xref: /linux/fs/f2fs/super.c (revision 17b121ad0c43342bc894632f6710b894849ca372)
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  * fs/f2fs/super.c
4  *
5  * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6  *             http://www.samsung.com/
7  */
8 #include <linux/module.h>
9 #include <linux/init.h>
10 #include <linux/fs.h>
11 #include <linux/statfs.h>
12 #include <linux/buffer_head.h>
13 #include <linux/backing-dev.h>
14 #include <linux/kthread.h>
15 #include <linux/parser.h>
16 #include <linux/mount.h>
17 #include <linux/seq_file.h>
18 #include <linux/proc_fs.h>
19 #include <linux/random.h>
20 #include <linux/exportfs.h>
21 #include <linux/blkdev.h>
22 #include <linux/quotaops.h>
23 #include <linux/f2fs_fs.h>
24 #include <linux/sysfs.h>
25 #include <linux/quota.h>
26 #include <linux/unicode.h>
27 #include <linux/part_stat.h>
28 #include <linux/zstd.h>
29 #include <linux/lz4.h>
30 
31 #include "f2fs.h"
32 #include "node.h"
33 #include "segment.h"
34 #include "xattr.h"
35 #include "gc.h"
36 
37 #define CREATE_TRACE_POINTS
38 #include <trace/events/f2fs.h>
39 
40 static struct kmem_cache *f2fs_inode_cachep;
41 
42 #ifdef CONFIG_F2FS_FAULT_INJECTION
43 
44 const char *f2fs_fault_name[FAULT_MAX] = {
45 	[FAULT_KMALLOC]		= "kmalloc",
46 	[FAULT_KVMALLOC]	= "kvmalloc",
47 	[FAULT_PAGE_ALLOC]	= "page alloc",
48 	[FAULT_PAGE_GET]	= "page get",
49 	[FAULT_ALLOC_NID]	= "alloc nid",
50 	[FAULT_ORPHAN]		= "orphan",
51 	[FAULT_BLOCK]		= "no more block",
52 	[FAULT_DIR_DEPTH]	= "too big dir depth",
53 	[FAULT_EVICT_INODE]	= "evict_inode fail",
54 	[FAULT_TRUNCATE]	= "truncate fail",
55 	[FAULT_READ_IO]		= "read IO error",
56 	[FAULT_CHECKPOINT]	= "checkpoint error",
57 	[FAULT_DISCARD]		= "discard error",
58 	[FAULT_WRITE_IO]	= "write IO error",
59 };
60 
61 void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate,
62 							unsigned int type)
63 {
64 	struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
65 
66 	if (rate) {
67 		atomic_set(&ffi->inject_ops, 0);
68 		ffi->inject_rate = rate;
69 	}
70 
71 	if (type)
72 		ffi->inject_type = type;
73 
74 	if (!rate && !type)
75 		memset(ffi, 0, sizeof(struct f2fs_fault_info));
76 }
77 #endif
78 
79 /* f2fs-wide shrinker description */
80 static struct shrinker f2fs_shrinker_info = {
81 	.scan_objects = f2fs_shrink_scan,
82 	.count_objects = f2fs_shrink_count,
83 	.seeks = DEFAULT_SEEKS,
84 };
85 
86 enum {
87 	Opt_gc_background,
88 	Opt_disable_roll_forward,
89 	Opt_norecovery,
90 	Opt_discard,
91 	Opt_nodiscard,
92 	Opt_noheap,
93 	Opt_heap,
94 	Opt_user_xattr,
95 	Opt_nouser_xattr,
96 	Opt_acl,
97 	Opt_noacl,
98 	Opt_active_logs,
99 	Opt_disable_ext_identify,
100 	Opt_inline_xattr,
101 	Opt_noinline_xattr,
102 	Opt_inline_xattr_size,
103 	Opt_inline_data,
104 	Opt_inline_dentry,
105 	Opt_noinline_dentry,
106 	Opt_flush_merge,
107 	Opt_noflush_merge,
108 	Opt_nobarrier,
109 	Opt_fastboot,
110 	Opt_extent_cache,
111 	Opt_noextent_cache,
112 	Opt_noinline_data,
113 	Opt_data_flush,
114 	Opt_reserve_root,
115 	Opt_resgid,
116 	Opt_resuid,
117 	Opt_mode,
118 	Opt_io_size_bits,
119 	Opt_fault_injection,
120 	Opt_fault_type,
121 	Opt_lazytime,
122 	Opt_nolazytime,
123 	Opt_quota,
124 	Opt_noquota,
125 	Opt_usrquota,
126 	Opt_grpquota,
127 	Opt_prjquota,
128 	Opt_usrjquota,
129 	Opt_grpjquota,
130 	Opt_prjjquota,
131 	Opt_offusrjquota,
132 	Opt_offgrpjquota,
133 	Opt_offprjjquota,
134 	Opt_jqfmt_vfsold,
135 	Opt_jqfmt_vfsv0,
136 	Opt_jqfmt_vfsv1,
137 	Opt_whint,
138 	Opt_alloc,
139 	Opt_fsync,
140 	Opt_test_dummy_encryption,
141 	Opt_inlinecrypt,
142 	Opt_checkpoint_disable,
143 	Opt_checkpoint_disable_cap,
144 	Opt_checkpoint_disable_cap_perc,
145 	Opt_checkpoint_enable,
146 	Opt_checkpoint_merge,
147 	Opt_nocheckpoint_merge,
148 	Opt_compress_algorithm,
149 	Opt_compress_log_size,
150 	Opt_compress_extension,
151 	Opt_nocompress_extension,
152 	Opt_compress_chksum,
153 	Opt_compress_mode,
154 	Opt_compress_cache,
155 	Opt_atgc,
156 	Opt_gc_merge,
157 	Opt_nogc_merge,
158 	Opt_err,
159 };
160 
161 static match_table_t f2fs_tokens = {
162 	{Opt_gc_background, "background_gc=%s"},
163 	{Opt_disable_roll_forward, "disable_roll_forward"},
164 	{Opt_norecovery, "norecovery"},
165 	{Opt_discard, "discard"},
166 	{Opt_nodiscard, "nodiscard"},
167 	{Opt_noheap, "no_heap"},
168 	{Opt_heap, "heap"},
169 	{Opt_user_xattr, "user_xattr"},
170 	{Opt_nouser_xattr, "nouser_xattr"},
171 	{Opt_acl, "acl"},
172 	{Opt_noacl, "noacl"},
173 	{Opt_active_logs, "active_logs=%u"},
174 	{Opt_disable_ext_identify, "disable_ext_identify"},
175 	{Opt_inline_xattr, "inline_xattr"},
176 	{Opt_noinline_xattr, "noinline_xattr"},
177 	{Opt_inline_xattr_size, "inline_xattr_size=%u"},
178 	{Opt_inline_data, "inline_data"},
179 	{Opt_inline_dentry, "inline_dentry"},
180 	{Opt_noinline_dentry, "noinline_dentry"},
181 	{Opt_flush_merge, "flush_merge"},
182 	{Opt_noflush_merge, "noflush_merge"},
183 	{Opt_nobarrier, "nobarrier"},
184 	{Opt_fastboot, "fastboot"},
185 	{Opt_extent_cache, "extent_cache"},
186 	{Opt_noextent_cache, "noextent_cache"},
187 	{Opt_noinline_data, "noinline_data"},
188 	{Opt_data_flush, "data_flush"},
189 	{Opt_reserve_root, "reserve_root=%u"},
190 	{Opt_resgid, "resgid=%u"},
191 	{Opt_resuid, "resuid=%u"},
192 	{Opt_mode, "mode=%s"},
193 	{Opt_io_size_bits, "io_bits=%u"},
194 	{Opt_fault_injection, "fault_injection=%u"},
195 	{Opt_fault_type, "fault_type=%u"},
196 	{Opt_lazytime, "lazytime"},
197 	{Opt_nolazytime, "nolazytime"},
198 	{Opt_quota, "quota"},
199 	{Opt_noquota, "noquota"},
200 	{Opt_usrquota, "usrquota"},
201 	{Opt_grpquota, "grpquota"},
202 	{Opt_prjquota, "prjquota"},
203 	{Opt_usrjquota, "usrjquota=%s"},
204 	{Opt_grpjquota, "grpjquota=%s"},
205 	{Opt_prjjquota, "prjjquota=%s"},
206 	{Opt_offusrjquota, "usrjquota="},
207 	{Opt_offgrpjquota, "grpjquota="},
208 	{Opt_offprjjquota, "prjjquota="},
209 	{Opt_jqfmt_vfsold, "jqfmt=vfsold"},
210 	{Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
211 	{Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
212 	{Opt_whint, "whint_mode=%s"},
213 	{Opt_alloc, "alloc_mode=%s"},
214 	{Opt_fsync, "fsync_mode=%s"},
215 	{Opt_test_dummy_encryption, "test_dummy_encryption=%s"},
216 	{Opt_test_dummy_encryption, "test_dummy_encryption"},
217 	{Opt_inlinecrypt, "inlinecrypt"},
218 	{Opt_checkpoint_disable, "checkpoint=disable"},
219 	{Opt_checkpoint_disable_cap, "checkpoint=disable:%u"},
220 	{Opt_checkpoint_disable_cap_perc, "checkpoint=disable:%u%%"},
221 	{Opt_checkpoint_enable, "checkpoint=enable"},
222 	{Opt_checkpoint_merge, "checkpoint_merge"},
223 	{Opt_nocheckpoint_merge, "nocheckpoint_merge"},
224 	{Opt_compress_algorithm, "compress_algorithm=%s"},
225 	{Opt_compress_log_size, "compress_log_size=%u"},
226 	{Opt_compress_extension, "compress_extension=%s"},
227 	{Opt_nocompress_extension, "nocompress_extension=%s"},
228 	{Opt_compress_chksum, "compress_chksum"},
229 	{Opt_compress_mode, "compress_mode=%s"},
230 	{Opt_compress_cache, "compress_cache"},
231 	{Opt_atgc, "atgc"},
232 	{Opt_gc_merge, "gc_merge"},
233 	{Opt_nogc_merge, "nogc_merge"},
234 	{Opt_err, NULL},
235 };
236 
237 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...)
238 {
239 	struct va_format vaf;
240 	va_list args;
241 	int level;
242 
243 	va_start(args, fmt);
244 
245 	level = printk_get_level(fmt);
246 	vaf.fmt = printk_skip_level(fmt);
247 	vaf.va = &args;
248 	printk("%c%cF2FS-fs (%s): %pV\n",
249 	       KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
250 
251 	va_end(args);
252 }
253 
254 #ifdef CONFIG_UNICODE
255 static const struct f2fs_sb_encodings {
256 	__u16 magic;
257 	char *name;
258 	char *version;
259 } f2fs_sb_encoding_map[] = {
260 	{F2FS_ENC_UTF8_12_1, "utf8", "12.1.0"},
261 };
262 
263 static int f2fs_sb_read_encoding(const struct f2fs_super_block *sb,
264 				 const struct f2fs_sb_encodings **encoding,
265 				 __u16 *flags)
266 {
267 	__u16 magic = le16_to_cpu(sb->s_encoding);
268 	int i;
269 
270 	for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
271 		if (magic == f2fs_sb_encoding_map[i].magic)
272 			break;
273 
274 	if (i >= ARRAY_SIZE(f2fs_sb_encoding_map))
275 		return -EINVAL;
276 
277 	*encoding = &f2fs_sb_encoding_map[i];
278 	*flags = le16_to_cpu(sb->s_encoding_flags);
279 
280 	return 0;
281 }
282 
283 struct kmem_cache *f2fs_cf_name_slab;
284 static int __init f2fs_create_casefold_cache(void)
285 {
286 	f2fs_cf_name_slab = f2fs_kmem_cache_create("f2fs_casefolded_name",
287 							F2FS_NAME_LEN);
288 	if (!f2fs_cf_name_slab)
289 		return -ENOMEM;
290 	return 0;
291 }
292 
293 static void f2fs_destroy_casefold_cache(void)
294 {
295 	kmem_cache_destroy(f2fs_cf_name_slab);
296 }
297 #else
298 static int __init f2fs_create_casefold_cache(void) { return 0; }
299 static void f2fs_destroy_casefold_cache(void) { }
300 #endif
301 
302 static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
303 {
304 	block_t limit = min((sbi->user_block_count << 1) / 1000,
305 			sbi->user_block_count - sbi->reserved_blocks);
306 
307 	/* limit is 0.2% */
308 	if (test_opt(sbi, RESERVE_ROOT) &&
309 			F2FS_OPTION(sbi).root_reserved_blocks > limit) {
310 		F2FS_OPTION(sbi).root_reserved_blocks = limit;
311 		f2fs_info(sbi, "Reduce reserved blocks for root = %u",
312 			  F2FS_OPTION(sbi).root_reserved_blocks);
313 	}
314 	if (!test_opt(sbi, RESERVE_ROOT) &&
315 		(!uid_eq(F2FS_OPTION(sbi).s_resuid,
316 				make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
317 		!gid_eq(F2FS_OPTION(sbi).s_resgid,
318 				make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
319 		f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
320 			  from_kuid_munged(&init_user_ns,
321 					   F2FS_OPTION(sbi).s_resuid),
322 			  from_kgid_munged(&init_user_ns,
323 					   F2FS_OPTION(sbi).s_resgid));
324 }
325 
326 static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
327 {
328 	if (!F2FS_OPTION(sbi).unusable_cap_perc)
329 		return;
330 
331 	if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
332 		F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
333 	else
334 		F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
335 					F2FS_OPTION(sbi).unusable_cap_perc;
336 
337 	f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
338 			F2FS_OPTION(sbi).unusable_cap,
339 			F2FS_OPTION(sbi).unusable_cap_perc);
340 }
341 
342 static void init_once(void *foo)
343 {
344 	struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
345 
346 	inode_init_once(&fi->vfs_inode);
347 }
348 
349 #ifdef CONFIG_QUOTA
350 static const char * const quotatypes[] = INITQFNAMES;
351 #define QTYPE2NAME(t) (quotatypes[t])
352 static int f2fs_set_qf_name(struct super_block *sb, int qtype,
353 							substring_t *args)
354 {
355 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
356 	char *qname;
357 	int ret = -EINVAL;
358 
359 	if (sb_any_quota_loaded(sb) && !F2FS_OPTION(sbi).s_qf_names[qtype]) {
360 		f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
361 		return -EINVAL;
362 	}
363 	if (f2fs_sb_has_quota_ino(sbi)) {
364 		f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
365 		return 0;
366 	}
367 
368 	qname = match_strdup(args);
369 	if (!qname) {
370 		f2fs_err(sbi, "Not enough memory for storing quotafile name");
371 		return -ENOMEM;
372 	}
373 	if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
374 		if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
375 			ret = 0;
376 		else
377 			f2fs_err(sbi, "%s quota file already specified",
378 				 QTYPE2NAME(qtype));
379 		goto errout;
380 	}
381 	if (strchr(qname, '/')) {
382 		f2fs_err(sbi, "quotafile must be on filesystem root");
383 		goto errout;
384 	}
385 	F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
386 	set_opt(sbi, QUOTA);
387 	return 0;
388 errout:
389 	kfree(qname);
390 	return ret;
391 }
392 
393 static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
394 {
395 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
396 
397 	if (sb_any_quota_loaded(sb) && F2FS_OPTION(sbi).s_qf_names[qtype]) {
398 		f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
399 		return -EINVAL;
400 	}
401 	kfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
402 	F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
403 	return 0;
404 }
405 
406 static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
407 {
408 	/*
409 	 * We do the test below only for project quotas. 'usrquota' and
410 	 * 'grpquota' mount options are allowed even without quota feature
411 	 * to support legacy quotas in quota files.
412 	 */
413 	if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi)) {
414 		f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
415 		return -1;
416 	}
417 	if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
418 			F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
419 			F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) {
420 		if (test_opt(sbi, USRQUOTA) &&
421 				F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
422 			clear_opt(sbi, USRQUOTA);
423 
424 		if (test_opt(sbi, GRPQUOTA) &&
425 				F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
426 			clear_opt(sbi, GRPQUOTA);
427 
428 		if (test_opt(sbi, PRJQUOTA) &&
429 				F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
430 			clear_opt(sbi, PRJQUOTA);
431 
432 		if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
433 				test_opt(sbi, PRJQUOTA)) {
434 			f2fs_err(sbi, "old and new quota format mixing");
435 			return -1;
436 		}
437 
438 		if (!F2FS_OPTION(sbi).s_jquota_fmt) {
439 			f2fs_err(sbi, "journaled quota format not specified");
440 			return -1;
441 		}
442 	}
443 
444 	if (f2fs_sb_has_quota_ino(sbi) && F2FS_OPTION(sbi).s_jquota_fmt) {
445 		f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
446 		F2FS_OPTION(sbi).s_jquota_fmt = 0;
447 	}
448 	return 0;
449 }
450 #endif
451 
452 static int f2fs_set_test_dummy_encryption(struct super_block *sb,
453 					  const char *opt,
454 					  const substring_t *arg,
455 					  bool is_remount)
456 {
457 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
458 #ifdef CONFIG_FS_ENCRYPTION
459 	int err;
460 
461 	if (!f2fs_sb_has_encrypt(sbi)) {
462 		f2fs_err(sbi, "Encrypt feature is off");
463 		return -EINVAL;
464 	}
465 
466 	/*
467 	 * This mount option is just for testing, and it's not worthwhile to
468 	 * implement the extra complexity (e.g. RCU protection) that would be
469 	 * needed to allow it to be set or changed during remount.  We do allow
470 	 * it to be specified during remount, but only if there is no change.
471 	 */
472 	if (is_remount && !F2FS_OPTION(sbi).dummy_enc_policy.policy) {
473 		f2fs_warn(sbi, "Can't set test_dummy_encryption on remount");
474 		return -EINVAL;
475 	}
476 	err = fscrypt_set_test_dummy_encryption(
477 		sb, arg->from, &F2FS_OPTION(sbi).dummy_enc_policy);
478 	if (err) {
479 		if (err == -EEXIST)
480 			f2fs_warn(sbi,
481 				  "Can't change test_dummy_encryption on remount");
482 		else if (err == -EINVAL)
483 			f2fs_warn(sbi, "Value of option \"%s\" is unrecognized",
484 				  opt);
485 		else
486 			f2fs_warn(sbi, "Error processing option \"%s\" [%d]",
487 				  opt, err);
488 		return -EINVAL;
489 	}
490 	f2fs_warn(sbi, "Test dummy encryption mode enabled");
491 #else
492 	f2fs_warn(sbi, "Test dummy encryption mount option ignored");
493 #endif
494 	return 0;
495 }
496 
497 #ifdef CONFIG_F2FS_FS_COMPRESSION
498 /*
499  * 1. The same extension name cannot not appear in both compress and non-compress extension
500  * at the same time.
501  * 2. If the compress extension specifies all files, the types specified by the non-compress
502  * extension will be treated as special cases and will not be compressed.
503  * 3. Don't allow the non-compress extension specifies all files.
504  */
505 static int f2fs_test_compress_extension(struct f2fs_sb_info *sbi)
506 {
507 	unsigned char (*ext)[F2FS_EXTENSION_LEN];
508 	unsigned char (*noext)[F2FS_EXTENSION_LEN];
509 	int ext_cnt, noext_cnt, index = 0, no_index = 0;
510 
511 	ext = F2FS_OPTION(sbi).extensions;
512 	ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
513 	noext = F2FS_OPTION(sbi).noextensions;
514 	noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
515 
516 	if (!noext_cnt)
517 		return 0;
518 
519 	for (no_index = 0; no_index < noext_cnt; no_index++) {
520 		if (!strcasecmp("*", noext[no_index])) {
521 			f2fs_info(sbi, "Don't allow the nocompress extension specifies all files");
522 			return -EINVAL;
523 		}
524 		for (index = 0; index < ext_cnt; index++) {
525 			if (!strcasecmp(ext[index], noext[no_index])) {
526 				f2fs_info(sbi, "Don't allow the same extension %s appear in both compress and nocompress extension",
527 						ext[index]);
528 				return -EINVAL;
529 			}
530 		}
531 	}
532 	return 0;
533 }
534 
535 #ifdef CONFIG_F2FS_FS_LZ4
536 static int f2fs_set_lz4hc_level(struct f2fs_sb_info *sbi, const char *str)
537 {
538 #ifdef CONFIG_F2FS_FS_LZ4HC
539 	unsigned int level;
540 #endif
541 
542 	if (strlen(str) == 3) {
543 		F2FS_OPTION(sbi).compress_level = 0;
544 		return 0;
545 	}
546 
547 #ifdef CONFIG_F2FS_FS_LZ4HC
548 	str += 3;
549 
550 	if (str[0] != ':') {
551 		f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
552 		return -EINVAL;
553 	}
554 	if (kstrtouint(str + 1, 10, &level))
555 		return -EINVAL;
556 
557 	if (level < LZ4HC_MIN_CLEVEL || level > LZ4HC_MAX_CLEVEL) {
558 		f2fs_info(sbi, "invalid lz4hc compress level: %d", level);
559 		return -EINVAL;
560 	}
561 
562 	F2FS_OPTION(sbi).compress_level = level;
563 	return 0;
564 #else
565 	f2fs_info(sbi, "kernel doesn't support lz4hc compression");
566 	return -EINVAL;
567 #endif
568 }
569 #endif
570 
571 #ifdef CONFIG_F2FS_FS_ZSTD
572 static int f2fs_set_zstd_level(struct f2fs_sb_info *sbi, const char *str)
573 {
574 	unsigned int level;
575 	int len = 4;
576 
577 	if (strlen(str) == len) {
578 		F2FS_OPTION(sbi).compress_level = 0;
579 		return 0;
580 	}
581 
582 	str += len;
583 
584 	if (str[0] != ':') {
585 		f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
586 		return -EINVAL;
587 	}
588 	if (kstrtouint(str + 1, 10, &level))
589 		return -EINVAL;
590 
591 	if (!level || level > ZSTD_maxCLevel()) {
592 		f2fs_info(sbi, "invalid zstd compress level: %d", level);
593 		return -EINVAL;
594 	}
595 
596 	F2FS_OPTION(sbi).compress_level = level;
597 	return 0;
598 }
599 #endif
600 #endif
601 
602 static int parse_options(struct super_block *sb, char *options, bool is_remount)
603 {
604 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
605 	substring_t args[MAX_OPT_ARGS];
606 #ifdef CONFIG_F2FS_FS_COMPRESSION
607 	unsigned char (*ext)[F2FS_EXTENSION_LEN];
608 	unsigned char (*noext)[F2FS_EXTENSION_LEN];
609 	int ext_cnt, noext_cnt;
610 #endif
611 	char *p, *name;
612 	int arg = 0;
613 	kuid_t uid;
614 	kgid_t gid;
615 	int ret;
616 
617 	if (!options)
618 		goto default_check;
619 
620 	while ((p = strsep(&options, ",")) != NULL) {
621 		int token;
622 
623 		if (!*p)
624 			continue;
625 		/*
626 		 * Initialize args struct so we know whether arg was
627 		 * found; some options take optional arguments.
628 		 */
629 		args[0].to = args[0].from = NULL;
630 		token = match_token(p, f2fs_tokens, args);
631 
632 		switch (token) {
633 		case Opt_gc_background:
634 			name = match_strdup(&args[0]);
635 
636 			if (!name)
637 				return -ENOMEM;
638 			if (!strcmp(name, "on")) {
639 				F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
640 			} else if (!strcmp(name, "off")) {
641 				F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_OFF;
642 			} else if (!strcmp(name, "sync")) {
643 				F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_SYNC;
644 			} else {
645 				kfree(name);
646 				return -EINVAL;
647 			}
648 			kfree(name);
649 			break;
650 		case Opt_disable_roll_forward:
651 			set_opt(sbi, DISABLE_ROLL_FORWARD);
652 			break;
653 		case Opt_norecovery:
654 			/* this option mounts f2fs with ro */
655 			set_opt(sbi, NORECOVERY);
656 			if (!f2fs_readonly(sb))
657 				return -EINVAL;
658 			break;
659 		case Opt_discard:
660 			set_opt(sbi, DISCARD);
661 			break;
662 		case Opt_nodiscard:
663 			if (f2fs_sb_has_blkzoned(sbi)) {
664 				f2fs_warn(sbi, "discard is required for zoned block devices");
665 				return -EINVAL;
666 			}
667 			clear_opt(sbi, DISCARD);
668 			break;
669 		case Opt_noheap:
670 			set_opt(sbi, NOHEAP);
671 			break;
672 		case Opt_heap:
673 			clear_opt(sbi, NOHEAP);
674 			break;
675 #ifdef CONFIG_F2FS_FS_XATTR
676 		case Opt_user_xattr:
677 			set_opt(sbi, XATTR_USER);
678 			break;
679 		case Opt_nouser_xattr:
680 			clear_opt(sbi, XATTR_USER);
681 			break;
682 		case Opt_inline_xattr:
683 			set_opt(sbi, INLINE_XATTR);
684 			break;
685 		case Opt_noinline_xattr:
686 			clear_opt(sbi, INLINE_XATTR);
687 			break;
688 		case Opt_inline_xattr_size:
689 			if (args->from && match_int(args, &arg))
690 				return -EINVAL;
691 			set_opt(sbi, INLINE_XATTR_SIZE);
692 			F2FS_OPTION(sbi).inline_xattr_size = arg;
693 			break;
694 #else
695 		case Opt_user_xattr:
696 			f2fs_info(sbi, "user_xattr options not supported");
697 			break;
698 		case Opt_nouser_xattr:
699 			f2fs_info(sbi, "nouser_xattr options not supported");
700 			break;
701 		case Opt_inline_xattr:
702 			f2fs_info(sbi, "inline_xattr options not supported");
703 			break;
704 		case Opt_noinline_xattr:
705 			f2fs_info(sbi, "noinline_xattr options not supported");
706 			break;
707 #endif
708 #ifdef CONFIG_F2FS_FS_POSIX_ACL
709 		case Opt_acl:
710 			set_opt(sbi, POSIX_ACL);
711 			break;
712 		case Opt_noacl:
713 			clear_opt(sbi, POSIX_ACL);
714 			break;
715 #else
716 		case Opt_acl:
717 			f2fs_info(sbi, "acl options not supported");
718 			break;
719 		case Opt_noacl:
720 			f2fs_info(sbi, "noacl options not supported");
721 			break;
722 #endif
723 		case Opt_active_logs:
724 			if (args->from && match_int(args, &arg))
725 				return -EINVAL;
726 			if (arg != 2 && arg != 4 &&
727 				arg != NR_CURSEG_PERSIST_TYPE)
728 				return -EINVAL;
729 			F2FS_OPTION(sbi).active_logs = arg;
730 			break;
731 		case Opt_disable_ext_identify:
732 			set_opt(sbi, DISABLE_EXT_IDENTIFY);
733 			break;
734 		case Opt_inline_data:
735 			set_opt(sbi, INLINE_DATA);
736 			break;
737 		case Opt_inline_dentry:
738 			set_opt(sbi, INLINE_DENTRY);
739 			break;
740 		case Opt_noinline_dentry:
741 			clear_opt(sbi, INLINE_DENTRY);
742 			break;
743 		case Opt_flush_merge:
744 			set_opt(sbi, FLUSH_MERGE);
745 			break;
746 		case Opt_noflush_merge:
747 			clear_opt(sbi, FLUSH_MERGE);
748 			break;
749 		case Opt_nobarrier:
750 			set_opt(sbi, NOBARRIER);
751 			break;
752 		case Opt_fastboot:
753 			set_opt(sbi, FASTBOOT);
754 			break;
755 		case Opt_extent_cache:
756 			set_opt(sbi, EXTENT_CACHE);
757 			break;
758 		case Opt_noextent_cache:
759 			clear_opt(sbi, EXTENT_CACHE);
760 			break;
761 		case Opt_noinline_data:
762 			clear_opt(sbi, INLINE_DATA);
763 			break;
764 		case Opt_data_flush:
765 			set_opt(sbi, DATA_FLUSH);
766 			break;
767 		case Opt_reserve_root:
768 			if (args->from && match_int(args, &arg))
769 				return -EINVAL;
770 			if (test_opt(sbi, RESERVE_ROOT)) {
771 				f2fs_info(sbi, "Preserve previous reserve_root=%u",
772 					  F2FS_OPTION(sbi).root_reserved_blocks);
773 			} else {
774 				F2FS_OPTION(sbi).root_reserved_blocks = arg;
775 				set_opt(sbi, RESERVE_ROOT);
776 			}
777 			break;
778 		case Opt_resuid:
779 			if (args->from && match_int(args, &arg))
780 				return -EINVAL;
781 			uid = make_kuid(current_user_ns(), arg);
782 			if (!uid_valid(uid)) {
783 				f2fs_err(sbi, "Invalid uid value %d", arg);
784 				return -EINVAL;
785 			}
786 			F2FS_OPTION(sbi).s_resuid = uid;
787 			break;
788 		case Opt_resgid:
789 			if (args->from && match_int(args, &arg))
790 				return -EINVAL;
791 			gid = make_kgid(current_user_ns(), arg);
792 			if (!gid_valid(gid)) {
793 				f2fs_err(sbi, "Invalid gid value %d", arg);
794 				return -EINVAL;
795 			}
796 			F2FS_OPTION(sbi).s_resgid = gid;
797 			break;
798 		case Opt_mode:
799 			name = match_strdup(&args[0]);
800 
801 			if (!name)
802 				return -ENOMEM;
803 			if (!strcmp(name, "adaptive")) {
804 				if (f2fs_sb_has_blkzoned(sbi)) {
805 					f2fs_warn(sbi, "adaptive mode is not allowed with zoned block device feature");
806 					kfree(name);
807 					return -EINVAL;
808 				}
809 				F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
810 			} else if (!strcmp(name, "lfs")) {
811 				F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
812 			} else {
813 				kfree(name);
814 				return -EINVAL;
815 			}
816 			kfree(name);
817 			break;
818 		case Opt_io_size_bits:
819 			if (args->from && match_int(args, &arg))
820 				return -EINVAL;
821 			if (arg <= 0 || arg > __ilog2_u32(BIO_MAX_VECS)) {
822 				f2fs_warn(sbi, "Not support %d, larger than %d",
823 					  1 << arg, BIO_MAX_VECS);
824 				return -EINVAL;
825 			}
826 			F2FS_OPTION(sbi).write_io_size_bits = arg;
827 			break;
828 #ifdef CONFIG_F2FS_FAULT_INJECTION
829 		case Opt_fault_injection:
830 			if (args->from && match_int(args, &arg))
831 				return -EINVAL;
832 			f2fs_build_fault_attr(sbi, arg, F2FS_ALL_FAULT_TYPE);
833 			set_opt(sbi, FAULT_INJECTION);
834 			break;
835 
836 		case Opt_fault_type:
837 			if (args->from && match_int(args, &arg))
838 				return -EINVAL;
839 			f2fs_build_fault_attr(sbi, 0, arg);
840 			set_opt(sbi, FAULT_INJECTION);
841 			break;
842 #else
843 		case Opt_fault_injection:
844 			f2fs_info(sbi, "fault_injection options not supported");
845 			break;
846 
847 		case Opt_fault_type:
848 			f2fs_info(sbi, "fault_type options not supported");
849 			break;
850 #endif
851 		case Opt_lazytime:
852 			sb->s_flags |= SB_LAZYTIME;
853 			break;
854 		case Opt_nolazytime:
855 			sb->s_flags &= ~SB_LAZYTIME;
856 			break;
857 #ifdef CONFIG_QUOTA
858 		case Opt_quota:
859 		case Opt_usrquota:
860 			set_opt(sbi, USRQUOTA);
861 			break;
862 		case Opt_grpquota:
863 			set_opt(sbi, GRPQUOTA);
864 			break;
865 		case Opt_prjquota:
866 			set_opt(sbi, PRJQUOTA);
867 			break;
868 		case Opt_usrjquota:
869 			ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
870 			if (ret)
871 				return ret;
872 			break;
873 		case Opt_grpjquota:
874 			ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
875 			if (ret)
876 				return ret;
877 			break;
878 		case Opt_prjjquota:
879 			ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
880 			if (ret)
881 				return ret;
882 			break;
883 		case Opt_offusrjquota:
884 			ret = f2fs_clear_qf_name(sb, USRQUOTA);
885 			if (ret)
886 				return ret;
887 			break;
888 		case Opt_offgrpjquota:
889 			ret = f2fs_clear_qf_name(sb, GRPQUOTA);
890 			if (ret)
891 				return ret;
892 			break;
893 		case Opt_offprjjquota:
894 			ret = f2fs_clear_qf_name(sb, PRJQUOTA);
895 			if (ret)
896 				return ret;
897 			break;
898 		case Opt_jqfmt_vfsold:
899 			F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
900 			break;
901 		case Opt_jqfmt_vfsv0:
902 			F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
903 			break;
904 		case Opt_jqfmt_vfsv1:
905 			F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
906 			break;
907 		case Opt_noquota:
908 			clear_opt(sbi, QUOTA);
909 			clear_opt(sbi, USRQUOTA);
910 			clear_opt(sbi, GRPQUOTA);
911 			clear_opt(sbi, PRJQUOTA);
912 			break;
913 #else
914 		case Opt_quota:
915 		case Opt_usrquota:
916 		case Opt_grpquota:
917 		case Opt_prjquota:
918 		case Opt_usrjquota:
919 		case Opt_grpjquota:
920 		case Opt_prjjquota:
921 		case Opt_offusrjquota:
922 		case Opt_offgrpjquota:
923 		case Opt_offprjjquota:
924 		case Opt_jqfmt_vfsold:
925 		case Opt_jqfmt_vfsv0:
926 		case Opt_jqfmt_vfsv1:
927 		case Opt_noquota:
928 			f2fs_info(sbi, "quota operations not supported");
929 			break;
930 #endif
931 		case Opt_whint:
932 			name = match_strdup(&args[0]);
933 			if (!name)
934 				return -ENOMEM;
935 			if (!strcmp(name, "user-based")) {
936 				F2FS_OPTION(sbi).whint_mode = WHINT_MODE_USER;
937 			} else if (!strcmp(name, "off")) {
938 				F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
939 			} else if (!strcmp(name, "fs-based")) {
940 				F2FS_OPTION(sbi).whint_mode = WHINT_MODE_FS;
941 			} else {
942 				kfree(name);
943 				return -EINVAL;
944 			}
945 			kfree(name);
946 			break;
947 		case Opt_alloc:
948 			name = match_strdup(&args[0]);
949 			if (!name)
950 				return -ENOMEM;
951 
952 			if (!strcmp(name, "default")) {
953 				F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
954 			} else if (!strcmp(name, "reuse")) {
955 				F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
956 			} else {
957 				kfree(name);
958 				return -EINVAL;
959 			}
960 			kfree(name);
961 			break;
962 		case Opt_fsync:
963 			name = match_strdup(&args[0]);
964 			if (!name)
965 				return -ENOMEM;
966 			if (!strcmp(name, "posix")) {
967 				F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
968 			} else if (!strcmp(name, "strict")) {
969 				F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_STRICT;
970 			} else if (!strcmp(name, "nobarrier")) {
971 				F2FS_OPTION(sbi).fsync_mode =
972 							FSYNC_MODE_NOBARRIER;
973 			} else {
974 				kfree(name);
975 				return -EINVAL;
976 			}
977 			kfree(name);
978 			break;
979 		case Opt_test_dummy_encryption:
980 			ret = f2fs_set_test_dummy_encryption(sb, p, &args[0],
981 							     is_remount);
982 			if (ret)
983 				return ret;
984 			break;
985 		case Opt_inlinecrypt:
986 #ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
987 			sb->s_flags |= SB_INLINECRYPT;
988 #else
989 			f2fs_info(sbi, "inline encryption not supported");
990 #endif
991 			break;
992 		case Opt_checkpoint_disable_cap_perc:
993 			if (args->from && match_int(args, &arg))
994 				return -EINVAL;
995 			if (arg < 0 || arg > 100)
996 				return -EINVAL;
997 			F2FS_OPTION(sbi).unusable_cap_perc = arg;
998 			set_opt(sbi, DISABLE_CHECKPOINT);
999 			break;
1000 		case Opt_checkpoint_disable_cap:
1001 			if (args->from && match_int(args, &arg))
1002 				return -EINVAL;
1003 			F2FS_OPTION(sbi).unusable_cap = arg;
1004 			set_opt(sbi, DISABLE_CHECKPOINT);
1005 			break;
1006 		case Opt_checkpoint_disable:
1007 			set_opt(sbi, DISABLE_CHECKPOINT);
1008 			break;
1009 		case Opt_checkpoint_enable:
1010 			clear_opt(sbi, DISABLE_CHECKPOINT);
1011 			break;
1012 		case Opt_checkpoint_merge:
1013 			set_opt(sbi, MERGE_CHECKPOINT);
1014 			break;
1015 		case Opt_nocheckpoint_merge:
1016 			clear_opt(sbi, MERGE_CHECKPOINT);
1017 			break;
1018 #ifdef CONFIG_F2FS_FS_COMPRESSION
1019 		case Opt_compress_algorithm:
1020 			if (!f2fs_sb_has_compression(sbi)) {
1021 				f2fs_info(sbi, "Image doesn't support compression");
1022 				break;
1023 			}
1024 			name = match_strdup(&args[0]);
1025 			if (!name)
1026 				return -ENOMEM;
1027 			if (!strcmp(name, "lzo")) {
1028 #ifdef CONFIG_F2FS_FS_LZO
1029 				F2FS_OPTION(sbi).compress_level = 0;
1030 				F2FS_OPTION(sbi).compress_algorithm =
1031 								COMPRESS_LZO;
1032 #else
1033 				f2fs_info(sbi, "kernel doesn't support lzo compression");
1034 #endif
1035 			} else if (!strncmp(name, "lz4", 3)) {
1036 #ifdef CONFIG_F2FS_FS_LZ4
1037 				ret = f2fs_set_lz4hc_level(sbi, name);
1038 				if (ret) {
1039 					kfree(name);
1040 					return -EINVAL;
1041 				}
1042 				F2FS_OPTION(sbi).compress_algorithm =
1043 								COMPRESS_LZ4;
1044 #else
1045 				f2fs_info(sbi, "kernel doesn't support lz4 compression");
1046 #endif
1047 			} else if (!strncmp(name, "zstd", 4)) {
1048 #ifdef CONFIG_F2FS_FS_ZSTD
1049 				ret = f2fs_set_zstd_level(sbi, name);
1050 				if (ret) {
1051 					kfree(name);
1052 					return -EINVAL;
1053 				}
1054 				F2FS_OPTION(sbi).compress_algorithm =
1055 								COMPRESS_ZSTD;
1056 #else
1057 				f2fs_info(sbi, "kernel doesn't support zstd compression");
1058 #endif
1059 			} else if (!strcmp(name, "lzo-rle")) {
1060 #ifdef CONFIG_F2FS_FS_LZORLE
1061 				F2FS_OPTION(sbi).compress_level = 0;
1062 				F2FS_OPTION(sbi).compress_algorithm =
1063 								COMPRESS_LZORLE;
1064 #else
1065 				f2fs_info(sbi, "kernel doesn't support lzorle compression");
1066 #endif
1067 			} else {
1068 				kfree(name);
1069 				return -EINVAL;
1070 			}
1071 			kfree(name);
1072 			break;
1073 		case Opt_compress_log_size:
1074 			if (!f2fs_sb_has_compression(sbi)) {
1075 				f2fs_info(sbi, "Image doesn't support compression");
1076 				break;
1077 			}
1078 			if (args->from && match_int(args, &arg))
1079 				return -EINVAL;
1080 			if (arg < MIN_COMPRESS_LOG_SIZE ||
1081 				arg > MAX_COMPRESS_LOG_SIZE) {
1082 				f2fs_err(sbi,
1083 					"Compress cluster log size is out of range");
1084 				return -EINVAL;
1085 			}
1086 			F2FS_OPTION(sbi).compress_log_size = arg;
1087 			break;
1088 		case Opt_compress_extension:
1089 			if (!f2fs_sb_has_compression(sbi)) {
1090 				f2fs_info(sbi, "Image doesn't support compression");
1091 				break;
1092 			}
1093 			name = match_strdup(&args[0]);
1094 			if (!name)
1095 				return -ENOMEM;
1096 
1097 			ext = F2FS_OPTION(sbi).extensions;
1098 			ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
1099 
1100 			if (strlen(name) >= F2FS_EXTENSION_LEN ||
1101 				ext_cnt >= COMPRESS_EXT_NUM) {
1102 				f2fs_err(sbi,
1103 					"invalid extension length/number");
1104 				kfree(name);
1105 				return -EINVAL;
1106 			}
1107 
1108 			strcpy(ext[ext_cnt], name);
1109 			F2FS_OPTION(sbi).compress_ext_cnt++;
1110 			kfree(name);
1111 			break;
1112 		case Opt_nocompress_extension:
1113 			if (!f2fs_sb_has_compression(sbi)) {
1114 				f2fs_info(sbi, "Image doesn't support compression");
1115 				break;
1116 			}
1117 			name = match_strdup(&args[0]);
1118 			if (!name)
1119 				return -ENOMEM;
1120 
1121 			noext = F2FS_OPTION(sbi).noextensions;
1122 			noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
1123 
1124 			if (strlen(name) >= F2FS_EXTENSION_LEN ||
1125 				noext_cnt >= COMPRESS_EXT_NUM) {
1126 				f2fs_err(sbi,
1127 					"invalid extension length/number");
1128 				kfree(name);
1129 				return -EINVAL;
1130 			}
1131 
1132 			strcpy(noext[noext_cnt], name);
1133 			F2FS_OPTION(sbi).nocompress_ext_cnt++;
1134 			kfree(name);
1135 			break;
1136 		case Opt_compress_chksum:
1137 			F2FS_OPTION(sbi).compress_chksum = true;
1138 			break;
1139 		case Opt_compress_mode:
1140 			name = match_strdup(&args[0]);
1141 			if (!name)
1142 				return -ENOMEM;
1143 			if (!strcmp(name, "fs")) {
1144 				F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
1145 			} else if (!strcmp(name, "user")) {
1146 				F2FS_OPTION(sbi).compress_mode = COMPR_MODE_USER;
1147 			} else {
1148 				kfree(name);
1149 				return -EINVAL;
1150 			}
1151 			kfree(name);
1152 			break;
1153 		case Opt_compress_cache:
1154 			set_opt(sbi, COMPRESS_CACHE);
1155 			break;
1156 #else
1157 		case Opt_compress_algorithm:
1158 		case Opt_compress_log_size:
1159 		case Opt_compress_extension:
1160 		case Opt_nocompress_extension:
1161 		case Opt_compress_chksum:
1162 		case Opt_compress_mode:
1163 		case Opt_compress_cache:
1164 			f2fs_info(sbi, "compression options not supported");
1165 			break;
1166 #endif
1167 		case Opt_atgc:
1168 			set_opt(sbi, ATGC);
1169 			break;
1170 		case Opt_gc_merge:
1171 			set_opt(sbi, GC_MERGE);
1172 			break;
1173 		case Opt_nogc_merge:
1174 			clear_opt(sbi, GC_MERGE);
1175 			break;
1176 		default:
1177 			f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
1178 				 p);
1179 			return -EINVAL;
1180 		}
1181 	}
1182 default_check:
1183 #ifdef CONFIG_QUOTA
1184 	if (f2fs_check_quota_options(sbi))
1185 		return -EINVAL;
1186 #else
1187 	if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sbi->sb)) {
1188 		f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1189 		return -EINVAL;
1190 	}
1191 	if (f2fs_sb_has_project_quota(sbi) && !f2fs_readonly(sbi->sb)) {
1192 		f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1193 		return -EINVAL;
1194 	}
1195 #endif
1196 #ifndef CONFIG_UNICODE
1197 	if (f2fs_sb_has_casefold(sbi)) {
1198 		f2fs_err(sbi,
1199 			"Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
1200 		return -EINVAL;
1201 	}
1202 #endif
1203 	/*
1204 	 * The BLKZONED feature indicates that the drive was formatted with
1205 	 * zone alignment optimization. This is optional for host-aware
1206 	 * devices, but mandatory for host-managed zoned block devices.
1207 	 */
1208 #ifndef CONFIG_BLK_DEV_ZONED
1209 	if (f2fs_sb_has_blkzoned(sbi)) {
1210 		f2fs_err(sbi, "Zoned block device support is not enabled");
1211 		return -EINVAL;
1212 	}
1213 #endif
1214 
1215 #ifdef CONFIG_F2FS_FS_COMPRESSION
1216 	if (f2fs_test_compress_extension(sbi)) {
1217 		f2fs_err(sbi, "invalid compress or nocompress extension");
1218 		return -EINVAL;
1219 	}
1220 #endif
1221 
1222 	if (F2FS_IO_SIZE_BITS(sbi) && !f2fs_lfs_mode(sbi)) {
1223 		f2fs_err(sbi, "Should set mode=lfs with %uKB-sized IO",
1224 			 F2FS_IO_SIZE_KB(sbi));
1225 		return -EINVAL;
1226 	}
1227 
1228 	if (test_opt(sbi, INLINE_XATTR_SIZE)) {
1229 		int min_size, max_size;
1230 
1231 		if (!f2fs_sb_has_extra_attr(sbi) ||
1232 			!f2fs_sb_has_flexible_inline_xattr(sbi)) {
1233 			f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
1234 			return -EINVAL;
1235 		}
1236 		if (!test_opt(sbi, INLINE_XATTR)) {
1237 			f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
1238 			return -EINVAL;
1239 		}
1240 
1241 		min_size = sizeof(struct f2fs_xattr_header) / sizeof(__le32);
1242 		max_size = MAX_INLINE_XATTR_SIZE;
1243 
1244 		if (F2FS_OPTION(sbi).inline_xattr_size < min_size ||
1245 				F2FS_OPTION(sbi).inline_xattr_size > max_size) {
1246 			f2fs_err(sbi, "inline xattr size is out of range: %d ~ %d",
1247 				 min_size, max_size);
1248 			return -EINVAL;
1249 		}
1250 	}
1251 
1252 	if (test_opt(sbi, DISABLE_CHECKPOINT) && f2fs_lfs_mode(sbi)) {
1253 		f2fs_err(sbi, "LFS not compatible with checkpoint=disable");
1254 		return -EINVAL;
1255 	}
1256 
1257 	/* Not pass down write hints if the number of active logs is lesser
1258 	 * than NR_CURSEG_PERSIST_TYPE.
1259 	 */
1260 	if (F2FS_OPTION(sbi).active_logs != NR_CURSEG_TYPE)
1261 		F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1262 
1263 	if (f2fs_sb_has_readonly(sbi) && !f2fs_readonly(sbi->sb)) {
1264 		f2fs_err(sbi, "Allow to mount readonly mode only");
1265 		return -EROFS;
1266 	}
1267 	return 0;
1268 }
1269 
1270 static struct inode *f2fs_alloc_inode(struct super_block *sb)
1271 {
1272 	struct f2fs_inode_info *fi;
1273 
1274 	fi = kmem_cache_alloc(f2fs_inode_cachep, GFP_F2FS_ZERO);
1275 	if (!fi)
1276 		return NULL;
1277 
1278 	init_once((void *) fi);
1279 
1280 	/* Initialize f2fs-specific inode info */
1281 	atomic_set(&fi->dirty_pages, 0);
1282 	atomic_set(&fi->i_compr_blocks, 0);
1283 	init_rwsem(&fi->i_sem);
1284 	spin_lock_init(&fi->i_size_lock);
1285 	INIT_LIST_HEAD(&fi->dirty_list);
1286 	INIT_LIST_HEAD(&fi->gdirty_list);
1287 	INIT_LIST_HEAD(&fi->inmem_ilist);
1288 	INIT_LIST_HEAD(&fi->inmem_pages);
1289 	mutex_init(&fi->inmem_lock);
1290 	init_rwsem(&fi->i_gc_rwsem[READ]);
1291 	init_rwsem(&fi->i_gc_rwsem[WRITE]);
1292 	init_rwsem(&fi->i_xattr_sem);
1293 
1294 	/* Will be used by directory only */
1295 	fi->i_dir_level = F2FS_SB(sb)->dir_level;
1296 
1297 	return &fi->vfs_inode;
1298 }
1299 
1300 static int f2fs_drop_inode(struct inode *inode)
1301 {
1302 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1303 	int ret;
1304 
1305 	/*
1306 	 * during filesystem shutdown, if checkpoint is disabled,
1307 	 * drop useless meta/node dirty pages.
1308 	 */
1309 	if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1310 		if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1311 			inode->i_ino == F2FS_META_INO(sbi)) {
1312 			trace_f2fs_drop_inode(inode, 1);
1313 			return 1;
1314 		}
1315 	}
1316 
1317 	/*
1318 	 * This is to avoid a deadlock condition like below.
1319 	 * writeback_single_inode(inode)
1320 	 *  - f2fs_write_data_page
1321 	 *    - f2fs_gc -> iput -> evict
1322 	 *       - inode_wait_for_writeback(inode)
1323 	 */
1324 	if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
1325 		if (!inode->i_nlink && !is_bad_inode(inode)) {
1326 			/* to avoid evict_inode call simultaneously */
1327 			atomic_inc(&inode->i_count);
1328 			spin_unlock(&inode->i_lock);
1329 
1330 			/* some remained atomic pages should discarded */
1331 			if (f2fs_is_atomic_file(inode))
1332 				f2fs_drop_inmem_pages(inode);
1333 
1334 			/* should remain fi->extent_tree for writepage */
1335 			f2fs_destroy_extent_node(inode);
1336 
1337 			sb_start_intwrite(inode->i_sb);
1338 			f2fs_i_size_write(inode, 0);
1339 
1340 			f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1341 					inode, NULL, 0, DATA);
1342 			truncate_inode_pages_final(inode->i_mapping);
1343 
1344 			if (F2FS_HAS_BLOCKS(inode))
1345 				f2fs_truncate(inode);
1346 
1347 			sb_end_intwrite(inode->i_sb);
1348 
1349 			spin_lock(&inode->i_lock);
1350 			atomic_dec(&inode->i_count);
1351 		}
1352 		trace_f2fs_drop_inode(inode, 0);
1353 		return 0;
1354 	}
1355 	ret = generic_drop_inode(inode);
1356 	if (!ret)
1357 		ret = fscrypt_drop_inode(inode);
1358 	trace_f2fs_drop_inode(inode, ret);
1359 	return ret;
1360 }
1361 
1362 int f2fs_inode_dirtied(struct inode *inode, bool sync)
1363 {
1364 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1365 	int ret = 0;
1366 
1367 	spin_lock(&sbi->inode_lock[DIRTY_META]);
1368 	if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1369 		ret = 1;
1370 	} else {
1371 		set_inode_flag(inode, FI_DIRTY_INODE);
1372 		stat_inc_dirty_inode(sbi, DIRTY_META);
1373 	}
1374 	if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1375 		list_add_tail(&F2FS_I(inode)->gdirty_list,
1376 				&sbi->inode_list[DIRTY_META]);
1377 		inc_page_count(sbi, F2FS_DIRTY_IMETA);
1378 	}
1379 	spin_unlock(&sbi->inode_lock[DIRTY_META]);
1380 	return ret;
1381 }
1382 
1383 void f2fs_inode_synced(struct inode *inode)
1384 {
1385 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1386 
1387 	spin_lock(&sbi->inode_lock[DIRTY_META]);
1388 	if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1389 		spin_unlock(&sbi->inode_lock[DIRTY_META]);
1390 		return;
1391 	}
1392 	if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1393 		list_del_init(&F2FS_I(inode)->gdirty_list);
1394 		dec_page_count(sbi, F2FS_DIRTY_IMETA);
1395 	}
1396 	clear_inode_flag(inode, FI_DIRTY_INODE);
1397 	clear_inode_flag(inode, FI_AUTO_RECOVER);
1398 	stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1399 	spin_unlock(&sbi->inode_lock[DIRTY_META]);
1400 }
1401 
1402 /*
1403  * f2fs_dirty_inode() is called from __mark_inode_dirty()
1404  *
1405  * We should call set_dirty_inode to write the dirty inode through write_inode.
1406  */
1407 static void f2fs_dirty_inode(struct inode *inode, int flags)
1408 {
1409 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1410 
1411 	if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1412 			inode->i_ino == F2FS_META_INO(sbi))
1413 		return;
1414 
1415 	if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1416 		clear_inode_flag(inode, FI_AUTO_RECOVER);
1417 
1418 	f2fs_inode_dirtied(inode, false);
1419 }
1420 
1421 static void f2fs_free_inode(struct inode *inode)
1422 {
1423 	fscrypt_free_inode(inode);
1424 	kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1425 }
1426 
1427 static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1428 {
1429 	percpu_counter_destroy(&sbi->alloc_valid_block_count);
1430 	percpu_counter_destroy(&sbi->total_valid_inode_count);
1431 }
1432 
1433 static void destroy_device_list(struct f2fs_sb_info *sbi)
1434 {
1435 	int i;
1436 
1437 	for (i = 0; i < sbi->s_ndevs; i++) {
1438 		blkdev_put(FDEV(i).bdev, FMODE_EXCL);
1439 #ifdef CONFIG_BLK_DEV_ZONED
1440 		kvfree(FDEV(i).blkz_seq);
1441 		kfree(FDEV(i).zone_capacity_blocks);
1442 #endif
1443 	}
1444 	kvfree(sbi->devs);
1445 }
1446 
1447 static void f2fs_put_super(struct super_block *sb)
1448 {
1449 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1450 	int i;
1451 	bool dropped;
1452 
1453 	/* unregister procfs/sysfs entries in advance to avoid race case */
1454 	f2fs_unregister_sysfs(sbi);
1455 
1456 	f2fs_quota_off_umount(sb);
1457 
1458 	/* prevent remaining shrinker jobs */
1459 	mutex_lock(&sbi->umount_mutex);
1460 
1461 	/*
1462 	 * flush all issued checkpoints and stop checkpoint issue thread.
1463 	 * after then, all checkpoints should be done by each process context.
1464 	 */
1465 	f2fs_stop_ckpt_thread(sbi);
1466 
1467 	/*
1468 	 * We don't need to do checkpoint when superblock is clean.
1469 	 * But, the previous checkpoint was not done by umount, it needs to do
1470 	 * clean checkpoint again.
1471 	 */
1472 	if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1473 			!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1474 		struct cp_control cpc = {
1475 			.reason = CP_UMOUNT,
1476 		};
1477 		f2fs_write_checkpoint(sbi, &cpc);
1478 	}
1479 
1480 	/* be sure to wait for any on-going discard commands */
1481 	dropped = f2fs_issue_discard_timeout(sbi);
1482 
1483 	if ((f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi)) &&
1484 					!sbi->discard_blks && !dropped) {
1485 		struct cp_control cpc = {
1486 			.reason = CP_UMOUNT | CP_TRIMMED,
1487 		};
1488 		f2fs_write_checkpoint(sbi, &cpc);
1489 	}
1490 
1491 	/*
1492 	 * normally superblock is clean, so we need to release this.
1493 	 * In addition, EIO will skip do checkpoint, we need this as well.
1494 	 */
1495 	f2fs_release_ino_entry(sbi, true);
1496 
1497 	f2fs_leave_shrinker(sbi);
1498 	mutex_unlock(&sbi->umount_mutex);
1499 
1500 	/* our cp_error case, we can wait for any writeback page */
1501 	f2fs_flush_merged_writes(sbi);
1502 
1503 	f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1504 
1505 	f2fs_bug_on(sbi, sbi->fsync_node_num);
1506 
1507 	f2fs_destroy_compress_inode(sbi);
1508 
1509 	iput(sbi->node_inode);
1510 	sbi->node_inode = NULL;
1511 
1512 	iput(sbi->meta_inode);
1513 	sbi->meta_inode = NULL;
1514 
1515 	/*
1516 	 * iput() can update stat information, if f2fs_write_checkpoint()
1517 	 * above failed with error.
1518 	 */
1519 	f2fs_destroy_stats(sbi);
1520 
1521 	/* destroy f2fs internal modules */
1522 	f2fs_destroy_node_manager(sbi);
1523 	f2fs_destroy_segment_manager(sbi);
1524 
1525 	f2fs_destroy_post_read_wq(sbi);
1526 
1527 	kvfree(sbi->ckpt);
1528 
1529 	sb->s_fs_info = NULL;
1530 	if (sbi->s_chksum_driver)
1531 		crypto_free_shash(sbi->s_chksum_driver);
1532 	kfree(sbi->raw_super);
1533 
1534 	destroy_device_list(sbi);
1535 	f2fs_destroy_page_array_cache(sbi);
1536 	f2fs_destroy_xattr_caches(sbi);
1537 	mempool_destroy(sbi->write_io_dummy);
1538 #ifdef CONFIG_QUOTA
1539 	for (i = 0; i < MAXQUOTAS; i++)
1540 		kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1541 #endif
1542 	fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
1543 	destroy_percpu_info(sbi);
1544 	for (i = 0; i < NR_PAGE_TYPE; i++)
1545 		kvfree(sbi->write_io[i]);
1546 #ifdef CONFIG_UNICODE
1547 	utf8_unload(sb->s_encoding);
1548 #endif
1549 	kfree(sbi);
1550 }
1551 
1552 int f2fs_sync_fs(struct super_block *sb, int sync)
1553 {
1554 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1555 	int err = 0;
1556 
1557 	if (unlikely(f2fs_cp_error(sbi)))
1558 		return 0;
1559 	if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1560 		return 0;
1561 
1562 	trace_f2fs_sync_fs(sb, sync);
1563 
1564 	if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1565 		return -EAGAIN;
1566 
1567 	if (sync)
1568 		err = f2fs_issue_checkpoint(sbi);
1569 
1570 	return err;
1571 }
1572 
1573 static int f2fs_freeze(struct super_block *sb)
1574 {
1575 	if (f2fs_readonly(sb))
1576 		return 0;
1577 
1578 	/* IO error happened before */
1579 	if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1580 		return -EIO;
1581 
1582 	/* must be clean, since sync_filesystem() was already called */
1583 	if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1584 		return -EINVAL;
1585 
1586 	/* ensure no checkpoint required */
1587 	if (!llist_empty(&F2FS_SB(sb)->cprc_info.issue_list))
1588 		return -EINVAL;
1589 	return 0;
1590 }
1591 
1592 static int f2fs_unfreeze(struct super_block *sb)
1593 {
1594 	return 0;
1595 }
1596 
1597 #ifdef CONFIG_QUOTA
1598 static int f2fs_statfs_project(struct super_block *sb,
1599 				kprojid_t projid, struct kstatfs *buf)
1600 {
1601 	struct kqid qid;
1602 	struct dquot *dquot;
1603 	u64 limit;
1604 	u64 curblock;
1605 
1606 	qid = make_kqid_projid(projid);
1607 	dquot = dqget(sb, qid);
1608 	if (IS_ERR(dquot))
1609 		return PTR_ERR(dquot);
1610 	spin_lock(&dquot->dq_dqb_lock);
1611 
1612 	limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1613 					dquot->dq_dqb.dqb_bhardlimit);
1614 	if (limit)
1615 		limit >>= sb->s_blocksize_bits;
1616 
1617 	if (limit && buf->f_blocks > limit) {
1618 		curblock = (dquot->dq_dqb.dqb_curspace +
1619 			    dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
1620 		buf->f_blocks = limit;
1621 		buf->f_bfree = buf->f_bavail =
1622 			(buf->f_blocks > curblock) ?
1623 			 (buf->f_blocks - curblock) : 0;
1624 	}
1625 
1626 	limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1627 					dquot->dq_dqb.dqb_ihardlimit);
1628 
1629 	if (limit && buf->f_files > limit) {
1630 		buf->f_files = limit;
1631 		buf->f_ffree =
1632 			(buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1633 			 (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1634 	}
1635 
1636 	spin_unlock(&dquot->dq_dqb_lock);
1637 	dqput(dquot);
1638 	return 0;
1639 }
1640 #endif
1641 
1642 static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1643 {
1644 	struct super_block *sb = dentry->d_sb;
1645 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1646 	u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
1647 	block_t total_count, user_block_count, start_count;
1648 	u64 avail_node_count;
1649 
1650 	total_count = le64_to_cpu(sbi->raw_super->block_count);
1651 	user_block_count = sbi->user_block_count;
1652 	start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
1653 	buf->f_type = F2FS_SUPER_MAGIC;
1654 	buf->f_bsize = sbi->blocksize;
1655 
1656 	buf->f_blocks = total_count - start_count;
1657 	buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1658 						sbi->current_reserved_blocks;
1659 
1660 	spin_lock(&sbi->stat_lock);
1661 	if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1662 		buf->f_bfree = 0;
1663 	else
1664 		buf->f_bfree -= sbi->unusable_block_count;
1665 	spin_unlock(&sbi->stat_lock);
1666 
1667 	if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1668 		buf->f_bavail = buf->f_bfree -
1669 				F2FS_OPTION(sbi).root_reserved_blocks;
1670 	else
1671 		buf->f_bavail = 0;
1672 
1673 	avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
1674 
1675 	if (avail_node_count > user_block_count) {
1676 		buf->f_files = user_block_count;
1677 		buf->f_ffree = buf->f_bavail;
1678 	} else {
1679 		buf->f_files = avail_node_count;
1680 		buf->f_ffree = min(avail_node_count - valid_node_count(sbi),
1681 					buf->f_bavail);
1682 	}
1683 
1684 	buf->f_namelen = F2FS_NAME_LEN;
1685 	buf->f_fsid    = u64_to_fsid(id);
1686 
1687 #ifdef CONFIG_QUOTA
1688 	if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1689 			sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1690 		f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1691 	}
1692 #endif
1693 	return 0;
1694 }
1695 
1696 static inline void f2fs_show_quota_options(struct seq_file *seq,
1697 					   struct super_block *sb)
1698 {
1699 #ifdef CONFIG_QUOTA
1700 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1701 
1702 	if (F2FS_OPTION(sbi).s_jquota_fmt) {
1703 		char *fmtname = "";
1704 
1705 		switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1706 		case QFMT_VFS_OLD:
1707 			fmtname = "vfsold";
1708 			break;
1709 		case QFMT_VFS_V0:
1710 			fmtname = "vfsv0";
1711 			break;
1712 		case QFMT_VFS_V1:
1713 			fmtname = "vfsv1";
1714 			break;
1715 		}
1716 		seq_printf(seq, ",jqfmt=%s", fmtname);
1717 	}
1718 
1719 	if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1720 		seq_show_option(seq, "usrjquota",
1721 			F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1722 
1723 	if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1724 		seq_show_option(seq, "grpjquota",
1725 			F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1726 
1727 	if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1728 		seq_show_option(seq, "prjjquota",
1729 			F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1730 #endif
1731 }
1732 
1733 #ifdef CONFIG_F2FS_FS_COMPRESSION
1734 static inline void f2fs_show_compress_options(struct seq_file *seq,
1735 							struct super_block *sb)
1736 {
1737 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
1738 	char *algtype = "";
1739 	int i;
1740 
1741 	if (!f2fs_sb_has_compression(sbi))
1742 		return;
1743 
1744 	switch (F2FS_OPTION(sbi).compress_algorithm) {
1745 	case COMPRESS_LZO:
1746 		algtype = "lzo";
1747 		break;
1748 	case COMPRESS_LZ4:
1749 		algtype = "lz4";
1750 		break;
1751 	case COMPRESS_ZSTD:
1752 		algtype = "zstd";
1753 		break;
1754 	case COMPRESS_LZORLE:
1755 		algtype = "lzo-rle";
1756 		break;
1757 	}
1758 	seq_printf(seq, ",compress_algorithm=%s", algtype);
1759 
1760 	if (F2FS_OPTION(sbi).compress_level)
1761 		seq_printf(seq, ":%d", F2FS_OPTION(sbi).compress_level);
1762 
1763 	seq_printf(seq, ",compress_log_size=%u",
1764 			F2FS_OPTION(sbi).compress_log_size);
1765 
1766 	for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1767 		seq_printf(seq, ",compress_extension=%s",
1768 			F2FS_OPTION(sbi).extensions[i]);
1769 	}
1770 
1771 	for (i = 0; i < F2FS_OPTION(sbi).nocompress_ext_cnt; i++) {
1772 		seq_printf(seq, ",nocompress_extension=%s",
1773 			F2FS_OPTION(sbi).noextensions[i]);
1774 	}
1775 
1776 	if (F2FS_OPTION(sbi).compress_chksum)
1777 		seq_puts(seq, ",compress_chksum");
1778 
1779 	if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_FS)
1780 		seq_printf(seq, ",compress_mode=%s", "fs");
1781 	else if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_USER)
1782 		seq_printf(seq, ",compress_mode=%s", "user");
1783 
1784 	if (test_opt(sbi, COMPRESS_CACHE))
1785 		seq_puts(seq, ",compress_cache");
1786 }
1787 #endif
1788 
1789 static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1790 {
1791 	struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1792 
1793 	if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1794 		seq_printf(seq, ",background_gc=%s", "sync");
1795 	else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1796 		seq_printf(seq, ",background_gc=%s", "on");
1797 	else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1798 		seq_printf(seq, ",background_gc=%s", "off");
1799 
1800 	if (test_opt(sbi, GC_MERGE))
1801 		seq_puts(seq, ",gc_merge");
1802 
1803 	if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1804 		seq_puts(seq, ",disable_roll_forward");
1805 	if (test_opt(sbi, NORECOVERY))
1806 		seq_puts(seq, ",norecovery");
1807 	if (test_opt(sbi, DISCARD))
1808 		seq_puts(seq, ",discard");
1809 	else
1810 		seq_puts(seq, ",nodiscard");
1811 	if (test_opt(sbi, NOHEAP))
1812 		seq_puts(seq, ",no_heap");
1813 	else
1814 		seq_puts(seq, ",heap");
1815 #ifdef CONFIG_F2FS_FS_XATTR
1816 	if (test_opt(sbi, XATTR_USER))
1817 		seq_puts(seq, ",user_xattr");
1818 	else
1819 		seq_puts(seq, ",nouser_xattr");
1820 	if (test_opt(sbi, INLINE_XATTR))
1821 		seq_puts(seq, ",inline_xattr");
1822 	else
1823 		seq_puts(seq, ",noinline_xattr");
1824 	if (test_opt(sbi, INLINE_XATTR_SIZE))
1825 		seq_printf(seq, ",inline_xattr_size=%u",
1826 					F2FS_OPTION(sbi).inline_xattr_size);
1827 #endif
1828 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1829 	if (test_opt(sbi, POSIX_ACL))
1830 		seq_puts(seq, ",acl");
1831 	else
1832 		seq_puts(seq, ",noacl");
1833 #endif
1834 	if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
1835 		seq_puts(seq, ",disable_ext_identify");
1836 	if (test_opt(sbi, INLINE_DATA))
1837 		seq_puts(seq, ",inline_data");
1838 	else
1839 		seq_puts(seq, ",noinline_data");
1840 	if (test_opt(sbi, INLINE_DENTRY))
1841 		seq_puts(seq, ",inline_dentry");
1842 	else
1843 		seq_puts(seq, ",noinline_dentry");
1844 	if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
1845 		seq_puts(seq, ",flush_merge");
1846 	if (test_opt(sbi, NOBARRIER))
1847 		seq_puts(seq, ",nobarrier");
1848 	if (test_opt(sbi, FASTBOOT))
1849 		seq_puts(seq, ",fastboot");
1850 	if (test_opt(sbi, EXTENT_CACHE))
1851 		seq_puts(seq, ",extent_cache");
1852 	else
1853 		seq_puts(seq, ",noextent_cache");
1854 	if (test_opt(sbi, DATA_FLUSH))
1855 		seq_puts(seq, ",data_flush");
1856 
1857 	seq_puts(seq, ",mode=");
1858 	if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
1859 		seq_puts(seq, "adaptive");
1860 	else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
1861 		seq_puts(seq, "lfs");
1862 	seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
1863 	if (test_opt(sbi, RESERVE_ROOT))
1864 		seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
1865 				F2FS_OPTION(sbi).root_reserved_blocks,
1866 				from_kuid_munged(&init_user_ns,
1867 					F2FS_OPTION(sbi).s_resuid),
1868 				from_kgid_munged(&init_user_ns,
1869 					F2FS_OPTION(sbi).s_resgid));
1870 	if (F2FS_IO_SIZE_BITS(sbi))
1871 		seq_printf(seq, ",io_bits=%u",
1872 				F2FS_OPTION(sbi).write_io_size_bits);
1873 #ifdef CONFIG_F2FS_FAULT_INJECTION
1874 	if (test_opt(sbi, FAULT_INJECTION)) {
1875 		seq_printf(seq, ",fault_injection=%u",
1876 				F2FS_OPTION(sbi).fault_info.inject_rate);
1877 		seq_printf(seq, ",fault_type=%u",
1878 				F2FS_OPTION(sbi).fault_info.inject_type);
1879 	}
1880 #endif
1881 #ifdef CONFIG_QUOTA
1882 	if (test_opt(sbi, QUOTA))
1883 		seq_puts(seq, ",quota");
1884 	if (test_opt(sbi, USRQUOTA))
1885 		seq_puts(seq, ",usrquota");
1886 	if (test_opt(sbi, GRPQUOTA))
1887 		seq_puts(seq, ",grpquota");
1888 	if (test_opt(sbi, PRJQUOTA))
1889 		seq_puts(seq, ",prjquota");
1890 #endif
1891 	f2fs_show_quota_options(seq, sbi->sb);
1892 	if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_USER)
1893 		seq_printf(seq, ",whint_mode=%s", "user-based");
1894 	else if (F2FS_OPTION(sbi).whint_mode == WHINT_MODE_FS)
1895 		seq_printf(seq, ",whint_mode=%s", "fs-based");
1896 
1897 	fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
1898 
1899 	if (sbi->sb->s_flags & SB_INLINECRYPT)
1900 		seq_puts(seq, ",inlinecrypt");
1901 
1902 	if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
1903 		seq_printf(seq, ",alloc_mode=%s", "default");
1904 	else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
1905 		seq_printf(seq, ",alloc_mode=%s", "reuse");
1906 
1907 	if (test_opt(sbi, DISABLE_CHECKPOINT))
1908 		seq_printf(seq, ",checkpoint=disable:%u",
1909 				F2FS_OPTION(sbi).unusable_cap);
1910 	if (test_opt(sbi, MERGE_CHECKPOINT))
1911 		seq_puts(seq, ",checkpoint_merge");
1912 	else
1913 		seq_puts(seq, ",nocheckpoint_merge");
1914 	if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
1915 		seq_printf(seq, ",fsync_mode=%s", "posix");
1916 	else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
1917 		seq_printf(seq, ",fsync_mode=%s", "strict");
1918 	else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
1919 		seq_printf(seq, ",fsync_mode=%s", "nobarrier");
1920 
1921 #ifdef CONFIG_F2FS_FS_COMPRESSION
1922 	f2fs_show_compress_options(seq, sbi->sb);
1923 #endif
1924 
1925 	if (test_opt(sbi, ATGC))
1926 		seq_puts(seq, ",atgc");
1927 	return 0;
1928 }
1929 
1930 static void default_options(struct f2fs_sb_info *sbi)
1931 {
1932 	/* init some FS parameters */
1933 	if (f2fs_sb_has_readonly(sbi))
1934 		F2FS_OPTION(sbi).active_logs = NR_CURSEG_RO_TYPE;
1935 	else
1936 		F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
1937 
1938 	F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
1939 	F2FS_OPTION(sbi).whint_mode = WHINT_MODE_OFF;
1940 	F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
1941 	F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
1942 	F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
1943 	F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
1944 	F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
1945 	F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
1946 	F2FS_OPTION(sbi).compress_ext_cnt = 0;
1947 	F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
1948 	F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
1949 
1950 	sbi->sb->s_flags &= ~SB_INLINECRYPT;
1951 
1952 	set_opt(sbi, INLINE_XATTR);
1953 	set_opt(sbi, INLINE_DATA);
1954 	set_opt(sbi, INLINE_DENTRY);
1955 	set_opt(sbi, EXTENT_CACHE);
1956 	set_opt(sbi, NOHEAP);
1957 	clear_opt(sbi, DISABLE_CHECKPOINT);
1958 	set_opt(sbi, MERGE_CHECKPOINT);
1959 	F2FS_OPTION(sbi).unusable_cap = 0;
1960 	sbi->sb->s_flags |= SB_LAZYTIME;
1961 	set_opt(sbi, FLUSH_MERGE);
1962 	set_opt(sbi, DISCARD);
1963 	if (f2fs_sb_has_blkzoned(sbi))
1964 		F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
1965 	else
1966 		F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
1967 
1968 #ifdef CONFIG_F2FS_FS_XATTR
1969 	set_opt(sbi, XATTR_USER);
1970 #endif
1971 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1972 	set_opt(sbi, POSIX_ACL);
1973 #endif
1974 
1975 	f2fs_build_fault_attr(sbi, 0, 0);
1976 }
1977 
1978 #ifdef CONFIG_QUOTA
1979 static int f2fs_enable_quotas(struct super_block *sb);
1980 #endif
1981 
1982 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
1983 {
1984 	unsigned int s_flags = sbi->sb->s_flags;
1985 	struct cp_control cpc;
1986 	int err = 0;
1987 	int ret;
1988 	block_t unusable;
1989 
1990 	if (s_flags & SB_RDONLY) {
1991 		f2fs_err(sbi, "checkpoint=disable on readonly fs");
1992 		return -EINVAL;
1993 	}
1994 	sbi->sb->s_flags |= SB_ACTIVE;
1995 
1996 	f2fs_update_time(sbi, DISABLE_TIME);
1997 
1998 	while (!f2fs_time_over(sbi, DISABLE_TIME)) {
1999 		down_write(&sbi->gc_lock);
2000 		err = f2fs_gc(sbi, true, false, false, NULL_SEGNO);
2001 		if (err == -ENODATA) {
2002 			err = 0;
2003 			break;
2004 		}
2005 		if (err && err != -EAGAIN)
2006 			break;
2007 	}
2008 
2009 	ret = sync_filesystem(sbi->sb);
2010 	if (ret || err) {
2011 		err = ret ? ret : err;
2012 		goto restore_flag;
2013 	}
2014 
2015 	unusable = f2fs_get_unusable_blocks(sbi);
2016 	if (f2fs_disable_cp_again(sbi, unusable)) {
2017 		err = -EAGAIN;
2018 		goto restore_flag;
2019 	}
2020 
2021 	down_write(&sbi->gc_lock);
2022 	cpc.reason = CP_PAUSE;
2023 	set_sbi_flag(sbi, SBI_CP_DISABLED);
2024 	err = f2fs_write_checkpoint(sbi, &cpc);
2025 	if (err)
2026 		goto out_unlock;
2027 
2028 	spin_lock(&sbi->stat_lock);
2029 	sbi->unusable_block_count = unusable;
2030 	spin_unlock(&sbi->stat_lock);
2031 
2032 out_unlock:
2033 	up_write(&sbi->gc_lock);
2034 restore_flag:
2035 	sbi->sb->s_flags = s_flags;	/* Restore SB_RDONLY status */
2036 	return err;
2037 }
2038 
2039 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
2040 {
2041 	/* we should flush all the data to keep data consistency */
2042 	sync_inodes_sb(sbi->sb);
2043 
2044 	down_write(&sbi->gc_lock);
2045 	f2fs_dirty_to_prefree(sbi);
2046 
2047 	clear_sbi_flag(sbi, SBI_CP_DISABLED);
2048 	set_sbi_flag(sbi, SBI_IS_DIRTY);
2049 	up_write(&sbi->gc_lock);
2050 
2051 	f2fs_sync_fs(sbi->sb, 1);
2052 }
2053 
2054 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
2055 {
2056 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2057 	struct f2fs_mount_info org_mount_opt;
2058 	unsigned long old_sb_flags;
2059 	int err;
2060 	bool need_restart_gc = false, need_stop_gc = false;
2061 	bool need_restart_ckpt = false, need_stop_ckpt = false;
2062 	bool need_restart_flush = false, need_stop_flush = false;
2063 	bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
2064 	bool disable_checkpoint = test_opt(sbi, DISABLE_CHECKPOINT);
2065 	bool no_io_align = !F2FS_IO_ALIGNED(sbi);
2066 	bool no_atgc = !test_opt(sbi, ATGC);
2067 	bool no_compress_cache = !test_opt(sbi, COMPRESS_CACHE);
2068 	bool checkpoint_changed;
2069 #ifdef CONFIG_QUOTA
2070 	int i, j;
2071 #endif
2072 
2073 	/*
2074 	 * Save the old mount options in case we
2075 	 * need to restore them.
2076 	 */
2077 	org_mount_opt = sbi->mount_opt;
2078 	old_sb_flags = sb->s_flags;
2079 
2080 #ifdef CONFIG_QUOTA
2081 	org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
2082 	for (i = 0; i < MAXQUOTAS; i++) {
2083 		if (F2FS_OPTION(sbi).s_qf_names[i]) {
2084 			org_mount_opt.s_qf_names[i] =
2085 				kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
2086 				GFP_KERNEL);
2087 			if (!org_mount_opt.s_qf_names[i]) {
2088 				for (j = 0; j < i; j++)
2089 					kfree(org_mount_opt.s_qf_names[j]);
2090 				return -ENOMEM;
2091 			}
2092 		} else {
2093 			org_mount_opt.s_qf_names[i] = NULL;
2094 		}
2095 	}
2096 #endif
2097 
2098 	/* recover superblocks we couldn't write due to previous RO mount */
2099 	if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
2100 		err = f2fs_commit_super(sbi, false);
2101 		f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
2102 			  err);
2103 		if (!err)
2104 			clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2105 	}
2106 
2107 	default_options(sbi);
2108 
2109 	/* parse mount options */
2110 	err = parse_options(sb, data, true);
2111 	if (err)
2112 		goto restore_opts;
2113 	checkpoint_changed =
2114 			disable_checkpoint != test_opt(sbi, DISABLE_CHECKPOINT);
2115 
2116 	/*
2117 	 * Previous and new state of filesystem is RO,
2118 	 * so skip checking GC and FLUSH_MERGE conditions.
2119 	 */
2120 	if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
2121 		goto skip;
2122 
2123 	if (f2fs_sb_has_readonly(sbi) && !(*flags & SB_RDONLY)) {
2124 		err = -EROFS;
2125 		goto restore_opts;
2126 	}
2127 
2128 #ifdef CONFIG_QUOTA
2129 	if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
2130 		err = dquot_suspend(sb, -1);
2131 		if (err < 0)
2132 			goto restore_opts;
2133 	} else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
2134 		/* dquot_resume needs RW */
2135 		sb->s_flags &= ~SB_RDONLY;
2136 		if (sb_any_quota_suspended(sb)) {
2137 			dquot_resume(sb, -1);
2138 		} else if (f2fs_sb_has_quota_ino(sbi)) {
2139 			err = f2fs_enable_quotas(sb);
2140 			if (err)
2141 				goto restore_opts;
2142 		}
2143 	}
2144 #endif
2145 	/* disallow enable atgc dynamically */
2146 	if (no_atgc == !!test_opt(sbi, ATGC)) {
2147 		err = -EINVAL;
2148 		f2fs_warn(sbi, "switch atgc option is not allowed");
2149 		goto restore_opts;
2150 	}
2151 
2152 	/* disallow enable/disable extent_cache dynamically */
2153 	if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
2154 		err = -EINVAL;
2155 		f2fs_warn(sbi, "switch extent_cache option is not allowed");
2156 		goto restore_opts;
2157 	}
2158 
2159 	if (no_io_align == !!F2FS_IO_ALIGNED(sbi)) {
2160 		err = -EINVAL;
2161 		f2fs_warn(sbi, "switch io_bits option is not allowed");
2162 		goto restore_opts;
2163 	}
2164 
2165 	if (no_compress_cache == !!test_opt(sbi, COMPRESS_CACHE)) {
2166 		err = -EINVAL;
2167 		f2fs_warn(sbi, "switch compress_cache option is not allowed");
2168 		goto restore_opts;
2169 	}
2170 
2171 	if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
2172 		err = -EINVAL;
2173 		f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
2174 		goto restore_opts;
2175 	}
2176 
2177 	/*
2178 	 * We stop the GC thread if FS is mounted as RO
2179 	 * or if background_gc = off is passed in mount
2180 	 * option. Also sync the filesystem.
2181 	 */
2182 	if ((*flags & SB_RDONLY) ||
2183 			(F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF &&
2184 			!test_opt(sbi, GC_MERGE))) {
2185 		if (sbi->gc_thread) {
2186 			f2fs_stop_gc_thread(sbi);
2187 			need_restart_gc = true;
2188 		}
2189 	} else if (!sbi->gc_thread) {
2190 		err = f2fs_start_gc_thread(sbi);
2191 		if (err)
2192 			goto restore_opts;
2193 		need_stop_gc = true;
2194 	}
2195 
2196 	if (*flags & SB_RDONLY ||
2197 		F2FS_OPTION(sbi).whint_mode != org_mount_opt.whint_mode) {
2198 		sync_inodes_sb(sb);
2199 
2200 		set_sbi_flag(sbi, SBI_IS_DIRTY);
2201 		set_sbi_flag(sbi, SBI_IS_CLOSE);
2202 		f2fs_sync_fs(sb, 1);
2203 		clear_sbi_flag(sbi, SBI_IS_CLOSE);
2204 	}
2205 
2206 	if ((*flags & SB_RDONLY) || test_opt(sbi, DISABLE_CHECKPOINT) ||
2207 			!test_opt(sbi, MERGE_CHECKPOINT)) {
2208 		f2fs_stop_ckpt_thread(sbi);
2209 		need_restart_ckpt = true;
2210 	} else {
2211 		err = f2fs_start_ckpt_thread(sbi);
2212 		if (err) {
2213 			f2fs_err(sbi,
2214 			    "Failed to start F2FS issue_checkpoint_thread (%d)",
2215 			    err);
2216 			goto restore_gc;
2217 		}
2218 		need_stop_ckpt = true;
2219 	}
2220 
2221 	/*
2222 	 * We stop issue flush thread if FS is mounted as RO
2223 	 * or if flush_merge is not passed in mount option.
2224 	 */
2225 	if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2226 		clear_opt(sbi, FLUSH_MERGE);
2227 		f2fs_destroy_flush_cmd_control(sbi, false);
2228 		need_restart_flush = true;
2229 	} else {
2230 		err = f2fs_create_flush_cmd_control(sbi);
2231 		if (err)
2232 			goto restore_ckpt;
2233 		need_stop_flush = true;
2234 	}
2235 
2236 	if (checkpoint_changed) {
2237 		if (test_opt(sbi, DISABLE_CHECKPOINT)) {
2238 			err = f2fs_disable_checkpoint(sbi);
2239 			if (err)
2240 				goto restore_flush;
2241 		} else {
2242 			f2fs_enable_checkpoint(sbi);
2243 		}
2244 	}
2245 
2246 skip:
2247 #ifdef CONFIG_QUOTA
2248 	/* Release old quota file names */
2249 	for (i = 0; i < MAXQUOTAS; i++)
2250 		kfree(org_mount_opt.s_qf_names[i]);
2251 #endif
2252 	/* Update the POSIXACL Flag */
2253 	sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
2254 		(test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
2255 
2256 	limit_reserve_root(sbi);
2257 	adjust_unusable_cap_perc(sbi);
2258 	*flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
2259 	return 0;
2260 restore_flush:
2261 	if (need_restart_flush) {
2262 		if (f2fs_create_flush_cmd_control(sbi))
2263 			f2fs_warn(sbi, "background flush thread has stopped");
2264 	} else if (need_stop_flush) {
2265 		clear_opt(sbi, FLUSH_MERGE);
2266 		f2fs_destroy_flush_cmd_control(sbi, false);
2267 	}
2268 restore_ckpt:
2269 	if (need_restart_ckpt) {
2270 		if (f2fs_start_ckpt_thread(sbi))
2271 			f2fs_warn(sbi, "background ckpt thread has stopped");
2272 	} else if (need_stop_ckpt) {
2273 		f2fs_stop_ckpt_thread(sbi);
2274 	}
2275 restore_gc:
2276 	if (need_restart_gc) {
2277 		if (f2fs_start_gc_thread(sbi))
2278 			f2fs_warn(sbi, "background gc thread has stopped");
2279 	} else if (need_stop_gc) {
2280 		f2fs_stop_gc_thread(sbi);
2281 	}
2282 restore_opts:
2283 #ifdef CONFIG_QUOTA
2284 	F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
2285 	for (i = 0; i < MAXQUOTAS; i++) {
2286 		kfree(F2FS_OPTION(sbi).s_qf_names[i]);
2287 		F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
2288 	}
2289 #endif
2290 	sbi->mount_opt = org_mount_opt;
2291 	sb->s_flags = old_sb_flags;
2292 	return err;
2293 }
2294 
2295 #ifdef CONFIG_QUOTA
2296 /* Read data from quotafile */
2297 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
2298 			       size_t len, loff_t off)
2299 {
2300 	struct inode *inode = sb_dqopt(sb)->files[type];
2301 	struct address_space *mapping = inode->i_mapping;
2302 	block_t blkidx = F2FS_BYTES_TO_BLK(off);
2303 	int offset = off & (sb->s_blocksize - 1);
2304 	int tocopy;
2305 	size_t toread;
2306 	loff_t i_size = i_size_read(inode);
2307 	struct page *page;
2308 	char *kaddr;
2309 
2310 	if (off > i_size)
2311 		return 0;
2312 
2313 	if (off + len > i_size)
2314 		len = i_size - off;
2315 	toread = len;
2316 	while (toread > 0) {
2317 		tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
2318 repeat:
2319 		page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
2320 		if (IS_ERR(page)) {
2321 			if (PTR_ERR(page) == -ENOMEM) {
2322 				congestion_wait(BLK_RW_ASYNC,
2323 						DEFAULT_IO_TIMEOUT);
2324 				goto repeat;
2325 			}
2326 			set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2327 			return PTR_ERR(page);
2328 		}
2329 
2330 		lock_page(page);
2331 
2332 		if (unlikely(page->mapping != mapping)) {
2333 			f2fs_put_page(page, 1);
2334 			goto repeat;
2335 		}
2336 		if (unlikely(!PageUptodate(page))) {
2337 			f2fs_put_page(page, 1);
2338 			set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2339 			return -EIO;
2340 		}
2341 
2342 		kaddr = kmap_atomic(page);
2343 		memcpy(data, kaddr + offset, tocopy);
2344 		kunmap_atomic(kaddr);
2345 		f2fs_put_page(page, 1);
2346 
2347 		offset = 0;
2348 		toread -= tocopy;
2349 		data += tocopy;
2350 		blkidx++;
2351 	}
2352 	return len;
2353 }
2354 
2355 /* Write to quotafile */
2356 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2357 				const char *data, size_t len, loff_t off)
2358 {
2359 	struct inode *inode = sb_dqopt(sb)->files[type];
2360 	struct address_space *mapping = inode->i_mapping;
2361 	const struct address_space_operations *a_ops = mapping->a_ops;
2362 	int offset = off & (sb->s_blocksize - 1);
2363 	size_t towrite = len;
2364 	struct page *page;
2365 	void *fsdata = NULL;
2366 	char *kaddr;
2367 	int err = 0;
2368 	int tocopy;
2369 
2370 	while (towrite > 0) {
2371 		tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2372 								towrite);
2373 retry:
2374 		err = a_ops->write_begin(NULL, mapping, off, tocopy, 0,
2375 							&page, &fsdata);
2376 		if (unlikely(err)) {
2377 			if (err == -ENOMEM) {
2378 				congestion_wait(BLK_RW_ASYNC,
2379 						DEFAULT_IO_TIMEOUT);
2380 				goto retry;
2381 			}
2382 			set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2383 			break;
2384 		}
2385 
2386 		kaddr = kmap_atomic(page);
2387 		memcpy(kaddr + offset, data, tocopy);
2388 		kunmap_atomic(kaddr);
2389 		flush_dcache_page(page);
2390 
2391 		a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2392 						page, fsdata);
2393 		offset = 0;
2394 		towrite -= tocopy;
2395 		off += tocopy;
2396 		data += tocopy;
2397 		cond_resched();
2398 	}
2399 
2400 	if (len == towrite)
2401 		return err;
2402 	inode->i_mtime = inode->i_ctime = current_time(inode);
2403 	f2fs_mark_inode_dirty_sync(inode, false);
2404 	return len - towrite;
2405 }
2406 
2407 static struct dquot **f2fs_get_dquots(struct inode *inode)
2408 {
2409 	return F2FS_I(inode)->i_dquot;
2410 }
2411 
2412 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2413 {
2414 	return &F2FS_I(inode)->i_reserved_quota;
2415 }
2416 
2417 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2418 {
2419 	if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2420 		f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2421 		return 0;
2422 	}
2423 
2424 	return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2425 					F2FS_OPTION(sbi).s_jquota_fmt, type);
2426 }
2427 
2428 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2429 {
2430 	int enabled = 0;
2431 	int i, err;
2432 
2433 	if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2434 		err = f2fs_enable_quotas(sbi->sb);
2435 		if (err) {
2436 			f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2437 			return 0;
2438 		}
2439 		return 1;
2440 	}
2441 
2442 	for (i = 0; i < MAXQUOTAS; i++) {
2443 		if (F2FS_OPTION(sbi).s_qf_names[i]) {
2444 			err = f2fs_quota_on_mount(sbi, i);
2445 			if (!err) {
2446 				enabled = 1;
2447 				continue;
2448 			}
2449 			f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2450 				 err, i);
2451 		}
2452 	}
2453 	return enabled;
2454 }
2455 
2456 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2457 			     unsigned int flags)
2458 {
2459 	struct inode *qf_inode;
2460 	unsigned long qf_inum;
2461 	int err;
2462 
2463 	BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2464 
2465 	qf_inum = f2fs_qf_ino(sb, type);
2466 	if (!qf_inum)
2467 		return -EPERM;
2468 
2469 	qf_inode = f2fs_iget(sb, qf_inum);
2470 	if (IS_ERR(qf_inode)) {
2471 		f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2472 		return PTR_ERR(qf_inode);
2473 	}
2474 
2475 	/* Don't account quota for quota files to avoid recursion */
2476 	qf_inode->i_flags |= S_NOQUOTA;
2477 	err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2478 	iput(qf_inode);
2479 	return err;
2480 }
2481 
2482 static int f2fs_enable_quotas(struct super_block *sb)
2483 {
2484 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2485 	int type, err = 0;
2486 	unsigned long qf_inum;
2487 	bool quota_mopt[MAXQUOTAS] = {
2488 		test_opt(sbi, USRQUOTA),
2489 		test_opt(sbi, GRPQUOTA),
2490 		test_opt(sbi, PRJQUOTA),
2491 	};
2492 
2493 	if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2494 		f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2495 		return 0;
2496 	}
2497 
2498 	sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2499 
2500 	for (type = 0; type < MAXQUOTAS; type++) {
2501 		qf_inum = f2fs_qf_ino(sb, type);
2502 		if (qf_inum) {
2503 			err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2504 				DQUOT_USAGE_ENABLED |
2505 				(quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2506 			if (err) {
2507 				f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2508 					 type, err);
2509 				for (type--; type >= 0; type--)
2510 					dquot_quota_off(sb, type);
2511 				set_sbi_flag(F2FS_SB(sb),
2512 						SBI_QUOTA_NEED_REPAIR);
2513 				return err;
2514 			}
2515 		}
2516 	}
2517 	return 0;
2518 }
2519 
2520 int f2fs_quota_sync(struct super_block *sb, int type)
2521 {
2522 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2523 	struct quota_info *dqopt = sb_dqopt(sb);
2524 	int cnt;
2525 	int ret;
2526 
2527 	/*
2528 	 * do_quotactl
2529 	 *  f2fs_quota_sync
2530 	 *  down_read(quota_sem)
2531 	 *  dquot_writeback_dquots()
2532 	 *  f2fs_dquot_commit
2533 	 *                            block_operation
2534 	 *                            down_read(quota_sem)
2535 	 */
2536 	f2fs_lock_op(sbi);
2537 
2538 	down_read(&sbi->quota_sem);
2539 	ret = dquot_writeback_dquots(sb, type);
2540 	if (ret)
2541 		goto out;
2542 
2543 	/*
2544 	 * Now when everything is written we can discard the pagecache so
2545 	 * that userspace sees the changes.
2546 	 */
2547 	for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2548 		struct address_space *mapping;
2549 
2550 		if (type != -1 && cnt != type)
2551 			continue;
2552 		if (!sb_has_quota_active(sb, cnt))
2553 			continue;
2554 
2555 		mapping = dqopt->files[cnt]->i_mapping;
2556 
2557 		ret = filemap_fdatawrite(mapping);
2558 		if (ret)
2559 			goto out;
2560 
2561 		/* if we are using journalled quota */
2562 		if (is_journalled_quota(sbi))
2563 			continue;
2564 
2565 		ret = filemap_fdatawait(mapping);
2566 		if (ret)
2567 			set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2568 
2569 		inode_lock(dqopt->files[cnt]);
2570 		truncate_inode_pages(&dqopt->files[cnt]->i_data, 0);
2571 		inode_unlock(dqopt->files[cnt]);
2572 	}
2573 out:
2574 	if (ret)
2575 		set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2576 	up_read(&sbi->quota_sem);
2577 	f2fs_unlock_op(sbi);
2578 	return ret;
2579 }
2580 
2581 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2582 							const struct path *path)
2583 {
2584 	struct inode *inode;
2585 	int err;
2586 
2587 	/* if quota sysfile exists, deny enabling quota with specific file */
2588 	if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2589 		f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2590 		return -EBUSY;
2591 	}
2592 
2593 	err = f2fs_quota_sync(sb, type);
2594 	if (err)
2595 		return err;
2596 
2597 	err = dquot_quota_on(sb, type, format_id, path);
2598 	if (err)
2599 		return err;
2600 
2601 	inode = d_inode(path->dentry);
2602 
2603 	inode_lock(inode);
2604 	F2FS_I(inode)->i_flags |= F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL;
2605 	f2fs_set_inode_flags(inode);
2606 	inode_unlock(inode);
2607 	f2fs_mark_inode_dirty_sync(inode, false);
2608 
2609 	return 0;
2610 }
2611 
2612 static int __f2fs_quota_off(struct super_block *sb, int type)
2613 {
2614 	struct inode *inode = sb_dqopt(sb)->files[type];
2615 	int err;
2616 
2617 	if (!inode || !igrab(inode))
2618 		return dquot_quota_off(sb, type);
2619 
2620 	err = f2fs_quota_sync(sb, type);
2621 	if (err)
2622 		goto out_put;
2623 
2624 	err = dquot_quota_off(sb, type);
2625 	if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
2626 		goto out_put;
2627 
2628 	inode_lock(inode);
2629 	F2FS_I(inode)->i_flags &= ~(F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL);
2630 	f2fs_set_inode_flags(inode);
2631 	inode_unlock(inode);
2632 	f2fs_mark_inode_dirty_sync(inode, false);
2633 out_put:
2634 	iput(inode);
2635 	return err;
2636 }
2637 
2638 static int f2fs_quota_off(struct super_block *sb, int type)
2639 {
2640 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2641 	int err;
2642 
2643 	err = __f2fs_quota_off(sb, type);
2644 
2645 	/*
2646 	 * quotactl can shutdown journalled quota, result in inconsistence
2647 	 * between quota record and fs data by following updates, tag the
2648 	 * flag to let fsck be aware of it.
2649 	 */
2650 	if (is_journalled_quota(sbi))
2651 		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2652 	return err;
2653 }
2654 
2655 void f2fs_quota_off_umount(struct super_block *sb)
2656 {
2657 	int type;
2658 	int err;
2659 
2660 	for (type = 0; type < MAXQUOTAS; type++) {
2661 		err = __f2fs_quota_off(sb, type);
2662 		if (err) {
2663 			int ret = dquot_quota_off(sb, type);
2664 
2665 			f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
2666 				 type, err, ret);
2667 			set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2668 		}
2669 	}
2670 	/*
2671 	 * In case of checkpoint=disable, we must flush quota blocks.
2672 	 * This can cause NULL exception for node_inode in end_io, since
2673 	 * put_super already dropped it.
2674 	 */
2675 	sync_filesystem(sb);
2676 }
2677 
2678 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
2679 {
2680 	struct quota_info *dqopt = sb_dqopt(sb);
2681 	int type;
2682 
2683 	for (type = 0; type < MAXQUOTAS; type++) {
2684 		if (!dqopt->files[type])
2685 			continue;
2686 		f2fs_inode_synced(dqopt->files[type]);
2687 	}
2688 }
2689 
2690 static int f2fs_dquot_commit(struct dquot *dquot)
2691 {
2692 	struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2693 	int ret;
2694 
2695 	down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
2696 	ret = dquot_commit(dquot);
2697 	if (ret < 0)
2698 		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2699 	up_read(&sbi->quota_sem);
2700 	return ret;
2701 }
2702 
2703 static int f2fs_dquot_acquire(struct dquot *dquot)
2704 {
2705 	struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2706 	int ret;
2707 
2708 	down_read(&sbi->quota_sem);
2709 	ret = dquot_acquire(dquot);
2710 	if (ret < 0)
2711 		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2712 	up_read(&sbi->quota_sem);
2713 	return ret;
2714 }
2715 
2716 static int f2fs_dquot_release(struct dquot *dquot)
2717 {
2718 	struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
2719 	int ret = dquot_release(dquot);
2720 
2721 	if (ret < 0)
2722 		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2723 	return ret;
2724 }
2725 
2726 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
2727 {
2728 	struct super_block *sb = dquot->dq_sb;
2729 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2730 	int ret = dquot_mark_dquot_dirty(dquot);
2731 
2732 	/* if we are using journalled quota */
2733 	if (is_journalled_quota(sbi))
2734 		set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
2735 
2736 	return ret;
2737 }
2738 
2739 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
2740 {
2741 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2742 	int ret = dquot_commit_info(sb, type);
2743 
2744 	if (ret < 0)
2745 		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2746 	return ret;
2747 }
2748 
2749 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
2750 {
2751 	*projid = F2FS_I(inode)->i_projid;
2752 	return 0;
2753 }
2754 
2755 static const struct dquot_operations f2fs_quota_operations = {
2756 	.get_reserved_space = f2fs_get_reserved_space,
2757 	.write_dquot	= f2fs_dquot_commit,
2758 	.acquire_dquot	= f2fs_dquot_acquire,
2759 	.release_dquot	= f2fs_dquot_release,
2760 	.mark_dirty	= f2fs_dquot_mark_dquot_dirty,
2761 	.write_info	= f2fs_dquot_commit_info,
2762 	.alloc_dquot	= dquot_alloc,
2763 	.destroy_dquot	= dquot_destroy,
2764 	.get_projid	= f2fs_get_projid,
2765 	.get_next_id	= dquot_get_next_id,
2766 };
2767 
2768 static const struct quotactl_ops f2fs_quotactl_ops = {
2769 	.quota_on	= f2fs_quota_on,
2770 	.quota_off	= f2fs_quota_off,
2771 	.quota_sync	= f2fs_quota_sync,
2772 	.get_state	= dquot_get_state,
2773 	.set_info	= dquot_set_dqinfo,
2774 	.get_dqblk	= dquot_get_dqblk,
2775 	.set_dqblk	= dquot_set_dqblk,
2776 	.get_nextdqblk	= dquot_get_next_dqblk,
2777 };
2778 #else
2779 int f2fs_quota_sync(struct super_block *sb, int type)
2780 {
2781 	return 0;
2782 }
2783 
2784 void f2fs_quota_off_umount(struct super_block *sb)
2785 {
2786 }
2787 #endif
2788 
2789 static const struct super_operations f2fs_sops = {
2790 	.alloc_inode	= f2fs_alloc_inode,
2791 	.free_inode	= f2fs_free_inode,
2792 	.drop_inode	= f2fs_drop_inode,
2793 	.write_inode	= f2fs_write_inode,
2794 	.dirty_inode	= f2fs_dirty_inode,
2795 	.show_options	= f2fs_show_options,
2796 #ifdef CONFIG_QUOTA
2797 	.quota_read	= f2fs_quota_read,
2798 	.quota_write	= f2fs_quota_write,
2799 	.get_dquots	= f2fs_get_dquots,
2800 #endif
2801 	.evict_inode	= f2fs_evict_inode,
2802 	.put_super	= f2fs_put_super,
2803 	.sync_fs	= f2fs_sync_fs,
2804 	.freeze_fs	= f2fs_freeze,
2805 	.unfreeze_fs	= f2fs_unfreeze,
2806 	.statfs		= f2fs_statfs,
2807 	.remount_fs	= f2fs_remount,
2808 };
2809 
2810 #ifdef CONFIG_FS_ENCRYPTION
2811 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
2812 {
2813 	return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2814 				F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2815 				ctx, len, NULL);
2816 }
2817 
2818 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
2819 							void *fs_data)
2820 {
2821 	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2822 
2823 	/*
2824 	 * Encrypting the root directory is not allowed because fsck
2825 	 * expects lost+found directory to exist and remain unencrypted
2826 	 * if LOST_FOUND feature is enabled.
2827 	 *
2828 	 */
2829 	if (f2fs_sb_has_lost_found(sbi) &&
2830 			inode->i_ino == F2FS_ROOT_INO(sbi))
2831 		return -EPERM;
2832 
2833 	return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
2834 				F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
2835 				ctx, len, fs_data, XATTR_CREATE);
2836 }
2837 
2838 static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
2839 {
2840 	return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
2841 }
2842 
2843 static bool f2fs_has_stable_inodes(struct super_block *sb)
2844 {
2845 	return true;
2846 }
2847 
2848 static void f2fs_get_ino_and_lblk_bits(struct super_block *sb,
2849 				       int *ino_bits_ret, int *lblk_bits_ret)
2850 {
2851 	*ino_bits_ret = 8 * sizeof(nid_t);
2852 	*lblk_bits_ret = 8 * sizeof(block_t);
2853 }
2854 
2855 static int f2fs_get_num_devices(struct super_block *sb)
2856 {
2857 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2858 
2859 	if (f2fs_is_multi_device(sbi))
2860 		return sbi->s_ndevs;
2861 	return 1;
2862 }
2863 
2864 static void f2fs_get_devices(struct super_block *sb,
2865 			     struct request_queue **devs)
2866 {
2867 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2868 	int i;
2869 
2870 	for (i = 0; i < sbi->s_ndevs; i++)
2871 		devs[i] = bdev_get_queue(FDEV(i).bdev);
2872 }
2873 
2874 static const struct fscrypt_operations f2fs_cryptops = {
2875 	.key_prefix		= "f2fs:",
2876 	.get_context		= f2fs_get_context,
2877 	.set_context		= f2fs_set_context,
2878 	.get_dummy_policy	= f2fs_get_dummy_policy,
2879 	.empty_dir		= f2fs_empty_dir,
2880 	.max_namelen		= F2FS_NAME_LEN,
2881 	.has_stable_inodes	= f2fs_has_stable_inodes,
2882 	.get_ino_and_lblk_bits	= f2fs_get_ino_and_lblk_bits,
2883 	.get_num_devices	= f2fs_get_num_devices,
2884 	.get_devices		= f2fs_get_devices,
2885 };
2886 #endif
2887 
2888 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
2889 		u64 ino, u32 generation)
2890 {
2891 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
2892 	struct inode *inode;
2893 
2894 	if (f2fs_check_nid_range(sbi, ino))
2895 		return ERR_PTR(-ESTALE);
2896 
2897 	/*
2898 	 * f2fs_iget isn't quite right if the inode is currently unallocated!
2899 	 * However f2fs_iget currently does appropriate checks to handle stale
2900 	 * inodes so everything is OK.
2901 	 */
2902 	inode = f2fs_iget(sb, ino);
2903 	if (IS_ERR(inode))
2904 		return ERR_CAST(inode);
2905 	if (unlikely(generation && inode->i_generation != generation)) {
2906 		/* we didn't find the right inode.. */
2907 		iput(inode);
2908 		return ERR_PTR(-ESTALE);
2909 	}
2910 	return inode;
2911 }
2912 
2913 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
2914 		int fh_len, int fh_type)
2915 {
2916 	return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
2917 				    f2fs_nfs_get_inode);
2918 }
2919 
2920 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
2921 		int fh_len, int fh_type)
2922 {
2923 	return generic_fh_to_parent(sb, fid, fh_len, fh_type,
2924 				    f2fs_nfs_get_inode);
2925 }
2926 
2927 static const struct export_operations f2fs_export_ops = {
2928 	.fh_to_dentry = f2fs_fh_to_dentry,
2929 	.fh_to_parent = f2fs_fh_to_parent,
2930 	.get_parent = f2fs_get_parent,
2931 };
2932 
2933 loff_t max_file_blocks(struct inode *inode)
2934 {
2935 	loff_t result = 0;
2936 	loff_t leaf_count;
2937 
2938 	/*
2939 	 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
2940 	 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
2941 	 * space in inode.i_addr, it will be more safe to reassign
2942 	 * result as zero.
2943 	 */
2944 
2945 	if (inode && f2fs_compressed_file(inode))
2946 		leaf_count = ADDRS_PER_BLOCK(inode);
2947 	else
2948 		leaf_count = DEF_ADDRS_PER_BLOCK;
2949 
2950 	/* two direct node blocks */
2951 	result += (leaf_count * 2);
2952 
2953 	/* two indirect node blocks */
2954 	leaf_count *= NIDS_PER_BLOCK;
2955 	result += (leaf_count * 2);
2956 
2957 	/* one double indirect node block */
2958 	leaf_count *= NIDS_PER_BLOCK;
2959 	result += leaf_count;
2960 
2961 	return result;
2962 }
2963 
2964 static int __f2fs_commit_super(struct buffer_head *bh,
2965 			struct f2fs_super_block *super)
2966 {
2967 	lock_buffer(bh);
2968 	if (super)
2969 		memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
2970 	set_buffer_dirty(bh);
2971 	unlock_buffer(bh);
2972 
2973 	/* it's rare case, we can do fua all the time */
2974 	return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
2975 }
2976 
2977 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
2978 					struct buffer_head *bh)
2979 {
2980 	struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
2981 					(bh->b_data + F2FS_SUPER_OFFSET);
2982 	struct super_block *sb = sbi->sb;
2983 	u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
2984 	u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
2985 	u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
2986 	u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
2987 	u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
2988 	u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
2989 	u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
2990 	u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
2991 	u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
2992 	u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
2993 	u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
2994 	u32 segment_count = le32_to_cpu(raw_super->segment_count);
2995 	u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
2996 	u64 main_end_blkaddr = main_blkaddr +
2997 				(segment_count_main << log_blocks_per_seg);
2998 	u64 seg_end_blkaddr = segment0_blkaddr +
2999 				(segment_count << log_blocks_per_seg);
3000 
3001 	if (segment0_blkaddr != cp_blkaddr) {
3002 		f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
3003 			  segment0_blkaddr, cp_blkaddr);
3004 		return true;
3005 	}
3006 
3007 	if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
3008 							sit_blkaddr) {
3009 		f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
3010 			  cp_blkaddr, sit_blkaddr,
3011 			  segment_count_ckpt << log_blocks_per_seg);
3012 		return true;
3013 	}
3014 
3015 	if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
3016 							nat_blkaddr) {
3017 		f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
3018 			  sit_blkaddr, nat_blkaddr,
3019 			  segment_count_sit << log_blocks_per_seg);
3020 		return true;
3021 	}
3022 
3023 	if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
3024 							ssa_blkaddr) {
3025 		f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
3026 			  nat_blkaddr, ssa_blkaddr,
3027 			  segment_count_nat << log_blocks_per_seg);
3028 		return true;
3029 	}
3030 
3031 	if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
3032 							main_blkaddr) {
3033 		f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
3034 			  ssa_blkaddr, main_blkaddr,
3035 			  segment_count_ssa << log_blocks_per_seg);
3036 		return true;
3037 	}
3038 
3039 	if (main_end_blkaddr > seg_end_blkaddr) {
3040 		f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
3041 			  main_blkaddr, seg_end_blkaddr,
3042 			  segment_count_main << log_blocks_per_seg);
3043 		return true;
3044 	} else if (main_end_blkaddr < seg_end_blkaddr) {
3045 		int err = 0;
3046 		char *res;
3047 
3048 		/* fix in-memory information all the time */
3049 		raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
3050 				segment0_blkaddr) >> log_blocks_per_seg);
3051 
3052 		if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
3053 			set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3054 			res = "internally";
3055 		} else {
3056 			err = __f2fs_commit_super(bh, NULL);
3057 			res = err ? "failed" : "done";
3058 		}
3059 		f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
3060 			  res, main_blkaddr, seg_end_blkaddr,
3061 			  segment_count_main << log_blocks_per_seg);
3062 		if (err)
3063 			return true;
3064 	}
3065 	return false;
3066 }
3067 
3068 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
3069 				struct buffer_head *bh)
3070 {
3071 	block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
3072 	block_t total_sections, blocks_per_seg;
3073 	struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3074 					(bh->b_data + F2FS_SUPER_OFFSET);
3075 	size_t crc_offset = 0;
3076 	__u32 crc = 0;
3077 
3078 	if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
3079 		f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
3080 			  F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
3081 		return -EINVAL;
3082 	}
3083 
3084 	/* Check checksum_offset and crc in superblock */
3085 	if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
3086 		crc_offset = le32_to_cpu(raw_super->checksum_offset);
3087 		if (crc_offset !=
3088 			offsetof(struct f2fs_super_block, crc)) {
3089 			f2fs_info(sbi, "Invalid SB checksum offset: %zu",
3090 				  crc_offset);
3091 			return -EFSCORRUPTED;
3092 		}
3093 		crc = le32_to_cpu(raw_super->crc);
3094 		if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
3095 			f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
3096 			return -EFSCORRUPTED;
3097 		}
3098 	}
3099 
3100 	/* Currently, support only 4KB block size */
3101 	if (le32_to_cpu(raw_super->log_blocksize) != F2FS_BLKSIZE_BITS) {
3102 		f2fs_info(sbi, "Invalid log_blocksize (%u), supports only %u",
3103 			  le32_to_cpu(raw_super->log_blocksize),
3104 			  F2FS_BLKSIZE_BITS);
3105 		return -EFSCORRUPTED;
3106 	}
3107 
3108 	/* check log blocks per segment */
3109 	if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
3110 		f2fs_info(sbi, "Invalid log blocks per segment (%u)",
3111 			  le32_to_cpu(raw_super->log_blocks_per_seg));
3112 		return -EFSCORRUPTED;
3113 	}
3114 
3115 	/* Currently, support 512/1024/2048/4096 bytes sector size */
3116 	if (le32_to_cpu(raw_super->log_sectorsize) >
3117 				F2FS_MAX_LOG_SECTOR_SIZE ||
3118 		le32_to_cpu(raw_super->log_sectorsize) <
3119 				F2FS_MIN_LOG_SECTOR_SIZE) {
3120 		f2fs_info(sbi, "Invalid log sectorsize (%u)",
3121 			  le32_to_cpu(raw_super->log_sectorsize));
3122 		return -EFSCORRUPTED;
3123 	}
3124 	if (le32_to_cpu(raw_super->log_sectors_per_block) +
3125 		le32_to_cpu(raw_super->log_sectorsize) !=
3126 			F2FS_MAX_LOG_SECTOR_SIZE) {
3127 		f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
3128 			  le32_to_cpu(raw_super->log_sectors_per_block),
3129 			  le32_to_cpu(raw_super->log_sectorsize));
3130 		return -EFSCORRUPTED;
3131 	}
3132 
3133 	segment_count = le32_to_cpu(raw_super->segment_count);
3134 	segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3135 	segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3136 	secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3137 	total_sections = le32_to_cpu(raw_super->section_count);
3138 
3139 	/* blocks_per_seg should be 512, given the above check */
3140 	blocks_per_seg = 1 << le32_to_cpu(raw_super->log_blocks_per_seg);
3141 
3142 	if (segment_count > F2FS_MAX_SEGMENT ||
3143 				segment_count < F2FS_MIN_SEGMENTS) {
3144 		f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
3145 		return -EFSCORRUPTED;
3146 	}
3147 
3148 	if (total_sections > segment_count_main || total_sections < 1 ||
3149 			segs_per_sec > segment_count || !segs_per_sec) {
3150 		f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
3151 			  segment_count, total_sections, segs_per_sec);
3152 		return -EFSCORRUPTED;
3153 	}
3154 
3155 	if (segment_count_main != total_sections * segs_per_sec) {
3156 		f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
3157 			  segment_count_main, total_sections, segs_per_sec);
3158 		return -EFSCORRUPTED;
3159 	}
3160 
3161 	if ((segment_count / segs_per_sec) < total_sections) {
3162 		f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
3163 			  segment_count, segs_per_sec, total_sections);
3164 		return -EFSCORRUPTED;
3165 	}
3166 
3167 	if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
3168 		f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
3169 			  segment_count, le64_to_cpu(raw_super->block_count));
3170 		return -EFSCORRUPTED;
3171 	}
3172 
3173 	if (RDEV(0).path[0]) {
3174 		block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
3175 		int i = 1;
3176 
3177 		while (i < MAX_DEVICES && RDEV(i).path[0]) {
3178 			dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
3179 			i++;
3180 		}
3181 		if (segment_count != dev_seg_count) {
3182 			f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
3183 					segment_count, dev_seg_count);
3184 			return -EFSCORRUPTED;
3185 		}
3186 	} else {
3187 		if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
3188 					!bdev_is_zoned(sbi->sb->s_bdev)) {
3189 			f2fs_info(sbi, "Zoned block device path is missing");
3190 			return -EFSCORRUPTED;
3191 		}
3192 	}
3193 
3194 	if (secs_per_zone > total_sections || !secs_per_zone) {
3195 		f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
3196 			  secs_per_zone, total_sections);
3197 		return -EFSCORRUPTED;
3198 	}
3199 	if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
3200 			raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
3201 			(le32_to_cpu(raw_super->extension_count) +
3202 			raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
3203 		f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
3204 			  le32_to_cpu(raw_super->extension_count),
3205 			  raw_super->hot_ext_count,
3206 			  F2FS_MAX_EXTENSION);
3207 		return -EFSCORRUPTED;
3208 	}
3209 
3210 	if (le32_to_cpu(raw_super->cp_payload) >
3211 				(blocks_per_seg - F2FS_CP_PACKS)) {
3212 		f2fs_info(sbi, "Insane cp_payload (%u > %u)",
3213 			  le32_to_cpu(raw_super->cp_payload),
3214 			  blocks_per_seg - F2FS_CP_PACKS);
3215 		return -EFSCORRUPTED;
3216 	}
3217 
3218 	/* check reserved ino info */
3219 	if (le32_to_cpu(raw_super->node_ino) != 1 ||
3220 		le32_to_cpu(raw_super->meta_ino) != 2 ||
3221 		le32_to_cpu(raw_super->root_ino) != 3) {
3222 		f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
3223 			  le32_to_cpu(raw_super->node_ino),
3224 			  le32_to_cpu(raw_super->meta_ino),
3225 			  le32_to_cpu(raw_super->root_ino));
3226 		return -EFSCORRUPTED;
3227 	}
3228 
3229 	/* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
3230 	if (sanity_check_area_boundary(sbi, bh))
3231 		return -EFSCORRUPTED;
3232 
3233 	return 0;
3234 }
3235 
3236 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
3237 {
3238 	unsigned int total, fsmeta;
3239 	struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3240 	struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
3241 	unsigned int ovp_segments, reserved_segments;
3242 	unsigned int main_segs, blocks_per_seg;
3243 	unsigned int sit_segs, nat_segs;
3244 	unsigned int sit_bitmap_size, nat_bitmap_size;
3245 	unsigned int log_blocks_per_seg;
3246 	unsigned int segment_count_main;
3247 	unsigned int cp_pack_start_sum, cp_payload;
3248 	block_t user_block_count, valid_user_blocks;
3249 	block_t avail_node_count, valid_node_count;
3250 	int i, j;
3251 
3252 	total = le32_to_cpu(raw_super->segment_count);
3253 	fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
3254 	sit_segs = le32_to_cpu(raw_super->segment_count_sit);
3255 	fsmeta += sit_segs;
3256 	nat_segs = le32_to_cpu(raw_super->segment_count_nat);
3257 	fsmeta += nat_segs;
3258 	fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
3259 	fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
3260 
3261 	if (unlikely(fsmeta >= total))
3262 		return 1;
3263 
3264 	ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
3265 	reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
3266 
3267 	if (!f2fs_sb_has_readonly(sbi) &&
3268 			unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
3269 			ovp_segments == 0 || reserved_segments == 0)) {
3270 		f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
3271 		return 1;
3272 	}
3273 	user_block_count = le64_to_cpu(ckpt->user_block_count);
3274 	segment_count_main = le32_to_cpu(raw_super->segment_count_main) +
3275 			(f2fs_sb_has_readonly(sbi) ? 1 : 0);
3276 	log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3277 	if (!user_block_count || user_block_count >=
3278 			segment_count_main << log_blocks_per_seg) {
3279 		f2fs_err(sbi, "Wrong user_block_count: %u",
3280 			 user_block_count);
3281 		return 1;
3282 	}
3283 
3284 	valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
3285 	if (valid_user_blocks > user_block_count) {
3286 		f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
3287 			 valid_user_blocks, user_block_count);
3288 		return 1;
3289 	}
3290 
3291 	valid_node_count = le32_to_cpu(ckpt->valid_node_count);
3292 	avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
3293 	if (valid_node_count > avail_node_count) {
3294 		f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
3295 			 valid_node_count, avail_node_count);
3296 		return 1;
3297 	}
3298 
3299 	main_segs = le32_to_cpu(raw_super->segment_count_main);
3300 	blocks_per_seg = sbi->blocks_per_seg;
3301 
3302 	for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3303 		if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
3304 			le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
3305 			return 1;
3306 
3307 		if (f2fs_sb_has_readonly(sbi))
3308 			goto check_data;
3309 
3310 		for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
3311 			if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3312 				le32_to_cpu(ckpt->cur_node_segno[j])) {
3313 				f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
3314 					 i, j,
3315 					 le32_to_cpu(ckpt->cur_node_segno[i]));
3316 				return 1;
3317 			}
3318 		}
3319 	}
3320 check_data:
3321 	for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
3322 		if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
3323 			le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
3324 			return 1;
3325 
3326 		if (f2fs_sb_has_readonly(sbi))
3327 			goto skip_cross;
3328 
3329 		for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3330 			if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3331 				le32_to_cpu(ckpt->cur_data_segno[j])) {
3332 				f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3333 					 i, j,
3334 					 le32_to_cpu(ckpt->cur_data_segno[i]));
3335 				return 1;
3336 			}
3337 		}
3338 	}
3339 	for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3340 		for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3341 			if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3342 				le32_to_cpu(ckpt->cur_data_segno[j])) {
3343 				f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3344 					 i, j,
3345 					 le32_to_cpu(ckpt->cur_node_segno[i]));
3346 				return 1;
3347 			}
3348 		}
3349 	}
3350 skip_cross:
3351 	sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3352 	nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3353 
3354 	if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3355 		nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3356 		f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3357 			 sit_bitmap_size, nat_bitmap_size);
3358 		return 1;
3359 	}
3360 
3361 	cp_pack_start_sum = __start_sum_addr(sbi);
3362 	cp_payload = __cp_payload(sbi);
3363 	if (cp_pack_start_sum < cp_payload + 1 ||
3364 		cp_pack_start_sum > blocks_per_seg - 1 -
3365 			NR_CURSEG_PERSIST_TYPE) {
3366 		f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3367 			 cp_pack_start_sum);
3368 		return 1;
3369 	}
3370 
3371 	if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3372 		le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3373 		f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3374 			  "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3375 			  "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3376 			  le32_to_cpu(ckpt->checksum_offset));
3377 		return 1;
3378 	}
3379 
3380 	if (unlikely(f2fs_cp_error(sbi))) {
3381 		f2fs_err(sbi, "A bug case: need to run fsck");
3382 		return 1;
3383 	}
3384 	return 0;
3385 }
3386 
3387 static void init_sb_info(struct f2fs_sb_info *sbi)
3388 {
3389 	struct f2fs_super_block *raw_super = sbi->raw_super;
3390 	int i;
3391 
3392 	sbi->log_sectors_per_block =
3393 		le32_to_cpu(raw_super->log_sectors_per_block);
3394 	sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3395 	sbi->blocksize = 1 << sbi->log_blocksize;
3396 	sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3397 	sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
3398 	sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3399 	sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3400 	sbi->total_sections = le32_to_cpu(raw_super->section_count);
3401 	sbi->total_node_count =
3402 		(le32_to_cpu(raw_super->segment_count_nat) / 2)
3403 			* sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
3404 	F2FS_ROOT_INO(sbi) = le32_to_cpu(raw_super->root_ino);
3405 	F2FS_NODE_INO(sbi) = le32_to_cpu(raw_super->node_ino);
3406 	F2FS_META_INO(sbi) = le32_to_cpu(raw_super->meta_ino);
3407 	sbi->cur_victim_sec = NULL_SECNO;
3408 	sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3409 	sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3410 	sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3411 	sbi->migration_granularity = sbi->segs_per_sec;
3412 
3413 	sbi->dir_level = DEF_DIR_LEVEL;
3414 	sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3415 	sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3416 	sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3417 	sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3418 	sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3419 	sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3420 				DEF_UMOUNT_DISCARD_TIMEOUT;
3421 	clear_sbi_flag(sbi, SBI_NEED_FSCK);
3422 
3423 	for (i = 0; i < NR_COUNT_TYPE; i++)
3424 		atomic_set(&sbi->nr_pages[i], 0);
3425 
3426 	for (i = 0; i < META; i++)
3427 		atomic_set(&sbi->wb_sync_req[i], 0);
3428 
3429 	INIT_LIST_HEAD(&sbi->s_list);
3430 	mutex_init(&sbi->umount_mutex);
3431 	init_rwsem(&sbi->io_order_lock);
3432 	spin_lock_init(&sbi->cp_lock);
3433 
3434 	sbi->dirty_device = 0;
3435 	spin_lock_init(&sbi->dev_lock);
3436 
3437 	init_rwsem(&sbi->sb_lock);
3438 	init_rwsem(&sbi->pin_sem);
3439 }
3440 
3441 static int init_percpu_info(struct f2fs_sb_info *sbi)
3442 {
3443 	int err;
3444 
3445 	err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3446 	if (err)
3447 		return err;
3448 
3449 	err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3450 								GFP_KERNEL);
3451 	if (err)
3452 		percpu_counter_destroy(&sbi->alloc_valid_block_count);
3453 
3454 	return err;
3455 }
3456 
3457 #ifdef CONFIG_BLK_DEV_ZONED
3458 
3459 struct f2fs_report_zones_args {
3460 	struct f2fs_dev_info *dev;
3461 	bool zone_cap_mismatch;
3462 };
3463 
3464 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3465 			      void *data)
3466 {
3467 	struct f2fs_report_zones_args *rz_args = data;
3468 
3469 	if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3470 		return 0;
3471 
3472 	set_bit(idx, rz_args->dev->blkz_seq);
3473 	rz_args->dev->zone_capacity_blocks[idx] = zone->capacity >>
3474 						F2FS_LOG_SECTORS_PER_BLOCK;
3475 	if (zone->len != zone->capacity && !rz_args->zone_cap_mismatch)
3476 		rz_args->zone_cap_mismatch = true;
3477 
3478 	return 0;
3479 }
3480 
3481 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3482 {
3483 	struct block_device *bdev = FDEV(devi).bdev;
3484 	sector_t nr_sectors = bdev_nr_sectors(bdev);
3485 	struct f2fs_report_zones_args rep_zone_arg;
3486 	int ret;
3487 
3488 	if (!f2fs_sb_has_blkzoned(sbi))
3489 		return 0;
3490 
3491 	if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3492 				SECTOR_TO_BLOCK(bdev_zone_sectors(bdev)))
3493 		return -EINVAL;
3494 	sbi->blocks_per_blkz = SECTOR_TO_BLOCK(bdev_zone_sectors(bdev));
3495 	if (sbi->log_blocks_per_blkz && sbi->log_blocks_per_blkz !=
3496 				__ilog2_u32(sbi->blocks_per_blkz))
3497 		return -EINVAL;
3498 	sbi->log_blocks_per_blkz = __ilog2_u32(sbi->blocks_per_blkz);
3499 	FDEV(devi).nr_blkz = SECTOR_TO_BLOCK(nr_sectors) >>
3500 					sbi->log_blocks_per_blkz;
3501 	if (nr_sectors & (bdev_zone_sectors(bdev) - 1))
3502 		FDEV(devi).nr_blkz++;
3503 
3504 	FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3505 					BITS_TO_LONGS(FDEV(devi).nr_blkz)
3506 					* sizeof(unsigned long),
3507 					GFP_KERNEL);
3508 	if (!FDEV(devi).blkz_seq)
3509 		return -ENOMEM;
3510 
3511 	/* Get block zones type and zone-capacity */
3512 	FDEV(devi).zone_capacity_blocks = f2fs_kzalloc(sbi,
3513 					FDEV(devi).nr_blkz * sizeof(block_t),
3514 					GFP_KERNEL);
3515 	if (!FDEV(devi).zone_capacity_blocks)
3516 		return -ENOMEM;
3517 
3518 	rep_zone_arg.dev = &FDEV(devi);
3519 	rep_zone_arg.zone_cap_mismatch = false;
3520 
3521 	ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3522 				  &rep_zone_arg);
3523 	if (ret < 0)
3524 		return ret;
3525 
3526 	if (!rep_zone_arg.zone_cap_mismatch) {
3527 		kfree(FDEV(devi).zone_capacity_blocks);
3528 		FDEV(devi).zone_capacity_blocks = NULL;
3529 	}
3530 
3531 	return 0;
3532 }
3533 #endif
3534 
3535 /*
3536  * Read f2fs raw super block.
3537  * Because we have two copies of super block, so read both of them
3538  * to get the first valid one. If any one of them is broken, we pass
3539  * them recovery flag back to the caller.
3540  */
3541 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3542 			struct f2fs_super_block **raw_super,
3543 			int *valid_super_block, int *recovery)
3544 {
3545 	struct super_block *sb = sbi->sb;
3546 	int block;
3547 	struct buffer_head *bh;
3548 	struct f2fs_super_block *super;
3549 	int err = 0;
3550 
3551 	super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3552 	if (!super)
3553 		return -ENOMEM;
3554 
3555 	for (block = 0; block < 2; block++) {
3556 		bh = sb_bread(sb, block);
3557 		if (!bh) {
3558 			f2fs_err(sbi, "Unable to read %dth superblock",
3559 				 block + 1);
3560 			err = -EIO;
3561 			*recovery = 1;
3562 			continue;
3563 		}
3564 
3565 		/* sanity checking of raw super */
3566 		err = sanity_check_raw_super(sbi, bh);
3567 		if (err) {
3568 			f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3569 				 block + 1);
3570 			brelse(bh);
3571 			*recovery = 1;
3572 			continue;
3573 		}
3574 
3575 		if (!*raw_super) {
3576 			memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3577 							sizeof(*super));
3578 			*valid_super_block = block;
3579 			*raw_super = super;
3580 		}
3581 		brelse(bh);
3582 	}
3583 
3584 	/* No valid superblock */
3585 	if (!*raw_super)
3586 		kfree(super);
3587 	else
3588 		err = 0;
3589 
3590 	return err;
3591 }
3592 
3593 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
3594 {
3595 	struct buffer_head *bh;
3596 	__u32 crc = 0;
3597 	int err;
3598 
3599 	if ((recover && f2fs_readonly(sbi->sb)) ||
3600 				bdev_read_only(sbi->sb->s_bdev)) {
3601 		set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3602 		return -EROFS;
3603 	}
3604 
3605 	/* we should update superblock crc here */
3606 	if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
3607 		crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
3608 				offsetof(struct f2fs_super_block, crc));
3609 		F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
3610 	}
3611 
3612 	/* write back-up superblock first */
3613 	bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
3614 	if (!bh)
3615 		return -EIO;
3616 	err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3617 	brelse(bh);
3618 
3619 	/* if we are in recovery path, skip writing valid superblock */
3620 	if (recover || err)
3621 		return err;
3622 
3623 	/* write current valid superblock */
3624 	bh = sb_bread(sbi->sb, sbi->valid_super_block);
3625 	if (!bh)
3626 		return -EIO;
3627 	err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3628 	brelse(bh);
3629 	return err;
3630 }
3631 
3632 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
3633 {
3634 	struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3635 	unsigned int max_devices = MAX_DEVICES;
3636 	int i;
3637 
3638 	/* Initialize single device information */
3639 	if (!RDEV(0).path[0]) {
3640 		if (!bdev_is_zoned(sbi->sb->s_bdev))
3641 			return 0;
3642 		max_devices = 1;
3643 	}
3644 
3645 	/*
3646 	 * Initialize multiple devices information, or single
3647 	 * zoned block device information.
3648 	 */
3649 	sbi->devs = f2fs_kzalloc(sbi,
3650 				 array_size(max_devices,
3651 					    sizeof(struct f2fs_dev_info)),
3652 				 GFP_KERNEL);
3653 	if (!sbi->devs)
3654 		return -ENOMEM;
3655 
3656 	for (i = 0; i < max_devices; i++) {
3657 
3658 		if (i > 0 && !RDEV(i).path[0])
3659 			break;
3660 
3661 		if (max_devices == 1) {
3662 			/* Single zoned block device mount */
3663 			FDEV(0).bdev =
3664 				blkdev_get_by_dev(sbi->sb->s_bdev->bd_dev,
3665 					sbi->sb->s_mode, sbi->sb->s_type);
3666 		} else {
3667 			/* Multi-device mount */
3668 			memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
3669 			FDEV(i).total_segments =
3670 				le32_to_cpu(RDEV(i).total_segments);
3671 			if (i == 0) {
3672 				FDEV(i).start_blk = 0;
3673 				FDEV(i).end_blk = FDEV(i).start_blk +
3674 				    (FDEV(i).total_segments <<
3675 				    sbi->log_blocks_per_seg) - 1 +
3676 				    le32_to_cpu(raw_super->segment0_blkaddr);
3677 			} else {
3678 				FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
3679 				FDEV(i).end_blk = FDEV(i).start_blk +
3680 					(FDEV(i).total_segments <<
3681 					sbi->log_blocks_per_seg) - 1;
3682 			}
3683 			FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
3684 					sbi->sb->s_mode, sbi->sb->s_type);
3685 		}
3686 		if (IS_ERR(FDEV(i).bdev))
3687 			return PTR_ERR(FDEV(i).bdev);
3688 
3689 		/* to release errored devices */
3690 		sbi->s_ndevs = i + 1;
3691 
3692 #ifdef CONFIG_BLK_DEV_ZONED
3693 		if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
3694 				!f2fs_sb_has_blkzoned(sbi)) {
3695 			f2fs_err(sbi, "Zoned block device feature not enabled");
3696 			return -EINVAL;
3697 		}
3698 		if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
3699 			if (init_blkz_info(sbi, i)) {
3700 				f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
3701 				return -EINVAL;
3702 			}
3703 			if (max_devices == 1)
3704 				break;
3705 			f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
3706 				  i, FDEV(i).path,
3707 				  FDEV(i).total_segments,
3708 				  FDEV(i).start_blk, FDEV(i).end_blk,
3709 				  bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
3710 				  "Host-aware" : "Host-managed");
3711 			continue;
3712 		}
3713 #endif
3714 		f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
3715 			  i, FDEV(i).path,
3716 			  FDEV(i).total_segments,
3717 			  FDEV(i).start_blk, FDEV(i).end_blk);
3718 	}
3719 	f2fs_info(sbi,
3720 		  "IO Block Size: %8d KB", F2FS_IO_SIZE_KB(sbi));
3721 	return 0;
3722 }
3723 
3724 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
3725 {
3726 #ifdef CONFIG_UNICODE
3727 	if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
3728 		const struct f2fs_sb_encodings *encoding_info;
3729 		struct unicode_map *encoding;
3730 		__u16 encoding_flags;
3731 
3732 		if (f2fs_sb_read_encoding(sbi->raw_super, &encoding_info,
3733 					  &encoding_flags)) {
3734 			f2fs_err(sbi,
3735 				 "Encoding requested by superblock is unknown");
3736 			return -EINVAL;
3737 		}
3738 
3739 		encoding = utf8_load(encoding_info->version);
3740 		if (IS_ERR(encoding)) {
3741 			f2fs_err(sbi,
3742 				 "can't mount with superblock charset: %s-%s "
3743 				 "not supported by the kernel. flags: 0x%x.",
3744 				 encoding_info->name, encoding_info->version,
3745 				 encoding_flags);
3746 			return PTR_ERR(encoding);
3747 		}
3748 		f2fs_info(sbi, "Using encoding defined by superblock: "
3749 			 "%s-%s with flags 0x%hx", encoding_info->name,
3750 			 encoding_info->version?:"\b", encoding_flags);
3751 
3752 		sbi->sb->s_encoding = encoding;
3753 		sbi->sb->s_encoding_flags = encoding_flags;
3754 	}
3755 #else
3756 	if (f2fs_sb_has_casefold(sbi)) {
3757 		f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
3758 		return -EINVAL;
3759 	}
3760 #endif
3761 	return 0;
3762 }
3763 
3764 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
3765 {
3766 	struct f2fs_sm_info *sm_i = SM_I(sbi);
3767 
3768 	/* adjust parameters according to the volume size */
3769 	if (sm_i->main_segments <= SMALL_VOLUME_SEGMENTS) {
3770 		F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
3771 		sm_i->dcc_info->discard_granularity = 1;
3772 		sm_i->ipu_policy = 1 << F2FS_IPU_FORCE;
3773 	}
3774 
3775 	sbi->readdir_ra = 1;
3776 }
3777 
3778 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
3779 {
3780 	struct f2fs_sb_info *sbi;
3781 	struct f2fs_super_block *raw_super;
3782 	struct inode *root;
3783 	int err;
3784 	bool skip_recovery = false, need_fsck = false;
3785 	char *options = NULL;
3786 	int recovery, i, valid_super_block;
3787 	struct curseg_info *seg_i;
3788 	int retry_cnt = 1;
3789 
3790 try_onemore:
3791 	err = -EINVAL;
3792 	raw_super = NULL;
3793 	valid_super_block = -1;
3794 	recovery = 0;
3795 
3796 	/* allocate memory for f2fs-specific super block info */
3797 	sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
3798 	if (!sbi)
3799 		return -ENOMEM;
3800 
3801 	sbi->sb = sb;
3802 
3803 	/* Load the checksum driver */
3804 	sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
3805 	if (IS_ERR(sbi->s_chksum_driver)) {
3806 		f2fs_err(sbi, "Cannot load crc32 driver.");
3807 		err = PTR_ERR(sbi->s_chksum_driver);
3808 		sbi->s_chksum_driver = NULL;
3809 		goto free_sbi;
3810 	}
3811 
3812 	/* set a block size */
3813 	if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
3814 		f2fs_err(sbi, "unable to set blocksize");
3815 		goto free_sbi;
3816 	}
3817 
3818 	err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
3819 								&recovery);
3820 	if (err)
3821 		goto free_sbi;
3822 
3823 	sb->s_fs_info = sbi;
3824 	sbi->raw_super = raw_super;
3825 
3826 	/* precompute checksum seed for metadata */
3827 	if (f2fs_sb_has_inode_chksum(sbi))
3828 		sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
3829 						sizeof(raw_super->uuid));
3830 
3831 	default_options(sbi);
3832 	/* parse mount options */
3833 	options = kstrdup((const char *)data, GFP_KERNEL);
3834 	if (data && !options) {
3835 		err = -ENOMEM;
3836 		goto free_sb_buf;
3837 	}
3838 
3839 	err = parse_options(sb, options, false);
3840 	if (err)
3841 		goto free_options;
3842 
3843 	sb->s_maxbytes = max_file_blocks(NULL) <<
3844 				le32_to_cpu(raw_super->log_blocksize);
3845 	sb->s_max_links = F2FS_LINK_MAX;
3846 
3847 	err = f2fs_setup_casefold(sbi);
3848 	if (err)
3849 		goto free_options;
3850 
3851 #ifdef CONFIG_QUOTA
3852 	sb->dq_op = &f2fs_quota_operations;
3853 	sb->s_qcop = &f2fs_quotactl_ops;
3854 	sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
3855 
3856 	if (f2fs_sb_has_quota_ino(sbi)) {
3857 		for (i = 0; i < MAXQUOTAS; i++) {
3858 			if (f2fs_qf_ino(sbi->sb, i))
3859 				sbi->nquota_files++;
3860 		}
3861 	}
3862 #endif
3863 
3864 	sb->s_op = &f2fs_sops;
3865 #ifdef CONFIG_FS_ENCRYPTION
3866 	sb->s_cop = &f2fs_cryptops;
3867 #endif
3868 #ifdef CONFIG_FS_VERITY
3869 	sb->s_vop = &f2fs_verityops;
3870 #endif
3871 	sb->s_xattr = f2fs_xattr_handlers;
3872 	sb->s_export_op = &f2fs_export_ops;
3873 	sb->s_magic = F2FS_SUPER_MAGIC;
3874 	sb->s_time_gran = 1;
3875 	sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
3876 		(test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
3877 	memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
3878 	sb->s_iflags |= SB_I_CGROUPWB;
3879 
3880 	/* init f2fs-specific super block info */
3881 	sbi->valid_super_block = valid_super_block;
3882 	init_rwsem(&sbi->gc_lock);
3883 	mutex_init(&sbi->writepages);
3884 	init_rwsem(&sbi->cp_global_sem);
3885 	init_rwsem(&sbi->node_write);
3886 	init_rwsem(&sbi->node_change);
3887 
3888 	/* disallow all the data/node/meta page writes */
3889 	set_sbi_flag(sbi, SBI_POR_DOING);
3890 	spin_lock_init(&sbi->stat_lock);
3891 
3892 	/* init iostat info */
3893 	spin_lock_init(&sbi->iostat_lock);
3894 	sbi->iostat_enable = false;
3895 	sbi->iostat_period_ms = DEFAULT_IOSTAT_PERIOD_MS;
3896 
3897 	for (i = 0; i < NR_PAGE_TYPE; i++) {
3898 		int n = (i == META) ? 1 : NR_TEMP_TYPE;
3899 		int j;
3900 
3901 		sbi->write_io[i] =
3902 			f2fs_kmalloc(sbi,
3903 				     array_size(n,
3904 						sizeof(struct f2fs_bio_info)),
3905 				     GFP_KERNEL);
3906 		if (!sbi->write_io[i]) {
3907 			err = -ENOMEM;
3908 			goto free_bio_info;
3909 		}
3910 
3911 		for (j = HOT; j < n; j++) {
3912 			init_rwsem(&sbi->write_io[i][j].io_rwsem);
3913 			sbi->write_io[i][j].sbi = sbi;
3914 			sbi->write_io[i][j].bio = NULL;
3915 			spin_lock_init(&sbi->write_io[i][j].io_lock);
3916 			INIT_LIST_HEAD(&sbi->write_io[i][j].io_list);
3917 			INIT_LIST_HEAD(&sbi->write_io[i][j].bio_list);
3918 			init_rwsem(&sbi->write_io[i][j].bio_list_lock);
3919 		}
3920 	}
3921 
3922 	init_rwsem(&sbi->cp_rwsem);
3923 	init_rwsem(&sbi->quota_sem);
3924 	init_waitqueue_head(&sbi->cp_wait);
3925 	init_sb_info(sbi);
3926 
3927 	err = init_percpu_info(sbi);
3928 	if (err)
3929 		goto free_bio_info;
3930 
3931 	if (F2FS_IO_ALIGNED(sbi)) {
3932 		sbi->write_io_dummy =
3933 			mempool_create_page_pool(2 * (F2FS_IO_SIZE(sbi) - 1), 0);
3934 		if (!sbi->write_io_dummy) {
3935 			err = -ENOMEM;
3936 			goto free_percpu;
3937 		}
3938 	}
3939 
3940 	/* init per sbi slab cache */
3941 	err = f2fs_init_xattr_caches(sbi);
3942 	if (err)
3943 		goto free_io_dummy;
3944 	err = f2fs_init_page_array_cache(sbi);
3945 	if (err)
3946 		goto free_xattr_cache;
3947 
3948 	/* get an inode for meta space */
3949 	sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
3950 	if (IS_ERR(sbi->meta_inode)) {
3951 		f2fs_err(sbi, "Failed to read F2FS meta data inode");
3952 		err = PTR_ERR(sbi->meta_inode);
3953 		goto free_page_array_cache;
3954 	}
3955 
3956 	err = f2fs_get_valid_checkpoint(sbi);
3957 	if (err) {
3958 		f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
3959 		goto free_meta_inode;
3960 	}
3961 
3962 	if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
3963 		set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3964 	if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
3965 		set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
3966 		sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
3967 	}
3968 
3969 	if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
3970 		set_sbi_flag(sbi, SBI_NEED_FSCK);
3971 
3972 	/* Initialize device list */
3973 	err = f2fs_scan_devices(sbi);
3974 	if (err) {
3975 		f2fs_err(sbi, "Failed to find devices");
3976 		goto free_devices;
3977 	}
3978 
3979 	err = f2fs_init_post_read_wq(sbi);
3980 	if (err) {
3981 		f2fs_err(sbi, "Failed to initialize post read workqueue");
3982 		goto free_devices;
3983 	}
3984 
3985 	sbi->total_valid_node_count =
3986 				le32_to_cpu(sbi->ckpt->valid_node_count);
3987 	percpu_counter_set(&sbi->total_valid_inode_count,
3988 				le32_to_cpu(sbi->ckpt->valid_inode_count));
3989 	sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
3990 	sbi->total_valid_block_count =
3991 				le64_to_cpu(sbi->ckpt->valid_block_count);
3992 	sbi->last_valid_block_count = sbi->total_valid_block_count;
3993 	sbi->reserved_blocks = 0;
3994 	sbi->current_reserved_blocks = 0;
3995 	limit_reserve_root(sbi);
3996 	adjust_unusable_cap_perc(sbi);
3997 
3998 	for (i = 0; i < NR_INODE_TYPE; i++) {
3999 		INIT_LIST_HEAD(&sbi->inode_list[i]);
4000 		spin_lock_init(&sbi->inode_lock[i]);
4001 	}
4002 	mutex_init(&sbi->flush_lock);
4003 
4004 	f2fs_init_extent_cache_info(sbi);
4005 
4006 	f2fs_init_ino_entry_info(sbi);
4007 
4008 	f2fs_init_fsync_node_info(sbi);
4009 
4010 	/* setup checkpoint request control and start checkpoint issue thread */
4011 	f2fs_init_ckpt_req_control(sbi);
4012 	if (!f2fs_readonly(sb) && !test_opt(sbi, DISABLE_CHECKPOINT) &&
4013 			test_opt(sbi, MERGE_CHECKPOINT)) {
4014 		err = f2fs_start_ckpt_thread(sbi);
4015 		if (err) {
4016 			f2fs_err(sbi,
4017 			    "Failed to start F2FS issue_checkpoint_thread (%d)",
4018 			    err);
4019 			goto stop_ckpt_thread;
4020 		}
4021 	}
4022 
4023 	/* setup f2fs internal modules */
4024 	err = f2fs_build_segment_manager(sbi);
4025 	if (err) {
4026 		f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
4027 			 err);
4028 		goto free_sm;
4029 	}
4030 	err = f2fs_build_node_manager(sbi);
4031 	if (err) {
4032 		f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
4033 			 err);
4034 		goto free_nm;
4035 	}
4036 
4037 	/* For write statistics */
4038 	sbi->sectors_written_start = f2fs_get_sectors_written(sbi);
4039 
4040 	/* Read accumulated write IO statistics if exists */
4041 	seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
4042 	if (__exist_node_summaries(sbi))
4043 		sbi->kbytes_written =
4044 			le64_to_cpu(seg_i->journal->info.kbytes_written);
4045 
4046 	f2fs_build_gc_manager(sbi);
4047 
4048 	err = f2fs_build_stats(sbi);
4049 	if (err)
4050 		goto free_nm;
4051 
4052 	/* get an inode for node space */
4053 	sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
4054 	if (IS_ERR(sbi->node_inode)) {
4055 		f2fs_err(sbi, "Failed to read node inode");
4056 		err = PTR_ERR(sbi->node_inode);
4057 		goto free_stats;
4058 	}
4059 
4060 	/* read root inode and dentry */
4061 	root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
4062 	if (IS_ERR(root)) {
4063 		f2fs_err(sbi, "Failed to read root inode");
4064 		err = PTR_ERR(root);
4065 		goto free_node_inode;
4066 	}
4067 	if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
4068 			!root->i_size || !root->i_nlink) {
4069 		iput(root);
4070 		err = -EINVAL;
4071 		goto free_node_inode;
4072 	}
4073 
4074 	sb->s_root = d_make_root(root); /* allocate root dentry */
4075 	if (!sb->s_root) {
4076 		err = -ENOMEM;
4077 		goto free_node_inode;
4078 	}
4079 
4080 	err = f2fs_init_compress_inode(sbi);
4081 	if (err)
4082 		goto free_root_inode;
4083 
4084 	err = f2fs_register_sysfs(sbi);
4085 	if (err)
4086 		goto free_compress_inode;
4087 
4088 #ifdef CONFIG_QUOTA
4089 	/* Enable quota usage during mount */
4090 	if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
4091 		err = f2fs_enable_quotas(sb);
4092 		if (err)
4093 			f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
4094 	}
4095 #endif
4096 	/* if there are any orphan inodes, free them */
4097 	err = f2fs_recover_orphan_inodes(sbi);
4098 	if (err)
4099 		goto free_meta;
4100 
4101 	if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
4102 		goto reset_checkpoint;
4103 
4104 	/* recover fsynced data */
4105 	if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
4106 			!test_opt(sbi, NORECOVERY)) {
4107 		/*
4108 		 * mount should be failed, when device has readonly mode, and
4109 		 * previous checkpoint was not done by clean system shutdown.
4110 		 */
4111 		if (f2fs_hw_is_readonly(sbi)) {
4112 			if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4113 				err = f2fs_recover_fsync_data(sbi, true);
4114 				if (err > 0) {
4115 					err = -EROFS;
4116 					f2fs_err(sbi, "Need to recover fsync data, but "
4117 						"write access unavailable, please try "
4118 						"mount w/ disable_roll_forward or norecovery");
4119 				}
4120 				if (err < 0)
4121 					goto free_meta;
4122 			}
4123 			f2fs_info(sbi, "write access unavailable, skipping recovery");
4124 			goto reset_checkpoint;
4125 		}
4126 
4127 		if (need_fsck)
4128 			set_sbi_flag(sbi, SBI_NEED_FSCK);
4129 
4130 		if (skip_recovery)
4131 			goto reset_checkpoint;
4132 
4133 		err = f2fs_recover_fsync_data(sbi, false);
4134 		if (err < 0) {
4135 			if (err != -ENOMEM)
4136 				skip_recovery = true;
4137 			need_fsck = true;
4138 			f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
4139 				 err);
4140 			goto free_meta;
4141 		}
4142 	} else {
4143 		err = f2fs_recover_fsync_data(sbi, true);
4144 
4145 		if (!f2fs_readonly(sb) && err > 0) {
4146 			err = -EINVAL;
4147 			f2fs_err(sbi, "Need to recover fsync data");
4148 			goto free_meta;
4149 		}
4150 	}
4151 
4152 	/*
4153 	 * If the f2fs is not readonly and fsync data recovery succeeds,
4154 	 * check zoned block devices' write pointer consistency.
4155 	 */
4156 	if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
4157 		err = f2fs_check_write_pointer(sbi);
4158 		if (err)
4159 			goto free_meta;
4160 	}
4161 
4162 reset_checkpoint:
4163 	f2fs_init_inmem_curseg(sbi);
4164 
4165 	/* f2fs_recover_fsync_data() cleared this already */
4166 	clear_sbi_flag(sbi, SBI_POR_DOING);
4167 
4168 	if (test_opt(sbi, DISABLE_CHECKPOINT)) {
4169 		err = f2fs_disable_checkpoint(sbi);
4170 		if (err)
4171 			goto sync_free_meta;
4172 	} else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
4173 		f2fs_enable_checkpoint(sbi);
4174 	}
4175 
4176 	/*
4177 	 * If filesystem is not mounted as read-only then
4178 	 * do start the gc_thread.
4179 	 */
4180 	if ((F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF ||
4181 		test_opt(sbi, GC_MERGE)) && !f2fs_readonly(sb)) {
4182 		/* After POR, we can run background GC thread.*/
4183 		err = f2fs_start_gc_thread(sbi);
4184 		if (err)
4185 			goto sync_free_meta;
4186 	}
4187 	kvfree(options);
4188 
4189 	/* recover broken superblock */
4190 	if (recovery) {
4191 		err = f2fs_commit_super(sbi, true);
4192 		f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
4193 			  sbi->valid_super_block ? 1 : 2, err);
4194 	}
4195 
4196 	f2fs_join_shrinker(sbi);
4197 
4198 	f2fs_tuning_parameters(sbi);
4199 
4200 	f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
4201 		    cur_cp_version(F2FS_CKPT(sbi)));
4202 	f2fs_update_time(sbi, CP_TIME);
4203 	f2fs_update_time(sbi, REQ_TIME);
4204 	clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4205 	return 0;
4206 
4207 sync_free_meta:
4208 	/* safe to flush all the data */
4209 	sync_filesystem(sbi->sb);
4210 	retry_cnt = 0;
4211 
4212 free_meta:
4213 #ifdef CONFIG_QUOTA
4214 	f2fs_truncate_quota_inode_pages(sb);
4215 	if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
4216 		f2fs_quota_off_umount(sbi->sb);
4217 #endif
4218 	/*
4219 	 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
4220 	 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
4221 	 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
4222 	 * falls into an infinite loop in f2fs_sync_meta_pages().
4223 	 */
4224 	truncate_inode_pages_final(META_MAPPING(sbi));
4225 	/* evict some inodes being cached by GC */
4226 	evict_inodes(sb);
4227 	f2fs_unregister_sysfs(sbi);
4228 free_compress_inode:
4229 	f2fs_destroy_compress_inode(sbi);
4230 free_root_inode:
4231 	dput(sb->s_root);
4232 	sb->s_root = NULL;
4233 free_node_inode:
4234 	f2fs_release_ino_entry(sbi, true);
4235 	truncate_inode_pages_final(NODE_MAPPING(sbi));
4236 	iput(sbi->node_inode);
4237 	sbi->node_inode = NULL;
4238 free_stats:
4239 	f2fs_destroy_stats(sbi);
4240 free_nm:
4241 	f2fs_destroy_node_manager(sbi);
4242 free_sm:
4243 	f2fs_destroy_segment_manager(sbi);
4244 	f2fs_destroy_post_read_wq(sbi);
4245 stop_ckpt_thread:
4246 	f2fs_stop_ckpt_thread(sbi);
4247 free_devices:
4248 	destroy_device_list(sbi);
4249 	kvfree(sbi->ckpt);
4250 free_meta_inode:
4251 	make_bad_inode(sbi->meta_inode);
4252 	iput(sbi->meta_inode);
4253 	sbi->meta_inode = NULL;
4254 free_page_array_cache:
4255 	f2fs_destroy_page_array_cache(sbi);
4256 free_xattr_cache:
4257 	f2fs_destroy_xattr_caches(sbi);
4258 free_io_dummy:
4259 	mempool_destroy(sbi->write_io_dummy);
4260 free_percpu:
4261 	destroy_percpu_info(sbi);
4262 free_bio_info:
4263 	for (i = 0; i < NR_PAGE_TYPE; i++)
4264 		kvfree(sbi->write_io[i]);
4265 
4266 #ifdef CONFIG_UNICODE
4267 	utf8_unload(sb->s_encoding);
4268 	sb->s_encoding = NULL;
4269 #endif
4270 free_options:
4271 #ifdef CONFIG_QUOTA
4272 	for (i = 0; i < MAXQUOTAS; i++)
4273 		kfree(F2FS_OPTION(sbi).s_qf_names[i]);
4274 #endif
4275 	fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
4276 	kvfree(options);
4277 free_sb_buf:
4278 	kfree(raw_super);
4279 free_sbi:
4280 	if (sbi->s_chksum_driver)
4281 		crypto_free_shash(sbi->s_chksum_driver);
4282 	kfree(sbi);
4283 
4284 	/* give only one another chance */
4285 	if (retry_cnt > 0 && skip_recovery) {
4286 		retry_cnt--;
4287 		shrink_dcache_sb(sb);
4288 		goto try_onemore;
4289 	}
4290 	return err;
4291 }
4292 
4293 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
4294 			const char *dev_name, void *data)
4295 {
4296 	return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
4297 }
4298 
4299 static void kill_f2fs_super(struct super_block *sb)
4300 {
4301 	if (sb->s_root) {
4302 		struct f2fs_sb_info *sbi = F2FS_SB(sb);
4303 
4304 		set_sbi_flag(sbi, SBI_IS_CLOSE);
4305 		f2fs_stop_gc_thread(sbi);
4306 		f2fs_stop_discard_thread(sbi);
4307 
4308 #ifdef CONFIG_F2FS_FS_COMPRESSION
4309 		/*
4310 		 * latter evict_inode() can bypass checking and invalidating
4311 		 * compress inode cache.
4312 		 */
4313 		if (test_opt(sbi, COMPRESS_CACHE))
4314 			truncate_inode_pages_final(COMPRESS_MAPPING(sbi));
4315 #endif
4316 
4317 		if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
4318 				!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4319 			struct cp_control cpc = {
4320 				.reason = CP_UMOUNT,
4321 			};
4322 			f2fs_write_checkpoint(sbi, &cpc);
4323 		}
4324 
4325 		if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
4326 			sb->s_flags &= ~SB_RDONLY;
4327 	}
4328 	kill_block_super(sb);
4329 }
4330 
4331 static struct file_system_type f2fs_fs_type = {
4332 	.owner		= THIS_MODULE,
4333 	.name		= "f2fs",
4334 	.mount		= f2fs_mount,
4335 	.kill_sb	= kill_f2fs_super,
4336 	.fs_flags	= FS_REQUIRES_DEV,
4337 };
4338 MODULE_ALIAS_FS("f2fs");
4339 
4340 static int __init init_inodecache(void)
4341 {
4342 	f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
4343 			sizeof(struct f2fs_inode_info), 0,
4344 			SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
4345 	if (!f2fs_inode_cachep)
4346 		return -ENOMEM;
4347 	return 0;
4348 }
4349 
4350 static void destroy_inodecache(void)
4351 {
4352 	/*
4353 	 * Make sure all delayed rcu free inodes are flushed before we
4354 	 * destroy cache.
4355 	 */
4356 	rcu_barrier();
4357 	kmem_cache_destroy(f2fs_inode_cachep);
4358 }
4359 
4360 static int __init init_f2fs_fs(void)
4361 {
4362 	int err;
4363 
4364 	if (PAGE_SIZE != F2FS_BLKSIZE) {
4365 		printk("F2FS not supported on PAGE_SIZE(%lu) != %d\n",
4366 				PAGE_SIZE, F2FS_BLKSIZE);
4367 		return -EINVAL;
4368 	}
4369 
4370 	err = init_inodecache();
4371 	if (err)
4372 		goto fail;
4373 	err = f2fs_create_node_manager_caches();
4374 	if (err)
4375 		goto free_inodecache;
4376 	err = f2fs_create_segment_manager_caches();
4377 	if (err)
4378 		goto free_node_manager_caches;
4379 	err = f2fs_create_checkpoint_caches();
4380 	if (err)
4381 		goto free_segment_manager_caches;
4382 	err = f2fs_create_recovery_cache();
4383 	if (err)
4384 		goto free_checkpoint_caches;
4385 	err = f2fs_create_extent_cache();
4386 	if (err)
4387 		goto free_recovery_cache;
4388 	err = f2fs_create_garbage_collection_cache();
4389 	if (err)
4390 		goto free_extent_cache;
4391 	err = f2fs_init_sysfs();
4392 	if (err)
4393 		goto free_garbage_collection_cache;
4394 	err = register_shrinker(&f2fs_shrinker_info);
4395 	if (err)
4396 		goto free_sysfs;
4397 	err = register_filesystem(&f2fs_fs_type);
4398 	if (err)
4399 		goto free_shrinker;
4400 	f2fs_create_root_stats();
4401 	err = f2fs_init_post_read_processing();
4402 	if (err)
4403 		goto free_root_stats;
4404 	err = f2fs_init_bio_entry_cache();
4405 	if (err)
4406 		goto free_post_read;
4407 	err = f2fs_init_bioset();
4408 	if (err)
4409 		goto free_bio_enrty_cache;
4410 	err = f2fs_init_compress_mempool();
4411 	if (err)
4412 		goto free_bioset;
4413 	err = f2fs_init_compress_cache();
4414 	if (err)
4415 		goto free_compress_mempool;
4416 	err = f2fs_create_casefold_cache();
4417 	if (err)
4418 		goto free_compress_cache;
4419 	return 0;
4420 free_compress_cache:
4421 	f2fs_destroy_compress_cache();
4422 free_compress_mempool:
4423 	f2fs_destroy_compress_mempool();
4424 free_bioset:
4425 	f2fs_destroy_bioset();
4426 free_bio_enrty_cache:
4427 	f2fs_destroy_bio_entry_cache();
4428 free_post_read:
4429 	f2fs_destroy_post_read_processing();
4430 free_root_stats:
4431 	f2fs_destroy_root_stats();
4432 	unregister_filesystem(&f2fs_fs_type);
4433 free_shrinker:
4434 	unregister_shrinker(&f2fs_shrinker_info);
4435 free_sysfs:
4436 	f2fs_exit_sysfs();
4437 free_garbage_collection_cache:
4438 	f2fs_destroy_garbage_collection_cache();
4439 free_extent_cache:
4440 	f2fs_destroy_extent_cache();
4441 free_recovery_cache:
4442 	f2fs_destroy_recovery_cache();
4443 free_checkpoint_caches:
4444 	f2fs_destroy_checkpoint_caches();
4445 free_segment_manager_caches:
4446 	f2fs_destroy_segment_manager_caches();
4447 free_node_manager_caches:
4448 	f2fs_destroy_node_manager_caches();
4449 free_inodecache:
4450 	destroy_inodecache();
4451 fail:
4452 	return err;
4453 }
4454 
4455 static void __exit exit_f2fs_fs(void)
4456 {
4457 	f2fs_destroy_casefold_cache();
4458 	f2fs_destroy_compress_cache();
4459 	f2fs_destroy_compress_mempool();
4460 	f2fs_destroy_bioset();
4461 	f2fs_destroy_bio_entry_cache();
4462 	f2fs_destroy_post_read_processing();
4463 	f2fs_destroy_root_stats();
4464 	unregister_filesystem(&f2fs_fs_type);
4465 	unregister_shrinker(&f2fs_shrinker_info);
4466 	f2fs_exit_sysfs();
4467 	f2fs_destroy_garbage_collection_cache();
4468 	f2fs_destroy_extent_cache();
4469 	f2fs_destroy_recovery_cache();
4470 	f2fs_destroy_checkpoint_caches();
4471 	f2fs_destroy_segment_manager_caches();
4472 	f2fs_destroy_node_manager_caches();
4473 	destroy_inodecache();
4474 }
4475 
4476 module_init(init_f2fs_fs)
4477 module_exit(exit_f2fs_fs)
4478 
4479 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
4480 MODULE_DESCRIPTION("Flash Friendly File System");
4481 MODULE_LICENSE("GPL");
4482 MODULE_SOFTDEP("pre: crc32");
4483 
4484