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]         = "invalid blkaddr",
  65 };
  66
  67 void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
  68                                                         unsigned int type)
  69 {
  70         struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
  71
  72         if (rate) {
  73                 atomic_set(&ffi->inject_ops, 0);
  74                 ffi->inject_rate = rate;
  75         }
  76
  77         if (type)
  78                 ffi->inject_type = type;
  79
  80         if (!rate && !type)
  81                 memset(ffi, 0, sizeof(struct f2fs_fault_info));
  82 }
  83 #endif
  84
  85 /* f2fs-wide shrinker description */
  86 static struct shrinker f2fs_shrinker_info = {
  87         .scan_objects = f2fs_shrink_scan,
  88         .count_objects = f2fs_shrink_count,
  89         .seeks = DEFAULT_SEEKS,
  90 };
  91
  92 enum {
  93         Opt_gc_background,
  94         Opt_disable_roll_forward,
  95         Opt_norecovery,
  96         Opt_discard,
  97         Opt_nodiscard,
  98         Opt_noheap,
  99         Opt_heap,
 100         Opt_user_xattr,
 101         Opt_nouser_xattr,
 102         Opt_acl,
 103         Opt_noacl,
 104         Opt_active_logs,
 105         Opt_disable_ext_identify,
 106         Opt_inline_xattr,
 107         Opt_noinline_xattr,
 108         Opt_inline_xattr_size,
 109         Opt_inline_data,
 110         Opt_inline_dentry,
 111         Opt_noinline_dentry,
 112         Opt_flush_merge,
 113         Opt_noflush_merge,
 114         Opt_barrier,
 115         Opt_nobarrier,
 116         Opt_fastboot,
 117         Opt_extent_cache,
 118         Opt_noextent_cache,
 119         Opt_noinline_data,
 120         Opt_data_flush,
 121         Opt_reserve_root,
 122         Opt_resgid,
 123         Opt_resuid,
 124         Opt_mode,
 125         Opt_io_size_bits,
 126         Opt_fault_injection,
 127         Opt_fault_type,
 128         Opt_lazytime,
 129         Opt_nolazytime,
 130         Opt_quota,
 131         Opt_noquota,
 132         Opt_usrquota,
 133         Opt_grpquota,
 134         Opt_prjquota,
 135         Opt_usrjquota,
 136         Opt_grpjquota,
 137         Opt_prjjquota,
 138         Opt_offusrjquota,
 139         Opt_offgrpjquota,
 140         Opt_offprjjquota,
 141         Opt_jqfmt_vfsold,
 142         Opt_jqfmt_vfsv0,
 143         Opt_jqfmt_vfsv1,
 144         Opt_alloc,
 145         Opt_fsync,
 146         Opt_test_dummy_encryption,
 147         Opt_inlinecrypt,
 148         Opt_checkpoint_disable,
 149         Opt_checkpoint_disable_cap,
 150         Opt_checkpoint_disable_cap_perc,
 151         Opt_checkpoint_enable,
 152         Opt_checkpoint_merge,
 153         Opt_nocheckpoint_merge,
 154         Opt_compress_algorithm,
 155         Opt_compress_log_size,
 156         Opt_compress_extension,
 157         Opt_nocompress_extension,
 158         Opt_compress_chksum,
 159         Opt_compress_mode,
 160         Opt_compress_cache,
 161         Opt_atgc,
 162         Opt_gc_merge,
 163         Opt_nogc_merge,
 164         Opt_discard_unit,
 165         Opt_memory_mode,
 166         Opt_age_extent_cache,
 167         Opt_err,
 168 };
 169
 170 static match_table_t f2fs_tokens = {
 171         {Opt_gc_background, "background_gc=%s"},
 172         {Opt_disable_roll_forward, "disable_roll_forward"},
 173         {Opt_norecovery, "norecovery"},
 174         {Opt_discard, "discard"},
 175         {Opt_nodiscard, "nodiscard"},
 176         {Opt_noheap, "no_heap"},
 177         {Opt_heap, "heap"},
 178         {Opt_user_xattr, "user_xattr"},
 179         {Opt_nouser_xattr, "nouser_xattr"},
 180         {Opt_acl, "acl"},
 181         {Opt_noacl, "noacl"},
 182         {Opt_active_logs, "active_logs=%u"},
 183         {Opt_disable_ext_identify, "disable_ext_identify"},
 184         {Opt_inline_xattr, "inline_xattr"},
 185         {Opt_noinline_xattr, "noinline_xattr"},
 186         {Opt_inline_xattr_size, "inline_xattr_size=%u"},
 187         {Opt_inline_data, "inline_data"},
 188         {Opt_inline_dentry, "inline_dentry"},
 189         {Opt_noinline_dentry, "noinline_dentry"},
 190         {Opt_flush_merge, "flush_merge"},
 191         {Opt_noflush_merge, "noflush_merge"},
 192         {Opt_barrier, "barrier"},
 193         {Opt_nobarrier, "nobarrier"},
 194         {Opt_fastboot, "fastboot"},
 195         {Opt_extent_cache, "extent_cache"},
 196         {Opt_noextent_cache, "noextent_cache"},
 197         {Opt_noinline_data, "noinline_data"},
 198         {Opt_data_flush, "data_flush"},
 199         {Opt_reserve_root, "reserve_root=%u"},
 200         {Opt_resgid, "resgid=%u"},
 201         {Opt_resuid, "resuid=%u"},
 202         {Opt_mode, "mode=%s"},
 203         {Opt_io_size_bits, "io_bits=%u"},
 204         {Opt_fault_injection, "fault_injection=%u"},
 205         {Opt_fault_type, "fault_type=%u"},
 206         {Opt_lazytime, "lazytime"},
 207         {Opt_nolazytime, "nolazytime"},
 208         {Opt_quota, "quota"},
 209         {Opt_noquota, "noquota"},
 210         {Opt_usrquota, "usrquota"},
 211         {Opt_grpquota, "grpquota"},
 212         {Opt_prjquota, "prjquota"},
 213         {Opt_usrjquota, "usrjquota=%s"},
 214         {Opt_grpjquota, "grpjquota=%s"},
 215         {Opt_prjjquota, "prjjquota=%s"},
 216         {Opt_offusrjquota, "usrjquota="},
 217         {Opt_offgrpjquota, "grpjquota="},
 218         {Opt_offprjjquota, "prjjquota="},
 219         {Opt_jqfmt_vfsold, "jqfmt=vfsold"},
 220         {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
 221         {Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
 222         {Opt_alloc, "alloc_mode=%s"},
 223         {Opt_fsync, "fsync_mode=%s"},
 224         {Opt_test_dummy_encryption, "test_dummy_encryption=%s"},
 225         {Opt_test_dummy_encryption, "test_dummy_encryption"},
 226         {Opt_inlinecrypt, "inlinecrypt"},
 227         {Opt_checkpoint_disable, "checkpoint=disable"},
 228         {Opt_checkpoint_disable_cap, "checkpoint=disable:%u"},
 229         {Opt_checkpoint_disable_cap_perc, "checkpoint=disable:%u%%"},
 230         {Opt_checkpoint_enable, "checkpoint=enable"},
 231         {Opt_checkpoint_merge, "checkpoint_merge"},
 232         {Opt_nocheckpoint_merge, "nocheckpoint_merge"},
 233         {Opt_compress_algorithm, "compress_algorithm=%s"},
 234         {Opt_compress_log_size, "compress_log_size=%u"},
 235         {Opt_compress_extension, "compress_extension=%s"},
 236         {Opt_nocompress_extension, "nocompress_extension=%s"},
 237         {Opt_compress_chksum, "compress_chksum"},
 238         {Opt_compress_mode, "compress_mode=%s"},
 239         {Opt_compress_cache, "compress_cache"},
 240         {Opt_atgc, "atgc"},
 241         {Opt_gc_merge, "gc_merge"},
 242         {Opt_nogc_merge, "nogc_merge"},
 243         {Opt_discard_unit, "discard_unit=%s"},
 244         {Opt_memory_mode, "memory=%s"},
 245         {Opt_age_extent_cache, "age_extent_cache"},
 246         {Opt_err, NULL},
 247 };
 248
 249 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...)
 250 {
 251         struct va_format vaf;
 252         va_list args;
 253         int level;
 254
 255         va_start(args, fmt);
 256
 257         level = printk_get_level(fmt);
 258         vaf.fmt = printk_skip_level(fmt);
 259         vaf.va = &args;
 260         printk("%c%cF2FS-fs (%s): %pV\n",
 261                KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
 262
 263         va_end(args);
 264 }
 265
 266 #if IS_ENABLED(CONFIG_UNICODE)
 267 static const struct f2fs_sb_encodings {
 268         __u16 magic;
 269         char *name;
 270         unsigned int version;
 271 } f2fs_sb_encoding_map[] = {
 272         {F2FS_ENC_UTF8_12_1, "utf8", UNICODE_AGE(12, 1, 0)},
 273 };
 274
 275 static const struct f2fs_sb_encodings *
 276 f2fs_sb_read_encoding(const struct f2fs_super_block *sb)
 277 {
 278         __u16 magic = le16_to_cpu(sb->s_encoding);
 279         int i;
 280
 281         for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
 282                 if (magic == f2fs_sb_encoding_map[i].magic)
 283                         return &f2fs_sb_encoding_map[i];
 284
 285         return NULL;
 286 }
 287
 288 struct kmem_cache *f2fs_cf_name_slab;
 289 static int __init f2fs_create_casefold_cache(void)
 290 {
 291         f2fs_cf_name_slab = f2fs_kmem_cache_create("f2fs_casefolded_name",
 292                                                         F2FS_NAME_LEN);
 293         return f2fs_cf_name_slab ? 0 : -ENOMEM;
 294 }
 295
 296 static void f2fs_destroy_casefold_cache(void)
 297 {
 298         kmem_cache_destroy(f2fs_cf_name_slab);
 299 }
 300 #else
 301 static int __init f2fs_create_casefold_cache(void) { return 0; }
 302 static void f2fs_destroy_casefold_cache(void) { }
 303 #endif
 304
 305 static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
 306 {
 307         block_t limit = min((sbi->user_block_count >> 3),
 308                         sbi->user_block_count - sbi->reserved_blocks);
 309
 310         /* limit is 12.5% */
 311         if (test_opt(sbi, RESERVE_ROOT) &&
 312                         F2FS_OPTION(sbi).root_reserved_blocks > limit) {
 313                 F2FS_OPTION(sbi).root_reserved_blocks = limit;
 314                 f2fs_info(sbi, "Reduce reserved blocks for root = %u",
 315                           F2FS_OPTION(sbi).root_reserved_blocks);
 316         }
 317         if (!test_opt(sbi, RESERVE_ROOT) &&
 318                 (!uid_eq(F2FS_OPTION(sbi).s_resuid,
 319                                 make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
 320                 !gid_eq(F2FS_OPTION(sbi).s_resgid,
 321                                 make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
 322                 f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
 323                           from_kuid_munged(&init_user_ns,
 324                                            F2FS_OPTION(sbi).s_resuid),
 325                           from_kgid_munged(&init_user_ns,
 326                                            F2FS_OPTION(sbi).s_resgid));
 327 }
 328
 329 static inline int adjust_reserved_segment(struct f2fs_sb_info *sbi)
 330 {
 331         unsigned int sec_blks = sbi->blocks_per_seg * sbi->segs_per_sec;
 332         unsigned int avg_vblocks;
 333         unsigned int wanted_reserved_segments;
 334         block_t avail_user_block_count;
 335
 336         if (!F2FS_IO_ALIGNED(sbi))
 337                 return 0;
 338
 339         /* average valid block count in section in worst case */
 340         avg_vblocks = sec_blks / F2FS_IO_SIZE(sbi);
 341
 342         /*
 343          * we need enough free space when migrating one section in worst case
 344          */
 345         wanted_reserved_segments = (F2FS_IO_SIZE(sbi) / avg_vblocks) *
 346                                                 reserved_segments(sbi);
 347         wanted_reserved_segments -= reserved_segments(sbi);
 348
 349         avail_user_block_count = sbi->user_block_count -
 350                                 sbi->current_reserved_blocks -
 351                                 F2FS_OPTION(sbi).root_reserved_blocks;
 352
 353         if (wanted_reserved_segments * sbi->blocks_per_seg >
 354                                         avail_user_block_count) {
 355                 f2fs_err(sbi, "IO align feature can't grab additional reserved segment: %u, available segments: %u",
 356                         wanted_reserved_segments,
 357                         avail_user_block_count >> sbi->log_blocks_per_seg);
 358                 return -ENOSPC;
 359         }
 360
 361         SM_I(sbi)->additional_reserved_segments = wanted_reserved_segments;
 362
 363         f2fs_info(sbi, "IO align feature needs additional reserved segment: %u",
 364                          wanted_reserved_segments);
 365
 366         return 0;
 367 }
 368
 369 static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
 370 {
 371         if (!F2FS_OPTION(sbi).unusable_cap_perc)
 372                 return;
 373
 374         if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
 375                 F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
 376         else
 377                 F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
 378                                         F2FS_OPTION(sbi).unusable_cap_perc;
 379
 380         f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
 381                         F2FS_OPTION(sbi).unusable_cap,
 382                         F2FS_OPTION(sbi).unusable_cap_perc);
 383 }
 384
 385 static void init_once(void *foo)
 386 {
 387         struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
 388
 389         inode_init_once(&fi->vfs_inode);
 390 }
 391
 392 #ifdef CONFIG_QUOTA
 393 static const char * const quotatypes[] = INITQFNAMES;
 394 #define QTYPE2NAME(t) (quotatypes[t])
 395 static int f2fs_set_qf_name(struct super_block *sb, int qtype,
 396                                                         substring_t *args)
 397 {
 398         struct f2fs_sb_info *sbi = F2FS_SB(sb);
 399         char *qname;
 400         int ret = -EINVAL;
 401
 402         if (sb_any_quota_loaded(sb) && !F2FS_OPTION(sbi).s_qf_names[qtype]) {
 403                 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
 404                 return -EINVAL;
 405         }
 406         if (f2fs_sb_has_quota_ino(sbi)) {
 407                 f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
 408                 return 0;
 409         }
 410
 411         qname = match_strdup(args);
 412         if (!qname) {
 413                 f2fs_err(sbi, "Not enough memory for storing quotafile name");
 414                 return -ENOMEM;
 415         }
 416         if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
 417                 if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
 418                         ret = 0;
 419                 else
 420                         f2fs_err(sbi, "%s quota file already specified",
 421                                  QTYPE2NAME(qtype));
 422                 goto errout;
 423         }
 424         if (strchr(qname, '/')) {
 425                 f2fs_err(sbi, "quotafile must be on filesystem root");
 426                 goto errout;
 427         }
 428         F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
 429         set_opt(sbi, QUOTA);
 430         return 0;
 431 errout:
 432         kfree(qname);
 433         return ret;
 434 }
 435
 436 static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
 437 {
 438         struct f2fs_sb_info *sbi = F2FS_SB(sb);
 439
 440         if (sb_any_quota_loaded(sb) && F2FS_OPTION(sbi).s_qf_names[qtype]) {
 441                 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
 442                 return -EINVAL;
 443         }
 444         kfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
 445         F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
 446         return 0;
 447 }
 448
 449 static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
 450 {
 451         /*
 452          * We do the test below only for project quotas. 'usrquota' and
 453          * 'grpquota' mount options are allowed even without quota feature
 454          * to support legacy quotas in quota files.
 455          */
 456         if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi)) {
 457                 f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
 458                 return -1;
 459         }
 460         if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
 461                         F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
 462                         F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) {
 463                 if (test_opt(sbi, USRQUOTA) &&
 464                                 F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
 465                         clear_opt(sbi, USRQUOTA);
 466
 467                 if (test_opt(sbi, GRPQUOTA) &&
 468                                 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
 469                         clear_opt(sbi, GRPQUOTA);
 470
 471                 if (test_opt(sbi, PRJQUOTA) &&
 472                                 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
 473                         clear_opt(sbi, PRJQUOTA);
 474
 475                 if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
 476                                 test_opt(sbi, PRJQUOTA)) {
 477                         f2fs_err(sbi, "old and new quota format mixing");
 478                         return -1;
 479                 }
 480
 481                 if (!F2FS_OPTION(sbi).s_jquota_fmt) {
 482                         f2fs_err(sbi, "journaled quota format not specified");
 483                         return -1;
 484                 }
 485         }
 486
 487         if (f2fs_sb_has_quota_ino(sbi) && F2FS_OPTION(sbi).s_jquota_fmt) {
 488                 f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
 489                 F2FS_OPTION(sbi).s_jquota_fmt = 0;
 490         }
 491         return 0;
 492 }
 493 #endif
 494
 495 static int f2fs_set_test_dummy_encryption(struct super_block *sb,
 496                                           const char *opt,
 497                                           const substring_t *arg,
 498                                           bool is_remount)
 499 {
 500         struct f2fs_sb_info *sbi = F2FS_SB(sb);
 501         struct fs_parameter param = {
 502                 .type = fs_value_is_string,
 503                 .string = arg->from ? arg->from : "",
 504         };
 505         struct fscrypt_dummy_policy *policy =
 506                 &F2FS_OPTION(sbi).dummy_enc_policy;
 507         int err;
 508
 509         if (!IS_ENABLED(CONFIG_FS_ENCRYPTION)) {
 510                 f2fs_warn(sbi, "test_dummy_encryption option not supported");
 511                 return -EINVAL;
 512         }
 513
 514         if (!f2fs_sb_has_encrypt(sbi)) {
 515                 f2fs_err(sbi, "Encrypt feature is off");
 516                 return -EINVAL;
 517         }
 518
 519         /*
 520          * This mount option is just for testing, and it's not worthwhile to
 521          * implement the extra complexity (e.g. RCU protection) that would be
 522          * needed to allow it to be set or changed during remount.  We do allow
 523          * it to be specified during remount, but only if there is no change.
 524          */
 525         if (is_remount && !fscrypt_is_dummy_policy_set(policy)) {
 526                 f2fs_warn(sbi, "Can't set test_dummy_encryption on remount");
 527                 return -EINVAL;
 528         }
 529
 530         err = fscrypt_parse_test_dummy_encryption(&param, policy);
 531         if (err) {
 532                 if (err == -EEXIST)
 533                         f2fs_warn(sbi,
 534                                   "Can't change test_dummy_encryption on remount");
 535                 else if (err == -EINVAL)
 536                         f2fs_warn(sbi, "Value of option \"%s\" is unrecognized",
 537                                   opt);
 538                 else
 539                         f2fs_warn(sbi, "Error processing option \"%s\" [%d]",
 540                                   opt, err);
 541                 return -EINVAL;
 542         }
 543         f2fs_warn(sbi, "Test dummy encryption mode enabled");
 544         return 0;
 545 }
 546
 547 #ifdef CONFIG_F2FS_FS_COMPRESSION
 548 /*
 549  * 1. The same extension name cannot not appear in both compress and non-compress extension
 550  * at the same time.
 551  * 2. If the compress extension specifies all files, the types specified by the non-compress
 552  * extension will be treated as special cases and will not be compressed.
 553  * 3. Don't allow the non-compress extension specifies all files.
 554  */
 555 static int f2fs_test_compress_extension(struct f2fs_sb_info *sbi)
 556 {
 557         unsigned char (*ext)[F2FS_EXTENSION_LEN];
 558         unsigned char (*noext)[F2FS_EXTENSION_LEN];
 559         int ext_cnt, noext_cnt, index = 0, no_index = 0;
 560
 561         ext = F2FS_OPTION(sbi).extensions;
 562         ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
 563         noext = F2FS_OPTION(sbi).noextensions;
 564         noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
 565
 566         if (!noext_cnt)
 567                 return 0;
 568
 569         for (no_index = 0; no_index < noext_cnt; no_index++) {
 570                 if (!strcasecmp("*", noext[no_index])) {
 571                         f2fs_info(sbi, "Don't allow the nocompress extension specifies all files");
 572                         return -EINVAL;
 573                 }
 574                 for (index = 0; index < ext_cnt; index++) {
 575                         if (!strcasecmp(ext[index], noext[no_index])) {
 576                                 f2fs_info(sbi, "Don't allow the same extension %s appear in both compress and nocompress extension",
 577                                                 ext[index]);
 578                                 return -EINVAL;
 579                         }
 580                 }
 581         }
 582         return 0;
 583 }
 584
 585 #ifdef CONFIG_F2FS_FS_LZ4
 586 static int f2fs_set_lz4hc_level(struct f2fs_sb_info *sbi, const char *str)
 587 {
 588 #ifdef CONFIG_F2FS_FS_LZ4HC
 589         unsigned int level;
 590 #endif
 591
 592         if (strlen(str) == 3) {
 593                 F2FS_OPTION(sbi).compress_level = 0;
 594                 return 0;
 595         }
 596
 597 #ifdef CONFIG_F2FS_FS_LZ4HC
 598         str += 3;
 599
 600         if (str[0] != ':') {
 601                 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
 602                 return -EINVAL;
 603         }
 604         if (kstrtouint(str + 1, 10, &level))
 605                 return -EINVAL;
 606
 607         if (level < LZ4HC_MIN_CLEVEL || level > LZ4HC_MAX_CLEVEL) {
 608                 f2fs_info(sbi, "invalid lz4hc compress level: %d", level);
 609                 return -EINVAL;
 610         }
 611
 612         F2FS_OPTION(sbi).compress_level = level;
 613         return 0;
 614 #else
 615         f2fs_info(sbi, "kernel doesn't support lz4hc compression");
 616         return -EINVAL;
 617 #endif
 618 }
 619 #endif
 620
 621 #ifdef CONFIG_F2FS_FS_ZSTD
 622 static int f2fs_set_zstd_level(struct f2fs_sb_info *sbi, const char *str)
 623 {
 624         unsigned int level;
 625         int len = 4;
 626
 627         if (strlen(str) == len) {
 628                 F2FS_OPTION(sbi).compress_level = 0;
 629                 return 0;
 630         }
 631
 632         str += len;
 633
 634         if (str[0] != ':') {
 635                 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
 636                 return -EINVAL;
 637         }
 638         if (kstrtouint(str + 1, 10, &level))
 639                 return -EINVAL;
 640
 641         if (!level || level > zstd_max_clevel()) {
 642                 f2fs_info(sbi, "invalid zstd compress level: %d", level);
 643                 return -EINVAL;
 644         }
 645
 646         F2FS_OPTION(sbi).compress_level = level;
 647         return 0;
 648 }
 649 #endif
 650 #endif
 651
 652 static int parse_options(struct super_block *sb, char *options, bool is_remount)
 653 {
 654         struct f2fs_sb_info *sbi = F2FS_SB(sb);
 655         substring_t args[MAX_OPT_ARGS];
 656 #ifdef CONFIG_F2FS_FS_COMPRESSION
 657         unsigned char (*ext)[F2FS_EXTENSION_LEN];
 658         unsigned char (*noext)[F2FS_EXTENSION_LEN];
 659         int ext_cnt, noext_cnt;
 660 #endif
 661         char *p, *name;
 662         int arg = 0;
 663         kuid_t uid;
 664         kgid_t gid;
 665         int ret;
 666
 667         if (!options)
 668                 goto default_check;
 669
 670         while ((p = strsep(&options, ",")) != NULL) {
 671                 int token;
 672
 673                 if (!*p)
 674                         continue;
 675                 /*
 676                  * Initialize args struct so we know whether arg was
 677                  * found; some options take optional arguments.
 678                  */
 679                 args[0].to = args[0].from = NULL;
 680                 token = match_token(p, f2fs_tokens, args);
 681
 682                 switch (token) {
 683                 case Opt_gc_background:
 684                         name = match_strdup(&args[0]);
 685
 686                         if (!name)
 687                                 return -ENOMEM;
 688                         if (!strcmp(name, "on")) {
 689                                 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
 690                         } else if (!strcmp(name, "off")) {
 691                                 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_OFF;
 692                         } else if (!strcmp(name, "sync")) {
 693                                 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_SYNC;
 694                         } else {
 695                                 kfree(name);
 696                                 return -EINVAL;
 697                         }
 698                         kfree(name);
 699                         break;
 700                 case Opt_disable_roll_forward:
 701                         set_opt(sbi, DISABLE_ROLL_FORWARD);
 702                         break;
 703                 case Opt_norecovery:
 704                         /* this option mounts f2fs with ro */
 705                         set_opt(sbi, NORECOVERY);
 706                         if (!f2fs_readonly(sb))
 707                                 return -EINVAL;
 708                         break;
 709                 case Opt_discard:
 710                         if (!f2fs_hw_support_discard(sbi)) {
 711                                 f2fs_warn(sbi, "device does not support discard");
 712                                 break;
 713                         }
 714                         set_opt(sbi, DISCARD);
 715                         break;
 716                 case Opt_nodiscard:
 717                         if (f2fs_hw_should_discard(sbi)) {
 718                                 f2fs_warn(sbi, "discard is required for zoned block devices");
 719                                 return -EINVAL;
 720                         }
 721                         clear_opt(sbi, DISCARD);
 722                         break;
 723                 case Opt_noheap:
 724                         set_opt(sbi, NOHEAP);
 725                         break;
 726                 case Opt_heap:
 727                         clear_opt(sbi, NOHEAP);
 728                         break;
 729 #ifdef CONFIG_F2FS_FS_XATTR
 730                 case Opt_user_xattr:
 731                         set_opt(sbi, XATTR_USER);
 732                         break;
 733                 case Opt_nouser_xattr:
 734                         clear_opt(sbi, XATTR_USER);
 735                         break;
 736                 case Opt_inline_xattr:
 737                         set_opt(sbi, INLINE_XATTR);
 738                         break;
 739                 case Opt_noinline_xattr:
 740                         clear_opt(sbi, INLINE_XATTR);
 741                         break;
 742                 case Opt_inline_xattr_size:
 743                         if (args->from && match_int(args, &arg))
 744                                 return -EINVAL;
 745                         set_opt(sbi, INLINE_XATTR_SIZE);
 746                         F2FS_OPTION(sbi).inline_xattr_size = arg;
 747                         break;
 748 #else
 749                 case Opt_user_xattr:
 750                         f2fs_info(sbi, "user_xattr options not supported");
 751                         break;
 752                 case Opt_nouser_xattr:
 753                         f2fs_info(sbi, "nouser_xattr options not supported");
 754                         break;
 755                 case Opt_inline_xattr:
 756                         f2fs_info(sbi, "inline_xattr options not supported");
 757                         break;
 758                 case Opt_noinline_xattr:
 759                         f2fs_info(sbi, "noinline_xattr options not supported");
 760                         break;
 761 #endif
 762 #ifdef CONFIG_F2FS_FS_POSIX_ACL
 763                 case Opt_acl:
 764                         set_opt(sbi, POSIX_ACL);
 765                         break;
 766                 case Opt_noacl:
 767                         clear_opt(sbi, POSIX_ACL);
 768                         break;
 769 #else
 770                 case Opt_acl:
 771                         f2fs_info(sbi, "acl options not supported");
 772                         break;
 773                 case Opt_noacl:
 774                         f2fs_info(sbi, "noacl options not supported");
 775                         break;
 776 #endif
 777                 case Opt_active_logs:
 778                         if (args->from && match_int(args, &arg))
 779                                 return -EINVAL;
 780                         if (arg != 2 && arg != 4 &&
 781                                 arg != NR_CURSEG_PERSIST_TYPE)
 782                                 return -EINVAL;
 783                         F2FS_OPTION(sbi).active_logs = arg;
 784                         break;
 785                 case Opt_disable_ext_identify:
 786                         set_opt(sbi, DISABLE_EXT_IDENTIFY);
 787                         break;
 788                 case Opt_inline_data:
 789                         set_opt(sbi, INLINE_DATA);
 790                         break;
 791                 case Opt_inline_dentry:
 792                         set_opt(sbi, INLINE_DENTRY);
 793                         break;
 794                 case Opt_noinline_dentry:
 795                         clear_opt(sbi, INLINE_DENTRY);
 796                         break;
 797                 case Opt_flush_merge:
 798                         set_opt(sbi, FLUSH_MERGE);
 799                         break;
 800                 case Opt_noflush_merge:
 801                         clear_opt(sbi, FLUSH_MERGE);
 802                         break;
 803                 case Opt_nobarrier:
 804                         set_opt(sbi, NOBARRIER);
 805                         break;
 806                 case Opt_barrier:
 807                         clear_opt(sbi, NOBARRIER);
 808                         break;
 809                 case Opt_fastboot:
 810                         set_opt(sbi, FASTBOOT);
 811                         break;
 812                 case Opt_extent_cache:
 813                         set_opt(sbi, READ_EXTENT_CACHE);
 814                         break;
 815                 case Opt_noextent_cache:
 816                         clear_opt(sbi, READ_EXTENT_CACHE);
 817                         break;
 818                 case Opt_noinline_data:
 819                         clear_opt(sbi, INLINE_DATA);
 820                         break;
 821                 case Opt_data_flush:
 822                         set_opt(sbi, DATA_FLUSH);
 823                         break;
 824                 case Opt_reserve_root:
 825                         if (args->from && match_int(args, &arg))
 826                                 return -EINVAL;
 827                         if (test_opt(sbi, RESERVE_ROOT)) {
 828                                 f2fs_info(sbi, "Preserve previous reserve_root=%u",
 829                                           F2FS_OPTION(sbi).root_reserved_blocks);
 830                         } else {
 831                                 F2FS_OPTION(sbi).root_reserved_blocks = arg;
 832                                 set_opt(sbi, RESERVE_ROOT);
 833                         }
 834                         break;
 835                 case Opt_resuid:
 836                         if (args->from && match_int(args, &arg))
 837                                 return -EINVAL;
 838                         uid = make_kuid(current_user_ns(), arg);
 839                         if (!uid_valid(uid)) {
 840                                 f2fs_err(sbi, "Invalid uid value %d", arg);
 841                                 return -EINVAL;
 842                         }
 843                         F2FS_OPTION(sbi).s_resuid = uid;
 844                         break;
 845                 case Opt_resgid:
 846                         if (args->from && match_int(args, &arg))
 847                                 return -EINVAL;
 848                         gid = make_kgid(current_user_ns(), arg);
 849                         if (!gid_valid(gid)) {
 850                                 f2fs_err(sbi, "Invalid gid value %d", arg);
 851                                 return -EINVAL;
 852                         }
 853                         F2FS_OPTION(sbi).s_resgid = gid;
 854                         break;
 855                 case Opt_mode:
 856                         name = match_strdup(&args[0]);
 857
 858                         if (!name)
 859                                 return -ENOMEM;
 860                         if (!strcmp(name, "adaptive")) {
 861                                 if (f2fs_sb_has_blkzoned(sbi)) {
 862                                         f2fs_warn(sbi, "adaptive mode is not allowed with zoned block device feature");
 863                                         kfree(name);
 864                                         return -EINVAL;
 865                                 }
 866                                 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
 867                         } else if (!strcmp(name, "lfs")) {
 868                                 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
 869                         } else if (!strcmp(name, "fragment:segment")) {
 870                                 F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_SEG;
 871                         } else if (!strcmp(name, "fragment:block")) {
 872                                 F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_BLK;
 873                         } else {
 874                                 kfree(name);
 875                                 return -EINVAL;
 876                         }
 877                         kfree(name);
 878                         break;
 879                 case Opt_io_size_bits:
 880                         if (args->from && match_int(args, &arg))
 881                                 return -EINVAL;
 882                         if (arg <= 0 || arg > __ilog2_u32(BIO_MAX_VECS)) {
 883                                 f2fs_warn(sbi, "Not support %ld, larger than %d",
 884                                         BIT(arg), BIO_MAX_VECS);
 885                                 return -EINVAL;
 886                         }
 887                         F2FS_OPTION(sbi).write_io_size_bits = arg;
 888                         break;
 889 #ifdef CONFIG_F2FS_FAULT_INJECTION
 890                 case Opt_fault_injection:
 891                         if (args->from && match_int(args, &arg))
 892                                 return -EINVAL;
 893                         f2fs_build_fault_attr(sbi, arg, F2FS_ALL_FAULT_TYPE);
 894                         set_opt(sbi, FAULT_INJECTION);
 895                         break;
 896
 897                 case Opt_fault_type:
 898                         if (args->from && match_int(args, &arg))
 899                                 return -EINVAL;
 900                         f2fs_build_fault_attr(sbi, 0, arg);
 901                         set_opt(sbi, FAULT_INJECTION);
 902                         break;
 903 #else
 904                 case Opt_fault_injection:
 905                         f2fs_info(sbi, "fault_injection options not supported");
 906                         break;
 907
 908                 case Opt_fault_type:
 909                         f2fs_info(sbi, "fault_type options not supported");
 910                         break;
 911 #endif
 912                 case Opt_lazytime:
 913                         sb->s_flags |= SB_LAZYTIME;
 914                         break;
 915                 case Opt_nolazytime:
 916                         sb->s_flags &= ~SB_LAZYTIME;
 917                         break;
 918 #ifdef CONFIG_QUOTA
 919                 case Opt_quota:
 920                 case Opt_usrquota:
 921                         set_opt(sbi, USRQUOTA);
 922                         break;
 923                 case Opt_grpquota:
 924                         set_opt(sbi, GRPQUOTA);
 925                         break;
 926                 case Opt_prjquota:
 927                         set_opt(sbi, PRJQUOTA);
 928                         break;
 929                 case Opt_usrjquota:
 930                         ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
 931                         if (ret)
 932                                 return ret;
 933                         break;
 934                 case Opt_grpjquota:
 935                         ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
 936                         if (ret)
 937                                 return ret;
 938                         break;
 939                 case Opt_prjjquota:
 940                         ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
 941                         if (ret)
 942                                 return ret;
 943                         break;
 944                 case Opt_offusrjquota:
 945                         ret = f2fs_clear_qf_name(sb, USRQUOTA);
 946                         if (ret)
 947                                 return ret;
 948                         break;
 949                 case Opt_offgrpjquota:
 950                         ret = f2fs_clear_qf_name(sb, GRPQUOTA);
 951                         if (ret)
 952                                 return ret;
 953                         break;
 954                 case Opt_offprjjquota:
 955                         ret = f2fs_clear_qf_name(sb, PRJQUOTA);
 956                         if (ret)
 957                                 return ret;
 958                         break;
 959                 case Opt_jqfmt_vfsold:
 960                         F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
 961                         break;
 962                 case Opt_jqfmt_vfsv0:
 963                         F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
 964                         break;
 965                 case Opt_jqfmt_vfsv1:
 966                         F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
 967                         break;
 968                 case Opt_noquota:
 969                         clear_opt(sbi, QUOTA);
 970                         clear_opt(sbi, USRQUOTA);
 971                         clear_opt(sbi, GRPQUOTA);
 972                         clear_opt(sbi, PRJQUOTA);
 973                         break;
 974 #else
 975                 case Opt_quota:
 976                 case Opt_usrquota:
 977                 case Opt_grpquota:
 978                 case Opt_prjquota:
 979                 case Opt_usrjquota:
 980                 case Opt_grpjquota:
 981                 case Opt_prjjquota:
 982                 case Opt_offusrjquota:
 983                 case Opt_offgrpjquota:
 984                 case Opt_offprjjquota:
 985                 case Opt_jqfmt_vfsold:
 986                 case Opt_jqfmt_vfsv0:
 987                 case Opt_jqfmt_vfsv1:
 988                 case Opt_noquota:
 989                         f2fs_info(sbi, "quota operations not supported");
 990                         break;
 991 #endif
 992                 case Opt_alloc:
 993                         name = match_strdup(&args[0]);
 994                         if (!name)
 995                                 return -ENOMEM;
 996
 997                         if (!strcmp(name, "default")) {
 998                                 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
 999                         } else if (!strcmp(name, "reuse")) {
1000                                 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
1001                         } else {
1002                                 kfree(name);
1003                                 return -EINVAL;
1004                         }
1005                         kfree(name);
1006                         break;
1007                 case Opt_fsync:
1008                         name = match_strdup(&args[0]);
1009                         if (!name)
1010                                 return -ENOMEM;
1011                         if (!strcmp(name, "posix")) {
1012                                 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
1013                         } else if (!strcmp(name, "strict")) {
1014                                 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_STRICT;
1015                         } else if (!strcmp(name, "nobarrier")) {
1016                                 F2FS_OPTION(sbi).fsync_mode =
1017                                                         FSYNC_MODE_NOBARRIER;
1018                         } else {
1019                                 kfree(name);
1020                                 return -EINVAL;
1021                         }
1022                         kfree(name);
1023                         break;
1024                 case Opt_test_dummy_encryption:
1025                         ret = f2fs_set_test_dummy_encryption(sb, p, &args[0],
1026                                                              is_remount);
1027                         if (ret)
1028                                 return ret;
1029                         break;
1030                 case Opt_inlinecrypt:
1031 #ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
1032                         sb->s_flags |= SB_INLINECRYPT;
1033 #else
1034                         f2fs_info(sbi, "inline encryption not supported");
1035 #endif
1036                         break;
1037                 case Opt_checkpoint_disable_cap_perc:
1038                         if (args->from && match_int(args, &arg))
1039                                 return -EINVAL;
1040                         if (arg < 0 || arg > 100)
1041                                 return -EINVAL;
1042                         F2FS_OPTION(sbi).unusable_cap_perc = arg;
1043                         set_opt(sbi, DISABLE_CHECKPOINT);
1044                         break;
1045                 case Opt_checkpoint_disable_cap:
1046                         if (args->from && match_int(args, &arg))
1047                                 return -EINVAL;
1048                         F2FS_OPTION(sbi).unusable_cap = arg;
1049                         set_opt(sbi, DISABLE_CHECKPOINT);
1050                         break;
1051                 case Opt_checkpoint_disable:
1052                         set_opt(sbi, DISABLE_CHECKPOINT);
1053                         break;
1054                 case Opt_checkpoint_enable:
1055                         clear_opt(sbi, DISABLE_CHECKPOINT);
1056                         break;
1057                 case Opt_checkpoint_merge:
1058                         set_opt(sbi, MERGE_CHECKPOINT);
1059                         break;
1060                 case Opt_nocheckpoint_merge:
1061                         clear_opt(sbi, MERGE_CHECKPOINT);
1062                         break;
1063 #ifdef CONFIG_F2FS_FS_COMPRESSION
1064                 case Opt_compress_algorithm:
1065                         if (!f2fs_sb_has_compression(sbi)) {
1066                                 f2fs_info(sbi, "Image doesn't support compression");
1067                                 break;
1068                         }
1069                         name = match_strdup(&args[0]);
1070                         if (!name)
1071                                 return -ENOMEM;
1072                         if (!strcmp(name, "lzo")) {
1073 #ifdef CONFIG_F2FS_FS_LZO
1074                                 F2FS_OPTION(sbi).compress_level = 0;
1075                                 F2FS_OPTION(sbi).compress_algorithm =
1076                                                                 COMPRESS_LZO;
1077 #else
1078                                 f2fs_info(sbi, "kernel doesn't support lzo compression");
1079 #endif
1080                         } else if (!strncmp(name, "lz4", 3)) {
1081 #ifdef CONFIG_F2FS_FS_LZ4
1082                                 ret = f2fs_set_lz4hc_level(sbi, name);
1083                                 if (ret) {
1084                                         kfree(name);
1085                                         return -EINVAL;
1086                                 }
1087                                 F2FS_OPTION(sbi).compress_algorithm =
1088                                                                 COMPRESS_LZ4;
1089 #else
1090                                 f2fs_info(sbi, "kernel doesn't support lz4 compression");
1091 #endif
1092                         } else if (!strncmp(name, "zstd", 4)) {
1093 #ifdef CONFIG_F2FS_FS_ZSTD
1094                                 ret = f2fs_set_zstd_level(sbi, name);
1095                                 if (ret) {
1096                                         kfree(name);
1097                                         return -EINVAL;
1098                                 }
1099                                 F2FS_OPTION(sbi).compress_algorithm =
1100                                                                 COMPRESS_ZSTD;
1101 #else
1102                                 f2fs_info(sbi, "kernel doesn't support zstd compression");
1103 #endif
1104                         } else if (!strcmp(name, "lzo-rle")) {
1105 #ifdef CONFIG_F2FS_FS_LZORLE
1106                                 F2FS_OPTION(sbi).compress_level = 0;
1107                                 F2FS_OPTION(sbi).compress_algorithm =
1108                                                                 COMPRESS_LZORLE;
1109 #else
1110                                 f2fs_info(sbi, "kernel doesn't support lzorle compression");
1111 #endif
1112                         } else {
1113                                 kfree(name);
1114                                 return -EINVAL;
1115                         }
1116                         kfree(name);
1117                         break;
1118                 case Opt_compress_log_size:
1119                         if (!f2fs_sb_has_compression(sbi)) {
1120                                 f2fs_info(sbi, "Image doesn't support compression");
1121                                 break;
1122                         }
1123                         if (args->from && match_int(args, &arg))
1124                                 return -EINVAL;
1125                         if (arg < MIN_COMPRESS_LOG_SIZE ||
1126                                 arg > MAX_COMPRESS_LOG_SIZE) {
1127                                 f2fs_err(sbi,
1128                                         "Compress cluster log size is out of range");
1129                                 return -EINVAL;
1130                         }
1131                         F2FS_OPTION(sbi).compress_log_size = arg;
1132                         break;
1133                 case Opt_compress_extension:
1134                         if (!f2fs_sb_has_compression(sbi)) {
1135                                 f2fs_info(sbi, "Image doesn't support compression");
1136                                 break;
1137                         }
1138                         name = match_strdup(&args[0]);
1139                         if (!name)
1140                                 return -ENOMEM;
1141
1142                         ext = F2FS_OPTION(sbi).extensions;
1143                         ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
1144
1145                         if (strlen(name) >= F2FS_EXTENSION_LEN ||
1146                                 ext_cnt >= COMPRESS_EXT_NUM) {
1147                                 f2fs_err(sbi,
1148                                         "invalid extension length/number");
1149                                 kfree(name);
1150                                 return -EINVAL;
1151                         }
1152
1153                         strcpy(ext[ext_cnt], name);
1154                         F2FS_OPTION(sbi).compress_ext_cnt++;
1155                         kfree(name);
1156                         break;
1157                 case Opt_nocompress_extension:
1158                         if (!f2fs_sb_has_compression(sbi)) {
1159                                 f2fs_info(sbi, "Image doesn't support compression");
1160                                 break;
1161                         }
1162                         name = match_strdup(&args[0]);
1163                         if (!name)
1164                                 return -ENOMEM;
1165
1166                         noext = F2FS_OPTION(sbi).noextensions;
1167                         noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
1168
1169                         if (strlen(name) >= F2FS_EXTENSION_LEN ||
1170                                 noext_cnt >= COMPRESS_EXT_NUM) {
1171                                 f2fs_err(sbi,
1172                                         "invalid extension length/number");
1173                                 kfree(name);
1174                                 return -EINVAL;
1175                         }
1176
1177                         strcpy(noext[noext_cnt], name);
1178                         F2FS_OPTION(sbi).nocompress_ext_cnt++;
1179                         kfree(name);
1180                         break;
1181                 case Opt_compress_chksum:
1182                         if (!f2fs_sb_has_compression(sbi)) {
1183                                 f2fs_info(sbi, "Image doesn't support compression");
1184                                 break;
1185                         }
1186                         F2FS_OPTION(sbi).compress_chksum = true;
1187                         break;
1188                 case Opt_compress_mode:
1189                         if (!f2fs_sb_has_compression(sbi)) {
1190                                 f2fs_info(sbi, "Image doesn't support compression");
1191                                 break;
1192                         }
1193                         name = match_strdup(&args[0]);
1194                         if (!name)
1195                                 return -ENOMEM;
1196                         if (!strcmp(name, "fs")) {
1197                                 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
1198                         } else if (!strcmp(name, "user")) {
1199                                 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_USER;
1200                         } else {
1201                                 kfree(name);
1202                                 return -EINVAL;
1203                         }
1204                         kfree(name);
1205                         break;
1206                 case Opt_compress_cache:
1207                         if (!f2fs_sb_has_compression(sbi)) {
1208                                 f2fs_info(sbi, "Image doesn't support compression");
1209                                 break;
1210                         }
1211                         set_opt(sbi, COMPRESS_CACHE);
1212                         break;
1213 #else
1214                 case Opt_compress_algorithm:
1215                 case Opt_compress_log_size:
1216                 case Opt_compress_extension:
1217                 case Opt_nocompress_extension:
1218                 case Opt_compress_chksum:
1219                 case Opt_compress_mode:
1220                 case Opt_compress_cache:
1221                         f2fs_info(sbi, "compression options not supported");
1222                         break;
1223 #endif
1224                 case Opt_atgc:
1225                         set_opt(sbi, ATGC);
1226                         break;
1227                 case Opt_gc_merge:
1228                         set_opt(sbi, GC_MERGE);
1229                         break;
1230                 case Opt_nogc_merge:
1231                         clear_opt(sbi, GC_MERGE);
1232                         break;
1233                 case Opt_discard_unit:
1234                         name = match_strdup(&args[0]);
1235                         if (!name)
1236                                 return -ENOMEM;
1237                         if (!strcmp(name, "block")) {
1238                                 F2FS_OPTION(sbi).discard_unit =
1239                                                 DISCARD_UNIT_BLOCK;
1240                         } else if (!strcmp(name, "segment")) {
1241                                 F2FS_OPTION(sbi).discard_unit =
1242                                                 DISCARD_UNIT_SEGMENT;
1243                         } else if (!strcmp(name, "section")) {
1244                                 F2FS_OPTION(sbi).discard_unit =
1245                                                 DISCARD_UNIT_SECTION;
1246                         } else {
1247                                 kfree(name);
1248                                 return -EINVAL;
1249                         }
1250                         kfree(name);
1251                         break;
1252                 case Opt_memory_mode:
1253                         name = match_strdup(&args[0]);
1254                         if (!name)
1255                                 return -ENOMEM;
1256                         if (!strcmp(name, "normal")) {
1257                                 F2FS_OPTION(sbi).memory_mode =
1258                                                 MEMORY_MODE_NORMAL;
1259                         } else if (!strcmp(name, "low")) {
1260                                 F2FS_OPTION(sbi).memory_mode =
1261                                                 MEMORY_MODE_LOW;
1262                         } else {
1263                                 kfree(name);
1264                                 return -EINVAL;
1265                         }
1266                         kfree(name);
1267                         break;
1268                 case Opt_age_extent_cache:
1269                         set_opt(sbi, AGE_EXTENT_CACHE);
1270                         break;
1271                 default:
1272                         f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
1273                                  p);
1274                         return -EINVAL;
1275                 }
1276         }
1277 default_check:
1278 #ifdef CONFIG_QUOTA
1279         if (f2fs_check_quota_options(sbi))
1280                 return -EINVAL;
1281 #else
1282         if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sbi->sb)) {
1283                 f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1284                 return -EINVAL;
1285         }
1286         if (f2fs_sb_has_project_quota(sbi) && !f2fs_readonly(sbi->sb)) {
1287                 f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1288                 return -EINVAL;
1289         }
1290 #endif
1291 #if !IS_ENABLED(CONFIG_UNICODE)
1292         if (f2fs_sb_has_casefold(sbi)) {
1293                 f2fs_err(sbi,
1294                         "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
1295                 return -EINVAL;
1296         }
1297 #endif
1298         /*
1299          * The BLKZONED feature indicates that the drive was formatted with
1300          * zone alignment optimization. This is optional for host-aware
1301          * devices, but mandatory for host-managed zoned block devices.
1302          */
1303         if (f2fs_sb_has_blkzoned(sbi)) {
1304 #ifdef CONFIG_BLK_DEV_ZONED
1305                 if (F2FS_OPTION(sbi).discard_unit !=
1306                                                 DISCARD_UNIT_SECTION) {
1307                         f2fs_info(sbi, "Zoned block device doesn't need small discard, set discard_unit=section by default");
1308                         F2FS_OPTION(sbi).discard_unit =
1309                                         DISCARD_UNIT_SECTION;
1310                 }
1311 #else
1312                 f2fs_err(sbi, "Zoned block device support is not enabled");
1313                 return -EINVAL;
1314 #endif
1315         }
1316
1317 #ifdef CONFIG_F2FS_FS_COMPRESSION
1318         if (f2fs_test_compress_extension(sbi)) {
1319                 f2fs_err(sbi, "invalid compress or nocompress extension");
1320                 return -EINVAL;
1321         }
1322 #endif
1323
1324         if (F2FS_IO_SIZE_BITS(sbi) && !f2fs_lfs_mode(sbi)) {
1325                 f2fs_err(sbi, "Should set mode=lfs with %luKB-sized IO",
1326                          F2FS_IO_SIZE_KB(sbi));
1327                 return -EINVAL;
1328         }
1329
1330         if (test_opt(sbi, INLINE_XATTR_SIZE)) {
1331                 int min_size, max_size;
1332
1333                 if (!f2fs_sb_has_extra_attr(sbi) ||
1334                         !f2fs_sb_has_flexible_inline_xattr(sbi)) {
1335                         f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
1336                         return -EINVAL;
1337                 }
1338                 if (!test_opt(sbi, INLINE_XATTR)) {
1339                         f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
1340                         return -EINVAL;
1341                 }
1342
1343                 min_size = sizeof(struct f2fs_xattr_header) / sizeof(__le32);
1344                 max_size = MAX_INLINE_XATTR_SIZE;
1345
1346                 if (F2FS_OPTION(sbi).inline_xattr_size < min_size ||
1347                                 F2FS_OPTION(sbi).inline_xattr_size > max_size) {
1348                         f2fs_err(sbi, "inline xattr size is out of range: %d ~ %d",
1349                                  min_size, max_size);
1350                         return -EINVAL;
1351                 }
1352         }
1353
1354         if (test_opt(sbi, DISABLE_CHECKPOINT) && f2fs_lfs_mode(sbi)) {
1355                 f2fs_err(sbi, "LFS is not compatible with checkpoint=disable");
1356                 return -EINVAL;
1357         }
1358
1359         if (test_opt(sbi, ATGC) && f2fs_lfs_mode(sbi)) {
1360                 f2fs_err(sbi, "LFS is not compatible with ATGC");
1361                 return -EINVAL;
1362         }
1363
1364         if (f2fs_is_readonly(sbi) && test_opt(sbi, FLUSH_MERGE)) {
1365                 f2fs_err(sbi, "FLUSH_MERGE not compatible with readonly mode");
1366                 return -EINVAL;
1367         }
1368
1369         if (f2fs_sb_has_readonly(sbi) && !f2fs_readonly(sbi->sb)) {
1370                 f2fs_err(sbi, "Allow to mount readonly mode only");
1371                 return -EROFS;
1372         }
1373         return 0;
1374 }
1375
1376 static struct inode *f2fs_alloc_inode(struct super_block *sb)
1377 {
1378         struct f2fs_inode_info *fi;
1379
1380         if (time_to_inject(F2FS_SB(sb), FAULT_SLAB_ALLOC))
1381                 return NULL;
1382
1383         fi = alloc_inode_sb(sb, f2fs_inode_cachep, GFP_F2FS_ZERO);
1384         if (!fi)
1385                 return NULL;
1386
1387         init_once((void *) fi);
1388
1389         /* Initialize f2fs-specific inode info */
1390         atomic_set(&fi->dirty_pages, 0);
1391         atomic_set(&fi->i_compr_blocks, 0);
1392         init_f2fs_rwsem(&fi->i_sem);
1393         spin_lock_init(&fi->i_size_lock);
1394         INIT_LIST_HEAD(&fi->dirty_list);
1395         INIT_LIST_HEAD(&fi->gdirty_list);
1396         init_f2fs_rwsem(&fi->i_gc_rwsem[READ]);
1397         init_f2fs_rwsem(&fi->i_gc_rwsem[WRITE]);
1398         init_f2fs_rwsem(&fi->i_xattr_sem);
1399
1400         /* Will be used by directory only */
1401         fi->i_dir_level = F2FS_SB(sb)->dir_level;
1402
1403         return &fi->vfs_inode;
1404 }
1405
1406 static int f2fs_drop_inode(struct inode *inode)
1407 {
1408         struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1409         int ret;
1410
1411         /*
1412          * during filesystem shutdown, if checkpoint is disabled,
1413          * drop useless meta/node dirty pages.
1414          */
1415         if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1416                 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1417                         inode->i_ino == F2FS_META_INO(sbi)) {
1418                         trace_f2fs_drop_inode(inode, 1);
1419                         return 1;
1420                 }
1421         }
1422
1423         /*
1424          * This is to avoid a deadlock condition like below.
1425          * writeback_single_inode(inode)
1426          *  - f2fs_write_data_page
1427          *    - f2fs_gc -> iput -> evict
1428          *       - inode_wait_for_writeback(inode)
1429          */
1430         if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
1431                 if (!inode->i_nlink && !is_bad_inode(inode)) {
1432                         /* to avoid evict_inode call simultaneously */
1433                         atomic_inc(&inode->i_count);
1434                         spin_unlock(&inode->i_lock);
1435
1436                         /* should remain fi->extent_tree for writepage */
1437                         f2fs_destroy_extent_node(inode);
1438
1439                         sb_start_intwrite(inode->i_sb);
1440                         f2fs_i_size_write(inode, 0);
1441
1442                         f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1443                                         inode, NULL, 0, DATA);
1444                         truncate_inode_pages_final(inode->i_mapping);
1445
1446                         if (F2FS_HAS_BLOCKS(inode))
1447                                 f2fs_truncate(inode);
1448
1449                         sb_end_intwrite(inode->i_sb);
1450
1451                         spin_lock(&inode->i_lock);
1452                         atomic_dec(&inode->i_count);
1453                 }
1454                 trace_f2fs_drop_inode(inode, 0);
1455                 return 0;
1456         }
1457         ret = generic_drop_inode(inode);
1458         if (!ret)
1459                 ret = fscrypt_drop_inode(inode);
1460         trace_f2fs_drop_inode(inode, ret);
1461         return ret;
1462 }
1463
1464 int f2fs_inode_dirtied(struct inode *inode, bool sync)
1465 {
1466         struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1467         int ret = 0;
1468
1469         spin_lock(&sbi->inode_lock[DIRTY_META]);
1470         if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1471                 ret = 1;
1472         } else {
1473                 set_inode_flag(inode, FI_DIRTY_INODE);
1474                 stat_inc_dirty_inode(sbi, DIRTY_META);
1475         }
1476         if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1477                 list_add_tail(&F2FS_I(inode)->gdirty_list,
1478                                 &sbi->inode_list[DIRTY_META]);
1479                 inc_page_count(sbi, F2FS_DIRTY_IMETA);
1480         }
1481         spin_unlock(&sbi->inode_lock[DIRTY_META]);
1482         return ret;
1483 }
1484
1485 void f2fs_inode_synced(struct inode *inode)
1486 {
1487         struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1488
1489         spin_lock(&sbi->inode_lock[DIRTY_META]);
1490         if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1491                 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1492                 return;
1493         }
1494         if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1495                 list_del_init(&F2FS_I(inode)->gdirty_list);
1496                 dec_page_count(sbi, F2FS_DIRTY_IMETA);
1497         }
1498         clear_inode_flag(inode, FI_DIRTY_INODE);
1499         clear_inode_flag(inode, FI_AUTO_RECOVER);
1500         stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1501         spin_unlock(&sbi->inode_lock[DIRTY_META]);
1502 }
1503
1504 /*
1505  * f2fs_dirty_inode() is called from __mark_inode_dirty()
1506  *
1507  * We should call set_dirty_inode to write the dirty inode through write_inode.
1508  */
1509 static void f2fs_dirty_inode(struct inode *inode, int flags)
1510 {
1511         struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1512
1513         if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1514                         inode->i_ino == F2FS_META_INO(sbi))
1515                 return;
1516
1517         if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1518                 clear_inode_flag(inode, FI_AUTO_RECOVER);
1519
1520         f2fs_inode_dirtied(inode, false);
1521 }
1522
1523 static void f2fs_free_inode(struct inode *inode)
1524 {
1525         fscrypt_free_inode(inode);
1526         kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1527 }
1528
1529 static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1530 {
1531         percpu_counter_destroy(&sbi->total_valid_inode_count);
1532         percpu_counter_destroy(&sbi->rf_node_block_count);
1533         percpu_counter_destroy(&sbi->alloc_valid_block_count);
1534 }
1535
1536 static void destroy_device_list(struct f2fs_sb_info *sbi)
1537 {
1538         int i;
1539
1540         for (i = 0; i < sbi->s_ndevs; i++) {
1541                 blkdev_put(FDEV(i).bdev, FMODE_EXCL);
1542 #ifdef CONFIG_BLK_DEV_ZONED
1543                 kvfree(FDEV(i).blkz_seq);
1544 #endif
1545         }
1546         kvfree(sbi->devs);
1547 }
1548
1549 static void f2fs_put_super(struct super_block *sb)
1550 {
1551         struct f2fs_sb_info *sbi = F2FS_SB(sb);
1552         int i;
1553         bool done;
1554
1555         /* unregister procfs/sysfs entries in advance to avoid race case */
1556         f2fs_unregister_sysfs(sbi);
1557
1558         f2fs_quota_off_umount(sb);
1559
1560         /* prevent remaining shrinker jobs */
1561         mutex_lock(&sbi->umount_mutex);
1562
1563         /*
1564          * flush all issued checkpoints and stop checkpoint issue thread.
1565          * after then, all checkpoints should be done by each process context.
1566          */
1567         f2fs_stop_ckpt_thread(sbi);
1568
1569         /*
1570          * We don't need to do checkpoint when superblock is clean.
1571          * But, the previous checkpoint was not done by umount, it needs to do
1572          * clean checkpoint again.
1573          */
1574         if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1575                         !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1576                 struct cp_control cpc = {
1577                         .reason = CP_UMOUNT,
1578                 };
1579                 f2fs_write_checkpoint(sbi, &cpc);
1580         }
1581
1582         /* be sure to wait for any on-going discard commands */
1583         done = f2fs_issue_discard_timeout(sbi);
1584         if (f2fs_realtime_discard_enable(sbi) && !sbi->discard_blks && done) {
1585                 struct cp_control cpc = {
1586                         .reason = CP_UMOUNT | CP_TRIMMED,
1587                 };
1588                 f2fs_write_checkpoint(sbi, &cpc);
1589         }
1590
1591         /*
1592          * normally superblock is clean, so we need to release this.
1593          * In addition, EIO will skip do checkpoint, we need this as well.
1594          */
1595         f2fs_release_ino_entry(sbi, true);
1596
1597         f2fs_leave_shrinker(sbi);
1598         mutex_unlock(&sbi->umount_mutex);
1599
1600         /* our cp_error case, we can wait for any writeback page */
1601         f2fs_flush_merged_writes(sbi);
1602
1603         f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1604
1605         f2fs_bug_on(sbi, sbi->fsync_node_num);
1606
1607         f2fs_destroy_compress_inode(sbi);
1608
1609         iput(sbi->node_inode);
1610         sbi->node_inode = NULL;
1611
1612         iput(sbi->meta_inode);
1613         sbi->meta_inode = NULL;
1614
1615         /*
1616          * iput() can update stat information, if f2fs_write_checkpoint()
1617          * above failed with error.
1618          */
1619         f2fs_destroy_stats(sbi);
1620
1621         /* destroy f2fs internal modules */
1622         f2fs_destroy_node_manager(sbi);
1623         f2fs_destroy_segment_manager(sbi);
1624
1625         f2fs_destroy_post_read_wq(sbi);
1626
1627         kvfree(sbi->ckpt);
1628
1629         sb->s_fs_info = NULL;
1630         if (sbi->s_chksum_driver)
1631                 crypto_free_shash(sbi->s_chksum_driver);
1632         kfree(sbi->raw_super);
1633
1634         destroy_device_list(sbi);
1635         f2fs_destroy_page_array_cache(sbi);
1636         f2fs_destroy_xattr_caches(sbi);
1637         mempool_destroy(sbi->write_io_dummy);
1638 #ifdef CONFIG_QUOTA
1639         for (i = 0; i < MAXQUOTAS; i++)
1640                 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1641 #endif
1642         fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
1643         destroy_percpu_info(sbi);
1644         f2fs_destroy_iostat(sbi);
1645         for (i = 0; i < NR_PAGE_TYPE; i++)
1646                 kvfree(sbi->write_io[i]);
1647 #if IS_ENABLED(CONFIG_UNICODE)
1648         utf8_unload(sb->s_encoding);
1649 #endif
1650         kfree(sbi);
1651 }
1652
1653 int f2fs_sync_fs(struct super_block *sb, int sync)
1654 {
1655         struct f2fs_sb_info *sbi = F2FS_SB(sb);
1656         int err = 0;
1657
1658         if (unlikely(f2fs_cp_error(sbi)))
1659                 return 0;
1660         if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1661                 return 0;
1662
1663         trace_f2fs_sync_fs(sb, sync);
1664
1665         if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1666                 return -EAGAIN;
1667
1668         if (sync)
1669                 err = f2fs_issue_checkpoint(sbi);
1670
1671         return err;
1672 }
1673
1674 static int f2fs_freeze(struct super_block *sb)
1675 {
1676         if (f2fs_readonly(sb))
1677                 return 0;
1678
1679         /* IO error happened before */
1680         if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1681                 return -EIO;
1682
1683         /* must be clean, since sync_filesystem() was already called */
1684         if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1685                 return -EINVAL;
1686
1687         /* Let's flush checkpoints and stop the thread. */
1688         f2fs_flush_ckpt_thread(F2FS_SB(sb));
1689
1690         /* to avoid deadlock on f2fs_evict_inode->SB_FREEZE_FS */
1691         set_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1692         return 0;
1693 }
1694
1695 static int f2fs_unfreeze(struct super_block *sb)
1696 {
1697         clear_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1698         return 0;
1699 }
1700
1701 #ifdef CONFIG_QUOTA
1702 static int f2fs_statfs_project(struct super_block *sb,
1703                                 kprojid_t projid, struct kstatfs *buf)
1704 {
1705         struct kqid qid;
1706         struct dquot *dquot;
1707         u64 limit;
1708         u64 curblock;
1709
1710         qid = make_kqid_projid(projid);
1711         dquot = dqget(sb, qid);
1712         if (IS_ERR(dquot))
1713                 return PTR_ERR(dquot);
1714         spin_lock(&dquot->dq_dqb_lock);
1715
1716         limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1717                                         dquot->dq_dqb.dqb_bhardlimit);
1718         if (limit)
1719                 limit >>= sb->s_blocksize_bits;
1720
1721         if (limit && buf->f_blocks > limit) {
1722                 curblock = (dquot->dq_dqb.dqb_curspace +
1723                             dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
1724                 buf->f_blocks = limit;
1725                 buf->f_bfree = buf->f_bavail =
1726                         (buf->f_blocks > curblock) ?
1727                          (buf->f_blocks - curblock) : 0;
1728         }
1729
1730         limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1731                                         dquot->dq_dqb.dqb_ihardlimit);
1732
1733         if (limit && buf->f_files > limit) {
1734                 buf->f_files = limit;
1735                 buf->f_ffree =
1736                         (buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1737                          (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1738         }
1739
1740         spin_unlock(&dquot->dq_dqb_lock);
1741         dqput(dquot);
1742         return 0;
1743 }
1744 #endif
1745
1746 static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1747 {
1748         struct super_block *sb = dentry->d_sb;
1749         struct f2fs_sb_info *sbi = F2FS_SB(sb);
1750         u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
1751         block_t total_count, user_block_count, start_count;
1752         u64 avail_node_count;
1753         unsigned int total_valid_node_count;
1754
1755         total_count = le64_to_cpu(sbi->raw_super->block_count);
1756         start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
1757         buf->f_type = F2FS_SUPER_MAGIC;
1758         buf->f_bsize = sbi->blocksize;
1759
1760         buf->f_blocks = total_count - start_count;
1761
1762         spin_lock(&sbi->stat_lock);
1763
1764         user_block_count = sbi->user_block_count;
1765         total_valid_node_count = valid_node_count(sbi);
1766         avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
1767         buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1768                                                 sbi->current_reserved_blocks;
1769
1770         if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1771                 buf->f_bfree = 0;
1772         else
1773                 buf->f_bfree -= sbi->unusable_block_count;
1774         spin_unlock(&sbi->stat_lock);
1775
1776         if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1777                 buf->f_bavail = buf->f_bfree -
1778                                 F2FS_OPTION(sbi).root_reserved_blocks;
1779         else
1780                 buf->f_bavail = 0;
1781
1782         if (avail_node_count > user_block_count) {
1783                 buf->f_files = user_block_count;
1784                 buf->f_ffree = buf->f_bavail;
1785         } else {
1786                 buf->f_files = avail_node_count;
1787                 buf->f_ffree = min(avail_node_count - total_valid_node_count,
1788                                         buf->f_bavail);
1789         }
1790
1791         buf->f_namelen = F2FS_NAME_LEN;
1792         buf->f_fsid    = u64_to_fsid(id);
1793
1794 #ifdef CONFIG_QUOTA
1795         if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1796                         sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1797                 f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1798         }
1799 #endif
1800         return 0;
1801 }
1802
1803 static inline void f2fs_show_quota_options(struct seq_file *seq,
1804                                            struct super_block *sb)
1805 {
1806 #ifdef CONFIG_QUOTA
1807         struct f2fs_sb_info *sbi = F2FS_SB(sb);
1808
1809         if (F2FS_OPTION(sbi).s_jquota_fmt) {
1810                 char *fmtname = "";
1811
1812                 switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1813                 case QFMT_VFS_OLD:
1814                         fmtname = "vfsold";
1815                         break;
1816                 case QFMT_VFS_V0:
1817                         fmtname = "vfsv0";
1818                         break;
1819                 case QFMT_VFS_V1:
1820                         fmtname = "vfsv1";
1821                         break;
1822                 }
1823                 seq_printf(seq, ",jqfmt=%s", fmtname);
1824         }
1825
1826         if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1827                 seq_show_option(seq, "usrjquota",
1828                         F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1829
1830         if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1831                 seq_show_option(seq, "grpjquota",
1832                         F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1833
1834         if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1835                 seq_show_option(seq, "prjjquota",
1836                         F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1837 #endif
1838 }
1839
1840 #ifdef CONFIG_F2FS_FS_COMPRESSION
1841 static inline void f2fs_show_compress_options(struct seq_file *seq,
1842                                                         struct super_block *sb)
1843 {
1844         struct f2fs_sb_info *sbi = F2FS_SB(sb);
1845         char *algtype = "";
1846         int i;
1847
1848         if (!f2fs_sb_has_compression(sbi))
1849                 return;
1850
1851         switch (F2FS_OPTION(sbi).compress_algorithm) {
1852         case COMPRESS_LZO:
1853                 algtype = "lzo";
1854                 break;
1855         case COMPRESS_LZ4:
1856                 algtype = "lz4";
1857                 break;
1858         case COMPRESS_ZSTD:
1859                 algtype = "zstd";
1860                 break;
1861         case COMPRESS_LZORLE:
1862                 algtype = "lzo-rle";
1863                 break;
1864         }
1865         seq_printf(seq, ",compress_algorithm=%s", algtype);
1866
1867         if (F2FS_OPTION(sbi).compress_level)
1868                 seq_printf(seq, ":%d", F2FS_OPTION(sbi).compress_level);
1869
1870         seq_printf(seq, ",compress_log_size=%u",
1871                         F2FS_OPTION(sbi).compress_log_size);
1872
1873         for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1874                 seq_printf(seq, ",compress_extension=%s",
1875                         F2FS_OPTION(sbi).extensions[i]);
1876         }
1877
1878         for (i = 0; i < F2FS_OPTION(sbi).nocompress_ext_cnt; i++) {
1879                 seq_printf(seq, ",nocompress_extension=%s",
1880                         F2FS_OPTION(sbi).noextensions[i]);
1881         }
1882
1883         if (F2FS_OPTION(sbi).compress_chksum)
1884                 seq_puts(seq, ",compress_chksum");
1885
1886         if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_FS)
1887                 seq_printf(seq, ",compress_mode=%s", "fs");
1888         else if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_USER)
1889                 seq_printf(seq, ",compress_mode=%s", "user");
1890
1891         if (test_opt(sbi, COMPRESS_CACHE))
1892                 seq_puts(seq, ",compress_cache");
1893 }
1894 #endif
1895
1896 static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1897 {
1898         struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1899
1900         if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1901                 seq_printf(seq, ",background_gc=%s", "sync");
1902         else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1903                 seq_printf(seq, ",background_gc=%s", "on");
1904         else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1905                 seq_printf(seq, ",background_gc=%s", "off");
1906
1907         if (test_opt(sbi, GC_MERGE))
1908                 seq_puts(seq, ",gc_merge");
1909         else
1910                 seq_puts(seq, ",nogc_merge");
1911
1912         if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1913                 seq_puts(seq, ",disable_roll_forward");
1914         if (test_opt(sbi, NORECOVERY))
1915                 seq_puts(seq, ",norecovery");
1916         if (test_opt(sbi, DISCARD)) {
1917                 seq_puts(seq, ",discard");
1918                 if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK)
1919                         seq_printf(seq, ",discard_unit=%s", "block");
1920                 else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SEGMENT)
1921                         seq_printf(seq, ",discard_unit=%s", "segment");
1922                 else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SECTION)
1923                         seq_printf(seq, ",discard_unit=%s", "section");
1924         } else {
1925                 seq_puts(seq, ",nodiscard");
1926         }
1927         if (test_opt(sbi, NOHEAP))
1928                 seq_puts(seq, ",no_heap");
1929         else
1930                 seq_puts(seq, ",heap");
1931 #ifdef CONFIG_F2FS_FS_XATTR
1932         if (test_opt(sbi, XATTR_USER))
1933                 seq_puts(seq, ",user_xattr");
1934         else
1935                 seq_puts(seq, ",nouser_xattr");
1936         if (test_opt(sbi, INLINE_XATTR))
1937                 seq_puts(seq, ",inline_xattr");
1938         else
1939                 seq_puts(seq, ",noinline_xattr");
1940         if (test_opt(sbi, INLINE_XATTR_SIZE))
1941                 seq_printf(seq, ",inline_xattr_size=%u",
1942                                         F2FS_OPTION(sbi).inline_xattr_size);
1943 #endif
1944 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1945         if (test_opt(sbi, POSIX_ACL))
1946                 seq_puts(seq, ",acl");
1947         else
1948                 seq_puts(seq, ",noacl");
1949 #endif
1950         if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
1951                 seq_puts(seq, ",disable_ext_identify");
1952         if (test_opt(sbi, INLINE_DATA))
1953                 seq_puts(seq, ",inline_data");
1954         else
1955                 seq_puts(seq, ",noinline_data");
1956         if (test_opt(sbi, INLINE_DENTRY))
1957                 seq_puts(seq, ",inline_dentry");
1958         else
1959                 seq_puts(seq, ",noinline_dentry");
1960         if (test_opt(sbi, FLUSH_MERGE))
1961                 seq_puts(seq, ",flush_merge");
1962         else
1963                 seq_puts(seq, ",noflush_merge");
1964         if (test_opt(sbi, NOBARRIER))
1965                 seq_puts(seq, ",nobarrier");
1966         else
1967                 seq_puts(seq, ",barrier");
1968         if (test_opt(sbi, FASTBOOT))
1969                 seq_puts(seq, ",fastboot");
1970         if (test_opt(sbi, READ_EXTENT_CACHE))
1971                 seq_puts(seq, ",extent_cache");
1972         else
1973                 seq_puts(seq, ",noextent_cache");
1974         if (test_opt(sbi, AGE_EXTENT_CACHE))
1975                 seq_puts(seq, ",age_extent_cache");
1976         if (test_opt(sbi, DATA_FLUSH))
1977                 seq_puts(seq, ",data_flush");
1978
1979         seq_puts(seq, ",mode=");
1980         if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
1981                 seq_puts(seq, "adaptive");
1982         else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
1983                 seq_puts(seq, "lfs");
1984         else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG)
1985                 seq_puts(seq, "fragment:segment");
1986         else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK)
1987                 seq_puts(seq, "fragment:block");
1988         seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
1989         if (test_opt(sbi, RESERVE_ROOT))
1990                 seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
1991                                 F2FS_OPTION(sbi).root_reserved_blocks,
1992                                 from_kuid_munged(&init_user_ns,
1993                                         F2FS_OPTION(sbi).s_resuid),
1994                                 from_kgid_munged(&init_user_ns,
1995                                         F2FS_OPTION(sbi).s_resgid));
1996         if (F2FS_IO_SIZE_BITS(sbi))
1997                 seq_printf(seq, ",io_bits=%u",
1998                                 F2FS_OPTION(sbi).write_io_size_bits);
1999 #ifdef CONFIG_F2FS_FAULT_INJECTION
2000         if (test_opt(sbi, FAULT_INJECTION)) {
2001                 seq_printf(seq, ",fault_injection=%u",
2002                                 F2FS_OPTION(sbi).fault_info.inject_rate);
2003                 seq_printf(seq, ",fault_type=%u",
2004                                 F2FS_OPTION(sbi).fault_info.inject_type);
2005         }
2006 #endif
2007 #ifdef CONFIG_QUOTA
2008         if (test_opt(sbi, QUOTA))
2009                 seq_puts(seq, ",quota");
2010         if (test_opt(sbi, USRQUOTA))
2011                 seq_puts(seq, ",usrquota");
2012         if (test_opt(sbi, GRPQUOTA))
2013                 seq_puts(seq, ",grpquota");
2014         if (test_opt(sbi, PRJQUOTA))
2015                 seq_puts(seq, ",prjquota");
2016 #endif
2017         f2fs_show_quota_options(seq, sbi->sb);
2018
2019         fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
2020
2021         if (sbi->sb->s_flags & SB_INLINECRYPT)
2022                 seq_puts(seq, ",inlinecrypt");
2023
2024         if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
2025                 seq_printf(seq, ",alloc_mode=%s", "default");
2026         else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
2027                 seq_printf(seq, ",alloc_mode=%s", "reuse");
2028
2029         if (test_opt(sbi, DISABLE_CHECKPOINT))
2030                 seq_printf(seq, ",checkpoint=disable:%u",
2031                                 F2FS_OPTION(sbi).unusable_cap);
2032         if (test_opt(sbi, MERGE_CHECKPOINT))
2033                 seq_puts(seq, ",checkpoint_merge");
2034         else
2035                 seq_puts(seq, ",nocheckpoint_merge");
2036         if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
2037                 seq_printf(seq, ",fsync_mode=%s", "posix");
2038         else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
2039                 seq_printf(seq, ",fsync_mode=%s", "strict");
2040         else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
2041                 seq_printf(seq, ",fsync_mode=%s", "nobarrier");
2042
2043 #ifdef CONFIG_F2FS_FS_COMPRESSION
2044         f2fs_show_compress_options(seq, sbi->sb);
2045 #endif
2046
2047         if (test_opt(sbi, ATGC))
2048                 seq_puts(seq, ",atgc");
2049
2050         if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_NORMAL)
2051                 seq_printf(seq, ",memory=%s", "normal");
2052         else if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW)
2053                 seq_printf(seq, ",memory=%s", "low");
2054
2055         return 0;
2056 }
2057
2058 static void default_options(struct f2fs_sb_info *sbi)
2059 {
2060         /* init some FS parameters */
2061         if (f2fs_sb_has_readonly(sbi))
2062                 F2FS_OPTION(sbi).active_logs = NR_CURSEG_RO_TYPE;
2063         else
2064                 F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
2065
2066         F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
2067         if (le32_to_cpu(F2FS_RAW_SUPER(sbi)->segment_count_main) <=
2068                                                         SMALL_VOLUME_SEGMENTS)
2069                 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
2070         else
2071                 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
2072         F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
2073         F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
2074         F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
2075         if (f2fs_sb_has_compression(sbi)) {
2076                 F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
2077                 F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
2078                 F2FS_OPTION(sbi).compress_ext_cnt = 0;
2079                 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
2080         }
2081         F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
2082         F2FS_OPTION(sbi).memory_mode = MEMORY_MODE_NORMAL;
2083
2084         sbi->sb->s_flags &= ~SB_INLINECRYPT;
2085
2086         set_opt(sbi, INLINE_XATTR);
2087         set_opt(sbi, INLINE_DATA);
2088         set_opt(sbi, INLINE_DENTRY);
2089         set_opt(sbi, READ_EXTENT_CACHE);
2090         set_opt(sbi, NOHEAP);
2091         clear_opt(sbi, DISABLE_CHECKPOINT);
2092         set_opt(sbi, MERGE_CHECKPOINT);
2093         F2FS_OPTION(sbi).unusable_cap = 0;
2094         sbi->sb->s_flags |= SB_LAZYTIME;
2095         if (!f2fs_is_readonly(sbi))
2096                 set_opt(sbi, FLUSH_MERGE);
2097         if (f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi))
2098                 set_opt(sbi, DISCARD);
2099         if (f2fs_sb_has_blkzoned(sbi)) {
2100                 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
2101                 F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_SECTION;
2102         } else {
2103                 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
2104                 F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_BLOCK;
2105         }
2106
2107 #ifdef CONFIG_F2FS_FS_XATTR
2108         set_opt(sbi, XATTR_USER);
2109 #endif
2110 #ifdef CONFIG_F2FS_FS_POSIX_ACL
2111         set_opt(sbi, POSIX_ACL);
2112 #endif
2113
2114         f2fs_build_fault_attr(sbi, 0, 0);
2115 }
2116
2117 #ifdef CONFIG_QUOTA
2118 static int f2fs_enable_quotas(struct super_block *sb);
2119 #endif
2120
2121 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
2122 {
2123         unsigned int s_flags = sbi->sb->s_flags;
2124         struct cp_control cpc;
2125         unsigned int gc_mode = sbi->gc_mode;
2126         int err = 0;
2127         int ret;
2128         block_t unusable;
2129
2130         if (s_flags & SB_RDONLY) {
2131                 f2fs_err(sbi, "checkpoint=disable on readonly fs");
2132                 return -EINVAL;
2133         }
2134         sbi->sb->s_flags |= SB_ACTIVE;
2135
2136         /* check if we need more GC first */
2137         unusable = f2fs_get_unusable_blocks(sbi);
2138         if (!f2fs_disable_cp_again(sbi, unusable))
2139                 goto skip_gc;
2140
2141         f2fs_update_time(sbi, DISABLE_TIME);
2142
2143         sbi->gc_mode = GC_URGENT_HIGH;
2144
2145         while (!f2fs_time_over(sbi, DISABLE_TIME)) {
2146                 struct f2fs_gc_control gc_control = {
2147                         .victim_segno = NULL_SEGNO,
2148                         .init_gc_type = FG_GC,
2149                         .should_migrate_blocks = false,
2150                         .err_gc_skipped = true,
2151                         .nr_free_secs = 1 };
2152
2153                 f2fs_down_write(&sbi->gc_lock);
2154                 err = f2fs_gc(sbi, &gc_control);
2155                 if (err == -ENODATA) {
2156                         err = 0;
2157                         break;
2158                 }
2159                 if (err && err != -EAGAIN)
2160                         break;
2161         }
2162
2163         ret = sync_filesystem(sbi->sb);
2164         if (ret || err) {
2165                 err = ret ? ret : err;
2166                 goto restore_flag;
2167         }
2168
2169         unusable = f2fs_get_unusable_blocks(sbi);
2170         if (f2fs_disable_cp_again(sbi, unusable)) {
2171                 err = -EAGAIN;
2172                 goto restore_flag;
2173         }
2174
2175 skip_gc:
2176         f2fs_down_write(&sbi->gc_lock);
2177         cpc.reason = CP_PAUSE;
2178         set_sbi_flag(sbi, SBI_CP_DISABLED);
2179         err = f2fs_write_checkpoint(sbi, &cpc);
2180         if (err)
2181                 goto out_unlock;
2182
2183         spin_lock(&sbi->stat_lock);
2184         sbi->unusable_block_count = unusable;
2185         spin_unlock(&sbi->stat_lock);
2186
2187 out_unlock:
2188         f2fs_up_write(&sbi->gc_lock);
2189 restore_flag:
2190         sbi->gc_mode = gc_mode;
2191         sbi->sb->s_flags = s_flags;     /* Restore SB_RDONLY status */
2192         return err;
2193 }
2194
2195 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
2196 {
2197         int retry = DEFAULT_RETRY_IO_COUNT;
2198
2199         /* we should flush all the data to keep data consistency */
2200         do {
2201                 sync_inodes_sb(sbi->sb);
2202                 f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2203         } while (get_pages(sbi, F2FS_DIRTY_DATA) && retry--);
2204
2205         if (unlikely(retry < 0))
2206                 f2fs_warn(sbi, "checkpoint=enable has some unwritten data.");
2207
2208         f2fs_down_write(&sbi->gc_lock);
2209         f2fs_dirty_to_prefree(sbi);
2210
2211         clear_sbi_flag(sbi, SBI_CP_DISABLED);
2212         set_sbi_flag(sbi, SBI_IS_DIRTY);
2213         f2fs_up_write(&sbi->gc_lock);
2214
2215         f2fs_sync_fs(sbi->sb, 1);
2216
2217         /* Let's ensure there's no pending checkpoint anymore */
2218         f2fs_flush_ckpt_thread(sbi);
2219 }
2220
2221 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
2222 {
2223         struct f2fs_sb_info *sbi = F2FS_SB(sb);
2224         struct f2fs_mount_info org_mount_opt;
2225         unsigned long old_sb_flags;
2226         int err;
2227         bool need_restart_gc = false, need_stop_gc = false;
2228         bool need_restart_ckpt = false, need_stop_ckpt = false;
2229         bool need_restart_flush = false, need_stop_flush = false;
2230         bool need_restart_discard = false, need_stop_discard = false;
2231         bool no_read_extent_cache = !test_opt(sbi, READ_EXTENT_CACHE);
2232         bool no_age_extent_cache = !test_opt(sbi, AGE_EXTENT_CACHE);
2233         bool enable_checkpoint = !test_opt(sbi, DISABLE_CHECKPOINT);
2234         bool no_io_align = !F2FS_IO_ALIGNED(sbi);
2235         bool no_atgc = !test_opt(sbi, ATGC);
2236         bool no_discard = !test_opt(sbi, DISCARD);
2237         bool no_compress_cache = !test_opt(sbi, COMPRESS_CACHE);
2238         bool block_unit_discard = f2fs_block_unit_discard(sbi);
2239 #ifdef CONFIG_QUOTA
2240         int i, j;
2241 #endif
2242
2243         /*
2244          * Save the old mount options in case we
2245          * need to restore them.
2246          */
2247         org_mount_opt = sbi->mount_opt;
2248         old_sb_flags = sb->s_flags;
2249
2250 #ifdef CONFIG_QUOTA
2251         org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
2252         for (i = 0; i < MAXQUOTAS; i++) {
2253                 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2254                         org_mount_opt.s_qf_names[i] =
2255                                 kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
2256                                 GFP_KERNEL);
2257                         if (!org_mount_opt.s_qf_names[i]) {
2258                                 for (j = 0; j < i; j++)
2259                                         kfree(org_mount_opt.s_qf_names[j]);
2260                                 return -ENOMEM;
2261                         }
2262                 } else {
2263                         org_mount_opt.s_qf_names[i] = NULL;
2264                 }
2265         }
2266 #endif
2267
2268         /* recover superblocks we couldn't write due to previous RO mount */
2269         if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
2270                 err = f2fs_commit_super(sbi, false);
2271                 f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
2272                           err);
2273                 if (!err)
2274                         clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2275         }
2276
2277         default_options(sbi);
2278
2279         /* parse mount options */
2280         err = parse_options(sb, data, true);
2281         if (err)
2282                 goto restore_opts;
2283
2284         /*
2285          * Previous and new state of filesystem is RO,
2286          * so skip checking GC and FLUSH_MERGE conditions.
2287          */
2288         if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
2289                 goto skip;
2290
2291         if (f2fs_dev_is_readonly(sbi) && !(*flags & SB_RDONLY)) {
2292                 err = -EROFS;
2293                 goto restore_opts;
2294         }
2295
2296 #ifdef CONFIG_QUOTA
2297         if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
2298                 err = dquot_suspend(sb, -1);
2299                 if (err < 0)
2300                         goto restore_opts;
2301         } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
2302                 /* dquot_resume needs RW */
2303                 sb->s_flags &= ~SB_RDONLY;
2304                 if (sb_any_quota_suspended(sb)) {
2305                         dquot_resume(sb, -1);
2306                 } else if (f2fs_sb_has_quota_ino(sbi)) {
2307                         err = f2fs_enable_quotas(sb);
2308                         if (err)
2309                                 goto restore_opts;
2310                 }
2311         }
2312 #endif
2313         if (f2fs_lfs_mode(sbi) && !IS_F2FS_IPU_DISABLE(sbi)) {
2314                 err = -EINVAL;
2315                 f2fs_warn(sbi, "LFS is not compatible with IPU");
2316                 goto restore_opts;
2317         }
2318
2319         /* disallow enable atgc dynamically */
2320         if (no_atgc == !!test_opt(sbi, ATGC)) {
2321                 err = -EINVAL;
2322                 f2fs_warn(sbi, "switch atgc option is not allowed");
2323                 goto restore_opts;
2324         }
2325
2326         /* disallow enable/disable extent_cache dynamically */
2327         if (no_read_extent_cache == !!test_opt(sbi, READ_EXTENT_CACHE)) {
2328                 err = -EINVAL;
2329                 f2fs_warn(sbi, "switch extent_cache option is not allowed");
2330                 goto restore_opts;
2331         }
2332         /* disallow enable/disable age extent_cache dynamically */
2333         if (no_age_extent_cache == !!test_opt(sbi, AGE_EXTENT_CACHE)) {
2334                 err = -EINVAL;
2335                 f2fs_warn(sbi, "switch age_extent_cache option is not allowed");
2336                 goto restore_opts;
2337         }
2338
2339         if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
2340                 err = -EINVAL;
2341                 f2fs_warn(sbi, "switch io_bits option is not allowed");
2342                 goto restore_opts;
2343         }
2344
2345         if (no_compress_cache == !!test_opt(sbi, COMPRESS_CACHE)) {
2346                 err = -EINVAL;
2347                 f2fs_warn(sbi, "switch compress_cache option is not allowed");
2348                 goto restore_opts;
2349         }
2350
2351         if (block_unit_discard != f2fs_block_unit_discard(sbi)) {
2352                 err = -EINVAL;
2353                 f2fs_warn(sbi, "switch discard_unit option is not allowed");
2354                 goto restore_opts;
2355         }
2356
2357         if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
2358                 err = -EINVAL;
2359                 f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
2360                 goto restore_opts;
2361         }
2362
2363         /*
2364          * We stop the GC thread if FS is mounted as RO
2365          * or if background_gc = off is passed in mount
2366          * option. Also sync the filesystem.
2367          */
2368         if ((*flags & SB_RDONLY) ||
2369                         (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF &&
2370                         !test_opt(sbi, GC_MERGE))) {
2371                 if (sbi->gc_thread) {
2372                         f2fs_stop_gc_thread(sbi);
2373                         need_restart_gc = true;
2374                 }
2375         } else if (!sbi->gc_thread) {
2376                 err = f2fs_start_gc_thread(sbi);
2377                 if (err)
2378                         goto restore_opts;
2379                 need_stop_gc = true;
2380         }
2381
2382         if (*flags & SB_RDONLY) {
2383                 sync_inodes_sb(sb);
2384
2385                 set_sbi_flag(sbi, SBI_IS_DIRTY);
2386                 set_sbi_flag(sbi, SBI_IS_CLOSE);
2387                 f2fs_sync_fs(sb, 1);
2388                 clear_sbi_flag(sbi, SBI_IS_CLOSE);
2389         }
2390
2391         if ((*flags & SB_RDONLY) || test_opt(sbi, DISABLE_CHECKPOINT) ||
2392                         !test_opt(sbi, MERGE_CHECKPOINT)) {
2393                 f2fs_stop_ckpt_thread(sbi);
2394                 need_restart_ckpt = true;
2395         } else {
2396                 /* Flush if the prevous checkpoint, if exists. */
2397                 f2fs_flush_ckpt_thread(sbi);
2398
2399                 err = f2fs_start_ckpt_thread(sbi);
2400                 if (err) {
2401                         f2fs_err(sbi,
2402                             "Failed to start F2FS issue_checkpoint_thread (%d)",
2403                             err);
2404                         goto restore_gc;
2405                 }
2406                 need_stop_ckpt = true;
2407         }
2408
2409         /*
2410          * We stop issue flush thread if FS is mounted as RO
2411          * or if flush_merge is not passed in mount option.
2412          */
2413         if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2414                 clear_opt(sbi, FLUSH_MERGE);
2415                 f2fs_destroy_flush_cmd_control(sbi, false);
2416                 need_restart_flush = true;
2417         } else {
2418                 err = f2fs_create_flush_cmd_control(sbi);
2419                 if (err)
2420                         goto restore_ckpt;
2421                 need_stop_flush = true;
2422         }
2423
2424         if (no_discard == !!test_opt(sbi, DISCARD)) {
2425                 if (test_opt(sbi, DISCARD)) {
2426                         err = f2fs_start_discard_thread(sbi);
2427                         if (err)
2428                                 goto restore_flush;
2429                         need_stop_discard = true;
2430                 } else {
2431                         f2fs_stop_discard_thread(sbi);
2432                         f2fs_issue_discard_timeout(sbi);
2433                         need_restart_discard = true;
2434                 }
2435         }
2436
2437         if (enable_checkpoint == !!test_opt(sbi, DISABLE_CHECKPOINT)) {
2438                 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
2439                         err = f2fs_disable_checkpoint(sbi);
2440                         if (err)
2441                                 goto restore_discard;
2442                 } else {
2443                         f2fs_enable_checkpoint(sbi);
2444                 }
2445         }
2446
2447 skip:
2448 #ifdef CONFIG_QUOTA
2449         /* Release old quota file names */
2450         for (i = 0; i < MAXQUOTAS; i++)
2451                 kfree(org_mount_opt.s_qf_names[i]);
2452 #endif
2453         /* Update the POSIXACL Flag */
2454         sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
2455                 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
2456
2457         limit_reserve_root(sbi);
2458         adjust_unusable_cap_perc(sbi);
2459         *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
2460         return 0;
2461 restore_discard:
2462         if (need_restart_discard) {
2463                 if (f2fs_start_discard_thread(sbi))
2464                         f2fs_warn(sbi, "discard has been stopped");
2465         } else if (need_stop_discard) {
2466                 f2fs_stop_discard_thread(sbi);
2467         }
2468 restore_flush:
2469         if (need_restart_flush) {
2470                 if (f2fs_create_flush_cmd_control(sbi))
2471                         f2fs_warn(sbi, "background flush thread has stopped");
2472         } else if (need_stop_flush) {
2473                 clear_opt(sbi, FLUSH_MERGE);
2474                 f2fs_destroy_flush_cmd_control(sbi, false);
2475         }
2476 restore_ckpt:
2477         if (need_restart_ckpt) {
2478                 if (f2fs_start_ckpt_thread(sbi))
2479                         f2fs_warn(sbi, "background ckpt thread has stopped");
2480         } else if (need_stop_ckpt) {
2481                 f2fs_stop_ckpt_thread(sbi);
2482         }
2483 restore_gc:
2484         if (need_restart_gc) {
2485                 if (f2fs_start_gc_thread(sbi))
2486                         f2fs_warn(sbi, "background gc thread has stopped");
2487         } else if (need_stop_gc) {
2488                 f2fs_stop_gc_thread(sbi);
2489         }
2490 restore_opts:
2491 #ifdef CONFIG_QUOTA
2492         F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
2493         for (i = 0; i < MAXQUOTAS; i++) {
2494                 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
2495                 F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
2496         }
2497 #endif
2498         sbi->mount_opt = org_mount_opt;
2499         sb->s_flags = old_sb_flags;
2500         return err;
2501 }
2502
2503 #ifdef CONFIG_QUOTA
2504 static bool f2fs_need_recovery(struct f2fs_sb_info *sbi)
2505 {
2506         /* need to recovery orphan */
2507         if (is_set_ckpt_flags(sbi, CP_ORPHAN_PRESENT_FLAG))
2508                 return true;
2509         /* need to recovery data */
2510         if (test_opt(sbi, DISABLE_ROLL_FORWARD))
2511                 return false;
2512         if (test_opt(sbi, NORECOVERY))
2513                 return false;
2514         return !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG);
2515 }
2516
2517 static bool f2fs_recover_quota_begin(struct f2fs_sb_info *sbi)
2518 {
2519         bool readonly = f2fs_readonly(sbi->sb);
2520
2521         if (!f2fs_need_recovery(sbi))
2522                 return false;
2523
2524         /* it doesn't need to check f2fs_sb_has_readonly() */
2525         if (f2fs_hw_is_readonly(sbi))
2526                 return false;
2527
2528         if (readonly) {
2529                 sbi->sb->s_flags &= ~SB_RDONLY;
2530                 set_sbi_flag(sbi, SBI_IS_WRITABLE);
2531         }
2532
2533         /*
2534          * Turn on quotas which were not enabled for read-only mounts if
2535          * filesystem has quota feature, so that they are updated correctly.
2536          */
2537         return f2fs_enable_quota_files(sbi, readonly);
2538 }
2539
2540 static void f2fs_recover_quota_end(struct f2fs_sb_info *sbi,
2541                                                 bool quota_enabled)
2542 {
2543         if (quota_enabled)
2544                 f2fs_quota_off_umount(sbi->sb);
2545
2546         if (is_sbi_flag_set(sbi, SBI_IS_WRITABLE)) {
2547                 clear_sbi_flag(sbi, SBI_IS_WRITABLE);
2548                 sbi->sb->s_flags |= SB_RDONLY;
2549         }
2550 }
2551
2552 /* Read data from quotafile */
2553 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
2554                                size_t len, loff_t off)
2555 {
2556         struct inode *inode = sb_dqopt(sb)->files[type];
2557         struct address_space *mapping = inode->i_mapping;
2558         block_t blkidx = F2FS_BYTES_TO_BLK(off);
2559         int offset = off & (sb->s_blocksize - 1);
2560         int tocopy;
2561         size_t toread;
2562         loff_t i_size = i_size_read(inode);
2563         struct page *page;
2564
2565         if (off > i_size)
2566                 return 0;
2567
2568         if (off + len > i_size)
2569                 len = i_size - off;
2570         toread = len;
2571         while (toread > 0) {
2572                 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
2573 repeat:
2574                 page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
2575                 if (IS_ERR(page)) {
2576                         if (PTR_ERR(page) == -ENOMEM) {
2577                                 memalloc_retry_wait(GFP_NOFS);
2578                                 goto repeat;
2579                         }
2580                         set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2581                         return PTR_ERR(page);
2582                 }
2583
2584                 lock_page(page);
2585
2586                 if (unlikely(page->mapping != mapping)) {
2587                         f2fs_put_page(page, 1);
2588                         goto repeat;
2589                 }
2590                 if (unlikely(!PageUptodate(page))) {
2591                         f2fs_put_page(page, 1);
2592                         set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2593                         return -EIO;
2594                 }
2595
2596                 memcpy_from_page(data, page, offset, tocopy);
2597                 f2fs_put_page(page, 1);
2598
2599                 offset = 0;
2600                 toread -= tocopy;
2601                 data += tocopy;
2602                 blkidx++;
2603         }
2604         return len;
2605 }
2606
2607 /* Write to quotafile */
2608 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2609                                 const char *data, size_t len, loff_t off)
2610 {
2611         struct inode *inode = sb_dqopt(sb)->files[type];
2612         struct address_space *mapping = inode->i_mapping;
2613         const struct address_space_operations *a_ops = mapping->a_ops;
2614         int offset = off & (sb->s_blocksize - 1);
2615         size_t towrite = len;
2616         struct page *page;
2617         void *fsdata = NULL;
2618         int err = 0;
2619         int tocopy;
2620
2621         while (towrite > 0) {
2622                 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2623                                                                 towrite);
2624 retry:
2625                 err = a_ops->write_begin(NULL, mapping, off, tocopy,
2626                                                         &page, &fsdata);
2627                 if (unlikely(err)) {
2628                         if (err == -ENOMEM) {
2629                                 f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2630                                 goto retry;
2631                         }
2632                         set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2633                         break;
2634                 }
2635
2636                 memcpy_to_page(page, offset, data, tocopy);
2637
2638                 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2639                                                 page, fsdata);
2640                 offset = 0;
2641                 towrite -= tocopy;
2642                 off += tocopy;
2643                 data += tocopy;
2644                 cond_resched();
2645         }
2646
2647         if (len == towrite)
2648                 return err;
2649         inode->i_mtime = inode->i_ctime = current_time(inode);
2650         f2fs_mark_inode_dirty_sync(inode, false);
2651         return len - towrite;
2652 }
2653
2654 int f2fs_dquot_initialize(struct inode *inode)
2655 {
2656         if (time_to_inject(F2FS_I_SB(inode), FAULT_DQUOT_INIT))
2657                 return -ESRCH;
2658
2659         return dquot_initialize(inode);
2660 }
2661
2662 static struct dquot **f2fs_get_dquots(struct inode *inode)
2663 {
2664         return F2FS_I(inode)->i_dquot;
2665 }
2666
2667 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2668 {
2669         return &F2FS_I(inode)->i_reserved_quota;
2670 }
2671
2672 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2673 {
2674         if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2675                 f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2676                 return 0;
2677         }
2678
2679         return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2680                                         F2FS_OPTION(sbi).s_jquota_fmt, type);
2681 }
2682
2683 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2684 {
2685         int enabled = 0;
2686         int i, err;
2687
2688         if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2689                 err = f2fs_enable_quotas(sbi->sb);
2690                 if (err) {
2691                         f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2692                         return 0;
2693                 }
2694                 return 1;
2695         }
2696
2697         for (i = 0; i < MAXQUOTAS; i++) {
2698                 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2699                         err = f2fs_quota_on_mount(sbi, i);
2700                         if (!err) {
2701                                 enabled = 1;
2702                                 continue;
2703                         }
2704                         f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2705                                  err, i);
2706                 }
2707         }
2708         return enabled;
2709 }
2710
2711 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2712                              unsigned int flags)
2713 {
2714         struct inode *qf_inode;
2715         unsigned long qf_inum;
2716         int err;
2717
2718         BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2719
2720         qf_inum = f2fs_qf_ino(sb, type);
2721         if (!qf_inum)
2722                 return -EPERM;
2723
2724         qf_inode = f2fs_iget(sb, qf_inum);
2725         if (IS_ERR(qf_inode)) {
2726                 f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2727                 return PTR_ERR(qf_inode);
2728         }
2729
2730         /* Don't account quota for quota files to avoid recursion */
2731         qf_inode->i_flags |= S_NOQUOTA;
2732         err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2733         iput(qf_inode);
2734         return err;
2735 }
2736
2737 static int f2fs_enable_quotas(struct super_block *sb)
2738 {
2739         struct f2fs_sb_info *sbi = F2FS_SB(sb);
2740         int type, err = 0;
2741         unsigned long qf_inum;
2742         bool quota_mopt[MAXQUOTAS] = {
2743                 test_opt(sbi, USRQUOTA),
2744                 test_opt(sbi, GRPQUOTA),
2745                 test_opt(sbi, PRJQUOTA),
2746         };
2747
2748         if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2749                 f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2750                 return 0;
2751         }
2752
2753         sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2754
2755         for (type = 0; type < MAXQUOTAS; type++) {
2756                 qf_inum = f2fs_qf_ino(sb, type);
2757                 if (qf_inum) {
2758                         err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2759                                 DQUOT_USAGE_ENABLED |
2760                                 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2761                         if (err) {
2762                                 f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2763                                          type, err);
2764                                 for (type--; type >= 0; type--)
2765                                         dquot_quota_off(sb, type);
2766                                 set_sbi_flag(F2FS_SB(sb),
2767                                                 SBI_QUOTA_NEED_REPAIR);
2768                                 return err;
2769                         }
2770                 }
2771         }
2772         return 0;
2773 }
2774
2775 static int f2fs_quota_sync_file(struct f2fs_sb_info *sbi, int type)
2776 {
2777         struct quota_info *dqopt = sb_dqopt(sbi->sb);
2778         struct address_space *mapping = dqopt->files[type]->i_mapping;
2779         int ret = 0;
2780
2781         ret = dquot_writeback_dquots(sbi->sb, type);
2782         if (ret)
2783                 goto out;
2784
2785         ret = filemap_fdatawrite(mapping);
2786         if (ret)
2787                 goto out;
2788
2789         /* if we are using journalled quota */
2790         if (is_journalled_quota(sbi))
2791                 goto out;
2792
2793         ret = filemap_fdatawait(mapping);
2794
2795         truncate_inode_pages(&dqopt->files[type]->i_data, 0);
2796 out:
2797         if (ret)
2798                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2799         return ret;
2800 }
2801
2802 int f2fs_quota_sync(struct super_block *sb, int type)
2803 {
2804         struct f2fs_sb_info *sbi = F2FS_SB(sb);
2805         struct quota_info *dqopt = sb_dqopt(sb);
2806         int cnt;
2807         int ret = 0;
2808
2809         /*
2810          * Now when everything is written we can discard the pagecache so
2811          * that userspace sees the changes.
2812          */
2813         for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2814
2815                 if (type != -1 && cnt != type)
2816                         continue;
2817
2818                 if (!sb_has_quota_active(sb, cnt))
2819                         continue;
2820
2821                 if (!f2fs_sb_has_quota_ino(sbi))
2822                         inode_lock(dqopt->files[cnt]);
2823
2824                 /*
2825                  * do_quotactl
2826                  *  f2fs_quota_sync
2827                  *  f2fs_down_read(quota_sem)
2828                  *  dquot_writeback_dquots()
2829                  *  f2fs_dquot_commit
2830                  *                            block_operation
2831                  *                            f2fs_down_read(quota_sem)
2832                  */
2833                 f2fs_lock_op(sbi);
2834                 f2fs_down_read(&sbi->quota_sem);
2835
2836                 ret = f2fs_quota_sync_file(sbi, cnt);
2837
2838                 f2fs_up_read(&sbi->quota_sem);
2839                 f2fs_unlock_op(sbi);
2840
2841                 if (!f2fs_sb_has_quota_ino(sbi))
2842                         inode_unlock(dqopt->files[cnt]);
2843
2844                 if (ret)
2845                         break;
2846         }
2847         return ret;
2848 }
2849
2850 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2851                                                         const struct path *path)
2852 {
2853         struct inode *inode;
2854         int err;
2855
2856         /* if quota sysfile exists, deny enabling quota with specific file */
2857         if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2858                 f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2859                 return -EBUSY;
2860         }
2861
2862         err = f2fs_quota_sync(sb, type);
2863         if (err)
2864                 return err;
2865
2866         err = dquot_quota_on(sb, type, format_id, path);
2867         if (err)
2868                 return err;
2869
2870         inode = d_inode(path->dentry);
2871
2872         inode_lock(inode);
2873         F2FS_I(inode)->i_flags |= F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL;
2874         f2fs_set_inode_flags(inode);
2875         inode_unlock(inode);
2876         f2fs_mark_inode_dirty_sync(inode, false);
2877
2878         return 0;
2879 }
2880
2881 static int __f2fs_quota_off(struct super_block *sb, int type)
2882 {
2883         struct inode *inode = sb_dqopt(sb)->files[type];
2884         int err;
2885
2886         if (!inode || !igrab(inode))
2887                 return dquot_quota_off(sb, type);
2888
2889         err = f2fs_quota_sync(sb, type);
2890         if (err)
2891                 goto out_put;
2892
2893         err = dquot_quota_off(sb, type);
2894         if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
2895                 goto out_put;
2896
2897         inode_lock(inode);
2898         F2FS_I(inode)->i_flags &= ~(F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL);
2899         f2fs_set_inode_flags(inode);
2900         inode_unlock(inode);
2901         f2fs_mark_inode_dirty_sync(inode, false);
2902 out_put:
2903         iput(inode);
2904         return err;
2905 }
2906
2907 static int f2fs_quota_off(struct super_block *sb, int type)
2908 {
2909         struct f2fs_sb_info *sbi = F2FS_SB(sb);
2910         int err;
2911
2912         err = __f2fs_quota_off(sb, type);
2913
2914         /*
2915          * quotactl can shutdown journalled quota, result in inconsistence
2916          * between quota record and fs data by following updates, tag the
2917          * flag to let fsck be aware of it.
2918          */
2919         if (is_journalled_quota(sbi))
2920                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2921         return err;
2922 }
2923
2924 void f2fs_quota_off_umount(struct super_block *sb)
2925 {
2926         int type;
2927         int err;
2928
2929         for (type = 0; type < MAXQUOTAS; type++) {
2930                 err = __f2fs_quota_off(sb, type);
2931                 if (err) {
2932                         int ret = dquot_quota_off(sb, type);
2933
2934                         f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
2935                                  type, err, ret);
2936                         set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2937                 }
2938         }
2939         /*
2940          * In case of checkpoint=disable, we must flush quota blocks.
2941          * This can cause NULL exception for node_inode in end_io, since
2942          * put_super already dropped it.
2943          */
2944         sync_filesystem(sb);
2945 }
2946
2947 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
2948 {
2949         struct quota_info *dqopt = sb_dqopt(sb);
2950         int type;
2951
2952         for (type = 0; type < MAXQUOTAS; type++) {
2953                 if (!dqopt->files[type])
2954                         continue;
2955                 f2fs_inode_synced(dqopt->files[type]);
2956         }
2957 }
2958
2959 static int f2fs_dquot_commit(struct dquot *dquot)
2960 {
2961         struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2962         int ret;
2963
2964         f2fs_down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
2965         ret = dquot_commit(dquot);
2966         if (ret < 0)
2967                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2968         f2fs_up_read(&sbi->quota_sem);
2969         return ret;
2970 }
2971
2972 static int f2fs_dquot_acquire(struct dquot *dquot)
2973 {
2974         struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2975         int ret;
2976
2977         f2fs_down_read(&sbi->quota_sem);
2978         ret = dquot_acquire(dquot);
2979         if (ret < 0)
2980                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2981         f2fs_up_read(&sbi->quota_sem);
2982         return ret;
2983 }
2984
2985 static int f2fs_dquot_release(struct dquot *dquot)
2986 {
2987         struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2988         int ret = dquot_release(dquot);
2989
2990         if (ret < 0)
2991                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2992         return ret;
2993 }
2994
2995 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
2996 {
2997         struct super_block *sb = dquot->dq_sb;
2998         struct f2fs_sb_info *sbi = F2FS_SB(sb);
2999         int ret = dquot_mark_dquot_dirty(dquot);
3000
3001         /* if we are using journalled quota */
3002         if (is_journalled_quota(sbi))
3003                 set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
3004
3005         return ret;
3006 }
3007
3008 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
3009 {
3010         struct f2fs_sb_info *sbi = F2FS_SB(sb);
3011         int ret = dquot_commit_info(sb, type);
3012
3013         if (ret < 0)
3014                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3015         return ret;
3016 }
3017
3018 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
3019 {
3020         *projid = F2FS_I(inode)->i_projid;
3021         return 0;
3022 }
3023
3024 static const struct dquot_operations f2fs_quota_operations = {
3025         .get_reserved_space = f2fs_get_reserved_space,
3026         .write_dquot    = f2fs_dquot_commit,
3027         .acquire_dquot  = f2fs_dquot_acquire,
3028         .release_dquot  = f2fs_dquot_release,
3029         .mark_dirty     = f2fs_dquot_mark_dquot_dirty,
3030         .write_info     = f2fs_dquot_commit_info,
3031         .alloc_dquot    = dquot_alloc,
3032         .destroy_dquot  = dquot_destroy,
3033         .get_projid     = f2fs_get_projid,
3034         .get_next_id    = dquot_get_next_id,
3035 };
3036
3037 static const struct quotactl_ops f2fs_quotactl_ops = {
3038         .quota_on       = f2fs_quota_on,
3039         .quota_off      = f2fs_quota_off,
3040         .quota_sync     = f2fs_quota_sync,
3041         .get_state      = dquot_get_state,
3042         .set_info       = dquot_set_dqinfo,
3043         .get_dqblk      = dquot_get_dqblk,
3044         .set_dqblk      = dquot_set_dqblk,
3045         .get_nextdqblk  = dquot_get_next_dqblk,
3046 };
3047 #else
3048 int f2fs_dquot_initialize(struct inode *inode)
3049 {
3050         return 0;
3051 }
3052
3053 int f2fs_quota_sync(struct super_block *sb, int type)
3054 {
3055         return 0;
3056 }
3057
3058 void f2fs_quota_off_umount(struct super_block *sb)
3059 {
3060 }
3061 #endif
3062
3063 static const struct super_operations f2fs_sops = {
3064         .alloc_inode    = f2fs_alloc_inode,
3065         .free_inode     = f2fs_free_inode,
3066         .drop_inode     = f2fs_drop_inode,
3067         .write_inode    = f2fs_write_inode,
3068         .dirty_inode    = f2fs_dirty_inode,
3069         .show_options   = f2fs_show_options,
3070 #ifdef CONFIG_QUOTA
3071         .quota_read     = f2fs_quota_read,
3072         .quota_write    = f2fs_quota_write,
3073         .get_dquots     = f2fs_get_dquots,
3074 #endif
3075         .evict_inode    = f2fs_evict_inode,
3076         .put_super      = f2fs_put_super,
3077         .sync_fs        = f2fs_sync_fs,
3078         .freeze_fs      = f2fs_freeze,
3079         .unfreeze_fs    = f2fs_unfreeze,
3080         .statfs         = f2fs_statfs,
3081         .remount_fs     = f2fs_remount,
3082 };
3083
3084 #ifdef CONFIG_FS_ENCRYPTION
3085 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
3086 {
3087         return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3088                                 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3089                                 ctx, len, NULL);
3090 }
3091
3092 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
3093                                                         void *fs_data)
3094 {
3095         struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3096
3097         /*
3098          * Encrypting the root directory is not allowed because fsck
3099          * expects lost+found directory to exist and remain unencrypted
3100          * if LOST_FOUND feature is enabled.
3101          *
3102          */
3103         if (f2fs_sb_has_lost_found(sbi) &&
3104                         inode->i_ino == F2FS_ROOT_INO(sbi))
3105                 return -EPERM;
3106
3107         return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3108                                 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3109                                 ctx, len, fs_data, XATTR_CREATE);
3110 }
3111
3112 static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
3113 {
3114         return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
3115 }
3116
3117 static bool f2fs_has_stable_inodes(struct super_block *sb)
3118 {
3119         return true;
3120 }
3121
3122 static void f2fs_get_ino_and_lblk_bits(struct super_block *sb,
3123                                        int *ino_bits_ret, int *lblk_bits_ret)
3124 {
3125         *ino_bits_ret = 8 * sizeof(nid_t);
3126         *lblk_bits_ret = 8 * sizeof(block_t);
3127 }
3128
3129 static struct block_device **f2fs_get_devices(struct super_block *sb,
3130                                               unsigned int *num_devs)
3131 {
3132         struct f2fs_sb_info *sbi = F2FS_SB(sb);
3133         struct block_device **devs;
3134         int i;
3135
3136         if (!f2fs_is_multi_device(sbi))
3137                 return NULL;
3138
3139         devs = kmalloc_array(sbi->s_ndevs, sizeof(*devs), GFP_KERNEL);
3140         if (!devs)
3141                 return ERR_PTR(-ENOMEM);
3142
3143         for (i = 0; i < sbi->s_ndevs; i++)
3144                 devs[i] = FDEV(i).bdev;
3145         *num_devs = sbi->s_ndevs;
3146         return devs;
3147 }
3148
3149 static const struct fscrypt_operations f2fs_cryptops = {
3150         .key_prefix             = "f2fs:",
3151         .get_context            = f2fs_get_context,
3152         .set_context            = f2fs_set_context,
3153         .get_dummy_policy       = f2fs_get_dummy_policy,
3154         .empty_dir              = f2fs_empty_dir,
3155         .has_stable_inodes      = f2fs_has_stable_inodes,
3156         .get_ino_and_lblk_bits  = f2fs_get_ino_and_lblk_bits,
3157         .get_devices            = f2fs_get_devices,
3158 };
3159 #endif
3160
3161 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
3162                 u64 ino, u32 generation)
3163 {
3164         struct f2fs_sb_info *sbi = F2FS_SB(sb);
3165         struct inode *inode;
3166
3167         if (f2fs_check_nid_range(sbi, ino))
3168                 return ERR_PTR(-ESTALE);
3169
3170         /*
3171          * f2fs_iget isn't quite right if the inode is currently unallocated!
3172          * However f2fs_iget currently does appropriate checks to handle stale
3173          * inodes so everything is OK.
3174          */
3175         inode = f2fs_iget(sb, ino);
3176         if (IS_ERR(inode))
3177                 return ERR_CAST(inode);
3178         if (unlikely(generation && inode->i_generation != generation)) {
3179                 /* we didn't find the right inode.. */
3180                 iput(inode);
3181                 return ERR_PTR(-ESTALE);
3182         }
3183         return inode;
3184 }
3185
3186 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
3187                 int fh_len, int fh_type)
3188 {
3189         return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
3190                                     f2fs_nfs_get_inode);
3191 }
3192
3193 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
3194                 int fh_len, int fh_type)
3195 {
3196         return generic_fh_to_parent(sb, fid, fh_len, fh_type,
3197                                     f2fs_nfs_get_inode);
3198 }
3199
3200 static const struct export_operations f2fs_export_ops = {
3201         .fh_to_dentry = f2fs_fh_to_dentry,
3202         .fh_to_parent = f2fs_fh_to_parent,
3203         .get_parent = f2fs_get_parent,
3204 };
3205
3206 loff_t max_file_blocks(struct inode *inode)
3207 {
3208         loff_t result = 0;
3209         loff_t leaf_count;
3210
3211         /*
3212          * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
3213          * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
3214          * space in inode.i_addr, it will be more safe to reassign
3215          * result as zero.
3216          */
3217
3218         if (inode && f2fs_compressed_file(inode))
3219                 leaf_count = ADDRS_PER_BLOCK(inode);
3220         else
3221                 leaf_count = DEF_ADDRS_PER_BLOCK;
3222
3223         /* two direct node blocks */
3224         result += (leaf_count * 2);
3225
3226         /* two indirect node blocks */
3227         leaf_count *= NIDS_PER_BLOCK;
3228         result += (leaf_count * 2);
3229
3230         /* one double indirect node block */
3231         leaf_count *= NIDS_PER_BLOCK;
3232         result += leaf_count;
3233
3234         return result;
3235 }
3236
3237 static int __f2fs_commit_super(struct buffer_head *bh,
3238                         struct f2fs_super_block *super)
3239 {
3240         lock_buffer(bh);
3241         if (super)
3242                 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
3243         set_buffer_dirty(bh);
3244         unlock_buffer(bh);
3245
3246         /* it's rare case, we can do fua all the time */
3247         return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
3248 }
3249
3250 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
3251                                         struct buffer_head *bh)
3252 {
3253         struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3254                                         (bh->b_data + F2FS_SUPER_OFFSET);
3255         struct super_block *sb = sbi->sb;
3256         u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
3257         u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
3258         u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
3259         u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
3260         u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
3261         u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
3262         u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
3263         u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
3264         u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
3265         u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
3266         u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3267         u32 segment_count = le32_to_cpu(raw_super->segment_count);
3268         u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3269         u64 main_end_blkaddr = main_blkaddr +
3270                                 (segment_count_main << log_blocks_per_seg);
3271         u64 seg_end_blkaddr = segment0_blkaddr +
3272                                 (segment_count << log_blocks_per_seg);
3273
3274         if (segment0_blkaddr != cp_blkaddr) {
3275                 f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
3276                           segment0_blkaddr, cp_blkaddr);
3277                 return true;
3278         }
3279
3280         if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
3281                                                         sit_blkaddr) {
3282                 f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
3283                           cp_blkaddr, sit_blkaddr,
3284                           segment_count_ckpt << log_blocks_per_seg);
3285                 return true;
3286         }
3287
3288         if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
3289                                                         nat_blkaddr) {
3290                 f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
3291                           sit_blkaddr, nat_blkaddr,
3292                           segment_count_sit << log_blocks_per_seg);
3293                 return true;
3294         }
3295
3296         if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
3297                                                         ssa_blkaddr) {
3298                 f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
3299                           nat_blkaddr, ssa_blkaddr,
3300                           segment_count_nat << log_blocks_per_seg);
3301                 return true;
3302         }
3303
3304         if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
3305                                                         main_blkaddr) {
3306                 f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
3307                           ssa_blkaddr, main_blkaddr,
3308                           segment_count_ssa << log_blocks_per_seg);
3309                 return true;
3310         }
3311
3312         if (main_end_blkaddr > seg_end_blkaddr) {
3313                 f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
3314                           main_blkaddr, seg_end_blkaddr,
3315                           segment_count_main << log_blocks_per_seg);
3316                 return true;
3317         } else if (main_end_blkaddr < seg_end_blkaddr) {
3318                 int err = 0;
3319                 char *res;
3320
3321                 /* fix in-memory information all the time */
3322                 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
3323                                 segment0_blkaddr) >> log_blocks_per_seg);
3324
3325                 if (f2fs_readonly(sb) || f2fs_hw_is_readonly(sbi)) {
3326                         set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3327                         res = "internally";
3328                 } else {
3329                         err = __f2fs_commit_super(bh, NULL);
3330                         res = err ? "failed" : "done";
3331                 }
3332                 f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
3333                           res, main_blkaddr, seg_end_blkaddr,
3334                           segment_count_main << log_blocks_per_seg);
3335                 if (err)
3336                         return true;
3337         }
3338         return false;
3339 }
3340
3341 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
3342                                 struct buffer_head *bh)
3343 {
3344         block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
3345         block_t total_sections, blocks_per_seg;
3346         struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3347                                         (bh->b_data + F2FS_SUPER_OFFSET);
3348         size_t crc_offset = 0;
3349         __u32 crc = 0;
3350
3351         if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
3352                 f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
3353                           F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
3354                 return -EINVAL;
3355         }
3356
3357         /* Check checksum_offset and crc in superblock */
3358         if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
3359                 crc_offset = le32_to_cpu(raw_super->checksum_offset);
3360                 if (crc_offset !=
3361                         offsetof(struct f2fs_super_block, crc)) {
3362                         f2fs_info(sbi, "Invalid SB checksum offset: %zu",
3363                                   crc_offset);
3364                         return -EFSCORRUPTED;
3365                 }
3366                 crc = le32_to_cpu(raw_super->crc);
3367                 if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
3368                         f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
3369                         return -EFSCORRUPTED;
3370                 }
3371         }
3372
3373         /* Currently, support only 4KB block size */
3374         if (le32_to_cpu(raw_super->log_blocksize) != F2FS_BLKSIZE_BITS) {
3375                 f2fs_info(sbi, "Invalid log_blocksize (%u), supports only %u",
3376                           le32_to_cpu(raw_super->log_blocksize),
3377                           F2FS_BLKSIZE_BITS);
3378                 return -EFSCORRUPTED;
3379         }
3380
3381         /* check log blocks per segment */
3382         if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
3383                 f2fs_info(sbi, "Invalid log blocks per segment (%u)",
3384                           le32_to_cpu(raw_super->log_blocks_per_seg));
3385                 return -EFSCORRUPTED;
3386         }
3387
3388         /* Currently, support 512/1024/2048/4096 bytes sector size */
3389         if (le32_to_cpu(raw_super->log_sectorsize) >
3390                                 F2FS_MAX_LOG_SECTOR_SIZE ||
3391                 le32_to_cpu(raw_super->log_sectorsize) <
3392                                 F2FS_MIN_LOG_SECTOR_SIZE) {
3393                 f2fs_info(sbi, "Invalid log sectorsize (%u)",
3394                           le32_to_cpu(raw_super->log_sectorsize));
3395                 return -EFSCORRUPTED;
3396         }
3397         if (le32_to_cpu(raw_super->log_sectors_per_block) +
3398                 le32_to_cpu(raw_super->log_sectorsize) !=
3399                         F2FS_MAX_LOG_SECTOR_SIZE) {
3400                 f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
3401                           le32_to_cpu(raw_super->log_sectors_per_block),
3402                           le32_to_cpu(raw_super->log_sectorsize));
3403                 return -EFSCORRUPTED;
3404         }
3405
3406         segment_count = le32_to_cpu(raw_super->segment_count);
3407         segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3408         segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3409         secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3410         total_sections = le32_to_cpu(raw_super->section_count);
3411
3412         /* blocks_per_seg should be 512, given the above check */
3413         blocks_per_seg = BIT(le32_to_cpu(raw_super->log_blocks_per_seg));
3414
3415         if (segment_count > F2FS_MAX_SEGMENT ||
3416                                 segment_count < F2FS_MIN_SEGMENTS) {
3417                 f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
3418                 return -EFSCORRUPTED;
3419         }
3420
3421         if (total_sections > segment_count_main || total_sections < 1 ||
3422                         segs_per_sec > segment_count || !segs_per_sec) {
3423                 f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
3424                           segment_count, total_sections, segs_per_sec);
3425                 return -EFSCORRUPTED;
3426         }
3427
3428         if (segment_count_main != total_sections * segs_per_sec) {
3429                 f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
3430                           segment_count_main, total_sections, segs_per_sec);
3431                 return -EFSCORRUPTED;
3432         }
3433
3434         if ((segment_count / segs_per_sec) < total_sections) {
3435                 f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
3436                           segment_count, segs_per_sec, total_sections);
3437                 return -EFSCORRUPTED;
3438         }
3439
3440         if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
3441                 f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
3442                           segment_count, le64_to_cpu(raw_super->block_count));
3443                 return -EFSCORRUPTED;
3444         }
3445
3446         if (RDEV(0).path[0]) {
3447                 block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
3448                 int i = 1;
3449
3450                 while (i < MAX_DEVICES && RDEV(i).path[0]) {
3451                         dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
3452                         i++;
3453                 }
3454                 if (segment_count != dev_seg_count) {
3455                         f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
3456                                         segment_count, dev_seg_count);
3457                         return -EFSCORRUPTED;
3458                 }
3459         } else {
3460                 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
3461                                         !bdev_is_zoned(sbi->sb->s_bdev)) {
3462                         f2fs_info(sbi, "Zoned block device path is missing");
3463                         return -EFSCORRUPTED;
3464                 }
3465         }
3466
3467         if (secs_per_zone > total_sections || !secs_per_zone) {
3468                 f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
3469                           secs_per_zone, total_sections);
3470                 return -EFSCORRUPTED;
3471         }
3472         if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
3473                         raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
3474                         (le32_to_cpu(raw_super->extension_count) +
3475                         raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
3476                 f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
3477                           le32_to_cpu(raw_super->extension_count),
3478                           raw_super->hot_ext_count,
3479                           F2FS_MAX_EXTENSION);
3480                 return -EFSCORRUPTED;
3481         }
3482
3483         if (le32_to_cpu(raw_super->cp_payload) >=
3484                                 (blocks_per_seg - F2FS_CP_PACKS -
3485                                 NR_CURSEG_PERSIST_TYPE)) {
3486                 f2fs_info(sbi, "Insane cp_payload (%u >= %u)",
3487                           le32_to_cpu(raw_super->cp_payload),
3488                           blocks_per_seg - F2FS_CP_PACKS -
3489                           NR_CURSEG_PERSIST_TYPE);
3490                 return -EFSCORRUPTED;
3491         }
3492
3493         /* check reserved ino info */
3494         if (le32_to_cpu(raw_super->node_ino) != 1 ||
3495                 le32_to_cpu(raw_super->meta_ino) != 2 ||
3496                 le32_to_cpu(raw_super->root_ino) != 3) {
3497                 f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
3498                           le32_to_cpu(raw_super->node_ino),
3499                           le32_to_cpu(raw_super->meta_ino),
3500                           le32_to_cpu(raw_super->root_ino));
3501                 return -EFSCORRUPTED;
3502         }
3503
3504         /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
3505         if (sanity_check_area_boundary(sbi, bh))
3506                 return -EFSCORRUPTED;
3507
3508         return 0;
3509 }
3510
3511 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
3512 {
3513         unsigned int total, fsmeta;
3514         struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3515         struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
3516         unsigned int ovp_segments, reserved_segments;
3517         unsigned int main_segs, blocks_per_seg;
3518         unsigned int sit_segs, nat_segs;
3519         unsigned int sit_bitmap_size, nat_bitmap_size;
3520         unsigned int log_blocks_per_seg;
3521         unsigned int segment_count_main;
3522         unsigned int cp_pack_start_sum, cp_payload;
3523         block_t user_block_count, valid_user_blocks;
3524         block_t avail_node_count, valid_node_count;
3525         unsigned int nat_blocks, nat_bits_bytes, nat_bits_blocks;
3526         int i, j;
3527
3528         total = le32_to_cpu(raw_super->segment_count);
3529         fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
3530         sit_segs = le32_to_cpu(raw_super->segment_count_sit);
3531         fsmeta += sit_segs;
3532         nat_segs = le32_to_cpu(raw_super->segment_count_nat);
3533         fsmeta += nat_segs;
3534         fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
3535         fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
3536
3537         if (unlikely(fsmeta >= total))
3538                 return 1;
3539
3540         ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
3541         reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
3542
3543         if (!f2fs_sb_has_readonly(sbi) &&
3544                         unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
3545                         ovp_segments == 0 || reserved_segments == 0)) {
3546                 f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
3547                 return 1;
3548         }
3549         user_block_count = le64_to_cpu(ckpt->user_block_count);
3550         segment_count_main = le32_to_cpu(raw_super->segment_count_main) +
3551                         (f2fs_sb_has_readonly(sbi) ? 1 : 0);
3552         log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3553         if (!user_block_count || user_block_count >=
3554                         segment_count_main << log_blocks_per_seg) {
3555                 f2fs_err(sbi, "Wrong user_block_count: %u",
3556                          user_block_count);
3557                 return 1;
3558         }
3559
3560         valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
3561         if (valid_user_blocks > user_block_count) {
3562                 f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
3563                          valid_user_blocks, user_block_count);
3564                 return 1;
3565         }
3566
3567         valid_node_count = le32_to_cpu(ckpt->valid_node_count);
3568         avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
3569         if (valid_node_count > avail_node_count) {
3570                 f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
3571                          valid_node_count, avail_node_count);
3572                 return 1;
3573         }
3574
3575         main_segs = le32_to_cpu(raw_super->segment_count_main);
3576         blocks_per_seg = sbi->blocks_per_seg;
3577
3578         for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3579                 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
3580                         le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
3581                         return 1;
3582
3583                 if (f2fs_sb_has_readonly(sbi))
3584                         goto check_data;
3585
3586                 for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
3587                         if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3588                                 le32_to_cpu(ckpt->cur_node_segno[j])) {
3589                                 f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
3590                                          i, j,
3591                                          le32_to_cpu(ckpt->cur_node_segno[i]));
3592                                 return 1;
3593                         }
3594                 }
3595         }
3596 check_data:
3597         for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
3598                 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
3599                         le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
3600                         return 1;
3601
3602                 if (f2fs_sb_has_readonly(sbi))
3603                         goto skip_cross;
3604
3605                 for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3606                         if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3607                                 le32_to_cpu(ckpt->cur_data_segno[j])) {
3608                                 f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3609                                          i, j,
3610                                          le32_to_cpu(ckpt->cur_data_segno[i]));
3611                                 return 1;
3612                         }
3613                 }
3614         }
3615         for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3616                 for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3617                         if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3618                                 le32_to_cpu(ckpt->cur_data_segno[j])) {
3619                                 f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3620                                          i, j,
3621                                          le32_to_cpu(ckpt->cur_node_segno[i]));
3622                                 return 1;
3623                         }
3624                 }
3625         }
3626 skip_cross:
3627         sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3628         nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3629
3630         if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3631                 nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3632                 f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3633                          sit_bitmap_size, nat_bitmap_size);
3634                 return 1;
3635         }
3636
3637         cp_pack_start_sum = __start_sum_addr(sbi);
3638         cp_payload = __cp_payload(sbi);
3639         if (cp_pack_start_sum < cp_payload + 1 ||
3640                 cp_pack_start_sum > blocks_per_seg - 1 -
3641                         NR_CURSEG_PERSIST_TYPE) {
3642                 f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3643                          cp_pack_start_sum);
3644                 return 1;
3645         }
3646
3647         if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3648                 le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3649                 f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3650                           "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3651                           "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3652                           le32_to_cpu(ckpt->checksum_offset));
3653                 return 1;
3654         }
3655
3656         nat_blocks = nat_segs << log_blocks_per_seg;
3657         nat_bits_bytes = nat_blocks / BITS_PER_BYTE;
3658         nat_bits_blocks = F2FS_BLK_ALIGN((nat_bits_bytes << 1) + 8);
3659         if (__is_set_ckpt_flags(ckpt, CP_NAT_BITS_FLAG) &&
3660                 (cp_payload + F2FS_CP_PACKS +
3661                 NR_CURSEG_PERSIST_TYPE + nat_bits_blocks >= blocks_per_seg)) {
3662                 f2fs_warn(sbi, "Insane cp_payload: %u, nat_bits_blocks: %u)",
3663                           cp_payload, nat_bits_blocks);
3664                 return 1;
3665         }
3666
3667         if (unlikely(f2fs_cp_error(sbi))) {
3668                 f2fs_err(sbi, "A bug case: need to run fsck");
3669                 return 1;
3670         }
3671         return 0;
3672 }
3673
3674 static void init_sb_info(struct f2fs_sb_info *sbi)
3675 {
3676         struct f2fs_super_block *raw_super = sbi->raw_super;
3677         int i;
3678
3679         sbi->log_sectors_per_block =
3680                 le32_to_cpu(raw_super->log_sectors_per_block);
3681         sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3682         sbi->blocksize = BIT(sbi->log_blocksize);
3683         sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3684         sbi->blocks_per_seg = BIT(sbi->log_blocks_per_seg);
3685         sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3686         sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3687         sbi->total_sections = le32_to_cpu(raw_super->section_count);
3688         sbi->total_node_count =
3689                 (le32_to_cpu(raw_super->segment_count_nat) / 2)
3690                         * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
3691         F2FS_ROOT_INO(sbi) = le32_to_cpu(raw_super->root_ino);
3692         F2FS_NODE_INO(sbi) = le32_to_cpu(raw_super->node_ino);
3693         F2FS_META_INO(sbi) = le32_to_cpu(raw_super->meta_ino);
3694         sbi->cur_victim_sec = NULL_SECNO;
3695         sbi->gc_mode = GC_NORMAL;
3696         sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3697         sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3698         sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3699         sbi->migration_granularity = sbi->segs_per_sec;
3700         sbi->seq_file_ra_mul = MIN_RA_MUL;
3701         sbi->max_fragment_chunk = DEF_FRAGMENT_SIZE;
3702         sbi->max_fragment_hole = DEF_FRAGMENT_SIZE;
3703         spin_lock_init(&sbi->gc_remaining_trials_lock);
3704         atomic64_set(&sbi->current_atomic_write, 0);
3705
3706         sbi->dir_level = DEF_DIR_LEVEL;
3707         sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3708         sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3709         sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3710         sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3711         sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3712         sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3713                                 DEF_UMOUNT_DISCARD_TIMEOUT;
3714         clear_sbi_flag(sbi, SBI_NEED_FSCK);
3715
3716         for (i = 0; i < NR_COUNT_TYPE; i++)
3717                 atomic_set(&sbi->nr_pages[i], 0);
3718
3719         for (i = 0; i < META; i++)
3720                 atomic_set(&sbi->wb_sync_req[i], 0);
3721
3722         INIT_LIST_HEAD(&sbi->s_list);
3723         mutex_init(&sbi->umount_mutex);
3724         init_f2fs_rwsem(&sbi->io_order_lock);
3725         spin_lock_init(&sbi->cp_lock);
3726
3727         sbi->dirty_device = 0;
3728         spin_lock_init(&sbi->dev_lock);
3729
3730         init_f2fs_rwsem(&sbi->sb_lock);
3731         init_f2fs_rwsem(&sbi->pin_sem);
3732 }
3733
3734 static int init_percpu_info(struct f2fs_sb_info *sbi)
3735 {
3736         int err;
3737
3738         err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3739         if (err)
3740                 return err;
3741
3742         err = percpu_counter_init(&sbi->rf_node_block_count, 0, GFP_KERNEL);
3743         if (err)
3744                 goto err_valid_block;
3745
3746         err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3747                                                                 GFP_KERNEL);
3748         if (err)
3749                 goto err_node_block;
3750         return 0;
3751
3752 err_node_block:
3753         percpu_counter_destroy(&sbi->rf_node_block_count);
3754 err_valid_block:
3755         percpu_counter_destroy(&sbi->alloc_valid_block_count);
3756         return err;
3757 }
3758
3759 #ifdef CONFIG_BLK_DEV_ZONED
3760
3761 struct f2fs_report_zones_args {
3762         struct f2fs_sb_info *sbi;
3763         struct f2fs_dev_info *dev;
3764 };
3765
3766 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3767                               void *data)
3768 {
3769         struct f2fs_report_zones_args *rz_args = data;
3770         block_t unusable_blocks = (zone->len - zone->capacity) >>
3771                                         F2FS_LOG_SECTORS_PER_BLOCK;
3772
3773         if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3774                 return 0;
3775
3776         set_bit(idx, rz_args->dev->blkz_seq);
3777         if (!rz_args->sbi->unusable_blocks_per_sec) {
3778                 rz_args->sbi->unusable_blocks_per_sec = unusable_blocks;
3779                 return 0;
3780         }
3781         if (rz_args->sbi->unusable_blocks_per_sec != unusable_blocks) {
3782                 f2fs_err(rz_args->sbi, "F2FS supports single zone capacity\n");
3783                 return -EINVAL;
3784         }
3785         return 0;
3786 }
3787
3788 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3789 {
3790         struct block_device *bdev = FDEV(devi).bdev;
3791         sector_t nr_sectors = bdev_nr_sectors(bdev);
3792         struct f2fs_report_zones_args rep_zone_arg;
3793         u64 zone_sectors;
3794         int ret;
3795
3796         if (!f2fs_sb_has_blkzoned(sbi))
3797                 return 0;
3798
3799         zone_sectors = bdev_zone_sectors(bdev);
3800         if (!is_power_of_2(zone_sectors)) {
3801                 f2fs_err(sbi, "F2FS does not support non power of 2 zone sizes\n");
3802                 return -EINVAL;
3803         }
3804
3805         if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3806                                 SECTOR_TO_BLOCK(zone_sectors))
3807                 return -EINVAL;
3808         sbi->blocks_per_blkz = SECTOR_TO_BLOCK(zone_sectors);
3809         FDEV(devi).nr_blkz = div_u64(SECTOR_TO_BLOCK(nr_sectors),
3810                                         sbi->blocks_per_blkz);
3811         if (nr_sectors & (zone_sectors - 1))
3812                 FDEV(devi).nr_blkz++;
3813
3814         FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3815                                         BITS_TO_LONGS(FDEV(devi).nr_blkz)
3816                                         * sizeof(unsigned long),
3817                                         GFP_KERNEL);
3818         if (!FDEV(devi).blkz_seq)
3819                 return -ENOMEM;
3820
3821         rep_zone_arg.sbi = sbi;
3822         rep_zone_arg.dev = &FDEV(devi);
3823
3824         ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3825                                   &rep_zone_arg);
3826         if (ret < 0)
3827                 return ret;
3828         return 0;
3829 }
3830 #endif
3831
3832 /*
3833  * Read f2fs raw super block.
3834  * Because we have two copies of super block, so read both of them
3835  * to get the first valid one. If any one of them is broken, we pass
3836  * them recovery flag back to the caller.
3837  */
3838 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3839                         struct f2fs_super_block **raw_super,
3840                         int *valid_super_block, int *recovery)
3841 {
3842         struct super_block *sb = sbi->sb;
3843         int block;
3844         struct buffer_head *bh;
3845         struct f2fs_super_block *super;
3846         int err = 0;
3847
3848         super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3849         if (!super)
3850                 return -ENOMEM;
3851
3852         for (block = 0; block < 2; block++) {
3853                 bh = sb_bread(sb, block);
3854                 if (!bh) {
3855                         f2fs_err(sbi, "Unable to read %dth superblock",
3856                                  block + 1);
3857                         err = -EIO;
3858                         *recovery = 1;
3859                         continue;
3860                 }
3861
3862                 /* sanity checking of raw super */
3863                 err = sanity_check_raw_super(sbi, bh);
3864                 if (err) {
3865                         f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3866                                  block + 1);
3867                         brelse(bh);
3868                         *recovery = 1;
3869                         continue;
3870                 }
3871
3872                 if (!*raw_super) {
3873                         memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3874                                                         sizeof(*super));
3875                         *valid_super_block = block;
3876                         *raw_super = super;
3877                 }
3878                 brelse(bh);
3879         }
3880
3881         /* No valid superblock */
3882         if (!*raw_super)
3883                 kfree(super);
3884         else
3885                 err = 0;
3886
3887         return err;
3888 }
3889
3890 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
3891 {
3892         struct buffer_head *bh;
3893         __u32 crc = 0;
3894         int err;
3895
3896         if ((recover && f2fs_readonly(sbi->sb)) ||
3897                                 f2fs_hw_is_readonly(sbi)) {
3898                 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3899                 return -EROFS;
3900         }
3901
3902         /* we should update superblock crc here */
3903         if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
3904                 crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
3905                                 offsetof(struct f2fs_super_block, crc));
3906                 F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
3907         }
3908
3909         /* write back-up superblock first */
3910         bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
3911         if (!bh)
3912                 return -EIO;
3913         err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3914         brelse(bh);
3915
3916         /* if we are in recovery path, skip writing valid superblock */
3917         if (recover || err)
3918                 return err;
3919
3920         /* write current valid superblock */
3921         bh = sb_bread(sbi->sb, sbi->valid_super_block);
3922         if (!bh)
3923                 return -EIO;
3924         err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3925         brelse(bh);
3926         return err;
3927 }
3928
3929 void f2fs_handle_stop(struct f2fs_sb_info *sbi, unsigned char reason)
3930 {
3931         struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3932         int err;
3933
3934         f2fs_down_write(&sbi->sb_lock);
3935
3936         if (raw_super->s_stop_reason[reason] < GENMASK(BITS_PER_BYTE - 1, 0))
3937                 raw_super->s_stop_reason[reason]++;
3938
3939         err = f2fs_commit_super(sbi, false);
3940         if (err)
3941                 f2fs_err(sbi, "f2fs_commit_super fails to record reason:%u err:%d",
3942                                                                 reason, err);
3943         f2fs_up_write(&sbi->sb_lock);
3944 }
3945
3946 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag)
3947 {
3948         spin_lock(&sbi->error_lock);
3949         if (!test_bit(flag, (unsigned long *)sbi->errors)) {
3950                 set_bit(flag, (unsigned long *)sbi->errors);
3951                 sbi->error_dirty = true;
3952         }
3953         spin_unlock(&sbi->error_lock);
3954 }
3955
3956 static bool f2fs_update_errors(struct f2fs_sb_info *sbi)
3957 {
3958         bool need_update = false;
3959
3960         spin_lock(&sbi->error_lock);
3961         if (sbi->error_dirty) {
3962                 memcpy(F2FS_RAW_SUPER(sbi)->s_errors, sbi->errors,
3963                                                         MAX_F2FS_ERRORS);
3964                 sbi->error_dirty = false;
3965                 need_update = true;
3966         }
3967         spin_unlock(&sbi->error_lock);
3968
3969         return need_update;
3970 }
3971
3972 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error)
3973 {
3974         int err;
3975
3976         f2fs_save_errors(sbi, error);
3977
3978         f2fs_down_write(&sbi->sb_lock);
3979
3980         if (!f2fs_update_errors(sbi))
3981                 goto out_unlock;
3982
3983         err = f2fs_commit_super(sbi, false);
3984         if (err)
3985                 f2fs_err(sbi, "f2fs_commit_super fails to record errors:%u, err:%d",
3986                                                                 error, err);
3987 out_unlock:
3988         f2fs_up_write(&sbi->sb_lock);
3989 }
3990
3991 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
3992 {
3993         struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3994         unsigned int max_devices = MAX_DEVICES;
3995         unsigned int logical_blksize;
3996         int i;
3997
3998         /* Initialize single device information */
3999         if (!RDEV(0).path[0]) {
4000                 if (!bdev_is_zoned(sbi->sb->s_bdev))
4001                         return 0;
4002                 max_devices = 1;
4003         }
4004
4005         /*
4006          * Initialize multiple devices information, or single
4007          * zoned block device information.
4008          */
4009         sbi->devs = f2fs_kzalloc(sbi,
4010                                  array_size(max_devices,
4011                                             sizeof(struct f2fs_dev_info)),
4012                                  GFP_KERNEL);
4013         if (!sbi->devs)
4014                 return -ENOMEM;
4015
4016         logical_blksize = bdev_logical_block_size(sbi->sb->s_bdev);
4017         sbi->aligned_blksize = true;
4018
4019         for (i = 0; i < max_devices; i++) {
4020
4021                 if (i > 0 && !RDEV(i).path[0])
4022                         break;
4023
4024                 if (max_devices == 1) {
4025                         /* Single zoned block device mount */
4026                         FDEV(0).bdev =
4027                                 blkdev_get_by_dev(sbi->sb->s_bdev->bd_dev,
4028                                         sbi->sb->s_mode, sbi->sb->s_type);
4029                 } else {
4030                         /* Multi-device mount */
4031                         memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
4032                         FDEV(i).total_segments =
4033                                 le32_to_cpu(RDEV(i).total_segments);
4034                         if (i == 0) {
4035                                 FDEV(i).start_blk = 0;
4036                                 FDEV(i).end_blk = FDEV(i).start_blk +
4037                                     (FDEV(i).total_segments <<
4038                                     sbi->log_blocks_per_seg) - 1 +
4039                                     le32_to_cpu(raw_super->segment0_blkaddr);
4040                         } else {
4041                                 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
4042                                 FDEV(i).end_blk = FDEV(i).start_blk +
4043                                         (FDEV(i).total_segments <<
4044                                         sbi->log_blocks_per_seg) - 1;
4045                         }
4046                         FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
4047                                         sbi->sb->s_mode, sbi->sb->s_type);
4048                 }
4049                 if (IS_ERR(FDEV(i).bdev))
4050                         return PTR_ERR(FDEV(i).bdev);
4051
4052                 /* to release errored devices */
4053                 sbi->s_ndevs = i + 1;
4054
4055                 if (logical_blksize != bdev_logical_block_size(FDEV(i).bdev))
4056                         sbi->aligned_blksize = false;
4057
4058 #ifdef CONFIG_BLK_DEV_ZONED
4059                 if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
4060                                 !f2fs_sb_has_blkzoned(sbi)) {
4061                         f2fs_err(sbi, "Zoned block device feature not enabled");
4062                         return -EINVAL;
4063                 }
4064                 if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
4065                         if (init_blkz_info(sbi, i)) {
4066                                 f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
4067                                 return -EINVAL;
4068                         }
4069                         if (max_devices == 1)
4070                                 break;
4071                         f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
4072                                   i, FDEV(i).path,
4073                                   FDEV(i).total_segments,
4074                                   FDEV(i).start_blk, FDEV(i).end_blk,
4075                                   bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
4076                                   "Host-aware" : "Host-managed");
4077                         continue;
4078                 }
4079 #endif
4080                 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
4081                           i, FDEV(i).path,
4082                           FDEV(i).total_segments,
4083                           FDEV(i).start_blk, FDEV(i).end_blk);
4084         }
4085         f2fs_info(sbi,
4086                   "IO Block Size: %8ld KB", F2FS_IO_SIZE_KB(sbi));
4087         return 0;
4088 }
4089
4090 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
4091 {
4092 #if IS_ENABLED(CONFIG_UNICODE)
4093         if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
4094                 const struct f2fs_sb_encodings *encoding_info;
4095                 struct unicode_map *encoding;
4096                 __u16 encoding_flags;
4097
4098                 encoding_info = f2fs_sb_read_encoding(sbi->raw_super);
4099                 if (!encoding_info) {
4100                         f2fs_err(sbi,
4101                                  "Encoding requested by superblock is unknown");
4102                         return -EINVAL;
4103                 }
4104
4105                 encoding_flags = le16_to_cpu(sbi->raw_super->s_encoding_flags);
4106                 encoding = utf8_load(encoding_info->version);
4107                 if (IS_ERR(encoding)) {
4108                         f2fs_err(sbi,
4109                                  "can't mount with superblock charset: %s-%u.%u.%u "
4110                                  "not supported by the kernel. flags: 0x%x.",
4111                                  encoding_info->name,
4112                                  unicode_major(encoding_info->version),
4113                                  unicode_minor(encoding_info->version),
4114                                  unicode_rev(encoding_info->version),
4115                                  encoding_flags);
4116                         return PTR_ERR(encoding);
4117                 }
4118                 f2fs_info(sbi, "Using encoding defined by superblock: "
4119                          "%s-%u.%u.%u with flags 0x%hx", encoding_info->name,
4120                          unicode_major(encoding_info->version),
4121                          unicode_minor(encoding_info->version),
4122                          unicode_rev(encoding_info->version),
4123                          encoding_flags);
4124
4125                 sbi->sb->s_encoding = encoding;
4126                 sbi->sb->s_encoding_flags = encoding_flags;
4127         }
4128 #else
4129         if (f2fs_sb_has_casefold(sbi)) {
4130                 f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
4131                 return -EINVAL;
4132         }
4133 #endif
4134         return 0;
4135 }
4136
4137 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
4138 {
4139         /* adjust parameters according to the volume size */
4140         if (MAIN_SEGS(sbi) <= SMALL_VOLUME_SEGMENTS) {
4141                 if (f2fs_block_unit_discard(sbi))
4142                         SM_I(sbi)->dcc_info->discard_granularity =
4143                                                 MIN_DISCARD_GRANULARITY;
4144                 if (!f2fs_lfs_mode(sbi))
4145                         SM_I(sbi)->ipu_policy = BIT(F2FS_IPU_FORCE) |
4146                                                 BIT(F2FS_IPU_HONOR_OPU_WRITE);
4147         }
4148
4149         sbi->readdir_ra = true;
4150 }
4151
4152 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
4153 {
4154         struct f2fs_sb_info *sbi;
4155         struct f2fs_super_block *raw_super;
4156         struct inode *root;
4157         int err;
4158         bool skip_recovery = false, need_fsck = false;
4159         char *options = NULL;
4160         int recovery, i, valid_super_block;
4161         struct curseg_info *seg_i;
4162         int retry_cnt = 1;
4163 #ifdef CONFIG_QUOTA
4164         bool quota_enabled = false;
4165 #endif
4166
4167 try_onemore:
4168         err = -EINVAL;
4169         raw_super = NULL;
4170         valid_super_block = -1;
4171         recovery = 0;
4172
4173         /* allocate memory for f2fs-specific super block info */
4174         sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
4175         if (!sbi)
4176                 return -ENOMEM;
4177
4178         sbi->sb = sb;
4179
4180         /* initialize locks within allocated memory */
4181         init_f2fs_rwsem(&sbi->gc_lock);
4182         mutex_init(&sbi->writepages);
4183         init_f2fs_rwsem(&sbi->cp_global_sem);
4184         init_f2fs_rwsem(&sbi->node_write);
4185         init_f2fs_rwsem(&sbi->node_change);
4186         spin_lock_init(&sbi->stat_lock);
4187         init_f2fs_rwsem(&sbi->cp_rwsem);
4188         init_f2fs_rwsem(&sbi->quota_sem);
4189         init_waitqueue_head(&sbi->cp_wait);
4190         spin_lock_init(&sbi->error_lock);
4191
4192         for (i = 0; i < NR_INODE_TYPE; i++) {
4193                 INIT_LIST_HEAD(&sbi->inode_list[i]);
4194                 spin_lock_init(&sbi->inode_lock[i]);
4195         }
4196         mutex_init(&sbi->flush_lock);
4197
4198         /* Load the checksum driver */
4199         sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
4200         if (IS_ERR(sbi->s_chksum_driver)) {
4201                 f2fs_err(sbi, "Cannot load crc32 driver.");
4202                 err = PTR_ERR(sbi->s_chksum_driver);
4203                 sbi->s_chksum_driver = NULL;
4204                 goto free_sbi;
4205         }
4206
4207         /* set a block size */
4208         if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
4209                 f2fs_err(sbi, "unable to set blocksize");
4210                 goto free_sbi;
4211         }
4212
4213         err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
4214                                                                 &recovery);
4215         if (err)
4216                 goto free_sbi;
4217
4218         sb->s_fs_info = sbi;
4219         sbi->raw_super = raw_super;
4220
4221         memcpy(sbi->errors, raw_super->s_errors, MAX_F2FS_ERRORS);
4222
4223         /* precompute checksum seed for metadata */
4224         if (f2fs_sb_has_inode_chksum(sbi))
4225                 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
4226                                                 sizeof(raw_super->uuid));
4227
4228         default_options(sbi);
4229         /* parse mount options */
4230         options = kstrdup((const char *)data, GFP_KERNEL);
4231         if (data && !options) {
4232                 err = -ENOMEM;
4233                 goto free_sb_buf;
4234         }
4235
4236         err = parse_options(sb, options, false);
4237         if (err)
4238                 goto free_options;
4239
4240         sb->s_maxbytes = max_file_blocks(NULL) <<
4241                                 le32_to_cpu(raw_super->log_blocksize);
4242         sb->s_max_links = F2FS_LINK_MAX;
4243
4244         err = f2fs_setup_casefold(sbi);
4245         if (err)
4246                 goto free_options;
4247
4248 #ifdef CONFIG_QUOTA
4249         sb->dq_op = &f2fs_quota_operations;
4250         sb->s_qcop = &f2fs_quotactl_ops;
4251         sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
4252
4253         if (f2fs_sb_has_quota_ino(sbi)) {
4254                 for (i = 0; i < MAXQUOTAS; i++) {
4255                         if (f2fs_qf_ino(sbi->sb, i))
4256                                 sbi->nquota_files++;
4257                 }
4258         }
4259 #endif
4260
4261         sb->s_op = &f2fs_sops;
4262 #ifdef CONFIG_FS_ENCRYPTION
4263         sb->s_cop = &f2fs_cryptops;
4264 #endif
4265 #ifdef CONFIG_FS_VERITY
4266         sb->s_vop = &f2fs_verityops;
4267 #endif
4268         sb->s_xattr = f2fs_xattr_handlers;
4269         sb->s_export_op = &f2fs_export_ops;
4270         sb->s_magic = F2FS_SUPER_MAGIC;
4271         sb->s_time_gran = 1;
4272         sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
4273                 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
4274         memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
4275         sb->s_iflags |= SB_I_CGROUPWB;
4276
4277         /* init f2fs-specific super block info */
4278         sbi->valid_super_block = valid_super_block;
4279
4280         /* disallow all the data/node/meta page writes */
4281         set_sbi_flag(sbi, SBI_POR_DOING);
4282
4283         err = f2fs_init_write_merge_io(sbi);
4284         if (err)
4285                 goto free_bio_info;
4286
4287         init_sb_info(sbi);
4288
4289         err = f2fs_init_iostat(sbi);
4290         if (err)
4291                 goto free_bio_info;
4292
4293         err = init_percpu_info(sbi);
4294         if (err)
4295                 goto free_iostat;
4296
4297         if (F2FS_IO_ALIGNED(sbi)) {
4298                 sbi->write_io_dummy =
4299                         mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
4300                 if (!sbi->write_io_dummy) {
4301                         err = -ENOMEM;
4302                         goto free_percpu;
4303                 }
4304         }
4305
4306         /* init per sbi slab cache */
4307         err = f2fs_init_xattr_caches(sbi);
4308         if (err)
4309                 goto free_io_dummy;
4310         err = f2fs_init_page_array_cache(sbi);
4311         if (err)
4312                 goto free_xattr_cache;
4313
4314         /* get an inode for meta space */
4315         sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
4316         if (IS_ERR(sbi->meta_inode)) {
4317                 f2fs_err(sbi, "Failed to read F2FS meta data inode");
4318                 err = PTR_ERR(sbi->meta_inode);
4319                 goto free_page_array_cache;
4320         }
4321
4322         err = f2fs_get_valid_checkpoint(sbi);
4323         if (err) {
4324                 f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
4325                 goto free_meta_inode;
4326         }
4327
4328         if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
4329                 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
4330         if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
4331                 set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4332                 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
4333         }
4334
4335         if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
4336                 set_sbi_flag(sbi, SBI_NEED_FSCK);
4337
4338         /* Initialize device list */
4339         err = f2fs_scan_devices(sbi);
4340         if (err) {
4341                 f2fs_err(sbi, "Failed to find devices");
4342                 goto free_devices;
4343         }
4344
4345         err = f2fs_init_post_read_wq(sbi);
4346         if (err) {
4347                 f2fs_err(sbi, "Failed to initialize post read workqueue");
4348                 goto free_devices;
4349         }
4350
4351         sbi->total_valid_node_count =
4352                                 le32_to_cpu(sbi->ckpt->valid_node_count);
4353         percpu_counter_set(&sbi->total_valid_inode_count,
4354                                 le32_to_cpu(sbi->ckpt->valid_inode_count));
4355         sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
4356         sbi->total_valid_block_count =
4357                                 le64_to_cpu(sbi->ckpt->valid_block_count);
4358         sbi->last_valid_block_count = sbi->total_valid_block_count;
4359         sbi->reserved_blocks = 0;
4360         sbi->current_reserved_blocks = 0;
4361         limit_reserve_root(sbi);
4362         adjust_unusable_cap_perc(sbi);
4363
4364         f2fs_init_extent_cache_info(sbi);
4365
4366         f2fs_init_ino_entry_info(sbi);
4367
4368         f2fs_init_fsync_node_info(sbi);
4369
4370         /* setup checkpoint request control and start checkpoint issue thread */
4371         f2fs_init_ckpt_req_control(sbi);
4372         if (!f2fs_readonly(sb) && !test_opt(sbi, DISABLE_CHECKPOINT) &&
4373                         test_opt(sbi, MERGE_CHECKPOINT)) {
4374                 err = f2fs_start_ckpt_thread(sbi);
4375                 if (err) {
4376                         f2fs_err(sbi,
4377                             "Failed to start F2FS issue_checkpoint_thread (%d)",
4378                             err);
4379                         goto stop_ckpt_thread;
4380                 }
4381         }
4382
4383         /* setup f2fs internal modules */
4384         err = f2fs_build_segment_manager(sbi);
4385         if (err) {
4386                 f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
4387                          err);
4388                 goto free_sm;
4389         }
4390         err = f2fs_build_node_manager(sbi);
4391         if (err) {
4392                 f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
4393                          err);
4394                 goto free_nm;
4395         }
4396
4397         err = adjust_reserved_segment(sbi);
4398         if (err)
4399                 goto free_nm;
4400
4401         /* For write statistics */
4402         sbi->sectors_written_start = f2fs_get_sectors_written(sbi);
4403
4404         /* Read accumulated write IO statistics if exists */
4405         seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
4406         if (__exist_node_summaries(sbi))
4407                 sbi->kbytes_written =
4408                         le64_to_cpu(seg_i->journal->info.kbytes_written);
4409
4410         f2fs_build_gc_manager(sbi);
4411
4412         err = f2fs_build_stats(sbi);
4413         if (err)
4414                 goto free_nm;
4415
4416         /* get an inode for node space */
4417         sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
4418         if (IS_ERR(sbi->node_inode)) {
4419                 f2fs_err(sbi, "Failed to read node inode");
4420                 err = PTR_ERR(sbi->node_inode);
4421                 goto free_stats;
4422         }
4423
4424         /* read root inode and dentry */
4425         root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
4426         if (IS_ERR(root)) {
4427                 f2fs_err(sbi, "Failed to read root inode");
4428                 err = PTR_ERR(root);
4429                 goto free_node_inode;
4430         }
4431         if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
4432                         !root->i_size || !root->i_nlink) {
4433                 iput(root);
4434                 err = -EINVAL;
4435                 goto free_node_inode;
4436         }
4437
4438         sb->s_root = d_make_root(root); /* allocate root dentry */
4439         if (!sb->s_root) {
4440                 err = -ENOMEM;
4441                 goto free_node_inode;
4442         }
4443
4444         err = f2fs_init_compress_inode(sbi);
4445         if (err)
4446                 goto free_root_inode;
4447
4448         err = f2fs_register_sysfs(sbi);
4449         if (err)
4450                 goto free_compress_inode;
4451
4452 #ifdef CONFIG_QUOTA
4453         /* Enable quota usage during mount */
4454         if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
4455                 err = f2fs_enable_quotas(sb);
4456                 if (err)
4457                         f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
4458         }
4459
4460         quota_enabled = f2fs_recover_quota_begin(sbi);
4461 #endif
4462         /* if there are any orphan inodes, free them */
4463         err = f2fs_recover_orphan_inodes(sbi);
4464         if (err)
4465                 goto free_meta;
4466
4467         if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
4468                 goto reset_checkpoint;
4469
4470         /* recover fsynced data */
4471         if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
4472                         !test_opt(sbi, NORECOVERY)) {
4473                 /*
4474                  * mount should be failed, when device has readonly mode, and
4475                  * previous checkpoint was not done by clean system shutdown.
4476                  */
4477                 if (f2fs_hw_is_readonly(sbi)) {
4478                         if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4479                                 err = f2fs_recover_fsync_data(sbi, true);
4480                                 if (err > 0) {
4481                                         err = -EROFS;
4482                                         f2fs_err(sbi, "Need to recover fsync data, but "
4483                                                 "write access unavailable, please try "
4484                                                 "mount w/ disable_roll_forward or norecovery");
4485                                 }
4486                                 if (err < 0)
4487                                         goto free_meta;
4488                         }
4489                         f2fs_info(sbi, "write access unavailable, skipping recovery");
4490                         goto reset_checkpoint;
4491                 }
4492
4493                 if (need_fsck)
4494                         set_sbi_flag(sbi, SBI_NEED_FSCK);
4495
4496                 if (skip_recovery)
4497                         goto reset_checkpoint;
4498
4499                 err = f2fs_recover_fsync_data(sbi, false);
4500                 if (err < 0) {
4501                         if (err != -ENOMEM)
4502                                 skip_recovery = true;
4503                         need_fsck = true;
4504                         f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
4505                                  err);
4506                         goto free_meta;
4507                 }
4508         } else {
4509                 err = f2fs_recover_fsync_data(sbi, true);
4510
4511                 if (!f2fs_readonly(sb) && err > 0) {
4512                         err = -EINVAL;
4513                         f2fs_err(sbi, "Need to recover fsync data");
4514                         goto free_meta;
4515                 }
4516         }
4517
4518 #ifdef CONFIG_QUOTA
4519         f2fs_recover_quota_end(sbi, quota_enabled);
4520 #endif
4521
4522         /*
4523          * If the f2fs is not readonly and fsync data recovery succeeds,
4524          * check zoned block devices' write pointer consistency.
4525          */
4526         if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
4527                 err = f2fs_check_write_pointer(sbi);
4528                 if (err)
4529                         goto free_meta;
4530         }
4531
4532 reset_checkpoint:
4533         f2fs_init_inmem_curseg(sbi);
4534
4535         /* f2fs_recover_fsync_data() cleared this already */
4536         clear_sbi_flag(sbi, SBI_POR_DOING);
4537
4538         if (test_opt(sbi, DISABLE_CHECKPOINT)) {
4539                 err = f2fs_disable_checkpoint(sbi);
4540                 if (err)
4541                         goto sync_free_meta;
4542         } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
4543                 f2fs_enable_checkpoint(sbi);
4544         }
4545
4546         /*
4547          * If filesystem is not mounted as read-only then
4548          * do start the gc_thread.
4549          */
4550         if ((F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF ||
4551                 test_opt(sbi, GC_MERGE)) && !f2fs_readonly(sb)) {
4552                 /* After POR, we can run background GC thread.*/
4553                 err = f2fs_start_gc_thread(sbi);
4554                 if (err)
4555                         goto sync_free_meta;
4556         }
4557         kvfree(options);
4558
4559         /* recover broken superblock */
4560         if (recovery) {
4561                 err = f2fs_commit_super(sbi, true);
4562                 f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
4563                           sbi->valid_super_block ? 1 : 2, err);
4564         }
4565
4566         f2fs_join_shrinker(sbi);
4567
4568         f2fs_tuning_parameters(sbi);
4569
4570         f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
4571                     cur_cp_version(F2FS_CKPT(sbi)));
4572         f2fs_update_time(sbi, CP_TIME);
4573         f2fs_update_time(sbi, REQ_TIME);
4574         clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4575         return 0;
4576
4577 sync_free_meta:
4578         /* safe to flush all the data */
4579         sync_filesystem(sbi->sb);
4580         retry_cnt = 0;
4581
4582 free_meta:
4583 #ifdef CONFIG_QUOTA
4584         f2fs_truncate_quota_inode_pages(sb);
4585         if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
4586                 f2fs_quota_off_umount(sbi->sb);
4587 #endif
4588         /*
4589          * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
4590          * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
4591          * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
4592          * falls into an infinite loop in f2fs_sync_meta_pages().
4593          */
4594         truncate_inode_pages_final(META_MAPPING(sbi));
4595         /* evict some inodes being cached by GC */
4596         evict_inodes(sb);
4597         f2fs_unregister_sysfs(sbi);
4598 free_compress_inode:
4599         f2fs_destroy_compress_inode(sbi);
4600 free_root_inode:
4601         dput(sb->s_root);
4602         sb->s_root = NULL;
4603 free_node_inode:
4604         f2fs_release_ino_entry(sbi, true);
4605         truncate_inode_pages_final(NODE_MAPPING(sbi));
4606         iput(sbi->node_inode);
4607         sbi->node_inode = NULL;
4608 free_stats:
4609         f2fs_destroy_stats(sbi);
4610 free_nm:
4611         /* stop discard thread before destroying node manager */
4612         f2fs_stop_discard_thread(sbi);
4613         f2fs_destroy_node_manager(sbi);
4614 free_sm:
4615         f2fs_destroy_segment_manager(sbi);
4616 stop_ckpt_thread:
4617         f2fs_stop_ckpt_thread(sbi);
4618         f2fs_destroy_post_read_wq(sbi);
4619 free_devices:
4620         destroy_device_list(sbi);
4621         kvfree(sbi->ckpt);
4622 free_meta_inode:
4623         make_bad_inode(sbi->meta_inode);
4624         iput(sbi->meta_inode);
4625         sbi->meta_inode = NULL;
4626 free_page_array_cache:
4627         f2fs_destroy_page_array_cache(sbi);
4628 free_xattr_cache:
4629         f2fs_destroy_xattr_caches(sbi);
4630 free_io_dummy:
4631         mempool_destroy(sbi->write_io_dummy);
4632 free_percpu:
4633         destroy_percpu_info(sbi);
4634 free_iostat:
4635         f2fs_destroy_iostat(sbi);
4636 free_bio_info:
4637         for (i = 0; i < NR_PAGE_TYPE; i++)
4638                 kvfree(sbi->write_io[i]);
4639
4640 #if IS_ENABLED(CONFIG_UNICODE)
4641         utf8_unload(sb->s_encoding);
4642         sb->s_encoding = NULL;
4643 #endif
4644 free_options:
4645 #ifdef CONFIG_QUOTA
4646         for (i = 0; i < MAXQUOTAS; i++)
4647                 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
4648 #endif
4649         fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
4650         kvfree(options);
4651 free_sb_buf:
4652         kfree(raw_super);
4653 free_sbi:
4654         if (sbi->s_chksum_driver)
4655                 crypto_free_shash(sbi->s_chksum_driver);
4656         kfree(sbi);
4657
4658         /* give only one another chance */
4659         if (retry_cnt > 0 && skip_recovery) {
4660                 retry_cnt--;
4661                 shrink_dcache_sb(sb);
4662                 goto try_onemore;
4663         }
4664         return err;
4665 }
4666
4667 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
4668                         const char *dev_name, void *data)
4669 {
4670         return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
4671 }
4672
4673 static void kill_f2fs_super(struct super_block *sb)
4674 {
4675         if (sb->s_root) {
4676                 struct f2fs_sb_info *sbi = F2FS_SB(sb);
4677
4678                 set_sbi_flag(sbi, SBI_IS_CLOSE);
4679                 f2fs_stop_gc_thread(sbi);
4680                 f2fs_stop_discard_thread(sbi);
4681
4682 #ifdef CONFIG_F2FS_FS_COMPRESSION
4683                 /*
4684                  * latter evict_inode() can bypass checking and invalidating
4685                  * compress inode cache.
4686                  */
4687                 if (test_opt(sbi, COMPRESS_CACHE))
4688                         truncate_inode_pages_final(COMPRESS_MAPPING(sbi));
4689 #endif
4690
4691                 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
4692                                 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4693                         struct cp_control cpc = {
4694                                 .reason = CP_UMOUNT,
4695                         };
4696                         f2fs_write_checkpoint(sbi, &cpc);
4697                 }
4698
4699                 if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
4700                         sb->s_flags &= ~SB_RDONLY;
4701         }
4702         kill_block_super(sb);
4703 }
4704
4705 static struct file_system_type f2fs_fs_type = {
4706         .owner          = THIS_MODULE,
4707         .name           = "f2fs",
4708         .mount          = f2fs_mount,
4709         .kill_sb        = kill_f2fs_super,
4710         .fs_flags       = FS_REQUIRES_DEV | FS_ALLOW_IDMAP,
4711 };
4712 MODULE_ALIAS_FS("f2fs");
4713
4714 static int __init init_inodecache(void)
4715 {
4716         f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
4717                         sizeof(struct f2fs_inode_info), 0,
4718                         SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
4719         return f2fs_inode_cachep ? 0 : -ENOMEM;
4720 }
4721
4722 static void destroy_inodecache(void)
4723 {
4724         /*
4725          * Make sure all delayed rcu free inodes are flushed before we
4726          * destroy cache.
4727          */
4728         rcu_barrier();
4729         kmem_cache_destroy(f2fs_inode_cachep);
4730 }
4731
4732 static int __init init_f2fs_fs(void)
4733 {
4734         int err;
4735
4736         if (PAGE_SIZE != F2FS_BLKSIZE) {
4737                 printk("F2FS not supported on PAGE_SIZE(%lu) != %d\n",
4738                                 PAGE_SIZE, F2FS_BLKSIZE);
4739                 return -EINVAL;
4740         }
4741
4742         err = init_inodecache();
4743         if (err)
4744                 goto fail;
4745         err = f2fs_create_node_manager_caches();
4746         if (err)
4747                 goto free_inodecache;
4748         err = f2fs_create_segment_manager_caches();
4749         if (err)
4750                 goto free_node_manager_caches;
4751         err = f2fs_create_checkpoint_caches();
4752         if (err)
4753                 goto free_segment_manager_caches;
4754         err = f2fs_create_recovery_cache();
4755         if (err)
4756                 goto free_checkpoint_caches;
4757         err = f2fs_create_extent_cache();
4758         if (err)
4759                 goto free_recovery_cache;
4760         err = f2fs_create_garbage_collection_cache();
4761         if (err)
4762                 goto free_extent_cache;
4763         err = f2fs_init_sysfs();
4764         if (err)
4765                 goto free_garbage_collection_cache;
4766         err = register_shrinker(&f2fs_shrinker_info, "f2fs-shrinker");
4767         if (err)
4768                 goto free_sysfs;
4769         err = register_filesystem(&f2fs_fs_type);
4770         if (err)
4771                 goto free_shrinker;
4772         f2fs_create_root_stats();
4773         err = f2fs_init_post_read_processing();
4774         if (err)
4775                 goto free_root_stats;
4776         err = f2fs_init_iostat_processing();
4777         if (err)
4778                 goto free_post_read;
4779         err = f2fs_init_bio_entry_cache();
4780         if (err)
4781                 goto free_iostat;
4782         err = f2fs_init_bioset();
4783         if (err)
4784                 goto free_bio_entry_cache;
4785         err = f2fs_init_compress_mempool();
4786         if (err)
4787                 goto free_bioset;
4788         err = f2fs_init_compress_cache();
4789         if (err)
4790                 goto free_compress_mempool;
4791         err = f2fs_create_casefold_cache();
4792         if (err)
4793                 goto free_compress_cache;
4794         return 0;
4795 free_compress_cache:
4796         f2fs_destroy_compress_cache();
4797 free_compress_mempool:
4798         f2fs_destroy_compress_mempool();
4799 free_bioset:
4800         f2fs_destroy_bioset();
4801 free_bio_entry_cache:
4802         f2fs_destroy_bio_entry_cache();
4803 free_iostat:
4804         f2fs_destroy_iostat_processing();
4805 free_post_read:
4806         f2fs_destroy_post_read_processing();
4807 free_root_stats:
4808         f2fs_destroy_root_stats();
4809         unregister_filesystem(&f2fs_fs_type);
4810 free_shrinker:
4811         unregister_shrinker(&f2fs_shrinker_info);
4812 free_sysfs:
4813         f2fs_exit_sysfs();
4814 free_garbage_collection_cache:
4815         f2fs_destroy_garbage_collection_cache();
4816 free_extent_cache:
4817         f2fs_destroy_extent_cache();
4818 free_recovery_cache:
4819         f2fs_destroy_recovery_cache();
4820 free_checkpoint_caches:
4821         f2fs_destroy_checkpoint_caches();
4822 free_segment_manager_caches:
4823         f2fs_destroy_segment_manager_caches();
4824 free_node_manager_caches:
4825         f2fs_destroy_node_manager_caches();
4826 free_inodecache:
4827         destroy_inodecache();
4828 fail:
4829         return err;
4830 }
4831
4832 static void __exit exit_f2fs_fs(void)
4833 {
4834         f2fs_destroy_casefold_cache();
4835         f2fs_destroy_compress_cache();
4836         f2fs_destroy_compress_mempool();
4837         f2fs_destroy_bioset();
4838         f2fs_destroy_bio_entry_cache();
4839         f2fs_destroy_iostat_processing();
4840         f2fs_destroy_post_read_processing();
4841         f2fs_destroy_root_stats();
4842         unregister_filesystem(&f2fs_fs_type);
4843         unregister_shrinker(&f2fs_shrinker_info);
4844         f2fs_exit_sysfs();
4845         f2fs_destroy_garbage_collection_cache();
4846         f2fs_destroy_extent_cache();
4847         f2fs_destroy_recovery_cache();
4848         f2fs_destroy_checkpoint_caches();
4849         f2fs_destroy_segment_manager_caches();
4850         f2fs_destroy_node_manager_caches();
4851         destroy_inodecache();
4852 }
4853
4854 module_init(init_f2fs_fs)
4855 module_exit(exit_f2fs_fs)
4856
4857 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
4858 MODULE_DESCRIPTION("Flash Friendly File System");
4859 MODULE_LICENSE("GPL");
4860 MODULE_SOFTDEP("pre: crc32");
4861