xref: /openbmc/linux/fs/f2fs/super.c (revision f4356947)
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  * fs/f2fs/super.c
4  *
5  * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6  *             http://www.samsung.com/
7  */
8 #include <linux/module.h>
9 #include <linux/init.h>
10 #include <linux/fs.h>
11 #include <linux/fs_context.h>
12 #include <linux/sched/mm.h>
13 #include <linux/statfs.h>
14 #include <linux/buffer_head.h>
15 #include <linux/kthread.h>
16 #include <linux/parser.h>
17 #include <linux/mount.h>
18 #include <linux/seq_file.h>
19 #include <linux/proc_fs.h>
20 #include <linux/random.h>
21 #include <linux/exportfs.h>
22 #include <linux/blkdev.h>
23 #include <linux/quotaops.h>
24 #include <linux/f2fs_fs.h>
25 #include <linux/sysfs.h>
26 #include <linux/quota.h>
27 #include <linux/unicode.h>
28 #include <linux/part_stat.h>
29 #include <linux/zstd.h>
30 #include <linux/lz4.h>
31 
32 #include "f2fs.h"
33 #include "node.h"
34 #include "segment.h"
35 #include "xattr.h"
36 #include "gc.h"
37 #include "iostat.h"
38 
39 #define CREATE_TRACE_POINTS
40 #include <trace/events/f2fs.h>
41 
42 static struct kmem_cache *f2fs_inode_cachep;
43 
44 #ifdef CONFIG_F2FS_FAULT_INJECTION
45 
46 const char *f2fs_fault_name[FAULT_MAX] = {
47 	[FAULT_KMALLOC]		= "kmalloc",
48 	[FAULT_KVMALLOC]	= "kvmalloc",
49 	[FAULT_PAGE_ALLOC]	= "page alloc",
50 	[FAULT_PAGE_GET]	= "page get",
51 	[FAULT_ALLOC_NID]	= "alloc nid",
52 	[FAULT_ORPHAN]		= "orphan",
53 	[FAULT_BLOCK]		= "no more block",
54 	[FAULT_DIR_DEPTH]	= "too big dir depth",
55 	[FAULT_EVICT_INODE]	= "evict_inode fail",
56 	[FAULT_TRUNCATE]	= "truncate fail",
57 	[FAULT_READ_IO]		= "read IO error",
58 	[FAULT_CHECKPOINT]	= "checkpoint error",
59 	[FAULT_DISCARD]		= "discard error",
60 	[FAULT_WRITE_IO]	= "write IO error",
61 	[FAULT_SLAB_ALLOC]	= "slab alloc",
62 	[FAULT_DQUOT_INIT]	= "dquot initialize",
63 	[FAULT_LOCK_OP]		= "lock_op",
64 	[FAULT_BLKADDR]		= "invalid blkaddr",
65 };
66 
67 void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
68 							unsigned int type)
69 {
70 	struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
71 
72 	if (rate) {
73 		atomic_set(&ffi->inject_ops, 0);
74 		ffi->inject_rate = rate;
75 	}
76 
77 	if (type)
78 		ffi->inject_type = type;
79 
80 	if (!rate && !type)
81 		memset(ffi, 0, sizeof(struct f2fs_fault_info));
82 }
83 #endif
84 
85 /* f2fs-wide shrinker description */
86 static struct shrinker f2fs_shrinker_info = {
87 	.scan_objects = f2fs_shrink_scan,
88 	.count_objects = f2fs_shrink_count,
89 	.seeks = DEFAULT_SEEKS,
90 };
91 
92 enum {
93 	Opt_gc_background,
94 	Opt_disable_roll_forward,
95 	Opt_norecovery,
96 	Opt_discard,
97 	Opt_nodiscard,
98 	Opt_noheap,
99 	Opt_heap,
100 	Opt_user_xattr,
101 	Opt_nouser_xattr,
102 	Opt_acl,
103 	Opt_noacl,
104 	Opt_active_logs,
105 	Opt_disable_ext_identify,
106 	Opt_inline_xattr,
107 	Opt_noinline_xattr,
108 	Opt_inline_xattr_size,
109 	Opt_inline_data,
110 	Opt_inline_dentry,
111 	Opt_noinline_dentry,
112 	Opt_flush_merge,
113 	Opt_noflush_merge,
114 	Opt_barrier,
115 	Opt_nobarrier,
116 	Opt_fastboot,
117 	Opt_extent_cache,
118 	Opt_noextent_cache,
119 	Opt_noinline_data,
120 	Opt_data_flush,
121 	Opt_reserve_root,
122 	Opt_resgid,
123 	Opt_resuid,
124 	Opt_mode,
125 	Opt_io_size_bits,
126 	Opt_fault_injection,
127 	Opt_fault_type,
128 	Opt_lazytime,
129 	Opt_nolazytime,
130 	Opt_quota,
131 	Opt_noquota,
132 	Opt_usrquota,
133 	Opt_grpquota,
134 	Opt_prjquota,
135 	Opt_usrjquota,
136 	Opt_grpjquota,
137 	Opt_prjjquota,
138 	Opt_offusrjquota,
139 	Opt_offgrpjquota,
140 	Opt_offprjjquota,
141 	Opt_jqfmt_vfsold,
142 	Opt_jqfmt_vfsv0,
143 	Opt_jqfmt_vfsv1,
144 	Opt_alloc,
145 	Opt_fsync,
146 	Opt_test_dummy_encryption,
147 	Opt_inlinecrypt,
148 	Opt_checkpoint_disable,
149 	Opt_checkpoint_disable_cap,
150 	Opt_checkpoint_disable_cap_perc,
151 	Opt_checkpoint_enable,
152 	Opt_checkpoint_merge,
153 	Opt_nocheckpoint_merge,
154 	Opt_compress_algorithm,
155 	Opt_compress_log_size,
156 	Opt_compress_extension,
157 	Opt_nocompress_extension,
158 	Opt_compress_chksum,
159 	Opt_compress_mode,
160 	Opt_compress_cache,
161 	Opt_atgc,
162 	Opt_gc_merge,
163 	Opt_nogc_merge,
164 	Opt_discard_unit,
165 	Opt_memory_mode,
166 	Opt_age_extent_cache,
167 	Opt_err,
168 };
169 
170 static match_table_t f2fs_tokens = {
171 	{Opt_gc_background, "background_gc=%s"},
172 	{Opt_disable_roll_forward, "disable_roll_forward"},
173 	{Opt_norecovery, "norecovery"},
174 	{Opt_discard, "discard"},
175 	{Opt_nodiscard, "nodiscard"},
176 	{Opt_noheap, "no_heap"},
177 	{Opt_heap, "heap"},
178 	{Opt_user_xattr, "user_xattr"},
179 	{Opt_nouser_xattr, "nouser_xattr"},
180 	{Opt_acl, "acl"},
181 	{Opt_noacl, "noacl"},
182 	{Opt_active_logs, "active_logs=%u"},
183 	{Opt_disable_ext_identify, "disable_ext_identify"},
184 	{Opt_inline_xattr, "inline_xattr"},
185 	{Opt_noinline_xattr, "noinline_xattr"},
186 	{Opt_inline_xattr_size, "inline_xattr_size=%u"},
187 	{Opt_inline_data, "inline_data"},
188 	{Opt_inline_dentry, "inline_dentry"},
189 	{Opt_noinline_dentry, "noinline_dentry"},
190 	{Opt_flush_merge, "flush_merge"},
191 	{Opt_noflush_merge, "noflush_merge"},
192 	{Opt_barrier, "barrier"},
193 	{Opt_nobarrier, "nobarrier"},
194 	{Opt_fastboot, "fastboot"},
195 	{Opt_extent_cache, "extent_cache"},
196 	{Opt_noextent_cache, "noextent_cache"},
197 	{Opt_noinline_data, "noinline_data"},
198 	{Opt_data_flush, "data_flush"},
199 	{Opt_reserve_root, "reserve_root=%u"},
200 	{Opt_resgid, "resgid=%u"},
201 	{Opt_resuid, "resuid=%u"},
202 	{Opt_mode, "mode=%s"},
203 	{Opt_io_size_bits, "io_bits=%u"},
204 	{Opt_fault_injection, "fault_injection=%u"},
205 	{Opt_fault_type, "fault_type=%u"},
206 	{Opt_lazytime, "lazytime"},
207 	{Opt_nolazytime, "nolazytime"},
208 	{Opt_quota, "quota"},
209 	{Opt_noquota, "noquota"},
210 	{Opt_usrquota, "usrquota"},
211 	{Opt_grpquota, "grpquota"},
212 	{Opt_prjquota, "prjquota"},
213 	{Opt_usrjquota, "usrjquota=%s"},
214 	{Opt_grpjquota, "grpjquota=%s"},
215 	{Opt_prjjquota, "prjjquota=%s"},
216 	{Opt_offusrjquota, "usrjquota="},
217 	{Opt_offgrpjquota, "grpjquota="},
218 	{Opt_offprjjquota, "prjjquota="},
219 	{Opt_jqfmt_vfsold, "jqfmt=vfsold"},
220 	{Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
221 	{Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
222 	{Opt_alloc, "alloc_mode=%s"},
223 	{Opt_fsync, "fsync_mode=%s"},
224 	{Opt_test_dummy_encryption, "test_dummy_encryption=%s"},
225 	{Opt_test_dummy_encryption, "test_dummy_encryption"},
226 	{Opt_inlinecrypt, "inlinecrypt"},
227 	{Opt_checkpoint_disable, "checkpoint=disable"},
228 	{Opt_checkpoint_disable_cap, "checkpoint=disable:%u"},
229 	{Opt_checkpoint_disable_cap_perc, "checkpoint=disable:%u%%"},
230 	{Opt_checkpoint_enable, "checkpoint=enable"},
231 	{Opt_checkpoint_merge, "checkpoint_merge"},
232 	{Opt_nocheckpoint_merge, "nocheckpoint_merge"},
233 	{Opt_compress_algorithm, "compress_algorithm=%s"},
234 	{Opt_compress_log_size, "compress_log_size=%u"},
235 	{Opt_compress_extension, "compress_extension=%s"},
236 	{Opt_nocompress_extension, "nocompress_extension=%s"},
237 	{Opt_compress_chksum, "compress_chksum"},
238 	{Opt_compress_mode, "compress_mode=%s"},
239 	{Opt_compress_cache, "compress_cache"},
240 	{Opt_atgc, "atgc"},
241 	{Opt_gc_merge, "gc_merge"},
242 	{Opt_nogc_merge, "nogc_merge"},
243 	{Opt_discard_unit, "discard_unit=%s"},
244 	{Opt_memory_mode, "memory=%s"},
245 	{Opt_age_extent_cache, "age_extent_cache"},
246 	{Opt_err, NULL},
247 };
248 
249 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...)
250 {
251 	struct va_format vaf;
252 	va_list args;
253 	int level;
254 
255 	va_start(args, fmt);
256 
257 	level = printk_get_level(fmt);
258 	vaf.fmt = printk_skip_level(fmt);
259 	vaf.va = &args;
260 	printk("%c%cF2FS-fs (%s): %pV\n",
261 	       KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
262 
263 	va_end(args);
264 }
265 
266 #if IS_ENABLED(CONFIG_UNICODE)
267 static const struct f2fs_sb_encodings {
268 	__u16 magic;
269 	char *name;
270 	unsigned int version;
271 } f2fs_sb_encoding_map[] = {
272 	{F2FS_ENC_UTF8_12_1, "utf8", UNICODE_AGE(12, 1, 0)},
273 };
274 
275 static const struct f2fs_sb_encodings *
276 f2fs_sb_read_encoding(const struct f2fs_super_block *sb)
277 {
278 	__u16 magic = le16_to_cpu(sb->s_encoding);
279 	int i;
280 
281 	for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
282 		if (magic == f2fs_sb_encoding_map[i].magic)
283 			return &f2fs_sb_encoding_map[i];
284 
285 	return NULL;
286 }
287 
288 struct kmem_cache *f2fs_cf_name_slab;
289 static int __init f2fs_create_casefold_cache(void)
290 {
291 	f2fs_cf_name_slab = f2fs_kmem_cache_create("f2fs_casefolded_name",
292 							F2FS_NAME_LEN);
293 	return f2fs_cf_name_slab ? 0 : -ENOMEM;
294 }
295 
296 static void f2fs_destroy_casefold_cache(void)
297 {
298 	kmem_cache_destroy(f2fs_cf_name_slab);
299 }
300 #else
301 static int __init f2fs_create_casefold_cache(void) { return 0; }
302 static void f2fs_destroy_casefold_cache(void) { }
303 #endif
304 
305 static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
306 {
307 	block_t limit = min((sbi->user_block_count >> 3),
308 			sbi->user_block_count - sbi->reserved_blocks);
309 
310 	/* limit is 12.5% */
311 	if (test_opt(sbi, RESERVE_ROOT) &&
312 			F2FS_OPTION(sbi).root_reserved_blocks > limit) {
313 		F2FS_OPTION(sbi).root_reserved_blocks = limit;
314 		f2fs_info(sbi, "Reduce reserved blocks for root = %u",
315 			  F2FS_OPTION(sbi).root_reserved_blocks);
316 	}
317 	if (!test_opt(sbi, RESERVE_ROOT) &&
318 		(!uid_eq(F2FS_OPTION(sbi).s_resuid,
319 				make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
320 		!gid_eq(F2FS_OPTION(sbi).s_resgid,
321 				make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
322 		f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
323 			  from_kuid_munged(&init_user_ns,
324 					   F2FS_OPTION(sbi).s_resuid),
325 			  from_kgid_munged(&init_user_ns,
326 					   F2FS_OPTION(sbi).s_resgid));
327 }
328 
329 static inline int adjust_reserved_segment(struct f2fs_sb_info *sbi)
330 {
331 	unsigned int sec_blks = sbi->blocks_per_seg * sbi->segs_per_sec;
332 	unsigned int avg_vblocks;
333 	unsigned int wanted_reserved_segments;
334 	block_t avail_user_block_count;
335 
336 	if (!F2FS_IO_ALIGNED(sbi))
337 		return 0;
338 
339 	/* average valid block count in section in worst case */
340 	avg_vblocks = sec_blks / F2FS_IO_SIZE(sbi);
341 
342 	/*
343 	 * we need enough free space when migrating one section in worst case
344 	 */
345 	wanted_reserved_segments = (F2FS_IO_SIZE(sbi) / avg_vblocks) *
346 						reserved_segments(sbi);
347 	wanted_reserved_segments -= reserved_segments(sbi);
348 
349 	avail_user_block_count = sbi->user_block_count -
350 				sbi->current_reserved_blocks -
351 				F2FS_OPTION(sbi).root_reserved_blocks;
352 
353 	if (wanted_reserved_segments * sbi->blocks_per_seg >
354 					avail_user_block_count) {
355 		f2fs_err(sbi, "IO align feature can't grab additional reserved segment: %u, available segments: %u",
356 			wanted_reserved_segments,
357 			avail_user_block_count >> sbi->log_blocks_per_seg);
358 		return -ENOSPC;
359 	}
360 
361 	SM_I(sbi)->additional_reserved_segments = wanted_reserved_segments;
362 
363 	f2fs_info(sbi, "IO align feature needs additional reserved segment: %u",
364 			 wanted_reserved_segments);
365 
366 	return 0;
367 }
368 
369 static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
370 {
371 	if (!F2FS_OPTION(sbi).unusable_cap_perc)
372 		return;
373 
374 	if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
375 		F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
376 	else
377 		F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
378 					F2FS_OPTION(sbi).unusable_cap_perc;
379 
380 	f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
381 			F2FS_OPTION(sbi).unusable_cap,
382 			F2FS_OPTION(sbi).unusable_cap_perc);
383 }
384 
385 static void init_once(void *foo)
386 {
387 	struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
388 
389 	inode_init_once(&fi->vfs_inode);
390 }
391 
392 #ifdef CONFIG_QUOTA
393 static const char * const quotatypes[] = INITQFNAMES;
394 #define QTYPE2NAME(t) (quotatypes[t])
395 static int f2fs_set_qf_name(struct super_block *sb, int qtype,
396 							substring_t *args)
397 {
398 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
399 	char *qname;
400 	int ret = -EINVAL;
401 
402 	if (sb_any_quota_loaded(sb) && !F2FS_OPTION(sbi).s_qf_names[qtype]) {
403 		f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
404 		return -EINVAL;
405 	}
406 	if (f2fs_sb_has_quota_ino(sbi)) {
407 		f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
408 		return 0;
409 	}
410 
411 	qname = match_strdup(args);
412 	if (!qname) {
413 		f2fs_err(sbi, "Not enough memory for storing quotafile name");
414 		return -ENOMEM;
415 	}
416 	if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
417 		if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
418 			ret = 0;
419 		else
420 			f2fs_err(sbi, "%s quota file already specified",
421 				 QTYPE2NAME(qtype));
422 		goto errout;
423 	}
424 	if (strchr(qname, '/')) {
425 		f2fs_err(sbi, "quotafile must be on filesystem root");
426 		goto errout;
427 	}
428 	F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
429 	set_opt(sbi, QUOTA);
430 	return 0;
431 errout:
432 	kfree(qname);
433 	return ret;
434 }
435 
436 static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
437 {
438 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
439 
440 	if (sb_any_quota_loaded(sb) && F2FS_OPTION(sbi).s_qf_names[qtype]) {
441 		f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
442 		return -EINVAL;
443 	}
444 	kfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
445 	F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
446 	return 0;
447 }
448 
449 static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
450 {
451 	/*
452 	 * We do the test below only for project quotas. 'usrquota' and
453 	 * 'grpquota' mount options are allowed even without quota feature
454 	 * to support legacy quotas in quota files.
455 	 */
456 	if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi)) {
457 		f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
458 		return -1;
459 	}
460 	if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
461 			F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
462 			F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) {
463 		if (test_opt(sbi, USRQUOTA) &&
464 				F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
465 			clear_opt(sbi, USRQUOTA);
466 
467 		if (test_opt(sbi, GRPQUOTA) &&
468 				F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
469 			clear_opt(sbi, GRPQUOTA);
470 
471 		if (test_opt(sbi, PRJQUOTA) &&
472 				F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
473 			clear_opt(sbi, PRJQUOTA);
474 
475 		if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
476 				test_opt(sbi, PRJQUOTA)) {
477 			f2fs_err(sbi, "old and new quota format mixing");
478 			return -1;
479 		}
480 
481 		if (!F2FS_OPTION(sbi).s_jquota_fmt) {
482 			f2fs_err(sbi, "journaled quota format not specified");
483 			return -1;
484 		}
485 	}
486 
487 	if (f2fs_sb_has_quota_ino(sbi) && F2FS_OPTION(sbi).s_jquota_fmt) {
488 		f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
489 		F2FS_OPTION(sbi).s_jquota_fmt = 0;
490 	}
491 	return 0;
492 }
493 #endif
494 
495 static int f2fs_set_test_dummy_encryption(struct super_block *sb,
496 					  const char *opt,
497 					  const substring_t *arg,
498 					  bool is_remount)
499 {
500 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
501 	struct fs_parameter param = {
502 		.type = fs_value_is_string,
503 		.string = arg->from ? arg->from : "",
504 	};
505 	struct fscrypt_dummy_policy *policy =
506 		&F2FS_OPTION(sbi).dummy_enc_policy;
507 	int err;
508 
509 	if (!IS_ENABLED(CONFIG_FS_ENCRYPTION)) {
510 		f2fs_warn(sbi, "test_dummy_encryption option not supported");
511 		return -EINVAL;
512 	}
513 
514 	if (!f2fs_sb_has_encrypt(sbi)) {
515 		f2fs_err(sbi, "Encrypt feature is off");
516 		return -EINVAL;
517 	}
518 
519 	/*
520 	 * This mount option is just for testing, and it's not worthwhile to
521 	 * implement the extra complexity (e.g. RCU protection) that would be
522 	 * needed to allow it to be set or changed during remount.  We do allow
523 	 * it to be specified during remount, but only if there is no change.
524 	 */
525 	if (is_remount && !fscrypt_is_dummy_policy_set(policy)) {
526 		f2fs_warn(sbi, "Can't set test_dummy_encryption on remount");
527 		return -EINVAL;
528 	}
529 
530 	err = fscrypt_parse_test_dummy_encryption(&param, policy);
531 	if (err) {
532 		if (err == -EEXIST)
533 			f2fs_warn(sbi,
534 				  "Can't change test_dummy_encryption on remount");
535 		else if (err == -EINVAL)
536 			f2fs_warn(sbi, "Value of option \"%s\" is unrecognized",
537 				  opt);
538 		else
539 			f2fs_warn(sbi, "Error processing option \"%s\" [%d]",
540 				  opt, err);
541 		return -EINVAL;
542 	}
543 	f2fs_warn(sbi, "Test dummy encryption mode enabled");
544 	return 0;
545 }
546 
547 #ifdef CONFIG_F2FS_FS_COMPRESSION
548 /*
549  * 1. The same extension name cannot not appear in both compress and non-compress extension
550  * at the same time.
551  * 2. If the compress extension specifies all files, the types specified by the non-compress
552  * extension will be treated as special cases and will not be compressed.
553  * 3. Don't allow the non-compress extension specifies all files.
554  */
555 static int f2fs_test_compress_extension(struct f2fs_sb_info *sbi)
556 {
557 	unsigned char (*ext)[F2FS_EXTENSION_LEN];
558 	unsigned char (*noext)[F2FS_EXTENSION_LEN];
559 	int ext_cnt, noext_cnt, index = 0, no_index = 0;
560 
561 	ext = F2FS_OPTION(sbi).extensions;
562 	ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
563 	noext = F2FS_OPTION(sbi).noextensions;
564 	noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
565 
566 	if (!noext_cnt)
567 		return 0;
568 
569 	for (no_index = 0; no_index < noext_cnt; no_index++) {
570 		if (!strcasecmp("*", noext[no_index])) {
571 			f2fs_info(sbi, "Don't allow the nocompress extension specifies all files");
572 			return -EINVAL;
573 		}
574 		for (index = 0; index < ext_cnt; index++) {
575 			if (!strcasecmp(ext[index], noext[no_index])) {
576 				f2fs_info(sbi, "Don't allow the same extension %s appear in both compress and nocompress extension",
577 						ext[index]);
578 				return -EINVAL;
579 			}
580 		}
581 	}
582 	return 0;
583 }
584 
585 #ifdef CONFIG_F2FS_FS_LZ4
586 static int f2fs_set_lz4hc_level(struct f2fs_sb_info *sbi, const char *str)
587 {
588 #ifdef CONFIG_F2FS_FS_LZ4HC
589 	unsigned int level;
590 #endif
591 
592 	if (strlen(str) == 3) {
593 		F2FS_OPTION(sbi).compress_level = 0;
594 		return 0;
595 	}
596 
597 #ifdef CONFIG_F2FS_FS_LZ4HC
598 	str += 3;
599 
600 	if (str[0] != ':') {
601 		f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
602 		return -EINVAL;
603 	}
604 	if (kstrtouint(str + 1, 10, &level))
605 		return -EINVAL;
606 
607 	if (level < LZ4HC_MIN_CLEVEL || level > LZ4HC_MAX_CLEVEL) {
608 		f2fs_info(sbi, "invalid lz4hc compress level: %d", level);
609 		return -EINVAL;
610 	}
611 
612 	F2FS_OPTION(sbi).compress_level = level;
613 	return 0;
614 #else
615 	f2fs_info(sbi, "kernel doesn't support lz4hc compression");
616 	return -EINVAL;
617 #endif
618 }
619 #endif
620 
621 #ifdef CONFIG_F2FS_FS_ZSTD
622 static int f2fs_set_zstd_level(struct f2fs_sb_info *sbi, const char *str)
623 {
624 	unsigned int level;
625 	int len = 4;
626 
627 	if (strlen(str) == len) {
628 		F2FS_OPTION(sbi).compress_level = 0;
629 		return 0;
630 	}
631 
632 	str += len;
633 
634 	if (str[0] != ':') {
635 		f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
636 		return -EINVAL;
637 	}
638 	if (kstrtouint(str + 1, 10, &level))
639 		return -EINVAL;
640 
641 	if (!level || level > zstd_max_clevel()) {
642 		f2fs_info(sbi, "invalid zstd compress level: %d", level);
643 		return -EINVAL;
644 	}
645 
646 	F2FS_OPTION(sbi).compress_level = level;
647 	return 0;
648 }
649 #endif
650 #endif
651 
652 static int parse_options(struct super_block *sb, char *options, bool is_remount)
653 {
654 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
655 	substring_t args[MAX_OPT_ARGS];
656 #ifdef CONFIG_F2FS_FS_COMPRESSION
657 	unsigned char (*ext)[F2FS_EXTENSION_LEN];
658 	unsigned char (*noext)[F2FS_EXTENSION_LEN];
659 	int ext_cnt, noext_cnt;
660 #endif
661 	char *p, *name;
662 	int arg = 0;
663 	kuid_t uid;
664 	kgid_t gid;
665 	int ret;
666 
667 	if (!options)
668 		goto default_check;
669 
670 	while ((p = strsep(&options, ",")) != NULL) {
671 		int token;
672 
673 		if (!*p)
674 			continue;
675 		/*
676 		 * Initialize args struct so we know whether arg was
677 		 * found; some options take optional arguments.
678 		 */
679 		args[0].to = args[0].from = NULL;
680 		token = match_token(p, f2fs_tokens, args);
681 
682 		switch (token) {
683 		case Opt_gc_background:
684 			name = match_strdup(&args[0]);
685 
686 			if (!name)
687 				return -ENOMEM;
688 			if (!strcmp(name, "on")) {
689 				F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
690 			} else if (!strcmp(name, "off")) {
691 				F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_OFF;
692 			} else if (!strcmp(name, "sync")) {
693 				F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_SYNC;
694 			} else {
695 				kfree(name);
696 				return -EINVAL;
697 			}
698 			kfree(name);
699 			break;
700 		case Opt_disable_roll_forward:
701 			set_opt(sbi, DISABLE_ROLL_FORWARD);
702 			break;
703 		case Opt_norecovery:
704 			/* this option mounts f2fs with ro */
705 			set_opt(sbi, NORECOVERY);
706 			if (!f2fs_readonly(sb))
707 				return -EINVAL;
708 			break;
709 		case Opt_discard:
710 			if (!f2fs_hw_support_discard(sbi)) {
711 				f2fs_warn(sbi, "device does not support discard");
712 				break;
713 			}
714 			set_opt(sbi, DISCARD);
715 			break;
716 		case Opt_nodiscard:
717 			if (f2fs_hw_should_discard(sbi)) {
718 				f2fs_warn(sbi, "discard is required for zoned block devices");
719 				return -EINVAL;
720 			}
721 			clear_opt(sbi, DISCARD);
722 			break;
723 		case Opt_noheap:
724 			set_opt(sbi, NOHEAP);
725 			break;
726 		case Opt_heap:
727 			clear_opt(sbi, NOHEAP);
728 			break;
729 #ifdef CONFIG_F2FS_FS_XATTR
730 		case Opt_user_xattr:
731 			set_opt(sbi, XATTR_USER);
732 			break;
733 		case Opt_nouser_xattr:
734 			clear_opt(sbi, XATTR_USER);
735 			break;
736 		case Opt_inline_xattr:
737 			set_opt(sbi, INLINE_XATTR);
738 			break;
739 		case Opt_noinline_xattr:
740 			clear_opt(sbi, INLINE_XATTR);
741 			break;
742 		case Opt_inline_xattr_size:
743 			if (args->from && match_int(args, &arg))
744 				return -EINVAL;
745 			set_opt(sbi, INLINE_XATTR_SIZE);
746 			F2FS_OPTION(sbi).inline_xattr_size = arg;
747 			break;
748 #else
749 		case Opt_user_xattr:
750 			f2fs_info(sbi, "user_xattr options not supported");
751 			break;
752 		case Opt_nouser_xattr:
753 			f2fs_info(sbi, "nouser_xattr options not supported");
754 			break;
755 		case Opt_inline_xattr:
756 			f2fs_info(sbi, "inline_xattr options not supported");
757 			break;
758 		case Opt_noinline_xattr:
759 			f2fs_info(sbi, "noinline_xattr options not supported");
760 			break;
761 #endif
762 #ifdef CONFIG_F2FS_FS_POSIX_ACL
763 		case Opt_acl:
764 			set_opt(sbi, POSIX_ACL);
765 			break;
766 		case Opt_noacl:
767 			clear_opt(sbi, POSIX_ACL);
768 			break;
769 #else
770 		case Opt_acl:
771 			f2fs_info(sbi, "acl options not supported");
772 			break;
773 		case Opt_noacl:
774 			f2fs_info(sbi, "noacl options not supported");
775 			break;
776 #endif
777 		case Opt_active_logs:
778 			if (args->from && match_int(args, &arg))
779 				return -EINVAL;
780 			if (arg != 2 && arg != 4 &&
781 				arg != NR_CURSEG_PERSIST_TYPE)
782 				return -EINVAL;
783 			F2FS_OPTION(sbi).active_logs = arg;
784 			break;
785 		case Opt_disable_ext_identify:
786 			set_opt(sbi, DISABLE_EXT_IDENTIFY);
787 			break;
788 		case Opt_inline_data:
789 			set_opt(sbi, INLINE_DATA);
790 			break;
791 		case Opt_inline_dentry:
792 			set_opt(sbi, INLINE_DENTRY);
793 			break;
794 		case Opt_noinline_dentry:
795 			clear_opt(sbi, INLINE_DENTRY);
796 			break;
797 		case Opt_flush_merge:
798 			set_opt(sbi, FLUSH_MERGE);
799 			break;
800 		case Opt_noflush_merge:
801 			clear_opt(sbi, FLUSH_MERGE);
802 			break;
803 		case Opt_nobarrier:
804 			set_opt(sbi, NOBARRIER);
805 			break;
806 		case Opt_barrier:
807 			clear_opt(sbi, NOBARRIER);
808 			break;
809 		case Opt_fastboot:
810 			set_opt(sbi, FASTBOOT);
811 			break;
812 		case Opt_extent_cache:
813 			set_opt(sbi, READ_EXTENT_CACHE);
814 			break;
815 		case Opt_noextent_cache:
816 			clear_opt(sbi, READ_EXTENT_CACHE);
817 			break;
818 		case Opt_noinline_data:
819 			clear_opt(sbi, INLINE_DATA);
820 			break;
821 		case Opt_data_flush:
822 			set_opt(sbi, DATA_FLUSH);
823 			break;
824 		case Opt_reserve_root:
825 			if (args->from && match_int(args, &arg))
826 				return -EINVAL;
827 			if (test_opt(sbi, RESERVE_ROOT)) {
828 				f2fs_info(sbi, "Preserve previous reserve_root=%u",
829 					  F2FS_OPTION(sbi).root_reserved_blocks);
830 			} else {
831 				F2FS_OPTION(sbi).root_reserved_blocks = arg;
832 				set_opt(sbi, RESERVE_ROOT);
833 			}
834 			break;
835 		case Opt_resuid:
836 			if (args->from && match_int(args, &arg))
837 				return -EINVAL;
838 			uid = make_kuid(current_user_ns(), arg);
839 			if (!uid_valid(uid)) {
840 				f2fs_err(sbi, "Invalid uid value %d", arg);
841 				return -EINVAL;
842 			}
843 			F2FS_OPTION(sbi).s_resuid = uid;
844 			break;
845 		case Opt_resgid:
846 			if (args->from && match_int(args, &arg))
847 				return -EINVAL;
848 			gid = make_kgid(current_user_ns(), arg);
849 			if (!gid_valid(gid)) {
850 				f2fs_err(sbi, "Invalid gid value %d", arg);
851 				return -EINVAL;
852 			}
853 			F2FS_OPTION(sbi).s_resgid = gid;
854 			break;
855 		case Opt_mode:
856 			name = match_strdup(&args[0]);
857 
858 			if (!name)
859 				return -ENOMEM;
860 			if (!strcmp(name, "adaptive")) {
861 				if (f2fs_sb_has_blkzoned(sbi)) {
862 					f2fs_warn(sbi, "adaptive mode is not allowed with zoned block device feature");
863 					kfree(name);
864 					return -EINVAL;
865 				}
866 				F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
867 			} else if (!strcmp(name, "lfs")) {
868 				F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
869 			} else if (!strcmp(name, "fragment:segment")) {
870 				F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_SEG;
871 			} else if (!strcmp(name, "fragment:block")) {
872 				F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_BLK;
873 			} else {
874 				kfree(name);
875 				return -EINVAL;
876 			}
877 			kfree(name);
878 			break;
879 		case Opt_io_size_bits:
880 			if (args->from && match_int(args, &arg))
881 				return -EINVAL;
882 			if (arg <= 0 || arg > __ilog2_u32(BIO_MAX_VECS)) {
883 				f2fs_warn(sbi, "Not support %ld, larger than %d",
884 					BIT(arg), BIO_MAX_VECS);
885 				return -EINVAL;
886 			}
887 			F2FS_OPTION(sbi).write_io_size_bits = arg;
888 			break;
889 #ifdef CONFIG_F2FS_FAULT_INJECTION
890 		case Opt_fault_injection:
891 			if (args->from && match_int(args, &arg))
892 				return -EINVAL;
893 			f2fs_build_fault_attr(sbi, arg, F2FS_ALL_FAULT_TYPE);
894 			set_opt(sbi, FAULT_INJECTION);
895 			break;
896 
897 		case Opt_fault_type:
898 			if (args->from && match_int(args, &arg))
899 				return -EINVAL;
900 			f2fs_build_fault_attr(sbi, 0, arg);
901 			set_opt(sbi, FAULT_INJECTION);
902 			break;
903 #else
904 		case Opt_fault_injection:
905 			f2fs_info(sbi, "fault_injection options not supported");
906 			break;
907 
908 		case Opt_fault_type:
909 			f2fs_info(sbi, "fault_type options not supported");
910 			break;
911 #endif
912 		case Opt_lazytime:
913 			sb->s_flags |= SB_LAZYTIME;
914 			break;
915 		case Opt_nolazytime:
916 			sb->s_flags &= ~SB_LAZYTIME;
917 			break;
918 #ifdef CONFIG_QUOTA
919 		case Opt_quota:
920 		case Opt_usrquota:
921 			set_opt(sbi, USRQUOTA);
922 			break;
923 		case Opt_grpquota:
924 			set_opt(sbi, GRPQUOTA);
925 			break;
926 		case Opt_prjquota:
927 			set_opt(sbi, PRJQUOTA);
928 			break;
929 		case Opt_usrjquota:
930 			ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
931 			if (ret)
932 				return ret;
933 			break;
934 		case Opt_grpjquota:
935 			ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
936 			if (ret)
937 				return ret;
938 			break;
939 		case Opt_prjjquota:
940 			ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
941 			if (ret)
942 				return ret;
943 			break;
944 		case Opt_offusrjquota:
945 			ret = f2fs_clear_qf_name(sb, USRQUOTA);
946 			if (ret)
947 				return ret;
948 			break;
949 		case Opt_offgrpjquota:
950 			ret = f2fs_clear_qf_name(sb, GRPQUOTA);
951 			if (ret)
952 				return ret;
953 			break;
954 		case Opt_offprjjquota:
955 			ret = f2fs_clear_qf_name(sb, PRJQUOTA);
956 			if (ret)
957 				return ret;
958 			break;
959 		case Opt_jqfmt_vfsold:
960 			F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
961 			break;
962 		case Opt_jqfmt_vfsv0:
963 			F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
964 			break;
965 		case Opt_jqfmt_vfsv1:
966 			F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
967 			break;
968 		case Opt_noquota:
969 			clear_opt(sbi, QUOTA);
970 			clear_opt(sbi, USRQUOTA);
971 			clear_opt(sbi, GRPQUOTA);
972 			clear_opt(sbi, PRJQUOTA);
973 			break;
974 #else
975 		case Opt_quota:
976 		case Opt_usrquota:
977 		case Opt_grpquota:
978 		case Opt_prjquota:
979 		case Opt_usrjquota:
980 		case Opt_grpjquota:
981 		case Opt_prjjquota:
982 		case Opt_offusrjquota:
983 		case Opt_offgrpjquota:
984 		case Opt_offprjjquota:
985 		case Opt_jqfmt_vfsold:
986 		case Opt_jqfmt_vfsv0:
987 		case Opt_jqfmt_vfsv1:
988 		case Opt_noquota:
989 			f2fs_info(sbi, "quota operations not supported");
990 			break;
991 #endif
992 		case Opt_alloc:
993 			name = match_strdup(&args[0]);
994 			if (!name)
995 				return -ENOMEM;
996 
997 			if (!strcmp(name, "default")) {
998 				F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
999 			} else if (!strcmp(name, "reuse")) {
1000 				F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
1001 			} else {
1002 				kfree(name);
1003 				return -EINVAL;
1004 			}
1005 			kfree(name);
1006 			break;
1007 		case Opt_fsync:
1008 			name = match_strdup(&args[0]);
1009 			if (!name)
1010 				return -ENOMEM;
1011 			if (!strcmp(name, "posix")) {
1012 				F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
1013 			} else if (!strcmp(name, "strict")) {
1014 				F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_STRICT;
1015 			} else if (!strcmp(name, "nobarrier")) {
1016 				F2FS_OPTION(sbi).fsync_mode =
1017 							FSYNC_MODE_NOBARRIER;
1018 			} else {
1019 				kfree(name);
1020 				return -EINVAL;
1021 			}
1022 			kfree(name);
1023 			break;
1024 		case Opt_test_dummy_encryption:
1025 			ret = f2fs_set_test_dummy_encryption(sb, p, &args[0],
1026 							     is_remount);
1027 			if (ret)
1028 				return ret;
1029 			break;
1030 		case Opt_inlinecrypt:
1031 #ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
1032 			sb->s_flags |= SB_INLINECRYPT;
1033 #else
1034 			f2fs_info(sbi, "inline encryption not supported");
1035 #endif
1036 			break;
1037 		case Opt_checkpoint_disable_cap_perc:
1038 			if (args->from && match_int(args, &arg))
1039 				return -EINVAL;
1040 			if (arg < 0 || arg > 100)
1041 				return -EINVAL;
1042 			F2FS_OPTION(sbi).unusable_cap_perc = arg;
1043 			set_opt(sbi, DISABLE_CHECKPOINT);
1044 			break;
1045 		case Opt_checkpoint_disable_cap:
1046 			if (args->from && match_int(args, &arg))
1047 				return -EINVAL;
1048 			F2FS_OPTION(sbi).unusable_cap = arg;
1049 			set_opt(sbi, DISABLE_CHECKPOINT);
1050 			break;
1051 		case Opt_checkpoint_disable:
1052 			set_opt(sbi, DISABLE_CHECKPOINT);
1053 			break;
1054 		case Opt_checkpoint_enable:
1055 			clear_opt(sbi, DISABLE_CHECKPOINT);
1056 			break;
1057 		case Opt_checkpoint_merge:
1058 			set_opt(sbi, MERGE_CHECKPOINT);
1059 			break;
1060 		case Opt_nocheckpoint_merge:
1061 			clear_opt(sbi, MERGE_CHECKPOINT);
1062 			break;
1063 #ifdef CONFIG_F2FS_FS_COMPRESSION
1064 		case Opt_compress_algorithm:
1065 			if (!f2fs_sb_has_compression(sbi)) {
1066 				f2fs_info(sbi, "Image doesn't support compression");
1067 				break;
1068 			}
1069 			name = match_strdup(&args[0]);
1070 			if (!name)
1071 				return -ENOMEM;
1072 			if (!strcmp(name, "lzo")) {
1073 #ifdef CONFIG_F2FS_FS_LZO
1074 				F2FS_OPTION(sbi).compress_level = 0;
1075 				F2FS_OPTION(sbi).compress_algorithm =
1076 								COMPRESS_LZO;
1077 #else
1078 				f2fs_info(sbi, "kernel doesn't support lzo compression");
1079 #endif
1080 			} else if (!strncmp(name, "lz4", 3)) {
1081 #ifdef CONFIG_F2FS_FS_LZ4
1082 				ret = f2fs_set_lz4hc_level(sbi, name);
1083 				if (ret) {
1084 					kfree(name);
1085 					return -EINVAL;
1086 				}
1087 				F2FS_OPTION(sbi).compress_algorithm =
1088 								COMPRESS_LZ4;
1089 #else
1090 				f2fs_info(sbi, "kernel doesn't support lz4 compression");
1091 #endif
1092 			} else if (!strncmp(name, "zstd", 4)) {
1093 #ifdef CONFIG_F2FS_FS_ZSTD
1094 				ret = f2fs_set_zstd_level(sbi, name);
1095 				if (ret) {
1096 					kfree(name);
1097 					return -EINVAL;
1098 				}
1099 				F2FS_OPTION(sbi).compress_algorithm =
1100 								COMPRESS_ZSTD;
1101 #else
1102 				f2fs_info(sbi, "kernel doesn't support zstd compression");
1103 #endif
1104 			} else if (!strcmp(name, "lzo-rle")) {
1105 #ifdef CONFIG_F2FS_FS_LZORLE
1106 				F2FS_OPTION(sbi).compress_level = 0;
1107 				F2FS_OPTION(sbi).compress_algorithm =
1108 								COMPRESS_LZORLE;
1109 #else
1110 				f2fs_info(sbi, "kernel doesn't support lzorle compression");
1111 #endif
1112 			} else {
1113 				kfree(name);
1114 				return -EINVAL;
1115 			}
1116 			kfree(name);
1117 			break;
1118 		case Opt_compress_log_size:
1119 			if (!f2fs_sb_has_compression(sbi)) {
1120 				f2fs_info(sbi, "Image doesn't support compression");
1121 				break;
1122 			}
1123 			if (args->from && match_int(args, &arg))
1124 				return -EINVAL;
1125 			if (arg < MIN_COMPRESS_LOG_SIZE ||
1126 				arg > MAX_COMPRESS_LOG_SIZE) {
1127 				f2fs_err(sbi,
1128 					"Compress cluster log size is out of range");
1129 				return -EINVAL;
1130 			}
1131 			F2FS_OPTION(sbi).compress_log_size = arg;
1132 			break;
1133 		case Opt_compress_extension:
1134 			if (!f2fs_sb_has_compression(sbi)) {
1135 				f2fs_info(sbi, "Image doesn't support compression");
1136 				break;
1137 			}
1138 			name = match_strdup(&args[0]);
1139 			if (!name)
1140 				return -ENOMEM;
1141 
1142 			ext = F2FS_OPTION(sbi).extensions;
1143 			ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
1144 
1145 			if (strlen(name) >= F2FS_EXTENSION_LEN ||
1146 				ext_cnt >= COMPRESS_EXT_NUM) {
1147 				f2fs_err(sbi,
1148 					"invalid extension length/number");
1149 				kfree(name);
1150 				return -EINVAL;
1151 			}
1152 
1153 			strcpy(ext[ext_cnt], name);
1154 			F2FS_OPTION(sbi).compress_ext_cnt++;
1155 			kfree(name);
1156 			break;
1157 		case Opt_nocompress_extension:
1158 			if (!f2fs_sb_has_compression(sbi)) {
1159 				f2fs_info(sbi, "Image doesn't support compression");
1160 				break;
1161 			}
1162 			name = match_strdup(&args[0]);
1163 			if (!name)
1164 				return -ENOMEM;
1165 
1166 			noext = F2FS_OPTION(sbi).noextensions;
1167 			noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
1168 
1169 			if (strlen(name) >= F2FS_EXTENSION_LEN ||
1170 				noext_cnt >= COMPRESS_EXT_NUM) {
1171 				f2fs_err(sbi,
1172 					"invalid extension length/number");
1173 				kfree(name);
1174 				return -EINVAL;
1175 			}
1176 
1177 			strcpy(noext[noext_cnt], name);
1178 			F2FS_OPTION(sbi).nocompress_ext_cnt++;
1179 			kfree(name);
1180 			break;
1181 		case Opt_compress_chksum:
1182 			if (!f2fs_sb_has_compression(sbi)) {
1183 				f2fs_info(sbi, "Image doesn't support compression");
1184 				break;
1185 			}
1186 			F2FS_OPTION(sbi).compress_chksum = true;
1187 			break;
1188 		case Opt_compress_mode:
1189 			if (!f2fs_sb_has_compression(sbi)) {
1190 				f2fs_info(sbi, "Image doesn't support compression");
1191 				break;
1192 			}
1193 			name = match_strdup(&args[0]);
1194 			if (!name)
1195 				return -ENOMEM;
1196 			if (!strcmp(name, "fs")) {
1197 				F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
1198 			} else if (!strcmp(name, "user")) {
1199 				F2FS_OPTION(sbi).compress_mode = COMPR_MODE_USER;
1200 			} else {
1201 				kfree(name);
1202 				return -EINVAL;
1203 			}
1204 			kfree(name);
1205 			break;
1206 		case Opt_compress_cache:
1207 			if (!f2fs_sb_has_compression(sbi)) {
1208 				f2fs_info(sbi, "Image doesn't support compression");
1209 				break;
1210 			}
1211 			set_opt(sbi, COMPRESS_CACHE);
1212 			break;
1213 #else
1214 		case Opt_compress_algorithm:
1215 		case Opt_compress_log_size:
1216 		case Opt_compress_extension:
1217 		case Opt_nocompress_extension:
1218 		case Opt_compress_chksum:
1219 		case Opt_compress_mode:
1220 		case Opt_compress_cache:
1221 			f2fs_info(sbi, "compression options not supported");
1222 			break;
1223 #endif
1224 		case Opt_atgc:
1225 			set_opt(sbi, ATGC);
1226 			break;
1227 		case Opt_gc_merge:
1228 			set_opt(sbi, GC_MERGE);
1229 			break;
1230 		case Opt_nogc_merge:
1231 			clear_opt(sbi, GC_MERGE);
1232 			break;
1233 		case Opt_discard_unit:
1234 			name = match_strdup(&args[0]);
1235 			if (!name)
1236 				return -ENOMEM;
1237 			if (!strcmp(name, "block")) {
1238 				F2FS_OPTION(sbi).discard_unit =
1239 						DISCARD_UNIT_BLOCK;
1240 			} else if (!strcmp(name, "segment")) {
1241 				F2FS_OPTION(sbi).discard_unit =
1242 						DISCARD_UNIT_SEGMENT;
1243 			} else if (!strcmp(name, "section")) {
1244 				F2FS_OPTION(sbi).discard_unit =
1245 						DISCARD_UNIT_SECTION;
1246 			} else {
1247 				kfree(name);
1248 				return -EINVAL;
1249 			}
1250 			kfree(name);
1251 			break;
1252 		case Opt_memory_mode:
1253 			name = match_strdup(&args[0]);
1254 			if (!name)
1255 				return -ENOMEM;
1256 			if (!strcmp(name, "normal")) {
1257 				F2FS_OPTION(sbi).memory_mode =
1258 						MEMORY_MODE_NORMAL;
1259 			} else if (!strcmp(name, "low")) {
1260 				F2FS_OPTION(sbi).memory_mode =
1261 						MEMORY_MODE_LOW;
1262 			} else {
1263 				kfree(name);
1264 				return -EINVAL;
1265 			}
1266 			kfree(name);
1267 			break;
1268 		case Opt_age_extent_cache:
1269 			set_opt(sbi, AGE_EXTENT_CACHE);
1270 			break;
1271 		default:
1272 			f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
1273 				 p);
1274 			return -EINVAL;
1275 		}
1276 	}
1277 default_check:
1278 #ifdef CONFIG_QUOTA
1279 	if (f2fs_check_quota_options(sbi))
1280 		return -EINVAL;
1281 #else
1282 	if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sbi->sb)) {
1283 		f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1284 		return -EINVAL;
1285 	}
1286 	if (f2fs_sb_has_project_quota(sbi) && !f2fs_readonly(sbi->sb)) {
1287 		f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1288 		return -EINVAL;
1289 	}
1290 #endif
1291 #if !IS_ENABLED(CONFIG_UNICODE)
1292 	if (f2fs_sb_has_casefold(sbi)) {
1293 		f2fs_err(sbi,
1294 			"Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
1295 		return -EINVAL;
1296 	}
1297 #endif
1298 	/*
1299 	 * The BLKZONED feature indicates that the drive was formatted with
1300 	 * zone alignment optimization. This is optional for host-aware
1301 	 * devices, but mandatory for host-managed zoned block devices.
1302 	 */
1303 	if (f2fs_sb_has_blkzoned(sbi)) {
1304 #ifdef CONFIG_BLK_DEV_ZONED
1305 		if (F2FS_OPTION(sbi).discard_unit !=
1306 						DISCARD_UNIT_SECTION) {
1307 			f2fs_info(sbi, "Zoned block device doesn't need small discard, set discard_unit=section by default");
1308 			F2FS_OPTION(sbi).discard_unit =
1309 					DISCARD_UNIT_SECTION;
1310 		}
1311 #else
1312 		f2fs_err(sbi, "Zoned block device support is not enabled");
1313 		return -EINVAL;
1314 #endif
1315 	}
1316 
1317 #ifdef CONFIG_F2FS_FS_COMPRESSION
1318 	if (f2fs_test_compress_extension(sbi)) {
1319 		f2fs_err(sbi, "invalid compress or nocompress extension");
1320 		return -EINVAL;
1321 	}
1322 #endif
1323 
1324 	if (F2FS_IO_SIZE_BITS(sbi) && !f2fs_lfs_mode(sbi)) {
1325 		f2fs_err(sbi, "Should set mode=lfs with %luKB-sized IO",
1326 			 F2FS_IO_SIZE_KB(sbi));
1327 		return -EINVAL;
1328 	}
1329 
1330 	if (test_opt(sbi, INLINE_XATTR_SIZE)) {
1331 		int min_size, max_size;
1332 
1333 		if (!f2fs_sb_has_extra_attr(sbi) ||
1334 			!f2fs_sb_has_flexible_inline_xattr(sbi)) {
1335 			f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
1336 			return -EINVAL;
1337 		}
1338 		if (!test_opt(sbi, INLINE_XATTR)) {
1339 			f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
1340 			return -EINVAL;
1341 		}
1342 
1343 		min_size = sizeof(struct f2fs_xattr_header) / sizeof(__le32);
1344 		max_size = MAX_INLINE_XATTR_SIZE;
1345 
1346 		if (F2FS_OPTION(sbi).inline_xattr_size < min_size ||
1347 				F2FS_OPTION(sbi).inline_xattr_size > max_size) {
1348 			f2fs_err(sbi, "inline xattr size is out of range: %d ~ %d",
1349 				 min_size, max_size);
1350 			return -EINVAL;
1351 		}
1352 	}
1353 
1354 	if (test_opt(sbi, DISABLE_CHECKPOINT) && f2fs_lfs_mode(sbi)) {
1355 		f2fs_err(sbi, "LFS is not compatible with checkpoint=disable");
1356 		return -EINVAL;
1357 	}
1358 
1359 	if (test_opt(sbi, ATGC) && f2fs_lfs_mode(sbi)) {
1360 		f2fs_err(sbi, "LFS is not compatible with ATGC");
1361 		return -EINVAL;
1362 	}
1363 
1364 	if (f2fs_is_readonly(sbi) && test_opt(sbi, FLUSH_MERGE)) {
1365 		f2fs_err(sbi, "FLUSH_MERGE not compatible with readonly mode");
1366 		return -EINVAL;
1367 	}
1368 
1369 	if (f2fs_sb_has_readonly(sbi) && !f2fs_readonly(sbi->sb)) {
1370 		f2fs_err(sbi, "Allow to mount readonly mode only");
1371 		return -EROFS;
1372 	}
1373 	return 0;
1374 }
1375 
1376 static struct inode *f2fs_alloc_inode(struct super_block *sb)
1377 {
1378 	struct f2fs_inode_info *fi;
1379 
1380 	if (time_to_inject(F2FS_SB(sb), FAULT_SLAB_ALLOC))
1381 		return NULL;
1382 
1383 	fi = alloc_inode_sb(sb, f2fs_inode_cachep, GFP_F2FS_ZERO);
1384 	if (!fi)
1385 		return NULL;
1386 
1387 	init_once((void *) fi);
1388 
1389 	/* Initialize f2fs-specific inode info */
1390 	atomic_set(&fi->dirty_pages, 0);
1391 	atomic_set(&fi->i_compr_blocks, 0);
1392 	init_f2fs_rwsem(&fi->i_sem);
1393 	spin_lock_init(&fi->i_size_lock);
1394 	INIT_LIST_HEAD(&fi->dirty_list);
1395 	INIT_LIST_HEAD(&fi->gdirty_list);
1396 	init_f2fs_rwsem(&fi->i_gc_rwsem[READ]);
1397 	init_f2fs_rwsem(&fi->i_gc_rwsem[WRITE]);
1398 	init_f2fs_rwsem(&fi->i_xattr_sem);
1399 
1400 	/* Will be used by directory only */
1401 	fi->i_dir_level = F2FS_SB(sb)->dir_level;
1402 
1403 	return &fi->vfs_inode;
1404 }
1405 
1406 static int f2fs_drop_inode(struct inode *inode)
1407 {
1408 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1409 	int ret;
1410 
1411 	/*
1412 	 * during filesystem shutdown, if checkpoint is disabled,
1413 	 * drop useless meta/node dirty pages.
1414 	 */
1415 	if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1416 		if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1417 			inode->i_ino == F2FS_META_INO(sbi)) {
1418 			trace_f2fs_drop_inode(inode, 1);
1419 			return 1;
1420 		}
1421 	}
1422 
1423 	/*
1424 	 * This is to avoid a deadlock condition like below.
1425 	 * writeback_single_inode(inode)
1426 	 *  - f2fs_write_data_page
1427 	 *    - f2fs_gc -> iput -> evict
1428 	 *       - inode_wait_for_writeback(inode)
1429 	 */
1430 	if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
1431 		if (!inode->i_nlink && !is_bad_inode(inode)) {
1432 			/* to avoid evict_inode call simultaneously */
1433 			atomic_inc(&inode->i_count);
1434 			spin_unlock(&inode->i_lock);
1435 
1436 			/* should remain fi->extent_tree for writepage */
1437 			f2fs_destroy_extent_node(inode);
1438 
1439 			sb_start_intwrite(inode->i_sb);
1440 			f2fs_i_size_write(inode, 0);
1441 
1442 			f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1443 					inode, NULL, 0, DATA);
1444 			truncate_inode_pages_final(inode->i_mapping);
1445 
1446 			if (F2FS_HAS_BLOCKS(inode))
1447 				f2fs_truncate(inode);
1448 
1449 			sb_end_intwrite(inode->i_sb);
1450 
1451 			spin_lock(&inode->i_lock);
1452 			atomic_dec(&inode->i_count);
1453 		}
1454 		trace_f2fs_drop_inode(inode, 0);
1455 		return 0;
1456 	}
1457 	ret = generic_drop_inode(inode);
1458 	if (!ret)
1459 		ret = fscrypt_drop_inode(inode);
1460 	trace_f2fs_drop_inode(inode, ret);
1461 	return ret;
1462 }
1463 
1464 int f2fs_inode_dirtied(struct inode *inode, bool sync)
1465 {
1466 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1467 	int ret = 0;
1468 
1469 	spin_lock(&sbi->inode_lock[DIRTY_META]);
1470 	if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1471 		ret = 1;
1472 	} else {
1473 		set_inode_flag(inode, FI_DIRTY_INODE);
1474 		stat_inc_dirty_inode(sbi, DIRTY_META);
1475 	}
1476 	if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1477 		list_add_tail(&F2FS_I(inode)->gdirty_list,
1478 				&sbi->inode_list[DIRTY_META]);
1479 		inc_page_count(sbi, F2FS_DIRTY_IMETA);
1480 	}
1481 	spin_unlock(&sbi->inode_lock[DIRTY_META]);
1482 	return ret;
1483 }
1484 
1485 void f2fs_inode_synced(struct inode *inode)
1486 {
1487 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1488 
1489 	spin_lock(&sbi->inode_lock[DIRTY_META]);
1490 	if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1491 		spin_unlock(&sbi->inode_lock[DIRTY_META]);
1492 		return;
1493 	}
1494 	if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1495 		list_del_init(&F2FS_I(inode)->gdirty_list);
1496 		dec_page_count(sbi, F2FS_DIRTY_IMETA);
1497 	}
1498 	clear_inode_flag(inode, FI_DIRTY_INODE);
1499 	clear_inode_flag(inode, FI_AUTO_RECOVER);
1500 	stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1501 	spin_unlock(&sbi->inode_lock[DIRTY_META]);
1502 }
1503 
1504 /*
1505  * f2fs_dirty_inode() is called from __mark_inode_dirty()
1506  *
1507  * We should call set_dirty_inode to write the dirty inode through write_inode.
1508  */
1509 static void f2fs_dirty_inode(struct inode *inode, int flags)
1510 {
1511 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1512 
1513 	if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1514 			inode->i_ino == F2FS_META_INO(sbi))
1515 		return;
1516 
1517 	if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1518 		clear_inode_flag(inode, FI_AUTO_RECOVER);
1519 
1520 	f2fs_inode_dirtied(inode, false);
1521 }
1522 
1523 static void f2fs_free_inode(struct inode *inode)
1524 {
1525 	fscrypt_free_inode(inode);
1526 	kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1527 }
1528 
1529 static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1530 {
1531 	percpu_counter_destroy(&sbi->total_valid_inode_count);
1532 	percpu_counter_destroy(&sbi->rf_node_block_count);
1533 	percpu_counter_destroy(&sbi->alloc_valid_block_count);
1534 }
1535 
1536 static void destroy_device_list(struct f2fs_sb_info *sbi)
1537 {
1538 	int i;
1539 
1540 	for (i = 0; i < sbi->s_ndevs; i++) {
1541 		blkdev_put(FDEV(i).bdev, FMODE_EXCL);
1542 #ifdef CONFIG_BLK_DEV_ZONED
1543 		kvfree(FDEV(i).blkz_seq);
1544 #endif
1545 	}
1546 	kvfree(sbi->devs);
1547 }
1548 
1549 static void f2fs_put_super(struct super_block *sb)
1550 {
1551 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1552 	int i;
1553 	bool done;
1554 
1555 	/* unregister procfs/sysfs entries in advance to avoid race case */
1556 	f2fs_unregister_sysfs(sbi);
1557 
1558 	f2fs_quota_off_umount(sb);
1559 
1560 	/* prevent remaining shrinker jobs */
1561 	mutex_lock(&sbi->umount_mutex);
1562 
1563 	/*
1564 	 * flush all issued checkpoints and stop checkpoint issue thread.
1565 	 * after then, all checkpoints should be done by each process context.
1566 	 */
1567 	f2fs_stop_ckpt_thread(sbi);
1568 
1569 	/*
1570 	 * We don't need to do checkpoint when superblock is clean.
1571 	 * But, the previous checkpoint was not done by umount, it needs to do
1572 	 * clean checkpoint again.
1573 	 */
1574 	if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1575 			!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1576 		struct cp_control cpc = {
1577 			.reason = CP_UMOUNT,
1578 		};
1579 		f2fs_write_checkpoint(sbi, &cpc);
1580 	}
1581 
1582 	/* be sure to wait for any on-going discard commands */
1583 	done = f2fs_issue_discard_timeout(sbi);
1584 	if (f2fs_realtime_discard_enable(sbi) && !sbi->discard_blks && done) {
1585 		struct cp_control cpc = {
1586 			.reason = CP_UMOUNT | CP_TRIMMED,
1587 		};
1588 		f2fs_write_checkpoint(sbi, &cpc);
1589 	}
1590 
1591 	/*
1592 	 * normally superblock is clean, so we need to release this.
1593 	 * In addition, EIO will skip do checkpoint, we need this as well.
1594 	 */
1595 	f2fs_release_ino_entry(sbi, true);
1596 
1597 	f2fs_leave_shrinker(sbi);
1598 	mutex_unlock(&sbi->umount_mutex);
1599 
1600 	/* our cp_error case, we can wait for any writeback page */
1601 	f2fs_flush_merged_writes(sbi);
1602 
1603 	f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1604 
1605 	f2fs_bug_on(sbi, sbi->fsync_node_num);
1606 
1607 	f2fs_destroy_compress_inode(sbi);
1608 
1609 	iput(sbi->node_inode);
1610 	sbi->node_inode = NULL;
1611 
1612 	iput(sbi->meta_inode);
1613 	sbi->meta_inode = NULL;
1614 
1615 	/*
1616 	 * iput() can update stat information, if f2fs_write_checkpoint()
1617 	 * above failed with error.
1618 	 */
1619 	f2fs_destroy_stats(sbi);
1620 
1621 	/* destroy f2fs internal modules */
1622 	f2fs_destroy_node_manager(sbi);
1623 	f2fs_destroy_segment_manager(sbi);
1624 
1625 	f2fs_destroy_post_read_wq(sbi);
1626 
1627 	kvfree(sbi->ckpt);
1628 
1629 	sb->s_fs_info = NULL;
1630 	if (sbi->s_chksum_driver)
1631 		crypto_free_shash(sbi->s_chksum_driver);
1632 	kfree(sbi->raw_super);
1633 
1634 	destroy_device_list(sbi);
1635 	f2fs_destroy_page_array_cache(sbi);
1636 	f2fs_destroy_xattr_caches(sbi);
1637 	mempool_destroy(sbi->write_io_dummy);
1638 #ifdef CONFIG_QUOTA
1639 	for (i = 0; i < MAXQUOTAS; i++)
1640 		kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1641 #endif
1642 	fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
1643 	destroy_percpu_info(sbi);
1644 	f2fs_destroy_iostat(sbi);
1645 	for (i = 0; i < NR_PAGE_TYPE; i++)
1646 		kvfree(sbi->write_io[i]);
1647 #if IS_ENABLED(CONFIG_UNICODE)
1648 	utf8_unload(sb->s_encoding);
1649 #endif
1650 	kfree(sbi);
1651 }
1652 
1653 int f2fs_sync_fs(struct super_block *sb, int sync)
1654 {
1655 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1656 	int err = 0;
1657 
1658 	if (unlikely(f2fs_cp_error(sbi)))
1659 		return 0;
1660 	if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1661 		return 0;
1662 
1663 	trace_f2fs_sync_fs(sb, sync);
1664 
1665 	if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1666 		return -EAGAIN;
1667 
1668 	if (sync)
1669 		err = f2fs_issue_checkpoint(sbi);
1670 
1671 	return err;
1672 }
1673 
1674 static int f2fs_freeze(struct super_block *sb)
1675 {
1676 	if (f2fs_readonly(sb))
1677 		return 0;
1678 
1679 	/* IO error happened before */
1680 	if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1681 		return -EIO;
1682 
1683 	/* must be clean, since sync_filesystem() was already called */
1684 	if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1685 		return -EINVAL;
1686 
1687 	/* Let's flush checkpoints and stop the thread. */
1688 	f2fs_flush_ckpt_thread(F2FS_SB(sb));
1689 
1690 	/* to avoid deadlock on f2fs_evict_inode->SB_FREEZE_FS */
1691 	set_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1692 	return 0;
1693 }
1694 
1695 static int f2fs_unfreeze(struct super_block *sb)
1696 {
1697 	clear_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1698 	return 0;
1699 }
1700 
1701 #ifdef CONFIG_QUOTA
1702 static int f2fs_statfs_project(struct super_block *sb,
1703 				kprojid_t projid, struct kstatfs *buf)
1704 {
1705 	struct kqid qid;
1706 	struct dquot *dquot;
1707 	u64 limit;
1708 	u64 curblock;
1709 
1710 	qid = make_kqid_projid(projid);
1711 	dquot = dqget(sb, qid);
1712 	if (IS_ERR(dquot))
1713 		return PTR_ERR(dquot);
1714 	spin_lock(&dquot->dq_dqb_lock);
1715 
1716 	limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1717 					dquot->dq_dqb.dqb_bhardlimit);
1718 	if (limit)
1719 		limit >>= sb->s_blocksize_bits;
1720 
1721 	if (limit && buf->f_blocks > limit) {
1722 		curblock = (dquot->dq_dqb.dqb_curspace +
1723 			    dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
1724 		buf->f_blocks = limit;
1725 		buf->f_bfree = buf->f_bavail =
1726 			(buf->f_blocks > curblock) ?
1727 			 (buf->f_blocks - curblock) : 0;
1728 	}
1729 
1730 	limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1731 					dquot->dq_dqb.dqb_ihardlimit);
1732 
1733 	if (limit && buf->f_files > limit) {
1734 		buf->f_files = limit;
1735 		buf->f_ffree =
1736 			(buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1737 			 (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1738 	}
1739 
1740 	spin_unlock(&dquot->dq_dqb_lock);
1741 	dqput(dquot);
1742 	return 0;
1743 }
1744 #endif
1745 
1746 static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1747 {
1748 	struct super_block *sb = dentry->d_sb;
1749 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1750 	u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
1751 	block_t total_count, user_block_count, start_count;
1752 	u64 avail_node_count;
1753 	unsigned int total_valid_node_count;
1754 
1755 	total_count = le64_to_cpu(sbi->raw_super->block_count);
1756 	start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
1757 	buf->f_type = F2FS_SUPER_MAGIC;
1758 	buf->f_bsize = sbi->blocksize;
1759 
1760 	buf->f_blocks = total_count - start_count;
1761 
1762 	spin_lock(&sbi->stat_lock);
1763 
1764 	user_block_count = sbi->user_block_count;
1765 	total_valid_node_count = valid_node_count(sbi);
1766 	avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
1767 	buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1768 						sbi->current_reserved_blocks;
1769 
1770 	if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1771 		buf->f_bfree = 0;
1772 	else
1773 		buf->f_bfree -= sbi->unusable_block_count;
1774 	spin_unlock(&sbi->stat_lock);
1775 
1776 	if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1777 		buf->f_bavail = buf->f_bfree -
1778 				F2FS_OPTION(sbi).root_reserved_blocks;
1779 	else
1780 		buf->f_bavail = 0;
1781 
1782 	if (avail_node_count > user_block_count) {
1783 		buf->f_files = user_block_count;
1784 		buf->f_ffree = buf->f_bavail;
1785 	} else {
1786 		buf->f_files = avail_node_count;
1787 		buf->f_ffree = min(avail_node_count - total_valid_node_count,
1788 					buf->f_bavail);
1789 	}
1790 
1791 	buf->f_namelen = F2FS_NAME_LEN;
1792 	buf->f_fsid    = u64_to_fsid(id);
1793 
1794 #ifdef CONFIG_QUOTA
1795 	if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1796 			sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1797 		f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1798 	}
1799 #endif
1800 	return 0;
1801 }
1802 
1803 static inline void f2fs_show_quota_options(struct seq_file *seq,
1804 					   struct super_block *sb)
1805 {
1806 #ifdef CONFIG_QUOTA
1807 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1808 
1809 	if (F2FS_OPTION(sbi).s_jquota_fmt) {
1810 		char *fmtname = "";
1811 
1812 		switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1813 		case QFMT_VFS_OLD:
1814 			fmtname = "vfsold";
1815 			break;
1816 		case QFMT_VFS_V0:
1817 			fmtname = "vfsv0";
1818 			break;
1819 		case QFMT_VFS_V1:
1820 			fmtname = "vfsv1";
1821 			break;
1822 		}
1823 		seq_printf(seq, ",jqfmt=%s", fmtname);
1824 	}
1825 
1826 	if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1827 		seq_show_option(seq, "usrjquota",
1828 			F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1829 
1830 	if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1831 		seq_show_option(seq, "grpjquota",
1832 			F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1833 
1834 	if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1835 		seq_show_option(seq, "prjjquota",
1836 			F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1837 #endif
1838 }
1839 
1840 #ifdef CONFIG_F2FS_FS_COMPRESSION
1841 static inline void f2fs_show_compress_options(struct seq_file *seq,
1842 							struct super_block *sb)
1843 {
1844 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1845 	char *algtype = "";
1846 	int i;
1847 
1848 	if (!f2fs_sb_has_compression(sbi))
1849 		return;
1850 
1851 	switch (F2FS_OPTION(sbi).compress_algorithm) {
1852 	case COMPRESS_LZO:
1853 		algtype = "lzo";
1854 		break;
1855 	case COMPRESS_LZ4:
1856 		algtype = "lz4";
1857 		break;
1858 	case COMPRESS_ZSTD:
1859 		algtype = "zstd";
1860 		break;
1861 	case COMPRESS_LZORLE:
1862 		algtype = "lzo-rle";
1863 		break;
1864 	}
1865 	seq_printf(seq, ",compress_algorithm=%s", algtype);
1866 
1867 	if (F2FS_OPTION(sbi).compress_level)
1868 		seq_printf(seq, ":%d", F2FS_OPTION(sbi).compress_level);
1869 
1870 	seq_printf(seq, ",compress_log_size=%u",
1871 			F2FS_OPTION(sbi).compress_log_size);
1872 
1873 	for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1874 		seq_printf(seq, ",compress_extension=%s",
1875 			F2FS_OPTION(sbi).extensions[i]);
1876 	}
1877 
1878 	for (i = 0; i < F2FS_OPTION(sbi).nocompress_ext_cnt; i++) {
1879 		seq_printf(seq, ",nocompress_extension=%s",
1880 			F2FS_OPTION(sbi).noextensions[i]);
1881 	}
1882 
1883 	if (F2FS_OPTION(sbi).compress_chksum)
1884 		seq_puts(seq, ",compress_chksum");
1885 
1886 	if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_FS)
1887 		seq_printf(seq, ",compress_mode=%s", "fs");
1888 	else if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_USER)
1889 		seq_printf(seq, ",compress_mode=%s", "user");
1890 
1891 	if (test_opt(sbi, COMPRESS_CACHE))
1892 		seq_puts(seq, ",compress_cache");
1893 }
1894 #endif
1895 
1896 static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1897 {
1898 	struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1899 
1900 	if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1901 		seq_printf(seq, ",background_gc=%s", "sync");
1902 	else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1903 		seq_printf(seq, ",background_gc=%s", "on");
1904 	else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1905 		seq_printf(seq, ",background_gc=%s", "off");
1906 
1907 	if (test_opt(sbi, GC_MERGE))
1908 		seq_puts(seq, ",gc_merge");
1909 	else
1910 		seq_puts(seq, ",nogc_merge");
1911 
1912 	if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1913 		seq_puts(seq, ",disable_roll_forward");
1914 	if (test_opt(sbi, NORECOVERY))
1915 		seq_puts(seq, ",norecovery");
1916 	if (test_opt(sbi, DISCARD)) {
1917 		seq_puts(seq, ",discard");
1918 		if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK)
1919 			seq_printf(seq, ",discard_unit=%s", "block");
1920 		else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SEGMENT)
1921 			seq_printf(seq, ",discard_unit=%s", "segment");
1922 		else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SECTION)
1923 			seq_printf(seq, ",discard_unit=%s", "section");
1924 	} else {
1925 		seq_puts(seq, ",nodiscard");
1926 	}
1927 	if (test_opt(sbi, NOHEAP))
1928 		seq_puts(seq, ",no_heap");
1929 	else
1930 		seq_puts(seq, ",heap");
1931 #ifdef CONFIG_F2FS_FS_XATTR
1932 	if (test_opt(sbi, XATTR_USER))
1933 		seq_puts(seq, ",user_xattr");
1934 	else
1935 		seq_puts(seq, ",nouser_xattr");
1936 	if (test_opt(sbi, INLINE_XATTR))
1937 		seq_puts(seq, ",inline_xattr");
1938 	else
1939 		seq_puts(seq, ",noinline_xattr");
1940 	if (test_opt(sbi, INLINE_XATTR_SIZE))
1941 		seq_printf(seq, ",inline_xattr_size=%u",
1942 					F2FS_OPTION(sbi).inline_xattr_size);
1943 #endif
1944 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1945 	if (test_opt(sbi, POSIX_ACL))
1946 		seq_puts(seq, ",acl");
1947 	else
1948 		seq_puts(seq, ",noacl");
1949 #endif
1950 	if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
1951 		seq_puts(seq, ",disable_ext_identify");
1952 	if (test_opt(sbi, INLINE_DATA))
1953 		seq_puts(seq, ",inline_data");
1954 	else
1955 		seq_puts(seq, ",noinline_data");
1956 	if (test_opt(sbi, INLINE_DENTRY))
1957 		seq_puts(seq, ",inline_dentry");
1958 	else
1959 		seq_puts(seq, ",noinline_dentry");
1960 	if (test_opt(sbi, FLUSH_MERGE))
1961 		seq_puts(seq, ",flush_merge");
1962 	else
1963 		seq_puts(seq, ",noflush_merge");
1964 	if (test_opt(sbi, NOBARRIER))
1965 		seq_puts(seq, ",nobarrier");
1966 	else
1967 		seq_puts(seq, ",barrier");
1968 	if (test_opt(sbi, FASTBOOT))
1969 		seq_puts(seq, ",fastboot");
1970 	if (test_opt(sbi, READ_EXTENT_CACHE))
1971 		seq_puts(seq, ",extent_cache");
1972 	else
1973 		seq_puts(seq, ",noextent_cache");
1974 	if (test_opt(sbi, AGE_EXTENT_CACHE))
1975 		seq_puts(seq, ",age_extent_cache");
1976 	if (test_opt(sbi, DATA_FLUSH))
1977 		seq_puts(seq, ",data_flush");
1978 
1979 	seq_puts(seq, ",mode=");
1980 	if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
1981 		seq_puts(seq, "adaptive");
1982 	else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
1983 		seq_puts(seq, "lfs");
1984 	else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG)
1985 		seq_puts(seq, "fragment:segment");
1986 	else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK)
1987 		seq_puts(seq, "fragment:block");
1988 	seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
1989 	if (test_opt(sbi, RESERVE_ROOT))
1990 		seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
1991 				F2FS_OPTION(sbi).root_reserved_blocks,
1992 				from_kuid_munged(&init_user_ns,
1993 					F2FS_OPTION(sbi).s_resuid),
1994 				from_kgid_munged(&init_user_ns,
1995 					F2FS_OPTION(sbi).s_resgid));
1996 	if (F2FS_IO_SIZE_BITS(sbi))
1997 		seq_printf(seq, ",io_bits=%u",
1998 				F2FS_OPTION(sbi).write_io_size_bits);
1999 #ifdef CONFIG_F2FS_FAULT_INJECTION
2000 	if (test_opt(sbi, FAULT_INJECTION)) {
2001 		seq_printf(seq, ",fault_injection=%u",
2002 				F2FS_OPTION(sbi).fault_info.inject_rate);
2003 		seq_printf(seq, ",fault_type=%u",
2004 				F2FS_OPTION(sbi).fault_info.inject_type);
2005 	}
2006 #endif
2007 #ifdef CONFIG_QUOTA
2008 	if (test_opt(sbi, QUOTA))
2009 		seq_puts(seq, ",quota");
2010 	if (test_opt(sbi, USRQUOTA))
2011 		seq_puts(seq, ",usrquota");
2012 	if (test_opt(sbi, GRPQUOTA))
2013 		seq_puts(seq, ",grpquota");
2014 	if (test_opt(sbi, PRJQUOTA))
2015 		seq_puts(seq, ",prjquota");
2016 #endif
2017 	f2fs_show_quota_options(seq, sbi->sb);
2018 
2019 	fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
2020 
2021 	if (sbi->sb->s_flags & SB_INLINECRYPT)
2022 		seq_puts(seq, ",inlinecrypt");
2023 
2024 	if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
2025 		seq_printf(seq, ",alloc_mode=%s", "default");
2026 	else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
2027 		seq_printf(seq, ",alloc_mode=%s", "reuse");
2028 
2029 	if (test_opt(sbi, DISABLE_CHECKPOINT))
2030 		seq_printf(seq, ",checkpoint=disable:%u",
2031 				F2FS_OPTION(sbi).unusable_cap);
2032 	if (test_opt(sbi, MERGE_CHECKPOINT))
2033 		seq_puts(seq, ",checkpoint_merge");
2034 	else
2035 		seq_puts(seq, ",nocheckpoint_merge");
2036 	if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
2037 		seq_printf(seq, ",fsync_mode=%s", "posix");
2038 	else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
2039 		seq_printf(seq, ",fsync_mode=%s", "strict");
2040 	else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
2041 		seq_printf(seq, ",fsync_mode=%s", "nobarrier");
2042 
2043 #ifdef CONFIG_F2FS_FS_COMPRESSION
2044 	f2fs_show_compress_options(seq, sbi->sb);
2045 #endif
2046 
2047 	if (test_opt(sbi, ATGC))
2048 		seq_puts(seq, ",atgc");
2049 
2050 	if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_NORMAL)
2051 		seq_printf(seq, ",memory=%s", "normal");
2052 	else if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW)
2053 		seq_printf(seq, ",memory=%s", "low");
2054 
2055 	return 0;
2056 }
2057 
2058 static void default_options(struct f2fs_sb_info *sbi)
2059 {
2060 	/* init some FS parameters */
2061 	if (f2fs_sb_has_readonly(sbi))
2062 		F2FS_OPTION(sbi).active_logs = NR_CURSEG_RO_TYPE;
2063 	else
2064 		F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
2065 
2066 	F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
2067 	if (le32_to_cpu(F2FS_RAW_SUPER(sbi)->segment_count_main) <=
2068 							SMALL_VOLUME_SEGMENTS)
2069 		F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
2070 	else
2071 		F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
2072 	F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
2073 	F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
2074 	F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
2075 	if (f2fs_sb_has_compression(sbi)) {
2076 		F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
2077 		F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
2078 		F2FS_OPTION(sbi).compress_ext_cnt = 0;
2079 		F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
2080 	}
2081 	F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
2082 	F2FS_OPTION(sbi).memory_mode = MEMORY_MODE_NORMAL;
2083 
2084 	sbi->sb->s_flags &= ~SB_INLINECRYPT;
2085 
2086 	set_opt(sbi, INLINE_XATTR);
2087 	set_opt(sbi, INLINE_DATA);
2088 	set_opt(sbi, INLINE_DENTRY);
2089 	set_opt(sbi, READ_EXTENT_CACHE);
2090 	set_opt(sbi, NOHEAP);
2091 	clear_opt(sbi, DISABLE_CHECKPOINT);
2092 	set_opt(sbi, MERGE_CHECKPOINT);
2093 	F2FS_OPTION(sbi).unusable_cap = 0;
2094 	sbi->sb->s_flags |= SB_LAZYTIME;
2095 	if (!f2fs_is_readonly(sbi))
2096 		set_opt(sbi, FLUSH_MERGE);
2097 	if (f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi))
2098 		set_opt(sbi, DISCARD);
2099 	if (f2fs_sb_has_blkzoned(sbi)) {
2100 		F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
2101 		F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_SECTION;
2102 	} else {
2103 		F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
2104 		F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_BLOCK;
2105 	}
2106 
2107 #ifdef CONFIG_F2FS_FS_XATTR
2108 	set_opt(sbi, XATTR_USER);
2109 #endif
2110 #ifdef CONFIG_F2FS_FS_POSIX_ACL
2111 	set_opt(sbi, POSIX_ACL);
2112 #endif
2113 
2114 	f2fs_build_fault_attr(sbi, 0, 0);
2115 }
2116 
2117 #ifdef CONFIG_QUOTA
2118 static int f2fs_enable_quotas(struct super_block *sb);
2119 #endif
2120 
2121 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
2122 {
2123 	unsigned int s_flags = sbi->sb->s_flags;
2124 	struct cp_control cpc;
2125 	unsigned int gc_mode = sbi->gc_mode;
2126 	int err = 0;
2127 	int ret;
2128 	block_t unusable;
2129 
2130 	if (s_flags & SB_RDONLY) {
2131 		f2fs_err(sbi, "checkpoint=disable on readonly fs");
2132 		return -EINVAL;
2133 	}
2134 	sbi->sb->s_flags |= SB_ACTIVE;
2135 
2136 	/* check if we need more GC first */
2137 	unusable = f2fs_get_unusable_blocks(sbi);
2138 	if (!f2fs_disable_cp_again(sbi, unusable))
2139 		goto skip_gc;
2140 
2141 	f2fs_update_time(sbi, DISABLE_TIME);
2142 
2143 	sbi->gc_mode = GC_URGENT_HIGH;
2144 
2145 	while (!f2fs_time_over(sbi, DISABLE_TIME)) {
2146 		struct f2fs_gc_control gc_control = {
2147 			.victim_segno = NULL_SEGNO,
2148 			.init_gc_type = FG_GC,
2149 			.should_migrate_blocks = false,
2150 			.err_gc_skipped = true,
2151 			.nr_free_secs = 1 };
2152 
2153 		f2fs_down_write(&sbi->gc_lock);
2154 		err = f2fs_gc(sbi, &gc_control);
2155 		if (err == -ENODATA) {
2156 			err = 0;
2157 			break;
2158 		}
2159 		if (err && err != -EAGAIN)
2160 			break;
2161 	}
2162 
2163 	ret = sync_filesystem(sbi->sb);
2164 	if (ret || err) {
2165 		err = ret ? ret : err;
2166 		goto restore_flag;
2167 	}
2168 
2169 	unusable = f2fs_get_unusable_blocks(sbi);
2170 	if (f2fs_disable_cp_again(sbi, unusable)) {
2171 		err = -EAGAIN;
2172 		goto restore_flag;
2173 	}
2174 
2175 skip_gc:
2176 	f2fs_down_write(&sbi->gc_lock);
2177 	cpc.reason = CP_PAUSE;
2178 	set_sbi_flag(sbi, SBI_CP_DISABLED);
2179 	err = f2fs_write_checkpoint(sbi, &cpc);
2180 	if (err)
2181 		goto out_unlock;
2182 
2183 	spin_lock(&sbi->stat_lock);
2184 	sbi->unusable_block_count = unusable;
2185 	spin_unlock(&sbi->stat_lock);
2186 
2187 out_unlock:
2188 	f2fs_up_write(&sbi->gc_lock);
2189 restore_flag:
2190 	sbi->gc_mode = gc_mode;
2191 	sbi->sb->s_flags = s_flags;	/* Restore SB_RDONLY status */
2192 	return err;
2193 }
2194 
2195 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
2196 {
2197 	int retry = DEFAULT_RETRY_IO_COUNT;
2198 
2199 	/* we should flush all the data to keep data consistency */
2200 	do {
2201 		sync_inodes_sb(sbi->sb);
2202 		f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2203 	} while (get_pages(sbi, F2FS_DIRTY_DATA) && retry--);
2204 
2205 	if (unlikely(retry < 0))
2206 		f2fs_warn(sbi, "checkpoint=enable has some unwritten data.");
2207 
2208 	f2fs_down_write(&sbi->gc_lock);
2209 	f2fs_dirty_to_prefree(sbi);
2210 
2211 	clear_sbi_flag(sbi, SBI_CP_DISABLED);
2212 	set_sbi_flag(sbi, SBI_IS_DIRTY);
2213 	f2fs_up_write(&sbi->gc_lock);
2214 
2215 	f2fs_sync_fs(sbi->sb, 1);
2216 
2217 	/* Let's ensure there's no pending checkpoint anymore */
2218 	f2fs_flush_ckpt_thread(sbi);
2219 }
2220 
2221 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
2222 {
2223 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2224 	struct f2fs_mount_info org_mount_opt;
2225 	unsigned long old_sb_flags;
2226 	int err;
2227 	bool need_restart_gc = false, need_stop_gc = false;
2228 	bool need_restart_ckpt = false, need_stop_ckpt = false;
2229 	bool need_restart_flush = false, need_stop_flush = false;
2230 	bool need_restart_discard = false, need_stop_discard = false;
2231 	bool no_read_extent_cache = !test_opt(sbi, READ_EXTENT_CACHE);
2232 	bool no_age_extent_cache = !test_opt(sbi, AGE_EXTENT_CACHE);
2233 	bool enable_checkpoint = !test_opt(sbi, DISABLE_CHECKPOINT);
2234 	bool no_io_align = !F2FS_IO_ALIGNED(sbi);
2235 	bool no_atgc = !test_opt(sbi, ATGC);
2236 	bool no_discard = !test_opt(sbi, DISCARD);
2237 	bool no_compress_cache = !test_opt(sbi, COMPRESS_CACHE);
2238 	bool block_unit_discard = f2fs_block_unit_discard(sbi);
2239 #ifdef CONFIG_QUOTA
2240 	int i, j;
2241 #endif
2242 
2243 	/*
2244 	 * Save the old mount options in case we
2245 	 * need to restore them.
2246 	 */
2247 	org_mount_opt = sbi->mount_opt;
2248 	old_sb_flags = sb->s_flags;
2249 
2250 #ifdef CONFIG_QUOTA
2251 	org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
2252 	for (i = 0; i < MAXQUOTAS; i++) {
2253 		if (F2FS_OPTION(sbi).s_qf_names[i]) {
2254 			org_mount_opt.s_qf_names[i] =
2255 				kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
2256 				GFP_KERNEL);
2257 			if (!org_mount_opt.s_qf_names[i]) {
2258 				for (j = 0; j < i; j++)
2259 					kfree(org_mount_opt.s_qf_names[j]);
2260 				return -ENOMEM;
2261 			}
2262 		} else {
2263 			org_mount_opt.s_qf_names[i] = NULL;
2264 		}
2265 	}
2266 #endif
2267 
2268 	/* recover superblocks we couldn't write due to previous RO mount */
2269 	if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
2270 		err = f2fs_commit_super(sbi, false);
2271 		f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
2272 			  err);
2273 		if (!err)
2274 			clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2275 	}
2276 
2277 	default_options(sbi);
2278 
2279 	/* parse mount options */
2280 	err = parse_options(sb, data, true);
2281 	if (err)
2282 		goto restore_opts;
2283 
2284 	/*
2285 	 * Previous and new state of filesystem is RO,
2286 	 * so skip checking GC and FLUSH_MERGE conditions.
2287 	 */
2288 	if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
2289 		goto skip;
2290 
2291 	if (f2fs_dev_is_readonly(sbi) && !(*flags & SB_RDONLY)) {
2292 		err = -EROFS;
2293 		goto restore_opts;
2294 	}
2295 
2296 #ifdef CONFIG_QUOTA
2297 	if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
2298 		err = dquot_suspend(sb, -1);
2299 		if (err < 0)
2300 			goto restore_opts;
2301 	} else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
2302 		/* dquot_resume needs RW */
2303 		sb->s_flags &= ~SB_RDONLY;
2304 		if (sb_any_quota_suspended(sb)) {
2305 			dquot_resume(sb, -1);
2306 		} else if (f2fs_sb_has_quota_ino(sbi)) {
2307 			err = f2fs_enable_quotas(sb);
2308 			if (err)
2309 				goto restore_opts;
2310 		}
2311 	}
2312 #endif
2313 	if (f2fs_lfs_mode(sbi) && !IS_F2FS_IPU_DISABLE(sbi)) {
2314 		err = -EINVAL;
2315 		f2fs_warn(sbi, "LFS is not compatible with IPU");
2316 		goto restore_opts;
2317 	}
2318 
2319 	/* disallow enable atgc dynamically */
2320 	if (no_atgc == !!test_opt(sbi, ATGC)) {
2321 		err = -EINVAL;
2322 		f2fs_warn(sbi, "switch atgc option is not allowed");
2323 		goto restore_opts;
2324 	}
2325 
2326 	/* disallow enable/disable extent_cache dynamically */
2327 	if (no_read_extent_cache == !!test_opt(sbi, READ_EXTENT_CACHE)) {
2328 		err = -EINVAL;
2329 		f2fs_warn(sbi, "switch extent_cache option is not allowed");
2330 		goto restore_opts;
2331 	}
2332 	/* disallow enable/disable age extent_cache dynamically */
2333 	if (no_age_extent_cache == !!test_opt(sbi, AGE_EXTENT_CACHE)) {
2334 		err = -EINVAL;
2335 		f2fs_warn(sbi, "switch age_extent_cache option is not allowed");
2336 		goto restore_opts;
2337 	}
2338 
2339 	if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
2340 		err = -EINVAL;
2341 		f2fs_warn(sbi, "switch io_bits option is not allowed");
2342 		goto restore_opts;
2343 	}
2344 
2345 	if (no_compress_cache == !!test_opt(sbi, COMPRESS_CACHE)) {
2346 		err = -EINVAL;
2347 		f2fs_warn(sbi, "switch compress_cache option is not allowed");
2348 		goto restore_opts;
2349 	}
2350 
2351 	if (block_unit_discard != f2fs_block_unit_discard(sbi)) {
2352 		err = -EINVAL;
2353 		f2fs_warn(sbi, "switch discard_unit option is not allowed");
2354 		goto restore_opts;
2355 	}
2356 
2357 	if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
2358 		err = -EINVAL;
2359 		f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
2360 		goto restore_opts;
2361 	}
2362 
2363 	/*
2364 	 * We stop the GC thread if FS is mounted as RO
2365 	 * or if background_gc = off is passed in mount
2366 	 * option. Also sync the filesystem.
2367 	 */
2368 	if ((*flags & SB_RDONLY) ||
2369 			(F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF &&
2370 			!test_opt(sbi, GC_MERGE))) {
2371 		if (sbi->gc_thread) {
2372 			f2fs_stop_gc_thread(sbi);
2373 			need_restart_gc = true;
2374 		}
2375 	} else if (!sbi->gc_thread) {
2376 		err = f2fs_start_gc_thread(sbi);
2377 		if (err)
2378 			goto restore_opts;
2379 		need_stop_gc = true;
2380 	}
2381 
2382 	if (*flags & SB_RDONLY) {
2383 		sync_inodes_sb(sb);
2384 
2385 		set_sbi_flag(sbi, SBI_IS_DIRTY);
2386 		set_sbi_flag(sbi, SBI_IS_CLOSE);
2387 		f2fs_sync_fs(sb, 1);
2388 		clear_sbi_flag(sbi, SBI_IS_CLOSE);
2389 	}
2390 
2391 	if ((*flags & SB_RDONLY) || test_opt(sbi, DISABLE_CHECKPOINT) ||
2392 			!test_opt(sbi, MERGE_CHECKPOINT)) {
2393 		f2fs_stop_ckpt_thread(sbi);
2394 		need_restart_ckpt = true;
2395 	} else {
2396 		/* Flush if the prevous checkpoint, if exists. */
2397 		f2fs_flush_ckpt_thread(sbi);
2398 
2399 		err = f2fs_start_ckpt_thread(sbi);
2400 		if (err) {
2401 			f2fs_err(sbi,
2402 			    "Failed to start F2FS issue_checkpoint_thread (%d)",
2403 			    err);
2404 			goto restore_gc;
2405 		}
2406 		need_stop_ckpt = true;
2407 	}
2408 
2409 	/*
2410 	 * We stop issue flush thread if FS is mounted as RO
2411 	 * or if flush_merge is not passed in mount option.
2412 	 */
2413 	if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2414 		clear_opt(sbi, FLUSH_MERGE);
2415 		f2fs_destroy_flush_cmd_control(sbi, false);
2416 		need_restart_flush = true;
2417 	} else {
2418 		err = f2fs_create_flush_cmd_control(sbi);
2419 		if (err)
2420 			goto restore_ckpt;
2421 		need_stop_flush = true;
2422 	}
2423 
2424 	if (no_discard == !!test_opt(sbi, DISCARD)) {
2425 		if (test_opt(sbi, DISCARD)) {
2426 			err = f2fs_start_discard_thread(sbi);
2427 			if (err)
2428 				goto restore_flush;
2429 			need_stop_discard = true;
2430 		} else {
2431 			f2fs_stop_discard_thread(sbi);
2432 			f2fs_issue_discard_timeout(sbi);
2433 			need_restart_discard = true;
2434 		}
2435 	}
2436 
2437 	if (enable_checkpoint == !!test_opt(sbi, DISABLE_CHECKPOINT)) {
2438 		if (test_opt(sbi, DISABLE_CHECKPOINT)) {
2439 			err = f2fs_disable_checkpoint(sbi);
2440 			if (err)
2441 				goto restore_discard;
2442 		} else {
2443 			f2fs_enable_checkpoint(sbi);
2444 		}
2445 	}
2446 
2447 skip:
2448 #ifdef CONFIG_QUOTA
2449 	/* Release old quota file names */
2450 	for (i = 0; i < MAXQUOTAS; i++)
2451 		kfree(org_mount_opt.s_qf_names[i]);
2452 #endif
2453 	/* Update the POSIXACL Flag */
2454 	sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
2455 		(test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
2456 
2457 	limit_reserve_root(sbi);
2458 	adjust_unusable_cap_perc(sbi);
2459 	*flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
2460 	return 0;
2461 restore_discard:
2462 	if (need_restart_discard) {
2463 		if (f2fs_start_discard_thread(sbi))
2464 			f2fs_warn(sbi, "discard has been stopped");
2465 	} else if (need_stop_discard) {
2466 		f2fs_stop_discard_thread(sbi);
2467 	}
2468 restore_flush:
2469 	if (need_restart_flush) {
2470 		if (f2fs_create_flush_cmd_control(sbi))
2471 			f2fs_warn(sbi, "background flush thread has stopped");
2472 	} else if (need_stop_flush) {
2473 		clear_opt(sbi, FLUSH_MERGE);
2474 		f2fs_destroy_flush_cmd_control(sbi, false);
2475 	}
2476 restore_ckpt:
2477 	if (need_restart_ckpt) {
2478 		if (f2fs_start_ckpt_thread(sbi))
2479 			f2fs_warn(sbi, "background ckpt thread has stopped");
2480 	} else if (need_stop_ckpt) {
2481 		f2fs_stop_ckpt_thread(sbi);
2482 	}
2483 restore_gc:
2484 	if (need_restart_gc) {
2485 		if (f2fs_start_gc_thread(sbi))
2486 			f2fs_warn(sbi, "background gc thread has stopped");
2487 	} else if (need_stop_gc) {
2488 		f2fs_stop_gc_thread(sbi);
2489 	}
2490 restore_opts:
2491 #ifdef CONFIG_QUOTA
2492 	F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
2493 	for (i = 0; i < MAXQUOTAS; i++) {
2494 		kfree(F2FS_OPTION(sbi).s_qf_names[i]);
2495 		F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
2496 	}
2497 #endif
2498 	sbi->mount_opt = org_mount_opt;
2499 	sb->s_flags = old_sb_flags;
2500 	return err;
2501 }
2502 
2503 #ifdef CONFIG_QUOTA
2504 static bool f2fs_need_recovery(struct f2fs_sb_info *sbi)
2505 {
2506 	/* need to recovery orphan */
2507 	if (is_set_ckpt_flags(sbi, CP_ORPHAN_PRESENT_FLAG))
2508 		return true;
2509 	/* need to recovery data */
2510 	if (test_opt(sbi, DISABLE_ROLL_FORWARD))
2511 		return false;
2512 	if (test_opt(sbi, NORECOVERY))
2513 		return false;
2514 	return !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG);
2515 }
2516 
2517 static bool f2fs_recover_quota_begin(struct f2fs_sb_info *sbi)
2518 {
2519 	bool readonly = f2fs_readonly(sbi->sb);
2520 
2521 	if (!f2fs_need_recovery(sbi))
2522 		return false;
2523 
2524 	/* it doesn't need to check f2fs_sb_has_readonly() */
2525 	if (f2fs_hw_is_readonly(sbi))
2526 		return false;
2527 
2528 	if (readonly) {
2529 		sbi->sb->s_flags &= ~SB_RDONLY;
2530 		set_sbi_flag(sbi, SBI_IS_WRITABLE);
2531 	}
2532 
2533 	/*
2534 	 * Turn on quotas which were not enabled for read-only mounts if
2535 	 * filesystem has quota feature, so that they are updated correctly.
2536 	 */
2537 	return f2fs_enable_quota_files(sbi, readonly);
2538 }
2539 
2540 static void f2fs_recover_quota_end(struct f2fs_sb_info *sbi,
2541 						bool quota_enabled)
2542 {
2543 	if (quota_enabled)
2544 		f2fs_quota_off_umount(sbi->sb);
2545 
2546 	if (is_sbi_flag_set(sbi, SBI_IS_WRITABLE)) {
2547 		clear_sbi_flag(sbi, SBI_IS_WRITABLE);
2548 		sbi->sb->s_flags |= SB_RDONLY;
2549 	}
2550 }
2551 
2552 /* Read data from quotafile */
2553 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
2554 			       size_t len, loff_t off)
2555 {
2556 	struct inode *inode = sb_dqopt(sb)->files[type];
2557 	struct address_space *mapping = inode->i_mapping;
2558 	block_t blkidx = F2FS_BYTES_TO_BLK(off);
2559 	int offset = off & (sb->s_blocksize - 1);
2560 	int tocopy;
2561 	size_t toread;
2562 	loff_t i_size = i_size_read(inode);
2563 	struct page *page;
2564 
2565 	if (off > i_size)
2566 		return 0;
2567 
2568 	if (off + len > i_size)
2569 		len = i_size - off;
2570 	toread = len;
2571 	while (toread > 0) {
2572 		tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
2573 repeat:
2574 		page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
2575 		if (IS_ERR(page)) {
2576 			if (PTR_ERR(page) == -ENOMEM) {
2577 				memalloc_retry_wait(GFP_NOFS);
2578 				goto repeat;
2579 			}
2580 			set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2581 			return PTR_ERR(page);
2582 		}
2583 
2584 		lock_page(page);
2585 
2586 		if (unlikely(page->mapping != mapping)) {
2587 			f2fs_put_page(page, 1);
2588 			goto repeat;
2589 		}
2590 		if (unlikely(!PageUptodate(page))) {
2591 			f2fs_put_page(page, 1);
2592 			set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2593 			return -EIO;
2594 		}
2595 
2596 		memcpy_from_page(data, page, offset, tocopy);
2597 		f2fs_put_page(page, 1);
2598 
2599 		offset = 0;
2600 		toread -= tocopy;
2601 		data += tocopy;
2602 		blkidx++;
2603 	}
2604 	return len;
2605 }
2606 
2607 /* Write to quotafile */
2608 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2609 				const char *data, size_t len, loff_t off)
2610 {
2611 	struct inode *inode = sb_dqopt(sb)->files[type];
2612 	struct address_space *mapping = inode->i_mapping;
2613 	const struct address_space_operations *a_ops = mapping->a_ops;
2614 	int offset = off & (sb->s_blocksize - 1);
2615 	size_t towrite = len;
2616 	struct page *page;
2617 	void *fsdata = NULL;
2618 	int err = 0;
2619 	int tocopy;
2620 
2621 	while (towrite > 0) {
2622 		tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2623 								towrite);
2624 retry:
2625 		err = a_ops->write_begin(NULL, mapping, off, tocopy,
2626 							&page, &fsdata);
2627 		if (unlikely(err)) {
2628 			if (err == -ENOMEM) {
2629 				f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2630 				goto retry;
2631 			}
2632 			set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2633 			break;
2634 		}
2635 
2636 		memcpy_to_page(page, offset, data, tocopy);
2637 
2638 		a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2639 						page, fsdata);
2640 		offset = 0;
2641 		towrite -= tocopy;
2642 		off += tocopy;
2643 		data += tocopy;
2644 		cond_resched();
2645 	}
2646 
2647 	if (len == towrite)
2648 		return err;
2649 	inode->i_mtime = inode->i_ctime = current_time(inode);
2650 	f2fs_mark_inode_dirty_sync(inode, false);
2651 	return len - towrite;
2652 }
2653 
2654 int f2fs_dquot_initialize(struct inode *inode)
2655 {
2656 	if (time_to_inject(F2FS_I_SB(inode), FAULT_DQUOT_INIT))
2657 		return -ESRCH;
2658 
2659 	return dquot_initialize(inode);
2660 }
2661 
2662 static struct dquot **f2fs_get_dquots(struct inode *inode)
2663 {
2664 	return F2FS_I(inode)->i_dquot;
2665 }
2666 
2667 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2668 {
2669 	return &F2FS_I(inode)->i_reserved_quota;
2670 }
2671 
2672 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2673 {
2674 	if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2675 		f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2676 		return 0;
2677 	}
2678 
2679 	return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2680 					F2FS_OPTION(sbi).s_jquota_fmt, type);
2681 }
2682 
2683 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2684 {
2685 	int enabled = 0;
2686 	int i, err;
2687 
2688 	if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2689 		err = f2fs_enable_quotas(sbi->sb);
2690 		if (err) {
2691 			f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2692 			return 0;
2693 		}
2694 		return 1;
2695 	}
2696 
2697 	for (i = 0; i < MAXQUOTAS; i++) {
2698 		if (F2FS_OPTION(sbi).s_qf_names[i]) {
2699 			err = f2fs_quota_on_mount(sbi, i);
2700 			if (!err) {
2701 				enabled = 1;
2702 				continue;
2703 			}
2704 			f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2705 				 err, i);
2706 		}
2707 	}
2708 	return enabled;
2709 }
2710 
2711 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2712 			     unsigned int flags)
2713 {
2714 	struct inode *qf_inode;
2715 	unsigned long qf_inum;
2716 	int err;
2717 
2718 	BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2719 
2720 	qf_inum = f2fs_qf_ino(sb, type);
2721 	if (!qf_inum)
2722 		return -EPERM;
2723 
2724 	qf_inode = f2fs_iget(sb, qf_inum);
2725 	if (IS_ERR(qf_inode)) {
2726 		f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2727 		return PTR_ERR(qf_inode);
2728 	}
2729 
2730 	/* Don't account quota for quota files to avoid recursion */
2731 	qf_inode->i_flags |= S_NOQUOTA;
2732 	err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2733 	iput(qf_inode);
2734 	return err;
2735 }
2736 
2737 static int f2fs_enable_quotas(struct super_block *sb)
2738 {
2739 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2740 	int type, err = 0;
2741 	unsigned long qf_inum;
2742 	bool quota_mopt[MAXQUOTAS] = {
2743 		test_opt(sbi, USRQUOTA),
2744 		test_opt(sbi, GRPQUOTA),
2745 		test_opt(sbi, PRJQUOTA),
2746 	};
2747 
2748 	if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2749 		f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2750 		return 0;
2751 	}
2752 
2753 	sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2754 
2755 	for (type = 0; type < MAXQUOTAS; type++) {
2756 		qf_inum = f2fs_qf_ino(sb, type);
2757 		if (qf_inum) {
2758 			err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2759 				DQUOT_USAGE_ENABLED |
2760 				(quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2761 			if (err) {
2762 				f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2763 					 type, err);
2764 				for (type--; type >= 0; type--)
2765 					dquot_quota_off(sb, type);
2766 				set_sbi_flag(F2FS_SB(sb),
2767 						SBI_QUOTA_NEED_REPAIR);
2768 				return err;
2769 			}
2770 		}
2771 	}
2772 	return 0;
2773 }
2774 
2775 static int f2fs_quota_sync_file(struct f2fs_sb_info *sbi, int type)
2776 {
2777 	struct quota_info *dqopt = sb_dqopt(sbi->sb);
2778 	struct address_space *mapping = dqopt->files[type]->i_mapping;
2779 	int ret = 0;
2780 
2781 	ret = dquot_writeback_dquots(sbi->sb, type);
2782 	if (ret)
2783 		goto out;
2784 
2785 	ret = filemap_fdatawrite(mapping);
2786 	if (ret)
2787 		goto out;
2788 
2789 	/* if we are using journalled quota */
2790 	if (is_journalled_quota(sbi))
2791 		goto out;
2792 
2793 	ret = filemap_fdatawait(mapping);
2794 
2795 	truncate_inode_pages(&dqopt->files[type]->i_data, 0);
2796 out:
2797 	if (ret)
2798 		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2799 	return ret;
2800 }
2801 
2802 int f2fs_quota_sync(struct super_block *sb, int type)
2803 {
2804 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2805 	struct quota_info *dqopt = sb_dqopt(sb);
2806 	int cnt;
2807 	int ret = 0;
2808 
2809 	/*
2810 	 * Now when everything is written we can discard the pagecache so
2811 	 * that userspace sees the changes.
2812 	 */
2813 	for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2814 
2815 		if (type != -1 && cnt != type)
2816 			continue;
2817 
2818 		if (!sb_has_quota_active(sb, cnt))
2819 			continue;
2820 
2821 		if (!f2fs_sb_has_quota_ino(sbi))
2822 			inode_lock(dqopt->files[cnt]);
2823 
2824 		/*
2825 		 * do_quotactl
2826 		 *  f2fs_quota_sync
2827 		 *  f2fs_down_read(quota_sem)
2828 		 *  dquot_writeback_dquots()
2829 		 *  f2fs_dquot_commit
2830 		 *			      block_operation
2831 		 *			      f2fs_down_read(quota_sem)
2832 		 */
2833 		f2fs_lock_op(sbi);
2834 		f2fs_down_read(&sbi->quota_sem);
2835 
2836 		ret = f2fs_quota_sync_file(sbi, cnt);
2837 
2838 		f2fs_up_read(&sbi->quota_sem);
2839 		f2fs_unlock_op(sbi);
2840 
2841 		if (!f2fs_sb_has_quota_ino(sbi))
2842 			inode_unlock(dqopt->files[cnt]);
2843 
2844 		if (ret)
2845 			break;
2846 	}
2847 	return ret;
2848 }
2849 
2850 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2851 							const struct path *path)
2852 {
2853 	struct inode *inode;
2854 	int err;
2855 
2856 	/* if quota sysfile exists, deny enabling quota with specific file */
2857 	if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2858 		f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2859 		return -EBUSY;
2860 	}
2861 
2862 	err = f2fs_quota_sync(sb, type);
2863 	if (err)
2864 		return err;
2865 
2866 	err = dquot_quota_on(sb, type, format_id, path);
2867 	if (err)
2868 		return err;
2869 
2870 	inode = d_inode(path->dentry);
2871 
2872 	inode_lock(inode);
2873 	F2FS_I(inode)->i_flags |= F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL;
2874 	f2fs_set_inode_flags(inode);
2875 	inode_unlock(inode);
2876 	f2fs_mark_inode_dirty_sync(inode, false);
2877 
2878 	return 0;
2879 }
2880 
2881 static int __f2fs_quota_off(struct super_block *sb, int type)
2882 {
2883 	struct inode *inode = sb_dqopt(sb)->files[type];
2884 	int err;
2885 
2886 	if (!inode || !igrab(inode))
2887 		return dquot_quota_off(sb, type);
2888 
2889 	err = f2fs_quota_sync(sb, type);
2890 	if (err)
2891 		goto out_put;
2892 
2893 	err = dquot_quota_off(sb, type);
2894 	if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
2895 		goto out_put;
2896 
2897 	inode_lock(inode);
2898 	F2FS_I(inode)->i_flags &= ~(F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL);
2899 	f2fs_set_inode_flags(inode);
2900 	inode_unlock(inode);
2901 	f2fs_mark_inode_dirty_sync(inode, false);
2902 out_put:
2903 	iput(inode);
2904 	return err;
2905 }
2906 
2907 static int f2fs_quota_off(struct super_block *sb, int type)
2908 {
2909 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2910 	int err;
2911 
2912 	err = __f2fs_quota_off(sb, type);
2913 
2914 	/*
2915 	 * quotactl can shutdown journalled quota, result in inconsistence
2916 	 * between quota record and fs data by following updates, tag the
2917 	 * flag to let fsck be aware of it.
2918 	 */
2919 	if (is_journalled_quota(sbi))
2920 		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2921 	return err;
2922 }
2923 
2924 void f2fs_quota_off_umount(struct super_block *sb)
2925 {
2926 	int type;
2927 	int err;
2928 
2929 	for (type = 0; type < MAXQUOTAS; type++) {
2930 		err = __f2fs_quota_off(sb, type);
2931 		if (err) {
2932 			int ret = dquot_quota_off(sb, type);
2933 
2934 			f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
2935 				 type, err, ret);
2936 			set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2937 		}
2938 	}
2939 	/*
2940 	 * In case of checkpoint=disable, we must flush quota blocks.
2941 	 * This can cause NULL exception for node_inode in end_io, since
2942 	 * put_super already dropped it.
2943 	 */
2944 	sync_filesystem(sb);
2945 }
2946 
2947 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
2948 {
2949 	struct quota_info *dqopt = sb_dqopt(sb);
2950 	int type;
2951 
2952 	for (type = 0; type < MAXQUOTAS; type++) {
2953 		if (!dqopt->files[type])
2954 			continue;
2955 		f2fs_inode_synced(dqopt->files[type]);
2956 	}
2957 }
2958 
2959 static int f2fs_dquot_commit(struct dquot *dquot)
2960 {
2961 	struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2962 	int ret;
2963 
2964 	f2fs_down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
2965 	ret = dquot_commit(dquot);
2966 	if (ret < 0)
2967 		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2968 	f2fs_up_read(&sbi->quota_sem);
2969 	return ret;
2970 }
2971 
2972 static int f2fs_dquot_acquire(struct dquot *dquot)
2973 {
2974 	struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2975 	int ret;
2976 
2977 	f2fs_down_read(&sbi->quota_sem);
2978 	ret = dquot_acquire(dquot);
2979 	if (ret < 0)
2980 		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2981 	f2fs_up_read(&sbi->quota_sem);
2982 	return ret;
2983 }
2984 
2985 static int f2fs_dquot_release(struct dquot *dquot)
2986 {
2987 	struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2988 	int ret = dquot_release(dquot);
2989 
2990 	if (ret < 0)
2991 		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2992 	return ret;
2993 }
2994 
2995 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
2996 {
2997 	struct super_block *sb = dquot->dq_sb;
2998 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2999 	int ret = dquot_mark_dquot_dirty(dquot);
3000 
3001 	/* if we are using journalled quota */
3002 	if (is_journalled_quota(sbi))
3003 		set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
3004 
3005 	return ret;
3006 }
3007 
3008 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
3009 {
3010 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
3011 	int ret = dquot_commit_info(sb, type);
3012 
3013 	if (ret < 0)
3014 		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3015 	return ret;
3016 }
3017 
3018 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
3019 {
3020 	*projid = F2FS_I(inode)->i_projid;
3021 	return 0;
3022 }
3023 
3024 static const struct dquot_operations f2fs_quota_operations = {
3025 	.get_reserved_space = f2fs_get_reserved_space,
3026 	.write_dquot	= f2fs_dquot_commit,
3027 	.acquire_dquot	= f2fs_dquot_acquire,
3028 	.release_dquot	= f2fs_dquot_release,
3029 	.mark_dirty	= f2fs_dquot_mark_dquot_dirty,
3030 	.write_info	= f2fs_dquot_commit_info,
3031 	.alloc_dquot	= dquot_alloc,
3032 	.destroy_dquot	= dquot_destroy,
3033 	.get_projid	= f2fs_get_projid,
3034 	.get_next_id	= dquot_get_next_id,
3035 };
3036 
3037 static const struct quotactl_ops f2fs_quotactl_ops = {
3038 	.quota_on	= f2fs_quota_on,
3039 	.quota_off	= f2fs_quota_off,
3040 	.quota_sync	= f2fs_quota_sync,
3041 	.get_state	= dquot_get_state,
3042 	.set_info	= dquot_set_dqinfo,
3043 	.get_dqblk	= dquot_get_dqblk,
3044 	.set_dqblk	= dquot_set_dqblk,
3045 	.get_nextdqblk	= dquot_get_next_dqblk,
3046 };
3047 #else
3048 int f2fs_dquot_initialize(struct inode *inode)
3049 {
3050 	return 0;
3051 }
3052 
3053 int f2fs_quota_sync(struct super_block *sb, int type)
3054 {
3055 	return 0;
3056 }
3057 
3058 void f2fs_quota_off_umount(struct super_block *sb)
3059 {
3060 }
3061 #endif
3062 
3063 static const struct super_operations f2fs_sops = {
3064 	.alloc_inode	= f2fs_alloc_inode,
3065 	.free_inode	= f2fs_free_inode,
3066 	.drop_inode	= f2fs_drop_inode,
3067 	.write_inode	= f2fs_write_inode,
3068 	.dirty_inode	= f2fs_dirty_inode,
3069 	.show_options	= f2fs_show_options,
3070 #ifdef CONFIG_QUOTA
3071 	.quota_read	= f2fs_quota_read,
3072 	.quota_write	= f2fs_quota_write,
3073 	.get_dquots	= f2fs_get_dquots,
3074 #endif
3075 	.evict_inode	= f2fs_evict_inode,
3076 	.put_super	= f2fs_put_super,
3077 	.sync_fs	= f2fs_sync_fs,
3078 	.freeze_fs	= f2fs_freeze,
3079 	.unfreeze_fs	= f2fs_unfreeze,
3080 	.statfs		= f2fs_statfs,
3081 	.remount_fs	= f2fs_remount,
3082 };
3083 
3084 #ifdef CONFIG_FS_ENCRYPTION
3085 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
3086 {
3087 	return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3088 				F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3089 				ctx, len, NULL);
3090 }
3091 
3092 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
3093 							void *fs_data)
3094 {
3095 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3096 
3097 	/*
3098 	 * Encrypting the root directory is not allowed because fsck
3099 	 * expects lost+found directory to exist and remain unencrypted
3100 	 * if LOST_FOUND feature is enabled.
3101 	 *
3102 	 */
3103 	if (f2fs_sb_has_lost_found(sbi) &&
3104 			inode->i_ino == F2FS_ROOT_INO(sbi))
3105 		return -EPERM;
3106 
3107 	return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3108 				F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3109 				ctx, len, fs_data, XATTR_CREATE);
3110 }
3111 
3112 static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
3113 {
3114 	return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
3115 }
3116 
3117 static bool f2fs_has_stable_inodes(struct super_block *sb)
3118 {
3119 	return true;
3120 }
3121 
3122 static void f2fs_get_ino_and_lblk_bits(struct super_block *sb,
3123 				       int *ino_bits_ret, int *lblk_bits_ret)
3124 {
3125 	*ino_bits_ret = 8 * sizeof(nid_t);
3126 	*lblk_bits_ret = 8 * sizeof(block_t);
3127 }
3128 
3129 static struct block_device **f2fs_get_devices(struct super_block *sb,
3130 					      unsigned int *num_devs)
3131 {
3132 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
3133 	struct block_device **devs;
3134 	int i;
3135 
3136 	if (!f2fs_is_multi_device(sbi))
3137 		return NULL;
3138 
3139 	devs = kmalloc_array(sbi->s_ndevs, sizeof(*devs), GFP_KERNEL);
3140 	if (!devs)
3141 		return ERR_PTR(-ENOMEM);
3142 
3143 	for (i = 0; i < sbi->s_ndevs; i++)
3144 		devs[i] = FDEV(i).bdev;
3145 	*num_devs = sbi->s_ndevs;
3146 	return devs;
3147 }
3148 
3149 static const struct fscrypt_operations f2fs_cryptops = {
3150 	.key_prefix		= "f2fs:",
3151 	.get_context		= f2fs_get_context,
3152 	.set_context		= f2fs_set_context,
3153 	.get_dummy_policy	= f2fs_get_dummy_policy,
3154 	.empty_dir		= f2fs_empty_dir,
3155 	.has_stable_inodes	= f2fs_has_stable_inodes,
3156 	.get_ino_and_lblk_bits	= f2fs_get_ino_and_lblk_bits,
3157 	.get_devices		= f2fs_get_devices,
3158 };
3159 #endif
3160 
3161 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
3162 		u64 ino, u32 generation)
3163 {
3164 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
3165 	struct inode *inode;
3166 
3167 	if (f2fs_check_nid_range(sbi, ino))
3168 		return ERR_PTR(-ESTALE);
3169 
3170 	/*
3171 	 * f2fs_iget isn't quite right if the inode is currently unallocated!
3172 	 * However f2fs_iget currently does appropriate checks to handle stale
3173 	 * inodes so everything is OK.
3174 	 */
3175 	inode = f2fs_iget(sb, ino);
3176 	if (IS_ERR(inode))
3177 		return ERR_CAST(inode);
3178 	if (unlikely(generation && inode->i_generation != generation)) {
3179 		/* we didn't find the right inode.. */
3180 		iput(inode);
3181 		return ERR_PTR(-ESTALE);
3182 	}
3183 	return inode;
3184 }
3185 
3186 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
3187 		int fh_len, int fh_type)
3188 {
3189 	return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
3190 				    f2fs_nfs_get_inode);
3191 }
3192 
3193 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
3194 		int fh_len, int fh_type)
3195 {
3196 	return generic_fh_to_parent(sb, fid, fh_len, fh_type,
3197 				    f2fs_nfs_get_inode);
3198 }
3199 
3200 static const struct export_operations f2fs_export_ops = {
3201 	.fh_to_dentry = f2fs_fh_to_dentry,
3202 	.fh_to_parent = f2fs_fh_to_parent,
3203 	.get_parent = f2fs_get_parent,
3204 };
3205 
3206 loff_t max_file_blocks(struct inode *inode)
3207 {
3208 	loff_t result = 0;
3209 	loff_t leaf_count;
3210 
3211 	/*
3212 	 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
3213 	 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
3214 	 * space in inode.i_addr, it will be more safe to reassign
3215 	 * result as zero.
3216 	 */
3217 
3218 	if (inode && f2fs_compressed_file(inode))
3219 		leaf_count = ADDRS_PER_BLOCK(inode);
3220 	else
3221 		leaf_count = DEF_ADDRS_PER_BLOCK;
3222 
3223 	/* two direct node blocks */
3224 	result += (leaf_count * 2);
3225 
3226 	/* two indirect node blocks */
3227 	leaf_count *= NIDS_PER_BLOCK;
3228 	result += (leaf_count * 2);
3229 
3230 	/* one double indirect node block */
3231 	leaf_count *= NIDS_PER_BLOCK;
3232 	result += leaf_count;
3233 
3234 	return result;
3235 }
3236 
3237 static int __f2fs_commit_super(struct buffer_head *bh,
3238 			struct f2fs_super_block *super)
3239 {
3240 	lock_buffer(bh);
3241 	if (super)
3242 		memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
3243 	set_buffer_dirty(bh);
3244 	unlock_buffer(bh);
3245 
3246 	/* it's rare case, we can do fua all the time */
3247 	return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
3248 }
3249 
3250 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
3251 					struct buffer_head *bh)
3252 {
3253 	struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3254 					(bh->b_data + F2FS_SUPER_OFFSET);
3255 	struct super_block *sb = sbi->sb;
3256 	u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
3257 	u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
3258 	u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
3259 	u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
3260 	u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
3261 	u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
3262 	u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
3263 	u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
3264 	u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
3265 	u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
3266 	u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3267 	u32 segment_count = le32_to_cpu(raw_super->segment_count);
3268 	u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3269 	u64 main_end_blkaddr = main_blkaddr +
3270 				(segment_count_main << log_blocks_per_seg);
3271 	u64 seg_end_blkaddr = segment0_blkaddr +
3272 				(segment_count << log_blocks_per_seg);
3273 
3274 	if (segment0_blkaddr != cp_blkaddr) {
3275 		f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
3276 			  segment0_blkaddr, cp_blkaddr);
3277 		return true;
3278 	}
3279 
3280 	if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
3281 							sit_blkaddr) {
3282 		f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
3283 			  cp_blkaddr, sit_blkaddr,
3284 			  segment_count_ckpt << log_blocks_per_seg);
3285 		return true;
3286 	}
3287 
3288 	if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
3289 							nat_blkaddr) {
3290 		f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
3291 			  sit_blkaddr, nat_blkaddr,
3292 			  segment_count_sit << log_blocks_per_seg);
3293 		return true;
3294 	}
3295 
3296 	if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
3297 							ssa_blkaddr) {
3298 		f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
3299 			  nat_blkaddr, ssa_blkaddr,
3300 			  segment_count_nat << log_blocks_per_seg);
3301 		return true;
3302 	}
3303 
3304 	if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
3305 							main_blkaddr) {
3306 		f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
3307 			  ssa_blkaddr, main_blkaddr,
3308 			  segment_count_ssa << log_blocks_per_seg);
3309 		return true;
3310 	}
3311 
3312 	if (main_end_blkaddr > seg_end_blkaddr) {
3313 		f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
3314 			  main_blkaddr, seg_end_blkaddr,
3315 			  segment_count_main << log_blocks_per_seg);
3316 		return true;
3317 	} else if (main_end_blkaddr < seg_end_blkaddr) {
3318 		int err = 0;
3319 		char *res;
3320 
3321 		/* fix in-memory information all the time */
3322 		raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
3323 				segment0_blkaddr) >> log_blocks_per_seg);
3324 
3325 		if (f2fs_readonly(sb) || f2fs_hw_is_readonly(sbi)) {
3326 			set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3327 			res = "internally";
3328 		} else {
3329 			err = __f2fs_commit_super(bh, NULL);
3330 			res = err ? "failed" : "done";
3331 		}
3332 		f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
3333 			  res, main_blkaddr, seg_end_blkaddr,
3334 			  segment_count_main << log_blocks_per_seg);
3335 		if (err)
3336 			return true;
3337 	}
3338 	return false;
3339 }
3340 
3341 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
3342 				struct buffer_head *bh)
3343 {
3344 	block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
3345 	block_t total_sections, blocks_per_seg;
3346 	struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3347 					(bh->b_data + F2FS_SUPER_OFFSET);
3348 	size_t crc_offset = 0;
3349 	__u32 crc = 0;
3350 
3351 	if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
3352 		f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
3353 			  F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
3354 		return -EINVAL;
3355 	}
3356 
3357 	/* Check checksum_offset and crc in superblock */
3358 	if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
3359 		crc_offset = le32_to_cpu(raw_super->checksum_offset);
3360 		if (crc_offset !=
3361 			offsetof(struct f2fs_super_block, crc)) {
3362 			f2fs_info(sbi, "Invalid SB checksum offset: %zu",
3363 				  crc_offset);
3364 			return -EFSCORRUPTED;
3365 		}
3366 		crc = le32_to_cpu(raw_super->crc);
3367 		if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
3368 			f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
3369 			return -EFSCORRUPTED;
3370 		}
3371 	}
3372 
3373 	/* Currently, support only 4KB block size */
3374 	if (le32_to_cpu(raw_super->log_blocksize) != F2FS_BLKSIZE_BITS) {
3375 		f2fs_info(sbi, "Invalid log_blocksize (%u), supports only %u",
3376 			  le32_to_cpu(raw_super->log_blocksize),
3377 			  F2FS_BLKSIZE_BITS);
3378 		return -EFSCORRUPTED;
3379 	}
3380 
3381 	/* check log blocks per segment */
3382 	if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
3383 		f2fs_info(sbi, "Invalid log blocks per segment (%u)",
3384 			  le32_to_cpu(raw_super->log_blocks_per_seg));
3385 		return -EFSCORRUPTED;
3386 	}
3387 
3388 	/* Currently, support 512/1024/2048/4096 bytes sector size */
3389 	if (le32_to_cpu(raw_super->log_sectorsize) >
3390 				F2FS_MAX_LOG_SECTOR_SIZE ||
3391 		le32_to_cpu(raw_super->log_sectorsize) <
3392 				F2FS_MIN_LOG_SECTOR_SIZE) {
3393 		f2fs_info(sbi, "Invalid log sectorsize (%u)",
3394 			  le32_to_cpu(raw_super->log_sectorsize));
3395 		return -EFSCORRUPTED;
3396 	}
3397 	if (le32_to_cpu(raw_super->log_sectors_per_block) +
3398 		le32_to_cpu(raw_super->log_sectorsize) !=
3399 			F2FS_MAX_LOG_SECTOR_SIZE) {
3400 		f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
3401 			  le32_to_cpu(raw_super->log_sectors_per_block),
3402 			  le32_to_cpu(raw_super->log_sectorsize));
3403 		return -EFSCORRUPTED;
3404 	}
3405 
3406 	segment_count = le32_to_cpu(raw_super->segment_count);
3407 	segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3408 	segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3409 	secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3410 	total_sections = le32_to_cpu(raw_super->section_count);
3411 
3412 	/* blocks_per_seg should be 512, given the above check */
3413 	blocks_per_seg = BIT(le32_to_cpu(raw_super->log_blocks_per_seg));
3414 
3415 	if (segment_count > F2FS_MAX_SEGMENT ||
3416 				segment_count < F2FS_MIN_SEGMENTS) {
3417 		f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
3418 		return -EFSCORRUPTED;
3419 	}
3420 
3421 	if (total_sections > segment_count_main || total_sections < 1 ||
3422 			segs_per_sec > segment_count || !segs_per_sec) {
3423 		f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
3424 			  segment_count, total_sections, segs_per_sec);
3425 		return -EFSCORRUPTED;
3426 	}
3427 
3428 	if (segment_count_main != total_sections * segs_per_sec) {
3429 		f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
3430 			  segment_count_main, total_sections, segs_per_sec);
3431 		return -EFSCORRUPTED;
3432 	}
3433 
3434 	if ((segment_count / segs_per_sec) < total_sections) {
3435 		f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
3436 			  segment_count, segs_per_sec, total_sections);
3437 		return -EFSCORRUPTED;
3438 	}
3439 
3440 	if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
3441 		f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
3442 			  segment_count, le64_to_cpu(raw_super->block_count));
3443 		return -EFSCORRUPTED;
3444 	}
3445 
3446 	if (RDEV(0).path[0]) {
3447 		block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
3448 		int i = 1;
3449 
3450 		while (i < MAX_DEVICES && RDEV(i).path[0]) {
3451 			dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
3452 			i++;
3453 		}
3454 		if (segment_count != dev_seg_count) {
3455 			f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
3456 					segment_count, dev_seg_count);
3457 			return -EFSCORRUPTED;
3458 		}
3459 	} else {
3460 		if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
3461 					!bdev_is_zoned(sbi->sb->s_bdev)) {
3462 			f2fs_info(sbi, "Zoned block device path is missing");
3463 			return -EFSCORRUPTED;
3464 		}
3465 	}
3466 
3467 	if (secs_per_zone > total_sections || !secs_per_zone) {
3468 		f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
3469 			  secs_per_zone, total_sections);
3470 		return -EFSCORRUPTED;
3471 	}
3472 	if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
3473 			raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
3474 			(le32_to_cpu(raw_super->extension_count) +
3475 			raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
3476 		f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
3477 			  le32_to_cpu(raw_super->extension_count),
3478 			  raw_super->hot_ext_count,
3479 			  F2FS_MAX_EXTENSION);
3480 		return -EFSCORRUPTED;
3481 	}
3482 
3483 	if (le32_to_cpu(raw_super->cp_payload) >=
3484 				(blocks_per_seg - F2FS_CP_PACKS -
3485 				NR_CURSEG_PERSIST_TYPE)) {
3486 		f2fs_info(sbi, "Insane cp_payload (%u >= %u)",
3487 			  le32_to_cpu(raw_super->cp_payload),
3488 			  blocks_per_seg - F2FS_CP_PACKS -
3489 			  NR_CURSEG_PERSIST_TYPE);
3490 		return -EFSCORRUPTED;
3491 	}
3492 
3493 	/* check reserved ino info */
3494 	if (le32_to_cpu(raw_super->node_ino) != 1 ||
3495 		le32_to_cpu(raw_super->meta_ino) != 2 ||
3496 		le32_to_cpu(raw_super->root_ino) != 3) {
3497 		f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
3498 			  le32_to_cpu(raw_super->node_ino),
3499 			  le32_to_cpu(raw_super->meta_ino),
3500 			  le32_to_cpu(raw_super->root_ino));
3501 		return -EFSCORRUPTED;
3502 	}
3503 
3504 	/* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
3505 	if (sanity_check_area_boundary(sbi, bh))
3506 		return -EFSCORRUPTED;
3507 
3508 	return 0;
3509 }
3510 
3511 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
3512 {
3513 	unsigned int total, fsmeta;
3514 	struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3515 	struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
3516 	unsigned int ovp_segments, reserved_segments;
3517 	unsigned int main_segs, blocks_per_seg;
3518 	unsigned int sit_segs, nat_segs;
3519 	unsigned int sit_bitmap_size, nat_bitmap_size;
3520 	unsigned int log_blocks_per_seg;
3521 	unsigned int segment_count_main;
3522 	unsigned int cp_pack_start_sum, cp_payload;
3523 	block_t user_block_count, valid_user_blocks;
3524 	block_t avail_node_count, valid_node_count;
3525 	unsigned int nat_blocks, nat_bits_bytes, nat_bits_blocks;
3526 	int i, j;
3527 
3528 	total = le32_to_cpu(raw_super->segment_count);
3529 	fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
3530 	sit_segs = le32_to_cpu(raw_super->segment_count_sit);
3531 	fsmeta += sit_segs;
3532 	nat_segs = le32_to_cpu(raw_super->segment_count_nat);
3533 	fsmeta += nat_segs;
3534 	fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
3535 	fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
3536 
3537 	if (unlikely(fsmeta >= total))
3538 		return 1;
3539 
3540 	ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
3541 	reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
3542 
3543 	if (!f2fs_sb_has_readonly(sbi) &&
3544 			unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
3545 			ovp_segments == 0 || reserved_segments == 0)) {
3546 		f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
3547 		return 1;
3548 	}
3549 	user_block_count = le64_to_cpu(ckpt->user_block_count);
3550 	segment_count_main = le32_to_cpu(raw_super->segment_count_main) +
3551 			(f2fs_sb_has_readonly(sbi) ? 1 : 0);
3552 	log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3553 	if (!user_block_count || user_block_count >=
3554 			segment_count_main << log_blocks_per_seg) {
3555 		f2fs_err(sbi, "Wrong user_block_count: %u",
3556 			 user_block_count);
3557 		return 1;
3558 	}
3559 
3560 	valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
3561 	if (valid_user_blocks > user_block_count) {
3562 		f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
3563 			 valid_user_blocks, user_block_count);
3564 		return 1;
3565 	}
3566 
3567 	valid_node_count = le32_to_cpu(ckpt->valid_node_count);
3568 	avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
3569 	if (valid_node_count > avail_node_count) {
3570 		f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
3571 			 valid_node_count, avail_node_count);
3572 		return 1;
3573 	}
3574 
3575 	main_segs = le32_to_cpu(raw_super->segment_count_main);
3576 	blocks_per_seg = sbi->blocks_per_seg;
3577 
3578 	for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3579 		if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
3580 			le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
3581 			return 1;
3582 
3583 		if (f2fs_sb_has_readonly(sbi))
3584 			goto check_data;
3585 
3586 		for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
3587 			if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3588 				le32_to_cpu(ckpt->cur_node_segno[j])) {
3589 				f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
3590 					 i, j,
3591 					 le32_to_cpu(ckpt->cur_node_segno[i]));
3592 				return 1;
3593 			}
3594 		}
3595 	}
3596 check_data:
3597 	for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
3598 		if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
3599 			le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
3600 			return 1;
3601 
3602 		if (f2fs_sb_has_readonly(sbi))
3603 			goto skip_cross;
3604 
3605 		for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3606 			if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3607 				le32_to_cpu(ckpt->cur_data_segno[j])) {
3608 				f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3609 					 i, j,
3610 					 le32_to_cpu(ckpt->cur_data_segno[i]));
3611 				return 1;
3612 			}
3613 		}
3614 	}
3615 	for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3616 		for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3617 			if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3618 				le32_to_cpu(ckpt->cur_data_segno[j])) {
3619 				f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3620 					 i, j,
3621 					 le32_to_cpu(ckpt->cur_node_segno[i]));
3622 				return 1;
3623 			}
3624 		}
3625 	}
3626 skip_cross:
3627 	sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3628 	nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3629 
3630 	if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3631 		nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3632 		f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3633 			 sit_bitmap_size, nat_bitmap_size);
3634 		return 1;
3635 	}
3636 
3637 	cp_pack_start_sum = __start_sum_addr(sbi);
3638 	cp_payload = __cp_payload(sbi);
3639 	if (cp_pack_start_sum < cp_payload + 1 ||
3640 		cp_pack_start_sum > blocks_per_seg - 1 -
3641 			NR_CURSEG_PERSIST_TYPE) {
3642 		f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3643 			 cp_pack_start_sum);
3644 		return 1;
3645 	}
3646 
3647 	if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3648 		le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3649 		f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3650 			  "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3651 			  "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3652 			  le32_to_cpu(ckpt->checksum_offset));
3653 		return 1;
3654 	}
3655 
3656 	nat_blocks = nat_segs << log_blocks_per_seg;
3657 	nat_bits_bytes = nat_blocks / BITS_PER_BYTE;
3658 	nat_bits_blocks = F2FS_BLK_ALIGN((nat_bits_bytes << 1) + 8);
3659 	if (__is_set_ckpt_flags(ckpt, CP_NAT_BITS_FLAG) &&
3660 		(cp_payload + F2FS_CP_PACKS +
3661 		NR_CURSEG_PERSIST_TYPE + nat_bits_blocks >= blocks_per_seg)) {
3662 		f2fs_warn(sbi, "Insane cp_payload: %u, nat_bits_blocks: %u)",
3663 			  cp_payload, nat_bits_blocks);
3664 		return 1;
3665 	}
3666 
3667 	if (unlikely(f2fs_cp_error(sbi))) {
3668 		f2fs_err(sbi, "A bug case: need to run fsck");
3669 		return 1;
3670 	}
3671 	return 0;
3672 }
3673 
3674 static void init_sb_info(struct f2fs_sb_info *sbi)
3675 {
3676 	struct f2fs_super_block *raw_super = sbi->raw_super;
3677 	int i;
3678 
3679 	sbi->log_sectors_per_block =
3680 		le32_to_cpu(raw_super->log_sectors_per_block);
3681 	sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3682 	sbi->blocksize = BIT(sbi->log_blocksize);
3683 	sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3684 	sbi->blocks_per_seg = BIT(sbi->log_blocks_per_seg);
3685 	sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3686 	sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3687 	sbi->total_sections = le32_to_cpu(raw_super->section_count);
3688 	sbi->total_node_count =
3689 		(le32_to_cpu(raw_super->segment_count_nat) / 2)
3690 			* sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
3691 	F2FS_ROOT_INO(sbi) = le32_to_cpu(raw_super->root_ino);
3692 	F2FS_NODE_INO(sbi) = le32_to_cpu(raw_super->node_ino);
3693 	F2FS_META_INO(sbi) = le32_to_cpu(raw_super->meta_ino);
3694 	sbi->cur_victim_sec = NULL_SECNO;
3695 	sbi->gc_mode = GC_NORMAL;
3696 	sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3697 	sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3698 	sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3699 	sbi->migration_granularity = sbi->segs_per_sec;
3700 	sbi->seq_file_ra_mul = MIN_RA_MUL;
3701 	sbi->max_fragment_chunk = DEF_FRAGMENT_SIZE;
3702 	sbi->max_fragment_hole = DEF_FRAGMENT_SIZE;
3703 	spin_lock_init(&sbi->gc_remaining_trials_lock);
3704 	atomic64_set(&sbi->current_atomic_write, 0);
3705 
3706 	sbi->dir_level = DEF_DIR_LEVEL;
3707 	sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3708 	sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3709 	sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3710 	sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3711 	sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3712 	sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3713 				DEF_UMOUNT_DISCARD_TIMEOUT;
3714 	clear_sbi_flag(sbi, SBI_NEED_FSCK);
3715 
3716 	for (i = 0; i < NR_COUNT_TYPE; i++)
3717 		atomic_set(&sbi->nr_pages[i], 0);
3718 
3719 	for (i = 0; i < META; i++)
3720 		atomic_set(&sbi->wb_sync_req[i], 0);
3721 
3722 	INIT_LIST_HEAD(&sbi->s_list);
3723 	mutex_init(&sbi->umount_mutex);
3724 	init_f2fs_rwsem(&sbi->io_order_lock);
3725 	spin_lock_init(&sbi->cp_lock);
3726 
3727 	sbi->dirty_device = 0;
3728 	spin_lock_init(&sbi->dev_lock);
3729 
3730 	init_f2fs_rwsem(&sbi->sb_lock);
3731 	init_f2fs_rwsem(&sbi->pin_sem);
3732 }
3733 
3734 static int init_percpu_info(struct f2fs_sb_info *sbi)
3735 {
3736 	int err;
3737 
3738 	err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3739 	if (err)
3740 		return err;
3741 
3742 	err = percpu_counter_init(&sbi->rf_node_block_count, 0, GFP_KERNEL);
3743 	if (err)
3744 		goto err_valid_block;
3745 
3746 	err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3747 								GFP_KERNEL);
3748 	if (err)
3749 		goto err_node_block;
3750 	return 0;
3751 
3752 err_node_block:
3753 	percpu_counter_destroy(&sbi->rf_node_block_count);
3754 err_valid_block:
3755 	percpu_counter_destroy(&sbi->alloc_valid_block_count);
3756 	return err;
3757 }
3758 
3759 #ifdef CONFIG_BLK_DEV_ZONED
3760 
3761 struct f2fs_report_zones_args {
3762 	struct f2fs_sb_info *sbi;
3763 	struct f2fs_dev_info *dev;
3764 };
3765 
3766 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3767 			      void *data)
3768 {
3769 	struct f2fs_report_zones_args *rz_args = data;
3770 	block_t unusable_blocks = (zone->len - zone->capacity) >>
3771 					F2FS_LOG_SECTORS_PER_BLOCK;
3772 
3773 	if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3774 		return 0;
3775 
3776 	set_bit(idx, rz_args->dev->blkz_seq);
3777 	if (!rz_args->sbi->unusable_blocks_per_sec) {
3778 		rz_args->sbi->unusable_blocks_per_sec = unusable_blocks;
3779 		return 0;
3780 	}
3781 	if (rz_args->sbi->unusable_blocks_per_sec != unusable_blocks) {
3782 		f2fs_err(rz_args->sbi, "F2FS supports single zone capacity\n");
3783 		return -EINVAL;
3784 	}
3785 	return 0;
3786 }
3787 
3788 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3789 {
3790 	struct block_device *bdev = FDEV(devi).bdev;
3791 	sector_t nr_sectors = bdev_nr_sectors(bdev);
3792 	struct f2fs_report_zones_args rep_zone_arg;
3793 	u64 zone_sectors;
3794 	int ret;
3795 
3796 	if (!f2fs_sb_has_blkzoned(sbi))
3797 		return 0;
3798 
3799 	zone_sectors = bdev_zone_sectors(bdev);
3800 	if (!is_power_of_2(zone_sectors)) {
3801 		f2fs_err(sbi, "F2FS does not support non power of 2 zone sizes\n");
3802 		return -EINVAL;
3803 	}
3804 
3805 	if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3806 				SECTOR_TO_BLOCK(zone_sectors))
3807 		return -EINVAL;
3808 	sbi->blocks_per_blkz = SECTOR_TO_BLOCK(zone_sectors);
3809 	FDEV(devi).nr_blkz = div_u64(SECTOR_TO_BLOCK(nr_sectors),
3810 					sbi->blocks_per_blkz);
3811 	if (nr_sectors & (zone_sectors - 1))
3812 		FDEV(devi).nr_blkz++;
3813 
3814 	FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3815 					BITS_TO_LONGS(FDEV(devi).nr_blkz)
3816 					* sizeof(unsigned long),
3817 					GFP_KERNEL);
3818 	if (!FDEV(devi).blkz_seq)
3819 		return -ENOMEM;
3820 
3821 	rep_zone_arg.sbi = sbi;
3822 	rep_zone_arg.dev = &FDEV(devi);
3823 
3824 	ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3825 				  &rep_zone_arg);
3826 	if (ret < 0)
3827 		return ret;
3828 	return 0;
3829 }
3830 #endif
3831 
3832 /*
3833  * Read f2fs raw super block.
3834  * Because we have two copies of super block, so read both of them
3835  * to get the first valid one. If any one of them is broken, we pass
3836  * them recovery flag back to the caller.
3837  */
3838 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3839 			struct f2fs_super_block **raw_super,
3840 			int *valid_super_block, int *recovery)
3841 {
3842 	struct super_block *sb = sbi->sb;
3843 	int block;
3844 	struct buffer_head *bh;
3845 	struct f2fs_super_block *super;
3846 	int err = 0;
3847 
3848 	super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3849 	if (!super)
3850 		return -ENOMEM;
3851 
3852 	for (block = 0; block < 2; block++) {
3853 		bh = sb_bread(sb, block);
3854 		if (!bh) {
3855 			f2fs_err(sbi, "Unable to read %dth superblock",
3856 				 block + 1);
3857 			err = -EIO;
3858 			*recovery = 1;
3859 			continue;
3860 		}
3861 
3862 		/* sanity checking of raw super */
3863 		err = sanity_check_raw_super(sbi, bh);
3864 		if (err) {
3865 			f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3866 				 block + 1);
3867 			brelse(bh);
3868 			*recovery = 1;
3869 			continue;
3870 		}
3871 
3872 		if (!*raw_super) {
3873 			memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3874 							sizeof(*super));
3875 			*valid_super_block = block;
3876 			*raw_super = super;
3877 		}
3878 		brelse(bh);
3879 	}
3880 
3881 	/* No valid superblock */
3882 	if (!*raw_super)
3883 		kfree(super);
3884 	else
3885 		err = 0;
3886 
3887 	return err;
3888 }
3889 
3890 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
3891 {
3892 	struct buffer_head *bh;
3893 	__u32 crc = 0;
3894 	int err;
3895 
3896 	if ((recover && f2fs_readonly(sbi->sb)) ||
3897 				f2fs_hw_is_readonly(sbi)) {
3898 		set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3899 		return -EROFS;
3900 	}
3901 
3902 	/* we should update superblock crc here */
3903 	if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
3904 		crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
3905 				offsetof(struct f2fs_super_block, crc));
3906 		F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
3907 	}
3908 
3909 	/* write back-up superblock first */
3910 	bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
3911 	if (!bh)
3912 		return -EIO;
3913 	err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3914 	brelse(bh);
3915 
3916 	/* if we are in recovery path, skip writing valid superblock */
3917 	if (recover || err)
3918 		return err;
3919 
3920 	/* write current valid superblock */
3921 	bh = sb_bread(sbi->sb, sbi->valid_super_block);
3922 	if (!bh)
3923 		return -EIO;
3924 	err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3925 	brelse(bh);
3926 	return err;
3927 }
3928 
3929 void f2fs_handle_stop(struct f2fs_sb_info *sbi, unsigned char reason)
3930 {
3931 	struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3932 	int err;
3933 
3934 	f2fs_down_write(&sbi->sb_lock);
3935 
3936 	if (raw_super->s_stop_reason[reason] < GENMASK(BITS_PER_BYTE - 1, 0))
3937 		raw_super->s_stop_reason[reason]++;
3938 
3939 	err = f2fs_commit_super(sbi, false);
3940 	if (err)
3941 		f2fs_err(sbi, "f2fs_commit_super fails to record reason:%u err:%d",
3942 								reason, err);
3943 	f2fs_up_write(&sbi->sb_lock);
3944 }
3945 
3946 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag)
3947 {
3948 	spin_lock(&sbi->error_lock);
3949 	if (!test_bit(flag, (unsigned long *)sbi->errors)) {
3950 		set_bit(flag, (unsigned long *)sbi->errors);
3951 		sbi->error_dirty = true;
3952 	}
3953 	spin_unlock(&sbi->error_lock);
3954 }
3955 
3956 static bool f2fs_update_errors(struct f2fs_sb_info *sbi)
3957 {
3958 	bool need_update = false;
3959 
3960 	spin_lock(&sbi->error_lock);
3961 	if (sbi->error_dirty) {
3962 		memcpy(F2FS_RAW_SUPER(sbi)->s_errors, sbi->errors,
3963 							MAX_F2FS_ERRORS);
3964 		sbi->error_dirty = false;
3965 		need_update = true;
3966 	}
3967 	spin_unlock(&sbi->error_lock);
3968 
3969 	return need_update;
3970 }
3971 
3972 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error)
3973 {
3974 	int err;
3975 
3976 	f2fs_save_errors(sbi, error);
3977 
3978 	f2fs_down_write(&sbi->sb_lock);
3979 
3980 	if (!f2fs_update_errors(sbi))
3981 		goto out_unlock;
3982 
3983 	err = f2fs_commit_super(sbi, false);
3984 	if (err)
3985 		f2fs_err(sbi, "f2fs_commit_super fails to record errors:%u, err:%d",
3986 								error, err);
3987 out_unlock:
3988 	f2fs_up_write(&sbi->sb_lock);
3989 }
3990 
3991 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
3992 {
3993 	struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3994 	unsigned int max_devices = MAX_DEVICES;
3995 	unsigned int logical_blksize;
3996 	int i;
3997 
3998 	/* Initialize single device information */
3999 	if (!RDEV(0).path[0]) {
4000 		if (!bdev_is_zoned(sbi->sb->s_bdev))
4001 			return 0;
4002 		max_devices = 1;
4003 	}
4004 
4005 	/*
4006 	 * Initialize multiple devices information, or single
4007 	 * zoned block device information.
4008 	 */
4009 	sbi->devs = f2fs_kzalloc(sbi,
4010 				 array_size(max_devices,
4011 					    sizeof(struct f2fs_dev_info)),
4012 				 GFP_KERNEL);
4013 	if (!sbi->devs)
4014 		return -ENOMEM;
4015 
4016 	logical_blksize = bdev_logical_block_size(sbi->sb->s_bdev);
4017 	sbi->aligned_blksize = true;
4018 
4019 	for (i = 0; i < max_devices; i++) {
4020 
4021 		if (i > 0 && !RDEV(i).path[0])
4022 			break;
4023 
4024 		if (max_devices == 1) {
4025 			/* Single zoned block device mount */
4026 			FDEV(0).bdev =
4027 				blkdev_get_by_dev(sbi->sb->s_bdev->bd_dev,
4028 					sbi->sb->s_mode, sbi->sb->s_type);
4029 		} else {
4030 			/* Multi-device mount */
4031 			memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
4032 			FDEV(i).total_segments =
4033 				le32_to_cpu(RDEV(i).total_segments);
4034 			if (i == 0) {
4035 				FDEV(i).start_blk = 0;
4036 				FDEV(i).end_blk = FDEV(i).start_blk +
4037 				    (FDEV(i).total_segments <<
4038 				    sbi->log_blocks_per_seg) - 1 +
4039 				    le32_to_cpu(raw_super->segment0_blkaddr);
4040 			} else {
4041 				FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
4042 				FDEV(i).end_blk = FDEV(i).start_blk +
4043 					(FDEV(i).total_segments <<
4044 					sbi->log_blocks_per_seg) - 1;
4045 			}
4046 			FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
4047 					sbi->sb->s_mode, sbi->sb->s_type);
4048 		}
4049 		if (IS_ERR(FDEV(i).bdev))
4050 			return PTR_ERR(FDEV(i).bdev);
4051 
4052 		/* to release errored devices */
4053 		sbi->s_ndevs = i + 1;
4054 
4055 		if (logical_blksize != bdev_logical_block_size(FDEV(i).bdev))
4056 			sbi->aligned_blksize = false;
4057 
4058 #ifdef CONFIG_BLK_DEV_ZONED
4059 		if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
4060 				!f2fs_sb_has_blkzoned(sbi)) {
4061 			f2fs_err(sbi, "Zoned block device feature not enabled");
4062 			return -EINVAL;
4063 		}
4064 		if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
4065 			if (init_blkz_info(sbi, i)) {
4066 				f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
4067 				return -EINVAL;
4068 			}
4069 			if (max_devices == 1)
4070 				break;
4071 			f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
4072 				  i, FDEV(i).path,
4073 				  FDEV(i).total_segments,
4074 				  FDEV(i).start_blk, FDEV(i).end_blk,
4075 				  bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
4076 				  "Host-aware" : "Host-managed");
4077 			continue;
4078 		}
4079 #endif
4080 		f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
4081 			  i, FDEV(i).path,
4082 			  FDEV(i).total_segments,
4083 			  FDEV(i).start_blk, FDEV(i).end_blk);
4084 	}
4085 	f2fs_info(sbi,
4086 		  "IO Block Size: %8ld KB", F2FS_IO_SIZE_KB(sbi));
4087 	return 0;
4088 }
4089 
4090 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
4091 {
4092 #if IS_ENABLED(CONFIG_UNICODE)
4093 	if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
4094 		const struct f2fs_sb_encodings *encoding_info;
4095 		struct unicode_map *encoding;
4096 		__u16 encoding_flags;
4097 
4098 		encoding_info = f2fs_sb_read_encoding(sbi->raw_super);
4099 		if (!encoding_info) {
4100 			f2fs_err(sbi,
4101 				 "Encoding requested by superblock is unknown");
4102 			return -EINVAL;
4103 		}
4104 
4105 		encoding_flags = le16_to_cpu(sbi->raw_super->s_encoding_flags);
4106 		encoding = utf8_load(encoding_info->version);
4107 		if (IS_ERR(encoding)) {
4108 			f2fs_err(sbi,
4109 				 "can't mount with superblock charset: %s-%u.%u.%u "
4110 				 "not supported by the kernel. flags: 0x%x.",
4111 				 encoding_info->name,
4112 				 unicode_major(encoding_info->version),
4113 				 unicode_minor(encoding_info->version),
4114 				 unicode_rev(encoding_info->version),
4115 				 encoding_flags);
4116 			return PTR_ERR(encoding);
4117 		}
4118 		f2fs_info(sbi, "Using encoding defined by superblock: "
4119 			 "%s-%u.%u.%u with flags 0x%hx", encoding_info->name,
4120 			 unicode_major(encoding_info->version),
4121 			 unicode_minor(encoding_info->version),
4122 			 unicode_rev(encoding_info->version),
4123 			 encoding_flags);
4124 
4125 		sbi->sb->s_encoding = encoding;
4126 		sbi->sb->s_encoding_flags = encoding_flags;
4127 	}
4128 #else
4129 	if (f2fs_sb_has_casefold(sbi)) {
4130 		f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
4131 		return -EINVAL;
4132 	}
4133 #endif
4134 	return 0;
4135 }
4136 
4137 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
4138 {
4139 	/* adjust parameters according to the volume size */
4140 	if (MAIN_SEGS(sbi) <= SMALL_VOLUME_SEGMENTS) {
4141 		if (f2fs_block_unit_discard(sbi))
4142 			SM_I(sbi)->dcc_info->discard_granularity =
4143 						MIN_DISCARD_GRANULARITY;
4144 		if (!f2fs_lfs_mode(sbi))
4145 			SM_I(sbi)->ipu_policy = BIT(F2FS_IPU_FORCE) |
4146 						BIT(F2FS_IPU_HONOR_OPU_WRITE);
4147 	}
4148 
4149 	sbi->readdir_ra = true;
4150 }
4151 
4152 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
4153 {
4154 	struct f2fs_sb_info *sbi;
4155 	struct f2fs_super_block *raw_super;
4156 	struct inode *root;
4157 	int err;
4158 	bool skip_recovery = false, need_fsck = false;
4159 	char *options = NULL;
4160 	int recovery, i, valid_super_block;
4161 	struct curseg_info *seg_i;
4162 	int retry_cnt = 1;
4163 #ifdef CONFIG_QUOTA
4164 	bool quota_enabled = false;
4165 #endif
4166 
4167 try_onemore:
4168 	err = -EINVAL;
4169 	raw_super = NULL;
4170 	valid_super_block = -1;
4171 	recovery = 0;
4172 
4173 	/* allocate memory for f2fs-specific super block info */
4174 	sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
4175 	if (!sbi)
4176 		return -ENOMEM;
4177 
4178 	sbi->sb = sb;
4179 
4180 	/* initialize locks within allocated memory */
4181 	init_f2fs_rwsem(&sbi->gc_lock);
4182 	mutex_init(&sbi->writepages);
4183 	init_f2fs_rwsem(&sbi->cp_global_sem);
4184 	init_f2fs_rwsem(&sbi->node_write);
4185 	init_f2fs_rwsem(&sbi->node_change);
4186 	spin_lock_init(&sbi->stat_lock);
4187 	init_f2fs_rwsem(&sbi->cp_rwsem);
4188 	init_f2fs_rwsem(&sbi->quota_sem);
4189 	init_waitqueue_head(&sbi->cp_wait);
4190 	spin_lock_init(&sbi->error_lock);
4191 
4192 	for (i = 0; i < NR_INODE_TYPE; i++) {
4193 		INIT_LIST_HEAD(&sbi->inode_list[i]);
4194 		spin_lock_init(&sbi->inode_lock[i]);
4195 	}
4196 	mutex_init(&sbi->flush_lock);
4197 
4198 	/* Load the checksum driver */
4199 	sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
4200 	if (IS_ERR(sbi->s_chksum_driver)) {
4201 		f2fs_err(sbi, "Cannot load crc32 driver.");
4202 		err = PTR_ERR(sbi->s_chksum_driver);
4203 		sbi->s_chksum_driver = NULL;
4204 		goto free_sbi;
4205 	}
4206 
4207 	/* set a block size */
4208 	if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
4209 		f2fs_err(sbi, "unable to set blocksize");
4210 		goto free_sbi;
4211 	}
4212 
4213 	err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
4214 								&recovery);
4215 	if (err)
4216 		goto free_sbi;
4217 
4218 	sb->s_fs_info = sbi;
4219 	sbi->raw_super = raw_super;
4220 
4221 	memcpy(sbi->errors, raw_super->s_errors, MAX_F2FS_ERRORS);
4222 
4223 	/* precompute checksum seed for metadata */
4224 	if (f2fs_sb_has_inode_chksum(sbi))
4225 		sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
4226 						sizeof(raw_super->uuid));
4227 
4228 	default_options(sbi);
4229 	/* parse mount options */
4230 	options = kstrdup((const char *)data, GFP_KERNEL);
4231 	if (data && !options) {
4232 		err = -ENOMEM;
4233 		goto free_sb_buf;
4234 	}
4235 
4236 	err = parse_options(sb, options, false);
4237 	if (err)
4238 		goto free_options;
4239 
4240 	sb->s_maxbytes = max_file_blocks(NULL) <<
4241 				le32_to_cpu(raw_super->log_blocksize);
4242 	sb->s_max_links = F2FS_LINK_MAX;
4243 
4244 	err = f2fs_setup_casefold(sbi);
4245 	if (err)
4246 		goto free_options;
4247 
4248 #ifdef CONFIG_QUOTA
4249 	sb->dq_op = &f2fs_quota_operations;
4250 	sb->s_qcop = &f2fs_quotactl_ops;
4251 	sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
4252 
4253 	if (f2fs_sb_has_quota_ino(sbi)) {
4254 		for (i = 0; i < MAXQUOTAS; i++) {
4255 			if (f2fs_qf_ino(sbi->sb, i))
4256 				sbi->nquota_files++;
4257 		}
4258 	}
4259 #endif
4260 
4261 	sb->s_op = &f2fs_sops;
4262 #ifdef CONFIG_FS_ENCRYPTION
4263 	sb->s_cop = &f2fs_cryptops;
4264 #endif
4265 #ifdef CONFIG_FS_VERITY
4266 	sb->s_vop = &f2fs_verityops;
4267 #endif
4268 	sb->s_xattr = f2fs_xattr_handlers;
4269 	sb->s_export_op = &f2fs_export_ops;
4270 	sb->s_magic = F2FS_SUPER_MAGIC;
4271 	sb->s_time_gran = 1;
4272 	sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
4273 		(test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
4274 	memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
4275 	sb->s_iflags |= SB_I_CGROUPWB;
4276 
4277 	/* init f2fs-specific super block info */
4278 	sbi->valid_super_block = valid_super_block;
4279 
4280 	/* disallow all the data/node/meta page writes */
4281 	set_sbi_flag(sbi, SBI_POR_DOING);
4282 
4283 	err = f2fs_init_write_merge_io(sbi);
4284 	if (err)
4285 		goto free_bio_info;
4286 
4287 	init_sb_info(sbi);
4288 
4289 	err = f2fs_init_iostat(sbi);
4290 	if (err)
4291 		goto free_bio_info;
4292 
4293 	err = init_percpu_info(sbi);
4294 	if (err)
4295 		goto free_iostat;
4296 
4297 	if (F2FS_IO_ALIGNED(sbi)) {
4298 		sbi->write_io_dummy =
4299 			mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
4300 		if (!sbi->write_io_dummy) {
4301 			err = -ENOMEM;
4302 			goto free_percpu;
4303 		}
4304 	}
4305 
4306 	/* init per sbi slab cache */
4307 	err = f2fs_init_xattr_caches(sbi);
4308 	if (err)
4309 		goto free_io_dummy;
4310 	err = f2fs_init_page_array_cache(sbi);
4311 	if (err)
4312 		goto free_xattr_cache;
4313 
4314 	/* get an inode for meta space */
4315 	sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
4316 	if (IS_ERR(sbi->meta_inode)) {
4317 		f2fs_err(sbi, "Failed to read F2FS meta data inode");
4318 		err = PTR_ERR(sbi->meta_inode);
4319 		goto free_page_array_cache;
4320 	}
4321 
4322 	err = f2fs_get_valid_checkpoint(sbi);
4323 	if (err) {
4324 		f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
4325 		goto free_meta_inode;
4326 	}
4327 
4328 	if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
4329 		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
4330 	if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
4331 		set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4332 		sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
4333 	}
4334 
4335 	if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
4336 		set_sbi_flag(sbi, SBI_NEED_FSCK);
4337 
4338 	/* Initialize device list */
4339 	err = f2fs_scan_devices(sbi);
4340 	if (err) {
4341 		f2fs_err(sbi, "Failed to find devices");
4342 		goto free_devices;
4343 	}
4344 
4345 	err = f2fs_init_post_read_wq(sbi);
4346 	if (err) {
4347 		f2fs_err(sbi, "Failed to initialize post read workqueue");
4348 		goto free_devices;
4349 	}
4350 
4351 	sbi->total_valid_node_count =
4352 				le32_to_cpu(sbi->ckpt->valid_node_count);
4353 	percpu_counter_set(&sbi->total_valid_inode_count,
4354 				le32_to_cpu(sbi->ckpt->valid_inode_count));
4355 	sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
4356 	sbi->total_valid_block_count =
4357 				le64_to_cpu(sbi->ckpt->valid_block_count);
4358 	sbi->last_valid_block_count = sbi->total_valid_block_count;
4359 	sbi->reserved_blocks = 0;
4360 	sbi->current_reserved_blocks = 0;
4361 	limit_reserve_root(sbi);
4362 	adjust_unusable_cap_perc(sbi);
4363 
4364 	f2fs_init_extent_cache_info(sbi);
4365 
4366 	f2fs_init_ino_entry_info(sbi);
4367 
4368 	f2fs_init_fsync_node_info(sbi);
4369 
4370 	/* setup checkpoint request control and start checkpoint issue thread */
4371 	f2fs_init_ckpt_req_control(sbi);
4372 	if (!f2fs_readonly(sb) && !test_opt(sbi, DISABLE_CHECKPOINT) &&
4373 			test_opt(sbi, MERGE_CHECKPOINT)) {
4374 		err = f2fs_start_ckpt_thread(sbi);
4375 		if (err) {
4376 			f2fs_err(sbi,
4377 			    "Failed to start F2FS issue_checkpoint_thread (%d)",
4378 			    err);
4379 			goto stop_ckpt_thread;
4380 		}
4381 	}
4382 
4383 	/* setup f2fs internal modules */
4384 	err = f2fs_build_segment_manager(sbi);
4385 	if (err) {
4386 		f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
4387 			 err);
4388 		goto free_sm;
4389 	}
4390 	err = f2fs_build_node_manager(sbi);
4391 	if (err) {
4392 		f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
4393 			 err);
4394 		goto free_nm;
4395 	}
4396 
4397 	err = adjust_reserved_segment(sbi);
4398 	if (err)
4399 		goto free_nm;
4400 
4401 	/* For write statistics */
4402 	sbi->sectors_written_start = f2fs_get_sectors_written(sbi);
4403 
4404 	/* Read accumulated write IO statistics if exists */
4405 	seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
4406 	if (__exist_node_summaries(sbi))
4407 		sbi->kbytes_written =
4408 			le64_to_cpu(seg_i->journal->info.kbytes_written);
4409 
4410 	f2fs_build_gc_manager(sbi);
4411 
4412 	err = f2fs_build_stats(sbi);
4413 	if (err)
4414 		goto free_nm;
4415 
4416 	/* get an inode for node space */
4417 	sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
4418 	if (IS_ERR(sbi->node_inode)) {
4419 		f2fs_err(sbi, "Failed to read node inode");
4420 		err = PTR_ERR(sbi->node_inode);
4421 		goto free_stats;
4422 	}
4423 
4424 	/* read root inode and dentry */
4425 	root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
4426 	if (IS_ERR(root)) {
4427 		f2fs_err(sbi, "Failed to read root inode");
4428 		err = PTR_ERR(root);
4429 		goto free_node_inode;
4430 	}
4431 	if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
4432 			!root->i_size || !root->i_nlink) {
4433 		iput(root);
4434 		err = -EINVAL;
4435 		goto free_node_inode;
4436 	}
4437 
4438 	sb->s_root = d_make_root(root); /* allocate root dentry */
4439 	if (!sb->s_root) {
4440 		err = -ENOMEM;
4441 		goto free_node_inode;
4442 	}
4443 
4444 	err = f2fs_init_compress_inode(sbi);
4445 	if (err)
4446 		goto free_root_inode;
4447 
4448 	err = f2fs_register_sysfs(sbi);
4449 	if (err)
4450 		goto free_compress_inode;
4451 
4452 #ifdef CONFIG_QUOTA
4453 	/* Enable quota usage during mount */
4454 	if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
4455 		err = f2fs_enable_quotas(sb);
4456 		if (err)
4457 			f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
4458 	}
4459 
4460 	quota_enabled = f2fs_recover_quota_begin(sbi);
4461 #endif
4462 	/* if there are any orphan inodes, free them */
4463 	err = f2fs_recover_orphan_inodes(sbi);
4464 	if (err)
4465 		goto free_meta;
4466 
4467 	if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
4468 		goto reset_checkpoint;
4469 
4470 	/* recover fsynced data */
4471 	if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
4472 			!test_opt(sbi, NORECOVERY)) {
4473 		/*
4474 		 * mount should be failed, when device has readonly mode, and
4475 		 * previous checkpoint was not done by clean system shutdown.
4476 		 */
4477 		if (f2fs_hw_is_readonly(sbi)) {
4478 			if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4479 				err = f2fs_recover_fsync_data(sbi, true);
4480 				if (err > 0) {
4481 					err = -EROFS;
4482 					f2fs_err(sbi, "Need to recover fsync data, but "
4483 						"write access unavailable, please try "
4484 						"mount w/ disable_roll_forward or norecovery");
4485 				}
4486 				if (err < 0)
4487 					goto free_meta;
4488 			}
4489 			f2fs_info(sbi, "write access unavailable, skipping recovery");
4490 			goto reset_checkpoint;
4491 		}
4492 
4493 		if (need_fsck)
4494 			set_sbi_flag(sbi, SBI_NEED_FSCK);
4495 
4496 		if (skip_recovery)
4497 			goto reset_checkpoint;
4498 
4499 		err = f2fs_recover_fsync_data(sbi, false);
4500 		if (err < 0) {
4501 			if (err != -ENOMEM)
4502 				skip_recovery = true;
4503 			need_fsck = true;
4504 			f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
4505 				 err);
4506 			goto free_meta;
4507 		}
4508 	} else {
4509 		err = f2fs_recover_fsync_data(sbi, true);
4510 
4511 		if (!f2fs_readonly(sb) && err > 0) {
4512 			err = -EINVAL;
4513 			f2fs_err(sbi, "Need to recover fsync data");
4514 			goto free_meta;
4515 		}
4516 	}
4517 
4518 #ifdef CONFIG_QUOTA
4519 	f2fs_recover_quota_end(sbi, quota_enabled);
4520 #endif
4521 
4522 	/*
4523 	 * If the f2fs is not readonly and fsync data recovery succeeds,
4524 	 * check zoned block devices' write pointer consistency.
4525 	 */
4526 	if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
4527 		err = f2fs_check_write_pointer(sbi);
4528 		if (err)
4529 			goto free_meta;
4530 	}
4531 
4532 reset_checkpoint:
4533 	f2fs_init_inmem_curseg(sbi);
4534 
4535 	/* f2fs_recover_fsync_data() cleared this already */
4536 	clear_sbi_flag(sbi, SBI_POR_DOING);
4537 
4538 	if (test_opt(sbi, DISABLE_CHECKPOINT)) {
4539 		err = f2fs_disable_checkpoint(sbi);
4540 		if (err)
4541 			goto sync_free_meta;
4542 	} else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
4543 		f2fs_enable_checkpoint(sbi);
4544 	}
4545 
4546 	/*
4547 	 * If filesystem is not mounted as read-only then
4548 	 * do start the gc_thread.
4549 	 */
4550 	if ((F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF ||
4551 		test_opt(sbi, GC_MERGE)) && !f2fs_readonly(sb)) {
4552 		/* After POR, we can run background GC thread.*/
4553 		err = f2fs_start_gc_thread(sbi);
4554 		if (err)
4555 			goto sync_free_meta;
4556 	}
4557 	kvfree(options);
4558 
4559 	/* recover broken superblock */
4560 	if (recovery) {
4561 		err = f2fs_commit_super(sbi, true);
4562 		f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
4563 			  sbi->valid_super_block ? 1 : 2, err);
4564 	}
4565 
4566 	f2fs_join_shrinker(sbi);
4567 
4568 	f2fs_tuning_parameters(sbi);
4569 
4570 	f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
4571 		    cur_cp_version(F2FS_CKPT(sbi)));
4572 	f2fs_update_time(sbi, CP_TIME);
4573 	f2fs_update_time(sbi, REQ_TIME);
4574 	clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4575 	return 0;
4576 
4577 sync_free_meta:
4578 	/* safe to flush all the data */
4579 	sync_filesystem(sbi->sb);
4580 	retry_cnt = 0;
4581 
4582 free_meta:
4583 #ifdef CONFIG_QUOTA
4584 	f2fs_truncate_quota_inode_pages(sb);
4585 	if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
4586 		f2fs_quota_off_umount(sbi->sb);
4587 #endif
4588 	/*
4589 	 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
4590 	 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
4591 	 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
4592 	 * falls into an infinite loop in f2fs_sync_meta_pages().
4593 	 */
4594 	truncate_inode_pages_final(META_MAPPING(sbi));
4595 	/* evict some inodes being cached by GC */
4596 	evict_inodes(sb);
4597 	f2fs_unregister_sysfs(sbi);
4598 free_compress_inode:
4599 	f2fs_destroy_compress_inode(sbi);
4600 free_root_inode:
4601 	dput(sb->s_root);
4602 	sb->s_root = NULL;
4603 free_node_inode:
4604 	f2fs_release_ino_entry(sbi, true);
4605 	truncate_inode_pages_final(NODE_MAPPING(sbi));
4606 	iput(sbi->node_inode);
4607 	sbi->node_inode = NULL;
4608 free_stats:
4609 	f2fs_destroy_stats(sbi);
4610 free_nm:
4611 	/* stop discard thread before destroying node manager */
4612 	f2fs_stop_discard_thread(sbi);
4613 	f2fs_destroy_node_manager(sbi);
4614 free_sm:
4615 	f2fs_destroy_segment_manager(sbi);
4616 stop_ckpt_thread:
4617 	f2fs_stop_ckpt_thread(sbi);
4618 	f2fs_destroy_post_read_wq(sbi);
4619 free_devices:
4620 	destroy_device_list(sbi);
4621 	kvfree(sbi->ckpt);
4622 free_meta_inode:
4623 	make_bad_inode(sbi->meta_inode);
4624 	iput(sbi->meta_inode);
4625 	sbi->meta_inode = NULL;
4626 free_page_array_cache:
4627 	f2fs_destroy_page_array_cache(sbi);
4628 free_xattr_cache:
4629 	f2fs_destroy_xattr_caches(sbi);
4630 free_io_dummy:
4631 	mempool_destroy(sbi->write_io_dummy);
4632 free_percpu:
4633 	destroy_percpu_info(sbi);
4634 free_iostat:
4635 	f2fs_destroy_iostat(sbi);
4636 free_bio_info:
4637 	for (i = 0; i < NR_PAGE_TYPE; i++)
4638 		kvfree(sbi->write_io[i]);
4639 
4640 #if IS_ENABLED(CONFIG_UNICODE)
4641 	utf8_unload(sb->s_encoding);
4642 	sb->s_encoding = NULL;
4643 #endif
4644 free_options:
4645 #ifdef CONFIG_QUOTA
4646 	for (i = 0; i < MAXQUOTAS; i++)
4647 		kfree(F2FS_OPTION(sbi).s_qf_names[i]);
4648 #endif
4649 	fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
4650 	kvfree(options);
4651 free_sb_buf:
4652 	kfree(raw_super);
4653 free_sbi:
4654 	if (sbi->s_chksum_driver)
4655 		crypto_free_shash(sbi->s_chksum_driver);
4656 	kfree(sbi);
4657 
4658 	/* give only one another chance */
4659 	if (retry_cnt > 0 && skip_recovery) {
4660 		retry_cnt--;
4661 		shrink_dcache_sb(sb);
4662 		goto try_onemore;
4663 	}
4664 	return err;
4665 }
4666 
4667 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
4668 			const char *dev_name, void *data)
4669 {
4670 	return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
4671 }
4672 
4673 static void kill_f2fs_super(struct super_block *sb)
4674 {
4675 	if (sb->s_root) {
4676 		struct f2fs_sb_info *sbi = F2FS_SB(sb);
4677 
4678 		set_sbi_flag(sbi, SBI_IS_CLOSE);
4679 		f2fs_stop_gc_thread(sbi);
4680 		f2fs_stop_discard_thread(sbi);
4681 
4682 #ifdef CONFIG_F2FS_FS_COMPRESSION
4683 		/*
4684 		 * latter evict_inode() can bypass checking and invalidating
4685 		 * compress inode cache.
4686 		 */
4687 		if (test_opt(sbi, COMPRESS_CACHE))
4688 			truncate_inode_pages_final(COMPRESS_MAPPING(sbi));
4689 #endif
4690 
4691 		if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
4692 				!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4693 			struct cp_control cpc = {
4694 				.reason = CP_UMOUNT,
4695 			};
4696 			f2fs_write_checkpoint(sbi, &cpc);
4697 		}
4698 
4699 		if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
4700 			sb->s_flags &= ~SB_RDONLY;
4701 	}
4702 	kill_block_super(sb);
4703 }
4704 
4705 static struct file_system_type f2fs_fs_type = {
4706 	.owner		= THIS_MODULE,
4707 	.name		= "f2fs",
4708 	.mount		= f2fs_mount,
4709 	.kill_sb	= kill_f2fs_super,
4710 	.fs_flags	= FS_REQUIRES_DEV | FS_ALLOW_IDMAP,
4711 };
4712 MODULE_ALIAS_FS("f2fs");
4713 
4714 static int __init init_inodecache(void)
4715 {
4716 	f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
4717 			sizeof(struct f2fs_inode_info), 0,
4718 			SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
4719 	return f2fs_inode_cachep ? 0 : -ENOMEM;
4720 }
4721 
4722 static void destroy_inodecache(void)
4723 {
4724 	/*
4725 	 * Make sure all delayed rcu free inodes are flushed before we
4726 	 * destroy cache.
4727 	 */
4728 	rcu_barrier();
4729 	kmem_cache_destroy(f2fs_inode_cachep);
4730 }
4731 
4732 static int __init init_f2fs_fs(void)
4733 {
4734 	int err;
4735 
4736 	if (PAGE_SIZE != F2FS_BLKSIZE) {
4737 		printk("F2FS not supported on PAGE_SIZE(%lu) != %d\n",
4738 				PAGE_SIZE, F2FS_BLKSIZE);
4739 		return -EINVAL;
4740 	}
4741 
4742 	err = init_inodecache();
4743 	if (err)
4744 		goto fail;
4745 	err = f2fs_create_node_manager_caches();
4746 	if (err)
4747 		goto free_inodecache;
4748 	err = f2fs_create_segment_manager_caches();
4749 	if (err)
4750 		goto free_node_manager_caches;
4751 	err = f2fs_create_checkpoint_caches();
4752 	if (err)
4753 		goto free_segment_manager_caches;
4754 	err = f2fs_create_recovery_cache();
4755 	if (err)
4756 		goto free_checkpoint_caches;
4757 	err = f2fs_create_extent_cache();
4758 	if (err)
4759 		goto free_recovery_cache;
4760 	err = f2fs_create_garbage_collection_cache();
4761 	if (err)
4762 		goto free_extent_cache;
4763 	err = f2fs_init_sysfs();
4764 	if (err)
4765 		goto free_garbage_collection_cache;
4766 	err = register_shrinker(&f2fs_shrinker_info, "f2fs-shrinker");
4767 	if (err)
4768 		goto free_sysfs;
4769 	err = register_filesystem(&f2fs_fs_type);
4770 	if (err)
4771 		goto free_shrinker;
4772 	f2fs_create_root_stats();
4773 	err = f2fs_init_post_read_processing();
4774 	if (err)
4775 		goto free_root_stats;
4776 	err = f2fs_init_iostat_processing();
4777 	if (err)
4778 		goto free_post_read;
4779 	err = f2fs_init_bio_entry_cache();
4780 	if (err)
4781 		goto free_iostat;
4782 	err = f2fs_init_bioset();
4783 	if (err)
4784 		goto free_bio_entry_cache;
4785 	err = f2fs_init_compress_mempool();
4786 	if (err)
4787 		goto free_bioset;
4788 	err = f2fs_init_compress_cache();
4789 	if (err)
4790 		goto free_compress_mempool;
4791 	err = f2fs_create_casefold_cache();
4792 	if (err)
4793 		goto free_compress_cache;
4794 	return 0;
4795 free_compress_cache:
4796 	f2fs_destroy_compress_cache();
4797 free_compress_mempool:
4798 	f2fs_destroy_compress_mempool();
4799 free_bioset:
4800 	f2fs_destroy_bioset();
4801 free_bio_entry_cache:
4802 	f2fs_destroy_bio_entry_cache();
4803 free_iostat:
4804 	f2fs_destroy_iostat_processing();
4805 free_post_read:
4806 	f2fs_destroy_post_read_processing();
4807 free_root_stats:
4808 	f2fs_destroy_root_stats();
4809 	unregister_filesystem(&f2fs_fs_type);
4810 free_shrinker:
4811 	unregister_shrinker(&f2fs_shrinker_info);
4812 free_sysfs:
4813 	f2fs_exit_sysfs();
4814 free_garbage_collection_cache:
4815 	f2fs_destroy_garbage_collection_cache();
4816 free_extent_cache:
4817 	f2fs_destroy_extent_cache();
4818 free_recovery_cache:
4819 	f2fs_destroy_recovery_cache();
4820 free_checkpoint_caches:
4821 	f2fs_destroy_checkpoint_caches();
4822 free_segment_manager_caches:
4823 	f2fs_destroy_segment_manager_caches();
4824 free_node_manager_caches:
4825 	f2fs_destroy_node_manager_caches();
4826 free_inodecache:
4827 	destroy_inodecache();
4828 fail:
4829 	return err;
4830 }
4831 
4832 static void __exit exit_f2fs_fs(void)
4833 {
4834 	f2fs_destroy_casefold_cache();
4835 	f2fs_destroy_compress_cache();
4836 	f2fs_destroy_compress_mempool();
4837 	f2fs_destroy_bioset();
4838 	f2fs_destroy_bio_entry_cache();
4839 	f2fs_destroy_iostat_processing();
4840 	f2fs_destroy_post_read_processing();
4841 	f2fs_destroy_root_stats();
4842 	unregister_filesystem(&f2fs_fs_type);
4843 	unregister_shrinker(&f2fs_shrinker_info);
4844 	f2fs_exit_sysfs();
4845 	f2fs_destroy_garbage_collection_cache();
4846 	f2fs_destroy_extent_cache();
4847 	f2fs_destroy_recovery_cache();
4848 	f2fs_destroy_checkpoint_caches();
4849 	f2fs_destroy_segment_manager_caches();
4850 	f2fs_destroy_node_manager_caches();
4851 	destroy_inodecache();
4852 }
4853 
4854 module_init(init_f2fs_fs)
4855 module_exit(exit_f2fs_fs)
4856 
4857 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
4858 MODULE_DESCRIPTION("Flash Friendly File System");
4859 MODULE_LICENSE("GPL");
4860 MODULE_SOFTDEP("pre: crc32");
4861 
4862