readinode.c 43.3 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3
/*
 * JFFS2 -- Journalling Flash File System, Version 2.
 *
4
 * Copyright © 2001-2007 Red Hat, Inc.
L
Linus Torvalds 已提交
5 6 7 8 9 10 11 12
 *
 * Created by David Woodhouse <dwmw2@infradead.org>
 *
 * For licensing information, see the file 'LICENCE' in this directory.
 *
 */

#include <linux/kernel.h>
13
#include <linux/sched.h>
L
Linus Torvalds 已提交
14 15 16 17 18 19 20 21
#include <linux/slab.h>
#include <linux/fs.h>
#include <linux/crc32.h>
#include <linux/pagemap.h>
#include <linux/mtd/mtd.h>
#include <linux/compiler.h>
#include "nodelist.h"

22
/*
23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215
 * Check the data CRC of the node.
 *
 * Returns: 0 if the data CRC is correct;
 * 	    1 - if incorrect;
 *	    error code if an error occured.
 */
static int check_node_data(struct jffs2_sb_info *c, struct jffs2_tmp_dnode_info *tn)
{
	struct jffs2_raw_node_ref *ref = tn->fn->raw;
	int err = 0, pointed = 0;
	struct jffs2_eraseblock *jeb;
	unsigned char *buffer;
	uint32_t crc, ofs, len;
	size_t retlen;

	BUG_ON(tn->csize == 0);

	if (!jffs2_is_writebuffered(c))
		goto adj_acc;

	/* Calculate how many bytes were already checked */
	ofs = ref_offset(ref) + sizeof(struct jffs2_raw_inode);
	len = ofs % c->wbuf_pagesize;
	if (likely(len))
		len = c->wbuf_pagesize - len;

	if (len >= tn->csize) {
		dbg_readinode("no need to check node at %#08x, data length %u, data starts at %#08x - it has already been checked.\n",
			ref_offset(ref), tn->csize, ofs);
		goto adj_acc;
	}

	ofs += len;
	len = tn->csize - len;

	dbg_readinode("check node at %#08x, data length %u, partial CRC %#08x, correct CRC %#08x, data starts at %#08x, start checking from %#08x - %u bytes.\n",
		ref_offset(ref), tn->csize, tn->partial_crc, tn->data_crc, ofs - len, ofs, len);

#ifndef __ECOS
	/* TODO: instead, incapsulate point() stuff to jffs2_flash_read(),
	 * adding and jffs2_flash_read_end() interface. */
	if (c->mtd->point) {
		err = c->mtd->point(c->mtd, ofs, len, &retlen, &buffer);
		if (!err && retlen < tn->csize) {
			JFFS2_WARNING("MTD point returned len too short: %zu instead of %u.\n", retlen, tn->csize);
			c->mtd->unpoint(c->mtd, buffer, ofs, len);
		} else if (err)
			JFFS2_WARNING("MTD point failed: error code %d.\n", err);
		else
			pointed = 1; /* succefully pointed to device */
	}
#endif

	if (!pointed) {
		buffer = kmalloc(len, GFP_KERNEL);
		if (unlikely(!buffer))
			return -ENOMEM;

		/* TODO: this is very frequent pattern, make it a separate
		 * routine */
		err = jffs2_flash_read(c, ofs, len, &retlen, buffer);
		if (err) {
			JFFS2_ERROR("can not read %d bytes from 0x%08x, error code: %d.\n", len, ofs, err);
			goto free_out;
		}

		if (retlen != len) {
			JFFS2_ERROR("short read at %#08x: %zd instead of %d.\n", ofs, retlen, len);
			err = -EIO;
			goto free_out;
		}
	}

	/* Continue calculating CRC */
	crc = crc32(tn->partial_crc, buffer, len);
	if(!pointed)
		kfree(buffer);
#ifndef __ECOS
	else
		c->mtd->unpoint(c->mtd, buffer, ofs, len);
#endif

	if (crc != tn->data_crc) {
		JFFS2_NOTICE("wrong data CRC in data node at 0x%08x: read %#08x, calculated %#08x.\n",
			ofs, tn->data_crc, crc);
		return 1;
	}

adj_acc:
	jeb = &c->blocks[ref->flash_offset / c->sector_size];
	len = ref_totlen(c, jeb, ref);
	/* If it should be REF_NORMAL, it'll get marked as such when
	   we build the fragtree, shortly. No need to worry about GC
	   moving it while it's marked REF_PRISTINE -- GC won't happen
	   till we've finished checking every inode anyway. */
	ref->flash_offset |= REF_PRISTINE;
	/*
	 * Mark the node as having been checked and fix the
	 * accounting accordingly.
	 */
	spin_lock(&c->erase_completion_lock);
	jeb->used_size += len;
	jeb->unchecked_size -= len;
	c->used_size += len;
	c->unchecked_size -= len;
	jffs2_dbg_acct_paranoia_check_nolock(c, jeb);
	spin_unlock(&c->erase_completion_lock);

	return 0;

free_out:
	if(!pointed)
		kfree(buffer);
#ifndef __ECOS
	else
		c->mtd->unpoint(c->mtd, buffer, ofs, len);
#endif
	return err;
}

/*
 * Helper function for jffs2_add_older_frag_to_fragtree().
 *
 * Checks the node if we are in the checking stage.
 */
static int check_tn_node(struct jffs2_sb_info *c, struct jffs2_tmp_dnode_info *tn)
{
	int ret;

	BUG_ON(ref_obsolete(tn->fn->raw));

	/* We only check the data CRC of unchecked nodes */
	if (ref_flags(tn->fn->raw) != REF_UNCHECKED)
		return 0;

	dbg_readinode("check node %#04x-%#04x, phys offs %#08x\n",
		      tn->fn->ofs, tn->fn->ofs + tn->fn->size, ref_offset(tn->fn->raw));

	ret = check_node_data(c, tn);
	if (unlikely(ret < 0)) {
		JFFS2_ERROR("check_node_data() returned error: %d.\n",
			ret);
	} else if (unlikely(ret > 0)) {
		dbg_readinode("CRC error, mark it obsolete.\n");
		jffs2_mark_node_obsolete(c, tn->fn->raw);
	}

	return ret;
}

static struct jffs2_tmp_dnode_info *jffs2_lookup_tn(struct rb_root *tn_root, uint32_t offset)
{
	struct rb_node *next;
	struct jffs2_tmp_dnode_info *tn = NULL;

	dbg_readinode("root %p, offset %d\n", tn_root, offset);

	next = tn_root->rb_node;

	while (next) {
		tn = rb_entry(next, struct jffs2_tmp_dnode_info, rb);

		if (tn->fn->ofs < offset)
			next = tn->rb.rb_right;
		else if (tn->fn->ofs >= offset)
			next = tn->rb.rb_left;
		else
			break;
	}

	return tn;
}


static void jffs2_kill_tn(struct jffs2_sb_info *c, struct jffs2_tmp_dnode_info *tn)
{
	jffs2_mark_node_obsolete(c, tn->fn->raw);
	jffs2_free_full_dnode(tn->fn);
	jffs2_free_tmp_dnode_info(tn);
}
/*
 * This function is used when we read an inode. Data nodes arrive in
 * arbitrary order -- they may be older or newer than the nodes which
 * are already in the tree. Where overlaps occur, the older node can
 * be discarded as long as the newer passes the CRC check. We don't
 * bother to keep track of holes in this rbtree, and neither do we deal
 * with frags -- we can have multiple entries starting at the same
 * offset, and the one with the smallest length will come first in the
 * ordering.
 *
 * Returns 0 if the node was inserted
 *         1 if the node is obsolete (because we can't mark it so yet)
 *         < 0 an if error occurred
L
Linus Torvalds 已提交
216
 */
217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280
static int jffs2_add_tn_to_tree(struct jffs2_sb_info *c,
				struct jffs2_readinode_info *rii,
				struct jffs2_tmp_dnode_info *tn)
{
	uint32_t fn_end = tn->fn->ofs + tn->fn->size;
	struct jffs2_tmp_dnode_info *insert_point = NULL, *this;

	dbg_readinode("insert fragment %#04x-%#04x, ver %u\n", tn->fn->ofs, fn_end, tn->version);

	/* If a node has zero dsize, we only have to keep if it if it might be the
	   node with highest version -- i.e. the one which will end up as f->metadata.
	   Note that such nodes won't be REF_UNCHECKED since there are no data to
	   check anyway. */
	if (!tn->fn->size) {
		if (rii->mdata_tn) {
			/* We had a candidate mdata node already */
			dbg_readinode("kill old mdata with ver %d\n", rii->mdata_tn->version);
			jffs2_kill_tn(c, rii->mdata_tn);
		}
		rii->mdata_tn = tn;
		dbg_readinode("keep new mdata with ver %d\n", tn->version);
		return 0;
	}

	/* Find the earliest node which _may_ be relevant to this one */
	this = jffs2_lookup_tn(&rii->tn_root, tn->fn->ofs);
	if (!this) {
		/* First addition to empty tree. $DEITY how I love the easy cases */
		rb_link_node(&tn->rb, NULL, &rii->tn_root.rb_node);
		rb_insert_color(&tn->rb, &rii->tn_root);
		dbg_readinode("keep new frag\n");
		return 0;
	}

	/* If we add a new node it'll be somewhere under here. */
	insert_point = this;

	/* If the node is coincident with another at a lower address,
	   back up until the other node is found. It may be relevant */
	while (tn->overlapped)
		tn = tn_prev(tn);

	dbg_readinode("'this' found %#04x-%#04x (%s)\n", this->fn->ofs, this->fn->ofs + this->fn->size, this->fn ? "data" : "hole");

	while (this) {
		if (this->fn->ofs > fn_end)
			break;
		dbg_readinode("Ponder this ver %d, 0x%x-0x%x\n",
			      this->version, this->fn->ofs, this->fn->size);

		if (this->version == tn->version) {
			/* Version number collision means REF_PRISTINE GC. Accept either of them
			   as long as the CRC is correct. Check the one we have already...  */
			if (!check_tn_node(c, this)) {
				/* The one we already had was OK. Keep it and throw away the new one */
				dbg_readinode("Like old node. Throw away new\n");
				jffs2_kill_tn(c, tn);
				return 0;
			} else {
				/* Who cares if the new one is good; keep it for now anyway. */
				rb_replace_node(&this->rb, &tn->rb, &rii->tn_root);
				/* Same overlapping from in front and behind */
				jffs2_kill_tn(c, this);
				dbg_readinode("Like new node. Throw away old\n");
281
				goto calc_overlaps;
282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309
			}
		}
		if (this->version < tn->version &&
		    this->fn->ofs >= tn->fn->ofs &&
		    this->fn->ofs + this->fn->size <= fn_end) {
			/* New node entirely overlaps 'this' */
			if (check_tn_node(c, tn)) {
				dbg_readinode("new node bad CRC\n");
				jffs2_kill_tn(c, tn);
				return 0;
			}
			/* ... and is good. Kill 'this'... */
			rb_replace_node(&this->rb, &tn->rb, &rii->tn_root);
			jffs2_kill_tn(c, this);
			/* ... and any subsequent nodes which are also overlapped */
			this = tn_next(tn);
			while (this && this->fn->ofs + this->fn->size < fn_end) {
				struct jffs2_tmp_dnode_info *next = tn_next(this);
				if (this->version < tn->version) {
					tn_erase(this, &rii->tn_root);
					dbg_readinode("Kill overlapped ver %d, 0x%x-0x%x\n",
						      this->version, this->fn->ofs,
						      this->fn->ofs+this->fn->size);
					jffs2_kill_tn(c, this);
				}
				this = next;
			}
			dbg_readinode("Done inserting new\n");
310
			goto calc_overlaps;
311 312 313 314 315 316 317 318 319 320 321
		}
		if (this->version > tn->version &&
		    this->fn->ofs <= tn->fn->ofs &&
		    this->fn->ofs+this->fn->size >= fn_end) {
			/* New node entirely overlapped by 'this' */
			if (!check_tn_node(c, this)) {
				dbg_readinode("Good CRC on old node. Kill new\n");
				jffs2_kill_tn(c, tn);
				return 0;
			}
			/* ... but 'this' was bad. Replace it... */
322
			tn->overlapped = this->overlapped;
323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360
			rb_replace_node(&this->rb, &tn->rb, &rii->tn_root);
			dbg_readinode("Bad CRC on old overlapping node. Kill it\n");
			jffs2_kill_tn(c, this);
			return 0;
		}
		/* We want to be inserted under the last node which is
		   either at a lower offset _or_ has a smaller range */
		if (this->fn->ofs < tn->fn->ofs ||
		    (this->fn->ofs == tn->fn->ofs &&
		     this->fn->size <= tn->fn->size))
			insert_point = this;

		this = tn_next(this);
	}
	dbg_readinode("insert_point %p, ver %d, 0x%x-0x%x, ov %d\n",
		      insert_point, insert_point->version, insert_point->fn->ofs,
		      insert_point->fn->ofs+insert_point->fn->size,
		      insert_point->overlapped);
	/* We neither completely obsoleted nor were completely
	   obsoleted by an earlier node. Insert under insert_point */
	{
		struct rb_node *parent = &insert_point->rb;
		struct rb_node **link = &parent;

		while (*link) {
			parent = *link;
			insert_point = rb_entry(parent, struct jffs2_tmp_dnode_info, rb);
			if (tn->fn->ofs > insert_point->fn->ofs)
				link = &insert_point->rb.rb_right;
			else if (tn->fn->ofs < insert_point->fn->ofs ||
				 tn->fn->size < insert_point->fn->size)
				link = &insert_point->rb.rb_left;
			else
				link = &insert_point->rb.rb_right;
		}
		rb_link_node(&tn->rb, &insert_point->rb, link);
		rb_insert_color(&tn->rb, &rii->tn_root);
	}
361 362

 calc_overlaps:
363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395
	/* If there's anything behind that overlaps us, note it */
	this = tn_prev(tn);
	if (this) {
		while (1) {
			if (this->fn->ofs + this->fn->size > tn->fn->ofs) {
				dbg_readinode("Node is overlapped by %p (v %d, 0x%x-0x%x)\n",
					      this, this->version, this->fn->ofs,
					      this->fn->ofs+this->fn->size);
				tn->overlapped = 1;
				break;
			}
			if (!this->overlapped)
				break;
			this = tn_prev(this);
		}
	}

	/* If the new node overlaps anything ahead, note it */
	this = tn_next(tn);
	while (this && this->fn->ofs < fn_end) {
		this->overlapped = 1;
		dbg_readinode("Node ver %d, 0x%x-0x%x is overlapped\n",
			      this->version, this->fn->ofs,
			      this->fn->ofs+this->fn->size);
		this = tn_next(this);
	}
	return 0;
}

/* Trivial function to remove the last node in the tree. Which by definition
   has no right-hand -- so can be removed just by making its only child (if
   any) take its place under its parent. */
static void eat_last(struct rb_root *root, struct rb_node *node)
L
Linus Torvalds 已提交
396
{
397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428
	struct rb_node *parent = rb_parent(node);
	struct rb_node **link;

	/* LAST! */
	BUG_ON(node->rb_right);

	if (!parent)
		link = &root->rb_node;
	else if (node == parent->rb_left)
		link = &parent->rb_left;
	else
		link = &parent->rb_right;

	*link = node->rb_left;
	/* Colour doesn't matter now. Only the parent pointer. */
	if (node->rb_left)
		node->rb_left->rb_parent_color = node->rb_parent_color;
}

/* We put this in reverse order, so we can just use eat_last */
static void ver_insert(struct rb_root *ver_root, struct jffs2_tmp_dnode_info *tn)
{
	struct rb_node **link = &ver_root->rb_node;
	struct rb_node *parent = NULL;
	struct jffs2_tmp_dnode_info *this_tn;

	while (*link) {
		parent = *link;
		this_tn = rb_entry(parent, struct jffs2_tmp_dnode_info, rb);

		if (tn->version > this_tn->version)
			link = &parent->rb_left;
429
		else
430
			link = &parent->rb_right;
431
	}
432 433 434 435
	dbg_readinode("Link new node at %p (root is %p)\n", link, ver_root);
	rb_link_node(&tn->rb, parent, link);
	rb_insert_color(&tn->rb, ver_root);
}
436

437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486
/* Build final, normal fragtree from tn tree. It doesn't matter which order
   we add nodes to the real fragtree, as long as they don't overlap. And
   having thrown away the majority of overlapped nodes as we went, there
   really shouldn't be many sets of nodes which do overlap. If we start at
   the end, we can use the overlap markers -- we can just eat nodes which
   aren't overlapped, and when we encounter nodes which _do_ overlap we
   sort them all into a temporary tree in version order before replaying them. */
static int jffs2_build_inode_fragtree(struct jffs2_sb_info *c,
				      struct jffs2_inode_info *f,
				      struct jffs2_readinode_info *rii)
{
	struct jffs2_tmp_dnode_info *pen, *last, *this;
	struct rb_root ver_root = RB_ROOT;
	uint32_t high_ver = 0;

	if (rii->mdata_tn) {
		dbg_readinode("potential mdata is ver %d at %p\n", rii->mdata_tn->version, rii->mdata_tn);
		high_ver = rii->mdata_tn->version;
		rii->latest_ref = rii->mdata_tn->fn->raw;
	}
#ifdef JFFS2_DBG_READINODE_MESSAGES
	this = tn_last(&rii->tn_root);
	while (this) {
		dbg_readinode("tn %p ver %d range 0x%x-0x%x ov %d\n", this, this->version, this->fn->ofs,
			     this->fn->ofs+this->fn->size, this->overlapped);
		this = tn_prev(this);
	}
#endif
	pen = tn_last(&rii->tn_root);
	while ((last = pen)) {
		pen = tn_prev(last);

		eat_last(&rii->tn_root, &last->rb);
		ver_insert(&ver_root, last);

		if (unlikely(last->overlapped))
			continue;

		/* Now we have a bunch of nodes in reverse version
		   order, in the tree at ver_root. Most of the time,
		   there'll actually be only one node in the 'tree',
		   in fact. */
		this = tn_last(&ver_root);

		while (this) {
			struct jffs2_tmp_dnode_info *vers_next;
			int ret;
			vers_next = tn_prev(this);
			eat_last(&ver_root, &this->rb);
			if (check_tn_node(c, this)) {
487
				dbg_readinode("node ver %d, 0x%x-0x%x failed CRC\n",
488 489 490 491 492 493 494 495 496 497 498 499
					     this->version, this->fn->ofs,
					     this->fn->ofs+this->fn->size);
				jffs2_kill_tn(c, this);
			} else {
				if (this->version > high_ver) {
					/* Note that this is different from the other
					   highest_version, because this one is only
					   counting _valid_ nodes which could give the
					   latest inode metadata */
					high_ver = this->version;
					rii->latest_ref = this->fn->raw;
				}
500
				dbg_readinode("Add %p (v %d, 0x%x-0x%x, ov %d) to fragtree\n",
501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527
					     this, this->version, this->fn->ofs,
					     this->fn->ofs+this->fn->size, this->overlapped);

				ret = jffs2_add_full_dnode_to_inode(c, f, this->fn);
				if (ret) {
					/* Free the nodes in vers_root; let the caller
					   deal with the rest */
					JFFS2_ERROR("Add node to tree failed %d\n", ret);
					while (1) {
						vers_next = tn_prev(this);
						if (check_tn_node(c, this))
							jffs2_mark_node_obsolete(c, this->fn->raw);
						jffs2_free_full_dnode(this->fn);
						jffs2_free_tmp_dnode_info(this);
						this = vers_next;
						if (!this)
							break;
						eat_last(&ver_root, &vers_next->rb);
					}
					return ret;
				}
				jffs2_free_tmp_dnode_info(this);
			}
			this = vers_next;
		}
	}
	return 0;
528
}
L
Linus Torvalds 已提交
529

530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547
static void jffs2_free_tmp_dnode_info_list(struct rb_root *list)
{
	struct rb_node *this;
	struct jffs2_tmp_dnode_info *tn;

	this = list->rb_node;

	/* Now at bottom of tree */
	while (this) {
		if (this->rb_left)
			this = this->rb_left;
		else if (this->rb_right)
			this = this->rb_right;
		else {
			tn = rb_entry(this, struct jffs2_tmp_dnode_info, rb);
			jffs2_free_full_dnode(tn->fn);
			jffs2_free_tmp_dnode_info(tn);

548
			this = rb_parent(this);
549 550 551 552 553 554 555 556 557 558 559 560
			if (!this)
				break;

			if (this->rb_left == &tn->rb)
				this->rb_left = NULL;
			else if (this->rb_right == &tn->rb)
				this->rb_right = NULL;
			else BUG();
		}
	}
	list->rb_node = NULL;
}
L
Linus Torvalds 已提交
561

562 563 564
static void jffs2_free_full_dirent_list(struct jffs2_full_dirent *fd)
{
	struct jffs2_full_dirent *next;
565

566 567 568 569 570 571
	while (fd) {
		next = fd->next;
		jffs2_free_full_dirent(fd);
		fd = next;
	}
}
L
Linus Torvalds 已提交
572

573 574 575 576 577 578
/* Returns first valid node after 'ref'. May return 'ref' */
static struct jffs2_raw_node_ref *jffs2_first_valid_node(struct jffs2_raw_node_ref *ref)
{
	while (ref && ref->next_in_ino) {
		if (!ref_obsolete(ref))
			return ref;
579
		dbg_noderef("node at 0x%08x is obsoleted. Ignoring.\n", ref_offset(ref));
580 581 582 583
		ref = ref->next_in_ino;
	}
	return NULL;
}
L
Linus Torvalds 已提交
584

585 586 587 588 589 590 591 592
/*
 * Helper function for jffs2_get_inode_nodes().
 * It is called every time an directory entry node is found.
 *
 * Returns: 0 on succes;
 * 	    1 if the node should be marked obsolete;
 * 	    negative error code on failure.
 */
593
static inline int read_direntry(struct jffs2_sb_info *c, struct jffs2_raw_node_ref *ref,
594 595
				struct jffs2_raw_dirent *rd, size_t read,
				struct jffs2_readinode_info *rii)
596 597
{
	struct jffs2_full_dirent *fd;
598
	uint32_t crc;
599

600 601
	/* Obsoleted. This cannot happen, surely? dwmw2 20020308 */
	BUG_ON(ref_obsolete(ref));
602

603 604 605 606
	crc = crc32(0, rd, sizeof(*rd) - 8);
	if (unlikely(crc != je32_to_cpu(rd->node_crc))) {
		JFFS2_NOTICE("header CRC failed on dirent node at %#08x: read %#08x, calculated %#08x\n",
			     ref_offset(ref), je32_to_cpu(rd->node_crc), crc);
607 608
		jffs2_mark_node_obsolete(c, ref);
		return 0;
609
	}
610

611 612 613 614 615 616 617 618 619
	/* If we've never checked the CRCs on this node, check them now */
	if (ref_flags(ref) == REF_UNCHECKED) {
		struct jffs2_eraseblock *jeb;
		int len;

		/* Sanity check */
		if (unlikely(PAD((rd->nsize + sizeof(*rd))) != PAD(je32_to_cpu(rd->totlen)))) {
			JFFS2_ERROR("illegal nsize in node at %#08x: nsize %#02x, totlen %#04x\n",
				    ref_offset(ref), rd->nsize, je32_to_cpu(rd->totlen));
620 621
			jffs2_mark_node_obsolete(c, ref);
			return 0;
622 623 624 625 626 627 628 629 630 631 632 633 634 635
		}

		jeb = &c->blocks[ref->flash_offset / c->sector_size];
		len = ref_totlen(c, jeb, ref);

		spin_lock(&c->erase_completion_lock);
		jeb->used_size += len;
		jeb->unchecked_size -= len;
		c->used_size += len;
		c->unchecked_size -= len;
		ref->flash_offset = ref_offset(ref) | REF_PRISTINE;
		spin_unlock(&c->erase_completion_lock);
	}

636 637 638
	fd = jffs2_alloc_full_dirent(rd->nsize + 1);
	if (unlikely(!fd))
		return -ENOMEM;
L
Linus Torvalds 已提交
639

640 641 642 643
	fd->raw = ref;
	fd->version = je32_to_cpu(rd->version);
	fd->ino = je32_to_cpu(rd->ino);
	fd->type = rd->type;
L
Linus Torvalds 已提交
644

645 646 647
	if (fd->version > rii->highest_version)
		rii->highest_version = fd->version;

648
	/* Pick out the mctime of the latest dirent */
649 650 651
	if(fd->version > rii->mctime_ver && je32_to_cpu(rd->mctime)) {
		rii->mctime_ver = fd->version;
		rii->latest_mctime = je32_to_cpu(rd->mctime);
L
Linus Torvalds 已提交
652 653
	}

654
	/*
655 656 657 658 659 660
	 * Copy as much of the name as possible from the raw
	 * dirent we've already read from the flash.
	 */
	if (read > sizeof(*rd))
		memcpy(&fd->name[0], &rd->name[0],
		       min_t(uint32_t, rd->nsize, (read - sizeof(*rd)) ));
661

662 663 664 665 666
	/* Do we need to copy any more of the name directly from the flash? */
	if (rd->nsize + sizeof(*rd) > read) {
		/* FIXME: point() */
		int err;
		int already = read - sizeof(*rd);
667 668

		err = jffs2_flash_read(c, (ref_offset(ref)) + read,
669 670 671
				rd->nsize - already, &read, &fd->name[already]);
		if (unlikely(read != rd->nsize - already) && likely(!err))
			return -EIO;
672

673
		if (unlikely(err)) {
674
			JFFS2_ERROR("read remainder of name: error %d\n", err);
675 676
			jffs2_free_full_dirent(fd);
			return -EIO;
L
Linus Torvalds 已提交
677 678
		}
	}
679

680 681 682
	fd->nhash = full_name_hash(fd->name, rd->nsize);
	fd->next = NULL;
	fd->name[rd->nsize] = '\0';
683

684 685
	/*
	 * Wheee. We now have a complete jffs2_full_dirent structure, with
686
	 * the name in it and everything. Link it into the list
687
	 */
688
	jffs2_add_fd_to_list(c, fd, &rii->fds);
689

L
Linus Torvalds 已提交
690 691 692
	return 0;
}

693 694 695 696
/*
 * Helper function for jffs2_get_inode_nodes().
 * It is called every time an inode node is found.
 *
697
 * Returns: 0 on success;
698 699 700
 * 	    1 if the node should be marked obsolete;
 * 	    negative error code on failure.
 */
701
static inline int read_dnode(struct jffs2_sb_info *c, struct jffs2_raw_node_ref *ref,
702 703
			     struct jffs2_raw_inode *rd, int rdlen,
			     struct jffs2_readinode_info *rii)
L
Linus Torvalds 已提交
704
{
705
	struct jffs2_tmp_dnode_info *tn;
706 707
	uint32_t len, csize;
	int ret = 1;
708
	uint32_t crc;
709

710 711 712
	/* Obsoleted. This cannot happen, surely? dwmw2 20020308 */
	BUG_ON(ref_obsolete(ref));

713 714 715 716
	crc = crc32(0, rd, sizeof(*rd) - 8);
	if (unlikely(crc != je32_to_cpu(rd->node_crc))) {
		JFFS2_NOTICE("node CRC failed on dnode at %#08x: read %#08x, calculated %#08x\n",
			     ref_offset(ref), je32_to_cpu(rd->node_crc), crc);
717 718
		jffs2_mark_node_obsolete(c, ref);
		return 0;
719 720
	}

721 722
	tn = jffs2_alloc_tmp_dnode_info();
	if (!tn) {
723
		JFFS2_ERROR("failed to allocate tn (%zu bytes).\n", sizeof(*tn));
724 725 726 727 728
		return -ENOMEM;
	}

	tn->partial_crc = 0;
	csize = je32_to_cpu(rd->csize);
729

730 731
	/* If we've never checked the CRCs on this node, check them now */
	if (ref_flags(ref) == REF_UNCHECKED) {
732

733 734 735
		/* Sanity checks */
		if (unlikely(je32_to_cpu(rd->offset) > je32_to_cpu(rd->isize)) ||
		    unlikely(PAD(je32_to_cpu(rd->csize) + sizeof(*rd)) != PAD(je32_to_cpu(rd->totlen)))) {
736
				JFFS2_WARNING("inode node header CRC is corrupted at %#08x\n", ref_offset(ref));
737
				jffs2_dbg_dump_node(c, ref_offset(ref));
738
			goto free_out;
L
Linus Torvalds 已提交
739 740
		}

741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782
		if (jffs2_is_writebuffered(c) && csize != 0) {
			/* At this point we are supposed to check the data CRC
			 * of our unchecked node. But thus far, we do not
			 * know whether the node is valid or obsolete. To
			 * figure this out, we need to walk all the nodes of
			 * the inode and build the inode fragtree. We don't
			 * want to spend time checking data of nodes which may
			 * later be found to be obsolete. So we put off the full
			 * data CRC checking until we have read all the inode
			 * nodes and have started building the fragtree.
			 *
			 * The fragtree is being built starting with nodes
			 * having the highest version number, so we'll be able
			 * to detect whether a node is valid (i.e., it is not
			 * overlapped by a node with higher version) or not.
			 * And we'll be able to check only those nodes, which
			 * are not obsolete.
			 *
			 * Of course, this optimization only makes sense in case
			 * of NAND flashes (or other flashes whith
			 * !jffs2_can_mark_obsolete()), since on NOR flashes
			 * nodes are marked obsolete physically.
			 *
			 * Since NAND flashes (or other flashes with
			 * jffs2_is_writebuffered(c)) are anyway read by
			 * fractions of c->wbuf_pagesize, and we have just read
			 * the node header, it is likely that the starting part
			 * of the node data is also read when we read the
			 * header. So we don't mind to check the CRC of the
			 * starting part of the data of the node now, and check
			 * the second part later (in jffs2_check_node_data()).
			 * Of course, we will not need to re-read and re-check
			 * the NAND page which we have just read. This is why we
			 * read the whole NAND page at jffs2_get_inode_nodes(),
			 * while we needed only the node header.
			 */
			unsigned char *buf;

			/* 'buf' will point to the start of data */
			buf = (unsigned char *)rd + sizeof(*rd);
			/* len will be the read data length */
			len = min_t(uint32_t, rdlen - sizeof(*rd), csize);
783 784
			tn->partial_crc = crc32(0, buf, len);

785
			dbg_readinode("Calculates CRC (%#08x) for %d bytes, csize %d\n", tn->partial_crc, len, csize);
786 787 788

			/* If we actually calculated the whole data CRC
			 * and it is wrong, drop the node. */
789
			if (len >= csize && unlikely(tn->partial_crc != je32_to_cpu(rd->data_crc))) {
790 791
				JFFS2_NOTICE("wrong data CRC in data node at 0x%08x: read %#08x, calculated %#08x.\n",
					ref_offset(ref), tn->partial_crc, je32_to_cpu(rd->data_crc));
792
				goto free_out;
793
			}
L
Linus Torvalds 已提交
794

795 796 797 798 799 800 801 802 803
		} else if (csize == 0) {
			/*
			 * We checked the header CRC. If the node has no data, adjust
			 * the space accounting now. For other nodes this will be done
			 * later either when the node is marked obsolete or when its
			 * data is checked.
			 */
			struct jffs2_eraseblock *jeb;

804
			dbg_readinode("the node has no data.\n");
805 806 807 808 809 810 811 812
			jeb = &c->blocks[ref->flash_offset / c->sector_size];
			len = ref_totlen(c, jeb, ref);

			spin_lock(&c->erase_completion_lock);
			jeb->used_size += len;
			jeb->unchecked_size -= len;
			c->used_size += len;
			c->unchecked_size -= len;
813
			ref->flash_offset = ref_offset(ref) | REF_NORMAL;
814
			spin_unlock(&c->erase_completion_lock);
L
Linus Torvalds 已提交
815 816 817
		}
	}

818 819
	tn->fn = jffs2_alloc_full_dnode();
	if (!tn->fn) {
820
		JFFS2_ERROR("alloc fn failed\n");
821 822
		ret = -ENOMEM;
		goto free_out;
823
	}
824

825 826
	tn->version = je32_to_cpu(rd->version);
	tn->fn->ofs = je32_to_cpu(rd->offset);
827 828
	tn->data_crc = je32_to_cpu(rd->data_crc);
	tn->csize = csize;
829
	tn->fn->raw = ref;
830 831 832 833
	tn->overlapped = 0;

	if (tn->version > rii->highest_version)
		rii->highest_version = tn->version;
834

835 836
	/* There was a bug where we wrote hole nodes out with
	   csize/dsize swapped. Deal with it */
837 838
	if (rd->compr == JFFS2_COMPR_ZERO && !je32_to_cpu(rd->dsize) && csize)
		tn->fn->size = csize;
839 840 841
	else // normal case...
		tn->fn->size = je32_to_cpu(rd->dsize);

842
	dbg_readinode("dnode @%08x: ver %u, offset %#04x, dsize %#04x, csize %#04x\n",
843
		  ref_offset(ref), je32_to_cpu(rd->version), je32_to_cpu(rd->offset), je32_to_cpu(rd->dsize), csize);
844

845
	ret = jffs2_add_tn_to_tree(c, rii, tn);
L
Linus Torvalds 已提交
846

847 848 849 850 851 852 853
	if (ret) {
		jffs2_free_full_dnode(tn->fn);
	free_out:
		jffs2_free_tmp_dnode_info(tn);
		return ret;
	}
#ifdef JFFS2_DBG_READINODE_MESSAGES
854
	dbg_readinode("After adding ver %d:\n", je32_to_cpu(rd->version));
855 856 857 858 859 860 861 862
	tn = tn_first(&rii->tn_root);
	while (tn) {
		dbg_readinode("%p: v %d r 0x%x-0x%x ov %d\n",
			     tn, tn->version, tn->fn->ofs,
			     tn->fn->ofs+tn->fn->size, tn->overlapped);
		tn = tn_next(tn);
	}
#endif
L
Linus Torvalds 已提交
863 864 865
	return 0;
}

866 867 868 869
/*
 * Helper function for jffs2_get_inode_nodes().
 * It is called every time an unknown node is found.
 *
870
 * Returns: 0 on success;
871 872 873
 * 	    1 if the node should be marked obsolete;
 * 	    negative error code on failure.
 */
874
static inline int read_unknown(struct jffs2_sb_info *c, struct jffs2_raw_node_ref *ref, struct jffs2_unknown_node *un)
L
Linus Torvalds 已提交
875
{
876
	/* We don't mark unknown nodes as REF_UNCHECKED */
877 878 879 880 881 882
	if (ref_flags(ref) == REF_UNCHECKED) {
		JFFS2_ERROR("REF_UNCHECKED but unknown node at %#08x\n",
			    ref_offset(ref));
		JFFS2_ERROR("Node is {%04x,%04x,%08x,%08x}. Please report this error.\n",
                            je16_to_cpu(un->magic), je16_to_cpu(un->nodetype),
                            je32_to_cpu(un->totlen), je32_to_cpu(un->hdr_crc));
883 884
		jffs2_mark_node_obsolete(c, ref);
		return 0;
885
	}
886

887
	un->nodetype = cpu_to_je16(JFFS2_NODE_ACCURATE | je16_to_cpu(un->nodetype));
L
Linus Torvalds 已提交
888

889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911
	switch(je16_to_cpu(un->nodetype) & JFFS2_COMPAT_MASK) {

	case JFFS2_FEATURE_INCOMPAT:
		JFFS2_ERROR("unknown INCOMPAT nodetype %#04X at %#08x\n",
			    je16_to_cpu(un->nodetype), ref_offset(ref));
		/* EEP */
		BUG();
		break;

	case JFFS2_FEATURE_ROCOMPAT:
		JFFS2_ERROR("unknown ROCOMPAT nodetype %#04X at %#08x\n",
			    je16_to_cpu(un->nodetype), ref_offset(ref));
		BUG_ON(!(c->flags & JFFS2_SB_FLAG_RO));
		break;

	case JFFS2_FEATURE_RWCOMPAT_COPY:
		JFFS2_NOTICE("unknown RWCOMPAT_COPY nodetype %#04X at %#08x\n",
			     je16_to_cpu(un->nodetype), ref_offset(ref));
		break;

	case JFFS2_FEATURE_RWCOMPAT_DELETE:
		JFFS2_NOTICE("unknown RWCOMPAT_DELETE nodetype %#04X at %#08x\n",
			     je16_to_cpu(un->nodetype), ref_offset(ref));
912 913
		jffs2_mark_node_obsolete(c, ref);
		return 0;
L
Linus Torvalds 已提交
914 915
	}

916
	return 0;
L
Linus Torvalds 已提交
917 918
}

919 920 921 922 923 924 925 926
/*
 * Helper function for jffs2_get_inode_nodes().
 * The function detects whether more data should be read and reads it if yes.
 *
 * Returns: 0 on succes;
 * 	    negative error code on failure.
 */
static int read_more(struct jffs2_sb_info *c, struct jffs2_raw_node_ref *ref,
927
		     int needed_len, int *rdlen, unsigned char *buf)
928
{
929
	int err, to_read = needed_len - *rdlen;
930 931 932 933
	size_t retlen;
	uint32_t offs;

	if (jffs2_is_writebuffered(c)) {
934
		int rem = to_read % c->wbuf_pagesize;
935

936 937 938
		if (rem)
			to_read += c->wbuf_pagesize - rem;
	}
939 940 941

	/* We need to read more data */
	offs = ref_offset(ref) + *rdlen;
942

943
	dbg_readinode("read more %d bytes\n", to_read);
944

945
	err = jffs2_flash_read(c, offs, to_read, &retlen, buf + *rdlen);
946 947
	if (err) {
		JFFS2_ERROR("can not read %d bytes from 0x%08x, "
948
			"error code: %d.\n", to_read, offs, err);
949 950
		return err;
	}
951

952
	if (retlen < to_read) {
953
		JFFS2_ERROR("short read at %#08x: %zu instead of %d.\n",
954
				offs, retlen, to_read);
955 956 957
		return -EIO;
	}

958
	*rdlen += to_read;
959 960 961
	return 0;
}

962
/* Get tmp_dnode_info and full_dirent for all non-obsolete nodes associated
963 964 965 966
   with this ino. Perform a preliminary ordering on data nodes, throwing away
   those which are completely obsoleted by newer ones. The naïve approach we
   use to take of just returning them _all_ in version order will cause us to
   run out of memory in certain degenerate cases. */
967
static int jffs2_get_inode_nodes(struct jffs2_sb_info *c, struct jffs2_inode_info *f,
968
				 struct jffs2_readinode_info *rii)
L
Linus Torvalds 已提交
969
{
970
	struct jffs2_raw_node_ref *ref, *valid_ref;
971 972
	unsigned char *buf = NULL;
	union jffs2_node_union *node;
973
	size_t retlen;
974
	int len, err;
L
Linus Torvalds 已提交
975

976
	rii->mctime_ver = 0;
977

978
	dbg_readinode("ino #%u\n", f->inocache->ino);
L
Linus Torvalds 已提交
979

980 981
	/* FIXME: in case of NOR and available ->point() this
	 * needs to be fixed. */
982
	len = sizeof(union jffs2_node_union) + c->wbuf_pagesize;
983 984 985
	buf = kmalloc(len, GFP_KERNEL);
	if (!buf)
		return -ENOMEM;
986

987
	spin_lock(&c->erase_completion_lock);
988
	valid_ref = jffs2_first_valid_node(f->inocache->nodes);
989 990
	if (!valid_ref && f->inocache->ino != 1)
		JFFS2_WARNING("Eep. No valid nodes for ino #%u.\n", f->inocache->ino);
991 992 993 994 995 996 997 998 999 1000 1001 1002 1003
	while (valid_ref) {
		/* We can hold a pointer to a non-obsolete node without the spinlock,
		   but _obsolete_ nodes may disappear at any time, if the block
		   they're in gets erased. So if we mark 'ref' obsolete while we're
		   not holding the lock, it can go away immediately. For that reason,
		   we find the next valid node first, before processing 'ref'.
		*/
		ref = valid_ref;
		valid_ref = jffs2_first_valid_node(ref->next_in_ino);
		spin_unlock(&c->erase_completion_lock);

		cond_resched();

1004 1005 1006
		/*
		 * At this point we don't know the type of the node we're going
		 * to read, so we do not know the size of its header. In order
1007 1008
		 * to minimize the amount of flash IO we assume the header is
		 * of size = JFFS2_MIN_NODE_HEADER.
1009
		 */
1010
		len = JFFS2_MIN_NODE_HEADER;
1011
		if (jffs2_is_writebuffered(c)) {
1012 1013
			int end, rem;

1014
			/*
1015 1016 1017 1018 1019
			 * We are about to read JFFS2_MIN_NODE_HEADER bytes,
			 * but this flash has some minimal I/O unit. It is
			 * possible that we'll need to read more soon, so read
			 * up to the next min. I/O unit, in order not to
			 * re-read the same min. I/O unit twice.
1020
			 */
1021 1022 1023 1024 1025
			end = ref_offset(ref) + len;
			rem = end % c->wbuf_pagesize;
			if (rem)
				end += c->wbuf_pagesize - rem;
			len = end - ref_offset(ref);
1026 1027
		}

1028
		dbg_readinode("read %d bytes at %#08x(%d).\n", len, ref_offset(ref), ref_flags(ref));
1029

1030
		/* FIXME: point() */
1031
		err = jffs2_flash_read(c, ref_offset(ref), len, &retlen, buf);
1032
		if (err) {
1033 1034 1035
			JFFS2_ERROR("can not read %d bytes from 0x%08x, " "error code: %d.\n", len, ref_offset(ref), err);
			goto free_out;
		}
1036

1037
		if (retlen < len) {
1038
			JFFS2_ERROR("short read at %#08x: %zu instead of %d.\n", ref_offset(ref), retlen, len);
1039
			err = -EIO;
1040 1041
			goto free_out;
		}
1042

1043
		node = (union jffs2_node_union *)buf;
1044

1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055
		/* No need to mask in the valid bit; it shouldn't be invalid */
		if (je32_to_cpu(node->u.hdr_crc) != crc32(0, node, sizeof(node->u)-4)) {
			JFFS2_NOTICE("Node header CRC failed at %#08x. {%04x,%04x,%08x,%08x}\n",
				     ref_offset(ref), je16_to_cpu(node->u.magic),
				     je16_to_cpu(node->u.nodetype),
				     je32_to_cpu(node->u.totlen),
				     je32_to_cpu(node->u.hdr_crc));
			jffs2_dbg_dump_node(c, ref_offset(ref));
			jffs2_mark_node_obsolete(c, ref);
			goto cont;
		}
1056 1057 1058 1059
		if (je16_to_cpu(node->u.magic) != JFFS2_MAGIC_BITMASK) {
			/* Not a JFFS2 node, whinge and move on */
			JFFS2_NOTICE("Wrong magic bitmask 0x%04x in node header at %#08x.\n",
				     je16_to_cpu(node->u.magic), ref_offset(ref));
1060 1061 1062
			jffs2_mark_node_obsolete(c, ref);
			goto cont;
		}
1063

1064
		switch (je16_to_cpu(node->u.nodetype)) {
1065

1066 1067
		case JFFS2_NODETYPE_DIRENT:

1068
			if (JFFS2_MIN_NODE_HEADER < sizeof(struct jffs2_raw_dirent)) {
1069
				err = read_more(c, ref, sizeof(struct jffs2_raw_dirent), &len, buf);
1070 1071 1072
				if (unlikely(err))
					goto free_out;
			}
1073

1074 1075
			err = read_direntry(c, ref, &node->d, retlen, rii);
			if (unlikely(err))
1076
				goto free_out;
1077

L
Linus Torvalds 已提交
1078 1079
			break;

1080
		case JFFS2_NODETYPE_INODE:
1081

1082
			if (JFFS2_MIN_NODE_HEADER < sizeof(struct jffs2_raw_inode)) {
1083
				err = read_more(c, ref, sizeof(struct jffs2_raw_inode), &len, buf);
1084 1085
				if (unlikely(err))
					goto free_out;
1086
			}
L
Linus Torvalds 已提交
1087

1088 1089
			err = read_dnode(c, ref, &node->i, len, rii);
			if (unlikely(err))
1090
				goto free_out;
L
Linus Torvalds 已提交
1091

1092
			break;
L
Linus Torvalds 已提交
1093

1094
		default:
1095
			if (JFFS2_MIN_NODE_HEADER < sizeof(struct jffs2_unknown_node)) {
1096
				err = read_more(c, ref, sizeof(struct jffs2_unknown_node), &len, buf);
1097 1098
				if (unlikely(err))
					goto free_out;
1099
			}
1100

1101
			err = read_unknown(c, ref, &node->u);
1102 1103 1104 1105 1106 1107 1108
			if (err == 1) {
				jffs2_mark_node_obsolete(c, ref);
				break;
			} else if (unlikely(err))
				goto free_out;

		}
1109
	cont:
1110
		spin_lock(&c->erase_completion_lock);
L
Linus Torvalds 已提交
1111
	}
1112

1113
	spin_unlock(&c->erase_completion_lock);
1114
	kfree(buf);
1115

1116 1117
	f->highest_version = rii->highest_version;

1118
	dbg_readinode("nodes of inode #%u were read, the highest version is %u, latest_mctime %u, mctime_ver %u.\n",
1119 1120
		      f->inocache->ino, rii->highest_version, rii->latest_mctime,
		      rii->mctime_ver);
1121 1122 1123
	return 0;

 free_out:
1124 1125 1126
	jffs2_free_tmp_dnode_info_list(&rii->tn_root);
	jffs2_free_full_dirent_list(rii->fds);
	rii->fds = NULL;
1127
	kfree(buf);
1128
	return err;
L
Linus Torvalds 已提交
1129 1130
}

1131
static int jffs2_do_read_inode_internal(struct jffs2_sb_info *c,
L
Linus Torvalds 已提交
1132 1133 1134
					struct jffs2_inode_info *f,
					struct jffs2_raw_inode *latest_node)
{
1135
	struct jffs2_readinode_info rii;
1136
	uint32_t crc, new_size;
L
Linus Torvalds 已提交
1137 1138 1139
	size_t retlen;
	int ret;

1140
	dbg_readinode("ino #%u nlink is %d\n", f->inocache->ino, f->inocache->nlink);
L
Linus Torvalds 已提交
1141

1142 1143
	memset(&rii, 0, sizeof(rii));

L
Linus Torvalds 已提交
1144
	/* Grab all nodes relevant to this ino */
1145
	ret = jffs2_get_inode_nodes(c, f, &rii);
L
Linus Torvalds 已提交
1146 1147

	if (ret) {
1148
		JFFS2_ERROR("cannot read nodes for ino %u, returned error is %d\n", f->inocache->ino, ret);
L
Linus Torvalds 已提交
1149 1150 1151 1152 1153
		if (f->inocache->state == INO_STATE_READING)
			jffs2_set_inocache_state(c, f->inocache, INO_STATE_CHECKEDABSENT);
		return ret;
	}

1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168
	ret = jffs2_build_inode_fragtree(c, f, &rii);
	if (ret) {
		JFFS2_ERROR("Failed to build final fragtree for inode #%u: error %d\n",
			    f->inocache->ino, ret);
		if (f->inocache->state == INO_STATE_READING)
			jffs2_set_inocache_state(c, f->inocache, INO_STATE_CHECKEDABSENT);
		jffs2_free_tmp_dnode_info_list(&rii.tn_root);
		/* FIXME: We could at least crc-check them all */
		if (rii.mdata_tn) {
			jffs2_free_full_dnode(rii.mdata_tn->fn);
			jffs2_free_tmp_dnode_info(rii.mdata_tn);
			rii.mdata_tn = NULL;
		}
		return ret;
	}
1169

1170 1171 1172 1173 1174 1175
	if (rii.mdata_tn) {
		if (rii.mdata_tn->fn->raw == rii.latest_ref) {
			f->metadata = rii.mdata_tn->fn;
			jffs2_free_tmp_dnode_info(rii.mdata_tn);
		} else {
			jffs2_kill_tn(c, rii.mdata_tn);
1176
		}
1177
		rii.mdata_tn = NULL;
L
Linus Torvalds 已提交
1178 1179
	}

1180 1181 1182
	f->dents = rii.fds;

	jffs2_dbg_fragtree_paranoia_check_nolock(f);
1183

1184
	if (unlikely(!rii.latest_ref)) {
L
Linus Torvalds 已提交
1185 1186
		/* No data nodes for this inode. */
		if (f->inocache->ino != 1) {
1187
			JFFS2_WARNING("no data nodes found for ino #%u\n", f->inocache->ino);
1188
			if (!rii.fds) {
L
Linus Torvalds 已提交
1189 1190 1191 1192
				if (f->inocache->state == INO_STATE_READING)
					jffs2_set_inocache_state(c, f->inocache, INO_STATE_CHECKEDABSENT);
				return -EIO;
			}
1193
			JFFS2_NOTICE("but it has children so we fake some modes for it\n");
L
Linus Torvalds 已提交
1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205
		}
		latest_node->mode = cpu_to_jemode(S_IFDIR|S_IRUGO|S_IWUSR|S_IXUGO);
		latest_node->version = cpu_to_je32(0);
		latest_node->atime = latest_node->ctime = latest_node->mtime = cpu_to_je32(0);
		latest_node->isize = cpu_to_je32(0);
		latest_node->gid = cpu_to_je16(0);
		latest_node->uid = cpu_to_je16(0);
		if (f->inocache->state == INO_STATE_READING)
			jffs2_set_inocache_state(c, f->inocache, INO_STATE_PRESENT);
		return 0;
	}

1206
	ret = jffs2_flash_read(c, ref_offset(rii.latest_ref), sizeof(*latest_node), &retlen, (void *)latest_node);
L
Linus Torvalds 已提交
1207
	if (ret || retlen != sizeof(*latest_node)) {
1208 1209
		JFFS2_ERROR("failed to read from flash: error %d, %zd of %zd bytes read\n",
			ret, retlen, sizeof(*latest_node));
L
Linus Torvalds 已提交
1210 1211 1212 1213 1214 1215 1216 1217
		/* FIXME: If this fails, there seems to be a memory leak. Find it. */
		up(&f->sem);
		jffs2_do_clear_inode(c, f);
		return ret?ret:-EIO;
	}

	crc = crc32(0, latest_node, sizeof(*latest_node)-8);
	if (crc != je32_to_cpu(latest_node->node_crc)) {
1218
		JFFS2_ERROR("CRC failed for read_inode of inode %u at physical location 0x%x\n",
1219
			f->inocache->ino, ref_offset(rii.latest_ref));
L
Linus Torvalds 已提交
1220 1221 1222 1223 1224 1225 1226
		up(&f->sem);
		jffs2_do_clear_inode(c, f);
		return -EIO;
	}

	switch(jemode_to_cpu(latest_node->mode) & S_IFMT) {
	case S_IFDIR:
1227
		if (rii.mctime_ver > je32_to_cpu(latest_node->version)) {
L
Linus Torvalds 已提交
1228 1229
			/* The times in the latest_node are actually older than
			   mctime in the latest dirent. Cheat. */
1230
			latest_node->ctime = latest_node->mtime = cpu_to_je32(rii.latest_mctime);
L
Linus Torvalds 已提交
1231 1232 1233
		}
		break;

1234

L
Linus Torvalds 已提交
1235 1236
	case S_IFREG:
		/* If it was a regular file, truncate it to the latest node's isize */
1237 1238 1239 1240 1241 1242
		new_size = jffs2_truncate_fragtree(c, &f->fragtree, je32_to_cpu(latest_node->isize));
		if (new_size != je32_to_cpu(latest_node->isize)) {
			JFFS2_WARNING("Truncating ino #%u to %d bytes failed because it only had %d bytes to start with!\n",
				      f->inocache->ino, je32_to_cpu(latest_node->isize), new_size);
			latest_node->isize = cpu_to_je32(new_size);
		}
L
Linus Torvalds 已提交
1243 1244 1245 1246 1247 1248 1249 1250 1251
		break;

	case S_IFLNK:
		/* Hack to work around broken isize in old symlink code.
		   Remove this when dwmw2 comes to his senses and stops
		   symlinks from being an entirely gratuitous special
		   case. */
		if (!je32_to_cpu(latest_node->isize))
			latest_node->isize = latest_node->dsize;
1252 1253 1254

		if (f->inocache->state != INO_STATE_CHECKING) {
			/* Symlink's inode data is the target path. Read it and
1255 1256 1257 1258
			 * keep in RAM to facilitate quick follow symlink
			 * operation. */
			f->target = kmalloc(je32_to_cpu(latest_node->csize) + 1, GFP_KERNEL);
			if (!f->target) {
1259
				JFFS2_ERROR("can't allocate %d bytes of memory for the symlink target path cache\n", je32_to_cpu(latest_node->csize));
1260 1261 1262 1263
				up(&f->sem);
				jffs2_do_clear_inode(c, f);
				return -ENOMEM;
			}
1264

1265
			ret = jffs2_flash_read(c, ref_offset(rii.latest_ref) + sizeof(*latest_node),
1266
						je32_to_cpu(latest_node->csize), &retlen, (char *)f->target);
1267

1268 1269 1270
			if (ret  || retlen != je32_to_cpu(latest_node->csize)) {
				if (retlen != je32_to_cpu(latest_node->csize))
					ret = -EIO;
1271 1272
				kfree(f->target);
				f->target = NULL;
1273 1274 1275 1276 1277
				up(&f->sem);
				jffs2_do_clear_inode(c, f);
				return -ret;
			}

1278
			f->target[je32_to_cpu(latest_node->csize)] = '\0';
1279
			dbg_readinode("symlink's target '%s' cached\n", f->target);
1280
		}
1281

L
Linus Torvalds 已提交
1282 1283 1284 1285 1286 1287 1288
		/* fall through... */

	case S_IFBLK:
	case S_IFCHR:
		/* Certain inode types should have only one data node, and it's
		   kept as the metadata node */
		if (f->metadata) {
1289
			JFFS2_ERROR("Argh. Special inode #%u with mode 0%o had metadata node\n",
L
Linus Torvalds 已提交
1290 1291 1292 1293 1294 1295
			       f->inocache->ino, jemode_to_cpu(latest_node->mode));
			up(&f->sem);
			jffs2_do_clear_inode(c, f);
			return -EIO;
		}
		if (!frag_first(&f->fragtree)) {
1296
			JFFS2_ERROR("Argh. Special inode #%u with mode 0%o has no fragments\n",
L
Linus Torvalds 已提交
1297 1298 1299 1300 1301 1302 1303
			       f->inocache->ino, jemode_to_cpu(latest_node->mode));
			up(&f->sem);
			jffs2_do_clear_inode(c, f);
			return -EIO;
		}
		/* ASSERT: f->fraglist != NULL */
		if (frag_next(frag_first(&f->fragtree))) {
1304
			JFFS2_ERROR("Argh. Special inode #%u with mode 0x%x had more than one node\n",
L
Linus Torvalds 已提交
1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322
			       f->inocache->ino, jemode_to_cpu(latest_node->mode));
			/* FIXME: Deal with it - check crc32, check for duplicate node, check times and discard the older one */
			up(&f->sem);
			jffs2_do_clear_inode(c, f);
			return -EIO;
		}
		/* OK. We're happy */
		f->metadata = frag_first(&f->fragtree)->node;
		jffs2_free_node_frag(frag_first(&f->fragtree));
		f->fragtree = RB_ROOT;
		break;
	}
	if (f->inocache->state == INO_STATE_READING)
		jffs2_set_inocache_state(c, f->inocache, INO_STATE_PRESENT);

	return 0;
}

1323
/* Scan the list of all nodes present for this ino, build map of versions, etc. */
1324
int jffs2_do_read_inode(struct jffs2_sb_info *c, struct jffs2_inode_info *f,
1325 1326
			uint32_t ino, struct jffs2_raw_inode *latest_node)
{
1327
	dbg_readinode("read inode #%u\n", ino);
1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339

 retry_inocache:
	spin_lock(&c->inocache_lock);
	f->inocache = jffs2_get_ino_cache(c, ino);

	if (f->inocache) {
		/* Check its state. We may need to wait before we can use it */
		switch(f->inocache->state) {
		case INO_STATE_UNCHECKED:
		case INO_STATE_CHECKEDABSENT:
			f->inocache->state = INO_STATE_READING;
			break;
1340

1341 1342 1343 1344 1345
		case INO_STATE_CHECKING:
		case INO_STATE_GC:
			/* If it's in either of these states, we need
			   to wait for whoever's got it to finish and
			   put it back. */
1346
			dbg_readinode("waiting for ino #%u in state %d\n", ino, f->inocache->state);
1347 1348 1349 1350 1351 1352 1353 1354
			sleep_on_spinunlock(&c->inocache_wq, &c->inocache_lock);
			goto retry_inocache;

		case INO_STATE_READING:
		case INO_STATE_PRESENT:
			/* Eep. This should never happen. It can
			happen if Linux calls read_inode() again
			before clear_inode() has finished though. */
1355
			JFFS2_ERROR("Eep. Trying to read_inode #%u when it's already in state %d!\n", ino, f->inocache->state);
1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369
			/* Fail. That's probably better than allowing it to succeed */
			f->inocache = NULL;
			break;

		default:
			BUG();
		}
	}
	spin_unlock(&c->inocache_lock);

	if (!f->inocache && ino == 1) {
		/* Special case - no root inode on medium */
		f->inocache = jffs2_alloc_inode_cache();
		if (!f->inocache) {
1370
			JFFS2_ERROR("cannot allocate inocache for root inode\n");
1371 1372
			return -ENOMEM;
		}
1373
		dbg_readinode("creating inocache for root inode\n");
1374 1375 1376 1377 1378 1379 1380
		memset(f->inocache, 0, sizeof(struct jffs2_inode_cache));
		f->inocache->ino = f->inocache->nlink = 1;
		f->inocache->nodes = (struct jffs2_raw_node_ref *)f->inocache;
		f->inocache->state = INO_STATE_READING;
		jffs2_add_ino_cache(c, f->inocache);
	}
	if (!f->inocache) {
1381
		JFFS2_ERROR("requestied to read an nonexistent ino %u\n", ino);
1382 1383 1384 1385 1386 1387 1388 1389 1390
		return -ENOENT;
	}

	return jffs2_do_read_inode_internal(c, f, latest_node);
}

int jffs2_do_crccheck_inode(struct jffs2_sb_info *c, struct jffs2_inode_cache *ic)
{
	struct jffs2_raw_inode n;
1391
	struct jffs2_inode_info *f = kzalloc(sizeof(*f), GFP_KERNEL);
1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408
	int ret;

	if (!f)
		return -ENOMEM;

	init_MUTEX_LOCKED(&f->sem);
	f->inocache = ic;

	ret = jffs2_do_read_inode_internal(c, f, &n);
	if (!ret) {
		up(&f->sem);
		jffs2_do_clear_inode(c, f);
	}
	kfree (f);
	return ret;
}

L
Linus Torvalds 已提交
1409 1410 1411 1412 1413
void jffs2_do_clear_inode(struct jffs2_sb_info *c, struct jffs2_inode_info *f)
{
	struct jffs2_full_dirent *fd, *fds;
	int deleted;

1414
	jffs2_clear_acl(f);
1415
	jffs2_xattr_delete_inode(c, f->inocache);
L
Linus Torvalds 已提交
1416 1417 1418
	down(&f->sem);
	deleted = f->inocache && !f->inocache->nlink;

1419 1420 1421
	if (f->inocache && f->inocache->state != INO_STATE_CHECKING)
		jffs2_set_inocache_state(c, f->inocache, INO_STATE_CLEARING);

L
Linus Torvalds 已提交
1422 1423 1424 1425 1426 1427 1428 1429
	if (f->metadata) {
		if (deleted)
			jffs2_mark_node_obsolete(c, f->metadata->raw);
		jffs2_free_full_dnode(f->metadata);
	}

	jffs2_kill_fragtree(&f->fragtree, deleted?c:NULL);

1430 1431 1432 1433
	if (f->target) {
		kfree(f->target);
		f->target = NULL;
	}
1434

1435 1436 1437 1438 1439
	fds = f->dents;
	while(fds) {
		fd = fds;
		fds = fd->next;
		jffs2_free_full_dirent(fd);
L
Linus Torvalds 已提交
1440 1441
	}

1442
	if (f->inocache && f->inocache->state != INO_STATE_CHECKING) {
L
Linus Torvalds 已提交
1443
		jffs2_set_inocache_state(c, f->inocache, INO_STATE_CHECKEDABSENT);
1444 1445 1446
		if (f->inocache->nodes == (void *)f->inocache)
			jffs2_del_ino_cache(c, f->inocache);
	}
L
Linus Torvalds 已提交
1447 1448 1449

	up(&f->sem);
}