]> git.ipfire.org Git - thirdparty/kernel/linux.git/blob - fs/f2fs/super.c
projects / thirdparty / kernel / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge tag 's390-6.8-1' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
[thirdparty/kernel/linux.git] / fs / f2fs / super.c
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * fs/f2fs/super.c
4 *
5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6 * http://www.samsung.com/
7 */
8 #include <linux/module.h>
9 #include <linux/init.h>
10 #include <linux/fs.h>
11 #include <linux/fs_context.h>
12 #include <linux/sched/mm.h>
13 #include <linux/statfs.h>
14 #include <linux/buffer_head.h>
15 #include <linux/kthread.h>
16 #include <linux/parser.h>
17 #include <linux/mount.h>
18 #include <linux/seq_file.h>
19 #include <linux/proc_fs.h>
20 #include <linux/random.h>
21 #include <linux/exportfs.h>
22 #include <linux/blkdev.h>
23 #include <linux/quotaops.h>
24 #include <linux/f2fs_fs.h>
25 #include <linux/sysfs.h>
26 #include <linux/quota.h>
27 #include <linux/unicode.h>
28 #include <linux/part_stat.h>
29 #include <linux/zstd.h>
30 #include <linux/lz4.h>
31
32 #include "f2fs.h"
33 #include "node.h"
34 #include "segment.h"
35 #include "xattr.h"
36 #include "gc.h"
37 #include "iostat.h"
38
39 #define CREATE_TRACE_POINTS
40 #include <trace/events/f2fs.h>
41
42 static struct kmem_cache *f2fs_inode_cachep;
43
44 #ifdef CONFIG_F2FS_FAULT_INJECTION
45
46 const char *f2fs_fault_name[FAULT_MAX] = {
47 [FAULT_KMALLOC] = "kmalloc",
48 [FAULT_KVMALLOC] = "kvmalloc",
49 [FAULT_PAGE_ALLOC] = "page alloc",
50 [FAULT_PAGE_GET] = "page get",
51 [FAULT_ALLOC_NID] = "alloc nid",
52 [FAULT_ORPHAN] = "orphan",
53 [FAULT_BLOCK] = "no more block",
54 [FAULT_DIR_DEPTH] = "too big dir depth",
55 [FAULT_EVICT_INODE] = "evict_inode fail",
56 [FAULT_TRUNCATE] = "truncate fail",
57 [FAULT_READ_IO] = "read IO error",
58 [FAULT_CHECKPOINT] = "checkpoint error",
59 [FAULT_DISCARD] = "discard error",
60 [FAULT_WRITE_IO] = "write IO error",
61 [FAULT_SLAB_ALLOC] = "slab alloc",
62 [FAULT_DQUOT_INIT] = "dquot initialize",
63 [FAULT_LOCK_OP] = "lock_op",
64 [FAULT_BLKADDR] = "invalid blkaddr",
65 };
66
67 void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
68 unsigned int type)
69 {
70 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
71
72 if (rate) {
73 atomic_set(&ffi->inject_ops, 0);
74 ffi->inject_rate = rate;
75 }
76
77 if (type)
78 ffi->inject_type = type;
79
80 if (!rate && !type)
81 memset(ffi, 0, sizeof(struct f2fs_fault_info));
82 }
83 #endif
84
85 /* f2fs-wide shrinker description */
86 static struct shrinker *f2fs_shrinker_info;
87
88 static int __init f2fs_init_shrinker(void)
89 {
90 f2fs_shrinker_info = shrinker_alloc(0, "f2fs-shrinker");
91 if (!f2fs_shrinker_info)
92 return -ENOMEM;
93
94 f2fs_shrinker_info->count_objects = f2fs_shrink_count;
95 f2fs_shrinker_info->scan_objects = f2fs_shrink_scan;
96
97 shrinker_register(f2fs_shrinker_info);
98
99 return 0;
100 }
101
102 static void f2fs_exit_shrinker(void)
103 {
104 shrinker_free(f2fs_shrinker_info);
105 }
106
107 enum {
108 Opt_gc_background,
109 Opt_disable_roll_forward,
110 Opt_norecovery,
111 Opt_discard,
112 Opt_nodiscard,
113 Opt_noheap,
114 Opt_heap,
115 Opt_user_xattr,
116 Opt_nouser_xattr,
117 Opt_acl,
118 Opt_noacl,
119 Opt_active_logs,
120 Opt_disable_ext_identify,
121 Opt_inline_xattr,
122 Opt_noinline_xattr,
123 Opt_inline_xattr_size,
124 Opt_inline_data,
125 Opt_inline_dentry,
126 Opt_noinline_dentry,
127 Opt_flush_merge,
128 Opt_noflush_merge,
129 Opt_barrier,
130 Opt_nobarrier,
131 Opt_fastboot,
132 Opt_extent_cache,
133 Opt_noextent_cache,
134 Opt_noinline_data,
135 Opt_data_flush,
136 Opt_reserve_root,
137 Opt_resgid,
138 Opt_resuid,
139 Opt_mode,
140 Opt_io_size_bits,
141 Opt_fault_injection,
142 Opt_fault_type,
143 Opt_lazytime,
144 Opt_nolazytime,
145 Opt_quota,
146 Opt_noquota,
147 Opt_usrquota,
148 Opt_grpquota,
149 Opt_prjquota,
150 Opt_usrjquota,
151 Opt_grpjquota,
152 Opt_prjjquota,
153 Opt_offusrjquota,
154 Opt_offgrpjquota,
155 Opt_offprjjquota,
156 Opt_jqfmt_vfsold,
157 Opt_jqfmt_vfsv0,
158 Opt_jqfmt_vfsv1,
159 Opt_alloc,
160 Opt_fsync,
161 Opt_test_dummy_encryption,
162 Opt_inlinecrypt,
163 Opt_checkpoint_disable,
164 Opt_checkpoint_disable_cap,
165 Opt_checkpoint_disable_cap_perc,
166 Opt_checkpoint_enable,
167 Opt_checkpoint_merge,
168 Opt_nocheckpoint_merge,
169 Opt_compress_algorithm,
170 Opt_compress_log_size,
171 Opt_compress_extension,
172 Opt_nocompress_extension,
173 Opt_compress_chksum,
174 Opt_compress_mode,
175 Opt_compress_cache,
176 Opt_atgc,
177 Opt_gc_merge,
178 Opt_nogc_merge,
179 Opt_discard_unit,
180 Opt_memory_mode,
181 Opt_age_extent_cache,
182 Opt_errors,
183 Opt_err,
184 };
185
186 static match_table_t f2fs_tokens = {
187 {Opt_gc_background, "background_gc=%s"},
188 {Opt_disable_roll_forward, "disable_roll_forward"},
189 {Opt_norecovery, "norecovery"},
190 {Opt_discard, "discard"},
191 {Opt_nodiscard, "nodiscard"},
192 {Opt_noheap, "no_heap"},
193 {Opt_heap, "heap"},
194 {Opt_user_xattr, "user_xattr"},
195 {Opt_nouser_xattr, "nouser_xattr"},
196 {Opt_acl, "acl"},
197 {Opt_noacl, "noacl"},
198 {Opt_active_logs, "active_logs=%u"},
199 {Opt_disable_ext_identify, "disable_ext_identify"},
200 {Opt_inline_xattr, "inline_xattr"},
201 {Opt_noinline_xattr, "noinline_xattr"},
202 {Opt_inline_xattr_size, "inline_xattr_size=%u"},
203 {Opt_inline_data, "inline_data"},
204 {Opt_inline_dentry, "inline_dentry"},
205 {Opt_noinline_dentry, "noinline_dentry"},
206 {Opt_flush_merge, "flush_merge"},
207 {Opt_noflush_merge, "noflush_merge"},
208 {Opt_barrier, "barrier"},
209 {Opt_nobarrier, "nobarrier"},
210 {Opt_fastboot, "fastboot"},
211 {Opt_extent_cache, "extent_cache"},
212 {Opt_noextent_cache, "noextent_cache"},
213 {Opt_noinline_data, "noinline_data"},
214 {Opt_data_flush, "data_flush"},
215 {Opt_reserve_root, "reserve_root=%u"},
216 {Opt_resgid, "resgid=%u"},
217 {Opt_resuid, "resuid=%u"},
218 {Opt_mode, "mode=%s"},
219 {Opt_io_size_bits, "io_bits=%u"},
220 {Opt_fault_injection, "fault_injection=%u"},
221 {Opt_fault_type, "fault_type=%u"},
222 {Opt_lazytime, "lazytime"},
223 {Opt_nolazytime, "nolazytime"},
224 {Opt_quota, "quota"},
225 {Opt_noquota, "noquota"},
226 {Opt_usrquota, "usrquota"},
227 {Opt_grpquota, "grpquota"},
228 {Opt_prjquota, "prjquota"},
229 {Opt_usrjquota, "usrjquota=%s"},
230 {Opt_grpjquota, "grpjquota=%s"},
231 {Opt_prjjquota, "prjjquota=%s"},
232 {Opt_offusrjquota, "usrjquota="},
233 {Opt_offgrpjquota, "grpjquota="},
234 {Opt_offprjjquota, "prjjquota="},
235 {Opt_jqfmt_vfsold, "jqfmt=vfsold"},
236 {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
237 {Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
238 {Opt_alloc, "alloc_mode=%s"},
239 {Opt_fsync, "fsync_mode=%s"},
240 {Opt_test_dummy_encryption, "test_dummy_encryption=%s"},
241 {Opt_test_dummy_encryption, "test_dummy_encryption"},
242 {Opt_inlinecrypt, "inlinecrypt"},
243 {Opt_checkpoint_disable, "checkpoint=disable"},
244 {Opt_checkpoint_disable_cap, "checkpoint=disable:%u"},
245 {Opt_checkpoint_disable_cap_perc, "checkpoint=disable:%u%%"},
246 {Opt_checkpoint_enable, "checkpoint=enable"},
247 {Opt_checkpoint_merge, "checkpoint_merge"},
248 {Opt_nocheckpoint_merge, "nocheckpoint_merge"},
249 {Opt_compress_algorithm, "compress_algorithm=%s"},
250 {Opt_compress_log_size, "compress_log_size=%u"},
251 {Opt_compress_extension, "compress_extension=%s"},
252 {Opt_nocompress_extension, "nocompress_extension=%s"},
253 {Opt_compress_chksum, "compress_chksum"},
254 {Opt_compress_mode, "compress_mode=%s"},
255 {Opt_compress_cache, "compress_cache"},
256 {Opt_atgc, "atgc"},
257 {Opt_gc_merge, "gc_merge"},
258 {Opt_nogc_merge, "nogc_merge"},
259 {Opt_discard_unit, "discard_unit=%s"},
260 {Opt_memory_mode, "memory=%s"},
261 {Opt_age_extent_cache, "age_extent_cache"},
262 {Opt_errors, "errors=%s"},
263 {Opt_err, NULL},
264 };
265
266 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...)
267 {
268 struct va_format vaf;
269 va_list args;
270 int level;
271
272 va_start(args, fmt);
273
274 level = printk_get_level(fmt);
275 vaf.fmt = printk_skip_level(fmt);
276 vaf.va = &args;
277 printk("%c%cF2FS-fs (%s): %pV\n",
278 KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
279
280 va_end(args);
281 }
282
283 #if IS_ENABLED(CONFIG_UNICODE)
284 static const struct f2fs_sb_encodings {
285 __u16 magic;
286 char *name;
287 unsigned int version;
288 } f2fs_sb_encoding_map[] = {
289 {F2FS_ENC_UTF8_12_1, "utf8", UNICODE_AGE(12, 1, 0)},
290 };
291
292 static const struct f2fs_sb_encodings *
293 f2fs_sb_read_encoding(const struct f2fs_super_block *sb)
294 {
295 __u16 magic = le16_to_cpu(sb->s_encoding);
296 int i;
297
298 for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
299 if (magic == f2fs_sb_encoding_map[i].magic)
300 return &f2fs_sb_encoding_map[i];
301
302 return NULL;
303 }
304
305 struct kmem_cache *f2fs_cf_name_slab;
306 static int __init f2fs_create_casefold_cache(void)
307 {
308 f2fs_cf_name_slab = f2fs_kmem_cache_create("f2fs_casefolded_name",
309 F2FS_NAME_LEN);
310 return f2fs_cf_name_slab ? 0 : -ENOMEM;
311 }
312
313 static void f2fs_destroy_casefold_cache(void)
314 {
315 kmem_cache_destroy(f2fs_cf_name_slab);
316 }
317 #else
318 static int __init f2fs_create_casefold_cache(void) { return 0; }
319 static void f2fs_destroy_casefold_cache(void) { }
320 #endif
321
322 static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
323 {
324 block_t limit = min((sbi->user_block_count >> 3),
325 sbi->user_block_count - sbi->reserved_blocks);
326
327 /* limit is 12.5% */
328 if (test_opt(sbi, RESERVE_ROOT) &&
329 F2FS_OPTION(sbi).root_reserved_blocks > limit) {
330 F2FS_OPTION(sbi).root_reserved_blocks = limit;
331 f2fs_info(sbi, "Reduce reserved blocks for root = %u",
332 F2FS_OPTION(sbi).root_reserved_blocks);
333 }
334 if (!test_opt(sbi, RESERVE_ROOT) &&
335 (!uid_eq(F2FS_OPTION(sbi).s_resuid,
336 make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
337 !gid_eq(F2FS_OPTION(sbi).s_resgid,
338 make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
339 f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
340 from_kuid_munged(&init_user_ns,
341 F2FS_OPTION(sbi).s_resuid),
342 from_kgid_munged(&init_user_ns,
343 F2FS_OPTION(sbi).s_resgid));
344 }
345
346 static inline int adjust_reserved_segment(struct f2fs_sb_info *sbi)
347 {
348 unsigned int sec_blks = sbi->blocks_per_seg * sbi->segs_per_sec;
349 unsigned int avg_vblocks;
350 unsigned int wanted_reserved_segments;
351 block_t avail_user_block_count;
352
353 if (!F2FS_IO_ALIGNED(sbi))
354 return 0;
355
356 /* average valid block count in section in worst case */
357 avg_vblocks = sec_blks / F2FS_IO_SIZE(sbi);
358
359 /*
360 * we need enough free space when migrating one section in worst case
361 */
362 wanted_reserved_segments = (F2FS_IO_SIZE(sbi) / avg_vblocks) *
363 reserved_segments(sbi);
364 wanted_reserved_segments -= reserved_segments(sbi);
365
366 avail_user_block_count = sbi->user_block_count -
367 sbi->current_reserved_blocks -
368 F2FS_OPTION(sbi).root_reserved_blocks;
369
370 if (wanted_reserved_segments * sbi->blocks_per_seg >
371 avail_user_block_count) {
372 f2fs_err(sbi, "IO align feature can't grab additional reserved segment: %u, available segments: %u",
373 wanted_reserved_segments,
374 avail_user_block_count >> sbi->log_blocks_per_seg);
375 return -ENOSPC;
376 }
377
378 SM_I(sbi)->additional_reserved_segments = wanted_reserved_segments;
379
380 f2fs_info(sbi, "IO align feature needs additional reserved segment: %u",
381 wanted_reserved_segments);
382
383 return 0;
384 }
385
386 static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
387 {
388 if (!F2FS_OPTION(sbi).unusable_cap_perc)
389 return;
390
391 if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
392 F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
393 else
394 F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
395 F2FS_OPTION(sbi).unusable_cap_perc;
396
397 f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
398 F2FS_OPTION(sbi).unusable_cap,
399 F2FS_OPTION(sbi).unusable_cap_perc);
400 }
401
402 static void init_once(void *foo)
403 {
404 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
405
406 inode_init_once(&fi->vfs_inode);
407 }
408
409 #ifdef CONFIG_QUOTA
410 static const char * const quotatypes[] = INITQFNAMES;
411 #define QTYPE2NAME(t) (quotatypes[t])
412 static int f2fs_set_qf_name(struct super_block *sb, int qtype,
413 substring_t *args)
414 {
415 struct f2fs_sb_info *sbi = F2FS_SB(sb);
416 char *qname;
417 int ret = -EINVAL;
418
419 if (sb_any_quota_loaded(sb) && !F2FS_OPTION(sbi).s_qf_names[qtype]) {
420 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
421 return -EINVAL;
422 }
423 if (f2fs_sb_has_quota_ino(sbi)) {
424 f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
425 return 0;
426 }
427
428 qname = match_strdup(args);
429 if (!qname) {
430 f2fs_err(sbi, "Not enough memory for storing quotafile name");
431 return -ENOMEM;
432 }
433 if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
434 if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
435 ret = 0;
436 else
437 f2fs_err(sbi, "%s quota file already specified",
438 QTYPE2NAME(qtype));
439 goto errout;
440 }
441 if (strchr(qname, '/')) {
442 f2fs_err(sbi, "quotafile must be on filesystem root");
443 goto errout;
444 }
445 F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
446 set_opt(sbi, QUOTA);
447 return 0;
448 errout:
449 kfree(qname);
450 return ret;
451 }
452
453 static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
454 {
455 struct f2fs_sb_info *sbi = F2FS_SB(sb);
456
457 if (sb_any_quota_loaded(sb) && F2FS_OPTION(sbi).s_qf_names[qtype]) {
458 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
459 return -EINVAL;
460 }
461 kfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
462 F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
463 return 0;
464 }
465
466 static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
467 {
468 /*
469 * We do the test below only for project quotas. 'usrquota' and
470 * 'grpquota' mount options are allowed even without quota feature
471 * to support legacy quotas in quota files.
472 */
473 if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi)) {
474 f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
475 return -1;
476 }
477 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
478 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
479 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) {
480 if (test_opt(sbi, USRQUOTA) &&
481 F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
482 clear_opt(sbi, USRQUOTA);
483
484 if (test_opt(sbi, GRPQUOTA) &&
485 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
486 clear_opt(sbi, GRPQUOTA);
487
488 if (test_opt(sbi, PRJQUOTA) &&
489 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
490 clear_opt(sbi, PRJQUOTA);
491
492 if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
493 test_opt(sbi, PRJQUOTA)) {
494 f2fs_err(sbi, "old and new quota format mixing");
495 return -1;
496 }
497
498 if (!F2FS_OPTION(sbi).s_jquota_fmt) {
499 f2fs_err(sbi, "journaled quota format not specified");
500 return -1;
501 }
502 }
503
504 if (f2fs_sb_has_quota_ino(sbi) && F2FS_OPTION(sbi).s_jquota_fmt) {
505 f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
506 F2FS_OPTION(sbi).s_jquota_fmt = 0;
507 }
508 return 0;
509 }
510 #endif
511
512 static int f2fs_set_test_dummy_encryption(struct super_block *sb,
513 const char *opt,
514 const substring_t *arg,
515 bool is_remount)
516 {
517 struct f2fs_sb_info *sbi = F2FS_SB(sb);
518 struct fs_parameter param = {
519 .type = fs_value_is_string,
520 .string = arg->from ? arg->from : "",
521 };
522 struct fscrypt_dummy_policy *policy =
523 &F2FS_OPTION(sbi).dummy_enc_policy;
524 int err;
525
526 if (!IS_ENABLED(CONFIG_FS_ENCRYPTION)) {
527 f2fs_warn(sbi, "test_dummy_encryption option not supported");
528 return -EINVAL;
529 }
530
531 if (!f2fs_sb_has_encrypt(sbi)) {
532 f2fs_err(sbi, "Encrypt feature is off");
533 return -EINVAL;
534 }
535
536 /*
537 * This mount option is just for testing, and it's not worthwhile to
538 * implement the extra complexity (e.g. RCU protection) that would be
539 * needed to allow it to be set or changed during remount. We do allow
540 * it to be specified during remount, but only if there is no change.
541 */
542 if (is_remount && !fscrypt_is_dummy_policy_set(policy)) {
543 f2fs_warn(sbi, "Can't set test_dummy_encryption on remount");
544 return -EINVAL;
545 }
546
547 err = fscrypt_parse_test_dummy_encryption(&param, policy);
548 if (err) {
549 if (err == -EEXIST)
550 f2fs_warn(sbi,
551 "Can't change test_dummy_encryption on remount");
552 else if (err == -EINVAL)
553 f2fs_warn(sbi, "Value of option \"%s\" is unrecognized",
554 opt);
555 else
556 f2fs_warn(sbi, "Error processing option \"%s\" [%d]",
557 opt, err);
558 return -EINVAL;
559 }
560 f2fs_warn(sbi, "Test dummy encryption mode enabled");
561 return 0;
562 }
563
564 #ifdef CONFIG_F2FS_FS_COMPRESSION
565 static bool is_compress_extension_exist(struct f2fs_sb_info *sbi,
566 const char *new_ext, bool is_ext)
567 {
568 unsigned char (*ext)[F2FS_EXTENSION_LEN];
569 int ext_cnt;
570 int i;
571
572 if (is_ext) {
573 ext = F2FS_OPTION(sbi).extensions;
574 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
575 } else {
576 ext = F2FS_OPTION(sbi).noextensions;
577 ext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
578 }
579
580 for (i = 0; i < ext_cnt; i++) {
581 if (!strcasecmp(new_ext, ext[i]))
582 return true;
583 }
584
585 return false;
586 }
587
588 /*
589 * 1. The same extension name cannot not appear in both compress and non-compress extension
590 * at the same time.
591 * 2. If the compress extension specifies all files, the types specified by the non-compress
592 * extension will be treated as special cases and will not be compressed.
593 * 3. Don't allow the non-compress extension specifies all files.
594 */
595 static int f2fs_test_compress_extension(struct f2fs_sb_info *sbi)
596 {
597 unsigned char (*ext)[F2FS_EXTENSION_LEN];
598 unsigned char (*noext)[F2FS_EXTENSION_LEN];
599 int ext_cnt, noext_cnt, index = 0, no_index = 0;
600
601 ext = F2FS_OPTION(sbi).extensions;
602 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
603 noext = F2FS_OPTION(sbi).noextensions;
604 noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
605
606 if (!noext_cnt)
607 return 0;
608
609 for (no_index = 0; no_index < noext_cnt; no_index++) {
610 if (!strcasecmp("*", noext[no_index])) {
611 f2fs_info(sbi, "Don't allow the nocompress extension specifies all files");
612 return -EINVAL;
613 }
614 for (index = 0; index < ext_cnt; index++) {
615 if (!strcasecmp(ext[index], noext[no_index])) {
616 f2fs_info(sbi, "Don't allow the same extension %s appear in both compress and nocompress extension",
617 ext[index]);
618 return -EINVAL;
619 }
620 }
621 }
622 return 0;
623 }
624
625 #ifdef CONFIG_F2FS_FS_LZ4
626 static int f2fs_set_lz4hc_level(struct f2fs_sb_info *sbi, const char *str)
627 {
628 #ifdef CONFIG_F2FS_FS_LZ4HC
629 unsigned int level;
630
631 if (strlen(str) == 3) {
632 F2FS_OPTION(sbi).compress_level = 0;
633 return 0;
634 }
635
636 str += 3;
637
638 if (str[0] != ':') {
639 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
640 return -EINVAL;
641 }
642 if (kstrtouint(str + 1, 10, &level))
643 return -EINVAL;
644
645 if (!f2fs_is_compress_level_valid(COMPRESS_LZ4, level)) {
646 f2fs_info(sbi, "invalid lz4hc compress level: %d", level);
647 return -EINVAL;
648 }
649
650 F2FS_OPTION(sbi).compress_level = level;
651 return 0;
652 #else
653 if (strlen(str) == 3) {
654 F2FS_OPTION(sbi).compress_level = 0;
655 return 0;
656 }
657 f2fs_info(sbi, "kernel doesn't support lz4hc compression");
658 return -EINVAL;
659 #endif
660 }
661 #endif
662
663 #ifdef CONFIG_F2FS_FS_ZSTD
664 static int f2fs_set_zstd_level(struct f2fs_sb_info *sbi, const char *str)
665 {
666 unsigned int level;
667 int len = 4;
668
669 if (strlen(str) == len) {
670 F2FS_OPTION(sbi).compress_level = F2FS_ZSTD_DEFAULT_CLEVEL;
671 return 0;
672 }
673
674 str += len;
675
676 if (str[0] != ':') {
677 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
678 return -EINVAL;
679 }
680 if (kstrtouint(str + 1, 10, &level))
681 return -EINVAL;
682
683 if (!f2fs_is_compress_level_valid(COMPRESS_ZSTD, level)) {
684 f2fs_info(sbi, "invalid zstd compress level: %d", level);
685 return -EINVAL;
686 }
687
688 F2FS_OPTION(sbi).compress_level = level;
689 return 0;
690 }
691 #endif
692 #endif
693
694 static int parse_options(struct super_block *sb, char *options, bool is_remount)
695 {
696 struct f2fs_sb_info *sbi = F2FS_SB(sb);
697 substring_t args[MAX_OPT_ARGS];
698 #ifdef CONFIG_F2FS_FS_COMPRESSION
699 unsigned char (*ext)[F2FS_EXTENSION_LEN];
700 unsigned char (*noext)[F2FS_EXTENSION_LEN];
701 int ext_cnt, noext_cnt;
702 #endif
703 char *p, *name;
704 int arg = 0;
705 kuid_t uid;
706 kgid_t gid;
707 int ret;
708
709 if (!options)
710 goto default_check;
711
712 while ((p = strsep(&options, ",")) != NULL) {
713 int token;
714
715 if (!*p)
716 continue;
717 /*
718 * Initialize args struct so we know whether arg was
719 * found; some options take optional arguments.
720 */
721 args[0].to = args[0].from = NULL;
722 token = match_token(p, f2fs_tokens, args);
723
724 switch (token) {
725 case Opt_gc_background:
726 name = match_strdup(&args[0]);
727
728 if (!name)
729 return -ENOMEM;
730 if (!strcmp(name, "on")) {
731 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
732 } else if (!strcmp(name, "off")) {
733 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_OFF;
734 } else if (!strcmp(name, "sync")) {
735 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_SYNC;
736 } else {
737 kfree(name);
738 return -EINVAL;
739 }
740 kfree(name);
741 break;
742 case Opt_disable_roll_forward:
743 set_opt(sbi, DISABLE_ROLL_FORWARD);
744 break;
745 case Opt_norecovery:
746 /* this option mounts f2fs with ro */
747 set_opt(sbi, NORECOVERY);
748 if (!f2fs_readonly(sb))
749 return -EINVAL;
750 break;
751 case Opt_discard:
752 if (!f2fs_hw_support_discard(sbi)) {
753 f2fs_warn(sbi, "device does not support discard");
754 break;
755 }
756 set_opt(sbi, DISCARD);
757 break;
758 case Opt_nodiscard:
759 if (f2fs_hw_should_discard(sbi)) {
760 f2fs_warn(sbi, "discard is required for zoned block devices");
761 return -EINVAL;
762 }
763 clear_opt(sbi, DISCARD);
764 break;
765 case Opt_noheap:
766 set_opt(sbi, NOHEAP);
767 break;
768 case Opt_heap:
769 clear_opt(sbi, NOHEAP);
770 break;
771 #ifdef CONFIG_F2FS_FS_XATTR
772 case Opt_user_xattr:
773 set_opt(sbi, XATTR_USER);
774 break;
775 case Opt_nouser_xattr:
776 clear_opt(sbi, XATTR_USER);
777 break;
778 case Opt_inline_xattr:
779 set_opt(sbi, INLINE_XATTR);
780 break;
781 case Opt_noinline_xattr:
782 clear_opt(sbi, INLINE_XATTR);
783 break;
784 case Opt_inline_xattr_size:
785 if (args->from && match_int(args, &arg))
786 return -EINVAL;
787 set_opt(sbi, INLINE_XATTR_SIZE);
788 F2FS_OPTION(sbi).inline_xattr_size = arg;
789 break;
790 #else
791 case Opt_user_xattr:
792 f2fs_info(sbi, "user_xattr options not supported");
793 break;
794 case Opt_nouser_xattr:
795 f2fs_info(sbi, "nouser_xattr options not supported");
796 break;
797 case Opt_inline_xattr:
798 f2fs_info(sbi, "inline_xattr options not supported");
799 break;
800 case Opt_noinline_xattr:
801 f2fs_info(sbi, "noinline_xattr options not supported");
802 break;
803 #endif
804 #ifdef CONFIG_F2FS_FS_POSIX_ACL
805 case Opt_acl:
806 set_opt(sbi, POSIX_ACL);
807 break;
808 case Opt_noacl:
809 clear_opt(sbi, POSIX_ACL);
810 break;
811 #else
812 case Opt_acl:
813 f2fs_info(sbi, "acl options not supported");
814 break;
815 case Opt_noacl:
816 f2fs_info(sbi, "noacl options not supported");
817 break;
818 #endif
819 case Opt_active_logs:
820 if (args->from && match_int(args, &arg))
821 return -EINVAL;
822 if (arg != 2 && arg != 4 &&
823 arg != NR_CURSEG_PERSIST_TYPE)
824 return -EINVAL;
825 F2FS_OPTION(sbi).active_logs = arg;
826 break;
827 case Opt_disable_ext_identify:
828 set_opt(sbi, DISABLE_EXT_IDENTIFY);
829 break;
830 case Opt_inline_data:
831 set_opt(sbi, INLINE_DATA);
832 break;
833 case Opt_inline_dentry:
834 set_opt(sbi, INLINE_DENTRY);
835 break;
836 case Opt_noinline_dentry:
837 clear_opt(sbi, INLINE_DENTRY);
838 break;
839 case Opt_flush_merge:
840 set_opt(sbi, FLUSH_MERGE);
841 break;
842 case Opt_noflush_merge:
843 clear_opt(sbi, FLUSH_MERGE);
844 break;
845 case Opt_nobarrier:
846 set_opt(sbi, NOBARRIER);
847 break;
848 case Opt_barrier:
849 clear_opt(sbi, NOBARRIER);
850 break;
851 case Opt_fastboot:
852 set_opt(sbi, FASTBOOT);
853 break;
854 case Opt_extent_cache:
855 set_opt(sbi, READ_EXTENT_CACHE);
856 break;
857 case Opt_noextent_cache:
858 clear_opt(sbi, READ_EXTENT_CACHE);
859 break;
860 case Opt_noinline_data:
861 clear_opt(sbi, INLINE_DATA);
862 break;
863 case Opt_data_flush:
864 set_opt(sbi, DATA_FLUSH);
865 break;
866 case Opt_reserve_root:
867 if (args->from && match_int(args, &arg))
868 return -EINVAL;
869 if (test_opt(sbi, RESERVE_ROOT)) {
870 f2fs_info(sbi, "Preserve previous reserve_root=%u",
871 F2FS_OPTION(sbi).root_reserved_blocks);
872 } else {
873 F2FS_OPTION(sbi).root_reserved_blocks = arg;
874 set_opt(sbi, RESERVE_ROOT);
875 }
876 break;
877 case Opt_resuid:
878 if (args->from && match_int(args, &arg))
879 return -EINVAL;
880 uid = make_kuid(current_user_ns(), arg);
881 if (!uid_valid(uid)) {
882 f2fs_err(sbi, "Invalid uid value %d", arg);
883 return -EINVAL;
884 }
885 F2FS_OPTION(sbi).s_resuid = uid;
886 break;
887 case Opt_resgid:
888 if (args->from && match_int(args, &arg))
889 return -EINVAL;
890 gid = make_kgid(current_user_ns(), arg);
891 if (!gid_valid(gid)) {
892 f2fs_err(sbi, "Invalid gid value %d", arg);
893 return -EINVAL;
894 }
895 F2FS_OPTION(sbi).s_resgid = gid;
896 break;
897 case Opt_mode:
898 name = match_strdup(&args[0]);
899
900 if (!name)
901 return -ENOMEM;
902 if (!strcmp(name, "adaptive")) {
903 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
904 } else if (!strcmp(name, "lfs")) {
905 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
906 } else if (!strcmp(name, "fragment:segment")) {
907 F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_SEG;
908 } else if (!strcmp(name, "fragment:block")) {
909 F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_BLK;
910 } else {
911 kfree(name);
912 return -EINVAL;
913 }
914 kfree(name);
915 break;
916 case Opt_io_size_bits:
917 if (args->from && match_int(args, &arg))
918 return -EINVAL;
919 if (arg <= 0 || arg > __ilog2_u32(BIO_MAX_VECS)) {
920 f2fs_warn(sbi, "Not support %ld, larger than %d",
921 BIT(arg), BIO_MAX_VECS);
922 return -EINVAL;
923 }
924 F2FS_OPTION(sbi).write_io_size_bits = arg;
925 break;
926 #ifdef CONFIG_F2FS_FAULT_INJECTION
927 case Opt_fault_injection:
928 if (args->from && match_int(args, &arg))
929 return -EINVAL;
930 f2fs_build_fault_attr(sbi, arg, F2FS_ALL_FAULT_TYPE);
931 set_opt(sbi, FAULT_INJECTION);
932 break;
933
934 case Opt_fault_type:
935 if (args->from && match_int(args, &arg))
936 return -EINVAL;
937 f2fs_build_fault_attr(sbi, 0, arg);
938 set_opt(sbi, FAULT_INJECTION);
939 break;
940 #else
941 case Opt_fault_injection:
942 f2fs_info(sbi, "fault_injection options not supported");
943 break;
944
945 case Opt_fault_type:
946 f2fs_info(sbi, "fault_type options not supported");
947 break;
948 #endif
949 case Opt_lazytime:
950 sb->s_flags |= SB_LAZYTIME;
951 break;
952 case Opt_nolazytime:
953 sb->s_flags &= ~SB_LAZYTIME;
954 break;
955 #ifdef CONFIG_QUOTA
956 case Opt_quota:
957 case Opt_usrquota:
958 set_opt(sbi, USRQUOTA);
959 break;
960 case Opt_grpquota:
961 set_opt(sbi, GRPQUOTA);
962 break;
963 case Opt_prjquota:
964 set_opt(sbi, PRJQUOTA);
965 break;
966 case Opt_usrjquota:
967 ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
968 if (ret)
969 return ret;
970 break;
971 case Opt_grpjquota:
972 ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
973 if (ret)
974 return ret;
975 break;
976 case Opt_prjjquota:
977 ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
978 if (ret)
979 return ret;
980 break;
981 case Opt_offusrjquota:
982 ret = f2fs_clear_qf_name(sb, USRQUOTA);
983 if (ret)
984 return ret;
985 break;
986 case Opt_offgrpjquota:
987 ret = f2fs_clear_qf_name(sb, GRPQUOTA);
988 if (ret)
989 return ret;
990 break;
991 case Opt_offprjjquota:
992 ret = f2fs_clear_qf_name(sb, PRJQUOTA);
993 if (ret)
994 return ret;
995 break;
996 case Opt_jqfmt_vfsold:
997 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
998 break;
999 case Opt_jqfmt_vfsv0:
1000 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
1001 break;
1002 case Opt_jqfmt_vfsv1:
1003 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
1004 break;
1005 case Opt_noquota:
1006 clear_opt(sbi, QUOTA);
1007 clear_opt(sbi, USRQUOTA);
1008 clear_opt(sbi, GRPQUOTA);
1009 clear_opt(sbi, PRJQUOTA);
1010 break;
1011 #else
1012 case Opt_quota:
1013 case Opt_usrquota:
1014 case Opt_grpquota:
1015 case Opt_prjquota:
1016 case Opt_usrjquota:
1017 case Opt_grpjquota:
1018 case Opt_prjjquota:
1019 case Opt_offusrjquota:
1020 case Opt_offgrpjquota:
1021 case Opt_offprjjquota:
1022 case Opt_jqfmt_vfsold:
1023 case Opt_jqfmt_vfsv0:
1024 case Opt_jqfmt_vfsv1:
1025 case Opt_noquota:
1026 f2fs_info(sbi, "quota operations not supported");
1027 break;
1028 #endif
1029 case Opt_alloc:
1030 name = match_strdup(&args[0]);
1031 if (!name)
1032 return -ENOMEM;
1033
1034 if (!strcmp(name, "default")) {
1035 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
1036 } else if (!strcmp(name, "reuse")) {
1037 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
1038 } else {
1039 kfree(name);
1040 return -EINVAL;
1041 }
1042 kfree(name);
1043 break;
1044 case Opt_fsync:
1045 name = match_strdup(&args[0]);
1046 if (!name)
1047 return -ENOMEM;
1048 if (!strcmp(name, "posix")) {
1049 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
1050 } else if (!strcmp(name, "strict")) {
1051 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_STRICT;
1052 } else if (!strcmp(name, "nobarrier")) {
1053 F2FS_OPTION(sbi).fsync_mode =
1054 FSYNC_MODE_NOBARRIER;
1055 } else {
1056 kfree(name);
1057 return -EINVAL;
1058 }
1059 kfree(name);
1060 break;
1061 case Opt_test_dummy_encryption:
1062 ret = f2fs_set_test_dummy_encryption(sb, p, &args[0],
1063 is_remount);
1064 if (ret)
1065 return ret;
1066 break;
1067 case Opt_inlinecrypt:
1068 #ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
1069 sb->s_flags |= SB_INLINECRYPT;
1070 #else
1071 f2fs_info(sbi, "inline encryption not supported");
1072 #endif
1073 break;
1074 case Opt_checkpoint_disable_cap_perc:
1075 if (args->from && match_int(args, &arg))
1076 return -EINVAL;
1077 if (arg < 0 || arg > 100)
1078 return -EINVAL;
1079 F2FS_OPTION(sbi).unusable_cap_perc = arg;
1080 set_opt(sbi, DISABLE_CHECKPOINT);
1081 break;
1082 case Opt_checkpoint_disable_cap:
1083 if (args->from && match_int(args, &arg))
1084 return -EINVAL;
1085 F2FS_OPTION(sbi).unusable_cap = arg;
1086 set_opt(sbi, DISABLE_CHECKPOINT);
1087 break;
1088 case Opt_checkpoint_disable:
1089 set_opt(sbi, DISABLE_CHECKPOINT);
1090 break;
1091 case Opt_checkpoint_enable:
1092 clear_opt(sbi, DISABLE_CHECKPOINT);
1093 break;
1094 case Opt_checkpoint_merge:
1095 set_opt(sbi, MERGE_CHECKPOINT);
1096 break;
1097 case Opt_nocheckpoint_merge:
1098 clear_opt(sbi, MERGE_CHECKPOINT);
1099 break;
1100 #ifdef CONFIG_F2FS_FS_COMPRESSION
1101 case Opt_compress_algorithm:
1102 if (!f2fs_sb_has_compression(sbi)) {
1103 f2fs_info(sbi, "Image doesn't support compression");
1104 break;
1105 }
1106 name = match_strdup(&args[0]);
1107 if (!name)
1108 return -ENOMEM;
1109 if (!strcmp(name, "lzo")) {
1110 #ifdef CONFIG_F2FS_FS_LZO
1111 F2FS_OPTION(sbi).compress_level = 0;
1112 F2FS_OPTION(sbi).compress_algorithm =
1113 COMPRESS_LZO;
1114 #else
1115 f2fs_info(sbi, "kernel doesn't support lzo compression");
1116 #endif
1117 } else if (!strncmp(name, "lz4", 3)) {
1118 #ifdef CONFIG_F2FS_FS_LZ4
1119 ret = f2fs_set_lz4hc_level(sbi, name);
1120 if (ret) {
1121 kfree(name);
1122 return -EINVAL;
1123 }
1124 F2FS_OPTION(sbi).compress_algorithm =
1125 COMPRESS_LZ4;
1126 #else
1127 f2fs_info(sbi, "kernel doesn't support lz4 compression");
1128 #endif
1129 } else if (!strncmp(name, "zstd", 4)) {
1130 #ifdef CONFIG_F2FS_FS_ZSTD
1131 ret = f2fs_set_zstd_level(sbi, name);
1132 if (ret) {
1133 kfree(name);
1134 return -EINVAL;
1135 }
1136 F2FS_OPTION(sbi).compress_algorithm =
1137 COMPRESS_ZSTD;
1138 #else
1139 f2fs_info(sbi, "kernel doesn't support zstd compression");
1140 #endif
1141 } else if (!strcmp(name, "lzo-rle")) {
1142 #ifdef CONFIG_F2FS_FS_LZORLE
1143 F2FS_OPTION(sbi).compress_level = 0;
1144 F2FS_OPTION(sbi).compress_algorithm =
1145 COMPRESS_LZORLE;
1146 #else
1147 f2fs_info(sbi, "kernel doesn't support lzorle compression");
1148 #endif
1149 } else {
1150 kfree(name);
1151 return -EINVAL;
1152 }
1153 kfree(name);
1154 break;
1155 case Opt_compress_log_size:
1156 if (!f2fs_sb_has_compression(sbi)) {
1157 f2fs_info(sbi, "Image doesn't support compression");
1158 break;
1159 }
1160 if (args->from && match_int(args, &arg))
1161 return -EINVAL;
1162 if (arg < MIN_COMPRESS_LOG_SIZE ||
1163 arg > MAX_COMPRESS_LOG_SIZE) {
1164 f2fs_err(sbi,
1165 "Compress cluster log size is out of range");
1166 return -EINVAL;
1167 }
1168 F2FS_OPTION(sbi).compress_log_size = arg;
1169 break;
1170 case Opt_compress_extension:
1171 if (!f2fs_sb_has_compression(sbi)) {
1172 f2fs_info(sbi, "Image doesn't support compression");
1173 break;
1174 }
1175 name = match_strdup(&args[0]);
1176 if (!name)
1177 return -ENOMEM;
1178
1179 ext = F2FS_OPTION(sbi).extensions;
1180 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
1181
1182 if (strlen(name) >= F2FS_EXTENSION_LEN ||
1183 ext_cnt >= COMPRESS_EXT_NUM) {
1184 f2fs_err(sbi,
1185 "invalid extension length/number");
1186 kfree(name);
1187 return -EINVAL;
1188 }
1189
1190 if (is_compress_extension_exist(sbi, name, true)) {
1191 kfree(name);
1192 break;
1193 }
1194
1195 strcpy(ext[ext_cnt], name);
1196 F2FS_OPTION(sbi).compress_ext_cnt++;
1197 kfree(name);
1198 break;
1199 case Opt_nocompress_extension:
1200 if (!f2fs_sb_has_compression(sbi)) {
1201 f2fs_info(sbi, "Image doesn't support compression");
1202 break;
1203 }
1204 name = match_strdup(&args[0]);
1205 if (!name)
1206 return -ENOMEM;
1207
1208 noext = F2FS_OPTION(sbi).noextensions;
1209 noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
1210
1211 if (strlen(name) >= F2FS_EXTENSION_LEN ||
1212 noext_cnt >= COMPRESS_EXT_NUM) {
1213 f2fs_err(sbi,
1214 "invalid extension length/number");
1215 kfree(name);
1216 return -EINVAL;
1217 }
1218
1219 if (is_compress_extension_exist(sbi, name, false)) {
1220 kfree(name);
1221 break;
1222 }
1223
1224 strcpy(noext[noext_cnt], name);
1225 F2FS_OPTION(sbi).nocompress_ext_cnt++;
1226 kfree(name);
1227 break;
1228 case Opt_compress_chksum:
1229 if (!f2fs_sb_has_compression(sbi)) {
1230 f2fs_info(sbi, "Image doesn't support compression");
1231 break;
1232 }
1233 F2FS_OPTION(sbi).compress_chksum = true;
1234 break;
1235 case Opt_compress_mode:
1236 if (!f2fs_sb_has_compression(sbi)) {
1237 f2fs_info(sbi, "Image doesn't support compression");
1238 break;
1239 }
1240 name = match_strdup(&args[0]);
1241 if (!name)
1242 return -ENOMEM;
1243 if (!strcmp(name, "fs")) {
1244 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
1245 } else if (!strcmp(name, "user")) {
1246 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_USER;
1247 } else {
1248 kfree(name);
1249 return -EINVAL;
1250 }
1251 kfree(name);
1252 break;
1253 case Opt_compress_cache:
1254 if (!f2fs_sb_has_compression(sbi)) {
1255 f2fs_info(sbi, "Image doesn't support compression");
1256 break;
1257 }
1258 set_opt(sbi, COMPRESS_CACHE);
1259 break;
1260 #else
1261 case Opt_compress_algorithm:
1262 case Opt_compress_log_size:
1263 case Opt_compress_extension:
1264 case Opt_nocompress_extension:
1265 case Opt_compress_chksum:
1266 case Opt_compress_mode:
1267 case Opt_compress_cache:
1268 f2fs_info(sbi, "compression options not supported");
1269 break;
1270 #endif
1271 case Opt_atgc:
1272 set_opt(sbi, ATGC);
1273 break;
1274 case Opt_gc_merge:
1275 set_opt(sbi, GC_MERGE);
1276 break;
1277 case Opt_nogc_merge:
1278 clear_opt(sbi, GC_MERGE);
1279 break;
1280 case Opt_discard_unit:
1281 name = match_strdup(&args[0]);
1282 if (!name)
1283 return -ENOMEM;
1284 if (!strcmp(name, "block")) {
1285 F2FS_OPTION(sbi).discard_unit =
1286 DISCARD_UNIT_BLOCK;
1287 } else if (!strcmp(name, "segment")) {
1288 F2FS_OPTION(sbi).discard_unit =
1289 DISCARD_UNIT_SEGMENT;
1290 } else if (!strcmp(name, "section")) {
1291 F2FS_OPTION(sbi).discard_unit =
1292 DISCARD_UNIT_SECTION;
1293 } else {
1294 kfree(name);
1295 return -EINVAL;
1296 }
1297 kfree(name);
1298 break;
1299 case Opt_memory_mode:
1300 name = match_strdup(&args[0]);
1301 if (!name)
1302 return -ENOMEM;
1303 if (!strcmp(name, "normal")) {
1304 F2FS_OPTION(sbi).memory_mode =
1305 MEMORY_MODE_NORMAL;
1306 } else if (!strcmp(name, "low")) {
1307 F2FS_OPTION(sbi).memory_mode =
1308 MEMORY_MODE_LOW;
1309 } else {
1310 kfree(name);
1311 return -EINVAL;
1312 }
1313 kfree(name);
1314 break;
1315 case Opt_age_extent_cache:
1316 set_opt(sbi, AGE_EXTENT_CACHE);
1317 break;
1318 case Opt_errors:
1319 name = match_strdup(&args[0]);
1320 if (!name)
1321 return -ENOMEM;
1322 if (!strcmp(name, "remount-ro")) {
1323 F2FS_OPTION(sbi).errors =
1324 MOUNT_ERRORS_READONLY;
1325 } else if (!strcmp(name, "continue")) {
1326 F2FS_OPTION(sbi).errors =
1327 MOUNT_ERRORS_CONTINUE;
1328 } else if (!strcmp(name, "panic")) {
1329 F2FS_OPTION(sbi).errors =
1330 MOUNT_ERRORS_PANIC;
1331 } else {
1332 kfree(name);
1333 return -EINVAL;
1334 }
1335 kfree(name);
1336 break;
1337 default:
1338 f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
1339 p);
1340 return -EINVAL;
1341 }
1342 }
1343 default_check:
1344 #ifdef CONFIG_QUOTA
1345 if (f2fs_check_quota_options(sbi))
1346 return -EINVAL;
1347 #else
1348 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sbi->sb)) {
1349 f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1350 return -EINVAL;
1351 }
1352 if (f2fs_sb_has_project_quota(sbi) && !f2fs_readonly(sbi->sb)) {
1353 f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1354 return -EINVAL;
1355 }
1356 #endif
1357 #if !IS_ENABLED(CONFIG_UNICODE)
1358 if (f2fs_sb_has_casefold(sbi)) {
1359 f2fs_err(sbi,
1360 "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
1361 return -EINVAL;
1362 }
1363 #endif
1364 /*
1365 * The BLKZONED feature indicates that the drive was formatted with
1366 * zone alignment optimization. This is optional for host-aware
1367 * devices, but mandatory for host-managed zoned block devices.
1368 */
1369 if (f2fs_sb_has_blkzoned(sbi)) {
1370 #ifdef CONFIG_BLK_DEV_ZONED
1371 if (F2FS_OPTION(sbi).discard_unit !=
1372 DISCARD_UNIT_SECTION) {
1373 f2fs_info(sbi, "Zoned block device doesn't need small discard, set discard_unit=section by default");
1374 F2FS_OPTION(sbi).discard_unit =
1375 DISCARD_UNIT_SECTION;
1376 }
1377
1378 if (F2FS_OPTION(sbi).fs_mode != FS_MODE_LFS) {
1379 f2fs_info(sbi, "Only lfs mode is allowed with zoned block device feature");
1380 return -EINVAL;
1381 }
1382 #else
1383 f2fs_err(sbi, "Zoned block device support is not enabled");
1384 return -EINVAL;
1385 #endif
1386 }
1387
1388 #ifdef CONFIG_F2FS_FS_COMPRESSION
1389 if (f2fs_test_compress_extension(sbi)) {
1390 f2fs_err(sbi, "invalid compress or nocompress extension");
1391 return -EINVAL;
1392 }
1393 #endif
1394
1395 if (F2FS_IO_SIZE_BITS(sbi) && !f2fs_lfs_mode(sbi)) {
1396 f2fs_err(sbi, "Should set mode=lfs with %luKB-sized IO",
1397 F2FS_IO_SIZE_KB(sbi));
1398 return -EINVAL;
1399 }
1400
1401 if (test_opt(sbi, INLINE_XATTR_SIZE)) {
1402 int min_size, max_size;
1403
1404 if (!f2fs_sb_has_extra_attr(sbi) ||
1405 !f2fs_sb_has_flexible_inline_xattr(sbi)) {
1406 f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
1407 return -EINVAL;
1408 }
1409 if (!test_opt(sbi, INLINE_XATTR)) {
1410 f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
1411 return -EINVAL;
1412 }
1413
1414 min_size = MIN_INLINE_XATTR_SIZE;
1415 max_size = MAX_INLINE_XATTR_SIZE;
1416
1417 if (F2FS_OPTION(sbi).inline_xattr_size < min_size ||
1418 F2FS_OPTION(sbi).inline_xattr_size > max_size) {
1419 f2fs_err(sbi, "inline xattr size is out of range: %d ~ %d",
1420 min_size, max_size);
1421 return -EINVAL;
1422 }
1423 }
1424
1425 if (test_opt(sbi, DISABLE_CHECKPOINT) && f2fs_lfs_mode(sbi)) {
1426 f2fs_err(sbi, "LFS is not compatible with checkpoint=disable");
1427 return -EINVAL;
1428 }
1429
1430 if (test_opt(sbi, ATGC) && f2fs_lfs_mode(sbi)) {
1431 f2fs_err(sbi, "LFS is not compatible with ATGC");
1432 return -EINVAL;
1433 }
1434
1435 if (f2fs_is_readonly(sbi) && test_opt(sbi, FLUSH_MERGE)) {
1436 f2fs_err(sbi, "FLUSH_MERGE not compatible with readonly mode");
1437 return -EINVAL;
1438 }
1439
1440 if (f2fs_sb_has_readonly(sbi) && !f2fs_readonly(sbi->sb)) {
1441 f2fs_err(sbi, "Allow to mount readonly mode only");
1442 return -EROFS;
1443 }
1444 return 0;
1445 }
1446
1447 static struct inode *f2fs_alloc_inode(struct super_block *sb)
1448 {
1449 struct f2fs_inode_info *fi;
1450
1451 if (time_to_inject(F2FS_SB(sb), FAULT_SLAB_ALLOC))
1452 return NULL;
1453
1454 fi = alloc_inode_sb(sb, f2fs_inode_cachep, GFP_F2FS_ZERO);
1455 if (!fi)
1456 return NULL;
1457
1458 init_once((void *) fi);
1459
1460 /* Initialize f2fs-specific inode info */
1461 atomic_set(&fi->dirty_pages, 0);
1462 atomic_set(&fi->i_compr_blocks, 0);
1463 init_f2fs_rwsem(&fi->i_sem);
1464 spin_lock_init(&fi->i_size_lock);
1465 INIT_LIST_HEAD(&fi->dirty_list);
1466 INIT_LIST_HEAD(&fi->gdirty_list);
1467 init_f2fs_rwsem(&fi->i_gc_rwsem[READ]);
1468 init_f2fs_rwsem(&fi->i_gc_rwsem[WRITE]);
1469 init_f2fs_rwsem(&fi->i_xattr_sem);
1470
1471 /* Will be used by directory only */
1472 fi->i_dir_level = F2FS_SB(sb)->dir_level;
1473
1474 return &fi->vfs_inode;
1475 }
1476
1477 static int f2fs_drop_inode(struct inode *inode)
1478 {
1479 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1480 int ret;
1481
1482 /*
1483 * during filesystem shutdown, if checkpoint is disabled,
1484 * drop useless meta/node dirty pages.
1485 */
1486 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1487 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1488 inode->i_ino == F2FS_META_INO(sbi)) {
1489 trace_f2fs_drop_inode(inode, 1);
1490 return 1;
1491 }
1492 }
1493
1494 /*
1495 * This is to avoid a deadlock condition like below.
1496 * writeback_single_inode(inode)
1497 * - f2fs_write_data_page
1498 * - f2fs_gc -> iput -> evict
1499 * - inode_wait_for_writeback(inode)
1500 */
1501 if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
1502 if (!inode->i_nlink && !is_bad_inode(inode)) {
1503 /* to avoid evict_inode call simultaneously */
1504 atomic_inc(&inode->i_count);
1505 spin_unlock(&inode->i_lock);
1506
1507 /* should remain fi->extent_tree for writepage */
1508 f2fs_destroy_extent_node(inode);
1509
1510 sb_start_intwrite(inode->i_sb);
1511 f2fs_i_size_write(inode, 0);
1512
1513 f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1514 inode, NULL, 0, DATA);
1515 truncate_inode_pages_final(inode->i_mapping);
1516
1517 if (F2FS_HAS_BLOCKS(inode))
1518 f2fs_truncate(inode);
1519
1520 sb_end_intwrite(inode->i_sb);
1521
1522 spin_lock(&inode->i_lock);
1523 atomic_dec(&inode->i_count);
1524 }
1525 trace_f2fs_drop_inode(inode, 0);
1526 return 0;
1527 }
1528 ret = generic_drop_inode(inode);
1529 if (!ret)
1530 ret = fscrypt_drop_inode(inode);
1531 trace_f2fs_drop_inode(inode, ret);
1532 return ret;
1533 }
1534
1535 int f2fs_inode_dirtied(struct inode *inode, bool sync)
1536 {
1537 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1538 int ret = 0;
1539
1540 spin_lock(&sbi->inode_lock[DIRTY_META]);
1541 if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1542 ret = 1;
1543 } else {
1544 set_inode_flag(inode, FI_DIRTY_INODE);
1545 stat_inc_dirty_inode(sbi, DIRTY_META);
1546 }
1547 if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1548 list_add_tail(&F2FS_I(inode)->gdirty_list,
1549 &sbi->inode_list[DIRTY_META]);
1550 inc_page_count(sbi, F2FS_DIRTY_IMETA);
1551 }
1552 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1553 return ret;
1554 }
1555
1556 void f2fs_inode_synced(struct inode *inode)
1557 {
1558 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1559
1560 spin_lock(&sbi->inode_lock[DIRTY_META]);
1561 if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1562 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1563 return;
1564 }
1565 if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1566 list_del_init(&F2FS_I(inode)->gdirty_list);
1567 dec_page_count(sbi, F2FS_DIRTY_IMETA);
1568 }
1569 clear_inode_flag(inode, FI_DIRTY_INODE);
1570 clear_inode_flag(inode, FI_AUTO_RECOVER);
1571 stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1572 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1573 }
1574
1575 /*
1576 * f2fs_dirty_inode() is called from __mark_inode_dirty()
1577 *
1578 * We should call set_dirty_inode to write the dirty inode through write_inode.
1579 */
1580 static void f2fs_dirty_inode(struct inode *inode, int flags)
1581 {
1582 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1583
1584 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1585 inode->i_ino == F2FS_META_INO(sbi))
1586 return;
1587
1588 if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1589 clear_inode_flag(inode, FI_AUTO_RECOVER);
1590
1591 f2fs_inode_dirtied(inode, false);
1592 }
1593
1594 static void f2fs_free_inode(struct inode *inode)
1595 {
1596 fscrypt_free_inode(inode);
1597 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1598 }
1599
1600 static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1601 {
1602 percpu_counter_destroy(&sbi->total_valid_inode_count);
1603 percpu_counter_destroy(&sbi->rf_node_block_count);
1604 percpu_counter_destroy(&sbi->alloc_valid_block_count);
1605 }
1606
1607 static void destroy_device_list(struct f2fs_sb_info *sbi)
1608 {
1609 int i;
1610
1611 for (i = 0; i < sbi->s_ndevs; i++) {
1612 if (i > 0)
1613 bdev_release(FDEV(i).bdev_handle);
1614 #ifdef CONFIG_BLK_DEV_ZONED
1615 kvfree(FDEV(i).blkz_seq);
1616 #endif
1617 }
1618 kvfree(sbi->devs);
1619 }
1620
1621 static void f2fs_put_super(struct super_block *sb)
1622 {
1623 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1624 int i;
1625 int err = 0;
1626 bool done;
1627
1628 /* unregister procfs/sysfs entries in advance to avoid race case */
1629 f2fs_unregister_sysfs(sbi);
1630
1631 f2fs_quota_off_umount(sb);
1632
1633 /* prevent remaining shrinker jobs */
1634 mutex_lock(&sbi->umount_mutex);
1635
1636 /*
1637 * flush all issued checkpoints and stop checkpoint issue thread.
1638 * after then, all checkpoints should be done by each process context.
1639 */
1640 f2fs_stop_ckpt_thread(sbi);
1641
1642 /*
1643 * We don't need to do checkpoint when superblock is clean.
1644 * But, the previous checkpoint was not done by umount, it needs to do
1645 * clean checkpoint again.
1646 */
1647 if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1648 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1649 struct cp_control cpc = {
1650 .reason = CP_UMOUNT,
1651 };
1652 stat_inc_cp_call_count(sbi, TOTAL_CALL);
1653 err = f2fs_write_checkpoint(sbi, &cpc);
1654 }
1655
1656 /* be sure to wait for any on-going discard commands */
1657 done = f2fs_issue_discard_timeout(sbi);
1658 if (f2fs_realtime_discard_enable(sbi) && !sbi->discard_blks && done) {
1659 struct cp_control cpc = {
1660 .reason = CP_UMOUNT | CP_TRIMMED,
1661 };
1662 stat_inc_cp_call_count(sbi, TOTAL_CALL);
1663 err = f2fs_write_checkpoint(sbi, &cpc);
1664 }
1665
1666 /*
1667 * normally superblock is clean, so we need to release this.
1668 * In addition, EIO will skip do checkpoint, we need this as well.
1669 */
1670 f2fs_release_ino_entry(sbi, true);
1671
1672 f2fs_leave_shrinker(sbi);
1673 mutex_unlock(&sbi->umount_mutex);
1674
1675 /* our cp_error case, we can wait for any writeback page */
1676 f2fs_flush_merged_writes(sbi);
1677
1678 f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1679
1680 if (err || f2fs_cp_error(sbi)) {
1681 truncate_inode_pages_final(NODE_MAPPING(sbi));
1682 truncate_inode_pages_final(META_MAPPING(sbi));
1683 }
1684
1685 for (i = 0; i < NR_COUNT_TYPE; i++) {
1686 if (!get_pages(sbi, i))
1687 continue;
1688 f2fs_err(sbi, "detect filesystem reference count leak during "
1689 "umount, type: %d, count: %lld", i, get_pages(sbi, i));
1690 f2fs_bug_on(sbi, 1);
1691 }
1692
1693 f2fs_bug_on(sbi, sbi->fsync_node_num);
1694
1695 f2fs_destroy_compress_inode(sbi);
1696
1697 iput(sbi->node_inode);
1698 sbi->node_inode = NULL;
1699
1700 iput(sbi->meta_inode);
1701 sbi->meta_inode = NULL;
1702
1703 /*
1704 * iput() can update stat information, if f2fs_write_checkpoint()
1705 * above failed with error.
1706 */
1707 f2fs_destroy_stats(sbi);
1708
1709 /* destroy f2fs internal modules */
1710 f2fs_destroy_node_manager(sbi);
1711 f2fs_destroy_segment_manager(sbi);
1712
1713 /* flush s_error_work before sbi destroy */
1714 flush_work(&sbi->s_error_work);
1715
1716 f2fs_destroy_post_read_wq(sbi);
1717
1718 kvfree(sbi->ckpt);
1719
1720 if (sbi->s_chksum_driver)
1721 crypto_free_shash(sbi->s_chksum_driver);
1722 kfree(sbi->raw_super);
1723
1724 f2fs_destroy_page_array_cache(sbi);
1725 f2fs_destroy_xattr_caches(sbi);
1726 mempool_destroy(sbi->write_io_dummy);
1727 #ifdef CONFIG_QUOTA
1728 for (i = 0; i < MAXQUOTAS; i++)
1729 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1730 #endif
1731 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
1732 destroy_percpu_info(sbi);
1733 f2fs_destroy_iostat(sbi);
1734 for (i = 0; i < NR_PAGE_TYPE; i++)
1735 kvfree(sbi->write_io[i]);
1736 #if IS_ENABLED(CONFIG_UNICODE)
1737 utf8_unload(sb->s_encoding);
1738 #endif
1739 }
1740
1741 int f2fs_sync_fs(struct super_block *sb, int sync)
1742 {
1743 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1744 int err = 0;
1745
1746 if (unlikely(f2fs_cp_error(sbi)))
1747 return 0;
1748 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1749 return 0;
1750
1751 trace_f2fs_sync_fs(sb, sync);
1752
1753 if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1754 return -EAGAIN;
1755
1756 if (sync) {
1757 stat_inc_cp_call_count(sbi, TOTAL_CALL);
1758 err = f2fs_issue_checkpoint(sbi);
1759 }
1760
1761 return err;
1762 }
1763
1764 static int f2fs_freeze(struct super_block *sb)
1765 {
1766 if (f2fs_readonly(sb))
1767 return 0;
1768
1769 /* IO error happened before */
1770 if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1771 return -EIO;
1772
1773 /* must be clean, since sync_filesystem() was already called */
1774 if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1775 return -EINVAL;
1776
1777 /* Let's flush checkpoints and stop the thread. */
1778 f2fs_flush_ckpt_thread(F2FS_SB(sb));
1779
1780 /* to avoid deadlock on f2fs_evict_inode->SB_FREEZE_FS */
1781 set_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1782 return 0;
1783 }
1784
1785 static int f2fs_unfreeze(struct super_block *sb)
1786 {
1787 clear_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1788 return 0;
1789 }
1790
1791 #ifdef CONFIG_QUOTA
1792 static int f2fs_statfs_project(struct super_block *sb,
1793 kprojid_t projid, struct kstatfs *buf)
1794 {
1795 struct kqid qid;
1796 struct dquot *dquot;
1797 u64 limit;
1798 u64 curblock;
1799
1800 qid = make_kqid_projid(projid);
1801 dquot = dqget(sb, qid);
1802 if (IS_ERR(dquot))
1803 return PTR_ERR(dquot);
1804 spin_lock(&dquot->dq_dqb_lock);
1805
1806 limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1807 dquot->dq_dqb.dqb_bhardlimit);
1808 if (limit)
1809 limit >>= sb->s_blocksize_bits;
1810
1811 if (limit && buf->f_blocks > limit) {
1812 curblock = (dquot->dq_dqb.dqb_curspace +
1813 dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
1814 buf->f_blocks = limit;
1815 buf->f_bfree = buf->f_bavail =
1816 (buf->f_blocks > curblock) ?
1817 (buf->f_blocks - curblock) : 0;
1818 }
1819
1820 limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1821 dquot->dq_dqb.dqb_ihardlimit);
1822
1823 if (limit && buf->f_files > limit) {
1824 buf->f_files = limit;
1825 buf->f_ffree =
1826 (buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1827 (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1828 }
1829
1830 spin_unlock(&dquot->dq_dqb_lock);
1831 dqput(dquot);
1832 return 0;
1833 }
1834 #endif
1835
1836 static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1837 {
1838 struct super_block *sb = dentry->d_sb;
1839 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1840 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
1841 block_t total_count, user_block_count, start_count;
1842 u64 avail_node_count;
1843 unsigned int total_valid_node_count;
1844
1845 total_count = le64_to_cpu(sbi->raw_super->block_count);
1846 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
1847 buf->f_type = F2FS_SUPER_MAGIC;
1848 buf->f_bsize = sbi->blocksize;
1849
1850 buf->f_blocks = total_count - start_count;
1851
1852 spin_lock(&sbi->stat_lock);
1853
1854 user_block_count = sbi->user_block_count;
1855 total_valid_node_count = valid_node_count(sbi);
1856 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
1857 buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1858 sbi->current_reserved_blocks;
1859
1860 if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1861 buf->f_bfree = 0;
1862 else
1863 buf->f_bfree -= sbi->unusable_block_count;
1864 spin_unlock(&sbi->stat_lock);
1865
1866 if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1867 buf->f_bavail = buf->f_bfree -
1868 F2FS_OPTION(sbi).root_reserved_blocks;
1869 else
1870 buf->f_bavail = 0;
1871
1872 if (avail_node_count > user_block_count) {
1873 buf->f_files = user_block_count;
1874 buf->f_ffree = buf->f_bavail;
1875 } else {
1876 buf->f_files = avail_node_count;
1877 buf->f_ffree = min(avail_node_count - total_valid_node_count,
1878 buf->f_bavail);
1879 }
1880
1881 buf->f_namelen = F2FS_NAME_LEN;
1882 buf->f_fsid = u64_to_fsid(id);
1883
1884 #ifdef CONFIG_QUOTA
1885 if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1886 sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1887 f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1888 }
1889 #endif
1890 return 0;
1891 }
1892
1893 static inline void f2fs_show_quota_options(struct seq_file *seq,
1894 struct super_block *sb)
1895 {
1896 #ifdef CONFIG_QUOTA
1897 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1898
1899 if (F2FS_OPTION(sbi).s_jquota_fmt) {
1900 char *fmtname = "";
1901
1902 switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1903 case QFMT_VFS_OLD:
1904 fmtname = "vfsold";
1905 break;
1906 case QFMT_VFS_V0:
1907 fmtname = "vfsv0";
1908 break;
1909 case QFMT_VFS_V1:
1910 fmtname = "vfsv1";
1911 break;
1912 }
1913 seq_printf(seq, ",jqfmt=%s", fmtname);
1914 }
1915
1916 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1917 seq_show_option(seq, "usrjquota",
1918 F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1919
1920 if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1921 seq_show_option(seq, "grpjquota",
1922 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1923
1924 if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1925 seq_show_option(seq, "prjjquota",
1926 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1927 #endif
1928 }
1929
1930 #ifdef CONFIG_F2FS_FS_COMPRESSION
1931 static inline void f2fs_show_compress_options(struct seq_file *seq,
1932 struct super_block *sb)
1933 {
1934 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1935 char *algtype = "";
1936 int i;
1937
1938 if (!f2fs_sb_has_compression(sbi))
1939 return;
1940
1941 switch (F2FS_OPTION(sbi).compress_algorithm) {
1942 case COMPRESS_LZO:
1943 algtype = "lzo";
1944 break;
1945 case COMPRESS_LZ4:
1946 algtype = "lz4";
1947 break;
1948 case COMPRESS_ZSTD:
1949 algtype = "zstd";
1950 break;
1951 case COMPRESS_LZORLE:
1952 algtype = "lzo-rle";
1953 break;
1954 }
1955 seq_printf(seq, ",compress_algorithm=%s", algtype);
1956
1957 if (F2FS_OPTION(sbi).compress_level)
1958 seq_printf(seq, ":%d", F2FS_OPTION(sbi).compress_level);
1959
1960 seq_printf(seq, ",compress_log_size=%u",
1961 F2FS_OPTION(sbi).compress_log_size);
1962
1963 for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1964 seq_printf(seq, ",compress_extension=%s",
1965 F2FS_OPTION(sbi).extensions[i]);
1966 }
1967
1968 for (i = 0; i < F2FS_OPTION(sbi).nocompress_ext_cnt; i++) {
1969 seq_printf(seq, ",nocompress_extension=%s",
1970 F2FS_OPTION(sbi).noextensions[i]);
1971 }
1972
1973 if (F2FS_OPTION(sbi).compress_chksum)
1974 seq_puts(seq, ",compress_chksum");
1975
1976 if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_FS)
1977 seq_printf(seq, ",compress_mode=%s", "fs");
1978 else if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_USER)
1979 seq_printf(seq, ",compress_mode=%s", "user");
1980
1981 if (test_opt(sbi, COMPRESS_CACHE))
1982 seq_puts(seq, ",compress_cache");
1983 }
1984 #endif
1985
1986 static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1987 {
1988 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1989
1990 if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1991 seq_printf(seq, ",background_gc=%s", "sync");
1992 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1993 seq_printf(seq, ",background_gc=%s", "on");
1994 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1995 seq_printf(seq, ",background_gc=%s", "off");
1996
1997 if (test_opt(sbi, GC_MERGE))
1998 seq_puts(seq, ",gc_merge");
1999 else
2000 seq_puts(seq, ",nogc_merge");
2001
2002 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
2003 seq_puts(seq, ",disable_roll_forward");
2004 if (test_opt(sbi, NORECOVERY))
2005 seq_puts(seq, ",norecovery");
2006 if (test_opt(sbi, DISCARD)) {
2007 seq_puts(seq, ",discard");
2008 if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK)
2009 seq_printf(seq, ",discard_unit=%s", "block");
2010 else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SEGMENT)
2011 seq_printf(seq, ",discard_unit=%s", "segment");
2012 else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SECTION)
2013 seq_printf(seq, ",discard_unit=%s", "section");
2014 } else {
2015 seq_puts(seq, ",nodiscard");
2016 }
2017 if (test_opt(sbi, NOHEAP))
2018 seq_puts(seq, ",no_heap");
2019 else
2020 seq_puts(seq, ",heap");
2021 #ifdef CONFIG_F2FS_FS_XATTR
2022 if (test_opt(sbi, XATTR_USER))
2023 seq_puts(seq, ",user_xattr");
2024 else
2025 seq_puts(seq, ",nouser_xattr");
2026 if (test_opt(sbi, INLINE_XATTR))
2027 seq_puts(seq, ",inline_xattr");
2028 else
2029 seq_puts(seq, ",noinline_xattr");
2030 if (test_opt(sbi, INLINE_XATTR_SIZE))
2031 seq_printf(seq, ",inline_xattr_size=%u",
2032 F2FS_OPTION(sbi).inline_xattr_size);
2033 #endif
2034 #ifdef CONFIG_F2FS_FS_POSIX_ACL
2035 if (test_opt(sbi, POSIX_ACL))
2036 seq_puts(seq, ",acl");
2037 else
2038 seq_puts(seq, ",noacl");
2039 #endif
2040 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
2041 seq_puts(seq, ",disable_ext_identify");
2042 if (test_opt(sbi, INLINE_DATA))
2043 seq_puts(seq, ",inline_data");
2044 else
2045 seq_puts(seq, ",noinline_data");
2046 if (test_opt(sbi, INLINE_DENTRY))
2047 seq_puts(seq, ",inline_dentry");
2048 else
2049 seq_puts(seq, ",noinline_dentry");
2050 if (test_opt(sbi, FLUSH_MERGE))
2051 seq_puts(seq, ",flush_merge");
2052 else
2053 seq_puts(seq, ",noflush_merge");
2054 if (test_opt(sbi, NOBARRIER))
2055 seq_puts(seq, ",nobarrier");
2056 else
2057 seq_puts(seq, ",barrier");
2058 if (test_opt(sbi, FASTBOOT))
2059 seq_puts(seq, ",fastboot");
2060 if (test_opt(sbi, READ_EXTENT_CACHE))
2061 seq_puts(seq, ",extent_cache");
2062 else
2063 seq_puts(seq, ",noextent_cache");
2064 if (test_opt(sbi, AGE_EXTENT_CACHE))
2065 seq_puts(seq, ",age_extent_cache");
2066 if (test_opt(sbi, DATA_FLUSH))
2067 seq_puts(seq, ",data_flush");
2068
2069 seq_puts(seq, ",mode=");
2070 if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
2071 seq_puts(seq, "adaptive");
2072 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
2073 seq_puts(seq, "lfs");
2074 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG)
2075 seq_puts(seq, "fragment:segment");
2076 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK)
2077 seq_puts(seq, "fragment:block");
2078 seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
2079 if (test_opt(sbi, RESERVE_ROOT))
2080 seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
2081 F2FS_OPTION(sbi).root_reserved_blocks,
2082 from_kuid_munged(&init_user_ns,
2083 F2FS_OPTION(sbi).s_resuid),
2084 from_kgid_munged(&init_user_ns,
2085 F2FS_OPTION(sbi).s_resgid));
2086 if (F2FS_IO_SIZE_BITS(sbi))
2087 seq_printf(seq, ",io_bits=%u",
2088 F2FS_OPTION(sbi).write_io_size_bits);
2089 #ifdef CONFIG_F2FS_FAULT_INJECTION
2090 if (test_opt(sbi, FAULT_INJECTION)) {
2091 seq_printf(seq, ",fault_injection=%u",
2092 F2FS_OPTION(sbi).fault_info.inject_rate);
2093 seq_printf(seq, ",fault_type=%u",
2094 F2FS_OPTION(sbi).fault_info.inject_type);
2095 }
2096 #endif
2097 #ifdef CONFIG_QUOTA
2098 if (test_opt(sbi, QUOTA))
2099 seq_puts(seq, ",quota");
2100 if (test_opt(sbi, USRQUOTA))
2101 seq_puts(seq, ",usrquota");
2102 if (test_opt(sbi, GRPQUOTA))
2103 seq_puts(seq, ",grpquota");
2104 if (test_opt(sbi, PRJQUOTA))
2105 seq_puts(seq, ",prjquota");
2106 #endif
2107 f2fs_show_quota_options(seq, sbi->sb);
2108
2109 fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
2110
2111 if (sbi->sb->s_flags & SB_INLINECRYPT)
2112 seq_puts(seq, ",inlinecrypt");
2113
2114 if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
2115 seq_printf(seq, ",alloc_mode=%s", "default");
2116 else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
2117 seq_printf(seq, ",alloc_mode=%s", "reuse");
2118
2119 if (test_opt(sbi, DISABLE_CHECKPOINT))
2120 seq_printf(seq, ",checkpoint=disable:%u",
2121 F2FS_OPTION(sbi).unusable_cap);
2122 if (test_opt(sbi, MERGE_CHECKPOINT))
2123 seq_puts(seq, ",checkpoint_merge");
2124 else
2125 seq_puts(seq, ",nocheckpoint_merge");
2126 if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
2127 seq_printf(seq, ",fsync_mode=%s", "posix");
2128 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
2129 seq_printf(seq, ",fsync_mode=%s", "strict");
2130 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
2131 seq_printf(seq, ",fsync_mode=%s", "nobarrier");
2132
2133 #ifdef CONFIG_F2FS_FS_COMPRESSION
2134 f2fs_show_compress_options(seq, sbi->sb);
2135 #endif
2136
2137 if (test_opt(sbi, ATGC))
2138 seq_puts(seq, ",atgc");
2139
2140 if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_NORMAL)
2141 seq_printf(seq, ",memory=%s", "normal");
2142 else if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW)
2143 seq_printf(seq, ",memory=%s", "low");
2144
2145 if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_READONLY)
2146 seq_printf(seq, ",errors=%s", "remount-ro");
2147 else if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_CONTINUE)
2148 seq_printf(seq, ",errors=%s", "continue");
2149 else if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_PANIC)
2150 seq_printf(seq, ",errors=%s", "panic");
2151
2152 return 0;
2153 }
2154
2155 static void default_options(struct f2fs_sb_info *sbi, bool remount)
2156 {
2157 /* init some FS parameters */
2158 if (!remount) {
2159 set_opt(sbi, READ_EXTENT_CACHE);
2160 clear_opt(sbi, DISABLE_CHECKPOINT);
2161
2162 if (f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi))
2163 set_opt(sbi, DISCARD);
2164
2165 if (f2fs_sb_has_blkzoned(sbi))
2166 F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_SECTION;
2167 else
2168 F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_BLOCK;
2169 }
2170
2171 if (f2fs_sb_has_readonly(sbi))
2172 F2FS_OPTION(sbi).active_logs = NR_CURSEG_RO_TYPE;
2173 else
2174 F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
2175
2176 F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
2177 if (le32_to_cpu(F2FS_RAW_SUPER(sbi)->segment_count_main) <=
2178 SMALL_VOLUME_SEGMENTS)
2179 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
2180 else
2181 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
2182 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
2183 F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
2184 F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
2185 if (f2fs_sb_has_compression(sbi)) {
2186 F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
2187 F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
2188 F2FS_OPTION(sbi).compress_ext_cnt = 0;
2189 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
2190 }
2191 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
2192 F2FS_OPTION(sbi).memory_mode = MEMORY_MODE_NORMAL;
2193 F2FS_OPTION(sbi).errors = MOUNT_ERRORS_CONTINUE;
2194
2195 sbi->sb->s_flags &= ~SB_INLINECRYPT;
2196
2197 set_opt(sbi, INLINE_XATTR);
2198 set_opt(sbi, INLINE_DATA);
2199 set_opt(sbi, INLINE_DENTRY);
2200 set_opt(sbi, NOHEAP);
2201 set_opt(sbi, MERGE_CHECKPOINT);
2202 F2FS_OPTION(sbi).unusable_cap = 0;
2203 sbi->sb->s_flags |= SB_LAZYTIME;
2204 if (!f2fs_is_readonly(sbi))
2205 set_opt(sbi, FLUSH_MERGE);
2206 if (f2fs_sb_has_blkzoned(sbi))
2207 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
2208 else
2209 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
2210
2211 #ifdef CONFIG_F2FS_FS_XATTR
2212 set_opt(sbi, XATTR_USER);
2213 #endif
2214 #ifdef CONFIG_F2FS_FS_POSIX_ACL
2215 set_opt(sbi, POSIX_ACL);
2216 #endif
2217
2218 f2fs_build_fault_attr(sbi, 0, 0);
2219 }
2220
2221 #ifdef CONFIG_QUOTA
2222 static int f2fs_enable_quotas(struct super_block *sb);
2223 #endif
2224
2225 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
2226 {
2227 unsigned int s_flags = sbi->sb->s_flags;
2228 struct cp_control cpc;
2229 unsigned int gc_mode = sbi->gc_mode;
2230 int err = 0;
2231 int ret;
2232 block_t unusable;
2233
2234 if (s_flags & SB_RDONLY) {
2235 f2fs_err(sbi, "checkpoint=disable on readonly fs");
2236 return -EINVAL;
2237 }
2238 sbi->sb->s_flags |= SB_ACTIVE;
2239
2240 /* check if we need more GC first */
2241 unusable = f2fs_get_unusable_blocks(sbi);
2242 if (!f2fs_disable_cp_again(sbi, unusable))
2243 goto skip_gc;
2244
2245 f2fs_update_time(sbi, DISABLE_TIME);
2246
2247 sbi->gc_mode = GC_URGENT_HIGH;
2248
2249 while (!f2fs_time_over(sbi, DISABLE_TIME)) {
2250 struct f2fs_gc_control gc_control = {
2251 .victim_segno = NULL_SEGNO,
2252 .init_gc_type = FG_GC,
2253 .should_migrate_blocks = false,
2254 .err_gc_skipped = true,
2255 .nr_free_secs = 1 };
2256
2257 f2fs_down_write(&sbi->gc_lock);
2258 stat_inc_gc_call_count(sbi, FOREGROUND);
2259 err = f2fs_gc(sbi, &gc_control);
2260 if (err == -ENODATA) {
2261 err = 0;
2262 break;
2263 }
2264 if (err && err != -EAGAIN)
2265 break;
2266 }
2267
2268 ret = sync_filesystem(sbi->sb);
2269 if (ret || err) {
2270 err = ret ? ret : err;
2271 goto restore_flag;
2272 }
2273
2274 unusable = f2fs_get_unusable_blocks(sbi);
2275 if (f2fs_disable_cp_again(sbi, unusable)) {
2276 err = -EAGAIN;
2277 goto restore_flag;
2278 }
2279
2280 skip_gc:
2281 f2fs_down_write(&sbi->gc_lock);
2282 cpc.reason = CP_PAUSE;
2283 set_sbi_flag(sbi, SBI_CP_DISABLED);
2284 stat_inc_cp_call_count(sbi, TOTAL_CALL);
2285 err = f2fs_write_checkpoint(sbi, &cpc);
2286 if (err)
2287 goto out_unlock;
2288
2289 spin_lock(&sbi->stat_lock);
2290 sbi->unusable_block_count = unusable;
2291 spin_unlock(&sbi->stat_lock);
2292
2293 out_unlock:
2294 f2fs_up_write(&sbi->gc_lock);
2295 restore_flag:
2296 sbi->gc_mode = gc_mode;
2297 sbi->sb->s_flags = s_flags; /* Restore SB_RDONLY status */
2298 return err;
2299 }
2300
2301 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
2302 {
2303 int retry = DEFAULT_RETRY_IO_COUNT;
2304
2305 /* we should flush all the data to keep data consistency */
2306 do {
2307 sync_inodes_sb(sbi->sb);
2308 f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2309 } while (get_pages(sbi, F2FS_DIRTY_DATA) && retry--);
2310
2311 if (unlikely(retry < 0))
2312 f2fs_warn(sbi, "checkpoint=enable has some unwritten data.");
2313
2314 f2fs_down_write(&sbi->gc_lock);
2315 f2fs_dirty_to_prefree(sbi);
2316
2317 clear_sbi_flag(sbi, SBI_CP_DISABLED);
2318 set_sbi_flag(sbi, SBI_IS_DIRTY);
2319 f2fs_up_write(&sbi->gc_lock);
2320
2321 f2fs_sync_fs(sbi->sb, 1);
2322
2323 /* Let's ensure there's no pending checkpoint anymore */
2324 f2fs_flush_ckpt_thread(sbi);
2325 }
2326
2327 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
2328 {
2329 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2330 struct f2fs_mount_info org_mount_opt;
2331 unsigned long old_sb_flags;
2332 int err;
2333 bool need_restart_gc = false, need_stop_gc = false;
2334 bool need_restart_flush = false, need_stop_flush = false;
2335 bool need_restart_discard = false, need_stop_discard = false;
2336 bool need_enable_checkpoint = false, need_disable_checkpoint = false;
2337 bool no_read_extent_cache = !test_opt(sbi, READ_EXTENT_CACHE);
2338 bool no_age_extent_cache = !test_opt(sbi, AGE_EXTENT_CACHE);
2339 bool enable_checkpoint = !test_opt(sbi, DISABLE_CHECKPOINT);
2340 bool no_io_align = !F2FS_IO_ALIGNED(sbi);
2341 bool no_atgc = !test_opt(sbi, ATGC);
2342 bool no_discard = !test_opt(sbi, DISCARD);
2343 bool no_compress_cache = !test_opt(sbi, COMPRESS_CACHE);
2344 bool block_unit_discard = f2fs_block_unit_discard(sbi);
2345 #ifdef CONFIG_QUOTA
2346 int i, j;
2347 #endif
2348
2349 /*
2350 * Save the old mount options in case we
2351 * need to restore them.
2352 */
2353 org_mount_opt = sbi->mount_opt;
2354 old_sb_flags = sb->s_flags;
2355
2356 #ifdef CONFIG_QUOTA
2357 org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
2358 for (i = 0; i < MAXQUOTAS; i++) {
2359 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2360 org_mount_opt.s_qf_names[i] =
2361 kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
2362 GFP_KERNEL);
2363 if (!org_mount_opt.s_qf_names[i]) {
2364 for (j = 0; j < i; j++)
2365 kfree(org_mount_opt.s_qf_names[j]);
2366 return -ENOMEM;
2367 }
2368 } else {
2369 org_mount_opt.s_qf_names[i] = NULL;
2370 }
2371 }
2372 #endif
2373
2374 /* recover superblocks we couldn't write due to previous RO mount */
2375 if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
2376 err = f2fs_commit_super(sbi, false);
2377 f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
2378 err);
2379 if (!err)
2380 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2381 }
2382
2383 default_options(sbi, true);
2384
2385 /* parse mount options */
2386 err = parse_options(sb, data, true);
2387 if (err)
2388 goto restore_opts;
2389
2390 /* flush outstanding errors before changing fs state */
2391 flush_work(&sbi->s_error_work);
2392
2393 /*
2394 * Previous and new state of filesystem is RO,
2395 * so skip checking GC and FLUSH_MERGE conditions.
2396 */
2397 if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
2398 goto skip;
2399
2400 if (f2fs_dev_is_readonly(sbi) && !(*flags & SB_RDONLY)) {
2401 err = -EROFS;
2402 goto restore_opts;
2403 }
2404
2405 #ifdef CONFIG_QUOTA
2406 if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
2407 err = dquot_suspend(sb, -1);
2408 if (err < 0)
2409 goto restore_opts;
2410 } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
2411 /* dquot_resume needs RW */
2412 sb->s_flags &= ~SB_RDONLY;
2413 if (sb_any_quota_suspended(sb)) {
2414 dquot_resume(sb, -1);
2415 } else if (f2fs_sb_has_quota_ino(sbi)) {
2416 err = f2fs_enable_quotas(sb);
2417 if (err)
2418 goto restore_opts;
2419 }
2420 }
2421 #endif
2422 if (f2fs_lfs_mode(sbi) && !IS_F2FS_IPU_DISABLE(sbi)) {
2423 err = -EINVAL;
2424 f2fs_warn(sbi, "LFS is not compatible with IPU");
2425 goto restore_opts;
2426 }
2427
2428 /* disallow enable atgc dynamically */
2429 if (no_atgc == !!test_opt(sbi, ATGC)) {
2430 err = -EINVAL;
2431 f2fs_warn(sbi, "switch atgc option is not allowed");
2432 goto restore_opts;
2433 }
2434
2435 /* disallow enable/disable extent_cache dynamically */
2436 if (no_read_extent_cache == !!test_opt(sbi, READ_EXTENT_CACHE)) {
2437 err = -EINVAL;
2438 f2fs_warn(sbi, "switch extent_cache option is not allowed");
2439 goto restore_opts;
2440 }
2441 /* disallow enable/disable age extent_cache dynamically */
2442 if (no_age_extent_cache == !!test_opt(sbi, AGE_EXTENT_CACHE)) {
2443 err = -EINVAL;
2444 f2fs_warn(sbi, "switch age_extent_cache option is not allowed");
2445 goto restore_opts;
2446 }
2447
2448 if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
2449 err = -EINVAL;
2450 f2fs_warn(sbi, "switch io_bits option is not allowed");
2451 goto restore_opts;
2452 }
2453
2454 if (no_compress_cache == !!test_opt(sbi, COMPRESS_CACHE)) {
2455 err = -EINVAL;
2456 f2fs_warn(sbi, "switch compress_cache option is not allowed");
2457 goto restore_opts;
2458 }
2459
2460 if (block_unit_discard != f2fs_block_unit_discard(sbi)) {
2461 err = -EINVAL;
2462 f2fs_warn(sbi, "switch discard_unit option is not allowed");
2463 goto restore_opts;
2464 }
2465
2466 if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
2467 err = -EINVAL;
2468 f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
2469 goto restore_opts;
2470 }
2471
2472 /*
2473 * We stop the GC thread if FS is mounted as RO
2474 * or if background_gc = off is passed in mount
2475 * option. Also sync the filesystem.
2476 */
2477 if ((*flags & SB_RDONLY) ||
2478 (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF &&
2479 !test_opt(sbi, GC_MERGE))) {
2480 if (sbi->gc_thread) {
2481 f2fs_stop_gc_thread(sbi);
2482 need_restart_gc = true;
2483 }
2484 } else if (!sbi->gc_thread) {
2485 err = f2fs_start_gc_thread(sbi);
2486 if (err)
2487 goto restore_opts;
2488 need_stop_gc = true;
2489 }
2490
2491 if (*flags & SB_RDONLY) {
2492 sync_inodes_sb(sb);
2493
2494 set_sbi_flag(sbi, SBI_IS_DIRTY);
2495 set_sbi_flag(sbi, SBI_IS_CLOSE);
2496 f2fs_sync_fs(sb, 1);
2497 clear_sbi_flag(sbi, SBI_IS_CLOSE);
2498 }
2499
2500 /*
2501 * We stop issue flush thread if FS is mounted as RO
2502 * or if flush_merge is not passed in mount option.
2503 */
2504 if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2505 clear_opt(sbi, FLUSH_MERGE);
2506 f2fs_destroy_flush_cmd_control(sbi, false);
2507 need_restart_flush = true;
2508 } else {
2509 err = f2fs_create_flush_cmd_control(sbi);
2510 if (err)
2511 goto restore_gc;
2512 need_stop_flush = true;
2513 }
2514
2515 if (no_discard == !!test_opt(sbi, DISCARD)) {
2516 if (test_opt(sbi, DISCARD)) {
2517 err = f2fs_start_discard_thread(sbi);
2518 if (err)
2519 goto restore_flush;
2520 need_stop_discard = true;
2521 } else {
2522 f2fs_stop_discard_thread(sbi);
2523 f2fs_issue_discard_timeout(sbi);
2524 need_restart_discard = true;
2525 }
2526 }
2527
2528 if (enable_checkpoint == !!test_opt(sbi, DISABLE_CHECKPOINT)) {
2529 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
2530 err = f2fs_disable_checkpoint(sbi);
2531 if (err)
2532 goto restore_discard;
2533 need_enable_checkpoint = true;
2534 } else {
2535 f2fs_enable_checkpoint(sbi);
2536 need_disable_checkpoint = true;
2537 }
2538 }
2539
2540 /*
2541 * Place this routine at the end, since a new checkpoint would be
2542 * triggered while remount and we need to take care of it before
2543 * returning from remount.
2544 */
2545 if ((*flags & SB_RDONLY) || test_opt(sbi, DISABLE_CHECKPOINT) ||
2546 !test_opt(sbi, MERGE_CHECKPOINT)) {
2547 f2fs_stop_ckpt_thread(sbi);
2548 } else {
2549 /* Flush if the prevous checkpoint, if exists. */
2550 f2fs_flush_ckpt_thread(sbi);
2551
2552 err = f2fs_start_ckpt_thread(sbi);
2553 if (err) {
2554 f2fs_err(sbi,
2555 "Failed to start F2FS issue_checkpoint_thread (%d)",
2556 err);
2557 goto restore_checkpoint;
2558 }
2559 }
2560
2561 skip:
2562 #ifdef CONFIG_QUOTA
2563 /* Release old quota file names */
2564 for (i = 0; i < MAXQUOTAS; i++)
2565 kfree(org_mount_opt.s_qf_names[i]);
2566 #endif
2567 /* Update the POSIXACL Flag */
2568 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
2569 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
2570
2571 limit_reserve_root(sbi);
2572 adjust_unusable_cap_perc(sbi);
2573 *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
2574 return 0;
2575 restore_checkpoint:
2576 if (need_enable_checkpoint) {
2577 f2fs_enable_checkpoint(sbi);
2578 } else if (need_disable_checkpoint) {
2579 if (f2fs_disable_checkpoint(sbi))
2580 f2fs_warn(sbi, "checkpoint has not been disabled");
2581 }
2582 restore_discard:
2583 if (need_restart_discard) {
2584 if (f2fs_start_discard_thread(sbi))
2585 f2fs_warn(sbi, "discard has been stopped");
2586 } else if (need_stop_discard) {
2587 f2fs_stop_discard_thread(sbi);
2588 }
2589 restore_flush:
2590 if (need_restart_flush) {
2591 if (f2fs_create_flush_cmd_control(sbi))
2592 f2fs_warn(sbi, "background flush thread has stopped");
2593 } else if (need_stop_flush) {
2594 clear_opt(sbi, FLUSH_MERGE);
2595 f2fs_destroy_flush_cmd_control(sbi, false);
2596 }
2597 restore_gc:
2598 if (need_restart_gc) {
2599 if (f2fs_start_gc_thread(sbi))
2600 f2fs_warn(sbi, "background gc thread has stopped");
2601 } else if (need_stop_gc) {
2602 f2fs_stop_gc_thread(sbi);
2603 }
2604 restore_opts:
2605 #ifdef CONFIG_QUOTA
2606 F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
2607 for (i = 0; i < MAXQUOTAS; i++) {
2608 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
2609 F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
2610 }
2611 #endif
2612 sbi->mount_opt = org_mount_opt;
2613 sb->s_flags = old_sb_flags;
2614 return err;
2615 }
2616
2617 #ifdef CONFIG_QUOTA
2618 static bool f2fs_need_recovery(struct f2fs_sb_info *sbi)
2619 {
2620 /* need to recovery orphan */
2621 if (is_set_ckpt_flags(sbi, CP_ORPHAN_PRESENT_FLAG))
2622 return true;
2623 /* need to recovery data */
2624 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
2625 return false;
2626 if (test_opt(sbi, NORECOVERY))
2627 return false;
2628 return !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG);
2629 }
2630
2631 static bool f2fs_recover_quota_begin(struct f2fs_sb_info *sbi)
2632 {
2633 bool readonly = f2fs_readonly(sbi->sb);
2634
2635 if (!f2fs_need_recovery(sbi))
2636 return false;
2637
2638 /* it doesn't need to check f2fs_sb_has_readonly() */
2639 if (f2fs_hw_is_readonly(sbi))
2640 return false;
2641
2642 if (readonly) {
2643 sbi->sb->s_flags &= ~SB_RDONLY;
2644 set_sbi_flag(sbi, SBI_IS_WRITABLE);
2645 }
2646
2647 /*
2648 * Turn on quotas which were not enabled for read-only mounts if
2649 * filesystem has quota feature, so that they are updated correctly.
2650 */
2651 return f2fs_enable_quota_files(sbi, readonly);
2652 }
2653
2654 static void f2fs_recover_quota_end(struct f2fs_sb_info *sbi,
2655 bool quota_enabled)
2656 {
2657 if (quota_enabled)
2658 f2fs_quota_off_umount(sbi->sb);
2659
2660 if (is_sbi_flag_set(sbi, SBI_IS_WRITABLE)) {
2661 clear_sbi_flag(sbi, SBI_IS_WRITABLE);
2662 sbi->sb->s_flags |= SB_RDONLY;
2663 }
2664 }
2665
2666 /* Read data from quotafile */
2667 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
2668 size_t len, loff_t off)
2669 {
2670 struct inode *inode = sb_dqopt(sb)->files[type];
2671 struct address_space *mapping = inode->i_mapping;
2672 block_t blkidx = F2FS_BYTES_TO_BLK(off);
2673 int offset = off & (sb->s_blocksize - 1);
2674 int tocopy;
2675 size_t toread;
2676 loff_t i_size = i_size_read(inode);
2677 struct page *page;
2678
2679 if (off > i_size)
2680 return 0;
2681
2682 if (off + len > i_size)
2683 len = i_size - off;
2684 toread = len;
2685 while (toread > 0) {
2686 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
2687 repeat:
2688 page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
2689 if (IS_ERR(page)) {
2690 if (PTR_ERR(page) == -ENOMEM) {
2691 memalloc_retry_wait(GFP_NOFS);
2692 goto repeat;
2693 }
2694 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2695 return PTR_ERR(page);
2696 }
2697
2698 lock_page(page);
2699
2700 if (unlikely(page->mapping != mapping)) {
2701 f2fs_put_page(page, 1);
2702 goto repeat;
2703 }
2704 if (unlikely(!PageUptodate(page))) {
2705 f2fs_put_page(page, 1);
2706 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2707 return -EIO;
2708 }
2709
2710 memcpy_from_page(data, page, offset, tocopy);
2711 f2fs_put_page(page, 1);
2712
2713 offset = 0;
2714 toread -= tocopy;
2715 data += tocopy;
2716 blkidx++;
2717 }
2718 return len;
2719 }
2720
2721 /* Write to quotafile */
2722 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2723 const char *data, size_t len, loff_t off)
2724 {
2725 struct inode *inode = sb_dqopt(sb)->files[type];
2726 struct address_space *mapping = inode->i_mapping;
2727 const struct address_space_operations *a_ops = mapping->a_ops;
2728 int offset = off & (sb->s_blocksize - 1);
2729 size_t towrite = len;
2730 struct page *page;
2731 void *fsdata = NULL;
2732 int err = 0;
2733 int tocopy;
2734
2735 while (towrite > 0) {
2736 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2737 towrite);
2738 retry:
2739 err = a_ops->write_begin(NULL, mapping, off, tocopy,
2740 &page, &fsdata);
2741 if (unlikely(err)) {
2742 if (err == -ENOMEM) {
2743 f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2744 goto retry;
2745 }
2746 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2747 break;
2748 }
2749
2750 memcpy_to_page(page, offset, data, tocopy);
2751
2752 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2753 page, fsdata);
2754 offset = 0;
2755 towrite -= tocopy;
2756 off += tocopy;
2757 data += tocopy;
2758 cond_resched();
2759 }
2760
2761 if (len == towrite)
2762 return err;
2763 inode_set_mtime_to_ts(inode, inode_set_ctime_current(inode));
2764 f2fs_mark_inode_dirty_sync(inode, false);
2765 return len - towrite;
2766 }
2767
2768 int f2fs_dquot_initialize(struct inode *inode)
2769 {
2770 if (time_to_inject(F2FS_I_SB(inode), FAULT_DQUOT_INIT))
2771 return -ESRCH;
2772
2773 return dquot_initialize(inode);
2774 }
2775
2776 static struct dquot **f2fs_get_dquots(struct inode *inode)
2777 {
2778 return F2FS_I(inode)->i_dquot;
2779 }
2780
2781 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2782 {
2783 return &F2FS_I(inode)->i_reserved_quota;
2784 }
2785
2786 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2787 {
2788 if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2789 f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2790 return 0;
2791 }
2792
2793 return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2794 F2FS_OPTION(sbi).s_jquota_fmt, type);
2795 }
2796
2797 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2798 {
2799 int enabled = 0;
2800 int i, err;
2801
2802 if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2803 err = f2fs_enable_quotas(sbi->sb);
2804 if (err) {
2805 f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2806 return 0;
2807 }
2808 return 1;
2809 }
2810
2811 for (i = 0; i < MAXQUOTAS; i++) {
2812 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2813 err = f2fs_quota_on_mount(sbi, i);
2814 if (!err) {
2815 enabled = 1;
2816 continue;
2817 }
2818 f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2819 err, i);
2820 }
2821 }
2822 return enabled;
2823 }
2824
2825 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2826 unsigned int flags)
2827 {
2828 struct inode *qf_inode;
2829 unsigned long qf_inum;
2830 unsigned long qf_flag = F2FS_QUOTA_DEFAULT_FL;
2831 int err;
2832
2833 BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2834
2835 qf_inum = f2fs_qf_ino(sb, type);
2836 if (!qf_inum)
2837 return -EPERM;
2838
2839 qf_inode = f2fs_iget(sb, qf_inum);
2840 if (IS_ERR(qf_inode)) {
2841 f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2842 return PTR_ERR(qf_inode);
2843 }
2844
2845 /* Don't account quota for quota files to avoid recursion */
2846 inode_lock(qf_inode);
2847 qf_inode->i_flags |= S_NOQUOTA;
2848
2849 if ((F2FS_I(qf_inode)->i_flags & qf_flag) != qf_flag) {
2850 F2FS_I(qf_inode)->i_flags |= qf_flag;
2851 f2fs_set_inode_flags(qf_inode);
2852 }
2853 inode_unlock(qf_inode);
2854
2855 err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2856 iput(qf_inode);
2857 return err;
2858 }
2859
2860 static int f2fs_enable_quotas(struct super_block *sb)
2861 {
2862 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2863 int type, err = 0;
2864 unsigned long qf_inum;
2865 bool quota_mopt[MAXQUOTAS] = {
2866 test_opt(sbi, USRQUOTA),
2867 test_opt(sbi, GRPQUOTA),
2868 test_opt(sbi, PRJQUOTA),
2869 };
2870
2871 if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2872 f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2873 return 0;
2874 }
2875
2876 sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2877
2878 for (type = 0; type < MAXQUOTAS; type++) {
2879 qf_inum = f2fs_qf_ino(sb, type);
2880 if (qf_inum) {
2881 err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2882 DQUOT_USAGE_ENABLED |
2883 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2884 if (err) {
2885 f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2886 type, err);
2887 for (type--; type >= 0; type--)
2888 dquot_quota_off(sb, type);
2889 set_sbi_flag(F2FS_SB(sb),
2890 SBI_QUOTA_NEED_REPAIR);
2891 return err;
2892 }
2893 }
2894 }
2895 return 0;
2896 }
2897
2898 static int f2fs_quota_sync_file(struct f2fs_sb_info *sbi, int type)
2899 {
2900 struct quota_info *dqopt = sb_dqopt(sbi->sb);
2901 struct address_space *mapping = dqopt->files[type]->i_mapping;
2902 int ret = 0;
2903
2904 ret = dquot_writeback_dquots(sbi->sb, type);
2905 if (ret)
2906 goto out;
2907
2908 ret = filemap_fdatawrite(mapping);
2909 if (ret)
2910 goto out;
2911
2912 /* if we are using journalled quota */
2913 if (is_journalled_quota(sbi))
2914 goto out;
2915
2916 ret = filemap_fdatawait(mapping);
2917
2918 truncate_inode_pages(&dqopt->files[type]->i_data, 0);
2919 out:
2920 if (ret)
2921 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2922 return ret;
2923 }
2924
2925 int f2fs_quota_sync(struct super_block *sb, int type)
2926 {
2927 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2928 struct quota_info *dqopt = sb_dqopt(sb);
2929 int cnt;
2930 int ret = 0;
2931
2932 /*
2933 * Now when everything is written we can discard the pagecache so
2934 * that userspace sees the changes.
2935 */
2936 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2937
2938 if (type != -1 && cnt != type)
2939 continue;
2940
2941 if (!sb_has_quota_active(sb, cnt))
2942 continue;
2943
2944 if (!f2fs_sb_has_quota_ino(sbi))
2945 inode_lock(dqopt->files[cnt]);
2946
2947 /*
2948 * do_quotactl
2949 * f2fs_quota_sync
2950 * f2fs_down_read(quota_sem)
2951 * dquot_writeback_dquots()
2952 * f2fs_dquot_commit
2953 * block_operation
2954 * f2fs_down_read(quota_sem)
2955 */
2956 f2fs_lock_op(sbi);
2957 f2fs_down_read(&sbi->quota_sem);
2958
2959 ret = f2fs_quota_sync_file(sbi, cnt);
2960
2961 f2fs_up_read(&sbi->quota_sem);
2962 f2fs_unlock_op(sbi);
2963
2964 if (!f2fs_sb_has_quota_ino(sbi))
2965 inode_unlock(dqopt->files[cnt]);
2966
2967 if (ret)
2968 break;
2969 }
2970 return ret;
2971 }
2972
2973 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2974 const struct path *path)
2975 {
2976 struct inode *inode;
2977 int err;
2978
2979 /* if quota sysfile exists, deny enabling quota with specific file */
2980 if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2981 f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2982 return -EBUSY;
2983 }
2984
2985 if (path->dentry->d_sb != sb)
2986 return -EXDEV;
2987
2988 err = f2fs_quota_sync(sb, type);
2989 if (err)
2990 return err;
2991
2992 inode = d_inode(path->dentry);
2993
2994 err = filemap_fdatawrite(inode->i_mapping);
2995 if (err)
2996 return err;
2997
2998 err = filemap_fdatawait(inode->i_mapping);
2999 if (err)
3000 return err;
3001
3002 err = dquot_quota_on(sb, type, format_id, path);
3003 if (err)
3004 return err;
3005
3006 inode_lock(inode);
3007 F2FS_I(inode)->i_flags |= F2FS_QUOTA_DEFAULT_FL;
3008 f2fs_set_inode_flags(inode);
3009 inode_unlock(inode);
3010 f2fs_mark_inode_dirty_sync(inode, false);
3011
3012 return 0;
3013 }
3014
3015 static int __f2fs_quota_off(struct super_block *sb, int type)
3016 {
3017 struct inode *inode = sb_dqopt(sb)->files[type];
3018 int err;
3019
3020 if (!inode || !igrab(inode))
3021 return dquot_quota_off(sb, type);
3022
3023 err = f2fs_quota_sync(sb, type);
3024 if (err)
3025 goto out_put;
3026
3027 err = dquot_quota_off(sb, type);
3028 if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
3029 goto out_put;
3030
3031 inode_lock(inode);
3032 F2FS_I(inode)->i_flags &= ~F2FS_QUOTA_DEFAULT_FL;
3033 f2fs_set_inode_flags(inode);
3034 inode_unlock(inode);
3035 f2fs_mark_inode_dirty_sync(inode, false);
3036 out_put:
3037 iput(inode);
3038 return err;
3039 }
3040
3041 static int f2fs_quota_off(struct super_block *sb, int type)
3042 {
3043 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3044 int err;
3045
3046 err = __f2fs_quota_off(sb, type);
3047
3048 /*
3049 * quotactl can shutdown journalled quota, result in inconsistence
3050 * between quota record and fs data by following updates, tag the
3051 * flag to let fsck be aware of it.
3052 */
3053 if (is_journalled_quota(sbi))
3054 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3055 return err;
3056 }
3057
3058 void f2fs_quota_off_umount(struct super_block *sb)
3059 {
3060 int type;
3061 int err;
3062
3063 for (type = 0; type < MAXQUOTAS; type++) {
3064 err = __f2fs_quota_off(sb, type);
3065 if (err) {
3066 int ret = dquot_quota_off(sb, type);
3067
3068 f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
3069 type, err, ret);
3070 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
3071 }
3072 }
3073 /*
3074 * In case of checkpoint=disable, we must flush quota blocks.
3075 * This can cause NULL exception for node_inode in end_io, since
3076 * put_super already dropped it.
3077 */
3078 sync_filesystem(sb);
3079 }
3080
3081 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
3082 {
3083 struct quota_info *dqopt = sb_dqopt(sb);
3084 int type;
3085
3086 for (type = 0; type < MAXQUOTAS; type++) {
3087 if (!dqopt->files[type])
3088 continue;
3089 f2fs_inode_synced(dqopt->files[type]);
3090 }
3091 }
3092
3093 static int f2fs_dquot_commit(struct dquot *dquot)
3094 {
3095 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3096 int ret;
3097
3098 f2fs_down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
3099 ret = dquot_commit(dquot);
3100 if (ret < 0)
3101 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3102 f2fs_up_read(&sbi->quota_sem);
3103 return ret;
3104 }
3105
3106 static int f2fs_dquot_acquire(struct dquot *dquot)
3107 {
3108 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3109 int ret;
3110
3111 f2fs_down_read(&sbi->quota_sem);
3112 ret = dquot_acquire(dquot);
3113 if (ret < 0)
3114 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3115 f2fs_up_read(&sbi->quota_sem);
3116 return ret;
3117 }
3118
3119 static int f2fs_dquot_release(struct dquot *dquot)
3120 {
3121 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3122 int ret = dquot_release(dquot);
3123
3124 if (ret < 0)
3125 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3126 return ret;
3127 }
3128
3129 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
3130 {
3131 struct super_block *sb = dquot->dq_sb;
3132 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3133 int ret = dquot_mark_dquot_dirty(dquot);
3134
3135 /* if we are using journalled quota */
3136 if (is_journalled_quota(sbi))
3137 set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
3138
3139 return ret;
3140 }
3141
3142 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
3143 {
3144 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3145 int ret = dquot_commit_info(sb, type);
3146
3147 if (ret < 0)
3148 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3149 return ret;
3150 }
3151
3152 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
3153 {
3154 *projid = F2FS_I(inode)->i_projid;
3155 return 0;
3156 }
3157
3158 static const struct dquot_operations f2fs_quota_operations = {
3159 .get_reserved_space = f2fs_get_reserved_space,
3160 .write_dquot = f2fs_dquot_commit,
3161 .acquire_dquot = f2fs_dquot_acquire,
3162 .release_dquot = f2fs_dquot_release,
3163 .mark_dirty = f2fs_dquot_mark_dquot_dirty,
3164 .write_info = f2fs_dquot_commit_info,
3165 .alloc_dquot = dquot_alloc,
3166 .destroy_dquot = dquot_destroy,
3167 .get_projid = f2fs_get_projid,
3168 .get_next_id = dquot_get_next_id,
3169 };
3170
3171 static const struct quotactl_ops f2fs_quotactl_ops = {
3172 .quota_on = f2fs_quota_on,
3173 .quota_off = f2fs_quota_off,
3174 .quota_sync = f2fs_quota_sync,
3175 .get_state = dquot_get_state,
3176 .set_info = dquot_set_dqinfo,
3177 .get_dqblk = dquot_get_dqblk,
3178 .set_dqblk = dquot_set_dqblk,
3179 .get_nextdqblk = dquot_get_next_dqblk,
3180 };
3181 #else
3182 int f2fs_dquot_initialize(struct inode *inode)
3183 {
3184 return 0;
3185 }
3186
3187 int f2fs_quota_sync(struct super_block *sb, int type)
3188 {
3189 return 0;
3190 }
3191
3192 void f2fs_quota_off_umount(struct super_block *sb)
3193 {
3194 }
3195 #endif
3196
3197 static const struct super_operations f2fs_sops = {
3198 .alloc_inode = f2fs_alloc_inode,
3199 .free_inode = f2fs_free_inode,
3200 .drop_inode = f2fs_drop_inode,
3201 .write_inode = f2fs_write_inode,
3202 .dirty_inode = f2fs_dirty_inode,
3203 .show_options = f2fs_show_options,
3204 #ifdef CONFIG_QUOTA
3205 .quota_read = f2fs_quota_read,
3206 .quota_write = f2fs_quota_write,
3207 .get_dquots = f2fs_get_dquots,
3208 #endif
3209 .evict_inode = f2fs_evict_inode,
3210 .put_super = f2fs_put_super,
3211 .sync_fs = f2fs_sync_fs,
3212 .freeze_fs = f2fs_freeze,
3213 .unfreeze_fs = f2fs_unfreeze,
3214 .statfs = f2fs_statfs,
3215 .remount_fs = f2fs_remount,
3216 };
3217
3218 #ifdef CONFIG_FS_ENCRYPTION
3219 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
3220 {
3221 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3222 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3223 ctx, len, NULL);
3224 }
3225
3226 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
3227 void *fs_data)
3228 {
3229 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3230
3231 /*
3232 * Encrypting the root directory is not allowed because fsck
3233 * expects lost+found directory to exist and remain unencrypted
3234 * if LOST_FOUND feature is enabled.
3235 *
3236 */
3237 if (f2fs_sb_has_lost_found(sbi) &&
3238 inode->i_ino == F2FS_ROOT_INO(sbi))
3239 return -EPERM;
3240
3241 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3242 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3243 ctx, len, fs_data, XATTR_CREATE);
3244 }
3245
3246 static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
3247 {
3248 return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
3249 }
3250
3251 static bool f2fs_has_stable_inodes(struct super_block *sb)
3252 {
3253 return true;
3254 }
3255
3256 static struct block_device **f2fs_get_devices(struct super_block *sb,
3257 unsigned int *num_devs)
3258 {
3259 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3260 struct block_device **devs;
3261 int i;
3262
3263 if (!f2fs_is_multi_device(sbi))
3264 return NULL;
3265
3266 devs = kmalloc_array(sbi->s_ndevs, sizeof(*devs), GFP_KERNEL);
3267 if (!devs)
3268 return ERR_PTR(-ENOMEM);
3269
3270 for (i = 0; i < sbi->s_ndevs; i++)
3271 devs[i] = FDEV(i).bdev;
3272 *num_devs = sbi->s_ndevs;
3273 return devs;
3274 }
3275
3276 static const struct fscrypt_operations f2fs_cryptops = {
3277 .needs_bounce_pages = 1,
3278 .has_32bit_inodes = 1,
3279 .supports_subblock_data_units = 1,
3280 .legacy_key_prefix = "f2fs:",
3281 .get_context = f2fs_get_context,
3282 .set_context = f2fs_set_context,
3283 .get_dummy_policy = f2fs_get_dummy_policy,
3284 .empty_dir = f2fs_empty_dir,
3285 .has_stable_inodes = f2fs_has_stable_inodes,
3286 .get_devices = f2fs_get_devices,
3287 };
3288 #endif
3289
3290 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
3291 u64 ino, u32 generation)
3292 {
3293 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3294 struct inode *inode;
3295
3296 if (f2fs_check_nid_range(sbi, ino))
3297 return ERR_PTR(-ESTALE);
3298
3299 /*
3300 * f2fs_iget isn't quite right if the inode is currently unallocated!
3301 * However f2fs_iget currently does appropriate checks to handle stale
3302 * inodes so everything is OK.
3303 */
3304 inode = f2fs_iget(sb, ino);
3305 if (IS_ERR(inode))
3306 return ERR_CAST(inode);
3307 if (unlikely(generation && inode->i_generation != generation)) {
3308 /* we didn't find the right inode.. */
3309 iput(inode);
3310 return ERR_PTR(-ESTALE);
3311 }
3312 return inode;
3313 }
3314
3315 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
3316 int fh_len, int fh_type)
3317 {
3318 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
3319 f2fs_nfs_get_inode);
3320 }
3321
3322 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
3323 int fh_len, int fh_type)
3324 {
3325 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
3326 f2fs_nfs_get_inode);
3327 }
3328
3329 static const struct export_operations f2fs_export_ops = {
3330 .encode_fh = generic_encode_ino32_fh,
3331 .fh_to_dentry = f2fs_fh_to_dentry,
3332 .fh_to_parent = f2fs_fh_to_parent,
3333 .get_parent = f2fs_get_parent,
3334 };
3335
3336 loff_t max_file_blocks(struct inode *inode)
3337 {
3338 loff_t result = 0;
3339 loff_t leaf_count;
3340
3341 /*
3342 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
3343 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
3344 * space in inode.i_addr, it will be more safe to reassign
3345 * result as zero.
3346 */
3347
3348 if (inode && f2fs_compressed_file(inode))
3349 leaf_count = ADDRS_PER_BLOCK(inode);
3350 else
3351 leaf_count = DEF_ADDRS_PER_BLOCK;
3352
3353 /* two direct node blocks */
3354 result += (leaf_count * 2);
3355
3356 /* two indirect node blocks */
3357 leaf_count *= NIDS_PER_BLOCK;
3358 result += (leaf_count * 2);
3359
3360 /* one double indirect node block */
3361 leaf_count *= NIDS_PER_BLOCK;
3362 result += leaf_count;
3363
3364 return result;
3365 }
3366
3367 static int __f2fs_commit_super(struct buffer_head *bh,
3368 struct f2fs_super_block *super)
3369 {
3370 lock_buffer(bh);
3371 if (super)
3372 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
3373 set_buffer_dirty(bh);
3374 unlock_buffer(bh);
3375
3376 /* it's rare case, we can do fua all the time */
3377 return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
3378 }
3379
3380 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
3381 struct buffer_head *bh)
3382 {
3383 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3384 (bh->b_data + F2FS_SUPER_OFFSET);
3385 struct super_block *sb = sbi->sb;
3386 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
3387 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
3388 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
3389 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
3390 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
3391 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
3392 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
3393 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
3394 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
3395 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
3396 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3397 u32 segment_count = le32_to_cpu(raw_super->segment_count);
3398 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3399 u64 main_end_blkaddr = main_blkaddr +
3400 (segment_count_main << log_blocks_per_seg);
3401 u64 seg_end_blkaddr = segment0_blkaddr +
3402 (segment_count << log_blocks_per_seg);
3403
3404 if (segment0_blkaddr != cp_blkaddr) {
3405 f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
3406 segment0_blkaddr, cp_blkaddr);
3407 return true;
3408 }
3409
3410 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
3411 sit_blkaddr) {
3412 f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
3413 cp_blkaddr, sit_blkaddr,
3414 segment_count_ckpt << log_blocks_per_seg);
3415 return true;
3416 }
3417
3418 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
3419 nat_blkaddr) {
3420 f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
3421 sit_blkaddr, nat_blkaddr,
3422 segment_count_sit << log_blocks_per_seg);
3423 return true;
3424 }
3425
3426 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
3427 ssa_blkaddr) {
3428 f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
3429 nat_blkaddr, ssa_blkaddr,
3430 segment_count_nat << log_blocks_per_seg);
3431 return true;
3432 }
3433
3434 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
3435 main_blkaddr) {
3436 f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
3437 ssa_blkaddr, main_blkaddr,
3438 segment_count_ssa << log_blocks_per_seg);
3439 return true;
3440 }
3441
3442 if (main_end_blkaddr > seg_end_blkaddr) {
3443 f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
3444 main_blkaddr, seg_end_blkaddr,
3445 segment_count_main << log_blocks_per_seg);
3446 return true;
3447 } else if (main_end_blkaddr < seg_end_blkaddr) {
3448 int err = 0;
3449 char *res;
3450
3451 /* fix in-memory information all the time */
3452 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
3453 segment0_blkaddr) >> log_blocks_per_seg);
3454
3455 if (f2fs_readonly(sb) || f2fs_hw_is_readonly(sbi)) {
3456 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3457 res = "internally";
3458 } else {
3459 err = __f2fs_commit_super(bh, NULL);
3460 res = err ? "failed" : "done";
3461 }
3462 f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
3463 res, main_blkaddr, seg_end_blkaddr,
3464 segment_count_main << log_blocks_per_seg);
3465 if (err)
3466 return true;
3467 }
3468 return false;
3469 }
3470
3471 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
3472 struct buffer_head *bh)
3473 {
3474 block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
3475 block_t total_sections, blocks_per_seg;
3476 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3477 (bh->b_data + F2FS_SUPER_OFFSET);
3478 size_t crc_offset = 0;
3479 __u32 crc = 0;
3480
3481 if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
3482 f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
3483 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
3484 return -EINVAL;
3485 }
3486
3487 /* Check checksum_offset and crc in superblock */
3488 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
3489 crc_offset = le32_to_cpu(raw_super->checksum_offset);
3490 if (crc_offset !=
3491 offsetof(struct f2fs_super_block, crc)) {
3492 f2fs_info(sbi, "Invalid SB checksum offset: %zu",
3493 crc_offset);
3494 return -EFSCORRUPTED;
3495 }
3496 crc = le32_to_cpu(raw_super->crc);
3497 if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
3498 f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
3499 return -EFSCORRUPTED;
3500 }
3501 }
3502
3503 /* Currently, support only 4KB block size */
3504 if (le32_to_cpu(raw_super->log_blocksize) != F2FS_BLKSIZE_BITS) {
3505 f2fs_info(sbi, "Invalid log_blocksize (%u), supports only %u",
3506 le32_to_cpu(raw_super->log_blocksize),
3507 F2FS_BLKSIZE_BITS);
3508 return -EFSCORRUPTED;
3509 }
3510
3511 /* check log blocks per segment */
3512 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
3513 f2fs_info(sbi, "Invalid log blocks per segment (%u)",
3514 le32_to_cpu(raw_super->log_blocks_per_seg));
3515 return -EFSCORRUPTED;
3516 }
3517
3518 /* Currently, support 512/1024/2048/4096/16K bytes sector size */
3519 if (le32_to_cpu(raw_super->log_sectorsize) >
3520 F2FS_MAX_LOG_SECTOR_SIZE ||
3521 le32_to_cpu(raw_super->log_sectorsize) <
3522 F2FS_MIN_LOG_SECTOR_SIZE) {
3523 f2fs_info(sbi, "Invalid log sectorsize (%u)",
3524 le32_to_cpu(raw_super->log_sectorsize));
3525 return -EFSCORRUPTED;
3526 }
3527 if (le32_to_cpu(raw_super->log_sectors_per_block) +
3528 le32_to_cpu(raw_super->log_sectorsize) !=
3529 F2FS_MAX_LOG_SECTOR_SIZE) {
3530 f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
3531 le32_to_cpu(raw_super->log_sectors_per_block),
3532 le32_to_cpu(raw_super->log_sectorsize));
3533 return -EFSCORRUPTED;
3534 }
3535
3536 segment_count = le32_to_cpu(raw_super->segment_count);
3537 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3538 segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3539 secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3540 total_sections = le32_to_cpu(raw_super->section_count);
3541
3542 /* blocks_per_seg should be 512, given the above check */
3543 blocks_per_seg = BIT(le32_to_cpu(raw_super->log_blocks_per_seg));
3544
3545 if (segment_count > F2FS_MAX_SEGMENT ||
3546 segment_count < F2FS_MIN_SEGMENTS) {
3547 f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
3548 return -EFSCORRUPTED;
3549 }
3550
3551 if (total_sections > segment_count_main || total_sections < 1 ||
3552 segs_per_sec > segment_count || !segs_per_sec) {
3553 f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
3554 segment_count, total_sections, segs_per_sec);
3555 return -EFSCORRUPTED;
3556 }
3557
3558 if (segment_count_main != total_sections * segs_per_sec) {
3559 f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
3560 segment_count_main, total_sections, segs_per_sec);
3561 return -EFSCORRUPTED;
3562 }
3563
3564 if ((segment_count / segs_per_sec) < total_sections) {
3565 f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
3566 segment_count, segs_per_sec, total_sections);
3567 return -EFSCORRUPTED;
3568 }
3569
3570 if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
3571 f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
3572 segment_count, le64_to_cpu(raw_super->block_count));
3573 return -EFSCORRUPTED;
3574 }
3575
3576 if (RDEV(0).path[0]) {
3577 block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
3578 int i = 1;
3579
3580 while (i < MAX_DEVICES && RDEV(i).path[0]) {
3581 dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
3582 i++;
3583 }
3584 if (segment_count != dev_seg_count) {
3585 f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
3586 segment_count, dev_seg_count);
3587 return -EFSCORRUPTED;
3588 }
3589 } else {
3590 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
3591 !bdev_is_zoned(sbi->sb->s_bdev)) {
3592 f2fs_info(sbi, "Zoned block device path is missing");
3593 return -EFSCORRUPTED;
3594 }
3595 }
3596
3597 if (secs_per_zone > total_sections || !secs_per_zone) {
3598 f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
3599 secs_per_zone, total_sections);
3600 return -EFSCORRUPTED;
3601 }
3602 if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
3603 raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
3604 (le32_to_cpu(raw_super->extension_count) +
3605 raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
3606 f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
3607 le32_to_cpu(raw_super->extension_count),
3608 raw_super->hot_ext_count,
3609 F2FS_MAX_EXTENSION);
3610 return -EFSCORRUPTED;
3611 }
3612
3613 if (le32_to_cpu(raw_super->cp_payload) >=
3614 (blocks_per_seg - F2FS_CP_PACKS -
3615 NR_CURSEG_PERSIST_TYPE)) {
3616 f2fs_info(sbi, "Insane cp_payload (%u >= %u)",
3617 le32_to_cpu(raw_super->cp_payload),
3618 blocks_per_seg - F2FS_CP_PACKS -
3619 NR_CURSEG_PERSIST_TYPE);
3620 return -EFSCORRUPTED;
3621 }
3622
3623 /* check reserved ino info */
3624 if (le32_to_cpu(raw_super->node_ino) != 1 ||
3625 le32_to_cpu(raw_super->meta_ino) != 2 ||
3626 le32_to_cpu(raw_super->root_ino) != 3) {
3627 f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
3628 le32_to_cpu(raw_super->node_ino),
3629 le32_to_cpu(raw_super->meta_ino),
3630 le32_to_cpu(raw_super->root_ino));
3631 return -EFSCORRUPTED;
3632 }
3633
3634 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
3635 if (sanity_check_area_boundary(sbi, bh))
3636 return -EFSCORRUPTED;
3637
3638 return 0;
3639 }
3640
3641 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
3642 {
3643 unsigned int total, fsmeta;
3644 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3645 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
3646 unsigned int ovp_segments, reserved_segments;
3647 unsigned int main_segs, blocks_per_seg;
3648 unsigned int sit_segs, nat_segs;
3649 unsigned int sit_bitmap_size, nat_bitmap_size;
3650 unsigned int log_blocks_per_seg;
3651 unsigned int segment_count_main;
3652 unsigned int cp_pack_start_sum, cp_payload;
3653 block_t user_block_count, valid_user_blocks;
3654 block_t avail_node_count, valid_node_count;
3655 unsigned int nat_blocks, nat_bits_bytes, nat_bits_blocks;
3656 int i, j;
3657
3658 total = le32_to_cpu(raw_super->segment_count);
3659 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
3660 sit_segs = le32_to_cpu(raw_super->segment_count_sit);
3661 fsmeta += sit_segs;
3662 nat_segs = le32_to_cpu(raw_super->segment_count_nat);
3663 fsmeta += nat_segs;
3664 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
3665 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
3666
3667 if (unlikely(fsmeta >= total))
3668 return 1;
3669
3670 ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
3671 reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
3672
3673 if (!f2fs_sb_has_readonly(sbi) &&
3674 unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
3675 ovp_segments == 0 || reserved_segments == 0)) {
3676 f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
3677 return 1;
3678 }
3679 user_block_count = le64_to_cpu(ckpt->user_block_count);
3680 segment_count_main = le32_to_cpu(raw_super->segment_count_main) +
3681 (f2fs_sb_has_readonly(sbi) ? 1 : 0);
3682 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3683 if (!user_block_count || user_block_count >=
3684 segment_count_main << log_blocks_per_seg) {
3685 f2fs_err(sbi, "Wrong user_block_count: %u",
3686 user_block_count);
3687 return 1;
3688 }
3689
3690 valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
3691 if (valid_user_blocks > user_block_count) {
3692 f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
3693 valid_user_blocks, user_block_count);
3694 return 1;
3695 }
3696
3697 valid_node_count = le32_to_cpu(ckpt->valid_node_count);
3698 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
3699 if (valid_node_count > avail_node_count) {
3700 f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
3701 valid_node_count, avail_node_count);
3702 return 1;
3703 }
3704
3705 main_segs = le32_to_cpu(raw_super->segment_count_main);
3706 blocks_per_seg = sbi->blocks_per_seg;
3707
3708 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3709 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
3710 le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
3711 return 1;
3712
3713 if (f2fs_sb_has_readonly(sbi))
3714 goto check_data;
3715
3716 for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
3717 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3718 le32_to_cpu(ckpt->cur_node_segno[j])) {
3719 f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
3720 i, j,
3721 le32_to_cpu(ckpt->cur_node_segno[i]));
3722 return 1;
3723 }
3724 }
3725 }
3726 check_data:
3727 for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
3728 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
3729 le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
3730 return 1;
3731
3732 if (f2fs_sb_has_readonly(sbi))
3733 goto skip_cross;
3734
3735 for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3736 if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3737 le32_to_cpu(ckpt->cur_data_segno[j])) {
3738 f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3739 i, j,
3740 le32_to_cpu(ckpt->cur_data_segno[i]));
3741 return 1;
3742 }
3743 }
3744 }
3745 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3746 for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3747 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3748 le32_to_cpu(ckpt->cur_data_segno[j])) {
3749 f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3750 i, j,
3751 le32_to_cpu(ckpt->cur_node_segno[i]));
3752 return 1;
3753 }
3754 }
3755 }
3756 skip_cross:
3757 sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3758 nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3759
3760 if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3761 nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3762 f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3763 sit_bitmap_size, nat_bitmap_size);
3764 return 1;
3765 }
3766
3767 cp_pack_start_sum = __start_sum_addr(sbi);
3768 cp_payload = __cp_payload(sbi);
3769 if (cp_pack_start_sum < cp_payload + 1 ||
3770 cp_pack_start_sum > blocks_per_seg - 1 -
3771 NR_CURSEG_PERSIST_TYPE) {
3772 f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3773 cp_pack_start_sum);
3774 return 1;
3775 }
3776
3777 if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3778 le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3779 f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3780 "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3781 "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3782 le32_to_cpu(ckpt->checksum_offset));
3783 return 1;
3784 }
3785
3786 nat_blocks = nat_segs << log_blocks_per_seg;
3787 nat_bits_bytes = nat_blocks / BITS_PER_BYTE;
3788 nat_bits_blocks = F2FS_BLK_ALIGN((nat_bits_bytes << 1) + 8);
3789 if (__is_set_ckpt_flags(ckpt, CP_NAT_BITS_FLAG) &&
3790 (cp_payload + F2FS_CP_PACKS +
3791 NR_CURSEG_PERSIST_TYPE + nat_bits_blocks >= blocks_per_seg)) {
3792 f2fs_warn(sbi, "Insane cp_payload: %u, nat_bits_blocks: %u)",
3793 cp_payload, nat_bits_blocks);
3794 return 1;
3795 }
3796
3797 if (unlikely(f2fs_cp_error(sbi))) {
3798 f2fs_err(sbi, "A bug case: need to run fsck");
3799 return 1;
3800 }
3801 return 0;
3802 }
3803
3804 static void init_sb_info(struct f2fs_sb_info *sbi)
3805 {
3806 struct f2fs_super_block *raw_super = sbi->raw_super;
3807 int i;
3808
3809 sbi->log_sectors_per_block =
3810 le32_to_cpu(raw_super->log_sectors_per_block);
3811 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3812 sbi->blocksize = BIT(sbi->log_blocksize);
3813 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3814 sbi->blocks_per_seg = BIT(sbi->log_blocks_per_seg);
3815 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3816 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3817 sbi->total_sections = le32_to_cpu(raw_super->section_count);
3818 sbi->total_node_count =
3819 (le32_to_cpu(raw_super->segment_count_nat) / 2)
3820 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
3821 F2FS_ROOT_INO(sbi) = le32_to_cpu(raw_super->root_ino);
3822 F2FS_NODE_INO(sbi) = le32_to_cpu(raw_super->node_ino);
3823 F2FS_META_INO(sbi) = le32_to_cpu(raw_super->meta_ino);
3824 sbi->cur_victim_sec = NULL_SECNO;
3825 sbi->gc_mode = GC_NORMAL;
3826 sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3827 sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3828 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3829 sbi->migration_granularity = sbi->segs_per_sec;
3830 sbi->seq_file_ra_mul = MIN_RA_MUL;
3831 sbi->max_fragment_chunk = DEF_FRAGMENT_SIZE;
3832 sbi->max_fragment_hole = DEF_FRAGMENT_SIZE;
3833 spin_lock_init(&sbi->gc_remaining_trials_lock);
3834 atomic64_set(&sbi->current_atomic_write, 0);
3835
3836 sbi->dir_level = DEF_DIR_LEVEL;
3837 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3838 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3839 sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3840 sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3841 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3842 sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3843 DEF_UMOUNT_DISCARD_TIMEOUT;
3844 clear_sbi_flag(sbi, SBI_NEED_FSCK);
3845
3846 for (i = 0; i < NR_COUNT_TYPE; i++)
3847 atomic_set(&sbi->nr_pages[i], 0);
3848
3849 for (i = 0; i < META; i++)
3850 atomic_set(&sbi->wb_sync_req[i], 0);
3851
3852 INIT_LIST_HEAD(&sbi->s_list);
3853 mutex_init(&sbi->umount_mutex);
3854 init_f2fs_rwsem(&sbi->io_order_lock);
3855 spin_lock_init(&sbi->cp_lock);
3856
3857 sbi->dirty_device = 0;
3858 spin_lock_init(&sbi->dev_lock);
3859
3860 init_f2fs_rwsem(&sbi->sb_lock);
3861 init_f2fs_rwsem(&sbi->pin_sem);
3862 }
3863
3864 static int init_percpu_info(struct f2fs_sb_info *sbi)
3865 {
3866 int err;
3867
3868 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3869 if (err)
3870 return err;
3871
3872 err = percpu_counter_init(&sbi->rf_node_block_count, 0, GFP_KERNEL);
3873 if (err)
3874 goto err_valid_block;
3875
3876 err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3877 GFP_KERNEL);
3878 if (err)
3879 goto err_node_block;
3880 return 0;
3881
3882 err_node_block:
3883 percpu_counter_destroy(&sbi->rf_node_block_count);
3884 err_valid_block:
3885 percpu_counter_destroy(&sbi->alloc_valid_block_count);
3886 return err;
3887 }
3888
3889 #ifdef CONFIG_BLK_DEV_ZONED
3890
3891 struct f2fs_report_zones_args {
3892 struct f2fs_sb_info *sbi;
3893 struct f2fs_dev_info *dev;
3894 };
3895
3896 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3897 void *data)
3898 {
3899 struct f2fs_report_zones_args *rz_args = data;
3900 block_t unusable_blocks = (zone->len - zone->capacity) >>
3901 F2FS_LOG_SECTORS_PER_BLOCK;
3902
3903 if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3904 return 0;
3905
3906 set_bit(idx, rz_args->dev->blkz_seq);
3907 if (!rz_args->sbi->unusable_blocks_per_sec) {
3908 rz_args->sbi->unusable_blocks_per_sec = unusable_blocks;
3909 return 0;
3910 }
3911 if (rz_args->sbi->unusable_blocks_per_sec != unusable_blocks) {
3912 f2fs_err(rz_args->sbi, "F2FS supports single zone capacity\n");
3913 return -EINVAL;
3914 }
3915 return 0;
3916 }
3917
3918 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3919 {
3920 struct block_device *bdev = FDEV(devi).bdev;
3921 sector_t nr_sectors = bdev_nr_sectors(bdev);
3922 struct f2fs_report_zones_args rep_zone_arg;
3923 u64 zone_sectors;
3924 int ret;
3925
3926 if (!f2fs_sb_has_blkzoned(sbi))
3927 return 0;
3928
3929 zone_sectors = bdev_zone_sectors(bdev);
3930 if (!is_power_of_2(zone_sectors)) {
3931 f2fs_err(sbi, "F2FS does not support non power of 2 zone sizes\n");
3932 return -EINVAL;
3933 }
3934
3935 if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3936 SECTOR_TO_BLOCK(zone_sectors))
3937 return -EINVAL;
3938 sbi->blocks_per_blkz = SECTOR_TO_BLOCK(zone_sectors);
3939 FDEV(devi).nr_blkz = div_u64(SECTOR_TO_BLOCK(nr_sectors),
3940 sbi->blocks_per_blkz);
3941 if (nr_sectors & (zone_sectors - 1))
3942 FDEV(devi).nr_blkz++;
3943
3944 FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3945 BITS_TO_LONGS(FDEV(devi).nr_blkz)
3946 * sizeof(unsigned long),
3947 GFP_KERNEL);
3948 if (!FDEV(devi).blkz_seq)
3949 return -ENOMEM;
3950
3951 rep_zone_arg.sbi = sbi;
3952 rep_zone_arg.dev = &FDEV(devi);
3953
3954 ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3955 &rep_zone_arg);
3956 if (ret < 0)
3957 return ret;
3958 return 0;
3959 }
3960 #endif
3961
3962 /*
3963 * Read f2fs raw super block.
3964 * Because we have two copies of super block, so read both of them
3965 * to get the first valid one. If any one of them is broken, we pass
3966 * them recovery flag back to the caller.
3967 */
3968 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3969 struct f2fs_super_block **raw_super,
3970 int *valid_super_block, int *recovery)
3971 {
3972 struct super_block *sb = sbi->sb;
3973 int block;
3974 struct buffer_head *bh;
3975 struct f2fs_super_block *super;
3976 int err = 0;
3977
3978 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3979 if (!super)
3980 return -ENOMEM;
3981
3982 for (block = 0; block < 2; block++) {
3983 bh = sb_bread(sb, block);
3984 if (!bh) {
3985 f2fs_err(sbi, "Unable to read %dth superblock",
3986 block + 1);
3987 err = -EIO;
3988 *recovery = 1;
3989 continue;
3990 }
3991
3992 /* sanity checking of raw super */
3993 err = sanity_check_raw_super(sbi, bh);
3994 if (err) {
3995 f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3996 block + 1);
3997 brelse(bh);
3998 *recovery = 1;
3999 continue;
4000 }
4001
4002 if (!*raw_super) {
4003 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
4004 sizeof(*super));
4005 *valid_super_block = block;
4006 *raw_super = super;
4007 }
4008 brelse(bh);
4009 }
4010
4011 /* No valid superblock */
4012 if (!*raw_super)
4013 kfree(super);
4014 else
4015 err = 0;
4016
4017 return err;
4018 }
4019
4020 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
4021 {
4022 struct buffer_head *bh;
4023 __u32 crc = 0;
4024 int err;
4025
4026 if ((recover && f2fs_readonly(sbi->sb)) ||
4027 f2fs_hw_is_readonly(sbi)) {
4028 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
4029 return -EROFS;
4030 }
4031
4032 /* we should update superblock crc here */
4033 if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
4034 crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
4035 offsetof(struct f2fs_super_block, crc));
4036 F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
4037 }
4038
4039 /* write back-up superblock first */
4040 bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
4041 if (!bh)
4042 return -EIO;
4043 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
4044 brelse(bh);
4045
4046 /* if we are in recovery path, skip writing valid superblock */
4047 if (recover || err)
4048 return err;
4049
4050 /* write current valid superblock */
4051 bh = sb_bread(sbi->sb, sbi->valid_super_block);
4052 if (!bh)
4053 return -EIO;
4054 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
4055 brelse(bh);
4056 return err;
4057 }
4058
4059 static void save_stop_reason(struct f2fs_sb_info *sbi, unsigned char reason)
4060 {
4061 unsigned long flags;
4062
4063 spin_lock_irqsave(&sbi->error_lock, flags);
4064 if (sbi->stop_reason[reason] < GENMASK(BITS_PER_BYTE - 1, 0))
4065 sbi->stop_reason[reason]++;
4066 spin_unlock_irqrestore(&sbi->error_lock, flags);
4067 }
4068
4069 static void f2fs_record_stop_reason(struct f2fs_sb_info *sbi)
4070 {
4071 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
4072 unsigned long flags;
4073 int err;
4074
4075 f2fs_down_write(&sbi->sb_lock);
4076
4077 spin_lock_irqsave(&sbi->error_lock, flags);
4078 if (sbi->error_dirty) {
4079 memcpy(F2FS_RAW_SUPER(sbi)->s_errors, sbi->errors,
4080 MAX_F2FS_ERRORS);
4081 sbi->error_dirty = false;
4082 }
4083 memcpy(raw_super->s_stop_reason, sbi->stop_reason, MAX_STOP_REASON);
4084 spin_unlock_irqrestore(&sbi->error_lock, flags);
4085
4086 err = f2fs_commit_super(sbi, false);
4087
4088 f2fs_up_write(&sbi->sb_lock);
4089 if (err)
4090 f2fs_err(sbi, "f2fs_commit_super fails to record err:%d", err);
4091 }
4092
4093 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag)
4094 {
4095 unsigned long flags;
4096
4097 spin_lock_irqsave(&sbi->error_lock, flags);
4098 if (!test_bit(flag, (unsigned long *)sbi->errors)) {
4099 set_bit(flag, (unsigned long *)sbi->errors);
4100 sbi->error_dirty = true;
4101 }
4102 spin_unlock_irqrestore(&sbi->error_lock, flags);
4103 }
4104
4105 static bool f2fs_update_errors(struct f2fs_sb_info *sbi)
4106 {
4107 unsigned long flags;
4108 bool need_update = false;
4109
4110 spin_lock_irqsave(&sbi->error_lock, flags);
4111 if (sbi->error_dirty) {
4112 memcpy(F2FS_RAW_SUPER(sbi)->s_errors, sbi->errors,
4113 MAX_F2FS_ERRORS);
4114 sbi->error_dirty = false;
4115 need_update = true;
4116 }
4117 spin_unlock_irqrestore(&sbi->error_lock, flags);
4118
4119 return need_update;
4120 }
4121
4122 static void f2fs_record_errors(struct f2fs_sb_info *sbi, unsigned char error)
4123 {
4124 int err;
4125
4126 f2fs_down_write(&sbi->sb_lock);
4127
4128 if (!f2fs_update_errors(sbi))
4129 goto out_unlock;
4130
4131 err = f2fs_commit_super(sbi, false);
4132 if (err)
4133 f2fs_err(sbi, "f2fs_commit_super fails to record errors:%u, err:%d",
4134 error, err);
4135 out_unlock:
4136 f2fs_up_write(&sbi->sb_lock);
4137 }
4138
4139 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error)
4140 {
4141 f2fs_save_errors(sbi, error);
4142 f2fs_record_errors(sbi, error);
4143 }
4144
4145 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error)
4146 {
4147 f2fs_save_errors(sbi, error);
4148
4149 if (!sbi->error_dirty)
4150 return;
4151 if (!test_bit(error, (unsigned long *)sbi->errors))
4152 return;
4153 schedule_work(&sbi->s_error_work);
4154 }
4155
4156 static bool system_going_down(void)
4157 {
4158 return system_state == SYSTEM_HALT || system_state == SYSTEM_POWER_OFF
4159 || system_state == SYSTEM_RESTART;
4160 }
4161
4162 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason,
4163 bool irq_context)
4164 {
4165 struct super_block *sb = sbi->sb;
4166 bool shutdown = reason == STOP_CP_REASON_SHUTDOWN;
4167 bool continue_fs = !shutdown &&
4168 F2FS_OPTION(sbi).errors == MOUNT_ERRORS_CONTINUE;
4169
4170 set_ckpt_flags(sbi, CP_ERROR_FLAG);
4171
4172 if (!f2fs_hw_is_readonly(sbi)) {
4173 save_stop_reason(sbi, reason);
4174
4175 if (irq_context && !shutdown)
4176 schedule_work(&sbi->s_error_work);
4177 else
4178 f2fs_record_stop_reason(sbi);
4179 }
4180
4181 /*
4182 * We force ERRORS_RO behavior when system is rebooting. Otherwise we
4183 * could panic during 'reboot -f' as the underlying device got already
4184 * disabled.
4185 */
4186 if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_PANIC &&
4187 !shutdown && !system_going_down() &&
4188 !is_sbi_flag_set(sbi, SBI_IS_SHUTDOWN))
4189 panic("F2FS-fs (device %s): panic forced after error\n",
4190 sb->s_id);
4191
4192 if (shutdown)
4193 set_sbi_flag(sbi, SBI_IS_SHUTDOWN);
4194
4195 /* continue filesystem operators if errors=continue */
4196 if (continue_fs || f2fs_readonly(sb))
4197 return;
4198
4199 f2fs_warn(sbi, "Remounting filesystem read-only");
4200 /*
4201 * Make sure updated value of ->s_mount_flags will be visible before
4202 * ->s_flags update
4203 */
4204 smp_wmb();
4205 sb->s_flags |= SB_RDONLY;
4206 }
4207
4208 static void f2fs_record_error_work(struct work_struct *work)
4209 {
4210 struct f2fs_sb_info *sbi = container_of(work,
4211 struct f2fs_sb_info, s_error_work);
4212
4213 f2fs_record_stop_reason(sbi);
4214 }
4215
4216 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
4217 {
4218 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
4219 unsigned int max_devices = MAX_DEVICES;
4220 unsigned int logical_blksize;
4221 blk_mode_t mode = sb_open_mode(sbi->sb->s_flags);
4222 int i;
4223
4224 /* Initialize single device information */
4225 if (!RDEV(0).path[0]) {
4226 if (!bdev_is_zoned(sbi->sb->s_bdev))
4227 return 0;
4228 max_devices = 1;
4229 }
4230
4231 /*
4232 * Initialize multiple devices information, or single
4233 * zoned block device information.
4234 */
4235 sbi->devs = f2fs_kzalloc(sbi,
4236 array_size(max_devices,
4237 sizeof(struct f2fs_dev_info)),
4238 GFP_KERNEL);
4239 if (!sbi->devs)
4240 return -ENOMEM;
4241
4242 logical_blksize = bdev_logical_block_size(sbi->sb->s_bdev);
4243 sbi->aligned_blksize = true;
4244
4245 for (i = 0; i < max_devices; i++) {
4246 if (i == 0)
4247 FDEV(0).bdev_handle = sbi->sb->s_bdev_handle;
4248 else if (!RDEV(i).path[0])
4249 break;
4250
4251 if (max_devices > 1) {
4252 /* Multi-device mount */
4253 memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
4254 FDEV(i).total_segments =
4255 le32_to_cpu(RDEV(i).total_segments);
4256 if (i == 0) {
4257 FDEV(i).start_blk = 0;
4258 FDEV(i).end_blk = FDEV(i).start_blk +
4259 (FDEV(i).total_segments <<
4260 sbi->log_blocks_per_seg) - 1 +
4261 le32_to_cpu(raw_super->segment0_blkaddr);
4262 } else {
4263 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
4264 FDEV(i).end_blk = FDEV(i).start_blk +
4265 (FDEV(i).total_segments <<
4266 sbi->log_blocks_per_seg) - 1;
4267 FDEV(i).bdev_handle = bdev_open_by_path(
4268 FDEV(i).path, mode, sbi->sb, NULL);
4269 }
4270 }
4271 if (IS_ERR(FDEV(i).bdev_handle))
4272 return PTR_ERR(FDEV(i).bdev_handle);
4273
4274 FDEV(i).bdev = FDEV(i).bdev_handle->bdev;
4275 /* to release errored devices */
4276 sbi->s_ndevs = i + 1;
4277
4278 if (logical_blksize != bdev_logical_block_size(FDEV(i).bdev))
4279 sbi->aligned_blksize = false;
4280
4281 #ifdef CONFIG_BLK_DEV_ZONED
4282 if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
4283 !f2fs_sb_has_blkzoned(sbi)) {
4284 f2fs_err(sbi, "Zoned block device feature not enabled");
4285 return -EINVAL;
4286 }
4287 if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
4288 if (init_blkz_info(sbi, i)) {
4289 f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
4290 return -EINVAL;
4291 }
4292 if (max_devices == 1)
4293 break;
4294 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
4295 i, FDEV(i).path,
4296 FDEV(i).total_segments,
4297 FDEV(i).start_blk, FDEV(i).end_blk,
4298 bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
4299 "Host-aware" : "Host-managed");
4300 continue;
4301 }
4302 #endif
4303 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
4304 i, FDEV(i).path,
4305 FDEV(i).total_segments,
4306 FDEV(i).start_blk, FDEV(i).end_blk);
4307 }
4308 f2fs_info(sbi,
4309 "IO Block Size: %8ld KB", F2FS_IO_SIZE_KB(sbi));
4310 return 0;
4311 }
4312
4313 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
4314 {
4315 #if IS_ENABLED(CONFIG_UNICODE)
4316 if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
4317 const struct f2fs_sb_encodings *encoding_info;
4318 struct unicode_map *encoding;
4319 __u16 encoding_flags;
4320
4321 encoding_info = f2fs_sb_read_encoding(sbi->raw_super);
4322 if (!encoding_info) {
4323 f2fs_err(sbi,
4324 "Encoding requested by superblock is unknown");
4325 return -EINVAL;
4326 }
4327
4328 encoding_flags = le16_to_cpu(sbi->raw_super->s_encoding_flags);
4329 encoding = utf8_load(encoding_info->version);
4330 if (IS_ERR(encoding)) {
4331 f2fs_err(sbi,
4332 "can't mount with superblock charset: %s-%u.%u.%u "
4333 "not supported by the kernel. flags: 0x%x.",
4334 encoding_info->name,
4335 unicode_major(encoding_info->version),
4336 unicode_minor(encoding_info->version),
4337 unicode_rev(encoding_info->version),
4338 encoding_flags);
4339 return PTR_ERR(encoding);
4340 }
4341 f2fs_info(sbi, "Using encoding defined by superblock: "
4342 "%s-%u.%u.%u with flags 0x%hx", encoding_info->name,
4343 unicode_major(encoding_info->version),
4344 unicode_minor(encoding_info->version),
4345 unicode_rev(encoding_info->version),
4346 encoding_flags);
4347
4348 sbi->sb->s_encoding = encoding;
4349 sbi->sb->s_encoding_flags = encoding_flags;
4350 }
4351 #else
4352 if (f2fs_sb_has_casefold(sbi)) {
4353 f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
4354 return -EINVAL;
4355 }
4356 #endif
4357 return 0;
4358 }
4359
4360 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
4361 {
4362 /* adjust parameters according to the volume size */
4363 if (MAIN_SEGS(sbi) <= SMALL_VOLUME_SEGMENTS) {
4364 if (f2fs_block_unit_discard(sbi))
4365 SM_I(sbi)->dcc_info->discard_granularity =
4366 MIN_DISCARD_GRANULARITY;
4367 if (!f2fs_lfs_mode(sbi))
4368 SM_I(sbi)->ipu_policy = BIT(F2FS_IPU_FORCE) |
4369 BIT(F2FS_IPU_HONOR_OPU_WRITE);
4370 }
4371
4372 sbi->readdir_ra = true;
4373 }
4374
4375 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
4376 {
4377 struct f2fs_sb_info *sbi;
4378 struct f2fs_super_block *raw_super;
4379 struct inode *root;
4380 int err;
4381 bool skip_recovery = false, need_fsck = false;
4382 char *options = NULL;
4383 int recovery, i, valid_super_block;
4384 struct curseg_info *seg_i;
4385 int retry_cnt = 1;
4386 #ifdef CONFIG_QUOTA
4387 bool quota_enabled = false;
4388 #endif
4389
4390 try_onemore:
4391 err = -EINVAL;
4392 raw_super = NULL;
4393 valid_super_block = -1;
4394 recovery = 0;
4395
4396 /* allocate memory for f2fs-specific super block info */
4397 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
4398 if (!sbi)
4399 return -ENOMEM;
4400
4401 sbi->sb = sb;
4402
4403 /* initialize locks within allocated memory */
4404 init_f2fs_rwsem(&sbi->gc_lock);
4405 mutex_init(&sbi->writepages);
4406 init_f2fs_rwsem(&sbi->cp_global_sem);
4407 init_f2fs_rwsem(&sbi->node_write);
4408 init_f2fs_rwsem(&sbi->node_change);
4409 spin_lock_init(&sbi->stat_lock);
4410 init_f2fs_rwsem(&sbi->cp_rwsem);
4411 init_f2fs_rwsem(&sbi->quota_sem);
4412 init_waitqueue_head(&sbi->cp_wait);
4413 spin_lock_init(&sbi->error_lock);
4414
4415 for (i = 0; i < NR_INODE_TYPE; i++) {
4416 INIT_LIST_HEAD(&sbi->inode_list[i]);
4417 spin_lock_init(&sbi->inode_lock[i]);
4418 }
4419 mutex_init(&sbi->flush_lock);
4420
4421 /* Load the checksum driver */
4422 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
4423 if (IS_ERR(sbi->s_chksum_driver)) {
4424 f2fs_err(sbi, "Cannot load crc32 driver.");
4425 err = PTR_ERR(sbi->s_chksum_driver);
4426 sbi->s_chksum_driver = NULL;
4427 goto free_sbi;
4428 }
4429
4430 /* set a block size */
4431 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
4432 f2fs_err(sbi, "unable to set blocksize");
4433 goto free_sbi;
4434 }
4435
4436 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
4437 &recovery);
4438 if (err)
4439 goto free_sbi;
4440
4441 sb->s_fs_info = sbi;
4442 sbi->raw_super = raw_super;
4443
4444 INIT_WORK(&sbi->s_error_work, f2fs_record_error_work);
4445 memcpy(sbi->errors, raw_super->s_errors, MAX_F2FS_ERRORS);
4446 memcpy(sbi->stop_reason, raw_super->s_stop_reason, MAX_STOP_REASON);
4447
4448 /* precompute checksum seed for metadata */
4449 if (f2fs_sb_has_inode_chksum(sbi))
4450 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
4451 sizeof(raw_super->uuid));
4452
4453 default_options(sbi, false);
4454 /* parse mount options */
4455 options = kstrdup((const char *)data, GFP_KERNEL);
4456 if (data && !options) {
4457 err = -ENOMEM;
4458 goto free_sb_buf;
4459 }
4460
4461 err = parse_options(sb, options, false);
4462 if (err)
4463 goto free_options;
4464
4465 sb->s_maxbytes = max_file_blocks(NULL) <<
4466 le32_to_cpu(raw_super->log_blocksize);
4467 sb->s_max_links = F2FS_LINK_MAX;
4468
4469 err = f2fs_setup_casefold(sbi);
4470 if (err)
4471 goto free_options;
4472
4473 #ifdef CONFIG_QUOTA
4474 sb->dq_op = &f2fs_quota_operations;
4475 sb->s_qcop = &f2fs_quotactl_ops;
4476 sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
4477
4478 if (f2fs_sb_has_quota_ino(sbi)) {
4479 for (i = 0; i < MAXQUOTAS; i++) {
4480 if (f2fs_qf_ino(sbi->sb, i))
4481 sbi->nquota_files++;
4482 }
4483 }
4484 #endif
4485
4486 sb->s_op = &f2fs_sops;
4487 #ifdef CONFIG_FS_ENCRYPTION
4488 sb->s_cop = &f2fs_cryptops;
4489 #endif
4490 #ifdef CONFIG_FS_VERITY
4491 sb->s_vop = &f2fs_verityops;
4492 #endif
4493 sb->s_xattr = f2fs_xattr_handlers;
4494 sb->s_export_op = &f2fs_export_ops;
4495 sb->s_magic = F2FS_SUPER_MAGIC;
4496 sb->s_time_gran = 1;
4497 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
4498 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
4499 memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
4500 sb->s_iflags |= SB_I_CGROUPWB;
4501
4502 /* init f2fs-specific super block info */
4503 sbi->valid_super_block = valid_super_block;
4504
4505 /* disallow all the data/node/meta page writes */
4506 set_sbi_flag(sbi, SBI_POR_DOING);
4507
4508 err = f2fs_init_write_merge_io(sbi);
4509 if (err)
4510 goto free_bio_info;
4511
4512 init_sb_info(sbi);
4513
4514 err = f2fs_init_iostat(sbi);
4515 if (err)
4516 goto free_bio_info;
4517
4518 err = init_percpu_info(sbi);
4519 if (err)
4520 goto free_iostat;
4521
4522 if (F2FS_IO_ALIGNED(sbi)) {
4523 sbi->write_io_dummy =
4524 mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
4525 if (!sbi->write_io_dummy) {
4526 err = -ENOMEM;
4527 goto free_percpu;
4528 }
4529 }
4530
4531 /* init per sbi slab cache */
4532 err = f2fs_init_xattr_caches(sbi);
4533 if (err)
4534 goto free_io_dummy;
4535 err = f2fs_init_page_array_cache(sbi);
4536 if (err)
4537 goto free_xattr_cache;
4538
4539 /* get an inode for meta space */
4540 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
4541 if (IS_ERR(sbi->meta_inode)) {
4542 f2fs_err(sbi, "Failed to read F2FS meta data inode");
4543 err = PTR_ERR(sbi->meta_inode);
4544 goto free_page_array_cache;
4545 }
4546
4547 err = f2fs_get_valid_checkpoint(sbi);
4548 if (err) {
4549 f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
4550 goto free_meta_inode;
4551 }
4552
4553 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
4554 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
4555 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
4556 set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4557 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
4558 }
4559
4560 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
4561 set_sbi_flag(sbi, SBI_NEED_FSCK);
4562
4563 /* Initialize device list */
4564 err = f2fs_scan_devices(sbi);
4565 if (err) {
4566 f2fs_err(sbi, "Failed to find devices");
4567 goto free_devices;
4568 }
4569
4570 err = f2fs_init_post_read_wq(sbi);
4571 if (err) {
4572 f2fs_err(sbi, "Failed to initialize post read workqueue");
4573 goto free_devices;
4574 }
4575
4576 sbi->total_valid_node_count =
4577 le32_to_cpu(sbi->ckpt->valid_node_count);
4578 percpu_counter_set(&sbi->total_valid_inode_count,
4579 le32_to_cpu(sbi->ckpt->valid_inode_count));
4580 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
4581 sbi->total_valid_block_count =
4582 le64_to_cpu(sbi->ckpt->valid_block_count);
4583 sbi->last_valid_block_count = sbi->total_valid_block_count;
4584 sbi->reserved_blocks = 0;
4585 sbi->current_reserved_blocks = 0;
4586 limit_reserve_root(sbi);
4587 adjust_unusable_cap_perc(sbi);
4588
4589 f2fs_init_extent_cache_info(sbi);
4590
4591 f2fs_init_ino_entry_info(sbi);
4592
4593 f2fs_init_fsync_node_info(sbi);
4594
4595 /* setup checkpoint request control and start checkpoint issue thread */
4596 f2fs_init_ckpt_req_control(sbi);
4597 if (!f2fs_readonly(sb) && !test_opt(sbi, DISABLE_CHECKPOINT) &&
4598 test_opt(sbi, MERGE_CHECKPOINT)) {
4599 err = f2fs_start_ckpt_thread(sbi);
4600 if (err) {
4601 f2fs_err(sbi,
4602 "Failed to start F2FS issue_checkpoint_thread (%d)",
4603 err);
4604 goto stop_ckpt_thread;
4605 }
4606 }
4607
4608 /* setup f2fs internal modules */
4609 err = f2fs_build_segment_manager(sbi);
4610 if (err) {
4611 f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
4612 err);
4613 goto free_sm;
4614 }
4615 err = f2fs_build_node_manager(sbi);
4616 if (err) {
4617 f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
4618 err);
4619 goto free_nm;
4620 }
4621
4622 err = adjust_reserved_segment(sbi);
4623 if (err)
4624 goto free_nm;
4625
4626 /* For write statistics */
4627 sbi->sectors_written_start = f2fs_get_sectors_written(sbi);
4628
4629 /* Read accumulated write IO statistics if exists */
4630 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
4631 if (__exist_node_summaries(sbi))
4632 sbi->kbytes_written =
4633 le64_to_cpu(seg_i->journal->info.kbytes_written);
4634
4635 f2fs_build_gc_manager(sbi);
4636
4637 err = f2fs_build_stats(sbi);
4638 if (err)
4639 goto free_nm;
4640
4641 /* get an inode for node space */
4642 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
4643 if (IS_ERR(sbi->node_inode)) {
4644 f2fs_err(sbi, "Failed to read node inode");
4645 err = PTR_ERR(sbi->node_inode);
4646 goto free_stats;
4647 }
4648
4649 /* read root inode and dentry */
4650 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
4651 if (IS_ERR(root)) {
4652 f2fs_err(sbi, "Failed to read root inode");
4653 err = PTR_ERR(root);
4654 goto free_node_inode;
4655 }
4656 if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
4657 !root->i_size || !root->i_nlink) {
4658 iput(root);
4659 err = -EINVAL;
4660 goto free_node_inode;
4661 }
4662
4663 sb->s_root = d_make_root(root); /* allocate root dentry */
4664 if (!sb->s_root) {
4665 err = -ENOMEM;
4666 goto free_node_inode;
4667 }
4668
4669 err = f2fs_init_compress_inode(sbi);
4670 if (err)
4671 goto free_root_inode;
4672
4673 err = f2fs_register_sysfs(sbi);
4674 if (err)
4675 goto free_compress_inode;
4676
4677 #ifdef CONFIG_QUOTA
4678 /* Enable quota usage during mount */
4679 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
4680 err = f2fs_enable_quotas(sb);
4681 if (err)
4682 f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
4683 }
4684
4685 quota_enabled = f2fs_recover_quota_begin(sbi);
4686 #endif
4687 /* if there are any orphan inodes, free them */
4688 err = f2fs_recover_orphan_inodes(sbi);
4689 if (err)
4690 goto free_meta;
4691
4692 if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
4693 goto reset_checkpoint;
4694
4695 /* recover fsynced data */
4696 if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
4697 !test_opt(sbi, NORECOVERY)) {
4698 /*
4699 * mount should be failed, when device has readonly mode, and
4700 * previous checkpoint was not done by clean system shutdown.
4701 */
4702 if (f2fs_hw_is_readonly(sbi)) {
4703 if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4704 err = f2fs_recover_fsync_data(sbi, true);
4705 if (err > 0) {
4706 err = -EROFS;
4707 f2fs_err(sbi, "Need to recover fsync data, but "
4708 "write access unavailable, please try "
4709 "mount w/ disable_roll_forward or norecovery");
4710 }
4711 if (err < 0)
4712 goto free_meta;
4713 }
4714 f2fs_info(sbi, "write access unavailable, skipping recovery");
4715 goto reset_checkpoint;
4716 }
4717
4718 if (need_fsck)
4719 set_sbi_flag(sbi, SBI_NEED_FSCK);
4720
4721 if (skip_recovery)
4722 goto reset_checkpoint;
4723
4724 err = f2fs_recover_fsync_data(sbi, false);
4725 if (err < 0) {
4726 if (err != -ENOMEM)
4727 skip_recovery = true;
4728 need_fsck = true;
4729 f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
4730 err);
4731 goto free_meta;
4732 }
4733 } else {
4734 err = f2fs_recover_fsync_data(sbi, true);
4735
4736 if (!f2fs_readonly(sb) && err > 0) {
4737 err = -EINVAL;
4738 f2fs_err(sbi, "Need to recover fsync data");
4739 goto free_meta;
4740 }
4741 }
4742
4743 #ifdef CONFIG_QUOTA
4744 f2fs_recover_quota_end(sbi, quota_enabled);
4745 #endif
4746
4747 /*
4748 * If the f2fs is not readonly and fsync data recovery succeeds,
4749 * check zoned block devices' write pointer consistency.
4750 */
4751 if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
4752 err = f2fs_check_write_pointer(sbi);
4753 if (err)
4754 goto free_meta;
4755 }
4756
4757 reset_checkpoint:
4758 f2fs_init_inmem_curseg(sbi);
4759
4760 /* f2fs_recover_fsync_data() cleared this already */
4761 clear_sbi_flag(sbi, SBI_POR_DOING);
4762
4763 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
4764 err = f2fs_disable_checkpoint(sbi);
4765 if (err)
4766 goto sync_free_meta;
4767 } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
4768 f2fs_enable_checkpoint(sbi);
4769 }
4770
4771 /*
4772 * If filesystem is not mounted as read-only then
4773 * do start the gc_thread.
4774 */
4775 if ((F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF ||
4776 test_opt(sbi, GC_MERGE)) && !f2fs_readonly(sb)) {
4777 /* After POR, we can run background GC thread.*/
4778 err = f2fs_start_gc_thread(sbi);
4779 if (err)
4780 goto sync_free_meta;
4781 }
4782 kvfree(options);
4783
4784 /* recover broken superblock */
4785 if (recovery) {
4786 err = f2fs_commit_super(sbi, true);
4787 f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
4788 sbi->valid_super_block ? 1 : 2, err);
4789 }
4790
4791 f2fs_join_shrinker(sbi);
4792
4793 f2fs_tuning_parameters(sbi);
4794
4795 f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
4796 cur_cp_version(F2FS_CKPT(sbi)));
4797 f2fs_update_time(sbi, CP_TIME);
4798 f2fs_update_time(sbi, REQ_TIME);
4799 clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4800 return 0;
4801
4802 sync_free_meta:
4803 /* safe to flush all the data */
4804 sync_filesystem(sbi->sb);
4805 retry_cnt = 0;
4806
4807 free_meta:
4808 #ifdef CONFIG_QUOTA
4809 f2fs_truncate_quota_inode_pages(sb);
4810 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
4811 f2fs_quota_off_umount(sbi->sb);
4812 #endif
4813 /*
4814 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
4815 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
4816 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
4817 * falls into an infinite loop in f2fs_sync_meta_pages().
4818 */
4819 truncate_inode_pages_final(META_MAPPING(sbi));
4820 /* evict some inodes being cached by GC */
4821 evict_inodes(sb);
4822 f2fs_unregister_sysfs(sbi);
4823 free_compress_inode:
4824 f2fs_destroy_compress_inode(sbi);
4825 free_root_inode:
4826 dput(sb->s_root);
4827 sb->s_root = NULL;
4828 free_node_inode:
4829 f2fs_release_ino_entry(sbi, true);
4830 truncate_inode_pages_final(NODE_MAPPING(sbi));
4831 iput(sbi->node_inode);
4832 sbi->node_inode = NULL;
4833 free_stats:
4834 f2fs_destroy_stats(sbi);
4835 free_nm:
4836 /* stop discard thread before destroying node manager */
4837 f2fs_stop_discard_thread(sbi);
4838 f2fs_destroy_node_manager(sbi);
4839 free_sm:
4840 f2fs_destroy_segment_manager(sbi);
4841 stop_ckpt_thread:
4842 f2fs_stop_ckpt_thread(sbi);
4843 /* flush s_error_work before sbi destroy */
4844 flush_work(&sbi->s_error_work);
4845 f2fs_destroy_post_read_wq(sbi);
4846 free_devices:
4847 destroy_device_list(sbi);
4848 kvfree(sbi->ckpt);
4849 free_meta_inode:
4850 make_bad_inode(sbi->meta_inode);
4851 iput(sbi->meta_inode);
4852 sbi->meta_inode = NULL;
4853 free_page_array_cache:
4854 f2fs_destroy_page_array_cache(sbi);
4855 free_xattr_cache:
4856 f2fs_destroy_xattr_caches(sbi);
4857 free_io_dummy:
4858 mempool_destroy(sbi->write_io_dummy);
4859 free_percpu:
4860 destroy_percpu_info(sbi);
4861 free_iostat:
4862 f2fs_destroy_iostat(sbi);
4863 free_bio_info:
4864 for (i = 0; i < NR_PAGE_TYPE; i++)
4865 kvfree(sbi->write_io[i]);
4866
4867 #if IS_ENABLED(CONFIG_UNICODE)
4868 utf8_unload(sb->s_encoding);
4869 sb->s_encoding = NULL;
4870 #endif
4871 free_options:
4872 #ifdef CONFIG_QUOTA
4873 for (i = 0; i < MAXQUOTAS; i++)
4874 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
4875 #endif
4876 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
4877 kvfree(options);
4878 free_sb_buf:
4879 kfree(raw_super);
4880 free_sbi:
4881 if (sbi->s_chksum_driver)
4882 crypto_free_shash(sbi->s_chksum_driver);
4883 kfree(sbi);
4884
4885 /* give only one another chance */
4886 if (retry_cnt > 0 && skip_recovery) {
4887 retry_cnt--;
4888 shrink_dcache_sb(sb);
4889 goto try_onemore;
4890 }
4891 return err;
4892 }
4893
4894 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
4895 const char *dev_name, void *data)
4896 {
4897 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
4898 }
4899
4900 static void kill_f2fs_super(struct super_block *sb)
4901 {
4902 struct f2fs_sb_info *sbi = F2FS_SB(sb);
4903
4904 if (sb->s_root) {
4905 set_sbi_flag(sbi, SBI_IS_CLOSE);
4906 f2fs_stop_gc_thread(sbi);
4907 f2fs_stop_discard_thread(sbi);
4908
4909 #ifdef CONFIG_F2FS_FS_COMPRESSION
4910 /*
4911 * latter evict_inode() can bypass checking and invalidating
4912 * compress inode cache.
4913 */
4914 if (test_opt(sbi, COMPRESS_CACHE))
4915 truncate_inode_pages_final(COMPRESS_MAPPING(sbi));
4916 #endif
4917
4918 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
4919 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4920 struct cp_control cpc = {
4921 .reason = CP_UMOUNT,
4922 };
4923 stat_inc_cp_call_count(sbi, TOTAL_CALL);
4924 f2fs_write_checkpoint(sbi, &cpc);
4925 }
4926
4927 if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
4928 sb->s_flags &= ~SB_RDONLY;
4929 }
4930 kill_block_super(sb);
4931 /* Release block devices last, after fscrypt_destroy_keyring(). */
4932 if (sbi) {
4933 destroy_device_list(sbi);
4934 kfree(sbi);
4935 sb->s_fs_info = NULL;
4936 }
4937 }
4938
4939 static struct file_system_type f2fs_fs_type = {
4940 .owner = THIS_MODULE,
4941 .name = "f2fs",
4942 .mount = f2fs_mount,
4943 .kill_sb = kill_f2fs_super,
4944 .fs_flags = FS_REQUIRES_DEV | FS_ALLOW_IDMAP,
4945 };
4946 MODULE_ALIAS_FS("f2fs");
4947
4948 static int __init init_inodecache(void)
4949 {
4950 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
4951 sizeof(struct f2fs_inode_info), 0,
4952 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
4953 return f2fs_inode_cachep ? 0 : -ENOMEM;
4954 }
4955
4956 static void destroy_inodecache(void)
4957 {
4958 /*
4959 * Make sure all delayed rcu free inodes are flushed before we
4960 * destroy cache.
4961 */
4962 rcu_barrier();
4963 kmem_cache_destroy(f2fs_inode_cachep);
4964 }
4965
4966 static int __init init_f2fs_fs(void)
4967 {
4968 int err;
4969
4970 if (PAGE_SIZE != F2FS_BLKSIZE) {
4971 printk("F2FS not supported on PAGE_SIZE(%lu) != BLOCK_SIZE(%lu)\n",
4972 PAGE_SIZE, F2FS_BLKSIZE);
4973 return -EINVAL;
4974 }
4975
4976 err = init_inodecache();
4977 if (err)
4978 goto fail;
4979 err = f2fs_create_node_manager_caches();
4980 if (err)
4981 goto free_inodecache;
4982 err = f2fs_create_segment_manager_caches();
4983 if (err)
4984 goto free_node_manager_caches;
4985 err = f2fs_create_checkpoint_caches();
4986 if (err)
4987 goto free_segment_manager_caches;
4988 err = f2fs_create_recovery_cache();
4989 if (err)
4990 goto free_checkpoint_caches;
4991 err = f2fs_create_extent_cache();
4992 if (err)
4993 goto free_recovery_cache;
4994 err = f2fs_create_garbage_collection_cache();
4995 if (err)
4996 goto free_extent_cache;
4997 err = f2fs_init_sysfs();
4998 if (err)
4999 goto free_garbage_collection_cache;
5000 err = f2fs_init_shrinker();
5001 if (err)
5002 goto free_sysfs;
5003 err = register_filesystem(&f2fs_fs_type);
5004 if (err)
5005 goto free_shrinker;
5006 f2fs_create_root_stats();
5007 err = f2fs_init_post_read_processing();
5008 if (err)
5009 goto free_root_stats;
5010 err = f2fs_init_iostat_processing();
5011 if (err)
5012 goto free_post_read;
5013 err = f2fs_init_bio_entry_cache();
5014 if (err)
5015 goto free_iostat;
5016 err = f2fs_init_bioset();
5017 if (err)
5018 goto free_bio_entry_cache;
5019 err = f2fs_init_compress_mempool();
5020 if (err)
5021 goto free_bioset;
5022 err = f2fs_init_compress_cache();
5023 if (err)
5024 goto free_compress_mempool;
5025 err = f2fs_create_casefold_cache();
5026 if (err)
5027 goto free_compress_cache;
5028 return 0;
5029 free_compress_cache:
5030 f2fs_destroy_compress_cache();
5031 free_compress_mempool:
5032 f2fs_destroy_compress_mempool();
5033 free_bioset:
5034 f2fs_destroy_bioset();
5035 free_bio_entry_cache:
5036 f2fs_destroy_bio_entry_cache();
5037 free_iostat:
5038 f2fs_destroy_iostat_processing();
5039 free_post_read:
5040 f2fs_destroy_post_read_processing();
5041 free_root_stats:
5042 f2fs_destroy_root_stats();
5043 unregister_filesystem(&f2fs_fs_type);
5044 free_shrinker:
5045 f2fs_exit_shrinker();
5046 free_sysfs:
5047 f2fs_exit_sysfs();
5048 free_garbage_collection_cache:
5049 f2fs_destroy_garbage_collection_cache();
5050 free_extent_cache:
5051 f2fs_destroy_extent_cache();
5052 free_recovery_cache:
5053 f2fs_destroy_recovery_cache();
5054 free_checkpoint_caches:
5055 f2fs_destroy_checkpoint_caches();
5056 free_segment_manager_caches:
5057 f2fs_destroy_segment_manager_caches();
5058 free_node_manager_caches:
5059 f2fs_destroy_node_manager_caches();
5060 free_inodecache:
5061 destroy_inodecache();
5062 fail:
5063 return err;
5064 }
5065
5066 static void __exit exit_f2fs_fs(void)
5067 {
5068 f2fs_destroy_casefold_cache();
5069 f2fs_destroy_compress_cache();
5070 f2fs_destroy_compress_mempool();
5071 f2fs_destroy_bioset();
5072 f2fs_destroy_bio_entry_cache();
5073 f2fs_destroy_iostat_processing();
5074 f2fs_destroy_post_read_processing();
5075 f2fs_destroy_root_stats();
5076 unregister_filesystem(&f2fs_fs_type);
5077 f2fs_exit_shrinker();
5078 f2fs_exit_sysfs();
5079 f2fs_destroy_garbage_collection_cache();
5080 f2fs_destroy_extent_cache();
5081 f2fs_destroy_recovery_cache();
5082 f2fs_destroy_checkpoint_caches();
5083 f2fs_destroy_segment_manager_caches();
5084 f2fs_destroy_node_manager_caches();
5085 destroy_inodecache();
5086 }
5087
5088 module_init(init_f2fs_fs)
5089 module_exit(exit_f2fs_fs)
5090
5091 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
5092 MODULE_DESCRIPTION("Flash Friendly File System");
5093 MODULE_LICENSE("GPL");
5094 MODULE_SOFTDEP("pre: crc32");
5095
A mirror of Linus' kernel repository