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#include <linux/kernel.h> 14#include <linux/mtd/mtd.h> 15#include <linux/slab.h> 16#include <linux/pagemap.h> 17#include <linux/crc32.h> 18#include <linux/compiler.h> 19#include <linux/stat.h> 20#include "nodelist.h" 21#include "compr.h" 22 23static int jffs2_garbage_collect_pristine(struct jffs2_sb_info *c, 24 struct jffs2_inode_cache *ic, 25 struct jffs2_raw_node_ref *raw); 26static int jffs2_garbage_collect_metadata(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb, 27 struct jffs2_inode_info *f, struct jffs2_full_dnode *fd); 28static int jffs2_garbage_collect_dirent(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb, 29 struct jffs2_inode_info *f, struct jffs2_full_dirent *fd); 30static int jffs2_garbage_collect_deletion_dirent(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb, 31 struct jffs2_inode_info *f, struct jffs2_full_dirent *fd); 32static int jffs2_garbage_collect_hole(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb, 33 struct jffs2_inode_info *f, struct jffs2_full_dnode *fn, 34 uint32_t start, uint32_t end); 35static int jffs2_garbage_collect_dnode(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb, 36 struct jffs2_inode_info *f, struct jffs2_full_dnode *fn, 37 uint32_t start, uint32_t end); 38static int jffs2_garbage_collect_live(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb, 39 struct jffs2_raw_node_ref *raw, struct jffs2_inode_info *f); 40 41/* Called with erase_completion_lock held */ 42static struct jffs2_eraseblock *jffs2_find_gc_block(struct jffs2_sb_info *c) 43{ 44 struct jffs2_eraseblock *ret; 45 struct list_head *nextlist = NULL; 46 int n = jiffies % 128; 47 48 /* Pick an eraseblock to garbage collect next. This is where we'll 49 put the clever wear-levelling algorithms. Eventually. */ 50 /* We possibly want to favour the dirtier blocks more when the 51 number of free blocks is low. */ 52again: 53 if (!list_empty(&c->bad_used_list) && c->nr_free_blocks > c->resv_blocks_gcbad) { 54 D1(printk(KERN_DEBUG "Picking block from bad_used_list to GC next\n")); 55 nextlist = &c->bad_used_list; 56 } else if (n < 50 && !list_empty(&c->erasable_list)) { 57 /* Note that most of them will have gone directly to be erased. 58 So don't favour the erasable_list _too_ much. */ 59 D1(printk(KERN_DEBUG "Picking block from erasable_list to GC next\n")); 60 nextlist = &c->erasable_list; 61 } else if (n < 110 && !list_empty(&c->very_dirty_list)) { 62 /* Most of the time, pick one off the very_dirty list */ 63 D1(printk(KERN_DEBUG "Picking block from very_dirty_list to GC next\n")); 64 nextlist = &c->very_dirty_list; 65 } else if (n < 126 && !list_empty(&c->dirty_list)) { 66 D1(printk(KERN_DEBUG "Picking block from dirty_list to GC next\n")); 67 nextlist = &c->dirty_list; 68 } else if (!list_empty(&c->clean_list)) { 69 D1(printk(KERN_DEBUG "Picking block from clean_list to GC next\n")); 70 nextlist = &c->clean_list; 71 } else if (!list_empty(&c->dirty_list)) { 72 D1(printk(KERN_DEBUG "Picking block from dirty_list to GC next (clean_list was empty)\n")); 73 74 nextlist = &c->dirty_list; 75 } else if (!list_empty(&c->very_dirty_list)) { 76 D1(printk(KERN_DEBUG "Picking block from very_dirty_list to GC next (clean_list and dirty_list were empty)\n")); 77 nextlist = &c->very_dirty_list; 78 } else if (!list_empty(&c->erasable_list)) { 79 D1(printk(KERN_DEBUG "Picking block from erasable_list to GC next (clean_list and {very_,}dirty_list were empty)\n")); 80 81 nextlist = &c->erasable_list; 82 } else if (!list_empty(&c->erasable_pending_wbuf_list)) { 83 /* There are blocks are wating for the wbuf sync */ 84 D1(printk(KERN_DEBUG "Synching wbuf in order to reuse erasable_pending_wbuf_list blocks\n")); 85 spin_unlock(&c->erase_completion_lock); 86 jffs2_flush_wbuf_pad(c); 87 spin_lock(&c->erase_completion_lock); 88 goto again; 89 } else { 90 /* Eep. All were empty */ 91 D1(printk(KERN_NOTICE "jffs2: No clean, dirty _or_ erasable blocks to GC from! Where are they all?\n")); 92 return NULL; 93 } 94 95 ret = list_entry(nextlist->next, struct jffs2_eraseblock, list); 96 list_del(&ret->list); 97 c->gcblock = ret; 98 ret->gc_node = ret->first_node; 99 if (!ret->gc_node) { 100 printk(KERN_WARNING "Eep. ret->gc_node for block at 0x%08x is NULL\n", ret->offset); 101 BUG(); 102 } 103 104 /* Have we accidentally picked a clean block with wasted space ? */ 105 if (ret->wasted_size) { 106 D1(printk(KERN_DEBUG "Converting wasted_size %08x to dirty_size\n", ret->wasted_size)); 107 ret->dirty_size += ret->wasted_size; 108 c->wasted_size -= ret->wasted_size; 109 c->dirty_size += ret->wasted_size; 110 ret->wasted_size = 0; 111 } 112 113 return ret; 114} 115 116/* jffs2_garbage_collect_pass 117 * Make a single attempt to progress GC. Move one node, and possibly 118 * start erasing one eraseblock. 119 */ 120int jffs2_garbage_collect_pass(struct jffs2_sb_info *c) 121{ 122 struct jffs2_inode_info *f; 123 struct jffs2_inode_cache *ic; 124 struct jffs2_eraseblock *jeb; 125 struct jffs2_raw_node_ref *raw; 126 uint32_t gcblock_dirty; 127 int ret = 0, inum, nlink; 128 int xattr = 0; 129 130 if (mutex_lock_interruptible(&c->alloc_sem)) 131 return -EINTR; 132 133 for (;;) { 134 spin_lock(&c->erase_completion_lock); 135 if (!c->unchecked_size) 136 break; 137 138 /* We can't start doing GC yet. We haven't finished checking 139 the node CRCs etc. Do it now. */ 140 141 /* checked_ino is protected by the alloc_sem */ 142 if (c->checked_ino > c->highest_ino && xattr) { 143 printk(KERN_CRIT "Checked all inodes but still 0x%x bytes of unchecked space?\n", 144 c->unchecked_size); 145 jffs2_dbg_dump_block_lists_nolock(c); 146 spin_unlock(&c->erase_completion_lock); 147 mutex_unlock(&c->alloc_sem); 148 return -ENOSPC; 149 } 150 151 spin_unlock(&c->erase_completion_lock); 152 153 if (!xattr) 154 xattr = jffs2_verify_xattr(c); 155 156 spin_lock(&c->inocache_lock); 157 158 ic = jffs2_get_ino_cache(c, c->checked_ino++); 159 160 if (!ic) { 161 spin_unlock(&c->inocache_lock); 162 continue; 163 } 164 165 if (!ic->pino_nlink) { 166 D1(printk(KERN_DEBUG "Skipping check of ino #%d with nlink/pino zero\n", 167 ic->ino)); 168 spin_unlock(&c->inocache_lock); 169 jffs2_xattr_delete_inode(c, ic); 170 continue; 171 } 172 switch(ic->state) { 173 case INO_STATE_CHECKEDABSENT: 174 case INO_STATE_PRESENT: 175 D1(printk(KERN_DEBUG "Skipping ino #%u already checked\n", ic->ino)); 176 spin_unlock(&c->inocache_lock); 177 continue; 178 179 case INO_STATE_GC: 180 case INO_STATE_CHECKING: 181 printk(KERN_WARNING "Inode #%u is in state %d during CRC check phase!\n", ic->ino, ic->state); 182 spin_unlock(&c->inocache_lock); 183 BUG(); 184 185 case INO_STATE_READING: 186 /* We need to wait for it to finish, lest we move on 187 and trigger the BUG() above while we haven't yet 188 finished checking all its nodes */ 189 D1(printk(KERN_DEBUG "Waiting for ino #%u to finish reading\n", ic->ino)); 190 /* We need to come back again for the _same_ inode. We've 191 made no progress in this case, but that should be OK */ 192 c->checked_ino--; 193 194 mutex_unlock(&c->alloc_sem); 195 sleep_on_spinunlock(&c->inocache_wq, &c->inocache_lock); 196 return 0; 197 198 default: 199 BUG(); 200 201 case INO_STATE_UNCHECKED: 202 ; 203 } 204 ic->state = INO_STATE_CHECKING; 205 spin_unlock(&c->inocache_lock); 206 207 D1(printk(KERN_DEBUG "jffs2_garbage_collect_pass() triggering inode scan of ino#%u\n", ic->ino)); 208 209 ret = jffs2_do_crccheck_inode(c, ic); 210 if (ret) 211 printk(KERN_WARNING "Returned error for crccheck of ino #%u. Expect badness...\n", ic->ino); 212 213 jffs2_set_inocache_state(c, ic, INO_STATE_CHECKEDABSENT); 214 mutex_unlock(&c->alloc_sem); 215 return ret; 216 } 217 218 /* If there are any blocks which need erasing, erase them now */ 219 if (!list_empty(&c->erase_complete_list) || 220 !list_empty(&c->erase_pending_list)) { 221 spin_unlock(&c->erase_completion_lock); 222 D1(printk(KERN_DEBUG "jffs2_garbage_collect_pass() erasing pending blocks\n")); 223 if (jffs2_erase_pending_blocks(c, 1)) { 224 mutex_unlock(&c->alloc_sem); 225 return 0; 226 } 227 D1(printk(KERN_DEBUG "No progress from erasing blocks; doing GC anyway\n")); 228 spin_lock(&c->erase_completion_lock); 229 } 230 231 /* First, work out which block we're garbage-collecting */ 232 jeb = c->gcblock; 233 234 if (!jeb) 235 jeb = jffs2_find_gc_block(c); 236 237 if (!jeb) { 238 /* Couldn't find a free block. But maybe we can just erase one and make 'progress'? */ 239 if (c->nr_erasing_blocks) { 240 spin_unlock(&c->erase_completion_lock); 241 mutex_unlock(&c->alloc_sem); 242 return -EAGAIN; 243 } 244 D1(printk(KERN_NOTICE "jffs2: Couldn't find erase block to garbage collect!\n")); 245 spin_unlock(&c->erase_completion_lock); 246 mutex_unlock(&c->alloc_sem); 247 return -EIO; 248 } 249 250 D1(printk(KERN_DEBUG "GC from block %08x, used_size %08x, dirty_size %08x, free_size %08x\n", jeb->offset, jeb->used_size, jeb->dirty_size, jeb->free_size)); 251 D1(if (c->nextblock) 252 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)); 253 254 if (!jeb->used_size) { 255 mutex_unlock(&c->alloc_sem); 256 goto eraseit; 257 } 258 259 raw = jeb->gc_node; 260 gcblock_dirty = jeb->dirty_size; 261 262 while(ref_obsolete(raw)) { 263 D1(printk(KERN_DEBUG "Node at 0x%08x is obsolete... skipping\n", ref_offset(raw))); 264 raw = ref_next(raw); 265 if (unlikely(!raw)) { 266 printk(KERN_WARNING "eep. End of raw list while still supposedly nodes to GC\n"); 267 printk(KERN_WARNING "erase block at 0x%08x. free_size 0x%08x, dirty_size 0x%08x, used_size 0x%08x\n", 268 jeb->offset, jeb->free_size, jeb->dirty_size, jeb->used_size); 269 jeb->gc_node = raw; 270 spin_unlock(&c->erase_completion_lock); 271 mutex_unlock(&c->alloc_sem); 272 BUG(); 273 } 274 } 275 jeb->gc_node = raw; 276 277 D1(printk(KERN_DEBUG "Going to garbage collect node at 0x%08x\n", ref_offset(raw))); 278 279 if (!raw->next_in_ino) { 280 /* Inode-less node. Clean marker, snapshot or something like that */ 281 spin_unlock(&c->erase_completion_lock); 282 if (ref_flags(raw) == REF_PRISTINE) { 283 /* It's an unknown node with JFFS2_FEATURE_RWCOMPAT_COPY */ 284 jffs2_garbage_collect_pristine(c, NULL, raw); 285 } else { 286 /* Just mark it obsolete */ 287 jffs2_mark_node_obsolete(c, raw); 288 } 289 mutex_unlock(&c->alloc_sem); 290 goto eraseit_lock; 291 } 292 293 ic = jffs2_raw_ref_to_ic(raw); 294 295#ifdef CONFIG_JFFS2_FS_XATTR 296 /* When 'ic' refers xattr_datum/xattr_ref, this node is GCed as xattr. 297 * We can decide whether this node is inode or xattr by ic->class. */ 298 if (ic->class == RAWNODE_CLASS_XATTR_DATUM 299 || ic->class == RAWNODE_CLASS_XATTR_REF) { 300 spin_unlock(&c->erase_completion_lock); 301 302 if (ic->class == RAWNODE_CLASS_XATTR_DATUM) { 303 ret = jffs2_garbage_collect_xattr_datum(c, (struct jffs2_xattr_datum *)ic, raw); 304 } else { 305 ret = jffs2_garbage_collect_xattr_ref(c, (struct jffs2_xattr_ref *)ic, raw); 306 } 307 goto test_gcnode; 308 } 309#endif 310 311 /* We need to hold the inocache. Either the erase_completion_lock or 312 the inocache_lock are sufficient; we trade down since the inocache_lock 313 causes less contention. */ 314 spin_lock(&c->inocache_lock); 315 316 spin_unlock(&c->erase_completion_lock); 317 318 D1(printk(KERN_DEBUG "jffs2_garbage_collect_pass collecting from block @0x%08x. Node @0x%08x(%d), ino #%u\n", jeb->offset, ref_offset(raw), ref_flags(raw), ic->ino)); 319 320 /* Three possibilities: 321 1. Inode is already in-core. We must iget it and do proper 322 updating to its fragtree, etc. 323 2. Inode is not in-core, node is REF_PRISTINE. We lock the 324 inocache to prevent a read_inode(), copy the node intact. 325 3. Inode is not in-core, node is not pristine. We must iget() 326 and take the slow path. 327 */ 328 329 switch(ic->state) { 330 case INO_STATE_CHECKEDABSENT: 331 /* It's been checked, but it's not currently in-core. 332 We can just copy any pristine nodes, but have 333 to prevent anyone else from doing read_inode() while 334 we're at it, so we set the state accordingly */ 335 if (ref_flags(raw) == REF_PRISTINE) 336 ic->state = INO_STATE_GC; 337 else { 338 D1(printk(KERN_DEBUG "Ino #%u is absent but node not REF_PRISTINE. Reading.\n", 339 ic->ino)); 340 } 341 break; 342 343 case INO_STATE_PRESENT: 344 /* It's in-core. GC must iget() it. */ 345 break; 346 347 case INO_STATE_UNCHECKED: 348 case INO_STATE_CHECKING: 349 case INO_STATE_GC: 350 /* Should never happen. We should have finished checking 351 by the time we actually start doing any GC, and since 352 we're holding the alloc_sem, no other garbage collection 353 can happen. 354 */ 355 printk(KERN_CRIT "Inode #%u already in state %d in jffs2_garbage_collect_pass()!\n", 356 ic->ino, ic->state); 357 mutex_unlock(&c->alloc_sem); 358 spin_unlock(&c->inocache_lock); 359 BUG(); 360 361 case INO_STATE_READING: 362 /* Someone's currently trying to read it. We must wait for 363 them to finish and then go through the full iget() route 364 to do the GC. However, sometimes read_inode() needs to get 365 the alloc_sem() (for marking nodes invalid) so we must 366 drop the alloc_sem before sleeping. */ 367 368 mutex_unlock(&c->alloc_sem); 369 D1(printk(KERN_DEBUG "jffs2_garbage_collect_pass() waiting for ino #%u in state %d\n", 370 ic->ino, ic->state)); 371 sleep_on_spinunlock(&c->inocache_wq, &c->inocache_lock); 372 /* And because we dropped the alloc_sem we must start again from the 373 beginning. Ponder chance of livelock here -- we're returning success 374 without actually making any progress. 375 376 Q: What are the chances that the inode is back in INO_STATE_READING 377 again by the time we next enter this function? And that this happens 378 enough times to cause a real delay? 379 380 A: Small enough that I don't care :) 381 */ 382 return 0; 383 } 384 385 /* OK. Now if the inode is in state INO_STATE_GC, we are going to copy the 386 node intact, and we don't have to muck about with the fragtree etc. 387 because we know it's not in-core. If it _was_ in-core, we go through 388 all the iget() crap anyway */ 389 390 if (ic->state == INO_STATE_GC) { 391 spin_unlock(&c->inocache_lock); 392 393 ret = jffs2_garbage_collect_pristine(c, ic, raw); 394 395 spin_lock(&c->inocache_lock); 396 ic->state = INO_STATE_CHECKEDABSENT; 397 wake_up(&c->inocache_wq); 398 399 if (ret != -EBADFD) { 400 spin_unlock(&c->inocache_lock); 401 goto test_gcnode; 402 } 403 404 /* Fall through if it wanted us to, with inocache_lock held */ 405 } 406 407 /* Prevent the fairly unlikely race where the gcblock is 408 entirely obsoleted by the final close of a file which had 409 the only valid nodes in the block, followed by erasure, 410 followed by freeing of the ic because the erased block(s) 411 held _all_ the nodes of that inode.... never been seen but 412 it's vaguely possible. */ 413 414 inum = ic->ino; 415 nlink = ic->pino_nlink; 416 spin_unlock(&c->inocache_lock); 417 418 f = jffs2_gc_fetch_inode(c, inum, !nlink); 419 if (IS_ERR(f)) { 420 ret = PTR_ERR(f); 421 goto release_sem; 422 } 423 if (!f) { 424 ret = 0; 425 goto release_sem; 426 } 427 428 ret = jffs2_garbage_collect_live(c, jeb, raw, f); 429 430 jffs2_gc_release_inode(c, f); 431 432 test_gcnode: 433 if (jeb->dirty_size == gcblock_dirty && !ref_obsolete(jeb->gc_node)) { 434 /* Eep. This really should never happen. GC is broken */ 435 printk(KERN_ERR "Error garbage collecting node at %08x!\n", ref_offset(jeb->gc_node)); 436 ret = -ENOSPC; 437 } 438 release_sem: 439 mutex_unlock(&c->alloc_sem); 440 441 eraseit_lock: 442 /* If we've finished this block, start it erasing */ 443 spin_lock(&c->erase_completion_lock); 444 445 eraseit: 446 if (c->gcblock && !c->gcblock->used_size) { 447 D1(printk(KERN_DEBUG "Block at 0x%08x completely obsoleted by GC. Moving to erase_pending_list\n", c->gcblock->offset)); 448 /* We're GC'ing an empty block? */ 449 list_add_tail(&c->gcblock->list, &c->erase_pending_list); 450 c->gcblock = NULL; 451 c->nr_erasing_blocks++; 452 jffs2_garbage_collect_trigger(c); 453 } 454 spin_unlock(&c->erase_completion_lock); 455 456 return ret; 457} 458 459static int jffs2_garbage_collect_live(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb, 460 struct jffs2_raw_node_ref *raw, struct jffs2_inode_info *f) 461{ 462 struct jffs2_node_frag *frag; 463 struct jffs2_full_dnode *fn = NULL; 464 struct jffs2_full_dirent *fd; 465 uint32_t start = 0, end = 0, nrfrags = 0; 466 int ret = 0; 467 468 mutex_lock(&f->sem); 469 470 /* Now we have the lock for this inode. Check that it's still the one at the head 471 of the list. */ 472 473 spin_lock(&c->erase_completion_lock); 474 475 if (c->gcblock != jeb) { 476 spin_unlock(&c->erase_completion_lock); 477 D1(printk(KERN_DEBUG "GC block is no longer gcblock. Restart\n")); 478 goto upnout; 479 } 480 if (ref_obsolete(raw)) { 481 spin_unlock(&c->erase_completion_lock); 482 D1(printk(KERN_DEBUG "node to be GC'd was obsoleted in the meantime.\n")); 483 /* They'll call again */ 484 goto upnout; 485 } 486 spin_unlock(&c->erase_completion_lock); 487 488 /* OK. Looks safe. And nobody can get us now because we have the semaphore. Move the block */ 489 if (f->metadata && f->metadata->raw == raw) { 490 fn = f->metadata; 491 ret = jffs2_garbage_collect_metadata(c, jeb, f, fn); 492 goto upnout; 493 } 494 495 for (frag = frag_first(&f->fragtree); frag; frag = frag_next(frag)) { 496 if (frag->node && frag->node->raw == raw) { 497 fn = frag->node; 498 end = frag->ofs + frag->size; 499 if (!nrfrags++) 500 start = frag->ofs; 501 if (nrfrags == frag->node->frags) 502 break; /* We've found them all */ 503 } 504 } 505 if (fn) { 506 if (ref_flags(raw) == REF_PRISTINE) { 507 ret = jffs2_garbage_collect_pristine(c, f->inocache, raw); 508 if (!ret) { 509 /* Urgh. Return it sensibly. */ 510 frag->node->raw = f->inocache->nodes; 511 } 512 if (ret != -EBADFD) 513 goto upnout; 514 } 515 /* We found a datanode. Do the GC */ 516 if((start >> PAGE_CACHE_SHIFT) < ((end-1) >> PAGE_CACHE_SHIFT)) { 517 /* It crosses a page boundary. Therefore, it must be a hole. */ 518 ret = jffs2_garbage_collect_hole(c, jeb, f, fn, start, end); 519 } else { 520 /* It could still be a hole. But we GC the page this way anyway */ 521 ret = jffs2_garbage_collect_dnode(c, jeb, f, fn, start, end); 522 } 523 goto upnout; 524 } 525 526 /* Wasn't a dnode. Try dirent */ 527 for (fd = f->dents; fd; fd=fd->next) { 528 if (fd->raw == raw) 529 break; 530 } 531 532 if (fd && fd->ino) { 533 ret = jffs2_garbage_collect_dirent(c, jeb, f, fd); 534 } else if (fd) { 535 ret = jffs2_garbage_collect_deletion_dirent(c, jeb, f, fd); 536 } else { 537 printk(KERN_WARNING "Raw node at 0x%08x wasn't in node lists for ino #%u\n", 538 ref_offset(raw), f->inocache->ino); 539 if (ref_obsolete(raw)) { 540 printk(KERN_WARNING "But it's obsolete so we don't mind too much\n"); 541 } else { 542 jffs2_dbg_dump_node(c, ref_offset(raw)); 543 BUG(); 544 } 545 } 546 upnout: 547 mutex_unlock(&f->sem); 548 549 return ret; 550} 551 552static int jffs2_garbage_collect_pristine(struct jffs2_sb_info *c, 553 struct jffs2_inode_cache *ic, 554 struct jffs2_raw_node_ref *raw) 555{ 556 union jffs2_node_union *node; 557 size_t retlen; 558 int ret; 559 uint32_t phys_ofs, alloclen; 560 uint32_t crc, rawlen; 561 int retried = 0; 562 563 D1(printk(KERN_DEBUG "Going to GC REF_PRISTINE node at 0x%08x\n", ref_offset(raw))); 564 565 alloclen = rawlen = ref_totlen(c, c->gcblock, raw); 566 567 /* Ask for a small amount of space (or the totlen if smaller) because we 568 don't want to force wastage of the end of a block if splitting would 569 work. */ 570 if (ic && alloclen > sizeof(struct jffs2_raw_inode) + JFFS2_MIN_DATA_LEN) 571 alloclen = sizeof(struct jffs2_raw_inode) + JFFS2_MIN_DATA_LEN; 572 573 ret = jffs2_reserve_space_gc(c, alloclen, &alloclen, rawlen); 574 /* 'rawlen' is not the exact summary size; it is only an upper estimation */ 575 576 if (ret) 577 return ret; 578 579 if (alloclen < rawlen) { 580 /* Doesn't fit untouched. We'll go the old route and split it */ 581 return -EBADFD; 582 } 583 584 node = kmalloc(rawlen, GFP_KERNEL); 585 if (!node) 586 return -ENOMEM; 587 588 ret = jffs2_flash_read(c, ref_offset(raw), rawlen, &retlen, (char *)node); 589 if (!ret && retlen != rawlen) 590 ret = -EIO; 591 if (ret) 592 goto out_node; 593 594 crc = crc32(0, node, sizeof(struct jffs2_unknown_node)-4); 595 if (je32_to_cpu(node->u.hdr_crc) != crc) { 596 printk(KERN_WARNING "Header CRC failed on REF_PRISTINE node at 0x%08x: Read 0x%08x, calculated 0x%08x\n", 597 ref_offset(raw), je32_to_cpu(node->u.hdr_crc), crc); 598 goto bail; 599 } 600 601 switch(je16_to_cpu(node->u.nodetype)) { 602 case JFFS2_NODETYPE_INODE: 603 crc = crc32(0, node, sizeof(node->i)-8); 604 if (je32_to_cpu(node->i.node_crc) != crc) { 605 printk(KERN_WARNING "Node CRC failed on REF_PRISTINE data node at 0x%08x: Read 0x%08x, calculated 0x%08x\n", 606 ref_offset(raw), je32_to_cpu(node->i.node_crc), crc); 607 goto bail; 608 } 609 610 if (je32_to_cpu(node->i.dsize)) { 611 crc = crc32(0, node->i.data, je32_to_cpu(node->i.csize)); 612 if (je32_to_cpu(node->i.data_crc) != crc) { 613 printk(KERN_WARNING "Data CRC failed on REF_PRISTINE data node at 0x%08x: Read 0x%08x, calculated 0x%08x\n", 614 ref_offset(raw), je32_to_cpu(node->i.data_crc), crc); 615 goto bail; 616 } 617 } 618 break; 619 620 case JFFS2_NODETYPE_DIRENT: 621 crc = crc32(0, node, sizeof(node->d)-8); 622 if (je32_to_cpu(node->d.node_crc) != crc) { 623 printk(KERN_WARNING "Node CRC failed on REF_PRISTINE dirent node at 0x%08x: Read 0x%08x, calculated 0x%08x\n", 624 ref_offset(raw), je32_to_cpu(node->d.node_crc), crc); 625 goto bail; 626 } 627 628 if (strnlen(node->d.name, node->d.nsize) != node->d.nsize) { 629 printk(KERN_WARNING "Name in dirent node at 0x%08x contains zeroes\n", ref_offset(raw)); 630 goto bail; 631 } 632 633 if (node->d.nsize) { 634 crc = crc32(0, node->d.name, node->d.nsize); 635 if (je32_to_cpu(node->d.name_crc) != crc) { 636 printk(KERN_WARNING "Name CRC failed on REF_PRISTINE dirent node at 0x%08x: Read 0x%08x, calculated 0x%08x\n", 637 ref_offset(raw), je32_to_cpu(node->d.name_crc), crc); 638 goto bail; 639 } 640 } 641 break; 642 default: 643 /* If it's inode-less, we don't _know_ what it is. Just copy it intact */ 644 if (ic) { 645 printk(KERN_WARNING "Unknown node type for REF_PRISTINE node at 0x%08x: 0x%04x\n", 646 ref_offset(raw), je16_to_cpu(node->u.nodetype)); 647 goto bail; 648 } 649 } 650 651 /* OK, all the CRCs are good; this node can just be copied as-is. */ 652 retry: 653 phys_ofs = write_ofs(c); 654 655 ret = jffs2_flash_write(c, phys_ofs, rawlen, &retlen, (char *)node); 656 657 if (ret || (retlen != rawlen)) { 658 printk(KERN_NOTICE "Write of %d bytes at 0x%08x failed. returned %d, retlen %zd\n", 659 rawlen, phys_ofs, ret, retlen); 660 if (retlen) { 661 jffs2_add_physical_node_ref(c, phys_ofs | REF_OBSOLETE, rawlen, NULL); 662 } else { 663 printk(KERN_NOTICE "Not marking the space at 0x%08x as dirty because the flash driver returned retlen zero\n", phys_ofs); 664 } 665 if (!retried) { 666 /* Try to reallocate space and retry */ 667 uint32_t dummy; 668 struct jffs2_eraseblock *jeb = &c->blocks[phys_ofs / c->sector_size]; 669 670 retried = 1; 671 672 D1(printk(KERN_DEBUG "Retrying failed write of REF_PRISTINE node.\n")); 673 674 jffs2_dbg_acct_sanity_check(c,jeb); 675 jffs2_dbg_acct_paranoia_check(c, jeb); 676 677 ret = jffs2_reserve_space_gc(c, rawlen, &dummy, rawlen); 678 /* this is not the exact summary size of it, 679 it is only an upper estimation */ 680 681 if (!ret) { 682 D1(printk(KERN_DEBUG "Allocated space at 0x%08x to retry failed write.\n", phys_ofs)); 683 684 jffs2_dbg_acct_sanity_check(c,jeb); 685 jffs2_dbg_acct_paranoia_check(c, jeb); 686 687 goto retry; 688 } 689 D1(printk(KERN_DEBUG "Failed to allocate space to retry failed write: %d!\n", ret)); 690 } 691 692 if (!ret) 693 ret = -EIO; 694 goto out_node; 695 } 696 jffs2_add_physical_node_ref(c, phys_ofs | REF_PRISTINE, rawlen, ic); 697 698 jffs2_mark_node_obsolete(c, raw); 699 D1(printk(KERN_DEBUG "WHEEE! GC REF_PRISTINE node at 0x%08x succeeded\n", ref_offset(raw))); 700 701 out_node: 702 kfree(node); 703 return ret; 704 bail: 705 ret = -EBADFD; 706 goto out_node; 707} 708 709static int jffs2_garbage_collect_metadata(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb, 710 struct jffs2_inode_info *f, struct jffs2_full_dnode *fn) 711{ 712 struct jffs2_full_dnode *new_fn; 713 struct jffs2_raw_inode ri; 714 struct jffs2_node_frag *last_frag; 715 union jffs2_device_node dev; 716 char *mdata = NULL; 717 int mdatalen = 0; 718 uint32_t alloclen, ilen; 719 int ret; 720 721 if (S_ISBLK(JFFS2_F_I_MODE(f)) || 722 S_ISCHR(JFFS2_F_I_MODE(f)) ) { 723 /* For these, we don't actually need to read the old node */ 724 mdatalen = jffs2_encode_dev(&dev, JFFS2_F_I_RDEV(f)); 725 mdata = (char *)&dev; 726 D1(printk(KERN_DEBUG "jffs2_garbage_collect_metadata(): Writing %d bytes of kdev_t\n", mdatalen)); 727 } else if (S_ISLNK(JFFS2_F_I_MODE(f))) { 728 mdatalen = fn->size; 729 mdata = kmalloc(fn->size, GFP_KERNEL); 730 if (!mdata) { 731 printk(KERN_WARNING "kmalloc of mdata failed in jffs2_garbage_collect_metadata()\n"); 732 return -ENOMEM; 733 } 734 ret = jffs2_read_dnode(c, f, fn, mdata, 0, mdatalen); 735 if (ret) { 736 printk(KERN_WARNING "read of old metadata failed in jffs2_garbage_collect_metadata(): %d\n", ret); 737 kfree(mdata); 738 return ret; 739 } 740 D1(printk(KERN_DEBUG "jffs2_garbage_collect_metadata(): Writing %d bites of symlink target\n", mdatalen)); 741 742 } 743 744 ret = jffs2_reserve_space_gc(c, sizeof(ri) + mdatalen, &alloclen, 745 JFFS2_SUMMARY_INODE_SIZE); 746 if (ret) { 747 printk(KERN_WARNING "jffs2_reserve_space_gc of %zd bytes for garbage_collect_metadata failed: %d\n", 748 sizeof(ri)+ mdatalen, ret); 749 goto out; 750 } 751 752 last_frag = frag_last(&f->fragtree); 753 if (last_frag) 754 /* Fetch the inode length from the fragtree rather then 755 * from i_size since i_size may have not been updated yet */ 756 ilen = last_frag->ofs + last_frag->size; 757 else 758 ilen = JFFS2_F_I_SIZE(f); 759 760 memset(&ri, 0, sizeof(ri)); 761 ri.magic = cpu_to_je16(JFFS2_MAGIC_BITMASK); 762 ri.nodetype = cpu_to_je16(JFFS2_NODETYPE_INODE); 763 ri.totlen = cpu_to_je32(sizeof(ri) + mdatalen); 764 ri.hdr_crc = cpu_to_je32(crc32(0, &ri, sizeof(struct jffs2_unknown_node)-4)); 765 766 ri.ino = cpu_to_je32(f->inocache->ino); 767 ri.version = cpu_to_je32(++f->highest_version); 768 ri.mode = cpu_to_jemode(JFFS2_F_I_MODE(f)); 769 ri.uid = cpu_to_je16(JFFS2_F_I_UID(f)); 770 ri.gid = cpu_to_je16(JFFS2_F_I_GID(f)); 771 ri.isize = cpu_to_je32(ilen); 772 ri.atime = cpu_to_je32(JFFS2_F_I_ATIME(f)); 773 ri.ctime = cpu_to_je32(JFFS2_F_I_CTIME(f)); 774 ri.mtime = cpu_to_je32(JFFS2_F_I_MTIME(f)); 775 ri.offset = cpu_to_je32(0); 776 ri.csize = cpu_to_je32(mdatalen); 777 ri.dsize = cpu_to_je32(mdatalen); 778 ri.compr = JFFS2_COMPR_NONE; 779 ri.node_crc = cpu_to_je32(crc32(0, &ri, sizeof(ri)-8)); 780 ri.data_crc = cpu_to_je32(crc32(0, mdata, mdatalen)); 781 782 new_fn = jffs2_write_dnode(c, f, &ri, mdata, mdatalen, ALLOC_GC); 783 784 if (IS_ERR(new_fn)) { 785 printk(KERN_WARNING "Error writing new dnode: %ld\n", PTR_ERR(new_fn)); 786 ret = PTR_ERR(new_fn); 787 goto out; 788 } 789 jffs2_mark_node_obsolete(c, fn->raw); 790 jffs2_free_full_dnode(fn); 791 f->metadata = new_fn; 792 out: 793 if (S_ISLNK(JFFS2_F_I_MODE(f))) 794 kfree(mdata); 795 return ret; 796} 797 798static int jffs2_garbage_collect_dirent(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb, 799 struct jffs2_inode_info *f, struct jffs2_full_dirent *fd) 800{ 801 struct jffs2_full_dirent *new_fd; 802 struct jffs2_raw_dirent rd; 803 uint32_t alloclen; 804 int ret; 805 806 rd.magic = cpu_to_je16(JFFS2_MAGIC_BITMASK); 807 rd.nodetype = cpu_to_je16(JFFS2_NODETYPE_DIRENT); 808 rd.nsize = strlen(fd->name); 809 rd.totlen = cpu_to_je32(sizeof(rd) + rd.nsize); 810 rd.hdr_crc = cpu_to_je32(crc32(0, &rd, sizeof(struct jffs2_unknown_node)-4)); 811 812 rd.pino = cpu_to_je32(f->inocache->ino); 813 rd.version = cpu_to_je32(++f->highest_version); 814 rd.ino = cpu_to_je32(fd->ino); 815 /* If the times on this inode were set by explicit utime() they can be different, 816 so refrain from splatting them. */ 817 if (JFFS2_F_I_MTIME(f) == JFFS2_F_I_CTIME(f)) 818 rd.mctime = cpu_to_je32(JFFS2_F_I_MTIME(f)); 819 else 820 rd.mctime = cpu_to_je32(0); 821 rd.type = fd->type; 822 rd.node_crc = cpu_to_je32(crc32(0, &rd, sizeof(rd)-8)); 823 rd.name_crc = cpu_to_je32(crc32(0, fd->name, rd.nsize)); 824 825 ret = jffs2_reserve_space_gc(c, sizeof(rd)+rd.nsize, &alloclen, 826 JFFS2_SUMMARY_DIRENT_SIZE(rd.nsize)); 827 if (ret) { 828 printk(KERN_WARNING "jffs2_reserve_space_gc of %zd bytes for garbage_collect_dirent failed: %d\n", 829 sizeof(rd)+rd.nsize, ret); 830 return ret; 831 } 832 new_fd = jffs2_write_dirent(c, f, &rd, fd->name, rd.nsize, ALLOC_GC); 833 834 if (IS_ERR(new_fd)) { 835 printk(KERN_WARNING "jffs2_write_dirent in garbage_collect_dirent failed: %ld\n", PTR_ERR(new_fd)); 836 return PTR_ERR(new_fd); 837 } 838 jffs2_add_fd_to_list(c, new_fd, &f->dents); 839 return 0; 840} 841 842static int jffs2_garbage_collect_deletion_dirent(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb, 843 struct jffs2_inode_info *f, struct jffs2_full_dirent *fd) 844{ 845 struct jffs2_full_dirent **fdp = &f->dents; 846 int found = 0; 847 848 /* On a medium where we can't actually mark nodes obsolete 849 pernamently, such as NAND flash, we need to work out 850 whether this deletion dirent is still needed to actively 851 delete a 'real' dirent with the same name that's still 852 somewhere else on the flash. */ 853 if (!jffs2_can_mark_obsolete(c)) { 854 struct jffs2_raw_dirent *rd; 855 struct jffs2_raw_node_ref *raw; 856 int ret; 857 size_t retlen; 858 int name_len = strlen(fd->name); 859 uint32_t name_crc = crc32(0, fd->name, name_len); 860 uint32_t rawlen = ref_totlen(c, jeb, fd->raw); 861 862 rd = kmalloc(rawlen, GFP_KERNEL); 863 if (!rd) 864 return -ENOMEM; 865 866 /* Prevent the erase code from nicking the obsolete node refs while 867 we're looking at them. I really don't like this extra lock but 868 can't see any alternative. Suggestions on a postcard to... */ 869 mutex_lock(&c->erase_free_sem); 870 871 for (raw = f->inocache->nodes; raw != (void *)f->inocache; raw = raw->next_in_ino) { 872 873 cond_resched(); 874 875 /* We only care about obsolete ones */ 876 if (!(ref_obsolete(raw))) 877 continue; 878 879 /* Any dirent with the same name is going to have the same length... */ 880 if (ref_totlen(c, NULL, raw) != rawlen) 881 continue; 882 883 /* Doesn't matter if there's one in the same erase block. We're going to 884 delete it too at the same time. */ 885 if (SECTOR_ADDR(raw->flash_offset) == SECTOR_ADDR(fd->raw->flash_offset)) 886 continue; 887 888 D1(printk(KERN_DEBUG "Check potential deletion dirent at %08x\n", ref_offset(raw))); 889 890 /* This is an obsolete node belonging to the same directory, and it's of the right 891 length. We need to take a closer look...*/ 892 ret = jffs2_flash_read(c, ref_offset(raw), rawlen, &retlen, (char *)rd); 893 if (ret) { 894 printk(KERN_WARNING "jffs2_g_c_deletion_dirent(): Read error (%d) reading obsolete node at %08x\n", ret, ref_offset(raw)); 895 /* If we can't read it, we don't need to continue to obsolete it. Continue */ 896 continue; 897 } 898 if (retlen != rawlen) { 899 printk(KERN_WARNING "jffs2_g_c_deletion_dirent(): Short read (%zd not %u) reading header from obsolete node at %08x\n", 900 retlen, rawlen, ref_offset(raw)); 901 continue; 902 } 903 904 if (je16_to_cpu(rd->nodetype) != JFFS2_NODETYPE_DIRENT) 905 continue; 906 907 /* If the name CRC doesn't match, skip */ 908 if (je32_to_cpu(rd->name_crc) != name_crc) 909 continue; 910 911 /* If the name length doesn't match, or it's another deletion dirent, skip */ 912 if (rd->nsize != name_len || !je32_to_cpu(rd->ino)) 913 continue; 914 915 /* OK, check the actual name now */ 916 if (memcmp(rd->name, fd->name, name_len)) 917 continue; 918 919 /* OK. The name really does match. There really is still an older node on 920 the flash which our deletion dirent obsoletes. So we have to write out 921 a new deletion dirent to replace it */ 922 mutex_unlock(&c->erase_free_sem); 923 924 D1(printk(KERN_DEBUG "Deletion dirent at %08x still obsoletes real dirent \"%s\" at %08x for ino #%u\n", 925 ref_offset(fd->raw), fd->name, ref_offset(raw), je32_to_cpu(rd->ino))); 926 kfree(rd); 927 928 return jffs2_garbage_collect_dirent(c, jeb, f, fd); 929 } 930 931 mutex_unlock(&c->erase_free_sem); 932 kfree(rd); 933 } 934 935 936 /* No need for it any more. Just mark it obsolete and remove it from the list */ 937 while (*fdp) { 938 if ((*fdp) == fd) { 939 found = 1; 940 *fdp = fd->next; 941 break; 942 } 943 fdp = &(*fdp)->next; 944 } 945 if (!found) { 946 printk(KERN_WARNING "Deletion dirent \"%s\" not found in list for ino #%u\n", fd->name, f->inocache->ino); 947 } 948 jffs2_mark_node_obsolete(c, fd->raw); 949 jffs2_free_full_dirent(fd); 950 return 0; 951} 952 953static int jffs2_garbage_collect_hole(struct jffs2_sb_info *c, struct jffs2_eraseblock *jeb, 954 struct jffs2_inode_info *f, struct jffs2_full_dnode *fn, 955 uint32_t start, uint32_t end) 956{ 957 struct jffs2_raw_inode ri; 958 struct jffs2_node_frag *frag; 959 struct jffs2_full_dnode *new_fn; 960 uint32_t alloclen, ilen; 961 int ret; 962 963 D1(printk(KERN_DEBUG "Writing replacement hole node for ino #%u from offset 0x%x to 0x%x\n", 964 f->inocache->ino, start, end)); 965 966 memset(&ri, 0, sizeof(ri)); 967 968 if(fn->frags > 1) { 969 size_t readlen; 970 uint32_t crc; 971 /* It's partially obsoleted by a later write. So we have to 972 write it out again with the _same_ version as before */ 973 ret = jffs2_flash_read(c, ref_offset(fn->raw), sizeof(ri), &readlen, (char *)&ri); 974 if (readlen != sizeof(ri) || ret) { 975 printk(KERN_WARNING "Node read failed in jffs2_garbage_collect_hole. Ret %d, retlen %zd. Data will be lost by writing new hole node\n", ret, readlen); 976 goto fill; 977 } 978 if (je16_to_cpu(ri.nodetype) != JFFS2_NODETYPE_INODE) { 979 printk(KERN_WARNING "jffs2_garbage_collect_hole: Node at 0x%08x had node type 0x%04x instead of JFFS2_NODETYPE_INODE(0x%04x)\n", 980 ref_offset(fn->raw), 981 je16_to_cpu(ri.nodetype), JFFS2_NODETYPE_INODE); 982 return -EIO; 983 } 984 if (je32_to_cpu(ri.totlen) != sizeof(ri)) { 985 printk(KERN_WARNING "jffs2_garbage_collect_hole: Node at 0x%08x had totlen 0x%x instead of expected 0x%zx\n", 986 ref_offset(fn->raw), 987 je32_to_cpu(ri.totlen), sizeof(ri)); 988 return -EIO; 989 } 990 crc = crc32(0, &ri, sizeof(ri)-8); 991 if (crc != je32_to_cpu(ri.node_crc)) { 992 printk(KERN_WARNING "jffs2_garbage_collect_hole: Node at 0x%08x had CRC 0x%08x which doesn't match calculated CRC 0x%08x\n", 993 ref_offset(fn->raw), 994 je32_to_cpu(ri.node_crc), crc); 995 printk(KERN_WARNING "Data in the range 0x%08x to 0x%08x of inode #%u will be lost\n", 996 start, end, f->inocache->ino); 997 goto fill; 998 } 999 if (ri.compr != JFFS2_COMPR_ZERO) { 1000 printk(KERN_WARNING "jffs2_garbage_collect_hole: Node 0x%08x wasn't a hole node!\n", ref_offset(fn->raw)); 1001 printk(KERN_WARNING "Data in the range 0x%08x to 0x%08x of inode #%u will be lost\n", 1002 start, end, f->inocache->ino); 1003 goto fill; 1004 } 1005 } else { 1006 fill: 1007 ri.magic = cpu_to_je16(JFFS2_MAGIC_BITMASK); 1008 ri.nodetype = cpu_to_je16(JFFS2_NODETYPE_INODE); 1009 ri.totlen = cpu_to_je32(sizeof(ri)); 1010 ri.hdr_crc = cpu_to_je32(crc32(0, &ri, sizeof(struct jffs2_unknown_node)-4)); 1011 1012 ri.ino = cpu_to_je32(f->inocache->ino); 1013 ri.version = cpu_to_je32(++f->highest_version); 1014 ri.offset = cpu_to_je32(start); 1015 ri.dsize = cpu_to_je32(end - start); 1016 ri.csize = cpu_to_je32(0); 1017 ri.compr = JFFS2_COMPR_ZERO; 1018 } 1019 1020 frag = frag_last(&f->fragtree); 1021 if (frag) 1022 /* Fetch the inode length from the fragtree rather then 1023 * from i_size since i_size may have not been updated yet */ 1024 ilen = frag->ofs + frag->size; 1025 else 1026 ilen = JFFS2_F_I_SIZE(f); 1027 1028 ri.mode = cpu_to_jemode(JFFS2_F_I_MODE(f)); 1029 ri.uid = cpu_to_je16(JFFS2_F_I_UID(f)); 1030 ri.gid = cpu_to_je16(JFFS2_F_I_GID(f)); 1031 ri.isize = cpu_to_je32(ilen); 1032 ri.atime = cpu_to_je32(JFFS2_F_I_ATIME(f)); 1033 ri.ctime = cpu_to_je32(JFFS2_F_I_CTIME(f)); 1034 ri.mtime = cpu_to_je32(JFFS2_F_I_MTIME(f)); 1035 ri.data_crc = cpu_to_je32(0); 1036 ri.node_crc = cpu_to_je32(crc32(0, &ri, sizeof(ri)-8)); 1037 1038 ret = jffs2_reserve_space_gc(c, sizeof(ri), &alloclen, 1039 JFFS2_SUMMARY_INODE_SIZE); 1040 if (ret) { 1041 printk(KERN_WARNING "jffs2_reserve_space_gc of %zd bytes for garbage_collect_hole failed: %d\n", 1042 sizeof(ri), ret); 1043 return ret; 1044 } 1045 new_fn = jffs2_write_dnode(c, f, &ri, NULL, 0, ALLOC_GC); 1046 1047 if (IS_ERR(new_fn)) { 1048 printk(KERN_WARNING "Error writing new hole node: %ld\n", PTR_ERR(new_fn)); 1049 return PTR_ERR(new_fn); 1050 } 1051 if (je32_to_cpu(ri.version) == f->highest_version) { 1052 jffs2_add_full_dnode_to_inode(c, f, new_fn); 1053 if (f->metadata) { 1054 jffs2_mark_node_obsolete(c, f->metadata->raw); 1055 jffs2_free_full_dnode(f->metadata); 1056 f->metadata = NULL; 1057 } 1058 return 0; 1059 } 1060 1061 /* 1062 * We should only get here in the case where the node we are 1063 * replacing had more than one frag, so we kept the same version 1064 * number as before. (Except in case of error -- see 'goto fill;' 1065 * above.) 1066 */ 1067 D1(if(unlikely(fn->frags <= 1)) { 1068 printk(KERN_WARNING "jffs2_garbage_collect_hole: Replacing fn with %d frag(s) but new ver %d != highest_version %d of ino #%d\n", 1069 fn->frags, je32_to_cpu(ri.version), f->highest_version, 1070 je32_to_cpu(ri.ino)); 1071 }); 1072 1073 /* This is a partially-overlapped hole node. Mark it REF_NORMAL not REF_PRISTINE */ 1074 mark_ref_normal(new_fn->raw); 1075 1076 for (frag = jffs2_lookup_node_frag(&f->fragtree, fn->ofs); 1077 frag; frag = frag_next(frag)) { 1078 if (frag->ofs > fn->size + fn->ofs) 1079 break; 1080 if (frag->node == fn) { 1081 frag->node = new_fn; 1082 new_fn->frags++; 1083 fn->frags--; 1084 } 1085 } 1086 if (fn->frags) { 1087 printk(KERN_WARNING "jffs2_garbage_collect_hole: Old node still has frags!\n"); 1088 BUG(); 1089 } 1090 if (!new_fn->frags) { 1091 printk(KERN_WARNING "jffs2_garbage_collect_hole: New node has no frags!\n"); 1092 BUG(); 1093 } 1094 1095 jffs2_mark_node_obsolete(c, fn->raw); 1096 jffs2_free_full_dnode(fn); 1097 1098 return 0; 1099} 1100 1101static int jffs2_garbage_collect_dnode(struct jffs2_sb_info *c, struct jffs2_eraseblock *orig_jeb, 1102 struct jffs2_inode_info *f, struct jffs2_full_dnode *fn, 1103 uint32_t start, uint32_t end) 1104{ 1105 struct jffs2_full_dnode *new_fn; 1106 struct jffs2_raw_inode ri; 1107 uint32_t alloclen, offset, orig_end, orig_start; 1108 int ret = 0; 1109 unsigned char *comprbuf = NULL, *writebuf; 1110 unsigned long pg; 1111 unsigned char *pg_ptr; 1112 1113 memset(&ri, 0, sizeof(ri)); 1114 1115 D1(printk(KERN_DEBUG "Writing replacement dnode for ino #%u from offset 0x%x to 0x%x\n", 1116 f->inocache->ino, start, end)); 1117 1118 orig_end = end; 1119 orig_start = start; 1120 1121 if (c->nr_free_blocks + c->nr_erasing_blocks > c->resv_blocks_gcmerge) { 1122 /* Attempt to do some merging. But only expand to cover logically 1123 adjacent frags if the block containing them is already considered 1124 to be dirty. Otherwise we end up with GC just going round in 1125 circles dirtying the nodes it already wrote out, especially 1126 on NAND where we have small eraseblocks and hence a much higher 1127 chance of nodes having to be split to cross boundaries. */ 1128 1129 struct jffs2_node_frag *frag; 1130 uint32_t min, max; 1131 1132 min = start & ~(PAGE_CACHE_SIZE-1); 1133 max = min + PAGE_CACHE_SIZE; 1134 1135 frag = jffs2_lookup_node_frag(&f->fragtree, start); 1136 1137 /* BUG_ON(!frag) but that'll happen anyway... */ 1138 1139 BUG_ON(frag->ofs != start); 1140 1141 /* First grow down... */ 1142 while((frag = frag_prev(frag)) && frag->ofs >= min) { 1143 1144 /* If the previous frag doesn't even reach the beginning, there's 1145 excessive fragmentation. Just merge. */ 1146 if (frag->ofs > min) { 1147 D1(printk(KERN_DEBUG "Expanding down to cover partial frag (0x%x-0x%x)\n", 1148 frag->ofs, frag->ofs+frag->size)); 1149 start = frag->ofs; 1150 continue; 1151 } 1152 /* OK. This frag holds the first byte of the page. */ 1153 if (!frag->node || !frag->node->raw) { 1154 D1(printk(KERN_DEBUG "First frag in page is hole (0x%x-0x%x). Not expanding down.\n", 1155 frag->ofs, frag->ofs+frag->size)); 1156 break; 1157 } else { 1158 1159 /* OK, it's a frag which extends to the beginning of the page. Does it live 1160 in a block which is still considered clean? If so, don't obsolete it. 1161 If not, cover it anyway. */ 1162 1163 struct jffs2_raw_node_ref *raw = frag->node->raw; 1164 struct jffs2_eraseblock *jeb; 1165 1166 jeb = &c->blocks[raw->flash_offset / c->sector_size]; 1167 1168 if (jeb == c->gcblock) { 1169 D1(printk(KERN_DEBUG "Expanding down to cover frag (0x%x-0x%x) in gcblock at %08x\n", 1170 frag->ofs, frag->ofs+frag->size, ref_offset(raw))); 1171 start = frag->ofs; 1172 break; 1173 } 1174 if (!ISDIRTY(jeb->dirty_size + jeb->wasted_size)) { 1175 D1(printk(KERN_DEBUG "Not expanding down to cover frag (0x%x-0x%x) in clean block %08x\n", 1176 frag->ofs, frag->ofs+frag->size, jeb->offset)); 1177 break; 1178 } 1179 1180 D1(printk(KERN_DEBUG "Expanding down to cover frag (0x%x-0x%x) in dirty block %08x\n", 1181 frag->ofs, frag->ofs+frag->size, jeb->offset)); 1182 start = frag->ofs; 1183 break; 1184 } 1185 } 1186 1187 /* ... then up */ 1188 1189 /* Find last frag which is actually part of the node we're to GC. */ 1190 frag = jffs2_lookup_node_frag(&f->fragtree, end-1); 1191 1192 while((frag = frag_next(frag)) && frag->ofs+frag->size <= max) { 1193 1194 /* If the previous frag doesn't even reach the beginning, there's lots 1195 of fragmentation. Just merge. */ 1196 if (frag->ofs+frag->size < max) { 1197 D1(printk(KERN_DEBUG "Expanding up to cover partial frag (0x%x-0x%x)\n", 1198 frag->ofs, frag->ofs+frag->size)); 1199 end = frag->ofs + frag->size; 1200 continue; 1201 } 1202 1203 if (!frag->node || !frag->node->raw) { 1204 D1(printk(KERN_DEBUG "Last frag in page is hole (0x%x-0x%x). Not expanding up.\n", 1205 frag->ofs, frag->ofs+frag->size)); 1206 break; 1207 } else { 1208 1209 /* OK, it's a frag which extends to the beginning of the page. Does it live 1210 in a block which is still considered clean? If so, don't obsolete it. 1211 If not, cover it anyway. */ 1212 1213 struct jffs2_raw_node_ref *raw = frag->node->raw; 1214 struct jffs2_eraseblock *jeb; 1215 1216 jeb = &c->blocks[raw->flash_offset / c->sector_size]; 1217 1218 if (jeb == c->gcblock) { 1219 D1(printk(KERN_DEBUG "Expanding up to cover frag (0x%x-0x%x) in gcblock at %08x\n", 1220 frag->ofs, frag->ofs+frag->size, ref_offset(raw))); 1221 end = frag->ofs + frag->size; 1222 break; 1223 } 1224 if (!ISDIRTY(jeb->dirty_size + jeb->wasted_size)) { 1225 D1(printk(KERN_DEBUG "Not expanding up to cover frag (0x%x-0x%x) in clean block %08x\n", 1226 frag->ofs, frag->ofs+frag->size, jeb->offset)); 1227 break; 1228 } 1229 1230 D1(printk(KERN_DEBUG "Expanding up to cover frag (0x%x-0x%x) in dirty block %08x\n", 1231 frag->ofs, frag->ofs+frag->size, jeb->offset)); 1232 end = frag->ofs + frag->size; 1233 break; 1234 } 1235 } 1236 D1(printk(KERN_DEBUG "Expanded dnode to write from (0x%x-0x%x) to (0x%x-0x%x)\n", 1237 orig_start, orig_end, start, end)); 1238 1239 D1(BUG_ON(end > frag_last(&f->fragtree)->ofs + frag_last(&f->fragtree)->size)); 1240 BUG_ON(end < orig_end); 1241 BUG_ON(start > orig_start); 1242 } 1243 1244 /* First, use readpage() to read the appropriate page into the page cache */ 1245 /* Q: What happens if we actually try to GC the _same_ page for which commit_write() 1246 * triggered garbage collection in the first place? 1247 * A: I _think_ it's OK. read_cache_page shouldn't deadlock, we'll write out the 1248 * page OK. We'll actually write it out again in commit_write, which is a little 1249 * suboptimal, but at least we're correct. 1250 */ 1251 pg_ptr = jffs2_gc_fetch_page(c, f, start, &pg); 1252 1253 if (IS_ERR(pg_ptr)) { 1254 printk(KERN_WARNING "read_cache_page() returned error: %ld\n", PTR_ERR(pg_ptr)); 1255 return PTR_ERR(pg_ptr); 1256 } 1257 1258 offset = start; 1259 while(offset < orig_end) { 1260 uint32_t datalen; 1261 uint32_t cdatalen; 1262 uint16_t comprtype = JFFS2_COMPR_NONE; 1263 1264 ret = jffs2_reserve_space_gc(c, sizeof(ri) + JFFS2_MIN_DATA_LEN, 1265 &alloclen, JFFS2_SUMMARY_INODE_SIZE); 1266 1267 if (ret) { 1268 printk(KERN_WARNING "jffs2_reserve_space_gc of %zd bytes for garbage_collect_dnode failed: %d\n", 1269 sizeof(ri)+ JFFS2_MIN_DATA_LEN, ret); 1270 break; 1271 } 1272 cdatalen = min_t(uint32_t, alloclen - sizeof(ri), end - offset); 1273 datalen = end - offset; 1274 1275 writebuf = pg_ptr + (offset & (PAGE_CACHE_SIZE -1)); 1276 1277 comprtype = jffs2_compress(c, f, writebuf, &comprbuf, &datalen, &cdatalen); 1278 1279 ri.magic = cpu_to_je16(JFFS2_MAGIC_BITMASK); 1280 ri.nodetype = cpu_to_je16(JFFS2_NODETYPE_INODE); 1281 ri.totlen = cpu_to_je32(sizeof(ri) + cdatalen); 1282 ri.hdr_crc = cpu_to_je32(crc32(0, &ri, sizeof(struct jffs2_unknown_node)-4)); 1283 1284 ri.ino = cpu_to_je32(f->inocache->ino); 1285 ri.version = cpu_to_je32(++f->highest_version); 1286 ri.mode = cpu_to_jemode(JFFS2_F_I_MODE(f)); 1287 ri.uid = cpu_to_je16(JFFS2_F_I_UID(f)); 1288 ri.gid = cpu_to_je16(JFFS2_F_I_GID(f)); 1289 ri.isize = cpu_to_je32(JFFS2_F_I_SIZE(f)); 1290 ri.atime = cpu_to_je32(JFFS2_F_I_ATIME(f)); 1291 ri.ctime = cpu_to_je32(JFFS2_F_I_CTIME(f)); 1292 ri.mtime = cpu_to_je32(JFFS2_F_I_MTIME(f)); 1293 ri.offset = cpu_to_je32(offset); 1294 ri.csize = cpu_to_je32(cdatalen); 1295 ri.dsize = cpu_to_je32(datalen); 1296 ri.compr = comprtype & 0xff; 1297 ri.usercompr = (comprtype >> 8) & 0xff; 1298 ri.node_crc = cpu_to_je32(crc32(0, &ri, sizeof(ri)-8)); 1299 ri.data_crc = cpu_to_je32(crc32(0, comprbuf, cdatalen)); 1300 1301 new_fn = jffs2_write_dnode(c, f, &ri, comprbuf, cdatalen, ALLOC_GC); 1302 1303 jffs2_free_comprbuf(comprbuf, writebuf); 1304 1305 if (IS_ERR(new_fn)) { 1306 printk(KERN_WARNING "Error writing new dnode: %ld\n", PTR_ERR(new_fn)); 1307 ret = PTR_ERR(new_fn); 1308 break; 1309 } 1310 ret = jffs2_add_full_dnode_to_inode(c, f, new_fn); 1311 offset += datalen; 1312 if (f->metadata) { 1313 jffs2_mark_node_obsolete(c, f->metadata->raw); 1314 jffs2_free_full_dnode(f->metadata); 1315 f->metadata = NULL; 1316 } 1317 } 1318 1319 jffs2_gc_release_page(c, pg_ptr, &pg); 1320 return ret; 1321} 1322