resize.c 33.1 KB
Newer Older
1
/*
2
 *  linux/fs/ext4/resize.c
3
 *
4
 * Support for resizing an ext4 filesystem while it is mounted.
5 6 7 8 9 10 11
 *
 * Copyright (C) 2001, 2002 Andreas Dilger <adilger@clusterfs.com>
 *
 * This could probably be made into a module, because it is not often in use.
 */


12
#define EXT4FS_DEBUG
13

14
#include <linux/ext4_jbd2.h>
15 16 17 18

#include <linux/errno.h>
#include <linux/slab.h>

A
Andreas Dilger 已提交
19
#include "group.h"
20 21 22 23 24

#define outside(b, first, last)	((b) < (first) || (b) >= (last))
#define inside(b, first, last)	((b) >= (first) && (b) < (last))

static int verify_group_input(struct super_block *sb,
25
			      struct ext4_new_group_data *input)
26
{
27 28
	struct ext4_sb_info *sbi = EXT4_SB(sb);
	struct ext4_super_block *es = sbi->s_es;
L
Laurent Vivier 已提交
29
	ext4_fsblk_t start = ext4_blocks_count(es);
30
	ext4_fsblk_t end = start + input->blocks_count;
31
	ext4_group_t group = input->group;
32 33 34
	ext4_fsblk_t itend = input->inode_table + sbi->s_itb_per_group;
	unsigned overhead = ext4_bg_has_super(sb, group) ?
		(1 + ext4_bg_num_gdb(sb, group) +
35
		 le16_to_cpu(es->s_reserved_gdt_blocks)) : 0;
36
	ext4_fsblk_t metaend = start + overhead;
37
	struct buffer_head *bh = NULL;
38
	ext4_grpblk_t free_blocks_count, offset;
39 40 41 42 43 44
	int err = -EINVAL;

	input->free_blocks_count = free_blocks_count =
		input->blocks_count - 2 - overhead - sbi->s_itb_per_group;

	if (test_opt(sb, DEBUG))
45
		printk(KERN_DEBUG "EXT4-fs: adding %s group %u: %u blocks "
46
		       "(%d free, %u reserved)\n",
47
		       ext4_bg_has_super(sb, input->group) ? "normal" :
48 49 50
		       "no-super", input->group, input->blocks_count,
		       free_blocks_count, input->reserved_blocks);

51
	ext4_get_group_no_and_offset(sb, start, NULL, &offset);
52
	if (group != sbi->s_groups_count)
53
		ext4_warning(sb, __FUNCTION__,
54 55
			     "Cannot add at group %u (only %lu groups)",
			     input->group, sbi->s_groups_count);
56 57
	else if (offset != 0)
			ext4_warning(sb, __FUNCTION__, "Last group not full");
58
	else if (input->reserved_blocks > input->blocks_count / 5)
59
		ext4_warning(sb, __FUNCTION__, "Reserved blocks too high (%u)",
60 61
			     input->reserved_blocks);
	else if (free_blocks_count < 0)
62
		ext4_warning(sb, __FUNCTION__, "Bad blocks count %u",
63 64
			     input->blocks_count);
	else if (!(bh = sb_bread(sb, end - 1)))
65
		ext4_warning(sb, __FUNCTION__,
66
			     "Cannot read last block (%llu)",
67 68
			     end - 1);
	else if (outside(input->block_bitmap, start, end))
69
		ext4_warning(sb, __FUNCTION__,
L
Laurent Vivier 已提交
70
			     "Block bitmap not in group (block %llu)",
71
			     (unsigned long long)input->block_bitmap);
72
	else if (outside(input->inode_bitmap, start, end))
73
		ext4_warning(sb, __FUNCTION__,
L
Laurent Vivier 已提交
74
			     "Inode bitmap not in group (block %llu)",
75
			     (unsigned long long)input->inode_bitmap);
76 77
	else if (outside(input->inode_table, start, end) ||
	         outside(itend - 1, start, end))
78
		ext4_warning(sb, __FUNCTION__,
L
Laurent Vivier 已提交
79
			     "Inode table not in group (blocks %llu-%llu)",
80
			     (unsigned long long)input->inode_table, itend - 1);
81
	else if (input->inode_bitmap == input->block_bitmap)
82
		ext4_warning(sb, __FUNCTION__,
L
Laurent Vivier 已提交
83
			     "Block bitmap same as inode bitmap (%llu)",
84
			     (unsigned long long)input->block_bitmap);
85
	else if (inside(input->block_bitmap, input->inode_table, itend))
86
		ext4_warning(sb, __FUNCTION__,
L
Laurent Vivier 已提交
87
			     "Block bitmap (%llu) in inode table (%llu-%llu)",
88 89
			     (unsigned long long)input->block_bitmap,
			     (unsigned long long)input->inode_table, itend - 1);
90
	else if (inside(input->inode_bitmap, input->inode_table, itend))
91
		ext4_warning(sb, __FUNCTION__,
L
Laurent Vivier 已提交
92
			     "Inode bitmap (%llu) in inode table (%llu-%llu)",
93 94
			     (unsigned long long)input->inode_bitmap,
			     (unsigned long long)input->inode_table, itend - 1);
95
	else if (inside(input->block_bitmap, start, metaend))
96
		ext4_warning(sb, __FUNCTION__,
L
Laurent Vivier 已提交
97
			     "Block bitmap (%llu) in GDT table"
98
			     " (%llu-%llu)",
99 100
			     (unsigned long long)input->block_bitmap,
			     start, metaend - 1);
101
	else if (inside(input->inode_bitmap, start, metaend))
102
		ext4_warning(sb, __FUNCTION__,
L
Laurent Vivier 已提交
103
			     "Inode bitmap (%llu) in GDT table"
104
			     " (%llu-%llu)",
105 106
			     (unsigned long long)input->inode_bitmap,
			     start, metaend - 1);
107 108
	else if (inside(input->inode_table, start, metaend) ||
	         inside(itend - 1, start, metaend))
109
		ext4_warning(sb, __FUNCTION__,
110 111
			     "Inode table (%llu-%llu) overlaps"
			     "GDT table (%llu-%llu)",
112 113
			     (unsigned long long)input->inode_table,
			     itend - 1, start, metaend - 1);
114 115 116 117 118 119 120 121
	else
		err = 0;
	brelse(bh);

	return err;
}

static struct buffer_head *bclean(handle_t *handle, struct super_block *sb,
122
				  ext4_fsblk_t blk)
123 124 125 126 127 128 129
{
	struct buffer_head *bh;
	int err;

	bh = sb_getblk(sb, blk);
	if (!bh)
		return ERR_PTR(-EIO);
130
	if ((err = ext4_journal_get_write_access(handle, bh))) {
131 132 133 134 135 136 137 138 139 140 141 142
		brelse(bh);
		bh = ERR_PTR(err);
	} else {
		lock_buffer(bh);
		memset(bh->b_data, 0, sb->s_blocksize);
		set_buffer_uptodate(bh);
		unlock_buffer(bh);
	}

	return bh;
}

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
/*
 * If we have fewer than thresh credits, extend by EXT4_MAX_TRANS_DATA.
 * If that fails, restart the transaction & regain write access for the
 * buffer head which is used for block_bitmap modifications.
 */
static int extend_or_restart_transaction(handle_t *handle, int thresh,
					 struct buffer_head *bh)
{
	int err;

	if (handle->h_buffer_credits >= thresh)
		return 0;

	err = ext4_journal_extend(handle, EXT4_MAX_TRANS_DATA);
	if (err < 0)
		return err;
	if (err) {
		if ((err = ext4_journal_restart(handle, EXT4_MAX_TRANS_DATA)))
			return err;
	        if ((err = ext4_journal_get_write_access(handle, bh)))
			return err;
        }

	return 0;
}

169 170 171 172 173 174 175 176
/*
 * Set up the block and inode bitmaps, and the inode table for the new group.
 * This doesn't need to be part of the main transaction, since we are only
 * changing blocks outside the actual filesystem.  We still do journaling to
 * ensure the recovery is correct in case of a failure just after resize.
 * If any part of this fails, we simply abort the resize.
 */
static int setup_new_group_blocks(struct super_block *sb,
177
				  struct ext4_new_group_data *input)
178
{
179 180 181
	struct ext4_sb_info *sbi = EXT4_SB(sb);
	ext4_fsblk_t start = ext4_group_first_block_no(sb, input->group);
	int reserved_gdb = ext4_bg_has_super(sb, input->group) ?
182
		le16_to_cpu(sbi->s_es->s_reserved_gdt_blocks) : 0;
183
	unsigned long gdblocks = ext4_bg_num_gdb(sb, input->group);
184 185
	struct buffer_head *bh;
	handle_t *handle;
186 187
	ext4_fsblk_t block;
	ext4_grpblk_t bit;
188 189 190
	int i;
	int err = 0, err2;

191 192 193
	/* This transaction may be extended/restarted along the way */
	handle = ext4_journal_start_sb(sb, EXT4_MAX_TRANS_DATA);

194 195 196 197 198 199 200 201 202 203 204 205 206 207
	if (IS_ERR(handle))
		return PTR_ERR(handle);

	lock_super(sb);
	if (input->group != sbi->s_groups_count) {
		err = -EBUSY;
		goto exit_journal;
	}

	if (IS_ERR(bh = bclean(handle, sb, input->block_bitmap))) {
		err = PTR_ERR(bh);
		goto exit_journal;
	}

208
	if (ext4_bg_has_super(sb, input->group)) {
E
Eric Sandeen 已提交
209
		ext4_debug("mark backup superblock %#04llx (+0)\n", start);
210
		ext4_set_bit(0, bh->b_data);
211 212 213 214 215 216 217
	}

	/* Copy all of the GDT blocks into the backup in this group */
	for (i = 0, bit = 1, block = start + 1;
	     i < gdblocks; i++, block++, bit++) {
		struct buffer_head *gdb;

E
Eric Sandeen 已提交
218
		ext4_debug("update backup group %#04llx (+%d)\n", block, bit);
219

220 221 222
		if ((err = extend_or_restart_transaction(handle, 1, bh)))
			goto exit_bh;

223 224 225 226 227
		gdb = sb_getblk(sb, block);
		if (!gdb) {
			err = -EIO;
			goto exit_bh;
		}
228
		if ((err = ext4_journal_get_write_access(handle, gdb))) {
229 230 231
			brelse(gdb);
			goto exit_bh;
		}
232 233
		lock_buffer(gdb);
		memcpy(gdb->b_data, sbi->s_group_desc[i]->b_data, gdb->b_size);
234
		set_buffer_uptodate(gdb);
235
		unlock_buffer(gdb);
236 237
		ext4_journal_dirty_metadata(handle, gdb);
		ext4_set_bit(bit, bh->b_data);
238 239 240 241 242 243 244 245
		brelse(gdb);
	}

	/* Zero out all of the reserved backup group descriptor table blocks */
	for (i = 0, bit = gdblocks + 1, block = start + bit;
	     i < reserved_gdb; i++, block++, bit++) {
		struct buffer_head *gdb;

E
Eric Sandeen 已提交
246
		ext4_debug("clear reserved block %#04llx (+%d)\n", block, bit);
247

248 249 250
		if ((err = extend_or_restart_transaction(handle, 1, bh)))
			goto exit_bh;

251 252 253 254
		if (IS_ERR(gdb = bclean(handle, sb, block))) {
			err = PTR_ERR(bh);
			goto exit_bh;
		}
255 256
		ext4_journal_dirty_metadata(handle, gdb);
		ext4_set_bit(bit, bh->b_data);
257 258
		brelse(gdb);
	}
E
Eric Sandeen 已提交
259
	ext4_debug("mark block bitmap %#04llx (+%llu)\n", input->block_bitmap,
260
		   input->block_bitmap - start);
261
	ext4_set_bit(input->block_bitmap - start, bh->b_data);
E
Eric Sandeen 已提交
262
	ext4_debug("mark inode bitmap %#04llx (+%llu)\n", input->inode_bitmap,
263
		   input->inode_bitmap - start);
264
	ext4_set_bit(input->inode_bitmap - start, bh->b_data);
265 266 267 268 269 270

	/* Zero out all of the inode table blocks */
	for (i = 0, block = input->inode_table, bit = block - start;
	     i < sbi->s_itb_per_group; i++, bit++, block++) {
		struct buffer_head *it;

E
Eric Sandeen 已提交
271
		ext4_debug("clear inode block %#04llx (+%d)\n", block, bit);
272 273 274 275

		if ((err = extend_or_restart_transaction(handle, 1, bh)))
			goto exit_bh;

276 277 278 279
		if (IS_ERR(it = bclean(handle, sb, block))) {
			err = PTR_ERR(it);
			goto exit_bh;
		}
280
		ext4_journal_dirty_metadata(handle, it);
281
		brelse(it);
282
		ext4_set_bit(bit, bh->b_data);
283
	}
284 285 286 287

	if ((err = extend_or_restart_transaction(handle, 2, bh)))
		goto exit_bh;

288
	mark_bitmap_end(input->blocks_count, EXT4_BLOCKS_PER_GROUP(sb),
289
			bh->b_data);
290
	ext4_journal_dirty_metadata(handle, bh);
291 292 293
	brelse(bh);

	/* Mark unused entries in inode bitmap used */
E
Eric Sandeen 已提交
294
	ext4_debug("clear inode bitmap %#04llx (+%llu)\n",
295 296 297 298 299 300
		   input->inode_bitmap, input->inode_bitmap - start);
	if (IS_ERR(bh = bclean(handle, sb, input->inode_bitmap))) {
		err = PTR_ERR(bh);
		goto exit_journal;
	}

301
	mark_bitmap_end(EXT4_INODES_PER_GROUP(sb), EXT4_BLOCKS_PER_GROUP(sb),
302
			bh->b_data);
303
	ext4_journal_dirty_metadata(handle, bh);
304 305 306 307 308
exit_bh:
	brelse(bh);

exit_journal:
	unlock_super(sb);
309
	if ((err2 = ext4_journal_stop(handle)) && !err)
310 311 312 313 314 315 316
		err = err2;

	return err;
}

/*
 * Iterate through the groups which hold BACKUP superblock/GDT copies in an
317
 * ext4 filesystem.  The counters should be initialized to 1, 5, and 7 before
318 319 320 321
 * calling this for the first time.  In a sparse filesystem it will be the
 * sequence of powers of 3, 5, and 7: 1, 3, 5, 7, 9, 25, 27, 49, 81, ...
 * For a non-sparse filesystem it will be every group: 1, 2, 3, 4, ...
 */
322
static unsigned ext4_list_backups(struct super_block *sb, unsigned *three,
323 324 325 326 327 328
				  unsigned *five, unsigned *seven)
{
	unsigned *min = three;
	int mult = 3;
	unsigned ret;

329 330
	if (!EXT4_HAS_RO_COMPAT_FEATURE(sb,
					EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER)) {
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
		ret = *min;
		*min += 1;
		return ret;
	}

	if (*five < *min) {
		min = five;
		mult = 5;
	}
	if (*seven < *min) {
		min = seven;
		mult = 7;
	}

	ret = *min;
	*min *= mult;

	return ret;
}

/*
 * Check that all of the backup GDT blocks are held in the primary GDT block.
 * It is assumed that they are stored in group order.  Returns the number of
 * groups in current filesystem that have BACKUPS, or -ve error code.
 */
static int verify_reserved_gdb(struct super_block *sb,
			       struct buffer_head *primary)
{
359
	const ext4_fsblk_t blk = primary->b_blocknr;
360
	const ext4_group_t end = EXT4_SB(sb)->s_groups_count;
361 362 363 364 365 366 367
	unsigned three = 1;
	unsigned five = 5;
	unsigned seven = 7;
	unsigned grp;
	__le32 *p = (__le32 *)primary->b_data;
	int gdbackups = 0;

368
	while ((grp = ext4_list_backups(sb, &three, &five, &seven)) < end) {
L
Laurent Vivier 已提交
369 370
		if (le32_to_cpu(*p++) !=
		    grp * EXT4_BLOCKS_PER_GROUP(sb) + blk){
371
			ext4_warning(sb, __FUNCTION__,
372 373
				     "reserved GDT %llu"
				     " missing grp %d (%llu)",
374
				     blk, grp,
L
Laurent Vivier 已提交
375 376 377
				     grp *
				     (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
				     blk);
378 379
			return -EINVAL;
		}
380
		if (++gdbackups > EXT4_ADDR_PER_BLOCK(sb))
381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
			return -EFBIG;
	}

	return gdbackups;
}

/*
 * Called when we need to bring a reserved group descriptor table block into
 * use from the resize inode.  The primary copy of the new GDT block currently
 * is an indirect block (under the double indirect block in the resize inode).
 * The new backup GDT blocks will be stored as leaf blocks in this indirect
 * block, in group order.  Even though we know all the block numbers we need,
 * we check to ensure that the resize inode has actually reserved these blocks.
 *
 * Don't need to update the block bitmaps because the blocks are still in use.
 *
 * We get all of the error cases out of the way, so that we are sure to not
 * fail once we start modifying the data on disk, because JBD has no rollback.
 */
static int add_new_gdb(handle_t *handle, struct inode *inode,
401
		       struct ext4_new_group_data *input,
402 403 404
		       struct buffer_head **primary)
{
	struct super_block *sb = inode->i_sb;
405 406 407
	struct ext4_super_block *es = EXT4_SB(sb)->s_es;
	unsigned long gdb_num = input->group / EXT4_DESC_PER_BLOCK(sb);
	ext4_fsblk_t gdblock = EXT4_SB(sb)->s_sbh->b_blocknr + 1 + gdb_num;
408 409 410
	struct buffer_head **o_group_desc, **n_group_desc;
	struct buffer_head *dind;
	int gdbackups;
411
	struct ext4_iloc iloc;
412 413 414 415 416
	__le32 *data;
	int err;

	if (test_opt(sb, DEBUG))
		printk(KERN_DEBUG
417
		       "EXT4-fs: ext4_add_new_gdb: adding group block %lu\n",
418 419 420 421 422 423 424
		       gdb_num);

	/*
	 * If we are not using the primary superblock/GDT copy don't resize,
	 * because the user tools have no way of handling this.  Probably a
	 * bad time to do it anyways.
	 */
425 426 427
	if (EXT4_SB(sb)->s_sbh->b_blocknr !=
	    le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block)) {
		ext4_warning(sb, __FUNCTION__,
428
			"won't resize using backup superblock at %llu",
429
			(unsigned long long)EXT4_SB(sb)->s_sbh->b_blocknr);
430 431 432 433 434 435 436 437 438 439 440 441
		return -EPERM;
	}

	*primary = sb_bread(sb, gdblock);
	if (!*primary)
		return -EIO;

	if ((gdbackups = verify_reserved_gdb(sb, *primary)) < 0) {
		err = gdbackups;
		goto exit_bh;
	}

442
	data = EXT4_I(inode)->i_data + EXT4_DIND_BLOCK;
443 444 445 446 447 448 449
	dind = sb_bread(sb, le32_to_cpu(*data));
	if (!dind) {
		err = -EIO;
		goto exit_bh;
	}

	data = (__le32 *)dind->b_data;
450 451
	if (le32_to_cpu(data[gdb_num % EXT4_ADDR_PER_BLOCK(sb)]) != gdblock) {
		ext4_warning(sb, __FUNCTION__,
452
			     "new group %u GDT block %llu not reserved",
453 454 455 456 457
			     input->group, gdblock);
		err = -EINVAL;
		goto exit_dind;
	}

458
	if ((err = ext4_journal_get_write_access(handle, EXT4_SB(sb)->s_sbh)))
459 460
		goto exit_dind;

461
	if ((err = ext4_journal_get_write_access(handle, *primary)))
462 463
		goto exit_sbh;

464
	if ((err = ext4_journal_get_write_access(handle, dind)))
465 466
		goto exit_primary;

467 468
	/* ext4_reserve_inode_write() gets a reference on the iloc */
	if ((err = ext4_reserve_inode_write(handle, inode, &iloc)))
469 470 471 472 473 474
		goto exit_dindj;

	n_group_desc = kmalloc((gdb_num + 1) * sizeof(struct buffer_head *),
			GFP_KERNEL);
	if (!n_group_desc) {
		err = -ENOMEM;
475
		ext4_warning (sb, __FUNCTION__,
476 477 478 479 480 481 482 483 484 485 486 487 488
			      "not enough memory for %lu groups", gdb_num + 1);
		goto exit_inode;
	}

	/*
	 * Finally, we have all of the possible failures behind us...
	 *
	 * Remove new GDT block from inode double-indirect block and clear out
	 * the new GDT block for use (which also "frees" the backup GDT blocks
	 * from the reserved inode).  We don't need to change the bitmaps for
	 * these blocks, because they are marked as in-use from being in the
	 * reserved inode, and will become GDT blocks (primary and backup).
	 */
489 490
	data[gdb_num % EXT4_ADDR_PER_BLOCK(sb)] = 0;
	ext4_journal_dirty_metadata(handle, dind);
491 492
	brelse(dind);
	inode->i_blocks -= (gdbackups + 1) * sb->s_blocksize >> 9;
493
	ext4_mark_iloc_dirty(handle, inode, &iloc);
494
	memset((*primary)->b_data, 0, sb->s_blocksize);
495
	ext4_journal_dirty_metadata(handle, *primary);
496

497
	o_group_desc = EXT4_SB(sb)->s_group_desc;
498
	memcpy(n_group_desc, o_group_desc,
499
	       EXT4_SB(sb)->s_gdb_count * sizeof(struct buffer_head *));
500
	n_group_desc[gdb_num] = *primary;
501 502
	EXT4_SB(sb)->s_group_desc = n_group_desc;
	EXT4_SB(sb)->s_gdb_count++;
503 504 505 506
	kfree(o_group_desc);

	es->s_reserved_gdt_blocks =
		cpu_to_le16(le16_to_cpu(es->s_reserved_gdt_blocks) - 1);
507
	ext4_journal_dirty_metadata(handle, EXT4_SB(sb)->s_sbh);
508 509 510 511

	return 0;

exit_inode:
512
	//ext4_journal_release_buffer(handle, iloc.bh);
513 514
	brelse(iloc.bh);
exit_dindj:
515
	//ext4_journal_release_buffer(handle, dind);
516
exit_primary:
517
	//ext4_journal_release_buffer(handle, *primary);
518
exit_sbh:
519
	//ext4_journal_release_buffer(handle, *primary);
520 521 522 523 524
exit_dind:
	brelse(dind);
exit_bh:
	brelse(*primary);

525
	ext4_debug("leaving with error %d\n", err);
526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542
	return err;
}

/*
 * Called when we are adding a new group which has a backup copy of each of
 * the GDT blocks (i.e. sparse group) and there are reserved GDT blocks.
 * We need to add these reserved backup GDT blocks to the resize inode, so
 * that they are kept for future resizing and not allocated to files.
 *
 * Each reserved backup GDT block will go into a different indirect block.
 * The indirect blocks are actually the primary reserved GDT blocks,
 * so we know in advance what their block numbers are.  We only get the
 * double-indirect block to verify it is pointing to the primary reserved
 * GDT blocks so we don't overwrite a data block by accident.  The reserved
 * backup GDT blocks are stored in their reserved primary GDT block.
 */
static int reserve_backup_gdb(handle_t *handle, struct inode *inode,
543
			      struct ext4_new_group_data *input)
544 545
{
	struct super_block *sb = inode->i_sb;
546
	int reserved_gdb =le16_to_cpu(EXT4_SB(sb)->s_es->s_reserved_gdt_blocks);
547 548
	struct buffer_head **primary;
	struct buffer_head *dind;
549 550
	struct ext4_iloc iloc;
	ext4_fsblk_t blk;
551 552 553 554 555 556 557 558 559
	__le32 *data, *end;
	int gdbackups = 0;
	int res, i;
	int err;

	primary = kmalloc(reserved_gdb * sizeof(*primary), GFP_KERNEL);
	if (!primary)
		return -ENOMEM;

560
	data = EXT4_I(inode)->i_data + EXT4_DIND_BLOCK;
561 562 563 564 565 566
	dind = sb_bread(sb, le32_to_cpu(*data));
	if (!dind) {
		err = -EIO;
		goto exit_free;
	}

567 568 569
	blk = EXT4_SB(sb)->s_sbh->b_blocknr + 1 + EXT4_SB(sb)->s_gdb_count;
	data = (__le32 *)dind->b_data + EXT4_SB(sb)->s_gdb_count;
	end = (__le32 *)dind->b_data + EXT4_ADDR_PER_BLOCK(sb);
570 571 572 573

	/* Get each reserved primary GDT block and verify it holds backups */
	for (res = 0; res < reserved_gdb; res++, blk++) {
		if (le32_to_cpu(*data) != blk) {
574
			ext4_warning(sb, __FUNCTION__,
575
				     "reserved block %llu"
576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596
				     " not at offset %ld",
				     blk,
				     (long)(data - (__le32 *)dind->b_data));
			err = -EINVAL;
			goto exit_bh;
		}
		primary[res] = sb_bread(sb, blk);
		if (!primary[res]) {
			err = -EIO;
			goto exit_bh;
		}
		if ((gdbackups = verify_reserved_gdb(sb, primary[res])) < 0) {
			brelse(primary[res]);
			err = gdbackups;
			goto exit_bh;
		}
		if (++data >= end)
			data = (__le32 *)dind->b_data;
	}

	for (i = 0; i < reserved_gdb; i++) {
597
		if ((err = ext4_journal_get_write_access(handle, primary[i]))) {
598 599 600
			/*
			int j;
			for (j = 0; j < i; j++)
601
				ext4_journal_release_buffer(handle, primary[j]);
602 603 604 605 606
			 */
			goto exit_bh;
		}
	}

607
	if ((err = ext4_reserve_inode_write(handle, inode, &iloc)))
608 609 610 611 612 613
		goto exit_bh;

	/*
	 * Finally we can add each of the reserved backup GDT blocks from
	 * the new group to its reserved primary GDT block.
	 */
614
	blk = input->group * EXT4_BLOCKS_PER_GROUP(sb);
615 616 617 618 619 620 621
	for (i = 0; i < reserved_gdb; i++) {
		int err2;
		data = (__le32 *)primary[i]->b_data;
		/* printk("reserving backup %lu[%u] = %lu\n",
		       primary[i]->b_blocknr, gdbackups,
		       blk + primary[i]->b_blocknr); */
		data[gdbackups] = cpu_to_le32(blk + primary[i]->b_blocknr);
622
		err2 = ext4_journal_dirty_metadata(handle, primary[i]);
623 624 625 626
		if (!err)
			err = err2;
	}
	inode->i_blocks += reserved_gdb * sb->s_blocksize >> 9;
627
	ext4_mark_iloc_dirty(handle, inode, &iloc);
628 629 630 631 632 633 634 635 636 637 638 639 640

exit_bh:
	while (--res >= 0)
		brelse(primary[res]);
	brelse(dind);

exit_free:
	kfree(primary);

	return err;
}

/*
641
 * Update the backup copies of the ext4 metadata.  These don't need to be part
642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657
 * of the main resize transaction, because e2fsck will re-write them if there
 * is a problem (basically only OOM will cause a problem).  However, we
 * _should_ update the backups if possible, in case the primary gets trashed
 * for some reason and we need to run e2fsck from a backup superblock.  The
 * important part is that the new block and inode counts are in the backup
 * superblocks, and the location of the new group metadata in the GDT backups.
 *
 * We do not need lock_super() for this, because these blocks are not
 * otherwise touched by the filesystem code when it is mounted.  We don't
 * need to worry about last changing from sbi->s_groups_count, because the
 * worst that can happen is that we do not copy the full number of backups
 * at this time.  The resize which changed s_groups_count will backup again.
 */
static void update_backups(struct super_block *sb,
			   int blk_off, char *data, int size)
{
658
	struct ext4_sb_info *sbi = EXT4_SB(sb);
659
	const ext4_group_t last = sbi->s_groups_count;
660
	const int bpg = EXT4_BLOCKS_PER_GROUP(sb);
661 662 663
	unsigned three = 1;
	unsigned five = 5;
	unsigned seven = 7;
664
	ext4_group_t group;
665 666 667 668
	int rest = sb->s_blocksize - size;
	handle_t *handle;
	int err = 0, err2;

669
	handle = ext4_journal_start_sb(sb, EXT4_MAX_TRANS_DATA);
670 671 672 673 674 675
	if (IS_ERR(handle)) {
		group = 1;
		err = PTR_ERR(handle);
		goto exit_err;
	}

676
	while ((group = ext4_list_backups(sb, &three, &five, &seven)) < last) {
677 678 679 680
		struct buffer_head *bh;

		/* Out of journal space, and can't get more - abort - so sad */
		if (handle->h_buffer_credits == 0 &&
681 682
		    ext4_journal_extend(handle, EXT4_MAX_TRANS_DATA) &&
		    (err = ext4_journal_restart(handle, EXT4_MAX_TRANS_DATA)))
683 684 685 686 687 688 689
			break;

		bh = sb_getblk(sb, group * bpg + blk_off);
		if (!bh) {
			err = -EIO;
			break;
		}
690
		ext4_debug("update metadata backup %#04lx\n",
691
			  (unsigned long)bh->b_blocknr);
692
		if ((err = ext4_journal_get_write_access(handle, bh)))
693 694 695 696 697 698 699
			break;
		lock_buffer(bh);
		memcpy(bh->b_data, data, size);
		if (rest)
			memset(bh->b_data + size, 0, rest);
		set_buffer_uptodate(bh);
		unlock_buffer(bh);
700
		ext4_journal_dirty_metadata(handle, bh);
701 702
		brelse(bh);
	}
703
	if ((err2 = ext4_journal_stop(handle)) && !err)
704 705 706 707 708 709 710 711 712 713 714 715 716 717
		err = err2;

	/*
	 * Ugh! Need to have e2fsck write the backup copies.  It is too
	 * late to revert the resize, we shouldn't fail just because of
	 * the backup copies (they are only needed in case of corruption).
	 *
	 * However, if we got here we have a journal problem too, so we
	 * can't really start a transaction to mark the superblock.
	 * Chicken out and just set the flag on the hope it will be written
	 * to disk, and if not - we will simply wait until next fsck.
	 */
exit_err:
	if (err) {
718
		ext4_warning(sb, __FUNCTION__,
719
			     "can't update backup for group %lu (err %d), "
720
			     "forcing fsck on next reboot", group, err);
721 722
		sbi->s_mount_state &= ~EXT4_VALID_FS;
		sbi->s_es->s_state &= cpu_to_le16(~EXT4_VALID_FS);
723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739
		mark_buffer_dirty(sbi->s_sbh);
	}
}

/* Add group descriptor data to an existing or new group descriptor block.
 * Ensure we handle all possible error conditions _before_ we start modifying
 * the filesystem, because we cannot abort the transaction and not have it
 * write the data to disk.
 *
 * If we are on a GDT block boundary, we need to get the reserved GDT block.
 * Otherwise, we may need to add backup GDT blocks for a sparse group.
 *
 * We only need to hold the superblock lock while we are actually adding
 * in the new group's counts to the superblock.  Prior to that we have
 * not really "added" the group at all.  We re-check that we are still
 * adding in the last group in case things have changed since verifying.
 */
740
int ext4_group_add(struct super_block *sb, struct ext4_new_group_data *input)
741
{
742 743 744
	struct ext4_sb_info *sbi = EXT4_SB(sb);
	struct ext4_super_block *es = sbi->s_es;
	int reserved_gdb = ext4_bg_has_super(sb, input->group) ?
745 746
		le16_to_cpu(es->s_reserved_gdt_blocks) : 0;
	struct buffer_head *primary = NULL;
747
	struct ext4_group_desc *gdp;
748 749 750 751 752
	struct inode *inode = NULL;
	handle_t *handle;
	int gdb_off, gdb_num;
	int err, err2;

753 754
	gdb_num = input->group / EXT4_DESC_PER_BLOCK(sb);
	gdb_off = input->group % EXT4_DESC_PER_BLOCK(sb);
755

756 757 758
	if (gdb_off == 0 && !EXT4_HAS_RO_COMPAT_FEATURE(sb,
					EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER)) {
		ext4_warning(sb, __FUNCTION__,
759 760 761 762
			     "Can't resize non-sparse filesystem further");
		return -EPERM;
	}

L
Laurent Vivier 已提交
763 764
	if (ext4_blocks_count(es) + input->blocks_count <
	    ext4_blocks_count(es)) {
765
		ext4_warning(sb, __FUNCTION__, "blocks_count overflow\n");
766 767 768
		return -EINVAL;
	}

769
	if (le32_to_cpu(es->s_inodes_count) + EXT4_INODES_PER_GROUP(sb) <
770
	    le32_to_cpu(es->s_inodes_count)) {
771
		ext4_warning(sb, __FUNCTION__, "inodes_count overflow\n");
772 773 774 775
		return -EINVAL;
	}

	if (reserved_gdb || gdb_off == 0) {
776 777 778
		if (!EXT4_HAS_COMPAT_FEATURE(sb,
					     EXT4_FEATURE_COMPAT_RESIZE_INODE)){
			ext4_warning(sb, __FUNCTION__,
779 780 781
				     "No reserved GDT blocks, can't resize");
			return -EPERM;
		}
782 783
		inode = ext4_iget(sb, EXT4_RESIZE_INO);
		if (IS_ERR(inode)) {
784
			ext4_warning(sb, __FUNCTION__,
785
				     "Error opening resize inode");
786
			return PTR_ERR(inode);
787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802
		}
	}

	if ((err = verify_group_input(sb, input)))
		goto exit_put;

	if ((err = setup_new_group_blocks(sb, input)))
		goto exit_put;

	/*
	 * We will always be modifying at least the superblock and a GDT
	 * block.  If we are adding a group past the last current GDT block,
	 * we will also modify the inode and the dindirect block.  If we
	 * are adding a group with superblock/GDT backups  we will also
	 * modify each of the reserved GDT dindirect blocks.
	 */
803 804
	handle = ext4_journal_start_sb(sb,
				       ext4_bg_has_super(sb, input->group) ?
805 806 807 808 809 810 811 812
				       3 + reserved_gdb : 4);
	if (IS_ERR(handle)) {
		err = PTR_ERR(handle);
		goto exit_put;
	}

	lock_super(sb);
	if (input->group != sbi->s_groups_count) {
813
		ext4_warning(sb, __FUNCTION__,
814 815 816 817 818
			     "multiple resizers run on filesystem!");
		err = -EBUSY;
		goto exit_journal;
	}

819
	if ((err = ext4_journal_get_write_access(handle, sbi->s_sbh)))
820 821 822 823 824 825 826 827 828 829
		goto exit_journal;

	/*
	 * We will only either add reserved group blocks to a backup group
	 * or remove reserved blocks for the first group in a new group block.
	 * Doing both would be mean more complex code, and sane people don't
	 * use non-sparse filesystems anymore.  This is already checked above.
	 */
	if (gdb_off) {
		primary = sbi->s_group_desc[gdb_num];
830
		if ((err = ext4_journal_get_write_access(handle, primary)))
831 832
			goto exit_journal;

833
		if (reserved_gdb && ext4_bg_num_gdb(sb, input->group) &&
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
		    (err = reserve_backup_gdb(handle, inode, input)))
			goto exit_journal;
	} else if ((err = add_new_gdb(handle, inode, input, &primary)))
		goto exit_journal;

	/*
	 * OK, now we've set up the new group.  Time to make it active.
	 *
	 * Current kernels don't lock all allocations via lock_super(),
	 * so we have to be safe wrt. concurrent accesses the group
	 * data.  So we need to be careful to set all of the relevant
	 * group descriptor data etc. *before* we enable the group.
	 *
	 * The key field here is sbi->s_groups_count: as long as
	 * that retains its old value, nobody is going to access the new
	 * group.
	 *
	 * So first we update all the descriptor metadata for the new
	 * group; then we update the total disk blocks count; then we
	 * update the groups count to enable the group; then finally we
	 * update the free space counts so that the system can start
	 * using the new disk blocks.
	 */

	/* Update group descriptor block for new group */
859
	gdp = (struct ext4_group_desc *)primary->b_data + gdb_off;
860

861 862 863
	ext4_block_bitmap_set(sb, gdp, input->block_bitmap); /* LV FIXME */
	ext4_inode_bitmap_set(sb, gdp, input->inode_bitmap); /* LV FIXME */
	ext4_inode_table_set(sb, gdp, input->inode_table); /* LV FIXME */
864
	gdp->bg_free_blocks_count = cpu_to_le16(input->free_blocks_count);
865
	gdp->bg_free_inodes_count = cpu_to_le16(EXT4_INODES_PER_GROUP(sb));
A
Andreas Dilger 已提交
866
	gdp->bg_checksum = ext4_group_desc_csum(sbi, input->group, gdp);
867 868 869 870 871 872 873 874 875 876 877

	/*
	 * Make the new blocks and inodes valid next.  We do this before
	 * increasing the group count so that once the group is enabled,
	 * all of its blocks and inodes are already valid.
	 *
	 * We always allocate group-by-group, then block-by-block or
	 * inode-by-inode within a group, so enabling these
	 * blocks/inodes before the group is live won't actually let us
	 * allocate the new space yet.
	 */
L
Laurent Vivier 已提交
878
	ext4_blocks_count_set(es, ext4_blocks_count(es) +
879 880
		input->blocks_count);
	es->s_inodes_count = cpu_to_le32(le32_to_cpu(es->s_inodes_count) +
881
		EXT4_INODES_PER_GROUP(sb));
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

	/*
	 * We need to protect s_groups_count against other CPUs seeing
	 * inconsistent state in the superblock.
	 *
	 * The precise rules we use are:
	 *
	 * * Writers of s_groups_count *must* hold lock_super
	 * AND
	 * * Writers must perform a smp_wmb() after updating all dependent
	 *   data and before modifying the groups count
	 *
	 * * Readers must hold lock_super() over the access
	 * OR
	 * * Readers must perform an smp_rmb() after reading the groups count
	 *   and before reading any dependent data.
	 *
	 * NB. These rules can be relaxed when checking the group count
	 * while freeing data, as we can only allocate from a block
	 * group after serialising against the group count, and we can
	 * only then free after serialising in turn against that
	 * allocation.
	 */
	smp_wmb();

	/* Update the global fs size fields */
	sbi->s_groups_count++;

910
	ext4_journal_dirty_metadata(handle, primary);
911 912 913

	/* Update the reserved block counts only once the new group is
	 * active. */
L
Laurent Vivier 已提交
914
	ext4_r_blocks_count_set(es, ext4_r_blocks_count(es) +
915 916 917
		input->reserved_blocks);

	/* Update the free space counts */
P
Peter Zijlstra 已提交
918
	percpu_counter_add(&sbi->s_freeblocks_counter,
919
			   input->free_blocks_count);
P
Peter Zijlstra 已提交
920
	percpu_counter_add(&sbi->s_freeinodes_counter,
921
			   EXT4_INODES_PER_GROUP(sb));
922

923
	ext4_journal_dirty_metadata(handle, sbi->s_sbh);
924 925 926 927
	sb->s_dirt = 1;

exit_journal:
	unlock_super(sb);
928
	if ((err2 = ext4_journal_stop(handle)) && !err)
929 930 931
		err = err2;
	if (!err) {
		update_backups(sb, sbi->s_sbh->b_blocknr, (char *)es,
932
			       sizeof(struct ext4_super_block));
933 934 935 936 937 938
		update_backups(sb, primary->b_blocknr, primary->b_data,
			       primary->b_size);
	}
exit_put:
	iput(inode);
	return err;
939
} /* ext4_group_add */
940 941 942 943 944 945

/* Extend the filesystem to the new number of blocks specified.  This entry
 * point is only used to extend the current filesystem to the end of the last
 * existing group.  It can be accessed via ioctl, or by "remount,resize=<size>"
 * for emergencies (because it has no dependencies on reserved blocks).
 *
946
 * If we _really_ wanted, we could use default values to call ext4_group_add()
947 948 949
 * allow the "remount" trick to work for arbitrary resizing, assuming enough
 * GDT blocks are reserved to grow to the desired size.
 */
950 951
int ext4_group_extend(struct super_block *sb, struct ext4_super_block *es,
		      ext4_fsblk_t n_blocks_count)
952
{
953
	ext4_fsblk_t o_blocks_count;
954
	ext4_group_t o_groups_count;
955 956
	ext4_grpblk_t last;
	ext4_grpblk_t add;
957 958 959 960 961 962 963 964
	struct buffer_head * bh;
	handle_t *handle;
	int err;
	unsigned long freed_blocks;

	/* We don't need to worry about locking wrt other resizers just
	 * yet: we're going to revalidate es->s_blocks_count after
	 * taking lock_super() below. */
L
Laurent Vivier 已提交
965
	o_blocks_count = ext4_blocks_count(es);
966
	o_groups_count = EXT4_SB(sb)->s_groups_count;
967 968

	if (test_opt(sb, DEBUG))
969
		printk(KERN_DEBUG "EXT4-fs: extending last group from %llu uto %llu blocks\n",
970 971 972 973 974 975
		       o_blocks_count, n_blocks_count);

	if (n_blocks_count == 0 || n_blocks_count == o_blocks_count)
		return 0;

	if (n_blocks_count > (sector_t)(~0ULL) >> (sb->s_blocksize_bits - 9)) {
976
		printk(KERN_ERR "EXT4-fs: filesystem on %s:"
977
			" too large to resize to %llu blocks safely\n",
978 979
			sb->s_id, n_blocks_count);
		if (sizeof(sector_t) < 8)
980
			ext4_warning(sb, __FUNCTION__,
981 982 983 984 985
			"CONFIG_LBD not enabled\n");
		return -EINVAL;
	}

	if (n_blocks_count < o_blocks_count) {
986
		ext4_warning(sb, __FUNCTION__,
987 988 989 990 991
			     "can't shrink FS - resize aborted");
		return -EBUSY;
	}

	/* Handle the remaining blocks in the last group only. */
992
	ext4_get_group_no_and_offset(sb, o_blocks_count, NULL, &last);
993 994

	if (last == 0) {
995
		ext4_warning(sb, __FUNCTION__,
996 997 998 999
			     "need to use ext2online to resize further");
		return -EPERM;
	}

1000
	add = EXT4_BLOCKS_PER_GROUP(sb) - last;
1001 1002

	if (o_blocks_count + add < o_blocks_count) {
1003
		ext4_warning(sb, __FUNCTION__, "blocks_count overflow");
1004 1005 1006 1007 1008 1009 1010
		return -EINVAL;
	}

	if (o_blocks_count + add > n_blocks_count)
		add = n_blocks_count - o_blocks_count;

	if (o_blocks_count + add < n_blocks_count)
1011
		ext4_warning(sb, __FUNCTION__,
1012
			     "will only finish group (%llu"
1013 1014 1015 1016 1017 1018
			     " blocks, %u new)",
			     o_blocks_count + add, add);

	/* See if the device is actually as big as what was requested */
	bh = sb_bread(sb, o_blocks_count + add -1);
	if (!bh) {
1019
		ext4_warning(sb, __FUNCTION__,
1020 1021 1022 1023 1024 1025
			     "can't read last block, resize aborted");
		return -ENOSPC;
	}
	brelse(bh);

	/* We will update the superblock, one block bitmap, and
1026
	 * one group descriptor via ext4_free_blocks().
1027
	 */
1028
	handle = ext4_journal_start_sb(sb, 3);
1029 1030
	if (IS_ERR(handle)) {
		err = PTR_ERR(handle);
1031
		ext4_warning(sb, __FUNCTION__, "error %d on journal start",err);
1032 1033 1034 1035
		goto exit_put;
	}

	lock_super(sb);
L
Laurent Vivier 已提交
1036
	if (o_blocks_count != ext4_blocks_count(es)) {
1037
		ext4_warning(sb, __FUNCTION__,
1038 1039 1040 1041 1042 1043
			     "multiple resizers run on filesystem!");
		unlock_super(sb);
		err = -EBUSY;
		goto exit_put;
	}

1044 1045 1046
	if ((err = ext4_journal_get_write_access(handle,
						 EXT4_SB(sb)->s_sbh))) {
		ext4_warning(sb, __FUNCTION__,
1047 1048
			     "error %d on journal write access", err);
		unlock_super(sb);
1049
		ext4_journal_stop(handle);
1050 1051
		goto exit_put;
	}
L
Laurent Vivier 已提交
1052
	ext4_blocks_count_set(es, o_blocks_count + add);
1053
	ext4_journal_dirty_metadata(handle, EXT4_SB(sb)->s_sbh);
1054 1055
	sb->s_dirt = 1;
	unlock_super(sb);
E
Eric Sandeen 已提交
1056
	ext4_debug("freeing blocks %llu through %llu\n", o_blocks_count,
1057
		   o_blocks_count + add);
1058
	ext4_free_blocks_sb(handle, sb, o_blocks_count, add, &freed_blocks);
1059
	ext4_debug("freed blocks %llu through %llu\n", o_blocks_count,
1060
		   o_blocks_count + add);
1061
	if ((err = ext4_journal_stop(handle)))
1062 1063
		goto exit_put;
	if (test_opt(sb, DEBUG))
L
Laurent Vivier 已提交
1064 1065
		printk(KERN_DEBUG "EXT4-fs: extended group to %llu blocks\n",
		       ext4_blocks_count(es));
1066 1067
	update_backups(sb, EXT4_SB(sb)->s_sbh->b_blocknr, (char *)es,
		       sizeof(struct ext4_super_block));
1068 1069
exit_put:
	return err;
1070
} /* ext4_group_extend */