f2fs: use generic EFSBADCRC/EFSCORRUPTED
[platform/kernel/linux-rpi.git] / fs / f2fs / gc.c
1 /*
2  * fs/f2fs/gc.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/fs.h>
12 #include <linux/module.h>
13 #include <linux/backing-dev.h>
14 #include <linux/init.h>
15 #include <linux/f2fs_fs.h>
16 #include <linux/kthread.h>
17 #include <linux/delay.h>
18 #include <linux/freezer.h>
19
20 #include "f2fs.h"
21 #include "node.h"
22 #include "segment.h"
23 #include "gc.h"
24 #include <trace/events/f2fs.h>
25
26 static int gc_thread_func(void *data)
27 {
28         struct f2fs_sb_info *sbi = data;
29         struct f2fs_gc_kthread *gc_th = sbi->gc_thread;
30         wait_queue_head_t *wq = &sbi->gc_thread->gc_wait_queue_head;
31         unsigned int wait_ms;
32
33         wait_ms = gc_th->min_sleep_time;
34
35         set_freezable();
36         do {
37                 wait_event_interruptible_timeout(*wq,
38                                 kthread_should_stop() || freezing(current) ||
39                                 gc_th->gc_wake,
40                                 msecs_to_jiffies(wait_ms));
41
42                 /* give it a try one time */
43                 if (gc_th->gc_wake)
44                         gc_th->gc_wake = 0;
45
46                 if (try_to_freeze())
47                         continue;
48                 if (kthread_should_stop())
49                         break;
50
51                 if (sbi->sb->s_writers.frozen >= SB_FREEZE_WRITE) {
52                         increase_sleep_time(gc_th, &wait_ms);
53                         continue;
54                 }
55
56                 if (time_to_inject(sbi, FAULT_CHECKPOINT)) {
57                         f2fs_show_injection_info(FAULT_CHECKPOINT);
58                         f2fs_stop_checkpoint(sbi, false);
59                 }
60
61                 if (!sb_start_write_trylock(sbi->sb))
62                         continue;
63
64                 /*
65                  * [GC triggering condition]
66                  * 0. GC is not conducted currently.
67                  * 1. There are enough dirty segments.
68                  * 2. IO subsystem is idle by checking the # of writeback pages.
69                  * 3. IO subsystem is idle by checking the # of requests in
70                  *    bdev's request list.
71                  *
72                  * Note) We have to avoid triggering GCs frequently.
73                  * Because it is possible that some segments can be
74                  * invalidated soon after by user update or deletion.
75                  * So, I'd like to wait some time to collect dirty segments.
76                  */
77                 if (sbi->gc_mode == GC_URGENT) {
78                         wait_ms = gc_th->urgent_sleep_time;
79                         mutex_lock(&sbi->gc_mutex);
80                         goto do_gc;
81                 }
82
83                 if (!mutex_trylock(&sbi->gc_mutex))
84                         goto next;
85
86                 if (!is_idle(sbi)) {
87                         increase_sleep_time(gc_th, &wait_ms);
88                         mutex_unlock(&sbi->gc_mutex);
89                         goto next;
90                 }
91
92                 if (has_enough_invalid_blocks(sbi))
93                         decrease_sleep_time(gc_th, &wait_ms);
94                 else
95                         increase_sleep_time(gc_th, &wait_ms);
96 do_gc:
97                 stat_inc_bggc_count(sbi);
98
99                 /* if return value is not zero, no victim was selected */
100                 if (f2fs_gc(sbi, test_opt(sbi, FORCE_FG_GC), true, NULL_SEGNO))
101                         wait_ms = gc_th->no_gc_sleep_time;
102
103                 trace_f2fs_background_gc(sbi->sb, wait_ms,
104                                 prefree_segments(sbi), free_segments(sbi));
105
106                 /* balancing f2fs's metadata periodically */
107                 f2fs_balance_fs_bg(sbi);
108 next:
109                 sb_end_write(sbi->sb);
110
111         } while (!kthread_should_stop());
112         return 0;
113 }
114
115 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi)
116 {
117         struct f2fs_gc_kthread *gc_th;
118         dev_t dev = sbi->sb->s_bdev->bd_dev;
119         int err = 0;
120
121         gc_th = f2fs_kmalloc(sbi, sizeof(struct f2fs_gc_kthread), GFP_KERNEL);
122         if (!gc_th) {
123                 err = -ENOMEM;
124                 goto out;
125         }
126
127         gc_th->urgent_sleep_time = DEF_GC_THREAD_URGENT_SLEEP_TIME;
128         gc_th->min_sleep_time = DEF_GC_THREAD_MIN_SLEEP_TIME;
129         gc_th->max_sleep_time = DEF_GC_THREAD_MAX_SLEEP_TIME;
130         gc_th->no_gc_sleep_time = DEF_GC_THREAD_NOGC_SLEEP_TIME;
131
132         gc_th->gc_wake= 0;
133
134         sbi->gc_thread = gc_th;
135         init_waitqueue_head(&sbi->gc_thread->gc_wait_queue_head);
136         sbi->gc_thread->f2fs_gc_task = kthread_run(gc_thread_func, sbi,
137                         "f2fs_gc-%u:%u", MAJOR(dev), MINOR(dev));
138         if (IS_ERR(gc_th->f2fs_gc_task)) {
139                 err = PTR_ERR(gc_th->f2fs_gc_task);
140                 kfree(gc_th);
141                 sbi->gc_thread = NULL;
142         }
143 out:
144         return err;
145 }
146
147 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi)
148 {
149         struct f2fs_gc_kthread *gc_th = sbi->gc_thread;
150         if (!gc_th)
151                 return;
152         kthread_stop(gc_th->f2fs_gc_task);
153         kfree(gc_th);
154         sbi->gc_thread = NULL;
155 }
156
157 static int select_gc_type(struct f2fs_sb_info *sbi, int gc_type)
158 {
159         int gc_mode = (gc_type == BG_GC) ? GC_CB : GC_GREEDY;
160
161         switch (sbi->gc_mode) {
162         case GC_IDLE_CB:
163                 gc_mode = GC_CB;
164                 break;
165         case GC_IDLE_GREEDY:
166         case GC_URGENT:
167                 gc_mode = GC_GREEDY;
168                 break;
169         }
170         return gc_mode;
171 }
172
173 static void select_policy(struct f2fs_sb_info *sbi, int gc_type,
174                         int type, struct victim_sel_policy *p)
175 {
176         struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
177
178         if (p->alloc_mode == SSR) {
179                 p->gc_mode = GC_GREEDY;
180                 p->dirty_segmap = dirty_i->dirty_segmap[type];
181                 p->max_search = dirty_i->nr_dirty[type];
182                 p->ofs_unit = 1;
183         } else {
184                 p->gc_mode = select_gc_type(sbi, gc_type);
185                 p->dirty_segmap = dirty_i->dirty_segmap[DIRTY];
186                 p->max_search = dirty_i->nr_dirty[DIRTY];
187                 p->ofs_unit = sbi->segs_per_sec;
188         }
189
190         /* we need to check every dirty segments in the FG_GC case */
191         if (gc_type != FG_GC &&
192                         (sbi->gc_mode != GC_URGENT) &&
193                         p->max_search > sbi->max_victim_search)
194                 p->max_search = sbi->max_victim_search;
195
196         /* let's select beginning hot/small space first in no_heap mode*/
197         if (test_opt(sbi, NOHEAP) &&
198                 (type == CURSEG_HOT_DATA || IS_NODESEG(type)))
199                 p->offset = 0;
200         else
201                 p->offset = SIT_I(sbi)->last_victim[p->gc_mode];
202 }
203
204 static unsigned int get_max_cost(struct f2fs_sb_info *sbi,
205                                 struct victim_sel_policy *p)
206 {
207         /* SSR allocates in a segment unit */
208         if (p->alloc_mode == SSR)
209                 return sbi->blocks_per_seg;
210         if (p->gc_mode == GC_GREEDY)
211                 return 2 * sbi->blocks_per_seg * p->ofs_unit;
212         else if (p->gc_mode == GC_CB)
213                 return UINT_MAX;
214         else /* No other gc_mode */
215                 return 0;
216 }
217
218 static unsigned int check_bg_victims(struct f2fs_sb_info *sbi)
219 {
220         struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
221         unsigned int secno;
222
223         /*
224          * If the gc_type is FG_GC, we can select victim segments
225          * selected by background GC before.
226          * Those segments guarantee they have small valid blocks.
227          */
228         for_each_set_bit(secno, dirty_i->victim_secmap, MAIN_SECS(sbi)) {
229                 if (sec_usage_check(sbi, secno))
230                         continue;
231                 clear_bit(secno, dirty_i->victim_secmap);
232                 return GET_SEG_FROM_SEC(sbi, secno);
233         }
234         return NULL_SEGNO;
235 }
236
237 static unsigned int get_cb_cost(struct f2fs_sb_info *sbi, unsigned int segno)
238 {
239         struct sit_info *sit_i = SIT_I(sbi);
240         unsigned int secno = GET_SEC_FROM_SEG(sbi, segno);
241         unsigned int start = GET_SEG_FROM_SEC(sbi, secno);
242         unsigned long long mtime = 0;
243         unsigned int vblocks;
244         unsigned char age = 0;
245         unsigned char u;
246         unsigned int i;
247
248         for (i = 0; i < sbi->segs_per_sec; i++)
249                 mtime += get_seg_entry(sbi, start + i)->mtime;
250         vblocks = get_valid_blocks(sbi, segno, true);
251
252         mtime = div_u64(mtime, sbi->segs_per_sec);
253         vblocks = div_u64(vblocks, sbi->segs_per_sec);
254
255         u = (vblocks * 100) >> sbi->log_blocks_per_seg;
256
257         /* Handle if the system time has changed by the user */
258         if (mtime < sit_i->min_mtime)
259                 sit_i->min_mtime = mtime;
260         if (mtime > sit_i->max_mtime)
261                 sit_i->max_mtime = mtime;
262         if (sit_i->max_mtime != sit_i->min_mtime)
263                 age = 100 - div64_u64(100 * (mtime - sit_i->min_mtime),
264                                 sit_i->max_mtime - sit_i->min_mtime);
265
266         return UINT_MAX - ((100 * (100 - u) * age) / (100 + u));
267 }
268
269 static inline unsigned int get_gc_cost(struct f2fs_sb_info *sbi,
270                         unsigned int segno, struct victim_sel_policy *p)
271 {
272         if (p->alloc_mode == SSR)
273                 return get_seg_entry(sbi, segno)->ckpt_valid_blocks;
274
275         /* alloc_mode == LFS */
276         if (p->gc_mode == GC_GREEDY)
277                 return get_valid_blocks(sbi, segno, true);
278         else
279                 return get_cb_cost(sbi, segno);
280 }
281
282 static unsigned int count_bits(const unsigned long *addr,
283                                 unsigned int offset, unsigned int len)
284 {
285         unsigned int end = offset + len, sum = 0;
286
287         while (offset < end) {
288                 if (test_bit(offset++, addr))
289                         ++sum;
290         }
291         return sum;
292 }
293
294 /*
295  * This function is called from two paths.
296  * One is garbage collection and the other is SSR segment selection.
297  * When it is called during GC, it just gets a victim segment
298  * and it does not remove it from dirty seglist.
299  * When it is called from SSR segment selection, it finds a segment
300  * which has minimum valid blocks and removes it from dirty seglist.
301  */
302 static int get_victim_by_default(struct f2fs_sb_info *sbi,
303                 unsigned int *result, int gc_type, int type, char alloc_mode)
304 {
305         struct dirty_seglist_info *dirty_i = DIRTY_I(sbi);
306         struct sit_info *sm = SIT_I(sbi);
307         struct victim_sel_policy p;
308         unsigned int secno, last_victim;
309         unsigned int last_segment = MAIN_SEGS(sbi);
310         unsigned int nsearched = 0;
311
312         mutex_lock(&dirty_i->seglist_lock);
313
314         p.alloc_mode = alloc_mode;
315         select_policy(sbi, gc_type, type, &p);
316
317         p.min_segno = NULL_SEGNO;
318         p.min_cost = get_max_cost(sbi, &p);
319
320         if (*result != NULL_SEGNO) {
321                 if (IS_DATASEG(get_seg_entry(sbi, *result)->type) &&
322                         get_valid_blocks(sbi, *result, false) &&
323                         !sec_usage_check(sbi, GET_SEC_FROM_SEG(sbi, *result)))
324                         p.min_segno = *result;
325                 goto out;
326         }
327
328         if (p.max_search == 0)
329                 goto out;
330
331         last_victim = sm->last_victim[p.gc_mode];
332         if (p.alloc_mode == LFS && gc_type == FG_GC) {
333                 p.min_segno = check_bg_victims(sbi);
334                 if (p.min_segno != NULL_SEGNO)
335                         goto got_it;
336         }
337
338         while (1) {
339                 unsigned long cost;
340                 unsigned int segno;
341
342                 segno = find_next_bit(p.dirty_segmap, last_segment, p.offset);
343                 if (segno >= last_segment) {
344                         if (sm->last_victim[p.gc_mode]) {
345                                 last_segment =
346                                         sm->last_victim[p.gc_mode];
347                                 sm->last_victim[p.gc_mode] = 0;
348                                 p.offset = 0;
349                                 continue;
350                         }
351                         break;
352                 }
353
354                 p.offset = segno + p.ofs_unit;
355                 if (p.ofs_unit > 1) {
356                         p.offset -= segno % p.ofs_unit;
357                         nsearched += count_bits(p.dirty_segmap,
358                                                 p.offset - p.ofs_unit,
359                                                 p.ofs_unit);
360                 } else {
361                         nsearched++;
362                 }
363
364                 secno = GET_SEC_FROM_SEG(sbi, segno);
365
366                 if (sec_usage_check(sbi, secno))
367                         goto next;
368                 if (gc_type == BG_GC && test_bit(secno, dirty_i->victim_secmap))
369                         goto next;
370
371                 cost = get_gc_cost(sbi, segno, &p);
372
373                 if (p.min_cost > cost) {
374                         p.min_segno = segno;
375                         p.min_cost = cost;
376                 }
377 next:
378                 if (nsearched >= p.max_search) {
379                         if (!sm->last_victim[p.gc_mode] && segno <= last_victim)
380                                 sm->last_victim[p.gc_mode] = last_victim + 1;
381                         else
382                                 sm->last_victim[p.gc_mode] = segno + 1;
383                         sm->last_victim[p.gc_mode] %= MAIN_SEGS(sbi);
384                         break;
385                 }
386         }
387         if (p.min_segno != NULL_SEGNO) {
388 got_it:
389                 if (p.alloc_mode == LFS) {
390                         secno = GET_SEC_FROM_SEG(sbi, p.min_segno);
391                         if (gc_type == FG_GC)
392                                 sbi->cur_victim_sec = secno;
393                         else
394                                 set_bit(secno, dirty_i->victim_secmap);
395                 }
396                 *result = (p.min_segno / p.ofs_unit) * p.ofs_unit;
397
398                 trace_f2fs_get_victim(sbi->sb, type, gc_type, &p,
399                                 sbi->cur_victim_sec,
400                                 prefree_segments(sbi), free_segments(sbi));
401         }
402 out:
403         mutex_unlock(&dirty_i->seglist_lock);
404
405         return (p.min_segno == NULL_SEGNO) ? 0 : 1;
406 }
407
408 static const struct victim_selection default_v_ops = {
409         .get_victim = get_victim_by_default,
410 };
411
412 static struct inode *find_gc_inode(struct gc_inode_list *gc_list, nid_t ino)
413 {
414         struct inode_entry *ie;
415
416         ie = radix_tree_lookup(&gc_list->iroot, ino);
417         if (ie)
418                 return ie->inode;
419         return NULL;
420 }
421
422 static void add_gc_inode(struct gc_inode_list *gc_list, struct inode *inode)
423 {
424         struct inode_entry *new_ie;
425
426         if (inode == find_gc_inode(gc_list, inode->i_ino)) {
427                 iput(inode);
428                 return;
429         }
430         new_ie = f2fs_kmem_cache_alloc(f2fs_inode_entry_slab, GFP_NOFS);
431         new_ie->inode = inode;
432
433         f2fs_radix_tree_insert(&gc_list->iroot, inode->i_ino, new_ie);
434         list_add_tail(&new_ie->list, &gc_list->ilist);
435 }
436
437 static void put_gc_inode(struct gc_inode_list *gc_list)
438 {
439         struct inode_entry *ie, *next_ie;
440         list_for_each_entry_safe(ie, next_ie, &gc_list->ilist, list) {
441                 radix_tree_delete(&gc_list->iroot, ie->inode->i_ino);
442                 iput(ie->inode);
443                 list_del(&ie->list);
444                 kmem_cache_free(f2fs_inode_entry_slab, ie);
445         }
446 }
447
448 static int check_valid_map(struct f2fs_sb_info *sbi,
449                                 unsigned int segno, int offset)
450 {
451         struct sit_info *sit_i = SIT_I(sbi);
452         struct seg_entry *sentry;
453         int ret;
454
455         down_read(&sit_i->sentry_lock);
456         sentry = get_seg_entry(sbi, segno);
457         ret = f2fs_test_bit(offset, sentry->cur_valid_map);
458         up_read(&sit_i->sentry_lock);
459         return ret;
460 }
461
462 /*
463  * This function compares node address got in summary with that in NAT.
464  * On validity, copy that node with cold status, otherwise (invalid node)
465  * ignore that.
466  */
467 static void gc_node_segment(struct f2fs_sb_info *sbi,
468                 struct f2fs_summary *sum, unsigned int segno, int gc_type)
469 {
470         struct f2fs_summary *entry;
471         block_t start_addr;
472         int off;
473         int phase = 0;
474         bool fggc = (gc_type == FG_GC);
475
476         start_addr = START_BLOCK(sbi, segno);
477
478 next_step:
479         entry = sum;
480
481         if (fggc && phase == 2)
482                 atomic_inc(&sbi->wb_sync_req[NODE]);
483
484         for (off = 0; off < sbi->blocks_per_seg; off++, entry++) {
485                 nid_t nid = le32_to_cpu(entry->nid);
486                 struct page *node_page;
487                 struct node_info ni;
488
489                 /* stop BG_GC if there is not enough free sections. */
490                 if (gc_type == BG_GC && has_not_enough_free_secs(sbi, 0, 0))
491                         return;
492
493                 if (check_valid_map(sbi, segno, off) == 0)
494                         continue;
495
496                 if (phase == 0) {
497                         f2fs_ra_meta_pages(sbi, NAT_BLOCK_OFFSET(nid), 1,
498                                                         META_NAT, true);
499                         continue;
500                 }
501
502                 if (phase == 1) {
503                         f2fs_ra_node_page(sbi, nid);
504                         continue;
505                 }
506
507                 /* phase == 2 */
508                 node_page = f2fs_get_node_page(sbi, nid);
509                 if (IS_ERR(node_page))
510                         continue;
511
512                 /* block may become invalid during f2fs_get_node_page */
513                 if (check_valid_map(sbi, segno, off) == 0) {
514                         f2fs_put_page(node_page, 1);
515                         continue;
516                 }
517
518                 if (f2fs_get_node_info(sbi, nid, &ni)) {
519                         f2fs_put_page(node_page, 1);
520                         continue;
521                 }
522
523                 if (ni.blk_addr != start_addr + off) {
524                         f2fs_put_page(node_page, 1);
525                         continue;
526                 }
527
528                 f2fs_move_node_page(node_page, gc_type);
529                 stat_inc_node_blk_count(sbi, 1, gc_type);
530         }
531
532         if (++phase < 3)
533                 goto next_step;
534
535         if (fggc)
536                 atomic_dec(&sbi->wb_sync_req[NODE]);
537 }
538
539 /*
540  * Calculate start block index indicating the given node offset.
541  * Be careful, caller should give this node offset only indicating direct node
542  * blocks. If any node offsets, which point the other types of node blocks such
543  * as indirect or double indirect node blocks, are given, it must be a caller's
544  * bug.
545  */
546 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode)
547 {
548         unsigned int indirect_blks = 2 * NIDS_PER_BLOCK + 4;
549         unsigned int bidx;
550
551         if (node_ofs == 0)
552                 return 0;
553
554         if (node_ofs <= 2) {
555                 bidx = node_ofs - 1;
556         } else if (node_ofs <= indirect_blks) {
557                 int dec = (node_ofs - 4) / (NIDS_PER_BLOCK + 1);
558                 bidx = node_ofs - 2 - dec;
559         } else {
560                 int dec = (node_ofs - indirect_blks - 3) / (NIDS_PER_BLOCK + 1);
561                 bidx = node_ofs - 5 - dec;
562         }
563         return bidx * ADDRS_PER_BLOCK + ADDRS_PER_INODE(inode);
564 }
565
566 static bool is_alive(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
567                 struct node_info *dni, block_t blkaddr, unsigned int *nofs)
568 {
569         struct page *node_page;
570         nid_t nid;
571         unsigned int ofs_in_node;
572         block_t source_blkaddr;
573
574         nid = le32_to_cpu(sum->nid);
575         ofs_in_node = le16_to_cpu(sum->ofs_in_node);
576
577         node_page = f2fs_get_node_page(sbi, nid);
578         if (IS_ERR(node_page))
579                 return false;
580
581         if (f2fs_get_node_info(sbi, nid, dni)) {
582                 f2fs_put_page(node_page, 1);
583                 return false;
584         }
585
586         if (sum->version != dni->version) {
587                 f2fs_msg(sbi->sb, KERN_WARNING,
588                                 "%s: valid data with mismatched node version.",
589                                 __func__);
590                 set_sbi_flag(sbi, SBI_NEED_FSCK);
591         }
592
593         *nofs = ofs_of_node(node_page);
594         source_blkaddr = datablock_addr(NULL, node_page, ofs_in_node);
595         f2fs_put_page(node_page, 1);
596
597         if (source_blkaddr != blkaddr)
598                 return false;
599         return true;
600 }
601
602 static int ra_data_block(struct inode *inode, pgoff_t index)
603 {
604         struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
605         struct address_space *mapping = inode->i_mapping;
606         struct dnode_of_data dn;
607         struct page *page;
608         struct extent_info ei = {0, 0, 0};
609         struct f2fs_io_info fio = {
610                 .sbi = sbi,
611                 .ino = inode->i_ino,
612                 .type = DATA,
613                 .temp = COLD,
614                 .op = REQ_OP_READ,
615                 .op_flags = 0,
616                 .encrypted_page = NULL,
617                 .in_list = false,
618                 .retry = false,
619         };
620         int err;
621
622         page = f2fs_grab_cache_page(mapping, index, true);
623         if (!page)
624                 return -ENOMEM;
625
626         if (f2fs_lookup_extent_cache(inode, index, &ei)) {
627                 dn.data_blkaddr = ei.blk + index - ei.fofs;
628                 goto got_it;
629         }
630
631         set_new_dnode(&dn, inode, NULL, NULL, 0);
632         err = f2fs_get_dnode_of_data(&dn, index, LOOKUP_NODE);
633         if (err)
634                 goto put_page;
635         f2fs_put_dnode(&dn);
636
637         if (unlikely(!f2fs_is_valid_blkaddr(sbi, dn.data_blkaddr,
638                                                 DATA_GENERIC))) {
639                 err = -EFSCORRUPTED;
640                 goto put_page;
641         }
642 got_it:
643         /* read page */
644         fio.page = page;
645         fio.new_blkaddr = fio.old_blkaddr = dn.data_blkaddr;
646
647         fio.encrypted_page = f2fs_pagecache_get_page(META_MAPPING(sbi),
648                                         dn.data_blkaddr,
649                                         FGP_LOCK | FGP_CREAT, GFP_NOFS);
650         if (!fio.encrypted_page) {
651                 err = -ENOMEM;
652                 goto put_page;
653         }
654
655         err = f2fs_submit_page_bio(&fio);
656         if (err)
657                 goto put_encrypted_page;
658         f2fs_put_page(fio.encrypted_page, 0);
659         f2fs_put_page(page, 1);
660         return 0;
661 put_encrypted_page:
662         f2fs_put_page(fio.encrypted_page, 1);
663 put_page:
664         f2fs_put_page(page, 1);
665         return err;
666 }
667
668 /*
669  * Move data block via META_MAPPING while keeping locked data page.
670  * This can be used to move blocks, aka LBAs, directly on disk.
671  */
672 static void move_data_block(struct inode *inode, block_t bidx,
673                                 int gc_type, unsigned int segno, int off)
674 {
675         struct f2fs_io_info fio = {
676                 .sbi = F2FS_I_SB(inode),
677                 .ino = inode->i_ino,
678                 .type = DATA,
679                 .temp = COLD,
680                 .op = REQ_OP_READ,
681                 .op_flags = 0,
682                 .encrypted_page = NULL,
683                 .in_list = false,
684                 .retry = false,
685         };
686         struct dnode_of_data dn;
687         struct f2fs_summary sum;
688         struct node_info ni;
689         struct page *page, *mpage;
690         block_t newaddr;
691         int err;
692         bool lfs_mode = test_opt(fio.sbi, LFS);
693
694         /* do not read out */
695         page = f2fs_grab_cache_page(inode->i_mapping, bidx, false);
696         if (!page)
697                 return;
698
699         if (!check_valid_map(F2FS_I_SB(inode), segno, off))
700                 goto out;
701
702         if (f2fs_is_atomic_file(inode)) {
703                 F2FS_I(inode)->i_gc_failures[GC_FAILURE_ATOMIC]++;
704                 F2FS_I_SB(inode)->skipped_atomic_files[gc_type]++;
705                 goto out;
706         }
707
708         if (f2fs_is_pinned_file(inode)) {
709                 f2fs_pin_file_control(inode, true);
710                 goto out;
711         }
712
713         set_new_dnode(&dn, inode, NULL, NULL, 0);
714         err = f2fs_get_dnode_of_data(&dn, bidx, LOOKUP_NODE);
715         if (err)
716                 goto out;
717
718         if (unlikely(dn.data_blkaddr == NULL_ADDR)) {
719                 ClearPageUptodate(page);
720                 goto put_out;
721         }
722
723         /*
724          * don't cache encrypted data into meta inode until previous dirty
725          * data were writebacked to avoid racing between GC and flush.
726          */
727         f2fs_wait_on_page_writeback(page, DATA, true);
728
729         err = f2fs_get_node_info(fio.sbi, dn.nid, &ni);
730         if (err)
731                 goto put_out;
732
733         set_summary(&sum, dn.nid, dn.ofs_in_node, ni.version);
734
735         /* read page */
736         fio.page = page;
737         fio.new_blkaddr = fio.old_blkaddr = dn.data_blkaddr;
738
739         if (lfs_mode)
740                 down_write(&fio.sbi->io_order_lock);
741
742         f2fs_allocate_data_block(fio.sbi, NULL, fio.old_blkaddr, &newaddr,
743                                         &sum, CURSEG_COLD_DATA, NULL, false);
744
745         fio.encrypted_page = f2fs_pagecache_get_page(META_MAPPING(fio.sbi),
746                                 newaddr, FGP_LOCK | FGP_CREAT, GFP_NOFS);
747         if (!fio.encrypted_page) {
748                 err = -ENOMEM;
749                 goto recover_block;
750         }
751
752         mpage = f2fs_pagecache_get_page(META_MAPPING(fio.sbi),
753                                         fio.old_blkaddr, FGP_LOCK, GFP_NOFS);
754         if (mpage) {
755                 bool updated = false;
756
757                 if (PageUptodate(mpage)) {
758                         memcpy(page_address(fio.encrypted_page),
759                                         page_address(mpage), PAGE_SIZE);
760                         updated = true;
761                 }
762                 f2fs_put_page(mpage, 1);
763                 invalidate_mapping_pages(META_MAPPING(fio.sbi),
764                                         fio.old_blkaddr, fio.old_blkaddr);
765                 if (updated)
766                         goto write_page;
767         }
768
769         err = f2fs_submit_page_bio(&fio);
770         if (err)
771                 goto put_page_out;
772
773         /* write page */
774         lock_page(fio.encrypted_page);
775
776         if (unlikely(fio.encrypted_page->mapping != META_MAPPING(fio.sbi))) {
777                 err = -EIO;
778                 goto put_page_out;
779         }
780         if (unlikely(!PageUptodate(fio.encrypted_page))) {
781                 err = -EIO;
782                 goto put_page_out;
783         }
784
785 write_page:
786         set_page_dirty(fio.encrypted_page);
787         f2fs_wait_on_page_writeback(fio.encrypted_page, DATA, true);
788         if (clear_page_dirty_for_io(fio.encrypted_page))
789                 dec_page_count(fio.sbi, F2FS_DIRTY_META);
790
791         set_page_writeback(fio.encrypted_page);
792         ClearPageError(page);
793
794         /* allocate block address */
795         f2fs_wait_on_page_writeback(dn.node_page, NODE, true);
796
797         fio.op = REQ_OP_WRITE;
798         fio.op_flags = REQ_SYNC;
799         fio.new_blkaddr = newaddr;
800         f2fs_submit_page_write(&fio);
801         if (fio.retry) {
802                 if (PageWriteback(fio.encrypted_page))
803                         end_page_writeback(fio.encrypted_page);
804                 goto put_page_out;
805         }
806
807         f2fs_update_iostat(fio.sbi, FS_GC_DATA_IO, F2FS_BLKSIZE);
808
809         f2fs_update_data_blkaddr(&dn, newaddr);
810         set_inode_flag(inode, FI_APPEND_WRITE);
811         if (page->index == 0)
812                 set_inode_flag(inode, FI_FIRST_BLOCK_WRITTEN);
813 put_page_out:
814         f2fs_put_page(fio.encrypted_page, 1);
815 recover_block:
816         if (lfs_mode)
817                 up_write(&fio.sbi->io_order_lock);
818         if (err)
819                 f2fs_do_replace_block(fio.sbi, &sum, newaddr, fio.old_blkaddr,
820                                                                 true, true);
821 put_out:
822         f2fs_put_dnode(&dn);
823 out:
824         f2fs_put_page(page, 1);
825 }
826
827 static void move_data_page(struct inode *inode, block_t bidx, int gc_type,
828                                                         unsigned int segno, int off)
829 {
830         struct page *page;
831
832         page = f2fs_get_lock_data_page(inode, bidx, true);
833         if (IS_ERR(page))
834                 return;
835
836         if (!check_valid_map(F2FS_I_SB(inode), segno, off))
837                 goto out;
838
839         if (f2fs_is_atomic_file(inode)) {
840                 F2FS_I(inode)->i_gc_failures[GC_FAILURE_ATOMIC]++;
841                 F2FS_I_SB(inode)->skipped_atomic_files[gc_type]++;
842                 goto out;
843         }
844         if (f2fs_is_pinned_file(inode)) {
845                 if (gc_type == FG_GC)
846                         f2fs_pin_file_control(inode, true);
847                 goto out;
848         }
849
850         if (gc_type == BG_GC) {
851                 if (PageWriteback(page))
852                         goto out;
853                 set_page_dirty(page);
854                 set_cold_data(page);
855         } else {
856                 struct f2fs_io_info fio = {
857                         .sbi = F2FS_I_SB(inode),
858                         .ino = inode->i_ino,
859                         .type = DATA,
860                         .temp = COLD,
861                         .op = REQ_OP_WRITE,
862                         .op_flags = REQ_SYNC,
863                         .old_blkaddr = NULL_ADDR,
864                         .page = page,
865                         .encrypted_page = NULL,
866                         .need_lock = LOCK_REQ,
867                         .io_type = FS_GC_DATA_IO,
868                 };
869                 bool is_dirty = PageDirty(page);
870                 int err;
871
872 retry:
873                 set_page_dirty(page);
874                 f2fs_wait_on_page_writeback(page, DATA, true);
875                 if (clear_page_dirty_for_io(page)) {
876                         inode_dec_dirty_pages(inode);
877                         f2fs_remove_dirty_inode(inode);
878                 }
879
880                 set_cold_data(page);
881
882                 err = f2fs_do_write_data_page(&fio);
883                 if (err) {
884                         clear_cold_data(page);
885                         if (err == -ENOMEM) {
886                                 congestion_wait(BLK_RW_ASYNC, HZ/50);
887                                 goto retry;
888                         }
889                         if (is_dirty)
890                                 set_page_dirty(page);
891                 }
892         }
893 out:
894         f2fs_put_page(page, 1);
895 }
896
897 /*
898  * This function tries to get parent node of victim data block, and identifies
899  * data block validity. If the block is valid, copy that with cold status and
900  * modify parent node.
901  * If the parent node is not valid or the data block address is different,
902  * the victim data block is ignored.
903  */
904 static void gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
905                 struct gc_inode_list *gc_list, unsigned int segno, int gc_type)
906 {
907         struct super_block *sb = sbi->sb;
908         struct f2fs_summary *entry;
909         block_t start_addr;
910         int off;
911         int phase = 0;
912
913         start_addr = START_BLOCK(sbi, segno);
914
915 next_step:
916         entry = sum;
917
918         for (off = 0; off < sbi->blocks_per_seg; off++, entry++) {
919                 struct page *data_page;
920                 struct inode *inode;
921                 struct node_info dni; /* dnode info for the data */
922                 unsigned int ofs_in_node, nofs;
923                 block_t start_bidx;
924                 nid_t nid = le32_to_cpu(entry->nid);
925
926                 /* stop BG_GC if there is not enough free sections. */
927                 if (gc_type == BG_GC && has_not_enough_free_secs(sbi, 0, 0))
928                         return;
929
930                 if (check_valid_map(sbi, segno, off) == 0)
931                         continue;
932
933                 if (phase == 0) {
934                         f2fs_ra_meta_pages(sbi, NAT_BLOCK_OFFSET(nid), 1,
935                                                         META_NAT, true);
936                         continue;
937                 }
938
939                 if (phase == 1) {
940                         f2fs_ra_node_page(sbi, nid);
941                         continue;
942                 }
943
944                 /* Get an inode by ino with checking validity */
945                 if (!is_alive(sbi, entry, &dni, start_addr + off, &nofs))
946                         continue;
947
948                 if (phase == 2) {
949                         f2fs_ra_node_page(sbi, dni.ino);
950                         continue;
951                 }
952
953                 ofs_in_node = le16_to_cpu(entry->ofs_in_node);
954
955                 if (phase == 3) {
956                         inode = f2fs_iget(sb, dni.ino);
957                         if (IS_ERR(inode) || is_bad_inode(inode))
958                                 continue;
959
960                         if (!down_write_trylock(
961                                 &F2FS_I(inode)->i_gc_rwsem[WRITE])) {
962                                 iput(inode);
963                                 sbi->skipped_gc_rwsem++;
964                                 continue;
965                         }
966
967                         start_bidx = f2fs_start_bidx_of_node(nofs, inode) +
968                                                                 ofs_in_node;
969
970                         if (f2fs_post_read_required(inode)) {
971                                 int err = ra_data_block(inode, start_bidx);
972
973                                 up_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
974                                 if (err) {
975                                         iput(inode);
976                                         continue;
977                                 }
978                                 add_gc_inode(gc_list, inode);
979                                 continue;
980                         }
981
982                         data_page = f2fs_get_read_data_page(inode,
983                                                 start_bidx, REQ_RAHEAD, true);
984                         up_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
985                         if (IS_ERR(data_page)) {
986                                 iput(inode);
987                                 continue;
988                         }
989
990                         f2fs_put_page(data_page, 0);
991                         add_gc_inode(gc_list, inode);
992                         continue;
993                 }
994
995                 /* phase 4 */
996                 inode = find_gc_inode(gc_list, dni.ino);
997                 if (inode) {
998                         struct f2fs_inode_info *fi = F2FS_I(inode);
999                         bool locked = false;
1000
1001                         if (S_ISREG(inode->i_mode)) {
1002                                 if (!down_write_trylock(&fi->i_gc_rwsem[READ]))
1003                                         continue;
1004                                 if (!down_write_trylock(
1005                                                 &fi->i_gc_rwsem[WRITE])) {
1006                                         sbi->skipped_gc_rwsem++;
1007                                         up_write(&fi->i_gc_rwsem[READ]);
1008                                         continue;
1009                                 }
1010                                 locked = true;
1011
1012                                 /* wait for all inflight aio data */
1013                                 inode_dio_wait(inode);
1014                         }
1015
1016                         start_bidx = f2fs_start_bidx_of_node(nofs, inode)
1017                                                                 + ofs_in_node;
1018                         if (f2fs_post_read_required(inode))
1019                                 move_data_block(inode, start_bidx, gc_type,
1020                                                                 segno, off);
1021                         else
1022                                 move_data_page(inode, start_bidx, gc_type,
1023                                                                 segno, off);
1024
1025                         if (locked) {
1026                                 up_write(&fi->i_gc_rwsem[WRITE]);
1027                                 up_write(&fi->i_gc_rwsem[READ]);
1028                         }
1029
1030                         stat_inc_data_blk_count(sbi, 1, gc_type);
1031                 }
1032         }
1033
1034         if (++phase < 5)
1035                 goto next_step;
1036 }
1037
1038 static int __get_victim(struct f2fs_sb_info *sbi, unsigned int *victim,
1039                         int gc_type)
1040 {
1041         struct sit_info *sit_i = SIT_I(sbi);
1042         int ret;
1043
1044         down_write(&sit_i->sentry_lock);
1045         ret = DIRTY_I(sbi)->v_ops->get_victim(sbi, victim, gc_type,
1046                                               NO_CHECK_TYPE, LFS);
1047         up_write(&sit_i->sentry_lock);
1048         return ret;
1049 }
1050
1051 static int do_garbage_collect(struct f2fs_sb_info *sbi,
1052                                 unsigned int start_segno,
1053                                 struct gc_inode_list *gc_list, int gc_type)
1054 {
1055         struct page *sum_page;
1056         struct f2fs_summary_block *sum;
1057         struct blk_plug plug;
1058         unsigned int segno = start_segno;
1059         unsigned int end_segno = start_segno + sbi->segs_per_sec;
1060         int seg_freed = 0;
1061         unsigned char type = IS_DATASEG(get_seg_entry(sbi, segno)->type) ?
1062                                                 SUM_TYPE_DATA : SUM_TYPE_NODE;
1063
1064         /* readahead multi ssa blocks those have contiguous address */
1065         if (sbi->segs_per_sec > 1)
1066                 f2fs_ra_meta_pages(sbi, GET_SUM_BLOCK(sbi, segno),
1067                                         sbi->segs_per_sec, META_SSA, true);
1068
1069         /* reference all summary page */
1070         while (segno < end_segno) {
1071                 sum_page = f2fs_get_sum_page(sbi, segno++);
1072                 unlock_page(sum_page);
1073         }
1074
1075         blk_start_plug(&plug);
1076
1077         for (segno = start_segno; segno < end_segno; segno++) {
1078
1079                 /* find segment summary of victim */
1080                 sum_page = find_get_page(META_MAPPING(sbi),
1081                                         GET_SUM_BLOCK(sbi, segno));
1082                 f2fs_put_page(sum_page, 0);
1083
1084                 if (get_valid_blocks(sbi, segno, false) == 0 ||
1085                                 !PageUptodate(sum_page) ||
1086                                 unlikely(f2fs_cp_error(sbi)))
1087                         goto next;
1088
1089                 sum = page_address(sum_page);
1090                 if (type != GET_SUM_TYPE((&sum->footer))) {
1091                         f2fs_msg(sbi->sb, KERN_ERR, "Inconsistent segment (%u) "
1092                                 "type [%d, %d] in SSA and SIT",
1093                                 segno, type, GET_SUM_TYPE((&sum->footer)));
1094                         set_sbi_flag(sbi, SBI_NEED_FSCK);
1095                         goto next;
1096                 }
1097
1098                 /*
1099                  * this is to avoid deadlock:
1100                  * - lock_page(sum_page)         - f2fs_replace_block
1101                  *  - check_valid_map()            - down_write(sentry_lock)
1102                  *   - down_read(sentry_lock)     - change_curseg()
1103                  *                                  - lock_page(sum_page)
1104                  */
1105                 if (type == SUM_TYPE_NODE)
1106                         gc_node_segment(sbi, sum->entries, segno, gc_type);
1107                 else
1108                         gc_data_segment(sbi, sum->entries, gc_list, segno,
1109                                                                 gc_type);
1110
1111                 stat_inc_seg_count(sbi, type, gc_type);
1112
1113                 if (gc_type == FG_GC &&
1114                                 get_valid_blocks(sbi, segno, false) == 0)
1115                         seg_freed++;
1116 next:
1117                 f2fs_put_page(sum_page, 0);
1118         }
1119
1120         if (gc_type == FG_GC)
1121                 f2fs_submit_merged_write(sbi,
1122                                 (type == SUM_TYPE_NODE) ? NODE : DATA);
1123
1124         blk_finish_plug(&plug);
1125
1126         stat_inc_call_count(sbi->stat_info);
1127
1128         return seg_freed;
1129 }
1130
1131 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync,
1132                         bool background, unsigned int segno)
1133 {
1134         int gc_type = sync ? FG_GC : BG_GC;
1135         int sec_freed = 0, seg_freed = 0, total_freed = 0;
1136         int ret = 0;
1137         struct cp_control cpc;
1138         unsigned int init_segno = segno;
1139         struct gc_inode_list gc_list = {
1140                 .ilist = LIST_HEAD_INIT(gc_list.ilist),
1141                 .iroot = RADIX_TREE_INIT(gc_list.iroot, GFP_NOFS),
1142         };
1143         unsigned long long last_skipped = sbi->skipped_atomic_files[FG_GC];
1144         unsigned long long first_skipped;
1145         unsigned int skipped_round = 0, round = 0;
1146
1147         trace_f2fs_gc_begin(sbi->sb, sync, background,
1148                                 get_pages(sbi, F2FS_DIRTY_NODES),
1149                                 get_pages(sbi, F2FS_DIRTY_DENTS),
1150                                 get_pages(sbi, F2FS_DIRTY_IMETA),
1151                                 free_sections(sbi),
1152                                 free_segments(sbi),
1153                                 reserved_segments(sbi),
1154                                 prefree_segments(sbi));
1155
1156         cpc.reason = __get_cp_reason(sbi);
1157         sbi->skipped_gc_rwsem = 0;
1158         first_skipped = last_skipped;
1159 gc_more:
1160         if (unlikely(!(sbi->sb->s_flags & SB_ACTIVE))) {
1161                 ret = -EINVAL;
1162                 goto stop;
1163         }
1164         if (unlikely(f2fs_cp_error(sbi))) {
1165                 ret = -EIO;
1166                 goto stop;
1167         }
1168
1169         if (gc_type == BG_GC && has_not_enough_free_secs(sbi, 0, 0)) {
1170                 /*
1171                  * For example, if there are many prefree_segments below given
1172                  * threshold, we can make them free by checkpoint. Then, we
1173                  * secure free segments which doesn't need fggc any more.
1174                  */
1175                 if (prefree_segments(sbi)) {
1176                         ret = f2fs_write_checkpoint(sbi, &cpc);
1177                         if (ret)
1178                                 goto stop;
1179                 }
1180                 if (has_not_enough_free_secs(sbi, 0, 0))
1181                         gc_type = FG_GC;
1182         }
1183
1184         /* f2fs_balance_fs doesn't need to do BG_GC in critical path. */
1185         if (gc_type == BG_GC && !background) {
1186                 ret = -EINVAL;
1187                 goto stop;
1188         }
1189         if (!__get_victim(sbi, &segno, gc_type)) {
1190                 ret = -ENODATA;
1191                 goto stop;
1192         }
1193
1194         seg_freed = do_garbage_collect(sbi, segno, &gc_list, gc_type);
1195         if (gc_type == FG_GC && seg_freed == sbi->segs_per_sec)
1196                 sec_freed++;
1197         total_freed += seg_freed;
1198
1199         if (gc_type == FG_GC) {
1200                 if (sbi->skipped_atomic_files[FG_GC] > last_skipped ||
1201                                                 sbi->skipped_gc_rwsem)
1202                         skipped_round++;
1203                 last_skipped = sbi->skipped_atomic_files[FG_GC];
1204                 round++;
1205         }
1206
1207         if (gc_type == FG_GC)
1208                 sbi->cur_victim_sec = NULL_SEGNO;
1209
1210         if (sync)
1211                 goto stop;
1212
1213         if (has_not_enough_free_secs(sbi, sec_freed, 0)) {
1214                 if (skipped_round <= MAX_SKIP_GC_COUNT ||
1215                                         skipped_round * 2 < round) {
1216                         segno = NULL_SEGNO;
1217                         goto gc_more;
1218                 }
1219
1220                 if (first_skipped < last_skipped &&
1221                                 (last_skipped - first_skipped) >
1222                                                 sbi->skipped_gc_rwsem) {
1223                         f2fs_drop_inmem_pages_all(sbi, true);
1224                         segno = NULL_SEGNO;
1225                         goto gc_more;
1226                 }
1227                 if (gc_type == FG_GC)
1228                         ret = f2fs_write_checkpoint(sbi, &cpc);
1229         }
1230 stop:
1231         SIT_I(sbi)->last_victim[ALLOC_NEXT] = 0;
1232         SIT_I(sbi)->last_victim[FLUSH_DEVICE] = init_segno;
1233
1234         trace_f2fs_gc_end(sbi->sb, ret, total_freed, sec_freed,
1235                                 get_pages(sbi, F2FS_DIRTY_NODES),
1236                                 get_pages(sbi, F2FS_DIRTY_DENTS),
1237                                 get_pages(sbi, F2FS_DIRTY_IMETA),
1238                                 free_sections(sbi),
1239                                 free_segments(sbi),
1240                                 reserved_segments(sbi),
1241                                 prefree_segments(sbi));
1242
1243         mutex_unlock(&sbi->gc_mutex);
1244
1245         put_gc_inode(&gc_list);
1246
1247         if (sync)
1248                 ret = sec_freed ? 0 : -EAGAIN;
1249         return ret;
1250 }
1251
1252 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi)
1253 {
1254         DIRTY_I(sbi)->v_ops = &default_v_ops;
1255
1256         sbi->gc_pin_file_threshold = DEF_GC_FAILED_PINNED_FILES;
1257
1258         /* give warm/cold data area from slower device */
1259         if (f2fs_is_multi_device(sbi) && sbi->segs_per_sec == 1)
1260                 SIT_I(sbi)->last_victim[ALLOC_NEXT] =
1261                                 GET_SEGNO(sbi, FDEV(0).end_blk) + 1;
1262 }