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,
153 static match_table_t f2fs_tokens = {
154 {Opt_gc_background, "background_gc=%s"},
155 {Opt_disable_roll_forward, "disable_roll_forward"},
156 {Opt_norecovery, "norecovery"},
157 {Opt_discard, "discard"},
158 {Opt_nodiscard, "nodiscard"},
159 {Opt_noheap, "no_heap"},
161 {Opt_user_xattr, "user_xattr"},
162 {Opt_nouser_xattr, "nouser_xattr"},
164 {Opt_noacl, "noacl"},
165 {Opt_active_logs, "active_logs=%u"},
166 {Opt_disable_ext_identify, "disable_ext_identify"},
167 {Opt_inline_xattr, "inline_xattr"},
168 {Opt_noinline_xattr, "noinline_xattr"},
169 {Opt_inline_xattr_size, "inline_xattr_size=%u"},
170 {Opt_inline_data, "inline_data"},
171 {Opt_inline_dentry, "inline_dentry"},
172 {Opt_noinline_dentry, "noinline_dentry"},
173 {Opt_flush_merge, "flush_merge"},
174 {Opt_noflush_merge, "noflush_merge"},
175 {Opt_nobarrier, "nobarrier"},
176 {Opt_fastboot, "fastboot"},
177 {Opt_extent_cache, "extent_cache"},
178 {Opt_noextent_cache, "noextent_cache"},
179 {Opt_noinline_data, "noinline_data"},
180 {Opt_data_flush, "data_flush"},
181 {Opt_reserve_root, "reserve_root=%u"},
182 {Opt_resgid, "resgid=%u"},
183 {Opt_resuid, "resuid=%u"},
184 {Opt_mode, "mode=%s"},
185 {Opt_io_size_bits, "io_bits=%u"},
186 {Opt_fault_injection, "fault_injection=%u"},
187 {Opt_fault_type, "fault_type=%u"},
188 {Opt_lazytime, "lazytime"},
189 {Opt_nolazytime, "nolazytime"},
190 {Opt_quota, "quota"},
191 {Opt_noquota, "noquota"},
192 {Opt_usrquota, "usrquota"},
193 {Opt_grpquota, "grpquota"},
194 {Opt_prjquota, "prjquota"},
195 {Opt_usrjquota, "usrjquota=%s"},
196 {Opt_grpjquota, "grpjquota=%s"},
197 {Opt_prjjquota, "prjjquota=%s"},
198 {Opt_offusrjquota, "usrjquota="},
199 {Opt_offgrpjquota, "grpjquota="},
200 {Opt_offprjjquota, "prjjquota="},
201 {Opt_jqfmt_vfsold, "jqfmt=vfsold"},
202 {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
203 {Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
204 {Opt_whint, "whint_mode=%s"},
205 {Opt_alloc, "alloc_mode=%s"},
206 {Opt_fsync, "fsync_mode=%s"},
207 {Opt_test_dummy_encryption, "test_dummy_encryption=%s"},
208 {Opt_test_dummy_encryption, "test_dummy_encryption"},
209 {Opt_inlinecrypt, "inlinecrypt"},
210 {Opt_checkpoint_disable, "checkpoint=disable"},
211 {Opt_checkpoint_disable_cap, "checkpoint=disable:%u"},
212 {Opt_checkpoint_disable_cap_perc, "checkpoint=disable:%u%%"},
213 {Opt_checkpoint_enable, "checkpoint=enable"},
214 {Opt_compress_algorithm, "compress_algorithm=%s"},
215 {Opt_compress_log_size, "compress_log_size=%u"},
216 {Opt_compress_extension, "compress_extension=%s"},
221 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...)
223 struct va_format vaf;
229 level = printk_get_level(fmt);
230 vaf.fmt = printk_skip_level(fmt);
232 printk("%c%cF2FS-fs (%s): %pV\n",
233 KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
238 #ifdef CONFIG_UNICODE
239 static const struct f2fs_sb_encodings {
243 } f2fs_sb_encoding_map[] = {
244 {F2FS_ENC_UTF8_12_1, "utf8", "12.1.0"},
247 static int f2fs_sb_read_encoding(const struct f2fs_super_block *sb,
248 const struct f2fs_sb_encodings **encoding,
251 __u16 magic = le16_to_cpu(sb->s_encoding);
254 for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
255 if (magic == f2fs_sb_encoding_map[i].magic)
258 if (i >= ARRAY_SIZE(f2fs_sb_encoding_map))
261 *encoding = &f2fs_sb_encoding_map[i];
262 *flags = le16_to_cpu(sb->s_encoding_flags);
268 static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
270 block_t limit = min((sbi->user_block_count << 1) / 1000,
271 sbi->user_block_count - sbi->reserved_blocks);
274 if (test_opt(sbi, RESERVE_ROOT) &&
275 F2FS_OPTION(sbi).root_reserved_blocks > limit) {
276 F2FS_OPTION(sbi).root_reserved_blocks = limit;
277 f2fs_info(sbi, "Reduce reserved blocks for root = %u",
278 F2FS_OPTION(sbi).root_reserved_blocks);
280 if (!test_opt(sbi, RESERVE_ROOT) &&
281 (!uid_eq(F2FS_OPTION(sbi).s_resuid,
282 make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
283 !gid_eq(F2FS_OPTION(sbi).s_resgid,
284 make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
285 f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
286 from_kuid_munged(&init_user_ns,
287 F2FS_OPTION(sbi).s_resuid),
288 from_kgid_munged(&init_user_ns,
289 F2FS_OPTION(sbi).s_resgid));
292 static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
294 if (!F2FS_OPTION(sbi).unusable_cap_perc)
297 if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
298 F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
300 F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
301 F2FS_OPTION(sbi).unusable_cap_perc;
303 f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
304 F2FS_OPTION(sbi).unusable_cap,
305 F2FS_OPTION(sbi).unusable_cap_perc);
308 static void init_once(void *foo)
310 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
312 inode_init_once(&fi->vfs_inode);
316 static const char * const quotatypes[] = INITQFNAMES;
317 #define QTYPE2NAME(t) (quotatypes[t])
318 static int f2fs_set_qf_name(struct super_block *sb, int qtype,
321 struct f2fs_sb_info *sbi = F2FS_SB(sb);
325 if (sb_any_quota_loaded(sb) && !F2FS_OPTION(sbi).s_qf_names[qtype]) {
326 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
329 if (f2fs_sb_has_quota_ino(sbi)) {
330 f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
334 qname = match_strdup(args);
336 f2fs_err(sbi, "Not enough memory for storing quotafile name");
339 if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
340 if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
343 f2fs_err(sbi, "%s quota file already specified",
347 if (strchr(qname, '/')) {
348 f2fs_err(sbi, "quotafile must be on filesystem root");
351 F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
359 static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
361 struct f2fs_sb_info *sbi = F2FS_SB(sb);
363 if (sb_any_quota_loaded(sb) && F2FS_OPTION(sbi).s_qf_names[qtype]) {
364 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
367 kfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
368 F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
372 static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
375 * We do the test below only for project quotas. 'usrquota' and
376 * 'grpquota' mount options are allowed even without quota feature
377 * to support legacy quotas in quota files.
379 if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi)) {
380 f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
383 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
384 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
385 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) {
386 if (test_opt(sbi, USRQUOTA) &&
387 F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
388 clear_opt(sbi, USRQUOTA);
390 if (test_opt(sbi, GRPQUOTA) &&
391 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
392 clear_opt(sbi, GRPQUOTA);
394 if (test_opt(sbi, PRJQUOTA) &&
395 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
396 clear_opt(sbi, PRJQUOTA);
398 if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
399 test_opt(sbi, PRJQUOTA)) {
400 f2fs_err(sbi, "old and new quota format mixing");
404 if (!F2FS_OPTION(sbi).s_jquota_fmt) {
405 f2fs_err(sbi, "journaled quota format not specified");
410 if (f2fs_sb_has_quota_ino(sbi) && F2FS_OPTION(sbi).s_jquota_fmt) {
411 f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
412 F2FS_OPTION(sbi).s_jquota_fmt = 0;
418 static int f2fs_set_test_dummy_encryption(struct super_block *sb,
420 const substring_t *arg,
423 struct f2fs_sb_info *sbi = F2FS_SB(sb);
424 #ifdef CONFIG_FS_ENCRYPTION
427 if (!f2fs_sb_has_encrypt(sbi)) {
428 f2fs_err(sbi, "Encrypt feature is off");
433 * This mount option is just for testing, and it's not worthwhile to
434 * implement the extra complexity (e.g. RCU protection) that would be
435 * needed to allow it to be set or changed during remount. We do allow
436 * it to be specified during remount, but only if there is no change.
438 if (is_remount && !F2FS_OPTION(sbi).dummy_enc_policy.policy) {
439 f2fs_warn(sbi, "Can't set test_dummy_encryption on remount");
442 err = fscrypt_set_test_dummy_encryption(
443 sb, arg->from, &F2FS_OPTION(sbi).dummy_enc_policy);
447 "Can't change test_dummy_encryption on remount");
448 else if (err == -EINVAL)
449 f2fs_warn(sbi, "Value of option \"%s\" is unrecognized",
452 f2fs_warn(sbi, "Error processing option \"%s\" [%d]",
456 f2fs_warn(sbi, "Test dummy encryption mode enabled");
458 f2fs_warn(sbi, "Test dummy encryption mount option ignored");
463 static int parse_options(struct super_block *sb, char *options, bool is_remount)
465 struct f2fs_sb_info *sbi = F2FS_SB(sb);
466 substring_t args[MAX_OPT_ARGS];
467 #ifdef CONFIG_F2FS_FS_COMPRESSION
468 unsigned char (*ext)[F2FS_EXTENSION_LEN];
480 while ((p = strsep(&options, ",")) != NULL) {
485 * Initialize args struct so we know whether arg was
486 * found; some options take optional arguments.
488 args[0].to = args[0].from = NULL;
489 token = match_token(p, f2fs_tokens, args);
492 case Opt_gc_background:
493 name = match_strdup(&args[0]);
497 if (!strcmp(name, "on")) {
498 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
499 } else if (!strcmp(name, "off")) {
500 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_OFF;
501 } else if (!strcmp(name, "sync")) {
502 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_SYNC;
509 case Opt_disable_roll_forward:
510 set_opt(sbi, DISABLE_ROLL_FORWARD);
513 /* this option mounts f2fs with ro */
514 set_opt(sbi, NORECOVERY);
515 if (!f2fs_readonly(sb))
519 set_opt(sbi, DISCARD);
522 if (f2fs_sb_has_blkzoned(sbi)) {
523 f2fs_warn(sbi, "discard is required for zoned block devices");
526 clear_opt(sbi, DISCARD);
529 set_opt(sbi, NOHEAP);
532 clear_opt(sbi, NOHEAP);
534 #ifdef CONFIG_F2FS_FS_XATTR
536 set_opt(sbi, XATTR_USER);
538 case Opt_nouser_xattr:
539 clear_opt(sbi, XATTR_USER);
541 case Opt_inline_xattr:
542 set_opt(sbi, INLINE_XATTR);
544 case Opt_noinline_xattr:
545 clear_opt(sbi, INLINE_XATTR);
547 case Opt_inline_xattr_size:
548 if (args->from && match_int(args, &arg))
550 set_opt(sbi, INLINE_XATTR_SIZE);
551 F2FS_OPTION(sbi).inline_xattr_size = arg;
555 f2fs_info(sbi, "user_xattr options not supported");
557 case Opt_nouser_xattr:
558 f2fs_info(sbi, "nouser_xattr options not supported");
560 case Opt_inline_xattr:
561 f2fs_info(sbi, "inline_xattr options not supported");
563 case Opt_noinline_xattr:
564 f2fs_info(sbi, "noinline_xattr options not supported");
567 #ifdef CONFIG_F2FS_FS_POSIX_ACL
569 set_opt(sbi, POSIX_ACL);
572 clear_opt(sbi, POSIX_ACL);
576 f2fs_info(sbi, "acl options not supported");
579 f2fs_info(sbi, "noacl options not supported");
582 case Opt_active_logs:
583 if (args->from && match_int(args, &arg))
585 if (arg != 2 && arg != 4 &&
586 arg != NR_CURSEG_PERSIST_TYPE)
588 F2FS_OPTION(sbi).active_logs = arg;
590 case Opt_disable_ext_identify:
591 set_opt(sbi, DISABLE_EXT_IDENTIFY);
593 case Opt_inline_data:
594 set_opt(sbi, INLINE_DATA);
596 case Opt_inline_dentry:
597 set_opt(sbi, INLINE_DENTRY);
599 case Opt_noinline_dentry:
600 clear_opt(sbi, INLINE_DENTRY);
602 case Opt_flush_merge:
603 set_opt(sbi, FLUSH_MERGE);
605 case Opt_noflush_merge:
606 clear_opt(sbi, FLUSH_MERGE);
609 set_opt(sbi, NOBARRIER);
612 set_opt(sbi, FASTBOOT);
614 case Opt_extent_cache:
615 set_opt(sbi, EXTENT_CACHE);
617 case Opt_noextent_cache:
618 clear_opt(sbi, EXTENT_CACHE);
620 case Opt_noinline_data:
621 clear_opt(sbi, INLINE_DATA);
624 set_opt(sbi, DATA_FLUSH);
626 case Opt_reserve_root:
627 if (args->from && match_int(args, &arg))
629 if (test_opt(sbi, RESERVE_ROOT)) {
630 f2fs_info(sbi, "Preserve previous reserve_root=%u",
631 F2FS_OPTION(sbi).root_reserved_blocks);
633 F2FS_OPTION(sbi).root_reserved_blocks = arg;
634 set_opt(sbi, RESERVE_ROOT);
638 if (args->from && match_int(args, &arg))
640 uid = make_kuid(current_user_ns(), arg);
641 if (!uid_valid(uid)) {
642 f2fs_err(sbi, "Invalid uid value %d", arg);
645 F2FS_OPTION(sbi).s_resuid = uid;
648 if (args->from && match_int(args, &arg))
650 gid = make_kgid(current_user_ns(), arg);
651 if (!gid_valid(gid)) {
652 f2fs_err(sbi, "Invalid gid value %d", arg);
655 F2FS_OPTION(sbi).s_resgid = gid;
658 name = match_strdup(&args[0]);
662 if (!strcmp(name, "adaptive")) {
663 if (f2fs_sb_has_blkzoned(sbi)) {
664 f2fs_warn(sbi, "adaptive mode is not allowed with zoned block device feature");
668 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
669 } else if (!strcmp(name, "lfs")) {
670 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
677 case Opt_io_size_bits:
678 if (args->from && match_int(args, &arg))
680 if (arg <= 0 || arg > __ilog2_u32(BIO_MAX_PAGES)) {
681 f2fs_warn(sbi, "Not support %d, larger than %d",
682 1 << arg, BIO_MAX_PAGES);
685 F2FS_OPTION(sbi).write_io_size_bits = arg;
687 #ifdef CONFIG_F2FS_FAULT_INJECTION
688 case Opt_fault_injection:
689 if (args->from && match_int(args, &arg))
691 f2fs_build_fault_attr(sbi, arg, F2FS_ALL_FAULT_TYPE);
692 set_opt(sbi, FAULT_INJECTION);
696 if (args->from && match_int(args, &arg))
698 f2fs_build_fault_attr(sbi, 0, arg);
699 set_opt(sbi, FAULT_INJECTION);
702 case Opt_fault_injection:
703 f2fs_info(sbi, "fault_injection options not supported");
707 f2fs_info(sbi, "fault_type options not supported");
711 sb->s_flags |= SB_LAZYTIME;
714 sb->s_flags &= ~SB_LAZYTIME;
719 set_opt(sbi, USRQUOTA);
722 set_opt(sbi, GRPQUOTA);
725 set_opt(sbi, PRJQUOTA);
728 ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
733 ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
738 ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
742 case Opt_offusrjquota:
743 ret = f2fs_clear_qf_name(sb, USRQUOTA);
747 case Opt_offgrpjquota:
748 ret = f2fs_clear_qf_name(sb, GRPQUOTA);
752 case Opt_offprjjquota:
753 ret = f2fs_clear_qf_name(sb, PRJQUOTA);
757 case Opt_jqfmt_vfsold:
758 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
760 case Opt_jqfmt_vfsv0:
761 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
763 case Opt_jqfmt_vfsv1:
764 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
767 clear_opt(sbi, QUOTA);
768 clear_opt(sbi, USRQUOTA);
769 clear_opt(sbi, GRPQUOTA);
770 clear_opt(sbi, PRJQUOTA);
780 case Opt_offusrjquota:
781 case Opt_offgrpjquota:
782 case Opt_offprjjquota:
783 case Opt_jqfmt_vfsold:
784 case Opt_jqfmt_vfsv0:
785 case Opt_jqfmt_vfsv1:
787 f2fs_info(sbi, "quota operations not supported");
791 name = match_strdup(&args[0]);
794 if (!strcmp(name, "user-based")) {
795 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_USER;
796 } else if (!strcmp(name, "off")) {
797 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
798 } else if (!strcmp(name, "fs-based")) {
799 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_FS;
807 name = match_strdup(&args[0]);
811 if (!strcmp(name, "default")) {
812 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
813 } else if (!strcmp(name, "reuse")) {
814 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
822 name = match_strdup(&args[0]);
825 if (!strcmp(name, "posix")) {
826 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
827 } else if (!strcmp(name, "strict")) {
828 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_STRICT;
829 } else if (!strcmp(name, "nobarrier")) {
830 F2FS_OPTION(sbi).fsync_mode =
831 FSYNC_MODE_NOBARRIER;
838 case Opt_test_dummy_encryption:
839 ret = f2fs_set_test_dummy_encryption(sb, p, &args[0],
844 case Opt_inlinecrypt:
845 #ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
846 sb->s_flags |= SB_INLINECRYPT;
848 f2fs_info(sbi, "inline encryption not supported");
851 case Opt_checkpoint_disable_cap_perc:
852 if (args->from && match_int(args, &arg))
854 if (arg < 0 || arg > 100)
856 F2FS_OPTION(sbi).unusable_cap_perc = arg;
857 set_opt(sbi, DISABLE_CHECKPOINT);
859 case Opt_checkpoint_disable_cap:
860 if (args->from && match_int(args, &arg))
862 F2FS_OPTION(sbi).unusable_cap = arg;
863 set_opt(sbi, DISABLE_CHECKPOINT);
865 case Opt_checkpoint_disable:
866 set_opt(sbi, DISABLE_CHECKPOINT);
868 case Opt_checkpoint_enable:
869 clear_opt(sbi, DISABLE_CHECKPOINT);
871 #ifdef CONFIG_F2FS_FS_COMPRESSION
872 case Opt_compress_algorithm:
873 if (!f2fs_sb_has_compression(sbi)) {
874 f2fs_info(sbi, "Image doesn't support compression");
877 name = match_strdup(&args[0]);
880 if (!strcmp(name, "lzo")) {
881 F2FS_OPTION(sbi).compress_algorithm =
883 } else if (!strcmp(name, "lz4")) {
884 F2FS_OPTION(sbi).compress_algorithm =
886 } else if (!strcmp(name, "zstd")) {
887 F2FS_OPTION(sbi).compress_algorithm =
889 } else if (!strcmp(name, "lzo-rle")) {
890 F2FS_OPTION(sbi).compress_algorithm =
898 case Opt_compress_log_size:
899 if (!f2fs_sb_has_compression(sbi)) {
900 f2fs_info(sbi, "Image doesn't support compression");
903 if (args->from && match_int(args, &arg))
905 if (arg < MIN_COMPRESS_LOG_SIZE ||
906 arg > MAX_COMPRESS_LOG_SIZE) {
908 "Compress cluster log size is out of range");
911 F2FS_OPTION(sbi).compress_log_size = arg;
913 case Opt_compress_extension:
914 if (!f2fs_sb_has_compression(sbi)) {
915 f2fs_info(sbi, "Image doesn't support compression");
918 name = match_strdup(&args[0]);
922 ext = F2FS_OPTION(sbi).extensions;
923 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
925 if (strlen(name) >= F2FS_EXTENSION_LEN ||
926 ext_cnt >= COMPRESS_EXT_NUM) {
928 "invalid extension length/number");
933 strcpy(ext[ext_cnt], name);
934 F2FS_OPTION(sbi).compress_ext_cnt++;
938 case Opt_compress_algorithm:
939 case Opt_compress_log_size:
940 case Opt_compress_extension:
941 f2fs_info(sbi, "compression options not supported");
948 f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
954 if (f2fs_check_quota_options(sbi))
957 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sbi->sb)) {
958 f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
961 if (f2fs_sb_has_project_quota(sbi) && !f2fs_readonly(sbi->sb)) {
962 f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
966 #ifndef CONFIG_UNICODE
967 if (f2fs_sb_has_casefold(sbi)) {
969 "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
974 * The BLKZONED feature indicates that the drive was formatted with
975 * zone alignment optimization. This is optional for host-aware
976 * devices, but mandatory for host-managed zoned block devices.
978 #ifndef CONFIG_BLK_DEV_ZONED
979 if (f2fs_sb_has_blkzoned(sbi)) {
980 f2fs_err(sbi, "Zoned block device support is not enabled");
985 if (F2FS_IO_SIZE_BITS(sbi) && !f2fs_lfs_mode(sbi)) {
986 f2fs_err(sbi, "Should set mode=lfs with %uKB-sized IO",
987 F2FS_IO_SIZE_KB(sbi));
991 if (test_opt(sbi, INLINE_XATTR_SIZE)) {
992 int min_size, max_size;
994 if (!f2fs_sb_has_extra_attr(sbi) ||
995 !f2fs_sb_has_flexible_inline_xattr(sbi)) {
996 f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
999 if (!test_opt(sbi, INLINE_XATTR)) {
1000 f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
1004 min_size = sizeof(struct f2fs_xattr_header) / sizeof(__le32);
1005 max_size = MAX_INLINE_XATTR_SIZE;
1007 if (F2FS_OPTION(sbi).inline_xattr_size < min_size ||
1008 F2FS_OPTION(sbi).inline_xattr_size > max_size) {
1009 f2fs_err(sbi, "inline xattr size is out of range: %d ~ %d",
1010 min_size, max_size);
1015 if (test_opt(sbi, DISABLE_CHECKPOINT) && f2fs_lfs_mode(sbi)) {
1016 f2fs_err(sbi, "LFS not compatible with checkpoint=disable\n");
1020 /* Not pass down write hints if the number of active logs is lesser
1021 * than NR_CURSEG_PERSIST_TYPE.
1023 if (F2FS_OPTION(sbi).active_logs != NR_CURSEG_TYPE)
1024 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1028 static struct inode *f2fs_alloc_inode(struct super_block *sb)
1030 struct f2fs_inode_info *fi;
1032 fi = kmem_cache_alloc(f2fs_inode_cachep, GFP_F2FS_ZERO);
1036 init_once((void *) fi);
1038 /* Initialize f2fs-specific inode info */
1039 atomic_set(&fi->dirty_pages, 0);
1040 atomic_set(&fi->i_compr_blocks, 0);
1041 init_rwsem(&fi->i_sem);
1042 spin_lock_init(&fi->i_size_lock);
1043 INIT_LIST_HEAD(&fi->dirty_list);
1044 INIT_LIST_HEAD(&fi->gdirty_list);
1045 INIT_LIST_HEAD(&fi->inmem_ilist);
1046 INIT_LIST_HEAD(&fi->inmem_pages);
1047 mutex_init(&fi->inmem_lock);
1048 init_rwsem(&fi->i_gc_rwsem[READ]);
1049 init_rwsem(&fi->i_gc_rwsem[WRITE]);
1050 init_rwsem(&fi->i_mmap_sem);
1051 init_rwsem(&fi->i_xattr_sem);
1053 /* Will be used by directory only */
1054 fi->i_dir_level = F2FS_SB(sb)->dir_level;
1058 return &fi->vfs_inode;
1061 static int f2fs_drop_inode(struct inode *inode)
1063 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1067 * during filesystem shutdown, if checkpoint is disabled,
1068 * drop useless meta/node dirty pages.
1070 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1071 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1072 inode->i_ino == F2FS_META_INO(sbi)) {
1073 trace_f2fs_drop_inode(inode, 1);
1079 * This is to avoid a deadlock condition like below.
1080 * writeback_single_inode(inode)
1081 * - f2fs_write_data_page
1082 * - f2fs_gc -> iput -> evict
1083 * - inode_wait_for_writeback(inode)
1085 if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
1086 if (!inode->i_nlink && !is_bad_inode(inode)) {
1087 /* to avoid evict_inode call simultaneously */
1088 atomic_inc(&inode->i_count);
1089 spin_unlock(&inode->i_lock);
1091 /* some remained atomic pages should discarded */
1092 if (f2fs_is_atomic_file(inode))
1093 f2fs_drop_inmem_pages(inode);
1095 /* should remain fi->extent_tree for writepage */
1096 f2fs_destroy_extent_node(inode);
1098 sb_start_intwrite(inode->i_sb);
1099 f2fs_i_size_write(inode, 0);
1101 f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1102 inode, NULL, 0, DATA);
1103 truncate_inode_pages_final(inode->i_mapping);
1105 if (F2FS_HAS_BLOCKS(inode))
1106 f2fs_truncate(inode);
1108 sb_end_intwrite(inode->i_sb);
1110 spin_lock(&inode->i_lock);
1111 atomic_dec(&inode->i_count);
1113 trace_f2fs_drop_inode(inode, 0);
1116 ret = generic_drop_inode(inode);
1118 ret = fscrypt_drop_inode(inode);
1119 trace_f2fs_drop_inode(inode, ret);
1123 int f2fs_inode_dirtied(struct inode *inode, bool sync)
1125 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1128 spin_lock(&sbi->inode_lock[DIRTY_META]);
1129 if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1132 set_inode_flag(inode, FI_DIRTY_INODE);
1133 stat_inc_dirty_inode(sbi, DIRTY_META);
1135 if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1136 list_add_tail(&F2FS_I(inode)->gdirty_list,
1137 &sbi->inode_list[DIRTY_META]);
1138 inc_page_count(sbi, F2FS_DIRTY_IMETA);
1140 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1144 void f2fs_inode_synced(struct inode *inode)
1146 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1148 spin_lock(&sbi->inode_lock[DIRTY_META]);
1149 if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1150 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1153 if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1154 list_del_init(&F2FS_I(inode)->gdirty_list);
1155 dec_page_count(sbi, F2FS_DIRTY_IMETA);
1157 clear_inode_flag(inode, FI_DIRTY_INODE);
1158 clear_inode_flag(inode, FI_AUTO_RECOVER);
1159 stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1160 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1164 * f2fs_dirty_inode() is called from __mark_inode_dirty()
1166 * We should call set_dirty_inode to write the dirty inode through write_inode.
1168 static void f2fs_dirty_inode(struct inode *inode, int flags)
1170 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1172 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1173 inode->i_ino == F2FS_META_INO(sbi))
1176 if (flags == I_DIRTY_TIME)
1179 if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1180 clear_inode_flag(inode, FI_AUTO_RECOVER);
1182 f2fs_inode_dirtied(inode, false);
1185 static void f2fs_free_inode(struct inode *inode)
1187 fscrypt_free_inode(inode);
1188 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1191 static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1193 percpu_counter_destroy(&sbi->alloc_valid_block_count);
1194 percpu_counter_destroy(&sbi->total_valid_inode_count);
1197 static void destroy_device_list(struct f2fs_sb_info *sbi)
1201 for (i = 0; i < sbi->s_ndevs; i++) {
1202 blkdev_put(FDEV(i).bdev, FMODE_EXCL);
1203 #ifdef CONFIG_BLK_DEV_ZONED
1204 kvfree(FDEV(i).blkz_seq);
1205 kfree(FDEV(i).zone_capacity_blocks);
1211 static void f2fs_put_super(struct super_block *sb)
1213 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1217 /* unregister procfs/sysfs entries in advance to avoid race case */
1218 f2fs_unregister_sysfs(sbi);
1220 f2fs_quota_off_umount(sb);
1222 /* prevent remaining shrinker jobs */
1223 mutex_lock(&sbi->umount_mutex);
1226 * We don't need to do checkpoint when superblock is clean.
1227 * But, the previous checkpoint was not done by umount, it needs to do
1228 * clean checkpoint again.
1230 if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1231 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1232 struct cp_control cpc = {
1233 .reason = CP_UMOUNT,
1235 f2fs_write_checkpoint(sbi, &cpc);
1238 /* be sure to wait for any on-going discard commands */
1239 dropped = f2fs_issue_discard_timeout(sbi);
1241 if ((f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi)) &&
1242 !sbi->discard_blks && !dropped) {
1243 struct cp_control cpc = {
1244 .reason = CP_UMOUNT | CP_TRIMMED,
1246 f2fs_write_checkpoint(sbi, &cpc);
1250 * normally superblock is clean, so we need to release this.
1251 * In addition, EIO will skip do checkpoint, we need this as well.
1253 f2fs_release_ino_entry(sbi, true);
1255 f2fs_leave_shrinker(sbi);
1256 mutex_unlock(&sbi->umount_mutex);
1258 /* our cp_error case, we can wait for any writeback page */
1259 f2fs_flush_merged_writes(sbi);
1261 f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1263 f2fs_bug_on(sbi, sbi->fsync_node_num);
1265 iput(sbi->node_inode);
1266 sbi->node_inode = NULL;
1268 iput(sbi->meta_inode);
1269 sbi->meta_inode = NULL;
1272 * iput() can update stat information, if f2fs_write_checkpoint()
1273 * above failed with error.
1275 f2fs_destroy_stats(sbi);
1277 /* destroy f2fs internal modules */
1278 f2fs_destroy_node_manager(sbi);
1279 f2fs_destroy_segment_manager(sbi);
1281 f2fs_destroy_post_read_wq(sbi);
1285 sb->s_fs_info = NULL;
1286 if (sbi->s_chksum_driver)
1287 crypto_free_shash(sbi->s_chksum_driver);
1288 kfree(sbi->raw_super);
1290 destroy_device_list(sbi);
1291 f2fs_destroy_page_array_cache(sbi);
1292 f2fs_destroy_xattr_caches(sbi);
1293 mempool_destroy(sbi->write_io_dummy);
1295 for (i = 0; i < MAXQUOTAS; i++)
1296 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1298 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
1299 destroy_percpu_info(sbi);
1300 for (i = 0; i < NR_PAGE_TYPE; i++)
1301 kvfree(sbi->write_io[i]);
1302 #ifdef CONFIG_UNICODE
1303 utf8_unload(sb->s_encoding);
1308 int f2fs_sync_fs(struct super_block *sb, int sync)
1310 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1313 if (unlikely(f2fs_cp_error(sbi)))
1315 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1318 trace_f2fs_sync_fs(sb, sync);
1320 if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1324 struct cp_control cpc;
1326 cpc.reason = __get_cp_reason(sbi);
1328 down_write(&sbi->gc_lock);
1329 err = f2fs_write_checkpoint(sbi, &cpc);
1330 up_write(&sbi->gc_lock);
1332 f2fs_trace_ios(NULL, 1);
1337 static int f2fs_freeze(struct super_block *sb)
1339 if (f2fs_readonly(sb))
1342 /* IO error happened before */
1343 if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1346 /* must be clean, since sync_filesystem() was already called */
1347 if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1352 static int f2fs_unfreeze(struct super_block *sb)
1358 static int f2fs_statfs_project(struct super_block *sb,
1359 kprojid_t projid, struct kstatfs *buf)
1362 struct dquot *dquot;
1366 qid = make_kqid_projid(projid);
1367 dquot = dqget(sb, qid);
1369 return PTR_ERR(dquot);
1370 spin_lock(&dquot->dq_dqb_lock);
1372 limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1373 dquot->dq_dqb.dqb_bhardlimit);
1375 limit >>= sb->s_blocksize_bits;
1377 if (limit && buf->f_blocks > limit) {
1378 curblock = (dquot->dq_dqb.dqb_curspace +
1379 dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
1380 buf->f_blocks = limit;
1381 buf->f_bfree = buf->f_bavail =
1382 (buf->f_blocks > curblock) ?
1383 (buf->f_blocks - curblock) : 0;
1386 limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1387 dquot->dq_dqb.dqb_ihardlimit);
1389 if (limit && buf->f_files > limit) {
1390 buf->f_files = limit;
1392 (buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1393 (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1396 spin_unlock(&dquot->dq_dqb_lock);
1402 static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1404 struct super_block *sb = dentry->d_sb;
1405 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1406 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
1407 block_t total_count, user_block_count, start_count;
1408 u64 avail_node_count;
1410 total_count = le64_to_cpu(sbi->raw_super->block_count);
1411 user_block_count = sbi->user_block_count;
1412 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
1413 buf->f_type = F2FS_SUPER_MAGIC;
1414 buf->f_bsize = sbi->blocksize;
1416 buf->f_blocks = total_count - start_count;
1417 buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1418 sbi->current_reserved_blocks;
1420 spin_lock(&sbi->stat_lock);
1421 if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1424 buf->f_bfree -= sbi->unusable_block_count;
1425 spin_unlock(&sbi->stat_lock);
1427 if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1428 buf->f_bavail = buf->f_bfree -
1429 F2FS_OPTION(sbi).root_reserved_blocks;
1433 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
1435 if (avail_node_count > user_block_count) {
1436 buf->f_files = user_block_count;
1437 buf->f_ffree = buf->f_bavail;
1439 buf->f_files = avail_node_count;
1440 buf->f_ffree = min(avail_node_count - valid_node_count(sbi),
1444 buf->f_namelen = F2FS_NAME_LEN;
1445 buf->f_fsid = u64_to_fsid(id);
1448 if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1449 sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1450 f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1456 static inline void f2fs_show_quota_options(struct seq_file *seq,
1457 struct super_block *sb)
1460 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1462 if (F2FS_OPTION(sbi).s_jquota_fmt) {
1465 switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1476 seq_printf(seq, ",jqfmt=%s", fmtname);
1479 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1480 seq_show_option(seq, "usrjquota",
1481 F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1483 if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1484 seq_show_option(seq, "grpjquota",
1485 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1487 if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1488 seq_show_option(seq, "prjjquota",
1489 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1493 static inline void f2fs_show_compress_options(struct seq_file *seq,
1494 struct super_block *sb)
1496 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1500 if (!f2fs_sb_has_compression(sbi))
1503 switch (F2FS_OPTION(sbi).compress_algorithm) {
1513 case COMPRESS_LZORLE:
1514 algtype = "lzo-rle";
1517 seq_printf(seq, ",compress_algorithm=%s", algtype);
1519 seq_printf(seq, ",compress_log_size=%u",
1520 F2FS_OPTION(sbi).compress_log_size);
1522 for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1523 seq_printf(seq, ",compress_extension=%s",
1524 F2FS_OPTION(sbi).extensions[i]);
1528 static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1530 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1532 if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1533 seq_printf(seq, ",background_gc=%s", "sync");
1534 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1535 seq_printf(seq, ",background_gc=%s", "on");
1536 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1537 seq_printf(seq, ",background_gc=%s", "off");
1539 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1540 seq_puts(seq, ",disable_roll_forward");
1541 if (test_opt(sbi, NORECOVERY))
1542 seq_puts(seq, ",norecovery");
1543 if (test_opt(sbi, DISCARD))
1544 seq_puts(seq, ",discard");
1546 seq_puts(seq, ",nodiscard");
1547 if (test_opt(sbi, NOHEAP))
1548 seq_puts(seq, ",no_heap");
1550 seq_puts(seq, ",heap");
1551 #ifdef CONFIG_F2FS_FS_XATTR
1552 if (test_opt(sbi, XATTR_USER))
1553 seq_puts(seq, ",user_xattr");
1555 seq_puts(seq, ",nouser_xattr");
1556 if (test_opt(sbi, INLINE_XATTR))
1557 seq_puts(seq, ",inline_xattr");
1559 seq_puts(seq, ",noinline_xattr");
1560 if (test_opt(sbi, INLINE_XATTR_SIZE))
1561 seq_printf(seq, ",inline_xattr_size=%u",
1562 F2FS_OPTION(sbi).inline_xattr_size);
1564 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1565 if (test_opt(sbi, POSIX_ACL))
1566 seq_puts(seq, ",acl");
1568 seq_puts(seq, ",noacl");
1570 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
1571 seq_puts(seq, ",disable_ext_identify");
1572 if (test_opt(sbi, INLINE_DATA))
1573 seq_puts(seq, ",inline_data");
1575 seq_puts(seq, ",noinline_data");
1576 if (test_opt(sbi, INLINE_DENTRY))
1577 seq_puts(seq, ",inline_dentry");
1579 seq_puts(seq, ",noinline_dentry");
1580 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
1581 seq_puts(seq, ",flush_merge");
1582 if (test_opt(sbi, NOBARRIER))
1583 seq_puts(seq, ",nobarrier");
1584 if (test_opt(sbi, FASTBOOT))
1585 seq_puts(seq, ",fastboot");
1586 if (test_opt(sbi, EXTENT_CACHE))
1587 seq_puts(seq, ",extent_cache");
1589 seq_puts(seq, ",noextent_cache");
1590 if (test_opt(sbi, DATA_FLUSH))
1591 seq_puts(seq, ",data_flush");
1593 seq_puts(seq, ",mode=");
1594 if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
1595 seq_puts(seq, "adaptive");
1596 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
1597 seq_puts(seq, "lfs");
1598 seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
1599 if (test_opt(sbi, RESERVE_ROOT))
1600 seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
1601 F2FS_OPTION(sbi).root_reserved_blocks,
1602 from_kuid_munged(&init_user_ns,
1603 F2FS_OPTION(sbi).s_resuid),
1604 from_kgid_munged(&init_user_ns,
1605 F2FS_OPTION(sbi).s_resgid));
1606 if (F2FS_IO_SIZE_BITS(sbi))
1607 seq_printf(seq, ",io_bits=%u",
1608 F2FS_OPTION(sbi).write_io_size_bits);
1609 #ifdef CONFIG_F2FS_FAULT_INJECTION
1610 if (test_opt(sbi, FAULT_INJECTION)) {
1611 seq_printf(seq, ",fault_injection=%u",
1612 F2FS_OPTION(sbi).fault_info.inject_rate);
1613 seq_printf(seq, ",fault_type=%u",
1614 F2FS_OPTION(sbi).fault_info.inject_type);
1618 if (test_opt(sbi, QUOTA))
1619 seq_puts(seq, ",quota");
1620 if (test_opt(sbi, USRQUOTA))
1621 seq_puts(seq, ",usrquota");
1622 if (test_opt(sbi, GRPQUOTA))
1623 seq_puts(seq, ",grpquota");
1624 if (test_opt(sbi, PRJQUOTA))
1625 seq_puts(seq, ",prjquota");
1627 f2fs_show_quota_options(seq, sbi->sb);
1628 if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_USER)
1629 seq_printf(seq, ",whint_mode=%s", "user-based");
1630 else if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_FS)
1631 seq_printf(seq, ",whint_mode=%s", "fs-based");
1633 fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
1635 if (sbi->sb->s_flags & SB_INLINECRYPT)
1636 seq_puts(seq, ",inlinecrypt");
1638 if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
1639 seq_printf(seq, ",alloc_mode=%s", "default");
1640 else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
1641 seq_printf(seq, ",alloc_mode=%s", "reuse");
1643 if (test_opt(sbi, DISABLE_CHECKPOINT))
1644 seq_printf(seq, ",checkpoint=disable:%u",
1645 F2FS_OPTION(sbi).unusable_cap);
1646 if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
1647 seq_printf(seq, ",fsync_mode=%s", "posix");
1648 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
1649 seq_printf(seq, ",fsync_mode=%s", "strict");
1650 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
1651 seq_printf(seq, ",fsync_mode=%s", "nobarrier");
1653 #ifdef CONFIG_F2FS_FS_COMPRESSION
1654 f2fs_show_compress_options(seq, sbi->sb);
1657 if (test_opt(sbi, ATGC))
1658 seq_puts(seq, ",atgc");
1662 static void default_options(struct f2fs_sb_info *sbi)
1664 /* init some FS parameters */
1665 F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
1666 F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
1667 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1668 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
1669 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_NOBARRIER;
1670 F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
1671 F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
1672 F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
1673 F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
1674 F2FS_OPTION(sbi).compress_ext_cnt = 0;
1675 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
1677 sbi->sb->s_flags &= ~SB_INLINECRYPT;
1679 set_opt(sbi, INLINE_XATTR);
1680 set_opt(sbi, INLINE_DATA);
1681 set_opt(sbi, INLINE_DENTRY);
1682 set_opt(sbi, EXTENT_CACHE);
1683 set_opt(sbi, NOHEAP);
1685 clear_opt(sbi, DISABLE_CHECKPOINT);
1686 F2FS_OPTION(sbi).unusable_cap = 0;
1687 sbi->sb->s_flags |= SB_LAZYTIME;
1688 set_opt(sbi, FLUSH_MERGE);
1689 set_opt(sbi, DISCARD);
1690 if (f2fs_sb_has_blkzoned(sbi))
1691 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
1693 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
1695 #ifdef CONFIG_F2FS_FS_XATTR
1696 set_opt(sbi, XATTR_USER);
1698 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1699 set_opt(sbi, POSIX_ACL);
1702 f2fs_build_fault_attr(sbi, 0, 0);
1706 static int f2fs_enable_quotas(struct super_block *sb);
1709 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
1711 unsigned int s_flags = sbi->sb->s_flags;
1712 struct cp_control cpc;
1717 if (s_flags & SB_RDONLY) {
1718 f2fs_err(sbi, "checkpoint=disable on readonly fs");
1721 sbi->sb->s_flags |= SB_ACTIVE;
1723 f2fs_update_time(sbi, DISABLE_TIME);
1725 while (!f2fs_time_over(sbi, DISABLE_TIME)) {
1726 down_write(&sbi->gc_lock);
1727 err = f2fs_gc(sbi, true, false, NULL_SEGNO);
1728 if (err == -ENODATA) {
1732 if (err && err != -EAGAIN)
1736 ret = sync_filesystem(sbi->sb);
1738 err = ret ? ret: err;
1742 unusable = f2fs_get_unusable_blocks(sbi);
1743 if (f2fs_disable_cp_again(sbi, unusable)) {
1748 down_write(&sbi->gc_lock);
1749 cpc.reason = CP_PAUSE;
1750 set_sbi_flag(sbi, SBI_CP_DISABLED);
1751 err = f2fs_write_checkpoint(sbi, &cpc);
1755 spin_lock(&sbi->stat_lock);
1756 sbi->unusable_block_count = unusable;
1757 spin_unlock(&sbi->stat_lock);
1760 up_write(&sbi->gc_lock);
1762 sbi->sb->s_flags = s_flags; /* Restore SB_RDONLY status */
1766 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
1768 /* we should flush all the data to keep data consistency */
1769 sync_inodes_sb(sbi->sb);
1771 down_write(&sbi->gc_lock);
1772 f2fs_dirty_to_prefree(sbi);
1774 clear_sbi_flag(sbi, SBI_CP_DISABLED);
1775 set_sbi_flag(sbi, SBI_IS_DIRTY);
1776 up_write(&sbi->gc_lock);
1778 f2fs_sync_fs(sbi->sb, 1);
1781 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
1783 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1784 struct f2fs_mount_info org_mount_opt;
1785 unsigned long old_sb_flags;
1787 bool need_restart_gc = false;
1788 bool need_stop_gc = false;
1789 bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
1790 bool disable_checkpoint = test_opt(sbi, DISABLE_CHECKPOINT);
1791 bool no_io_align = !F2FS_IO_ALIGNED(sbi);
1792 bool no_atgc = !test_opt(sbi, ATGC);
1793 bool checkpoint_changed;
1799 * Save the old mount options in case we
1800 * need to restore them.
1802 org_mount_opt = sbi->mount_opt;
1803 old_sb_flags = sb->s_flags;
1806 org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
1807 for (i = 0; i < MAXQUOTAS; i++) {
1808 if (F2FS_OPTION(sbi).s_qf_names[i]) {
1809 org_mount_opt.s_qf_names[i] =
1810 kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
1812 if (!org_mount_opt.s_qf_names[i]) {
1813 for (j = 0; j < i; j++)
1814 kfree(org_mount_opt.s_qf_names[j]);
1818 org_mount_opt.s_qf_names[i] = NULL;
1823 /* recover superblocks we couldn't write due to previous RO mount */
1824 if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
1825 err = f2fs_commit_super(sbi, false);
1826 f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
1829 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
1832 default_options(sbi);
1834 /* parse mount options */
1835 err = parse_options(sb, data, true);
1838 checkpoint_changed =
1839 disable_checkpoint != test_opt(sbi, DISABLE_CHECKPOINT);
1842 * Previous and new state of filesystem is RO,
1843 * so skip checking GC and FLUSH_MERGE conditions.
1845 if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
1849 if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
1850 err = dquot_suspend(sb, -1);
1853 } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
1854 /* dquot_resume needs RW */
1855 sb->s_flags &= ~SB_RDONLY;
1856 if (sb_any_quota_suspended(sb)) {
1857 dquot_resume(sb, -1);
1858 } else if (f2fs_sb_has_quota_ino(sbi)) {
1859 err = f2fs_enable_quotas(sb);
1865 /* disallow enable atgc dynamically */
1866 if (no_atgc == !!test_opt(sbi, ATGC)) {
1868 f2fs_warn(sbi, "switch atgc option is not allowed");
1872 /* disallow enable/disable extent_cache dynamically */
1873 if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
1875 f2fs_warn(sbi, "switch extent_cache option is not allowed");
1879 if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
1881 f2fs_warn(sbi, "switch io_bits option is not allowed");
1885 if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
1887 f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
1892 * We stop the GC thread if FS is mounted as RO
1893 * or if background_gc = off is passed in mount
1894 * option. Also sync the filesystem.
1896 if ((*flags & SB_RDONLY) ||
1897 F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF) {
1898 if (sbi->gc_thread) {
1899 f2fs_stop_gc_thread(sbi);
1900 need_restart_gc = true;
1902 } else if (!sbi->gc_thread) {
1903 err = f2fs_start_gc_thread(sbi);
1906 need_stop_gc = true;
1909 if (*flags & SB_RDONLY ||
1910 F2FS_OPTION(sbi).whint_mode != org_mount_opt.whint_mode) {
1911 writeback_inodes_sb(sb, WB_REASON_SYNC);
1914 set_sbi_flag(sbi, SBI_IS_DIRTY);
1915 set_sbi_flag(sbi, SBI_IS_CLOSE);
1916 f2fs_sync_fs(sb, 1);
1917 clear_sbi_flag(sbi, SBI_IS_CLOSE);
1920 if (checkpoint_changed) {
1921 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
1922 err = f2fs_disable_checkpoint(sbi);
1926 f2fs_enable_checkpoint(sbi);
1931 * We stop issue flush thread if FS is mounted as RO
1932 * or if flush_merge is not passed in mount option.
1934 if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
1935 clear_opt(sbi, FLUSH_MERGE);
1936 f2fs_destroy_flush_cmd_control(sbi, false);
1938 err = f2fs_create_flush_cmd_control(sbi);
1944 /* Release old quota file names */
1945 for (i = 0; i < MAXQUOTAS; i++)
1946 kfree(org_mount_opt.s_qf_names[i]);
1948 /* Update the POSIXACL Flag */
1949 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
1950 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
1952 limit_reserve_root(sbi);
1953 adjust_unusable_cap_perc(sbi);
1954 *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
1957 if (need_restart_gc) {
1958 if (f2fs_start_gc_thread(sbi))
1959 f2fs_warn(sbi, "background gc thread has stopped");
1960 } else if (need_stop_gc) {
1961 f2fs_stop_gc_thread(sbi);
1965 F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
1966 for (i = 0; i < MAXQUOTAS; i++) {
1967 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1968 F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
1971 sbi->mount_opt = org_mount_opt;
1972 sb->s_flags = old_sb_flags;
1977 /* Read data from quotafile */
1978 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
1979 size_t len, loff_t off)
1981 struct inode *inode = sb_dqopt(sb)->files[type];
1982 struct address_space *mapping = inode->i_mapping;
1983 block_t blkidx = F2FS_BYTES_TO_BLK(off);
1984 int offset = off & (sb->s_blocksize - 1);
1987 loff_t i_size = i_size_read(inode);
1994 if (off + len > i_size)
1997 while (toread > 0) {
1998 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
2000 page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
2002 if (PTR_ERR(page) == -ENOMEM) {
2003 congestion_wait(BLK_RW_ASYNC,
2004 DEFAULT_IO_TIMEOUT);
2007 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2008 return PTR_ERR(page);
2013 if (unlikely(page->mapping != mapping)) {
2014 f2fs_put_page(page, 1);
2017 if (unlikely(!PageUptodate(page))) {
2018 f2fs_put_page(page, 1);
2019 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2023 kaddr = kmap_atomic(page);
2024 memcpy(data, kaddr + offset, tocopy);
2025 kunmap_atomic(kaddr);
2026 f2fs_put_page(page, 1);
2036 /* Write to quotafile */
2037 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2038 const char *data, size_t len, loff_t off)
2040 struct inode *inode = sb_dqopt(sb)->files[type];
2041 struct address_space *mapping = inode->i_mapping;
2042 const struct address_space_operations *a_ops = mapping->a_ops;
2043 int offset = off & (sb->s_blocksize - 1);
2044 size_t towrite = len;
2046 void *fsdata = NULL;
2051 while (towrite > 0) {
2052 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2055 err = a_ops->write_begin(NULL, mapping, off, tocopy, 0,
2057 if (unlikely(err)) {
2058 if (err == -ENOMEM) {
2059 congestion_wait(BLK_RW_ASYNC,
2060 DEFAULT_IO_TIMEOUT);
2063 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2067 kaddr = kmap_atomic(page);
2068 memcpy(kaddr + offset, data, tocopy);
2069 kunmap_atomic(kaddr);
2070 flush_dcache_page(page);
2072 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2083 inode->i_mtime = inode->i_ctime = current_time(inode);
2084 f2fs_mark_inode_dirty_sync(inode, false);
2085 return len - towrite;
2088 static struct dquot **f2fs_get_dquots(struct inode *inode)
2090 return F2FS_I(inode)->i_dquot;
2093 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2095 return &F2FS_I(inode)->i_reserved_quota;
2098 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2100 if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2101 f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2105 return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2106 F2FS_OPTION(sbi).s_jquota_fmt, type);
2109 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2114 if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2115 err = f2fs_enable_quotas(sbi->sb);
2117 f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2123 for (i = 0; i < MAXQUOTAS; i++) {
2124 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2125 err = f2fs_quota_on_mount(sbi, i);
2130 f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2137 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2140 struct inode *qf_inode;
2141 unsigned long qf_inum;
2144 BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2146 qf_inum = f2fs_qf_ino(sb, type);
2150 qf_inode = f2fs_iget(sb, qf_inum);
2151 if (IS_ERR(qf_inode)) {
2152 f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2153 return PTR_ERR(qf_inode);
2156 /* Don't account quota for quota files to avoid recursion */
2157 qf_inode->i_flags |= S_NOQUOTA;
2158 err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2163 static int f2fs_enable_quotas(struct super_block *sb)
2165 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2167 unsigned long qf_inum;
2168 bool quota_mopt[MAXQUOTAS] = {
2169 test_opt(sbi, USRQUOTA),
2170 test_opt(sbi, GRPQUOTA),
2171 test_opt(sbi, PRJQUOTA),
2174 if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2175 f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2179 sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2181 for (type = 0; type < MAXQUOTAS; type++) {
2182 qf_inum = f2fs_qf_ino(sb, type);
2184 err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2185 DQUOT_USAGE_ENABLED |
2186 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2188 f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2190 for (type--; type >= 0; type--)
2191 dquot_quota_off(sb, type);
2192 set_sbi_flag(F2FS_SB(sb),
2193 SBI_QUOTA_NEED_REPAIR);
2201 int f2fs_quota_sync(struct super_block *sb, int type)
2203 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2204 struct quota_info *dqopt = sb_dqopt(sb);
2211 * down_read(quota_sem)
2212 * dquot_writeback_dquots()
2215 * down_read(quota_sem)
2219 down_read(&sbi->quota_sem);
2220 ret = dquot_writeback_dquots(sb, type);
2225 * Now when everything is written we can discard the pagecache so
2226 * that userspace sees the changes.
2228 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2229 struct address_space *mapping;
2231 if (type != -1 && cnt != type)
2233 if (!sb_has_quota_active(sb, cnt))
2236 mapping = dqopt->files[cnt]->i_mapping;
2238 ret = filemap_fdatawrite(mapping);
2242 /* if we are using journalled quota */
2243 if (is_journalled_quota(sbi))
2246 ret = filemap_fdatawait(mapping);
2248 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2250 inode_lock(dqopt->files[cnt]);
2251 truncate_inode_pages(&dqopt->files[cnt]->i_data, 0);
2252 inode_unlock(dqopt->files[cnt]);
2256 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2257 up_read(&sbi->quota_sem);
2258 f2fs_unlock_op(sbi);
2262 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2263 const struct path *path)
2265 struct inode *inode;
2268 /* if quota sysfile exists, deny enabling quota with specific file */
2269 if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2270 f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2274 err = f2fs_quota_sync(sb, type);
2278 err = dquot_quota_on(sb, type, format_id, path);
2282 inode = d_inode(path->dentry);
2285 F2FS_I(inode)->i_flags |= F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL;
2286 f2fs_set_inode_flags(inode);
2287 inode_unlock(inode);
2288 f2fs_mark_inode_dirty_sync(inode, false);
2293 static int __f2fs_quota_off(struct super_block *sb, int type)
2295 struct inode *inode = sb_dqopt(sb)->files[type];
2298 if (!inode || !igrab(inode))
2299 return dquot_quota_off(sb, type);
2301 err = f2fs_quota_sync(sb, type);
2305 err = dquot_quota_off(sb, type);
2306 if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
2310 F2FS_I(inode)->i_flags &= ~(F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL);
2311 f2fs_set_inode_flags(inode);
2312 inode_unlock(inode);
2313 f2fs_mark_inode_dirty_sync(inode, false);
2319 static int f2fs_quota_off(struct super_block *sb, int type)
2321 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2324 err = __f2fs_quota_off(sb, type);
2327 * quotactl can shutdown journalled quota, result in inconsistence
2328 * between quota record and fs data by following updates, tag the
2329 * flag to let fsck be aware of it.
2331 if (is_journalled_quota(sbi))
2332 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2336 void f2fs_quota_off_umount(struct super_block *sb)
2341 for (type = 0; type < MAXQUOTAS; type++) {
2342 err = __f2fs_quota_off(sb, type);
2344 int ret = dquot_quota_off(sb, type);
2346 f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
2348 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2352 * In case of checkpoint=disable, we must flush quota blocks.
2353 * This can cause NULL exception for node_inode in end_io, since
2354 * put_super already dropped it.
2356 sync_filesystem(sb);
2359 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
2361 struct quota_info *dqopt = sb_dqopt(sb);
2364 for (type = 0; type < MAXQUOTAS; type++) {
2365 if (!dqopt->files[type])
2367 f2fs_inode_synced(dqopt->files[type]);
2371 static int f2fs_dquot_commit(struct dquot *dquot)
2373 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2376 down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
2377 ret = dquot_commit(dquot);
2379 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2380 up_read(&sbi->quota_sem);
2384 static int f2fs_dquot_acquire(struct dquot *dquot)
2386 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2389 down_read(&sbi->quota_sem);
2390 ret = dquot_acquire(dquot);
2392 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2393 up_read(&sbi->quota_sem);
2397 static int f2fs_dquot_release(struct dquot *dquot)
2399 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2400 int ret = dquot_release(dquot);
2403 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2407 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
2409 struct super_block *sb = dquot->dq_sb;
2410 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2411 int ret = dquot_mark_dquot_dirty(dquot);
2413 /* if we are using journalled quota */
2414 if (is_journalled_quota(sbi))
2415 set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
2420 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
2422 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2423 int ret = dquot_commit_info(sb, type);
2426 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2430 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
2432 *projid = F2FS_I(inode)->i_projid;
2436 static const struct dquot_operations f2fs_quota_operations = {
2437 .get_reserved_space = f2fs_get_reserved_space,
2438 .write_dquot = f2fs_dquot_commit,
2439 .acquire_dquot = f2fs_dquot_acquire,
2440 .release_dquot = f2fs_dquot_release,
2441 .mark_dirty = f2fs_dquot_mark_dquot_dirty,
2442 .write_info = f2fs_dquot_commit_info,
2443 .alloc_dquot = dquot_alloc,
2444 .destroy_dquot = dquot_destroy,
2445 .get_projid = f2fs_get_projid,
2446 .get_next_id = dquot_get_next_id,
2449 static const struct quotactl_ops f2fs_quotactl_ops = {
2450 .quota_on = f2fs_quota_on,
2451 .quota_off = f2fs_quota_off,
2452 .quota_sync = f2fs_quota_sync,
2453 .get_state = dquot_get_state,
2454 .set_info = dquot_set_dqinfo,
2455 .get_dqblk = dquot_get_dqblk,
2456 .set_dqblk = dquot_set_dqblk,
2457 .get_nextdqblk = dquot_get_next_dqblk,
2460 int f2fs_quota_sync(struct super_block *sb, int type)
2465 void f2fs_quota_off_umount(struct super_block *sb)
2470 static const struct super_operations f2fs_sops = {
2471 .alloc_inode = f2fs_alloc_inode,
2472 .free_inode = f2fs_free_inode,
2473 .drop_inode = f2fs_drop_inode,
2474 .write_inode = f2fs_write_inode,
2475 .dirty_inode = f2fs_dirty_inode,
2476 .show_options = f2fs_show_options,
2478 .quota_read = f2fs_quota_read,
2479 .quota_write = f2fs_quota_write,
2480 .get_dquots = f2fs_get_dquots,
2482 .evict_inode = f2fs_evict_inode,
2483 .put_super = f2fs_put_super,
2484 .sync_fs = f2fs_sync_fs,
2485 .freeze_fs = f2fs_freeze,
2486 .unfreeze_fs = f2fs_unfreeze,
2487 .statfs = f2fs_statfs,
2488 .remount_fs = f2fs_remount,
2491 #ifdef CONFIG_FS_ENCRYPTION
2492 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
2494 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2495 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2499 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
2502 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2505 * Encrypting the root directory is not allowed because fsck
2506 * expects lost+found directory to exist and remain unencrypted
2507 * if LOST_FOUND feature is enabled.
2510 if (f2fs_sb_has_lost_found(sbi) &&
2511 inode->i_ino == F2FS_ROOT_INO(sbi))
2514 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2515 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2516 ctx, len, fs_data, XATTR_CREATE);
2519 static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
2521 return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
2524 static bool f2fs_has_stable_inodes(struct super_block *sb)
2529 static void f2fs_get_ino_and_lblk_bits(struct super_block *sb,
2530 int *ino_bits_ret, int *lblk_bits_ret)
2532 *ino_bits_ret = 8 * sizeof(nid_t);
2533 *lblk_bits_ret = 8 * sizeof(block_t);
2536 static int f2fs_get_num_devices(struct super_block *sb)
2538 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2540 if (f2fs_is_multi_device(sbi))
2541 return sbi->s_ndevs;
2545 static void f2fs_get_devices(struct super_block *sb,
2546 struct request_queue **devs)
2548 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2551 for (i = 0; i < sbi->s_ndevs; i++)
2552 devs[i] = bdev_get_queue(FDEV(i).bdev);
2555 static const struct fscrypt_operations f2fs_cryptops = {
2556 .key_prefix = "f2fs:",
2557 .get_context = f2fs_get_context,
2558 .set_context = f2fs_set_context,
2559 .get_dummy_policy = f2fs_get_dummy_policy,
2560 .empty_dir = f2fs_empty_dir,
2561 .max_namelen = F2FS_NAME_LEN,
2562 .has_stable_inodes = f2fs_has_stable_inodes,
2563 .get_ino_and_lblk_bits = f2fs_get_ino_and_lblk_bits,
2564 .get_num_devices = f2fs_get_num_devices,
2565 .get_devices = f2fs_get_devices,
2569 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
2570 u64 ino, u32 generation)
2572 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2573 struct inode *inode;
2575 if (f2fs_check_nid_range(sbi, ino))
2576 return ERR_PTR(-ESTALE);
2579 * f2fs_iget isn't quite right if the inode is currently unallocated!
2580 * However f2fs_iget currently does appropriate checks to handle stale
2581 * inodes so everything is OK.
2583 inode = f2fs_iget(sb, ino);
2585 return ERR_CAST(inode);
2586 if (unlikely(generation && inode->i_generation != generation)) {
2587 /* we didn't find the right inode.. */
2589 return ERR_PTR(-ESTALE);
2594 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
2595 int fh_len, int fh_type)
2597 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
2598 f2fs_nfs_get_inode);
2601 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
2602 int fh_len, int fh_type)
2604 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
2605 f2fs_nfs_get_inode);
2608 static const struct export_operations f2fs_export_ops = {
2609 .fh_to_dentry = f2fs_fh_to_dentry,
2610 .fh_to_parent = f2fs_fh_to_parent,
2611 .get_parent = f2fs_get_parent,
2614 static loff_t max_file_blocks(void)
2617 loff_t leaf_count = DEF_ADDRS_PER_BLOCK;
2620 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
2621 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
2622 * space in inode.i_addr, it will be more safe to reassign
2626 /* two direct node blocks */
2627 result += (leaf_count * 2);
2629 /* two indirect node blocks */
2630 leaf_count *= NIDS_PER_BLOCK;
2631 result += (leaf_count * 2);
2633 /* one double indirect node block */
2634 leaf_count *= NIDS_PER_BLOCK;
2635 result += leaf_count;
2640 static int __f2fs_commit_super(struct buffer_head *bh,
2641 struct f2fs_super_block *super)
2645 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
2646 set_buffer_dirty(bh);
2649 /* it's rare case, we can do fua all the time */
2650 return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
2653 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
2654 struct buffer_head *bh)
2656 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2657 (bh->b_data + F2FS_SUPER_OFFSET);
2658 struct super_block *sb = sbi->sb;
2659 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
2660 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
2661 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
2662 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
2663 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
2664 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
2665 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
2666 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
2667 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
2668 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
2669 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2670 u32 segment_count = le32_to_cpu(raw_super->segment_count);
2671 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2672 u64 main_end_blkaddr = main_blkaddr +
2673 (segment_count_main << log_blocks_per_seg);
2674 u64 seg_end_blkaddr = segment0_blkaddr +
2675 (segment_count << log_blocks_per_seg);
2677 if (segment0_blkaddr != cp_blkaddr) {
2678 f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
2679 segment0_blkaddr, cp_blkaddr);
2683 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
2685 f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
2686 cp_blkaddr, sit_blkaddr,
2687 segment_count_ckpt << log_blocks_per_seg);
2691 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
2693 f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
2694 sit_blkaddr, nat_blkaddr,
2695 segment_count_sit << log_blocks_per_seg);
2699 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
2701 f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
2702 nat_blkaddr, ssa_blkaddr,
2703 segment_count_nat << log_blocks_per_seg);
2707 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
2709 f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
2710 ssa_blkaddr, main_blkaddr,
2711 segment_count_ssa << log_blocks_per_seg);
2715 if (main_end_blkaddr > seg_end_blkaddr) {
2716 f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
2717 main_blkaddr, seg_end_blkaddr,
2718 segment_count_main << log_blocks_per_seg);
2720 } else if (main_end_blkaddr < seg_end_blkaddr) {
2724 /* fix in-memory information all the time */
2725 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
2726 segment0_blkaddr) >> log_blocks_per_seg);
2728 if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
2729 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2732 err = __f2fs_commit_super(bh, NULL);
2733 res = err ? "failed" : "done";
2735 f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
2736 res, main_blkaddr, seg_end_blkaddr,
2737 segment_count_main << log_blocks_per_seg);
2744 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
2745 struct buffer_head *bh)
2747 block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
2748 block_t total_sections, blocks_per_seg;
2749 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2750 (bh->b_data + F2FS_SUPER_OFFSET);
2751 size_t crc_offset = 0;
2754 if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
2755 f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
2756 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
2760 /* Check checksum_offset and crc in superblock */
2761 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
2762 crc_offset = le32_to_cpu(raw_super->checksum_offset);
2764 offsetof(struct f2fs_super_block, crc)) {
2765 f2fs_info(sbi, "Invalid SB checksum offset: %zu",
2767 return -EFSCORRUPTED;
2769 crc = le32_to_cpu(raw_super->crc);
2770 if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
2771 f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
2772 return -EFSCORRUPTED;
2776 /* Currently, support only 4KB page cache size */
2777 if (F2FS_BLKSIZE != PAGE_SIZE) {
2778 f2fs_info(sbi, "Invalid page_cache_size (%lu), supports only 4KB",
2780 return -EFSCORRUPTED;
2783 /* Currently, support only 4KB block size */
2784 if (le32_to_cpu(raw_super->log_blocksize) != F2FS_BLKSIZE_BITS) {
2785 f2fs_info(sbi, "Invalid log_blocksize (%u), supports only %u",
2786 le32_to_cpu(raw_super->log_blocksize),
2788 return -EFSCORRUPTED;
2791 /* check log blocks per segment */
2792 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
2793 f2fs_info(sbi, "Invalid log blocks per segment (%u)",
2794 le32_to_cpu(raw_super->log_blocks_per_seg));
2795 return -EFSCORRUPTED;
2798 /* Currently, support 512/1024/2048/4096 bytes sector size */
2799 if (le32_to_cpu(raw_super->log_sectorsize) >
2800 F2FS_MAX_LOG_SECTOR_SIZE ||
2801 le32_to_cpu(raw_super->log_sectorsize) <
2802 F2FS_MIN_LOG_SECTOR_SIZE) {
2803 f2fs_info(sbi, "Invalid log sectorsize (%u)",
2804 le32_to_cpu(raw_super->log_sectorsize));
2805 return -EFSCORRUPTED;
2807 if (le32_to_cpu(raw_super->log_sectors_per_block) +
2808 le32_to_cpu(raw_super->log_sectorsize) !=
2809 F2FS_MAX_LOG_SECTOR_SIZE) {
2810 f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
2811 le32_to_cpu(raw_super->log_sectors_per_block),
2812 le32_to_cpu(raw_super->log_sectorsize));
2813 return -EFSCORRUPTED;
2816 segment_count = le32_to_cpu(raw_super->segment_count);
2817 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2818 segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
2819 secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
2820 total_sections = le32_to_cpu(raw_super->section_count);
2822 /* blocks_per_seg should be 512, given the above check */
2823 blocks_per_seg = 1 << le32_to_cpu(raw_super->log_blocks_per_seg);
2825 if (segment_count > F2FS_MAX_SEGMENT ||
2826 segment_count < F2FS_MIN_SEGMENTS) {
2827 f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
2828 return -EFSCORRUPTED;
2831 if (total_sections > segment_count_main || total_sections < 1 ||
2832 segs_per_sec > segment_count || !segs_per_sec) {
2833 f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
2834 segment_count, total_sections, segs_per_sec);
2835 return -EFSCORRUPTED;
2838 if (segment_count_main != total_sections * segs_per_sec) {
2839 f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
2840 segment_count_main, total_sections, segs_per_sec);
2841 return -EFSCORRUPTED;
2844 if ((segment_count / segs_per_sec) < total_sections) {
2845 f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
2846 segment_count, segs_per_sec, total_sections);
2847 return -EFSCORRUPTED;
2850 if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
2851 f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
2852 segment_count, le64_to_cpu(raw_super->block_count));
2853 return -EFSCORRUPTED;
2856 if (RDEV(0).path[0]) {
2857 block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
2860 while (i < MAX_DEVICES && RDEV(i).path[0]) {
2861 dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
2864 if (segment_count != dev_seg_count) {
2865 f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
2866 segment_count, dev_seg_count);
2867 return -EFSCORRUPTED;
2870 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
2871 !bdev_is_zoned(sbi->sb->s_bdev)) {
2872 f2fs_info(sbi, "Zoned block device path is missing");
2873 return -EFSCORRUPTED;
2877 if (secs_per_zone > total_sections || !secs_per_zone) {
2878 f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
2879 secs_per_zone, total_sections);
2880 return -EFSCORRUPTED;
2882 if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
2883 raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
2884 (le32_to_cpu(raw_super->extension_count) +
2885 raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
2886 f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
2887 le32_to_cpu(raw_super->extension_count),
2888 raw_super->hot_ext_count,
2889 F2FS_MAX_EXTENSION);
2890 return -EFSCORRUPTED;
2893 if (le32_to_cpu(raw_super->cp_payload) >
2894 (blocks_per_seg - F2FS_CP_PACKS)) {
2895 f2fs_info(sbi, "Insane cp_payload (%u > %u)",
2896 le32_to_cpu(raw_super->cp_payload),
2897 blocks_per_seg - F2FS_CP_PACKS);
2898 return -EFSCORRUPTED;
2901 /* check reserved ino info */
2902 if (le32_to_cpu(raw_super->node_ino) != 1 ||
2903 le32_to_cpu(raw_super->meta_ino) != 2 ||
2904 le32_to_cpu(raw_super->root_ino) != 3) {
2905 f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
2906 le32_to_cpu(raw_super->node_ino),
2907 le32_to_cpu(raw_super->meta_ino),
2908 le32_to_cpu(raw_super->root_ino));
2909 return -EFSCORRUPTED;
2912 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
2913 if (sanity_check_area_boundary(sbi, bh))
2914 return -EFSCORRUPTED;
2919 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
2921 unsigned int total, fsmeta;
2922 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
2923 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2924 unsigned int ovp_segments, reserved_segments;
2925 unsigned int main_segs, blocks_per_seg;
2926 unsigned int sit_segs, nat_segs;
2927 unsigned int sit_bitmap_size, nat_bitmap_size;
2928 unsigned int log_blocks_per_seg;
2929 unsigned int segment_count_main;
2930 unsigned int cp_pack_start_sum, cp_payload;
2931 block_t user_block_count, valid_user_blocks;
2932 block_t avail_node_count, valid_node_count;
2935 total = le32_to_cpu(raw_super->segment_count);
2936 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
2937 sit_segs = le32_to_cpu(raw_super->segment_count_sit);
2939 nat_segs = le32_to_cpu(raw_super->segment_count_nat);
2941 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
2942 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
2944 if (unlikely(fsmeta >= total))
2947 ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
2948 reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
2950 if (unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
2951 ovp_segments == 0 || reserved_segments == 0)) {
2952 f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
2956 user_block_count = le64_to_cpu(ckpt->user_block_count);
2957 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2958 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2959 if (!user_block_count || user_block_count >=
2960 segment_count_main << log_blocks_per_seg) {
2961 f2fs_err(sbi, "Wrong user_block_count: %u",
2966 valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
2967 if (valid_user_blocks > user_block_count) {
2968 f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
2969 valid_user_blocks, user_block_count);
2973 valid_node_count = le32_to_cpu(ckpt->valid_node_count);
2974 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
2975 if (valid_node_count > avail_node_count) {
2976 f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
2977 valid_node_count, avail_node_count);
2981 main_segs = le32_to_cpu(raw_super->segment_count_main);
2982 blocks_per_seg = sbi->blocks_per_seg;
2984 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
2985 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
2986 le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
2988 for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
2989 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
2990 le32_to_cpu(ckpt->cur_node_segno[j])) {
2991 f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
2993 le32_to_cpu(ckpt->cur_node_segno[i]));
2998 for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
2999 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
3000 le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
3002 for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3003 if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3004 le32_to_cpu(ckpt->cur_data_segno[j])) {
3005 f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3007 le32_to_cpu(ckpt->cur_data_segno[i]));
3012 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3013 for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3014 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3015 le32_to_cpu(ckpt->cur_data_segno[j])) {
3016 f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3018 le32_to_cpu(ckpt->cur_node_segno[i]));
3024 sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3025 nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3027 if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3028 nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3029 f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3030 sit_bitmap_size, nat_bitmap_size);
3034 cp_pack_start_sum = __start_sum_addr(sbi);
3035 cp_payload = __cp_payload(sbi);
3036 if (cp_pack_start_sum < cp_payload + 1 ||
3037 cp_pack_start_sum > blocks_per_seg - 1 -
3038 NR_CURSEG_PERSIST_TYPE) {
3039 f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3044 if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3045 le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3046 f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3047 "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3048 "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3049 le32_to_cpu(ckpt->checksum_offset));
3053 if (unlikely(f2fs_cp_error(sbi))) {
3054 f2fs_err(sbi, "A bug case: need to run fsck");
3060 static void init_sb_info(struct f2fs_sb_info *sbi)
3062 struct f2fs_super_block *raw_super = sbi->raw_super;
3065 sbi->log_sectors_per_block =
3066 le32_to_cpu(raw_super->log_sectors_per_block);
3067 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3068 sbi->blocksize = 1 << sbi->log_blocksize;
3069 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3070 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
3071 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3072 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3073 sbi->total_sections = le32_to_cpu(raw_super->section_count);
3074 sbi->total_node_count =
3075 (le32_to_cpu(raw_super->segment_count_nat) / 2)
3076 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
3077 sbi->root_ino_num = le32_to_cpu(raw_super->root_ino);
3078 sbi->node_ino_num = le32_to_cpu(raw_super->node_ino);
3079 sbi->meta_ino_num = le32_to_cpu(raw_super->meta_ino);
3080 sbi->cur_victim_sec = NULL_SECNO;
3081 sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3082 sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3083 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3084 sbi->migration_granularity = sbi->segs_per_sec;
3086 sbi->dir_level = DEF_DIR_LEVEL;
3087 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3088 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3089 sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3090 sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3091 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3092 sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3093 DEF_UMOUNT_DISCARD_TIMEOUT;
3094 clear_sbi_flag(sbi, SBI_NEED_FSCK);
3096 for (i = 0; i < NR_COUNT_TYPE; i++)
3097 atomic_set(&sbi->nr_pages[i], 0);
3099 for (i = 0; i < META; i++)
3100 atomic_set(&sbi->wb_sync_req[i], 0);
3102 INIT_LIST_HEAD(&sbi->s_list);
3103 mutex_init(&sbi->umount_mutex);
3104 init_rwsem(&sbi->io_order_lock);
3105 spin_lock_init(&sbi->cp_lock);
3107 sbi->dirty_device = 0;
3108 spin_lock_init(&sbi->dev_lock);
3110 init_rwsem(&sbi->sb_lock);
3111 init_rwsem(&sbi->pin_sem);
3114 static int init_percpu_info(struct f2fs_sb_info *sbi)
3118 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3122 err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3125 percpu_counter_destroy(&sbi->alloc_valid_block_count);
3130 #ifdef CONFIG_BLK_DEV_ZONED
3132 struct f2fs_report_zones_args {
3133 struct f2fs_dev_info *dev;
3134 bool zone_cap_mismatch;
3137 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3140 struct f2fs_report_zones_args *rz_args = data;
3142 if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3145 set_bit(idx, rz_args->dev->blkz_seq);
3146 rz_args->dev->zone_capacity_blocks[idx] = zone->capacity >>
3147 F2FS_LOG_SECTORS_PER_BLOCK;
3148 if (zone->len != zone->capacity && !rz_args->zone_cap_mismatch)
3149 rz_args->zone_cap_mismatch = true;
3154 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3156 struct block_device *bdev = FDEV(devi).bdev;
3157 sector_t nr_sectors = bdev->bd_part->nr_sects;
3158 struct f2fs_report_zones_args rep_zone_arg;
3161 if (!f2fs_sb_has_blkzoned(sbi))
3164 if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3165 SECTOR_TO_BLOCK(bdev_zone_sectors(bdev)))
3167 sbi->blocks_per_blkz = SECTOR_TO_BLOCK(bdev_zone_sectors(bdev));
3168 if (sbi->log_blocks_per_blkz && sbi->log_blocks_per_blkz !=
3169 __ilog2_u32(sbi->blocks_per_blkz))
3171 sbi->log_blocks_per_blkz = __ilog2_u32(sbi->blocks_per_blkz);
3172 FDEV(devi).nr_blkz = SECTOR_TO_BLOCK(nr_sectors) >>
3173 sbi->log_blocks_per_blkz;
3174 if (nr_sectors & (bdev_zone_sectors(bdev) - 1))
3175 FDEV(devi).nr_blkz++;
3177 FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3178 BITS_TO_LONGS(FDEV(devi).nr_blkz)
3179 * sizeof(unsigned long),
3181 if (!FDEV(devi).blkz_seq)
3184 /* Get block zones type and zone-capacity */
3185 FDEV(devi).zone_capacity_blocks = f2fs_kzalloc(sbi,
3186 FDEV(devi).nr_blkz * sizeof(block_t),
3188 if (!FDEV(devi).zone_capacity_blocks)
3191 rep_zone_arg.dev = &FDEV(devi);
3192 rep_zone_arg.zone_cap_mismatch = false;
3194 ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3199 if (!rep_zone_arg.zone_cap_mismatch) {
3200 kfree(FDEV(devi).zone_capacity_blocks);
3201 FDEV(devi).zone_capacity_blocks = NULL;
3209 * Read f2fs raw super block.
3210 * Because we have two copies of super block, so read both of them
3211 * to get the first valid one. If any one of them is broken, we pass
3212 * them recovery flag back to the caller.
3214 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3215 struct f2fs_super_block **raw_super,
3216 int *valid_super_block, int *recovery)
3218 struct super_block *sb = sbi->sb;
3220 struct buffer_head *bh;
3221 struct f2fs_super_block *super;
3224 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3228 for (block = 0; block < 2; block++) {
3229 bh = sb_bread(sb, block);
3231 f2fs_err(sbi, "Unable to read %dth superblock",
3238 /* sanity checking of raw super */
3239 err = sanity_check_raw_super(sbi, bh);
3241 f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3249 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3251 *valid_super_block = block;
3257 /* No valid superblock */
3266 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
3268 struct buffer_head *bh;
3272 if ((recover && f2fs_readonly(sbi->sb)) ||
3273 bdev_read_only(sbi->sb->s_bdev)) {
3274 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3278 /* we should update superblock crc here */
3279 if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
3280 crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
3281 offsetof(struct f2fs_super_block, crc));
3282 F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
3285 /* write back-up superblock first */
3286 bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
3289 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3292 /* if we are in recovery path, skip writing valid superblock */
3296 /* write current valid superblock */
3297 bh = sb_bread(sbi->sb, sbi->valid_super_block);
3300 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3305 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
3307 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3308 unsigned int max_devices = MAX_DEVICES;
3311 /* Initialize single device information */
3312 if (!RDEV(0).path[0]) {
3313 if (!bdev_is_zoned(sbi->sb->s_bdev))
3319 * Initialize multiple devices information, or single
3320 * zoned block device information.
3322 sbi->devs = f2fs_kzalloc(sbi,
3323 array_size(max_devices,
3324 sizeof(struct f2fs_dev_info)),
3329 for (i = 0; i < max_devices; i++) {
3331 if (i > 0 && !RDEV(i).path[0])
3334 if (max_devices == 1) {
3335 /* Single zoned block device mount */
3337 blkdev_get_by_dev(sbi->sb->s_bdev->bd_dev,
3338 sbi->sb->s_mode, sbi->sb->s_type);
3340 /* Multi-device mount */
3341 memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
3342 FDEV(i).total_segments =
3343 le32_to_cpu(RDEV(i).total_segments);
3345 FDEV(i).start_blk = 0;
3346 FDEV(i).end_blk = FDEV(i).start_blk +
3347 (FDEV(i).total_segments <<
3348 sbi->log_blocks_per_seg) - 1 +
3349 le32_to_cpu(raw_super->segment0_blkaddr);
3351 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
3352 FDEV(i).end_blk = FDEV(i).start_blk +
3353 (FDEV(i).total_segments <<
3354 sbi->log_blocks_per_seg) - 1;
3356 FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
3357 sbi->sb->s_mode, sbi->sb->s_type);
3359 if (IS_ERR(FDEV(i).bdev))
3360 return PTR_ERR(FDEV(i).bdev);
3362 /* to release errored devices */
3363 sbi->s_ndevs = i + 1;
3365 #ifdef CONFIG_BLK_DEV_ZONED
3366 if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
3367 !f2fs_sb_has_blkzoned(sbi)) {
3368 f2fs_err(sbi, "Zoned block device feature not enabled\n");
3371 if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
3372 if (init_blkz_info(sbi, i)) {
3373 f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
3376 if (max_devices == 1)
3378 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
3380 FDEV(i).total_segments,
3381 FDEV(i).start_blk, FDEV(i).end_blk,
3382 bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
3383 "Host-aware" : "Host-managed");
3387 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
3389 FDEV(i).total_segments,
3390 FDEV(i).start_blk, FDEV(i).end_blk);
3393 "IO Block Size: %8d KB", F2FS_IO_SIZE_KB(sbi));
3397 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
3399 #ifdef CONFIG_UNICODE
3400 if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
3401 const struct f2fs_sb_encodings *encoding_info;
3402 struct unicode_map *encoding;
3403 __u16 encoding_flags;
3405 if (f2fs_sb_has_encrypt(sbi)) {
3407 "Can't mount with encoding and encryption");
3411 if (f2fs_sb_read_encoding(sbi->raw_super, &encoding_info,
3414 "Encoding requested by superblock is unknown");
3418 encoding = utf8_load(encoding_info->version);
3419 if (IS_ERR(encoding)) {
3421 "can't mount with superblock charset: %s-%s "
3422 "not supported by the kernel. flags: 0x%x.",
3423 encoding_info->name, encoding_info->version,
3425 return PTR_ERR(encoding);
3427 f2fs_info(sbi, "Using encoding defined by superblock: "
3428 "%s-%s with flags 0x%hx", encoding_info->name,
3429 encoding_info->version?:"\b", encoding_flags);
3431 sbi->sb->s_encoding = encoding;
3432 sbi->sb->s_encoding_flags = encoding_flags;
3433 sbi->sb->s_d_op = &f2fs_dentry_ops;
3436 if (f2fs_sb_has_casefold(sbi)) {
3437 f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
3444 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
3446 struct f2fs_sm_info *sm_i = SM_I(sbi);
3448 /* adjust parameters according to the volume size */
3449 if (sm_i->main_segments <= SMALL_VOLUME_SEGMENTS) {
3450 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
3451 sm_i->dcc_info->discard_granularity = 1;
3452 sm_i->ipu_policy = 1 << F2FS_IPU_FORCE;
3455 sbi->readdir_ra = 1;
3458 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
3460 struct f2fs_sb_info *sbi;
3461 struct f2fs_super_block *raw_super;
3464 bool skip_recovery = false, need_fsck = false;
3465 char *options = NULL;
3466 int recovery, i, valid_super_block;
3467 struct curseg_info *seg_i;
3473 valid_super_block = -1;
3476 /* allocate memory for f2fs-specific super block info */
3477 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
3483 /* Load the checksum driver */
3484 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
3485 if (IS_ERR(sbi->s_chksum_driver)) {
3486 f2fs_err(sbi, "Cannot load crc32 driver.");
3487 err = PTR_ERR(sbi->s_chksum_driver);
3488 sbi->s_chksum_driver = NULL;
3492 /* set a block size */
3493 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
3494 f2fs_err(sbi, "unable to set blocksize");
3498 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
3503 sb->s_fs_info = sbi;
3504 sbi->raw_super = raw_super;
3506 /* precompute checksum seed for metadata */
3507 if (f2fs_sb_has_inode_chksum(sbi))
3508 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
3509 sizeof(raw_super->uuid));
3511 default_options(sbi);
3512 /* parse mount options */
3513 options = kstrdup((const char *)data, GFP_KERNEL);
3514 if (data && !options) {
3519 err = parse_options(sb, options, false);
3523 sbi->max_file_blocks = max_file_blocks();
3524 sb->s_maxbytes = sbi->max_file_blocks <<
3525 le32_to_cpu(raw_super->log_blocksize);
3526 sb->s_max_links = F2FS_LINK_MAX;
3528 err = f2fs_setup_casefold(sbi);
3533 sb->dq_op = &f2fs_quota_operations;
3534 sb->s_qcop = &f2fs_quotactl_ops;
3535 sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
3537 if (f2fs_sb_has_quota_ino(sbi)) {
3538 for (i = 0; i < MAXQUOTAS; i++) {
3539 if (f2fs_qf_ino(sbi->sb, i))
3540 sbi->nquota_files++;
3545 sb->s_op = &f2fs_sops;
3546 #ifdef CONFIG_FS_ENCRYPTION
3547 sb->s_cop = &f2fs_cryptops;
3549 #ifdef CONFIG_FS_VERITY
3550 sb->s_vop = &f2fs_verityops;
3552 sb->s_xattr = f2fs_xattr_handlers;
3553 sb->s_export_op = &f2fs_export_ops;
3554 sb->s_magic = F2FS_SUPER_MAGIC;
3555 sb->s_time_gran = 1;
3556 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
3557 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
3558 memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
3559 sb->s_iflags |= SB_I_CGROUPWB;
3561 /* init f2fs-specific super block info */
3562 sbi->valid_super_block = valid_super_block;
3563 init_rwsem(&sbi->gc_lock);
3564 mutex_init(&sbi->writepages);
3565 mutex_init(&sbi->cp_mutex);
3566 init_rwsem(&sbi->node_write);
3567 init_rwsem(&sbi->node_change);
3569 /* disallow all the data/node/meta page writes */
3570 set_sbi_flag(sbi, SBI_POR_DOING);
3571 spin_lock_init(&sbi->stat_lock);
3573 /* init iostat info */
3574 spin_lock_init(&sbi->iostat_lock);
3575 sbi->iostat_enable = false;
3576 sbi->iostat_period_ms = DEFAULT_IOSTAT_PERIOD_MS;
3578 for (i = 0; i < NR_PAGE_TYPE; i++) {
3579 int n = (i == META) ? 1: NR_TEMP_TYPE;
3585 sizeof(struct f2fs_bio_info)),
3587 if (!sbi->write_io[i]) {
3592 for (j = HOT; j < n; j++) {
3593 init_rwsem(&sbi->write_io[i][j].io_rwsem);
3594 sbi->write_io[i][j].sbi = sbi;
3595 sbi->write_io[i][j].bio = NULL;
3596 spin_lock_init(&sbi->write_io[i][j].io_lock);
3597 INIT_LIST_HEAD(&sbi->write_io[i][j].io_list);
3598 INIT_LIST_HEAD(&sbi->write_io[i][j].bio_list);
3599 init_rwsem(&sbi->write_io[i][j].bio_list_lock);
3603 init_rwsem(&sbi->cp_rwsem);
3604 init_rwsem(&sbi->quota_sem);
3605 init_waitqueue_head(&sbi->cp_wait);
3608 err = init_percpu_info(sbi);
3612 if (F2FS_IO_ALIGNED(sbi)) {
3613 sbi->write_io_dummy =
3614 mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
3615 if (!sbi->write_io_dummy) {
3621 /* init per sbi slab cache */
3622 err = f2fs_init_xattr_caches(sbi);
3625 err = f2fs_init_page_array_cache(sbi);
3627 goto free_xattr_cache;
3629 /* get an inode for meta space */
3630 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
3631 if (IS_ERR(sbi->meta_inode)) {
3632 f2fs_err(sbi, "Failed to read F2FS meta data inode");
3633 err = PTR_ERR(sbi->meta_inode);
3634 goto free_page_array_cache;
3637 err = f2fs_get_valid_checkpoint(sbi);
3639 f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
3640 goto free_meta_inode;
3643 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
3644 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3645 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
3646 set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
3647 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
3650 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
3651 set_sbi_flag(sbi, SBI_NEED_FSCK);
3653 /* Initialize device list */
3654 err = f2fs_scan_devices(sbi);
3656 f2fs_err(sbi, "Failed to find devices");
3660 err = f2fs_init_post_read_wq(sbi);
3662 f2fs_err(sbi, "Failed to initialize post read workqueue");
3666 sbi->total_valid_node_count =
3667 le32_to_cpu(sbi->ckpt->valid_node_count);
3668 percpu_counter_set(&sbi->total_valid_inode_count,
3669 le32_to_cpu(sbi->ckpt->valid_inode_count));
3670 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
3671 sbi->total_valid_block_count =
3672 le64_to_cpu(sbi->ckpt->valid_block_count);
3673 sbi->last_valid_block_count = sbi->total_valid_block_count;
3674 sbi->reserved_blocks = 0;
3675 sbi->current_reserved_blocks = 0;
3676 limit_reserve_root(sbi);
3677 adjust_unusable_cap_perc(sbi);
3679 for (i = 0; i < NR_INODE_TYPE; i++) {
3680 INIT_LIST_HEAD(&sbi->inode_list[i]);
3681 spin_lock_init(&sbi->inode_lock[i]);
3683 mutex_init(&sbi->flush_lock);
3685 f2fs_init_extent_cache_info(sbi);
3687 f2fs_init_ino_entry_info(sbi);
3689 f2fs_init_fsync_node_info(sbi);
3691 /* setup f2fs internal modules */
3692 err = f2fs_build_segment_manager(sbi);
3694 f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
3698 err = f2fs_build_node_manager(sbi);
3700 f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
3705 /* For write statistics */
3706 if (sb->s_bdev->bd_part)
3707 sbi->sectors_written_start =
3708 (u64)part_stat_read(sb->s_bdev->bd_part,
3709 sectors[STAT_WRITE]);
3711 /* Read accumulated write IO statistics if exists */
3712 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
3713 if (__exist_node_summaries(sbi))
3714 sbi->kbytes_written =
3715 le64_to_cpu(seg_i->journal->info.kbytes_written);
3717 f2fs_build_gc_manager(sbi);
3719 err = f2fs_build_stats(sbi);
3723 /* get an inode for node space */
3724 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
3725 if (IS_ERR(sbi->node_inode)) {
3726 f2fs_err(sbi, "Failed to read node inode");
3727 err = PTR_ERR(sbi->node_inode);
3731 /* read root inode and dentry */
3732 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
3734 f2fs_err(sbi, "Failed to read root inode");
3735 err = PTR_ERR(root);
3736 goto free_node_inode;
3738 if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
3739 !root->i_size || !root->i_nlink) {
3742 goto free_node_inode;
3745 sb->s_root = d_make_root(root); /* allocate root dentry */
3748 goto free_node_inode;
3751 err = f2fs_register_sysfs(sbi);
3753 goto free_root_inode;
3756 /* Enable quota usage during mount */
3757 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
3758 err = f2fs_enable_quotas(sb);
3760 f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
3763 /* if there are any orphan inodes, free them */
3764 err = f2fs_recover_orphan_inodes(sbi);
3768 if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
3769 goto reset_checkpoint;
3771 /* recover fsynced data */
3772 if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
3773 !test_opt(sbi, NORECOVERY)) {
3775 * mount should be failed, when device has readonly mode, and
3776 * previous checkpoint was not done by clean system shutdown.
3778 if (f2fs_hw_is_readonly(sbi)) {
3779 if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
3781 f2fs_err(sbi, "Need to recover fsync data, but write access unavailable");
3784 f2fs_info(sbi, "write access unavailable, skipping recovery");
3785 goto reset_checkpoint;
3789 set_sbi_flag(sbi, SBI_NEED_FSCK);
3792 goto reset_checkpoint;
3794 err = f2fs_recover_fsync_data(sbi, false);
3797 skip_recovery = true;
3799 f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
3804 err = f2fs_recover_fsync_data(sbi, true);
3806 if (!f2fs_readonly(sb) && err > 0) {
3808 f2fs_err(sbi, "Need to recover fsync data");
3814 * If the f2fs is not readonly and fsync data recovery succeeds,
3815 * check zoned block devices' write pointer consistency.
3817 if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
3818 err = f2fs_check_write_pointer(sbi);
3824 f2fs_init_inmem_curseg(sbi);
3826 /* f2fs_recover_fsync_data() cleared this already */
3827 clear_sbi_flag(sbi, SBI_POR_DOING);
3829 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
3830 err = f2fs_disable_checkpoint(sbi);
3832 goto sync_free_meta;
3833 } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
3834 f2fs_enable_checkpoint(sbi);
3838 * If filesystem is not mounted as read-only then
3839 * do start the gc_thread.
3841 if (F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF && !f2fs_readonly(sb)) {
3842 /* After POR, we can run background GC thread.*/
3843 err = f2fs_start_gc_thread(sbi);
3845 goto sync_free_meta;
3849 /* recover broken superblock */
3851 err = f2fs_commit_super(sbi, true);
3852 f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
3853 sbi->valid_super_block ? 1 : 2, err);
3856 f2fs_join_shrinker(sbi);
3858 f2fs_tuning_parameters(sbi);
3860 f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
3861 cur_cp_version(F2FS_CKPT(sbi)));
3862 f2fs_update_time(sbi, CP_TIME);
3863 f2fs_update_time(sbi, REQ_TIME);
3864 clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
3868 /* safe to flush all the data */
3869 sync_filesystem(sbi->sb);
3874 f2fs_truncate_quota_inode_pages(sb);
3875 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
3876 f2fs_quota_off_umount(sbi->sb);
3879 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
3880 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
3881 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
3882 * falls into an infinite loop in f2fs_sync_meta_pages().
3884 truncate_inode_pages_final(META_MAPPING(sbi));
3885 /* evict some inodes being cached by GC */
3887 f2fs_unregister_sysfs(sbi);
3892 f2fs_release_ino_entry(sbi, true);
3893 truncate_inode_pages_final(NODE_MAPPING(sbi));
3894 iput(sbi->node_inode);
3895 sbi->node_inode = NULL;
3897 f2fs_destroy_stats(sbi);
3899 f2fs_destroy_node_manager(sbi);
3901 f2fs_destroy_segment_manager(sbi);
3902 f2fs_destroy_post_read_wq(sbi);
3904 destroy_device_list(sbi);
3907 make_bad_inode(sbi->meta_inode);
3908 iput(sbi->meta_inode);
3909 sbi->meta_inode = NULL;
3910 free_page_array_cache:
3911 f2fs_destroy_page_array_cache(sbi);
3913 f2fs_destroy_xattr_caches(sbi);
3915 mempool_destroy(sbi->write_io_dummy);
3917 destroy_percpu_info(sbi);
3919 for (i = 0; i < NR_PAGE_TYPE; i++)
3920 kvfree(sbi->write_io[i]);
3922 #ifdef CONFIG_UNICODE
3923 utf8_unload(sb->s_encoding);
3924 sb->s_encoding = NULL;
3928 for (i = 0; i < MAXQUOTAS; i++)
3929 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
3931 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
3936 if (sbi->s_chksum_driver)
3937 crypto_free_shash(sbi->s_chksum_driver);
3940 /* give only one another chance */
3941 if (retry_cnt > 0 && skip_recovery) {
3943 shrink_dcache_sb(sb);
3949 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
3950 const char *dev_name, void *data)
3952 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
3955 static void kill_f2fs_super(struct super_block *sb)
3958 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3960 set_sbi_flag(sbi, SBI_IS_CLOSE);
3961 f2fs_stop_gc_thread(sbi);
3962 f2fs_stop_discard_thread(sbi);
3964 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
3965 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
3966 struct cp_control cpc = {
3967 .reason = CP_UMOUNT,
3969 f2fs_write_checkpoint(sbi, &cpc);
3972 if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
3973 sb->s_flags &= ~SB_RDONLY;
3975 kill_block_super(sb);
3978 static struct file_system_type f2fs_fs_type = {
3979 .owner = THIS_MODULE,
3981 .mount = f2fs_mount,
3982 .kill_sb = kill_f2fs_super,
3983 .fs_flags = FS_REQUIRES_DEV,
3985 MODULE_ALIAS_FS("f2fs");
3987 static int __init init_inodecache(void)
3989 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
3990 sizeof(struct f2fs_inode_info), 0,
3991 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
3992 if (!f2fs_inode_cachep)
3997 static void destroy_inodecache(void)
4000 * Make sure all delayed rcu free inodes are flushed before we
4004 kmem_cache_destroy(f2fs_inode_cachep);
4007 static int __init init_f2fs_fs(void)
4011 if (PAGE_SIZE != F2FS_BLKSIZE) {
4012 printk("F2FS not supported on PAGE_SIZE(%lu) != %d\n",
4013 PAGE_SIZE, F2FS_BLKSIZE);
4017 f2fs_build_trace_ios();
4019 err = init_inodecache();
4022 err = f2fs_create_node_manager_caches();
4024 goto free_inodecache;
4025 err = f2fs_create_segment_manager_caches();
4027 goto free_node_manager_caches;
4028 err = f2fs_create_checkpoint_caches();
4030 goto free_segment_manager_caches;
4031 err = f2fs_create_extent_cache();
4033 goto free_checkpoint_caches;
4034 err = f2fs_create_garbage_collection_cache();
4036 goto free_extent_cache;
4037 err = f2fs_init_sysfs();
4039 goto free_garbage_collection_cache;
4040 err = register_shrinker(&f2fs_shrinker_info);
4043 err = register_filesystem(&f2fs_fs_type);
4046 f2fs_create_root_stats();
4047 err = f2fs_init_post_read_processing();
4049 goto free_root_stats;
4050 err = f2fs_init_bio_entry_cache();
4052 goto free_post_read;
4053 err = f2fs_init_bioset();
4055 goto free_bio_enrty_cache;
4056 err = f2fs_init_compress_mempool();
4059 err = f2fs_init_compress_cache();
4061 goto free_compress_mempool;
4063 free_compress_mempool:
4064 f2fs_destroy_compress_mempool();
4066 f2fs_destroy_bioset();
4067 free_bio_enrty_cache:
4068 f2fs_destroy_bio_entry_cache();
4070 f2fs_destroy_post_read_processing();
4072 f2fs_destroy_root_stats();
4073 unregister_filesystem(&f2fs_fs_type);
4075 unregister_shrinker(&f2fs_shrinker_info);
4078 free_garbage_collection_cache:
4079 f2fs_destroy_garbage_collection_cache();
4081 f2fs_destroy_extent_cache();
4082 free_checkpoint_caches:
4083 f2fs_destroy_checkpoint_caches();
4084 free_segment_manager_caches:
4085 f2fs_destroy_segment_manager_caches();
4086 free_node_manager_caches:
4087 f2fs_destroy_node_manager_caches();
4089 destroy_inodecache();
4094 static void __exit exit_f2fs_fs(void)
4096 f2fs_destroy_compress_cache();
4097 f2fs_destroy_compress_mempool();
4098 f2fs_destroy_bioset();
4099 f2fs_destroy_bio_entry_cache();
4100 f2fs_destroy_post_read_processing();
4101 f2fs_destroy_root_stats();
4102 unregister_filesystem(&f2fs_fs_type);
4103 unregister_shrinker(&f2fs_shrinker_info);
4105 f2fs_destroy_garbage_collection_cache();
4106 f2fs_destroy_extent_cache();
4107 f2fs_destroy_checkpoint_caches();
4108 f2fs_destroy_segment_manager_caches();
4109 f2fs_destroy_node_manager_caches();
4110 destroy_inodecache();
4111 f2fs_destroy_trace_ios();
4114 module_init(init_f2fs_fs)
4115 module_exit(exit_f2fs_fs)
4117 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
4118 MODULE_DESCRIPTION("Flash Friendly File System");
4119 MODULE_LICENSE("GPL");