?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
dm snapshot: fix metadata corruption
[linux/fpc-iii.git] / fs / jffs2 / gc.c
blob5a2dec2b064c945aba23acac93dcee2820019d68
1 /*
2 * JFFS2 -- Journalling Flash File System, Version 2.
3 *
4 * Copyright © 2001-2007 Red Hat, Inc.
5 * Copyright © 2004-2010 David Woodhouse <dwmw2@infradead.org>
6 *
7 * Created by David Woodhouse <dwmw2@infradead.org>
8 *
9 * For licensing information, see the file 'LICENCE' in this directory.
10 *
11 */
12
13 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
14
15 #include <linux/kernel.h>
16 #include <linux/mtd/mtd.h>
17 #include <linux/slab.h>
18 #include <linux/pagemap.h>
19 #include <linux/crc32.h>
20 #include <linux/compiler.h>
21 #include <linux/stat.h>
22 #include "nodelist.h"
23 #include "compr.h"
24
25 static int jffs2_garbage_collect_pristine(struct jffs2_sb_info *c,
26 struct jffs2_inode_cache *ic,
27 struct jffs2_raw_node_ref *raw);
28 static int jffs2_garbage_collect_metadata(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb,
29 struct jffs2_inode_info *f, struct jffs2_full_dnode *fd);
30 static int jffs2_garbage_collect_dirent(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb,
31 struct jffs2_inode_info *f, struct jffs2_full_dirent *fd);
32 static int jffs2_garbage_collect_deletion_dirent(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb,
33 struct jffs2_inode_info *f, struct jffs2_full_dirent *fd);
34 static int jffs2_garbage_collect_hole(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb,
35 struct jffs2_inode_info *f, struct jffs2_full_dnode *fn,
36 uint32_t start, uint32_t end);
37 static int jffs2_garbage_collect_dnode(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb,
38 struct jffs2_inode_info *f, struct jffs2_full_dnode *fn,
39 uint32_t start, uint32_t end);
40 static int jffs2_garbage_collect_live(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb,
41 struct jffs2_raw_node_ref *raw, struct jffs2_inode_info *f);
42
43 /* Called with erase_completion_lock held */
44 static struct jffs2_eraseblock *jffs2_find_gc_block(struct jffs2_sb_info *c)
45 {
46 struct jffs2_eraseblock *ret;
47 struct list_head *nextlist = NULL;
48 int n = jiffies % 128;
49
50 /* Pick an eraseblock to garbage collect next. This is where we'll
51 put the clever wear-levelling algorithms. Eventually. */
52 /* We possibly want to favour the dirtier blocks more when the
53 number of free blocks is low. */
54 again:
55 if (!list_empty(&c->bad_used_list) && c->nr_free_blocks > c->resv_blocks_gcbad) {
56 jffs2_dbg(1, "Picking block from bad_used_list to GC next\n");
57 nextlist = &c->bad_used_list;
58 } else if (n < 50 && !list_empty(&c->erasable_list)) {
59 /* Note that most of them will have gone directly to be erased.
60 So don't favour the erasable_list _too_ much. */
61 jffs2_dbg(1, "Picking block from erasable_list to GC next\n");
62 nextlist = &c->erasable_list;
63 } else if (n < 110 && !list_empty(&c->very_dirty_list)) {
64 /* Most of the time, pick one off the very_dirty list */
65 jffs2_dbg(1, "Picking block from very_dirty_list to GC next\n");
66 nextlist = &c->very_dirty_list;
67 } else if (n < 126 && !list_empty(&c->dirty_list)) {
68 jffs2_dbg(1, "Picking block from dirty_list to GC next\n");
69 nextlist = &c->dirty_list;
70 } else if (!list_empty(&c->clean_list)) {
71 jffs2_dbg(1, "Picking block from clean_list to GC next\n");
72 nextlist = &c->clean_list;
73 } else if (!list_empty(&c->dirty_list)) {
74 jffs2_dbg(1, "Picking block from dirty_list to GC next (clean_list was empty)\n");
75
76 nextlist = &c->dirty_list;
77 } else if (!list_empty(&c->very_dirty_list)) {
78 jffs2_dbg(1, "Picking block from very_dirty_list to GC next (clean_list and dirty_list were empty)\n");
79 nextlist = &c->very_dirty_list;
80 } else if (!list_empty(&c->erasable_list)) {
81 jffs2_dbg(1, "Picking block from erasable_list to GC next (clean_list and {very_,}dirty_list were empty)\n");
82
83 nextlist = &c->erasable_list;
84 } else if (!list_empty(&c->erasable_pending_wbuf_list)) {
85 /* There are blocks are wating for the wbuf sync */
86 jffs2_dbg(1, "Synching wbuf in order to reuse erasable_pending_wbuf_list blocks\n");
87 spin_unlock(&c->erase_completion_lock);
88 jffs2_flush_wbuf_pad(c);
89 spin_lock(&c->erase_completion_lock);
90 goto again;
91 } else {
92 /* Eep. All were empty */
93 jffs2_dbg(1, "No clean, dirty _or_ erasable blocks to GC from! Where are they all?\n");
94 return NULL;
95 }
96
97 ret = list_entry(nextlist->next, struct jffs2_eraseblock, list);
98 list_del(&ret->list);
99 c->gcblock = ret;
100 ret->gc_node = ret->first_node;
101 if (!ret->gc_node) {
102 pr_warn("Eep. ret->gc_node for block at 0x%08x is NULL\n",
103 ret->offset);
104 BUG();
105 }
106
107 /* Have we accidentally picked a clean block with wasted space ? */
108 if (ret->wasted_size) {
109 jffs2_dbg(1, "Converting wasted_size %08x to dirty_size\n",
110 ret->wasted_size);
111 ret->dirty_size += ret->wasted_size;
112 c->wasted_size -= ret->wasted_size;
113 c->dirty_size += ret->wasted_size;
114 ret->wasted_size = 0;
115 }
116
117 return ret;
118 }
119
120 /* jffs2_garbage_collect_pass
121 * Make a single attempt to progress GC. Move one node, and possibly
122 * start erasing one eraseblock.
123 */
124 int jffs2_garbage_collect_pass(struct jffs2_sb_info *c)
125 {
126 struct jffs2_inode_info *f;
127 struct jffs2_inode_cache *ic;
128 struct jffs2_eraseblock *jeb;
129 struct jffs2_raw_node_ref *raw;
130 uint32_t gcblock_dirty;
131 int ret = 0, inum, nlink;
132 int xattr = 0;
133
134 if (mutex_lock_interruptible(&c->alloc_sem))
135 return -EINTR;
136
137 for (;;) {
138 spin_lock(&c->erase_completion_lock);
139 if (!c->unchecked_size)
140 break;
141
142 /* We can't start doing GC yet. We haven't finished checking
143 the node CRCs etc. Do it now. */
144
145 /* checked_ino is protected by the alloc_sem */
146 if (c->checked_ino > c->highest_ino && xattr) {
147 pr_crit("Checked all inodes but still 0x%x bytes of unchecked space?\n",
148 c->unchecked_size);
149 jffs2_dbg_dump_block_lists_nolock(c);
150 spin_unlock(&c->erase_completion_lock);
151 mutex_unlock(&c->alloc_sem);
152 return -ENOSPC;
153 }
154
155 spin_unlock(&c->erase_completion_lock);
156
157 if (!xattr)
158 xattr = jffs2_verify_xattr(c);
159
160 spin_lock(&c->inocache_lock);
161
162 ic = jffs2_get_ino_cache(c, c->checked_ino++);
163
164 if (!ic) {
165 spin_unlock(&c->inocache_lock);
166 continue;
167 }
168
169 if (!ic->pino_nlink) {
170 jffs2_dbg(1, "Skipping check of ino #%d with nlink/pino zero\n",
171 ic->ino);
172 spin_unlock(&c->inocache_lock);
173 jffs2_xattr_delete_inode(c, ic);
174 continue;
175 }
176 switch(ic->state) {
177 case INO_STATE_CHECKEDABSENT:
178 case INO_STATE_PRESENT:
179 jffs2_dbg(1, "Skipping ino #%u already checked\n",
180 ic->ino);
181 spin_unlock(&c->inocache_lock);
182 continue;
183
184 case INO_STATE_GC:
185 case INO_STATE_CHECKING:
186 pr_warn("Inode #%u is in state %d during CRC check phase!\n",
187 ic->ino, ic->state);
188 spin_unlock(&c->inocache_lock);
189 BUG();
190
191 case INO_STATE_READING:
192 /* We need to wait for it to finish, lest we move on
193 and trigger the BUG() above while we haven't yet
194 finished checking all its nodes */
195 jffs2_dbg(1, "Waiting for ino #%u to finish reading\n",
196 ic->ino);
197 /* We need to come back again for the _same_ inode. We've
198 made no progress in this case, but that should be OK */
199 c->checked_ino--;
200
201 mutex_unlock(&c->alloc_sem);
202 sleep_on_spinunlock(&c->inocache_wq, &c->inocache_lock);
203 return 0;
204
205 default:
206 BUG();
207
208 case INO_STATE_UNCHECKED:
209 ;
210 }
211 ic->state = INO_STATE_CHECKING;
212 spin_unlock(&c->inocache_lock);
213
214 jffs2_dbg(1, "%s(): triggering inode scan of ino#%u\n",
215 __func__, ic->ino);
216
217 ret = jffs2_do_crccheck_inode(c, ic);
218 if (ret)
219 pr_warn("Returned error for crccheck of ino #%u. Expect badness...\n",
220 ic->ino);
221
222 jffs2_set_inocache_state(c, ic, INO_STATE_CHECKEDABSENT);
223 mutex_unlock(&c->alloc_sem);
224 return ret;
225 }
226
227 /* If there are any blocks which need erasing, erase them now */
228 if (!list_empty(&c->erase_complete_list) ||
229 !list_empty(&c->erase_pending_list)) {
230 spin_unlock(&c->erase_completion_lock);
231 mutex_unlock(&c->alloc_sem);
232 jffs2_dbg(1, "%s(): erasing pending blocks\n", __func__);
233 if (jffs2_erase_pending_blocks(c, 1))
234 return 0;
235
236 jffs2_dbg(1, "No progress from erasing block; doing GC anyway\n");
237 mutex_lock(&c->alloc_sem);
238 spin_lock(&c->erase_completion_lock);
239 }
240
241 /* First, work out which block we're garbage-collecting */
242 jeb = c->gcblock;
243
244 if (!jeb)
245 jeb = jffs2_find_gc_block(c);
246
247 if (!jeb) {
248 /* Couldn't find a free block. But maybe we can just erase one and make 'progress'? */
249 if (c->nr_erasing_blocks) {
250 spin_unlock(&c->erase_completion_lock);
251 mutex_unlock(&c->alloc_sem);
252 return -EAGAIN;
253 }
254 jffs2_dbg(1, "Couldn't find erase block to garbage collect!\n");
255 spin_unlock(&c->erase_completion_lock);
256 mutex_unlock(&c->alloc_sem);
257 return -EIO;
258 }
259
260 jffs2_dbg(1, "GC from block %08x, used_size %08x, dirty_size %08x, free_size %08x\n",
261 jeb->offset, jeb->used_size, jeb->dirty_size, jeb->free_size);
262 D1(if (c->nextblock)
263 printk(KERN_DEBUG "Nextblock at %08x, used_size %08x, dirty_size %08x, wasted_size %08x, free_size %08x\n", c->nextblock->offset, c->nextblock->used_size, c->nextblock->dirty_size, c->nextblock->wasted_size, c->nextblock->free_size));
264
265 if (!jeb->used_size) {
266 mutex_unlock(&c->alloc_sem);
267 goto eraseit;
268 }
269
270 raw = jeb->gc_node;
271 gcblock_dirty = jeb->dirty_size;
272
273 while(ref_obsolete(raw)) {
274 jffs2_dbg(1, "Node at 0x%08x is obsolete... skipping\n",
275 ref_offset(raw));
276 raw = ref_next(raw);
277 if (unlikely(!raw)) {
278 pr_warn("eep. End of raw list while still supposedly nodes to GC\n");
279 pr_warn("erase block at 0x%08x. free_size 0x%08x, dirty_size 0x%08x, used_size 0x%08x\n",
280 jeb->offset, jeb->free_size,
281 jeb->dirty_size, jeb->used_size);
282 jeb->gc_node = raw;
283 spin_unlock(&c->erase_completion_lock);
284 mutex_unlock(&c->alloc_sem);
285 BUG();
286 }
287 }
288 jeb->gc_node = raw;
289
290 jffs2_dbg(1, "Going to garbage collect node at 0x%08x\n",
291 ref_offset(raw));
292
293 if (!raw->next_in_ino) {
294 /* Inode-less node. Clean marker, snapshot or something like that */
295 spin_unlock(&c->erase_completion_lock);
296 if (ref_flags(raw) == REF_PRISTINE) {
297 /* It's an unknown node with JFFS2_FEATURE_RWCOMPAT_COPY */
298 jffs2_garbage_collect_pristine(c, NULL, raw);
299 } else {
300 /* Just mark it obsolete */
301 jffs2_mark_node_obsolete(c, raw);
302 }
303 mutex_unlock(&c->alloc_sem);
304 goto eraseit_lock;
305 }
306
307 ic = jffs2_raw_ref_to_ic(raw);
308
309 #ifdef CONFIG_JFFS2_FS_XATTR
310 /* When 'ic' refers xattr_datum/xattr_ref, this node is GCed as xattr.
311 * We can decide whether this node is inode or xattr by ic->class. */
312 if (ic->class == RAWNODE_CLASS_XATTR_DATUM
313 || ic->class == RAWNODE_CLASS_XATTR_REF) {
314 spin_unlock(&c->erase_completion_lock);
315
316 if (ic->class == RAWNODE_CLASS_XATTR_DATUM) {
317 ret = jffs2_garbage_collect_xattr_datum(c, (struct jffs2_xattr_datum *)ic, raw);
318 } else {
319 ret = jffs2_garbage_collect_xattr_ref(c, (struct jffs2_xattr_ref *)ic, raw);
320 }
321 goto test_gcnode;
322 }
323 #endif
324
325 /* We need to hold the inocache. Either the erase_completion_lock or
326 the inocache_lock are sufficient; we trade down since the inocache_lock
327 causes less contention. */
328 spin_lock(&c->inocache_lock);
329
330 spin_unlock(&c->erase_completion_lock);
331
332 jffs2_dbg(1, "%s(): collecting from block @0x%08x. Node @0x%08x(%d), ino #%u\n",
333 __func__, jeb->offset, ref_offset(raw), ref_flags(raw),
334 ic->ino);
335
336 /* Three possibilities:
337 1. Inode is already in-core. We must iget it and do proper
338 updating to its fragtree, etc.
339 2. Inode is not in-core, node is REF_PRISTINE. We lock the
340 inocache to prevent a read_inode(), copy the node intact.
341 3. Inode is not in-core, node is not pristine. We must iget()
342 and take the slow path.
343 */
344
345 switch(ic->state) {
346 case INO_STATE_CHECKEDABSENT:
347 /* It's been checked, but it's not currently in-core.
348 We can just copy any pristine nodes, but have
349 to prevent anyone else from doing read_inode() while
350 we're at it, so we set the state accordingly */
351 if (ref_flags(raw) == REF_PRISTINE)
352 ic->state = INO_STATE_GC;
353 else {
354 jffs2_dbg(1, "Ino #%u is absent but node not REF_PRISTINE. Reading.\n",
355 ic->ino);
356 }
357 break;
358
359 case INO_STATE_PRESENT:
360 /* It's in-core. GC must iget() it. */
361 break;
362
363 case INO_STATE_UNCHECKED:
364 case INO_STATE_CHECKING:
365 case INO_STATE_GC:
366 /* Should never happen. We should have finished checking
367 by the time we actually start doing any GC, and since
368 we're holding the alloc_sem, no other garbage collection
369 can happen.
370 */
371 pr_crit("Inode #%u already in state %d in jffs2_garbage_collect_pass()!\n",
372 ic->ino, ic->state);
373 mutex_unlock(&c->alloc_sem);
374 spin_unlock(&c->inocache_lock);
375 BUG();
376
377 case INO_STATE_READING:
378 /* Someone's currently trying to read it. We must wait for
379 them to finish and then go through the full iget() route
380 to do the GC. However, sometimes read_inode() needs to get
381 the alloc_sem() (for marking nodes invalid) so we must
382 drop the alloc_sem before sleeping. */
383
384 mutex_unlock(&c->alloc_sem);
385 jffs2_dbg(1, "%s(): waiting for ino #%u in state %d\n",
386 __func__, ic->ino, ic->state);
387 sleep_on_spinunlock(&c->inocache_wq, &c->inocache_lock);
388 /* And because we dropped the alloc_sem we must start again from the
389 beginning. Ponder chance of livelock here -- we're returning success
390 without actually making any progress.
391
392 Q: What are the chances that the inode is back in INO_STATE_READING
393 again by the time we next enter this function? And that this happens
394 enough times to cause a real delay?
395
396 A: Small enough that I don't care :)
397 */
398 return 0;
399 }
400
401 /* OK. Now if the inode is in state INO_STATE_GC, we are going to copy the
402 node intact, and we don't have to muck about with the fragtree etc.
403 because we know it's not in-core. If it _was_ in-core, we go through
404 all the iget() crap anyway */
405
406 if (ic->state == INO_STATE_GC) {
407 spin_unlock(&c->inocache_lock);
408
409 ret = jffs2_garbage_collect_pristine(c, ic, raw);
410
411 spin_lock(&c->inocache_lock);
412 ic->state = INO_STATE_CHECKEDABSENT;
413 wake_up(&c->inocache_wq);
414
415 if (ret != -EBADFD) {
416 spin_unlock(&c->inocache_lock);
417 goto test_gcnode;
418 }
419
420 /* Fall through if it wanted us to, with inocache_lock held */
421 }
422
423 /* Prevent the fairly unlikely race where the gcblock is
424 entirely obsoleted by the final close of a file which had
425 the only valid nodes in the block, followed by erasure,
426 followed by freeing of the ic because the erased block(s)
427 held _all_ the nodes of that inode.... never been seen but
428 it's vaguely possible. */
429
430 inum = ic->ino;
431 nlink = ic->pino_nlink;
432 spin_unlock(&c->inocache_lock);
433
434 f = jffs2_gc_fetch_inode(c, inum, !nlink);
435 if (IS_ERR(f)) {
436 ret = PTR_ERR(f);
437 goto release_sem;
438 }
439 if (!f) {
440 ret = 0;
441 goto release_sem;
442 }
443
444 ret = jffs2_garbage_collect_live(c, jeb, raw, f);
445
446 jffs2_gc_release_inode(c, f);
447
448 test_gcnode:
449 if (jeb->dirty_size == gcblock_dirty && !ref_obsolete(jeb->gc_node)) {
450 /* Eep. This really should never happen. GC is broken */
451 pr_err("Error garbage collecting node at %08x!\n",
452 ref_offset(jeb->gc_node));
453 ret = -ENOSPC;
454 }
455 release_sem:
456 mutex_unlock(&c->alloc_sem);
457
458 eraseit_lock:
459 /* If we've finished this block, start it erasing */
460 spin_lock(&c->erase_completion_lock);
461
462 eraseit:
463 if (c->gcblock && !c->gcblock->used_size) {
464 jffs2_dbg(1, "Block at 0x%08x completely obsoleted by GC. Moving to erase_pending_list\n",
465 c->gcblock->offset);
466 /* We're GC'ing an empty block? */
467 list_add_tail(&c->gcblock->list, &c->erase_pending_list);
468 c->gcblock = NULL;
469 c->nr_erasing_blocks++;
470 jffs2_garbage_collect_trigger(c);
471 }
472 spin_unlock(&c->erase_completion_lock);
473
474 return ret;
475 }
476
477 static int jffs2_garbage_collect_live(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb,
478 struct jffs2_raw_node_ref *raw, struct jffs2_inode_info *f)
479 {
480 struct jffs2_node_frag *frag;
481 struct jffs2_full_dnode *fn = NULL;
482 struct jffs2_full_dirent *fd;
483 uint32_t start = 0, end = 0, nrfrags = 0;
484 int ret = 0;
485
486 mutex_lock(&f->sem);
487
488 /* Now we have the lock for this inode. Check that it's still the one at the head
489 of the list. */
490
491 spin_lock(&c->erase_completion_lock);
492
493 if (c->gcblock != jeb) {
494 spin_unlock(&c->erase_completion_lock);
495 jffs2_dbg(1, "GC block is no longer gcblock. Restart\n");
496 goto upnout;
497 }
498 if (ref_obsolete(raw)) {
499 spin_unlock(&c->erase_completion_lock);
500 jffs2_dbg(1, "node to be GC'd was obsoleted in the meantime.\n");
501 /* They'll call again */
502 goto upnout;
503 }
504 spin_unlock(&c->erase_completion_lock);
505
506 /* OK. Looks safe. And nobody can get us now because we have the semaphore. Move the block */
507 if (f->metadata && f->metadata->raw == raw) {
508 fn = f->metadata;
509 ret = jffs2_garbage_collect_metadata(c, jeb, f, fn);
510 goto upnout;
511 }
512
513 /* FIXME. Read node and do lookup? */
514 for (frag = frag_first(&f->fragtree); frag; frag = frag_next(frag)) {
515 if (frag->node && frag->node->raw == raw) {
516 fn = frag->node;
517 end = frag->ofs + frag->size;
518 if (!nrfrags++)
519 start = frag->ofs;
520 if (nrfrags == frag->node->frags)
521 break; /* We've found them all */
522 }
523 }
524 if (fn) {
525 if (ref_flags(raw) == REF_PRISTINE) {
526 ret = jffs2_garbage_collect_pristine(c, f->inocache, raw);
527 if (!ret) {
528 /* Urgh. Return it sensibly. */
529 frag->node->raw = f->inocache->nodes;
530 }
531 if (ret != -EBADFD)
532 goto upnout;
533 }
534 /* We found a datanode. Do the GC */
535 if((start >> PAGE_CACHE_SHIFT) < ((end-1) >> PAGE_CACHE_SHIFT)) {
536 /* It crosses a page boundary. Therefore, it must be a hole. */
537 ret = jffs2_garbage_collect_hole(c, jeb, f, fn, start, end);
538 } else {
539 /* It could still be a hole. But we GC the page this way anyway */
540 ret = jffs2_garbage_collect_dnode(c, jeb, f, fn, start, end);
541 }
542 goto upnout;
543 }
544
545 /* Wasn't a dnode. Try dirent */
546 for (fd = f->dents; fd; fd=fd->next) {
547 if (fd->raw == raw)
548 break;
549 }
550
551 if (fd && fd->ino) {
552 ret = jffs2_garbage_collect_dirent(c, jeb, f, fd);
553 } else if (fd) {
554 ret = jffs2_garbage_collect_deletion_dirent(c, jeb, f, fd);
555 } else {
556 pr_warn("Raw node at 0x%08x wasn't in node lists for ino #%u\n",
557 ref_offset(raw), f->inocache->ino);
558 if (ref_obsolete(raw)) {
559 pr_warn("But it's obsolete so we don't mind too much\n");
560 } else {
561 jffs2_dbg_dump_node(c, ref_offset(raw));
562 BUG();
563 }
564 }
565 upnout:
566 mutex_unlock(&f->sem);
567
568 return ret;
569 }
570
571 static int jffs2_garbage_collect_pristine(struct jffs2_sb_info *c,
572 struct jffs2_inode_cache *ic,
573 struct jffs2_raw_node_ref *raw)
574 {
575 union jffs2_node_union *node;
576 size_t retlen;
577 int ret;
578 uint32_t phys_ofs, alloclen;
579 uint32_t crc, rawlen;
580 int retried = 0;
581
582 jffs2_dbg(1, "Going to GC REF_PRISTINE node at 0x%08x\n",
583 ref_offset(raw));
584
585 alloclen = rawlen = ref_totlen(c, c->gcblock, raw);
586
587 /* Ask for a small amount of space (or the totlen if smaller) because we
588 don't want to force wastage of the end of a block if splitting would
589 work. */
590 if (ic && alloclen > sizeof(struct jffs2_raw_inode) + JFFS2_MIN_DATA_LEN)
591 alloclen = sizeof(struct jffs2_raw_inode) + JFFS2_MIN_DATA_LEN;
592
593 ret = jffs2_reserve_space_gc(c, alloclen, &alloclen, rawlen);
594 /* 'rawlen' is not the exact summary size; it is only an upper estimation */
595
596 if (ret)
597 return ret;
598
599 if (alloclen < rawlen) {
600 /* Doesn't fit untouched. We'll go the old route and split it */
601 return -EBADFD;
602 }
603
604 node = kmalloc(rawlen, GFP_KERNEL);
605 if (!node)
606 return -ENOMEM;
607
608 ret = jffs2_flash_read(c, ref_offset(raw), rawlen, &retlen, (char *)node);
609 if (!ret && retlen != rawlen)
610 ret = -EIO;
611 if (ret)
612 goto out_node;
613
614 crc = crc32(0, node, sizeof(struct jffs2_unknown_node)-4);
615 if (je32_to_cpu(node->u.hdr_crc) != crc) {
616 pr_warn("Header CRC failed on REF_PRISTINE node at 0x%08x: Read 0x%08x, calculated 0x%08x\n",
617 ref_offset(raw), je32_to_cpu(node->u.hdr_crc), crc);
618 goto bail;
619 }
620
621 switch(je16_to_cpu(node->u.nodetype)) {
622 case JFFS2_NODETYPE_INODE:
623 crc = crc32(0, node, sizeof(node->i)-8);
624 if (je32_to_cpu(node->i.node_crc) != crc) {
625 pr_warn("Node CRC failed on REF_PRISTINE data node at 0x%08x: Read 0x%08x, calculated 0x%08x\n",
626 ref_offset(raw), je32_to_cpu(node->i.node_crc),
627 crc);
628 goto bail;
629 }
630
631 if (je32_to_cpu(node->i.dsize)) {
632 crc = crc32(0, node->i.data, je32_to_cpu(node->i.csize));
633 if (je32_to_cpu(node->i.data_crc) != crc) {
634 pr_warn("Data CRC failed on REF_PRISTINE data node at 0x%08x: Read 0x%08x, calculated 0x%08x\n",
635 ref_offset(raw),
636 je32_to_cpu(node->i.data_crc), crc);
637 goto bail;
638 }
639 }
640 break;
641
642 case JFFS2_NODETYPE_DIRENT:
643 crc = crc32(0, node, sizeof(node->d)-8);
644 if (je32_to_cpu(node->d.node_crc) != crc) {
645 pr_warn("Node CRC failed on REF_PRISTINE dirent node at 0x%08x: Read 0x%08x, calculated 0x%08x\n",
646 ref_offset(raw),
647 je32_to_cpu(node->d.node_crc), crc);
648 goto bail;
649 }
650
651 if (strnlen(node->d.name, node->d.nsize) != node->d.nsize) {
652 pr_warn("Name in dirent node at 0x%08x contains zeroes\n",
653 ref_offset(raw));
654 goto bail;
655 }
656
657 if (node->d.nsize) {
658 crc = crc32(0, node->d.name, node->d.nsize);
659 if (je32_to_cpu(node->d.name_crc) != crc) {
660 pr_warn("Name CRC failed on REF_PRISTINE dirent node at 0x%08x: Read 0x%08x, calculated 0x%08x\n",
661 ref_offset(raw),
662 je32_to_cpu(node->d.name_crc), crc);
663 goto bail;
664 }
665 }
666 break;
667 default:
668 /* If it's inode-less, we don't _know_ what it is. Just copy it intact */
669 if (ic) {
670 pr_warn("Unknown node type for REF_PRISTINE node at 0x%08x: 0x%04x\n",
671 ref_offset(raw), je16_to_cpu(node->u.nodetype));
672 goto bail;
673 }
674 }
675
676 /* OK, all the CRCs are good; this node can just be copied as-is. */
677 retry:
678 phys_ofs = write_ofs(c);
679
680 ret = jffs2_flash_write(c, phys_ofs, rawlen, &retlen, (char *)node);
681
682 if (ret || (retlen != rawlen)) {
683 pr_notice("Write of %d bytes at 0x%08x failed. returned %d, retlen %zd\n",
684 rawlen, phys_ofs, ret, retlen);
685 if (retlen) {
686 jffs2_add_physical_node_ref(c, phys_ofs | REF_OBSOLETE, rawlen, NULL);
687 } else {
688 pr_notice("Not marking the space at 0x%08x as dirty because the flash driver returned retlen zero\n",
689 phys_ofs);
690 }
691 if (!retried) {
692 /* Try to reallocate space and retry */
693 uint32_t dummy;
694 struct jffs2_eraseblock *jeb = &c->blocks[phys_ofs / c->sector_size];
695
696 retried = 1;
697
698 jffs2_dbg(1, "Retrying failed write of REF_PRISTINE node.\n");
699
700 jffs2_dbg_acct_sanity_check(c,jeb);
701 jffs2_dbg_acct_paranoia_check(c, jeb);
702
703 ret = jffs2_reserve_space_gc(c, rawlen, &dummy, rawlen);
704 /* this is not the exact summary size of it,
705 it is only an upper estimation */
706
707 if (!ret) {
708 jffs2_dbg(1, "Allocated space at 0x%08x to retry failed write.\n",
709 phys_ofs);
710
711 jffs2_dbg_acct_sanity_check(c,jeb);
712 jffs2_dbg_acct_paranoia_check(c, jeb);
713
714 goto retry;
715 }
716 jffs2_dbg(1, "Failed to allocate space to retry failed write: %d!\n",
717 ret);
718 }
719
720 if (!ret)
721 ret = -EIO;
722 goto out_node;
723 }
724 jffs2_add_physical_node_ref(c, phys_ofs | REF_PRISTINE, rawlen, ic);
725
726 jffs2_mark_node_obsolete(c, raw);
727 jffs2_dbg(1, "WHEEE! GC REF_PRISTINE node at 0x%08x succeeded\n",
728 ref_offset(raw));
729
730 out_node:
731 kfree(node);
732 return ret;
733 bail:
734 ret = -EBADFD;
735 goto out_node;
736 }
737
738 static int jffs2_garbage_collect_metadata(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb,
739 struct jffs2_inode_info *f, struct jffs2_full_dnode *fn)
740 {
741 struct jffs2_full_dnode *new_fn;
742 struct jffs2_raw_inode ri;
743 struct jffs2_node_frag *last_frag;
744 union jffs2_device_node dev;
745 char *mdata = NULL;
746 int mdatalen = 0;
747 uint32_t alloclen, ilen;
748 int ret;
749
750 if (S_ISBLK(JFFS2_F_I_MODE(f)) ||
751 S_ISCHR(JFFS2_F_I_MODE(f)) ) {
752 /* For these, we don't actually need to read the old node */
753 mdatalen = jffs2_encode_dev(&dev, JFFS2_F_I_RDEV(f));
754 mdata = (char *)&dev;
755 jffs2_dbg(1, "%s(): Writing %d bytes of kdev_t\n",
756 __func__, mdatalen);
757 } else if (S_ISLNK(JFFS2_F_I_MODE(f))) {
758 mdatalen = fn->size;
759 mdata = kmalloc(fn->size, GFP_KERNEL);
760 if (!mdata) {
761 pr_warn("kmalloc of mdata failed in jffs2_garbage_collect_metadata()\n");
762 return -ENOMEM;
763 }
764 ret = jffs2_read_dnode(c, f, fn, mdata, 0, mdatalen);
765 if (ret) {
766 pr_warn("read of old metadata failed in jffs2_garbage_collect_metadata(): %d\n",
767 ret);
768 kfree(mdata);
769 return ret;
770 }
771 jffs2_dbg(1, "%s(): Writing %d bites of symlink target\n",
772 __func__, mdatalen);
773
774 }
775
776 ret = jffs2_reserve_space_gc(c, sizeof(ri) + mdatalen, &alloclen,
777 JFFS2_SUMMARY_INODE_SIZE);
778 if (ret) {
779 pr_warn("jffs2_reserve_space_gc of %zd bytes for garbage_collect_metadata failed: %d\n",
780 sizeof(ri) + mdatalen, ret);
781 goto out;
782 }
783
784 last_frag = frag_last(&f->fragtree);
785 if (last_frag)
786 /* Fetch the inode length from the fragtree rather then
787 * from i_size since i_size may have not been updated yet */
788 ilen = last_frag->ofs + last_frag->size;
789 else
790 ilen = JFFS2_F_I_SIZE(f);
791
792 memset(&ri, 0, sizeof(ri));
793 ri.magic = cpu_to_je16(JFFS2_MAGIC_BITMASK);
794 ri.nodetype = cpu_to_je16(JFFS2_NODETYPE_INODE);
795 ri.totlen = cpu_to_je32(sizeof(ri) + mdatalen);
796 ri.hdr_crc = cpu_to_je32(crc32(0, &ri, sizeof(struct jffs2_unknown_node)-4));
797
798 ri.ino = cpu_to_je32(f->inocache->ino);
799 ri.version = cpu_to_je32(++f->highest_version);
800 ri.mode = cpu_to_jemode(JFFS2_F_I_MODE(f));
801 ri.uid = cpu_to_je16(JFFS2_F_I_UID(f));
802 ri.gid = cpu_to_je16(JFFS2_F_I_GID(f));
803 ri.isize = cpu_to_je32(ilen);
804 ri.atime = cpu_to_je32(JFFS2_F_I_ATIME(f));
805 ri.ctime = cpu_to_je32(JFFS2_F_I_CTIME(f));
806 ri.mtime = cpu_to_je32(JFFS2_F_I_MTIME(f));
807 ri.offset = cpu_to_je32(0);
808 ri.csize = cpu_to_je32(mdatalen);
809 ri.dsize = cpu_to_je32(mdatalen);
810 ri.compr = JFFS2_COMPR_NONE;
811 ri.node_crc = cpu_to_je32(crc32(0, &ri, sizeof(ri)-8));
812 ri.data_crc = cpu_to_je32(crc32(0, mdata, mdatalen));
813
814 new_fn = jffs2_write_dnode(c, f, &ri, mdata, mdatalen, ALLOC_GC);
815
816 if (IS_ERR(new_fn)) {
817 pr_warn("Error writing new dnode: %ld\n", PTR_ERR(new_fn));
818 ret = PTR_ERR(new_fn);
819 goto out;
820 }
821 jffs2_mark_node_obsolete(c, fn->raw);
822 jffs2_free_full_dnode(fn);
823 f->metadata = new_fn;
824 out:
825 if (S_ISLNK(JFFS2_F_I_MODE(f)))
826 kfree(mdata);
827 return ret;
828 }
829
830 static int jffs2_garbage_collect_dirent(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb,
831 struct jffs2_inode_info *f, struct jffs2_full_dirent *fd)
832 {
833 struct jffs2_full_dirent *new_fd;
834 struct jffs2_raw_dirent rd;
835 uint32_t alloclen;
836 int ret;
837
838 rd.magic = cpu_to_je16(JFFS2_MAGIC_BITMASK);
839 rd.nodetype = cpu_to_je16(JFFS2_NODETYPE_DIRENT);
840 rd.nsize = strlen(fd->name);
841 rd.totlen = cpu_to_je32(sizeof(rd) + rd.nsize);
842 rd.hdr_crc = cpu_to_je32(crc32(0, &rd, sizeof(struct jffs2_unknown_node)-4));
843
844 rd.pino = cpu_to_je32(f->inocache->ino);
845 rd.version = cpu_to_je32(++f->highest_version);
846 rd.ino = cpu_to_je32(fd->ino);
847 /* If the times on this inode were set by explicit utime() they can be different,
848 so refrain from splatting them. */
849 if (JFFS2_F_I_MTIME(f) == JFFS2_F_I_CTIME(f))
850 rd.mctime = cpu_to_je32(JFFS2_F_I_MTIME(f));
851 else
852 rd.mctime = cpu_to_je32(0);
853 rd.type = fd->type;
854 rd.node_crc = cpu_to_je32(crc32(0, &rd, sizeof(rd)-8));
855 rd.name_crc = cpu_to_je32(crc32(0, fd->name, rd.nsize));
856
857 ret = jffs2_reserve_space_gc(c, sizeof(rd)+rd.nsize, &alloclen,
858 JFFS2_SUMMARY_DIRENT_SIZE(rd.nsize));
859 if (ret) {
860 pr_warn("jffs2_reserve_space_gc of %zd bytes for garbage_collect_dirent failed: %d\n",
861 sizeof(rd)+rd.nsize, ret);
862 return ret;
863 }
864 new_fd = jffs2_write_dirent(c, f, &rd, fd->name, rd.nsize, ALLOC_GC);
865
866 if (IS_ERR(new_fd)) {
867 pr_warn("jffs2_write_dirent in garbage_collect_dirent failed: %ld\n",
868 PTR_ERR(new_fd));
869 return PTR_ERR(new_fd);
870 }
871 jffs2_add_fd_to_list(c, new_fd, &f->dents);
872 return 0;
873 }
874
875 static int jffs2_garbage_collect_deletion_dirent(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb,
876 struct jffs2_inode_info *f, struct jffs2_full_dirent *fd)
877 {
878 struct jffs2_full_dirent **fdp = &f->dents;
879 int found = 0;
880
881 /* On a medium where we can't actually mark nodes obsolete
882 pernamently, such as NAND flash, we need to work out
883 whether this deletion dirent is still needed to actively
884 delete a 'real' dirent with the same name that's still
885 somewhere else on the flash. */
886 if (!jffs2_can_mark_obsolete(c)) {
887 struct jffs2_raw_dirent *rd;
888 struct jffs2_raw_node_ref *raw;
889 int ret;
890 size_t retlen;
891 int name_len = strlen(fd->name);
892 uint32_t name_crc = crc32(0, fd->name, name_len);
893 uint32_t rawlen = ref_totlen(c, jeb, fd->raw);
894
895 rd = kmalloc(rawlen, GFP_KERNEL);
896 if (!rd)
897 return -ENOMEM;
898
899 /* Prevent the erase code from nicking the obsolete node refs while
900 we're looking at them. I really don't like this extra lock but
901 can't see any alternative. Suggestions on a postcard to... */
902 mutex_lock(&c->erase_free_sem);
903
904 for (raw = f->inocache->nodes; raw != (void *)f->inocache; raw = raw->next_in_ino) {
905
906 cond_resched();
907
908 /* We only care about obsolete ones */
909 if (!(ref_obsolete(raw)))
910 continue;
911
912 /* Any dirent with the same name is going to have the same length... */
913 if (ref_totlen(c, NULL, raw) != rawlen)
914 continue;
915
916 /* Doesn't matter if there's one in the same erase block. We're going to
917 delete it too at the same time. */
918 if (SECTOR_ADDR(raw->flash_offset) == SECTOR_ADDR(fd->raw->flash_offset))
919 continue;
920
921 jffs2_dbg(1, "Check potential deletion dirent at %08x\n",
922 ref_offset(raw));
923
924 /* This is an obsolete node belonging to the same directory, and it's of the right
925 length. We need to take a closer look...*/
926 ret = jffs2_flash_read(c, ref_offset(raw), rawlen, &retlen, (char *)rd);
927 if (ret) {
928 pr_warn("%s(): Read error (%d) reading obsolete node at %08x\n",
929 __func__, ret, ref_offset(raw));
930 /* If we can't read it, we don't need to continue to obsolete it. Continue */
931 continue;
932 }
933 if (retlen != rawlen) {
934 pr_warn("%s(): Short read (%zd not %u) reading header from obsolete node at %08x\n",
935 __func__, retlen, rawlen,
936 ref_offset(raw));
937 continue;
938 }
939
940 if (je16_to_cpu(rd->nodetype) != JFFS2_NODETYPE_DIRENT)
941 continue;
942
943 /* If the name CRC doesn't match, skip */
944 if (je32_to_cpu(rd->name_crc) != name_crc)
945 continue;
946
947 /* If the name length doesn't match, or it's another deletion dirent, skip */
948 if (rd->nsize != name_len || !je32_to_cpu(rd->ino))
949 continue;
950
951 /* OK, check the actual name now */
952 if (memcmp(rd->name, fd->name, name_len))
953 continue;
954
955 /* OK. The name really does match. There really is still an older node on
956 the flash which our deletion dirent obsoletes. So we have to write out
957 a new deletion dirent to replace it */
958 mutex_unlock(&c->erase_free_sem);
959
960 jffs2_dbg(1, "Deletion dirent at %08x still obsoletes real dirent \"%s\" at %08x for ino #%u\n",
961 ref_offset(fd->raw), fd->name,
962 ref_offset(raw), je32_to_cpu(rd->ino));
963 kfree(rd);
964
965 return jffs2_garbage_collect_dirent(c, jeb, f, fd);
966 }
967
968 mutex_unlock(&c->erase_free_sem);
969 kfree(rd);
970 }
971
972 /* FIXME: If we're deleting a dirent which contains the current mtime and ctime,
973 we should update the metadata node with those times accordingly */
974
975 /* No need for it any more. Just mark it obsolete and remove it from the list */
976 while (*fdp) {
977 if ((*fdp) == fd) {
978 found = 1;
979 *fdp = fd->next;
980 break;
981 }
982 fdp = &(*fdp)->next;
983 }
984 if (!found) {
985 pr_warn("Deletion dirent \"%s\" not found in list for ino #%u\n",
986 fd->name, f->inocache->ino);
987 }
988 jffs2_mark_node_obsolete(c, fd->raw);
989 jffs2_free_full_dirent(fd);
990 return 0;
991 }
992
993 static int jffs2_garbage_collect_hole(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb,
994 struct jffs2_inode_info *f, struct jffs2_full_dnode *fn,
995 uint32_t start, uint32_t end)
996 {
997 struct jffs2_raw_inode ri;
998 struct jffs2_node_frag *frag;
999 struct jffs2_full_dnode *new_fn;
1000 uint32_t alloclen, ilen;
1001 int ret;
1002
1003 jffs2_dbg(1, "Writing replacement hole node for ino #%u from offset 0x%x to 0x%x\n",
1004 f->inocache->ino, start, end);
1005
1006 memset(&ri, 0, sizeof(ri));
1007
1008 if(fn->frags > 1) {
1009 size_t readlen;
1010 uint32_t crc;
1011 /* It's partially obsoleted by a later write. So we have to
1012 write it out again with the _same_ version as before */
1013 ret = jffs2_flash_read(c, ref_offset(fn->raw), sizeof(ri), &readlen, (char *)&ri);
1014 if (readlen != sizeof(ri) || ret) {
1015 pr_warn("Node read failed in jffs2_garbage_collect_hole. Ret %d, retlen %zd. Data will be lost by writing new hole node\n",
1016 ret, readlen);
1017 goto fill;
1018 }
1019 if (je16_to_cpu(ri.nodetype) != JFFS2_NODETYPE_INODE) {
1020 pr_warn("%s(): Node at 0x%08x had node type 0x%04x instead of JFFS2_NODETYPE_INODE(0x%04x)\n",
1021 __func__, ref_offset(fn->raw),
1022 je16_to_cpu(ri.nodetype), JFFS2_NODETYPE_INODE);
1023 return -EIO;
1024 }
1025 if (je32_to_cpu(ri.totlen) != sizeof(ri)) {
1026 pr_warn("%s(): Node at 0x%08x had totlen 0x%x instead of expected 0x%zx\n",
1027 __func__, ref_offset(fn->raw),
1028 je32_to_cpu(ri.totlen), sizeof(ri));
1029 return -EIO;
1030 }
1031 crc = crc32(0, &ri, sizeof(ri)-8);
1032 if (crc != je32_to_cpu(ri.node_crc)) {
1033 pr_warn("%s: Node at 0x%08x had CRC 0x%08x which doesn't match calculated CRC 0x%08x\n",
1034 __func__, ref_offset(fn->raw),
1035 je32_to_cpu(ri.node_crc), crc);
1036 /* FIXME: We could possibly deal with this by writing new holes for each frag */
1037 pr_warn("Data in the range 0x%08x to 0x%08x of inode #%u will be lost\n",
1038 start, end, f->inocache->ino);
1039 goto fill;
1040 }
1041 if (ri.compr != JFFS2_COMPR_ZERO) {
1042 pr_warn("%s(): Node 0x%08x wasn't a hole node!\n",
1043 __func__, ref_offset(fn->raw));
1044 pr_warn("Data in the range 0x%08x to 0x%08x of inode #%u will be lost\n",
1045 start, end, f->inocache->ino);
1046 goto fill;
1047 }
1048 } else {
1049 fill:
1050 ri.magic = cpu_to_je16(JFFS2_MAGIC_BITMASK);
1051 ri.nodetype = cpu_to_je16(JFFS2_NODETYPE_INODE);
1052 ri.totlen = cpu_to_je32(sizeof(ri));
1053 ri.hdr_crc = cpu_to_je32(crc32(0, &ri, sizeof(struct jffs2_unknown_node)-4));
1054
1055 ri.ino = cpu_to_je32(f->inocache->ino);
1056 ri.version = cpu_to_je32(++f->highest_version);
1057 ri.offset = cpu_to_je32(start);
1058 ri.dsize = cpu_to_je32(end - start);
1059 ri.csize = cpu_to_je32(0);
1060 ri.compr = JFFS2_COMPR_ZERO;
1061 }
1062
1063 frag = frag_last(&f->fragtree);
1064 if (frag)
1065 /* Fetch the inode length from the fragtree rather then
1066 * from i_size since i_size may have not been updated yet */
1067 ilen = frag->ofs + frag->size;
1068 else
1069 ilen = JFFS2_F_I_SIZE(f);
1070
1071 ri.mode = cpu_to_jemode(JFFS2_F_I_MODE(f));
1072 ri.uid = cpu_to_je16(JFFS2_F_I_UID(f));
1073 ri.gid = cpu_to_je16(JFFS2_F_I_GID(f));
1074 ri.isize = cpu_to_je32(ilen);
1075 ri.atime = cpu_to_je32(JFFS2_F_I_ATIME(f));
1076 ri.ctime = cpu_to_je32(JFFS2_F_I_CTIME(f));
1077 ri.mtime = cpu_to_je32(JFFS2_F_I_MTIME(f));
1078 ri.data_crc = cpu_to_je32(0);
1079 ri.node_crc = cpu_to_je32(crc32(0, &ri, sizeof(ri)-8));
1080
1081 ret = jffs2_reserve_space_gc(c, sizeof(ri), &alloclen,
1082 JFFS2_SUMMARY_INODE_SIZE);
1083 if (ret) {
1084 pr_warn("jffs2_reserve_space_gc of %zd bytes for garbage_collect_hole failed: %d\n",
1085 sizeof(ri), ret);
1086 return ret;
1087 }
1088 new_fn = jffs2_write_dnode(c, f, &ri, NULL, 0, ALLOC_GC);
1089
1090 if (IS_ERR(new_fn)) {
1091 pr_warn("Error writing new hole node: %ld\n", PTR_ERR(new_fn));
1092 return PTR_ERR(new_fn);
1093 }
1094 if (je32_to_cpu(ri.version) == f->highest_version) {
1095 jffs2_add_full_dnode_to_inode(c, f, new_fn);
1096 if (f->metadata) {
1097 jffs2_mark_node_obsolete(c, f->metadata->raw);
1098 jffs2_free_full_dnode(f->metadata);
1099 f->metadata = NULL;
1100 }
1101 return 0;
1102 }
1103
1104 /*
1105 * We should only get here in the case where the node we are
1106 * replacing had more than one frag, so we kept the same version
1107 * number as before. (Except in case of error -- see 'goto fill;'
1108 * above.)
1109 */
1110 D1(if(unlikely(fn->frags <= 1)) {
1111 pr_warn("%s(): Replacing fn with %d frag(s) but new ver %d != highest_version %d of ino #%d\n",
1112 __func__, fn->frags, je32_to_cpu(ri.version),
1113 f->highest_version, je32_to_cpu(ri.ino));
1114 });
1115
1116 /* This is a partially-overlapped hole node. Mark it REF_NORMAL not REF_PRISTINE */
1117 mark_ref_normal(new_fn->raw);
1118
1119 for (frag = jffs2_lookup_node_frag(&f->fragtree, fn->ofs);
1120 frag; frag = frag_next(frag)) {
1121 if (frag->ofs > fn->size + fn->ofs)
1122 break;
1123 if (frag->node == fn) {
1124 frag->node = new_fn;
1125 new_fn->frags++;
1126 fn->frags--;
1127 }
1128 }
1129 if (fn->frags) {
1130 pr_warn("%s(): Old node still has frags!\n", __func__);
1131 BUG();
1132 }
1133 if (!new_fn->frags) {
1134 pr_warn("%s(): New node has no frags!\n", __func__);
1135 BUG();
1136 }
1137
1138 jffs2_mark_node_obsolete(c, fn->raw);
1139 jffs2_free_full_dnode(fn);
1140
1141 return 0;
1142 }
1143
1144 static int jffs2_garbage_collect_dnode(struct jffs2_sb_info *c, struct jffs2_eraseblock *orig_jeb,
1145 struct jffs2_inode_info *f, struct jffs2_full_dnode *fn,
1146 uint32_t start, uint32_t end)
1147 {
1148 struct jffs2_full_dnode *new_fn;
1149 struct jffs2_raw_inode ri;
1150 uint32_t alloclen, offset, orig_end, orig_start;
1151 int ret = 0;
1152 unsigned char *comprbuf = NULL, *writebuf;
1153 unsigned long pg;
1154 unsigned char *pg_ptr;
1155
1156 memset(&ri, 0, sizeof(ri));
1157
1158 jffs2_dbg(1, "Writing replacement dnode for ino #%u from offset 0x%x to 0x%x\n",
1159 f->inocache->ino, start, end);
1160
1161 orig_end = end;
1162 orig_start = start;
1163
1164 if (c->nr_free_blocks + c->nr_erasing_blocks > c->resv_blocks_gcmerge) {
1165 /* Attempt to do some merging. But only expand to cover logically
1166 adjacent frags if the block containing them is already considered
1167 to be dirty. Otherwise we end up with GC just going round in
1168 circles dirtying the nodes it already wrote out, especially
1169 on NAND where we have small eraseblocks and hence a much higher
1170 chance of nodes having to be split to cross boundaries. */
1171
1172 struct jffs2_node_frag *frag;
1173 uint32_t min, max;
1174
1175 min = start & ~(PAGE_CACHE_SIZE-1);
1176 max = min + PAGE_CACHE_SIZE;
1177
1178 frag = jffs2_lookup_node_frag(&f->fragtree, start);
1179
1180 /* BUG_ON(!frag) but that'll happen anyway... */
1181
1182 BUG_ON(frag->ofs != start);
1183
1184 /* First grow down... */
1185 while((frag = frag_prev(frag)) && frag->ofs >= min) {
1186
1187 /* If the previous frag doesn't even reach the beginning, there's
1188 excessive fragmentation. Just merge. */
1189 if (frag->ofs > min) {
1190 jffs2_dbg(1, "Expanding down to cover partial frag (0x%x-0x%x)\n",
1191 frag->ofs, frag->ofs+frag->size);
1192 start = frag->ofs;
1193 continue;
1194 }
1195 /* OK. This frag holds the first byte of the page. */
1196 if (!frag->node || !frag->node->raw) {
1197 jffs2_dbg(1, "First frag in page is hole (0x%x-0x%x). Not expanding down.\n",
1198 frag->ofs, frag->ofs+frag->size);
1199 break;
1200 } else {
1201
1202 /* OK, it's a frag which extends to the beginning of the page. Does it live
1203 in a block which is still considered clean? If so, don't obsolete it.
1204 If not, cover it anyway. */
1205
1206 struct jffs2_raw_node_ref *raw = frag->node->raw;
1207 struct jffs2_eraseblock *jeb;
1208
1209 jeb = &c->blocks[raw->flash_offset / c->sector_size];
1210
1211 if (jeb == c->gcblock) {
1212 jffs2_dbg(1, "Expanding down to cover frag (0x%x-0x%x) in gcblock at %08x\n",
1213 frag->ofs,
1214 frag->ofs + frag->size,
1215 ref_offset(raw));
1216 start = frag->ofs;
1217 break;
1218 }
1219 if (!ISDIRTY(jeb->dirty_size + jeb->wasted_size)) {
1220 jffs2_dbg(1, "Not expanding down to cover frag (0x%x-0x%x) in clean block %08x\n",
1221 frag->ofs,
1222 frag->ofs + frag->size,
1223 jeb->offset);
1224 break;
1225 }
1226
1227 jffs2_dbg(1, "Expanding down to cover frag (0x%x-0x%x) in dirty block %08x\n",
1228 frag->ofs,
1229 frag->ofs + frag->size,
1230 jeb->offset);
1231 start = frag->ofs;
1232 break;
1233 }
1234 }
1235
1236 /* ... then up */
1237
1238 /* Find last frag which is actually part of the node we're to GC. */
1239 frag = jffs2_lookup_node_frag(&f->fragtree, end-1);
1240
1241 while((frag = frag_next(frag)) && frag->ofs+frag->size <= max) {
1242
1243 /* If the previous frag doesn't even reach the beginning, there's lots
1244 of fragmentation. Just merge. */
1245 if (frag->ofs+frag->size < max) {
1246 jffs2_dbg(1, "Expanding up to cover partial frag (0x%x-0x%x)\n",
1247 frag->ofs, frag->ofs+frag->size);
1248 end = frag->ofs + frag->size;
1249 continue;
1250 }
1251
1252 if (!frag->node || !frag->node->raw) {
1253 jffs2_dbg(1, "Last frag in page is hole (0x%x-0x%x). Not expanding up.\n",
1254 frag->ofs, frag->ofs+frag->size);
1255 break;
1256 } else {
1257
1258 /* OK, it's a frag which extends to the beginning of the page. Does it live
1259 in a block which is still considered clean? If so, don't obsolete it.
1260 If not, cover it anyway. */
1261
1262 struct jffs2_raw_node_ref *raw = frag->node->raw;
1263 struct jffs2_eraseblock *jeb;
1264
1265 jeb = &c->blocks[raw->flash_offset / c->sector_size];
1266
1267 if (jeb == c->gcblock) {
1268 jffs2_dbg(1, "Expanding up to cover frag (0x%x-0x%x) in gcblock at %08x\n",
1269 frag->ofs,
1270 frag->ofs + frag->size,
1271 ref_offset(raw));
1272 end = frag->ofs + frag->size;
1273 break;
1274 }
1275 if (!ISDIRTY(jeb->dirty_size + jeb->wasted_size)) {
1276 jffs2_dbg(1, "Not expanding up to cover frag (0x%x-0x%x) in clean block %08x\n",
1277 frag->ofs,
1278 frag->ofs + frag->size,
1279 jeb->offset);
1280 break;
1281 }
1282
1283 jffs2_dbg(1, "Expanding up to cover frag (0x%x-0x%x) in dirty block %08x\n",
1284 frag->ofs,
1285 frag->ofs + frag->size,
1286 jeb->offset);
1287 end = frag->ofs + frag->size;
1288 break;
1289 }
1290 }
1291 jffs2_dbg(1, "Expanded dnode to write from (0x%x-0x%x) to (0x%x-0x%x)\n",
1292 orig_start, orig_end, start, end);
1293
1294 D1(BUG_ON(end > frag_last(&f->fragtree)->ofs + frag_last(&f->fragtree)->size));
1295 BUG_ON(end < orig_end);
1296 BUG_ON(start > orig_start);
1297 }
1298
1299 /* First, use readpage() to read the appropriate page into the page cache */
1300 /* Q: What happens if we actually try to GC the _same_ page for which commit_write()
1301 * triggered garbage collection in the first place?
1302 * A: I _think_ it's OK. read_cache_page shouldn't deadlock, we'll write out the
1303 * page OK. We'll actually write it out again in commit_write, which is a little
1304 * suboptimal, but at least we're correct.
1305 */
1306 pg_ptr = jffs2_gc_fetch_page(c, f, start, &pg);
1307
1308 if (IS_ERR(pg_ptr)) {
1309 pr_warn("read_cache_page() returned error: %ld\n",
1310 PTR_ERR(pg_ptr));
1311 return PTR_ERR(pg_ptr);
1312 }
1313
1314 offset = start;
1315 while(offset < orig_end) {
1316 uint32_t datalen;
1317 uint32_t cdatalen;
1318 uint16_t comprtype = JFFS2_COMPR_NONE;
1319
1320 ret = jffs2_reserve_space_gc(c, sizeof(ri) + JFFS2_MIN_DATA_LEN,
1321 &alloclen, JFFS2_SUMMARY_INODE_SIZE);
1322
1323 if (ret) {
1324 pr_warn("jffs2_reserve_space_gc of %zd bytes for garbage_collect_dnode failed: %d\n",
1325 sizeof(ri) + JFFS2_MIN_DATA_LEN, ret);
1326 break;
1327 }
1328 cdatalen = min_t(uint32_t, alloclen - sizeof(ri), end - offset);
1329 datalen = end - offset;
1330
1331 writebuf = pg_ptr + (offset & (PAGE_CACHE_SIZE -1));
1332
1333 comprtype = jffs2_compress(c, f, writebuf, &comprbuf, &datalen, &cdatalen);
1334
1335 ri.magic = cpu_to_je16(JFFS2_MAGIC_BITMASK);
1336 ri.nodetype = cpu_to_je16(JFFS2_NODETYPE_INODE);
1337 ri.totlen = cpu_to_je32(sizeof(ri) + cdatalen);
1338 ri.hdr_crc = cpu_to_je32(crc32(0, &ri, sizeof(struct jffs2_unknown_node)-4));
1339
1340 ri.ino = cpu_to_je32(f->inocache->ino);
1341 ri.version = cpu_to_je32(++f->highest_version);
1342 ri.mode = cpu_to_jemode(JFFS2_F_I_MODE(f));
1343 ri.uid = cpu_to_je16(JFFS2_F_I_UID(f));
1344 ri.gid = cpu_to_je16(JFFS2_F_I_GID(f));
1345 ri.isize = cpu_to_je32(JFFS2_F_I_SIZE(f));
1346 ri.atime = cpu_to_je32(JFFS2_F_I_ATIME(f));
1347 ri.ctime = cpu_to_je32(JFFS2_F_I_CTIME(f));
1348 ri.mtime = cpu_to_je32(JFFS2_F_I_MTIME(f));
1349 ri.offset = cpu_to_je32(offset);
1350 ri.csize = cpu_to_je32(cdatalen);
1351 ri.dsize = cpu_to_je32(datalen);
1352 ri.compr = comprtype & 0xff;
1353 ri.usercompr = (comprtype >> 8) & 0xff;
1354 ri.node_crc = cpu_to_je32(crc32(0, &ri, sizeof(ri)-8));
1355 ri.data_crc = cpu_to_je32(crc32(0, comprbuf, cdatalen));
1356
1357 new_fn = jffs2_write_dnode(c, f, &ri, comprbuf, cdatalen, ALLOC_GC);
1358
1359 jffs2_free_comprbuf(comprbuf, writebuf);
1360
1361 if (IS_ERR(new_fn)) {
1362 pr_warn("Error writing new dnode: %ld\n",
1363 PTR_ERR(new_fn));
1364 ret = PTR_ERR(new_fn);
1365 break;
1366 }
1367 ret = jffs2_add_full_dnode_to_inode(c, f, new_fn);
1368 offset += datalen;
1369 if (f->metadata) {
1370 jffs2_mark_node_obsolete(c, f->metadata->raw);
1371 jffs2_free_full_dnode(f->metadata);
1372 f->metadata = NULL;
1373 }
1374 }
1375
1376 jffs2_gc_release_page(c, pg_ptr, &pg);
1377 return ret;
1378 }
Linux with added FPC-III support