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",
66 void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
69 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
72 atomic_set(&ffi->inject_ops, 0);
73 ffi->inject_rate = rate;
77 ffi->inject_type = type;
80 memset(ffi, 0, sizeof(struct f2fs_fault_info));
84 /* f2fs-wide shrinker description */
85 static struct shrinker f2fs_shrinker_info = {
86 .scan_objects = f2fs_shrink_scan,
87 .count_objects = f2fs_shrink_count,
88 .seeks = DEFAULT_SEEKS,
93 Opt_disable_roll_forward,
104 Opt_disable_ext_identify,
107 Opt_inline_xattr_size,
144 Opt_test_dummy_encryption,
146 Opt_checkpoint_disable,
147 Opt_checkpoint_disable_cap,
148 Opt_checkpoint_disable_cap_perc,
149 Opt_checkpoint_enable,
150 Opt_checkpoint_merge,
151 Opt_nocheckpoint_merge,
152 Opt_compress_algorithm,
153 Opt_compress_log_size,
154 Opt_compress_extension,
155 Opt_nocompress_extension,
167 static match_table_t f2fs_tokens = {
168 {Opt_gc_background, "background_gc=%s"},
169 {Opt_disable_roll_forward, "disable_roll_forward"},
170 {Opt_norecovery, "norecovery"},
171 {Opt_discard, "discard"},
172 {Opt_nodiscard, "nodiscard"},
173 {Opt_noheap, "no_heap"},
175 {Opt_user_xattr, "user_xattr"},
176 {Opt_nouser_xattr, "nouser_xattr"},
178 {Opt_noacl, "noacl"},
179 {Opt_active_logs, "active_logs=%u"},
180 {Opt_disable_ext_identify, "disable_ext_identify"},
181 {Opt_inline_xattr, "inline_xattr"},
182 {Opt_noinline_xattr, "noinline_xattr"},
183 {Opt_inline_xattr_size, "inline_xattr_size=%u"},
184 {Opt_inline_data, "inline_data"},
185 {Opt_inline_dentry, "inline_dentry"},
186 {Opt_noinline_dentry, "noinline_dentry"},
187 {Opt_flush_merge, "flush_merge"},
188 {Opt_noflush_merge, "noflush_merge"},
189 {Opt_nobarrier, "nobarrier"},
190 {Opt_fastboot, "fastboot"},
191 {Opt_extent_cache, "extent_cache"},
192 {Opt_noextent_cache, "noextent_cache"},
193 {Opt_noinline_data, "noinline_data"},
194 {Opt_data_flush, "data_flush"},
195 {Opt_reserve_root, "reserve_root=%u"},
196 {Opt_resgid, "resgid=%u"},
197 {Opt_resuid, "resuid=%u"},
198 {Opt_mode, "mode=%s"},
199 {Opt_io_size_bits, "io_bits=%u"},
200 {Opt_fault_injection, "fault_injection=%u"},
201 {Opt_fault_type, "fault_type=%u"},
202 {Opt_lazytime, "lazytime"},
203 {Opt_nolazytime, "nolazytime"},
204 {Opt_quota, "quota"},
205 {Opt_noquota, "noquota"},
206 {Opt_usrquota, "usrquota"},
207 {Opt_grpquota, "grpquota"},
208 {Opt_prjquota, "prjquota"},
209 {Opt_usrjquota, "usrjquota=%s"},
210 {Opt_grpjquota, "grpjquota=%s"},
211 {Opt_prjjquota, "prjjquota=%s"},
212 {Opt_offusrjquota, "usrjquota="},
213 {Opt_offgrpjquota, "grpjquota="},
214 {Opt_offprjjquota, "prjjquota="},
215 {Opt_jqfmt_vfsold, "jqfmt=vfsold"},
216 {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
217 {Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
218 {Opt_alloc, "alloc_mode=%s"},
219 {Opt_fsync, "fsync_mode=%s"},
220 {Opt_test_dummy_encryption, "test_dummy_encryption=%s"},
221 {Opt_test_dummy_encryption, "test_dummy_encryption"},
222 {Opt_inlinecrypt, "inlinecrypt"},
223 {Opt_checkpoint_disable, "checkpoint=disable"},
224 {Opt_checkpoint_disable_cap, "checkpoint=disable:%u"},
225 {Opt_checkpoint_disable_cap_perc, "checkpoint=disable:%u%%"},
226 {Opt_checkpoint_enable, "checkpoint=enable"},
227 {Opt_checkpoint_merge, "checkpoint_merge"},
228 {Opt_nocheckpoint_merge, "nocheckpoint_merge"},
229 {Opt_compress_algorithm, "compress_algorithm=%s"},
230 {Opt_compress_log_size, "compress_log_size=%u"},
231 {Opt_compress_extension, "compress_extension=%s"},
232 {Opt_nocompress_extension, "nocompress_extension=%s"},
233 {Opt_compress_chksum, "compress_chksum"},
234 {Opt_compress_mode, "compress_mode=%s"},
235 {Opt_compress_cache, "compress_cache"},
237 {Opt_gc_merge, "gc_merge"},
238 {Opt_nogc_merge, "nogc_merge"},
239 {Opt_discard_unit, "discard_unit=%s"},
240 {Opt_memory_mode, "memory=%s"},
244 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...)
246 struct va_format vaf;
252 level = printk_get_level(fmt);
253 vaf.fmt = printk_skip_level(fmt);
255 printk("%c%cF2FS-fs (%s): %pV\n",
256 KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
261 #if IS_ENABLED(CONFIG_UNICODE)
262 static const struct f2fs_sb_encodings {
265 unsigned int version;
266 } f2fs_sb_encoding_map[] = {
267 {F2FS_ENC_UTF8_12_1, "utf8", UNICODE_AGE(12, 1, 0)},
270 static const struct f2fs_sb_encodings *
271 f2fs_sb_read_encoding(const struct f2fs_super_block *sb)
273 __u16 magic = le16_to_cpu(sb->s_encoding);
276 for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
277 if (magic == f2fs_sb_encoding_map[i].magic)
278 return &f2fs_sb_encoding_map[i];
283 struct kmem_cache *f2fs_cf_name_slab;
284 static int __init f2fs_create_casefold_cache(void)
286 f2fs_cf_name_slab = f2fs_kmem_cache_create("f2fs_casefolded_name",
288 if (!f2fs_cf_name_slab)
293 static void f2fs_destroy_casefold_cache(void)
295 kmem_cache_destroy(f2fs_cf_name_slab);
298 static int __init f2fs_create_casefold_cache(void) { return 0; }
299 static void f2fs_destroy_casefold_cache(void) { }
302 static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
304 block_t limit = min((sbi->user_block_count << 1) / 1000,
305 sbi->user_block_count - sbi->reserved_blocks);
308 if (test_opt(sbi, RESERVE_ROOT) &&
309 F2FS_OPTION(sbi).root_reserved_blocks > limit) {
310 F2FS_OPTION(sbi).root_reserved_blocks = limit;
311 f2fs_info(sbi, "Reduce reserved blocks for root = %u",
312 F2FS_OPTION(sbi).root_reserved_blocks);
314 if (!test_opt(sbi, RESERVE_ROOT) &&
315 (!uid_eq(F2FS_OPTION(sbi).s_resuid,
316 make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
317 !gid_eq(F2FS_OPTION(sbi).s_resgid,
318 make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
319 f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
320 from_kuid_munged(&init_user_ns,
321 F2FS_OPTION(sbi).s_resuid),
322 from_kgid_munged(&init_user_ns,
323 F2FS_OPTION(sbi).s_resgid));
326 static inline int adjust_reserved_segment(struct f2fs_sb_info *sbi)
328 unsigned int sec_blks = sbi->blocks_per_seg * sbi->segs_per_sec;
329 unsigned int avg_vblocks;
330 unsigned int wanted_reserved_segments;
331 block_t avail_user_block_count;
333 if (!F2FS_IO_ALIGNED(sbi))
336 /* average valid block count in section in worst case */
337 avg_vblocks = sec_blks / F2FS_IO_SIZE(sbi);
340 * we need enough free space when migrating one section in worst case
342 wanted_reserved_segments = (F2FS_IO_SIZE(sbi) / avg_vblocks) *
343 reserved_segments(sbi);
344 wanted_reserved_segments -= reserved_segments(sbi);
346 avail_user_block_count = sbi->user_block_count -
347 sbi->current_reserved_blocks -
348 F2FS_OPTION(sbi).root_reserved_blocks;
350 if (wanted_reserved_segments * sbi->blocks_per_seg >
351 avail_user_block_count) {
352 f2fs_err(sbi, "IO align feature can't grab additional reserved segment: %u, available segments: %u",
353 wanted_reserved_segments,
354 avail_user_block_count >> sbi->log_blocks_per_seg);
358 SM_I(sbi)->additional_reserved_segments = wanted_reserved_segments;
360 f2fs_info(sbi, "IO align feature needs additional reserved segment: %u",
361 wanted_reserved_segments);
366 static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
368 if (!F2FS_OPTION(sbi).unusable_cap_perc)
371 if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
372 F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
374 F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
375 F2FS_OPTION(sbi).unusable_cap_perc;
377 f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
378 F2FS_OPTION(sbi).unusable_cap,
379 F2FS_OPTION(sbi).unusable_cap_perc);
382 static void init_once(void *foo)
384 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
386 inode_init_once(&fi->vfs_inode);
390 static const char * const quotatypes[] = INITQFNAMES;
391 #define QTYPE2NAME(t) (quotatypes[t])
392 static int f2fs_set_qf_name(struct super_block *sb, int qtype,
395 struct f2fs_sb_info *sbi = F2FS_SB(sb);
399 if (sb_any_quota_loaded(sb) && !F2FS_OPTION(sbi).s_qf_names[qtype]) {
400 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
403 if (f2fs_sb_has_quota_ino(sbi)) {
404 f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
408 qname = match_strdup(args);
410 f2fs_err(sbi, "Not enough memory for storing quotafile name");
413 if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
414 if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
417 f2fs_err(sbi, "%s quota file already specified",
421 if (strchr(qname, '/')) {
422 f2fs_err(sbi, "quotafile must be on filesystem root");
425 F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
433 static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
435 struct f2fs_sb_info *sbi = F2FS_SB(sb);
437 if (sb_any_quota_loaded(sb) && F2FS_OPTION(sbi).s_qf_names[qtype]) {
438 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
441 kfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
442 F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
446 static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
449 * We do the test below only for project quotas. 'usrquota' and
450 * 'grpquota' mount options are allowed even without quota feature
451 * to support legacy quotas in quota files.
453 if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi)) {
454 f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
457 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
458 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
459 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) {
460 if (test_opt(sbi, USRQUOTA) &&
461 F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
462 clear_opt(sbi, USRQUOTA);
464 if (test_opt(sbi, GRPQUOTA) &&
465 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
466 clear_opt(sbi, GRPQUOTA);
468 if (test_opt(sbi, PRJQUOTA) &&
469 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
470 clear_opt(sbi, PRJQUOTA);
472 if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
473 test_opt(sbi, PRJQUOTA)) {
474 f2fs_err(sbi, "old and new quota format mixing");
478 if (!F2FS_OPTION(sbi).s_jquota_fmt) {
479 f2fs_err(sbi, "journaled quota format not specified");
484 if (f2fs_sb_has_quota_ino(sbi) && F2FS_OPTION(sbi).s_jquota_fmt) {
485 f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
486 F2FS_OPTION(sbi).s_jquota_fmt = 0;
492 static int f2fs_set_test_dummy_encryption(struct super_block *sb,
494 const substring_t *arg,
497 struct f2fs_sb_info *sbi = F2FS_SB(sb);
498 struct fs_parameter param = {
499 .type = fs_value_is_string,
500 .string = arg->from ? arg->from : "",
502 struct fscrypt_dummy_policy *policy =
503 &F2FS_OPTION(sbi).dummy_enc_policy;
506 if (!IS_ENABLED(CONFIG_FS_ENCRYPTION)) {
507 f2fs_warn(sbi, "test_dummy_encryption option not supported");
511 if (!f2fs_sb_has_encrypt(sbi)) {
512 f2fs_err(sbi, "Encrypt feature is off");
517 * This mount option is just for testing, and it's not worthwhile to
518 * implement the extra complexity (e.g. RCU protection) that would be
519 * needed to allow it to be set or changed during remount. We do allow
520 * it to be specified during remount, but only if there is no change.
522 if (is_remount && !fscrypt_is_dummy_policy_set(policy)) {
523 f2fs_warn(sbi, "Can't set test_dummy_encryption on remount");
527 err = fscrypt_parse_test_dummy_encryption(¶m, policy);
531 "Can't change test_dummy_encryption on remount");
532 else if (err == -EINVAL)
533 f2fs_warn(sbi, "Value of option \"%s\" is unrecognized",
536 f2fs_warn(sbi, "Error processing option \"%s\" [%d]",
540 err = fscrypt_add_test_dummy_key(sb, policy);
542 f2fs_warn(sbi, "Error adding test dummy encryption key [%d]",
546 f2fs_warn(sbi, "Test dummy encryption mode enabled");
550 #ifdef CONFIG_F2FS_FS_COMPRESSION
552 * 1. The same extension name cannot not appear in both compress and non-compress extension
554 * 2. If the compress extension specifies all files, the types specified by the non-compress
555 * extension will be treated as special cases and will not be compressed.
556 * 3. Don't allow the non-compress extension specifies all files.
558 static int f2fs_test_compress_extension(struct f2fs_sb_info *sbi)
560 unsigned char (*ext)[F2FS_EXTENSION_LEN];
561 unsigned char (*noext)[F2FS_EXTENSION_LEN];
562 int ext_cnt, noext_cnt, index = 0, no_index = 0;
564 ext = F2FS_OPTION(sbi).extensions;
565 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
566 noext = F2FS_OPTION(sbi).noextensions;
567 noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
572 for (no_index = 0; no_index < noext_cnt; no_index++) {
573 if (!strcasecmp("*", noext[no_index])) {
574 f2fs_info(sbi, "Don't allow the nocompress extension specifies all files");
577 for (index = 0; index < ext_cnt; index++) {
578 if (!strcasecmp(ext[index], noext[no_index])) {
579 f2fs_info(sbi, "Don't allow the same extension %s appear in both compress and nocompress extension",
588 #ifdef CONFIG_F2FS_FS_LZ4
589 static int f2fs_set_lz4hc_level(struct f2fs_sb_info *sbi, const char *str)
591 #ifdef CONFIG_F2FS_FS_LZ4HC
595 if (strlen(str) == 3) {
596 F2FS_OPTION(sbi).compress_level = 0;
600 #ifdef CONFIG_F2FS_FS_LZ4HC
604 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
607 if (kstrtouint(str + 1, 10, &level))
610 if (level < LZ4HC_MIN_CLEVEL || level > LZ4HC_MAX_CLEVEL) {
611 f2fs_info(sbi, "invalid lz4hc compress level: %d", level);
615 F2FS_OPTION(sbi).compress_level = level;
618 f2fs_info(sbi, "kernel doesn't support lz4hc compression");
624 #ifdef CONFIG_F2FS_FS_ZSTD
625 static int f2fs_set_zstd_level(struct f2fs_sb_info *sbi, const char *str)
630 if (strlen(str) == len) {
631 F2FS_OPTION(sbi).compress_level = 0;
638 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
641 if (kstrtouint(str + 1, 10, &level))
644 if (!level || level > zstd_max_clevel()) {
645 f2fs_info(sbi, "invalid zstd compress level: %d", level);
649 F2FS_OPTION(sbi).compress_level = level;
655 static int parse_options(struct super_block *sb, char *options, bool is_remount)
657 struct f2fs_sb_info *sbi = F2FS_SB(sb);
658 substring_t args[MAX_OPT_ARGS];
659 #ifdef CONFIG_F2FS_FS_COMPRESSION
660 unsigned char (*ext)[F2FS_EXTENSION_LEN];
661 unsigned char (*noext)[F2FS_EXTENSION_LEN];
662 int ext_cnt, noext_cnt;
673 while ((p = strsep(&options, ",")) != NULL) {
679 * Initialize args struct so we know whether arg was
680 * found; some options take optional arguments.
682 args[0].to = args[0].from = NULL;
683 token = match_token(p, f2fs_tokens, args);
686 case Opt_gc_background:
687 name = match_strdup(&args[0]);
691 if (!strcmp(name, "on")) {
692 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
693 } else if (!strcmp(name, "off")) {
694 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_OFF;
695 } else if (!strcmp(name, "sync")) {
696 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_SYNC;
703 case Opt_disable_roll_forward:
704 set_opt(sbi, DISABLE_ROLL_FORWARD);
707 /* this option mounts f2fs with ro */
708 set_opt(sbi, NORECOVERY);
709 if (!f2fs_readonly(sb))
713 if (!f2fs_hw_support_discard(sbi)) {
714 f2fs_warn(sbi, "device does not support discard");
717 set_opt(sbi, DISCARD);
720 if (f2fs_hw_should_discard(sbi)) {
721 f2fs_warn(sbi, "discard is required for zoned block devices");
724 clear_opt(sbi, DISCARD);
727 set_opt(sbi, NOHEAP);
730 clear_opt(sbi, NOHEAP);
732 #ifdef CONFIG_F2FS_FS_XATTR
734 set_opt(sbi, XATTR_USER);
736 case Opt_nouser_xattr:
737 clear_opt(sbi, XATTR_USER);
739 case Opt_inline_xattr:
740 set_opt(sbi, INLINE_XATTR);
742 case Opt_noinline_xattr:
743 clear_opt(sbi, INLINE_XATTR);
745 case Opt_inline_xattr_size:
746 if (args->from && match_int(args, &arg))
748 set_opt(sbi, INLINE_XATTR_SIZE);
749 F2FS_OPTION(sbi).inline_xattr_size = arg;
753 f2fs_info(sbi, "user_xattr options not supported");
755 case Opt_nouser_xattr:
756 f2fs_info(sbi, "nouser_xattr options not supported");
758 case Opt_inline_xattr:
759 f2fs_info(sbi, "inline_xattr options not supported");
761 case Opt_noinline_xattr:
762 f2fs_info(sbi, "noinline_xattr options not supported");
765 #ifdef CONFIG_F2FS_FS_POSIX_ACL
767 set_opt(sbi, POSIX_ACL);
770 clear_opt(sbi, POSIX_ACL);
774 f2fs_info(sbi, "acl options not supported");
777 f2fs_info(sbi, "noacl options not supported");
780 case Opt_active_logs:
781 if (args->from && match_int(args, &arg))
783 if (arg != 2 && arg != 4 &&
784 arg != NR_CURSEG_PERSIST_TYPE)
786 F2FS_OPTION(sbi).active_logs = arg;
788 case Opt_disable_ext_identify:
789 set_opt(sbi, DISABLE_EXT_IDENTIFY);
791 case Opt_inline_data:
792 set_opt(sbi, INLINE_DATA);
794 case Opt_inline_dentry:
795 set_opt(sbi, INLINE_DENTRY);
797 case Opt_noinline_dentry:
798 clear_opt(sbi, INLINE_DENTRY);
800 case Opt_flush_merge:
801 set_opt(sbi, FLUSH_MERGE);
803 case Opt_noflush_merge:
804 clear_opt(sbi, FLUSH_MERGE);
807 set_opt(sbi, NOBARRIER);
810 set_opt(sbi, FASTBOOT);
812 case Opt_extent_cache:
813 set_opt(sbi, EXTENT_CACHE);
815 case Opt_noextent_cache:
816 clear_opt(sbi, EXTENT_CACHE);
818 case Opt_noinline_data:
819 clear_opt(sbi, INLINE_DATA);
822 set_opt(sbi, DATA_FLUSH);
824 case Opt_reserve_root:
825 if (args->from && match_int(args, &arg))
827 if (test_opt(sbi, RESERVE_ROOT)) {
828 f2fs_info(sbi, "Preserve previous reserve_root=%u",
829 F2FS_OPTION(sbi).root_reserved_blocks);
831 F2FS_OPTION(sbi).root_reserved_blocks = arg;
832 set_opt(sbi, RESERVE_ROOT);
836 if (args->from && match_int(args, &arg))
838 uid = make_kuid(current_user_ns(), arg);
839 if (!uid_valid(uid)) {
840 f2fs_err(sbi, "Invalid uid value %d", arg);
843 F2FS_OPTION(sbi).s_resuid = uid;
846 if (args->from && match_int(args, &arg))
848 gid = make_kgid(current_user_ns(), arg);
849 if (!gid_valid(gid)) {
850 f2fs_err(sbi, "Invalid gid value %d", arg);
853 F2FS_OPTION(sbi).s_resgid = gid;
856 name = match_strdup(&args[0]);
860 if (!strcmp(name, "adaptive")) {
861 if (f2fs_sb_has_blkzoned(sbi)) {
862 f2fs_warn(sbi, "adaptive mode is not allowed with zoned block device feature");
866 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
867 } else if (!strcmp(name, "lfs")) {
868 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
869 } else if (!strcmp(name, "fragment:segment")) {
870 F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_SEG;
871 } else if (!strcmp(name, "fragment:block")) {
872 F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_BLK;
879 case Opt_io_size_bits:
880 if (args->from && match_int(args, &arg))
882 if (arg <= 0 || arg > __ilog2_u32(BIO_MAX_VECS)) {
883 f2fs_warn(sbi, "Not support %d, larger than %d",
884 1 << arg, BIO_MAX_VECS);
887 F2FS_OPTION(sbi).write_io_size_bits = arg;
889 #ifdef CONFIG_F2FS_FAULT_INJECTION
890 case Opt_fault_injection:
891 if (args->from && match_int(args, &arg))
893 f2fs_build_fault_attr(sbi, arg, F2FS_ALL_FAULT_TYPE);
894 set_opt(sbi, FAULT_INJECTION);
898 if (args->from && match_int(args, &arg))
900 f2fs_build_fault_attr(sbi, 0, arg);
901 set_opt(sbi, FAULT_INJECTION);
904 case Opt_fault_injection:
905 f2fs_info(sbi, "fault_injection options not supported");
909 f2fs_info(sbi, "fault_type options not supported");
913 sb->s_flags |= SB_LAZYTIME;
916 sb->s_flags &= ~SB_LAZYTIME;
921 set_opt(sbi, USRQUOTA);
924 set_opt(sbi, GRPQUOTA);
927 set_opt(sbi, PRJQUOTA);
930 ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
935 ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
940 ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
944 case Opt_offusrjquota:
945 ret = f2fs_clear_qf_name(sb, USRQUOTA);
949 case Opt_offgrpjquota:
950 ret = f2fs_clear_qf_name(sb, GRPQUOTA);
954 case Opt_offprjjquota:
955 ret = f2fs_clear_qf_name(sb, PRJQUOTA);
959 case Opt_jqfmt_vfsold:
960 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
962 case Opt_jqfmt_vfsv0:
963 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
965 case Opt_jqfmt_vfsv1:
966 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
969 clear_opt(sbi, QUOTA);
970 clear_opt(sbi, USRQUOTA);
971 clear_opt(sbi, GRPQUOTA);
972 clear_opt(sbi, PRJQUOTA);
982 case Opt_offusrjquota:
983 case Opt_offgrpjquota:
984 case Opt_offprjjquota:
985 case Opt_jqfmt_vfsold:
986 case Opt_jqfmt_vfsv0:
987 case Opt_jqfmt_vfsv1:
989 f2fs_info(sbi, "quota operations not supported");
993 name = match_strdup(&args[0]);
997 if (!strcmp(name, "default")) {
998 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
999 } else if (!strcmp(name, "reuse")) {
1000 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
1008 name = match_strdup(&args[0]);
1011 if (!strcmp(name, "posix")) {
1012 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
1013 } else if (!strcmp(name, "strict")) {
1014 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_STRICT;
1015 } else if (!strcmp(name, "nobarrier")) {
1016 F2FS_OPTION(sbi).fsync_mode =
1017 FSYNC_MODE_NOBARRIER;
1024 case Opt_test_dummy_encryption:
1025 ret = f2fs_set_test_dummy_encryption(sb, p, &args[0],
1030 case Opt_inlinecrypt:
1031 #ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
1032 sb->s_flags |= SB_INLINECRYPT;
1034 f2fs_info(sbi, "inline encryption not supported");
1037 case Opt_checkpoint_disable_cap_perc:
1038 if (args->from && match_int(args, &arg))
1040 if (arg < 0 || arg > 100)
1042 F2FS_OPTION(sbi).unusable_cap_perc = arg;
1043 set_opt(sbi, DISABLE_CHECKPOINT);
1045 case Opt_checkpoint_disable_cap:
1046 if (args->from && match_int(args, &arg))
1048 F2FS_OPTION(sbi).unusable_cap = arg;
1049 set_opt(sbi, DISABLE_CHECKPOINT);
1051 case Opt_checkpoint_disable:
1052 set_opt(sbi, DISABLE_CHECKPOINT);
1054 case Opt_checkpoint_enable:
1055 clear_opt(sbi, DISABLE_CHECKPOINT);
1057 case Opt_checkpoint_merge:
1058 set_opt(sbi, MERGE_CHECKPOINT);
1060 case Opt_nocheckpoint_merge:
1061 clear_opt(sbi, MERGE_CHECKPOINT);
1063 #ifdef CONFIG_F2FS_FS_COMPRESSION
1064 case Opt_compress_algorithm:
1065 if (!f2fs_sb_has_compression(sbi)) {
1066 f2fs_info(sbi, "Image doesn't support compression");
1069 name = match_strdup(&args[0]);
1072 if (!strcmp(name, "lzo")) {
1073 #ifdef CONFIG_F2FS_FS_LZO
1074 F2FS_OPTION(sbi).compress_level = 0;
1075 F2FS_OPTION(sbi).compress_algorithm =
1078 f2fs_info(sbi, "kernel doesn't support lzo compression");
1080 } else if (!strncmp(name, "lz4", 3)) {
1081 #ifdef CONFIG_F2FS_FS_LZ4
1082 ret = f2fs_set_lz4hc_level(sbi, name);
1087 F2FS_OPTION(sbi).compress_algorithm =
1090 f2fs_info(sbi, "kernel doesn't support lz4 compression");
1092 } else if (!strncmp(name, "zstd", 4)) {
1093 #ifdef CONFIG_F2FS_FS_ZSTD
1094 ret = f2fs_set_zstd_level(sbi, name);
1099 F2FS_OPTION(sbi).compress_algorithm =
1102 f2fs_info(sbi, "kernel doesn't support zstd compression");
1104 } else if (!strcmp(name, "lzo-rle")) {
1105 #ifdef CONFIG_F2FS_FS_LZORLE
1106 F2FS_OPTION(sbi).compress_level = 0;
1107 F2FS_OPTION(sbi).compress_algorithm =
1110 f2fs_info(sbi, "kernel doesn't support lzorle compression");
1118 case Opt_compress_log_size:
1119 if (!f2fs_sb_has_compression(sbi)) {
1120 f2fs_info(sbi, "Image doesn't support compression");
1123 if (args->from && match_int(args, &arg))
1125 if (arg < MIN_COMPRESS_LOG_SIZE ||
1126 arg > MAX_COMPRESS_LOG_SIZE) {
1128 "Compress cluster log size is out of range");
1131 F2FS_OPTION(sbi).compress_log_size = arg;
1133 case Opt_compress_extension:
1134 if (!f2fs_sb_has_compression(sbi)) {
1135 f2fs_info(sbi, "Image doesn't support compression");
1138 name = match_strdup(&args[0]);
1142 ext = F2FS_OPTION(sbi).extensions;
1143 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
1145 if (strlen(name) >= F2FS_EXTENSION_LEN ||
1146 ext_cnt >= COMPRESS_EXT_NUM) {
1148 "invalid extension length/number");
1153 strcpy(ext[ext_cnt], name);
1154 F2FS_OPTION(sbi).compress_ext_cnt++;
1157 case Opt_nocompress_extension:
1158 if (!f2fs_sb_has_compression(sbi)) {
1159 f2fs_info(sbi, "Image doesn't support compression");
1162 name = match_strdup(&args[0]);
1166 noext = F2FS_OPTION(sbi).noextensions;
1167 noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
1169 if (strlen(name) >= F2FS_EXTENSION_LEN ||
1170 noext_cnt >= COMPRESS_EXT_NUM) {
1172 "invalid extension length/number");
1177 strcpy(noext[noext_cnt], name);
1178 F2FS_OPTION(sbi).nocompress_ext_cnt++;
1181 case Opt_compress_chksum:
1182 F2FS_OPTION(sbi).compress_chksum = true;
1184 case Opt_compress_mode:
1185 name = match_strdup(&args[0]);
1188 if (!strcmp(name, "fs")) {
1189 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
1190 } else if (!strcmp(name, "user")) {
1191 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_USER;
1198 case Opt_compress_cache:
1199 set_opt(sbi, COMPRESS_CACHE);
1202 case Opt_compress_algorithm:
1203 case Opt_compress_log_size:
1204 case Opt_compress_extension:
1205 case Opt_nocompress_extension:
1206 case Opt_compress_chksum:
1207 case Opt_compress_mode:
1208 case Opt_compress_cache:
1209 f2fs_info(sbi, "compression options not supported");
1216 set_opt(sbi, GC_MERGE);
1218 case Opt_nogc_merge:
1219 clear_opt(sbi, GC_MERGE);
1221 case Opt_discard_unit:
1222 name = match_strdup(&args[0]);
1225 if (!strcmp(name, "block")) {
1226 F2FS_OPTION(sbi).discard_unit =
1228 } else if (!strcmp(name, "segment")) {
1229 F2FS_OPTION(sbi).discard_unit =
1230 DISCARD_UNIT_SEGMENT;
1231 } else if (!strcmp(name, "section")) {
1232 F2FS_OPTION(sbi).discard_unit =
1233 DISCARD_UNIT_SECTION;
1240 case Opt_memory_mode:
1241 name = match_strdup(&args[0]);
1244 if (!strcmp(name, "normal")) {
1245 F2FS_OPTION(sbi).memory_mode =
1247 } else if (!strcmp(name, "low")) {
1248 F2FS_OPTION(sbi).memory_mode =
1257 f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
1264 if (f2fs_check_quota_options(sbi))
1267 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sbi->sb)) {
1268 f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1271 if (f2fs_sb_has_project_quota(sbi) && !f2fs_readonly(sbi->sb)) {
1272 f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1276 #if !IS_ENABLED(CONFIG_UNICODE)
1277 if (f2fs_sb_has_casefold(sbi)) {
1279 "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
1284 * The BLKZONED feature indicates that the drive was formatted with
1285 * zone alignment optimization. This is optional for host-aware
1286 * devices, but mandatory for host-managed zoned block devices.
1288 #ifndef CONFIG_BLK_DEV_ZONED
1289 if (f2fs_sb_has_blkzoned(sbi)) {
1290 f2fs_err(sbi, "Zoned block device support is not enabled");
1294 if (f2fs_sb_has_blkzoned(sbi)) {
1295 if (F2FS_OPTION(sbi).discard_unit !=
1296 DISCARD_UNIT_SECTION) {
1297 f2fs_info(sbi, "Zoned block device doesn't need small discard, set discard_unit=section by default");
1298 F2FS_OPTION(sbi).discard_unit =
1299 DISCARD_UNIT_SECTION;
1303 #ifdef CONFIG_F2FS_FS_COMPRESSION
1304 if (f2fs_test_compress_extension(sbi)) {
1305 f2fs_err(sbi, "invalid compress or nocompress extension");
1310 if (F2FS_IO_SIZE_BITS(sbi) && !f2fs_lfs_mode(sbi)) {
1311 f2fs_err(sbi, "Should set mode=lfs with %uKB-sized IO",
1312 F2FS_IO_SIZE_KB(sbi));
1316 if (test_opt(sbi, INLINE_XATTR_SIZE)) {
1317 int min_size, max_size;
1319 if (!f2fs_sb_has_extra_attr(sbi) ||
1320 !f2fs_sb_has_flexible_inline_xattr(sbi)) {
1321 f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
1324 if (!test_opt(sbi, INLINE_XATTR)) {
1325 f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
1329 min_size = sizeof(struct f2fs_xattr_header) / sizeof(__le32);
1330 max_size = MAX_INLINE_XATTR_SIZE;
1332 if (F2FS_OPTION(sbi).inline_xattr_size < min_size ||
1333 F2FS_OPTION(sbi).inline_xattr_size > max_size) {
1334 f2fs_err(sbi, "inline xattr size is out of range: %d ~ %d",
1335 min_size, max_size);
1340 if (test_opt(sbi, DISABLE_CHECKPOINT) && f2fs_lfs_mode(sbi)) {
1341 f2fs_err(sbi, "LFS not compatible with checkpoint=disable");
1345 if (f2fs_sb_has_readonly(sbi) && !f2fs_readonly(sbi->sb)) {
1346 f2fs_err(sbi, "Allow to mount readonly mode only");
1352 static struct inode *f2fs_alloc_inode(struct super_block *sb)
1354 struct f2fs_inode_info *fi;
1356 if (time_to_inject(F2FS_SB(sb), FAULT_SLAB_ALLOC)) {
1357 f2fs_show_injection_info(F2FS_SB(sb), FAULT_SLAB_ALLOC);
1361 fi = alloc_inode_sb(sb, f2fs_inode_cachep, GFP_F2FS_ZERO);
1365 init_once((void *) fi);
1367 /* Initialize f2fs-specific inode info */
1368 atomic_set(&fi->dirty_pages, 0);
1369 atomic_set(&fi->i_compr_blocks, 0);
1370 init_f2fs_rwsem(&fi->i_sem);
1371 spin_lock_init(&fi->i_size_lock);
1372 INIT_LIST_HEAD(&fi->dirty_list);
1373 INIT_LIST_HEAD(&fi->gdirty_list);
1374 init_f2fs_rwsem(&fi->i_gc_rwsem[READ]);
1375 init_f2fs_rwsem(&fi->i_gc_rwsem[WRITE]);
1376 init_f2fs_rwsem(&fi->i_xattr_sem);
1378 /* Will be used by directory only */
1379 fi->i_dir_level = F2FS_SB(sb)->dir_level;
1381 return &fi->vfs_inode;
1384 static int f2fs_drop_inode(struct inode *inode)
1386 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1390 * during filesystem shutdown, if checkpoint is disabled,
1391 * drop useless meta/node dirty pages.
1393 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1394 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1395 inode->i_ino == F2FS_META_INO(sbi)) {
1396 trace_f2fs_drop_inode(inode, 1);
1402 * This is to avoid a deadlock condition like below.
1403 * writeback_single_inode(inode)
1404 * - f2fs_write_data_page
1405 * - f2fs_gc -> iput -> evict
1406 * - inode_wait_for_writeback(inode)
1408 if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
1409 if (!inode->i_nlink && !is_bad_inode(inode)) {
1410 /* to avoid evict_inode call simultaneously */
1411 atomic_inc(&inode->i_count);
1412 spin_unlock(&inode->i_lock);
1414 f2fs_abort_atomic_write(inode, true);
1416 /* should remain fi->extent_tree for writepage */
1417 f2fs_destroy_extent_node(inode);
1419 sb_start_intwrite(inode->i_sb);
1420 f2fs_i_size_write(inode, 0);
1422 f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1423 inode, NULL, 0, DATA);
1424 truncate_inode_pages_final(inode->i_mapping);
1426 if (F2FS_HAS_BLOCKS(inode))
1427 f2fs_truncate(inode);
1429 sb_end_intwrite(inode->i_sb);
1431 spin_lock(&inode->i_lock);
1432 atomic_dec(&inode->i_count);
1434 trace_f2fs_drop_inode(inode, 0);
1437 ret = generic_drop_inode(inode);
1439 ret = fscrypt_drop_inode(inode);
1440 trace_f2fs_drop_inode(inode, ret);
1444 int f2fs_inode_dirtied(struct inode *inode, bool sync)
1446 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1449 spin_lock(&sbi->inode_lock[DIRTY_META]);
1450 if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1453 set_inode_flag(inode, FI_DIRTY_INODE);
1454 stat_inc_dirty_inode(sbi, DIRTY_META);
1456 if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1457 list_add_tail(&F2FS_I(inode)->gdirty_list,
1458 &sbi->inode_list[DIRTY_META]);
1459 inc_page_count(sbi, F2FS_DIRTY_IMETA);
1461 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1465 void f2fs_inode_synced(struct inode *inode)
1467 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1469 spin_lock(&sbi->inode_lock[DIRTY_META]);
1470 if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1471 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1474 if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1475 list_del_init(&F2FS_I(inode)->gdirty_list);
1476 dec_page_count(sbi, F2FS_DIRTY_IMETA);
1478 clear_inode_flag(inode, FI_DIRTY_INODE);
1479 clear_inode_flag(inode, FI_AUTO_RECOVER);
1480 stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1481 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1485 * f2fs_dirty_inode() is called from __mark_inode_dirty()
1487 * We should call set_dirty_inode to write the dirty inode through write_inode.
1489 static void f2fs_dirty_inode(struct inode *inode, int flags)
1491 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1493 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1494 inode->i_ino == F2FS_META_INO(sbi))
1497 if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1498 clear_inode_flag(inode, FI_AUTO_RECOVER);
1500 f2fs_inode_dirtied(inode, false);
1503 static void f2fs_free_inode(struct inode *inode)
1505 fscrypt_free_inode(inode);
1506 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1509 static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1511 percpu_counter_destroy(&sbi->total_valid_inode_count);
1512 percpu_counter_destroy(&sbi->rf_node_block_count);
1513 percpu_counter_destroy(&sbi->alloc_valid_block_count);
1516 static void destroy_device_list(struct f2fs_sb_info *sbi)
1520 for (i = 0; i < sbi->s_ndevs; i++) {
1521 blkdev_put(FDEV(i).bdev, FMODE_EXCL);
1522 #ifdef CONFIG_BLK_DEV_ZONED
1523 kvfree(FDEV(i).blkz_seq);
1529 static void f2fs_put_super(struct super_block *sb)
1531 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1535 /* unregister procfs/sysfs entries in advance to avoid race case */
1536 f2fs_unregister_sysfs(sbi);
1538 f2fs_quota_off_umount(sb);
1540 /* prevent remaining shrinker jobs */
1541 mutex_lock(&sbi->umount_mutex);
1544 * flush all issued checkpoints and stop checkpoint issue thread.
1545 * after then, all checkpoints should be done by each process context.
1547 f2fs_stop_ckpt_thread(sbi);
1550 * We don't need to do checkpoint when superblock is clean.
1551 * But, the previous checkpoint was not done by umount, it needs to do
1552 * clean checkpoint again.
1554 if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1555 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1556 struct cp_control cpc = {
1557 .reason = CP_UMOUNT,
1559 f2fs_write_checkpoint(sbi, &cpc);
1562 /* be sure to wait for any on-going discard commands */
1563 dropped = f2fs_issue_discard_timeout(sbi);
1565 if ((f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi)) &&
1566 !sbi->discard_blks && !dropped) {
1567 struct cp_control cpc = {
1568 .reason = CP_UMOUNT | CP_TRIMMED,
1570 f2fs_write_checkpoint(sbi, &cpc);
1574 * normally superblock is clean, so we need to release this.
1575 * In addition, EIO will skip do checkpoint, we need this as well.
1577 f2fs_release_ino_entry(sbi, true);
1579 f2fs_leave_shrinker(sbi);
1580 mutex_unlock(&sbi->umount_mutex);
1582 /* our cp_error case, we can wait for any writeback page */
1583 f2fs_flush_merged_writes(sbi);
1585 f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1587 f2fs_bug_on(sbi, sbi->fsync_node_num);
1589 f2fs_destroy_compress_inode(sbi);
1591 iput(sbi->node_inode);
1592 sbi->node_inode = NULL;
1594 iput(sbi->meta_inode);
1595 sbi->meta_inode = NULL;
1598 * iput() can update stat information, if f2fs_write_checkpoint()
1599 * above failed with error.
1601 f2fs_destroy_stats(sbi);
1603 /* destroy f2fs internal modules */
1604 f2fs_destroy_node_manager(sbi);
1605 f2fs_destroy_segment_manager(sbi);
1607 f2fs_destroy_post_read_wq(sbi);
1611 sb->s_fs_info = NULL;
1612 if (sbi->s_chksum_driver)
1613 crypto_free_shash(sbi->s_chksum_driver);
1614 kfree(sbi->raw_super);
1616 destroy_device_list(sbi);
1617 f2fs_destroy_page_array_cache(sbi);
1618 f2fs_destroy_xattr_caches(sbi);
1619 mempool_destroy(sbi->write_io_dummy);
1621 for (i = 0; i < MAXQUOTAS; i++)
1622 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1624 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
1625 destroy_percpu_info(sbi);
1626 f2fs_destroy_iostat(sbi);
1627 for (i = 0; i < NR_PAGE_TYPE; i++)
1628 kvfree(sbi->write_io[i]);
1629 #if IS_ENABLED(CONFIG_UNICODE)
1630 utf8_unload(sb->s_encoding);
1635 int f2fs_sync_fs(struct super_block *sb, int sync)
1637 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1640 if (unlikely(f2fs_cp_error(sbi)))
1642 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1645 trace_f2fs_sync_fs(sb, sync);
1647 if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1651 err = f2fs_issue_checkpoint(sbi);
1656 static int f2fs_freeze(struct super_block *sb)
1658 if (f2fs_readonly(sb))
1661 /* IO error happened before */
1662 if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1665 /* must be clean, since sync_filesystem() was already called */
1666 if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1669 /* ensure no checkpoint required */
1670 if (!llist_empty(&F2FS_SB(sb)->cprc_info.issue_list))
1673 /* to avoid deadlock on f2fs_evict_inode->SB_FREEZE_FS */
1674 set_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1678 static int f2fs_unfreeze(struct super_block *sb)
1680 clear_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1685 static int f2fs_statfs_project(struct super_block *sb,
1686 kprojid_t projid, struct kstatfs *buf)
1689 struct dquot *dquot;
1693 qid = make_kqid_projid(projid);
1694 dquot = dqget(sb, qid);
1696 return PTR_ERR(dquot);
1697 spin_lock(&dquot->dq_dqb_lock);
1699 limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1700 dquot->dq_dqb.dqb_bhardlimit);
1702 limit >>= sb->s_blocksize_bits;
1704 if (limit && buf->f_blocks > limit) {
1705 curblock = (dquot->dq_dqb.dqb_curspace +
1706 dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
1707 buf->f_blocks = limit;
1708 buf->f_bfree = buf->f_bavail =
1709 (buf->f_blocks > curblock) ?
1710 (buf->f_blocks - curblock) : 0;
1713 limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1714 dquot->dq_dqb.dqb_ihardlimit);
1716 if (limit && buf->f_files > limit) {
1717 buf->f_files = limit;
1719 (buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1720 (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1723 spin_unlock(&dquot->dq_dqb_lock);
1729 static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1731 struct super_block *sb = dentry->d_sb;
1732 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1733 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
1734 block_t total_count, user_block_count, start_count;
1735 u64 avail_node_count;
1736 unsigned int total_valid_node_count;
1738 total_count = le64_to_cpu(sbi->raw_super->block_count);
1739 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
1740 buf->f_type = F2FS_SUPER_MAGIC;
1741 buf->f_bsize = sbi->blocksize;
1743 buf->f_blocks = total_count - start_count;
1745 spin_lock(&sbi->stat_lock);
1747 user_block_count = sbi->user_block_count;
1748 total_valid_node_count = valid_node_count(sbi);
1749 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
1750 buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1751 sbi->current_reserved_blocks;
1753 if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1756 buf->f_bfree -= sbi->unusable_block_count;
1757 spin_unlock(&sbi->stat_lock);
1759 if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1760 buf->f_bavail = buf->f_bfree -
1761 F2FS_OPTION(sbi).root_reserved_blocks;
1765 if (avail_node_count > user_block_count) {
1766 buf->f_files = user_block_count;
1767 buf->f_ffree = buf->f_bavail;
1769 buf->f_files = avail_node_count;
1770 buf->f_ffree = min(avail_node_count - total_valid_node_count,
1774 buf->f_namelen = F2FS_NAME_LEN;
1775 buf->f_fsid = u64_to_fsid(id);
1778 if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1779 sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1780 f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1786 static inline void f2fs_show_quota_options(struct seq_file *seq,
1787 struct super_block *sb)
1790 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1792 if (F2FS_OPTION(sbi).s_jquota_fmt) {
1795 switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1806 seq_printf(seq, ",jqfmt=%s", fmtname);
1809 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1810 seq_show_option(seq, "usrjquota",
1811 F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1813 if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1814 seq_show_option(seq, "grpjquota",
1815 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1817 if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1818 seq_show_option(seq, "prjjquota",
1819 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1823 #ifdef CONFIG_F2FS_FS_COMPRESSION
1824 static inline void f2fs_show_compress_options(struct seq_file *seq,
1825 struct super_block *sb)
1827 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1831 if (!f2fs_sb_has_compression(sbi))
1834 switch (F2FS_OPTION(sbi).compress_algorithm) {
1844 case COMPRESS_LZORLE:
1845 algtype = "lzo-rle";
1848 seq_printf(seq, ",compress_algorithm=%s", algtype);
1850 if (F2FS_OPTION(sbi).compress_level)
1851 seq_printf(seq, ":%d", F2FS_OPTION(sbi).compress_level);
1853 seq_printf(seq, ",compress_log_size=%u",
1854 F2FS_OPTION(sbi).compress_log_size);
1856 for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1857 seq_printf(seq, ",compress_extension=%s",
1858 F2FS_OPTION(sbi).extensions[i]);
1861 for (i = 0; i < F2FS_OPTION(sbi).nocompress_ext_cnt; i++) {
1862 seq_printf(seq, ",nocompress_extension=%s",
1863 F2FS_OPTION(sbi).noextensions[i]);
1866 if (F2FS_OPTION(sbi).compress_chksum)
1867 seq_puts(seq, ",compress_chksum");
1869 if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_FS)
1870 seq_printf(seq, ",compress_mode=%s", "fs");
1871 else if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_USER)
1872 seq_printf(seq, ",compress_mode=%s", "user");
1874 if (test_opt(sbi, COMPRESS_CACHE))
1875 seq_puts(seq, ",compress_cache");
1879 static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1881 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1883 if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1884 seq_printf(seq, ",background_gc=%s", "sync");
1885 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1886 seq_printf(seq, ",background_gc=%s", "on");
1887 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1888 seq_printf(seq, ",background_gc=%s", "off");
1890 if (test_opt(sbi, GC_MERGE))
1891 seq_puts(seq, ",gc_merge");
1893 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1894 seq_puts(seq, ",disable_roll_forward");
1895 if (test_opt(sbi, NORECOVERY))
1896 seq_puts(seq, ",norecovery");
1897 if (test_opt(sbi, DISCARD))
1898 seq_puts(seq, ",discard");
1900 seq_puts(seq, ",nodiscard");
1901 if (test_opt(sbi, NOHEAP))
1902 seq_puts(seq, ",no_heap");
1904 seq_puts(seq, ",heap");
1905 #ifdef CONFIG_F2FS_FS_XATTR
1906 if (test_opt(sbi, XATTR_USER))
1907 seq_puts(seq, ",user_xattr");
1909 seq_puts(seq, ",nouser_xattr");
1910 if (test_opt(sbi, INLINE_XATTR))
1911 seq_puts(seq, ",inline_xattr");
1913 seq_puts(seq, ",noinline_xattr");
1914 if (test_opt(sbi, INLINE_XATTR_SIZE))
1915 seq_printf(seq, ",inline_xattr_size=%u",
1916 F2FS_OPTION(sbi).inline_xattr_size);
1918 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1919 if (test_opt(sbi, POSIX_ACL))
1920 seq_puts(seq, ",acl");
1922 seq_puts(seq, ",noacl");
1924 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
1925 seq_puts(seq, ",disable_ext_identify");
1926 if (test_opt(sbi, INLINE_DATA))
1927 seq_puts(seq, ",inline_data");
1929 seq_puts(seq, ",noinline_data");
1930 if (test_opt(sbi, INLINE_DENTRY))
1931 seq_puts(seq, ",inline_dentry");
1933 seq_puts(seq, ",noinline_dentry");
1934 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
1935 seq_puts(seq, ",flush_merge");
1936 if (test_opt(sbi, NOBARRIER))
1937 seq_puts(seq, ",nobarrier");
1938 if (test_opt(sbi, FASTBOOT))
1939 seq_puts(seq, ",fastboot");
1940 if (test_opt(sbi, EXTENT_CACHE))
1941 seq_puts(seq, ",extent_cache");
1943 seq_puts(seq, ",noextent_cache");
1944 if (test_opt(sbi, DATA_FLUSH))
1945 seq_puts(seq, ",data_flush");
1947 seq_puts(seq, ",mode=");
1948 if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
1949 seq_puts(seq, "adaptive");
1950 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
1951 seq_puts(seq, "lfs");
1952 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG)
1953 seq_puts(seq, "fragment:segment");
1954 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK)
1955 seq_puts(seq, "fragment:block");
1956 seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
1957 if (test_opt(sbi, RESERVE_ROOT))
1958 seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
1959 F2FS_OPTION(sbi).root_reserved_blocks,
1960 from_kuid_munged(&init_user_ns,
1961 F2FS_OPTION(sbi).s_resuid),
1962 from_kgid_munged(&init_user_ns,
1963 F2FS_OPTION(sbi).s_resgid));
1964 if (F2FS_IO_SIZE_BITS(sbi))
1965 seq_printf(seq, ",io_bits=%u",
1966 F2FS_OPTION(sbi).write_io_size_bits);
1967 #ifdef CONFIG_F2FS_FAULT_INJECTION
1968 if (test_opt(sbi, FAULT_INJECTION)) {
1969 seq_printf(seq, ",fault_injection=%u",
1970 F2FS_OPTION(sbi).fault_info.inject_rate);
1971 seq_printf(seq, ",fault_type=%u",
1972 F2FS_OPTION(sbi).fault_info.inject_type);
1976 if (test_opt(sbi, QUOTA))
1977 seq_puts(seq, ",quota");
1978 if (test_opt(sbi, USRQUOTA))
1979 seq_puts(seq, ",usrquota");
1980 if (test_opt(sbi, GRPQUOTA))
1981 seq_puts(seq, ",grpquota");
1982 if (test_opt(sbi, PRJQUOTA))
1983 seq_puts(seq, ",prjquota");
1985 f2fs_show_quota_options(seq, sbi->sb);
1987 fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
1989 if (sbi->sb->s_flags & SB_INLINECRYPT)
1990 seq_puts(seq, ",inlinecrypt");
1992 if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
1993 seq_printf(seq, ",alloc_mode=%s", "default");
1994 else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
1995 seq_printf(seq, ",alloc_mode=%s", "reuse");
1997 if (test_opt(sbi, DISABLE_CHECKPOINT))
1998 seq_printf(seq, ",checkpoint=disable:%u",
1999 F2FS_OPTION(sbi).unusable_cap);
2000 if (test_opt(sbi, MERGE_CHECKPOINT))
2001 seq_puts(seq, ",checkpoint_merge");
2003 seq_puts(seq, ",nocheckpoint_merge");
2004 if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
2005 seq_printf(seq, ",fsync_mode=%s", "posix");
2006 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
2007 seq_printf(seq, ",fsync_mode=%s", "strict");
2008 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
2009 seq_printf(seq, ",fsync_mode=%s", "nobarrier");
2011 #ifdef CONFIG_F2FS_FS_COMPRESSION
2012 f2fs_show_compress_options(seq, sbi->sb);
2015 if (test_opt(sbi, ATGC))
2016 seq_puts(seq, ",atgc");
2018 if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK)
2019 seq_printf(seq, ",discard_unit=%s", "block");
2020 else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SEGMENT)
2021 seq_printf(seq, ",discard_unit=%s", "segment");
2022 else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SECTION)
2023 seq_printf(seq, ",discard_unit=%s", "section");
2025 if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_NORMAL)
2026 seq_printf(seq, ",memory=%s", "normal");
2027 else if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW)
2028 seq_printf(seq, ",memory=%s", "low");
2033 static void default_options(struct f2fs_sb_info *sbi)
2035 /* init some FS parameters */
2036 if (f2fs_sb_has_readonly(sbi))
2037 F2FS_OPTION(sbi).active_logs = NR_CURSEG_RO_TYPE;
2039 F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
2041 F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
2042 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
2043 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
2044 F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
2045 F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
2046 F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
2047 F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
2048 F2FS_OPTION(sbi).compress_ext_cnt = 0;
2049 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
2050 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
2051 F2FS_OPTION(sbi).memory_mode = MEMORY_MODE_NORMAL;
2053 sbi->sb->s_flags &= ~SB_INLINECRYPT;
2055 set_opt(sbi, INLINE_XATTR);
2056 set_opt(sbi, INLINE_DATA);
2057 set_opt(sbi, INLINE_DENTRY);
2058 set_opt(sbi, EXTENT_CACHE);
2059 set_opt(sbi, NOHEAP);
2060 clear_opt(sbi, DISABLE_CHECKPOINT);
2061 set_opt(sbi, MERGE_CHECKPOINT);
2062 F2FS_OPTION(sbi).unusable_cap = 0;
2063 sbi->sb->s_flags |= SB_LAZYTIME;
2064 set_opt(sbi, FLUSH_MERGE);
2065 if (f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi))
2066 set_opt(sbi, DISCARD);
2067 if (f2fs_sb_has_blkzoned(sbi)) {
2068 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
2069 F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_SECTION;
2071 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
2072 F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_BLOCK;
2075 #ifdef CONFIG_F2FS_FS_XATTR
2076 set_opt(sbi, XATTR_USER);
2078 #ifdef CONFIG_F2FS_FS_POSIX_ACL
2079 set_opt(sbi, POSIX_ACL);
2082 f2fs_build_fault_attr(sbi, 0, 0);
2086 static int f2fs_enable_quotas(struct super_block *sb);
2089 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
2091 unsigned int s_flags = sbi->sb->s_flags;
2092 struct cp_control cpc;
2093 unsigned int gc_mode = sbi->gc_mode;
2098 if (s_flags & SB_RDONLY) {
2099 f2fs_err(sbi, "checkpoint=disable on readonly fs");
2102 sbi->sb->s_flags |= SB_ACTIVE;
2104 /* check if we need more GC first */
2105 unusable = f2fs_get_unusable_blocks(sbi);
2106 if (!f2fs_disable_cp_again(sbi, unusable))
2109 f2fs_update_time(sbi, DISABLE_TIME);
2111 sbi->gc_mode = GC_URGENT_HIGH;
2113 while (!f2fs_time_over(sbi, DISABLE_TIME)) {
2114 struct f2fs_gc_control gc_control = {
2115 .victim_segno = NULL_SEGNO,
2116 .init_gc_type = FG_GC,
2117 .should_migrate_blocks = false,
2118 .err_gc_skipped = true,
2119 .nr_free_secs = 1 };
2121 f2fs_down_write(&sbi->gc_lock);
2122 err = f2fs_gc(sbi, &gc_control);
2123 if (err == -ENODATA) {
2127 if (err && err != -EAGAIN)
2131 ret = sync_filesystem(sbi->sb);
2133 err = ret ? ret : err;
2137 unusable = f2fs_get_unusable_blocks(sbi);
2138 if (f2fs_disable_cp_again(sbi, unusable)) {
2144 f2fs_down_write(&sbi->gc_lock);
2145 cpc.reason = CP_PAUSE;
2146 set_sbi_flag(sbi, SBI_CP_DISABLED);
2147 err = f2fs_write_checkpoint(sbi, &cpc);
2151 spin_lock(&sbi->stat_lock);
2152 sbi->unusable_block_count = unusable;
2153 spin_unlock(&sbi->stat_lock);
2156 f2fs_up_write(&sbi->gc_lock);
2158 sbi->gc_mode = gc_mode;
2159 sbi->sb->s_flags = s_flags; /* Restore SB_RDONLY status */
2163 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
2165 int retry = DEFAULT_RETRY_IO_COUNT;
2167 /* we should flush all the data to keep data consistency */
2169 sync_inodes_sb(sbi->sb);
2170 f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2171 } while (get_pages(sbi, F2FS_DIRTY_DATA) && retry--);
2173 if (unlikely(retry < 0))
2174 f2fs_warn(sbi, "checkpoint=enable has some unwritten data.");
2176 f2fs_down_write(&sbi->gc_lock);
2177 f2fs_dirty_to_prefree(sbi);
2179 clear_sbi_flag(sbi, SBI_CP_DISABLED);
2180 set_sbi_flag(sbi, SBI_IS_DIRTY);
2181 f2fs_up_write(&sbi->gc_lock);
2183 f2fs_sync_fs(sbi->sb, 1);
2186 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
2188 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2189 struct f2fs_mount_info org_mount_opt;
2190 unsigned long old_sb_flags;
2192 bool need_restart_gc = false, need_stop_gc = false;
2193 bool need_restart_ckpt = false, need_stop_ckpt = false;
2194 bool need_restart_flush = false, need_stop_flush = false;
2195 bool need_restart_discard = false, need_stop_discard = false;
2196 bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
2197 bool enable_checkpoint = !test_opt(sbi, DISABLE_CHECKPOINT);
2198 bool no_io_align = !F2FS_IO_ALIGNED(sbi);
2199 bool no_atgc = !test_opt(sbi, ATGC);
2200 bool no_discard = !test_opt(sbi, DISCARD);
2201 bool no_compress_cache = !test_opt(sbi, COMPRESS_CACHE);
2202 bool block_unit_discard = f2fs_block_unit_discard(sbi);
2203 struct discard_cmd_control *dcc;
2209 * Save the old mount options in case we
2210 * need to restore them.
2212 org_mount_opt = sbi->mount_opt;
2213 old_sb_flags = sb->s_flags;
2216 org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
2217 for (i = 0; i < MAXQUOTAS; i++) {
2218 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2219 org_mount_opt.s_qf_names[i] =
2220 kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
2222 if (!org_mount_opt.s_qf_names[i]) {
2223 for (j = 0; j < i; j++)
2224 kfree(org_mount_opt.s_qf_names[j]);
2228 org_mount_opt.s_qf_names[i] = NULL;
2233 /* recover superblocks we couldn't write due to previous RO mount */
2234 if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
2235 err = f2fs_commit_super(sbi, false);
2236 f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
2239 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2242 default_options(sbi);
2244 /* parse mount options */
2245 err = parse_options(sb, data, true);
2250 * Previous and new state of filesystem is RO,
2251 * so skip checking GC and FLUSH_MERGE conditions.
2253 if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
2256 if (f2fs_sb_has_readonly(sbi) && !(*flags & SB_RDONLY)) {
2262 if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
2263 err = dquot_suspend(sb, -1);
2266 } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
2267 /* dquot_resume needs RW */
2268 sb->s_flags &= ~SB_RDONLY;
2269 if (sb_any_quota_suspended(sb)) {
2270 dquot_resume(sb, -1);
2271 } else if (f2fs_sb_has_quota_ino(sbi)) {
2272 err = f2fs_enable_quotas(sb);
2278 /* disallow enable atgc dynamically */
2279 if (no_atgc == !!test_opt(sbi, ATGC)) {
2281 f2fs_warn(sbi, "switch atgc option is not allowed");
2285 /* disallow enable/disable extent_cache dynamically */
2286 if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
2288 f2fs_warn(sbi, "switch extent_cache option is not allowed");
2292 if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
2294 f2fs_warn(sbi, "switch io_bits option is not allowed");
2298 if (no_compress_cache == !!test_opt(sbi, COMPRESS_CACHE)) {
2300 f2fs_warn(sbi, "switch compress_cache option is not allowed");
2304 if (block_unit_discard != f2fs_block_unit_discard(sbi)) {
2306 f2fs_warn(sbi, "switch discard_unit option is not allowed");
2310 if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
2312 f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
2317 * We stop the GC thread if FS is mounted as RO
2318 * or if background_gc = off is passed in mount
2319 * option. Also sync the filesystem.
2321 if ((*flags & SB_RDONLY) ||
2322 (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF &&
2323 !test_opt(sbi, GC_MERGE))) {
2324 if (sbi->gc_thread) {
2325 f2fs_stop_gc_thread(sbi);
2326 need_restart_gc = true;
2328 } else if (!sbi->gc_thread) {
2329 err = f2fs_start_gc_thread(sbi);
2332 need_stop_gc = true;
2335 if (*flags & SB_RDONLY) {
2338 set_sbi_flag(sbi, SBI_IS_DIRTY);
2339 set_sbi_flag(sbi, SBI_IS_CLOSE);
2340 f2fs_sync_fs(sb, 1);
2341 clear_sbi_flag(sbi, SBI_IS_CLOSE);
2344 if ((*flags & SB_RDONLY) || test_opt(sbi, DISABLE_CHECKPOINT) ||
2345 !test_opt(sbi, MERGE_CHECKPOINT)) {
2346 f2fs_stop_ckpt_thread(sbi);
2347 need_restart_ckpt = true;
2349 err = f2fs_start_ckpt_thread(sbi);
2352 "Failed to start F2FS issue_checkpoint_thread (%d)",
2356 need_stop_ckpt = true;
2360 * We stop issue flush thread if FS is mounted as RO
2361 * or if flush_merge is not passed in mount option.
2363 if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2364 clear_opt(sbi, FLUSH_MERGE);
2365 f2fs_destroy_flush_cmd_control(sbi, false);
2366 need_restart_flush = true;
2368 err = f2fs_create_flush_cmd_control(sbi);
2371 need_stop_flush = true;
2374 if (no_discard == !!test_opt(sbi, DISCARD)) {
2375 if (test_opt(sbi, DISCARD)) {
2376 err = f2fs_start_discard_thread(sbi);
2379 need_stop_discard = true;
2381 dcc = SM_I(sbi)->dcc_info;
2382 f2fs_stop_discard_thread(sbi);
2383 if (atomic_read(&dcc->discard_cmd_cnt))
2384 f2fs_issue_discard_timeout(sbi);
2385 need_restart_discard = true;
2389 if (enable_checkpoint == !!test_opt(sbi, DISABLE_CHECKPOINT)) {
2390 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
2391 err = f2fs_disable_checkpoint(sbi);
2393 goto restore_discard;
2395 f2fs_enable_checkpoint(sbi);
2401 /* Release old quota file names */
2402 for (i = 0; i < MAXQUOTAS; i++)
2403 kfree(org_mount_opt.s_qf_names[i]);
2405 /* Update the POSIXACL Flag */
2406 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
2407 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
2409 limit_reserve_root(sbi);
2410 adjust_unusable_cap_perc(sbi);
2411 *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
2414 if (need_restart_discard) {
2415 if (f2fs_start_discard_thread(sbi))
2416 f2fs_warn(sbi, "discard has been stopped");
2417 } else if (need_stop_discard) {
2418 f2fs_stop_discard_thread(sbi);
2421 if (need_restart_flush) {
2422 if (f2fs_create_flush_cmd_control(sbi))
2423 f2fs_warn(sbi, "background flush thread has stopped");
2424 } else if (need_stop_flush) {
2425 clear_opt(sbi, FLUSH_MERGE);
2426 f2fs_destroy_flush_cmd_control(sbi, false);
2429 if (need_restart_ckpt) {
2430 if (f2fs_start_ckpt_thread(sbi))
2431 f2fs_warn(sbi, "background ckpt thread has stopped");
2432 } else if (need_stop_ckpt) {
2433 f2fs_stop_ckpt_thread(sbi);
2436 if (need_restart_gc) {
2437 if (f2fs_start_gc_thread(sbi))
2438 f2fs_warn(sbi, "background gc thread has stopped");
2439 } else if (need_stop_gc) {
2440 f2fs_stop_gc_thread(sbi);
2444 F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
2445 for (i = 0; i < MAXQUOTAS; i++) {
2446 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
2447 F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
2450 sbi->mount_opt = org_mount_opt;
2451 sb->s_flags = old_sb_flags;
2456 /* Read data from quotafile */
2457 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
2458 size_t len, loff_t off)
2460 struct inode *inode = sb_dqopt(sb)->files[type];
2461 struct address_space *mapping = inode->i_mapping;
2462 block_t blkidx = F2FS_BYTES_TO_BLK(off);
2463 int offset = off & (sb->s_blocksize - 1);
2466 loff_t i_size = i_size_read(inode);
2473 if (off + len > i_size)
2476 while (toread > 0) {
2477 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
2479 page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
2481 if (PTR_ERR(page) == -ENOMEM) {
2482 memalloc_retry_wait(GFP_NOFS);
2485 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2486 return PTR_ERR(page);
2491 if (unlikely(page->mapping != mapping)) {
2492 f2fs_put_page(page, 1);
2495 if (unlikely(!PageUptodate(page))) {
2496 f2fs_put_page(page, 1);
2497 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2501 kaddr = kmap_atomic(page);
2502 memcpy(data, kaddr + offset, tocopy);
2503 kunmap_atomic(kaddr);
2504 f2fs_put_page(page, 1);
2514 /* Write to quotafile */
2515 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2516 const char *data, size_t len, loff_t off)
2518 struct inode *inode = sb_dqopt(sb)->files[type];
2519 struct address_space *mapping = inode->i_mapping;
2520 const struct address_space_operations *a_ops = mapping->a_ops;
2521 int offset = off & (sb->s_blocksize - 1);
2522 size_t towrite = len;
2524 void *fsdata = NULL;
2529 while (towrite > 0) {
2530 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2533 err = a_ops->write_begin(NULL, mapping, off, tocopy,
2535 if (unlikely(err)) {
2536 if (err == -ENOMEM) {
2537 f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2540 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2544 kaddr = kmap_atomic(page);
2545 memcpy(kaddr + offset, data, tocopy);
2546 kunmap_atomic(kaddr);
2547 flush_dcache_page(page);
2549 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2560 inode->i_mtime = inode->i_ctime = current_time(inode);
2561 f2fs_mark_inode_dirty_sync(inode, false);
2562 return len - towrite;
2565 int f2fs_dquot_initialize(struct inode *inode)
2567 if (time_to_inject(F2FS_I_SB(inode), FAULT_DQUOT_INIT)) {
2568 f2fs_show_injection_info(F2FS_I_SB(inode), FAULT_DQUOT_INIT);
2572 return dquot_initialize(inode);
2575 static struct dquot **f2fs_get_dquots(struct inode *inode)
2577 return F2FS_I(inode)->i_dquot;
2580 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2582 return &F2FS_I(inode)->i_reserved_quota;
2585 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2587 if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2588 f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2592 return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2593 F2FS_OPTION(sbi).s_jquota_fmt, type);
2596 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2601 if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2602 err = f2fs_enable_quotas(sbi->sb);
2604 f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2610 for (i = 0; i < MAXQUOTAS; i++) {
2611 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2612 err = f2fs_quota_on_mount(sbi, i);
2617 f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2624 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2627 struct inode *qf_inode;
2628 unsigned long qf_inum;
2631 BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2633 qf_inum = f2fs_qf_ino(sb, type);
2637 qf_inode = f2fs_iget(sb, qf_inum);
2638 if (IS_ERR(qf_inode)) {
2639 f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2640 return PTR_ERR(qf_inode);
2643 /* Don't account quota for quota files to avoid recursion */
2644 qf_inode->i_flags |= S_NOQUOTA;
2645 err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2650 static int f2fs_enable_quotas(struct super_block *sb)
2652 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2654 unsigned long qf_inum;
2655 bool quota_mopt[MAXQUOTAS] = {
2656 test_opt(sbi, USRQUOTA),
2657 test_opt(sbi, GRPQUOTA),
2658 test_opt(sbi, PRJQUOTA),
2661 if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2662 f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2666 sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2668 for (type = 0; type < MAXQUOTAS; type++) {
2669 qf_inum = f2fs_qf_ino(sb, type);
2671 err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2672 DQUOT_USAGE_ENABLED |
2673 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2675 f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2677 for (type--; type >= 0; type--)
2678 dquot_quota_off(sb, type);
2679 set_sbi_flag(F2FS_SB(sb),
2680 SBI_QUOTA_NEED_REPAIR);
2688 static int f2fs_quota_sync_file(struct f2fs_sb_info *sbi, int type)
2690 struct quota_info *dqopt = sb_dqopt(sbi->sb);
2691 struct address_space *mapping = dqopt->files[type]->i_mapping;
2694 ret = dquot_writeback_dquots(sbi->sb, type);
2698 ret = filemap_fdatawrite(mapping);
2702 /* if we are using journalled quota */
2703 if (is_journalled_quota(sbi))
2706 ret = filemap_fdatawait(mapping);
2708 truncate_inode_pages(&dqopt->files[type]->i_data, 0);
2711 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2715 int f2fs_quota_sync(struct super_block *sb, int type)
2717 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2718 struct quota_info *dqopt = sb_dqopt(sb);
2723 * Now when everything is written we can discard the pagecache so
2724 * that userspace sees the changes.
2726 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2728 if (type != -1 && cnt != type)
2731 if (!sb_has_quota_active(sb, cnt))
2734 if (!f2fs_sb_has_quota_ino(sbi))
2735 inode_lock(dqopt->files[cnt]);
2740 * f2fs_down_read(quota_sem)
2741 * dquot_writeback_dquots()
2744 * f2fs_down_read(quota_sem)
2747 f2fs_down_read(&sbi->quota_sem);
2749 ret = f2fs_quota_sync_file(sbi, cnt);
2751 f2fs_up_read(&sbi->quota_sem);
2752 f2fs_unlock_op(sbi);
2754 if (!f2fs_sb_has_quota_ino(sbi))
2755 inode_unlock(dqopt->files[cnt]);
2763 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2764 const struct path *path)
2766 struct inode *inode;
2769 /* if quota sysfile exists, deny enabling quota with specific file */
2770 if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2771 f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2775 err = f2fs_quota_sync(sb, type);
2779 err = dquot_quota_on(sb, type, format_id, path);
2783 inode = d_inode(path->dentry);
2786 F2FS_I(inode)->i_flags |= F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL;
2787 f2fs_set_inode_flags(inode);
2788 inode_unlock(inode);
2789 f2fs_mark_inode_dirty_sync(inode, false);
2794 static int __f2fs_quota_off(struct super_block *sb, int type)
2796 struct inode *inode = sb_dqopt(sb)->files[type];
2799 if (!inode || !igrab(inode))
2800 return dquot_quota_off(sb, type);
2802 err = f2fs_quota_sync(sb, type);
2806 err = dquot_quota_off(sb, type);
2807 if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
2811 F2FS_I(inode)->i_flags &= ~(F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL);
2812 f2fs_set_inode_flags(inode);
2813 inode_unlock(inode);
2814 f2fs_mark_inode_dirty_sync(inode, false);
2820 static int f2fs_quota_off(struct super_block *sb, int type)
2822 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2825 err = __f2fs_quota_off(sb, type);
2828 * quotactl can shutdown journalled quota, result in inconsistence
2829 * between quota record and fs data by following updates, tag the
2830 * flag to let fsck be aware of it.
2832 if (is_journalled_quota(sbi))
2833 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2837 void f2fs_quota_off_umount(struct super_block *sb)
2842 for (type = 0; type < MAXQUOTAS; type++) {
2843 err = __f2fs_quota_off(sb, type);
2845 int ret = dquot_quota_off(sb, type);
2847 f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
2849 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2853 * In case of checkpoint=disable, we must flush quota blocks.
2854 * This can cause NULL exception for node_inode in end_io, since
2855 * put_super already dropped it.
2857 sync_filesystem(sb);
2860 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
2862 struct quota_info *dqopt = sb_dqopt(sb);
2865 for (type = 0; type < MAXQUOTAS; type++) {
2866 if (!dqopt->files[type])
2868 f2fs_inode_synced(dqopt->files[type]);
2872 static int f2fs_dquot_commit(struct dquot *dquot)
2874 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2877 f2fs_down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
2878 ret = dquot_commit(dquot);
2880 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2881 f2fs_up_read(&sbi->quota_sem);
2885 static int f2fs_dquot_acquire(struct dquot *dquot)
2887 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2890 f2fs_down_read(&sbi->quota_sem);
2891 ret = dquot_acquire(dquot);
2893 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2894 f2fs_up_read(&sbi->quota_sem);
2898 static int f2fs_dquot_release(struct dquot *dquot)
2900 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2901 int ret = dquot_release(dquot);
2904 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2908 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
2910 struct super_block *sb = dquot->dq_sb;
2911 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2912 int ret = dquot_mark_dquot_dirty(dquot);
2914 /* if we are using journalled quota */
2915 if (is_journalled_quota(sbi))
2916 set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
2921 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
2923 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2924 int ret = dquot_commit_info(sb, type);
2927 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2931 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
2933 *projid = F2FS_I(inode)->i_projid;
2937 static const struct dquot_operations f2fs_quota_operations = {
2938 .get_reserved_space = f2fs_get_reserved_space,
2939 .write_dquot = f2fs_dquot_commit,
2940 .acquire_dquot = f2fs_dquot_acquire,
2941 .release_dquot = f2fs_dquot_release,
2942 .mark_dirty = f2fs_dquot_mark_dquot_dirty,
2943 .write_info = f2fs_dquot_commit_info,
2944 .alloc_dquot = dquot_alloc,
2945 .destroy_dquot = dquot_destroy,
2946 .get_projid = f2fs_get_projid,
2947 .get_next_id = dquot_get_next_id,
2950 static const struct quotactl_ops f2fs_quotactl_ops = {
2951 .quota_on = f2fs_quota_on,
2952 .quota_off = f2fs_quota_off,
2953 .quota_sync = f2fs_quota_sync,
2954 .get_state = dquot_get_state,
2955 .set_info = dquot_set_dqinfo,
2956 .get_dqblk = dquot_get_dqblk,
2957 .set_dqblk = dquot_set_dqblk,
2958 .get_nextdqblk = dquot_get_next_dqblk,
2961 int f2fs_dquot_initialize(struct inode *inode)
2966 int f2fs_quota_sync(struct super_block *sb, int type)
2971 void f2fs_quota_off_umount(struct super_block *sb)
2976 static const struct super_operations f2fs_sops = {
2977 .alloc_inode = f2fs_alloc_inode,
2978 .free_inode = f2fs_free_inode,
2979 .drop_inode = f2fs_drop_inode,
2980 .write_inode = f2fs_write_inode,
2981 .dirty_inode = f2fs_dirty_inode,
2982 .show_options = f2fs_show_options,
2984 .quota_read = f2fs_quota_read,
2985 .quota_write = f2fs_quota_write,
2986 .get_dquots = f2fs_get_dquots,
2988 .evict_inode = f2fs_evict_inode,
2989 .put_super = f2fs_put_super,
2990 .sync_fs = f2fs_sync_fs,
2991 .freeze_fs = f2fs_freeze,
2992 .unfreeze_fs = f2fs_unfreeze,
2993 .statfs = f2fs_statfs,
2994 .remount_fs = f2fs_remount,
2997 #ifdef CONFIG_FS_ENCRYPTION
2998 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
3000 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3001 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3005 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
3008 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3011 * Encrypting the root directory is not allowed because fsck
3012 * expects lost+found directory to exist and remain unencrypted
3013 * if LOST_FOUND feature is enabled.
3016 if (f2fs_sb_has_lost_found(sbi) &&
3017 inode->i_ino == F2FS_ROOT_INO(sbi))
3020 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3021 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3022 ctx, len, fs_data, XATTR_CREATE);
3025 static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
3027 return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
3030 static bool f2fs_has_stable_inodes(struct super_block *sb)
3035 static void f2fs_get_ino_and_lblk_bits(struct super_block *sb,
3036 int *ino_bits_ret, int *lblk_bits_ret)
3038 *ino_bits_ret = 8 * sizeof(nid_t);
3039 *lblk_bits_ret = 8 * sizeof(block_t);
3042 static int f2fs_get_num_devices(struct super_block *sb)
3044 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3046 if (f2fs_is_multi_device(sbi))
3047 return sbi->s_ndevs;
3051 static void f2fs_get_devices(struct super_block *sb,
3052 struct request_queue **devs)
3054 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3057 for (i = 0; i < sbi->s_ndevs; i++)
3058 devs[i] = bdev_get_queue(FDEV(i).bdev);
3061 static const struct fscrypt_operations f2fs_cryptops = {
3062 .key_prefix = "f2fs:",
3063 .get_context = f2fs_get_context,
3064 .set_context = f2fs_set_context,
3065 .get_dummy_policy = f2fs_get_dummy_policy,
3066 .empty_dir = f2fs_empty_dir,
3067 .has_stable_inodes = f2fs_has_stable_inodes,
3068 .get_ino_and_lblk_bits = f2fs_get_ino_and_lblk_bits,
3069 .get_num_devices = f2fs_get_num_devices,
3070 .get_devices = f2fs_get_devices,
3074 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
3075 u64 ino, u32 generation)
3077 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3078 struct inode *inode;
3080 if (f2fs_check_nid_range(sbi, ino))
3081 return ERR_PTR(-ESTALE);
3084 * f2fs_iget isn't quite right if the inode is currently unallocated!
3085 * However f2fs_iget currently does appropriate checks to handle stale
3086 * inodes so everything is OK.
3088 inode = f2fs_iget(sb, ino);
3090 return ERR_CAST(inode);
3091 if (unlikely(generation && inode->i_generation != generation)) {
3092 /* we didn't find the right inode.. */
3094 return ERR_PTR(-ESTALE);
3099 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
3100 int fh_len, int fh_type)
3102 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
3103 f2fs_nfs_get_inode);
3106 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
3107 int fh_len, int fh_type)
3109 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
3110 f2fs_nfs_get_inode);
3113 static const struct export_operations f2fs_export_ops = {
3114 .fh_to_dentry = f2fs_fh_to_dentry,
3115 .fh_to_parent = f2fs_fh_to_parent,
3116 .get_parent = f2fs_get_parent,
3119 loff_t max_file_blocks(struct inode *inode)
3125 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
3126 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
3127 * space in inode.i_addr, it will be more safe to reassign
3131 if (inode && f2fs_compressed_file(inode))
3132 leaf_count = ADDRS_PER_BLOCK(inode);
3134 leaf_count = DEF_ADDRS_PER_BLOCK;
3136 /* two direct node blocks */
3137 result += (leaf_count * 2);
3139 /* two indirect node blocks */
3140 leaf_count *= NIDS_PER_BLOCK;
3141 result += (leaf_count * 2);
3143 /* one double indirect node block */
3144 leaf_count *= NIDS_PER_BLOCK;
3145 result += leaf_count;
3150 static int __f2fs_commit_super(struct buffer_head *bh,
3151 struct f2fs_super_block *super)
3155 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
3156 set_buffer_dirty(bh);
3159 /* it's rare case, we can do fua all the time */
3160 return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
3163 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
3164 struct buffer_head *bh)
3166 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3167 (bh->b_data + F2FS_SUPER_OFFSET);
3168 struct super_block *sb = sbi->sb;
3169 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
3170 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
3171 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
3172 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
3173 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
3174 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
3175 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
3176 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
3177 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
3178 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
3179 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3180 u32 segment_count = le32_to_cpu(raw_super->segment_count);
3181 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3182 u64 main_end_blkaddr = main_blkaddr +
3183 (segment_count_main << log_blocks_per_seg);
3184 u64 seg_end_blkaddr = segment0_blkaddr +
3185 (segment_count << log_blocks_per_seg);
3187 if (segment0_blkaddr != cp_blkaddr) {
3188 f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
3189 segment0_blkaddr, cp_blkaddr);
3193 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
3195 f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
3196 cp_blkaddr, sit_blkaddr,
3197 segment_count_ckpt << log_blocks_per_seg);
3201 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
3203 f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
3204 sit_blkaddr, nat_blkaddr,
3205 segment_count_sit << log_blocks_per_seg);
3209 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
3211 f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
3212 nat_blkaddr, ssa_blkaddr,
3213 segment_count_nat << log_blocks_per_seg);
3217 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
3219 f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
3220 ssa_blkaddr, main_blkaddr,
3221 segment_count_ssa << log_blocks_per_seg);
3225 if (main_end_blkaddr > seg_end_blkaddr) {
3226 f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
3227 main_blkaddr, seg_end_blkaddr,
3228 segment_count_main << log_blocks_per_seg);
3230 } else if (main_end_blkaddr < seg_end_blkaddr) {
3234 /* fix in-memory information all the time */
3235 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
3236 segment0_blkaddr) >> log_blocks_per_seg);
3238 if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
3239 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3242 err = __f2fs_commit_super(bh, NULL);
3243 res = err ? "failed" : "done";
3245 f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
3246 res, main_blkaddr, seg_end_blkaddr,
3247 segment_count_main << log_blocks_per_seg);
3254 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
3255 struct buffer_head *bh)
3257 block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
3258 block_t total_sections, blocks_per_seg;
3259 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3260 (bh->b_data + F2FS_SUPER_OFFSET);
3261 size_t crc_offset = 0;
3264 if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
3265 f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
3266 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
3270 /* Check checksum_offset and crc in superblock */
3271 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
3272 crc_offset = le32_to_cpu(raw_super->checksum_offset);
3274 offsetof(struct f2fs_super_block, crc)) {
3275 f2fs_info(sbi, "Invalid SB checksum offset: %zu",
3277 return -EFSCORRUPTED;
3279 crc = le32_to_cpu(raw_super->crc);
3280 if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
3281 f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
3282 return -EFSCORRUPTED;
3286 /* Currently, support only 4KB block size */
3287 if (le32_to_cpu(raw_super->log_blocksize) != F2FS_BLKSIZE_BITS) {
3288 f2fs_info(sbi, "Invalid log_blocksize (%u), supports only %u",
3289 le32_to_cpu(raw_super->log_blocksize),
3291 return -EFSCORRUPTED;
3294 /* check log blocks per segment */
3295 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
3296 f2fs_info(sbi, "Invalid log blocks per segment (%u)",
3297 le32_to_cpu(raw_super->log_blocks_per_seg));
3298 return -EFSCORRUPTED;
3301 /* Currently, support 512/1024/2048/4096 bytes sector size */
3302 if (le32_to_cpu(raw_super->log_sectorsize) >
3303 F2FS_MAX_LOG_SECTOR_SIZE ||
3304 le32_to_cpu(raw_super->log_sectorsize) <
3305 F2FS_MIN_LOG_SECTOR_SIZE) {
3306 f2fs_info(sbi, "Invalid log sectorsize (%u)",
3307 le32_to_cpu(raw_super->log_sectorsize));
3308 return -EFSCORRUPTED;
3310 if (le32_to_cpu(raw_super->log_sectors_per_block) +
3311 le32_to_cpu(raw_super->log_sectorsize) !=
3312 F2FS_MAX_LOG_SECTOR_SIZE) {
3313 f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
3314 le32_to_cpu(raw_super->log_sectors_per_block),
3315 le32_to_cpu(raw_super->log_sectorsize));
3316 return -EFSCORRUPTED;
3319 segment_count = le32_to_cpu(raw_super->segment_count);
3320 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3321 segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3322 secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3323 total_sections = le32_to_cpu(raw_super->section_count);
3325 /* blocks_per_seg should be 512, given the above check */
3326 blocks_per_seg = 1 << le32_to_cpu(raw_super->log_blocks_per_seg);
3328 if (segment_count > F2FS_MAX_SEGMENT ||
3329 segment_count < F2FS_MIN_SEGMENTS) {
3330 f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
3331 return -EFSCORRUPTED;
3334 if (total_sections > segment_count_main || total_sections < 1 ||
3335 segs_per_sec > segment_count || !segs_per_sec) {
3336 f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
3337 segment_count, total_sections, segs_per_sec);
3338 return -EFSCORRUPTED;
3341 if (segment_count_main != total_sections * segs_per_sec) {
3342 f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
3343 segment_count_main, total_sections, segs_per_sec);
3344 return -EFSCORRUPTED;
3347 if ((segment_count / segs_per_sec) < total_sections) {
3348 f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
3349 segment_count, segs_per_sec, total_sections);
3350 return -EFSCORRUPTED;
3353 if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
3354 f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
3355 segment_count, le64_to_cpu(raw_super->block_count));
3356 return -EFSCORRUPTED;
3359 if (RDEV(0).path[0]) {
3360 block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
3363 while (i < MAX_DEVICES && RDEV(i).path[0]) {
3364 dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
3367 if (segment_count != dev_seg_count) {
3368 f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
3369 segment_count, dev_seg_count);
3370 return -EFSCORRUPTED;
3373 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
3374 !bdev_is_zoned(sbi->sb->s_bdev)) {
3375 f2fs_info(sbi, "Zoned block device path is missing");
3376 return -EFSCORRUPTED;
3380 if (secs_per_zone > total_sections || !secs_per_zone) {
3381 f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
3382 secs_per_zone, total_sections);
3383 return -EFSCORRUPTED;
3385 if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
3386 raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
3387 (le32_to_cpu(raw_super->extension_count) +
3388 raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
3389 f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
3390 le32_to_cpu(raw_super->extension_count),
3391 raw_super->hot_ext_count,
3392 F2FS_MAX_EXTENSION);
3393 return -EFSCORRUPTED;
3396 if (le32_to_cpu(raw_super->cp_payload) >=
3397 (blocks_per_seg - F2FS_CP_PACKS -
3398 NR_CURSEG_PERSIST_TYPE)) {
3399 f2fs_info(sbi, "Insane cp_payload (%u >= %u)",
3400 le32_to_cpu(raw_super->cp_payload),
3401 blocks_per_seg - F2FS_CP_PACKS -
3402 NR_CURSEG_PERSIST_TYPE);
3403 return -EFSCORRUPTED;
3406 /* check reserved ino info */
3407 if (le32_to_cpu(raw_super->node_ino) != 1 ||
3408 le32_to_cpu(raw_super->meta_ino) != 2 ||
3409 le32_to_cpu(raw_super->root_ino) != 3) {
3410 f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
3411 le32_to_cpu(raw_super->node_ino),
3412 le32_to_cpu(raw_super->meta_ino),
3413 le32_to_cpu(raw_super->root_ino));
3414 return -EFSCORRUPTED;
3417 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
3418 if (sanity_check_area_boundary(sbi, bh))
3419 return -EFSCORRUPTED;
3424 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
3426 unsigned int total, fsmeta;
3427 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3428 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
3429 unsigned int ovp_segments, reserved_segments;
3430 unsigned int main_segs, blocks_per_seg;
3431 unsigned int sit_segs, nat_segs;
3432 unsigned int sit_bitmap_size, nat_bitmap_size;
3433 unsigned int log_blocks_per_seg;
3434 unsigned int segment_count_main;
3435 unsigned int cp_pack_start_sum, cp_payload;
3436 block_t user_block_count, valid_user_blocks;
3437 block_t avail_node_count, valid_node_count;
3438 unsigned int nat_blocks, nat_bits_bytes, nat_bits_blocks;
3441 total = le32_to_cpu(raw_super->segment_count);
3442 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
3443 sit_segs = le32_to_cpu(raw_super->segment_count_sit);
3445 nat_segs = le32_to_cpu(raw_super->segment_count_nat);
3447 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
3448 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
3450 if (unlikely(fsmeta >= total))
3453 ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
3454 reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
3456 if (!f2fs_sb_has_readonly(sbi) &&
3457 unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
3458 ovp_segments == 0 || reserved_segments == 0)) {
3459 f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
3462 user_block_count = le64_to_cpu(ckpt->user_block_count);
3463 segment_count_main = le32_to_cpu(raw_super->segment_count_main) +
3464 (f2fs_sb_has_readonly(sbi) ? 1 : 0);
3465 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3466 if (!user_block_count || user_block_count >=
3467 segment_count_main << log_blocks_per_seg) {
3468 f2fs_err(sbi, "Wrong user_block_count: %u",
3473 valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
3474 if (valid_user_blocks > user_block_count) {
3475 f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
3476 valid_user_blocks, user_block_count);
3480 valid_node_count = le32_to_cpu(ckpt->valid_node_count);
3481 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
3482 if (valid_node_count > avail_node_count) {
3483 f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
3484 valid_node_count, avail_node_count);
3488 main_segs = le32_to_cpu(raw_super->segment_count_main);
3489 blocks_per_seg = sbi->blocks_per_seg;
3491 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3492 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
3493 le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
3496 if (f2fs_sb_has_readonly(sbi))
3499 for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
3500 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3501 le32_to_cpu(ckpt->cur_node_segno[j])) {
3502 f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
3504 le32_to_cpu(ckpt->cur_node_segno[i]));
3510 for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
3511 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
3512 le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
3515 if (f2fs_sb_has_readonly(sbi))
3518 for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3519 if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3520 le32_to_cpu(ckpt->cur_data_segno[j])) {
3521 f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3523 le32_to_cpu(ckpt->cur_data_segno[i]));
3528 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3529 for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3530 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3531 le32_to_cpu(ckpt->cur_data_segno[j])) {
3532 f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3534 le32_to_cpu(ckpt->cur_node_segno[i]));
3540 sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3541 nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3543 if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3544 nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3545 f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3546 sit_bitmap_size, nat_bitmap_size);
3550 cp_pack_start_sum = __start_sum_addr(sbi);
3551 cp_payload = __cp_payload(sbi);
3552 if (cp_pack_start_sum < cp_payload + 1 ||
3553 cp_pack_start_sum > blocks_per_seg - 1 -
3554 NR_CURSEG_PERSIST_TYPE) {
3555 f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3560 if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3561 le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3562 f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3563 "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3564 "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3565 le32_to_cpu(ckpt->checksum_offset));
3569 nat_blocks = nat_segs << log_blocks_per_seg;
3570 nat_bits_bytes = nat_blocks / BITS_PER_BYTE;
3571 nat_bits_blocks = F2FS_BLK_ALIGN((nat_bits_bytes << 1) + 8);
3572 if (__is_set_ckpt_flags(ckpt, CP_NAT_BITS_FLAG) &&
3573 (cp_payload + F2FS_CP_PACKS +
3574 NR_CURSEG_PERSIST_TYPE + nat_bits_blocks >= blocks_per_seg)) {
3575 f2fs_warn(sbi, "Insane cp_payload: %u, nat_bits_blocks: %u)",
3576 cp_payload, nat_bits_blocks);
3580 if (unlikely(f2fs_cp_error(sbi))) {
3581 f2fs_err(sbi, "A bug case: need to run fsck");
3587 static void init_sb_info(struct f2fs_sb_info *sbi)
3589 struct f2fs_super_block *raw_super = sbi->raw_super;
3592 sbi->log_sectors_per_block =
3593 le32_to_cpu(raw_super->log_sectors_per_block);
3594 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3595 sbi->blocksize = 1 << sbi->log_blocksize;
3596 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3597 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
3598 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3599 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3600 sbi->total_sections = le32_to_cpu(raw_super->section_count);
3601 sbi->total_node_count =
3602 (le32_to_cpu(raw_super->segment_count_nat) / 2)
3603 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
3604 F2FS_ROOT_INO(sbi) = le32_to_cpu(raw_super->root_ino);
3605 F2FS_NODE_INO(sbi) = le32_to_cpu(raw_super->node_ino);
3606 F2FS_META_INO(sbi) = le32_to_cpu(raw_super->meta_ino);
3607 sbi->cur_victim_sec = NULL_SECNO;
3608 sbi->gc_mode = GC_NORMAL;
3609 sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3610 sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3611 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3612 sbi->migration_granularity = sbi->segs_per_sec;
3613 sbi->seq_file_ra_mul = MIN_RA_MUL;
3614 sbi->max_fragment_chunk = DEF_FRAGMENT_SIZE;
3615 sbi->max_fragment_hole = DEF_FRAGMENT_SIZE;
3616 spin_lock_init(&sbi->gc_urgent_high_lock);
3617 atomic64_set(&sbi->current_atomic_write, 0);
3619 sbi->dir_level = DEF_DIR_LEVEL;
3620 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3621 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3622 sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3623 sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3624 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3625 sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3626 DEF_UMOUNT_DISCARD_TIMEOUT;
3627 clear_sbi_flag(sbi, SBI_NEED_FSCK);
3629 for (i = 0; i < NR_COUNT_TYPE; i++)
3630 atomic_set(&sbi->nr_pages[i], 0);
3632 for (i = 0; i < META; i++)
3633 atomic_set(&sbi->wb_sync_req[i], 0);
3635 INIT_LIST_HEAD(&sbi->s_list);
3636 mutex_init(&sbi->umount_mutex);
3637 init_f2fs_rwsem(&sbi->io_order_lock);
3638 spin_lock_init(&sbi->cp_lock);
3640 sbi->dirty_device = 0;
3641 spin_lock_init(&sbi->dev_lock);
3643 init_f2fs_rwsem(&sbi->sb_lock);
3644 init_f2fs_rwsem(&sbi->pin_sem);
3647 static int init_percpu_info(struct f2fs_sb_info *sbi)
3651 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3655 err = percpu_counter_init(&sbi->rf_node_block_count, 0, GFP_KERNEL);
3657 goto err_valid_block;
3659 err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3662 goto err_node_block;
3666 percpu_counter_destroy(&sbi->rf_node_block_count);
3668 percpu_counter_destroy(&sbi->alloc_valid_block_count);
3672 #ifdef CONFIG_BLK_DEV_ZONED
3674 struct f2fs_report_zones_args {
3675 struct f2fs_sb_info *sbi;
3676 struct f2fs_dev_info *dev;
3679 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3682 struct f2fs_report_zones_args *rz_args = data;
3683 block_t unusable_blocks = (zone->len - zone->capacity) >>
3684 F2FS_LOG_SECTORS_PER_BLOCK;
3686 if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3689 set_bit(idx, rz_args->dev->blkz_seq);
3690 if (!rz_args->sbi->unusable_blocks_per_sec) {
3691 rz_args->sbi->unusable_blocks_per_sec = unusable_blocks;
3694 if (rz_args->sbi->unusable_blocks_per_sec != unusable_blocks) {
3695 f2fs_err(rz_args->sbi, "F2FS supports single zone capacity\n");
3701 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3703 struct block_device *bdev = FDEV(devi).bdev;
3704 sector_t nr_sectors = bdev_nr_sectors(bdev);
3705 struct f2fs_report_zones_args rep_zone_arg;
3709 if (!f2fs_sb_has_blkzoned(sbi))
3712 zone_sectors = bdev_zone_sectors(bdev);
3713 if (!is_power_of_2(zone_sectors)) {
3714 f2fs_err(sbi, "F2FS does not support non power of 2 zone sizes\n");
3718 if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3719 SECTOR_TO_BLOCK(zone_sectors))
3721 sbi->blocks_per_blkz = SECTOR_TO_BLOCK(zone_sectors);
3722 if (sbi->log_blocks_per_blkz && sbi->log_blocks_per_blkz !=
3723 __ilog2_u32(sbi->blocks_per_blkz))
3725 sbi->log_blocks_per_blkz = __ilog2_u32(sbi->blocks_per_blkz);
3726 FDEV(devi).nr_blkz = SECTOR_TO_BLOCK(nr_sectors) >>
3727 sbi->log_blocks_per_blkz;
3728 if (nr_sectors & (zone_sectors - 1))
3729 FDEV(devi).nr_blkz++;
3731 FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3732 BITS_TO_LONGS(FDEV(devi).nr_blkz)
3733 * sizeof(unsigned long),
3735 if (!FDEV(devi).blkz_seq)
3738 rep_zone_arg.sbi = sbi;
3739 rep_zone_arg.dev = &FDEV(devi);
3741 ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3750 * Read f2fs raw super block.
3751 * Because we have two copies of super block, so read both of them
3752 * to get the first valid one. If any one of them is broken, we pass
3753 * them recovery flag back to the caller.
3755 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3756 struct f2fs_super_block **raw_super,
3757 int *valid_super_block, int *recovery)
3759 struct super_block *sb = sbi->sb;
3761 struct buffer_head *bh;
3762 struct f2fs_super_block *super;
3765 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3769 for (block = 0; block < 2; block++) {
3770 bh = sb_bread(sb, block);
3772 f2fs_err(sbi, "Unable to read %dth superblock",
3779 /* sanity checking of raw super */
3780 err = sanity_check_raw_super(sbi, bh);
3782 f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3790 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3792 *valid_super_block = block;
3798 /* No valid superblock */
3807 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
3809 struct buffer_head *bh;
3813 if ((recover && f2fs_readonly(sbi->sb)) ||
3814 bdev_read_only(sbi->sb->s_bdev)) {
3815 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3819 /* we should update superblock crc here */
3820 if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
3821 crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
3822 offsetof(struct f2fs_super_block, crc));
3823 F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
3826 /* write back-up superblock first */
3827 bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
3830 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3833 /* if we are in recovery path, skip writing valid superblock */
3837 /* write current valid superblock */
3838 bh = sb_bread(sbi->sb, sbi->valid_super_block);
3841 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3846 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
3848 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3849 unsigned int max_devices = MAX_DEVICES;
3850 unsigned int logical_blksize;
3853 /* Initialize single device information */
3854 if (!RDEV(0).path[0]) {
3855 if (!bdev_is_zoned(sbi->sb->s_bdev))
3861 * Initialize multiple devices information, or single
3862 * zoned block device information.
3864 sbi->devs = f2fs_kzalloc(sbi,
3865 array_size(max_devices,
3866 sizeof(struct f2fs_dev_info)),
3871 logical_blksize = bdev_logical_block_size(sbi->sb->s_bdev);
3872 sbi->aligned_blksize = true;
3874 for (i = 0; i < max_devices; i++) {
3876 if (i > 0 && !RDEV(i).path[0])
3879 if (max_devices == 1) {
3880 /* Single zoned block device mount */
3882 blkdev_get_by_dev(sbi->sb->s_bdev->bd_dev,
3883 sbi->sb->s_mode, sbi->sb->s_type);
3885 /* Multi-device mount */
3886 memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
3887 FDEV(i).total_segments =
3888 le32_to_cpu(RDEV(i).total_segments);
3890 FDEV(i).start_blk = 0;
3891 FDEV(i).end_blk = FDEV(i).start_blk +
3892 (FDEV(i).total_segments <<
3893 sbi->log_blocks_per_seg) - 1 +
3894 le32_to_cpu(raw_super->segment0_blkaddr);
3896 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
3897 FDEV(i).end_blk = FDEV(i).start_blk +
3898 (FDEV(i).total_segments <<
3899 sbi->log_blocks_per_seg) - 1;
3901 FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
3902 sbi->sb->s_mode, sbi->sb->s_type);
3904 if (IS_ERR(FDEV(i).bdev))
3905 return PTR_ERR(FDEV(i).bdev);
3907 /* to release errored devices */
3908 sbi->s_ndevs = i + 1;
3910 if (logical_blksize != bdev_logical_block_size(FDEV(i).bdev))
3911 sbi->aligned_blksize = false;
3913 #ifdef CONFIG_BLK_DEV_ZONED
3914 if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
3915 !f2fs_sb_has_blkzoned(sbi)) {
3916 f2fs_err(sbi, "Zoned block device feature not enabled");
3919 if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
3920 if (init_blkz_info(sbi, i)) {
3921 f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
3924 if (max_devices == 1)
3926 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
3928 FDEV(i).total_segments,
3929 FDEV(i).start_blk, FDEV(i).end_blk,
3930 bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
3931 "Host-aware" : "Host-managed");
3935 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
3937 FDEV(i).total_segments,
3938 FDEV(i).start_blk, FDEV(i).end_blk);
3941 "IO Block Size: %8d KB", F2FS_IO_SIZE_KB(sbi));
3945 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
3947 #if IS_ENABLED(CONFIG_UNICODE)
3948 if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
3949 const struct f2fs_sb_encodings *encoding_info;
3950 struct unicode_map *encoding;
3951 __u16 encoding_flags;
3953 encoding_info = f2fs_sb_read_encoding(sbi->raw_super);
3954 if (!encoding_info) {
3956 "Encoding requested by superblock is unknown");
3960 encoding_flags = le16_to_cpu(sbi->raw_super->s_encoding_flags);
3961 encoding = utf8_load(encoding_info->version);
3962 if (IS_ERR(encoding)) {
3964 "can't mount with superblock charset: %s-%u.%u.%u "
3965 "not supported by the kernel. flags: 0x%x.",
3966 encoding_info->name,
3967 unicode_major(encoding_info->version),
3968 unicode_minor(encoding_info->version),
3969 unicode_rev(encoding_info->version),
3971 return PTR_ERR(encoding);
3973 f2fs_info(sbi, "Using encoding defined by superblock: "
3974 "%s-%u.%u.%u with flags 0x%hx", encoding_info->name,
3975 unicode_major(encoding_info->version),
3976 unicode_minor(encoding_info->version),
3977 unicode_rev(encoding_info->version),
3980 sbi->sb->s_encoding = encoding;
3981 sbi->sb->s_encoding_flags = encoding_flags;
3984 if (f2fs_sb_has_casefold(sbi)) {
3985 f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
3992 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
3994 struct f2fs_sm_info *sm_i = SM_I(sbi);
3996 /* adjust parameters according to the volume size */
3997 if (sm_i->main_segments <= SMALL_VOLUME_SEGMENTS) {
3998 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
3999 if (f2fs_block_unit_discard(sbi))
4000 sm_i->dcc_info->discard_granularity = 1;
4001 sm_i->ipu_policy = 1 << F2FS_IPU_FORCE |
4002 1 << F2FS_IPU_HONOR_OPU_WRITE;
4005 sbi->readdir_ra = 1;
4008 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
4010 struct f2fs_sb_info *sbi;
4011 struct f2fs_super_block *raw_super;
4014 bool skip_recovery = false, need_fsck = false;
4015 char *options = NULL;
4016 int recovery, i, valid_super_block;
4017 struct curseg_info *seg_i;
4023 valid_super_block = -1;
4026 /* allocate memory for f2fs-specific super block info */
4027 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
4033 /* Load the checksum driver */
4034 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
4035 if (IS_ERR(sbi->s_chksum_driver)) {
4036 f2fs_err(sbi, "Cannot load crc32 driver.");
4037 err = PTR_ERR(sbi->s_chksum_driver);
4038 sbi->s_chksum_driver = NULL;
4042 /* set a block size */
4043 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
4044 f2fs_err(sbi, "unable to set blocksize");
4048 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
4053 sb->s_fs_info = sbi;
4054 sbi->raw_super = raw_super;
4056 /* precompute checksum seed for metadata */
4057 if (f2fs_sb_has_inode_chksum(sbi))
4058 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
4059 sizeof(raw_super->uuid));
4061 default_options(sbi);
4062 /* parse mount options */
4063 options = kstrdup((const char *)data, GFP_KERNEL);
4064 if (data && !options) {
4069 err = parse_options(sb, options, false);
4073 sb->s_maxbytes = max_file_blocks(NULL) <<
4074 le32_to_cpu(raw_super->log_blocksize);
4075 sb->s_max_links = F2FS_LINK_MAX;
4077 err = f2fs_setup_casefold(sbi);
4082 sb->dq_op = &f2fs_quota_operations;
4083 sb->s_qcop = &f2fs_quotactl_ops;
4084 sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
4086 if (f2fs_sb_has_quota_ino(sbi)) {
4087 for (i = 0; i < MAXQUOTAS; i++) {
4088 if (f2fs_qf_ino(sbi->sb, i))
4089 sbi->nquota_files++;
4094 sb->s_op = &f2fs_sops;
4095 #ifdef CONFIG_FS_ENCRYPTION
4096 sb->s_cop = &f2fs_cryptops;
4098 #ifdef CONFIG_FS_VERITY
4099 sb->s_vop = &f2fs_verityops;
4101 sb->s_xattr = f2fs_xattr_handlers;
4102 sb->s_export_op = &f2fs_export_ops;
4103 sb->s_magic = F2FS_SUPER_MAGIC;
4104 sb->s_time_gran = 1;
4105 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
4106 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
4107 memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
4108 sb->s_iflags |= SB_I_CGROUPWB;
4110 /* init f2fs-specific super block info */
4111 sbi->valid_super_block = valid_super_block;
4112 init_f2fs_rwsem(&sbi->gc_lock);
4113 mutex_init(&sbi->writepages);
4114 init_f2fs_rwsem(&sbi->cp_global_sem);
4115 init_f2fs_rwsem(&sbi->node_write);
4116 init_f2fs_rwsem(&sbi->node_change);
4118 /* disallow all the data/node/meta page writes */
4119 set_sbi_flag(sbi, SBI_POR_DOING);
4120 spin_lock_init(&sbi->stat_lock);
4122 err = f2fs_init_write_merge_io(sbi);
4126 init_f2fs_rwsem(&sbi->cp_rwsem);
4127 init_f2fs_rwsem(&sbi->quota_sem);
4128 init_waitqueue_head(&sbi->cp_wait);
4131 err = f2fs_init_iostat(sbi);
4135 err = init_percpu_info(sbi);
4139 if (F2FS_IO_ALIGNED(sbi)) {
4140 sbi->write_io_dummy =
4141 mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
4142 if (!sbi->write_io_dummy) {
4148 /* init per sbi slab cache */
4149 err = f2fs_init_xattr_caches(sbi);
4152 err = f2fs_init_page_array_cache(sbi);
4154 goto free_xattr_cache;
4156 /* get an inode for meta space */
4157 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
4158 if (IS_ERR(sbi->meta_inode)) {
4159 f2fs_err(sbi, "Failed to read F2FS meta data inode");
4160 err = PTR_ERR(sbi->meta_inode);
4161 goto free_page_array_cache;
4164 err = f2fs_get_valid_checkpoint(sbi);
4166 f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
4167 goto free_meta_inode;
4170 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
4171 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
4172 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
4173 set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4174 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
4177 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
4178 set_sbi_flag(sbi, SBI_NEED_FSCK);
4180 /* Initialize device list */
4181 err = f2fs_scan_devices(sbi);
4183 f2fs_err(sbi, "Failed to find devices");
4187 err = f2fs_init_post_read_wq(sbi);
4189 f2fs_err(sbi, "Failed to initialize post read workqueue");
4193 sbi->total_valid_node_count =
4194 le32_to_cpu(sbi->ckpt->valid_node_count);
4195 percpu_counter_set(&sbi->total_valid_inode_count,
4196 le32_to_cpu(sbi->ckpt->valid_inode_count));
4197 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
4198 sbi->total_valid_block_count =
4199 le64_to_cpu(sbi->ckpt->valid_block_count);
4200 sbi->last_valid_block_count = sbi->total_valid_block_count;
4201 sbi->reserved_blocks = 0;
4202 sbi->current_reserved_blocks = 0;
4203 limit_reserve_root(sbi);
4204 adjust_unusable_cap_perc(sbi);
4206 for (i = 0; i < NR_INODE_TYPE; i++) {
4207 INIT_LIST_HEAD(&sbi->inode_list[i]);
4208 spin_lock_init(&sbi->inode_lock[i]);
4210 mutex_init(&sbi->flush_lock);
4212 f2fs_init_extent_cache_info(sbi);
4214 f2fs_init_ino_entry_info(sbi);
4216 f2fs_init_fsync_node_info(sbi);
4218 /* setup checkpoint request control and start checkpoint issue thread */
4219 f2fs_init_ckpt_req_control(sbi);
4220 if (!f2fs_readonly(sb) && !test_opt(sbi, DISABLE_CHECKPOINT) &&
4221 test_opt(sbi, MERGE_CHECKPOINT)) {
4222 err = f2fs_start_ckpt_thread(sbi);
4225 "Failed to start F2FS issue_checkpoint_thread (%d)",
4227 goto stop_ckpt_thread;
4231 /* setup f2fs internal modules */
4232 err = f2fs_build_segment_manager(sbi);
4234 f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
4238 err = f2fs_build_node_manager(sbi);
4240 f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
4245 err = adjust_reserved_segment(sbi);
4249 /* For write statistics */
4250 sbi->sectors_written_start = f2fs_get_sectors_written(sbi);
4252 /* Read accumulated write IO statistics if exists */
4253 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
4254 if (__exist_node_summaries(sbi))
4255 sbi->kbytes_written =
4256 le64_to_cpu(seg_i->journal->info.kbytes_written);
4258 f2fs_build_gc_manager(sbi);
4260 err = f2fs_build_stats(sbi);
4264 /* get an inode for node space */
4265 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
4266 if (IS_ERR(sbi->node_inode)) {
4267 f2fs_err(sbi, "Failed to read node inode");
4268 err = PTR_ERR(sbi->node_inode);
4272 /* read root inode and dentry */
4273 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
4275 f2fs_err(sbi, "Failed to read root inode");
4276 err = PTR_ERR(root);
4277 goto free_node_inode;
4279 if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
4280 !root->i_size || !root->i_nlink) {
4283 goto free_node_inode;
4286 sb->s_root = d_make_root(root); /* allocate root dentry */
4289 goto free_node_inode;
4292 err = f2fs_init_compress_inode(sbi);
4294 goto free_root_inode;
4296 err = f2fs_register_sysfs(sbi);
4298 goto free_compress_inode;
4301 /* Enable quota usage during mount */
4302 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
4303 err = f2fs_enable_quotas(sb);
4305 f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
4308 /* if there are any orphan inodes, free them */
4309 err = f2fs_recover_orphan_inodes(sbi);
4313 if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
4314 goto reset_checkpoint;
4316 /* recover fsynced data */
4317 if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
4318 !test_opt(sbi, NORECOVERY)) {
4320 * mount should be failed, when device has readonly mode, and
4321 * previous checkpoint was not done by clean system shutdown.
4323 if (f2fs_hw_is_readonly(sbi)) {
4324 if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4325 err = f2fs_recover_fsync_data(sbi, true);
4328 f2fs_err(sbi, "Need to recover fsync data, but "
4329 "write access unavailable, please try "
4330 "mount w/ disable_roll_forward or norecovery");
4335 f2fs_info(sbi, "write access unavailable, skipping recovery");
4336 goto reset_checkpoint;
4340 set_sbi_flag(sbi, SBI_NEED_FSCK);
4343 goto reset_checkpoint;
4345 err = f2fs_recover_fsync_data(sbi, false);
4348 skip_recovery = true;
4350 f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
4355 err = f2fs_recover_fsync_data(sbi, true);
4357 if (!f2fs_readonly(sb) && err > 0) {
4359 f2fs_err(sbi, "Need to recover fsync data");
4365 * If the f2fs is not readonly and fsync data recovery succeeds,
4366 * check zoned block devices' write pointer consistency.
4368 if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
4369 err = f2fs_check_write_pointer(sbi);
4375 f2fs_init_inmem_curseg(sbi);
4377 /* f2fs_recover_fsync_data() cleared this already */
4378 clear_sbi_flag(sbi, SBI_POR_DOING);
4380 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
4381 err = f2fs_disable_checkpoint(sbi);
4383 goto sync_free_meta;
4384 } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
4385 f2fs_enable_checkpoint(sbi);
4389 * If filesystem is not mounted as read-only then
4390 * do start the gc_thread.
4392 if ((F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF ||
4393 test_opt(sbi, GC_MERGE)) && !f2fs_readonly(sb)) {
4394 /* After POR, we can run background GC thread.*/
4395 err = f2fs_start_gc_thread(sbi);
4397 goto sync_free_meta;
4401 /* recover broken superblock */
4403 err = f2fs_commit_super(sbi, true);
4404 f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
4405 sbi->valid_super_block ? 1 : 2, err);
4408 f2fs_join_shrinker(sbi);
4410 f2fs_tuning_parameters(sbi);
4412 f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
4413 cur_cp_version(F2FS_CKPT(sbi)));
4414 f2fs_update_time(sbi, CP_TIME);
4415 f2fs_update_time(sbi, REQ_TIME);
4416 clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4420 /* safe to flush all the data */
4421 sync_filesystem(sbi->sb);
4426 f2fs_truncate_quota_inode_pages(sb);
4427 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
4428 f2fs_quota_off_umount(sbi->sb);
4431 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
4432 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
4433 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
4434 * falls into an infinite loop in f2fs_sync_meta_pages().
4436 truncate_inode_pages_final(META_MAPPING(sbi));
4437 /* evict some inodes being cached by GC */
4439 f2fs_unregister_sysfs(sbi);
4440 free_compress_inode:
4441 f2fs_destroy_compress_inode(sbi);
4446 f2fs_release_ino_entry(sbi, true);
4447 truncate_inode_pages_final(NODE_MAPPING(sbi));
4448 iput(sbi->node_inode);
4449 sbi->node_inode = NULL;
4451 f2fs_destroy_stats(sbi);
4453 /* stop discard thread before destroying node manager */
4454 f2fs_stop_discard_thread(sbi);
4455 f2fs_destroy_node_manager(sbi);
4457 f2fs_destroy_segment_manager(sbi);
4458 f2fs_destroy_post_read_wq(sbi);
4460 f2fs_stop_ckpt_thread(sbi);
4462 destroy_device_list(sbi);
4465 make_bad_inode(sbi->meta_inode);
4466 iput(sbi->meta_inode);
4467 sbi->meta_inode = NULL;
4468 free_page_array_cache:
4469 f2fs_destroy_page_array_cache(sbi);
4471 f2fs_destroy_xattr_caches(sbi);
4473 mempool_destroy(sbi->write_io_dummy);
4475 destroy_percpu_info(sbi);
4477 f2fs_destroy_iostat(sbi);
4479 for (i = 0; i < NR_PAGE_TYPE; i++)
4480 kvfree(sbi->write_io[i]);
4482 #if IS_ENABLED(CONFIG_UNICODE)
4483 utf8_unload(sb->s_encoding);
4484 sb->s_encoding = NULL;
4488 for (i = 0; i < MAXQUOTAS; i++)
4489 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
4491 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
4496 if (sbi->s_chksum_driver)
4497 crypto_free_shash(sbi->s_chksum_driver);
4500 /* give only one another chance */
4501 if (retry_cnt > 0 && skip_recovery) {
4503 shrink_dcache_sb(sb);
4509 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
4510 const char *dev_name, void *data)
4512 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
4515 static void kill_f2fs_super(struct super_block *sb)
4518 struct f2fs_sb_info *sbi = F2FS_SB(sb);
4520 set_sbi_flag(sbi, SBI_IS_CLOSE);
4521 f2fs_stop_gc_thread(sbi);
4522 f2fs_stop_discard_thread(sbi);
4524 #ifdef CONFIG_F2FS_FS_COMPRESSION
4526 * latter evict_inode() can bypass checking and invalidating
4527 * compress inode cache.
4529 if (test_opt(sbi, COMPRESS_CACHE))
4530 truncate_inode_pages_final(COMPRESS_MAPPING(sbi));
4533 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
4534 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4535 struct cp_control cpc = {
4536 .reason = CP_UMOUNT,
4538 f2fs_write_checkpoint(sbi, &cpc);
4541 if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
4542 sb->s_flags &= ~SB_RDONLY;
4544 kill_block_super(sb);
4547 static struct file_system_type f2fs_fs_type = {
4548 .owner = THIS_MODULE,
4550 .mount = f2fs_mount,
4551 .kill_sb = kill_f2fs_super,
4552 .fs_flags = FS_REQUIRES_DEV | FS_ALLOW_IDMAP,
4554 MODULE_ALIAS_FS("f2fs");
4556 static int __init init_inodecache(void)
4558 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
4559 sizeof(struct f2fs_inode_info), 0,
4560 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
4561 if (!f2fs_inode_cachep)
4566 static void destroy_inodecache(void)
4569 * Make sure all delayed rcu free inodes are flushed before we
4573 kmem_cache_destroy(f2fs_inode_cachep);
4576 static int __init init_f2fs_fs(void)
4580 if (PAGE_SIZE != F2FS_BLKSIZE) {
4581 printk("F2FS not supported on PAGE_SIZE(%lu) != %d\n",
4582 PAGE_SIZE, F2FS_BLKSIZE);
4586 err = init_inodecache();
4589 err = f2fs_create_node_manager_caches();
4591 goto free_inodecache;
4592 err = f2fs_create_segment_manager_caches();
4594 goto free_node_manager_caches;
4595 err = f2fs_create_checkpoint_caches();
4597 goto free_segment_manager_caches;
4598 err = f2fs_create_recovery_cache();
4600 goto free_checkpoint_caches;
4601 err = f2fs_create_extent_cache();
4603 goto free_recovery_cache;
4604 err = f2fs_create_garbage_collection_cache();
4606 goto free_extent_cache;
4607 err = f2fs_init_sysfs();
4609 goto free_garbage_collection_cache;
4610 err = register_shrinker(&f2fs_shrinker_info, "f2fs-shrinker");
4613 err = register_filesystem(&f2fs_fs_type);
4616 f2fs_create_root_stats();
4617 err = f2fs_init_post_read_processing();
4619 goto free_root_stats;
4620 err = f2fs_init_iostat_processing();
4622 goto free_post_read;
4623 err = f2fs_init_bio_entry_cache();
4626 err = f2fs_init_bioset();
4628 goto free_bio_enrty_cache;
4629 err = f2fs_init_compress_mempool();
4632 err = f2fs_init_compress_cache();
4634 goto free_compress_mempool;
4635 err = f2fs_create_casefold_cache();
4637 goto free_compress_cache;
4639 free_compress_cache:
4640 f2fs_destroy_compress_cache();
4641 free_compress_mempool:
4642 f2fs_destroy_compress_mempool();
4644 f2fs_destroy_bioset();
4645 free_bio_enrty_cache:
4646 f2fs_destroy_bio_entry_cache();
4648 f2fs_destroy_iostat_processing();
4650 f2fs_destroy_post_read_processing();
4652 f2fs_destroy_root_stats();
4653 unregister_filesystem(&f2fs_fs_type);
4655 unregister_shrinker(&f2fs_shrinker_info);
4658 free_garbage_collection_cache:
4659 f2fs_destroy_garbage_collection_cache();
4661 f2fs_destroy_extent_cache();
4662 free_recovery_cache:
4663 f2fs_destroy_recovery_cache();
4664 free_checkpoint_caches:
4665 f2fs_destroy_checkpoint_caches();
4666 free_segment_manager_caches:
4667 f2fs_destroy_segment_manager_caches();
4668 free_node_manager_caches:
4669 f2fs_destroy_node_manager_caches();
4671 destroy_inodecache();
4676 static void __exit exit_f2fs_fs(void)
4678 f2fs_destroy_casefold_cache();
4679 f2fs_destroy_compress_cache();
4680 f2fs_destroy_compress_mempool();
4681 f2fs_destroy_bioset();
4682 f2fs_destroy_bio_entry_cache();
4683 f2fs_destroy_iostat_processing();
4684 f2fs_destroy_post_read_processing();
4685 f2fs_destroy_root_stats();
4686 unregister_filesystem(&f2fs_fs_type);
4687 unregister_shrinker(&f2fs_shrinker_info);
4689 f2fs_destroy_garbage_collection_cache();
4690 f2fs_destroy_extent_cache();
4691 f2fs_destroy_recovery_cache();
4692 f2fs_destroy_checkpoint_caches();
4693 f2fs_destroy_segment_manager_caches();
4694 f2fs_destroy_node_manager_caches();
4695 destroy_inodecache();
4698 module_init(init_f2fs_fs)
4699 module_exit(exit_f2fs_fs)
4701 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
4702 MODULE_DESCRIPTION("Flash Friendly File System");
4703 MODULE_LICENSE("GPL");
4704 MODULE_SOFTDEP("pre: crc32");