linux/fs/f2fs/super.c
<<
>>
Prefs
   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/statfs.h>
  12#include <linux/buffer_head.h>
  13#include <linux/backing-dev.h>
  14#include <linux/kthread.h>
  15#include <linux/parser.h>
  16#include <linux/mount.h>
  17#include <linux/seq_file.h>
  18#include <linux/proc_fs.h>
  19#include <linux/random.h>
  20#include <linux/exportfs.h>
  21#include <linux/blkdev.h>
  22#include <linux/quotaops.h>
  23#include <linux/f2fs_fs.h>
  24#include <linux/sysfs.h>
  25#include <linux/quota.h>
  26#include <linux/unicode.h>
  27#include <linux/part_stat.h>
  28#include <linux/zstd.h>
  29#include <linux/lz4.h>
  30
  31#include "f2fs.h"
  32#include "node.h"
  33#include "segment.h"
  34#include "xattr.h"
  35#include "gc.h"
  36#include "iostat.h"
  37
  38#define CREATE_TRACE_POINTS
  39#include <trace/events/f2fs.h>
  40
  41static struct kmem_cache *f2fs_inode_cachep;
  42
  43#ifdef CONFIG_F2FS_FAULT_INJECTION
  44
  45const char *f2fs_fault_name[FAULT_MAX] = {
  46        [FAULT_KMALLOC]         = "kmalloc",
  47        [FAULT_KVMALLOC]        = "kvmalloc",
  48        [FAULT_PAGE_ALLOC]      = "page alloc",
  49        [FAULT_PAGE_GET]        = "page get",
  50        [FAULT_ALLOC_NID]       = "alloc nid",
  51        [FAULT_ORPHAN]          = "orphan",
  52        [FAULT_BLOCK]           = "no more block",
  53        [FAULT_DIR_DEPTH]       = "too big dir depth",
  54        [FAULT_EVICT_INODE]     = "evict_inode fail",
  55        [FAULT_TRUNCATE]        = "truncate fail",
  56        [FAULT_READ_IO]         = "read IO error",
  57        [FAULT_CHECKPOINT]      = "checkpoint error",
  58        [FAULT_DISCARD]         = "discard error",
  59        [FAULT_WRITE_IO]        = "write IO error",
  60        [FAULT_SLAB_ALLOC]      = "slab alloc",
  61        [FAULT_DQUOT_INIT]      = "dquot initialize",
  62};
  63
  64void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
  65                                                        unsigned int type)
  66{
  67        struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
  68
  69        if (rate) {
  70                atomic_set(&ffi->inject_ops, 0);
  71                ffi->inject_rate = rate;
  72        }
  73
  74        if (type)
  75                ffi->inject_type = type;
  76
  77        if (!rate && !type)
  78                memset(ffi, 0, sizeof(struct f2fs_fault_info));
  79}
  80#endif
  81
  82/* f2fs-wide shrinker description */
  83static struct shrinker f2fs_shrinker_info = {
  84        .scan_objects = f2fs_shrink_scan,
  85        .count_objects = f2fs_shrink_count,
  86        .seeks = DEFAULT_SEEKS,
  87};
  88
  89enum {
  90        Opt_gc_background,
  91        Opt_disable_roll_forward,
  92        Opt_norecovery,
  93        Opt_discard,
  94        Opt_nodiscard,
  95        Opt_noheap,
  96        Opt_heap,
  97        Opt_user_xattr,
  98        Opt_nouser_xattr,
  99        Opt_acl,
 100        Opt_noacl,
 101        Opt_active_logs,
 102        Opt_disable_ext_identify,
 103        Opt_inline_xattr,
 104        Opt_noinline_xattr,
 105        Opt_inline_xattr_size,
 106        Opt_inline_data,
 107        Opt_inline_dentry,
 108        Opt_noinline_dentry,
 109        Opt_flush_merge,
 110        Opt_noflush_merge,
 111        Opt_nobarrier,
 112        Opt_fastboot,
 113        Opt_extent_cache,
 114        Opt_noextent_cache,
 115        Opt_noinline_data,
 116        Opt_data_flush,
 117        Opt_reserve_root,
 118        Opt_resgid,
 119        Opt_resuid,
 120        Opt_mode,
 121        Opt_io_size_bits,
 122        Opt_fault_injection,
 123        Opt_fault_type,
 124        Opt_lazytime,
 125        Opt_nolazytime,
 126        Opt_quota,
 127        Opt_noquota,
 128        Opt_usrquota,
 129        Opt_grpquota,
 130        Opt_prjquota,
 131        Opt_usrjquota,
 132        Opt_grpjquota,
 133        Opt_prjjquota,
 134        Opt_offusrjquota,
 135        Opt_offgrpjquota,
 136        Opt_offprjjquota,
 137        Opt_jqfmt_vfsold,
 138        Opt_jqfmt_vfsv0,
 139        Opt_jqfmt_vfsv1,
 140        Opt_whint,
 141        Opt_alloc,
 142        Opt_fsync,
 143        Opt_test_dummy_encryption,
 144        Opt_inlinecrypt,
 145        Opt_checkpoint_disable,
 146        Opt_checkpoint_disable_cap,
 147        Opt_checkpoint_disable_cap_perc,
 148        Opt_checkpoint_enable,
 149        Opt_checkpoint_merge,
 150        Opt_nocheckpoint_merge,
 151        Opt_compress_algorithm,
 152        Opt_compress_log_size,
 153        Opt_compress_extension,
 154        Opt_nocompress_extension,
 155        Opt_compress_chksum,
 156        Opt_compress_mode,
 157        Opt_compress_cache,
 158        Opt_atgc,
 159        Opt_gc_merge,
 160        Opt_nogc_merge,
 161        Opt_discard_unit,
 162        Opt_err,
 163};
 164
 165static match_table_t f2fs_tokens = {
 166        {Opt_gc_background, "background_gc=%s"},
 167        {Opt_disable_roll_forward, "disable_roll_forward"},
 168        {Opt_norecovery, "norecovery"},
 169        {Opt_discard, "discard"},
 170        {Opt_nodiscard, "nodiscard"},
 171        {Opt_noheap, "no_heap"},
 172        {Opt_heap, "heap"},
 173        {Opt_user_xattr, "user_xattr"},
 174        {Opt_nouser_xattr, "nouser_xattr"},
 175        {Opt_acl, "acl"},
 176        {Opt_noacl, "noacl"},
 177        {Opt_active_logs, "active_logs=%u"},
 178        {Opt_disable_ext_identify, "disable_ext_identify"},
 179        {Opt_inline_xattr, "inline_xattr"},
 180        {Opt_noinline_xattr, "noinline_xattr"},
 181        {Opt_inline_xattr_size, "inline_xattr_size=%u"},
 182        {Opt_inline_data, "inline_data"},
 183        {Opt_inline_dentry, "inline_dentry"},
 184        {Opt_noinline_dentry, "noinline_dentry"},
 185        {Opt_flush_merge, "flush_merge"},
 186        {Opt_noflush_merge, "noflush_merge"},
 187        {Opt_nobarrier, "nobarrier"},
 188        {Opt_fastboot, "fastboot"},
 189        {Opt_extent_cache, "extent_cache"},
 190        {Opt_noextent_cache, "noextent_cache"},
 191        {Opt_noinline_data, "noinline_data"},
 192        {Opt_data_flush, "data_flush"},
 193        {Opt_reserve_root, "reserve_root=%u"},
 194        {Opt_resgid, "resgid=%u"},
 195        {Opt_resuid, "resuid=%u"},
 196        {Opt_mode, "mode=%s"},
 197        {Opt_io_size_bits, "io_bits=%u"},
 198        {Opt_fault_injection, "fault_injection=%u"},
 199        {Opt_fault_type, "fault_type=%u"},
 200        {Opt_lazytime, "lazytime"},
 201        {Opt_nolazytime, "nolazytime"},
 202        {Opt_quota, "quota"},
 203        {Opt_noquota, "noquota"},
 204        {Opt_usrquota, "usrquota"},
 205        {Opt_grpquota, "grpquota"},
 206        {Opt_prjquota, "prjquota"},
 207        {Opt_usrjquota, "usrjquota=%s"},
 208        {Opt_grpjquota, "grpjquota=%s"},
 209        {Opt_prjjquota, "prjjquota=%s"},
 210        {Opt_offusrjquota, "usrjquota="},
 211        {Opt_offgrpjquota, "grpjquota="},
 212        {Opt_offprjjquota, "prjjquota="},
 213        {Opt_jqfmt_vfsold, "jqfmt=vfsold"},
 214        {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
 215        {Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
 216        {Opt_whint, "whint_mode=%s"},
 217        {Opt_alloc, "alloc_mode=%s"},
 218        {Opt_fsync, "fsync_mode=%s"},
 219        {Opt_test_dummy_encryption, "test_dummy_encryption=%s"},
 220        {Opt_test_dummy_encryption, "test_dummy_encryption"},
 221        {Opt_inlinecrypt, "inlinecrypt"},
 222        {Opt_checkpoint_disable, "checkpoint=disable"},
 223        {Opt_checkpoint_disable_cap, "checkpoint=disable:%u"},
 224        {Opt_checkpoint_disable_cap_perc, "checkpoint=disable:%u%%"},
 225        {Opt_checkpoint_enable, "checkpoint=enable"},
 226        {Opt_checkpoint_merge, "checkpoint_merge"},
 227        {Opt_nocheckpoint_merge, "nocheckpoint_merge"},
 228        {Opt_compress_algorithm, "compress_algorithm=%s"},
 229        {Opt_compress_log_size, "compress_log_size=%u"},
 230        {Opt_compress_extension, "compress_extension=%s"},
 231        {Opt_nocompress_extension, "nocompress_extension=%s"},
 232        {Opt_compress_chksum, "compress_chksum"},
 233        {Opt_compress_mode, "compress_mode=%s"},
 234        {Opt_compress_cache, "compress_cache"},
 235        {Opt_atgc, "atgc"},
 236        {Opt_gc_merge, "gc_merge"},
 237        {Opt_nogc_merge, "nogc_merge"},
 238        {Opt_discard_unit, "discard_unit=%s"},
 239        {Opt_err, NULL},
 240};
 241
 242void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...)
 243{
 244        struct va_format vaf;
 245        va_list args;
 246        int level;
 247
 248        va_start(args, fmt);
 249
 250        level = printk_get_level(fmt);
 251        vaf.fmt = printk_skip_level(fmt);
 252        vaf.va = &args;
 253        printk("%c%cF2FS-fs (%s): %pV\n",
 254               KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
 255
 256        va_end(args);
 257}
 258
 259#ifdef CONFIG_UNICODE
 260static const struct f2fs_sb_encodings {
 261        __u16 magic;
 262        char *name;
 263        char *version;
 264} f2fs_sb_encoding_map[] = {
 265        {F2FS_ENC_UTF8_12_1, "utf8", "12.1.0"},
 266};
 267
 268static int f2fs_sb_read_encoding(const struct f2fs_super_block *sb,
 269                                 const struct f2fs_sb_encodings **encoding,
 270                                 __u16 *flags)
 271{
 272        __u16 magic = le16_to_cpu(sb->s_encoding);
 273        int i;
 274
 275        for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
 276                if (magic == f2fs_sb_encoding_map[i].magic)
 277                        break;
 278
 279        if (i >= ARRAY_SIZE(f2fs_sb_encoding_map))
 280                return -EINVAL;
 281
 282        *encoding = &f2fs_sb_encoding_map[i];
 283        *flags = le16_to_cpu(sb->s_encoding_flags);
 284
 285        return 0;
 286}
 287
 288struct kmem_cache *f2fs_cf_name_slab;
 289static 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        if (!f2fs_cf_name_slab)
 294                return -ENOMEM;
 295        return 0;
 296}
 297
 298static void f2fs_destroy_casefold_cache(void)
 299{
 300        kmem_cache_destroy(f2fs_cf_name_slab);
 301}
 302#else
 303static int __init f2fs_create_casefold_cache(void) { return 0; }
 304static void f2fs_destroy_casefold_cache(void) { }
 305#endif
 306
 307static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
 308{
 309        block_t limit = min((sbi->user_block_count << 1) / 1000,
 310                        sbi->user_block_count - sbi->reserved_blocks);
 311
 312        /* limit is 0.2% */
 313        if (test_opt(sbi, RESERVE_ROOT) &&
 314                        F2FS_OPTION(sbi).root_reserved_blocks > limit) {
 315                F2FS_OPTION(sbi).root_reserved_blocks = limit;
 316                f2fs_info(sbi, "Reduce reserved blocks for root = %u",
 317                          F2FS_OPTION(sbi).root_reserved_blocks);
 318        }
 319        if (!test_opt(sbi, RESERVE_ROOT) &&
 320                (!uid_eq(F2FS_OPTION(sbi).s_resuid,
 321                                make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
 322                !gid_eq(F2FS_OPTION(sbi).s_resgid,
 323                                make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
 324                f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
 325                          from_kuid_munged(&init_user_ns,
 326                                           F2FS_OPTION(sbi).s_resuid),
 327                          from_kgid_munged(&init_user_ns,
 328                                           F2FS_OPTION(sbi).s_resgid));
 329}
 330
 331static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
 332{
 333        if (!F2FS_OPTION(sbi).unusable_cap_perc)
 334                return;
 335
 336        if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
 337                F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
 338        else
 339                F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
 340                                        F2FS_OPTION(sbi).unusable_cap_perc;
 341
 342        f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
 343                        F2FS_OPTION(sbi).unusable_cap,
 344                        F2FS_OPTION(sbi).unusable_cap_perc);
 345}
 346
 347static void init_once(void *foo)
 348{
 349        struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
 350
 351        inode_init_once(&fi->vfs_inode);
 352}
 353
 354#ifdef CONFIG_QUOTA
 355static const char * const quotatypes[] = INITQFNAMES;
 356#define QTYPE2NAME(t) (quotatypes[t])
 357static int f2fs_set_qf_name(struct super_block *sb, int qtype,
 358                                                        substring_t *args)
 359{
 360        struct f2fs_sb_info *sbi = F2FS_SB(sb);
 361        char *qname;
 362        int ret = -EINVAL;
 363
 364        if (sb_any_quota_loaded(sb) && !F2FS_OPTION(sbi).s_qf_names[qtype]) {
 365                f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
 366                return -EINVAL;
 367        }
 368        if (f2fs_sb_has_quota_ino(sbi)) {
 369                f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
 370                return 0;
 371        }
 372
 373        qname = match_strdup(args);
 374        if (!qname) {
 375                f2fs_err(sbi, "Not enough memory for storing quotafile name");
 376                return -ENOMEM;
 377        }
 378        if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
 379                if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
 380                        ret = 0;
 381                else
 382                        f2fs_err(sbi, "%s quota file already specified",
 383                                 QTYPE2NAME(qtype));
 384                goto errout;
 385        }
 386        if (strchr(qname, '/')) {
 387                f2fs_err(sbi, "quotafile must be on filesystem root");
 388                goto errout;
 389        }
 390        F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
 391        set_opt(sbi, QUOTA);
 392        return 0;
 393errout:
 394        kfree(qname);
 395        return ret;
 396}
 397
 398static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
 399{
 400        struct f2fs_sb_info *sbi = F2FS_SB(sb);
 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        kfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
 407        F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
 408        return 0;
 409}
 410
 411static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
 412{
 413        /*
 414         * We do the test below only for project quotas. 'usrquota' and
 415         * 'grpquota' mount options are allowed even without quota feature
 416         * to support legacy quotas in quota files.
 417         */
 418        if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi)) {
 419                f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
 420                return -1;
 421        }
 422        if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
 423                        F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
 424                        F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) {
 425                if (test_opt(sbi, USRQUOTA) &&
 426                                F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
 427                        clear_opt(sbi, USRQUOTA);
 428
 429                if (test_opt(sbi, GRPQUOTA) &&
 430                                F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
 431                        clear_opt(sbi, GRPQUOTA);
 432
 433                if (test_opt(sbi, PRJQUOTA) &&
 434                                F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
 435                        clear_opt(sbi, PRJQUOTA);
 436
 437                if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
 438                                test_opt(sbi, PRJQUOTA)) {
 439                        f2fs_err(sbi, "old and new quota format mixing");
 440                        return -1;
 441                }
 442
 443                if (!F2FS_OPTION(sbi).s_jquota_fmt) {
 444                        f2fs_err(sbi, "journaled quota format not specified");
 445                        return -1;
 446                }
 447        }
 448
 449        if (f2fs_sb_has_quota_ino(sbi) && F2FS_OPTION(sbi).s_jquota_fmt) {
 450                f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
 451                F2FS_OPTION(sbi).s_jquota_fmt = 0;
 452        }
 453        return 0;
 454}
 455#endif
 456
 457static int f2fs_set_test_dummy_encryption(struct super_block *sb,
 458                                          const char *opt,
 459                                          const substring_t *arg,
 460                                          bool is_remount)
 461{
 462        struct f2fs_sb_info *sbi = F2FS_SB(sb);
 463#ifdef CONFIG_FS_ENCRYPTION
 464        int err;
 465
 466        if (!f2fs_sb_has_encrypt(sbi)) {
 467                f2fs_err(sbi, "Encrypt feature is off");
 468                return -EINVAL;
 469        }
 470
 471        /*
 472         * This mount option is just for testing, and it's not worthwhile to
 473         * implement the extra complexity (e.g. RCU protection) that would be
 474         * needed to allow it to be set or changed during remount.  We do allow
 475         * it to be specified during remount, but only if there is no change.
 476         */
 477        if (is_remount && !F2FS_OPTION(sbi).dummy_enc_policy.policy) {
 478                f2fs_warn(sbi, "Can't set test_dummy_encryption on remount");
 479                return -EINVAL;
 480        }
 481        err = fscrypt_set_test_dummy_encryption(
 482                sb, arg->from, &F2FS_OPTION(sbi).dummy_enc_policy);
 483        if (err) {
 484                if (err == -EEXIST)
 485                        f2fs_warn(sbi,
 486                                  "Can't change test_dummy_encryption on remount");
 487                else if (err == -EINVAL)
 488                        f2fs_warn(sbi, "Value of option \"%s\" is unrecognized",
 489                                  opt);
 490                else
 491                        f2fs_warn(sbi, "Error processing option \"%s\" [%d]",
 492                                  opt, err);
 493                return -EINVAL;
 494        }
 495        f2fs_warn(sbi, "Test dummy encryption mode enabled");
 496#else
 497        f2fs_warn(sbi, "Test dummy encryption mount option ignored");
 498#endif
 499        return 0;
 500}
 501
 502#ifdef CONFIG_F2FS_FS_COMPRESSION
 503/*
 504 * 1. The same extension name cannot not appear in both compress and non-compress extension
 505 * at the same time.
 506 * 2. If the compress extension specifies all files, the types specified by the non-compress
 507 * extension will be treated as special cases and will not be compressed.
 508 * 3. Don't allow the non-compress extension specifies all files.
 509 */
 510static int f2fs_test_compress_extension(struct f2fs_sb_info *sbi)
 511{
 512        unsigned char (*ext)[F2FS_EXTENSION_LEN];
 513        unsigned char (*noext)[F2FS_EXTENSION_LEN];
 514        int ext_cnt, noext_cnt, index = 0, no_index = 0;
 515
 516        ext = F2FS_OPTION(sbi).extensions;
 517        ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
 518        noext = F2FS_OPTION(sbi).noextensions;
 519        noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
 520
 521        if (!noext_cnt)
 522                return 0;
 523
 524        for (no_index = 0; no_index < noext_cnt; no_index++) {
 525                if (!strcasecmp("*", noext[no_index])) {
 526                        f2fs_info(sbi, "Don't allow the nocompress extension specifies all files");
 527                        return -EINVAL;
 528                }
 529                for (index = 0; index < ext_cnt; index++) {
 530                        if (!strcasecmp(ext[index], noext[no_index])) {
 531                                f2fs_info(sbi, "Don't allow the same extension %s appear in both compress and nocompress extension",
 532                                                ext[index]);
 533                                return -EINVAL;
 534                        }
 535                }
 536        }
 537        return 0;
 538}
 539
 540#ifdef CONFIG_F2FS_FS_LZ4
 541static int f2fs_set_lz4hc_level(struct f2fs_sb_info *sbi, const char *str)
 542{
 543#ifdef CONFIG_F2FS_FS_LZ4HC
 544        unsigned int level;
 545#endif
 546
 547        if (strlen(str) == 3) {
 548                F2FS_OPTION(sbi).compress_level = 0;
 549                return 0;
 550        }
 551
 552#ifdef CONFIG_F2FS_FS_LZ4HC
 553        str += 3;
 554
 555        if (str[0] != ':') {
 556                f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
 557                return -EINVAL;
 558        }
 559        if (kstrtouint(str + 1, 10, &level))
 560                return -EINVAL;
 561
 562        if (level < LZ4HC_MIN_CLEVEL || level > LZ4HC_MAX_CLEVEL) {
 563                f2fs_info(sbi, "invalid lz4hc compress level: %d", level);
 564                return -EINVAL;
 565        }
 566
 567        F2FS_OPTION(sbi).compress_level = level;
 568        return 0;
 569#else
 570        f2fs_info(sbi, "kernel doesn't support lz4hc compression");
 571        return -EINVAL;
 572#endif
 573}
 574#endif
 575
 576#ifdef CONFIG_F2FS_FS_ZSTD
 577static int f2fs_set_zstd_level(struct f2fs_sb_info *sbi, const char *str)
 578{
 579        unsigned int level;
 580        int len = 4;
 581
 582        if (strlen(str) == len) {
 583                F2FS_OPTION(sbi).compress_level = 0;
 584                return 0;
 585        }
 586
 587        str += len;
 588
 589        if (str[0] != ':') {
 590                f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
 591                return -EINVAL;
 592        }
 593        if (kstrtouint(str + 1, 10, &level))
 594                return -EINVAL;
 595
 596        if (!level || level > zstd_max_clevel()) {
 597                f2fs_info(sbi, "invalid zstd compress level: %d", level);
 598                return -EINVAL;
 599        }
 600
 601        F2FS_OPTION(sbi).compress_level = level;
 602        return 0;
 603}
 604#endif
 605#endif
 606
 607static int parse_options(struct super_block *sb, char *options, bool is_remount)
 608{
 609        struct f2fs_sb_info *sbi = F2FS_SB(sb);
 610        substring_t args[MAX_OPT_ARGS];
 611#ifdef CONFIG_F2FS_FS_COMPRESSION
 612        unsigned char (*ext)[F2FS_EXTENSION_LEN];
 613        unsigned char (*noext)[F2FS_EXTENSION_LEN];
 614        int ext_cnt, noext_cnt;
 615#endif
 616        char *p, *name;
 617        int arg = 0;
 618        kuid_t uid;
 619        kgid_t gid;
 620        int ret;
 621
 622        if (!options)
 623                goto default_check;
 624
 625        while ((p = strsep(&options, ",")) != NULL) {
 626                int token;
 627
 628                if (!*p)
 629                        continue;
 630                /*
 631                 * Initialize args struct so we know whether arg was
 632                 * found; some options take optional arguments.
 633                 */
 634                args[0].to = args[0].from = NULL;
 635                token = match_token(p, f2fs_tokens, args);
 636
 637                switch (token) {
 638                case Opt_gc_background:
 639                        name = match_strdup(&args[0]);
 640
 641                        if (!name)
 642                                return -ENOMEM;
 643                        if (!strcmp(name, "on")) {
 644                                F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
 645                        } else if (!strcmp(name, "off")) {
 646                                F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_OFF;
 647                        } else if (!strcmp(name, "sync")) {
 648                                F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_SYNC;
 649                        } else {
 650                                kfree(name);
 651                                return -EINVAL;
 652                        }
 653                        kfree(name);
 654                        break;
 655                case Opt_disable_roll_forward:
 656                        set_opt(sbi, DISABLE_ROLL_FORWARD);
 657                        break;
 658                case Opt_norecovery:
 659                        /* this option mounts f2fs with ro */
 660                        set_opt(sbi, NORECOVERY);
 661                        if (!f2fs_readonly(sb))
 662                                return -EINVAL;
 663                        break;
 664                case Opt_discard:
 665                        if (!f2fs_hw_support_discard(sbi)) {
 666                                f2fs_warn(sbi, "device does not support discard");
 667                                break;
 668                        }
 669                        set_opt(sbi, DISCARD);
 670                        break;
 671                case Opt_nodiscard:
 672                        if (f2fs_hw_should_discard(sbi)) {
 673                                f2fs_warn(sbi, "discard is required for zoned block devices");
 674                                return -EINVAL;
 675                        }
 676                        clear_opt(sbi, DISCARD);
 677                        break;
 678                case Opt_noheap:
 679                        set_opt(sbi, NOHEAP);
 680                        break;
 681                case Opt_heap:
 682                        clear_opt(sbi, NOHEAP);
 683                        break;
 684#ifdef CONFIG_F2FS_FS_XATTR
 685                case Opt_user_xattr:
 686                        set_opt(sbi, XATTR_USER);
 687                        break;
 688                case Opt_nouser_xattr:
 689                        clear_opt(sbi, XATTR_USER);
 690                        break;
 691                case Opt_inline_xattr:
 692                        set_opt(sbi, INLINE_XATTR);
 693                        break;
 694                case Opt_noinline_xattr:
 695                        clear_opt(sbi, INLINE_XATTR);
 696                        break;
 697                case Opt_inline_xattr_size:
 698                        if (args->from && match_int(args, &arg))
 699                                return -EINVAL;
 700                        set_opt(sbi, INLINE_XATTR_SIZE);
 701                        F2FS_OPTION(sbi).inline_xattr_size = arg;
 702                        break;
 703#else
 704                case Opt_user_xattr:
 705                        f2fs_info(sbi, "user_xattr options not supported");
 706                        break;
 707                case Opt_nouser_xattr:
 708                        f2fs_info(sbi, "nouser_xattr options not supported");
 709                        break;
 710                case Opt_inline_xattr:
 711                        f2fs_info(sbi, "inline_xattr options not supported");
 712                        break;
 713                case Opt_noinline_xattr:
 714                        f2fs_info(sbi, "noinline_xattr options not supported");
 715                        break;
 716#endif
 717#ifdef CONFIG_F2FS_FS_POSIX_ACL
 718                case Opt_acl:
 719                        set_opt(sbi, POSIX_ACL);
 720                        break;
 721                case Opt_noacl:
 722                        clear_opt(sbi, POSIX_ACL);
 723                        break;
 724#else
 725                case Opt_acl:
 726                        f2fs_info(sbi, "acl options not supported");
 727                        break;
 728                case Opt_noacl:
 729                        f2fs_info(sbi, "noacl options not supported");
 730                        break;
 731#endif
 732                case Opt_active_logs:
 733                        if (args->from && match_int(args, &arg))
 734                                return -EINVAL;
 735                        if (arg != 2 && arg != 4 &&
 736                                arg != NR_CURSEG_PERSIST_TYPE)
 737                                return -EINVAL;
 738                        F2FS_OPTION(sbi).active_logs = arg;
 739                        break;
 740                case Opt_disable_ext_identify:
 741                        set_opt(sbi, DISABLE_EXT_IDENTIFY);
 742                        break;
 743                case Opt_inline_data:
 744                        set_opt(sbi, INLINE_DATA);
 745                        break;
 746                case Opt_inline_dentry:
 747                        set_opt(sbi, INLINE_DENTRY);
 748                        break;
 749                case Opt_noinline_dentry:
 750                        clear_opt(sbi, INLINE_DENTRY);
 751                        break;
 752                case Opt_flush_merge:
 753                        set_opt(sbi, FLUSH_MERGE);
 754                        break;
 755                case Opt_noflush_merge:
 756                        clear_opt(sbi, FLUSH_MERGE);
 757                        break;
 758                case Opt_nobarrier:
 759                        set_opt(sbi, NOBARRIER);
 760                        break;
 761                case Opt_fastboot:
 762                        set_opt(sbi, FASTBOOT);
 763                        break;
 764                case Opt_extent_cache:
 765                        set_opt(sbi, EXTENT_CACHE);
 766                        break;
 767                case Opt_noextent_cache:
 768                        clear_opt(sbi, EXTENT_CACHE);
 769                        break;
 770                case Opt_noinline_data:
 771                        clear_opt(sbi, INLINE_DATA);
 772                        break;
 773                case Opt_data_flush:
 774                        set_opt(sbi, DATA_FLUSH);
 775                        break;
 776                case Opt_reserve_root:
 777                        if (args->from && match_int(args, &arg))
 778                                return -EINVAL;
 779                        if (test_opt(sbi, RESERVE_ROOT)) {
 780                                f2fs_info(sbi, "Preserve previous reserve_root=%u",
 781                                          F2FS_OPTION(sbi).root_reserved_blocks);
 782                        } else {
 783                                F2FS_OPTION(sbi).root_reserved_blocks = arg;
 784                                set_opt(sbi, RESERVE_ROOT);
 785                        }
 786                        break;
 787                case Opt_resuid:
 788                        if (args->from && match_int(args, &arg))
 789                                return -EINVAL;
 790                        uid = make_kuid(current_user_ns(), arg);
 791                        if (!uid_valid(uid)) {
 792                                f2fs_err(sbi, "Invalid uid value %d", arg);
 793                                return -EINVAL;
 794                        }
 795                        F2FS_OPTION(sbi).s_resuid = uid;
 796                        break;
 797                case Opt_resgid:
 798                        if (args->from && match_int(args, &arg))
 799                                return -EINVAL;
 800                        gid = make_kgid(current_user_ns(), arg);
 801                        if (!gid_valid(gid)) {
 802                                f2fs_err(sbi, "Invalid gid value %d", arg);
 803                                return -EINVAL;
 804                        }
 805                        F2FS_OPTION(sbi).s_resgid = gid;
 806                        break;
 807                case Opt_mode:
 808                        name = match_strdup(&args[0]);
 809
 810                        if (!name)
 811                                return -ENOMEM;
 812                        if (!strcmp(name, "adaptive")) {
 813                                if (f2fs_sb_has_blkzoned(sbi)) {
 814                                        f2fs_warn(sbi, "adaptive mode is not allowed with zoned block device feature");
 815                                        kfree(name);
 816                                        return -EINVAL;
 817                                }
 818                                F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
 819                        } else if (!strcmp(name, "lfs")) {
 820                                F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
 821                        } else if (!strcmp(name, "fragment:segment")) {
 822                                F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_SEG;
 823                        } else if (!strcmp(name, "fragment:block")) {
 824                                F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_BLK;
 825                        } else {
 826                                kfree(name);
 827                                return -EINVAL;
 828                        }
 829                        kfree(name);
 830                        break;
 831                case Opt_io_size_bits:
 832                        if (args->from && match_int(args, &arg))
 833                                return -EINVAL;
 834                        if (arg <= 0 || arg > __ilog2_u32(BIO_MAX_VECS)) {
 835                                f2fs_warn(sbi, "Not support %d, larger than %d",
 836                                          1 << arg, BIO_MAX_VECS);
 837                                return -EINVAL;
 838                        }
 839                        F2FS_OPTION(sbi).write_io_size_bits = arg;
 840                        break;
 841#ifdef CONFIG_F2FS_FAULT_INJECTION
 842                case Opt_fault_injection:
 843                        if (args->from && match_int(args, &arg))
 844                                return -EINVAL;
 845                        f2fs_build_fault_attr(sbi, arg, F2FS_ALL_FAULT_TYPE);
 846                        set_opt(sbi, FAULT_INJECTION);
 847                        break;
 848
 849                case Opt_fault_type:
 850                        if (args->from && match_int(args, &arg))
 851                                return -EINVAL;
 852                        f2fs_build_fault_attr(sbi, 0, arg);
 853                        set_opt(sbi, FAULT_INJECTION);
 854                        break;
 855#else
 856                case Opt_fault_injection:
 857                        f2fs_info(sbi, "fault_injection options not supported");
 858                        break;
 859
 860                case Opt_fault_type:
 861                        f2fs_info(sbi, "fault_type options not supported");
 862                        break;
 863#endif
 864                case Opt_lazytime:
 865                        sb->s_flags |= SB_LAZYTIME;
 866                        break;
 867                case Opt_nolazytime:
 868                        sb->s_flags &= ~SB_LAZYTIME;
 869                        break;
 870#ifdef CONFIG_QUOTA
 871                case Opt_quota:
 872                case Opt_usrquota:
 873                        set_opt(sbi, USRQUOTA);
 874                        break;
 875                case Opt_grpquota:
 876                        set_opt(sbi, GRPQUOTA);
 877                        break;
 878                case Opt_prjquota:
 879                        set_opt(sbi, PRJQUOTA);
 880                        break;
 881                case Opt_usrjquota:
 882                        ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
 883                        if (ret)
 884                                return ret;
 885                        break;
 886                case Opt_grpjquota:
 887                        ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
 888                        if (ret)
 889                                return ret;
 890                        break;
 891                case Opt_prjjquota:
 892                        ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
 893                        if (ret)
 894                                return ret;
 895                        break;
 896                case Opt_offusrjquota:
 897                        ret = f2fs_clear_qf_name(sb, USRQUOTA);
 898                        if (ret)
 899                                return ret;
 900                        break;
 901                case Opt_offgrpjquota:
 902                        ret = f2fs_clear_qf_name(sb, GRPQUOTA);
 903                        if (ret)
 904                                return ret;
 905                        break;
 906                case Opt_offprjjquota:
 907                        ret = f2fs_clear_qf_name(sb, PRJQUOTA);
 908                        if (ret)
 909                                return ret;
 910                        break;
 911                case Opt_jqfmt_vfsold:
 912                        F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
 913                        break;
 914                case Opt_jqfmt_vfsv0:
 915                        F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
 916                        break;
 917                case Opt_jqfmt_vfsv1:
 918                        F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
 919                        break;
 920                case Opt_noquota:
 921                        clear_opt(sbi, QUOTA);
 922                        clear_opt(sbi, USRQUOTA);
 923                        clear_opt(sbi, GRPQUOTA);
 924                        clear_opt(sbi, PRJQUOTA);
 925                        break;
 926#else
 927                case Opt_quota:
 928                case Opt_usrquota:
 929                case Opt_grpquota:
 930                case Opt_prjquota:
 931                case Opt_usrjquota:
 932                case Opt_grpjquota:
 933                case Opt_prjjquota:
 934                case Opt_offusrjquota:
 935                case Opt_offgrpjquota:
 936                case Opt_offprjjquota:
 937                case Opt_jqfmt_vfsold:
 938                case Opt_jqfmt_vfsv0:
 939                case Opt_jqfmt_vfsv1:
 940                case Opt_noquota:
 941                        f2fs_info(sbi, "quota operations not supported");
 942                        break;
 943#endif
 944                case Opt_whint:
 945                        name = match_strdup(&args[0]);
 946                        if (!name)
 947                                return -ENOMEM;
 948                        if (!strcmp(name, "user-based")) {
 949                                F2FS_OPTION(sbi).whint_mode = WHINT_MODE_USER;
 950                        } else if (!strcmp(name, "off")) {
 951                                F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
 952                        } else if (!strcmp(name, "fs-based")) {
 953                                F2FS_OPTION(sbi).whint_mode = WHINT_MODE_FS;
 954                        } else {
 955                                kfree(name);
 956                                return -EINVAL;
 957                        }
 958                        kfree(name);
 959                        break;
 960                case Opt_alloc:
 961                        name = match_strdup(&args[0]);
 962                        if (!name)
 963                                return -ENOMEM;
 964
 965                        if (!strcmp(name, "default")) {
 966                                F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
 967                        } else if (!strcmp(name, "reuse")) {
 968                                F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
 969                        } else {
 970                                kfree(name);
 971                                return -EINVAL;
 972                        }
 973                        kfree(name);
 974                        break;
 975                case Opt_fsync:
 976                        name = match_strdup(&args[0]);
 977                        if (!name)
 978                                return -ENOMEM;
 979                        if (!strcmp(name, "posix")) {
 980                                F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
 981                        } else if (!strcmp(name, "strict")) {
 982                                F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_STRICT;
 983                        } else if (!strcmp(name, "nobarrier")) {
 984                                F2FS_OPTION(sbi).fsync_mode =
 985                                                        FSYNC_MODE_NOBARRIER;
 986                        } else {
 987                                kfree(name);
 988                                return -EINVAL;
 989                        }
 990                        kfree(name);
 991                        break;
 992                case Opt_test_dummy_encryption:
 993                        ret = f2fs_set_test_dummy_encryption(sb, p, &args[0],
 994                                                             is_remount);
 995                        if (ret)
 996                                return ret;
 997                        break;
 998                case Opt_inlinecrypt:
 999#ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
1000                        sb->s_flags |= SB_INLINECRYPT;
1001#else
1002                        f2fs_info(sbi, "inline encryption not supported");
1003#endif
1004                        break;
1005                case Opt_checkpoint_disable_cap_perc:
1006                        if (args->from && match_int(args, &arg))
1007                                return -EINVAL;
1008                        if (arg < 0 || arg > 100)
1009                                return -EINVAL;
1010                        F2FS_OPTION(sbi).unusable_cap_perc = arg;
1011                        set_opt(sbi, DISABLE_CHECKPOINT);
1012                        break;
1013                case Opt_checkpoint_disable_cap:
1014                        if (args->from && match_int(args, &arg))
1015                                return -EINVAL;
1016                        F2FS_OPTION(sbi).unusable_cap = arg;
1017                        set_opt(sbi, DISABLE_CHECKPOINT);
1018                        break;
1019                case Opt_checkpoint_disable:
1020                        set_opt(sbi, DISABLE_CHECKPOINT);
1021                        break;
1022                case Opt_checkpoint_enable:
1023                        clear_opt(sbi, DISABLE_CHECKPOINT);
1024                        break;
1025                case Opt_checkpoint_merge:
1026                        set_opt(sbi, MERGE_CHECKPOINT);
1027                        break;
1028                case Opt_nocheckpoint_merge:
1029                        clear_opt(sbi, MERGE_CHECKPOINT);
1030                        break;
1031#ifdef CONFIG_F2FS_FS_COMPRESSION
1032                case Opt_compress_algorithm:
1033                        if (!f2fs_sb_has_compression(sbi)) {
1034                                f2fs_info(sbi, "Image doesn't support compression");
1035                                break;
1036                        }
1037                        name = match_strdup(&args[0]);
1038                        if (!name)
1039                                return -ENOMEM;
1040                        if (!strcmp(name, "lzo")) {
1041#ifdef CONFIG_F2FS_FS_LZO
1042                                F2FS_OPTION(sbi).compress_level = 0;
1043                                F2FS_OPTION(sbi).compress_algorithm =
1044                                                                COMPRESS_LZO;
1045#else
1046                                f2fs_info(sbi, "kernel doesn't support lzo compression");
1047#endif
1048                        } else if (!strncmp(name, "lz4", 3)) {
1049#ifdef CONFIG_F2FS_FS_LZ4
1050                                ret = f2fs_set_lz4hc_level(sbi, name);
1051                                if (ret) {
1052                                        kfree(name);
1053                                        return -EINVAL;
1054                                }
1055                                F2FS_OPTION(sbi).compress_algorithm =
1056                                                                COMPRESS_LZ4;
1057#else
1058                                f2fs_info(sbi, "kernel doesn't support lz4 compression");
1059#endif
1060                        } else if (!strncmp(name, "zstd", 4)) {
1061#ifdef CONFIG_F2FS_FS_ZSTD
1062                                ret = f2fs_set_zstd_level(sbi, name);
1063                                if (ret) {
1064                                        kfree(name);
1065                                        return -EINVAL;
1066                                }
1067                                F2FS_OPTION(sbi).compress_algorithm =
1068                                                                COMPRESS_ZSTD;
1069#else
1070                                f2fs_info(sbi, "kernel doesn't support zstd compression");
1071#endif
1072                        } else if (!strcmp(name, "lzo-rle")) {
1073#ifdef CONFIG_F2FS_FS_LZORLE
1074                                F2FS_OPTION(sbi).compress_level = 0;
1075                                F2FS_OPTION(sbi).compress_algorithm =
1076                                                                COMPRESS_LZORLE;
1077#else
1078                                f2fs_info(sbi, "kernel doesn't support lzorle compression");
1079#endif
1080                        } else {
1081                                kfree(name);
1082                                return -EINVAL;
1083                        }
1084                        kfree(name);
1085                        break;
1086                case Opt_compress_log_size:
1087                        if (!f2fs_sb_has_compression(sbi)) {
1088                                f2fs_info(sbi, "Image doesn't support compression");
1089                                break;
1090                        }
1091                        if (args->from && match_int(args, &arg))
1092                                return -EINVAL;
1093                        if (arg < MIN_COMPRESS_LOG_SIZE ||
1094                                arg > MAX_COMPRESS_LOG_SIZE) {
1095                                f2fs_err(sbi,
1096                                        "Compress cluster log size is out of range");
1097                                return -EINVAL;
1098                        }
1099                        F2FS_OPTION(sbi).compress_log_size = arg;
1100                        break;
1101                case Opt_compress_extension:
1102                        if (!f2fs_sb_has_compression(sbi)) {
1103                                f2fs_info(sbi, "Image doesn't support compression");
1104                                break;
1105                        }
1106                        name = match_strdup(&args[0]);
1107                        if (!name)
1108                                return -ENOMEM;
1109
1110                        ext = F2FS_OPTION(sbi).extensions;
1111                        ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
1112
1113                        if (strlen(name) >= F2FS_EXTENSION_LEN ||
1114                                ext_cnt >= COMPRESS_EXT_NUM) {
1115                                f2fs_err(sbi,
1116                                        "invalid extension length/number");
1117                                kfree(name);
1118                                return -EINVAL;
1119                        }
1120
1121                        strcpy(ext[ext_cnt], name);
1122                        F2FS_OPTION(sbi).compress_ext_cnt++;
1123                        kfree(name);
1124                        break;
1125                case Opt_nocompress_extension:
1126                        if (!f2fs_sb_has_compression(sbi)) {
1127                                f2fs_info(sbi, "Image doesn't support compression");
1128                                break;
1129                        }
1130                        name = match_strdup(&args[0]);
1131                        if (!name)
1132                                return -ENOMEM;
1133
1134                        noext = F2FS_OPTION(sbi).noextensions;
1135                        noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
1136
1137                        if (strlen(name) >= F2FS_EXTENSION_LEN ||
1138                                noext_cnt >= COMPRESS_EXT_NUM) {
1139                                f2fs_err(sbi,
1140                                        "invalid extension length/number");
1141                                kfree(name);
1142                                return -EINVAL;
1143                        }
1144
1145                        strcpy(noext[noext_cnt], name);
1146                        F2FS_OPTION(sbi).nocompress_ext_cnt++;
1147                        kfree(name);
1148                        break;
1149                case Opt_compress_chksum:
1150                        F2FS_OPTION(sbi).compress_chksum = true;
1151                        break;
1152                case Opt_compress_mode:
1153                        name = match_strdup(&args[0]);
1154                        if (!name)
1155                                return -ENOMEM;
1156                        if (!strcmp(name, "fs")) {
1157                                F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
1158                        } else if (!strcmp(name, "user")) {
1159                                F2FS_OPTION(sbi).compress_mode = COMPR_MODE_USER;
1160                        } else {
1161                                kfree(name);
1162                                return -EINVAL;
1163                        }
1164                        kfree(name);
1165                        break;
1166                case Opt_compress_cache:
1167                        set_opt(sbi, COMPRESS_CACHE);
1168                        break;
1169#else
1170                case Opt_compress_algorithm:
1171                case Opt_compress_log_size:
1172                case Opt_compress_extension:
1173                case Opt_nocompress_extension:
1174                case Opt_compress_chksum:
1175                case Opt_compress_mode:
1176                case Opt_compress_cache:
1177                        f2fs_info(sbi, "compression options not supported");
1178                        break;
1179#endif
1180                case Opt_atgc:
1181                        set_opt(sbi, ATGC);
1182                        break;
1183                case Opt_gc_merge:
1184                        set_opt(sbi, GC_MERGE);
1185                        break;
1186                case Opt_nogc_merge:
1187                        clear_opt(sbi, GC_MERGE);
1188                        break;
1189                case Opt_discard_unit:
1190                        name = match_strdup(&args[0]);
1191                        if (!name)
1192                                return -ENOMEM;
1193                        if (!strcmp(name, "block")) {
1194                                F2FS_OPTION(sbi).discard_unit =
1195                                                DISCARD_UNIT_BLOCK;
1196                        } else if (!strcmp(name, "segment")) {
1197                                F2FS_OPTION(sbi).discard_unit =
1198                                                DISCARD_UNIT_SEGMENT;
1199                        } else if (!strcmp(name, "section")) {
1200                                F2FS_OPTION(sbi).discard_unit =
1201                                                DISCARD_UNIT_SECTION;
1202                        } else {
1203                                kfree(name);
1204                                return -EINVAL;
1205                        }
1206                        kfree(name);
1207                        break;
1208                default:
1209                        f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
1210                                 p);
1211                        return -EINVAL;
1212                }
1213        }
1214default_check:
1215#ifdef CONFIG_QUOTA
1216        if (f2fs_check_quota_options(sbi))
1217                return -EINVAL;
1218#else
1219        if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sbi->sb)) {
1220                f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1221                return -EINVAL;
1222        }
1223        if (f2fs_sb_has_project_quota(sbi) && !f2fs_readonly(sbi->sb)) {
1224                f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1225                return -EINVAL;
1226        }
1227#endif
1228#ifndef CONFIG_UNICODE
1229        if (f2fs_sb_has_casefold(sbi)) {
1230                f2fs_err(sbi,
1231                        "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
1232                return -EINVAL;
1233        }
1234#endif
1235        /*
1236         * The BLKZONED feature indicates that the drive was formatted with
1237         * zone alignment optimization. This is optional for host-aware
1238         * devices, but mandatory for host-managed zoned block devices.
1239         */
1240#ifndef CONFIG_BLK_DEV_ZONED
1241        if (f2fs_sb_has_blkzoned(sbi)) {
1242                f2fs_err(sbi, "Zoned block device support is not enabled");
1243                return -EINVAL;
1244        }
1245#endif
1246        if (f2fs_sb_has_blkzoned(sbi)) {
1247                if (F2FS_OPTION(sbi).discard_unit !=
1248                                                DISCARD_UNIT_SECTION) {
1249                        f2fs_info(sbi, "Zoned block device doesn't need small discard, set discard_unit=section by default");
1250                        F2FS_OPTION(sbi).discard_unit =
1251                                        DISCARD_UNIT_SECTION;
1252                }
1253        }
1254
1255#ifdef CONFIG_F2FS_FS_COMPRESSION
1256        if (f2fs_test_compress_extension(sbi)) {
1257                f2fs_err(sbi, "invalid compress or nocompress extension");
1258                return -EINVAL;
1259        }
1260#endif
1261
1262        if (F2FS_IO_SIZE_BITS(sbi) && !f2fs_lfs_mode(sbi)) {
1263                f2fs_err(sbi, "Should set mode=lfs with %uKB-sized IO",
1264                         F2FS_IO_SIZE_KB(sbi));
1265                return -EINVAL;
1266        }
1267
1268        if (test_opt(sbi, INLINE_XATTR_SIZE)) {
1269                int min_size, max_size;
1270
1271                if (!f2fs_sb_has_extra_attr(sbi) ||
1272                        !f2fs_sb_has_flexible_inline_xattr(sbi)) {
1273                        f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
1274                        return -EINVAL;
1275                }
1276                if (!test_opt(sbi, INLINE_XATTR)) {
1277                        f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
1278                        return -EINVAL;
1279                }
1280
1281                min_size = sizeof(struct f2fs_xattr_header) / sizeof(__le32);
1282                max_size = MAX_INLINE_XATTR_SIZE;
1283
1284                if (F2FS_OPTION(sbi).inline_xattr_size < min_size ||
1285                                F2FS_OPTION(sbi).inline_xattr_size > max_size) {
1286                        f2fs_err(sbi, "inline xattr size is out of range: %d ~ %d",
1287                                 min_size, max_size);
1288                        return -EINVAL;
1289                }
1290        }
1291
1292        if (test_opt(sbi, DISABLE_CHECKPOINT) && f2fs_lfs_mode(sbi)) {
1293                f2fs_err(sbi, "LFS not compatible with checkpoint=disable");
1294                return -EINVAL;
1295        }
1296
1297        /* Not pass down write hints if the number of active logs is lesser
1298         * than NR_CURSEG_PERSIST_TYPE.
1299         */
1300        if (F2FS_OPTION(sbi).active_logs != NR_CURSEG_PERSIST_TYPE)
1301                F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1302
1303        if (f2fs_sb_has_readonly(sbi) && !f2fs_readonly(sbi->sb)) {
1304                f2fs_err(sbi, "Allow to mount readonly mode only");
1305                return -EROFS;
1306        }
1307        return 0;
1308}
1309
1310static struct inode *f2fs_alloc_inode(struct super_block *sb)
1311{
1312        struct f2fs_inode_info *fi;
1313
1314        fi = f2fs_kmem_cache_alloc(f2fs_inode_cachep,
1315                                GFP_F2FS_ZERO, false, F2FS_SB(sb));
1316        if (!fi)
1317                return NULL;
1318
1319        init_once((void *) fi);
1320
1321        /* Initialize f2fs-specific inode info */
1322        atomic_set(&fi->dirty_pages, 0);
1323        atomic_set(&fi->i_compr_blocks, 0);
1324        init_rwsem(&fi->i_sem);
1325        spin_lock_init(&fi->i_size_lock);
1326        INIT_LIST_HEAD(&fi->dirty_list);
1327        INIT_LIST_HEAD(&fi->gdirty_list);
1328        INIT_LIST_HEAD(&fi->inmem_ilist);
1329        INIT_LIST_HEAD(&fi->inmem_pages);
1330        mutex_init(&fi->inmem_lock);
1331        init_rwsem(&fi->i_gc_rwsem[READ]);
1332        init_rwsem(&fi->i_gc_rwsem[WRITE]);
1333        init_rwsem(&fi->i_xattr_sem);
1334
1335        /* Will be used by directory only */
1336        fi->i_dir_level = F2FS_SB(sb)->dir_level;
1337
1338        return &fi->vfs_inode;
1339}
1340
1341static int f2fs_drop_inode(struct inode *inode)
1342{
1343        struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1344        int ret;
1345
1346        /*
1347         * during filesystem shutdown, if checkpoint is disabled,
1348         * drop useless meta/node dirty pages.
1349         */
1350        if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1351                if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1352                        inode->i_ino == F2FS_META_INO(sbi)) {
1353                        trace_f2fs_drop_inode(inode, 1);
1354                        return 1;
1355                }
1356        }
1357
1358        /*
1359         * This is to avoid a deadlock condition like below.
1360         * writeback_single_inode(inode)
1361         *  - f2fs_write_data_page
1362         *    - f2fs_gc -> iput -> evict
1363         *       - inode_wait_for_writeback(inode)
1364         */
1365        if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
1366                if (!inode->i_nlink && !is_bad_inode(inode)) {
1367                        /* to avoid evict_inode call simultaneously */
1368                        atomic_inc(&inode->i_count);
1369                        spin_unlock(&inode->i_lock);
1370
1371                        /* some remained atomic pages should discarded */
1372                        if (f2fs_is_atomic_file(inode))
1373                                f2fs_drop_inmem_pages(inode);
1374
1375                        /* should remain fi->extent_tree for writepage */
1376                        f2fs_destroy_extent_node(inode);
1377
1378                        sb_start_intwrite(inode->i_sb);
1379                        f2fs_i_size_write(inode, 0);
1380
1381                        f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1382                                        inode, NULL, 0, DATA);
1383                        truncate_inode_pages_final(inode->i_mapping);
1384
1385                        if (F2FS_HAS_BLOCKS(inode))
1386                                f2fs_truncate(inode);
1387
1388                        sb_end_intwrite(inode->i_sb);
1389
1390                        spin_lock(&inode->i_lock);
1391                        atomic_dec(&inode->i_count);
1392                }
1393                trace_f2fs_drop_inode(inode, 0);
1394                return 0;
1395        }
1396        ret = generic_drop_inode(inode);
1397        if (!ret)
1398                ret = fscrypt_drop_inode(inode);
1399        trace_f2fs_drop_inode(inode, ret);
1400        return ret;
1401}
1402
1403int f2fs_inode_dirtied(struct inode *inode, bool sync)
1404{
1405        struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1406        int ret = 0;
1407
1408        spin_lock(&sbi->inode_lock[DIRTY_META]);
1409        if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1410                ret = 1;
1411        } else {
1412                set_inode_flag(inode, FI_DIRTY_INODE);
1413                stat_inc_dirty_inode(sbi, DIRTY_META);
1414        }
1415        if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1416                list_add_tail(&F2FS_I(inode)->gdirty_list,
1417                                &sbi->inode_list[DIRTY_META]);
1418                inc_page_count(sbi, F2FS_DIRTY_IMETA);
1419        }
1420        spin_unlock(&sbi->inode_lock[DIRTY_META]);
1421        return ret;
1422}
1423
1424void f2fs_inode_synced(struct inode *inode)
1425{
1426        struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1427
1428        spin_lock(&sbi->inode_lock[DIRTY_META]);
1429        if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1430                spin_unlock(&sbi->inode_lock[DIRTY_META]);
1431                return;
1432        }
1433        if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1434                list_del_init(&F2FS_I(inode)->gdirty_list);
1435                dec_page_count(sbi, F2FS_DIRTY_IMETA);
1436        }
1437        clear_inode_flag(inode, FI_DIRTY_INODE);
1438        clear_inode_flag(inode, FI_AUTO_RECOVER);
1439        stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1440        spin_unlock(&sbi->inode_lock[DIRTY_META]);
1441}
1442
1443/*
1444 * f2fs_dirty_inode() is called from __mark_inode_dirty()
1445 *
1446 * We should call set_dirty_inode to write the dirty inode through write_inode.
1447 */
1448static void f2fs_dirty_inode(struct inode *inode, int flags)
1449{
1450        struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1451
1452        if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1453                        inode->i_ino == F2FS_META_INO(sbi))
1454                return;
1455
1456        if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1457                clear_inode_flag(inode, FI_AUTO_RECOVER);
1458
1459        f2fs_inode_dirtied(inode, false);
1460}
1461
1462static void f2fs_free_inode(struct inode *inode)
1463{
1464        fscrypt_free_inode(inode);
1465        kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1466}
1467
1468static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1469{
1470        percpu_counter_destroy(&sbi->alloc_valid_block_count);
1471        percpu_counter_destroy(&sbi->total_valid_inode_count);
1472}
1473
1474static void destroy_device_list(struct f2fs_sb_info *sbi)
1475{
1476        int i;
1477
1478        for (i = 0; i < sbi->s_ndevs; i++) {
1479                blkdev_put(FDEV(i).bdev, FMODE_EXCL);
1480#ifdef CONFIG_BLK_DEV_ZONED
1481                kvfree(FDEV(i).blkz_seq);
1482                kfree(FDEV(i).zone_capacity_blocks);
1483#endif
1484        }
1485        kvfree(sbi->devs);
1486}
1487
1488static void f2fs_put_super(struct super_block *sb)
1489{
1490        struct f2fs_sb_info *sbi = F2FS_SB(sb);
1491        int i;
1492        bool dropped;
1493
1494        /* unregister procfs/sysfs entries in advance to avoid race case */
1495        f2fs_unregister_sysfs(sbi);
1496
1497        f2fs_quota_off_umount(sb);
1498
1499        /* prevent remaining shrinker jobs */
1500        mutex_lock(&sbi->umount_mutex);
1501
1502        /*
1503         * flush all issued checkpoints and stop checkpoint issue thread.
1504         * after then, all checkpoints should be done by each process context.
1505         */
1506        f2fs_stop_ckpt_thread(sbi);
1507
1508        /*
1509         * We don't need to do checkpoint when superblock is clean.
1510         * But, the previous checkpoint was not done by umount, it needs to do
1511         * clean checkpoint again.
1512         */
1513        if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1514                        !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1515                struct cp_control cpc = {
1516                        .reason = CP_UMOUNT,
1517                };
1518                f2fs_write_checkpoint(sbi, &cpc);
1519        }
1520
1521        /* be sure to wait for any on-going discard commands */
1522        dropped = f2fs_issue_discard_timeout(sbi);
1523
1524        if ((f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi)) &&
1525                                        !sbi->discard_blks && !dropped) {
1526                struct cp_control cpc = {
1527                        .reason = CP_UMOUNT | CP_TRIMMED,
1528                };
1529                f2fs_write_checkpoint(sbi, &cpc);
1530        }
1531
1532        /*
1533         * normally superblock is clean, so we need to release this.
1534         * In addition, EIO will skip do checkpoint, we need this as well.
1535         */
1536        f2fs_release_ino_entry(sbi, true);
1537
1538        f2fs_leave_shrinker(sbi);
1539        mutex_unlock(&sbi->umount_mutex);
1540
1541        /* our cp_error case, we can wait for any writeback page */
1542        f2fs_flush_merged_writes(sbi);
1543
1544        f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1545
1546        f2fs_bug_on(sbi, sbi->fsync_node_num);
1547
1548        f2fs_destroy_compress_inode(sbi);
1549
1550        iput(sbi->node_inode);
1551        sbi->node_inode = NULL;
1552
1553        iput(sbi->meta_inode);
1554        sbi->meta_inode = NULL;
1555
1556        /*
1557         * iput() can update stat information, if f2fs_write_checkpoint()
1558         * above failed with error.
1559         */
1560        f2fs_destroy_stats(sbi);
1561
1562        /* destroy f2fs internal modules */
1563        f2fs_destroy_node_manager(sbi);
1564        f2fs_destroy_segment_manager(sbi);
1565
1566        f2fs_destroy_post_read_wq(sbi);
1567
1568        kvfree(sbi->ckpt);
1569
1570        sb->s_fs_info = NULL;
1571        if (sbi->s_chksum_driver)
1572                crypto_free_shash(sbi->s_chksum_driver);
1573        kfree(sbi->raw_super);
1574
1575        destroy_device_list(sbi);
1576        f2fs_destroy_page_array_cache(sbi);
1577        f2fs_destroy_xattr_caches(sbi);
1578        mempool_destroy(sbi->write_io_dummy);
1579#ifdef CONFIG_QUOTA
1580        for (i = 0; i < MAXQUOTAS; i++)
1581                kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1582#endif
1583        fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
1584        destroy_percpu_info(sbi);
1585        f2fs_destroy_iostat(sbi);
1586        for (i = 0; i < NR_PAGE_TYPE; i++)
1587                kvfree(sbi->write_io[i]);
1588#ifdef CONFIG_UNICODE
1589        utf8_unload(sb->s_encoding);
1590#endif
1591        kfree(sbi);
1592}
1593
1594int f2fs_sync_fs(struct super_block *sb, int sync)
1595{
1596        struct f2fs_sb_info *sbi = F2FS_SB(sb);
1597        int err = 0;
1598
1599        if (unlikely(f2fs_cp_error(sbi)))
1600                return 0;
1601        if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1602                return 0;
1603
1604        trace_f2fs_sync_fs(sb, sync);
1605
1606        if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1607                return -EAGAIN;
1608
1609        if (sync)
1610                err = f2fs_issue_checkpoint(sbi);
1611
1612        return err;
1613}
1614
1615static int f2fs_freeze(struct super_block *sb)
1616{
1617        if (f2fs_readonly(sb))
1618                return 0;
1619
1620        /* IO error happened before */
1621        if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1622                return -EIO;
1623
1624        /* must be clean, since sync_filesystem() was already called */
1625        if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1626                return -EINVAL;
1627
1628        /* ensure no checkpoint required */
1629        if (!llist_empty(&F2FS_SB(sb)->cprc_info.issue_list))
1630                return -EINVAL;
1631        return 0;
1632}
1633
1634static int f2fs_unfreeze(struct super_block *sb)
1635{
1636        return 0;
1637}
1638
1639#ifdef CONFIG_QUOTA
1640static int f2fs_statfs_project(struct super_block *sb,
1641                                kprojid_t projid, struct kstatfs *buf)
1642{
1643        struct kqid qid;
1644        struct dquot *dquot;
1645        u64 limit;
1646        u64 curblock;
1647
1648        qid = make_kqid_projid(projid);
1649        dquot = dqget(sb, qid);
1650        if (IS_ERR(dquot))
1651                return PTR_ERR(dquot);
1652        spin_lock(&dquot->dq_dqb_lock);
1653
1654        limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1655                                        dquot->dq_dqb.dqb_bhardlimit);
1656        if (limit)
1657                limit >>= sb->s_blocksize_bits;
1658
1659        if (limit && buf->f_blocks > limit) {
1660                curblock = (dquot->dq_dqb.dqb_curspace +
1661                            dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
1662                buf->f_blocks = limit;
1663                buf->f_bfree = buf->f_bavail =
1664                        (buf->f_blocks > curblock) ?
1665                         (buf->f_blocks - curblock) : 0;
1666        }
1667
1668        limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1669                                        dquot->dq_dqb.dqb_ihardlimit);
1670
1671        if (limit && buf->f_files > limit) {
1672                buf->f_files = limit;
1673                buf->f_ffree =
1674                        (buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1675                         (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1676        }
1677
1678        spin_unlock(&dquot->dq_dqb_lock);
1679        dqput(dquot);
1680        return 0;
1681}
1682#endif
1683
1684static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1685{
1686        struct super_block *sb = dentry->d_sb;
1687        struct f2fs_sb_info *sbi = F2FS_SB(sb);
1688        u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
1689        block_t total_count, user_block_count, start_count;
1690        u64 avail_node_count;
1691
1692        total_count = le64_to_cpu(sbi->raw_super->block_count);
1693        user_block_count = sbi->user_block_count;
1694        start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
1695        buf->f_type = F2FS_SUPER_MAGIC;
1696        buf->f_bsize = sbi->blocksize;
1697
1698        buf->f_blocks = total_count - start_count;
1699        buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1700                                                sbi->current_reserved_blocks;
1701
1702        spin_lock(&sbi->stat_lock);
1703        if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1704                buf->f_bfree = 0;
1705        else
1706                buf->f_bfree -= sbi->unusable_block_count;
1707        spin_unlock(&sbi->stat_lock);
1708
1709        if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1710                buf->f_bavail = buf->f_bfree -
1711                                F2FS_OPTION(sbi).root_reserved_blocks;
1712        else
1713                buf->f_bavail = 0;
1714
1715        avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
1716
1717        if (avail_node_count > user_block_count) {
1718                buf->f_files = user_block_count;
1719                buf->f_ffree = buf->f_bavail;
1720        } else {
1721                buf->f_files = avail_node_count;
1722                buf->f_ffree = min(avail_node_count - valid_node_count(sbi),
1723                                        buf->f_bavail);
1724        }
1725
1726        buf->f_namelen = F2FS_NAME_LEN;
1727        buf->f_fsid    = u64_to_fsid(id);
1728
1729#ifdef CONFIG_QUOTA
1730        if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1731                        sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1732                f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1733        }
1734#endif
1735        return 0;
1736}
1737
1738static inline void f2fs_show_quota_options(struct seq_file *seq,
1739                                           struct super_block *sb)
1740{
1741#ifdef CONFIG_QUOTA
1742        struct f2fs_sb_info *sbi = F2FS_SB(sb);
1743
1744        if (F2FS_OPTION(sbi).s_jquota_fmt) {
1745                char *fmtname = "";
1746
1747                switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1748                case QFMT_VFS_OLD:
1749                        fmtname = "vfsold";
1750                        break;
1751                case QFMT_VFS_V0:
1752                        fmtname = "vfsv0";
1753                        break;
1754                case QFMT_VFS_V1:
1755                        fmtname = "vfsv1";
1756                        break;
1757                }
1758                seq_printf(seq, ",jqfmt=%s", fmtname);
1759        }
1760
1761        if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1762                seq_show_option(seq, "usrjquota",
1763                        F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1764
1765        if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1766                seq_show_option(seq, "grpjquota",
1767                        F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1768
1769        if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1770                seq_show_option(seq, "prjjquota",
1771                        F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1772#endif
1773}
1774
1775#ifdef CONFIG_F2FS_FS_COMPRESSION
1776static inline void f2fs_show_compress_options(struct seq_file *seq,
1777                                                        struct super_block *sb)
1778{
1779        struct f2fs_sb_info *sbi = F2FS_SB(sb);
1780        char *algtype = "";
1781        int i;
1782
1783        if (!f2fs_sb_has_compression(sbi))
1784                return;
1785
1786        switch (F2FS_OPTION(sbi).compress_algorithm) {
1787        case COMPRESS_LZO:
1788                algtype = "lzo";
1789                break;
1790        case COMPRESS_LZ4:
1791                algtype = "lz4";
1792                break;
1793        case COMPRESS_ZSTD:
1794                algtype = "zstd";
1795                break;
1796        case COMPRESS_LZORLE:
1797                algtype = "lzo-rle";
1798                break;
1799        }
1800        seq_printf(seq, ",compress_algorithm=%s", algtype);
1801
1802        if (F2FS_OPTION(sbi).compress_level)
1803                seq_printf(seq, ":%d", F2FS_OPTION(sbi).compress_level);
1804
1805        seq_printf(seq, ",compress_log_size=%u",
1806                        F2FS_OPTION(sbi).compress_log_size);
1807
1808        for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1809                seq_printf(seq, ",compress_extension=%s",
1810                        F2FS_OPTION(sbi).extensions[i]);
1811        }
1812
1813        for (i = 0; i < F2FS_OPTION(sbi).nocompress_ext_cnt; i++) {
1814                seq_printf(seq, ",nocompress_extension=%s",
1815                        F2FS_OPTION(sbi).noextensions[i]);
1816        }
1817
1818        if (F2FS_OPTION(sbi).compress_chksum)
1819                seq_puts(seq, ",compress_chksum");
1820
1821        if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_FS)
1822                seq_printf(seq, ",compress_mode=%s", "fs");
1823        else if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_USER)
1824                seq_printf(seq, ",compress_mode=%s", "user");
1825
1826        if (test_opt(sbi, COMPRESS_CACHE))
1827                seq_puts(seq, ",compress_cache");
1828}
1829#endif
1830
1831static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1832{
1833        struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1834
1835        if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1836                seq_printf(seq, ",background_gc=%s", "sync");
1837        else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1838                seq_printf(seq, ",background_gc=%s", "on");
1839        else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1840                seq_printf(seq, ",background_gc=%s", "off");
1841
1842        if (test_opt(sbi, GC_MERGE))
1843                seq_puts(seq, ",gc_merge");
1844
1845        if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1846                seq_puts(seq, ",disable_roll_forward");
1847        if (test_opt(sbi, NORECOVERY))
1848                seq_puts(seq, ",norecovery");
1849        if (test_opt(sbi, DISCARD))
1850                seq_puts(seq, ",discard");
1851        else
1852                seq_puts(seq, ",nodiscard");
1853        if (test_opt(sbi, NOHEAP))
1854                seq_puts(seq, ",no_heap");
1855        else
1856                seq_puts(seq, ",heap");
1857#ifdef CONFIG_F2FS_FS_XATTR
1858        if (test_opt(sbi, XATTR_USER))
1859                seq_puts(seq, ",user_xattr");
1860        else
1861                seq_puts(seq, ",nouser_xattr");
1862        if (test_opt(sbi, INLINE_XATTR))
1863                seq_puts(seq, ",inline_xattr");
1864        else
1865                seq_puts(seq, ",noinline_xattr");
1866        if (test_opt(sbi, INLINE_XATTR_SIZE))
1867                seq_printf(seq, ",inline_xattr_size=%u",
1868                                        F2FS_OPTION(sbi).inline_xattr_size);
1869#endif
1870#ifdef CONFIG_F2FS_FS_POSIX_ACL
1871        if (test_opt(sbi, POSIX_ACL))
1872                seq_puts(seq, ",acl");
1873        else
1874                seq_puts(seq, ",noacl");
1875#endif
1876        if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
1877                seq_puts(seq, ",disable_ext_identify");
1878        if (test_opt(sbi, INLINE_DATA))
1879                seq_puts(seq, ",inline_data");
1880        else
1881                seq_puts(seq, ",noinline_data");
1882        if (test_opt(sbi, INLINE_DENTRY))
1883                seq_puts(seq, ",inline_dentry");
1884        else
1885                seq_puts(seq, ",noinline_dentry");
1886        if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
1887                seq_puts(seq, ",flush_merge");
1888        if (test_opt(sbi, NOBARRIER))
1889                seq_puts(seq, ",nobarrier");
1890        if (test_opt(sbi, FASTBOOT))
1891                seq_puts(seq, ",fastboot");
1892        if (test_opt(sbi, EXTENT_CACHE))
1893                seq_puts(seq, ",extent_cache");
1894        else
1895                seq_puts(seq, ",noextent_cache");
1896        if (test_opt(sbi, DATA_FLUSH))
1897                seq_puts(seq, ",data_flush");
1898
1899        seq_puts(seq, ",mode=");
1900        if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
1901                seq_puts(seq, "adaptive");
1902        else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
1903                seq_puts(seq, "lfs");
1904        else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG)
1905                seq_puts(seq, "fragment:segment");
1906        else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK)
1907                seq_puts(seq, "fragment:block");
1908        seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
1909        if (test_opt(sbi, RESERVE_ROOT))
1910                seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
1911                                F2FS_OPTION(sbi).root_reserved_blocks,
1912                                from_kuid_munged(&init_user_ns,
1913                                        F2FS_OPTION(sbi).s_resuid),
1914                                from_kgid_munged(&init_user_ns,
1915                                        F2FS_OPTION(sbi).s_resgid));
1916        if (F2FS_IO_SIZE_BITS(sbi))
1917                seq_printf(seq, ",io_bits=%u",
1918                                F2FS_OPTION(sbi).write_io_size_bits);
1919#ifdef CONFIG_F2FS_FAULT_INJECTION
1920        if (test_opt(sbi, FAULT_INJECTION)) {
1921                seq_printf(seq, ",fault_injection=%u",
1922                                F2FS_OPTION(sbi).fault_info.inject_rate);
1923                seq_printf(seq, ",fault_type=%u",
1924                                F2FS_OPTION(sbi).fault_info.inject_type);
1925        }
1926#endif
1927#ifdef CONFIG_QUOTA
1928        if (test_opt(sbi, QUOTA))
1929                seq_puts(seq, ",quota");
1930        if (test_opt(sbi, USRQUOTA))
1931                seq_puts(seq, ",usrquota");
1932        if (test_opt(sbi, GRPQUOTA))
1933                seq_puts(seq, ",grpquota");
1934        if (test_opt(sbi, PRJQUOTA))
1935                seq_puts(seq, ",prjquota");
1936#endif
1937        f2fs_show_quota_options(seq, sbi->sb);
1938        if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_USER)
1939                seq_printf(seq, ",whint_mode=%s", "user-based");
1940        else if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_FS)
1941                seq_printf(seq, ",whint_mode=%s", "fs-based");
1942
1943        fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
1944
1945        if (sbi->sb->s_flags & SB_INLINECRYPT)
1946                seq_puts(seq, ",inlinecrypt");
1947
1948        if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
1949                seq_printf(seq, ",alloc_mode=%s", "default");
1950        else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
1951                seq_printf(seq, ",alloc_mode=%s", "reuse");
1952
1953        if (test_opt(sbi, DISABLE_CHECKPOINT))
1954                seq_printf(seq, ",checkpoint=disable:%u",
1955                                F2FS_OPTION(sbi).unusable_cap);
1956        if (test_opt(sbi, MERGE_CHECKPOINT))
1957                seq_puts(seq, ",checkpoint_merge");
1958        else
1959                seq_puts(seq, ",nocheckpoint_merge");
1960        if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
1961                seq_printf(seq, ",fsync_mode=%s", "posix");
1962        else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
1963                seq_printf(seq, ",fsync_mode=%s", "strict");
1964        else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
1965                seq_printf(seq, ",fsync_mode=%s", "nobarrier");
1966
1967#ifdef CONFIG_F2FS_FS_COMPRESSION
1968        f2fs_show_compress_options(seq, sbi->sb);
1969#endif
1970
1971        if (test_opt(sbi, ATGC))
1972                seq_puts(seq, ",atgc");
1973
1974        if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK)
1975                seq_printf(seq, ",discard_unit=%s", "block");
1976        else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SEGMENT)
1977                seq_printf(seq, ",discard_unit=%s", "segment");
1978        else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SECTION)
1979                seq_printf(seq, ",discard_unit=%s", "section");
1980
1981        return 0;
1982}
1983
1984static void default_options(struct f2fs_sb_info *sbi)
1985{
1986        /* init some FS parameters */
1987        if (f2fs_sb_has_readonly(sbi))
1988                F2FS_OPTION(sbi).active_logs = NR_CURSEG_RO_TYPE;
1989        else
1990                F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
1991
1992        F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
1993        F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1994        F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
1995        F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
1996        F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
1997        F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
1998        F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
1999        F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
2000        F2FS_OPTION(sbi).compress_ext_cnt = 0;
2001        F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
2002        F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
2003
2004        sbi->sb->s_flags &= ~SB_INLINECRYPT;
2005
2006        set_opt(sbi, INLINE_XATTR);
2007        set_opt(sbi, INLINE_DATA);
2008        set_opt(sbi, INLINE_DENTRY);
2009        set_opt(sbi, EXTENT_CACHE);
2010        set_opt(sbi, NOHEAP);
2011        clear_opt(sbi, DISABLE_CHECKPOINT);
2012        set_opt(sbi, MERGE_CHECKPOINT);
2013        F2FS_OPTION(sbi).unusable_cap = 0;
2014        sbi->sb->s_flags |= SB_LAZYTIME;
2015        set_opt(sbi, FLUSH_MERGE);
2016        if (f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi))
2017                set_opt(sbi, DISCARD);
2018        if (f2fs_sb_has_blkzoned(sbi)) {
2019                F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
2020                F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_SECTION;
2021        } else {
2022                F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
2023                F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_BLOCK;
2024        }
2025
2026#ifdef CONFIG_F2FS_FS_XATTR
2027        set_opt(sbi, XATTR_USER);
2028#endif
2029#ifdef CONFIG_F2FS_FS_POSIX_ACL
2030        set_opt(sbi, POSIX_ACL);
2031#endif
2032
2033        f2fs_build_fault_attr(sbi, 0, 0);
2034}
2035
2036#ifdef CONFIG_QUOTA
2037static int f2fs_enable_quotas(struct super_block *sb);
2038#endif
2039
2040static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
2041{
2042        unsigned int s_flags = sbi->sb->s_flags;
2043        struct cp_control cpc;
2044        int err = 0;
2045        int ret;
2046        block_t unusable;
2047
2048        if (s_flags & SB_RDONLY) {
2049                f2fs_err(sbi, "checkpoint=disable on readonly fs");
2050                return -EINVAL;
2051        }
2052        sbi->sb->s_flags |= SB_ACTIVE;
2053
2054        f2fs_update_time(sbi, DISABLE_TIME);
2055
2056        while (!f2fs_time_over(sbi, DISABLE_TIME)) {
2057                down_write(&sbi->gc_lock);
2058                err = f2fs_gc(sbi, true, false, false, NULL_SEGNO);
2059                if (err == -ENODATA) {
2060                        err = 0;
2061                        break;
2062                }
2063                if (err && err != -EAGAIN)
2064                        break;
2065        }
2066
2067        ret = sync_filesystem(sbi->sb);
2068        if (ret || err) {
2069                err = ret ? ret : err;
2070                goto restore_flag;
2071        }
2072
2073        unusable = f2fs_get_unusable_blocks(sbi);
2074        if (f2fs_disable_cp_again(sbi, unusable)) {
2075                err = -EAGAIN;
2076                goto restore_flag;
2077        }
2078
2079        down_write(&sbi->gc_lock);
2080        cpc.reason = CP_PAUSE;
2081        set_sbi_flag(sbi, SBI_CP_DISABLED);
2082        err = f2fs_write_checkpoint(sbi, &cpc);
2083        if (err)
2084                goto out_unlock;
2085
2086        spin_lock(&sbi->stat_lock);
2087        sbi->unusable_block_count = unusable;
2088        spin_unlock(&sbi->stat_lock);
2089
2090out_unlock:
2091        up_write(&sbi->gc_lock);
2092restore_flag:
2093        sbi->sb->s_flags = s_flags;     /* Restore SB_RDONLY status */
2094        return err;
2095}
2096
2097static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
2098{
2099        int retry = DEFAULT_RETRY_IO_COUNT;
2100
2101        /* we should flush all the data to keep data consistency */
2102        do {
2103                sync_inodes_sb(sbi->sb);
2104                cond_resched();
2105                congestion_wait(BLK_RW_ASYNC, DEFAULT_IO_TIMEOUT);
2106        } while (get_pages(sbi, F2FS_DIRTY_DATA) && retry--);
2107
2108        if (unlikely(retry < 0))
2109                f2fs_warn(sbi, "checkpoint=enable has some unwritten data.");
2110
2111        down_write(&sbi->gc_lock);
2112        f2fs_dirty_to_prefree(sbi);
2113
2114        clear_sbi_flag(sbi, SBI_CP_DISABLED);
2115        set_sbi_flag(sbi, SBI_IS_DIRTY);
2116        up_write(&sbi->gc_lock);
2117
2118        f2fs_sync_fs(sbi->sb, 1);
2119}
2120
2121static int f2fs_remount(struct super_block *sb, int *flags, char *data)
2122{
2123        struct f2fs_sb_info *sbi = F2FS_SB(sb);
2124        struct f2fs_mount_info org_mount_opt;
2125        unsigned long old_sb_flags;
2126        int err;
2127        bool need_restart_gc = false, need_stop_gc = false;
2128        bool need_restart_ckpt = false, need_stop_ckpt = false;
2129        bool need_restart_flush = false, need_stop_flush = false;
2130        bool need_restart_discard = false, need_stop_discard = false;
2131        bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
2132        bool enable_checkpoint = !test_opt(sbi, DISABLE_CHECKPOINT);
2133        bool no_io_align = !F2FS_IO_ALIGNED(sbi);
2134        bool no_atgc = !test_opt(sbi, ATGC);
2135        bool no_discard = !test_opt(sbi, DISCARD);
2136        bool no_compress_cache = !test_opt(sbi, COMPRESS_CACHE);
2137        bool block_unit_discard = f2fs_block_unit_discard(sbi);
2138        struct discard_cmd_control *dcc;
2139#ifdef CONFIG_QUOTA
2140        int i, j;
2141#endif
2142
2143        /*
2144         * Save the old mount options in case we
2145         * need to restore them.
2146         */
2147        org_mount_opt = sbi->mount_opt;
2148        old_sb_flags = sb->s_flags;
2149
2150#ifdef CONFIG_QUOTA
2151        org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
2152        for (i = 0; i < MAXQUOTAS; i++) {
2153                if (F2FS_OPTION(sbi).s_qf_names[i]) {
2154                        org_mount_opt.s_qf_names[i] =
2155                                kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
2156                                GFP_KERNEL);
2157                        if (!org_mount_opt.s_qf_names[i]) {
2158                                for (j = 0; j < i; j++)
2159                                        kfree(org_mount_opt.s_qf_names[j]);
2160                                return -ENOMEM;
2161                        }
2162                } else {
2163                        org_mount_opt.s_qf_names[i] = NULL;
2164                }
2165        }
2166#endif
2167
2168        /* recover superblocks we couldn't write due to previous RO mount */
2169        if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
2170                err = f2fs_commit_super(sbi, false);
2171                f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
2172                          err);
2173                if (!err)
2174                        clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2175        }
2176
2177        default_options(sbi);
2178
2179        /* parse mount options */
2180        err = parse_options(sb, data, true);
2181        if (err)
2182                goto restore_opts;
2183
2184        /*
2185         * Previous and new state of filesystem is RO,
2186         * so skip checking GC and FLUSH_MERGE conditions.
2187         */
2188        if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
2189                goto skip;
2190
2191        if (f2fs_sb_has_readonly(sbi) && !(*flags & SB_RDONLY)) {
2192                err = -EROFS;
2193                goto restore_opts;
2194        }
2195
2196#ifdef CONFIG_QUOTA
2197        if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
2198                err = dquot_suspend(sb, -1);
2199                if (err < 0)
2200                        goto restore_opts;
2201        } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
2202                /* dquot_resume needs RW */
2203                sb->s_flags &= ~SB_RDONLY;
2204                if (sb_any_quota_suspended(sb)) {
2205                        dquot_resume(sb, -1);
2206                } else if (f2fs_sb_has_quota_ino(sbi)) {
2207                        err = f2fs_enable_quotas(sb);
2208                        if (err)
2209                                goto restore_opts;
2210                }
2211        }
2212#endif
2213        /* disallow enable atgc dynamically */
2214        if (no_atgc == !!test_opt(sbi, ATGC)) {
2215                err = -EINVAL;
2216                f2fs_warn(sbi, "switch atgc option is not allowed");
2217                goto restore_opts;
2218        }
2219
2220        /* disallow enable/disable extent_cache dynamically */
2221        if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
2222                err = -EINVAL;
2223                f2fs_warn(sbi, "switch extent_cache option is not allowed");
2224                goto restore_opts;
2225        }
2226
2227        if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
2228                err = -EINVAL;
2229                f2fs_warn(sbi, "switch io_bits option is not allowed");
2230                goto restore_opts;
2231        }
2232
2233        if (no_compress_cache == !!test_opt(sbi, COMPRESS_CACHE)) {
2234                err = -EINVAL;
2235                f2fs_warn(sbi, "switch compress_cache option is not allowed");
2236                goto restore_opts;
2237        }
2238
2239        if (block_unit_discard != f2fs_block_unit_discard(sbi)) {
2240                err = -EINVAL;
2241                f2fs_warn(sbi, "switch discard_unit option is not allowed");
2242                goto restore_opts;
2243        }
2244
2245        if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
2246                err = -EINVAL;
2247                f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
2248                goto restore_opts;
2249        }
2250
2251        /*
2252         * We stop the GC thread if FS is mounted as RO
2253         * or if background_gc = off is passed in mount
2254         * option. Also sync the filesystem.
2255         */
2256        if ((*flags & SB_RDONLY) ||
2257                        (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF &&
2258                        !test_opt(sbi, GC_MERGE))) {
2259                if (sbi->gc_thread) {
2260                        f2fs_stop_gc_thread(sbi);
2261                        need_restart_gc = true;
2262                }
2263        } else if (!sbi->gc_thread) {
2264                err = f2fs_start_gc_thread(sbi);
2265                if (err)
2266                        goto restore_opts;
2267                need_stop_gc = true;
2268        }
2269
2270        if (*flags & SB_RDONLY ||
2271                F2FS_OPTION(sbi).whint_mode != org_mount_opt.whint_mode) {
2272                sync_inodes_sb(sb);
2273
2274                set_sbi_flag(sbi, SBI_IS_DIRTY);
2275                set_sbi_flag(sbi, SBI_IS_CLOSE);
2276                f2fs_sync_fs(sb, 1);
2277                clear_sbi_flag(sbi, SBI_IS_CLOSE);
2278        }
2279
2280        if ((*flags & SB_RDONLY) || test_opt(sbi, DISABLE_CHECKPOINT) ||
2281                        !test_opt(sbi, MERGE_CHECKPOINT)) {
2282                f2fs_stop_ckpt_thread(sbi);
2283                need_restart_ckpt = true;
2284        } else {
2285                err = f2fs_start_ckpt_thread(sbi);
2286                if (err) {
2287                        f2fs_err(sbi,
2288                            "Failed to start F2FS issue_checkpoint_thread (%d)",
2289                            err);
2290                        goto restore_gc;
2291                }
2292                need_stop_ckpt = true;
2293        }
2294
2295        /*
2296         * We stop issue flush thread if FS is mounted as RO
2297         * or if flush_merge is not passed in mount option.
2298         */
2299        if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2300                clear_opt(sbi, FLUSH_MERGE);
2301                f2fs_destroy_flush_cmd_control(sbi, false);
2302                need_restart_flush = true;
2303        } else {
2304                err = f2fs_create_flush_cmd_control(sbi);
2305                if (err)
2306                        goto restore_ckpt;
2307                need_stop_flush = true;
2308        }
2309
2310        if (no_discard == !!test_opt(sbi, DISCARD)) {
2311                if (test_opt(sbi, DISCARD)) {
2312                        err = f2fs_start_discard_thread(sbi);
2313                        if (err)
2314                                goto restore_flush;
2315                        need_stop_discard = true;
2316                } else {
2317                        dcc = SM_I(sbi)->dcc_info;
2318                        f2fs_stop_discard_thread(sbi);
2319                        if (atomic_read(&dcc->discard_cmd_cnt))
2320                                f2fs_issue_discard_timeout(sbi);
2321                        need_restart_discard = true;
2322                }
2323        }
2324
2325        if (enable_checkpoint == !!test_opt(sbi, DISABLE_CHECKPOINT)) {
2326                if (test_opt(sbi, DISABLE_CHECKPOINT)) {
2327                        err = f2fs_disable_checkpoint(sbi);
2328                        if (err)
2329                                goto restore_discard;
2330                } else {
2331                        f2fs_enable_checkpoint(sbi);
2332                }
2333        }
2334
2335skip:
2336#ifdef CONFIG_QUOTA
2337        /* Release old quota file names */
2338        for (i = 0; i < MAXQUOTAS; i++)
2339                kfree(org_mount_opt.s_qf_names[i]);
2340#endif
2341        /* Update the POSIXACL Flag */
2342        sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
2343                (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
2344
2345        limit_reserve_root(sbi);
2346        adjust_unusable_cap_perc(sbi);
2347        *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
2348        return 0;
2349restore_discard:
2350        if (need_restart_discard) {
2351                if (f2fs_start_discard_thread(sbi))
2352                        f2fs_warn(sbi, "discard has been stopped");
2353        } else if (need_stop_discard) {
2354                f2fs_stop_discard_thread(sbi);
2355        }
2356restore_flush:
2357        if (need_restart_flush) {
2358                if (f2fs_create_flush_cmd_control(sbi))
2359                        f2fs_warn(sbi, "background flush thread has stopped");
2360        } else if (need_stop_flush) {
2361                clear_opt(sbi, FLUSH_MERGE);
2362                f2fs_destroy_flush_cmd_control(sbi, false);
2363        }
2364restore_ckpt:
2365        if (need_restart_ckpt) {
2366                if (f2fs_start_ckpt_thread(sbi))
2367                        f2fs_warn(sbi, "background ckpt thread has stopped");
2368        } else if (need_stop_ckpt) {
2369                f2fs_stop_ckpt_thread(sbi);
2370        }
2371restore_gc:
2372        if (need_restart_gc) {
2373                if (f2fs_start_gc_thread(sbi))
2374                        f2fs_warn(sbi, "background gc thread has stopped");
2375        } else if (need_stop_gc) {
2376                f2fs_stop_gc_thread(sbi);
2377        }
2378restore_opts:
2379#ifdef CONFIG_QUOTA
2380        F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
2381        for (i = 0; i < MAXQUOTAS; i++) {
2382                kfree(F2FS_OPTION(sbi).s_qf_names[i]);
2383                F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
2384        }
2385#endif
2386        sbi->mount_opt = org_mount_opt;
2387        sb->s_flags = old_sb_flags;
2388        return err;
2389}
2390
2391#ifdef CONFIG_QUOTA
2392/* Read data from quotafile */
2393static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
2394                               size_t len, loff_t off)
2395{
2396        struct inode *inode = sb_dqopt(sb)->files[type];
2397        struct address_space *mapping = inode->i_mapping;
2398        block_t blkidx = F2FS_BYTES_TO_BLK(off);
2399        int offset = off & (sb->s_blocksize - 1);
2400        int tocopy;
2401        size_t toread;
2402        loff_t i_size = i_size_read(inode);
2403        struct page *page;
2404        char *kaddr;
2405
2406        if (off > i_size)
2407                return 0;
2408
2409        if (off + len > i_size)
2410                len = i_size - off;
2411        toread = len;
2412        while (toread > 0) {
2413                tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
2414repeat:
2415                page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
2416                if (IS_ERR(page)) {
2417                        if (PTR_ERR(page) == -ENOMEM) {
2418                                congestion_wait(BLK_RW_ASYNC,
2419                                                DEFAULT_IO_TIMEOUT);
2420                                goto repeat;
2421                        }
2422                        set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2423                        return PTR_ERR(page);
2424                }
2425
2426                lock_page(page);
2427
2428                if (unlikely(page->mapping != mapping)) {
2429                        f2fs_put_page(page, 1);
2430                        goto repeat;
2431                }
2432                if (unlikely(!PageUptodate(page))) {
2433                        f2fs_put_page(page, 1);
2434                        set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2435                        return -EIO;
2436                }
2437
2438                kaddr = kmap_atomic(page);
2439                memcpy(data, kaddr + offset, tocopy);
2440                kunmap_atomic(kaddr);
2441                f2fs_put_page(page, 1);
2442
2443                offset = 0;
2444                toread -= tocopy;
2445                data += tocopy;
2446                blkidx++;
2447        }
2448        return len;
2449}
2450
2451/* Write to quotafile */
2452static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2453                                const char *data, size_t len, loff_t off)
2454{
2455        struct inode *inode = sb_dqopt(sb)->files[type];
2456        struct address_space *mapping = inode->i_mapping;
2457        const struct address_space_operations *a_ops = mapping->a_ops;
2458        int offset = off & (sb->s_blocksize - 1);
2459        size_t towrite = len;
2460        struct page *page;
2461        void *fsdata = NULL;
2462        char *kaddr;
2463        int err = 0;
2464        int tocopy;
2465
2466        while (towrite > 0) {
2467                tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2468                                                                towrite);
2469retry:
2470                err = a_ops->write_begin(NULL, mapping, off, tocopy, 0,
2471                                                        &page, &fsdata);
2472                if (unlikely(err)) {
2473                        if (err == -ENOMEM) {
2474                                congestion_wait(BLK_RW_ASYNC,
2475                                                DEFAULT_IO_TIMEOUT);
2476                                goto retry;
2477                        }
2478                        set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2479                        break;
2480                }
2481
2482                kaddr = kmap_atomic(page);
2483                memcpy(kaddr + offset, data, tocopy);
2484                kunmap_atomic(kaddr);
2485                flush_dcache_page(page);
2486
2487                a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2488                                                page, fsdata);
2489                offset = 0;
2490                towrite -= tocopy;
2491                off += tocopy;
2492                data += tocopy;
2493                cond_resched();
2494        }
2495
2496        if (len == towrite)
2497                return err;
2498        inode->i_mtime = inode->i_ctime = current_time(inode);
2499        f2fs_mark_inode_dirty_sync(inode, false);
2500        return len - towrite;
2501}
2502
2503int f2fs_dquot_initialize(struct inode *inode)
2504{
2505        if (time_to_inject(F2FS_I_SB(inode), FAULT_DQUOT_INIT)) {
2506                f2fs_show_injection_info(F2FS_I_SB(inode), FAULT_DQUOT_INIT);
2507                return -ESRCH;
2508        }
2509
2510        return dquot_initialize(inode);
2511}
2512
2513static struct dquot **f2fs_get_dquots(struct inode *inode)
2514{
2515        return F2FS_I(inode)->i_dquot;
2516}
2517
2518static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2519{
2520        return &F2FS_I(inode)->i_reserved_quota;
2521}
2522
2523static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2524{
2525        if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2526                f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2527                return 0;
2528        }
2529
2530        return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2531                                        F2FS_OPTION(sbi).s_jquota_fmt, type);
2532}
2533
2534int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2535{
2536        int enabled = 0;
2537        int i, err;
2538
2539        if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2540                err = f2fs_enable_quotas(sbi->sb);
2541                if (err) {
2542                        f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2543                        return 0;
2544                }
2545                return 1;
2546        }
2547
2548        for (i = 0; i < MAXQUOTAS; i++) {
2549                if (F2FS_OPTION(sbi).s_qf_names[i]) {
2550                        err = f2fs_quota_on_mount(sbi, i);
2551                        if (!err) {
2552                                enabled = 1;
2553                                continue;
2554                        }
2555                        f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2556                                 err, i);
2557                }
2558        }
2559        return enabled;
2560}
2561
2562static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2563                             unsigned int flags)
2564{
2565        struct inode *qf_inode;
2566        unsigned long qf_inum;
2567        int err;
2568
2569        BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2570
2571        qf_inum = f2fs_qf_ino(sb, type);
2572        if (!qf_inum)
2573                return -EPERM;
2574
2575        qf_inode = f2fs_iget(sb, qf_inum);
2576        if (IS_ERR(qf_inode)) {
2577                f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2578                return PTR_ERR(qf_inode);
2579        }
2580
2581        /* Don't account quota for quota files to avoid recursion */
2582        qf_inode->i_flags |= S_NOQUOTA;
2583        err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2584        iput(qf_inode);
2585        return err;
2586}
2587
2588static int f2fs_enable_quotas(struct super_block *sb)
2589{
2590        struct f2fs_sb_info *sbi = F2FS_SB(sb);
2591        int type, err = 0;
2592        unsigned long qf_inum;
2593        bool quota_mopt[MAXQUOTAS] = {
2594                test_opt(sbi, USRQUOTA),
2595                test_opt(sbi, GRPQUOTA),
2596                test_opt(sbi, PRJQUOTA),
2597        };
2598
2599        if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2600                f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2601                return 0;
2602        }
2603
2604        sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2605
2606        for (type = 0; type < MAXQUOTAS; type++) {
2607                qf_inum = f2fs_qf_ino(sb, type);
2608                if (qf_inum) {
2609                        err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2610                                DQUOT_USAGE_ENABLED |
2611                                (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2612                        if (err) {
2613                                f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2614                                         type, err);
2615                                for (type--; type >= 0; type--)
2616                                        dquot_quota_off(sb, type);
2617                                set_sbi_flag(F2FS_SB(sb),
2618                                                SBI_QUOTA_NEED_REPAIR);
2619                                return err;
2620                        }
2621                }
2622        }
2623        return 0;
2624}
2625
2626static int f2fs_quota_sync_file(struct f2fs_sb_info *sbi, int type)
2627{
2628        struct quota_info *dqopt = sb_dqopt(sbi->sb);
2629        struct address_space *mapping = dqopt->files[type]->i_mapping;
2630        int ret = 0;
2631
2632        ret = dquot_writeback_dquots(sbi->sb, type);
2633        if (ret)
2634                goto out;
2635
2636        ret = filemap_fdatawrite(mapping);
2637        if (ret)
2638                goto out;
2639
2640        /* if we are using journalled quota */
2641        if (is_journalled_quota(sbi))
2642                goto out;
2643
2644        ret = filemap_fdatawait(mapping);
2645
2646        truncate_inode_pages(&dqopt->files[type]->i_data, 0);
2647out:
2648        if (ret)
2649                set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2650        return ret;
2651}
2652
2653int f2fs_quota_sync(struct super_block *sb, int type)
2654{
2655        struct f2fs_sb_info *sbi = F2FS_SB(sb);
2656        struct quota_info *dqopt = sb_dqopt(sb);
2657        int cnt;
2658        int ret;
2659
2660        /*
2661         * Now when everything is written we can discard the pagecache so
2662         * that userspace sees the changes.
2663         */
2664        for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2665
2666                if (type != -1 && cnt != type)
2667                        continue;
2668
2669                if (!sb_has_quota_active(sb, type))
2670                        return 0;
2671
2672                inode_lock(dqopt->files[cnt]);
2673
2674                /*
2675                 * do_quotactl
2676                 *  f2fs_quota_sync
2677                 *  down_read(quota_sem)
2678                 *  dquot_writeback_dquots()
2679                 *  f2fs_dquot_commit
2680                 *                            block_operation
2681                 *                            down_read(quota_sem)
2682                 */
2683                f2fs_lock_op(sbi);
2684                down_read(&sbi->quota_sem);
2685
2686                ret = f2fs_quota_sync_file(sbi, cnt);
2687
2688                up_read(&sbi->quota_sem);
2689                f2fs_unlock_op(sbi);
2690
2691                inode_unlock(dqopt->files[cnt]);
2692
2693                if (ret)
2694                        break;
2695        }
2696        return ret;
2697}
2698
2699static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2700                                                        const struct path *path)
2701{
2702        struct inode *inode;
2703        int err;
2704
2705        /* if quota sysfile exists, deny enabling quota with specific file */
2706        if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2707                f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2708                return -EBUSY;
2709        }
2710
2711        err = f2fs_quota_sync(sb, type);
2712        if (err)
2713                return err;
2714
2715        err = dquot_quota_on(sb, type, format_id, path);
2716        if (err)
2717                return err;
2718
2719        inode = d_inode(path->dentry);
2720
2721        inode_lock(inode);
2722        F2FS_I(inode)->i_flags |= F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL;
2723        f2fs_set_inode_flags(inode);
2724        inode_unlock(inode);
2725        f2fs_mark_inode_dirty_sync(inode, false);
2726
2727        return 0;
2728}
2729
2730static int __f2fs_quota_off(struct super_block *sb, int type)
2731{
2732        struct inode *inode = sb_dqopt(sb)->files[type];
2733        int err;
2734
2735        if (!inode || !igrab(inode))
2736                return dquot_quota_off(sb, type);
2737
2738        err = f2fs_quota_sync(sb, type);
2739        if (err)
2740                goto out_put;
2741
2742        err = dquot_quota_off(sb, type);
2743        if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
2744                goto out_put;
2745
2746        inode_lock(inode);
2747        F2FS_I(inode)->i_flags &= ~(F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL);
2748        f2fs_set_inode_flags(inode);
2749        inode_unlock(inode);
2750        f2fs_mark_inode_dirty_sync(inode, false);
2751out_put:
2752        iput(inode);
2753        return err;
2754}
2755
2756static int f2fs_quota_off(struct super_block *sb, int type)
2757{
2758        struct f2fs_sb_info *sbi = F2FS_SB(sb);
2759        int err;
2760
2761        err = __f2fs_quota_off(sb, type);
2762
2763        /*
2764         * quotactl can shutdown journalled quota, result in inconsistence
2765         * between quota record and fs data by following updates, tag the
2766         * flag to let fsck be aware of it.
2767         */
2768        if (is_journalled_quota(sbi))
2769                set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2770        return err;
2771}
2772
2773void f2fs_quota_off_umount(struct super_block *sb)
2774{
2775        int type;
2776        int err;
2777
2778        for (type = 0; type < MAXQUOTAS; type++) {
2779                err = __f2fs_quota_off(sb, type);
2780                if (err) {
2781                        int ret = dquot_quota_off(sb, type);
2782
2783                        f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
2784                                 type, err, ret);
2785                        set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2786                }
2787        }
2788        /*
2789         * In case of checkpoint=disable, we must flush quota blocks.
2790         * This can cause NULL exception for node_inode in end_io, since
2791         * put_super already dropped it.
2792         */
2793        sync_filesystem(sb);
2794}
2795
2796static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
2797{
2798        struct quota_info *dqopt = sb_dqopt(sb);
2799        int type;
2800
2801        for (type = 0; type < MAXQUOTAS; type++) {
2802                if (!dqopt->files[type])
2803                        continue;
2804                f2fs_inode_synced(dqopt->files[type]);
2805        }
2806}
2807
2808static int f2fs_dquot_commit(struct dquot *dquot)
2809{
2810        struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2811        int ret;
2812
2813        down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
2814        ret = dquot_commit(dquot);
2815        if (ret < 0)
2816                set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2817        up_read(&sbi->quota_sem);
2818        return ret;
2819}
2820
2821static int f2fs_dquot_acquire(struct dquot *dquot)
2822{
2823        struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2824        int ret;
2825
2826        down_read(&sbi->quota_sem);
2827        ret = dquot_acquire(dquot);
2828        if (ret < 0)
2829                set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2830        up_read(&sbi->quota_sem);
2831        return ret;
2832}
2833
2834static int f2fs_dquot_release(struct dquot *dquot)
2835{
2836        struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2837        int ret = dquot_release(dquot);
2838
2839        if (ret < 0)
2840                set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2841        return ret;
2842}
2843
2844static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
2845{
2846        struct super_block *sb = dquot->dq_sb;
2847        struct f2fs_sb_info *sbi = F2FS_SB(sb);
2848        int ret = dquot_mark_dquot_dirty(dquot);
2849
2850        /* if we are using journalled quota */
2851        if (is_journalled_quota(sbi))
2852                set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
2853
2854        return ret;
2855}
2856
2857static int f2fs_dquot_commit_info(struct super_block *sb, int type)
2858{
2859        struct f2fs_sb_info *sbi = F2FS_SB(sb);
2860        int ret = dquot_commit_info(sb, type);
2861
2862        if (ret < 0)
2863                set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2864        return ret;
2865}
2866
2867static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
2868{
2869        *projid = F2FS_I(inode)->i_projid;
2870        return 0;
2871}
2872
2873static const struct dquot_operations f2fs_quota_operations = {
2874        .get_reserved_space = f2fs_get_reserved_space,
2875        .write_dquot    = f2fs_dquot_commit,
2876        .acquire_dquot  = f2fs_dquot_acquire,
2877        .release_dquot  = f2fs_dquot_release,
2878        .mark_dirty     = f2fs_dquot_mark_dquot_dirty,
2879        .write_info     = f2fs_dquot_commit_info,
2880        .alloc_dquot    = dquot_alloc,
2881        .destroy_dquot  = dquot_destroy,
2882        .get_projid     = f2fs_get_projid,
2883        .get_next_id    = dquot_get_next_id,
2884};
2885
2886static const struct quotactl_ops f2fs_quotactl_ops = {
2887        .quota_on       = f2fs_quota_on,
2888        .quota_off      = f2fs_quota_off,
2889        .quota_sync     = f2fs_quota_sync,
2890        .get_state      = dquot_get_state,
2891        .set_info       = dquot_set_dqinfo,
2892        .get_dqblk      = dquot_get_dqblk,
2893        .set_dqblk      = dquot_set_dqblk,
2894        .get_nextdqblk  = dquot_get_next_dqblk,
2895};
2896#else
2897int f2fs_dquot_initialize(struct inode *inode)
2898{
2899        return 0;
2900}
2901
2902int f2fs_quota_sync(struct super_block *sb, int type)
2903{
2904        return 0;
2905}
2906
2907void f2fs_quota_off_umount(struct super_block *sb)
2908{
2909}
2910#endif
2911
2912static const struct super_operations f2fs_sops = {
2913        .alloc_inode    = f2fs_alloc_inode,
2914        .free_inode     = f2fs_free_inode,
2915        .drop_inode     = f2fs_drop_inode,
2916        .write_inode    = f2fs_write_inode,
2917        .dirty_inode    = f2fs_dirty_inode,
2918        .show_options   = f2fs_show_options,
2919#ifdef CONFIG_QUOTA
2920        .quota_read     = f2fs_quota_read,
2921        .quota_write    = f2fs_quota_write,
2922        .get_dquots     = f2fs_get_dquots,
2923#endif
2924        .evict_inode    = f2fs_evict_inode,
2925        .put_super      = f2fs_put_super,
2926        .sync_fs        = f2fs_sync_fs,
2927        .freeze_fs      = f2fs_freeze,
2928        .unfreeze_fs    = f2fs_unfreeze,
2929        .statfs         = f2fs_statfs,
2930        .remount_fs     = f2fs_remount,
2931};
2932
2933#ifdef CONFIG_FS_ENCRYPTION
2934static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
2935{
2936        return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2937                                F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2938                                ctx, len, NULL);
2939}
2940
2941static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
2942                                                        void *fs_data)
2943{
2944        struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2945
2946        /*
2947         * Encrypting the root directory is not allowed because fsck
2948         * expects lost+found directory to exist and remain unencrypted
2949         * if LOST_FOUND feature is enabled.
2950         *
2951         */
2952        if (f2fs_sb_has_lost_found(sbi) &&
2953                        inode->i_ino == F2FS_ROOT_INO(sbi))
2954                return -EPERM;
2955
2956        return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2957                                F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2958                                ctx, len, fs_data, XATTR_CREATE);
2959}
2960
2961static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
2962{
2963        return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
2964}
2965
2966static bool f2fs_has_stable_inodes(struct super_block *sb)
2967{
2968        return true;
2969}
2970
2971static void f2fs_get_ino_and_lblk_bits(struct super_block *sb,
2972                                       int *ino_bits_ret, int *lblk_bits_ret)
2973{
2974        *ino_bits_ret = 8 * sizeof(nid_t);
2975        *lblk_bits_ret = 8 * sizeof(block_t);
2976}
2977
2978static int f2fs_get_num_devices(struct super_block *sb)
2979{
2980        struct f2fs_sb_info *sbi = F2FS_SB(sb);
2981
2982        if (f2fs_is_multi_device(sbi))
2983                return sbi->s_ndevs;
2984        return 1;
2985}
2986
2987static void f2fs_get_devices(struct super_block *sb,
2988                             struct request_queue **devs)
2989{
2990        struct f2fs_sb_info *sbi = F2FS_SB(sb);
2991        int i;
2992
2993        for (i = 0; i < sbi->s_ndevs; i++)
2994                devs[i] = bdev_get_queue(FDEV(i).bdev);
2995}
2996
2997static const struct fscrypt_operations f2fs_cryptops = {
2998        .key_prefix             = "f2fs:",
2999        .get_context            = f2fs_get_context,
3000        .set_context            = f2fs_set_context,
3001        .get_dummy_policy       = f2fs_get_dummy_policy,
3002        .empty_dir              = f2fs_empty_dir,
3003        .has_stable_inodes      = f2fs_has_stable_inodes,
3004        .get_ino_and_lblk_bits  = f2fs_get_ino_and_lblk_bits,
3005        .get_num_devices        = f2fs_get_num_devices,
3006        .get_devices            = f2fs_get_devices,
3007};
3008#endif
3009
3010static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
3011                u64 ino, u32 generation)
3012{
3013        struct f2fs_sb_info *sbi = F2FS_SB(sb);
3014        struct inode *inode;
3015
3016        if (f2fs_check_nid_range(sbi, ino))
3017                return ERR_PTR(-ESTALE);
3018
3019        /*
3020         * f2fs_iget isn't quite right if the inode is currently unallocated!
3021         * However f2fs_iget currently does appropriate checks to handle stale
3022         * inodes so everything is OK.
3023         */
3024        inode = f2fs_iget(sb, ino);
3025        if (IS_ERR(inode))
3026                return ERR_CAST(inode);
3027        if (unlikely(generation && inode->i_generation != generation)) {
3028                /* we didn't find the right inode.. */
3029                iput(inode);
3030                return ERR_PTR(-ESTALE);
3031        }
3032        return inode;
3033}
3034
3035static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
3036                int fh_len, int fh_type)
3037{
3038        return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
3039                                    f2fs_nfs_get_inode);
3040}
3041
3042static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
3043                int fh_len, int fh_type)
3044{
3045        return generic_fh_to_parent(sb, fid, fh_len, fh_type,
3046                                    f2fs_nfs_get_inode);
3047}
3048
3049static const struct export_operations f2fs_export_ops = {
3050        .fh_to_dentry = f2fs_fh_to_dentry,
3051        .fh_to_parent = f2fs_fh_to_parent,
3052        .get_parent = f2fs_get_parent,
3053};
3054
3055loff_t max_file_blocks(struct inode *inode)
3056{
3057        loff_t result = 0;
3058        loff_t leaf_count;
3059
3060        /*
3061         * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
3062         * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
3063         * space in inode.i_addr, it will be more safe to reassign
3064         * result as zero.
3065         */
3066
3067        if (inode && f2fs_compressed_file(inode))
3068                leaf_count = ADDRS_PER_BLOCK(inode);
3069        else
3070                leaf_count = DEF_ADDRS_PER_BLOCK;
3071
3072        /* two direct node blocks */
3073        result += (leaf_count * 2);
3074
3075        /* two indirect node blocks */
3076        leaf_count *= NIDS_PER_BLOCK;
3077        result += (leaf_count * 2);
3078
3079        /* one double indirect node block */
3080        leaf_count *= NIDS_PER_BLOCK;
3081        result += leaf_count;
3082
3083        return result;
3084}
3085
3086static int __f2fs_commit_super(struct buffer_head *bh,
3087                        struct f2fs_super_block *super)
3088{
3089        lock_buffer(bh);
3090        if (super)
3091                memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
3092        set_buffer_dirty(bh);
3093        unlock_buffer(bh);
3094
3095        /* it's rare case, we can do fua all the time */
3096        return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
3097}
3098
3099static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
3100                                        struct buffer_head *bh)
3101{
3102        struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3103                                        (bh->b_data + F2FS_SUPER_OFFSET);
3104        struct super_block *sb = sbi->sb;
3105        u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
3106        u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
3107        u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
3108        u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
3109        u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
3110        u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
3111        u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
3112        u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
3113        u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
3114        u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
3115        u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3116        u32 segment_count = le32_to_cpu(raw_super->segment_count);
3117        u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3118        u64 main_end_blkaddr = main_blkaddr +
3119                                (segment_count_main << log_blocks_per_seg);
3120        u64 seg_end_blkaddr = segment0_blkaddr +
3121                                (segment_count << log_blocks_per_seg);
3122
3123        if (segment0_blkaddr != cp_blkaddr) {
3124                f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
3125                          segment0_blkaddr, cp_blkaddr);
3126                return true;
3127        }
3128
3129        if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
3130                                                        sit_blkaddr) {
3131                f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
3132                          cp_blkaddr, sit_blkaddr,
3133                          segment_count_ckpt << log_blocks_per_seg);
3134                return true;
3135        }
3136
3137        if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
3138                                                        nat_blkaddr) {
3139                f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
3140                          sit_blkaddr, nat_blkaddr,
3141                          segment_count_sit << log_blocks_per_seg);
3142                return true;
3143        }
3144
3145        if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
3146                                                        ssa_blkaddr) {
3147                f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
3148                          nat_blkaddr, ssa_blkaddr,
3149                          segment_count_nat << log_blocks_per_seg);
3150                return true;
3151        }
3152
3153        if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
3154                                                        main_blkaddr) {
3155                f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
3156                          ssa_blkaddr, main_blkaddr,
3157                          segment_count_ssa << log_blocks_per_seg);
3158                return true;
3159        }
3160
3161        if (main_end_blkaddr > seg_end_blkaddr) {
3162                f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
3163                          main_blkaddr, seg_end_blkaddr,
3164                          segment_count_main << log_blocks_per_seg);
3165                return true;
3166        } else if (main_end_blkaddr < seg_end_blkaddr) {
3167                int err = 0;
3168                char *res;
3169
3170                /* fix in-memory information all the time */
3171                raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
3172                                segment0_blkaddr) >> log_blocks_per_seg);
3173
3174                if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
3175                        set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3176                        res = "internally";
3177                } else {
3178                        err = __f2fs_commit_super(bh, NULL);
3179                        res = err ? "failed" : "done";
3180                }
3181                f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
3182                          res, main_blkaddr, seg_end_blkaddr,
3183                          segment_count_main << log_blocks_per_seg);
3184                if (err)
3185                        return true;
3186        }
3187        return false;
3188}
3189
3190static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
3191                                struct buffer_head *bh)
3192{
3193        block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
3194        block_t total_sections, blocks_per_seg;
3195        struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3196                                        (bh->b_data + F2FS_SUPER_OFFSET);
3197        size_t crc_offset = 0;
3198        __u32 crc = 0;
3199
3200        if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
3201                f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
3202                          F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
3203                return -EINVAL;
3204        }
3205
3206        /* Check checksum_offset and crc in superblock */
3207        if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
3208                crc_offset = le32_to_cpu(raw_super->checksum_offset);
3209                if (crc_offset !=
3210                        offsetof(struct f2fs_super_block, crc)) {
3211                        f2fs_info(sbi, "Invalid SB checksum offset: %zu",
3212                                  crc_offset);
3213                        return -EFSCORRUPTED;
3214                }
3215                crc = le32_to_cpu(raw_super->crc);
3216                if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
3217                        f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
3218                        return -EFSCORRUPTED;
3219                }
3220        }
3221
3222        /* Currently, support only 4KB block size */
3223        if (le32_to_cpu(raw_super->log_blocksize) != F2FS_BLKSIZE_BITS) {
3224                f2fs_info(sbi, "Invalid log_blocksize (%u), supports only %u",
3225                          le32_to_cpu(raw_super->log_blocksize),
3226                          F2FS_BLKSIZE_BITS);
3227                return -EFSCORRUPTED;
3228        }
3229
3230        /* check log blocks per segment */
3231        if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
3232                f2fs_info(sbi, "Invalid log blocks per segment (%u)",
3233                          le32_to_cpu(raw_super->log_blocks_per_seg));
3234                return -EFSCORRUPTED;
3235        }
3236
3237        /* Currently, support 512/1024/2048/4096 bytes sector size */
3238        if (le32_to_cpu(raw_super->log_sectorsize) >
3239                                F2FS_MAX_LOG_SECTOR_SIZE ||
3240                le32_to_cpu(raw_super->log_sectorsize) <
3241                                F2FS_MIN_LOG_SECTOR_SIZE) {
3242                f2fs_info(sbi, "Invalid log sectorsize (%u)",
3243                          le32_to_cpu(raw_super->log_sectorsize));
3244                return -EFSCORRUPTED;
3245        }
3246        if (le32_to_cpu(raw_super->log_sectors_per_block) +
3247                le32_to_cpu(raw_super->log_sectorsize) !=
3248                        F2FS_MAX_LOG_SECTOR_SIZE) {
3249                f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
3250                          le32_to_cpu(raw_super->log_sectors_per_block),
3251                          le32_to_cpu(raw_super->log_sectorsize));
3252                return -EFSCORRUPTED;
3253        }
3254
3255        segment_count = le32_to_cpu(raw_super->segment_count);
3256        segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3257        segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3258        secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3259        total_sections = le32_to_cpu(raw_super->section_count);
3260
3261        /* blocks_per_seg should be 512, given the above check */
3262        blocks_per_seg = 1 << le32_to_cpu(raw_super->log_blocks_per_seg);
3263
3264        if (segment_count > F2FS_MAX_SEGMENT ||
3265                                segment_count < F2FS_MIN_SEGMENTS) {
3266                f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
3267                return -EFSCORRUPTED;
3268        }
3269
3270        if (total_sections > segment_count_main || total_sections < 1 ||
3271                        segs_per_sec > segment_count || !segs_per_sec) {
3272                f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
3273                          segment_count, total_sections, segs_per_sec);
3274                return -EFSCORRUPTED;
3275        }
3276
3277        if (segment_count_main != total_sections * segs_per_sec) {
3278                f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
3279                          segment_count_main, total_sections, segs_per_sec);
3280                return -EFSCORRUPTED;
3281        }
3282
3283        if ((segment_count / segs_per_sec) < total_sections) {
3284                f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
3285                          segment_count, segs_per_sec, total_sections);
3286                return -EFSCORRUPTED;
3287        }
3288
3289        if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
3290                f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
3291                          segment_count, le64_to_cpu(raw_super->block_count));
3292                return -EFSCORRUPTED;
3293        }
3294
3295        if (RDEV(0).path[0]) {
3296                block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
3297                int i = 1;
3298
3299                while (i < MAX_DEVICES && RDEV(i).path[0]) {
3300                        dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
3301                        i++;
3302                }
3303                if (segment_count != dev_seg_count) {
3304                        f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
3305                                        segment_count, dev_seg_count);
3306                        return -EFSCORRUPTED;
3307                }
3308        } else {
3309                if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
3310                                        !bdev_is_zoned(sbi->sb->s_bdev)) {
3311                        f2fs_info(sbi, "Zoned block device path is missing");
3312                        return -EFSCORRUPTED;
3313                }
3314        }
3315
3316        if (secs_per_zone > total_sections || !secs_per_zone) {
3317                f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
3318                          secs_per_zone, total_sections);
3319                return -EFSCORRUPTED;
3320        }
3321        if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
3322                        raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
3323                        (le32_to_cpu(raw_super->extension_count) +
3324                        raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
3325                f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
3326                          le32_to_cpu(raw_super->extension_count),
3327                          raw_super->hot_ext_count,
3328                          F2FS_MAX_EXTENSION);
3329                return -EFSCORRUPTED;
3330        }
3331
3332        if (le32_to_cpu(raw_super->cp_payload) >=
3333                                (blocks_per_seg - F2FS_CP_PACKS -
3334                                NR_CURSEG_PERSIST_TYPE)) {
3335                f2fs_info(sbi, "Insane cp_payload (%u >= %u)",
3336                          le32_to_cpu(raw_super->cp_payload),
3337                          blocks_per_seg - F2FS_CP_PACKS -
3338                          NR_CURSEG_PERSIST_TYPE);
3339                return -EFSCORRUPTED;
3340        }
3341
3342        /* check reserved ino info */
3343        if (le32_to_cpu(raw_super->node_ino) != 1 ||
3344                le32_to_cpu(raw_super->meta_ino) != 2 ||
3345                le32_to_cpu(raw_super->root_ino) != 3) {
3346                f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
3347                          le32_to_cpu(raw_super->node_ino),
3348                          le32_to_cpu(raw_super->meta_ino),
3349                          le32_to_cpu(raw_super->root_ino));
3350                return -EFSCORRUPTED;
3351        }
3352
3353        /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
3354        if (sanity_check_area_boundary(sbi, bh))
3355                return -EFSCORRUPTED;
3356
3357        return 0;
3358}
3359
3360int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
3361{
3362        unsigned int total, fsmeta;
3363        struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3364        struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
3365        unsigned int ovp_segments, reserved_segments;
3366        unsigned int main_segs, blocks_per_seg;
3367        unsigned int sit_segs, nat_segs;
3368        unsigned int sit_bitmap_size, nat_bitmap_size;
3369        unsigned int log_blocks_per_seg;
3370        unsigned int segment_count_main;
3371        unsigned int cp_pack_start_sum, cp_payload;
3372        block_t user_block_count, valid_user_blocks;
3373        block_t avail_node_count, valid_node_count;
3374        unsigned int nat_blocks, nat_bits_bytes, nat_bits_blocks;
3375        int i, j;
3376
3377        total = le32_to_cpu(raw_super->segment_count);
3378        fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
3379        sit_segs = le32_to_cpu(raw_super->segment_count_sit);
3380        fsmeta += sit_segs;
3381        nat_segs = le32_to_cpu(raw_super->segment_count_nat);
3382        fsmeta += nat_segs;
3383        fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
3384        fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
3385
3386        if (unlikely(fsmeta >= total))
3387                return 1;
3388
3389        ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
3390        reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
3391
3392        if (!f2fs_sb_has_readonly(sbi) &&
3393                        unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
3394                        ovp_segments == 0 || reserved_segments == 0)) {
3395                f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
3396                return 1;
3397        }
3398        user_block_count = le64_to_cpu(ckpt->user_block_count);
3399        segment_count_main = le32_to_cpu(raw_super->segment_count_main) +
3400                        (f2fs_sb_has_readonly(sbi) ? 1 : 0);
3401        log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3402        if (!user_block_count || user_block_count >=
3403                        segment_count_main << log_blocks_per_seg) {
3404                f2fs_err(sbi, "Wrong user_block_count: %u",
3405                         user_block_count);
3406                return 1;
3407        }
3408
3409        valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
3410        if (valid_user_blocks > user_block_count) {
3411                f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
3412                         valid_user_blocks, user_block_count);
3413                return 1;
3414        }
3415
3416        valid_node_count = le32_to_cpu(ckpt->valid_node_count);
3417        avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
3418        if (valid_node_count > avail_node_count) {
3419                f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
3420                         valid_node_count, avail_node_count);
3421                return 1;
3422        }
3423
3424        main_segs = le32_to_cpu(raw_super->segment_count_main);
3425        blocks_per_seg = sbi->blocks_per_seg;
3426
3427        for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3428                if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
3429                        le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
3430                        return 1;
3431
3432                if (f2fs_sb_has_readonly(sbi))
3433                        goto check_data;
3434
3435                for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
3436                        if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3437                                le32_to_cpu(ckpt->cur_node_segno[j])) {
3438                                f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
3439                                         i, j,
3440                                         le32_to_cpu(ckpt->cur_node_segno[i]));
3441                                return 1;
3442                        }
3443                }
3444        }
3445check_data:
3446        for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
3447                if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
3448                        le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
3449                        return 1;
3450
3451                if (f2fs_sb_has_readonly(sbi))
3452                        goto skip_cross;
3453
3454                for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3455                        if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3456                                le32_to_cpu(ckpt->cur_data_segno[j])) {
3457                                f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3458                                         i, j,
3459                                         le32_to_cpu(ckpt->cur_data_segno[i]));
3460                                return 1;
3461                        }
3462                }
3463        }
3464        for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3465                for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3466                        if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3467                                le32_to_cpu(ckpt->cur_data_segno[j])) {
3468                                f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3469                                         i, j,
3470                                         le32_to_cpu(ckpt->cur_node_segno[i]));
3471                                return 1;
3472                        }
3473                }
3474        }
3475skip_cross:
3476        sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3477        nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3478
3479        if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3480                nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3481                f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3482                         sit_bitmap_size, nat_bitmap_size);
3483                return 1;
3484        }
3485
3486        cp_pack_start_sum = __start_sum_addr(sbi);
3487        cp_payload = __cp_payload(sbi);
3488        if (cp_pack_start_sum < cp_payload + 1 ||
3489                cp_pack_start_sum > blocks_per_seg - 1 -
3490                        NR_CURSEG_PERSIST_TYPE) {
3491                f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3492                         cp_pack_start_sum);
3493                return 1;
3494        }
3495
3496        if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3497                le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3498                f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3499                          "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3500                          "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3501                          le32_to_cpu(ckpt->checksum_offset));
3502                return 1;
3503        }
3504
3505        nat_blocks = nat_segs << log_blocks_per_seg;
3506        nat_bits_bytes = nat_blocks / BITS_PER_BYTE;
3507        nat_bits_blocks = F2FS_BLK_ALIGN((nat_bits_bytes << 1) + 8);
3508        if (__is_set_ckpt_flags(ckpt, CP_NAT_BITS_FLAG) &&
3509                (cp_payload + F2FS_CP_PACKS +
3510                NR_CURSEG_PERSIST_TYPE + nat_bits_blocks >= blocks_per_seg)) {
3511                f2fs_warn(sbi, "Insane cp_payload: %u, nat_bits_blocks: %u)",
3512                          cp_payload, nat_bits_blocks);
3513                return 1;
3514        }
3515
3516        if (unlikely(f2fs_cp_error(sbi))) {
3517                f2fs_err(sbi, "A bug case: need to run fsck");
3518                return 1;
3519        }
3520        return 0;
3521}
3522
3523static void init_sb_info(struct f2fs_sb_info *sbi)
3524{
3525        struct f2fs_super_block *raw_super = sbi->raw_super;
3526        int i;
3527
3528        sbi->log_sectors_per_block =
3529                le32_to_cpu(raw_super->log_sectors_per_block);
3530        sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3531        sbi->blocksize = 1 << sbi->log_blocksize;
3532        sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3533        sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
3534        sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3535        sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3536        sbi->total_sections = le32_to_cpu(raw_super->section_count);
3537        sbi->total_node_count =
3538                (le32_to_cpu(raw_super->segment_count_nat) / 2)
3539                        * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
3540        F2FS_ROOT_INO(sbi) = le32_to_cpu(raw_super->root_ino);
3541        F2FS_NODE_INO(sbi) = le32_to_cpu(raw_super->node_ino);
3542        F2FS_META_INO(sbi) = le32_to_cpu(raw_super->meta_ino);
3543        sbi->cur_victim_sec = NULL_SECNO;
3544        sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3545        sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3546        sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3547        sbi->migration_granularity = sbi->segs_per_sec;
3548        sbi->seq_file_ra_mul = MIN_RA_MUL;
3549        sbi->max_fragment_chunk = DEF_FRAGMENT_SIZE;
3550        sbi->max_fragment_hole = DEF_FRAGMENT_SIZE;
3551
3552        sbi->dir_level = DEF_DIR_LEVEL;
3553        sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3554        sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3555        sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3556        sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3557        sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3558        sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3559                                DEF_UMOUNT_DISCARD_TIMEOUT;
3560        clear_sbi_flag(sbi, SBI_NEED_FSCK);
3561
3562        for (i = 0; i < NR_COUNT_TYPE; i++)
3563                atomic_set(&sbi->nr_pages[i], 0);
3564
3565        for (i = 0; i < META; i++)
3566                atomic_set(&sbi->wb_sync_req[i], 0);
3567
3568        INIT_LIST_HEAD(&sbi->s_list);
3569        mutex_init(&sbi->umount_mutex);
3570        init_rwsem(&sbi->io_order_lock);
3571        spin_lock_init(&sbi->cp_lock);
3572
3573        sbi->dirty_device = 0;
3574        spin_lock_init(&sbi->dev_lock);
3575
3576        init_rwsem(&sbi->sb_lock);
3577        init_rwsem(&sbi->pin_sem);
3578}
3579
3580static int init_percpu_info(struct f2fs_sb_info *sbi)
3581{
3582        int err;
3583
3584        err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3585        if (err)
3586                return err;
3587
3588        err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3589                                                                GFP_KERNEL);
3590        if (err)
3591                percpu_counter_destroy(&sbi->alloc_valid_block_count);
3592
3593        return err;
3594}
3595
3596#ifdef CONFIG_BLK_DEV_ZONED
3597
3598struct f2fs_report_zones_args {
3599        struct f2fs_dev_info *dev;
3600        bool zone_cap_mismatch;
3601};
3602
3603static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3604                              void *data)
3605{
3606        struct f2fs_report_zones_args *rz_args = data;
3607
3608        if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3609                return 0;
3610
3611        set_bit(idx, rz_args->dev->blkz_seq);
3612        rz_args->dev->zone_capacity_blocks[idx] = zone->capacity >>
3613                                                F2FS_LOG_SECTORS_PER_BLOCK;
3614        if (zone->len != zone->capacity && !rz_args->zone_cap_mismatch)
3615                rz_args->zone_cap_mismatch = true;
3616
3617        return 0;
3618}
3619
3620static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3621{
3622        struct block_device *bdev = FDEV(devi).bdev;
3623        sector_t nr_sectors = bdev_nr_sectors(bdev);
3624        struct f2fs_report_zones_args rep_zone_arg;
3625        int ret;
3626
3627        if (!f2fs_sb_has_blkzoned(sbi))
3628                return 0;
3629
3630        if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3631                                SECTOR_TO_BLOCK(bdev_zone_sectors(bdev)))
3632                return -EINVAL;
3633        sbi->blocks_per_blkz = SECTOR_TO_BLOCK(bdev_zone_sectors(bdev));
3634        if (sbi->log_blocks_per_blkz && sbi->log_blocks_per_blkz !=
3635                                __ilog2_u32(sbi->blocks_per_blkz))
3636                return -EINVAL;
3637        sbi->log_blocks_per_blkz = __ilog2_u32(sbi->blocks_per_blkz);
3638        FDEV(devi).nr_blkz = SECTOR_TO_BLOCK(nr_sectors) >>
3639                                        sbi->log_blocks_per_blkz;
3640        if (nr_sectors & (bdev_zone_sectors(bdev) - 1))
3641                FDEV(devi).nr_blkz++;
3642
3643        FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3644                                        BITS_TO_LONGS(FDEV(devi).nr_blkz)
3645                                        * sizeof(unsigned long),
3646                                        GFP_KERNEL);
3647        if (!FDEV(devi).blkz_seq)
3648                return -ENOMEM;
3649
3650        /* Get block zones type and zone-capacity */
3651        FDEV(devi).zone_capacity_blocks = f2fs_kzalloc(sbi,
3652                                        FDEV(devi).nr_blkz * sizeof(block_t),
3653                                        GFP_KERNEL);
3654        if (!FDEV(devi).zone_capacity_blocks)
3655                return -ENOMEM;
3656
3657        rep_zone_arg.dev = &FDEV(devi);
3658        rep_zone_arg.zone_cap_mismatch = false;
3659
3660        ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3661                                  &rep_zone_arg);
3662        if (ret < 0)
3663                return ret;
3664
3665        if (!rep_zone_arg.zone_cap_mismatch) {
3666                kfree(FDEV(devi).zone_capacity_blocks);
3667                FDEV(devi).zone_capacity_blocks = NULL;
3668        }
3669
3670        return 0;
3671}
3672#endif
3673
3674/*
3675 * Read f2fs raw super block.
3676 * Because we have two copies of super block, so read both of them
3677 * to get the first valid one. If any one of them is broken, we pass
3678 * them recovery flag back to the caller.
3679 */
3680static int read_raw_super_block(struct f2fs_sb_info *sbi,
3681                        struct f2fs_super_block **raw_super,
3682                        int *valid_super_block, int *recovery)
3683{
3684        struct super_block *sb = sbi->sb;
3685        int block;
3686        struct buffer_head *bh;
3687        struct f2fs_super_block *super;
3688        int err = 0;
3689
3690        super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3691        if (!super)
3692                return -ENOMEM;
3693
3694        for (block = 0; block < 2; block++) {
3695                bh = sb_bread(sb, block);
3696                if (!bh) {
3697                        f2fs_err(sbi, "Unable to read %dth superblock",
3698                                 block + 1);
3699                        err = -EIO;
3700                        *recovery = 1;
3701                        continue;
3702                }
3703
3704                /* sanity checking of raw super */
3705                err = sanity_check_raw_super(sbi, bh);
3706                if (err) {
3707                        f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3708                                 block + 1);
3709                        brelse(bh);
3710                        *recovery = 1;
3711                        continue;
3712                }
3713
3714                if (!*raw_super) {
3715                        memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3716                                                        sizeof(*super));
3717                        *valid_super_block = block;
3718                        *raw_super = super;
3719                }
3720                brelse(bh);
3721        }
3722
3723        /* No valid superblock */
3724        if (!*raw_super)
3725                kfree(super);
3726        else
3727                err = 0;
3728
3729        return err;
3730}
3731
3732int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
3733{
3734        struct buffer_head *bh;
3735        __u32 crc = 0;
3736        int err;
3737
3738        if ((recover && f2fs_readonly(sbi->sb)) ||
3739                                bdev_read_only(sbi->sb->s_bdev)) {
3740                set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3741                return -EROFS;
3742        }
3743
3744        /* we should update superblock crc here */
3745        if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
3746                crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
3747                                offsetof(struct f2fs_super_block, crc));
3748                F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
3749        }
3750
3751        /* write back-up superblock first */
3752        bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
3753        if (!bh)
3754                return -EIO;
3755        err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3756        brelse(bh);
3757
3758        /* if we are in recovery path, skip writing valid superblock */
3759        if (recover || err)
3760                return err;
3761
3762        /* write current valid superblock */
3763        bh = sb_bread(sbi->sb, sbi->valid_super_block);
3764        if (!bh)
3765                return -EIO;
3766        err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3767        brelse(bh);
3768        return err;
3769}
3770
3771static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
3772{
3773        struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3774        unsigned int max_devices = MAX_DEVICES;
3775        unsigned int logical_blksize;
3776        int i;
3777
3778        /* Initialize single device information */
3779        if (!RDEV(0).path[0]) {
3780                if (!bdev_is_zoned(sbi->sb->s_bdev))
3781                        return 0;
3782                max_devices = 1;
3783        }
3784
3785        /*
3786         * Initialize multiple devices information, or single
3787         * zoned block device information.
3788         */
3789        sbi->devs = f2fs_kzalloc(sbi,
3790                                 array_size(max_devices,
3791                                            sizeof(struct f2fs_dev_info)),
3792                                 GFP_KERNEL);
3793        if (!sbi->devs)
3794                return -ENOMEM;
3795
3796        logical_blksize = bdev_logical_block_size(sbi->sb->s_bdev);
3797        sbi->aligned_blksize = true;
3798
3799        for (i = 0; i < max_devices; i++) {
3800
3801                if (i > 0 && !RDEV(i).path[0])
3802                        break;
3803
3804                if (max_devices == 1) {
3805                        /* Single zoned block device mount */
3806                        FDEV(0).bdev =
3807                                blkdev_get_by_dev(sbi->sb->s_bdev->bd_dev,
3808                                        sbi->sb->s_mode, sbi->sb->s_type);
3809                } else {
3810                        /* Multi-device mount */
3811                        memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
3812                        FDEV(i).total_segments =
3813                                le32_to_cpu(RDEV(i).total_segments);
3814                        if (i == 0) {
3815                                FDEV(i).start_blk = 0;
3816                                FDEV(i).end_blk = FDEV(i).start_blk +
3817                                    (FDEV(i).total_segments <<
3818                                    sbi->log_blocks_per_seg) - 1 +
3819                                    le32_to_cpu(raw_super->segment0_blkaddr);
3820                        } else {
3821                                FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
3822                                FDEV(i).end_blk = FDEV(i).start_blk +
3823                                        (FDEV(i).total_segments <<
3824                                        sbi->log_blocks_per_seg) - 1;
3825                        }
3826                        FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
3827                                        sbi->sb->s_mode, sbi->sb->s_type);
3828                }
3829                if (IS_ERR(FDEV(i).bdev))
3830                        return PTR_ERR(FDEV(i).bdev);
3831
3832                /* to release errored devices */
3833                sbi->s_ndevs = i + 1;
3834
3835                if (logical_blksize != bdev_logical_block_size(FDEV(i).bdev))
3836                        sbi->aligned_blksize = false;
3837
3838#ifdef CONFIG_BLK_DEV_ZONED
3839                if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
3840                                !f2fs_sb_has_blkzoned(sbi)) {
3841                        f2fs_err(sbi, "Zoned block device feature not enabled");
3842                        return -EINVAL;
3843                }
3844                if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
3845                        if (init_blkz_info(sbi, i)) {
3846                                f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
3847                                return -EINVAL;
3848                        }
3849                        if (max_devices == 1)
3850                                break;
3851                        f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
3852                                  i, FDEV(i).path,
3853                                  FDEV(i).total_segments,
3854                                  FDEV(i).start_blk, FDEV(i).end_blk,
3855                                  bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
3856                                  "Host-aware" : "Host-managed");
3857                        continue;
3858                }
3859#endif
3860                f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
3861                          i, FDEV(i).path,
3862                          FDEV(i).total_segments,
3863                          FDEV(i).start_blk, FDEV(i).end_blk);
3864        }
3865        f2fs_info(sbi,
3866                  "IO Block Size: %8d KB", F2FS_IO_SIZE_KB(sbi));
3867        return 0;
3868}
3869
3870static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
3871{
3872#ifdef CONFIG_UNICODE
3873        if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
3874                const struct f2fs_sb_encodings *encoding_info;
3875                struct unicode_map *encoding;
3876                __u16 encoding_flags;
3877
3878                if (f2fs_sb_read_encoding(sbi->raw_super, &encoding_info,
3879                                          &encoding_flags)) {
3880                        f2fs_err(sbi,
3881                                 "Encoding requested by superblock is unknown");
3882                        return -EINVAL;
3883                }
3884
3885                encoding = utf8_load(encoding_info->version);
3886                if (IS_ERR(encoding)) {
3887                        f2fs_err(sbi,
3888                                 "can't mount with superblock charset: %s-%s "
3889                                 "not supported by the kernel. flags: 0x%x.",
3890                                 encoding_info->name, encoding_info->version,
3891                                 encoding_flags);
3892                        return PTR_ERR(encoding);
3893                }
3894                f2fs_info(sbi, "Using encoding defined by superblock: "
3895                         "%s-%s with flags 0x%hx", encoding_info->name,
3896                         encoding_info->version?:"\b", encoding_flags);
3897
3898                sbi->sb->s_encoding = encoding;
3899                sbi->sb->s_encoding_flags = encoding_flags;
3900        }
3901#else
3902        if (f2fs_sb_has_casefold(sbi)) {
3903                f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
3904                return -EINVAL;
3905        }
3906#endif
3907        return 0;
3908}
3909
3910static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
3911{
3912        struct f2fs_sm_info *sm_i = SM_I(sbi);
3913
3914        /* adjust parameters according to the volume size */
3915        if (sm_i->main_segments <= SMALL_VOLUME_SEGMENTS) {
3916                F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
3917                if (f2fs_block_unit_discard(sbi))
3918                        sm_i->dcc_info->discard_granularity = 1;
3919                sm_i->ipu_policy = 1 << F2FS_IPU_FORCE;
3920        }
3921
3922        sbi->readdir_ra = 1;
3923}
3924
3925static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
3926{
3927        struct f2fs_sb_info *sbi;
3928        struct f2fs_super_block *raw_super;
3929        struct inode *root;
3930        int err;
3931        bool skip_recovery = false, need_fsck = false;
3932        char *options = NULL;
3933        int recovery, i, valid_super_block;
3934        struct curseg_info *seg_i;
3935        int retry_cnt = 1;
3936
3937try_onemore:
3938        err = -EINVAL;
3939        raw_super = NULL;
3940        valid_super_block = -1;
3941        recovery = 0;
3942
3943        /* allocate memory for f2fs-specific super block info */
3944        sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
3945        if (!sbi)
3946                return -ENOMEM;
3947
3948        sbi->sb = sb;
3949
3950        /* Load the checksum driver */
3951        sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
3952        if (IS_ERR(sbi->s_chksum_driver)) {
3953                f2fs_err(sbi, "Cannot load crc32 driver.");
3954                err = PTR_ERR(sbi->s_chksum_driver);
3955                sbi->s_chksum_driver = NULL;
3956                goto free_sbi;
3957        }
3958
3959        /* set a block size */
3960        if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
3961                f2fs_err(sbi, "unable to set blocksize");
3962                goto free_sbi;
3963        }
3964
3965        err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
3966                                                                &recovery);
3967        if (err)
3968                goto free_sbi;
3969
3970        sb->s_fs_info = sbi;
3971        sbi->raw_super = raw_super;
3972
3973        /* precompute checksum seed for metadata */
3974        if (f2fs_sb_has_inode_chksum(sbi))
3975                sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
3976                                                sizeof(raw_super->uuid));
3977
3978        default_options(sbi);
3979        /* parse mount options */
3980        options = kstrdup((const char *)data, GFP_KERNEL);
3981        if (data && !options) {
3982                err = -ENOMEM;
3983                goto free_sb_buf;
3984        }
3985
3986        err = parse_options(sb, options, false);
3987        if (err)
3988                goto free_options;
3989
3990        sb->s_maxbytes = max_file_blocks(NULL) <<
3991                                le32_to_cpu(raw_super->log_blocksize);
3992        sb->s_max_links = F2FS_LINK_MAX;
3993
3994        err = f2fs_setup_casefold(sbi);
3995        if (err)
3996                goto free_options;
3997
3998#ifdef CONFIG_QUOTA
3999        sb->dq_op = &f2fs_quota_operations;
4000        sb->s_qcop = &f2fs_quotactl_ops;
4001        sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
4002
4003        if (f2fs_sb_has_quota_ino(sbi)) {
4004                for (i = 0; i < MAXQUOTAS; i++) {
4005                        if (f2fs_qf_ino(sbi->sb, i))
4006                                sbi->nquota_files++;
4007                }
4008        }
4009#endif
4010
4011        sb->s_op = &f2fs_sops;
4012#ifdef CONFIG_FS_ENCRYPTION
4013        sb->s_cop = &f2fs_cryptops;
4014#endif
4015#ifdef CONFIG_FS_VERITY
4016        sb->s_vop = &f2fs_verityops;
4017#endif
4018        sb->s_xattr = f2fs_xattr_handlers;
4019        sb->s_export_op = &f2fs_export_ops;
4020        sb->s_magic = F2FS_SUPER_MAGIC;
4021        sb->s_time_gran = 1;
4022        sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
4023                (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
4024        memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
4025        sb->s_iflags |= SB_I_CGROUPWB;
4026
4027        /* init f2fs-specific super block info */
4028        sbi->valid_super_block = valid_super_block;
4029        init_rwsem(&sbi->gc_lock);
4030        mutex_init(&sbi->writepages);
4031        init_rwsem(&sbi->cp_global_sem);
4032        init_rwsem(&sbi->node_write);
4033        init_rwsem(&sbi->node_change);
4034
4035        /* disallow all the data/node/meta page writes */
4036        set_sbi_flag(sbi, SBI_POR_DOING);
4037        spin_lock_init(&sbi->stat_lock);
4038
4039        for (i = 0; i < NR_PAGE_TYPE; i++) {
4040                int n = (i == META) ? 1 : NR_TEMP_TYPE;
4041                int j;
4042
4043                sbi->write_io[i] =
4044                        f2fs_kmalloc(sbi,
4045                                     array_size(n,
4046                                                sizeof(struct f2fs_bio_info)),
4047                                     GFP_KERNEL);
4048                if (!sbi->write_io[i]) {
4049                        err = -ENOMEM;
4050                        goto free_bio_info;
4051                }
4052
4053                for (j = HOT; j < n; j++) {
4054                        init_rwsem(&sbi->write_io[i][j].io_rwsem);
4055                        sbi->write_io[i][j].sbi = sbi;
4056                        sbi->write_io[i][j].bio = NULL;
4057                        spin_lock_init(&sbi->write_io[i][j].io_lock);
4058                        INIT_LIST_HEAD(&sbi->write_io[i][j].io_list);
4059                        INIT_LIST_HEAD(&sbi->write_io[i][j].bio_list);
4060                        init_rwsem(&sbi->write_io[i][j].bio_list_lock);
4061                }
4062        }
4063
4064        init_rwsem(&sbi->cp_rwsem);
4065        init_rwsem(&sbi->quota_sem);
4066        init_waitqueue_head(&sbi->cp_wait);
4067        init_sb_info(sbi);
4068
4069        err = f2fs_init_iostat(sbi);
4070        if (err)
4071                goto free_bio_info;
4072
4073        err = init_percpu_info(sbi);
4074        if (err)
4075                goto free_iostat;
4076
4077        if (F2FS_IO_ALIGNED(sbi)) {
4078                sbi->write_io_dummy =
4079                        mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
4080                if (!sbi->write_io_dummy) {
4081                        err = -ENOMEM;
4082                        goto free_percpu;
4083                }
4084        }
4085
4086        /* init per sbi slab cache */
4087        err = f2fs_init_xattr_caches(sbi);
4088        if (err)
4089                goto free_io_dummy;
4090        err = f2fs_init_page_array_cache(sbi);
4091        if (err)
4092                goto free_xattr_cache;
4093
4094        /* get an inode for meta space */
4095        sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
4096        if (IS_ERR(sbi->meta_inode)) {
4097                f2fs_err(sbi, "Failed to read F2FS meta data inode");
4098                err = PTR_ERR(sbi->meta_inode);
4099                goto free_page_array_cache;
4100        }
4101
4102        err = f2fs_get_valid_checkpoint(sbi);
4103        if (err) {
4104                f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
4105                goto free_meta_inode;
4106        }
4107
4108        if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
4109                set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
4110        if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
4111                set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4112                sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
4113        }
4114
4115        if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
4116                set_sbi_flag(sbi, SBI_NEED_FSCK);
4117
4118        /* Initialize device list */
4119        err = f2fs_scan_devices(sbi);
4120        if (err) {
4121                f2fs_err(sbi, "Failed to find devices");
4122                goto free_devices;
4123        }
4124
4125        err = f2fs_init_post_read_wq(sbi);
4126        if (err) {
4127                f2fs_err(sbi, "Failed to initialize post read workqueue");
4128                goto free_devices;
4129        }
4130
4131        sbi->total_valid_node_count =
4132                                le32_to_cpu(sbi->ckpt->valid_node_count);
4133        percpu_counter_set(&sbi->total_valid_inode_count,
4134                                le32_to_cpu(sbi->ckpt->valid_inode_count));
4135        sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
4136        sbi->total_valid_block_count =
4137                                le64_to_cpu(sbi->ckpt->valid_block_count);
4138        sbi->last_valid_block_count = sbi->total_valid_block_count;
4139        sbi->reserved_blocks = 0;
4140        sbi->current_reserved_blocks = 0;
4141        limit_reserve_root(sbi);
4142        adjust_unusable_cap_perc(sbi);
4143
4144        for (i = 0; i < NR_INODE_TYPE; i++) {
4145                INIT_LIST_HEAD(&sbi->inode_list[i]);
4146                spin_lock_init(&sbi->inode_lock[i]);
4147        }
4148        mutex_init(&sbi->flush_lock);
4149
4150        f2fs_init_extent_cache_info(sbi);
4151
4152        f2fs_init_ino_entry_info(sbi);
4153
4154        f2fs_init_fsync_node_info(sbi);
4155
4156        /* setup checkpoint request control and start checkpoint issue thread */
4157        f2fs_init_ckpt_req_control(sbi);
4158        if (!f2fs_readonly(sb) && !test_opt(sbi, DISABLE_CHECKPOINT) &&
4159                        test_opt(sbi, MERGE_CHECKPOINT)) {
4160                err = f2fs_start_ckpt_thread(sbi);
4161                if (err) {
4162                        f2fs_err(sbi,
4163                            "Failed to start F2FS issue_checkpoint_thread (%d)",
4164                            err);
4165                        goto stop_ckpt_thread;
4166                }
4167        }
4168
4169        /* setup f2fs internal modules */
4170        err = f2fs_build_segment_manager(sbi);
4171        if (err) {
4172                f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
4173                         err);
4174                goto free_sm;
4175        }
4176        err = f2fs_build_node_manager(sbi);
4177        if (err) {
4178                f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
4179                         err);
4180                goto free_nm;
4181        }
4182
4183        /* For write statistics */
4184        sbi->sectors_written_start = f2fs_get_sectors_written(sbi);
4185
4186        /* Read accumulated write IO statistics if exists */
4187        seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
4188        if (__exist_node_summaries(sbi))
4189                sbi->kbytes_written =
4190                        le64_to_cpu(seg_i->journal->info.kbytes_written);
4191
4192        f2fs_build_gc_manager(sbi);
4193
4194        err = f2fs_build_stats(sbi);
4195        if (err)
4196                goto free_nm;
4197
4198        /* get an inode for node space */
4199        sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
4200        if (IS_ERR(sbi->node_inode)) {
4201                f2fs_err(sbi, "Failed to read node inode");
4202                err = PTR_ERR(sbi->node_inode);
4203                goto free_stats;
4204        }
4205
4206        /* read root inode and dentry */
4207        root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
4208        if (IS_ERR(root)) {
4209                f2fs_err(sbi, "Failed to read root inode");
4210                err = PTR_ERR(root);
4211                goto free_node_inode;
4212        }
4213        if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
4214                        !root->i_size || !root->i_nlink) {
4215                iput(root);
4216                err = -EINVAL;
4217                goto free_node_inode;
4218        }
4219
4220        sb->s_root = d_make_root(root); /* allocate root dentry */
4221        if (!sb->s_root) {
4222                err = -ENOMEM;
4223                goto free_node_inode;
4224        }
4225
4226        err = f2fs_init_compress_inode(sbi);
4227        if (err)
4228                goto free_root_inode;
4229
4230        err = f2fs_register_sysfs(sbi);
4231        if (err)
4232                goto free_compress_inode;
4233
4234#ifdef CONFIG_QUOTA
4235        /* Enable quota usage during mount */
4236        if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
4237                err = f2fs_enable_quotas(sb);
4238                if (err)
4239                        f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
4240        }
4241#endif
4242        /* if there are any orphan inodes, free them */
4243        err = f2fs_recover_orphan_inodes(sbi);
4244        if (err)
4245                goto free_meta;
4246
4247        if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
4248                goto reset_checkpoint;
4249
4250        /* recover fsynced data */
4251        if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
4252                        !test_opt(sbi, NORECOVERY)) {
4253                /*
4254                 * mount should be failed, when device has readonly mode, and
4255                 * previous checkpoint was not done by clean system shutdown.
4256                 */
4257                if (f2fs_hw_is_readonly(sbi)) {
4258                        if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4259                                err = f2fs_recover_fsync_data(sbi, true);
4260                                if (err > 0) {
4261                                        err = -EROFS;
4262                                        f2fs_err(sbi, "Need to recover fsync data, but "
4263                                                "write access unavailable, please try "
4264                                                "mount w/ disable_roll_forward or norecovery");
4265                                }
4266                                if (err < 0)
4267                                        goto free_meta;
4268                        }
4269                        f2fs_info(sbi, "write access unavailable, skipping recovery");
4270                        goto reset_checkpoint;
4271                }
4272
4273                if (need_fsck)
4274                        set_sbi_flag(sbi, SBI_NEED_FSCK);
4275
4276                if (skip_recovery)
4277                        goto reset_checkpoint;
4278
4279                err = f2fs_recover_fsync_data(sbi, false);
4280                if (err < 0) {
4281                        if (err != -ENOMEM)
4282                                skip_recovery = true;
4283                        need_fsck = true;
4284                        f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
4285                                 err);
4286                        goto free_meta;
4287                }
4288        } else {
4289                err = f2fs_recover_fsync_data(sbi, true);
4290
4291                if (!f2fs_readonly(sb) && err > 0) {
4292                        err = -EINVAL;
4293                        f2fs_err(sbi, "Need to recover fsync data");
4294                        goto free_meta;
4295                }
4296        }
4297
4298        /*
4299         * If the f2fs is not readonly and fsync data recovery succeeds,
4300         * check zoned block devices' write pointer consistency.
4301         */
4302        if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
4303                err = f2fs_check_write_pointer(sbi);
4304                if (err)
4305                        goto free_meta;
4306        }
4307
4308reset_checkpoint:
4309        f2fs_init_inmem_curseg(sbi);
4310
4311        /* f2fs_recover_fsync_data() cleared this already */
4312        clear_sbi_flag(sbi, SBI_POR_DOING);
4313
4314        if (test_opt(sbi, DISABLE_CHECKPOINT)) {
4315                err = f2fs_disable_checkpoint(sbi);
4316                if (err)
4317                        goto sync_free_meta;
4318        } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
4319                f2fs_enable_checkpoint(sbi);
4320        }
4321
4322        /*
4323         * If filesystem is not mounted as read-only then
4324         * do start the gc_thread.
4325         */
4326        if ((F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF ||
4327                test_opt(sbi, GC_MERGE)) && !f2fs_readonly(sb)) {
4328                /* After POR, we can run background GC thread.*/
4329                err = f2fs_start_gc_thread(sbi);
4330                if (err)
4331                        goto sync_free_meta;
4332        }
4333        kvfree(options);
4334
4335        /* recover broken superblock */
4336        if (recovery) {
4337                err = f2fs_commit_super(sbi, true);
4338                f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
4339                          sbi->valid_super_block ? 1 : 2, err);
4340        }
4341
4342        f2fs_join_shrinker(sbi);
4343
4344        f2fs_tuning_parameters(sbi);
4345
4346        f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
4347                    cur_cp_version(F2FS_CKPT(sbi)));
4348        f2fs_update_time(sbi, CP_TIME);
4349        f2fs_update_time(sbi, REQ_TIME);
4350        clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4351        return 0;
4352
4353sync_free_meta:
4354        /* safe to flush all the data */
4355        sync_filesystem(sbi->sb);
4356        retry_cnt = 0;
4357
4358free_meta:
4359#ifdef CONFIG_QUOTA
4360        f2fs_truncate_quota_inode_pages(sb);
4361        if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
4362                f2fs_quota_off_umount(sbi->sb);
4363#endif
4364        /*
4365         * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
4366         * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
4367         * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
4368         * falls into an infinite loop in f2fs_sync_meta_pages().
4369         */
4370        truncate_inode_pages_final(META_MAPPING(sbi));
4371        /* evict some inodes being cached by GC */
4372        evict_inodes(sb);
4373        f2fs_unregister_sysfs(sbi);
4374free_compress_inode:
4375        f2fs_destroy_compress_inode(sbi);
4376free_root_inode:
4377        dput(sb->s_root);
4378        sb->s_root = NULL;
4379free_node_inode:
4380        f2fs_release_ino_entry(sbi, true);
4381        truncate_inode_pages_final(NODE_MAPPING(sbi));
4382        iput(sbi->node_inode);
4383        sbi->node_inode = NULL;
4384free_stats:
4385        f2fs_destroy_stats(sbi);
4386free_nm:
4387        /* stop discard thread before destroying node manager */
4388        f2fs_stop_discard_thread(sbi);
4389        f2fs_destroy_node_manager(sbi);
4390free_sm:
4391        f2fs_destroy_segment_manager(sbi);
4392        f2fs_destroy_post_read_wq(sbi);
4393stop_ckpt_thread:
4394        f2fs_stop_ckpt_thread(sbi);
4395free_devices:
4396        destroy_device_list(sbi);
4397        kvfree(sbi->ckpt);
4398free_meta_inode:
4399        make_bad_inode(sbi->meta_inode);
4400        iput(sbi->meta_inode);
4401        sbi->meta_inode = NULL;
4402free_page_array_cache:
4403        f2fs_destroy_page_array_cache(sbi);
4404free_xattr_cache:
4405        f2fs_destroy_xattr_caches(sbi);
4406free_io_dummy:
4407        mempool_destroy(sbi->write_io_dummy);
4408free_percpu:
4409        destroy_percpu_info(sbi);
4410free_iostat:
4411        f2fs_destroy_iostat(sbi);
4412free_bio_info:
4413        for (i = 0; i < NR_PAGE_TYPE; i++)
4414                kvfree(sbi->write_io[i]);
4415
4416#ifdef CONFIG_UNICODE
4417        utf8_unload(sb->s_encoding);
4418        sb->s_encoding = NULL;
4419#endif
4420free_options:
4421#ifdef CONFIG_QUOTA
4422        for (i = 0; i < MAXQUOTAS; i++)
4423                kfree(F2FS_OPTION(sbi).s_qf_names[i]);
4424#endif
4425        fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
4426        kvfree(options);
4427free_sb_buf:
4428        kfree(raw_super);
4429free_sbi:
4430        if (sbi->s_chksum_driver)
4431                crypto_free_shash(sbi->s_chksum_driver);
4432        kfree(sbi);
4433
4434        /* give only one another chance */
4435        if (retry_cnt > 0 && skip_recovery) {
4436                retry_cnt--;
4437                shrink_dcache_sb(sb);
4438                goto try_onemore;
4439        }
4440        return err;
4441}
4442
4443static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
4444                        const char *dev_name, void *data)
4445{
4446        return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
4447}
4448
4449static void kill_f2fs_super(struct super_block *sb)
4450{
4451        if (sb->s_root) {
4452                struct f2fs_sb_info *sbi = F2FS_SB(sb);
4453
4454                set_sbi_flag(sbi, SBI_IS_CLOSE);
4455                f2fs_stop_gc_thread(sbi);
4456                f2fs_stop_discard_thread(sbi);
4457
4458#ifdef CONFIG_F2FS_FS_COMPRESSION
4459                /*
4460                 * latter evict_inode() can bypass checking and invalidating
4461                 * compress inode cache.
4462                 */
4463                if (test_opt(sbi, COMPRESS_CACHE))
4464                        truncate_inode_pages_final(COMPRESS_MAPPING(sbi));
4465#endif
4466
4467                if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
4468                                !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4469                        struct cp_control cpc = {
4470                                .reason = CP_UMOUNT,
4471                        };
4472                        f2fs_write_checkpoint(sbi, &cpc);
4473                }
4474
4475                if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
4476                        sb->s_flags &= ~SB_RDONLY;
4477        }
4478        kill_block_super(sb);
4479}
4480
4481static struct file_system_type f2fs_fs_type = {
4482        .owner          = THIS_MODULE,
4483        .name           = "f2fs",
4484        .mount          = f2fs_mount,
4485        .kill_sb        = kill_f2fs_super,
4486        .fs_flags       = FS_REQUIRES_DEV,
4487};
4488MODULE_ALIAS_FS("f2fs");
4489
4490static int __init init_inodecache(void)
4491{
4492        f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
4493                        sizeof(struct f2fs_inode_info), 0,
4494                        SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
4495        if (!f2fs_inode_cachep)
4496                return -ENOMEM;
4497        return 0;
4498}
4499
4500static void destroy_inodecache(void)
4501{
4502        /*
4503         * Make sure all delayed rcu free inodes are flushed before we
4504         * destroy cache.
4505         */
4506        rcu_barrier();
4507        kmem_cache_destroy(f2fs_inode_cachep);
4508}
4509
4510static int __init init_f2fs_fs(void)
4511{
4512        int err;
4513
4514        if (PAGE_SIZE != F2FS_BLKSIZE) {
4515                printk("F2FS not supported on PAGE_SIZE(%lu) != %d\n",
4516                                PAGE_SIZE, F2FS_BLKSIZE);
4517                return -EINVAL;
4518        }
4519
4520        err = init_inodecache();
4521        if (err)
4522                goto fail;
4523        err = f2fs_create_node_manager_caches();
4524        if (err)
4525                goto free_inodecache;
4526        err = f2fs_create_segment_manager_caches();
4527        if (err)
4528                goto free_node_manager_caches;
4529        err = f2fs_create_checkpoint_caches();
4530        if (err)
4531                goto free_segment_manager_caches;
4532        err = f2fs_create_recovery_cache();
4533        if (err)
4534                goto free_checkpoint_caches;
4535        err = f2fs_create_extent_cache();
4536        if (err)
4537                goto free_recovery_cache;
4538        err = f2fs_create_garbage_collection_cache();
4539        if (err)
4540                goto free_extent_cache;
4541        err = f2fs_init_sysfs();
4542        if (err)
4543                goto free_garbage_collection_cache;
4544        err = register_shrinker(&f2fs_shrinker_info);
4545        if (err)
4546                goto free_sysfs;
4547        err = register_filesystem(&f2fs_fs_type);
4548        if (err)
4549                goto free_shrinker;
4550        f2fs_create_root_stats();
4551        err = f2fs_init_post_read_processing();
4552        if (err)
4553                goto free_root_stats;
4554        err = f2fs_init_iostat_processing();
4555        if (err)
4556                goto free_post_read;
4557        err = f2fs_init_bio_entry_cache();
4558        if (err)
4559                goto free_iostat;
4560        err = f2fs_init_bioset();
4561        if (err)
4562                goto free_bio_enrty_cache;
4563        err = f2fs_init_compress_mempool();
4564        if (err)
4565                goto free_bioset;
4566        err = f2fs_init_compress_cache();
4567        if (err)
4568                goto free_compress_mempool;
4569        err = f2fs_create_casefold_cache();
4570        if (err)
4571                goto free_compress_cache;
4572        return 0;
4573free_compress_cache:
4574        f2fs_destroy_compress_cache();
4575free_compress_mempool:
4576        f2fs_destroy_compress_mempool();
4577free_bioset:
4578        f2fs_destroy_bioset();
4579free_bio_enrty_cache:
4580        f2fs_destroy_bio_entry_cache();
4581free_iostat:
4582        f2fs_destroy_iostat_processing();
4583free_post_read:
4584        f2fs_destroy_post_read_processing();
4585free_root_stats:
4586        f2fs_destroy_root_stats();
4587        unregister_filesystem(&f2fs_fs_type);
4588free_shrinker:
4589        unregister_shrinker(&f2fs_shrinker_info);
4590free_sysfs:
4591        f2fs_exit_sysfs();
4592free_garbage_collection_cache:
4593        f2fs_destroy_garbage_collection_cache();
4594free_extent_cache:
4595        f2fs_destroy_extent_cache();
4596free_recovery_cache:
4597        f2fs_destroy_recovery_cache();
4598free_checkpoint_caches:
4599        f2fs_destroy_checkpoint_caches();
4600free_segment_manager_caches:
4601        f2fs_destroy_segment_manager_caches();
4602free_node_manager_caches:
4603        f2fs_destroy_node_manager_caches();
4604free_inodecache:
4605        destroy_inodecache();
4606fail:
4607        return err;
4608}
4609
4610static void __exit exit_f2fs_fs(void)
4611{
4612        f2fs_destroy_casefold_cache();
4613        f2fs_destroy_compress_cache();
4614        f2fs_destroy_compress_mempool();
4615        f2fs_destroy_bioset();
4616        f2fs_destroy_bio_entry_cache();
4617        f2fs_destroy_iostat_processing();
4618        f2fs_destroy_post_read_processing();
4619        f2fs_destroy_root_stats();
4620        unregister_filesystem(&f2fs_fs_type);
4621        unregister_shrinker(&f2fs_shrinker_info);
4622        f2fs_exit_sysfs();
4623        f2fs_destroy_garbage_collection_cache();
4624        f2fs_destroy_extent_cache();
4625        f2fs_destroy_recovery_cache();
4626        f2fs_destroy_checkpoint_caches();
4627        f2fs_destroy_segment_manager_caches();
4628        f2fs_destroy_node_manager_caches();
4629        destroy_inodecache();
4630}
4631
4632module_init(init_f2fs_fs)
4633module_exit(exit_f2fs_fs)
4634
4635MODULE_AUTHOR("Samsung Electronics's Praesto Team");
4636MODULE_DESCRIPTION("Flash Friendly File System");
4637MODULE_LICENSE("GPL");
4638MODULE_SOFTDEP("pre: crc32");
4639
4640