super.c

   1// SPDX-License-Identifier: GPL-2.0
   2/*
   3 * fs/f2fs/super.c
   4 *
   5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
   6 *             http://www.samsung.com/
   7 */
   8#include <linux/module.h>
   9#include <linux/init.h>
  10#include <linux/fs.h>
  11#include <linux/fs_context.h>
  12#include <linux/sched/mm.h>
  13#include <linux/statfs.h>
  14#include <linux/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#include <linux/ctype.h>
  31#include <linux/fs_parser.h>
  32
  33#include "f2fs.h"
  34#include "node.h"
  35#include "segment.h"
  36#include "xattr.h"
  37#include "gc.h"
  38#include "iostat.h"
  39
  40#define CREATE_TRACE_POINTS
  41#include <trace/events/f2fs.h>
  42
  43static struct kmem_cache *f2fs_inode_cachep;
  44
  45#ifdef CONFIG_F2FS_FAULT_INJECTION
  46
  47const char *f2fs_fault_name[FAULT_MAX] = {
  48	[FAULT_KMALLOC]			= "kmalloc",
  49	[FAULT_KVMALLOC]		= "kvmalloc",
  50	[FAULT_PAGE_ALLOC]		= "page alloc",
  51	[FAULT_PAGE_GET]		= "page get",
  52	[FAULT_ALLOC_BIO]		= "alloc bio(obsolete)",
  53	[FAULT_ALLOC_NID]		= "alloc nid",
  54	[FAULT_ORPHAN]			= "orphan",
  55	[FAULT_BLOCK]			= "no more block",
  56	[FAULT_DIR_DEPTH]		= "too big dir depth",
  57	[FAULT_EVICT_INODE]		= "evict_inode fail",
  58	[FAULT_TRUNCATE]		= "truncate fail",
  59	[FAULT_READ_IO]			= "read IO error",
  60	[FAULT_CHECKPOINT]		= "checkpoint error",
  61	[FAULT_DISCARD]			= "discard error",
  62	[FAULT_WRITE_IO]		= "write IO error",
  63	[FAULT_SLAB_ALLOC]		= "slab alloc",
  64	[FAULT_DQUOT_INIT]		= "dquot initialize",
  65	[FAULT_LOCK_OP]			= "lock_op",
  66	[FAULT_BLKADDR_VALIDITY]	= "invalid blkaddr",
  67	[FAULT_BLKADDR_CONSISTENCE]	= "inconsistent blkaddr",
  68	[FAULT_NO_SEGMENT]		= "no free segment",
  69	[FAULT_INCONSISTENT_FOOTER]	= "inconsistent footer",
  70	[FAULT_TIMEOUT]			= "timeout",
  71	[FAULT_VMALLOC]			= "vmalloc",
  72};
  73
  74int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate,
  75				unsigned long type, enum fault_option fo)
  76{
  77	struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
  78
  79	if (fo & FAULT_ALL) {
  80		memset(ffi, 0, sizeof(struct f2fs_fault_info));
  81		return 0;
  82	}
  83
  84	if (fo & FAULT_RATE) {
  85		if (rate > INT_MAX)
  86			return -EINVAL;
  87		atomic_set(&ffi->inject_ops, 0);
  88		ffi->inject_rate = (int)rate;
  89		f2fs_info(sbi, "build fault injection rate: %lu", rate);
  90	}
  91
  92	if (fo & FAULT_TYPE) {
  93		if (type >= BIT(FAULT_MAX))
  94			return -EINVAL;
  95		ffi->inject_type = (unsigned int)type;
  96		f2fs_info(sbi, "build fault injection type: 0x%lx", type);
  97	}
  98
  99	return 0;
 100}
 101#endif
 102
 103/* f2fs-wide shrinker description */
 104static struct shrinker *f2fs_shrinker_info;
 105
 106static int __init f2fs_init_shrinker(void)
 107{
 108	f2fs_shrinker_info = shrinker_alloc(0, "f2fs-shrinker");
 109	if (!f2fs_shrinker_info)
 110		return -ENOMEM;
 111
 112	f2fs_shrinker_info->count_objects = f2fs_shrink_count;
 113	f2fs_shrinker_info->scan_objects = f2fs_shrink_scan;
 114
 115	shrinker_register(f2fs_shrinker_info);
 116
 117	return 0;
 118}
 119
 120static void f2fs_exit_shrinker(void)
 121{
 122	shrinker_free(f2fs_shrinker_info);
 123}
 124
 125enum {
 126	Opt_gc_background,
 127	Opt_disable_roll_forward,
 128	Opt_norecovery,
 129	Opt_discard,
 130	Opt_noheap,
 131	Opt_heap,
 132	Opt_user_xattr,
 133	Opt_acl,
 134	Opt_active_logs,
 135	Opt_disable_ext_identify,
 136	Opt_inline_xattr,
 137	Opt_inline_xattr_size,
 138	Opt_inline_data,
 139	Opt_inline_dentry,
 140	Opt_flush_merge,
 141	Opt_barrier,
 142	Opt_fastboot,
 143	Opt_extent_cache,
 144	Opt_data_flush,
 145	Opt_reserve_root,
 146	Opt_reserve_node,
 147	Opt_resgid,
 148	Opt_resuid,
 149	Opt_mode,
 150	Opt_fault_injection,
 151	Opt_fault_type,
 152	Opt_lazytime,
 153	Opt_quota,
 154	Opt_usrquota,
 155	Opt_grpquota,
 156	Opt_prjquota,
 157	Opt_usrjquota,
 158	Opt_grpjquota,
 159	Opt_prjjquota,
 160	Opt_alloc,
 161	Opt_fsync,
 162	Opt_test_dummy_encryption,
 163	Opt_inlinecrypt,
 164	Opt_checkpoint_disable,
 165	Opt_checkpoint_disable_cap,
 166	Opt_checkpoint_disable_cap_perc,
 167	Opt_checkpoint_enable,
 168	Opt_checkpoint_merge,
 169	Opt_compress_algorithm,
 170	Opt_compress_log_size,
 171	Opt_nocompress_extension,
 172	Opt_compress_extension,
 173	Opt_compress_chksum,
 174	Opt_compress_mode,
 175	Opt_compress_cache,
 176	Opt_atgc,
 177	Opt_gc_merge,
 178	Opt_discard_unit,
 179	Opt_memory_mode,
 180	Opt_age_extent_cache,
 181	Opt_errors,
 182	Opt_nat_bits,
 183	Opt_jqfmt,
 184	Opt_checkpoint,
 185	Opt_lookup_mode,
 186	Opt_err,
 187};
 188
 189static const struct constant_table f2fs_param_background_gc[] = {
 190	{"on",		BGGC_MODE_ON},
 191	{"off",		BGGC_MODE_OFF},
 192	{"sync",	BGGC_MODE_SYNC},
 193	{}
 194};
 195
 196static const struct constant_table f2fs_param_mode[] = {
 197	{"adaptive",		FS_MODE_ADAPTIVE},
 198	{"lfs",			FS_MODE_LFS},
 199	{"fragment:segment",	FS_MODE_FRAGMENT_SEG},
 200	{"fragment:block",	FS_MODE_FRAGMENT_BLK},
 201	{}
 202};
 203
 204static const struct constant_table f2fs_param_jqfmt[] = {
 205	{"vfsold",	QFMT_VFS_OLD},
 206	{"vfsv0",	QFMT_VFS_V0},
 207	{"vfsv1",	QFMT_VFS_V1},
 208	{}
 209};
 210
 211static const struct constant_table f2fs_param_alloc_mode[] = {
 212	{"default",	ALLOC_MODE_DEFAULT},
 213	{"reuse",	ALLOC_MODE_REUSE},
 214	{}
 215};
 216static const struct constant_table f2fs_param_fsync_mode[] = {
 217	{"posix",	FSYNC_MODE_POSIX},
 218	{"strict",	FSYNC_MODE_STRICT},
 219	{"nobarrier",	FSYNC_MODE_NOBARRIER},
 220	{}
 221};
 222
 223static const struct constant_table f2fs_param_compress_mode[] = {
 224	{"fs",		COMPR_MODE_FS},
 225	{"user",	COMPR_MODE_USER},
 226	{}
 227};
 228
 229static const struct constant_table f2fs_param_discard_unit[] = {
 230	{"block",	DISCARD_UNIT_BLOCK},
 231	{"segment",	DISCARD_UNIT_SEGMENT},
 232	{"section",	DISCARD_UNIT_SECTION},
 233	{}
 234};
 235
 236static const struct constant_table f2fs_param_memory_mode[] = {
 237	{"normal",	MEMORY_MODE_NORMAL},
 238	{"low",		MEMORY_MODE_LOW},
 239	{}
 240};
 241
 242static const struct constant_table f2fs_param_errors[] = {
 243	{"remount-ro",	MOUNT_ERRORS_READONLY},
 244	{"continue",	MOUNT_ERRORS_CONTINUE},
 245	{"panic",	MOUNT_ERRORS_PANIC},
 246	{}
 247};
 248
 249static const struct constant_table f2fs_param_lookup_mode[] = {
 250	{"perf",	LOOKUP_PERF},
 251	{"compat",	LOOKUP_COMPAT},
 252	{"auto",	LOOKUP_AUTO},
 253	{}
 254};
 255
 256static const struct fs_parameter_spec f2fs_param_specs[] = {
 257	fsparam_enum("background_gc", Opt_gc_background, f2fs_param_background_gc),
 258	fsparam_flag("disable_roll_forward", Opt_disable_roll_forward),
 259	fsparam_flag("norecovery", Opt_norecovery),
 260	fsparam_flag_no("discard", Opt_discard),
 261	fsparam_flag("no_heap", Opt_noheap),
 262	fsparam_flag("heap", Opt_heap),
 263	fsparam_flag_no("user_xattr", Opt_user_xattr),
 264	fsparam_flag_no("acl", Opt_acl),
 265	fsparam_s32("active_logs", Opt_active_logs),
 266	fsparam_flag("disable_ext_identify", Opt_disable_ext_identify),
 267	fsparam_flag_no("inline_xattr", Opt_inline_xattr),
 268	fsparam_s32("inline_xattr_size", Opt_inline_xattr_size),
 269	fsparam_flag_no("inline_data", Opt_inline_data),
 270	fsparam_flag_no("inline_dentry", Opt_inline_dentry),
 271	fsparam_flag_no("flush_merge", Opt_flush_merge),
 272	fsparam_flag_no("barrier", Opt_barrier),
 273	fsparam_flag("fastboot", Opt_fastboot),
 274	fsparam_flag_no("extent_cache", Opt_extent_cache),
 275	fsparam_flag("data_flush", Opt_data_flush),
 276	fsparam_u32("reserve_root", Opt_reserve_root),
 277	fsparam_u32("reserve_node", Opt_reserve_node),
 278	fsparam_gid("resgid", Opt_resgid),
 279	fsparam_uid("resuid", Opt_resuid),
 280	fsparam_enum("mode", Opt_mode, f2fs_param_mode),
 281	fsparam_s32("fault_injection", Opt_fault_injection),
 282	fsparam_u32("fault_type", Opt_fault_type),
 283	fsparam_flag_no("lazytime", Opt_lazytime),
 284	fsparam_flag_no("quota", Opt_quota),
 285	fsparam_flag("usrquota", Opt_usrquota),
 286	fsparam_flag("grpquota", Opt_grpquota),
 287	fsparam_flag("prjquota", Opt_prjquota),
 288	fsparam_string_empty("usrjquota", Opt_usrjquota),
 289	fsparam_string_empty("grpjquota", Opt_grpjquota),
 290	fsparam_string_empty("prjjquota", Opt_prjjquota),
 291	fsparam_flag("nat_bits", Opt_nat_bits),
 292	fsparam_enum("jqfmt", Opt_jqfmt, f2fs_param_jqfmt),
 293	fsparam_enum("alloc_mode", Opt_alloc, f2fs_param_alloc_mode),
 294	fsparam_enum("fsync_mode", Opt_fsync, f2fs_param_fsync_mode),
 295	fsparam_string("test_dummy_encryption", Opt_test_dummy_encryption),
 296	fsparam_flag("test_dummy_encryption", Opt_test_dummy_encryption),
 297	fsparam_flag("inlinecrypt", Opt_inlinecrypt),
 298	fsparam_string("checkpoint", Opt_checkpoint),
 299	fsparam_flag_no("checkpoint_merge", Opt_checkpoint_merge),
 300	fsparam_string("compress_algorithm", Opt_compress_algorithm),
 301	fsparam_u32("compress_log_size", Opt_compress_log_size),
 302	fsparam_string("compress_extension", Opt_compress_extension),
 303	fsparam_string("nocompress_extension", Opt_nocompress_extension),
 304	fsparam_flag("compress_chksum", Opt_compress_chksum),
 305	fsparam_enum("compress_mode", Opt_compress_mode, f2fs_param_compress_mode),
 306	fsparam_flag("compress_cache", Opt_compress_cache),
 307	fsparam_flag("atgc", Opt_atgc),
 308	fsparam_flag_no("gc_merge", Opt_gc_merge),
 309	fsparam_enum("discard_unit", Opt_discard_unit, f2fs_param_discard_unit),
 310	fsparam_enum("memory", Opt_memory_mode, f2fs_param_memory_mode),
 311	fsparam_flag("age_extent_cache", Opt_age_extent_cache),
 312	fsparam_enum("errors", Opt_errors, f2fs_param_errors),
 313	fsparam_enum("lookup_mode", Opt_lookup_mode, f2fs_param_lookup_mode),
 314	{}
 315};
 316
 317/* Resort to a match_table for this interestingly formatted option */
 318static match_table_t f2fs_checkpoint_tokens = {
 319	{Opt_checkpoint_disable, "disable"},
 320	{Opt_checkpoint_disable_cap, "disable:%u"},
 321	{Opt_checkpoint_disable_cap_perc, "disable:%u%%"},
 322	{Opt_checkpoint_enable, "enable"},
 323	{Opt_err, NULL},
 324};
 325
 326#define F2FS_SPEC_background_gc			(1 << 0)
 327#define F2FS_SPEC_inline_xattr_size		(1 << 1)
 328#define F2FS_SPEC_active_logs			(1 << 2)
 329#define F2FS_SPEC_reserve_root			(1 << 3)
 330#define F2FS_SPEC_resgid			(1 << 4)
 331#define F2FS_SPEC_resuid			(1 << 5)
 332#define F2FS_SPEC_mode				(1 << 6)
 333#define F2FS_SPEC_fault_injection		(1 << 7)
 334#define F2FS_SPEC_fault_type			(1 << 8)
 335#define F2FS_SPEC_jqfmt				(1 << 9)
 336#define F2FS_SPEC_alloc_mode			(1 << 10)
 337#define F2FS_SPEC_fsync_mode			(1 << 11)
 338#define F2FS_SPEC_checkpoint_disable_cap	(1 << 12)
 339#define F2FS_SPEC_checkpoint_disable_cap_perc	(1 << 13)
 340#define F2FS_SPEC_compress_level		(1 << 14)
 341#define F2FS_SPEC_compress_algorithm		(1 << 15)
 342#define F2FS_SPEC_compress_log_size		(1 << 16)
 343#define F2FS_SPEC_compress_extension		(1 << 17)
 344#define F2FS_SPEC_nocompress_extension		(1 << 18)
 345#define F2FS_SPEC_compress_chksum		(1 << 19)
 346#define F2FS_SPEC_compress_mode			(1 << 20)
 347#define F2FS_SPEC_discard_unit			(1 << 21)
 348#define F2FS_SPEC_memory_mode			(1 << 22)
 349#define F2FS_SPEC_errors			(1 << 23)
 350#define F2FS_SPEC_lookup_mode			(1 << 24)
 351#define F2FS_SPEC_reserve_node			(1 << 25)
 352
 353struct f2fs_fs_context {
 354	struct f2fs_mount_info info;
 355	unsigned long long opt_mask;	/* Bits changed */
 356	unsigned int	spec_mask;
 357	unsigned short	qname_mask;
 358};
 359
 360#define F2FS_CTX_INFO(ctx)	((ctx)->info)
 361
 362static inline void ctx_set_opt(struct f2fs_fs_context *ctx,
 363			       enum f2fs_mount_opt flag)
 364{
 365	ctx->info.opt |= BIT(flag);
 366	ctx->opt_mask |= BIT(flag);
 367}
 368
 369static inline void ctx_clear_opt(struct f2fs_fs_context *ctx,
 370				 enum f2fs_mount_opt flag)
 371{
 372	ctx->info.opt &= ~BIT(flag);
 373	ctx->opt_mask |= BIT(flag);
 374}
 375
 376static inline bool ctx_test_opt(struct f2fs_fs_context *ctx,
 377				enum f2fs_mount_opt flag)
 378{
 379	return ctx->info.opt & BIT(flag);
 380}
 381
 382void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate,
 383					const char *fmt, ...)
 384{
 385	struct va_format vaf;
 386	va_list args;
 387	int level;
 388
 389	va_start(args, fmt);
 390
 391	level = printk_get_level(fmt);
 392	vaf.fmt = printk_skip_level(fmt);
 393	vaf.va = &args;
 394	if (limit_rate)
 395		if (sbi)
 396			printk_ratelimited("%c%cF2FS-fs (%s): %pV\n",
 397				KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
 398		else
 399			printk_ratelimited("%c%cF2FS-fs: %pV\n",
 400				KERN_SOH_ASCII, level, &vaf);
 401	else
 402		if (sbi)
 403			printk("%c%cF2FS-fs (%s): %pV\n",
 404				KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
 405		else
 406			printk("%c%cF2FS-fs: %pV\n",
 407				KERN_SOH_ASCII, level, &vaf);
 408
 409	va_end(args);
 410}
 411
 412#if IS_ENABLED(CONFIG_UNICODE)
 413static const struct f2fs_sb_encodings {
 414	__u16 magic;
 415	char *name;
 416	unsigned int version;
 417} f2fs_sb_encoding_map[] = {
 418	{F2FS_ENC_UTF8_12_1, "utf8", UNICODE_AGE(12, 1, 0)},
 419};
 420
 421static const struct f2fs_sb_encodings *
 422f2fs_sb_read_encoding(const struct f2fs_super_block *sb)
 423{
 424	__u16 magic = le16_to_cpu(sb->s_encoding);
 425	int i;
 426
 427	for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
 428		if (magic == f2fs_sb_encoding_map[i].magic)
 429			return &f2fs_sb_encoding_map[i];
 430
 431	return NULL;
 432}
 433
 434struct kmem_cache *f2fs_cf_name_slab;
 435static int __init f2fs_create_casefold_cache(void)
 436{
 437	f2fs_cf_name_slab = f2fs_kmem_cache_create("f2fs_casefolded_name",
 438						   F2FS_NAME_LEN);
 439	return f2fs_cf_name_slab ? 0 : -ENOMEM;
 440}
 441
 442static void f2fs_destroy_casefold_cache(void)
 443{
 444	kmem_cache_destroy(f2fs_cf_name_slab);
 445}
 446#else
 447static int __init f2fs_create_casefold_cache(void) { return 0; }
 448static void f2fs_destroy_casefold_cache(void) { }
 449#endif
 450
 451static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
 452{
 453	block_t block_limit = min((sbi->user_block_count >> 3),
 454			sbi->user_block_count - sbi->reserved_blocks);
 455	block_t node_limit = sbi->total_node_count >> 3;
 456
 457	/* limit is 12.5% */
 458	if (test_opt(sbi, RESERVE_ROOT) &&
 459			F2FS_OPTION(sbi).root_reserved_blocks > block_limit) {
 460		F2FS_OPTION(sbi).root_reserved_blocks = block_limit;
 461		f2fs_info(sbi, "Reduce reserved blocks for root = %u",
 462			  F2FS_OPTION(sbi).root_reserved_blocks);
 463	}
 464	if (test_opt(sbi, RESERVE_NODE) &&
 465			F2FS_OPTION(sbi).root_reserved_nodes > node_limit) {
 466		F2FS_OPTION(sbi).root_reserved_nodes = node_limit;
 467		f2fs_info(sbi, "Reduce reserved nodes for root = %u",
 468			  F2FS_OPTION(sbi).root_reserved_nodes);
 469	}
 470	if (!test_opt(sbi, RESERVE_ROOT) && !test_opt(sbi, RESERVE_NODE) &&
 471		(!uid_eq(F2FS_OPTION(sbi).s_resuid,
 472				make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
 473		!gid_eq(F2FS_OPTION(sbi).s_resgid,
 474				make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
 475		f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root"
 476				" and reserve_node",
 477			  from_kuid_munged(&init_user_ns,
 478					   F2FS_OPTION(sbi).s_resuid),
 479			  from_kgid_munged(&init_user_ns,
 480					   F2FS_OPTION(sbi).s_resgid));
 481}
 482
 483static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
 484{
 485	if (!F2FS_OPTION(sbi).unusable_cap_perc)
 486		return;
 487
 488	if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
 489		F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
 490	else
 491		F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
 492					F2FS_OPTION(sbi).unusable_cap_perc;
 493
 494	f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
 495			F2FS_OPTION(sbi).unusable_cap,
 496			F2FS_OPTION(sbi).unusable_cap_perc);
 497}
 498
 499static void init_once(void *foo)
 500{
 501	struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
 502
 503	inode_init_once(&fi->vfs_inode);
 504#ifdef CONFIG_FS_ENCRYPTION
 505	fi->i_crypt_info = NULL;
 506#endif
 507#ifdef CONFIG_FS_VERITY
 508	fi->i_verity_info = NULL;
 509#endif
 510}
 511
 512#ifdef CONFIG_QUOTA
 513static const char * const quotatypes[] = INITQFNAMES;
 514#define QTYPE2NAME(t) (quotatypes[t])
 515/*
 516 * Note the name of the specified quota file.
 517 */
 518static int f2fs_note_qf_name(struct fs_context *fc, int qtype,
 519			     struct fs_parameter *param)
 520{
 521	struct f2fs_fs_context *ctx = fc->fs_private;
 522	char *qname;
 523
 524	if (param->size < 1) {
 525		f2fs_err(NULL, "Missing quota name");
 526		return -EINVAL;
 527	}
 528	if (strchr(param->string, '/')) {
 529		f2fs_err(NULL, "quotafile must be on filesystem root");
 530		return -EINVAL;
 531	}
 532	if (ctx->info.s_qf_names[qtype]) {
 533		if (strcmp(ctx->info.s_qf_names[qtype], param->string) != 0) {
 534			f2fs_err(NULL, "Quota file already specified");
 535			return -EINVAL;
 536		}
 537		return 0;
 538	}
 539
 540	qname = kmemdup_nul(param->string, param->size, GFP_KERNEL);
 541	if (!qname) {
 542		f2fs_err(NULL, "Not enough memory for storing quotafile name");
 543		return -ENOMEM;
 544	}
 545	F2FS_CTX_INFO(ctx).s_qf_names[qtype] = qname;
 546	ctx->qname_mask |= 1 << qtype;
 547	return 0;
 548}
 549
 550/*
 551 * Clear the name of the specified quota file.
 552 */
 553static int f2fs_unnote_qf_name(struct fs_context *fc, int qtype)
 554{
 555	struct f2fs_fs_context *ctx = fc->fs_private;
 556
 557	kfree(ctx->info.s_qf_names[qtype]);
 558	ctx->info.s_qf_names[qtype] = NULL;
 559	ctx->qname_mask |= 1 << qtype;
 560	return 0;
 561}
 562
 563static void f2fs_unnote_qf_name_all(struct fs_context *fc)
 564{
 565	int i;
 566
 567	for (i = 0; i < MAXQUOTAS; i++)
 568		f2fs_unnote_qf_name(fc, i);
 569}
 570#endif
 571
 572static int f2fs_parse_test_dummy_encryption(const struct fs_parameter *param,
 573					    struct f2fs_fs_context *ctx)
 574{
 575	int err;
 576
 577	if (!IS_ENABLED(CONFIG_FS_ENCRYPTION)) {
 578		f2fs_warn(NULL, "test_dummy_encryption option not supported");
 579		return -EINVAL;
 580	}
 581	err = fscrypt_parse_test_dummy_encryption(param,
 582					&ctx->info.dummy_enc_policy);
 583	if (err) {
 584		if (err == -EINVAL)
 585			f2fs_warn(NULL, "Value of option \"%s\" is unrecognized",
 586				  param->key);
 587		else if (err == -EEXIST)
 588			f2fs_warn(NULL, "Conflicting test_dummy_encryption options");
 589		else
 590			f2fs_warn(NULL, "Error processing option \"%s\" [%d]",
 591				  param->key, err);
 592		return -EINVAL;
 593	}
 594	return 0;
 595}
 596
 597#ifdef CONFIG_F2FS_FS_COMPRESSION
 598static bool is_compress_extension_exist(struct f2fs_mount_info *info,
 599					const char *new_ext, bool is_ext)
 600{
 601	unsigned char (*ext)[F2FS_EXTENSION_LEN];
 602	int ext_cnt;
 603	int i;
 604
 605	if (is_ext) {
 606		ext = info->extensions;
 607		ext_cnt = info->compress_ext_cnt;
 608	} else {
 609		ext = info->noextensions;
 610		ext_cnt = info->nocompress_ext_cnt;
 611	}
 612
 613	for (i = 0; i < ext_cnt; i++) {
 614		if (!strcasecmp(new_ext, ext[i]))
 615			return true;
 616	}
 617
 618	return false;
 619}
 620
 621/*
 622 * 1. The same extension name cannot not appear in both compress and non-compress extension
 623 * at the same time.
 624 * 2. If the compress extension specifies all files, the types specified by the non-compress
 625 * extension will be treated as special cases and will not be compressed.
 626 * 3. Don't allow the non-compress extension specifies all files.
 627 */
 628static int f2fs_test_compress_extension(unsigned char (*noext)[F2FS_EXTENSION_LEN],
 629					int noext_cnt,
 630					unsigned char (*ext)[F2FS_EXTENSION_LEN],
 631					int ext_cnt)
 632{
 633	int index = 0, no_index = 0;
 634
 635	if (!noext_cnt)
 636		return 0;
 637
 638	for (no_index = 0; no_index < noext_cnt; no_index++) {
 639		if (strlen(noext[no_index]) == 0)
 640			continue;
 641		if (!strcasecmp("*", noext[no_index])) {
 642			f2fs_info(NULL, "Don't allow the nocompress extension specifies all files");
 643			return -EINVAL;
 644		}
 645		for (index = 0; index < ext_cnt; index++) {
 646			if (strlen(ext[index]) == 0)
 647				continue;
 648			if (!strcasecmp(ext[index], noext[no_index])) {
 649				f2fs_info(NULL, "Don't allow the same extension %s appear in both compress and nocompress extension",
 650						ext[index]);
 651				return -EINVAL;
 652			}
 653		}
 654	}
 655	return 0;
 656}
 657
 658#ifdef CONFIG_F2FS_FS_LZ4
 659static int f2fs_set_lz4hc_level(struct f2fs_fs_context *ctx, const char *str)
 660{
 661#ifdef CONFIG_F2FS_FS_LZ4HC
 662	unsigned int level;
 663
 664	if (strlen(str) == 3) {
 665		F2FS_CTX_INFO(ctx).compress_level = 0;
 666		ctx->spec_mask |= F2FS_SPEC_compress_level;
 667		return 0;
 668	}
 669
 670	str += 3;
 671
 672	if (str[0] != ':') {
 673		f2fs_info(NULL, "wrong format, e.g. <alg_name>:<compr_level>");
 674		return -EINVAL;
 675	}
 676	if (kstrtouint(str + 1, 10, &level))
 677		return -EINVAL;
 678
 679	if (!f2fs_is_compress_level_valid(COMPRESS_LZ4, level)) {
 680		f2fs_info(NULL, "invalid lz4hc compress level: %d", level);
 681		return -EINVAL;
 682	}
 683
 684	F2FS_CTX_INFO(ctx).compress_level = level;
 685	ctx->spec_mask |= F2FS_SPEC_compress_level;
 686	return 0;
 687#else
 688	if (strlen(str) == 3) {
 689		F2FS_CTX_INFO(ctx).compress_level = 0;
 690		ctx->spec_mask |= F2FS_SPEC_compress_level;
 691		return 0;
 692	}
 693	f2fs_info(NULL, "kernel doesn't support lz4hc compression");
 694	return -EINVAL;
 695#endif
 696}
 697#endif
 698
 699#ifdef CONFIG_F2FS_FS_ZSTD
 700static int f2fs_set_zstd_level(struct f2fs_fs_context *ctx, const char *str)
 701{
 702	int level;
 703	int len = 4;
 704
 705	if (strlen(str) == len) {
 706		F2FS_CTX_INFO(ctx).compress_level = F2FS_ZSTD_DEFAULT_CLEVEL;
 707		ctx->spec_mask |= F2FS_SPEC_compress_level;
 708		return 0;
 709	}
 710
 711	str += len;
 712
 713	if (str[0] != ':') {
 714		f2fs_info(NULL, "wrong format, e.g. <alg_name>:<compr_level>");
 715		return -EINVAL;
 716	}
 717	if (kstrtoint(str + 1, 10, &level))
 718		return -EINVAL;
 719
 720	/* f2fs does not support negative compress level now */
 721	if (level < 0) {
 722		f2fs_info(NULL, "do not support negative compress level: %d", level);
 723		return -ERANGE;
 724	}
 725
 726	if (!f2fs_is_compress_level_valid(COMPRESS_ZSTD, level)) {
 727		f2fs_info(NULL, "invalid zstd compress level: %d", level);
 728		return -EINVAL;
 729	}
 730
 731	F2FS_CTX_INFO(ctx).compress_level = level;
 732	ctx->spec_mask |= F2FS_SPEC_compress_level;
 733	return 0;
 734}
 735#endif
 736#endif
 737
 738static int f2fs_parse_param(struct fs_context *fc, struct fs_parameter *param)
 739{
 740	struct f2fs_fs_context *ctx = fc->fs_private;
 741#ifdef CONFIG_F2FS_FS_COMPRESSION
 742	unsigned char (*ext)[F2FS_EXTENSION_LEN];
 743	unsigned char (*noext)[F2FS_EXTENSION_LEN];
 744	int ext_cnt, noext_cnt;
 745	char *name;
 746#endif
 747	substring_t args[MAX_OPT_ARGS];
 748	struct fs_parse_result result;
 749	int token, ret, arg;
 750
 751	token = fs_parse(fc, f2fs_param_specs, param, &result);
 752	if (token < 0)
 753		return token;
 754
 755	switch (token) {
 756	case Opt_gc_background:
 757		F2FS_CTX_INFO(ctx).bggc_mode = result.uint_32;
 758		ctx->spec_mask |= F2FS_SPEC_background_gc;
 759		break;
 760	case Opt_disable_roll_forward:
 761		ctx_set_opt(ctx, F2FS_MOUNT_DISABLE_ROLL_FORWARD);
 762		break;
 763	case Opt_norecovery:
 764		/* requires ro mount, checked in f2fs_validate_options */
 765		ctx_set_opt(ctx, F2FS_MOUNT_NORECOVERY);
 766		break;
 767	case Opt_discard:
 768		if (result.negated)
 769			ctx_clear_opt(ctx, F2FS_MOUNT_DISCARD);
 770		else
 771			ctx_set_opt(ctx, F2FS_MOUNT_DISCARD);
 772		break;
 773	case Opt_noheap:
 774	case Opt_heap:
 775		f2fs_warn(NULL, "heap/no_heap options were deprecated");
 776		break;
 777#ifdef CONFIG_F2FS_FS_XATTR
 778	case Opt_user_xattr:
 779		if (result.negated)
 780			ctx_clear_opt(ctx, F2FS_MOUNT_XATTR_USER);
 781		else
 782			ctx_set_opt(ctx, F2FS_MOUNT_XATTR_USER);
 783		break;
 784	case Opt_inline_xattr:
 785		if (result.negated)
 786			ctx_clear_opt(ctx, F2FS_MOUNT_INLINE_XATTR);
 787		else
 788			ctx_set_opt(ctx, F2FS_MOUNT_INLINE_XATTR);
 789		break;
 790	case Opt_inline_xattr_size:
 791		if (result.int_32 < MIN_INLINE_XATTR_SIZE ||
 792			result.int_32 > MAX_INLINE_XATTR_SIZE) {
 793			f2fs_err(NULL, "inline xattr size is out of range: %u ~ %u",
 794				 (u32)MIN_INLINE_XATTR_SIZE, (u32)MAX_INLINE_XATTR_SIZE);
 795			return -EINVAL;
 796		}
 797		ctx_set_opt(ctx, F2FS_MOUNT_INLINE_XATTR_SIZE);
 798		F2FS_CTX_INFO(ctx).inline_xattr_size = result.int_32;
 799		ctx->spec_mask |= F2FS_SPEC_inline_xattr_size;
 800		break;
 801#else
 802	case Opt_user_xattr:
 803	case Opt_inline_xattr:
 804	case Opt_inline_xattr_size:
 805		f2fs_info(NULL, "%s options not supported", param->key);
 806		break;
 807#endif
 808#ifdef CONFIG_F2FS_FS_POSIX_ACL
 809	case Opt_acl:
 810		if (result.negated)
 811			ctx_clear_opt(ctx, F2FS_MOUNT_POSIX_ACL);
 812		else
 813			ctx_set_opt(ctx, F2FS_MOUNT_POSIX_ACL);
 814		break;
 815#else
 816	case Opt_acl:
 817		f2fs_info(NULL, "%s options not supported", param->key);
 818		break;
 819#endif
 820	case Opt_active_logs:
 821		if (result.int_32 != 2 && result.int_32 != 4 &&
 822			result.int_32 != NR_CURSEG_PERSIST_TYPE)
 823			return -EINVAL;
 824		ctx->spec_mask |= F2FS_SPEC_active_logs;
 825		F2FS_CTX_INFO(ctx).active_logs = result.int_32;
 826		break;
 827	case Opt_disable_ext_identify:
 828		ctx_set_opt(ctx, F2FS_MOUNT_DISABLE_EXT_IDENTIFY);
 829		break;
 830	case Opt_inline_data:
 831		if (result.negated)
 832			ctx_clear_opt(ctx, F2FS_MOUNT_INLINE_DATA);
 833		else
 834			ctx_set_opt(ctx, F2FS_MOUNT_INLINE_DATA);
 835		break;
 836	case Opt_inline_dentry:
 837		if (result.negated)
 838			ctx_clear_opt(ctx, F2FS_MOUNT_INLINE_DENTRY);
 839		else
 840			ctx_set_opt(ctx, F2FS_MOUNT_INLINE_DENTRY);
 841		break;
 842	case Opt_flush_merge:
 843		if (result.negated)
 844			ctx_clear_opt(ctx, F2FS_MOUNT_FLUSH_MERGE);
 845		else
 846			ctx_set_opt(ctx, F2FS_MOUNT_FLUSH_MERGE);
 847		break;
 848	case Opt_barrier:
 849		if (result.negated)
 850			ctx_set_opt(ctx, F2FS_MOUNT_NOBARRIER);
 851		else
 852			ctx_clear_opt(ctx, F2FS_MOUNT_NOBARRIER);
 853		break;
 854	case Opt_fastboot:
 855		ctx_set_opt(ctx, F2FS_MOUNT_FASTBOOT);
 856		break;
 857	case Opt_extent_cache:
 858		if (result.negated)
 859			ctx_clear_opt(ctx, F2FS_MOUNT_READ_EXTENT_CACHE);
 860		else
 861			ctx_set_opt(ctx, F2FS_MOUNT_READ_EXTENT_CACHE);
 862		break;
 863	case Opt_data_flush:
 864		ctx_set_opt(ctx, F2FS_MOUNT_DATA_FLUSH);
 865		break;
 866	case Opt_reserve_root:
 867		ctx_set_opt(ctx, F2FS_MOUNT_RESERVE_ROOT);
 868		F2FS_CTX_INFO(ctx).root_reserved_blocks = result.uint_32;
 869		ctx->spec_mask |= F2FS_SPEC_reserve_root;
 870		break;
 871	case Opt_reserve_node:
 872		ctx_set_opt(ctx, F2FS_MOUNT_RESERVE_NODE);
 873		F2FS_CTX_INFO(ctx).root_reserved_nodes = result.uint_32;
 874		ctx->spec_mask |= F2FS_SPEC_reserve_node;
 875		break;
 876	case Opt_resuid:
 877		F2FS_CTX_INFO(ctx).s_resuid = result.uid;
 878		ctx->spec_mask |= F2FS_SPEC_resuid;
 879		break;
 880	case Opt_resgid:
 881		F2FS_CTX_INFO(ctx).s_resgid = result.gid;
 882		ctx->spec_mask |= F2FS_SPEC_resgid;
 883		break;
 884	case Opt_mode:
 885		F2FS_CTX_INFO(ctx).fs_mode = result.uint_32;
 886		ctx->spec_mask |= F2FS_SPEC_mode;
 887		break;
 888#ifdef CONFIG_F2FS_FAULT_INJECTION
 889	case Opt_fault_injection:
 890		F2FS_CTX_INFO(ctx).fault_info.inject_rate = result.int_32;
 891		ctx->spec_mask |= F2FS_SPEC_fault_injection;
 892		ctx_set_opt(ctx, F2FS_MOUNT_FAULT_INJECTION);
 893		break;
 894
 895	case Opt_fault_type:
 896		if (result.uint_32 > BIT(FAULT_MAX))
 897			return -EINVAL;
 898		F2FS_CTX_INFO(ctx).fault_info.inject_type = result.uint_32;
 899		ctx->spec_mask |= F2FS_SPEC_fault_type;
 900		ctx_set_opt(ctx, F2FS_MOUNT_FAULT_INJECTION);
 901		break;
 902#else
 903	case Opt_fault_injection:
 904	case Opt_fault_type:
 905		f2fs_info(NULL, "%s options not supported", param->key);
 906		break;
 907#endif
 908	case Opt_lazytime:
 909		if (result.negated)
 910			ctx_clear_opt(ctx, F2FS_MOUNT_LAZYTIME);
 911		else
 912			ctx_set_opt(ctx, F2FS_MOUNT_LAZYTIME);
 913		break;
 914#ifdef CONFIG_QUOTA
 915	case Opt_quota:
 916		if (result.negated) {
 917			ctx_clear_opt(ctx, F2FS_MOUNT_QUOTA);
 918			ctx_clear_opt(ctx, F2FS_MOUNT_USRQUOTA);
 919			ctx_clear_opt(ctx, F2FS_MOUNT_GRPQUOTA);
 920			ctx_clear_opt(ctx, F2FS_MOUNT_PRJQUOTA);
 921		} else
 922			ctx_set_opt(ctx, F2FS_MOUNT_USRQUOTA);
 923		break;
 924	case Opt_usrquota:
 925		ctx_set_opt(ctx, F2FS_MOUNT_USRQUOTA);
 926		break;
 927	case Opt_grpquota:
 928		ctx_set_opt(ctx, F2FS_MOUNT_GRPQUOTA);
 929		break;
 930	case Opt_prjquota:
 931		ctx_set_opt(ctx, F2FS_MOUNT_PRJQUOTA);
 932		break;
 933	case Opt_usrjquota:
 934		if (!*param->string)
 935			ret = f2fs_unnote_qf_name(fc, USRQUOTA);
 936		else
 937			ret = f2fs_note_qf_name(fc, USRQUOTA, param);
 938		if (ret)
 939			return ret;
 940		break;
 941	case Opt_grpjquota:
 942		if (!*param->string)
 943			ret = f2fs_unnote_qf_name(fc, GRPQUOTA);
 944		else
 945			ret = f2fs_note_qf_name(fc, GRPQUOTA, param);
 946		if (ret)
 947			return ret;
 948		break;
 949	case Opt_prjjquota:
 950		if (!*param->string)
 951			ret = f2fs_unnote_qf_name(fc, PRJQUOTA);
 952		else
 953			ret = f2fs_note_qf_name(fc, PRJQUOTA, param);
 954		if (ret)
 955			return ret;
 956		break;
 957	case Opt_jqfmt:
 958		F2FS_CTX_INFO(ctx).s_jquota_fmt = result.int_32;
 959		ctx->spec_mask |= F2FS_SPEC_jqfmt;
 960		break;
 961#else
 962	case Opt_quota:
 963	case Opt_usrquota:
 964	case Opt_grpquota:
 965	case Opt_prjquota:
 966	case Opt_usrjquota:
 967	case Opt_grpjquota:
 968	case Opt_prjjquota:
 969		f2fs_info(NULL, "quota operations not supported");
 970		break;
 971#endif
 972	case Opt_alloc:
 973		F2FS_CTX_INFO(ctx).alloc_mode = result.uint_32;
 974		ctx->spec_mask |= F2FS_SPEC_alloc_mode;
 975		break;
 976	case Opt_fsync:
 977		F2FS_CTX_INFO(ctx).fsync_mode = result.uint_32;
 978		ctx->spec_mask |= F2FS_SPEC_fsync_mode;
 979		break;
 980	case Opt_test_dummy_encryption:
 981		ret = f2fs_parse_test_dummy_encryption(param, ctx);
 982		if (ret)
 983			return ret;
 984		break;
 985	case Opt_inlinecrypt:
 986#ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
 987		ctx_set_opt(ctx, F2FS_MOUNT_INLINECRYPT);
 988#else
 989		f2fs_info(NULL, "inline encryption not supported");
 990#endif
 991		break;
 992	case Opt_checkpoint:
 993		/*
 994		 * Initialize args struct so we know whether arg was
 995		 * found; some options take optional arguments.
 996		 */
 997		args[0].from = args[0].to = NULL;
 998		arg = 0;
 999
1000		/* revert to match_table for checkpoint= options */
1001		token = match_token(param->string, f2fs_checkpoint_tokens, args);
1002		switch (token) {
1003		case Opt_checkpoint_disable_cap_perc:
1004			if (args->from && match_int(args, &arg))
1005				return -EINVAL;
1006			if (arg < 0 || arg > 100)
1007				return -EINVAL;
1008			F2FS_CTX_INFO(ctx).unusable_cap_perc = arg;
1009			ctx->spec_mask |= F2FS_SPEC_checkpoint_disable_cap_perc;
1010			ctx_set_opt(ctx, F2FS_MOUNT_DISABLE_CHECKPOINT);
1011			break;
1012		case Opt_checkpoint_disable_cap:
1013			if (args->from && match_int(args, &arg))
1014				return -EINVAL;
1015			F2FS_CTX_INFO(ctx).unusable_cap = arg;
1016			ctx->spec_mask |= F2FS_SPEC_checkpoint_disable_cap;
1017			ctx_set_opt(ctx, F2FS_MOUNT_DISABLE_CHECKPOINT);
1018			break;
1019		case Opt_checkpoint_disable:
1020			ctx_set_opt(ctx, F2FS_MOUNT_DISABLE_CHECKPOINT);
1021			break;
1022		case Opt_checkpoint_enable:
1023			F2FS_CTX_INFO(ctx).unusable_cap_perc = 0;
1024			ctx->spec_mask |= F2FS_SPEC_checkpoint_disable_cap_perc;
1025			F2FS_CTX_INFO(ctx).unusable_cap = 0;
1026			ctx->spec_mask |= F2FS_SPEC_checkpoint_disable_cap;
1027			ctx_clear_opt(ctx, F2FS_MOUNT_DISABLE_CHECKPOINT);
1028			break;
1029		default:
1030			return -EINVAL;
1031		}
1032		break;
1033	case Opt_checkpoint_merge:
1034		if (result.negated)
1035			ctx_clear_opt(ctx, F2FS_MOUNT_MERGE_CHECKPOINT);
1036		else
1037			ctx_set_opt(ctx, F2FS_MOUNT_MERGE_CHECKPOINT);
1038		break;
1039#ifdef CONFIG_F2FS_FS_COMPRESSION
1040	case Opt_compress_algorithm:
1041		name = param->string;
1042		if (!strcmp(name, "lzo")) {
1043#ifdef CONFIG_F2FS_FS_LZO
1044			F2FS_CTX_INFO(ctx).compress_level = 0;
1045			F2FS_CTX_INFO(ctx).compress_algorithm = COMPRESS_LZO;
1046			ctx->spec_mask |= F2FS_SPEC_compress_level;
1047			ctx->spec_mask |= F2FS_SPEC_compress_algorithm;
1048#else
1049			f2fs_info(NULL, "kernel doesn't support lzo compression");
1050#endif
1051		} else if (!strncmp(name, "lz4", 3)) {
1052#ifdef CONFIG_F2FS_FS_LZ4
1053			ret = f2fs_set_lz4hc_level(ctx, name);
1054			if (ret)
1055				return -EINVAL;
1056			F2FS_CTX_INFO(ctx).compress_algorithm = COMPRESS_LZ4;
1057			ctx->spec_mask |= F2FS_SPEC_compress_algorithm;
1058#else
1059			f2fs_info(NULL, "kernel doesn't support lz4 compression");
1060#endif
1061		} else if (!strncmp(name, "zstd", 4)) {
1062#ifdef CONFIG_F2FS_FS_ZSTD
1063			ret = f2fs_set_zstd_level(ctx, name);
1064			if (ret)
1065				return -EINVAL;
1066			F2FS_CTX_INFO(ctx).compress_algorithm = COMPRESS_ZSTD;
1067			ctx->spec_mask |= F2FS_SPEC_compress_algorithm;
1068#else
1069			f2fs_info(NULL, "kernel doesn't support zstd compression");
1070#endif
1071		} else if (!strcmp(name, "lzo-rle")) {
1072#ifdef CONFIG_F2FS_FS_LZORLE
1073			F2FS_CTX_INFO(ctx).compress_level = 0;
1074			F2FS_CTX_INFO(ctx).compress_algorithm = COMPRESS_LZORLE;
1075			ctx->spec_mask |= F2FS_SPEC_compress_level;
1076			ctx->spec_mask |= F2FS_SPEC_compress_algorithm;
1077#else
1078			f2fs_info(NULL, "kernel doesn't support lzorle compression");
1079#endif
1080		} else
1081			return -EINVAL;
1082		break;
1083	case Opt_compress_log_size:
1084		if (result.uint_32 < MIN_COMPRESS_LOG_SIZE ||
1085		    result.uint_32 > MAX_COMPRESS_LOG_SIZE) {
1086			f2fs_err(NULL,
1087				"Compress cluster log size is out of range");
1088			return -EINVAL;
1089		}
1090		F2FS_CTX_INFO(ctx).compress_log_size = result.uint_32;
1091		ctx->spec_mask |= F2FS_SPEC_compress_log_size;
1092		break;
1093	case Opt_compress_extension:
1094		name = param->string;
1095		ext = F2FS_CTX_INFO(ctx).extensions;
1096		ext_cnt = F2FS_CTX_INFO(ctx).compress_ext_cnt;
1097
1098		if (strlen(name) >= F2FS_EXTENSION_LEN ||
1099		    ext_cnt >= COMPRESS_EXT_NUM) {
1100			f2fs_err(NULL, "invalid extension length/number");
1101			return -EINVAL;
1102		}
1103
1104		if (is_compress_extension_exist(&ctx->info, name, true))
1105			break;
1106
1107		ret = strscpy(ext[ext_cnt], name, F2FS_EXTENSION_LEN);
1108		if (ret < 0)
1109			return ret;
1110		F2FS_CTX_INFO(ctx).compress_ext_cnt++;
1111		ctx->spec_mask |= F2FS_SPEC_compress_extension;
1112		break;
1113	case Opt_nocompress_extension:
1114		name = param->string;
1115		noext = F2FS_CTX_INFO(ctx).noextensions;
1116		noext_cnt = F2FS_CTX_INFO(ctx).nocompress_ext_cnt;
1117
1118		if (strlen(name) >= F2FS_EXTENSION_LEN ||
1119			noext_cnt >= COMPRESS_EXT_NUM) {
1120			f2fs_err(NULL, "invalid extension length/number");
1121			return -EINVAL;
1122		}
1123
1124		if (is_compress_extension_exist(&ctx->info, name, false))
1125			break;
1126
1127		ret = strscpy(noext[noext_cnt], name, F2FS_EXTENSION_LEN);
1128		if (ret < 0)
1129			return ret;
1130		F2FS_CTX_INFO(ctx).nocompress_ext_cnt++;
1131		ctx->spec_mask |= F2FS_SPEC_nocompress_extension;
1132		break;
1133	case Opt_compress_chksum:
1134		F2FS_CTX_INFO(ctx).compress_chksum = true;
1135		ctx->spec_mask |= F2FS_SPEC_compress_chksum;
1136		break;
1137	case Opt_compress_mode:
1138		F2FS_CTX_INFO(ctx).compress_mode = result.uint_32;
1139		ctx->spec_mask |= F2FS_SPEC_compress_mode;
1140		break;
1141	case Opt_compress_cache:
1142		ctx_set_opt(ctx, F2FS_MOUNT_COMPRESS_CACHE);
1143		break;
1144#else
1145	case Opt_compress_algorithm:
1146	case Opt_compress_log_size:
1147	case Opt_compress_extension:
1148	case Opt_nocompress_extension:
1149	case Opt_compress_chksum:
1150	case Opt_compress_mode:
1151	case Opt_compress_cache:
1152		f2fs_info(NULL, "compression options not supported");
1153		break;
1154#endif
1155	case Opt_atgc:
1156		ctx_set_opt(ctx, F2FS_MOUNT_ATGC);
1157		break;
1158	case Opt_gc_merge:
1159		if (result.negated)
1160			ctx_clear_opt(ctx, F2FS_MOUNT_GC_MERGE);
1161		else
1162			ctx_set_opt(ctx, F2FS_MOUNT_GC_MERGE);
1163		break;
1164	case Opt_discard_unit:
1165		F2FS_CTX_INFO(ctx).discard_unit = result.uint_32;
1166		ctx->spec_mask |= F2FS_SPEC_discard_unit;
1167		break;
1168	case Opt_memory_mode:
1169		F2FS_CTX_INFO(ctx).memory_mode = result.uint_32;
1170		ctx->spec_mask |= F2FS_SPEC_memory_mode;
1171		break;
1172	case Opt_age_extent_cache:
1173		ctx_set_opt(ctx, F2FS_MOUNT_AGE_EXTENT_CACHE);
1174		break;
1175	case Opt_errors:
1176		F2FS_CTX_INFO(ctx).errors = result.uint_32;
1177		ctx->spec_mask |= F2FS_SPEC_errors;
1178		break;
1179	case Opt_nat_bits:
1180		ctx_set_opt(ctx, F2FS_MOUNT_NAT_BITS);
1181		break;
1182	case Opt_lookup_mode:
1183		F2FS_CTX_INFO(ctx).lookup_mode = result.uint_32;
1184		ctx->spec_mask |= F2FS_SPEC_lookup_mode;
1185		break;
1186	}
1187	return 0;
1188}
1189
1190/*
1191 * Check quota settings consistency.
1192 */
1193static int f2fs_check_quota_consistency(struct fs_context *fc,
1194					struct super_block *sb)
1195{
1196	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1197 #ifdef CONFIG_QUOTA
1198	struct f2fs_fs_context *ctx = fc->fs_private;
1199	bool quota_feature = f2fs_sb_has_quota_ino(sbi);
1200	bool quota_turnon = sb_any_quota_loaded(sb);
1201	char *old_qname, *new_qname;
1202	bool usr_qf_name, grp_qf_name, prj_qf_name, usrquota, grpquota, prjquota;
1203	int i;
1204
1205	/*
1206	 * We do the test below only for project quotas. 'usrquota' and
1207	 * 'grpquota' mount options are allowed even without quota feature
1208	 * to support legacy quotas in quota files.
1209	 */
1210	if (ctx_test_opt(ctx, F2FS_MOUNT_PRJQUOTA) &&
1211			!f2fs_sb_has_project_quota(sbi)) {
1212		f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
1213		return -EINVAL;
1214	}
1215
1216	if (ctx->qname_mask) {
1217		for (i = 0; i < MAXQUOTAS; i++) {
1218			if (!(ctx->qname_mask & (1 << i)))
1219				continue;
1220
1221			old_qname = F2FS_OPTION(sbi).s_qf_names[i];
1222			new_qname = F2FS_CTX_INFO(ctx).s_qf_names[i];
1223			if (quota_turnon &&
1224				!!old_qname != !!new_qname)
1225				goto err_jquota_change;
1226
1227			if (old_qname) {
1228				if (!new_qname) {
1229					f2fs_info(sbi, "remove qf_name %s",
1230								old_qname);
1231					continue;
1232				} else if (strcmp(old_qname, new_qname) == 0) {
1233					ctx->qname_mask &= ~(1 << i);
1234					continue;
1235				}
1236				goto err_jquota_specified;
1237			}
1238
1239			if (quota_feature) {
1240				f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
1241				ctx->qname_mask &= ~(1 << i);
1242				kfree(F2FS_CTX_INFO(ctx).s_qf_names[i]);
1243				F2FS_CTX_INFO(ctx).s_qf_names[i] = NULL;
1244			}
1245		}
1246	}
1247
1248	/* Make sure we don't mix old and new quota format */
1249	usr_qf_name = F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
1250			F2FS_CTX_INFO(ctx).s_qf_names[USRQUOTA];
1251	grp_qf_name = F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
1252			F2FS_CTX_INFO(ctx).s_qf_names[GRPQUOTA];
1253	prj_qf_name = F2FS_OPTION(sbi).s_qf_names[PRJQUOTA] ||
1254			F2FS_CTX_INFO(ctx).s_qf_names[PRJQUOTA];
1255	usrquota = test_opt(sbi, USRQUOTA) ||
1256			ctx_test_opt(ctx, F2FS_MOUNT_USRQUOTA);
1257	grpquota = test_opt(sbi, GRPQUOTA) ||
1258			ctx_test_opt(ctx, F2FS_MOUNT_GRPQUOTA);
1259	prjquota = test_opt(sbi, PRJQUOTA) ||
1260			ctx_test_opt(ctx, F2FS_MOUNT_PRJQUOTA);
1261
1262	if (usr_qf_name) {
1263		ctx_clear_opt(ctx, F2FS_MOUNT_USRQUOTA);
1264		usrquota = false;
1265	}
1266	if (grp_qf_name) {
1267		ctx_clear_opt(ctx, F2FS_MOUNT_GRPQUOTA);
1268		grpquota = false;
1269	}
1270	if (prj_qf_name) {
1271		ctx_clear_opt(ctx, F2FS_MOUNT_PRJQUOTA);
1272		prjquota = false;
1273	}
1274	if (usr_qf_name || grp_qf_name || prj_qf_name) {
1275		if (grpquota || usrquota || prjquota) {
1276			f2fs_err(sbi, "old and new quota format mixing");
1277			return -EINVAL;
1278		}
1279		if (!(ctx->spec_mask & F2FS_SPEC_jqfmt ||
1280				F2FS_OPTION(sbi).s_jquota_fmt)) {
1281			f2fs_err(sbi, "journaled quota format not specified");
1282			return -EINVAL;
1283		}
1284	}
1285	return 0;
1286
1287err_jquota_change:
1288	f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
1289	return -EINVAL;
1290err_jquota_specified:
1291	f2fs_err(sbi, "%s quota file already specified",
1292		 QTYPE2NAME(i));
1293	return -EINVAL;
1294
1295#else
1296	if (f2fs_readonly(sbi->sb))
1297		return 0;
1298	if (f2fs_sb_has_quota_ino(sbi)) {
1299		f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1300		return -EINVAL;
1301	}
1302	if (f2fs_sb_has_project_quota(sbi)) {
1303		f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1304		return -EINVAL;
1305	}
1306
1307	return 0;
1308#endif
1309}
1310
1311static int f2fs_check_test_dummy_encryption(struct fs_context *fc,
1312					    struct super_block *sb)
1313{
1314	struct f2fs_fs_context *ctx = fc->fs_private;
1315	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1316
1317	if (!fscrypt_is_dummy_policy_set(&F2FS_CTX_INFO(ctx).dummy_enc_policy))
1318		return 0;
1319
1320	if (!f2fs_sb_has_encrypt(sbi)) {
1321		f2fs_err(sbi, "Encrypt feature is off");
1322		return -EINVAL;
1323	}
1324
1325	/*
1326	 * This mount option is just for testing, and it's not worthwhile to
1327	 * implement the extra complexity (e.g. RCU protection) that would be
1328	 * needed to allow it to be set or changed during remount.  We do allow
1329	 * it to be specified during remount, but only if there is no change.
1330	 */
1331	if (fc->purpose == FS_CONTEXT_FOR_RECONFIGURE) {
1332		if (fscrypt_dummy_policies_equal(&F2FS_OPTION(sbi).dummy_enc_policy,
1333				&F2FS_CTX_INFO(ctx).dummy_enc_policy))
1334			return 0;
1335		f2fs_warn(sbi, "Can't set or change test_dummy_encryption on remount");
1336		return -EINVAL;
1337	}
1338	return 0;
1339}
1340
1341static inline bool test_compression_spec(unsigned int mask)
1342{
1343	return mask & (F2FS_SPEC_compress_algorithm
1344			| F2FS_SPEC_compress_log_size
1345			| F2FS_SPEC_compress_extension
1346			| F2FS_SPEC_nocompress_extension
1347			| F2FS_SPEC_compress_chksum
1348			| F2FS_SPEC_compress_mode);
1349}
1350
1351static inline void clear_compression_spec(struct f2fs_fs_context *ctx)
1352{
1353	ctx->spec_mask &= ~(F2FS_SPEC_compress_algorithm
1354						| F2FS_SPEC_compress_log_size
1355						| F2FS_SPEC_compress_extension
1356						| F2FS_SPEC_nocompress_extension
1357						| F2FS_SPEC_compress_chksum
1358						| F2FS_SPEC_compress_mode);
1359}
1360
1361static int f2fs_check_compression(struct fs_context *fc,
1362				  struct super_block *sb)
1363{
1364#ifdef CONFIG_F2FS_FS_COMPRESSION
1365	struct f2fs_fs_context *ctx = fc->fs_private;
1366	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1367	int i, cnt;
1368
1369	if (!f2fs_sb_has_compression(sbi)) {
1370		if (test_compression_spec(ctx->spec_mask) ||
1371			ctx_test_opt(ctx, F2FS_MOUNT_COMPRESS_CACHE))
1372			f2fs_info(sbi, "Image doesn't support compression");
1373		clear_compression_spec(ctx);
1374		ctx->opt_mask &= ~BIT(F2FS_MOUNT_COMPRESS_CACHE);
1375		return 0;
1376	}
1377	if (ctx->spec_mask & F2FS_SPEC_compress_extension) {
1378		cnt = F2FS_CTX_INFO(ctx).compress_ext_cnt;
1379		for (i = 0; i < F2FS_CTX_INFO(ctx).compress_ext_cnt; i++) {
1380			if (is_compress_extension_exist(&F2FS_OPTION(sbi),
1381					F2FS_CTX_INFO(ctx).extensions[i], true)) {
1382				F2FS_CTX_INFO(ctx).extensions[i][0] = '\0';
1383				cnt--;
1384			}
1385		}
1386		if (F2FS_OPTION(sbi).compress_ext_cnt + cnt > COMPRESS_EXT_NUM) {
1387			f2fs_err(sbi, "invalid extension length/number");
1388			return -EINVAL;
1389		}
1390	}
1391	if (ctx->spec_mask & F2FS_SPEC_nocompress_extension) {
1392		cnt = F2FS_CTX_INFO(ctx).nocompress_ext_cnt;
1393		for (i = 0; i < F2FS_CTX_INFO(ctx).nocompress_ext_cnt; i++) {
1394			if (is_compress_extension_exist(&F2FS_OPTION(sbi),
1395					F2FS_CTX_INFO(ctx).noextensions[i], false)) {
1396				F2FS_CTX_INFO(ctx).noextensions[i][0] = '\0';
1397				cnt--;
1398			}
1399		}
1400		if (F2FS_OPTION(sbi).nocompress_ext_cnt + cnt > COMPRESS_EXT_NUM) {
1401			f2fs_err(sbi, "invalid noextension length/number");
1402			return -EINVAL;
1403		}
1404	}
1405
1406	if (f2fs_test_compress_extension(F2FS_CTX_INFO(ctx).noextensions,
1407				F2FS_CTX_INFO(ctx).nocompress_ext_cnt,
1408				F2FS_CTX_INFO(ctx).extensions,
1409				F2FS_CTX_INFO(ctx).compress_ext_cnt)) {
1410		f2fs_err(sbi, "new noextensions conflicts with new extensions");
1411		return -EINVAL;
1412	}
1413	if (f2fs_test_compress_extension(F2FS_CTX_INFO(ctx).noextensions,
1414				F2FS_CTX_INFO(ctx).nocompress_ext_cnt,
1415				F2FS_OPTION(sbi).extensions,
1416				F2FS_OPTION(sbi).compress_ext_cnt)) {
1417		f2fs_err(sbi, "new noextensions conflicts with old extensions");
1418		return -EINVAL;
1419	}
1420	if (f2fs_test_compress_extension(F2FS_OPTION(sbi).noextensions,
1421				F2FS_OPTION(sbi).nocompress_ext_cnt,
1422				F2FS_CTX_INFO(ctx).extensions,
1423				F2FS_CTX_INFO(ctx).compress_ext_cnt)) {
1424		f2fs_err(sbi, "new extensions conflicts with old noextensions");
1425		return -EINVAL;
1426	}
1427#endif
1428	return 0;
1429}
1430
1431static int f2fs_check_opt_consistency(struct fs_context *fc,
1432				      struct super_block *sb)
1433{
1434	struct f2fs_fs_context *ctx = fc->fs_private;
1435	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1436	int err;
1437
1438	if (ctx_test_opt(ctx, F2FS_MOUNT_NORECOVERY) && !f2fs_readonly(sb))
1439		return -EINVAL;
1440
1441	if (f2fs_hw_should_discard(sbi) &&
1442			(ctx->opt_mask & BIT(F2FS_MOUNT_DISCARD)) &&
1443			!ctx_test_opt(ctx, F2FS_MOUNT_DISCARD)) {
1444		f2fs_warn(sbi, "discard is required for zoned block devices");
1445		return -EINVAL;
1446	}
1447
1448	if (!f2fs_hw_support_discard(sbi) &&
1449			(ctx->opt_mask & BIT(F2FS_MOUNT_DISCARD)) &&
1450			ctx_test_opt(ctx, F2FS_MOUNT_DISCARD)) {
1451		f2fs_warn(sbi, "device does not support discard");
1452		ctx_clear_opt(ctx, F2FS_MOUNT_DISCARD);
1453		ctx->opt_mask &= ~BIT(F2FS_MOUNT_DISCARD);
1454	}
1455
1456	if (f2fs_sb_has_device_alias(sbi) &&
1457			(ctx->opt_mask & BIT(F2FS_MOUNT_READ_EXTENT_CACHE)) &&
1458			!ctx_test_opt(ctx, F2FS_MOUNT_READ_EXTENT_CACHE)) {
1459		f2fs_err(sbi, "device aliasing requires extent cache");
1460		return -EINVAL;
1461	}
1462
1463	if (test_opt(sbi, RESERVE_ROOT) &&
1464			(ctx->opt_mask & BIT(F2FS_MOUNT_RESERVE_ROOT)) &&
1465			ctx_test_opt(ctx, F2FS_MOUNT_RESERVE_ROOT)) {
1466		f2fs_info(sbi, "Preserve previous reserve_root=%u",
1467			F2FS_OPTION(sbi).root_reserved_blocks);
1468		ctx_clear_opt(ctx, F2FS_MOUNT_RESERVE_ROOT);
1469		ctx->opt_mask &= ~BIT(F2FS_MOUNT_RESERVE_ROOT);
1470	}
1471	if (test_opt(sbi, RESERVE_NODE) &&
1472			(ctx->opt_mask & BIT(F2FS_MOUNT_RESERVE_NODE)) &&
1473			ctx_test_opt(ctx, F2FS_MOUNT_RESERVE_NODE)) {
1474		f2fs_info(sbi, "Preserve previous reserve_node=%u",
1475			F2FS_OPTION(sbi).root_reserved_nodes);
1476		ctx_clear_opt(ctx, F2FS_MOUNT_RESERVE_NODE);
1477		ctx->opt_mask &= ~BIT(F2FS_MOUNT_RESERVE_NODE);
1478	}
1479
1480	err = f2fs_check_test_dummy_encryption(fc, sb);
1481	if (err)
1482		return err;
1483
1484	err = f2fs_check_compression(fc, sb);
1485	if (err)
1486		return err;
1487
1488	err = f2fs_check_quota_consistency(fc, sb);
1489	if (err)
1490		return err;
1491
1492	if (!IS_ENABLED(CONFIG_UNICODE) && f2fs_sb_has_casefold(sbi)) {
1493		f2fs_err(sbi,
1494			"Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
1495		return -EINVAL;
1496	}
1497
1498	/*
1499	 * The BLKZONED feature indicates that the drive was formatted with
1500	 * zone alignment optimization. This is optional for host-aware
1501	 * devices, but mandatory for host-managed zoned block devices.
1502	 */
1503	if (f2fs_sb_has_blkzoned(sbi)) {
1504		if (F2FS_CTX_INFO(ctx).bggc_mode == BGGC_MODE_OFF) {
1505			f2fs_warn(sbi, "zoned devices need bggc");
1506			return -EINVAL;
1507		}
1508#ifdef CONFIG_BLK_DEV_ZONED
1509		if ((ctx->spec_mask & F2FS_SPEC_discard_unit) &&
1510		F2FS_CTX_INFO(ctx).discard_unit != DISCARD_UNIT_SECTION) {
1511			f2fs_info(sbi, "Zoned block device doesn't need small discard, set discard_unit=section by default");
1512			F2FS_CTX_INFO(ctx).discard_unit = DISCARD_UNIT_SECTION;
1513		}
1514
1515		if ((ctx->spec_mask & F2FS_SPEC_mode) &&
1516		F2FS_CTX_INFO(ctx).fs_mode != FS_MODE_LFS) {
1517			f2fs_info(sbi, "Only lfs mode is allowed with zoned block device feature");
1518			return -EINVAL;
1519		}
1520#else
1521		f2fs_err(sbi, "Zoned block device support is not enabled");
1522		return -EINVAL;
1523#endif
1524	}
1525
1526	if (ctx_test_opt(ctx, F2FS_MOUNT_INLINE_XATTR_SIZE)) {
1527		if (!f2fs_sb_has_extra_attr(sbi) ||
1528			!f2fs_sb_has_flexible_inline_xattr(sbi)) {
1529			f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
1530			return -EINVAL;
1531		}
1532		if (!ctx_test_opt(ctx, F2FS_MOUNT_INLINE_XATTR) && !test_opt(sbi, INLINE_XATTR)) {
1533			f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
1534			return -EINVAL;
1535		}
1536	}
1537
1538	if (ctx_test_opt(ctx, F2FS_MOUNT_ATGC) &&
1539	    F2FS_CTX_INFO(ctx).fs_mode == FS_MODE_LFS) {
1540		f2fs_err(sbi, "LFS is not compatible with ATGC");
1541		return -EINVAL;
1542	}
1543
1544	if (f2fs_is_readonly(sbi) && ctx_test_opt(ctx, F2FS_MOUNT_FLUSH_MERGE)) {
1545		f2fs_err(sbi, "FLUSH_MERGE not compatible with readonly mode");
1546		return -EINVAL;
1547	}
1548
1549	if (f2fs_sb_has_readonly(sbi) && !f2fs_readonly(sbi->sb)) {
1550		f2fs_err(sbi, "Allow to mount readonly mode only");
1551		return -EROFS;
1552	}
1553	return 0;
1554}
1555
1556static void f2fs_apply_quota_options(struct fs_context *fc,
1557				     struct super_block *sb)
1558{
1559#ifdef CONFIG_QUOTA
1560	struct f2fs_fs_context *ctx = fc->fs_private;
1561	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1562	bool quota_feature = f2fs_sb_has_quota_ino(sbi);
1563	char *qname;
1564	int i;
1565
1566	if (quota_feature)
1567		return;
1568
1569	for (i = 0; i < MAXQUOTAS; i++) {
1570		if (!(ctx->qname_mask & (1 << i)))
1571			continue;
1572
1573		qname = F2FS_CTX_INFO(ctx).s_qf_names[i];
1574		if (qname) {
1575			qname = kstrdup(F2FS_CTX_INFO(ctx).s_qf_names[i],
1576					GFP_KERNEL | __GFP_NOFAIL);
1577			set_opt(sbi, QUOTA);
1578		}
1579		F2FS_OPTION(sbi).s_qf_names[i] = qname;
1580	}
1581
1582	if (ctx->spec_mask & F2FS_SPEC_jqfmt)
1583		F2FS_OPTION(sbi).s_jquota_fmt = F2FS_CTX_INFO(ctx).s_jquota_fmt;
1584
1585	if (quota_feature && F2FS_OPTION(sbi).s_jquota_fmt) {
1586		f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
1587		F2FS_OPTION(sbi).s_jquota_fmt = 0;
1588	}
1589#endif
1590}
1591
1592static void f2fs_apply_test_dummy_encryption(struct fs_context *fc,
1593					     struct super_block *sb)
1594{
1595	struct f2fs_fs_context *ctx = fc->fs_private;
1596	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1597
1598	if (!fscrypt_is_dummy_policy_set(&F2FS_CTX_INFO(ctx).dummy_enc_policy) ||
1599		/* if already set, it was already verified to be the same */
1600		fscrypt_is_dummy_policy_set(&F2FS_OPTION(sbi).dummy_enc_policy))
1601		return;
1602	swap(F2FS_OPTION(sbi).dummy_enc_policy, F2FS_CTX_INFO(ctx).dummy_enc_policy);
1603	f2fs_warn(sbi, "Test dummy encryption mode enabled");
1604}
1605
1606static void f2fs_apply_compression(struct fs_context *fc,
1607				   struct super_block *sb)
1608{
1609#ifdef CONFIG_F2FS_FS_COMPRESSION
1610	struct f2fs_fs_context *ctx = fc->fs_private;
1611	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1612	unsigned char (*ctx_ext)[F2FS_EXTENSION_LEN];
1613	unsigned char (*sbi_ext)[F2FS_EXTENSION_LEN];
1614	int ctx_cnt, sbi_cnt, i;
1615
1616	if (ctx->spec_mask & F2FS_SPEC_compress_level)
1617		F2FS_OPTION(sbi).compress_level =
1618					F2FS_CTX_INFO(ctx).compress_level;
1619	if (ctx->spec_mask & F2FS_SPEC_compress_algorithm)
1620		F2FS_OPTION(sbi).compress_algorithm =
1621					F2FS_CTX_INFO(ctx).compress_algorithm;
1622	if (ctx->spec_mask & F2FS_SPEC_compress_log_size)
1623		F2FS_OPTION(sbi).compress_log_size =
1624					F2FS_CTX_INFO(ctx).compress_log_size;
1625	if (ctx->spec_mask & F2FS_SPEC_compress_chksum)
1626		F2FS_OPTION(sbi).compress_chksum =
1627					F2FS_CTX_INFO(ctx).compress_chksum;
1628	if (ctx->spec_mask & F2FS_SPEC_compress_mode)
1629		F2FS_OPTION(sbi).compress_mode =
1630					F2FS_CTX_INFO(ctx).compress_mode;
1631	if (ctx->spec_mask & F2FS_SPEC_compress_extension) {
1632		ctx_ext = F2FS_CTX_INFO(ctx).extensions;
1633		ctx_cnt = F2FS_CTX_INFO(ctx).compress_ext_cnt;
1634		sbi_ext = F2FS_OPTION(sbi).extensions;
1635		sbi_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
1636		for (i = 0; i < ctx_cnt; i++) {
1637			if (strlen(ctx_ext[i]) == 0)
1638				continue;
1639			strscpy(sbi_ext[sbi_cnt], ctx_ext[i]);
1640			sbi_cnt++;
1641		}
1642		F2FS_OPTION(sbi).compress_ext_cnt = sbi_cnt;
1643	}
1644	if (ctx->spec_mask & F2FS_SPEC_nocompress_extension) {
1645		ctx_ext = F2FS_CTX_INFO(ctx).noextensions;
1646		ctx_cnt = F2FS_CTX_INFO(ctx).nocompress_ext_cnt;
1647		sbi_ext = F2FS_OPTION(sbi).noextensions;
1648		sbi_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
1649		for (i = 0; i < ctx_cnt; i++) {
1650			if (strlen(ctx_ext[i]) == 0)
1651				continue;
1652			strscpy(sbi_ext[sbi_cnt], ctx_ext[i]);
1653			sbi_cnt++;
1654		}
1655		F2FS_OPTION(sbi).nocompress_ext_cnt = sbi_cnt;
1656	}
1657#endif
1658}
1659
1660static void f2fs_apply_options(struct fs_context *fc, struct super_block *sb)
1661{
1662	struct f2fs_fs_context *ctx = fc->fs_private;
1663	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1664
1665	F2FS_OPTION(sbi).opt &= ~ctx->opt_mask;
1666	F2FS_OPTION(sbi).opt |= F2FS_CTX_INFO(ctx).opt;
1667
1668	if (ctx->spec_mask & F2FS_SPEC_background_gc)
1669		F2FS_OPTION(sbi).bggc_mode = F2FS_CTX_INFO(ctx).bggc_mode;
1670	if (ctx->spec_mask & F2FS_SPEC_inline_xattr_size)
1671		F2FS_OPTION(sbi).inline_xattr_size =
1672					F2FS_CTX_INFO(ctx).inline_xattr_size;
1673	if (ctx->spec_mask & F2FS_SPEC_active_logs)
1674		F2FS_OPTION(sbi).active_logs = F2FS_CTX_INFO(ctx).active_logs;
1675	if (ctx->spec_mask & F2FS_SPEC_reserve_root)
1676		F2FS_OPTION(sbi).root_reserved_blocks =
1677					F2FS_CTX_INFO(ctx).root_reserved_blocks;
1678	if (ctx->spec_mask & F2FS_SPEC_reserve_node)
1679		F2FS_OPTION(sbi).root_reserved_nodes =
1680					F2FS_CTX_INFO(ctx).root_reserved_nodes;
1681	if (ctx->spec_mask & F2FS_SPEC_resgid)
1682		F2FS_OPTION(sbi).s_resgid = F2FS_CTX_INFO(ctx).s_resgid;
1683	if (ctx->spec_mask & F2FS_SPEC_resuid)
1684		F2FS_OPTION(sbi).s_resuid = F2FS_CTX_INFO(ctx).s_resuid;
1685	if (ctx->spec_mask & F2FS_SPEC_mode)
1686		F2FS_OPTION(sbi).fs_mode = F2FS_CTX_INFO(ctx).fs_mode;
1687#ifdef CONFIG_F2FS_FAULT_INJECTION
1688	if (ctx->spec_mask & F2FS_SPEC_fault_injection)
1689		(void)f2fs_build_fault_attr(sbi,
1690		F2FS_CTX_INFO(ctx).fault_info.inject_rate, 0, FAULT_RATE);
1691	if (ctx->spec_mask & F2FS_SPEC_fault_type)
1692		(void)f2fs_build_fault_attr(sbi, 0,
1693			F2FS_CTX_INFO(ctx).fault_info.inject_type, FAULT_TYPE);
1694#endif
1695	if (ctx->spec_mask & F2FS_SPEC_alloc_mode)
1696		F2FS_OPTION(sbi).alloc_mode = F2FS_CTX_INFO(ctx).alloc_mode;
1697	if (ctx->spec_mask & F2FS_SPEC_fsync_mode)
1698		F2FS_OPTION(sbi).fsync_mode = F2FS_CTX_INFO(ctx).fsync_mode;
1699	if (ctx->spec_mask & F2FS_SPEC_checkpoint_disable_cap)
1700		F2FS_OPTION(sbi).unusable_cap = F2FS_CTX_INFO(ctx).unusable_cap;
1701	if (ctx->spec_mask & F2FS_SPEC_checkpoint_disable_cap_perc)
1702		F2FS_OPTION(sbi).unusable_cap_perc =
1703					F2FS_CTX_INFO(ctx).unusable_cap_perc;
1704	if (ctx->spec_mask & F2FS_SPEC_discard_unit)
1705		F2FS_OPTION(sbi).discard_unit = F2FS_CTX_INFO(ctx).discard_unit;
1706	if (ctx->spec_mask & F2FS_SPEC_memory_mode)
1707		F2FS_OPTION(sbi).memory_mode = F2FS_CTX_INFO(ctx).memory_mode;
1708	if (ctx->spec_mask & F2FS_SPEC_errors)
1709		F2FS_OPTION(sbi).errors = F2FS_CTX_INFO(ctx).errors;
1710	if (ctx->spec_mask & F2FS_SPEC_lookup_mode)
1711		F2FS_OPTION(sbi).lookup_mode = F2FS_CTX_INFO(ctx).lookup_mode;
1712
1713	f2fs_apply_compression(fc, sb);
1714	f2fs_apply_test_dummy_encryption(fc, sb);
1715	f2fs_apply_quota_options(fc, sb);
1716}
1717
1718static int f2fs_sanity_check_options(struct f2fs_sb_info *sbi, bool remount)
1719{
1720	if (f2fs_sb_has_device_alias(sbi) &&
1721	    !test_opt(sbi, READ_EXTENT_CACHE)) {
1722		f2fs_err(sbi, "device aliasing requires extent cache");
1723		return -EINVAL;
1724	}
1725
1726	if (!remount)
1727		return 0;
1728
1729#ifdef CONFIG_BLK_DEV_ZONED
1730	if (f2fs_sb_has_blkzoned(sbi) &&
1731	    sbi->max_open_zones < F2FS_OPTION(sbi).active_logs) {
1732		f2fs_err(sbi,
1733			"zoned: max open zones %u is too small, need at least %u open zones",
1734				 sbi->max_open_zones, F2FS_OPTION(sbi).active_logs);
1735		return -EINVAL;
1736	}
1737#endif
1738	if (f2fs_lfs_mode(sbi) && !IS_F2FS_IPU_DISABLE(sbi)) {
1739		f2fs_warn(sbi, "LFS is not compatible with IPU");
1740		return -EINVAL;
1741	}
1742	return 0;
1743}
1744
1745static struct inode *f2fs_alloc_inode(struct super_block *sb)
1746{
1747	struct f2fs_inode_info *fi;
1748
1749	if (time_to_inject(F2FS_SB(sb), FAULT_SLAB_ALLOC))
1750		return NULL;
1751
1752	fi = alloc_inode_sb(sb, f2fs_inode_cachep, GFP_F2FS_ZERO);
1753	if (!fi)
1754		return NULL;
1755
1756	init_once((void *) fi);
1757
1758	/* Initialize f2fs-specific inode info */
1759	atomic_set(&fi->dirty_pages, 0);
1760	atomic_set(&fi->i_compr_blocks, 0);
1761	atomic_set(&fi->open_count, 0);
1762	atomic_set(&fi->writeback, 0);
1763	init_f2fs_rwsem(&fi->i_sem);
1764	spin_lock_init(&fi->i_size_lock);
1765	INIT_LIST_HEAD(&fi->dirty_list);
1766	INIT_LIST_HEAD(&fi->gdirty_list);
1767	INIT_LIST_HEAD(&fi->gdonate_list);
1768	init_f2fs_rwsem(&fi->i_gc_rwsem[READ]);
1769	init_f2fs_rwsem(&fi->i_gc_rwsem[WRITE]);
1770	init_f2fs_rwsem(&fi->i_xattr_sem);
1771
1772	/* Will be used by directory only */
1773	fi->i_dir_level = F2FS_SB(sb)->dir_level;
1774
1775	return &fi->vfs_inode;
1776}
1777
1778static int f2fs_drop_inode(struct inode *inode)
1779{
1780	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1781	int ret;
1782
1783	/*
1784	 * during filesystem shutdown, if checkpoint is disabled,
1785	 * drop useless meta/node dirty pages.
1786	 */
1787	if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1788		if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1789			inode->i_ino == F2FS_META_INO(sbi)) {
1790			trace_f2fs_drop_inode(inode, 1);
1791			return 1;
1792		}
1793	}
1794
1795	/*
1796	 * This is to avoid a deadlock condition like below.
1797	 * writeback_single_inode(inode)
1798	 *  - f2fs_write_data_page
1799	 *    - f2fs_gc -> iput -> evict
1800	 *       - inode_wait_for_writeback(inode)
1801	 */
1802	if ((!inode_unhashed(inode) && inode_state_read(inode) & I_SYNC)) {
1803		if (!inode->i_nlink && !is_bad_inode(inode)) {
1804			/* to avoid evict_inode call simultaneously */
1805			__iget(inode);
1806			spin_unlock(&inode->i_lock);
1807
1808			/* should remain fi->extent_tree for writepage */
1809			f2fs_destroy_extent_node(inode);
1810
1811			sb_start_intwrite(inode->i_sb);
1812			f2fs_i_size_write(inode, 0);
1813
1814			f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1815					inode, NULL, 0, DATA);
1816			truncate_inode_pages_final(inode->i_mapping);
1817
1818			if (F2FS_HAS_BLOCKS(inode))
1819				f2fs_truncate(inode);
1820
1821			sb_end_intwrite(inode->i_sb);
1822
1823			spin_lock(&inode->i_lock);
1824			atomic_dec(&inode->i_count);
1825		}
1826		trace_f2fs_drop_inode(inode, 0);
1827		return 0;
1828	}
1829	ret = inode_generic_drop(inode);
1830	if (!ret)
1831		ret = fscrypt_drop_inode(inode);
1832	trace_f2fs_drop_inode(inode, ret);
1833	return ret;
1834}
1835
1836int f2fs_inode_dirtied(struct inode *inode, bool sync)
1837{
1838	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1839	int ret = 0;
1840
1841	spin_lock(&sbi->inode_lock[DIRTY_META]);
1842	if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1843		ret = 1;
1844	} else {
1845		set_inode_flag(inode, FI_DIRTY_INODE);
1846		stat_inc_dirty_inode(sbi, DIRTY_META);
1847	}
1848	if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1849		list_add_tail(&F2FS_I(inode)->gdirty_list,
1850				&sbi->inode_list[DIRTY_META]);
1851		inc_page_count(sbi, F2FS_DIRTY_IMETA);
1852	}
1853	spin_unlock(&sbi->inode_lock[DIRTY_META]);
1854
1855	/* if atomic write is not committed, set inode w/ atomic dirty */
1856	if (!ret && f2fs_is_atomic_file(inode) &&
1857			!is_inode_flag_set(inode, FI_ATOMIC_COMMITTED))
1858		set_inode_flag(inode, FI_ATOMIC_DIRTIED);
1859
1860	return ret;
1861}
1862
1863void f2fs_inode_synced(struct inode *inode)
1864{
1865	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1866
1867	spin_lock(&sbi->inode_lock[DIRTY_META]);
1868	if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1869		spin_unlock(&sbi->inode_lock[DIRTY_META]);
1870		return;
1871	}
1872	if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1873		list_del_init(&F2FS_I(inode)->gdirty_list);
1874		dec_page_count(sbi, F2FS_DIRTY_IMETA);
1875	}
1876	clear_inode_flag(inode, FI_DIRTY_INODE);
1877	clear_inode_flag(inode, FI_AUTO_RECOVER);
1878	stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1879	spin_unlock(&sbi->inode_lock[DIRTY_META]);
1880}
1881
1882/*
1883 * f2fs_dirty_inode() is called from __mark_inode_dirty()
1884 *
1885 * We should call set_dirty_inode to write the dirty inode through write_inode.
1886 */
1887static void f2fs_dirty_inode(struct inode *inode, int flags)
1888{
1889	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1890
1891	if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1892			inode->i_ino == F2FS_META_INO(sbi))
1893		return;
1894
1895	if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1896		clear_inode_flag(inode, FI_AUTO_RECOVER);
1897
1898	f2fs_inode_dirtied(inode, false);
1899}
1900
1901static void f2fs_free_inode(struct inode *inode)
1902{
1903	fscrypt_free_inode(inode);
1904	kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1905}
1906
1907static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1908{
1909	percpu_counter_destroy(&sbi->total_valid_inode_count);
1910	percpu_counter_destroy(&sbi->rf_node_block_count);
1911	percpu_counter_destroy(&sbi->alloc_valid_block_count);
1912}
1913
1914static void destroy_device_list(struct f2fs_sb_info *sbi)
1915{
1916	int i;
1917
1918	for (i = 0; i < sbi->s_ndevs; i++) {
1919		if (i > 0)
1920			bdev_fput(FDEV(i).bdev_file);
1921#ifdef CONFIG_BLK_DEV_ZONED
1922		kvfree(FDEV(i).blkz_seq);
1923#endif
1924	}
1925	kvfree(sbi->devs);
1926}
1927
1928static void f2fs_put_super(struct super_block *sb)
1929{
1930	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1931	int i;
1932	int err = 0;
1933	bool done;
1934
1935	/* unregister procfs/sysfs entries in advance to avoid race case */
1936	f2fs_unregister_sysfs(sbi);
1937
1938	f2fs_quota_off_umount(sb);
1939
1940	/* prevent remaining shrinker jobs */
1941	mutex_lock(&sbi->umount_mutex);
1942
1943	/*
1944	 * flush all issued checkpoints and stop checkpoint issue thread.
1945	 * after then, all checkpoints should be done by each process context.
1946	 */
1947	f2fs_stop_ckpt_thread(sbi);
1948
1949	/*
1950	 * We don't need to do checkpoint when superblock is clean.
1951	 * But, the previous checkpoint was not done by umount, it needs to do
1952	 * clean checkpoint again.
1953	 */
1954	if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1955			!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1956		struct cp_control cpc = {
1957			.reason = CP_UMOUNT,
1958		};
1959		stat_inc_cp_call_count(sbi, TOTAL_CALL);
1960		err = f2fs_write_checkpoint(sbi, &cpc);
1961	}
1962
1963	/* be sure to wait for any on-going discard commands */
1964	done = f2fs_issue_discard_timeout(sbi);
1965	if (f2fs_realtime_discard_enable(sbi) && !sbi->discard_blks && done) {
1966		struct cp_control cpc = {
1967			.reason = CP_UMOUNT | CP_TRIMMED,
1968		};
1969		stat_inc_cp_call_count(sbi, TOTAL_CALL);
1970		err = f2fs_write_checkpoint(sbi, &cpc);
1971	}
1972
1973	/*
1974	 * normally superblock is clean, so we need to release this.
1975	 * In addition, EIO will skip do checkpoint, we need this as well.
1976	 */
1977	f2fs_release_ino_entry(sbi, true);
1978
1979	f2fs_leave_shrinker(sbi);
1980	mutex_unlock(&sbi->umount_mutex);
1981
1982	/* our cp_error case, we can wait for any writeback page */
1983	f2fs_flush_merged_writes(sbi);
1984
1985	f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1986
1987	if (err || f2fs_cp_error(sbi)) {
1988		truncate_inode_pages_final(NODE_MAPPING(sbi));
1989		truncate_inode_pages_final(META_MAPPING(sbi));
1990	}
1991
1992	f2fs_bug_on(sbi, sbi->fsync_node_num);
1993
1994	f2fs_destroy_compress_inode(sbi);
1995
1996	iput(sbi->node_inode);
1997	sbi->node_inode = NULL;
1998
1999	iput(sbi->meta_inode);
2000	sbi->meta_inode = NULL;
2001
2002	/* Should check the page counts after dropping all node/meta pages */
2003	for (i = 0; i < NR_COUNT_TYPE; i++) {
2004		if (!get_pages(sbi, i))
2005			continue;
2006		f2fs_err(sbi, "detect filesystem reference count leak during "
2007			"umount, type: %d, count: %lld", i, get_pages(sbi, i));
2008		f2fs_bug_on(sbi, 1);
2009	}
2010
2011	/*
2012	 * iput() can update stat information, if f2fs_write_checkpoint()
2013	 * above failed with error.
2014	 */
2015	f2fs_destroy_stats(sbi);
2016
2017	/* destroy f2fs internal modules */
2018	f2fs_destroy_node_manager(sbi);
2019	f2fs_destroy_segment_manager(sbi);
2020
2021	/* flush s_error_work before sbi destroy */
2022	flush_work(&sbi->s_error_work);
2023
2024	f2fs_destroy_post_read_wq(sbi);
2025
2026	kvfree(sbi->ckpt);
2027
2028	kfree(sbi->raw_super);
2029
2030	f2fs_destroy_page_array_cache(sbi);
2031#ifdef CONFIG_QUOTA
2032	for (i = 0; i < MAXQUOTAS; i++)
2033		kfree(F2FS_OPTION(sbi).s_qf_names[i]);
2034#endif
2035	fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
2036	destroy_percpu_info(sbi);
2037	f2fs_destroy_iostat(sbi);
2038	for (i = 0; i < NR_PAGE_TYPE; i++)
2039		kfree(sbi->write_io[i]);
2040#if IS_ENABLED(CONFIG_UNICODE)
2041	utf8_unload(sb->s_encoding);
2042#endif
2043}
2044
2045int f2fs_sync_fs(struct super_block *sb, int sync)
2046{
2047	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2048	int err = 0;
2049
2050	if (unlikely(f2fs_cp_error(sbi)))
2051		return 0;
2052	if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
2053		return 0;
2054
2055	trace_f2fs_sync_fs(sb, sync);
2056
2057	if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
2058		return -EAGAIN;
2059
2060	if (sync) {
2061		stat_inc_cp_call_count(sbi, TOTAL_CALL);
2062		err = f2fs_issue_checkpoint(sbi);
2063	}
2064
2065	return err;
2066}
2067
2068static int f2fs_freeze(struct super_block *sb)
2069{
2070	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2071
2072	if (f2fs_readonly(sb))
2073		return 0;
2074
2075	/* IO error happened before */
2076	if (unlikely(f2fs_cp_error(sbi)))
2077		return -EIO;
2078
2079	/* must be clean, since sync_filesystem() was already called */
2080	if (is_sbi_flag_set(sbi, SBI_IS_DIRTY))
2081		return -EINVAL;
2082
2083	sbi->umount_lock_holder = current;
2084
2085	/* Let's flush checkpoints and stop the thread. */
2086	f2fs_flush_ckpt_thread(sbi);
2087
2088	sbi->umount_lock_holder = NULL;
2089
2090	/* to avoid deadlock on f2fs_evict_inode->SB_FREEZE_FS */
2091	set_sbi_flag(sbi, SBI_IS_FREEZING);
2092	return 0;
2093}
2094
2095static int f2fs_unfreeze(struct super_block *sb)
2096{
2097	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2098
2099	/*
2100	 * It will update discard_max_bytes of mounted lvm device to zero
2101	 * after creating snapshot on this lvm device, let's drop all
2102	 * remained discards.
2103	 * We don't need to disable real-time discard because discard_max_bytes
2104	 * will recover after removal of snapshot.
2105	 */
2106	if (test_opt(sbi, DISCARD) && !f2fs_hw_support_discard(sbi))
2107		f2fs_issue_discard_timeout(sbi);
2108
2109	clear_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
2110	return 0;
2111}
2112
2113#ifdef CONFIG_QUOTA
2114static int f2fs_statfs_project(struct super_block *sb,
2115				kprojid_t projid, struct kstatfs *buf)
2116{
2117	struct kqid qid;
2118	struct dquot *dquot;
2119	u64 limit;
2120	u64 curblock;
2121
2122	qid = make_kqid_projid(projid);
2123	dquot = dqget(sb, qid);
2124	if (IS_ERR(dquot))
2125		return PTR_ERR(dquot);
2126	spin_lock(&dquot->dq_dqb_lock);
2127
2128	limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
2129					dquot->dq_dqb.dqb_bhardlimit);
2130	limit >>= sb->s_blocksize_bits;
2131
2132	if (limit) {
2133		uint64_t remaining = 0;
2134
2135		curblock = (dquot->dq_dqb.dqb_curspace +
2136			    dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
2137		if (limit > curblock)
2138			remaining = limit - curblock;
2139
2140		buf->f_blocks = min(buf->f_blocks, limit);
2141		buf->f_bfree = min(buf->f_bfree, remaining);
2142		buf->f_bavail = min(buf->f_bavail, remaining);
2143	}
2144
2145	limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
2146					dquot->dq_dqb.dqb_ihardlimit);
2147
2148	if (limit) {
2149		uint64_t remaining = 0;
2150
2151		if (limit > dquot->dq_dqb.dqb_curinodes)
2152			remaining = limit - dquot->dq_dqb.dqb_curinodes;
2153
2154		buf->f_files = min(buf->f_files, limit);
2155		buf->f_ffree = min(buf->f_ffree, remaining);
2156	}
2157
2158	spin_unlock(&dquot->dq_dqb_lock);
2159	dqput(dquot);
2160	return 0;
2161}
2162#endif
2163
2164static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
2165{
2166	struct super_block *sb = dentry->d_sb;
2167	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2168	u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
2169	block_t total_count, user_block_count, start_count;
2170	u64 avail_node_count;
2171	unsigned int total_valid_node_count;
2172
2173	total_count = le64_to_cpu(sbi->raw_super->block_count);
2174	start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
2175	buf->f_type = F2FS_SUPER_MAGIC;
2176	buf->f_bsize = sbi->blocksize;
2177
2178	buf->f_blocks = total_count - start_count;
2179
2180	spin_lock(&sbi->stat_lock);
2181	if (sbi->carve_out)
2182		buf->f_blocks -= sbi->current_reserved_blocks;
2183	user_block_count = sbi->user_block_count;
2184	total_valid_node_count = valid_node_count(sbi);
2185	avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
2186	buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
2187						sbi->current_reserved_blocks;
2188
2189	if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
2190		buf->f_bfree = 0;
2191	else
2192		buf->f_bfree -= sbi->unusable_block_count;
2193	spin_unlock(&sbi->stat_lock);
2194
2195	if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
2196		buf->f_bavail = buf->f_bfree -
2197				F2FS_OPTION(sbi).root_reserved_blocks;
2198	else
2199		buf->f_bavail = 0;
2200
2201	if (avail_node_count > user_block_count) {
2202		buf->f_files = user_block_count;
2203		buf->f_ffree = buf->f_bavail;
2204	} else {
2205		buf->f_files = avail_node_count;
2206		buf->f_ffree = min(avail_node_count - total_valid_node_count,
2207					buf->f_bavail);
2208	}
2209
2210	buf->f_namelen = F2FS_NAME_LEN;
2211	buf->f_fsid    = u64_to_fsid(id);
2212
2213#ifdef CONFIG_QUOTA
2214	if (is_inode_flag_set(d_inode(dentry), FI_PROJ_INHERIT) &&
2215			sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
2216		f2fs_statfs_project(sb, F2FS_I(d_inode(dentry))->i_projid, buf);
2217	}
2218#endif
2219	return 0;
2220}
2221
2222static inline void f2fs_show_quota_options(struct seq_file *seq,
2223					   struct super_block *sb)
2224{
2225#ifdef CONFIG_QUOTA
2226	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2227
2228	if (F2FS_OPTION(sbi).s_jquota_fmt) {
2229		char *fmtname = "";
2230
2231		switch (F2FS_OPTION(sbi).s_jquota_fmt) {
2232		case QFMT_VFS_OLD:
2233			fmtname = "vfsold";
2234			break;
2235		case QFMT_VFS_V0:
2236			fmtname = "vfsv0";
2237			break;
2238		case QFMT_VFS_V1:
2239			fmtname = "vfsv1";
2240			break;
2241		}
2242		seq_printf(seq, ",jqfmt=%s", fmtname);
2243	}
2244
2245	if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
2246		seq_show_option(seq, "usrjquota",
2247			F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
2248
2249	if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
2250		seq_show_option(seq, "grpjquota",
2251			F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
2252
2253	if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
2254		seq_show_option(seq, "prjjquota",
2255			F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
2256#endif
2257}
2258
2259#ifdef CONFIG_F2FS_FS_COMPRESSION
2260static inline void f2fs_show_compress_options(struct seq_file *seq,
2261							struct super_block *sb)
2262{
2263	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2264	char *algtype = "";
2265	int i;
2266
2267	if (!f2fs_sb_has_compression(sbi))
2268		return;
2269
2270	switch (F2FS_OPTION(sbi).compress_algorithm) {
2271	case COMPRESS_LZO:
2272		algtype = "lzo";
2273		break;
2274	case COMPRESS_LZ4:
2275		algtype = "lz4";
2276		break;
2277	case COMPRESS_ZSTD:
2278		algtype = "zstd";
2279		break;
2280	case COMPRESS_LZORLE:
2281		algtype = "lzo-rle";
2282		break;
2283	}
2284	seq_printf(seq, ",compress_algorithm=%s", algtype);
2285
2286	if (F2FS_OPTION(sbi).compress_level)
2287		seq_printf(seq, ":%d", F2FS_OPTION(sbi).compress_level);
2288
2289	seq_printf(seq, ",compress_log_size=%u",
2290			F2FS_OPTION(sbi).compress_log_size);
2291
2292	for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
2293		seq_printf(seq, ",compress_extension=%s",
2294			F2FS_OPTION(sbi).extensions[i]);
2295	}
2296
2297	for (i = 0; i < F2FS_OPTION(sbi).nocompress_ext_cnt; i++) {
2298		seq_printf(seq, ",nocompress_extension=%s",
2299			F2FS_OPTION(sbi).noextensions[i]);
2300	}
2301
2302	if (F2FS_OPTION(sbi).compress_chksum)
2303		seq_puts(seq, ",compress_chksum");
2304
2305	if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_FS)
2306		seq_printf(seq, ",compress_mode=%s", "fs");
2307	else if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_USER)
2308		seq_printf(seq, ",compress_mode=%s", "user");
2309
2310	if (test_opt(sbi, COMPRESS_CACHE))
2311		seq_puts(seq, ",compress_cache");
2312}
2313#endif
2314
2315static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
2316{
2317	struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
2318
2319	if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
2320		seq_printf(seq, ",background_gc=%s", "sync");
2321	else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
2322		seq_printf(seq, ",background_gc=%s", "on");
2323	else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
2324		seq_printf(seq, ",background_gc=%s", "off");
2325
2326	if (test_opt(sbi, GC_MERGE))
2327		seq_puts(seq, ",gc_merge");
2328	else
2329		seq_puts(seq, ",nogc_merge");
2330
2331	if (test_opt(sbi, DISABLE_ROLL_FORWARD))
2332		seq_puts(seq, ",disable_roll_forward");
2333	if (test_opt(sbi, NORECOVERY))
2334		seq_puts(seq, ",norecovery");
2335	if (test_opt(sbi, DISCARD)) {
2336		seq_puts(seq, ",discard");
2337		if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK)
2338			seq_printf(seq, ",discard_unit=%s", "block");
2339		else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SEGMENT)
2340			seq_printf(seq, ",discard_unit=%s", "segment");
2341		else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SECTION)
2342			seq_printf(seq, ",discard_unit=%s", "section");
2343	} else {
2344		seq_puts(seq, ",nodiscard");
2345	}
2346#ifdef CONFIG_F2FS_FS_XATTR
2347	if (test_opt(sbi, XATTR_USER))
2348		seq_puts(seq, ",user_xattr");
2349	else
2350		seq_puts(seq, ",nouser_xattr");
2351	if (test_opt(sbi, INLINE_XATTR))
2352		seq_puts(seq, ",inline_xattr");
2353	else
2354		seq_puts(seq, ",noinline_xattr");
2355	if (test_opt(sbi, INLINE_XATTR_SIZE))
2356		seq_printf(seq, ",inline_xattr_size=%u",
2357					F2FS_OPTION(sbi).inline_xattr_size);
2358#endif
2359#ifdef CONFIG_F2FS_FS_POSIX_ACL
2360	if (test_opt(sbi, POSIX_ACL))
2361		seq_puts(seq, ",acl");
2362	else
2363		seq_puts(seq, ",noacl");
2364#endif
2365	if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
2366		seq_puts(seq, ",disable_ext_identify");
2367	if (test_opt(sbi, INLINE_DATA))
2368		seq_puts(seq, ",inline_data");
2369	else
2370		seq_puts(seq, ",noinline_data");
2371	if (test_opt(sbi, INLINE_DENTRY))
2372		seq_puts(seq, ",inline_dentry");
2373	else
2374		seq_puts(seq, ",noinline_dentry");
2375	if (test_opt(sbi, FLUSH_MERGE))
2376		seq_puts(seq, ",flush_merge");
2377	else
2378		seq_puts(seq, ",noflush_merge");
2379	if (test_opt(sbi, NOBARRIER))
2380		seq_puts(seq, ",nobarrier");
2381	else
2382		seq_puts(seq, ",barrier");
2383	if (test_opt(sbi, FASTBOOT))
2384		seq_puts(seq, ",fastboot");
2385	if (test_opt(sbi, READ_EXTENT_CACHE))
2386		seq_puts(seq, ",extent_cache");
2387	else
2388		seq_puts(seq, ",noextent_cache");
2389	if (test_opt(sbi, AGE_EXTENT_CACHE))
2390		seq_puts(seq, ",age_extent_cache");
2391	if (test_opt(sbi, DATA_FLUSH))
2392		seq_puts(seq, ",data_flush");
2393
2394	seq_puts(seq, ",mode=");
2395	if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
2396		seq_puts(seq, "adaptive");
2397	else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
2398		seq_puts(seq, "lfs");
2399	else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG)
2400		seq_puts(seq, "fragment:segment");
2401	else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK)
2402		seq_puts(seq, "fragment:block");
2403	seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
2404	if (test_opt(sbi, RESERVE_ROOT) || test_opt(sbi, RESERVE_NODE))
2405		seq_printf(seq, ",reserve_root=%u,reserve_node=%u,resuid=%u,"
2406				"resgid=%u",
2407				F2FS_OPTION(sbi).root_reserved_blocks,
2408				F2FS_OPTION(sbi).root_reserved_nodes,
2409				from_kuid_munged(&init_user_ns,
2410					F2FS_OPTION(sbi).s_resuid),
2411				from_kgid_munged(&init_user_ns,
2412					F2FS_OPTION(sbi).s_resgid));
2413#ifdef CONFIG_F2FS_FAULT_INJECTION
2414	if (test_opt(sbi, FAULT_INJECTION)) {
2415		seq_printf(seq, ",fault_injection=%u",
2416				F2FS_OPTION(sbi).fault_info.inject_rate);
2417		seq_printf(seq, ",fault_type=%u",
2418				F2FS_OPTION(sbi).fault_info.inject_type);
2419	}
2420#endif
2421#ifdef CONFIG_QUOTA
2422	if (test_opt(sbi, QUOTA))
2423		seq_puts(seq, ",quota");
2424	if (test_opt(sbi, USRQUOTA))
2425		seq_puts(seq, ",usrquota");
2426	if (test_opt(sbi, GRPQUOTA))
2427		seq_puts(seq, ",grpquota");
2428	if (test_opt(sbi, PRJQUOTA))
2429		seq_puts(seq, ",prjquota");
2430#endif
2431	f2fs_show_quota_options(seq, sbi->sb);
2432
2433	fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
2434
2435	if (sbi->sb->s_flags & SB_INLINECRYPT)
2436		seq_puts(seq, ",inlinecrypt");
2437
2438	if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
2439		seq_printf(seq, ",alloc_mode=%s", "default");
2440	else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
2441		seq_printf(seq, ",alloc_mode=%s", "reuse");
2442
2443	if (test_opt(sbi, DISABLE_CHECKPOINT))
2444		seq_printf(seq, ",checkpoint=disable:%u",
2445				F2FS_OPTION(sbi).unusable_cap);
2446	if (test_opt(sbi, MERGE_CHECKPOINT))
2447		seq_puts(seq, ",checkpoint_merge");
2448	else
2449		seq_puts(seq, ",nocheckpoint_merge");
2450	if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
2451		seq_printf(seq, ",fsync_mode=%s", "posix");
2452	else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
2453		seq_printf(seq, ",fsync_mode=%s", "strict");
2454	else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
2455		seq_printf(seq, ",fsync_mode=%s", "nobarrier");
2456
2457#ifdef CONFIG_F2FS_FS_COMPRESSION
2458	f2fs_show_compress_options(seq, sbi->sb);
2459#endif
2460
2461	if (test_opt(sbi, ATGC))
2462		seq_puts(seq, ",atgc");
2463
2464	if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_NORMAL)
2465		seq_printf(seq, ",memory=%s", "normal");
2466	else if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW)
2467		seq_printf(seq, ",memory=%s", "low");
2468
2469	if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_READONLY)
2470		seq_printf(seq, ",errors=%s", "remount-ro");
2471	else if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_CONTINUE)
2472		seq_printf(seq, ",errors=%s", "continue");
2473	else if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_PANIC)
2474		seq_printf(seq, ",errors=%s", "panic");
2475
2476	if (test_opt(sbi, NAT_BITS))
2477		seq_puts(seq, ",nat_bits");
2478
2479	if (F2FS_OPTION(sbi).lookup_mode == LOOKUP_PERF)
2480		seq_show_option(seq, "lookup_mode", "perf");
2481	else if (F2FS_OPTION(sbi).lookup_mode == LOOKUP_COMPAT)
2482		seq_show_option(seq, "lookup_mode", "compat");
2483	else if (F2FS_OPTION(sbi).lookup_mode == LOOKUP_AUTO)
2484		seq_show_option(seq, "lookup_mode", "auto");
2485
2486	return 0;
2487}
2488
2489static void default_options(struct f2fs_sb_info *sbi, bool remount)
2490{
2491	/* init some FS parameters */
2492	if (!remount) {
2493		set_opt(sbi, READ_EXTENT_CACHE);
2494		clear_opt(sbi, DISABLE_CHECKPOINT);
2495
2496		if (f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi))
2497			set_opt(sbi, DISCARD);
2498
2499		if (f2fs_sb_has_blkzoned(sbi))
2500			F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_SECTION;
2501		else
2502			F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_BLOCK;
2503	}
2504
2505	if (f2fs_sb_has_readonly(sbi))
2506		F2FS_OPTION(sbi).active_logs = NR_CURSEG_RO_TYPE;
2507	else
2508		F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
2509
2510	F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
2511	if (le32_to_cpu(F2FS_RAW_SUPER(sbi)->segment_count_main) <=
2512							SMALL_VOLUME_SEGMENTS)
2513		F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
2514	else
2515		F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
2516	F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
2517	F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
2518	F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
2519	if (f2fs_sb_has_compression(sbi)) {
2520		F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
2521		F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
2522		F2FS_OPTION(sbi).compress_ext_cnt = 0;
2523		F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
2524	}
2525	F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
2526	F2FS_OPTION(sbi).memory_mode = MEMORY_MODE_NORMAL;
2527	F2FS_OPTION(sbi).errors = MOUNT_ERRORS_CONTINUE;
2528
2529	set_opt(sbi, INLINE_XATTR);
2530	set_opt(sbi, INLINE_DATA);
2531	set_opt(sbi, INLINE_DENTRY);
2532	set_opt(sbi, MERGE_CHECKPOINT);
2533	set_opt(sbi, LAZYTIME);
2534	F2FS_OPTION(sbi).unusable_cap = 0;
2535	if (!f2fs_is_readonly(sbi))
2536		set_opt(sbi, FLUSH_MERGE);
2537	if (f2fs_sb_has_blkzoned(sbi))
2538		F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
2539	else
2540		F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
2541
2542#ifdef CONFIG_F2FS_FS_XATTR
2543	set_opt(sbi, XATTR_USER);
2544#endif
2545#ifdef CONFIG_F2FS_FS_POSIX_ACL
2546	set_opt(sbi, POSIX_ACL);
2547#endif
2548
2549	f2fs_build_fault_attr(sbi, 0, 0, FAULT_ALL);
2550
2551	F2FS_OPTION(sbi).lookup_mode = LOOKUP_PERF;
2552}
2553
2554#ifdef CONFIG_QUOTA
2555static int f2fs_enable_quotas(struct super_block *sb);
2556#endif
2557
2558static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
2559{
2560	unsigned int s_flags = sbi->sb->s_flags;
2561	struct cp_control cpc;
2562	unsigned int gc_mode = sbi->gc_mode;
2563	int err = 0;
2564	int ret;
2565	block_t unusable;
2566
2567	if (s_flags & SB_RDONLY) {
2568		f2fs_err(sbi, "checkpoint=disable on readonly fs");
2569		return -EINVAL;
2570	}
2571	sbi->sb->s_flags |= SB_ACTIVE;
2572
2573	/* check if we need more GC first */
2574	unusable = f2fs_get_unusable_blocks(sbi);
2575	if (!f2fs_disable_cp_again(sbi, unusable))
2576		goto skip_gc;
2577
2578	f2fs_update_time(sbi, DISABLE_TIME);
2579
2580	sbi->gc_mode = GC_URGENT_HIGH;
2581
2582	while (!f2fs_time_over(sbi, DISABLE_TIME)) {
2583		struct f2fs_gc_control gc_control = {
2584			.victim_segno = NULL_SEGNO,
2585			.init_gc_type = FG_GC,
2586			.should_migrate_blocks = false,
2587			.err_gc_skipped = true,
2588			.no_bg_gc = true,
2589			.nr_free_secs = 1 };
2590
2591		f2fs_down_write(&sbi->gc_lock);
2592		stat_inc_gc_call_count(sbi, FOREGROUND);
2593		err = f2fs_gc(sbi, &gc_control);
2594		if (err == -ENODATA) {
2595			err = 0;
2596			break;
2597		}
2598		if (err && err != -EAGAIN)
2599			break;
2600	}
2601
2602	ret = sync_filesystem(sbi->sb);
2603	if (ret || err) {
2604		err = ret ? ret : err;
2605		goto restore_flag;
2606	}
2607
2608	unusable = f2fs_get_unusable_blocks(sbi);
2609	if (f2fs_disable_cp_again(sbi, unusable)) {
2610		err = -EAGAIN;
2611		goto restore_flag;
2612	}
2613
2614skip_gc:
2615	f2fs_down_write(&sbi->gc_lock);
2616	cpc.reason = CP_PAUSE;
2617	set_sbi_flag(sbi, SBI_CP_DISABLED);
2618	stat_inc_cp_call_count(sbi, TOTAL_CALL);
2619	err = f2fs_write_checkpoint(sbi, &cpc);
2620	if (err)
2621		goto out_unlock;
2622
2623	spin_lock(&sbi->stat_lock);
2624	sbi->unusable_block_count = unusable;
2625	spin_unlock(&sbi->stat_lock);
2626
2627out_unlock:
2628	f2fs_up_write(&sbi->gc_lock);
2629restore_flag:
2630	sbi->gc_mode = gc_mode;
2631	sbi->sb->s_flags = s_flags;	/* Restore SB_RDONLY status */
2632	f2fs_info(sbi, "f2fs_disable_checkpoint() finish, err:%d", err);
2633	return err;
2634}
2635
2636static int f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
2637{
2638	unsigned int nr_pages = get_pages(sbi, F2FS_DIRTY_DATA) / 16;
2639	long long start, writeback, end;
2640	int ret;
2641
2642	f2fs_info(sbi, "f2fs_enable_checkpoint() starts, meta: %lld, node: %lld, data: %lld",
2643					get_pages(sbi, F2FS_DIRTY_META),
2644					get_pages(sbi, F2FS_DIRTY_NODES),
2645					get_pages(sbi, F2FS_DIRTY_DATA));
2646
2647	f2fs_update_time(sbi, ENABLE_TIME);
2648
2649	start = ktime_get();
2650
2651	/* we should flush all the data to keep data consistency */
2652	while (get_pages(sbi, F2FS_DIRTY_DATA)) {
2653		writeback_inodes_sb_nr(sbi->sb, nr_pages, WB_REASON_SYNC);
2654		f2fs_io_schedule_timeout(DEFAULT_SCHEDULE_TIMEOUT);
2655
2656		if (f2fs_time_over(sbi, ENABLE_TIME))
2657			break;
2658	}
2659	writeback = ktime_get();
2660
2661	sync_inodes_sb(sbi->sb);
2662
2663	if (unlikely(get_pages(sbi, F2FS_DIRTY_DATA)))
2664		f2fs_warn(sbi, "checkpoint=enable has some unwritten data: %lld",
2665					get_pages(sbi, F2FS_DIRTY_DATA));
2666
2667	f2fs_down_write(&sbi->gc_lock);
2668	f2fs_dirty_to_prefree(sbi);
2669
2670	clear_sbi_flag(sbi, SBI_CP_DISABLED);
2671	set_sbi_flag(sbi, SBI_IS_DIRTY);
2672	f2fs_up_write(&sbi->gc_lock);
2673
2674	ret = f2fs_sync_fs(sbi->sb, 1);
2675	if (ret)
2676		f2fs_err(sbi, "%s sync_fs failed, ret: %d", __func__, ret);
2677
2678	/* Let's ensure there's no pending checkpoint anymore */
2679	f2fs_flush_ckpt_thread(sbi);
2680
2681	end = ktime_get();
2682
2683	f2fs_info(sbi, "f2fs_enable_checkpoint() finishes, writeback:%llu, sync:%llu",
2684					ktime_ms_delta(writeback, start),
2685					ktime_ms_delta(end, writeback));
2686	return ret;
2687}
2688
2689static int __f2fs_remount(struct fs_context *fc, struct super_block *sb)
2690{
2691	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2692	struct f2fs_mount_info org_mount_opt;
2693	unsigned long old_sb_flags;
2694	unsigned int flags = fc->sb_flags;
2695	int err;
2696	bool need_restart_gc = false, need_stop_gc = false;
2697	bool need_restart_flush = false, need_stop_flush = false;
2698	bool need_restart_discard = false, need_stop_discard = false;
2699	bool need_enable_checkpoint = false, need_disable_checkpoint = false;
2700	bool no_read_extent_cache = !test_opt(sbi, READ_EXTENT_CACHE);
2701	bool no_age_extent_cache = !test_opt(sbi, AGE_EXTENT_CACHE);
2702	bool enable_checkpoint = !test_opt(sbi, DISABLE_CHECKPOINT);
2703	bool no_atgc = !test_opt(sbi, ATGC);
2704	bool no_discard = !test_opt(sbi, DISCARD);
2705	bool no_compress_cache = !test_opt(sbi, COMPRESS_CACHE);
2706	bool block_unit_discard = f2fs_block_unit_discard(sbi);
2707	bool no_nat_bits = !test_opt(sbi, NAT_BITS);
2708#ifdef CONFIG_QUOTA
2709	int i, j;
2710#endif
2711
2712	/*
2713	 * Save the old mount options in case we
2714	 * need to restore them.
2715	 */
2716	org_mount_opt = sbi->mount_opt;
2717	old_sb_flags = sb->s_flags;
2718
2719	sbi->umount_lock_holder = current;
2720
2721#ifdef CONFIG_QUOTA
2722	org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
2723	for (i = 0; i < MAXQUOTAS; i++) {
2724		if (F2FS_OPTION(sbi).s_qf_names[i]) {
2725			org_mount_opt.s_qf_names[i] =
2726				kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
2727				GFP_KERNEL);
2728			if (!org_mount_opt.s_qf_names[i]) {
2729				for (j = 0; j < i; j++)
2730					kfree(org_mount_opt.s_qf_names[j]);
2731				return -ENOMEM;
2732			}
2733		} else {
2734			org_mount_opt.s_qf_names[i] = NULL;
2735		}
2736	}
2737#endif
2738
2739	/* recover superblocks we couldn't write due to previous RO mount */
2740	if (!(flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
2741		err = f2fs_commit_super(sbi, false);
2742		f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
2743			  err);
2744		if (!err)
2745			clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2746	}
2747
2748	default_options(sbi, true);
2749
2750	err = f2fs_check_opt_consistency(fc, sb);
2751	if (err)
2752		goto restore_opts;
2753
2754	f2fs_apply_options(fc, sb);
2755
2756	err = f2fs_sanity_check_options(sbi, true);
2757	if (err)
2758		goto restore_opts;
2759
2760	/* flush outstanding errors before changing fs state */
2761	flush_work(&sbi->s_error_work);
2762
2763	/*
2764	 * Previous and new state of filesystem is RO,
2765	 * so skip checking GC and FLUSH_MERGE conditions.
2766	 */
2767	if (f2fs_readonly(sb) && (flags & SB_RDONLY))
2768		goto skip;
2769
2770	if (f2fs_dev_is_readonly(sbi) && !(flags & SB_RDONLY)) {
2771		err = -EROFS;
2772		goto restore_opts;
2773	}
2774
2775#ifdef CONFIG_QUOTA
2776	if (!f2fs_readonly(sb) && (flags & SB_RDONLY)) {
2777		err = dquot_suspend(sb, -1);
2778		if (err < 0)
2779			goto restore_opts;
2780	} else if (f2fs_readonly(sb) && !(flags & SB_RDONLY)) {
2781		/* dquot_resume needs RW */
2782		sb->s_flags &= ~SB_RDONLY;
2783		if (sb_any_quota_suspended(sb)) {
2784			dquot_resume(sb, -1);
2785		} else if (f2fs_sb_has_quota_ino(sbi)) {
2786			err = f2fs_enable_quotas(sb);
2787			if (err)
2788				goto restore_opts;
2789		}
2790	}
2791#endif
2792	/* disallow enable atgc dynamically */
2793	if (no_atgc == !!test_opt(sbi, ATGC)) {
2794		err = -EINVAL;
2795		f2fs_warn(sbi, "switch atgc option is not allowed");
2796		goto restore_opts;
2797	}
2798
2799	/* disallow enable/disable extent_cache dynamically */
2800	if (no_read_extent_cache == !!test_opt(sbi, READ_EXTENT_CACHE)) {
2801		err = -EINVAL;
2802		f2fs_warn(sbi, "switch extent_cache option is not allowed");
2803		goto restore_opts;
2804	}
2805	/* disallow enable/disable age extent_cache dynamically */
2806	if (no_age_extent_cache == !!test_opt(sbi, AGE_EXTENT_CACHE)) {
2807		err = -EINVAL;
2808		f2fs_warn(sbi, "switch age_extent_cache option is not allowed");
2809		goto restore_opts;
2810	}
2811
2812	if (no_compress_cache == !!test_opt(sbi, COMPRESS_CACHE)) {
2813		err = -EINVAL;
2814		f2fs_warn(sbi, "switch compress_cache option is not allowed");
2815		goto restore_opts;
2816	}
2817
2818	if (block_unit_discard != f2fs_block_unit_discard(sbi)) {
2819		err = -EINVAL;
2820		f2fs_warn(sbi, "switch discard_unit option is not allowed");
2821		goto restore_opts;
2822	}
2823
2824	if (no_nat_bits == !!test_opt(sbi, NAT_BITS)) {
2825		err = -EINVAL;
2826		f2fs_warn(sbi, "switch nat_bits option is not allowed");
2827		goto restore_opts;
2828	}
2829
2830	if ((flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
2831		err = -EINVAL;
2832		f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
2833		goto restore_opts;
2834	}
2835
2836	/*
2837	 * We stop the GC thread if FS is mounted as RO
2838	 * or if background_gc = off is passed in mount
2839	 * option. Also sync the filesystem.
2840	 */
2841	if ((flags & SB_RDONLY) ||
2842			(F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF &&
2843			!test_opt(sbi, GC_MERGE))) {
2844		if (sbi->gc_thread) {
2845			f2fs_stop_gc_thread(sbi);
2846			need_restart_gc = true;
2847		}
2848	} else if (!sbi->gc_thread) {
2849		err = f2fs_start_gc_thread(sbi);
2850		if (err)
2851			goto restore_opts;
2852		need_stop_gc = true;
2853	}
2854
2855	if (flags & SB_RDONLY) {
2856		sync_inodes_sb(sb);
2857
2858		set_sbi_flag(sbi, SBI_IS_DIRTY);
2859		set_sbi_flag(sbi, SBI_IS_CLOSE);
2860		f2fs_sync_fs(sb, 1);
2861		clear_sbi_flag(sbi, SBI_IS_CLOSE);
2862	}
2863
2864	/*
2865	 * We stop issue flush thread if FS is mounted as RO
2866	 * or if flush_merge is not passed in mount option.
2867	 */
2868	if ((flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2869		clear_opt(sbi, FLUSH_MERGE);
2870		f2fs_destroy_flush_cmd_control(sbi, false);
2871		need_restart_flush = true;
2872	} else {
2873		err = f2fs_create_flush_cmd_control(sbi);
2874		if (err)
2875			goto restore_gc;
2876		need_stop_flush = true;
2877	}
2878
2879	if (no_discard == !!test_opt(sbi, DISCARD)) {
2880		if (test_opt(sbi, DISCARD)) {
2881			err = f2fs_start_discard_thread(sbi);
2882			if (err)
2883				goto restore_flush;
2884			need_stop_discard = true;
2885		} else {
2886			f2fs_stop_discard_thread(sbi);
2887			f2fs_issue_discard_timeout(sbi);
2888			need_restart_discard = true;
2889		}
2890	}
2891
2892	adjust_unusable_cap_perc(sbi);
2893	if (enable_checkpoint == !!test_opt(sbi, DISABLE_CHECKPOINT)) {
2894		if (test_opt(sbi, DISABLE_CHECKPOINT)) {
2895			err = f2fs_disable_checkpoint(sbi);
2896			if (err)
2897				goto restore_discard;
2898			need_enable_checkpoint = true;
2899		} else {
2900			err = f2fs_enable_checkpoint(sbi);
2901			if (err)
2902				goto restore_discard;
2903			need_disable_checkpoint = true;
2904		}
2905	}
2906
2907	/*
2908	 * Place this routine at the end, since a new checkpoint would be
2909	 * triggered while remount and we need to take care of it before
2910	 * returning from remount.
2911	 */
2912	if ((flags & SB_RDONLY) || test_opt(sbi, DISABLE_CHECKPOINT) ||
2913			!test_opt(sbi, MERGE_CHECKPOINT)) {
2914		f2fs_stop_ckpt_thread(sbi);
2915	} else {
2916		/* Flush if the previous checkpoint, if exists. */
2917		f2fs_flush_ckpt_thread(sbi);
2918
2919		err = f2fs_start_ckpt_thread(sbi);
2920		if (err) {
2921			f2fs_err(sbi,
2922			    "Failed to start F2FS issue_checkpoint_thread (%d)",
2923			    err);
2924			goto restore_checkpoint;
2925		}
2926	}
2927
2928skip:
2929#ifdef CONFIG_QUOTA
2930	/* Release old quota file names */
2931	for (i = 0; i < MAXQUOTAS; i++)
2932		kfree(org_mount_opt.s_qf_names[i]);
2933#endif
2934	/* Update the POSIXACL Flag */
2935	sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
2936		(test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
2937
2938	limit_reserve_root(sbi);
2939	fc->sb_flags = (flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
2940
2941	sbi->umount_lock_holder = NULL;
2942	return 0;
2943restore_checkpoint:
2944	if (need_enable_checkpoint) {
2945		if (f2fs_enable_checkpoint(sbi))
2946			f2fs_warn(sbi, "checkpoint has not been enabled");
2947	} else if (need_disable_checkpoint) {
2948		if (f2fs_disable_checkpoint(sbi))
2949			f2fs_warn(sbi, "checkpoint has not been disabled");
2950	}
2951restore_discard:
2952	if (need_restart_discard) {
2953		if (f2fs_start_discard_thread(sbi))
2954			f2fs_warn(sbi, "discard has been stopped");
2955	} else if (need_stop_discard) {
2956		f2fs_stop_discard_thread(sbi);
2957	}
2958restore_flush:
2959	if (need_restart_flush) {
2960		if (f2fs_create_flush_cmd_control(sbi))
2961			f2fs_warn(sbi, "background flush thread has stopped");
2962	} else if (need_stop_flush) {
2963		clear_opt(sbi, FLUSH_MERGE);
2964		f2fs_destroy_flush_cmd_control(sbi, false);
2965	}
2966restore_gc:
2967	if (need_restart_gc) {
2968		if (f2fs_start_gc_thread(sbi))
2969			f2fs_warn(sbi, "background gc thread has stopped");
2970	} else if (need_stop_gc) {
2971		f2fs_stop_gc_thread(sbi);
2972	}
2973restore_opts:
2974#ifdef CONFIG_QUOTA
2975	F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
2976	for (i = 0; i < MAXQUOTAS; i++) {
2977		kfree(F2FS_OPTION(sbi).s_qf_names[i]);
2978		F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
2979	}
2980#endif
2981	sbi->mount_opt = org_mount_opt;
2982	sb->s_flags = old_sb_flags;
2983
2984	sbi->umount_lock_holder = NULL;
2985	return err;
2986}
2987
2988static void f2fs_shutdown(struct super_block *sb)
2989{
2990	f2fs_do_shutdown(F2FS_SB(sb), F2FS_GOING_DOWN_NOSYNC, false, false);
2991}
2992
2993#ifdef CONFIG_QUOTA
2994static bool f2fs_need_recovery(struct f2fs_sb_info *sbi)
2995{
2996	/* need to recovery orphan */
2997	if (is_set_ckpt_flags(sbi, CP_ORPHAN_PRESENT_FLAG))
2998		return true;
2999	/* need to recovery data */
3000	if (test_opt(sbi, DISABLE_ROLL_FORWARD))
3001		return false;
3002	if (test_opt(sbi, NORECOVERY))
3003		return false;
3004	return !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG);
3005}
3006
3007static bool f2fs_recover_quota_begin(struct f2fs_sb_info *sbi)
3008{
3009	bool readonly = f2fs_readonly(sbi->sb);
3010
3011	if (!f2fs_need_recovery(sbi))
3012		return false;
3013
3014	/* it doesn't need to check f2fs_sb_has_readonly() */
3015	if (f2fs_hw_is_readonly(sbi))
3016		return false;
3017
3018	if (readonly) {
3019		sbi->sb->s_flags &= ~SB_RDONLY;
3020		set_sbi_flag(sbi, SBI_IS_WRITABLE);
3021	}
3022
3023	/*
3024	 * Turn on quotas which were not enabled for read-only mounts if
3025	 * filesystem has quota feature, so that they are updated correctly.
3026	 */
3027	return f2fs_enable_quota_files(sbi, readonly);
3028}
3029
3030static void f2fs_recover_quota_end(struct f2fs_sb_info *sbi,
3031						bool quota_enabled)
3032{
3033	if (quota_enabled)
3034		f2fs_quota_off_umount(sbi->sb);
3035
3036	if (is_sbi_flag_set(sbi, SBI_IS_WRITABLE)) {
3037		clear_sbi_flag(sbi, SBI_IS_WRITABLE);
3038		sbi->sb->s_flags |= SB_RDONLY;
3039	}
3040}
3041
3042/* Read data from quotafile */
3043static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
3044			       size_t len, loff_t off)
3045{
3046	struct inode *inode = sb_dqopt(sb)->files[type];
3047	struct address_space *mapping = inode->i_mapping;
3048	int tocopy;
3049	size_t toread;
3050	loff_t i_size = i_size_read(inode);
3051
3052	if (off > i_size)
3053		return 0;
3054
3055	if (off + len > i_size)
3056		len = i_size - off;
3057	toread = len;
3058	while (toread > 0) {
3059		struct folio *folio;
3060		size_t offset;
3061
3062repeat:
3063		folio = mapping_read_folio_gfp(mapping, off >> PAGE_SHIFT,
3064				GFP_NOFS);
3065		if (IS_ERR(folio)) {
3066			if (PTR_ERR(folio) == -ENOMEM) {
3067				memalloc_retry_wait(GFP_NOFS);
3068				goto repeat;
3069			}
3070			set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
3071			return PTR_ERR(folio);
3072		}
3073		offset = offset_in_folio(folio, off);
3074		tocopy = min(folio_size(folio) - offset, toread);
3075
3076		folio_lock(folio);
3077
3078		if (unlikely(folio->mapping != mapping)) {
3079			f2fs_folio_put(folio, true);
3080			goto repeat;
3081		}
3082
3083		/*
3084		 * should never happen, just leave f2fs_bug_on() here to catch
3085		 * any potential bug.
3086		 */
3087		f2fs_bug_on(F2FS_SB(sb), !folio_test_uptodate(folio));
3088
3089		memcpy_from_folio(data, folio, offset, tocopy);
3090		f2fs_folio_put(folio, true);
3091
3092		toread -= tocopy;
3093		data += tocopy;
3094		off += tocopy;
3095	}
3096	return len;
3097}
3098
3099/* Write to quotafile */
3100static ssize_t f2fs_quota_write(struct super_block *sb, int type,
3101				const char *data, size_t len, loff_t off)
3102{
3103	struct inode *inode = sb_dqopt(sb)->files[type];
3104	struct address_space *mapping = inode->i_mapping;
3105	const struct address_space_operations *a_ops = mapping->a_ops;
3106	int offset = off & (sb->s_blocksize - 1);
3107	size_t towrite = len;
3108	struct folio *folio;
3109	void *fsdata = NULL;
3110	int err = 0;
3111	int tocopy;
3112
3113	while (towrite > 0) {
3114		tocopy = min_t(unsigned long, sb->s_blocksize - offset,
3115								towrite);
3116retry:
3117		err = a_ops->write_begin(NULL, mapping, off, tocopy,
3118							&folio, &fsdata);
3119		if (unlikely(err)) {
3120			if (err == -ENOMEM) {
3121				memalloc_retry_wait(GFP_NOFS);
3122				goto retry;
3123			}
3124			set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
3125			break;
3126		}
3127
3128		memcpy_to_folio(folio, offset_in_folio(folio, off), data, tocopy);
3129
3130		a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
3131						folio, fsdata);
3132		offset = 0;
3133		towrite -= tocopy;
3134		off += tocopy;
3135		data += tocopy;
3136		cond_resched();
3137	}
3138
3139	if (len == towrite)
3140		return err;
3141	inode_set_mtime_to_ts(inode, inode_set_ctime_current(inode));
3142	f2fs_mark_inode_dirty_sync(inode, false);
3143	return len - towrite;
3144}
3145
3146int f2fs_dquot_initialize(struct inode *inode)
3147{
3148	if (time_to_inject(F2FS_I_SB(inode), FAULT_DQUOT_INIT))
3149		return -ESRCH;
3150
3151	return dquot_initialize(inode);
3152}
3153
3154static struct dquot __rcu **f2fs_get_dquots(struct inode *inode)
3155{
3156	return F2FS_I(inode)->i_dquot;
3157}
3158
3159static qsize_t *f2fs_get_reserved_space(struct inode *inode)
3160{
3161	return &F2FS_I(inode)->i_reserved_quota;
3162}
3163
3164static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
3165{
3166	if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
3167		f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
3168		return 0;
3169	}
3170
3171	return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
3172					F2FS_OPTION(sbi).s_jquota_fmt, type);
3173}
3174
3175int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
3176{
3177	int enabled = 0;
3178	int i, err;
3179
3180	if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
3181		err = f2fs_enable_quotas(sbi->sb);
3182		if (err) {
3183			f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
3184			return 0;
3185		}
3186		return 1;
3187	}
3188
3189	for (i = 0; i < MAXQUOTAS; i++) {
3190		if (F2FS_OPTION(sbi).s_qf_names[i]) {
3191			err = f2fs_quota_on_mount(sbi, i);
3192			if (!err) {
3193				enabled = 1;
3194				continue;
3195			}
3196			f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
3197				 err, i);
3198		}
3199	}
3200	return enabled;
3201}
3202
3203static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
3204			     unsigned int flags)
3205{
3206	struct inode *qf_inode;
3207	unsigned long qf_inum;
3208	unsigned long qf_flag = F2FS_QUOTA_DEFAULT_FL;
3209	int err;
3210
3211	BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
3212
3213	qf_inum = f2fs_qf_ino(sb, type);
3214	if (!qf_inum)
3215		return -EPERM;
3216
3217	qf_inode = f2fs_iget(sb, qf_inum);
3218	if (IS_ERR(qf_inode)) {
3219		f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
3220		return PTR_ERR(qf_inode);
3221	}
3222
3223	/* Don't account quota for quota files to avoid recursion */
3224	inode_lock(qf_inode);
3225	qf_inode->i_flags |= S_NOQUOTA;
3226
3227	if ((F2FS_I(qf_inode)->i_flags & qf_flag) != qf_flag) {
3228		F2FS_I(qf_inode)->i_flags |= qf_flag;
3229		f2fs_set_inode_flags(qf_inode);
3230	}
3231	inode_unlock(qf_inode);
3232
3233	err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
3234	iput(qf_inode);
3235	return err;
3236}
3237
3238static int f2fs_enable_quotas(struct super_block *sb)
3239{
3240	struct f2fs_sb_info *sbi = F2FS_SB(sb);
3241	int type, err = 0;
3242	unsigned long qf_inum;
3243	bool quota_mopt[MAXQUOTAS] = {
3244		test_opt(sbi, USRQUOTA),
3245		test_opt(sbi, GRPQUOTA),
3246		test_opt(sbi, PRJQUOTA),
3247	};
3248
3249	if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
3250		f2fs_err(sbi, "quota file may be corrupted, skip loading it");
3251		return 0;
3252	}
3253
3254	sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
3255
3256	for (type = 0; type < MAXQUOTAS; type++) {
3257		qf_inum = f2fs_qf_ino(sb, type);
3258		if (qf_inum) {
3259			err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
3260				DQUOT_USAGE_ENABLED |
3261				(quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
3262			if (err) {
3263				f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
3264					 type, err);
3265				for (type--; type >= 0; type--)
3266					dquot_quota_off(sb, type);
3267				set_sbi_flag(F2FS_SB(sb),
3268						SBI_QUOTA_NEED_REPAIR);
3269				return err;
3270			}
3271		}
3272	}
3273	return 0;
3274}
3275
3276static int f2fs_quota_sync_file(struct f2fs_sb_info *sbi, int type)
3277{
3278	struct quota_info *dqopt = sb_dqopt(sbi->sb);
3279	struct address_space *mapping = dqopt->files[type]->i_mapping;
3280	int ret = 0;
3281
3282	ret = dquot_writeback_dquots(sbi->sb, type);
3283	if (ret)
3284		goto out;
3285
3286	ret = filemap_fdatawrite(mapping);
3287	if (ret)
3288		goto out;
3289
3290	/* if we are using journalled quota */
3291	if (is_journalled_quota(sbi))
3292		goto out;
3293
3294	ret = filemap_fdatawait(mapping);
3295
3296	truncate_inode_pages(&dqopt->files[type]->i_data, 0);
3297out:
3298	if (ret)
3299		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3300	return ret;
3301}
3302
3303int f2fs_do_quota_sync(struct super_block *sb, int type)
3304{
3305	struct f2fs_sb_info *sbi = F2FS_SB(sb);
3306	struct quota_info *dqopt = sb_dqopt(sb);
3307	int cnt;
3308	int ret = 0;
3309
3310	/*
3311	 * Now when everything is written we can discard the pagecache so
3312	 * that userspace sees the changes.
3313	 */
3314	for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
3315
3316		if (type != -1 && cnt != type)
3317			continue;
3318
3319		if (!sb_has_quota_active(sb, cnt))
3320			continue;
3321
3322		if (!f2fs_sb_has_quota_ino(sbi))
3323			inode_lock(dqopt->files[cnt]);
3324
3325		/*
3326		 * do_quotactl
3327		 *  f2fs_quota_sync
3328		 *  f2fs_down_read(quota_sem)
3329		 *  dquot_writeback_dquots()
3330		 *  f2fs_dquot_commit
3331		 *			      block_operation
3332		 *			      f2fs_down_read(quota_sem)
3333		 */
3334		f2fs_lock_op(sbi);
3335		f2fs_down_read(&sbi->quota_sem);
3336
3337		ret = f2fs_quota_sync_file(sbi, cnt);
3338
3339		f2fs_up_read(&sbi->quota_sem);
3340		f2fs_unlock_op(sbi);
3341
3342		if (!f2fs_sb_has_quota_ino(sbi))
3343			inode_unlock(dqopt->files[cnt]);
3344
3345		if (ret)
3346			break;
3347	}
3348	return ret;
3349}
3350
3351static int f2fs_quota_sync(struct super_block *sb, int type)
3352{
3353	int ret;
3354
3355	F2FS_SB(sb)->umount_lock_holder = current;
3356	ret = f2fs_do_quota_sync(sb, type);
3357	F2FS_SB(sb)->umount_lock_holder = NULL;
3358	return ret;
3359}
3360
3361static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
3362							const struct path *path)
3363{
3364	struct inode *inode;
3365	int err = 0;
3366
3367	/* if quota sysfile exists, deny enabling quota with specific file */
3368	if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
3369		f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
3370		return -EBUSY;
3371	}
3372
3373	if (path->dentry->d_sb != sb)
3374		return -EXDEV;
3375
3376	F2FS_SB(sb)->umount_lock_holder = current;
3377
3378	err = f2fs_do_quota_sync(sb, type);
3379	if (err)
3380		goto out;
3381
3382	inode = d_inode(path->dentry);
3383
3384	err = filemap_fdatawrite(inode->i_mapping);
3385	if (err)
3386		goto out;
3387
3388	err = filemap_fdatawait(inode->i_mapping);
3389	if (err)
3390		goto out;
3391
3392	err = dquot_quota_on(sb, type, format_id, path);
3393	if (err)
3394		goto out;
3395
3396	inode_lock(inode);
3397	F2FS_I(inode)->i_flags |= F2FS_QUOTA_DEFAULT_FL;
3398	f2fs_set_inode_flags(inode);
3399	inode_unlock(inode);
3400	f2fs_mark_inode_dirty_sync(inode, false);
3401out:
3402	F2FS_SB(sb)->umount_lock_holder = NULL;
3403	return err;
3404}
3405
3406static int __f2fs_quota_off(struct super_block *sb, int type)
3407{
3408	struct inode *inode = sb_dqopt(sb)->files[type];
3409	int err;
3410
3411	if (!inode || !igrab(inode))
3412		return dquot_quota_off(sb, type);
3413
3414	err = f2fs_do_quota_sync(sb, type);
3415	if (err)
3416		goto out_put;
3417
3418	err = dquot_quota_off(sb, type);
3419	if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
3420		goto out_put;
3421
3422	inode_lock(inode);
3423	F2FS_I(inode)->i_flags &= ~F2FS_QUOTA_DEFAULT_FL;
3424	f2fs_set_inode_flags(inode);
3425	inode_unlock(inode);
3426	f2fs_mark_inode_dirty_sync(inode, false);
3427out_put:
3428	iput(inode);
3429	return err;
3430}
3431
3432static int f2fs_quota_off(struct super_block *sb, int type)
3433{
3434	struct f2fs_sb_info *sbi = F2FS_SB(sb);
3435	int err;
3436
3437	F2FS_SB(sb)->umount_lock_holder = current;
3438
3439	err = __f2fs_quota_off(sb, type);
3440
3441	/*
3442	 * quotactl can shutdown journalled quota, result in inconsistence
3443	 * between quota record and fs data by following updates, tag the
3444	 * flag to let fsck be aware of it.
3445	 */
3446	if (is_journalled_quota(sbi))
3447		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3448
3449	F2FS_SB(sb)->umount_lock_holder = NULL;
3450
3451	return err;
3452}
3453
3454void f2fs_quota_off_umount(struct super_block *sb)
3455{
3456	int type;
3457	int err;
3458
3459	for (type = 0; type < MAXQUOTAS; type++) {
3460		err = __f2fs_quota_off(sb, type);
3461		if (err) {
3462			int ret = dquot_quota_off(sb, type);
3463
3464			f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
3465				 type, err, ret);
3466			set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
3467		}
3468	}
3469	/*
3470	 * In case of checkpoint=disable, we must flush quota blocks.
3471	 * This can cause NULL exception for node_inode in end_io, since
3472	 * put_super already dropped it.
3473	 */
3474	sync_filesystem(sb);
3475}
3476
3477static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
3478{
3479	struct quota_info *dqopt = sb_dqopt(sb);
3480	int type;
3481
3482	for (type = 0; type < MAXQUOTAS; type++) {
3483		if (!dqopt->files[type])
3484			continue;
3485		f2fs_inode_synced(dqopt->files[type]);
3486	}
3487}
3488
3489static int f2fs_dquot_commit(struct dquot *dquot)
3490{
3491	struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3492	int ret;
3493
3494	f2fs_down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
3495	ret = dquot_commit(dquot);
3496	if (ret < 0)
3497		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3498	f2fs_up_read(&sbi->quota_sem);
3499	return ret;
3500}
3501
3502static int f2fs_dquot_acquire(struct dquot *dquot)
3503{
3504	struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3505	int ret;
3506
3507	f2fs_down_read(&sbi->quota_sem);
3508	ret = dquot_acquire(dquot);
3509	if (ret < 0)
3510		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3511	f2fs_up_read(&sbi->quota_sem);
3512	return ret;
3513}
3514
3515static int f2fs_dquot_release(struct dquot *dquot)
3516{
3517	struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3518	int ret = dquot_release(dquot);
3519
3520	if (ret < 0)
3521		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3522	return ret;
3523}
3524
3525static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
3526{
3527	struct super_block *sb = dquot->dq_sb;
3528	struct f2fs_sb_info *sbi = F2FS_SB(sb);
3529	int ret = dquot_mark_dquot_dirty(dquot);
3530
3531	/* if we are using journalled quota */
3532	if (is_journalled_quota(sbi))
3533		set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
3534
3535	return ret;
3536}
3537
3538static int f2fs_dquot_commit_info(struct super_block *sb, int type)
3539{
3540	struct f2fs_sb_info *sbi = F2FS_SB(sb);
3541	int ret = dquot_commit_info(sb, type);
3542
3543	if (ret < 0)
3544		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3545	return ret;
3546}
3547
3548static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
3549{
3550	*projid = F2FS_I(inode)->i_projid;
3551	return 0;
3552}
3553
3554static const struct dquot_operations f2fs_quota_operations = {
3555	.get_reserved_space = f2fs_get_reserved_space,
3556	.write_dquot	= f2fs_dquot_commit,
3557	.acquire_dquot	= f2fs_dquot_acquire,
3558	.release_dquot	= f2fs_dquot_release,
3559	.mark_dirty	= f2fs_dquot_mark_dquot_dirty,
3560	.write_info	= f2fs_dquot_commit_info,
3561	.alloc_dquot	= dquot_alloc,
3562	.destroy_dquot	= dquot_destroy,
3563	.get_projid	= f2fs_get_projid,
3564	.get_next_id	= dquot_get_next_id,
3565};
3566
3567static const struct quotactl_ops f2fs_quotactl_ops = {
3568	.quota_on	= f2fs_quota_on,
3569	.quota_off	= f2fs_quota_off,
3570	.quota_sync	= f2fs_quota_sync,
3571	.get_state	= dquot_get_state,
3572	.set_info	= dquot_set_dqinfo,
3573	.get_dqblk	= dquot_get_dqblk,
3574	.set_dqblk	= dquot_set_dqblk,
3575	.get_nextdqblk	= dquot_get_next_dqblk,
3576};
3577#else
3578int f2fs_dquot_initialize(struct inode *inode)
3579{
3580	return 0;
3581}
3582
3583int f2fs_do_quota_sync(struct super_block *sb, int type)
3584{
3585	return 0;
3586}
3587
3588void f2fs_quota_off_umount(struct super_block *sb)
3589{
3590}
3591#endif
3592
3593static const struct super_operations f2fs_sops = {
3594	.alloc_inode	= f2fs_alloc_inode,
3595	.free_inode	= f2fs_free_inode,
3596	.drop_inode	= f2fs_drop_inode,
3597	.write_inode	= f2fs_write_inode,
3598	.dirty_inode	= f2fs_dirty_inode,
3599	.show_options	= f2fs_show_options,
3600#ifdef CONFIG_QUOTA
3601	.quota_read	= f2fs_quota_read,
3602	.quota_write	= f2fs_quota_write,
3603	.get_dquots	= f2fs_get_dquots,
3604#endif
3605	.evict_inode	= f2fs_evict_inode,
3606	.put_super	= f2fs_put_super,
3607	.sync_fs	= f2fs_sync_fs,
3608	.freeze_fs	= f2fs_freeze,
3609	.unfreeze_fs	= f2fs_unfreeze,
3610	.statfs		= f2fs_statfs,
3611	.shutdown	= f2fs_shutdown,
3612};
3613
3614#ifdef CONFIG_FS_ENCRYPTION
3615static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
3616{
3617	return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3618				F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3619				ctx, len, NULL);
3620}
3621
3622static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
3623							void *fs_data)
3624{
3625	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3626
3627	/*
3628	 * Encrypting the root directory is not allowed because fsck
3629	 * expects lost+found directory to exist and remain unencrypted
3630	 * if LOST_FOUND feature is enabled.
3631	 *
3632	 */
3633	if (f2fs_sb_has_lost_found(sbi) &&
3634			inode->i_ino == F2FS_ROOT_INO(sbi))
3635		return -EPERM;
3636
3637	return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3638				F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3639				ctx, len, fs_data, XATTR_CREATE);
3640}
3641
3642static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
3643{
3644	return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
3645}
3646
3647static bool f2fs_has_stable_inodes(struct super_block *sb)
3648{
3649	return true;
3650}
3651
3652static struct block_device **f2fs_get_devices(struct super_block *sb,
3653					      unsigned int *num_devs)
3654{
3655	struct f2fs_sb_info *sbi = F2FS_SB(sb);
3656	struct block_device **devs;
3657	int i;
3658
3659	if (!f2fs_is_multi_device(sbi))
3660		return NULL;
3661
3662	devs = kmalloc_array(sbi->s_ndevs, sizeof(*devs), GFP_KERNEL);
3663	if (!devs)
3664		return ERR_PTR(-ENOMEM);
3665
3666	for (i = 0; i < sbi->s_ndevs; i++)
3667		devs[i] = FDEV(i).bdev;
3668	*num_devs = sbi->s_ndevs;
3669	return devs;
3670}
3671
3672static const struct fscrypt_operations f2fs_cryptops = {
3673	.inode_info_offs	= (int)offsetof(struct f2fs_inode_info, i_crypt_info) -
3674				  (int)offsetof(struct f2fs_inode_info, vfs_inode),
3675	.needs_bounce_pages	= 1,
3676	.has_32bit_inodes	= 1,
3677	.supports_subblock_data_units = 1,
3678	.legacy_key_prefix	= "f2fs:",
3679	.get_context		= f2fs_get_context,
3680	.set_context		= f2fs_set_context,
3681	.get_dummy_policy	= f2fs_get_dummy_policy,
3682	.empty_dir		= f2fs_empty_dir,
3683	.has_stable_inodes	= f2fs_has_stable_inodes,
3684	.get_devices		= f2fs_get_devices,
3685};
3686#endif /* CONFIG_FS_ENCRYPTION */
3687
3688static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
3689		u64 ino, u32 generation)
3690{
3691	struct f2fs_sb_info *sbi = F2FS_SB(sb);
3692	struct inode *inode;
3693
3694	if (f2fs_check_nid_range(sbi, ino))
3695		return ERR_PTR(-ESTALE);
3696
3697	/*
3698	 * f2fs_iget isn't quite right if the inode is currently unallocated!
3699	 * However f2fs_iget currently does appropriate checks to handle stale
3700	 * inodes so everything is OK.
3701	 */
3702	inode = f2fs_iget(sb, ino);
3703	if (IS_ERR(inode))
3704		return ERR_CAST(inode);
3705	if (unlikely(generation && inode->i_generation != generation)) {
3706		/* we didn't find the right inode.. */
3707		iput(inode);
3708		return ERR_PTR(-ESTALE);
3709	}
3710	return inode;
3711}
3712
3713static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
3714		int fh_len, int fh_type)
3715{
3716	return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
3717				    f2fs_nfs_get_inode);
3718}
3719
3720static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
3721		int fh_len, int fh_type)
3722{
3723	return generic_fh_to_parent(sb, fid, fh_len, fh_type,
3724				    f2fs_nfs_get_inode);
3725}
3726
3727static const struct export_operations f2fs_export_ops = {
3728	.encode_fh = generic_encode_ino32_fh,
3729	.fh_to_dentry = f2fs_fh_to_dentry,
3730	.fh_to_parent = f2fs_fh_to_parent,
3731	.get_parent = f2fs_get_parent,
3732};
3733
3734loff_t max_file_blocks(struct inode *inode)
3735{
3736	loff_t result = 0;
3737	loff_t leaf_count;
3738
3739	/*
3740	 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
3741	 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
3742	 * space in inode.i_addr, it will be more safe to reassign
3743	 * result as zero.
3744	 */
3745
3746	if (inode && f2fs_compressed_file(inode))
3747		leaf_count = ADDRS_PER_BLOCK(inode);
3748	else
3749		leaf_count = DEF_ADDRS_PER_BLOCK;
3750
3751	/* two direct node blocks */
3752	result += (leaf_count * 2);
3753
3754	/* two indirect node blocks */
3755	leaf_count *= NIDS_PER_BLOCK;
3756	result += (leaf_count * 2);
3757
3758	/* one double indirect node block */
3759	leaf_count *= NIDS_PER_BLOCK;
3760	result += leaf_count;
3761
3762	/*
3763	 * For compatibility with FSCRYPT_POLICY_FLAG_IV_INO_LBLK_{64,32} with
3764	 * a 4K crypto data unit, we must restrict the max filesize to what can
3765	 * fit within U32_MAX + 1 data units.
3766	 */
3767
3768	result = umin(result, F2FS_BYTES_TO_BLK(((loff_t)U32_MAX + 1) * 4096));
3769
3770	return result;
3771}
3772
3773static int __f2fs_commit_super(struct f2fs_sb_info *sbi, struct folio *folio,
3774						pgoff_t index, bool update)
3775{
3776	struct bio *bio;
3777	/* it's rare case, we can do fua all the time */
3778	blk_opf_t opf = REQ_OP_WRITE | REQ_SYNC | REQ_PREFLUSH | REQ_FUA;
3779	int ret;
3780
3781	folio_lock(folio);
3782	folio_wait_writeback(folio);
3783	if (update)
3784		memcpy(F2FS_SUPER_BLOCK(folio, index), F2FS_RAW_SUPER(sbi),
3785					sizeof(struct f2fs_super_block));
3786	folio_mark_dirty(folio);
3787	folio_clear_dirty_for_io(folio);
3788	folio_start_writeback(folio);
3789	folio_unlock(folio);
3790
3791	bio = bio_alloc(sbi->sb->s_bdev, 1, opf, GFP_NOFS);
3792
3793	/* it doesn't need to set crypto context for superblock update */
3794	bio->bi_iter.bi_sector = SECTOR_FROM_BLOCK(folio->index);
3795
3796	if (!bio_add_folio(bio, folio, folio_size(folio), 0))
3797		f2fs_bug_on(sbi, 1);
3798
3799	ret = submit_bio_wait(bio);
3800	bio_put(bio);
3801	folio_end_writeback(folio);
3802
3803	return ret;
3804}
3805
3806static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
3807					struct folio *folio, pgoff_t index)
3808{
3809	struct f2fs_super_block *raw_super = F2FS_SUPER_BLOCK(folio, index);
3810	struct super_block *sb = sbi->sb;
3811	u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
3812	u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
3813	u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
3814	u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
3815	u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
3816	u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
3817	u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
3818	u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
3819	u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
3820	u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
3821	u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3822	u32 segment_count = le32_to_cpu(raw_super->segment_count);
3823	u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3824	u64 main_end_blkaddr = main_blkaddr +
3825				((u64)segment_count_main << log_blocks_per_seg);
3826	u64 seg_end_blkaddr = segment0_blkaddr +
3827				((u64)segment_count << log_blocks_per_seg);
3828
3829	if (segment0_blkaddr != cp_blkaddr) {
3830		f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
3831			  segment0_blkaddr, cp_blkaddr);
3832		return true;
3833	}
3834
3835	if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
3836							sit_blkaddr) {
3837		f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
3838			  cp_blkaddr, sit_blkaddr,
3839			  segment_count_ckpt << log_blocks_per_seg);
3840		return true;
3841	}
3842
3843	if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
3844							nat_blkaddr) {
3845		f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
3846			  sit_blkaddr, nat_blkaddr,
3847			  segment_count_sit << log_blocks_per_seg);
3848		return true;
3849	}
3850
3851	if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
3852							ssa_blkaddr) {
3853		f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
3854			  nat_blkaddr, ssa_blkaddr,
3855			  segment_count_nat << log_blocks_per_seg);
3856		return true;
3857	}
3858
3859	if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
3860							main_blkaddr) {
3861		f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
3862			  ssa_blkaddr, main_blkaddr,
3863			  segment_count_ssa << log_blocks_per_seg);
3864		return true;
3865	}
3866
3867	if (main_end_blkaddr > seg_end_blkaddr) {
3868		f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
3869			  main_blkaddr, seg_end_blkaddr,
3870			  segment_count_main << log_blocks_per_seg);
3871		return true;
3872	} else if (main_end_blkaddr < seg_end_blkaddr) {
3873		int err = 0;
3874		char *res;
3875
3876		/* fix in-memory information all the time */
3877		raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
3878				segment0_blkaddr) >> log_blocks_per_seg);
3879
3880		if (f2fs_readonly(sb) || f2fs_hw_is_readonly(sbi)) {
3881			set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3882			res = "internally";
3883		} else {
3884			err = __f2fs_commit_super(sbi, folio, index, false);
3885			res = err ? "failed" : "done";
3886		}
3887		f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
3888			  res, main_blkaddr, seg_end_blkaddr,
3889			  segment_count_main << log_blocks_per_seg);
3890		if (err)
3891			return true;
3892	}
3893	return false;
3894}
3895
3896static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
3897					struct folio *folio, pgoff_t index)
3898{
3899	block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
3900	block_t total_sections, blocks_per_seg;
3901	struct f2fs_super_block *raw_super = F2FS_SUPER_BLOCK(folio, index);
3902	size_t crc_offset = 0;
3903	__u32 crc = 0;
3904
3905	if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
3906		f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
3907			  F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
3908		return -EINVAL;
3909	}
3910
3911	/* Check checksum_offset and crc in superblock */
3912	if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
3913		crc_offset = le32_to_cpu(raw_super->checksum_offset);
3914		if (crc_offset !=
3915			offsetof(struct f2fs_super_block, crc)) {
3916			f2fs_info(sbi, "Invalid SB checksum offset: %zu",
3917				  crc_offset);
3918			return -EFSCORRUPTED;
3919		}
3920		crc = le32_to_cpu(raw_super->crc);
3921		if (crc != f2fs_crc32(raw_super, crc_offset)) {
3922			f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
3923			return -EFSCORRUPTED;
3924		}
3925	}
3926
3927	/* only support block_size equals to PAGE_SIZE */
3928	if (le32_to_cpu(raw_super->log_blocksize) != F2FS_BLKSIZE_BITS) {
3929		f2fs_info(sbi, "Invalid log_blocksize (%u), supports only %u",
3930			  le32_to_cpu(raw_super->log_blocksize),
3931			  F2FS_BLKSIZE_BITS);
3932		return -EFSCORRUPTED;
3933	}
3934
3935	/* check log blocks per segment */
3936	if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
3937		f2fs_info(sbi, "Invalid log blocks per segment (%u)",
3938			  le32_to_cpu(raw_super->log_blocks_per_seg));
3939		return -EFSCORRUPTED;
3940	}
3941
3942	/* Currently, support 512/1024/2048/4096/16K bytes sector size */
3943	if (le32_to_cpu(raw_super->log_sectorsize) >
3944				F2FS_MAX_LOG_SECTOR_SIZE ||
3945		le32_to_cpu(raw_super->log_sectorsize) <
3946				F2FS_MIN_LOG_SECTOR_SIZE) {
3947		f2fs_info(sbi, "Invalid log sectorsize (%u)",
3948			  le32_to_cpu(raw_super->log_sectorsize));
3949		return -EFSCORRUPTED;
3950	}
3951	if (le32_to_cpu(raw_super->log_sectors_per_block) +
3952		le32_to_cpu(raw_super->log_sectorsize) !=
3953			F2FS_MAX_LOG_SECTOR_SIZE) {
3954		f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
3955			  le32_to_cpu(raw_super->log_sectors_per_block),
3956			  le32_to_cpu(raw_super->log_sectorsize));
3957		return -EFSCORRUPTED;
3958	}
3959
3960	segment_count = le32_to_cpu(raw_super->segment_count);
3961	segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3962	segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3963	secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3964	total_sections = le32_to_cpu(raw_super->section_count);
3965
3966	/* blocks_per_seg should be 512, given the above check */
3967	blocks_per_seg = BIT(le32_to_cpu(raw_super->log_blocks_per_seg));
3968
3969	if (segment_count > F2FS_MAX_SEGMENT ||
3970				segment_count < F2FS_MIN_SEGMENTS) {
3971		f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
3972		return -EFSCORRUPTED;
3973	}
3974
3975	if (total_sections > segment_count_main || total_sections < 1 ||
3976			segs_per_sec > segment_count || !segs_per_sec) {
3977		f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
3978			  segment_count, total_sections, segs_per_sec);
3979		return -EFSCORRUPTED;
3980	}
3981
3982	if (segment_count_main != total_sections * segs_per_sec) {
3983		f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
3984			  segment_count_main, total_sections, segs_per_sec);
3985		return -EFSCORRUPTED;
3986	}
3987
3988	if ((segment_count / segs_per_sec) < total_sections) {
3989		f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
3990			  segment_count, segs_per_sec, total_sections);
3991		return -EFSCORRUPTED;
3992	}
3993
3994	if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
3995		f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
3996			  segment_count, le64_to_cpu(raw_super->block_count));
3997		return -EFSCORRUPTED;
3998	}
3999
4000	if (RDEV(0).path[0]) {
4001		block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
4002		int i = 1;
4003
4004		while (i < MAX_DEVICES && RDEV(i).path[0]) {
4005			dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
4006			i++;
4007		}
4008		if (segment_count != dev_seg_count) {
4009			f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
4010					segment_count, dev_seg_count);
4011			return -EFSCORRUPTED;
4012		}
4013	} else {
4014		if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
4015					!bdev_is_zoned(sbi->sb->s_bdev)) {
4016			f2fs_info(sbi, "Zoned block device path is missing");
4017			return -EFSCORRUPTED;
4018		}
4019	}
4020
4021	if (secs_per_zone > total_sections || !secs_per_zone) {
4022		f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
4023			  secs_per_zone, total_sections);
4024		return -EFSCORRUPTED;
4025	}
4026	if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
4027			raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
4028			(le32_to_cpu(raw_super->extension_count) +
4029			raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
4030		f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
4031			  le32_to_cpu(raw_super->extension_count),
4032			  raw_super->hot_ext_count,
4033			  F2FS_MAX_EXTENSION);
4034		return -EFSCORRUPTED;
4035	}
4036
4037	if (le32_to_cpu(raw_super->cp_payload) >=
4038				(blocks_per_seg - F2FS_CP_PACKS -
4039				NR_CURSEG_PERSIST_TYPE)) {
4040		f2fs_info(sbi, "Insane cp_payload (%u >= %u)",
4041			  le32_to_cpu(raw_super->cp_payload),
4042			  blocks_per_seg - F2FS_CP_PACKS -
4043			  NR_CURSEG_PERSIST_TYPE);
4044		return -EFSCORRUPTED;
4045	}
4046
4047	/* check reserved ino info */
4048	if (le32_to_cpu(raw_super->node_ino) != 1 ||
4049		le32_to_cpu(raw_super->meta_ino) != 2 ||
4050		le32_to_cpu(raw_super->root_ino) != 3) {
4051		f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
4052			  le32_to_cpu(raw_super->node_ino),
4053			  le32_to_cpu(raw_super->meta_ino),
4054			  le32_to_cpu(raw_super->root_ino));
4055		return -EFSCORRUPTED;
4056	}
4057
4058	/* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
4059	if (sanity_check_area_boundary(sbi, folio, index))
4060		return -EFSCORRUPTED;
4061
4062	return 0;
4063}
4064
4065int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
4066{
4067	unsigned int total, fsmeta;
4068	struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
4069	struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
4070	unsigned int ovp_segments, reserved_segments;
4071	unsigned int main_segs, blocks_per_seg;
4072	unsigned int sit_segs, nat_segs;
4073	unsigned int sit_bitmap_size, nat_bitmap_size;
4074	unsigned int log_blocks_per_seg;
4075	unsigned int segment_count_main;
4076	unsigned int cp_pack_start_sum, cp_payload;
4077	block_t user_block_count, valid_user_blocks;
4078	block_t avail_node_count, valid_node_count;
4079	unsigned int nat_blocks, nat_bits_bytes, nat_bits_blocks;
4080	unsigned int sit_blk_cnt;
4081	int i, j;
4082
4083	total = le32_to_cpu(raw_super->segment_count);
4084	fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
4085	sit_segs = le32_to_cpu(raw_super->segment_count_sit);
4086	fsmeta += sit_segs;
4087	nat_segs = le32_to_cpu(raw_super->segment_count_nat);
4088	fsmeta += nat_segs;
4089	fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
4090	fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
4091
4092	if (unlikely(fsmeta >= total))
4093		return 1;
4094
4095	ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
4096	reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
4097
4098	if (!f2fs_sb_has_readonly(sbi) &&
4099			unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
4100			ovp_segments == 0 || reserved_segments == 0)) {
4101		f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
4102		return 1;
4103	}
4104	user_block_count = le64_to_cpu(ckpt->user_block_count);
4105	segment_count_main = le32_to_cpu(raw_super->segment_count_main) +
4106			(f2fs_sb_has_readonly(sbi) ? 1 : 0);
4107	log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
4108	if (!user_block_count || user_block_count >=
4109			segment_count_main << log_blocks_per_seg) {
4110		f2fs_err(sbi, "Wrong user_block_count: %u",
4111			 user_block_count);
4112		return 1;
4113	}
4114
4115	valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
4116	if (valid_user_blocks > user_block_count) {
4117		f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
4118			 valid_user_blocks, user_block_count);
4119		return 1;
4120	}
4121
4122	valid_node_count = le32_to_cpu(ckpt->valid_node_count);
4123	avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
4124	if (valid_node_count > avail_node_count) {
4125		f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
4126			 valid_node_count, avail_node_count);
4127		return 1;
4128	}
4129
4130	main_segs = le32_to_cpu(raw_super->segment_count_main);
4131	blocks_per_seg = BLKS_PER_SEG(sbi);
4132
4133	for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
4134		if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
4135			le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
4136			return 1;
4137
4138		if (f2fs_sb_has_readonly(sbi))
4139			goto check_data;
4140
4141		for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
4142			if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
4143				le32_to_cpu(ckpt->cur_node_segno[j])) {
4144				f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
4145					 i, j,
4146					 le32_to_cpu(ckpt->cur_node_segno[i]));
4147				return 1;
4148			}
4149		}
4150	}
4151check_data:
4152	for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
4153		if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
4154			le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
4155			return 1;
4156
4157		if (f2fs_sb_has_readonly(sbi))
4158			goto skip_cross;
4159
4160		for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
4161			if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
4162				le32_to_cpu(ckpt->cur_data_segno[j])) {
4163				f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
4164					 i, j,
4165					 le32_to_cpu(ckpt->cur_data_segno[i]));
4166				return 1;
4167			}
4168		}
4169	}
4170	for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
4171		for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
4172			if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
4173				le32_to_cpu(ckpt->cur_data_segno[j])) {
4174				f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
4175					 i, j,
4176					 le32_to_cpu(ckpt->cur_node_segno[i]));
4177				return 1;
4178			}
4179		}
4180	}
4181skip_cross:
4182	sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
4183	nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
4184
4185	if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
4186		nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
4187		f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
4188			 sit_bitmap_size, nat_bitmap_size);
4189		return 1;
4190	}
4191
4192	sit_blk_cnt = DIV_ROUND_UP(main_segs, SIT_ENTRY_PER_BLOCK);
4193	if (sit_bitmap_size * 8 < sit_blk_cnt) {
4194		f2fs_err(sbi, "Wrong bitmap size: sit: %u, sit_blk_cnt:%u",
4195			 sit_bitmap_size, sit_blk_cnt);
4196		return 1;
4197	}
4198
4199	cp_pack_start_sum = __start_sum_addr(sbi);
4200	cp_payload = __cp_payload(sbi);
4201	if (cp_pack_start_sum < cp_payload + 1 ||
4202		cp_pack_start_sum > blocks_per_seg - 1 -
4203			NR_CURSEG_PERSIST_TYPE) {
4204		f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
4205			 cp_pack_start_sum);
4206		return 1;
4207	}
4208
4209	if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
4210		le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
4211		f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
4212			  "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
4213			  "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
4214			  le32_to_cpu(ckpt->checksum_offset));
4215		return 1;
4216	}
4217
4218	nat_blocks = nat_segs << log_blocks_per_seg;
4219	nat_bits_bytes = nat_blocks / BITS_PER_BYTE;
4220	nat_bits_blocks = F2FS_BLK_ALIGN((nat_bits_bytes << 1) + 8);
4221	if (__is_set_ckpt_flags(ckpt, CP_NAT_BITS_FLAG) &&
4222		(cp_payload + F2FS_CP_PACKS +
4223		NR_CURSEG_PERSIST_TYPE + nat_bits_blocks >= blocks_per_seg)) {
4224		f2fs_warn(sbi, "Insane cp_payload: %u, nat_bits_blocks: %u)",
4225			  cp_payload, nat_bits_blocks);
4226		return 1;
4227	}
4228
4229	if (unlikely(f2fs_cp_error(sbi))) {
4230		f2fs_err(sbi, "A bug case: need to run fsck");
4231		return 1;
4232	}
4233	return 0;
4234}
4235
4236static void init_sb_info(struct f2fs_sb_info *sbi)
4237{
4238	struct f2fs_super_block *raw_super = sbi->raw_super;
4239	int i;
4240
4241	sbi->log_sectors_per_block =
4242		le32_to_cpu(raw_super->log_sectors_per_block);
4243	sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
4244	sbi->blocksize = BIT(sbi->log_blocksize);
4245	sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
4246	sbi->blocks_per_seg = BIT(sbi->log_blocks_per_seg);
4247	sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
4248	sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
4249	sbi->total_sections = le32_to_cpu(raw_super->section_count);
4250	sbi->total_node_count = SEGS_TO_BLKS(sbi,
4251			((le32_to_cpu(raw_super->segment_count_nat) / 2) *
4252			NAT_ENTRY_PER_BLOCK));
4253	sbi->allocate_section_hint = le32_to_cpu(raw_super->section_count);
4254	sbi->allocate_section_policy = ALLOCATE_FORWARD_NOHINT;
4255	F2FS_ROOT_INO(sbi) = le32_to_cpu(raw_super->root_ino);
4256	F2FS_NODE_INO(sbi) = le32_to_cpu(raw_super->node_ino);
4257	F2FS_META_INO(sbi) = le32_to_cpu(raw_super->meta_ino);
4258	sbi->cur_victim_sec = NULL_SECNO;
4259	sbi->gc_mode = GC_NORMAL;
4260	sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
4261	sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
4262	sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
4263	sbi->migration_granularity = SEGS_PER_SEC(sbi);
4264	sbi->migration_window_granularity = f2fs_sb_has_blkzoned(sbi) ?
4265		DEF_MIGRATION_WINDOW_GRANULARITY_ZONED : SEGS_PER_SEC(sbi);
4266	sbi->seq_file_ra_mul = MIN_RA_MUL;
4267	sbi->max_fragment_chunk = DEF_FRAGMENT_SIZE;
4268	sbi->max_fragment_hole = DEF_FRAGMENT_SIZE;
4269	spin_lock_init(&sbi->gc_remaining_trials_lock);
4270	atomic64_set(&sbi->current_atomic_write, 0);
4271
4272	sbi->sum_blocksize = f2fs_sb_has_packed_ssa(sbi) ?
4273		4096 : sbi->blocksize;
4274	sbi->sums_per_block = sbi->blocksize / sbi->sum_blocksize;
4275	sbi->entries_in_sum = sbi->sum_blocksize / 8;
4276	sbi->sum_entry_size = SUMMARY_SIZE * sbi->entries_in_sum;
4277	sbi->sum_journal_size = sbi->sum_blocksize - SUM_FOOTER_SIZE -
4278		sbi->sum_entry_size;
4279	sbi->nat_journal_entries = (sbi->sum_journal_size - 2) /
4280		sizeof(struct nat_journal_entry);
4281	sbi->sit_journal_entries = (sbi->sum_journal_size - 2) /
4282		sizeof(struct sit_journal_entry);
4283
4284	sbi->dir_level = DEF_DIR_LEVEL;
4285	sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
4286	sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
4287	sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
4288	sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
4289	sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
4290	sbi->interval_time[ENABLE_TIME] = DEF_ENABLE_INTERVAL;
4291	sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
4292				DEF_UMOUNT_DISCARD_TIMEOUT;
4293	clear_sbi_flag(sbi, SBI_NEED_FSCK);
4294
4295	for (i = 0; i < NR_COUNT_TYPE; i++)
4296		atomic_set(&sbi->nr_pages[i], 0);
4297
4298	for (i = 0; i < META; i++)
4299		atomic_set(&sbi->wb_sync_req[i], 0);
4300
4301	INIT_LIST_HEAD(&sbi->s_list);
4302	mutex_init(&sbi->umount_mutex);
4303	init_f2fs_rwsem(&sbi->io_order_lock);
4304	spin_lock_init(&sbi->cp_lock);
4305
4306	sbi->dirty_device = 0;
4307	spin_lock_init(&sbi->dev_lock);
4308
4309	init_f2fs_rwsem(&sbi->sb_lock);
4310	init_f2fs_rwsem(&sbi->pin_sem);
4311}
4312
4313static int init_percpu_info(struct f2fs_sb_info *sbi)
4314{
4315	int err;
4316
4317	err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
4318	if (err)
4319		return err;
4320
4321	err = percpu_counter_init(&sbi->rf_node_block_count, 0, GFP_KERNEL);
4322	if (err)
4323		goto err_valid_block;
4324
4325	err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
4326								GFP_KERNEL);
4327	if (err)
4328		goto err_node_block;
4329	return 0;
4330
4331err_node_block:
4332	percpu_counter_destroy(&sbi->rf_node_block_count);
4333err_valid_block:
4334	percpu_counter_destroy(&sbi->alloc_valid_block_count);
4335	return err;
4336}
4337
4338#ifdef CONFIG_BLK_DEV_ZONED
4339
4340struct f2fs_report_zones_args {
4341	struct f2fs_sb_info *sbi;
4342	struct f2fs_dev_info *dev;
4343};
4344
4345static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
4346			      void *data)
4347{
4348	struct f2fs_report_zones_args *rz_args = data;
4349	block_t unusable_blocks = (zone->len - zone->capacity) >>
4350					F2FS_LOG_SECTORS_PER_BLOCK;
4351
4352	if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
4353		return 0;
4354
4355	set_bit(idx, rz_args->dev->blkz_seq);
4356	if (!rz_args->sbi->unusable_blocks_per_sec) {
4357		rz_args->sbi->unusable_blocks_per_sec = unusable_blocks;
4358		return 0;
4359	}
4360	if (rz_args->sbi->unusable_blocks_per_sec != unusable_blocks) {
4361		f2fs_err(rz_args->sbi, "F2FS supports single zone capacity\n");
4362		return -EINVAL;
4363	}
4364	return 0;
4365}
4366
4367static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
4368{
4369	struct block_device *bdev = FDEV(devi).bdev;
4370	sector_t nr_sectors = bdev_nr_sectors(bdev);
4371	struct f2fs_report_zones_args rep_zone_arg;
4372	u64 zone_sectors;
4373	unsigned int max_open_zones;
4374	int ret;
4375
4376	if (!f2fs_sb_has_blkzoned(sbi))
4377		return 0;
4378
4379	if (bdev_is_zoned(FDEV(devi).bdev)) {
4380		max_open_zones = bdev_max_open_zones(bdev);
4381		if (max_open_zones && (max_open_zones < sbi->max_open_zones))
4382			sbi->max_open_zones = max_open_zones;
4383		if (sbi->max_open_zones < F2FS_OPTION(sbi).active_logs) {
4384			f2fs_err(sbi,
4385				"zoned: max open zones %u is too small, need at least %u open zones",
4386				sbi->max_open_zones, F2FS_OPTION(sbi).active_logs);
4387			return -EINVAL;
4388		}
4389	}
4390
4391	zone_sectors = bdev_zone_sectors(bdev);
4392	if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
4393				SECTOR_TO_BLOCK(zone_sectors))
4394		return -EINVAL;
4395	sbi->blocks_per_blkz = SECTOR_TO_BLOCK(zone_sectors);
4396	FDEV(devi).nr_blkz = div_u64(SECTOR_TO_BLOCK(nr_sectors),
4397					sbi->blocks_per_blkz);
4398	if (nr_sectors & (zone_sectors - 1))
4399		FDEV(devi).nr_blkz++;
4400
4401	FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
4402					BITS_TO_LONGS(FDEV(devi).nr_blkz)
4403					* sizeof(unsigned long),
4404					GFP_KERNEL);
4405	if (!FDEV(devi).blkz_seq)
4406		return -ENOMEM;
4407
4408	rep_zone_arg.sbi = sbi;
4409	rep_zone_arg.dev = &FDEV(devi);
4410
4411	ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
4412				  &rep_zone_arg);
4413	if (ret < 0)
4414		return ret;
4415	return 0;
4416}
4417#endif
4418
4419/*
4420 * Read f2fs raw super block.
4421 * Because we have two copies of super block, so read both of them
4422 * to get the first valid one. If any one of them is broken, we pass
4423 * them recovery flag back to the caller.
4424 */
4425static int read_raw_super_block(struct f2fs_sb_info *sbi,
4426			struct f2fs_super_block **raw_super,
4427			int *valid_super_block, int *recovery)
4428{
4429	struct super_block *sb = sbi->sb;
4430	int block;
4431	struct folio *folio;
4432	struct f2fs_super_block *super;
4433	int err = 0;
4434
4435	super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
4436	if (!super)
4437		return -ENOMEM;
4438
4439	for (block = 0; block < 2; block++) {
4440		folio = read_mapping_folio(sb->s_bdev->bd_mapping, block, NULL);
4441		if (IS_ERR(folio)) {
4442			f2fs_err(sbi, "Unable to read %dth superblock",
4443				 block + 1);
4444			err = PTR_ERR(folio);
4445			*recovery = 1;
4446			continue;
4447		}
4448
4449		/* sanity checking of raw super */
4450		err = sanity_check_raw_super(sbi, folio, block);
4451		if (err) {
4452			f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
4453				 block + 1);
4454			folio_put(folio);
4455			*recovery = 1;
4456			continue;
4457		}
4458
4459		if (!*raw_super) {
4460			memcpy(super, F2FS_SUPER_BLOCK(folio, block),
4461							sizeof(*super));
4462			*valid_super_block = block;
4463			*raw_super = super;
4464		}
4465		folio_put(folio);
4466	}
4467
4468	/* No valid superblock */
4469	if (!*raw_super)
4470		kfree(super);
4471	else
4472		err = 0;
4473
4474	return err;
4475}
4476
4477int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
4478{
4479	struct folio *folio;
4480	pgoff_t index;
4481	__u32 crc = 0;
4482	int err;
4483
4484	if ((recover && f2fs_readonly(sbi->sb)) ||
4485				f2fs_hw_is_readonly(sbi)) {
4486		set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
4487		return -EROFS;
4488	}
4489
4490	/* we should update superblock crc here */
4491	if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
4492		crc = f2fs_crc32(F2FS_RAW_SUPER(sbi),
4493				offsetof(struct f2fs_super_block, crc));
4494		F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
4495	}
4496
4497	/* write back-up superblock first */
4498	index = sbi->valid_super_block ? 0 : 1;
4499	folio = read_mapping_folio(sbi->sb->s_bdev->bd_mapping, index, NULL);
4500	if (IS_ERR(folio))
4501		return PTR_ERR(folio);
4502	err = __f2fs_commit_super(sbi, folio, index, true);
4503	folio_put(folio);
4504
4505	/* if we are in recovery path, skip writing valid superblock */
4506	if (recover || err)
4507		return err;
4508
4509	/* write current valid superblock */
4510	index = sbi->valid_super_block;
4511	folio = read_mapping_folio(sbi->sb->s_bdev->bd_mapping, index, NULL);
4512	if (IS_ERR(folio))
4513		return PTR_ERR(folio);
4514	err = __f2fs_commit_super(sbi, folio, index, true);
4515	folio_put(folio);
4516	return err;
4517}
4518
4519static void save_stop_reason(struct f2fs_sb_info *sbi, unsigned char reason)
4520{
4521	unsigned long flags;
4522
4523	spin_lock_irqsave(&sbi->error_lock, flags);
4524	if (sbi->stop_reason[reason] < GENMASK(BITS_PER_BYTE - 1, 0))
4525		sbi->stop_reason[reason]++;
4526	spin_unlock_irqrestore(&sbi->error_lock, flags);
4527}
4528
4529static void f2fs_record_stop_reason(struct f2fs_sb_info *sbi)
4530{
4531	struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
4532	unsigned long flags;
4533	int err;
4534
4535	f2fs_down_write(&sbi->sb_lock);
4536
4537	spin_lock_irqsave(&sbi->error_lock, flags);
4538	if (sbi->error_dirty) {
4539		memcpy(F2FS_RAW_SUPER(sbi)->s_errors, sbi->errors,
4540							MAX_F2FS_ERRORS);
4541		sbi->error_dirty = false;
4542	}
4543	memcpy(raw_super->s_stop_reason, sbi->stop_reason, MAX_STOP_REASON);
4544	spin_unlock_irqrestore(&sbi->error_lock, flags);
4545
4546	err = f2fs_commit_super(sbi, false);
4547
4548	f2fs_up_write(&sbi->sb_lock);
4549	if (err)
4550		f2fs_err_ratelimited(sbi,
4551			"f2fs_commit_super fails to record stop_reason, err:%d",
4552			err);
4553}
4554
4555void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag)
4556{
4557	unsigned long flags;
4558
4559	spin_lock_irqsave(&sbi->error_lock, flags);
4560	if (!test_bit(flag, (unsigned long *)sbi->errors)) {
4561		set_bit(flag, (unsigned long *)sbi->errors);
4562		sbi->error_dirty = true;
4563	}
4564	spin_unlock_irqrestore(&sbi->error_lock, flags);
4565}
4566
4567void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error)
4568{
4569	f2fs_save_errors(sbi, error);
4570
4571	if (!sbi->error_dirty)
4572		return;
4573	if (!test_bit(error, (unsigned long *)sbi->errors))
4574		return;
4575	schedule_work(&sbi->s_error_work);
4576}
4577
4578static bool system_going_down(void)
4579{
4580	return system_state == SYSTEM_HALT || system_state == SYSTEM_POWER_OFF
4581		|| system_state == SYSTEM_RESTART;
4582}
4583
4584void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason)
4585{
4586	struct super_block *sb = sbi->sb;
4587	bool shutdown = reason == STOP_CP_REASON_SHUTDOWN;
4588	bool continue_fs = !shutdown &&
4589			F2FS_OPTION(sbi).errors == MOUNT_ERRORS_CONTINUE;
4590
4591	set_ckpt_flags(sbi, CP_ERROR_FLAG);
4592
4593	if (!f2fs_hw_is_readonly(sbi)) {
4594		save_stop_reason(sbi, reason);
4595
4596		/*
4597		 * always create an asynchronous task to record stop_reason
4598		 * in order to avoid potential deadlock when running into
4599		 * f2fs_record_stop_reason() synchronously.
4600		 */
4601		schedule_work(&sbi->s_error_work);
4602	}
4603
4604	/*
4605	 * We force ERRORS_RO behavior when system is rebooting. Otherwise we
4606	 * could panic during 'reboot -f' as the underlying device got already
4607	 * disabled.
4608	 */
4609	if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_PANIC &&
4610				!shutdown && !system_going_down() &&
4611				!is_sbi_flag_set(sbi, SBI_IS_SHUTDOWN))
4612		panic("F2FS-fs (device %s): panic forced after error\n",
4613							sb->s_id);
4614
4615	if (shutdown)
4616		set_sbi_flag(sbi, SBI_IS_SHUTDOWN);
4617	else
4618		dump_stack();
4619
4620	/*
4621	 * Continue filesystem operators if errors=continue. Should not set
4622	 * RO by shutdown, since RO bypasses thaw_super which can hang the
4623	 * system.
4624	 */
4625	if (continue_fs || f2fs_readonly(sb) || shutdown) {
4626		f2fs_warn(sbi, "Stopped filesystem due to reason: %d", reason);
4627		return;
4628	}
4629
4630	f2fs_warn(sbi, "Remounting filesystem read-only");
4631
4632	/*
4633	 * We have already set CP_ERROR_FLAG flag to stop all updates
4634	 * to filesystem, so it doesn't need to set SB_RDONLY flag here
4635	 * because the flag should be set covered w/ sb->s_umount semaphore
4636	 * via remount procedure, otherwise, it will confuse code like
4637	 * freeze_super() which will lead to deadlocks and other problems.
4638	 */
4639}
4640
4641static void f2fs_record_error_work(struct work_struct *work)
4642{
4643	struct f2fs_sb_info *sbi = container_of(work,
4644					struct f2fs_sb_info, s_error_work);
4645
4646	f2fs_record_stop_reason(sbi);
4647}
4648
4649static inline unsigned int get_first_seq_zone_segno(struct f2fs_sb_info *sbi)
4650{
4651#ifdef CONFIG_BLK_DEV_ZONED
4652	unsigned int zoneno, total_zones;
4653	int devi;
4654
4655	if (!f2fs_sb_has_blkzoned(sbi))
4656		return NULL_SEGNO;
4657
4658	for (devi = 0; devi < sbi->s_ndevs; devi++) {
4659		if (!bdev_is_zoned(FDEV(devi).bdev))
4660			continue;
4661
4662		total_zones = GET_ZONE_FROM_SEG(sbi, FDEV(devi).total_segments);
4663
4664		for (zoneno = 0; zoneno < total_zones; zoneno++) {
4665			unsigned int segs, blks;
4666
4667			if (!f2fs_zone_is_seq(sbi, devi, zoneno))
4668				continue;
4669
4670			segs = GET_SEG_FROM_SEC(sbi,
4671					zoneno * sbi->secs_per_zone);
4672			blks = SEGS_TO_BLKS(sbi, segs);
4673			return GET_SEGNO(sbi, FDEV(devi).start_blk + blks);
4674		}
4675	}
4676#endif
4677	return NULL_SEGNO;
4678}
4679
4680static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
4681{
4682	struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
4683	unsigned int max_devices = MAX_DEVICES;
4684	unsigned int logical_blksize;
4685	blk_mode_t mode = sb_open_mode(sbi->sb->s_flags);
4686	int i;
4687
4688	/* Initialize single device information */
4689	if (!RDEV(0).path[0]) {
4690		if (!bdev_is_zoned(sbi->sb->s_bdev))
4691			return 0;
4692		max_devices = 1;
4693	}
4694
4695	/*
4696	 * Initialize multiple devices information, or single
4697	 * zoned block device information.
4698	 */
4699	sbi->devs = f2fs_kzalloc(sbi,
4700				 array_size(max_devices,
4701					    sizeof(struct f2fs_dev_info)),
4702				 GFP_KERNEL);
4703	if (!sbi->devs)
4704		return -ENOMEM;
4705
4706	logical_blksize = bdev_logical_block_size(sbi->sb->s_bdev);
4707	sbi->aligned_blksize = true;
4708	sbi->bggc_io_aware = AWARE_ALL_IO;
4709#ifdef CONFIG_BLK_DEV_ZONED
4710	sbi->max_open_zones = UINT_MAX;
4711	sbi->blkzone_alloc_policy = BLKZONE_ALLOC_PRIOR_SEQ;
4712	sbi->bggc_io_aware = AWARE_READ_IO;
4713#endif
4714
4715	for (i = 0; i < max_devices; i++) {
4716		if (max_devices == 1) {
4717			FDEV(i).total_segments =
4718				le32_to_cpu(raw_super->segment_count_main);
4719			FDEV(i).start_blk = 0;
4720			FDEV(i).end_blk = FDEV(i).total_segments *
4721						BLKS_PER_SEG(sbi);
4722		}
4723
4724		if (i == 0)
4725			FDEV(0).bdev_file = sbi->sb->s_bdev_file;
4726		else if (!RDEV(i).path[0])
4727			break;
4728
4729		if (max_devices > 1) {
4730			/* Multi-device mount */
4731			memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
4732			FDEV(i).total_segments =
4733				le32_to_cpu(RDEV(i).total_segments);
4734			if (i == 0) {
4735				FDEV(i).start_blk = 0;
4736				FDEV(i).end_blk = FDEV(i).start_blk +
4737					SEGS_TO_BLKS(sbi,
4738					FDEV(i).total_segments) - 1 +
4739					le32_to_cpu(raw_super->segment0_blkaddr);
4740				sbi->allocate_section_hint = FDEV(i).total_segments /
4741							SEGS_PER_SEC(sbi);
4742			} else {
4743				FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
4744				FDEV(i).end_blk = FDEV(i).start_blk +
4745						SEGS_TO_BLKS(sbi,
4746						FDEV(i).total_segments) - 1;
4747				FDEV(i).bdev_file = bdev_file_open_by_path(
4748					FDEV(i).path, mode, sbi->sb, NULL);
4749			}
4750		}
4751		if (IS_ERR(FDEV(i).bdev_file))
4752			return PTR_ERR(FDEV(i).bdev_file);
4753
4754		FDEV(i).bdev = file_bdev(FDEV(i).bdev_file);
4755		/* to release errored devices */
4756		sbi->s_ndevs = i + 1;
4757
4758		if (logical_blksize != bdev_logical_block_size(FDEV(i).bdev))
4759			sbi->aligned_blksize = false;
4760
4761#ifdef CONFIG_BLK_DEV_ZONED
4762		if (bdev_is_zoned(FDEV(i).bdev)) {
4763			if (!f2fs_sb_has_blkzoned(sbi)) {
4764				f2fs_err(sbi, "Zoned block device feature not enabled");
4765				return -EINVAL;
4766			}
4767			if (init_blkz_info(sbi, i)) {
4768				f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
4769				return -EINVAL;
4770			}
4771			if (max_devices == 1)
4772				break;
4773			f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: Host-managed)",
4774				  i, FDEV(i).path,
4775				  FDEV(i).total_segments,
4776				  FDEV(i).start_blk, FDEV(i).end_blk);
4777			continue;
4778		}
4779#endif
4780		f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
4781			  i, FDEV(i).path,
4782			  FDEV(i).total_segments,
4783			  FDEV(i).start_blk, FDEV(i).end_blk);
4784	}
4785	return 0;
4786}
4787
4788static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
4789{
4790#if IS_ENABLED(CONFIG_UNICODE)
4791	if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
4792		const struct f2fs_sb_encodings *encoding_info;
4793		struct unicode_map *encoding;
4794		__u16 encoding_flags;
4795
4796		encoding_info = f2fs_sb_read_encoding(sbi->raw_super);
4797		if (!encoding_info) {
4798			f2fs_err(sbi,
4799				 "Encoding requested by superblock is unknown");
4800			return -EINVAL;
4801		}
4802
4803		encoding_flags = le16_to_cpu(sbi->raw_super->s_encoding_flags);
4804		encoding = utf8_load(encoding_info->version);
4805		if (IS_ERR(encoding)) {
4806			f2fs_err(sbi,
4807				 "can't mount with superblock charset: %s-%u.%u.%u "
4808				 "not supported by the kernel. flags: 0x%x.",
4809				 encoding_info->name,
4810				 unicode_major(encoding_info->version),
4811				 unicode_minor(encoding_info->version),
4812				 unicode_rev(encoding_info->version),
4813				 encoding_flags);
4814			return PTR_ERR(encoding);
4815		}
4816		f2fs_info(sbi, "Using encoding defined by superblock: "
4817			 "%s-%u.%u.%u with flags 0x%hx", encoding_info->name,
4818			 unicode_major(encoding_info->version),
4819			 unicode_minor(encoding_info->version),
4820			 unicode_rev(encoding_info->version),
4821			 encoding_flags);
4822
4823		sbi->sb->s_encoding = encoding;
4824		sbi->sb->s_encoding_flags = encoding_flags;
4825	}
4826#else
4827	if (f2fs_sb_has_casefold(sbi)) {
4828		f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
4829		return -EINVAL;
4830	}
4831#endif
4832	return 0;
4833}
4834
4835static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
4836{
4837	/* adjust parameters according to the volume size */
4838	if (MAIN_SEGS(sbi) <= SMALL_VOLUME_SEGMENTS) {
4839		if (f2fs_block_unit_discard(sbi))
4840			SM_I(sbi)->dcc_info->discard_granularity =
4841						MIN_DISCARD_GRANULARITY;
4842		if (!f2fs_lfs_mode(sbi))
4843			SM_I(sbi)->ipu_policy = BIT(F2FS_IPU_FORCE) |
4844						BIT(F2FS_IPU_HONOR_OPU_WRITE);
4845	}
4846
4847	sbi->readdir_ra = true;
4848}
4849
4850static int f2fs_fill_super(struct super_block *sb, struct fs_context *fc)
4851{
4852	struct f2fs_fs_context *ctx = fc->fs_private;
4853	struct f2fs_sb_info *sbi;
4854	struct f2fs_super_block *raw_super;
4855	struct inode *root;
4856	int err;
4857	bool skip_recovery = false, need_fsck = false;
4858	int recovery, i, valid_super_block;
4859	struct curseg_info *seg_i;
4860	int retry_cnt = 1;
4861#ifdef CONFIG_QUOTA
4862	bool quota_enabled = false;
4863#endif
4864
4865try_onemore:
4866	err = -EINVAL;
4867	raw_super = NULL;
4868	valid_super_block = -1;
4869	recovery = 0;
4870
4871	/* allocate memory for f2fs-specific super block info */
4872	sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
4873	if (!sbi)
4874		return -ENOMEM;
4875
4876	sbi->sb = sb;
4877
4878	/* initialize locks within allocated memory */
4879	init_f2fs_rwsem(&sbi->gc_lock);
4880	mutex_init(&sbi->writepages);
4881	init_f2fs_rwsem(&sbi->cp_global_sem);
4882	init_f2fs_rwsem(&sbi->node_write);
4883	init_f2fs_rwsem(&sbi->node_change);
4884	spin_lock_init(&sbi->stat_lock);
4885	init_f2fs_rwsem(&sbi->cp_rwsem);
4886	init_f2fs_rwsem(&sbi->quota_sem);
4887	init_waitqueue_head(&sbi->cp_wait);
4888	spin_lock_init(&sbi->error_lock);
4889
4890	for (i = 0; i < NR_INODE_TYPE; i++) {
4891		INIT_LIST_HEAD(&sbi->inode_list[i]);
4892		spin_lock_init(&sbi->inode_lock[i]);
4893	}
4894	mutex_init(&sbi->flush_lock);
4895
4896	/* set a block size */
4897	if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
4898		f2fs_err(sbi, "unable to set blocksize");
4899		goto free_sbi;
4900	}
4901
4902	err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
4903								&recovery);
4904	if (err)
4905		goto free_sbi;
4906
4907	sb->s_fs_info = sbi;
4908	sbi->raw_super = raw_super;
4909
4910	INIT_WORK(&sbi->s_error_work, f2fs_record_error_work);
4911	memcpy(sbi->errors, raw_super->s_errors, MAX_F2FS_ERRORS);
4912	memcpy(sbi->stop_reason, raw_super->s_stop_reason, MAX_STOP_REASON);
4913
4914	/* precompute checksum seed for metadata */
4915	if (f2fs_sb_has_inode_chksum(sbi))
4916		sbi->s_chksum_seed = f2fs_chksum(~0, raw_super->uuid,
4917						 sizeof(raw_super->uuid));
4918
4919	default_options(sbi, false);
4920
4921	err = f2fs_check_opt_consistency(fc, sb);
4922	if (err)
4923		goto free_sb_buf;
4924
4925	f2fs_apply_options(fc, sb);
4926
4927	err = f2fs_sanity_check_options(sbi, false);
4928	if (err)
4929		goto free_options;
4930
4931	sb->s_maxbytes = max_file_blocks(NULL) <<
4932				le32_to_cpu(raw_super->log_blocksize);
4933	sb->s_max_links = F2FS_LINK_MAX;
4934
4935	err = f2fs_setup_casefold(sbi);
4936	if (err)
4937		goto free_options;
4938
4939#ifdef CONFIG_QUOTA
4940	sb->dq_op = &f2fs_quota_operations;
4941	sb->s_qcop = &f2fs_quotactl_ops;
4942	sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
4943
4944	if (f2fs_sb_has_quota_ino(sbi)) {
4945		for (i = 0; i < MAXQUOTAS; i++) {
4946			if (f2fs_qf_ino(sbi->sb, i))
4947				sbi->nquota_files++;
4948		}
4949	}
4950#endif
4951
4952	sb->s_op = &f2fs_sops;
4953#ifdef CONFIG_FS_ENCRYPTION
4954	sb->s_cop = &f2fs_cryptops;
4955#endif
4956#ifdef CONFIG_FS_VERITY
4957	sb->s_vop = &f2fs_verityops;
4958#endif
4959	sb->s_xattr = f2fs_xattr_handlers;
4960	sb->s_export_op = &f2fs_export_ops;
4961	sb->s_magic = F2FS_SUPER_MAGIC;
4962	sb->s_time_gran = 1;
4963	sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
4964		(test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
4965	if (test_opt(sbi, INLINECRYPT))
4966		sb->s_flags |= SB_INLINECRYPT;
4967
4968	if (test_opt(sbi, LAZYTIME))
4969		sb->s_flags |= SB_LAZYTIME;
4970	else
4971		sb->s_flags &= ~SB_LAZYTIME;
4972
4973	super_set_uuid(sb, (void *) raw_super->uuid, sizeof(raw_super->uuid));
4974	super_set_sysfs_name_bdev(sb);
4975	sb->s_iflags |= SB_I_CGROUPWB;
4976
4977	/* init f2fs-specific super block info */
4978	sbi->valid_super_block = valid_super_block;
4979
4980	/* disallow all the data/node/meta page writes */
4981	set_sbi_flag(sbi, SBI_POR_DOING);
4982
4983	err = f2fs_init_write_merge_io(sbi);
4984	if (err)
4985		goto free_bio_info;
4986
4987	init_sb_info(sbi);
4988
4989	err = f2fs_init_iostat(sbi);
4990	if (err)
4991		goto free_bio_info;
4992
4993	err = init_percpu_info(sbi);
4994	if (err)
4995		goto free_iostat;
4996
4997	err = f2fs_init_page_array_cache(sbi);
4998	if (err)
4999		goto free_percpu;
5000
5001	/* get an inode for meta space */
5002	sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
5003	if (IS_ERR(sbi->meta_inode)) {
5004		f2fs_err(sbi, "Failed to read F2FS meta data inode");
5005		err = PTR_ERR(sbi->meta_inode);
5006		goto free_page_array_cache;
5007	}
5008
5009	err = f2fs_get_valid_checkpoint(sbi);
5010	if (err) {
5011		f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
5012		goto free_meta_inode;
5013	}
5014
5015	if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
5016		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
5017	if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
5018		set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
5019		sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
5020	}
5021
5022	if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
5023		set_sbi_flag(sbi, SBI_NEED_FSCK);
5024
5025	/* Initialize device list */
5026	err = f2fs_scan_devices(sbi);
5027	if (err) {
5028		f2fs_err(sbi, "Failed to find devices");
5029		goto free_devices;
5030	}
5031
5032	err = f2fs_init_post_read_wq(sbi);
5033	if (err) {
5034		f2fs_err(sbi, "Failed to initialize post read workqueue");
5035		goto free_devices;
5036	}
5037
5038	sbi->total_valid_node_count =
5039				le32_to_cpu(sbi->ckpt->valid_node_count);
5040	percpu_counter_set(&sbi->total_valid_inode_count,
5041				le32_to_cpu(sbi->ckpt->valid_inode_count));
5042	sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
5043	sbi->total_valid_block_count =
5044				le64_to_cpu(sbi->ckpt->valid_block_count);
5045	sbi->last_valid_block_count = sbi->total_valid_block_count;
5046	sbi->reserved_blocks = 0;
5047	sbi->current_reserved_blocks = 0;
5048	limit_reserve_root(sbi);
5049	adjust_unusable_cap_perc(sbi);
5050
5051	f2fs_init_extent_cache_info(sbi);
5052
5053	f2fs_init_ino_entry_info(sbi);
5054
5055	f2fs_init_fsync_node_info(sbi);
5056
5057	/* setup checkpoint request control and start checkpoint issue thread */
5058	f2fs_init_ckpt_req_control(sbi);
5059	if (!f2fs_readonly(sb) && !test_opt(sbi, DISABLE_CHECKPOINT) &&
5060			test_opt(sbi, MERGE_CHECKPOINT)) {
5061		err = f2fs_start_ckpt_thread(sbi);
5062		if (err) {
5063			f2fs_err(sbi,
5064			    "Failed to start F2FS issue_checkpoint_thread (%d)",
5065			    err);
5066			goto stop_ckpt_thread;
5067		}
5068	}
5069
5070	/* setup f2fs internal modules */
5071	err = f2fs_build_segment_manager(sbi);
5072	if (err) {
5073		f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
5074			 err);
5075		goto free_sm;
5076	}
5077	err = f2fs_build_node_manager(sbi);
5078	if (err) {
5079		f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
5080			 err);
5081		goto free_nm;
5082	}
5083
5084	/* For write statistics */
5085	sbi->sectors_written_start = f2fs_get_sectors_written(sbi);
5086
5087	/* get segno of first zoned block device */
5088	sbi->first_seq_zone_segno = get_first_seq_zone_segno(sbi);
5089
5090	sbi->reserved_pin_section = f2fs_sb_has_blkzoned(sbi) ?
5091			ZONED_PIN_SEC_REQUIRED_COUNT :
5092			GET_SEC_FROM_SEG(sbi, overprovision_segments(sbi));
5093
5094	/* Read accumulated write IO statistics if exists */
5095	seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
5096	if (__exist_node_summaries(sbi))
5097		sbi->kbytes_written =
5098			le64_to_cpu(seg_i->journal->info.kbytes_written);
5099
5100	f2fs_build_gc_manager(sbi);
5101
5102	err = f2fs_build_stats(sbi);
5103	if (err)
5104		goto free_nm;
5105
5106	/* get an inode for node space */
5107	sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
5108	if (IS_ERR(sbi->node_inode)) {
5109		f2fs_err(sbi, "Failed to read node inode");
5110		err = PTR_ERR(sbi->node_inode);
5111		goto free_stats;
5112	}
5113
5114	/* read root inode and dentry */
5115	root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
5116	if (IS_ERR(root)) {
5117		f2fs_err(sbi, "Failed to read root inode");
5118		err = PTR_ERR(root);
5119		goto free_node_inode;
5120	}
5121	if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
5122			!root->i_size || !root->i_nlink) {
5123		iput(root);
5124		err = -EINVAL;
5125		goto free_node_inode;
5126	}
5127
5128	generic_set_sb_d_ops(sb);
5129	sb->s_root = d_make_root(root); /* allocate root dentry */
5130	if (!sb->s_root) {
5131		err = -ENOMEM;
5132		goto free_node_inode;
5133	}
5134
5135	err = f2fs_init_compress_inode(sbi);
5136	if (err)
5137		goto free_root_inode;
5138
5139	err = f2fs_register_sysfs(sbi);
5140	if (err)
5141		goto free_compress_inode;
5142
5143	sbi->umount_lock_holder = current;
5144#ifdef CONFIG_QUOTA
5145	/* Enable quota usage during mount */
5146	if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
5147		err = f2fs_enable_quotas(sb);
5148		if (err)
5149			f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
5150	}
5151
5152	quota_enabled = f2fs_recover_quota_begin(sbi);
5153#endif
5154	/* if there are any orphan inodes, free them */
5155	err = f2fs_recover_orphan_inodes(sbi);
5156	if (err)
5157		goto free_meta;
5158
5159	if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG))) {
5160		skip_recovery = true;
5161		goto reset_checkpoint;
5162	}
5163
5164	/* recover fsynced data */
5165	if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
5166			!test_opt(sbi, NORECOVERY)) {
5167		/*
5168		 * mount should be failed, when device has readonly mode, and
5169		 * previous checkpoint was not done by clean system shutdown.
5170		 */
5171		if (f2fs_hw_is_readonly(sbi)) {
5172			if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
5173				err = f2fs_recover_fsync_data(sbi, true);
5174				if (err > 0) {
5175					err = -EROFS;
5176					f2fs_err(sbi, "Need to recover fsync data, but "
5177						"write access unavailable, please try "
5178						"mount w/ disable_roll_forward or norecovery");
5179				}
5180				if (err < 0)
5181					goto free_meta;
5182			}
5183			f2fs_info(sbi, "write access unavailable, skipping recovery");
5184			goto reset_checkpoint;
5185		}
5186
5187		if (need_fsck)
5188			set_sbi_flag(sbi, SBI_NEED_FSCK);
5189
5190		if (skip_recovery)
5191			goto reset_checkpoint;
5192
5193		err = f2fs_recover_fsync_data(sbi, false);
5194		if (err < 0) {
5195			if (err != -ENOMEM)
5196				skip_recovery = true;
5197			need_fsck = true;
5198			f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
5199				 err);
5200			goto free_meta;
5201		}
5202	} else {
5203		err = f2fs_recover_fsync_data(sbi, true);
5204		if (err > 0) {
5205			if (!f2fs_readonly(sb)) {
5206				f2fs_err(sbi, "Need to recover fsync data");
5207				err = -EINVAL;
5208				goto free_meta;
5209			} else {
5210				f2fs_info(sbi, "drop all fsynced data");
5211				err = 0;
5212			}
5213		}
5214	}
5215
5216reset_checkpoint:
5217#ifdef CONFIG_QUOTA
5218	f2fs_recover_quota_end(sbi, quota_enabled);
5219#endif
5220	/*
5221	 * If the f2fs is not readonly and fsync data recovery succeeds,
5222	 * write pointer consistency of cursegs and other zones are already
5223	 * checked and fixed during recovery. However, if recovery fails,
5224	 * write pointers are left untouched, and retry-mount should check
5225	 * them here.
5226	 */
5227	if (skip_recovery)
5228		err = f2fs_check_and_fix_write_pointer(sbi);
5229	if (err)
5230		goto free_meta;
5231
5232	/* f2fs_recover_fsync_data() cleared this already */
5233	clear_sbi_flag(sbi, SBI_POR_DOING);
5234
5235	err = f2fs_init_inmem_curseg(sbi);
5236	if (err)
5237		goto sync_free_meta;
5238
5239	if (test_opt(sbi, DISABLE_CHECKPOINT))
5240		err = f2fs_disable_checkpoint(sbi);
5241	else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG))
5242		err = f2fs_enable_checkpoint(sbi);
5243	if (err)
5244		goto sync_free_meta;
5245
5246	/*
5247	 * If filesystem is not mounted as read-only then
5248	 * do start the gc_thread.
5249	 */
5250	if ((F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF ||
5251		test_opt(sbi, GC_MERGE)) && !f2fs_readonly(sb)) {
5252		/* After POR, we can run background GC thread.*/
5253		err = f2fs_start_gc_thread(sbi);
5254		if (err)
5255			goto sync_free_meta;
5256	}
5257
5258	/* recover broken superblock */
5259	if (recovery) {
5260		err = f2fs_commit_super(sbi, true);
5261		f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
5262			  sbi->valid_super_block ? 1 : 2, err);
5263	}
5264
5265	f2fs_join_shrinker(sbi);
5266
5267	f2fs_tuning_parameters(sbi);
5268
5269	f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
5270		    cur_cp_version(F2FS_CKPT(sbi)));
5271	f2fs_update_time(sbi, CP_TIME);
5272	f2fs_update_time(sbi, REQ_TIME);
5273	clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
5274
5275	sbi->umount_lock_holder = NULL;
5276	return 0;
5277
5278sync_free_meta:
5279	/* safe to flush all the data */
5280	sync_filesystem(sbi->sb);
5281	retry_cnt = 0;
5282
5283free_meta:
5284#ifdef CONFIG_QUOTA
5285	f2fs_truncate_quota_inode_pages(sb);
5286	if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
5287		f2fs_quota_off_umount(sbi->sb);
5288#endif
5289	/*
5290	 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
5291	 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
5292	 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
5293	 * falls into an infinite loop in f2fs_sync_meta_pages().
5294	 */
5295	truncate_inode_pages_final(META_MAPPING(sbi));
5296	/* evict some inodes being cached by GC */
5297	evict_inodes(sb);
5298	f2fs_unregister_sysfs(sbi);
5299free_compress_inode:
5300	f2fs_destroy_compress_inode(sbi);
5301free_root_inode:
5302	dput(sb->s_root);
5303	sb->s_root = NULL;
5304free_node_inode:
5305	f2fs_release_ino_entry(sbi, true);
5306	truncate_inode_pages_final(NODE_MAPPING(sbi));
5307	iput(sbi->node_inode);
5308	sbi->node_inode = NULL;
5309free_stats:
5310	f2fs_destroy_stats(sbi);
5311free_nm:
5312	/* stop discard thread before destroying node manager */
5313	f2fs_stop_discard_thread(sbi);
5314	f2fs_destroy_node_manager(sbi);
5315free_sm:
5316	f2fs_destroy_segment_manager(sbi);
5317stop_ckpt_thread:
5318	f2fs_stop_ckpt_thread(sbi);
5319	/* flush s_error_work before sbi destroy */
5320	flush_work(&sbi->s_error_work);
5321	f2fs_destroy_post_read_wq(sbi);
5322free_devices:
5323	destroy_device_list(sbi);
5324	kvfree(sbi->ckpt);
5325free_meta_inode:
5326	make_bad_inode(sbi->meta_inode);
5327	iput(sbi->meta_inode);
5328	sbi->meta_inode = NULL;
5329free_page_array_cache:
5330	f2fs_destroy_page_array_cache(sbi);
5331free_percpu:
5332	destroy_percpu_info(sbi);
5333free_iostat:
5334	f2fs_destroy_iostat(sbi);
5335free_bio_info:
5336	for (i = 0; i < NR_PAGE_TYPE; i++)
5337		kfree(sbi->write_io[i]);
5338
5339#if IS_ENABLED(CONFIG_UNICODE)
5340	utf8_unload(sb->s_encoding);
5341	sb->s_encoding = NULL;
5342#endif
5343free_options:
5344#ifdef CONFIG_QUOTA
5345	for (i = 0; i < MAXQUOTAS; i++)
5346		kfree(F2FS_OPTION(sbi).s_qf_names[i]);
5347#endif
5348	/* no need to free dummy_enc_policy, we just keep it in ctx when failed */
5349	swap(F2FS_CTX_INFO(ctx).dummy_enc_policy, F2FS_OPTION(sbi).dummy_enc_policy);
5350free_sb_buf:
5351	kfree(raw_super);
5352free_sbi:
5353	kfree(sbi);
5354	sb->s_fs_info = NULL;
5355
5356	/* give only one another chance */
5357	if (retry_cnt > 0 && skip_recovery) {
5358		retry_cnt--;
5359		shrink_dcache_sb(sb);
5360		goto try_onemore;
5361	}
5362	return err;
5363}
5364
5365static int f2fs_get_tree(struct fs_context *fc)
5366{
5367	return get_tree_bdev(fc, f2fs_fill_super);
5368}
5369
5370static int f2fs_reconfigure(struct fs_context *fc)
5371{
5372	struct super_block *sb = fc->root->d_sb;
5373
5374	return __f2fs_remount(fc, sb);
5375}
5376
5377static void f2fs_fc_free(struct fs_context *fc)
5378{
5379	struct f2fs_fs_context *ctx = fc->fs_private;
5380
5381	if (!ctx)
5382		return;
5383
5384#ifdef CONFIG_QUOTA
5385	f2fs_unnote_qf_name_all(fc);
5386#endif
5387	fscrypt_free_dummy_policy(&F2FS_CTX_INFO(ctx).dummy_enc_policy);
5388	kfree(ctx);
5389}
5390
5391static const struct fs_context_operations f2fs_context_ops = {
5392	.parse_param	= f2fs_parse_param,
5393	.get_tree	= f2fs_get_tree,
5394	.reconfigure = f2fs_reconfigure,
5395	.free	= f2fs_fc_free,
5396};
5397
5398static void kill_f2fs_super(struct super_block *sb)
5399{
5400	struct f2fs_sb_info *sbi = F2FS_SB(sb);
5401
5402	if (sb->s_root) {
5403		sbi->umount_lock_holder = current;
5404
5405		set_sbi_flag(sbi, SBI_IS_CLOSE);
5406		f2fs_stop_gc_thread(sbi);
5407		f2fs_stop_discard_thread(sbi);
5408
5409#ifdef CONFIG_F2FS_FS_COMPRESSION
5410		/*
5411		 * latter evict_inode() can bypass checking and invalidating
5412		 * compress inode cache.
5413		 */
5414		if (test_opt(sbi, COMPRESS_CACHE))
5415			truncate_inode_pages_final(COMPRESS_MAPPING(sbi));
5416#endif
5417
5418		if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
5419				!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
5420			struct cp_control cpc = {
5421				.reason = CP_UMOUNT,
5422			};
5423			stat_inc_cp_call_count(sbi, TOTAL_CALL);
5424			f2fs_write_checkpoint(sbi, &cpc);
5425		}
5426
5427		if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
5428			sb->s_flags &= ~SB_RDONLY;
5429	}
5430	kill_block_super(sb);
5431	/* Release block devices last, after fscrypt_destroy_keyring(). */
5432	if (sbi) {
5433		destroy_device_list(sbi);
5434		kfree(sbi);
5435		sb->s_fs_info = NULL;
5436	}
5437}
5438
5439static int f2fs_init_fs_context(struct fs_context *fc)
5440{
5441	struct f2fs_fs_context *ctx;
5442
5443	ctx = kzalloc(sizeof(struct f2fs_fs_context), GFP_KERNEL);
5444	if (!ctx)
5445		return -ENOMEM;
5446
5447	fc->fs_private = ctx;
5448	fc->ops = &f2fs_context_ops;
5449
5450	return 0;
5451}
5452
5453static struct file_system_type f2fs_fs_type = {
5454	.owner		= THIS_MODULE,
5455	.name		= "f2fs",
5456	.init_fs_context = f2fs_init_fs_context,
5457	.kill_sb	= kill_f2fs_super,
5458	.fs_flags	= FS_REQUIRES_DEV | FS_ALLOW_IDMAP,
5459};
5460MODULE_ALIAS_FS("f2fs");
5461
5462static int __init init_inodecache(void)
5463{
5464	f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
5465			sizeof(struct f2fs_inode_info), 0,
5466			SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
5467	return f2fs_inode_cachep ? 0 : -ENOMEM;
5468}
5469
5470static void destroy_inodecache(void)
5471{
5472	/*
5473	 * Make sure all delayed rcu free inodes are flushed before we
5474	 * destroy cache.
5475	 */
5476	rcu_barrier();
5477	kmem_cache_destroy(f2fs_inode_cachep);
5478}
5479
5480static int __init init_f2fs_fs(void)
5481{
5482	int err;
5483
5484	err = init_inodecache();
5485	if (err)
5486		goto fail;
5487	err = f2fs_create_node_manager_caches();
5488	if (err)
5489		goto free_inodecache;
5490	err = f2fs_create_segment_manager_caches();
5491	if (err)
5492		goto free_node_manager_caches;
5493	err = f2fs_create_checkpoint_caches();
5494	if (err)
5495		goto free_segment_manager_caches;
5496	err = f2fs_create_recovery_cache();
5497	if (err)
5498		goto free_checkpoint_caches;
5499	err = f2fs_create_extent_cache();
5500	if (err)
5501		goto free_recovery_cache;
5502	err = f2fs_create_garbage_collection_cache();
5503	if (err)
5504		goto free_extent_cache;
5505	err = f2fs_init_sysfs();
5506	if (err)
5507		goto free_garbage_collection_cache;
5508	err = f2fs_init_shrinker();
5509	if (err)
5510		goto free_sysfs;
5511	f2fs_create_root_stats();
5512	err = f2fs_init_post_read_processing();
5513	if (err)
5514		goto free_root_stats;
5515	err = f2fs_init_iostat_processing();
5516	if (err)
5517		goto free_post_read;
5518	err = f2fs_init_bio_entry_cache();
5519	if (err)
5520		goto free_iostat;
5521	err = f2fs_init_bioset();
5522	if (err)
5523		goto free_bio_entry_cache;
5524	err = f2fs_init_compress_mempool();
5525	if (err)
5526		goto free_bioset;
5527	err = f2fs_init_compress_cache();
5528	if (err)
5529		goto free_compress_mempool;
5530	err = f2fs_create_casefold_cache();
5531	if (err)
5532		goto free_compress_cache;
5533	err = f2fs_init_xattr_cache();
5534	if (err)
5535		goto free_casefold_cache;
5536	err = register_filesystem(&f2fs_fs_type);
5537	if (err)
5538		goto free_xattr_cache;
5539	return 0;
5540free_xattr_cache:
5541	f2fs_destroy_xattr_cache();
5542free_casefold_cache:
5543	f2fs_destroy_casefold_cache();
5544free_compress_cache:
5545	f2fs_destroy_compress_cache();
5546free_compress_mempool:
5547	f2fs_destroy_compress_mempool();
5548free_bioset:
5549	f2fs_destroy_bioset();
5550free_bio_entry_cache:
5551	f2fs_destroy_bio_entry_cache();
5552free_iostat:
5553	f2fs_destroy_iostat_processing();
5554free_post_read:
5555	f2fs_destroy_post_read_processing();
5556free_root_stats:
5557	f2fs_destroy_root_stats();
5558	f2fs_exit_shrinker();
5559free_sysfs:
5560	f2fs_exit_sysfs();
5561free_garbage_collection_cache:
5562	f2fs_destroy_garbage_collection_cache();
5563free_extent_cache:
5564	f2fs_destroy_extent_cache();
5565free_recovery_cache:
5566	f2fs_destroy_recovery_cache();
5567free_checkpoint_caches:
5568	f2fs_destroy_checkpoint_caches();
5569free_segment_manager_caches:
5570	f2fs_destroy_segment_manager_caches();
5571free_node_manager_caches:
5572	f2fs_destroy_node_manager_caches();
5573free_inodecache:
5574	destroy_inodecache();
5575fail:
5576	return err;
5577}
5578
5579static void __exit exit_f2fs_fs(void)
5580{
5581	unregister_filesystem(&f2fs_fs_type);
5582	f2fs_destroy_xattr_cache();
5583	f2fs_destroy_casefold_cache();
5584	f2fs_destroy_compress_cache();
5585	f2fs_destroy_compress_mempool();
5586	f2fs_destroy_bioset();
5587	f2fs_destroy_bio_entry_cache();
5588	f2fs_destroy_iostat_processing();
5589	f2fs_destroy_post_read_processing();
5590	f2fs_destroy_root_stats();
5591	f2fs_exit_shrinker();
5592	f2fs_exit_sysfs();
5593	f2fs_destroy_garbage_collection_cache();
5594	f2fs_destroy_extent_cache();
5595	f2fs_destroy_recovery_cache();
5596	f2fs_destroy_checkpoint_caches();
5597	f2fs_destroy_segment_manager_caches();
5598	f2fs_destroy_node_manager_caches();
5599	destroy_inodecache();
5600}
5601
5602module_init(init_f2fs_fs)
5603module_exit(exit_f2fs_fs)
5604
5605MODULE_AUTHOR("Samsung Electronics's Praesto Team");
5606MODULE_DESCRIPTION("Flash Friendly File System");
5607MODULE_LICENSE("GPL");