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,
155 static match_table_t f2fs_tokens = {
156 {Opt_gc_background, "background_gc=%s"},
157 {Opt_disable_roll_forward, "disable_roll_forward"},
158 {Opt_norecovery, "norecovery"},
159 {Opt_discard, "discard"},
160 {Opt_nodiscard, "nodiscard"},
161 {Opt_noheap, "no_heap"},
163 {Opt_user_xattr, "user_xattr"},
164 {Opt_nouser_xattr, "nouser_xattr"},
166 {Opt_noacl, "noacl"},
167 {Opt_active_logs, "active_logs=%u"},
168 {Opt_disable_ext_identify, "disable_ext_identify"},
169 {Opt_inline_xattr, "inline_xattr"},
170 {Opt_noinline_xattr, "noinline_xattr"},
171 {Opt_inline_xattr_size, "inline_xattr_size=%u"},
172 {Opt_inline_data, "inline_data"},
173 {Opt_inline_dentry, "inline_dentry"},
174 {Opt_noinline_dentry, "noinline_dentry"},
175 {Opt_flush_merge, "flush_merge"},
176 {Opt_noflush_merge, "noflush_merge"},
177 {Opt_nobarrier, "nobarrier"},
178 {Opt_fastboot, "fastboot"},
179 {Opt_extent_cache, "extent_cache"},
180 {Opt_noextent_cache, "noextent_cache"},
181 {Opt_noinline_data, "noinline_data"},
182 {Opt_data_flush, "data_flush"},
183 {Opt_reserve_root, "reserve_root=%u"},
184 {Opt_resgid, "resgid=%u"},
185 {Opt_resuid, "resuid=%u"},
186 {Opt_mode, "mode=%s"},
187 {Opt_io_size_bits, "io_bits=%u"},
188 {Opt_fault_injection, "fault_injection=%u"},
189 {Opt_fault_type, "fault_type=%u"},
190 {Opt_lazytime, "lazytime"},
191 {Opt_nolazytime, "nolazytime"},
192 {Opt_quota, "quota"},
193 {Opt_noquota, "noquota"},
194 {Opt_usrquota, "usrquota"},
195 {Opt_grpquota, "grpquota"},
196 {Opt_prjquota, "prjquota"},
197 {Opt_usrjquota, "usrjquota=%s"},
198 {Opt_grpjquota, "grpjquota=%s"},
199 {Opt_prjjquota, "prjjquota=%s"},
200 {Opt_offusrjquota, "usrjquota="},
201 {Opt_offgrpjquota, "grpjquota="},
202 {Opt_offprjjquota, "prjjquota="},
203 {Opt_jqfmt_vfsold, "jqfmt=vfsold"},
204 {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
205 {Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
206 {Opt_whint, "whint_mode=%s"},
207 {Opt_alloc, "alloc_mode=%s"},
208 {Opt_fsync, "fsync_mode=%s"},
209 {Opt_test_dummy_encryption, "test_dummy_encryption=%s"},
210 {Opt_test_dummy_encryption, "test_dummy_encryption"},
211 {Opt_inlinecrypt, "inlinecrypt"},
212 {Opt_checkpoint_disable, "checkpoint=disable"},
213 {Opt_checkpoint_disable_cap, "checkpoint=disable:%u"},
214 {Opt_checkpoint_disable_cap_perc, "checkpoint=disable:%u%%"},
215 {Opt_checkpoint_enable, "checkpoint=enable"},
216 {Opt_compress_algorithm, "compress_algorithm=%s"},
217 {Opt_compress_log_size, "compress_log_size=%u"},
218 {Opt_compress_extension, "compress_extension=%s"},
219 {Opt_compress_chksum, "compress_chksum"},
220 {Opt_compress_mode, "compress_mode=%s"},
225 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...)
227 struct va_format vaf;
233 level = printk_get_level(fmt);
234 vaf.fmt = printk_skip_level(fmt);
236 printk("%c%cF2FS-fs (%s): %pV\n",
237 KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
242 #ifdef CONFIG_UNICODE
243 static const struct f2fs_sb_encodings {
247 } f2fs_sb_encoding_map[] = {
248 {F2FS_ENC_UTF8_12_1, "utf8", "12.1.0"},
251 static int f2fs_sb_read_encoding(const struct f2fs_super_block *sb,
252 const struct f2fs_sb_encodings **encoding,
255 __u16 magic = le16_to_cpu(sb->s_encoding);
258 for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
259 if (magic == f2fs_sb_encoding_map[i].magic)
262 if (i >= ARRAY_SIZE(f2fs_sb_encoding_map))
265 *encoding = &f2fs_sb_encoding_map[i];
266 *flags = le16_to_cpu(sb->s_encoding_flags);
272 static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
274 block_t limit = min((sbi->user_block_count >> 3),
275 sbi->user_block_count - sbi->reserved_blocks);
278 if (test_opt(sbi, RESERVE_ROOT) &&
279 F2FS_OPTION(sbi).root_reserved_blocks > limit) {
280 F2FS_OPTION(sbi).root_reserved_blocks = limit;
281 f2fs_info(sbi, "Reduce reserved blocks for root = %u",
282 F2FS_OPTION(sbi).root_reserved_blocks);
284 if (!test_opt(sbi, RESERVE_ROOT) &&
285 (!uid_eq(F2FS_OPTION(sbi).s_resuid,
286 make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
287 !gid_eq(F2FS_OPTION(sbi).s_resgid,
288 make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
289 f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
290 from_kuid_munged(&init_user_ns,
291 F2FS_OPTION(sbi).s_resuid),
292 from_kgid_munged(&init_user_ns,
293 F2FS_OPTION(sbi).s_resgid));
296 static inline int adjust_reserved_segment(struct f2fs_sb_info *sbi)
298 unsigned int sec_blks = sbi->blocks_per_seg * sbi->segs_per_sec;
299 unsigned int avg_vblocks;
300 unsigned int wanted_reserved_segments;
301 block_t avail_user_block_count;
303 if (!F2FS_IO_ALIGNED(sbi))
306 /* average valid block count in section in worst case */
307 avg_vblocks = sec_blks / F2FS_IO_SIZE(sbi);
310 * we need enough free space when migrating one section in worst case
312 wanted_reserved_segments = (F2FS_IO_SIZE(sbi) / avg_vblocks) *
313 reserved_segments(sbi);
314 wanted_reserved_segments -= reserved_segments(sbi);
316 avail_user_block_count = sbi->user_block_count -
317 sbi->current_reserved_blocks -
318 F2FS_OPTION(sbi).root_reserved_blocks;
320 if (wanted_reserved_segments * sbi->blocks_per_seg >
321 avail_user_block_count) {
322 f2fs_err(sbi, "IO align feature can't grab additional reserved segment: %u, available segments: %u",
323 wanted_reserved_segments,
324 avail_user_block_count >> sbi->log_blocks_per_seg);
328 SM_I(sbi)->additional_reserved_segments = wanted_reserved_segments;
330 f2fs_info(sbi, "IO align feature needs additional reserved segment: %u",
331 wanted_reserved_segments);
336 static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
338 if (!F2FS_OPTION(sbi).unusable_cap_perc)
341 if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
342 F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
344 F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
345 F2FS_OPTION(sbi).unusable_cap_perc;
347 f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
348 F2FS_OPTION(sbi).unusable_cap,
349 F2FS_OPTION(sbi).unusable_cap_perc);
352 static void init_once(void *foo)
354 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
356 inode_init_once(&fi->vfs_inode);
360 static const char * const quotatypes[] = INITQFNAMES;
361 #define QTYPE2NAME(t) (quotatypes[t])
362 static int f2fs_set_qf_name(struct super_block *sb, int qtype,
365 struct f2fs_sb_info *sbi = F2FS_SB(sb);
369 if (sb_any_quota_loaded(sb) && !F2FS_OPTION(sbi).s_qf_names[qtype]) {
370 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
373 if (f2fs_sb_has_quota_ino(sbi)) {
374 f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
378 qname = match_strdup(args);
380 f2fs_err(sbi, "Not enough memory for storing quotafile name");
383 if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
384 if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
387 f2fs_err(sbi, "%s quota file already specified",
391 if (strchr(qname, '/')) {
392 f2fs_err(sbi, "quotafile must be on filesystem root");
395 F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
403 static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
405 struct f2fs_sb_info *sbi = F2FS_SB(sb);
407 if (sb_any_quota_loaded(sb) && F2FS_OPTION(sbi).s_qf_names[qtype]) {
408 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
411 kfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
412 F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
416 static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
419 * We do the test below only for project quotas. 'usrquota' and
420 * 'grpquota' mount options are allowed even without quota feature
421 * to support legacy quotas in quota files.
423 if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi)) {
424 f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
427 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
428 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
429 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) {
430 if (test_opt(sbi, USRQUOTA) &&
431 F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
432 clear_opt(sbi, USRQUOTA);
434 if (test_opt(sbi, GRPQUOTA) &&
435 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
436 clear_opt(sbi, GRPQUOTA);
438 if (test_opt(sbi, PRJQUOTA) &&
439 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
440 clear_opt(sbi, PRJQUOTA);
442 if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
443 test_opt(sbi, PRJQUOTA)) {
444 f2fs_err(sbi, "old and new quota format mixing");
448 if (!F2FS_OPTION(sbi).s_jquota_fmt) {
449 f2fs_err(sbi, "journaled quota format not specified");
454 if (f2fs_sb_has_quota_ino(sbi) && F2FS_OPTION(sbi).s_jquota_fmt) {
455 f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
456 F2FS_OPTION(sbi).s_jquota_fmt = 0;
462 static int f2fs_set_test_dummy_encryption(struct super_block *sb,
464 const substring_t *arg,
467 struct f2fs_sb_info *sbi = F2FS_SB(sb);
468 #ifdef CONFIG_FS_ENCRYPTION
471 if (!f2fs_sb_has_encrypt(sbi)) {
472 f2fs_err(sbi, "Encrypt feature is off");
477 * This mount option is just for testing, and it's not worthwhile to
478 * implement the extra complexity (e.g. RCU protection) that would be
479 * needed to allow it to be set or changed during remount. We do allow
480 * it to be specified during remount, but only if there is no change.
482 if (is_remount && !F2FS_OPTION(sbi).dummy_enc_policy.policy) {
483 f2fs_warn(sbi, "Can't set test_dummy_encryption on remount");
486 err = fscrypt_set_test_dummy_encryption(
487 sb, arg->from, &F2FS_OPTION(sbi).dummy_enc_policy);
491 "Can't change test_dummy_encryption on remount");
492 else if (err == -EINVAL)
493 f2fs_warn(sbi, "Value of option \"%s\" is unrecognized",
496 f2fs_warn(sbi, "Error processing option \"%s\" [%d]",
500 f2fs_warn(sbi, "Test dummy encryption mode enabled");
502 f2fs_warn(sbi, "Test dummy encryption mount option ignored");
507 static int parse_options(struct super_block *sb, char *options, bool is_remount)
509 struct f2fs_sb_info *sbi = F2FS_SB(sb);
510 substring_t args[MAX_OPT_ARGS];
511 #ifdef CONFIG_F2FS_FS_COMPRESSION
512 unsigned char (*ext)[F2FS_EXTENSION_LEN];
524 while ((p = strsep(&options, ",")) != NULL) {
529 * Initialize args struct so we know whether arg was
530 * found; some options take optional arguments.
532 args[0].to = args[0].from = NULL;
533 token = match_token(p, f2fs_tokens, args);
536 case Opt_gc_background:
537 name = match_strdup(&args[0]);
541 if (!strcmp(name, "on")) {
542 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
543 } else if (!strcmp(name, "off")) {
544 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_OFF;
545 } else if (!strcmp(name, "sync")) {
546 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_SYNC;
553 case Opt_disable_roll_forward:
554 set_opt(sbi, DISABLE_ROLL_FORWARD);
557 /* this option mounts f2fs with ro */
558 set_opt(sbi, NORECOVERY);
559 if (!f2fs_readonly(sb))
563 set_opt(sbi, DISCARD);
566 if (f2fs_sb_has_blkzoned(sbi)) {
567 f2fs_warn(sbi, "discard is required for zoned block devices");
570 clear_opt(sbi, DISCARD);
573 set_opt(sbi, NOHEAP);
576 clear_opt(sbi, NOHEAP);
578 #ifdef CONFIG_F2FS_FS_XATTR
580 set_opt(sbi, XATTR_USER);
582 case Opt_nouser_xattr:
583 clear_opt(sbi, XATTR_USER);
585 case Opt_inline_xattr:
586 set_opt(sbi, INLINE_XATTR);
588 case Opt_noinline_xattr:
589 clear_opt(sbi, INLINE_XATTR);
591 case Opt_inline_xattr_size:
592 if (args->from && match_int(args, &arg))
594 set_opt(sbi, INLINE_XATTR_SIZE);
595 F2FS_OPTION(sbi).inline_xattr_size = arg;
599 f2fs_info(sbi, "user_xattr options not supported");
601 case Opt_nouser_xattr:
602 f2fs_info(sbi, "nouser_xattr options not supported");
604 case Opt_inline_xattr:
605 f2fs_info(sbi, "inline_xattr options not supported");
607 case Opt_noinline_xattr:
608 f2fs_info(sbi, "noinline_xattr options not supported");
611 #ifdef CONFIG_F2FS_FS_POSIX_ACL
613 set_opt(sbi, POSIX_ACL);
616 clear_opt(sbi, POSIX_ACL);
620 f2fs_info(sbi, "acl options not supported");
623 f2fs_info(sbi, "noacl options not supported");
626 case Opt_active_logs:
627 if (args->from && match_int(args, &arg))
629 if (arg != 2 && arg != 4 &&
630 arg != NR_CURSEG_PERSIST_TYPE)
632 F2FS_OPTION(sbi).active_logs = arg;
634 case Opt_disable_ext_identify:
635 set_opt(sbi, DISABLE_EXT_IDENTIFY);
637 case Opt_inline_data:
638 set_opt(sbi, INLINE_DATA);
640 case Opt_inline_dentry:
641 set_opt(sbi, INLINE_DENTRY);
643 case Opt_noinline_dentry:
644 clear_opt(sbi, INLINE_DENTRY);
646 case Opt_flush_merge:
647 set_opt(sbi, FLUSH_MERGE);
649 case Opt_noflush_merge:
650 clear_opt(sbi, FLUSH_MERGE);
653 set_opt(sbi, NOBARRIER);
656 set_opt(sbi, FASTBOOT);
658 case Opt_extent_cache:
659 set_opt(sbi, EXTENT_CACHE);
661 case Opt_noextent_cache:
662 clear_opt(sbi, EXTENT_CACHE);
664 case Opt_noinline_data:
665 clear_opt(sbi, INLINE_DATA);
668 set_opt(sbi, DATA_FLUSH);
670 case Opt_reserve_root:
671 if (args->from && match_int(args, &arg))
673 if (test_opt(sbi, RESERVE_ROOT)) {
674 f2fs_info(sbi, "Preserve previous reserve_root=%u",
675 F2FS_OPTION(sbi).root_reserved_blocks);
677 F2FS_OPTION(sbi).root_reserved_blocks = arg;
678 set_opt(sbi, RESERVE_ROOT);
682 if (args->from && match_int(args, &arg))
684 uid = make_kuid(current_user_ns(), arg);
685 if (!uid_valid(uid)) {
686 f2fs_err(sbi, "Invalid uid value %d", arg);
689 F2FS_OPTION(sbi).s_resuid = uid;
692 if (args->from && match_int(args, &arg))
694 gid = make_kgid(current_user_ns(), arg);
695 if (!gid_valid(gid)) {
696 f2fs_err(sbi, "Invalid gid value %d", arg);
699 F2FS_OPTION(sbi).s_resgid = gid;
702 name = match_strdup(&args[0]);
706 if (!strcmp(name, "adaptive")) {
707 if (f2fs_sb_has_blkzoned(sbi)) {
708 f2fs_warn(sbi, "adaptive mode is not allowed with zoned block device feature");
712 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
713 } else if (!strcmp(name, "lfs")) {
714 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
721 case Opt_io_size_bits:
722 if (args->from && match_int(args, &arg))
724 if (arg <= 0 || arg > __ilog2_u32(BIO_MAX_PAGES)) {
725 f2fs_warn(sbi, "Not support %d, larger than %d",
726 1 << arg, BIO_MAX_PAGES);
729 F2FS_OPTION(sbi).write_io_size_bits = arg;
731 #ifdef CONFIG_F2FS_FAULT_INJECTION
732 case Opt_fault_injection:
733 if (args->from && match_int(args, &arg))
735 f2fs_build_fault_attr(sbi, arg, F2FS_ALL_FAULT_TYPE);
736 set_opt(sbi, FAULT_INJECTION);
740 if (args->from && match_int(args, &arg))
742 f2fs_build_fault_attr(sbi, 0, arg);
743 set_opt(sbi, FAULT_INJECTION);
746 case Opt_fault_injection:
747 f2fs_info(sbi, "fault_injection options not supported");
751 f2fs_info(sbi, "fault_type options not supported");
755 sb->s_flags |= SB_LAZYTIME;
758 sb->s_flags &= ~SB_LAZYTIME;
763 set_opt(sbi, USRQUOTA);
766 set_opt(sbi, GRPQUOTA);
769 set_opt(sbi, PRJQUOTA);
772 ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
777 ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
782 ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
786 case Opt_offusrjquota:
787 ret = f2fs_clear_qf_name(sb, USRQUOTA);
791 case Opt_offgrpjquota:
792 ret = f2fs_clear_qf_name(sb, GRPQUOTA);
796 case Opt_offprjjquota:
797 ret = f2fs_clear_qf_name(sb, PRJQUOTA);
801 case Opt_jqfmt_vfsold:
802 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
804 case Opt_jqfmt_vfsv0:
805 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
807 case Opt_jqfmt_vfsv1:
808 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
811 clear_opt(sbi, QUOTA);
812 clear_opt(sbi, USRQUOTA);
813 clear_opt(sbi, GRPQUOTA);
814 clear_opt(sbi, PRJQUOTA);
824 case Opt_offusrjquota:
825 case Opt_offgrpjquota:
826 case Opt_offprjjquota:
827 case Opt_jqfmt_vfsold:
828 case Opt_jqfmt_vfsv0:
829 case Opt_jqfmt_vfsv1:
831 f2fs_info(sbi, "quota operations not supported");
835 name = match_strdup(&args[0]);
838 if (!strcmp(name, "user-based")) {
839 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_USER;
840 } else if (!strcmp(name, "off")) {
841 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
842 } else if (!strcmp(name, "fs-based")) {
843 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_FS;
851 name = match_strdup(&args[0]);
855 if (!strcmp(name, "default")) {
856 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
857 } else if (!strcmp(name, "reuse")) {
858 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
866 name = match_strdup(&args[0]);
869 if (!strcmp(name, "posix")) {
870 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
871 } else if (!strcmp(name, "strict")) {
872 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_STRICT;
873 } else if (!strcmp(name, "nobarrier")) {
874 F2FS_OPTION(sbi).fsync_mode =
875 FSYNC_MODE_NOBARRIER;
882 case Opt_test_dummy_encryption:
883 ret = f2fs_set_test_dummy_encryption(sb, p, &args[0],
888 case Opt_inlinecrypt:
889 #ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
890 sb->s_flags |= SB_INLINECRYPT;
892 f2fs_info(sbi, "inline encryption not supported");
895 case Opt_checkpoint_disable_cap_perc:
896 if (args->from && match_int(args, &arg))
898 if (arg < 0 || arg > 100)
900 F2FS_OPTION(sbi).unusable_cap_perc = arg;
901 set_opt(sbi, DISABLE_CHECKPOINT);
903 case Opt_checkpoint_disable_cap:
904 if (args->from && match_int(args, &arg))
906 F2FS_OPTION(sbi).unusable_cap = arg;
907 set_opt(sbi, DISABLE_CHECKPOINT);
909 case Opt_checkpoint_disable:
910 set_opt(sbi, DISABLE_CHECKPOINT);
912 case Opt_checkpoint_enable:
913 clear_opt(sbi, DISABLE_CHECKPOINT);
915 #ifdef CONFIG_F2FS_FS_COMPRESSION
916 case Opt_compress_algorithm:
917 if (!f2fs_sb_has_compression(sbi)) {
918 f2fs_info(sbi, "Image doesn't support compression");
921 name = match_strdup(&args[0]);
924 if (!strcmp(name, "lzo")) {
925 F2FS_OPTION(sbi).compress_algorithm =
927 } else if (!strcmp(name, "lz4")) {
928 F2FS_OPTION(sbi).compress_algorithm =
930 } else if (!strcmp(name, "zstd")) {
931 F2FS_OPTION(sbi).compress_algorithm =
933 } else if (!strcmp(name, "lzo-rle")) {
934 F2FS_OPTION(sbi).compress_algorithm =
942 case Opt_compress_log_size:
943 if (!f2fs_sb_has_compression(sbi)) {
944 f2fs_info(sbi, "Image doesn't support compression");
947 if (args->from && match_int(args, &arg))
949 if (arg < MIN_COMPRESS_LOG_SIZE ||
950 arg > MAX_COMPRESS_LOG_SIZE) {
952 "Compress cluster log size is out of range");
955 F2FS_OPTION(sbi).compress_log_size = arg;
957 case Opt_compress_extension:
958 if (!f2fs_sb_has_compression(sbi)) {
959 f2fs_info(sbi, "Image doesn't support compression");
962 name = match_strdup(&args[0]);
966 ext = F2FS_OPTION(sbi).extensions;
967 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
969 if (strlen(name) >= F2FS_EXTENSION_LEN ||
970 ext_cnt >= COMPRESS_EXT_NUM) {
972 "invalid extension length/number");
977 strcpy(ext[ext_cnt], name);
978 F2FS_OPTION(sbi).compress_ext_cnt++;
981 case Opt_compress_chksum:
982 F2FS_OPTION(sbi).compress_chksum = true;
984 case Opt_compress_mode:
985 name = match_strdup(&args[0]);
988 if (!strcmp(name, "fs")) {
989 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
990 } else if (!strcmp(name, "user")) {
991 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_USER;
999 case Opt_compress_algorithm:
1000 case Opt_compress_log_size:
1001 case Opt_compress_extension:
1002 case Opt_compress_chksum:
1003 case Opt_compress_mode:
1004 f2fs_info(sbi, "compression options not supported");
1011 f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
1017 if (f2fs_check_quota_options(sbi))
1020 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sbi->sb)) {
1021 f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1024 if (f2fs_sb_has_project_quota(sbi) && !f2fs_readonly(sbi->sb)) {
1025 f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1029 #ifndef CONFIG_UNICODE
1030 if (f2fs_sb_has_casefold(sbi)) {
1032 "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
1037 * The BLKZONED feature indicates that the drive was formatted with
1038 * zone alignment optimization. This is optional for host-aware
1039 * devices, but mandatory for host-managed zoned block devices.
1041 #ifndef CONFIG_BLK_DEV_ZONED
1042 if (f2fs_sb_has_blkzoned(sbi)) {
1043 f2fs_err(sbi, "Zoned block device support is not enabled");
1048 if (F2FS_IO_SIZE_BITS(sbi) && !f2fs_lfs_mode(sbi)) {
1049 f2fs_err(sbi, "Should set mode=lfs with %uKB-sized IO",
1050 F2FS_IO_SIZE_KB(sbi));
1054 if (test_opt(sbi, INLINE_XATTR_SIZE)) {
1055 int min_size, max_size;
1057 if (!f2fs_sb_has_extra_attr(sbi) ||
1058 !f2fs_sb_has_flexible_inline_xattr(sbi)) {
1059 f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
1062 if (!test_opt(sbi, INLINE_XATTR)) {
1063 f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
1067 min_size = sizeof(struct f2fs_xattr_header) / sizeof(__le32);
1068 max_size = MAX_INLINE_XATTR_SIZE;
1070 if (F2FS_OPTION(sbi).inline_xattr_size < min_size ||
1071 F2FS_OPTION(sbi).inline_xattr_size > max_size) {
1072 f2fs_err(sbi, "inline xattr size is out of range: %d ~ %d",
1073 min_size, max_size);
1078 if (test_opt(sbi, DISABLE_CHECKPOINT) && f2fs_lfs_mode(sbi)) {
1079 f2fs_err(sbi, "LFS not compatible with checkpoint=disable\n");
1083 /* Not pass down write hints if the number of active logs is lesser
1084 * than NR_CURSEG_PERSIST_TYPE.
1086 if (F2FS_OPTION(sbi).active_logs != NR_CURSEG_PERSIST_TYPE)
1087 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1091 static struct inode *f2fs_alloc_inode(struct super_block *sb)
1093 struct f2fs_inode_info *fi;
1095 fi = kmem_cache_alloc(f2fs_inode_cachep, GFP_F2FS_ZERO);
1099 init_once((void *) fi);
1101 /* Initialize f2fs-specific inode info */
1102 atomic_set(&fi->dirty_pages, 0);
1103 atomic_set(&fi->i_compr_blocks, 0);
1104 init_rwsem(&fi->i_sem);
1105 spin_lock_init(&fi->i_size_lock);
1106 INIT_LIST_HEAD(&fi->dirty_list);
1107 INIT_LIST_HEAD(&fi->gdirty_list);
1108 INIT_LIST_HEAD(&fi->inmem_ilist);
1109 INIT_LIST_HEAD(&fi->inmem_pages);
1110 mutex_init(&fi->inmem_lock);
1111 init_rwsem(&fi->i_gc_rwsem[READ]);
1112 init_rwsem(&fi->i_gc_rwsem[WRITE]);
1113 init_rwsem(&fi->i_mmap_sem);
1114 init_rwsem(&fi->i_xattr_sem);
1116 /* Will be used by directory only */
1117 fi->i_dir_level = F2FS_SB(sb)->dir_level;
1121 return &fi->vfs_inode;
1124 static int f2fs_drop_inode(struct inode *inode)
1126 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1130 * during filesystem shutdown, if checkpoint is disabled,
1131 * drop useless meta/node dirty pages.
1133 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1134 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1135 inode->i_ino == F2FS_META_INO(sbi)) {
1136 trace_f2fs_drop_inode(inode, 1);
1142 * This is to avoid a deadlock condition like below.
1143 * writeback_single_inode(inode)
1144 * - f2fs_write_data_page
1145 * - f2fs_gc -> iput -> evict
1146 * - inode_wait_for_writeback(inode)
1148 if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
1149 if (!inode->i_nlink && !is_bad_inode(inode)) {
1150 /* to avoid evict_inode call simultaneously */
1151 atomic_inc(&inode->i_count);
1152 spin_unlock(&inode->i_lock);
1154 /* some remained atomic pages should discarded */
1155 if (f2fs_is_atomic_file(inode))
1156 f2fs_drop_inmem_pages(inode);
1158 /* should remain fi->extent_tree for writepage */
1159 f2fs_destroy_extent_node(inode);
1161 sb_start_intwrite(inode->i_sb);
1162 f2fs_i_size_write(inode, 0);
1164 f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1165 inode, NULL, 0, DATA);
1166 truncate_inode_pages_final(inode->i_mapping);
1168 if (F2FS_HAS_BLOCKS(inode))
1169 f2fs_truncate(inode);
1171 sb_end_intwrite(inode->i_sb);
1173 spin_lock(&inode->i_lock);
1174 atomic_dec(&inode->i_count);
1176 trace_f2fs_drop_inode(inode, 0);
1179 ret = generic_drop_inode(inode);
1181 ret = fscrypt_drop_inode(inode);
1182 trace_f2fs_drop_inode(inode, ret);
1186 int f2fs_inode_dirtied(struct inode *inode, bool sync)
1188 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1191 spin_lock(&sbi->inode_lock[DIRTY_META]);
1192 if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1195 set_inode_flag(inode, FI_DIRTY_INODE);
1196 stat_inc_dirty_inode(sbi, DIRTY_META);
1198 if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1199 list_add_tail(&F2FS_I(inode)->gdirty_list,
1200 &sbi->inode_list[DIRTY_META]);
1201 inc_page_count(sbi, F2FS_DIRTY_IMETA);
1203 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1207 void f2fs_inode_synced(struct inode *inode)
1209 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1211 spin_lock(&sbi->inode_lock[DIRTY_META]);
1212 if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1213 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1216 if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1217 list_del_init(&F2FS_I(inode)->gdirty_list);
1218 dec_page_count(sbi, F2FS_DIRTY_IMETA);
1220 clear_inode_flag(inode, FI_DIRTY_INODE);
1221 clear_inode_flag(inode, FI_AUTO_RECOVER);
1222 stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1223 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1227 * f2fs_dirty_inode() is called from __mark_inode_dirty()
1229 * We should call set_dirty_inode to write the dirty inode through write_inode.
1231 static void f2fs_dirty_inode(struct inode *inode, int flags)
1233 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1235 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1236 inode->i_ino == F2FS_META_INO(sbi))
1239 if (flags == I_DIRTY_TIME)
1242 if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1243 clear_inode_flag(inode, FI_AUTO_RECOVER);
1245 f2fs_inode_dirtied(inode, false);
1248 static void f2fs_free_inode(struct inode *inode)
1250 fscrypt_free_inode(inode);
1251 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1254 static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1256 percpu_counter_destroy(&sbi->alloc_valid_block_count);
1257 percpu_counter_destroy(&sbi->total_valid_inode_count);
1260 static void destroy_device_list(struct f2fs_sb_info *sbi)
1264 for (i = 0; i < sbi->s_ndevs; i++) {
1265 blkdev_put(FDEV(i).bdev, FMODE_EXCL);
1266 #ifdef CONFIG_BLK_DEV_ZONED
1267 kvfree(FDEV(i).blkz_seq);
1273 static void f2fs_put_super(struct super_block *sb)
1275 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1279 /* unregister procfs/sysfs entries in advance to avoid race case */
1280 f2fs_unregister_sysfs(sbi);
1282 f2fs_quota_off_umount(sb);
1284 /* prevent remaining shrinker jobs */
1285 mutex_lock(&sbi->umount_mutex);
1288 * We don't need to do checkpoint when superblock is clean.
1289 * But, the previous checkpoint was not done by umount, it needs to do
1290 * clean checkpoint again.
1292 if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1293 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1294 struct cp_control cpc = {
1295 .reason = CP_UMOUNT,
1297 f2fs_write_checkpoint(sbi, &cpc);
1300 /* be sure to wait for any on-going discard commands */
1301 dropped = f2fs_issue_discard_timeout(sbi);
1303 if ((f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi)) &&
1304 !sbi->discard_blks && !dropped) {
1305 struct cp_control cpc = {
1306 .reason = CP_UMOUNT | CP_TRIMMED,
1308 f2fs_write_checkpoint(sbi, &cpc);
1312 * normally superblock is clean, so we need to release this.
1313 * In addition, EIO will skip do checkpoint, we need this as well.
1315 f2fs_release_ino_entry(sbi, true);
1317 f2fs_leave_shrinker(sbi);
1318 mutex_unlock(&sbi->umount_mutex);
1320 /* our cp_error case, we can wait for any writeback page */
1321 f2fs_flush_merged_writes(sbi);
1323 f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1325 f2fs_bug_on(sbi, sbi->fsync_node_num);
1327 iput(sbi->node_inode);
1328 sbi->node_inode = NULL;
1330 iput(sbi->meta_inode);
1331 sbi->meta_inode = NULL;
1334 * iput() can update stat information, if f2fs_write_checkpoint()
1335 * above failed with error.
1337 f2fs_destroy_stats(sbi);
1339 /* destroy f2fs internal modules */
1340 f2fs_destroy_node_manager(sbi);
1341 f2fs_destroy_segment_manager(sbi);
1343 f2fs_destroy_post_read_wq(sbi);
1347 sb->s_fs_info = NULL;
1348 if (sbi->s_chksum_driver)
1349 crypto_free_shash(sbi->s_chksum_driver);
1350 kfree(sbi->raw_super);
1352 destroy_device_list(sbi);
1353 f2fs_destroy_page_array_cache(sbi);
1354 f2fs_destroy_xattr_caches(sbi);
1355 mempool_destroy(sbi->write_io_dummy);
1357 for (i = 0; i < MAXQUOTAS; i++)
1358 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1360 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
1361 destroy_percpu_info(sbi);
1362 for (i = 0; i < NR_PAGE_TYPE; i++)
1363 kvfree(sbi->write_io[i]);
1364 #ifdef CONFIG_UNICODE
1365 utf8_unload(sb->s_encoding);
1370 int f2fs_sync_fs(struct super_block *sb, int sync)
1372 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1375 if (unlikely(f2fs_cp_error(sbi)))
1377 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1380 trace_f2fs_sync_fs(sb, sync);
1382 if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1386 struct cp_control cpc;
1388 cpc.reason = __get_cp_reason(sbi);
1390 down_write(&sbi->gc_lock);
1391 err = f2fs_write_checkpoint(sbi, &cpc);
1392 up_write(&sbi->gc_lock);
1394 f2fs_trace_ios(NULL, 1);
1399 static int f2fs_freeze(struct super_block *sb)
1401 if (f2fs_readonly(sb))
1404 /* IO error happened before */
1405 if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1408 /* must be clean, since sync_filesystem() was already called */
1409 if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1414 static int f2fs_unfreeze(struct super_block *sb)
1420 static int f2fs_statfs_project(struct super_block *sb,
1421 kprojid_t projid, struct kstatfs *buf)
1424 struct dquot *dquot;
1428 qid = make_kqid_projid(projid);
1429 dquot = dqget(sb, qid);
1431 return PTR_ERR(dquot);
1432 spin_lock(&dquot->dq_dqb_lock);
1434 limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1435 dquot->dq_dqb.dqb_bhardlimit);
1437 limit >>= sb->s_blocksize_bits;
1439 if (limit && buf->f_blocks > limit) {
1440 curblock = (dquot->dq_dqb.dqb_curspace +
1441 dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
1442 buf->f_blocks = limit;
1443 buf->f_bfree = buf->f_bavail =
1444 (buf->f_blocks > curblock) ?
1445 (buf->f_blocks - curblock) : 0;
1448 limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1449 dquot->dq_dqb.dqb_ihardlimit);
1451 if (limit && buf->f_files > limit) {
1452 buf->f_files = limit;
1454 (buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1455 (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1458 spin_unlock(&dquot->dq_dqb_lock);
1464 static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1466 struct super_block *sb = dentry->d_sb;
1467 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1468 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
1469 block_t total_count, user_block_count, start_count;
1470 u64 avail_node_count;
1472 total_count = le64_to_cpu(sbi->raw_super->block_count);
1473 user_block_count = sbi->user_block_count;
1474 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
1475 buf->f_type = F2FS_SUPER_MAGIC;
1476 buf->f_bsize = sbi->blocksize;
1478 buf->f_blocks = total_count - start_count;
1479 buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1480 sbi->current_reserved_blocks;
1482 spin_lock(&sbi->stat_lock);
1483 if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1486 buf->f_bfree -= sbi->unusable_block_count;
1487 spin_unlock(&sbi->stat_lock);
1489 if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1490 buf->f_bavail = buf->f_bfree -
1491 F2FS_OPTION(sbi).root_reserved_blocks;
1495 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
1497 if (avail_node_count > user_block_count) {
1498 buf->f_files = user_block_count;
1499 buf->f_ffree = buf->f_bavail;
1501 buf->f_files = avail_node_count;
1502 buf->f_ffree = min(avail_node_count - valid_node_count(sbi),
1506 buf->f_namelen = F2FS_NAME_LEN;
1507 buf->f_fsid = u64_to_fsid(id);
1510 if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1511 sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1512 f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1518 static inline void f2fs_show_quota_options(struct seq_file *seq,
1519 struct super_block *sb)
1522 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1524 if (F2FS_OPTION(sbi).s_jquota_fmt) {
1527 switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1538 seq_printf(seq, ",jqfmt=%s", fmtname);
1541 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1542 seq_show_option(seq, "usrjquota",
1543 F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1545 if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1546 seq_show_option(seq, "grpjquota",
1547 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1549 if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1550 seq_show_option(seq, "prjjquota",
1551 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1555 static inline void f2fs_show_compress_options(struct seq_file *seq,
1556 struct super_block *sb)
1558 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1562 if (!f2fs_sb_has_compression(sbi))
1565 switch (F2FS_OPTION(sbi).compress_algorithm) {
1575 case COMPRESS_LZORLE:
1576 algtype = "lzo-rle";
1579 seq_printf(seq, ",compress_algorithm=%s", algtype);
1581 seq_printf(seq, ",compress_log_size=%u",
1582 F2FS_OPTION(sbi).compress_log_size);
1584 for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1585 seq_printf(seq, ",compress_extension=%s",
1586 F2FS_OPTION(sbi).extensions[i]);
1589 if (F2FS_OPTION(sbi).compress_chksum)
1590 seq_puts(seq, ",compress_chksum");
1592 if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_FS)
1593 seq_printf(seq, ",compress_mode=%s", "fs");
1594 else if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_USER)
1595 seq_printf(seq, ",compress_mode=%s", "user");
1598 static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1600 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1602 if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1603 seq_printf(seq, ",background_gc=%s", "sync");
1604 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1605 seq_printf(seq, ",background_gc=%s", "on");
1606 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1607 seq_printf(seq, ",background_gc=%s", "off");
1609 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1610 seq_puts(seq, ",disable_roll_forward");
1611 if (test_opt(sbi, NORECOVERY))
1612 seq_puts(seq, ",norecovery");
1613 if (test_opt(sbi, DISCARD))
1614 seq_puts(seq, ",discard");
1616 seq_puts(seq, ",nodiscard");
1617 if (test_opt(sbi, NOHEAP))
1618 seq_puts(seq, ",no_heap");
1620 seq_puts(seq, ",heap");
1621 #ifdef CONFIG_F2FS_FS_XATTR
1622 if (test_opt(sbi, XATTR_USER))
1623 seq_puts(seq, ",user_xattr");
1625 seq_puts(seq, ",nouser_xattr");
1626 if (test_opt(sbi, INLINE_XATTR))
1627 seq_puts(seq, ",inline_xattr");
1629 seq_puts(seq, ",noinline_xattr");
1630 if (test_opt(sbi, INLINE_XATTR_SIZE))
1631 seq_printf(seq, ",inline_xattr_size=%u",
1632 F2FS_OPTION(sbi).inline_xattr_size);
1634 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1635 if (test_opt(sbi, POSIX_ACL))
1636 seq_puts(seq, ",acl");
1638 seq_puts(seq, ",noacl");
1640 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
1641 seq_puts(seq, ",disable_ext_identify");
1642 if (test_opt(sbi, INLINE_DATA))
1643 seq_puts(seq, ",inline_data");
1645 seq_puts(seq, ",noinline_data");
1646 if (test_opt(sbi, INLINE_DENTRY))
1647 seq_puts(seq, ",inline_dentry");
1649 seq_puts(seq, ",noinline_dentry");
1650 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
1651 seq_puts(seq, ",flush_merge");
1652 if (test_opt(sbi, NOBARRIER))
1653 seq_puts(seq, ",nobarrier");
1654 if (test_opt(sbi, FASTBOOT))
1655 seq_puts(seq, ",fastboot");
1656 if (test_opt(sbi, EXTENT_CACHE))
1657 seq_puts(seq, ",extent_cache");
1659 seq_puts(seq, ",noextent_cache");
1660 if (test_opt(sbi, DATA_FLUSH))
1661 seq_puts(seq, ",data_flush");
1663 seq_puts(seq, ",mode=");
1664 if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
1665 seq_puts(seq, "adaptive");
1666 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
1667 seq_puts(seq, "lfs");
1668 seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
1669 if (test_opt(sbi, RESERVE_ROOT))
1670 seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
1671 F2FS_OPTION(sbi).root_reserved_blocks,
1672 from_kuid_munged(&init_user_ns,
1673 F2FS_OPTION(sbi).s_resuid),
1674 from_kgid_munged(&init_user_ns,
1675 F2FS_OPTION(sbi).s_resgid));
1676 if (F2FS_IO_SIZE_BITS(sbi))
1677 seq_printf(seq, ",io_bits=%u",
1678 F2FS_OPTION(sbi).write_io_size_bits);
1679 #ifdef CONFIG_F2FS_FAULT_INJECTION
1680 if (test_opt(sbi, FAULT_INJECTION)) {
1681 seq_printf(seq, ",fault_injection=%u",
1682 F2FS_OPTION(sbi).fault_info.inject_rate);
1683 seq_printf(seq, ",fault_type=%u",
1684 F2FS_OPTION(sbi).fault_info.inject_type);
1688 if (test_opt(sbi, QUOTA))
1689 seq_puts(seq, ",quota");
1690 if (test_opt(sbi, USRQUOTA))
1691 seq_puts(seq, ",usrquota");
1692 if (test_opt(sbi, GRPQUOTA))
1693 seq_puts(seq, ",grpquota");
1694 if (test_opt(sbi, PRJQUOTA))
1695 seq_puts(seq, ",prjquota");
1697 f2fs_show_quota_options(seq, sbi->sb);
1698 if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_USER)
1699 seq_printf(seq, ",whint_mode=%s", "user-based");
1700 else if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_FS)
1701 seq_printf(seq, ",whint_mode=%s", "fs-based");
1703 fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
1705 if (sbi->sb->s_flags & SB_INLINECRYPT)
1706 seq_puts(seq, ",inlinecrypt");
1708 if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
1709 seq_printf(seq, ",alloc_mode=%s", "default");
1710 else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
1711 seq_printf(seq, ",alloc_mode=%s", "reuse");
1713 if (test_opt(sbi, DISABLE_CHECKPOINT))
1714 seq_printf(seq, ",checkpoint=disable:%u",
1715 F2FS_OPTION(sbi).unusable_cap);
1716 if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
1717 seq_printf(seq, ",fsync_mode=%s", "posix");
1718 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
1719 seq_printf(seq, ",fsync_mode=%s", "strict");
1720 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
1721 seq_printf(seq, ",fsync_mode=%s", "nobarrier");
1723 #ifdef CONFIG_F2FS_FS_COMPRESSION
1724 f2fs_show_compress_options(seq, sbi->sb);
1727 if (test_opt(sbi, ATGC))
1728 seq_puts(seq, ",atgc");
1732 static void default_options(struct f2fs_sb_info *sbi)
1734 /* init some FS parameters */
1735 F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
1736 F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
1737 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1738 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
1739 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
1740 F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
1741 F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
1742 F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
1743 F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
1744 F2FS_OPTION(sbi).compress_ext_cnt = 0;
1745 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
1746 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
1748 sbi->sb->s_flags &= ~SB_INLINECRYPT;
1750 set_opt(sbi, INLINE_XATTR);
1751 set_opt(sbi, INLINE_DATA);
1752 set_opt(sbi, INLINE_DENTRY);
1753 set_opt(sbi, EXTENT_CACHE);
1754 set_opt(sbi, NOHEAP);
1755 clear_opt(sbi, DISABLE_CHECKPOINT);
1756 F2FS_OPTION(sbi).unusable_cap = 0;
1757 sbi->sb->s_flags |= SB_LAZYTIME;
1758 set_opt(sbi, FLUSH_MERGE);
1759 set_opt(sbi, DISCARD);
1760 if (f2fs_sb_has_blkzoned(sbi))
1761 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
1763 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
1765 #ifdef CONFIG_F2FS_FS_XATTR
1766 set_opt(sbi, XATTR_USER);
1768 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1769 set_opt(sbi, POSIX_ACL);
1772 f2fs_build_fault_attr(sbi, 0, 0);
1776 static int f2fs_enable_quotas(struct super_block *sb);
1779 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
1781 unsigned int s_flags = sbi->sb->s_flags;
1782 struct cp_control cpc;
1787 if (s_flags & SB_RDONLY) {
1788 f2fs_err(sbi, "checkpoint=disable on readonly fs");
1791 sbi->sb->s_flags |= SB_ACTIVE;
1793 f2fs_update_time(sbi, DISABLE_TIME);
1795 while (!f2fs_time_over(sbi, DISABLE_TIME)) {
1796 down_write(&sbi->gc_lock);
1797 err = f2fs_gc(sbi, true, false, false, NULL_SEGNO);
1798 if (err == -ENODATA) {
1802 if (err && err != -EAGAIN)
1806 ret = sync_filesystem(sbi->sb);
1808 err = ret ? ret: err;
1812 unusable = f2fs_get_unusable_blocks(sbi);
1813 if (f2fs_disable_cp_again(sbi, unusable)) {
1818 down_write(&sbi->gc_lock);
1819 cpc.reason = CP_PAUSE;
1820 set_sbi_flag(sbi, SBI_CP_DISABLED);
1821 err = f2fs_write_checkpoint(sbi, &cpc);
1825 spin_lock(&sbi->stat_lock);
1826 sbi->unusable_block_count = unusable;
1827 spin_unlock(&sbi->stat_lock);
1830 up_write(&sbi->gc_lock);
1832 sbi->sb->s_flags = s_flags; /* Restore SB_RDONLY status */
1836 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
1838 int retry = DEFAULT_RETRY_IO_COUNT;
1840 /* we should flush all the data to keep data consistency */
1842 sync_inodes_sb(sbi->sb);
1844 congestion_wait(BLK_RW_ASYNC, DEFAULT_IO_TIMEOUT);
1845 } while (get_pages(sbi, F2FS_DIRTY_DATA) && retry--);
1847 if (unlikely(retry < 0))
1848 f2fs_warn(sbi, "checkpoint=enable has some unwritten data.");
1850 down_write(&sbi->gc_lock);
1851 f2fs_dirty_to_prefree(sbi);
1853 clear_sbi_flag(sbi, SBI_CP_DISABLED);
1854 set_sbi_flag(sbi, SBI_IS_DIRTY);
1855 up_write(&sbi->gc_lock);
1857 f2fs_sync_fs(sbi->sb, 1);
1860 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
1862 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1863 struct f2fs_mount_info org_mount_opt;
1864 unsigned long old_sb_flags;
1866 bool need_restart_gc = false;
1867 bool need_stop_gc = false;
1868 bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
1869 bool disable_checkpoint = test_opt(sbi, DISABLE_CHECKPOINT);
1870 bool no_io_align = !F2FS_IO_ALIGNED(sbi);
1871 bool no_atgc = !test_opt(sbi, ATGC);
1872 bool checkpoint_changed;
1878 * Save the old mount options in case we
1879 * need to restore them.
1881 org_mount_opt = sbi->mount_opt;
1882 old_sb_flags = sb->s_flags;
1885 org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
1886 for (i = 0; i < MAXQUOTAS; i++) {
1887 if (F2FS_OPTION(sbi).s_qf_names[i]) {
1888 org_mount_opt.s_qf_names[i] =
1889 kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
1891 if (!org_mount_opt.s_qf_names[i]) {
1892 for (j = 0; j < i; j++)
1893 kfree(org_mount_opt.s_qf_names[j]);
1897 org_mount_opt.s_qf_names[i] = NULL;
1902 /* recover superblocks we couldn't write due to previous RO mount */
1903 if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
1904 err = f2fs_commit_super(sbi, false);
1905 f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
1908 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
1911 default_options(sbi);
1913 /* parse mount options */
1914 err = parse_options(sb, data, true);
1917 checkpoint_changed =
1918 disable_checkpoint != test_opt(sbi, DISABLE_CHECKPOINT);
1921 * Previous and new state of filesystem is RO,
1922 * so skip checking GC and FLUSH_MERGE conditions.
1924 if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
1928 if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
1929 err = dquot_suspend(sb, -1);
1932 } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
1933 /* dquot_resume needs RW */
1934 sb->s_flags &= ~SB_RDONLY;
1935 if (sb_any_quota_suspended(sb)) {
1936 dquot_resume(sb, -1);
1937 } else if (f2fs_sb_has_quota_ino(sbi)) {
1938 err = f2fs_enable_quotas(sb);
1944 /* disallow enable atgc dynamically */
1945 if (no_atgc == !!test_opt(sbi, ATGC)) {
1947 f2fs_warn(sbi, "switch atgc option is not allowed");
1951 /* disallow enable/disable extent_cache dynamically */
1952 if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
1954 f2fs_warn(sbi, "switch extent_cache option is not allowed");
1958 if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
1960 f2fs_warn(sbi, "switch io_bits option is not allowed");
1964 if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
1966 f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
1971 * We stop the GC thread if FS is mounted as RO
1972 * or if background_gc = off is passed in mount
1973 * option. Also sync the filesystem.
1975 if ((*flags & SB_RDONLY) ||
1976 F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF) {
1977 if (sbi->gc_thread) {
1978 f2fs_stop_gc_thread(sbi);
1979 need_restart_gc = true;
1981 } else if (!sbi->gc_thread) {
1982 err = f2fs_start_gc_thread(sbi);
1985 need_stop_gc = true;
1988 if (*flags & SB_RDONLY ||
1989 F2FS_OPTION(sbi).whint_mode != org_mount_opt.whint_mode) {
1990 writeback_inodes_sb(sb, WB_REASON_SYNC);
1993 set_sbi_flag(sbi, SBI_IS_DIRTY);
1994 set_sbi_flag(sbi, SBI_IS_CLOSE);
1995 f2fs_sync_fs(sb, 1);
1996 clear_sbi_flag(sbi, SBI_IS_CLOSE);
1999 if (checkpoint_changed) {
2000 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
2001 err = f2fs_disable_checkpoint(sbi);
2005 f2fs_enable_checkpoint(sbi);
2010 * We stop issue flush thread if FS is mounted as RO
2011 * or if flush_merge is not passed in mount option.
2013 if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2014 clear_opt(sbi, FLUSH_MERGE);
2015 f2fs_destroy_flush_cmd_control(sbi, false);
2017 err = f2fs_create_flush_cmd_control(sbi);
2023 /* Release old quota file names */
2024 for (i = 0; i < MAXQUOTAS; i++)
2025 kfree(org_mount_opt.s_qf_names[i]);
2027 /* Update the POSIXACL Flag */
2028 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
2029 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
2031 limit_reserve_root(sbi);
2032 adjust_unusable_cap_perc(sbi);
2033 *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
2036 if (need_restart_gc) {
2037 if (f2fs_start_gc_thread(sbi))
2038 f2fs_warn(sbi, "background gc thread has stopped");
2039 } else if (need_stop_gc) {
2040 f2fs_stop_gc_thread(sbi);
2044 F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
2045 for (i = 0; i < MAXQUOTAS; i++) {
2046 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
2047 F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
2050 sbi->mount_opt = org_mount_opt;
2051 sb->s_flags = old_sb_flags;
2056 /* Read data from quotafile */
2057 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
2058 size_t len, loff_t off)
2060 struct inode *inode = sb_dqopt(sb)->files[type];
2061 struct address_space *mapping = inode->i_mapping;
2062 block_t blkidx = F2FS_BYTES_TO_BLK(off);
2063 int offset = off & (sb->s_blocksize - 1);
2066 loff_t i_size = i_size_read(inode);
2072 if (off + len > i_size)
2075 while (toread > 0) {
2076 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
2078 page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
2080 if (PTR_ERR(page) == -ENOMEM) {
2081 congestion_wait(BLK_RW_ASYNC,
2082 DEFAULT_IO_TIMEOUT);
2085 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2086 return PTR_ERR(page);
2091 if (unlikely(page->mapping != mapping)) {
2092 f2fs_put_page(page, 1);
2095 if (unlikely(!PageUptodate(page))) {
2096 f2fs_put_page(page, 1);
2097 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2101 memcpy_from_page(data, page, offset, tocopy);
2102 f2fs_put_page(page, 1);
2112 /* Write to quotafile */
2113 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2114 const char *data, size_t len, loff_t off)
2116 struct inode *inode = sb_dqopt(sb)->files[type];
2117 struct address_space *mapping = inode->i_mapping;
2118 const struct address_space_operations *a_ops = mapping->a_ops;
2119 int offset = off & (sb->s_blocksize - 1);
2120 size_t towrite = len;
2122 void *fsdata = NULL;
2126 while (towrite > 0) {
2127 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2130 err = a_ops->write_begin(NULL, mapping, off, tocopy, 0,
2132 if (unlikely(err)) {
2133 if (err == -ENOMEM) {
2134 congestion_wait(BLK_RW_ASYNC,
2135 DEFAULT_IO_TIMEOUT);
2138 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2142 memcpy_to_page(page, offset, data, tocopy);
2144 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2155 inode->i_mtime = inode->i_ctime = current_time(inode);
2156 f2fs_mark_inode_dirty_sync(inode, false);
2157 return len - towrite;
2160 static struct dquot **f2fs_get_dquots(struct inode *inode)
2162 return F2FS_I(inode)->i_dquot;
2165 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2167 return &F2FS_I(inode)->i_reserved_quota;
2170 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2172 if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2173 f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2177 return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2178 F2FS_OPTION(sbi).s_jquota_fmt, type);
2181 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2186 if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2187 err = f2fs_enable_quotas(sbi->sb);
2189 f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2195 for (i = 0; i < MAXQUOTAS; i++) {
2196 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2197 err = f2fs_quota_on_mount(sbi, i);
2202 f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2209 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2212 struct inode *qf_inode;
2213 unsigned long qf_inum;
2216 BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2218 qf_inum = f2fs_qf_ino(sb, type);
2222 qf_inode = f2fs_iget(sb, qf_inum);
2223 if (IS_ERR(qf_inode)) {
2224 f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2225 return PTR_ERR(qf_inode);
2228 /* Don't account quota for quota files to avoid recursion */
2229 qf_inode->i_flags |= S_NOQUOTA;
2230 err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2235 static int f2fs_enable_quotas(struct super_block *sb)
2237 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2239 unsigned long qf_inum;
2240 bool quota_mopt[MAXQUOTAS] = {
2241 test_opt(sbi, USRQUOTA),
2242 test_opt(sbi, GRPQUOTA),
2243 test_opt(sbi, PRJQUOTA),
2246 if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2247 f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2251 sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2253 for (type = 0; type < MAXQUOTAS; type++) {
2254 qf_inum = f2fs_qf_ino(sb, type);
2256 err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2257 DQUOT_USAGE_ENABLED |
2258 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2260 f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2262 for (type--; type >= 0; type--)
2263 dquot_quota_off(sb, type);
2264 set_sbi_flag(F2FS_SB(sb),
2265 SBI_QUOTA_NEED_REPAIR);
2273 static int f2fs_quota_sync_file(struct f2fs_sb_info *sbi, int type)
2275 struct quota_info *dqopt = sb_dqopt(sbi->sb);
2276 struct address_space *mapping = dqopt->files[type]->i_mapping;
2279 ret = dquot_writeback_dquots(sbi->sb, type);
2283 ret = filemap_fdatawrite(mapping);
2287 /* if we are using journalled quota */
2288 if (is_journalled_quota(sbi))
2291 ret = filemap_fdatawait(mapping);
2293 truncate_inode_pages(&dqopt->files[type]->i_data, 0);
2296 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2300 int f2fs_quota_sync(struct super_block *sb, int type)
2302 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2303 struct quota_info *dqopt = sb_dqopt(sb);
2308 * Now when everything is written we can discard the pagecache so
2309 * that userspace sees the changes.
2311 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2313 if (type != -1 && cnt != type)
2316 if (!sb_has_quota_active(sb, cnt))
2319 if (!f2fs_sb_has_quota_ino(sbi))
2320 inode_lock(dqopt->files[cnt]);
2325 * down_read(quota_sem)
2326 * dquot_writeback_dquots()
2329 * down_read(quota_sem)
2332 down_read(&sbi->quota_sem);
2334 ret = f2fs_quota_sync_file(sbi, cnt);
2336 up_read(&sbi->quota_sem);
2337 f2fs_unlock_op(sbi);
2339 if (!f2fs_sb_has_quota_ino(sbi))
2340 inode_unlock(dqopt->files[cnt]);
2348 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2349 const struct path *path)
2351 struct inode *inode;
2354 /* if quota sysfile exists, deny enabling quota with specific file */
2355 if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2356 f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2360 err = f2fs_quota_sync(sb, type);
2364 err = dquot_quota_on(sb, type, format_id, path);
2368 inode = d_inode(path->dentry);
2371 F2FS_I(inode)->i_flags |= F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL;
2372 f2fs_set_inode_flags(inode);
2373 inode_unlock(inode);
2374 f2fs_mark_inode_dirty_sync(inode, false);
2379 static int __f2fs_quota_off(struct super_block *sb, int type)
2381 struct inode *inode = sb_dqopt(sb)->files[type];
2384 if (!inode || !igrab(inode))
2385 return dquot_quota_off(sb, type);
2387 err = f2fs_quota_sync(sb, type);
2391 err = dquot_quota_off(sb, type);
2392 if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
2396 F2FS_I(inode)->i_flags &= ~(F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL);
2397 f2fs_set_inode_flags(inode);
2398 inode_unlock(inode);
2399 f2fs_mark_inode_dirty_sync(inode, false);
2405 static int f2fs_quota_off(struct super_block *sb, int type)
2407 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2410 err = __f2fs_quota_off(sb, type);
2413 * quotactl can shutdown journalled quota, result in inconsistence
2414 * between quota record and fs data by following updates, tag the
2415 * flag to let fsck be aware of it.
2417 if (is_journalled_quota(sbi))
2418 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2422 void f2fs_quota_off_umount(struct super_block *sb)
2427 for (type = 0; type < MAXQUOTAS; type++) {
2428 err = __f2fs_quota_off(sb, type);
2430 int ret = dquot_quota_off(sb, type);
2432 f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
2434 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2438 * In case of checkpoint=disable, we must flush quota blocks.
2439 * This can cause NULL exception for node_inode in end_io, since
2440 * put_super already dropped it.
2442 sync_filesystem(sb);
2445 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
2447 struct quota_info *dqopt = sb_dqopt(sb);
2450 for (type = 0; type < MAXQUOTAS; type++) {
2451 if (!dqopt->files[type])
2453 f2fs_inode_synced(dqopt->files[type]);
2457 static int f2fs_dquot_commit(struct dquot *dquot)
2459 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2462 down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
2463 ret = dquot_commit(dquot);
2465 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2466 up_read(&sbi->quota_sem);
2470 static int f2fs_dquot_acquire(struct dquot *dquot)
2472 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2475 down_read(&sbi->quota_sem);
2476 ret = dquot_acquire(dquot);
2478 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2479 up_read(&sbi->quota_sem);
2483 static int f2fs_dquot_release(struct dquot *dquot)
2485 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2486 int ret = dquot_release(dquot);
2489 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2493 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
2495 struct super_block *sb = dquot->dq_sb;
2496 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2497 int ret = dquot_mark_dquot_dirty(dquot);
2499 /* if we are using journalled quota */
2500 if (is_journalled_quota(sbi))
2501 set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
2506 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
2508 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2509 int ret = dquot_commit_info(sb, type);
2512 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2516 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
2518 *projid = F2FS_I(inode)->i_projid;
2522 static const struct dquot_operations f2fs_quota_operations = {
2523 .get_reserved_space = f2fs_get_reserved_space,
2524 .write_dquot = f2fs_dquot_commit,
2525 .acquire_dquot = f2fs_dquot_acquire,
2526 .release_dquot = f2fs_dquot_release,
2527 .mark_dirty = f2fs_dquot_mark_dquot_dirty,
2528 .write_info = f2fs_dquot_commit_info,
2529 .alloc_dquot = dquot_alloc,
2530 .destroy_dquot = dquot_destroy,
2531 .get_projid = f2fs_get_projid,
2532 .get_next_id = dquot_get_next_id,
2535 static const struct quotactl_ops f2fs_quotactl_ops = {
2536 .quota_on = f2fs_quota_on,
2537 .quota_off = f2fs_quota_off,
2538 .quota_sync = f2fs_quota_sync,
2539 .get_state = dquot_get_state,
2540 .set_info = dquot_set_dqinfo,
2541 .get_dqblk = dquot_get_dqblk,
2542 .set_dqblk = dquot_set_dqblk,
2543 .get_nextdqblk = dquot_get_next_dqblk,
2546 int f2fs_quota_sync(struct super_block *sb, int type)
2551 void f2fs_quota_off_umount(struct super_block *sb)
2556 static const struct super_operations f2fs_sops = {
2557 .alloc_inode = f2fs_alloc_inode,
2558 .free_inode = f2fs_free_inode,
2559 .drop_inode = f2fs_drop_inode,
2560 .write_inode = f2fs_write_inode,
2561 .dirty_inode = f2fs_dirty_inode,
2562 .show_options = f2fs_show_options,
2564 .quota_read = f2fs_quota_read,
2565 .quota_write = f2fs_quota_write,
2566 .get_dquots = f2fs_get_dquots,
2568 .evict_inode = f2fs_evict_inode,
2569 .put_super = f2fs_put_super,
2570 .sync_fs = f2fs_sync_fs,
2571 .freeze_fs = f2fs_freeze,
2572 .unfreeze_fs = f2fs_unfreeze,
2573 .statfs = f2fs_statfs,
2574 .remount_fs = f2fs_remount,
2577 #ifdef CONFIG_FS_ENCRYPTION
2578 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
2580 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2581 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2585 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
2588 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2591 * Encrypting the root directory is not allowed because fsck
2592 * expects lost+found directory to exist and remain unencrypted
2593 * if LOST_FOUND feature is enabled.
2596 if (f2fs_sb_has_lost_found(sbi) &&
2597 inode->i_ino == F2FS_ROOT_INO(sbi))
2600 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2601 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2602 ctx, len, fs_data, XATTR_CREATE);
2605 static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
2607 return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
2610 static bool f2fs_has_stable_inodes(struct super_block *sb)
2615 static void f2fs_get_ino_and_lblk_bits(struct super_block *sb,
2616 int *ino_bits_ret, int *lblk_bits_ret)
2618 *ino_bits_ret = 8 * sizeof(nid_t);
2619 *lblk_bits_ret = 8 * sizeof(block_t);
2622 static int f2fs_get_num_devices(struct super_block *sb)
2624 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2626 if (f2fs_is_multi_device(sbi))
2627 return sbi->s_ndevs;
2631 static void f2fs_get_devices(struct super_block *sb,
2632 struct request_queue **devs)
2634 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2637 for (i = 0; i < sbi->s_ndevs; i++)
2638 devs[i] = bdev_get_queue(FDEV(i).bdev);
2641 static const struct fscrypt_operations f2fs_cryptops = {
2642 .key_prefix = "f2fs:",
2643 .get_context = f2fs_get_context,
2644 .set_context = f2fs_set_context,
2645 .get_dummy_policy = f2fs_get_dummy_policy,
2646 .empty_dir = f2fs_empty_dir,
2647 .max_namelen = F2FS_NAME_LEN,
2648 .has_stable_inodes = f2fs_has_stable_inodes,
2649 .get_ino_and_lblk_bits = f2fs_get_ino_and_lblk_bits,
2650 .get_num_devices = f2fs_get_num_devices,
2651 .get_devices = f2fs_get_devices,
2655 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
2656 u64 ino, u32 generation)
2658 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2659 struct inode *inode;
2661 if (f2fs_check_nid_range(sbi, ino))
2662 return ERR_PTR(-ESTALE);
2665 * f2fs_iget isn't quite right if the inode is currently unallocated!
2666 * However f2fs_iget currently does appropriate checks to handle stale
2667 * inodes so everything is OK.
2669 inode = f2fs_iget(sb, ino);
2671 return ERR_CAST(inode);
2672 if (unlikely(generation && inode->i_generation != generation)) {
2673 /* we didn't find the right inode.. */
2675 return ERR_PTR(-ESTALE);
2680 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
2681 int fh_len, int fh_type)
2683 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
2684 f2fs_nfs_get_inode);
2687 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
2688 int fh_len, int fh_type)
2690 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
2691 f2fs_nfs_get_inode);
2694 static const struct export_operations f2fs_export_ops = {
2695 .fh_to_dentry = f2fs_fh_to_dentry,
2696 .fh_to_parent = f2fs_fh_to_parent,
2697 .get_parent = f2fs_get_parent,
2700 static loff_t max_file_blocks(void)
2703 loff_t leaf_count = DEF_ADDRS_PER_BLOCK;
2706 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
2707 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
2708 * space in inode.i_addr, it will be more safe to reassign
2712 /* two direct node blocks */
2713 result += (leaf_count * 2);
2715 /* two indirect node blocks */
2716 leaf_count *= NIDS_PER_BLOCK;
2717 result += (leaf_count * 2);
2719 /* one double indirect node block */
2720 leaf_count *= NIDS_PER_BLOCK;
2721 result += leaf_count;
2726 static int __f2fs_commit_super(struct buffer_head *bh,
2727 struct f2fs_super_block *super)
2731 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
2732 set_buffer_dirty(bh);
2735 /* it's rare case, we can do fua all the time */
2736 return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
2739 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
2740 struct buffer_head *bh)
2742 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2743 (bh->b_data + F2FS_SUPER_OFFSET);
2744 struct super_block *sb = sbi->sb;
2745 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
2746 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
2747 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
2748 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
2749 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
2750 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
2751 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
2752 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
2753 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
2754 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
2755 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2756 u32 segment_count = le32_to_cpu(raw_super->segment_count);
2757 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2758 u64 main_end_blkaddr = main_blkaddr +
2759 (segment_count_main << log_blocks_per_seg);
2760 u64 seg_end_blkaddr = segment0_blkaddr +
2761 (segment_count << log_blocks_per_seg);
2763 if (segment0_blkaddr != cp_blkaddr) {
2764 f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
2765 segment0_blkaddr, cp_blkaddr);
2769 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
2771 f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
2772 cp_blkaddr, sit_blkaddr,
2773 segment_count_ckpt << log_blocks_per_seg);
2777 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
2779 f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
2780 sit_blkaddr, nat_blkaddr,
2781 segment_count_sit << log_blocks_per_seg);
2785 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
2787 f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
2788 nat_blkaddr, ssa_blkaddr,
2789 segment_count_nat << log_blocks_per_seg);
2793 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
2795 f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
2796 ssa_blkaddr, main_blkaddr,
2797 segment_count_ssa << log_blocks_per_seg);
2801 if (main_end_blkaddr > seg_end_blkaddr) {
2802 f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
2803 main_blkaddr, seg_end_blkaddr,
2804 segment_count_main << log_blocks_per_seg);
2806 } else if (main_end_blkaddr < seg_end_blkaddr) {
2810 /* fix in-memory information all the time */
2811 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
2812 segment0_blkaddr) >> log_blocks_per_seg);
2814 if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
2815 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2818 err = __f2fs_commit_super(bh, NULL);
2819 res = err ? "failed" : "done";
2821 f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
2822 res, main_blkaddr, seg_end_blkaddr,
2823 segment_count_main << log_blocks_per_seg);
2830 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
2831 struct buffer_head *bh)
2833 block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
2834 block_t total_sections, blocks_per_seg;
2835 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2836 (bh->b_data + F2FS_SUPER_OFFSET);
2837 size_t crc_offset = 0;
2840 if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
2841 f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
2842 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
2846 /* Check checksum_offset and crc in superblock */
2847 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
2848 crc_offset = le32_to_cpu(raw_super->checksum_offset);
2850 offsetof(struct f2fs_super_block, crc)) {
2851 f2fs_info(sbi, "Invalid SB checksum offset: %zu",
2853 return -EFSCORRUPTED;
2855 crc = le32_to_cpu(raw_super->crc);
2856 if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
2857 f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
2858 return -EFSCORRUPTED;
2862 /* Currently, support only 4KB page cache size */
2863 if (F2FS_BLKSIZE != PAGE_SIZE) {
2864 f2fs_info(sbi, "Invalid page_cache_size (%lu), supports only 4KB",
2866 return -EFSCORRUPTED;
2869 /* Currently, support only 4KB block size */
2870 if (le32_to_cpu(raw_super->log_blocksize) != F2FS_BLKSIZE_BITS) {
2871 f2fs_info(sbi, "Invalid log_blocksize (%u), supports only %u",
2872 le32_to_cpu(raw_super->log_blocksize),
2874 return -EFSCORRUPTED;
2877 /* check log blocks per segment */
2878 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
2879 f2fs_info(sbi, "Invalid log blocks per segment (%u)",
2880 le32_to_cpu(raw_super->log_blocks_per_seg));
2881 return -EFSCORRUPTED;
2884 /* Currently, support 512/1024/2048/4096 bytes sector size */
2885 if (le32_to_cpu(raw_super->log_sectorsize) >
2886 F2FS_MAX_LOG_SECTOR_SIZE ||
2887 le32_to_cpu(raw_super->log_sectorsize) <
2888 F2FS_MIN_LOG_SECTOR_SIZE) {
2889 f2fs_info(sbi, "Invalid log sectorsize (%u)",
2890 le32_to_cpu(raw_super->log_sectorsize));
2891 return -EFSCORRUPTED;
2893 if (le32_to_cpu(raw_super->log_sectors_per_block) +
2894 le32_to_cpu(raw_super->log_sectorsize) !=
2895 F2FS_MAX_LOG_SECTOR_SIZE) {
2896 f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
2897 le32_to_cpu(raw_super->log_sectors_per_block),
2898 le32_to_cpu(raw_super->log_sectorsize));
2899 return -EFSCORRUPTED;
2902 segment_count = le32_to_cpu(raw_super->segment_count);
2903 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2904 segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
2905 secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
2906 total_sections = le32_to_cpu(raw_super->section_count);
2908 /* blocks_per_seg should be 512, given the above check */
2909 blocks_per_seg = 1 << le32_to_cpu(raw_super->log_blocks_per_seg);
2911 if (segment_count > F2FS_MAX_SEGMENT ||
2912 segment_count < F2FS_MIN_SEGMENTS) {
2913 f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
2914 return -EFSCORRUPTED;
2917 if (total_sections > segment_count_main || total_sections < 1 ||
2918 segs_per_sec > segment_count || !segs_per_sec) {
2919 f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
2920 segment_count, total_sections, segs_per_sec);
2921 return -EFSCORRUPTED;
2924 if (segment_count_main != total_sections * segs_per_sec) {
2925 f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
2926 segment_count_main, total_sections, segs_per_sec);
2927 return -EFSCORRUPTED;
2930 if ((segment_count / segs_per_sec) < total_sections) {
2931 f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
2932 segment_count, segs_per_sec, total_sections);
2933 return -EFSCORRUPTED;
2936 if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
2937 f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
2938 segment_count, le64_to_cpu(raw_super->block_count));
2939 return -EFSCORRUPTED;
2942 if (RDEV(0).path[0]) {
2943 block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
2946 while (i < MAX_DEVICES && RDEV(i).path[0]) {
2947 dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
2950 if (segment_count != dev_seg_count) {
2951 f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
2952 segment_count, dev_seg_count);
2953 return -EFSCORRUPTED;
2956 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
2957 !bdev_is_zoned(sbi->sb->s_bdev)) {
2958 f2fs_info(sbi, "Zoned block device path is missing");
2959 return -EFSCORRUPTED;
2963 if (secs_per_zone > total_sections || !secs_per_zone) {
2964 f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
2965 secs_per_zone, total_sections);
2966 return -EFSCORRUPTED;
2968 if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
2969 raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
2970 (le32_to_cpu(raw_super->extension_count) +
2971 raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
2972 f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
2973 le32_to_cpu(raw_super->extension_count),
2974 raw_super->hot_ext_count,
2975 F2FS_MAX_EXTENSION);
2976 return -EFSCORRUPTED;
2979 if (le32_to_cpu(raw_super->cp_payload) >=
2980 (blocks_per_seg - F2FS_CP_PACKS -
2981 NR_CURSEG_PERSIST_TYPE)) {
2982 f2fs_info(sbi, "Insane cp_payload (%u >= %u)",
2983 le32_to_cpu(raw_super->cp_payload),
2984 blocks_per_seg - F2FS_CP_PACKS -
2985 NR_CURSEG_PERSIST_TYPE);
2986 return -EFSCORRUPTED;
2989 /* check reserved ino info */
2990 if (le32_to_cpu(raw_super->node_ino) != 1 ||
2991 le32_to_cpu(raw_super->meta_ino) != 2 ||
2992 le32_to_cpu(raw_super->root_ino) != 3) {
2993 f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
2994 le32_to_cpu(raw_super->node_ino),
2995 le32_to_cpu(raw_super->meta_ino),
2996 le32_to_cpu(raw_super->root_ino));
2997 return -EFSCORRUPTED;
3000 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
3001 if (sanity_check_area_boundary(sbi, bh))
3002 return -EFSCORRUPTED;
3007 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
3009 unsigned int total, fsmeta;
3010 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3011 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
3012 unsigned int ovp_segments, reserved_segments;
3013 unsigned int main_segs, blocks_per_seg;
3014 unsigned int sit_segs, nat_segs;
3015 unsigned int sit_bitmap_size, nat_bitmap_size;
3016 unsigned int log_blocks_per_seg;
3017 unsigned int segment_count_main;
3018 unsigned int cp_pack_start_sum, cp_payload;
3019 block_t user_block_count, valid_user_blocks;
3020 block_t avail_node_count, valid_node_count;
3021 unsigned int nat_blocks, nat_bits_bytes, nat_bits_blocks;
3024 total = le32_to_cpu(raw_super->segment_count);
3025 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
3026 sit_segs = le32_to_cpu(raw_super->segment_count_sit);
3028 nat_segs = le32_to_cpu(raw_super->segment_count_nat);
3030 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
3031 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
3033 if (unlikely(fsmeta >= total))
3036 ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
3037 reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
3039 if (unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
3040 ovp_segments == 0 || reserved_segments == 0)) {
3041 f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
3045 user_block_count = le64_to_cpu(ckpt->user_block_count);
3046 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3047 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3048 if (!user_block_count || user_block_count >=
3049 segment_count_main << log_blocks_per_seg) {
3050 f2fs_err(sbi, "Wrong user_block_count: %u",
3055 valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
3056 if (valid_user_blocks > user_block_count) {
3057 f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
3058 valid_user_blocks, user_block_count);
3062 valid_node_count = le32_to_cpu(ckpt->valid_node_count);
3063 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
3064 if (valid_node_count > avail_node_count) {
3065 f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
3066 valid_node_count, avail_node_count);
3070 main_segs = le32_to_cpu(raw_super->segment_count_main);
3071 blocks_per_seg = sbi->blocks_per_seg;
3073 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3074 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
3075 le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
3077 for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
3078 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3079 le32_to_cpu(ckpt->cur_node_segno[j])) {
3080 f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
3082 le32_to_cpu(ckpt->cur_node_segno[i]));
3087 for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
3088 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
3089 le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
3091 for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3092 if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3093 le32_to_cpu(ckpt->cur_data_segno[j])) {
3094 f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3096 le32_to_cpu(ckpt->cur_data_segno[i]));
3101 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3102 for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3103 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3104 le32_to_cpu(ckpt->cur_data_segno[j])) {
3105 f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3107 le32_to_cpu(ckpt->cur_node_segno[i]));
3113 sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3114 nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3116 if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3117 nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3118 f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3119 sit_bitmap_size, nat_bitmap_size);
3123 cp_pack_start_sum = __start_sum_addr(sbi);
3124 cp_payload = __cp_payload(sbi);
3125 if (cp_pack_start_sum < cp_payload + 1 ||
3126 cp_pack_start_sum > blocks_per_seg - 1 -
3127 NR_CURSEG_PERSIST_TYPE) {
3128 f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3133 if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3134 le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3135 f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3136 "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3137 "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3138 le32_to_cpu(ckpt->checksum_offset));
3142 nat_blocks = nat_segs << log_blocks_per_seg;
3143 nat_bits_bytes = nat_blocks / BITS_PER_BYTE;
3144 nat_bits_blocks = F2FS_BLK_ALIGN((nat_bits_bytes << 1) + 8);
3145 if (__is_set_ckpt_flags(ckpt, CP_NAT_BITS_FLAG) &&
3146 (cp_payload + F2FS_CP_PACKS +
3147 NR_CURSEG_PERSIST_TYPE + nat_bits_blocks >= blocks_per_seg)) {
3148 f2fs_warn(sbi, "Insane cp_payload: %u, nat_bits_blocks: %u)",
3149 cp_payload, nat_bits_blocks);
3153 if (unlikely(f2fs_cp_error(sbi))) {
3154 f2fs_err(sbi, "A bug case: need to run fsck");
3160 static void init_sb_info(struct f2fs_sb_info *sbi)
3162 struct f2fs_super_block *raw_super = sbi->raw_super;
3165 sbi->log_sectors_per_block =
3166 le32_to_cpu(raw_super->log_sectors_per_block);
3167 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3168 sbi->blocksize = 1 << sbi->log_blocksize;
3169 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3170 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
3171 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3172 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3173 sbi->total_sections = le32_to_cpu(raw_super->section_count);
3174 sbi->total_node_count =
3175 (le32_to_cpu(raw_super->segment_count_nat) / 2)
3176 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
3177 sbi->root_ino_num = le32_to_cpu(raw_super->root_ino);
3178 sbi->node_ino_num = le32_to_cpu(raw_super->node_ino);
3179 sbi->meta_ino_num = le32_to_cpu(raw_super->meta_ino);
3180 sbi->cur_victim_sec = NULL_SECNO;
3181 sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3182 sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3183 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3184 sbi->migration_granularity = sbi->segs_per_sec;
3186 sbi->dir_level = DEF_DIR_LEVEL;
3187 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3188 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3189 sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3190 sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3191 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3192 sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3193 DEF_UMOUNT_DISCARD_TIMEOUT;
3194 clear_sbi_flag(sbi, SBI_NEED_FSCK);
3196 for (i = 0; i < NR_COUNT_TYPE; i++)
3197 atomic_set(&sbi->nr_pages[i], 0);
3199 for (i = 0; i < META; i++)
3200 atomic_set(&sbi->wb_sync_req[i], 0);
3202 INIT_LIST_HEAD(&sbi->s_list);
3203 mutex_init(&sbi->umount_mutex);
3204 init_rwsem(&sbi->io_order_lock);
3205 spin_lock_init(&sbi->cp_lock);
3207 sbi->dirty_device = 0;
3208 spin_lock_init(&sbi->dev_lock);
3210 init_rwsem(&sbi->sb_lock);
3211 init_rwsem(&sbi->pin_sem);
3214 static int init_percpu_info(struct f2fs_sb_info *sbi)
3218 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3222 err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3225 percpu_counter_destroy(&sbi->alloc_valid_block_count);
3230 #ifdef CONFIG_BLK_DEV_ZONED
3232 struct f2fs_report_zones_args {
3233 struct f2fs_sb_info *sbi;
3234 struct f2fs_dev_info *dev;
3237 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3240 struct f2fs_report_zones_args *rz_args = data;
3241 block_t unusable_blocks = (zone->len - zone->capacity) >>
3242 F2FS_LOG_SECTORS_PER_BLOCK;
3244 if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3247 set_bit(idx, rz_args->dev->blkz_seq);
3248 if (!rz_args->sbi->unusable_blocks_per_sec) {
3249 rz_args->sbi->unusable_blocks_per_sec = unusable_blocks;
3252 if (rz_args->sbi->unusable_blocks_per_sec != unusable_blocks) {
3253 f2fs_err(rz_args->sbi, "F2FS supports single zone capacity\n");
3259 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3261 struct block_device *bdev = FDEV(devi).bdev;
3262 sector_t nr_sectors = bdev->bd_part->nr_sects;
3263 struct f2fs_report_zones_args rep_zone_arg;
3266 if (!f2fs_sb_has_blkzoned(sbi))
3269 if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3270 SECTOR_TO_BLOCK(bdev_zone_sectors(bdev)))
3272 sbi->blocks_per_blkz = SECTOR_TO_BLOCK(bdev_zone_sectors(bdev));
3273 if (sbi->log_blocks_per_blkz && sbi->log_blocks_per_blkz !=
3274 __ilog2_u32(sbi->blocks_per_blkz))
3276 sbi->log_blocks_per_blkz = __ilog2_u32(sbi->blocks_per_blkz);
3277 FDEV(devi).nr_blkz = SECTOR_TO_BLOCK(nr_sectors) >>
3278 sbi->log_blocks_per_blkz;
3279 if (nr_sectors & (bdev_zone_sectors(bdev) - 1))
3280 FDEV(devi).nr_blkz++;
3282 FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3283 BITS_TO_LONGS(FDEV(devi).nr_blkz)
3284 * sizeof(unsigned long),
3286 if (!FDEV(devi).blkz_seq)
3289 rep_zone_arg.sbi = sbi;
3290 rep_zone_arg.dev = &FDEV(devi);
3292 ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3301 * Read f2fs raw super block.
3302 * Because we have two copies of super block, so read both of them
3303 * to get the first valid one. If any one of them is broken, we pass
3304 * them recovery flag back to the caller.
3306 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3307 struct f2fs_super_block **raw_super,
3308 int *valid_super_block, int *recovery)
3310 struct super_block *sb = sbi->sb;
3312 struct buffer_head *bh;
3313 struct f2fs_super_block *super;
3316 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3320 for (block = 0; block < 2; block++) {
3321 bh = sb_bread(sb, block);
3323 f2fs_err(sbi, "Unable to read %dth superblock",
3330 /* sanity checking of raw super */
3331 err = sanity_check_raw_super(sbi, bh);
3333 f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3341 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3343 *valid_super_block = block;
3349 /* No valid superblock */
3358 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
3360 struct buffer_head *bh;
3364 if ((recover && f2fs_readonly(sbi->sb)) ||
3365 bdev_read_only(sbi->sb->s_bdev)) {
3366 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3370 /* we should update superblock crc here */
3371 if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
3372 crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
3373 offsetof(struct f2fs_super_block, crc));
3374 F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
3377 /* write back-up superblock first */
3378 bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
3381 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3384 /* if we are in recovery path, skip writing valid superblock */
3388 /* write current valid superblock */
3389 bh = sb_bread(sbi->sb, sbi->valid_super_block);
3392 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3397 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
3399 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3400 unsigned int max_devices = MAX_DEVICES;
3403 /* Initialize single device information */
3404 if (!RDEV(0).path[0]) {
3405 if (!bdev_is_zoned(sbi->sb->s_bdev))
3411 * Initialize multiple devices information, or single
3412 * zoned block device information.
3414 sbi->devs = f2fs_kzalloc(sbi,
3415 array_size(max_devices,
3416 sizeof(struct f2fs_dev_info)),
3421 for (i = 0; i < max_devices; i++) {
3423 if (i > 0 && !RDEV(i).path[0])
3426 if (max_devices == 1) {
3427 /* Single zoned block device mount */
3429 blkdev_get_by_dev(sbi->sb->s_bdev->bd_dev,
3430 sbi->sb->s_mode, sbi->sb->s_type);
3432 /* Multi-device mount */
3433 memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
3434 FDEV(i).total_segments =
3435 le32_to_cpu(RDEV(i).total_segments);
3437 FDEV(i).start_blk = 0;
3438 FDEV(i).end_blk = FDEV(i).start_blk +
3439 (FDEV(i).total_segments <<
3440 sbi->log_blocks_per_seg) - 1 +
3441 le32_to_cpu(raw_super->segment0_blkaddr);
3443 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
3444 FDEV(i).end_blk = FDEV(i).start_blk +
3445 (FDEV(i).total_segments <<
3446 sbi->log_blocks_per_seg) - 1;
3448 FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
3449 sbi->sb->s_mode, sbi->sb->s_type);
3451 if (IS_ERR(FDEV(i).bdev))
3452 return PTR_ERR(FDEV(i).bdev);
3454 /* to release errored devices */
3455 sbi->s_ndevs = i + 1;
3457 #ifdef CONFIG_BLK_DEV_ZONED
3458 if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
3459 !f2fs_sb_has_blkzoned(sbi)) {
3460 f2fs_err(sbi, "Zoned block device feature not enabled\n");
3463 if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
3464 if (init_blkz_info(sbi, i)) {
3465 f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
3468 if (max_devices == 1)
3470 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
3472 FDEV(i).total_segments,
3473 FDEV(i).start_blk, FDEV(i).end_blk,
3474 bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
3475 "Host-aware" : "Host-managed");
3479 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
3481 FDEV(i).total_segments,
3482 FDEV(i).start_blk, FDEV(i).end_blk);
3485 "IO Block Size: %8d KB", F2FS_IO_SIZE_KB(sbi));
3489 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
3491 #ifdef CONFIG_UNICODE
3492 if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
3493 const struct f2fs_sb_encodings *encoding_info;
3494 struct unicode_map *encoding;
3495 __u16 encoding_flags;
3497 if (f2fs_sb_has_encrypt(sbi)) {
3499 "Can't mount with encoding and encryption");
3503 if (f2fs_sb_read_encoding(sbi->raw_super, &encoding_info,
3506 "Encoding requested by superblock is unknown");
3510 encoding = utf8_load(encoding_info->version);
3511 if (IS_ERR(encoding)) {
3513 "can't mount with superblock charset: %s-%s "
3514 "not supported by the kernel. flags: 0x%x.",
3515 encoding_info->name, encoding_info->version,
3517 return PTR_ERR(encoding);
3519 f2fs_info(sbi, "Using encoding defined by superblock: "
3520 "%s-%s with flags 0x%hx", encoding_info->name,
3521 encoding_info->version?:"\b", encoding_flags);
3523 sbi->sb->s_encoding = encoding;
3524 sbi->sb->s_encoding_flags = encoding_flags;
3525 sbi->sb->s_d_op = &f2fs_dentry_ops;
3528 if (f2fs_sb_has_casefold(sbi)) {
3529 f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
3536 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
3538 struct f2fs_sm_info *sm_i = SM_I(sbi);
3540 /* adjust parameters according to the volume size */
3541 if (sm_i->main_segments <= SMALL_VOLUME_SEGMENTS) {
3542 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
3543 sm_i->dcc_info->discard_granularity = 1;
3544 sm_i->ipu_policy = 1 << F2FS_IPU_FORCE;
3547 sbi->readdir_ra = 1;
3550 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
3552 struct f2fs_sb_info *sbi;
3553 struct f2fs_super_block *raw_super;
3556 bool skip_recovery = false, need_fsck = false;
3557 char *options = NULL;
3558 int recovery, i, valid_super_block;
3559 struct curseg_info *seg_i;
3565 valid_super_block = -1;
3568 /* allocate memory for f2fs-specific super block info */
3569 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
3575 /* Load the checksum driver */
3576 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
3577 if (IS_ERR(sbi->s_chksum_driver)) {
3578 f2fs_err(sbi, "Cannot load crc32 driver.");
3579 err = PTR_ERR(sbi->s_chksum_driver);
3580 sbi->s_chksum_driver = NULL;
3584 /* set a block size */
3585 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
3586 f2fs_err(sbi, "unable to set blocksize");
3590 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
3595 sb->s_fs_info = sbi;
3596 sbi->raw_super = raw_super;
3598 /* precompute checksum seed for metadata */
3599 if (f2fs_sb_has_inode_chksum(sbi))
3600 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
3601 sizeof(raw_super->uuid));
3603 default_options(sbi);
3604 /* parse mount options */
3605 options = kstrdup((const char *)data, GFP_KERNEL);
3606 if (data && !options) {
3611 err = parse_options(sb, options, false);
3615 sbi->max_file_blocks = max_file_blocks();
3616 sb->s_maxbytes = sbi->max_file_blocks <<
3617 le32_to_cpu(raw_super->log_blocksize);
3618 sb->s_max_links = F2FS_LINK_MAX;
3620 err = f2fs_setup_casefold(sbi);
3625 sb->dq_op = &f2fs_quota_operations;
3626 sb->s_qcop = &f2fs_quotactl_ops;
3627 sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
3629 if (f2fs_sb_has_quota_ino(sbi)) {
3630 for (i = 0; i < MAXQUOTAS; i++) {
3631 if (f2fs_qf_ino(sbi->sb, i))
3632 sbi->nquota_files++;
3637 sb->s_op = &f2fs_sops;
3638 #ifdef CONFIG_FS_ENCRYPTION
3639 sb->s_cop = &f2fs_cryptops;
3641 #ifdef CONFIG_FS_VERITY
3642 sb->s_vop = &f2fs_verityops;
3644 sb->s_xattr = f2fs_xattr_handlers;
3645 sb->s_export_op = &f2fs_export_ops;
3646 sb->s_magic = F2FS_SUPER_MAGIC;
3647 sb->s_time_gran = 1;
3648 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
3649 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
3650 memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
3651 sb->s_iflags |= SB_I_CGROUPWB;
3653 /* init f2fs-specific super block info */
3654 sbi->valid_super_block = valid_super_block;
3655 init_rwsem(&sbi->gc_lock);
3656 mutex_init(&sbi->writepages);
3657 mutex_init(&sbi->cp_mutex);
3658 init_rwsem(&sbi->node_write);
3659 init_rwsem(&sbi->node_change);
3661 /* disallow all the data/node/meta page writes */
3662 set_sbi_flag(sbi, SBI_POR_DOING);
3663 spin_lock_init(&sbi->stat_lock);
3665 /* init iostat info */
3666 spin_lock_init(&sbi->iostat_lock);
3667 sbi->iostat_enable = false;
3668 sbi->iostat_period_ms = DEFAULT_IOSTAT_PERIOD_MS;
3670 for (i = 0; i < NR_PAGE_TYPE; i++) {
3671 int n = (i == META) ? 1: NR_TEMP_TYPE;
3677 sizeof(struct f2fs_bio_info)),
3679 if (!sbi->write_io[i]) {
3684 for (j = HOT; j < n; j++) {
3685 init_rwsem(&sbi->write_io[i][j].io_rwsem);
3686 sbi->write_io[i][j].sbi = sbi;
3687 sbi->write_io[i][j].bio = NULL;
3688 spin_lock_init(&sbi->write_io[i][j].io_lock);
3689 INIT_LIST_HEAD(&sbi->write_io[i][j].io_list);
3690 INIT_LIST_HEAD(&sbi->write_io[i][j].bio_list);
3691 init_rwsem(&sbi->write_io[i][j].bio_list_lock);
3695 init_rwsem(&sbi->cp_rwsem);
3696 init_rwsem(&sbi->quota_sem);
3697 init_waitqueue_head(&sbi->cp_wait);
3700 err = init_percpu_info(sbi);
3704 if (F2FS_IO_ALIGNED(sbi)) {
3705 sbi->write_io_dummy =
3706 mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
3707 if (!sbi->write_io_dummy) {
3713 /* init per sbi slab cache */
3714 err = f2fs_init_xattr_caches(sbi);
3717 err = f2fs_init_page_array_cache(sbi);
3719 goto free_xattr_cache;
3721 /* get an inode for meta space */
3722 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
3723 if (IS_ERR(sbi->meta_inode)) {
3724 f2fs_err(sbi, "Failed to read F2FS meta data inode");
3725 err = PTR_ERR(sbi->meta_inode);
3726 goto free_page_array_cache;
3729 err = f2fs_get_valid_checkpoint(sbi);
3731 f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
3732 goto free_meta_inode;
3735 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
3736 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3737 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
3738 set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
3739 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
3742 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
3743 set_sbi_flag(sbi, SBI_NEED_FSCK);
3745 /* Initialize device list */
3746 err = f2fs_scan_devices(sbi);
3748 f2fs_err(sbi, "Failed to find devices");
3752 err = f2fs_init_post_read_wq(sbi);
3754 f2fs_err(sbi, "Failed to initialize post read workqueue");
3758 sbi->total_valid_node_count =
3759 le32_to_cpu(sbi->ckpt->valid_node_count);
3760 percpu_counter_set(&sbi->total_valid_inode_count,
3761 le32_to_cpu(sbi->ckpt->valid_inode_count));
3762 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
3763 sbi->total_valid_block_count =
3764 le64_to_cpu(sbi->ckpt->valid_block_count);
3765 sbi->last_valid_block_count = sbi->total_valid_block_count;
3766 sbi->reserved_blocks = 0;
3767 sbi->current_reserved_blocks = 0;
3768 limit_reserve_root(sbi);
3769 adjust_unusable_cap_perc(sbi);
3771 for (i = 0; i < NR_INODE_TYPE; i++) {
3772 INIT_LIST_HEAD(&sbi->inode_list[i]);
3773 spin_lock_init(&sbi->inode_lock[i]);
3775 mutex_init(&sbi->flush_lock);
3777 f2fs_init_extent_cache_info(sbi);
3779 f2fs_init_ino_entry_info(sbi);
3781 f2fs_init_fsync_node_info(sbi);
3783 /* setup f2fs internal modules */
3784 err = f2fs_build_segment_manager(sbi);
3786 f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
3790 err = f2fs_build_node_manager(sbi);
3792 f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
3797 err = adjust_reserved_segment(sbi);
3801 /* For write statistics */
3802 if (sb->s_bdev->bd_part)
3803 sbi->sectors_written_start =
3804 (u64)part_stat_read(sb->s_bdev->bd_part,
3805 sectors[STAT_WRITE]);
3807 /* Read accumulated write IO statistics if exists */
3808 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
3809 if (__exist_node_summaries(sbi))
3810 sbi->kbytes_written =
3811 le64_to_cpu(seg_i->journal->info.kbytes_written);
3813 f2fs_build_gc_manager(sbi);
3815 err = f2fs_build_stats(sbi);
3819 /* get an inode for node space */
3820 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
3821 if (IS_ERR(sbi->node_inode)) {
3822 f2fs_err(sbi, "Failed to read node inode");
3823 err = PTR_ERR(sbi->node_inode);
3827 /* read root inode and dentry */
3828 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
3830 f2fs_err(sbi, "Failed to read root inode");
3831 err = PTR_ERR(root);
3832 goto free_node_inode;
3834 if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
3835 !root->i_size || !root->i_nlink) {
3838 goto free_node_inode;
3841 sb->s_root = d_make_root(root); /* allocate root dentry */
3844 goto free_node_inode;
3847 err = f2fs_register_sysfs(sbi);
3849 goto free_root_inode;
3852 /* Enable quota usage during mount */
3853 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
3854 err = f2fs_enable_quotas(sb);
3856 f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
3859 /* if there are any orphan inodes, free them */
3860 err = f2fs_recover_orphan_inodes(sbi);
3864 if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
3865 goto reset_checkpoint;
3867 /* recover fsynced data */
3868 if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
3869 !test_opt(sbi, NORECOVERY)) {
3871 * mount should be failed, when device has readonly mode, and
3872 * previous checkpoint was not done by clean system shutdown.
3874 if (f2fs_hw_is_readonly(sbi)) {
3875 if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
3877 f2fs_err(sbi, "Need to recover fsync data, but write access unavailable");
3880 f2fs_info(sbi, "write access unavailable, skipping recovery");
3881 goto reset_checkpoint;
3885 set_sbi_flag(sbi, SBI_NEED_FSCK);
3888 goto reset_checkpoint;
3890 err = f2fs_recover_fsync_data(sbi, false);
3893 skip_recovery = true;
3895 f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
3900 err = f2fs_recover_fsync_data(sbi, true);
3902 if (!f2fs_readonly(sb) && err > 0) {
3904 f2fs_err(sbi, "Need to recover fsync data");
3910 * If the f2fs is not readonly and fsync data recovery succeeds,
3911 * check zoned block devices' write pointer consistency.
3913 if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
3914 err = f2fs_check_write_pointer(sbi);
3920 f2fs_init_inmem_curseg(sbi);
3922 /* f2fs_recover_fsync_data() cleared this already */
3923 clear_sbi_flag(sbi, SBI_POR_DOING);
3925 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
3926 err = f2fs_disable_checkpoint(sbi);
3928 goto sync_free_meta;
3929 } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
3930 f2fs_enable_checkpoint(sbi);
3934 * If filesystem is not mounted as read-only then
3935 * do start the gc_thread.
3937 if (F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF && !f2fs_readonly(sb)) {
3938 /* After POR, we can run background GC thread.*/
3939 err = f2fs_start_gc_thread(sbi);
3941 goto sync_free_meta;
3945 /* recover broken superblock */
3947 err = f2fs_commit_super(sbi, true);
3948 f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
3949 sbi->valid_super_block ? 1 : 2, err);
3952 f2fs_join_shrinker(sbi);
3954 f2fs_tuning_parameters(sbi);
3956 f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
3957 cur_cp_version(F2FS_CKPT(sbi)));
3958 f2fs_update_time(sbi, CP_TIME);
3959 f2fs_update_time(sbi, REQ_TIME);
3960 clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
3964 /* safe to flush all the data */
3965 sync_filesystem(sbi->sb);
3970 f2fs_truncate_quota_inode_pages(sb);
3971 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
3972 f2fs_quota_off_umount(sbi->sb);
3975 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
3976 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
3977 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
3978 * falls into an infinite loop in f2fs_sync_meta_pages().
3980 truncate_inode_pages_final(META_MAPPING(sbi));
3981 /* evict some inodes being cached by GC */
3983 f2fs_unregister_sysfs(sbi);
3988 f2fs_release_ino_entry(sbi, true);
3989 truncate_inode_pages_final(NODE_MAPPING(sbi));
3990 iput(sbi->node_inode);
3991 sbi->node_inode = NULL;
3993 f2fs_destroy_stats(sbi);
3995 f2fs_destroy_node_manager(sbi);
3997 f2fs_destroy_segment_manager(sbi);
3998 f2fs_destroy_post_read_wq(sbi);
4000 destroy_device_list(sbi);
4003 make_bad_inode(sbi->meta_inode);
4004 iput(sbi->meta_inode);
4005 sbi->meta_inode = NULL;
4006 free_page_array_cache:
4007 f2fs_destroy_page_array_cache(sbi);
4009 f2fs_destroy_xattr_caches(sbi);
4011 mempool_destroy(sbi->write_io_dummy);
4013 destroy_percpu_info(sbi);
4015 for (i = 0; i < NR_PAGE_TYPE; i++)
4016 kvfree(sbi->write_io[i]);
4018 #ifdef CONFIG_UNICODE
4019 utf8_unload(sb->s_encoding);
4020 sb->s_encoding = NULL;
4024 for (i = 0; i < MAXQUOTAS; i++)
4025 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
4027 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
4032 if (sbi->s_chksum_driver)
4033 crypto_free_shash(sbi->s_chksum_driver);
4036 /* give only one another chance */
4037 if (retry_cnt > 0 && skip_recovery) {
4039 shrink_dcache_sb(sb);
4045 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
4046 const char *dev_name, void *data)
4048 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
4051 static void kill_f2fs_super(struct super_block *sb)
4054 struct f2fs_sb_info *sbi = F2FS_SB(sb);
4056 set_sbi_flag(sbi, SBI_IS_CLOSE);
4057 f2fs_stop_gc_thread(sbi);
4058 f2fs_stop_discard_thread(sbi);
4060 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
4061 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4062 struct cp_control cpc = {
4063 .reason = CP_UMOUNT,
4065 f2fs_write_checkpoint(sbi, &cpc);
4068 if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
4069 sb->s_flags &= ~SB_RDONLY;
4071 kill_block_super(sb);
4074 static struct file_system_type f2fs_fs_type = {
4075 .owner = THIS_MODULE,
4077 .mount = f2fs_mount,
4078 .kill_sb = kill_f2fs_super,
4079 .fs_flags = FS_REQUIRES_DEV,
4081 MODULE_ALIAS_FS("f2fs");
4083 static int __init init_inodecache(void)
4085 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
4086 sizeof(struct f2fs_inode_info), 0,
4087 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
4088 if (!f2fs_inode_cachep)
4093 static void destroy_inodecache(void)
4096 * Make sure all delayed rcu free inodes are flushed before we
4100 kmem_cache_destroy(f2fs_inode_cachep);
4103 static int __init init_f2fs_fs(void)
4107 if (PAGE_SIZE != F2FS_BLKSIZE) {
4108 printk("F2FS not supported on PAGE_SIZE(%lu) != %d\n",
4109 PAGE_SIZE, F2FS_BLKSIZE);
4113 f2fs_build_trace_ios();
4115 err = init_inodecache();
4118 err = f2fs_create_node_manager_caches();
4120 goto free_inodecache;
4121 err = f2fs_create_segment_manager_caches();
4123 goto free_node_manager_caches;
4124 err = f2fs_create_checkpoint_caches();
4126 goto free_segment_manager_caches;
4127 err = f2fs_create_recovery_cache();
4129 goto free_checkpoint_caches;
4130 err = f2fs_create_extent_cache();
4132 goto free_recovery_cache;
4133 err = f2fs_create_garbage_collection_cache();
4135 goto free_extent_cache;
4136 err = f2fs_init_sysfs();
4138 goto free_garbage_collection_cache;
4139 err = register_shrinker(&f2fs_shrinker_info);
4142 err = register_filesystem(&f2fs_fs_type);
4145 f2fs_create_root_stats();
4146 err = f2fs_init_post_read_processing();
4148 goto free_root_stats;
4149 err = f2fs_init_bio_entry_cache();
4151 goto free_post_read;
4152 err = f2fs_init_bioset();
4154 goto free_bio_enrty_cache;
4155 err = f2fs_init_compress_mempool();
4158 err = f2fs_init_compress_cache();
4160 goto free_compress_mempool;
4162 free_compress_mempool:
4163 f2fs_destroy_compress_mempool();
4165 f2fs_destroy_bioset();
4166 free_bio_enrty_cache:
4167 f2fs_destroy_bio_entry_cache();
4169 f2fs_destroy_post_read_processing();
4171 f2fs_destroy_root_stats();
4172 unregister_filesystem(&f2fs_fs_type);
4174 unregister_shrinker(&f2fs_shrinker_info);
4177 free_garbage_collection_cache:
4178 f2fs_destroy_garbage_collection_cache();
4180 f2fs_destroy_extent_cache();
4181 free_recovery_cache:
4182 f2fs_destroy_recovery_cache();
4183 free_checkpoint_caches:
4184 f2fs_destroy_checkpoint_caches();
4185 free_segment_manager_caches:
4186 f2fs_destroy_segment_manager_caches();
4187 free_node_manager_caches:
4188 f2fs_destroy_node_manager_caches();
4190 destroy_inodecache();
4195 static void __exit exit_f2fs_fs(void)
4197 f2fs_destroy_compress_cache();
4198 f2fs_destroy_compress_mempool();
4199 f2fs_destroy_bioset();
4200 f2fs_destroy_bio_entry_cache();
4201 f2fs_destroy_post_read_processing();
4202 f2fs_destroy_root_stats();
4203 unregister_filesystem(&f2fs_fs_type);
4204 unregister_shrinker(&f2fs_shrinker_info);
4206 f2fs_destroy_garbage_collection_cache();
4207 f2fs_destroy_extent_cache();
4208 f2fs_destroy_recovery_cache();
4209 f2fs_destroy_checkpoint_caches();
4210 f2fs_destroy_segment_manager_caches();
4211 f2fs_destroy_node_manager_caches();
4212 destroy_inodecache();
4213 f2fs_destroy_trace_ios();
4216 module_init(init_f2fs_fs)
4217 module_exit(exit_f2fs_fs)
4219 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
4220 MODULE_DESCRIPTION("Flash Friendly File System");
4221 MODULE_LICENSE("GPL");
4222 MODULE_SOFTDEP("pre: crc32");