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.val[0] = (u32)id;
1446 buf->f_fsid.val[1] = (u32)(id >> 32);
1449 if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1450 sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1451 f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1457 static inline void f2fs_show_quota_options(struct seq_file *seq,
1458 struct super_block *sb)
1461 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1463 if (F2FS_OPTION(sbi).s_jquota_fmt) {
1466 switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1477 seq_printf(seq, ",jqfmt=%s", fmtname);
1480 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1481 seq_show_option(seq, "usrjquota",
1482 F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1484 if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1485 seq_show_option(seq, "grpjquota",
1486 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1488 if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1489 seq_show_option(seq, "prjjquota",
1490 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1494 static inline void f2fs_show_compress_options(struct seq_file *seq,
1495 struct super_block *sb)
1497 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1501 if (!f2fs_sb_has_compression(sbi))
1504 switch (F2FS_OPTION(sbi).compress_algorithm) {
1514 case COMPRESS_LZORLE:
1515 algtype = "lzo-rle";
1518 seq_printf(seq, ",compress_algorithm=%s", algtype);
1520 seq_printf(seq, ",compress_log_size=%u",
1521 F2FS_OPTION(sbi).compress_log_size);
1523 for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1524 seq_printf(seq, ",compress_extension=%s",
1525 F2FS_OPTION(sbi).extensions[i]);
1529 static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1531 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1533 if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1534 seq_printf(seq, ",background_gc=%s", "sync");
1535 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1536 seq_printf(seq, ",background_gc=%s", "on");
1537 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1538 seq_printf(seq, ",background_gc=%s", "off");
1540 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1541 seq_puts(seq, ",disable_roll_forward");
1542 if (test_opt(sbi, NORECOVERY))
1543 seq_puts(seq, ",norecovery");
1544 if (test_opt(sbi, DISCARD))
1545 seq_puts(seq, ",discard");
1547 seq_puts(seq, ",nodiscard");
1548 if (test_opt(sbi, NOHEAP))
1549 seq_puts(seq, ",no_heap");
1551 seq_puts(seq, ",heap");
1552 #ifdef CONFIG_F2FS_FS_XATTR
1553 if (test_opt(sbi, XATTR_USER))
1554 seq_puts(seq, ",user_xattr");
1556 seq_puts(seq, ",nouser_xattr");
1557 if (test_opt(sbi, INLINE_XATTR))
1558 seq_puts(seq, ",inline_xattr");
1560 seq_puts(seq, ",noinline_xattr");
1561 if (test_opt(sbi, INLINE_XATTR_SIZE))
1562 seq_printf(seq, ",inline_xattr_size=%u",
1563 F2FS_OPTION(sbi).inline_xattr_size);
1565 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1566 if (test_opt(sbi, POSIX_ACL))
1567 seq_puts(seq, ",acl");
1569 seq_puts(seq, ",noacl");
1571 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
1572 seq_puts(seq, ",disable_ext_identify");
1573 if (test_opt(sbi, INLINE_DATA))
1574 seq_puts(seq, ",inline_data");
1576 seq_puts(seq, ",noinline_data");
1577 if (test_opt(sbi, INLINE_DENTRY))
1578 seq_puts(seq, ",inline_dentry");
1580 seq_puts(seq, ",noinline_dentry");
1581 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
1582 seq_puts(seq, ",flush_merge");
1583 if (test_opt(sbi, NOBARRIER))
1584 seq_puts(seq, ",nobarrier");
1585 if (test_opt(sbi, FASTBOOT))
1586 seq_puts(seq, ",fastboot");
1587 if (test_opt(sbi, EXTENT_CACHE))
1588 seq_puts(seq, ",extent_cache");
1590 seq_puts(seq, ",noextent_cache");
1591 if (test_opt(sbi, DATA_FLUSH))
1592 seq_puts(seq, ",data_flush");
1594 seq_puts(seq, ",mode=");
1595 if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
1596 seq_puts(seq, "adaptive");
1597 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
1598 seq_puts(seq, "lfs");
1599 seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
1600 if (test_opt(sbi, RESERVE_ROOT))
1601 seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
1602 F2FS_OPTION(sbi).root_reserved_blocks,
1603 from_kuid_munged(&init_user_ns,
1604 F2FS_OPTION(sbi).s_resuid),
1605 from_kgid_munged(&init_user_ns,
1606 F2FS_OPTION(sbi).s_resgid));
1607 if (F2FS_IO_SIZE_BITS(sbi))
1608 seq_printf(seq, ",io_bits=%u",
1609 F2FS_OPTION(sbi).write_io_size_bits);
1610 #ifdef CONFIG_F2FS_FAULT_INJECTION
1611 if (test_opt(sbi, FAULT_INJECTION)) {
1612 seq_printf(seq, ",fault_injection=%u",
1613 F2FS_OPTION(sbi).fault_info.inject_rate);
1614 seq_printf(seq, ",fault_type=%u",
1615 F2FS_OPTION(sbi).fault_info.inject_type);
1619 if (test_opt(sbi, QUOTA))
1620 seq_puts(seq, ",quota");
1621 if (test_opt(sbi, USRQUOTA))
1622 seq_puts(seq, ",usrquota");
1623 if (test_opt(sbi, GRPQUOTA))
1624 seq_puts(seq, ",grpquota");
1625 if (test_opt(sbi, PRJQUOTA))
1626 seq_puts(seq, ",prjquota");
1628 f2fs_show_quota_options(seq, sbi->sb);
1629 if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_USER)
1630 seq_printf(seq, ",whint_mode=%s", "user-based");
1631 else if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_FS)
1632 seq_printf(seq, ",whint_mode=%s", "fs-based");
1634 fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
1636 if (sbi->sb->s_flags & SB_INLINECRYPT)
1637 seq_puts(seq, ",inlinecrypt");
1639 if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
1640 seq_printf(seq, ",alloc_mode=%s", "default");
1641 else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
1642 seq_printf(seq, ",alloc_mode=%s", "reuse");
1644 if (test_opt(sbi, DISABLE_CHECKPOINT))
1645 seq_printf(seq, ",checkpoint=disable:%u",
1646 F2FS_OPTION(sbi).unusable_cap);
1647 if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
1648 seq_printf(seq, ",fsync_mode=%s", "posix");
1649 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
1650 seq_printf(seq, ",fsync_mode=%s", "strict");
1651 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
1652 seq_printf(seq, ",fsync_mode=%s", "nobarrier");
1654 #ifdef CONFIG_F2FS_FS_COMPRESSION
1655 f2fs_show_compress_options(seq, sbi->sb);
1658 if (test_opt(sbi, ATGC))
1659 seq_puts(seq, ",atgc");
1663 static void default_options(struct f2fs_sb_info *sbi)
1665 /* init some FS parameters */
1666 F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
1667 F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
1668 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1669 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
1670 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
1671 F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
1672 F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
1673 F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
1674 F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
1675 F2FS_OPTION(sbi).compress_ext_cnt = 0;
1676 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
1678 sbi->sb->s_flags &= ~SB_INLINECRYPT;
1680 set_opt(sbi, INLINE_XATTR);
1681 set_opt(sbi, INLINE_DATA);
1682 set_opt(sbi, INLINE_DENTRY);
1683 set_opt(sbi, EXTENT_CACHE);
1684 set_opt(sbi, NOHEAP);
1685 clear_opt(sbi, DISABLE_CHECKPOINT);
1686 F2FS_OPTION(sbi).unusable_cap = 0;
1687 sbi->sb->s_flags |= SB_LAZYTIME;
1688 set_opt(sbi, FLUSH_MERGE);
1689 set_opt(sbi, DISCARD);
1690 if (f2fs_sb_has_blkzoned(sbi))
1691 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
1693 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
1695 #ifdef CONFIG_F2FS_FS_XATTR
1696 set_opt(sbi, XATTR_USER);
1698 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1699 set_opt(sbi, POSIX_ACL);
1702 f2fs_build_fault_attr(sbi, 0, 0);
1706 static int f2fs_enable_quotas(struct super_block *sb);
1709 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
1711 unsigned int s_flags = sbi->sb->s_flags;
1712 struct cp_control cpc;
1717 if (s_flags & SB_RDONLY) {
1718 f2fs_err(sbi, "checkpoint=disable on readonly fs");
1721 sbi->sb->s_flags |= SB_ACTIVE;
1723 f2fs_update_time(sbi, DISABLE_TIME);
1725 while (!f2fs_time_over(sbi, DISABLE_TIME)) {
1726 down_write(&sbi->gc_lock);
1727 err = f2fs_gc(sbi, true, false, NULL_SEGNO);
1728 if (err == -ENODATA) {
1732 if (err && err != -EAGAIN)
1736 ret = sync_filesystem(sbi->sb);
1738 err = ret ? ret: err;
1742 unusable = f2fs_get_unusable_blocks(sbi);
1743 if (f2fs_disable_cp_again(sbi, unusable)) {
1748 down_write(&sbi->gc_lock);
1749 cpc.reason = CP_PAUSE;
1750 set_sbi_flag(sbi, SBI_CP_DISABLED);
1751 err = f2fs_write_checkpoint(sbi, &cpc);
1755 spin_lock(&sbi->stat_lock);
1756 sbi->unusable_block_count = unusable;
1757 spin_unlock(&sbi->stat_lock);
1760 up_write(&sbi->gc_lock);
1762 sbi->sb->s_flags = s_flags; /* Restore SB_RDONLY status */
1766 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
1768 down_write(&sbi->gc_lock);
1769 f2fs_dirty_to_prefree(sbi);
1771 clear_sbi_flag(sbi, SBI_CP_DISABLED);
1772 set_sbi_flag(sbi, SBI_IS_DIRTY);
1773 up_write(&sbi->gc_lock);
1775 f2fs_sync_fs(sbi->sb, 1);
1778 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
1780 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1781 struct f2fs_mount_info org_mount_opt;
1782 unsigned long old_sb_flags;
1784 bool need_restart_gc = false;
1785 bool need_stop_gc = false;
1786 bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
1787 bool disable_checkpoint = test_opt(sbi, DISABLE_CHECKPOINT);
1788 bool no_io_align = !F2FS_IO_ALIGNED(sbi);
1789 bool no_atgc = !test_opt(sbi, ATGC);
1790 bool checkpoint_changed;
1796 * Save the old mount options in case we
1797 * need to restore them.
1799 org_mount_opt = sbi->mount_opt;
1800 old_sb_flags = sb->s_flags;
1803 org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
1804 for (i = 0; i < MAXQUOTAS; i++) {
1805 if (F2FS_OPTION(sbi).s_qf_names[i]) {
1806 org_mount_opt.s_qf_names[i] =
1807 kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
1809 if (!org_mount_opt.s_qf_names[i]) {
1810 for (j = 0; j < i; j++)
1811 kfree(org_mount_opt.s_qf_names[j]);
1815 org_mount_opt.s_qf_names[i] = NULL;
1820 /* recover superblocks we couldn't write due to previous RO mount */
1821 if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
1822 err = f2fs_commit_super(sbi, false);
1823 f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
1826 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
1829 default_options(sbi);
1831 /* parse mount options */
1832 err = parse_options(sb, data, true);
1835 checkpoint_changed =
1836 disable_checkpoint != test_opt(sbi, DISABLE_CHECKPOINT);
1839 * Previous and new state of filesystem is RO,
1840 * so skip checking GC and FLUSH_MERGE conditions.
1842 if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
1846 if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
1847 err = dquot_suspend(sb, -1);
1850 } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
1851 /* dquot_resume needs RW */
1852 sb->s_flags &= ~SB_RDONLY;
1853 if (sb_any_quota_suspended(sb)) {
1854 dquot_resume(sb, -1);
1855 } else if (f2fs_sb_has_quota_ino(sbi)) {
1856 err = f2fs_enable_quotas(sb);
1862 /* disallow enable atgc dynamically */
1863 if (no_atgc == !!test_opt(sbi, ATGC)) {
1865 f2fs_warn(sbi, "switch atgc option is not allowed");
1869 /* disallow enable/disable extent_cache dynamically */
1870 if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
1872 f2fs_warn(sbi, "switch extent_cache option is not allowed");
1876 if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
1878 f2fs_warn(sbi, "switch io_bits option is not allowed");
1882 if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
1884 f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
1889 * We stop the GC thread if FS is mounted as RO
1890 * or if background_gc = off is passed in mount
1891 * option. Also sync the filesystem.
1893 if ((*flags & SB_RDONLY) ||
1894 F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF) {
1895 if (sbi->gc_thread) {
1896 f2fs_stop_gc_thread(sbi);
1897 need_restart_gc = true;
1899 } else if (!sbi->gc_thread) {
1900 err = f2fs_start_gc_thread(sbi);
1903 need_stop_gc = true;
1906 if (*flags & SB_RDONLY ||
1907 F2FS_OPTION(sbi).whint_mode != org_mount_opt.whint_mode) {
1908 writeback_inodes_sb(sb, WB_REASON_SYNC);
1911 set_sbi_flag(sbi, SBI_IS_DIRTY);
1912 set_sbi_flag(sbi, SBI_IS_CLOSE);
1913 f2fs_sync_fs(sb, 1);
1914 clear_sbi_flag(sbi, SBI_IS_CLOSE);
1917 if (checkpoint_changed) {
1918 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
1919 err = f2fs_disable_checkpoint(sbi);
1923 f2fs_enable_checkpoint(sbi);
1928 * We stop issue flush thread if FS is mounted as RO
1929 * or if flush_merge is not passed in mount option.
1931 if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
1932 clear_opt(sbi, FLUSH_MERGE);
1933 f2fs_destroy_flush_cmd_control(sbi, false);
1935 err = f2fs_create_flush_cmd_control(sbi);
1941 /* Release old quota file names */
1942 for (i = 0; i < MAXQUOTAS; i++)
1943 kfree(org_mount_opt.s_qf_names[i]);
1945 /* Update the POSIXACL Flag */
1946 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
1947 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
1949 limit_reserve_root(sbi);
1950 adjust_unusable_cap_perc(sbi);
1951 *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
1954 if (need_restart_gc) {
1955 if (f2fs_start_gc_thread(sbi))
1956 f2fs_warn(sbi, "background gc thread has stopped");
1957 } else if (need_stop_gc) {
1958 f2fs_stop_gc_thread(sbi);
1962 F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
1963 for (i = 0; i < MAXQUOTAS; i++) {
1964 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1965 F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
1968 sbi->mount_opt = org_mount_opt;
1969 sb->s_flags = old_sb_flags;
1974 /* Read data from quotafile */
1975 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
1976 size_t len, loff_t off)
1978 struct inode *inode = sb_dqopt(sb)->files[type];
1979 struct address_space *mapping = inode->i_mapping;
1980 block_t blkidx = F2FS_BYTES_TO_BLK(off);
1981 int offset = off & (sb->s_blocksize - 1);
1984 loff_t i_size = i_size_read(inode);
1991 if (off + len > i_size)
1994 while (toread > 0) {
1995 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
1997 page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
1999 if (PTR_ERR(page) == -ENOMEM) {
2000 congestion_wait(BLK_RW_ASYNC,
2001 DEFAULT_IO_TIMEOUT);
2004 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2005 return PTR_ERR(page);
2010 if (unlikely(page->mapping != mapping)) {
2011 f2fs_put_page(page, 1);
2014 if (unlikely(!PageUptodate(page))) {
2015 f2fs_put_page(page, 1);
2016 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2020 kaddr = kmap_atomic(page);
2021 memcpy(data, kaddr + offset, tocopy);
2022 kunmap_atomic(kaddr);
2023 f2fs_put_page(page, 1);
2033 /* Write to quotafile */
2034 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2035 const char *data, size_t len, loff_t off)
2037 struct inode *inode = sb_dqopt(sb)->files[type];
2038 struct address_space *mapping = inode->i_mapping;
2039 const struct address_space_operations *a_ops = mapping->a_ops;
2040 int offset = off & (sb->s_blocksize - 1);
2041 size_t towrite = len;
2043 void *fsdata = NULL;
2048 while (towrite > 0) {
2049 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2052 err = a_ops->write_begin(NULL, mapping, off, tocopy, 0,
2054 if (unlikely(err)) {
2055 if (err == -ENOMEM) {
2056 congestion_wait(BLK_RW_ASYNC,
2057 DEFAULT_IO_TIMEOUT);
2060 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2064 kaddr = kmap_atomic(page);
2065 memcpy(kaddr + offset, data, tocopy);
2066 kunmap_atomic(kaddr);
2067 flush_dcache_page(page);
2069 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2080 inode->i_mtime = inode->i_ctime = current_time(inode);
2081 f2fs_mark_inode_dirty_sync(inode, false);
2082 return len - towrite;
2085 static struct dquot **f2fs_get_dquots(struct inode *inode)
2087 return F2FS_I(inode)->i_dquot;
2090 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2092 return &F2FS_I(inode)->i_reserved_quota;
2095 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2097 if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2098 f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2102 return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2103 F2FS_OPTION(sbi).s_jquota_fmt, type);
2106 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2111 if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2112 err = f2fs_enable_quotas(sbi->sb);
2114 f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2120 for (i = 0; i < MAXQUOTAS; i++) {
2121 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2122 err = f2fs_quota_on_mount(sbi, i);
2127 f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2134 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2137 struct inode *qf_inode;
2138 unsigned long qf_inum;
2141 BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2143 qf_inum = f2fs_qf_ino(sb, type);
2147 qf_inode = f2fs_iget(sb, qf_inum);
2148 if (IS_ERR(qf_inode)) {
2149 f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2150 return PTR_ERR(qf_inode);
2153 /* Don't account quota for quota files to avoid recursion */
2154 qf_inode->i_flags |= S_NOQUOTA;
2155 err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2160 static int f2fs_enable_quotas(struct super_block *sb)
2162 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2164 unsigned long qf_inum;
2165 bool quota_mopt[MAXQUOTAS] = {
2166 test_opt(sbi, USRQUOTA),
2167 test_opt(sbi, GRPQUOTA),
2168 test_opt(sbi, PRJQUOTA),
2171 if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2172 f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2176 sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2178 for (type = 0; type < MAXQUOTAS; type++) {
2179 qf_inum = f2fs_qf_ino(sb, type);
2181 err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2182 DQUOT_USAGE_ENABLED |
2183 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2185 f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2187 for (type--; type >= 0; type--)
2188 dquot_quota_off(sb, type);
2189 set_sbi_flag(F2FS_SB(sb),
2190 SBI_QUOTA_NEED_REPAIR);
2198 int f2fs_quota_sync(struct super_block *sb, int type)
2200 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2201 struct quota_info *dqopt = sb_dqopt(sb);
2208 * down_read(quota_sem)
2209 * dquot_writeback_dquots()
2212 * down_read(quota_sem)
2216 down_read(&sbi->quota_sem);
2217 ret = dquot_writeback_dquots(sb, type);
2222 * Now when everything is written we can discard the pagecache so
2223 * that userspace sees the changes.
2225 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2226 struct address_space *mapping;
2228 if (type != -1 && cnt != type)
2230 if (!sb_has_quota_active(sb, cnt))
2233 mapping = dqopt->files[cnt]->i_mapping;
2235 ret = filemap_fdatawrite(mapping);
2239 /* if we are using journalled quota */
2240 if (is_journalled_quota(sbi))
2243 ret = filemap_fdatawait(mapping);
2245 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2247 inode_lock(dqopt->files[cnt]);
2248 truncate_inode_pages(&dqopt->files[cnt]->i_data, 0);
2249 inode_unlock(dqopt->files[cnt]);
2253 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2254 up_read(&sbi->quota_sem);
2255 f2fs_unlock_op(sbi);
2259 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2260 const struct path *path)
2262 struct inode *inode;
2265 /* if quota sysfile exists, deny enabling quota with specific file */
2266 if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2267 f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2271 err = f2fs_quota_sync(sb, type);
2275 err = dquot_quota_on(sb, type, format_id, path);
2279 inode = d_inode(path->dentry);
2282 F2FS_I(inode)->i_flags |= F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL;
2283 f2fs_set_inode_flags(inode);
2284 inode_unlock(inode);
2285 f2fs_mark_inode_dirty_sync(inode, false);
2290 static int __f2fs_quota_off(struct super_block *sb, int type)
2292 struct inode *inode = sb_dqopt(sb)->files[type];
2295 if (!inode || !igrab(inode))
2296 return dquot_quota_off(sb, type);
2298 err = f2fs_quota_sync(sb, type);
2302 err = dquot_quota_off(sb, type);
2303 if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
2307 F2FS_I(inode)->i_flags &= ~(F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL);
2308 f2fs_set_inode_flags(inode);
2309 inode_unlock(inode);
2310 f2fs_mark_inode_dirty_sync(inode, false);
2316 static int f2fs_quota_off(struct super_block *sb, int type)
2318 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2321 err = __f2fs_quota_off(sb, type);
2324 * quotactl can shutdown journalled quota, result in inconsistence
2325 * between quota record and fs data by following updates, tag the
2326 * flag to let fsck be aware of it.
2328 if (is_journalled_quota(sbi))
2329 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2333 void f2fs_quota_off_umount(struct super_block *sb)
2338 for (type = 0; type < MAXQUOTAS; type++) {
2339 err = __f2fs_quota_off(sb, type);
2341 int ret = dquot_quota_off(sb, type);
2343 f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
2345 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2349 * In case of checkpoint=disable, we must flush quota blocks.
2350 * This can cause NULL exception for node_inode in end_io, since
2351 * put_super already dropped it.
2353 sync_filesystem(sb);
2356 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
2358 struct quota_info *dqopt = sb_dqopt(sb);
2361 for (type = 0; type < MAXQUOTAS; type++) {
2362 if (!dqopt->files[type])
2364 f2fs_inode_synced(dqopt->files[type]);
2368 static int f2fs_dquot_commit(struct dquot *dquot)
2370 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2373 down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
2374 ret = dquot_commit(dquot);
2376 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2377 up_read(&sbi->quota_sem);
2381 static int f2fs_dquot_acquire(struct dquot *dquot)
2383 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2386 down_read(&sbi->quota_sem);
2387 ret = dquot_acquire(dquot);
2389 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2390 up_read(&sbi->quota_sem);
2394 static int f2fs_dquot_release(struct dquot *dquot)
2396 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2397 int ret = dquot_release(dquot);
2400 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2404 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
2406 struct super_block *sb = dquot->dq_sb;
2407 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2408 int ret = dquot_mark_dquot_dirty(dquot);
2410 /* if we are using journalled quota */
2411 if (is_journalled_quota(sbi))
2412 set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
2417 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
2419 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2420 int ret = dquot_commit_info(sb, type);
2423 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2427 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
2429 *projid = F2FS_I(inode)->i_projid;
2433 static const struct dquot_operations f2fs_quota_operations = {
2434 .get_reserved_space = f2fs_get_reserved_space,
2435 .write_dquot = f2fs_dquot_commit,
2436 .acquire_dquot = f2fs_dquot_acquire,
2437 .release_dquot = f2fs_dquot_release,
2438 .mark_dirty = f2fs_dquot_mark_dquot_dirty,
2439 .write_info = f2fs_dquot_commit_info,
2440 .alloc_dquot = dquot_alloc,
2441 .destroy_dquot = dquot_destroy,
2442 .get_projid = f2fs_get_projid,
2443 .get_next_id = dquot_get_next_id,
2446 static const struct quotactl_ops f2fs_quotactl_ops = {
2447 .quota_on = f2fs_quota_on,
2448 .quota_off = f2fs_quota_off,
2449 .quota_sync = f2fs_quota_sync,
2450 .get_state = dquot_get_state,
2451 .set_info = dquot_set_dqinfo,
2452 .get_dqblk = dquot_get_dqblk,
2453 .set_dqblk = dquot_set_dqblk,
2454 .get_nextdqblk = dquot_get_next_dqblk,
2457 int f2fs_quota_sync(struct super_block *sb, int type)
2462 void f2fs_quota_off_umount(struct super_block *sb)
2467 static const struct super_operations f2fs_sops = {
2468 .alloc_inode = f2fs_alloc_inode,
2469 .free_inode = f2fs_free_inode,
2470 .drop_inode = f2fs_drop_inode,
2471 .write_inode = f2fs_write_inode,
2472 .dirty_inode = f2fs_dirty_inode,
2473 .show_options = f2fs_show_options,
2475 .quota_read = f2fs_quota_read,
2476 .quota_write = f2fs_quota_write,
2477 .get_dquots = f2fs_get_dquots,
2479 .evict_inode = f2fs_evict_inode,
2480 .put_super = f2fs_put_super,
2481 .sync_fs = f2fs_sync_fs,
2482 .freeze_fs = f2fs_freeze,
2483 .unfreeze_fs = f2fs_unfreeze,
2484 .statfs = f2fs_statfs,
2485 .remount_fs = f2fs_remount,
2488 #ifdef CONFIG_FS_ENCRYPTION
2489 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
2491 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2492 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2496 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
2499 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2502 * Encrypting the root directory is not allowed because fsck
2503 * expects lost+found directory to exist and remain unencrypted
2504 * if LOST_FOUND feature is enabled.
2507 if (f2fs_sb_has_lost_found(sbi) &&
2508 inode->i_ino == F2FS_ROOT_INO(sbi))
2511 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2512 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2513 ctx, len, fs_data, XATTR_CREATE);
2516 static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
2518 return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
2521 static bool f2fs_has_stable_inodes(struct super_block *sb)
2526 static void f2fs_get_ino_and_lblk_bits(struct super_block *sb,
2527 int *ino_bits_ret, int *lblk_bits_ret)
2529 *ino_bits_ret = 8 * sizeof(nid_t);
2530 *lblk_bits_ret = 8 * sizeof(block_t);
2533 static int f2fs_get_num_devices(struct super_block *sb)
2535 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2537 if (f2fs_is_multi_device(sbi))
2538 return sbi->s_ndevs;
2542 static void f2fs_get_devices(struct super_block *sb,
2543 struct request_queue **devs)
2545 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2548 for (i = 0; i < sbi->s_ndevs; i++)
2549 devs[i] = bdev_get_queue(FDEV(i).bdev);
2552 static const struct fscrypt_operations f2fs_cryptops = {
2553 .key_prefix = "f2fs:",
2554 .get_context = f2fs_get_context,
2555 .set_context = f2fs_set_context,
2556 .get_dummy_policy = f2fs_get_dummy_policy,
2557 .empty_dir = f2fs_empty_dir,
2558 .max_namelen = F2FS_NAME_LEN,
2559 .has_stable_inodes = f2fs_has_stable_inodes,
2560 .get_ino_and_lblk_bits = f2fs_get_ino_and_lblk_bits,
2561 .get_num_devices = f2fs_get_num_devices,
2562 .get_devices = f2fs_get_devices,
2566 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
2567 u64 ino, u32 generation)
2569 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2570 struct inode *inode;
2572 if (f2fs_check_nid_range(sbi, ino))
2573 return ERR_PTR(-ESTALE);
2576 * f2fs_iget isn't quite right if the inode is currently unallocated!
2577 * However f2fs_iget currently does appropriate checks to handle stale
2578 * inodes so everything is OK.
2580 inode = f2fs_iget(sb, ino);
2582 return ERR_CAST(inode);
2583 if (unlikely(generation && inode->i_generation != generation)) {
2584 /* we didn't find the right inode.. */
2586 return ERR_PTR(-ESTALE);
2591 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
2592 int fh_len, int fh_type)
2594 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
2595 f2fs_nfs_get_inode);
2598 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
2599 int fh_len, int fh_type)
2601 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
2602 f2fs_nfs_get_inode);
2605 static const struct export_operations f2fs_export_ops = {
2606 .fh_to_dentry = f2fs_fh_to_dentry,
2607 .fh_to_parent = f2fs_fh_to_parent,
2608 .get_parent = f2fs_get_parent,
2611 static loff_t max_file_blocks(void)
2614 loff_t leaf_count = DEF_ADDRS_PER_BLOCK;
2617 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
2618 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
2619 * space in inode.i_addr, it will be more safe to reassign
2623 /* two direct node blocks */
2624 result += (leaf_count * 2);
2626 /* two indirect node blocks */
2627 leaf_count *= NIDS_PER_BLOCK;
2628 result += (leaf_count * 2);
2630 /* one double indirect node block */
2631 leaf_count *= NIDS_PER_BLOCK;
2632 result += leaf_count;
2637 static int __f2fs_commit_super(struct buffer_head *bh,
2638 struct f2fs_super_block *super)
2642 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
2643 set_buffer_dirty(bh);
2646 /* it's rare case, we can do fua all the time */
2647 return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
2650 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
2651 struct buffer_head *bh)
2653 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2654 (bh->b_data + F2FS_SUPER_OFFSET);
2655 struct super_block *sb = sbi->sb;
2656 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
2657 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
2658 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
2659 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
2660 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
2661 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
2662 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
2663 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
2664 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
2665 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
2666 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2667 u32 segment_count = le32_to_cpu(raw_super->segment_count);
2668 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2669 u64 main_end_blkaddr = main_blkaddr +
2670 (segment_count_main << log_blocks_per_seg);
2671 u64 seg_end_blkaddr = segment0_blkaddr +
2672 (segment_count << log_blocks_per_seg);
2674 if (segment0_blkaddr != cp_blkaddr) {
2675 f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
2676 segment0_blkaddr, cp_blkaddr);
2680 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
2682 f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
2683 cp_blkaddr, sit_blkaddr,
2684 segment_count_ckpt << log_blocks_per_seg);
2688 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
2690 f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
2691 sit_blkaddr, nat_blkaddr,
2692 segment_count_sit << log_blocks_per_seg);
2696 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
2698 f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
2699 nat_blkaddr, ssa_blkaddr,
2700 segment_count_nat << log_blocks_per_seg);
2704 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
2706 f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
2707 ssa_blkaddr, main_blkaddr,
2708 segment_count_ssa << log_blocks_per_seg);
2712 if (main_end_blkaddr > seg_end_blkaddr) {
2713 f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
2714 main_blkaddr, seg_end_blkaddr,
2715 segment_count_main << log_blocks_per_seg);
2717 } else if (main_end_blkaddr < seg_end_blkaddr) {
2721 /* fix in-memory information all the time */
2722 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
2723 segment0_blkaddr) >> log_blocks_per_seg);
2725 if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
2726 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2729 err = __f2fs_commit_super(bh, NULL);
2730 res = err ? "failed" : "done";
2732 f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
2733 res, main_blkaddr, seg_end_blkaddr,
2734 segment_count_main << log_blocks_per_seg);
2741 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
2742 struct buffer_head *bh)
2744 block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
2745 block_t total_sections, blocks_per_seg;
2746 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2747 (bh->b_data + F2FS_SUPER_OFFSET);
2748 unsigned int blocksize;
2749 size_t crc_offset = 0;
2752 if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
2753 f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
2754 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
2758 /* Check checksum_offset and crc in superblock */
2759 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
2760 crc_offset = le32_to_cpu(raw_super->checksum_offset);
2762 offsetof(struct f2fs_super_block, crc)) {
2763 f2fs_info(sbi, "Invalid SB checksum offset: %zu",
2765 return -EFSCORRUPTED;
2767 crc = le32_to_cpu(raw_super->crc);
2768 if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
2769 f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
2770 return -EFSCORRUPTED;
2774 /* Currently, support only 4KB page cache size */
2775 if (F2FS_BLKSIZE != PAGE_SIZE) {
2776 f2fs_info(sbi, "Invalid page_cache_size (%lu), supports only 4KB",
2778 return -EFSCORRUPTED;
2781 /* Currently, support only 4KB block size */
2782 blocksize = 1 << le32_to_cpu(raw_super->log_blocksize);
2783 if (blocksize != F2FS_BLKSIZE) {
2784 f2fs_info(sbi, "Invalid blocksize (%u), supports only 4KB",
2786 return -EFSCORRUPTED;
2789 /* check log blocks per segment */
2790 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
2791 f2fs_info(sbi, "Invalid log blocks per segment (%u)",
2792 le32_to_cpu(raw_super->log_blocks_per_seg));
2793 return -EFSCORRUPTED;
2796 /* Currently, support 512/1024/2048/4096 bytes sector size */
2797 if (le32_to_cpu(raw_super->log_sectorsize) >
2798 F2FS_MAX_LOG_SECTOR_SIZE ||
2799 le32_to_cpu(raw_super->log_sectorsize) <
2800 F2FS_MIN_LOG_SECTOR_SIZE) {
2801 f2fs_info(sbi, "Invalid log sectorsize (%u)",
2802 le32_to_cpu(raw_super->log_sectorsize));
2803 return -EFSCORRUPTED;
2805 if (le32_to_cpu(raw_super->log_sectors_per_block) +
2806 le32_to_cpu(raw_super->log_sectorsize) !=
2807 F2FS_MAX_LOG_SECTOR_SIZE) {
2808 f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
2809 le32_to_cpu(raw_super->log_sectors_per_block),
2810 le32_to_cpu(raw_super->log_sectorsize));
2811 return -EFSCORRUPTED;
2814 segment_count = le32_to_cpu(raw_super->segment_count);
2815 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2816 segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
2817 secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
2818 total_sections = le32_to_cpu(raw_super->section_count);
2820 /* blocks_per_seg should be 512, given the above check */
2821 blocks_per_seg = 1 << le32_to_cpu(raw_super->log_blocks_per_seg);
2823 if (segment_count > F2FS_MAX_SEGMENT ||
2824 segment_count < F2FS_MIN_SEGMENTS) {
2825 f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
2826 return -EFSCORRUPTED;
2829 if (total_sections > segment_count_main || total_sections < 1 ||
2830 segs_per_sec > segment_count || !segs_per_sec) {
2831 f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
2832 segment_count, total_sections, segs_per_sec);
2833 return -EFSCORRUPTED;
2836 if (segment_count_main != total_sections * segs_per_sec) {
2837 f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
2838 segment_count_main, total_sections, segs_per_sec);
2839 return -EFSCORRUPTED;
2842 if ((segment_count / segs_per_sec) < total_sections) {
2843 f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
2844 segment_count, segs_per_sec, total_sections);
2845 return -EFSCORRUPTED;
2848 if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
2849 f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
2850 segment_count, le64_to_cpu(raw_super->block_count));
2851 return -EFSCORRUPTED;
2854 if (RDEV(0).path[0]) {
2855 block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
2858 while (i < MAX_DEVICES && RDEV(i).path[0]) {
2859 dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
2862 if (segment_count != dev_seg_count) {
2863 f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
2864 segment_count, dev_seg_count);
2865 return -EFSCORRUPTED;
2868 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
2869 !bdev_is_zoned(sbi->sb->s_bdev)) {
2870 f2fs_info(sbi, "Zoned block device path is missing");
2871 return -EFSCORRUPTED;
2875 if (secs_per_zone > total_sections || !secs_per_zone) {
2876 f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
2877 secs_per_zone, total_sections);
2878 return -EFSCORRUPTED;
2880 if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
2881 raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
2882 (le32_to_cpu(raw_super->extension_count) +
2883 raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
2884 f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
2885 le32_to_cpu(raw_super->extension_count),
2886 raw_super->hot_ext_count,
2887 F2FS_MAX_EXTENSION);
2888 return -EFSCORRUPTED;
2891 if (le32_to_cpu(raw_super->cp_payload) >
2892 (blocks_per_seg - F2FS_CP_PACKS)) {
2893 f2fs_info(sbi, "Insane cp_payload (%u > %u)",
2894 le32_to_cpu(raw_super->cp_payload),
2895 blocks_per_seg - F2FS_CP_PACKS);
2896 return -EFSCORRUPTED;
2899 /* check reserved ino info */
2900 if (le32_to_cpu(raw_super->node_ino) != 1 ||
2901 le32_to_cpu(raw_super->meta_ino) != 2 ||
2902 le32_to_cpu(raw_super->root_ino) != 3) {
2903 f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
2904 le32_to_cpu(raw_super->node_ino),
2905 le32_to_cpu(raw_super->meta_ino),
2906 le32_to_cpu(raw_super->root_ino));
2907 return -EFSCORRUPTED;
2910 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
2911 if (sanity_check_area_boundary(sbi, bh))
2912 return -EFSCORRUPTED;
2917 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
2919 unsigned int total, fsmeta;
2920 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
2921 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
2922 unsigned int ovp_segments, reserved_segments;
2923 unsigned int main_segs, blocks_per_seg;
2924 unsigned int sit_segs, nat_segs;
2925 unsigned int sit_bitmap_size, nat_bitmap_size;
2926 unsigned int log_blocks_per_seg;
2927 unsigned int segment_count_main;
2928 unsigned int cp_pack_start_sum, cp_payload;
2929 block_t user_block_count, valid_user_blocks;
2930 block_t avail_node_count, valid_node_count;
2933 total = le32_to_cpu(raw_super->segment_count);
2934 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
2935 sit_segs = le32_to_cpu(raw_super->segment_count_sit);
2937 nat_segs = le32_to_cpu(raw_super->segment_count_nat);
2939 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
2940 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
2942 if (unlikely(fsmeta >= total))
2945 ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
2946 reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
2948 if (unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
2949 ovp_segments == 0 || reserved_segments == 0)) {
2950 f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
2954 user_block_count = le64_to_cpu(ckpt->user_block_count);
2955 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2956 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2957 if (!user_block_count || user_block_count >=
2958 segment_count_main << log_blocks_per_seg) {
2959 f2fs_err(sbi, "Wrong user_block_count: %u",
2964 valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
2965 if (valid_user_blocks > user_block_count) {
2966 f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
2967 valid_user_blocks, user_block_count);
2971 valid_node_count = le32_to_cpu(ckpt->valid_node_count);
2972 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
2973 if (valid_node_count > avail_node_count) {
2974 f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
2975 valid_node_count, avail_node_count);
2979 main_segs = le32_to_cpu(raw_super->segment_count_main);
2980 blocks_per_seg = sbi->blocks_per_seg;
2982 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
2983 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
2984 le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
2986 for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
2987 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
2988 le32_to_cpu(ckpt->cur_node_segno[j])) {
2989 f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
2991 le32_to_cpu(ckpt->cur_node_segno[i]));
2996 for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
2997 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
2998 le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
3000 for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3001 if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3002 le32_to_cpu(ckpt->cur_data_segno[j])) {
3003 f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3005 le32_to_cpu(ckpt->cur_data_segno[i]));
3010 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3011 for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3012 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3013 le32_to_cpu(ckpt->cur_data_segno[j])) {
3014 f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3016 le32_to_cpu(ckpt->cur_node_segno[i]));
3022 sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3023 nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3025 if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3026 nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3027 f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3028 sit_bitmap_size, nat_bitmap_size);
3032 cp_pack_start_sum = __start_sum_addr(sbi);
3033 cp_payload = __cp_payload(sbi);
3034 if (cp_pack_start_sum < cp_payload + 1 ||
3035 cp_pack_start_sum > blocks_per_seg - 1 -
3036 NR_CURSEG_PERSIST_TYPE) {
3037 f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3042 if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3043 le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3044 f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3045 "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3046 "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3047 le32_to_cpu(ckpt->checksum_offset));
3051 if (unlikely(f2fs_cp_error(sbi))) {
3052 f2fs_err(sbi, "A bug case: need to run fsck");
3058 static void init_sb_info(struct f2fs_sb_info *sbi)
3060 struct f2fs_super_block *raw_super = sbi->raw_super;
3063 sbi->log_sectors_per_block =
3064 le32_to_cpu(raw_super->log_sectors_per_block);
3065 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3066 sbi->blocksize = 1 << sbi->log_blocksize;
3067 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3068 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
3069 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3070 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3071 sbi->total_sections = le32_to_cpu(raw_super->section_count);
3072 sbi->total_node_count =
3073 (le32_to_cpu(raw_super->segment_count_nat) / 2)
3074 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
3075 sbi->root_ino_num = le32_to_cpu(raw_super->root_ino);
3076 sbi->node_ino_num = le32_to_cpu(raw_super->node_ino);
3077 sbi->meta_ino_num = le32_to_cpu(raw_super->meta_ino);
3078 sbi->cur_victim_sec = NULL_SECNO;
3079 sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3080 sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3081 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3082 sbi->migration_granularity = sbi->segs_per_sec;
3084 sbi->dir_level = DEF_DIR_LEVEL;
3085 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3086 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3087 sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3088 sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3089 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3090 sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3091 DEF_UMOUNT_DISCARD_TIMEOUT;
3092 clear_sbi_flag(sbi, SBI_NEED_FSCK);
3094 for (i = 0; i < NR_COUNT_TYPE; i++)
3095 atomic_set(&sbi->nr_pages[i], 0);
3097 for (i = 0; i < META; i++)
3098 atomic_set(&sbi->wb_sync_req[i], 0);
3100 INIT_LIST_HEAD(&sbi->s_list);
3101 mutex_init(&sbi->umount_mutex);
3102 init_rwsem(&sbi->io_order_lock);
3103 spin_lock_init(&sbi->cp_lock);
3105 sbi->dirty_device = 0;
3106 spin_lock_init(&sbi->dev_lock);
3108 init_rwsem(&sbi->sb_lock);
3109 init_rwsem(&sbi->pin_sem);
3112 static int init_percpu_info(struct f2fs_sb_info *sbi)
3116 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3120 err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3123 percpu_counter_destroy(&sbi->alloc_valid_block_count);
3128 #ifdef CONFIG_BLK_DEV_ZONED
3130 struct f2fs_report_zones_args {
3131 struct f2fs_dev_info *dev;
3132 bool zone_cap_mismatch;
3135 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3138 struct f2fs_report_zones_args *rz_args = data;
3140 if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3143 set_bit(idx, rz_args->dev->blkz_seq);
3144 rz_args->dev->zone_capacity_blocks[idx] = zone->capacity >>
3145 F2FS_LOG_SECTORS_PER_BLOCK;
3146 if (zone->len != zone->capacity && !rz_args->zone_cap_mismatch)
3147 rz_args->zone_cap_mismatch = true;
3152 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3154 struct block_device *bdev = FDEV(devi).bdev;
3155 sector_t nr_sectors = bdev->bd_part->nr_sects;
3156 struct f2fs_report_zones_args rep_zone_arg;
3159 if (!f2fs_sb_has_blkzoned(sbi))
3162 if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3163 SECTOR_TO_BLOCK(bdev_zone_sectors(bdev)))
3165 sbi->blocks_per_blkz = SECTOR_TO_BLOCK(bdev_zone_sectors(bdev));
3166 if (sbi->log_blocks_per_blkz && sbi->log_blocks_per_blkz !=
3167 __ilog2_u32(sbi->blocks_per_blkz))
3169 sbi->log_blocks_per_blkz = __ilog2_u32(sbi->blocks_per_blkz);
3170 FDEV(devi).nr_blkz = SECTOR_TO_BLOCK(nr_sectors) >>
3171 sbi->log_blocks_per_blkz;
3172 if (nr_sectors & (bdev_zone_sectors(bdev) - 1))
3173 FDEV(devi).nr_blkz++;
3175 FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3176 BITS_TO_LONGS(FDEV(devi).nr_blkz)
3177 * sizeof(unsigned long),
3179 if (!FDEV(devi).blkz_seq)
3182 /* Get block zones type and zone-capacity */
3183 FDEV(devi).zone_capacity_blocks = f2fs_kzalloc(sbi,
3184 FDEV(devi).nr_blkz * sizeof(block_t),
3186 if (!FDEV(devi).zone_capacity_blocks)
3189 rep_zone_arg.dev = &FDEV(devi);
3190 rep_zone_arg.zone_cap_mismatch = false;
3192 ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3197 if (!rep_zone_arg.zone_cap_mismatch) {
3198 kfree(FDEV(devi).zone_capacity_blocks);
3199 FDEV(devi).zone_capacity_blocks = NULL;
3207 * Read f2fs raw super block.
3208 * Because we have two copies of super block, so read both of them
3209 * to get the first valid one. If any one of them is broken, we pass
3210 * them recovery flag back to the caller.
3212 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3213 struct f2fs_super_block **raw_super,
3214 int *valid_super_block, int *recovery)
3216 struct super_block *sb = sbi->sb;
3218 struct buffer_head *bh;
3219 struct f2fs_super_block *super;
3222 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3226 for (block = 0; block < 2; block++) {
3227 bh = sb_bread(sb, block);
3229 f2fs_err(sbi, "Unable to read %dth superblock",
3236 /* sanity checking of raw super */
3237 err = sanity_check_raw_super(sbi, bh);
3239 f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3247 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3249 *valid_super_block = block;
3255 /* No valid superblock */
3264 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
3266 struct buffer_head *bh;
3270 if ((recover && f2fs_readonly(sbi->sb)) ||
3271 bdev_read_only(sbi->sb->s_bdev)) {
3272 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3276 /* we should update superblock crc here */
3277 if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
3278 crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
3279 offsetof(struct f2fs_super_block, crc));
3280 F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
3283 /* write back-up superblock first */
3284 bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
3287 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3290 /* if we are in recovery path, skip writing valid superblock */
3294 /* write current valid superblock */
3295 bh = sb_bread(sbi->sb, sbi->valid_super_block);
3298 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3303 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
3305 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3306 unsigned int max_devices = MAX_DEVICES;
3309 /* Initialize single device information */
3310 if (!RDEV(0).path[0]) {
3311 if (!bdev_is_zoned(sbi->sb->s_bdev))
3317 * Initialize multiple devices information, or single
3318 * zoned block device information.
3320 sbi->devs = f2fs_kzalloc(sbi,
3321 array_size(max_devices,
3322 sizeof(struct f2fs_dev_info)),
3327 for (i = 0; i < max_devices; i++) {
3329 if (i > 0 && !RDEV(i).path[0])
3332 if (max_devices == 1) {
3333 /* Single zoned block device mount */
3335 blkdev_get_by_dev(sbi->sb->s_bdev->bd_dev,
3336 sbi->sb->s_mode, sbi->sb->s_type);
3338 /* Multi-device mount */
3339 memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
3340 FDEV(i).total_segments =
3341 le32_to_cpu(RDEV(i).total_segments);
3343 FDEV(i).start_blk = 0;
3344 FDEV(i).end_blk = FDEV(i).start_blk +
3345 (FDEV(i).total_segments <<
3346 sbi->log_blocks_per_seg) - 1 +
3347 le32_to_cpu(raw_super->segment0_blkaddr);
3349 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
3350 FDEV(i).end_blk = FDEV(i).start_blk +
3351 (FDEV(i).total_segments <<
3352 sbi->log_blocks_per_seg) - 1;
3354 FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
3355 sbi->sb->s_mode, sbi->sb->s_type);
3357 if (IS_ERR(FDEV(i).bdev))
3358 return PTR_ERR(FDEV(i).bdev);
3360 /* to release errored devices */
3361 sbi->s_ndevs = i + 1;
3363 #ifdef CONFIG_BLK_DEV_ZONED
3364 if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
3365 !f2fs_sb_has_blkzoned(sbi)) {
3366 f2fs_err(sbi, "Zoned block device feature not enabled\n");
3369 if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
3370 if (init_blkz_info(sbi, i)) {
3371 f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
3374 if (max_devices == 1)
3376 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
3378 FDEV(i).total_segments,
3379 FDEV(i).start_blk, FDEV(i).end_blk,
3380 bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
3381 "Host-aware" : "Host-managed");
3385 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
3387 FDEV(i).total_segments,
3388 FDEV(i).start_blk, FDEV(i).end_blk);
3391 "IO Block Size: %8d KB", F2FS_IO_SIZE_KB(sbi));
3395 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
3397 #ifdef CONFIG_UNICODE
3398 if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
3399 const struct f2fs_sb_encodings *encoding_info;
3400 struct unicode_map *encoding;
3401 __u16 encoding_flags;
3403 if (f2fs_sb_has_encrypt(sbi)) {
3405 "Can't mount with encoding and encryption");
3409 if (f2fs_sb_read_encoding(sbi->raw_super, &encoding_info,
3412 "Encoding requested by superblock is unknown");
3416 encoding = utf8_load(encoding_info->version);
3417 if (IS_ERR(encoding)) {
3419 "can't mount with superblock charset: %s-%s "
3420 "not supported by the kernel. flags: 0x%x.",
3421 encoding_info->name, encoding_info->version,
3423 return PTR_ERR(encoding);
3425 f2fs_info(sbi, "Using encoding defined by superblock: "
3426 "%s-%s with flags 0x%hx", encoding_info->name,
3427 encoding_info->version?:"\b", encoding_flags);
3429 sbi->sb->s_encoding = encoding;
3430 sbi->sb->s_encoding_flags = encoding_flags;
3431 sbi->sb->s_d_op = &f2fs_dentry_ops;
3434 if (f2fs_sb_has_casefold(sbi)) {
3435 f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
3442 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
3444 struct f2fs_sm_info *sm_i = SM_I(sbi);
3446 /* adjust parameters according to the volume size */
3447 if (sm_i->main_segments <= SMALL_VOLUME_SEGMENTS) {
3448 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
3449 sm_i->dcc_info->discard_granularity = 1;
3450 sm_i->ipu_policy = 1 << F2FS_IPU_FORCE;
3453 sbi->readdir_ra = 1;
3456 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
3458 struct f2fs_sb_info *sbi;
3459 struct f2fs_super_block *raw_super;
3462 bool skip_recovery = false, need_fsck = false;
3463 char *options = NULL;
3464 int recovery, i, valid_super_block;
3465 struct curseg_info *seg_i;
3471 valid_super_block = -1;
3474 /* allocate memory for f2fs-specific super block info */
3475 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
3481 /* Load the checksum driver */
3482 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
3483 if (IS_ERR(sbi->s_chksum_driver)) {
3484 f2fs_err(sbi, "Cannot load crc32 driver.");
3485 err = PTR_ERR(sbi->s_chksum_driver);
3486 sbi->s_chksum_driver = NULL;
3490 /* set a block size */
3491 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
3492 f2fs_err(sbi, "unable to set blocksize");
3496 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
3501 sb->s_fs_info = sbi;
3502 sbi->raw_super = raw_super;
3504 /* precompute checksum seed for metadata */
3505 if (f2fs_sb_has_inode_chksum(sbi))
3506 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
3507 sizeof(raw_super->uuid));
3509 default_options(sbi);
3510 /* parse mount options */
3511 options = kstrdup((const char *)data, GFP_KERNEL);
3512 if (data && !options) {
3517 err = parse_options(sb, options, false);
3521 sbi->max_file_blocks = max_file_blocks();
3522 sb->s_maxbytes = sbi->max_file_blocks <<
3523 le32_to_cpu(raw_super->log_blocksize);
3524 sb->s_max_links = F2FS_LINK_MAX;
3526 err = f2fs_setup_casefold(sbi);
3531 sb->dq_op = &f2fs_quota_operations;
3532 sb->s_qcop = &f2fs_quotactl_ops;
3533 sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
3535 if (f2fs_sb_has_quota_ino(sbi)) {
3536 for (i = 0; i < MAXQUOTAS; i++) {
3537 if (f2fs_qf_ino(sbi->sb, i))
3538 sbi->nquota_files++;
3543 sb->s_op = &f2fs_sops;
3544 #ifdef CONFIG_FS_ENCRYPTION
3545 sb->s_cop = &f2fs_cryptops;
3547 #ifdef CONFIG_FS_VERITY
3548 sb->s_vop = &f2fs_verityops;
3550 sb->s_xattr = f2fs_xattr_handlers;
3551 sb->s_export_op = &f2fs_export_ops;
3552 sb->s_magic = F2FS_SUPER_MAGIC;
3553 sb->s_time_gran = 1;
3554 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
3555 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
3556 memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
3557 sb->s_iflags |= SB_I_CGROUPWB;
3559 /* init f2fs-specific super block info */
3560 sbi->valid_super_block = valid_super_block;
3561 init_rwsem(&sbi->gc_lock);
3562 mutex_init(&sbi->writepages);
3563 mutex_init(&sbi->cp_mutex);
3564 init_rwsem(&sbi->node_write);
3565 init_rwsem(&sbi->node_change);
3567 /* disallow all the data/node/meta page writes */
3568 set_sbi_flag(sbi, SBI_POR_DOING);
3569 spin_lock_init(&sbi->stat_lock);
3571 /* init iostat info */
3572 spin_lock_init(&sbi->iostat_lock);
3573 sbi->iostat_enable = false;
3574 sbi->iostat_period_ms = DEFAULT_IOSTAT_PERIOD_MS;
3576 for (i = 0; i < NR_PAGE_TYPE; i++) {
3577 int n = (i == META) ? 1: NR_TEMP_TYPE;
3583 sizeof(struct f2fs_bio_info)),
3585 if (!sbi->write_io[i]) {
3590 for (j = HOT; j < n; j++) {
3591 init_rwsem(&sbi->write_io[i][j].io_rwsem);
3592 sbi->write_io[i][j].sbi = sbi;
3593 sbi->write_io[i][j].bio = NULL;
3594 spin_lock_init(&sbi->write_io[i][j].io_lock);
3595 INIT_LIST_HEAD(&sbi->write_io[i][j].io_list);
3596 INIT_LIST_HEAD(&sbi->write_io[i][j].bio_list);
3597 init_rwsem(&sbi->write_io[i][j].bio_list_lock);
3601 init_rwsem(&sbi->cp_rwsem);
3602 init_rwsem(&sbi->quota_sem);
3603 init_waitqueue_head(&sbi->cp_wait);
3606 err = init_percpu_info(sbi);
3610 if (F2FS_IO_ALIGNED(sbi)) {
3611 sbi->write_io_dummy =
3612 mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
3613 if (!sbi->write_io_dummy) {
3619 /* init per sbi slab cache */
3620 err = f2fs_init_xattr_caches(sbi);
3623 err = f2fs_init_page_array_cache(sbi);
3625 goto free_xattr_cache;
3627 /* get an inode for meta space */
3628 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
3629 if (IS_ERR(sbi->meta_inode)) {
3630 f2fs_err(sbi, "Failed to read F2FS meta data inode");
3631 err = PTR_ERR(sbi->meta_inode);
3632 goto free_page_array_cache;
3635 err = f2fs_get_valid_checkpoint(sbi);
3637 f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
3638 goto free_meta_inode;
3641 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
3642 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3643 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
3644 set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
3645 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
3648 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
3649 set_sbi_flag(sbi, SBI_NEED_FSCK);
3651 /* Initialize device list */
3652 err = f2fs_scan_devices(sbi);
3654 f2fs_err(sbi, "Failed to find devices");
3658 err = f2fs_init_post_read_wq(sbi);
3660 f2fs_err(sbi, "Failed to initialize post read workqueue");
3664 sbi->total_valid_node_count =
3665 le32_to_cpu(sbi->ckpt->valid_node_count);
3666 percpu_counter_set(&sbi->total_valid_inode_count,
3667 le32_to_cpu(sbi->ckpt->valid_inode_count));
3668 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
3669 sbi->total_valid_block_count =
3670 le64_to_cpu(sbi->ckpt->valid_block_count);
3671 sbi->last_valid_block_count = sbi->total_valid_block_count;
3672 sbi->reserved_blocks = 0;
3673 sbi->current_reserved_blocks = 0;
3674 limit_reserve_root(sbi);
3675 adjust_unusable_cap_perc(sbi);
3677 for (i = 0; i < NR_INODE_TYPE; i++) {
3678 INIT_LIST_HEAD(&sbi->inode_list[i]);
3679 spin_lock_init(&sbi->inode_lock[i]);
3681 mutex_init(&sbi->flush_lock);
3683 f2fs_init_extent_cache_info(sbi);
3685 f2fs_init_ino_entry_info(sbi);
3687 f2fs_init_fsync_node_info(sbi);
3689 /* setup f2fs internal modules */
3690 err = f2fs_build_segment_manager(sbi);
3692 f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
3696 err = f2fs_build_node_manager(sbi);
3698 f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
3703 /* For write statistics */
3704 if (sb->s_bdev->bd_part)
3705 sbi->sectors_written_start =
3706 (u64)part_stat_read(sb->s_bdev->bd_part,
3707 sectors[STAT_WRITE]);
3709 /* Read accumulated write IO statistics if exists */
3710 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
3711 if (__exist_node_summaries(sbi))
3712 sbi->kbytes_written =
3713 le64_to_cpu(seg_i->journal->info.kbytes_written);
3715 f2fs_build_gc_manager(sbi);
3717 err = f2fs_build_stats(sbi);
3721 /* get an inode for node space */
3722 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
3723 if (IS_ERR(sbi->node_inode)) {
3724 f2fs_err(sbi, "Failed to read node inode");
3725 err = PTR_ERR(sbi->node_inode);
3729 /* read root inode and dentry */
3730 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
3732 f2fs_err(sbi, "Failed to read root inode");
3733 err = PTR_ERR(root);
3734 goto free_node_inode;
3736 if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
3737 !root->i_size || !root->i_nlink) {
3740 goto free_node_inode;
3743 sb->s_root = d_make_root(root); /* allocate root dentry */
3746 goto free_node_inode;
3749 err = f2fs_register_sysfs(sbi);
3751 goto free_root_inode;
3754 /* Enable quota usage during mount */
3755 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
3756 err = f2fs_enable_quotas(sb);
3758 f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
3761 /* if there are any orphan inodes, free them */
3762 err = f2fs_recover_orphan_inodes(sbi);
3766 if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
3767 goto reset_checkpoint;
3769 /* recover fsynced data */
3770 if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
3771 !test_opt(sbi, NORECOVERY)) {
3773 * mount should be failed, when device has readonly mode, and
3774 * previous checkpoint was not done by clean system shutdown.
3776 if (f2fs_hw_is_readonly(sbi)) {
3777 if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
3779 f2fs_err(sbi, "Need to recover fsync data, but write access unavailable");
3782 f2fs_info(sbi, "write access unavailable, skipping recovery");
3783 goto reset_checkpoint;
3787 set_sbi_flag(sbi, SBI_NEED_FSCK);
3790 goto reset_checkpoint;
3792 err = f2fs_recover_fsync_data(sbi, false);
3795 skip_recovery = true;
3797 f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
3802 err = f2fs_recover_fsync_data(sbi, true);
3804 if (!f2fs_readonly(sb) && err > 0) {
3806 f2fs_err(sbi, "Need to recover fsync data");
3812 * If the f2fs is not readonly and fsync data recovery succeeds,
3813 * check zoned block devices' write pointer consistency.
3815 if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
3816 err = f2fs_check_write_pointer(sbi);
3822 f2fs_init_inmem_curseg(sbi);
3824 /* f2fs_recover_fsync_data() cleared this already */
3825 clear_sbi_flag(sbi, SBI_POR_DOING);
3827 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
3828 err = f2fs_disable_checkpoint(sbi);
3830 goto sync_free_meta;
3831 } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
3832 f2fs_enable_checkpoint(sbi);
3836 * If filesystem is not mounted as read-only then
3837 * do start the gc_thread.
3839 if (F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF && !f2fs_readonly(sb)) {
3840 /* After POR, we can run background GC thread.*/
3841 err = f2fs_start_gc_thread(sbi);
3843 goto sync_free_meta;
3847 /* recover broken superblock */
3849 err = f2fs_commit_super(sbi, true);
3850 f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
3851 sbi->valid_super_block ? 1 : 2, err);
3854 f2fs_join_shrinker(sbi);
3856 f2fs_tuning_parameters(sbi);
3858 f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
3859 cur_cp_version(F2FS_CKPT(sbi)));
3860 f2fs_update_time(sbi, CP_TIME);
3861 f2fs_update_time(sbi, REQ_TIME);
3862 clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
3866 /* safe to flush all the data */
3867 sync_filesystem(sbi->sb);
3872 f2fs_truncate_quota_inode_pages(sb);
3873 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
3874 f2fs_quota_off_umount(sbi->sb);
3877 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
3878 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
3879 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
3880 * falls into an infinite loop in f2fs_sync_meta_pages().
3882 truncate_inode_pages_final(META_MAPPING(sbi));
3883 /* evict some inodes being cached by GC */
3885 f2fs_unregister_sysfs(sbi);
3890 f2fs_release_ino_entry(sbi, true);
3891 truncate_inode_pages_final(NODE_MAPPING(sbi));
3892 iput(sbi->node_inode);
3893 sbi->node_inode = NULL;
3895 f2fs_destroy_stats(sbi);
3897 f2fs_destroy_node_manager(sbi);
3899 f2fs_destroy_segment_manager(sbi);
3900 f2fs_destroy_post_read_wq(sbi);
3902 destroy_device_list(sbi);
3905 make_bad_inode(sbi->meta_inode);
3906 iput(sbi->meta_inode);
3907 sbi->meta_inode = NULL;
3908 free_page_array_cache:
3909 f2fs_destroy_page_array_cache(sbi);
3911 f2fs_destroy_xattr_caches(sbi);
3913 mempool_destroy(sbi->write_io_dummy);
3915 destroy_percpu_info(sbi);
3917 for (i = 0; i < NR_PAGE_TYPE; i++)
3918 kvfree(sbi->write_io[i]);
3920 #ifdef CONFIG_UNICODE
3921 utf8_unload(sb->s_encoding);
3925 for (i = 0; i < MAXQUOTAS; i++)
3926 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
3928 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
3933 if (sbi->s_chksum_driver)
3934 crypto_free_shash(sbi->s_chksum_driver);
3937 /* give only one another chance */
3938 if (retry_cnt > 0 && skip_recovery) {
3940 shrink_dcache_sb(sb);
3946 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
3947 const char *dev_name, void *data)
3949 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
3952 static void kill_f2fs_super(struct super_block *sb)
3955 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3957 set_sbi_flag(sbi, SBI_IS_CLOSE);
3958 f2fs_stop_gc_thread(sbi);
3959 f2fs_stop_discard_thread(sbi);
3961 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
3962 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
3963 struct cp_control cpc = {
3964 .reason = CP_UMOUNT,
3966 f2fs_write_checkpoint(sbi, &cpc);
3969 if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
3970 sb->s_flags &= ~SB_RDONLY;
3972 kill_block_super(sb);
3975 static struct file_system_type f2fs_fs_type = {
3976 .owner = THIS_MODULE,
3978 .mount = f2fs_mount,
3979 .kill_sb = kill_f2fs_super,
3980 .fs_flags = FS_REQUIRES_DEV,
3982 MODULE_ALIAS_FS("f2fs");
3984 static int __init init_inodecache(void)
3986 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
3987 sizeof(struct f2fs_inode_info), 0,
3988 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
3989 if (!f2fs_inode_cachep)
3994 static void destroy_inodecache(void)
3997 * Make sure all delayed rcu free inodes are flushed before we
4001 kmem_cache_destroy(f2fs_inode_cachep);
4004 static int __init init_f2fs_fs(void)
4008 if (PAGE_SIZE != F2FS_BLKSIZE) {
4009 printk("F2FS not supported on PAGE_SIZE(%lu) != %d\n",
4010 PAGE_SIZE, F2FS_BLKSIZE);
4014 f2fs_build_trace_ios();
4016 err = init_inodecache();
4019 err = f2fs_create_node_manager_caches();
4021 goto free_inodecache;
4022 err = f2fs_create_segment_manager_caches();
4024 goto free_node_manager_caches;
4025 err = f2fs_create_checkpoint_caches();
4027 goto free_segment_manager_caches;
4028 err = f2fs_create_extent_cache();
4030 goto free_checkpoint_caches;
4031 err = f2fs_create_garbage_collection_cache();
4033 goto free_extent_cache;
4034 err = f2fs_init_sysfs();
4036 goto free_garbage_collection_cache;
4037 err = register_shrinker(&f2fs_shrinker_info);
4040 err = register_filesystem(&f2fs_fs_type);
4043 f2fs_create_root_stats();
4044 err = f2fs_init_post_read_processing();
4046 goto free_root_stats;
4047 err = f2fs_init_bio_entry_cache();
4049 goto free_post_read;
4050 err = f2fs_init_bioset();
4052 goto free_bio_enrty_cache;
4053 err = f2fs_init_compress_mempool();
4056 err = f2fs_init_compress_cache();
4058 goto free_compress_mempool;
4060 free_compress_mempool:
4061 f2fs_destroy_compress_mempool();
4063 f2fs_destroy_bioset();
4064 free_bio_enrty_cache:
4065 f2fs_destroy_bio_entry_cache();
4067 f2fs_destroy_post_read_processing();
4069 f2fs_destroy_root_stats();
4070 unregister_filesystem(&f2fs_fs_type);
4072 unregister_shrinker(&f2fs_shrinker_info);
4075 free_garbage_collection_cache:
4076 f2fs_destroy_garbage_collection_cache();
4078 f2fs_destroy_extent_cache();
4079 free_checkpoint_caches:
4080 f2fs_destroy_checkpoint_caches();
4081 free_segment_manager_caches:
4082 f2fs_destroy_segment_manager_caches();
4083 free_node_manager_caches:
4084 f2fs_destroy_node_manager_caches();
4086 destroy_inodecache();
4091 static void __exit exit_f2fs_fs(void)
4093 f2fs_destroy_compress_cache();
4094 f2fs_destroy_compress_mempool();
4095 f2fs_destroy_bioset();
4096 f2fs_destroy_bio_entry_cache();
4097 f2fs_destroy_post_read_processing();
4098 f2fs_destroy_root_stats();
4099 unregister_filesystem(&f2fs_fs_type);
4100 unregister_shrinker(&f2fs_shrinker_info);
4102 f2fs_destroy_garbage_collection_cache();
4103 f2fs_destroy_extent_cache();
4104 f2fs_destroy_checkpoint_caches();
4105 f2fs_destroy_segment_manager_caches();
4106 f2fs_destroy_node_manager_caches();
4107 destroy_inodecache();
4108 f2fs_destroy_trace_ios();
4111 module_init(init_f2fs_fs)
4112 module_exit(exit_f2fs_fs)
4114 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
4115 MODULE_DESCRIPTION("Flash Friendly File System");
4116 MODULE_LICENSE("GPL");