projects / deliverable / linux.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
f2fs: use percpu_rw_semaphore
[deliverable/linux.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 42#ifdef CONFIG_F2FS_FAULT_INJECTION
08796897 43struct f2fs_fault_info f2fs_fault;
2c63fead
JK		 44
45char *fault_name[FAULT_MAX] = {
46 [FAULT_KMALLOC] = "kmalloc",
c41f3cc3 47 [FAULT_PAGE_ALLOC] = "page alloc",
cb78942b
JK		 48 [FAULT_ALLOC_NID] = "alloc nid",
49 [FAULT_ORPHAN] = "orphan",
50 [FAULT_BLOCK] = "no more block",
51 [FAULT_DIR_DEPTH] = "too big dir depth",
53aa6bbf 52 [FAULT_EVICT_INODE] = "evict_inode fail",
2c63fead 53};
08796897
SY		 54
55static void f2fs_build_fault_attr(unsigned int rate)
56{
57 if (rate) {
58 atomic_set(&f2fs_fault.inject_ops, 0);
59 f2fs_fault.inject_rate = rate;
60 f2fs_fault.inject_type = (1 << FAULT_MAX) - 1;
61 } else {
62 memset(&f2fs_fault, 0, sizeof(struct f2fs_fault_info));
63 }
64}
73faec4d
JK		 65#endif
66
2658e50d
JK		 67/* f2fs-wide shrinker description */
68static struct shrinker f2fs_shrinker_info = {
69 .scan_objects = f2fs_shrink_scan,
70 .count_objects = f2fs_shrink_count,
71 .seeks = DEFAULT_SEEKS,
72};
73
aff063e2 74enum {
696c018c 75 Opt_gc_background,
aff063e2 76 Opt_disable_roll_forward,
2d834bf9 77 Opt_norecovery,
aff063e2 78 Opt_discard,
64058be9 79 Opt_nodiscard,
aff063e2 80 Opt_noheap,
4058c511 81 Opt_user_xattr,
aff063e2 82 Opt_nouser_xattr,
4058c511 83 Opt_acl,
aff063e2
JK		 84 Opt_noacl,
85 Opt_active_logs,
86 Opt_disable_ext_identify,
444c580f 87 Opt_inline_xattr,
8274de77 88 Opt_inline_data,
5efd3c6f 89 Opt_inline_dentry,
6b4afdd7 90 Opt_flush_merge,
69e9e427 91 Opt_noflush_merge,
0f7b2abd 92 Opt_nobarrier,
d5053a34 93 Opt_fastboot,
89672159 94 Opt_extent_cache,
7daaea25 95 Opt_noextent_cache,
75342797 96 Opt_noinline_data,
343f40f0 97 Opt_data_flush,
36abef4e 98 Opt_mode,
73faec4d 99 Opt_fault_injection,
6d94c74a
JK		 100 Opt_lazytime,
101 Opt_nolazytime,
aff063e2
JK		 102 Opt_err,
103};
104
105static match_table_t f2fs_tokens = {
696c018c 106 {Opt_gc_background, "background_gc=%s"},
aff063e2 107 {Opt_disable_roll_forward, "disable_roll_forward"},
2d834bf9 108 {Opt_norecovery, "norecovery"},
aff063e2 109 {Opt_discard, "discard"},
64058be9 110 {Opt_nodiscard, "nodiscard"},
aff063e2 111 {Opt_noheap, "no_heap"},
4058c511 112 {Opt_user_xattr, "user_xattr"},
aff063e2 113 {Opt_nouser_xattr, "nouser_xattr"},
4058c511 114 {Opt_acl, "acl"},
aff063e2
JK		 115 {Opt_noacl, "noacl"},
116 {Opt_active_logs, "active_logs=%u"},
117 {Opt_disable_ext_identify, "disable_ext_identify"},
444c580f 118 {Opt_inline_xattr, "inline_xattr"},
8274de77 119 {Opt_inline_data, "inline_data"},
5efd3c6f 120 {Opt_inline_dentry, "inline_dentry"},
6b4afdd7 121 {Opt_flush_merge, "flush_merge"},
69e9e427 122 {Opt_noflush_merge, "noflush_merge"},
0f7b2abd 123 {Opt_nobarrier, "nobarrier"},
d5053a34 124 {Opt_fastboot, "fastboot"},
89672159 125 {Opt_extent_cache, "extent_cache"},
7daaea25 126 {Opt_noextent_cache, "noextent_cache"},
75342797 127 {Opt_noinline_data, "noinline_data"},
343f40f0 128 {Opt_data_flush, "data_flush"},
36abef4e 129 {Opt_mode, "mode=%s"},
73faec4d 130 {Opt_fault_injection, "fault_injection=%u"},
6d94c74a
JK		 131 {Opt_lazytime, "lazytime"},
132 {Opt_nolazytime, "nolazytime"},
aff063e2
JK		 133 {Opt_err, NULL},
134};
135
b59d0bae 136/* Sysfs support for f2fs */
ea91e9b0
JK		 137enum {
138 GC_THREAD, /* struct f2fs_gc_thread */
139 SM_INFO, /* struct f2fs_sm_info */
cdfc41c1 140 NM_INFO, /* struct f2fs_nm_info */
b1c57c1c 141 F2FS_SBI, /* struct f2fs_sb_info */
08796897
SY		 142#ifdef CONFIG_F2FS_FAULT_INJECTION
143 FAULT_INFO_RATE, /* struct f2fs_fault_info */
144 FAULT_INFO_TYPE, /* struct f2fs_fault_info */
145#endif
ea91e9b0
JK		 146};
147
b59d0bae
NJ		 148struct f2fs_attr {
149 struct attribute attr;
150 ssize_t (*show)(struct f2fs_attr *, struct f2fs_sb_info *, char *);
151 ssize_t (*store)(struct f2fs_attr *, struct f2fs_sb_info *,
152 const char *, size_t);
ea91e9b0 153 int struct_type;
b59d0bae
NJ		 154 int offset;
155};
156
ea91e9b0
JK		 157static unsigned char *__struct_ptr(struct f2fs_sb_info *sbi, int struct_type)
158{
159 if (struct_type == GC_THREAD)
160 return (unsigned char *)sbi->gc_thread;
161 else if (struct_type == SM_INFO)
162 return (unsigned char *)SM_I(sbi);
cdfc41c1
JK		 163 else if (struct_type == NM_INFO)
164 return (unsigned char *)NM_I(sbi);
b1c57c1c
JK		 165 else if (struct_type == F2FS_SBI)
166 return (unsigned char *)sbi;
08796897
SY		 167#ifdef CONFIG_F2FS_FAULT_INJECTION
168 else if (struct_type == FAULT_INFO_RATE ||
169 struct_type == FAULT_INFO_TYPE)
170 return (unsigned char *)&f2fs_fault;
171#endif
ea91e9b0
JK		 172 return NULL;
173}
174
8f1dbbbb
SL		 175static ssize_t lifetime_write_kbytes_show(struct f2fs_attr *a,
176 struct f2fs_sb_info *sbi, char *buf)
177{
178 struct super_block *sb = sbi->sb;
179
180 if (!sb->s_bdev->bd_part)
181 return snprintf(buf, PAGE_SIZE, "0\n");
182
183 return snprintf(buf, PAGE_SIZE, "%llu\n",
184 (unsigned long long)(sbi->kbytes_written +
185 BD_PART_WRITTEN(sbi)));
186}
187
b59d0bae
NJ		 188static ssize_t f2fs_sbi_show(struct f2fs_attr *a,
189 struct f2fs_sb_info *sbi, char *buf)
190{
ea91e9b0 191 unsigned char *ptr = NULL;
b59d0bae
NJ		 192 unsigned int *ui;
193
ea91e9b0
JK		 194 ptr = __struct_ptr(sbi, a->struct_type);
195 if (!ptr)
b59d0bae
NJ		 196 return -EINVAL;
197
ea91e9b0 198 ui = (unsigned int *)(ptr + a->offset);
b59d0bae
NJ		 199
200 return snprintf(buf, PAGE_SIZE, "%u\n", *ui);
201}
202
203static ssize_t f2fs_sbi_store(struct f2fs_attr *a,
204 struct f2fs_sb_info *sbi,
205 const char *buf, size_t count)
206{
ea91e9b0 207 unsigned char *ptr;
b59d0bae
NJ		 208 unsigned long t;
209 unsigned int *ui;
210 ssize_t ret;
211
ea91e9b0
JK		 212 ptr = __struct_ptr(sbi, a->struct_type);
213 if (!ptr)
b59d0bae
NJ		 214 return -EINVAL;
215
ea91e9b0 216 ui = (unsigned int *)(ptr + a->offset);
b59d0bae
NJ		 217
218 ret = kstrtoul(skip_spaces(buf), 0, &t);
219 if (ret < 0)
220 return ret;
08796897
SY		 221#ifdef CONFIG_F2FS_FAULT_INJECTION
222 if (a->struct_type == FAULT_INFO_TYPE && t >= (1 << FAULT_MAX))
223 return -EINVAL;
224#endif
b59d0bae
NJ		 225 *ui = t;
226 return count;
227}
228
229static ssize_t f2fs_attr_show(struct kobject *kobj,
230 struct attribute *attr, char *buf)
231{
232 struct f2fs_sb_info *sbi = container_of(kobj, struct f2fs_sb_info,
233 s_kobj);
234 struct f2fs_attr *a = container_of(attr, struct f2fs_attr, attr);
235
236 return a->show ? a->show(a, sbi, buf) : 0;
237}
238
239static ssize_t f2fs_attr_store(struct kobject *kobj, struct attribute *attr,
240 const char *buf, size_t len)
241{
242 struct f2fs_sb_info *sbi = container_of(kobj, struct f2fs_sb_info,
243 s_kobj);
244 struct f2fs_attr *a = container_of(attr, struct f2fs_attr, attr);
245
246 return a->store ? a->store(a, sbi, buf, len) : 0;
247}
248
249static void f2fs_sb_release(struct kobject *kobj)
250{
251 struct f2fs_sb_info *sbi = container_of(kobj, struct f2fs_sb_info,
252 s_kobj);
253 complete(&sbi->s_kobj_unregister);
254}
255
ea91e9b0 256#define F2FS_ATTR_OFFSET(_struct_type, _name, _mode, _show, _store, _offset) \
b59d0bae
NJ		 257static struct f2fs_attr f2fs_attr_##_name = { \
258 .attr = {.name = __stringify(_name), .mode = _mode }, \
259 .show = _show, \
260 .store = _store, \
ea91e9b0
JK		 261 .struct_type = _struct_type, \
262 .offset = _offset \
b59d0bae
NJ		 263}
264
ea91e9b0
JK		 265#define F2FS_RW_ATTR(struct_type, struct_name, name, elname) \
266 F2FS_ATTR_OFFSET(struct_type, name, 0644, \
267 f2fs_sbi_show, f2fs_sbi_store, \
268 offsetof(struct struct_name, elname))
b59d0bae 269
8f1dbbbb
SL		 270#define F2FS_GENERAL_RO_ATTR(name) \
271static struct f2fs_attr f2fs_attr_##name = __ATTR(name, 0444, name##_show, NULL)
272
ea91e9b0
JK		 273F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_min_sleep_time, min_sleep_time);
274F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_max_sleep_time, max_sleep_time);
275F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_no_gc_sleep_time, no_gc_sleep_time);
276F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_idle, gc_idle);
277F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, reclaim_segments, rec_prefree_segments);
7ac8c3b0 278F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, max_small_discards, max_discards);
bba681cb 279F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, batched_trim_sections, trim_sections);
216fbd64
JK		 280F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, ipu_policy, ipu_policy);
281F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, min_ipu_util, min_ipu_util);
c1ce1b02 282F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, min_fsync_blocks, min_fsync_blocks);
cdfc41c1 283F2FS_RW_ATTR(NM_INFO, f2fs_nm_info, ram_thresh, ram_thresh);
ea1a29a0 284F2FS_RW_ATTR(NM_INFO, f2fs_nm_info, ra_nid_pages, ra_nid_pages);
2304cb0c 285F2FS_RW_ATTR(NM_INFO, f2fs_nm_info, dirty_nats_ratio, dirty_nats_ratio);
b1c57c1c 286F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, max_victim_search, max_victim_search);
ab9fa662 287F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, dir_level, dir_level);
6beceb54 288F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, cp_interval, interval_time[CP_TIME]);
d0239e1b 289F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, idle_interval, interval_time[REQ_TIME]);
08796897
SY		 290#ifdef CONFIG_F2FS_FAULT_INJECTION
291F2FS_RW_ATTR(FAULT_INFO_RATE, f2fs_fault_info, inject_rate, inject_rate);
292F2FS_RW_ATTR(FAULT_INFO_TYPE, f2fs_fault_info, inject_type, inject_type);
293#endif
8f1dbbbb 294F2FS_GENERAL_RO_ATTR(lifetime_write_kbytes);
b59d0bae
NJ		 295
296#define ATTR_LIST(name) (&f2fs_attr_##name.attr)
297static struct attribute *f2fs_attrs[] = {
298 ATTR_LIST(gc_min_sleep_time),
299 ATTR_LIST(gc_max_sleep_time),
300 ATTR_LIST(gc_no_gc_sleep_time),
d2dc095f 301 ATTR_LIST(gc_idle),
ea91e9b0 302 ATTR_LIST(reclaim_segments),
7ac8c3b0 303 ATTR_LIST(max_small_discards),
bba681cb 304 ATTR_LIST(batched_trim_sections),
216fbd64
JK		 305 ATTR_LIST(ipu_policy),
306 ATTR_LIST(min_ipu_util),
c1ce1b02 307 ATTR_LIST(min_fsync_blocks),
b1c57c1c 308 ATTR_LIST(max_victim_search),
ab9fa662 309 ATTR_LIST(dir_level),
cdfc41c1 310 ATTR_LIST(ram_thresh),
ea1a29a0 311 ATTR_LIST(ra_nid_pages),
2304cb0c 312 ATTR_LIST(dirty_nats_ratio),
60b99b48 313 ATTR_LIST(cp_interval),
d0239e1b 314 ATTR_LIST(idle_interval),
8f1dbbbb 315 ATTR_LIST(lifetime_write_kbytes),
b59d0bae
NJ		 316 NULL,
317};
318
319static const struct sysfs_ops f2fs_attr_ops = {
320 .show = f2fs_attr_show,
321 .store = f2fs_attr_store,
322};
323
324static struct kobj_type f2fs_ktype = {
325 .default_attrs = f2fs_attrs,
326 .sysfs_ops = &f2fs_attr_ops,
327 .release = f2fs_sb_release,
328};
329
08796897
SY		 330#ifdef CONFIG_F2FS_FAULT_INJECTION
331/* sysfs for f2fs fault injection */
332static struct kobject f2fs_fault_inject;
333
334static struct attribute *f2fs_fault_attrs[] = {
335 ATTR_LIST(inject_rate),
336 ATTR_LIST(inject_type),
337 NULL
338};
339
340static struct kobj_type f2fs_fault_ktype = {
341 .default_attrs = f2fs_fault_attrs,
342 .sysfs_ops = &f2fs_attr_ops,
343};
344#endif
345
a07ef784
NJ		 346void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...)
347{
348 struct va_format vaf;
349 va_list args;
350
351 va_start(args, fmt);
352 vaf.fmt = fmt;
353 vaf.va = &args;
354 printk("%sF2FS-fs (%s): %pV\n", level, sb->s_id, &vaf);
355 va_end(args);
356}
357
aff063e2
JK		 358static void init_once(void *foo)
359{
360 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
361
aff063e2
JK		 362 inode_init_once(&fi->vfs_inode);
363}
364
696c018c
NJ		 365static int parse_options(struct super_block *sb, char *options)
366{
367 struct f2fs_sb_info *sbi = F2FS_SB(sb);
09d54cdd 368 struct request_queue *q;
696c018c
NJ		 369 substring_t args[MAX_OPT_ARGS];
370 char *p, *name;
371 int arg = 0;
372
73faec4d 373#ifdef CONFIG_F2FS_FAULT_INJECTION
08796897 374 f2fs_build_fault_attr(0);
73faec4d 375#endif
08796897 376
696c018c
NJ		 377 if (!options)
378 return 0;
379
380 while ((p = strsep(&options, ",")) != NULL) {
381 int token;
382 if (!*p)
383 continue;
384 /*
385 * Initialize args struct so we know whether arg was
386 * found; some options take optional arguments.
387 */
388 args[0].to = args[0].from = NULL;
389 token = match_token(p, f2fs_tokens, args);
390
391 switch (token) {
392 case Opt_gc_background:
393 name = match_strdup(&args[0]);
394
395 if (!name)
396 return -ENOMEM;
6aefd93b 397 if (strlen(name) == 2 && !strncmp(name, "on", 2)) {
696c018c 398 set_opt(sbi, BG_GC);
6aefd93b
JK		 399 clear_opt(sbi, FORCE_FG_GC);
400 } else if (strlen(name) == 3 && !strncmp(name, "off", 3)) {
696c018c 401 clear_opt(sbi, BG_GC);
6aefd93b
JK		 402 clear_opt(sbi, FORCE_FG_GC);
403 } else if (strlen(name) == 4 && !strncmp(name, "sync", 4)) {
404 set_opt(sbi, BG_GC);
405 set_opt(sbi, FORCE_FG_GC);
406 } else {
696c018c
NJ		 407 kfree(name);
408 return -EINVAL;
409 }
410 kfree(name);
411 break;
412 case Opt_disable_roll_forward:
413 set_opt(sbi, DISABLE_ROLL_FORWARD);
414 break;
2d834bf9
JK		 415 case Opt_norecovery:
416 /* this option mounts f2fs with ro */
417 set_opt(sbi, DISABLE_ROLL_FORWARD);
418 if (!f2fs_readonly(sb))
419 return -EINVAL;
420 break;
696c018c 421 case Opt_discard:
09d54cdd
CY		 422 q = bdev_get_queue(sb->s_bdev);
423 if (blk_queue_discard(q)) {
424 set_opt(sbi, DISCARD);
425 } else {
426 f2fs_msg(sb, KERN_WARNING,
427 "mounting with \"discard\" option, but "
428 "the device does not support discard");
429 }
696c018c 430 break;
64058be9
CY		 431 case Opt_nodiscard:
432 clear_opt(sbi, DISCARD);
696c018c
NJ		 433 case Opt_noheap:
434 set_opt(sbi, NOHEAP);
435 break;
436#ifdef CONFIG_F2FS_FS_XATTR
4058c511
KA		 437 case Opt_user_xattr:
438 set_opt(sbi, XATTR_USER);
439 break;
696c018c
NJ		 440 case Opt_nouser_xattr:
441 clear_opt(sbi, XATTR_USER);
442 break;
444c580f
JK		 443 case Opt_inline_xattr:
444 set_opt(sbi, INLINE_XATTR);
445 break;
696c018c 446#else
4058c511
KA		 447 case Opt_user_xattr:
448 f2fs_msg(sb, KERN_INFO,
449 "user_xattr options not supported");
450 break;
696c018c
NJ		 451 case Opt_nouser_xattr:
452 f2fs_msg(sb, KERN_INFO,
453 "nouser_xattr options not supported");
454 break;
444c580f
JK		 455 case Opt_inline_xattr:
456 f2fs_msg(sb, KERN_INFO,
457 "inline_xattr options not supported");
458 break;
696c018c
NJ		 459#endif
460#ifdef CONFIG_F2FS_FS_POSIX_ACL
4058c511
KA		 461 case Opt_acl:
462 set_opt(sbi, POSIX_ACL);
463 break;
696c018c
NJ		 464 case Opt_noacl:
465 clear_opt(sbi, POSIX_ACL);
466 break;
467#else
4058c511
KA		 468 case Opt_acl:
469 f2fs_msg(sb, KERN_INFO, "acl options not supported");
470 break;
696c018c
NJ		 471 case Opt_noacl:
472 f2fs_msg(sb, KERN_INFO, "noacl options not supported");
473 break;
474#endif
475 case Opt_active_logs:
476 if (args->from && match_int(args, &arg))
477 return -EINVAL;
478 if (arg != 2 && arg != 4 && arg != NR_CURSEG_TYPE)
479 return -EINVAL;
480 sbi->active_logs = arg;
481 break;
482 case Opt_disable_ext_identify:
483 set_opt(sbi, DISABLE_EXT_IDENTIFY);
484 break;
8274de77
HL		 485 case Opt_inline_data:
486 set_opt(sbi, INLINE_DATA);
487 break;
5efd3c6f
CY		 488 case Opt_inline_dentry:
489 set_opt(sbi, INLINE_DENTRY);
490 break;
6b4afdd7
JK		 491 case Opt_flush_merge:
492 set_opt(sbi, FLUSH_MERGE);
493 break;
69e9e427
JK		 494 case Opt_noflush_merge:
495 clear_opt(sbi, FLUSH_MERGE);
496 break;
0f7b2abd
JK		 497 case Opt_nobarrier:
498 set_opt(sbi, NOBARRIER);
499 break;
d5053a34
JK		 500 case Opt_fastboot:
501 set_opt(sbi, FASTBOOT);
502 break;
89672159
CY		 503 case Opt_extent_cache:
504 set_opt(sbi, EXTENT_CACHE);
505 break;
7daaea25
JK		 506 case Opt_noextent_cache:
507 clear_opt(sbi, EXTENT_CACHE);
508 break;
75342797
WL		 509 case Opt_noinline_data:
510 clear_opt(sbi, INLINE_DATA);
511 break;
343f40f0
CY		 512 case Opt_data_flush:
513 set_opt(sbi, DATA_FLUSH);
514 break;
36abef4e
JK		 515 case Opt_mode:
516 name = match_strdup(&args[0]);
517
518 if (!name)
519 return -ENOMEM;
520 if (strlen(name) == 8 &&
521 !strncmp(name, "adaptive", 8)) {
52763a4b 522 set_opt_mode(sbi, F2FS_MOUNT_ADAPTIVE);
36abef4e
JK		 523 } else if (strlen(name) == 3 &&
524 !strncmp(name, "lfs", 3)) {
52763a4b 525 set_opt_mode(sbi, F2FS_MOUNT_LFS);
36abef4e
JK		 526 } else {
527 kfree(name);
528 return -EINVAL;
529 }
530 kfree(name);
531 break;
73faec4d
JK		 532 case Opt_fault_injection:
533 if (args->from && match_int(args, &arg))
534 return -EINVAL;
535#ifdef CONFIG_F2FS_FAULT_INJECTION
08796897 536 f2fs_build_fault_attr(arg);
73faec4d
JK		 537#else
538 f2fs_msg(sb, KERN_INFO,
539 "FAULT_INJECTION was not selected");
540#endif
541 break;
6d94c74a
JK		 542 case Opt_lazytime:
543 sb->s_flags |= MS_LAZYTIME;
544 break;
545 case Opt_nolazytime:
546 sb->s_flags &= ~MS_LAZYTIME;
547 break;
696c018c
NJ		 548 default:
549 f2fs_msg(sb, KERN_ERR,
550 "Unrecognized mount option \"%s\" or missing value",
551 p);
552 return -EINVAL;
553 }
554 }
555 return 0;
556}
557
aff063e2
JK		 558static struct inode *f2fs_alloc_inode(struct super_block *sb)
559{
560 struct f2fs_inode_info *fi;
561
a0acdfe0 562 fi = kmem_cache_alloc(f2fs_inode_cachep, GFP_F2FS_ZERO);
aff063e2
JK		 563 if (!fi)
564 return NULL;
565
566 init_once((void *) fi);
567
1beba1b3
JK		 568 if (percpu_counter_init(&fi->dirty_pages, 0, GFP_NOFS)) {
569 kmem_cache_free(f2fs_inode_cachep, fi);
570 return NULL;
571 }
572
434720fa 573 /* Initialize f2fs-specific inode info */
aff063e2 574 fi->vfs_inode.i_version = 1;
aff063e2
JK		 575 fi->i_current_depth = 1;
576 fi->i_advise = 0;
d928bfbf 577 init_rwsem(&fi->i_sem);
2710fd7e 578 INIT_LIST_HEAD(&fi->dirty_list);
0f18b462 579 INIT_LIST_HEAD(&fi->gdirty_list);
88b88a66
JK		 580 INIT_LIST_HEAD(&fi->inmem_pages);
581 mutex_init(&fi->inmem_lock);
aff063e2 582
ab9fa662
JK		 583 /* Will be used by directory only */
584 fi->i_dir_level = F2FS_SB(sb)->dir_level;
aff063e2
JK		 585 return &fi->vfs_inode;
586}
587
531ad7d5
JK		 588static int f2fs_drop_inode(struct inode *inode)
589{
590 /*
591 * This is to avoid a deadlock condition like below.
592 * writeback_single_inode(inode)
593 * - f2fs_write_data_page
594 * - f2fs_gc -> iput -> evict
595 * - inode_wait_for_writeback(inode)
596 */
0f18b462 597 if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
06e1bc05 598 if (!inode->i_nlink && !is_bad_inode(inode)) {
3e72f721
JK		 599 /* to avoid evict_inode call simultaneously */
600 atomic_inc(&inode->i_count);
06e1bc05
JK		 601 spin_unlock(&inode->i_lock);
602
603 /* some remained atomic pages should discarded */
604 if (f2fs_is_atomic_file(inode))
29b96b54 605 drop_inmem_pages(inode);
06e1bc05 606
3e72f721
JK		 607 /* should remain fi->extent_tree for writepage */
608 f2fs_destroy_extent_node(inode);
609
06e1bc05 610 sb_start_intwrite(inode->i_sb);
fc9581c8 611 f2fs_i_size_write(inode, 0);
06e1bc05
JK		 612
613 if (F2FS_HAS_BLOCKS(inode))
9a449e9c 614 f2fs_truncate(inode);
06e1bc05
JK		 615
616 sb_end_intwrite(inode->i_sb);
617
0b81d077 618 fscrypt_put_encryption_info(inode, NULL);
06e1bc05 619 spin_lock(&inode->i_lock);
3e72f721 620 atomic_dec(&inode->i_count);
06e1bc05 621 }
531ad7d5 622 return 0;
06e1bc05 623 }
0f18b462 624
67c3758d 625 return generic_drop_inode(inode);
531ad7d5
JK		 626}
627
b3783873
JK		 628/*
629 * f2fs_dirty_inode() is called from __mark_inode_dirty()
630 *
631 * We should call set_dirty_inode to write the dirty inode through write_inode.
632 */
633static void f2fs_dirty_inode(struct inode *inode, int flags)
634{
0f18b462
JK		 635 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
636
637 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
638 inode->i_ino == F2FS_META_INO(sbi))
639 return;
640
6d94c74a
JK		 641 if (flags == I_DIRTY_TIME)
642 return;
643
26de9b11
JK		 644 if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
645 clear_inode_flag(inode, FI_AUTO_RECOVER);
646
0f18b462
JK		 647 spin_lock(&sbi->inode_lock[DIRTY_META]);
648 if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
649 spin_unlock(&sbi->inode_lock[DIRTY_META]);
650 return;
651 }
652
91942321 653 set_inode_flag(inode, FI_DIRTY_INODE);
0f18b462
JK		 654 list_add_tail(&F2FS_I(inode)->gdirty_list,
655 &sbi->inode_list[DIRTY_META]);
656 inc_page_count(sbi, F2FS_DIRTY_IMETA);
0f18b462 657 stat_inc_dirty_inode(sbi, DIRTY_META);
338bbfa0 658 spin_unlock(&sbi->inode_lock[DIRTY_META]);
0f18b462
JK		 659}
660
661void f2fs_inode_synced(struct inode *inode)
662{
663 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
664
665 spin_lock(&sbi->inode_lock[DIRTY_META]);
666 if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
667 spin_unlock(&sbi->inode_lock[DIRTY_META]);
668 return;
669 }
670 list_del_init(&F2FS_I(inode)->gdirty_list);
671 clear_inode_flag(inode, FI_DIRTY_INODE);
26de9b11 672 clear_inode_flag(inode, FI_AUTO_RECOVER);
0f18b462 673 dec_page_count(sbi, F2FS_DIRTY_IMETA);
0f18b462 674 stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
338bbfa0 675 spin_unlock(&sbi->inode_lock[DIRTY_META]);
b3783873
JK		 676}
677
aff063e2
JK		 678static void f2fs_i_callback(struct rcu_head *head)
679{
680 struct inode *inode = container_of(head, struct inode, i_rcu);
681 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
682}
683
25ca923b 684static void f2fs_destroy_inode(struct inode *inode)
aff063e2 685{
1beba1b3 686 percpu_counter_destroy(&F2FS_I(inode)->dirty_pages);
aff063e2
JK		 687 call_rcu(&inode->i_rcu, f2fs_i_callback);
688}
689
523be8a6
JK		 690static void destroy_percpu_info(struct f2fs_sb_info *sbi)
691{
692 int i;
693
694 for (i = 0; i < NR_COUNT_TYPE; i++)
695 percpu_counter_destroy(&sbi->nr_pages[i]);
41382ec4 696 percpu_counter_destroy(&sbi->alloc_valid_block_count);
513c5f37 697 percpu_counter_destroy(&sbi->total_valid_inode_count);
ec795418
JK		 698
699 percpu_free_rwsem(&sbi->cp_rwsem);
523be8a6
JK		 700}
701
aff063e2
JK		 702static void f2fs_put_super(struct super_block *sb)
703{
704 struct f2fs_sb_info *sbi = F2FS_SB(sb);
705
5e176d54
JK		 706 if (sbi->s_proc) {
707 remove_proc_entry("segment_info", sbi->s_proc);
f00d6fa7 708 remove_proc_entry("segment_bits", sbi->s_proc);
5e176d54
JK		 709 remove_proc_entry(sb->s_id, f2fs_proc_root);
710 }
b59d0bae 711 kobject_del(&sbi->s_kobj);
5e176d54 712
aff063e2
JK		 713 stop_gc_thread(sbi);
714
2658e50d
JK		 715 /* prevent remaining shrinker jobs */
716 mutex_lock(&sbi->umount_mutex);
717
85dc2f2c
JK		 718 /*
719 * We don't need to do checkpoint when superblock is clean.
720 * But, the previous checkpoint was not done by umount, it needs to do
721 * clean checkpoint again.
722 */
caf0047e 723 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
85dc2f2c 724 !is_set_ckpt_flags(F2FS_CKPT(sbi), CP_UMOUNT_FLAG)) {
75ab4cb8
JK		 725 struct cp_control cpc = {
726 .reason = CP_UMOUNT,
727 };
728 write_checkpoint(sbi, &cpc);
729 }
aff063e2 730
eca616f8
JK		 731 /* write_checkpoint can update stat informaion */
732 f2fs_destroy_stats(sbi);
733
cf779cab
JK		 734 /*
735 * normally superblock is clean, so we need to release this.
736 * In addition, EIO will skip do checkpoint, we need this as well.
737 */
74ef9241 738 release_ino_entry(sbi, true);
4b2fecc8 739 release_discard_addrs(sbi);
6f12ac25 740
2658e50d
JK		 741 f2fs_leave_shrinker(sbi);
742 mutex_unlock(&sbi->umount_mutex);
743
17c19120 744 /* our cp_error case, we can wait for any writeback page */
f5730184 745 f2fs_flush_merged_bios(sbi);
17c19120 746
aff063e2
JK		 747 iput(sbi->node_inode);
748 iput(sbi->meta_inode);
749
750 /* destroy f2fs internal modules */
751 destroy_node_manager(sbi);
752 destroy_segment_manager(sbi);
753
754 kfree(sbi->ckpt);
b59d0bae
NJ		 755 kobject_put(&sbi->s_kobj);
756 wait_for_completion(&sbi->s_kobj_unregister);
aff063e2
JK		 757
758 sb->s_fs_info = NULL;
43b6573b
KM		 759 if (sbi->s_chksum_driver)
760 crypto_free_shash(sbi->s_chksum_driver);
b39f0de2 761 kfree(sbi->raw_super);
523be8a6
JK		 762
763 destroy_percpu_info(sbi);
aff063e2
JK		 764 kfree(sbi);
765}
766
767int f2fs_sync_fs(struct super_block *sb, int sync)
768{
769 struct f2fs_sb_info *sbi = F2FS_SB(sb);
c34f42e2 770 int err = 0;
aff063e2 771
a2a4a7e4
NJ		 772 trace_f2fs_sync_fs(sb, sync);
773
b7473754 774 if (sync) {
d5053a34
JK		 775 struct cp_control cpc;
776
119ee914
JK		 777 cpc.reason = __get_cp_reason(sbi);
778
b7473754 779 mutex_lock(&sbi->gc_mutex);
c34f42e2 780 err = write_checkpoint(sbi, &cpc);
b7473754 781 mutex_unlock(&sbi->gc_mutex);
b7473754 782 }
05ca3632 783 f2fs_trace_ios(NULL, 1);
aff063e2 784
c34f42e2 785 return err;
aff063e2
JK		 786}
787
d6212a5f
CL		 788static int f2fs_freeze(struct super_block *sb)
789{
790 int err;
791
77888c1e 792 if (f2fs_readonly(sb))
d6212a5f
CL		 793 return 0;
794
795 err = f2fs_sync_fs(sb, 1);
796 return err;
797}
798
799static int f2fs_unfreeze(struct super_block *sb)
800{
801 return 0;
802}
803
aff063e2
JK		 804static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
805{
806 struct super_block *sb = dentry->d_sb;
807 struct f2fs_sb_info *sbi = F2FS_SB(sb);
808 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
809 block_t total_count, user_block_count, start_count, ovp_count;
810
811 total_count = le64_to_cpu(sbi->raw_super->block_count);
812 user_block_count = sbi->user_block_count;
813 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
814 ovp_count = SM_I(sbi)->ovp_segments << sbi->log_blocks_per_seg;
815 buf->f_type = F2FS_SUPER_MAGIC;
816 buf->f_bsize = sbi->blocksize;
817
818 buf->f_blocks = total_count - start_count;
819 buf->f_bfree = buf->f_blocks - valid_user_blocks(sbi) - ovp_count;
820 buf->f_bavail = user_block_count - valid_user_blocks(sbi);
821
c200b1aa
CY		 822 buf->f_files = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
823 buf->f_ffree = buf->f_files - valid_inode_count(sbi);
aff063e2 824
5a20d339 825 buf->f_namelen = F2FS_NAME_LEN;
aff063e2
JK		 826 buf->f_fsid.val[0] = (u32)id;
827 buf->f_fsid.val[1] = (u32)(id >> 32);
828
829 return 0;
830}
831
832static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
833{
834 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
835
6aefd93b
JK		 836 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, BG_GC)) {
837 if (test_opt(sbi, FORCE_FG_GC))
838 seq_printf(seq, ",background_gc=%s", "sync");
839 else
840 seq_printf(seq, ",background_gc=%s", "on");
841 } else {
696c018c 842 seq_printf(seq, ",background_gc=%s", "off");
6aefd93b 843 }
aff063e2
JK		 844 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
845 seq_puts(seq, ",disable_roll_forward");
846 if (test_opt(sbi, DISCARD))
847 seq_puts(seq, ",discard");
848 if (test_opt(sbi, NOHEAP))
849 seq_puts(seq, ",no_heap_alloc");
850#ifdef CONFIG_F2FS_FS_XATTR
851 if (test_opt(sbi, XATTR_USER))
852 seq_puts(seq, ",user_xattr");
853 else
854 seq_puts(seq, ",nouser_xattr");
444c580f
JK		 855 if (test_opt(sbi, INLINE_XATTR))
856 seq_puts(seq, ",inline_xattr");
aff063e2
JK		 857#endif
858#ifdef CONFIG_F2FS_FS_POSIX_ACL
859 if (test_opt(sbi, POSIX_ACL))
860 seq_puts(seq, ",acl");
861 else
862 seq_puts(seq, ",noacl");
863#endif
864 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
aa43507f 865 seq_puts(seq, ",disable_ext_identify");
8274de77
HL		 866 if (test_opt(sbi, INLINE_DATA))
867 seq_puts(seq, ",inline_data");
75342797
WL		 868 else
869 seq_puts(seq, ",noinline_data");
5efd3c6f
CY		 870 if (test_opt(sbi, INLINE_DENTRY))
871 seq_puts(seq, ",inline_dentry");
b270ad6f 872 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
6b4afdd7 873 seq_puts(seq, ",flush_merge");
0f7b2abd
JK		 874 if (test_opt(sbi, NOBARRIER))
875 seq_puts(seq, ",nobarrier");
d5053a34
JK		 876 if (test_opt(sbi, FASTBOOT))
877 seq_puts(seq, ",fastboot");
89672159
CY		 878 if (test_opt(sbi, EXTENT_CACHE))
879 seq_puts(seq, ",extent_cache");
7daaea25
JK		 880 else
881 seq_puts(seq, ",noextent_cache");
343f40f0
CY		 882 if (test_opt(sbi, DATA_FLUSH))
883 seq_puts(seq, ",data_flush");
36abef4e
JK		 884
885 seq_puts(seq, ",mode=");
886 if (test_opt(sbi, ADAPTIVE))
887 seq_puts(seq, "adaptive");
888 else if (test_opt(sbi, LFS))
889 seq_puts(seq, "lfs");
aff063e2
JK		 890 seq_printf(seq, ",active_logs=%u", sbi->active_logs);
891
892 return 0;
893}
894
5e176d54
JK		 895static int segment_info_seq_show(struct seq_file *seq, void *offset)
896{
897 struct super_block *sb = seq->private;
898 struct f2fs_sb_info *sbi = F2FS_SB(sb);
6c311ec6
CF		 899 unsigned int total_segs =
900 le32_to_cpu(sbi->raw_super->segment_count_main);
5e176d54
JK		 901 int i;
902
90aa6dc9
CY		 903 seq_puts(seq, "format: segment_type|valid_blocks\n"
904 "segment_type(0:HD, 1:WD, 2:CD, 3:HN, 4:WN, 5:CN)\n");
905
5e176d54 906 for (i = 0; i < total_segs; i++) {
90aa6dc9
CY		 907 struct seg_entry *se = get_seg_entry(sbi, i);
908
909 if ((i % 10) == 0)
01a5ad82 910 seq_printf(seq, "%-10d", i);
90aa6dc9
CY		 911 seq_printf(seq, "%d|%-3u", se->type,
912 get_valid_blocks(sbi, i, 1));
46c04366
GZ		 913 if ((i % 10) == 9 || i == (total_segs - 1))
914 seq_putc(seq, '\n');
5e176d54 915 else
46c04366 916 seq_putc(seq, ' ');
5e176d54 917 }
46c04366 918
5e176d54
JK		 919 return 0;
920}
921
f00d6fa7
JK		 922static int segment_bits_seq_show(struct seq_file *seq, void *offset)
923{
924 struct super_block *sb = seq->private;
925 struct f2fs_sb_info *sbi = F2FS_SB(sb);
926 unsigned int total_segs =
927 le32_to_cpu(sbi->raw_super->segment_count_main);
928 int i, j;
929
930 seq_puts(seq, "format: segment_type|valid_blocks|bitmaps\n"
931 "segment_type(0:HD, 1:WD, 2:CD, 3:HN, 4:WN, 5:CN)\n");
932
933 for (i = 0; i < total_segs; i++) {
934 struct seg_entry *se = get_seg_entry(sbi, i);
935
936 seq_printf(seq, "%-10d", i);
937 seq_printf(seq, "%d|%-3u|", se->type,
938 get_valid_blocks(sbi, i, 1));
939 for (j = 0; j < SIT_VBLOCK_MAP_SIZE; j++)
940 seq_printf(seq, "%x ", se->cur_valid_map[j]);
941 seq_putc(seq, '\n');
942 }
943 return 0;
944}
945
b7a15f3d
JK		 946#define F2FS_PROC_FILE_DEF(_name) \
947static int _name##_open_fs(struct inode *inode, struct file *file) \
948{ \
949 return single_open(file, _name##_seq_show, PDE_DATA(inode)); \
950} \
951 \
952static const struct file_operations f2fs_seq_##_name##_fops = { \
953 .owner = THIS_MODULE, \
954 .open = _name##_open_fs, \
955 .read = seq_read, \
956 .llseek = seq_lseek, \
957 .release = single_release, \
5e176d54
JK		 958};
959
b7a15f3d 960F2FS_PROC_FILE_DEF(segment_info);
f00d6fa7 961F2FS_PROC_FILE_DEF(segment_bits);
b7a15f3d 962
498c5e9f
YH		 963static void default_options(struct f2fs_sb_info *sbi)
964{
965 /* init some FS parameters */
966 sbi->active_logs = NR_CURSEG_TYPE;
967
968 set_opt(sbi, BG_GC);
969 set_opt(sbi, INLINE_DATA);
3e72f721 970 set_opt(sbi, EXTENT_CACHE);
6d94c74a 971 sbi->sb->s_flags |= MS_LAZYTIME;
69e9e427 972 set_opt(sbi, FLUSH_MERGE);
52763a4b
JK		 973 if (f2fs_sb_mounted_hmsmr(sbi->sb)) {
974 set_opt_mode(sbi, F2FS_MOUNT_LFS);
975 set_opt(sbi, DISCARD);
976 } else {
977 set_opt_mode(sbi, F2FS_MOUNT_ADAPTIVE);
978 }
498c5e9f
YH		 979
980#ifdef CONFIG_F2FS_FS_XATTR
981 set_opt(sbi, XATTR_USER);
982#endif
983#ifdef CONFIG_F2FS_FS_POSIX_ACL
984 set_opt(sbi, POSIX_ACL);
985#endif
986}
987
696c018c
NJ		 988static int f2fs_remount(struct super_block *sb, int *flags, char *data)
989{
990 struct f2fs_sb_info *sbi = F2FS_SB(sb);
991 struct f2fs_mount_info org_mount_opt;
992 int err, active_logs;
876dc59e
GZ		 993 bool need_restart_gc = false;
994 bool need_stop_gc = false;
9cd81ce3 995 bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
696c018c
NJ		 996
997 /*
998 * Save the old mount options in case we
999 * need to restore them.
1000 */
1001 org_mount_opt = sbi->mount_opt;
1002 active_logs = sbi->active_logs;
1003
df728b0f
JK		 1004 /* recover superblocks we couldn't write due to previous RO mount */
1005 if (!(*flags & MS_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
1006 err = f2fs_commit_super(sbi, false);
1007 f2fs_msg(sb, KERN_INFO,
1008 "Try to recover all the superblocks, ret: %d", err);
1009 if (!err)
1010 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
1011 }
1012
26666c8a 1013 sbi->mount_opt.opt = 0;
498c5e9f 1014 default_options(sbi);
26666c8a 1015
696c018c
NJ		 1016 /* parse mount options */
1017 err = parse_options(sb, data);
1018 if (err)
1019 goto restore_opts;
1020
1021 /*
1022 * Previous and new state of filesystem is RO,
876dc59e 1023 * so skip checking GC and FLUSH_MERGE conditions.
696c018c 1024 */
6b2920a5 1025 if (f2fs_readonly(sb) && (*flags & MS_RDONLY))
696c018c
NJ		 1026 goto skip;
1027
9cd81ce3
CY		 1028 /* disallow enable/disable extent_cache dynamically */
1029 if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
1030 err = -EINVAL;
1031 f2fs_msg(sbi->sb, KERN_WARNING,
1032 "switch extent_cache option is not allowed");
1033 goto restore_opts;
1034 }
1035
696c018c
NJ		 1036 /*
1037 * We stop the GC thread if FS is mounted as RO
1038 * or if background_gc = off is passed in mount
1039 * option. Also sync the filesystem.
1040 */
1041 if ((*flags & MS_RDONLY) || !test_opt(sbi, BG_GC)) {
1042 if (sbi->gc_thread) {
1043 stop_gc_thread(sbi);
876dc59e 1044 need_restart_gc = true;
696c018c 1045 }
aba291b3 1046 } else if (!sbi->gc_thread) {
696c018c
NJ		 1047 err = start_gc_thread(sbi);
1048 if (err)
1049 goto restore_opts;
876dc59e
GZ		 1050 need_stop_gc = true;
1051 }
1052
faa0e55b
JK		 1053 if (*flags & MS_RDONLY) {
1054 writeback_inodes_sb(sb, WB_REASON_SYNC);
1055 sync_inodes_sb(sb);
1056
1057 set_sbi_flag(sbi, SBI_IS_DIRTY);
1058 set_sbi_flag(sbi, SBI_IS_CLOSE);
1059 f2fs_sync_fs(sb, 1);
1060 clear_sbi_flag(sbi, SBI_IS_CLOSE);
1061 }
1062
876dc59e
GZ		 1063 /*
1064 * We stop issue flush thread if FS is mounted as RO
1065 * or if flush_merge is not passed in mount option.
1066 */
1067 if ((*flags & MS_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2163d198 1068 destroy_flush_cmd_control(sbi);
aba291b3 1069 } else if (!SM_I(sbi)->cmd_control_info) {
2163d198
GZ		 1070 err = create_flush_cmd_control(sbi);
1071 if (err)
a688b9d9 1072 goto restore_gc;
696c018c
NJ		 1073 }
1074skip:
1075 /* Update the POSIXACL Flag */
df728b0f 1076 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
696c018c 1077 (test_opt(sbi, POSIX_ACL) ? MS_POSIXACL : 0);
df728b0f 1078
696c018c 1079 return 0;
876dc59e
GZ		 1080restore_gc:
1081 if (need_restart_gc) {
1082 if (start_gc_thread(sbi))
1083 f2fs_msg(sbi->sb, KERN_WARNING,
e1c42045 1084 "background gc thread has stopped");
876dc59e
GZ		 1085 } else if (need_stop_gc) {
1086 stop_gc_thread(sbi);
1087 }
696c018c
NJ		 1088restore_opts:
1089 sbi->mount_opt = org_mount_opt;
1090 sbi->active_logs = active_logs;
1091 return err;
1092}
1093
aff063e2
JK		 1094static struct super_operations f2fs_sops = {
1095 .alloc_inode = f2fs_alloc_inode,
531ad7d5 1096 .drop_inode = f2fs_drop_inode,
aff063e2
JK		 1097 .destroy_inode = f2fs_destroy_inode,
1098 .write_inode = f2fs_write_inode,
b3783873 1099 .dirty_inode = f2fs_dirty_inode,
aff063e2
JK		 1100 .show_options = f2fs_show_options,
1101 .evict_inode = f2fs_evict_inode,
1102 .put_super = f2fs_put_super,
1103 .sync_fs = f2fs_sync_fs,
d6212a5f
CL		 1104 .freeze_fs = f2fs_freeze,
1105 .unfreeze_fs = f2fs_unfreeze,
aff063e2 1106 .statfs = f2fs_statfs,
696c018c 1107 .remount_fs = f2fs_remount,
aff063e2
JK		 1108};
1109
0b81d077
JK		 1110#ifdef CONFIG_F2FS_FS_ENCRYPTION
1111static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
1112{
1113 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
1114 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
1115 ctx, len, NULL);
1116}
1117
b5a7aef1
JK		 1118static int f2fs_key_prefix(struct inode *inode, u8 **key)
1119{
1120 *key = F2FS_I_SB(inode)->key_prefix;
1121 return F2FS_I_SB(inode)->key_prefix_size;
1122}
1123
0b81d077
JK		 1124static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
1125 void *fs_data)
1126{
1127 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
1128 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
1129 ctx, len, fs_data, XATTR_CREATE);
1130}
1131
1132static unsigned f2fs_max_namelen(struct inode *inode)
1133{
1134 return S_ISLNK(inode->i_mode) ?
1135 inode->i_sb->s_blocksize : F2FS_NAME_LEN;
1136}
1137
1138static struct fscrypt_operations f2fs_cryptops = {
1139 .get_context = f2fs_get_context,
b5a7aef1 1140 .key_prefix = f2fs_key_prefix,
0b81d077
JK		 1141 .set_context = f2fs_set_context,
1142 .is_encrypted = f2fs_encrypted_inode,
1143 .empty_dir = f2fs_empty_dir,
1144 .max_namelen = f2fs_max_namelen,
1145};
1146#else
1147static struct fscrypt_operations f2fs_cryptops = {
1148 .is_encrypted = f2fs_encrypted_inode,
1149};
1150#endif
1151
aff063e2
JK		 1152static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
1153 u64 ino, u32 generation)
1154{
1155 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1156 struct inode *inode;
1157
d6b7d4b3 1158 if (check_nid_range(sbi, ino))
910bb12d 1159 return ERR_PTR(-ESTALE);
aff063e2
JK		 1160
1161 /*
1162 * f2fs_iget isn't quite right if the inode is currently unallocated!
1163 * However f2fs_iget currently does appropriate checks to handle stale
1164 * inodes so everything is OK.
1165 */
1166 inode = f2fs_iget(sb, ino);
1167 if (IS_ERR(inode))
1168 return ERR_CAST(inode);
6bacf52f 1169 if (unlikely(generation && inode->i_generation != generation)) {
aff063e2
JK		 1170 /* we didn't find the right inode.. */
1171 iput(inode);
1172 return ERR_PTR(-ESTALE);
1173 }
1174 return inode;
1175}
1176
1177static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
1178 int fh_len, int fh_type)
1179{
1180 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
1181 f2fs_nfs_get_inode);
1182}
1183
1184static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
1185 int fh_len, int fh_type)
1186{
1187 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
1188 f2fs_nfs_get_inode);
1189}
1190
1191static const struct export_operations f2fs_export_ops = {
1192 .fh_to_dentry = f2fs_fh_to_dentry,
1193 .fh_to_parent = f2fs_fh_to_parent,
1194 .get_parent = f2fs_get_parent,
1195};
1196
e0afc4d6 1197static loff_t max_file_blocks(void)
aff063e2 1198{
de93653f 1199 loff_t result = (DEF_ADDRS_PER_INODE - F2FS_INLINE_XATTR_ADDRS);
aff063e2
JK		 1200 loff_t leaf_count = ADDRS_PER_BLOCK;
1201
1202 /* two direct node blocks */
1203 result += (leaf_count * 2);
1204
1205 /* two indirect node blocks */
1206 leaf_count *= NIDS_PER_BLOCK;
1207 result += (leaf_count * 2);
1208
1209 /* one double indirect node block */
1210 leaf_count *= NIDS_PER_BLOCK;
1211 result += leaf_count;
1212
aff063e2
JK		 1213 return result;
1214}
1215
fd694733
JK		 1216static int __f2fs_commit_super(struct buffer_head *bh,
1217 struct f2fs_super_block *super)
1218{
1219 lock_buffer(bh);
1220 if (super)
1221 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
1222 set_buffer_uptodate(bh);
1223 set_buffer_dirty(bh);
1224 unlock_buffer(bh);
1225
1226 /* it's rare case, we can do fua all the time */
1227 return __sync_dirty_buffer(bh, WRITE_FLUSH_FUA);
1228}
1229
df728b0f 1230static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
fd694733 1231 struct buffer_head *bh)
9a59b62f 1232{
fd694733
JK		 1233 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
1234 (bh->b_data + F2FS_SUPER_OFFSET);
df728b0f 1235 struct super_block *sb = sbi->sb;
9a59b62f
CY		 1236 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
1237 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
1238 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
1239 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
1240 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
1241 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
1242 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
1243 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
1244 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
1245 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
1246 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
1247 u32 segment_count = le32_to_cpu(raw_super->segment_count);
1248 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
fd694733
JK		 1249 u64 main_end_blkaddr = main_blkaddr +
1250 (segment_count_main << log_blocks_per_seg);
1251 u64 seg_end_blkaddr = segment0_blkaddr +
1252 (segment_count << log_blocks_per_seg);
9a59b62f
CY		 1253
1254 if (segment0_blkaddr != cp_blkaddr) {
1255 f2fs_msg(sb, KERN_INFO,
1256 "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
1257 segment0_blkaddr, cp_blkaddr);
1258 return true;
1259 }
1260
1261 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
1262 sit_blkaddr) {
1263 f2fs_msg(sb, KERN_INFO,
1264 "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
1265 cp_blkaddr, sit_blkaddr,
1266 segment_count_ckpt << log_blocks_per_seg);
1267 return true;
1268 }
1269
1270 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
1271 nat_blkaddr) {
1272 f2fs_msg(sb, KERN_INFO,
1273 "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
1274 sit_blkaddr, nat_blkaddr,
1275 segment_count_sit << log_blocks_per_seg);
1276 return true;
1277 }
1278
1279 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
1280 ssa_blkaddr) {
1281 f2fs_msg(sb, KERN_INFO,
1282 "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
1283 nat_blkaddr, ssa_blkaddr,
1284 segment_count_nat << log_blocks_per_seg);
1285 return true;
1286 }
1287
1288 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
1289 main_blkaddr) {
1290 f2fs_msg(sb, KERN_INFO,
1291 "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
1292 ssa_blkaddr, main_blkaddr,
1293 segment_count_ssa << log_blocks_per_seg);
1294 return true;
1295 }
1296
fd694733 1297 if (main_end_blkaddr > seg_end_blkaddr) {
9a59b62f 1298 f2fs_msg(sb, KERN_INFO,
fd694733 1299 "Wrong MAIN_AREA boundary, start(%u) end(%u) block(%u)",
9a59b62f 1300 main_blkaddr,
fd694733
JK		 1301 segment0_blkaddr +
1302 (segment_count << log_blocks_per_seg),
9a59b62f
CY		 1303 segment_count_main << log_blocks_per_seg);
1304 return true;
fd694733
JK		 1305 } else if (main_end_blkaddr < seg_end_blkaddr) {
1306 int err = 0;
1307 char *res;
1308
1309 /* fix in-memory information all the time */
1310 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
1311 segment0_blkaddr) >> log_blocks_per_seg);
1312
1313 if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
df728b0f 1314 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
fd694733
JK		 1315 res = "internally";
1316 } else {
1317 err = __f2fs_commit_super(bh, NULL);
1318 res = err ? "failed" : "done";
1319 }
1320 f2fs_msg(sb, KERN_INFO,
1321 "Fix alignment : %s, start(%u) end(%u) block(%u)",
1322 res, main_blkaddr,
1323 segment0_blkaddr +
1324 (segment_count << log_blocks_per_seg),
1325 segment_count_main << log_blocks_per_seg);
1326 if (err)
1327 return true;
9a59b62f 1328 }
9a59b62f
CY		 1329 return false;
1330}
1331
df728b0f 1332static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
fd694733 1333 struct buffer_head *bh)
aff063e2 1334{
fd694733
JK		 1335 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
1336 (bh->b_data + F2FS_SUPER_OFFSET);
df728b0f 1337 struct super_block *sb = sbi->sb;
aff063e2
JK		 1338 unsigned int blocksize;
1339
a07ef784
NJ		 1340 if (F2FS_SUPER_MAGIC != le32_to_cpu(raw_super->magic)) {
1341 f2fs_msg(sb, KERN_INFO,
1342 "Magic Mismatch, valid(0x%x) - read(0x%x)",
1343 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
aff063e2 1344 return 1;
a07ef784 1345 }
aff063e2 1346
5c9b4692 1347 /* Currently, support only 4KB page cache size */
09cbfeaf 1348 if (F2FS_BLKSIZE != PAGE_SIZE) {
5c9b4692 1349 f2fs_msg(sb, KERN_INFO,
14d7e9de 1350 "Invalid page_cache_size (%lu), supports only 4KB\n",
09cbfeaf 1351 PAGE_SIZE);
5c9b4692 1352 return 1;
1353 }
1354
aff063e2
JK		 1355 /* Currently, support only 4KB block size */
1356 blocksize = 1 << le32_to_cpu(raw_super->log_blocksize);
5c9b4692 1357 if (blocksize != F2FS_BLKSIZE) {
a07ef784
NJ		 1358 f2fs_msg(sb, KERN_INFO,
1359 "Invalid blocksize (%u), supports only 4KB\n",
1360 blocksize);
aff063e2 1361 return 1;
a07ef784 1362 }
5c9b4692 1363
9a59b62f
CY		 1364 /* check log blocks per segment */
1365 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
1366 f2fs_msg(sb, KERN_INFO,
1367 "Invalid log blocks per segment (%u)\n",
1368 le32_to_cpu(raw_super->log_blocks_per_seg));
1369 return 1;
1370 }
1371
55cf9cb6
CY		 1372 /* Currently, support 512/1024/2048/4096 bytes sector size */
1373 if (le32_to_cpu(raw_super->log_sectorsize) >
1374 F2FS_MAX_LOG_SECTOR_SIZE ||
1375 le32_to_cpu(raw_super->log_sectorsize) <
1376 F2FS_MIN_LOG_SECTOR_SIZE) {
1377 f2fs_msg(sb, KERN_INFO, "Invalid log sectorsize (%u)",
1378 le32_to_cpu(raw_super->log_sectorsize));
aff063e2 1379 return 1;
a07ef784 1380 }
55cf9cb6
CY		 1381 if (le32_to_cpu(raw_super->log_sectors_per_block) +
1382 le32_to_cpu(raw_super->log_sectorsize) !=
1383 F2FS_MAX_LOG_SECTOR_SIZE) {
1384 f2fs_msg(sb, KERN_INFO,
1385 "Invalid log sectors per block(%u) log sectorsize(%u)",
1386 le32_to_cpu(raw_super->log_sectors_per_block),
1387 le32_to_cpu(raw_super->log_sectorsize));
aff063e2 1388 return 1;
a07ef784 1389 }
9a59b62f
CY		 1390
1391 /* check reserved ino info */
1392 if (le32_to_cpu(raw_super->node_ino) != 1 ||
1393 le32_to_cpu(raw_super->meta_ino) != 2 ||
1394 le32_to_cpu(raw_super->root_ino) != 3) {
1395 f2fs_msg(sb, KERN_INFO,
1396 "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
1397 le32_to_cpu(raw_super->node_ino),
1398 le32_to_cpu(raw_super->meta_ino),
1399 le32_to_cpu(raw_super->root_ino));
1400 return 1;
1401 }
1402
1403 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
df728b0f 1404 if (sanity_check_area_boundary(sbi, bh))
9a59b62f
CY		 1405 return 1;
1406
aff063e2
JK		 1407 return 0;
1408}
1409
984ec63c 1410int sanity_check_ckpt(struct f2fs_sb_info *sbi)
aff063e2
JK		 1411{
1412 unsigned int total, fsmeta;
577e3495
JK		 1413 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
1414 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
aff063e2
JK		 1415
1416 total = le32_to_cpu(raw_super->segment_count);
1417 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
1418 fsmeta += le32_to_cpu(raw_super->segment_count_sit);
1419 fsmeta += le32_to_cpu(raw_super->segment_count_nat);
1420 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
1421 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
1422
6bacf52f 1423 if (unlikely(fsmeta >= total))
aff063e2 1424 return 1;
577e3495 1425
1e968fdf 1426 if (unlikely(f2fs_cp_error(sbi))) {
577e3495
JK		 1427 f2fs_msg(sbi->sb, KERN_ERR, "A bug case: need to run fsck");
1428 return 1;
1429 }
aff063e2
JK		 1430 return 0;
1431}
1432
1433static void init_sb_info(struct f2fs_sb_info *sbi)
1434{
1435 struct f2fs_super_block *raw_super = sbi->raw_super;
aff063e2
JK		 1436
1437 sbi->log_sectors_per_block =
1438 le32_to_cpu(raw_super->log_sectors_per_block);
1439 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
1440 sbi->blocksize = 1 << sbi->log_blocksize;
1441 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
1442 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
1443 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
1444 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
1445 sbi->total_sections = le32_to_cpu(raw_super->section_count);
1446 sbi->total_node_count =
1447 (le32_to_cpu(raw_super->segment_count_nat) / 2)
1448 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
1449 sbi->root_ino_num = le32_to_cpu(raw_super->root_ino);
1450 sbi->node_ino_num = le32_to_cpu(raw_super->node_ino);
1451 sbi->meta_ino_num = le32_to_cpu(raw_super->meta_ino);
5ec4e49f 1452 sbi->cur_victim_sec = NULL_SECNO;
b1c57c1c 1453 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
aff063e2 1454
ab9fa662 1455 sbi->dir_level = DEF_DIR_LEVEL;
6beceb54 1456 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
d0239e1b 1457 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
caf0047e 1458 clear_sbi_flag(sbi, SBI_NEED_FSCK);
2658e50d
JK		 1459
1460 INIT_LIST_HEAD(&sbi->s_list);
1461 mutex_init(&sbi->umount_mutex);
7dfeaa32
JK		 1462 mutex_init(&sbi->wio_mutex[NODE]);
1463 mutex_init(&sbi->wio_mutex[DATA]);
b5a7aef1
JK		 1464
1465#ifdef CONFIG_F2FS_FS_ENCRYPTION
1466 memcpy(sbi->key_prefix, F2FS_KEY_DESC_PREFIX,
1467 F2FS_KEY_DESC_PREFIX_SIZE);
1468 sbi->key_prefix_size = F2FS_KEY_DESC_PREFIX_SIZE;
1469#endif
aff063e2
JK		 1470}
1471
523be8a6
JK		 1472static int init_percpu_info(struct f2fs_sb_info *sbi)
1473{
1474 int i, err;
1475
ec795418
JK		 1476 if (percpu_init_rwsem(&sbi->cp_rwsem))
1477 return -ENOMEM;
1478
523be8a6
JK		 1479 for (i = 0; i < NR_COUNT_TYPE; i++) {
1480 err = percpu_counter_init(&sbi->nr_pages[i], 0, GFP_KERNEL);
1481 if (err)
1482 return err;
1483 }
41382ec4 1484
513c5f37
JK		 1485 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
1486 if (err)
1487 return err;
1488
1489 return percpu_counter_init(&sbi->total_valid_inode_count, 0,
41382ec4 1490 GFP_KERNEL);
523be8a6
JK		 1491}
1492
9076a75f
GZ		 1493/*
1494 * Read f2fs raw super block.
2b39e907
SL		 1495 * Because we have two copies of super block, so read both of them
1496 * to get the first valid one. If any one of them is broken, we pass
1497 * them recovery flag back to the caller.
9076a75f 1498 */
df728b0f 1499static int read_raw_super_block(struct f2fs_sb_info *sbi,
9076a75f 1500 struct f2fs_super_block **raw_super,
e8240f65 1501 int *valid_super_block, int *recovery)
14d7e9de 1502{
df728b0f 1503 struct super_block *sb = sbi->sb;
2b39e907 1504 int block;
e8240f65 1505 struct buffer_head *bh;
fd694733 1506 struct f2fs_super_block *super;
da554e48 1507 int err = 0;
14d7e9de 1508
b39f0de2
YH		 1509 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
1510 if (!super)
1511 return -ENOMEM;
2b39e907
SL		 1512
1513 for (block = 0; block < 2; block++) {
1514 bh = sb_bread(sb, block);
1515 if (!bh) {
1516 f2fs_msg(sb, KERN_ERR, "Unable to read %dth superblock",
9076a75f 1517 block + 1);
2b39e907
SL		 1518 err = -EIO;
1519 continue;
1520 }
14d7e9de 1521
2b39e907 1522 /* sanity checking of raw super */
df728b0f 1523 if (sanity_check_raw_super(sbi, bh)) {
2b39e907
SL		 1524 f2fs_msg(sb, KERN_ERR,
1525 "Can't find valid F2FS filesystem in %dth superblock",
1526 block + 1);
1527 err = -EINVAL;
1528 brelse(bh);
1529 continue;
1530 }
14d7e9de 1531
2b39e907 1532 if (!*raw_super) {
fd694733
JK		 1533 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
1534 sizeof(*super));
2b39e907
SL		 1535 *valid_super_block = block;
1536 *raw_super = super;
1537 }
1538 brelse(bh);
da554e48 1539 }
1540
2b39e907
SL		 1541 /* Fail to read any one of the superblocks*/
1542 if (err < 0)
1543 *recovery = 1;
da554e48 1544
da554e48 1545 /* No valid superblock */
2b39e907 1546 if (!*raw_super)
b39f0de2 1547 kfree(super);
2b39e907
SL		 1548 else
1549 err = 0;
da554e48 1550
2b39e907 1551 return err;
14d7e9de 1552}
1553
fd694733 1554int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
26d815ad 1555{
5d909cdb 1556 struct buffer_head *bh;
26d815ad
JK		 1557 int err;
1558
df728b0f
JK		 1559 if ((recover && f2fs_readonly(sbi->sb)) ||
1560 bdev_read_only(sbi->sb->s_bdev)) {
1561 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
f2353d7b 1562 return -EROFS;
df728b0f 1563 }
f2353d7b 1564
fd694733
JK		 1565 /* write back-up superblock first */
1566 bh = sb_getblk(sbi->sb, sbi->valid_super_block ? 0: 1);
5d909cdb
JK		 1567 if (!bh)
1568 return -EIO;
fd694733 1569 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
5d909cdb 1570 brelse(bh);
c5bda1c8
CY		 1571
1572 /* if we are in recovery path, skip writing valid superblock */
1573 if (recover || err)
5d909cdb 1574 return err;
26d815ad
JK		 1575
1576 /* write current valid superblock */
fd694733
JK		 1577 bh = sb_getblk(sbi->sb, sbi->valid_super_block);
1578 if (!bh)
1579 return -EIO;
1580 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
1581 brelse(bh);
1582 return err;
26d815ad
JK		 1583}
1584
aff063e2
JK		 1585static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
1586{
1587 struct f2fs_sb_info *sbi;
da554e48 1588 struct f2fs_super_block *raw_super;
aff063e2 1589 struct inode *root;
99e3e858 1590 int err;
2adc3505 1591 bool retry = true, need_fsck = false;
dabc4a5c 1592 char *options = NULL;
e8240f65 1593 int recovery, i, valid_super_block;
8f1dbbbb 1594 struct curseg_info *seg_i;
aff063e2 1595
ed2e621a 1596try_onemore:
da554e48 1597 err = -EINVAL;
1598 raw_super = NULL;
e8240f65 1599 valid_super_block = -1;
da554e48 1600 recovery = 0;
1601
aff063e2
JK		 1602 /* allocate memory for f2fs-specific super block info */
1603 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
1604 if (!sbi)
1605 return -ENOMEM;
1606
df728b0f
JK		 1607 sbi->sb = sb;
1608
43b6573b
KM		 1609 /* Load the checksum driver */
1610 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
1611 if (IS_ERR(sbi->s_chksum_driver)) {
1612 f2fs_msg(sb, KERN_ERR, "Cannot load crc32 driver.");
1613 err = PTR_ERR(sbi->s_chksum_driver);
1614 sbi->s_chksum_driver = NULL;
1615 goto free_sbi;
1616 }
1617
ff9234ad 1618 /* set a block size */
6bacf52f 1619 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
a07ef784 1620 f2fs_msg(sb, KERN_ERR, "unable to set blocksize");
aff063e2 1621 goto free_sbi;
a07ef784 1622 }
aff063e2 1623
df728b0f 1624 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
e8240f65 1625 &recovery);
9076a75f
GZ		 1626 if (err)
1627 goto free_sbi;
1628
5fb08372 1629 sb->s_fs_info = sbi;
52763a4b
JK		 1630 sbi->raw_super = raw_super;
1631
498c5e9f 1632 default_options(sbi);
aff063e2 1633 /* parse mount options */
dabc4a5c
JK		 1634 options = kstrdup((const char *)data, GFP_KERNEL);
1635 if (data && !options) {
1636 err = -ENOMEM;
aff063e2 1637 goto free_sb_buf;
dabc4a5c
JK		 1638 }
1639
1640 err = parse_options(sb, options);
1641 if (err)
1642 goto free_options;
aff063e2 1643
e0afc4d6
CY		 1644 sbi->max_file_blocks = max_file_blocks();
1645 sb->s_maxbytes = sbi->max_file_blocks <<
1646 le32_to_cpu(raw_super->log_blocksize);
aff063e2
JK		 1647 sb->s_max_links = F2FS_LINK_MAX;
1648 get_random_bytes(&sbi->s_next_generation, sizeof(u32));
1649
1650 sb->s_op = &f2fs_sops;
0b81d077 1651 sb->s_cop = &f2fs_cryptops;
aff063e2
JK		 1652 sb->s_xattr = f2fs_xattr_handlers;
1653 sb->s_export_op = &f2fs_export_ops;
1654 sb->s_magic = F2FS_SUPER_MAGIC;
aff063e2
JK		 1655 sb->s_time_gran = 1;
1656 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
1657 (test_opt(sbi, POSIX_ACL) ? MS_POSIXACL : 0);
1658 memcpy(sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
1659
1660 /* init f2fs-specific super block info */
e8240f65 1661 sbi->valid_super_block = valid_super_block;
aff063e2 1662 mutex_init(&sbi->gc_mutex);
aff063e2 1663 mutex_init(&sbi->cp_mutex);
b3582c68 1664 init_rwsem(&sbi->node_write);
315df839
JK		 1665
1666 /* disallow all the data/node/meta page writes */
1667 set_sbi_flag(sbi, SBI_POR_DOING);
aff063e2 1668 spin_lock_init(&sbi->stat_lock);
971767ca 1669
df0f8dc0 1670 init_rwsem(&sbi->read_io.io_rwsem);
458e6197
JK		 1671 sbi->read_io.sbi = sbi;
1672 sbi->read_io.bio = NULL;
1673 for (i = 0; i < NR_PAGE_TYPE; i++) {
df0f8dc0 1674 init_rwsem(&sbi->write_io[i].io_rwsem);
458e6197
JK		 1675 sbi->write_io[i].sbi = sbi;
1676 sbi->write_io[i].bio = NULL;
1677 }
971767ca 1678
fb51b5ef 1679 init_waitqueue_head(&sbi->cp_wait);
aff063e2
JK		 1680 init_sb_info(sbi);
1681
523be8a6
JK		 1682 err = init_percpu_info(sbi);
1683 if (err)
1684 goto free_options;
1685
aff063e2
JK		 1686 /* get an inode for meta space */
1687 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
1688 if (IS_ERR(sbi->meta_inode)) {
a07ef784 1689 f2fs_msg(sb, KERN_ERR, "Failed to read F2FS meta data inode");
aff063e2 1690 err = PTR_ERR(sbi->meta_inode);
dabc4a5c 1691 goto free_options;
aff063e2
JK		 1692 }
1693
1694 err = get_valid_checkpoint(sbi);
a07ef784
NJ		 1695 if (err) {
1696 f2fs_msg(sb, KERN_ERR, "Failed to get valid F2FS checkpoint");
aff063e2 1697 goto free_meta_inode;
a07ef784 1698 }
aff063e2 1699
aff063e2
JK		 1700 sbi->total_valid_node_count =
1701 le32_to_cpu(sbi->ckpt->valid_node_count);
513c5f37
JK		 1702 percpu_counter_set(&sbi->total_valid_inode_count,
1703 le32_to_cpu(sbi->ckpt->valid_inode_count));
aff063e2
JK		 1704 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
1705 sbi->total_valid_block_count =
1706 le64_to_cpu(sbi->ckpt->valid_block_count);
1707 sbi->last_valid_block_count = sbi->total_valid_block_count;
41382ec4 1708
c227f912
CY		 1709 for (i = 0; i < NR_INODE_TYPE; i++) {
1710 INIT_LIST_HEAD(&sbi->inode_list[i]);
1711 spin_lock_init(&sbi->inode_lock[i]);
1712 }
aff063e2 1713
1dcc336b
CY		 1714 init_extent_cache_info(sbi);
1715
6451e041 1716 init_ino_entry_info(sbi);
aff063e2
JK		 1717
1718 /* setup f2fs internal modules */
1719 err = build_segment_manager(sbi);
a07ef784
NJ		 1720 if (err) {
1721 f2fs_msg(sb, KERN_ERR,
1722 "Failed to initialize F2FS segment manager");
aff063e2 1723 goto free_sm;
a07ef784 1724 }
aff063e2 1725 err = build_node_manager(sbi);
a07ef784
NJ		 1726 if (err) {
1727 f2fs_msg(sb, KERN_ERR,
1728 "Failed to initialize F2FS node manager");
aff063e2 1729 goto free_nm;
a07ef784 1730 }
aff063e2 1731
8f1dbbbb
SL		 1732 /* For write statistics */
1733 if (sb->s_bdev->bd_part)
1734 sbi->sectors_written_start =
1735 (u64)part_stat_read(sb->s_bdev->bd_part, sectors[1]);
1736
1737 /* Read accumulated write IO statistics if exists */
1738 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
1739 if (__exist_node_summaries(sbi))
1740 sbi->kbytes_written =
b2dde6fc 1741 le64_to_cpu(seg_i->journal->info.kbytes_written);
8f1dbbbb 1742
aff063e2
JK		 1743 build_gc_manager(sbi);
1744
1745 /* get an inode for node space */
1746 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
1747 if (IS_ERR(sbi->node_inode)) {
a07ef784 1748 f2fs_msg(sb, KERN_ERR, "Failed to read node inode");
aff063e2
JK		 1749 err = PTR_ERR(sbi->node_inode);
1750 goto free_nm;
1751 }
1752
2658e50d
JK		 1753 f2fs_join_shrinker(sbi);
1754
aff063e2 1755 /* if there are nt orphan nodes free them */
8c14bfad
CY		 1756 err = recover_orphan_inodes(sbi);
1757 if (err)
1758 goto free_node_inode;
aff063e2
JK		 1759
1760 /* read root inode and dentry */
1761 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
1762 if (IS_ERR(root)) {
a07ef784 1763 f2fs_msg(sb, KERN_ERR, "Failed to read root inode");
aff063e2
JK		 1764 err = PTR_ERR(root);
1765 goto free_node_inode;
1766 }
8f99a946 1767 if (!S_ISDIR(root->i_mode) || !root->i_blocks || !root->i_size) {
9d847950 1768 iput(root);
8f99a946 1769 err = -EINVAL;
9d847950 1770 goto free_node_inode;
8f99a946 1771 }
aff063e2
JK		 1772
1773 sb->s_root = d_make_root(root); /* allocate root dentry */
1774 if (!sb->s_root) {
1775 err = -ENOMEM;
1776 goto free_root_inode;
1777 }
1778
aff063e2
JK		 1779 err = f2fs_build_stats(sbi);
1780 if (err)
6437d1b0 1781 goto free_root_inode;
aff063e2 1782
5e176d54
JK		 1783 if (f2fs_proc_root)
1784 sbi->s_proc = proc_mkdir(sb->s_id, f2fs_proc_root);
1785
f00d6fa7 1786 if (sbi->s_proc) {
5e176d54
JK		 1787 proc_create_data("segment_info", S_IRUGO, sbi->s_proc,
1788 &f2fs_seq_segment_info_fops, sb);
f00d6fa7
JK		 1789 proc_create_data("segment_bits", S_IRUGO, sbi->s_proc,
1790 &f2fs_seq_segment_bits_fops, sb);
1791 }
5e176d54 1792
b59d0bae
NJ		 1793 sbi->s_kobj.kset = f2fs_kset;
1794 init_completion(&sbi->s_kobj_unregister);
1795 err = kobject_init_and_add(&sbi->s_kobj, &f2fs_ktype, NULL,
1796 "%s", sb->s_id);
1797 if (err)
6437d1b0 1798 goto free_proc;
b59d0bae 1799
6437d1b0
JK		 1800 /* recover fsynced data */
1801 if (!test_opt(sbi, DISABLE_ROLL_FORWARD)) {
081d78c2
JK		 1802 /*
1803 * mount should be failed, when device has readonly mode, and
1804 * previous checkpoint was not done by clean system shutdown.
1805 */
1806 if (bdev_read_only(sb->s_bdev) &&
1807 !is_set_ckpt_flags(sbi->ckpt, CP_UMOUNT_FLAG)) {
1808 err = -EROFS;
1809 goto free_kobj;
1810 }
2adc3505
CY		 1811
1812 if (need_fsck)
1813 set_sbi_flag(sbi, SBI_NEED_FSCK);
1814
6781eabb
JK		 1815 err = recover_fsync_data(sbi, false);
1816 if (err < 0) {
2adc3505 1817 need_fsck = true;
6437d1b0 1818 f2fs_msg(sb, KERN_ERR,
99e3e858 1819 "Cannot recover all fsync data errno=%d", err);
ed2e621a
JK		 1820 goto free_kobj;
1821 }
6781eabb
JK		 1822 } else {
1823 err = recover_fsync_data(sbi, true);
1824
1825 if (!f2fs_readonly(sb) && err > 0) {
1826 err = -EINVAL;
1827 f2fs_msg(sb, KERN_ERR,
1828 "Need to recover fsync data");
1829 goto free_kobj;
1830 }
6437d1b0 1831 }
6781eabb 1832
315df839
JK		 1833 /* recover_fsync_data() cleared this already */
1834 clear_sbi_flag(sbi, SBI_POR_DOING);
b59d0bae 1835
6437d1b0
JK		 1836 /*
1837 * If filesystem is not mounted as read-only then
1838 * do start the gc_thread.
1839 */
6c029932 1840 if (test_opt(sbi, BG_GC) && !f2fs_readonly(sb)) {
6437d1b0
JK		 1841 /* After POR, we can run background GC thread.*/
1842 err = start_gc_thread(sbi);
1843 if (err)
1844 goto free_kobj;
1845 }
dabc4a5c 1846 kfree(options);
da554e48 1847
1848 /* recover broken superblock */
f2353d7b 1849 if (recovery) {
41214b3c
CY		 1850 err = f2fs_commit_super(sbi, true);
1851 f2fs_msg(sb, KERN_INFO,
99e3e858 1852 "Try to recover %dth superblock, ret: %d",
41214b3c 1853 sbi->valid_super_block ? 1 : 2, err);
da554e48 1854 }
1855
6beceb54 1856 f2fs_update_time(sbi, CP_TIME);
d0239e1b 1857 f2fs_update_time(sbi, REQ_TIME);
aff063e2 1858 return 0;
6437d1b0
JK		 1859
1860free_kobj:
0f18b462 1861 f2fs_sync_inode_meta(sbi);
6437d1b0 1862 kobject_del(&sbi->s_kobj);
29ba108d
CY		 1863 kobject_put(&sbi->s_kobj);
1864 wait_for_completion(&sbi->s_kobj_unregister);
6437d1b0 1865free_proc:
1d15bd20
CY		 1866 if (sbi->s_proc) {
1867 remove_proc_entry("segment_info", sbi->s_proc);
f00d6fa7 1868 remove_proc_entry("segment_bits", sbi->s_proc);
1d15bd20
CY		 1869 remove_proc_entry(sb->s_id, f2fs_proc_root);
1870 }
1871 f2fs_destroy_stats(sbi);
aff063e2
JK		 1872free_root_inode:
1873 dput(sb->s_root);
1874 sb->s_root = NULL;
1875free_node_inode:
2658e50d
JK		 1876 mutex_lock(&sbi->umount_mutex);
1877 f2fs_leave_shrinker(sbi);
aff063e2 1878 iput(sbi->node_inode);
2658e50d 1879 mutex_unlock(&sbi->umount_mutex);
aff063e2
JK		 1880free_nm:
1881 destroy_node_manager(sbi);
1882free_sm:
1883 destroy_segment_manager(sbi);
aff063e2
JK		 1884 kfree(sbi->ckpt);
1885free_meta_inode:
1886 make_bad_inode(sbi->meta_inode);
1887 iput(sbi->meta_inode);
dabc4a5c 1888free_options:
523be8a6 1889 destroy_percpu_info(sbi);
dabc4a5c 1890 kfree(options);
aff063e2 1891free_sb_buf:
b39f0de2 1892 kfree(raw_super);
aff063e2 1893free_sbi:
43b6573b
KM		 1894 if (sbi->s_chksum_driver)
1895 crypto_free_shash(sbi->s_chksum_driver);
aff063e2 1896 kfree(sbi);
ed2e621a
JK		 1897
1898 /* give only one another chance */
1899 if (retry) {
9df47ba7 1900 retry = false;
ed2e621a
JK		 1901 shrink_dcache_sb(sb);
1902 goto try_onemore;
1903 }
aff063e2
JK		 1904 return err;
1905}
1906
1907static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
1908 const char *dev_name, void *data)
1909{
1910 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
1911}
1912
30a5537f
JK		 1913static void kill_f2fs_super(struct super_block *sb)
1914{
1915 if (sb->s_root)
caf0047e 1916 set_sbi_flag(F2FS_SB(sb), SBI_IS_CLOSE);
30a5537f
JK		 1917 kill_block_super(sb);
1918}
1919
aff063e2
JK		 1920static struct file_system_type f2fs_fs_type = {
1921 .owner = THIS_MODULE,
1922 .name = "f2fs",
1923 .mount = f2fs_mount,
30a5537f 1924 .kill_sb = kill_f2fs_super,
aff063e2
JK		 1925 .fs_flags = FS_REQUIRES_DEV,
1926};
7f78e035 1927MODULE_ALIAS_FS("f2fs");
aff063e2 1928
6e6093a8 1929static int __init init_inodecache(void)
aff063e2 1930{
5d097056
VD		 1931 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
1932 sizeof(struct f2fs_inode_info), 0,
1933 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
6bacf52f 1934 if (!f2fs_inode_cachep)
aff063e2
JK		 1935 return -ENOMEM;
1936 return 0;
1937}
1938
1939static void destroy_inodecache(void)
1940{
1941 /*
1942 * Make sure all delayed rcu free inodes are flushed before we
1943 * destroy cache.
1944 */
1945 rcu_barrier();
1946 kmem_cache_destroy(f2fs_inode_cachep);
1947}
1948
1949static int __init init_f2fs_fs(void)
1950{
1951 int err;
1952
c0508650
JK		 1953 f2fs_build_trace_ios();
1954
aff063e2
JK		 1955 err = init_inodecache();
1956 if (err)
1957 goto fail;
1958 err = create_node_manager_caches();
1959 if (err)
9890ff3f 1960 goto free_inodecache;
7fd9e544 1961 err = create_segment_manager_caches();
aff063e2 1962 if (err)
9890ff3f 1963 goto free_node_manager_caches;
aff063e2
JK		 1964 err = create_checkpoint_caches();
1965 if (err)
06292073 1966 goto free_segment_manager_caches;
1dcc336b
CY		 1967 err = create_extent_cache();
1968 if (err)
1969 goto free_checkpoint_caches;
b59d0bae 1970 f2fs_kset = kset_create_and_add("f2fs", NULL, fs_kobj);
6e6b978c
WY		 1971 if (!f2fs_kset) {
1972 err = -ENOMEM;
1dcc336b 1973 goto free_extent_cache;
6e6b978c 1974 }
08796897
SY		 1975#ifdef CONFIG_F2FS_FAULT_INJECTION
1976 f2fs_fault_inject.kset = f2fs_kset;
1977 f2fs_build_fault_attr(0);
1978 err = kobject_init_and_add(&f2fs_fault_inject, &f2fs_fault_ktype,
1979 NULL, "fault_injection");
1980 if (err) {
1981 f2fs_fault_inject.kset = NULL;
1982 goto free_kset;
1983 }
1984#endif
2658e50d 1985 err = register_shrinker(&f2fs_shrinker_info);
cfc4d971 1986 if (err)
0b81d077 1987 goto free_kset;
2658e50d
JK		 1988
1989 err = register_filesystem(&f2fs_fs_type);
1990 if (err)
1991 goto free_shrinker;
787c7b8c
CY		 1992 err = f2fs_create_root_stats();
1993 if (err)
1994 goto free_filesystem;
5e176d54 1995 f2fs_proc_root = proc_mkdir("fs/f2fs", NULL);
9890ff3f
ZH		 1996 return 0;
1997
787c7b8c
CY		 1998free_filesystem:
1999 unregister_filesystem(&f2fs_fs_type);
2658e50d
JK		 2000free_shrinker:
2001 unregister_shrinker(&f2fs_shrinker_info);
9890ff3f 2002free_kset:
08796897
SY		 2003#ifdef CONFIG_F2FS_FAULT_INJECTION
2004 if (f2fs_fault_inject.kset)
2005 kobject_put(&f2fs_fault_inject);
2006#endif
9890ff3f 2007 kset_unregister(f2fs_kset);
1dcc336b
CY		 2008free_extent_cache:
2009 destroy_extent_cache();
9890ff3f
ZH		 2010free_checkpoint_caches:
2011 destroy_checkpoint_caches();
7fd9e544
JK		 2012free_segment_manager_caches:
2013 destroy_segment_manager_caches();
9890ff3f
ZH		 2014free_node_manager_caches:
2015 destroy_node_manager_caches();
2016free_inodecache:
2017 destroy_inodecache();
aff063e2
JK		 2018fail:
2019 return err;
2020}
2021
2022static void __exit exit_f2fs_fs(void)
2023{
5e176d54 2024 remove_proc_entry("fs/f2fs", NULL);
4589d25d 2025 f2fs_destroy_root_stats();
aff063e2 2026 unregister_filesystem(&f2fs_fs_type);
b8bef79d
TY		 2027 unregister_shrinker(&f2fs_shrinker_info);
2028#ifdef CONFIG_F2FS_FAULT_INJECTION
2029 kobject_put(&f2fs_fault_inject);
2030#endif
2031 kset_unregister(f2fs_kset);
fdf6c8be 2032 destroy_extent_cache();
aff063e2 2033 destroy_checkpoint_caches();
5dcd8a71 2034 destroy_segment_manager_caches();
aff063e2
JK		 2035 destroy_node_manager_caches();
2036 destroy_inodecache();
351f4fba 2037 f2fs_destroy_trace_ios();
aff063e2
JK		 2038}
2039
2040module_init(init_f2fs_fs)
2041module_exit(exit_f2fs_fs)
2042
2043MODULE_AUTHOR("Samsung Electronics's Praesto Team");
2044MODULE_DESCRIPTION("Flash Friendly File System");
2045MODULE_LICENSE("GPL");
Linux kernel - Deliverables
This page took 0.288065 seconds and 5 git commands to generate.