projects / linux-2.6-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
f2fs: avoid panic when truncating to max filesize
[linux-2.6-block.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
73faec4d
JK		 42#ifdef CONFIG_F2FS_FAULT_INJECTION
43u32 f2fs_fault_rate = 0;
2c63fead
JK		 44atomic_t f2fs_ops;
45
46char *fault_name[FAULT_MAX] = {
47 [FAULT_KMALLOC] = "kmalloc",
c41f3cc3 48 [FAULT_PAGE_ALLOC] = "page alloc",
cb78942b
JK		 49 [FAULT_ALLOC_NID] = "alloc nid",
50 [FAULT_ORPHAN] = "orphan",
51 [FAULT_BLOCK] = "no more block",
52 [FAULT_DIR_DEPTH] = "too big dir depth",
2c63fead 53};
73faec4d
JK		 54#endif
55
2658e50d
JK		 56/* f2fs-wide shrinker description */
57static struct shrinker f2fs_shrinker_info = {
58 .scan_objects = f2fs_shrink_scan,
59 .count_objects = f2fs_shrink_count,
60 .seeks = DEFAULT_SEEKS,
61};
62
aff063e2 63enum {
696c018c 64 Opt_gc_background,
aff063e2 65 Opt_disable_roll_forward,
2d834bf9 66 Opt_norecovery,
aff063e2
JK		 67 Opt_discard,
68 Opt_noheap,
4058c511 69 Opt_user_xattr,
aff063e2 70 Opt_nouser_xattr,
4058c511 71 Opt_acl,
aff063e2
JK		 72 Opt_noacl,
73 Opt_active_logs,
74 Opt_disable_ext_identify,
444c580f 75 Opt_inline_xattr,
8274de77 76 Opt_inline_data,
5efd3c6f 77 Opt_inline_dentry,
6b4afdd7 78 Opt_flush_merge,
0f7b2abd 79 Opt_nobarrier,
d5053a34 80 Opt_fastboot,
89672159 81 Opt_extent_cache,
7daaea25 82 Opt_noextent_cache,
75342797 83 Opt_noinline_data,
343f40f0 84 Opt_data_flush,
73faec4d 85 Opt_fault_injection,
aff063e2
JK		 86 Opt_err,
87};
88
89static match_table_t f2fs_tokens = {
696c018c 90 {Opt_gc_background, "background_gc=%s"},
aff063e2 91 {Opt_disable_roll_forward, "disable_roll_forward"},
2d834bf9 92 {Opt_norecovery, "norecovery"},
aff063e2
JK		 93 {Opt_discard, "discard"},
94 {Opt_noheap, "no_heap"},
4058c511 95 {Opt_user_xattr, "user_xattr"},
aff063e2 96 {Opt_nouser_xattr, "nouser_xattr"},
4058c511 97 {Opt_acl, "acl"},
aff063e2
JK		 98 {Opt_noacl, "noacl"},
99 {Opt_active_logs, "active_logs=%u"},
100 {Opt_disable_ext_identify, "disable_ext_identify"},
444c580f 101 {Opt_inline_xattr, "inline_xattr"},
8274de77 102 {Opt_inline_data, "inline_data"},
5efd3c6f 103 {Opt_inline_dentry, "inline_dentry"},
6b4afdd7 104 {Opt_flush_merge, "flush_merge"},
0f7b2abd 105 {Opt_nobarrier, "nobarrier"},
d5053a34 106 {Opt_fastboot, "fastboot"},
89672159 107 {Opt_extent_cache, "extent_cache"},
7daaea25 108 {Opt_noextent_cache, "noextent_cache"},
75342797 109 {Opt_noinline_data, "noinline_data"},
343f40f0 110 {Opt_data_flush, "data_flush"},
73faec4d 111 {Opt_fault_injection, "fault_injection=%u"},
aff063e2
JK		 112 {Opt_err, NULL},
113};
114
b59d0bae 115/* Sysfs support for f2fs */
ea91e9b0
JK		 116enum {
117 GC_THREAD, /* struct f2fs_gc_thread */
118 SM_INFO, /* struct f2fs_sm_info */
cdfc41c1 119 NM_INFO, /* struct f2fs_nm_info */
b1c57c1c 120 F2FS_SBI, /* struct f2fs_sb_info */
ea91e9b0
JK		 121};
122
b59d0bae
NJ		 123struct f2fs_attr {
124 struct attribute attr;
125 ssize_t (*show)(struct f2fs_attr *, struct f2fs_sb_info *, char *);
126 ssize_t (*store)(struct f2fs_attr *, struct f2fs_sb_info *,
127 const char *, size_t);
ea91e9b0 128 int struct_type;
b59d0bae
NJ		 129 int offset;
130};
131
ea91e9b0
JK		 132static unsigned char *__struct_ptr(struct f2fs_sb_info *sbi, int struct_type)
133{
134 if (struct_type == GC_THREAD)
135 return (unsigned char *)sbi->gc_thread;
136 else if (struct_type == SM_INFO)
137 return (unsigned char *)SM_I(sbi);
cdfc41c1
JK		 138 else if (struct_type == NM_INFO)
139 return (unsigned char *)NM_I(sbi);
b1c57c1c
JK		 140 else if (struct_type == F2FS_SBI)
141 return (unsigned char *)sbi;
ea91e9b0
JK		 142 return NULL;
143}
144
8f1dbbbb
SL		 145static ssize_t lifetime_write_kbytes_show(struct f2fs_attr *a,
146 struct f2fs_sb_info *sbi, char *buf)
147{
148 struct super_block *sb = sbi->sb;
149
150 if (!sb->s_bdev->bd_part)
151 return snprintf(buf, PAGE_SIZE, "0\n");
152
153 return snprintf(buf, PAGE_SIZE, "%llu\n",
154 (unsigned long long)(sbi->kbytes_written +
155 BD_PART_WRITTEN(sbi)));
156}
157
b59d0bae
NJ		 158static ssize_t f2fs_sbi_show(struct f2fs_attr *a,
159 struct f2fs_sb_info *sbi, char *buf)
160{
ea91e9b0 161 unsigned char *ptr = NULL;
b59d0bae
NJ		 162 unsigned int *ui;
163
ea91e9b0
JK		 164 ptr = __struct_ptr(sbi, a->struct_type);
165 if (!ptr)
b59d0bae
NJ		 166 return -EINVAL;
167
ea91e9b0 168 ui = (unsigned int *)(ptr + a->offset);
b59d0bae
NJ		 169
170 return snprintf(buf, PAGE_SIZE, "%u\n", *ui);
171}
172
173static ssize_t f2fs_sbi_store(struct f2fs_attr *a,
174 struct f2fs_sb_info *sbi,
175 const char *buf, size_t count)
176{
ea91e9b0 177 unsigned char *ptr;
b59d0bae
NJ		 178 unsigned long t;
179 unsigned int *ui;
180 ssize_t ret;
181
ea91e9b0
JK		 182 ptr = __struct_ptr(sbi, a->struct_type);
183 if (!ptr)
b59d0bae
NJ		 184 return -EINVAL;
185
ea91e9b0 186 ui = (unsigned int *)(ptr + a->offset);
b59d0bae
NJ		 187
188 ret = kstrtoul(skip_spaces(buf), 0, &t);
189 if (ret < 0)
190 return ret;
191 *ui = t;
192 return count;
193}
194
195static ssize_t f2fs_attr_show(struct kobject *kobj,
196 struct attribute *attr, char *buf)
197{
198 struct f2fs_sb_info *sbi = container_of(kobj, struct f2fs_sb_info,
199 s_kobj);
200 struct f2fs_attr *a = container_of(attr, struct f2fs_attr, attr);
201
202 return a->show ? a->show(a, sbi, buf) : 0;
203}
204
205static ssize_t f2fs_attr_store(struct kobject *kobj, struct attribute *attr,
206 const char *buf, size_t len)
207{
208 struct f2fs_sb_info *sbi = container_of(kobj, struct f2fs_sb_info,
209 s_kobj);
210 struct f2fs_attr *a = container_of(attr, struct f2fs_attr, attr);
211
212 return a->store ? a->store(a, sbi, buf, len) : 0;
213}
214
215static void f2fs_sb_release(struct kobject *kobj)
216{
217 struct f2fs_sb_info *sbi = container_of(kobj, struct f2fs_sb_info,
218 s_kobj);
219 complete(&sbi->s_kobj_unregister);
220}
221
ea91e9b0 222#define F2FS_ATTR_OFFSET(_struct_type, _name, _mode, _show, _store, _offset) \
b59d0bae
NJ		 223static struct f2fs_attr f2fs_attr_##_name = { \
224 .attr = {.name = __stringify(_name), .mode = _mode }, \
225 .show = _show, \
226 .store = _store, \
ea91e9b0
JK		 227 .struct_type = _struct_type, \
228 .offset = _offset \
b59d0bae
NJ		 229}
230
ea91e9b0
JK		 231#define F2FS_RW_ATTR(struct_type, struct_name, name, elname) \
232 F2FS_ATTR_OFFSET(struct_type, name, 0644, \
233 f2fs_sbi_show, f2fs_sbi_store, \
234 offsetof(struct struct_name, elname))
b59d0bae 235
8f1dbbbb
SL		 236#define F2FS_GENERAL_RO_ATTR(name) \
237static struct f2fs_attr f2fs_attr_##name = __ATTR(name, 0444, name##_show, NULL)
238
ea91e9b0
JK		 239F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_min_sleep_time, min_sleep_time);
240F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_max_sleep_time, max_sleep_time);
241F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_no_gc_sleep_time, no_gc_sleep_time);
242F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_idle, gc_idle);
243F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, reclaim_segments, rec_prefree_segments);
7ac8c3b0 244F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, max_small_discards, max_discards);
bba681cb 245F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, batched_trim_sections, trim_sections);
216fbd64
JK		 246F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, ipu_policy, ipu_policy);
247F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, min_ipu_util, min_ipu_util);
c1ce1b02 248F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, min_fsync_blocks, min_fsync_blocks);
cdfc41c1 249F2FS_RW_ATTR(NM_INFO, f2fs_nm_info, ram_thresh, ram_thresh);
ea1a29a0 250F2FS_RW_ATTR(NM_INFO, f2fs_nm_info, ra_nid_pages, ra_nid_pages);
2304cb0c 251F2FS_RW_ATTR(NM_INFO, f2fs_nm_info, dirty_nats_ratio, dirty_nats_ratio);
b1c57c1c 252F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, max_victim_search, max_victim_search);
ab9fa662 253F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, dir_level, dir_level);
6beceb54 254F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, cp_interval, interval_time[CP_TIME]);
d0239e1b 255F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, idle_interval, interval_time[REQ_TIME]);
8f1dbbbb 256F2FS_GENERAL_RO_ATTR(lifetime_write_kbytes);
b59d0bae
NJ		 257
258#define ATTR_LIST(name) (&f2fs_attr_##name.attr)
259static struct attribute *f2fs_attrs[] = {
260 ATTR_LIST(gc_min_sleep_time),
261 ATTR_LIST(gc_max_sleep_time),
262 ATTR_LIST(gc_no_gc_sleep_time),
d2dc095f 263 ATTR_LIST(gc_idle),
ea91e9b0 264 ATTR_LIST(reclaim_segments),
7ac8c3b0 265 ATTR_LIST(max_small_discards),
bba681cb 266 ATTR_LIST(batched_trim_sections),
216fbd64
JK		 267 ATTR_LIST(ipu_policy),
268 ATTR_LIST(min_ipu_util),
c1ce1b02 269 ATTR_LIST(min_fsync_blocks),
b1c57c1c 270 ATTR_LIST(max_victim_search),
ab9fa662 271 ATTR_LIST(dir_level),
cdfc41c1 272 ATTR_LIST(ram_thresh),
ea1a29a0 273 ATTR_LIST(ra_nid_pages),
2304cb0c 274 ATTR_LIST(dirty_nats_ratio),
60b99b48 275 ATTR_LIST(cp_interval),
d0239e1b 276 ATTR_LIST(idle_interval),
8f1dbbbb 277 ATTR_LIST(lifetime_write_kbytes),
b59d0bae
NJ		 278 NULL,
279};
280
281static const struct sysfs_ops f2fs_attr_ops = {
282 .show = f2fs_attr_show,
283 .store = f2fs_attr_store,
284};
285
286static struct kobj_type f2fs_ktype = {
287 .default_attrs = f2fs_attrs,
288 .sysfs_ops = &f2fs_attr_ops,
289 .release = f2fs_sb_release,
290};
291
a07ef784
NJ		 292void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...)
293{
294 struct va_format vaf;
295 va_list args;
296
297 va_start(args, fmt);
298 vaf.fmt = fmt;
299 vaf.va = &args;
300 printk("%sF2FS-fs (%s): %pV\n", level, sb->s_id, &vaf);
301 va_end(args);
302}
303
aff063e2
JK		 304static void init_once(void *foo)
305{
306 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
307
aff063e2
JK		 308 inode_init_once(&fi->vfs_inode);
309}
310
696c018c
NJ		 311static int parse_options(struct super_block *sb, char *options)
312{
313 struct f2fs_sb_info *sbi = F2FS_SB(sb);
09d54cdd 314 struct request_queue *q;
696c018c
NJ		 315 substring_t args[MAX_OPT_ARGS];
316 char *p, *name;
317 int arg = 0;
318
73faec4d
JK		 319#ifdef CONFIG_F2FS_FAULT_INJECTION
320 f2fs_fault_rate = 0;
321#endif
696c018c
NJ		 322 if (!options)
323 return 0;
324
325 while ((p = strsep(&options, ",")) != NULL) {
326 int token;
327 if (!*p)
328 continue;
329 /*
330 * Initialize args struct so we know whether arg was
331 * found; some options take optional arguments.
332 */
333 args[0].to = args[0].from = NULL;
334 token = match_token(p, f2fs_tokens, args);
335
336 switch (token) {
337 case Opt_gc_background:
338 name = match_strdup(&args[0]);
339
340 if (!name)
341 return -ENOMEM;
6aefd93b 342 if (strlen(name) == 2 && !strncmp(name, "on", 2)) {
696c018c 343 set_opt(sbi, BG_GC);
6aefd93b
JK		 344 clear_opt(sbi, FORCE_FG_GC);
345 } else if (strlen(name) == 3 && !strncmp(name, "off", 3)) {
696c018c 346 clear_opt(sbi, BG_GC);
6aefd93b
JK		 347 clear_opt(sbi, FORCE_FG_GC);
348 } else if (strlen(name) == 4 && !strncmp(name, "sync", 4)) {
349 set_opt(sbi, BG_GC);
350 set_opt(sbi, FORCE_FG_GC);
351 } else {
696c018c
NJ		 352 kfree(name);
353 return -EINVAL;
354 }
355 kfree(name);
356 break;
357 case Opt_disable_roll_forward:
358 set_opt(sbi, DISABLE_ROLL_FORWARD);
359 break;
2d834bf9
JK		 360 case Opt_norecovery:
361 /* this option mounts f2fs with ro */
362 set_opt(sbi, DISABLE_ROLL_FORWARD);
363 if (!f2fs_readonly(sb))
364 return -EINVAL;
365 break;
696c018c 366 case Opt_discard:
09d54cdd
CY		 367 q = bdev_get_queue(sb->s_bdev);
368 if (blk_queue_discard(q)) {
369 set_opt(sbi, DISCARD);
370 } else {
371 f2fs_msg(sb, KERN_WARNING,
372 "mounting with \"discard\" option, but "
373 "the device does not support discard");
374 }
696c018c
NJ		 375 break;
376 case Opt_noheap:
377 set_opt(sbi, NOHEAP);
378 break;
379#ifdef CONFIG_F2FS_FS_XATTR
4058c511
KA		 380 case Opt_user_xattr:
381 set_opt(sbi, XATTR_USER);
382 break;
696c018c
NJ		 383 case Opt_nouser_xattr:
384 clear_opt(sbi, XATTR_USER);
385 break;
444c580f
JK		 386 case Opt_inline_xattr:
387 set_opt(sbi, INLINE_XATTR);
388 break;
696c018c 389#else
4058c511
KA		 390 case Opt_user_xattr:
391 f2fs_msg(sb, KERN_INFO,
392 "user_xattr options not supported");
393 break;
696c018c
NJ		 394 case Opt_nouser_xattr:
395 f2fs_msg(sb, KERN_INFO,
396 "nouser_xattr options not supported");
397 break;
444c580f
JK		 398 case Opt_inline_xattr:
399 f2fs_msg(sb, KERN_INFO,
400 "inline_xattr options not supported");
401 break;
696c018c
NJ		 402#endif
403#ifdef CONFIG_F2FS_FS_POSIX_ACL
4058c511
KA		 404 case Opt_acl:
405 set_opt(sbi, POSIX_ACL);
406 break;
696c018c
NJ		 407 case Opt_noacl:
408 clear_opt(sbi, POSIX_ACL);
409 break;
410#else
4058c511
KA		 411 case Opt_acl:
412 f2fs_msg(sb, KERN_INFO, "acl options not supported");
413 break;
696c018c
NJ		 414 case Opt_noacl:
415 f2fs_msg(sb, KERN_INFO, "noacl options not supported");
416 break;
417#endif
418 case Opt_active_logs:
419 if (args->from && match_int(args, &arg))
420 return -EINVAL;
421 if (arg != 2 && arg != 4 && arg != NR_CURSEG_TYPE)
422 return -EINVAL;
423 sbi->active_logs = arg;
424 break;
425 case Opt_disable_ext_identify:
426 set_opt(sbi, DISABLE_EXT_IDENTIFY);
427 break;
8274de77
HL		 428 case Opt_inline_data:
429 set_opt(sbi, INLINE_DATA);
430 break;
5efd3c6f
CY		 431 case Opt_inline_dentry:
432 set_opt(sbi, INLINE_DENTRY);
433 break;
6b4afdd7
JK		 434 case Opt_flush_merge:
435 set_opt(sbi, FLUSH_MERGE);
436 break;
0f7b2abd
JK		 437 case Opt_nobarrier:
438 set_opt(sbi, NOBARRIER);
439 break;
d5053a34
JK		 440 case Opt_fastboot:
441 set_opt(sbi, FASTBOOT);
442 break;
89672159
CY		 443 case Opt_extent_cache:
444 set_opt(sbi, EXTENT_CACHE);
445 break;
7daaea25
JK		 446 case Opt_noextent_cache:
447 clear_opt(sbi, EXTENT_CACHE);
448 break;
75342797
WL		 449 case Opt_noinline_data:
450 clear_opt(sbi, INLINE_DATA);
451 break;
343f40f0
CY		 452 case Opt_data_flush:
453 set_opt(sbi, DATA_FLUSH);
454 break;
73faec4d
JK		 455 case Opt_fault_injection:
456 if (args->from && match_int(args, &arg))
457 return -EINVAL;
458#ifdef CONFIG_F2FS_FAULT_INJECTION
459 f2fs_fault_rate = arg;
2c63fead 460 atomic_set(&f2fs_ops, 0);
73faec4d
JK		 461#else
462 f2fs_msg(sb, KERN_INFO,
463 "FAULT_INJECTION was not selected");
464#endif
465 break;
696c018c
NJ		 466 default:
467 f2fs_msg(sb, KERN_ERR,
468 "Unrecognized mount option \"%s\" or missing value",
469 p);
470 return -EINVAL;
471 }
472 }
473 return 0;
474}
475
aff063e2
JK		 476static struct inode *f2fs_alloc_inode(struct super_block *sb)
477{
478 struct f2fs_inode_info *fi;
479
a0acdfe0 480 fi = kmem_cache_alloc(f2fs_inode_cachep, GFP_F2FS_ZERO);
aff063e2
JK		 481 if (!fi)
482 return NULL;
483
484 init_once((void *) fi);
485
434720fa 486 /* Initialize f2fs-specific inode info */
aff063e2 487 fi->vfs_inode.i_version = 1;
a7ffdbe2 488 atomic_set(&fi->dirty_pages, 0);
aff063e2
JK		 489 fi->i_current_depth = 1;
490 fi->i_advise = 0;
d928bfbf 491 init_rwsem(&fi->i_sem);
2710fd7e 492 INIT_LIST_HEAD(&fi->dirty_list);
88b88a66
JK		 493 INIT_LIST_HEAD(&fi->inmem_pages);
494 mutex_init(&fi->inmem_lock);
aff063e2
JK		 495
496 set_inode_flag(fi, FI_NEW_INODE);
497
444c580f
JK		 498 if (test_opt(F2FS_SB(sb), INLINE_XATTR))
499 set_inode_flag(fi, FI_INLINE_XATTR);
500
ab9fa662
JK		 501 /* Will be used by directory only */
502 fi->i_dir_level = F2FS_SB(sb)->dir_level;
aff063e2
JK		 503 return &fi->vfs_inode;
504}
505
531ad7d5
JK		 506static int f2fs_drop_inode(struct inode *inode)
507{
508 /*
509 * This is to avoid a deadlock condition like below.
510 * writeback_single_inode(inode)
511 * - f2fs_write_data_page
512 * - f2fs_gc -> iput -> evict
513 * - inode_wait_for_writeback(inode)
514 */
06e1bc05
JK		 515 if (!inode_unhashed(inode) && inode->i_state & I_SYNC) {
516 if (!inode->i_nlink && !is_bad_inode(inode)) {
3e72f721
JK		 517 /* to avoid evict_inode call simultaneously */
518 atomic_inc(&inode->i_count);
06e1bc05
JK		 519 spin_unlock(&inode->i_lock);
520
521 /* some remained atomic pages should discarded */
522 if (f2fs_is_atomic_file(inode))
29b96b54 523 drop_inmem_pages(inode);
06e1bc05 524
3e72f721
JK		 525 /* should remain fi->extent_tree for writepage */
526 f2fs_destroy_extent_node(inode);
527
06e1bc05
JK		 528 sb_start_intwrite(inode->i_sb);
529 i_size_write(inode, 0);
530
531 if (F2FS_HAS_BLOCKS(inode))
55f57d2c 532 f2fs_truncate(inode, true);
06e1bc05
JK		 533
534 sb_end_intwrite(inode->i_sb);
535
0b81d077 536 fscrypt_put_encryption_info(inode, NULL);
06e1bc05 537 spin_lock(&inode->i_lock);
3e72f721 538 atomic_dec(&inode->i_count);
06e1bc05 539 }
531ad7d5 540 return 0;
06e1bc05 541 }
531ad7d5
JK		 542 return generic_drop_inode(inode);
543}
544
b3783873
JK		 545/*
546 * f2fs_dirty_inode() is called from __mark_inode_dirty()
547 *
548 * We should call set_dirty_inode to write the dirty inode through write_inode.
549 */
550static void f2fs_dirty_inode(struct inode *inode, int flags)
551{
552 set_inode_flag(F2FS_I(inode), FI_DIRTY_INODE);
b3783873
JK		 553}
554
aff063e2
JK		 555static void f2fs_i_callback(struct rcu_head *head)
556{
557 struct inode *inode = container_of(head, struct inode, i_rcu);
558 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
559}
560
25ca923b 561static void f2fs_destroy_inode(struct inode *inode)
aff063e2
JK		 562{
563 call_rcu(&inode->i_rcu, f2fs_i_callback);
564}
565
566static void f2fs_put_super(struct super_block *sb)
567{
568 struct f2fs_sb_info *sbi = F2FS_SB(sb);
569
5e176d54
JK		 570 if (sbi->s_proc) {
571 remove_proc_entry("segment_info", sbi->s_proc);
f00d6fa7 572 remove_proc_entry("segment_bits", sbi->s_proc);
5e176d54
JK		 573 remove_proc_entry(sb->s_id, f2fs_proc_root);
574 }
b59d0bae 575 kobject_del(&sbi->s_kobj);
5e176d54 576
aff063e2
JK		 577 stop_gc_thread(sbi);
578
2658e50d
JK		 579 /* prevent remaining shrinker jobs */
580 mutex_lock(&sbi->umount_mutex);
581
85dc2f2c
JK		 582 /*
583 * We don't need to do checkpoint when superblock is clean.
584 * But, the previous checkpoint was not done by umount, it needs to do
585 * clean checkpoint again.
586 */
caf0047e 587 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
85dc2f2c 588 !is_set_ckpt_flags(F2FS_CKPT(sbi), CP_UMOUNT_FLAG)) {
75ab4cb8
JK		 589 struct cp_control cpc = {
590 .reason = CP_UMOUNT,
591 };
592 write_checkpoint(sbi, &cpc);
593 }
aff063e2 594
eca616f8
JK		 595 /* write_checkpoint can update stat informaion */
596 f2fs_destroy_stats(sbi);
597
cf779cab
JK		 598 /*
599 * normally superblock is clean, so we need to release this.
600 * In addition, EIO will skip do checkpoint, we need this as well.
601 */
74ef9241 602 release_ino_entry(sbi, true);
4b2fecc8 603 release_discard_addrs(sbi);
6f12ac25 604
2658e50d
JK		 605 f2fs_leave_shrinker(sbi);
606 mutex_unlock(&sbi->umount_mutex);
607
17c19120 608 /* our cp_error case, we can wait for any writeback page */
406657dd
CY		 609 if (get_pages(sbi, F2FS_WRITEBACK))
610 f2fs_flush_merged_bios(sbi);
17c19120 611
aff063e2
JK		 612 iput(sbi->node_inode);
613 iput(sbi->meta_inode);
614
615 /* destroy f2fs internal modules */
616 destroy_node_manager(sbi);
617 destroy_segment_manager(sbi);
618
619 kfree(sbi->ckpt);
b59d0bae
NJ		 620 kobject_put(&sbi->s_kobj);
621 wait_for_completion(&sbi->s_kobj_unregister);
aff063e2
JK		 622
623 sb->s_fs_info = NULL;
43b6573b
KM		 624 if (sbi->s_chksum_driver)
625 crypto_free_shash(sbi->s_chksum_driver);
b39f0de2 626 kfree(sbi->raw_super);
aff063e2
JK		 627 kfree(sbi);
628}
629
630int f2fs_sync_fs(struct super_block *sb, int sync)
631{
632 struct f2fs_sb_info *sbi = F2FS_SB(sb);
c34f42e2 633 int err = 0;
aff063e2 634
a2a4a7e4
NJ		 635 trace_f2fs_sync_fs(sb, sync);
636
b7473754 637 if (sync) {
d5053a34
JK		 638 struct cp_control cpc;
639
119ee914
JK		 640 cpc.reason = __get_cp_reason(sbi);
641
b7473754 642 mutex_lock(&sbi->gc_mutex);
c34f42e2 643 err = write_checkpoint(sbi, &cpc);
b7473754 644 mutex_unlock(&sbi->gc_mutex);
b7473754 645 }
05ca3632 646 f2fs_trace_ios(NULL, 1);
aff063e2 647
c34f42e2 648 return err;
aff063e2
JK		 649}
650
d6212a5f
CL		 651static int f2fs_freeze(struct super_block *sb)
652{
653 int err;
654
77888c1e 655 if (f2fs_readonly(sb))
d6212a5f
CL		 656 return 0;
657
658 err = f2fs_sync_fs(sb, 1);
659 return err;
660}
661
662static int f2fs_unfreeze(struct super_block *sb)
663{
664 return 0;
665}
666
aff063e2
JK		 667static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
668{
669 struct super_block *sb = dentry->d_sb;
670 struct f2fs_sb_info *sbi = F2FS_SB(sb);
671 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
672 block_t total_count, user_block_count, start_count, ovp_count;
673
674 total_count = le64_to_cpu(sbi->raw_super->block_count);
675 user_block_count = sbi->user_block_count;
676 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
677 ovp_count = SM_I(sbi)->ovp_segments << sbi->log_blocks_per_seg;
678 buf->f_type = F2FS_SUPER_MAGIC;
679 buf->f_bsize = sbi->blocksize;
680
681 buf->f_blocks = total_count - start_count;
682 buf->f_bfree = buf->f_blocks - valid_user_blocks(sbi) - ovp_count;
683 buf->f_bavail = user_block_count - valid_user_blocks(sbi);
684
c200b1aa
CY		 685 buf->f_files = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
686 buf->f_ffree = buf->f_files - valid_inode_count(sbi);
aff063e2 687
5a20d339 688 buf->f_namelen = F2FS_NAME_LEN;
aff063e2
JK		 689 buf->f_fsid.val[0] = (u32)id;
690 buf->f_fsid.val[1] = (u32)(id >> 32);
691
692 return 0;
693}
694
695static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
696{
697 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
698
6aefd93b
JK		 699 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, BG_GC)) {
700 if (test_opt(sbi, FORCE_FG_GC))
701 seq_printf(seq, ",background_gc=%s", "sync");
702 else
703 seq_printf(seq, ",background_gc=%s", "on");
704 } else {
696c018c 705 seq_printf(seq, ",background_gc=%s", "off");
6aefd93b 706 }
aff063e2
JK		 707 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
708 seq_puts(seq, ",disable_roll_forward");
709 if (test_opt(sbi, DISCARD))
710 seq_puts(seq, ",discard");
711 if (test_opt(sbi, NOHEAP))
712 seq_puts(seq, ",no_heap_alloc");
713#ifdef CONFIG_F2FS_FS_XATTR
714 if (test_opt(sbi, XATTR_USER))
715 seq_puts(seq, ",user_xattr");
716 else
717 seq_puts(seq, ",nouser_xattr");
444c580f
JK		 718 if (test_opt(sbi, INLINE_XATTR))
719 seq_puts(seq, ",inline_xattr");
aff063e2
JK		 720#endif
721#ifdef CONFIG_F2FS_FS_POSIX_ACL
722 if (test_opt(sbi, POSIX_ACL))
723 seq_puts(seq, ",acl");
724 else
725 seq_puts(seq, ",noacl");
726#endif
727 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
aa43507f 728 seq_puts(seq, ",disable_ext_identify");
8274de77
HL		 729 if (test_opt(sbi, INLINE_DATA))
730 seq_puts(seq, ",inline_data");
75342797
WL		 731 else
732 seq_puts(seq, ",noinline_data");
5efd3c6f
CY		 733 if (test_opt(sbi, INLINE_DENTRY))
734 seq_puts(seq, ",inline_dentry");
b270ad6f 735 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
6b4afdd7 736 seq_puts(seq, ",flush_merge");
0f7b2abd
JK		 737 if (test_opt(sbi, NOBARRIER))
738 seq_puts(seq, ",nobarrier");
d5053a34
JK		 739 if (test_opt(sbi, FASTBOOT))
740 seq_puts(seq, ",fastboot");
89672159
CY		 741 if (test_opt(sbi, EXTENT_CACHE))
742 seq_puts(seq, ",extent_cache");
7daaea25
JK		 743 else
744 seq_puts(seq, ",noextent_cache");
343f40f0
CY		 745 if (test_opt(sbi, DATA_FLUSH))
746 seq_puts(seq, ",data_flush");
aff063e2
JK		 747 seq_printf(seq, ",active_logs=%u", sbi->active_logs);
748
749 return 0;
750}
751
5e176d54
JK		 752static int segment_info_seq_show(struct seq_file *seq, void *offset)
753{
754 struct super_block *sb = seq->private;
755 struct f2fs_sb_info *sbi = F2FS_SB(sb);
6c311ec6
CF		 756 unsigned int total_segs =
757 le32_to_cpu(sbi->raw_super->segment_count_main);
5e176d54
JK		 758 int i;
759
90aa6dc9
CY		 760 seq_puts(seq, "format: segment_type|valid_blocks\n"
761 "segment_type(0:HD, 1:WD, 2:CD, 3:HN, 4:WN, 5:CN)\n");
762
5e176d54 763 for (i = 0; i < total_segs; i++) {
90aa6dc9
CY		 764 struct seg_entry *se = get_seg_entry(sbi, i);
765
766 if ((i % 10) == 0)
01a5ad82 767 seq_printf(seq, "%-10d", i);
90aa6dc9
CY		 768 seq_printf(seq, "%d|%-3u", se->type,
769 get_valid_blocks(sbi, i, 1));
46c04366
GZ		 770 if ((i % 10) == 9 || i == (total_segs - 1))
771 seq_putc(seq, '\n');
5e176d54 772 else
46c04366 773 seq_putc(seq, ' ');
5e176d54 774 }
46c04366 775
5e176d54
JK		 776 return 0;
777}
778
f00d6fa7
JK		 779static int segment_bits_seq_show(struct seq_file *seq, void *offset)
780{
781 struct super_block *sb = seq->private;
782 struct f2fs_sb_info *sbi = F2FS_SB(sb);
783 unsigned int total_segs =
784 le32_to_cpu(sbi->raw_super->segment_count_main);
785 int i, j;
786
787 seq_puts(seq, "format: segment_type|valid_blocks|bitmaps\n"
788 "segment_type(0:HD, 1:WD, 2:CD, 3:HN, 4:WN, 5:CN)\n");
789
790 for (i = 0; i < total_segs; i++) {
791 struct seg_entry *se = get_seg_entry(sbi, i);
792
793 seq_printf(seq, "%-10d", i);
794 seq_printf(seq, "%d|%-3u|", se->type,
795 get_valid_blocks(sbi, i, 1));
796 for (j = 0; j < SIT_VBLOCK_MAP_SIZE; j++)
797 seq_printf(seq, "%x ", se->cur_valid_map[j]);
798 seq_putc(seq, '\n');
799 }
800 return 0;
801}
802
b7a15f3d
JK		 803#define F2FS_PROC_FILE_DEF(_name) \
804static int _name##_open_fs(struct inode *inode, struct file *file) \
805{ \
806 return single_open(file, _name##_seq_show, PDE_DATA(inode)); \
807} \
808 \
809static const struct file_operations f2fs_seq_##_name##_fops = { \
810 .owner = THIS_MODULE, \
811 .open = _name##_open_fs, \
812 .read = seq_read, \
813 .llseek = seq_lseek, \
814 .release = single_release, \
5e176d54
JK		 815};
816
b7a15f3d 817F2FS_PROC_FILE_DEF(segment_info);
f00d6fa7 818F2FS_PROC_FILE_DEF(segment_bits);
b7a15f3d 819
498c5e9f
YH		 820static void default_options(struct f2fs_sb_info *sbi)
821{
822 /* init some FS parameters */
823 sbi->active_logs = NR_CURSEG_TYPE;
824
825 set_opt(sbi, BG_GC);
826 set_opt(sbi, INLINE_DATA);
3e72f721 827 set_opt(sbi, EXTENT_CACHE);
498c5e9f
YH		 828
829#ifdef CONFIG_F2FS_FS_XATTR
830 set_opt(sbi, XATTR_USER);
831#endif
832#ifdef CONFIG_F2FS_FS_POSIX_ACL
833 set_opt(sbi, POSIX_ACL);
834#endif
835}
836
696c018c
NJ		 837static int f2fs_remount(struct super_block *sb, int *flags, char *data)
838{
839 struct f2fs_sb_info *sbi = F2FS_SB(sb);
840 struct f2fs_mount_info org_mount_opt;
841 int err, active_logs;
876dc59e
GZ		 842 bool need_restart_gc = false;
843 bool need_stop_gc = false;
9cd81ce3 844 bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
696c018c
NJ		 845
846 /*
847 * Save the old mount options in case we
848 * need to restore them.
849 */
850 org_mount_opt = sbi->mount_opt;
851 active_logs = sbi->active_logs;
852
df728b0f
JK		 853 /* recover superblocks we couldn't write due to previous RO mount */
854 if (!(*flags & MS_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
855 err = f2fs_commit_super(sbi, false);
856 f2fs_msg(sb, KERN_INFO,
857 "Try to recover all the superblocks, ret: %d", err);
858 if (!err)
859 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
860 }
861
26666c8a 862 sbi->mount_opt.opt = 0;
498c5e9f 863 default_options(sbi);
26666c8a 864
696c018c
NJ		 865 /* parse mount options */
866 err = parse_options(sb, data);
867 if (err)
868 goto restore_opts;
869
870 /*
871 * Previous and new state of filesystem is RO,
876dc59e 872 * so skip checking GC and FLUSH_MERGE conditions.
696c018c 873 */
6b2920a5 874 if (f2fs_readonly(sb) && (*flags & MS_RDONLY))
696c018c
NJ		 875 goto skip;
876
9cd81ce3
CY		 877 /* disallow enable/disable extent_cache dynamically */
878 if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
879 err = -EINVAL;
880 f2fs_msg(sbi->sb, KERN_WARNING,
881 "switch extent_cache option is not allowed");
882 goto restore_opts;
883 }
884
696c018c
NJ		 885 /*
886 * We stop the GC thread if FS is mounted as RO
887 * or if background_gc = off is passed in mount
888 * option. Also sync the filesystem.
889 */
890 if ((*flags & MS_RDONLY) || !test_opt(sbi, BG_GC)) {
891 if (sbi->gc_thread) {
892 stop_gc_thread(sbi);
876dc59e 893 need_restart_gc = true;
696c018c 894 }
aba291b3 895 } else if (!sbi->gc_thread) {
696c018c
NJ		 896 err = start_gc_thread(sbi);
897 if (err)
898 goto restore_opts;
876dc59e
GZ		 899 need_stop_gc = true;
900 }
901
faa0e55b
JK		 902 if (*flags & MS_RDONLY) {
903 writeback_inodes_sb(sb, WB_REASON_SYNC);
904 sync_inodes_sb(sb);
905
906 set_sbi_flag(sbi, SBI_IS_DIRTY);
907 set_sbi_flag(sbi, SBI_IS_CLOSE);
908 f2fs_sync_fs(sb, 1);
909 clear_sbi_flag(sbi, SBI_IS_CLOSE);
910 }
911
876dc59e
GZ		 912 /*
913 * We stop issue flush thread if FS is mounted as RO
914 * or if flush_merge is not passed in mount option.
915 */
916 if ((*flags & MS_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2163d198 917 destroy_flush_cmd_control(sbi);
aba291b3 918 } else if (!SM_I(sbi)->cmd_control_info) {
2163d198
GZ		 919 err = create_flush_cmd_control(sbi);
920 if (err)
a688b9d9 921 goto restore_gc;
696c018c
NJ		 922 }
923skip:
924 /* Update the POSIXACL Flag */
df728b0f 925 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
696c018c 926 (test_opt(sbi, POSIX_ACL) ? MS_POSIXACL : 0);
df728b0f 927
696c018c 928 return 0;
876dc59e
GZ		 929restore_gc:
930 if (need_restart_gc) {
931 if (start_gc_thread(sbi))
932 f2fs_msg(sbi->sb, KERN_WARNING,
e1c42045 933 "background gc thread has stopped");
876dc59e
GZ		 934 } else if (need_stop_gc) {
935 stop_gc_thread(sbi);
936 }
696c018c
NJ		 937restore_opts:
938 sbi->mount_opt = org_mount_opt;
939 sbi->active_logs = active_logs;
940 return err;
941}
942
aff063e2
JK		 943static struct super_operations f2fs_sops = {
944 .alloc_inode = f2fs_alloc_inode,
531ad7d5 945 .drop_inode = f2fs_drop_inode,
aff063e2
JK		 946 .destroy_inode = f2fs_destroy_inode,
947 .write_inode = f2fs_write_inode,
b3783873 948 .dirty_inode = f2fs_dirty_inode,
aff063e2
JK		 949 .show_options = f2fs_show_options,
950 .evict_inode = f2fs_evict_inode,
951 .put_super = f2fs_put_super,
952 .sync_fs = f2fs_sync_fs,
d6212a5f
CL		 953 .freeze_fs = f2fs_freeze,
954 .unfreeze_fs = f2fs_unfreeze,
aff063e2 955 .statfs = f2fs_statfs,
696c018c 956 .remount_fs = f2fs_remount,
aff063e2
JK		 957};
958
0b81d077
JK		 959#ifdef CONFIG_F2FS_FS_ENCRYPTION
960static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
961{
962 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
963 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
964 ctx, len, NULL);
965}
966
967static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
968 void *fs_data)
969{
970 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
971 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
972 ctx, len, fs_data, XATTR_CREATE);
973}
974
975static unsigned f2fs_max_namelen(struct inode *inode)
976{
977 return S_ISLNK(inode->i_mode) ?
978 inode->i_sb->s_blocksize : F2FS_NAME_LEN;
979}
980
981static struct fscrypt_operations f2fs_cryptops = {
982 .get_context = f2fs_get_context,
983 .set_context = f2fs_set_context,
984 .is_encrypted = f2fs_encrypted_inode,
985 .empty_dir = f2fs_empty_dir,
986 .max_namelen = f2fs_max_namelen,
987};
988#else
989static struct fscrypt_operations f2fs_cryptops = {
990 .is_encrypted = f2fs_encrypted_inode,
991};
992#endif
993
aff063e2
JK		 994static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
995 u64 ino, u32 generation)
996{
997 struct f2fs_sb_info *sbi = F2FS_SB(sb);
998 struct inode *inode;
999
d6b7d4b3 1000 if (check_nid_range(sbi, ino))
910bb12d 1001 return ERR_PTR(-ESTALE);
aff063e2
JK		 1002
1003 /*
1004 * f2fs_iget isn't quite right if the inode is currently unallocated!
1005 * However f2fs_iget currently does appropriate checks to handle stale
1006 * inodes so everything is OK.
1007 */
1008 inode = f2fs_iget(sb, ino);
1009 if (IS_ERR(inode))
1010 return ERR_CAST(inode);
6bacf52f 1011 if (unlikely(generation && inode->i_generation != generation)) {
aff063e2
JK		 1012 /* we didn't find the right inode.. */
1013 iput(inode);
1014 return ERR_PTR(-ESTALE);
1015 }
1016 return inode;
1017}
1018
1019static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
1020 int fh_len, int fh_type)
1021{
1022 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
1023 f2fs_nfs_get_inode);
1024}
1025
1026static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
1027 int fh_len, int fh_type)
1028{
1029 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
1030 f2fs_nfs_get_inode);
1031}
1032
1033static const struct export_operations f2fs_export_ops = {
1034 .fh_to_dentry = f2fs_fh_to_dentry,
1035 .fh_to_parent = f2fs_fh_to_parent,
1036 .get_parent = f2fs_get_parent,
1037};
1038
e0afc4d6 1039static loff_t max_file_blocks(void)
aff063e2 1040{
de93653f 1041 loff_t result = (DEF_ADDRS_PER_INODE - F2FS_INLINE_XATTR_ADDRS);
aff063e2
JK		 1042 loff_t leaf_count = ADDRS_PER_BLOCK;
1043
1044 /* two direct node blocks */
1045 result += (leaf_count * 2);
1046
1047 /* two indirect node blocks */
1048 leaf_count *= NIDS_PER_BLOCK;
1049 result += (leaf_count * 2);
1050
1051 /* one double indirect node block */
1052 leaf_count *= NIDS_PER_BLOCK;
1053 result += leaf_count;
1054
aff063e2
JK		 1055 return result;
1056}
1057
fd694733
JK		 1058static int __f2fs_commit_super(struct buffer_head *bh,
1059 struct f2fs_super_block *super)
1060{
1061 lock_buffer(bh);
1062 if (super)
1063 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
1064 set_buffer_uptodate(bh);
1065 set_buffer_dirty(bh);
1066 unlock_buffer(bh);
1067
1068 /* it's rare case, we can do fua all the time */
1069 return __sync_dirty_buffer(bh, WRITE_FLUSH_FUA);
1070}
1071
df728b0f 1072static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
fd694733 1073 struct buffer_head *bh)
9a59b62f 1074{
fd694733
JK		 1075 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
1076 (bh->b_data + F2FS_SUPER_OFFSET);
df728b0f 1077 struct super_block *sb = sbi->sb;
9a59b62f
CY		 1078 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
1079 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
1080 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
1081 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
1082 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
1083 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
1084 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
1085 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
1086 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
1087 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
1088 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
1089 u32 segment_count = le32_to_cpu(raw_super->segment_count);
1090 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
fd694733
JK		 1091 u64 main_end_blkaddr = main_blkaddr +
1092 (segment_count_main << log_blocks_per_seg);
1093 u64 seg_end_blkaddr = segment0_blkaddr +
1094 (segment_count << log_blocks_per_seg);
9a59b62f
CY		 1095
1096 if (segment0_blkaddr != cp_blkaddr) {
1097 f2fs_msg(sb, KERN_INFO,
1098 "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
1099 segment0_blkaddr, cp_blkaddr);
1100 return true;
1101 }
1102
1103 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
1104 sit_blkaddr) {
1105 f2fs_msg(sb, KERN_INFO,
1106 "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
1107 cp_blkaddr, sit_blkaddr,
1108 segment_count_ckpt << log_blocks_per_seg);
1109 return true;
1110 }
1111
1112 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
1113 nat_blkaddr) {
1114 f2fs_msg(sb, KERN_INFO,
1115 "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
1116 sit_blkaddr, nat_blkaddr,
1117 segment_count_sit << log_blocks_per_seg);
1118 return true;
1119 }
1120
1121 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
1122 ssa_blkaddr) {
1123 f2fs_msg(sb, KERN_INFO,
1124 "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
1125 nat_blkaddr, ssa_blkaddr,
1126 segment_count_nat << log_blocks_per_seg);
1127 return true;
1128 }
1129
1130 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
1131 main_blkaddr) {
1132 f2fs_msg(sb, KERN_INFO,
1133 "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
1134 ssa_blkaddr, main_blkaddr,
1135 segment_count_ssa << log_blocks_per_seg);
1136 return true;
1137 }
1138
fd694733 1139 if (main_end_blkaddr > seg_end_blkaddr) {
9a59b62f 1140 f2fs_msg(sb, KERN_INFO,
fd694733 1141 "Wrong MAIN_AREA boundary, start(%u) end(%u) block(%u)",
9a59b62f 1142 main_blkaddr,
fd694733
JK		 1143 segment0_blkaddr +
1144 (segment_count << log_blocks_per_seg),
9a59b62f
CY		 1145 segment_count_main << log_blocks_per_seg);
1146 return true;
fd694733
JK		 1147 } else if (main_end_blkaddr < seg_end_blkaddr) {
1148 int err = 0;
1149 char *res;
1150
1151 /* fix in-memory information all the time */
1152 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
1153 segment0_blkaddr) >> log_blocks_per_seg);
1154
1155 if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
df728b0f 1156 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
fd694733
JK		 1157 res = "internally";
1158 } else {
1159 err = __f2fs_commit_super(bh, NULL);
1160 res = err ? "failed" : "done";
1161 }
1162 f2fs_msg(sb, KERN_INFO,
1163 "Fix alignment : %s, start(%u) end(%u) block(%u)",
1164 res, main_blkaddr,
1165 segment0_blkaddr +
1166 (segment_count << log_blocks_per_seg),
1167 segment_count_main << log_blocks_per_seg);
1168 if (err)
1169 return true;
9a59b62f 1170 }
9a59b62f
CY		 1171 return false;
1172}
1173
df728b0f 1174static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
fd694733 1175 struct buffer_head *bh)
aff063e2 1176{
fd694733
JK		 1177 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
1178 (bh->b_data + F2FS_SUPER_OFFSET);
df728b0f 1179 struct super_block *sb = sbi->sb;
aff063e2
JK		 1180 unsigned int blocksize;
1181
a07ef784
NJ		 1182 if (F2FS_SUPER_MAGIC != le32_to_cpu(raw_super->magic)) {
1183 f2fs_msg(sb, KERN_INFO,
1184 "Magic Mismatch, valid(0x%x) - read(0x%x)",
1185 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
aff063e2 1186 return 1;
a07ef784 1187 }
aff063e2 1188
5c9b4692 1189 /* Currently, support only 4KB page cache size */
09cbfeaf 1190 if (F2FS_BLKSIZE != PAGE_SIZE) {
5c9b4692 1191 f2fs_msg(sb, KERN_INFO,
14d7e9de 1192 "Invalid page_cache_size (%lu), supports only 4KB\n",
09cbfeaf 1193 PAGE_SIZE);
5c9b4692 1194 return 1;
1195 }
1196
aff063e2
JK		 1197 /* Currently, support only 4KB block size */
1198 blocksize = 1 << le32_to_cpu(raw_super->log_blocksize);
5c9b4692 1199 if (blocksize != F2FS_BLKSIZE) {
a07ef784
NJ		 1200 f2fs_msg(sb, KERN_INFO,
1201 "Invalid blocksize (%u), supports only 4KB\n",
1202 blocksize);
aff063e2 1203 return 1;
a07ef784 1204 }
5c9b4692 1205
9a59b62f
CY		 1206 /* check log blocks per segment */
1207 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
1208 f2fs_msg(sb, KERN_INFO,
1209 "Invalid log blocks per segment (%u)\n",
1210 le32_to_cpu(raw_super->log_blocks_per_seg));
1211 return 1;
1212 }
1213
55cf9cb6
CY		 1214 /* Currently, support 512/1024/2048/4096 bytes sector size */
1215 if (le32_to_cpu(raw_super->log_sectorsize) >
1216 F2FS_MAX_LOG_SECTOR_SIZE ||
1217 le32_to_cpu(raw_super->log_sectorsize) <
1218 F2FS_MIN_LOG_SECTOR_SIZE) {
1219 f2fs_msg(sb, KERN_INFO, "Invalid log sectorsize (%u)",
1220 le32_to_cpu(raw_super->log_sectorsize));
aff063e2 1221 return 1;
a07ef784 1222 }
55cf9cb6
CY		 1223 if (le32_to_cpu(raw_super->log_sectors_per_block) +
1224 le32_to_cpu(raw_super->log_sectorsize) !=
1225 F2FS_MAX_LOG_SECTOR_SIZE) {
1226 f2fs_msg(sb, KERN_INFO,
1227 "Invalid log sectors per block(%u) log sectorsize(%u)",
1228 le32_to_cpu(raw_super->log_sectors_per_block),
1229 le32_to_cpu(raw_super->log_sectorsize));
aff063e2 1230 return 1;
a07ef784 1231 }
9a59b62f
CY		 1232
1233 /* check reserved ino info */
1234 if (le32_to_cpu(raw_super->node_ino) != 1 ||
1235 le32_to_cpu(raw_super->meta_ino) != 2 ||
1236 le32_to_cpu(raw_super->root_ino) != 3) {
1237 f2fs_msg(sb, KERN_INFO,
1238 "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
1239 le32_to_cpu(raw_super->node_ino),
1240 le32_to_cpu(raw_super->meta_ino),
1241 le32_to_cpu(raw_super->root_ino));
1242 return 1;
1243 }
1244
1245 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
df728b0f 1246 if (sanity_check_area_boundary(sbi, bh))
9a59b62f
CY		 1247 return 1;
1248
aff063e2
JK		 1249 return 0;
1250}
1251
984ec63c 1252int sanity_check_ckpt(struct f2fs_sb_info *sbi)
aff063e2
JK		 1253{
1254 unsigned int total, fsmeta;
577e3495
JK		 1255 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
1256 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
aff063e2
JK		 1257
1258 total = le32_to_cpu(raw_super->segment_count);
1259 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
1260 fsmeta += le32_to_cpu(raw_super->segment_count_sit);
1261 fsmeta += le32_to_cpu(raw_super->segment_count_nat);
1262 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
1263 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
1264
6bacf52f 1265 if (unlikely(fsmeta >= total))
aff063e2 1266 return 1;
577e3495 1267
1e968fdf 1268 if (unlikely(f2fs_cp_error(sbi))) {
577e3495
JK		 1269 f2fs_msg(sbi->sb, KERN_ERR, "A bug case: need to run fsck");
1270 return 1;
1271 }
aff063e2
JK		 1272 return 0;
1273}
1274
1275static void init_sb_info(struct f2fs_sb_info *sbi)
1276{
1277 struct f2fs_super_block *raw_super = sbi->raw_super;
1278 int i;
1279
1280 sbi->log_sectors_per_block =
1281 le32_to_cpu(raw_super->log_sectors_per_block);
1282 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
1283 sbi->blocksize = 1 << sbi->log_blocksize;
1284 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
1285 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
1286 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
1287 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
1288 sbi->total_sections = le32_to_cpu(raw_super->section_count);
1289 sbi->total_node_count =
1290 (le32_to_cpu(raw_super->segment_count_nat) / 2)
1291 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
1292 sbi->root_ino_num = le32_to_cpu(raw_super->root_ino);
1293 sbi->node_ino_num = le32_to_cpu(raw_super->node_ino);
1294 sbi->meta_ino_num = le32_to_cpu(raw_super->meta_ino);
5ec4e49f 1295 sbi->cur_victim_sec = NULL_SECNO;
b1c57c1c 1296 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
aff063e2
JK		 1297
1298 for (i = 0; i < NR_COUNT_TYPE; i++)
1299 atomic_set(&sbi->nr_pages[i], 0);
ab9fa662
JK		 1300
1301 sbi->dir_level = DEF_DIR_LEVEL;
6beceb54 1302 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
d0239e1b 1303 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
caf0047e 1304 clear_sbi_flag(sbi, SBI_NEED_FSCK);
2658e50d
JK		 1305
1306 INIT_LIST_HEAD(&sbi->s_list);
1307 mutex_init(&sbi->umount_mutex);
aff063e2
JK		 1308}
1309
9076a75f
GZ		 1310/*
1311 * Read f2fs raw super block.
2b39e907
SL		 1312 * Because we have two copies of super block, so read both of them
1313 * to get the first valid one. If any one of them is broken, we pass
1314 * them recovery flag back to the caller.
9076a75f 1315 */
df728b0f 1316static int read_raw_super_block(struct f2fs_sb_info *sbi,
9076a75f 1317 struct f2fs_super_block **raw_super,
e8240f65 1318 int *valid_super_block, int *recovery)
14d7e9de 1319{
df728b0f 1320 struct super_block *sb = sbi->sb;
2b39e907 1321 int block;
e8240f65 1322 struct buffer_head *bh;
fd694733 1323 struct f2fs_super_block *super;
da554e48 1324 int err = 0;
14d7e9de 1325
b39f0de2
YH		 1326 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
1327 if (!super)
1328 return -ENOMEM;
2b39e907
SL		 1329
1330 for (block = 0; block < 2; block++) {
1331 bh = sb_bread(sb, block);
1332 if (!bh) {
1333 f2fs_msg(sb, KERN_ERR, "Unable to read %dth superblock",
9076a75f 1334 block + 1);
2b39e907
SL		 1335 err = -EIO;
1336 continue;
1337 }
14d7e9de 1338
2b39e907 1339 /* sanity checking of raw super */
df728b0f 1340 if (sanity_check_raw_super(sbi, bh)) {
2b39e907
SL		 1341 f2fs_msg(sb, KERN_ERR,
1342 "Can't find valid F2FS filesystem in %dth superblock",
1343 block + 1);
1344 err = -EINVAL;
1345 brelse(bh);
1346 continue;
1347 }
14d7e9de 1348
2b39e907 1349 if (!*raw_super) {
fd694733
JK		 1350 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
1351 sizeof(*super));
2b39e907
SL		 1352 *valid_super_block = block;
1353 *raw_super = super;
1354 }
1355 brelse(bh);
da554e48 1356 }
1357
2b39e907
SL		 1358 /* Fail to read any one of the superblocks*/
1359 if (err < 0)
1360 *recovery = 1;
da554e48 1361
da554e48 1362 /* No valid superblock */
2b39e907 1363 if (!*raw_super)
b39f0de2 1364 kfree(super);
2b39e907
SL		 1365 else
1366 err = 0;
da554e48 1367
2b39e907 1368 return err;
14d7e9de 1369}
1370
fd694733 1371int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
26d815ad 1372{
5d909cdb 1373 struct buffer_head *bh;
26d815ad
JK		 1374 int err;
1375
df728b0f
JK		 1376 if ((recover && f2fs_readonly(sbi->sb)) ||
1377 bdev_read_only(sbi->sb->s_bdev)) {
1378 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
f2353d7b 1379 return -EROFS;
df728b0f 1380 }
f2353d7b 1381
fd694733
JK		 1382 /* write back-up superblock first */
1383 bh = sb_getblk(sbi->sb, sbi->valid_super_block ? 0: 1);
5d909cdb
JK		 1384 if (!bh)
1385 return -EIO;
fd694733 1386 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
5d909cdb 1387 brelse(bh);
c5bda1c8
CY		 1388
1389 /* if we are in recovery path, skip writing valid superblock */
1390 if (recover || err)
5d909cdb 1391 return err;
26d815ad
JK		 1392
1393 /* write current valid superblock */
fd694733
JK		 1394 bh = sb_getblk(sbi->sb, sbi->valid_super_block);
1395 if (!bh)
1396 return -EIO;
1397 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
1398 brelse(bh);
1399 return err;
26d815ad
JK		 1400}
1401
aff063e2
JK		 1402static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
1403{
1404 struct f2fs_sb_info *sbi;
da554e48 1405 struct f2fs_super_block *raw_super;
aff063e2 1406 struct inode *root;
da554e48 1407 long err;
2adc3505 1408 bool retry = true, need_fsck = false;
dabc4a5c 1409 char *options = NULL;
e8240f65 1410 int recovery, i, valid_super_block;
8f1dbbbb 1411 struct curseg_info *seg_i;
aff063e2 1412
ed2e621a 1413try_onemore:
da554e48 1414 err = -EINVAL;
1415 raw_super = NULL;
e8240f65 1416 valid_super_block = -1;
da554e48 1417 recovery = 0;
1418
aff063e2
JK		 1419 /* allocate memory for f2fs-specific super block info */
1420 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
1421 if (!sbi)
1422 return -ENOMEM;
1423
df728b0f
JK		 1424 sbi->sb = sb;
1425
43b6573b
KM		 1426 /* Load the checksum driver */
1427 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
1428 if (IS_ERR(sbi->s_chksum_driver)) {
1429 f2fs_msg(sb, KERN_ERR, "Cannot load crc32 driver.");
1430 err = PTR_ERR(sbi->s_chksum_driver);
1431 sbi->s_chksum_driver = NULL;
1432 goto free_sbi;
1433 }
1434
ff9234ad 1435 /* set a block size */
6bacf52f 1436 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
a07ef784 1437 f2fs_msg(sb, KERN_ERR, "unable to set blocksize");
aff063e2 1438 goto free_sbi;
a07ef784 1439 }
aff063e2 1440
df728b0f 1441 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
e8240f65 1442 &recovery);
9076a75f
GZ		 1443 if (err)
1444 goto free_sbi;
1445
5fb08372 1446 sb->s_fs_info = sbi;
498c5e9f 1447 default_options(sbi);
aff063e2 1448 /* parse mount options */
dabc4a5c
JK		 1449 options = kstrdup((const char *)data, GFP_KERNEL);
1450 if (data && !options) {
1451 err = -ENOMEM;
aff063e2 1452 goto free_sb_buf;
dabc4a5c
JK		 1453 }
1454
1455 err = parse_options(sb, options);
1456 if (err)
1457 goto free_options;
aff063e2 1458
e0afc4d6
CY		 1459 sbi->max_file_blocks = max_file_blocks();
1460 sb->s_maxbytes = sbi->max_file_blocks <<
1461 le32_to_cpu(raw_super->log_blocksize);
aff063e2
JK		 1462 sb->s_max_links = F2FS_LINK_MAX;
1463 get_random_bytes(&sbi->s_next_generation, sizeof(u32));
1464
1465 sb->s_op = &f2fs_sops;
0b81d077 1466 sb->s_cop = &f2fs_cryptops;
aff063e2
JK		 1467 sb->s_xattr = f2fs_xattr_handlers;
1468 sb->s_export_op = &f2fs_export_ops;
1469 sb->s_magic = F2FS_SUPER_MAGIC;
aff063e2
JK		 1470 sb->s_time_gran = 1;
1471 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
1472 (test_opt(sbi, POSIX_ACL) ? MS_POSIXACL : 0);
1473 memcpy(sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
1474
1475 /* init f2fs-specific super block info */
aff063e2 1476 sbi->raw_super = raw_super;
e8240f65 1477 sbi->valid_super_block = valid_super_block;
aff063e2 1478 mutex_init(&sbi->gc_mutex);
5463e7c1 1479 mutex_init(&sbi->writepages);
aff063e2 1480 mutex_init(&sbi->cp_mutex);
b3582c68 1481 init_rwsem(&sbi->node_write);
315df839
JK		 1482
1483 /* disallow all the data/node/meta page writes */
1484 set_sbi_flag(sbi, SBI_POR_DOING);
aff063e2 1485 spin_lock_init(&sbi->stat_lock);
971767ca 1486
df0f8dc0 1487 init_rwsem(&sbi->read_io.io_rwsem);
458e6197
JK		 1488 sbi->read_io.sbi = sbi;
1489 sbi->read_io.bio = NULL;
1490 for (i = 0; i < NR_PAGE_TYPE; i++) {
df0f8dc0 1491 init_rwsem(&sbi->write_io[i].io_rwsem);
458e6197
JK		 1492 sbi->write_io[i].sbi = sbi;
1493 sbi->write_io[i].bio = NULL;
1494 }
971767ca 1495
e479556b 1496 init_rwsem(&sbi->cp_rwsem);
fb51b5ef 1497 init_waitqueue_head(&sbi->cp_wait);
aff063e2
JK		 1498 init_sb_info(sbi);
1499
1500 /* get an inode for meta space */
1501 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
1502 if (IS_ERR(sbi->meta_inode)) {
a07ef784 1503 f2fs_msg(sb, KERN_ERR, "Failed to read F2FS meta data inode");
aff063e2 1504 err = PTR_ERR(sbi->meta_inode);
dabc4a5c 1505 goto free_options;
aff063e2
JK		 1506 }
1507
1508 err = get_valid_checkpoint(sbi);
a07ef784
NJ		 1509 if (err) {
1510 f2fs_msg(sb, KERN_ERR, "Failed to get valid F2FS checkpoint");
aff063e2 1511 goto free_meta_inode;
a07ef784 1512 }
aff063e2 1513
aff063e2
JK		 1514 sbi->total_valid_node_count =
1515 le32_to_cpu(sbi->ckpt->valid_node_count);
1516 sbi->total_valid_inode_count =
1517 le32_to_cpu(sbi->ckpt->valid_inode_count);
1518 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
1519 sbi->total_valid_block_count =
1520 le64_to_cpu(sbi->ckpt->valid_block_count);
1521 sbi->last_valid_block_count = sbi->total_valid_block_count;
1522 sbi->alloc_valid_block_count = 0;
c227f912
CY		 1523 for (i = 0; i < NR_INODE_TYPE; i++) {
1524 INIT_LIST_HEAD(&sbi->inode_list[i]);
1525 spin_lock_init(&sbi->inode_lock[i]);
1526 }
aff063e2 1527
1dcc336b
CY		 1528 init_extent_cache_info(sbi);
1529
6451e041 1530 init_ino_entry_info(sbi);
aff063e2
JK		 1531
1532 /* setup f2fs internal modules */
1533 err = build_segment_manager(sbi);
a07ef784
NJ		 1534 if (err) {
1535 f2fs_msg(sb, KERN_ERR,
1536 "Failed to initialize F2FS segment manager");
aff063e2 1537 goto free_sm;
a07ef784 1538 }
aff063e2 1539 err = build_node_manager(sbi);
a07ef784
NJ		 1540 if (err) {
1541 f2fs_msg(sb, KERN_ERR,
1542 "Failed to initialize F2FS node manager");
aff063e2 1543 goto free_nm;
a07ef784 1544 }
aff063e2 1545
8f1dbbbb
SL		 1546 /* For write statistics */
1547 if (sb->s_bdev->bd_part)
1548 sbi->sectors_written_start =
1549 (u64)part_stat_read(sb->s_bdev->bd_part, sectors[1]);
1550
1551 /* Read accumulated write IO statistics if exists */
1552 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
1553 if (__exist_node_summaries(sbi))
1554 sbi->kbytes_written =
b2dde6fc 1555 le64_to_cpu(seg_i->journal->info.kbytes_written);
8f1dbbbb 1556
aff063e2
JK		 1557 build_gc_manager(sbi);
1558
1559 /* get an inode for node space */
1560 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
1561 if (IS_ERR(sbi->node_inode)) {
a07ef784 1562 f2fs_msg(sb, KERN_ERR, "Failed to read node inode");
aff063e2
JK		 1563 err = PTR_ERR(sbi->node_inode);
1564 goto free_nm;
1565 }
1566
2658e50d
JK		 1567 f2fs_join_shrinker(sbi);
1568
aff063e2 1569 /* if there are nt orphan nodes free them */
8c14bfad
CY		 1570 err = recover_orphan_inodes(sbi);
1571 if (err)
1572 goto free_node_inode;
aff063e2
JK		 1573
1574 /* read root inode and dentry */
1575 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
1576 if (IS_ERR(root)) {
a07ef784 1577 f2fs_msg(sb, KERN_ERR, "Failed to read root inode");
aff063e2
JK		 1578 err = PTR_ERR(root);
1579 goto free_node_inode;
1580 }
8f99a946 1581 if (!S_ISDIR(root->i_mode) || !root->i_blocks || !root->i_size) {
9d847950 1582 iput(root);
8f99a946 1583 err = -EINVAL;
9d847950 1584 goto free_node_inode;
8f99a946 1585 }
aff063e2
JK		 1586
1587 sb->s_root = d_make_root(root); /* allocate root dentry */
1588 if (!sb->s_root) {
1589 err = -ENOMEM;
1590 goto free_root_inode;
1591 }
1592
aff063e2
JK		 1593 err = f2fs_build_stats(sbi);
1594 if (err)
6437d1b0 1595 goto free_root_inode;
aff063e2 1596
5e176d54
JK		 1597 if (f2fs_proc_root)
1598 sbi->s_proc = proc_mkdir(sb->s_id, f2fs_proc_root);
1599
f00d6fa7 1600 if (sbi->s_proc) {
5e176d54
JK		 1601 proc_create_data("segment_info", S_IRUGO, sbi->s_proc,
1602 &f2fs_seq_segment_info_fops, sb);
f00d6fa7
JK		 1603 proc_create_data("segment_bits", S_IRUGO, sbi->s_proc,
1604 &f2fs_seq_segment_bits_fops, sb);
1605 }
5e176d54 1606
b59d0bae
NJ		 1607 sbi->s_kobj.kset = f2fs_kset;
1608 init_completion(&sbi->s_kobj_unregister);
1609 err = kobject_init_and_add(&sbi->s_kobj, &f2fs_ktype, NULL,
1610 "%s", sb->s_id);
1611 if (err)
6437d1b0 1612 goto free_proc;
b59d0bae 1613
6437d1b0
JK		 1614 /* recover fsynced data */
1615 if (!test_opt(sbi, DISABLE_ROLL_FORWARD)) {
081d78c2
JK		 1616 /*
1617 * mount should be failed, when device has readonly mode, and
1618 * previous checkpoint was not done by clean system shutdown.
1619 */
1620 if (bdev_read_only(sb->s_bdev) &&
1621 !is_set_ckpt_flags(sbi->ckpt, CP_UMOUNT_FLAG)) {
1622 err = -EROFS;
1623 goto free_kobj;
1624 }
2adc3505
CY		 1625
1626 if (need_fsck)
1627 set_sbi_flag(sbi, SBI_NEED_FSCK);
1628
6781eabb
JK		 1629 err = recover_fsync_data(sbi, false);
1630 if (err < 0) {
2adc3505 1631 need_fsck = true;
6437d1b0
JK		 1632 f2fs_msg(sb, KERN_ERR,
1633 "Cannot recover all fsync data errno=%ld", err);
ed2e621a
JK		 1634 goto free_kobj;
1635 }
6781eabb
JK		 1636 } else {
1637 err = recover_fsync_data(sbi, true);
1638
1639 if (!f2fs_readonly(sb) && err > 0) {
1640 err = -EINVAL;
1641 f2fs_msg(sb, KERN_ERR,
1642 "Need to recover fsync data");
1643 goto free_kobj;
1644 }
6437d1b0 1645 }
6781eabb 1646
315df839
JK		 1647 /* recover_fsync_data() cleared this already */
1648 clear_sbi_flag(sbi, SBI_POR_DOING);
b59d0bae 1649
6437d1b0
JK		 1650 /*
1651 * If filesystem is not mounted as read-only then
1652 * do start the gc_thread.
1653 */
6c029932 1654 if (test_opt(sbi, BG_GC) && !f2fs_readonly(sb)) {
6437d1b0
JK		 1655 /* After POR, we can run background GC thread.*/
1656 err = start_gc_thread(sbi);
1657 if (err)
1658 goto free_kobj;
1659 }
dabc4a5c 1660 kfree(options);
da554e48 1661
1662 /* recover broken superblock */
f2353d7b 1663 if (recovery) {
41214b3c
CY		 1664 err = f2fs_commit_super(sbi, true);
1665 f2fs_msg(sb, KERN_INFO,
1666 "Try to recover %dth superblock, ret: %ld",
1667 sbi->valid_super_block ? 1 : 2, err);
da554e48 1668 }
1669
6beceb54 1670 f2fs_update_time(sbi, CP_TIME);
d0239e1b 1671 f2fs_update_time(sbi, REQ_TIME);
aff063e2 1672 return 0;
6437d1b0
JK		 1673
1674free_kobj:
1675 kobject_del(&sbi->s_kobj);
29ba108d
CY		 1676 kobject_put(&sbi->s_kobj);
1677 wait_for_completion(&sbi->s_kobj_unregister);
6437d1b0 1678free_proc:
1d15bd20
CY		 1679 if (sbi->s_proc) {
1680 remove_proc_entry("segment_info", sbi->s_proc);
f00d6fa7 1681 remove_proc_entry("segment_bits", sbi->s_proc);
1d15bd20
CY		 1682 remove_proc_entry(sb->s_id, f2fs_proc_root);
1683 }
1684 f2fs_destroy_stats(sbi);
aff063e2
JK		 1685free_root_inode:
1686 dput(sb->s_root);
1687 sb->s_root = NULL;
1688free_node_inode:
2658e50d
JK		 1689 mutex_lock(&sbi->umount_mutex);
1690 f2fs_leave_shrinker(sbi);
aff063e2 1691 iput(sbi->node_inode);
2658e50d 1692 mutex_unlock(&sbi->umount_mutex);
aff063e2
JK		 1693free_nm:
1694 destroy_node_manager(sbi);
1695free_sm:
1696 destroy_segment_manager(sbi);
aff063e2
JK		 1697 kfree(sbi->ckpt);
1698free_meta_inode:
1699 make_bad_inode(sbi->meta_inode);
1700 iput(sbi->meta_inode);
dabc4a5c
JK		 1701free_options:
1702 kfree(options);
aff063e2 1703free_sb_buf:
b39f0de2 1704 kfree(raw_super);
aff063e2 1705free_sbi:
43b6573b
KM		 1706 if (sbi->s_chksum_driver)
1707 crypto_free_shash(sbi->s_chksum_driver);
aff063e2 1708 kfree(sbi);
ed2e621a
JK		 1709
1710 /* give only one another chance */
1711 if (retry) {
9df47ba7 1712 retry = false;
ed2e621a
JK		 1713 shrink_dcache_sb(sb);
1714 goto try_onemore;
1715 }
aff063e2
JK		 1716 return err;
1717}
1718
1719static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
1720 const char *dev_name, void *data)
1721{
1722 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
1723}
1724
30a5537f
JK		 1725static void kill_f2fs_super(struct super_block *sb)
1726{
1727 if (sb->s_root)
caf0047e 1728 set_sbi_flag(F2FS_SB(sb), SBI_IS_CLOSE);
30a5537f
JK		 1729 kill_block_super(sb);
1730}
1731
aff063e2
JK		 1732static struct file_system_type f2fs_fs_type = {
1733 .owner = THIS_MODULE,
1734 .name = "f2fs",
1735 .mount = f2fs_mount,
30a5537f 1736 .kill_sb = kill_f2fs_super,
aff063e2
JK		 1737 .fs_flags = FS_REQUIRES_DEV,
1738};
7f78e035 1739MODULE_ALIAS_FS("f2fs");
aff063e2 1740
6e6093a8 1741static int __init init_inodecache(void)
aff063e2 1742{
5d097056
VD		 1743 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
1744 sizeof(struct f2fs_inode_info), 0,
1745 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
6bacf52f 1746 if (!f2fs_inode_cachep)
aff063e2
JK		 1747 return -ENOMEM;
1748 return 0;
1749}
1750
1751static void destroy_inodecache(void)
1752{
1753 /*
1754 * Make sure all delayed rcu free inodes are flushed before we
1755 * destroy cache.
1756 */
1757 rcu_barrier();
1758 kmem_cache_destroy(f2fs_inode_cachep);
1759}
1760
1761static int __init init_f2fs_fs(void)
1762{
1763 int err;
1764
c0508650
JK		 1765 f2fs_build_trace_ios();
1766
aff063e2
JK		 1767 err = init_inodecache();
1768 if (err)
1769 goto fail;
1770 err = create_node_manager_caches();
1771 if (err)
9890ff3f 1772 goto free_inodecache;
7fd9e544 1773 err = create_segment_manager_caches();
aff063e2 1774 if (err)
9890ff3f 1775 goto free_node_manager_caches;
aff063e2
JK		 1776 err = create_checkpoint_caches();
1777 if (err)
06292073 1778 goto free_segment_manager_caches;
1dcc336b
CY		 1779 err = create_extent_cache();
1780 if (err)
1781 goto free_checkpoint_caches;
b59d0bae 1782 f2fs_kset = kset_create_and_add("f2fs", NULL, fs_kobj);
6e6b978c
WY		 1783 if (!f2fs_kset) {
1784 err = -ENOMEM;
1dcc336b 1785 goto free_extent_cache;
6e6b978c 1786 }
2658e50d 1787 err = register_shrinker(&f2fs_shrinker_info);
cfc4d971 1788 if (err)
0b81d077 1789 goto free_kset;
2658e50d
JK		 1790
1791 err = register_filesystem(&f2fs_fs_type);
1792 if (err)
1793 goto free_shrinker;
787c7b8c
CY		 1794 err = f2fs_create_root_stats();
1795 if (err)
1796 goto free_filesystem;
5e176d54 1797 f2fs_proc_root = proc_mkdir("fs/f2fs", NULL);
9890ff3f
ZH		 1798 return 0;
1799
787c7b8c
CY		 1800free_filesystem:
1801 unregister_filesystem(&f2fs_fs_type);
2658e50d
JK		 1802free_shrinker:
1803 unregister_shrinker(&f2fs_shrinker_info);
9890ff3f
ZH		 1804free_kset:
1805 kset_unregister(f2fs_kset);
1dcc336b
CY		 1806free_extent_cache:
1807 destroy_extent_cache();
9890ff3f
ZH		 1808free_checkpoint_caches:
1809 destroy_checkpoint_caches();
7fd9e544
JK		 1810free_segment_manager_caches:
1811 destroy_segment_manager_caches();
9890ff3f
ZH		 1812free_node_manager_caches:
1813 destroy_node_manager_caches();
1814free_inodecache:
1815 destroy_inodecache();
aff063e2
JK		 1816fail:
1817 return err;
1818}
1819
1820static void __exit exit_f2fs_fs(void)
1821{
5e176d54 1822 remove_proc_entry("fs/f2fs", NULL);
4589d25d 1823 f2fs_destroy_root_stats();
2658e50d 1824 unregister_shrinker(&f2fs_shrinker_info);
aff063e2 1825 unregister_filesystem(&f2fs_fs_type);
fdf6c8be 1826 destroy_extent_cache();
aff063e2 1827 destroy_checkpoint_caches();
5dcd8a71 1828 destroy_segment_manager_caches();
aff063e2
JK		 1829 destroy_node_manager_caches();
1830 destroy_inodecache();
b59d0bae 1831 kset_unregister(f2fs_kset);
351f4fba 1832 f2fs_destroy_trace_ios();
aff063e2
JK		 1833}
1834
1835module_init(init_f2fs_fs)
1836module_exit(exit_f2fs_fs)
1837
1838MODULE_AUTHOR("Samsung Electronics's Praesto Team");
1839MODULE_DESCRIPTION("Flash Friendly File System");
1840MODULE_LICENSE("GPL");
Linux 6.x block layer and io_uring tree(s)