recovery.c 42.9 KB
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554
/*
 * This file is part of UBIFS.
 *
 * Copyright (C) 2006-2008 Nokia Corporation
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License version 2 as published by
 * the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
 * more details.
 *
 * You should have received a copy of the GNU General Public License along with
 * this program; if not, write to the Free Software Foundation, Inc., 51
 * Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
 *
 * Authors: Adrian Hunter
 *          Artem Bityutskiy (Битюцкий Артём)
 */

/*
 * This file implements functions needed to recover from unclean un-mounts.
 * When UBIFS is mounted, it checks a flag on the master node to determine if
 * an un-mount was completed successfully. If not, the process of mounting
 * incorporates additional checking and fixing of on-flash data structures.
 * UBIFS always cleans away all remnants of an unclean un-mount, so that
 * errors do not accumulate. However UBIFS defers recovery if it is mounted
 * read-only, and the flash is not modified in that case.
 *
 * The general UBIFS approach to the recovery is that it recovers from
 * corruptions which could be caused by power cuts, but it refuses to recover
 * from corruption caused by other reasons. And UBIFS tries to distinguish
 * between these 2 reasons of corruptions and silently recover in the former
 * case and loudly complain in the latter case.
 *
 * UBIFS writes only to erased LEBs, so it writes only to the flash space
 * containing only 0xFFs. UBIFS also always writes strictly from the beginning
 * of the LEB to the end. And UBIFS assumes that the underlying flash media
 * writes in @c->max_write_size bytes at a time.
 *
 * Hence, if UBIFS finds a corrupted node at offset X, it expects only the min.
 * I/O unit corresponding to offset X to contain corrupted data, all the
 * following min. I/O units have to contain empty space (all 0xFFs). If this is
 * not true, the corruption cannot be the result of a power cut, and UBIFS
 * refuses to mount.
 */

#include <linux/crc32.h>
#include <linux/slab.h>
#include "ubifs.h"

/**
 * is_empty - determine whether a buffer is empty (contains all 0xff).
 * @buf: buffer to clean
 * @len: length of buffer
 *
 * This function returns %1 if the buffer is empty (contains all 0xff) otherwise
 * %0 is returned.
 */
static int is_empty(void *buf, int len)
{
	uint8_t *p = buf;
	int i;

	for (i = 0; i < len; i++)
		if (*p++ != 0xff)
			return 0;
	return 1;
}

/**
 * first_non_ff - find offset of the first non-0xff byte.
 * @buf: buffer to search in
 * @len: length of buffer
 *
 * This function returns offset of the first non-0xff byte in @buf or %-1 if
 * the buffer contains only 0xff bytes.
 */
static int first_non_ff(void *buf, int len)
{
	uint8_t *p = buf;
	int i;

	for (i = 0; i < len; i++)
		if (*p++ != 0xff)
			return i;
	return -1;
}

/**
 * get_master_node - get the last valid master node allowing for corruption.
 * @c: UBIFS file-system description object
 * @lnum: LEB number
 * @pbuf: buffer containing the LEB read, is returned here
 * @mst: master node, if found, is returned here
 * @cor: corruption, if found, is returned here
 *
 * This function allocates a buffer, reads the LEB into it, and finds and
 * returns the last valid master node allowing for one area of corruption.
 * The corrupt area, if there is one, must be consistent with the assumption
 * that it is the result of an unclean unmount while the master node was being
 * written. Under those circumstances, it is valid to use the previously written
 * master node.
 *
 * This function returns %0 on success and a negative error code on failure.
 */
static int get_master_node(const struct ubifs_info *c, int lnum, void **pbuf,
			   struct ubifs_mst_node **mst, void **cor)
{
	const int sz = c->mst_node_alsz;
	int err, offs, len;
	void *sbuf, *buf;

	sbuf = vmalloc(c->leb_size);
	if (!sbuf)
		return -ENOMEM;

	err = ubifs_leb_read(c, lnum, sbuf, 0, c->leb_size, 0);
	if (err && err != -EBADMSG)
		goto out_free;

	/* Find the first position that is definitely not a node */
	offs = 0;
	buf = sbuf;
	len = c->leb_size;
	while (offs + UBIFS_MST_NODE_SZ <= c->leb_size) {
		struct ubifs_ch *ch = buf;

		if (le32_to_cpu(ch->magic) != UBIFS_NODE_MAGIC)
			break;
		offs += sz;
		buf  += sz;
		len  -= sz;
	}
	/* See if there was a valid master node before that */
	if (offs) {
		int ret;

		offs -= sz;
		buf  -= sz;
		len  += sz;
		ret = ubifs_scan_a_node(c, buf, len, lnum, offs, 1);
		if (ret != SCANNED_A_NODE && offs) {
			/* Could have been corruption so check one place back */
			offs -= sz;
			buf  -= sz;
			len  += sz;
			ret = ubifs_scan_a_node(c, buf, len, lnum, offs, 1);
			if (ret != SCANNED_A_NODE)
				/*
				 * We accept only one area of corruption because
				 * we are assuming that it was caused while
				 * trying to write a master node.
				 */
				goto out_err;
		}
		if (ret == SCANNED_A_NODE) {
			struct ubifs_ch *ch = buf;

			if (ch->node_type != UBIFS_MST_NODE)
				goto out_err;
			dbg_rcvry("found a master node at %d:%d", lnum, offs);
			*mst = buf;
			offs += sz;
			buf  += sz;
			len  -= sz;
		}
	}
	/* Check for corruption */
	if (offs < c->leb_size) {
		if (!is_empty(buf, min_t(int, len, sz))) {
			*cor = buf;
			dbg_rcvry("found corruption at %d:%d", lnum, offs);
		}
		offs += sz;
		buf  += sz;
		len  -= sz;
	}
	/* Check remaining empty space */
	if (offs < c->leb_size)
		if (!is_empty(buf, len))
			goto out_err;
	*pbuf = sbuf;
	return 0;

out_err:
	err = -EINVAL;
out_free:
	vfree(sbuf);
	*mst = NULL;
	*cor = NULL;
	return err;
}

/**
 * write_rcvrd_mst_node - write recovered master node.
 * @c: UBIFS file-system description object
 * @mst: master node
 *
 * This function returns %0 on success and a negative error code on failure.
 */
static int write_rcvrd_mst_node(struct ubifs_info *c,
				struct ubifs_mst_node *mst)
{
	int err = 0, lnum = UBIFS_MST_LNUM, sz = c->mst_node_alsz;
	__le32 save_flags;

	dbg_rcvry("recovery");

	save_flags = mst->flags;
	mst->flags |= cpu_to_le32(UBIFS_MST_RCVRY);

	ubifs_prepare_node(c, mst, UBIFS_MST_NODE_SZ, 1);
	err = ubifs_leb_change(c, lnum, mst, sz);
	if (err)
		goto out;
	err = ubifs_leb_change(c, lnum + 1, mst, sz);
	if (err)
		goto out;
out:
	mst->flags = save_flags;
	return err;
}

/**
 * ubifs_recover_master_node - recover the master node.
 * @c: UBIFS file-system description object
 *
 * This function recovers the master node from corruption that may occur due to
 * an unclean unmount.
 *
 * This function returns %0 on success and a negative error code on failure.
 */
int ubifs_recover_master_node(struct ubifs_info *c)
{
	void *buf1 = NULL, *buf2 = NULL, *cor1 = NULL, *cor2 = NULL;
	struct ubifs_mst_node *mst1 = NULL, *mst2 = NULL, *mst;
	const int sz = c->mst_node_alsz;
	int err, offs1, offs2;

	dbg_rcvry("recovery");

	err = get_master_node(c, UBIFS_MST_LNUM, &buf1, &mst1, &cor1);
	if (err)
		goto out_free;

	err = get_master_node(c, UBIFS_MST_LNUM + 1, &buf2, &mst2, &cor2);
	if (err)
		goto out_free;

	if (mst1) {
		offs1 = (void *)mst1 - buf1;
		if ((le32_to_cpu(mst1->flags) & UBIFS_MST_RCVRY) &&
		    (offs1 == 0 && !cor1)) {
			/*
			 * mst1 was written by recovery at offset 0 with no
			 * corruption.
			 */
			dbg_rcvry("recovery recovery");
			mst = mst1;
		} else if (mst2) {
			offs2 = (void *)mst2 - buf2;
			if (offs1 == offs2) {
				/* Same offset, so must be the same */
				if (memcmp((void *)mst1 + UBIFS_CH_SZ,
					   (void *)mst2 + UBIFS_CH_SZ,
					   UBIFS_MST_NODE_SZ - UBIFS_CH_SZ))
					goto out_err;
				mst = mst1;
			} else if (offs2 + sz == offs1) {
				/* 1st LEB was written, 2nd was not */
				if (cor1)
					goto out_err;
				mst = mst1;
			} else if (offs1 == 0 &&
				   c->leb_size - offs2 - sz < sz) {
				/* 1st LEB was unmapped and written, 2nd not */
				if (cor1)
					goto out_err;
				mst = mst1;
			} else
				goto out_err;
		} else {
			/*
			 * 2nd LEB was unmapped and about to be written, so
			 * there must be only one master node in the first LEB
			 * and no corruption.
			 */
			if (offs1 != 0 || cor1)
				goto out_err;
			mst = mst1;
		}
	} else {
		if (!mst2)
			goto out_err;
		/*
		 * 1st LEB was unmapped and about to be written, so there must
		 * be no room left in 2nd LEB.
		 */
		offs2 = (void *)mst2 - buf2;
		if (offs2 + sz + sz <= c->leb_size)
			goto out_err;
		mst = mst2;
	}

	ubifs_msg("recovered master node from LEB %d",
		  (mst == mst1 ? UBIFS_MST_LNUM : UBIFS_MST_LNUM + 1));

	memcpy(c->mst_node, mst, UBIFS_MST_NODE_SZ);

	if (c->ro_mount) {
		/* Read-only mode. Keep a copy for switching to rw mode */
		c->rcvrd_mst_node = kmalloc(sz, GFP_KERNEL);
		if (!c->rcvrd_mst_node) {
			err = -ENOMEM;
			goto out_free;
		}
		memcpy(c->rcvrd_mst_node, c->mst_node, UBIFS_MST_NODE_SZ);

		/*
		 * We had to recover the master node, which means there was an
		 * unclean reboot. However, it is possible that the master node
		 * is clean at this point, i.e., %UBIFS_MST_DIRTY is not set.
		 * E.g., consider the following chain of events:
		 *
		 * 1. UBIFS was cleanly unmounted, so the master node is clean
		 * 2. UBIFS is being mounted R/W and starts changing the master
		 *    node in the first (%UBIFS_MST_LNUM). A power cut happens,
		 *    so this LEB ends up with some amount of garbage at the
		 *    end.
		 * 3. UBIFS is being mounted R/O. We reach this place and
		 *    recover the master node from the second LEB
		 *    (%UBIFS_MST_LNUM + 1). But we cannot update the media
		 *    because we are being mounted R/O. We have to defer the
		 *    operation.
		 * 4. However, this master node (@c->mst_node) is marked as
		 *    clean (since the step 1). And if we just return, the
		 *    mount code will be confused and won't recover the master
		 *    node when it is re-mounter R/W later.
		 *
		 *    Thus, to force the recovery by marking the master node as
		 *    dirty.
		 */
		c->mst_node->flags |= cpu_to_le32(UBIFS_MST_DIRTY);
	} else {
		/* Write the recovered master node */
		c->max_sqnum = le64_to_cpu(mst->ch.sqnum) - 1;
		err = write_rcvrd_mst_node(c, c->mst_node);
		if (err)
			goto out_free;
	}

	vfree(buf2);
	vfree(buf1);

	return 0;

out_err:
	err = -EINVAL;
out_free:
	ubifs_err("failed to recover master node");
	if (mst1) {
		ubifs_err("dumping first master node");
		ubifs_dump_node(c, mst1);
	}
	if (mst2) {
		ubifs_err("dumping second master node");
		ubifs_dump_node(c, mst2);
	}
	vfree(buf2);
	vfree(buf1);
	return err;
}

/**
 * ubifs_write_rcvrd_mst_node - write the recovered master node.
 * @c: UBIFS file-system description object
 *
 * This function writes the master node that was recovered during mounting in
 * read-only mode and must now be written because we are remounting rw.
 *
 * This function returns %0 on success and a negative error code on failure.
 */
int ubifs_write_rcvrd_mst_node(struct ubifs_info *c)
{
	int err;

	if (!c->rcvrd_mst_node)
		return 0;
	c->rcvrd_mst_node->flags |= cpu_to_le32(UBIFS_MST_DIRTY);
	c->mst_node->flags |= cpu_to_le32(UBIFS_MST_DIRTY);
	err = write_rcvrd_mst_node(c, c->rcvrd_mst_node);
	if (err)
		return err;
	kfree(c->rcvrd_mst_node);
	c->rcvrd_mst_node = NULL;
	return 0;
}

/**
 * is_last_write - determine if an offset was in the last write to a LEB.
 * @c: UBIFS file-system description object
 * @buf: buffer to check
 * @offs: offset to check
 *
 * This function returns %1 if @offs was in the last write to the LEB whose data
 * is in @buf, otherwise %0 is returned. The determination is made by checking
 * for subsequent empty space starting from the next @c->max_write_size
 * boundary.
 */
static int is_last_write(const struct ubifs_info *c, void *buf, int offs)
{
	int empty_offs, check_len;
	uint8_t *p;

	/*
	 * Round up to the next @c->max_write_size boundary i.e. @offs is in
	 * the last wbuf written. After that should be empty space.
	 */
	empty_offs = ALIGN(offs + 1, c->max_write_size);
	check_len = c->leb_size - empty_offs;
	p = buf + empty_offs - offs;
	return is_empty(p, check_len);
}

/**
 * clean_buf - clean the data from an LEB sitting in a buffer.
 * @c: UBIFS file-system description object
 * @buf: buffer to clean
 * @lnum: LEB number to clean
 * @offs: offset from which to clean
 * @len: length of buffer
 *
 * This function pads up to the next min_io_size boundary (if there is one) and
 * sets empty space to all 0xff. @buf, @offs and @len are updated to the next
 * @c->min_io_size boundary.
 */
static void clean_buf(const struct ubifs_info *c, void **buf, int lnum,
		      int *offs, int *len)
{
	int empty_offs, pad_len;

	lnum = lnum;
	dbg_rcvry("cleaning corruption at %d:%d", lnum, *offs);

	ubifs_assert(!(*offs & 7));
	empty_offs = ALIGN(*offs, c->min_io_size);
	pad_len = empty_offs - *offs;
	ubifs_pad(c, *buf, pad_len);
	*offs += pad_len;
	*buf += pad_len;
	*len -= pad_len;
	memset(*buf, 0xff, c->leb_size - empty_offs);
}

/**
 * no_more_nodes - determine if there are no more nodes in a buffer.
 * @c: UBIFS file-system description object
 * @buf: buffer to check
 * @len: length of buffer
 * @lnum: LEB number of the LEB from which @buf was read
 * @offs: offset from which @buf was read
 *
 * This function ensures that the corrupted node at @offs is the last thing
 * written to a LEB. This function returns %1 if more data is not found and
 * %0 if more data is found.
 */
static int no_more_nodes(const struct ubifs_info *c, void *buf, int len,
			int lnum, int offs)
{
	struct ubifs_ch *ch = buf;
	int skip, dlen = le32_to_cpu(ch->len);

	/* Check for empty space after the corrupt node's common header */
	skip = ALIGN(offs + UBIFS_CH_SZ, c->max_write_size) - offs;
	if (is_empty(buf + skip, len - skip))
		return 1;
	/*
	 * The area after the common header size is not empty, so the common
	 * header must be intact. Check it.
	 */
	if (ubifs_check_node(c, buf, lnum, offs, 1, 0) != -EUCLEAN) {
		dbg_rcvry("unexpected bad common header at %d:%d", lnum, offs);
		return 0;
	}
	/* Now we know the corrupt node's length we can skip over it */
	skip = ALIGN(offs + dlen, c->max_write_size) - offs;
	/* After which there should be empty space */
	if (is_empty(buf + skip, len - skip))
		return 1;
	dbg_rcvry("unexpected data at %d:%d", lnum, offs + skip);
	return 0;
}

/**
 * fix_unclean_leb - fix an unclean LEB.
 * @c: UBIFS file-system description object
 * @sleb: scanned LEB information
 * @start: offset where scan started
 */
static int fix_unclean_leb(struct ubifs_info *c, struct ubifs_scan_leb *sleb,
			   int start)
{
	int lnum = sleb->lnum, endpt = start;

	/* Get the end offset of the last node we are keeping */
	if (!list_empty(&sleb->nodes)) {
		struct ubifs_scan_node *snod;

		snod = list_entry(sleb->nodes.prev,
				  struct ubifs_scan_node, list);
		endpt = snod->offs + snod->len;
	}

	if (c->ro_mount && !c->remounting_rw) {
		/* Add to recovery list */
		struct ubifs_unclean_leb *ucleb;

		dbg_rcvry("need to fix LEB %d start %d endpt %d",
			  lnum, start, sleb->endpt);
		ucleb = kzalloc(sizeof(struct ubifs_unclean_leb), GFP_NOFS);
		if (!ucleb)
			return -ENOMEM;
		ucleb->lnum = lnum;
		ucleb->endpt = endpt;
		list_add_tail(&ucleb->list, &c->unclean_leb_list);
	} else {
		/* Write the fixed LEB back to flash */
		int err;

		dbg_rcvry("fixing LEB %d start %d endpt %d",
			  lnum, start, sleb->endpt);
		if (endpt == 0) {
			err = ubifs_leb_unmap(c, lnum);
			if (err)
				return err;
		} else {
			int len = ALIGN(endpt, c->min_io_size);

			if (start) {
				err = ubifs_leb_read(c, lnum, sleb->buf, 0,
						     start, 1);
				if (err)
					return err;
			}
			/* Pad to min_io_size */
			if (len > endpt) {
				int pad_len = len - ALIGN(endpt, 8);

				if (pad_len > 0) {
					void *buf = sleb->buf + len - pad_len;

					ubifs_pad(c, buf, pad_len);
				}
			}
			err = ubifs_leb_change(c, lnum, sleb->buf, len);
			if (err)
				return err;
		}
	}
	return 0;
}

/**
 * drop_last_group - drop the last group of nodes.
 * @sleb: scanned LEB information
 * @offs: offset of dropped nodes is returned here
 *
 * This is a helper function for 'ubifs_recover_leb()' which drops the last
 * group of nodes of the scanned LEB.
 */
static void drop_last_group(struct ubifs_scan_leb *sleb, int *offs)
{
	while (!list_empty(&sleb->nodes)) {
		struct ubifs_scan_node *snod;
		struct ubifs_ch *ch;

		snod = list_entry(sleb->nodes.prev, struct ubifs_scan_node,
				  list);
		ch = snod->node;
		if (ch->group_type != UBIFS_IN_NODE_GROUP)
			break;

		dbg_rcvry("dropping grouped node at %d:%d",
			  sleb->lnum, snod->offs);
		*offs = snod->offs;
		list_del(&snod->list);
		kfree(snod);
		sleb->nodes_cnt -= 1;
	}
}

/**
 * drop_last_node - drop the last node.
 * @sleb: scanned LEB information
 * @offs: offset of dropped nodes is returned here
 * @grouped: non-zero if whole group of nodes have to be dropped
 *
 * This is a helper function for 'ubifs_recover_leb()' which drops the last
 * node of the scanned LEB.
 */
static void drop_last_node(struct ubifs_scan_leb *sleb, int *offs)
{
	struct ubifs_scan_node *snod;

	if (!list_empty(&sleb->nodes)) {
		snod = list_entry(sleb->nodes.prev, struct ubifs_scan_node,
				  list);

		dbg_rcvry("dropping last node at %d:%d",
			  sleb->lnum, snod->offs);
		*offs = snod->offs;
		list_del(&snod->list);
		kfree(snod);
		sleb->nodes_cnt -= 1;
	}
}

/**
 * ubifs_recover_leb - scan and recover a LEB.
 * @c: UBIFS file-system description object
 * @lnum: LEB number
 * @offs: offset
 * @sbuf: LEB-sized buffer to use
 * @jhead: journal head number this LEB belongs to (%-1 if the LEB does not
 *         belong to any journal head)
 *
 * This function does a scan of a LEB, but caters for errors that might have
 * been caused by the unclean unmount from which we are attempting to recover.
 * Returns %0 in case of success, %-EUCLEAN if an unrecoverable corruption is
 * found, and a negative error code in case of failure.
 */
struct ubifs_scan_leb *ubifs_recover_leb(struct ubifs_info *c, int lnum,
					 int offs, void *sbuf, int jhead)
{
	int ret = 0, err, len = c->leb_size - offs, start = offs, min_io_unit;
	int grouped = jhead == -1 ? 0 : c->jheads[jhead].grouped;
	struct ubifs_scan_leb *sleb;
	void *buf = sbuf + offs;

	dbg_rcvry("%d:%d, jhead %d, grouped %d", lnum, offs, jhead, grouped);

	sleb = ubifs_start_scan(c, lnum, offs, sbuf);
	if (IS_ERR(sleb))
		return sleb;

	ubifs_assert(len >= 8);
	while (len >= 8) {
		dbg_scan("look at LEB %d:%d (%d bytes left)",
			 lnum, offs, len);

		cond_resched();

		/*
		 * Scan quietly until there is an error from which we cannot
		 * recover
		 */
		ret = ubifs_scan_a_node(c, buf, len, lnum, offs, 1);
		if (ret == SCANNED_A_NODE) {
			/* A valid node, and not a padding node */
			struct ubifs_ch *ch = buf;
			int node_len;

			err = ubifs_add_snod(c, sleb, buf, offs);
			if (err)
				goto error;
			node_len = ALIGN(le32_to_cpu(ch->len), 8);
			offs += node_len;
			buf += node_len;
			len -= node_len;
		} else if (ret > 0) {
			/* Padding bytes or a valid padding node */
			offs += ret;
			buf += ret;
			len -= ret;
		} else if (ret == SCANNED_EMPTY_SPACE ||
			   ret == SCANNED_GARBAGE     ||
			   ret == SCANNED_A_BAD_PAD_NODE ||
			   ret == SCANNED_A_CORRUPT_NODE) {
			dbg_rcvry("found corruption (%d) at %d:%d",
				  ret, lnum, offs);
			break;
		} else {
			ubifs_err("unexpected return value %d", ret);
			err = -EINVAL;
			goto error;
		}
	}

	if (ret == SCANNED_GARBAGE || ret == SCANNED_A_BAD_PAD_NODE) {
		if (!is_last_write(c, buf, offs))
			goto corrupted_rescan;
	} else if (ret == SCANNED_A_CORRUPT_NODE) {
		if (!no_more_nodes(c, buf, len, lnum, offs))
			goto corrupted_rescan;
	} else if (!is_empty(buf, len)) {
		if (!is_last_write(c, buf, offs)) {
			int corruption = first_non_ff(buf, len);

			/*
			 * See header comment for this file for more
			 * explanations about the reasons we have this check.
			 */
			ubifs_err("corrupt empty space LEB %d:%d, corruption starts at %d",
				  lnum, offs, corruption);
			/* Make sure we dump interesting non-0xFF data */
			offs += corruption;
			buf += corruption;
			goto corrupted;
		}
	}

	min_io_unit = round_down(offs, c->min_io_size);
	if (grouped)
		/*
		 * If nodes are grouped, always drop the incomplete group at
		 * the end.
		 */
		drop_last_group(sleb, &offs);

	if (jhead == GCHD) {
		/*
		 * If this LEB belongs to the GC head then while we are in the
		 * middle of the same min. I/O unit keep dropping nodes. So
		 * basically, what we want is to make sure that the last min.
		 * I/O unit where we saw the corruption is dropped completely
		 * with all the uncorrupted nodes which may possibly sit there.
		 *
		 * In other words, let's name the min. I/O unit where the
		 * corruption starts B, and the previous min. I/O unit A. The
		 * below code tries to deal with a situation when half of B
		 * contains valid nodes or the end of a valid node, and the
		 * second half of B contains corrupted data or garbage. This
		 * means that UBIFS had been writing to B just before the power
		 * cut happened. I do not know how realistic is this scenario
		 * that half of the min. I/O unit had been written successfully
		 * and the other half not, but this is possible in our 'failure
		 * mode emulation' infrastructure at least.
		 *
		 * So what is the problem, why we need to drop those nodes? Why
		 * can't we just clean-up the second half of B by putting a
		 * padding node there? We can, and this works fine with one
		 * exception which was reproduced with power cut emulation
		 * testing and happens extremely rarely.
		 *
		 * Imagine the file-system is full, we run GC which starts
		 * moving valid nodes from LEB X to LEB Y (obviously, LEB Y is
		 * the current GC head LEB). The @c->gc_lnum is -1, which means
		 * that GC will retain LEB X and will try to continue. Imagine
		 * that LEB X is currently the dirtiest LEB, and the amount of
		 * used space in LEB Y is exactly the same as amount of free
		 * space in LEB X.
		 *
		 * And a power cut happens when nodes are moved from LEB X to
		 * LEB Y. We are here trying to recover LEB Y which is the GC
		 * head LEB. We find the min. I/O unit B as described above.
		 * Then we clean-up LEB Y by padding min. I/O unit. And later
		 * 'ubifs_rcvry_gc_commit()' function fails, because it cannot
		 * find a dirty LEB which could be GC'd into LEB Y! Even LEB X
		 * does not match because the amount of valid nodes there does
		 * not fit the free space in LEB Y any more! And this is
		 * because of the padding node which we added to LEB Y. The
		 * user-visible effect of this which I once observed and
		 * analysed is that we cannot mount the file-system with
		 * -ENOSPC error.
		 *
		 * So obviously, to make sure that situation does not happen we
		 * should free min. I/O unit B in LEB Y completely and the last
		 * used min. I/O unit in LEB Y should be A. This is basically
		 * what the below code tries to do.
		 */
		while (offs > min_io_unit)
			drop_last_node(sleb, &offs);
	}

	buf = sbuf + offs;
	len = c->leb_size - offs;

	clean_buf(c, &buf, lnum, &offs, &len);
	ubifs_end_scan(c, sleb, lnum, offs);

	err = fix_unclean_leb(c, sleb, start);
	if (err)
		goto error;

	return sleb;

corrupted_rescan:
	/* Re-scan the corrupted data with verbose messages */
	ubifs_err("corruption %d", ret);
	ubifs_scan_a_node(c, buf, len, lnum, offs, 1);
corrupted:
	ubifs_scanned_corruption(c, lnum, offs, buf);
	err = -EUCLEAN;
error:
	ubifs_err("LEB %d scanning failed", lnum);
	ubifs_scan_destroy(sleb);
	return ERR_PTR(err);
}

/**
 * get_cs_sqnum - get commit start sequence number.
 * @c: UBIFS file-system description object
 * @lnum: LEB number of commit start node
 * @offs: offset of commit start node
 * @cs_sqnum: commit start sequence number is returned here
 *
 * This function returns %0 on success and a negative error code on failure.
 */
static int get_cs_sqnum(struct ubifs_info *c, int lnum, int offs,
			unsigned long long *cs_sqnum)
{
	struct ubifs_cs_node *cs_node = NULL;
	int err, ret;

	dbg_rcvry("at %d:%d", lnum, offs);
	cs_node = kmalloc(UBIFS_CS_NODE_SZ, GFP_KERNEL);
	if (!cs_node)
		return -ENOMEM;
	if (c->leb_size - offs < UBIFS_CS_NODE_SZ)
		goto out_err;
	err = ubifs_leb_read(c, lnum, (void *)cs_node, offs,
			     UBIFS_CS_NODE_SZ, 0);
	if (err && err != -EBADMSG)
		goto out_free;
	ret = ubifs_scan_a_node(c, cs_node, UBIFS_CS_NODE_SZ, lnum, offs, 0);
	if (ret != SCANNED_A_NODE) {
		ubifs_err("Not a valid node");
		goto out_err;
	}
	if (cs_node->ch.node_type != UBIFS_CS_NODE) {
		ubifs_err("Node a CS node, type is %d", cs_node->ch.node_type);
		goto out_err;
	}
	if (le64_to_cpu(cs_node->cmt_no) != c->cmt_no) {
		ubifs_err("CS node cmt_no %llu != current cmt_no %llu",
			  (unsigned long long)le64_to_cpu(cs_node->cmt_no),
			  c->cmt_no);
		goto out_err;
	}
	*cs_sqnum = le64_to_cpu(cs_node->ch.sqnum);
	dbg_rcvry("commit start sqnum %llu", *cs_sqnum);
	kfree(cs_node);
	return 0;

out_err:
	err = -EINVAL;
out_free:
	ubifs_err("failed to get CS sqnum");
	kfree(cs_node);
	return err;
}

/**
 * ubifs_recover_log_leb - scan and recover a log LEB.
 * @c: UBIFS file-system description object
 * @lnum: LEB number
 * @offs: offset
 * @sbuf: LEB-sized buffer to use
 *
 * This function does a scan of a LEB, but caters for errors that might have
 * been caused by unclean reboots from which we are attempting to recover
 * (assume that only the last log LEB can be corrupted by an unclean reboot).
 *
 * This function returns %0 on success and a negative error code on failure.
 */
struct ubifs_scan_leb *ubifs_recover_log_leb(struct ubifs_info *c, int lnum,
					     int offs, void *sbuf)
{
	struct ubifs_scan_leb *sleb;
	int next_lnum;

	dbg_rcvry("LEB %d", lnum);
	next_lnum = lnum + 1;
	if (next_lnum >= UBIFS_LOG_LNUM + c->log_lebs)
		next_lnum = UBIFS_LOG_LNUM;
	if (next_lnum != c->ltail_lnum) {
		/*
		 * We can only recover at the end of the log, so check that the
		 * next log LEB is empty or out of date.
		 */
		sleb = ubifs_scan(c, next_lnum, 0, sbuf, 0);
		if (IS_ERR(sleb))
			return sleb;
		if (sleb->nodes_cnt) {
			struct ubifs_scan_node *snod;
			unsigned long long cs_sqnum = c->cs_sqnum;

			snod = list_entry(sleb->nodes.next,
					  struct ubifs_scan_node, list);
			if (cs_sqnum == 0) {
				int err;

				err = get_cs_sqnum(c, lnum, offs, &cs_sqnum);
				if (err) {
					ubifs_scan_destroy(sleb);
					return ERR_PTR(err);
				}
			}
			if (snod->sqnum > cs_sqnum) {
				ubifs_err("unrecoverable log corruption in LEB %d",
					  lnum);
				ubifs_scan_destroy(sleb);
				return ERR_PTR(-EUCLEAN);
			}
		}
		ubifs_scan_destroy(sleb);
	}
	return ubifs_recover_leb(c, lnum, offs, sbuf, -1);
}

/**
 * recover_head - recover a head.
 * @c: UBIFS file-system description object
 * @lnum: LEB number of head to recover
 * @offs: offset of head to recover
 * @sbuf: LEB-sized buffer to use
 *
 * This function ensures that there is no data on the flash at a head location.
 *
 * This function returns %0 on success and a negative error code on failure.
 */
static int recover_head(struct ubifs_info *c, int lnum, int offs, void *sbuf)
{
	int len = c->max_write_size, err;

	if (offs + len > c->leb_size)
		len = c->leb_size - offs;

	if (!len)
		return 0;

	/* Read at the head location and check it is empty flash */
	err = ubifs_leb_read(c, lnum, sbuf, offs, len, 1);
	if (err || !is_empty(sbuf, len)) {
		dbg_rcvry("cleaning head at %d:%d", lnum, offs);
		if (offs == 0)
			return ubifs_leb_unmap(c, lnum);
		err = ubifs_leb_read(c, lnum, sbuf, 0, offs, 1);
		if (err)
			return err;
		return ubifs_leb_change(c, lnum, sbuf, offs);
	}

	return 0;
}

/**
 * ubifs_recover_inl_heads - recover index and LPT heads.
 * @c: UBIFS file-system description object
 * @sbuf: LEB-sized buffer to use
 *
 * This function ensures that there is no data on the flash at the index and
 * LPT head locations.
 *
 * This deals with the recovery of a half-completed journal commit. UBIFS is
 * careful never to overwrite the last version of the index or the LPT. Because
 * the index and LPT are wandering trees, data from a half-completed commit will
 * not be referenced anywhere in UBIFS. The data will be either in LEBs that are
 * assumed to be empty and will be unmapped anyway before use, or in the index
 * and LPT heads.
 *
 * This function returns %0 on success and a negative error code on failure.
 */
int ubifs_recover_inl_heads(struct ubifs_info *c, void *sbuf)
{
	int err;

	ubifs_assert(!c->ro_mount || c->remounting_rw);

	dbg_rcvry("checking index head at %d:%d", c->ihead_lnum, c->ihead_offs);
	err = recover_head(c, c->ihead_lnum, c->ihead_offs, sbuf);
	if (err)
		return err;

	dbg_rcvry("checking LPT head at %d:%d", c->nhead_lnum, c->nhead_offs);
	err = recover_head(c, c->nhead_lnum, c->nhead_offs, sbuf);
	if (err)
		return err;

	return 0;
}

/**
 * clean_an_unclean_leb - read and write a LEB to remove corruption.
 * @c: UBIFS file-system description object
 * @ucleb: unclean LEB information
 * @sbuf: LEB-sized buffer to use
 *
 * This function reads a LEB up to a point pre-determined by the mount recovery,
 * checks the nodes, and writes the result back to the flash, thereby cleaning
 * off any following corruption, or non-fatal ECC errors.
 *
 * This function returns %0 on success and a negative error code on failure.
 */
static int clean_an_unclean_leb(struct ubifs_info *c,
				struct ubifs_unclean_leb *ucleb, void *sbuf)
{
	int err, lnum = ucleb->lnum, offs = 0, len = ucleb->endpt, quiet = 1;
	void *buf = sbuf;

	dbg_rcvry("LEB %d len %d", lnum, len);

	if (len == 0) {
		/* Nothing to read, just unmap it */
		err = ubifs_leb_unmap(c, lnum);
		if (err)
			return err;
		return 0;
	}

	err = ubifs_leb_read(c, lnum, buf, offs, len, 0);
	if (err && err != -EBADMSG)
		return err;

	while (len >= 8) {
		int ret;

		cond_resched();

		/* Scan quietly until there is an error */
		ret = ubifs_scan_a_node(c, buf, len, lnum, offs, quiet);

		if (ret == SCANNED_A_NODE) {
			/* A valid node, and not a padding node */
			struct ubifs_ch *ch = buf;
			int node_len;

			node_len = ALIGN(le32_to_cpu(ch->len), 8);
			offs += node_len;
			buf += node_len;
			len -= node_len;
			continue;
		}

		if (ret > 0) {
			/* Padding bytes or a valid padding node */
			offs += ret;
			buf += ret;
			len -= ret;
			continue;
		}

		if (ret == SCANNED_EMPTY_SPACE) {
			ubifs_err("unexpected empty space at %d:%d",
				  lnum, offs);
			return -EUCLEAN;
		}

		if (quiet) {
			/* Redo the last scan but noisily */
			quiet = 0;
			continue;
		}

		ubifs_scanned_corruption(c, lnum, offs, buf);
		return -EUCLEAN;
	}

	/* Pad to min_io_size */
	len = ALIGN(ucleb->endpt, c->min_io_size);
	if (len > ucleb->endpt) {
		int pad_len = len - ALIGN(ucleb->endpt, 8);

		if (pad_len > 0) {
			buf = c->sbuf + len - pad_len;
			ubifs_pad(c, buf, pad_len);
		}
	}

	/* Write back the LEB atomically */
	err = ubifs_leb_change(c, lnum, sbuf, len);
	if (err)
		return err;

	dbg_rcvry("cleaned LEB %d", lnum);

	return 0;
}

/**
 * ubifs_clean_lebs - clean LEBs recovered during read-only mount.
 * @c: UBIFS file-system description object
 * @sbuf: LEB-sized buffer to use
 *
 * This function cleans a LEB identified during recovery that needs to be
 * written but was not because UBIFS was mounted read-only. This happens when
 * remounting to read-write mode.
 *
 * This function returns %0 on success and a negative error code on failure.
 */
int ubifs_clean_lebs(struct ubifs_info *c, void *sbuf)
{
	dbg_rcvry("recovery");
	while (!list_empty(&c->unclean_leb_list)) {
		struct ubifs_unclean_leb *ucleb;
		int err;

		ucleb = list_entry(c->unclean_leb_list.next,
				   struct ubifs_unclean_leb, list);
		err = clean_an_unclean_leb(c, ucleb, sbuf);
		if (err)
			return err;
		list_del(&ucleb->list);
		kfree(ucleb);
	}
	return 0;
}

/**
 * grab_empty_leb - grab an empty LEB to use as GC LEB and run commit.
 * @c: UBIFS file-system description object
 *
 * This is a helper function for 'ubifs_rcvry_gc_commit()' which grabs an empty
 * LEB to be used as GC LEB (@c->gc_lnum), and then runs the commit. Returns
 * zero in case of success and a negative error code in case of failure.
 */
static int grab_empty_leb(struct ubifs_info *c)
{
	int lnum, err;

	/*
	 * Note, it is very important to first search for an empty LEB and then
	 * run the commit, not vice-versa. The reason is that there might be
	 * only one empty LEB at the moment, the one which has been the
	 * @c->gc_lnum just before the power cut happened. During the regular
	 * UBIFS operation (not now) @c->gc_lnum is marked as "taken", so no
	 * one but GC can grab it. But at this moment this single empty LEB is
	 * not marked as taken, so if we run commit - what happens? Right, the
	 * commit will grab it and write the index there. Remember that the
	 * index always expands as long as there is free space, and it only
	 * starts consolidating when we run out of space.
	 *
	 * IOW, if we run commit now, we might not be able to find a free LEB
	 * after this.
	 */
	lnum = ubifs_find_free_leb_for_idx(c);
	if (lnum < 0) {
		ubifs_err("could not find an empty LEB");
		ubifs_dump_lprops(c);
		ubifs_dump_budg(c, &c->bi);
		return lnum;
	}

	/* Reset the index flag */
	err = ubifs_change_one_lp(c, lnum, LPROPS_NC, LPROPS_NC, 0,
				  LPROPS_INDEX, 0);
	if (err)
		return err;

	c->gc_lnum = lnum;
	dbg_rcvry("found empty LEB %d, run commit", lnum);

	return ubifs_run_commit(c);
}

/**
 * ubifs_rcvry_gc_commit - recover the GC LEB number and run the commit.
 * @c: UBIFS file-system description object
 *
 * Out-of-place garbage collection requires always one empty LEB with which to
 * start garbage collection. The LEB number is recorded in c->gc_lnum and is
 * written to the master node on unmounting. In the case of an unclean unmount
 * the value of gc_lnum recorded in the master node is out of date and cannot
 * be used. Instead, recovery must allocate an empty LEB for this purpose.
 * However, there may not be enough empty space, in which case it must be
 * possible to GC the dirtiest LEB into the GC head LEB.
 *
 * This function also runs the commit which causes the TNC updates from
 * size-recovery and orphans to be written to the flash. That is important to
 * ensure correct replay order for subsequent mounts.
 *
 * This function returns %0 on success and a negative error code on failure.
 */
int ubifs_rcvry_gc_commit(struct ubifs_info *c)
{
	struct ubifs_wbuf *wbuf = &c->jheads[GCHD].wbuf;
	struct ubifs_lprops lp;
	int err;

	dbg_rcvry("GC head LEB %d, offs %d", wbuf->lnum, wbuf->offs);

	c->gc_lnum = -1;
	if (wbuf->lnum == -1 || wbuf->offs == c->leb_size)
		return grab_empty_leb(c);

	err = ubifs_find_dirty_leb(c, &lp, wbuf->offs, 2);
	if (err) {
		if (err != -ENOSPC)
			return err;

		dbg_rcvry("could not find a dirty LEB");
		return grab_empty_leb(c);
	}

	ubifs_assert(!(lp.flags & LPROPS_INDEX));
	ubifs_assert(lp.free + lp.dirty >= wbuf->offs);

	/*
	 * We run the commit before garbage collection otherwise subsequent
	 * mounts will see the GC and orphan deletion in a different order.
	 */
	dbg_rcvry("committing");
	err = ubifs_run_commit(c);
	if (err)
		return err;

	dbg_rcvry("GC'ing LEB %d", lp.lnum);
	mutex_lock_nested(&wbuf->io_mutex, wbuf->jhead);
	err = ubifs_garbage_collect_leb(c, &lp);
	if (err >= 0) {
		int err2 = ubifs_wbuf_sync_nolock(wbuf);

		if (err2)
			err = err2;
	}
	mutex_unlock(&wbuf->io_mutex);
	if (err < 0) {
		ubifs_err("GC failed, error %d", err);
		if (err == -EAGAIN)
			err = -EINVAL;
		return err;
	}

	ubifs_assert(err == LEB_RETAINED);
	if (err != LEB_RETAINED)
		return -EINVAL;

	err = ubifs_leb_unmap(c, c->gc_lnum);
	if (err)
		return err;

	dbg_rcvry("allocated LEB %d for GC", lp.lnum);
	return 0;
}

/**
 * struct size_entry - inode size information for recovery.
 * @rb: link in the RB-tree of sizes
 * @inum: inode number
 * @i_size: size on inode
 * @d_size: maximum size based on data nodes
 * @exists: indicates whether the inode exists
 * @inode: inode if pinned in memory awaiting rw mode to fix it
 */
struct size_entry {
	struct rb_node rb;
	ino_t inum;
	loff_t i_size;
	loff_t d_size;
	int exists;
	struct inode *inode;
};

/**
 * add_ino - add an entry to the size tree.
 * @c: UBIFS file-system description object
 * @inum: inode number
 * @i_size: size on inode
 * @d_size: maximum size based on data nodes
 * @exists: indicates whether the inode exists
 */
static int add_ino(struct ubifs_info *c, ino_t inum, loff_t i_size,
		   loff_t d_size, int exists)
{
	struct rb_node **p = &c->size_tree.rb_node, *parent = NULL;
	struct size_entry *e;

	while (*p) {
		parent = *p;
		e = rb_entry(parent, struct size_entry, rb);
		if (inum < e->inum)
			p = &(*p)->rb_left;
		else
			p = &(*p)->rb_right;
	}

	e = kzalloc(sizeof(struct size_entry), GFP_KERNEL);
	if (!e)
		return -ENOMEM;

	e->inum = inum;
	e->i_size = i_size;
	e->d_size = d_size;
	e->exists = exists;

	rb_link_node(&e->rb, parent, p);
	rb_insert_color(&e->rb, &c->size_tree);

	return 0;
}

/**
 * find_ino - find an entry on the size tree.
 * @c: UBIFS file-system description object
 * @inum: inode number
 */
static struct size_entry *find_ino(struct ubifs_info *c, ino_t inum)
{
	struct rb_node *p = c->size_tree.rb_node;
	struct size_entry *e;

	while (p) {
		e = rb_entry(p, struct size_entry, rb);
		if (inum < e->inum)
			p = p->rb_left;
		else if (inum > e->inum)
			p = p->rb_right;
		else
			return e;
	}
	return NULL;
}

/**
 * remove_ino - remove an entry from the size tree.
 * @c: UBIFS file-system description object
 * @inum: inode number
 */
static void remove_ino(struct ubifs_info *c, ino_t inum)
{
	struct size_entry *e = find_ino(c, inum);

	if (!e)
		return;
	rb_erase(&e->rb, &c->size_tree);
	kfree(e);
}

/**
 * ubifs_destroy_size_tree - free resources related to the size tree.
 * @c: UBIFS file-system description object
 */
void ubifs_destroy_size_tree(struct ubifs_info *c)
{
	struct size_entry *e, *n;

	rbtree_postorder_for_each_entry_safe(e, n, &c->size_tree, rb) {
		if (e->inode)
			iput(e->inode);
		kfree(e);
	}

	c->size_tree = RB_ROOT;
}

/**
 * ubifs_recover_size_accum - accumulate inode sizes for recovery.
 * @c: UBIFS file-system description object
 * @key: node key
 * @deletion: node is for a deletion
 * @new_size: inode size
 *
 * This function has two purposes:
 *     1) to ensure there are no data nodes that fall outside the inode size
 *     2) to ensure there are no data nodes for inodes that do not exist
 * To accomplish those purposes, a rb-tree is constructed containing an entry
 * for each inode number in the journal that has not been deleted, and recording
 * the size from the inode node, the maximum size of any data node (also altered
 * by truncations) and a flag indicating a inode number for which no inode node
 * was present in the journal.
 *
 * Note that there is still the possibility that there are data nodes that have
 * been committed that are beyond the inode size, however the only way to find
 * them would be to scan the entire index. Alternatively, some provision could
 * be made to record the size of inodes at the start of commit, which would seem
 * very cumbersome for a scenario that is quite unlikely and the only negative
 * consequence of which is wasted space.
 *
 * This functions returns %0 on success and a negative error code on failure.
 */
int ubifs_recover_size_accum(struct ubifs_info *c, union ubifs_key *key,
			     int deletion, loff_t new_size)
{
	ino_t inum = key_inum(c, key);
	struct size_entry *e;
	int err;

	switch (key_type(c, key)) {
	case UBIFS_INO_KEY:
		if (deletion)
			remove_ino(c, inum);
		else {
			e = find_ino(c, inum);
			if (e) {
				e->i_size = new_size;
				e->exists = 1;
			} else {
				err = add_ino(c, inum, new_size, 0, 1);
				if (err)
					return err;
			}
		}
		break;
	case UBIFS_DATA_KEY:
		e = find_ino(c, inum);
		if (e) {
			if (new_size > e->d_size)
				e->d_size = new_size;
		} else {
			err = add_ino(c, inum, 0, new_size, 0);
			if (err)
				return err;
		}
		break;
	case UBIFS_TRUN_KEY:
		e = find_ino(c, inum);
		if (e)
			e->d_size = new_size;
		break;
	}
	return 0;
}

/**
 * fix_size_in_place - fix inode size in place on flash.
 * @c: UBIFS file-system description object
 * @e: inode size information for recovery
 */
static int fix_size_in_place(struct ubifs_info *c, struct size_entry *e)
{
	struct ubifs_ino_node *ino = c->sbuf;
	unsigned char *p;
	union ubifs_key key;
	int err, lnum, offs, len;
	loff_t i_size;
	uint32_t crc;

	/* Locate the inode node LEB number and offset */
	ino_key_init(c, &key, e->inum);
	err = ubifs_tnc_locate(c, &key, ino, &lnum, &offs);
	if (err)
		goto out;
	/*
	 * If the size recorded on the inode node is greater than the size that
	 * was calculated from nodes in the journal then don't change the inode.
	 */
	i_size = le64_to_cpu(ino->size);
	if (i_size >= e->d_size)
		return 0;
	/* Read the LEB */
	err = ubifs_leb_read(c, lnum, c->sbuf, 0, c->leb_size, 1);
	if (err)
		goto out;
	/* Change the size field and recalculate the CRC */
	ino = c->sbuf + offs;
	ino->size = cpu_to_le64(e->d_size);
	len = le32_to_cpu(ino->ch.len);
	crc = crc32(UBIFS_CRC32_INIT, (void *)ino + 8, len - 8);
	ino->ch.crc = cpu_to_le32(crc);
	/* Work out where data in the LEB ends and free space begins */
	p = c->sbuf;
	len = c->leb_size - 1;
	while (p[len] == 0xff)
		len -= 1;
	len = ALIGN(len + 1, c->min_io_size);
	/* Atomically write the fixed LEB back again */
	err = ubifs_leb_change(c, lnum, c->sbuf, len);
	if (err)
		goto out;
	dbg_rcvry("inode %lu at %d:%d size %lld -> %lld",
		  (unsigned long)e->inum, lnum, offs, i_size, e->d_size);
	return 0;

out:
	ubifs_warn("inode %lu failed to fix size %lld -> %lld error %d",
		   (unsigned long)e->inum, e->i_size, e->d_size, err);
	return err;
}

/**
 * ubifs_recover_size - recover inode size.
 * @c: UBIFS file-system description object
 *
 * This function attempts to fix inode size discrepancies identified by the
 * 'ubifs_recover_size_accum()' function.
 *
 * This functions returns %0 on success and a negative error code on failure.
 */
int ubifs_recover_size(struct ubifs_info *c)
{
	struct rb_node *this = rb_first(&c->size_tree);

	while (this) {
		struct size_entry *e;
		int err;

		e = rb_entry(this, struct size_entry, rb);
		if (!e->exists) {
			union ubifs_key key;

			ino_key_init(c, &key, e->inum);
			err = ubifs_tnc_lookup(c, &key, c->sbuf);
			if (err && err != -ENOENT)
				return err;
			if (err == -ENOENT) {
				/* Remove data nodes that have no inode */
				dbg_rcvry("removing ino %lu",
					  (unsigned long)e->inum);
				err = ubifs_tnc_remove_ino(c, e->inum);
				if (err)
					return err;
			} else {
				struct ubifs_ino_node *ino = c->sbuf;

				e->exists = 1;
				e->i_size = le64_to_cpu(ino->size);
			}
		}

		if (e->exists && e->i_size < e->d_size) {
			if (c->ro_mount) {
				/* Fix the inode size and pin it in memory */
				struct inode *inode;
				struct ubifs_inode *ui;

				ubifs_assert(!e->inode);

				inode = ubifs_iget(c->vfs_sb, e->inum);
				if (IS_ERR(inode))
					return PTR_ERR(inode);

				ui = ubifs_inode(inode);
				if (inode->i_size < e->d_size) {
					dbg_rcvry("ino %lu size %lld -> %lld",
						  (unsigned long)e->inum,
						  inode->i_size, e->d_size);
					inode->i_size = e->d_size;
					ui->ui_size = e->d_size;
					ui->synced_i_size = e->d_size;
					e->inode = inode;
					this = rb_next(this);
					continue;
				}
				iput(inode);
			} else {
				/* Fix the size in place */
				err = fix_size_in_place(c, e);
				if (err)
					return err;
				if (e->inode)
					iput(e->inode);
			}
		}

		this = rb_next(this);
		rb_erase(&e->rb, &c->size_tree);
		kfree(e);
	}

	return 0;
}