1 // SPDX-License-Identifier: GPL-2.0
2 /*
3 * fs/f2fs/super.c
4 *
5 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
6 * http://www.samsung.com/
7 */
8 #include <linux/module.h>
9 #include <linux/init.h>
10 #include <linux/fs.h>
11 #include <linux/fs_context.h>
12 #include <linux/sched/mm.h>
13 #include <linux/statfs.h>
14 #include <linux/buffer_head.h>
15 #include <linux/kthread.h>
16 #include <linux/parser.h>
17 #include <linux/mount.h>
18 #include <linux/seq_file.h>
19 #include <linux/proc_fs.h>
20 #include <linux/random.h>
21 #include <linux/exportfs.h>
22 #include <linux/blkdev.h>
23 #include <linux/quotaops.h>
24 #include <linux/f2fs_fs.h>
25 #include <linux/sysfs.h>
26 #include <linux/quota.h>
27 #include <linux/unicode.h>
28 #include <linux/part_stat.h>
29 #include <linux/zstd.h>
30 #include <linux/lz4.h>
31
32 #include "f2fs.h"
33 #include "node.h"
34 #include "segment.h"
35 #include "xattr.h"
36 #include "gc.h"
37 #include "iostat.h"
38
39 #define CREATE_TRACE_POINTS
40 #include <trace/events/f2fs.h>
41
42 static struct kmem_cache *f2fs_inode_cachep;
43
44 #ifdef CONFIG_F2FS_FAULT_INJECTION
45
46 const char *f2fs_fault_name[FAULT_MAX] = {
47 [FAULT_KMALLOC] = "kmalloc",
48 [FAULT_KVMALLOC] = "kvmalloc",
49 [FAULT_PAGE_ALLOC] = "page alloc",
50 [FAULT_PAGE_GET] = "page get",
51 [FAULT_ALLOC_NID] = "alloc nid",
52 [FAULT_ORPHAN] = "orphan",
53 [FAULT_BLOCK] = "no more block",
54 [FAULT_DIR_DEPTH] = "too big dir depth",
55 [FAULT_EVICT_INODE] = "evict_inode fail",
56 [FAULT_TRUNCATE] = "truncate fail",
57 [FAULT_READ_IO] = "read IO error",
58 [FAULT_CHECKPOINT] = "checkpoint error",
59 [FAULT_DISCARD] = "discard error",
60 [FAULT_WRITE_IO] = "write IO error",
61 [FAULT_SLAB_ALLOC] = "slab alloc",
62 [FAULT_DQUOT_INIT] = "dquot initialize",
63 [FAULT_LOCK_OP] = "lock_op",
64 [FAULT_BLKADDR] = "invalid blkaddr",
65 };
66
f2fs_build_fault_attr(struct f2fs_sb_info * sbi,unsigned long rate,unsigned long type)67 int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate,
68 unsigned long type)
69 {
70 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info;
71
72 if (rate) {
73 if (rate > INT_MAX)
74 return -EINVAL;
75 atomic_set(&ffi->inject_ops, 0);
76 ffi->inject_rate = (int)rate;
77 }
78
79 if (type) {
80 if (type >= BIT(FAULT_MAX))
81 return -EINVAL;
82 ffi->inject_type = (unsigned int)type;
83 }
84
85 if (!rate && !type)
86 memset(ffi, 0, sizeof(struct f2fs_fault_info));
87 else
88 f2fs_info(sbi,
89 "build fault injection attr: rate: %lu, type: 0x%lx",
90 rate, type);
91 return 0;
92 }
93 #endif
94
95 /* f2fs-wide shrinker description */
96 static struct shrinker f2fs_shrinker_info = {
97 .scan_objects = f2fs_shrink_scan,
98 .count_objects = f2fs_shrink_count,
99 .seeks = DEFAULT_SEEKS,
100 };
101
102 enum {
103 Opt_gc_background,
104 Opt_disable_roll_forward,
105 Opt_norecovery,
106 Opt_discard,
107 Opt_nodiscard,
108 Opt_noheap,
109 Opt_heap,
110 Opt_user_xattr,
111 Opt_nouser_xattr,
112 Opt_acl,
113 Opt_noacl,
114 Opt_active_logs,
115 Opt_disable_ext_identify,
116 Opt_inline_xattr,
117 Opt_noinline_xattr,
118 Opt_inline_xattr_size,
119 Opt_inline_data,
120 Opt_inline_dentry,
121 Opt_noinline_dentry,
122 Opt_flush_merge,
123 Opt_noflush_merge,
124 Opt_barrier,
125 Opt_nobarrier,
126 Opt_fastboot,
127 Opt_extent_cache,
128 Opt_noextent_cache,
129 Opt_noinline_data,
130 Opt_data_flush,
131 Opt_reserve_root,
132 Opt_resgid,
133 Opt_resuid,
134 Opt_mode,
135 Opt_fault_injection,
136 Opt_fault_type,
137 Opt_lazytime,
138 Opt_nolazytime,
139 Opt_quota,
140 Opt_noquota,
141 Opt_usrquota,
142 Opt_grpquota,
143 Opt_prjquota,
144 Opt_usrjquota,
145 Opt_grpjquota,
146 Opt_prjjquota,
147 Opt_offusrjquota,
148 Opt_offgrpjquota,
149 Opt_offprjjquota,
150 Opt_jqfmt_vfsold,
151 Opt_jqfmt_vfsv0,
152 Opt_jqfmt_vfsv1,
153 Opt_alloc,
154 Opt_fsync,
155 Opt_test_dummy_encryption,
156 Opt_inlinecrypt,
157 Opt_checkpoint_disable,
158 Opt_checkpoint_disable_cap,
159 Opt_checkpoint_disable_cap_perc,
160 Opt_checkpoint_enable,
161 Opt_checkpoint_merge,
162 Opt_nocheckpoint_merge,
163 Opt_compress_algorithm,
164 Opt_compress_log_size,
165 Opt_compress_extension,
166 Opt_nocompress_extension,
167 Opt_compress_chksum,
168 Opt_compress_mode,
169 Opt_compress_cache,
170 Opt_atgc,
171 Opt_gc_merge,
172 Opt_nogc_merge,
173 Opt_discard_unit,
174 Opt_memory_mode,
175 Opt_age_extent_cache,
176 Opt_errors,
177 Opt_err,
178 };
179
180 static match_table_t f2fs_tokens = {
181 {Opt_gc_background, "background_gc=%s"},
182 {Opt_disable_roll_forward, "disable_roll_forward"},
183 {Opt_norecovery, "norecovery"},
184 {Opt_discard, "discard"},
185 {Opt_nodiscard, "nodiscard"},
186 {Opt_noheap, "no_heap"},
187 {Opt_heap, "heap"},
188 {Opt_user_xattr, "user_xattr"},
189 {Opt_nouser_xattr, "nouser_xattr"},
190 {Opt_acl, "acl"},
191 {Opt_noacl, "noacl"},
192 {Opt_active_logs, "active_logs=%u"},
193 {Opt_disable_ext_identify, "disable_ext_identify"},
194 {Opt_inline_xattr, "inline_xattr"},
195 {Opt_noinline_xattr, "noinline_xattr"},
196 {Opt_inline_xattr_size, "inline_xattr_size=%u"},
197 {Opt_inline_data, "inline_data"},
198 {Opt_inline_dentry, "inline_dentry"},
199 {Opt_noinline_dentry, "noinline_dentry"},
200 {Opt_flush_merge, "flush_merge"},
201 {Opt_noflush_merge, "noflush_merge"},
202 {Opt_barrier, "barrier"},
203 {Opt_nobarrier, "nobarrier"},
204 {Opt_fastboot, "fastboot"},
205 {Opt_extent_cache, "extent_cache"},
206 {Opt_noextent_cache, "noextent_cache"},
207 {Opt_noinline_data, "noinline_data"},
208 {Opt_data_flush, "data_flush"},
209 {Opt_reserve_root, "reserve_root=%u"},
210 {Opt_resgid, "resgid=%u"},
211 {Opt_resuid, "resuid=%u"},
212 {Opt_mode, "mode=%s"},
213 {Opt_fault_injection, "fault_injection=%u"},
214 {Opt_fault_type, "fault_type=%u"},
215 {Opt_lazytime, "lazytime"},
216 {Opt_nolazytime, "nolazytime"},
217 {Opt_quota, "quota"},
218 {Opt_noquota, "noquota"},
219 {Opt_usrquota, "usrquota"},
220 {Opt_grpquota, "grpquota"},
221 {Opt_prjquota, "prjquota"},
222 {Opt_usrjquota, "usrjquota=%s"},
223 {Opt_grpjquota, "grpjquota=%s"},
224 {Opt_prjjquota, "prjjquota=%s"},
225 {Opt_offusrjquota, "usrjquota="},
226 {Opt_offgrpjquota, "grpjquota="},
227 {Opt_offprjjquota, "prjjquota="},
228 {Opt_jqfmt_vfsold, "jqfmt=vfsold"},
229 {Opt_jqfmt_vfsv0, "jqfmt=vfsv0"},
230 {Opt_jqfmt_vfsv1, "jqfmt=vfsv1"},
231 {Opt_alloc, "alloc_mode=%s"},
232 {Opt_fsync, "fsync_mode=%s"},
233 {Opt_test_dummy_encryption, "test_dummy_encryption=%s"},
234 {Opt_test_dummy_encryption, "test_dummy_encryption"},
235 {Opt_inlinecrypt, "inlinecrypt"},
236 {Opt_checkpoint_disable, "checkpoint=disable"},
237 {Opt_checkpoint_disable_cap, "checkpoint=disable:%u"},
238 {Opt_checkpoint_disable_cap_perc, "checkpoint=disable:%u%%"},
239 {Opt_checkpoint_enable, "checkpoint=enable"},
240 {Opt_checkpoint_merge, "checkpoint_merge"},
241 {Opt_nocheckpoint_merge, "nocheckpoint_merge"},
242 {Opt_compress_algorithm, "compress_algorithm=%s"},
243 {Opt_compress_log_size, "compress_log_size=%u"},
244 {Opt_compress_extension, "compress_extension=%s"},
245 {Opt_nocompress_extension, "nocompress_extension=%s"},
246 {Opt_compress_chksum, "compress_chksum"},
247 {Opt_compress_mode, "compress_mode=%s"},
248 {Opt_compress_cache, "compress_cache"},
249 {Opt_atgc, "atgc"},
250 {Opt_gc_merge, "gc_merge"},
251 {Opt_nogc_merge, "nogc_merge"},
252 {Opt_discard_unit, "discard_unit=%s"},
253 {Opt_memory_mode, "memory=%s"},
254 {Opt_age_extent_cache, "age_extent_cache"},
255 {Opt_errors, "errors=%s"},
256 {Opt_err, NULL},
257 };
258
f2fs_printk(struct f2fs_sb_info * sbi,bool limit_rate,const char * fmt,...)259 void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate,
260 const char *fmt, ...)
261 {
262 struct va_format vaf;
263 va_list args;
264 int level;
265
266 va_start(args, fmt);
267
268 level = printk_get_level(fmt);
269 vaf.fmt = printk_skip_level(fmt);
270 vaf.va = &args;
271 if (limit_rate)
272 printk_ratelimited("%c%cF2FS-fs (%s): %pV\n",
273 KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
274 else
275 printk("%c%cF2FS-fs (%s): %pV\n",
276 KERN_SOH_ASCII, level, sbi->sb->s_id, &vaf);
277
278 va_end(args);
279 }
280
281 #if IS_ENABLED(CONFIG_UNICODE)
282 static const struct f2fs_sb_encodings {
283 __u16 magic;
284 char *name;
285 unsigned int version;
286 } f2fs_sb_encoding_map[] = {
287 {F2FS_ENC_UTF8_12_1, "utf8", UNICODE_AGE(12, 1, 0)},
288 };
289
290 static const struct f2fs_sb_encodings *
f2fs_sb_read_encoding(const struct f2fs_super_block * sb)291 f2fs_sb_read_encoding(const struct f2fs_super_block *sb)
292 {
293 __u16 magic = le16_to_cpu(sb->s_encoding);
294 int i;
295
296 for (i = 0; i < ARRAY_SIZE(f2fs_sb_encoding_map); i++)
297 if (magic == f2fs_sb_encoding_map[i].magic)
298 return &f2fs_sb_encoding_map[i];
299
300 return NULL;
301 }
302
303 struct kmem_cache *f2fs_cf_name_slab;
f2fs_create_casefold_cache(void)304 static int __init f2fs_create_casefold_cache(void)
305 {
306 f2fs_cf_name_slab = f2fs_kmem_cache_create("f2fs_casefolded_name",
307 F2FS_NAME_LEN);
308 return f2fs_cf_name_slab ? 0 : -ENOMEM;
309 }
310
f2fs_destroy_casefold_cache(void)311 static void f2fs_destroy_casefold_cache(void)
312 {
313 kmem_cache_destroy(f2fs_cf_name_slab);
314 }
315 #else
f2fs_create_casefold_cache(void)316 static int __init f2fs_create_casefold_cache(void) { return 0; }
f2fs_destroy_casefold_cache(void)317 static void f2fs_destroy_casefold_cache(void) { }
318 #endif
319
limit_reserve_root(struct f2fs_sb_info * sbi)320 static inline void limit_reserve_root(struct f2fs_sb_info *sbi)
321 {
322 block_t limit = min((sbi->user_block_count >> 3),
323 sbi->user_block_count - sbi->reserved_blocks);
324
325 /* limit is 12.5% */
326 if (test_opt(sbi, RESERVE_ROOT) &&
327 F2FS_OPTION(sbi).root_reserved_blocks > limit) {
328 F2FS_OPTION(sbi).root_reserved_blocks = limit;
329 f2fs_info(sbi, "Reduce reserved blocks for root = %u",
330 F2FS_OPTION(sbi).root_reserved_blocks);
331 }
332 if (!test_opt(sbi, RESERVE_ROOT) &&
333 (!uid_eq(F2FS_OPTION(sbi).s_resuid,
334 make_kuid(&init_user_ns, F2FS_DEF_RESUID)) ||
335 !gid_eq(F2FS_OPTION(sbi).s_resgid,
336 make_kgid(&init_user_ns, F2FS_DEF_RESGID))))
337 f2fs_info(sbi, "Ignore s_resuid=%u, s_resgid=%u w/o reserve_root",
338 from_kuid_munged(&init_user_ns,
339 F2FS_OPTION(sbi).s_resuid),
340 from_kgid_munged(&init_user_ns,
341 F2FS_OPTION(sbi).s_resgid));
342 }
343
adjust_unusable_cap_perc(struct f2fs_sb_info * sbi)344 static inline void adjust_unusable_cap_perc(struct f2fs_sb_info *sbi)
345 {
346 if (!F2FS_OPTION(sbi).unusable_cap_perc)
347 return;
348
349 if (F2FS_OPTION(sbi).unusable_cap_perc == 100)
350 F2FS_OPTION(sbi).unusable_cap = sbi->user_block_count;
351 else
352 F2FS_OPTION(sbi).unusable_cap = (sbi->user_block_count / 100) *
353 F2FS_OPTION(sbi).unusable_cap_perc;
354
355 f2fs_info(sbi, "Adjust unusable cap for checkpoint=disable = %u / %u%%",
356 F2FS_OPTION(sbi).unusable_cap,
357 F2FS_OPTION(sbi).unusable_cap_perc);
358 }
359
init_once(void * foo)360 static void init_once(void *foo)
361 {
362 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
363
364 inode_init_once(&fi->vfs_inode);
365 }
366
367 #ifdef CONFIG_QUOTA
368 static const char * const quotatypes[] = INITQFNAMES;
369 #define QTYPE2NAME(t) (quotatypes[t])
f2fs_set_qf_name(struct super_block * sb,int qtype,substring_t * args)370 static int f2fs_set_qf_name(struct super_block *sb, int qtype,
371 substring_t *args)
372 {
373 struct f2fs_sb_info *sbi = F2FS_SB(sb);
374 char *qname;
375 int ret = -EINVAL;
376
377 if (sb_any_quota_loaded(sb) && !F2FS_OPTION(sbi).s_qf_names[qtype]) {
378 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
379 return -EINVAL;
380 }
381 if (f2fs_sb_has_quota_ino(sbi)) {
382 f2fs_info(sbi, "QUOTA feature is enabled, so ignore qf_name");
383 return 0;
384 }
385
386 qname = match_strdup(args);
387 if (!qname) {
388 f2fs_err(sbi, "Not enough memory for storing quotafile name");
389 return -ENOMEM;
390 }
391 if (F2FS_OPTION(sbi).s_qf_names[qtype]) {
392 if (strcmp(F2FS_OPTION(sbi).s_qf_names[qtype], qname) == 0)
393 ret = 0;
394 else
395 f2fs_err(sbi, "%s quota file already specified",
396 QTYPE2NAME(qtype));
397 goto errout;
398 }
399 if (strchr(qname, '/')) {
400 f2fs_err(sbi, "quotafile must be on filesystem root");
401 goto errout;
402 }
403 F2FS_OPTION(sbi).s_qf_names[qtype] = qname;
404 set_opt(sbi, QUOTA);
405 return 0;
406 errout:
407 kfree(qname);
408 return ret;
409 }
410
f2fs_clear_qf_name(struct super_block * sb,int qtype)411 static int f2fs_clear_qf_name(struct super_block *sb, int qtype)
412 {
413 struct f2fs_sb_info *sbi = F2FS_SB(sb);
414
415 if (sb_any_quota_loaded(sb) && F2FS_OPTION(sbi).s_qf_names[qtype]) {
416 f2fs_err(sbi, "Cannot change journaled quota options when quota turned on");
417 return -EINVAL;
418 }
419 kfree(F2FS_OPTION(sbi).s_qf_names[qtype]);
420 F2FS_OPTION(sbi).s_qf_names[qtype] = NULL;
421 return 0;
422 }
423
f2fs_check_quota_options(struct f2fs_sb_info * sbi)424 static int f2fs_check_quota_options(struct f2fs_sb_info *sbi)
425 {
426 /*
427 * We do the test below only for project quotas. 'usrquota' and
428 * 'grpquota' mount options are allowed even without quota feature
429 * to support legacy quotas in quota files.
430 */
431 if (test_opt(sbi, PRJQUOTA) && !f2fs_sb_has_project_quota(sbi)) {
432 f2fs_err(sbi, "Project quota feature not enabled. Cannot enable project quota enforcement.");
433 return -1;
434 }
435 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] ||
436 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] ||
437 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) {
438 if (test_opt(sbi, USRQUOTA) &&
439 F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
440 clear_opt(sbi, USRQUOTA);
441
442 if (test_opt(sbi, GRPQUOTA) &&
443 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
444 clear_opt(sbi, GRPQUOTA);
445
446 if (test_opt(sbi, PRJQUOTA) &&
447 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
448 clear_opt(sbi, PRJQUOTA);
449
450 if (test_opt(sbi, GRPQUOTA) || test_opt(sbi, USRQUOTA) ||
451 test_opt(sbi, PRJQUOTA)) {
452 f2fs_err(sbi, "old and new quota format mixing");
453 return -1;
454 }
455
456 if (!F2FS_OPTION(sbi).s_jquota_fmt) {
457 f2fs_err(sbi, "journaled quota format not specified");
458 return -1;
459 }
460 }
461
462 if (f2fs_sb_has_quota_ino(sbi) && F2FS_OPTION(sbi).s_jquota_fmt) {
463 f2fs_info(sbi, "QUOTA feature is enabled, so ignore jquota_fmt");
464 F2FS_OPTION(sbi).s_jquota_fmt = 0;
465 }
466 return 0;
467 }
468 #endif
469
f2fs_set_test_dummy_encryption(struct super_block * sb,const char * opt,const substring_t * arg,bool is_remount)470 static int f2fs_set_test_dummy_encryption(struct super_block *sb,
471 const char *opt,
472 const substring_t *arg,
473 bool is_remount)
474 {
475 struct f2fs_sb_info *sbi = F2FS_SB(sb);
476 struct fs_parameter param = {
477 .type = fs_value_is_string,
478 .string = arg->from ? arg->from : "",
479 };
480 struct fscrypt_dummy_policy *policy =
481 &F2FS_OPTION(sbi).dummy_enc_policy;
482 int err;
483
484 if (!IS_ENABLED(CONFIG_FS_ENCRYPTION)) {
485 f2fs_warn(sbi, "test_dummy_encryption option not supported");
486 return -EINVAL;
487 }
488
489 if (!f2fs_sb_has_encrypt(sbi)) {
490 f2fs_err(sbi, "Encrypt feature is off");
491 return -EINVAL;
492 }
493
494 /*
495 * This mount option is just for testing, and it's not worthwhile to
496 * implement the extra complexity (e.g. RCU protection) that would be
497 * needed to allow it to be set or changed during remount. We do allow
498 * it to be specified during remount, but only if there is no change.
499 */
500 if (is_remount && !fscrypt_is_dummy_policy_set(policy)) {
501 f2fs_warn(sbi, "Can't set test_dummy_encryption on remount");
502 return -EINVAL;
503 }
504
505 err = fscrypt_parse_test_dummy_encryption(¶m, policy);
506 if (err) {
507 if (err == -EEXIST)
508 f2fs_warn(sbi,
509 "Can't change test_dummy_encryption on remount");
510 else if (err == -EINVAL)
511 f2fs_warn(sbi, "Value of option \"%s\" is unrecognized",
512 opt);
513 else
514 f2fs_warn(sbi, "Error processing option \"%s\" [%d]",
515 opt, err);
516 return -EINVAL;
517 }
518 f2fs_warn(sbi, "Test dummy encryption mode enabled");
519 return 0;
520 }
521
522 #ifdef CONFIG_F2FS_FS_COMPRESSION
is_compress_extension_exist(struct f2fs_sb_info * sbi,const char * new_ext,bool is_ext)523 static bool is_compress_extension_exist(struct f2fs_sb_info *sbi,
524 const char *new_ext, bool is_ext)
525 {
526 unsigned char (*ext)[F2FS_EXTENSION_LEN];
527 int ext_cnt;
528 int i;
529
530 if (is_ext) {
531 ext = F2FS_OPTION(sbi).extensions;
532 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
533 } else {
534 ext = F2FS_OPTION(sbi).noextensions;
535 ext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
536 }
537
538 for (i = 0; i < ext_cnt; i++) {
539 if (!strcasecmp(new_ext, ext[i]))
540 return true;
541 }
542
543 return false;
544 }
545
546 /*
547 * 1. The same extension name cannot not appear in both compress and non-compress extension
548 * at the same time.
549 * 2. If the compress extension specifies all files, the types specified by the non-compress
550 * extension will be treated as special cases and will not be compressed.
551 * 3. Don't allow the non-compress extension specifies all files.
552 */
f2fs_test_compress_extension(struct f2fs_sb_info * sbi)553 static int f2fs_test_compress_extension(struct f2fs_sb_info *sbi)
554 {
555 unsigned char (*ext)[F2FS_EXTENSION_LEN];
556 unsigned char (*noext)[F2FS_EXTENSION_LEN];
557 int ext_cnt, noext_cnt, index = 0, no_index = 0;
558
559 ext = F2FS_OPTION(sbi).extensions;
560 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
561 noext = F2FS_OPTION(sbi).noextensions;
562 noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
563
564 if (!noext_cnt)
565 return 0;
566
567 for (no_index = 0; no_index < noext_cnt; no_index++) {
568 if (!strcasecmp("*", noext[no_index])) {
569 f2fs_info(sbi, "Don't allow the nocompress extension specifies all files");
570 return -EINVAL;
571 }
572 for (index = 0; index < ext_cnt; index++) {
573 if (!strcasecmp(ext[index], noext[no_index])) {
574 f2fs_info(sbi, "Don't allow the same extension %s appear in both compress and nocompress extension",
575 ext[index]);
576 return -EINVAL;
577 }
578 }
579 }
580 return 0;
581 }
582
583 #ifdef CONFIG_F2FS_FS_LZ4
f2fs_set_lz4hc_level(struct f2fs_sb_info * sbi,const char * str)584 static int f2fs_set_lz4hc_level(struct f2fs_sb_info *sbi, const char *str)
585 {
586 #ifdef CONFIG_F2FS_FS_LZ4HC
587 unsigned int level;
588
589 if (strlen(str) == 3) {
590 F2FS_OPTION(sbi).compress_level = 0;
591 return 0;
592 }
593
594 str += 3;
595
596 if (str[0] != ':') {
597 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
598 return -EINVAL;
599 }
600 if (kstrtouint(str + 1, 10, &level))
601 return -EINVAL;
602
603 if (!f2fs_is_compress_level_valid(COMPRESS_LZ4, level)) {
604 f2fs_info(sbi, "invalid lz4hc compress level: %d", level);
605 return -EINVAL;
606 }
607
608 F2FS_OPTION(sbi).compress_level = level;
609 return 0;
610 #else
611 if (strlen(str) == 3) {
612 F2FS_OPTION(sbi).compress_level = 0;
613 return 0;
614 }
615 f2fs_info(sbi, "kernel doesn't support lz4hc compression");
616 return -EINVAL;
617 #endif
618 }
619 #endif
620
621 #ifdef CONFIG_F2FS_FS_ZSTD
f2fs_set_zstd_level(struct f2fs_sb_info * sbi,const char * str)622 static int f2fs_set_zstd_level(struct f2fs_sb_info *sbi, const char *str)
623 {
624 int level;
625 int len = 4;
626
627 if (strlen(str) == len) {
628 F2FS_OPTION(sbi).compress_level = F2FS_ZSTD_DEFAULT_CLEVEL;
629 return 0;
630 }
631
632 str += len;
633
634 if (str[0] != ':') {
635 f2fs_info(sbi, "wrong format, e.g. <alg_name>:<compr_level>");
636 return -EINVAL;
637 }
638 if (kstrtoint(str + 1, 10, &level))
639 return -EINVAL;
640
641 /* f2fs does not support negative compress level now */
642 if (level < 0) {
643 f2fs_info(sbi, "do not support negative compress level: %d", level);
644 return -ERANGE;
645 }
646
647 if (!f2fs_is_compress_level_valid(COMPRESS_ZSTD, level)) {
648 f2fs_info(sbi, "invalid zstd compress level: %d", level);
649 return -EINVAL;
650 }
651
652 F2FS_OPTION(sbi).compress_level = level;
653 return 0;
654 }
655 #endif
656 #endif
657
parse_options(struct super_block * sb,char * options,bool is_remount)658 static int parse_options(struct super_block *sb, char *options, bool is_remount)
659 {
660 struct f2fs_sb_info *sbi = F2FS_SB(sb);
661 substring_t args[MAX_OPT_ARGS];
662 #ifdef CONFIG_F2FS_FS_COMPRESSION
663 unsigned char (*ext)[F2FS_EXTENSION_LEN];
664 unsigned char (*noext)[F2FS_EXTENSION_LEN];
665 int ext_cnt, noext_cnt;
666 #endif
667 char *p, *name;
668 int arg = 0;
669 kuid_t uid;
670 kgid_t gid;
671 int ret;
672
673 if (!options)
674 goto default_check;
675
676 while ((p = strsep(&options, ",")) != NULL) {
677 int token;
678
679 if (!*p)
680 continue;
681 /*
682 * Initialize args struct so we know whether arg was
683 * found; some options take optional arguments.
684 */
685 args[0].to = args[0].from = NULL;
686 token = match_token(p, f2fs_tokens, args);
687
688 switch (token) {
689 case Opt_gc_background:
690 name = match_strdup(&args[0]);
691
692 if (!name)
693 return -ENOMEM;
694 if (!strcmp(name, "on")) {
695 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
696 } else if (!strcmp(name, "off")) {
697 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_OFF;
698 } else if (!strcmp(name, "sync")) {
699 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_SYNC;
700 } else {
701 kfree(name);
702 return -EINVAL;
703 }
704 kfree(name);
705 break;
706 case Opt_disable_roll_forward:
707 set_opt(sbi, DISABLE_ROLL_FORWARD);
708 break;
709 case Opt_norecovery:
710 /* this option mounts f2fs with ro */
711 set_opt(sbi, NORECOVERY);
712 if (!f2fs_readonly(sb))
713 return -EINVAL;
714 break;
715 case Opt_discard:
716 if (!f2fs_hw_support_discard(sbi)) {
717 f2fs_warn(sbi, "device does not support discard");
718 break;
719 }
720 set_opt(sbi, DISCARD);
721 break;
722 case Opt_nodiscard:
723 if (f2fs_hw_should_discard(sbi)) {
724 f2fs_warn(sbi, "discard is required for zoned block devices");
725 return -EINVAL;
726 }
727 clear_opt(sbi, DISCARD);
728 break;
729 case Opt_noheap:
730 case Opt_heap:
731 f2fs_warn(sbi, "heap/no_heap options were deprecated");
732 break;
733 #ifdef CONFIG_F2FS_FS_XATTR
734 case Opt_user_xattr:
735 set_opt(sbi, XATTR_USER);
736 break;
737 case Opt_nouser_xattr:
738 clear_opt(sbi, XATTR_USER);
739 break;
740 case Opt_inline_xattr:
741 set_opt(sbi, INLINE_XATTR);
742 break;
743 case Opt_noinline_xattr:
744 clear_opt(sbi, INLINE_XATTR);
745 break;
746 case Opt_inline_xattr_size:
747 if (args->from && match_int(args, &arg))
748 return -EINVAL;
749 set_opt(sbi, INLINE_XATTR_SIZE);
750 F2FS_OPTION(sbi).inline_xattr_size = arg;
751 break;
752 #else
753 case Opt_user_xattr:
754 f2fs_info(sbi, "user_xattr options not supported");
755 break;
756 case Opt_nouser_xattr:
757 f2fs_info(sbi, "nouser_xattr options not supported");
758 break;
759 case Opt_inline_xattr:
760 f2fs_info(sbi, "inline_xattr options not supported");
761 break;
762 case Opt_noinline_xattr:
763 f2fs_info(sbi, "noinline_xattr options not supported");
764 break;
765 #endif
766 #ifdef CONFIG_F2FS_FS_POSIX_ACL
767 case Opt_acl:
768 set_opt(sbi, POSIX_ACL);
769 break;
770 case Opt_noacl:
771 clear_opt(sbi, POSIX_ACL);
772 break;
773 #else
774 case Opt_acl:
775 f2fs_info(sbi, "acl options not supported");
776 break;
777 case Opt_noacl:
778 f2fs_info(sbi, "noacl options not supported");
779 break;
780 #endif
781 case Opt_active_logs:
782 if (args->from && match_int(args, &arg))
783 return -EINVAL;
784 if (arg != 2 && arg != 4 &&
785 arg != NR_CURSEG_PERSIST_TYPE)
786 return -EINVAL;
787 F2FS_OPTION(sbi).active_logs = arg;
788 break;
789 case Opt_disable_ext_identify:
790 set_opt(sbi, DISABLE_EXT_IDENTIFY);
791 break;
792 case Opt_inline_data:
793 set_opt(sbi, INLINE_DATA);
794 break;
795 case Opt_inline_dentry:
796 set_opt(sbi, INLINE_DENTRY);
797 break;
798 case Opt_noinline_dentry:
799 clear_opt(sbi, INLINE_DENTRY);
800 break;
801 case Opt_flush_merge:
802 set_opt(sbi, FLUSH_MERGE);
803 break;
804 case Opt_noflush_merge:
805 clear_opt(sbi, FLUSH_MERGE);
806 break;
807 case Opt_nobarrier:
808 set_opt(sbi, NOBARRIER);
809 break;
810 case Opt_barrier:
811 clear_opt(sbi, NOBARRIER);
812 break;
813 case Opt_fastboot:
814 set_opt(sbi, FASTBOOT);
815 break;
816 case Opt_extent_cache:
817 set_opt(sbi, READ_EXTENT_CACHE);
818 break;
819 case Opt_noextent_cache:
820 clear_opt(sbi, READ_EXTENT_CACHE);
821 break;
822 case Opt_noinline_data:
823 clear_opt(sbi, INLINE_DATA);
824 break;
825 case Opt_data_flush:
826 set_opt(sbi, DATA_FLUSH);
827 break;
828 case Opt_reserve_root:
829 if (args->from && match_int(args, &arg))
830 return -EINVAL;
831 if (test_opt(sbi, RESERVE_ROOT)) {
832 f2fs_info(sbi, "Preserve previous reserve_root=%u",
833 F2FS_OPTION(sbi).root_reserved_blocks);
834 } else {
835 F2FS_OPTION(sbi).root_reserved_blocks = arg;
836 set_opt(sbi, RESERVE_ROOT);
837 }
838 break;
839 case Opt_resuid:
840 if (args->from && match_int(args, &arg))
841 return -EINVAL;
842 uid = make_kuid(current_user_ns(), arg);
843 if (!uid_valid(uid)) {
844 f2fs_err(sbi, "Invalid uid value %d", arg);
845 return -EINVAL;
846 }
847 F2FS_OPTION(sbi).s_resuid = uid;
848 break;
849 case Opt_resgid:
850 if (args->from && match_int(args, &arg))
851 return -EINVAL;
852 gid = make_kgid(current_user_ns(), arg);
853 if (!gid_valid(gid)) {
854 f2fs_err(sbi, "Invalid gid value %d", arg);
855 return -EINVAL;
856 }
857 F2FS_OPTION(sbi).s_resgid = gid;
858 break;
859 case Opt_mode:
860 name = match_strdup(&args[0]);
861
862 if (!name)
863 return -ENOMEM;
864 if (!strcmp(name, "adaptive")) {
865 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
866 } else if (!strcmp(name, "lfs")) {
867 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
868 } else if (!strcmp(name, "fragment:segment")) {
869 F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_SEG;
870 } else if (!strcmp(name, "fragment:block")) {
871 F2FS_OPTION(sbi).fs_mode = FS_MODE_FRAGMENT_BLK;
872 } else {
873 kfree(name);
874 return -EINVAL;
875 }
876 kfree(name);
877 break;
878 #ifdef CONFIG_F2FS_FAULT_INJECTION
879 case Opt_fault_injection:
880 if (args->from && match_int(args, &arg))
881 return -EINVAL;
882 if (f2fs_build_fault_attr(sbi, arg,
883 F2FS_ALL_FAULT_TYPE))
884 return -EINVAL;
885 set_opt(sbi, FAULT_INJECTION);
886 break;
887
888 case Opt_fault_type:
889 if (args->from && match_int(args, &arg))
890 return -EINVAL;
891 if (f2fs_build_fault_attr(sbi, 0, arg))
892 return -EINVAL;
893 set_opt(sbi, FAULT_INJECTION);
894 break;
895 #else
896 case Opt_fault_injection:
897 f2fs_info(sbi, "fault_injection options not supported");
898 break;
899
900 case Opt_fault_type:
901 f2fs_info(sbi, "fault_type options not supported");
902 break;
903 #endif
904 case Opt_lazytime:
905 sb->s_flags |= SB_LAZYTIME;
906 break;
907 case Opt_nolazytime:
908 sb->s_flags &= ~SB_LAZYTIME;
909 break;
910 #ifdef CONFIG_QUOTA
911 case Opt_quota:
912 case Opt_usrquota:
913 set_opt(sbi, USRQUOTA);
914 break;
915 case Opt_grpquota:
916 set_opt(sbi, GRPQUOTA);
917 break;
918 case Opt_prjquota:
919 set_opt(sbi, PRJQUOTA);
920 break;
921 case Opt_usrjquota:
922 ret = f2fs_set_qf_name(sb, USRQUOTA, &args[0]);
923 if (ret)
924 return ret;
925 break;
926 case Opt_grpjquota:
927 ret = f2fs_set_qf_name(sb, GRPQUOTA, &args[0]);
928 if (ret)
929 return ret;
930 break;
931 case Opt_prjjquota:
932 ret = f2fs_set_qf_name(sb, PRJQUOTA, &args[0]);
933 if (ret)
934 return ret;
935 break;
936 case Opt_offusrjquota:
937 ret = f2fs_clear_qf_name(sb, USRQUOTA);
938 if (ret)
939 return ret;
940 break;
941 case Opt_offgrpjquota:
942 ret = f2fs_clear_qf_name(sb, GRPQUOTA);
943 if (ret)
944 return ret;
945 break;
946 case Opt_offprjjquota:
947 ret = f2fs_clear_qf_name(sb, PRJQUOTA);
948 if (ret)
949 return ret;
950 break;
951 case Opt_jqfmt_vfsold:
952 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_OLD;
953 break;
954 case Opt_jqfmt_vfsv0:
955 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V0;
956 break;
957 case Opt_jqfmt_vfsv1:
958 F2FS_OPTION(sbi).s_jquota_fmt = QFMT_VFS_V1;
959 break;
960 case Opt_noquota:
961 clear_opt(sbi, QUOTA);
962 clear_opt(sbi, USRQUOTA);
963 clear_opt(sbi, GRPQUOTA);
964 clear_opt(sbi, PRJQUOTA);
965 break;
966 #else
967 case Opt_quota:
968 case Opt_usrquota:
969 case Opt_grpquota:
970 case Opt_prjquota:
971 case Opt_usrjquota:
972 case Opt_grpjquota:
973 case Opt_prjjquota:
974 case Opt_offusrjquota:
975 case Opt_offgrpjquota:
976 case Opt_offprjjquota:
977 case Opt_jqfmt_vfsold:
978 case Opt_jqfmt_vfsv0:
979 case Opt_jqfmt_vfsv1:
980 case Opt_noquota:
981 f2fs_info(sbi, "quota operations not supported");
982 break;
983 #endif
984 case Opt_alloc:
985 name = match_strdup(&args[0]);
986 if (!name)
987 return -ENOMEM;
988
989 if (!strcmp(name, "default")) {
990 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
991 } else if (!strcmp(name, "reuse")) {
992 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
993 } else {
994 kfree(name);
995 return -EINVAL;
996 }
997 kfree(name);
998 break;
999 case Opt_fsync:
1000 name = match_strdup(&args[0]);
1001 if (!name)
1002 return -ENOMEM;
1003 if (!strcmp(name, "posix")) {
1004 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
1005 } else if (!strcmp(name, "strict")) {
1006 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_STRICT;
1007 } else if (!strcmp(name, "nobarrier")) {
1008 F2FS_OPTION(sbi).fsync_mode =
1009 FSYNC_MODE_NOBARRIER;
1010 } else {
1011 kfree(name);
1012 return -EINVAL;
1013 }
1014 kfree(name);
1015 break;
1016 case Opt_test_dummy_encryption:
1017 ret = f2fs_set_test_dummy_encryption(sb, p, &args[0],
1018 is_remount);
1019 if (ret)
1020 return ret;
1021 break;
1022 case Opt_inlinecrypt:
1023 #ifdef CONFIG_FS_ENCRYPTION_INLINE_CRYPT
1024 sb->s_flags |= SB_INLINECRYPT;
1025 #else
1026 f2fs_info(sbi, "inline encryption not supported");
1027 #endif
1028 break;
1029 case Opt_checkpoint_disable_cap_perc:
1030 if (args->from && match_int(args, &arg))
1031 return -EINVAL;
1032 if (arg < 0 || arg > 100)
1033 return -EINVAL;
1034 F2FS_OPTION(sbi).unusable_cap_perc = arg;
1035 set_opt(sbi, DISABLE_CHECKPOINT);
1036 break;
1037 case Opt_checkpoint_disable_cap:
1038 if (args->from && match_int(args, &arg))
1039 return -EINVAL;
1040 F2FS_OPTION(sbi).unusable_cap = arg;
1041 set_opt(sbi, DISABLE_CHECKPOINT);
1042 break;
1043 case Opt_checkpoint_disable:
1044 set_opt(sbi, DISABLE_CHECKPOINT);
1045 break;
1046 case Opt_checkpoint_enable:
1047 clear_opt(sbi, DISABLE_CHECKPOINT);
1048 break;
1049 case Opt_checkpoint_merge:
1050 set_opt(sbi, MERGE_CHECKPOINT);
1051 break;
1052 case Opt_nocheckpoint_merge:
1053 clear_opt(sbi, MERGE_CHECKPOINT);
1054 break;
1055 #ifdef CONFIG_F2FS_FS_COMPRESSION
1056 case Opt_compress_algorithm:
1057 if (!f2fs_sb_has_compression(sbi)) {
1058 f2fs_info(sbi, "Image doesn't support compression");
1059 break;
1060 }
1061 name = match_strdup(&args[0]);
1062 if (!name)
1063 return -ENOMEM;
1064 if (!strcmp(name, "lzo")) {
1065 #ifdef CONFIG_F2FS_FS_LZO
1066 F2FS_OPTION(sbi).compress_level = 0;
1067 F2FS_OPTION(sbi).compress_algorithm =
1068 COMPRESS_LZO;
1069 #else
1070 f2fs_info(sbi, "kernel doesn't support lzo compression");
1071 #endif
1072 } else if (!strncmp(name, "lz4", 3)) {
1073 #ifdef CONFIG_F2FS_FS_LZ4
1074 ret = f2fs_set_lz4hc_level(sbi, name);
1075 if (ret) {
1076 kfree(name);
1077 return -EINVAL;
1078 }
1079 F2FS_OPTION(sbi).compress_algorithm =
1080 COMPRESS_LZ4;
1081 #else
1082 f2fs_info(sbi, "kernel doesn't support lz4 compression");
1083 #endif
1084 } else if (!strncmp(name, "zstd", 4)) {
1085 #ifdef CONFIG_F2FS_FS_ZSTD
1086 ret = f2fs_set_zstd_level(sbi, name);
1087 if (ret) {
1088 kfree(name);
1089 return -EINVAL;
1090 }
1091 F2FS_OPTION(sbi).compress_algorithm =
1092 COMPRESS_ZSTD;
1093 #else
1094 f2fs_info(sbi, "kernel doesn't support zstd compression");
1095 #endif
1096 } else if (!strcmp(name, "lzo-rle")) {
1097 #ifdef CONFIG_F2FS_FS_LZORLE
1098 F2FS_OPTION(sbi).compress_level = 0;
1099 F2FS_OPTION(sbi).compress_algorithm =
1100 COMPRESS_LZORLE;
1101 #else
1102 f2fs_info(sbi, "kernel doesn't support lzorle compression");
1103 #endif
1104 } else {
1105 kfree(name);
1106 return -EINVAL;
1107 }
1108 kfree(name);
1109 break;
1110 case Opt_compress_log_size:
1111 if (!f2fs_sb_has_compression(sbi)) {
1112 f2fs_info(sbi, "Image doesn't support compression");
1113 break;
1114 }
1115 if (args->from && match_int(args, &arg))
1116 return -EINVAL;
1117 if (arg < MIN_COMPRESS_LOG_SIZE ||
1118 arg > MAX_COMPRESS_LOG_SIZE) {
1119 f2fs_err(sbi,
1120 "Compress cluster log size is out of range");
1121 return -EINVAL;
1122 }
1123 F2FS_OPTION(sbi).compress_log_size = arg;
1124 break;
1125 case Opt_compress_extension:
1126 if (!f2fs_sb_has_compression(sbi)) {
1127 f2fs_info(sbi, "Image doesn't support compression");
1128 break;
1129 }
1130 name = match_strdup(&args[0]);
1131 if (!name)
1132 return -ENOMEM;
1133
1134 ext = F2FS_OPTION(sbi).extensions;
1135 ext_cnt = F2FS_OPTION(sbi).compress_ext_cnt;
1136
1137 if (strlen(name) >= F2FS_EXTENSION_LEN ||
1138 ext_cnt >= COMPRESS_EXT_NUM) {
1139 f2fs_err(sbi,
1140 "invalid extension length/number");
1141 kfree(name);
1142 return -EINVAL;
1143 }
1144
1145 if (is_compress_extension_exist(sbi, name, true)) {
1146 kfree(name);
1147 break;
1148 }
1149
1150 strcpy(ext[ext_cnt], name);
1151 F2FS_OPTION(sbi).compress_ext_cnt++;
1152 kfree(name);
1153 break;
1154 case Opt_nocompress_extension:
1155 if (!f2fs_sb_has_compression(sbi)) {
1156 f2fs_info(sbi, "Image doesn't support compression");
1157 break;
1158 }
1159 name = match_strdup(&args[0]);
1160 if (!name)
1161 return -ENOMEM;
1162
1163 noext = F2FS_OPTION(sbi).noextensions;
1164 noext_cnt = F2FS_OPTION(sbi).nocompress_ext_cnt;
1165
1166 if (strlen(name) >= F2FS_EXTENSION_LEN ||
1167 noext_cnt >= COMPRESS_EXT_NUM) {
1168 f2fs_err(sbi,
1169 "invalid extension length/number");
1170 kfree(name);
1171 return -EINVAL;
1172 }
1173
1174 if (is_compress_extension_exist(sbi, name, false)) {
1175 kfree(name);
1176 break;
1177 }
1178
1179 strcpy(noext[noext_cnt], name);
1180 F2FS_OPTION(sbi).nocompress_ext_cnt++;
1181 kfree(name);
1182 break;
1183 case Opt_compress_chksum:
1184 if (!f2fs_sb_has_compression(sbi)) {
1185 f2fs_info(sbi, "Image doesn't support compression");
1186 break;
1187 }
1188 F2FS_OPTION(sbi).compress_chksum = true;
1189 break;
1190 case Opt_compress_mode:
1191 if (!f2fs_sb_has_compression(sbi)) {
1192 f2fs_info(sbi, "Image doesn't support compression");
1193 break;
1194 }
1195 name = match_strdup(&args[0]);
1196 if (!name)
1197 return -ENOMEM;
1198 if (!strcmp(name, "fs")) {
1199 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
1200 } else if (!strcmp(name, "user")) {
1201 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_USER;
1202 } else {
1203 kfree(name);
1204 return -EINVAL;
1205 }
1206 kfree(name);
1207 break;
1208 case Opt_compress_cache:
1209 if (!f2fs_sb_has_compression(sbi)) {
1210 f2fs_info(sbi, "Image doesn't support compression");
1211 break;
1212 }
1213 set_opt(sbi, COMPRESS_CACHE);
1214 break;
1215 #else
1216 case Opt_compress_algorithm:
1217 case Opt_compress_log_size:
1218 case Opt_compress_extension:
1219 case Opt_nocompress_extension:
1220 case Opt_compress_chksum:
1221 case Opt_compress_mode:
1222 case Opt_compress_cache:
1223 f2fs_info(sbi, "compression options not supported");
1224 break;
1225 #endif
1226 case Opt_atgc:
1227 set_opt(sbi, ATGC);
1228 break;
1229 case Opt_gc_merge:
1230 set_opt(sbi, GC_MERGE);
1231 break;
1232 case Opt_nogc_merge:
1233 clear_opt(sbi, GC_MERGE);
1234 break;
1235 case Opt_discard_unit:
1236 name = match_strdup(&args[0]);
1237 if (!name)
1238 return -ENOMEM;
1239 if (!strcmp(name, "block")) {
1240 F2FS_OPTION(sbi).discard_unit =
1241 DISCARD_UNIT_BLOCK;
1242 } else if (!strcmp(name, "segment")) {
1243 F2FS_OPTION(sbi).discard_unit =
1244 DISCARD_UNIT_SEGMENT;
1245 } else if (!strcmp(name, "section")) {
1246 F2FS_OPTION(sbi).discard_unit =
1247 DISCARD_UNIT_SECTION;
1248 } else {
1249 kfree(name);
1250 return -EINVAL;
1251 }
1252 kfree(name);
1253 break;
1254 case Opt_memory_mode:
1255 name = match_strdup(&args[0]);
1256 if (!name)
1257 return -ENOMEM;
1258 if (!strcmp(name, "normal")) {
1259 F2FS_OPTION(sbi).memory_mode =
1260 MEMORY_MODE_NORMAL;
1261 } else if (!strcmp(name, "low")) {
1262 F2FS_OPTION(sbi).memory_mode =
1263 MEMORY_MODE_LOW;
1264 } else {
1265 kfree(name);
1266 return -EINVAL;
1267 }
1268 kfree(name);
1269 break;
1270 case Opt_age_extent_cache:
1271 set_opt(sbi, AGE_EXTENT_CACHE);
1272 break;
1273 case Opt_errors:
1274 name = match_strdup(&args[0]);
1275 if (!name)
1276 return -ENOMEM;
1277 if (!strcmp(name, "remount-ro")) {
1278 F2FS_OPTION(sbi).errors =
1279 MOUNT_ERRORS_READONLY;
1280 } else if (!strcmp(name, "continue")) {
1281 F2FS_OPTION(sbi).errors =
1282 MOUNT_ERRORS_CONTINUE;
1283 } else if (!strcmp(name, "panic")) {
1284 F2FS_OPTION(sbi).errors =
1285 MOUNT_ERRORS_PANIC;
1286 } else {
1287 kfree(name);
1288 return -EINVAL;
1289 }
1290 kfree(name);
1291 break;
1292 default:
1293 f2fs_err(sbi, "Unrecognized mount option \"%s\" or missing value",
1294 p);
1295 return -EINVAL;
1296 }
1297 }
1298 default_check:
1299 #ifdef CONFIG_QUOTA
1300 if (f2fs_check_quota_options(sbi))
1301 return -EINVAL;
1302 #else
1303 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sbi->sb)) {
1304 f2fs_info(sbi, "Filesystem with quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1305 return -EINVAL;
1306 }
1307 if (f2fs_sb_has_project_quota(sbi) && !f2fs_readonly(sbi->sb)) {
1308 f2fs_err(sbi, "Filesystem with project quota feature cannot be mounted RDWR without CONFIG_QUOTA");
1309 return -EINVAL;
1310 }
1311 #endif
1312 #if !IS_ENABLED(CONFIG_UNICODE)
1313 if (f2fs_sb_has_casefold(sbi)) {
1314 f2fs_err(sbi,
1315 "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
1316 return -EINVAL;
1317 }
1318 #endif
1319 /*
1320 * The BLKZONED feature indicates that the drive was formatted with
1321 * zone alignment optimization. This is optional for host-aware
1322 * devices, but mandatory for host-managed zoned block devices.
1323 */
1324 if (f2fs_sb_has_blkzoned(sbi)) {
1325 #ifdef CONFIG_BLK_DEV_ZONED
1326 if (F2FS_OPTION(sbi).discard_unit !=
1327 DISCARD_UNIT_SECTION) {
1328 f2fs_info(sbi, "Zoned block device doesn't need small discard, set discard_unit=section by default");
1329 F2FS_OPTION(sbi).discard_unit =
1330 DISCARD_UNIT_SECTION;
1331 }
1332
1333 if (F2FS_OPTION(sbi).fs_mode != FS_MODE_LFS) {
1334 f2fs_info(sbi, "Only lfs mode is allowed with zoned block device feature");
1335 return -EINVAL;
1336 }
1337 #else
1338 f2fs_err(sbi, "Zoned block device support is not enabled");
1339 return -EINVAL;
1340 #endif
1341 }
1342
1343 #ifdef CONFIG_F2FS_FS_COMPRESSION
1344 if (f2fs_test_compress_extension(sbi)) {
1345 f2fs_err(sbi, "invalid compress or nocompress extension");
1346 return -EINVAL;
1347 }
1348 #endif
1349
1350 if (test_opt(sbi, INLINE_XATTR_SIZE)) {
1351 int min_size, max_size;
1352
1353 if (!f2fs_sb_has_extra_attr(sbi) ||
1354 !f2fs_sb_has_flexible_inline_xattr(sbi)) {
1355 f2fs_err(sbi, "extra_attr or flexible_inline_xattr feature is off");
1356 return -EINVAL;
1357 }
1358 if (!test_opt(sbi, INLINE_XATTR)) {
1359 f2fs_err(sbi, "inline_xattr_size option should be set with inline_xattr option");
1360 return -EINVAL;
1361 }
1362
1363 min_size = MIN_INLINE_XATTR_SIZE;
1364 max_size = MAX_INLINE_XATTR_SIZE;
1365
1366 if (F2FS_OPTION(sbi).inline_xattr_size < min_size ||
1367 F2FS_OPTION(sbi).inline_xattr_size > max_size) {
1368 f2fs_err(sbi, "inline xattr size is out of range: %d ~ %d",
1369 min_size, max_size);
1370 return -EINVAL;
1371 }
1372 }
1373
1374 if (test_opt(sbi, DISABLE_CHECKPOINT) && f2fs_lfs_mode(sbi)) {
1375 f2fs_err(sbi, "LFS is not compatible with checkpoint=disable");
1376 return -EINVAL;
1377 }
1378
1379 if (test_opt(sbi, ATGC) && f2fs_lfs_mode(sbi)) {
1380 f2fs_err(sbi, "LFS is not compatible with ATGC");
1381 return -EINVAL;
1382 }
1383
1384 if (f2fs_is_readonly(sbi) && test_opt(sbi, FLUSH_MERGE)) {
1385 f2fs_err(sbi, "FLUSH_MERGE not compatible with readonly mode");
1386 return -EINVAL;
1387 }
1388
1389 if (f2fs_sb_has_readonly(sbi) && !f2fs_readonly(sbi->sb)) {
1390 f2fs_err(sbi, "Allow to mount readonly mode only");
1391 return -EROFS;
1392 }
1393 return 0;
1394 }
1395
f2fs_alloc_inode(struct super_block * sb)1396 static struct inode *f2fs_alloc_inode(struct super_block *sb)
1397 {
1398 struct f2fs_inode_info *fi;
1399
1400 if (time_to_inject(F2FS_SB(sb), FAULT_SLAB_ALLOC))
1401 return NULL;
1402
1403 fi = alloc_inode_sb(sb, f2fs_inode_cachep, GFP_F2FS_ZERO);
1404 if (!fi)
1405 return NULL;
1406
1407 init_once((void *) fi);
1408
1409 /* Initialize f2fs-specific inode info */
1410 atomic_set(&fi->dirty_pages, 0);
1411 atomic_set(&fi->i_compr_blocks, 0);
1412 init_f2fs_rwsem(&fi->i_sem);
1413 spin_lock_init(&fi->i_size_lock);
1414 INIT_LIST_HEAD(&fi->dirty_list);
1415 INIT_LIST_HEAD(&fi->gdirty_list);
1416 init_f2fs_rwsem(&fi->i_gc_rwsem[READ]);
1417 init_f2fs_rwsem(&fi->i_gc_rwsem[WRITE]);
1418 init_f2fs_rwsem(&fi->i_xattr_sem);
1419
1420 /* Will be used by directory only */
1421 fi->i_dir_level = F2FS_SB(sb)->dir_level;
1422
1423 return &fi->vfs_inode;
1424 }
1425
f2fs_drop_inode(struct inode * inode)1426 static int f2fs_drop_inode(struct inode *inode)
1427 {
1428 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1429 int ret;
1430
1431 /*
1432 * during filesystem shutdown, if checkpoint is disabled,
1433 * drop useless meta/node dirty pages.
1434 */
1435 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) {
1436 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1437 inode->i_ino == F2FS_META_INO(sbi)) {
1438 trace_f2fs_drop_inode(inode, 1);
1439 return 1;
1440 }
1441 }
1442
1443 /*
1444 * This is to avoid a deadlock condition like below.
1445 * writeback_single_inode(inode)
1446 * - f2fs_write_data_page
1447 * - f2fs_gc -> iput -> evict
1448 * - inode_wait_for_writeback(inode)
1449 */
1450 if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
1451 if (!inode->i_nlink && !is_bad_inode(inode)) {
1452 /* to avoid evict_inode call simultaneously */
1453 atomic_inc(&inode->i_count);
1454 spin_unlock(&inode->i_lock);
1455
1456 /* should remain fi->extent_tree for writepage */
1457 f2fs_destroy_extent_node(inode);
1458
1459 sb_start_intwrite(inode->i_sb);
1460 f2fs_i_size_write(inode, 0);
1461
1462 f2fs_submit_merged_write_cond(F2FS_I_SB(inode),
1463 inode, NULL, 0, DATA);
1464 truncate_inode_pages_final(inode->i_mapping);
1465
1466 if (F2FS_HAS_BLOCKS(inode))
1467 f2fs_truncate(inode);
1468
1469 sb_end_intwrite(inode->i_sb);
1470
1471 spin_lock(&inode->i_lock);
1472 atomic_dec(&inode->i_count);
1473 }
1474 trace_f2fs_drop_inode(inode, 0);
1475 return 0;
1476 }
1477 ret = generic_drop_inode(inode);
1478 if (!ret)
1479 ret = fscrypt_drop_inode(inode);
1480 trace_f2fs_drop_inode(inode, ret);
1481 return ret;
1482 }
1483
f2fs_inode_dirtied(struct inode * inode,bool sync)1484 int f2fs_inode_dirtied(struct inode *inode, bool sync)
1485 {
1486 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1487 int ret = 0;
1488
1489 spin_lock(&sbi->inode_lock[DIRTY_META]);
1490 if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1491 ret = 1;
1492 } else {
1493 set_inode_flag(inode, FI_DIRTY_INODE);
1494 stat_inc_dirty_inode(sbi, DIRTY_META);
1495 }
1496 if (sync && list_empty(&F2FS_I(inode)->gdirty_list)) {
1497 list_add_tail(&F2FS_I(inode)->gdirty_list,
1498 &sbi->inode_list[DIRTY_META]);
1499 inc_page_count(sbi, F2FS_DIRTY_IMETA);
1500 }
1501 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1502 return ret;
1503 }
1504
f2fs_inode_synced(struct inode * inode)1505 void f2fs_inode_synced(struct inode *inode)
1506 {
1507 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1508
1509 spin_lock(&sbi->inode_lock[DIRTY_META]);
1510 if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
1511 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1512 return;
1513 }
1514 if (!list_empty(&F2FS_I(inode)->gdirty_list)) {
1515 list_del_init(&F2FS_I(inode)->gdirty_list);
1516 dec_page_count(sbi, F2FS_DIRTY_IMETA);
1517 }
1518 clear_inode_flag(inode, FI_DIRTY_INODE);
1519 clear_inode_flag(inode, FI_AUTO_RECOVER);
1520 stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
1521 spin_unlock(&sbi->inode_lock[DIRTY_META]);
1522 }
1523
1524 /*
1525 * f2fs_dirty_inode() is called from __mark_inode_dirty()
1526 *
1527 * We should call set_dirty_inode to write the dirty inode through write_inode.
1528 */
f2fs_dirty_inode(struct inode * inode,int flags)1529 static void f2fs_dirty_inode(struct inode *inode, int flags)
1530 {
1531 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1532
1533 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
1534 inode->i_ino == F2FS_META_INO(sbi))
1535 return;
1536
1537 if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
1538 clear_inode_flag(inode, FI_AUTO_RECOVER);
1539
1540 f2fs_inode_dirtied(inode, false);
1541 }
1542
f2fs_free_inode(struct inode * inode)1543 static void f2fs_free_inode(struct inode *inode)
1544 {
1545 fscrypt_free_inode(inode);
1546 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
1547 }
1548
destroy_percpu_info(struct f2fs_sb_info * sbi)1549 static void destroy_percpu_info(struct f2fs_sb_info *sbi)
1550 {
1551 percpu_counter_destroy(&sbi->total_valid_inode_count);
1552 percpu_counter_destroy(&sbi->rf_node_block_count);
1553 percpu_counter_destroy(&sbi->alloc_valid_block_count);
1554 }
1555
destroy_device_list(struct f2fs_sb_info * sbi)1556 static void destroy_device_list(struct f2fs_sb_info *sbi)
1557 {
1558 int i;
1559
1560 for (i = 0; i < sbi->s_ndevs; i++) {
1561 if (i > 0)
1562 blkdev_put(FDEV(i).bdev, sbi->sb);
1563 #ifdef CONFIG_BLK_DEV_ZONED
1564 kvfree(FDEV(i).blkz_seq);
1565 #endif
1566 }
1567 kvfree(sbi->devs);
1568 }
1569
f2fs_put_super(struct super_block * sb)1570 static void f2fs_put_super(struct super_block *sb)
1571 {
1572 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1573 int i;
1574 int err = 0;
1575 bool done;
1576
1577 /* unregister procfs/sysfs entries in advance to avoid race case */
1578 f2fs_unregister_sysfs(sbi);
1579
1580 f2fs_quota_off_umount(sb);
1581
1582 /* prevent remaining shrinker jobs */
1583 mutex_lock(&sbi->umount_mutex);
1584
1585 /*
1586 * flush all issued checkpoints and stop checkpoint issue thread.
1587 * after then, all checkpoints should be done by each process context.
1588 */
1589 f2fs_stop_ckpt_thread(sbi);
1590
1591 /*
1592 * We don't need to do checkpoint when superblock is clean.
1593 * But, the previous checkpoint was not done by umount, it needs to do
1594 * clean checkpoint again.
1595 */
1596 if ((is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
1597 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG))) {
1598 struct cp_control cpc = {
1599 .reason = CP_UMOUNT,
1600 };
1601 stat_inc_cp_call_count(sbi, TOTAL_CALL);
1602 err = f2fs_write_checkpoint(sbi, &cpc);
1603 }
1604
1605 /* be sure to wait for any on-going discard commands */
1606 done = f2fs_issue_discard_timeout(sbi);
1607 if (f2fs_realtime_discard_enable(sbi) && !sbi->discard_blks && done) {
1608 struct cp_control cpc = {
1609 .reason = CP_UMOUNT | CP_TRIMMED,
1610 };
1611 stat_inc_cp_call_count(sbi, TOTAL_CALL);
1612 err = f2fs_write_checkpoint(sbi, &cpc);
1613 }
1614
1615 /*
1616 * normally superblock is clean, so we need to release this.
1617 * In addition, EIO will skip do checkpoint, we need this as well.
1618 */
1619 f2fs_release_ino_entry(sbi, true);
1620
1621 f2fs_leave_shrinker(sbi);
1622 mutex_unlock(&sbi->umount_mutex);
1623
1624 /* our cp_error case, we can wait for any writeback page */
1625 f2fs_flush_merged_writes(sbi);
1626
1627 f2fs_wait_on_all_pages(sbi, F2FS_WB_CP_DATA);
1628
1629 if (err || f2fs_cp_error(sbi)) {
1630 truncate_inode_pages_final(NODE_MAPPING(sbi));
1631 truncate_inode_pages_final(META_MAPPING(sbi));
1632 }
1633
1634 for (i = 0; i < NR_COUNT_TYPE; i++) {
1635 if (!get_pages(sbi, i))
1636 continue;
1637 f2fs_err(sbi, "detect filesystem reference count leak during "
1638 "umount, type: %d, count: %lld", i, get_pages(sbi, i));
1639 f2fs_bug_on(sbi, 1);
1640 }
1641
1642 f2fs_bug_on(sbi, sbi->fsync_node_num);
1643
1644 f2fs_destroy_compress_inode(sbi);
1645
1646 iput(sbi->node_inode);
1647 sbi->node_inode = NULL;
1648
1649 iput(sbi->meta_inode);
1650 sbi->meta_inode = NULL;
1651
1652 /*
1653 * iput() can update stat information, if f2fs_write_checkpoint()
1654 * above failed with error.
1655 */
1656 f2fs_destroy_stats(sbi);
1657
1658 /* destroy f2fs internal modules */
1659 f2fs_destroy_node_manager(sbi);
1660 f2fs_destroy_segment_manager(sbi);
1661
1662 /* flush s_error_work before sbi destroy */
1663 flush_work(&sbi->s_error_work);
1664
1665 f2fs_destroy_post_read_wq(sbi);
1666
1667 kvfree(sbi->ckpt);
1668
1669 sb->s_fs_info = NULL;
1670 if (sbi->s_chksum_driver)
1671 crypto_free_shash(sbi->s_chksum_driver);
1672 kfree(sbi->raw_super);
1673
1674 destroy_device_list(sbi);
1675 f2fs_destroy_page_array_cache(sbi);
1676 f2fs_destroy_xattr_caches(sbi);
1677 #ifdef CONFIG_QUOTA
1678 for (i = 0; i < MAXQUOTAS; i++)
1679 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
1680 #endif
1681 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
1682 destroy_percpu_info(sbi);
1683 f2fs_destroy_iostat(sbi);
1684 for (i = 0; i < NR_PAGE_TYPE; i++)
1685 kvfree(sbi->write_io[i]);
1686 #if IS_ENABLED(CONFIG_UNICODE)
1687 utf8_unload(sb->s_encoding);
1688 #endif
1689 kfree(sbi);
1690 }
1691
f2fs_sync_fs(struct super_block * sb,int sync)1692 int f2fs_sync_fs(struct super_block *sb, int sync)
1693 {
1694 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1695 int err = 0;
1696
1697 if (unlikely(f2fs_cp_error(sbi)))
1698 return 0;
1699 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED)))
1700 return 0;
1701
1702 trace_f2fs_sync_fs(sb, sync);
1703
1704 if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING)))
1705 return -EAGAIN;
1706
1707 if (sync) {
1708 stat_inc_cp_call_count(sbi, TOTAL_CALL);
1709 err = f2fs_issue_checkpoint(sbi);
1710 }
1711
1712 return err;
1713 }
1714
f2fs_freeze(struct super_block * sb)1715 static int f2fs_freeze(struct super_block *sb)
1716 {
1717 if (f2fs_readonly(sb))
1718 return 0;
1719
1720 /* IO error happened before */
1721 if (unlikely(f2fs_cp_error(F2FS_SB(sb))))
1722 return -EIO;
1723
1724 /* must be clean, since sync_filesystem() was already called */
1725 if (is_sbi_flag_set(F2FS_SB(sb), SBI_IS_DIRTY))
1726 return -EINVAL;
1727
1728 /* Let's flush checkpoints and stop the thread. */
1729 f2fs_flush_ckpt_thread(F2FS_SB(sb));
1730
1731 /* to avoid deadlock on f2fs_evict_inode->SB_FREEZE_FS */
1732 set_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1733 return 0;
1734 }
1735
f2fs_unfreeze(struct super_block * sb)1736 static int f2fs_unfreeze(struct super_block *sb)
1737 {
1738 clear_sbi_flag(F2FS_SB(sb), SBI_IS_FREEZING);
1739 return 0;
1740 }
1741
1742 #ifdef CONFIG_QUOTA
f2fs_statfs_project(struct super_block * sb,kprojid_t projid,struct kstatfs * buf)1743 static int f2fs_statfs_project(struct super_block *sb,
1744 kprojid_t projid, struct kstatfs *buf)
1745 {
1746 struct kqid qid;
1747 struct dquot *dquot;
1748 u64 limit;
1749 u64 curblock;
1750
1751 qid = make_kqid_projid(projid);
1752 dquot = dqget(sb, qid);
1753 if (IS_ERR(dquot))
1754 return PTR_ERR(dquot);
1755 spin_lock(&dquot->dq_dqb_lock);
1756
1757 limit = min_not_zero(dquot->dq_dqb.dqb_bsoftlimit,
1758 dquot->dq_dqb.dqb_bhardlimit);
1759 if (limit)
1760 limit >>= sb->s_blocksize_bits;
1761
1762 if (limit && buf->f_blocks > limit) {
1763 curblock = (dquot->dq_dqb.dqb_curspace +
1764 dquot->dq_dqb.dqb_rsvspace) >> sb->s_blocksize_bits;
1765 buf->f_blocks = limit;
1766 buf->f_bfree = buf->f_bavail =
1767 (buf->f_blocks > curblock) ?
1768 (buf->f_blocks - curblock) : 0;
1769 }
1770
1771 limit = min_not_zero(dquot->dq_dqb.dqb_isoftlimit,
1772 dquot->dq_dqb.dqb_ihardlimit);
1773
1774 if (limit && buf->f_files > limit) {
1775 buf->f_files = limit;
1776 buf->f_ffree =
1777 (buf->f_files > dquot->dq_dqb.dqb_curinodes) ?
1778 (buf->f_files - dquot->dq_dqb.dqb_curinodes) : 0;
1779 }
1780
1781 spin_unlock(&dquot->dq_dqb_lock);
1782 dqput(dquot);
1783 return 0;
1784 }
1785 #endif
1786
f2fs_statfs(struct dentry * dentry,struct kstatfs * buf)1787 static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
1788 {
1789 struct super_block *sb = dentry->d_sb;
1790 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1791 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
1792 block_t total_count, user_block_count, start_count;
1793 u64 avail_node_count;
1794 unsigned int total_valid_node_count;
1795
1796 total_count = le64_to_cpu(sbi->raw_super->block_count);
1797 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
1798 buf->f_type = F2FS_SUPER_MAGIC;
1799 buf->f_bsize = sbi->blocksize;
1800
1801 buf->f_blocks = total_count - start_count;
1802
1803 spin_lock(&sbi->stat_lock);
1804
1805 user_block_count = sbi->user_block_count;
1806 total_valid_node_count = valid_node_count(sbi);
1807 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
1808 buf->f_bfree = user_block_count - valid_user_blocks(sbi) -
1809 sbi->current_reserved_blocks;
1810
1811 if (unlikely(buf->f_bfree <= sbi->unusable_block_count))
1812 buf->f_bfree = 0;
1813 else
1814 buf->f_bfree -= sbi->unusable_block_count;
1815 spin_unlock(&sbi->stat_lock);
1816
1817 if (buf->f_bfree > F2FS_OPTION(sbi).root_reserved_blocks)
1818 buf->f_bavail = buf->f_bfree -
1819 F2FS_OPTION(sbi).root_reserved_blocks;
1820 else
1821 buf->f_bavail = 0;
1822
1823 if (avail_node_count > user_block_count) {
1824 buf->f_files = user_block_count;
1825 buf->f_ffree = buf->f_bavail;
1826 } else {
1827 buf->f_files = avail_node_count;
1828 buf->f_ffree = min(avail_node_count - total_valid_node_count,
1829 buf->f_bavail);
1830 }
1831
1832 buf->f_namelen = F2FS_NAME_LEN;
1833 buf->f_fsid = u64_to_fsid(id);
1834
1835 #ifdef CONFIG_QUOTA
1836 if (is_inode_flag_set(dentry->d_inode, FI_PROJ_INHERIT) &&
1837 sb_has_quota_limits_enabled(sb, PRJQUOTA)) {
1838 f2fs_statfs_project(sb, F2FS_I(dentry->d_inode)->i_projid, buf);
1839 }
1840 #endif
1841 return 0;
1842 }
1843
f2fs_show_quota_options(struct seq_file * seq,struct super_block * sb)1844 static inline void f2fs_show_quota_options(struct seq_file *seq,
1845 struct super_block *sb)
1846 {
1847 #ifdef CONFIG_QUOTA
1848 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1849
1850 if (F2FS_OPTION(sbi).s_jquota_fmt) {
1851 char *fmtname = "";
1852
1853 switch (F2FS_OPTION(sbi).s_jquota_fmt) {
1854 case QFMT_VFS_OLD:
1855 fmtname = "vfsold";
1856 break;
1857 case QFMT_VFS_V0:
1858 fmtname = "vfsv0";
1859 break;
1860 case QFMT_VFS_V1:
1861 fmtname = "vfsv1";
1862 break;
1863 }
1864 seq_printf(seq, ",jqfmt=%s", fmtname);
1865 }
1866
1867 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA])
1868 seq_show_option(seq, "usrjquota",
1869 F2FS_OPTION(sbi).s_qf_names[USRQUOTA]);
1870
1871 if (F2FS_OPTION(sbi).s_qf_names[GRPQUOTA])
1872 seq_show_option(seq, "grpjquota",
1873 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA]);
1874
1875 if (F2FS_OPTION(sbi).s_qf_names[PRJQUOTA])
1876 seq_show_option(seq, "prjjquota",
1877 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]);
1878 #endif
1879 }
1880
1881 #ifdef CONFIG_F2FS_FS_COMPRESSION
f2fs_show_compress_options(struct seq_file * seq,struct super_block * sb)1882 static inline void f2fs_show_compress_options(struct seq_file *seq,
1883 struct super_block *sb)
1884 {
1885 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1886 char *algtype = "";
1887 int i;
1888
1889 if (!f2fs_sb_has_compression(sbi))
1890 return;
1891
1892 switch (F2FS_OPTION(sbi).compress_algorithm) {
1893 case COMPRESS_LZO:
1894 algtype = "lzo";
1895 break;
1896 case COMPRESS_LZ4:
1897 algtype = "lz4";
1898 break;
1899 case COMPRESS_ZSTD:
1900 algtype = "zstd";
1901 break;
1902 case COMPRESS_LZORLE:
1903 algtype = "lzo-rle";
1904 break;
1905 }
1906 seq_printf(seq, ",compress_algorithm=%s", algtype);
1907
1908 if (F2FS_OPTION(sbi).compress_level)
1909 seq_printf(seq, ":%d", F2FS_OPTION(sbi).compress_level);
1910
1911 seq_printf(seq, ",compress_log_size=%u",
1912 F2FS_OPTION(sbi).compress_log_size);
1913
1914 for (i = 0; i < F2FS_OPTION(sbi).compress_ext_cnt; i++) {
1915 seq_printf(seq, ",compress_extension=%s",
1916 F2FS_OPTION(sbi).extensions[i]);
1917 }
1918
1919 for (i = 0; i < F2FS_OPTION(sbi).nocompress_ext_cnt; i++) {
1920 seq_printf(seq, ",nocompress_extension=%s",
1921 F2FS_OPTION(sbi).noextensions[i]);
1922 }
1923
1924 if (F2FS_OPTION(sbi).compress_chksum)
1925 seq_puts(seq, ",compress_chksum");
1926
1927 if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_FS)
1928 seq_printf(seq, ",compress_mode=%s", "fs");
1929 else if (F2FS_OPTION(sbi).compress_mode == COMPR_MODE_USER)
1930 seq_printf(seq, ",compress_mode=%s", "user");
1931
1932 if (test_opt(sbi, COMPRESS_CACHE))
1933 seq_puts(seq, ",compress_cache");
1934 }
1935 #endif
1936
f2fs_show_options(struct seq_file * seq,struct dentry * root)1937 static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
1938 {
1939 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
1940
1941 if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_SYNC)
1942 seq_printf(seq, ",background_gc=%s", "sync");
1943 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_ON)
1944 seq_printf(seq, ",background_gc=%s", "on");
1945 else if (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF)
1946 seq_printf(seq, ",background_gc=%s", "off");
1947
1948 if (test_opt(sbi, GC_MERGE))
1949 seq_puts(seq, ",gc_merge");
1950 else
1951 seq_puts(seq, ",nogc_merge");
1952
1953 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
1954 seq_puts(seq, ",disable_roll_forward");
1955 if (test_opt(sbi, NORECOVERY))
1956 seq_puts(seq, ",norecovery");
1957 if (test_opt(sbi, DISCARD)) {
1958 seq_puts(seq, ",discard");
1959 if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK)
1960 seq_printf(seq, ",discard_unit=%s", "block");
1961 else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SEGMENT)
1962 seq_printf(seq, ",discard_unit=%s", "segment");
1963 else if (F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_SECTION)
1964 seq_printf(seq, ",discard_unit=%s", "section");
1965 } else {
1966 seq_puts(seq, ",nodiscard");
1967 }
1968 #ifdef CONFIG_F2FS_FS_XATTR
1969 if (test_opt(sbi, XATTR_USER))
1970 seq_puts(seq, ",user_xattr");
1971 else
1972 seq_puts(seq, ",nouser_xattr");
1973 if (test_opt(sbi, INLINE_XATTR))
1974 seq_puts(seq, ",inline_xattr");
1975 else
1976 seq_puts(seq, ",noinline_xattr");
1977 if (test_opt(sbi, INLINE_XATTR_SIZE))
1978 seq_printf(seq, ",inline_xattr_size=%u",
1979 F2FS_OPTION(sbi).inline_xattr_size);
1980 #endif
1981 #ifdef CONFIG_F2FS_FS_POSIX_ACL
1982 if (test_opt(sbi, POSIX_ACL))
1983 seq_puts(seq, ",acl");
1984 else
1985 seq_puts(seq, ",noacl");
1986 #endif
1987 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
1988 seq_puts(seq, ",disable_ext_identify");
1989 if (test_opt(sbi, INLINE_DATA))
1990 seq_puts(seq, ",inline_data");
1991 else
1992 seq_puts(seq, ",noinline_data");
1993 if (test_opt(sbi, INLINE_DENTRY))
1994 seq_puts(seq, ",inline_dentry");
1995 else
1996 seq_puts(seq, ",noinline_dentry");
1997 if (test_opt(sbi, FLUSH_MERGE))
1998 seq_puts(seq, ",flush_merge");
1999 else
2000 seq_puts(seq, ",noflush_merge");
2001 if (test_opt(sbi, NOBARRIER))
2002 seq_puts(seq, ",nobarrier");
2003 else
2004 seq_puts(seq, ",barrier");
2005 if (test_opt(sbi, FASTBOOT))
2006 seq_puts(seq, ",fastboot");
2007 if (test_opt(sbi, READ_EXTENT_CACHE))
2008 seq_puts(seq, ",extent_cache");
2009 else
2010 seq_puts(seq, ",noextent_cache");
2011 if (test_opt(sbi, AGE_EXTENT_CACHE))
2012 seq_puts(seq, ",age_extent_cache");
2013 if (test_opt(sbi, DATA_FLUSH))
2014 seq_puts(seq, ",data_flush");
2015
2016 seq_puts(seq, ",mode=");
2017 if (F2FS_OPTION(sbi).fs_mode == FS_MODE_ADAPTIVE)
2018 seq_puts(seq, "adaptive");
2019 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS)
2020 seq_puts(seq, "lfs");
2021 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG)
2022 seq_puts(seq, "fragment:segment");
2023 else if (F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK)
2024 seq_puts(seq, "fragment:block");
2025 seq_printf(seq, ",active_logs=%u", F2FS_OPTION(sbi).active_logs);
2026 if (test_opt(sbi, RESERVE_ROOT))
2027 seq_printf(seq, ",reserve_root=%u,resuid=%u,resgid=%u",
2028 F2FS_OPTION(sbi).root_reserved_blocks,
2029 from_kuid_munged(&init_user_ns,
2030 F2FS_OPTION(sbi).s_resuid),
2031 from_kgid_munged(&init_user_ns,
2032 F2FS_OPTION(sbi).s_resgid));
2033 #ifdef CONFIG_F2FS_FAULT_INJECTION
2034 if (test_opt(sbi, FAULT_INJECTION)) {
2035 seq_printf(seq, ",fault_injection=%u",
2036 F2FS_OPTION(sbi).fault_info.inject_rate);
2037 seq_printf(seq, ",fault_type=%u",
2038 F2FS_OPTION(sbi).fault_info.inject_type);
2039 }
2040 #endif
2041 #ifdef CONFIG_QUOTA
2042 if (test_opt(sbi, QUOTA))
2043 seq_puts(seq, ",quota");
2044 if (test_opt(sbi, USRQUOTA))
2045 seq_puts(seq, ",usrquota");
2046 if (test_opt(sbi, GRPQUOTA))
2047 seq_puts(seq, ",grpquota");
2048 if (test_opt(sbi, PRJQUOTA))
2049 seq_puts(seq, ",prjquota");
2050 #endif
2051 f2fs_show_quota_options(seq, sbi->sb);
2052
2053 fscrypt_show_test_dummy_encryption(seq, ',', sbi->sb);
2054
2055 if (sbi->sb->s_flags & SB_INLINECRYPT)
2056 seq_puts(seq, ",inlinecrypt");
2057
2058 if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_DEFAULT)
2059 seq_printf(seq, ",alloc_mode=%s", "default");
2060 else if (F2FS_OPTION(sbi).alloc_mode == ALLOC_MODE_REUSE)
2061 seq_printf(seq, ",alloc_mode=%s", "reuse");
2062
2063 if (test_opt(sbi, DISABLE_CHECKPOINT))
2064 seq_printf(seq, ",checkpoint=disable:%u",
2065 F2FS_OPTION(sbi).unusable_cap);
2066 if (test_opt(sbi, MERGE_CHECKPOINT))
2067 seq_puts(seq, ",checkpoint_merge");
2068 else
2069 seq_puts(seq, ",nocheckpoint_merge");
2070 if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_POSIX)
2071 seq_printf(seq, ",fsync_mode=%s", "posix");
2072 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT)
2073 seq_printf(seq, ",fsync_mode=%s", "strict");
2074 else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_NOBARRIER)
2075 seq_printf(seq, ",fsync_mode=%s", "nobarrier");
2076
2077 #ifdef CONFIG_F2FS_FS_COMPRESSION
2078 f2fs_show_compress_options(seq, sbi->sb);
2079 #endif
2080
2081 if (test_opt(sbi, ATGC))
2082 seq_puts(seq, ",atgc");
2083
2084 if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_NORMAL)
2085 seq_printf(seq, ",memory=%s", "normal");
2086 else if (F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW)
2087 seq_printf(seq, ",memory=%s", "low");
2088
2089 if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_READONLY)
2090 seq_printf(seq, ",errors=%s", "remount-ro");
2091 else if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_CONTINUE)
2092 seq_printf(seq, ",errors=%s", "continue");
2093 else if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_PANIC)
2094 seq_printf(seq, ",errors=%s", "panic");
2095
2096 return 0;
2097 }
2098
default_options(struct f2fs_sb_info * sbi,bool remount)2099 static void default_options(struct f2fs_sb_info *sbi, bool remount)
2100 {
2101 /* init some FS parameters */
2102 if (!remount) {
2103 set_opt(sbi, READ_EXTENT_CACHE);
2104 clear_opt(sbi, DISABLE_CHECKPOINT);
2105
2106 if (f2fs_hw_support_discard(sbi) || f2fs_hw_should_discard(sbi))
2107 set_opt(sbi, DISCARD);
2108
2109 if (f2fs_sb_has_blkzoned(sbi))
2110 F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_SECTION;
2111 else
2112 F2FS_OPTION(sbi).discard_unit = DISCARD_UNIT_BLOCK;
2113 }
2114
2115 if (f2fs_sb_has_readonly(sbi))
2116 F2FS_OPTION(sbi).active_logs = NR_CURSEG_RO_TYPE;
2117 else
2118 F2FS_OPTION(sbi).active_logs = NR_CURSEG_PERSIST_TYPE;
2119
2120 F2FS_OPTION(sbi).inline_xattr_size = DEFAULT_INLINE_XATTR_ADDRS;
2121 if (le32_to_cpu(F2FS_RAW_SUPER(sbi)->segment_count_main) <=
2122 SMALL_VOLUME_SEGMENTS)
2123 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_REUSE;
2124 else
2125 F2FS_OPTION(sbi).alloc_mode = ALLOC_MODE_DEFAULT;
2126 F2FS_OPTION(sbi).fsync_mode = FSYNC_MODE_POSIX;
2127 F2FS_OPTION(sbi).s_resuid = make_kuid(&init_user_ns, F2FS_DEF_RESUID);
2128 F2FS_OPTION(sbi).s_resgid = make_kgid(&init_user_ns, F2FS_DEF_RESGID);
2129 if (f2fs_sb_has_compression(sbi)) {
2130 F2FS_OPTION(sbi).compress_algorithm = COMPRESS_LZ4;
2131 F2FS_OPTION(sbi).compress_log_size = MIN_COMPRESS_LOG_SIZE;
2132 F2FS_OPTION(sbi).compress_ext_cnt = 0;
2133 F2FS_OPTION(sbi).compress_mode = COMPR_MODE_FS;
2134 }
2135 F2FS_OPTION(sbi).bggc_mode = BGGC_MODE_ON;
2136 F2FS_OPTION(sbi).memory_mode = MEMORY_MODE_NORMAL;
2137 F2FS_OPTION(sbi).errors = MOUNT_ERRORS_CONTINUE;
2138
2139 set_opt(sbi, INLINE_XATTR);
2140 set_opt(sbi, INLINE_DATA);
2141 set_opt(sbi, INLINE_DENTRY);
2142 set_opt(sbi, MERGE_CHECKPOINT);
2143 F2FS_OPTION(sbi).unusable_cap = 0;
2144 sbi->sb->s_flags |= SB_LAZYTIME;
2145 if (!f2fs_is_readonly(sbi))
2146 set_opt(sbi, FLUSH_MERGE);
2147 if (f2fs_sb_has_blkzoned(sbi))
2148 F2FS_OPTION(sbi).fs_mode = FS_MODE_LFS;
2149 else
2150 F2FS_OPTION(sbi).fs_mode = FS_MODE_ADAPTIVE;
2151
2152 #ifdef CONFIG_F2FS_FS_XATTR
2153 set_opt(sbi, XATTR_USER);
2154 #endif
2155 #ifdef CONFIG_F2FS_FS_POSIX_ACL
2156 set_opt(sbi, POSIX_ACL);
2157 #endif
2158
2159 f2fs_build_fault_attr(sbi, 0, 0);
2160 }
2161
2162 #ifdef CONFIG_QUOTA
2163 static int f2fs_enable_quotas(struct super_block *sb);
2164 #endif
2165
f2fs_disable_checkpoint(struct f2fs_sb_info * sbi)2166 static int f2fs_disable_checkpoint(struct f2fs_sb_info *sbi)
2167 {
2168 unsigned int s_flags = sbi->sb->s_flags;
2169 struct cp_control cpc;
2170 unsigned int gc_mode = sbi->gc_mode;
2171 int err = 0;
2172 int ret;
2173 block_t unusable;
2174
2175 if (s_flags & SB_RDONLY) {
2176 f2fs_err(sbi, "checkpoint=disable on readonly fs");
2177 return -EINVAL;
2178 }
2179 sbi->sb->s_flags |= SB_ACTIVE;
2180
2181 /* check if we need more GC first */
2182 unusable = f2fs_get_unusable_blocks(sbi);
2183 if (!f2fs_disable_cp_again(sbi, unusable))
2184 goto skip_gc;
2185
2186 f2fs_update_time(sbi, DISABLE_TIME);
2187
2188 sbi->gc_mode = GC_URGENT_HIGH;
2189
2190 while (!f2fs_time_over(sbi, DISABLE_TIME)) {
2191 struct f2fs_gc_control gc_control = {
2192 .victim_segno = NULL_SEGNO,
2193 .init_gc_type = FG_GC,
2194 .should_migrate_blocks = false,
2195 .err_gc_skipped = true,
2196 .nr_free_secs = 1 };
2197
2198 f2fs_down_write(&sbi->gc_lock);
2199 stat_inc_gc_call_count(sbi, FOREGROUND);
2200 err = f2fs_gc(sbi, &gc_control);
2201 if (err == -ENODATA) {
2202 err = 0;
2203 break;
2204 }
2205 if (err && err != -EAGAIN)
2206 break;
2207 }
2208
2209 ret = sync_filesystem(sbi->sb);
2210 if (ret || err) {
2211 err = ret ? ret : err;
2212 goto restore_flag;
2213 }
2214
2215 unusable = f2fs_get_unusable_blocks(sbi);
2216 if (f2fs_disable_cp_again(sbi, unusable)) {
2217 err = -EAGAIN;
2218 goto restore_flag;
2219 }
2220
2221 skip_gc:
2222 f2fs_down_write(&sbi->gc_lock);
2223 cpc.reason = CP_PAUSE;
2224 set_sbi_flag(sbi, SBI_CP_DISABLED);
2225 stat_inc_cp_call_count(sbi, TOTAL_CALL);
2226 err = f2fs_write_checkpoint(sbi, &cpc);
2227 if (err)
2228 goto out_unlock;
2229
2230 spin_lock(&sbi->stat_lock);
2231 sbi->unusable_block_count = unusable;
2232 spin_unlock(&sbi->stat_lock);
2233
2234 out_unlock:
2235 f2fs_up_write(&sbi->gc_lock);
2236 restore_flag:
2237 sbi->gc_mode = gc_mode;
2238 sbi->sb->s_flags = s_flags; /* Restore SB_RDONLY status */
2239 return err;
2240 }
2241
f2fs_enable_checkpoint(struct f2fs_sb_info * sbi)2242 static void f2fs_enable_checkpoint(struct f2fs_sb_info *sbi)
2243 {
2244 int retry = DEFAULT_RETRY_IO_COUNT;
2245
2246 /* we should flush all the data to keep data consistency */
2247 do {
2248 sync_inodes_sb(sbi->sb);
2249 f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2250 } while (get_pages(sbi, F2FS_DIRTY_DATA) && retry--);
2251
2252 if (unlikely(retry < 0))
2253 f2fs_warn(sbi, "checkpoint=enable has some unwritten data.");
2254
2255 f2fs_down_write(&sbi->gc_lock);
2256 f2fs_dirty_to_prefree(sbi);
2257
2258 clear_sbi_flag(sbi, SBI_CP_DISABLED);
2259 set_sbi_flag(sbi, SBI_IS_DIRTY);
2260 f2fs_up_write(&sbi->gc_lock);
2261
2262 f2fs_sync_fs(sbi->sb, 1);
2263
2264 /* Let's ensure there's no pending checkpoint anymore */
2265 f2fs_flush_ckpt_thread(sbi);
2266 }
2267
f2fs_remount(struct super_block * sb,int * flags,char * data)2268 static int f2fs_remount(struct super_block *sb, int *flags, char *data)
2269 {
2270 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2271 struct f2fs_mount_info org_mount_opt;
2272 unsigned long old_sb_flags;
2273 int err;
2274 bool need_restart_gc = false, need_stop_gc = false;
2275 bool need_restart_ckpt = false, need_stop_ckpt = false;
2276 bool need_restart_flush = false, need_stop_flush = false;
2277 bool need_restart_discard = false, need_stop_discard = false;
2278 bool no_read_extent_cache = !test_opt(sbi, READ_EXTENT_CACHE);
2279 bool no_age_extent_cache = !test_opt(sbi, AGE_EXTENT_CACHE);
2280 bool enable_checkpoint = !test_opt(sbi, DISABLE_CHECKPOINT);
2281 bool no_atgc = !test_opt(sbi, ATGC);
2282 bool no_discard = !test_opt(sbi, DISCARD);
2283 bool no_compress_cache = !test_opt(sbi, COMPRESS_CACHE);
2284 bool block_unit_discard = f2fs_block_unit_discard(sbi);
2285 #ifdef CONFIG_QUOTA
2286 int i, j;
2287 #endif
2288
2289 /*
2290 * Save the old mount options in case we
2291 * need to restore them.
2292 */
2293 org_mount_opt = sbi->mount_opt;
2294 old_sb_flags = sb->s_flags;
2295
2296 #ifdef CONFIG_QUOTA
2297 org_mount_opt.s_jquota_fmt = F2FS_OPTION(sbi).s_jquota_fmt;
2298 for (i = 0; i < MAXQUOTAS; i++) {
2299 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2300 org_mount_opt.s_qf_names[i] =
2301 kstrdup(F2FS_OPTION(sbi).s_qf_names[i],
2302 GFP_KERNEL);
2303 if (!org_mount_opt.s_qf_names[i]) {
2304 for (j = 0; j < i; j++)
2305 kfree(org_mount_opt.s_qf_names[j]);
2306 return -ENOMEM;
2307 }
2308 } else {
2309 org_mount_opt.s_qf_names[i] = NULL;
2310 }
2311 }
2312 #endif
2313
2314 /* recover superblocks we couldn't write due to previous RO mount */
2315 if (!(*flags & SB_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
2316 err = f2fs_commit_super(sbi, false);
2317 f2fs_info(sbi, "Try to recover all the superblocks, ret: %d",
2318 err);
2319 if (!err)
2320 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
2321 }
2322
2323 default_options(sbi, true);
2324
2325 /* parse mount options */
2326 err = parse_options(sb, data, true);
2327 if (err)
2328 goto restore_opts;
2329
2330 /* flush outstanding errors before changing fs state */
2331 flush_work(&sbi->s_error_work);
2332
2333 /*
2334 * Previous and new state of filesystem is RO,
2335 * so skip checking GC and FLUSH_MERGE conditions.
2336 */
2337 if (f2fs_readonly(sb) && (*flags & SB_RDONLY))
2338 goto skip;
2339
2340 if (f2fs_dev_is_readonly(sbi) && !(*flags & SB_RDONLY)) {
2341 err = -EROFS;
2342 goto restore_opts;
2343 }
2344
2345 #ifdef CONFIG_QUOTA
2346 if (!f2fs_readonly(sb) && (*flags & SB_RDONLY)) {
2347 err = dquot_suspend(sb, -1);
2348 if (err < 0)
2349 goto restore_opts;
2350 } else if (f2fs_readonly(sb) && !(*flags & SB_RDONLY)) {
2351 /* dquot_resume needs RW */
2352 sb->s_flags &= ~SB_RDONLY;
2353 if (sb_any_quota_suspended(sb)) {
2354 dquot_resume(sb, -1);
2355 } else if (f2fs_sb_has_quota_ino(sbi)) {
2356 err = f2fs_enable_quotas(sb);
2357 if (err)
2358 goto restore_opts;
2359 }
2360 }
2361 #endif
2362 if (f2fs_lfs_mode(sbi) && !IS_F2FS_IPU_DISABLE(sbi)) {
2363 err = -EINVAL;
2364 f2fs_warn(sbi, "LFS is not compatible with IPU");
2365 goto restore_opts;
2366 }
2367
2368 /* disallow enable atgc dynamically */
2369 if (no_atgc == !!test_opt(sbi, ATGC)) {
2370 err = -EINVAL;
2371 f2fs_warn(sbi, "switch atgc option is not allowed");
2372 goto restore_opts;
2373 }
2374
2375 /* disallow enable/disable extent_cache dynamically */
2376 if (no_read_extent_cache == !!test_opt(sbi, READ_EXTENT_CACHE)) {
2377 err = -EINVAL;
2378 f2fs_warn(sbi, "switch extent_cache option is not allowed");
2379 goto restore_opts;
2380 }
2381 /* disallow enable/disable age extent_cache dynamically */
2382 if (no_age_extent_cache == !!test_opt(sbi, AGE_EXTENT_CACHE)) {
2383 err = -EINVAL;
2384 f2fs_warn(sbi, "switch age_extent_cache option is not allowed");
2385 goto restore_opts;
2386 }
2387
2388 if (no_compress_cache == !!test_opt(sbi, COMPRESS_CACHE)) {
2389 err = -EINVAL;
2390 f2fs_warn(sbi, "switch compress_cache option is not allowed");
2391 goto restore_opts;
2392 }
2393
2394 if (block_unit_discard != f2fs_block_unit_discard(sbi)) {
2395 err = -EINVAL;
2396 f2fs_warn(sbi, "switch discard_unit option is not allowed");
2397 goto restore_opts;
2398 }
2399
2400 if ((*flags & SB_RDONLY) && test_opt(sbi, DISABLE_CHECKPOINT)) {
2401 err = -EINVAL;
2402 f2fs_warn(sbi, "disabling checkpoint not compatible with read-only");
2403 goto restore_opts;
2404 }
2405
2406 /*
2407 * We stop the GC thread if FS is mounted as RO
2408 * or if background_gc = off is passed in mount
2409 * option. Also sync the filesystem.
2410 */
2411 if ((*flags & SB_RDONLY) ||
2412 (F2FS_OPTION(sbi).bggc_mode == BGGC_MODE_OFF &&
2413 !test_opt(sbi, GC_MERGE))) {
2414 if (sbi->gc_thread) {
2415 f2fs_stop_gc_thread(sbi);
2416 need_restart_gc = true;
2417 }
2418 } else if (!sbi->gc_thread) {
2419 err = f2fs_start_gc_thread(sbi);
2420 if (err)
2421 goto restore_opts;
2422 need_stop_gc = true;
2423 }
2424
2425 if (*flags & SB_RDONLY) {
2426 sync_inodes_sb(sb);
2427
2428 set_sbi_flag(sbi, SBI_IS_DIRTY);
2429 set_sbi_flag(sbi, SBI_IS_CLOSE);
2430 f2fs_sync_fs(sb, 1);
2431 clear_sbi_flag(sbi, SBI_IS_CLOSE);
2432 }
2433
2434 if ((*flags & SB_RDONLY) || test_opt(sbi, DISABLE_CHECKPOINT) ||
2435 !test_opt(sbi, MERGE_CHECKPOINT)) {
2436 f2fs_stop_ckpt_thread(sbi);
2437 need_restart_ckpt = true;
2438 } else {
2439 /* Flush if the prevous checkpoint, if exists. */
2440 f2fs_flush_ckpt_thread(sbi);
2441
2442 err = f2fs_start_ckpt_thread(sbi);
2443 if (err) {
2444 f2fs_err(sbi,
2445 "Failed to start F2FS issue_checkpoint_thread (%d)",
2446 err);
2447 goto restore_gc;
2448 }
2449 need_stop_ckpt = true;
2450 }
2451
2452 /*
2453 * We stop issue flush thread if FS is mounted as RO
2454 * or if flush_merge is not passed in mount option.
2455 */
2456 if ((*flags & SB_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2457 clear_opt(sbi, FLUSH_MERGE);
2458 f2fs_destroy_flush_cmd_control(sbi, false);
2459 need_restart_flush = true;
2460 } else {
2461 err = f2fs_create_flush_cmd_control(sbi);
2462 if (err)
2463 goto restore_ckpt;
2464 need_stop_flush = true;
2465 }
2466
2467 if (no_discard == !!test_opt(sbi, DISCARD)) {
2468 if (test_opt(sbi, DISCARD)) {
2469 err = f2fs_start_discard_thread(sbi);
2470 if (err)
2471 goto restore_flush;
2472 need_stop_discard = true;
2473 } else {
2474 f2fs_stop_discard_thread(sbi);
2475 f2fs_issue_discard_timeout(sbi);
2476 need_restart_discard = true;
2477 }
2478 }
2479
2480 if (enable_checkpoint == !!test_opt(sbi, DISABLE_CHECKPOINT)) {
2481 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
2482 err = f2fs_disable_checkpoint(sbi);
2483 if (err)
2484 goto restore_discard;
2485 } else {
2486 f2fs_enable_checkpoint(sbi);
2487 }
2488 }
2489
2490 skip:
2491 #ifdef CONFIG_QUOTA
2492 /* Release old quota file names */
2493 for (i = 0; i < MAXQUOTAS; i++)
2494 kfree(org_mount_opt.s_qf_names[i]);
2495 #endif
2496 /* Update the POSIXACL Flag */
2497 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
2498 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
2499
2500 limit_reserve_root(sbi);
2501 adjust_unusable_cap_perc(sbi);
2502 *flags = (*flags & ~SB_LAZYTIME) | (sb->s_flags & SB_LAZYTIME);
2503 return 0;
2504 restore_discard:
2505 if (need_restart_discard) {
2506 if (f2fs_start_discard_thread(sbi))
2507 f2fs_warn(sbi, "discard has been stopped");
2508 } else if (need_stop_discard) {
2509 f2fs_stop_discard_thread(sbi);
2510 }
2511 restore_flush:
2512 if (need_restart_flush) {
2513 if (f2fs_create_flush_cmd_control(sbi))
2514 f2fs_warn(sbi, "background flush thread has stopped");
2515 } else if (need_stop_flush) {
2516 clear_opt(sbi, FLUSH_MERGE);
2517 f2fs_destroy_flush_cmd_control(sbi, false);
2518 }
2519 restore_ckpt:
2520 if (need_restart_ckpt) {
2521 if (f2fs_start_ckpt_thread(sbi))
2522 f2fs_warn(sbi, "background ckpt thread has stopped");
2523 } else if (need_stop_ckpt) {
2524 f2fs_stop_ckpt_thread(sbi);
2525 }
2526 restore_gc:
2527 if (need_restart_gc) {
2528 if (f2fs_start_gc_thread(sbi))
2529 f2fs_warn(sbi, "background gc thread has stopped");
2530 } else if (need_stop_gc) {
2531 f2fs_stop_gc_thread(sbi);
2532 }
2533 restore_opts:
2534 #ifdef CONFIG_QUOTA
2535 F2FS_OPTION(sbi).s_jquota_fmt = org_mount_opt.s_jquota_fmt;
2536 for (i = 0; i < MAXQUOTAS; i++) {
2537 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
2538 F2FS_OPTION(sbi).s_qf_names[i] = org_mount_opt.s_qf_names[i];
2539 }
2540 #endif
2541 sbi->mount_opt = org_mount_opt;
2542 sb->s_flags = old_sb_flags;
2543 return err;
2544 }
2545
2546 #ifdef CONFIG_QUOTA
f2fs_need_recovery(struct f2fs_sb_info * sbi)2547 static bool f2fs_need_recovery(struct f2fs_sb_info *sbi)
2548 {
2549 /* need to recovery orphan */
2550 if (is_set_ckpt_flags(sbi, CP_ORPHAN_PRESENT_FLAG))
2551 return true;
2552 /* need to recovery data */
2553 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
2554 return false;
2555 if (test_opt(sbi, NORECOVERY))
2556 return false;
2557 return !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG);
2558 }
2559
f2fs_recover_quota_begin(struct f2fs_sb_info * sbi)2560 static bool f2fs_recover_quota_begin(struct f2fs_sb_info *sbi)
2561 {
2562 bool readonly = f2fs_readonly(sbi->sb);
2563
2564 if (!f2fs_need_recovery(sbi))
2565 return false;
2566
2567 /* it doesn't need to check f2fs_sb_has_readonly() */
2568 if (f2fs_hw_is_readonly(sbi))
2569 return false;
2570
2571 if (readonly) {
2572 sbi->sb->s_flags &= ~SB_RDONLY;
2573 set_sbi_flag(sbi, SBI_IS_WRITABLE);
2574 }
2575
2576 /*
2577 * Turn on quotas which were not enabled for read-only mounts if
2578 * filesystem has quota feature, so that they are updated correctly.
2579 */
2580 return f2fs_enable_quota_files(sbi, readonly);
2581 }
2582
f2fs_recover_quota_end(struct f2fs_sb_info * sbi,bool quota_enabled)2583 static void f2fs_recover_quota_end(struct f2fs_sb_info *sbi,
2584 bool quota_enabled)
2585 {
2586 if (quota_enabled)
2587 f2fs_quota_off_umount(sbi->sb);
2588
2589 if (is_sbi_flag_set(sbi, SBI_IS_WRITABLE)) {
2590 clear_sbi_flag(sbi, SBI_IS_WRITABLE);
2591 sbi->sb->s_flags |= SB_RDONLY;
2592 }
2593 }
2594
2595 /* Read data from quotafile */
f2fs_quota_read(struct super_block * sb,int type,char * data,size_t len,loff_t off)2596 static ssize_t f2fs_quota_read(struct super_block *sb, int type, char *data,
2597 size_t len, loff_t off)
2598 {
2599 struct inode *inode = sb_dqopt(sb)->files[type];
2600 struct address_space *mapping = inode->i_mapping;
2601 block_t blkidx = F2FS_BYTES_TO_BLK(off);
2602 int offset = off & (sb->s_blocksize - 1);
2603 int tocopy;
2604 size_t toread;
2605 loff_t i_size = i_size_read(inode);
2606 struct page *page;
2607
2608 if (off > i_size)
2609 return 0;
2610
2611 if (off + len > i_size)
2612 len = i_size - off;
2613 toread = len;
2614 while (toread > 0) {
2615 tocopy = min_t(unsigned long, sb->s_blocksize - offset, toread);
2616 repeat:
2617 page = read_cache_page_gfp(mapping, blkidx, GFP_NOFS);
2618 if (IS_ERR(page)) {
2619 if (PTR_ERR(page) == -ENOMEM) {
2620 memalloc_retry_wait(GFP_NOFS);
2621 goto repeat;
2622 }
2623 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2624 return PTR_ERR(page);
2625 }
2626
2627 lock_page(page);
2628
2629 if (unlikely(page->mapping != mapping)) {
2630 f2fs_put_page(page, 1);
2631 goto repeat;
2632 }
2633 if (unlikely(!PageUptodate(page))) {
2634 f2fs_put_page(page, 1);
2635 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2636 return -EIO;
2637 }
2638
2639 memcpy_from_page(data, page, offset, tocopy);
2640 f2fs_put_page(page, 1);
2641
2642 offset = 0;
2643 toread -= tocopy;
2644 data += tocopy;
2645 blkidx++;
2646 }
2647 return len;
2648 }
2649
2650 /* Write to quotafile */
f2fs_quota_write(struct super_block * sb,int type,const char * data,size_t len,loff_t off)2651 static ssize_t f2fs_quota_write(struct super_block *sb, int type,
2652 const char *data, size_t len, loff_t off)
2653 {
2654 struct inode *inode = sb_dqopt(sb)->files[type];
2655 struct address_space *mapping = inode->i_mapping;
2656 const struct address_space_operations *a_ops = mapping->a_ops;
2657 int offset = off & (sb->s_blocksize - 1);
2658 size_t towrite = len;
2659 struct page *page;
2660 void *fsdata = NULL;
2661 int err = 0;
2662 int tocopy;
2663
2664 while (towrite > 0) {
2665 tocopy = min_t(unsigned long, sb->s_blocksize - offset,
2666 towrite);
2667 retry:
2668 err = a_ops->write_begin(NULL, mapping, off, tocopy,
2669 &page, &fsdata);
2670 if (unlikely(err)) {
2671 if (err == -ENOMEM) {
2672 f2fs_io_schedule_timeout(DEFAULT_IO_TIMEOUT);
2673 goto retry;
2674 }
2675 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
2676 break;
2677 }
2678
2679 memcpy_to_page(page, offset, data, tocopy);
2680
2681 a_ops->write_end(NULL, mapping, off, tocopy, tocopy,
2682 page, fsdata);
2683 offset = 0;
2684 towrite -= tocopy;
2685 off += tocopy;
2686 data += tocopy;
2687 cond_resched();
2688 }
2689
2690 if (len == towrite)
2691 return err;
2692 inode->i_mtime = inode_set_ctime_current(inode);
2693 f2fs_mark_inode_dirty_sync(inode, false);
2694 return len - towrite;
2695 }
2696
f2fs_dquot_initialize(struct inode * inode)2697 int f2fs_dquot_initialize(struct inode *inode)
2698 {
2699 if (time_to_inject(F2FS_I_SB(inode), FAULT_DQUOT_INIT))
2700 return -ESRCH;
2701
2702 return dquot_initialize(inode);
2703 }
2704
f2fs_get_dquots(struct inode * inode)2705 static struct dquot __rcu **f2fs_get_dquots(struct inode *inode)
2706 {
2707 return F2FS_I(inode)->i_dquot;
2708 }
2709
f2fs_get_reserved_space(struct inode * inode)2710 static qsize_t *f2fs_get_reserved_space(struct inode *inode)
2711 {
2712 return &F2FS_I(inode)->i_reserved_quota;
2713 }
2714
f2fs_quota_on_mount(struct f2fs_sb_info * sbi,int type)2715 static int f2fs_quota_on_mount(struct f2fs_sb_info *sbi, int type)
2716 {
2717 if (is_set_ckpt_flags(sbi, CP_QUOTA_NEED_FSCK_FLAG)) {
2718 f2fs_err(sbi, "quota sysfile may be corrupted, skip loading it");
2719 return 0;
2720 }
2721
2722 return dquot_quota_on_mount(sbi->sb, F2FS_OPTION(sbi).s_qf_names[type],
2723 F2FS_OPTION(sbi).s_jquota_fmt, type);
2724 }
2725
f2fs_enable_quota_files(struct f2fs_sb_info * sbi,bool rdonly)2726 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly)
2727 {
2728 int enabled = 0;
2729 int i, err;
2730
2731 if (f2fs_sb_has_quota_ino(sbi) && rdonly) {
2732 err = f2fs_enable_quotas(sbi->sb);
2733 if (err) {
2734 f2fs_err(sbi, "Cannot turn on quota_ino: %d", err);
2735 return 0;
2736 }
2737 return 1;
2738 }
2739
2740 for (i = 0; i < MAXQUOTAS; i++) {
2741 if (F2FS_OPTION(sbi).s_qf_names[i]) {
2742 err = f2fs_quota_on_mount(sbi, i);
2743 if (!err) {
2744 enabled = 1;
2745 continue;
2746 }
2747 f2fs_err(sbi, "Cannot turn on quotas: %d on %d",
2748 err, i);
2749 }
2750 }
2751 return enabled;
2752 }
2753
f2fs_quota_enable(struct super_block * sb,int type,int format_id,unsigned int flags)2754 static int f2fs_quota_enable(struct super_block *sb, int type, int format_id,
2755 unsigned int flags)
2756 {
2757 struct inode *qf_inode;
2758 unsigned long qf_inum;
2759 unsigned long qf_flag = F2FS_QUOTA_DEFAULT_FL;
2760 int err;
2761
2762 BUG_ON(!f2fs_sb_has_quota_ino(F2FS_SB(sb)));
2763
2764 qf_inum = f2fs_qf_ino(sb, type);
2765 if (!qf_inum)
2766 return -EPERM;
2767
2768 qf_inode = f2fs_iget(sb, qf_inum);
2769 if (IS_ERR(qf_inode)) {
2770 f2fs_err(F2FS_SB(sb), "Bad quota inode %u:%lu", type, qf_inum);
2771 return PTR_ERR(qf_inode);
2772 }
2773
2774 /* Don't account quota for quota files to avoid recursion */
2775 inode_lock(qf_inode);
2776 qf_inode->i_flags |= S_NOQUOTA;
2777
2778 if ((F2FS_I(qf_inode)->i_flags & qf_flag) != qf_flag) {
2779 F2FS_I(qf_inode)->i_flags |= qf_flag;
2780 f2fs_set_inode_flags(qf_inode);
2781 }
2782 inode_unlock(qf_inode);
2783
2784 err = dquot_load_quota_inode(qf_inode, type, format_id, flags);
2785 iput(qf_inode);
2786 return err;
2787 }
2788
f2fs_enable_quotas(struct super_block * sb)2789 static int f2fs_enable_quotas(struct super_block *sb)
2790 {
2791 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2792 int type, err = 0;
2793 unsigned long qf_inum;
2794 bool quota_mopt[MAXQUOTAS] = {
2795 test_opt(sbi, USRQUOTA),
2796 test_opt(sbi, GRPQUOTA),
2797 test_opt(sbi, PRJQUOTA),
2798 };
2799
2800 if (is_set_ckpt_flags(F2FS_SB(sb), CP_QUOTA_NEED_FSCK_FLAG)) {
2801 f2fs_err(sbi, "quota file may be corrupted, skip loading it");
2802 return 0;
2803 }
2804
2805 sb_dqopt(sb)->flags |= DQUOT_QUOTA_SYS_FILE;
2806
2807 for (type = 0; type < MAXQUOTAS; type++) {
2808 qf_inum = f2fs_qf_ino(sb, type);
2809 if (qf_inum) {
2810 err = f2fs_quota_enable(sb, type, QFMT_VFS_V1,
2811 DQUOT_USAGE_ENABLED |
2812 (quota_mopt[type] ? DQUOT_LIMITS_ENABLED : 0));
2813 if (err) {
2814 f2fs_err(sbi, "Failed to enable quota tracking (type=%d, err=%d). Please run fsck to fix.",
2815 type, err);
2816 for (type--; type >= 0; type--)
2817 dquot_quota_off(sb, type);
2818 set_sbi_flag(F2FS_SB(sb),
2819 SBI_QUOTA_NEED_REPAIR);
2820 return err;
2821 }
2822 }
2823 }
2824 return 0;
2825 }
2826
f2fs_quota_sync_file(struct f2fs_sb_info * sbi,int type)2827 static int f2fs_quota_sync_file(struct f2fs_sb_info *sbi, int type)
2828 {
2829 struct quota_info *dqopt = sb_dqopt(sbi->sb);
2830 struct address_space *mapping = dqopt->files[type]->i_mapping;
2831 int ret = 0;
2832
2833 ret = dquot_writeback_dquots(sbi->sb, type);
2834 if (ret)
2835 goto out;
2836
2837 ret = filemap_fdatawrite(mapping);
2838 if (ret)
2839 goto out;
2840
2841 /* if we are using journalled quota */
2842 if (is_journalled_quota(sbi))
2843 goto out;
2844
2845 ret = filemap_fdatawait(mapping);
2846
2847 truncate_inode_pages(&dqopt->files[type]->i_data, 0);
2848 out:
2849 if (ret)
2850 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2851 return ret;
2852 }
2853
f2fs_quota_sync(struct super_block * sb,int type)2854 int f2fs_quota_sync(struct super_block *sb, int type)
2855 {
2856 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2857 struct quota_info *dqopt = sb_dqopt(sb);
2858 int cnt;
2859 int ret = 0;
2860
2861 /*
2862 * Now when everything is written we can discard the pagecache so
2863 * that userspace sees the changes.
2864 */
2865 for (cnt = 0; cnt < MAXQUOTAS; cnt++) {
2866
2867 if (type != -1 && cnt != type)
2868 continue;
2869
2870 if (!sb_has_quota_active(sb, cnt))
2871 continue;
2872
2873 if (!f2fs_sb_has_quota_ino(sbi))
2874 inode_lock(dqopt->files[cnt]);
2875
2876 /*
2877 * do_quotactl
2878 * f2fs_quota_sync
2879 * f2fs_down_read(quota_sem)
2880 * dquot_writeback_dquots()
2881 * f2fs_dquot_commit
2882 * block_operation
2883 * f2fs_down_read(quota_sem)
2884 */
2885 f2fs_lock_op(sbi);
2886 f2fs_down_read(&sbi->quota_sem);
2887
2888 ret = f2fs_quota_sync_file(sbi, cnt);
2889
2890 f2fs_up_read(&sbi->quota_sem);
2891 f2fs_unlock_op(sbi);
2892
2893 if (!f2fs_sb_has_quota_ino(sbi))
2894 inode_unlock(dqopt->files[cnt]);
2895
2896 if (ret)
2897 break;
2898 }
2899 return ret;
2900 }
2901
f2fs_quota_on(struct super_block * sb,int type,int format_id,const struct path * path)2902 static int f2fs_quota_on(struct super_block *sb, int type, int format_id,
2903 const struct path *path)
2904 {
2905 struct inode *inode;
2906 int err;
2907
2908 /* if quota sysfile exists, deny enabling quota with specific file */
2909 if (f2fs_sb_has_quota_ino(F2FS_SB(sb))) {
2910 f2fs_err(F2FS_SB(sb), "quota sysfile already exists");
2911 return -EBUSY;
2912 }
2913
2914 if (path->dentry->d_sb != sb)
2915 return -EXDEV;
2916
2917 err = f2fs_quota_sync(sb, type);
2918 if (err)
2919 return err;
2920
2921 inode = d_inode(path->dentry);
2922
2923 err = filemap_fdatawrite(inode->i_mapping);
2924 if (err)
2925 return err;
2926
2927 err = filemap_fdatawait(inode->i_mapping);
2928 if (err)
2929 return err;
2930
2931 err = dquot_quota_on(sb, type, format_id, path);
2932 if (err)
2933 return err;
2934
2935 inode_lock(inode);
2936 F2FS_I(inode)->i_flags |= F2FS_QUOTA_DEFAULT_FL;
2937 f2fs_set_inode_flags(inode);
2938 inode_unlock(inode);
2939 f2fs_mark_inode_dirty_sync(inode, false);
2940
2941 return 0;
2942 }
2943
__f2fs_quota_off(struct super_block * sb,int type)2944 static int __f2fs_quota_off(struct super_block *sb, int type)
2945 {
2946 struct inode *inode = sb_dqopt(sb)->files[type];
2947 int err;
2948
2949 if (!inode || !igrab(inode))
2950 return dquot_quota_off(sb, type);
2951
2952 err = f2fs_quota_sync(sb, type);
2953 if (err)
2954 goto out_put;
2955
2956 err = dquot_quota_off(sb, type);
2957 if (err || f2fs_sb_has_quota_ino(F2FS_SB(sb)))
2958 goto out_put;
2959
2960 inode_lock(inode);
2961 F2FS_I(inode)->i_flags &= ~F2FS_QUOTA_DEFAULT_FL;
2962 f2fs_set_inode_flags(inode);
2963 inode_unlock(inode);
2964 f2fs_mark_inode_dirty_sync(inode, false);
2965 out_put:
2966 iput(inode);
2967 return err;
2968 }
2969
f2fs_quota_off(struct super_block * sb,int type)2970 static int f2fs_quota_off(struct super_block *sb, int type)
2971 {
2972 struct f2fs_sb_info *sbi = F2FS_SB(sb);
2973 int err;
2974
2975 err = __f2fs_quota_off(sb, type);
2976
2977 /*
2978 * quotactl can shutdown journalled quota, result in inconsistence
2979 * between quota record and fs data by following updates, tag the
2980 * flag to let fsck be aware of it.
2981 */
2982 if (is_journalled_quota(sbi))
2983 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
2984 return err;
2985 }
2986
f2fs_quota_off_umount(struct super_block * sb)2987 void f2fs_quota_off_umount(struct super_block *sb)
2988 {
2989 int type;
2990 int err;
2991
2992 for (type = 0; type < MAXQUOTAS; type++) {
2993 err = __f2fs_quota_off(sb, type);
2994 if (err) {
2995 int ret = dquot_quota_off(sb, type);
2996
2997 f2fs_err(F2FS_SB(sb), "Fail to turn off disk quota (type: %d, err: %d, ret:%d), Please run fsck to fix it.",
2998 type, err, ret);
2999 set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR);
3000 }
3001 }
3002 /*
3003 * In case of checkpoint=disable, we must flush quota blocks.
3004 * This can cause NULL exception for node_inode in end_io, since
3005 * put_super already dropped it.
3006 */
3007 sync_filesystem(sb);
3008 }
3009
f2fs_truncate_quota_inode_pages(struct super_block * sb)3010 static void f2fs_truncate_quota_inode_pages(struct super_block *sb)
3011 {
3012 struct quota_info *dqopt = sb_dqopt(sb);
3013 int type;
3014
3015 for (type = 0; type < MAXQUOTAS; type++) {
3016 if (!dqopt->files[type])
3017 continue;
3018 f2fs_inode_synced(dqopt->files[type]);
3019 }
3020 }
3021
f2fs_dquot_commit(struct dquot * dquot)3022 static int f2fs_dquot_commit(struct dquot *dquot)
3023 {
3024 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3025 int ret;
3026
3027 f2fs_down_read_nested(&sbi->quota_sem, SINGLE_DEPTH_NESTING);
3028 ret = dquot_commit(dquot);
3029 if (ret < 0)
3030 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3031 f2fs_up_read(&sbi->quota_sem);
3032 return ret;
3033 }
3034
f2fs_dquot_acquire(struct dquot * dquot)3035 static int f2fs_dquot_acquire(struct dquot *dquot)
3036 {
3037 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3038 int ret;
3039
3040 f2fs_down_read(&sbi->quota_sem);
3041 ret = dquot_acquire(dquot);
3042 if (ret < 0)
3043 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3044 f2fs_up_read(&sbi->quota_sem);
3045 return ret;
3046 }
3047
f2fs_dquot_release(struct dquot * dquot)3048 static int f2fs_dquot_release(struct dquot *dquot)
3049 {
3050 struct f2fs_sb_info *sbi = F2FS_SB(dquot->dq_sb);
3051 int ret = dquot_release(dquot);
3052
3053 if (ret < 0)
3054 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3055 return ret;
3056 }
3057
f2fs_dquot_mark_dquot_dirty(struct dquot * dquot)3058 static int f2fs_dquot_mark_dquot_dirty(struct dquot *dquot)
3059 {
3060 struct super_block *sb = dquot->dq_sb;
3061 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3062 int ret = dquot_mark_dquot_dirty(dquot);
3063
3064 /* if we are using journalled quota */
3065 if (is_journalled_quota(sbi))
3066 set_sbi_flag(sbi, SBI_QUOTA_NEED_FLUSH);
3067
3068 return ret;
3069 }
3070
f2fs_dquot_commit_info(struct super_block * sb,int type)3071 static int f2fs_dquot_commit_info(struct super_block *sb, int type)
3072 {
3073 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3074 int ret = dquot_commit_info(sb, type);
3075
3076 if (ret < 0)
3077 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
3078 return ret;
3079 }
3080
f2fs_get_projid(struct inode * inode,kprojid_t * projid)3081 static int f2fs_get_projid(struct inode *inode, kprojid_t *projid)
3082 {
3083 *projid = F2FS_I(inode)->i_projid;
3084 return 0;
3085 }
3086
3087 static const struct dquot_operations f2fs_quota_operations = {
3088 .get_reserved_space = f2fs_get_reserved_space,
3089 .write_dquot = f2fs_dquot_commit,
3090 .acquire_dquot = f2fs_dquot_acquire,
3091 .release_dquot = f2fs_dquot_release,
3092 .mark_dirty = f2fs_dquot_mark_dquot_dirty,
3093 .write_info = f2fs_dquot_commit_info,
3094 .alloc_dquot = dquot_alloc,
3095 .destroy_dquot = dquot_destroy,
3096 .get_projid = f2fs_get_projid,
3097 .get_next_id = dquot_get_next_id,
3098 };
3099
3100 static const struct quotactl_ops f2fs_quotactl_ops = {
3101 .quota_on = f2fs_quota_on,
3102 .quota_off = f2fs_quota_off,
3103 .quota_sync = f2fs_quota_sync,
3104 .get_state = dquot_get_state,
3105 .set_info = dquot_set_dqinfo,
3106 .get_dqblk = dquot_get_dqblk,
3107 .set_dqblk = dquot_set_dqblk,
3108 .get_nextdqblk = dquot_get_next_dqblk,
3109 };
3110 #else
f2fs_dquot_initialize(struct inode * inode)3111 int f2fs_dquot_initialize(struct inode *inode)
3112 {
3113 return 0;
3114 }
3115
f2fs_quota_sync(struct super_block * sb,int type)3116 int f2fs_quota_sync(struct super_block *sb, int type)
3117 {
3118 return 0;
3119 }
3120
f2fs_quota_off_umount(struct super_block * sb)3121 void f2fs_quota_off_umount(struct super_block *sb)
3122 {
3123 }
3124 #endif
3125
3126 static const struct super_operations f2fs_sops = {
3127 .alloc_inode = f2fs_alloc_inode,
3128 .free_inode = f2fs_free_inode,
3129 .drop_inode = f2fs_drop_inode,
3130 .write_inode = f2fs_write_inode,
3131 .dirty_inode = f2fs_dirty_inode,
3132 .show_options = f2fs_show_options,
3133 #ifdef CONFIG_QUOTA
3134 .quota_read = f2fs_quota_read,
3135 .quota_write = f2fs_quota_write,
3136 .get_dquots = f2fs_get_dquots,
3137 #endif
3138 .evict_inode = f2fs_evict_inode,
3139 .put_super = f2fs_put_super,
3140 .sync_fs = f2fs_sync_fs,
3141 .freeze_fs = f2fs_freeze,
3142 .unfreeze_fs = f2fs_unfreeze,
3143 .statfs = f2fs_statfs,
3144 .remount_fs = f2fs_remount,
3145 };
3146
3147 #ifdef CONFIG_FS_ENCRYPTION
f2fs_get_context(struct inode * inode,void * ctx,size_t len)3148 static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
3149 {
3150 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3151 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3152 ctx, len, NULL);
3153 }
3154
f2fs_set_context(struct inode * inode,const void * ctx,size_t len,void * fs_data)3155 static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
3156 void *fs_data)
3157 {
3158 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
3159
3160 /*
3161 * Encrypting the root directory is not allowed because fsck
3162 * expects lost+found directory to exist and remain unencrypted
3163 * if LOST_FOUND feature is enabled.
3164 *
3165 */
3166 if (f2fs_sb_has_lost_found(sbi) &&
3167 inode->i_ino == F2FS_ROOT_INO(sbi))
3168 return -EPERM;
3169
3170 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
3171 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
3172 ctx, len, fs_data, XATTR_CREATE);
3173 }
3174
f2fs_get_dummy_policy(struct super_block * sb)3175 static const union fscrypt_policy *f2fs_get_dummy_policy(struct super_block *sb)
3176 {
3177 return F2FS_OPTION(F2FS_SB(sb)).dummy_enc_policy.policy;
3178 }
3179
f2fs_has_stable_inodes(struct super_block * sb)3180 static bool f2fs_has_stable_inodes(struct super_block *sb)
3181 {
3182 return true;
3183 }
3184
f2fs_get_ino_and_lblk_bits(struct super_block * sb,int * ino_bits_ret,int * lblk_bits_ret)3185 static void f2fs_get_ino_and_lblk_bits(struct super_block *sb,
3186 int *ino_bits_ret, int *lblk_bits_ret)
3187 {
3188 *ino_bits_ret = 8 * sizeof(nid_t);
3189 *lblk_bits_ret = 8 * sizeof(block_t);
3190 }
3191
f2fs_get_devices(struct super_block * sb,unsigned int * num_devs)3192 static struct block_device **f2fs_get_devices(struct super_block *sb,
3193 unsigned int *num_devs)
3194 {
3195 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3196 struct block_device **devs;
3197 int i;
3198
3199 if (!f2fs_is_multi_device(sbi))
3200 return NULL;
3201
3202 devs = kmalloc_array(sbi->s_ndevs, sizeof(*devs), GFP_KERNEL);
3203 if (!devs)
3204 return ERR_PTR(-ENOMEM);
3205
3206 for (i = 0; i < sbi->s_ndevs; i++)
3207 devs[i] = FDEV(i).bdev;
3208 *num_devs = sbi->s_ndevs;
3209 return devs;
3210 }
3211
3212 static const struct fscrypt_operations f2fs_cryptops = {
3213 .key_prefix = "f2fs:",
3214 .get_context = f2fs_get_context,
3215 .set_context = f2fs_set_context,
3216 .get_dummy_policy = f2fs_get_dummy_policy,
3217 .empty_dir = f2fs_empty_dir,
3218 .has_stable_inodes = f2fs_has_stable_inodes,
3219 .get_ino_and_lblk_bits = f2fs_get_ino_and_lblk_bits,
3220 .get_devices = f2fs_get_devices,
3221 };
3222 #endif
3223
f2fs_nfs_get_inode(struct super_block * sb,u64 ino,u32 generation)3224 static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
3225 u64 ino, u32 generation)
3226 {
3227 struct f2fs_sb_info *sbi = F2FS_SB(sb);
3228 struct inode *inode;
3229
3230 if (f2fs_check_nid_range(sbi, ino))
3231 return ERR_PTR(-ESTALE);
3232
3233 /*
3234 * f2fs_iget isn't quite right if the inode is currently unallocated!
3235 * However f2fs_iget currently does appropriate checks to handle stale
3236 * inodes so everything is OK.
3237 */
3238 inode = f2fs_iget(sb, ino);
3239 if (IS_ERR(inode))
3240 return ERR_CAST(inode);
3241 if (unlikely(generation && inode->i_generation != generation)) {
3242 /* we didn't find the right inode.. */
3243 iput(inode);
3244 return ERR_PTR(-ESTALE);
3245 }
3246 return inode;
3247 }
3248
f2fs_fh_to_dentry(struct super_block * sb,struct fid * fid,int fh_len,int fh_type)3249 static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
3250 int fh_len, int fh_type)
3251 {
3252 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
3253 f2fs_nfs_get_inode);
3254 }
3255
f2fs_fh_to_parent(struct super_block * sb,struct fid * fid,int fh_len,int fh_type)3256 static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
3257 int fh_len, int fh_type)
3258 {
3259 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
3260 f2fs_nfs_get_inode);
3261 }
3262
3263 static const struct export_operations f2fs_export_ops = {
3264 .fh_to_dentry = f2fs_fh_to_dentry,
3265 .fh_to_parent = f2fs_fh_to_parent,
3266 .get_parent = f2fs_get_parent,
3267 };
3268
max_file_blocks(struct inode * inode)3269 loff_t max_file_blocks(struct inode *inode)
3270 {
3271 loff_t result = 0;
3272 loff_t leaf_count;
3273
3274 /*
3275 * note: previously, result is equal to (DEF_ADDRS_PER_INODE -
3276 * DEFAULT_INLINE_XATTR_ADDRS), but now f2fs try to reserve more
3277 * space in inode.i_addr, it will be more safe to reassign
3278 * result as zero.
3279 */
3280
3281 if (inode && f2fs_compressed_file(inode))
3282 leaf_count = ADDRS_PER_BLOCK(inode);
3283 else
3284 leaf_count = DEF_ADDRS_PER_BLOCK;
3285
3286 /* two direct node blocks */
3287 result += (leaf_count * 2);
3288
3289 /* two indirect node blocks */
3290 leaf_count *= NIDS_PER_BLOCK;
3291 result += (leaf_count * 2);
3292
3293 /* one double indirect node block */
3294 leaf_count *= NIDS_PER_BLOCK;
3295 result += leaf_count;
3296
3297 return result;
3298 }
3299
__f2fs_commit_super(struct buffer_head * bh,struct f2fs_super_block * super)3300 static int __f2fs_commit_super(struct buffer_head *bh,
3301 struct f2fs_super_block *super)
3302 {
3303 lock_buffer(bh);
3304 if (super)
3305 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
3306 set_buffer_dirty(bh);
3307 unlock_buffer(bh);
3308
3309 /* it's rare case, we can do fua all the time */
3310 return __sync_dirty_buffer(bh, REQ_SYNC | REQ_PREFLUSH | REQ_FUA);
3311 }
3312
sanity_check_area_boundary(struct f2fs_sb_info * sbi,struct buffer_head * bh)3313 static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
3314 struct buffer_head *bh)
3315 {
3316 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3317 (bh->b_data + F2FS_SUPER_OFFSET);
3318 struct super_block *sb = sbi->sb;
3319 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
3320 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
3321 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
3322 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
3323 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
3324 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
3325 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
3326 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
3327 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
3328 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
3329 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3330 u32 segment_count = le32_to_cpu(raw_super->segment_count);
3331 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3332 u64 main_end_blkaddr = main_blkaddr +
3333 (segment_count_main << log_blocks_per_seg);
3334 u64 seg_end_blkaddr = segment0_blkaddr +
3335 (segment_count << log_blocks_per_seg);
3336
3337 if (segment0_blkaddr != cp_blkaddr) {
3338 f2fs_info(sbi, "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
3339 segment0_blkaddr, cp_blkaddr);
3340 return true;
3341 }
3342
3343 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
3344 sit_blkaddr) {
3345 f2fs_info(sbi, "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
3346 cp_blkaddr, sit_blkaddr,
3347 segment_count_ckpt << log_blocks_per_seg);
3348 return true;
3349 }
3350
3351 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
3352 nat_blkaddr) {
3353 f2fs_info(sbi, "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
3354 sit_blkaddr, nat_blkaddr,
3355 segment_count_sit << log_blocks_per_seg);
3356 return true;
3357 }
3358
3359 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
3360 ssa_blkaddr) {
3361 f2fs_info(sbi, "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
3362 nat_blkaddr, ssa_blkaddr,
3363 segment_count_nat << log_blocks_per_seg);
3364 return true;
3365 }
3366
3367 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
3368 main_blkaddr) {
3369 f2fs_info(sbi, "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
3370 ssa_blkaddr, main_blkaddr,
3371 segment_count_ssa << log_blocks_per_seg);
3372 return true;
3373 }
3374
3375 if (main_end_blkaddr > seg_end_blkaddr) {
3376 f2fs_info(sbi, "Wrong MAIN_AREA boundary, start(%u) end(%llu) block(%u)",
3377 main_blkaddr, seg_end_blkaddr,
3378 segment_count_main << log_blocks_per_seg);
3379 return true;
3380 } else if (main_end_blkaddr < seg_end_blkaddr) {
3381 int err = 0;
3382 char *res;
3383
3384 /* fix in-memory information all the time */
3385 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
3386 segment0_blkaddr) >> log_blocks_per_seg);
3387
3388 if (f2fs_readonly(sb) || f2fs_hw_is_readonly(sbi)) {
3389 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3390 res = "internally";
3391 } else {
3392 err = __f2fs_commit_super(bh, NULL);
3393 res = err ? "failed" : "done";
3394 }
3395 f2fs_info(sbi, "Fix alignment : %s, start(%u) end(%llu) block(%u)",
3396 res, main_blkaddr, seg_end_blkaddr,
3397 segment_count_main << log_blocks_per_seg);
3398 if (err)
3399 return true;
3400 }
3401 return false;
3402 }
3403
sanity_check_raw_super(struct f2fs_sb_info * sbi,struct buffer_head * bh)3404 static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
3405 struct buffer_head *bh)
3406 {
3407 block_t segment_count, segs_per_sec, secs_per_zone, segment_count_main;
3408 block_t total_sections, blocks_per_seg;
3409 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
3410 (bh->b_data + F2FS_SUPER_OFFSET);
3411 size_t crc_offset = 0;
3412 __u32 crc = 0;
3413
3414 if (le32_to_cpu(raw_super->magic) != F2FS_SUPER_MAGIC) {
3415 f2fs_info(sbi, "Magic Mismatch, valid(0x%x) - read(0x%x)",
3416 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
3417 return -EINVAL;
3418 }
3419
3420 /* Check checksum_offset and crc in superblock */
3421 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_SB_CHKSUM)) {
3422 crc_offset = le32_to_cpu(raw_super->checksum_offset);
3423 if (crc_offset !=
3424 offsetof(struct f2fs_super_block, crc)) {
3425 f2fs_info(sbi, "Invalid SB checksum offset: %zu",
3426 crc_offset);
3427 return -EFSCORRUPTED;
3428 }
3429 crc = le32_to_cpu(raw_super->crc);
3430 if (!f2fs_crc_valid(sbi, crc, raw_super, crc_offset)) {
3431 f2fs_info(sbi, "Invalid SB checksum value: %u", crc);
3432 return -EFSCORRUPTED;
3433 }
3434 }
3435
3436 /* Currently, support only 4KB block size */
3437 if (le32_to_cpu(raw_super->log_blocksize) != F2FS_BLKSIZE_BITS) {
3438 f2fs_info(sbi, "Invalid log_blocksize (%u), supports only %u",
3439 le32_to_cpu(raw_super->log_blocksize),
3440 F2FS_BLKSIZE_BITS);
3441 return -EFSCORRUPTED;
3442 }
3443
3444 /* check log blocks per segment */
3445 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
3446 f2fs_info(sbi, "Invalid log blocks per segment (%u)",
3447 le32_to_cpu(raw_super->log_blocks_per_seg));
3448 return -EFSCORRUPTED;
3449 }
3450
3451 /* Currently, support 512/1024/2048/4096 bytes sector size */
3452 if (le32_to_cpu(raw_super->log_sectorsize) >
3453 F2FS_MAX_LOG_SECTOR_SIZE ||
3454 le32_to_cpu(raw_super->log_sectorsize) <
3455 F2FS_MIN_LOG_SECTOR_SIZE) {
3456 f2fs_info(sbi, "Invalid log sectorsize (%u)",
3457 le32_to_cpu(raw_super->log_sectorsize));
3458 return -EFSCORRUPTED;
3459 }
3460 if (le32_to_cpu(raw_super->log_sectors_per_block) +
3461 le32_to_cpu(raw_super->log_sectorsize) !=
3462 F2FS_MAX_LOG_SECTOR_SIZE) {
3463 f2fs_info(sbi, "Invalid log sectors per block(%u) log sectorsize(%u)",
3464 le32_to_cpu(raw_super->log_sectors_per_block),
3465 le32_to_cpu(raw_super->log_sectorsize));
3466 return -EFSCORRUPTED;
3467 }
3468
3469 segment_count = le32_to_cpu(raw_super->segment_count);
3470 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
3471 segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3472 secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3473 total_sections = le32_to_cpu(raw_super->section_count);
3474
3475 /* blocks_per_seg should be 512, given the above check */
3476 blocks_per_seg = BIT(le32_to_cpu(raw_super->log_blocks_per_seg));
3477
3478 if (segment_count > F2FS_MAX_SEGMENT ||
3479 segment_count < F2FS_MIN_SEGMENTS) {
3480 f2fs_info(sbi, "Invalid segment count (%u)", segment_count);
3481 return -EFSCORRUPTED;
3482 }
3483
3484 if (total_sections > segment_count_main || total_sections < 1 ||
3485 segs_per_sec > segment_count || !segs_per_sec) {
3486 f2fs_info(sbi, "Invalid segment/section count (%u, %u x %u)",
3487 segment_count, total_sections, segs_per_sec);
3488 return -EFSCORRUPTED;
3489 }
3490
3491 if (segment_count_main != total_sections * segs_per_sec) {
3492 f2fs_info(sbi, "Invalid segment/section count (%u != %u * %u)",
3493 segment_count_main, total_sections, segs_per_sec);
3494 return -EFSCORRUPTED;
3495 }
3496
3497 if ((segment_count / segs_per_sec) < total_sections) {
3498 f2fs_info(sbi, "Small segment_count (%u < %u * %u)",
3499 segment_count, segs_per_sec, total_sections);
3500 return -EFSCORRUPTED;
3501 }
3502
3503 if (segment_count > (le64_to_cpu(raw_super->block_count) >> 9)) {
3504 f2fs_info(sbi, "Wrong segment_count / block_count (%u > %llu)",
3505 segment_count, le64_to_cpu(raw_super->block_count));
3506 return -EFSCORRUPTED;
3507 }
3508
3509 if (RDEV(0).path[0]) {
3510 block_t dev_seg_count = le32_to_cpu(RDEV(0).total_segments);
3511 int i = 1;
3512
3513 while (i < MAX_DEVICES && RDEV(i).path[0]) {
3514 dev_seg_count += le32_to_cpu(RDEV(i).total_segments);
3515 i++;
3516 }
3517 if (segment_count != dev_seg_count) {
3518 f2fs_info(sbi, "Segment count (%u) mismatch with total segments from devices (%u)",
3519 segment_count, dev_seg_count);
3520 return -EFSCORRUPTED;
3521 }
3522 } else {
3523 if (__F2FS_HAS_FEATURE(raw_super, F2FS_FEATURE_BLKZONED) &&
3524 !bdev_is_zoned(sbi->sb->s_bdev)) {
3525 f2fs_info(sbi, "Zoned block device path is missing");
3526 return -EFSCORRUPTED;
3527 }
3528 }
3529
3530 if (secs_per_zone > total_sections || !secs_per_zone) {
3531 f2fs_info(sbi, "Wrong secs_per_zone / total_sections (%u, %u)",
3532 secs_per_zone, total_sections);
3533 return -EFSCORRUPTED;
3534 }
3535 if (le32_to_cpu(raw_super->extension_count) > F2FS_MAX_EXTENSION ||
3536 raw_super->hot_ext_count > F2FS_MAX_EXTENSION ||
3537 (le32_to_cpu(raw_super->extension_count) +
3538 raw_super->hot_ext_count) > F2FS_MAX_EXTENSION) {
3539 f2fs_info(sbi, "Corrupted extension count (%u + %u > %u)",
3540 le32_to_cpu(raw_super->extension_count),
3541 raw_super->hot_ext_count,
3542 F2FS_MAX_EXTENSION);
3543 return -EFSCORRUPTED;
3544 }
3545
3546 if (le32_to_cpu(raw_super->cp_payload) >=
3547 (blocks_per_seg - F2FS_CP_PACKS -
3548 NR_CURSEG_PERSIST_TYPE)) {
3549 f2fs_info(sbi, "Insane cp_payload (%u >= %u)",
3550 le32_to_cpu(raw_super->cp_payload),
3551 blocks_per_seg - F2FS_CP_PACKS -
3552 NR_CURSEG_PERSIST_TYPE);
3553 return -EFSCORRUPTED;
3554 }
3555
3556 /* check reserved ino info */
3557 if (le32_to_cpu(raw_super->node_ino) != 1 ||
3558 le32_to_cpu(raw_super->meta_ino) != 2 ||
3559 le32_to_cpu(raw_super->root_ino) != 3) {
3560 f2fs_info(sbi, "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
3561 le32_to_cpu(raw_super->node_ino),
3562 le32_to_cpu(raw_super->meta_ino),
3563 le32_to_cpu(raw_super->root_ino));
3564 return -EFSCORRUPTED;
3565 }
3566
3567 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
3568 if (sanity_check_area_boundary(sbi, bh))
3569 return -EFSCORRUPTED;
3570
3571 return 0;
3572 }
3573
f2fs_sanity_check_ckpt(struct f2fs_sb_info * sbi)3574 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi)
3575 {
3576 unsigned int total, fsmeta;
3577 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
3578 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
3579 unsigned int ovp_segments, reserved_segments;
3580 unsigned int main_segs, blocks_per_seg;
3581 unsigned int sit_segs, nat_segs;
3582 unsigned int sit_bitmap_size, nat_bitmap_size;
3583 unsigned int log_blocks_per_seg;
3584 unsigned int segment_count_main;
3585 unsigned int cp_pack_start_sum, cp_payload;
3586 block_t user_block_count, valid_user_blocks;
3587 block_t avail_node_count, valid_node_count;
3588 unsigned int nat_blocks, nat_bits_bytes, nat_bits_blocks;
3589 int i, j;
3590
3591 total = le32_to_cpu(raw_super->segment_count);
3592 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
3593 sit_segs = le32_to_cpu(raw_super->segment_count_sit);
3594 fsmeta += sit_segs;
3595 nat_segs = le32_to_cpu(raw_super->segment_count_nat);
3596 fsmeta += nat_segs;
3597 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
3598 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
3599
3600 if (unlikely(fsmeta >= total))
3601 return 1;
3602
3603 ovp_segments = le32_to_cpu(ckpt->overprov_segment_count);
3604 reserved_segments = le32_to_cpu(ckpt->rsvd_segment_count);
3605
3606 if (!f2fs_sb_has_readonly(sbi) &&
3607 unlikely(fsmeta < F2FS_MIN_META_SEGMENTS ||
3608 ovp_segments == 0 || reserved_segments == 0)) {
3609 f2fs_err(sbi, "Wrong layout: check mkfs.f2fs version");
3610 return 1;
3611 }
3612 user_block_count = le64_to_cpu(ckpt->user_block_count);
3613 segment_count_main = le32_to_cpu(raw_super->segment_count_main) +
3614 (f2fs_sb_has_readonly(sbi) ? 1 : 0);
3615 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3616 if (!user_block_count || user_block_count >=
3617 segment_count_main << log_blocks_per_seg) {
3618 f2fs_err(sbi, "Wrong user_block_count: %u",
3619 user_block_count);
3620 return 1;
3621 }
3622
3623 valid_user_blocks = le64_to_cpu(ckpt->valid_block_count);
3624 if (valid_user_blocks > user_block_count) {
3625 f2fs_err(sbi, "Wrong valid_user_blocks: %u, user_block_count: %u",
3626 valid_user_blocks, user_block_count);
3627 return 1;
3628 }
3629
3630 valid_node_count = le32_to_cpu(ckpt->valid_node_count);
3631 avail_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
3632 if (valid_node_count > avail_node_count) {
3633 f2fs_err(sbi, "Wrong valid_node_count: %u, avail_node_count: %u",
3634 valid_node_count, avail_node_count);
3635 return 1;
3636 }
3637
3638 main_segs = le32_to_cpu(raw_super->segment_count_main);
3639 blocks_per_seg = BLKS_PER_SEG(sbi);
3640
3641 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3642 if (le32_to_cpu(ckpt->cur_node_segno[i]) >= main_segs ||
3643 le16_to_cpu(ckpt->cur_node_blkoff[i]) >= blocks_per_seg)
3644 return 1;
3645
3646 if (f2fs_sb_has_readonly(sbi))
3647 goto check_data;
3648
3649 for (j = i + 1; j < NR_CURSEG_NODE_TYPE; j++) {
3650 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3651 le32_to_cpu(ckpt->cur_node_segno[j])) {
3652 f2fs_err(sbi, "Node segment (%u, %u) has the same segno: %u",
3653 i, j,
3654 le32_to_cpu(ckpt->cur_node_segno[i]));
3655 return 1;
3656 }
3657 }
3658 }
3659 check_data:
3660 for (i = 0; i < NR_CURSEG_DATA_TYPE; i++) {
3661 if (le32_to_cpu(ckpt->cur_data_segno[i]) >= main_segs ||
3662 le16_to_cpu(ckpt->cur_data_blkoff[i]) >= blocks_per_seg)
3663 return 1;
3664
3665 if (f2fs_sb_has_readonly(sbi))
3666 goto skip_cross;
3667
3668 for (j = i + 1; j < NR_CURSEG_DATA_TYPE; j++) {
3669 if (le32_to_cpu(ckpt->cur_data_segno[i]) ==
3670 le32_to_cpu(ckpt->cur_data_segno[j])) {
3671 f2fs_err(sbi, "Data segment (%u, %u) has the same segno: %u",
3672 i, j,
3673 le32_to_cpu(ckpt->cur_data_segno[i]));
3674 return 1;
3675 }
3676 }
3677 }
3678 for (i = 0; i < NR_CURSEG_NODE_TYPE; i++) {
3679 for (j = 0; j < NR_CURSEG_DATA_TYPE; j++) {
3680 if (le32_to_cpu(ckpt->cur_node_segno[i]) ==
3681 le32_to_cpu(ckpt->cur_data_segno[j])) {
3682 f2fs_err(sbi, "Node segment (%u) and Data segment (%u) has the same segno: %u",
3683 i, j,
3684 le32_to_cpu(ckpt->cur_node_segno[i]));
3685 return 1;
3686 }
3687 }
3688 }
3689 skip_cross:
3690 sit_bitmap_size = le32_to_cpu(ckpt->sit_ver_bitmap_bytesize);
3691 nat_bitmap_size = le32_to_cpu(ckpt->nat_ver_bitmap_bytesize);
3692
3693 if (sit_bitmap_size != ((sit_segs / 2) << log_blocks_per_seg) / 8 ||
3694 nat_bitmap_size != ((nat_segs / 2) << log_blocks_per_seg) / 8) {
3695 f2fs_err(sbi, "Wrong bitmap size: sit: %u, nat:%u",
3696 sit_bitmap_size, nat_bitmap_size);
3697 return 1;
3698 }
3699
3700 cp_pack_start_sum = __start_sum_addr(sbi);
3701 cp_payload = __cp_payload(sbi);
3702 if (cp_pack_start_sum < cp_payload + 1 ||
3703 cp_pack_start_sum > blocks_per_seg - 1 -
3704 NR_CURSEG_PERSIST_TYPE) {
3705 f2fs_err(sbi, "Wrong cp_pack_start_sum: %u",
3706 cp_pack_start_sum);
3707 return 1;
3708 }
3709
3710 if (__is_set_ckpt_flags(ckpt, CP_LARGE_NAT_BITMAP_FLAG) &&
3711 le32_to_cpu(ckpt->checksum_offset) != CP_MIN_CHKSUM_OFFSET) {
3712 f2fs_warn(sbi, "using deprecated layout of large_nat_bitmap, "
3713 "please run fsck v1.13.0 or higher to repair, chksum_offset: %u, "
3714 "fixed with patch: \"f2fs-tools: relocate chksum_offset for large_nat_bitmap feature\"",
3715 le32_to_cpu(ckpt->checksum_offset));
3716 return 1;
3717 }
3718
3719 nat_blocks = nat_segs << log_blocks_per_seg;
3720 nat_bits_bytes = nat_blocks / BITS_PER_BYTE;
3721 nat_bits_blocks = F2FS_BLK_ALIGN((nat_bits_bytes << 1) + 8);
3722 if (__is_set_ckpt_flags(ckpt, CP_NAT_BITS_FLAG) &&
3723 (cp_payload + F2FS_CP_PACKS +
3724 NR_CURSEG_PERSIST_TYPE + nat_bits_blocks >= blocks_per_seg)) {
3725 f2fs_warn(sbi, "Insane cp_payload: %u, nat_bits_blocks: %u)",
3726 cp_payload, nat_bits_blocks);
3727 return 1;
3728 }
3729
3730 if (unlikely(f2fs_cp_error(sbi))) {
3731 f2fs_err(sbi, "A bug case: need to run fsck");
3732 return 1;
3733 }
3734 return 0;
3735 }
3736
init_sb_info(struct f2fs_sb_info * sbi)3737 static void init_sb_info(struct f2fs_sb_info *sbi)
3738 {
3739 struct f2fs_super_block *raw_super = sbi->raw_super;
3740 int i;
3741
3742 sbi->log_sectors_per_block =
3743 le32_to_cpu(raw_super->log_sectors_per_block);
3744 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
3745 sbi->blocksize = BIT(sbi->log_blocksize);
3746 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
3747 sbi->blocks_per_seg = BIT(sbi->log_blocks_per_seg);
3748 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
3749 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
3750 sbi->total_sections = le32_to_cpu(raw_super->section_count);
3751 sbi->total_node_count =
3752 ((le32_to_cpu(raw_super->segment_count_nat) / 2) *
3753 NAT_ENTRY_PER_BLOCK) << sbi->log_blocks_per_seg;
3754 F2FS_ROOT_INO(sbi) = le32_to_cpu(raw_super->root_ino);
3755 F2FS_NODE_INO(sbi) = le32_to_cpu(raw_super->node_ino);
3756 F2FS_META_INO(sbi) = le32_to_cpu(raw_super->meta_ino);
3757 sbi->cur_victim_sec = NULL_SECNO;
3758 sbi->gc_mode = GC_NORMAL;
3759 sbi->next_victim_seg[BG_GC] = NULL_SEGNO;
3760 sbi->next_victim_seg[FG_GC] = NULL_SEGNO;
3761 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
3762 sbi->migration_granularity = SEGS_PER_SEC(sbi);
3763 sbi->seq_file_ra_mul = MIN_RA_MUL;
3764 sbi->max_fragment_chunk = DEF_FRAGMENT_SIZE;
3765 sbi->max_fragment_hole = DEF_FRAGMENT_SIZE;
3766 spin_lock_init(&sbi->gc_remaining_trials_lock);
3767 atomic64_set(&sbi->current_atomic_write, 0);
3768
3769 sbi->dir_level = DEF_DIR_LEVEL;
3770 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
3771 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
3772 sbi->interval_time[DISCARD_TIME] = DEF_IDLE_INTERVAL;
3773 sbi->interval_time[GC_TIME] = DEF_IDLE_INTERVAL;
3774 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_INTERVAL;
3775 sbi->interval_time[UMOUNT_DISCARD_TIMEOUT] =
3776 DEF_UMOUNT_DISCARD_TIMEOUT;
3777 clear_sbi_flag(sbi, SBI_NEED_FSCK);
3778
3779 for (i = 0; i < NR_COUNT_TYPE; i++)
3780 atomic_set(&sbi->nr_pages[i], 0);
3781
3782 for (i = 0; i < META; i++)
3783 atomic_set(&sbi->wb_sync_req[i], 0);
3784
3785 INIT_LIST_HEAD(&sbi->s_list);
3786 mutex_init(&sbi->umount_mutex);
3787 init_f2fs_rwsem(&sbi->io_order_lock);
3788 spin_lock_init(&sbi->cp_lock);
3789
3790 sbi->dirty_device = 0;
3791 spin_lock_init(&sbi->dev_lock);
3792
3793 init_f2fs_rwsem(&sbi->sb_lock);
3794 init_f2fs_rwsem(&sbi->pin_sem);
3795 }
3796
init_percpu_info(struct f2fs_sb_info * sbi)3797 static int init_percpu_info(struct f2fs_sb_info *sbi)
3798 {
3799 int err;
3800
3801 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
3802 if (err)
3803 return err;
3804
3805 err = percpu_counter_init(&sbi->rf_node_block_count, 0, GFP_KERNEL);
3806 if (err)
3807 goto err_valid_block;
3808
3809 err = percpu_counter_init(&sbi->total_valid_inode_count, 0,
3810 GFP_KERNEL);
3811 if (err)
3812 goto err_node_block;
3813 return 0;
3814
3815 err_node_block:
3816 percpu_counter_destroy(&sbi->rf_node_block_count);
3817 err_valid_block:
3818 percpu_counter_destroy(&sbi->alloc_valid_block_count);
3819 return err;
3820 }
3821
3822 #ifdef CONFIG_BLK_DEV_ZONED
3823
3824 struct f2fs_report_zones_args {
3825 struct f2fs_sb_info *sbi;
3826 struct f2fs_dev_info *dev;
3827 };
3828
f2fs_report_zone_cb(struct blk_zone * zone,unsigned int idx,void * data)3829 static int f2fs_report_zone_cb(struct blk_zone *zone, unsigned int idx,
3830 void *data)
3831 {
3832 struct f2fs_report_zones_args *rz_args = data;
3833 block_t unusable_blocks = (zone->len - zone->capacity) >>
3834 F2FS_LOG_SECTORS_PER_BLOCK;
3835
3836 if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
3837 return 0;
3838
3839 set_bit(idx, rz_args->dev->blkz_seq);
3840 if (!rz_args->sbi->unusable_blocks_per_sec) {
3841 rz_args->sbi->unusable_blocks_per_sec = unusable_blocks;
3842 return 0;
3843 }
3844 if (rz_args->sbi->unusable_blocks_per_sec != unusable_blocks) {
3845 f2fs_err(rz_args->sbi, "F2FS supports single zone capacity\n");
3846 return -EINVAL;
3847 }
3848 return 0;
3849 }
3850
init_blkz_info(struct f2fs_sb_info * sbi,int devi)3851 static int init_blkz_info(struct f2fs_sb_info *sbi, int devi)
3852 {
3853 struct block_device *bdev = FDEV(devi).bdev;
3854 sector_t nr_sectors = bdev_nr_sectors(bdev);
3855 struct f2fs_report_zones_args rep_zone_arg;
3856 u64 zone_sectors;
3857 int ret;
3858
3859 if (!f2fs_sb_has_blkzoned(sbi))
3860 return 0;
3861
3862 zone_sectors = bdev_zone_sectors(bdev);
3863 if (sbi->blocks_per_blkz && sbi->blocks_per_blkz !=
3864 SECTOR_TO_BLOCK(zone_sectors))
3865 return -EINVAL;
3866 sbi->blocks_per_blkz = SECTOR_TO_BLOCK(zone_sectors);
3867 FDEV(devi).nr_blkz = div_u64(SECTOR_TO_BLOCK(nr_sectors),
3868 sbi->blocks_per_blkz);
3869 if (nr_sectors & (zone_sectors - 1))
3870 FDEV(devi).nr_blkz++;
3871
3872 FDEV(devi).blkz_seq = f2fs_kvzalloc(sbi,
3873 BITS_TO_LONGS(FDEV(devi).nr_blkz)
3874 * sizeof(unsigned long),
3875 GFP_KERNEL);
3876 if (!FDEV(devi).blkz_seq)
3877 return -ENOMEM;
3878
3879 rep_zone_arg.sbi = sbi;
3880 rep_zone_arg.dev = &FDEV(devi);
3881
3882 ret = blkdev_report_zones(bdev, 0, BLK_ALL_ZONES, f2fs_report_zone_cb,
3883 &rep_zone_arg);
3884 if (ret < 0)
3885 return ret;
3886 return 0;
3887 }
3888 #endif
3889
3890 /*
3891 * Read f2fs raw super block.
3892 * Because we have two copies of super block, so read both of them
3893 * to get the first valid one. If any one of them is broken, we pass
3894 * them recovery flag back to the caller.
3895 */
read_raw_super_block(struct f2fs_sb_info * sbi,struct f2fs_super_block ** raw_super,int * valid_super_block,int * recovery)3896 static int read_raw_super_block(struct f2fs_sb_info *sbi,
3897 struct f2fs_super_block **raw_super,
3898 int *valid_super_block, int *recovery)
3899 {
3900 struct super_block *sb = sbi->sb;
3901 int block;
3902 struct buffer_head *bh;
3903 struct f2fs_super_block *super;
3904 int err = 0;
3905
3906 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
3907 if (!super)
3908 return -ENOMEM;
3909
3910 for (block = 0; block < 2; block++) {
3911 bh = sb_bread(sb, block);
3912 if (!bh) {
3913 f2fs_err(sbi, "Unable to read %dth superblock",
3914 block + 1);
3915 err = -EIO;
3916 *recovery = 1;
3917 continue;
3918 }
3919
3920 /* sanity checking of raw super */
3921 err = sanity_check_raw_super(sbi, bh);
3922 if (err) {
3923 f2fs_err(sbi, "Can't find valid F2FS filesystem in %dth superblock",
3924 block + 1);
3925 brelse(bh);
3926 *recovery = 1;
3927 continue;
3928 }
3929
3930 if (!*raw_super) {
3931 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
3932 sizeof(*super));
3933 *valid_super_block = block;
3934 *raw_super = super;
3935 }
3936 brelse(bh);
3937 }
3938
3939 /* No valid superblock */
3940 if (!*raw_super)
3941 kfree(super);
3942 else
3943 err = 0;
3944
3945 return err;
3946 }
3947
f2fs_commit_super(struct f2fs_sb_info * sbi,bool recover)3948 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
3949 {
3950 struct buffer_head *bh;
3951 __u32 crc = 0;
3952 int err;
3953
3954 if ((recover && f2fs_readonly(sbi->sb)) ||
3955 f2fs_hw_is_readonly(sbi)) {
3956 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
3957 return -EROFS;
3958 }
3959
3960 /* we should update superblock crc here */
3961 if (!recover && f2fs_sb_has_sb_chksum(sbi)) {
3962 crc = f2fs_crc32(sbi, F2FS_RAW_SUPER(sbi),
3963 offsetof(struct f2fs_super_block, crc));
3964 F2FS_RAW_SUPER(sbi)->crc = cpu_to_le32(crc);
3965 }
3966
3967 /* write back-up superblock first */
3968 bh = sb_bread(sbi->sb, sbi->valid_super_block ? 0 : 1);
3969 if (!bh)
3970 return -EIO;
3971 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3972 brelse(bh);
3973
3974 /* if we are in recovery path, skip writing valid superblock */
3975 if (recover || err)
3976 return err;
3977
3978 /* write current valid superblock */
3979 bh = sb_bread(sbi->sb, sbi->valid_super_block);
3980 if (!bh)
3981 return -EIO;
3982 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
3983 brelse(bh);
3984 return err;
3985 }
3986
save_stop_reason(struct f2fs_sb_info * sbi,unsigned char reason)3987 static void save_stop_reason(struct f2fs_sb_info *sbi, unsigned char reason)
3988 {
3989 unsigned long flags;
3990
3991 spin_lock_irqsave(&sbi->error_lock, flags);
3992 if (sbi->stop_reason[reason] < GENMASK(BITS_PER_BYTE - 1, 0))
3993 sbi->stop_reason[reason]++;
3994 spin_unlock_irqrestore(&sbi->error_lock, flags);
3995 }
3996
f2fs_record_stop_reason(struct f2fs_sb_info * sbi)3997 static void f2fs_record_stop_reason(struct f2fs_sb_info *sbi)
3998 {
3999 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
4000 unsigned long flags;
4001 int err;
4002
4003 f2fs_down_write(&sbi->sb_lock);
4004
4005 spin_lock_irqsave(&sbi->error_lock, flags);
4006 if (sbi->error_dirty) {
4007 memcpy(F2FS_RAW_SUPER(sbi)->s_errors, sbi->errors,
4008 MAX_F2FS_ERRORS);
4009 sbi->error_dirty = false;
4010 }
4011 memcpy(raw_super->s_stop_reason, sbi->stop_reason, MAX_STOP_REASON);
4012 spin_unlock_irqrestore(&sbi->error_lock, flags);
4013
4014 err = f2fs_commit_super(sbi, false);
4015
4016 f2fs_up_write(&sbi->sb_lock);
4017 if (err)
4018 f2fs_err(sbi, "f2fs_commit_super fails to record err:%d", err);
4019 }
4020
f2fs_save_errors(struct f2fs_sb_info * sbi,unsigned char flag)4021 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag)
4022 {
4023 unsigned long flags;
4024
4025 spin_lock_irqsave(&sbi->error_lock, flags);
4026 if (!test_bit(flag, (unsigned long *)sbi->errors)) {
4027 set_bit(flag, (unsigned long *)sbi->errors);
4028 sbi->error_dirty = true;
4029 }
4030 spin_unlock_irqrestore(&sbi->error_lock, flags);
4031 }
4032
f2fs_update_errors(struct f2fs_sb_info * sbi)4033 static bool f2fs_update_errors(struct f2fs_sb_info *sbi)
4034 {
4035 unsigned long flags;
4036 bool need_update = false;
4037
4038 spin_lock_irqsave(&sbi->error_lock, flags);
4039 if (sbi->error_dirty) {
4040 memcpy(F2FS_RAW_SUPER(sbi)->s_errors, sbi->errors,
4041 MAX_F2FS_ERRORS);
4042 sbi->error_dirty = false;
4043 need_update = true;
4044 }
4045 spin_unlock_irqrestore(&sbi->error_lock, flags);
4046
4047 return need_update;
4048 }
4049
f2fs_record_errors(struct f2fs_sb_info * sbi,unsigned char error)4050 static void f2fs_record_errors(struct f2fs_sb_info *sbi, unsigned char error)
4051 {
4052 int err;
4053
4054 f2fs_down_write(&sbi->sb_lock);
4055
4056 if (!f2fs_update_errors(sbi))
4057 goto out_unlock;
4058
4059 err = f2fs_commit_super(sbi, false);
4060 if (err)
4061 f2fs_err(sbi, "f2fs_commit_super fails to record errors:%u, err:%d",
4062 error, err);
4063 out_unlock:
4064 f2fs_up_write(&sbi->sb_lock);
4065 }
4066
f2fs_handle_error(struct f2fs_sb_info * sbi,unsigned char error)4067 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error)
4068 {
4069 f2fs_save_errors(sbi, error);
4070 f2fs_record_errors(sbi, error);
4071 }
4072
f2fs_handle_error_async(struct f2fs_sb_info * sbi,unsigned char error)4073 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error)
4074 {
4075 f2fs_save_errors(sbi, error);
4076
4077 if (!sbi->error_dirty)
4078 return;
4079 if (!test_bit(error, (unsigned long *)sbi->errors))
4080 return;
4081 schedule_work(&sbi->s_error_work);
4082 }
4083
system_going_down(void)4084 static bool system_going_down(void)
4085 {
4086 return system_state == SYSTEM_HALT || system_state == SYSTEM_POWER_OFF
4087 || system_state == SYSTEM_RESTART;
4088 }
4089
f2fs_handle_critical_error(struct f2fs_sb_info * sbi,unsigned char reason,bool irq_context)4090 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason,
4091 bool irq_context)
4092 {
4093 struct super_block *sb = sbi->sb;
4094 bool shutdown = reason == STOP_CP_REASON_SHUTDOWN;
4095 bool continue_fs = !shutdown &&
4096 F2FS_OPTION(sbi).errors == MOUNT_ERRORS_CONTINUE;
4097
4098 set_ckpt_flags(sbi, CP_ERROR_FLAG);
4099
4100 if (!f2fs_hw_is_readonly(sbi)) {
4101 save_stop_reason(sbi, reason);
4102
4103 if (irq_context && !shutdown)
4104 schedule_work(&sbi->s_error_work);
4105 else
4106 f2fs_record_stop_reason(sbi);
4107 }
4108
4109 /*
4110 * We force ERRORS_RO behavior when system is rebooting. Otherwise we
4111 * could panic during 'reboot -f' as the underlying device got already
4112 * disabled.
4113 */
4114 if (F2FS_OPTION(sbi).errors == MOUNT_ERRORS_PANIC &&
4115 !shutdown && !system_going_down() &&
4116 !is_sbi_flag_set(sbi, SBI_IS_SHUTDOWN))
4117 panic("F2FS-fs (device %s): panic forced after error\n",
4118 sb->s_id);
4119
4120 if (shutdown)
4121 set_sbi_flag(sbi, SBI_IS_SHUTDOWN);
4122
4123 /*
4124 * Continue filesystem operators if errors=continue. Should not set
4125 * RO by shutdown, since RO bypasses thaw_super which can hang the
4126 * system.
4127 */
4128 if (continue_fs || f2fs_readonly(sb) || shutdown) {
4129 f2fs_warn(sbi, "Stopped filesystem due to reason: %d", reason);
4130 return;
4131 }
4132
4133 f2fs_warn(sbi, "Remounting filesystem read-only");
4134 /*
4135 * Make sure updated value of ->s_mount_flags will be visible before
4136 * ->s_flags update
4137 */
4138 smp_wmb();
4139 sb->s_flags |= SB_RDONLY;
4140 }
4141
f2fs_record_error_work(struct work_struct * work)4142 static void f2fs_record_error_work(struct work_struct *work)
4143 {
4144 struct f2fs_sb_info *sbi = container_of(work,
4145 struct f2fs_sb_info, s_error_work);
4146
4147 f2fs_record_stop_reason(sbi);
4148 }
4149
f2fs_scan_devices(struct f2fs_sb_info * sbi)4150 static int f2fs_scan_devices(struct f2fs_sb_info *sbi)
4151 {
4152 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
4153 unsigned int max_devices = MAX_DEVICES;
4154 unsigned int logical_blksize;
4155 blk_mode_t mode = sb_open_mode(sbi->sb->s_flags);
4156 int i;
4157
4158 /* Initialize single device information */
4159 if (!RDEV(0).path[0]) {
4160 if (!bdev_is_zoned(sbi->sb->s_bdev))
4161 return 0;
4162 max_devices = 1;
4163 }
4164
4165 /*
4166 * Initialize multiple devices information, or single
4167 * zoned block device information.
4168 */
4169 sbi->devs = f2fs_kzalloc(sbi,
4170 array_size(max_devices,
4171 sizeof(struct f2fs_dev_info)),
4172 GFP_KERNEL);
4173 if (!sbi->devs)
4174 return -ENOMEM;
4175
4176 logical_blksize = bdev_logical_block_size(sbi->sb->s_bdev);
4177 sbi->aligned_blksize = true;
4178
4179 for (i = 0; i < max_devices; i++) {
4180 if (i == 0)
4181 FDEV(0).bdev = sbi->sb->s_bdev;
4182 else if (!RDEV(i).path[0])
4183 break;
4184
4185 if (max_devices > 1) {
4186 /* Multi-device mount */
4187 memcpy(FDEV(i).path, RDEV(i).path, MAX_PATH_LEN);
4188 FDEV(i).total_segments =
4189 le32_to_cpu(RDEV(i).total_segments);
4190 if (i == 0) {
4191 FDEV(i).start_blk = 0;
4192 FDEV(i).end_blk = FDEV(i).start_blk +
4193 (FDEV(i).total_segments <<
4194 sbi->log_blocks_per_seg) - 1 +
4195 le32_to_cpu(raw_super->segment0_blkaddr);
4196 } else {
4197 FDEV(i).start_blk = FDEV(i - 1).end_blk + 1;
4198 FDEV(i).end_blk = FDEV(i).start_blk +
4199 (FDEV(i).total_segments <<
4200 sbi->log_blocks_per_seg) - 1;
4201 FDEV(i).bdev = blkdev_get_by_path(FDEV(i).path,
4202 mode, sbi->sb, NULL);
4203 }
4204 }
4205 if (IS_ERR(FDEV(i).bdev))
4206 return PTR_ERR(FDEV(i).bdev);
4207
4208 /* to release errored devices */
4209 sbi->s_ndevs = i + 1;
4210
4211 if (logical_blksize != bdev_logical_block_size(FDEV(i).bdev))
4212 sbi->aligned_blksize = false;
4213
4214 #ifdef CONFIG_BLK_DEV_ZONED
4215 if (bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HM &&
4216 !f2fs_sb_has_blkzoned(sbi)) {
4217 f2fs_err(sbi, "Zoned block device feature not enabled");
4218 return -EINVAL;
4219 }
4220 if (bdev_zoned_model(FDEV(i).bdev) != BLK_ZONED_NONE) {
4221 if (init_blkz_info(sbi, i)) {
4222 f2fs_err(sbi, "Failed to initialize F2FS blkzone information");
4223 return -EINVAL;
4224 }
4225 if (max_devices == 1)
4226 break;
4227 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x (zone: %s)",
4228 i, FDEV(i).path,
4229 FDEV(i).total_segments,
4230 FDEV(i).start_blk, FDEV(i).end_blk,
4231 bdev_zoned_model(FDEV(i).bdev) == BLK_ZONED_HA ?
4232 "Host-aware" : "Host-managed");
4233 continue;
4234 }
4235 #endif
4236 f2fs_info(sbi, "Mount Device [%2d]: %20s, %8u, %8x - %8x",
4237 i, FDEV(i).path,
4238 FDEV(i).total_segments,
4239 FDEV(i).start_blk, FDEV(i).end_blk);
4240 }
4241 return 0;
4242 }
4243
f2fs_setup_casefold(struct f2fs_sb_info * sbi)4244 static int f2fs_setup_casefold(struct f2fs_sb_info *sbi)
4245 {
4246 #if IS_ENABLED(CONFIG_UNICODE)
4247 if (f2fs_sb_has_casefold(sbi) && !sbi->sb->s_encoding) {
4248 const struct f2fs_sb_encodings *encoding_info;
4249 struct unicode_map *encoding;
4250 __u16 encoding_flags;
4251
4252 encoding_info = f2fs_sb_read_encoding(sbi->raw_super);
4253 if (!encoding_info) {
4254 f2fs_err(sbi,
4255 "Encoding requested by superblock is unknown");
4256 return -EINVAL;
4257 }
4258
4259 encoding_flags = le16_to_cpu(sbi->raw_super->s_encoding_flags);
4260 encoding = utf8_load(encoding_info->version);
4261 if (IS_ERR(encoding)) {
4262 f2fs_err(sbi,
4263 "can't mount with superblock charset: %s-%u.%u.%u "
4264 "not supported by the kernel. flags: 0x%x.",
4265 encoding_info->name,
4266 unicode_major(encoding_info->version),
4267 unicode_minor(encoding_info->version),
4268 unicode_rev(encoding_info->version),
4269 encoding_flags);
4270 return PTR_ERR(encoding);
4271 }
4272 f2fs_info(sbi, "Using encoding defined by superblock: "
4273 "%s-%u.%u.%u with flags 0x%hx", encoding_info->name,
4274 unicode_major(encoding_info->version),
4275 unicode_minor(encoding_info->version),
4276 unicode_rev(encoding_info->version),
4277 encoding_flags);
4278
4279 sbi->sb->s_encoding = encoding;
4280 sbi->sb->s_encoding_flags = encoding_flags;
4281 }
4282 #else
4283 if (f2fs_sb_has_casefold(sbi)) {
4284 f2fs_err(sbi, "Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE");
4285 return -EINVAL;
4286 }
4287 #endif
4288 return 0;
4289 }
4290
f2fs_tuning_parameters(struct f2fs_sb_info * sbi)4291 static void f2fs_tuning_parameters(struct f2fs_sb_info *sbi)
4292 {
4293 /* adjust parameters according to the volume size */
4294 if (MAIN_SEGS(sbi) <= SMALL_VOLUME_SEGMENTS) {
4295 if (f2fs_block_unit_discard(sbi))
4296 SM_I(sbi)->dcc_info->discard_granularity =
4297 MIN_DISCARD_GRANULARITY;
4298 if (!f2fs_lfs_mode(sbi))
4299 SM_I(sbi)->ipu_policy = BIT(F2FS_IPU_FORCE) |
4300 BIT(F2FS_IPU_HONOR_OPU_WRITE);
4301 }
4302
4303 sbi->readdir_ra = true;
4304 }
4305
f2fs_fill_super(struct super_block * sb,void * data,int silent)4306 static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
4307 {
4308 struct f2fs_sb_info *sbi;
4309 struct f2fs_super_block *raw_super;
4310 struct inode *root;
4311 int err;
4312 bool skip_recovery = false, need_fsck = false;
4313 char *options = NULL;
4314 int recovery, i, valid_super_block;
4315 struct curseg_info *seg_i;
4316 int retry_cnt = 1;
4317 #ifdef CONFIG_QUOTA
4318 bool quota_enabled = false;
4319 #endif
4320
4321 try_onemore:
4322 err = -EINVAL;
4323 raw_super = NULL;
4324 valid_super_block = -1;
4325 recovery = 0;
4326
4327 /* allocate memory for f2fs-specific super block info */
4328 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
4329 if (!sbi)
4330 return -ENOMEM;
4331
4332 sbi->sb = sb;
4333
4334 /* initialize locks within allocated memory */
4335 init_f2fs_rwsem(&sbi->gc_lock);
4336 mutex_init(&sbi->writepages);
4337 init_f2fs_rwsem(&sbi->cp_global_sem);
4338 init_f2fs_rwsem(&sbi->node_write);
4339 init_f2fs_rwsem(&sbi->node_change);
4340 spin_lock_init(&sbi->stat_lock);
4341 init_f2fs_rwsem(&sbi->cp_rwsem);
4342 init_f2fs_rwsem(&sbi->quota_sem);
4343 init_waitqueue_head(&sbi->cp_wait);
4344 spin_lock_init(&sbi->error_lock);
4345
4346 for (i = 0; i < NR_INODE_TYPE; i++) {
4347 INIT_LIST_HEAD(&sbi->inode_list[i]);
4348 spin_lock_init(&sbi->inode_lock[i]);
4349 }
4350 mutex_init(&sbi->flush_lock);
4351
4352 /* Load the checksum driver */
4353 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
4354 if (IS_ERR(sbi->s_chksum_driver)) {
4355 f2fs_err(sbi, "Cannot load crc32 driver.");
4356 err = PTR_ERR(sbi->s_chksum_driver);
4357 sbi->s_chksum_driver = NULL;
4358 goto free_sbi;
4359 }
4360
4361 /* set a block size */
4362 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
4363 f2fs_err(sbi, "unable to set blocksize");
4364 goto free_sbi;
4365 }
4366
4367 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
4368 &recovery);
4369 if (err)
4370 goto free_sbi;
4371
4372 sb->s_fs_info = sbi;
4373 sbi->raw_super = raw_super;
4374
4375 INIT_WORK(&sbi->s_error_work, f2fs_record_error_work);
4376 memcpy(sbi->errors, raw_super->s_errors, MAX_F2FS_ERRORS);
4377 memcpy(sbi->stop_reason, raw_super->s_stop_reason, MAX_STOP_REASON);
4378
4379 /* precompute checksum seed for metadata */
4380 if (f2fs_sb_has_inode_chksum(sbi))
4381 sbi->s_chksum_seed = f2fs_chksum(sbi, ~0, raw_super->uuid,
4382 sizeof(raw_super->uuid));
4383
4384 default_options(sbi, false);
4385 /* parse mount options */
4386 options = kstrdup((const char *)data, GFP_KERNEL);
4387 if (data && !options) {
4388 err = -ENOMEM;
4389 goto free_sb_buf;
4390 }
4391
4392 err = parse_options(sb, options, false);
4393 if (err)
4394 goto free_options;
4395
4396 sb->s_maxbytes = max_file_blocks(NULL) <<
4397 le32_to_cpu(raw_super->log_blocksize);
4398 sb->s_max_links = F2FS_LINK_MAX;
4399
4400 err = f2fs_setup_casefold(sbi);
4401 if (err)
4402 goto free_options;
4403
4404 #ifdef CONFIG_QUOTA
4405 sb->dq_op = &f2fs_quota_operations;
4406 sb->s_qcop = &f2fs_quotactl_ops;
4407 sb->s_quota_types = QTYPE_MASK_USR | QTYPE_MASK_GRP | QTYPE_MASK_PRJ;
4408
4409 if (f2fs_sb_has_quota_ino(sbi)) {
4410 for (i = 0; i < MAXQUOTAS; i++) {
4411 if (f2fs_qf_ino(sbi->sb, i))
4412 sbi->nquota_files++;
4413 }
4414 }
4415 #endif
4416
4417 sb->s_op = &f2fs_sops;
4418 #ifdef CONFIG_FS_ENCRYPTION
4419 sb->s_cop = &f2fs_cryptops;
4420 #endif
4421 #ifdef CONFIG_FS_VERITY
4422 sb->s_vop = &f2fs_verityops;
4423 #endif
4424 sb->s_xattr = f2fs_xattr_handlers;
4425 sb->s_export_op = &f2fs_export_ops;
4426 sb->s_magic = F2FS_SUPER_MAGIC;
4427 sb->s_time_gran = 1;
4428 sb->s_flags = (sb->s_flags & ~SB_POSIXACL) |
4429 (test_opt(sbi, POSIX_ACL) ? SB_POSIXACL : 0);
4430 memcpy(&sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
4431 sb->s_iflags |= SB_I_CGROUPWB;
4432
4433 /* init f2fs-specific super block info */
4434 sbi->valid_super_block = valid_super_block;
4435
4436 /* disallow all the data/node/meta page writes */
4437 set_sbi_flag(sbi, SBI_POR_DOING);
4438
4439 err = f2fs_init_write_merge_io(sbi);
4440 if (err)
4441 goto free_bio_info;
4442
4443 init_sb_info(sbi);
4444
4445 err = f2fs_init_iostat(sbi);
4446 if (err)
4447 goto free_bio_info;
4448
4449 err = init_percpu_info(sbi);
4450 if (err)
4451 goto free_iostat;
4452
4453 /* init per sbi slab cache */
4454 err = f2fs_init_xattr_caches(sbi);
4455 if (err)
4456 goto free_percpu;
4457 err = f2fs_init_page_array_cache(sbi);
4458 if (err)
4459 goto free_xattr_cache;
4460
4461 /* get an inode for meta space */
4462 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
4463 if (IS_ERR(sbi->meta_inode)) {
4464 f2fs_err(sbi, "Failed to read F2FS meta data inode");
4465 err = PTR_ERR(sbi->meta_inode);
4466 goto free_page_array_cache;
4467 }
4468
4469 err = f2fs_get_valid_checkpoint(sbi);
4470 if (err) {
4471 f2fs_err(sbi, "Failed to get valid F2FS checkpoint");
4472 goto free_meta_inode;
4473 }
4474
4475 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_QUOTA_NEED_FSCK_FLAG))
4476 set_sbi_flag(sbi, SBI_QUOTA_NEED_REPAIR);
4477 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_DISABLED_QUICK_FLAG)) {
4478 set_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4479 sbi->interval_time[DISABLE_TIME] = DEF_DISABLE_QUICK_INTERVAL;
4480 }
4481
4482 if (__is_set_ckpt_flags(F2FS_CKPT(sbi), CP_FSCK_FLAG))
4483 set_sbi_flag(sbi, SBI_NEED_FSCK);
4484
4485 /* Initialize device list */
4486 err = f2fs_scan_devices(sbi);
4487 if (err) {
4488 f2fs_err(sbi, "Failed to find devices");
4489 goto free_devices;
4490 }
4491
4492 err = f2fs_init_post_read_wq(sbi);
4493 if (err) {
4494 f2fs_err(sbi, "Failed to initialize post read workqueue");
4495 goto free_devices;
4496 }
4497
4498 sbi->total_valid_node_count =
4499 le32_to_cpu(sbi->ckpt->valid_node_count);
4500 percpu_counter_set(&sbi->total_valid_inode_count,
4501 le32_to_cpu(sbi->ckpt->valid_inode_count));
4502 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
4503 sbi->total_valid_block_count =
4504 le64_to_cpu(sbi->ckpt->valid_block_count);
4505 sbi->last_valid_block_count = sbi->total_valid_block_count;
4506 sbi->reserved_blocks = 0;
4507 sbi->current_reserved_blocks = 0;
4508 limit_reserve_root(sbi);
4509 adjust_unusable_cap_perc(sbi);
4510
4511 f2fs_init_extent_cache_info(sbi);
4512
4513 f2fs_init_ino_entry_info(sbi);
4514
4515 f2fs_init_fsync_node_info(sbi);
4516
4517 /* setup checkpoint request control and start checkpoint issue thread */
4518 f2fs_init_ckpt_req_control(sbi);
4519 if (!f2fs_readonly(sb) && !test_opt(sbi, DISABLE_CHECKPOINT) &&
4520 test_opt(sbi, MERGE_CHECKPOINT)) {
4521 err = f2fs_start_ckpt_thread(sbi);
4522 if (err) {
4523 f2fs_err(sbi,
4524 "Failed to start F2FS issue_checkpoint_thread (%d)",
4525 err);
4526 goto stop_ckpt_thread;
4527 }
4528 }
4529
4530 /* setup f2fs internal modules */
4531 err = f2fs_build_segment_manager(sbi);
4532 if (err) {
4533 f2fs_err(sbi, "Failed to initialize F2FS segment manager (%d)",
4534 err);
4535 goto free_sm;
4536 }
4537 err = f2fs_build_node_manager(sbi);
4538 if (err) {
4539 f2fs_err(sbi, "Failed to initialize F2FS node manager (%d)",
4540 err);
4541 goto free_nm;
4542 }
4543
4544 /* For write statistics */
4545 sbi->sectors_written_start = f2fs_get_sectors_written(sbi);
4546
4547 /* Read accumulated write IO statistics if exists */
4548 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
4549 if (__exist_node_summaries(sbi))
4550 sbi->kbytes_written =
4551 le64_to_cpu(seg_i->journal->info.kbytes_written);
4552
4553 f2fs_build_gc_manager(sbi);
4554
4555 err = f2fs_build_stats(sbi);
4556 if (err)
4557 goto free_nm;
4558
4559 /* get an inode for node space */
4560 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
4561 if (IS_ERR(sbi->node_inode)) {
4562 f2fs_err(sbi, "Failed to read node inode");
4563 err = PTR_ERR(sbi->node_inode);
4564 goto free_stats;
4565 }
4566
4567 /* read root inode and dentry */
4568 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
4569 if (IS_ERR(root)) {
4570 f2fs_err(sbi, "Failed to read root inode");
4571 err = PTR_ERR(root);
4572 goto free_node_inode;
4573 }
4574 if (!S_ISDIR(root->i_mode) || !root->i_blocks ||
4575 !root->i_size || !root->i_nlink) {
4576 iput(root);
4577 err = -EINVAL;
4578 goto free_node_inode;
4579 }
4580
4581 sb->s_root = d_make_root(root); /* allocate root dentry */
4582 if (!sb->s_root) {
4583 err = -ENOMEM;
4584 goto free_node_inode;
4585 }
4586
4587 err = f2fs_init_compress_inode(sbi);
4588 if (err)
4589 goto free_root_inode;
4590
4591 err = f2fs_register_sysfs(sbi);
4592 if (err)
4593 goto free_compress_inode;
4594
4595 #ifdef CONFIG_QUOTA
4596 /* Enable quota usage during mount */
4597 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb)) {
4598 err = f2fs_enable_quotas(sb);
4599 if (err)
4600 f2fs_err(sbi, "Cannot turn on quotas: error %d", err);
4601 }
4602
4603 quota_enabled = f2fs_recover_quota_begin(sbi);
4604 #endif
4605 /* if there are any orphan inodes, free them */
4606 err = f2fs_recover_orphan_inodes(sbi);
4607 if (err)
4608 goto free_meta;
4609
4610 if (unlikely(is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)))
4611 goto reset_checkpoint;
4612
4613 /* recover fsynced data */
4614 if (!test_opt(sbi, DISABLE_ROLL_FORWARD) &&
4615 !test_opt(sbi, NORECOVERY)) {
4616 /*
4617 * mount should be failed, when device has readonly mode, and
4618 * previous checkpoint was not done by clean system shutdown.
4619 */
4620 if (f2fs_hw_is_readonly(sbi)) {
4621 if (!is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4622 err = f2fs_recover_fsync_data(sbi, true);
4623 if (err > 0) {
4624 err = -EROFS;
4625 f2fs_err(sbi, "Need to recover fsync data, but "
4626 "write access unavailable, please try "
4627 "mount w/ disable_roll_forward or norecovery");
4628 }
4629 if (err < 0)
4630 goto free_meta;
4631 }
4632 f2fs_info(sbi, "write access unavailable, skipping recovery");
4633 goto reset_checkpoint;
4634 }
4635
4636 if (need_fsck)
4637 set_sbi_flag(sbi, SBI_NEED_FSCK);
4638
4639 if (skip_recovery)
4640 goto reset_checkpoint;
4641
4642 err = f2fs_recover_fsync_data(sbi, false);
4643 if (err < 0) {
4644 if (err != -ENOMEM)
4645 skip_recovery = true;
4646 need_fsck = true;
4647 f2fs_err(sbi, "Cannot recover all fsync data errno=%d",
4648 err);
4649 goto free_meta;
4650 }
4651 } else {
4652 err = f2fs_recover_fsync_data(sbi, true);
4653
4654 if (!f2fs_readonly(sb) && err > 0) {
4655 err = -EINVAL;
4656 f2fs_err(sbi, "Need to recover fsync data");
4657 goto free_meta;
4658 }
4659 }
4660
4661 #ifdef CONFIG_QUOTA
4662 f2fs_recover_quota_end(sbi, quota_enabled);
4663 #endif
4664
4665 /*
4666 * If the f2fs is not readonly and fsync data recovery succeeds,
4667 * check zoned block devices' write pointer consistency.
4668 */
4669 if (!err && !f2fs_readonly(sb) && f2fs_sb_has_blkzoned(sbi)) {
4670 err = f2fs_check_write_pointer(sbi);
4671 if (err)
4672 goto free_meta;
4673 }
4674
4675 reset_checkpoint:
4676 f2fs_init_inmem_curseg(sbi);
4677
4678 /* f2fs_recover_fsync_data() cleared this already */
4679 clear_sbi_flag(sbi, SBI_POR_DOING);
4680
4681 if (test_opt(sbi, DISABLE_CHECKPOINT)) {
4682 err = f2fs_disable_checkpoint(sbi);
4683 if (err)
4684 goto sync_free_meta;
4685 } else if (is_set_ckpt_flags(sbi, CP_DISABLED_FLAG)) {
4686 f2fs_enable_checkpoint(sbi);
4687 }
4688
4689 /*
4690 * If filesystem is not mounted as read-only then
4691 * do start the gc_thread.
4692 */
4693 if ((F2FS_OPTION(sbi).bggc_mode != BGGC_MODE_OFF ||
4694 test_opt(sbi, GC_MERGE)) && !f2fs_readonly(sb)) {
4695 /* After POR, we can run background GC thread.*/
4696 err = f2fs_start_gc_thread(sbi);
4697 if (err)
4698 goto sync_free_meta;
4699 }
4700 kvfree(options);
4701
4702 /* recover broken superblock */
4703 if (recovery) {
4704 err = f2fs_commit_super(sbi, true);
4705 f2fs_info(sbi, "Try to recover %dth superblock, ret: %d",
4706 sbi->valid_super_block ? 1 : 2, err);
4707 }
4708
4709 f2fs_join_shrinker(sbi);
4710
4711 f2fs_tuning_parameters(sbi);
4712
4713 f2fs_notice(sbi, "Mounted with checkpoint version = %llx",
4714 cur_cp_version(F2FS_CKPT(sbi)));
4715 f2fs_update_time(sbi, CP_TIME);
4716 f2fs_update_time(sbi, REQ_TIME);
4717 clear_sbi_flag(sbi, SBI_CP_DISABLED_QUICK);
4718 return 0;
4719
4720 sync_free_meta:
4721 /* safe to flush all the data */
4722 sync_filesystem(sbi->sb);
4723 retry_cnt = 0;
4724
4725 free_meta:
4726 #ifdef CONFIG_QUOTA
4727 f2fs_truncate_quota_inode_pages(sb);
4728 if (f2fs_sb_has_quota_ino(sbi) && !f2fs_readonly(sb))
4729 f2fs_quota_off_umount(sbi->sb);
4730 #endif
4731 /*
4732 * Some dirty meta pages can be produced by f2fs_recover_orphan_inodes()
4733 * failed by EIO. Then, iput(node_inode) can trigger balance_fs_bg()
4734 * followed by f2fs_write_checkpoint() through f2fs_write_node_pages(), which
4735 * falls into an infinite loop in f2fs_sync_meta_pages().
4736 */
4737 truncate_inode_pages_final(META_MAPPING(sbi));
4738 /* evict some inodes being cached by GC */
4739 evict_inodes(sb);
4740 f2fs_unregister_sysfs(sbi);
4741 free_compress_inode:
4742 f2fs_destroy_compress_inode(sbi);
4743 free_root_inode:
4744 dput(sb->s_root);
4745 sb->s_root = NULL;
4746 free_node_inode:
4747 f2fs_release_ino_entry(sbi, true);
4748 truncate_inode_pages_final(NODE_MAPPING(sbi));
4749 iput(sbi->node_inode);
4750 sbi->node_inode = NULL;
4751 free_stats:
4752 f2fs_destroy_stats(sbi);
4753 free_nm:
4754 /* stop discard thread before destroying node manager */
4755 f2fs_stop_discard_thread(sbi);
4756 f2fs_destroy_node_manager(sbi);
4757 free_sm:
4758 f2fs_destroy_segment_manager(sbi);
4759 stop_ckpt_thread:
4760 f2fs_stop_ckpt_thread(sbi);
4761 /* flush s_error_work before sbi destroy */
4762 flush_work(&sbi->s_error_work);
4763 f2fs_destroy_post_read_wq(sbi);
4764 free_devices:
4765 destroy_device_list(sbi);
4766 kvfree(sbi->ckpt);
4767 free_meta_inode:
4768 make_bad_inode(sbi->meta_inode);
4769 iput(sbi->meta_inode);
4770 sbi->meta_inode = NULL;
4771 free_page_array_cache:
4772 f2fs_destroy_page_array_cache(sbi);
4773 free_xattr_cache:
4774 f2fs_destroy_xattr_caches(sbi);
4775 free_percpu:
4776 destroy_percpu_info(sbi);
4777 free_iostat:
4778 f2fs_destroy_iostat(sbi);
4779 free_bio_info:
4780 for (i = 0; i < NR_PAGE_TYPE; i++)
4781 kvfree(sbi->write_io[i]);
4782
4783 #if IS_ENABLED(CONFIG_UNICODE)
4784 utf8_unload(sb->s_encoding);
4785 sb->s_encoding = NULL;
4786 #endif
4787 free_options:
4788 #ifdef CONFIG_QUOTA
4789 for (i = 0; i < MAXQUOTAS; i++)
4790 kfree(F2FS_OPTION(sbi).s_qf_names[i]);
4791 #endif
4792 fscrypt_free_dummy_policy(&F2FS_OPTION(sbi).dummy_enc_policy);
4793 kvfree(options);
4794 free_sb_buf:
4795 kfree(raw_super);
4796 free_sbi:
4797 if (sbi->s_chksum_driver)
4798 crypto_free_shash(sbi->s_chksum_driver);
4799 kfree(sbi);
4800
4801 /* give only one another chance */
4802 if (retry_cnt > 0 && skip_recovery) {
4803 retry_cnt--;
4804 shrink_dcache_sb(sb);
4805 goto try_onemore;
4806 }
4807 return err;
4808 }
4809
f2fs_mount(struct file_system_type * fs_type,int flags,const char * dev_name,void * data)4810 static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
4811 const char *dev_name, void *data)
4812 {
4813 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
4814 }
4815
kill_f2fs_super(struct super_block * sb)4816 static void kill_f2fs_super(struct super_block *sb)
4817 {
4818 if (sb->s_root) {
4819 struct f2fs_sb_info *sbi = F2FS_SB(sb);
4820
4821 set_sbi_flag(sbi, SBI_IS_CLOSE);
4822 f2fs_stop_gc_thread(sbi);
4823 f2fs_stop_discard_thread(sbi);
4824
4825 #ifdef CONFIG_F2FS_FS_COMPRESSION
4826 /*
4827 * latter evict_inode() can bypass checking and invalidating
4828 * compress inode cache.
4829 */
4830 if (test_opt(sbi, COMPRESS_CACHE))
4831 truncate_inode_pages_final(COMPRESS_MAPPING(sbi));
4832 #endif
4833
4834 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
4835 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
4836 struct cp_control cpc = {
4837 .reason = CP_UMOUNT,
4838 };
4839 stat_inc_cp_call_count(sbi, TOTAL_CALL);
4840 f2fs_write_checkpoint(sbi, &cpc);
4841 }
4842
4843 if (is_sbi_flag_set(sbi, SBI_IS_RECOVERED) && f2fs_readonly(sb))
4844 sb->s_flags &= ~SB_RDONLY;
4845 }
4846 kill_block_super(sb);
4847 }
4848
4849 static struct file_system_type f2fs_fs_type = {
4850 .owner = THIS_MODULE,
4851 .name = "f2fs",
4852 .mount = f2fs_mount,
4853 .kill_sb = kill_f2fs_super,
4854 .fs_flags = FS_REQUIRES_DEV | FS_ALLOW_IDMAP,
4855 };
4856 MODULE_ALIAS_FS("f2fs");
4857
init_inodecache(void)4858 static int __init init_inodecache(void)
4859 {
4860 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
4861 sizeof(struct f2fs_inode_info), 0,
4862 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
4863 return f2fs_inode_cachep ? 0 : -ENOMEM;
4864 }
4865
destroy_inodecache(void)4866 static void destroy_inodecache(void)
4867 {
4868 /*
4869 * Make sure all delayed rcu free inodes are flushed before we
4870 * destroy cache.
4871 */
4872 rcu_barrier();
4873 kmem_cache_destroy(f2fs_inode_cachep);
4874 }
4875
init_f2fs_fs(void)4876 static int __init init_f2fs_fs(void)
4877 {
4878 int err;
4879
4880 if (PAGE_SIZE != F2FS_BLKSIZE) {
4881 printk("F2FS not supported on PAGE_SIZE(%lu) != %d\n",
4882 PAGE_SIZE, F2FS_BLKSIZE);
4883 return -EINVAL;
4884 }
4885
4886 err = init_inodecache();
4887 if (err)
4888 goto fail;
4889 err = f2fs_create_node_manager_caches();
4890 if (err)
4891 goto free_inodecache;
4892 err = f2fs_create_segment_manager_caches();
4893 if (err)
4894 goto free_node_manager_caches;
4895 err = f2fs_create_checkpoint_caches();
4896 if (err)
4897 goto free_segment_manager_caches;
4898 err = f2fs_create_recovery_cache();
4899 if (err)
4900 goto free_checkpoint_caches;
4901 err = f2fs_create_extent_cache();
4902 if (err)
4903 goto free_recovery_cache;
4904 err = f2fs_create_garbage_collection_cache();
4905 if (err)
4906 goto free_extent_cache;
4907 err = f2fs_init_sysfs();
4908 if (err)
4909 goto free_garbage_collection_cache;
4910 err = register_shrinker(&f2fs_shrinker_info, "f2fs-shrinker");
4911 if (err)
4912 goto free_sysfs;
4913 err = register_filesystem(&f2fs_fs_type);
4914 if (err)
4915 goto free_shrinker;
4916 f2fs_create_root_stats();
4917 err = f2fs_init_post_read_processing();
4918 if (err)
4919 goto free_root_stats;
4920 err = f2fs_init_iostat_processing();
4921 if (err)
4922 goto free_post_read;
4923 err = f2fs_init_bio_entry_cache();
4924 if (err)
4925 goto free_iostat;
4926 err = f2fs_init_bioset();
4927 if (err)
4928 goto free_bio_entry_cache;
4929 err = f2fs_init_compress_mempool();
4930 if (err)
4931 goto free_bioset;
4932 err = f2fs_init_compress_cache();
4933 if (err)
4934 goto free_compress_mempool;
4935 err = f2fs_create_casefold_cache();
4936 if (err)
4937 goto free_compress_cache;
4938 return 0;
4939 free_compress_cache:
4940 f2fs_destroy_compress_cache();
4941 free_compress_mempool:
4942 f2fs_destroy_compress_mempool();
4943 free_bioset:
4944 f2fs_destroy_bioset();
4945 free_bio_entry_cache:
4946 f2fs_destroy_bio_entry_cache();
4947 free_iostat:
4948 f2fs_destroy_iostat_processing();
4949 free_post_read:
4950 f2fs_destroy_post_read_processing();
4951 free_root_stats:
4952 f2fs_destroy_root_stats();
4953 unregister_filesystem(&f2fs_fs_type);
4954 free_shrinker:
4955 unregister_shrinker(&f2fs_shrinker_info);
4956 free_sysfs:
4957 f2fs_exit_sysfs();
4958 free_garbage_collection_cache:
4959 f2fs_destroy_garbage_collection_cache();
4960 free_extent_cache:
4961 f2fs_destroy_extent_cache();
4962 free_recovery_cache:
4963 f2fs_destroy_recovery_cache();
4964 free_checkpoint_caches:
4965 f2fs_destroy_checkpoint_caches();
4966 free_segment_manager_caches:
4967 f2fs_destroy_segment_manager_caches();
4968 free_node_manager_caches:
4969 f2fs_destroy_node_manager_caches();
4970 free_inodecache:
4971 destroy_inodecache();
4972 fail:
4973 return err;
4974 }
4975
exit_f2fs_fs(void)4976 static void __exit exit_f2fs_fs(void)
4977 {
4978 f2fs_destroy_casefold_cache();
4979 f2fs_destroy_compress_cache();
4980 f2fs_destroy_compress_mempool();
4981 f2fs_destroy_bioset();
4982 f2fs_destroy_bio_entry_cache();
4983 f2fs_destroy_iostat_processing();
4984 f2fs_destroy_post_read_processing();
4985 f2fs_destroy_root_stats();
4986 unregister_filesystem(&f2fs_fs_type);
4987 unregister_shrinker(&f2fs_shrinker_info);
4988 f2fs_exit_sysfs();
4989 f2fs_destroy_garbage_collection_cache();
4990 f2fs_destroy_extent_cache();
4991 f2fs_destroy_recovery_cache();
4992 f2fs_destroy_checkpoint_caches();
4993 f2fs_destroy_segment_manager_caches();
4994 f2fs_destroy_node_manager_caches();
4995 destroy_inodecache();
4996 }
4997
4998 module_init(init_f2fs_fs)
4999 module_exit(exit_f2fs_fs)
5000
5001 MODULE_AUTHOR("Samsung Electronics's Praesto Team");
5002 MODULE_DESCRIPTION("Flash Friendly File System");
5003 MODULE_LICENSE("GPL");
5004 MODULE_SOFTDEP("pre: crc32");
5005
5006