]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blame - fs/f2fs/super.c
f2fs: add noextent_cache mount option
[mirror_ubuntu-jammy-kernel.git] / fs / f2fs / super.c
CommitLineData
0a8165d7 1/*
aff063e2
JK
2 * fs/f2fs/super.c
3 *
4 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
5 * http://www.samsung.com/
6 *
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License version 2 as
9 * published by the Free Software Foundation.
10 */
11#include <linux/module.h>
12#include <linux/init.h>
13#include <linux/fs.h>
14#include <linux/statfs.h>
aff063e2
JK
15#include <linux/buffer_head.h>
16#include <linux/backing-dev.h>
17#include <linux/kthread.h>
18#include <linux/parser.h>
19#include <linux/mount.h>
20#include <linux/seq_file.h>
5e176d54 21#include <linux/proc_fs.h>
aff063e2
JK
22#include <linux/random.h>
23#include <linux/exportfs.h>
d3ee456d 24#include <linux/blkdev.h>
aff063e2 25#include <linux/f2fs_fs.h>
b59d0bae 26#include <linux/sysfs.h>
aff063e2
JK
27
28#include "f2fs.h"
29#include "node.h"
5ec4e49f 30#include "segment.h"
aff063e2 31#include "xattr.h"
b59d0bae 32#include "gc.h"
db9f7c1a 33#include "trace.h"
aff063e2 34
a2a4a7e4
NJ
35#define CREATE_TRACE_POINTS
36#include <trace/events/f2fs.h>
37
5e176d54 38static struct proc_dir_entry *f2fs_proc_root;
aff063e2 39static struct kmem_cache *f2fs_inode_cachep;
b59d0bae 40static struct kset *f2fs_kset;
aff063e2 41
2658e50d
JK
42/* f2fs-wide shrinker description */
43static struct shrinker f2fs_shrinker_info = {
44 .scan_objects = f2fs_shrink_scan,
45 .count_objects = f2fs_shrink_count,
46 .seeks = DEFAULT_SEEKS,
47};
48
aff063e2 49enum {
696c018c 50 Opt_gc_background,
aff063e2 51 Opt_disable_roll_forward,
2d834bf9 52 Opt_norecovery,
aff063e2
JK
53 Opt_discard,
54 Opt_noheap,
4058c511 55 Opt_user_xattr,
aff063e2 56 Opt_nouser_xattr,
4058c511 57 Opt_acl,
aff063e2
JK
58 Opt_noacl,
59 Opt_active_logs,
60 Opt_disable_ext_identify,
444c580f 61 Opt_inline_xattr,
8274de77 62 Opt_inline_data,
5efd3c6f 63 Opt_inline_dentry,
6b4afdd7 64 Opt_flush_merge,
0f7b2abd 65 Opt_nobarrier,
d5053a34 66 Opt_fastboot,
89672159 67 Opt_extent_cache,
7daaea25 68 Opt_noextent_cache,
75342797 69 Opt_noinline_data,
aff063e2
JK
70 Opt_err,
71};
72
73static match_table_t f2fs_tokens = {
696c018c 74 {Opt_gc_background, "background_gc=%s"},
aff063e2 75 {Opt_disable_roll_forward, "disable_roll_forward"},
2d834bf9 76 {Opt_norecovery, "norecovery"},
aff063e2
JK
77 {Opt_discard, "discard"},
78 {Opt_noheap, "no_heap"},
4058c511 79 {Opt_user_xattr, "user_xattr"},
aff063e2 80 {Opt_nouser_xattr, "nouser_xattr"},
4058c511 81 {Opt_acl, "acl"},
aff063e2
JK
82 {Opt_noacl, "noacl"},
83 {Opt_active_logs, "active_logs=%u"},
84 {Opt_disable_ext_identify, "disable_ext_identify"},
444c580f 85 {Opt_inline_xattr, "inline_xattr"},
8274de77 86 {Opt_inline_data, "inline_data"},
5efd3c6f 87 {Opt_inline_dentry, "inline_dentry"},
6b4afdd7 88 {Opt_flush_merge, "flush_merge"},
0f7b2abd 89 {Opt_nobarrier, "nobarrier"},
d5053a34 90 {Opt_fastboot, "fastboot"},
89672159 91 {Opt_extent_cache, "extent_cache"},
7daaea25 92 {Opt_noextent_cache, "noextent_cache"},
75342797 93 {Opt_noinline_data, "noinline_data"},
aff063e2
JK
94 {Opt_err, NULL},
95};
96
b59d0bae 97/* Sysfs support for f2fs */
ea91e9b0
JK
98enum {
99 GC_THREAD, /* struct f2fs_gc_thread */
100 SM_INFO, /* struct f2fs_sm_info */
cdfc41c1 101 NM_INFO, /* struct f2fs_nm_info */
b1c57c1c 102 F2FS_SBI, /* struct f2fs_sb_info */
ea91e9b0
JK
103};
104
b59d0bae
NJ
105struct f2fs_attr {
106 struct attribute attr;
107 ssize_t (*show)(struct f2fs_attr *, struct f2fs_sb_info *, char *);
108 ssize_t (*store)(struct f2fs_attr *, struct f2fs_sb_info *,
109 const char *, size_t);
ea91e9b0 110 int struct_type;
b59d0bae
NJ
111 int offset;
112};
113
ea91e9b0
JK
114static unsigned char *__struct_ptr(struct f2fs_sb_info *sbi, int struct_type)
115{
116 if (struct_type == GC_THREAD)
117 return (unsigned char *)sbi->gc_thread;
118 else if (struct_type == SM_INFO)
119 return (unsigned char *)SM_I(sbi);
cdfc41c1
JK
120 else if (struct_type == NM_INFO)
121 return (unsigned char *)NM_I(sbi);
b1c57c1c
JK
122 else if (struct_type == F2FS_SBI)
123 return (unsigned char *)sbi;
ea91e9b0
JK
124 return NULL;
125}
126
b59d0bae
NJ
127static ssize_t f2fs_sbi_show(struct f2fs_attr *a,
128 struct f2fs_sb_info *sbi, char *buf)
129{
ea91e9b0 130 unsigned char *ptr = NULL;
b59d0bae
NJ
131 unsigned int *ui;
132
ea91e9b0
JK
133 ptr = __struct_ptr(sbi, a->struct_type);
134 if (!ptr)
b59d0bae
NJ
135 return -EINVAL;
136
ea91e9b0 137 ui = (unsigned int *)(ptr + a->offset);
b59d0bae
NJ
138
139 return snprintf(buf, PAGE_SIZE, "%u\n", *ui);
140}
141
142static ssize_t f2fs_sbi_store(struct f2fs_attr *a,
143 struct f2fs_sb_info *sbi,
144 const char *buf, size_t count)
145{
ea91e9b0 146 unsigned char *ptr;
b59d0bae
NJ
147 unsigned long t;
148 unsigned int *ui;
149 ssize_t ret;
150
ea91e9b0
JK
151 ptr = __struct_ptr(sbi, a->struct_type);
152 if (!ptr)
b59d0bae
NJ
153 return -EINVAL;
154
ea91e9b0 155 ui = (unsigned int *)(ptr + a->offset);
b59d0bae
NJ
156
157 ret = kstrtoul(skip_spaces(buf), 0, &t);
158 if (ret < 0)
159 return ret;
160 *ui = t;
161 return count;
162}
163
164static ssize_t f2fs_attr_show(struct kobject *kobj,
165 struct attribute *attr, char *buf)
166{
167 struct f2fs_sb_info *sbi = container_of(kobj, struct f2fs_sb_info,
168 s_kobj);
169 struct f2fs_attr *a = container_of(attr, struct f2fs_attr, attr);
170
171 return a->show ? a->show(a, sbi, buf) : 0;
172}
173
174static ssize_t f2fs_attr_store(struct kobject *kobj, struct attribute *attr,
175 const char *buf, size_t len)
176{
177 struct f2fs_sb_info *sbi = container_of(kobj, struct f2fs_sb_info,
178 s_kobj);
179 struct f2fs_attr *a = container_of(attr, struct f2fs_attr, attr);
180
181 return a->store ? a->store(a, sbi, buf, len) : 0;
182}
183
184static void f2fs_sb_release(struct kobject *kobj)
185{
186 struct f2fs_sb_info *sbi = container_of(kobj, struct f2fs_sb_info,
187 s_kobj);
188 complete(&sbi->s_kobj_unregister);
189}
190
ea91e9b0 191#define F2FS_ATTR_OFFSET(_struct_type, _name, _mode, _show, _store, _offset) \
b59d0bae
NJ
192static struct f2fs_attr f2fs_attr_##_name = { \
193 .attr = {.name = __stringify(_name), .mode = _mode }, \
194 .show = _show, \
195 .store = _store, \
ea91e9b0
JK
196 .struct_type = _struct_type, \
197 .offset = _offset \
b59d0bae
NJ
198}
199
ea91e9b0
JK
200#define F2FS_RW_ATTR(struct_type, struct_name, name, elname) \
201 F2FS_ATTR_OFFSET(struct_type, name, 0644, \
202 f2fs_sbi_show, f2fs_sbi_store, \
203 offsetof(struct struct_name, elname))
b59d0bae 204
ea91e9b0
JK
205F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_min_sleep_time, min_sleep_time);
206F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_max_sleep_time, max_sleep_time);
207F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_no_gc_sleep_time, no_gc_sleep_time);
208F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_idle, gc_idle);
209F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, reclaim_segments, rec_prefree_segments);
7ac8c3b0 210F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, max_small_discards, max_discards);
bba681cb 211F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, batched_trim_sections, trim_sections);
216fbd64
JK
212F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, ipu_policy, ipu_policy);
213F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, min_ipu_util, min_ipu_util);
c1ce1b02 214F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, min_fsync_blocks, min_fsync_blocks);
cdfc41c1 215F2FS_RW_ATTR(NM_INFO, f2fs_nm_info, ram_thresh, ram_thresh);
b1c57c1c 216F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, max_victim_search, max_victim_search);
ab9fa662 217F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, dir_level, dir_level);
b59d0bae
NJ
218
219#define ATTR_LIST(name) (&f2fs_attr_##name.attr)
220static struct attribute *f2fs_attrs[] = {
221 ATTR_LIST(gc_min_sleep_time),
222 ATTR_LIST(gc_max_sleep_time),
223 ATTR_LIST(gc_no_gc_sleep_time),
d2dc095f 224 ATTR_LIST(gc_idle),
ea91e9b0 225 ATTR_LIST(reclaim_segments),
7ac8c3b0 226 ATTR_LIST(max_small_discards),
bba681cb 227 ATTR_LIST(batched_trim_sections),
216fbd64
JK
228 ATTR_LIST(ipu_policy),
229 ATTR_LIST(min_ipu_util),
c1ce1b02 230 ATTR_LIST(min_fsync_blocks),
b1c57c1c 231 ATTR_LIST(max_victim_search),
ab9fa662 232 ATTR_LIST(dir_level),
cdfc41c1 233 ATTR_LIST(ram_thresh),
b59d0bae
NJ
234 NULL,
235};
236
237static const struct sysfs_ops f2fs_attr_ops = {
238 .show = f2fs_attr_show,
239 .store = f2fs_attr_store,
240};
241
242static struct kobj_type f2fs_ktype = {
243 .default_attrs = f2fs_attrs,
244 .sysfs_ops = &f2fs_attr_ops,
245 .release = f2fs_sb_release,
246};
247
a07ef784
NJ
248void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...)
249{
250 struct va_format vaf;
251 va_list args;
252
253 va_start(args, fmt);
254 vaf.fmt = fmt;
255 vaf.va = &args;
256 printk("%sF2FS-fs (%s): %pV\n", level, sb->s_id, &vaf);
257 va_end(args);
258}
259
aff063e2
JK
260static void init_once(void *foo)
261{
262 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
263
aff063e2
JK
264 inode_init_once(&fi->vfs_inode);
265}
266
696c018c
NJ
267static int parse_options(struct super_block *sb, char *options)
268{
269 struct f2fs_sb_info *sbi = F2FS_SB(sb);
09d54cdd 270 struct request_queue *q;
696c018c
NJ
271 substring_t args[MAX_OPT_ARGS];
272 char *p, *name;
273 int arg = 0;
274
275 if (!options)
276 return 0;
277
278 while ((p = strsep(&options, ",")) != NULL) {
279 int token;
280 if (!*p)
281 continue;
282 /*
283 * Initialize args struct so we know whether arg was
284 * found; some options take optional arguments.
285 */
286 args[0].to = args[0].from = NULL;
287 token = match_token(p, f2fs_tokens, args);
288
289 switch (token) {
290 case Opt_gc_background:
291 name = match_strdup(&args[0]);
292
293 if (!name)
294 return -ENOMEM;
04c09388 295 if (strlen(name) == 2 && !strncmp(name, "on", 2))
696c018c 296 set_opt(sbi, BG_GC);
04c09388 297 else if (strlen(name) == 3 && !strncmp(name, "off", 3))
696c018c
NJ
298 clear_opt(sbi, BG_GC);
299 else {
300 kfree(name);
301 return -EINVAL;
302 }
303 kfree(name);
304 break;
305 case Opt_disable_roll_forward:
306 set_opt(sbi, DISABLE_ROLL_FORWARD);
307 break;
2d834bf9
JK
308 case Opt_norecovery:
309 /* this option mounts f2fs with ro */
310 set_opt(sbi, DISABLE_ROLL_FORWARD);
311 if (!f2fs_readonly(sb))
312 return -EINVAL;
313 break;
696c018c 314 case Opt_discard:
09d54cdd
CY
315 q = bdev_get_queue(sb->s_bdev);
316 if (blk_queue_discard(q)) {
317 set_opt(sbi, DISCARD);
318 } else {
319 f2fs_msg(sb, KERN_WARNING,
320 "mounting with \"discard\" option, but "
321 "the device does not support discard");
322 }
696c018c
NJ
323 break;
324 case Opt_noheap:
325 set_opt(sbi, NOHEAP);
326 break;
327#ifdef CONFIG_F2FS_FS_XATTR
4058c511
KA
328 case Opt_user_xattr:
329 set_opt(sbi, XATTR_USER);
330 break;
696c018c
NJ
331 case Opt_nouser_xattr:
332 clear_opt(sbi, XATTR_USER);
333 break;
444c580f
JK
334 case Opt_inline_xattr:
335 set_opt(sbi, INLINE_XATTR);
336 break;
696c018c 337#else
4058c511
KA
338 case Opt_user_xattr:
339 f2fs_msg(sb, KERN_INFO,
340 "user_xattr options not supported");
341 break;
696c018c
NJ
342 case Opt_nouser_xattr:
343 f2fs_msg(sb, KERN_INFO,
344 "nouser_xattr options not supported");
345 break;
444c580f
JK
346 case Opt_inline_xattr:
347 f2fs_msg(sb, KERN_INFO,
348 "inline_xattr options not supported");
349 break;
696c018c
NJ
350#endif
351#ifdef CONFIG_F2FS_FS_POSIX_ACL
4058c511
KA
352 case Opt_acl:
353 set_opt(sbi, POSIX_ACL);
354 break;
696c018c
NJ
355 case Opt_noacl:
356 clear_opt(sbi, POSIX_ACL);
357 break;
358#else
4058c511
KA
359 case Opt_acl:
360 f2fs_msg(sb, KERN_INFO, "acl options not supported");
361 break;
696c018c
NJ
362 case Opt_noacl:
363 f2fs_msg(sb, KERN_INFO, "noacl options not supported");
364 break;
365#endif
366 case Opt_active_logs:
367 if (args->from && match_int(args, &arg))
368 return -EINVAL;
369 if (arg != 2 && arg != 4 && arg != NR_CURSEG_TYPE)
370 return -EINVAL;
371 sbi->active_logs = arg;
372 break;
373 case Opt_disable_ext_identify:
374 set_opt(sbi, DISABLE_EXT_IDENTIFY);
375 break;
8274de77
HL
376 case Opt_inline_data:
377 set_opt(sbi, INLINE_DATA);
378 break;
5efd3c6f
CY
379 case Opt_inline_dentry:
380 set_opt(sbi, INLINE_DENTRY);
381 break;
6b4afdd7
JK
382 case Opt_flush_merge:
383 set_opt(sbi, FLUSH_MERGE);
384 break;
0f7b2abd
JK
385 case Opt_nobarrier:
386 set_opt(sbi, NOBARRIER);
387 break;
d5053a34
JK
388 case Opt_fastboot:
389 set_opt(sbi, FASTBOOT);
390 break;
89672159
CY
391 case Opt_extent_cache:
392 set_opt(sbi, EXTENT_CACHE);
393 break;
7daaea25
JK
394 case Opt_noextent_cache:
395 clear_opt(sbi, EXTENT_CACHE);
396 break;
75342797
WL
397 case Opt_noinline_data:
398 clear_opt(sbi, INLINE_DATA);
399 break;
696c018c
NJ
400 default:
401 f2fs_msg(sb, KERN_ERR,
402 "Unrecognized mount option \"%s\" or missing value",
403 p);
404 return -EINVAL;
405 }
406 }
407 return 0;
408}
409
aff063e2
JK
410static struct inode *f2fs_alloc_inode(struct super_block *sb)
411{
412 struct f2fs_inode_info *fi;
413
a0acdfe0 414 fi = kmem_cache_alloc(f2fs_inode_cachep, GFP_F2FS_ZERO);
aff063e2
JK
415 if (!fi)
416 return NULL;
417
418 init_once((void *) fi);
419
434720fa 420 /* Initialize f2fs-specific inode info */
aff063e2 421 fi->vfs_inode.i_version = 1;
a7ffdbe2 422 atomic_set(&fi->dirty_pages, 0);
aff063e2
JK
423 fi->i_current_depth = 1;
424 fi->i_advise = 0;
0c872e2d 425 rwlock_init(&fi->ext_lock);
d928bfbf 426 init_rwsem(&fi->i_sem);
34ba94ba 427 INIT_RADIX_TREE(&fi->inmem_root, GFP_NOFS);
88b88a66
JK
428 INIT_LIST_HEAD(&fi->inmem_pages);
429 mutex_init(&fi->inmem_lock);
aff063e2
JK
430
431 set_inode_flag(fi, FI_NEW_INODE);
432
444c580f
JK
433 if (test_opt(F2FS_SB(sb), INLINE_XATTR))
434 set_inode_flag(fi, FI_INLINE_XATTR);
435
ab9fa662
JK
436 /* Will be used by directory only */
437 fi->i_dir_level = F2FS_SB(sb)->dir_level;
438
57e5055b
JK
439#ifdef CONFIG_F2FS_FS_ENCRYPTION
440 fi->i_crypt_info = NULL;
441#endif
aff063e2
JK
442 return &fi->vfs_inode;
443}
444
531ad7d5
JK
445static int f2fs_drop_inode(struct inode *inode)
446{
447 /*
448 * This is to avoid a deadlock condition like below.
449 * writeback_single_inode(inode)
450 * - f2fs_write_data_page
451 * - f2fs_gc -> iput -> evict
452 * - inode_wait_for_writeback(inode)
453 */
06e1bc05
JK
454 if (!inode_unhashed(inode) && inode->i_state & I_SYNC) {
455 if (!inode->i_nlink && !is_bad_inode(inode)) {
456 spin_unlock(&inode->i_lock);
457
458 /* some remained atomic pages should discarded */
459 if (f2fs_is_atomic_file(inode))
460 commit_inmem_pages(inode, true);
461
462 sb_start_intwrite(inode->i_sb);
463 i_size_write(inode, 0);
464
465 if (F2FS_HAS_BLOCKS(inode))
466 f2fs_truncate(inode);
467
468 sb_end_intwrite(inode->i_sb);
469
470#ifdef CONFIG_F2FS_FS_ENCRYPTION
471 if (F2FS_I(inode)->i_crypt_info)
26bf3dc7
JK
472 f2fs_free_encryption_info(inode,
473 F2FS_I(inode)->i_crypt_info);
06e1bc05
JK
474#endif
475 spin_lock(&inode->i_lock);
476 }
531ad7d5 477 return 0;
06e1bc05 478 }
531ad7d5
JK
479 return generic_drop_inode(inode);
480}
481
b3783873
JK
482/*
483 * f2fs_dirty_inode() is called from __mark_inode_dirty()
484 *
485 * We should call set_dirty_inode to write the dirty inode through write_inode.
486 */
487static void f2fs_dirty_inode(struct inode *inode, int flags)
488{
489 set_inode_flag(F2FS_I(inode), FI_DIRTY_INODE);
b3783873
JK
490}
491
aff063e2
JK
492static void f2fs_i_callback(struct rcu_head *head)
493{
494 struct inode *inode = container_of(head, struct inode, i_rcu);
495 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
496}
497
25ca923b 498static void f2fs_destroy_inode(struct inode *inode)
aff063e2
JK
499{
500 call_rcu(&inode->i_rcu, f2fs_i_callback);
501}
502
503static void f2fs_put_super(struct super_block *sb)
504{
505 struct f2fs_sb_info *sbi = F2FS_SB(sb);
506
5e176d54
JK
507 if (sbi->s_proc) {
508 remove_proc_entry("segment_info", sbi->s_proc);
509 remove_proc_entry(sb->s_id, f2fs_proc_root);
510 }
b59d0bae 511 kobject_del(&sbi->s_kobj);
5e176d54 512
aff063e2
JK
513 stop_gc_thread(sbi);
514
2658e50d
JK
515 /* prevent remaining shrinker jobs */
516 mutex_lock(&sbi->umount_mutex);
517
85dc2f2c
JK
518 /*
519 * We don't need to do checkpoint when superblock is clean.
520 * But, the previous checkpoint was not done by umount, it needs to do
521 * clean checkpoint again.
522 */
caf0047e 523 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
85dc2f2c 524 !is_set_ckpt_flags(F2FS_CKPT(sbi), CP_UMOUNT_FLAG)) {
75ab4cb8
JK
525 struct cp_control cpc = {
526 .reason = CP_UMOUNT,
527 };
528 write_checkpoint(sbi, &cpc);
529 }
aff063e2 530
eca616f8
JK
531 /* write_checkpoint can update stat informaion */
532 f2fs_destroy_stats(sbi);
533
cf779cab
JK
534 /*
535 * normally superblock is clean, so we need to release this.
536 * In addition, EIO will skip do checkpoint, we need this as well.
537 */
6f12ac25 538 release_dirty_inode(sbi);
4b2fecc8 539 release_discard_addrs(sbi);
6f12ac25 540
2658e50d
JK
541 f2fs_leave_shrinker(sbi);
542 mutex_unlock(&sbi->umount_mutex);
543
aff063e2
JK
544 iput(sbi->node_inode);
545 iput(sbi->meta_inode);
546
547 /* destroy f2fs internal modules */
548 destroy_node_manager(sbi);
549 destroy_segment_manager(sbi);
550
551 kfree(sbi->ckpt);
b59d0bae
NJ
552 kobject_put(&sbi->s_kobj);
553 wait_for_completion(&sbi->s_kobj_unregister);
aff063e2
JK
554
555 sb->s_fs_info = NULL;
556 brelse(sbi->raw_super_buf);
557 kfree(sbi);
558}
559
560int f2fs_sync_fs(struct super_block *sb, int sync)
561{
562 struct f2fs_sb_info *sbi = F2FS_SB(sb);
aff063e2 563
a2a4a7e4
NJ
564 trace_f2fs_sync_fs(sb, sync);
565
b7473754 566 if (sync) {
d5053a34
JK
567 struct cp_control cpc;
568
119ee914
JK
569 cpc.reason = __get_cp_reason(sbi);
570
b7473754 571 mutex_lock(&sbi->gc_mutex);
75ab4cb8 572 write_checkpoint(sbi, &cpc);
b7473754
JK
573 mutex_unlock(&sbi->gc_mutex);
574 } else {
7d82db83 575 f2fs_balance_fs(sbi);
b7473754 576 }
05ca3632 577 f2fs_trace_ios(NULL, 1);
aff063e2 578
64c576fe 579 return 0;
aff063e2
JK
580}
581
d6212a5f
CL
582static int f2fs_freeze(struct super_block *sb)
583{
584 int err;
585
77888c1e 586 if (f2fs_readonly(sb))
d6212a5f
CL
587 return 0;
588
589 err = f2fs_sync_fs(sb, 1);
590 return err;
591}
592
593static int f2fs_unfreeze(struct super_block *sb)
594{
595 return 0;
596}
597
aff063e2
JK
598static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
599{
600 struct super_block *sb = dentry->d_sb;
601 struct f2fs_sb_info *sbi = F2FS_SB(sb);
602 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
603 block_t total_count, user_block_count, start_count, ovp_count;
604
605 total_count = le64_to_cpu(sbi->raw_super->block_count);
606 user_block_count = sbi->user_block_count;
607 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
608 ovp_count = SM_I(sbi)->ovp_segments << sbi->log_blocks_per_seg;
609 buf->f_type = F2FS_SUPER_MAGIC;
610 buf->f_bsize = sbi->blocksize;
611
612 buf->f_blocks = total_count - start_count;
613 buf->f_bfree = buf->f_blocks - valid_user_blocks(sbi) - ovp_count;
614 buf->f_bavail = user_block_count - valid_user_blocks(sbi);
615
c200b1aa
CY
616 buf->f_files = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
617 buf->f_ffree = buf->f_files - valid_inode_count(sbi);
aff063e2 618
5a20d339 619 buf->f_namelen = F2FS_NAME_LEN;
aff063e2
JK
620 buf->f_fsid.val[0] = (u32)id;
621 buf->f_fsid.val[1] = (u32)(id >> 32);
622
623 return 0;
624}
625
626static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
627{
628 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
629
b270ad6f 630 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, BG_GC))
696c018c 631 seq_printf(seq, ",background_gc=%s", "on");
aff063e2 632 else
696c018c 633 seq_printf(seq, ",background_gc=%s", "off");
aff063e2
JK
634 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
635 seq_puts(seq, ",disable_roll_forward");
636 if (test_opt(sbi, DISCARD))
637 seq_puts(seq, ",discard");
638 if (test_opt(sbi, NOHEAP))
639 seq_puts(seq, ",no_heap_alloc");
640#ifdef CONFIG_F2FS_FS_XATTR
641 if (test_opt(sbi, XATTR_USER))
642 seq_puts(seq, ",user_xattr");
643 else
644 seq_puts(seq, ",nouser_xattr");
444c580f
JK
645 if (test_opt(sbi, INLINE_XATTR))
646 seq_puts(seq, ",inline_xattr");
aff063e2
JK
647#endif
648#ifdef CONFIG_F2FS_FS_POSIX_ACL
649 if (test_opt(sbi, POSIX_ACL))
650 seq_puts(seq, ",acl");
651 else
652 seq_puts(seq, ",noacl");
653#endif
654 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
aa43507f 655 seq_puts(seq, ",disable_ext_identify");
8274de77
HL
656 if (test_opt(sbi, INLINE_DATA))
657 seq_puts(seq, ",inline_data");
75342797
WL
658 else
659 seq_puts(seq, ",noinline_data");
5efd3c6f
CY
660 if (test_opt(sbi, INLINE_DENTRY))
661 seq_puts(seq, ",inline_dentry");
b270ad6f 662 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
6b4afdd7 663 seq_puts(seq, ",flush_merge");
0f7b2abd
JK
664 if (test_opt(sbi, NOBARRIER))
665 seq_puts(seq, ",nobarrier");
d5053a34
JK
666 if (test_opt(sbi, FASTBOOT))
667 seq_puts(seq, ",fastboot");
89672159
CY
668 if (test_opt(sbi, EXTENT_CACHE))
669 seq_puts(seq, ",extent_cache");
7daaea25
JK
670 else
671 seq_puts(seq, ",noextent_cache");
aff063e2
JK
672 seq_printf(seq, ",active_logs=%u", sbi->active_logs);
673
674 return 0;
675}
676
5e176d54
JK
677static int segment_info_seq_show(struct seq_file *seq, void *offset)
678{
679 struct super_block *sb = seq->private;
680 struct f2fs_sb_info *sbi = F2FS_SB(sb);
6c311ec6
CF
681 unsigned int total_segs =
682 le32_to_cpu(sbi->raw_super->segment_count_main);
5e176d54
JK
683 int i;
684
90aa6dc9
CY
685 seq_puts(seq, "format: segment_type|valid_blocks\n"
686 "segment_type(0:HD, 1:WD, 2:CD, 3:HN, 4:WN, 5:CN)\n");
687
5e176d54 688 for (i = 0; i < total_segs; i++) {
90aa6dc9
CY
689 struct seg_entry *se = get_seg_entry(sbi, i);
690
691 if ((i % 10) == 0)
692 seq_printf(seq, "%-5d", i);
693 seq_printf(seq, "%d|%-3u", se->type,
694 get_valid_blocks(sbi, i, 1));
46c04366
GZ
695 if ((i % 10) == 9 || i == (total_segs - 1))
696 seq_putc(seq, '\n');
5e176d54 697 else
46c04366 698 seq_putc(seq, ' ');
5e176d54 699 }
46c04366 700
5e176d54
JK
701 return 0;
702}
703
704static int segment_info_open_fs(struct inode *inode, struct file *file)
705{
706 return single_open(file, segment_info_seq_show, PDE_DATA(inode));
707}
708
709static const struct file_operations f2fs_seq_segment_info_fops = {
710 .owner = THIS_MODULE,
711 .open = segment_info_open_fs,
712 .read = seq_read,
713 .llseek = seq_lseek,
714 .release = single_release,
715};
716
498c5e9f
YH
717static void default_options(struct f2fs_sb_info *sbi)
718{
719 /* init some FS parameters */
720 sbi->active_logs = NR_CURSEG_TYPE;
721
722 set_opt(sbi, BG_GC);
723 set_opt(sbi, INLINE_DATA);
724
725#ifdef CONFIG_F2FS_FS_XATTR
726 set_opt(sbi, XATTR_USER);
727#endif
728#ifdef CONFIG_F2FS_FS_POSIX_ACL
729 set_opt(sbi, POSIX_ACL);
730#endif
731}
732
696c018c
NJ
733static int f2fs_remount(struct super_block *sb, int *flags, char *data)
734{
735 struct f2fs_sb_info *sbi = F2FS_SB(sb);
736 struct f2fs_mount_info org_mount_opt;
737 int err, active_logs;
876dc59e
GZ
738 bool need_restart_gc = false;
739 bool need_stop_gc = false;
696c018c 740
02b9984d
TT
741 sync_filesystem(sb);
742
696c018c
NJ
743 /*
744 * Save the old mount options in case we
745 * need to restore them.
746 */
747 org_mount_opt = sbi->mount_opt;
748 active_logs = sbi->active_logs;
749
26666c8a 750 sbi->mount_opt.opt = 0;
498c5e9f 751 default_options(sbi);
26666c8a 752
696c018c
NJ
753 /* parse mount options */
754 err = parse_options(sb, data);
755 if (err)
756 goto restore_opts;
757
758 /*
759 * Previous and new state of filesystem is RO,
876dc59e 760 * so skip checking GC and FLUSH_MERGE conditions.
696c018c 761 */
6b2920a5 762 if (f2fs_readonly(sb) && (*flags & MS_RDONLY))
696c018c
NJ
763 goto skip;
764
765 /*
766 * We stop the GC thread if FS is mounted as RO
767 * or if background_gc = off is passed in mount
768 * option. Also sync the filesystem.
769 */
770 if ((*flags & MS_RDONLY) || !test_opt(sbi, BG_GC)) {
771 if (sbi->gc_thread) {
772 stop_gc_thread(sbi);
773 f2fs_sync_fs(sb, 1);
876dc59e 774 need_restart_gc = true;
696c018c 775 }
aba291b3 776 } else if (!sbi->gc_thread) {
696c018c
NJ
777 err = start_gc_thread(sbi);
778 if (err)
779 goto restore_opts;
876dc59e
GZ
780 need_stop_gc = true;
781 }
782
783 /*
784 * We stop issue flush thread if FS is mounted as RO
785 * or if flush_merge is not passed in mount option.
786 */
787 if ((*flags & MS_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2163d198 788 destroy_flush_cmd_control(sbi);
aba291b3 789 } else if (!SM_I(sbi)->cmd_control_info) {
2163d198
GZ
790 err = create_flush_cmd_control(sbi);
791 if (err)
a688b9d9 792 goto restore_gc;
696c018c
NJ
793 }
794skip:
795 /* Update the POSIXACL Flag */
796 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
797 (test_opt(sbi, POSIX_ACL) ? MS_POSIXACL : 0);
798 return 0;
876dc59e
GZ
799restore_gc:
800 if (need_restart_gc) {
801 if (start_gc_thread(sbi))
802 f2fs_msg(sbi->sb, KERN_WARNING,
e1c42045 803 "background gc thread has stopped");
876dc59e
GZ
804 } else if (need_stop_gc) {
805 stop_gc_thread(sbi);
806 }
696c018c
NJ
807restore_opts:
808 sbi->mount_opt = org_mount_opt;
809 sbi->active_logs = active_logs;
810 return err;
811}
812
aff063e2
JK
813static struct super_operations f2fs_sops = {
814 .alloc_inode = f2fs_alloc_inode,
531ad7d5 815 .drop_inode = f2fs_drop_inode,
aff063e2
JK
816 .destroy_inode = f2fs_destroy_inode,
817 .write_inode = f2fs_write_inode,
b3783873 818 .dirty_inode = f2fs_dirty_inode,
aff063e2
JK
819 .show_options = f2fs_show_options,
820 .evict_inode = f2fs_evict_inode,
821 .put_super = f2fs_put_super,
822 .sync_fs = f2fs_sync_fs,
d6212a5f
CL
823 .freeze_fs = f2fs_freeze,
824 .unfreeze_fs = f2fs_unfreeze,
aff063e2 825 .statfs = f2fs_statfs,
696c018c 826 .remount_fs = f2fs_remount,
aff063e2
JK
827};
828
829static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
830 u64 ino, u32 generation)
831{
832 struct f2fs_sb_info *sbi = F2FS_SB(sb);
833 struct inode *inode;
834
d6b7d4b3 835 if (check_nid_range(sbi, ino))
910bb12d 836 return ERR_PTR(-ESTALE);
aff063e2
JK
837
838 /*
839 * f2fs_iget isn't quite right if the inode is currently unallocated!
840 * However f2fs_iget currently does appropriate checks to handle stale
841 * inodes so everything is OK.
842 */
843 inode = f2fs_iget(sb, ino);
844 if (IS_ERR(inode))
845 return ERR_CAST(inode);
6bacf52f 846 if (unlikely(generation && inode->i_generation != generation)) {
aff063e2
JK
847 /* we didn't find the right inode.. */
848 iput(inode);
849 return ERR_PTR(-ESTALE);
850 }
851 return inode;
852}
853
854static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
855 int fh_len, int fh_type)
856{
857 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
858 f2fs_nfs_get_inode);
859}
860
861static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
862 int fh_len, int fh_type)
863{
864 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
865 f2fs_nfs_get_inode);
866}
867
868static const struct export_operations f2fs_export_ops = {
869 .fh_to_dentry = f2fs_fh_to_dentry,
870 .fh_to_parent = f2fs_fh_to_parent,
871 .get_parent = f2fs_get_parent,
872};
873
aff063e2
JK
874static loff_t max_file_size(unsigned bits)
875{
de93653f 876 loff_t result = (DEF_ADDRS_PER_INODE - F2FS_INLINE_XATTR_ADDRS);
aff063e2
JK
877 loff_t leaf_count = ADDRS_PER_BLOCK;
878
879 /* two direct node blocks */
880 result += (leaf_count * 2);
881
882 /* two indirect node blocks */
883 leaf_count *= NIDS_PER_BLOCK;
884 result += (leaf_count * 2);
885
886 /* one double indirect node block */
887 leaf_count *= NIDS_PER_BLOCK;
888 result += leaf_count;
889
890 result <<= bits;
891 return result;
892}
893
a07ef784
NJ
894static int sanity_check_raw_super(struct super_block *sb,
895 struct f2fs_super_block *raw_super)
aff063e2
JK
896{
897 unsigned int blocksize;
898
a07ef784
NJ
899 if (F2FS_SUPER_MAGIC != le32_to_cpu(raw_super->magic)) {
900 f2fs_msg(sb, KERN_INFO,
901 "Magic Mismatch, valid(0x%x) - read(0x%x)",
902 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
aff063e2 903 return 1;
a07ef784 904 }
aff063e2 905
5c9b4692 906 /* Currently, support only 4KB page cache size */
907 if (F2FS_BLKSIZE != PAGE_CACHE_SIZE) {
908 f2fs_msg(sb, KERN_INFO,
14d7e9de 909 "Invalid page_cache_size (%lu), supports only 4KB\n",
5c9b4692 910 PAGE_CACHE_SIZE);
911 return 1;
912 }
913
aff063e2
JK
914 /* Currently, support only 4KB block size */
915 blocksize = 1 << le32_to_cpu(raw_super->log_blocksize);
5c9b4692 916 if (blocksize != F2FS_BLKSIZE) {
a07ef784
NJ
917 f2fs_msg(sb, KERN_INFO,
918 "Invalid blocksize (%u), supports only 4KB\n",
919 blocksize);
aff063e2 920 return 1;
a07ef784 921 }
5c9b4692 922
55cf9cb6
CY
923 /* Currently, support 512/1024/2048/4096 bytes sector size */
924 if (le32_to_cpu(raw_super->log_sectorsize) >
925 F2FS_MAX_LOG_SECTOR_SIZE ||
926 le32_to_cpu(raw_super->log_sectorsize) <
927 F2FS_MIN_LOG_SECTOR_SIZE) {
928 f2fs_msg(sb, KERN_INFO, "Invalid log sectorsize (%u)",
929 le32_to_cpu(raw_super->log_sectorsize));
aff063e2 930 return 1;
a07ef784 931 }
55cf9cb6
CY
932 if (le32_to_cpu(raw_super->log_sectors_per_block) +
933 le32_to_cpu(raw_super->log_sectorsize) !=
934 F2FS_MAX_LOG_SECTOR_SIZE) {
935 f2fs_msg(sb, KERN_INFO,
936 "Invalid log sectors per block(%u) log sectorsize(%u)",
937 le32_to_cpu(raw_super->log_sectors_per_block),
938 le32_to_cpu(raw_super->log_sectorsize));
aff063e2 939 return 1;
a07ef784 940 }
aff063e2
JK
941 return 0;
942}
943
577e3495 944static int sanity_check_ckpt(struct f2fs_sb_info *sbi)
aff063e2
JK
945{
946 unsigned int total, fsmeta;
577e3495
JK
947 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
948 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
aff063e2
JK
949
950 total = le32_to_cpu(raw_super->segment_count);
951 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
952 fsmeta += le32_to_cpu(raw_super->segment_count_sit);
953 fsmeta += le32_to_cpu(raw_super->segment_count_nat);
954 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
955 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
956
6bacf52f 957 if (unlikely(fsmeta >= total))
aff063e2 958 return 1;
577e3495 959
1e968fdf 960 if (unlikely(f2fs_cp_error(sbi))) {
577e3495
JK
961 f2fs_msg(sbi->sb, KERN_ERR, "A bug case: need to run fsck");
962 return 1;
963 }
aff063e2
JK
964 return 0;
965}
966
967static void init_sb_info(struct f2fs_sb_info *sbi)
968{
969 struct f2fs_super_block *raw_super = sbi->raw_super;
970 int i;
971
972 sbi->log_sectors_per_block =
973 le32_to_cpu(raw_super->log_sectors_per_block);
974 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
975 sbi->blocksize = 1 << sbi->log_blocksize;
976 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
977 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
978 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
979 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
980 sbi->total_sections = le32_to_cpu(raw_super->section_count);
981 sbi->total_node_count =
982 (le32_to_cpu(raw_super->segment_count_nat) / 2)
983 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
984 sbi->root_ino_num = le32_to_cpu(raw_super->root_ino);
985 sbi->node_ino_num = le32_to_cpu(raw_super->node_ino);
986 sbi->meta_ino_num = le32_to_cpu(raw_super->meta_ino);
5ec4e49f 987 sbi->cur_victim_sec = NULL_SECNO;
b1c57c1c 988 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
aff063e2
JK
989
990 for (i = 0; i < NR_COUNT_TYPE; i++)
991 atomic_set(&sbi->nr_pages[i], 0);
ab9fa662
JK
992
993 sbi->dir_level = DEF_DIR_LEVEL;
caf0047e 994 clear_sbi_flag(sbi, SBI_NEED_FSCK);
2658e50d
JK
995
996 INIT_LIST_HEAD(&sbi->s_list);
997 mutex_init(&sbi->umount_mutex);
aff063e2
JK
998}
999
9076a75f
GZ
1000/*
1001 * Read f2fs raw super block.
1002 * Because we have two copies of super block, so read the first one at first,
1003 * if the first one is invalid, move to read the second one.
1004 */
1005static int read_raw_super_block(struct super_block *sb,
1006 struct f2fs_super_block **raw_super,
da554e48 1007 struct buffer_head **raw_super_buf,
1008 int *recovery)
14d7e9de 1009{
9076a75f 1010 int block = 0;
da554e48 1011 struct buffer_head *buffer;
1012 struct f2fs_super_block *super;
1013 int err = 0;
14d7e9de 1014
9076a75f 1015retry:
da554e48 1016 buffer = sb_bread(sb, block);
1017 if (!buffer) {
1018 *recovery = 1;
9076a75f
GZ
1019 f2fs_msg(sb, KERN_ERR, "Unable to read %dth superblock",
1020 block + 1);
1021 if (block == 0) {
1022 block++;
1023 goto retry;
1024 } else {
da554e48 1025 err = -EIO;
1026 goto out;
9076a75f 1027 }
14d7e9de 1028 }
1029
da554e48 1030 super = (struct f2fs_super_block *)
1031 ((char *)(buffer)->b_data + F2FS_SUPER_OFFSET);
14d7e9de 1032
1033 /* sanity checking of raw super */
da554e48 1034 if (sanity_check_raw_super(sb, super)) {
1035 brelse(buffer);
1036 *recovery = 1;
6c311ec6
CF
1037 f2fs_msg(sb, KERN_ERR,
1038 "Can't find valid F2FS filesystem in %dth superblock",
1039 block + 1);
6bacf52f 1040 if (block == 0) {
9076a75f
GZ
1041 block++;
1042 goto retry;
1043 } else {
da554e48 1044 err = -EINVAL;
1045 goto out;
9076a75f
GZ
1046 }
1047 }
14d7e9de 1048
da554e48 1049 if (!*raw_super) {
1050 *raw_super_buf = buffer;
1051 *raw_super = super;
1052 } else {
1053 /* already have a valid superblock */
1054 brelse(buffer);
1055 }
1056
1057 /* check the validity of the second superblock */
1058 if (block == 0) {
1059 block++;
1060 goto retry;
1061 }
1062
1063out:
1064 /* No valid superblock */
1065 if (!*raw_super)
1066 return err;
1067
9076a75f 1068 return 0;
14d7e9de 1069}
1070
c5bda1c8 1071int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
26d815ad
JK
1072{
1073 struct buffer_head *sbh = sbi->raw_super_buf;
1074 sector_t block = sbh->b_blocknr;
1075 int err;
1076
1077 /* write back-up superblock first */
1078 sbh->b_blocknr = block ? 0 : 1;
1079 mark_buffer_dirty(sbh);
1080 err = sync_dirty_buffer(sbh);
1081
1082 sbh->b_blocknr = block;
c5bda1c8
CY
1083
1084 /* if we are in recovery path, skip writing valid superblock */
1085 if (recover || err)
26d815ad
JK
1086 goto out;
1087
1088 /* write current valid superblock */
1089 mark_buffer_dirty(sbh);
1090 err = sync_dirty_buffer(sbh);
1091out:
1092 clear_buffer_write_io_error(sbh);
1093 set_buffer_uptodate(sbh);
1094 return err;
1095}
1096
aff063e2
JK
1097static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
1098{
1099 struct f2fs_sb_info *sbi;
da554e48 1100 struct f2fs_super_block *raw_super;
aff063e2
JK
1101 struct buffer_head *raw_super_buf;
1102 struct inode *root;
da554e48 1103 long err;
2adc3505 1104 bool retry = true, need_fsck = false;
dabc4a5c 1105 char *options = NULL;
da554e48 1106 int recovery, i;
aff063e2 1107
ed2e621a 1108try_onemore:
da554e48 1109 err = -EINVAL;
1110 raw_super = NULL;
1111 raw_super_buf = NULL;
1112 recovery = 0;
1113
aff063e2
JK
1114 /* allocate memory for f2fs-specific super block info */
1115 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
1116 if (!sbi)
1117 return -ENOMEM;
1118
ff9234ad 1119 /* set a block size */
6bacf52f 1120 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
a07ef784 1121 f2fs_msg(sb, KERN_ERR, "unable to set blocksize");
aff063e2 1122 goto free_sbi;
a07ef784 1123 }
aff063e2 1124
da554e48 1125 err = read_raw_super_block(sb, &raw_super, &raw_super_buf, &recovery);
9076a75f
GZ
1126 if (err)
1127 goto free_sbi;
1128
5fb08372 1129 sb->s_fs_info = sbi;
498c5e9f 1130 default_options(sbi);
aff063e2 1131 /* parse mount options */
dabc4a5c
JK
1132 options = kstrdup((const char *)data, GFP_KERNEL);
1133 if (data && !options) {
1134 err = -ENOMEM;
aff063e2 1135 goto free_sb_buf;
dabc4a5c
JK
1136 }
1137
1138 err = parse_options(sb, options);
1139 if (err)
1140 goto free_options;
aff063e2 1141
25ca923b 1142 sb->s_maxbytes = max_file_size(le32_to_cpu(raw_super->log_blocksize));
aff063e2
JK
1143 sb->s_max_links = F2FS_LINK_MAX;
1144 get_random_bytes(&sbi->s_next_generation, sizeof(u32));
1145
1146 sb->s_op = &f2fs_sops;
1147 sb->s_xattr = f2fs_xattr_handlers;
1148 sb->s_export_op = &f2fs_export_ops;
1149 sb->s_magic = F2FS_SUPER_MAGIC;
aff063e2
JK
1150 sb->s_time_gran = 1;
1151 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
1152 (test_opt(sbi, POSIX_ACL) ? MS_POSIXACL : 0);
1153 memcpy(sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
1154
1155 /* init f2fs-specific super block info */
1156 sbi->sb = sb;
1157 sbi->raw_super = raw_super;
1158 sbi->raw_super_buf = raw_super_buf;
1159 mutex_init(&sbi->gc_mutex);
5463e7c1 1160 mutex_init(&sbi->writepages);
aff063e2 1161 mutex_init(&sbi->cp_mutex);
b3582c68 1162 init_rwsem(&sbi->node_write);
caf0047e 1163 clear_sbi_flag(sbi, SBI_POR_DOING);
aff063e2 1164 spin_lock_init(&sbi->stat_lock);
971767ca 1165
df0f8dc0 1166 init_rwsem(&sbi->read_io.io_rwsem);
458e6197
JK
1167 sbi->read_io.sbi = sbi;
1168 sbi->read_io.bio = NULL;
1169 for (i = 0; i < NR_PAGE_TYPE; i++) {
df0f8dc0 1170 init_rwsem(&sbi->write_io[i].io_rwsem);
458e6197
JK
1171 sbi->write_io[i].sbi = sbi;
1172 sbi->write_io[i].bio = NULL;
1173 }
971767ca 1174
e479556b 1175 init_rwsem(&sbi->cp_rwsem);
fb51b5ef 1176 init_waitqueue_head(&sbi->cp_wait);
aff063e2
JK
1177 init_sb_info(sbi);
1178
1179 /* get an inode for meta space */
1180 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
1181 if (IS_ERR(sbi->meta_inode)) {
a07ef784 1182 f2fs_msg(sb, KERN_ERR, "Failed to read F2FS meta data inode");
aff063e2 1183 err = PTR_ERR(sbi->meta_inode);
dabc4a5c 1184 goto free_options;
aff063e2
JK
1185 }
1186
1187 err = get_valid_checkpoint(sbi);
a07ef784
NJ
1188 if (err) {
1189 f2fs_msg(sb, KERN_ERR, "Failed to get valid F2FS checkpoint");
aff063e2 1190 goto free_meta_inode;
a07ef784 1191 }
aff063e2
JK
1192
1193 /* sanity checking of checkpoint */
1194 err = -EINVAL;
577e3495 1195 if (sanity_check_ckpt(sbi)) {
a07ef784 1196 f2fs_msg(sb, KERN_ERR, "Invalid F2FS checkpoint");
aff063e2 1197 goto free_cp;
a07ef784 1198 }
aff063e2
JK
1199
1200 sbi->total_valid_node_count =
1201 le32_to_cpu(sbi->ckpt->valid_node_count);
1202 sbi->total_valid_inode_count =
1203 le32_to_cpu(sbi->ckpt->valid_inode_count);
1204 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
1205 sbi->total_valid_block_count =
1206 le64_to_cpu(sbi->ckpt->valid_block_count);
1207 sbi->last_valid_block_count = sbi->total_valid_block_count;
1208 sbi->alloc_valid_block_count = 0;
1209 INIT_LIST_HEAD(&sbi->dir_inode_list);
1210 spin_lock_init(&sbi->dir_inode_lock);
1211
1dcc336b
CY
1212 init_extent_cache_info(sbi);
1213
6451e041 1214 init_ino_entry_info(sbi);
aff063e2
JK
1215
1216 /* setup f2fs internal modules */
1217 err = build_segment_manager(sbi);
a07ef784
NJ
1218 if (err) {
1219 f2fs_msg(sb, KERN_ERR,
1220 "Failed to initialize F2FS segment manager");
aff063e2 1221 goto free_sm;
a07ef784 1222 }
aff063e2 1223 err = build_node_manager(sbi);
a07ef784
NJ
1224 if (err) {
1225 f2fs_msg(sb, KERN_ERR,
1226 "Failed to initialize F2FS node manager");
aff063e2 1227 goto free_nm;
a07ef784 1228 }
aff063e2
JK
1229
1230 build_gc_manager(sbi);
1231
1232 /* get an inode for node space */
1233 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
1234 if (IS_ERR(sbi->node_inode)) {
a07ef784 1235 f2fs_msg(sb, KERN_ERR, "Failed to read node inode");
aff063e2
JK
1236 err = PTR_ERR(sbi->node_inode);
1237 goto free_nm;
1238 }
1239
2658e50d
JK
1240 f2fs_join_shrinker(sbi);
1241
aff063e2 1242 /* if there are nt orphan nodes free them */
8f99a946 1243 recover_orphan_inodes(sbi);
aff063e2
JK
1244
1245 /* read root inode and dentry */
1246 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
1247 if (IS_ERR(root)) {
a07ef784 1248 f2fs_msg(sb, KERN_ERR, "Failed to read root inode");
aff063e2
JK
1249 err = PTR_ERR(root);
1250 goto free_node_inode;
1251 }
8f99a946 1252 if (!S_ISDIR(root->i_mode) || !root->i_blocks || !root->i_size) {
9d847950 1253 iput(root);
8f99a946 1254 err = -EINVAL;
9d847950 1255 goto free_node_inode;
8f99a946 1256 }
aff063e2
JK
1257
1258 sb->s_root = d_make_root(root); /* allocate root dentry */
1259 if (!sb->s_root) {
1260 err = -ENOMEM;
1261 goto free_root_inode;
1262 }
1263
aff063e2
JK
1264 err = f2fs_build_stats(sbi);
1265 if (err)
6437d1b0 1266 goto free_root_inode;
aff063e2 1267
5e176d54
JK
1268 if (f2fs_proc_root)
1269 sbi->s_proc = proc_mkdir(sb->s_id, f2fs_proc_root);
1270
1271 if (sbi->s_proc)
1272 proc_create_data("segment_info", S_IRUGO, sbi->s_proc,
1273 &f2fs_seq_segment_info_fops, sb);
1274
b59d0bae
NJ
1275 sbi->s_kobj.kset = f2fs_kset;
1276 init_completion(&sbi->s_kobj_unregister);
1277 err = kobject_init_and_add(&sbi->s_kobj, &f2fs_ktype, NULL,
1278 "%s", sb->s_id);
1279 if (err)
6437d1b0 1280 goto free_proc;
b59d0bae 1281
6437d1b0
JK
1282 /* recover fsynced data */
1283 if (!test_opt(sbi, DISABLE_ROLL_FORWARD)) {
081d78c2
JK
1284 /*
1285 * mount should be failed, when device has readonly mode, and
1286 * previous checkpoint was not done by clean system shutdown.
1287 */
1288 if (bdev_read_only(sb->s_bdev) &&
1289 !is_set_ckpt_flags(sbi->ckpt, CP_UMOUNT_FLAG)) {
1290 err = -EROFS;
1291 goto free_kobj;
1292 }
2adc3505
CY
1293
1294 if (need_fsck)
1295 set_sbi_flag(sbi, SBI_NEED_FSCK);
1296
6437d1b0 1297 err = recover_fsync_data(sbi);
ed2e621a 1298 if (err) {
2adc3505 1299 need_fsck = true;
6437d1b0
JK
1300 f2fs_msg(sb, KERN_ERR,
1301 "Cannot recover all fsync data errno=%ld", err);
ed2e621a
JK
1302 goto free_kobj;
1303 }
6437d1b0 1304 }
b59d0bae 1305
6437d1b0
JK
1306 /*
1307 * If filesystem is not mounted as read-only then
1308 * do start the gc_thread.
1309 */
6c029932 1310 if (test_opt(sbi, BG_GC) && !f2fs_readonly(sb)) {
6437d1b0
JK
1311 /* After POR, we can run background GC thread.*/
1312 err = start_gc_thread(sbi);
1313 if (err)
1314 goto free_kobj;
1315 }
dabc4a5c 1316 kfree(options);
da554e48 1317
1318 /* recover broken superblock */
1319 if (recovery && !f2fs_readonly(sb) && !bdev_read_only(sb->s_bdev)) {
1320 f2fs_msg(sb, KERN_INFO, "Recover invalid superblock");
c5bda1c8 1321 f2fs_commit_super(sbi, true);
da554e48 1322 }
1323
aff063e2 1324 return 0;
6437d1b0
JK
1325
1326free_kobj:
1327 kobject_del(&sbi->s_kobj);
1328free_proc:
1d15bd20
CY
1329 if (sbi->s_proc) {
1330 remove_proc_entry("segment_info", sbi->s_proc);
1331 remove_proc_entry(sb->s_id, f2fs_proc_root);
1332 }
1333 f2fs_destroy_stats(sbi);
aff063e2
JK
1334free_root_inode:
1335 dput(sb->s_root);
1336 sb->s_root = NULL;
1337free_node_inode:
2658e50d
JK
1338 mutex_lock(&sbi->umount_mutex);
1339 f2fs_leave_shrinker(sbi);
aff063e2 1340 iput(sbi->node_inode);
2658e50d 1341 mutex_unlock(&sbi->umount_mutex);
aff063e2
JK
1342free_nm:
1343 destroy_node_manager(sbi);
1344free_sm:
1345 destroy_segment_manager(sbi);
1346free_cp:
1347 kfree(sbi->ckpt);
1348free_meta_inode:
1349 make_bad_inode(sbi->meta_inode);
1350 iput(sbi->meta_inode);
dabc4a5c
JK
1351free_options:
1352 kfree(options);
aff063e2
JK
1353free_sb_buf:
1354 brelse(raw_super_buf);
1355free_sbi:
1356 kfree(sbi);
ed2e621a
JK
1357
1358 /* give only one another chance */
1359 if (retry) {
9df47ba7 1360 retry = false;
ed2e621a
JK
1361 shrink_dcache_sb(sb);
1362 goto try_onemore;
1363 }
aff063e2
JK
1364 return err;
1365}
1366
1367static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
1368 const char *dev_name, void *data)
1369{
1370 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
1371}
1372
30a5537f
JK
1373static void kill_f2fs_super(struct super_block *sb)
1374{
1375 if (sb->s_root)
caf0047e 1376 set_sbi_flag(F2FS_SB(sb), SBI_IS_CLOSE);
30a5537f
JK
1377 kill_block_super(sb);
1378}
1379
aff063e2
JK
1380static struct file_system_type f2fs_fs_type = {
1381 .owner = THIS_MODULE,
1382 .name = "f2fs",
1383 .mount = f2fs_mount,
30a5537f 1384 .kill_sb = kill_f2fs_super,
aff063e2
JK
1385 .fs_flags = FS_REQUIRES_DEV,
1386};
7f78e035 1387MODULE_ALIAS_FS("f2fs");
aff063e2 1388
6e6093a8 1389static int __init init_inodecache(void)
aff063e2
JK
1390{
1391 f2fs_inode_cachep = f2fs_kmem_cache_create("f2fs_inode_cache",
e8512d2e 1392 sizeof(struct f2fs_inode_info));
6bacf52f 1393 if (!f2fs_inode_cachep)
aff063e2
JK
1394 return -ENOMEM;
1395 return 0;
1396}
1397
1398static void destroy_inodecache(void)
1399{
1400 /*
1401 * Make sure all delayed rcu free inodes are flushed before we
1402 * destroy cache.
1403 */
1404 rcu_barrier();
1405 kmem_cache_destroy(f2fs_inode_cachep);
1406}
1407
1408static int __init init_f2fs_fs(void)
1409{
1410 int err;
1411
c0508650
JK
1412 f2fs_build_trace_ios();
1413
aff063e2
JK
1414 err = init_inodecache();
1415 if (err)
1416 goto fail;
1417 err = create_node_manager_caches();
1418 if (err)
9890ff3f 1419 goto free_inodecache;
7fd9e544 1420 err = create_segment_manager_caches();
aff063e2 1421 if (err)
9890ff3f 1422 goto free_node_manager_caches;
aff063e2
JK
1423 err = create_checkpoint_caches();
1424 if (err)
06292073 1425 goto free_segment_manager_caches;
1dcc336b
CY
1426 err = create_extent_cache();
1427 if (err)
1428 goto free_checkpoint_caches;
b59d0bae 1429 f2fs_kset = kset_create_and_add("f2fs", NULL, fs_kobj);
6e6b978c
WY
1430 if (!f2fs_kset) {
1431 err = -ENOMEM;
1dcc336b 1432 goto free_extent_cache;
6e6b978c 1433 }
cfc4d971 1434 err = f2fs_init_crypto();
4589d25d 1435 if (err)
9890ff3f 1436 goto free_kset;
2658e50d
JK
1437
1438 err = register_shrinker(&f2fs_shrinker_info);
cfc4d971
JK
1439 if (err)
1440 goto free_crypto;
2658e50d
JK
1441
1442 err = register_filesystem(&f2fs_fs_type);
1443 if (err)
1444 goto free_shrinker;
4589d25d 1445 f2fs_create_root_stats();
5e176d54 1446 f2fs_proc_root = proc_mkdir("fs/f2fs", NULL);
9890ff3f
ZH
1447 return 0;
1448
2658e50d
JK
1449free_shrinker:
1450 unregister_shrinker(&f2fs_shrinker_info);
cfc4d971
JK
1451free_crypto:
1452 f2fs_exit_crypto();
9890ff3f
ZH
1453free_kset:
1454 kset_unregister(f2fs_kset);
1dcc336b
CY
1455free_extent_cache:
1456 destroy_extent_cache();
9890ff3f
ZH
1457free_checkpoint_caches:
1458 destroy_checkpoint_caches();
7fd9e544
JK
1459free_segment_manager_caches:
1460 destroy_segment_manager_caches();
9890ff3f
ZH
1461free_node_manager_caches:
1462 destroy_node_manager_caches();
1463free_inodecache:
1464 destroy_inodecache();
aff063e2
JK
1465fail:
1466 return err;
1467}
1468
1469static void __exit exit_f2fs_fs(void)
1470{
5e176d54 1471 remove_proc_entry("fs/f2fs", NULL);
4589d25d 1472 f2fs_destroy_root_stats();
2658e50d 1473 unregister_shrinker(&f2fs_shrinker_info);
aff063e2 1474 unregister_filesystem(&f2fs_fs_type);
cfc4d971 1475 f2fs_exit_crypto();
fdf6c8be 1476 destroy_extent_cache();
aff063e2 1477 destroy_checkpoint_caches();
5dcd8a71 1478 destroy_segment_manager_caches();
aff063e2
JK
1479 destroy_node_manager_caches();
1480 destroy_inodecache();
b59d0bae 1481 kset_unregister(f2fs_kset);
351f4fba 1482 f2fs_destroy_trace_ios();
aff063e2
JK
1483}
1484
1485module_init(init_f2fs_fs)
1486module_exit(exit_f2fs_fs)
1487
1488MODULE_AUTHOR("Samsung Electronics's Praesto Team");
1489MODULE_DESCRIPTION("Flash Friendly File System");
1490MODULE_LICENSE("GPL");