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_POSIX;
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);
1684 clear_opt(sbi, DISABLE_CHECKPOINT);
1685 F2FS_OPTION(sbi).unusable_cap = 0;
1686 sbi->sb->s_flags |= SB_LAZYTIME;
1687 set_opt(sbi, FLUSH_MERGE);
1688 set_opt(sbi, DISCARD);
1689 if (f2fs_sb_has_blkzoned(sbi))
1690 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
1692 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
1694 #ifdef CONFIG_F2FS_FS_XATTR
1695 set_opt(sbi, XATTR_USER);
1697 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1698 set_opt(sbi, POSIX_ACL);
1701 f2fs_build_fault_attr(sbi, 0, 0);
1705 static int f2fs_enable_quotas(struct super_block *sb);
1708 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
1710 unsigned int s_flags = sbi->sb->s_flags;
1711 struct cp_control cpc;
1716 if (s_flags & SB_RDONLY) {
1717 f2fs_err(sbi, "checkpoint=disable on readonly fs");
1720 sbi->sb->s_flags |= SB_ACTIVE;
1722 f2fs_update_time(sbi, DISABLE_TIME);
1724 while (!f2fs_time_over(sbi, DISABLE_TIME)) {
1725 down_write(&sbi->gc_lock);
1726 err = f2fs_gc(sbi, true, false, NULL_SEGNO);
1727 if (err == -ENODATA) {
1731 if (err && err != -EAGAIN)
1735 ret = sync_filesystem(sbi->sb);
1737 err = ret ? ret: err;
1741 unusable = f2fs_get_unusable_blocks(sbi);
1742 if (f2fs_disable_cp_again(sbi, unusable)) {
1747 down_write(&sbi->gc_lock);
1748 cpc.reason = CP_PAUSE;
1749 set_sbi_flag(sbi, SBI_CP_DISABLED);
1750 err = f2fs_write_checkpoint(sbi, &cpc);
1754 spin_lock(&sbi->stat_lock);
1755 sbi->unusable_block_count = unusable;
1756 spin_unlock(&sbi->stat_lock);
1759 up_write(&sbi->gc_lock);
1761 sbi->sb->s_flags = s_flags; /* Restore SB_RDONLY status */
1765 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
1767 down_write(&sbi->gc_lock);
1768 f2fs_dirty_to_prefree(sbi);
1770 clear_sbi_flag(sbi, SBI_CP_DISABLED);
1771 set_sbi_flag(sbi, SBI_IS_DIRTY);
1772 up_write(&sbi->gc_lock);
1774 f2fs_sync_fs(sbi->sb, 1);
1777 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
1779 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1780 struct f2fs_mount_info org_mount_opt;
1781 unsigned long old_sb_flags;
1783 bool need_restart_gc = false;
1784 bool need_stop_gc = false;
1785 bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
1786 bool disable_checkpoint = test_opt(sbi, DISABLE_CHECKPOINT);
1787 bool no_io_align = !F2FS_IO_ALIGNED(sbi);
1788 bool no_atgc = !test_opt(sbi, ATGC);
1789 bool checkpoint_changed;
1795 * Save the old mount options in case we
1796 * need to restore them.
1798 org_mount_opt = sbi->mount_opt;
1799 old_sb_flags = sb->s_flags;
1802 org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
1803 for (i = 0; i < MAXQUOTAS; i++) {
1804 if (F2FS_OPTION(sbi).s_qf_names[i]) {
1805 org_mount_opt.s_qf_names[i] =
1806 kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
1808 if (!org_mount_opt.s_qf_names[i]) {
1809 for (j = 0; j < i; j++)
1810 kfree(org_mount_opt.s_qf_names[j]);
1814 org_mount_opt.s_qf_names[i] = NULL;
1819 /* recover superblocks we couldn't write due to previous RO mount */
1820 if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
1821 err = f2fs_commit_super(sbi, false);
1822 f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
1825 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
1828 default_options(sbi);
1830 /* parse mount options */
1831 err = parse_options(sb, data, true);
1834 checkpoint_changed =
1835 disable_checkpoint != test_opt(sbi, DISABLE_CHECKPOINT);
1838 * Previous and new state of filesystem is RO,
1839 * so skip checking GC and FLUSH_MERGE conditions.
1841 if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
1845 if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
1846 err = dquot_suspend(sb, -1);
1849 } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
1850 /* dquot_resume needs RW */
1851 sb->s_flags &= ~SB_RDONLY;
1852 if (sb_any_quota_suspended(sb)) {
1853 dquot_resume(sb, -1);
1854 } else if (f2fs_sb_has_quota_ino(sbi)) {
1855 err = f2fs_enable_quotas(sb);
1861 /* disallow enable atgc dynamically */
1862 if (no_atgc == !!test_opt(sbi, ATGC)) {
1864 f2fs_warn(sbi, "switch atgc option is not allowed");
1868 /* disallow enable/disable extent_cache dynamically */
1869 if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
1871 f2fs_warn(sbi, "switch extent_cache option is not allowed");
1875 if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
1877 f2fs_warn(sbi, "switch io_bits option is not allowed");
1881 if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
1883 f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
1888 * We stop the GC thread if FS is mounted as RO
1889 * or if background_gc = off is passed in mount
1890 * option. Also sync the filesystem.
1892 if ((*flags & SB_RDONLY) ||
1893 F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF) {
1894 if (sbi->gc_thread) {
1895 f2fs_stop_gc_thread(sbi);
1896 need_restart_gc = true;
1898 } else if (!sbi->gc_thread) {
1899 err = f2fs_start_gc_thread(sbi);
1902 need_stop_gc = true;
1905 if (*flags & SB_RDONLY ||
1906 F2FS_OPTION(sbi).whint_mode != org_mount_opt.whint_mode) {
1907 writeback_inodes_sb(sb, WB_REASON_SYNC);
1910 set_sbi_flag(sbi, SBI_IS_DIRTY);
1911 set_sbi_flag(sbi, SBI_IS_CLOSE);
1912 f2fs_sync_fs(sb, 1);
1913 clear_sbi_flag(sbi, SBI_IS_CLOSE);
1916 if (checkpoint_changed) {
1917 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
1918 err = f2fs_disable_checkpoint(sbi);
1922 f2fs_enable_checkpoint(sbi);
1927 * We stop issue flush thread if FS is mounted as RO
1928 * or if flush_merge is not passed in mount option.
1930 if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
1931 clear_opt(sbi, FLUSH_MERGE);
1932 f2fs_destroy_flush_cmd_control(sbi, false);
1934 err = f2fs_create_flush_cmd_control(sbi);
1940 /* Release old quota file names */
1941 for (i = 0; i < MAXQUOTAS; i++)
1942 kfree(org_mount_opt.s_qf_names[i]);
1944 /* Update the POSIXACL Flag */
1945 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
1946 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
1948 limit_reserve_root(sbi);
1949 adjust_unusable_cap_perc(sbi);
1950 *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
1953 if (need_restart_gc) {
1954 if (f2fs_start_gc_thread(sbi))
1955 f2fs_warn(sbi, "background gc thread has stopped");
1956 } else if (need_stop_gc) {
1957 f2fs_stop_gc_thread(sbi);
1961 F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
1962 for (i = 0; i < MAXQUOTAS; i++) {
1963 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1964 F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
1967 sbi->mount_opt = org_mount_opt;
1968 sb->s_flags = old_sb_flags;
1973 /* Read data from quotafile */
1974 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
1975 size_t len, loff_t off)
1977 struct inode *inode = sb_dqopt(sb)->files[type];
1978 struct address_space *mapping = inode->i_mapping;
1979 block_t blkidx = F2FS_BYTES_TO_BLK(off);
1980 int offset = off & (sb->s_blocksize - 1);
1983 loff_t i_size = i_size_read(inode);
1990 if (off + len > i_size)
1993 while (toread > 0) {
1994 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
1996 page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
1998 if (PTR_ERR(page) == -ENOMEM) {
1999 congestion_wait(BLK_RW_ASYNC,
2000 DEFAULT_IO_TIMEOUT);
2003 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2004 return PTR_ERR(page);
2009 if (unlikely(page->mapping != mapping)) {
2010 f2fs_put_page(page, 1);
2013 if (unlikely(!PageUptodate(page))) {
2014 f2fs_put_page(page, 1);
2015 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2019 kaddr = kmap_atomic(page);
2020 memcpy(data, kaddr + offset, tocopy);
2021 kunmap_atomic(kaddr);
2022 f2fs_put_page(page, 1);
2032 /* Write to quotafile */
2033 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2034 const char *data, size_t len, loff_t off)
2036 struct inode *inode = sb_dqopt(sb)->files[type];
2037 struct address_space *mapping = inode->i_mapping;
2038 const struct address_space_operations *a_ops = mapping->a_ops;
2039 int offset = off & (sb->s_blocksize - 1);
2040 size_t towrite = len;
2042 void *fsdata = NULL;
2047 while (towrite > 0) {
2048 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2051 err = a_ops->write_begin(NULL, mapping, off, tocopy, 0,
2053 if (unlikely(err)) {
2054 if (err == -ENOMEM) {
2055 congestion_wait(BLK_RW_ASYNC,
2056 DEFAULT_IO_TIMEOUT);
2059 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2063 kaddr = kmap_atomic(page);
2064 memcpy(kaddr + offset, data, tocopy);
2065 kunmap_atomic(kaddr);
2066 flush_dcache_page(page);
2068 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2079 inode->i_mtime = inode->i_ctime = current_time(inode);
2080 f2fs_mark_inode_dirty_sync(inode, false);
2081 return len - towrite;
2084 static struct dquot **f2fs_get_dquots(struct inode *inode)
2086 return F2FS_I(inode)->i_dquot;
2089 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2091 return &F2FS_I(inode)->i_reserved_quota;
2094 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2096 if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2097 f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2101 return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2102 F2FS_OPTION(sbi).s_jquota_fmt, type);
2105 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2110 if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2111 err = f2fs_enable_quotas(sbi->sb);
2113 f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2119 for (i = 0; i < MAXQUOTAS; i++) {
2120 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2121 err = f2fs_quota_on_mount(sbi, i);
2126 f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2133 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2136 struct inode *qf_inode;
2137 unsigned long qf_inum;
2140 BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2142 qf_inum = f2fs_qf_ino(sb, type);
2146 qf_inode = f2fs_iget(sb, qf_inum);
2147 if (IS_ERR(qf_inode)) {
2148 f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2149 return PTR_ERR(qf_inode);
2152 /* Don't account quota for quota files to avoid recursion */
2153 qf_inode->i_flags |= S_NOQUOTA;
2154 err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2159 static int f2fs_enable_quotas(struct super_block *sb)
2161 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2163 unsigned long qf_inum;
2164 bool quota_mopt[MAXQUOTAS] = {
2165 test_opt(sbi, USRQUOTA),
2166 test_opt(sbi, GRPQUOTA),
2167 test_opt(sbi, PRJQUOTA),
2170 if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2171 f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2175 sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2177 for (type = 0; type < MAXQUOTAS; type++) {
2178 qf_inum = f2fs_qf_ino(sb, type);
2180 err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2181 DQUOT_USAGE_ENABLED |
2182 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2184 f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2186 for (type--; type >= 0; type--)
2187 dquot_quota_off(sb, type);
2188 set_sbi_flag(F2FS_SB(sb),
2189 SBI_QUOTA_NEED_REPAIR);
2197 int f2fs_quota_sync(struct super_block *sb, int type)
2199 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2200 struct quota_info *dqopt = sb_dqopt(sb);
2207 * down_read(quota_sem)
2208 * dquot_writeback_dquots()
2211 * down_read(quota_sem)
2215 down_read(&sbi->quota_sem);
2216 ret = dquot_writeback_dquots(sb, type);
2221 * Now when everything is written we can discard the pagecache so
2222 * that userspace sees the changes.
2224 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2225 struct address_space *mapping;
2227 if (type != -1 && cnt != type)
2229 if (!sb_has_quota_active(sb, cnt))
2232 mapping = dqopt->files[cnt]->i_mapping;
2234 ret = filemap_fdatawrite(mapping);
2238 /* if we are using journalled quota */
2239 if (is_journalled_quota(sbi))
2242 ret = filemap_fdatawait(mapping);
2244 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2246 inode_lock(dqopt->files[cnt]);
2247 truncate_inode_pages(&dqopt->files[cnt]->i_data, 0);
2248 inode_unlock(dqopt->files[cnt]);
2252 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2253 up_read(&sbi->quota_sem);
2254 f2fs_unlock_op(sbi);
2258 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2259 const struct path *path)
2261 struct inode *inode;
2264 /* if quota sysfile exists, deny enabling quota with specific file */
2265 if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2266 f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2270 err = f2fs_quota_sync(sb, type);
2274 err = dquot_quota_on(sb, type, format_id, path);
2278 inode = d_inode(path->dentry);
2281 F2FS_I(inode)->i_flags |= F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL;
2282 f2fs_set_inode_flags(inode);
2283 inode_unlock(inode);
2284 f2fs_mark_inode_dirty_sync(inode, false);
2289 static int __f2fs_quota_off(struct super_block *sb, int type)
2291 struct inode *inode = sb_dqopt(sb)->files[type];
2294 if (!inode || !igrab(inode))
2295 return dquot_quota_off(sb, type);
2297 err = f2fs_quota_sync(sb, type);
2301 err = dquot_quota_off(sb, type);
2302 if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
2306 F2FS_I(inode)->i_flags &= ~(F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL);
2307 f2fs_set_inode_flags(inode);
2308 inode_unlock(inode);
2309 f2fs_mark_inode_dirty_sync(inode, false);
2315 static int f2fs_quota_off(struct super_block *sb, int type)
2317 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2320 err = __f2fs_quota_off(sb, type);
2323 * quotactl can shutdown journalled quota, result in inconsistence
2324 * between quota record and fs data by following updates, tag the
2325 * flag to let fsck be aware of it.
2327 if (is_journalled_quota(sbi))
2328 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2332 void f2fs_quota_off_umount(struct super_block *sb)
2337 for (type = 0; type < MAXQUOTAS; type++) {
2338 err = __f2fs_quota_off(sb, type);
2340 int ret = dquot_quota_off(sb, type);
2342 f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
2344 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2348 * In case of checkpoint=disable, we must flush quota blocks.
2349 * This can cause NULL exception for node_inode in end_io, since
2350 * put_super already dropped it.
2352 sync_filesystem(sb);
2355 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
2357 struct quota_info *dqopt = sb_dqopt(sb);
2360 for (type = 0; type < MAXQUOTAS; type++) {
2361 if (!dqopt->files[type])
2363 f2fs_inode_synced(dqopt->files[type]);
2367 static int f2fs_dquot_commit(struct dquot *dquot)
2369 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2372 down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
2373 ret = dquot_commit(dquot);
2375 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2376 up_read(&sbi->quota_sem);
2380 static int f2fs_dquot_acquire(struct dquot *dquot)
2382 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2385 down_read(&sbi->quota_sem);
2386 ret = dquot_acquire(dquot);
2388 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2389 up_read(&sbi->quota_sem);
2393 static int f2fs_dquot_release(struct dquot *dquot)
2395 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2396 int ret = dquot_release(dquot);
2399 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2403 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
2405 struct super_block *sb = dquot->dq_sb;
2406 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2407 int ret = dquot_mark_dquot_dirty(dquot);
2409 /* if we are using journalled quota */
2410 if (is_journalled_quota(sbi))
2411 set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
2416 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
2418 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2419 int ret = dquot_commit_info(sb, type);
2422 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2426 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
2428 *projid = F2FS_I(inode)->i_projid;
2432 static const struct dquot_operations f2fs_quota_operations = {
2433 .get_reserved_space = f2fs_get_reserved_space,
2434 .write_dquot = f2fs_dquot_commit,
2435 .acquire_dquot = f2fs_dquot_acquire,
2436 .release_dquot = f2fs_dquot_release,
2437 .mark_dirty = f2fs_dquot_mark_dquot_dirty,
2438 .write_info = f2fs_dquot_commit_info,
2439 .alloc_dquot = dquot_alloc,
2440 .destroy_dquot = dquot_destroy,
2441 .get_projid = f2fs_get_projid,
2442 .get_next_id = dquot_get_next_id,
2445 static const struct quotactl_ops f2fs_quotactl_ops = {
2446 .quota_on = f2fs_quota_on,
2447 .quota_off = f2fs_quota_off,
2448 .quota_sync = f2fs_quota_sync,
2449 .get_state = dquot_get_state,
2450 .set_info = dquot_set_dqinfo,
2451 .get_dqblk = dquot_get_dqblk,
2452 .set_dqblk = dquot_set_dqblk,
2453 .get_nextdqblk = dquot_get_next_dqblk,
2456 int f2fs_quota_sync(struct super_block *sb, int type)
2461 void f2fs_quota_off_umount(struct super_block *sb)
2466 static const struct super_operations f2fs_sops = {
2467 .alloc_inode = f2fs_alloc_inode,
2468 .free_inode = f2fs_free_inode,
2469 .drop_inode = f2fs_drop_inode,
2470 .write_inode = f2fs_write_inode,
2471 .dirty_inode = f2fs_dirty_inode,
2472 .show_options = f2fs_show_options,
2474 .quota_read = f2fs_quota_read,
2475 .quota_write = f2fs_quota_write,
2476 .get_dquots = f2fs_get_dquots,
2478 .evict_inode = f2fs_evict_inode,
2479 .put_super = f2fs_put_super,
2480 .sync_fs = f2fs_sync_fs,
2481 .freeze_fs = f2fs_freeze,
2482 .unfreeze_fs = f2fs_unfreeze,
2483 .statfs = f2fs_statfs,
2484 .remount_fs = f2fs_remount,
2487 #ifdef CONFIG_FS_ENCRYPTION
2488 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
2490 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2491 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2495 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
2498 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2501 * Encrypting the root directory is not allowed because fsck
2502 * expects lost+found directory to exist and remain unencrypted
2503 * if LOST_FOUND feature is enabled.
2506 if (f2fs_sb_has_lost_found(sbi) &&
2507 inode->i_ino == F2FS_ROOT_INO(sbi))
2510 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2511 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2512 ctx, len, fs_data, XATTR_CREATE);
2515 static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
2517 return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
2520 static bool f2fs_has_stable_inodes(struct super_block *sb)
2525 static void f2fs_get_ino_and_lblk_bits(struct super_block *sb,
2526 int *ino_bits_ret, int *lblk_bits_ret)
2528 *ino_bits_ret = 8 * sizeof(nid_t);
2529 *lblk_bits_ret = 8 * sizeof(block_t);
2532 static int f2fs_get_num_devices(struct super_block *sb)
2534 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2536 if (f2fs_is_multi_device(sbi))
2537 return sbi->s_ndevs;
2541 static void f2fs_get_devices(struct super_block *sb,
2542 struct request_queue **devs)
2544 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2547 for (i = 0; i < sbi->s_ndevs; i++)
2548 devs[i] = bdev_get_queue(FDEV(i).bdev);
2551 static const struct fscrypt_operations f2fs_cryptops = {
2552 .key_prefix = "f2fs:",
2553 .get_context = f2fs_get_context,
2554 .set_context = f2fs_set_context,
2555 .get_dummy_policy = f2fs_get_dummy_policy,
2556 .empty_dir = f2fs_empty_dir,
2557 .max_namelen = F2FS_NAME_LEN,
2558 .has_stable_inodes = f2fs_has_stable_inodes,
2559 .get_ino_and_lblk_bits = f2fs_get_ino_and_lblk_bits,
2560 .get_num_devices = f2fs_get_num_devices,
2561 .get_devices = f2fs_get_devices,
2565 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
2566 u64 ino, u32 generation)
2568 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2569 struct inode *inode;
2571 if (f2fs_check_nid_range(sbi, ino))
2572 return ERR_PTR(-ESTALE);
2575 * f2fs_iget isn't quite right if the inode is currently unallocated!
2576 * However f2fs_iget currently does appropriate checks to handle stale
2577 * inodes so everything is OK.
2579 inode = f2fs_iget(sb, ino);
2581 return ERR_CAST(inode);
2582 if (unlikely(generation && inode->i_generation != generation)) {
2583 /* we didn't find the right inode.. */
2585 return ERR_PTR(-ESTALE);
2590 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
2591 int fh_len, int fh_type)
2593 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
2594 f2fs_nfs_get_inode);
2597 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
2598 int fh_len, int fh_type)
2600 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
2601 f2fs_nfs_get_inode);
2604 static const struct export_operations f2fs_export_ops = {
2605 .fh_to_dentry = f2fs_fh_to_dentry,
2606 .fh_to_parent = f2fs_fh_to_parent,
2607 .get_parent = f2fs_get_parent,
2610 static loff_t max_file_blocks(void)
2613 loff_t leaf_count = DEF_ADDRS_PER_BLOCK;
2616 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
2617 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
2618 * space in inode.i_addr, it will be more safe to reassign
2622 /* two direct node blocks */
2623 result += (leaf_count * 2);
2625 /* two indirect node blocks */
2626 leaf_count *= NIDS_PER_BLOCK;
2627 result += (leaf_count * 2);
2629 /* one double indirect node block */
2630 leaf_count *= NIDS_PER_BLOCK;
2631 result += leaf_count;
2636 static int __f2fs_commit_super(struct buffer_head *bh,
2637 struct f2fs_super_block *super)
2641 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
2642 set_buffer_dirty(bh);
2645 /* it's rare case, we can do fua all the time */
2646 return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
2649 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
2650 struct buffer_head *bh)
2652 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2653 (bh->b_data + F2FS_SUPER_OFFSET);
2654 struct super_block *sb = sbi->sb;
2655 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
2656 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
2657 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
2658 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
2659 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
2660 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
2661 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
2662 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
2663 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
2664 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
2665 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2666 u32 segment_count = le32_to_cpu(raw_super->segment_count);
2667 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2668 u64 main_end_blkaddr = main_blkaddr +
2669 (segment_count_main << log_blocks_per_seg);
2670 u64 seg_end_blkaddr = segment0_blkaddr +
2671 (segment_count << log_blocks_per_seg);
2673 if (segment0_blkaddr != cp_blkaddr) {
2674 f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
2675 segment0_blkaddr, cp_blkaddr);
2679 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
2681 f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
2682 cp_blkaddr, sit_blkaddr,
2683 segment_count_ckpt << log_blocks_per_seg);
2687 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
2689 f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
2690 sit_blkaddr, nat_blkaddr,
2691 segment_count_sit << log_blocks_per_seg);
2695 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
2697 f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
2698 nat_blkaddr, ssa_blkaddr,
2699 segment_count_nat << log_blocks_per_seg);
2703 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
2705 f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
2706 ssa_blkaddr, main_blkaddr,
2707 segment_count_ssa << log_blocks_per_seg);
2711 if (main_end_blkaddr > seg_end_blkaddr) {
2712 f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
2713 main_blkaddr, seg_end_blkaddr,
2714 segment_count_main << log_blocks_per_seg);
2716 } else if (main_end_blkaddr < seg_end_blkaddr) {
2720 /* fix in-memory information all the time */
2721 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
2722 segment0_blkaddr) >> log_blocks_per_seg);
2724 if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
2725 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2728 err = __f2fs_commit_super(bh, NULL);
2729 res = err ? "failed" : "done";
2731 f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
2732 res, main_blkaddr, seg_end_blkaddr,
2733 segment_count_main << log_blocks_per_seg);
2740 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
2741 struct buffer_head *bh)
2743 block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
2744 block_t total_sections, blocks_per_seg;
2745 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2746 (bh->b_data + F2FS_SUPER_OFFSET);
2747 unsigned int blocksize;
2748 size_t crc_offset = 0;
2751 if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
2752 f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
2753 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
2757 /* Check checksum_offset and crc in superblock */
2758 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
2759 crc_offset = le32_to_cpu(raw_super->checksum_offset);
2761 offsetof(struct f2fs_super_block, crc)) {
2762 f2fs_info(sbi, "Invalid SB checksum offset: %zu",
2764 return -EFSCORRUPTED;
2766 crc = le32_to_cpu(raw_super->crc);
2767 if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
2768 f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
2769 return -EFSCORRUPTED;
2773 /* Currently, support only 4KB page cache size */
2774 if (F2FS_BLKSIZE != PAGE_SIZE) {
2775 f2fs_info(sbi, "Invalid page_cache_size (%lu), supports only 4KB",
2777 return -EFSCORRUPTED;
2780 /* Currently, support only 4KB block size */
2781 blocksize = 1 << le32_to_cpu(raw_super->log_blocksize);
2782 if (blocksize != F2FS_BLKSIZE) {
2783 f2fs_info(sbi, "Invalid blocksize (%u), supports only 4KB",
2785 return -EFSCORRUPTED;
2788 /* check log blocks per segment */
2789 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
2790 f2fs_info(sbi, "Invalid log blocks per segment (%u)",
2791 le32_to_cpu(raw_super->log_blocks_per_seg));
2792 return -EFSCORRUPTED;
2795 /* Currently, support 512/1024/2048/4096 bytes sector size */
2796 if (le32_to_cpu(raw_super->log_sectorsize) >
2797 F2FS_MAX_LOG_SECTOR_SIZE ||
2798 le32_to_cpu(raw_super->log_sectorsize) <
2799 F2FS_MIN_LOG_SECTOR_SIZE) {
2800 f2fs_info(sbi, "Invalid log sectorsize (%u)",
2801 le32_to_cpu(raw_super->log_sectorsize));
2802 return -EFSCORRUPTED;
2804 if (le32_to_cpu(raw_super->log_sectors_per_block) +
2805 le32_to_cpu(raw_super->log_sectorsize) !=
2806 F2FS_MAX_LOG_SECTOR_SIZE) {
2807 f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
2808 le32_to_cpu(raw_super->log_sectors_per_block),
2809 le32_to_cpu(raw_super->log_sectorsize));
2810 return -EFSCORRUPTED;
2813 segment_count = le32_to_cpu(raw_super->segment_count);
2814 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2815 segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
2816 secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
2817 total_sections = le32_to_cpu(raw_super->section_count);
2819 /* blocks_per_seg should be 512, given the above check */
2820 blocks_per_seg = 1 << le32_to_cpu(raw_super->log_blocks_per_seg);
2822 if (segment_count > F2FS_MAX_SEGMENT ||
2823 segment_count < F2FS_MIN_SEGMENTS) {
2824 f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
2825 return -EFSCORRUPTED;
2828 if (total_sections > segment_count_main || total_sections < 1 ||
2829 segs_per_sec > segment_count || !segs_per_sec) {
2830 f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
2831 segment_count, total_sections, segs_per_sec);
2832 return -EFSCORRUPTED;
2835 if (segment_count_main != total_sections * segs_per_sec) {
2836 f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
2837 segment_count_main, total_sections, segs_per_sec);
2838 return -EFSCORRUPTED;
2841 if ((segment_count / segs_per_sec) < total_sections) {
2842 f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
2843 segment_count, segs_per_sec, total_sections);
2844 return -EFSCORRUPTED;
2847 if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
2848 f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
2849 segment_count, le64_to_cpu(raw_super->block_count));
2850 return -EFSCORRUPTED;
2853 if (RDEV(0).path[0]) {
2854 block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
2857 while (i < MAX_DEVICES && RDEV(i).path[0]) {
2858 dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
2861 if (segment_count != dev_seg_count) {
2862 f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
2863 segment_count, dev_seg_count);
2864 return -EFSCORRUPTED;
2867 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
2868 !bdev_is_zoned(sbi->sb->s_bdev)) {
2869 f2fs_info(sbi, "Zoned block device path is missing");
2870 return -EFSCORRUPTED;
2874 if (secs_per_zone > total_sections || !secs_per_zone) {
2875 f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
2876 secs_per_zone, total_sections);
2877 return -EFSCORRUPTED;
2879 if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
2880 raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
2881 (le32_to_cpu(raw_super->extension_count) +
2882 raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
2883 f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
2884 le32_to_cpu(raw_super->extension_count),
2885 raw_super->hot_ext_count,
2886 F2FS_MAX_EXTENSION);
2887 return -EFSCORRUPTED;
2890 if (le32_to_cpu(raw_super->cp_payload) >
2891 (blocks_per_seg - F2FS_CP_PACKS)) {
2892 f2fs_info(sbi, "Insane cp_payload (%u > %u)",
2893 le32_to_cpu(raw_super->cp_payload),
2894 blocks_per_seg - F2FS_CP_PACKS);
2895 return -EFSCORRUPTED;
2898 /* check reserved ino info */
2899 if (le32_to_cpu(raw_super->node_ino) != 1 ||
2900 le32_to_cpu(raw_super->meta_ino) != 2 ||
2901 le32_to_cpu(raw_super->root_ino) != 3) {
2902 f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
2903 le32_to_cpu(raw_super->node_ino),
2904 le32_to_cpu(raw_super->meta_ino),
2905 le32_to_cpu(raw_super->root_ino));
2906 return -EFSCORRUPTED;
2909 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
2910 if (sanity_check_area_boundary(sbi, bh))
2911 return -EFSCORRUPTED;
2916 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
2918 unsigned int total, fsmeta;
2919 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
2920 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2921 unsigned int ovp_segments, reserved_segments;
2922 unsigned int main_segs, blocks_per_seg;
2923 unsigned int sit_segs, nat_segs;
2924 unsigned int sit_bitmap_size, nat_bitmap_size;
2925 unsigned int log_blocks_per_seg;
2926 unsigned int segment_count_main;
2927 unsigned int cp_pack_start_sum, cp_payload;
2928 block_t user_block_count, valid_user_blocks;
2929 block_t avail_node_count, valid_node_count;
2932 total = le32_to_cpu(raw_super->segment_count);
2933 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
2934 sit_segs = le32_to_cpu(raw_super->segment_count_sit);
2936 nat_segs = le32_to_cpu(raw_super->segment_count_nat);
2938 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
2939 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
2941 if (unlikely(fsmeta >= total))
2944 ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
2945 reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
2947 if (unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
2948 ovp_segments == 0 || reserved_segments == 0)) {
2949 f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
2953 user_block_count = le64_to_cpu(ckpt->user_block_count);
2954 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2955 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2956 if (!user_block_count || user_block_count >=
2957 segment_count_main << log_blocks_per_seg) {
2958 f2fs_err(sbi, "Wrong user_block_count: %u",
2963 valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
2964 if (valid_user_blocks > user_block_count) {
2965 f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
2966 valid_user_blocks, user_block_count);
2970 valid_node_count = le32_to_cpu(ckpt->valid_node_count);
2971 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
2972 if (valid_node_count > avail_node_count) {
2973 f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
2974 valid_node_count, avail_node_count);
2978 main_segs = le32_to_cpu(raw_super->segment_count_main);
2979 blocks_per_seg = sbi->blocks_per_seg;
2981 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
2982 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
2983 le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
2985 for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
2986 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
2987 le32_to_cpu(ckpt->cur_node_segno[j])) {
2988 f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
2990 le32_to_cpu(ckpt->cur_node_segno[i]));
2995 for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
2996 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
2997 le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
2999 for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3000 if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3001 le32_to_cpu(ckpt->cur_data_segno[j])) {
3002 f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3004 le32_to_cpu(ckpt->cur_data_segno[i]));
3009 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3010 for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3011 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3012 le32_to_cpu(ckpt->cur_data_segno[j])) {
3013 f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3015 le32_to_cpu(ckpt->cur_node_segno[i]));
3021 sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3022 nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3024 if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3025 nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3026 f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3027 sit_bitmap_size, nat_bitmap_size);
3031 cp_pack_start_sum = __start_sum_addr(sbi);
3032 cp_payload = __cp_payload(sbi);
3033 if (cp_pack_start_sum < cp_payload + 1 ||
3034 cp_pack_start_sum > blocks_per_seg - 1 -
3035 NR_CURSEG_PERSIST_TYPE) {
3036 f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3041 if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3042 le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3043 f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3044 "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3045 "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3046 le32_to_cpu(ckpt->checksum_offset));
3050 if (unlikely(f2fs_cp_error(sbi))) {
3051 f2fs_err(sbi, "A bug case: need to run fsck");
3057 static void init_sb_info(struct f2fs_sb_info *sbi)
3059 struct f2fs_super_block *raw_super = sbi->raw_super;
3062 sbi->log_sectors_per_block =
3063 le32_to_cpu(raw_super->log_sectors_per_block);
3064 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3065 sbi->blocksize = 1 << sbi->log_blocksize;
3066 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3067 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
3068 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3069 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3070 sbi->total_sections = le32_to_cpu(raw_super->section_count);
3071 sbi->total_node_count =
3072 (le32_to_cpu(raw_super->segment_count_nat) / 2)
3073 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
3074 sbi->root_ino_num = le32_to_cpu(raw_super->root_ino);
3075 sbi->node_ino_num = le32_to_cpu(raw_super->node_ino);
3076 sbi->meta_ino_num = le32_to_cpu(raw_super->meta_ino);
3077 sbi->cur_victim_sec = NULL_SECNO;
3078 sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3079 sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3080 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3081 sbi->migration_granularity = sbi->segs_per_sec;
3083 sbi->dir_level = DEF_DIR_LEVEL;
3084 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3085 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3086 sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3087 sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3088 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3089 sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3090 DEF_UMOUNT_DISCARD_TIMEOUT;
3091 clear_sbi_flag(sbi, SBI_NEED_FSCK);
3093 for (i = 0; i < NR_COUNT_TYPE; i++)
3094 atomic_set(&sbi->nr_pages[i], 0);
3096 for (i = 0; i < META; i++)
3097 atomic_set(&sbi->wb_sync_req[i], 0);
3099 INIT_LIST_HEAD(&sbi->s_list);
3100 mutex_init(&sbi->umount_mutex);
3101 init_rwsem(&sbi->io_order_lock);
3102 spin_lock_init(&sbi->cp_lock);
3104 sbi->dirty_device = 0;
3105 spin_lock_init(&sbi->dev_lock);
3107 init_rwsem(&sbi->sb_lock);
3108 init_rwsem(&sbi->pin_sem);
3111 static int init_percpu_info(struct f2fs_sb_info *sbi)
3115 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3119 err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3122 percpu_counter_destroy(&sbi->alloc_valid_block_count);
3127 #ifdef CONFIG_BLK_DEV_ZONED
3129 struct f2fs_report_zones_args {
3130 struct f2fs_dev_info *dev;
3131 bool zone_cap_mismatch;
3134 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3137 struct f2fs_report_zones_args *rz_args = data;
3139 if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3142 set_bit(idx, rz_args->dev->blkz_seq);
3143 rz_args->dev->zone_capacity_blocks[idx] = zone->capacity >>
3144 F2FS_LOG_SECTORS_PER_BLOCK;
3145 if (zone->len != zone->capacity && !rz_args->zone_cap_mismatch)
3146 rz_args->zone_cap_mismatch = true;
3151 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3153 struct block_device *bdev = FDEV(devi).bdev;
3154 sector_t nr_sectors = bdev->bd_part->nr_sects;
3155 struct f2fs_report_zones_args rep_zone_arg;
3158 if (!f2fs_sb_has_blkzoned(sbi))
3161 if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3162 SECTOR_TO_BLOCK(bdev_zone_sectors(bdev)))
3164 sbi->blocks_per_blkz = SECTOR_TO_BLOCK(bdev_zone_sectors(bdev));
3165 if (sbi->log_blocks_per_blkz && sbi->log_blocks_per_blkz !=
3166 __ilog2_u32(sbi->blocks_per_blkz))
3168 sbi->log_blocks_per_blkz = __ilog2_u32(sbi->blocks_per_blkz);
3169 FDEV(devi).nr_blkz = SECTOR_TO_BLOCK(nr_sectors) >>
3170 sbi->log_blocks_per_blkz;
3171 if (nr_sectors & (bdev_zone_sectors(bdev) - 1))
3172 FDEV(devi).nr_blkz++;
3174 FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3175 BITS_TO_LONGS(FDEV(devi).nr_blkz)
3176 * sizeof(unsigned long),
3178 if (!FDEV(devi).blkz_seq)
3181 /* Get block zones type and zone-capacity */
3182 FDEV(devi).zone_capacity_blocks = f2fs_kzalloc(sbi,
3183 FDEV(devi).nr_blkz * sizeof(block_t),
3185 if (!FDEV(devi).zone_capacity_blocks)
3188 rep_zone_arg.dev = &FDEV(devi);
3189 rep_zone_arg.zone_cap_mismatch = false;
3191 ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3196 if (!rep_zone_arg.zone_cap_mismatch) {
3197 kfree(FDEV(devi).zone_capacity_blocks);
3198 FDEV(devi).zone_capacity_blocks = NULL;
3206 * Read f2fs raw super block.
3207 * Because we have two copies of super block, so read both of them
3208 * to get the first valid one. If any one of them is broken, we pass
3209 * them recovery flag back to the caller.
3211 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3212 struct f2fs_super_block **raw_super,
3213 int *valid_super_block, int *recovery)
3215 struct super_block *sb = sbi->sb;
3217 struct buffer_head *bh;
3218 struct f2fs_super_block *super;
3221 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3225 for (block = 0; block < 2; block++) {
3226 bh = sb_bread(sb, block);
3228 f2fs_err(sbi, "Unable to read %dth superblock",
3235 /* sanity checking of raw super */
3236 err = sanity_check_raw_super(sbi, bh);
3238 f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3246 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3248 *valid_super_block = block;
3254 /* No valid superblock */
3263 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
3265 struct buffer_head *bh;
3269 if ((recover && f2fs_readonly(sbi->sb)) ||
3270 bdev_read_only(sbi->sb->s_bdev)) {
3271 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3275 /* we should update superblock crc here */
3276 if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
3277 crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
3278 offsetof(struct f2fs_super_block, crc));
3279 F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
3282 /* write back-up superblock first */
3283 bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
3286 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3289 /* if we are in recovery path, skip writing valid superblock */
3293 /* write current valid superblock */
3294 bh = sb_bread(sbi->sb, sbi->valid_super_block);
3297 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3302 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
3304 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3305 unsigned int max_devices = MAX_DEVICES;
3308 /* Initialize single device information */
3309 if (!RDEV(0).path[0]) {
3310 if (!bdev_is_zoned(sbi->sb->s_bdev))
3316 * Initialize multiple devices information, or single
3317 * zoned block device information.
3319 sbi->devs = f2fs_kzalloc(sbi,
3320 array_size(max_devices,
3321 sizeof(struct f2fs_dev_info)),
3326 for (i = 0; i < max_devices; i++) {
3328 if (i > 0 && !RDEV(i).path[0])
3331 if (max_devices == 1) {
3332 /* Single zoned block device mount */
3334 blkdev_get_by_dev(sbi->sb->s_bdev->bd_dev,
3335 sbi->sb->s_mode, sbi->sb->s_type);
3337 /* Multi-device mount */
3338 memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
3339 FDEV(i).total_segments =
3340 le32_to_cpu(RDEV(i).total_segments);
3342 FDEV(i).start_blk = 0;
3343 FDEV(i).end_blk = FDEV(i).start_blk +
3344 (FDEV(i).total_segments <<
3345 sbi->log_blocks_per_seg) - 1 +
3346 le32_to_cpu(raw_super->segment0_blkaddr);
3348 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
3349 FDEV(i).end_blk = FDEV(i).start_blk +
3350 (FDEV(i).total_segments <<
3351 sbi->log_blocks_per_seg) - 1;
3353 FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
3354 sbi->sb->s_mode, sbi->sb->s_type);
3356 if (IS_ERR(FDEV(i).bdev))
3357 return PTR_ERR(FDEV(i).bdev);
3359 /* to release errored devices */
3360 sbi->s_ndevs = i + 1;
3362 #ifdef CONFIG_BLK_DEV_ZONED
3363 if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
3364 !f2fs_sb_has_blkzoned(sbi)) {
3365 f2fs_err(sbi, "Zoned block device feature not enabled\n");
3368 if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
3369 if (init_blkz_info(sbi, i)) {
3370 f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
3373 if (max_devices == 1)
3375 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
3377 FDEV(i).total_segments,
3378 FDEV(i).start_blk, FDEV(i).end_blk,
3379 bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
3380 "Host-aware" : "Host-managed");
3384 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
3386 FDEV(i).total_segments,
3387 FDEV(i).start_blk, FDEV(i).end_blk);
3390 "IO Block Size: %8d KB", F2FS_IO_SIZE_KB(sbi));
3394 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
3396 #ifdef CONFIG_UNICODE
3397 if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
3398 const struct f2fs_sb_encodings *encoding_info;
3399 struct unicode_map *encoding;
3400 __u16 encoding_flags;
3402 if (f2fs_sb_has_encrypt(sbi)) {
3404 "Can't mount with encoding and encryption");
3408 if (f2fs_sb_read_encoding(sbi->raw_super, &encoding_info,
3411 "Encoding requested by superblock is unknown");
3415 encoding = utf8_load(encoding_info->version);
3416 if (IS_ERR(encoding)) {
3418 "can't mount with superblock charset: %s-%s "
3419 "not supported by the kernel. flags: 0x%x.",
3420 encoding_info->name, encoding_info->version,
3422 return PTR_ERR(encoding);
3424 f2fs_info(sbi, "Using encoding defined by superblock: "
3425 "%s-%s with flags 0x%hx", encoding_info->name,
3426 encoding_info->version?:"\b", encoding_flags);
3428 sbi->sb->s_encoding = encoding;
3429 sbi->sb->s_encoding_flags = encoding_flags;
3430 sbi->sb->s_d_op = &f2fs_dentry_ops;
3433 if (f2fs_sb_has_casefold(sbi)) {
3434 f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
3441 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
3443 struct f2fs_sm_info *sm_i = SM_I(sbi);
3445 /* adjust parameters according to the volume size */
3446 if (sm_i->main_segments <= SMALL_VOLUME_SEGMENTS) {
3447 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
3448 sm_i->dcc_info->discard_granularity = 1;
3449 sm_i->ipu_policy = 1 << F2FS_IPU_FORCE;
3452 sbi->readdir_ra = 1;
3455 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
3457 struct f2fs_sb_info *sbi;
3458 struct f2fs_super_block *raw_super;
3461 bool skip_recovery = false, need_fsck = false;
3462 char *options = NULL;
3463 int recovery, i, valid_super_block;
3464 struct curseg_info *seg_i;
3470 valid_super_block = -1;
3473 /* allocate memory for f2fs-specific super block info */
3474 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
3480 /* Load the checksum driver */
3481 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
3482 if (IS_ERR(sbi->s_chksum_driver)) {
3483 f2fs_err(sbi, "Cannot load crc32 driver.");
3484 err = PTR_ERR(sbi->s_chksum_driver);
3485 sbi->s_chksum_driver = NULL;
3489 /* set a block size */
3490 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
3491 f2fs_err(sbi, "unable to set blocksize");
3495 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
3500 sb->s_fs_info = sbi;
3501 sbi->raw_super = raw_super;
3503 /* precompute checksum seed for metadata */
3504 if (f2fs_sb_has_inode_chksum(sbi))
3505 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
3506 sizeof(raw_super->uuid));
3508 default_options(sbi);
3509 /* parse mount options */
3510 options = kstrdup((const char *)data, GFP_KERNEL);
3511 if (data && !options) {
3516 err = parse_options(sb, options, false);
3520 sbi->max_file_blocks = max_file_blocks();
3521 sb->s_maxbytes = sbi->max_file_blocks <<
3522 le32_to_cpu(raw_super->log_blocksize);
3523 sb->s_max_links = F2FS_LINK_MAX;
3525 err = f2fs_setup_casefold(sbi);
3530 sb->dq_op = &f2fs_quota_operations;
3531 sb->s_qcop = &f2fs_quotactl_ops;
3532 sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
3534 if (f2fs_sb_has_quota_ino(sbi)) {
3535 for (i = 0; i < MAXQUOTAS; i++) {
3536 if (f2fs_qf_ino(sbi->sb, i))
3537 sbi->nquota_files++;
3542 sb->s_op = &f2fs_sops;
3543 #ifdef CONFIG_FS_ENCRYPTION
3544 sb->s_cop = &f2fs_cryptops;
3546 #ifdef CONFIG_FS_VERITY
3547 sb->s_vop = &f2fs_verityops;
3549 sb->s_xattr = f2fs_xattr_handlers;
3550 sb->s_export_op = &f2fs_export_ops;
3551 sb->s_magic = F2FS_SUPER_MAGIC;
3552 sb->s_time_gran = 1;
3553 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
3554 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
3555 memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
3556 sb->s_iflags |= SB_I_CGROUPWB;
3558 /* init f2fs-specific super block info */
3559 sbi->valid_super_block = valid_super_block;
3560 init_rwsem(&sbi->gc_lock);
3561 mutex_init(&sbi->writepages);
3562 mutex_init(&sbi->cp_mutex);
3563 init_rwsem(&sbi->node_write);
3564 init_rwsem(&sbi->node_change);
3566 /* disallow all the data/node/meta page writes */
3567 set_sbi_flag(sbi, SBI_POR_DOING);
3568 spin_lock_init(&sbi->stat_lock);
3570 /* init iostat info */
3571 spin_lock_init(&sbi->iostat_lock);
3572 sbi->iostat_enable = false;
3573 sbi->iostat_period_ms = DEFAULT_IOSTAT_PERIOD_MS;
3575 for (i = 0; i < NR_PAGE_TYPE; i++) {
3576 int n = (i == META) ? 1: NR_TEMP_TYPE;
3582 sizeof(struct f2fs_bio_info)),
3584 if (!sbi->write_io[i]) {
3589 for (j = HOT; j < n; j++) {
3590 init_rwsem(&sbi->write_io[i][j].io_rwsem);
3591 sbi->write_io[i][j].sbi = sbi;
3592 sbi->write_io[i][j].bio = NULL;
3593 spin_lock_init(&sbi->write_io[i][j].io_lock);
3594 INIT_LIST_HEAD(&sbi->write_io[i][j].io_list);
3595 INIT_LIST_HEAD(&sbi->write_io[i][j].bio_list);
3596 init_rwsem(&sbi->write_io[i][j].bio_list_lock);
3600 init_rwsem(&sbi->cp_rwsem);
3601 init_rwsem(&sbi->quota_sem);
3602 init_waitqueue_head(&sbi->cp_wait);
3605 err = init_percpu_info(sbi);
3609 if (F2FS_IO_ALIGNED(sbi)) {
3610 sbi->write_io_dummy =
3611 mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
3612 if (!sbi->write_io_dummy) {
3618 /* init per sbi slab cache */
3619 err = f2fs_init_xattr_caches(sbi);
3622 err = f2fs_init_page_array_cache(sbi);
3624 goto free_xattr_cache;
3626 /* get an inode for meta space */
3627 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
3628 if (IS_ERR(sbi->meta_inode)) {
3629 f2fs_err(sbi, "Failed to read F2FS meta data inode");
3630 err = PTR_ERR(sbi->meta_inode);
3631 goto free_page_array_cache;
3634 err = f2fs_get_valid_checkpoint(sbi);
3636 f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
3637 goto free_meta_inode;
3640 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
3641 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3642 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
3643 set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
3644 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
3647 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
3648 set_sbi_flag(sbi, SBI_NEED_FSCK);
3650 /* Initialize device list */
3651 err = f2fs_scan_devices(sbi);
3653 f2fs_err(sbi, "Failed to find devices");
3657 err = f2fs_init_post_read_wq(sbi);
3659 f2fs_err(sbi, "Failed to initialize post read workqueue");
3663 sbi->total_valid_node_count =
3664 le32_to_cpu(sbi->ckpt->valid_node_count);
3665 percpu_counter_set(&sbi->total_valid_inode_count,
3666 le32_to_cpu(sbi->ckpt->valid_inode_count));
3667 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
3668 sbi->total_valid_block_count =
3669 le64_to_cpu(sbi->ckpt->valid_block_count);
3670 sbi->last_valid_block_count = sbi->total_valid_block_count;
3671 sbi->reserved_blocks = 0;
3672 sbi->current_reserved_blocks = 0;
3673 limit_reserve_root(sbi);
3674 adjust_unusable_cap_perc(sbi);
3676 for (i = 0; i < NR_INODE_TYPE; i++) {
3677 INIT_LIST_HEAD(&sbi->inode_list[i]);
3678 spin_lock_init(&sbi->inode_lock[i]);
3680 mutex_init(&sbi->flush_lock);
3682 f2fs_init_extent_cache_info(sbi);
3684 f2fs_init_ino_entry_info(sbi);
3686 f2fs_init_fsync_node_info(sbi);
3688 /* setup f2fs internal modules */
3689 err = f2fs_build_segment_manager(sbi);
3691 f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
3695 err = f2fs_build_node_manager(sbi);
3697 f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
3702 /* For write statistics */
3703 if (sb->s_bdev->bd_part)
3704 sbi->sectors_written_start =
3705 (u64)part_stat_read(sb->s_bdev->bd_part,
3706 sectors[STAT_WRITE]);
3708 /* Read accumulated write IO statistics if exists */
3709 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
3710 if (__exist_node_summaries(sbi))
3711 sbi->kbytes_written =
3712 le64_to_cpu(seg_i->journal->info.kbytes_written);
3714 f2fs_build_gc_manager(sbi);
3716 err = f2fs_build_stats(sbi);
3720 /* get an inode for node space */
3721 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
3722 if (IS_ERR(sbi->node_inode)) {
3723 f2fs_err(sbi, "Failed to read node inode");
3724 err = PTR_ERR(sbi->node_inode);
3728 /* read root inode and dentry */
3729 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
3731 f2fs_err(sbi, "Failed to read root inode");
3732 err = PTR_ERR(root);
3733 goto free_node_inode;
3735 if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
3736 !root->i_size || !root->i_nlink) {
3739 goto free_node_inode;
3742 sb->s_root = d_make_root(root); /* allocate root dentry */
3745 goto free_node_inode;
3748 err = f2fs_register_sysfs(sbi);
3750 goto free_root_inode;
3753 /* Enable quota usage during mount */
3754 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
3755 err = f2fs_enable_quotas(sb);
3757 f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
3760 /* if there are any orphan inodes, free them */
3761 err = f2fs_recover_orphan_inodes(sbi);
3765 if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
3766 goto reset_checkpoint;
3768 /* recover fsynced data */
3769 if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
3770 !test_opt(sbi, NORECOVERY)) {
3772 * mount should be failed, when device has readonly mode, and
3773 * previous checkpoint was not done by clean system shutdown.
3775 if (f2fs_hw_is_readonly(sbi)) {
3776 if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
3778 f2fs_err(sbi, "Need to recover fsync data, but write access unavailable");
3781 f2fs_info(sbi, "write access unavailable, skipping recovery");
3782 goto reset_checkpoint;
3786 set_sbi_flag(sbi, SBI_NEED_FSCK);
3789 goto reset_checkpoint;
3791 err = f2fs_recover_fsync_data(sbi, false);
3794 skip_recovery = true;
3796 f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
3801 err = f2fs_recover_fsync_data(sbi, true);
3803 if (!f2fs_readonly(sb) && err > 0) {
3805 f2fs_err(sbi, "Need to recover fsync data");
3811 * If the f2fs is not readonly and fsync data recovery succeeds,
3812 * check zoned block devices' write pointer consistency.
3814 if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
3815 err = f2fs_check_write_pointer(sbi);
3821 f2fs_init_inmem_curseg(sbi);
3823 /* f2fs_recover_fsync_data() cleared this already */
3824 clear_sbi_flag(sbi, SBI_POR_DOING);
3826 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
3827 err = f2fs_disable_checkpoint(sbi);
3829 goto sync_free_meta;
3830 } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
3831 f2fs_enable_checkpoint(sbi);
3835 * If filesystem is not mounted as read-only then
3836 * do start the gc_thread.
3838 if (F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF && !f2fs_readonly(sb)) {
3839 /* After POR, we can run background GC thread.*/
3840 err = f2fs_start_gc_thread(sbi);
3842 goto sync_free_meta;
3846 /* recover broken superblock */
3848 err = f2fs_commit_super(sbi, true);
3849 f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
3850 sbi->valid_super_block ? 1 : 2, err);
3853 f2fs_join_shrinker(sbi);
3855 f2fs_tuning_parameters(sbi);
3857 f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
3858 cur_cp_version(F2FS_CKPT(sbi)));
3859 f2fs_update_time(sbi, CP_TIME);
3860 f2fs_update_time(sbi, REQ_TIME);
3861 clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
3865 /* safe to flush all the data */
3866 sync_filesystem(sbi->sb);
3871 f2fs_truncate_quota_inode_pages(sb);
3872 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
3873 f2fs_quota_off_umount(sbi->sb);
3876 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
3877 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
3878 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
3879 * falls into an infinite loop in f2fs_sync_meta_pages().
3881 truncate_inode_pages_final(META_MAPPING(sbi));
3882 /* evict some inodes being cached by GC */
3884 f2fs_unregister_sysfs(sbi);
3889 f2fs_release_ino_entry(sbi, true);
3890 truncate_inode_pages_final(NODE_MAPPING(sbi));
3891 iput(sbi->node_inode);
3892 sbi->node_inode = NULL;
3894 f2fs_destroy_stats(sbi);
3896 f2fs_destroy_node_manager(sbi);
3898 f2fs_destroy_segment_manager(sbi);
3899 f2fs_destroy_post_read_wq(sbi);
3901 destroy_device_list(sbi);
3904 make_bad_inode(sbi->meta_inode);
3905 iput(sbi->meta_inode);
3906 sbi->meta_inode = NULL;
3907 free_page_array_cache:
3908 f2fs_destroy_page_array_cache(sbi);
3910 f2fs_destroy_xattr_caches(sbi);
3912 mempool_destroy(sbi->write_io_dummy);
3914 destroy_percpu_info(sbi);
3916 for (i = 0; i < NR_PAGE_TYPE; i++)
3917 kvfree(sbi->write_io[i]);
3919 #ifdef CONFIG_UNICODE
3920 utf8_unload(sb->s_encoding);
3924 for (i = 0; i < MAXQUOTAS; i++)
3925 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
3927 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
3932 if (sbi->s_chksum_driver)
3933 crypto_free_shash(sbi->s_chksum_driver);
3936 /* give only one another chance */
3937 if (retry_cnt > 0 && skip_recovery) {
3939 shrink_dcache_sb(sb);
3945 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
3946 const char *dev_name, void *data)
3948 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
3951 static void kill_f2fs_super(struct super_block *sb)
3954 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3956 set_sbi_flag(sbi, SBI_IS_CLOSE);
3957 f2fs_stop_gc_thread(sbi);
3958 f2fs_stop_discard_thread(sbi);
3960 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
3961 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
3962 struct cp_control cpc = {
3963 .reason = CP_UMOUNT,
3965 f2fs_write_checkpoint(sbi, &cpc);
3968 if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
3969 sb->s_flags &= ~SB_RDONLY;
3971 kill_block_super(sb);
3974 static struct file_system_type f2fs_fs_type = {
3975 .owner = THIS_MODULE,
3977 .mount = f2fs_mount,
3978 .kill_sb = kill_f2fs_super,
3979 .fs_flags = FS_REQUIRES_DEV,
3981 MODULE_ALIAS_FS("f2fs");
3983 static int __init init_inodecache(void)
3985 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
3986 sizeof(struct f2fs_inode_info), 0,
3987 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
3988 if (!f2fs_inode_cachep)
3993 static void destroy_inodecache(void)
3996 * Make sure all delayed rcu free inodes are flushed before we
4000 kmem_cache_destroy(f2fs_inode_cachep);
4003 static int __init init_f2fs_fs(void)
4007 if (PAGE_SIZE != F2FS_BLKSIZE) {
4008 printk("F2FS not supported on PAGE_SIZE(%lu) != %d\n",
4009 PAGE_SIZE, F2FS_BLKSIZE);
4013 f2fs_build_trace_ios();
4015 err = init_inodecache();
4018 err = f2fs_create_node_manager_caches();
4020 goto free_inodecache;
4021 err = f2fs_create_segment_manager_caches();
4023 goto free_node_manager_caches;
4024 err = f2fs_create_checkpoint_caches();
4026 goto free_segment_manager_caches;
4027 err = f2fs_create_extent_cache();
4029 goto free_checkpoint_caches;
4030 err = f2fs_create_garbage_collection_cache();
4032 goto free_extent_cache;
4033 err = f2fs_init_sysfs();
4035 goto free_garbage_collection_cache;
4036 err = register_shrinker(&f2fs_shrinker_info);
4039 err = register_filesystem(&f2fs_fs_type);
4042 f2fs_create_root_stats();
4043 err = f2fs_init_post_read_processing();
4045 goto free_root_stats;
4046 err = f2fs_init_bio_entry_cache();
4048 goto free_post_read;
4049 err = f2fs_init_bioset();
4051 goto free_bio_enrty_cache;
4052 err = f2fs_init_compress_mempool();
4055 err = f2fs_init_compress_cache();
4057 goto free_compress_mempool;
4059 free_compress_mempool:
4060 f2fs_destroy_compress_mempool();
4062 f2fs_destroy_bioset();
4063 free_bio_enrty_cache:
4064 f2fs_destroy_bio_entry_cache();
4066 f2fs_destroy_post_read_processing();
4068 f2fs_destroy_root_stats();
4069 unregister_filesystem(&f2fs_fs_type);
4071 unregister_shrinker(&f2fs_shrinker_info);
4074 free_garbage_collection_cache:
4075 f2fs_destroy_garbage_collection_cache();
4077 f2fs_destroy_extent_cache();
4078 free_checkpoint_caches:
4079 f2fs_destroy_checkpoint_caches();
4080 free_segment_manager_caches:
4081 f2fs_destroy_segment_manager_caches();
4082 free_node_manager_caches:
4083 f2fs_destroy_node_manager_caches();
4085 destroy_inodecache();
4090 static void __exit exit_f2fs_fs(void)
4092 f2fs_destroy_compress_cache();
4093 f2fs_destroy_compress_mempool();
4094 f2fs_destroy_bioset();
4095 f2fs_destroy_bio_entry_cache();
4096 f2fs_destroy_post_read_processing();
4097 f2fs_destroy_root_stats();
4098 unregister_filesystem(&f2fs_fs_type);
4099 unregister_shrinker(&f2fs_shrinker_info);
4101 f2fs_destroy_garbage_collection_cache();
4102 f2fs_destroy_extent_cache();
4103 f2fs_destroy_checkpoint_caches();
4104 f2fs_destroy_segment_manager_caches();
4105 f2fs_destroy_node_manager_caches();
4106 destroy_inodecache();
4107 f2fs_destroy_trace_ios();
4110 module_init(init_f2fs_fs)
4111 module_exit(exit_f2fs_fs)
4113 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
4114 MODULE_DESCRIPTION("Flash Friendly File System");
4115 MODULE_LICENSE("GPL");