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>
28 #include <linux/zstd.h>
29 #include <linux/lz4.h>
37 #define CREATE_TRACE_POINTS
38 #include <trace/events/f2fs.h>
40 static struct kmem_cache *f2fs_inode_cachep;
42 #ifdef CONFIG_F2FS_FAULT_INJECTION
44 const char *f2fs_fault_name[FAULT_MAX] = {
45 [FAULT_KMALLOC] = "kmalloc",
46 [FAULT_KVMALLOC] = "kvmalloc",
47 [FAULT_PAGE_ALLOC] = "page alloc",
48 [FAULT_PAGE_GET] = "page get",
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_checkpoint_merge,
147 Opt_nocheckpoint_merge,
148 Opt_compress_algorithm,
149 Opt_compress_log_size,
150 Opt_compress_extension,
157 static match_table_t f2fs_tokens = {
158 {Opt_gc_background, "background_gc=%s"},
159 {Opt_disable_roll_forward, "disable_roll_forward"},
160 {Opt_norecovery, "norecovery"},
161 {Opt_discard, "discard"},
162 {Opt_nodiscard, "nodiscard"},
163 {Opt_noheap, "no_heap"},
165 {Opt_user_xattr, "user_xattr"},
166 {Opt_nouser_xattr, "nouser_xattr"},
168 {Opt_noacl, "noacl"},
169 {Opt_active_logs, "active_logs=%u"},
170 {Opt_disable_ext_identify, "disable_ext_identify"},
171 {Opt_inline_xattr, "inline_xattr"},
172 {Opt_noinline_xattr, "noinline_xattr"},
173 {Opt_inline_xattr_size, "inline_xattr_size=%u"},
174 {Opt_inline_data, "inline_data"},
175 {Opt_inline_dentry, "inline_dentry"},
176 {Opt_noinline_dentry, "noinline_dentry"},
177 {Opt_flush_merge, "flush_merge"},
178 {Opt_noflush_merge, "noflush_merge"},
179 {Opt_nobarrier, "nobarrier"},
180 {Opt_fastboot, "fastboot"},
181 {Opt_extent_cache, "extent_cache"},
182 {Opt_noextent_cache, "noextent_cache"},
183 {Opt_noinline_data, "noinline_data"},
184 {Opt_data_flush, "data_flush"},
185 {Opt_reserve_root, "reserve_root=%u"},
186 {Opt_resgid, "resgid=%u"},
187 {Opt_resuid, "resuid=%u"},
188 {Opt_mode, "mode=%s"},
189 {Opt_io_size_bits, "io_bits=%u"},
190 {Opt_fault_injection, "fault_injection=%u"},
191 {Opt_fault_type, "fault_type=%u"},
192 {Opt_lazytime, "lazytime"},
193 {Opt_nolazytime, "nolazytime"},
194 {Opt_quota, "quota"},
195 {Opt_noquota, "noquota"},
196 {Opt_usrquota, "usrquota"},
197 {Opt_grpquota, "grpquota"},
198 {Opt_prjquota, "prjquota"},
199 {Opt_usrjquota, "usrjquota=%s"},
200 {Opt_grpjquota, "grpjquota=%s"},
201 {Opt_prjjquota, "prjjquota=%s"},
202 {Opt_offusrjquota, "usrjquota="},
203 {Opt_offgrpjquota, "grpjquota="},
204 {Opt_offprjjquota, "prjjquota="},
205 {Opt_jqfmt_vfsold, "jqfmt=vfsold"},
206 {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
207 {Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
208 {Opt_whint, "whint_mode=%s"},
209 {Opt_alloc, "alloc_mode=%s"},
210 {Opt_fsync, "fsync_mode=%s"},
211 {Opt_test_dummy_encryption, "test_dummy_encryption=%s"},
212 {Opt_test_dummy_encryption, "test_dummy_encryption"},
213 {Opt_inlinecrypt, "inlinecrypt"},
214 {Opt_checkpoint_disable, "checkpoint=disable"},
215 {Opt_checkpoint_disable_cap, "checkpoint=disable:%u"},
216 {Opt_checkpoint_disable_cap_perc, "checkpoint=disable:%u%%"},
217 {Opt_checkpoint_enable, "checkpoint=enable"},
218 {Opt_checkpoint_merge, "checkpoint_merge"},
219 {Opt_nocheckpoint_merge, "nocheckpoint_merge"},
220 {Opt_compress_algorithm, "compress_algorithm=%s"},
221 {Opt_compress_log_size, "compress_log_size=%u"},
222 {Opt_compress_extension, "compress_extension=%s"},
223 {Opt_compress_chksum, "compress_chksum"},
224 {Opt_compress_mode, "compress_mode=%s"},
229 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...)
231 struct va_format vaf;
237 level = printk_get_level(fmt);
238 vaf.fmt = printk_skip_level(fmt);
240 printk("%c%cF2FS-fs (%s): %pV\n",
241 KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
246 #ifdef CONFIG_UNICODE
247 static const struct f2fs_sb_encodings {
251 } f2fs_sb_encoding_map[] = {
252 {F2FS_ENC_UTF8_12_1, "utf8", "12.1.0"},
255 static int f2fs_sb_read_encoding(const struct f2fs_super_block *sb,
256 const struct f2fs_sb_encodings **encoding,
259 __u16 magic = le16_to_cpu(sb->s_encoding);
262 for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
263 if (magic == f2fs_sb_encoding_map[i].magic)
266 if (i >= ARRAY_SIZE(f2fs_sb_encoding_map))
269 *encoding = &f2fs_sb_encoding_map[i];
270 *flags = le16_to_cpu(sb->s_encoding_flags);
276 static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
278 block_t limit = min((sbi->user_block_count << 1) / 1000,
279 sbi->user_block_count - sbi->reserved_blocks);
282 if (test_opt(sbi, RESERVE_ROOT) &&
283 F2FS_OPTION(sbi).root_reserved_blocks > limit) {
284 F2FS_OPTION(sbi).root_reserved_blocks = limit;
285 f2fs_info(sbi, "Reduce reserved blocks for root = %u",
286 F2FS_OPTION(sbi).root_reserved_blocks);
288 if (!test_opt(sbi, RESERVE_ROOT) &&
289 (!uid_eq(F2FS_OPTION(sbi).s_resuid,
290 make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
291 !gid_eq(F2FS_OPTION(sbi).s_resgid,
292 make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
293 f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
294 from_kuid_munged(&init_user_ns,
295 F2FS_OPTION(sbi).s_resuid),
296 from_kgid_munged(&init_user_ns,
297 F2FS_OPTION(sbi).s_resgid));
300 static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
302 if (!F2FS_OPTION(sbi).unusable_cap_perc)
305 if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
306 F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
308 F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
309 F2FS_OPTION(sbi).unusable_cap_perc;
311 f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
312 F2FS_OPTION(sbi).unusable_cap,
313 F2FS_OPTION(sbi).unusable_cap_perc);
316 static void init_once(void *foo)
318 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
320 inode_init_once(&fi->vfs_inode);
324 static const char * const quotatypes[] = INITQFNAMES;
325 #define QTYPE2NAME(t) (quotatypes[t])
326 static int f2fs_set_qf_name(struct super_block *sb, int qtype,
329 struct f2fs_sb_info *sbi = F2FS_SB(sb);
333 if (sb_any_quota_loaded(sb) && !F2FS_OPTION(sbi).s_qf_names[qtype]) {
334 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
337 if (f2fs_sb_has_quota_ino(sbi)) {
338 f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
342 qname = match_strdup(args);
344 f2fs_err(sbi, "Not enough memory for storing quotafile name");
347 if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
348 if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
351 f2fs_err(sbi, "%s quota file already specified",
355 if (strchr(qname, '/')) {
356 f2fs_err(sbi, "quotafile must be on filesystem root");
359 F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
367 static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
369 struct f2fs_sb_info *sbi = F2FS_SB(sb);
371 if (sb_any_quota_loaded(sb) && F2FS_OPTION(sbi).s_qf_names[qtype]) {
372 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
375 kfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
376 F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
380 static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
383 * We do the test below only for project quotas. 'usrquota' and
384 * 'grpquota' mount options are allowed even without quota feature
385 * to support legacy quotas in quota files.
387 if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi)) {
388 f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
391 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
392 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
393 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) {
394 if (test_opt(sbi, USRQUOTA) &&
395 F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
396 clear_opt(sbi, USRQUOTA);
398 if (test_opt(sbi, GRPQUOTA) &&
399 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
400 clear_opt(sbi, GRPQUOTA);
402 if (test_opt(sbi, PRJQUOTA) &&
403 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
404 clear_opt(sbi, PRJQUOTA);
406 if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
407 test_opt(sbi, PRJQUOTA)) {
408 f2fs_err(sbi, "old and new quota format mixing");
412 if (!F2FS_OPTION(sbi).s_jquota_fmt) {
413 f2fs_err(sbi, "journaled quota format not specified");
418 if (f2fs_sb_has_quota_ino(sbi) && F2FS_OPTION(sbi).s_jquota_fmt) {
419 f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
420 F2FS_OPTION(sbi).s_jquota_fmt = 0;
426 static int f2fs_set_test_dummy_encryption(struct super_block *sb,
428 const substring_t *arg,
431 struct f2fs_sb_info *sbi = F2FS_SB(sb);
432 #ifdef CONFIG_FS_ENCRYPTION
435 if (!f2fs_sb_has_encrypt(sbi)) {
436 f2fs_err(sbi, "Encrypt feature is off");
441 * This mount option is just for testing, and it's not worthwhile to
442 * implement the extra complexity (e.g. RCU protection) that would be
443 * needed to allow it to be set or changed during remount. We do allow
444 * it to be specified during remount, but only if there is no change.
446 if (is_remount && !F2FS_OPTION(sbi).dummy_enc_policy.policy) {
447 f2fs_warn(sbi, "Can't set test_dummy_encryption on remount");
450 err = fscrypt_set_test_dummy_encryption(
451 sb, arg->from, &F2FS_OPTION(sbi).dummy_enc_policy);
455 "Can't change test_dummy_encryption on remount");
456 else if (err == -EINVAL)
457 f2fs_warn(sbi, "Value of option \"%s\" is unrecognized",
460 f2fs_warn(sbi, "Error processing option \"%s\" [%d]",
464 f2fs_warn(sbi, "Test dummy encryption mode enabled");
466 f2fs_warn(sbi, "Test dummy encryption mount option ignored");
471 #ifdef CONFIG_F2FS_FS_COMPRESSION
472 #ifdef CONFIG_F2FS_FS_LZ4
473 static int f2fs_set_lz4hc_level(struct f2fs_sb_info *sbi, const char *str)
475 #ifdef CONFIG_F2FS_FS_LZ4HC
479 if (strlen(str) == 3) {
480 F2FS_OPTION(sbi).compress_level = 0;
484 #ifdef CONFIG_F2FS_FS_LZ4HC
488 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
491 if (kstrtouint(str + 1, 10, &level))
494 if (level < LZ4HC_MIN_CLEVEL || level > LZ4HC_MAX_CLEVEL) {
495 f2fs_info(sbi, "invalid lz4hc compress level: %d", level);
499 F2FS_OPTION(sbi).compress_level = level;
502 f2fs_info(sbi, "kernel doesn't support lz4hc compression");
508 #ifdef CONFIG_F2FS_FS_ZSTD
509 static int f2fs_set_zstd_level(struct f2fs_sb_info *sbi, const char *str)
514 if (strlen(str) == len) {
515 F2FS_OPTION(sbi).compress_level = 0;
522 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
525 if (kstrtouint(str + 1, 10, &level))
528 if (!level || level > ZSTD_maxCLevel()) {
529 f2fs_info(sbi, "invalid zstd compress level: %d", level);
533 F2FS_OPTION(sbi).compress_level = level;
539 static int parse_options(struct super_block *sb, char *options, bool is_remount)
541 struct f2fs_sb_info *sbi = F2FS_SB(sb);
542 substring_t args[MAX_OPT_ARGS];
543 #ifdef CONFIG_F2FS_FS_COMPRESSION
544 unsigned char (*ext)[F2FS_EXTENSION_LEN];
556 while ((p = strsep(&options, ",")) != NULL) {
561 * Initialize args struct so we know whether arg was
562 * found; some options take optional arguments.
564 args[0].to = args[0].from = NULL;
565 token = match_token(p, f2fs_tokens, args);
568 case Opt_gc_background:
569 name = match_strdup(&args[0]);
573 if (!strcmp(name, "on")) {
574 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
575 } else if (!strcmp(name, "off")) {
576 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_OFF;
577 } else if (!strcmp(name, "sync")) {
578 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_SYNC;
585 case Opt_disable_roll_forward:
586 set_opt(sbi, DISABLE_ROLL_FORWARD);
589 /* this option mounts f2fs with ro */
590 set_opt(sbi, NORECOVERY);
591 if (!f2fs_readonly(sb))
595 set_opt(sbi, DISCARD);
598 if (f2fs_sb_has_blkzoned(sbi)) {
599 f2fs_warn(sbi, "discard is required for zoned block devices");
602 clear_opt(sbi, DISCARD);
605 set_opt(sbi, NOHEAP);
608 clear_opt(sbi, NOHEAP);
610 #ifdef CONFIG_F2FS_FS_XATTR
612 set_opt(sbi, XATTR_USER);
614 case Opt_nouser_xattr:
615 clear_opt(sbi, XATTR_USER);
617 case Opt_inline_xattr:
618 set_opt(sbi, INLINE_XATTR);
620 case Opt_noinline_xattr:
621 clear_opt(sbi, INLINE_XATTR);
623 case Opt_inline_xattr_size:
624 if (args->from && match_int(args, &arg))
626 set_opt(sbi, INLINE_XATTR_SIZE);
627 F2FS_OPTION(sbi).inline_xattr_size = arg;
631 f2fs_info(sbi, "user_xattr options not supported");
633 case Opt_nouser_xattr:
634 f2fs_info(sbi, "nouser_xattr options not supported");
636 case Opt_inline_xattr:
637 f2fs_info(sbi, "inline_xattr options not supported");
639 case Opt_noinline_xattr:
640 f2fs_info(sbi, "noinline_xattr options not supported");
643 #ifdef CONFIG_F2FS_FS_POSIX_ACL
645 set_opt(sbi, POSIX_ACL);
648 clear_opt(sbi, POSIX_ACL);
652 f2fs_info(sbi, "acl options not supported");
655 f2fs_info(sbi, "noacl options not supported");
658 case Opt_active_logs:
659 if (args->from && match_int(args, &arg))
661 if (arg != 2 && arg != 4 &&
662 arg != NR_CURSEG_PERSIST_TYPE)
664 F2FS_OPTION(sbi).active_logs = arg;
666 case Opt_disable_ext_identify:
667 set_opt(sbi, DISABLE_EXT_IDENTIFY);
669 case Opt_inline_data:
670 set_opt(sbi, INLINE_DATA);
672 case Opt_inline_dentry:
673 set_opt(sbi, INLINE_DENTRY);
675 case Opt_noinline_dentry:
676 clear_opt(sbi, INLINE_DENTRY);
678 case Opt_flush_merge:
679 set_opt(sbi, FLUSH_MERGE);
681 case Opt_noflush_merge:
682 clear_opt(sbi, FLUSH_MERGE);
685 set_opt(sbi, NOBARRIER);
688 set_opt(sbi, FASTBOOT);
690 case Opt_extent_cache:
691 set_opt(sbi, EXTENT_CACHE);
693 case Opt_noextent_cache:
694 clear_opt(sbi, EXTENT_CACHE);
696 case Opt_noinline_data:
697 clear_opt(sbi, INLINE_DATA);
700 set_opt(sbi, DATA_FLUSH);
702 case Opt_reserve_root:
703 if (args->from && match_int(args, &arg))
705 if (test_opt(sbi, RESERVE_ROOT)) {
706 f2fs_info(sbi, "Preserve previous reserve_root=%u",
707 F2FS_OPTION(sbi).root_reserved_blocks);
709 F2FS_OPTION(sbi).root_reserved_blocks = arg;
710 set_opt(sbi, RESERVE_ROOT);
714 if (args->from && match_int(args, &arg))
716 uid = make_kuid(current_user_ns(), arg);
717 if (!uid_valid(uid)) {
718 f2fs_err(sbi, "Invalid uid value %d", arg);
721 F2FS_OPTION(sbi).s_resuid = uid;
724 if (args->from && match_int(args, &arg))
726 gid = make_kgid(current_user_ns(), arg);
727 if (!gid_valid(gid)) {
728 f2fs_err(sbi, "Invalid gid value %d", arg);
731 F2FS_OPTION(sbi).s_resgid = gid;
734 name = match_strdup(&args[0]);
738 if (!strcmp(name, "adaptive")) {
739 if (f2fs_sb_has_blkzoned(sbi)) {
740 f2fs_warn(sbi, "adaptive mode is not allowed with zoned block device feature");
744 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
745 } else if (!strcmp(name, "lfs")) {
746 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
753 case Opt_io_size_bits:
754 if (args->from && match_int(args, &arg))
756 if (arg <= 0 || arg > __ilog2_u32(BIO_MAX_VECS)) {
757 f2fs_warn(sbi, "Not support %d, larger than %d",
758 1 << arg, BIO_MAX_VECS);
761 F2FS_OPTION(sbi).write_io_size_bits = arg;
763 #ifdef CONFIG_F2FS_FAULT_INJECTION
764 case Opt_fault_injection:
765 if (args->from && match_int(args, &arg))
767 f2fs_build_fault_attr(sbi, arg, F2FS_ALL_FAULT_TYPE);
768 set_opt(sbi, FAULT_INJECTION);
772 if (args->from && match_int(args, &arg))
774 f2fs_build_fault_attr(sbi, 0, arg);
775 set_opt(sbi, FAULT_INJECTION);
778 case Opt_fault_injection:
779 f2fs_info(sbi, "fault_injection options not supported");
783 f2fs_info(sbi, "fault_type options not supported");
787 sb->s_flags |= SB_LAZYTIME;
790 sb->s_flags &= ~SB_LAZYTIME;
795 set_opt(sbi, USRQUOTA);
798 set_opt(sbi, GRPQUOTA);
801 set_opt(sbi, PRJQUOTA);
804 ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
809 ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
814 ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
818 case Opt_offusrjquota:
819 ret = f2fs_clear_qf_name(sb, USRQUOTA);
823 case Opt_offgrpjquota:
824 ret = f2fs_clear_qf_name(sb, GRPQUOTA);
828 case Opt_offprjjquota:
829 ret = f2fs_clear_qf_name(sb, PRJQUOTA);
833 case Opt_jqfmt_vfsold:
834 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
836 case Opt_jqfmt_vfsv0:
837 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
839 case Opt_jqfmt_vfsv1:
840 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
843 clear_opt(sbi, QUOTA);
844 clear_opt(sbi, USRQUOTA);
845 clear_opt(sbi, GRPQUOTA);
846 clear_opt(sbi, PRJQUOTA);
856 case Opt_offusrjquota:
857 case Opt_offgrpjquota:
858 case Opt_offprjjquota:
859 case Opt_jqfmt_vfsold:
860 case Opt_jqfmt_vfsv0:
861 case Opt_jqfmt_vfsv1:
863 f2fs_info(sbi, "quota operations not supported");
867 name = match_strdup(&args[0]);
870 if (!strcmp(name, "user-based")) {
871 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_USER;
872 } else if (!strcmp(name, "off")) {
873 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
874 } else if (!strcmp(name, "fs-based")) {
875 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_FS;
883 name = match_strdup(&args[0]);
887 if (!strcmp(name, "default")) {
888 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
889 } else if (!strcmp(name, "reuse")) {
890 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
898 name = match_strdup(&args[0]);
901 if (!strcmp(name, "posix")) {
902 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
903 } else if (!strcmp(name, "strict")) {
904 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_STRICT;
905 } else if (!strcmp(name, "nobarrier")) {
906 F2FS_OPTION(sbi).fsync_mode =
907 FSYNC_MODE_NOBARRIER;
914 case Opt_test_dummy_encryption:
915 ret = f2fs_set_test_dummy_encryption(sb, p, &args[0],
920 case Opt_inlinecrypt:
921 #ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
922 sb->s_flags |= SB_INLINECRYPT;
924 f2fs_info(sbi, "inline encryption not supported");
927 case Opt_checkpoint_disable_cap_perc:
928 if (args->from && match_int(args, &arg))
930 if (arg < 0 || arg > 100)
932 F2FS_OPTION(sbi).unusable_cap_perc = arg;
933 set_opt(sbi, DISABLE_CHECKPOINT);
935 case Opt_checkpoint_disable_cap:
936 if (args->from && match_int(args, &arg))
938 F2FS_OPTION(sbi).unusable_cap = arg;
939 set_opt(sbi, DISABLE_CHECKPOINT);
941 case Opt_checkpoint_disable:
942 set_opt(sbi, DISABLE_CHECKPOINT);
944 case Opt_checkpoint_enable:
945 clear_opt(sbi, DISABLE_CHECKPOINT);
947 case Opt_checkpoint_merge:
948 set_opt(sbi, MERGE_CHECKPOINT);
950 case Opt_nocheckpoint_merge:
951 clear_opt(sbi, MERGE_CHECKPOINT);
953 #ifdef CONFIG_F2FS_FS_COMPRESSION
954 case Opt_compress_algorithm:
955 if (!f2fs_sb_has_compression(sbi)) {
956 f2fs_info(sbi, "Image doesn't support compression");
959 name = match_strdup(&args[0]);
962 if (!strcmp(name, "lzo")) {
963 #ifdef CONFIG_F2FS_FS_LZO
964 F2FS_OPTION(sbi).compress_level = 0;
965 F2FS_OPTION(sbi).compress_algorithm =
968 f2fs_info(sbi, "kernel doesn't support lzo compression");
970 } else if (!strncmp(name, "lz4", 3)) {
971 #ifdef CONFIG_F2FS_FS_LZ4
972 ret = f2fs_set_lz4hc_level(sbi, name);
977 F2FS_OPTION(sbi).compress_algorithm =
980 f2fs_info(sbi, "kernel doesn't support lz4 compression");
982 } else if (!strncmp(name, "zstd", 4)) {
983 #ifdef CONFIG_F2FS_FS_ZSTD
984 ret = f2fs_set_zstd_level(sbi, name);
989 F2FS_OPTION(sbi).compress_algorithm =
992 f2fs_info(sbi, "kernel doesn't support zstd compression");
994 } else if (!strcmp(name, "lzo-rle")) {
995 #ifdef CONFIG_F2FS_FS_LZORLE
996 F2FS_OPTION(sbi).compress_level = 0;
997 F2FS_OPTION(sbi).compress_algorithm =
1000 f2fs_info(sbi, "kernel doesn't support lzorle compression");
1008 case Opt_compress_log_size:
1009 if (!f2fs_sb_has_compression(sbi)) {
1010 f2fs_info(sbi, "Image doesn't support compression");
1013 if (args->from && match_int(args, &arg))
1015 if (arg < MIN_COMPRESS_LOG_SIZE ||
1016 arg > MAX_COMPRESS_LOG_SIZE) {
1018 "Compress cluster log size is out of range");
1021 F2FS_OPTION(sbi).compress_log_size = arg;
1023 case Opt_compress_extension:
1024 if (!f2fs_sb_has_compression(sbi)) {
1025 f2fs_info(sbi, "Image doesn't support compression");
1028 name = match_strdup(&args[0]);
1032 ext = F2FS_OPTION(sbi).extensions;
1033 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
1035 if (strlen(name) >= F2FS_EXTENSION_LEN ||
1036 ext_cnt >= COMPRESS_EXT_NUM) {
1038 "invalid extension length/number");
1043 strcpy(ext[ext_cnt], name);
1044 F2FS_OPTION(sbi).compress_ext_cnt++;
1047 case Opt_compress_chksum:
1048 F2FS_OPTION(sbi).compress_chksum = true;
1050 case Opt_compress_mode:
1051 name = match_strdup(&args[0]);
1054 if (!strcmp(name, "fs")) {
1055 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
1056 } else if (!strcmp(name, "user")) {
1057 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_USER;
1065 case Opt_compress_algorithm:
1066 case Opt_compress_log_size:
1067 case Opt_compress_extension:
1068 case Opt_compress_chksum:
1069 case Opt_compress_mode:
1070 f2fs_info(sbi, "compression options not supported");
1077 f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
1083 if (f2fs_check_quota_options(sbi))
1086 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sbi->sb)) {
1087 f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1090 if (f2fs_sb_has_project_quota(sbi) && !f2fs_readonly(sbi->sb)) {
1091 f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1095 #ifndef CONFIG_UNICODE
1096 if (f2fs_sb_has_casefold(sbi)) {
1098 "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
1103 * The BLKZONED feature indicates that the drive was formatted with
1104 * zone alignment optimization. This is optional for host-aware
1105 * devices, but mandatory for host-managed zoned block devices.
1107 #ifndef CONFIG_BLK_DEV_ZONED
1108 if (f2fs_sb_has_blkzoned(sbi)) {
1109 f2fs_err(sbi, "Zoned block device support is not enabled");
1114 if (F2FS_IO_SIZE_BITS(sbi) && !f2fs_lfs_mode(sbi)) {
1115 f2fs_err(sbi, "Should set mode=lfs with %uKB-sized IO",
1116 F2FS_IO_SIZE_KB(sbi));
1120 if (test_opt(sbi, INLINE_XATTR_SIZE)) {
1121 int min_size, max_size;
1123 if (!f2fs_sb_has_extra_attr(sbi) ||
1124 !f2fs_sb_has_flexible_inline_xattr(sbi)) {
1125 f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
1128 if (!test_opt(sbi, INLINE_XATTR)) {
1129 f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
1133 min_size = sizeof(struct f2fs_xattr_header) / sizeof(__le32);
1134 max_size = MAX_INLINE_XATTR_SIZE;
1136 if (F2FS_OPTION(sbi).inline_xattr_size < min_size ||
1137 F2FS_OPTION(sbi).inline_xattr_size > max_size) {
1138 f2fs_err(sbi, "inline xattr size is out of range: %d ~ %d",
1139 min_size, max_size);
1144 if (test_opt(sbi, DISABLE_CHECKPOINT) && f2fs_lfs_mode(sbi)) {
1145 f2fs_err(sbi, "LFS not compatible with checkpoint=disable\n");
1149 /* Not pass down write hints if the number of active logs is lesser
1150 * than NR_CURSEG_PERSIST_TYPE.
1152 if (F2FS_OPTION(sbi).active_logs != NR_CURSEG_TYPE)
1153 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1157 static struct inode *f2fs_alloc_inode(struct super_block *sb)
1159 struct f2fs_inode_info *fi;
1161 fi = kmem_cache_alloc(f2fs_inode_cachep, GFP_F2FS_ZERO);
1165 init_once((void *) fi);
1167 /* Initialize f2fs-specific inode info */
1168 atomic_set(&fi->dirty_pages, 0);
1169 atomic_set(&fi->i_compr_blocks, 0);
1170 init_rwsem(&fi->i_sem);
1171 spin_lock_init(&fi->i_size_lock);
1172 INIT_LIST_HEAD(&fi->dirty_list);
1173 INIT_LIST_HEAD(&fi->gdirty_list);
1174 INIT_LIST_HEAD(&fi->inmem_ilist);
1175 INIT_LIST_HEAD(&fi->inmem_pages);
1176 mutex_init(&fi->inmem_lock);
1177 init_rwsem(&fi->i_gc_rwsem[READ]);
1178 init_rwsem(&fi->i_gc_rwsem[WRITE]);
1179 init_rwsem(&fi->i_mmap_sem);
1180 init_rwsem(&fi->i_xattr_sem);
1182 /* Will be used by directory only */
1183 fi->i_dir_level = F2FS_SB(sb)->dir_level;
1185 return &fi->vfs_inode;
1188 static int f2fs_drop_inode(struct inode *inode)
1190 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1194 * during filesystem shutdown, if checkpoint is disabled,
1195 * drop useless meta/node dirty pages.
1197 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1198 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1199 inode->i_ino == F2FS_META_INO(sbi)) {
1200 trace_f2fs_drop_inode(inode, 1);
1206 * This is to avoid a deadlock condition like below.
1207 * writeback_single_inode(inode)
1208 * - f2fs_write_data_page
1209 * - f2fs_gc -> iput -> evict
1210 * - inode_wait_for_writeback(inode)
1212 if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
1213 if (!inode->i_nlink && !is_bad_inode(inode)) {
1214 /* to avoid evict_inode call simultaneously */
1215 atomic_inc(&inode->i_count);
1216 spin_unlock(&inode->i_lock);
1218 /* some remained atomic pages should discarded */
1219 if (f2fs_is_atomic_file(inode))
1220 f2fs_drop_inmem_pages(inode);
1222 /* should remain fi->extent_tree for writepage */
1223 f2fs_destroy_extent_node(inode);
1225 sb_start_intwrite(inode->i_sb);
1226 f2fs_i_size_write(inode, 0);
1228 f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1229 inode, NULL, 0, DATA);
1230 truncate_inode_pages_final(inode->i_mapping);
1232 if (F2FS_HAS_BLOCKS(inode))
1233 f2fs_truncate(inode);
1235 sb_end_intwrite(inode->i_sb);
1237 spin_lock(&inode->i_lock);
1238 atomic_dec(&inode->i_count);
1240 trace_f2fs_drop_inode(inode, 0);
1243 ret = generic_drop_inode(inode);
1245 ret = fscrypt_drop_inode(inode);
1246 trace_f2fs_drop_inode(inode, ret);
1250 int f2fs_inode_dirtied(struct inode *inode, bool sync)
1252 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1255 spin_lock(&sbi->inode_lock[DIRTY_META]);
1256 if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1259 set_inode_flag(inode, FI_DIRTY_INODE);
1260 stat_inc_dirty_inode(sbi, DIRTY_META);
1262 if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1263 list_add_tail(&F2FS_I(inode)->gdirty_list,
1264 &sbi->inode_list[DIRTY_META]);
1265 inc_page_count(sbi, F2FS_DIRTY_IMETA);
1267 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1271 void f2fs_inode_synced(struct inode *inode)
1273 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1275 spin_lock(&sbi->inode_lock[DIRTY_META]);
1276 if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1277 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1280 if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1281 list_del_init(&F2FS_I(inode)->gdirty_list);
1282 dec_page_count(sbi, F2FS_DIRTY_IMETA);
1284 clear_inode_flag(inode, FI_DIRTY_INODE);
1285 clear_inode_flag(inode, FI_AUTO_RECOVER);
1286 stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1287 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1291 * f2fs_dirty_inode() is called from __mark_inode_dirty()
1293 * We should call set_dirty_inode to write the dirty inode through write_inode.
1295 static void f2fs_dirty_inode(struct inode *inode, int flags)
1297 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1299 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1300 inode->i_ino == F2FS_META_INO(sbi))
1303 if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1304 clear_inode_flag(inode, FI_AUTO_RECOVER);
1306 f2fs_inode_dirtied(inode, false);
1309 static void f2fs_free_inode(struct inode *inode)
1311 fscrypt_free_inode(inode);
1312 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1315 static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1317 percpu_counter_destroy(&sbi->alloc_valid_block_count);
1318 percpu_counter_destroy(&sbi->total_valid_inode_count);
1321 static void destroy_device_list(struct f2fs_sb_info *sbi)
1325 for (i = 0; i < sbi->s_ndevs; i++) {
1326 blkdev_put(FDEV(i).bdev, FMODE_EXCL);
1327 #ifdef CONFIG_BLK_DEV_ZONED
1328 kvfree(FDEV(i).blkz_seq);
1329 kfree(FDEV(i).zone_capacity_blocks);
1335 static void f2fs_put_super(struct super_block *sb)
1337 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1341 /* unregister procfs/sysfs entries in advance to avoid race case */
1342 f2fs_unregister_sysfs(sbi);
1344 f2fs_quota_off_umount(sb);
1346 /* prevent remaining shrinker jobs */
1347 mutex_lock(&sbi->umount_mutex);
1350 * flush all issued checkpoints and stop checkpoint issue thread.
1351 * after then, all checkpoints should be done by each process context.
1353 f2fs_stop_ckpt_thread(sbi);
1356 * We don't need to do checkpoint when superblock is clean.
1357 * But, the previous checkpoint was not done by umount, it needs to do
1358 * clean checkpoint again.
1360 if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1361 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1362 struct cp_control cpc = {
1363 .reason = CP_UMOUNT,
1365 f2fs_write_checkpoint(sbi, &cpc);
1368 /* be sure to wait for any on-going discard commands */
1369 dropped = f2fs_issue_discard_timeout(sbi);
1371 if ((f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi)) &&
1372 !sbi->discard_blks && !dropped) {
1373 struct cp_control cpc = {
1374 .reason = CP_UMOUNT | CP_TRIMMED,
1376 f2fs_write_checkpoint(sbi, &cpc);
1380 * normally superblock is clean, so we need to release this.
1381 * In addition, EIO will skip do checkpoint, we need this as well.
1383 f2fs_release_ino_entry(sbi, true);
1385 f2fs_leave_shrinker(sbi);
1386 mutex_unlock(&sbi->umount_mutex);
1388 /* our cp_error case, we can wait for any writeback page */
1389 f2fs_flush_merged_writes(sbi);
1391 f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1393 f2fs_bug_on(sbi, sbi->fsync_node_num);
1395 iput(sbi->node_inode);
1396 sbi->node_inode = NULL;
1398 iput(sbi->meta_inode);
1399 sbi->meta_inode = NULL;
1402 * iput() can update stat information, if f2fs_write_checkpoint()
1403 * above failed with error.
1405 f2fs_destroy_stats(sbi);
1407 /* destroy f2fs internal modules */
1408 f2fs_destroy_node_manager(sbi);
1409 f2fs_destroy_segment_manager(sbi);
1411 f2fs_destroy_post_read_wq(sbi);
1415 sb->s_fs_info = NULL;
1416 if (sbi->s_chksum_driver)
1417 crypto_free_shash(sbi->s_chksum_driver);
1418 kfree(sbi->raw_super);
1420 destroy_device_list(sbi);
1421 f2fs_destroy_page_array_cache(sbi);
1422 f2fs_destroy_xattr_caches(sbi);
1423 mempool_destroy(sbi->write_io_dummy);
1425 for (i = 0; i < MAXQUOTAS; i++)
1426 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1428 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
1429 destroy_percpu_info(sbi);
1430 for (i = 0; i < NR_PAGE_TYPE; i++)
1431 kvfree(sbi->write_io[i]);
1432 #ifdef CONFIG_UNICODE
1433 utf8_unload(sb->s_encoding);
1438 int f2fs_sync_fs(struct super_block *sb, int sync)
1440 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1443 if (unlikely(f2fs_cp_error(sbi)))
1445 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1448 trace_f2fs_sync_fs(sb, sync);
1450 if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1454 err = f2fs_issue_checkpoint(sbi);
1459 static int f2fs_freeze(struct super_block *sb)
1461 if (f2fs_readonly(sb))
1464 /* IO error happened before */
1465 if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1468 /* must be clean, since sync_filesystem() was already called */
1469 if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1472 /* ensure no checkpoint required */
1473 if (!llist_empty(&F2FS_SB(sb)->cprc_info.issue_list))
1478 static int f2fs_unfreeze(struct super_block *sb)
1484 static int f2fs_statfs_project(struct super_block *sb,
1485 kprojid_t projid, struct kstatfs *buf)
1488 struct dquot *dquot;
1492 qid = make_kqid_projid(projid);
1493 dquot = dqget(sb, qid);
1495 return PTR_ERR(dquot);
1496 spin_lock(&dquot->dq_dqb_lock);
1498 limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1499 dquot->dq_dqb.dqb_bhardlimit);
1501 limit >>= sb->s_blocksize_bits;
1503 if (limit && buf->f_blocks > limit) {
1504 curblock = (dquot->dq_dqb.dqb_curspace +
1505 dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
1506 buf->f_blocks = limit;
1507 buf->f_bfree = buf->f_bavail =
1508 (buf->f_blocks > curblock) ?
1509 (buf->f_blocks - curblock) : 0;
1512 limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1513 dquot->dq_dqb.dqb_ihardlimit);
1515 if (limit && buf->f_files > limit) {
1516 buf->f_files = limit;
1518 (buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1519 (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1522 spin_unlock(&dquot->dq_dqb_lock);
1528 static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1530 struct super_block *sb = dentry->d_sb;
1531 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1532 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
1533 block_t total_count, user_block_count, start_count;
1534 u64 avail_node_count;
1536 total_count = le64_to_cpu(sbi->raw_super->block_count);
1537 user_block_count = sbi->user_block_count;
1538 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
1539 buf->f_type = F2FS_SUPER_MAGIC;
1540 buf->f_bsize = sbi->blocksize;
1542 buf->f_blocks = total_count - start_count;
1543 buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1544 sbi->current_reserved_blocks;
1546 spin_lock(&sbi->stat_lock);
1547 if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1550 buf->f_bfree -= sbi->unusable_block_count;
1551 spin_unlock(&sbi->stat_lock);
1553 if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1554 buf->f_bavail = buf->f_bfree -
1555 F2FS_OPTION(sbi).root_reserved_blocks;
1559 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
1561 if (avail_node_count > user_block_count) {
1562 buf->f_files = user_block_count;
1563 buf->f_ffree = buf->f_bavail;
1565 buf->f_files = avail_node_count;
1566 buf->f_ffree = min(avail_node_count - valid_node_count(sbi),
1570 buf->f_namelen = F2FS_NAME_LEN;
1571 buf->f_fsid = u64_to_fsid(id);
1574 if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1575 sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1576 f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1582 static inline void f2fs_show_quota_options(struct seq_file *seq,
1583 struct super_block *sb)
1586 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1588 if (F2FS_OPTION(sbi).s_jquota_fmt) {
1591 switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1602 seq_printf(seq, ",jqfmt=%s", fmtname);
1605 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1606 seq_show_option(seq, "usrjquota",
1607 F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1609 if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1610 seq_show_option(seq, "grpjquota",
1611 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1613 if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1614 seq_show_option(seq, "prjjquota",
1615 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1619 static inline void f2fs_show_compress_options(struct seq_file *seq,
1620 struct super_block *sb)
1622 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1626 if (!f2fs_sb_has_compression(sbi))
1629 switch (F2FS_OPTION(sbi).compress_algorithm) {
1639 case COMPRESS_LZORLE:
1640 algtype = "lzo-rle";
1643 seq_printf(seq, ",compress_algorithm=%s", algtype);
1645 if (F2FS_OPTION(sbi).compress_level)
1646 seq_printf(seq, ":%d", F2FS_OPTION(sbi).compress_level);
1648 seq_printf(seq, ",compress_log_size=%u",
1649 F2FS_OPTION(sbi).compress_log_size);
1651 for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1652 seq_printf(seq, ",compress_extension=%s",
1653 F2FS_OPTION(sbi).extensions[i]);
1656 if (F2FS_OPTION(sbi).compress_chksum)
1657 seq_puts(seq, ",compress_chksum");
1659 if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_FS)
1660 seq_printf(seq, ",compress_mode=%s", "fs");
1661 else if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_USER)
1662 seq_printf(seq, ",compress_mode=%s", "user");
1665 static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1667 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1669 if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1670 seq_printf(seq, ",background_gc=%s", "sync");
1671 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1672 seq_printf(seq, ",background_gc=%s", "on");
1673 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1674 seq_printf(seq, ",background_gc=%s", "off");
1676 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1677 seq_puts(seq, ",disable_roll_forward");
1678 if (test_opt(sbi, NORECOVERY))
1679 seq_puts(seq, ",norecovery");
1680 if (test_opt(sbi, DISCARD))
1681 seq_puts(seq, ",discard");
1683 seq_puts(seq, ",nodiscard");
1684 if (test_opt(sbi, NOHEAP))
1685 seq_puts(seq, ",no_heap");
1687 seq_puts(seq, ",heap");
1688 #ifdef CONFIG_F2FS_FS_XATTR
1689 if (test_opt(sbi, XATTR_USER))
1690 seq_puts(seq, ",user_xattr");
1692 seq_puts(seq, ",nouser_xattr");
1693 if (test_opt(sbi, INLINE_XATTR))
1694 seq_puts(seq, ",inline_xattr");
1696 seq_puts(seq, ",noinline_xattr");
1697 if (test_opt(sbi, INLINE_XATTR_SIZE))
1698 seq_printf(seq, ",inline_xattr_size=%u",
1699 F2FS_OPTION(sbi).inline_xattr_size);
1701 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1702 if (test_opt(sbi, POSIX_ACL))
1703 seq_puts(seq, ",acl");
1705 seq_puts(seq, ",noacl");
1707 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
1708 seq_puts(seq, ",disable_ext_identify");
1709 if (test_opt(sbi, INLINE_DATA))
1710 seq_puts(seq, ",inline_data");
1712 seq_puts(seq, ",noinline_data");
1713 if (test_opt(sbi, INLINE_DENTRY))
1714 seq_puts(seq, ",inline_dentry");
1716 seq_puts(seq, ",noinline_dentry");
1717 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
1718 seq_puts(seq, ",flush_merge");
1719 if (test_opt(sbi, NOBARRIER))
1720 seq_puts(seq, ",nobarrier");
1721 if (test_opt(sbi, FASTBOOT))
1722 seq_puts(seq, ",fastboot");
1723 if (test_opt(sbi, EXTENT_CACHE))
1724 seq_puts(seq, ",extent_cache");
1726 seq_puts(seq, ",noextent_cache");
1727 if (test_opt(sbi, DATA_FLUSH))
1728 seq_puts(seq, ",data_flush");
1730 seq_puts(seq, ",mode=");
1731 if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
1732 seq_puts(seq, "adaptive");
1733 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
1734 seq_puts(seq, "lfs");
1735 seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
1736 if (test_opt(sbi, RESERVE_ROOT))
1737 seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
1738 F2FS_OPTION(sbi).root_reserved_blocks,
1739 from_kuid_munged(&init_user_ns,
1740 F2FS_OPTION(sbi).s_resuid),
1741 from_kgid_munged(&init_user_ns,
1742 F2FS_OPTION(sbi).s_resgid));
1743 if (F2FS_IO_SIZE_BITS(sbi))
1744 seq_printf(seq, ",io_bits=%u",
1745 F2FS_OPTION(sbi).write_io_size_bits);
1746 #ifdef CONFIG_F2FS_FAULT_INJECTION
1747 if (test_opt(sbi, FAULT_INJECTION)) {
1748 seq_printf(seq, ",fault_injection=%u",
1749 F2FS_OPTION(sbi).fault_info.inject_rate);
1750 seq_printf(seq, ",fault_type=%u",
1751 F2FS_OPTION(sbi).fault_info.inject_type);
1755 if (test_opt(sbi, QUOTA))
1756 seq_puts(seq, ",quota");
1757 if (test_opt(sbi, USRQUOTA))
1758 seq_puts(seq, ",usrquota");
1759 if (test_opt(sbi, GRPQUOTA))
1760 seq_puts(seq, ",grpquota");
1761 if (test_opt(sbi, PRJQUOTA))
1762 seq_puts(seq, ",prjquota");
1764 f2fs_show_quota_options(seq, sbi->sb);
1765 if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_USER)
1766 seq_printf(seq, ",whint_mode=%s", "user-based");
1767 else if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_FS)
1768 seq_printf(seq, ",whint_mode=%s", "fs-based");
1770 fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
1772 if (sbi->sb->s_flags & SB_INLINECRYPT)
1773 seq_puts(seq, ",inlinecrypt");
1775 if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
1776 seq_printf(seq, ",alloc_mode=%s", "default");
1777 else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
1778 seq_printf(seq, ",alloc_mode=%s", "reuse");
1780 if (test_opt(sbi, DISABLE_CHECKPOINT))
1781 seq_printf(seq, ",checkpoint=disable:%u",
1782 F2FS_OPTION(sbi).unusable_cap);
1783 if (test_opt(sbi, MERGE_CHECKPOINT))
1784 seq_puts(seq, ",checkpoint_merge");
1786 seq_puts(seq, ",nocheckpoint_merge");
1787 if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
1788 seq_printf(seq, ",fsync_mode=%s", "posix");
1789 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
1790 seq_printf(seq, ",fsync_mode=%s", "strict");
1791 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
1792 seq_printf(seq, ",fsync_mode=%s", "nobarrier");
1794 #ifdef CONFIG_F2FS_FS_COMPRESSION
1795 f2fs_show_compress_options(seq, sbi->sb);
1798 if (test_opt(sbi, ATGC))
1799 seq_puts(seq, ",atgc");
1803 static void default_options(struct f2fs_sb_info *sbi)
1805 /* init some FS parameters */
1806 F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
1807 F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
1808 F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1809 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
1810 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
1811 F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
1812 F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
1813 F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
1814 F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
1815 F2FS_OPTION(sbi).compress_ext_cnt = 0;
1816 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
1817 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
1819 sbi->sb->s_flags &= ~SB_INLINECRYPT;
1821 set_opt(sbi, INLINE_XATTR);
1822 set_opt(sbi, INLINE_DATA);
1823 set_opt(sbi, INLINE_DENTRY);
1824 set_opt(sbi, EXTENT_CACHE);
1825 set_opt(sbi, NOHEAP);
1826 clear_opt(sbi, DISABLE_CHECKPOINT);
1827 F2FS_OPTION(sbi).unusable_cap = 0;
1828 sbi->sb->s_flags |= SB_LAZYTIME;
1829 set_opt(sbi, FLUSH_MERGE);
1830 set_opt(sbi, DISCARD);
1831 if (f2fs_sb_has_blkzoned(sbi))
1832 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
1834 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
1836 #ifdef CONFIG_F2FS_FS_XATTR
1837 set_opt(sbi, XATTR_USER);
1839 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1840 set_opt(sbi, POSIX_ACL);
1843 f2fs_build_fault_attr(sbi, 0, 0);
1847 static int f2fs_enable_quotas(struct super_block *sb);
1850 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
1852 unsigned int s_flags = sbi->sb->s_flags;
1853 struct cp_control cpc;
1858 if (s_flags & SB_RDONLY) {
1859 f2fs_err(sbi, "checkpoint=disable on readonly fs");
1862 sbi->sb->s_flags |= SB_ACTIVE;
1864 f2fs_update_time(sbi, DISABLE_TIME);
1866 while (!f2fs_time_over(sbi, DISABLE_TIME)) {
1867 down_write(&sbi->gc_lock);
1868 err = f2fs_gc(sbi, true, false, NULL_SEGNO);
1869 if (err == -ENODATA) {
1873 if (err && err != -EAGAIN)
1877 ret = sync_filesystem(sbi->sb);
1879 err = ret ? ret: err;
1883 unusable = f2fs_get_unusable_blocks(sbi);
1884 if (f2fs_disable_cp_again(sbi, unusable)) {
1889 down_write(&sbi->gc_lock);
1890 cpc.reason = CP_PAUSE;
1891 set_sbi_flag(sbi, SBI_CP_DISABLED);
1892 err = f2fs_write_checkpoint(sbi, &cpc);
1896 spin_lock(&sbi->stat_lock);
1897 sbi->unusable_block_count = unusable;
1898 spin_unlock(&sbi->stat_lock);
1901 up_write(&sbi->gc_lock);
1903 sbi->sb->s_flags = s_flags; /* Restore SB_RDONLY status */
1907 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
1909 /* we should flush all the data to keep data consistency */
1910 sync_inodes_sb(sbi->sb);
1912 down_write(&sbi->gc_lock);
1913 f2fs_dirty_to_prefree(sbi);
1915 clear_sbi_flag(sbi, SBI_CP_DISABLED);
1916 set_sbi_flag(sbi, SBI_IS_DIRTY);
1917 up_write(&sbi->gc_lock);
1919 f2fs_sync_fs(sbi->sb, 1);
1922 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
1924 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1925 struct f2fs_mount_info org_mount_opt;
1926 unsigned long old_sb_flags;
1928 bool need_restart_gc = false;
1929 bool need_stop_gc = false;
1930 bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
1931 bool disable_checkpoint = test_opt(sbi, DISABLE_CHECKPOINT);
1932 bool no_io_align = !F2FS_IO_ALIGNED(sbi);
1933 bool no_atgc = !test_opt(sbi, ATGC);
1934 bool checkpoint_changed;
1940 * Save the old mount options in case we
1941 * need to restore them.
1943 org_mount_opt = sbi->mount_opt;
1944 old_sb_flags = sb->s_flags;
1947 org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
1948 for (i = 0; i < MAXQUOTAS; i++) {
1949 if (F2FS_OPTION(sbi).s_qf_names[i]) {
1950 org_mount_opt.s_qf_names[i] =
1951 kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
1953 if (!org_mount_opt.s_qf_names[i]) {
1954 for (j = 0; j < i; j++)
1955 kfree(org_mount_opt.s_qf_names[j]);
1959 org_mount_opt.s_qf_names[i] = NULL;
1964 /* recover superblocks we couldn't write due to previous RO mount */
1965 if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
1966 err = f2fs_commit_super(sbi, false);
1967 f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
1970 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
1973 default_options(sbi);
1975 /* parse mount options */
1976 err = parse_options(sb, data, true);
1979 checkpoint_changed =
1980 disable_checkpoint != test_opt(sbi, DISABLE_CHECKPOINT);
1983 * Previous and new state of filesystem is RO,
1984 * so skip checking GC and FLUSH_MERGE conditions.
1986 if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
1990 if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
1991 err = dquot_suspend(sb, -1);
1994 } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
1995 /* dquot_resume needs RW */
1996 sb->s_flags &= ~SB_RDONLY;
1997 if (sb_any_quota_suspended(sb)) {
1998 dquot_resume(sb, -1);
1999 } else if (f2fs_sb_has_quota_ino(sbi)) {
2000 err = f2fs_enable_quotas(sb);
2006 /* disallow enable atgc dynamically */
2007 if (no_atgc == !!test_opt(sbi, ATGC)) {
2009 f2fs_warn(sbi, "switch atgc option is not allowed");
2013 /* disallow enable/disable extent_cache dynamically */
2014 if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
2016 f2fs_warn(sbi, "switch extent_cache option is not allowed");
2020 if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
2022 f2fs_warn(sbi, "switch io_bits option is not allowed");
2026 if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
2028 f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
2033 * We stop the GC thread if FS is mounted as RO
2034 * or if background_gc = off is passed in mount
2035 * option. Also sync the filesystem.
2037 if ((*flags & SB_RDONLY) ||
2038 F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF) {
2039 if (sbi->gc_thread) {
2040 f2fs_stop_gc_thread(sbi);
2041 need_restart_gc = true;
2043 } else if (!sbi->gc_thread) {
2044 err = f2fs_start_gc_thread(sbi);
2047 need_stop_gc = true;
2050 if (*flags & SB_RDONLY ||
2051 F2FS_OPTION(sbi).whint_mode != org_mount_opt.whint_mode) {
2054 set_sbi_flag(sbi, SBI_IS_DIRTY);
2055 set_sbi_flag(sbi, SBI_IS_CLOSE);
2056 f2fs_sync_fs(sb, 1);
2057 clear_sbi_flag(sbi, SBI_IS_CLOSE);
2060 if (checkpoint_changed) {
2061 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
2062 err = f2fs_disable_checkpoint(sbi);
2066 f2fs_enable_checkpoint(sbi);
2070 if (!test_opt(sbi, DISABLE_CHECKPOINT) &&
2071 test_opt(sbi, MERGE_CHECKPOINT)) {
2072 err = f2fs_start_ckpt_thread(sbi);
2075 "Failed to start F2FS issue_checkpoint_thread (%d)",
2080 f2fs_stop_ckpt_thread(sbi);
2084 * We stop issue flush thread if FS is mounted as RO
2085 * or if flush_merge is not passed in mount option.
2087 if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2088 clear_opt(sbi, FLUSH_MERGE);
2089 f2fs_destroy_flush_cmd_control(sbi, false);
2091 err = f2fs_create_flush_cmd_control(sbi);
2097 /* Release old quota file names */
2098 for (i = 0; i < MAXQUOTAS; i++)
2099 kfree(org_mount_opt.s_qf_names[i]);
2101 /* Update the POSIXACL Flag */
2102 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
2103 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
2105 limit_reserve_root(sbi);
2106 adjust_unusable_cap_perc(sbi);
2107 *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
2110 if (need_restart_gc) {
2111 if (f2fs_start_gc_thread(sbi))
2112 f2fs_warn(sbi, "background gc thread has stopped");
2113 } else if (need_stop_gc) {
2114 f2fs_stop_gc_thread(sbi);
2118 F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
2119 for (i = 0; i < MAXQUOTAS; i++) {
2120 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
2121 F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
2124 sbi->mount_opt = org_mount_opt;
2125 sb->s_flags = old_sb_flags;
2130 /* Read data from quotafile */
2131 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
2132 size_t len, loff_t off)
2134 struct inode *inode = sb_dqopt(sb)->files[type];
2135 struct address_space *mapping = inode->i_mapping;
2136 block_t blkidx = F2FS_BYTES_TO_BLK(off);
2137 int offset = off & (sb->s_blocksize - 1);
2140 loff_t i_size = i_size_read(inode);
2147 if (off + len > i_size)
2150 while (toread > 0) {
2151 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
2153 page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
2155 if (PTR_ERR(page) == -ENOMEM) {
2156 congestion_wait(BLK_RW_ASYNC,
2157 DEFAULT_IO_TIMEOUT);
2160 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2161 return PTR_ERR(page);
2166 if (unlikely(page->mapping != mapping)) {
2167 f2fs_put_page(page, 1);
2170 if (unlikely(!PageUptodate(page))) {
2171 f2fs_put_page(page, 1);
2172 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2176 kaddr = kmap_atomic(page);
2177 memcpy(data, kaddr + offset, tocopy);
2178 kunmap_atomic(kaddr);
2179 f2fs_put_page(page, 1);
2189 /* Write to quotafile */
2190 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2191 const char *data, size_t len, loff_t off)
2193 struct inode *inode = sb_dqopt(sb)->files[type];
2194 struct address_space *mapping = inode->i_mapping;
2195 const struct address_space_operations *a_ops = mapping->a_ops;
2196 int offset = off & (sb->s_blocksize - 1);
2197 size_t towrite = len;
2199 void *fsdata = NULL;
2204 while (towrite > 0) {
2205 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2208 err = a_ops->write_begin(NULL, mapping, off, tocopy, 0,
2210 if (unlikely(err)) {
2211 if (err == -ENOMEM) {
2212 congestion_wait(BLK_RW_ASYNC,
2213 DEFAULT_IO_TIMEOUT);
2216 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2220 kaddr = kmap_atomic(page);
2221 memcpy(kaddr + offset, data, tocopy);
2222 kunmap_atomic(kaddr);
2223 flush_dcache_page(page);
2225 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2236 inode->i_mtime = inode->i_ctime = current_time(inode);
2237 f2fs_mark_inode_dirty_sync(inode, false);
2238 return len - towrite;
2241 static struct dquot **f2fs_get_dquots(struct inode *inode)
2243 return F2FS_I(inode)->i_dquot;
2246 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2248 return &F2FS_I(inode)->i_reserved_quota;
2251 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2253 if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2254 f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2258 return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2259 F2FS_OPTION(sbi).s_jquota_fmt, type);
2262 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2267 if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2268 err = f2fs_enable_quotas(sbi->sb);
2270 f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2276 for (i = 0; i < MAXQUOTAS; i++) {
2277 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2278 err = f2fs_quota_on_mount(sbi, i);
2283 f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2290 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2293 struct inode *qf_inode;
2294 unsigned long qf_inum;
2297 BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2299 qf_inum = f2fs_qf_ino(sb, type);
2303 qf_inode = f2fs_iget(sb, qf_inum);
2304 if (IS_ERR(qf_inode)) {
2305 f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2306 return PTR_ERR(qf_inode);
2309 /* Don't account quota for quota files to avoid recursion */
2310 qf_inode->i_flags |= S_NOQUOTA;
2311 err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2316 static int f2fs_enable_quotas(struct super_block *sb)
2318 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2320 unsigned long qf_inum;
2321 bool quota_mopt[MAXQUOTAS] = {
2322 test_opt(sbi, USRQUOTA),
2323 test_opt(sbi, GRPQUOTA),
2324 test_opt(sbi, PRJQUOTA),
2327 if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2328 f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2332 sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2334 for (type = 0; type < MAXQUOTAS; type++) {
2335 qf_inum = f2fs_qf_ino(sb, type);
2337 err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2338 DQUOT_USAGE_ENABLED |
2339 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2341 f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2343 for (type--; type >= 0; type--)
2344 dquot_quota_off(sb, type);
2345 set_sbi_flag(F2FS_SB(sb),
2346 SBI_QUOTA_NEED_REPAIR);
2354 int f2fs_quota_sync(struct super_block *sb, int type)
2356 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2357 struct quota_info *dqopt = sb_dqopt(sb);
2364 * down_read(quota_sem)
2365 * dquot_writeback_dquots()
2368 * down_read(quota_sem)
2372 down_read(&sbi->quota_sem);
2373 ret = dquot_writeback_dquots(sb, type);
2378 * Now when everything is written we can discard the pagecache so
2379 * that userspace sees the changes.
2381 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2382 struct address_space *mapping;
2384 if (type != -1 && cnt != type)
2386 if (!sb_has_quota_active(sb, cnt))
2389 mapping = dqopt->files[cnt]->i_mapping;
2391 ret = filemap_fdatawrite(mapping);
2395 /* if we are using journalled quota */
2396 if (is_journalled_quota(sbi))
2399 ret = filemap_fdatawait(mapping);
2401 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2403 inode_lock(dqopt->files[cnt]);
2404 truncate_inode_pages(&dqopt->files[cnt]->i_data, 0);
2405 inode_unlock(dqopt->files[cnt]);
2409 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2410 up_read(&sbi->quota_sem);
2411 f2fs_unlock_op(sbi);
2415 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2416 const struct path *path)
2418 struct inode *inode;
2421 /* if quota sysfile exists, deny enabling quota with specific file */
2422 if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2423 f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2427 err = f2fs_quota_sync(sb, type);
2431 err = dquot_quota_on(sb, type, format_id, path);
2435 inode = d_inode(path->dentry);
2438 F2FS_I(inode)->i_flags |= F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL;
2439 f2fs_set_inode_flags(inode);
2440 inode_unlock(inode);
2441 f2fs_mark_inode_dirty_sync(inode, false);
2446 static int __f2fs_quota_off(struct super_block *sb, int type)
2448 struct inode *inode = sb_dqopt(sb)->files[type];
2451 if (!inode || !igrab(inode))
2452 return dquot_quota_off(sb, type);
2454 err = f2fs_quota_sync(sb, type);
2458 err = dquot_quota_off(sb, type);
2459 if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
2463 F2FS_I(inode)->i_flags &= ~(F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL);
2464 f2fs_set_inode_flags(inode);
2465 inode_unlock(inode);
2466 f2fs_mark_inode_dirty_sync(inode, false);
2472 static int f2fs_quota_off(struct super_block *sb, int type)
2474 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2477 err = __f2fs_quota_off(sb, type);
2480 * quotactl can shutdown journalled quota, result in inconsistence
2481 * between quota record and fs data by following updates, tag the
2482 * flag to let fsck be aware of it.
2484 if (is_journalled_quota(sbi))
2485 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2489 void f2fs_quota_off_umount(struct super_block *sb)
2494 for (type = 0; type < MAXQUOTAS; type++) {
2495 err = __f2fs_quota_off(sb, type);
2497 int ret = dquot_quota_off(sb, type);
2499 f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
2501 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2505 * In case of checkpoint=disable, we must flush quota blocks.
2506 * This can cause NULL exception for node_inode in end_io, since
2507 * put_super already dropped it.
2509 sync_filesystem(sb);
2512 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
2514 struct quota_info *dqopt = sb_dqopt(sb);
2517 for (type = 0; type < MAXQUOTAS; type++) {
2518 if (!dqopt->files[type])
2520 f2fs_inode_synced(dqopt->files[type]);
2524 static int f2fs_dquot_commit(struct dquot *dquot)
2526 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2529 down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
2530 ret = dquot_commit(dquot);
2532 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2533 up_read(&sbi->quota_sem);
2537 static int f2fs_dquot_acquire(struct dquot *dquot)
2539 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2542 down_read(&sbi->quota_sem);
2543 ret = dquot_acquire(dquot);
2545 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2546 up_read(&sbi->quota_sem);
2550 static int f2fs_dquot_release(struct dquot *dquot)
2552 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2553 int ret = dquot_release(dquot);
2556 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2560 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
2562 struct super_block *sb = dquot->dq_sb;
2563 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2564 int ret = dquot_mark_dquot_dirty(dquot);
2566 /* if we are using journalled quota */
2567 if (is_journalled_quota(sbi))
2568 set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
2573 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
2575 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2576 int ret = dquot_commit_info(sb, type);
2579 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2583 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
2585 *projid = F2FS_I(inode)->i_projid;
2589 static const struct dquot_operations f2fs_quota_operations = {
2590 .get_reserved_space = f2fs_get_reserved_space,
2591 .write_dquot = f2fs_dquot_commit,
2592 .acquire_dquot = f2fs_dquot_acquire,
2593 .release_dquot = f2fs_dquot_release,
2594 .mark_dirty = f2fs_dquot_mark_dquot_dirty,
2595 .write_info = f2fs_dquot_commit_info,
2596 .alloc_dquot = dquot_alloc,
2597 .destroy_dquot = dquot_destroy,
2598 .get_projid = f2fs_get_projid,
2599 .get_next_id = dquot_get_next_id,
2602 static const struct quotactl_ops f2fs_quotactl_ops = {
2603 .quota_on = f2fs_quota_on,
2604 .quota_off = f2fs_quota_off,
2605 .quota_sync = f2fs_quota_sync,
2606 .get_state = dquot_get_state,
2607 .set_info = dquot_set_dqinfo,
2608 .get_dqblk = dquot_get_dqblk,
2609 .set_dqblk = dquot_set_dqblk,
2610 .get_nextdqblk = dquot_get_next_dqblk,
2613 int f2fs_quota_sync(struct super_block *sb, int type)
2618 void f2fs_quota_off_umount(struct super_block *sb)
2623 static const struct super_operations f2fs_sops = {
2624 .alloc_inode = f2fs_alloc_inode,
2625 .free_inode = f2fs_free_inode,
2626 .drop_inode = f2fs_drop_inode,
2627 .write_inode = f2fs_write_inode,
2628 .dirty_inode = f2fs_dirty_inode,
2629 .show_options = f2fs_show_options,
2631 .quota_read = f2fs_quota_read,
2632 .quota_write = f2fs_quota_write,
2633 .get_dquots = f2fs_get_dquots,
2635 .evict_inode = f2fs_evict_inode,
2636 .put_super = f2fs_put_super,
2637 .sync_fs = f2fs_sync_fs,
2638 .freeze_fs = f2fs_freeze,
2639 .unfreeze_fs = f2fs_unfreeze,
2640 .statfs = f2fs_statfs,
2641 .remount_fs = f2fs_remount,
2644 #ifdef CONFIG_FS_ENCRYPTION
2645 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
2647 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2648 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2652 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
2655 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2658 * Encrypting the root directory is not allowed because fsck
2659 * expects lost+found directory to exist and remain unencrypted
2660 * if LOST_FOUND feature is enabled.
2663 if (f2fs_sb_has_lost_found(sbi) &&
2664 inode->i_ino == F2FS_ROOT_INO(sbi))
2667 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2668 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2669 ctx, len, fs_data, XATTR_CREATE);
2672 static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
2674 return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
2677 static bool f2fs_has_stable_inodes(struct super_block *sb)
2682 static void f2fs_get_ino_and_lblk_bits(struct super_block *sb,
2683 int *ino_bits_ret, int *lblk_bits_ret)
2685 *ino_bits_ret = 8 * sizeof(nid_t);
2686 *lblk_bits_ret = 8 * sizeof(block_t);
2689 static int f2fs_get_num_devices(struct super_block *sb)
2691 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2693 if (f2fs_is_multi_device(sbi))
2694 return sbi->s_ndevs;
2698 static void f2fs_get_devices(struct super_block *sb,
2699 struct request_queue **devs)
2701 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2704 for (i = 0; i < sbi->s_ndevs; i++)
2705 devs[i] = bdev_get_queue(FDEV(i).bdev);
2708 static const struct fscrypt_operations f2fs_cryptops = {
2709 .key_prefix = "f2fs:",
2710 .get_context = f2fs_get_context,
2711 .set_context = f2fs_set_context,
2712 .get_dummy_policy = f2fs_get_dummy_policy,
2713 .empty_dir = f2fs_empty_dir,
2714 .max_namelen = F2FS_NAME_LEN,
2715 .has_stable_inodes = f2fs_has_stable_inodes,
2716 .get_ino_and_lblk_bits = f2fs_get_ino_and_lblk_bits,
2717 .get_num_devices = f2fs_get_num_devices,
2718 .get_devices = f2fs_get_devices,
2722 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
2723 u64 ino, u32 generation)
2725 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2726 struct inode *inode;
2728 if (f2fs_check_nid_range(sbi, ino))
2729 return ERR_PTR(-ESTALE);
2732 * f2fs_iget isn't quite right if the inode is currently unallocated!
2733 * However f2fs_iget currently does appropriate checks to handle stale
2734 * inodes so everything is OK.
2736 inode = f2fs_iget(sb, ino);
2738 return ERR_CAST(inode);
2739 if (unlikely(generation && inode->i_generation != generation)) {
2740 /* we didn't find the right inode.. */
2742 return ERR_PTR(-ESTALE);
2747 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
2748 int fh_len, int fh_type)
2750 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
2751 f2fs_nfs_get_inode);
2754 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
2755 int fh_len, int fh_type)
2757 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
2758 f2fs_nfs_get_inode);
2761 static const struct export_operations f2fs_export_ops = {
2762 .fh_to_dentry = f2fs_fh_to_dentry,
2763 .fh_to_parent = f2fs_fh_to_parent,
2764 .get_parent = f2fs_get_parent,
2767 loff_t max_file_blocks(struct inode *inode)
2773 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
2774 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
2775 * space in inode.i_addr, it will be more safe to reassign
2779 if (inode && f2fs_compressed_file(inode))
2780 leaf_count = ADDRS_PER_BLOCK(inode);
2782 leaf_count = DEF_ADDRS_PER_BLOCK;
2784 /* two direct node blocks */
2785 result += (leaf_count * 2);
2787 /* two indirect node blocks */
2788 leaf_count *= NIDS_PER_BLOCK;
2789 result += (leaf_count * 2);
2791 /* one double indirect node block */
2792 leaf_count *= NIDS_PER_BLOCK;
2793 result += leaf_count;
2798 static int __f2fs_commit_super(struct buffer_head *bh,
2799 struct f2fs_super_block *super)
2803 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
2804 set_buffer_dirty(bh);
2807 /* it's rare case, we can do fua all the time */
2808 return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
2811 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
2812 struct buffer_head *bh)
2814 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2815 (bh->b_data + F2FS_SUPER_OFFSET);
2816 struct super_block *sb = sbi->sb;
2817 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
2818 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
2819 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
2820 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
2821 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
2822 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
2823 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
2824 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
2825 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
2826 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
2827 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2828 u32 segment_count = le32_to_cpu(raw_super->segment_count);
2829 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2830 u64 main_end_blkaddr = main_blkaddr +
2831 (segment_count_main << log_blocks_per_seg);
2832 u64 seg_end_blkaddr = segment0_blkaddr +
2833 (segment_count << log_blocks_per_seg);
2835 if (segment0_blkaddr != cp_blkaddr) {
2836 f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
2837 segment0_blkaddr, cp_blkaddr);
2841 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
2843 f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
2844 cp_blkaddr, sit_blkaddr,
2845 segment_count_ckpt << log_blocks_per_seg);
2849 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
2851 f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
2852 sit_blkaddr, nat_blkaddr,
2853 segment_count_sit << log_blocks_per_seg);
2857 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
2859 f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
2860 nat_blkaddr, ssa_blkaddr,
2861 segment_count_nat << log_blocks_per_seg);
2865 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
2867 f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
2868 ssa_blkaddr, main_blkaddr,
2869 segment_count_ssa << log_blocks_per_seg);
2873 if (main_end_blkaddr > seg_end_blkaddr) {
2874 f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
2875 main_blkaddr, seg_end_blkaddr,
2876 segment_count_main << log_blocks_per_seg);
2878 } else if (main_end_blkaddr < seg_end_blkaddr) {
2882 /* fix in-memory information all the time */
2883 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
2884 segment0_blkaddr) >> log_blocks_per_seg);
2886 if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
2887 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2890 err = __f2fs_commit_super(bh, NULL);
2891 res = err ? "failed" : "done";
2893 f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
2894 res, main_blkaddr, seg_end_blkaddr,
2895 segment_count_main << log_blocks_per_seg);
2902 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
2903 struct buffer_head *bh)
2905 block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
2906 block_t total_sections, blocks_per_seg;
2907 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2908 (bh->b_data + F2FS_SUPER_OFFSET);
2909 size_t crc_offset = 0;
2912 if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
2913 f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
2914 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
2918 /* Check checksum_offset and crc in superblock */
2919 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
2920 crc_offset = le32_to_cpu(raw_super->checksum_offset);
2922 offsetof(struct f2fs_super_block, crc)) {
2923 f2fs_info(sbi, "Invalid SB checksum offset: %zu",
2925 return -EFSCORRUPTED;
2927 crc = le32_to_cpu(raw_super->crc);
2928 if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
2929 f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
2930 return -EFSCORRUPTED;
2934 /* Currently, support only 4KB block size */
2935 if (le32_to_cpu(raw_super->log_blocksize) != F2FS_BLKSIZE_BITS) {
2936 f2fs_info(sbi, "Invalid log_blocksize (%u), supports only %u",
2937 le32_to_cpu(raw_super->log_blocksize),
2939 return -EFSCORRUPTED;
2942 /* check log blocks per segment */
2943 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
2944 f2fs_info(sbi, "Invalid log blocks per segment (%u)",
2945 le32_to_cpu(raw_super->log_blocks_per_seg));
2946 return -EFSCORRUPTED;
2949 /* Currently, support 512/1024/2048/4096 bytes sector size */
2950 if (le32_to_cpu(raw_super->log_sectorsize) >
2951 F2FS_MAX_LOG_SECTOR_SIZE ||
2952 le32_to_cpu(raw_super->log_sectorsize) <
2953 F2FS_MIN_LOG_SECTOR_SIZE) {
2954 f2fs_info(sbi, "Invalid log sectorsize (%u)",
2955 le32_to_cpu(raw_super->log_sectorsize));
2956 return -EFSCORRUPTED;
2958 if (le32_to_cpu(raw_super->log_sectors_per_block) +
2959 le32_to_cpu(raw_super->log_sectorsize) !=
2960 F2FS_MAX_LOG_SECTOR_SIZE) {
2961 f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
2962 le32_to_cpu(raw_super->log_sectors_per_block),
2963 le32_to_cpu(raw_super->log_sectorsize));
2964 return -EFSCORRUPTED;
2967 segment_count = le32_to_cpu(raw_super->segment_count);
2968 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2969 segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
2970 secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
2971 total_sections = le32_to_cpu(raw_super->section_count);
2973 /* blocks_per_seg should be 512, given the above check */
2974 blocks_per_seg = 1 << le32_to_cpu(raw_super->log_blocks_per_seg);
2976 if (segment_count > F2FS_MAX_SEGMENT ||
2977 segment_count < F2FS_MIN_SEGMENTS) {
2978 f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
2979 return -EFSCORRUPTED;
2982 if (total_sections > segment_count_main || total_sections < 1 ||
2983 segs_per_sec > segment_count || !segs_per_sec) {
2984 f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
2985 segment_count, total_sections, segs_per_sec);
2986 return -EFSCORRUPTED;
2989 if (segment_count_main != total_sections * segs_per_sec) {
2990 f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
2991 segment_count_main, total_sections, segs_per_sec);
2992 return -EFSCORRUPTED;
2995 if ((segment_count / segs_per_sec) < total_sections) {
2996 f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
2997 segment_count, segs_per_sec, total_sections);
2998 return -EFSCORRUPTED;
3001 if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
3002 f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
3003 segment_count, le64_to_cpu(raw_super->block_count));
3004 return -EFSCORRUPTED;
3007 if (RDEV(0).path[0]) {
3008 block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
3011 while (i < MAX_DEVICES && RDEV(i).path[0]) {
3012 dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
3015 if (segment_count != dev_seg_count) {
3016 f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
3017 segment_count, dev_seg_count);
3018 return -EFSCORRUPTED;
3021 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
3022 !bdev_is_zoned(sbi->sb->s_bdev)) {
3023 f2fs_info(sbi, "Zoned block device path is missing");
3024 return -EFSCORRUPTED;
3028 if (secs_per_zone > total_sections || !secs_per_zone) {
3029 f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
3030 secs_per_zone, total_sections);
3031 return -EFSCORRUPTED;
3033 if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
3034 raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
3035 (le32_to_cpu(raw_super->extension_count) +
3036 raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
3037 f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
3038 le32_to_cpu(raw_super->extension_count),
3039 raw_super->hot_ext_count,
3040 F2FS_MAX_EXTENSION);
3041 return -EFSCORRUPTED;
3044 if (le32_to_cpu(raw_super->cp_payload) >
3045 (blocks_per_seg - F2FS_CP_PACKS)) {
3046 f2fs_info(sbi, "Insane cp_payload (%u > %u)",
3047 le32_to_cpu(raw_super->cp_payload),
3048 blocks_per_seg - F2FS_CP_PACKS);
3049 return -EFSCORRUPTED;
3052 /* check reserved ino info */
3053 if (le32_to_cpu(raw_super->node_ino) != 1 ||
3054 le32_to_cpu(raw_super->meta_ino) != 2 ||
3055 le32_to_cpu(raw_super->root_ino) != 3) {
3056 f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
3057 le32_to_cpu(raw_super->node_ino),
3058 le32_to_cpu(raw_super->meta_ino),
3059 le32_to_cpu(raw_super->root_ino));
3060 return -EFSCORRUPTED;
3063 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
3064 if (sanity_check_area_boundary(sbi, bh))
3065 return -EFSCORRUPTED;
3070 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
3072 unsigned int total, fsmeta;
3073 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3074 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
3075 unsigned int ovp_segments, reserved_segments;
3076 unsigned int main_segs, blocks_per_seg;
3077 unsigned int sit_segs, nat_segs;
3078 unsigned int sit_bitmap_size, nat_bitmap_size;
3079 unsigned int log_blocks_per_seg;
3080 unsigned int segment_count_main;
3081 unsigned int cp_pack_start_sum, cp_payload;
3082 block_t user_block_count, valid_user_blocks;
3083 block_t avail_node_count, valid_node_count;
3086 total = le32_to_cpu(raw_super->segment_count);
3087 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
3088 sit_segs = le32_to_cpu(raw_super->segment_count_sit);
3090 nat_segs = le32_to_cpu(raw_super->segment_count_nat);
3092 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
3093 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
3095 if (unlikely(fsmeta >= total))
3098 ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
3099 reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
3101 if (unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
3102 ovp_segments == 0 || reserved_segments == 0)) {
3103 f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
3107 user_block_count = le64_to_cpu(ckpt->user_block_count);
3108 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3109 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3110 if (!user_block_count || user_block_count >=
3111 segment_count_main << log_blocks_per_seg) {
3112 f2fs_err(sbi, "Wrong user_block_count: %u",
3117 valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
3118 if (valid_user_blocks > user_block_count) {
3119 f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
3120 valid_user_blocks, user_block_count);
3124 valid_node_count = le32_to_cpu(ckpt->valid_node_count);
3125 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
3126 if (valid_node_count > avail_node_count) {
3127 f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
3128 valid_node_count, avail_node_count);
3132 main_segs = le32_to_cpu(raw_super->segment_count_main);
3133 blocks_per_seg = sbi->blocks_per_seg;
3135 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3136 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
3137 le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
3139 for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
3140 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3141 le32_to_cpu(ckpt->cur_node_segno[j])) {
3142 f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
3144 le32_to_cpu(ckpt->cur_node_segno[i]));
3149 for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
3150 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
3151 le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
3153 for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3154 if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3155 le32_to_cpu(ckpt->cur_data_segno[j])) {
3156 f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3158 le32_to_cpu(ckpt->cur_data_segno[i]));
3163 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3164 for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3165 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3166 le32_to_cpu(ckpt->cur_data_segno[j])) {
3167 f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3169 le32_to_cpu(ckpt->cur_node_segno[i]));
3175 sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3176 nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3178 if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3179 nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3180 f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3181 sit_bitmap_size, nat_bitmap_size);
3185 cp_pack_start_sum = __start_sum_addr(sbi);
3186 cp_payload = __cp_payload(sbi);
3187 if (cp_pack_start_sum < cp_payload + 1 ||
3188 cp_pack_start_sum > blocks_per_seg - 1 -
3189 NR_CURSEG_PERSIST_TYPE) {
3190 f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3195 if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3196 le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3197 f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3198 "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3199 "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3200 le32_to_cpu(ckpt->checksum_offset));
3204 if (unlikely(f2fs_cp_error(sbi))) {
3205 f2fs_err(sbi, "A bug case: need to run fsck");
3211 static void init_sb_info(struct f2fs_sb_info *sbi)
3213 struct f2fs_super_block *raw_super = sbi->raw_super;
3216 sbi->log_sectors_per_block =
3217 le32_to_cpu(raw_super->log_sectors_per_block);
3218 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3219 sbi->blocksize = 1 << sbi->log_blocksize;
3220 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3221 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
3222 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3223 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3224 sbi->total_sections = le32_to_cpu(raw_super->section_count);
3225 sbi->total_node_count =
3226 (le32_to_cpu(raw_super->segment_count_nat) / 2)
3227 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
3228 F2FS_ROOT_INO(sbi) = le32_to_cpu(raw_super->root_ino);
3229 F2FS_NODE_INO(sbi) = le32_to_cpu(raw_super->node_ino);
3230 F2FS_META_INO(sbi) = le32_to_cpu(raw_super->meta_ino);
3231 sbi->cur_victim_sec = NULL_SECNO;
3232 sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3233 sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3234 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3235 sbi->migration_granularity = sbi->segs_per_sec;
3237 sbi->dir_level = DEF_DIR_LEVEL;
3238 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3239 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3240 sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3241 sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3242 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3243 sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3244 DEF_UMOUNT_DISCARD_TIMEOUT;
3245 clear_sbi_flag(sbi, SBI_NEED_FSCK);
3247 for (i = 0; i < NR_COUNT_TYPE; i++)
3248 atomic_set(&sbi->nr_pages[i], 0);
3250 for (i = 0; i < META; i++)
3251 atomic_set(&sbi->wb_sync_req[i], 0);
3253 INIT_LIST_HEAD(&sbi->s_list);
3254 mutex_init(&sbi->umount_mutex);
3255 init_rwsem(&sbi->io_order_lock);
3256 spin_lock_init(&sbi->cp_lock);
3258 sbi->dirty_device = 0;
3259 spin_lock_init(&sbi->dev_lock);
3261 init_rwsem(&sbi->sb_lock);
3262 init_rwsem(&sbi->pin_sem);
3265 static int init_percpu_info(struct f2fs_sb_info *sbi)
3269 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3273 err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3276 percpu_counter_destroy(&sbi->alloc_valid_block_count);
3281 #ifdef CONFIG_BLK_DEV_ZONED
3283 struct f2fs_report_zones_args {
3284 struct f2fs_dev_info *dev;
3285 bool zone_cap_mismatch;
3288 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3291 struct f2fs_report_zones_args *rz_args = data;
3293 if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3296 set_bit(idx, rz_args->dev->blkz_seq);
3297 rz_args->dev->zone_capacity_blocks[idx] = zone->capacity >>
3298 F2FS_LOG_SECTORS_PER_BLOCK;
3299 if (zone->len != zone->capacity && !rz_args->zone_cap_mismatch)
3300 rz_args->zone_cap_mismatch = true;
3305 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3307 struct block_device *bdev = FDEV(devi).bdev;
3308 sector_t nr_sectors = bdev_nr_sectors(bdev);
3309 struct f2fs_report_zones_args rep_zone_arg;
3312 if (!f2fs_sb_has_blkzoned(sbi))
3315 if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3316 SECTOR_TO_BLOCK(bdev_zone_sectors(bdev)))
3318 sbi->blocks_per_blkz = SECTOR_TO_BLOCK(bdev_zone_sectors(bdev));
3319 if (sbi->log_blocks_per_blkz && sbi->log_blocks_per_blkz !=
3320 __ilog2_u32(sbi->blocks_per_blkz))
3322 sbi->log_blocks_per_blkz = __ilog2_u32(sbi->blocks_per_blkz);
3323 FDEV(devi).nr_blkz = SECTOR_TO_BLOCK(nr_sectors) >>
3324 sbi->log_blocks_per_blkz;
3325 if (nr_sectors & (bdev_zone_sectors(bdev) - 1))
3326 FDEV(devi).nr_blkz++;
3328 FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3329 BITS_TO_LONGS(FDEV(devi).nr_blkz)
3330 * sizeof(unsigned long),
3332 if (!FDEV(devi).blkz_seq)
3335 /* Get block zones type and zone-capacity */
3336 FDEV(devi).zone_capacity_blocks = f2fs_kzalloc(sbi,
3337 FDEV(devi).nr_blkz * sizeof(block_t),
3339 if (!FDEV(devi).zone_capacity_blocks)
3342 rep_zone_arg.dev = &FDEV(devi);
3343 rep_zone_arg.zone_cap_mismatch = false;
3345 ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3350 if (!rep_zone_arg.zone_cap_mismatch) {
3351 kfree(FDEV(devi).zone_capacity_blocks);
3352 FDEV(devi).zone_capacity_blocks = NULL;
3360 * Read f2fs raw super block.
3361 * Because we have two copies of super block, so read both of them
3362 * to get the first valid one. If any one of them is broken, we pass
3363 * them recovery flag back to the caller.
3365 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3366 struct f2fs_super_block **raw_super,
3367 int *valid_super_block, int *recovery)
3369 struct super_block *sb = sbi->sb;
3371 struct buffer_head *bh;
3372 struct f2fs_super_block *super;
3375 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3379 for (block = 0; block < 2; block++) {
3380 bh = sb_bread(sb, block);
3382 f2fs_err(sbi, "Unable to read %dth superblock",
3389 /* sanity checking of raw super */
3390 err = sanity_check_raw_super(sbi, bh);
3392 f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3400 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3402 *valid_super_block = block;
3408 /* No valid superblock */
3417 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
3419 struct buffer_head *bh;
3423 if ((recover && f2fs_readonly(sbi->sb)) ||
3424 bdev_read_only(sbi->sb->s_bdev)) {
3425 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3429 /* we should update superblock crc here */
3430 if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
3431 crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
3432 offsetof(struct f2fs_super_block, crc));
3433 F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
3436 /* write back-up superblock first */
3437 bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
3440 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3443 /* if we are in recovery path, skip writing valid superblock */
3447 /* write current valid superblock */
3448 bh = sb_bread(sbi->sb, sbi->valid_super_block);
3451 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3456 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
3458 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3459 unsigned int max_devices = MAX_DEVICES;
3462 /* Initialize single device information */
3463 if (!RDEV(0).path[0]) {
3464 if (!bdev_is_zoned(sbi->sb->s_bdev))
3470 * Initialize multiple devices information, or single
3471 * zoned block device information.
3473 sbi->devs = f2fs_kzalloc(sbi,
3474 array_size(max_devices,
3475 sizeof(struct f2fs_dev_info)),
3480 for (i = 0; i < max_devices; i++) {
3482 if (i > 0 && !RDEV(i).path[0])
3485 if (max_devices == 1) {
3486 /* Single zoned block device mount */
3488 blkdev_get_by_dev(sbi->sb->s_bdev->bd_dev,
3489 sbi->sb->s_mode, sbi->sb->s_type);
3491 /* Multi-device mount */
3492 memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
3493 FDEV(i).total_segments =
3494 le32_to_cpu(RDEV(i).total_segments);
3496 FDEV(i).start_blk = 0;
3497 FDEV(i).end_blk = FDEV(i).start_blk +
3498 (FDEV(i).total_segments <<
3499 sbi->log_blocks_per_seg) - 1 +
3500 le32_to_cpu(raw_super->segment0_blkaddr);
3502 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
3503 FDEV(i).end_blk = FDEV(i).start_blk +
3504 (FDEV(i).total_segments <<
3505 sbi->log_blocks_per_seg) - 1;
3507 FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
3508 sbi->sb->s_mode, sbi->sb->s_type);
3510 if (IS_ERR(FDEV(i).bdev))
3511 return PTR_ERR(FDEV(i).bdev);
3513 /* to release errored devices */
3514 sbi->s_ndevs = i + 1;
3516 #ifdef CONFIG_BLK_DEV_ZONED
3517 if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
3518 !f2fs_sb_has_blkzoned(sbi)) {
3519 f2fs_err(sbi, "Zoned block device feature not enabled\n");
3522 if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
3523 if (init_blkz_info(sbi, i)) {
3524 f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
3527 if (max_devices == 1)
3529 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
3531 FDEV(i).total_segments,
3532 FDEV(i).start_blk, FDEV(i).end_blk,
3533 bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
3534 "Host-aware" : "Host-managed");
3538 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
3540 FDEV(i).total_segments,
3541 FDEV(i).start_blk, FDEV(i).end_blk);
3544 "IO Block Size: %8d KB", F2FS_IO_SIZE_KB(sbi));
3548 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
3550 #ifdef CONFIG_UNICODE
3551 if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
3552 const struct f2fs_sb_encodings *encoding_info;
3553 struct unicode_map *encoding;
3554 __u16 encoding_flags;
3556 if (f2fs_sb_read_encoding(sbi->raw_super, &encoding_info,
3559 "Encoding requested by superblock is unknown");
3563 encoding = utf8_load(encoding_info->version);
3564 if (IS_ERR(encoding)) {
3566 "can't mount with superblock charset: %s-%s "
3567 "not supported by the kernel. flags: 0x%x.",
3568 encoding_info->name, encoding_info->version,
3570 return PTR_ERR(encoding);
3572 f2fs_info(sbi, "Using encoding defined by superblock: "
3573 "%s-%s with flags 0x%hx", encoding_info->name,
3574 encoding_info->version?:"\b", encoding_flags);
3576 sbi->sb->s_encoding = encoding;
3577 sbi->sb->s_encoding_flags = encoding_flags;
3580 if (f2fs_sb_has_casefold(sbi)) {
3581 f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
3588 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
3590 struct f2fs_sm_info *sm_i = SM_I(sbi);
3592 /* adjust parameters according to the volume size */
3593 if (sm_i->main_segments <= SMALL_VOLUME_SEGMENTS) {
3594 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
3595 sm_i->dcc_info->discard_granularity = 1;
3596 sm_i->ipu_policy = 1 << F2FS_IPU_FORCE;
3599 sbi->readdir_ra = 1;
3602 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
3604 struct f2fs_sb_info *sbi;
3605 struct f2fs_super_block *raw_super;
3608 bool skip_recovery = false, need_fsck = false;
3609 char *options = NULL;
3610 int recovery, i, valid_super_block;
3611 struct curseg_info *seg_i;
3617 valid_super_block = -1;
3620 /* allocate memory for f2fs-specific super block info */
3621 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
3627 /* Load the checksum driver */
3628 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
3629 if (IS_ERR(sbi->s_chksum_driver)) {
3630 f2fs_err(sbi, "Cannot load crc32 driver.");
3631 err = PTR_ERR(sbi->s_chksum_driver);
3632 sbi->s_chksum_driver = NULL;
3636 /* set a block size */
3637 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
3638 f2fs_err(sbi, "unable to set blocksize");
3642 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
3647 sb->s_fs_info = sbi;
3648 sbi->raw_super = raw_super;
3650 /* precompute checksum seed for metadata */
3651 if (f2fs_sb_has_inode_chksum(sbi))
3652 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
3653 sizeof(raw_super->uuid));
3655 default_options(sbi);
3656 /* parse mount options */
3657 options = kstrdup((const char *)data, GFP_KERNEL);
3658 if (data && !options) {
3663 err = parse_options(sb, options, false);
3667 sb->s_maxbytes = max_file_blocks(NULL) <<
3668 le32_to_cpu(raw_super->log_blocksize);
3669 sb->s_max_links = F2FS_LINK_MAX;
3671 err = f2fs_setup_casefold(sbi);
3676 sb->dq_op = &f2fs_quota_operations;
3677 sb->s_qcop = &f2fs_quotactl_ops;
3678 sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
3680 if (f2fs_sb_has_quota_ino(sbi)) {
3681 for (i = 0; i < MAXQUOTAS; i++) {
3682 if (f2fs_qf_ino(sbi->sb, i))
3683 sbi->nquota_files++;
3688 sb->s_op = &f2fs_sops;
3689 #ifdef CONFIG_FS_ENCRYPTION
3690 sb->s_cop = &f2fs_cryptops;
3692 #ifdef CONFIG_FS_VERITY
3693 sb->s_vop = &f2fs_verityops;
3695 sb->s_xattr = f2fs_xattr_handlers;
3696 sb->s_export_op = &f2fs_export_ops;
3697 sb->s_magic = F2FS_SUPER_MAGIC;
3698 sb->s_time_gran = 1;
3699 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
3700 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
3701 memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
3702 sb->s_iflags |= SB_I_CGROUPWB;
3704 /* init f2fs-specific super block info */
3705 sbi->valid_super_block = valid_super_block;
3706 init_rwsem(&sbi->gc_lock);
3707 mutex_init(&sbi->writepages);
3708 init_rwsem(&sbi->cp_global_sem);
3709 init_rwsem(&sbi->node_write);
3710 init_rwsem(&sbi->node_change);
3712 /* disallow all the data/node/meta page writes */
3713 set_sbi_flag(sbi, SBI_POR_DOING);
3714 spin_lock_init(&sbi->stat_lock);
3716 /* init iostat info */
3717 spin_lock_init(&sbi->iostat_lock);
3718 sbi->iostat_enable = false;
3719 sbi->iostat_period_ms = DEFAULT_IOSTAT_PERIOD_MS;
3721 for (i = 0; i < NR_PAGE_TYPE; i++) {
3722 int n = (i == META) ? 1: NR_TEMP_TYPE;
3728 sizeof(struct f2fs_bio_info)),
3730 if (!sbi->write_io[i]) {
3735 for (j = HOT; j < n; j++) {
3736 init_rwsem(&sbi->write_io[i][j].io_rwsem);
3737 sbi->write_io[i][j].sbi = sbi;
3738 sbi->write_io[i][j].bio = NULL;
3739 spin_lock_init(&sbi->write_io[i][j].io_lock);
3740 INIT_LIST_HEAD(&sbi->write_io[i][j].io_list);
3741 INIT_LIST_HEAD(&sbi->write_io[i][j].bio_list);
3742 init_rwsem(&sbi->write_io[i][j].bio_list_lock);
3746 init_rwsem(&sbi->cp_rwsem);
3747 init_rwsem(&sbi->quota_sem);
3748 init_waitqueue_head(&sbi->cp_wait);
3751 err = init_percpu_info(sbi);
3755 if (F2FS_IO_ALIGNED(sbi)) {
3756 sbi->write_io_dummy =
3757 mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
3758 if (!sbi->write_io_dummy) {
3764 /* init per sbi slab cache */
3765 err = f2fs_init_xattr_caches(sbi);
3768 err = f2fs_init_page_array_cache(sbi);
3770 goto free_xattr_cache;
3772 /* get an inode for meta space */
3773 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
3774 if (IS_ERR(sbi->meta_inode)) {
3775 f2fs_err(sbi, "Failed to read F2FS meta data inode");
3776 err = PTR_ERR(sbi->meta_inode);
3777 goto free_page_array_cache;
3780 err = f2fs_get_valid_checkpoint(sbi);
3782 f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
3783 goto free_meta_inode;
3786 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
3787 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3788 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
3789 set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
3790 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
3793 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
3794 set_sbi_flag(sbi, SBI_NEED_FSCK);
3796 /* Initialize device list */
3797 err = f2fs_scan_devices(sbi);
3799 f2fs_err(sbi, "Failed to find devices");
3803 err = f2fs_init_post_read_wq(sbi);
3805 f2fs_err(sbi, "Failed to initialize post read workqueue");
3809 sbi->total_valid_node_count =
3810 le32_to_cpu(sbi->ckpt->valid_node_count);
3811 percpu_counter_set(&sbi->total_valid_inode_count,
3812 le32_to_cpu(sbi->ckpt->valid_inode_count));
3813 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
3814 sbi->total_valid_block_count =
3815 le64_to_cpu(sbi->ckpt->valid_block_count);
3816 sbi->last_valid_block_count = sbi->total_valid_block_count;
3817 sbi->reserved_blocks = 0;
3818 sbi->current_reserved_blocks = 0;
3819 limit_reserve_root(sbi);
3820 adjust_unusable_cap_perc(sbi);
3822 for (i = 0; i < NR_INODE_TYPE; i++) {
3823 INIT_LIST_HEAD(&sbi->inode_list[i]);
3824 spin_lock_init(&sbi->inode_lock[i]);
3826 mutex_init(&sbi->flush_lock);
3828 f2fs_init_extent_cache_info(sbi);
3830 f2fs_init_ino_entry_info(sbi);
3832 f2fs_init_fsync_node_info(sbi);
3834 /* setup checkpoint request control and start checkpoint issue thread */
3835 f2fs_init_ckpt_req_control(sbi);
3836 if (!test_opt(sbi, DISABLE_CHECKPOINT) &&
3837 test_opt(sbi, MERGE_CHECKPOINT)) {
3838 err = f2fs_start_ckpt_thread(sbi);
3841 "Failed to start F2FS issue_checkpoint_thread (%d)",
3843 goto stop_ckpt_thread;
3847 /* setup f2fs internal modules */
3848 err = f2fs_build_segment_manager(sbi);
3850 f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
3854 err = f2fs_build_node_manager(sbi);
3856 f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
3861 /* For write statistics */
3862 sbi->sectors_written_start = f2fs_get_sectors_written(sbi);
3864 /* Read accumulated write IO statistics if exists */
3865 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
3866 if (__exist_node_summaries(sbi))
3867 sbi->kbytes_written =
3868 le64_to_cpu(seg_i->journal->info.kbytes_written);
3870 f2fs_build_gc_manager(sbi);
3872 err = f2fs_build_stats(sbi);
3876 /* get an inode for node space */
3877 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
3878 if (IS_ERR(sbi->node_inode)) {
3879 f2fs_err(sbi, "Failed to read node inode");
3880 err = PTR_ERR(sbi->node_inode);
3884 /* read root inode and dentry */
3885 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
3887 f2fs_err(sbi, "Failed to read root inode");
3888 err = PTR_ERR(root);
3889 goto free_node_inode;
3891 if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
3892 !root->i_size || !root->i_nlink) {
3895 goto free_node_inode;
3898 sb->s_root = d_make_root(root); /* allocate root dentry */
3901 goto free_node_inode;
3904 err = f2fs_register_sysfs(sbi);
3906 goto free_root_inode;
3909 /* Enable quota usage during mount */
3910 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
3911 err = f2fs_enable_quotas(sb);
3913 f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
3916 /* if there are any orphan inodes, free them */
3917 err = f2fs_recover_orphan_inodes(sbi);
3921 if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
3922 goto reset_checkpoint;
3924 /* recover fsynced data */
3925 if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
3926 !test_opt(sbi, NORECOVERY)) {
3928 * mount should be failed, when device has readonly mode, and
3929 * previous checkpoint was not done by clean system shutdown.
3931 if (f2fs_hw_is_readonly(sbi)) {
3932 if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))
3933 f2fs_err(sbi, "Need to recover fsync data, but write access unavailable");
3935 f2fs_info(sbi, "write access unavailable, skipping recovery");
3936 goto reset_checkpoint;
3940 set_sbi_flag(sbi, SBI_NEED_FSCK);
3943 goto reset_checkpoint;
3945 err = f2fs_recover_fsync_data(sbi, false);
3948 skip_recovery = true;
3950 f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
3955 err = f2fs_recover_fsync_data(sbi, true);
3957 if (!f2fs_readonly(sb) && err > 0) {
3959 f2fs_err(sbi, "Need to recover fsync data");
3965 * If the f2fs is not readonly and fsync data recovery succeeds,
3966 * check zoned block devices' write pointer consistency.
3968 if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
3969 err = f2fs_check_write_pointer(sbi);
3975 f2fs_init_inmem_curseg(sbi);
3977 /* f2fs_recover_fsync_data() cleared this already */
3978 clear_sbi_flag(sbi, SBI_POR_DOING);
3980 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
3981 err = f2fs_disable_checkpoint(sbi);
3983 goto sync_free_meta;
3984 } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
3985 f2fs_enable_checkpoint(sbi);
3989 * If filesystem is not mounted as read-only then
3990 * do start the gc_thread.
3992 if (F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF && !f2fs_readonly(sb)) {
3993 /* After POR, we can run background GC thread.*/
3994 err = f2fs_start_gc_thread(sbi);
3996 goto sync_free_meta;
4000 /* recover broken superblock */
4002 err = f2fs_commit_super(sbi, true);
4003 f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
4004 sbi->valid_super_block ? 1 : 2, err);
4007 f2fs_join_shrinker(sbi);
4009 f2fs_tuning_parameters(sbi);
4011 f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
4012 cur_cp_version(F2FS_CKPT(sbi)));
4013 f2fs_update_time(sbi, CP_TIME);
4014 f2fs_update_time(sbi, REQ_TIME);
4015 clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4019 /* safe to flush all the data */
4020 sync_filesystem(sbi->sb);
4025 f2fs_truncate_quota_inode_pages(sb);
4026 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
4027 f2fs_quota_off_umount(sbi->sb);
4030 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
4031 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
4032 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
4033 * falls into an infinite loop in f2fs_sync_meta_pages().
4035 truncate_inode_pages_final(META_MAPPING(sbi));
4036 /* evict some inodes being cached by GC */
4038 f2fs_unregister_sysfs(sbi);
4043 f2fs_release_ino_entry(sbi, true);
4044 truncate_inode_pages_final(NODE_MAPPING(sbi));
4045 iput(sbi->node_inode);
4046 sbi->node_inode = NULL;
4048 f2fs_destroy_stats(sbi);
4050 f2fs_destroy_node_manager(sbi);
4052 f2fs_destroy_segment_manager(sbi);
4053 f2fs_destroy_post_read_wq(sbi);
4055 f2fs_stop_ckpt_thread(sbi);
4057 destroy_device_list(sbi);
4060 make_bad_inode(sbi->meta_inode);
4061 iput(sbi->meta_inode);
4062 sbi->meta_inode = NULL;
4063 free_page_array_cache:
4064 f2fs_destroy_page_array_cache(sbi);
4066 f2fs_destroy_xattr_caches(sbi);
4068 mempool_destroy(sbi->write_io_dummy);
4070 destroy_percpu_info(sbi);
4072 for (i = 0; i < NR_PAGE_TYPE; i++)
4073 kvfree(sbi->write_io[i]);
4075 #ifdef CONFIG_UNICODE
4076 utf8_unload(sb->s_encoding);
4077 sb->s_encoding = NULL;
4081 for (i = 0; i < MAXQUOTAS; i++)
4082 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
4084 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
4089 if (sbi->s_chksum_driver)
4090 crypto_free_shash(sbi->s_chksum_driver);
4093 /* give only one another chance */
4094 if (retry_cnt > 0 && skip_recovery) {
4096 shrink_dcache_sb(sb);
4102 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
4103 const char *dev_name, void *data)
4105 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
4108 static void kill_f2fs_super(struct super_block *sb)
4111 struct f2fs_sb_info *sbi = F2FS_SB(sb);
4113 set_sbi_flag(sbi, SBI_IS_CLOSE);
4114 f2fs_stop_gc_thread(sbi);
4115 f2fs_stop_discard_thread(sbi);
4117 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
4118 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4119 struct cp_control cpc = {
4120 .reason = CP_UMOUNT,
4122 f2fs_write_checkpoint(sbi, &cpc);
4125 if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
4126 sb->s_flags &= ~SB_RDONLY;
4128 kill_block_super(sb);
4131 static struct file_system_type f2fs_fs_type = {
4132 .owner = THIS_MODULE,
4134 .mount = f2fs_mount,
4135 .kill_sb = kill_f2fs_super,
4136 .fs_flags = FS_REQUIRES_DEV,
4138 MODULE_ALIAS_FS("f2fs");
4140 static int __init init_inodecache(void)
4142 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
4143 sizeof(struct f2fs_inode_info), 0,
4144 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
4145 if (!f2fs_inode_cachep)
4150 static void destroy_inodecache(void)
4153 * Make sure all delayed rcu free inodes are flushed before we
4157 kmem_cache_destroy(f2fs_inode_cachep);
4160 static int __init init_f2fs_fs(void)
4164 if (PAGE_SIZE != F2FS_BLKSIZE) {
4165 printk("F2FS not supported on PAGE_SIZE(%lu) != %d\n",
4166 PAGE_SIZE, F2FS_BLKSIZE);
4170 err = init_inodecache();
4173 err = f2fs_create_node_manager_caches();
4175 goto free_inodecache;
4176 err = f2fs_create_segment_manager_caches();
4178 goto free_node_manager_caches;
4179 err = f2fs_create_checkpoint_caches();
4181 goto free_segment_manager_caches;
4182 err = f2fs_create_extent_cache();
4184 goto free_checkpoint_caches;
4185 err = f2fs_create_garbage_collection_cache();
4187 goto free_extent_cache;
4188 err = f2fs_init_sysfs();
4190 goto free_garbage_collection_cache;
4191 err = register_shrinker(&f2fs_shrinker_info);
4194 err = register_filesystem(&f2fs_fs_type);
4197 f2fs_create_root_stats();
4198 err = f2fs_init_post_read_processing();
4200 goto free_root_stats;
4201 err = f2fs_init_bio_entry_cache();
4203 goto free_post_read;
4204 err = f2fs_init_bioset();
4206 goto free_bio_enrty_cache;
4207 err = f2fs_init_compress_mempool();
4210 err = f2fs_init_compress_cache();
4212 goto free_compress_mempool;
4214 free_compress_mempool:
4215 f2fs_destroy_compress_mempool();
4217 f2fs_destroy_bioset();
4218 free_bio_enrty_cache:
4219 f2fs_destroy_bio_entry_cache();
4221 f2fs_destroy_post_read_processing();
4223 f2fs_destroy_root_stats();
4224 unregister_filesystem(&f2fs_fs_type);
4226 unregister_shrinker(&f2fs_shrinker_info);
4229 free_garbage_collection_cache:
4230 f2fs_destroy_garbage_collection_cache();
4232 f2fs_destroy_extent_cache();
4233 free_checkpoint_caches:
4234 f2fs_destroy_checkpoint_caches();
4235 free_segment_manager_caches:
4236 f2fs_destroy_segment_manager_caches();
4237 free_node_manager_caches:
4238 f2fs_destroy_node_manager_caches();
4240 destroy_inodecache();
4245 static void __exit exit_f2fs_fs(void)
4247 f2fs_destroy_compress_cache();
4248 f2fs_destroy_compress_mempool();
4249 f2fs_destroy_bioset();
4250 f2fs_destroy_bio_entry_cache();
4251 f2fs_destroy_post_read_processing();
4252 f2fs_destroy_root_stats();
4253 unregister_filesystem(&f2fs_fs_type);
4254 unregister_shrinker(&f2fs_shrinker_info);
4256 f2fs_destroy_garbage_collection_cache();
4257 f2fs_destroy_extent_cache();
4258 f2fs_destroy_checkpoint_caches();
4259 f2fs_destroy_segment_manager_caches();
4260 f2fs_destroy_node_manager_caches();
4261 destroy_inodecache();
4264 module_init(init_f2fs_fs)
4265 module_exit(exit_f2fs_fs)
4267 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
4268 MODULE_DESCRIPTION("Flash Friendly File System");
4269 MODULE_LICENSE("GPL");