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/statfs.h>
12 #include <linux/buffer_head.h>
13 #include <linux/backing-dev.h>
14 #include <linux/kthread.h>
15 #include <linux/parser.h>
16 #include <linux/mount.h>
17 #include <linux/seq_file.h>
18 #include <linux/proc_fs.h>
19 #include <linux/random.h>
20 #include <linux/exportfs.h>
21 #include <linux/blkdev.h>
22 #include <linux/quotaops.h>
23 #include <linux/f2fs_fs.h>
24 #include <linux/sysfs.h>
25 #include <linux/quota.h>
26 #include <linux/unicode.h>
27 #include <linux/part_stat.h>
36 #define CREATE_TRACE_POINTS
37 #include <trace/events/f2fs.h>
39 static struct kmem_cache *f2fs_inode_cachep;
41 #ifdef CONFIG_F2FS_FAULT_INJECTION
43 const char *f2fs_fault_name[FAULT_MAX] = {
44 [FAULT_KMALLOC] = "kmalloc",
45 [FAULT_KVMALLOC] = "kvmalloc",
46 [FAULT_PAGE_ALLOC] = "page alloc",
47 [FAULT_PAGE_GET] = "page get",
48 [FAULT_ALLOC_BIO] = "alloc bio",
49 [FAULT_ALLOC_NID] = "alloc nid",
50 [FAULT_ORPHAN] = "orphan",
51 [FAULT_BLOCK] = "no more block",
52 [FAULT_DIR_DEPTH] = "too big dir depth",
53 [FAULT_EVICT_INODE] = "evict_inode fail",
54 [FAULT_TRUNCATE] = "truncate fail",
55 [FAULT_READ_IO] = "read IO error",
56 [FAULT_CHECKPOINT] = "checkpoint error",
57 [FAULT_DISCARD] = "discard error",
58 [FAULT_WRITE_IO] = "write IO error",
61 void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
64 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
67 atomic_set(&ffi->inject_ops, 0);
68 ffi->inject_rate = rate;
72 ffi->inject_type = type;
75 memset(ffi, 0, sizeof(struct f2fs_fault_info));
79 /* f2fs-wide shrinker description */
80 static struct shrinker f2fs_shrinker_info = {
81 .scan_objects = f2fs_shrink_scan,
82 .count_objects = f2fs_shrink_count,
83 .seeks = DEFAULT_SEEKS,
88 Opt_disable_roll_forward,
99 Opt_disable_ext_identify,
102 Opt_inline_xattr_size,
140 Opt_test_dummy_encryption,
142 Opt_checkpoint_disable,
143 Opt_checkpoint_disable_cap,
144 Opt_checkpoint_disable_cap_perc,
145 Opt_checkpoint_enable,
146 Opt_compress_algorithm,
147 Opt_compress_log_size,
148 Opt_compress_extension,
152 static match_table_t f2fs_tokens = {
153 {Opt_gc_background, "background_gc=%s"},
154 {Opt_disable_roll_forward, "disable_roll_forward"},
155 {Opt_norecovery, "norecovery"},
156 {Opt_discard, "discard"},
157 {Opt_nodiscard, "nodiscard"},
158 {Opt_noheap, "no_heap"},
160 {Opt_user_xattr, "user_xattr"},
161 {Opt_nouser_xattr, "nouser_xattr"},
163 {Opt_noacl, "noacl"},
164 {Opt_active_logs, "active_logs=%u"},
165 {Opt_disable_ext_identify, "disable_ext_identify"},
166 {Opt_inline_xattr, "inline_xattr"},
167 {Opt_noinline_xattr, "noinline_xattr"},
168 {Opt_inline_xattr_size, "inline_xattr_size=%u"},
169 {Opt_inline_data, "inline_data"},
170 {Opt_inline_dentry, "inline_dentry"},
171 {Opt_noinline_dentry, "noinline_dentry"},
172 {Opt_flush_merge, "flush_merge"},
173 {Opt_noflush_merge, "noflush_merge"},
174 {Opt_nobarrier, "nobarrier"},
175 {Opt_fastboot, "fastboot"},
176 {Opt_extent_cache, "extent_cache"},
177 {Opt_noextent_cache, "noextent_cache"},
178 {Opt_noinline_data, "noinline_data"},
179 {Opt_data_flush, "data_flush"},
180 {Opt_reserve_root, "reserve_root=%u"},
181 {Opt_resgid, "resgid=%u"},
182 {Opt_resuid, "resuid=%u"},
183 {Opt_mode, "mode=%s"},
184 {Opt_io_size_bits, "io_bits=%u"},
185 {Opt_fault_injection, "fault_injection=%u"},
186 {Opt_fault_type, "fault_type=%u"},
187 {Opt_lazytime, "lazytime"},
188 {Opt_nolazytime, "nolazytime"},
189 {Opt_quota, "quota"},
190 {Opt_noquota, "noquota"},
191 {Opt_usrquota, "usrquota"},
192 {Opt_grpquota, "grpquota"},
193 {Opt_prjquota, "prjquota"},
194 {Opt_usrjquota, "usrjquota=%s"},
195 {Opt_grpjquota, "grpjquota=%s"},
196 {Opt_prjjquota, "prjjquota=%s"},
197 {Opt_offusrjquota, "usrjquota="},
198 {Opt_offgrpjquota, "grpjquota="},
199 {Opt_offprjjquota, "prjjquota="},
200 {Opt_jqfmt_vfsold, "jqfmt=vfsold"},
201 {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
202 {Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
203 {Opt_whint, "whint_mode=%s"},
204 {Opt_alloc, "alloc_mode=%s"},
205 {Opt_fsync, "fsync_mode=%s"},
206 {Opt_test_dummy_encryption, "test_dummy_encryption=%s"},
207 {Opt_test_dummy_encryption, "test_dummy_encryption"},
208 {Opt_inlinecrypt, "inlinecrypt"},
209 {Opt_checkpoint_disable, "checkpoint=disable"},
210 {Opt_checkpoint_disable_cap, "checkpoint=disable:%u"},
211 {Opt_checkpoint_disable_cap_perc, "checkpoint=disable:%u%%"},
212 {Opt_checkpoint_enable, "checkpoint=enable"},
213 {Opt_compress_algorithm, "compress_algorithm=%s"},
214 {Opt_compress_log_size, "compress_log_size=%u"},
215 {Opt_compress_extension, "compress_extension=%s"},
219 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...)
221 struct va_format vaf;
227 level = printk_get_level(fmt);
228 vaf.fmt = printk_skip_level(fmt);
230 printk("%c%cF2FS-fs (%s): %pV\n",
231 KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
236 #ifdef CONFIG_UNICODE
237 static const struct f2fs_sb_encodings {
241 } f2fs_sb_encoding_map[] = {
242 {F2FS_ENC_UTF8_12_1, "utf8", "12.1.0"},
245 static int f2fs_sb_read_encoding(const struct f2fs_super_block *sb,
246 const struct f2fs_sb_encodings **encoding,
249 __u16 magic = le16_to_cpu(sb->s_encoding);
252 for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
253 if (magic == f2fs_sb_encoding_map[i].magic)
256 if (i >= ARRAY_SIZE(f2fs_sb_encoding_map))
259 *encoding = &f2fs_sb_encoding_map[i];
260 *flags = le16_to_cpu(sb->s_encoding_flags);
266 static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
268 block_t limit = min((sbi->user_block_count << 1) / 1000,
269 sbi->user_block_count - sbi->reserved_blocks);
272 if (test_opt(sbi, RESERVE_ROOT) &&
273 F2FS_OPTION(sbi).root_reserved_blocks > limit) {
274 F2FS_OPTION(sbi).root_reserved_blocks = limit;
275 f2fs_info(sbi, "Reduce reserved blocks for root = %u",
276 F2FS_OPTION(sbi).root_reserved_blocks);
278 if (!test_opt(sbi, RESERVE_ROOT) &&
279 (!uid_eq(F2FS_OPTION(sbi).s_resuid,
280 make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
281 !gid_eq(F2FS_OPTION(sbi).s_resgid,
282 make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
283 f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
284 from_kuid_munged(&init_user_ns,
285 F2FS_OPTION(sbi).s_resuid),
286 from_kgid_munged(&init_user_ns,
287 F2FS_OPTION(sbi).s_resgid));
290 static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
292 if (!F2FS_OPTION(sbi).unusable_cap_perc)
295 if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
296 F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
298 F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
299 F2FS_OPTION(sbi).unusable_cap_perc;
301 f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
302 F2FS_OPTION(sbi).unusable_cap,
303 F2FS_OPTION(sbi).unusable_cap_perc);
306 static void init_once(void *foo)
308 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
310 inode_init_once(&fi->vfs_inode);
314 static const char * const quotatypes[] = INITQFNAMES;
315 #define QTYPE2NAME(t) (quotatypes[t])
316 static int f2fs_set_qf_name(struct super_block *sb, int qtype,
319 struct f2fs_sb_info *sbi = F2FS_SB(sb);
323 if (sb_any_quota_loaded(sb) && !F2FS_OPTION(sbi).s_qf_names[qtype]) {
324 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
327 if (f2fs_sb_has_quota_ino(sbi)) {
328 f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
332 qname = match_strdup(args);
334 f2fs_err(sbi, "Not enough memory for storing quotafile name");
337 if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
338 if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
341 f2fs_err(sbi, "%s quota file already specified",
345 if (strchr(qname, '/')) {
346 f2fs_err(sbi, "quotafile must be on filesystem root");
349 F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
357 static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
359 struct f2fs_sb_info *sbi = F2FS_SB(sb);
361 if (sb_any_quota_loaded(sb) && F2FS_OPTION(sbi).s_qf_names[qtype]) {
362 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
365 kfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
366 F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
370 static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
373 * We do the test below only for project quotas. 'usrquota' and
374 * 'grpquota' mount options are allowed even without quota feature
375 * to support legacy quotas in quota files.
377 if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi)) {
378 f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
381 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
382 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
383 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) {
384 if (test_opt(sbi, USRQUOTA) &&
385 F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
386 clear_opt(sbi, USRQUOTA);
388 if (test_opt(sbi, GRPQUOTA) &&
389 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
390 clear_opt(sbi, GRPQUOTA);
392 if (test_opt(sbi, PRJQUOTA) &&
393 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
394 clear_opt(sbi, PRJQUOTA);
396 if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
397 test_opt(sbi, PRJQUOTA)) {
398 f2fs_err(sbi, "old and new quota format mixing");
402 if (!F2FS_OPTION(sbi).s_jquota_fmt) {
403 f2fs_err(sbi, "journaled quota format not specified");
408 if (f2fs_sb_has_quota_ino(sbi) && F2FS_OPTION(sbi).s_jquota_fmt) {
409 f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
410 F2FS_OPTION(sbi).s_jquota_fmt = 0;
416 static int f2fs_set_test_dummy_encryption(struct super_block *sb,
418 const substring_t *arg,
421 struct f2fs_sb_info *sbi = F2FS_SB(sb);
422 #ifdef CONFIG_FS_ENCRYPTION
425 if (!f2fs_sb_has_encrypt(sbi)) {
426 f2fs_err(sbi, "Encrypt feature is off");
431 * This mount option is just for testing, and it's not worthwhile to
432 * implement the extra complexity (e.g. RCU protection) that would be
433 * needed to allow it to be set or changed during remount. We do allow
434 * it to be specified during remount, but only if there is no change.
436 if (is_remount && !F2FS_OPTION(sbi).dummy_enc_ctx.ctx) {
437 f2fs_warn(sbi, "Can't set test_dummy_encryption on remount");
440 err = fscrypt_set_test_dummy_encryption(
441 sb, arg, &F2FS_OPTION(sbi).dummy_enc_ctx);
445 "Can't change test_dummy_encryption on remount");
446 else if (err == -EINVAL)
447 f2fs_warn(sbi, "Value of option \"%s\" is unrecognized",
450 f2fs_warn(sbi, "Error processing option \"%s\" [%d]",
454 f2fs_warn(sbi, "Test dummy encryption mode enabled");
456 f2fs_warn(sbi, "Test dummy encryption mount option ignored");
461 static int parse_options(struct super_block *sb, char *options, bool is_remount)
463 struct f2fs_sb_info *sbi = F2FS_SB(sb);
464 substring_t args[MAX_OPT_ARGS];
465 #ifdef CONFIG_F2FS_FS_COMPRESSION
466 unsigned char (*ext)[F2FS_EXTENSION_LEN];
478 while ((p = strsep(&options, ",")) != NULL) {
483 * Initialize args struct so we know whether arg was
484 * found; some options take optional arguments.
486 args[0].to = args[0].from = NULL;
487 token = match_token(p, f2fs_tokens, args);
490 case Opt_gc_background:
491 name = match_strdup(&args[0]);
495 if (!strcmp(name, "on")) {
496 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
497 } else if (!strcmp(name, "off")) {
498 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_OFF;
499 } else if (!strcmp(name, "sync")) {
500 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_SYNC;
507 case Opt_disable_roll_forward:
508 set_opt(sbi, DISABLE_ROLL_FORWARD);
511 /* this option mounts f2fs with ro */
512 set_opt(sbi, NORECOVERY);
513 if (!f2fs_readonly(sb))
517 set_opt(sbi, DISCARD);
520 if (f2fs_sb_has_blkzoned(sbi)) {
521 f2fs_warn(sbi, "discard is required for zoned block devices");
524 clear_opt(sbi, DISCARD);
527 set_opt(sbi, NOHEAP);
530 clear_opt(sbi, NOHEAP);
532 #ifdef CONFIG_F2FS_FS_XATTR
534 set_opt(sbi, XATTR_USER);
536 case Opt_nouser_xattr:
537 clear_opt(sbi, XATTR_USER);
539 case Opt_inline_xattr:
540 set_opt(sbi, INLINE_XATTR);
542 case Opt_noinline_xattr:
543 clear_opt(sbi, INLINE_XATTR);
545 case Opt_inline_xattr_size:
546 if (args->from && match_int(args, &arg))
548 set_opt(sbi, INLINE_XATTR_SIZE);
549 F2FS_OPTION(sbi).inline_xattr_size = arg;
553 f2fs_info(sbi, "user_xattr options not supported");
555 case Opt_nouser_xattr:
556 f2fs_info(sbi, "nouser_xattr options not supported");
558 case Opt_inline_xattr:
559 f2fs_info(sbi, "inline_xattr options not supported");
561 case Opt_noinline_xattr:
562 f2fs_info(sbi, "noinline_xattr options not supported");
565 #ifdef CONFIG_F2FS_FS_POSIX_ACL
567 set_opt(sbi, POSIX_ACL);
570 clear_opt(sbi, POSIX_ACL);
574 f2fs_info(sbi, "acl options not supported");
577 f2fs_info(sbi, "noacl options not supported");
580 case Opt_active_logs:
581 if (args->from && match_int(args, &arg))
583 if (arg != 2 && arg != 4 && arg != NR_CURSEG_TYPE)
585 F2FS_OPTION(sbi).active_logs = arg;
587 case Opt_disable_ext_identify:
588 set_opt(sbi, DISABLE_EXT_IDENTIFY);
590 case Opt_inline_data:
591 set_opt(sbi, INLINE_DATA);
593 case Opt_inline_dentry:
594 set_opt(sbi, INLINE_DENTRY);
596 case Opt_noinline_dentry:
597 clear_opt(sbi, INLINE_DENTRY);
599 case Opt_flush_merge:
600 set_opt(sbi, FLUSH_MERGE);
602 case Opt_noflush_merge:
603 clear_opt(sbi, FLUSH_MERGE);
606 set_opt(sbi, NOBARRIER);
609 set_opt(sbi, FASTBOOT);
611 case Opt_extent_cache:
612 set_opt(sbi, EXTENT_CACHE);
614 case Opt_noextent_cache:
615 clear_opt(sbi, EXTENT_CACHE);
617 case Opt_noinline_data:
618 clear_opt(sbi, INLINE_DATA);
621 set_opt(sbi, DATA_FLUSH);
623 case Opt_reserve_root:
624 if (args->from && match_int(args, &arg))
626 if (test_opt(sbi, RESERVE_ROOT)) {
627 f2fs_info(sbi, "Preserve previous reserve_root=%u",
628 F2FS_OPTION(sbi).root_reserved_blocks);
630 F2FS_OPTION(sbi).root_reserved_blocks = arg;
631 set_opt(sbi, RESERVE_ROOT);
635 if (args->from && match_int(args, &arg))
637 uid = make_kuid(current_user_ns(), arg);
638 if (!uid_valid(uid)) {
639 f2fs_err(sbi, "Invalid uid value %d", arg);
642 F2FS_OPTION(sbi).s_resuid = uid;
645 if (args->from && match_int(args, &arg))
647 gid = make_kgid(current_user_ns(), arg);
648 if (!gid_valid(gid)) {
649 f2fs_err(sbi, "Invalid gid value %d", arg);
652 F2FS_OPTION(sbi).s_resgid = gid;
655 name = match_strdup(&args[0]);
659 if (!strcmp(name, "adaptive")) {
660 if (f2fs_sb_has_blkzoned(sbi)) {
661 f2fs_warn(sbi, "adaptive mode is not allowed with zoned block device feature");
665 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
666 } else if (!strcmp(name, "lfs")) {
667 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
674 case Opt_io_size_bits:
675 if (args->from && match_int(args, &arg))
677 if (arg <= 0 || arg > __ilog2_u32(BIO_MAX_PAGES)) {
678 f2fs_warn(sbi, "Not support %d, larger than %d",
679 1 << arg, BIO_MAX_PAGES);
682 F2FS_OPTION(sbi).write_io_size_bits = arg;
684 #ifdef CONFIG_F2FS_FAULT_INJECTION
685 case Opt_fault_injection:
686 if (args->from && match_int(args, &arg))
688 f2fs_build_fault_attr(sbi, arg, F2FS_ALL_FAULT_TYPE);
689 set_opt(sbi, FAULT_INJECTION);
693 if (args->from && match_int(args, &arg))
695 f2fs_build_fault_attr(sbi, 0, arg);
696 set_opt(sbi, FAULT_INJECTION);
699 case Opt_fault_injection:
700 f2fs_info(sbi, "fault_injection options not supported");
704 f2fs_info(sbi, "fault_type options not supported");
708 sb->s_flags |= SB_LAZYTIME;
711 sb->s_flags &= ~SB_LAZYTIME;
716 set_opt(sbi, USRQUOTA);
719 set_opt(sbi, GRPQUOTA);
722 set_opt(sbi, PRJQUOTA);
725 ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
730 ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
735 ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
739 case Opt_offusrjquota:
740 ret = f2fs_clear_qf_name(sb, USRQUOTA);
744 case Opt_offgrpjquota:
745 ret = f2fs_clear_qf_name(sb, GRPQUOTA);
749 case Opt_offprjjquota:
750 ret = f2fs_clear_qf_name(sb, PRJQUOTA);
754 case Opt_jqfmt_vfsold:
755 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
757 case Opt_jqfmt_vfsv0:
758 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
760 case Opt_jqfmt_vfsv1:
761 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
764 clear_opt(sbi, QUOTA);
765 clear_opt(sbi, USRQUOTA);
766 clear_opt(sbi, GRPQUOTA);
767 clear_opt(sbi, PRJQUOTA);
777 case Opt_offusrjquota:
778 case Opt_offgrpjquota:
779 case Opt_offprjjquota:
780 case Opt_jqfmt_vfsold:
781 case Opt_jqfmt_vfsv0:
782 case Opt_jqfmt_vfsv1:
784 f2fs_info(sbi, "quota operations not supported");
788 name = match_strdup(&args[0]);
791 if (!strcmp(name, "user-based")) {
792 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_USER;
793 } else if (!strcmp(name, "off")) {
794 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
795 } else if (!strcmp(name, "fs-based")) {
796 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_FS;
804 name = match_strdup(&args[0]);
808 if (!strcmp(name, "default")) {
809 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
810 } else if (!strcmp(name, "reuse")) {
811 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
819 name = match_strdup(&args[0]);
822 if (!strcmp(name, "posix")) {
823 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
824 } else if (!strcmp(name, "strict")) {
825 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_STRICT;
826 } else if (!strcmp(name, "nobarrier")) {
827 F2FS_OPTION(sbi).fsync_mode =
828 FSYNC_MODE_NOBARRIER;
835 case Opt_test_dummy_encryption:
836 ret = f2fs_set_test_dummy_encryption(sb, p, &args[0],
841 case Opt_inlinecrypt:
842 #ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
843 sb->s_flags |= SB_INLINECRYPT;
845 f2fs_info(sbi, "inline encryption not supported");
848 case Opt_checkpoint_disable_cap_perc:
849 if (args->from && match_int(args, &arg))
851 if (arg < 0 || arg > 100)
853 F2FS_OPTION(sbi).unusable_cap_perc = arg;
854 set_opt(sbi, DISABLE_CHECKPOINT);
856 case Opt_checkpoint_disable_cap:
857 if (args->from && match_int(args, &arg))
859 F2FS_OPTION(sbi).unusable_cap = arg;
860 set_opt(sbi, DISABLE_CHECKPOINT);
862 case Opt_checkpoint_disable:
863 set_opt(sbi, DISABLE_CHECKPOINT);
865 case Opt_checkpoint_enable:
866 clear_opt(sbi, DISABLE_CHECKPOINT);
868 #ifdef CONFIG_F2FS_FS_COMPRESSION
869 case Opt_compress_algorithm:
870 if (!f2fs_sb_has_compression(sbi)) {
871 f2fs_err(sbi, "Compression feature if off");
874 name = match_strdup(&args[0]);
877 if (!strcmp(name, "lzo")) {
878 F2FS_OPTION(sbi).compress_algorithm =
880 } else if (!strcmp(name, "lz4")) {
881 F2FS_OPTION(sbi).compress_algorithm =
883 } else if (!strcmp(name, "zstd")) {
884 F2FS_OPTION(sbi).compress_algorithm =
886 } else if (!strcmp(name, "lzo-rle")) {
887 F2FS_OPTION(sbi).compress_algorithm =
895 case Opt_compress_log_size:
896 if (!f2fs_sb_has_compression(sbi)) {
897 f2fs_err(sbi, "Compression feature is off");
900 if (args->from && match_int(args, &arg))
902 if (arg < MIN_COMPRESS_LOG_SIZE ||
903 arg > MAX_COMPRESS_LOG_SIZE) {
905 "Compress cluster log size is out of range");
908 F2FS_OPTION(sbi).compress_log_size = arg;
910 case Opt_compress_extension:
911 if (!f2fs_sb_has_compression(sbi)) {
912 f2fs_err(sbi, "Compression feature is off");
915 name = match_strdup(&args[0]);
919 ext = F2FS_OPTION(sbi).extensions;
920 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
922 if (strlen(name) >= F2FS_EXTENSION_LEN ||
923 ext_cnt >= COMPRESS_EXT_NUM) {
925 "invalid extension length/number");
930 strcpy(ext[ext_cnt], name);
931 F2FS_OPTION(sbi).compress_ext_cnt++;
935 case Opt_compress_algorithm:
936 case Opt_compress_log_size:
937 case Opt_compress_extension:
938 f2fs_info(sbi, "compression options not supported");
942 f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
948 if (f2fs_check_quota_options(sbi))
951 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sbi->sb)) {
952 f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
955 if (f2fs_sb_has_project_quota(sbi) && !f2fs_readonly(sbi->sb)) {
956 f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
960 #ifndef CONFIG_UNICODE
961 if (f2fs_sb_has_casefold(sbi)) {
963 "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
968 if (F2FS_IO_SIZE_BITS(sbi) && !f2fs_lfs_mode(sbi)) {
969 f2fs_err(sbi, "Should set mode=lfs with %uKB-sized IO",
970 F2FS_IO_SIZE_KB(sbi));
974 if (test_opt(sbi, INLINE_XATTR_SIZE)) {
975 int min_size, max_size;
977 if (!f2fs_sb_has_extra_attr(sbi) ||
978 !f2fs_sb_has_flexible_inline_xattr(sbi)) {
979 f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
982 if (!test_opt(sbi, INLINE_XATTR)) {
983 f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
987 min_size = sizeof(struct f2fs_xattr_header) / sizeof(__le32);
988 max_size = MAX_INLINE_XATTR_SIZE;
990 if (F2FS_OPTION(sbi).inline_xattr_size < min_size ||
991 F2FS_OPTION(sbi).inline_xattr_size > max_size) {
992 f2fs_err(sbi, "inline xattr size is out of range: %d ~ %d",
998 if (test_opt(sbi, DISABLE_CHECKPOINT) && f2fs_lfs_mode(sbi)) {
999 f2fs_err(sbi, "LFS not compatible with checkpoint=disable\n");
1003 /* Not pass down write hints if the number of active logs is lesser
1004 * than NR_CURSEG_TYPE.
1006 if (F2FS_OPTION(sbi).active_logs != NR_CURSEG_TYPE)
1007 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1011 static struct inode *f2fs_alloc_inode(struct super_block *sb)
1013 struct f2fs_inode_info *fi;
1015 fi = kmem_cache_alloc(f2fs_inode_cachep, GFP_F2FS_ZERO);
1019 init_once((void *) fi);
1021 /* Initialize f2fs-specific inode info */
1022 atomic_set(&fi->dirty_pages, 0);
1023 init_rwsem(&fi->i_sem);
1024 spin_lock_init(&fi->i_size_lock);
1025 INIT_LIST_HEAD(&fi->dirty_list);
1026 INIT_LIST_HEAD(&fi->gdirty_list);
1027 INIT_LIST_HEAD(&fi->inmem_ilist);
1028 INIT_LIST_HEAD(&fi->inmem_pages);
1029 mutex_init(&fi->inmem_lock);
1030 init_rwsem(&fi->i_gc_rwsem[READ]);
1031 init_rwsem(&fi->i_gc_rwsem[WRITE]);
1032 init_rwsem(&fi->i_mmap_sem);
1033 init_rwsem(&fi->i_xattr_sem);
1035 /* Will be used by directory only */
1036 fi->i_dir_level = F2FS_SB(sb)->dir_level;
1040 return &fi->vfs_inode;
1043 static int f2fs_drop_inode(struct inode *inode)
1045 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1049 * during filesystem shutdown, if checkpoint is disabled,
1050 * drop useless meta/node dirty pages.
1052 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1053 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1054 inode->i_ino == F2FS_META_INO(sbi)) {
1055 trace_f2fs_drop_inode(inode, 1);
1061 * This is to avoid a deadlock condition like below.
1062 * writeback_single_inode(inode)
1063 * - f2fs_write_data_page
1064 * - f2fs_gc -> iput -> evict
1065 * - inode_wait_for_writeback(inode)
1067 if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
1068 if (!inode->i_nlink && !is_bad_inode(inode)) {
1069 /* to avoid evict_inode call simultaneously */
1070 atomic_inc(&inode->i_count);
1071 spin_unlock(&inode->i_lock);
1073 /* some remained atomic pages should discarded */
1074 if (f2fs_is_atomic_file(inode))
1075 f2fs_drop_inmem_pages(inode);
1077 /* should remain fi->extent_tree for writepage */
1078 f2fs_destroy_extent_node(inode);
1080 sb_start_intwrite(inode->i_sb);
1081 f2fs_i_size_write(inode, 0);
1083 f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1084 inode, NULL, 0, DATA);
1085 truncate_inode_pages_final(inode->i_mapping);
1087 if (F2FS_HAS_BLOCKS(inode))
1088 f2fs_truncate(inode);
1090 sb_end_intwrite(inode->i_sb);
1092 spin_lock(&inode->i_lock);
1093 atomic_dec(&inode->i_count);
1095 trace_f2fs_drop_inode(inode, 0);
1098 ret = generic_drop_inode(inode);
1100 ret = fscrypt_drop_inode(inode);
1101 trace_f2fs_drop_inode(inode, ret);
1105 int f2fs_inode_dirtied(struct inode *inode, bool sync)
1107 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1110 spin_lock(&sbi->inode_lock[DIRTY_META]);
1111 if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1114 set_inode_flag(inode, FI_DIRTY_INODE);
1115 stat_inc_dirty_inode(sbi, DIRTY_META);
1117 if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1118 list_add_tail(&F2FS_I(inode)->gdirty_list,
1119 &sbi->inode_list[DIRTY_META]);
1120 inc_page_count(sbi, F2FS_DIRTY_IMETA);
1122 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1126 void f2fs_inode_synced(struct inode *inode)
1128 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1130 spin_lock(&sbi->inode_lock[DIRTY_META]);
1131 if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1132 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1135 if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1136 list_del_init(&F2FS_I(inode)->gdirty_list);
1137 dec_page_count(sbi, F2FS_DIRTY_IMETA);
1139 clear_inode_flag(inode, FI_DIRTY_INODE);
1140 clear_inode_flag(inode, FI_AUTO_RECOVER);
1141 stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1142 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1146 * f2fs_dirty_inode() is called from __mark_inode_dirty()
1148 * We should call set_dirty_inode to write the dirty inode through write_inode.
1150 static void f2fs_dirty_inode(struct inode *inode, int flags)
1152 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1154 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1155 inode->i_ino == F2FS_META_INO(sbi))
1158 if (flags == I_DIRTY_TIME)
1161 if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1162 clear_inode_flag(inode, FI_AUTO_RECOVER);
1164 f2fs_inode_dirtied(inode, false);
1167 static void f2fs_free_inode(struct inode *inode)
1169 fscrypt_free_inode(inode);
1170 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1173 static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1175 percpu_counter_destroy(&sbi->alloc_valid_block_count);
1176 percpu_counter_destroy(&sbi->total_valid_inode_count);
1179 static void destroy_device_list(struct f2fs_sb_info *sbi)
1183 for (i = 0; i < sbi->s_ndevs; i++) {
1184 blkdev_put(FDEV(i).bdev, FMODE_EXCL);
1185 #ifdef CONFIG_BLK_DEV_ZONED
1186 kvfree(FDEV(i).blkz_seq);
1192 static void f2fs_put_super(struct super_block *sb)
1194 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1198 /* unregister procfs/sysfs entries in advance to avoid race case */
1199 f2fs_unregister_sysfs(sbi);
1201 f2fs_quota_off_umount(sb);
1203 /* prevent remaining shrinker jobs */
1204 mutex_lock(&sbi->umount_mutex);
1207 * We don't need to do checkpoint when superblock is clean.
1208 * But, the previous checkpoint was not done by umount, it needs to do
1209 * clean checkpoint again.
1211 if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1212 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1213 struct cp_control cpc = {
1214 .reason = CP_UMOUNT,
1216 f2fs_write_checkpoint(sbi, &cpc);
1219 /* be sure to wait for any on-going discard commands */
1220 dropped = f2fs_issue_discard_timeout(sbi);
1222 if ((f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi)) &&
1223 !sbi->discard_blks && !dropped) {
1224 struct cp_control cpc = {
1225 .reason = CP_UMOUNT | CP_TRIMMED,
1227 f2fs_write_checkpoint(sbi, &cpc);
1231 * normally superblock is clean, so we need to release this.
1232 * In addition, EIO will skip do checkpoint, we need this as well.
1234 f2fs_release_ino_entry(sbi, true);
1236 f2fs_leave_shrinker(sbi);
1237 mutex_unlock(&sbi->umount_mutex);
1239 /* our cp_error case, we can wait for any writeback page */
1240 f2fs_flush_merged_writes(sbi);
1242 f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1244 f2fs_bug_on(sbi, sbi->fsync_node_num);
1246 iput(sbi->node_inode);
1247 sbi->node_inode = NULL;
1249 iput(sbi->meta_inode);
1250 sbi->meta_inode = NULL;
1253 * iput() can update stat information, if f2fs_write_checkpoint()
1254 * above failed with error.
1256 f2fs_destroy_stats(sbi);
1258 /* destroy f2fs internal modules */
1259 f2fs_destroy_node_manager(sbi);
1260 f2fs_destroy_segment_manager(sbi);
1262 f2fs_destroy_post_read_wq(sbi);
1266 sb->s_fs_info = NULL;
1267 if (sbi->s_chksum_driver)
1268 crypto_free_shash(sbi->s_chksum_driver);
1269 kfree(sbi->raw_super);
1271 destroy_device_list(sbi);
1272 f2fs_destroy_xattr_caches(sbi);
1273 mempool_destroy(sbi->write_io_dummy);
1275 for (i = 0; i < MAXQUOTAS; i++)
1276 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1278 fscrypt_free_dummy_context(&F2FS_OPTION(sbi).dummy_enc_ctx);
1279 destroy_percpu_info(sbi);
1280 for (i = 0; i < NR_PAGE_TYPE; i++)
1281 kvfree(sbi->write_io[i]);
1282 #ifdef CONFIG_UNICODE
1283 utf8_unload(sbi->s_encoding);
1288 int f2fs_sync_fs(struct super_block *sb, int sync)
1290 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1293 if (unlikely(f2fs_cp_error(sbi)))
1295 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1298 trace_f2fs_sync_fs(sb, sync);
1300 if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1304 struct cp_control cpc;
1306 cpc.reason = __get_cp_reason(sbi);
1308 down_write(&sbi->gc_lock);
1309 err = f2fs_write_checkpoint(sbi, &cpc);
1310 up_write(&sbi->gc_lock);
1312 f2fs_trace_ios(NULL, 1);
1317 static int f2fs_freeze(struct super_block *sb)
1319 if (f2fs_readonly(sb))
1322 /* IO error happened before */
1323 if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1326 /* must be clean, since sync_filesystem() was already called */
1327 if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1332 static int f2fs_unfreeze(struct super_block *sb)
1338 static int f2fs_statfs_project(struct super_block *sb,
1339 kprojid_t projid, struct kstatfs *buf)
1342 struct dquot *dquot;
1346 qid = make_kqid_projid(projid);
1347 dquot = dqget(sb, qid);
1349 return PTR_ERR(dquot);
1350 spin_lock(&dquot->dq_dqb_lock);
1352 limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1353 dquot->dq_dqb.dqb_bhardlimit);
1355 limit >>= sb->s_blocksize_bits;
1357 if (limit && buf->f_blocks > limit) {
1358 curblock = (dquot->dq_dqb.dqb_curspace +
1359 dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
1360 buf->f_blocks = limit;
1361 buf->f_bfree = buf->f_bavail =
1362 (buf->f_blocks > curblock) ?
1363 (buf->f_blocks - curblock) : 0;
1366 limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1367 dquot->dq_dqb.dqb_ihardlimit);
1369 if (limit && buf->f_files > limit) {
1370 buf->f_files = limit;
1372 (buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1373 (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1376 spin_unlock(&dquot->dq_dqb_lock);
1382 static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1384 struct super_block *sb = dentry->d_sb;
1385 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1386 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
1387 block_t total_count, user_block_count, start_count;
1388 u64 avail_node_count;
1390 total_count = le64_to_cpu(sbi->raw_super->block_count);
1391 user_block_count = sbi->user_block_count;
1392 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
1393 buf->f_type = F2FS_SUPER_MAGIC;
1394 buf->f_bsize = sbi->blocksize;
1396 buf->f_blocks = total_count - start_count;
1397 buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1398 sbi->current_reserved_blocks;
1400 spin_lock(&sbi->stat_lock);
1401 if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1404 buf->f_bfree -= sbi->unusable_block_count;
1405 spin_unlock(&sbi->stat_lock);
1407 if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1408 buf->f_bavail = buf->f_bfree -
1409 F2FS_OPTION(sbi).root_reserved_blocks;
1413 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
1415 if (avail_node_count > user_block_count) {
1416 buf->f_files = user_block_count;
1417 buf->f_ffree = buf->f_bavail;
1419 buf->f_files = avail_node_count;
1420 buf->f_ffree = min(avail_node_count - valid_node_count(sbi),
1424 buf->f_namelen = F2FS_NAME_LEN;
1425 buf->f_fsid.val[0] = (u32)id;
1426 buf->f_fsid.val[1] = (u32)(id >> 32);
1429 if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1430 sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1431 f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1437 static inline void f2fs_show_quota_options(struct seq_file *seq,
1438 struct super_block *sb)
1441 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1443 if (F2FS_OPTION(sbi).s_jquota_fmt) {
1446 switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1457 seq_printf(seq, ",jqfmt=%s", fmtname);
1460 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1461 seq_show_option(seq, "usrjquota",
1462 F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1464 if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1465 seq_show_option(seq, "grpjquota",
1466 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1468 if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1469 seq_show_option(seq, "prjjquota",
1470 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1474 static inline void f2fs_show_compress_options(struct seq_file *seq,
1475 struct super_block *sb)
1477 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1481 if (!f2fs_sb_has_compression(sbi))
1484 switch (F2FS_OPTION(sbi).compress_algorithm) {
1494 case COMPRESS_LZORLE:
1495 algtype = "lzo-rle";
1498 seq_printf(seq, ",compress_algorithm=%s", algtype);
1500 seq_printf(seq, ",compress_log_size=%u",
1501 F2FS_OPTION(sbi).compress_log_size);
1503 for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1504 seq_printf(seq, ",compress_extension=%s",
1505 F2FS_OPTION(sbi).extensions[i]);
1509 static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1511 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1513 if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1514 seq_printf(seq, ",background_gc=%s", "sync");
1515 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1516 seq_printf(seq, ",background_gc=%s", "on");
1517 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1518 seq_printf(seq, ",background_gc=%s", "off");
1520 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1521 seq_puts(seq, ",disable_roll_forward");
1522 if (test_opt(sbi, NORECOVERY))
1523 seq_puts(seq, ",norecovery");
1524 if (test_opt(sbi, DISCARD))
1525 seq_puts(seq, ",discard");
1527 seq_puts(seq, ",nodiscard");
1528 if (test_opt(sbi, NOHEAP))
1529 seq_puts(seq, ",no_heap");
1531 seq_puts(seq, ",heap");
1532 #ifdef CONFIG_F2FS_FS_XATTR
1533 if (test_opt(sbi, XATTR_USER))
1534 seq_puts(seq, ",user_xattr");
1536 seq_puts(seq, ",nouser_xattr");
1537 if (test_opt(sbi, INLINE_XATTR))
1538 seq_puts(seq, ",inline_xattr");
1540 seq_puts(seq, ",noinline_xattr");
1541 if (test_opt(sbi, INLINE_XATTR_SIZE))
1542 seq_printf(seq, ",inline_xattr_size=%u",
1543 F2FS_OPTION(sbi).inline_xattr_size);
1545 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1546 if (test_opt(sbi, POSIX_ACL))
1547 seq_puts(seq, ",acl");
1549 seq_puts(seq, ",noacl");
1551 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
1552 seq_puts(seq, ",disable_ext_identify");
1553 if (test_opt(sbi, INLINE_DATA))
1554 seq_puts(seq, ",inline_data");
1556 seq_puts(seq, ",noinline_data");
1557 if (test_opt(sbi, INLINE_DENTRY))
1558 seq_puts(seq, ",inline_dentry");
1560 seq_puts(seq, ",noinline_dentry");
1561 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
1562 seq_puts(seq, ",flush_merge");
1563 if (test_opt(sbi, NOBARRIER))
1564 seq_puts(seq, ",nobarrier");
1565 if (test_opt(sbi, FASTBOOT))
1566 seq_puts(seq, ",fastboot");
1567 if (test_opt(sbi, EXTENT_CACHE))
1568 seq_puts(seq, ",extent_cache");
1570 seq_puts(seq, ",noextent_cache");
1571 if (test_opt(sbi, DATA_FLUSH))
1572 seq_puts(seq, ",data_flush");
1574 seq_puts(seq, ",mode=");
1575 if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
1576 seq_puts(seq, "adaptive");
1577 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
1578 seq_puts(seq, "lfs");
1579 seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
1580 if (test_opt(sbi, RESERVE_ROOT))
1581 seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
1582 F2FS_OPTION(sbi).root_reserved_blocks,
1583 from_kuid_munged(&init_user_ns,
1584 F2FS_OPTION(sbi).s_resuid),
1585 from_kgid_munged(&init_user_ns,
1586 F2FS_OPTION(sbi).s_resgid));
1587 if (F2FS_IO_SIZE_BITS(sbi))
1588 seq_printf(seq, ",io_bits=%u",
1589 F2FS_OPTION(sbi).write_io_size_bits);
1590 #ifdef CONFIG_F2FS_FAULT_INJECTION
1591 if (test_opt(sbi, FAULT_INJECTION)) {
1592 seq_printf(seq, ",fault_injection=%u",
1593 F2FS_OPTION(sbi).fault_info.inject_rate);
1594 seq_printf(seq, ",fault_type=%u",
1595 F2FS_OPTION(sbi).fault_info.inject_type);
1599 if (test_opt(sbi, QUOTA))
1600 seq_puts(seq, ",quota");
1601 if (test_opt(sbi, USRQUOTA))
1602 seq_puts(seq, ",usrquota");
1603 if (test_opt(sbi, GRPQUOTA))
1604 seq_puts(seq, ",grpquota");
1605 if (test_opt(sbi, PRJQUOTA))
1606 seq_puts(seq, ",prjquota");
1608 f2fs_show_quota_options(seq, sbi->sb);
1609 if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_USER)
1610 seq_printf(seq, ",whint_mode=%s", "user-based");
1611 else if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_FS)
1612 seq_printf(seq, ",whint_mode=%s", "fs-based");
1614 fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
1616 if (sbi->sb->s_flags & SB_INLINECRYPT)
1617 seq_puts(seq, ",inlinecrypt");
1619 if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
1620 seq_printf(seq, ",alloc_mode=%s", "default");
1621 else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
1622 seq_printf(seq, ",alloc_mode=%s", "reuse");
1624 if (test_opt(sbi, DISABLE_CHECKPOINT))
1625 seq_printf(seq, ",checkpoint=disable:%u",
1626 F2FS_OPTION(sbi).unusable_cap);
1627 if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
1628 seq_printf(seq, ",fsync_mode=%s", "posix");
1629 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
1630 seq_printf(seq, ",fsync_mode=%s", "strict");
1631 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
1632 seq_printf(seq, ",fsync_mode=%s", "nobarrier");
1634 #ifdef CONFIG_F2FS_FS_COMPRESSION
1635 f2fs_show_compress_options(seq, sbi->sb);
1640 static void default_options(struct f2fs_sb_info *sbi)
1642 /* init some FS parameters */
1643 F2FS_OPTION(sbi).active_logs = NR_CURSEG_TYPE;
1644 F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
1645 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1646 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
1647 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
1648 F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
1649 F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
1650 F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
1651 F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
1652 F2FS_OPTION(sbi).compress_ext_cnt = 0;
1653 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
1655 sbi->sb->s_flags &= ~SB_INLINECRYPT;
1657 set_opt(sbi, INLINE_XATTR);
1658 set_opt(sbi, INLINE_DATA);
1659 set_opt(sbi, INLINE_DENTRY);
1660 set_opt(sbi, EXTENT_CACHE);
1661 set_opt(sbi, NOHEAP);
1662 clear_opt(sbi, DISABLE_CHECKPOINT);
1663 F2FS_OPTION(sbi).unusable_cap = 0;
1664 sbi->sb->s_flags |= SB_LAZYTIME;
1665 set_opt(sbi, FLUSH_MERGE);
1666 set_opt(sbi, DISCARD);
1667 if (f2fs_sb_has_blkzoned(sbi))
1668 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
1670 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
1672 #ifdef CONFIG_F2FS_FS_XATTR
1673 set_opt(sbi, XATTR_USER);
1675 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1676 set_opt(sbi, POSIX_ACL);
1679 f2fs_build_fault_attr(sbi, 0, 0);
1683 static int f2fs_enable_quotas(struct super_block *sb);
1686 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
1688 unsigned int s_flags = sbi->sb->s_flags;
1689 struct cp_control cpc;
1694 if (s_flags & SB_RDONLY) {
1695 f2fs_err(sbi, "checkpoint=disable on readonly fs");
1698 sbi->sb->s_flags |= SB_ACTIVE;
1700 f2fs_update_time(sbi, DISABLE_TIME);
1702 while (!f2fs_time_over(sbi, DISABLE_TIME)) {
1703 down_write(&sbi->gc_lock);
1704 err = f2fs_gc(sbi, true, false, NULL_SEGNO);
1705 if (err == -ENODATA) {
1709 if (err && err != -EAGAIN)
1713 ret = sync_filesystem(sbi->sb);
1715 err = ret ? ret: err;
1719 unusable = f2fs_get_unusable_blocks(sbi);
1720 if (f2fs_disable_cp_again(sbi, unusable)) {
1725 down_write(&sbi->gc_lock);
1726 cpc.reason = CP_PAUSE;
1727 set_sbi_flag(sbi, SBI_CP_DISABLED);
1728 err = f2fs_write_checkpoint(sbi, &cpc);
1732 spin_lock(&sbi->stat_lock);
1733 sbi->unusable_block_count = unusable;
1734 spin_unlock(&sbi->stat_lock);
1737 up_write(&sbi->gc_lock);
1739 sbi->sb->s_flags = s_flags; /* Restore SB_RDONLY status */
1743 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
1745 down_write(&sbi->gc_lock);
1746 f2fs_dirty_to_prefree(sbi);
1748 clear_sbi_flag(sbi, SBI_CP_DISABLED);
1749 set_sbi_flag(sbi, SBI_IS_DIRTY);
1750 up_write(&sbi->gc_lock);
1752 f2fs_sync_fs(sbi->sb, 1);
1755 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
1757 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1758 struct f2fs_mount_info org_mount_opt;
1759 unsigned long old_sb_flags;
1761 bool need_restart_gc = false;
1762 bool need_stop_gc = false;
1763 bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
1764 bool disable_checkpoint = test_opt(sbi, DISABLE_CHECKPOINT);
1765 bool no_io_align = !F2FS_IO_ALIGNED(sbi);
1766 bool checkpoint_changed;
1772 * Save the old mount options in case we
1773 * need to restore them.
1775 org_mount_opt = sbi->mount_opt;
1776 old_sb_flags = sb->s_flags;
1779 org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
1780 for (i = 0; i < MAXQUOTAS; i++) {
1781 if (F2FS_OPTION(sbi).s_qf_names[i]) {
1782 org_mount_opt.s_qf_names[i] =
1783 kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
1785 if (!org_mount_opt.s_qf_names[i]) {
1786 for (j = 0; j < i; j++)
1787 kfree(org_mount_opt.s_qf_names[j]);
1791 org_mount_opt.s_qf_names[i] = NULL;
1796 /* recover superblocks we couldn't write due to previous RO mount */
1797 if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
1798 err = f2fs_commit_super(sbi, false);
1799 f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
1802 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
1805 default_options(sbi);
1807 /* parse mount options */
1808 err = parse_options(sb, data, true);
1811 checkpoint_changed =
1812 disable_checkpoint != test_opt(sbi, DISABLE_CHECKPOINT);
1815 * Previous and new state of filesystem is RO,
1816 * so skip checking GC and FLUSH_MERGE conditions.
1818 if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
1822 if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
1823 err = dquot_suspend(sb, -1);
1826 } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
1827 /* dquot_resume needs RW */
1828 sb->s_flags &= ~SB_RDONLY;
1829 if (sb_any_quota_suspended(sb)) {
1830 dquot_resume(sb, -1);
1831 } else if (f2fs_sb_has_quota_ino(sbi)) {
1832 err = f2fs_enable_quotas(sb);
1838 /* disallow enable/disable extent_cache dynamically */
1839 if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
1841 f2fs_warn(sbi, "switch extent_cache option is not allowed");
1845 if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
1847 f2fs_warn(sbi, "switch io_bits option is not allowed");
1851 if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
1853 f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
1858 * We stop the GC thread if FS is mounted as RO
1859 * or if background_gc = off is passed in mount
1860 * option. Also sync the filesystem.
1862 if ((*flags & SB_RDONLY) ||
1863 F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF) {
1864 if (sbi->gc_thread) {
1865 f2fs_stop_gc_thread(sbi);
1866 need_restart_gc = true;
1868 } else if (!sbi->gc_thread) {
1869 err = f2fs_start_gc_thread(sbi);
1872 need_stop_gc = true;
1875 if (*flags & SB_RDONLY ||
1876 F2FS_OPTION(sbi).whint_mode != org_mount_opt.whint_mode) {
1877 writeback_inodes_sb(sb, WB_REASON_SYNC);
1880 set_sbi_flag(sbi, SBI_IS_DIRTY);
1881 set_sbi_flag(sbi, SBI_IS_CLOSE);
1882 f2fs_sync_fs(sb, 1);
1883 clear_sbi_flag(sbi, SBI_IS_CLOSE);
1886 if (checkpoint_changed) {
1887 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
1888 err = f2fs_disable_checkpoint(sbi);
1892 f2fs_enable_checkpoint(sbi);
1897 * We stop issue flush thread if FS is mounted as RO
1898 * or if flush_merge is not passed in mount option.
1900 if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
1901 clear_opt(sbi, FLUSH_MERGE);
1902 f2fs_destroy_flush_cmd_control(sbi, false);
1904 err = f2fs_create_flush_cmd_control(sbi);
1910 /* Release old quota file names */
1911 for (i = 0; i < MAXQUOTAS; i++)
1912 kfree(org_mount_opt.s_qf_names[i]);
1914 /* Update the POSIXACL Flag */
1915 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
1916 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
1918 limit_reserve_root(sbi);
1919 adjust_unusable_cap_perc(sbi);
1920 *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
1923 if (need_restart_gc) {
1924 if (f2fs_start_gc_thread(sbi))
1925 f2fs_warn(sbi, "background gc thread has stopped");
1926 } else if (need_stop_gc) {
1927 f2fs_stop_gc_thread(sbi);
1931 F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
1932 for (i = 0; i < MAXQUOTAS; i++) {
1933 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1934 F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
1937 sbi->mount_opt = org_mount_opt;
1938 sb->s_flags = old_sb_flags;
1943 /* Read data from quotafile */
1944 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
1945 size_t len, loff_t off)
1947 struct inode *inode = sb_dqopt(sb)->files[type];
1948 struct address_space *mapping = inode->i_mapping;
1949 block_t blkidx = F2FS_BYTES_TO_BLK(off);
1950 int offset = off & (sb->s_blocksize - 1);
1953 loff_t i_size = i_size_read(inode);
1960 if (off + len > i_size)
1963 while (toread > 0) {
1964 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
1966 page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
1968 if (PTR_ERR(page) == -ENOMEM) {
1969 congestion_wait(BLK_RW_ASYNC,
1970 DEFAULT_IO_TIMEOUT);
1973 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
1974 return PTR_ERR(page);
1979 if (unlikely(page->mapping != mapping)) {
1980 f2fs_put_page(page, 1);
1983 if (unlikely(!PageUptodate(page))) {
1984 f2fs_put_page(page, 1);
1985 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
1989 kaddr = kmap_atomic(page);
1990 memcpy(data, kaddr + offset, tocopy);
1991 kunmap_atomic(kaddr);
1992 f2fs_put_page(page, 1);
2002 /* Write to quotafile */
2003 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2004 const char *data, size_t len, loff_t off)
2006 struct inode *inode = sb_dqopt(sb)->files[type];
2007 struct address_space *mapping = inode->i_mapping;
2008 const struct address_space_operations *a_ops = mapping->a_ops;
2009 int offset = off & (sb->s_blocksize - 1);
2010 size_t towrite = len;
2012 void *fsdata = NULL;
2017 while (towrite > 0) {
2018 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2021 err = a_ops->write_begin(NULL, mapping, off, tocopy, 0,
2023 if (unlikely(err)) {
2024 if (err == -ENOMEM) {
2025 congestion_wait(BLK_RW_ASYNC,
2026 DEFAULT_IO_TIMEOUT);
2029 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2033 kaddr = kmap_atomic(page);
2034 memcpy(kaddr + offset, data, tocopy);
2035 kunmap_atomic(kaddr);
2036 flush_dcache_page(page);
2038 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2049 inode->i_mtime = inode->i_ctime = current_time(inode);
2050 f2fs_mark_inode_dirty_sync(inode, false);
2051 return len - towrite;
2054 static struct dquot **f2fs_get_dquots(struct inode *inode)
2056 return F2FS_I(inode)->i_dquot;
2059 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2061 return &F2FS_I(inode)->i_reserved_quota;
2064 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2066 if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2067 f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2071 return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2072 F2FS_OPTION(sbi).s_jquota_fmt, type);
2075 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2080 if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2081 err = f2fs_enable_quotas(sbi->sb);
2083 f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2089 for (i = 0; i < MAXQUOTAS; i++) {
2090 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2091 err = f2fs_quota_on_mount(sbi, i);
2096 f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2103 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2106 struct inode *qf_inode;
2107 unsigned long qf_inum;
2110 BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2112 qf_inum = f2fs_qf_ino(sb, type);
2116 qf_inode = f2fs_iget(sb, qf_inum);
2117 if (IS_ERR(qf_inode)) {
2118 f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2119 return PTR_ERR(qf_inode);
2122 /* Don't account quota for quota files to avoid recursion */
2123 qf_inode->i_flags |= S_NOQUOTA;
2124 err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2129 static int f2fs_enable_quotas(struct super_block *sb)
2131 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2133 unsigned long qf_inum;
2134 bool quota_mopt[MAXQUOTAS] = {
2135 test_opt(sbi, USRQUOTA),
2136 test_opt(sbi, GRPQUOTA),
2137 test_opt(sbi, PRJQUOTA),
2140 if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2141 f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2145 sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2147 for (type = 0; type < MAXQUOTAS; type++) {
2148 qf_inum = f2fs_qf_ino(sb, type);
2150 err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2151 DQUOT_USAGE_ENABLED |
2152 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2154 f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2156 for (type--; type >= 0; type--)
2157 dquot_quota_off(sb, type);
2158 set_sbi_flag(F2FS_SB(sb),
2159 SBI_QUOTA_NEED_REPAIR);
2167 int f2fs_quota_sync(struct super_block *sb, int type)
2169 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2170 struct quota_info *dqopt = sb_dqopt(sb);
2177 * down_read(quota_sem)
2178 * dquot_writeback_dquots()
2181 * down_read(quota_sem)
2185 down_read(&sbi->quota_sem);
2186 ret = dquot_writeback_dquots(sb, type);
2191 * Now when everything is written we can discard the pagecache so
2192 * that userspace sees the changes.
2194 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2195 struct address_space *mapping;
2197 if (type != -1 && cnt != type)
2199 if (!sb_has_quota_active(sb, cnt))
2202 mapping = dqopt->files[cnt]->i_mapping;
2204 ret = filemap_fdatawrite(mapping);
2208 /* if we are using journalled quota */
2209 if (is_journalled_quota(sbi))
2212 ret = filemap_fdatawait(mapping);
2214 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2216 inode_lock(dqopt->files[cnt]);
2217 truncate_inode_pages(&dqopt->files[cnt]->i_data, 0);
2218 inode_unlock(dqopt->files[cnt]);
2222 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2223 up_read(&sbi->quota_sem);
2224 f2fs_unlock_op(sbi);
2228 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2229 const struct path *path)
2231 struct inode *inode;
2234 /* if quota sysfile exists, deny enabling quota with specific file */
2235 if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2236 f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2240 err = f2fs_quota_sync(sb, type);
2244 err = dquot_quota_on(sb, type, format_id, path);
2248 inode = d_inode(path->dentry);
2251 F2FS_I(inode)->i_flags |= F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL;
2252 f2fs_set_inode_flags(inode);
2253 inode_unlock(inode);
2254 f2fs_mark_inode_dirty_sync(inode, false);
2259 static int __f2fs_quota_off(struct super_block *sb, int type)
2261 struct inode *inode = sb_dqopt(sb)->files[type];
2264 if (!inode || !igrab(inode))
2265 return dquot_quota_off(sb, type);
2267 err = f2fs_quota_sync(sb, type);
2271 err = dquot_quota_off(sb, type);
2272 if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
2276 F2FS_I(inode)->i_flags &= ~(F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL);
2277 f2fs_set_inode_flags(inode);
2278 inode_unlock(inode);
2279 f2fs_mark_inode_dirty_sync(inode, false);
2285 static int f2fs_quota_off(struct super_block *sb, int type)
2287 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2290 err = __f2fs_quota_off(sb, type);
2293 * quotactl can shutdown journalled quota, result in inconsistence
2294 * between quota record and fs data by following updates, tag the
2295 * flag to let fsck be aware of it.
2297 if (is_journalled_quota(sbi))
2298 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2302 void f2fs_quota_off_umount(struct super_block *sb)
2307 for (type = 0; type < MAXQUOTAS; type++) {
2308 err = __f2fs_quota_off(sb, type);
2310 int ret = dquot_quota_off(sb, type);
2312 f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
2314 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2318 * In case of checkpoint=disable, we must flush quota blocks.
2319 * This can cause NULL exception for node_inode in end_io, since
2320 * put_super already dropped it.
2322 sync_filesystem(sb);
2325 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
2327 struct quota_info *dqopt = sb_dqopt(sb);
2330 for (type = 0; type < MAXQUOTAS; type++) {
2331 if (!dqopt->files[type])
2333 f2fs_inode_synced(dqopt->files[type]);
2337 static int f2fs_dquot_commit(struct dquot *dquot)
2339 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2342 down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
2343 ret = dquot_commit(dquot);
2345 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2346 up_read(&sbi->quota_sem);
2350 static int f2fs_dquot_acquire(struct dquot *dquot)
2352 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2355 down_read(&sbi->quota_sem);
2356 ret = dquot_acquire(dquot);
2358 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2359 up_read(&sbi->quota_sem);
2363 static int f2fs_dquot_release(struct dquot *dquot)
2365 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2366 int ret = dquot_release(dquot);
2369 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2373 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
2375 struct super_block *sb = dquot->dq_sb;
2376 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2377 int ret = dquot_mark_dquot_dirty(dquot);
2379 /* if we are using journalled quota */
2380 if (is_journalled_quota(sbi))
2381 set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
2386 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
2388 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2389 int ret = dquot_commit_info(sb, type);
2392 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2396 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
2398 *projid = F2FS_I(inode)->i_projid;
2402 static const struct dquot_operations f2fs_quota_operations = {
2403 .get_reserved_space = f2fs_get_reserved_space,
2404 .write_dquot = f2fs_dquot_commit,
2405 .acquire_dquot = f2fs_dquot_acquire,
2406 .release_dquot = f2fs_dquot_release,
2407 .mark_dirty = f2fs_dquot_mark_dquot_dirty,
2408 .write_info = f2fs_dquot_commit_info,
2409 .alloc_dquot = dquot_alloc,
2410 .destroy_dquot = dquot_destroy,
2411 .get_projid = f2fs_get_projid,
2412 .get_next_id = dquot_get_next_id,
2415 static const struct quotactl_ops f2fs_quotactl_ops = {
2416 .quota_on = f2fs_quota_on,
2417 .quota_off = f2fs_quota_off,
2418 .quota_sync = f2fs_quota_sync,
2419 .get_state = dquot_get_state,
2420 .set_info = dquot_set_dqinfo,
2421 .get_dqblk = dquot_get_dqblk,
2422 .set_dqblk = dquot_set_dqblk,
2423 .get_nextdqblk = dquot_get_next_dqblk,
2426 int f2fs_quota_sync(struct super_block *sb, int type)
2431 void f2fs_quota_off_umount(struct super_block *sb)
2436 static const struct super_operations f2fs_sops = {
2437 .alloc_inode = f2fs_alloc_inode,
2438 .free_inode = f2fs_free_inode,
2439 .drop_inode = f2fs_drop_inode,
2440 .write_inode = f2fs_write_inode,
2441 .dirty_inode = f2fs_dirty_inode,
2442 .show_options = f2fs_show_options,
2444 .quota_read = f2fs_quota_read,
2445 .quota_write = f2fs_quota_write,
2446 .get_dquots = f2fs_get_dquots,
2448 .evict_inode = f2fs_evict_inode,
2449 .put_super = f2fs_put_super,
2450 .sync_fs = f2fs_sync_fs,
2451 .freeze_fs = f2fs_freeze,
2452 .unfreeze_fs = f2fs_unfreeze,
2453 .statfs = f2fs_statfs,
2454 .remount_fs = f2fs_remount,
2457 #ifdef CONFIG_FS_ENCRYPTION
2458 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
2460 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2461 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2465 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
2468 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2471 * Encrypting the root directory is not allowed because fsck
2472 * expects lost+found directory to exist and remain unencrypted
2473 * if LOST_FOUND feature is enabled.
2476 if (f2fs_sb_has_lost_found(sbi) &&
2477 inode->i_ino == F2FS_ROOT_INO(sbi))
2480 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2481 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2482 ctx, len, fs_data, XATTR_CREATE);
2485 static const union fscrypt_context *
2486 f2fs_get_dummy_context(struct super_block *sb)
2488 return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_ctx.ctx;
2491 static bool f2fs_has_stable_inodes(struct super_block *sb)
2496 static void f2fs_get_ino_and_lblk_bits(struct super_block *sb,
2497 int *ino_bits_ret, int *lblk_bits_ret)
2499 *ino_bits_ret = 8 * sizeof(nid_t);
2500 *lblk_bits_ret = 8 * sizeof(block_t);
2503 static int f2fs_get_num_devices(struct super_block *sb)
2505 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2507 if (f2fs_is_multi_device(sbi))
2508 return sbi->s_ndevs;
2512 static void f2fs_get_devices(struct super_block *sb,
2513 struct request_queue **devs)
2515 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2518 for (i = 0; i < sbi->s_ndevs; i++)
2519 devs[i] = bdev_get_queue(FDEV(i).bdev);
2522 static const struct fscrypt_operations f2fs_cryptops = {
2523 .key_prefix = "f2fs:",
2524 .get_context = f2fs_get_context,
2525 .set_context = f2fs_set_context,
2526 .get_dummy_context = f2fs_get_dummy_context,
2527 .empty_dir = f2fs_empty_dir,
2528 .max_namelen = F2FS_NAME_LEN,
2529 .has_stable_inodes = f2fs_has_stable_inodes,
2530 .get_ino_and_lblk_bits = f2fs_get_ino_and_lblk_bits,
2531 .get_num_devices = f2fs_get_num_devices,
2532 .get_devices = f2fs_get_devices,
2536 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
2537 u64 ino, u32 generation)
2539 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2540 struct inode *inode;
2542 if (f2fs_check_nid_range(sbi, ino))
2543 return ERR_PTR(-ESTALE);
2546 * f2fs_iget isn't quite right if the inode is currently unallocated!
2547 * However f2fs_iget currently does appropriate checks to handle stale
2548 * inodes so everything is OK.
2550 inode = f2fs_iget(sb, ino);
2552 return ERR_CAST(inode);
2553 if (unlikely(generation && inode->i_generation != generation)) {
2554 /* we didn't find the right inode.. */
2556 return ERR_PTR(-ESTALE);
2561 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
2562 int fh_len, int fh_type)
2564 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
2565 f2fs_nfs_get_inode);
2568 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
2569 int fh_len, int fh_type)
2571 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
2572 f2fs_nfs_get_inode);
2575 static const struct export_operations f2fs_export_ops = {
2576 .fh_to_dentry = f2fs_fh_to_dentry,
2577 .fh_to_parent = f2fs_fh_to_parent,
2578 .get_parent = f2fs_get_parent,
2581 static loff_t max_file_blocks(void)
2584 loff_t leaf_count = DEF_ADDRS_PER_BLOCK;
2587 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
2588 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
2589 * space in inode.i_addr, it will be more safe to reassign
2593 /* two direct node blocks */
2594 result += (leaf_count * 2);
2596 /* two indirect node blocks */
2597 leaf_count *= NIDS_PER_BLOCK;
2598 result += (leaf_count * 2);
2600 /* one double indirect node block */
2601 leaf_count *= NIDS_PER_BLOCK;
2602 result += leaf_count;
2607 static int __f2fs_commit_super(struct buffer_head *bh,
2608 struct f2fs_super_block *super)
2612 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
2613 set_buffer_dirty(bh);
2616 /* it's rare case, we can do fua all the time */
2617 return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
2620 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
2621 struct buffer_head *bh)
2623 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2624 (bh->b_data + F2FS_SUPER_OFFSET);
2625 struct super_block *sb = sbi->sb;
2626 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
2627 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
2628 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
2629 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
2630 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
2631 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
2632 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
2633 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
2634 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
2635 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
2636 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2637 u32 segment_count = le32_to_cpu(raw_super->segment_count);
2638 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2639 u64 main_end_blkaddr = main_blkaddr +
2640 (segment_count_main << log_blocks_per_seg);
2641 u64 seg_end_blkaddr = segment0_blkaddr +
2642 (segment_count << log_blocks_per_seg);
2644 if (segment0_blkaddr != cp_blkaddr) {
2645 f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
2646 segment0_blkaddr, cp_blkaddr);
2650 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
2652 f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
2653 cp_blkaddr, sit_blkaddr,
2654 segment_count_ckpt << log_blocks_per_seg);
2658 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
2660 f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
2661 sit_blkaddr, nat_blkaddr,
2662 segment_count_sit << log_blocks_per_seg);
2666 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
2668 f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
2669 nat_blkaddr, ssa_blkaddr,
2670 segment_count_nat << log_blocks_per_seg);
2674 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
2676 f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
2677 ssa_blkaddr, main_blkaddr,
2678 segment_count_ssa << log_blocks_per_seg);
2682 if (main_end_blkaddr > seg_end_blkaddr) {
2683 f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%u) block(%u)",
2686 (segment_count << log_blocks_per_seg),
2687 segment_count_main << log_blocks_per_seg);
2689 } else if (main_end_blkaddr < seg_end_blkaddr) {
2693 /* fix in-memory information all the time */
2694 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
2695 segment0_blkaddr) >> log_blocks_per_seg);
2697 if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
2698 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2701 err = __f2fs_commit_super(bh, NULL);
2702 res = err ? "failed" : "done";
2704 f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%u) block(%u)",
2707 (segment_count << log_blocks_per_seg),
2708 segment_count_main << log_blocks_per_seg);
2715 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
2716 struct buffer_head *bh)
2718 block_t segment_count, segs_per_sec, secs_per_zone;
2719 block_t total_sections, blocks_per_seg;
2720 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2721 (bh->b_data + F2FS_SUPER_OFFSET);
2722 unsigned int blocksize;
2723 size_t crc_offset = 0;
2726 if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
2727 f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
2728 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
2732 /* Check checksum_offset and crc in superblock */
2733 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
2734 crc_offset = le32_to_cpu(raw_super->checksum_offset);
2736 offsetof(struct f2fs_super_block, crc)) {
2737 f2fs_info(sbi, "Invalid SB checksum offset: %zu",
2739 return -EFSCORRUPTED;
2741 crc = le32_to_cpu(raw_super->crc);
2742 if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
2743 f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
2744 return -EFSCORRUPTED;
2748 /* Currently, support only 4KB page cache size */
2749 if (F2FS_BLKSIZE != PAGE_SIZE) {
2750 f2fs_info(sbi, "Invalid page_cache_size (%lu), supports only 4KB",
2752 return -EFSCORRUPTED;
2755 /* Currently, support only 4KB block size */
2756 blocksize = 1 << le32_to_cpu(raw_super->log_blocksize);
2757 if (blocksize != F2FS_BLKSIZE) {
2758 f2fs_info(sbi, "Invalid blocksize (%u), supports only 4KB",
2760 return -EFSCORRUPTED;
2763 /* check log blocks per segment */
2764 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
2765 f2fs_info(sbi, "Invalid log blocks per segment (%u)",
2766 le32_to_cpu(raw_super->log_blocks_per_seg));
2767 return -EFSCORRUPTED;
2770 /* Currently, support 512/1024/2048/4096 bytes sector size */
2771 if (le32_to_cpu(raw_super->log_sectorsize) >
2772 F2FS_MAX_LOG_SECTOR_SIZE ||
2773 le32_to_cpu(raw_super->log_sectorsize) <
2774 F2FS_MIN_LOG_SECTOR_SIZE) {
2775 f2fs_info(sbi, "Invalid log sectorsize (%u)",
2776 le32_to_cpu(raw_super->log_sectorsize));
2777 return -EFSCORRUPTED;
2779 if (le32_to_cpu(raw_super->log_sectors_per_block) +
2780 le32_to_cpu(raw_super->log_sectorsize) !=
2781 F2FS_MAX_LOG_SECTOR_SIZE) {
2782 f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
2783 le32_to_cpu(raw_super->log_sectors_per_block),
2784 le32_to_cpu(raw_super->log_sectorsize));
2785 return -EFSCORRUPTED;
2788 segment_count = le32_to_cpu(raw_super->segment_count);
2789 segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
2790 secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
2791 total_sections = le32_to_cpu(raw_super->section_count);
2793 /* blocks_per_seg should be 512, given the above check */
2794 blocks_per_seg = 1 << le32_to_cpu(raw_super->log_blocks_per_seg);
2796 if (segment_count > F2FS_MAX_SEGMENT ||
2797 segment_count < F2FS_MIN_SEGMENTS) {
2798 f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
2799 return -EFSCORRUPTED;
2802 if (total_sections > segment_count ||
2803 total_sections < F2FS_MIN_SEGMENTS ||
2804 segs_per_sec > segment_count || !segs_per_sec) {
2805 f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
2806 segment_count, total_sections, segs_per_sec);
2807 return -EFSCORRUPTED;
2810 if ((segment_count / segs_per_sec) < total_sections) {
2811 f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
2812 segment_count, segs_per_sec, total_sections);
2813 return -EFSCORRUPTED;
2816 if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
2817 f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
2818 segment_count, le64_to_cpu(raw_super->block_count));
2819 return -EFSCORRUPTED;
2822 if (RDEV(0).path[0]) {
2823 block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
2826 while (i < MAX_DEVICES && RDEV(i).path[0]) {
2827 dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
2830 if (segment_count != dev_seg_count) {
2831 f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
2832 segment_count, dev_seg_count);
2833 return -EFSCORRUPTED;
2837 if (secs_per_zone > total_sections || !secs_per_zone) {
2838 f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
2839 secs_per_zone, total_sections);
2840 return -EFSCORRUPTED;
2842 if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
2843 raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
2844 (le32_to_cpu(raw_super->extension_count) +
2845 raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
2846 f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
2847 le32_to_cpu(raw_super->extension_count),
2848 raw_super->hot_ext_count,
2849 F2FS_MAX_EXTENSION);
2850 return -EFSCORRUPTED;
2853 if (le32_to_cpu(raw_super->cp_payload) >
2854 (blocks_per_seg - F2FS_CP_PACKS)) {
2855 f2fs_info(sbi, "Insane cp_payload (%u > %u)",
2856 le32_to_cpu(raw_super->cp_payload),
2857 blocks_per_seg - F2FS_CP_PACKS);
2858 return -EFSCORRUPTED;
2861 /* check reserved ino info */
2862 if (le32_to_cpu(raw_super->node_ino) != 1 ||
2863 le32_to_cpu(raw_super->meta_ino) != 2 ||
2864 le32_to_cpu(raw_super->root_ino) != 3) {
2865 f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
2866 le32_to_cpu(raw_super->node_ino),
2867 le32_to_cpu(raw_super->meta_ino),
2868 le32_to_cpu(raw_super->root_ino));
2869 return -EFSCORRUPTED;
2872 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
2873 if (sanity_check_area_boundary(sbi, bh))
2874 return -EFSCORRUPTED;
2879 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
2881 unsigned int total, fsmeta;
2882 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
2883 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2884 unsigned int ovp_segments, reserved_segments;
2885 unsigned int main_segs, blocks_per_seg;
2886 unsigned int sit_segs, nat_segs;
2887 unsigned int sit_bitmap_size, nat_bitmap_size;
2888 unsigned int log_blocks_per_seg;
2889 unsigned int segment_count_main;
2890 unsigned int cp_pack_start_sum, cp_payload;
2891 block_t user_block_count, valid_user_blocks;
2892 block_t avail_node_count, valid_node_count;
2895 total = le32_to_cpu(raw_super->segment_count);
2896 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
2897 sit_segs = le32_to_cpu(raw_super->segment_count_sit);
2899 nat_segs = le32_to_cpu(raw_super->segment_count_nat);
2901 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
2902 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
2904 if (unlikely(fsmeta >= total))
2907 ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
2908 reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
2910 if (unlikely(fsmeta < F2FS_MIN_SEGMENTS ||
2911 ovp_segments == 0 || reserved_segments == 0)) {
2912 f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
2916 user_block_count = le64_to_cpu(ckpt->user_block_count);
2917 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2918 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2919 if (!user_block_count || user_block_count >=
2920 segment_count_main << log_blocks_per_seg) {
2921 f2fs_err(sbi, "Wrong user_block_count: %u",
2926 valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
2927 if (valid_user_blocks > user_block_count) {
2928 f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
2929 valid_user_blocks, user_block_count);
2933 valid_node_count = le32_to_cpu(ckpt->valid_node_count);
2934 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
2935 if (valid_node_count > avail_node_count) {
2936 f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
2937 valid_node_count, avail_node_count);
2941 main_segs = le32_to_cpu(raw_super->segment_count_main);
2942 blocks_per_seg = sbi->blocks_per_seg;
2944 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
2945 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
2946 le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
2948 for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
2949 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
2950 le32_to_cpu(ckpt->cur_node_segno[j])) {
2951 f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
2953 le32_to_cpu(ckpt->cur_node_segno[i]));
2958 for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
2959 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
2960 le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
2962 for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
2963 if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
2964 le32_to_cpu(ckpt->cur_data_segno[j])) {
2965 f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
2967 le32_to_cpu(ckpt->cur_data_segno[i]));
2972 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
2973 for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
2974 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
2975 le32_to_cpu(ckpt->cur_data_segno[j])) {
2976 f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
2978 le32_to_cpu(ckpt->cur_node_segno[i]));
2984 sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
2985 nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
2987 if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
2988 nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
2989 f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
2990 sit_bitmap_size, nat_bitmap_size);
2994 cp_pack_start_sum = __start_sum_addr(sbi);
2995 cp_payload = __cp_payload(sbi);
2996 if (cp_pack_start_sum < cp_payload + 1 ||
2997 cp_pack_start_sum > blocks_per_seg - 1 -
2999 f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3004 if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3005 le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3006 f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3007 "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3008 "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3009 le32_to_cpu(ckpt->checksum_offset));
3013 if (unlikely(f2fs_cp_error(sbi))) {
3014 f2fs_err(sbi, "A bug case: need to run fsck");
3020 static void init_sb_info(struct f2fs_sb_info *sbi)
3022 struct f2fs_super_block *raw_super = sbi->raw_super;
3025 sbi->log_sectors_per_block =
3026 le32_to_cpu(raw_super->log_sectors_per_block);
3027 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3028 sbi->blocksize = 1 << sbi->log_blocksize;
3029 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3030 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
3031 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3032 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3033 sbi->total_sections = le32_to_cpu(raw_super->section_count);
3034 sbi->total_node_count =
3035 (le32_to_cpu(raw_super->segment_count_nat) / 2)
3036 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
3037 sbi->root_ino_num = le32_to_cpu(raw_super->root_ino);
3038 sbi->node_ino_num = le32_to_cpu(raw_super->node_ino);
3039 sbi->meta_ino_num = le32_to_cpu(raw_super->meta_ino);
3040 sbi->cur_victim_sec = NULL_SECNO;
3041 sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3042 sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3043 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3044 sbi->migration_granularity = sbi->segs_per_sec;
3046 sbi->dir_level = DEF_DIR_LEVEL;
3047 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3048 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3049 sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3050 sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3051 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3052 sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3053 DEF_UMOUNT_DISCARD_TIMEOUT;
3054 clear_sbi_flag(sbi, SBI_NEED_FSCK);
3056 for (i = 0; i < NR_COUNT_TYPE; i++)
3057 atomic_set(&sbi->nr_pages[i], 0);
3059 for (i = 0; i < META; i++)
3060 atomic_set(&sbi->wb_sync_req[i], 0);
3062 INIT_LIST_HEAD(&sbi->s_list);
3063 mutex_init(&sbi->umount_mutex);
3064 init_rwsem(&sbi->io_order_lock);
3065 spin_lock_init(&sbi->cp_lock);
3067 sbi->dirty_device = 0;
3068 spin_lock_init(&sbi->dev_lock);
3070 init_rwsem(&sbi->sb_lock);
3071 init_rwsem(&sbi->pin_sem);
3074 static int init_percpu_info(struct f2fs_sb_info *sbi)
3078 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3082 err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3085 percpu_counter_destroy(&sbi->alloc_valid_block_count);
3090 #ifdef CONFIG_BLK_DEV_ZONED
3091 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3094 struct f2fs_dev_info *dev = data;
3096 if (zone->type != BLK_ZONE_TYPE_CONVENTIONAL)
3097 set_bit(idx, dev->blkz_seq);
3101 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3103 struct block_device *bdev = FDEV(devi).bdev;
3104 sector_t nr_sectors = bdev->bd_part->nr_sects;
3107 if (!f2fs_sb_has_blkzoned(sbi))
3110 if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3111 SECTOR_TO_BLOCK(bdev_zone_sectors(bdev)))
3113 sbi->blocks_per_blkz = SECTOR_TO_BLOCK(bdev_zone_sectors(bdev));
3114 if (sbi->log_blocks_per_blkz && sbi->log_blocks_per_blkz !=
3115 __ilog2_u32(sbi->blocks_per_blkz))
3117 sbi->log_blocks_per_blkz = __ilog2_u32(sbi->blocks_per_blkz);
3118 FDEV(devi).nr_blkz = SECTOR_TO_BLOCK(nr_sectors) >>
3119 sbi->log_blocks_per_blkz;
3120 if (nr_sectors & (bdev_zone_sectors(bdev) - 1))
3121 FDEV(devi).nr_blkz++;
3123 FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3124 BITS_TO_LONGS(FDEV(devi).nr_blkz)
3125 * sizeof(unsigned long),
3127 if (!FDEV(devi).blkz_seq)
3130 /* Get block zones type */
3131 ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3141 * Read f2fs raw super block.
3142 * Because we have two copies of super block, so read both of them
3143 * to get the first valid one. If any one of them is broken, we pass
3144 * them recovery flag back to the caller.
3146 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3147 struct f2fs_super_block **raw_super,
3148 int *valid_super_block, int *recovery)
3150 struct super_block *sb = sbi->sb;
3152 struct buffer_head *bh;
3153 struct f2fs_super_block *super;
3156 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3160 for (block = 0; block < 2; block++) {
3161 bh = sb_bread(sb, block);
3163 f2fs_err(sbi, "Unable to read %dth superblock",
3170 /* sanity checking of raw super */
3171 err = sanity_check_raw_super(sbi, bh);
3173 f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3181 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3183 *valid_super_block = block;
3189 /* No valid superblock */
3198 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
3200 struct buffer_head *bh;
3204 if ((recover && f2fs_readonly(sbi->sb)) ||
3205 bdev_read_only(sbi->sb->s_bdev)) {
3206 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3210 /* we should update superblock crc here */
3211 if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
3212 crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
3213 offsetof(struct f2fs_super_block, crc));
3214 F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
3217 /* write back-up superblock first */
3218 bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
3221 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3224 /* if we are in recovery path, skip writing valid superblock */
3228 /* write current valid superblock */
3229 bh = sb_bread(sbi->sb, sbi->valid_super_block);
3232 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3237 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
3239 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3240 unsigned int max_devices = MAX_DEVICES;
3243 /* Initialize single device information */
3244 if (!RDEV(0).path[0]) {
3245 if (!bdev_is_zoned(sbi->sb->s_bdev))
3251 * Initialize multiple devices information, or single
3252 * zoned block device information.
3254 sbi->devs = f2fs_kzalloc(sbi,
3255 array_size(max_devices,
3256 sizeof(struct f2fs_dev_info)),
3261 for (i = 0; i < max_devices; i++) {
3263 if (i > 0 && !RDEV(i).path[0])
3266 if (max_devices == 1) {
3267 /* Single zoned block device mount */
3269 blkdev_get_by_dev(sbi->sb->s_bdev->bd_dev,
3270 sbi->sb->s_mode, sbi->sb->s_type);
3272 /* Multi-device mount */
3273 memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
3274 FDEV(i).total_segments =
3275 le32_to_cpu(RDEV(i).total_segments);
3277 FDEV(i).start_blk = 0;
3278 FDEV(i).end_blk = FDEV(i).start_blk +
3279 (FDEV(i).total_segments <<
3280 sbi->log_blocks_per_seg) - 1 +
3281 le32_to_cpu(raw_super->segment0_blkaddr);
3283 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
3284 FDEV(i).end_blk = FDEV(i).start_blk +
3285 (FDEV(i).total_segments <<
3286 sbi->log_blocks_per_seg) - 1;
3288 FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
3289 sbi->sb->s_mode, sbi->sb->s_type);
3291 if (IS_ERR(FDEV(i).bdev))
3292 return PTR_ERR(FDEV(i).bdev);
3294 /* to release errored devices */
3295 sbi->s_ndevs = i + 1;
3297 #ifdef CONFIG_BLK_DEV_ZONED
3298 if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
3299 !f2fs_sb_has_blkzoned(sbi)) {
3300 f2fs_err(sbi, "Zoned block device feature not enabled\n");
3303 if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
3304 if (init_blkz_info(sbi, i)) {
3305 f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
3308 if (max_devices == 1)
3310 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
3312 FDEV(i).total_segments,
3313 FDEV(i).start_blk, FDEV(i).end_blk,
3314 bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
3315 "Host-aware" : "Host-managed");
3319 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
3321 FDEV(i).total_segments,
3322 FDEV(i).start_blk, FDEV(i).end_blk);
3325 "IO Block Size: %8d KB", F2FS_IO_SIZE_KB(sbi));
3329 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
3331 #ifdef CONFIG_UNICODE
3332 if (f2fs_sb_has_casefold(sbi) && !sbi->s_encoding) {
3333 const struct f2fs_sb_encodings *encoding_info;
3334 struct unicode_map *encoding;
3335 __u16 encoding_flags;
3337 if (f2fs_sb_has_encrypt(sbi)) {
3339 "Can't mount with encoding and encryption");
3343 if (f2fs_sb_read_encoding(sbi->raw_super, &encoding_info,
3346 "Encoding requested by superblock is unknown");
3350 encoding = utf8_load(encoding_info->version);
3351 if (IS_ERR(encoding)) {
3353 "can't mount with superblock charset: %s-%s "
3354 "not supported by the kernel. flags: 0x%x.",
3355 encoding_info->name, encoding_info->version,
3357 return PTR_ERR(encoding);
3359 f2fs_info(sbi, "Using encoding defined by superblock: "
3360 "%s-%s with flags 0x%hx", encoding_info->name,
3361 encoding_info->version?:"\b", encoding_flags);
3363 sbi->s_encoding = encoding;
3364 sbi->s_encoding_flags = encoding_flags;
3365 sbi->sb->s_d_op = &f2fs_dentry_ops;
3368 if (f2fs_sb_has_casefold(sbi)) {
3369 f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
3376 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
3378 struct f2fs_sm_info *sm_i = SM_I(sbi);
3380 /* adjust parameters according to the volume size */
3381 if (sm_i->main_segments <= SMALL_VOLUME_SEGMENTS) {
3382 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
3383 sm_i->dcc_info->discard_granularity = 1;
3384 sm_i->ipu_policy = 1 << F2FS_IPU_FORCE;
3387 sbi->readdir_ra = 1;
3390 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
3392 struct f2fs_sb_info *sbi;
3393 struct f2fs_super_block *raw_super;
3396 bool skip_recovery = false, need_fsck = false;
3397 char *options = NULL;
3398 int recovery, i, valid_super_block;
3399 struct curseg_info *seg_i;
3405 valid_super_block = -1;
3408 /* allocate memory for f2fs-specific super block info */
3409 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
3415 /* Load the checksum driver */
3416 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
3417 if (IS_ERR(sbi->s_chksum_driver)) {
3418 f2fs_err(sbi, "Cannot load crc32 driver.");
3419 err = PTR_ERR(sbi->s_chksum_driver);
3420 sbi->s_chksum_driver = NULL;
3424 /* set a block size */
3425 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
3426 f2fs_err(sbi, "unable to set blocksize");
3430 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
3435 sb->s_fs_info = sbi;
3436 sbi->raw_super = raw_super;
3438 /* precompute checksum seed for metadata */
3439 if (f2fs_sb_has_inode_chksum(sbi))
3440 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
3441 sizeof(raw_super->uuid));
3444 * The BLKZONED feature indicates that the drive was formatted with
3445 * zone alignment optimization. This is optional for host-aware
3446 * devices, but mandatory for host-managed zoned block devices.
3448 #ifndef CONFIG_BLK_DEV_ZONED
3449 if (f2fs_sb_has_blkzoned(sbi)) {
3450 f2fs_err(sbi, "Zoned block device support is not enabled");
3455 default_options(sbi);
3456 /* parse mount options */
3457 options = kstrdup((const char *)data, GFP_KERNEL);
3458 if (data && !options) {
3463 err = parse_options(sb, options, false);
3467 sbi->max_file_blocks = max_file_blocks();
3468 sb->s_maxbytes = sbi->max_file_blocks <<
3469 le32_to_cpu(raw_super->log_blocksize);
3470 sb->s_max_links = F2FS_LINK_MAX;
3472 err = f2fs_setup_casefold(sbi);
3477 sb->dq_op = &f2fs_quota_operations;
3478 sb->s_qcop = &f2fs_quotactl_ops;
3479 sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
3481 if (f2fs_sb_has_quota_ino(sbi)) {
3482 for (i = 0; i < MAXQUOTAS; i++) {
3483 if (f2fs_qf_ino(sbi->sb, i))
3484 sbi->nquota_files++;
3489 sb->s_op = &f2fs_sops;
3490 #ifdef CONFIG_FS_ENCRYPTION
3491 sb->s_cop = &f2fs_cryptops;
3493 #ifdef CONFIG_FS_VERITY
3494 sb->s_vop = &f2fs_verityops;
3496 sb->s_xattr = f2fs_xattr_handlers;
3497 sb->s_export_op = &f2fs_export_ops;
3498 sb->s_magic = F2FS_SUPER_MAGIC;
3499 sb->s_time_gran = 1;
3500 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
3501 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
3502 memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
3503 sb->s_iflags |= SB_I_CGROUPWB;
3505 /* init f2fs-specific super block info */
3506 sbi->valid_super_block = valid_super_block;
3507 init_rwsem(&sbi->gc_lock);
3508 mutex_init(&sbi->writepages);
3509 mutex_init(&sbi->cp_mutex);
3510 init_rwsem(&sbi->node_write);
3511 init_rwsem(&sbi->node_change);
3513 /* disallow all the data/node/meta page writes */
3514 set_sbi_flag(sbi, SBI_POR_DOING);
3515 spin_lock_init(&sbi->stat_lock);
3517 /* init iostat info */
3518 spin_lock_init(&sbi->iostat_lock);
3519 sbi->iostat_enable = false;
3520 sbi->iostat_period_ms = DEFAULT_IOSTAT_PERIOD_MS;
3522 for (i = 0; i < NR_PAGE_TYPE; i++) {
3523 int n = (i == META) ? 1: NR_TEMP_TYPE;
3529 sizeof(struct f2fs_bio_info)),
3531 if (!sbi->write_io[i]) {
3536 for (j = HOT; j < n; j++) {
3537 init_rwsem(&sbi->write_io[i][j].io_rwsem);
3538 sbi->write_io[i][j].sbi = sbi;
3539 sbi->write_io[i][j].bio = NULL;
3540 spin_lock_init(&sbi->write_io[i][j].io_lock);
3541 INIT_LIST_HEAD(&sbi->write_io[i][j].io_list);
3542 INIT_LIST_HEAD(&sbi->write_io[i][j].bio_list);
3543 init_rwsem(&sbi->write_io[i][j].bio_list_lock);
3547 init_rwsem(&sbi->cp_rwsem);
3548 init_rwsem(&sbi->quota_sem);
3549 init_waitqueue_head(&sbi->cp_wait);
3552 err = init_percpu_info(sbi);
3556 if (F2FS_IO_ALIGNED(sbi)) {
3557 sbi->write_io_dummy =
3558 mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
3559 if (!sbi->write_io_dummy) {
3565 /* init per sbi slab cache */
3566 err = f2fs_init_xattr_caches(sbi);
3570 /* get an inode for meta space */
3571 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
3572 if (IS_ERR(sbi->meta_inode)) {
3573 f2fs_err(sbi, "Failed to read F2FS meta data inode");
3574 err = PTR_ERR(sbi->meta_inode);
3575 goto free_xattr_cache;
3578 err = f2fs_get_valid_checkpoint(sbi);
3580 f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
3581 goto free_meta_inode;
3584 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
3585 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3586 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
3587 set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
3588 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
3591 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
3592 set_sbi_flag(sbi, SBI_NEED_FSCK);
3594 /* Initialize device list */
3595 err = f2fs_scan_devices(sbi);
3597 f2fs_err(sbi, "Failed to find devices");
3601 err = f2fs_init_post_read_wq(sbi);
3603 f2fs_err(sbi, "Failed to initialize post read workqueue");
3607 sbi->total_valid_node_count =
3608 le32_to_cpu(sbi->ckpt->valid_node_count);
3609 percpu_counter_set(&sbi->total_valid_inode_count,
3610 le32_to_cpu(sbi->ckpt->valid_inode_count));
3611 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
3612 sbi->total_valid_block_count =
3613 le64_to_cpu(sbi->ckpt->valid_block_count);
3614 sbi->last_valid_block_count = sbi->total_valid_block_count;
3615 sbi->reserved_blocks = 0;
3616 sbi->current_reserved_blocks = 0;
3617 limit_reserve_root(sbi);
3618 adjust_unusable_cap_perc(sbi);
3620 for (i = 0; i < NR_INODE_TYPE; i++) {
3621 INIT_LIST_HEAD(&sbi->inode_list[i]);
3622 spin_lock_init(&sbi->inode_lock[i]);
3624 mutex_init(&sbi->flush_lock);
3626 f2fs_init_extent_cache_info(sbi);
3628 f2fs_init_ino_entry_info(sbi);
3630 f2fs_init_fsync_node_info(sbi);
3632 /* setup f2fs internal modules */
3633 err = f2fs_build_segment_manager(sbi);
3635 f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
3639 err = f2fs_build_node_manager(sbi);
3641 f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
3646 /* For write statistics */
3647 if (sb->s_bdev->bd_part)
3648 sbi->sectors_written_start =
3649 (u64)part_stat_read(sb->s_bdev->bd_part,
3650 sectors[STAT_WRITE]);
3652 /* Read accumulated write IO statistics if exists */
3653 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
3654 if (__exist_node_summaries(sbi))
3655 sbi->kbytes_written =
3656 le64_to_cpu(seg_i->journal->info.kbytes_written);
3658 f2fs_build_gc_manager(sbi);
3660 err = f2fs_build_stats(sbi);
3664 /* get an inode for node space */
3665 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
3666 if (IS_ERR(sbi->node_inode)) {
3667 f2fs_err(sbi, "Failed to read node inode");
3668 err = PTR_ERR(sbi->node_inode);
3672 /* read root inode and dentry */
3673 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
3675 f2fs_err(sbi, "Failed to read root inode");
3676 err = PTR_ERR(root);
3677 goto free_node_inode;
3679 if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
3680 !root->i_size || !root->i_nlink) {
3683 goto free_node_inode;
3686 sb->s_root = d_make_root(root); /* allocate root dentry */
3689 goto free_node_inode;
3692 err = f2fs_register_sysfs(sbi);
3694 goto free_root_inode;
3697 /* Enable quota usage during mount */
3698 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
3699 err = f2fs_enable_quotas(sb);
3701 f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
3704 /* if there are any orphan inodes, free them */
3705 err = f2fs_recover_orphan_inodes(sbi);
3709 if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
3710 goto reset_checkpoint;
3712 /* recover fsynced data */
3713 if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
3714 !test_opt(sbi, NORECOVERY)) {
3716 * mount should be failed, when device has readonly mode, and
3717 * previous checkpoint was not done by clean system shutdown.
3719 if (f2fs_hw_is_readonly(sbi)) {
3720 if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
3722 f2fs_err(sbi, "Need to recover fsync data, but write access unavailable");
3725 f2fs_info(sbi, "write access unavailable, skipping recovery");
3726 goto reset_checkpoint;
3730 set_sbi_flag(sbi, SBI_NEED_FSCK);
3733 goto reset_checkpoint;
3735 err = f2fs_recover_fsync_data(sbi, false);
3738 skip_recovery = true;
3740 f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
3745 err = f2fs_recover_fsync_data(sbi, true);
3747 if (!f2fs_readonly(sb) && err > 0) {
3749 f2fs_err(sbi, "Need to recover fsync data");
3755 * If the f2fs is not readonly and fsync data recovery succeeds,
3756 * check zoned block devices' write pointer consistency.
3758 if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
3759 err = f2fs_check_write_pointer(sbi);
3765 /* f2fs_recover_fsync_data() cleared this already */
3766 clear_sbi_flag(sbi, SBI_POR_DOING);
3768 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
3769 err = f2fs_disable_checkpoint(sbi);
3771 goto sync_free_meta;
3772 } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
3773 f2fs_enable_checkpoint(sbi);
3777 * If filesystem is not mounted as read-only then
3778 * do start the gc_thread.
3780 if (F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF && !f2fs_readonly(sb)) {
3781 /* After POR, we can run background GC thread.*/
3782 err = f2fs_start_gc_thread(sbi);
3784 goto sync_free_meta;
3788 /* recover broken superblock */
3790 err = f2fs_commit_super(sbi, true);
3791 f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
3792 sbi->valid_super_block ? 1 : 2, err);
3795 f2fs_join_shrinker(sbi);
3797 f2fs_tuning_parameters(sbi);
3799 f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
3800 cur_cp_version(F2FS_CKPT(sbi)));
3801 f2fs_update_time(sbi, CP_TIME);
3802 f2fs_update_time(sbi, REQ_TIME);
3803 clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
3807 /* safe to flush all the data */
3808 sync_filesystem(sbi->sb);
3813 f2fs_truncate_quota_inode_pages(sb);
3814 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
3815 f2fs_quota_off_umount(sbi->sb);
3818 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
3819 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
3820 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
3821 * falls into an infinite loop in f2fs_sync_meta_pages().
3823 truncate_inode_pages_final(META_MAPPING(sbi));
3824 /* evict some inodes being cached by GC */
3826 f2fs_unregister_sysfs(sbi);
3831 f2fs_release_ino_entry(sbi, true);
3832 truncate_inode_pages_final(NODE_MAPPING(sbi));
3833 iput(sbi->node_inode);
3834 sbi->node_inode = NULL;
3836 f2fs_destroy_stats(sbi);
3838 f2fs_destroy_node_manager(sbi);
3840 f2fs_destroy_segment_manager(sbi);
3841 f2fs_destroy_post_read_wq(sbi);
3843 destroy_device_list(sbi);
3846 make_bad_inode(sbi->meta_inode);
3847 iput(sbi->meta_inode);
3848 sbi->meta_inode = NULL;
3850 f2fs_destroy_xattr_caches(sbi);
3852 mempool_destroy(sbi->write_io_dummy);
3854 destroy_percpu_info(sbi);
3856 for (i = 0; i < NR_PAGE_TYPE; i++)
3857 kvfree(sbi->write_io[i]);
3859 #ifdef CONFIG_UNICODE
3860 utf8_unload(sbi->s_encoding);
3864 for (i = 0; i < MAXQUOTAS; i++)
3865 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
3867 fscrypt_free_dummy_context(&F2FS_OPTION(sbi).dummy_enc_ctx);
3872 if (sbi->s_chksum_driver)
3873 crypto_free_shash(sbi->s_chksum_driver);
3876 /* give only one another chance */
3877 if (retry_cnt > 0 && skip_recovery) {
3879 shrink_dcache_sb(sb);
3885 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
3886 const char *dev_name, void *data)
3888 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
3891 static void kill_f2fs_super(struct super_block *sb)
3894 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3896 set_sbi_flag(sbi, SBI_IS_CLOSE);
3897 f2fs_stop_gc_thread(sbi);
3898 f2fs_stop_discard_thread(sbi);
3900 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
3901 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
3902 struct cp_control cpc = {
3903 .reason = CP_UMOUNT,
3905 f2fs_write_checkpoint(sbi, &cpc);
3908 if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
3909 sb->s_flags &= ~SB_RDONLY;
3911 kill_block_super(sb);
3914 static struct file_system_type f2fs_fs_type = {
3915 .owner = THIS_MODULE,
3917 .mount = f2fs_mount,
3918 .kill_sb = kill_f2fs_super,
3919 .fs_flags = FS_REQUIRES_DEV,
3921 MODULE_ALIAS_FS("f2fs");
3923 static int __init init_inodecache(void)
3925 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
3926 sizeof(struct f2fs_inode_info), 0,
3927 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
3928 if (!f2fs_inode_cachep)
3933 static void destroy_inodecache(void)
3936 * Make sure all delayed rcu free inodes are flushed before we
3940 kmem_cache_destroy(f2fs_inode_cachep);
3943 static int __init init_f2fs_fs(void)
3947 if (PAGE_SIZE != F2FS_BLKSIZE) {
3948 printk("F2FS not supported on PAGE_SIZE(%lu) != %d\n",
3949 PAGE_SIZE, F2FS_BLKSIZE);
3953 f2fs_build_trace_ios();
3955 err = init_inodecache();
3958 err = f2fs_create_node_manager_caches();
3960 goto free_inodecache;
3961 err = f2fs_create_segment_manager_caches();
3963 goto free_node_manager_caches;
3964 err = f2fs_create_checkpoint_caches();
3966 goto free_segment_manager_caches;
3967 err = f2fs_create_extent_cache();
3969 goto free_checkpoint_caches;
3970 err = f2fs_init_sysfs();
3972 goto free_extent_cache;
3973 err = register_shrinker(&f2fs_shrinker_info);
3976 err = register_filesystem(&f2fs_fs_type);
3979 f2fs_create_root_stats();
3980 err = f2fs_init_post_read_processing();
3982 goto free_root_stats;
3983 err = f2fs_init_bio_entry_cache();
3985 goto free_post_read;
3986 err = f2fs_init_bioset();
3988 goto free_bio_enrty_cache;
3989 err = f2fs_init_compress_mempool();
3994 f2fs_destroy_bioset();
3995 free_bio_enrty_cache:
3996 f2fs_destroy_bio_entry_cache();
3998 f2fs_destroy_post_read_processing();
4000 f2fs_destroy_root_stats();
4001 unregister_filesystem(&f2fs_fs_type);
4003 unregister_shrinker(&f2fs_shrinker_info);
4007 f2fs_destroy_extent_cache();
4008 free_checkpoint_caches:
4009 f2fs_destroy_checkpoint_caches();
4010 free_segment_manager_caches:
4011 f2fs_destroy_segment_manager_caches();
4012 free_node_manager_caches:
4013 f2fs_destroy_node_manager_caches();
4015 destroy_inodecache();
4020 static void __exit exit_f2fs_fs(void)
4022 f2fs_destroy_compress_mempool();
4023 f2fs_destroy_bioset();
4024 f2fs_destroy_bio_entry_cache();
4025 f2fs_destroy_post_read_processing();
4026 f2fs_destroy_root_stats();
4027 unregister_filesystem(&f2fs_fs_type);
4028 unregister_shrinker(&f2fs_shrinker_info);
4030 f2fs_destroy_extent_cache();
4031 f2fs_destroy_checkpoint_caches();
4032 f2fs_destroy_segment_manager_caches();
4033 f2fs_destroy_node_manager_caches();
4034 destroy_inodecache();
4035 f2fs_destroy_trace_ios();
4038 module_init(init_f2fs_fs)
4039 module_exit(exit_f2fs_fs)
4041 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
4042 MODULE_DESCRIPTION("Flash Friendly File System");
4043 MODULE_LICENSE("GPL");