 * between quota record and fs data by following updates, tag the
2230 * flag to let fsck be aware of it.
2231 */
2232 if (is_journalled_quota(sbi))
2233 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2234 return err;
2235}
2236
4b2414d0 2237void f2fs_quota_off_umount(struct super_block *sb)
0abd675e
CY
2238{
2239 int type;
cda9cc59
YH
2240 int err;
2241
2242 for (type = 0; type < MAXQUOTAS; type++) {
fe973b06 2243 err = __f2fs_quota_off(sb, type);
cda9cc59
YH
2244 if (err) {
2245 int ret = dquot_quota_off(sb, type);
0abd675e 2246
dcbb4c10
JP
2247 f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
2248 type, err, ret);
af033b2a 2249 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
cda9cc59
YH
2250 }
2251 }
0e0667b6
JK
2252 /*
2253 * In case of checkpoint=disable, we must flush quota blocks.
2254 * This can cause NULL exception for node_inode in end_io, since
2255 * put_super already dropped it.
2256 */
2257 sync_filesystem(sb);
0abd675e
CY
2258}
2259
26b5a079
SY
2260static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
2261{
2262 struct quota_info *dqopt = sb_dqopt(sb);
2263 int type;
2264
2265 for (type = 0; type < MAXQUOTAS; type++) {
2266 if (!dqopt->files[type])
2267 continue;
2268 f2fs_inode_synced(dqopt->files[type]);
2269 }
2270}
2271
af033b2a
CY
2272static int f2fs_dquot_commit(struct dquot *dquot)
2273{
db6ec53b 2274 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
af033b2a
CY
2275 int ret;
2276
2c4e0c52 2277 down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
af033b2a
CY
2278 ret = dquot_commit(dquot);
2279 if (ret < 0)
db6ec53b
JK
2280 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2281 up_read(&sbi->quota_sem);
af033b2a
CY
2282 return ret;
2283}
2284
2285static int f2fs_dquot_acquire(struct dquot *dquot)
2286{
db6ec53b 2287 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
af033b2a
CY
2288 int ret;
2289
db6ec53b 2290 down_read(&sbi->quota_sem);
af033b2a
CY
2291 ret = dquot_acquire(dquot);
2292 if (ret < 0)
db6ec53b
JK
2293 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2294 up_read(&sbi->quota_sem);
af033b2a
CY
2295 return ret;
2296}
2297
2298static int f2fs_dquot_release(struct dquot *dquot)
2299{
db6ec53b 2300 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2c4e0c52 2301 int ret = dquot_release(dquot);
af033b2a 2302
af033b2a 2303 if (ret < 0)
db6ec53b 2304 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
af033b2a
CY
2305 return ret;
2306}
2307
2308static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
2309{
2310 struct super_block *sb = dquot->dq_sb;
2311 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2c4e0c52 2312 int ret = dquot_mark_dquot_dirty(dquot);
af033b2a
CY
2313
2314 /* if we are using journalled quota */
2315 if (is_journalled_quota(sbi))
2316 set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
2317
2318 return ret;
2319}
2320
2321static int f2fs_dquot_commit_info(struct super_block *sb, int type)
2322{
db6ec53b 2323 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2c4e0c52 2324 int ret = dquot_commit_info(sb, type);
af033b2a 2325
af033b2a 2326 if (ret < 0)
db6ec53b 2327 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
af033b2a
CY
2328 return ret;
2329}
26b5a079 2330
94b1e10e 2331static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
5c57132e
CY
2332{
2333 *projid = F2FS_I(inode)->i_projid;
2334 return 0;
2335}
2336
0abd675e
CY
2337static const struct dquot_operations f2fs_quota_operations = {
2338 .get_reserved_space = f2fs_get_reserved_space,
af033b2a
CY
2339 .write_dquot = f2fs_dquot_commit,
2340 .acquire_dquot = f2fs_dquot_acquire,
2341 .release_dquot = f2fs_dquot_release,
2342 .mark_dirty = f2fs_dquot_mark_dquot_dirty,
2343 .write_info = f2fs_dquot_commit_info,
0abd675e
CY
2344 .alloc_dquot = dquot_alloc,
2345 .destroy_dquot = dquot_destroy,
5c57132e 2346 .get_projid = f2fs_get_projid,
0abd675e
CY
2347 .get_next_id = dquot_get_next_id,
2348};
2349
2350static const struct quotactl_ops f2fs_quotactl_ops = {
2351 .quota_on = f2fs_quota_on,
2352 .quota_off = f2fs_quota_off,
2353 .quota_sync = f2fs_quota_sync,
2354 .get_state = dquot_get_state,
2355 .set_info = dquot_set_dqinfo,
2356 .get_dqblk = dquot_get_dqblk,
2357 .set_dqblk = dquot_set_dqblk,
2358 .get_nextdqblk = dquot_get_next_dqblk,
2359};
2360#else
af033b2a
CY
2361int f2fs_quota_sync(struct super_block *sb, int type)
2362{
2363 return 0;
2364}
2365
4b2414d0 2366void f2fs_quota_off_umount(struct super_block *sb)
0abd675e
CY
2367{
2368}
2369#endif
2370
f62fc9f9 2371static const struct super_operations f2fs_sops = {
aff063e2 2372 .alloc_inode = f2fs_alloc_inode,
d01718a0 2373 .free_inode = f2fs_free_inode,
531ad7d5 2374 .drop_inode = f2fs_drop_inode,
aff063e2 2375 .write_inode = f2fs_write_inode,
b3783873 2376 .dirty_inode = f2fs_dirty_inode,
aff063e2 2377 .show_options = f2fs_show_options,
0abd675e
CY
2378#ifdef CONFIG_QUOTA
2379 .quota_read = f2fs_quota_read,
2380 .quota_write = f2fs_quota_write,
2381 .get_dquots = f2fs_get_dquots,
2382#endif
aff063e2
JK
2383 .evict_inode = f2fs_evict_inode,
2384 .put_super = f2fs_put_super,
2385 .sync_fs = f2fs_sync_fs,
d6212a5f
CL
2386 .freeze_fs = f2fs_freeze,
2387 .unfreeze_fs = f2fs_unfreeze,
aff063e2 2388 .statfs = f2fs_statfs,
696c018c 2389 .remount_fs = f2fs_remount,
aff063e2
JK
2390};
2391
643fa961 2392#ifdef CONFIG_FS_ENCRYPTION
0b81d077
JK
2393static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
2394{
2395 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2396 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2397 ctx, len, NULL);
2398}
2399
2400static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
2401 void *fs_data)
2402{
b7c409de
SY
2403 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2404
2405 /*
2406 * Encrypting the root directory is not allowed because fsck
2407 * expects lost+found directory to exist and remain unencrypted
2408 * if LOST_FOUND feature is enabled.
2409 *
2410 */
7beb01f7 2411 if (f2fs_sb_has_lost_found(sbi) &&
b7c409de
SY
2412 inode->i_ino == F2FS_ROOT_INO(sbi))
2413 return -EPERM;
2414
0b81d077
JK
2415 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2416 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2417 ctx, len, fs_data, XATTR_CREATE);
2418}
2419
ff62af20
SY
2420static bool f2fs_dummy_context(struct inode *inode)
2421{
2422 return DUMMY_ENCRYPTION_ENABLED(F2FS_I_SB(inode));
2423}
2424
0eee17e3
EB
2425static bool f2fs_has_stable_inodes(struct super_block *sb)
2426{
2427 return true;
2428}
2429
2430static void f2fs_get_ino_and_lblk_bits(struct super_block *sb,
2431 int *ino_bits_ret, int *lblk_bits_ret)
2432{
2433 *ino_bits_ret = 8 * sizeof(nid_t);
2434 *lblk_bits_ret = 8 * sizeof(block_t);
2435}
2436
6f69f0ed 2437static const struct fscrypt_operations f2fs_cryptops = {
0eee17e3
EB
2438 .key_prefix = "f2fs:",
2439 .get_context = f2fs_get_context,
2440 .set_context = f2fs_set_context,
2441 .dummy_context = f2fs_dummy_context,
2442 .empty_dir = f2fs_empty_dir,
2443 .max_namelen = F2FS_NAME_LEN,
2444 .has_stable_inodes = f2fs_has_stable_inodes,
2445 .get_ino_and_lblk_bits = f2fs_get_ino_and_lblk_bits,
0b81d077 2446};
0b81d077
JK
2447#endif
2448
aff063e2
JK
2449static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
2450 u64 ino, u32 generation)
2451{
2452 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2453 struct inode *inode;
2454
4d57b86d 2455 if (f2fs_check_nid_range(sbi, ino))
910bb12d 2456 return ERR_PTR(-ESTALE);
aff063e2
JK
2457
2458 /*
2459 * f2fs_iget isn't quite right if the inode is currently unallocated!
2460 * However f2fs_iget currently does appropriate checks to handle stale
2461 * inodes so everything is OK.
2462 */
2463 inode = f2fs_iget(sb, ino);
2464 if (IS_ERR(inode))
2465 return ERR_CAST(inode);
6bacf52f 2466 if (unlikely(generation && inode->i_generation != generation)) {
aff063e2
JK
2467 /* we didn't find the right inode.. */
2468 iput(inode);
2469 return ERR_PTR(-ESTALE);
2470 }
2471 return inode;
2472}
2473
2474static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
2475 int fh_len, int fh_type)
2476{
2477 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
2478 f2fs_nfs_get_inode);
2479}
2480
2481static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
2482 int fh_len, int fh_type)
2483{
2484 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
2485 f2fs_nfs_get_inode);
2486}
2487
2488static const struct export_operations f2fs_export_ops = {
2489 .fh_to_dentry = f2fs_fh_to_dentry,
2490 .fh_to_parent = f2fs_fh_to_parent,
2491 .get_parent = f2fs_get_parent,
2492};
2493
e0afc4d6 2494static loff_t max_file_blocks(void)
aff063e2 2495{
7a2af766 2496 loff_t result = 0;
d02a6e61 2497 loff_t leaf_count = DEF_ADDRS_PER_BLOCK;
aff063e2 2498
7a2af766
CY
2499 /*
2500 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
6afc662e 2501 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
7a2af766
CY
2502 * space in inode.i_addr, it will be more safe to reassign
2503 * result as zero.
2504 */
2505
aff063e2
JK
2506 /* two direct node blocks */
2507 result += (leaf_count * 2);
2508
2509 /* two indirect node blocks */
2510 leaf_count *= NIDS_PER_BLOCK;
2511 result += (leaf_count * 2);
2512
2513 /* one double indirect node block */
2514 leaf_count *= NIDS_PER_BLOCK;
2515 result += leaf_count;
2516
aff063e2
JK
2517 return result;
2518}
2519
fd694733
JK
2520static int __f2fs_commit_super(struct buffer_head *bh,
2521 struct f2fs_super_block *super)
2522{
2523 lock_buffer(bh);
2524 if (super)
2525 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
fd694733
JK
2526 set_buffer_dirty(bh);
2527 unlock_buffer(bh);
2528
2529 /* it's rare case, we can do fua all the time */
3adc5fcb 2530 return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
fd694733
JK
2531}
2532
df728b0f 2533static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
fd694733 2534 struct buffer_head *bh)
9a59b62f 2535{
fd694733
JK
2536 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2537 (bh->b_data + F2FS_SUPER_OFFSET);
df728b0f 2538 struct super_block *sb = sbi->sb;
9a59b62f
CY
2539 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
2540 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
2541 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
2542 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
2543 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
2544 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
2545 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
2546 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
2547 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
2548 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
2549 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2550 u32 segment_count = le32_to_cpu(raw_super->segment_count);
2551 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
fd694733
JK
2552 u64 main_end_blkaddr = main_blkaddr +
2553 (segment_count_main << log_blocks_per_seg);
2554 u64 seg_end_blkaddr = segment0_blkaddr +
2555 (segment_count << log_blocks_per_seg);
9a59b62f
CY
2556
2557 if (segment0_blkaddr != cp_blkaddr) {
dcbb4c10
JP
2558 f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
2559 segment0_blkaddr, cp_blkaddr);
9a59b62f
CY
2560 return true;
2561 }
2562
2563 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
2564 sit_blkaddr) {
dcbb4c10
JP
2565 f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
2566 cp_blkaddr, sit_blkaddr,
2567 segment_count_ckpt << log_blocks_per_seg);
9a59b62f
CY
2568 return true;
2569 }
2570
2571 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
2572 nat_blkaddr) {
dcbb4c10
JP
2573 f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
2574 sit_blkaddr, nat_blkaddr,
2575 segment_count_sit << log_blocks_per_seg);
9a59b62f
CY
2576 return true;
2577 }
2578
2579 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
2580 ssa_blkaddr) {
dcbb4c10
JP
2581 f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
2582 nat_blkaddr, ssa_blkaddr,
2583 segment_count_nat << log_blocks_per_seg);
9a59b62f
CY
2584 return true;
2585 }
2586
2587 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
2588 main_blkaddr) {
dcbb4c10
JP
2589 f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
2590 ssa_blkaddr, main_blkaddr,
2591 segment_count_ssa << log_blocks_per_seg);
9a59b62f
CY
2592 return true;
2593 }
2594
fd694733 2595 if (main_end_blkaddr > seg_end_blkaddr) {
dcbb4c10
JP
2596 f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%u) block(%u)",
2597 main_blkaddr,
2598 segment0_blkaddr +
2599 (segment_count << log_blocks_per_seg),
2600 segment_count_main << log_blocks_per_seg);
9a59b62f 2601 return true;
fd694733
JK
2602 } else if (main_end_blkaddr < seg_end_blkaddr) {
2603 int err = 0;
2604 char *res;
2605
2606 /* fix in-memory information all the time */
2607 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
2608 segment0_blkaddr) >> log_blocks_per_seg);
2609
2610 if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
df728b0f 2611 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
fd694733
JK
2612 res = "internally";
2613 } else {
2614 err = __f2fs_commit_super(bh, NULL);
2615 res = err ? "failed" : "done";
2616 }
dcbb4c10
JP
2617 f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%u) block(%u)",
2618 res, main_blkaddr,
2619 segment0_blkaddr +
2620 (segment_count << log_blocks_per_seg),
2621 segment_count_main << log_blocks_per_seg);
fd694733
JK
2622 if (err)
2623 return true;
9a59b62f 2624 }
9a59b62f
CY
2625 return false;
2626}
2627
df728b0f 2628static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
fd694733 2629 struct buffer_head *bh)
aff063e2 2630{
0cfe75c5
JK
2631 block_t segment_count, segs_per_sec, secs_per_zone;
2632 block_t total_sections, blocks_per_seg;
fd694733
JK
2633 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2634 (bh->b_data + F2FS_SUPER_OFFSET);
aff063e2 2635 unsigned int blocksize;
d440c52d
JZ
2636 size_t crc_offset = 0;
2637 __u32 crc = 0;
2638
38fb6d0e
IZ
2639 if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
2640 f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
2641 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
2642 return -EINVAL;
2643 }
2644
d440c52d 2645 /* Check checksum_offset and crc in superblock */
7beb01f7 2646 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
d440c52d
JZ
2647 crc_offset = le32_to_cpu(raw_super->checksum_offset);
2648 if (crc_offset !=
2649 offsetof(struct f2fs_super_block, crc)) {
dcbb4c10
JP
2650 f2fs_info(sbi, "Invalid SB checksum offset: %zu",
2651 crc_offset);
38fb6d0e 2652 return -EFSCORRUPTED;
d440c52d
JZ
2653 }
2654 crc = le32_to_cpu(raw_super->crc);
2655 if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
dcbb4c10 2656 f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
38fb6d0e 2657 return -EFSCORRUPTED;
d440c52d
JZ
2658 }
2659 }
aff063e2 2660
5c9b4692 2661 /* Currently, support only 4KB page cache size */
09cbfeaf 2662 if (F2FS_BLKSIZE != PAGE_SIZE) {
dcbb4c10
JP
2663 f2fs_info(sbi, "Invalid page_cache_size (%lu), supports only 4KB",
2664 PAGE_SIZE);
38fb6d0e 2665 return -EFSCORRUPTED;
5c9b4692 2666 }
2667
aff063e2
JK
2668 /* Currently, support only 4KB block size */
2669 blocksize = 1 << le32_to_cpu(raw_super->log_blocksize);
5c9b4692 2670 if (blocksize != F2FS_BLKSIZE) {
dcbb4c10
JP
2671 f2fs_info(sbi, "Invalid blocksize (%u), supports only 4KB",
2672 blocksize);
38fb6d0e 2673 return -EFSCORRUPTED;
a07ef784 2674 }
5c9b4692 2675
9a59b62f
CY
2676 /* check log blocks per segment */
2677 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
dcbb4c10
JP
2678 f2fs_info(sbi, "Invalid log blocks per segment (%u)",
2679 le32_to_cpu(raw_super->log_blocks_per_seg));
38fb6d0e 2680 return -EFSCORRUPTED;
9a59b62f
CY
2681 }
2682
55cf9cb6
CY
2683 /* Currently, support 512/1024/2048/4096 bytes sector size */
2684 if (le32_to_cpu(raw_super->log_sectorsize) >
2685 F2FS_MAX_LOG_SECTOR_SIZE ||
2686 le32_to_cpu(raw_super->log_sectorsize) <
2687 F2FS_MIN_LOG_SECTOR_SIZE) {
dcbb4c10
JP
2688 f2fs_info(sbi, "Invalid log sectorsize (%u)",
2689 le32_to_cpu(raw_super->log_sectorsize));
38fb6d0e 2690 return -EFSCORRUPTED;
a07ef784 2691 }
55cf9cb6
CY
2692 if (le32_to_cpu(raw_super->log_sectors_per_block) +
2693 le32_to_cpu(raw_super->log_sectorsize) !=
2694 F2FS_MAX_LOG_SECTOR_SIZE) {
dcbb4c10
JP
2695 f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
2696 le32_to_cpu(raw_super->log_sectors_per_block),
2697 le32_to_cpu(raw_super->log_sectorsize));
38fb6d0e 2698 return -EFSCORRUPTED;
a07ef784 2699 }
9a59b62f 2700
0cfe75c5
JK
2701 segment_count = le32_to_cpu(raw_super->segment_count);
2702 segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
2703 secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
2704 total_sections = le32_to_cpu(raw_super->section_count);
2705
2706 /* blocks_per_seg should be 512, given the above check */
2707 blocks_per_seg = 1 << le32_to_cpu(raw_super->log_blocks_per_seg);
2708
2709 if (segment_count > F2FS_MAX_SEGMENT ||
2710 segment_count < F2FS_MIN_SEGMENTS) {
dcbb4c10 2711 f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
38fb6d0e 2712 return -EFSCORRUPTED;
0cfe75c5
JK
2713 }
2714
2715 if (total_sections > segment_count ||
2716 total_sections < F2FS_MIN_SEGMENTS ||
2717 segs_per_sec > segment_count || !segs_per_sec) {
dcbb4c10
JP
2718 f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
2719 segment_count, total_sections, segs_per_sec);
38fb6d0e 2720 return -EFSCORRUPTED;
0cfe75c5
JK
2721 }
2722
2723 if ((segment_count / segs_per_sec) < total_sections) {
dcbb4c10
JP
2724 f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
2725 segment_count, segs_per_sec, total_sections);
38fb6d0e 2726 return -EFSCORRUPTED;
0cfe75c5
JK
2727 }
2728
88960068 2729 if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
dcbb4c10
JP
2730 f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
2731 segment_count, le64_to_cpu(raw_super->block_count));
38fb6d0e 2732 return -EFSCORRUPTED;
0cfe75c5
JK
2733 }
2734
9f701f6c
QS
2735 if (RDEV(0).path[0]) {
2736 block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
2737 int i = 1;
2738
2739 while (i < MAX_DEVICES && RDEV(i).path[0]) {
2740 dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
2741 i++;
2742 }
2743 if (segment_count != dev_seg_count) {
2744 f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
2745 segment_count, dev_seg_count);
2746 return -EFSCORRUPTED;
2747 }
2748 }
2749
42bf546c 2750 if (secs_per_zone > total_sections || !secs_per_zone) {
dcbb4c10
JP
2751 f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
2752 secs_per_zone, total_sections);
38fb6d0e 2753 return -EFSCORRUPTED;
0cfe75c5
JK
2754 }
2755 if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
2756 raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
2757 (le32_to_cpu(raw_super->extension_count) +
2758 raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
dcbb4c10
JP
2759 f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
2760 le32_to_cpu(raw_super->extension_count),
2761 raw_super->hot_ext_count,
2762 F2FS_MAX_EXTENSION);
38fb6d0e 2763 return -EFSCORRUPTED;
0cfe75c5
JK
2764 }
2765
2766 if (le32_to_cpu(raw_super->cp_payload) >
2767 (blocks_per_seg - F2FS_CP_PACKS)) {
dcbb4c10
JP
2768 f2fs_info(sbi, "Insane cp_payload (%u > %u)",
2769 le32_to_cpu(raw_super->cp_payload),
2770 blocks_per_seg - F2FS_CP_PACKS);
38fb6d0e 2771 return -EFSCORRUPTED;
0cfe75c5
JK
2772 }
2773
9a59b62f
CY
2774 /* check reserved ino info */
2775 if (le32_to_cpu(raw_super->node_ino) != 1 ||
2776 le32_to_cpu(raw_super->meta_ino) != 2 ||
2777 le32_to_cpu(raw_super->root_ino) != 3) {
dcbb4c10
JP
2778 f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
2779 le32_to_cpu(raw_super->node_ino),
2780 le32_to_cpu(raw_super->meta_ino),
2781 le32_to_cpu(raw_super->root_ino));
38fb6d0e 2782 return -EFSCORRUPTED;
9a59b62f
CY
2783 }
2784
2785 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
df728b0f 2786 if (sanity_check_area_boundary(sbi, bh))
38fb6d0e 2787 return -EFSCORRUPTED;
9a59b62f 2788
aff063e2
JK
2789 return 0;
2790}
2791
4d57b86d 2792int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
aff063e2
JK
2793{
2794 unsigned int total, fsmeta;
577e3495
JK
2795 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
2796 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2040fce8 2797 unsigned int ovp_segments, reserved_segments;
15d3042a 2798 unsigned int main_segs, blocks_per_seg;
c77ec61c
CY
2799 unsigned int sit_segs, nat_segs;
2800 unsigned int sit_bitmap_size, nat_bitmap_size;
2801 unsigned int log_blocks_per_seg;
9dc956b2 2802 unsigned int segment_count_main;
e494c2f9 2803 unsigned int cp_pack_start_sum, cp_payload;
7b63f72f
CY
2804 block_t user_block_count, valid_user_blocks;
2805 block_t avail_node_count, valid_node_count;
042be0f8 2806 int i, j;
aff063e2
JK
2807
2808 total = le32_to_cpu(raw_super->segment_count);
2809 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
c77ec61c
CY
2810 sit_segs = le32_to_cpu(raw_super->segment_count_sit);
2811 fsmeta += sit_segs;
2812 nat_segs = le32_to_cpu(raw_super->segment_count_nat);
2813 fsmeta += nat_segs;
aff063e2
JK
2814 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
2815 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
2816
6bacf52f 2817 if (unlikely(fsmeta >= total))
aff063e2 2818 return 1;
577e3495 2819
2040fce8
JK
2820 ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
2821 reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
2822
2823 if (unlikely(fsmeta < F2FS_MIN_SEGMENTS ||
2824 ovp_segments == 0 || reserved_segments == 0)) {
dcbb4c10 2825 f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
2040fce8
JK
2826 return 1;
2827 }
2828
9dc956b2
CY
2829 user_block_count = le64_to_cpu(ckpt->user_block_count);
2830 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2831 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2832 if (!user_block_count || user_block_count >=
2833 segment_count_main << log_blocks_per_seg) {
dcbb4c10
JP
2834 f2fs_err(sbi, "Wrong user_block_count: %u",
2835 user_block_count);
9dc956b2
CY
2836 return 1;
2837 }
2838
7b63f72f
CY
2839 valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
2840 if (valid_user_blocks > user_block_count) {
dcbb4c10
JP
2841 f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
2842 valid_user_blocks, user_block_count);
7b63f72f
CY
2843 return 1;
2844 }
2845
2846 valid_node_count = le32_to_cpu(ckpt->valid_node_count);
27cae0bc 2847 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
7b63f72f 2848 if (valid_node_count > avail_node_count) {
dcbb4c10
JP
2849 f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
2850 valid_node_count, avail_node_count);
7b63f72f
CY
2851 return 1;
2852 }
2853
15d3042a
JQ
2854 main_segs = le32_to_cpu(raw_super->segment_count_main);
2855 blocks_per_seg = sbi->blocks_per_seg;
2856
2857 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
2858 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
2859 le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
2860 return 1;
042be0f8
CY
2861 for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
2862 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
2863 le32_to_cpu(ckpt->cur_node_segno[j])) {
dcbb4c10
JP
2864 f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
2865 i, j,
2866 le32_to_cpu(ckpt->cur_node_segno[i]));
042be0f8
CY
2867 return 1;
2868 }
2869 }
15d3042a
JQ
2870 }
2871 for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
2872 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
2873 le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
2874 return 1;
042be0f8
CY
2875 for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
2876 if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
2877 le32_to_cpu(ckpt->cur_data_segno[j])) {
dcbb4c10
JP
2878 f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
2879 i, j,
2880 le32_to_cpu(ckpt->cur_data_segno[i]));
042be0f8
CY
2881 return 1;
2882 }
2883 }
2884 }
2885 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
1166c1f2 2886 for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
042be0f8
CY
2887 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
2888 le32_to_cpu(ckpt->cur_data_segno[j])) {
1166c1f2 2889 f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
dcbb4c10
JP
2890 i, j,
2891 le32_to_cpu(ckpt->cur_node_segno[i]));
042be0f8
CY
2892 return 1;
2893 }
2894 }
15d3042a
JQ
2895 }
2896
c77ec61c
CY
2897 sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2898 nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
c77ec61c
CY
2899
2900 if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
2901 nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
dcbb4c10
JP
2902 f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
2903 sit_bitmap_size, nat_bitmap_size);
c77ec61c
CY
2904 return 1;
2905 }
e494c2f9
CY
2906
2907 cp_pack_start_sum = __start_sum_addr(sbi);
2908 cp_payload = __cp_payload(sbi);
2909 if (cp_pack_start_sum < cp_payload + 1 ||
2910 cp_pack_start_sum > blocks_per_seg - 1 -
2911 NR_CURSEG_TYPE) {
dcbb4c10
JP
2912 f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
2913 cp_pack_start_sum);
e494c2f9
CY
2914 return 1;
2915 }
c77ec61c 2916
5dae2d39
CY
2917 if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
2918 le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
2d008835
CY
2919 f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
2920 "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
2921 "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
dcbb4c10 2922 le32_to_cpu(ckpt->checksum_offset));
5dae2d39
CY
2923 return 1;
2924 }
2925
1e968fdf 2926 if (unlikely(f2fs_cp_error(sbi))) {
dcbb4c10 2927 f2fs_err(sbi, "A bug case: need to run fsck");
577e3495
JK
2928 return 1;
2929 }
aff063e2
JK
2930 return 0;
2931}
2932
2933static void init_sb_info(struct f2fs_sb_info *sbi)
2934{
2935 struct f2fs_super_block *raw_super = sbi->raw_super;
089842de 2936 int i;
aff063e2
JK
2937
2938 sbi->log_sectors_per_block =
2939 le32_to_cpu(raw_super->log_sectors_per_block);
2940 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
2941 sbi->blocksize = 1 << sbi->log_blocksize;
2942 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2943 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
2944 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
2945 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
2946 sbi->total_sections = le32_to_cpu(raw_super->section_count);
2947 sbi->total_node_count =
2948 (le32_to_cpu(raw_super->segment_count_nat) / 2)
2949 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
2950 sbi->root_ino_num = le32_to_cpu(raw_super->root_ino);
2951 sbi->node_ino_num = le32_to_cpu(raw_super->node_ino);
2952 sbi->meta_ino_num = le32_to_cpu(raw_super->meta_ino);
5ec4e49f 2953 sbi->cur_victim_sec = NULL_SECNO;
e3080b01
CY
2954 sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
2955 sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
b1c57c1c 2956 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
e3080b01 2957 sbi->migration_granularity = sbi->segs_per_sec;
aff063e2 2958
ab9fa662 2959 sbi->dir_level = DEF_DIR_LEVEL;
6beceb54 2960 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
d0239e1b 2961 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
a7d10cf3
ST
2962 sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
2963 sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
4354994f 2964 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
03f2c02d
JK
2965 sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
2966 DEF_UMOUNT_DISCARD_TIMEOUT;
caf0047e 2967 clear_sbi_flag(sbi, SBI_NEED_FSCK);
2658e50d 2968
35782b23
JK
2969 for (i = 0; i < NR_COUNT_TYPE; i++)
2970 atomic_set(&sbi->nr_pages[i], 0);
2971
c29fd0c0
CY
2972 for (i = 0; i < META; i++)
2973 atomic_set(&sbi->wb_sync_req[i], 0);
687de7f1 2974
2658e50d
JK
2975 INIT_LIST_HEAD(&sbi->s_list);
2976 mutex_init(&sbi->umount_mutex);
107a805d 2977 init_rwsem(&sbi->io_order_lock);
aaec2b1d 2978 spin_lock_init(&sbi->cp_lock);
1228b482
CY
2979
2980 sbi->dirty_device = 0;
2981 spin_lock_init(&sbi->dev_lock);
d0d3f1b3 2982
846ae671 2983 init_rwsem(&sbi->sb_lock);
f5a53edc 2984 init_rwsem(&sbi->pin_sem);
aff063e2
JK
2985}
2986
523be8a6
JK
2987static int init_percpu_info(struct f2fs_sb_info *sbi)
2988{
35782b23 2989 int err;
41382ec4 2990
513c5f37
JK
2991 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
2992 if (err)
2993 return err;
2994
4a70e255 2995 err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
41382ec4 2996 GFP_KERNEL);
4a70e255
CY
2997 if (err)
2998 percpu_counter_destroy(&sbi->alloc_valid_block_count);
2999
3000 return err;
523be8a6
JK
3001}
3002
178053e2 3003#ifdef CONFIG_BLK_DEV_ZONED
d4100351
CH
3004static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3005 void *data)
3006{
3007 struct f2fs_dev_info *dev = data;
3008
3009 if (zone->type != BLK_ZONE_TYPE_CONVENTIONAL)
3010 set_bit(idx, dev->blkz_seq);
3011 return 0;
3012}
3013
3c62be17 3014static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
178053e2 3015{
3c62be17 3016 struct block_device *bdev = FDEV(devi).bdev;
178053e2 3017 sector_t nr_sectors = bdev->bd_part->nr_sects;
d4100351 3018 int ret;
178053e2 3019
7beb01f7 3020 if (!f2fs_sb_has_blkzoned(sbi))
178053e2
DLM
3021 return 0;
3022
3c62be17 3023 if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
f99e8648 3024 SECTOR_TO_BLOCK(bdev_zone_sectors(bdev)))
3c62be17 3025 return -EINVAL;
f99e8648 3026 sbi->blocks_per_blkz = SECTOR_TO_BLOCK(bdev_zone_sectors(bdev));
3c62be17
JK
3027 if (sbi->log_blocks_per_blkz && sbi->log_blocks_per_blkz !=
3028 __ilog2_u32(sbi->blocks_per_blkz))
3029 return -EINVAL;
178053e2 3030 sbi->log_blocks_per_blkz = __ilog2_u32(sbi->blocks_per_blkz);
3c62be17
JK
3031 FDEV(devi).nr_blkz = SECTOR_TO_BLOCK(nr_sectors) >>
3032 sbi->log_blocks_per_blkz;
f99e8648 3033 if (nr_sectors & (bdev_zone_sectors(bdev) - 1))
3c62be17 3034 FDEV(devi).nr_blkz++;
178053e2 3035
95175daf
DLM
3036 FDEV(devi).blkz_seq = f2fs_kzalloc(sbi,
3037 BITS_TO_LONGS(FDEV(devi).nr_blkz)
3038 * sizeof(unsigned long),
3039 GFP_KERNEL);
3040 if (!FDEV(devi).blkz_seq)
178053e2
DLM
3041 return -ENOMEM;
3042
178053e2 3043 /* Get block zones type */
d4100351
CH
3044 ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3045 &FDEV(devi));
3046 if (ret < 0)
3047 return ret;
178053e2 3048
d4100351 3049 return 0;
178053e2
DLM
3050}
3051#endif
3052
9076a75f
GZ
3053/*
3054 * Read f2fs raw super block.
2b39e907
SL
3055 * Because we have two copies of super block, so read both of them
3056 * to get the first valid one. If any one of them is broken, we pass
3057 * them recovery flag back to the caller.
9076a75f 3058 */
df728b0f 3059static int read_raw_super_block(struct f2fs_sb_info *sbi,
9076a75f 3060 struct f2fs_super_block **raw_super,
e8240f65 3061 int *valid_super_block, int *recovery)
14d7e9de 3062{
df728b0f 3063 struct super_block *sb = sbi->sb;
2b39e907 3064 int block;
e8240f65 3065 struct buffer_head *bh;
fd694733 3066 struct f2fs_super_block *super;
da554e48 3067 int err = 0;
14d7e9de 3068
b39f0de2
YH
3069 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3070 if (!super)
3071 return -ENOMEM;
2b39e907
SL
3072
3073 for (block = 0; block < 2; block++) {
3074 bh = sb_bread(sb, block);
3075 if (!bh) {
dcbb4c10
JP
3076 f2fs_err(sbi, "Unable to read %dth superblock",
3077 block + 1);
2b39e907 3078 err = -EIO;
ed352042 3079 *recovery = 1;
2b39e907
SL
3080 continue;
3081 }
14d7e9de 3082
2b39e907 3083 /* sanity checking of raw super */
38fb6d0e
IZ
3084 err = sanity_check_raw_super(sbi, bh);
3085 if (err) {
dcbb4c10
JP
3086 f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3087 block + 1);
2b39e907 3088 brelse(bh);
ed352042 3089 *recovery = 1;
2b39e907
SL
3090 continue;
3091 }
14d7e9de 3092
2b39e907 3093 if (!*raw_super) {
fd694733
JK
3094 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3095 sizeof(*super));
2b39e907
SL
3096 *valid_super_block = block;
3097 *raw_super = super;
3098 }
3099 brelse(bh);
da554e48 3100 }
3101
da554e48 3102 /* No valid superblock */
2b39e907 3103 if (!*raw_super)
5222595d 3104 kvfree(super);
2b39e907
SL
3105 else
3106 err = 0;
da554e48 3107
2b39e907 3108 return err;
14d7e9de 3109}
3110
fd694733 3111int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
26d815ad 3112{
5d909cdb 3113 struct buffer_head *bh;
d440c52d 3114 __u32 crc = 0;
26d815ad
JK
3115 int err;
3116
df728b0f
JK
3117 if ((recover && f2fs_readonly(sbi->sb)) ||
3118 bdev_read_only(sbi->sb->s_bdev)) {
3119 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
f2353d7b 3120 return -EROFS;
df728b0f 3121 }
f2353d7b 3122
d440c52d 3123 /* we should update superblock crc here */
7beb01f7 3124 if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
d440c52d
JZ
3125 crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
3126 offsetof(struct f2fs_super_block, crc));
3127 F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
3128 }
3129
fd694733 3130 /* write back-up superblock first */
0964fc1a 3131 bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
5d909cdb
JK
3132 if (!bh)
3133 return -EIO;
fd694733 3134 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
5d909cdb 3135 brelse(bh);
c5bda1c8
CY
3136
3137 /* if we are in recovery path, skip writing valid superblock */
3138 if (recover || err)
5d909cdb 3139 return err;
26d815ad
JK
3140
3141 /* write current valid superblock */
0964fc1a 3142 bh = sb_bread(sbi->sb, sbi->valid_super_block);
fd694733
JK
3143 if (!bh)
3144 return -EIO;
3145 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3146 brelse(bh);
3147 return err;
26d815ad
JK
3148}
3149
3c62be17
JK
3150static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
3151{
3152 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
7bb3a371 3153 unsigned int max_devices = MAX_DEVICES;
3c62be17
JK
3154 int i;
3155
7bb3a371
MS
3156 /* Initialize single device information */
3157 if (!RDEV(0).path[0]) {
3158 if (!bdev_is_zoned(sbi->sb->s_bdev))
3c62be17 3159 return 0;
7bb3a371
MS
3160 max_devices = 1;
3161 }
3c62be17 3162
7bb3a371
MS
3163 /*
3164 * Initialize multiple devices information, or single
3165 * zoned block device information.
3166 */
026f0507
KC
3167 sbi->devs = f2fs_kzalloc(sbi,
3168 array_size(max_devices,
3169 sizeof(struct f2fs_dev_info)),
3170 GFP_KERNEL);
7bb3a371
MS
3171 if (!sbi->devs)
3172 return -ENOMEM;
3c62be17 3173
7bb3a371 3174 for (i = 0; i < max_devices; i++) {
3c62be17 3175
7bb3a371
MS
3176 if (i > 0 && !RDEV(i).path[0])
3177 break;
3178
3179 if (max_devices == 1) {
3180 /* Single zoned block device mount */
3181 FDEV(0).bdev =
3182 blkdev_get_by_dev(sbi->sb->s_bdev->bd_dev,
3c62be17 3183 sbi->sb->s_mode, sbi->sb->s_type);
7bb3a371
MS
3184 } else {
3185 /* Multi-device mount */
3186 memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
3187 FDEV(i).total_segments =
3188 le32_to_cpu(RDEV(i).total_segments);
3189 if (i == 0) {
3190 FDEV(i).start_blk = 0;
3191 FDEV(i).end_blk = FDEV(i).start_blk +
3192 (FDEV(i).total_segments <<
3193 sbi->log_blocks_per_seg) - 1 +
3194 le32_to_cpu(raw_super->segment0_blkaddr);
3195 } else {
3196 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
3197 FDEV(i).end_blk = FDEV(i).start_blk +
3198 (FDEV(i).total_segments <<
3199 sbi->log_blocks_per_seg) - 1;
3200 }
3201 FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
3c62be17 3202 sbi->sb->s_mode, sbi->sb->s_type);
7bb3a371 3203 }
3c62be17
JK
3204 if (IS_ERR(FDEV(i).bdev))
3205 return PTR_ERR(FDEV(i).bdev);
3206
3207 /* to release errored devices */
3208 sbi->s_ndevs = i + 1;
3209
3210#ifdef CONFIG_BLK_DEV_ZONED
3211 if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
7beb01f7 3212 !f2fs_sb_has_blkzoned(sbi)) {
dcbb4c10 3213 f2fs_err(sbi, "Zoned block device feature not enabled\n");
3c62be17
JK
3214 return -EINVAL;
3215 }
3216 if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
3217 if (init_blkz_info(sbi, i)) {
dcbb4c10 3218 f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
3c62be17
JK
3219 return -EINVAL;
3220 }
7bb3a371
MS
3221 if (max_devices == 1)
3222 break;
dcbb4c10
JP
3223 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
3224 i, FDEV(i).path,
3225 FDEV(i).total_segments,
3226 FDEV(i).start_blk, FDEV(i).end_blk,
3227 bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
3228 "Host-aware" : "Host-managed");
3c62be17
JK
3229 continue;
3230 }
3231#endif
dcbb4c10
JP
3232 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
3233 i, FDEV(i).path,
3234 FDEV(i).total_segments,
3235 FDEV(i).start_blk, FDEV(i).end_blk);
3236 }
3237 f2fs_info(sbi,
3238 "IO Block Size: %8d KB", F2FS_IO_SIZE_KB(sbi));
3c62be17
JK
3239 return 0;
3240}
3241
5aba5430
DR
3242static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
3243{
3244#ifdef CONFIG_UNICODE
3245 if (f2fs_sb_has_casefold(sbi) && !sbi->s_encoding) {
3246 const struct f2fs_sb_encodings *encoding_info;
3247 struct unicode_map *encoding;
3248 __u16 encoding_flags;
3249
3250 if (f2fs_sb_has_encrypt(sbi)) {
3251 f2fs_err(sbi,
3252 "Can't mount with encoding and encryption");
3253 return -EINVAL;
3254 }
3255
3256 if (f2fs_sb_read_encoding(sbi->raw_super, &encoding_info,
3257 &encoding_flags)) {
3258 f2fs_err(sbi,
3259 "Encoding requested by superblock is unknown");
3260 return -EINVAL;
3261 }
3262
3263 encoding = utf8_load(encoding_info->version);
3264 if (IS_ERR(encoding)) {
3265 f2fs_err(sbi,
3266 "can't mount with superblock charset: %s-%s "
3267 "not supported by the kernel. flags: 0x%x.",
3268 encoding_info->name, encoding_info->version,
3269 encoding_flags);
3270 return PTR_ERR(encoding);
3271 }
3272 f2fs_info(sbi, "Using encoding defined by superblock: "
3273 "%s-%s with flags 0x%hx", encoding_info->name,
3274 encoding_info->version?:"\b", encoding_flags);
3275
3276 sbi->s_encoding = encoding;
3277 sbi->s_encoding_flags = encoding_flags;
2c2eb7a3 3278 sbi->sb->s_d_op = &f2fs_dentry_ops;
5aba5430
DR
3279 }
3280#else
3281 if (f2fs_sb_has_casefold(sbi)) {
3282 f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
3283 return -EINVAL;
3284 }
3285#endif
3286 return 0;
3287}
3288
84b89e5d
JK
3289static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
3290{
3291 struct f2fs_sm_info *sm_i = SM_I(sbi);
3292
3293 /* adjust parameters according to the volume size */
3294 if (sm_i->main_segments <= SMALL_VOLUME_SEGMENTS) {
63189b78 3295 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
84b89e5d
JK
3296 sm_i->dcc_info->discard_granularity = 1;
3297 sm_i->ipu_policy = 1 << F2FS_IPU_FORCE;
3298 }
4cac90d5
CY
3299
3300 sbi->readdir_ra = 1;
84b89e5d
JK
3301}
3302
aff063e2
JK
3303static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
3304{
3305 struct f2fs_sb_info *sbi;
da554e48 3306 struct f2fs_super_block *raw_super;
aff063e2 3307 struct inode *root;
99e3e858 3308 int err;
aa2c8c43 3309 bool skip_recovery = false, need_fsck = false;
dabc4a5c 3310 char *options = NULL;
e8240f65 3311 int recovery, i, valid_super_block;
8f1dbbbb 3312 struct curseg_info *seg_i;
aa2c8c43 3313 int retry_cnt = 1;
aff063e2 3314
ed2e621a 3315try_onemore:
da554e48 3316 err = -EINVAL;
3317 raw_super = NULL;
e8240f65 3318 valid_super_block = -1;
da554e48 3319 recovery = 0;
3320
aff063e2
JK
3321 /* allocate memory for f2fs-specific super block info */
3322 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
3323 if (!sbi)
3324 return -ENOMEM;
3325
df728b0f
JK
3326 sbi->sb = sb;
3327
43b6573b
KM
3328 /* Load the checksum driver */
3329 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
3330 if (IS_ERR(sbi->s_chksum_driver)) {
dcbb4c10 3331 f2fs_err(sbi, "Cannot load crc32 driver.");
43b6573b
KM
3332 err = PTR_ERR(sbi->s_chksum_driver);
3333 sbi->s_chksum_driver = NULL;
3334 goto free_sbi;
3335 }
3336
ff9234ad 3337 /* set a block size */
6bacf52f 3338 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
dcbb4c10 3339 f2fs_err(sbi, "unable to set blocksize");
aff063e2 3340 goto free_sbi;
a07ef784 3341 }
aff063e2 3342
df728b0f 3343 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
e8240f65 3344 &recovery);
9076a75f
GZ
3345 if (err)
3346 goto free_sbi;
3347
5fb08372 3348 sb->s_fs_info = sbi;
52763a4b
JK
3349 sbi->raw_super = raw_super;
3350
704956ec 3351 /* precompute checksum seed for metadata */
7beb01f7 3352 if (f2fs_sb_has_inode_chksum(sbi))
704956ec
CY
3353 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
3354 sizeof(raw_super->uuid));
3355
d1b959c8
DLM
3356 /*
3357 * The BLKZONED feature indicates that the drive was formatted with
3358 * zone alignment optimization. This is optional for host-aware
3359 * devices, but mandatory for host-managed zoned block devices.
3360 */
3361#ifndef CONFIG_BLK_DEV_ZONED
7beb01f7 3362 if (f2fs_sb_has_blkzoned(sbi)) {
dcbb4c10 3363 f2fs_err(sbi, "Zoned block device support is not enabled");
1727f317 3364 err = -EOPNOTSUPP;
d1b959c8
DLM
3365 goto free_sb_buf;
3366 }
d1b959c8 3367#endif
498c5e9f 3368 default_options(sbi);
aff063e2 3369 /* parse mount options */
dabc4a5c
JK
3370 options = kstrdup((const char *)data, GFP_KERNEL);
3371 if (data && !options) {
3372 err = -ENOMEM;
aff063e2 3373 goto free_sb_buf;
dabc4a5c
JK
3374 }
3375
3376 err = parse_options(sb, options);
3377 if (err)
3378 goto free_options;
aff063e2 3379
e0afc4d6
CY
3380 sbi->max_file_blocks = max_file_blocks();
3381 sb->s_maxbytes = sbi->max_file_blocks <<
3382 le32_to_cpu(raw_super->log_blocksize);
aff063e2 3383 sb->s_max_links = F2FS_LINK_MAX;
aff063e2 3384
5aba5430
DR
3385 err = f2fs_setup_casefold(sbi);
3386 if (err)
3387 goto free_options;
3388
0abd675e
CY
3389#ifdef CONFIG_QUOTA
3390 sb->dq_op = &f2fs_quota_operations;
bc88ac96 3391 sb->s_qcop = &f2fs_quotactl_ops;
5c57132e 3392 sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
292c196a 3393
7beb01f7 3394 if (f2fs_sb_has_quota_ino(sbi)) {
292c196a
CY
3395 for (i = 0; i < MAXQUOTAS; i++) {
3396 if (f2fs_qf_ino(sbi->sb, i))
3397 sbi->nquota_files++;
3398 }
3399 }
0abd675e
CY
3400#endif
3401
aff063e2 3402 sb->s_op = &f2fs_sops;
643fa961 3403#ifdef CONFIG_FS_ENCRYPTION
0b81d077 3404 sb->s_cop = &f2fs_cryptops;
95ae251f
EB
3405#endif
3406#ifdef CONFIG_FS_VERITY
3407 sb->s_vop = &f2fs_verityops;
ffcc4182 3408#endif
aff063e2
JK
3409 sb->s_xattr = f2fs_xattr_handlers;
3410 sb->s_export_op = &f2fs_export_ops;
3411 sb->s_magic = F2FS_SUPER_MAGIC;
aff063e2 3412 sb->s_time_gran = 1;
1751e8a6
LT
3413 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
3414 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
85787090 3415 memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
578c6478 3416 sb->s_iflags |= SB_I_CGROUPWB;
aff063e2
JK
3417
3418 /* init f2fs-specific super block info */
e8240f65 3419 sbi->valid_super_block = valid_super_block;
fb24fea7 3420 init_rwsem(&sbi->gc_lock);
853137ce 3421 mutex_init(&sbi->writepages);
aff063e2 3422 mutex_init(&sbi->cp_mutex);
b3582c68 3423 init_rwsem(&sbi->node_write);
59c9081b 3424 init_rwsem(&sbi->node_change);
315df839
JK
3425
3426 /* disallow all the data/node/meta page writes */
3427 set_sbi_flag(sbi, SBI_POR_DOING);
aff063e2 3428 spin_lock_init(&sbi->stat_lock);
971767ca 3429
b0af6d49
CY
3430 /* init iostat info */
3431 spin_lock_init(&sbi->iostat_lock);
3432 sbi->iostat_enable = false;
2bc4bea3 3433 sbi->iostat_period_ms = DEFAULT_IOSTAT_PERIOD_MS;
b0af6d49 3434
458e6197 3435 for (i = 0; i < NR_PAGE_TYPE; i++) {
a912b54d
JK
3436 int n = (i == META) ? 1: NR_TEMP_TYPE;
3437 int j;
3438
c8606593
KC
3439 sbi->write_io[i] =
3440 f2fs_kmalloc(sbi,
3441 array_size(n,
3442 sizeof(struct f2fs_bio_info)),
3443 GFP_KERNEL);
b63def91
CJ
3444 if (!sbi->write_io[i]) {
3445 err = -ENOMEM;
0b2103e8 3446 goto free_bio_info;
b63def91 3447 }
a912b54d
JK
3448
3449 for (j = HOT; j < n; j++) {
3450 init_rwsem(&sbi->write_io[i][j].io_rwsem);
3451 sbi->write_io[i][j].sbi = sbi;
3452 sbi->write_io[i][j].bio = NULL;
fb830fc5
CY
3453 spin_lock_init(&sbi->write_io[i][j].io_lock);
3454 INIT_LIST_HEAD(&sbi->write_io[i][j].io_list);
0b20fcec
CY
3455 INIT_LIST_HEAD(&sbi->write_io[i][j].bio_list);
3456 init_rwsem(&sbi->write_io[i][j].bio_list_lock);
a912b54d 3457 }
458e6197 3458 }
971767ca 3459
b873b798 3460 init_rwsem(&sbi->cp_rwsem);
db6ec53b 3461 init_rwsem(&sbi->quota_sem);
fb51b5ef 3462 init_waitqueue_head(&sbi->cp_wait);
aff063e2
JK
3463 init_sb_info(sbi);
3464
523be8a6
JK
3465 err = init_percpu_info(sbi);
3466 if (err)
d7997e63 3467 goto free_bio_info;
523be8a6 3468
c72db71e 3469 if (F2FS_IO_ALIGNED(sbi)) {
0a595eba 3470 sbi->write_io_dummy =
a3ebfe4f 3471 mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
1727f317
CY
3472 if (!sbi->write_io_dummy) {
3473 err = -ENOMEM;
d7997e63 3474 goto free_percpu;
1727f317 3475 }
0a595eba
JK
3476 }
3477
a999150f
CY
3478 /* init per sbi slab cache */
3479 err = f2fs_init_xattr_caches(sbi);
3480 if (err)
3481 goto free_io_dummy;
3482
aff063e2
JK
3483 /* get an inode for meta space */
3484 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
3485 if (IS_ERR(sbi->meta_inode)) {
dcbb4c10 3486 f2fs_err(sbi, "Failed to read F2FS meta data inode");
aff063e2 3487 err = PTR_ERR(sbi->meta_inode);
a999150f 3488 goto free_xattr_cache;
aff063e2
JK
3489 }
3490
4d57b86d 3491 err = f2fs_get_valid_checkpoint(sbi);
a07ef784 3492 if (err) {
dcbb4c10 3493 f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
aff063e2 3494 goto free_meta_inode;
a07ef784 3495 }
aff063e2 3496
af033b2a
CY
3497 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
3498 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
db610a64
JK
3499 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
3500 set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
3501 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
3502 }
af033b2a 3503
04f0b2ea
QS
3504 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
3505 set_sbi_flag(sbi, SBI_NEED_FSCK);
3506
3c62be17
JK
3507 /* Initialize device list */
3508 err = f2fs_scan_devices(sbi);
3509 if (err) {
dcbb4c10 3510 f2fs_err(sbi, "Failed to find devices");
3c62be17
JK
3511 goto free_devices;
3512 }
3513
4c8ff709
CY
3514 err = f2fs_init_post_read_wq(sbi);
3515 if (err) {
3516 f2fs_err(sbi, "Failed to initialize post read workqueue");
3517 goto free_devices;
3518 }
3519
aff063e2
JK
3520 sbi->total_valid_node_count =
3521 le32_to_cpu(sbi->ckpt->valid_node_count);
513c5f37
JK
3522 percpu_counter_set(&sbi->total_valid_inode_count,
3523 le32_to_cpu(sbi->ckpt->valid_inode_count));
aff063e2
JK
3524 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
3525 sbi->total_valid_block_count =
3526 le64_to_cpu(sbi->ckpt->valid_block_count);
3527 sbi->last_valid_block_count = sbi->total_valid_block_count;
daeb433e 3528 sbi->reserved_blocks = 0;
80d42145 3529 sbi->current_reserved_blocks = 0;
7e65be49 3530 limit_reserve_root(sbi);
1ae18f71 3531 adjust_unusable_cap_perc(sbi);
41382ec4 3532
c227f912
CY
3533 for (i = 0; i < NR_INODE_TYPE; i++) {
3534 INIT_LIST_HEAD(&sbi->inode_list[i]);
3535 spin_lock_init(&sbi->inode_lock[i]);
3536 }
040d2bb3 3537 mutex_init(&sbi->flush_lock);
aff063e2 3538
4d57b86d 3539 f2fs_init_extent_cache_info(sbi);
1dcc336b 3540
4d57b86d 3541 f2fs_init_ino_entry_info(sbi);
aff063e2 3542
50fa53ec
CY
3543 f2fs_init_fsync_node_info(sbi);
3544
aff063e2 3545 /* setup f2fs internal modules */
4d57b86d 3546 err = f2fs_build_segment_manager(sbi);
a07ef784 3547 if (err) {
dcbb4c10
JP
3548 f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
3549 err);
aff063e2 3550 goto free_sm;
a07ef784 3551 }
4d57b86d 3552 err = f2fs_build_node_manager(sbi);
a07ef784 3553 if (err) {
dcbb4c10
JP
3554 f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
3555 err);
aff063e2 3556 goto free_nm;
a07ef784 3557 }
aff063e2 3558
8f1dbbbb
SL
3559 /* For write statistics */
3560 if (sb->s_bdev->bd_part)
3561 sbi->sectors_written_start =
dbae2c55
MC
3562 (u64)part_stat_read(sb->s_bdev->bd_part,
3563 sectors[STAT_WRITE]);
8f1dbbbb
SL
3564
3565 /* Read accumulated write IO statistics if exists */
3566 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
3567 if (__exist_node_summaries(sbi))
3568 sbi->kbytes_written =
b2dde6fc 3569 le64_to_cpu(seg_i->journal->info.kbytes_written);
8f1dbbbb 3570
4d57b86d 3571 f2fs_build_gc_manager(sbi);
aff063e2 3572
60aa4d55
ST
3573 err = f2fs_build_stats(sbi);
3574 if (err)
3575 goto free_nm;
3576
aff063e2
JK
3577 /* get an inode for node space */
3578 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
3579 if (IS_ERR(sbi->node_inode)) {
dcbb4c10 3580 f2fs_err(sbi, "Failed to read node inode");
aff063e2 3581 err = PTR_ERR(sbi->node_inode);
60aa4d55 3582 goto free_stats;
aff063e2
JK
3583 }
3584
aff063e2
JK
3585 /* read root inode and dentry */
3586 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
3587 if (IS_ERR(root)) {
dcbb4c10 3588 f2fs_err(sbi, "Failed to read root inode");
aff063e2 3589 err = PTR_ERR(root);
60aa4d55 3590 goto free_node_inode;
aff063e2 3591 }
bcbfbd60
CY
3592 if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
3593 !root->i_size || !root->i_nlink) {
9d847950 3594 iput(root);
8f99a946 3595 err = -EINVAL;
60aa4d55 3596 goto free_node_inode;
8f99a946 3597 }
aff063e2
JK
3598
3599 sb->s_root = d_make_root(root); /* allocate root dentry */
3600 if (!sb->s_root) {
3601 err = -ENOMEM;
025cdb16 3602 goto free_node_inode;
aff063e2
JK
3603 }
3604
dc6b2055 3605 err = f2fs_register_sysfs(sbi);
b59d0bae 3606 if (err)
a398101a 3607 goto free_root_inode;
b59d0bae 3608
ea676733 3609#ifdef CONFIG_QUOTA
76cf05d7 3610 /* Enable quota usage during mount */
7beb01f7 3611 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
ea676733 3612 err = f2fs_enable_quotas(sb);
730746ce 3613 if (err)
dcbb4c10 3614 f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
ea676733
JK
3615 }
3616#endif
7a88ddb5 3617 /* if there are any orphan inodes, free them */
4d57b86d 3618 err = f2fs_recover_orphan_inodes(sbi);
4b2414d0 3619 if (err)
ea676733 3620 goto free_meta;
4b2414d0 3621
4354994f 3622 if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
aa2c8c43 3623 goto reset_checkpoint;
4354994f 3624
6437d1b0 3625 /* recover fsynced data */
a9117eca
CY
3626 if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
3627 !test_opt(sbi, NORECOVERY)) {
081d78c2
JK
3628 /*
3629 * mount should be failed, when device has readonly mode, and
3630 * previous checkpoint was not done by clean system shutdown.
3631 */
b61af314
CY
3632 if (f2fs_hw_is_readonly(sbi)) {
3633 if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
3634 err = -EROFS;
dcbb4c10 3635 f2fs_err(sbi, "Need to recover fsync data, but write access unavailable");
b61af314
CY
3636 goto free_meta;
3637 }
dcbb4c10 3638 f2fs_info(sbi, "write access unavailable, skipping recovery");
b61af314 3639 goto reset_checkpoint;
081d78c2 3640 }
2adc3505
CY
3641
3642 if (need_fsck)
3643 set_sbi_flag(sbi, SBI_NEED_FSCK);
3644
aa2c8c43
CY
3645 if (skip_recovery)
3646 goto reset_checkpoint;
a468f0ef 3647
4d57b86d 3648 err = f2fs_recover_fsync_data(sbi, false);
6781eabb 3649 if (err < 0) {
aa2c8c43
CY
3650 if (err != -ENOMEM)
3651 skip_recovery = true;
2adc3505 3652 need_fsck = true;
dcbb4c10
JP
3653 f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
3654 err);
4b2414d0 3655 goto free_meta;
ed2e621a 3656 }
6781eabb 3657 } else {
4d57b86d 3658 err = f2fs_recover_fsync_data(sbi, true);
6781eabb
JK
3659
3660 if (!f2fs_readonly(sb) && err > 0) {
3661 err = -EINVAL;
dcbb4c10 3662 f2fs_err(sbi, "Need to recover fsync data");
ea676733 3663 goto free_meta;
6781eabb 3664 }
6437d1b0 3665 }
d508c94e
SK
3666
3667 /*
3668 * If the f2fs is not readonly and fsync data recovery succeeds,
3669 * check zoned block devices' write pointer consistency.
3670 */
3671 if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
3672 err = f2fs_check_write_pointer(sbi);
3673 if (err)
3674 goto free_meta;
3675 }
3676
aa2c8c43 3677reset_checkpoint:
4d57b86d 3678 /* f2fs_recover_fsync_data() cleared this already */
315df839 3679 clear_sbi_flag(sbi, SBI_POR_DOING);
b59d0bae 3680
4354994f
DR
3681 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
3682 err = f2fs_disable_checkpoint(sbi);
3683 if (err)
812a9597 3684 goto sync_free_meta;
4354994f
DR
3685 } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
3686 f2fs_enable_checkpoint(sbi);
3687 }
3688
6437d1b0
JK
3689 /*
3690 * If filesystem is not mounted as read-only then
3691 * do start the gc_thread.
3692 */
bbbc34fd 3693 if (F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF && !f2fs_readonly(sb)) {
6437d1b0 3694 /* After POR, we can run background GC thread.*/
4d57b86d 3695 err = f2fs_start_gc_thread(sbi);
6437d1b0 3696 if (err)
812a9597 3697 goto sync_free_meta;
6437d1b0 3698 }
5222595d 3699 kvfree(options);
da554e48 3700
3701 /* recover broken superblock */
f2353d7b 3702 if (recovery) {
41214b3c 3703 err = f2fs_commit_super(sbi, true);
dcbb4c10
JP
3704 f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
3705 sbi->valid_super_block ? 1 : 2, err);
da554e48 3706 }
3707
bae01eda
CY
3708 f2fs_join_shrinker(sbi);
3709
84b89e5d
JK
3710 f2fs_tuning_parameters(sbi);
3711
dcbb4c10
JP
3712 f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
3713 cur_cp_version(F2FS_CKPT(sbi)));
6beceb54 3714 f2fs_update_time(sbi, CP_TIME);
d0239e1b 3715 f2fs_update_time(sbi, REQ_TIME);
db610a64 3716 clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
aff063e2 3717 return 0;
6437d1b0 3718
812a9597
JK
3719sync_free_meta:
3720 /* safe to flush all the data */
3721 sync_filesystem(sbi->sb);
aa2c8c43 3722 retry_cnt = 0;
812a9597 3723
4b2414d0 3724free_meta:
ea676733 3725#ifdef CONFIG_QUOTA
26b5a079 3726 f2fs_truncate_quota_inode_pages(sb);
7beb01f7 3727 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
ea676733
JK
3728 f2fs_quota_off_umount(sbi->sb);
3729#endif
4b2414d0 3730 /*
4d57b86d 3731 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
4b2414d0 3732 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
4d57b86d
CY
3733 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
3734 * falls into an infinite loop in f2fs_sync_meta_pages().
4b2414d0
CY
3735 */
3736 truncate_inode_pages_final(META_MAPPING(sbi));
812a9597
JK
3737 /* evict some inodes being cached by GC */
3738 evict_inodes(sb);
dc6b2055 3739 f2fs_unregister_sysfs(sbi);
aff063e2
JK
3740free_root_inode:
3741 dput(sb->s_root);
3742 sb->s_root = NULL;
3743free_node_inode:
4d57b86d 3744 f2fs_release_ino_entry(sbi, true);
bae01eda 3745 truncate_inode_pages_final(NODE_MAPPING(sbi));
aff063e2 3746 iput(sbi->node_inode);
7c77bf7d 3747 sbi->node_inode = NULL;
60aa4d55
ST
3748free_stats:
3749 f2fs_destroy_stats(sbi);
aff063e2 3750free_nm:
4d57b86d 3751 f2fs_destroy_node_manager(sbi);
aff063e2 3752free_sm:
4d57b86d 3753 f2fs_destroy_segment_manager(sbi);
4c8ff709 3754 f2fs_destroy_post_read_wq(sbi);
3c62be17
JK
3755free_devices:
3756 destroy_device_list(sbi);
5222595d 3757 kvfree(sbi->ckpt);
aff063e2
JK
3758free_meta_inode:
3759 make_bad_inode(sbi->meta_inode);
3760 iput(sbi->meta_inode);
7c77bf7d 3761 sbi->meta_inode = NULL;
a999150f
CY
3762free_xattr_cache:
3763 f2fs_destroy_xattr_caches(sbi);
0a595eba
JK
3764free_io_dummy:
3765 mempool_destroy(sbi->write_io_dummy);
d7997e63
CY
3766free_percpu:
3767 destroy_percpu_info(sbi);
3768free_bio_info:
a912b54d 3769 for (i = 0; i < NR_PAGE_TYPE; i++)
5222595d 3770 kvfree(sbi->write_io[i]);
5aba5430
DR
3771
3772#ifdef CONFIG_UNICODE
3773 utf8_unload(sbi->s_encoding);
3774#endif
d7997e63 3775free_options:
4b2414d0
CY
3776#ifdef CONFIG_QUOTA
3777 for (i = 0; i < MAXQUOTAS; i++)
5222595d 3778 kvfree(F2FS_OPTION(sbi).s_qf_names[i]);
4b2414d0 3779#endif
5222595d 3780 kvfree(options);
aff063e2 3781free_sb_buf:
5222595d 3782 kvfree(raw_super);
aff063e2 3783free_sbi:
43b6573b
KM
3784 if (sbi->s_chksum_driver)
3785 crypto_free_shash(sbi->s_chksum_driver);
5222595d 3786 kvfree(sbi);
ed2e621a
JK
3787
3788 /* give only one another chance */
aa2c8c43
CY
3789 if (retry_cnt > 0 && skip_recovery) {
3790 retry_cnt--;
ed2e621a
JK
3791 shrink_dcache_sb(sb);
3792 goto try_onemore;
3793 }
aff063e2
JK
3794 return err;
3795}
3796
3797static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
3798 const char *dev_name, void *data)
3799{
3800 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
3801}
3802
30a5537f
JK
3803static void kill_f2fs_super(struct super_block *sb)
3804{
cce13252 3805 if (sb->s_root) {
1cb50f87
JK
3806 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3807
3808 set_sbi_flag(sbi, SBI_IS_CLOSE);
3809 f2fs_stop_gc_thread(sbi);
3810 f2fs_stop_discard_thread(sbi);
3811
3812 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
3813 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
3814 struct cp_control cpc = {
3815 .reason = CP_UMOUNT,
3816 };
3817 f2fs_write_checkpoint(sbi, &cpc);
3818 }
1378752b
CY
3819
3820 if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
3821 sb->s_flags &= ~SB_RDONLY;
cce13252 3822 }
30a5537f
JK
3823 kill_block_super(sb);
3824}
3825
aff063e2
JK
3826static struct file_system_type f2fs_fs_type = {
3827 .owner = THIS_MODULE,
3828 .name = "f2fs",
3829 .mount = f2fs_mount,
30a5537f 3830 .kill_sb = kill_f2fs_super,
aff063e2
JK
3831 .fs_flags = FS_REQUIRES_DEV,
3832};
7f78e035 3833MODULE_ALIAS_FS("f2fs");
aff063e2 3834
6e6093a8 3835static int __init init_inodecache(void)
aff063e2 3836{
5d097056
VD
3837 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
3838 sizeof(struct f2fs_inode_info), 0,
3839 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
6bacf52f 3840 if (!f2fs_inode_cachep)
aff063e2
JK
3841 return -ENOMEM;
3842 return 0;
3843}
3844
3845static void destroy_inodecache(void)
3846{
3847 /*
3848 * Make sure all delayed rcu free inodes are flushed before we
3849 * destroy cache.
3850 */
3851 rcu_barrier();
3852 kmem_cache_destroy(f2fs_inode_cachep);
3853}
3854
3855static int __init init_f2fs_fs(void)
3856{
3857 int err;
3858
4071e67c
AP
3859 if (PAGE_SIZE != F2FS_BLKSIZE) {
3860 printk("F2FS not supported on PAGE_SIZE(%lu) != %d\n",
3861 PAGE_SIZE, F2FS_BLKSIZE);
3862 return -EINVAL;
3863 }
3864
c0508650
JK
3865 f2fs_build_trace_ios();
3866
aff063e2
JK
3867 err = init_inodecache();
3868 if (err)
3869 goto fail;
4d57b86d 3870 err = f2fs_create_node_manager_caches();
aff063e2 3871 if (err)
9890ff3f 3872 goto free_inodecache;
4d57b86d 3873 err = f2fs_create_segment_manager_caches();
aff063e2 3874 if (err)
9890ff3f 3875 goto free_node_manager_caches;
4d57b86d 3876 err = f2fs_create_checkpoint_caches();
aff063e2 3877 if (err)
06292073 3878 goto free_segment_manager_caches;
4d57b86d 3879 err = f2fs_create_extent_cache();
1dcc336b
CY
3880 if (err)
3881 goto free_checkpoint_caches;
dc6b2055 3882 err = f2fs_init_sysfs();
a398101a 3883 if (err)
1dcc336b 3884 goto free_extent_cache;
2658e50d 3885 err = register_shrinker(&f2fs_shrinker_info);
cfc4d971 3886 if (err)
a398101a 3887 goto free_sysfs;
2658e50d
JK
3888 err = register_filesystem(&f2fs_fs_type);
3889 if (err)
3890 goto free_shrinker;
21acc07d 3891 f2fs_create_root_stats();
6dbb1796
EB
3892 err = f2fs_init_post_read_processing();
3893 if (err)
3894 goto free_root_stats;
0b20fcec
CY
3895 err = f2fs_init_bio_entry_cache();
3896 if (err)
3897 goto free_post_read;
f543805f
CY
3898 err = f2fs_init_bioset();
3899 if (err)
3900 goto free_bio_enrty_cache;
5e6bbde9
CY
3901 err = f2fs_init_compress_mempool();
3902 if (err)
3903 goto free_bioset;
9890ff3f 3904 return 0;
5e6bbde9
CY
3905free_bioset:
3906 f2fs_destroy_bioset();
f543805f
CY
3907free_bio_enrty_cache:
3908 f2fs_destroy_bio_entry_cache();
0b20fcec
CY
3909free_post_read:
3910 f2fs_destroy_post_read_processing();
6dbb1796
EB
3911free_root_stats:
3912 f2fs_destroy_root_stats();
787c7b8c 3913 unregister_filesystem(&f2fs_fs_type);
2658e50d
JK
3914free_shrinker:
3915 unregister_shrinker(&f2fs_shrinker_info);
a398101a 3916free_sysfs:
dc6b2055 3917 f2fs_exit_sysfs();
1dcc336b 3918free_extent_cache:
4d57b86d 3919 f2fs_destroy_extent_cache();
9890ff3f 3920free_checkpoint_caches:
4d57b86d 3921 f2fs_destroy_checkpoint_caches();
7fd9e544 3922free_segment_manager_caches:
4d57b86d 3923 f2fs_destroy_segment_manager_caches();
9890ff3f 3924free_node_manager_caches:
4d57b86d 3925 f2fs_destroy_node_manager_caches();
9890ff3f
ZH
3926free_inodecache:
3927 destroy_inodecache();
aff063e2
JK
3928fail:
3929 return err;
3930}
3931
3932static void __exit exit_f2fs_fs(void)
3933{
5e6bbde9 3934 f2fs_destroy_compress_mempool();
f543805f 3935 f2fs_destroy_bioset();
0b20fcec 3936 f2fs_destroy_bio_entry_cache();
6dbb1796 3937 f2fs_destroy_post_read_processing();
4589d25d 3938 f2fs_destroy_root_stats();
aff063e2 3939 unregister_filesystem(&f2fs_fs_type);
b8bef79d 3940 unregister_shrinker(&f2fs_shrinker_info);
dc6b2055 3941 f2fs_exit_sysfs();
4d57b86d
CY
3942 f2fs_destroy_extent_cache();
3943 f2fs_destroy_checkpoint_caches();
3944 f2fs_destroy_segment_manager_caches();
3945 f2fs_destroy_node_manager_caches();
aff063e2 3946 destroy_inodecache();
351f4fba 3947 f2fs_destroy_trace_ios();
aff063e2
JK
3948}
3949
3950module_init(init_f2fs_fs)
3951module_exit(exit_f2fs_fs)
3952
3953MODULE_AUTHOR("Samsung Electronics's Praesto Team");
3954MODULE_DESCRIPTION("Flash Friendly File System");
3955MODULE_LICENSE("GPL");
b4b9d34c 3956