]> git.proxmox.com Git - mirror_ubuntu-bionic-kernel.git/blame - fs/f2fs/super.c
f2fs: readahead for free nids building
[mirror_ubuntu-bionic-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);
60b99b48 218F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, cp_interval, cp_interval);
b59d0bae
NJ
219
220#define ATTR_LIST(name) (&f2fs_attr_##name.attr)
221static struct attribute *f2fs_attrs[] = {
222 ATTR_LIST(gc_min_sleep_time),
223 ATTR_LIST(gc_max_sleep_time),
224 ATTR_LIST(gc_no_gc_sleep_time),
d2dc095f 225 ATTR_LIST(gc_idle),
ea91e9b0 226 ATTR_LIST(reclaim_segments),
7ac8c3b0 227 ATTR_LIST(max_small_discards),
bba681cb 228 ATTR_LIST(batched_trim_sections),
216fbd64
JK
229 ATTR_LIST(ipu_policy),
230 ATTR_LIST(min_ipu_util),
c1ce1b02 231 ATTR_LIST(min_fsync_blocks),
b1c57c1c 232 ATTR_LIST(max_victim_search),
ab9fa662 233 ATTR_LIST(dir_level),
cdfc41c1 234 ATTR_LIST(ram_thresh),
60b99b48 235 ATTR_LIST(cp_interval),
b59d0bae
NJ
236 NULL,
237};
238
239static const struct sysfs_ops f2fs_attr_ops = {
240 .show = f2fs_attr_show,
241 .store = f2fs_attr_store,
242};
243
244static struct kobj_type f2fs_ktype = {
245 .default_attrs = f2fs_attrs,
246 .sysfs_ops = &f2fs_attr_ops,
247 .release = f2fs_sb_release,
248};
249
a07ef784
NJ
250void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...)
251{
252 struct va_format vaf;
253 va_list args;
254
255 va_start(args, fmt);
256 vaf.fmt = fmt;
257 vaf.va = &args;
258 printk("%sF2FS-fs (%s): %pV\n", level, sb->s_id, &vaf);
259 va_end(args);
260}
261
aff063e2
JK
262static void init_once(void *foo)
263{
264 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
265
aff063e2
JK
266 inode_init_once(&fi->vfs_inode);
267}
268
696c018c
NJ
269static int parse_options(struct super_block *sb, char *options)
270{
271 struct f2fs_sb_info *sbi = F2FS_SB(sb);
09d54cdd 272 struct request_queue *q;
696c018c
NJ
273 substring_t args[MAX_OPT_ARGS];
274 char *p, *name;
275 int arg = 0;
276
277 if (!options)
278 return 0;
279
280 while ((p = strsep(&options, ",")) != NULL) {
281 int token;
282 if (!*p)
283 continue;
284 /*
285 * Initialize args struct so we know whether arg was
286 * found; some options take optional arguments.
287 */
288 args[0].to = args[0].from = NULL;
289 token = match_token(p, f2fs_tokens, args);
290
291 switch (token) {
292 case Opt_gc_background:
293 name = match_strdup(&args[0]);
294
295 if (!name)
296 return -ENOMEM;
6aefd93b 297 if (strlen(name) == 2 && !strncmp(name, "on", 2)) {
696c018c 298 set_opt(sbi, BG_GC);
6aefd93b
JK
299 clear_opt(sbi, FORCE_FG_GC);
300 } else if (strlen(name) == 3 && !strncmp(name, "off", 3)) {
696c018c 301 clear_opt(sbi, BG_GC);
6aefd93b
JK
302 clear_opt(sbi, FORCE_FG_GC);
303 } else if (strlen(name) == 4 && !strncmp(name, "sync", 4)) {
304 set_opt(sbi, BG_GC);
305 set_opt(sbi, FORCE_FG_GC);
306 } else {
696c018c
NJ
307 kfree(name);
308 return -EINVAL;
309 }
310 kfree(name);
311 break;
312 case Opt_disable_roll_forward:
313 set_opt(sbi, DISABLE_ROLL_FORWARD);
314 break;
2d834bf9
JK
315 case Opt_norecovery:
316 /* this option mounts f2fs with ro */
317 set_opt(sbi, DISABLE_ROLL_FORWARD);
318 if (!f2fs_readonly(sb))
319 return -EINVAL;
320 break;
696c018c 321 case Opt_discard:
09d54cdd
CY
322 q = bdev_get_queue(sb->s_bdev);
323 if (blk_queue_discard(q)) {
324 set_opt(sbi, DISCARD);
325 } else {
326 f2fs_msg(sb, KERN_WARNING,
327 "mounting with \"discard\" option, but "
328 "the device does not support discard");
329 }
696c018c
NJ
330 break;
331 case Opt_noheap:
332 set_opt(sbi, NOHEAP);
333 break;
334#ifdef CONFIG_F2FS_FS_XATTR
4058c511
KA
335 case Opt_user_xattr:
336 set_opt(sbi, XATTR_USER);
337 break;
696c018c
NJ
338 case Opt_nouser_xattr:
339 clear_opt(sbi, XATTR_USER);
340 break;
444c580f
JK
341 case Opt_inline_xattr:
342 set_opt(sbi, INLINE_XATTR);
343 break;
696c018c 344#else
4058c511
KA
345 case Opt_user_xattr:
346 f2fs_msg(sb, KERN_INFO,
347 "user_xattr options not supported");
348 break;
696c018c
NJ
349 case Opt_nouser_xattr:
350 f2fs_msg(sb, KERN_INFO,
351 "nouser_xattr options not supported");
352 break;
444c580f
JK
353 case Opt_inline_xattr:
354 f2fs_msg(sb, KERN_INFO,
355 "inline_xattr options not supported");
356 break;
696c018c
NJ
357#endif
358#ifdef CONFIG_F2FS_FS_POSIX_ACL
4058c511
KA
359 case Opt_acl:
360 set_opt(sbi, POSIX_ACL);
361 break;
696c018c
NJ
362 case Opt_noacl:
363 clear_opt(sbi, POSIX_ACL);
364 break;
365#else
4058c511
KA
366 case Opt_acl:
367 f2fs_msg(sb, KERN_INFO, "acl options not supported");
368 break;
696c018c
NJ
369 case Opt_noacl:
370 f2fs_msg(sb, KERN_INFO, "noacl options not supported");
371 break;
372#endif
373 case Opt_active_logs:
374 if (args->from && match_int(args, &arg))
375 return -EINVAL;
376 if (arg != 2 && arg != 4 && arg != NR_CURSEG_TYPE)
377 return -EINVAL;
378 sbi->active_logs = arg;
379 break;
380 case Opt_disable_ext_identify:
381 set_opt(sbi, DISABLE_EXT_IDENTIFY);
382 break;
8274de77
HL
383 case Opt_inline_data:
384 set_opt(sbi, INLINE_DATA);
385 break;
5efd3c6f
CY
386 case Opt_inline_dentry:
387 set_opt(sbi, INLINE_DENTRY);
388 break;
6b4afdd7
JK
389 case Opt_flush_merge:
390 set_opt(sbi, FLUSH_MERGE);
391 break;
0f7b2abd
JK
392 case Opt_nobarrier:
393 set_opt(sbi, NOBARRIER);
394 break;
d5053a34
JK
395 case Opt_fastboot:
396 set_opt(sbi, FASTBOOT);
397 break;
89672159
CY
398 case Opt_extent_cache:
399 set_opt(sbi, EXTENT_CACHE);
400 break;
7daaea25
JK
401 case Opt_noextent_cache:
402 clear_opt(sbi, EXTENT_CACHE);
403 break;
75342797
WL
404 case Opt_noinline_data:
405 clear_opt(sbi, INLINE_DATA);
406 break;
696c018c
NJ
407 default:
408 f2fs_msg(sb, KERN_ERR,
409 "Unrecognized mount option \"%s\" or missing value",
410 p);
411 return -EINVAL;
412 }
413 }
414 return 0;
415}
416
aff063e2
JK
417static struct inode *f2fs_alloc_inode(struct super_block *sb)
418{
419 struct f2fs_inode_info *fi;
420
a0acdfe0 421 fi = kmem_cache_alloc(f2fs_inode_cachep, GFP_F2FS_ZERO);
aff063e2
JK
422 if (!fi)
423 return NULL;
424
425 init_once((void *) fi);
426
434720fa 427 /* Initialize f2fs-specific inode info */
aff063e2 428 fi->vfs_inode.i_version = 1;
a7ffdbe2 429 atomic_set(&fi->dirty_pages, 0);
aff063e2
JK
430 fi->i_current_depth = 1;
431 fi->i_advise = 0;
d928bfbf 432 init_rwsem(&fi->i_sem);
88b88a66
JK
433 INIT_LIST_HEAD(&fi->inmem_pages);
434 mutex_init(&fi->inmem_lock);
aff063e2
JK
435
436 set_inode_flag(fi, FI_NEW_INODE);
437
444c580f
JK
438 if (test_opt(F2FS_SB(sb), INLINE_XATTR))
439 set_inode_flag(fi, FI_INLINE_XATTR);
440
ab9fa662
JK
441 /* Will be used by directory only */
442 fi->i_dir_level = F2FS_SB(sb)->dir_level;
443
57e5055b
JK
444#ifdef CONFIG_F2FS_FS_ENCRYPTION
445 fi->i_crypt_info = NULL;
446#endif
aff063e2
JK
447 return &fi->vfs_inode;
448}
449
531ad7d5
JK
450static int f2fs_drop_inode(struct inode *inode)
451{
452 /*
453 * This is to avoid a deadlock condition like below.
454 * writeback_single_inode(inode)
455 * - f2fs_write_data_page
456 * - f2fs_gc -> iput -> evict
457 * - inode_wait_for_writeback(inode)
458 */
06e1bc05
JK
459 if (!inode_unhashed(inode) && inode->i_state & I_SYNC) {
460 if (!inode->i_nlink && !is_bad_inode(inode)) {
3e72f721
JK
461 /* to avoid evict_inode call simultaneously */
462 atomic_inc(&inode->i_count);
06e1bc05
JK
463 spin_unlock(&inode->i_lock);
464
465 /* some remained atomic pages should discarded */
466 if (f2fs_is_atomic_file(inode))
467 commit_inmem_pages(inode, true);
468
3e72f721
JK
469 /* should remain fi->extent_tree for writepage */
470 f2fs_destroy_extent_node(inode);
471
06e1bc05
JK
472 sb_start_intwrite(inode->i_sb);
473 i_size_write(inode, 0);
474
475 if (F2FS_HAS_BLOCKS(inode))
55f57d2c 476 f2fs_truncate(inode, true);
06e1bc05
JK
477
478 sb_end_intwrite(inode->i_sb);
479
480#ifdef CONFIG_F2FS_FS_ENCRYPTION
481 if (F2FS_I(inode)->i_crypt_info)
26bf3dc7
JK
482 f2fs_free_encryption_info(inode,
483 F2FS_I(inode)->i_crypt_info);
06e1bc05
JK
484#endif
485 spin_lock(&inode->i_lock);
3e72f721 486 atomic_dec(&inode->i_count);
06e1bc05 487 }
531ad7d5 488 return 0;
06e1bc05 489 }
531ad7d5
JK
490 return generic_drop_inode(inode);
491}
492
b3783873
JK
493/*
494 * f2fs_dirty_inode() is called from __mark_inode_dirty()
495 *
496 * We should call set_dirty_inode to write the dirty inode through write_inode.
497 */
498static void f2fs_dirty_inode(struct inode *inode, int flags)
499{
500 set_inode_flag(F2FS_I(inode), FI_DIRTY_INODE);
b3783873
JK
501}
502
aff063e2
JK
503static void f2fs_i_callback(struct rcu_head *head)
504{
505 struct inode *inode = container_of(head, struct inode, i_rcu);
506 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
507}
508
25ca923b 509static void f2fs_destroy_inode(struct inode *inode)
aff063e2
JK
510{
511 call_rcu(&inode->i_rcu, f2fs_i_callback);
512}
513
514static void f2fs_put_super(struct super_block *sb)
515{
516 struct f2fs_sb_info *sbi = F2FS_SB(sb);
517
5e176d54
JK
518 if (sbi->s_proc) {
519 remove_proc_entry("segment_info", sbi->s_proc);
520 remove_proc_entry(sb->s_id, f2fs_proc_root);
521 }
b59d0bae 522 kobject_del(&sbi->s_kobj);
5e176d54 523
aff063e2
JK
524 stop_gc_thread(sbi);
525
2658e50d
JK
526 /* prevent remaining shrinker jobs */
527 mutex_lock(&sbi->umount_mutex);
528
85dc2f2c
JK
529 /*
530 * We don't need to do checkpoint when superblock is clean.
531 * But, the previous checkpoint was not done by umount, it needs to do
532 * clean checkpoint again.
533 */
caf0047e 534 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
85dc2f2c 535 !is_set_ckpt_flags(F2FS_CKPT(sbi), CP_UMOUNT_FLAG)) {
75ab4cb8
JK
536 struct cp_control cpc = {
537 .reason = CP_UMOUNT,
538 };
539 write_checkpoint(sbi, &cpc);
540 }
aff063e2 541
eca616f8
JK
542 /* write_checkpoint can update stat informaion */
543 f2fs_destroy_stats(sbi);
544
cf779cab
JK
545 /*
546 * normally superblock is clean, so we need to release this.
547 * In addition, EIO will skip do checkpoint, we need this as well.
548 */
6f12ac25 549 release_dirty_inode(sbi);
4b2fecc8 550 release_discard_addrs(sbi);
6f12ac25 551
2658e50d
JK
552 f2fs_leave_shrinker(sbi);
553 mutex_unlock(&sbi->umount_mutex);
554
aff063e2
JK
555 iput(sbi->node_inode);
556 iput(sbi->meta_inode);
557
558 /* destroy f2fs internal modules */
559 destroy_node_manager(sbi);
560 destroy_segment_manager(sbi);
561
562 kfree(sbi->ckpt);
b59d0bae
NJ
563 kobject_put(&sbi->s_kobj);
564 wait_for_completion(&sbi->s_kobj_unregister);
aff063e2
JK
565
566 sb->s_fs_info = NULL;
567 brelse(sbi->raw_super_buf);
568 kfree(sbi);
569}
570
571int f2fs_sync_fs(struct super_block *sb, int sync)
572{
573 struct f2fs_sb_info *sbi = F2FS_SB(sb);
aff063e2 574
a2a4a7e4
NJ
575 trace_f2fs_sync_fs(sb, sync);
576
b7473754 577 if (sync) {
d5053a34
JK
578 struct cp_control cpc;
579
119ee914
JK
580 cpc.reason = __get_cp_reason(sbi);
581
b7473754 582 mutex_lock(&sbi->gc_mutex);
75ab4cb8 583 write_checkpoint(sbi, &cpc);
b7473754
JK
584 mutex_unlock(&sbi->gc_mutex);
585 } else {
7d82db83 586 f2fs_balance_fs(sbi);
b7473754 587 }
05ca3632 588 f2fs_trace_ios(NULL, 1);
aff063e2 589
64c576fe 590 return 0;
aff063e2
JK
591}
592
d6212a5f
CL
593static int f2fs_freeze(struct super_block *sb)
594{
595 int err;
596
77888c1e 597 if (f2fs_readonly(sb))
d6212a5f
CL
598 return 0;
599
600 err = f2fs_sync_fs(sb, 1);
601 return err;
602}
603
604static int f2fs_unfreeze(struct super_block *sb)
605{
606 return 0;
607}
608
aff063e2
JK
609static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
610{
611 struct super_block *sb = dentry->d_sb;
612 struct f2fs_sb_info *sbi = F2FS_SB(sb);
613 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
614 block_t total_count, user_block_count, start_count, ovp_count;
615
616 total_count = le64_to_cpu(sbi->raw_super->block_count);
617 user_block_count = sbi->user_block_count;
618 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
619 ovp_count = SM_I(sbi)->ovp_segments << sbi->log_blocks_per_seg;
620 buf->f_type = F2FS_SUPER_MAGIC;
621 buf->f_bsize = sbi->blocksize;
622
623 buf->f_blocks = total_count - start_count;
624 buf->f_bfree = buf->f_blocks - valid_user_blocks(sbi) - ovp_count;
625 buf->f_bavail = user_block_count - valid_user_blocks(sbi);
626
c200b1aa
CY
627 buf->f_files = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
628 buf->f_ffree = buf->f_files - valid_inode_count(sbi);
aff063e2 629
5a20d339 630 buf->f_namelen = F2FS_NAME_LEN;
aff063e2
JK
631 buf->f_fsid.val[0] = (u32)id;
632 buf->f_fsid.val[1] = (u32)(id >> 32);
633
634 return 0;
635}
636
637static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
638{
639 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
640
6aefd93b
JK
641 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, BG_GC)) {
642 if (test_opt(sbi, FORCE_FG_GC))
643 seq_printf(seq, ",background_gc=%s", "sync");
644 else
645 seq_printf(seq, ",background_gc=%s", "on");
646 } else {
696c018c 647 seq_printf(seq, ",background_gc=%s", "off");
6aefd93b 648 }
aff063e2
JK
649 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
650 seq_puts(seq, ",disable_roll_forward");
651 if (test_opt(sbi, DISCARD))
652 seq_puts(seq, ",discard");
653 if (test_opt(sbi, NOHEAP))
654 seq_puts(seq, ",no_heap_alloc");
655#ifdef CONFIG_F2FS_FS_XATTR
656 if (test_opt(sbi, XATTR_USER))
657 seq_puts(seq, ",user_xattr");
658 else
659 seq_puts(seq, ",nouser_xattr");
444c580f
JK
660 if (test_opt(sbi, INLINE_XATTR))
661 seq_puts(seq, ",inline_xattr");
aff063e2
JK
662#endif
663#ifdef CONFIG_F2FS_FS_POSIX_ACL
664 if (test_opt(sbi, POSIX_ACL))
665 seq_puts(seq, ",acl");
666 else
667 seq_puts(seq, ",noacl");
668#endif
669 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
aa43507f 670 seq_puts(seq, ",disable_ext_identify");
8274de77
HL
671 if (test_opt(sbi, INLINE_DATA))
672 seq_puts(seq, ",inline_data");
75342797
WL
673 else
674 seq_puts(seq, ",noinline_data");
5efd3c6f
CY
675 if (test_opt(sbi, INLINE_DENTRY))
676 seq_puts(seq, ",inline_dentry");
b270ad6f 677 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
6b4afdd7 678 seq_puts(seq, ",flush_merge");
0f7b2abd
JK
679 if (test_opt(sbi, NOBARRIER))
680 seq_puts(seq, ",nobarrier");
d5053a34
JK
681 if (test_opt(sbi, FASTBOOT))
682 seq_puts(seq, ",fastboot");
89672159
CY
683 if (test_opt(sbi, EXTENT_CACHE))
684 seq_puts(seq, ",extent_cache");
7daaea25
JK
685 else
686 seq_puts(seq, ",noextent_cache");
aff063e2
JK
687 seq_printf(seq, ",active_logs=%u", sbi->active_logs);
688
689 return 0;
690}
691
5e176d54
JK
692static int segment_info_seq_show(struct seq_file *seq, void *offset)
693{
694 struct super_block *sb = seq->private;
695 struct f2fs_sb_info *sbi = F2FS_SB(sb);
6c311ec6
CF
696 unsigned int total_segs =
697 le32_to_cpu(sbi->raw_super->segment_count_main);
5e176d54
JK
698 int i;
699
90aa6dc9
CY
700 seq_puts(seq, "format: segment_type|valid_blocks\n"
701 "segment_type(0:HD, 1:WD, 2:CD, 3:HN, 4:WN, 5:CN)\n");
702
5e176d54 703 for (i = 0; i < total_segs; i++) {
90aa6dc9
CY
704 struct seg_entry *se = get_seg_entry(sbi, i);
705
706 if ((i % 10) == 0)
01a5ad82 707 seq_printf(seq, "%-10d", i);
90aa6dc9
CY
708 seq_printf(seq, "%d|%-3u", se->type,
709 get_valid_blocks(sbi, i, 1));
46c04366
GZ
710 if ((i % 10) == 9 || i == (total_segs - 1))
711 seq_putc(seq, '\n');
5e176d54 712 else
46c04366 713 seq_putc(seq, ' ');
5e176d54 714 }
46c04366 715
5e176d54
JK
716 return 0;
717}
718
719static int segment_info_open_fs(struct inode *inode, struct file *file)
720{
721 return single_open(file, segment_info_seq_show, PDE_DATA(inode));
722}
723
724static const struct file_operations f2fs_seq_segment_info_fops = {
725 .owner = THIS_MODULE,
726 .open = segment_info_open_fs,
727 .read = seq_read,
728 .llseek = seq_lseek,
729 .release = single_release,
730};
731
498c5e9f
YH
732static void default_options(struct f2fs_sb_info *sbi)
733{
734 /* init some FS parameters */
735 sbi->active_logs = NR_CURSEG_TYPE;
736
737 set_opt(sbi, BG_GC);
738 set_opt(sbi, INLINE_DATA);
3e72f721 739 set_opt(sbi, EXTENT_CACHE);
498c5e9f
YH
740
741#ifdef CONFIG_F2FS_FS_XATTR
742 set_opt(sbi, XATTR_USER);
743#endif
744#ifdef CONFIG_F2FS_FS_POSIX_ACL
745 set_opt(sbi, POSIX_ACL);
746#endif
747}
748
696c018c
NJ
749static int f2fs_remount(struct super_block *sb, int *flags, char *data)
750{
751 struct f2fs_sb_info *sbi = F2FS_SB(sb);
752 struct f2fs_mount_info org_mount_opt;
753 int err, active_logs;
876dc59e
GZ
754 bool need_restart_gc = false;
755 bool need_stop_gc = false;
9cd81ce3 756 bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
696c018c 757
02b9984d
TT
758 sync_filesystem(sb);
759
696c018c
NJ
760 /*
761 * Save the old mount options in case we
762 * need to restore them.
763 */
764 org_mount_opt = sbi->mount_opt;
765 active_logs = sbi->active_logs;
766
26666c8a 767 sbi->mount_opt.opt = 0;
498c5e9f 768 default_options(sbi);
26666c8a 769
696c018c
NJ
770 /* parse mount options */
771 err = parse_options(sb, data);
772 if (err)
773 goto restore_opts;
774
775 /*
776 * Previous and new state of filesystem is RO,
876dc59e 777 * so skip checking GC and FLUSH_MERGE conditions.
696c018c 778 */
6b2920a5 779 if (f2fs_readonly(sb) && (*flags & MS_RDONLY))
696c018c
NJ
780 goto skip;
781
9cd81ce3
CY
782 /* disallow enable/disable extent_cache dynamically */
783 if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
784 err = -EINVAL;
785 f2fs_msg(sbi->sb, KERN_WARNING,
786 "switch extent_cache option is not allowed");
787 goto restore_opts;
788 }
789
696c018c
NJ
790 /*
791 * We stop the GC thread if FS is mounted as RO
792 * or if background_gc = off is passed in mount
793 * option. Also sync the filesystem.
794 */
795 if ((*flags & MS_RDONLY) || !test_opt(sbi, BG_GC)) {
796 if (sbi->gc_thread) {
797 stop_gc_thread(sbi);
798 f2fs_sync_fs(sb, 1);
876dc59e 799 need_restart_gc = true;
696c018c 800 }
aba291b3 801 } else if (!sbi->gc_thread) {
696c018c
NJ
802 err = start_gc_thread(sbi);
803 if (err)
804 goto restore_opts;
876dc59e
GZ
805 need_stop_gc = true;
806 }
807
808 /*
809 * We stop issue flush thread if FS is mounted as RO
810 * or if flush_merge is not passed in mount option.
811 */
812 if ((*flags & MS_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2163d198 813 destroy_flush_cmd_control(sbi);
aba291b3 814 } else if (!SM_I(sbi)->cmd_control_info) {
2163d198
GZ
815 err = create_flush_cmd_control(sbi);
816 if (err)
a688b9d9 817 goto restore_gc;
696c018c
NJ
818 }
819skip:
820 /* Update the POSIXACL Flag */
821 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
822 (test_opt(sbi, POSIX_ACL) ? MS_POSIXACL : 0);
823 return 0;
876dc59e
GZ
824restore_gc:
825 if (need_restart_gc) {
826 if (start_gc_thread(sbi))
827 f2fs_msg(sbi->sb, KERN_WARNING,
e1c42045 828 "background gc thread has stopped");
876dc59e
GZ
829 } else if (need_stop_gc) {
830 stop_gc_thread(sbi);
831 }
696c018c
NJ
832restore_opts:
833 sbi->mount_opt = org_mount_opt;
834 sbi->active_logs = active_logs;
835 return err;
836}
837
aff063e2
JK
838static struct super_operations f2fs_sops = {
839 .alloc_inode = f2fs_alloc_inode,
531ad7d5 840 .drop_inode = f2fs_drop_inode,
aff063e2
JK
841 .destroy_inode = f2fs_destroy_inode,
842 .write_inode = f2fs_write_inode,
b3783873 843 .dirty_inode = f2fs_dirty_inode,
aff063e2
JK
844 .show_options = f2fs_show_options,
845 .evict_inode = f2fs_evict_inode,
846 .put_super = f2fs_put_super,
847 .sync_fs = f2fs_sync_fs,
d6212a5f
CL
848 .freeze_fs = f2fs_freeze,
849 .unfreeze_fs = f2fs_unfreeze,
aff063e2 850 .statfs = f2fs_statfs,
696c018c 851 .remount_fs = f2fs_remount,
aff063e2
JK
852};
853
854static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
855 u64 ino, u32 generation)
856{
857 struct f2fs_sb_info *sbi = F2FS_SB(sb);
858 struct inode *inode;
859
d6b7d4b3 860 if (check_nid_range(sbi, ino))
910bb12d 861 return ERR_PTR(-ESTALE);
aff063e2
JK
862
863 /*
864 * f2fs_iget isn't quite right if the inode is currently unallocated!
865 * However f2fs_iget currently does appropriate checks to handle stale
866 * inodes so everything is OK.
867 */
868 inode = f2fs_iget(sb, ino);
869 if (IS_ERR(inode))
870 return ERR_CAST(inode);
6bacf52f 871 if (unlikely(generation && inode->i_generation != generation)) {
aff063e2
JK
872 /* we didn't find the right inode.. */
873 iput(inode);
874 return ERR_PTR(-ESTALE);
875 }
876 return inode;
877}
878
879static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
880 int fh_len, int fh_type)
881{
882 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
883 f2fs_nfs_get_inode);
884}
885
886static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
887 int fh_len, int fh_type)
888{
889 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
890 f2fs_nfs_get_inode);
891}
892
893static const struct export_operations f2fs_export_ops = {
894 .fh_to_dentry = f2fs_fh_to_dentry,
895 .fh_to_parent = f2fs_fh_to_parent,
896 .get_parent = f2fs_get_parent,
897};
898
aff063e2
JK
899static loff_t max_file_size(unsigned bits)
900{
de93653f 901 loff_t result = (DEF_ADDRS_PER_INODE - F2FS_INLINE_XATTR_ADDRS);
aff063e2
JK
902 loff_t leaf_count = ADDRS_PER_BLOCK;
903
904 /* two direct node blocks */
905 result += (leaf_count * 2);
906
907 /* two indirect node blocks */
908 leaf_count *= NIDS_PER_BLOCK;
909 result += (leaf_count * 2);
910
911 /* one double indirect node block */
912 leaf_count *= NIDS_PER_BLOCK;
913 result += leaf_count;
914
915 result <<= bits;
916 return result;
917}
918
a07ef784
NJ
919static int sanity_check_raw_super(struct super_block *sb,
920 struct f2fs_super_block *raw_super)
aff063e2
JK
921{
922 unsigned int blocksize;
923
a07ef784
NJ
924 if (F2FS_SUPER_MAGIC != le32_to_cpu(raw_super->magic)) {
925 f2fs_msg(sb, KERN_INFO,
926 "Magic Mismatch, valid(0x%x) - read(0x%x)",
927 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
aff063e2 928 return 1;
a07ef784 929 }
aff063e2 930
5c9b4692 931 /* Currently, support only 4KB page cache size */
932 if (F2FS_BLKSIZE != PAGE_CACHE_SIZE) {
933 f2fs_msg(sb, KERN_INFO,
14d7e9de 934 "Invalid page_cache_size (%lu), supports only 4KB\n",
5c9b4692 935 PAGE_CACHE_SIZE);
936 return 1;
937 }
938
aff063e2
JK
939 /* Currently, support only 4KB block size */
940 blocksize = 1 << le32_to_cpu(raw_super->log_blocksize);
5c9b4692 941 if (blocksize != F2FS_BLKSIZE) {
a07ef784
NJ
942 f2fs_msg(sb, KERN_INFO,
943 "Invalid blocksize (%u), supports only 4KB\n",
944 blocksize);
aff063e2 945 return 1;
a07ef784 946 }
5c9b4692 947
55cf9cb6
CY
948 /* Currently, support 512/1024/2048/4096 bytes sector size */
949 if (le32_to_cpu(raw_super->log_sectorsize) >
950 F2FS_MAX_LOG_SECTOR_SIZE ||
951 le32_to_cpu(raw_super->log_sectorsize) <
952 F2FS_MIN_LOG_SECTOR_SIZE) {
953 f2fs_msg(sb, KERN_INFO, "Invalid log sectorsize (%u)",
954 le32_to_cpu(raw_super->log_sectorsize));
aff063e2 955 return 1;
a07ef784 956 }
55cf9cb6
CY
957 if (le32_to_cpu(raw_super->log_sectors_per_block) +
958 le32_to_cpu(raw_super->log_sectorsize) !=
959 F2FS_MAX_LOG_SECTOR_SIZE) {
960 f2fs_msg(sb, KERN_INFO,
961 "Invalid log sectors per block(%u) log sectorsize(%u)",
962 le32_to_cpu(raw_super->log_sectors_per_block),
963 le32_to_cpu(raw_super->log_sectorsize));
aff063e2 964 return 1;
a07ef784 965 }
aff063e2
JK
966 return 0;
967}
968
577e3495 969static int sanity_check_ckpt(struct f2fs_sb_info *sbi)
aff063e2
JK
970{
971 unsigned int total, fsmeta;
577e3495
JK
972 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
973 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
aff063e2
JK
974
975 total = le32_to_cpu(raw_super->segment_count);
976 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
977 fsmeta += le32_to_cpu(raw_super->segment_count_sit);
978 fsmeta += le32_to_cpu(raw_super->segment_count_nat);
979 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
980 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
981
6bacf52f 982 if (unlikely(fsmeta >= total))
aff063e2 983 return 1;
577e3495 984
1e968fdf 985 if (unlikely(f2fs_cp_error(sbi))) {
577e3495
JK
986 f2fs_msg(sbi->sb, KERN_ERR, "A bug case: need to run fsck");
987 return 1;
988 }
aff063e2
JK
989 return 0;
990}
991
992static void init_sb_info(struct f2fs_sb_info *sbi)
993{
994 struct f2fs_super_block *raw_super = sbi->raw_super;
995 int i;
996
997 sbi->log_sectors_per_block =
998 le32_to_cpu(raw_super->log_sectors_per_block);
999 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
1000 sbi->blocksize = 1 << sbi->log_blocksize;
1001 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
1002 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
1003 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
1004 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
1005 sbi->total_sections = le32_to_cpu(raw_super->section_count);
1006 sbi->total_node_count =
1007 (le32_to_cpu(raw_super->segment_count_nat) / 2)
1008 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
1009 sbi->root_ino_num = le32_to_cpu(raw_super->root_ino);
1010 sbi->node_ino_num = le32_to_cpu(raw_super->node_ino);
1011 sbi->meta_ino_num = le32_to_cpu(raw_super->meta_ino);
5ec4e49f 1012 sbi->cur_victim_sec = NULL_SECNO;
b1c57c1c 1013 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
aff063e2
JK
1014
1015 for (i = 0; i < NR_COUNT_TYPE; i++)
1016 atomic_set(&sbi->nr_pages[i], 0);
ab9fa662
JK
1017
1018 sbi->dir_level = DEF_DIR_LEVEL;
60b99b48 1019 sbi->cp_interval = DEF_CP_INTERVAL;
caf0047e 1020 clear_sbi_flag(sbi, SBI_NEED_FSCK);
2658e50d
JK
1021
1022 INIT_LIST_HEAD(&sbi->s_list);
1023 mutex_init(&sbi->umount_mutex);
aff063e2
JK
1024}
1025
9076a75f
GZ
1026/*
1027 * Read f2fs raw super block.
1028 * Because we have two copies of super block, so read the first one at first,
1029 * if the first one is invalid, move to read the second one.
1030 */
1031static int read_raw_super_block(struct super_block *sb,
1032 struct f2fs_super_block **raw_super,
da554e48 1033 struct buffer_head **raw_super_buf,
1034 int *recovery)
14d7e9de 1035{
9076a75f 1036 int block = 0;
da554e48 1037 struct buffer_head *buffer;
1038 struct f2fs_super_block *super;
1039 int err = 0;
14d7e9de 1040
9076a75f 1041retry:
da554e48 1042 buffer = sb_bread(sb, block);
1043 if (!buffer) {
1044 *recovery = 1;
9076a75f
GZ
1045 f2fs_msg(sb, KERN_ERR, "Unable to read %dth superblock",
1046 block + 1);
1047 if (block == 0) {
1048 block++;
1049 goto retry;
1050 } else {
da554e48 1051 err = -EIO;
1052 goto out;
9076a75f 1053 }
14d7e9de 1054 }
1055
da554e48 1056 super = (struct f2fs_super_block *)
1057 ((char *)(buffer)->b_data + F2FS_SUPER_OFFSET);
14d7e9de 1058
1059 /* sanity checking of raw super */
da554e48 1060 if (sanity_check_raw_super(sb, super)) {
1061 brelse(buffer);
1062 *recovery = 1;
6c311ec6
CF
1063 f2fs_msg(sb, KERN_ERR,
1064 "Can't find valid F2FS filesystem in %dth superblock",
1065 block + 1);
6bacf52f 1066 if (block == 0) {
9076a75f
GZ
1067 block++;
1068 goto retry;
1069 } else {
da554e48 1070 err = -EINVAL;
1071 goto out;
9076a75f
GZ
1072 }
1073 }
14d7e9de 1074
da554e48 1075 if (!*raw_super) {
1076 *raw_super_buf = buffer;
1077 *raw_super = super;
1078 } else {
1079 /* already have a valid superblock */
1080 brelse(buffer);
1081 }
1082
1083 /* check the validity of the second superblock */
1084 if (block == 0) {
1085 block++;
1086 goto retry;
1087 }
1088
1089out:
1090 /* No valid superblock */
1091 if (!*raw_super)
1092 return err;
1093
9076a75f 1094 return 0;
14d7e9de 1095}
1096
c5bda1c8 1097int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
26d815ad
JK
1098{
1099 struct buffer_head *sbh = sbi->raw_super_buf;
1100 sector_t block = sbh->b_blocknr;
1101 int err;
1102
1103 /* write back-up superblock first */
1104 sbh->b_blocknr = block ? 0 : 1;
1105 mark_buffer_dirty(sbh);
1106 err = sync_dirty_buffer(sbh);
1107
1108 sbh->b_blocknr = block;
c5bda1c8
CY
1109
1110 /* if we are in recovery path, skip writing valid superblock */
1111 if (recover || err)
26d815ad
JK
1112 goto out;
1113
1114 /* write current valid superblock */
1115 mark_buffer_dirty(sbh);
1116 err = sync_dirty_buffer(sbh);
1117out:
1118 clear_buffer_write_io_error(sbh);
1119 set_buffer_uptodate(sbh);
1120 return err;
1121}
1122
aff063e2
JK
1123static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
1124{
1125 struct f2fs_sb_info *sbi;
da554e48 1126 struct f2fs_super_block *raw_super;
aff063e2
JK
1127 struct buffer_head *raw_super_buf;
1128 struct inode *root;
da554e48 1129 long err;
2adc3505 1130 bool retry = true, need_fsck = false;
dabc4a5c 1131 char *options = NULL;
da554e48 1132 int recovery, i;
aff063e2 1133
ed2e621a 1134try_onemore:
da554e48 1135 err = -EINVAL;
1136 raw_super = NULL;
1137 raw_super_buf = NULL;
1138 recovery = 0;
1139
aff063e2
JK
1140 /* allocate memory for f2fs-specific super block info */
1141 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
1142 if (!sbi)
1143 return -ENOMEM;
1144
ff9234ad 1145 /* set a block size */
6bacf52f 1146 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
a07ef784 1147 f2fs_msg(sb, KERN_ERR, "unable to set blocksize");
aff063e2 1148 goto free_sbi;
a07ef784 1149 }
aff063e2 1150
da554e48 1151 err = read_raw_super_block(sb, &raw_super, &raw_super_buf, &recovery);
9076a75f
GZ
1152 if (err)
1153 goto free_sbi;
1154
5fb08372 1155 sb->s_fs_info = sbi;
498c5e9f 1156 default_options(sbi);
aff063e2 1157 /* parse mount options */
dabc4a5c
JK
1158 options = kstrdup((const char *)data, GFP_KERNEL);
1159 if (data && !options) {
1160 err = -ENOMEM;
aff063e2 1161 goto free_sb_buf;
dabc4a5c
JK
1162 }
1163
1164 err = parse_options(sb, options);
1165 if (err)
1166 goto free_options;
aff063e2 1167
25ca923b 1168 sb->s_maxbytes = max_file_size(le32_to_cpu(raw_super->log_blocksize));
aff063e2
JK
1169 sb->s_max_links = F2FS_LINK_MAX;
1170 get_random_bytes(&sbi->s_next_generation, sizeof(u32));
1171
1172 sb->s_op = &f2fs_sops;
1173 sb->s_xattr = f2fs_xattr_handlers;
1174 sb->s_export_op = &f2fs_export_ops;
1175 sb->s_magic = F2FS_SUPER_MAGIC;
aff063e2
JK
1176 sb->s_time_gran = 1;
1177 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
1178 (test_opt(sbi, POSIX_ACL) ? MS_POSIXACL : 0);
1179 memcpy(sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
1180
1181 /* init f2fs-specific super block info */
1182 sbi->sb = sb;
1183 sbi->raw_super = raw_super;
1184 sbi->raw_super_buf = raw_super_buf;
1185 mutex_init(&sbi->gc_mutex);
5463e7c1 1186 mutex_init(&sbi->writepages);
aff063e2 1187 mutex_init(&sbi->cp_mutex);
b3582c68 1188 init_rwsem(&sbi->node_write);
315df839
JK
1189
1190 /* disallow all the data/node/meta page writes */
1191 set_sbi_flag(sbi, SBI_POR_DOING);
aff063e2 1192 spin_lock_init(&sbi->stat_lock);
971767ca 1193
df0f8dc0 1194 init_rwsem(&sbi->read_io.io_rwsem);
458e6197
JK
1195 sbi->read_io.sbi = sbi;
1196 sbi->read_io.bio = NULL;
1197 for (i = 0; i < NR_PAGE_TYPE; i++) {
df0f8dc0 1198 init_rwsem(&sbi->write_io[i].io_rwsem);
458e6197
JK
1199 sbi->write_io[i].sbi = sbi;
1200 sbi->write_io[i].bio = NULL;
1201 }
971767ca 1202
e479556b 1203 init_rwsem(&sbi->cp_rwsem);
fb51b5ef 1204 init_waitqueue_head(&sbi->cp_wait);
aff063e2
JK
1205 init_sb_info(sbi);
1206
1207 /* get an inode for meta space */
1208 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
1209 if (IS_ERR(sbi->meta_inode)) {
a07ef784 1210 f2fs_msg(sb, KERN_ERR, "Failed to read F2FS meta data inode");
aff063e2 1211 err = PTR_ERR(sbi->meta_inode);
dabc4a5c 1212 goto free_options;
aff063e2
JK
1213 }
1214
1215 err = get_valid_checkpoint(sbi);
a07ef784
NJ
1216 if (err) {
1217 f2fs_msg(sb, KERN_ERR, "Failed to get valid F2FS checkpoint");
aff063e2 1218 goto free_meta_inode;
a07ef784 1219 }
aff063e2
JK
1220
1221 /* sanity checking of checkpoint */
1222 err = -EINVAL;
577e3495 1223 if (sanity_check_ckpt(sbi)) {
a07ef784 1224 f2fs_msg(sb, KERN_ERR, "Invalid F2FS checkpoint");
aff063e2 1225 goto free_cp;
a07ef784 1226 }
aff063e2
JK
1227
1228 sbi->total_valid_node_count =
1229 le32_to_cpu(sbi->ckpt->valid_node_count);
1230 sbi->total_valid_inode_count =
1231 le32_to_cpu(sbi->ckpt->valid_inode_count);
1232 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
1233 sbi->total_valid_block_count =
1234 le64_to_cpu(sbi->ckpt->valid_block_count);
1235 sbi->last_valid_block_count = sbi->total_valid_block_count;
1236 sbi->alloc_valid_block_count = 0;
1237 INIT_LIST_HEAD(&sbi->dir_inode_list);
1238 spin_lock_init(&sbi->dir_inode_lock);
1239
1dcc336b
CY
1240 init_extent_cache_info(sbi);
1241
6451e041 1242 init_ino_entry_info(sbi);
aff063e2
JK
1243
1244 /* setup f2fs internal modules */
1245 err = build_segment_manager(sbi);
a07ef784
NJ
1246 if (err) {
1247 f2fs_msg(sb, KERN_ERR,
1248 "Failed to initialize F2FS segment manager");
aff063e2 1249 goto free_sm;
a07ef784 1250 }
aff063e2 1251 err = build_node_manager(sbi);
a07ef784
NJ
1252 if (err) {
1253 f2fs_msg(sb, KERN_ERR,
1254 "Failed to initialize F2FS node manager");
aff063e2 1255 goto free_nm;
a07ef784 1256 }
aff063e2
JK
1257
1258 build_gc_manager(sbi);
1259
1260 /* get an inode for node space */
1261 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
1262 if (IS_ERR(sbi->node_inode)) {
a07ef784 1263 f2fs_msg(sb, KERN_ERR, "Failed to read node inode");
aff063e2
JK
1264 err = PTR_ERR(sbi->node_inode);
1265 goto free_nm;
1266 }
1267
2658e50d
JK
1268 f2fs_join_shrinker(sbi);
1269
aff063e2 1270 /* if there are nt orphan nodes free them */
8c14bfad
CY
1271 err = recover_orphan_inodes(sbi);
1272 if (err)
1273 goto free_node_inode;
aff063e2
JK
1274
1275 /* read root inode and dentry */
1276 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
1277 if (IS_ERR(root)) {
a07ef784 1278 f2fs_msg(sb, KERN_ERR, "Failed to read root inode");
aff063e2
JK
1279 err = PTR_ERR(root);
1280 goto free_node_inode;
1281 }
8f99a946 1282 if (!S_ISDIR(root->i_mode) || !root->i_blocks || !root->i_size) {
9d847950 1283 iput(root);
8f99a946 1284 err = -EINVAL;
9d847950 1285 goto free_node_inode;
8f99a946 1286 }
aff063e2
JK
1287
1288 sb->s_root = d_make_root(root); /* allocate root dentry */
1289 if (!sb->s_root) {
1290 err = -ENOMEM;
1291 goto free_root_inode;
1292 }
1293
aff063e2
JK
1294 err = f2fs_build_stats(sbi);
1295 if (err)
6437d1b0 1296 goto free_root_inode;
aff063e2 1297
5e176d54
JK
1298 if (f2fs_proc_root)
1299 sbi->s_proc = proc_mkdir(sb->s_id, f2fs_proc_root);
1300
1301 if (sbi->s_proc)
1302 proc_create_data("segment_info", S_IRUGO, sbi->s_proc,
1303 &f2fs_seq_segment_info_fops, sb);
1304
b59d0bae
NJ
1305 sbi->s_kobj.kset = f2fs_kset;
1306 init_completion(&sbi->s_kobj_unregister);
1307 err = kobject_init_and_add(&sbi->s_kobj, &f2fs_ktype, NULL,
1308 "%s", sb->s_id);
1309 if (err)
6437d1b0 1310 goto free_proc;
b59d0bae 1311
6437d1b0
JK
1312 /* recover fsynced data */
1313 if (!test_opt(sbi, DISABLE_ROLL_FORWARD)) {
081d78c2
JK
1314 /*
1315 * mount should be failed, when device has readonly mode, and
1316 * previous checkpoint was not done by clean system shutdown.
1317 */
1318 if (bdev_read_only(sb->s_bdev) &&
1319 !is_set_ckpt_flags(sbi->ckpt, CP_UMOUNT_FLAG)) {
1320 err = -EROFS;
1321 goto free_kobj;
1322 }
2adc3505
CY
1323
1324 if (need_fsck)
1325 set_sbi_flag(sbi, SBI_NEED_FSCK);
1326
6437d1b0 1327 err = recover_fsync_data(sbi);
ed2e621a 1328 if (err) {
2adc3505 1329 need_fsck = true;
6437d1b0
JK
1330 f2fs_msg(sb, KERN_ERR,
1331 "Cannot recover all fsync data errno=%ld", err);
ed2e621a
JK
1332 goto free_kobj;
1333 }
6437d1b0 1334 }
315df839
JK
1335 /* recover_fsync_data() cleared this already */
1336 clear_sbi_flag(sbi, SBI_POR_DOING);
b59d0bae 1337
6437d1b0
JK
1338 /*
1339 * If filesystem is not mounted as read-only then
1340 * do start the gc_thread.
1341 */
6c029932 1342 if (test_opt(sbi, BG_GC) && !f2fs_readonly(sb)) {
6437d1b0
JK
1343 /* After POR, we can run background GC thread.*/
1344 err = start_gc_thread(sbi);
1345 if (err)
1346 goto free_kobj;
1347 }
dabc4a5c 1348 kfree(options);
da554e48 1349
1350 /* recover broken superblock */
1351 if (recovery && !f2fs_readonly(sb) && !bdev_read_only(sb->s_bdev)) {
1352 f2fs_msg(sb, KERN_INFO, "Recover invalid superblock");
c5bda1c8 1353 f2fs_commit_super(sbi, true);
da554e48 1354 }
1355
60b99b48
JK
1356 sbi->cp_expires = round_jiffies_up(jiffies);
1357
aff063e2 1358 return 0;
6437d1b0
JK
1359
1360free_kobj:
1361 kobject_del(&sbi->s_kobj);
1362free_proc:
1d15bd20
CY
1363 if (sbi->s_proc) {
1364 remove_proc_entry("segment_info", sbi->s_proc);
1365 remove_proc_entry(sb->s_id, f2fs_proc_root);
1366 }
1367 f2fs_destroy_stats(sbi);
aff063e2
JK
1368free_root_inode:
1369 dput(sb->s_root);
1370 sb->s_root = NULL;
1371free_node_inode:
2658e50d
JK
1372 mutex_lock(&sbi->umount_mutex);
1373 f2fs_leave_shrinker(sbi);
aff063e2 1374 iput(sbi->node_inode);
2658e50d 1375 mutex_unlock(&sbi->umount_mutex);
aff063e2
JK
1376free_nm:
1377 destroy_node_manager(sbi);
1378free_sm:
1379 destroy_segment_manager(sbi);
1380free_cp:
1381 kfree(sbi->ckpt);
1382free_meta_inode:
1383 make_bad_inode(sbi->meta_inode);
1384 iput(sbi->meta_inode);
dabc4a5c
JK
1385free_options:
1386 kfree(options);
aff063e2
JK
1387free_sb_buf:
1388 brelse(raw_super_buf);
1389free_sbi:
1390 kfree(sbi);
ed2e621a
JK
1391
1392 /* give only one another chance */
1393 if (retry) {
9df47ba7 1394 retry = false;
ed2e621a
JK
1395 shrink_dcache_sb(sb);
1396 goto try_onemore;
1397 }
aff063e2
JK
1398 return err;
1399}
1400
1401static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
1402 const char *dev_name, void *data)
1403{
1404 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
1405}
1406
30a5537f
JK
1407static void kill_f2fs_super(struct super_block *sb)
1408{
1409 if (sb->s_root)
caf0047e 1410 set_sbi_flag(F2FS_SB(sb), SBI_IS_CLOSE);
30a5537f
JK
1411 kill_block_super(sb);
1412}
1413
aff063e2
JK
1414static struct file_system_type f2fs_fs_type = {
1415 .owner = THIS_MODULE,
1416 .name = "f2fs",
1417 .mount = f2fs_mount,
30a5537f 1418 .kill_sb = kill_f2fs_super,
aff063e2
JK
1419 .fs_flags = FS_REQUIRES_DEV,
1420};
7f78e035 1421MODULE_ALIAS_FS("f2fs");
aff063e2 1422
6e6093a8 1423static int __init init_inodecache(void)
aff063e2
JK
1424{
1425 f2fs_inode_cachep = f2fs_kmem_cache_create("f2fs_inode_cache",
e8512d2e 1426 sizeof(struct f2fs_inode_info));
6bacf52f 1427 if (!f2fs_inode_cachep)
aff063e2
JK
1428 return -ENOMEM;
1429 return 0;
1430}
1431
1432static void destroy_inodecache(void)
1433{
1434 /*
1435 * Make sure all delayed rcu free inodes are flushed before we
1436 * destroy cache.
1437 */
1438 rcu_barrier();
1439 kmem_cache_destroy(f2fs_inode_cachep);
1440}
1441
1442static int __init init_f2fs_fs(void)
1443{
1444 int err;
1445
c0508650
JK
1446 f2fs_build_trace_ios();
1447
aff063e2
JK
1448 err = init_inodecache();
1449 if (err)
1450 goto fail;
1451 err = create_node_manager_caches();
1452 if (err)
9890ff3f 1453 goto free_inodecache;
7fd9e544 1454 err = create_segment_manager_caches();
aff063e2 1455 if (err)
9890ff3f 1456 goto free_node_manager_caches;
aff063e2
JK
1457 err = create_checkpoint_caches();
1458 if (err)
06292073 1459 goto free_segment_manager_caches;
1dcc336b
CY
1460 err = create_extent_cache();
1461 if (err)
1462 goto free_checkpoint_caches;
b59d0bae 1463 f2fs_kset = kset_create_and_add("f2fs", NULL, fs_kobj);
6e6b978c
WY
1464 if (!f2fs_kset) {
1465 err = -ENOMEM;
1dcc336b 1466 goto free_extent_cache;
6e6b978c 1467 }
cfc4d971 1468 err = f2fs_init_crypto();
4589d25d 1469 if (err)
9890ff3f 1470 goto free_kset;
2658e50d
JK
1471
1472 err = register_shrinker(&f2fs_shrinker_info);
cfc4d971
JK
1473 if (err)
1474 goto free_crypto;
2658e50d
JK
1475
1476 err = register_filesystem(&f2fs_fs_type);
1477 if (err)
1478 goto free_shrinker;
4589d25d 1479 f2fs_create_root_stats();
5e176d54 1480 f2fs_proc_root = proc_mkdir("fs/f2fs", NULL);
9890ff3f
ZH
1481 return 0;
1482
2658e50d
JK
1483free_shrinker:
1484 unregister_shrinker(&f2fs_shrinker_info);
cfc4d971
JK
1485free_crypto:
1486 f2fs_exit_crypto();
9890ff3f
ZH
1487free_kset:
1488 kset_unregister(f2fs_kset);
1dcc336b
CY
1489free_extent_cache:
1490 destroy_extent_cache();
9890ff3f
ZH
1491free_checkpoint_caches:
1492 destroy_checkpoint_caches();
7fd9e544
JK
1493free_segment_manager_caches:
1494 destroy_segment_manager_caches();
9890ff3f
ZH
1495free_node_manager_caches:
1496 destroy_node_manager_caches();
1497free_inodecache:
1498 destroy_inodecache();
aff063e2
JK
1499fail:
1500 return err;
1501}
1502
1503static void __exit exit_f2fs_fs(void)
1504{
5e176d54 1505 remove_proc_entry("fs/f2fs", NULL);
4589d25d 1506 f2fs_destroy_root_stats();
2658e50d 1507 unregister_shrinker(&f2fs_shrinker_info);
aff063e2 1508 unregister_filesystem(&f2fs_fs_type);
cfc4d971 1509 f2fs_exit_crypto();
fdf6c8be 1510 destroy_extent_cache();
aff063e2 1511 destroy_checkpoint_caches();
5dcd8a71 1512 destroy_segment_manager_caches();
aff063e2
JK
1513 destroy_node_manager_caches();
1514 destroy_inodecache();
b59d0bae 1515 kset_unregister(f2fs_kset);
351f4fba 1516 f2fs_destroy_trace_ios();
aff063e2
JK
1517}
1518
1519module_init(init_f2fs_fs)
1520module_exit(exit_f2fs_fs)
1521
1522MODULE_AUTHOR("Samsung Electronics's Praesto Team");
1523MODULE_DESCRIPTION("Flash Friendly File System");
1524MODULE_LICENSE("GPL");