projects / linux-2.6-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
f2fs: do fault injection initialization in default_options
[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 42#ifdef CONFIG_F2FS_FAULT_INJECTION
2c63fead
JK		 43
44char *fault_name[FAULT_MAX] = {
45 [FAULT_KMALLOC] = "kmalloc",
c41f3cc3 46 [FAULT_PAGE_ALLOC] = "page alloc",
cb78942b
JK		 47 [FAULT_ALLOC_NID] = "alloc nid",
48 [FAULT_ORPHAN] = "orphan",
49 [FAULT_BLOCK] = "no more block",
50 [FAULT_DIR_DEPTH] = "too big dir depth",
53aa6bbf 51 [FAULT_EVICT_INODE] = "evict_inode fail",
8b038c70 52 [FAULT_IO] = "IO error",
2c63fead 53};
08796897 54
1ecc0c5c
CY		 55static void f2fs_build_fault_attr(struct f2fs_sb_info *sbi,
56 unsigned int rate)
08796897 57{
1ecc0c5c
CY		 58 struct f2fs_fault_info *ffi = &sbi->fault_info;
59
08796897 60 if (rate) {
1ecc0c5c
CY		 61 atomic_set(&ffi->inject_ops, 0);
62 ffi->inject_rate = rate;
63 ffi->inject_type = (1 << FAULT_MAX) - 1;
08796897 64 } else {
1ecc0c5c 65 memset(ffi, 0, sizeof(struct f2fs_fault_info));
08796897
SY		 66 }
67}
73faec4d
JK		 68#endif
69
2658e50d
JK		 70/* f2fs-wide shrinker description */
71static struct shrinker f2fs_shrinker_info = {
72 .scan_objects = f2fs_shrink_scan,
73 .count_objects = f2fs_shrink_count,
74 .seeks = DEFAULT_SEEKS,
75};
76
aff063e2 77enum {
696c018c 78 Opt_gc_background,
aff063e2 79 Opt_disable_roll_forward,
2d834bf9 80 Opt_norecovery,
aff063e2 81 Opt_discard,
64058be9 82 Opt_nodiscard,
aff063e2 83 Opt_noheap,
4058c511 84 Opt_user_xattr,
aff063e2 85 Opt_nouser_xattr,
4058c511 86 Opt_acl,
aff063e2
JK		 87 Opt_noacl,
88 Opt_active_logs,
89 Opt_disable_ext_identify,
444c580f 90 Opt_inline_xattr,
8274de77 91 Opt_inline_data,
5efd3c6f 92 Opt_inline_dentry,
97c1794a 93 Opt_noinline_dentry,
6b4afdd7 94 Opt_flush_merge,
69e9e427 95 Opt_noflush_merge,
0f7b2abd 96 Opt_nobarrier,
d5053a34 97 Opt_fastboot,
89672159 98 Opt_extent_cache,
7daaea25 99 Opt_noextent_cache,
75342797 100 Opt_noinline_data,
343f40f0 101 Opt_data_flush,
36abef4e 102 Opt_mode,
73faec4d 103 Opt_fault_injection,
6d94c74a
JK		 104 Opt_lazytime,
105 Opt_nolazytime,
aff063e2
JK		 106 Opt_err,
107};
108
109static match_table_t f2fs_tokens = {
696c018c 110 {Opt_gc_background, "background_gc=%s"},
aff063e2 111 {Opt_disable_roll_forward, "disable_roll_forward"},
2d834bf9 112 {Opt_norecovery, "norecovery"},
aff063e2 113 {Opt_discard, "discard"},
64058be9 114 {Opt_nodiscard, "nodiscard"},
aff063e2 115 {Opt_noheap, "no_heap"},
4058c511 116 {Opt_user_xattr, "user_xattr"},
aff063e2 117 {Opt_nouser_xattr, "nouser_xattr"},
4058c511 118 {Opt_acl, "acl"},
aff063e2
JK		 119 {Opt_noacl, "noacl"},
120 {Opt_active_logs, "active_logs=%u"},
121 {Opt_disable_ext_identify, "disable_ext_identify"},
444c580f 122 {Opt_inline_xattr, "inline_xattr"},
8274de77 123 {Opt_inline_data, "inline_data"},
5efd3c6f 124 {Opt_inline_dentry, "inline_dentry"},
97c1794a 125 {Opt_noinline_dentry, "noinline_dentry"},
6b4afdd7 126 {Opt_flush_merge, "flush_merge"},
69e9e427 127 {Opt_noflush_merge, "noflush_merge"},
0f7b2abd 128 {Opt_nobarrier, "nobarrier"},
d5053a34 129 {Opt_fastboot, "fastboot"},
89672159 130 {Opt_extent_cache, "extent_cache"},
7daaea25 131 {Opt_noextent_cache, "noextent_cache"},
75342797 132 {Opt_noinline_data, "noinline_data"},
343f40f0 133 {Opt_data_flush, "data_flush"},
36abef4e 134 {Opt_mode, "mode=%s"},
73faec4d 135 {Opt_fault_injection, "fault_injection=%u"},
6d94c74a
JK		 136 {Opt_lazytime, "lazytime"},
137 {Opt_nolazytime, "nolazytime"},
aff063e2
JK		 138 {Opt_err, NULL},
139};
140
b59d0bae 141/* Sysfs support for f2fs */
ea91e9b0
JK		 142enum {
143 GC_THREAD, /* struct f2fs_gc_thread */
144 SM_INFO, /* struct f2fs_sm_info */
cdfc41c1 145 NM_INFO, /* struct f2fs_nm_info */
b1c57c1c 146 F2FS_SBI, /* struct f2fs_sb_info */
08796897
SY		 147#ifdef CONFIG_F2FS_FAULT_INJECTION
148 FAULT_INFO_RATE, /* struct f2fs_fault_info */
149 FAULT_INFO_TYPE, /* struct f2fs_fault_info */
150#endif
ea91e9b0
JK		 151};
152
b59d0bae
NJ		 153struct f2fs_attr {
154 struct attribute attr;
155 ssize_t (*show)(struct f2fs_attr *, struct f2fs_sb_info *, char *);
156 ssize_t (*store)(struct f2fs_attr *, struct f2fs_sb_info *,
157 const char *, size_t);
ea91e9b0 158 int struct_type;
b59d0bae
NJ		 159 int offset;
160};
161
ea91e9b0
JK		 162static unsigned char *__struct_ptr(struct f2fs_sb_info *sbi, int struct_type)
163{
164 if (struct_type == GC_THREAD)
165 return (unsigned char *)sbi->gc_thread;
166 else if (struct_type == SM_INFO)
167 return (unsigned char *)SM_I(sbi);
cdfc41c1
JK		 168 else if (struct_type == NM_INFO)
169 return (unsigned char *)NM_I(sbi);
b1c57c1c
JK		 170 else if (struct_type == F2FS_SBI)
171 return (unsigned char *)sbi;
08796897
SY		 172#ifdef CONFIG_F2FS_FAULT_INJECTION
173 else if (struct_type == FAULT_INFO_RATE ||
174 struct_type == FAULT_INFO_TYPE)
1ecc0c5c 175 return (unsigned char *)&sbi->fault_info;
08796897 176#endif
ea91e9b0
JK		 177 return NULL;
178}
179
8f1dbbbb
SL		 180static ssize_t lifetime_write_kbytes_show(struct f2fs_attr *a,
181 struct f2fs_sb_info *sbi, char *buf)
182{
183 struct super_block *sb = sbi->sb;
184
185 if (!sb->s_bdev->bd_part)
186 return snprintf(buf, PAGE_SIZE, "0\n");
187
188 return snprintf(buf, PAGE_SIZE, "%llu\n",
189 (unsigned long long)(sbi->kbytes_written +
190 BD_PART_WRITTEN(sbi)));
191}
192
b59d0bae
NJ		 193static ssize_t f2fs_sbi_show(struct f2fs_attr *a,
194 struct f2fs_sb_info *sbi, char *buf)
195{
ea91e9b0 196 unsigned char *ptr = NULL;
b59d0bae
NJ		 197 unsigned int *ui;
198
ea91e9b0
JK		 199 ptr = __struct_ptr(sbi, a->struct_type);
200 if (!ptr)
b59d0bae
NJ		 201 return -EINVAL;
202
ea91e9b0 203 ui = (unsigned int *)(ptr + a->offset);
b59d0bae
NJ		 204
205 return snprintf(buf, PAGE_SIZE, "%u\n", *ui);
206}
207
208static ssize_t f2fs_sbi_store(struct f2fs_attr *a,
209 struct f2fs_sb_info *sbi,
210 const char *buf, size_t count)
211{
ea91e9b0 212 unsigned char *ptr;
b59d0bae
NJ		 213 unsigned long t;
214 unsigned int *ui;
215 ssize_t ret;
216
ea91e9b0
JK		 217 ptr = __struct_ptr(sbi, a->struct_type);
218 if (!ptr)
b59d0bae
NJ		 219 return -EINVAL;
220
ea91e9b0 221 ui = (unsigned int *)(ptr + a->offset);
b59d0bae
NJ		 222
223 ret = kstrtoul(skip_spaces(buf), 0, &t);
224 if (ret < 0)
225 return ret;
08796897
SY		 226#ifdef CONFIG_F2FS_FAULT_INJECTION
227 if (a->struct_type == FAULT_INFO_TYPE && t >= (1 << FAULT_MAX))
228 return -EINVAL;
229#endif
b59d0bae
NJ		 230 *ui = t;
231 return count;
232}
233
234static ssize_t f2fs_attr_show(struct kobject *kobj,
235 struct attribute *attr, char *buf)
236{
237 struct f2fs_sb_info *sbi = container_of(kobj, struct f2fs_sb_info,
238 s_kobj);
239 struct f2fs_attr *a = container_of(attr, struct f2fs_attr, attr);
240
241 return a->show ? a->show(a, sbi, buf) : 0;
242}
243
244static ssize_t f2fs_attr_store(struct kobject *kobj, struct attribute *attr,
245 const char *buf, size_t len)
246{
247 struct f2fs_sb_info *sbi = container_of(kobj, struct f2fs_sb_info,
248 s_kobj);
249 struct f2fs_attr *a = container_of(attr, struct f2fs_attr, attr);
250
251 return a->store ? a->store(a, sbi, buf, len) : 0;
252}
253
254static void f2fs_sb_release(struct kobject *kobj)
255{
256 struct f2fs_sb_info *sbi = container_of(kobj, struct f2fs_sb_info,
257 s_kobj);
258 complete(&sbi->s_kobj_unregister);
259}
260
ea91e9b0 261#define F2FS_ATTR_OFFSET(_struct_type, _name, _mode, _show, _store, _offset) \
b59d0bae
NJ		 262static struct f2fs_attr f2fs_attr_##_name = { \
263 .attr = {.name = __stringify(_name), .mode = _mode }, \
264 .show = _show, \
265 .store = _store, \
ea91e9b0
JK		 266 .struct_type = _struct_type, \
267 .offset = _offset \
b59d0bae
NJ		 268}
269
ea91e9b0
JK		 270#define F2FS_RW_ATTR(struct_type, struct_name, name, elname) \
271 F2FS_ATTR_OFFSET(struct_type, name, 0644, \
272 f2fs_sbi_show, f2fs_sbi_store, \
273 offsetof(struct struct_name, elname))
b59d0bae 274
8f1dbbbb
SL		 275#define F2FS_GENERAL_RO_ATTR(name) \
276static struct f2fs_attr f2fs_attr_##name = __ATTR(name, 0444, name##_show, NULL)
277
ea91e9b0
JK		 278F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_min_sleep_time, min_sleep_time);
279F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_max_sleep_time, max_sleep_time);
280F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_no_gc_sleep_time, no_gc_sleep_time);
281F2FS_RW_ATTR(GC_THREAD, f2fs_gc_kthread, gc_idle, gc_idle);
282F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, reclaim_segments, rec_prefree_segments);
7ac8c3b0 283F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, max_small_discards, max_discards);
bba681cb 284F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, batched_trim_sections, trim_sections);
216fbd64
JK		 285F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, ipu_policy, ipu_policy);
286F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, min_ipu_util, min_ipu_util);
c1ce1b02 287F2FS_RW_ATTR(SM_INFO, f2fs_sm_info, min_fsync_blocks, min_fsync_blocks);
cdfc41c1 288F2FS_RW_ATTR(NM_INFO, f2fs_nm_info, ram_thresh, ram_thresh);
ea1a29a0 289F2FS_RW_ATTR(NM_INFO, f2fs_nm_info, ra_nid_pages, ra_nid_pages);
2304cb0c 290F2FS_RW_ATTR(NM_INFO, f2fs_nm_info, dirty_nats_ratio, dirty_nats_ratio);
b1c57c1c 291F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, max_victim_search, max_victim_search);
ab9fa662 292F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, dir_level, dir_level);
6beceb54 293F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, cp_interval, interval_time[CP_TIME]);
d0239e1b 294F2FS_RW_ATTR(F2FS_SBI, f2fs_sb_info, idle_interval, interval_time[REQ_TIME]);
08796897
SY		 295#ifdef CONFIG_F2FS_FAULT_INJECTION
296F2FS_RW_ATTR(FAULT_INFO_RATE, f2fs_fault_info, inject_rate, inject_rate);
297F2FS_RW_ATTR(FAULT_INFO_TYPE, f2fs_fault_info, inject_type, inject_type);
298#endif
8f1dbbbb 299F2FS_GENERAL_RO_ATTR(lifetime_write_kbytes);
b59d0bae
NJ		 300
301#define ATTR_LIST(name) (&f2fs_attr_##name.attr)
302static struct attribute *f2fs_attrs[] = {
303 ATTR_LIST(gc_min_sleep_time),
304 ATTR_LIST(gc_max_sleep_time),
305 ATTR_LIST(gc_no_gc_sleep_time),
d2dc095f 306 ATTR_LIST(gc_idle),
ea91e9b0 307 ATTR_LIST(reclaim_segments),
7ac8c3b0 308 ATTR_LIST(max_small_discards),
bba681cb 309 ATTR_LIST(batched_trim_sections),
216fbd64
JK		 310 ATTR_LIST(ipu_policy),
311 ATTR_LIST(min_ipu_util),
c1ce1b02 312 ATTR_LIST(min_fsync_blocks),
b1c57c1c 313 ATTR_LIST(max_victim_search),
ab9fa662 314 ATTR_LIST(dir_level),
cdfc41c1 315 ATTR_LIST(ram_thresh),
ea1a29a0 316 ATTR_LIST(ra_nid_pages),
2304cb0c 317 ATTR_LIST(dirty_nats_ratio),
60b99b48 318 ATTR_LIST(cp_interval),
d0239e1b 319 ATTR_LIST(idle_interval),
1ecc0c5c
CY		 320#ifdef CONFIG_F2FS_FAULT_INJECTION
321 ATTR_LIST(inject_rate),
322 ATTR_LIST(inject_type),
323#endif
8f1dbbbb 324 ATTR_LIST(lifetime_write_kbytes),
b59d0bae
NJ		 325 NULL,
326};
327
328static const struct sysfs_ops f2fs_attr_ops = {
329 .show = f2fs_attr_show,
330 .store = f2fs_attr_store,
331};
332
333static struct kobj_type f2fs_ktype = {
334 .default_attrs = f2fs_attrs,
335 .sysfs_ops = &f2fs_attr_ops,
336 .release = f2fs_sb_release,
337};
338
a07ef784
NJ		 339void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...)
340{
341 struct va_format vaf;
342 va_list args;
343
344 va_start(args, fmt);
345 vaf.fmt = fmt;
346 vaf.va = &args;
347 printk("%sF2FS-fs (%s): %pV\n", level, sb->s_id, &vaf);
348 va_end(args);
349}
350
aff063e2
JK		 351static void init_once(void *foo)
352{
353 struct f2fs_inode_info *fi = (struct f2fs_inode_info *) foo;
354
aff063e2
JK		 355 inode_init_once(&fi->vfs_inode);
356}
357
696c018c
NJ		 358static int parse_options(struct super_block *sb, char *options)
359{
360 struct f2fs_sb_info *sbi = F2FS_SB(sb);
09d54cdd 361 struct request_queue *q;
696c018c
NJ		 362 substring_t args[MAX_OPT_ARGS];
363 char *p, *name;
364 int arg = 0;
365
366 if (!options)
367 return 0;
368
369 while ((p = strsep(&options, ",")) != NULL) {
370 int token;
371 if (!*p)
372 continue;
373 /*
374 * Initialize args struct so we know whether arg was
375 * found; some options take optional arguments.
376 */
377 args[0].to = args[0].from = NULL;
378 token = match_token(p, f2fs_tokens, args);
379
380 switch (token) {
381 case Opt_gc_background:
382 name = match_strdup(&args[0]);
383
384 if (!name)
385 return -ENOMEM;
6aefd93b 386 if (strlen(name) == 2 && !strncmp(name, "on", 2)) {
696c018c 387 set_opt(sbi, BG_GC);
6aefd93b
JK		 388 clear_opt(sbi, FORCE_FG_GC);
389 } else if (strlen(name) == 3 && !strncmp(name, "off", 3)) {
696c018c 390 clear_opt(sbi, BG_GC);
6aefd93b
JK		 391 clear_opt(sbi, FORCE_FG_GC);
392 } else if (strlen(name) == 4 && !strncmp(name, "sync", 4)) {
393 set_opt(sbi, BG_GC);
394 set_opt(sbi, FORCE_FG_GC);
395 } else {
696c018c
NJ		 396 kfree(name);
397 return -EINVAL;
398 }
399 kfree(name);
400 break;
401 case Opt_disable_roll_forward:
402 set_opt(sbi, DISABLE_ROLL_FORWARD);
403 break;
2d834bf9
JK		 404 case Opt_norecovery:
405 /* this option mounts f2fs with ro */
406 set_opt(sbi, DISABLE_ROLL_FORWARD);
407 if (!f2fs_readonly(sb))
408 return -EINVAL;
409 break;
696c018c 410 case Opt_discard:
09d54cdd
CY		 411 q = bdev_get_queue(sb->s_bdev);
412 if (blk_queue_discard(q)) {
413 set_opt(sbi, DISCARD);
414 } else {
415 f2fs_msg(sb, KERN_WARNING,
416 "mounting with \"discard\" option, but "
417 "the device does not support discard");
418 }
696c018c 419 break;
64058be9
CY		 420 case Opt_nodiscard:
421 clear_opt(sbi, DISCARD);
696c018c
NJ		 422 case Opt_noheap:
423 set_opt(sbi, NOHEAP);
424 break;
425#ifdef CONFIG_F2FS_FS_XATTR
4058c511
KA		 426 case Opt_user_xattr:
427 set_opt(sbi, XATTR_USER);
428 break;
696c018c
NJ		 429 case Opt_nouser_xattr:
430 clear_opt(sbi, XATTR_USER);
431 break;
444c580f
JK		 432 case Opt_inline_xattr:
433 set_opt(sbi, INLINE_XATTR);
434 break;
696c018c 435#else
4058c511
KA		 436 case Opt_user_xattr:
437 f2fs_msg(sb, KERN_INFO,
438 "user_xattr options not supported");
439 break;
696c018c
NJ		 440 case Opt_nouser_xattr:
441 f2fs_msg(sb, KERN_INFO,
442 "nouser_xattr options not supported");
443 break;
444c580f
JK		 444 case Opt_inline_xattr:
445 f2fs_msg(sb, KERN_INFO,
446 "inline_xattr options not supported");
447 break;
696c018c
NJ		 448#endif
449#ifdef CONFIG_F2FS_FS_POSIX_ACL
4058c511
KA		 450 case Opt_acl:
451 set_opt(sbi, POSIX_ACL);
452 break;
696c018c
NJ		 453 case Opt_noacl:
454 clear_opt(sbi, POSIX_ACL);
455 break;
456#else
4058c511
KA		 457 case Opt_acl:
458 f2fs_msg(sb, KERN_INFO, "acl options not supported");
459 break;
696c018c
NJ		 460 case Opt_noacl:
461 f2fs_msg(sb, KERN_INFO, "noacl options not supported");
462 break;
463#endif
464 case Opt_active_logs:
465 if (args->from && match_int(args, &arg))
466 return -EINVAL;
467 if (arg != 2 && arg != 4 && arg != NR_CURSEG_TYPE)
468 return -EINVAL;
469 sbi->active_logs = arg;
470 break;
471 case Opt_disable_ext_identify:
472 set_opt(sbi, DISABLE_EXT_IDENTIFY);
473 break;
8274de77
HL		 474 case Opt_inline_data:
475 set_opt(sbi, INLINE_DATA);
476 break;
5efd3c6f
CY		 477 case Opt_inline_dentry:
478 set_opt(sbi, INLINE_DENTRY);
479 break;
97c1794a
CY		 480 case Opt_noinline_dentry:
481 clear_opt(sbi, INLINE_DENTRY);
482 break;
6b4afdd7
JK		 483 case Opt_flush_merge:
484 set_opt(sbi, FLUSH_MERGE);
485 break;
69e9e427
JK		 486 case Opt_noflush_merge:
487 clear_opt(sbi, FLUSH_MERGE);
488 break;
0f7b2abd
JK		 489 case Opt_nobarrier:
490 set_opt(sbi, NOBARRIER);
491 break;
d5053a34
JK		 492 case Opt_fastboot:
493 set_opt(sbi, FASTBOOT);
494 break;
89672159
CY		 495 case Opt_extent_cache:
496 set_opt(sbi, EXTENT_CACHE);
497 break;
7daaea25
JK		 498 case Opt_noextent_cache:
499 clear_opt(sbi, EXTENT_CACHE);
500 break;
75342797
WL		 501 case Opt_noinline_data:
502 clear_opt(sbi, INLINE_DATA);
503 break;
343f40f0
CY		 504 case Opt_data_flush:
505 set_opt(sbi, DATA_FLUSH);
506 break;
36abef4e
JK		 507 case Opt_mode:
508 name = match_strdup(&args[0]);
509
510 if (!name)
511 return -ENOMEM;
512 if (strlen(name) == 8 &&
513 !strncmp(name, "adaptive", 8)) {
52763a4b 514 set_opt_mode(sbi, F2FS_MOUNT_ADAPTIVE);
36abef4e
JK		 515 } else if (strlen(name) == 3 &&
516 !strncmp(name, "lfs", 3)) {
52763a4b 517 set_opt_mode(sbi, F2FS_MOUNT_LFS);
36abef4e
JK		 518 } else {
519 kfree(name);
520 return -EINVAL;
521 }
522 kfree(name);
523 break;
73faec4d
JK		 524 case Opt_fault_injection:
525 if (args->from && match_int(args, &arg))
526 return -EINVAL;
527#ifdef CONFIG_F2FS_FAULT_INJECTION
1ecc0c5c 528 f2fs_build_fault_attr(sbi, arg);
73faec4d
JK		 529#else
530 f2fs_msg(sb, KERN_INFO,
531 "FAULT_INJECTION was not selected");
532#endif
533 break;
6d94c74a
JK		 534 case Opt_lazytime:
535 sb->s_flags |= MS_LAZYTIME;
536 break;
537 case Opt_nolazytime:
538 sb->s_flags &= ~MS_LAZYTIME;
539 break;
696c018c
NJ		 540 default:
541 f2fs_msg(sb, KERN_ERR,
542 "Unrecognized mount option \"%s\" or missing value",
543 p);
544 return -EINVAL;
545 }
546 }
547 return 0;
548}
549
aff063e2
JK		 550static struct inode *f2fs_alloc_inode(struct super_block *sb)
551{
552 struct f2fs_inode_info *fi;
553
a0acdfe0 554 fi = kmem_cache_alloc(f2fs_inode_cachep, GFP_F2FS_ZERO);
aff063e2
JK		 555 if (!fi)
556 return NULL;
557
558 init_once((void *) fi);
559
1beba1b3
JK		 560 if (percpu_counter_init(&fi->dirty_pages, 0, GFP_NOFS)) {
561 kmem_cache_free(f2fs_inode_cachep, fi);
562 return NULL;
563 }
564
434720fa 565 /* Initialize f2fs-specific inode info */
aff063e2 566 fi->vfs_inode.i_version = 1;
aff063e2
JK		 567 fi->i_current_depth = 1;
568 fi->i_advise = 0;
d928bfbf 569 init_rwsem(&fi->i_sem);
2710fd7e 570 INIT_LIST_HEAD(&fi->dirty_list);
0f18b462 571 INIT_LIST_HEAD(&fi->gdirty_list);
88b88a66
JK		 572 INIT_LIST_HEAD(&fi->inmem_pages);
573 mutex_init(&fi->inmem_lock);
82e0a5aa
CY		 574 init_rwsem(&fi->dio_rwsem[READ]);
575 init_rwsem(&fi->dio_rwsem[WRITE]);
aff063e2 576
ab9fa662
JK		 577 /* Will be used by directory only */
578 fi->i_dir_level = F2FS_SB(sb)->dir_level;
aff063e2
JK		 579 return &fi->vfs_inode;
580}
581
531ad7d5
JK		 582static int f2fs_drop_inode(struct inode *inode)
583{
584 /*
585 * This is to avoid a deadlock condition like below.
586 * writeback_single_inode(inode)
587 * - f2fs_write_data_page
588 * - f2fs_gc -> iput -> evict
589 * - inode_wait_for_writeback(inode)
590 */
0f18b462 591 if ((!inode_unhashed(inode) && inode->i_state & I_SYNC)) {
06e1bc05 592 if (!inode->i_nlink && !is_bad_inode(inode)) {
3e72f721
JK		 593 /* to avoid evict_inode call simultaneously */
594 atomic_inc(&inode->i_count);
06e1bc05
JK		 595 spin_unlock(&inode->i_lock);
596
597 /* some remained atomic pages should discarded */
598 if (f2fs_is_atomic_file(inode))
29b96b54 599 drop_inmem_pages(inode);
06e1bc05 600
3e72f721
JK		 601 /* should remain fi->extent_tree for writepage */
602 f2fs_destroy_extent_node(inode);
603
06e1bc05 604 sb_start_intwrite(inode->i_sb);
fc9581c8 605 f2fs_i_size_write(inode, 0);
06e1bc05
JK		 606
607 if (F2FS_HAS_BLOCKS(inode))
9a449e9c 608 f2fs_truncate(inode);
06e1bc05
JK		 609
610 sb_end_intwrite(inode->i_sb);
611
0b81d077 612 fscrypt_put_encryption_info(inode, NULL);
06e1bc05 613 spin_lock(&inode->i_lock);
3e72f721 614 atomic_dec(&inode->i_count);
06e1bc05 615 }
531ad7d5 616 return 0;
06e1bc05 617 }
0f18b462 618
67c3758d 619 return generic_drop_inode(inode);
531ad7d5
JK		 620}
621
b56ab837 622int f2fs_inode_dirtied(struct inode *inode)
b3783873 623{
0f18b462
JK		 624 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
625
0f18b462
JK		 626 spin_lock(&sbi->inode_lock[DIRTY_META]);
627 if (is_inode_flag_set(inode, FI_DIRTY_INODE)) {
628 spin_unlock(&sbi->inode_lock[DIRTY_META]);
b56ab837 629 return 1;
0f18b462
JK		 630 }
631
91942321 632 set_inode_flag(inode, FI_DIRTY_INODE);
0f18b462
JK		 633 list_add_tail(&F2FS_I(inode)->gdirty_list,
634 &sbi->inode_list[DIRTY_META]);
635 inc_page_count(sbi, F2FS_DIRTY_IMETA);
0f18b462 636 stat_inc_dirty_inode(sbi, DIRTY_META);
338bbfa0 637 spin_unlock(&sbi->inode_lock[DIRTY_META]);
b56ab837
JK		 638
639 return 0;
0f18b462
JK		 640}
641
642void f2fs_inode_synced(struct inode *inode)
643{
644 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
645
646 spin_lock(&sbi->inode_lock[DIRTY_META]);
647 if (!is_inode_flag_set(inode, FI_DIRTY_INODE)) {
648 spin_unlock(&sbi->inode_lock[DIRTY_META]);
649 return;
650 }
651 list_del_init(&F2FS_I(inode)->gdirty_list);
652 clear_inode_flag(inode, FI_DIRTY_INODE);
26de9b11 653 clear_inode_flag(inode, FI_AUTO_RECOVER);
0f18b462 654 dec_page_count(sbi, F2FS_DIRTY_IMETA);
0f18b462 655 stat_dec_dirty_inode(F2FS_I_SB(inode), DIRTY_META);
338bbfa0 656 spin_unlock(&sbi->inode_lock[DIRTY_META]);
b3783873
JK		 657}
658
b56ab837
JK		 659/*
660 * f2fs_dirty_inode() is called from __mark_inode_dirty()
661 *
662 * We should call set_dirty_inode to write the dirty inode through write_inode.
663 */
664static void f2fs_dirty_inode(struct inode *inode, int flags)
665{
666 struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
667
668 if (inode->i_ino == F2FS_NODE_INO(sbi) ||
669 inode->i_ino == F2FS_META_INO(sbi))
670 return;
671
672 if (flags == I_DIRTY_TIME)
673 return;
674
675 if (is_inode_flag_set(inode, FI_AUTO_RECOVER))
676 clear_inode_flag(inode, FI_AUTO_RECOVER);
677
678 f2fs_inode_dirtied(inode);
679}
680
aff063e2
JK		 681static void f2fs_i_callback(struct rcu_head *head)
682{
683 struct inode *inode = container_of(head, struct inode, i_rcu);
684 kmem_cache_free(f2fs_inode_cachep, F2FS_I(inode));
685}
686
25ca923b 687static void f2fs_destroy_inode(struct inode *inode)
aff063e2 688{
1beba1b3 689 percpu_counter_destroy(&F2FS_I(inode)->dirty_pages);
aff063e2
JK		 690 call_rcu(&inode->i_rcu, f2fs_i_callback);
691}
692
523be8a6
JK		 693static void destroy_percpu_info(struct f2fs_sb_info *sbi)
694{
695 int i;
696
697 for (i = 0; i < NR_COUNT_TYPE; i++)
698 percpu_counter_destroy(&sbi->nr_pages[i]);
41382ec4 699 percpu_counter_destroy(&sbi->alloc_valid_block_count);
513c5f37 700 percpu_counter_destroy(&sbi->total_valid_inode_count);
523be8a6
JK		 701}
702
aff063e2
JK		 703static void f2fs_put_super(struct super_block *sb)
704{
705 struct f2fs_sb_info *sbi = F2FS_SB(sb);
706
5e176d54
JK		 707 if (sbi->s_proc) {
708 remove_proc_entry("segment_info", sbi->s_proc);
f00d6fa7 709 remove_proc_entry("segment_bits", sbi->s_proc);
5e176d54
JK		 710 remove_proc_entry(sb->s_id, f2fs_proc_root);
711 }
b59d0bae 712 kobject_del(&sbi->s_kobj);
5e176d54 713
aff063e2
JK		 714 stop_gc_thread(sbi);
715
2658e50d
JK		 716 /* prevent remaining shrinker jobs */
717 mutex_lock(&sbi->umount_mutex);
718
85dc2f2c
JK		 719 /*
720 * We don't need to do checkpoint when superblock is clean.
721 * But, the previous checkpoint was not done by umount, it needs to do
722 * clean checkpoint again.
723 */
caf0047e 724 if (is_sbi_flag_set(sbi, SBI_IS_DIRTY) ||
aaec2b1d 725 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
75ab4cb8
JK		 726 struct cp_control cpc = {
727 .reason = CP_UMOUNT,
728 };
729 write_checkpoint(sbi, &cpc);
730 }
aff063e2 731
eca616f8
JK		 732 /* write_checkpoint can update stat informaion */
733 f2fs_destroy_stats(sbi);
734
cf779cab
JK		 735 /*
736 * normally superblock is clean, so we need to release this.
737 * In addition, EIO will skip do checkpoint, we need this as well.
738 */
74ef9241 739 release_ino_entry(sbi, true);
4b2fecc8 740 release_discard_addrs(sbi);
6f12ac25 741
2658e50d
JK		 742 f2fs_leave_shrinker(sbi);
743 mutex_unlock(&sbi->umount_mutex);
744
17c19120 745 /* our cp_error case, we can wait for any writeback page */
f5730184 746 f2fs_flush_merged_bios(sbi);
17c19120 747
aff063e2
JK		 748 iput(sbi->node_inode);
749 iput(sbi->meta_inode);
750
751 /* destroy f2fs internal modules */
752 destroy_node_manager(sbi);
753 destroy_segment_manager(sbi);
754
755 kfree(sbi->ckpt);
b59d0bae
NJ		 756 kobject_put(&sbi->s_kobj);
757 wait_for_completion(&sbi->s_kobj_unregister);
aff063e2
JK		 758
759 sb->s_fs_info = NULL;
43b6573b
KM		 760 if (sbi->s_chksum_driver)
761 crypto_free_shash(sbi->s_chksum_driver);
b39f0de2 762 kfree(sbi->raw_super);
523be8a6
JK		 763
764 destroy_percpu_info(sbi);
aff063e2
JK		 765 kfree(sbi);
766}
767
768int f2fs_sync_fs(struct super_block *sb, int sync)
769{
770 struct f2fs_sb_info *sbi = F2FS_SB(sb);
c34f42e2 771 int err = 0;
aff063e2 772
a2a4a7e4
NJ		 773 trace_f2fs_sync_fs(sb, sync);
774
b7473754 775 if (sync) {
d5053a34
JK		 776 struct cp_control cpc;
777
119ee914
JK		 778 cpc.reason = __get_cp_reason(sbi);
779
b7473754 780 mutex_lock(&sbi->gc_mutex);
c34f42e2 781 err = write_checkpoint(sbi, &cpc);
b7473754 782 mutex_unlock(&sbi->gc_mutex);
b7473754 783 }
05ca3632 784 f2fs_trace_ios(NULL, 1);
aff063e2 785
c34f42e2 786 return err;
aff063e2
JK		 787}
788
d6212a5f
CL		 789static int f2fs_freeze(struct super_block *sb)
790{
791 int err;
792
77888c1e 793 if (f2fs_readonly(sb))
d6212a5f
CL		 794 return 0;
795
796 err = f2fs_sync_fs(sb, 1);
797 return err;
798}
799
800static int f2fs_unfreeze(struct super_block *sb)
801{
802 return 0;
803}
804
aff063e2
JK		 805static int f2fs_statfs(struct dentry *dentry, struct kstatfs *buf)
806{
807 struct super_block *sb = dentry->d_sb;
808 struct f2fs_sb_info *sbi = F2FS_SB(sb);
809 u64 id = huge_encode_dev(sb->s_bdev->bd_dev);
810 block_t total_count, user_block_count, start_count, ovp_count;
811
812 total_count = le64_to_cpu(sbi->raw_super->block_count);
813 user_block_count = sbi->user_block_count;
814 start_count = le32_to_cpu(sbi->raw_super->segment0_blkaddr);
815 ovp_count = SM_I(sbi)->ovp_segments << sbi->log_blocks_per_seg;
816 buf->f_type = F2FS_SUPER_MAGIC;
817 buf->f_bsize = sbi->blocksize;
818
819 buf->f_blocks = total_count - start_count;
3e6d0b4d 820 buf->f_bfree = user_block_count - valid_user_blocks(sbi) + ovp_count;
aff063e2
JK		 821 buf->f_bavail = user_block_count - valid_user_blocks(sbi);
822
c200b1aa
CY		 823 buf->f_files = sbi->total_node_count - F2FS_RESERVED_NODE_NUM;
824 buf->f_ffree = buf->f_files - valid_inode_count(sbi);
aff063e2 825
5a20d339 826 buf->f_namelen = F2FS_NAME_LEN;
aff063e2
JK		 827 buf->f_fsid.val[0] = (u32)id;
828 buf->f_fsid.val[1] = (u32)(id >> 32);
829
830 return 0;
831}
832
833static int f2fs_show_options(struct seq_file *seq, struct dentry *root)
834{
835 struct f2fs_sb_info *sbi = F2FS_SB(root->d_sb);
836
6aefd93b
JK		 837 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, BG_GC)) {
838 if (test_opt(sbi, FORCE_FG_GC))
839 seq_printf(seq, ",background_gc=%s", "sync");
840 else
841 seq_printf(seq, ",background_gc=%s", "on");
842 } else {
696c018c 843 seq_printf(seq, ",background_gc=%s", "off");
6aefd93b 844 }
aff063e2
JK		 845 if (test_opt(sbi, DISABLE_ROLL_FORWARD))
846 seq_puts(seq, ",disable_roll_forward");
847 if (test_opt(sbi, DISCARD))
848 seq_puts(seq, ",discard");
849 if (test_opt(sbi, NOHEAP))
850 seq_puts(seq, ",no_heap_alloc");
851#ifdef CONFIG_F2FS_FS_XATTR
852 if (test_opt(sbi, XATTR_USER))
853 seq_puts(seq, ",user_xattr");
854 else
855 seq_puts(seq, ",nouser_xattr");
444c580f
JK		 856 if (test_opt(sbi, INLINE_XATTR))
857 seq_puts(seq, ",inline_xattr");
aff063e2
JK		 858#endif
859#ifdef CONFIG_F2FS_FS_POSIX_ACL
860 if (test_opt(sbi, POSIX_ACL))
861 seq_puts(seq, ",acl");
862 else
863 seq_puts(seq, ",noacl");
864#endif
865 if (test_opt(sbi, DISABLE_EXT_IDENTIFY))
aa43507f 866 seq_puts(seq, ",disable_ext_identify");
8274de77
HL		 867 if (test_opt(sbi, INLINE_DATA))
868 seq_puts(seq, ",inline_data");
75342797
WL		 869 else
870 seq_puts(seq, ",noinline_data");
5efd3c6f
CY		 871 if (test_opt(sbi, INLINE_DENTRY))
872 seq_puts(seq, ",inline_dentry");
97c1794a
CY		 873 else
874 seq_puts(seq, ",noinline_dentry");
b270ad6f 875 if (!f2fs_readonly(sbi->sb) && test_opt(sbi, FLUSH_MERGE))
6b4afdd7 876 seq_puts(seq, ",flush_merge");
0f7b2abd
JK		 877 if (test_opt(sbi, NOBARRIER))
878 seq_puts(seq, ",nobarrier");
d5053a34
JK		 879 if (test_opt(sbi, FASTBOOT))
880 seq_puts(seq, ",fastboot");
89672159
CY		 881 if (test_opt(sbi, EXTENT_CACHE))
882 seq_puts(seq, ",extent_cache");
7daaea25
JK		 883 else
884 seq_puts(seq, ",noextent_cache");
343f40f0
CY		 885 if (test_opt(sbi, DATA_FLUSH))
886 seq_puts(seq, ",data_flush");
36abef4e
JK		 887
888 seq_puts(seq, ",mode=");
889 if (test_opt(sbi, ADAPTIVE))
890 seq_puts(seq, "adaptive");
891 else if (test_opt(sbi, LFS))
892 seq_puts(seq, "lfs");
aff063e2
JK		 893 seq_printf(seq, ",active_logs=%u", sbi->active_logs);
894
895 return 0;
896}
897
5e176d54
JK		 898static int segment_info_seq_show(struct seq_file *seq, void *offset)
899{
900 struct super_block *sb = seq->private;
901 struct f2fs_sb_info *sbi = F2FS_SB(sb);
6c311ec6
CF		 902 unsigned int total_segs =
903 le32_to_cpu(sbi->raw_super->segment_count_main);
5e176d54
JK		 904 int i;
905
90aa6dc9
CY		 906 seq_puts(seq, "format: segment_type|valid_blocks\n"
907 "segment_type(0:HD, 1:WD, 2:CD, 3:HN, 4:WN, 5:CN)\n");
908
5e176d54 909 for (i = 0; i < total_segs; i++) {
90aa6dc9
CY		 910 struct seg_entry *se = get_seg_entry(sbi, i);
911
912 if ((i % 10) == 0)
01a5ad82 913 seq_printf(seq, "%-10d", i);
90aa6dc9
CY		 914 seq_printf(seq, "%d|%-3u", se->type,
915 get_valid_blocks(sbi, i, 1));
46c04366
GZ		 916 if ((i % 10) == 9 || i == (total_segs - 1))
917 seq_putc(seq, '\n');
5e176d54 918 else
46c04366 919 seq_putc(seq, ' ');
5e176d54 920 }
46c04366 921
5e176d54
JK		 922 return 0;
923}
924
f00d6fa7
JK		 925static int segment_bits_seq_show(struct seq_file *seq, void *offset)
926{
927 struct super_block *sb = seq->private;
928 struct f2fs_sb_info *sbi = F2FS_SB(sb);
929 unsigned int total_segs =
930 le32_to_cpu(sbi->raw_super->segment_count_main);
931 int i, j;
932
933 seq_puts(seq, "format: segment_type|valid_blocks|bitmaps\n"
934 "segment_type(0:HD, 1:WD, 2:CD, 3:HN, 4:WN, 5:CN)\n");
935
936 for (i = 0; i < total_segs; i++) {
937 struct seg_entry *se = get_seg_entry(sbi, i);
938
939 seq_printf(seq, "%-10d", i);
940 seq_printf(seq, "%d|%-3u|", se->type,
941 get_valid_blocks(sbi, i, 1));
942 for (j = 0; j < SIT_VBLOCK_MAP_SIZE; j++)
d32853de 943 seq_printf(seq, " %.2x", se->cur_valid_map[j]);
f00d6fa7
JK		 944 seq_putc(seq, '\n');
945 }
946 return 0;
947}
948
b7a15f3d
JK		 949#define F2FS_PROC_FILE_DEF(_name) \
950static int _name##_open_fs(struct inode *inode, struct file *file) \
951{ \
952 return single_open(file, _name##_seq_show, PDE_DATA(inode)); \
953} \
954 \
955static const struct file_operations f2fs_seq_##_name##_fops = { \
b7a15f3d
JK		 956 .open = _name##_open_fs, \
957 .read = seq_read, \
958 .llseek = seq_lseek, \
959 .release = single_release, \
5e176d54
JK		 960};
961
b7a15f3d 962F2FS_PROC_FILE_DEF(segment_info);
f00d6fa7 963F2FS_PROC_FILE_DEF(segment_bits);
b7a15f3d 964
498c5e9f
YH		 965static void default_options(struct f2fs_sb_info *sbi)
966{
967 /* init some FS parameters */
968 sbi->active_logs = NR_CURSEG_TYPE;
969
970 set_opt(sbi, BG_GC);
971 set_opt(sbi, INLINE_DATA);
97c1794a 972 set_opt(sbi, INLINE_DENTRY);
3e72f721 973 set_opt(sbi, EXTENT_CACHE);
6d94c74a 974 sbi->sb->s_flags |= MS_LAZYTIME;
69e9e427 975 set_opt(sbi, FLUSH_MERGE);
52763a4b
JK		 976 if (f2fs_sb_mounted_hmsmr(sbi->sb)) {
977 set_opt_mode(sbi, F2FS_MOUNT_LFS);
978 set_opt(sbi, DISCARD);
979 } else {
980 set_opt_mode(sbi, F2FS_MOUNT_ADAPTIVE);
981 }
498c5e9f
YH		 982
983#ifdef CONFIG_F2FS_FS_XATTR
984 set_opt(sbi, XATTR_USER);
985#endif
986#ifdef CONFIG_F2FS_FS_POSIX_ACL
987 set_opt(sbi, POSIX_ACL);
988#endif
36dbd328
CY		 989
990#ifdef CONFIG_F2FS_FAULT_INJECTION
991 f2fs_build_fault_attr(sbi, 0);
992#endif
498c5e9f
YH		 993}
994
696c018c
NJ		 995static int f2fs_remount(struct super_block *sb, int *flags, char *data)
996{
997 struct f2fs_sb_info *sbi = F2FS_SB(sb);
998 struct f2fs_mount_info org_mount_opt;
999 int err, active_logs;
876dc59e
GZ		 1000 bool need_restart_gc = false;
1001 bool need_stop_gc = false;
9cd81ce3 1002 bool no_extent_cache = !test_opt(sbi, EXTENT_CACHE);
696c018c
NJ		 1003
1004 /*
1005 * Save the old mount options in case we
1006 * need to restore them.
1007 */
1008 org_mount_opt = sbi->mount_opt;
1009 active_logs = sbi->active_logs;
1010
df728b0f
JK		 1011 /* recover superblocks we couldn't write due to previous RO mount */
1012 if (!(*flags & MS_RDONLY) && is_sbi_flag_set(sbi, SBI_NEED_SB_WRITE)) {
1013 err = f2fs_commit_super(sbi, false);
1014 f2fs_msg(sb, KERN_INFO,
1015 "Try to recover all the superblocks, ret: %d", err);
1016 if (!err)
1017 clear_sbi_flag(sbi, SBI_NEED_SB_WRITE);
1018 }
1019
26666c8a 1020 sbi->mount_opt.opt = 0;
498c5e9f 1021 default_options(sbi);
26666c8a 1022
696c018c
NJ		 1023 /* parse mount options */
1024 err = parse_options(sb, data);
1025 if (err)
1026 goto restore_opts;
1027
1028 /*
1029 * Previous and new state of filesystem is RO,
876dc59e 1030 * so skip checking GC and FLUSH_MERGE conditions.
696c018c 1031 */
6b2920a5 1032 if (f2fs_readonly(sb) && (*flags & MS_RDONLY))
696c018c
NJ		 1033 goto skip;
1034
9cd81ce3
CY		 1035 /* disallow enable/disable extent_cache dynamically */
1036 if (no_extent_cache == !!test_opt(sbi, EXTENT_CACHE)) {
1037 err = -EINVAL;
1038 f2fs_msg(sbi->sb, KERN_WARNING,
1039 "switch extent_cache option is not allowed");
1040 goto restore_opts;
1041 }
1042
696c018c
NJ		 1043 /*
1044 * We stop the GC thread if FS is mounted as RO
1045 * or if background_gc = off is passed in mount
1046 * option. Also sync the filesystem.
1047 */
1048 if ((*flags & MS_RDONLY) || !test_opt(sbi, BG_GC)) {
1049 if (sbi->gc_thread) {
1050 stop_gc_thread(sbi);
876dc59e 1051 need_restart_gc = true;
696c018c 1052 }
aba291b3 1053 } else if (!sbi->gc_thread) {
696c018c
NJ		 1054 err = start_gc_thread(sbi);
1055 if (err)
1056 goto restore_opts;
876dc59e
GZ		 1057 need_stop_gc = true;
1058 }
1059
faa0e55b
JK		 1060 if (*flags & MS_RDONLY) {
1061 writeback_inodes_sb(sb, WB_REASON_SYNC);
1062 sync_inodes_sb(sb);
1063
1064 set_sbi_flag(sbi, SBI_IS_DIRTY);
1065 set_sbi_flag(sbi, SBI_IS_CLOSE);
1066 f2fs_sync_fs(sb, 1);
1067 clear_sbi_flag(sbi, SBI_IS_CLOSE);
1068 }
1069
876dc59e
GZ		 1070 /*
1071 * We stop issue flush thread if FS is mounted as RO
1072 * or if flush_merge is not passed in mount option.
1073 */
1074 if ((*flags & MS_RDONLY) || !test_opt(sbi, FLUSH_MERGE)) {
2163d198 1075 destroy_flush_cmd_control(sbi);
aba291b3 1076 } else if (!SM_I(sbi)->cmd_control_info) {
2163d198
GZ		 1077 err = create_flush_cmd_control(sbi);
1078 if (err)
a688b9d9 1079 goto restore_gc;
696c018c
NJ		 1080 }
1081skip:
1082 /* Update the POSIXACL Flag */
df728b0f 1083 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
696c018c 1084 (test_opt(sbi, POSIX_ACL) ? MS_POSIXACL : 0);
df728b0f 1085
696c018c 1086 return 0;
876dc59e
GZ		 1087restore_gc:
1088 if (need_restart_gc) {
1089 if (start_gc_thread(sbi))
1090 f2fs_msg(sbi->sb, KERN_WARNING,
e1c42045 1091 "background gc thread has stopped");
876dc59e
GZ		 1092 } else if (need_stop_gc) {
1093 stop_gc_thread(sbi);
1094 }
696c018c
NJ		 1095restore_opts:
1096 sbi->mount_opt = org_mount_opt;
1097 sbi->active_logs = active_logs;
1098 return err;
1099}
1100
aff063e2
JK		 1101static struct super_operations f2fs_sops = {
1102 .alloc_inode = f2fs_alloc_inode,
531ad7d5 1103 .drop_inode = f2fs_drop_inode,
aff063e2
JK		 1104 .destroy_inode = f2fs_destroy_inode,
1105 .write_inode = f2fs_write_inode,
b3783873 1106 .dirty_inode = f2fs_dirty_inode,
aff063e2
JK		 1107 .show_options = f2fs_show_options,
1108 .evict_inode = f2fs_evict_inode,
1109 .put_super = f2fs_put_super,
1110 .sync_fs = f2fs_sync_fs,
d6212a5f
CL		 1111 .freeze_fs = f2fs_freeze,
1112 .unfreeze_fs = f2fs_unfreeze,
aff063e2 1113 .statfs = f2fs_statfs,
696c018c 1114 .remount_fs = f2fs_remount,
aff063e2
JK		 1115};
1116
0b81d077
JK		 1117#ifdef CONFIG_F2FS_FS_ENCRYPTION
1118static int f2fs_get_context(struct inode *inode, void *ctx, size_t len)
1119{
1120 return f2fs_getxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
1121 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
1122 ctx, len, NULL);
1123}
1124
b5a7aef1
JK		 1125static int f2fs_key_prefix(struct inode *inode, u8 **key)
1126{
1127 *key = F2FS_I_SB(inode)->key_prefix;
1128 return F2FS_I_SB(inode)->key_prefix_size;
1129}
1130
0b81d077
JK		 1131static int f2fs_set_context(struct inode *inode, const void *ctx, size_t len,
1132 void *fs_data)
1133{
1134 return f2fs_setxattr(inode, F2FS_XATTR_INDEX_ENCRYPTION,
1135 F2FS_XATTR_NAME_ENCRYPTION_CONTEXT,
1136 ctx, len, fs_data, XATTR_CREATE);
1137}
1138
1139static unsigned f2fs_max_namelen(struct inode *inode)
1140{
1141 return S_ISLNK(inode->i_mode) ?
1142 inode->i_sb->s_blocksize : F2FS_NAME_LEN;
1143}
1144
1145static struct fscrypt_operations f2fs_cryptops = {
1146 .get_context = f2fs_get_context,
b5a7aef1 1147 .key_prefix = f2fs_key_prefix,
0b81d077
JK		 1148 .set_context = f2fs_set_context,
1149 .is_encrypted = f2fs_encrypted_inode,
1150 .empty_dir = f2fs_empty_dir,
1151 .max_namelen = f2fs_max_namelen,
1152};
1153#else
1154static struct fscrypt_operations f2fs_cryptops = {
1155 .is_encrypted = f2fs_encrypted_inode,
1156};
1157#endif
1158
aff063e2
JK		 1159static struct inode *f2fs_nfs_get_inode(struct super_block *sb,
1160 u64 ino, u32 generation)
1161{
1162 struct f2fs_sb_info *sbi = F2FS_SB(sb);
1163 struct inode *inode;
1164
d6b7d4b3 1165 if (check_nid_range(sbi, ino))
910bb12d 1166 return ERR_PTR(-ESTALE);
aff063e2
JK		 1167
1168 /*
1169 * f2fs_iget isn't quite right if the inode is currently unallocated!
1170 * However f2fs_iget currently does appropriate checks to handle stale
1171 * inodes so everything is OK.
1172 */
1173 inode = f2fs_iget(sb, ino);
1174 if (IS_ERR(inode))
1175 return ERR_CAST(inode);
6bacf52f 1176 if (unlikely(generation && inode->i_generation != generation)) {
aff063e2
JK		 1177 /* we didn't find the right inode.. */
1178 iput(inode);
1179 return ERR_PTR(-ESTALE);
1180 }
1181 return inode;
1182}
1183
1184static struct dentry *f2fs_fh_to_dentry(struct super_block *sb, struct fid *fid,
1185 int fh_len, int fh_type)
1186{
1187 return generic_fh_to_dentry(sb, fid, fh_len, fh_type,
1188 f2fs_nfs_get_inode);
1189}
1190
1191static struct dentry *f2fs_fh_to_parent(struct super_block *sb, struct fid *fid,
1192 int fh_len, int fh_type)
1193{
1194 return generic_fh_to_parent(sb, fid, fh_len, fh_type,
1195 f2fs_nfs_get_inode);
1196}
1197
1198static const struct export_operations f2fs_export_ops = {
1199 .fh_to_dentry = f2fs_fh_to_dentry,
1200 .fh_to_parent = f2fs_fh_to_parent,
1201 .get_parent = f2fs_get_parent,
1202};
1203
e0afc4d6 1204static loff_t max_file_blocks(void)
aff063e2 1205{
de93653f 1206 loff_t result = (DEF_ADDRS_PER_INODE - F2FS_INLINE_XATTR_ADDRS);
aff063e2
JK		 1207 loff_t leaf_count = ADDRS_PER_BLOCK;
1208
1209 /* two direct node blocks */
1210 result += (leaf_count * 2);
1211
1212 /* two indirect node blocks */
1213 leaf_count *= NIDS_PER_BLOCK;
1214 result += (leaf_count * 2);
1215
1216 /* one double indirect node block */
1217 leaf_count *= NIDS_PER_BLOCK;
1218 result += leaf_count;
1219
aff063e2
JK		 1220 return result;
1221}
1222
fd694733
JK		 1223static int __f2fs_commit_super(struct buffer_head *bh,
1224 struct f2fs_super_block *super)
1225{
1226 lock_buffer(bh);
1227 if (super)
1228 memcpy(bh->b_data + F2FS_SUPER_OFFSET, super, sizeof(*super));
1229 set_buffer_uptodate(bh);
1230 set_buffer_dirty(bh);
1231 unlock_buffer(bh);
1232
1233 /* it's rare case, we can do fua all the time */
1234 return __sync_dirty_buffer(bh, WRITE_FLUSH_FUA);
1235}
1236
df728b0f 1237static inline bool sanity_check_area_boundary(struct f2fs_sb_info *sbi,
fd694733 1238 struct buffer_head *bh)
9a59b62f 1239{
fd694733
JK		 1240 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
1241 (bh->b_data + F2FS_SUPER_OFFSET);
df728b0f 1242 struct super_block *sb = sbi->sb;
9a59b62f
CY		 1243 u32 segment0_blkaddr = le32_to_cpu(raw_super->segment0_blkaddr);
1244 u32 cp_blkaddr = le32_to_cpu(raw_super->cp_blkaddr);
1245 u32 sit_blkaddr = le32_to_cpu(raw_super->sit_blkaddr);
1246 u32 nat_blkaddr = le32_to_cpu(raw_super->nat_blkaddr);
1247 u32 ssa_blkaddr = le32_to_cpu(raw_super->ssa_blkaddr);
1248 u32 main_blkaddr = le32_to_cpu(raw_super->main_blkaddr);
1249 u32 segment_count_ckpt = le32_to_cpu(raw_super->segment_count_ckpt);
1250 u32 segment_count_sit = le32_to_cpu(raw_super->segment_count_sit);
1251 u32 segment_count_nat = le32_to_cpu(raw_super->segment_count_nat);
1252 u32 segment_count_ssa = le32_to_cpu(raw_super->segment_count_ssa);
1253 u32 segment_count_main = le32_to_cpu(raw_super->segment_count_main);
1254 u32 segment_count = le32_to_cpu(raw_super->segment_count);
1255 u32 log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
fd694733
JK		 1256 u64 main_end_blkaddr = main_blkaddr +
1257 (segment_count_main << log_blocks_per_seg);
1258 u64 seg_end_blkaddr = segment0_blkaddr +
1259 (segment_count << log_blocks_per_seg);
9a59b62f
CY		 1260
1261 if (segment0_blkaddr != cp_blkaddr) {
1262 f2fs_msg(sb, KERN_INFO,
1263 "Mismatch start address, segment0(%u) cp_blkaddr(%u)",
1264 segment0_blkaddr, cp_blkaddr);
1265 return true;
1266 }
1267
1268 if (cp_blkaddr + (segment_count_ckpt << log_blocks_per_seg) !=
1269 sit_blkaddr) {
1270 f2fs_msg(sb, KERN_INFO,
1271 "Wrong CP boundary, start(%u) end(%u) blocks(%u)",
1272 cp_blkaddr, sit_blkaddr,
1273 segment_count_ckpt << log_blocks_per_seg);
1274 return true;
1275 }
1276
1277 if (sit_blkaddr + (segment_count_sit << log_blocks_per_seg) !=
1278 nat_blkaddr) {
1279 f2fs_msg(sb, KERN_INFO,
1280 "Wrong SIT boundary, start(%u) end(%u) blocks(%u)",
1281 sit_blkaddr, nat_blkaddr,
1282 segment_count_sit << log_blocks_per_seg);
1283 return true;
1284 }
1285
1286 if (nat_blkaddr + (segment_count_nat << log_blocks_per_seg) !=
1287 ssa_blkaddr) {
1288 f2fs_msg(sb, KERN_INFO,
1289 "Wrong NAT boundary, start(%u) end(%u) blocks(%u)",
1290 nat_blkaddr, ssa_blkaddr,
1291 segment_count_nat << log_blocks_per_seg);
1292 return true;
1293 }
1294
1295 if (ssa_blkaddr + (segment_count_ssa << log_blocks_per_seg) !=
1296 main_blkaddr) {
1297 f2fs_msg(sb, KERN_INFO,
1298 "Wrong SSA boundary, start(%u) end(%u) blocks(%u)",
1299 ssa_blkaddr, main_blkaddr,
1300 segment_count_ssa << log_blocks_per_seg);
1301 return true;
1302 }
1303
fd694733 1304 if (main_end_blkaddr > seg_end_blkaddr) {
9a59b62f 1305 f2fs_msg(sb, KERN_INFO,
fd694733 1306 "Wrong MAIN_AREA boundary, start(%u) end(%u) block(%u)",
9a59b62f 1307 main_blkaddr,
fd694733
JK		 1308 segment0_blkaddr +
1309 (segment_count << log_blocks_per_seg),
9a59b62f
CY		 1310 segment_count_main << log_blocks_per_seg);
1311 return true;
fd694733
JK		 1312 } else if (main_end_blkaddr < seg_end_blkaddr) {
1313 int err = 0;
1314 char *res;
1315
1316 /* fix in-memory information all the time */
1317 raw_super->segment_count = cpu_to_le32((main_end_blkaddr -
1318 segment0_blkaddr) >> log_blocks_per_seg);
1319
1320 if (f2fs_readonly(sb) || bdev_read_only(sb->s_bdev)) {
df728b0f 1321 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
fd694733
JK		 1322 res = "internally";
1323 } else {
1324 err = __f2fs_commit_super(bh, NULL);
1325 res = err ? "failed" : "done";
1326 }
1327 f2fs_msg(sb, KERN_INFO,
1328 "Fix alignment : %s, start(%u) end(%u) block(%u)",
1329 res, main_blkaddr,
1330 segment0_blkaddr +
1331 (segment_count << log_blocks_per_seg),
1332 segment_count_main << log_blocks_per_seg);
1333 if (err)
1334 return true;
9a59b62f 1335 }
9a59b62f
CY		 1336 return false;
1337}
1338
df728b0f 1339static int sanity_check_raw_super(struct f2fs_sb_info *sbi,
fd694733 1340 struct buffer_head *bh)
aff063e2 1341{
fd694733
JK		 1342 struct f2fs_super_block *raw_super = (struct f2fs_super_block *)
1343 (bh->b_data + F2FS_SUPER_OFFSET);
df728b0f 1344 struct super_block *sb = sbi->sb;
aff063e2
JK		 1345 unsigned int blocksize;
1346
a07ef784
NJ		 1347 if (F2FS_SUPER_MAGIC != le32_to_cpu(raw_super->magic)) {
1348 f2fs_msg(sb, KERN_INFO,
1349 "Magic Mismatch, valid(0x%x) - read(0x%x)",
1350 F2FS_SUPER_MAGIC, le32_to_cpu(raw_super->magic));
aff063e2 1351 return 1;
a07ef784 1352 }
aff063e2 1353
5c9b4692 1354 /* Currently, support only 4KB page cache size */
09cbfeaf 1355 if (F2FS_BLKSIZE != PAGE_SIZE) {
5c9b4692 1356 f2fs_msg(sb, KERN_INFO,
14d7e9de 1357 "Invalid page_cache_size (%lu), supports only 4KB\n",
09cbfeaf 1358 PAGE_SIZE);
5c9b4692 1359 return 1;
1360 }
1361
aff063e2
JK		 1362 /* Currently, support only 4KB block size */
1363 blocksize = 1 << le32_to_cpu(raw_super->log_blocksize);
5c9b4692 1364 if (blocksize != F2FS_BLKSIZE) {
a07ef784
NJ		 1365 f2fs_msg(sb, KERN_INFO,
1366 "Invalid blocksize (%u), supports only 4KB\n",
1367 blocksize);
aff063e2 1368 return 1;
a07ef784 1369 }
5c9b4692 1370
9a59b62f
CY		 1371 /* check log blocks per segment */
1372 if (le32_to_cpu(raw_super->log_blocks_per_seg) != 9) {
1373 f2fs_msg(sb, KERN_INFO,
1374 "Invalid log blocks per segment (%u)\n",
1375 le32_to_cpu(raw_super->log_blocks_per_seg));
1376 return 1;
1377 }
1378
55cf9cb6
CY		 1379 /* Currently, support 512/1024/2048/4096 bytes sector size */
1380 if (le32_to_cpu(raw_super->log_sectorsize) >
1381 F2FS_MAX_LOG_SECTOR_SIZE ||
1382 le32_to_cpu(raw_super->log_sectorsize) <
1383 F2FS_MIN_LOG_SECTOR_SIZE) {
1384 f2fs_msg(sb, KERN_INFO, "Invalid log sectorsize (%u)",
1385 le32_to_cpu(raw_super->log_sectorsize));
aff063e2 1386 return 1;
a07ef784 1387 }
55cf9cb6
CY		 1388 if (le32_to_cpu(raw_super->log_sectors_per_block) +
1389 le32_to_cpu(raw_super->log_sectorsize) !=
1390 F2FS_MAX_LOG_SECTOR_SIZE) {
1391 f2fs_msg(sb, KERN_INFO,
1392 "Invalid log sectors per block(%u) log sectorsize(%u)",
1393 le32_to_cpu(raw_super->log_sectors_per_block),
1394 le32_to_cpu(raw_super->log_sectorsize));
aff063e2 1395 return 1;
a07ef784 1396 }
9a59b62f
CY		 1397
1398 /* check reserved ino info */
1399 if (le32_to_cpu(raw_super->node_ino) != 1 ||
1400 le32_to_cpu(raw_super->meta_ino) != 2 ||
1401 le32_to_cpu(raw_super->root_ino) != 3) {
1402 f2fs_msg(sb, KERN_INFO,
1403 "Invalid Fs Meta Ino: node(%u) meta(%u) root(%u)",
1404 le32_to_cpu(raw_super->node_ino),
1405 le32_to_cpu(raw_super->meta_ino),
1406 le32_to_cpu(raw_super->root_ino));
1407 return 1;
1408 }
1409
1410 /* check CP/SIT/NAT/SSA/MAIN_AREA area boundary */
df728b0f 1411 if (sanity_check_area_boundary(sbi, bh))
9a59b62f
CY		 1412 return 1;
1413
aff063e2
JK		 1414 return 0;
1415}
1416
984ec63c 1417int sanity_check_ckpt(struct f2fs_sb_info *sbi)
aff063e2
JK		 1418{
1419 unsigned int total, fsmeta;
577e3495
JK		 1420 struct f2fs_super_block *raw_super = F2FS_RAW_SUPER(sbi);
1421 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi);
aff063e2
JK		 1422
1423 total = le32_to_cpu(raw_super->segment_count);
1424 fsmeta = le32_to_cpu(raw_super->segment_count_ckpt);
1425 fsmeta += le32_to_cpu(raw_super->segment_count_sit);
1426 fsmeta += le32_to_cpu(raw_super->segment_count_nat);
1427 fsmeta += le32_to_cpu(ckpt->rsvd_segment_count);
1428 fsmeta += le32_to_cpu(raw_super->segment_count_ssa);
1429
6bacf52f 1430 if (unlikely(fsmeta >= total))
aff063e2 1431 return 1;
577e3495 1432
1e968fdf 1433 if (unlikely(f2fs_cp_error(sbi))) {
577e3495
JK		 1434 f2fs_msg(sbi->sb, KERN_ERR, "A bug case: need to run fsck");
1435 return 1;
1436 }
aff063e2
JK		 1437 return 0;
1438}
1439
1440static void init_sb_info(struct f2fs_sb_info *sbi)
1441{
1442 struct f2fs_super_block *raw_super = sbi->raw_super;
aff063e2
JK		 1443
1444 sbi->log_sectors_per_block =
1445 le32_to_cpu(raw_super->log_sectors_per_block);
1446 sbi->log_blocksize = le32_to_cpu(raw_super->log_blocksize);
1447 sbi->blocksize = 1 << sbi->log_blocksize;
1448 sbi->log_blocks_per_seg = le32_to_cpu(raw_super->log_blocks_per_seg);
1449 sbi->blocks_per_seg = 1 << sbi->log_blocks_per_seg;
1450 sbi->segs_per_sec = le32_to_cpu(raw_super->segs_per_sec);
1451 sbi->secs_per_zone = le32_to_cpu(raw_super->secs_per_zone);
1452 sbi->total_sections = le32_to_cpu(raw_super->section_count);
1453 sbi->total_node_count =
1454 (le32_to_cpu(raw_super->segment_count_nat) / 2)
1455 * sbi->blocks_per_seg * NAT_ENTRY_PER_BLOCK;
1456 sbi->root_ino_num = le32_to_cpu(raw_super->root_ino);
1457 sbi->node_ino_num = le32_to_cpu(raw_super->node_ino);
1458 sbi->meta_ino_num = le32_to_cpu(raw_super->meta_ino);
5ec4e49f 1459 sbi->cur_victim_sec = NULL_SECNO;
b1c57c1c 1460 sbi->max_victim_search = DEF_MAX_VICTIM_SEARCH;
aff063e2 1461
ab9fa662 1462 sbi->dir_level = DEF_DIR_LEVEL;
6beceb54 1463 sbi->interval_time[CP_TIME] = DEF_CP_INTERVAL;
d0239e1b 1464 sbi->interval_time[REQ_TIME] = DEF_IDLE_INTERVAL;
caf0047e 1465 clear_sbi_flag(sbi, SBI_NEED_FSCK);
2658e50d
JK		 1466
1467 INIT_LIST_HEAD(&sbi->s_list);
1468 mutex_init(&sbi->umount_mutex);
7dfeaa32
JK		 1469 mutex_init(&sbi->wio_mutex[NODE]);
1470 mutex_init(&sbi->wio_mutex[DATA]);
aaec2b1d 1471 spin_lock_init(&sbi->cp_lock);
b5a7aef1
JK		 1472
1473#ifdef CONFIG_F2FS_FS_ENCRYPTION
1474 memcpy(sbi->key_prefix, F2FS_KEY_DESC_PREFIX,
1475 F2FS_KEY_DESC_PREFIX_SIZE);
1476 sbi->key_prefix_size = F2FS_KEY_DESC_PREFIX_SIZE;
1477#endif
aff063e2
JK		 1478}
1479
523be8a6
JK		 1480static int init_percpu_info(struct f2fs_sb_info *sbi)
1481{
1482 int i, err;
1483
1484 for (i = 0; i < NR_COUNT_TYPE; i++) {
1485 err = percpu_counter_init(&sbi->nr_pages[i], 0, GFP_KERNEL);
1486 if (err)
1487 return err;
1488 }
41382ec4 1489
513c5f37
JK		 1490 err = percpu_counter_init(&sbi->alloc_valid_block_count, 0, GFP_KERNEL);
1491 if (err)
1492 return err;
1493
1494 return percpu_counter_init(&sbi->total_valid_inode_count, 0,
41382ec4 1495 GFP_KERNEL);
523be8a6
JK		 1496}
1497
9076a75f
GZ		 1498/*
1499 * Read f2fs raw super block.
2b39e907
SL		 1500 * Because we have two copies of super block, so read both of them
1501 * to get the first valid one. If any one of them is broken, we pass
1502 * them recovery flag back to the caller.
9076a75f 1503 */
df728b0f 1504static int read_raw_super_block(struct f2fs_sb_info *sbi,
9076a75f 1505 struct f2fs_super_block **raw_super,
e8240f65 1506 int *valid_super_block, int *recovery)
14d7e9de 1507{
df728b0f 1508 struct super_block *sb = sbi->sb;
2b39e907 1509 int block;
e8240f65 1510 struct buffer_head *bh;
fd694733 1511 struct f2fs_super_block *super;
da554e48 1512 int err = 0;
14d7e9de 1513
b39f0de2
YH		 1514 super = kzalloc(sizeof(struct f2fs_super_block), GFP_KERNEL);
1515 if (!super)
1516 return -ENOMEM;
2b39e907
SL		 1517
1518 for (block = 0; block < 2; block++) {
1519 bh = sb_bread(sb, block);
1520 if (!bh) {
1521 f2fs_msg(sb, KERN_ERR, "Unable to read %dth superblock",
9076a75f 1522 block + 1);
2b39e907
SL		 1523 err = -EIO;
1524 continue;
1525 }
14d7e9de 1526
2b39e907 1527 /* sanity checking of raw super */
df728b0f 1528 if (sanity_check_raw_super(sbi, bh)) {
2b39e907
SL		 1529 f2fs_msg(sb, KERN_ERR,
1530 "Can't find valid F2FS filesystem in %dth superblock",
1531 block + 1);
1532 err = -EINVAL;
1533 brelse(bh);
1534 continue;
1535 }
14d7e9de 1536
2b39e907 1537 if (!*raw_super) {
fd694733
JK		 1538 memcpy(super, bh->b_data + F2FS_SUPER_OFFSET,
1539 sizeof(*super));
2b39e907
SL		 1540 *valid_super_block = block;
1541 *raw_super = super;
1542 }
1543 brelse(bh);
da554e48 1544 }
1545
2b39e907
SL		 1546 /* Fail to read any one of the superblocks*/
1547 if (err < 0)
1548 *recovery = 1;
da554e48 1549
da554e48 1550 /* No valid superblock */
2b39e907 1551 if (!*raw_super)
b39f0de2 1552 kfree(super);
2b39e907
SL		 1553 else
1554 err = 0;
da554e48 1555
2b39e907 1556 return err;
14d7e9de 1557}
1558
fd694733 1559int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover)
26d815ad 1560{
5d909cdb 1561 struct buffer_head *bh;
26d815ad
JK		 1562 int err;
1563
df728b0f
JK		 1564 if ((recover && f2fs_readonly(sbi->sb)) ||
1565 bdev_read_only(sbi->sb->s_bdev)) {
1566 set_sbi_flag(sbi, SBI_NEED_SB_WRITE);
f2353d7b 1567 return -EROFS;
df728b0f 1568 }
f2353d7b 1569
fd694733
JK		 1570 /* write back-up superblock first */
1571 bh = sb_getblk(sbi->sb, sbi->valid_super_block ? 0: 1);
5d909cdb
JK		 1572 if (!bh)
1573 return -EIO;
fd694733 1574 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
5d909cdb 1575 brelse(bh);
c5bda1c8
CY		 1576
1577 /* if we are in recovery path, skip writing valid superblock */
1578 if (recover || err)
5d909cdb 1579 return err;
26d815ad
JK		 1580
1581 /* write current valid superblock */
fd694733
JK		 1582 bh = sb_getblk(sbi->sb, sbi->valid_super_block);
1583 if (!bh)
1584 return -EIO;
1585 err = __f2fs_commit_super(bh, F2FS_RAW_SUPER(sbi));
1586 brelse(bh);
1587 return err;
26d815ad
JK		 1588}
1589
aff063e2
JK		 1590static int f2fs_fill_super(struct super_block *sb, void *data, int silent)
1591{
1592 struct f2fs_sb_info *sbi;
da554e48 1593 struct f2fs_super_block *raw_super;
aff063e2 1594 struct inode *root;
99e3e858 1595 int err;
2adc3505 1596 bool retry = true, need_fsck = false;
dabc4a5c 1597 char *options = NULL;
e8240f65 1598 int recovery, i, valid_super_block;
8f1dbbbb 1599 struct curseg_info *seg_i;
aff063e2 1600
ed2e621a 1601try_onemore:
da554e48 1602 err = -EINVAL;
1603 raw_super = NULL;
e8240f65 1604 valid_super_block = -1;
da554e48 1605 recovery = 0;
1606
aff063e2
JK		 1607 /* allocate memory for f2fs-specific super block info */
1608 sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
1609 if (!sbi)
1610 return -ENOMEM;
1611
df728b0f
JK		 1612 sbi->sb = sb;
1613
43b6573b
KM		 1614 /* Load the checksum driver */
1615 sbi->s_chksum_driver = crypto_alloc_shash("crc32", 0, 0);
1616 if (IS_ERR(sbi->s_chksum_driver)) {
1617 f2fs_msg(sb, KERN_ERR, "Cannot load crc32 driver.");
1618 err = PTR_ERR(sbi->s_chksum_driver);
1619 sbi->s_chksum_driver = NULL;
1620 goto free_sbi;
1621 }
1622
ff9234ad 1623 /* set a block size */
6bacf52f 1624 if (unlikely(!sb_set_blocksize(sb, F2FS_BLKSIZE))) {
a07ef784 1625 f2fs_msg(sb, KERN_ERR, "unable to set blocksize");
aff063e2 1626 goto free_sbi;
a07ef784 1627 }
aff063e2 1628
df728b0f 1629 err = read_raw_super_block(sbi, &raw_super, &valid_super_block,
e8240f65 1630 &recovery);
9076a75f
GZ		 1631 if (err)
1632 goto free_sbi;
1633
5fb08372 1634 sb->s_fs_info = sbi;
52763a4b
JK		 1635 sbi->raw_super = raw_super;
1636
498c5e9f 1637 default_options(sbi);
aff063e2 1638 /* parse mount options */
dabc4a5c
JK		 1639 options = kstrdup((const char *)data, GFP_KERNEL);
1640 if (data && !options) {
1641 err = -ENOMEM;
aff063e2 1642 goto free_sb_buf;
dabc4a5c
JK		 1643 }
1644
1645 err = parse_options(sb, options);
1646 if (err)
1647 goto free_options;
aff063e2 1648
e0afc4d6
CY		 1649 sbi->max_file_blocks = max_file_blocks();
1650 sb->s_maxbytes = sbi->max_file_blocks <<
1651 le32_to_cpu(raw_super->log_blocksize);
aff063e2
JK		 1652 sb->s_max_links = F2FS_LINK_MAX;
1653 get_random_bytes(&sbi->s_next_generation, sizeof(u32));
1654
1655 sb->s_op = &f2fs_sops;
0b81d077 1656 sb->s_cop = &f2fs_cryptops;
aff063e2
JK		 1657 sb->s_xattr = f2fs_xattr_handlers;
1658 sb->s_export_op = &f2fs_export_ops;
1659 sb->s_magic = F2FS_SUPER_MAGIC;
aff063e2
JK		 1660 sb->s_time_gran = 1;
1661 sb->s_flags = (sb->s_flags & ~MS_POSIXACL) |
1662 (test_opt(sbi, POSIX_ACL) ? MS_POSIXACL : 0);
1663 memcpy(sb->s_uuid, raw_super->uuid, sizeof(raw_super->uuid));
1664
1665 /* init f2fs-specific super block info */
e8240f65 1666 sbi->valid_super_block = valid_super_block;
aff063e2 1667 mutex_init(&sbi->gc_mutex);
aff063e2 1668 mutex_init(&sbi->cp_mutex);
b3582c68 1669 init_rwsem(&sbi->node_write);
315df839
JK		 1670
1671 /* disallow all the data/node/meta page writes */
1672 set_sbi_flag(sbi, SBI_POR_DOING);
aff063e2 1673 spin_lock_init(&sbi->stat_lock);
971767ca 1674
df0f8dc0 1675 init_rwsem(&sbi->read_io.io_rwsem);
458e6197
JK		 1676 sbi->read_io.sbi = sbi;
1677 sbi->read_io.bio = NULL;
1678 for (i = 0; i < NR_PAGE_TYPE; i++) {
df0f8dc0 1679 init_rwsem(&sbi->write_io[i].io_rwsem);
458e6197
JK		 1680 sbi->write_io[i].sbi = sbi;
1681 sbi->write_io[i].bio = NULL;
1682 }
971767ca 1683
b873b798 1684 init_rwsem(&sbi->cp_rwsem);
fb51b5ef 1685 init_waitqueue_head(&sbi->cp_wait);
aff063e2
JK		 1686 init_sb_info(sbi);
1687
523be8a6
JK		 1688 err = init_percpu_info(sbi);
1689 if (err)
1690 goto free_options;
1691
aff063e2
JK		 1692 /* get an inode for meta space */
1693 sbi->meta_inode = f2fs_iget(sb, F2FS_META_INO(sbi));
1694 if (IS_ERR(sbi->meta_inode)) {
a07ef784 1695 f2fs_msg(sb, KERN_ERR, "Failed to read F2FS meta data inode");
aff063e2 1696 err = PTR_ERR(sbi->meta_inode);
dabc4a5c 1697 goto free_options;
aff063e2
JK		 1698 }
1699
1700 err = get_valid_checkpoint(sbi);
a07ef784
NJ		 1701 if (err) {
1702 f2fs_msg(sb, KERN_ERR, "Failed to get valid F2FS checkpoint");
aff063e2 1703 goto free_meta_inode;
a07ef784 1704 }
aff063e2 1705
aff063e2
JK		 1706 sbi->total_valid_node_count =
1707 le32_to_cpu(sbi->ckpt->valid_node_count);
513c5f37
JK		 1708 percpu_counter_set(&sbi->total_valid_inode_count,
1709 le32_to_cpu(sbi->ckpt->valid_inode_count));
aff063e2
JK		 1710 sbi->user_block_count = le64_to_cpu(sbi->ckpt->user_block_count);
1711 sbi->total_valid_block_count =
1712 le64_to_cpu(sbi->ckpt->valid_block_count);
1713 sbi->last_valid_block_count = sbi->total_valid_block_count;
41382ec4 1714
c227f912
CY		 1715 for (i = 0; i < NR_INODE_TYPE; i++) {
1716 INIT_LIST_HEAD(&sbi->inode_list[i]);
1717 spin_lock_init(&sbi->inode_lock[i]);
1718 }
aff063e2 1719
1dcc336b
CY		 1720 init_extent_cache_info(sbi);
1721
6451e041 1722 init_ino_entry_info(sbi);
aff063e2
JK		 1723
1724 /* setup f2fs internal modules */
1725 err = build_segment_manager(sbi);
a07ef784
NJ		 1726 if (err) {
1727 f2fs_msg(sb, KERN_ERR,
1728 "Failed to initialize F2FS segment manager");
aff063e2 1729 goto free_sm;
a07ef784 1730 }
aff063e2 1731 err = build_node_manager(sbi);
a07ef784
NJ		 1732 if (err) {
1733 f2fs_msg(sb, KERN_ERR,
1734 "Failed to initialize F2FS node manager");
aff063e2 1735 goto free_nm;
a07ef784 1736 }
aff063e2 1737
8f1dbbbb
SL		 1738 /* For write statistics */
1739 if (sb->s_bdev->bd_part)
1740 sbi->sectors_written_start =
1741 (u64)part_stat_read(sb->s_bdev->bd_part, sectors[1]);
1742
1743 /* Read accumulated write IO statistics if exists */
1744 seg_i = CURSEG_I(sbi, CURSEG_HOT_NODE);
1745 if (__exist_node_summaries(sbi))
1746 sbi->kbytes_written =
b2dde6fc 1747 le64_to_cpu(seg_i->journal->info.kbytes_written);
8f1dbbbb 1748
aff063e2
JK		 1749 build_gc_manager(sbi);
1750
1751 /* get an inode for node space */
1752 sbi->node_inode = f2fs_iget(sb, F2FS_NODE_INO(sbi));
1753 if (IS_ERR(sbi->node_inode)) {
a07ef784 1754 f2fs_msg(sb, KERN_ERR, "Failed to read node inode");
aff063e2
JK		 1755 err = PTR_ERR(sbi->node_inode);
1756 goto free_nm;
1757 }
1758
2658e50d
JK		 1759 f2fs_join_shrinker(sbi);
1760
aff063e2 1761 /* if there are nt orphan nodes free them */
8c14bfad
CY		 1762 err = recover_orphan_inodes(sbi);
1763 if (err)
1764 goto free_node_inode;
aff063e2
JK		 1765
1766 /* read root inode and dentry */
1767 root = f2fs_iget(sb, F2FS_ROOT_INO(sbi));
1768 if (IS_ERR(root)) {
a07ef784 1769 f2fs_msg(sb, KERN_ERR, "Failed to read root inode");
aff063e2
JK		 1770 err = PTR_ERR(root);
1771 goto free_node_inode;
1772 }
8f99a946 1773 if (!S_ISDIR(root->i_mode) || !root->i_blocks || !root->i_size) {
9d847950 1774 iput(root);
8f99a946 1775 err = -EINVAL;
9d847950 1776 goto free_node_inode;
8f99a946 1777 }
aff063e2
JK		 1778
1779 sb->s_root = d_make_root(root); /* allocate root dentry */
1780 if (!sb->s_root) {
1781 err = -ENOMEM;
1782 goto free_root_inode;
1783 }
1784
aff063e2
JK		 1785 err = f2fs_build_stats(sbi);
1786 if (err)
6437d1b0 1787 goto free_root_inode;
aff063e2 1788
5e176d54
JK		 1789 if (f2fs_proc_root)
1790 sbi->s_proc = proc_mkdir(sb->s_id, f2fs_proc_root);
1791
f00d6fa7 1792 if (sbi->s_proc) {
5e176d54
JK		 1793 proc_create_data("segment_info", S_IRUGO, sbi->s_proc,
1794 &f2fs_seq_segment_info_fops, sb);
f00d6fa7
JK		 1795 proc_create_data("segment_bits", S_IRUGO, sbi->s_proc,
1796 &f2fs_seq_segment_bits_fops, sb);
1797 }
5e176d54 1798
b59d0bae
NJ		 1799 sbi->s_kobj.kset = f2fs_kset;
1800 init_completion(&sbi->s_kobj_unregister);
1801 err = kobject_init_and_add(&sbi->s_kobj, &f2fs_ktype, NULL,
1802 "%s", sb->s_id);
1803 if (err)
6437d1b0 1804 goto free_proc;
b59d0bae 1805
6437d1b0
JK		 1806 /* recover fsynced data */
1807 if (!test_opt(sbi, DISABLE_ROLL_FORWARD)) {
081d78c2
JK		 1808 /*
1809 * mount should be failed, when device has readonly mode, and
1810 * previous checkpoint was not done by clean system shutdown.
1811 */
1812 if (bdev_read_only(sb->s_bdev) &&
aaec2b1d 1813 !is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG)) {
081d78c2
JK		 1814 err = -EROFS;
1815 goto free_kobj;
1816 }
2adc3505
CY		 1817
1818 if (need_fsck)
1819 set_sbi_flag(sbi, SBI_NEED_FSCK);
1820
a468f0ef
JK		 1821 if (!retry)
1822 goto skip_recovery;
1823
6781eabb
JK		 1824 err = recover_fsync_data(sbi, false);
1825 if (err < 0) {
2adc3505 1826 need_fsck = true;
6437d1b0 1827 f2fs_msg(sb, KERN_ERR,
99e3e858 1828 "Cannot recover all fsync data errno=%d", err);
ed2e621a
JK		 1829 goto free_kobj;
1830 }
6781eabb
JK		 1831 } else {
1832 err = recover_fsync_data(sbi, true);
1833
1834 if (!f2fs_readonly(sb) && err > 0) {
1835 err = -EINVAL;
1836 f2fs_msg(sb, KERN_ERR,
1837 "Need to recover fsync data");
1838 goto free_kobj;
1839 }
6437d1b0 1840 }
a468f0ef 1841skip_recovery:
315df839
JK		 1842 /* recover_fsync_data() cleared this already */
1843 clear_sbi_flag(sbi, SBI_POR_DOING);
b59d0bae 1844
6437d1b0
JK		 1845 /*
1846 * If filesystem is not mounted as read-only then
1847 * do start the gc_thread.
1848 */
6c029932 1849 if (test_opt(sbi, BG_GC) && !f2fs_readonly(sb)) {
6437d1b0
JK		 1850 /* After POR, we can run background GC thread.*/
1851 err = start_gc_thread(sbi);
1852 if (err)
1853 goto free_kobj;
1854 }
dabc4a5c 1855 kfree(options);
da554e48 1856
1857 /* recover broken superblock */
f2353d7b 1858 if (recovery) {
41214b3c
CY		 1859 err = f2fs_commit_super(sbi, true);
1860 f2fs_msg(sb, KERN_INFO,
99e3e858 1861 "Try to recover %dth superblock, ret: %d",
41214b3c 1862 sbi->valid_super_block ? 1 : 2, err);
da554e48 1863 }
1864
6beceb54 1865 f2fs_update_time(sbi, CP_TIME);
d0239e1b 1866 f2fs_update_time(sbi, REQ_TIME);
aff063e2 1867 return 0;
6437d1b0
JK		 1868
1869free_kobj:
0f18b462 1870 f2fs_sync_inode_meta(sbi);
6437d1b0 1871 kobject_del(&sbi->s_kobj);
29ba108d
CY		 1872 kobject_put(&sbi->s_kobj);
1873 wait_for_completion(&sbi->s_kobj_unregister);
6437d1b0 1874free_proc:
1d15bd20
CY		 1875 if (sbi->s_proc) {
1876 remove_proc_entry("segment_info", sbi->s_proc);
f00d6fa7 1877 remove_proc_entry("segment_bits", sbi->s_proc);
1d15bd20
CY		 1878 remove_proc_entry(sb->s_id, f2fs_proc_root);
1879 }
1880 f2fs_destroy_stats(sbi);
aff063e2
JK		 1881free_root_inode:
1882 dput(sb->s_root);
1883 sb->s_root = NULL;
1884free_node_inode:
bb5dada7 1885 truncate_inode_pages_final(NODE_MAPPING(sbi));
2658e50d 1886 mutex_lock(&sbi->umount_mutex);
d41065e2 1887 release_ino_entry(sbi, true);
2658e50d 1888 f2fs_leave_shrinker(sbi);
aff063e2 1889 iput(sbi->node_inode);
2658e50d 1890 mutex_unlock(&sbi->umount_mutex);
aff063e2
JK		 1891free_nm:
1892 destroy_node_manager(sbi);
1893free_sm:
1894 destroy_segment_manager(sbi);
aff063e2
JK		 1895 kfree(sbi->ckpt);
1896free_meta_inode:
1897 make_bad_inode(sbi->meta_inode);
1898 iput(sbi->meta_inode);
dabc4a5c 1899free_options:
523be8a6 1900 destroy_percpu_info(sbi);
dabc4a5c 1901 kfree(options);
aff063e2 1902free_sb_buf:
b39f0de2 1903 kfree(raw_super);
aff063e2 1904free_sbi:
43b6573b
KM		 1905 if (sbi->s_chksum_driver)
1906 crypto_free_shash(sbi->s_chksum_driver);
aff063e2 1907 kfree(sbi);
ed2e621a
JK		 1908
1909 /* give only one another chance */
1910 if (retry) {
9df47ba7 1911 retry = false;
ed2e621a
JK		 1912 shrink_dcache_sb(sb);
1913 goto try_onemore;
1914 }
aff063e2
JK		 1915 return err;
1916}
1917
1918static struct dentry *f2fs_mount(struct file_system_type *fs_type, int flags,
1919 const char *dev_name, void *data)
1920{
1921 return mount_bdev(fs_type, flags, dev_name, data, f2fs_fill_super);
1922}
1923
30a5537f
JK		 1924static void kill_f2fs_super(struct super_block *sb)
1925{
1926 if (sb->s_root)
caf0047e 1927 set_sbi_flag(F2FS_SB(sb), SBI_IS_CLOSE);
30a5537f
JK		 1928 kill_block_super(sb);
1929}
1930
aff063e2
JK		 1931static struct file_system_type f2fs_fs_type = {
1932 .owner = THIS_MODULE,
1933 .name = "f2fs",
1934 .mount = f2fs_mount,
30a5537f 1935 .kill_sb = kill_f2fs_super,
aff063e2
JK		 1936 .fs_flags = FS_REQUIRES_DEV,
1937};
7f78e035 1938MODULE_ALIAS_FS("f2fs");
aff063e2 1939
6e6093a8 1940static int __init init_inodecache(void)
aff063e2 1941{
5d097056
VD		 1942 f2fs_inode_cachep = kmem_cache_create("f2fs_inode_cache",
1943 sizeof(struct f2fs_inode_info), 0,
1944 SLAB_RECLAIM_ACCOUNT|SLAB_ACCOUNT, NULL);
6bacf52f 1945 if (!f2fs_inode_cachep)
aff063e2
JK		 1946 return -ENOMEM;
1947 return 0;
1948}
1949
1950static void destroy_inodecache(void)
1951{
1952 /*
1953 * Make sure all delayed rcu free inodes are flushed before we
1954 * destroy cache.
1955 */
1956 rcu_barrier();
1957 kmem_cache_destroy(f2fs_inode_cachep);
1958}
1959
1960static int __init init_f2fs_fs(void)
1961{
1962 int err;
1963
c0508650
JK		 1964 f2fs_build_trace_ios();
1965
aff063e2
JK		 1966 err = init_inodecache();
1967 if (err)
1968 goto fail;
1969 err = create_node_manager_caches();
1970 if (err)
9890ff3f 1971 goto free_inodecache;
7fd9e544 1972 err = create_segment_manager_caches();
aff063e2 1973 if (err)
9890ff3f 1974 goto free_node_manager_caches;
aff063e2
JK		 1975 err = create_checkpoint_caches();
1976 if (err)
06292073 1977 goto free_segment_manager_caches;
1dcc336b
CY		 1978 err = create_extent_cache();
1979 if (err)
1980 goto free_checkpoint_caches;
b59d0bae 1981 f2fs_kset = kset_create_and_add("f2fs", NULL, fs_kobj);
6e6b978c
WY		 1982 if (!f2fs_kset) {
1983 err = -ENOMEM;
1dcc336b 1984 goto free_extent_cache;
6e6b978c 1985 }
2658e50d 1986 err = register_shrinker(&f2fs_shrinker_info);
cfc4d971 1987 if (err)
0b81d077 1988 goto free_kset;
2658e50d
JK		 1989
1990 err = register_filesystem(&f2fs_fs_type);
1991 if (err)
1992 goto free_shrinker;
787c7b8c
CY		 1993 err = f2fs_create_root_stats();
1994 if (err)
1995 goto free_filesystem;
5e176d54 1996 f2fs_proc_root = proc_mkdir("fs/f2fs", NULL);
9890ff3f
ZH		 1997 return 0;
1998
787c7b8c
CY		 1999free_filesystem:
2000 unregister_filesystem(&f2fs_fs_type);
2658e50d
JK		 2001free_shrinker:
2002 unregister_shrinker(&f2fs_shrinker_info);
9890ff3f
ZH		 2003free_kset:
2004 kset_unregister(f2fs_kset);
1dcc336b
CY		 2005free_extent_cache:
2006 destroy_extent_cache();
9890ff3f
ZH		 2007free_checkpoint_caches:
2008 destroy_checkpoint_caches();
7fd9e544
JK		 2009free_segment_manager_caches:
2010 destroy_segment_manager_caches();
9890ff3f
ZH		 2011free_node_manager_caches:
2012 destroy_node_manager_caches();
2013free_inodecache:
2014 destroy_inodecache();
aff063e2
JK		 2015fail:
2016 return err;
2017}
2018
2019static void __exit exit_f2fs_fs(void)
2020{
5e176d54 2021 remove_proc_entry("fs/f2fs", NULL);
4589d25d 2022 f2fs_destroy_root_stats();
aff063e2 2023 unregister_filesystem(&f2fs_fs_type);
b8bef79d 2024 unregister_shrinker(&f2fs_shrinker_info);
b8bef79d 2025 kset_unregister(f2fs_kset);
fdf6c8be 2026 destroy_extent_cache();
aff063e2 2027 destroy_checkpoint_caches();
5dcd8a71 2028 destroy_segment_manager_caches();
aff063e2
JK		 2029 destroy_node_manager_caches();
2030 destroy_inodecache();
351f4fba 2031 f2fs_destroy_trace_ios();
aff063e2
JK		 2032}
2033
2034module_init(init_f2fs_fs)
2035module_exit(exit_f2fs_fs)
2036
2037MODULE_AUTHOR("Samsung Electronics's Praesto Team");
2038MODULE_DESCRIPTION("Flash Friendly File System");
2039MODULE_LICENSE("GPL");
Linux 6.x block layer and io_uring tree(s)