fs/f2fs/super.c

   1 // SPDX-License-Identifier: GPL-2.0
   2 /*
   3  * fs/f2fs/super.c
   4  *
   5  * Copyright (c) 2012 Samsung Electronics Co., Ltd.
   6  *             http://www.samsung.com/
   7  */
   8 #include <linux/module.h>
   9 #include <linux/init.h>
  10 #include <linux/fs.h>
  11 #include <linux/fs_context.h>
  12 #include <linux/sched/mm.h>
  13 #include <linux/statfs.h>
  14 #include <linux/buffer_head.h>
  15 #include <linux/kthread.h>
  16 #include <linux/parser.h>
  17 #include <linux/mount.h>
  18 #include <linux/seq_file.h>
  19 #include <linux/proc_fs.h>
  20 #include <linux/random.h>
  21 #include <linux/exportfs.h>
  22 #include <linux/blkdev.h>
  23 #include <linux/quotaops.h>
  24 #include <linux/f2fs_fs.h>
  25 #include <linux/sysfs.h>
  26 #include <linux/quota.h>
  27 #include <linux/unicode.h>
  28 #include <linux/part_stat.h>
  29 #include <linux/zstd.h>
  30 #include <linux/lz4.h>
  31
  32 #include "f2fs.h"
  33 #include "node.h"
  34 #include "segment.h"
  35 #include "xattr.h"
  36 #include "gc.h"
  37 #include "iostat.h"
  38
  39 #define CREATE_TRACE_POINTS
  40 #include <trace/events/f2fs.h>
  41
  42 static struct kmem_cache *f2fs_inode_cachep;
  43
  44 #ifdef CONFIG_F2FS_FAULT_INJECTION
  45
  46 const char *f2fs_fault_name[FAULT_MAX] = {
  47         [FAULT_KMALLOC]                 = "kmalloc",
  48         [FAULT_KVMALLOC]                = "kvmalloc",
  49         [FAULT_PAGE_ALLOC]              = "page alloc",
  50         [FAULT_PAGE_GET]                = "page get",
  51         [FAULT_ALLOC_NID]               = "alloc nid",
  52         [FAULT_ORPHAN]                  = "orphan",
  53         [FAULT_BLOCK]                   = "no more block",
  54         [FAULT_DIR_DEPTH]               = "too big dir depth",
  55         [FAULT_EVICT_INODE]             = "evict_inode fail",
  56         [FAULT_TRUNCATE]                = "truncate fail",
  57         [FAULT_READ_IO]                 = "read IO error",
  58         [FAULT_CHECKPOINT]              = "checkpoint error",
  59         [FAULT_DISCARD]                 = "discard error",
  60         [FAULT_WRITE_IO]                = "write IO error",
  61         [FAULT_SLAB_ALLOC]              = "slab alloc",
  62         [FAULT_DQUOT_INIT]              = "dquot initialize",
  63         [FAULT_LOCK_OP]                 = "lock_op",
  64         [FAULT_BLKADDR_VALIDITY]        = "invalid blkaddr",
  65         [FAULT_BLKADDR_CONSISTENCE]     = "inconsistent blkaddr",
  66         [FAULT_NO_SEGMENT]              = "no free segment",
  67 };
  68
  69 int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate,
  70                                                         unsigned long type)
  71 {
  72         struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
  73
  74         if (rate) {
  75                 if (rate > INT_MAX)
  76                         return -EINVAL;
  77                 atomic_set(&ffi->inject_ops, 0);
  78                 ffi->inject_rate = (int)rate;
  79         }
  80
  81         if (type) {
  82                 if (type >= BIT(FAULT_MAX))
  83                         return -EINVAL;
  84                 ffi->inject_type = (unsigned int)type;
  85         }
  86
  87         if (!rate && !type)
  88                 memset(ffi, 0, sizeof(struct f2fs_fault_info));
  89         else
  90                 f2fs_info(sbi,
  91                         "build fault injection attr: rate: %lu, type: 0x%lx",
  92                                                                 rate, type);
  93         return 0;
  94 }
  95 #endif
  96
  97 /* f2fs-wide shrinker description */
  98 static struct shrinker *f2fs_shrinker_info;
  99
 100 static int __init f2fs_init_shrinker(void)
 101 {
 102         f2fs_shrinker_info = shrinker_alloc(0, "f2fs-shrinker");
 103         if (!f2fs_shrinker_info)
 104                 return -ENOMEM;
 105
 106         f2fs_shrinker_info->count_objects = f2fs_shrink_count;
 107         f2fs_shrinker_info->scan_objects = f2fs_shrink_scan;
 108
 109         shrinker_register(f2fs_shrinker_info);
 110
 111         return 0;
 112 }
 113
 114 static void f2fs_exit_shrinker(void)
 115 {
 116         shrinker_free(f2fs_shrinker_info);
 117 }
 118
 119 enum {
 120         Opt_gc_background,
 121         Opt_disable_roll_forward,
 122         Opt_norecovery,
 123         Opt_discard,
 124         Opt_nodiscard,
 125         Opt_noheap,
 126         Opt_heap,
 127         Opt_user_xattr,
 128         Opt_nouser_xattr,
 129         Opt_acl,
 130         Opt_noacl,
 131         Opt_active_logs,
 132         Opt_disable_ext_identify,
 133         Opt_inline_xattr,
 134         Opt_noinline_xattr,
 135         Opt_inline_xattr_size,
 136         Opt_inline_data,
 137         Opt_inline_dentry,
 138         Opt_noinline_dentry,
 139         Opt_flush_merge,
 140         Opt_noflush_merge,
 141         Opt_barrier,
 142         Opt_nobarrier,
 143         Opt_fastboot,
 144         Opt_extent_cache,
 145         Opt_noextent_cache,
 146         Opt_noinline_data,
 147         Opt_data_flush,
 148         Opt_reserve_root,
 149         Opt_resgid,
 150         Opt_resuid,
 151         Opt_mode,
 152         Opt_fault_injection,
 153         Opt_fault_type,
 154         Opt_lazytime,
 155         Opt_nolazytime,
 156         Opt_quota,
 157         Opt_noquota,
 158         Opt_usrquota,
 159         Opt_grpquota,
 160         Opt_prjquota,
 161         Opt_usrjquota,
 162         Opt_grpjquota,
 163         Opt_prjjquota,
 164         Opt_offusrjquota,
 165         Opt_offgrpjquota,
 166         Opt_offprjjquota,
 167         Opt_jqfmt_vfsold,
 168         Opt_jqfmt_vfsv0,
 169         Opt_jqfmt_vfsv1,
 170         Opt_alloc,
 171         Opt_fsync,
 172         Opt_test_dummy_encryption,
 173         Opt_inlinecrypt,
 174         Opt_checkpoint_disable,
 175         Opt_checkpoint_disable_cap,
 176         Opt_checkpoint_disable_cap_perc,
 177         Opt_checkpoint_enable,
 178         Opt_checkpoint_merge,
 179         Opt_nocheckpoint_merge,
 180         Opt_compress_algorithm,
 181         Opt_compress_log_size,
 182         Opt_compress_extension,
 183         Opt_nocompress_extension,
 184         Opt_compress_chksum,
 185         Opt_compress_mode,
 186         Opt_compress_cache,
 187         Opt_atgc,
 188         Opt_gc_merge,
 189         Opt_nogc_merge,
 190         Opt_discard_unit,
 191         Opt_memory_mode,
 192         Opt_age_extent_cache,
 193         Opt_errors,
 194         Opt_err,
 195 };
 196
 197 static match_table_t f2fs_tokens = {
 198         {Opt_gc_background, "background_gc=%s"},
 199         {Opt_disable_roll_forward, "disable_roll_forward"},
 200         {Opt_norecovery, "norecovery"},
 201         {Opt_discard, "discard"},
 202         {Opt_nodiscard, "nodiscard"},
 203         {Opt_noheap, "no_heap"},
 204         {Opt_heap, "heap"},
 205         {Opt_user_xattr, "user_xattr"},
 206         {Opt_nouser_xattr, "nouser_xattr"},
 207         {Opt_acl, "acl"},
 208         {Opt_noacl, "noacl"},
 209         {Opt_active_logs, "active_logs=%u"},
 210         {Opt_disable_ext_identify, "disable_ext_identify"},
 211         {Opt_inline_xattr, "inline_xattr"},
 212         {Opt_noinline_xattr, "noinline_xattr"},
 213         {Opt_inline_xattr_size, "inline_xattr_size=%u"},
 214         {Opt_inline_data, "inline_data"},
 215         {Opt_inline_dentry, "inline_dentry"},
 216         {Opt_noinline_dentry, "noinline_dentry"},
 217         {Opt_flush_merge, "flush_merge"},
 218         {Opt_noflush_merge, "noflush_merge"},
 219         {Opt_barrier, "barrier"},
 220         {Opt_nobarrier, "nobarrier"},
 221         {Opt_fastboot, "fastboot"},
 222         {Opt_extent_cache, "extent_cache"},
 223         {Opt_noextent_cache, "noextent_cache"},
 224         {Opt_noinline_data, "noinline_data"},
 225         {Opt_data_flush, "data_flush"},
 226         {Opt_reserve_root, "reserve_root=%u"},
 227         {Opt_resgid, "resgid=%u"},
 228         {Opt_resuid, "resuid=%u"},
 229         {Opt_mode, "mode=%s"},
 230         {Opt_fault_injection, "fault_injection=%u"},
 231         {Opt_fault_type, "fault_type=%u"},
 232         {Opt_lazytime, "lazytime"},
 233         {Opt_nolazytime, "nolazytime"},
 234         {Opt_quota, "quota"},
 235         {Opt_noquota, "noquota"},
 236         {Opt_usrquota, "usrquota"},
 237         {Opt_grpquota, "grpquota"},
 238         {Opt_prjquota, "prjquota"},
 239         {Opt_usrjquota, "usrjquota=%s"},
 240         {Opt_grpjquota, "grpjquota=%s"},
 241         {Opt_prjjquota, "prjjquota=%s"},
 242         {Opt_offusrjquota, "usrjquota="},
 243         {Opt_offgrpjquota, "grpjquota="},
 244         {Opt_offprjjquota, "prjjquota="},
 245         {Opt_jqfmt_vfsold, "jqfmt=vfsold"},
 246         {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
 247         {Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
 248         {Opt_alloc, "alloc_mode=%s"},
 249         {Opt_fsync, "fsync_mode=%s"},
 250         {Opt_test_dummy_encryption, "test_dummy_encryption=%s"},
 251         {Opt_test_dummy_encryption, "test_dummy_encryption"},
 252         {Opt_inlinecrypt, "inlinecrypt"},
 253         {Opt_checkpoint_disable, "checkpoint=disable"},
 254         {Opt_checkpoint_disable_cap, "checkpoint=disable:%u"},
 255         {Opt_checkpoint_disable_cap_perc, "checkpoint=disable:%u%%"},
 256         {Opt_checkpoint_enable, "checkpoint=enable"},
 257         {Opt_checkpoint_merge, "checkpoint_merge"},
 258         {Opt_nocheckpoint_merge, "nocheckpoint_merge"},
 259         {Opt_compress_algorithm, "compress_algorithm=%s"},
 260         {Opt_compress_log_size, "compress_log_size=%u"},
 261         {Opt_compress_extension, "compress_extension=%s"},
 262         {Opt_nocompress_extension, "nocompress_extension=%s"},
 263         {Opt_compress_chksum, "compress_chksum"},
 264         {Opt_compress_mode, "compress_mode=%s"},
 265         {Opt_compress_cache, "compress_cache"},
 266         {Opt_atgc, "atgc"},
 267         {Opt_gc_merge, "gc_merge"},
 268         {Opt_nogc_merge, "nogc_merge"},
 269         {Opt_discard_unit, "discard_unit=%s"},
 270         {Opt_memory_mode, "memory=%s"},
 271         {Opt_age_extent_cache, "age_extent_cache"},
 272         {Opt_errors, "errors=%s"},
 273         {Opt_err, NULL},
 274 };
 275
 276 void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate,
 277                                                 const char *fmt, ...)
 278 {
 279         struct va_format vaf;
 280         va_list args;
 281         int level;
 282
 283         va_start(args, fmt);
 284
 285         level = printk_get_level(fmt);
 286         vaf.fmt = printk_skip_level(fmt);
 287         vaf.va = &args;
 288         if (limit_rate)
 289                 printk_ratelimited("%c%cF2FS-fs (%s): %pV\n",
 290                         KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
 291         else
 292                 printk("%c%cF2FS-fs (%s): %pV\n",
 293                         KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
 294
 295         va_end(args);
 296 }
 297
 298 #if IS_ENABLED(CONFIG_UNICODE)
 299 static const struct f2fs_sb_encodings {
 300         __u16 magic;
 301         char *name;
 302         unsigned int version;
 303 } f2fs_sb_encoding_map[] = {
 304         {F2FS_ENC_UTF8_12_1, "utf8", UNICODE_AGE(12, 1, 0)},
 305 };
 306
 307 static const struct f2fs_sb_encodings *
 308 f2fs_sb_read_encoding(const struct f2fs_super_block *sb)
 309 {
 310         __u16 magic = le16_to_cpu(sb->s_encoding);
 311         int i;
 312
 313         for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
 314                 if (magic == f2fs_sb_encoding_map[i].magic)
 315                         return &f2fs_sb_encoding_map[i];
 316
 317         return NULL;
 318 }
 319
 320 struct kmem_cache *f2fs_cf_name_slab;
 321 static int __init f2fs_create_casefold_cache(void)
 322 {
 323         f2fs_cf_name_slab = f2fs_kmem_cache_create("f2fs_casefolded_name",
 324                                                         F2FS_NAME_LEN);
 325         return f2fs_cf_name_slab ? 0 : -ENOMEM;
 326 }
 327
 328 static void f2fs_destroy_casefold_cache(void)
 329 {
 330         kmem_cache_destroy(f2fs_cf_name_slab);
 331 }
 332 #else
 333 static int __init f2fs_create_casefold_cache(void) { return 0; }
 334 static void f2fs_destroy_casefold_cache(void) { }
 335 #endif
 336
 337 static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
 338 {
 339         block_t limit = min((sbi->user_block_count >> 3),
 340                         sbi->user_block_count - sbi->reserved_blocks);
 341
 342         /* limit is 12.5% */
 343         if (test_opt(sbi, RESERVE_ROOT) &&
 344                         F2FS_OPTION(sbi).root_reserved_blocks > limit) {
 345                 F2FS_OPTION(sbi).root_reserved_blocks = limit;
 346                 f2fs_info(sbi, "Reduce reserved blocks for root = %u",
 347                           F2FS_OPTION(sbi).root_reserved_blocks);
 348         }
 349         if (!test_opt(sbi, RESERVE_ROOT) &&
 350                 (!uid_eq(F2FS_OPTION(sbi).s_resuid,
 351                                 make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
 352                 !gid_eq(F2FS_OPTION(sbi).s_resgid,
 353                                 make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
 354                 f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
 355                           from_kuid_munged(&init_user_ns,
 356                                            F2FS_OPTION(sbi).s_resuid),
 357                           from_kgid_munged(&init_user_ns,
 358                                            F2FS_OPTION(sbi).s_resgid));
 359 }
 360
 361 static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
 362 {
 363         if (!F2FS_OPTION(sbi).unusable_cap_perc)
 364                 return;
 365
 366         if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
 367                 F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
 368         else
 369                 F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
 370                                         F2FS_OPTION(sbi).unusable_cap_perc;
 371
 372         f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
 373                         F2FS_OPTION(sbi).unusable_cap,
 374                         F2FS_OPTION(sbi).unusable_cap_perc);
 375 }
 376
 377 static void init_once(void *foo)
 378 {
 379         struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
 380
 381         inode_init_once(&fi->vfs_inode);
 382 }
 383
 384 #ifdef CONFIG_QUOTA
 385 static const char * const quotatypes[] = INITQFNAMES;
 386 #define QTYPE2NAME(t) (quotatypes[t])
 387 static int f2fs_set_qf_name(struct super_block *sb, int qtype,
 388                                                         substring_t *args)
 389 {
 390         struct f2fs_sb_info *sbi = F2FS_SB(sb);
 391         char *qname;
 392         int ret = -EINVAL;
 393
 394         if (sb_any_quota_loaded(sb) && !F2FS_OPTION(sbi).s_qf_names[qtype]) {
 395                 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
 396                 return -EINVAL;
 397         }
 398         if (f2fs_sb_has_quota_ino(sbi)) {
 399                 f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
 400                 return 0;
 401         }
 402
 403         qname = match_strdup(args);
 404         if (!qname) {
 405                 f2fs_err(sbi, "Not enough memory for storing quotafile name");
 406                 return -ENOMEM;
 407         }
 408         if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
 409                 if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
 410                         ret = 0;
 411                 else
 412                         f2fs_err(sbi, "%s quota file already specified",
 413                                  QTYPE2NAME(qtype));
 414                 goto errout;
 415         }
 416         if (strchr(qname, '/')) {
 417                 f2fs_err(sbi, "quotafile must be on filesystem root");
 418                 goto errout;
 419         }
 420         F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
 421         set_opt(sbi, QUOTA);
 422         return 0;
 423 errout:
 424         kfree(qname);
 425         return ret;
 426 }
 427
 428 static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
 429 {
 430         struct f2fs_sb_info *sbi = F2FS_SB(sb);
 431
 432         if (sb_any_quota_loaded(sb) && F2FS_OPTION(sbi).s_qf_names[qtype]) {
 433                 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
 434                 return -EINVAL;
 435         }
 436         kfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
 437         F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
 438         return 0;
 439 }
 440
 441 static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
 442 {
 443         /*
 444          * We do the test below only for project quotas. 'usrquota' and
 445          * 'grpquota' mount options are allowed even without quota feature
 446          * to support legacy quotas in quota files.
 447          */
 448         if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi)) {
 449                 f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
 450                 return -1;
 451         }
 452         if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
 453                         F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
 454                         F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) {
 455                 if (test_opt(sbi, USRQUOTA) &&
 456                                 F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
 457                         clear_opt(sbi, USRQUOTA);
 458
 459                 if (test_opt(sbi, GRPQUOTA) &&
 460                                 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
 461                         clear_opt(sbi, GRPQUOTA);
 462
 463                 if (test_opt(sbi, PRJQUOTA) &&
 464                                 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
 465                         clear_opt(sbi, PRJQUOTA);
 466
 467                 if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
 468                                 test_opt(sbi, PRJQUOTA)) {
 469                         f2fs_err(sbi, "old and new quota format mixing");
 470                         return -1;
 471                 }
 472
 473                 if (!F2FS_OPTION(sbi).s_jquota_fmt) {
 474                         f2fs_err(sbi, "journaled quota format not specified");
 475                         return -1;
 476                 }
 477         }
 478
 479         if (f2fs_sb_has_quota_ino(sbi) && F2FS_OPTION(sbi).s_jquota_fmt) {
 480                 f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
 481                 F2FS_OPTION(sbi).s_jquota_fmt = 0;
 482         }
 483         return 0;
 484 }
 485 #endif
 486
 487 static int f2fs_set_test_dummy_encryption(struct super_block *sb,
 488                                           const char *opt,
 489                                           const substring_t *arg,
 490                                           bool is_remount)
 491 {
 492         struct f2fs_sb_info *sbi = F2FS_SB(sb);
 493         struct fs_parameter param = {
 494                 .type = fs_value_is_string,
 495                 .string = arg->from ? arg->from : "",
 496         };
 497         struct fscrypt_dummy_policy *policy =
 498                 &F2FS_OPTION(sbi).dummy_enc_policy;
 499         int err;
 500
 501         if (!IS_ENABLED(CONFIG_FS_ENCRYPTION)) {
 502                 f2fs_warn(sbi, "test_dummy_encryption option not supported");
 503                 return -EINVAL;
 504         }
 505
 506         if (!f2fs_sb_has_encrypt(sbi)) {
 507                 f2fs_err(sbi, "Encrypt feature is off");
 508                 return -EINVAL;
 509         }
 510
 511         /*
 512          * This mount option is just for testing, and it's not worthwhile to
 513          * implement the extra complexity (e.g. RCU protection) that would be
 514          * needed to allow it to be set or changed during remount.  We do allow
 515          * it to be specified during remount, but only if there is no change.
 516          */
 517         if (is_remount && !fscrypt_is_dummy_policy_set(policy)) {
 518                 f2fs_warn(sbi, "Can't set test_dummy_encryption on remount");
 519                 return -EINVAL;
 520         }
 521
 522         err = fscrypt_parse_test_dummy_encryption(&param, policy);
 523         if (err) {
 524                 if (err == -EEXIST)
 525                         f2fs_warn(sbi,
 526                                   "Can't change test_dummy_encryption on remount");
 527                 else if (err == -EINVAL)
 528                         f2fs_warn(sbi, "Value of option \"%s\" is unrecognized",
 529                                   opt);
 530                 else
 531                         f2fs_warn(sbi, "Error processing option \"%s\" [%d]",
 532                                   opt, err);
 533                 return -EINVAL;
 534         }
 535         f2fs_warn(sbi, "Test dummy encryption mode enabled");
 536         return 0;
 537 }
 538
 539 #ifdef CONFIG_F2FS_FS_COMPRESSION
 540 static bool is_compress_extension_exist(struct f2fs_sb_info *sbi,
 541                                         const char *new_ext, bool is_ext)
 542 {
 543         unsigned char (*ext)[F2FS_EXTENSION_LEN];
 544         int ext_cnt;
 545         int i;
 546
 547         if (is_ext) {
 548                 ext = F2FS_OPTION(sbi).extensions;
 549                 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
 550         } else {
 551                 ext = F2FS_OPTION(sbi).noextensions;
 552                 ext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
 553         }
 554
 555         for (i = 0; i < ext_cnt; i++) {
 556                 if (!strcasecmp(new_ext, ext[i]))
 557                         return true;
 558         }
 559
 560         return false;
 561 }
 562
 563 /*
 564  * 1. The same extension name cannot not appear in both compress and non-compress extension
 565  * at the same time.
 566  * 2. If the compress extension specifies all files, the types specified by the non-compress
 567  * extension will be treated as special cases and will not be compressed.
 568  * 3. Don't allow the non-compress extension specifies all files.
 569  */
 570 static int f2fs_test_compress_extension(struct f2fs_sb_info *sbi)
 571 {
 572         unsigned char (*ext)[F2FS_EXTENSION_LEN];
 573         unsigned char (*noext)[F2FS_EXTENSION_LEN];
 574         int ext_cnt, noext_cnt, index = 0, no_index = 0;
 575
 576         ext = F2FS_OPTION(sbi).extensions;
 577         ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
 578         noext = F2FS_OPTION(sbi).noextensions;
 579         noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
 580
 581         if (!noext_cnt)
 582                 return 0;
 583
 584         for (no_index = 0; no_index < noext_cnt; no_index++) {
 585                 if (!strcasecmp("*", noext[no_index])) {
 586                         f2fs_info(sbi, "Don't allow the nocompress extension specifies all files");
 587                         return -EINVAL;
 588                 }
 589                 for (index = 0; index < ext_cnt; index++) {
 590                         if (!strcasecmp(ext[index], noext[no_index])) {
 591                                 f2fs_info(sbi, "Don't allow the same extension %s appear in both compress and nocompress extension",
 592                                                 ext[index]);
 593                                 return -EINVAL;
 594                         }
 595                 }
 596         }
 597         return 0;
 598 }
 599
 600 #ifdef CONFIG_F2FS_FS_LZ4
 601 static int f2fs_set_lz4hc_level(struct f2fs_sb_info *sbi, const char *str)
 602 {
 603 #ifdef CONFIG_F2FS_FS_LZ4HC
 604         unsigned int level;
 605
 606         if (strlen(str) == 3) {
 607                 F2FS_OPTION(sbi).compress_level = 0;
 608                 return 0;
 609         }
 610
 611         str += 3;
 612
 613         if (str[0] != ':') {
 614                 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
 615                 return -EINVAL;
 616         }
 617         if (kstrtouint(str + 1, 10, &level))
 618                 return -EINVAL;
 619
 620         if (!f2fs_is_compress_level_valid(COMPRESS_LZ4, level)) {
 621                 f2fs_info(sbi, "invalid lz4hc compress level: %d", level);
 622                 return -EINVAL;
 623         }
 624
 625         F2FS_OPTION(sbi).compress_level = level;
 626         return 0;
 627 #else
 628         if (strlen(str) == 3) {
 629                 F2FS_OPTION(sbi).compress_level = 0;
 630                 return 0;
 631         }
 632         f2fs_info(sbi, "kernel doesn't support lz4hc compression");
 633         return -EINVAL;
 634 #endif
 635 }
 636 #endif
 637
 638 #ifdef CONFIG_F2FS_FS_ZSTD
 639 static int f2fs_set_zstd_level(struct f2fs_sb_info *sbi, const char *str)
 640 {
 641         int level;
 642         int len = 4;
 643
 644         if (strlen(str) == len) {
 645                 F2FS_OPTION(sbi).compress_level = F2FS_ZSTD_DEFAULT_CLEVEL;
 646                 return 0;
 647         }
 648
 649         str += len;
 650
 651         if (str[0] != ':') {
 652                 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
 653                 return -EINVAL;
 654         }
 655         if (kstrtoint(str + 1, 10, &level))
 656                 return -EINVAL;
 657
 658         /* f2fs does not support negative compress level now */
 659         if (level < 0) {
 660                 f2fs_info(sbi, "do not support negative compress level: %d", level);
 661                 return -ERANGE;
 662         }
 663
 664         if (!f2fs_is_compress_level_valid(COMPRESS_ZSTD, level)) {
 665                 f2fs_info(sbi, "invalid zstd compress level: %d", level);
 666                 return -EINVAL;
 667         }
 668
 669         F2FS_OPTION(sbi).compress_level = level;
 670         return 0;
 671 }
 672 #endif
 673 #endif
 674
 675 static int parse_options(struct super_block *sb, char *options, bool is_remount)
 676 {
 677         struct f2fs_sb_info *sbi = F2FS_SB(sb);
 678         substring_t args[MAX_OPT_ARGS];
 679 #ifdef CONFIG_F2FS_FS_COMPRESSION
 680         unsigned char (*ext)[F2FS_EXTENSION_LEN];
 681         unsigned char (*noext)[F2FS_EXTENSION_LEN];
 682         int ext_cnt, noext_cnt;
 683 #endif
 684         char *p, *name;
 685         int arg = 0;
 686         kuid_t uid;
 687         kgid_t gid;
 688         int ret;
 689
 690         if (!options)
 691                 goto default_check;
 692
 693         while ((p = strsep(&options, ",")) != NULL) {
 694                 int token;
 695
 696                 if (!*p)
 697                         continue;
 698                 /*
 699                  * Initialize args struct so we know whether arg was
 700                  * found; some options take optional arguments.
 701                  */
 702                 args[0].to = args[0].from = NULL;
 703                 token = match_token(p, f2fs_tokens, args);
 704
 705                 switch (token) {
 706                 case Opt_gc_background:
 707                         name = match_strdup(&args[0]);
 708
 709                         if (!name)
 710                                 return -ENOMEM;
 711                         if (!strcmp(name, "on")) {
 712                                 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
 713                         } else if (!strcmp(name, "off")) {
 714                                 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_OFF;
 715                         } else if (!strcmp(name, "sync")) {
 716                                 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_SYNC;
 717                         } else {
 718                                 kfree(name);
 719                                 return -EINVAL;
 720                         }
 721                         kfree(name);
 722                         break;
 723                 case Opt_disable_roll_forward:
 724                         set_opt(sbi, DISABLE_ROLL_FORWARD);
 725                         break;
 726                 case Opt_norecovery:
 727                         /* this option mounts f2fs with ro */
 728                         set_opt(sbi, NORECOVERY);
 729                         if (!f2fs_readonly(sb))
 730                                 return -EINVAL;
 731                         break;
 732                 case Opt_discard:
 733                         if (!f2fs_hw_support_discard(sbi)) {
 734                                 f2fs_warn(sbi, "device does not support discard");
 735                                 break;
 736                         }
 737                         set_opt(sbi, DISCARD);
 738                         break;
 739                 case Opt_nodiscard:
 740                         if (f2fs_hw_should_discard(sbi)) {
 741                                 f2fs_warn(sbi, "discard is required for zoned block devices");
 742                                 return -EINVAL;
 743                         }
 744                         clear_opt(sbi, DISCARD);
 745                         break;
 746                 case Opt_noheap:
 747                 case Opt_heap:
 748                         f2fs_warn(sbi, "heap/no_heap options were deprecated");
 749                         break;
 750 #ifdef CONFIG_F2FS_FS_XATTR
 751                 case Opt_user_xattr:
 752                         set_opt(sbi, XATTR_USER);
 753                         break;
 754                 case Opt_nouser_xattr:
 755                         clear_opt(sbi, XATTR_USER);
 756                         break;
 757                 case Opt_inline_xattr:
 758                         set_opt(sbi, INLINE_XATTR);
 759                         break;
 760                 case Opt_noinline_xattr:
 761                         clear_opt(sbi, INLINE_XATTR);
 762                         break;
 763                 case Opt_inline_xattr_size:
 764                         if (args->from && match_int(args, &arg))
 765                                 return -EINVAL;
 766                         set_opt(sbi, INLINE_XATTR_SIZE);
 767                         F2FS_OPTION(sbi).inline_xattr_size = arg;
 768                         break;
 769 #else
 770                 case Opt_user_xattr:
 771                         f2fs_info(sbi, "user_xattr options not supported");
 772                         break;
 773                 case Opt_nouser_xattr:
 774                         f2fs_info(sbi, "nouser_xattr options not supported");
 775                         break;
 776                 case Opt_inline_xattr:
 777                         f2fs_info(sbi, "inline_xattr options not supported");
 778                         break;
 779                 case Opt_noinline_xattr:
 780                         f2fs_info(sbi, "noinline_xattr options not supported");
 781                         break;
 782 #endif
 783 #ifdef CONFIG_F2FS_FS_POSIX_ACL
 784                 case Opt_acl:
 785                         set_opt(sbi, POSIX_ACL);
 786                         break;
 787                 case Opt_noacl:
 788                         clear_opt(sbi, POSIX_ACL);
 789                         break;
 790 #else
 791                 case Opt_acl:
 792                         f2fs_info(sbi, "acl options not supported");
 793                         break;
 794                 case Opt_noacl:
 795                         f2fs_info(sbi, "noacl options not supported");
 796                         break;
 797 #endif
 798                 case Opt_active_logs:
 799                         if (args->from && match_int(args, &arg))
 800                                 return -EINVAL;
 801                         if (arg != 2 && arg != 4 &&
 802                                 arg != NR_CURSEG_PERSIST_TYPE)
 803                                 return -EINVAL;
 804                         F2FS_OPTION(sbi).active_logs = arg;
 805                         break;
 806                 case Opt_disable_ext_identify:
 807                         set_opt(sbi, DISABLE_EXT_IDENTIFY);
 808                         break;
 809                 case Opt_inline_data:
 810                         set_opt(sbi, INLINE_DATA);
 811                         break;
 812                 case Opt_inline_dentry:
 813                         set_opt(sbi, INLINE_DENTRY);
 814                         break;
 815                 case Opt_noinline_dentry:
 816                         clear_opt(sbi, INLINE_DENTRY);
 817                         break;
 818                 case Opt_flush_merge:
 819                         set_opt(sbi, FLUSH_MERGE);
 820                         break;
 821                 case Opt_noflush_merge:
 822                         clear_opt(sbi, FLUSH_MERGE);
 823                         break;
 824                 case Opt_nobarrier:
 825                         set_opt(sbi, NOBARRIER);
 826                         break;
 827                 case Opt_barrier:
 828                         clear_opt(sbi, NOBARRIER);
 829                         break;
 830                 case Opt_fastboot:
 831                         set_opt(sbi, FASTBOOT);
 832                         break;
 833                 case Opt_extent_cache:
 834                         set_opt(sbi, READ_EXTENT_CACHE);
 835                         break;
 836                 case Opt_noextent_cache:
 837                         clear_opt(sbi, READ_EXTENT_CACHE);
 838                         break;
 839                 case Opt_noinline_data:
 840                         clear_opt(sbi, INLINE_DATA);
 841                         break;
 842                 case Opt_data_flush:
 843                         set_opt(sbi, DATA_FLUSH);
 844                         break;
 845                 case Opt_reserve_root:
 846                         if (args->from && match_int(args, &arg))
 847                                 return -EINVAL;
 848                         if (test_opt(sbi, RESERVE_ROOT)) {
 849                                 f2fs_info(sbi, "Preserve previous reserve_root=%u",
 850                                           F2FS_OPTION(sbi).root_reserved_blocks);
 851                         } else {
 852                                 F2FS_OPTION(sbi).root_reserved_blocks = arg;
 853                                 set_opt(sbi, RESERVE_ROOT);
 854                         }
 855                         break;
 856                 case Opt_resuid:
 857                         if (args->from && match_int(args, &arg))
 858                                 return -EINVAL;
 859                         uid = make_kuid(current_user_ns(), arg);
 860                         if (!uid_valid(uid)) {
 861                                 f2fs_err(sbi, "Invalid uid value %d", arg);
 862                                 return -EINVAL;
 863                         }
 864                         F2FS_OPTION(sbi).s_resuid = uid;
 865                         break;
 866                 case Opt_resgid:
 867                         if (args->from && match_int(args, &arg))
 868                                 return -EINVAL;
 869                         gid = make_kgid(current_user_ns(), arg);
 870                         if (!gid_valid(gid)) {
 871                                 f2fs_err(sbi, "Invalid gid value %d", arg);
 872                                 return -EINVAL;
 873                         }
 874                         F2FS_OPTION(sbi).s_resgid = gid;
 875                         break;
 876                 case Opt_mode:
 877                         name = match_strdup(&args[0]);
 878
 879                         if (!name)
 880                                 return -ENOMEM;
 881                         if (!strcmp(name, "adaptive")) {
 882                                 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
 883                         } else if (!strcmp(name, "lfs")) {
 884                                 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
 885                         } else if (!strcmp(name, "fragment:segment")) {
 886                                 F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_SEG;
 887                         } else if (!strcmp(name, "fragment:block")) {
 888                                 F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_BLK;
 889                         } else {
 890                                 kfree(name);
 891                                 return -EINVAL;
 892                         }
 893                         kfree(name);
 894                         break;
 895 #ifdef CONFIG_F2FS_FAULT_INJECTION
 896                 case Opt_fault_injection:
 897                         if (args->from && match_int(args, &arg))
 898                                 return -EINVAL;
 899                         if (f2fs_build_fault_attr(sbi, arg,
 900                                         F2FS_ALL_FAULT_TYPE))
 901                                 return -EINVAL;
 902                         set_opt(sbi, FAULT_INJECTION);
 903                         break;
 904
 905                 case Opt_fault_type:
 906                         if (args->from && match_int(args, &arg))
 907                                 return -EINVAL;
 908                         if (f2fs_build_fault_attr(sbi, 0, arg))
 909                                 return -EINVAL;
 910                         set_opt(sbi, FAULT_INJECTION);
 911                         break;
 912 #else
 913                 case Opt_fault_injection:
 914                         f2fs_info(sbi, "fault_injection options not supported");
 915                         break;
 916
 917                 case Opt_fault_type:
 918                         f2fs_info(sbi, "fault_type options not supported");
 919                         break;
 920 #endif
 921                 case Opt_lazytime:
 922                         sb->s_flags |= SB_LAZYTIME;
 923                         break;
 924                 case Opt_nolazytime:
 925                         sb->s_flags &= ~SB_LAZYTIME;
 926                         break;
 927 #ifdef CONFIG_QUOTA
 928                 case Opt_quota:
 929                 case Opt_usrquota:
 930                         set_opt(sbi, USRQUOTA);
 931                         break;
 932                 case Opt_grpquota:
 933                         set_opt(sbi, GRPQUOTA);
 934                         break;
 935                 case Opt_prjquota:
 936                         set_opt(sbi, PRJQUOTA);
 937                         break;
 938                 case Opt_usrjquota:
 939                         ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
 940                         if (ret)
 941                                 return ret;
 942                         break;
 943                 case Opt_grpjquota:
 944                         ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
 945                         if (ret)
 946                                 return ret;
 947                         break;
 948                 case Opt_prjjquota:
 949                         ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
 950                         if (ret)
 951                                 return ret;
 952                         break;
 953                 case Opt_offusrjquota:
 954                         ret = f2fs_clear_qf_name(sb, USRQUOTA);
 955                         if (ret)
 956                                 return ret;
 957                         break;
 958                 case Opt_offgrpjquota:
 959                         ret = f2fs_clear_qf_name(sb, GRPQUOTA);
 960                         if (ret)
 961                                 return ret;
 962                         break;
 963                 case Opt_offprjjquota:
 964                         ret = f2fs_clear_qf_name(sb, PRJQUOTA);
 965                         if (ret)
 966                                 return ret;
 967                         break;
 968                 case Opt_jqfmt_vfsold:
 969                         F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
 970                         break;
 971                 case Opt_jqfmt_vfsv0:
 972                         F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
 973                         break;
 974                 case Opt_jqfmt_vfsv1:
 975                         F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
 976                         break;
 977                 case Opt_noquota:
 978                         clear_opt(sbi, QUOTA);
 979                         clear_opt(sbi, USRQUOTA);
 980                         clear_opt(sbi, GRPQUOTA);
 981                         clear_opt(sbi, PRJQUOTA);
 982                         break;
 983 #else
 984                 case Opt_quota:
 985                 case Opt_usrquota:
 986                 case Opt_grpquota:
 987                 case Opt_prjquota:
 988                 case Opt_usrjquota:
 989                 case Opt_grpjquota:
 990                 case Opt_prjjquota:
 991                 case Opt_offusrjquota:
 992                 case Opt_offgrpjquota:
 993                 case Opt_offprjjquota:
 994                 case Opt_jqfmt_vfsold:
 995                 case Opt_jqfmt_vfsv0:
 996                 case Opt_jqfmt_vfsv1:
 997                 case Opt_noquota:
 998                         f2fs_info(sbi, "quota operations not supported");
 999                         break;
1000 #endif
1001                 case Opt_alloc:
1002                         name = match_strdup(&args[0]);
1003                         if (!name)
1004                                 return -ENOMEM;
1005
1006                         if (!strcmp(name, "default")) {
1007                                 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
1008                         } else if (!strcmp(name, "reuse")) {
1009                                 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
1010                         } else {
1011                                 kfree(name);
1012                                 return -EINVAL;
1013                         }
1014                         kfree(name);
1015                         break;
1016                 case Opt_fsync:
1017                         name = match_strdup(&args[0]);
1018                         if (!name)
1019                                 return -ENOMEM;
1020                         if (!strcmp(name, "posix")) {
1021                                 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
1022                         } else if (!strcmp(name, "strict")) {
1023                                 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_STRICT;
1024                         } else if (!strcmp(name, "nobarrier")) {
1025                                 F2FS_OPTION(sbi).fsync_mode =
1026                                                         FSYNC_MODE_NOBARRIER;
1027                         } else {
1028                                 kfree(name);
1029                                 return -EINVAL;
1030                         }
1031                         kfree(name);
1032                         break;
1033                 case Opt_test_dummy_encryption:
1034                         ret = f2fs_set_test_dummy_encryption(sb, p, &args[0],
1035                                                              is_remount);
1036                         if (ret)
1037                                 return ret;
1038                         break;
1039                 case Opt_inlinecrypt:
1040 #ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
1041                         sb->s_flags |= SB_INLINECRYPT;
1042 #else
1043                         f2fs_info(sbi, "inline encryption not supported");
1044 #endif
1045                         break;
1046                 case Opt_checkpoint_disable_cap_perc:
1047                         if (args->from && match_int(args, &arg))
1048                                 return -EINVAL;
1049                         if (arg < 0 || arg > 100)
1050                                 return -EINVAL;
1051                         F2FS_OPTION(sbi).unusable_cap_perc = arg;
1052                         set_opt(sbi, DISABLE_CHECKPOINT);
1053                         break;
1054                 case Opt_checkpoint_disable_cap:
1055                         if (args->from && match_int(args, &arg))
1056                                 return -EINVAL;
1057                         F2FS_OPTION(sbi).unusable_cap = arg;
1058                         set_opt(sbi, DISABLE_CHECKPOINT);
1059                         break;
1060                 case Opt_checkpoint_disable:
1061                         set_opt(sbi, DISABLE_CHECKPOINT);
1062                         break;
1063                 case Opt_checkpoint_enable:
1064                         clear_opt(sbi, DISABLE_CHECKPOINT);
1065                         break;
1066                 case Opt_checkpoint_merge:
1067                         set_opt(sbi, MERGE_CHECKPOINT);
1068                         break;
1069                 case Opt_nocheckpoint_merge:
1070                         clear_opt(sbi, MERGE_CHECKPOINT);
1071                         break;
1072 #ifdef CONFIG_F2FS_FS_COMPRESSION
1073                 case Opt_compress_algorithm:
1074                         if (!f2fs_sb_has_compression(sbi)) {
1075                                 f2fs_info(sbi, "Image doesn't support compression");
1076                                 break;
1077                         }
1078                         name = match_strdup(&args[0]);
1079                         if (!name)
1080                                 return -ENOMEM;
1081                         if (!strcmp(name, "lzo")) {
1082 #ifdef CONFIG_F2FS_FS_LZO
1083                                 F2FS_OPTION(sbi).compress_level = 0;
1084                                 F2FS_OPTION(sbi).compress_algorithm =
1085                                                                 COMPRESS_LZO;
1086 #else
1087                                 f2fs_info(sbi, "kernel doesn't support lzo compression");
1088 #endif
1089                         } else if (!strncmp(name, "lz4", 3)) {
1090 #ifdef CONFIG_F2FS_FS_LZ4
1091                                 ret = f2fs_set_lz4hc_level(sbi, name);
1092                                 if (ret) {
1093                                         kfree(name);
1094                                         return -EINVAL;
1095                                 }
1096                                 F2FS_OPTION(sbi).compress_algorithm =
1097                                                                 COMPRESS_LZ4;
1098 #else
1099                                 f2fs_info(sbi, "kernel doesn't support lz4 compression");
1100 #endif
1101                         } else if (!strncmp(name, "zstd", 4)) {
1102 #ifdef CONFIG_F2FS_FS_ZSTD
1103                                 ret = f2fs_set_zstd_level(sbi, name);
1104                                 if (ret) {
1105                                         kfree(name);
1106                                         return -EINVAL;
1107                                 }
1108                                 F2FS_OPTION(sbi).compress_algorithm =
1109                                                                 COMPRESS_ZSTD;
1110 #else
1111                                 f2fs_info(sbi, "kernel doesn't support zstd compression");
1112 #endif
1113                         } else if (!strcmp(name, "lzo-rle")) {
1114 #ifdef CONFIG_F2FS_FS_LZORLE
1115                                 F2FS_OPTION(sbi).compress_level = 0;
1116                                 F2FS_OPTION(sbi).compress_algorithm =
1117                                                                 COMPRESS_LZORLE;
1118 #else
1119                                 f2fs_info(sbi, "kernel doesn't support lzorle compression");
1120 #endif
1121                         } else {
1122                                 kfree(name);
1123                                 return -EINVAL;
1124                         }
1125                         kfree(name);
1126                         break;
1127                 case Opt_compress_log_size:
1128                         if (!f2fs_sb_has_compression(sbi)) {
1129                                 f2fs_info(sbi, "Image doesn't support compression");
1130                                 break;
1131                         }
1132                         if (args->from && match_int(args, &arg))
1133                                 return -EINVAL;
1134                         if (arg < MIN_COMPRESS_LOG_SIZE ||
1135                                 arg > MAX_COMPRESS_LOG_SIZE) {
1136                                 f2fs_err(sbi,
1137                                         "Compress cluster log size is out of range");
1138                                 return -EINVAL;
1139                         }
1140                         F2FS_OPTION(sbi).compress_log_size = arg;
1141                         break;
1142                 case Opt_compress_extension:
1143                         if (!f2fs_sb_has_compression(sbi)) {
1144                                 f2fs_info(sbi, "Image doesn't support compression");
1145                                 break;
1146                         }
1147                         name = match_strdup(&args[0]);
1148                         if (!name)
1149                                 return -ENOMEM;
1150
1151                         ext = F2FS_OPTION(sbi).extensions;
1152                         ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
1153
1154                         if (strlen(name) >= F2FS_EXTENSION_LEN ||
1155                                 ext_cnt >= COMPRESS_EXT_NUM) {
1156                                 f2fs_err(sbi,
1157                                         "invalid extension length/number");
1158                                 kfree(name);
1159                                 return -EINVAL;
1160                         }
1161
1162                         if (is_compress_extension_exist(sbi, name, true)) {
1163                                 kfree(name);
1164                                 break;
1165                         }
1166
1167                         strcpy(ext[ext_cnt], name);
1168                         F2FS_OPTION(sbi).compress_ext_cnt++;
1169                         kfree(name);
1170                         break;
1171                 case Opt_nocompress_extension:
1172                         if (!f2fs_sb_has_compression(sbi)) {
1173                                 f2fs_info(sbi, "Image doesn't support compression");
1174                                 break;
1175                         }
1176                         name = match_strdup(&args[0]);
1177                         if (!name)
1178                                 return -ENOMEM;
1179
1180                         noext = F2FS_OPTION(sbi).noextensions;
1181                         noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
1182
1183                         if (strlen(name) >= F2FS_EXTENSION_LEN ||
1184                                 noext_cnt >= COMPRESS_EXT_NUM) {
1185                                 f2fs_err(sbi,
1186                                         "invalid extension length/number");
1187                                 kfree(name);
1188                                 return -EINVAL;
1189                         }
1190
1191                         if (is_compress_extension_exist(sbi, name, false)) {
1192                                 kfree(name);
1193                                 break;
1194                         }
1195
1196                         strcpy(noext[noext_cnt], name);
1197                         F2FS_OPTION(sbi).nocompress_ext_cnt++;
1198                         kfree(name);
1199                         break;
1200                 case Opt_compress_chksum:
1201                         if (!f2fs_sb_has_compression(sbi)) {
1202                                 f2fs_info(sbi, "Image doesn't support compression");
1203                                 break;
1204                         }
1205                         F2FS_OPTION(sbi).compress_chksum = true;
1206                         break;
1207                 case Opt_compress_mode:
1208                         if (!f2fs_sb_has_compression(sbi)) {
1209                                 f2fs_info(sbi, "Image doesn't support compression");
1210                                 break;
1211                         }
1212                         name = match_strdup(&args[0]);
1213                         if (!name)
1214                                 return -ENOMEM;
1215                         if (!strcmp(name, "fs")) {
1216                                 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
1217                         } else if (!strcmp(name, "user")) {
1218                                 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_USER;
1219                         } else {
1220                                 kfree(name);
1221                                 return -EINVAL;
1222                         }
1223                         kfree(name);
1224                         break;
1225                 case Opt_compress_cache:
1226                         if (!f2fs_sb_has_compression(sbi)) {
1227                                 f2fs_info(sbi, "Image doesn't support compression");
1228                                 break;
1229                         }
1230                         set_opt(sbi, COMPRESS_CACHE);
1231                         break;
1232 #else
1233                 case Opt_compress_algorithm:
1234                 case Opt_compress_log_size:
1235                 case Opt_compress_extension:
1236                 case Opt_nocompress_extension:
1237                 case Opt_compress_chksum:
1238                 case Opt_compress_mode:
1239                 case Opt_compress_cache:
1240                         f2fs_info(sbi, "compression options not supported");
1241                         break;
1242 #endif
1243                 case Opt_atgc:
1244                         set_opt(sbi, ATGC);
1245                         break;
1246                 case Opt_gc_merge:
1247                         set_opt(sbi, GC_MERGE);
1248                         break;
1249                 case Opt_nogc_merge:
1250                         clear_opt(sbi, GC_MERGE);
1251                         break;
1252                 case Opt_discard_unit:
1253                         name = match_strdup(&args[0]);
1254                         if (!name)
1255                                 return -ENOMEM;
1256                         if (!strcmp(name, "block")) {
1257                                 F2FS_OPTION(sbi).discard_unit =
1258                                                 DISCARD_UNIT_BLOCK;
1259                         } else if (!strcmp(name, "segment")) {
1260                                 F2FS_OPTION(sbi).discard_unit =
1261                                                 DISCARD_UNIT_SEGMENT;
1262                         } else if (!strcmp(name, "section")) {
1263                                 F2FS_OPTION(sbi).discard_unit =
1264                                                 DISCARD_UNIT_SECTION;
1265                         } else {
1266                                 kfree(name);
1267                                 return -EINVAL;
1268                         }
1269                         kfree(name);
1270                         break;
1271                 case Opt_memory_mode:
1272                         name = match_strdup(&args[0]);
1273                         if (!name)
1274                                 return -ENOMEM;
1275                         if (!strcmp(name, "normal")) {
1276                                 F2FS_OPTION(sbi).memory_mode =
1277                                                 MEMORY_MODE_NORMAL;
1278                         } else if (!strcmp(name, "low")) {
1279                                 F2FS_OPTION(sbi).memory_mode =
1280                                                 MEMORY_MODE_LOW;
1281                         } else {
1282                                 kfree(name);
1283                                 return -EINVAL;
1284                         }
1285                         kfree(name);
1286                         break;
1287                 case Opt_age_extent_cache:
1288                         set_opt(sbi, AGE_EXTENT_CACHE);
1289                         break;
1290                 case Opt_errors:
1291                         name = match_strdup(&args[0]);
1292                         if (!name)
1293                                 return -ENOMEM;
1294                         if (!strcmp(name, "remount-ro")) {
1295                                 F2FS_OPTION(sbi).errors =
1296                                                 MOUNT_ERRORS_READONLY;
1297                         } else if (!strcmp(name, "continue")) {
1298                                 F2FS_OPTION(sbi).errors =
1299                                                 MOUNT_ERRORS_CONTINUE;
1300                         } else if (!strcmp(name, "panic")) {
1301                                 F2FS_OPTION(sbi).errors =
1302                                                 MOUNT_ERRORS_PANIC;
1303                         } else {
1304                                 kfree(name);
1305                                 return -EINVAL;
1306                         }
1307                         kfree(name);
1308                         break;
1309                 default:
1310                         f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
1311                                  p);
1312                         return -EINVAL;
1313                 }
1314         }
1315 default_check:
1316 #ifdef CONFIG_QUOTA
1317         if (f2fs_check_quota_options(sbi))
1318                 return -EINVAL;
1319 #else
1320         if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sbi->sb)) {
1321                 f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1322                 return -EINVAL;
1323         }
1324         if (f2fs_sb_has_project_quota(sbi) && !f2fs_readonly(sbi->sb)) {
1325                 f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1326                 return -EINVAL;
1327         }
1328 #endif
1329 #if !IS_ENABLED(CONFIG_UNICODE)
1330         if (f2fs_sb_has_casefold(sbi)) {
1331                 f2fs_err(sbi,
1332                         "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
1333                 return -EINVAL;
1334         }
1335 #endif
1336         /*
1337          * The BLKZONED feature indicates that the drive was formatted with
1338          * zone alignment optimization. This is optional for host-aware
1339          * devices, but mandatory for host-managed zoned block devices.
1340          */
1341         if (f2fs_sb_has_blkzoned(sbi)) {
1342 #ifdef CONFIG_BLK_DEV_ZONED
1343                 if (F2FS_OPTION(sbi).discard_unit !=
1344                                                 DISCARD_UNIT_SECTION) {
1345                         f2fs_info(sbi, "Zoned block device doesn't need small discard, set discard_unit=section by default");
1346                         F2FS_OPTION(sbi).discard_unit =
1347                                         DISCARD_UNIT_SECTION;
1348                 }
1349
1350                 if (F2FS_OPTION(sbi).fs_mode != FS_MODE_LFS) {
1351                         f2fs_info(sbi, "Only lfs mode is allowed with zoned block device feature");
1352                         return -EINVAL;
1353                 }
1354 #else
1355                 f2fs_err(sbi, "Zoned block device support is not enabled");
1356                 return -EINVAL;
1357 #endif
1358         }
1359
1360 #ifdef CONFIG_F2FS_FS_COMPRESSION
1361         if (f2fs_test_compress_extension(sbi)) {
1362                 f2fs_err(sbi, "invalid compress or nocompress extension");
1363                 return -EINVAL;
1364         }
1365 #endif
1366
1367         if (test_opt(sbi, INLINE_XATTR_SIZE)) {
1368                 int min_size, max_size;
1369
1370                 if (!f2fs_sb_has_extra_attr(sbi) ||
1371                         !f2fs_sb_has_flexible_inline_xattr(sbi)) {
1372                         f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
1373                         return -EINVAL;
1374                 }
1375                 if (!test_opt(sbi, INLINE_XATTR)) {
1376                         f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
1377                         return -EINVAL;
1378                 }
1379
1380                 min_size = MIN_INLINE_XATTR_SIZE;
1381                 max_size = MAX_INLINE_XATTR_SIZE;
1382
1383                 if (F2FS_OPTION(sbi).inline_xattr_size < min_size ||
1384                                 F2FS_OPTION(sbi).inline_xattr_size > max_size) {
1385                         f2fs_err(sbi, "inline xattr size is out of range: %d ~ %d",
1386                                  min_size, max_size);
1387                         return -EINVAL;
1388                 }
1389         }
1390
1391         if (test_opt(sbi, ATGC) && f2fs_lfs_mode(sbi)) {
1392                 f2fs_err(sbi, "LFS is not compatible with ATGC");
1393                 return -EINVAL;
1394         }
1395
1396         if (f2fs_is_readonly(sbi) && test_opt(sbi, FLUSH_MERGE)) {
1397                 f2fs_err(sbi, "FLUSH_MERGE not compatible with readonly mode");
1398                 return -EINVAL;
1399         }
1400
1401         if (f2fs_sb_has_readonly(sbi) && !f2fs_readonly(sbi->sb)) {
1402                 f2fs_err(sbi, "Allow to mount readonly mode only");
1403                 return -EROFS;
1404         }
1405         return 0;
1406 }
1407
1408 static struct inode *f2fs_alloc_inode(struct super_block *sb)
1409 {
1410         struct f2fs_inode_info *fi;
1411
1412         if (time_to_inject(F2FS_SB(sb), FAULT_SLAB_ALLOC))
1413                 return NULL;
1414
1415         fi = alloc_inode_sb(sb, f2fs_inode_cachep, GFP_F2FS_ZERO);
1416         if (!fi)
1417                 return NULL;
1418
1419         init_once((void *) fi);
1420
1421         /* Initialize f2fs-specific inode info */
1422         atomic_set(&fi->dirty_pages, 0);
1423         atomic_set(&fi->i_compr_blocks, 0);
1424         init_f2fs_rwsem(&fi->i_sem);
1425         spin_lock_init(&fi->i_size_lock);
1426         INIT_LIST_HEAD(&fi->dirty_list);
1427         INIT_LIST_HEAD(&fi->gdirty_list);
1428         init_f2fs_rwsem(&fi->i_gc_rwsem[READ]);
1429         init_f2fs_rwsem(&fi->i_gc_rwsem[WRITE]);
1430         init_f2fs_rwsem(&fi->i_xattr_sem);
1431
1432         /* Will be used by directory only */
1433         fi->i_dir_level = F2FS_SB(sb)->dir_level;
1434
1435         return &fi->vfs_inode;
1436 }
1437
1438 static int f2fs_drop_inode(struct inode *inode)
1439 {
1440         struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1441         int ret;
1442
1443         /*
1444          * during filesystem shutdown, if checkpoint is disabled,
1445          * drop useless meta/node dirty pages.
1446          */
1447         if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1448                 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1449                         inode->i_ino == F2FS_META_INO(sbi)) {
1450                         trace_f2fs_drop_inode(inode, 1);
1451                         return 1;
1452                 }
1453         }
1454
1455         /*
1456          * This is to avoid a deadlock condition like below.
1457          * writeback_single_inode(inode)
1458          *  - f2fs_write_data_page
1459          *    - f2fs_gc -> iput -> evict
1460          *       - inode_wait_for_writeback(inode)
1461          */
1462         if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
1463                 if (!inode->i_nlink && !is_bad_inode(inode)) {
1464                         /* to avoid evict_inode call simultaneously */
1465                         atomic_inc(&inode->i_count);
1466                         spin_unlock(&inode->i_lock);
1467
1468                         /* should remain fi->extent_tree for writepage */
1469                         f2fs_destroy_extent_node(inode);
1470
1471                         sb_start_intwrite(inode->i_sb);
1472                         f2fs_i_size_write(inode, 0);
1473
1474                         f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1475                                         inode, NULL, 0, DATA);
1476                         truncate_inode_pages_final(inode->i_mapping);
1477
1478                         if (F2FS_HAS_BLOCKS(inode))
1479                                 f2fs_truncate(inode);
1480
1481                         sb_end_intwrite(inode->i_sb);
1482
1483                         spin_lock(&inode->i_lock);
1484                         atomic_dec(&inode->i_count);
1485                 }
1486                 trace_f2fs_drop_inode(inode, 0);
1487                 return 0;
1488         }
1489         ret = generic_drop_inode(inode);
1490         if (!ret)
1491                 ret = fscrypt_drop_inode(inode);
1492         trace_f2fs_drop_inode(inode, ret);
1493         return ret;
1494 }
1495
1496 int f2fs_inode_dirtied(struct inode *inode, bool sync)
1497 {
1498         struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1499         int ret = 0;
1500
1501         spin_lock(&sbi->inode_lock[DIRTY_META]);
1502         if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1503                 ret = 1;
1504         } else {
1505                 set_inode_flag(inode, FI_DIRTY_INODE);
1506                 stat_inc_dirty_inode(sbi, DIRTY_META);
1507         }
1508         if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1509                 list_add_tail(&F2FS_I(inode)->gdirty_list,
1510                                 &sbi->inode_list[DIRTY_META]);
1511                 inc_page_count(sbi, F2FS_DIRTY_IMETA);
1512         }
1513         spin_unlock(&sbi->inode_lock[DIRTY_META]);
1514         return ret;
1515 }
1516
1517 void f2fs_inode_synced(struct inode *inode)
1518 {
1519         struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1520
1521         spin_lock(&sbi->inode_lock[DIRTY_META]);
1522         if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1523                 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1524                 return;
1525         }
1526         if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1527                 list_del_init(&F2FS_I(inode)->gdirty_list);
1528                 dec_page_count(sbi, F2FS_DIRTY_IMETA);
1529         }
1530         clear_inode_flag(inode, FI_DIRTY_INODE);
1531         clear_inode_flag(inode, FI_AUTO_RECOVER);
1532         stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1533         spin_unlock(&sbi->inode_lock[DIRTY_META]);
1534 }
1535
1536 /*
1537  * f2fs_dirty_inode() is called from __mark_inode_dirty()
1538  *
1539  * We should call set_dirty_inode to write the dirty inode through write_inode.
1540  */
1541 static void f2fs_dirty_inode(struct inode *inode, int flags)
1542 {
1543         struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1544
1545         if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1546                         inode->i_ino == F2FS_META_INO(sbi))
1547                 return;
1548
1549         if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1550                 clear_inode_flag(inode, FI_AUTO_RECOVER);
1551
1552         f2fs_inode_dirtied(inode, false);
1553 }
1554
1555 static void f2fs_free_inode(struct inode *inode)
1556 {
1557         fscrypt_free_inode(inode);
1558         kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1559 }
1560
1561 static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1562 {
1563         percpu_counter_destroy(&sbi->total_valid_inode_count);
1564         percpu_counter_destroy(&sbi->rf_node_block_count);
1565         percpu_counter_destroy(&sbi->alloc_valid_block_count);
1566 }
1567
1568 static void destroy_device_list(struct f2fs_sb_info *sbi)
1569 {
1570         int i;
1571
1572         for (i = 0; i < sbi->s_ndevs; i++) {
1573                 if (i > 0)
1574                         bdev_fput(FDEV(i).bdev_file);
1575 #ifdef CONFIG_BLK_DEV_ZONED
1576                 kvfree(FDEV(i).blkz_seq);
1577 #endif
1578         }
1579         kvfree(sbi->devs);
1580 }
1581
1582 static void f2fs_put_super(struct super_block *sb)
1583 {
1584         struct f2fs_sb_info *sbi = F2FS_SB(sb);
1585         int i;
1586         int err = 0;
1587         bool done;
1588
1589         /* unregister procfs/sysfs entries in advance to avoid race case */
1590         f2fs_unregister_sysfs(sbi);
1591
1592         f2fs_quota_off_umount(sb);
1593
1594         /* prevent remaining shrinker jobs */
1595         mutex_lock(&sbi->umount_mutex);
1596
1597         /*
1598          * flush all issued checkpoints and stop checkpoint issue thread.
1599          * after then, all checkpoints should be done by each process context.
1600          */
1601         f2fs_stop_ckpt_thread(sbi);
1602
1603         /*
1604          * We don't need to do checkpoint when superblock is clean.
1605          * But, the previous checkpoint was not done by umount, it needs to do
1606          * clean checkpoint again.
1607          */
1608         if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1609                         !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1610                 struct cp_control cpc = {
1611                         .reason = CP_UMOUNT,
1612                 };
1613                 stat_inc_cp_call_count(sbi, TOTAL_CALL);
1614                 err = f2fs_write_checkpoint(sbi, &cpc);
1615         }
1616
1617         /* be sure to wait for any on-going discard commands */
1618         done = f2fs_issue_discard_timeout(sbi);
1619         if (f2fs_realtime_discard_enable(sbi) && !sbi->discard_blks && done) {
1620                 struct cp_control cpc = {
1621                         .reason = CP_UMOUNT | CP_TRIMMED,
1622                 };
1623                 stat_inc_cp_call_count(sbi, TOTAL_CALL);
1624                 err = f2fs_write_checkpoint(sbi, &cpc);
1625         }
1626
1627         /*
1628          * normally superblock is clean, so we need to release this.
1629          * In addition, EIO will skip do checkpoint, we need this as well.
1630          */
1631         f2fs_release_ino_entry(sbi, true);
1632
1633         f2fs_leave_shrinker(sbi);
1634         mutex_unlock(&sbi->umount_mutex);
1635
1636         /* our cp_error case, we can wait for any writeback page */
1637         f2fs_flush_merged_writes(sbi);
1638
1639         f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1640
1641         if (err || f2fs_cp_error(sbi)) {
1642                 truncate_inode_pages_final(NODE_MAPPING(sbi));
1643                 truncate_inode_pages_final(META_MAPPING(sbi));
1644         }
1645
1646         for (i = 0; i < NR_COUNT_TYPE; i++) {
1647                 if (!get_pages(sbi, i))
1648                         continue;
1649                 f2fs_err(sbi, "detect filesystem reference count leak during "
1650                         "umount, type: %d, count: %lld", i, get_pages(sbi, i));
1651                 f2fs_bug_on(sbi, 1);
1652         }
1653
1654         f2fs_bug_on(sbi, sbi->fsync_node_num);
1655
1656         f2fs_destroy_compress_inode(sbi);
1657
1658         iput(sbi->node_inode);
1659         sbi->node_inode = NULL;
1660
1661         iput(sbi->meta_inode);
1662         sbi->meta_inode = NULL;
1663
1664         /*
1665          * iput() can update stat information, if f2fs_write_checkpoint()
1666          * above failed with error.
1667          */
1668         f2fs_destroy_stats(sbi);
1669
1670         /* destroy f2fs internal modules */
1671         f2fs_destroy_node_manager(sbi);
1672         f2fs_destroy_segment_manager(sbi);
1673
1674         /* flush s_error_work before sbi destroy */
1675         flush_work(&sbi->s_error_work);
1676
1677         f2fs_destroy_post_read_wq(sbi);
1678
1679         kvfree(sbi->ckpt);
1680
1681         if (sbi->s_chksum_driver)
1682                 crypto_free_shash(sbi->s_chksum_driver);
1683         kfree(sbi->raw_super);
1684
1685         f2fs_destroy_page_array_cache(sbi);
1686         f2fs_destroy_xattr_caches(sbi);
1687 #ifdef CONFIG_QUOTA
1688         for (i = 0; i < MAXQUOTAS; i++)
1689                 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1690 #endif
1691         fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
1692         destroy_percpu_info(sbi);
1693         f2fs_destroy_iostat(sbi);
1694         for (i = 0; i < NR_PAGE_TYPE; i++)
1695                 kvfree(sbi->write_io[i]);
1696 #if IS_ENABLED(CONFIG_UNICODE)
1697         utf8_unload(sb->s_encoding);
1698 #endif
1699 }
1700
1701 int f2fs_sync_fs(struct super_block *sb, int sync)
1702 {
1703         struct f2fs_sb_info *sbi = F2FS_SB(sb);
1704         int err = 0;
1705
1706         if (unlikely(f2fs_cp_error(sbi)))
1707                 return 0;
1708         if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1709                 return 0;
1710
1711         trace_f2fs_sync_fs(sb, sync);
1712
1713         if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1714                 return -EAGAIN;
1715
1716         if (sync) {
1717                 stat_inc_cp_call_count(sbi, TOTAL_CALL);
1718                 err = f2fs_issue_checkpoint(sbi);
1719         }
1720
1721         return err;
1722 }
1723
1724 static int f2fs_freeze(struct super_block *sb)
1725 {
1726         if (f2fs_readonly(sb))
1727                 return 0;
1728
1729         /* IO error happened before */
1730         if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1731                 return -EIO;
1732
1733         /* must be clean, since sync_filesystem() was already called */
1734         if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1735                 return -EINVAL;
1736
1737         /* Let's flush checkpoints and stop the thread. */
1738         f2fs_flush_ckpt_thread(F2FS_SB(sb));
1739
1740         /* to avoid deadlock on f2fs_evict_inode->SB_FREEZE_FS */
1741         set_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1742         return 0;
1743 }
1744
1745 static int f2fs_unfreeze(struct super_block *sb)
1746 {
1747         clear_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1748         return 0;
1749 }
1750
1751 #ifdef CONFIG_QUOTA
1752 static int f2fs_statfs_project(struct super_block *sb,
1753                                 kprojid_t projid, struct kstatfs *buf)
1754 {
1755         struct kqid qid;
1756         struct dquot *dquot;
1757         u64 limit;
1758         u64 curblock;
1759
1760         qid = make_kqid_projid(projid);
1761         dquot = dqget(sb, qid);
1762         if (IS_ERR(dquot))
1763                 return PTR_ERR(dquot);
1764         spin_lock(&dquot->dq_dqb_lock);
1765
1766         limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1767                                         dquot->dq_dqb.dqb_bhardlimit);
1768         if (limit)
1769                 limit >>= sb->s_blocksize_bits;
1770
1771         if (limit && buf->f_blocks > limit) {
1772                 curblock = (dquot->dq_dqb.dqb_curspace +
1773                             dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
1774                 buf->f_blocks = limit;
1775                 buf->f_bfree = buf->f_bavail =
1776                         (buf->f_blocks > curblock) ?
1777                          (buf->f_blocks - curblock) : 0;
1778         }
1779
1780         limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1781                                         dquot->dq_dqb.dqb_ihardlimit);
1782
1783         if (limit && buf->f_files > limit) {
1784                 buf->f_files = limit;
1785                 buf->f_ffree =
1786                         (buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1787                          (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1788         }
1789
1790         spin_unlock(&dquot->dq_dqb_lock);
1791         dqput(dquot);
1792         return 0;
1793 }
1794 #endif
1795
1796 static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1797 {
1798         struct super_block *sb = dentry->d_sb;
1799         struct f2fs_sb_info *sbi = F2FS_SB(sb);
1800         u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
1801         block_t total_count, user_block_count, start_count;
1802         u64 avail_node_count;
1803         unsigned int total_valid_node_count;
1804
1805         total_count = le64_to_cpu(sbi->raw_super->block_count);
1806         start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
1807         buf->f_type = F2FS_SUPER_MAGIC;
1808         buf->f_bsize = sbi->blocksize;
1809
1810         buf->f_blocks = total_count - start_count;
1811
1812         spin_lock(&sbi->stat_lock);
1813
1814         user_block_count = sbi->user_block_count;
1815         total_valid_node_count = valid_node_count(sbi);
1816         avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
1817         buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1818                                                 sbi->current_reserved_blocks;
1819
1820         if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1821                 buf->f_bfree = 0;
1822         else
1823                 buf->f_bfree -= sbi->unusable_block_count;
1824         spin_unlock(&sbi->stat_lock);
1825
1826         if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1827                 buf->f_bavail = buf->f_bfree -
1828                                 F2FS_OPTION(sbi).root_reserved_blocks;
1829         else
1830                 buf->f_bavail = 0;
1831
1832         if (avail_node_count > user_block_count) {
1833                 buf->f_files = user_block_count;
1834                 buf->f_ffree = buf->f_bavail;
1835         } else {
1836                 buf->f_files = avail_node_count;
1837                 buf->f_ffree = min(avail_node_count - total_valid_node_count,
1838                                         buf->f_bavail);
1839         }
1840
1841         buf->f_namelen = F2FS_NAME_LEN;
1842         buf->f_fsid    = u64_to_fsid(id);
1843
1844 #ifdef CONFIG_QUOTA
1845         if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1846                         sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1847                 f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1848         }
1849 #endif
1850         return 0;
1851 }
1852
1853 static inline void f2fs_show_quota_options(struct seq_file *seq,
1854                                            struct super_block *sb)
1855 {
1856 #ifdef CONFIG_QUOTA
1857         struct f2fs_sb_info *sbi = F2FS_SB(sb);
1858
1859         if (F2FS_OPTION(sbi).s_jquota_fmt) {
1860                 char *fmtname = "";
1861
1862                 switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1863                 case QFMT_VFS_OLD:
1864                         fmtname = "vfsold";
1865                         break;
1866                 case QFMT_VFS_V0:
1867                         fmtname = "vfsv0";
1868                         break;
1869                 case QFMT_VFS_V1:
1870                         fmtname = "vfsv1";
1871                         break;
1872                 }
1873                 seq_printf(seq, ",jqfmt=%s", fmtname);
1874         }
1875
1876         if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1877                 seq_show_option(seq, "usrjquota",
1878                         F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1879
1880         if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1881                 seq_show_option(seq, "grpjquota",
1882                         F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1883
1884         if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1885                 seq_show_option(seq, "prjjquota",
1886                         F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1887 #endif
1888 }
1889
1890 #ifdef CONFIG_F2FS_FS_COMPRESSION
1891 static inline void f2fs_show_compress_options(struct seq_file *seq,
1892                                                         struct super_block *sb)
1893 {
1894         struct f2fs_sb_info *sbi = F2FS_SB(sb);
1895         char *algtype = "";
1896         int i;
1897
1898         if (!f2fs_sb_has_compression(sbi))
1899                 return;
1900
1901         switch (F2FS_OPTION(sbi).compress_algorithm) {
1902         case COMPRESS_LZO:
1903                 algtype = "lzo";
1904                 break;
1905         case COMPRESS_LZ4:
1906                 algtype = "lz4";
1907                 break;
1908         case COMPRESS_ZSTD:
1909                 algtype = "zstd";
1910                 break;
1911         case COMPRESS_LZORLE:
1912                 algtype = "lzo-rle";
1913                 break;
1914         }
1915         seq_printf(seq, ",compress_algorithm=%s", algtype);
1916
1917         if (F2FS_OPTION(sbi).compress_level)
1918                 seq_printf(seq, ":%d", F2FS_OPTION(sbi).compress_level);
1919
1920         seq_printf(seq, ",compress_log_size=%u",
1921                         F2FS_OPTION(sbi).compress_log_size);
1922
1923         for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1924                 seq_printf(seq, ",compress_extension=%s",
1925                         F2FS_OPTION(sbi).extensions[i]);
1926         }
1927
1928         for (i = 0; i < F2FS_OPTION(sbi).nocompress_ext_cnt; i++) {
1929                 seq_printf(seq, ",nocompress_extension=%s",
1930                         F2FS_OPTION(sbi).noextensions[i]);
1931         }
1932
1933         if (F2FS_OPTION(sbi).compress_chksum)
1934                 seq_puts(seq, ",compress_chksum");
1935
1936         if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_FS)
1937                 seq_printf(seq, ",compress_mode=%s", "fs");
1938         else if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_USER)
1939                 seq_printf(seq, ",compress_mode=%s", "user");
1940
1941         if (test_opt(sbi, COMPRESS_CACHE))
1942                 seq_puts(seq, ",compress_cache");
1943 }
1944 #endif
1945
1946 static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1947 {
1948         struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1949
1950         if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1951                 seq_printf(seq, ",background_gc=%s", "sync");
1952         else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1953                 seq_printf(seq, ",background_gc=%s", "on");
1954         else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1955                 seq_printf(seq, ",background_gc=%s", "off");
1956
1957         if (test_opt(sbi, GC_MERGE))
1958                 seq_puts(seq, ",gc_merge");
1959         else
1960                 seq_puts(seq, ",nogc_merge");
1961
1962         if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1963                 seq_puts(seq, ",disable_roll_forward");
1964         if (test_opt(sbi, NORECOVERY))
1965                 seq_puts(seq, ",norecovery");
1966         if (test_opt(sbi, DISCARD)) {
1967                 seq_puts(seq, ",discard");
1968                 if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK)
1969                         seq_printf(seq, ",discard_unit=%s", "block");
1970                 else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SEGMENT)
1971                         seq_printf(seq, ",discard_unit=%s", "segment");
1972                 else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SECTION)
1973                         seq_printf(seq, ",discard_unit=%s", "section");
1974         } else {
1975                 seq_puts(seq, ",nodiscard");
1976         }
1977 #ifdef CONFIG_F2FS_FS_XATTR
1978         if (test_opt(sbi, XATTR_USER))
1979                 seq_puts(seq, ",user_xattr");
1980         else
1981                 seq_puts(seq, ",nouser_xattr");
1982         if (test_opt(sbi, INLINE_XATTR))
1983                 seq_puts(seq, ",inline_xattr");
1984         else
1985                 seq_puts(seq, ",noinline_xattr");
1986         if (test_opt(sbi, INLINE_XATTR_SIZE))
1987                 seq_printf(seq, ",inline_xattr_size=%u",
1988                                         F2FS_OPTION(sbi).inline_xattr_size);
1989 #endif
1990 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1991         if (test_opt(sbi, POSIX_ACL))
1992                 seq_puts(seq, ",acl");
1993         else
1994                 seq_puts(seq, ",noacl");
1995 #endif
1996         if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
1997                 seq_puts(seq, ",disable_ext_identify");
1998         if (test_opt(sbi, INLINE_DATA))
1999                 seq_puts(seq, ",inline_data");
2000         else
2001                 seq_puts(seq, ",noinline_data");
2002         if (test_opt(sbi, INLINE_DENTRY))
2003                 seq_puts(seq, ",inline_dentry");
2004         else
2005                 seq_puts(seq, ",noinline_dentry");
2006         if (test_opt(sbi, FLUSH_MERGE))
2007                 seq_puts(seq, ",flush_merge");
2008         else
2009                 seq_puts(seq, ",noflush_merge");
2010         if (test_opt(sbi, NOBARRIER))
2011                 seq_puts(seq, ",nobarrier");
2012         else
2013                 seq_puts(seq, ",barrier");
2014         if (test_opt(sbi, FASTBOOT))
2015                 seq_puts(seq, ",fastboot");
2016         if (test_opt(sbi, READ_EXTENT_CACHE))
2017                 seq_puts(seq, ",extent_cache");
2018         else
2019                 seq_puts(seq, ",noextent_cache");
2020         if (test_opt(sbi, AGE_EXTENT_CACHE))
2021                 seq_puts(seq, ",age_extent_cache");
2022         if (test_opt(sbi, DATA_FLUSH))
2023                 seq_puts(seq, ",data_flush");
2024
2025         seq_puts(seq, ",mode=");
2026         if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
2027                 seq_puts(seq, "adaptive");
2028         else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
2029                 seq_puts(seq, "lfs");
2030         else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG)
2031                 seq_puts(seq, "fragment:segment");
2032         else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK)
2033                 seq_puts(seq, "fragment:block");
2034         seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
2035         if (test_opt(sbi, RESERVE_ROOT))
2036                 seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
2037                                 F2FS_OPTION(sbi).root_reserved_blocks,
2038                                 from_kuid_munged(&init_user_ns,
2039                                         F2FS_OPTION(sbi).s_resuid),
2040                                 from_kgid_munged(&init_user_ns,
2041                                         F2FS_OPTION(sbi).s_resgid));
2042 #ifdef CONFIG_F2FS_FAULT_INJECTION
2043         if (test_opt(sbi, FAULT_INJECTION)) {
2044                 seq_printf(seq, ",fault_injection=%u",
2045                                 F2FS_OPTION(sbi).fault_info.inject_rate);
2046                 seq_printf(seq, ",fault_type=%u",
2047                                 F2FS_OPTION(sbi).fault_info.inject_type);
2048         }
2049 #endif
2050 #ifdef CONFIG_QUOTA
2051         if (test_opt(sbi, QUOTA))
2052                 seq_puts(seq, ",quota");
2053         if (test_opt(sbi, USRQUOTA))
2054                 seq_puts(seq, ",usrquota");
2055         if (test_opt(sbi, GRPQUOTA))
2056                 seq_puts(seq, ",grpquota");
2057         if (test_opt(sbi, PRJQUOTA))
2058                 seq_puts(seq, ",prjquota");
2059 #endif
2060         f2fs_show_quota_options(seq, sbi->sb);
2061
2062         fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
2063
2064         if (sbi->sb->s_flags & SB_INLINECRYPT)
2065                 seq_puts(seq, ",inlinecrypt");
2066
2067         if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
2068                 seq_printf(seq, ",alloc_mode=%s", "default");
2069         else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
2070                 seq_printf(seq, ",alloc_mode=%s", "reuse");
2071
2072         if (test_opt(sbi, DISABLE_CHECKPOINT))
2073                 seq_printf(seq, ",checkpoint=disable:%u",
2074                                 F2FS_OPTION(sbi).unusable_cap);
2075         if (test_opt(sbi, MERGE_CHECKPOINT))
2076                 seq_puts(seq, ",checkpoint_merge");
2077         else
2078                 seq_puts(seq, ",nocheckpoint_merge");
2079         if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
2080                 seq_printf(seq, ",fsync_mode=%s", "posix");
2081         else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
2082                 seq_printf(seq, ",fsync_mode=%s", "strict");
2083         else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
2084                 seq_printf(seq, ",fsync_mode=%s", "nobarrier");
2085
2086 #ifdef CONFIG_F2FS_FS_COMPRESSION
2087         f2fs_show_compress_options(seq, sbi->sb);
2088 #endif
2089
2090         if (test_opt(sbi, ATGC))
2091                 seq_puts(seq, ",atgc");
2092
2093         if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_NORMAL)
2094                 seq_printf(seq, ",memory=%s", "normal");
2095         else if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW)
2096                 seq_printf(seq, ",memory=%s", "low");
2097
2098         if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_READONLY)
2099                 seq_printf(seq, ",errors=%s", "remount-ro");
2100         else if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_CONTINUE)
2101                 seq_printf(seq, ",errors=%s", "continue");
2102         else if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_PANIC)
2103                 seq_printf(seq, ",errors=%s", "panic");
2104
2105         return 0;
2106 }
2107
2108 static void default_options(struct f2fs_sb_info *sbi, bool remount)
2109 {
2110         /* init some FS parameters */
2111         if (!remount) {
2112                 set_opt(sbi, READ_EXTENT_CACHE);
2113                 clear_opt(sbi, DISABLE_CHECKPOINT);
2114
2115                 if (f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi))
2116                         set_opt(sbi, DISCARD);
2117
2118                 if (f2fs_sb_has_blkzoned(sbi))
2119                         F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_SECTION;
2120                 else
2121                         F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_BLOCK;
2122         }
2123
2124         if (f2fs_sb_has_readonly(sbi))
2125                 F2FS_OPTION(sbi).active_logs = NR_CURSEG_RO_TYPE;
2126         else
2127                 F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
2128
2129         F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
2130         if (le32_to_cpu(F2FS_RAW_SUPER(sbi)->segment_count_main) <=
2131                                                         SMALL_VOLUME_SEGMENTS)
2132                 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
2133         else
2134                 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
2135         F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
2136         F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
2137         F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
2138         if (f2fs_sb_has_compression(sbi)) {
2139                 F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
2140                 F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
2141                 F2FS_OPTION(sbi).compress_ext_cnt = 0;
2142                 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
2143         }
2144         F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
2145         F2FS_OPTION(sbi).memory_mode = MEMORY_MODE_NORMAL;
2146         F2FS_OPTION(sbi).errors = MOUNT_ERRORS_CONTINUE;
2147
2148         set_opt(sbi, INLINE_XATTR);
2149         set_opt(sbi, INLINE_DATA);
2150         set_opt(sbi, INLINE_DENTRY);
2151         set_opt(sbi, MERGE_CHECKPOINT);
2152         F2FS_OPTION(sbi).unusable_cap = 0;
2153         sbi->sb->s_flags |= SB_LAZYTIME;
2154         if (!f2fs_is_readonly(sbi))
2155                 set_opt(sbi, FLUSH_MERGE);
2156         if (f2fs_sb_has_blkzoned(sbi))
2157                 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
2158         else
2159                 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
2160
2161 #ifdef CONFIG_F2FS_FS_XATTR
2162         set_opt(sbi, XATTR_USER);
2163 #endif
2164 #ifdef CONFIG_F2FS_FS_POSIX_ACL
2165         set_opt(sbi, POSIX_ACL);
2166 #endif
2167
2168         f2fs_build_fault_attr(sbi, 0, 0);
2169 }
2170
2171 #ifdef CONFIG_QUOTA
2172 static int f2fs_enable_quotas(struct super_block *sb);
2173 #endif
2174
2175 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
2176 {
2177         unsigned int s_flags = sbi->sb->s_flags;
2178         struct cp_control cpc;
2179         unsigned int gc_mode = sbi->gc_mode;
2180         int err = 0;
2181         int ret;
2182         block_t unusable;
2183
2184         if (s_flags & SB_RDONLY) {
2185                 f2fs_err(sbi, "checkpoint=disable on readonly fs");
2186                 return -EINVAL;
2187         }
2188         sbi->sb->s_flags |= SB_ACTIVE;
2189
2190         /* check if we need more GC first */
2191         unusable = f2fs_get_unusable_blocks(sbi);
2192         if (!f2fs_disable_cp_again(sbi, unusable))
2193                 goto skip_gc;
2194
2195         f2fs_update_time(sbi, DISABLE_TIME);
2196
2197         sbi->gc_mode = GC_URGENT_HIGH;
2198
2199         while (!f2fs_time_over(sbi, DISABLE_TIME)) {
2200                 struct f2fs_gc_control gc_control = {
2201                         .victim_segno = NULL_SEGNO,
2202                         .init_gc_type = FG_GC,
2203                         .should_migrate_blocks = false,
2204                         .err_gc_skipped = true,
2205                         .no_bg_gc = true,
2206                         .nr_free_secs = 1 };
2207
2208                 f2fs_down_write(&sbi->gc_lock);
2209                 stat_inc_gc_call_count(sbi, FOREGROUND);
2210                 err = f2fs_gc(sbi, &gc_control);
2211                 if (err == -ENODATA) {
2212                         err = 0;
2213                         break;
2214                 }
2215                 if (err && err != -EAGAIN)
2216                         break;
2217         }
2218
2219         ret = sync_filesystem(sbi->sb);
2220         if (ret || err) {
2221                 err = ret ? ret : err;
2222                 goto restore_flag;
2223         }
2224
2225         unusable = f2fs_get_unusable_blocks(sbi);
2226         if (f2fs_disable_cp_again(sbi, unusable)) {
2227                 err = -EAGAIN;
2228                 goto restore_flag;
2229         }
2230
2231 skip_gc:
2232         f2fs_down_write(&sbi->gc_lock);
2233         cpc.reason = CP_PAUSE;
2234         set_sbi_flag(sbi, SBI_CP_DISABLED);
2235         stat_inc_cp_call_count(sbi, TOTAL_CALL);
2236         err = f2fs_write_checkpoint(sbi, &cpc);
2237         if (err)
2238                 goto out_unlock;
2239
2240         spin_lock(&sbi->stat_lock);
2241         sbi->unusable_block_count = unusable;
2242         spin_unlock(&sbi->stat_lock);
2243
2244 out_unlock:
2245         f2fs_up_write(&sbi->gc_lock);
2246 restore_flag:
2247         sbi->gc_mode = gc_mode;
2248         sbi->sb->s_flags = s_flags;     /* Restore SB_RDONLY status */
2249         return err;
2250 }
2251
2252 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
2253 {
2254         int retry = DEFAULT_RETRY_IO_COUNT;
2255
2256         /* we should flush all the data to keep data consistency */
2257         do {
2258                 sync_inodes_sb(sbi->sb);
2259                 f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2260         } while (get_pages(sbi, F2FS_DIRTY_DATA) && retry--);
2261
2262         if (unlikely(retry < 0))
2263                 f2fs_warn(sbi, "checkpoint=enable has some unwritten data.");
2264
2265         f2fs_down_write(&sbi->gc_lock);
2266         f2fs_dirty_to_prefree(sbi);
2267
2268         clear_sbi_flag(sbi, SBI_CP_DISABLED);
2269         set_sbi_flag(sbi, SBI_IS_DIRTY);
2270         f2fs_up_write(&sbi->gc_lock);
2271
2272         f2fs_sync_fs(sbi->sb, 1);
2273
2274         /* Let's ensure there's no pending checkpoint anymore */
2275         f2fs_flush_ckpt_thread(sbi);
2276 }
2277
2278 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
2279 {
2280         struct f2fs_sb_info *sbi = F2FS_SB(sb);
2281         struct f2fs_mount_info org_mount_opt;
2282         unsigned long old_sb_flags;
2283         int err;
2284         bool need_restart_gc = false, need_stop_gc = false;
2285         bool need_restart_flush = false, need_stop_flush = false;
2286         bool need_restart_discard = false, need_stop_discard = false;
2287         bool need_enable_checkpoint = false, need_disable_checkpoint = false;
2288         bool no_read_extent_cache = !test_opt(sbi, READ_EXTENT_CACHE);
2289         bool no_age_extent_cache = !test_opt(sbi, AGE_EXTENT_CACHE);
2290         bool enable_checkpoint = !test_opt(sbi, DISABLE_CHECKPOINT);
2291         bool no_atgc = !test_opt(sbi, ATGC);
2292         bool no_discard = !test_opt(sbi, DISCARD);
2293         bool no_compress_cache = !test_opt(sbi, COMPRESS_CACHE);
2294         bool block_unit_discard = f2fs_block_unit_discard(sbi);
2295 #ifdef CONFIG_QUOTA
2296         int i, j;
2297 #endif
2298
2299         /*
2300          * Save the old mount options in case we
2301          * need to restore them.
2302          */
2303         org_mount_opt = sbi->mount_opt;
2304         old_sb_flags = sb->s_flags;
2305
2306 #ifdef CONFIG_QUOTA
2307         org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
2308         for (i = 0; i < MAXQUOTAS; i++) {
2309                 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2310                         org_mount_opt.s_qf_names[i] =
2311                                 kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
2312                                 GFP_KERNEL);
2313                         if (!org_mount_opt.s_qf_names[i]) {
2314                                 for (j = 0; j < i; j++)
2315                                         kfree(org_mount_opt.s_qf_names[j]);
2316                                 return -ENOMEM;
2317                         }
2318                 } else {
2319                         org_mount_opt.s_qf_names[i] = NULL;
2320                 }
2321         }
2322 #endif
2323
2324         /* recover superblocks we couldn't write due to previous RO mount */
2325         if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
2326                 err = f2fs_commit_super(sbi, false);
2327                 f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
2328                           err);
2329                 if (!err)
2330                         clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2331         }
2332
2333         default_options(sbi, true);
2334
2335         /* parse mount options */
2336         err = parse_options(sb, data, true);
2337         if (err)
2338                 goto restore_opts;
2339
2340 #ifdef CONFIG_BLK_DEV_ZONED
2341         if (f2fs_sb_has_blkzoned(sbi) &&
2342                 sbi->max_open_zones < F2FS_OPTION(sbi).active_logs) {
2343                 f2fs_err(sbi,
2344                         "zoned: max open zones %u is too small, need at least %u open zones",
2345                                  sbi->max_open_zones, F2FS_OPTION(sbi).active_logs);
2346                 err = -EINVAL;
2347                 goto restore_opts;
2348         }
2349 #endif
2350
2351         /* flush outstanding errors before changing fs state */
2352         flush_work(&sbi->s_error_work);
2353
2354         /*
2355          * Previous and new state of filesystem is RO,
2356          * so skip checking GC and FLUSH_MERGE conditions.
2357          */
2358         if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
2359                 goto skip;
2360
2361         if (f2fs_dev_is_readonly(sbi) && !(*flags & SB_RDONLY)) {
2362                 err = -EROFS;
2363                 goto restore_opts;
2364         }
2365
2366 #ifdef CONFIG_QUOTA
2367         if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
2368                 err = dquot_suspend(sb, -1);
2369                 if (err < 0)
2370                         goto restore_opts;
2371         } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
2372                 /* dquot_resume needs RW */
2373                 sb->s_flags &= ~SB_RDONLY;
2374                 if (sb_any_quota_suspended(sb)) {
2375                         dquot_resume(sb, -1);
2376                 } else if (f2fs_sb_has_quota_ino(sbi)) {
2377                         err = f2fs_enable_quotas(sb);
2378                         if (err)
2379                                 goto restore_opts;
2380                 }
2381         }
2382 #endif
2383         if (f2fs_lfs_mode(sbi) && !IS_F2FS_IPU_DISABLE(sbi)) {
2384                 err = -EINVAL;
2385                 f2fs_warn(sbi, "LFS is not compatible with IPU");
2386                 goto restore_opts;
2387         }
2388
2389         /* disallow enable atgc dynamically */
2390         if (no_atgc == !!test_opt(sbi, ATGC)) {
2391                 err = -EINVAL;
2392                 f2fs_warn(sbi, "switch atgc option is not allowed");
2393                 goto restore_opts;
2394         }
2395
2396         /* disallow enable/disable extent_cache dynamically */
2397         if (no_read_extent_cache == !!test_opt(sbi, READ_EXTENT_CACHE)) {
2398                 err = -EINVAL;
2399                 f2fs_warn(sbi, "switch extent_cache option is not allowed");
2400                 goto restore_opts;
2401         }
2402         /* disallow enable/disable age extent_cache dynamically */
2403         if (no_age_extent_cache == !!test_opt(sbi, AGE_EXTENT_CACHE)) {
2404                 err = -EINVAL;
2405                 f2fs_warn(sbi, "switch age_extent_cache option is not allowed");
2406                 goto restore_opts;
2407         }
2408
2409         if (no_compress_cache == !!test_opt(sbi, COMPRESS_CACHE)) {
2410                 err = -EINVAL;
2411                 f2fs_warn(sbi, "switch compress_cache option is not allowed");
2412                 goto restore_opts;
2413         }
2414
2415         if (block_unit_discard != f2fs_block_unit_discard(sbi)) {
2416                 err = -EINVAL;
2417                 f2fs_warn(sbi, "switch discard_unit option is not allowed");
2418                 goto restore_opts;
2419         }
2420
2421         if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
2422                 err = -EINVAL;
2423                 f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
2424                 goto restore_opts;
2425         }
2426
2427         /*
2428          * We stop the GC thread if FS is mounted as RO
2429          * or if background_gc = off is passed in mount
2430          * option. Also sync the filesystem.
2431          */
2432         if ((*flags & SB_RDONLY) ||
2433                         (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF &&
2434                         !test_opt(sbi, GC_MERGE))) {
2435                 if (sbi->gc_thread) {
2436                         f2fs_stop_gc_thread(sbi);
2437                         need_restart_gc = true;
2438                 }
2439         } else if (!sbi->gc_thread) {
2440                 err = f2fs_start_gc_thread(sbi);
2441                 if (err)
2442                         goto restore_opts;
2443                 need_stop_gc = true;
2444         }
2445
2446         if (*flags & SB_RDONLY) {
2447                 sync_inodes_sb(sb);
2448
2449                 set_sbi_flag(sbi, SBI_IS_DIRTY);
2450                 set_sbi_flag(sbi, SBI_IS_CLOSE);
2451                 f2fs_sync_fs(sb, 1);
2452                 clear_sbi_flag(sbi, SBI_IS_CLOSE);
2453         }
2454
2455         /*
2456          * We stop issue flush thread if FS is mounted as RO
2457          * or if flush_merge is not passed in mount option.
2458          */
2459         if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2460                 clear_opt(sbi, FLUSH_MERGE);
2461                 f2fs_destroy_flush_cmd_control(sbi, false);
2462                 need_restart_flush = true;
2463         } else {
2464                 err = f2fs_create_flush_cmd_control(sbi);
2465                 if (err)
2466                         goto restore_gc;
2467                 need_stop_flush = true;
2468         }
2469
2470         if (no_discard == !!test_opt(sbi, DISCARD)) {
2471                 if (test_opt(sbi, DISCARD)) {
2472                         err = f2fs_start_discard_thread(sbi);
2473                         if (err)
2474                                 goto restore_flush;
2475                         need_stop_discard = true;
2476                 } else {
2477                         f2fs_stop_discard_thread(sbi);
2478                         f2fs_issue_discard_timeout(sbi);
2479                         need_restart_discard = true;
2480                 }
2481         }
2482
2483         if (enable_checkpoint == !!test_opt(sbi, DISABLE_CHECKPOINT)) {
2484                 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
2485                         err = f2fs_disable_checkpoint(sbi);
2486                         if (err)
2487                                 goto restore_discard;
2488                         need_enable_checkpoint = true;
2489                 } else {
2490                         f2fs_enable_checkpoint(sbi);
2491                         need_disable_checkpoint = true;
2492                 }
2493         }
2494
2495         /*
2496          * Place this routine at the end, since a new checkpoint would be
2497          * triggered while remount and we need to take care of it before
2498          * returning from remount.
2499          */
2500         if ((*flags & SB_RDONLY) || test_opt(sbi, DISABLE_CHECKPOINT) ||
2501                         !test_opt(sbi, MERGE_CHECKPOINT)) {
2502                 f2fs_stop_ckpt_thread(sbi);
2503         } else {
2504                 /* Flush if the prevous checkpoint, if exists. */
2505                 f2fs_flush_ckpt_thread(sbi);
2506
2507                 err = f2fs_start_ckpt_thread(sbi);
2508                 if (err) {
2509                         f2fs_err(sbi,
2510                             "Failed to start F2FS issue_checkpoint_thread (%d)",
2511                             err);
2512                         goto restore_checkpoint;
2513                 }
2514         }
2515
2516 skip:
2517 #ifdef CONFIG_QUOTA
2518         /* Release old quota file names */
2519         for (i = 0; i < MAXQUOTAS; i++)
2520                 kfree(org_mount_opt.s_qf_names[i]);
2521 #endif
2522         /* Update the POSIXACL Flag */
2523         sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
2524                 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
2525
2526         limit_reserve_root(sbi);
2527         adjust_unusable_cap_perc(sbi);
2528         *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
2529         return 0;
2530 restore_checkpoint:
2531         if (need_enable_checkpoint) {
2532                 f2fs_enable_checkpoint(sbi);
2533         } else if (need_disable_checkpoint) {
2534                 if (f2fs_disable_checkpoint(sbi))
2535                         f2fs_warn(sbi, "checkpoint has not been disabled");
2536         }
2537 restore_discard:
2538         if (need_restart_discard) {
2539                 if (f2fs_start_discard_thread(sbi))
2540                         f2fs_warn(sbi, "discard has been stopped");
2541         } else if (need_stop_discard) {
2542                 f2fs_stop_discard_thread(sbi);
2543         }
2544 restore_flush:
2545         if (need_restart_flush) {
2546                 if (f2fs_create_flush_cmd_control(sbi))
2547                         f2fs_warn(sbi, "background flush thread has stopped");
2548         } else if (need_stop_flush) {
2549                 clear_opt(sbi, FLUSH_MERGE);
2550                 f2fs_destroy_flush_cmd_control(sbi, false);
2551         }
2552 restore_gc:
2553         if (need_restart_gc) {
2554                 if (f2fs_start_gc_thread(sbi))
2555                         f2fs_warn(sbi, "background gc thread has stopped");
2556         } else if (need_stop_gc) {
2557                 f2fs_stop_gc_thread(sbi);
2558         }
2559 restore_opts:
2560 #ifdef CONFIG_QUOTA
2561         F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
2562         for (i = 0; i < MAXQUOTAS; i++) {
2563                 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
2564                 F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
2565         }
2566 #endif
2567         sbi->mount_opt = org_mount_opt;
2568         sb->s_flags = old_sb_flags;
2569         return err;
2570 }
2571
2572 static void f2fs_shutdown(struct super_block *sb)
2573 {
2574         f2fs_do_shutdown(F2FS_SB(sb), F2FS_GOING_DOWN_NOSYNC, false);
2575 }
2576
2577 #ifdef CONFIG_QUOTA
2578 static bool f2fs_need_recovery(struct f2fs_sb_info *sbi)
2579 {
2580         /* need to recovery orphan */
2581         if (is_set_ckpt_flags(sbi, CP_ORPHAN_PRESENT_FLAG))
2582                 return true;
2583         /* need to recovery data */
2584         if (test_opt(sbi, DISABLE_ROLL_FORWARD))
2585                 return false;
2586         if (test_opt(sbi, NORECOVERY))
2587                 return false;
2588         return !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG);
2589 }
2590
2591 static bool f2fs_recover_quota_begin(struct f2fs_sb_info *sbi)
2592 {
2593         bool readonly = f2fs_readonly(sbi->sb);
2594
2595         if (!f2fs_need_recovery(sbi))
2596                 return false;
2597
2598         /* it doesn't need to check f2fs_sb_has_readonly() */
2599         if (f2fs_hw_is_readonly(sbi))
2600                 return false;
2601
2602         if (readonly) {
2603                 sbi->sb->s_flags &= ~SB_RDONLY;
2604                 set_sbi_flag(sbi, SBI_IS_WRITABLE);
2605         }
2606
2607         /*
2608          * Turn on quotas which were not enabled for read-only mounts if
2609          * filesystem has quota feature, so that they are updated correctly.
2610          */
2611         return f2fs_enable_quota_files(sbi, readonly);
2612 }
2613
2614 static void f2fs_recover_quota_end(struct f2fs_sb_info *sbi,
2615                                                 bool quota_enabled)
2616 {
2617         if (quota_enabled)
2618                 f2fs_quota_off_umount(sbi->sb);
2619
2620         if (is_sbi_flag_set(sbi, SBI_IS_WRITABLE)) {
2621                 clear_sbi_flag(sbi, SBI_IS_WRITABLE);
2622                 sbi->sb->s_flags |= SB_RDONLY;
2623         }
2624 }
2625
2626 /* Read data from quotafile */
2627 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
2628                                size_t len, loff_t off)
2629 {
2630         struct inode *inode = sb_dqopt(sb)->files[type];
2631         struct address_space *mapping = inode->i_mapping;
2632         block_t blkidx = F2FS_BYTES_TO_BLK(off);
2633         int offset = off & (sb->s_blocksize - 1);
2634         int tocopy;
2635         size_t toread;
2636         loff_t i_size = i_size_read(inode);
2637         struct page *page;
2638
2639         if (off > i_size)
2640                 return 0;
2641
2642         if (off + len > i_size)
2643                 len = i_size - off;
2644         toread = len;
2645         while (toread > 0) {
2646                 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
2647 repeat:
2648                 page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
2649                 if (IS_ERR(page)) {
2650                         if (PTR_ERR(page) == -ENOMEM) {
2651                                 memalloc_retry_wait(GFP_NOFS);
2652                                 goto repeat;
2653                         }
2654                         set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2655                         return PTR_ERR(page);
2656                 }
2657
2658                 lock_page(page);
2659
2660                 if (unlikely(page->mapping != mapping)) {
2661                         f2fs_put_page(page, 1);
2662                         goto repeat;
2663                 }
2664                 if (unlikely(!PageUptodate(page))) {
2665                         f2fs_put_page(page, 1);
2666                         set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2667                         return -EIO;
2668                 }
2669
2670                 memcpy_from_page(data, page, offset, tocopy);
2671                 f2fs_put_page(page, 1);
2672
2673                 offset = 0;
2674                 toread -= tocopy;
2675                 data += tocopy;
2676                 blkidx++;
2677         }
2678         return len;
2679 }
2680
2681 /* Write to quotafile */
2682 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2683                                 const char *data, size_t len, loff_t off)
2684 {
2685         struct inode *inode = sb_dqopt(sb)->files[type];
2686         struct address_space *mapping = inode->i_mapping;
2687         const struct address_space_operations *a_ops = mapping->a_ops;
2688         int offset = off & (sb->s_blocksize - 1);
2689         size_t towrite = len;
2690         struct page *page;
2691         void *fsdata = NULL;
2692         int err = 0;
2693         int tocopy;
2694
2695         while (towrite > 0) {
2696                 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2697                                                                 towrite);
2698 retry:
2699                 err = a_ops->write_begin(NULL, mapping, off, tocopy,
2700                                                         &page, &fsdata);
2701                 if (unlikely(err)) {
2702                         if (err == -ENOMEM) {
2703                                 f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2704                                 goto retry;
2705                         }
2706                         set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2707                         break;
2708                 }
2709
2710                 memcpy_to_page(page, offset, data, tocopy);
2711
2712                 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2713                                                 page, fsdata);
2714                 offset = 0;
2715                 towrite -= tocopy;
2716                 off += tocopy;
2717                 data += tocopy;
2718                 cond_resched();
2719         }
2720
2721         if (len == towrite)
2722                 return err;
2723         inode_set_mtime_to_ts(inode, inode_set_ctime_current(inode));
2724         f2fs_mark_inode_dirty_sync(inode, false);
2725         return len - towrite;
2726 }
2727
2728 int f2fs_dquot_initialize(struct inode *inode)
2729 {
2730         if (time_to_inject(F2FS_I_SB(inode), FAULT_DQUOT_INIT))
2731                 return -ESRCH;
2732
2733         return dquot_initialize(inode);
2734 }
2735
2736 static struct dquot __rcu **f2fs_get_dquots(struct inode *inode)
2737 {
2738         return F2FS_I(inode)->i_dquot;
2739 }
2740
2741 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2742 {
2743         return &F2FS_I(inode)->i_reserved_quota;
2744 }
2745
2746 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2747 {
2748         if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2749                 f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2750                 return 0;
2751         }
2752
2753         return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2754                                         F2FS_OPTION(sbi).s_jquota_fmt, type);
2755 }
2756
2757 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2758 {
2759         int enabled = 0;
2760         int i, err;
2761
2762         if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2763                 err = f2fs_enable_quotas(sbi->sb);
2764                 if (err) {
2765                         f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2766                         return 0;
2767                 }
2768                 return 1;
2769         }
2770
2771         for (i = 0; i < MAXQUOTAS; i++) {
2772                 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2773                         err = f2fs_quota_on_mount(sbi, i);
2774                         if (!err) {
2775                                 enabled = 1;
2776                                 continue;
2777                         }
2778                         f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2779                                  err, i);
2780                 }
2781         }
2782         return enabled;
2783 }
2784
2785 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2786                              unsigned int flags)
2787 {
2788         struct inode *qf_inode;
2789         unsigned long qf_inum;
2790         unsigned long qf_flag = F2FS_QUOTA_DEFAULT_FL;
2791         int err;
2792
2793         BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2794
2795         qf_inum = f2fs_qf_ino(sb, type);
2796         if (!qf_inum)
2797                 return -EPERM;
2798
2799         qf_inode = f2fs_iget(sb, qf_inum);
2800         if (IS_ERR(qf_inode)) {
2801                 f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2802                 return PTR_ERR(qf_inode);
2803         }
2804
2805         /* Don't account quota for quota files to avoid recursion */
2806         inode_lock(qf_inode);
2807         qf_inode->i_flags |= S_NOQUOTA;
2808
2809         if ((F2FS_I(qf_inode)->i_flags & qf_flag) != qf_flag) {
2810                 F2FS_I(qf_inode)->i_flags |= qf_flag;
2811                 f2fs_set_inode_flags(qf_inode);
2812         }
2813         inode_unlock(qf_inode);
2814
2815         err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2816         iput(qf_inode);
2817         return err;
2818 }
2819
2820 static int f2fs_enable_quotas(struct super_block *sb)
2821 {
2822         struct f2fs_sb_info *sbi = F2FS_SB(sb);
2823         int type, err = 0;
2824         unsigned long qf_inum;
2825         bool quota_mopt[MAXQUOTAS] = {
2826                 test_opt(sbi, USRQUOTA),
2827                 test_opt(sbi, GRPQUOTA),
2828                 test_opt(sbi, PRJQUOTA),
2829         };
2830
2831         if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2832                 f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2833                 return 0;
2834         }
2835
2836         sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2837
2838         for (type = 0; type < MAXQUOTAS; type++) {
2839                 qf_inum = f2fs_qf_ino(sb, type);
2840                 if (qf_inum) {
2841                         err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2842                                 DQUOT_USAGE_ENABLED |
2843                                 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2844                         if (err) {
2845                                 f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2846                                          type, err);
2847                                 for (type--; type >= 0; type--)
2848                                         dquot_quota_off(sb, type);
2849                                 set_sbi_flag(F2FS_SB(sb),
2850                                                 SBI_QUOTA_NEED_REPAIR);
2851                                 return err;
2852                         }
2853                 }
2854         }
2855         return 0;
2856 }
2857
2858 static int f2fs_quota_sync_file(struct f2fs_sb_info *sbi, int type)
2859 {
2860         struct quota_info *dqopt = sb_dqopt(sbi->sb);
2861         struct address_space *mapping = dqopt->files[type]->i_mapping;
2862         int ret = 0;
2863
2864         ret = dquot_writeback_dquots(sbi->sb, type);
2865         if (ret)
2866                 goto out;
2867
2868         ret = filemap_fdatawrite(mapping);
2869         if (ret)
2870                 goto out;
2871
2872         /* if we are using journalled quota */
2873         if (is_journalled_quota(sbi))
2874                 goto out;
2875
2876         ret = filemap_fdatawait(mapping);
2877
2878         truncate_inode_pages(&dqopt->files[type]->i_data, 0);
2879 out:
2880         if (ret)
2881                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2882         return ret;
2883 }
2884
2885 int f2fs_quota_sync(struct super_block *sb, int type)
2886 {
2887         struct f2fs_sb_info *sbi = F2FS_SB(sb);
2888         struct quota_info *dqopt = sb_dqopt(sb);
2889         int cnt;
2890         int ret = 0;
2891
2892         /*
2893          * Now when everything is written we can discard the pagecache so
2894          * that userspace sees the changes.
2895          */
2896         for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2897
2898                 if (type != -1 && cnt != type)
2899                         continue;
2900
2901                 if (!sb_has_quota_active(sb, cnt))
2902                         continue;
2903
2904                 if (!f2fs_sb_has_quota_ino(sbi))
2905                         inode_lock(dqopt->files[cnt]);
2906
2907                 /*
2908                  * do_quotactl
2909                  *  f2fs_quota_sync
2910                  *  f2fs_down_read(quota_sem)
2911                  *  dquot_writeback_dquots()
2912                  *  f2fs_dquot_commit
2913                  *                            block_operation
2914                  *                            f2fs_down_read(quota_sem)
2915                  */
2916                 f2fs_lock_op(sbi);
2917                 f2fs_down_read(&sbi->quota_sem);
2918
2919                 ret = f2fs_quota_sync_file(sbi, cnt);
2920
2921                 f2fs_up_read(&sbi->quota_sem);
2922                 f2fs_unlock_op(sbi);
2923
2924                 if (!f2fs_sb_has_quota_ino(sbi))
2925                         inode_unlock(dqopt->files[cnt]);
2926
2927                 if (ret)
2928                         break;
2929         }
2930         return ret;
2931 }
2932
2933 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2934                                                         const struct path *path)
2935 {
2936         struct inode *inode;
2937         int err;
2938
2939         /* if quota sysfile exists, deny enabling quota with specific file */
2940         if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2941                 f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2942                 return -EBUSY;
2943         }
2944
2945         if (path->dentry->d_sb != sb)
2946                 return -EXDEV;
2947
2948         err = f2fs_quota_sync(sb, type);
2949         if (err)
2950                 return err;
2951
2952         inode = d_inode(path->dentry);
2953
2954         err = filemap_fdatawrite(inode->i_mapping);
2955         if (err)
2956                 return err;
2957
2958         err = filemap_fdatawait(inode->i_mapping);
2959         if (err)
2960                 return err;
2961
2962         err = dquot_quota_on(sb, type, format_id, path);
2963         if (err)
2964                 return err;
2965
2966         inode_lock(inode);
2967         F2FS_I(inode)->i_flags |= F2FS_QUOTA_DEFAULT_FL;
2968         f2fs_set_inode_flags(inode);
2969         inode_unlock(inode);
2970         f2fs_mark_inode_dirty_sync(inode, false);
2971
2972         return 0;
2973 }
2974
2975 static int __f2fs_quota_off(struct super_block *sb, int type)
2976 {
2977         struct inode *inode = sb_dqopt(sb)->files[type];
2978         int err;
2979
2980         if (!inode || !igrab(inode))
2981                 return dquot_quota_off(sb, type);
2982
2983         err = f2fs_quota_sync(sb, type);
2984         if (err)
2985                 goto out_put;
2986
2987         err = dquot_quota_off(sb, type);
2988         if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
2989                 goto out_put;
2990
2991         inode_lock(inode);
2992         F2FS_I(inode)->i_flags &= ~F2FS_QUOTA_DEFAULT_FL;
2993         f2fs_set_inode_flags(inode);
2994         inode_unlock(inode);
2995         f2fs_mark_inode_dirty_sync(inode, false);
2996 out_put:
2997         iput(inode);
2998         return err;
2999 }
3000
3001 static int f2fs_quota_off(struct super_block *sb, int type)
3002 {
3003         struct f2fs_sb_info *sbi = F2FS_SB(sb);
3004         int err;
3005
3006         err = __f2fs_quota_off(sb, type);
3007
3008         /*
3009          * quotactl can shutdown journalled quota, result in inconsistence
3010          * between quota record and fs data by following updates, tag the
3011          * flag to let fsck be aware of it.
3012          */
3013         if (is_journalled_quota(sbi))
3014                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3015         return err;
3016 }
3017
3018 void f2fs_quota_off_umount(struct super_block *sb)
3019 {
3020         int type;
3021         int err;
3022
3023         for (type = 0; type < MAXQUOTAS; type++) {
3024                 err = __f2fs_quota_off(sb, type);
3025                 if (err) {
3026                         int ret = dquot_quota_off(sb, type);
3027
3028                         f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
3029                                  type, err, ret);
3030                         set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
3031                 }
3032         }
3033         /*
3034          * In case of checkpoint=disable, we must flush quota blocks.
3035          * This can cause NULL exception for node_inode in end_io, since
3036          * put_super already dropped it.
3037          */
3038         sync_filesystem(sb);
3039 }
3040
3041 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
3042 {
3043         struct quota_info *dqopt = sb_dqopt(sb);
3044         int type;
3045
3046         for (type = 0; type < MAXQUOTAS; type++) {
3047                 if (!dqopt->files[type])
3048                         continue;
3049                 f2fs_inode_synced(dqopt->files[type]);
3050         }
3051 }
3052
3053 static int f2fs_dquot_commit(struct dquot *dquot)
3054 {
3055         struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3056         int ret;
3057
3058         f2fs_down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
3059         ret = dquot_commit(dquot);
3060         if (ret < 0)
3061                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3062         f2fs_up_read(&sbi->quota_sem);
3063         return ret;
3064 }
3065
3066 static int f2fs_dquot_acquire(struct dquot *dquot)
3067 {
3068         struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3069         int ret;
3070
3071         f2fs_down_read(&sbi->quota_sem);
3072         ret = dquot_acquire(dquot);
3073         if (ret < 0)
3074                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3075         f2fs_up_read(&sbi->quota_sem);
3076         return ret;
3077 }
3078
3079 static int f2fs_dquot_release(struct dquot *dquot)
3080 {
3081         struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3082         int ret = dquot_release(dquot);
3083
3084         if (ret < 0)
3085                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3086         return ret;
3087 }
3088
3089 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
3090 {
3091         struct super_block *sb = dquot->dq_sb;
3092         struct f2fs_sb_info *sbi = F2FS_SB(sb);
3093         int ret = dquot_mark_dquot_dirty(dquot);
3094
3095         /* if we are using journalled quota */
3096         if (is_journalled_quota(sbi))
3097                 set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
3098
3099         return ret;
3100 }
3101
3102 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
3103 {
3104         struct f2fs_sb_info *sbi = F2FS_SB(sb);
3105         int ret = dquot_commit_info(sb, type);
3106
3107         if (ret < 0)
3108                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3109         return ret;
3110 }
3111
3112 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
3113 {
3114         *projid = F2FS_I(inode)->i_projid;
3115         return 0;
3116 }
3117
3118 static const struct dquot_operations f2fs_quota_operations = {
3119         .get_reserved_space = f2fs_get_reserved_space,
3120         .write_dquot    = f2fs_dquot_commit,
3121         .acquire_dquot  = f2fs_dquot_acquire,
3122         .release_dquot  = f2fs_dquot_release,
3123         .mark_dirty     = f2fs_dquot_mark_dquot_dirty,
3124         .write_info     = f2fs_dquot_commit_info,
3125         .alloc_dquot    = dquot_alloc,
3126         .destroy_dquot  = dquot_destroy,
3127         .get_projid     = f2fs_get_projid,
3128         .get_next_id    = dquot_get_next_id,
3129 };
3130
3131 static const struct quotactl_ops f2fs_quotactl_ops = {
3132         .quota_on       = f2fs_quota_on,
3133         .quota_off      = f2fs_quota_off,
3134         .quota_sync     = f2fs_quota_sync,
3135         .get_state      = dquot_get_state,
3136         .set_info       = dquot_set_dqinfo,
3137         .get_dqblk      = dquot_get_dqblk,
3138         .set_dqblk      = dquot_set_dqblk,
3139         .get_nextdqblk  = dquot_get_next_dqblk,
3140 };
3141 #else
3142 int f2fs_dquot_initialize(struct inode *inode)
3143 {
3144         return 0;
3145 }
3146
3147 int f2fs_quota_sync(struct super_block *sb, int type)
3148 {
3149         return 0;
3150 }
3151
3152 void f2fs_quota_off_umount(struct super_block *sb)
3153 {
3154 }
3155 #endif
3156
3157 static const struct super_operations f2fs_sops = {
3158         .alloc_inode    = f2fs_alloc_inode,
3159         .free_inode     = f2fs_free_inode,
3160         .drop_inode     = f2fs_drop_inode,
3161         .write_inode    = f2fs_write_inode,
3162         .dirty_inode    = f2fs_dirty_inode,
3163         .show_options   = f2fs_show_options,
3164 #ifdef CONFIG_QUOTA
3165         .quota_read     = f2fs_quota_read,
3166         .quota_write    = f2fs_quota_write,
3167         .get_dquots     = f2fs_get_dquots,
3168 #endif
3169         .evict_inode    = f2fs_evict_inode,
3170         .put_super      = f2fs_put_super,
3171         .sync_fs        = f2fs_sync_fs,
3172         .freeze_fs      = f2fs_freeze,
3173         .unfreeze_fs    = f2fs_unfreeze,
3174         .statfs         = f2fs_statfs,
3175         .remount_fs     = f2fs_remount,
3176         .shutdown       = f2fs_shutdown,
3177 };
3178
3179 #ifdef CONFIG_FS_ENCRYPTION
3180 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
3181 {
3182         return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3183                                 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3184                                 ctx, len, NULL);
3185 }
3186
3187 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
3188                                                         void *fs_data)
3189 {
3190         struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3191
3192         /*
3193          * Encrypting the root directory is not allowed because fsck
3194          * expects lost+found directory to exist and remain unencrypted
3195          * if LOST_FOUND feature is enabled.
3196          *
3197          */
3198         if (f2fs_sb_has_lost_found(sbi) &&
3199                         inode->i_ino == F2FS_ROOT_INO(sbi))
3200                 return -EPERM;
3201
3202         return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3203                                 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3204                                 ctx, len, fs_data, XATTR_CREATE);
3205 }
3206
3207 static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
3208 {
3209         return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
3210 }
3211
3212 static bool f2fs_has_stable_inodes(struct super_block *sb)
3213 {
3214         return true;
3215 }
3216
3217 static struct block_device **f2fs_get_devices(struct super_block *sb,
3218                                               unsigned int *num_devs)
3219 {
3220         struct f2fs_sb_info *sbi = F2FS_SB(sb);
3221         struct block_device **devs;
3222         int i;
3223
3224         if (!f2fs_is_multi_device(sbi))
3225                 return NULL;
3226
3227         devs = kmalloc_array(sbi->s_ndevs, sizeof(*devs), GFP_KERNEL);
3228         if (!devs)
3229                 return ERR_PTR(-ENOMEM);
3230
3231         for (i = 0; i < sbi->s_ndevs; i++)
3232                 devs[i] = FDEV(i).bdev;
3233         *num_devs = sbi->s_ndevs;
3234         return devs;
3235 }
3236
3237 static const struct fscrypt_operations f2fs_cryptops = {
3238         .needs_bounce_pages     = 1,
3239         .has_32bit_inodes       = 1,
3240         .supports_subblock_data_units = 1,
3241         .legacy_key_prefix      = "f2fs:",
3242         .get_context            = f2fs_get_context,
3243         .set_context            = f2fs_set_context,
3244         .get_dummy_policy       = f2fs_get_dummy_policy,
3245         .empty_dir              = f2fs_empty_dir,
3246         .has_stable_inodes      = f2fs_has_stable_inodes,
3247         .get_devices            = f2fs_get_devices,
3248 };
3249 #endif
3250
3251 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
3252                 u64 ino, u32 generation)
3253 {
3254         struct f2fs_sb_info *sbi = F2FS_SB(sb);
3255         struct inode *inode;
3256
3257         if (f2fs_check_nid_range(sbi, ino))
3258                 return ERR_PTR(-ESTALE);
3259
3260         /*
3261          * f2fs_iget isn't quite right if the inode is currently unallocated!
3262          * However f2fs_iget currently does appropriate checks to handle stale
3263          * inodes so everything is OK.
3264          */
3265         inode = f2fs_iget(sb, ino);
3266         if (IS_ERR(inode))
3267                 return ERR_CAST(inode);
3268         if (unlikely(generation && inode->i_generation != generation)) {
3269                 /* we didn't find the right inode.. */
3270                 iput(inode);
3271                 return ERR_PTR(-ESTALE);
3272         }
3273         return inode;
3274 }
3275
3276 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
3277                 int fh_len, int fh_type)
3278 {
3279         return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
3280                                     f2fs_nfs_get_inode);
3281 }
3282
3283 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
3284                 int fh_len, int fh_type)
3285 {
3286         return generic_fh_to_parent(sb, fid, fh_len, fh_type,
3287                                     f2fs_nfs_get_inode);
3288 }
3289
3290 static const struct export_operations f2fs_export_ops = {
3291         .encode_fh = generic_encode_ino32_fh,
3292         .fh_to_dentry = f2fs_fh_to_dentry,
3293         .fh_to_parent = f2fs_fh_to_parent,
3294         .get_parent = f2fs_get_parent,
3295 };
3296
3297 loff_t max_file_blocks(struct inode *inode)
3298 {
3299         loff_t result = 0;
3300         loff_t leaf_count;
3301
3302         /*
3303          * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
3304          * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
3305          * space in inode.i_addr, it will be more safe to reassign
3306          * result as zero.
3307          */
3308
3309         if (inode && f2fs_compressed_file(inode))
3310                 leaf_count = ADDRS_PER_BLOCK(inode);
3311         else
3312                 leaf_count = DEF_ADDRS_PER_BLOCK;
3313
3314         /* two direct node blocks */
3315         result += (leaf_count * 2);
3316
3317         /* two indirect node blocks */
3318         leaf_count *= NIDS_PER_BLOCK;
3319         result += (leaf_count * 2);
3320
3321         /* one double indirect node block */
3322         leaf_count *= NIDS_PER_BLOCK;
3323         result += leaf_count;
3324
3325         /*
3326          * For compatibility with FSCRYPT_POLICY_FLAG_IV_INO_LBLK_{64,32} with
3327          * a 4K crypto data unit, we must restrict the max filesize to what can
3328          * fit within U32_MAX + 1 data units.
3329          */
3330
3331         result = min(result, (((loff_t)U32_MAX + 1) * 4096) >> F2FS_BLKSIZE_BITS);
3332
3333         return result;
3334 }
3335
3336 static int __f2fs_commit_super(struct buffer_head *bh,
3337                         struct f2fs_super_block *super)
3338 {
3339         lock_buffer(bh);
3340         if (super)
3341                 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
3342         set_buffer_dirty(bh);
3343         unlock_buffer(bh);
3344
3345         /* it's rare case, we can do fua all the time */
3346         return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
3347 }
3348
3349 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
3350                                         struct buffer_head *bh)
3351 {
3352         struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3353                                         (bh->b_data + F2FS_SUPER_OFFSET);
3354         struct super_block *sb = sbi->sb;
3355         u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
3356         u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
3357         u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
3358         u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
3359         u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
3360         u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
3361         u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
3362         u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
3363         u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
3364         u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
3365         u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3366         u32 segment_count = le32_to_cpu(raw_super->segment_count);
3367         u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3368         u64 main_end_blkaddr = main_blkaddr +
3369                                 (segment_count_main << log_blocks_per_seg);
3370         u64 seg_end_blkaddr = segment0_blkaddr +
3371                                 (segment_count << log_blocks_per_seg);
3372
3373         if (segment0_blkaddr != cp_blkaddr) {
3374                 f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
3375                           segment0_blkaddr, cp_blkaddr);
3376                 return true;
3377         }
3378
3379         if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
3380                                                         sit_blkaddr) {
3381                 f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
3382                           cp_blkaddr, sit_blkaddr,
3383                           segment_count_ckpt << log_blocks_per_seg);
3384                 return true;
3385         }
3386
3387         if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
3388                                                         nat_blkaddr) {
3389                 f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
3390                           sit_blkaddr, nat_blkaddr,
3391                           segment_count_sit << log_blocks_per_seg);
3392                 return true;
3393         }
3394
3395         if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
3396                                                         ssa_blkaddr) {
3397                 f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
3398                           nat_blkaddr, ssa_blkaddr,
3399                           segment_count_nat << log_blocks_per_seg);
3400                 return true;
3401         }
3402
3403         if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
3404                                                         main_blkaddr) {
3405                 f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
3406                           ssa_blkaddr, main_blkaddr,
3407                           segment_count_ssa << log_blocks_per_seg);
3408                 return true;
3409         }
3410
3411         if (main_end_blkaddr > seg_end_blkaddr) {
3412                 f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
3413                           main_blkaddr, seg_end_blkaddr,
3414                           segment_count_main << log_blocks_per_seg);
3415                 return true;
3416         } else if (main_end_blkaddr < seg_end_blkaddr) {
3417                 int err = 0;
3418                 char *res;
3419
3420                 /* fix in-memory information all the time */
3421                 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
3422                                 segment0_blkaddr) >> log_blocks_per_seg);
3423
3424                 if (f2fs_readonly(sb) || f2fs_hw_is_readonly(sbi)) {
3425                         set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3426                         res = "internally";
3427                 } else {
3428                         err = __f2fs_commit_super(bh, NULL);
3429                         res = err ? "failed" : "done";
3430                 }
3431                 f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
3432                           res, main_blkaddr, seg_end_blkaddr,
3433                           segment_count_main << log_blocks_per_seg);
3434                 if (err)
3435                         return true;
3436         }
3437         return false;
3438 }
3439
3440 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
3441                                 struct buffer_head *bh)
3442 {
3443         block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
3444         block_t total_sections, blocks_per_seg;
3445         struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3446                                         (bh->b_data + F2FS_SUPER_OFFSET);
3447         size_t crc_offset = 0;
3448         __u32 crc = 0;
3449
3450         if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
3451                 f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
3452                           F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
3453                 return -EINVAL;
3454         }
3455
3456         /* Check checksum_offset and crc in superblock */
3457         if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
3458                 crc_offset = le32_to_cpu(raw_super->checksum_offset);
3459                 if (crc_offset !=
3460                         offsetof(struct f2fs_super_block, crc)) {
3461                         f2fs_info(sbi, "Invalid SB checksum offset: %zu",
3462                                   crc_offset);
3463                         return -EFSCORRUPTED;
3464                 }
3465                 crc = le32_to_cpu(raw_super->crc);
3466                 if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
3467                         f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
3468                         return -EFSCORRUPTED;
3469                 }
3470         }
3471
3472         /* only support block_size equals to PAGE_SIZE */
3473         if (le32_to_cpu(raw_super->log_blocksize) != F2FS_BLKSIZE_BITS) {
3474                 f2fs_info(sbi, "Invalid log_blocksize (%u), supports only %u",
3475                           le32_to_cpu(raw_super->log_blocksize),
3476                           F2FS_BLKSIZE_BITS);
3477                 return -EFSCORRUPTED;
3478         }
3479
3480         /* check log blocks per segment */
3481         if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
3482                 f2fs_info(sbi, "Invalid log blocks per segment (%u)",
3483                           le32_to_cpu(raw_super->log_blocks_per_seg));
3484                 return -EFSCORRUPTED;
3485         }
3486
3487         /* Currently, support 512/1024/2048/4096/16K bytes sector size */
3488         if (le32_to_cpu(raw_super->log_sectorsize) >
3489                                 F2FS_MAX_LOG_SECTOR_SIZE ||
3490                 le32_to_cpu(raw_super->log_sectorsize) <
3491                                 F2FS_MIN_LOG_SECTOR_SIZE) {
3492                 f2fs_info(sbi, "Invalid log sectorsize (%u)",
3493                           le32_to_cpu(raw_super->log_sectorsize));
3494                 return -EFSCORRUPTED;
3495         }
3496         if (le32_to_cpu(raw_super->log_sectors_per_block) +
3497                 le32_to_cpu(raw_super->log_sectorsize) !=
3498                         F2FS_MAX_LOG_SECTOR_SIZE) {
3499                 f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
3500                           le32_to_cpu(raw_super->log_sectors_per_block),
3501                           le32_to_cpu(raw_super->log_sectorsize));
3502                 return -EFSCORRUPTED;
3503         }
3504
3505         segment_count = le32_to_cpu(raw_super->segment_count);
3506         segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3507         segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3508         secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3509         total_sections = le32_to_cpu(raw_super->section_count);
3510
3511         /* blocks_per_seg should be 512, given the above check */
3512         blocks_per_seg = BIT(le32_to_cpu(raw_super->log_blocks_per_seg));
3513
3514         if (segment_count > F2FS_MAX_SEGMENT ||
3515                                 segment_count < F2FS_MIN_SEGMENTS) {
3516                 f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
3517                 return -EFSCORRUPTED;
3518         }
3519
3520         if (total_sections > segment_count_main || total_sections < 1 ||
3521                         segs_per_sec > segment_count || !segs_per_sec) {
3522                 f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
3523                           segment_count, total_sections, segs_per_sec);
3524                 return -EFSCORRUPTED;
3525         }
3526
3527         if (segment_count_main != total_sections * segs_per_sec) {
3528                 f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
3529                           segment_count_main, total_sections, segs_per_sec);
3530                 return -EFSCORRUPTED;
3531         }
3532
3533         if ((segment_count / segs_per_sec) < total_sections) {
3534                 f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
3535                           segment_count, segs_per_sec, total_sections);
3536                 return -EFSCORRUPTED;
3537         }
3538
3539         if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
3540                 f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
3541                           segment_count, le64_to_cpu(raw_super->block_count));
3542                 return -EFSCORRUPTED;
3543         }
3544
3545         if (RDEV(0).path[0]) {
3546                 block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
3547                 int i = 1;
3548
3549                 while (i < MAX_DEVICES && RDEV(i).path[0]) {
3550                         dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
3551                         i++;
3552                 }
3553                 if (segment_count != dev_seg_count) {
3554                         f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
3555                                         segment_count, dev_seg_count);
3556                         return -EFSCORRUPTED;
3557                 }
3558         } else {
3559                 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
3560                                         !bdev_is_zoned(sbi->sb->s_bdev)) {
3561                         f2fs_info(sbi, "Zoned block device path is missing");
3562                         return -EFSCORRUPTED;
3563                 }
3564         }
3565
3566         if (secs_per_zone > total_sections || !secs_per_zone) {
3567                 f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
3568                           secs_per_zone, total_sections);
3569                 return -EFSCORRUPTED;
3570         }
3571         if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
3572                         raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
3573                         (le32_to_cpu(raw_super->extension_count) +
3574                         raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
3575                 f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
3576                           le32_to_cpu(raw_super->extension_count),
3577                           raw_super->hot_ext_count,
3578                           F2FS_MAX_EXTENSION);
3579                 return -EFSCORRUPTED;
3580         }
3581
3582         if (le32_to_cpu(raw_super->cp_payload) >=
3583                                 (blocks_per_seg - F2FS_CP_PACKS -
3584                                 NR_CURSEG_PERSIST_TYPE)) {
3585                 f2fs_info(sbi, "Insane cp_payload (%u >= %u)",
3586                           le32_to_cpu(raw_super->cp_payload),
3587                           blocks_per_seg - F2FS_CP_PACKS -
3588                           NR_CURSEG_PERSIST_TYPE);
3589                 return -EFSCORRUPTED;
3590         }
3591
3592         /* check reserved ino info */
3593         if (le32_to_cpu(raw_super->node_ino) != 1 ||
3594                 le32_to_cpu(raw_super->meta_ino) != 2 ||
3595                 le32_to_cpu(raw_super->root_ino) != 3) {
3596                 f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
3597                           le32_to_cpu(raw_super->node_ino),
3598                           le32_to_cpu(raw_super->meta_ino),
3599                           le32_to_cpu(raw_super->root_ino));
3600                 return -EFSCORRUPTED;
3601         }
3602
3603         /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
3604         if (sanity_check_area_boundary(sbi, bh))
3605                 return -EFSCORRUPTED;
3606
3607         return 0;
3608 }
3609
3610 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
3611 {
3612         unsigned int total, fsmeta;
3613         struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3614         struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
3615         unsigned int ovp_segments, reserved_segments;
3616         unsigned int main_segs, blocks_per_seg;
3617         unsigned int sit_segs, nat_segs;
3618         unsigned int sit_bitmap_size, nat_bitmap_size;
3619         unsigned int log_blocks_per_seg;
3620         unsigned int segment_count_main;
3621         unsigned int cp_pack_start_sum, cp_payload;
3622         block_t user_block_count, valid_user_blocks;
3623         block_t avail_node_count, valid_node_count;
3624         unsigned int nat_blocks, nat_bits_bytes, nat_bits_blocks;
3625         int i, j;
3626
3627         total = le32_to_cpu(raw_super->segment_count);
3628         fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
3629         sit_segs = le32_to_cpu(raw_super->segment_count_sit);
3630         fsmeta += sit_segs;
3631         nat_segs = le32_to_cpu(raw_super->segment_count_nat);
3632         fsmeta += nat_segs;
3633         fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
3634         fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
3635
3636         if (unlikely(fsmeta >= total))
3637                 return 1;
3638
3639         ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
3640         reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
3641
3642         if (!f2fs_sb_has_readonly(sbi) &&
3643                         unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
3644                         ovp_segments == 0 || reserved_segments == 0)) {
3645                 f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
3646                 return 1;
3647         }
3648         user_block_count = le64_to_cpu(ckpt->user_block_count);
3649         segment_count_main = le32_to_cpu(raw_super->segment_count_main) +
3650                         (f2fs_sb_has_readonly(sbi) ? 1 : 0);
3651         log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3652         if (!user_block_count || user_block_count >=
3653                         segment_count_main << log_blocks_per_seg) {
3654                 f2fs_err(sbi, "Wrong user_block_count: %u",
3655                          user_block_count);
3656                 return 1;
3657         }
3658
3659         valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
3660         if (valid_user_blocks > user_block_count) {
3661                 f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
3662                          valid_user_blocks, user_block_count);
3663                 return 1;
3664         }
3665
3666         valid_node_count = le32_to_cpu(ckpt->valid_node_count);
3667         avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
3668         if (valid_node_count > avail_node_count) {
3669                 f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
3670                          valid_node_count, avail_node_count);
3671                 return 1;
3672         }
3673
3674         main_segs = le32_to_cpu(raw_super->segment_count_main);
3675         blocks_per_seg = BLKS_PER_SEG(sbi);
3676
3677         for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3678                 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
3679                         le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
3680                         return 1;
3681
3682                 if (f2fs_sb_has_readonly(sbi))
3683                         goto check_data;
3684
3685                 for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
3686                         if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3687                                 le32_to_cpu(ckpt->cur_node_segno[j])) {
3688                                 f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
3689                                          i, j,
3690                                          le32_to_cpu(ckpt->cur_node_segno[i]));
3691                                 return 1;
3692                         }
3693                 }
3694         }
3695 check_data:
3696         for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
3697                 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
3698                         le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
3699                         return 1;
3700
3701                 if (f2fs_sb_has_readonly(sbi))
3702                         goto skip_cross;
3703
3704                 for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3705                         if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3706                                 le32_to_cpu(ckpt->cur_data_segno[j])) {
3707                                 f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3708                                          i, j,
3709                                          le32_to_cpu(ckpt->cur_data_segno[i]));
3710                                 return 1;
3711                         }
3712                 }
3713         }
3714         for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3715                 for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3716                         if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3717                                 le32_to_cpu(ckpt->cur_data_segno[j])) {
3718                                 f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3719                                          i, j,
3720                                          le32_to_cpu(ckpt->cur_node_segno[i]));
3721                                 return 1;
3722                         }
3723                 }
3724         }
3725 skip_cross:
3726         sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3727         nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3728
3729         if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3730                 nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3731                 f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3732                          sit_bitmap_size, nat_bitmap_size);
3733                 return 1;
3734         }
3735
3736         cp_pack_start_sum = __start_sum_addr(sbi);
3737         cp_payload = __cp_payload(sbi);
3738         if (cp_pack_start_sum < cp_payload + 1 ||
3739                 cp_pack_start_sum > blocks_per_seg - 1 -
3740                         NR_CURSEG_PERSIST_TYPE) {
3741                 f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3742                          cp_pack_start_sum);
3743                 return 1;
3744         }
3745
3746         if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3747                 le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3748                 f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3749                           "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3750                           "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3751                           le32_to_cpu(ckpt->checksum_offset));
3752                 return 1;
3753         }
3754
3755         nat_blocks = nat_segs << log_blocks_per_seg;
3756         nat_bits_bytes = nat_blocks / BITS_PER_BYTE;
3757         nat_bits_blocks = F2FS_BLK_ALIGN((nat_bits_bytes << 1) + 8);
3758         if (__is_set_ckpt_flags(ckpt, CP_NAT_BITS_FLAG) &&
3759                 (cp_payload + F2FS_CP_PACKS +
3760                 NR_CURSEG_PERSIST_TYPE + nat_bits_blocks >= blocks_per_seg)) {
3761                 f2fs_warn(sbi, "Insane cp_payload: %u, nat_bits_blocks: %u)",
3762                           cp_payload, nat_bits_blocks);
3763                 return 1;
3764         }
3765
3766         if (unlikely(f2fs_cp_error(sbi))) {
3767                 f2fs_err(sbi, "A bug case: need to run fsck");
3768                 return 1;
3769         }
3770         return 0;
3771 }
3772
3773 static void init_sb_info(struct f2fs_sb_info *sbi)
3774 {
3775         struct f2fs_super_block *raw_super = sbi->raw_super;
3776         int i;
3777
3778         sbi->log_sectors_per_block =
3779                 le32_to_cpu(raw_super->log_sectors_per_block);
3780         sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3781         sbi->blocksize = BIT(sbi->log_blocksize);
3782         sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3783         sbi->blocks_per_seg = BIT(sbi->log_blocks_per_seg);
3784         sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3785         sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3786         sbi->total_sections = le32_to_cpu(raw_super->section_count);
3787         sbi->total_node_count = SEGS_TO_BLKS(sbi,
3788                         ((le32_to_cpu(raw_super->segment_count_nat) / 2) *
3789                         NAT_ENTRY_PER_BLOCK));
3790         F2FS_ROOT_INO(sbi) = le32_to_cpu(raw_super->root_ino);
3791         F2FS_NODE_INO(sbi) = le32_to_cpu(raw_super->node_ino);
3792         F2FS_META_INO(sbi) = le32_to_cpu(raw_super->meta_ino);
3793         sbi->cur_victim_sec = NULL_SECNO;
3794         sbi->gc_mode = GC_NORMAL;
3795         sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3796         sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3797         sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3798         sbi->migration_granularity = SEGS_PER_SEC(sbi);
3799         sbi->seq_file_ra_mul = MIN_RA_MUL;
3800         sbi->max_fragment_chunk = DEF_FRAGMENT_SIZE;
3801         sbi->max_fragment_hole = DEF_FRAGMENT_SIZE;
3802         spin_lock_init(&sbi->gc_remaining_trials_lock);
3803         atomic64_set(&sbi->current_atomic_write, 0);
3804
3805         sbi->dir_level = DEF_DIR_LEVEL;
3806         sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3807         sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3808         sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3809         sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3810         sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3811         sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3812                                 DEF_UMOUNT_DISCARD_TIMEOUT;
3813         clear_sbi_flag(sbi, SBI_NEED_FSCK);
3814
3815         for (i = 0; i < NR_COUNT_TYPE; i++)
3816                 atomic_set(&sbi->nr_pages[i], 0);
3817
3818         for (i = 0; i < META; i++)
3819                 atomic_set(&sbi->wb_sync_req[i], 0);
3820
3821         INIT_LIST_HEAD(&sbi->s_list);
3822         mutex_init(&sbi->umount_mutex);
3823         init_f2fs_rwsem(&sbi->io_order_lock);
3824         spin_lock_init(&sbi->cp_lock);
3825
3826         sbi->dirty_device = 0;
3827         spin_lock_init(&sbi->dev_lock);
3828
3829         init_f2fs_rwsem(&sbi->sb_lock);
3830         init_f2fs_rwsem(&sbi->pin_sem);
3831 }
3832
3833 static int init_percpu_info(struct f2fs_sb_info *sbi)
3834 {
3835         int err;
3836
3837         err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3838         if (err)
3839                 return err;
3840
3841         err = percpu_counter_init(&sbi->rf_node_block_count, 0, GFP_KERNEL);
3842         if (err)
3843                 goto err_valid_block;
3844
3845         err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3846                                                                 GFP_KERNEL);
3847         if (err)
3848                 goto err_node_block;
3849         return 0;
3850
3851 err_node_block:
3852         percpu_counter_destroy(&sbi->rf_node_block_count);
3853 err_valid_block:
3854         percpu_counter_destroy(&sbi->alloc_valid_block_count);
3855         return err;
3856 }
3857
3858 #ifdef CONFIG_BLK_DEV_ZONED
3859
3860 struct f2fs_report_zones_args {
3861         struct f2fs_sb_info *sbi;
3862         struct f2fs_dev_info *dev;
3863 };
3864
3865 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3866                               void *data)
3867 {
3868         struct f2fs_report_zones_args *rz_args = data;
3869         block_t unusable_blocks = (zone->len - zone->capacity) >>
3870                                         F2FS_LOG_SECTORS_PER_BLOCK;
3871
3872         if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3873                 return 0;
3874
3875         set_bit(idx, rz_args->dev->blkz_seq);
3876         if (!rz_args->sbi->unusable_blocks_per_sec) {
3877                 rz_args->sbi->unusable_blocks_per_sec = unusable_blocks;
3878                 return 0;
3879         }
3880         if (rz_args->sbi->unusable_blocks_per_sec != unusable_blocks) {
3881                 f2fs_err(rz_args->sbi, "F2FS supports single zone capacity\n");
3882                 return -EINVAL;
3883         }
3884         return 0;
3885 }
3886
3887 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3888 {
3889         struct block_device *bdev = FDEV(devi).bdev;
3890         sector_t nr_sectors = bdev_nr_sectors(bdev);
3891         struct f2fs_report_zones_args rep_zone_arg;
3892         u64 zone_sectors;
3893         unsigned int max_open_zones;
3894         int ret;
3895
3896         if (!f2fs_sb_has_blkzoned(sbi))
3897                 return 0;
3898
3899         if (bdev_is_zoned(FDEV(devi).bdev)) {
3900                 max_open_zones = bdev_max_open_zones(bdev);
3901                 if (max_open_zones && (max_open_zones < sbi->max_open_zones))
3902                         sbi->max_open_zones = max_open_zones;
3903                 if (sbi->max_open_zones < F2FS_OPTION(sbi).active_logs) {
3904                         f2fs_err(sbi,
3905                                 "zoned: max open zones %u is too small, need at least %u open zones",
3906                                 sbi->max_open_zones, F2FS_OPTION(sbi).active_logs);
3907                         return -EINVAL;
3908                 }
3909         }
3910
3911         zone_sectors = bdev_zone_sectors(bdev);
3912         if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3913                                 SECTOR_TO_BLOCK(zone_sectors))
3914                 return -EINVAL;
3915         sbi->blocks_per_blkz = SECTOR_TO_BLOCK(zone_sectors);
3916         FDEV(devi).nr_blkz = div_u64(SECTOR_TO_BLOCK(nr_sectors),
3917                                         sbi->blocks_per_blkz);
3918         if (nr_sectors & (zone_sectors - 1))
3919                 FDEV(devi).nr_blkz++;
3920
3921         FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3922                                         BITS_TO_LONGS(FDEV(devi).nr_blkz)
3923                                         * sizeof(unsigned long),
3924                                         GFP_KERNEL);
3925         if (!FDEV(devi).blkz_seq)
3926                 return -ENOMEM;
3927
3928         rep_zone_arg.sbi = sbi;
3929         rep_zone_arg.dev = &FDEV(devi);
3930
3931         ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3932                                   &rep_zone_arg);
3933         if (ret < 0)
3934                 return ret;
3935         return 0;
3936 }
3937 #endif
3938
3939 /*
3940  * Read f2fs raw super block.
3941  * Because we have two copies of super block, so read both of them
3942  * to get the first valid one. If any one of them is broken, we pass
3943  * them recovery flag back to the caller.
3944  */
3945 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3946                         struct f2fs_super_block **raw_super,
3947                         int *valid_super_block, int *recovery)
3948 {
3949         struct super_block *sb = sbi->sb;
3950         int block;
3951         struct buffer_head *bh;
3952         struct f2fs_super_block *super;
3953         int err = 0;
3954
3955         super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3956         if (!super)
3957                 return -ENOMEM;
3958
3959         for (block = 0; block < 2; block++) {
3960                 bh = sb_bread(sb, block);
3961                 if (!bh) {
3962                         f2fs_err(sbi, "Unable to read %dth superblock",
3963                                  block + 1);
3964                         err = -EIO;
3965                         *recovery = 1;
3966                         continue;
3967                 }
3968
3969                 /* sanity checking of raw super */
3970                 err = sanity_check_raw_super(sbi, bh);
3971                 if (err) {
3972                         f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3973                                  block + 1);
3974                         brelse(bh);
3975                         *recovery = 1;
3976                         continue;
3977                 }
3978
3979                 if (!*raw_super) {
3980                         memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3981                                                         sizeof(*super));
3982                         *valid_super_block = block;
3983                         *raw_super = super;
3984                 }
3985                 brelse(bh);
3986         }
3987
3988         /* No valid superblock */
3989         if (!*raw_super)
3990                 kfree(super);
3991         else
3992                 err = 0;
3993
3994         return err;
3995 }
3996
3997 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
3998 {
3999         struct buffer_head *bh;
4000         __u32 crc = 0;
4001         int err;
4002
4003         if ((recover && f2fs_readonly(sbi->sb)) ||
4004                                 f2fs_hw_is_readonly(sbi)) {
4005                 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
4006                 return -EROFS;
4007         }
4008
4009         /* we should update superblock crc here */
4010         if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
4011                 crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
4012                                 offsetof(struct f2fs_super_block, crc));
4013                 F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
4014         }
4015
4016         /* write back-up superblock first */
4017         bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
4018         if (!bh)
4019                 return -EIO;
4020         err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
4021         brelse(bh);
4022
4023         /* if we are in recovery path, skip writing valid superblock */
4024         if (recover || err)
4025                 return err;
4026
4027         /* write current valid superblock */
4028         bh = sb_bread(sbi->sb, sbi->valid_super_block);
4029         if (!bh)
4030                 return -EIO;
4031         err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
4032         brelse(bh);
4033         return err;
4034 }
4035
4036 static void save_stop_reason(struct f2fs_sb_info *sbi, unsigned char reason)
4037 {
4038         unsigned long flags;
4039
4040         spin_lock_irqsave(&sbi->error_lock, flags);
4041         if (sbi->stop_reason[reason] < GENMASK(BITS_PER_BYTE - 1, 0))
4042                 sbi->stop_reason[reason]++;
4043         spin_unlock_irqrestore(&sbi->error_lock, flags);
4044 }
4045
4046 static void f2fs_record_stop_reason(struct f2fs_sb_info *sbi)
4047 {
4048         struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
4049         unsigned long flags;
4050         int err;
4051
4052         f2fs_down_write(&sbi->sb_lock);
4053
4054         spin_lock_irqsave(&sbi->error_lock, flags);
4055         if (sbi->error_dirty) {
4056                 memcpy(F2FS_RAW_SUPER(sbi)->s_errors, sbi->errors,
4057                                                         MAX_F2FS_ERRORS);
4058                 sbi->error_dirty = false;
4059         }
4060         memcpy(raw_super->s_stop_reason, sbi->stop_reason, MAX_STOP_REASON);
4061         spin_unlock_irqrestore(&sbi->error_lock, flags);
4062
4063         err = f2fs_commit_super(sbi, false);
4064
4065         f2fs_up_write(&sbi->sb_lock);
4066         if (err)
4067                 f2fs_err_ratelimited(sbi,
4068                         "f2fs_commit_super fails to record stop_reason, err:%d",
4069                         err);
4070 }
4071
4072 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag)
4073 {
4074         unsigned long flags;
4075
4076         spin_lock_irqsave(&sbi->error_lock, flags);
4077         if (!test_bit(flag, (unsigned long *)sbi->errors)) {
4078                 set_bit(flag, (unsigned long *)sbi->errors);
4079                 sbi->error_dirty = true;
4080         }
4081         spin_unlock_irqrestore(&sbi->error_lock, flags);
4082 }
4083
4084 static bool f2fs_update_errors(struct f2fs_sb_info *sbi)
4085 {
4086         unsigned long flags;
4087         bool need_update = false;
4088
4089         spin_lock_irqsave(&sbi->error_lock, flags);
4090         if (sbi->error_dirty) {
4091                 memcpy(F2FS_RAW_SUPER(sbi)->s_errors, sbi->errors,
4092                                                         MAX_F2FS_ERRORS);
4093                 sbi->error_dirty = false;
4094                 need_update = true;
4095         }
4096         spin_unlock_irqrestore(&sbi->error_lock, flags);
4097
4098         return need_update;
4099 }
4100
4101 static void f2fs_record_errors(struct f2fs_sb_info *sbi, unsigned char error)
4102 {
4103         int err;
4104
4105         f2fs_down_write(&sbi->sb_lock);
4106
4107         if (!f2fs_update_errors(sbi))
4108                 goto out_unlock;
4109
4110         err = f2fs_commit_super(sbi, false);
4111         if (err)
4112                 f2fs_err_ratelimited(sbi,
4113                         "f2fs_commit_super fails to record errors:%u, err:%d",
4114                         error, err);
4115 out_unlock:
4116         f2fs_up_write(&sbi->sb_lock);
4117 }
4118
4119 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error)
4120 {
4121         f2fs_save_errors(sbi, error);
4122         f2fs_record_errors(sbi, error);
4123 }
4124
4125 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error)
4126 {
4127         f2fs_save_errors(sbi, error);
4128
4129         if (!sbi->error_dirty)
4130                 return;
4131         if (!test_bit(error, (unsigned long *)sbi->errors))
4132                 return;
4133         schedule_work(&sbi->s_error_work);
4134 }
4135
4136 static bool system_going_down(void)
4137 {
4138         return system_state == SYSTEM_HALT || system_state == SYSTEM_POWER_OFF
4139                 || system_state == SYSTEM_RESTART;
4140 }
4141
4142 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason,
4143                                                         bool irq_context)
4144 {
4145         struct super_block *sb = sbi->sb;
4146         bool shutdown = reason == STOP_CP_REASON_SHUTDOWN;
4147         bool continue_fs = !shutdown &&
4148                         F2FS_OPTION(sbi).errors == MOUNT_ERRORS_CONTINUE;
4149
4150         set_ckpt_flags(sbi, CP_ERROR_FLAG);
4151
4152         if (!f2fs_hw_is_readonly(sbi)) {
4153                 save_stop_reason(sbi, reason);
4154
4155                 if (irq_context && !shutdown)
4156                         schedule_work(&sbi->s_error_work);
4157                 else
4158                         f2fs_record_stop_reason(sbi);
4159         }
4160
4161         /*
4162          * We force ERRORS_RO behavior when system is rebooting. Otherwise we
4163          * could panic during 'reboot -f' as the underlying device got already
4164          * disabled.
4165          */
4166         if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_PANIC &&
4167                                 !shutdown && !system_going_down() &&
4168                                 !is_sbi_flag_set(sbi, SBI_IS_SHUTDOWN))
4169                 panic("F2FS-fs (device %s): panic forced after error\n",
4170                                                         sb->s_id);
4171
4172         if (shutdown)
4173                 set_sbi_flag(sbi, SBI_IS_SHUTDOWN);
4174
4175         /*
4176          * Continue filesystem operators if errors=continue. Should not set
4177          * RO by shutdown, since RO bypasses thaw_super which can hang the
4178          * system.
4179          */
4180         if (continue_fs || f2fs_readonly(sb) || shutdown) {
4181                 f2fs_warn(sbi, "Stopped filesystem due to reason: %d", reason);
4182                 return;
4183         }
4184
4185         f2fs_warn(sbi, "Remounting filesystem read-only");
4186         /*
4187          * Make sure updated value of ->s_mount_flags will be visible before
4188          * ->s_flags update
4189          */
4190         smp_wmb();
4191         sb->s_flags |= SB_RDONLY;
4192 }
4193
4194 static void f2fs_record_error_work(struct work_struct *work)
4195 {
4196         struct f2fs_sb_info *sbi = container_of(work,
4197                                         struct f2fs_sb_info, s_error_work);
4198
4199         f2fs_record_stop_reason(sbi);
4200 }
4201
4202 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
4203 {
4204         struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
4205         unsigned int max_devices = MAX_DEVICES;
4206         unsigned int logical_blksize;
4207         blk_mode_t mode = sb_open_mode(sbi->sb->s_flags);
4208         int i;
4209
4210         /* Initialize single device information */
4211         if (!RDEV(0).path[0]) {
4212                 if (!bdev_is_zoned(sbi->sb->s_bdev))
4213                         return 0;
4214                 max_devices = 1;
4215         }
4216
4217         /*
4218          * Initialize multiple devices information, or single
4219          * zoned block device information.
4220          */
4221         sbi->devs = f2fs_kzalloc(sbi,
4222                                  array_size(max_devices,
4223                                             sizeof(struct f2fs_dev_info)),
4224                                  GFP_KERNEL);
4225         if (!sbi->devs)
4226                 return -ENOMEM;
4227
4228         logical_blksize = bdev_logical_block_size(sbi->sb->s_bdev);
4229         sbi->aligned_blksize = true;
4230 #ifdef CONFIG_BLK_DEV_ZONED
4231         sbi->max_open_zones = UINT_MAX;
4232 #endif
4233
4234         for (i = 0; i < max_devices; i++) {
4235                 if (i == 0)
4236                         FDEV(0).bdev_file = sbi->sb->s_bdev_file;
4237                 else if (!RDEV(i).path[0])
4238                         break;
4239
4240                 if (max_devices > 1) {
4241                         /* Multi-device mount */
4242                         memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
4243                         FDEV(i).total_segments =
4244                                 le32_to_cpu(RDEV(i).total_segments);
4245                         if (i == 0) {
4246                                 FDEV(i).start_blk = 0;
4247                                 FDEV(i).end_blk = FDEV(i).start_blk +
4248                                         SEGS_TO_BLKS(sbi,
4249                                         FDEV(i).total_segments) - 1 +
4250                                         le32_to_cpu(raw_super->segment0_blkaddr);
4251                         } else {
4252                                 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
4253                                 FDEV(i).end_blk = FDEV(i).start_blk +
4254                                                 SEGS_TO_BLKS(sbi,
4255                                                 FDEV(i).total_segments) - 1;
4256                                 FDEV(i).bdev_file = bdev_file_open_by_path(
4257                                         FDEV(i).path, mode, sbi->sb, NULL);
4258                         }
4259                 }
4260                 if (IS_ERR(FDEV(i).bdev_file))
4261                         return PTR_ERR(FDEV(i).bdev_file);
4262
4263                 FDEV(i).bdev = file_bdev(FDEV(i).bdev_file);
4264                 /* to release errored devices */
4265                 sbi->s_ndevs = i + 1;
4266
4267                 if (logical_blksize != bdev_logical_block_size(FDEV(i).bdev))
4268                         sbi->aligned_blksize = false;
4269
4270 #ifdef CONFIG_BLK_DEV_ZONED
4271                 if (bdev_is_zoned(FDEV(i).bdev)) {
4272                         if (!f2fs_sb_has_blkzoned(sbi)) {
4273                                 f2fs_err(sbi, "Zoned block device feature not enabled");
4274                                 return -EINVAL;
4275                         }
4276                         if (init_blkz_info(sbi, i)) {
4277                                 f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
4278                                 return -EINVAL;
4279                         }
4280                         if (max_devices == 1)
4281                                 break;
4282                         f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: Host-managed)",
4283                                   i, FDEV(i).path,
4284                                   FDEV(i).total_segments,
4285                                   FDEV(i).start_blk, FDEV(i).end_blk);
4286                         continue;
4287                 }
4288 #endif
4289                 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
4290                           i, FDEV(i).path,
4291                           FDEV(i).total_segments,
4292                           FDEV(i).start_blk, FDEV(i).end_blk);
4293         }
4294         return 0;
4295 }
4296
4297 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
4298 {
4299 #if IS_ENABLED(CONFIG_UNICODE)
4300         if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
4301                 const struct f2fs_sb_encodings *encoding_info;
4302                 struct unicode_map *encoding;
4303                 __u16 encoding_flags;
4304
4305                 encoding_info = f2fs_sb_read_encoding(sbi->raw_super);
4306                 if (!encoding_info) {
4307                         f2fs_err(sbi,
4308                                  "Encoding requested by superblock is unknown");
4309                         return -EINVAL;
4310                 }
4311
4312                 encoding_flags = le16_to_cpu(sbi->raw_super->s_encoding_flags);
4313                 encoding = utf8_load(encoding_info->version);
4314                 if (IS_ERR(encoding)) {
4315                         f2fs_err(sbi,
4316                                  "can't mount with superblock charset: %s-%u.%u.%u "
4317                                  "not supported by the kernel. flags: 0x%x.",
4318                                  encoding_info->name,
4319                                  unicode_major(encoding_info->version),
4320                                  unicode_minor(encoding_info->version),
4321                                  unicode_rev(encoding_info->version),
4322                                  encoding_flags);
4323                         return PTR_ERR(encoding);
4324                 }
4325                 f2fs_info(sbi, "Using encoding defined by superblock: "
4326                          "%s-%u.%u.%u with flags 0x%hx", encoding_info->name,
4327                          unicode_major(encoding_info->version),
4328                          unicode_minor(encoding_info->version),
4329                          unicode_rev(encoding_info->version),
4330                          encoding_flags);
4331
4332                 sbi->sb->s_encoding = encoding;
4333                 sbi->sb->s_encoding_flags = encoding_flags;
4334         }
4335 #else
4336         if (f2fs_sb_has_casefold(sbi)) {
4337                 f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
4338                 return -EINVAL;
4339         }
4340 #endif
4341         return 0;
4342 }
4343
4344 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
4345 {
4346         /* adjust parameters according to the volume size */
4347         if (MAIN_SEGS(sbi) <= SMALL_VOLUME_SEGMENTS) {
4348                 if (f2fs_block_unit_discard(sbi))
4349                         SM_I(sbi)->dcc_info->discard_granularity =
4350                                                 MIN_DISCARD_GRANULARITY;
4351                 if (!f2fs_lfs_mode(sbi))
4352                         SM_I(sbi)->ipu_policy = BIT(F2FS_IPU_FORCE) |
4353                                                 BIT(F2FS_IPU_HONOR_OPU_WRITE);
4354         }
4355
4356         sbi->readdir_ra = true;
4357 }
4358
4359 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
4360 {
4361         struct f2fs_sb_info *sbi;
4362         struct f2fs_super_block *raw_super;
4363         struct inode *root;
4364         int err;
4365         bool skip_recovery = false, need_fsck = false;
4366         char *options = NULL;
4367         int recovery, i, valid_super_block;
4368         struct curseg_info *seg_i;
4369         int retry_cnt = 1;
4370 #ifdef CONFIG_QUOTA
4371         bool quota_enabled = false;
4372 #endif
4373
4374 try_onemore:
4375         err = -EINVAL;
4376         raw_super = NULL;
4377         valid_super_block = -1;
4378         recovery = 0;
4379
4380         /* allocate memory for f2fs-specific super block info */
4381         sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
4382         if (!sbi)
4383                 return -ENOMEM;
4384
4385         sbi->sb = sb;
4386
4387         /* initialize locks within allocated memory */
4388         init_f2fs_rwsem(&sbi->gc_lock);
4389         mutex_init(&sbi->writepages);
4390         init_f2fs_rwsem(&sbi->cp_global_sem);
4391         init_f2fs_rwsem(&sbi->node_write);
4392         init_f2fs_rwsem(&sbi->node_change);
4393         spin_lock_init(&sbi->stat_lock);
4394         init_f2fs_rwsem(&sbi->cp_rwsem);
4395         init_f2fs_rwsem(&sbi->quota_sem);
4396         init_waitqueue_head(&sbi->cp_wait);
4397         spin_lock_init(&sbi->error_lock);
4398
4399         for (i = 0; i < NR_INODE_TYPE; i++) {
4400                 INIT_LIST_HEAD(&sbi->inode_list[i]);
4401                 spin_lock_init(&sbi->inode_lock[i]);
4402         }
4403         mutex_init(&sbi->flush_lock);
4404
4405         /* Load the checksum driver */
4406         sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
4407         if (IS_ERR(sbi->s_chksum_driver)) {
4408                 f2fs_err(sbi, "Cannot load crc32 driver.");
4409                 err = PTR_ERR(sbi->s_chksum_driver);
4410                 sbi->s_chksum_driver = NULL;
4411                 goto free_sbi;
4412         }
4413
4414         /* set a block size */
4415         if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
4416                 f2fs_err(sbi, "unable to set blocksize");
4417                 goto free_sbi;
4418         }
4419
4420         err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
4421                                                                 &recovery);
4422         if (err)
4423                 goto free_sbi;
4424
4425         sb->s_fs_info = sbi;
4426         sbi->raw_super = raw_super;
4427
4428         INIT_WORK(&sbi->s_error_work, f2fs_record_error_work);
4429         memcpy(sbi->errors, raw_super->s_errors, MAX_F2FS_ERRORS);
4430         memcpy(sbi->stop_reason, raw_super->s_stop_reason, MAX_STOP_REASON);
4431
4432         /* precompute checksum seed for metadata */
4433         if (f2fs_sb_has_inode_chksum(sbi))
4434                 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
4435                                                 sizeof(raw_super->uuid));
4436
4437         default_options(sbi, false);
4438         /* parse mount options */
4439         options = kstrdup((const char *)data, GFP_KERNEL);
4440         if (data && !options) {
4441                 err = -ENOMEM;
4442                 goto free_sb_buf;
4443         }
4444
4445         err = parse_options(sb, options, false);
4446         if (err)
4447                 goto free_options;
4448
4449         sb->s_maxbytes = max_file_blocks(NULL) <<
4450                                 le32_to_cpu(raw_super->log_blocksize);
4451         sb->s_max_links = F2FS_LINK_MAX;
4452
4453         err = f2fs_setup_casefold(sbi);
4454         if (err)
4455                 goto free_options;
4456
4457 #ifdef CONFIG_QUOTA
4458         sb->dq_op = &f2fs_quota_operations;
4459         sb->s_qcop = &f2fs_quotactl_ops;
4460         sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
4461
4462         if (f2fs_sb_has_quota_ino(sbi)) {
4463                 for (i = 0; i < MAXQUOTAS; i++) {
4464                         if (f2fs_qf_ino(sbi->sb, i))
4465                                 sbi->nquota_files++;
4466                 }
4467         }
4468 #endif
4469
4470         sb->s_op = &f2fs_sops;
4471 #ifdef CONFIG_FS_ENCRYPTION
4472         sb->s_cop = &f2fs_cryptops;
4473 #endif
4474 #ifdef CONFIG_FS_VERITY
4475         sb->s_vop = &f2fs_verityops;
4476 #endif
4477         sb->s_xattr = f2fs_xattr_handlers;
4478         sb->s_export_op = &f2fs_export_ops;
4479         sb->s_magic = F2FS_SUPER_MAGIC;
4480         sb->s_time_gran = 1;
4481         sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
4482                 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
4483         super_set_uuid(sb, (void *) raw_super->uuid, sizeof(raw_super->uuid));
4484         sb->s_iflags |= SB_I_CGROUPWB;
4485
4486         /* init f2fs-specific super block info */
4487         sbi->valid_super_block = valid_super_block;
4488
4489         /* disallow all the data/node/meta page writes */
4490         set_sbi_flag(sbi, SBI_POR_DOING);
4491
4492         err = f2fs_init_write_merge_io(sbi);
4493         if (err)
4494                 goto free_bio_info;
4495
4496         init_sb_info(sbi);
4497
4498         err = f2fs_init_iostat(sbi);
4499         if (err)
4500                 goto free_bio_info;
4501
4502         err = init_percpu_info(sbi);
4503         if (err)
4504                 goto free_iostat;
4505
4506         /* init per sbi slab cache */
4507         err = f2fs_init_xattr_caches(sbi);
4508         if (err)
4509                 goto free_percpu;
4510         err = f2fs_init_page_array_cache(sbi);
4511         if (err)
4512                 goto free_xattr_cache;
4513
4514         /* get an inode for meta space */
4515         sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
4516         if (IS_ERR(sbi->meta_inode)) {
4517                 f2fs_err(sbi, "Failed to read F2FS meta data inode");
4518                 err = PTR_ERR(sbi->meta_inode);
4519                 goto free_page_array_cache;
4520         }
4521
4522         err = f2fs_get_valid_checkpoint(sbi);
4523         if (err) {
4524                 f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
4525                 goto free_meta_inode;
4526         }
4527
4528         if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
4529                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
4530         if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
4531                 set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4532                 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
4533         }
4534
4535         if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
4536                 set_sbi_flag(sbi, SBI_NEED_FSCK);
4537
4538         /* Initialize device list */
4539         err = f2fs_scan_devices(sbi);
4540         if (err) {
4541                 f2fs_err(sbi, "Failed to find devices");
4542                 goto free_devices;
4543         }
4544
4545         err = f2fs_init_post_read_wq(sbi);
4546         if (err) {
4547                 f2fs_err(sbi, "Failed to initialize post read workqueue");
4548                 goto free_devices;
4549         }
4550
4551         sbi->total_valid_node_count =
4552                                 le32_to_cpu(sbi->ckpt->valid_node_count);
4553         percpu_counter_set(&sbi->total_valid_inode_count,
4554                                 le32_to_cpu(sbi->ckpt->valid_inode_count));
4555         sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
4556         sbi->total_valid_block_count =
4557                                 le64_to_cpu(sbi->ckpt->valid_block_count);
4558         sbi->last_valid_block_count = sbi->total_valid_block_count;
4559         sbi->reserved_blocks = 0;
4560         sbi->current_reserved_blocks = 0;
4561         limit_reserve_root(sbi);
4562         adjust_unusable_cap_perc(sbi);
4563
4564         f2fs_init_extent_cache_info(sbi);
4565
4566         f2fs_init_ino_entry_info(sbi);
4567
4568         f2fs_init_fsync_node_info(sbi);
4569
4570         /* setup checkpoint request control and start checkpoint issue thread */
4571         f2fs_init_ckpt_req_control(sbi);
4572         if (!f2fs_readonly(sb) && !test_opt(sbi, DISABLE_CHECKPOINT) &&
4573                         test_opt(sbi, MERGE_CHECKPOINT)) {
4574                 err = f2fs_start_ckpt_thread(sbi);
4575                 if (err) {
4576                         f2fs_err(sbi,
4577                             "Failed to start F2FS issue_checkpoint_thread (%d)",
4578                             err);
4579                         goto stop_ckpt_thread;
4580                 }
4581         }
4582
4583         /* setup f2fs internal modules */
4584         err = f2fs_build_segment_manager(sbi);
4585         if (err) {
4586                 f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
4587                          err);
4588                 goto free_sm;
4589         }
4590         err = f2fs_build_node_manager(sbi);
4591         if (err) {
4592                 f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
4593                          err);
4594                 goto free_nm;
4595         }
4596
4597         /* For write statistics */
4598         sbi->sectors_written_start = f2fs_get_sectors_written(sbi);
4599
4600         /* Read accumulated write IO statistics if exists */
4601         seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
4602         if (__exist_node_summaries(sbi))
4603                 sbi->kbytes_written =
4604                         le64_to_cpu(seg_i->journal->info.kbytes_written);
4605
4606         f2fs_build_gc_manager(sbi);
4607
4608         err = f2fs_build_stats(sbi);
4609         if (err)
4610                 goto free_nm;
4611
4612         /* get an inode for node space */
4613         sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
4614         if (IS_ERR(sbi->node_inode)) {
4615                 f2fs_err(sbi, "Failed to read node inode");
4616                 err = PTR_ERR(sbi->node_inode);
4617                 goto free_stats;
4618         }
4619
4620         /* read root inode and dentry */
4621         root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
4622         if (IS_ERR(root)) {
4623                 f2fs_err(sbi, "Failed to read root inode");
4624                 err = PTR_ERR(root);
4625                 goto free_node_inode;
4626         }
4627         if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
4628                         !root->i_size || !root->i_nlink) {
4629                 iput(root);
4630                 err = -EINVAL;
4631                 goto free_node_inode;
4632         }
4633
4634         generic_set_sb_d_ops(sb);
4635         sb->s_root = d_make_root(root); /* allocate root dentry */
4636         if (!sb->s_root) {
4637                 err = -ENOMEM;
4638                 goto free_node_inode;
4639         }
4640
4641         err = f2fs_init_compress_inode(sbi);
4642         if (err)
4643                 goto free_root_inode;
4644
4645         err = f2fs_register_sysfs(sbi);
4646         if (err)
4647                 goto free_compress_inode;
4648
4649 #ifdef CONFIG_QUOTA
4650         /* Enable quota usage during mount */
4651         if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
4652                 err = f2fs_enable_quotas(sb);
4653                 if (err)
4654                         f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
4655         }
4656
4657         quota_enabled = f2fs_recover_quota_begin(sbi);
4658 #endif
4659         /* if there are any orphan inodes, free them */
4660         err = f2fs_recover_orphan_inodes(sbi);
4661         if (err)
4662                 goto free_meta;
4663
4664         if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
4665                 goto reset_checkpoint;
4666
4667         /* recover fsynced data */
4668         if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
4669                         !test_opt(sbi, NORECOVERY)) {
4670                 /*
4671                  * mount should be failed, when device has readonly mode, and
4672                  * previous checkpoint was not done by clean system shutdown.
4673                  */
4674                 if (f2fs_hw_is_readonly(sbi)) {
4675                         if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4676                                 err = f2fs_recover_fsync_data(sbi, true);
4677                                 if (err > 0) {
4678                                         err = -EROFS;
4679                                         f2fs_err(sbi, "Need to recover fsync data, but "
4680                                                 "write access unavailable, please try "
4681                                                 "mount w/ disable_roll_forward or norecovery");
4682                                 }
4683                                 if (err < 0)
4684                                         goto free_meta;
4685                         }
4686                         f2fs_info(sbi, "write access unavailable, skipping recovery");
4687                         goto reset_checkpoint;
4688                 }
4689
4690                 if (need_fsck)
4691                         set_sbi_flag(sbi, SBI_NEED_FSCK);
4692
4693                 if (skip_recovery)
4694                         goto reset_checkpoint;
4695
4696                 err = f2fs_recover_fsync_data(sbi, false);
4697                 if (err < 0) {
4698                         if (err != -ENOMEM)
4699                                 skip_recovery = true;
4700                         need_fsck = true;
4701                         f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
4702                                  err);
4703                         goto free_meta;
4704                 }
4705         } else {
4706                 err = f2fs_recover_fsync_data(sbi, true);
4707
4708                 if (!f2fs_readonly(sb) && err > 0) {
4709                         err = -EINVAL;
4710                         f2fs_err(sbi, "Need to recover fsync data");
4711                         goto free_meta;
4712                 }
4713         }
4714
4715 #ifdef CONFIG_QUOTA
4716         f2fs_recover_quota_end(sbi, quota_enabled);
4717 #endif
4718 reset_checkpoint:
4719         /*
4720          * If the f2fs is not readonly and fsync data recovery succeeds,
4721          * check zoned block devices' write pointer consistency.
4722          */
4723         if (f2fs_sb_has_blkzoned(sbi) && !f2fs_readonly(sb)) {
4724                 int err2;
4725
4726                 f2fs_notice(sbi, "Checking entire write pointers");
4727                 err2 = f2fs_check_write_pointer(sbi);
4728                 if (err2)
4729                         err = err2;
4730         }
4731         if (err)
4732                 goto free_meta;
4733
4734         err = f2fs_init_inmem_curseg(sbi);
4735         if (err)
4736                 goto sync_free_meta;
4737
4738         /* f2fs_recover_fsync_data() cleared this already */
4739         clear_sbi_flag(sbi, SBI_POR_DOING);
4740
4741         if (test_opt(sbi, DISABLE_CHECKPOINT)) {
4742                 err = f2fs_disable_checkpoint(sbi);
4743                 if (err)
4744                         goto sync_free_meta;
4745         } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
4746                 f2fs_enable_checkpoint(sbi);
4747         }
4748
4749         /*
4750          * If filesystem is not mounted as read-only then
4751          * do start the gc_thread.
4752          */
4753         if ((F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF ||
4754                 test_opt(sbi, GC_MERGE)) && !f2fs_readonly(sb)) {
4755                 /* After POR, we can run background GC thread.*/
4756                 err = f2fs_start_gc_thread(sbi);
4757                 if (err)
4758                         goto sync_free_meta;
4759         }
4760         kvfree(options);
4761
4762         /* recover broken superblock */
4763         if (recovery) {
4764                 err = f2fs_commit_super(sbi, true);
4765                 f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
4766                           sbi->valid_super_block ? 1 : 2, err);
4767         }
4768
4769         f2fs_join_shrinker(sbi);
4770
4771         f2fs_tuning_parameters(sbi);
4772
4773         f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
4774                     cur_cp_version(F2FS_CKPT(sbi)));
4775         f2fs_update_time(sbi, CP_TIME);
4776         f2fs_update_time(sbi, REQ_TIME);
4777         clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4778         return 0;
4779
4780 sync_free_meta:
4781         /* safe to flush all the data */
4782         sync_filesystem(sbi->sb);
4783         retry_cnt = 0;
4784
4785 free_meta:
4786 #ifdef CONFIG_QUOTA
4787         f2fs_truncate_quota_inode_pages(sb);
4788         if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
4789                 f2fs_quota_off_umount(sbi->sb);
4790 #endif
4791         /*
4792          * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
4793          * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
4794          * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
4795          * falls into an infinite loop in f2fs_sync_meta_pages().
4796          */
4797         truncate_inode_pages_final(META_MAPPING(sbi));
4798         /* evict some inodes being cached by GC */
4799         evict_inodes(sb);
4800         f2fs_unregister_sysfs(sbi);
4801 free_compress_inode:
4802         f2fs_destroy_compress_inode(sbi);
4803 free_root_inode:
4804         dput(sb->s_root);
4805         sb->s_root = NULL;
4806 free_node_inode:
4807         f2fs_release_ino_entry(sbi, true);
4808         truncate_inode_pages_final(NODE_MAPPING(sbi));
4809         iput(sbi->node_inode);
4810         sbi->node_inode = NULL;
4811 free_stats:
4812         f2fs_destroy_stats(sbi);
4813 free_nm:
4814         /* stop discard thread before destroying node manager */
4815         f2fs_stop_discard_thread(sbi);
4816         f2fs_destroy_node_manager(sbi);
4817 free_sm:
4818         f2fs_destroy_segment_manager(sbi);
4819 stop_ckpt_thread:
4820         f2fs_stop_ckpt_thread(sbi);
4821         /* flush s_error_work before sbi destroy */
4822         flush_work(&sbi->s_error_work);
4823         f2fs_destroy_post_read_wq(sbi);
4824 free_devices:
4825         destroy_device_list(sbi);
4826         kvfree(sbi->ckpt);
4827 free_meta_inode:
4828         make_bad_inode(sbi->meta_inode);
4829         iput(sbi->meta_inode);
4830         sbi->meta_inode = NULL;
4831 free_page_array_cache:
4832         f2fs_destroy_page_array_cache(sbi);
4833 free_xattr_cache:
4834         f2fs_destroy_xattr_caches(sbi);
4835 free_percpu:
4836         destroy_percpu_info(sbi);
4837 free_iostat:
4838         f2fs_destroy_iostat(sbi);
4839 free_bio_info:
4840         for (i = 0; i < NR_PAGE_TYPE; i++)
4841                 kvfree(sbi->write_io[i]);
4842
4843 #if IS_ENABLED(CONFIG_UNICODE)
4844         utf8_unload(sb->s_encoding);
4845         sb->s_encoding = NULL;
4846 #endif
4847 free_options:
4848 #ifdef CONFIG_QUOTA
4849         for (i = 0; i < MAXQUOTAS; i++)
4850                 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
4851 #endif
4852         fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
4853         kvfree(options);
4854 free_sb_buf:
4855         kfree(raw_super);
4856 free_sbi:
4857         if (sbi->s_chksum_driver)
4858                 crypto_free_shash(sbi->s_chksum_driver);
4859         kfree(sbi);
4860         sb->s_fs_info = NULL;
4861
4862         /* give only one another chance */
4863         if (retry_cnt > 0 && skip_recovery) {
4864                 retry_cnt--;
4865                 shrink_dcache_sb(sb);
4866                 goto try_onemore;
4867         }
4868         return err;
4869 }
4870
4871 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
4872                         const char *dev_name, void *data)
4873 {
4874         return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
4875 }
4876
4877 static void kill_f2fs_super(struct super_block *sb)
4878 {
4879         struct f2fs_sb_info *sbi = F2FS_SB(sb);
4880
4881         if (sb->s_root) {
4882                 set_sbi_flag(sbi, SBI_IS_CLOSE);
4883                 f2fs_stop_gc_thread(sbi);
4884                 f2fs_stop_discard_thread(sbi);
4885
4886 #ifdef CONFIG_F2FS_FS_COMPRESSION
4887                 /*
4888                  * latter evict_inode() can bypass checking and invalidating
4889                  * compress inode cache.
4890                  */
4891                 if (test_opt(sbi, COMPRESS_CACHE))
4892                         truncate_inode_pages_final(COMPRESS_MAPPING(sbi));
4893 #endif
4894
4895                 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
4896                                 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4897                         struct cp_control cpc = {
4898                                 .reason = CP_UMOUNT,
4899                         };
4900                         stat_inc_cp_call_count(sbi, TOTAL_CALL);
4901                         f2fs_write_checkpoint(sbi, &cpc);
4902                 }
4903
4904                 if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
4905                         sb->s_flags &= ~SB_RDONLY;
4906         }
4907         kill_block_super(sb);
4908         /* Release block devices last, after fscrypt_destroy_keyring(). */
4909         if (sbi) {
4910                 destroy_device_list(sbi);
4911                 kfree(sbi);
4912                 sb->s_fs_info = NULL;
4913         }
4914 }
4915
4916 static struct file_system_type f2fs_fs_type = {
4917         .owner          = THIS_MODULE,
4918         .name           = "f2fs",
4919         .mount          = f2fs_mount,
4920         .kill_sb        = kill_f2fs_super,
4921         .fs_flags       = FS_REQUIRES_DEV | FS_ALLOW_IDMAP,
4922 };
4923 MODULE_ALIAS_FS("f2fs");
4924
4925 static int __init init_inodecache(void)
4926 {
4927         f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
4928                         sizeof(struct f2fs_inode_info), 0,
4929                         SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
4930         return f2fs_inode_cachep ? 0 : -ENOMEM;
4931 }
4932
4933 static void destroy_inodecache(void)
4934 {
4935         /*
4936          * Make sure all delayed rcu free inodes are flushed before we
4937          * destroy cache.
4938          */
4939         rcu_barrier();
4940         kmem_cache_destroy(f2fs_inode_cachep);
4941 }
4942
4943 static int __init init_f2fs_fs(void)
4944 {
4945         int err;
4946
4947         err = init_inodecache();
4948         if (err)
4949                 goto fail;
4950         err = f2fs_create_node_manager_caches();
4951         if (err)
4952                 goto free_inodecache;
4953         err = f2fs_create_segment_manager_caches();
4954         if (err)
4955                 goto free_node_manager_caches;
4956         err = f2fs_create_checkpoint_caches();
4957         if (err)
4958                 goto free_segment_manager_caches;
4959         err = f2fs_create_recovery_cache();
4960         if (err)
4961                 goto free_checkpoint_caches;
4962         err = f2fs_create_extent_cache();
4963         if (err)
4964                 goto free_recovery_cache;
4965         err = f2fs_create_garbage_collection_cache();
4966         if (err)
4967                 goto free_extent_cache;
4968         err = f2fs_init_sysfs();
4969         if (err)
4970                 goto free_garbage_collection_cache;
4971         err = f2fs_init_shrinker();
4972         if (err)
4973                 goto free_sysfs;
4974         err = register_filesystem(&f2fs_fs_type);
4975         if (err)
4976                 goto free_shrinker;
4977         f2fs_create_root_stats();
4978         err = f2fs_init_post_read_processing();
4979         if (err)
4980                 goto free_root_stats;
4981         err = f2fs_init_iostat_processing();
4982         if (err)
4983                 goto free_post_read;
4984         err = f2fs_init_bio_entry_cache();
4985         if (err)
4986                 goto free_iostat;
4987         err = f2fs_init_bioset();
4988         if (err)
4989                 goto free_bio_entry_cache;
4990         err = f2fs_init_compress_mempool();
4991         if (err)
4992                 goto free_bioset;
4993         err = f2fs_init_compress_cache();
4994         if (err)
4995                 goto free_compress_mempool;
4996         err = f2fs_create_casefold_cache();
4997         if (err)
4998                 goto free_compress_cache;
4999         return 0;
5000 free_compress_cache:
5001         f2fs_destroy_compress_cache();
5002 free_compress_mempool:
5003         f2fs_destroy_compress_mempool();
5004 free_bioset:
5005         f2fs_destroy_bioset();
5006 free_bio_entry_cache:
5007         f2fs_destroy_bio_entry_cache();
5008 free_iostat:
5009         f2fs_destroy_iostat_processing();
5010 free_post_read:
5011         f2fs_destroy_post_read_processing();
5012 free_root_stats:
5013         f2fs_destroy_root_stats();
5014         unregister_filesystem(&f2fs_fs_type);
5015 free_shrinker:
5016         f2fs_exit_shrinker();
5017 free_sysfs:
5018         f2fs_exit_sysfs();
5019 free_garbage_collection_cache:
5020         f2fs_destroy_garbage_collection_cache();
5021 free_extent_cache:
5022         f2fs_destroy_extent_cache();
5023 free_recovery_cache:
5024         f2fs_destroy_recovery_cache();
5025 free_checkpoint_caches:
5026         f2fs_destroy_checkpoint_caches();
5027 free_segment_manager_caches:
5028         f2fs_destroy_segment_manager_caches();
5029 free_node_manager_caches:
5030         f2fs_destroy_node_manager_caches();
5031 free_inodecache:
5032         destroy_inodecache();
5033 fail:
5034         return err;
5035 }
5036
5037 static void __exit exit_f2fs_fs(void)
5038 {
5039         f2fs_destroy_casefold_cache();
5040         f2fs_destroy_compress_cache();
5041         f2fs_destroy_compress_mempool();
5042         f2fs_destroy_bioset();
5043         f2fs_destroy_bio_entry_cache();
5044         f2fs_destroy_iostat_processing();
5045         f2fs_destroy_post_read_processing();
5046         f2fs_destroy_root_stats();
5047         unregister_filesystem(&f2fs_fs_type);
5048         f2fs_exit_shrinker();
5049         f2fs_exit_sysfs();
5050         f2fs_destroy_garbage_collection_cache();
5051         f2fs_destroy_extent_cache();
5052         f2fs_destroy_recovery_cache();
5053         f2fs_destroy_checkpoint_caches();
5054         f2fs_destroy_segment_manager_caches();
5055         f2fs_destroy_node_manager_caches();
5056         destroy_inodecache();
5057 }
5058
5059 module_init(init_f2fs_fs)
5060 module_exit(exit_f2fs_fs)
5061
5062 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
5063 MODULE_DESCRIPTION("Flash Friendly File System");
5064 MODULE_LICENSE("GPL");
5065 MODULE_SOFTDEP("pre: crc32");
5066