1 // SPDX-License-Identifier: GPL-2.0
5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6 * http://www.samsung.com/
8 #include <linux/module.h>
9 #include <linux/init.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>
39 #define CREATE_TRACE_POINTS
40 #include <trace/events/f2fs.h>
42 static struct kmem_cache *f2fs_inode_cachep;
44 #ifdef CONFIG_F2FS_FAULT_INJECTION
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",
67 void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
70 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
73 atomic_set(&ffi->inject_ops, 0);
74 ffi->inject_rate = rate;
78 ffi->inject_type = type;
81 memset(ffi, 0, sizeof(struct f2fs_fault_info));
85 /* f2fs-wide shrinker description */
86 static struct shrinker *f2fs_shrinker_info;
88 static int __init f2fs_init_shrinker(void)
90 f2fs_shrinker_info = shrinker_alloc(0, "f2fs-shrinker");
91 if (!f2fs_shrinker_info)
94 f2fs_shrinker_info->count_objects = f2fs_shrink_count;
95 f2fs_shrinker_info->scan_objects = f2fs_shrink_scan;
97 shrinker_register(f2fs_shrinker_info);
102 static void f2fs_exit_shrinker(void)
104 shrinker_free(f2fs_shrinker_info);
109 Opt_disable_roll_forward,
120 Opt_disable_ext_identify,
123 Opt_inline_xattr_size,
161 Opt_test_dummy_encryption,
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,
181 Opt_age_extent_cache,
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"},
194 {Opt_user_xattr, "user_xattr"},
195 {Opt_nouser_xattr, "nouser_xattr"},
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"},
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"},
266 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...)
268 struct va_format vaf;
274 level = printk_get_level(fmt);
275 vaf.fmt = printk_skip_level(fmt);
277 printk("%c%cF2FS-fs (%s): %pV\n",
278 KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
283 #if IS_ENABLED(CONFIG_UNICODE)
284 static const struct f2fs_sb_encodings {
287 unsigned int version;
288 } f2fs_sb_encoding_map[] = {
289 {F2FS_ENC_UTF8_12_1, "utf8", UNICODE_AGE(12, 1, 0)},
292 static const struct f2fs_sb_encodings *
293 f2fs_sb_read_encoding(const struct f2fs_super_block *sb)
295 __u16 magic = le16_to_cpu(sb->s_encoding);
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];
305 struct kmem_cache *f2fs_cf_name_slab;
306 static int __init f2fs_create_casefold_cache(void)
308 f2fs_cf_name_slab = f2fs_kmem_cache_create("f2fs_casefolded_name",
310 return f2fs_cf_name_slab ? 0 : -ENOMEM;
313 static void f2fs_destroy_casefold_cache(void)
315 kmem_cache_destroy(f2fs_cf_name_slab);
318 static int __init f2fs_create_casefold_cache(void) { return 0; }
319 static void f2fs_destroy_casefold_cache(void) { }
322 static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
324 block_t limit = min((sbi->user_block_count >> 3),
325 sbi->user_block_count - sbi->reserved_blocks);
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);
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));
346 static inline int adjust_reserved_segment(struct f2fs_sb_info *sbi)
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;
353 if (!F2FS_IO_ALIGNED(sbi))
356 /* average valid block count in section in worst case */
357 avg_vblocks = sec_blks / F2FS_IO_SIZE(sbi);
360 * we need enough free space when migrating one section in worst case
362 wanted_reserved_segments = (F2FS_IO_SIZE(sbi) / avg_vblocks) *
363 reserved_segments(sbi);
364 wanted_reserved_segments -= reserved_segments(sbi);
366 avail_user_block_count = sbi->user_block_count -
367 sbi->current_reserved_blocks -
368 F2FS_OPTION(sbi).root_reserved_blocks;
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);
378 SM_I(sbi)->additional_reserved_segments = wanted_reserved_segments;
380 f2fs_info(sbi, "IO align feature needs additional reserved segment: %u",
381 wanted_reserved_segments);
386 static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
388 if (!F2FS_OPTION(sbi).unusable_cap_perc)
391 if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
392 F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
394 F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
395 F2FS_OPTION(sbi).unusable_cap_perc;
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);
402 static void init_once(void *foo)
404 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
406 inode_init_once(&fi->vfs_inode);
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,
415 struct f2fs_sb_info *sbi = F2FS_SB(sb);
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");
423 if (f2fs_sb_has_quota_ino(sbi)) {
424 f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
428 qname = match_strdup(args);
430 f2fs_err(sbi, "Not enough memory for storing quotafile name");
433 if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
434 if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
437 f2fs_err(sbi, "%s quota file already specified",
441 if (strchr(qname, '/')) {
442 f2fs_err(sbi, "quotafile must be on filesystem root");
445 F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
453 static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
455 struct f2fs_sb_info *sbi = F2FS_SB(sb);
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");
461 kfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
462 F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
466 static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
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.
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.");
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);
484 if (test_opt(sbi, GRPQUOTA) &&
485 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
486 clear_opt(sbi, GRPQUOTA);
488 if (test_opt(sbi, PRJQUOTA) &&
489 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
490 clear_opt(sbi, PRJQUOTA);
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");
498 if (!F2FS_OPTION(sbi).s_jquota_fmt) {
499 f2fs_err(sbi, "journaled quota format not specified");
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;
512 static int f2fs_set_test_dummy_encryption(struct super_block *sb,
514 const substring_t *arg,
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 : "",
522 struct fscrypt_dummy_policy *policy =
523 &F2FS_OPTION(sbi).dummy_enc_policy;
526 if (!IS_ENABLED(CONFIG_FS_ENCRYPTION)) {
527 f2fs_warn(sbi, "test_dummy_encryption option not supported");
531 if (!f2fs_sb_has_encrypt(sbi)) {
532 f2fs_err(sbi, "Encrypt feature is off");
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.
542 if (is_remount && !fscrypt_is_dummy_policy_set(policy)) {
543 f2fs_warn(sbi, "Can't set test_dummy_encryption on remount");
547 err = fscrypt_parse_test_dummy_encryption(¶m, policy);
551 "Can't change test_dummy_encryption on remount");
552 else if (err == -EINVAL)
553 f2fs_warn(sbi, "Value of option \"%s\" is unrecognized",
556 f2fs_warn(sbi, "Error processing option \"%s\" [%d]",
560 f2fs_warn(sbi, "Test dummy encryption mode enabled");
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)
568 unsigned char (*ext)[F2FS_EXTENSION_LEN];
573 ext = F2FS_OPTION(sbi).extensions;
574 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
576 ext = F2FS_OPTION(sbi).noextensions;
577 ext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
580 for (i = 0; i < ext_cnt; i++) {
581 if (!strcasecmp(new_ext, ext[i]))
589 * 1. The same extension name cannot not appear in both compress and non-compress extension
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.
595 static int f2fs_test_compress_extension(struct f2fs_sb_info *sbi)
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;
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;
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");
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",
625 #ifdef CONFIG_F2FS_FS_LZ4
626 static int f2fs_set_lz4hc_level(struct f2fs_sb_info *sbi, const char *str)
628 #ifdef CONFIG_F2FS_FS_LZ4HC
631 if (strlen(str) == 3) {
632 F2FS_OPTION(sbi).compress_level = 0;
639 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
642 if (kstrtouint(str + 1, 10, &level))
645 if (!f2fs_is_compress_level_valid(COMPRESS_LZ4, level)) {
646 f2fs_info(sbi, "invalid lz4hc compress level: %d", level);
650 F2FS_OPTION(sbi).compress_level = level;
653 if (strlen(str) == 3) {
654 F2FS_OPTION(sbi).compress_level = 0;
657 f2fs_info(sbi, "kernel doesn't support lz4hc compression");
663 #ifdef CONFIG_F2FS_FS_ZSTD
664 static int f2fs_set_zstd_level(struct f2fs_sb_info *sbi, const char *str)
669 if (strlen(str) == len) {
670 F2FS_OPTION(sbi).compress_level = F2FS_ZSTD_DEFAULT_CLEVEL;
677 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
680 if (kstrtouint(str + 1, 10, &level))
683 if (!f2fs_is_compress_level_valid(COMPRESS_ZSTD, level)) {
684 f2fs_info(sbi, "invalid zstd compress level: %d", level);
688 F2FS_OPTION(sbi).compress_level = level;
694 static int parse_options(struct super_block *sb, char *options, bool is_remount)
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;
712 while ((p = strsep(&options, ",")) != NULL) {
718 * Initialize args struct so we know whether arg was
719 * found; some options take optional arguments.
721 args[0].to = args[0].from = NULL;
722 token = match_token(p, f2fs_tokens, args);
725 case Opt_gc_background:
726 name = match_strdup(&args[0]);
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;
742 case Opt_disable_roll_forward:
743 set_opt(sbi, DISABLE_ROLL_FORWARD);
746 /* this option mounts f2fs with ro */
747 set_opt(sbi, NORECOVERY);
748 if (!f2fs_readonly(sb))
752 if (!f2fs_hw_support_discard(sbi)) {
753 f2fs_warn(sbi, "device does not support discard");
756 set_opt(sbi, DISCARD);
759 if (f2fs_hw_should_discard(sbi)) {
760 f2fs_warn(sbi, "discard is required for zoned block devices");
763 clear_opt(sbi, DISCARD);
766 set_opt(sbi, NOHEAP);
769 clear_opt(sbi, NOHEAP);
771 #ifdef CONFIG_F2FS_FS_XATTR
773 set_opt(sbi, XATTR_USER);
775 case Opt_nouser_xattr:
776 clear_opt(sbi, XATTR_USER);
778 case Opt_inline_xattr:
779 set_opt(sbi, INLINE_XATTR);
781 case Opt_noinline_xattr:
782 clear_opt(sbi, INLINE_XATTR);
784 case Opt_inline_xattr_size:
785 if (args->from && match_int(args, &arg))
787 set_opt(sbi, INLINE_XATTR_SIZE);
788 F2FS_OPTION(sbi).inline_xattr_size = arg;
792 f2fs_info(sbi, "user_xattr options not supported");
794 case Opt_nouser_xattr:
795 f2fs_info(sbi, "nouser_xattr options not supported");
797 case Opt_inline_xattr:
798 f2fs_info(sbi, "inline_xattr options not supported");
800 case Opt_noinline_xattr:
801 f2fs_info(sbi, "noinline_xattr options not supported");
804 #ifdef CONFIG_F2FS_FS_POSIX_ACL
806 set_opt(sbi, POSIX_ACL);
809 clear_opt(sbi, POSIX_ACL);
813 f2fs_info(sbi, "acl options not supported");
816 f2fs_info(sbi, "noacl options not supported");
819 case Opt_active_logs:
820 if (args->from && match_int(args, &arg))
822 if (arg != 2 && arg != 4 &&
823 arg != NR_CURSEG_PERSIST_TYPE)
825 F2FS_OPTION(sbi).active_logs = arg;
827 case Opt_disable_ext_identify:
828 set_opt(sbi, DISABLE_EXT_IDENTIFY);
830 case Opt_inline_data:
831 set_opt(sbi, INLINE_DATA);
833 case Opt_inline_dentry:
834 set_opt(sbi, INLINE_DENTRY);
836 case Opt_noinline_dentry:
837 clear_opt(sbi, INLINE_DENTRY);
839 case Opt_flush_merge:
840 set_opt(sbi, FLUSH_MERGE);
842 case Opt_noflush_merge:
843 clear_opt(sbi, FLUSH_MERGE);
846 set_opt(sbi, NOBARRIER);
849 clear_opt(sbi, NOBARRIER);
852 set_opt(sbi, FASTBOOT);
854 case Opt_extent_cache:
855 set_opt(sbi, READ_EXTENT_CACHE);
857 case Opt_noextent_cache:
858 clear_opt(sbi, READ_EXTENT_CACHE);
860 case Opt_noinline_data:
861 clear_opt(sbi, INLINE_DATA);
864 set_opt(sbi, DATA_FLUSH);
866 case Opt_reserve_root:
867 if (args->from && match_int(args, &arg))
869 if (test_opt(sbi, RESERVE_ROOT)) {
870 f2fs_info(sbi, "Preserve previous reserve_root=%u",
871 F2FS_OPTION(sbi).root_reserved_blocks);
873 F2FS_OPTION(sbi).root_reserved_blocks = arg;
874 set_opt(sbi, RESERVE_ROOT);
878 if (args->from && match_int(args, &arg))
880 uid = make_kuid(current_user_ns(), arg);
881 if (!uid_valid(uid)) {
882 f2fs_err(sbi, "Invalid uid value %d", arg);
885 F2FS_OPTION(sbi).s_resuid = uid;
888 if (args->from && match_int(args, &arg))
890 gid = make_kgid(current_user_ns(), arg);
891 if (!gid_valid(gid)) {
892 f2fs_err(sbi, "Invalid gid value %d", arg);
895 F2FS_OPTION(sbi).s_resgid = gid;
898 name = match_strdup(&args[0]);
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;
916 case Opt_io_size_bits:
917 if (args->from && match_int(args, &arg))
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);
924 F2FS_OPTION(sbi).write_io_size_bits = arg;
926 #ifdef CONFIG_F2FS_FAULT_INJECTION
927 case Opt_fault_injection:
928 if (args->from && match_int(args, &arg))
930 f2fs_build_fault_attr(sbi, arg, F2FS_ALL_FAULT_TYPE);
931 set_opt(sbi, FAULT_INJECTION);
935 if (args->from && match_int(args, &arg))
937 f2fs_build_fault_attr(sbi, 0, arg);
938 set_opt(sbi, FAULT_INJECTION);
941 case Opt_fault_injection:
942 f2fs_info(sbi, "fault_injection options not supported");
946 f2fs_info(sbi, "fault_type options not supported");
950 sb->s_flags |= SB_LAZYTIME;
953 sb->s_flags &= ~SB_LAZYTIME;
958 set_opt(sbi, USRQUOTA);
961 set_opt(sbi, GRPQUOTA);
964 set_opt(sbi, PRJQUOTA);
967 ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
972 ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
977 ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
981 case Opt_offusrjquota:
982 ret = f2fs_clear_qf_name(sb, USRQUOTA);
986 case Opt_offgrpjquota:
987 ret = f2fs_clear_qf_name(sb, GRPQUOTA);
991 case Opt_offprjjquota:
992 ret = f2fs_clear_qf_name(sb, PRJQUOTA);
996 case Opt_jqfmt_vfsold:
997 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
999 case Opt_jqfmt_vfsv0:
1000 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
1002 case Opt_jqfmt_vfsv1:
1003 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
1006 clear_opt(sbi, QUOTA);
1007 clear_opt(sbi, USRQUOTA);
1008 clear_opt(sbi, GRPQUOTA);
1009 clear_opt(sbi, PRJQUOTA);
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:
1026 f2fs_info(sbi, "quota operations not supported");
1030 name = match_strdup(&args[0]);
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;
1045 name = match_strdup(&args[0]);
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;
1061 case Opt_test_dummy_encryption:
1062 ret = f2fs_set_test_dummy_encryption(sb, p, &args[0],
1067 case Opt_inlinecrypt:
1068 #ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
1069 sb->s_flags |= SB_INLINECRYPT;
1071 f2fs_info(sbi, "inline encryption not supported");
1074 case Opt_checkpoint_disable_cap_perc:
1075 if (args->from && match_int(args, &arg))
1077 if (arg < 0 || arg > 100)
1079 F2FS_OPTION(sbi).unusable_cap_perc = arg;
1080 set_opt(sbi, DISABLE_CHECKPOINT);
1082 case Opt_checkpoint_disable_cap:
1083 if (args->from && match_int(args, &arg))
1085 F2FS_OPTION(sbi).unusable_cap = arg;
1086 set_opt(sbi, DISABLE_CHECKPOINT);
1088 case Opt_checkpoint_disable:
1089 set_opt(sbi, DISABLE_CHECKPOINT);
1091 case Opt_checkpoint_enable:
1092 clear_opt(sbi, DISABLE_CHECKPOINT);
1094 case Opt_checkpoint_merge:
1095 set_opt(sbi, MERGE_CHECKPOINT);
1097 case Opt_nocheckpoint_merge:
1098 clear_opt(sbi, MERGE_CHECKPOINT);
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");
1106 name = match_strdup(&args[0]);
1109 if (!strcmp(name, "lzo")) {
1110 #ifdef CONFIG_F2FS_FS_LZO
1111 F2FS_OPTION(sbi).compress_level = 0;
1112 F2FS_OPTION(sbi).compress_algorithm =
1115 f2fs_info(sbi, "kernel doesn't support lzo compression");
1117 } else if (!strncmp(name, "lz4", 3)) {
1118 #ifdef CONFIG_F2FS_FS_LZ4
1119 ret = f2fs_set_lz4hc_level(sbi, name);
1124 F2FS_OPTION(sbi).compress_algorithm =
1127 f2fs_info(sbi, "kernel doesn't support lz4 compression");
1129 } else if (!strncmp(name, "zstd", 4)) {
1130 #ifdef CONFIG_F2FS_FS_ZSTD
1131 ret = f2fs_set_zstd_level(sbi, name);
1136 F2FS_OPTION(sbi).compress_algorithm =
1139 f2fs_info(sbi, "kernel doesn't support zstd compression");
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 =
1147 f2fs_info(sbi, "kernel doesn't support lzorle compression");
1155 case Opt_compress_log_size:
1156 if (!f2fs_sb_has_compression(sbi)) {
1157 f2fs_info(sbi, "Image doesn't support compression");
1160 if (args->from && match_int(args, &arg))
1162 if (arg < MIN_COMPRESS_LOG_SIZE ||
1163 arg > MAX_COMPRESS_LOG_SIZE) {
1165 "Compress cluster log size is out of range");
1168 F2FS_OPTION(sbi).compress_log_size = arg;
1170 case Opt_compress_extension:
1171 if (!f2fs_sb_has_compression(sbi)) {
1172 f2fs_info(sbi, "Image doesn't support compression");
1175 name = match_strdup(&args[0]);
1179 ext = F2FS_OPTION(sbi).extensions;
1180 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
1182 if (strlen(name) >= F2FS_EXTENSION_LEN ||
1183 ext_cnt >= COMPRESS_EXT_NUM) {
1185 "invalid extension length/number");
1190 if (is_compress_extension_exist(sbi, name, true)) {
1195 strcpy(ext[ext_cnt], name);
1196 F2FS_OPTION(sbi).compress_ext_cnt++;
1199 case Opt_nocompress_extension:
1200 if (!f2fs_sb_has_compression(sbi)) {
1201 f2fs_info(sbi, "Image doesn't support compression");
1204 name = match_strdup(&args[0]);
1208 noext = F2FS_OPTION(sbi).noextensions;
1209 noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
1211 if (strlen(name) >= F2FS_EXTENSION_LEN ||
1212 noext_cnt >= COMPRESS_EXT_NUM) {
1214 "invalid extension length/number");
1219 if (is_compress_extension_exist(sbi, name, false)) {
1224 strcpy(noext[noext_cnt], name);
1225 F2FS_OPTION(sbi).nocompress_ext_cnt++;
1228 case Opt_compress_chksum:
1229 if (!f2fs_sb_has_compression(sbi)) {
1230 f2fs_info(sbi, "Image doesn't support compression");
1233 F2FS_OPTION(sbi).compress_chksum = true;
1235 case Opt_compress_mode:
1236 if (!f2fs_sb_has_compression(sbi)) {
1237 f2fs_info(sbi, "Image doesn't support compression");
1240 name = match_strdup(&args[0]);
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;
1253 case Opt_compress_cache:
1254 if (!f2fs_sb_has_compression(sbi)) {
1255 f2fs_info(sbi, "Image doesn't support compression");
1258 set_opt(sbi, COMPRESS_CACHE);
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");
1275 set_opt(sbi, GC_MERGE);
1277 case Opt_nogc_merge:
1278 clear_opt(sbi, GC_MERGE);
1280 case Opt_discard_unit:
1281 name = match_strdup(&args[0]);
1284 if (!strcmp(name, "block")) {
1285 F2FS_OPTION(sbi).discard_unit =
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;
1299 case Opt_memory_mode:
1300 name = match_strdup(&args[0]);
1303 if (!strcmp(name, "normal")) {
1304 F2FS_OPTION(sbi).memory_mode =
1306 } else if (!strcmp(name, "low")) {
1307 F2FS_OPTION(sbi).memory_mode =
1315 case Opt_age_extent_cache:
1316 set_opt(sbi, AGE_EXTENT_CACHE);
1319 name = match_strdup(&args[0]);
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 =
1338 f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
1345 if (f2fs_check_quota_options(sbi))
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");
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");
1357 #if !IS_ENABLED(CONFIG_UNICODE)
1358 if (f2fs_sb_has_casefold(sbi)) {
1360 "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
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.
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;
1378 if (F2FS_OPTION(sbi).fs_mode != FS_MODE_LFS) {
1379 f2fs_info(sbi, "Only lfs mode is allowed with zoned block device feature");
1383 f2fs_err(sbi, "Zoned block device support is not enabled");
1388 #ifdef CONFIG_F2FS_FS_COMPRESSION
1389 if (f2fs_test_compress_extension(sbi)) {
1390 f2fs_err(sbi, "invalid compress or nocompress extension");
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));
1401 if (test_opt(sbi, INLINE_XATTR_SIZE)) {
1402 int min_size, max_size;
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");
1409 if (!test_opt(sbi, INLINE_XATTR)) {
1410 f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
1414 min_size = MIN_INLINE_XATTR_SIZE;
1415 max_size = MAX_INLINE_XATTR_SIZE;
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);
1425 if (test_opt(sbi, ATGC) && f2fs_lfs_mode(sbi)) {
1426 f2fs_err(sbi, "LFS is not compatible with ATGC");
1430 if (f2fs_is_readonly(sbi) && test_opt(sbi, FLUSH_MERGE)) {
1431 f2fs_err(sbi, "FLUSH_MERGE not compatible with readonly mode");
1435 if (f2fs_sb_has_readonly(sbi) && !f2fs_readonly(sbi->sb)) {
1436 f2fs_err(sbi, "Allow to mount readonly mode only");
1442 static struct inode *f2fs_alloc_inode(struct super_block *sb)
1444 struct f2fs_inode_info *fi;
1446 if (time_to_inject(F2FS_SB(sb), FAULT_SLAB_ALLOC))
1449 fi = alloc_inode_sb(sb, f2fs_inode_cachep, GFP_F2FS_ZERO);
1453 init_once((void *) fi);
1455 /* Initialize f2fs-specific inode info */
1456 atomic_set(&fi->dirty_pages, 0);
1457 atomic_set(&fi->i_compr_blocks, 0);
1458 init_f2fs_rwsem(&fi->i_sem);
1459 spin_lock_init(&fi->i_size_lock);
1460 INIT_LIST_HEAD(&fi->dirty_list);
1461 INIT_LIST_HEAD(&fi->gdirty_list);
1462 init_f2fs_rwsem(&fi->i_gc_rwsem[READ]);
1463 init_f2fs_rwsem(&fi->i_gc_rwsem[WRITE]);
1464 init_f2fs_rwsem(&fi->i_xattr_sem);
1466 /* Will be used by directory only */
1467 fi->i_dir_level = F2FS_SB(sb)->dir_level;
1469 return &fi->vfs_inode;
1472 static int f2fs_drop_inode(struct inode *inode)
1474 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1478 * during filesystem shutdown, if checkpoint is disabled,
1479 * drop useless meta/node dirty pages.
1481 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1482 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1483 inode->i_ino == F2FS_META_INO(sbi)) {
1484 trace_f2fs_drop_inode(inode, 1);
1490 * This is to avoid a deadlock condition like below.
1491 * writeback_single_inode(inode)
1492 * - f2fs_write_data_page
1493 * - f2fs_gc -> iput -> evict
1494 * - inode_wait_for_writeback(inode)
1496 if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
1497 if (!inode->i_nlink && !is_bad_inode(inode)) {
1498 /* to avoid evict_inode call simultaneously */
1499 atomic_inc(&inode->i_count);
1500 spin_unlock(&inode->i_lock);
1502 /* should remain fi->extent_tree for writepage */
1503 f2fs_destroy_extent_node(inode);
1505 sb_start_intwrite(inode->i_sb);
1506 f2fs_i_size_write(inode, 0);
1508 f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1509 inode, NULL, 0, DATA);
1510 truncate_inode_pages_final(inode->i_mapping);
1512 if (F2FS_HAS_BLOCKS(inode))
1513 f2fs_truncate(inode);
1515 sb_end_intwrite(inode->i_sb);
1517 spin_lock(&inode->i_lock);
1518 atomic_dec(&inode->i_count);
1520 trace_f2fs_drop_inode(inode, 0);
1523 ret = generic_drop_inode(inode);
1525 ret = fscrypt_drop_inode(inode);
1526 trace_f2fs_drop_inode(inode, ret);
1530 int f2fs_inode_dirtied(struct inode *inode, bool sync)
1532 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1535 spin_lock(&sbi->inode_lock[DIRTY_META]);
1536 if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1539 set_inode_flag(inode, FI_DIRTY_INODE);
1540 stat_inc_dirty_inode(sbi, DIRTY_META);
1542 if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1543 list_add_tail(&F2FS_I(inode)->gdirty_list,
1544 &sbi->inode_list[DIRTY_META]);
1545 inc_page_count(sbi, F2FS_DIRTY_IMETA);
1547 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1551 void f2fs_inode_synced(struct inode *inode)
1553 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1555 spin_lock(&sbi->inode_lock[DIRTY_META]);
1556 if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1557 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1560 if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1561 list_del_init(&F2FS_I(inode)->gdirty_list);
1562 dec_page_count(sbi, F2FS_DIRTY_IMETA);
1564 clear_inode_flag(inode, FI_DIRTY_INODE);
1565 clear_inode_flag(inode, FI_AUTO_RECOVER);
1566 stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1567 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1571 * f2fs_dirty_inode() is called from __mark_inode_dirty()
1573 * We should call set_dirty_inode to write the dirty inode through write_inode.
1575 static void f2fs_dirty_inode(struct inode *inode, int flags)
1577 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1579 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1580 inode->i_ino == F2FS_META_INO(sbi))
1583 if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1584 clear_inode_flag(inode, FI_AUTO_RECOVER);
1586 f2fs_inode_dirtied(inode, false);
1589 static void f2fs_free_inode(struct inode *inode)
1591 fscrypt_free_inode(inode);
1592 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1595 static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1597 percpu_counter_destroy(&sbi->total_valid_inode_count);
1598 percpu_counter_destroy(&sbi->rf_node_block_count);
1599 percpu_counter_destroy(&sbi->alloc_valid_block_count);
1602 static void destroy_device_list(struct f2fs_sb_info *sbi)
1606 for (i = 0; i < sbi->s_ndevs; i++) {
1608 bdev_release(FDEV(i).bdev_handle);
1609 #ifdef CONFIG_BLK_DEV_ZONED
1610 kvfree(FDEV(i).blkz_seq);
1616 static void f2fs_put_super(struct super_block *sb)
1618 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1623 /* unregister procfs/sysfs entries in advance to avoid race case */
1624 f2fs_unregister_sysfs(sbi);
1626 f2fs_quota_off_umount(sb);
1628 /* prevent remaining shrinker jobs */
1629 mutex_lock(&sbi->umount_mutex);
1632 * flush all issued checkpoints and stop checkpoint issue thread.
1633 * after then, all checkpoints should be done by each process context.
1635 f2fs_stop_ckpt_thread(sbi);
1638 * We don't need to do checkpoint when superblock is clean.
1639 * But, the previous checkpoint was not done by umount, it needs to do
1640 * clean checkpoint again.
1642 if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1643 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1644 struct cp_control cpc = {
1645 .reason = CP_UMOUNT,
1647 stat_inc_cp_call_count(sbi, TOTAL_CALL);
1648 err = f2fs_write_checkpoint(sbi, &cpc);
1651 /* be sure to wait for any on-going discard commands */
1652 done = f2fs_issue_discard_timeout(sbi);
1653 if (f2fs_realtime_discard_enable(sbi) && !sbi->discard_blks && done) {
1654 struct cp_control cpc = {
1655 .reason = CP_UMOUNT | CP_TRIMMED,
1657 stat_inc_cp_call_count(sbi, TOTAL_CALL);
1658 err = f2fs_write_checkpoint(sbi, &cpc);
1662 * normally superblock is clean, so we need to release this.
1663 * In addition, EIO will skip do checkpoint, we need this as well.
1665 f2fs_release_ino_entry(sbi, true);
1667 f2fs_leave_shrinker(sbi);
1668 mutex_unlock(&sbi->umount_mutex);
1670 /* our cp_error case, we can wait for any writeback page */
1671 f2fs_flush_merged_writes(sbi);
1673 f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1675 if (err || f2fs_cp_error(sbi)) {
1676 truncate_inode_pages_final(NODE_MAPPING(sbi));
1677 truncate_inode_pages_final(META_MAPPING(sbi));
1680 for (i = 0; i < NR_COUNT_TYPE; i++) {
1681 if (!get_pages(sbi, i))
1683 f2fs_err(sbi, "detect filesystem reference count leak during "
1684 "umount, type: %d, count: %lld", i, get_pages(sbi, i));
1685 f2fs_bug_on(sbi, 1);
1688 f2fs_bug_on(sbi, sbi->fsync_node_num);
1690 f2fs_destroy_compress_inode(sbi);
1692 iput(sbi->node_inode);
1693 sbi->node_inode = NULL;
1695 iput(sbi->meta_inode);
1696 sbi->meta_inode = NULL;
1699 * iput() can update stat information, if f2fs_write_checkpoint()
1700 * above failed with error.
1702 f2fs_destroy_stats(sbi);
1704 /* destroy f2fs internal modules */
1705 f2fs_destroy_node_manager(sbi);
1706 f2fs_destroy_segment_manager(sbi);
1708 /* flush s_error_work before sbi destroy */
1709 flush_work(&sbi->s_error_work);
1711 f2fs_destroy_post_read_wq(sbi);
1715 sb->s_fs_info = NULL;
1716 if (sbi->s_chksum_driver)
1717 crypto_free_shash(sbi->s_chksum_driver);
1718 kfree(sbi->raw_super);
1720 destroy_device_list(sbi);
1721 f2fs_destroy_page_array_cache(sbi);
1722 f2fs_destroy_xattr_caches(sbi);
1723 mempool_destroy(sbi->write_io_dummy);
1725 for (i = 0; i < MAXQUOTAS; i++)
1726 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1728 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
1729 destroy_percpu_info(sbi);
1730 f2fs_destroy_iostat(sbi);
1731 for (i = 0; i < NR_PAGE_TYPE; i++)
1732 kvfree(sbi->write_io[i]);
1733 #if IS_ENABLED(CONFIG_UNICODE)
1734 utf8_unload(sb->s_encoding);
1739 int f2fs_sync_fs(struct super_block *sb, int sync)
1741 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1744 if (unlikely(f2fs_cp_error(sbi)))
1746 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1749 trace_f2fs_sync_fs(sb, sync);
1751 if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1755 stat_inc_cp_call_count(sbi, TOTAL_CALL);
1756 err = f2fs_issue_checkpoint(sbi);
1762 static int f2fs_freeze(struct super_block *sb)
1764 if (f2fs_readonly(sb))
1767 /* IO error happened before */
1768 if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1771 /* must be clean, since sync_filesystem() was already called */
1772 if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1775 /* Let's flush checkpoints and stop the thread. */
1776 f2fs_flush_ckpt_thread(F2FS_SB(sb));
1778 /* to avoid deadlock on f2fs_evict_inode->SB_FREEZE_FS */
1779 set_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1783 static int f2fs_unfreeze(struct super_block *sb)
1785 clear_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1790 static int f2fs_statfs_project(struct super_block *sb,
1791 kprojid_t projid, struct kstatfs *buf)
1794 struct dquot *dquot;
1798 qid = make_kqid_projid(projid);
1799 dquot = dqget(sb, qid);
1801 return PTR_ERR(dquot);
1802 spin_lock(&dquot->dq_dqb_lock);
1804 limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1805 dquot->dq_dqb.dqb_bhardlimit);
1807 limit >>= sb->s_blocksize_bits;
1809 if (limit && buf->f_blocks > limit) {
1810 curblock = (dquot->dq_dqb.dqb_curspace +
1811 dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
1812 buf->f_blocks = limit;
1813 buf->f_bfree = buf->f_bavail =
1814 (buf->f_blocks > curblock) ?
1815 (buf->f_blocks - curblock) : 0;
1818 limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1819 dquot->dq_dqb.dqb_ihardlimit);
1821 if (limit && buf->f_files > limit) {
1822 buf->f_files = limit;
1824 (buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1825 (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1828 spin_unlock(&dquot->dq_dqb_lock);
1834 static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1836 struct super_block *sb = dentry->d_sb;
1837 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1838 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
1839 block_t total_count, user_block_count, start_count;
1840 u64 avail_node_count;
1841 unsigned int total_valid_node_count;
1843 total_count = le64_to_cpu(sbi->raw_super->block_count);
1844 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
1845 buf->f_type = F2FS_SUPER_MAGIC;
1846 buf->f_bsize = sbi->blocksize;
1848 buf->f_blocks = total_count - start_count;
1850 spin_lock(&sbi->stat_lock);
1852 user_block_count = sbi->user_block_count;
1853 total_valid_node_count = valid_node_count(sbi);
1854 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
1855 buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1856 sbi->current_reserved_blocks;
1858 if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1861 buf->f_bfree -= sbi->unusable_block_count;
1862 spin_unlock(&sbi->stat_lock);
1864 if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1865 buf->f_bavail = buf->f_bfree -
1866 F2FS_OPTION(sbi).root_reserved_blocks;
1870 if (avail_node_count > user_block_count) {
1871 buf->f_files = user_block_count;
1872 buf->f_ffree = buf->f_bavail;
1874 buf->f_files = avail_node_count;
1875 buf->f_ffree = min(avail_node_count - total_valid_node_count,
1879 buf->f_namelen = F2FS_NAME_LEN;
1880 buf->f_fsid = u64_to_fsid(id);
1883 if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1884 sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1885 f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1891 static inline void f2fs_show_quota_options(struct seq_file *seq,
1892 struct super_block *sb)
1895 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1897 if (F2FS_OPTION(sbi).s_jquota_fmt) {
1900 switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1911 seq_printf(seq, ",jqfmt=%s", fmtname);
1914 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1915 seq_show_option(seq, "usrjquota",
1916 F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1918 if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1919 seq_show_option(seq, "grpjquota",
1920 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1922 if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1923 seq_show_option(seq, "prjjquota",
1924 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1928 #ifdef CONFIG_F2FS_FS_COMPRESSION
1929 static inline void f2fs_show_compress_options(struct seq_file *seq,
1930 struct super_block *sb)
1932 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1936 if (!f2fs_sb_has_compression(sbi))
1939 switch (F2FS_OPTION(sbi).compress_algorithm) {
1949 case COMPRESS_LZORLE:
1950 algtype = "lzo-rle";
1953 seq_printf(seq, ",compress_algorithm=%s", algtype);
1955 if (F2FS_OPTION(sbi).compress_level)
1956 seq_printf(seq, ":%d", F2FS_OPTION(sbi).compress_level);
1958 seq_printf(seq, ",compress_log_size=%u",
1959 F2FS_OPTION(sbi).compress_log_size);
1961 for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1962 seq_printf(seq, ",compress_extension=%s",
1963 F2FS_OPTION(sbi).extensions[i]);
1966 for (i = 0; i < F2FS_OPTION(sbi).nocompress_ext_cnt; i++) {
1967 seq_printf(seq, ",nocompress_extension=%s",
1968 F2FS_OPTION(sbi).noextensions[i]);
1971 if (F2FS_OPTION(sbi).compress_chksum)
1972 seq_puts(seq, ",compress_chksum");
1974 if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_FS)
1975 seq_printf(seq, ",compress_mode=%s", "fs");
1976 else if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_USER)
1977 seq_printf(seq, ",compress_mode=%s", "user");
1979 if (test_opt(sbi, COMPRESS_CACHE))
1980 seq_puts(seq, ",compress_cache");
1984 static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1986 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1988 if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1989 seq_printf(seq, ",background_gc=%s", "sync");
1990 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1991 seq_printf(seq, ",background_gc=%s", "on");
1992 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1993 seq_printf(seq, ",background_gc=%s", "off");
1995 if (test_opt(sbi, GC_MERGE))
1996 seq_puts(seq, ",gc_merge");
1998 seq_puts(seq, ",nogc_merge");
2000 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
2001 seq_puts(seq, ",disable_roll_forward");
2002 if (test_opt(sbi, NORECOVERY))
2003 seq_puts(seq, ",norecovery");
2004 if (test_opt(sbi, DISCARD)) {
2005 seq_puts(seq, ",discard");
2006 if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK)
2007 seq_printf(seq, ",discard_unit=%s", "block");
2008 else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SEGMENT)
2009 seq_printf(seq, ",discard_unit=%s", "segment");
2010 else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SECTION)
2011 seq_printf(seq, ",discard_unit=%s", "section");
2013 seq_puts(seq, ",nodiscard");
2015 if (test_opt(sbi, NOHEAP))
2016 seq_puts(seq, ",no_heap");
2018 seq_puts(seq, ",heap");
2019 #ifdef CONFIG_F2FS_FS_XATTR
2020 if (test_opt(sbi, XATTR_USER))
2021 seq_puts(seq, ",user_xattr");
2023 seq_puts(seq, ",nouser_xattr");
2024 if (test_opt(sbi, INLINE_XATTR))
2025 seq_puts(seq, ",inline_xattr");
2027 seq_puts(seq, ",noinline_xattr");
2028 if (test_opt(sbi, INLINE_XATTR_SIZE))
2029 seq_printf(seq, ",inline_xattr_size=%u",
2030 F2FS_OPTION(sbi).inline_xattr_size);
2032 #ifdef CONFIG_F2FS_FS_POSIX_ACL
2033 if (test_opt(sbi, POSIX_ACL))
2034 seq_puts(seq, ",acl");
2036 seq_puts(seq, ",noacl");
2038 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
2039 seq_puts(seq, ",disable_ext_identify");
2040 if (test_opt(sbi, INLINE_DATA))
2041 seq_puts(seq, ",inline_data");
2043 seq_puts(seq, ",noinline_data");
2044 if (test_opt(sbi, INLINE_DENTRY))
2045 seq_puts(seq, ",inline_dentry");
2047 seq_puts(seq, ",noinline_dentry");
2048 if (test_opt(sbi, FLUSH_MERGE))
2049 seq_puts(seq, ",flush_merge");
2051 seq_puts(seq, ",noflush_merge");
2052 if (test_opt(sbi, NOBARRIER))
2053 seq_puts(seq, ",nobarrier");
2055 seq_puts(seq, ",barrier");
2056 if (test_opt(sbi, FASTBOOT))
2057 seq_puts(seq, ",fastboot");
2058 if (test_opt(sbi, READ_EXTENT_CACHE))
2059 seq_puts(seq, ",extent_cache");
2061 seq_puts(seq, ",noextent_cache");
2062 if (test_opt(sbi, AGE_EXTENT_CACHE))
2063 seq_puts(seq, ",age_extent_cache");
2064 if (test_opt(sbi, DATA_FLUSH))
2065 seq_puts(seq, ",data_flush");
2067 seq_puts(seq, ",mode=");
2068 if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
2069 seq_puts(seq, "adaptive");
2070 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
2071 seq_puts(seq, "lfs");
2072 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG)
2073 seq_puts(seq, "fragment:segment");
2074 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK)
2075 seq_puts(seq, "fragment:block");
2076 seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
2077 if (test_opt(sbi, RESERVE_ROOT))
2078 seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
2079 F2FS_OPTION(sbi).root_reserved_blocks,
2080 from_kuid_munged(&init_user_ns,
2081 F2FS_OPTION(sbi).s_resuid),
2082 from_kgid_munged(&init_user_ns,
2083 F2FS_OPTION(sbi).s_resgid));
2084 if (F2FS_IO_SIZE_BITS(sbi))
2085 seq_printf(seq, ",io_bits=%u",
2086 F2FS_OPTION(sbi).write_io_size_bits);
2087 #ifdef CONFIG_F2FS_FAULT_INJECTION
2088 if (test_opt(sbi, FAULT_INJECTION)) {
2089 seq_printf(seq, ",fault_injection=%u",
2090 F2FS_OPTION(sbi).fault_info.inject_rate);
2091 seq_printf(seq, ",fault_type=%u",
2092 F2FS_OPTION(sbi).fault_info.inject_type);
2096 if (test_opt(sbi, QUOTA))
2097 seq_puts(seq, ",quota");
2098 if (test_opt(sbi, USRQUOTA))
2099 seq_puts(seq, ",usrquota");
2100 if (test_opt(sbi, GRPQUOTA))
2101 seq_puts(seq, ",grpquota");
2102 if (test_opt(sbi, PRJQUOTA))
2103 seq_puts(seq, ",prjquota");
2105 f2fs_show_quota_options(seq, sbi->sb);
2107 fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
2109 if (sbi->sb->s_flags & SB_INLINECRYPT)
2110 seq_puts(seq, ",inlinecrypt");
2112 if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
2113 seq_printf(seq, ",alloc_mode=%s", "default");
2114 else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
2115 seq_printf(seq, ",alloc_mode=%s", "reuse");
2117 if (test_opt(sbi, DISABLE_CHECKPOINT))
2118 seq_printf(seq, ",checkpoint=disable:%u",
2119 F2FS_OPTION(sbi).unusable_cap);
2120 if (test_opt(sbi, MERGE_CHECKPOINT))
2121 seq_puts(seq, ",checkpoint_merge");
2123 seq_puts(seq, ",nocheckpoint_merge");
2124 if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
2125 seq_printf(seq, ",fsync_mode=%s", "posix");
2126 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
2127 seq_printf(seq, ",fsync_mode=%s", "strict");
2128 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
2129 seq_printf(seq, ",fsync_mode=%s", "nobarrier");
2131 #ifdef CONFIG_F2FS_FS_COMPRESSION
2132 f2fs_show_compress_options(seq, sbi->sb);
2135 if (test_opt(sbi, ATGC))
2136 seq_puts(seq, ",atgc");
2138 if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_NORMAL)
2139 seq_printf(seq, ",memory=%s", "normal");
2140 else if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW)
2141 seq_printf(seq, ",memory=%s", "low");
2143 if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_READONLY)
2144 seq_printf(seq, ",errors=%s", "remount-ro");
2145 else if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_CONTINUE)
2146 seq_printf(seq, ",errors=%s", "continue");
2147 else if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_PANIC)
2148 seq_printf(seq, ",errors=%s", "panic");
2153 static void default_options(struct f2fs_sb_info *sbi, bool remount)
2155 /* init some FS parameters */
2157 set_opt(sbi, READ_EXTENT_CACHE);
2158 clear_opt(sbi, DISABLE_CHECKPOINT);
2160 if (f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi))
2161 set_opt(sbi, DISCARD);
2163 if (f2fs_sb_has_blkzoned(sbi))
2164 F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_SECTION;
2166 F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_BLOCK;
2169 if (f2fs_sb_has_readonly(sbi))
2170 F2FS_OPTION(sbi).active_logs = NR_CURSEG_RO_TYPE;
2172 F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
2174 F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
2175 if (le32_to_cpu(F2FS_RAW_SUPER(sbi)->segment_count_main) <=
2176 SMALL_VOLUME_SEGMENTS)
2177 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
2179 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
2180 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
2181 F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
2182 F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
2183 if (f2fs_sb_has_compression(sbi)) {
2184 F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
2185 F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
2186 F2FS_OPTION(sbi).compress_ext_cnt = 0;
2187 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
2189 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
2190 F2FS_OPTION(sbi).memory_mode = MEMORY_MODE_NORMAL;
2191 F2FS_OPTION(sbi).errors = MOUNT_ERRORS_CONTINUE;
2193 sbi->sb->s_flags &= ~SB_INLINECRYPT;
2195 set_opt(sbi, INLINE_XATTR);
2196 set_opt(sbi, INLINE_DATA);
2197 set_opt(sbi, INLINE_DENTRY);
2198 set_opt(sbi, NOHEAP);
2199 set_opt(sbi, MERGE_CHECKPOINT);
2200 F2FS_OPTION(sbi).unusable_cap = 0;
2201 sbi->sb->s_flags |= SB_LAZYTIME;
2202 if (!f2fs_is_readonly(sbi))
2203 set_opt(sbi, FLUSH_MERGE);
2204 if (f2fs_sb_has_blkzoned(sbi))
2205 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
2207 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
2209 #ifdef CONFIG_F2FS_FS_XATTR
2210 set_opt(sbi, XATTR_USER);
2212 #ifdef CONFIG_F2FS_FS_POSIX_ACL
2213 set_opt(sbi, POSIX_ACL);
2216 f2fs_build_fault_attr(sbi, 0, 0);
2220 static int f2fs_enable_quotas(struct super_block *sb);
2223 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
2225 unsigned int s_flags = sbi->sb->s_flags;
2226 struct cp_control cpc;
2227 unsigned int gc_mode = sbi->gc_mode;
2232 if (s_flags & SB_RDONLY) {
2233 f2fs_err(sbi, "checkpoint=disable on readonly fs");
2236 sbi->sb->s_flags |= SB_ACTIVE;
2238 /* check if we need more GC first */
2239 unusable = f2fs_get_unusable_blocks(sbi);
2240 if (!f2fs_disable_cp_again(sbi, unusable))
2243 f2fs_update_time(sbi, DISABLE_TIME);
2245 sbi->gc_mode = GC_URGENT_HIGH;
2247 while (!f2fs_time_over(sbi, DISABLE_TIME)) {
2248 struct f2fs_gc_control gc_control = {
2249 .victim_segno = NULL_SEGNO,
2250 .init_gc_type = FG_GC,
2251 .should_migrate_blocks = false,
2252 .err_gc_skipped = true,
2253 .nr_free_secs = 1 };
2255 f2fs_down_write(&sbi->gc_lock);
2256 stat_inc_gc_call_count(sbi, FOREGROUND);
2257 err = f2fs_gc(sbi, &gc_control);
2258 if (err == -ENODATA) {
2262 if (err && err != -EAGAIN)
2266 ret = sync_filesystem(sbi->sb);
2268 err = ret ? ret : err;
2272 unusable = f2fs_get_unusable_blocks(sbi);
2273 if (f2fs_disable_cp_again(sbi, unusable)) {
2279 f2fs_down_write(&sbi->gc_lock);
2280 cpc.reason = CP_PAUSE;
2281 set_sbi_flag(sbi, SBI_CP_DISABLED);
2282 stat_inc_cp_call_count(sbi, TOTAL_CALL);
2283 err = f2fs_write_checkpoint(sbi, &cpc);
2287 spin_lock(&sbi->stat_lock);
2288 sbi->unusable_block_count = unusable;
2289 spin_unlock(&sbi->stat_lock);
2292 f2fs_up_write(&sbi->gc_lock);
2294 sbi->gc_mode = gc_mode;
2295 sbi->sb->s_flags = s_flags; /* Restore SB_RDONLY status */
2299 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
2301 int retry = DEFAULT_RETRY_IO_COUNT;
2303 /* we should flush all the data to keep data consistency */
2305 sync_inodes_sb(sbi->sb);
2306 f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2307 } while (get_pages(sbi, F2FS_DIRTY_DATA) && retry--);
2309 if (unlikely(retry < 0))
2310 f2fs_warn(sbi, "checkpoint=enable has some unwritten data.");
2312 f2fs_down_write(&sbi->gc_lock);
2313 f2fs_dirty_to_prefree(sbi);
2315 clear_sbi_flag(sbi, SBI_CP_DISABLED);
2316 set_sbi_flag(sbi, SBI_IS_DIRTY);
2317 f2fs_up_write(&sbi->gc_lock);
2319 f2fs_sync_fs(sbi->sb, 1);
2321 /* Let's ensure there's no pending checkpoint anymore */
2322 f2fs_flush_ckpt_thread(sbi);
2325 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
2327 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2328 struct f2fs_mount_info org_mount_opt;
2329 unsigned long old_sb_flags;
2331 bool need_restart_gc = false, need_stop_gc = false;
2332 bool need_restart_flush = false, need_stop_flush = false;
2333 bool need_restart_discard = false, need_stop_discard = false;
2334 bool need_enable_checkpoint = false, need_disable_checkpoint = false;
2335 bool no_read_extent_cache = !test_opt(sbi, READ_EXTENT_CACHE);
2336 bool no_age_extent_cache = !test_opt(sbi, AGE_EXTENT_CACHE);
2337 bool enable_checkpoint = !test_opt(sbi, DISABLE_CHECKPOINT);
2338 bool no_io_align = !F2FS_IO_ALIGNED(sbi);
2339 bool no_atgc = !test_opt(sbi, ATGC);
2340 bool no_discard = !test_opt(sbi, DISCARD);
2341 bool no_compress_cache = !test_opt(sbi, COMPRESS_CACHE);
2342 bool block_unit_discard = f2fs_block_unit_discard(sbi);
2348 * Save the old mount options in case we
2349 * need to restore them.
2351 org_mount_opt = sbi->mount_opt;
2352 old_sb_flags = sb->s_flags;
2355 org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
2356 for (i = 0; i < MAXQUOTAS; i++) {
2357 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2358 org_mount_opt.s_qf_names[i] =
2359 kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
2361 if (!org_mount_opt.s_qf_names[i]) {
2362 for (j = 0; j < i; j++)
2363 kfree(org_mount_opt.s_qf_names[j]);
2367 org_mount_opt.s_qf_names[i] = NULL;
2372 /* recover superblocks we couldn't write due to previous RO mount */
2373 if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
2374 err = f2fs_commit_super(sbi, false);
2375 f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
2378 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2381 default_options(sbi, true);
2383 /* parse mount options */
2384 err = parse_options(sb, data, true);
2388 /* flush outstanding errors before changing fs state */
2389 flush_work(&sbi->s_error_work);
2392 * Previous and new state of filesystem is RO,
2393 * so skip checking GC and FLUSH_MERGE conditions.
2395 if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
2398 if (f2fs_dev_is_readonly(sbi) && !(*flags & SB_RDONLY)) {
2404 if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
2405 err = dquot_suspend(sb, -1);
2408 } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
2409 /* dquot_resume needs RW */
2410 sb->s_flags &= ~SB_RDONLY;
2411 if (sb_any_quota_suspended(sb)) {
2412 dquot_resume(sb, -1);
2413 } else if (f2fs_sb_has_quota_ino(sbi)) {
2414 err = f2fs_enable_quotas(sb);
2420 if (f2fs_lfs_mode(sbi) && !IS_F2FS_IPU_DISABLE(sbi)) {
2422 f2fs_warn(sbi, "LFS is not compatible with IPU");
2426 /* disallow enable atgc dynamically */
2427 if (no_atgc == !!test_opt(sbi, ATGC)) {
2429 f2fs_warn(sbi, "switch atgc option is not allowed");
2433 /* disallow enable/disable extent_cache dynamically */
2434 if (no_read_extent_cache == !!test_opt(sbi, READ_EXTENT_CACHE)) {
2436 f2fs_warn(sbi, "switch extent_cache option is not allowed");
2439 /* disallow enable/disable age extent_cache dynamically */
2440 if (no_age_extent_cache == !!test_opt(sbi, AGE_EXTENT_CACHE)) {
2442 f2fs_warn(sbi, "switch age_extent_cache option is not allowed");
2446 if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
2448 f2fs_warn(sbi, "switch io_bits option is not allowed");
2452 if (no_compress_cache == !!test_opt(sbi, COMPRESS_CACHE)) {
2454 f2fs_warn(sbi, "switch compress_cache option is not allowed");
2458 if (block_unit_discard != f2fs_block_unit_discard(sbi)) {
2460 f2fs_warn(sbi, "switch discard_unit option is not allowed");
2464 if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
2466 f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
2471 * We stop the GC thread if FS is mounted as RO
2472 * or if background_gc = off is passed in mount
2473 * option. Also sync the filesystem.
2475 if ((*flags & SB_RDONLY) ||
2476 (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF &&
2477 !test_opt(sbi, GC_MERGE))) {
2478 if (sbi->gc_thread) {
2479 f2fs_stop_gc_thread(sbi);
2480 need_restart_gc = true;
2482 } else if (!sbi->gc_thread) {
2483 err = f2fs_start_gc_thread(sbi);
2486 need_stop_gc = true;
2489 if (*flags & SB_RDONLY) {
2492 set_sbi_flag(sbi, SBI_IS_DIRTY);
2493 set_sbi_flag(sbi, SBI_IS_CLOSE);
2494 f2fs_sync_fs(sb, 1);
2495 clear_sbi_flag(sbi, SBI_IS_CLOSE);
2499 * We stop issue flush thread if FS is mounted as RO
2500 * or if flush_merge is not passed in mount option.
2502 if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2503 clear_opt(sbi, FLUSH_MERGE);
2504 f2fs_destroy_flush_cmd_control(sbi, false);
2505 need_restart_flush = true;
2507 err = f2fs_create_flush_cmd_control(sbi);
2510 need_stop_flush = true;
2513 if (no_discard == !!test_opt(sbi, DISCARD)) {
2514 if (test_opt(sbi, DISCARD)) {
2515 err = f2fs_start_discard_thread(sbi);
2518 need_stop_discard = true;
2520 f2fs_stop_discard_thread(sbi);
2521 f2fs_issue_discard_timeout(sbi);
2522 need_restart_discard = true;
2526 if (enable_checkpoint == !!test_opt(sbi, DISABLE_CHECKPOINT)) {
2527 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
2528 err = f2fs_disable_checkpoint(sbi);
2530 goto restore_discard;
2531 need_enable_checkpoint = true;
2533 f2fs_enable_checkpoint(sbi);
2534 need_disable_checkpoint = true;
2539 * Place this routine at the end, since a new checkpoint would be
2540 * triggered while remount and we need to take care of it before
2541 * returning from remount.
2543 if ((*flags & SB_RDONLY) || test_opt(sbi, DISABLE_CHECKPOINT) ||
2544 !test_opt(sbi, MERGE_CHECKPOINT)) {
2545 f2fs_stop_ckpt_thread(sbi);
2547 /* Flush if the prevous checkpoint, if exists. */
2548 f2fs_flush_ckpt_thread(sbi);
2550 err = f2fs_start_ckpt_thread(sbi);
2553 "Failed to start F2FS issue_checkpoint_thread (%d)",
2555 goto restore_checkpoint;
2561 /* Release old quota file names */
2562 for (i = 0; i < MAXQUOTAS; i++)
2563 kfree(org_mount_opt.s_qf_names[i]);
2565 /* Update the POSIXACL Flag */
2566 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
2567 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
2569 limit_reserve_root(sbi);
2570 adjust_unusable_cap_perc(sbi);
2571 *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
2574 if (need_enable_checkpoint) {
2575 f2fs_enable_checkpoint(sbi);
2576 } else if (need_disable_checkpoint) {
2577 if (f2fs_disable_checkpoint(sbi))
2578 f2fs_warn(sbi, "checkpoint has not been disabled");
2581 if (need_restart_discard) {
2582 if (f2fs_start_discard_thread(sbi))
2583 f2fs_warn(sbi, "discard has been stopped");
2584 } else if (need_stop_discard) {
2585 f2fs_stop_discard_thread(sbi);
2588 if (need_restart_flush) {
2589 if (f2fs_create_flush_cmd_control(sbi))
2590 f2fs_warn(sbi, "background flush thread has stopped");
2591 } else if (need_stop_flush) {
2592 clear_opt(sbi, FLUSH_MERGE);
2593 f2fs_destroy_flush_cmd_control(sbi, false);
2596 if (need_restart_gc) {
2597 if (f2fs_start_gc_thread(sbi))
2598 f2fs_warn(sbi, "background gc thread has stopped");
2599 } else if (need_stop_gc) {
2600 f2fs_stop_gc_thread(sbi);
2604 F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
2605 for (i = 0; i < MAXQUOTAS; i++) {
2606 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
2607 F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
2610 sbi->mount_opt = org_mount_opt;
2611 sb->s_flags = old_sb_flags;
2616 static bool f2fs_need_recovery(struct f2fs_sb_info *sbi)
2618 /* need to recovery orphan */
2619 if (is_set_ckpt_flags(sbi, CP_ORPHAN_PRESENT_FLAG))
2621 /* need to recovery data */
2622 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
2624 if (test_opt(sbi, NORECOVERY))
2626 return !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG);
2629 static bool f2fs_recover_quota_begin(struct f2fs_sb_info *sbi)
2631 bool readonly = f2fs_readonly(sbi->sb);
2633 if (!f2fs_need_recovery(sbi))
2636 /* it doesn't need to check f2fs_sb_has_readonly() */
2637 if (f2fs_hw_is_readonly(sbi))
2641 sbi->sb->s_flags &= ~SB_RDONLY;
2642 set_sbi_flag(sbi, SBI_IS_WRITABLE);
2646 * Turn on quotas which were not enabled for read-only mounts if
2647 * filesystem has quota feature, so that they are updated correctly.
2649 return f2fs_enable_quota_files(sbi, readonly);
2652 static void f2fs_recover_quota_end(struct f2fs_sb_info *sbi,
2656 f2fs_quota_off_umount(sbi->sb);
2658 if (is_sbi_flag_set(sbi, SBI_IS_WRITABLE)) {
2659 clear_sbi_flag(sbi, SBI_IS_WRITABLE);
2660 sbi->sb->s_flags |= SB_RDONLY;
2664 /* Read data from quotafile */
2665 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
2666 size_t len, loff_t off)
2668 struct inode *inode = sb_dqopt(sb)->files[type];
2669 struct address_space *mapping = inode->i_mapping;
2670 block_t blkidx = F2FS_BYTES_TO_BLK(off);
2671 int offset = off & (sb->s_blocksize - 1);
2674 loff_t i_size = i_size_read(inode);
2680 if (off + len > i_size)
2683 while (toread > 0) {
2684 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
2686 page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
2688 if (PTR_ERR(page) == -ENOMEM) {
2689 memalloc_retry_wait(GFP_NOFS);
2692 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2693 return PTR_ERR(page);
2698 if (unlikely(page->mapping != mapping)) {
2699 f2fs_put_page(page, 1);
2702 if (unlikely(!PageUptodate(page))) {
2703 f2fs_put_page(page, 1);
2704 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2708 memcpy_from_page(data, page, offset, tocopy);
2709 f2fs_put_page(page, 1);
2719 /* Write to quotafile */
2720 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2721 const char *data, size_t len, loff_t off)
2723 struct inode *inode = sb_dqopt(sb)->files[type];
2724 struct address_space *mapping = inode->i_mapping;
2725 const struct address_space_operations *a_ops = mapping->a_ops;
2726 int offset = off & (sb->s_blocksize - 1);
2727 size_t towrite = len;
2729 void *fsdata = NULL;
2733 while (towrite > 0) {
2734 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2737 err = a_ops->write_begin(NULL, mapping, off, tocopy,
2739 if (unlikely(err)) {
2740 if (err == -ENOMEM) {
2741 f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2744 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2748 memcpy_to_page(page, offset, data, tocopy);
2750 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2761 inode_set_mtime_to_ts(inode, inode_set_ctime_current(inode));
2762 f2fs_mark_inode_dirty_sync(inode, false);
2763 return len - towrite;
2766 int f2fs_dquot_initialize(struct inode *inode)
2768 if (time_to_inject(F2FS_I_SB(inode), FAULT_DQUOT_INIT))
2771 return dquot_initialize(inode);
2774 static struct dquot **f2fs_get_dquots(struct inode *inode)
2776 return F2FS_I(inode)->i_dquot;
2779 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2781 return &F2FS_I(inode)->i_reserved_quota;
2784 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2786 if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2787 f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2791 return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2792 F2FS_OPTION(sbi).s_jquota_fmt, type);
2795 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2800 if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2801 err = f2fs_enable_quotas(sbi->sb);
2803 f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2809 for (i = 0; i < MAXQUOTAS; i++) {
2810 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2811 err = f2fs_quota_on_mount(sbi, i);
2816 f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2823 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2826 struct inode *qf_inode;
2827 unsigned long qf_inum;
2828 unsigned long qf_flag = F2FS_QUOTA_DEFAULT_FL;
2831 BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2833 qf_inum = f2fs_qf_ino(sb, type);
2837 qf_inode = f2fs_iget(sb, qf_inum);
2838 if (IS_ERR(qf_inode)) {
2839 f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2840 return PTR_ERR(qf_inode);
2843 /* Don't account quota for quota files to avoid recursion */
2844 inode_lock(qf_inode);
2845 qf_inode->i_flags |= S_NOQUOTA;
2847 if ((F2FS_I(qf_inode)->i_flags & qf_flag) != qf_flag) {
2848 F2FS_I(qf_inode)->i_flags |= qf_flag;
2849 f2fs_set_inode_flags(qf_inode);
2851 inode_unlock(qf_inode);
2853 err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2858 static int f2fs_enable_quotas(struct super_block *sb)
2860 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2862 unsigned long qf_inum;
2863 bool quota_mopt[MAXQUOTAS] = {
2864 test_opt(sbi, USRQUOTA),
2865 test_opt(sbi, GRPQUOTA),
2866 test_opt(sbi, PRJQUOTA),
2869 if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2870 f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2874 sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2876 for (type = 0; type < MAXQUOTAS; type++) {
2877 qf_inum = f2fs_qf_ino(sb, type);
2879 err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2880 DQUOT_USAGE_ENABLED |
2881 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2883 f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2885 for (type--; type >= 0; type--)
2886 dquot_quota_off(sb, type);
2887 set_sbi_flag(F2FS_SB(sb),
2888 SBI_QUOTA_NEED_REPAIR);
2896 static int f2fs_quota_sync_file(struct f2fs_sb_info *sbi, int type)
2898 struct quota_info *dqopt = sb_dqopt(sbi->sb);
2899 struct address_space *mapping = dqopt->files[type]->i_mapping;
2902 ret = dquot_writeback_dquots(sbi->sb, type);
2906 ret = filemap_fdatawrite(mapping);
2910 /* if we are using journalled quota */
2911 if (is_journalled_quota(sbi))
2914 ret = filemap_fdatawait(mapping);
2916 truncate_inode_pages(&dqopt->files[type]->i_data, 0);
2919 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2923 int f2fs_quota_sync(struct super_block *sb, int type)
2925 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2926 struct quota_info *dqopt = sb_dqopt(sb);
2931 * Now when everything is written we can discard the pagecache so
2932 * that userspace sees the changes.
2934 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2936 if (type != -1 && cnt != type)
2939 if (!sb_has_quota_active(sb, cnt))
2942 if (!f2fs_sb_has_quota_ino(sbi))
2943 inode_lock(dqopt->files[cnt]);
2948 * f2fs_down_read(quota_sem)
2949 * dquot_writeback_dquots()
2952 * f2fs_down_read(quota_sem)
2955 f2fs_down_read(&sbi->quota_sem);
2957 ret = f2fs_quota_sync_file(sbi, cnt);
2959 f2fs_up_read(&sbi->quota_sem);
2960 f2fs_unlock_op(sbi);
2962 if (!f2fs_sb_has_quota_ino(sbi))
2963 inode_unlock(dqopt->files[cnt]);
2971 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2972 const struct path *path)
2974 struct inode *inode;
2977 /* if quota sysfile exists, deny enabling quota with specific file */
2978 if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2979 f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2983 if (path->dentry->d_sb != sb)
2986 err = f2fs_quota_sync(sb, type);
2990 inode = d_inode(path->dentry);
2992 err = filemap_fdatawrite(inode->i_mapping);
2996 err = filemap_fdatawait(inode->i_mapping);
3000 err = dquot_quota_on(sb, type, format_id, path);
3005 F2FS_I(inode)->i_flags |= F2FS_QUOTA_DEFAULT_FL;
3006 f2fs_set_inode_flags(inode);
3007 inode_unlock(inode);
3008 f2fs_mark_inode_dirty_sync(inode, false);
3013 static int __f2fs_quota_off(struct super_block *sb, int type)
3015 struct inode *inode = sb_dqopt(sb)->files[type];
3018 if (!inode || !igrab(inode))
3019 return dquot_quota_off(sb, type);
3021 err = f2fs_quota_sync(sb, type);
3025 err = dquot_quota_off(sb, type);
3026 if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
3030 F2FS_I(inode)->i_flags &= ~F2FS_QUOTA_DEFAULT_FL;
3031 f2fs_set_inode_flags(inode);
3032 inode_unlock(inode);
3033 f2fs_mark_inode_dirty_sync(inode, false);
3039 static int f2fs_quota_off(struct super_block *sb, int type)
3041 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3044 err = __f2fs_quota_off(sb, type);
3047 * quotactl can shutdown journalled quota, result in inconsistence
3048 * between quota record and fs data by following updates, tag the
3049 * flag to let fsck be aware of it.
3051 if (is_journalled_quota(sbi))
3052 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3056 void f2fs_quota_off_umount(struct super_block *sb)
3061 for (type = 0; type < MAXQUOTAS; type++) {
3062 err = __f2fs_quota_off(sb, type);
3064 int ret = dquot_quota_off(sb, type);
3066 f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
3068 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
3072 * In case of checkpoint=disable, we must flush quota blocks.
3073 * This can cause NULL exception for node_inode in end_io, since
3074 * put_super already dropped it.
3076 sync_filesystem(sb);
3079 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
3081 struct quota_info *dqopt = sb_dqopt(sb);
3084 for (type = 0; type < MAXQUOTAS; type++) {
3085 if (!dqopt->files[type])
3087 f2fs_inode_synced(dqopt->files[type]);
3091 static int f2fs_dquot_commit(struct dquot *dquot)
3093 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3096 f2fs_down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
3097 ret = dquot_commit(dquot);
3099 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3100 f2fs_up_read(&sbi->quota_sem);
3104 static int f2fs_dquot_acquire(struct dquot *dquot)
3106 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3109 f2fs_down_read(&sbi->quota_sem);
3110 ret = dquot_acquire(dquot);
3112 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3113 f2fs_up_read(&sbi->quota_sem);
3117 static int f2fs_dquot_release(struct dquot *dquot)
3119 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3120 int ret = dquot_release(dquot);
3123 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3127 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
3129 struct super_block *sb = dquot->dq_sb;
3130 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3131 int ret = dquot_mark_dquot_dirty(dquot);
3133 /* if we are using journalled quota */
3134 if (is_journalled_quota(sbi))
3135 set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
3140 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
3142 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3143 int ret = dquot_commit_info(sb, type);
3146 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3150 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
3152 *projid = F2FS_I(inode)->i_projid;
3156 static const struct dquot_operations f2fs_quota_operations = {
3157 .get_reserved_space = f2fs_get_reserved_space,
3158 .write_dquot = f2fs_dquot_commit,
3159 .acquire_dquot = f2fs_dquot_acquire,
3160 .release_dquot = f2fs_dquot_release,
3161 .mark_dirty = f2fs_dquot_mark_dquot_dirty,
3162 .write_info = f2fs_dquot_commit_info,
3163 .alloc_dquot = dquot_alloc,
3164 .destroy_dquot = dquot_destroy,
3165 .get_projid = f2fs_get_projid,
3166 .get_next_id = dquot_get_next_id,
3169 static const struct quotactl_ops f2fs_quotactl_ops = {
3170 .quota_on = f2fs_quota_on,
3171 .quota_off = f2fs_quota_off,
3172 .quota_sync = f2fs_quota_sync,
3173 .get_state = dquot_get_state,
3174 .set_info = dquot_set_dqinfo,
3175 .get_dqblk = dquot_get_dqblk,
3176 .set_dqblk = dquot_set_dqblk,
3177 .get_nextdqblk = dquot_get_next_dqblk,
3180 int f2fs_dquot_initialize(struct inode *inode)
3185 int f2fs_quota_sync(struct super_block *sb, int type)
3190 void f2fs_quota_off_umount(struct super_block *sb)
3195 static const struct super_operations f2fs_sops = {
3196 .alloc_inode = f2fs_alloc_inode,
3197 .free_inode = f2fs_free_inode,
3198 .drop_inode = f2fs_drop_inode,
3199 .write_inode = f2fs_write_inode,
3200 .dirty_inode = f2fs_dirty_inode,
3201 .show_options = f2fs_show_options,
3203 .quota_read = f2fs_quota_read,
3204 .quota_write = f2fs_quota_write,
3205 .get_dquots = f2fs_get_dquots,
3207 .evict_inode = f2fs_evict_inode,
3208 .put_super = f2fs_put_super,
3209 .sync_fs = f2fs_sync_fs,
3210 .freeze_fs = f2fs_freeze,
3211 .unfreeze_fs = f2fs_unfreeze,
3212 .statfs = f2fs_statfs,
3213 .remount_fs = f2fs_remount,
3216 #ifdef CONFIG_FS_ENCRYPTION
3217 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
3219 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3220 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3224 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
3227 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3230 * Encrypting the root directory is not allowed because fsck
3231 * expects lost+found directory to exist and remain unencrypted
3232 * if LOST_FOUND feature is enabled.
3235 if (f2fs_sb_has_lost_found(sbi) &&
3236 inode->i_ino == F2FS_ROOT_INO(sbi))
3239 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3240 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3241 ctx, len, fs_data, XATTR_CREATE);
3244 static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
3246 return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
3249 static bool f2fs_has_stable_inodes(struct super_block *sb)
3254 static struct block_device **f2fs_get_devices(struct super_block *sb,
3255 unsigned int *num_devs)
3257 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3258 struct block_device **devs;
3261 if (!f2fs_is_multi_device(sbi))
3264 devs = kmalloc_array(sbi->s_ndevs, sizeof(*devs), GFP_KERNEL);
3266 return ERR_PTR(-ENOMEM);
3268 for (i = 0; i < sbi->s_ndevs; i++)
3269 devs[i] = FDEV(i).bdev;
3270 *num_devs = sbi->s_ndevs;
3274 static const struct fscrypt_operations f2fs_cryptops = {
3275 .needs_bounce_pages = 1,
3276 .has_32bit_inodes = 1,
3277 .supports_subblock_data_units = 1,
3278 .legacy_key_prefix = "f2fs:",
3279 .get_context = f2fs_get_context,
3280 .set_context = f2fs_set_context,
3281 .get_dummy_policy = f2fs_get_dummy_policy,
3282 .empty_dir = f2fs_empty_dir,
3283 .has_stable_inodes = f2fs_has_stable_inodes,
3284 .get_devices = f2fs_get_devices,
3288 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
3289 u64 ino, u32 generation)
3291 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3292 struct inode *inode;
3294 if (f2fs_check_nid_range(sbi, ino))
3295 return ERR_PTR(-ESTALE);
3298 * f2fs_iget isn't quite right if the inode is currently unallocated!
3299 * However f2fs_iget currently does appropriate checks to handle stale
3300 * inodes so everything is OK.
3302 inode = f2fs_iget(sb, ino);
3304 return ERR_CAST(inode);
3305 if (unlikely(generation && inode->i_generation != generation)) {
3306 /* we didn't find the right inode.. */
3308 return ERR_PTR(-ESTALE);
3313 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
3314 int fh_len, int fh_type)
3316 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
3317 f2fs_nfs_get_inode);
3320 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
3321 int fh_len, int fh_type)
3323 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
3324 f2fs_nfs_get_inode);
3327 static const struct export_operations f2fs_export_ops = {
3328 .encode_fh = generic_encode_ino32_fh,
3329 .fh_to_dentry = f2fs_fh_to_dentry,
3330 .fh_to_parent = f2fs_fh_to_parent,
3331 .get_parent = f2fs_get_parent,
3334 loff_t max_file_blocks(struct inode *inode)
3340 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
3341 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
3342 * space in inode.i_addr, it will be more safe to reassign
3346 if (inode && f2fs_compressed_file(inode))
3347 leaf_count = ADDRS_PER_BLOCK(inode);
3349 leaf_count = DEF_ADDRS_PER_BLOCK;
3351 /* two direct node blocks */
3352 result += (leaf_count * 2);
3354 /* two indirect node blocks */
3355 leaf_count *= NIDS_PER_BLOCK;
3356 result += (leaf_count * 2);
3358 /* one double indirect node block */
3359 leaf_count *= NIDS_PER_BLOCK;
3360 result += leaf_count;
3365 static int __f2fs_commit_super(struct buffer_head *bh,
3366 struct f2fs_super_block *super)
3370 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
3371 set_buffer_dirty(bh);
3374 /* it's rare case, we can do fua all the time */
3375 return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
3378 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
3379 struct buffer_head *bh)
3381 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3382 (bh->b_data + F2FS_SUPER_OFFSET);
3383 struct super_block *sb = sbi->sb;
3384 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
3385 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
3386 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
3387 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
3388 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
3389 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
3390 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
3391 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
3392 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
3393 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
3394 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3395 u32 segment_count = le32_to_cpu(raw_super->segment_count);
3396 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3397 u64 main_end_blkaddr = main_blkaddr +
3398 (segment_count_main << log_blocks_per_seg);
3399 u64 seg_end_blkaddr = segment0_blkaddr +
3400 (segment_count << log_blocks_per_seg);
3402 if (segment0_blkaddr != cp_blkaddr) {
3403 f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
3404 segment0_blkaddr, cp_blkaddr);
3408 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
3410 f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
3411 cp_blkaddr, sit_blkaddr,
3412 segment_count_ckpt << log_blocks_per_seg);
3416 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
3418 f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
3419 sit_blkaddr, nat_blkaddr,
3420 segment_count_sit << log_blocks_per_seg);
3424 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
3426 f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
3427 nat_blkaddr, ssa_blkaddr,
3428 segment_count_nat << log_blocks_per_seg);
3432 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
3434 f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
3435 ssa_blkaddr, main_blkaddr,
3436 segment_count_ssa << log_blocks_per_seg);
3440 if (main_end_blkaddr > seg_end_blkaddr) {
3441 f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
3442 main_blkaddr, seg_end_blkaddr,
3443 segment_count_main << log_blocks_per_seg);
3445 } else if (main_end_blkaddr < seg_end_blkaddr) {
3449 /* fix in-memory information all the time */
3450 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
3451 segment0_blkaddr) >> log_blocks_per_seg);
3453 if (f2fs_readonly(sb) || f2fs_hw_is_readonly(sbi)) {
3454 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3457 err = __f2fs_commit_super(bh, NULL);
3458 res = err ? "failed" : "done";
3460 f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
3461 res, main_blkaddr, seg_end_blkaddr,
3462 segment_count_main << log_blocks_per_seg);
3469 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
3470 struct buffer_head *bh)
3472 block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
3473 block_t total_sections, blocks_per_seg;
3474 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3475 (bh->b_data + F2FS_SUPER_OFFSET);
3476 size_t crc_offset = 0;
3479 if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
3480 f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
3481 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
3485 /* Check checksum_offset and crc in superblock */
3486 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
3487 crc_offset = le32_to_cpu(raw_super->checksum_offset);
3489 offsetof(struct f2fs_super_block, crc)) {
3490 f2fs_info(sbi, "Invalid SB checksum offset: %zu",
3492 return -EFSCORRUPTED;
3494 crc = le32_to_cpu(raw_super->crc);
3495 if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
3496 f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
3497 return -EFSCORRUPTED;
3501 /* Currently, support only 4KB block size */
3502 if (le32_to_cpu(raw_super->log_blocksize) != F2FS_BLKSIZE_BITS) {
3503 f2fs_info(sbi, "Invalid log_blocksize (%u), supports only %u",
3504 le32_to_cpu(raw_super->log_blocksize),
3506 return -EFSCORRUPTED;
3509 /* check log blocks per segment */
3510 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
3511 f2fs_info(sbi, "Invalid log blocks per segment (%u)",
3512 le32_to_cpu(raw_super->log_blocks_per_seg));
3513 return -EFSCORRUPTED;
3516 /* Currently, support 512/1024/2048/4096/16K bytes sector size */
3517 if (le32_to_cpu(raw_super->log_sectorsize) >
3518 F2FS_MAX_LOG_SECTOR_SIZE ||
3519 le32_to_cpu(raw_super->log_sectorsize) <
3520 F2FS_MIN_LOG_SECTOR_SIZE) {
3521 f2fs_info(sbi, "Invalid log sectorsize (%u)",
3522 le32_to_cpu(raw_super->log_sectorsize));
3523 return -EFSCORRUPTED;
3525 if (le32_to_cpu(raw_super->log_sectors_per_block) +
3526 le32_to_cpu(raw_super->log_sectorsize) !=
3527 F2FS_MAX_LOG_SECTOR_SIZE) {
3528 f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
3529 le32_to_cpu(raw_super->log_sectors_per_block),
3530 le32_to_cpu(raw_super->log_sectorsize));
3531 return -EFSCORRUPTED;
3534 segment_count = le32_to_cpu(raw_super->segment_count);
3535 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3536 segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3537 secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3538 total_sections = le32_to_cpu(raw_super->section_count);
3540 /* blocks_per_seg should be 512, given the above check */
3541 blocks_per_seg = BIT(le32_to_cpu(raw_super->log_blocks_per_seg));
3543 if (segment_count > F2FS_MAX_SEGMENT ||
3544 segment_count < F2FS_MIN_SEGMENTS) {
3545 f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
3546 return -EFSCORRUPTED;
3549 if (total_sections > segment_count_main || total_sections < 1 ||
3550 segs_per_sec > segment_count || !segs_per_sec) {
3551 f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
3552 segment_count, total_sections, segs_per_sec);
3553 return -EFSCORRUPTED;
3556 if (segment_count_main != total_sections * segs_per_sec) {
3557 f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
3558 segment_count_main, total_sections, segs_per_sec);
3559 return -EFSCORRUPTED;
3562 if ((segment_count / segs_per_sec) < total_sections) {
3563 f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
3564 segment_count, segs_per_sec, total_sections);
3565 return -EFSCORRUPTED;
3568 if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
3569 f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
3570 segment_count, le64_to_cpu(raw_super->block_count));
3571 return -EFSCORRUPTED;
3574 if (RDEV(0).path[0]) {
3575 block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
3578 while (i < MAX_DEVICES && RDEV(i).path[0]) {
3579 dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
3582 if (segment_count != dev_seg_count) {
3583 f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
3584 segment_count, dev_seg_count);
3585 return -EFSCORRUPTED;
3588 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
3589 !bdev_is_zoned(sbi->sb->s_bdev)) {
3590 f2fs_info(sbi, "Zoned block device path is missing");
3591 return -EFSCORRUPTED;
3595 if (secs_per_zone > total_sections || !secs_per_zone) {
3596 f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
3597 secs_per_zone, total_sections);
3598 return -EFSCORRUPTED;
3600 if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
3601 raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
3602 (le32_to_cpu(raw_super->extension_count) +
3603 raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
3604 f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
3605 le32_to_cpu(raw_super->extension_count),
3606 raw_super->hot_ext_count,
3607 F2FS_MAX_EXTENSION);
3608 return -EFSCORRUPTED;
3611 if (le32_to_cpu(raw_super->cp_payload) >=
3612 (blocks_per_seg - F2FS_CP_PACKS -
3613 NR_CURSEG_PERSIST_TYPE)) {
3614 f2fs_info(sbi, "Insane cp_payload (%u >= %u)",
3615 le32_to_cpu(raw_super->cp_payload),
3616 blocks_per_seg - F2FS_CP_PACKS -
3617 NR_CURSEG_PERSIST_TYPE);
3618 return -EFSCORRUPTED;
3621 /* check reserved ino info */
3622 if (le32_to_cpu(raw_super->node_ino) != 1 ||
3623 le32_to_cpu(raw_super->meta_ino) != 2 ||
3624 le32_to_cpu(raw_super->root_ino) != 3) {
3625 f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
3626 le32_to_cpu(raw_super->node_ino),
3627 le32_to_cpu(raw_super->meta_ino),
3628 le32_to_cpu(raw_super->root_ino));
3629 return -EFSCORRUPTED;
3632 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
3633 if (sanity_check_area_boundary(sbi, bh))
3634 return -EFSCORRUPTED;
3639 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
3641 unsigned int total, fsmeta;
3642 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3643 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
3644 unsigned int ovp_segments, reserved_segments;
3645 unsigned int main_segs, blocks_per_seg;
3646 unsigned int sit_segs, nat_segs;
3647 unsigned int sit_bitmap_size, nat_bitmap_size;
3648 unsigned int log_blocks_per_seg;
3649 unsigned int segment_count_main;
3650 unsigned int cp_pack_start_sum, cp_payload;
3651 block_t user_block_count, valid_user_blocks;
3652 block_t avail_node_count, valid_node_count;
3653 unsigned int nat_blocks, nat_bits_bytes, nat_bits_blocks;
3656 total = le32_to_cpu(raw_super->segment_count);
3657 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
3658 sit_segs = le32_to_cpu(raw_super->segment_count_sit);
3660 nat_segs = le32_to_cpu(raw_super->segment_count_nat);
3662 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
3663 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
3665 if (unlikely(fsmeta >= total))
3668 ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
3669 reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
3671 if (!f2fs_sb_has_readonly(sbi) &&
3672 unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
3673 ovp_segments == 0 || reserved_segments == 0)) {
3674 f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
3677 user_block_count = le64_to_cpu(ckpt->user_block_count);
3678 segment_count_main = le32_to_cpu(raw_super->segment_count_main) +
3679 (f2fs_sb_has_readonly(sbi) ? 1 : 0);
3680 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3681 if (!user_block_count || user_block_count >=
3682 segment_count_main << log_blocks_per_seg) {
3683 f2fs_err(sbi, "Wrong user_block_count: %u",
3688 valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
3689 if (valid_user_blocks > user_block_count) {
3690 f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
3691 valid_user_blocks, user_block_count);
3695 valid_node_count = le32_to_cpu(ckpt->valid_node_count);
3696 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
3697 if (valid_node_count > avail_node_count) {
3698 f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
3699 valid_node_count, avail_node_count);
3703 main_segs = le32_to_cpu(raw_super->segment_count_main);
3704 blocks_per_seg = sbi->blocks_per_seg;
3706 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3707 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
3708 le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
3711 if (f2fs_sb_has_readonly(sbi))
3714 for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
3715 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3716 le32_to_cpu(ckpt->cur_node_segno[j])) {
3717 f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
3719 le32_to_cpu(ckpt->cur_node_segno[i]));
3725 for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
3726 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
3727 le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
3730 if (f2fs_sb_has_readonly(sbi))
3733 for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3734 if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3735 le32_to_cpu(ckpt->cur_data_segno[j])) {
3736 f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3738 le32_to_cpu(ckpt->cur_data_segno[i]));
3743 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3744 for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3745 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3746 le32_to_cpu(ckpt->cur_data_segno[j])) {
3747 f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3749 le32_to_cpu(ckpt->cur_node_segno[i]));
3755 sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3756 nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3758 if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3759 nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3760 f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3761 sit_bitmap_size, nat_bitmap_size);
3765 cp_pack_start_sum = __start_sum_addr(sbi);
3766 cp_payload = __cp_payload(sbi);
3767 if (cp_pack_start_sum < cp_payload + 1 ||
3768 cp_pack_start_sum > blocks_per_seg - 1 -
3769 NR_CURSEG_PERSIST_TYPE) {
3770 f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3775 if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3776 le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3777 f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3778 "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3779 "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3780 le32_to_cpu(ckpt->checksum_offset));
3784 nat_blocks = nat_segs << log_blocks_per_seg;
3785 nat_bits_bytes = nat_blocks / BITS_PER_BYTE;
3786 nat_bits_blocks = F2FS_BLK_ALIGN((nat_bits_bytes << 1) + 8);
3787 if (__is_set_ckpt_flags(ckpt, CP_NAT_BITS_FLAG) &&
3788 (cp_payload + F2FS_CP_PACKS +
3789 NR_CURSEG_PERSIST_TYPE + nat_bits_blocks >= blocks_per_seg)) {
3790 f2fs_warn(sbi, "Insane cp_payload: %u, nat_bits_blocks: %u)",
3791 cp_payload, nat_bits_blocks);
3795 if (unlikely(f2fs_cp_error(sbi))) {
3796 f2fs_err(sbi, "A bug case: need to run fsck");
3802 static void init_sb_info(struct f2fs_sb_info *sbi)
3804 struct f2fs_super_block *raw_super = sbi->raw_super;
3807 sbi->log_sectors_per_block =
3808 le32_to_cpu(raw_super->log_sectors_per_block);
3809 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3810 sbi->blocksize = BIT(sbi->log_blocksize);
3811 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3812 sbi->blocks_per_seg = BIT(sbi->log_blocks_per_seg);
3813 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3814 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3815 sbi->total_sections = le32_to_cpu(raw_super->section_count);
3816 sbi->total_node_count =
3817 (le32_to_cpu(raw_super->segment_count_nat) / 2)
3818 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
3819 F2FS_ROOT_INO(sbi) = le32_to_cpu(raw_super->root_ino);
3820 F2FS_NODE_INO(sbi) = le32_to_cpu(raw_super->node_ino);
3821 F2FS_META_INO(sbi) = le32_to_cpu(raw_super->meta_ino);
3822 sbi->cur_victim_sec = NULL_SECNO;
3823 sbi->gc_mode = GC_NORMAL;
3824 sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3825 sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3826 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3827 sbi->migration_granularity = sbi->segs_per_sec;
3828 sbi->seq_file_ra_mul = MIN_RA_MUL;
3829 sbi->max_fragment_chunk = DEF_FRAGMENT_SIZE;
3830 sbi->max_fragment_hole = DEF_FRAGMENT_SIZE;
3831 spin_lock_init(&sbi->gc_remaining_trials_lock);
3832 atomic64_set(&sbi->current_atomic_write, 0);
3834 sbi->dir_level = DEF_DIR_LEVEL;
3835 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3836 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3837 sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3838 sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3839 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3840 sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3841 DEF_UMOUNT_DISCARD_TIMEOUT;
3842 clear_sbi_flag(sbi, SBI_NEED_FSCK);
3844 for (i = 0; i < NR_COUNT_TYPE; i++)
3845 atomic_set(&sbi->nr_pages[i], 0);
3847 for (i = 0; i < META; i++)
3848 atomic_set(&sbi->wb_sync_req[i], 0);
3850 INIT_LIST_HEAD(&sbi->s_list);
3851 mutex_init(&sbi->umount_mutex);
3852 init_f2fs_rwsem(&sbi->io_order_lock);
3853 spin_lock_init(&sbi->cp_lock);
3855 sbi->dirty_device = 0;
3856 spin_lock_init(&sbi->dev_lock);
3858 init_f2fs_rwsem(&sbi->sb_lock);
3859 init_f2fs_rwsem(&sbi->pin_sem);
3862 static int init_percpu_info(struct f2fs_sb_info *sbi)
3866 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3870 err = percpu_counter_init(&sbi->rf_node_block_count, 0, GFP_KERNEL);
3872 goto err_valid_block;
3874 err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3877 goto err_node_block;
3881 percpu_counter_destroy(&sbi->rf_node_block_count);
3883 percpu_counter_destroy(&sbi->alloc_valid_block_count);
3887 #ifdef CONFIG_BLK_DEV_ZONED
3889 struct f2fs_report_zones_args {
3890 struct f2fs_sb_info *sbi;
3891 struct f2fs_dev_info *dev;
3894 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3897 struct f2fs_report_zones_args *rz_args = data;
3898 block_t unusable_blocks = (zone->len - zone->capacity) >>
3899 F2FS_LOG_SECTORS_PER_BLOCK;
3901 if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3904 set_bit(idx, rz_args->dev->blkz_seq);
3905 if (!rz_args->sbi->unusable_blocks_per_sec) {
3906 rz_args->sbi->unusable_blocks_per_sec = unusable_blocks;
3909 if (rz_args->sbi->unusable_blocks_per_sec != unusable_blocks) {
3910 f2fs_err(rz_args->sbi, "F2FS supports single zone capacity\n");
3916 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3918 struct block_device *bdev = FDEV(devi).bdev;
3919 sector_t nr_sectors = bdev_nr_sectors(bdev);
3920 struct f2fs_report_zones_args rep_zone_arg;
3924 if (!f2fs_sb_has_blkzoned(sbi))
3927 zone_sectors = bdev_zone_sectors(bdev);
3928 if (!is_power_of_2(zone_sectors)) {
3929 f2fs_err(sbi, "F2FS does not support non power of 2 zone sizes\n");
3933 if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3934 SECTOR_TO_BLOCK(zone_sectors))
3936 sbi->blocks_per_blkz = SECTOR_TO_BLOCK(zone_sectors);
3937 FDEV(devi).nr_blkz = div_u64(SECTOR_TO_BLOCK(nr_sectors),
3938 sbi->blocks_per_blkz);
3939 if (nr_sectors & (zone_sectors - 1))
3940 FDEV(devi).nr_blkz++;
3942 FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3943 BITS_TO_LONGS(FDEV(devi).nr_blkz)
3944 * sizeof(unsigned long),
3946 if (!FDEV(devi).blkz_seq)
3949 rep_zone_arg.sbi = sbi;
3950 rep_zone_arg.dev = &FDEV(devi);
3952 ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3961 * Read f2fs raw super block.
3962 * Because we have two copies of super block, so read both of them
3963 * to get the first valid one. If any one of them is broken, we pass
3964 * them recovery flag back to the caller.
3966 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3967 struct f2fs_super_block **raw_super,
3968 int *valid_super_block, int *recovery)
3970 struct super_block *sb = sbi->sb;
3972 struct buffer_head *bh;
3973 struct f2fs_super_block *super;
3976 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3980 for (block = 0; block < 2; block++) {
3981 bh = sb_bread(sb, block);
3983 f2fs_err(sbi, "Unable to read %dth superblock",
3990 /* sanity checking of raw super */
3991 err = sanity_check_raw_super(sbi, bh);
3993 f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
4001 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
4003 *valid_super_block = block;
4009 /* No valid superblock */
4018 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
4020 struct buffer_head *bh;
4024 if ((recover && f2fs_readonly(sbi->sb)) ||
4025 f2fs_hw_is_readonly(sbi)) {
4026 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
4030 /* we should update superblock crc here */
4031 if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
4032 crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
4033 offsetof(struct f2fs_super_block, crc));
4034 F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
4037 /* write back-up superblock first */
4038 bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
4041 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
4044 /* if we are in recovery path, skip writing valid superblock */
4048 /* write current valid superblock */
4049 bh = sb_bread(sbi->sb, sbi->valid_super_block);
4052 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
4057 static void save_stop_reason(struct f2fs_sb_info *sbi, unsigned char reason)
4059 unsigned long flags;
4061 spin_lock_irqsave(&sbi->error_lock, flags);
4062 if (sbi->stop_reason[reason] < GENMASK(BITS_PER_BYTE - 1, 0))
4063 sbi->stop_reason[reason]++;
4064 spin_unlock_irqrestore(&sbi->error_lock, flags);
4067 static void f2fs_record_stop_reason(struct f2fs_sb_info *sbi)
4069 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
4070 unsigned long flags;
4073 f2fs_down_write(&sbi->sb_lock);
4075 spin_lock_irqsave(&sbi->error_lock, flags);
4076 if (sbi->error_dirty) {
4077 memcpy(F2FS_RAW_SUPER(sbi)->s_errors, sbi->errors,
4079 sbi->error_dirty = false;
4081 memcpy(raw_super->s_stop_reason, sbi->stop_reason, MAX_STOP_REASON);
4082 spin_unlock_irqrestore(&sbi->error_lock, flags);
4084 err = f2fs_commit_super(sbi, false);
4086 f2fs_up_write(&sbi->sb_lock);
4088 f2fs_err(sbi, "f2fs_commit_super fails to record err:%d", err);
4091 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag)
4093 unsigned long flags;
4095 spin_lock_irqsave(&sbi->error_lock, flags);
4096 if (!test_bit(flag, (unsigned long *)sbi->errors)) {
4097 set_bit(flag, (unsigned long *)sbi->errors);
4098 sbi->error_dirty = true;
4100 spin_unlock_irqrestore(&sbi->error_lock, flags);
4103 static bool f2fs_update_errors(struct f2fs_sb_info *sbi)
4105 unsigned long flags;
4106 bool need_update = false;
4108 spin_lock_irqsave(&sbi->error_lock, flags);
4109 if (sbi->error_dirty) {
4110 memcpy(F2FS_RAW_SUPER(sbi)->s_errors, sbi->errors,
4112 sbi->error_dirty = false;
4115 spin_unlock_irqrestore(&sbi->error_lock, flags);
4120 static void f2fs_record_errors(struct f2fs_sb_info *sbi, unsigned char error)
4124 f2fs_down_write(&sbi->sb_lock);
4126 if (!f2fs_update_errors(sbi))
4129 err = f2fs_commit_super(sbi, false);
4131 f2fs_err(sbi, "f2fs_commit_super fails to record errors:%u, err:%d",
4134 f2fs_up_write(&sbi->sb_lock);
4137 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error)
4139 f2fs_save_errors(sbi, error);
4140 f2fs_record_errors(sbi, error);
4143 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error)
4145 f2fs_save_errors(sbi, error);
4147 if (!sbi->error_dirty)
4149 if (!test_bit(error, (unsigned long *)sbi->errors))
4151 schedule_work(&sbi->s_error_work);
4154 static bool system_going_down(void)
4156 return system_state == SYSTEM_HALT || system_state == SYSTEM_POWER_OFF
4157 || system_state == SYSTEM_RESTART;
4160 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason,
4163 struct super_block *sb = sbi->sb;
4164 bool shutdown = reason == STOP_CP_REASON_SHUTDOWN;
4165 bool continue_fs = !shutdown &&
4166 F2FS_OPTION(sbi).errors == MOUNT_ERRORS_CONTINUE;
4168 set_ckpt_flags(sbi, CP_ERROR_FLAG);
4170 if (!f2fs_hw_is_readonly(sbi)) {
4171 save_stop_reason(sbi, reason);
4173 if (irq_context && !shutdown)
4174 schedule_work(&sbi->s_error_work);
4176 f2fs_record_stop_reason(sbi);
4180 * We force ERRORS_RO behavior when system is rebooting. Otherwise we
4181 * could panic during 'reboot -f' as the underlying device got already
4184 if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_PANIC &&
4185 !shutdown && !system_going_down() &&
4186 !is_sbi_flag_set(sbi, SBI_IS_SHUTDOWN))
4187 panic("F2FS-fs (device %s): panic forced after error\n",
4191 set_sbi_flag(sbi, SBI_IS_SHUTDOWN);
4193 /* continue filesystem operators if errors=continue */
4194 if (continue_fs || f2fs_readonly(sb))
4197 f2fs_warn(sbi, "Remounting filesystem read-only");
4199 * Make sure updated value of ->s_mount_flags will be visible before
4203 sb->s_flags |= SB_RDONLY;
4206 static void f2fs_record_error_work(struct work_struct *work)
4208 struct f2fs_sb_info *sbi = container_of(work,
4209 struct f2fs_sb_info, s_error_work);
4211 f2fs_record_stop_reason(sbi);
4214 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
4216 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
4217 unsigned int max_devices = MAX_DEVICES;
4218 unsigned int logical_blksize;
4219 blk_mode_t mode = sb_open_mode(sbi->sb->s_flags);
4222 /* Initialize single device information */
4223 if (!RDEV(0).path[0]) {
4224 if (!bdev_is_zoned(sbi->sb->s_bdev))
4230 * Initialize multiple devices information, or single
4231 * zoned block device information.
4233 sbi->devs = f2fs_kzalloc(sbi,
4234 array_size(max_devices,
4235 sizeof(struct f2fs_dev_info)),
4240 logical_blksize = bdev_logical_block_size(sbi->sb->s_bdev);
4241 sbi->aligned_blksize = true;
4243 for (i = 0; i < max_devices; i++) {
4245 FDEV(0).bdev_handle = sbi->sb->s_bdev_handle;
4246 else if (!RDEV(i).path[0])
4249 if (max_devices > 1) {
4250 /* Multi-device mount */
4251 memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
4252 FDEV(i).total_segments =
4253 le32_to_cpu(RDEV(i).total_segments);
4255 FDEV(i).start_blk = 0;
4256 FDEV(i).end_blk = FDEV(i).start_blk +
4257 (FDEV(i).total_segments <<
4258 sbi->log_blocks_per_seg) - 1 +
4259 le32_to_cpu(raw_super->segment0_blkaddr);
4261 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
4262 FDEV(i).end_blk = FDEV(i).start_blk +
4263 (FDEV(i).total_segments <<
4264 sbi->log_blocks_per_seg) - 1;
4265 FDEV(i).bdev_handle = bdev_open_by_path(
4266 FDEV(i).path, mode, sbi->sb, NULL);
4269 if (IS_ERR(FDEV(i).bdev_handle))
4270 return PTR_ERR(FDEV(i).bdev_handle);
4272 FDEV(i).bdev = FDEV(i).bdev_handle->bdev;
4273 /* to release errored devices */
4274 sbi->s_ndevs = i + 1;
4276 if (logical_blksize != bdev_logical_block_size(FDEV(i).bdev))
4277 sbi->aligned_blksize = false;
4279 #ifdef CONFIG_BLK_DEV_ZONED
4280 if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
4281 !f2fs_sb_has_blkzoned(sbi)) {
4282 f2fs_err(sbi, "Zoned block device feature not enabled");
4285 if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
4286 if (init_blkz_info(sbi, i)) {
4287 f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
4290 if (max_devices == 1)
4292 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
4294 FDEV(i).total_segments,
4295 FDEV(i).start_blk, FDEV(i).end_blk,
4296 bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
4297 "Host-aware" : "Host-managed");
4301 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
4303 FDEV(i).total_segments,
4304 FDEV(i).start_blk, FDEV(i).end_blk);
4307 "IO Block Size: %8ld KB", F2FS_IO_SIZE_KB(sbi));
4311 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
4313 #if IS_ENABLED(CONFIG_UNICODE)
4314 if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
4315 const struct f2fs_sb_encodings *encoding_info;
4316 struct unicode_map *encoding;
4317 __u16 encoding_flags;
4319 encoding_info = f2fs_sb_read_encoding(sbi->raw_super);
4320 if (!encoding_info) {
4322 "Encoding requested by superblock is unknown");
4326 encoding_flags = le16_to_cpu(sbi->raw_super->s_encoding_flags);
4327 encoding = utf8_load(encoding_info->version);
4328 if (IS_ERR(encoding)) {
4330 "can't mount with superblock charset: %s-%u.%u.%u "
4331 "not supported by the kernel. flags: 0x%x.",
4332 encoding_info->name,
4333 unicode_major(encoding_info->version),
4334 unicode_minor(encoding_info->version),
4335 unicode_rev(encoding_info->version),
4337 return PTR_ERR(encoding);
4339 f2fs_info(sbi, "Using encoding defined by superblock: "
4340 "%s-%u.%u.%u with flags 0x%hx", encoding_info->name,
4341 unicode_major(encoding_info->version),
4342 unicode_minor(encoding_info->version),
4343 unicode_rev(encoding_info->version),
4346 sbi->sb->s_encoding = encoding;
4347 sbi->sb->s_encoding_flags = encoding_flags;
4350 if (f2fs_sb_has_casefold(sbi)) {
4351 f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
4358 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
4360 /* adjust parameters according to the volume size */
4361 if (MAIN_SEGS(sbi) <= SMALL_VOLUME_SEGMENTS) {
4362 if (f2fs_block_unit_discard(sbi))
4363 SM_I(sbi)->dcc_info->discard_granularity =
4364 MIN_DISCARD_GRANULARITY;
4365 if (!f2fs_lfs_mode(sbi))
4366 SM_I(sbi)->ipu_policy = BIT(F2FS_IPU_FORCE) |
4367 BIT(F2FS_IPU_HONOR_OPU_WRITE);
4370 sbi->readdir_ra = true;
4373 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
4375 struct f2fs_sb_info *sbi;
4376 struct f2fs_super_block *raw_super;
4379 bool skip_recovery = false, need_fsck = false;
4380 char *options = NULL;
4381 int recovery, i, valid_super_block;
4382 struct curseg_info *seg_i;
4385 bool quota_enabled = false;
4391 valid_super_block = -1;
4394 /* allocate memory for f2fs-specific super block info */
4395 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
4401 /* initialize locks within allocated memory */
4402 init_f2fs_rwsem(&sbi->gc_lock);
4403 mutex_init(&sbi->writepages);
4404 init_f2fs_rwsem(&sbi->cp_global_sem);
4405 init_f2fs_rwsem(&sbi->node_write);
4406 init_f2fs_rwsem(&sbi->node_change);
4407 spin_lock_init(&sbi->stat_lock);
4408 init_f2fs_rwsem(&sbi->cp_rwsem);
4409 init_f2fs_rwsem(&sbi->quota_sem);
4410 init_waitqueue_head(&sbi->cp_wait);
4411 spin_lock_init(&sbi->error_lock);
4413 for (i = 0; i < NR_INODE_TYPE; i++) {
4414 INIT_LIST_HEAD(&sbi->inode_list[i]);
4415 spin_lock_init(&sbi->inode_lock[i]);
4417 mutex_init(&sbi->flush_lock);
4419 /* Load the checksum driver */
4420 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
4421 if (IS_ERR(sbi->s_chksum_driver)) {
4422 f2fs_err(sbi, "Cannot load crc32 driver.");
4423 err = PTR_ERR(sbi->s_chksum_driver);
4424 sbi->s_chksum_driver = NULL;
4428 /* set a block size */
4429 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
4430 f2fs_err(sbi, "unable to set blocksize");
4434 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
4439 sb->s_fs_info = sbi;
4440 sbi->raw_super = raw_super;
4442 INIT_WORK(&sbi->s_error_work, f2fs_record_error_work);
4443 memcpy(sbi->errors, raw_super->s_errors, MAX_F2FS_ERRORS);
4444 memcpy(sbi->stop_reason, raw_super->s_stop_reason, MAX_STOP_REASON);
4446 /* precompute checksum seed for metadata */
4447 if (f2fs_sb_has_inode_chksum(sbi))
4448 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
4449 sizeof(raw_super->uuid));
4451 default_options(sbi, false);
4452 /* parse mount options */
4453 options = kstrdup((const char *)data, GFP_KERNEL);
4454 if (data && !options) {
4459 err = parse_options(sb, options, false);
4463 sb->s_maxbytes = max_file_blocks(NULL) <<
4464 le32_to_cpu(raw_super->log_blocksize);
4465 sb->s_max_links = F2FS_LINK_MAX;
4467 err = f2fs_setup_casefold(sbi);
4472 sb->dq_op = &f2fs_quota_operations;
4473 sb->s_qcop = &f2fs_quotactl_ops;
4474 sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
4476 if (f2fs_sb_has_quota_ino(sbi)) {
4477 for (i = 0; i < MAXQUOTAS; i++) {
4478 if (f2fs_qf_ino(sbi->sb, i))
4479 sbi->nquota_files++;
4484 sb->s_op = &f2fs_sops;
4485 #ifdef CONFIG_FS_ENCRYPTION
4486 sb->s_cop = &f2fs_cryptops;
4488 #ifdef CONFIG_FS_VERITY
4489 sb->s_vop = &f2fs_verityops;
4491 sb->s_xattr = f2fs_xattr_handlers;
4492 sb->s_export_op = &f2fs_export_ops;
4493 sb->s_magic = F2FS_SUPER_MAGIC;
4494 sb->s_time_gran = 1;
4495 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
4496 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
4497 memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
4498 sb->s_iflags |= SB_I_CGROUPWB;
4500 /* init f2fs-specific super block info */
4501 sbi->valid_super_block = valid_super_block;
4503 /* disallow all the data/node/meta page writes */
4504 set_sbi_flag(sbi, SBI_POR_DOING);
4506 err = f2fs_init_write_merge_io(sbi);
4512 err = f2fs_init_iostat(sbi);
4516 err = init_percpu_info(sbi);
4520 if (F2FS_IO_ALIGNED(sbi)) {
4521 sbi->write_io_dummy =
4522 mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
4523 if (!sbi->write_io_dummy) {
4529 /* init per sbi slab cache */
4530 err = f2fs_init_xattr_caches(sbi);
4533 err = f2fs_init_page_array_cache(sbi);
4535 goto free_xattr_cache;
4537 /* get an inode for meta space */
4538 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
4539 if (IS_ERR(sbi->meta_inode)) {
4540 f2fs_err(sbi, "Failed to read F2FS meta data inode");
4541 err = PTR_ERR(sbi->meta_inode);
4542 goto free_page_array_cache;
4545 err = f2fs_get_valid_checkpoint(sbi);
4547 f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
4548 goto free_meta_inode;
4551 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
4552 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
4553 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
4554 set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4555 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
4558 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
4559 set_sbi_flag(sbi, SBI_NEED_FSCK);
4561 /* Initialize device list */
4562 err = f2fs_scan_devices(sbi);
4564 f2fs_err(sbi, "Failed to find devices");
4568 err = f2fs_init_post_read_wq(sbi);
4570 f2fs_err(sbi, "Failed to initialize post read workqueue");
4574 sbi->total_valid_node_count =
4575 le32_to_cpu(sbi->ckpt->valid_node_count);
4576 percpu_counter_set(&sbi->total_valid_inode_count,
4577 le32_to_cpu(sbi->ckpt->valid_inode_count));
4578 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
4579 sbi->total_valid_block_count =
4580 le64_to_cpu(sbi->ckpt->valid_block_count);
4581 sbi->last_valid_block_count = sbi->total_valid_block_count;
4582 sbi->reserved_blocks = 0;
4583 sbi->current_reserved_blocks = 0;
4584 limit_reserve_root(sbi);
4585 adjust_unusable_cap_perc(sbi);
4587 f2fs_init_extent_cache_info(sbi);
4589 f2fs_init_ino_entry_info(sbi);
4591 f2fs_init_fsync_node_info(sbi);
4593 /* setup checkpoint request control and start checkpoint issue thread */
4594 f2fs_init_ckpt_req_control(sbi);
4595 if (!f2fs_readonly(sb) && !test_opt(sbi, DISABLE_CHECKPOINT) &&
4596 test_opt(sbi, MERGE_CHECKPOINT)) {
4597 err = f2fs_start_ckpt_thread(sbi);
4600 "Failed to start F2FS issue_checkpoint_thread (%d)",
4602 goto stop_ckpt_thread;
4606 /* setup f2fs internal modules */
4607 err = f2fs_build_segment_manager(sbi);
4609 f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
4613 err = f2fs_build_node_manager(sbi);
4615 f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
4620 err = adjust_reserved_segment(sbi);
4624 /* For write statistics */
4625 sbi->sectors_written_start = f2fs_get_sectors_written(sbi);
4627 /* Read accumulated write IO statistics if exists */
4628 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
4629 if (__exist_node_summaries(sbi))
4630 sbi->kbytes_written =
4631 le64_to_cpu(seg_i->journal->info.kbytes_written);
4633 f2fs_build_gc_manager(sbi);
4635 err = f2fs_build_stats(sbi);
4639 /* get an inode for node space */
4640 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
4641 if (IS_ERR(sbi->node_inode)) {
4642 f2fs_err(sbi, "Failed to read node inode");
4643 err = PTR_ERR(sbi->node_inode);
4647 /* read root inode and dentry */
4648 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
4650 f2fs_err(sbi, "Failed to read root inode");
4651 err = PTR_ERR(root);
4652 goto free_node_inode;
4654 if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
4655 !root->i_size || !root->i_nlink) {
4658 goto free_node_inode;
4661 sb->s_root = d_make_root(root); /* allocate root dentry */
4664 goto free_node_inode;
4667 err = f2fs_init_compress_inode(sbi);
4669 goto free_root_inode;
4671 err = f2fs_register_sysfs(sbi);
4673 goto free_compress_inode;
4676 /* Enable quota usage during mount */
4677 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
4678 err = f2fs_enable_quotas(sb);
4680 f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
4683 quota_enabled = f2fs_recover_quota_begin(sbi);
4685 /* if there are any orphan inodes, free them */
4686 err = f2fs_recover_orphan_inodes(sbi);
4690 if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
4691 goto reset_checkpoint;
4693 /* recover fsynced data */
4694 if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
4695 !test_opt(sbi, NORECOVERY)) {
4697 * mount should be failed, when device has readonly mode, and
4698 * previous checkpoint was not done by clean system shutdown.
4700 if (f2fs_hw_is_readonly(sbi)) {
4701 if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4702 err = f2fs_recover_fsync_data(sbi, true);
4705 f2fs_err(sbi, "Need to recover fsync data, but "
4706 "write access unavailable, please try "
4707 "mount w/ disable_roll_forward or norecovery");
4712 f2fs_info(sbi, "write access unavailable, skipping recovery");
4713 goto reset_checkpoint;
4717 set_sbi_flag(sbi, SBI_NEED_FSCK);
4720 goto reset_checkpoint;
4722 err = f2fs_recover_fsync_data(sbi, false);
4725 skip_recovery = true;
4727 f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
4732 err = f2fs_recover_fsync_data(sbi, true);
4734 if (!f2fs_readonly(sb) && err > 0) {
4736 f2fs_err(sbi, "Need to recover fsync data");
4742 f2fs_recover_quota_end(sbi, quota_enabled);
4746 * If the f2fs is not readonly and fsync data recovery succeeds,
4747 * check zoned block devices' write pointer consistency.
4749 if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
4750 err = f2fs_check_write_pointer(sbi);
4755 f2fs_init_inmem_curseg(sbi);
4757 /* f2fs_recover_fsync_data() cleared this already */
4758 clear_sbi_flag(sbi, SBI_POR_DOING);
4760 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
4761 err = f2fs_disable_checkpoint(sbi);
4763 goto sync_free_meta;
4764 } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
4765 f2fs_enable_checkpoint(sbi);
4769 * If filesystem is not mounted as read-only then
4770 * do start the gc_thread.
4772 if ((F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF ||
4773 test_opt(sbi, GC_MERGE)) && !f2fs_readonly(sb)) {
4774 /* After POR, we can run background GC thread.*/
4775 err = f2fs_start_gc_thread(sbi);
4777 goto sync_free_meta;
4781 /* recover broken superblock */
4783 err = f2fs_commit_super(sbi, true);
4784 f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
4785 sbi->valid_super_block ? 1 : 2, err);
4788 f2fs_join_shrinker(sbi);
4790 f2fs_tuning_parameters(sbi);
4792 f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
4793 cur_cp_version(F2FS_CKPT(sbi)));
4794 f2fs_update_time(sbi, CP_TIME);
4795 f2fs_update_time(sbi, REQ_TIME);
4796 clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4800 /* safe to flush all the data */
4801 sync_filesystem(sbi->sb);
4806 f2fs_truncate_quota_inode_pages(sb);
4807 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
4808 f2fs_quota_off_umount(sbi->sb);
4811 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
4812 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
4813 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
4814 * falls into an infinite loop in f2fs_sync_meta_pages().
4816 truncate_inode_pages_final(META_MAPPING(sbi));
4817 /* evict some inodes being cached by GC */
4819 f2fs_unregister_sysfs(sbi);
4820 free_compress_inode:
4821 f2fs_destroy_compress_inode(sbi);
4826 f2fs_release_ino_entry(sbi, true);
4827 truncate_inode_pages_final(NODE_MAPPING(sbi));
4828 iput(sbi->node_inode);
4829 sbi->node_inode = NULL;
4831 f2fs_destroy_stats(sbi);
4833 /* stop discard thread before destroying node manager */
4834 f2fs_stop_discard_thread(sbi);
4835 f2fs_destroy_node_manager(sbi);
4837 f2fs_destroy_segment_manager(sbi);
4839 f2fs_stop_ckpt_thread(sbi);
4840 /* flush s_error_work before sbi destroy */
4841 flush_work(&sbi->s_error_work);
4842 f2fs_destroy_post_read_wq(sbi);
4844 destroy_device_list(sbi);
4847 make_bad_inode(sbi->meta_inode);
4848 iput(sbi->meta_inode);
4849 sbi->meta_inode = NULL;
4850 free_page_array_cache:
4851 f2fs_destroy_page_array_cache(sbi);
4853 f2fs_destroy_xattr_caches(sbi);
4855 mempool_destroy(sbi->write_io_dummy);
4857 destroy_percpu_info(sbi);
4859 f2fs_destroy_iostat(sbi);
4861 for (i = 0; i < NR_PAGE_TYPE; i++)
4862 kvfree(sbi->write_io[i]);
4864 #if IS_ENABLED(CONFIG_UNICODE)
4865 utf8_unload(sb->s_encoding);
4866 sb->s_encoding = NULL;
4870 for (i = 0; i < MAXQUOTAS; i++)
4871 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
4873 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
4878 if (sbi->s_chksum_driver)
4879 crypto_free_shash(sbi->s_chksum_driver);
4882 /* give only one another chance */
4883 if (retry_cnt > 0 && skip_recovery) {
4885 shrink_dcache_sb(sb);
4891 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
4892 const char *dev_name, void *data)
4894 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
4897 static void kill_f2fs_super(struct super_block *sb)
4900 struct f2fs_sb_info *sbi = F2FS_SB(sb);
4902 set_sbi_flag(sbi, SBI_IS_CLOSE);
4903 f2fs_stop_gc_thread(sbi);
4904 f2fs_stop_discard_thread(sbi);
4906 #ifdef CONFIG_F2FS_FS_COMPRESSION
4908 * latter evict_inode() can bypass checking and invalidating
4909 * compress inode cache.
4911 if (test_opt(sbi, COMPRESS_CACHE))
4912 truncate_inode_pages_final(COMPRESS_MAPPING(sbi));
4915 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
4916 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4917 struct cp_control cpc = {
4918 .reason = CP_UMOUNT,
4920 stat_inc_cp_call_count(sbi, TOTAL_CALL);
4921 f2fs_write_checkpoint(sbi, &cpc);
4924 if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
4925 sb->s_flags &= ~SB_RDONLY;
4927 kill_block_super(sb);
4930 static struct file_system_type f2fs_fs_type = {
4931 .owner = THIS_MODULE,
4933 .mount = f2fs_mount,
4934 .kill_sb = kill_f2fs_super,
4935 .fs_flags = FS_REQUIRES_DEV | FS_ALLOW_IDMAP,
4937 MODULE_ALIAS_FS("f2fs");
4939 static int __init init_inodecache(void)
4941 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
4942 sizeof(struct f2fs_inode_info), 0,
4943 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
4944 return f2fs_inode_cachep ? 0 : -ENOMEM;
4947 static void destroy_inodecache(void)
4950 * Make sure all delayed rcu free inodes are flushed before we
4954 kmem_cache_destroy(f2fs_inode_cachep);
4957 static int __init init_f2fs_fs(void)
4961 if (PAGE_SIZE != F2FS_BLKSIZE) {
4962 printk("F2FS not supported on PAGE_SIZE(%lu) != BLOCK_SIZE(%lu)\n",
4963 PAGE_SIZE, F2FS_BLKSIZE);
4967 err = init_inodecache();
4970 err = f2fs_create_node_manager_caches();
4972 goto free_inodecache;
4973 err = f2fs_create_segment_manager_caches();
4975 goto free_node_manager_caches;
4976 err = f2fs_create_checkpoint_caches();
4978 goto free_segment_manager_caches;
4979 err = f2fs_create_recovery_cache();
4981 goto free_checkpoint_caches;
4982 err = f2fs_create_extent_cache();
4984 goto free_recovery_cache;
4985 err = f2fs_create_garbage_collection_cache();
4987 goto free_extent_cache;
4988 err = f2fs_init_sysfs();
4990 goto free_garbage_collection_cache;
4991 err = f2fs_init_shrinker();
4994 err = register_filesystem(&f2fs_fs_type);
4997 f2fs_create_root_stats();
4998 err = f2fs_init_post_read_processing();
5000 goto free_root_stats;
5001 err = f2fs_init_iostat_processing();
5003 goto free_post_read;
5004 err = f2fs_init_bio_entry_cache();
5007 err = f2fs_init_bioset();
5009 goto free_bio_entry_cache;
5010 err = f2fs_init_compress_mempool();
5013 err = f2fs_init_compress_cache();
5015 goto free_compress_mempool;
5016 err = f2fs_create_casefold_cache();
5018 goto free_compress_cache;
5020 free_compress_cache:
5021 f2fs_destroy_compress_cache();
5022 free_compress_mempool:
5023 f2fs_destroy_compress_mempool();
5025 f2fs_destroy_bioset();
5026 free_bio_entry_cache:
5027 f2fs_destroy_bio_entry_cache();
5029 f2fs_destroy_iostat_processing();
5031 f2fs_destroy_post_read_processing();
5033 f2fs_destroy_root_stats();
5034 unregister_filesystem(&f2fs_fs_type);
5036 f2fs_exit_shrinker();
5039 free_garbage_collection_cache:
5040 f2fs_destroy_garbage_collection_cache();
5042 f2fs_destroy_extent_cache();
5043 free_recovery_cache:
5044 f2fs_destroy_recovery_cache();
5045 free_checkpoint_caches:
5046 f2fs_destroy_checkpoint_caches();
5047 free_segment_manager_caches:
5048 f2fs_destroy_segment_manager_caches();
5049 free_node_manager_caches:
5050 f2fs_destroy_node_manager_caches();
5052 destroy_inodecache();
5057 static void __exit exit_f2fs_fs(void)
5059 f2fs_destroy_casefold_cache();
5060 f2fs_destroy_compress_cache();
5061 f2fs_destroy_compress_mempool();
5062 f2fs_destroy_bioset();
5063 f2fs_destroy_bio_entry_cache();
5064 f2fs_destroy_iostat_processing();
5065 f2fs_destroy_post_read_processing();
5066 f2fs_destroy_root_stats();
5067 unregister_filesystem(&f2fs_fs_type);
5068 f2fs_exit_shrinker();
5070 f2fs_destroy_garbage_collection_cache();
5071 f2fs_destroy_extent_cache();
5072 f2fs_destroy_recovery_cache();
5073 f2fs_destroy_checkpoint_caches();
5074 f2fs_destroy_segment_manager_caches();
5075 f2fs_destroy_node_manager_caches();
5076 destroy_inodecache();
5079 module_init(init_f2fs_fs)
5080 module_exit(exit_f2fs_fs)
5082 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
5083 MODULE_DESCRIPTION("Flash Friendly File System");
5084 MODULE_LICENSE("GPL");
5085 MODULE_SOFTDEP("pre: crc32");