resize.c 33.3 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 15 16

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

17
#include "ext4_jbd2.h"
18

19 20 21 22 23 24 25
int ext4_resize_begin(struct super_block *sb)
{
	int ret = 0;

	if (!capable(CAP_SYS_RESOURCE))
		return -EPERM;

26 27 28 29 30 31 32 33 34 35
	/*
	 * We are not allowed to do online-resizing on a filesystem mounted
	 * with error, because it can destroy the filesystem easily.
	 */
	if (EXT4_SB(sb)->s_mount_state & EXT4_ERROR_FS) {
		ext4_warning(sb, "There are errors in the filesystem, "
			     "so online resizing is not allowed\n");
		return -EPERM;
	}

36 37 38 39 40 41 42 43 44 45 46 47
	if (test_and_set_bit_lock(EXT4_RESIZING, &EXT4_SB(sb)->s_resize_flags))
		ret = -EBUSY;

	return ret;
}

void ext4_resize_end(struct super_block *sb)
{
	clear_bit_unlock(EXT4_RESIZING, &EXT4_SB(sb)->s_resize_flags);
	smp_mb__after_clear_bit();
}

48 49 50 51
#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,
52
			      struct ext4_new_group_data *input)
53
{
54 55
	struct ext4_sb_info *sbi = EXT4_SB(sb);
	struct ext4_super_block *es = sbi->s_es;
L
Laurent Vivier 已提交
56
	ext4_fsblk_t start = ext4_blocks_count(es);
57
	ext4_fsblk_t end = start + input->blocks_count;
58
	ext4_group_t group = input->group;
59 60 61
	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) +
62
		 le16_to_cpu(es->s_reserved_gdt_blocks)) : 0;
63
	ext4_fsblk_t metaend = start + overhead;
64
	struct buffer_head *bh = NULL;
65
	ext4_grpblk_t free_blocks_count, offset;
66 67 68 69 70 71
	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))
72
		printk(KERN_DEBUG "EXT4-fs: adding %s group %u: %u blocks "
73
		       "(%d free, %u reserved)\n",
74
		       ext4_bg_has_super(sb, input->group) ? "normal" :
75 76 77
		       "no-super", input->group, input->blocks_count,
		       free_blocks_count, input->reserved_blocks);

78
	ext4_get_group_no_and_offset(sb, start, NULL, &offset);
79
	if (group != sbi->s_groups_count)
80
		ext4_warning(sb, "Cannot add at group %u (only %u groups)",
81
			     input->group, sbi->s_groups_count);
82
	else if (offset != 0)
83
			ext4_warning(sb, "Last group not full");
84
	else if (input->reserved_blocks > input->blocks_count / 5)
85
		ext4_warning(sb, "Reserved blocks too high (%u)",
86 87
			     input->reserved_blocks);
	else if (free_blocks_count < 0)
88
		ext4_warning(sb, "Bad blocks count %u",
89 90
			     input->blocks_count);
	else if (!(bh = sb_bread(sb, end - 1)))
91
		ext4_warning(sb, "Cannot read last block (%llu)",
92 93
			     end - 1);
	else if (outside(input->block_bitmap, start, end))
94
		ext4_warning(sb, "Block bitmap not in group (block %llu)",
95
			     (unsigned long long)input->block_bitmap);
96
	else if (outside(input->inode_bitmap, start, end))
97
		ext4_warning(sb, "Inode bitmap not in group (block %llu)",
98
			     (unsigned long long)input->inode_bitmap);
99
	else if (outside(input->inode_table, start, end) ||
100
		 outside(itend - 1, start, end))
101
		ext4_warning(sb, "Inode table not in group (blocks %llu-%llu)",
102
			     (unsigned long long)input->inode_table, itend - 1);
103
	else if (input->inode_bitmap == input->block_bitmap)
104
		ext4_warning(sb, "Block bitmap same as inode bitmap (%llu)",
105
			     (unsigned long long)input->block_bitmap);
106
	else if (inside(input->block_bitmap, input->inode_table, itend))
107 108
		ext4_warning(sb, "Block bitmap (%llu) in inode table "
			     "(%llu-%llu)",
109 110
			     (unsigned long long)input->block_bitmap,
			     (unsigned long long)input->inode_table, itend - 1);
111
	else if (inside(input->inode_bitmap, input->inode_table, itend))
112 113
		ext4_warning(sb, "Inode bitmap (%llu) in inode table "
			     "(%llu-%llu)",
114 115
			     (unsigned long long)input->inode_bitmap,
			     (unsigned long long)input->inode_table, itend - 1);
116
	else if (inside(input->block_bitmap, start, metaend))
117
		ext4_warning(sb, "Block bitmap (%llu) in GDT table (%llu-%llu)",
118 119
			     (unsigned long long)input->block_bitmap,
			     start, metaend - 1);
120
	else if (inside(input->inode_bitmap, start, metaend))
121
		ext4_warning(sb, "Inode bitmap (%llu) in GDT table (%llu-%llu)",
122 123
			     (unsigned long long)input->inode_bitmap,
			     start, metaend - 1);
124
	else if (inside(input->inode_table, start, metaend) ||
125
		 inside(itend - 1, start, metaend))
126 127
		ext4_warning(sb, "Inode table (%llu-%llu) overlaps GDT table "
			     "(%llu-%llu)",
128 129
			     (unsigned long long)input->inode_table,
			     itend - 1, start, metaend - 1);
130 131 132 133 134 135 136 137
	else
		err = 0;
	brelse(bh);

	return err;
}

static struct buffer_head *bclean(handle_t *handle, struct super_block *sb,
138
				  ext4_fsblk_t blk)
139 140 141 142 143 144 145
{
	struct buffer_head *bh;
	int err;

	bh = sb_getblk(sb, blk);
	if (!bh)
		return ERR_PTR(-EIO);
146
	if ((err = ext4_journal_get_write_access(handle, bh))) {
147 148 149 150 151 152 153 154 155 156
		brelse(bh);
		bh = ERR_PTR(err);
	} else {
		memset(bh->b_data, 0, sb->s_blocksize);
		set_buffer_uptodate(bh);
	}

	return bh;
}

157 158 159 160 161
/*
 * 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.
 */
162
static int extend_or_restart_transaction(handle_t *handle, int thresh)
163 164 165
{
	int err;

166
	if (ext4_handle_has_enough_credits(handle, thresh))
167 168 169 170 171 172
		return 0;

	err = ext4_journal_extend(handle, EXT4_MAX_TRANS_DATA);
	if (err < 0)
		return err;
	if (err) {
173 174
		err = ext4_journal_restart(handle, EXT4_MAX_TRANS_DATA);
		if (err)
175
			return err;
176
	}
177 178 179 180

	return 0;
}

181 182 183 184 185 186 187 188
/*
 * 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,
189
				  struct ext4_new_group_data *input)
190
{
191 192 193
	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) ?
194
		le16_to_cpu(sbi->s_es->s_reserved_gdt_blocks) : 0;
195
	unsigned long gdblocks = ext4_bg_num_gdb(sb, input->group);
196 197
	struct buffer_head *bh;
	handle_t *handle;
198 199
	ext4_fsblk_t block;
	ext4_grpblk_t bit;
200 201 202
	int i;
	int err = 0, err2;

203 204 205
	/* This transaction may be extended/restarted along the way */
	handle = ext4_journal_start_sb(sb, EXT4_MAX_TRANS_DATA);

206 207 208
	if (IS_ERR(handle))
		return PTR_ERR(handle);

209
	BUG_ON(input->group != sbi->s_groups_count);
210 211 212 213 214 215

	/* 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
216
		ext4_debug("update backup group %#04llx (+%d)\n", block, bit);
217 218 219
		err = extend_or_restart_transaction(handle, 1);
		if (err)
			goto exit_journal;
220

221 222 223
		gdb = sb_getblk(sb, block);
		if (!gdb) {
			err = -EIO;
224
			goto exit_journal;
225
		}
226
		if ((err = ext4_journal_get_write_access(handle, gdb))) {
227
			brelse(gdb);
228
			goto exit_journal;
229
		}
230
		memcpy(gdb->b_data, sbi->s_group_desc[i]->b_data, gdb->b_size);
231
		set_buffer_uptodate(gdb);
232 233 234
		err = ext4_handle_dirty_metadata(handle, NULL, gdb);
		if (unlikely(err)) {
			brelse(gdb);
235
			goto exit_journal;
236
		}
237 238 239 240
		brelse(gdb);
	}

	/* Zero out all of the reserved backup group descriptor table blocks */
241
	ext4_debug("clear inode table blocks %#04llx -> %#04lx\n",
242 243
			block, sbi->s_itb_per_group);
	err = sb_issue_zeroout(sb, gdblocks + start + 1, reserved_gdb,
244
			       GFP_NOFS);
245
	if (err)
246 247 248 249 250 251 252 253 254 255 256
		goto exit_journal;

	err = extend_or_restart_transaction(handle, 2);
	if (err)
		goto exit_journal;

	bh = bclean(handle, sb, input->block_bitmap);
	if (IS_ERR(bh)) {
		err = PTR_ERR(bh);
		goto exit_journal;
	}
257 258 259 260 261

	if (ext4_bg_has_super(sb, input->group)) {
		ext4_debug("mark backup group tables %#04llx (+0)\n", start);
		ext4_set_bits(bh->b_data, 0, gdblocks + reserved_gdb + 1);
	}
262

E
Eric Sandeen 已提交
263
	ext4_debug("mark block bitmap %#04llx (+%llu)\n", input->block_bitmap,
264
		   input->block_bitmap - start);
265
	ext4_set_bit(input->block_bitmap - start, bh->b_data);
E
Eric Sandeen 已提交
266
	ext4_debug("mark inode bitmap %#04llx (+%llu)\n", input->inode_bitmap,
267
		   input->inode_bitmap - start);
268
	ext4_set_bit(input->inode_bitmap - start, bh->b_data);
269 270

	/* Zero out all of the inode table blocks */
271
	block = input->inode_table;
272
	ext4_debug("clear inode table blocks %#04llx -> %#04lx\n",
273
			block, sbi->s_itb_per_group);
274
	err = sb_issue_zeroout(sb, block, sbi->s_itb_per_group, GFP_NOFS);
275 276
	if (err)
		goto exit_bh;
277 278
	ext4_set_bits(bh->b_data, input->inode_table - start,
		      sbi->s_itb_per_group);
279 280


281 282
	ext4_mark_bitmap_end(input->blocks_count, sb->s_blocksize * 8,
			     bh->b_data);
283 284 285 286 287
	err = ext4_handle_dirty_metadata(handle, NULL, bh);
	if (unlikely(err)) {
		ext4_std_error(sb, err);
		goto exit_bh;
	}
288 289
	brelse(bh);
	/* Mark unused entries in inode bitmap used */
E
Eric Sandeen 已提交
290
	ext4_debug("clear inode bitmap %#04llx (+%llu)\n",
291 292 293 294 295 296
		   input->inode_bitmap, input->inode_bitmap - start);
	if (IS_ERR(bh = bclean(handle, sb, input->inode_bitmap))) {
		err = PTR_ERR(bh);
		goto exit_journal;
	}

297 298
	ext4_mark_bitmap_end(EXT4_INODES_PER_GROUP(sb), sb->s_blocksize * 8,
			     bh->b_data);
299 300 301
	err = ext4_handle_dirty_metadata(handle, NULL, bh);
	if (unlikely(err))
		ext4_std_error(sb, err);
302 303 304 305
exit_bh:
	brelse(bh);

exit_journal:
306
	if ((err2 = ext4_journal_stop(handle)) && !err)
307 308 309 310 311 312 313
		err = err2;

	return err;
}

/*
 * Iterate through the groups which hold BACKUP superblock/GDT copies in an
314
 * ext4 filesystem.  The counters should be initialized to 1, 5, and 7 before
315 316 317 318
 * 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, ...
 */
319
static unsigned ext4_list_backups(struct super_block *sb, unsigned *three,
320 321 322 323 324 325
				  unsigned *five, unsigned *seven)
{
	unsigned *min = three;
	int mult = 3;
	unsigned ret;

326 327
	if (!EXT4_HAS_RO_COMPAT_FEATURE(sb,
					EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER)) {
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
		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)
{
356
	const ext4_fsblk_t blk = primary->b_blocknr;
357
	const ext4_group_t end = EXT4_SB(sb)->s_groups_count;
358 359 360 361 362 363 364
	unsigned three = 1;
	unsigned five = 5;
	unsigned seven = 7;
	unsigned grp;
	__le32 *p = (__le32 *)primary->b_data;
	int gdbackups = 0;

365
	while ((grp = ext4_list_backups(sb, &three, &five, &seven)) < end) {
L
Laurent Vivier 已提交
366 367
		if (le32_to_cpu(*p++) !=
		    grp * EXT4_BLOCKS_PER_GROUP(sb) + blk){
368
			ext4_warning(sb, "reserved GDT %llu"
369
				     " missing grp %d (%llu)",
370
				     blk, grp,
L
Laurent Vivier 已提交
371 372 373
				     grp *
				     (ext4_fsblk_t)EXT4_BLOCKS_PER_GROUP(sb) +
				     blk);
374 375
			return -EINVAL;
		}
376
		if (++gdbackups > EXT4_ADDR_PER_BLOCK(sb))
377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396
			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,
397
		       ext4_group_t group)
398 399
{
	struct super_block *sb = inode->i_sb;
400
	struct ext4_super_block *es = EXT4_SB(sb)->s_es;
401
	unsigned long gdb_num = group / EXT4_DESC_PER_BLOCK(sb);
402
	ext4_fsblk_t gdblock = EXT4_SB(sb)->s_sbh->b_blocknr + 1 + gdb_num;
403 404
	struct buffer_head **o_group_desc, **n_group_desc;
	struct buffer_head *dind;
405
	struct buffer_head *gdb_bh;
406
	int gdbackups;
407
	struct ext4_iloc iloc;
408 409 410 411 412
	__le32 *data;
	int err;

	if (test_opt(sb, DEBUG))
		printk(KERN_DEBUG
413
		       "EXT4-fs: ext4_add_new_gdb: adding group block %lu\n",
414 415
		       gdb_num);

416 417
	/*
	 * If we are not using the primary superblock/GDT copy don't resize,
418 419 420
         * because the user tools have no way of handling this.  Probably a
         * bad time to do it anyways.
         */
421 422
	if (EXT4_SB(sb)->s_sbh->b_blocknr !=
	    le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block)) {
423
		ext4_warning(sb, "won't resize using backup superblock at %llu",
424
			(unsigned long long)EXT4_SB(sb)->s_sbh->b_blocknr);
425 426 427
		return -EPERM;
	}

428 429
	gdb_bh = sb_bread(sb, gdblock);
	if (!gdb_bh)
430 431
		return -EIO;

432 433
	gdbackups = verify_reserved_gdb(sb, gdb_bh);
	if (gdbackups < 0) {
434 435 436 437
		err = gdbackups;
		goto exit_bh;
	}

438
	data = EXT4_I(inode)->i_data + EXT4_DIND_BLOCK;
439 440 441 442 443 444 445
	dind = sb_bread(sb, le32_to_cpu(*data));
	if (!dind) {
		err = -EIO;
		goto exit_bh;
	}

	data = (__le32 *)dind->b_data;
446
	if (le32_to_cpu(data[gdb_num % EXT4_ADDR_PER_BLOCK(sb)]) != gdblock) {
447
		ext4_warning(sb, "new group %u GDT block %llu not reserved",
448
			     group, gdblock);
449 450 451 452
		err = -EINVAL;
		goto exit_dind;
	}

453 454
	err = ext4_journal_get_write_access(handle, EXT4_SB(sb)->s_sbh);
	if (unlikely(err))
455 456
		goto exit_dind;

457
	err = ext4_journal_get_write_access(handle, gdb_bh);
458
	if (unlikely(err))
459 460
		goto exit_sbh;

461 462 463
	err = ext4_journal_get_write_access(handle, dind);
	if (unlikely(err))
		ext4_std_error(sb, err);
464

465
	/* ext4_reserve_inode_write() gets a reference on the iloc */
466 467
	err = ext4_reserve_inode_write(handle, inode, &iloc);
	if (unlikely(err))
468 469
		goto exit_dindj;

470 471 472
	n_group_desc = ext4_kvmalloc((gdb_num + 1) *
				     sizeof(struct buffer_head *),
				     GFP_NOFS);
473 474
	if (!n_group_desc) {
		err = -ENOMEM;
475 476
		ext4_warning(sb, "not enough memory for %lu groups",
			     gdb_num + 1);
477 478 479 480 481 482 483 484 485 486 487 488
		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
	data[gdb_num % EXT4_ADDR_PER_BLOCK(sb)] = 0;
490 491 492 493 494
	err = ext4_handle_dirty_metadata(handle, NULL, dind);
	if (unlikely(err)) {
		ext4_std_error(sb, err);
		goto exit_inode;
	}
495
	inode->i_blocks -= (gdbackups + 1) * sb->s_blocksize >> 9;
496
	ext4_mark_iloc_dirty(handle, inode, &iloc);
497 498
	memset(gdb_bh->b_data, 0, sb->s_blocksize);
	err = ext4_handle_dirty_metadata(handle, NULL, gdb_bh);
499 500 501 502 503
	if (unlikely(err)) {
		ext4_std_error(sb, err);
		goto exit_inode;
	}
	brelse(dind);
504

505
	o_group_desc = EXT4_SB(sb)->s_group_desc;
506
	memcpy(n_group_desc, o_group_desc,
507
	       EXT4_SB(sb)->s_gdb_count * sizeof(struct buffer_head *));
508
	n_group_desc[gdb_num] = gdb_bh;
509 510
	EXT4_SB(sb)->s_group_desc = n_group_desc;
	EXT4_SB(sb)->s_gdb_count++;
511
	ext4_kvfree(o_group_desc);
512

M
Marcin Slusarz 已提交
513
	le16_add_cpu(&es->s_reserved_gdt_blocks, -1);
514 515 516
	err = ext4_handle_dirty_metadata(handle, NULL, EXT4_SB(sb)->s_sbh);
	if (err)
		ext4_std_error(sb, err);
517

518
	return err;
519 520

exit_inode:
521
	ext4_kvfree(n_group_desc);
522
	/* ext4_handle_release_buffer(handle, iloc.bh); */
523 524
	brelse(iloc.bh);
exit_dindj:
525
	/* ext4_handle_release_buffer(handle, dind); */
526
exit_sbh:
527
	/* ext4_handle_release_buffer(handle, EXT4_SB(sb)->s_sbh); */
528 529 530
exit_dind:
	brelse(dind);
exit_bh:
531
	brelse(gdb_bh);
532

533
	ext4_debug("leaving with error %d\n", err);
534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550
	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,
551
			      ext4_group_t group)
552 553
{
	struct super_block *sb = inode->i_sb;
554
	int reserved_gdb =le16_to_cpu(EXT4_SB(sb)->s_es->s_reserved_gdt_blocks);
555 556
	struct buffer_head **primary;
	struct buffer_head *dind;
557 558
	struct ext4_iloc iloc;
	ext4_fsblk_t blk;
559 560 561 562 563
	__le32 *data, *end;
	int gdbackups = 0;
	int res, i;
	int err;

564
	primary = kmalloc(reserved_gdb * sizeof(*primary), GFP_NOFS);
565 566 567
	if (!primary)
		return -ENOMEM;

568
	data = EXT4_I(inode)->i_data + EXT4_DIND_BLOCK;
569 570 571 572 573 574
	dind = sb_bread(sb, le32_to_cpu(*data));
	if (!dind) {
		err = -EIO;
		goto exit_free;
	}

575
	blk = EXT4_SB(sb)->s_sbh->b_blocknr + 1 + EXT4_SB(sb)->s_gdb_count;
J
Josef Bacik 已提交
576 577
	data = (__le32 *)dind->b_data + (EXT4_SB(sb)->s_gdb_count %
					 EXT4_ADDR_PER_BLOCK(sb));
578
	end = (__le32 *)dind->b_data + EXT4_ADDR_PER_BLOCK(sb);
579 580 581 582

	/* 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) {
583
			ext4_warning(sb, "reserved block %llu"
584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604
				     " 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++) {
605
		if ((err = ext4_journal_get_write_access(handle, primary[i]))) {
606 607 608
			/*
			int j;
			for (j = 0; j < i; j++)
609
				ext4_handle_release_buffer(handle, primary[j]);
610 611 612 613 614
			 */
			goto exit_bh;
		}
	}

615
	if ((err = ext4_reserve_inode_write(handle, inode, &iloc)))
616 617 618 619 620 621
		goto exit_bh;

	/*
	 * Finally we can add each of the reserved backup GDT blocks from
	 * the new group to its reserved primary GDT block.
	 */
622
	blk = group * EXT4_BLOCKS_PER_GROUP(sb);
623 624 625 626 627 628 629
	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);
630
		err2 = ext4_handle_dirty_metadata(handle, NULL, primary[i]);
631 632 633 634
		if (!err)
			err = err2;
	}
	inode->i_blocks += reserved_gdb * sb->s_blocksize >> 9;
635
	ext4_mark_iloc_dirty(handle, inode, &iloc);
636 637 638 639 640 641 642 643 644 645 646 647 648

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

exit_free:
	kfree(primary);

	return err;
}

/*
649
 * Update the backup copies of the ext4 metadata.  These don't need to be part
650 651 652 653 654 655 656
 * 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.
 *
657 658 659 660 661 662
 * We do not need take the s_resize_lock 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.
663 664 665 666
 */
static void update_backups(struct super_block *sb,
			   int blk_off, char *data, int size)
{
667
	struct ext4_sb_info *sbi = EXT4_SB(sb);
668
	const ext4_group_t last = sbi->s_groups_count;
669
	const int bpg = EXT4_BLOCKS_PER_GROUP(sb);
670 671 672
	unsigned three = 1;
	unsigned five = 5;
	unsigned seven = 7;
673
	ext4_group_t group;
674 675 676 677
	int rest = sb->s_blocksize - size;
	handle_t *handle;
	int err = 0, err2;

678
	handle = ext4_journal_start_sb(sb, EXT4_MAX_TRANS_DATA);
679 680 681 682 683 684
	if (IS_ERR(handle)) {
		group = 1;
		err = PTR_ERR(handle);
		goto exit_err;
	}

685
	while ((group = ext4_list_backups(sb, &three, &five, &seven)) < last) {
686 687 688
		struct buffer_head *bh;

		/* Out of journal space, and can't get more - abort - so sad */
689 690
		if (ext4_handle_valid(handle) &&
		    handle->h_buffer_credits == 0 &&
691 692
		    ext4_journal_extend(handle, EXT4_MAX_TRANS_DATA) &&
		    (err = ext4_journal_restart(handle, EXT4_MAX_TRANS_DATA)))
693 694 695 696 697 698 699
			break;

		bh = sb_getblk(sb, group * bpg + blk_off);
		if (!bh) {
			err = -EIO;
			break;
		}
700
		ext4_debug("update metadata backup %#04lx\n",
701
			  (unsigned long)bh->b_blocknr);
702
		if ((err = ext4_journal_get_write_access(handle, bh)))
703 704 705 706 707 708 709
			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);
710 711 712
		err = ext4_handle_dirty_metadata(handle, NULL, bh);
		if (unlikely(err))
			ext4_std_error(sb, err);
713 714
		brelse(bh);
	}
715
	if ((err2 = ext4_journal_stop(handle)) && !err)
716 717 718 719 720 721 722 723 724 725 726 727 728 729
		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) {
730
		ext4_warning(sb, "can't update backup for group %u (err %d), "
731
			     "forcing fsck on next reboot", group, err);
732 733
		sbi->s_mount_state &= ~EXT4_VALID_FS;
		sbi->s_es->s_state &= cpu_to_le16(~EXT4_VALID_FS);
734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750
		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.
 */
751
int ext4_group_add(struct super_block *sb, struct ext4_new_group_data *input)
752
{
753 754 755
	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) ?
756 757
		le16_to_cpu(es->s_reserved_gdt_blocks) : 0;
	struct buffer_head *primary = NULL;
758
	struct ext4_group_desc *gdp;
759 760 761 762 763
	struct inode *inode = NULL;
	handle_t *handle;
	int gdb_off, gdb_num;
	int err, err2;

764 765
	gdb_num = input->group / EXT4_DESC_PER_BLOCK(sb);
	gdb_off = input->group % EXT4_DESC_PER_BLOCK(sb);
766

767 768
	if (gdb_off == 0 && !EXT4_HAS_RO_COMPAT_FEATURE(sb,
					EXT4_FEATURE_RO_COMPAT_SPARSE_SUPER)) {
769
		ext4_warning(sb, "Can't resize non-sparse filesystem further");
770 771 772
		return -EPERM;
	}

L
Laurent Vivier 已提交
773 774
	if (ext4_blocks_count(es) + input->blocks_count <
	    ext4_blocks_count(es)) {
775
		ext4_warning(sb, "blocks_count overflow");
776 777 778
		return -EINVAL;
	}

779
	if (le32_to_cpu(es->s_inodes_count) + EXT4_INODES_PER_GROUP(sb) <
780
	    le32_to_cpu(es->s_inodes_count)) {
781
		ext4_warning(sb, "inodes_count overflow");
782 783 784 785
		return -EINVAL;
	}

	if (reserved_gdb || gdb_off == 0) {
786
		if (!EXT4_HAS_COMPAT_FEATURE(sb,
787 788
					     EXT4_FEATURE_COMPAT_RESIZE_INODE)
		    || !le16_to_cpu(es->s_reserved_gdt_blocks)) {
789
			ext4_warning(sb,
790 791 792
				     "No reserved GDT blocks, can't resize");
			return -EPERM;
		}
793 794
		inode = ext4_iget(sb, EXT4_RESIZE_INO);
		if (IS_ERR(inode)) {
795
			ext4_warning(sb, "Error opening resize inode");
796
			return PTR_ERR(inode);
797 798 799
		}
	}

800

801 802 803 804 805 806 807 808 809 810 811 812 813
	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.
	 */
814 815
	handle = ext4_journal_start_sb(sb,
				       ext4_bg_has_super(sb, input->group) ?
816 817 818 819 820 821
				       3 + reserved_gdb : 4);
	if (IS_ERR(handle)) {
		err = PTR_ERR(handle);
		goto exit_put;
	}

822
	if ((err = ext4_journal_get_write_access(handle, sbi->s_sbh)))
823 824
		goto exit_journal;

825 826 827 828 829 830
        /*
         * 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.
         */
831 832
	if (gdb_off) {
		primary = sbi->s_group_desc[gdb_num];
833
		if ((err = ext4_journal_get_write_access(handle, primary)))
834 835
			goto exit_journal;

836 837 838 839 840
		if (reserved_gdb && ext4_bg_num_gdb(sb, input->group)) {
			err = reserve_backup_gdb(handle, inode, input->group);
			if (err)
				goto exit_journal;
		}
841 842 843 844 845 846 847 848 849 850
	} else {
		/*
		 * Note that we can access new group descriptor block safely
		 * only if add_new_gdb() succeeds.
		 */
		err = add_new_gdb(handle, inode, input->group);
		if (err)
			goto exit_journal;
		primary = sbi->s_group_desc[gdb_num];
	}
851

852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868
        /*
         * OK, now we've set up the new group.  Time to make it active.
         *
         * 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.
         */
869 870

	/* Update group descriptor block for new group */
871 872
	gdp = (struct ext4_group_desc *)((char *)primary->b_data +
					 gdb_off * EXT4_DESC_SIZE(sb));
873

874
	memset(gdp, 0, EXT4_DESC_SIZE(sb));
875 876 877
	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 */
878 879
	ext4_free_blks_set(sb, gdp, input->free_blocks_count);
	ext4_free_inodes_set(sb, gdp, EXT4_INODES_PER_GROUP(sb));
880
	gdp->bg_flags = cpu_to_le16(EXT4_BG_INODE_ZEROED);
A
Andreas Dilger 已提交
881
	gdp->bg_checksum = ext4_group_desc_csum(sbi, input->group, gdp);
882

883 884 885 886
	/*
	 * We can allocate memory for mb_alloc based on the new group
	 * descriptor
	 */
887
	err = ext4_mb_add_groupinfo(sb, input->group, gdp);
888
	if (err)
889 890
		goto exit_journal;

891 892 893 894 895 896 897 898 899 900
	/*
	 * 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
901
	ext4_blocks_count_set(es, ext4_blocks_count(es) +
902
		input->blocks_count);
M
Marcin Slusarz 已提交
903
	le32_add_cpu(&es->s_inodes_count, EXT4_INODES_PER_GROUP(sb));
904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927

	/*
	 * We need to protect s_groups_count against other CPUs seeing
	 * inconsistent state in the superblock.
	 *
	 * The precise rules we use are:
	 *
	 * * Writers must perform a smp_wmb() after updating all dependent
	 *   data and before modifying the groups count
	 *
	 * * 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++;

928 929 930 931 932
	err = ext4_handle_dirty_metadata(handle, NULL, primary);
	if (unlikely(err)) {
		ext4_std_error(sb, err);
		goto exit_journal;
	}
933 934 935

	/* Update the reserved block counts only once the new group is
	 * active. */
L
Laurent Vivier 已提交
936
	ext4_r_blocks_count_set(es, ext4_r_blocks_count(es) +
937 938 939
		input->reserved_blocks);

	/* Update the free space counts */
P
Peter Zijlstra 已提交
940
	percpu_counter_add(&sbi->s_freeblocks_counter,
941
			   input->free_blocks_count);
P
Peter Zijlstra 已提交
942
	percpu_counter_add(&sbi->s_freeinodes_counter,
943
			   EXT4_INODES_PER_GROUP(sb));
944

945 946
	if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_FLEX_BG) &&
	    sbi->s_log_groups_per_flex) {
947 948
		ext4_group_t flex_group;
		flex_group = ext4_flex_group(sbi, input->group);
949 950 951 952
		atomic_add(input->free_blocks_count,
			   &sbi->s_flex_groups[flex_group].free_blocks);
		atomic_add(EXT4_INODES_PER_GROUP(sb),
			   &sbi->s_flex_groups[flex_group].free_inodes);
953 954
	}

T
Theodore Ts'o 已提交
955
	ext4_handle_dirty_super(handle, sb);
956 957

exit_journal:
958
	if ((err2 = ext4_journal_stop(handle)) && !err)
959
		err = err2;
960
	if (!err && primary) {
961
		update_backups(sb, sbi->s_sbh->b_blocknr, (char *)es,
962
			       sizeof(struct ext4_super_block));
963 964 965 966 967 968
		update_backups(sb, primary->b_blocknr, primary->b_data,
			       primary->b_size);
	}
exit_put:
	iput(inode);
	return err;
969
} /* ext4_group_add */
970

971 972
/*
 * Extend the filesystem to the new number of blocks specified.  This entry
973 974 975 976
 * 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).
 *
977
 * If we _really_ wanted, we could use default values to call ext4_group_add()
978 979 980
 * allow the "remount" trick to work for arbitrary resizing, assuming enough
 * GDT blocks are reserved to grow to the desired size.
 */
981 982
int ext4_group_extend(struct super_block *sb, struct ext4_super_block *es,
		      ext4_fsblk_t n_blocks_count)
983
{
984 985 986
	ext4_fsblk_t o_blocks_count;
	ext4_grpblk_t last;
	ext4_grpblk_t add;
987
	struct buffer_head *bh;
988
	handle_t *handle;
989
	int err, err2;
990
	ext4_group_t group;
991

L
Laurent Vivier 已提交
992
	o_blocks_count = ext4_blocks_count(es);
993 994

	if (test_opt(sb, DEBUG))
995
		printk(KERN_DEBUG "EXT4-fs: extending last group from %llu to %llu blocks\n",
996 997 998 999 1000 1001
		       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)) {
1002
		printk(KERN_ERR "EXT4-fs: filesystem on %s:"
1003
			" too large to resize to %llu blocks safely\n",
1004 1005
			sb->s_id, n_blocks_count);
		if (sizeof(sector_t) < 8)
1006
			ext4_warning(sb, "CONFIG_LBDAF not enabled");
1007 1008 1009 1010
		return -EINVAL;
	}

	if (n_blocks_count < o_blocks_count) {
1011
		ext4_warning(sb, "can't shrink FS - resize aborted");
1012
		return -EINVAL;
1013 1014 1015
	}

	/* Handle the remaining blocks in the last group only. */
1016
	ext4_get_group_no_and_offset(sb, o_blocks_count, &group, &last);
1017 1018

	if (last == 0) {
1019
		ext4_warning(sb, "need to use ext2online to resize further");
1020 1021 1022
		return -EPERM;
	}

1023
	add = EXT4_BLOCKS_PER_GROUP(sb) - last;
1024 1025

	if (o_blocks_count + add < o_blocks_count) {
1026
		ext4_warning(sb, "blocks_count overflow");
1027 1028 1029 1030 1031 1032 1033
		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)
1034
		ext4_warning(sb, "will only finish group (%llu blocks, %u new)",
1035 1036 1037
			     o_blocks_count + add, add);

	/* See if the device is actually as big as what was requested */
1038
	bh = sb_bread(sb, o_blocks_count + add - 1);
1039
	if (!bh) {
1040
		ext4_warning(sb, "can't read last block, resize aborted");
1041 1042 1043 1044 1045
		return -ENOSPC;
	}
	brelse(bh);

	/* We will update the superblock, one block bitmap, and
1046
	 * one group descriptor via ext4_free_blocks().
1047
	 */
1048
	handle = ext4_journal_start_sb(sb, 3);
1049 1050
	if (IS_ERR(handle)) {
		err = PTR_ERR(handle);
1051
		ext4_warning(sb, "error %d on journal start", err);
1052 1053 1054
		goto exit_put;
	}

1055 1056
	if ((err = ext4_journal_get_write_access(handle,
						 EXT4_SB(sb)->s_sbh))) {
1057
		ext4_warning(sb, "error %d on journal write access", err);
1058
		ext4_journal_stop(handle);
1059 1060
		goto exit_put;
	}
L
Laurent Vivier 已提交
1061
	ext4_blocks_count_set(es, o_blocks_count + add);
E
Eric Sandeen 已提交
1062
	ext4_debug("freeing blocks %llu through %llu\n", o_blocks_count,
1063
		   o_blocks_count + add);
1064
	/* We add the blocks to the bitmap and set the group need init bit */
1065
	err = ext4_group_add_blocks(handle, sb, o_blocks_count, add);
T
Theodore Ts'o 已提交
1066
	ext4_handle_dirty_super(handle, sb);
1067
	ext4_debug("freed blocks %llu through %llu\n", o_blocks_count,
1068
		   o_blocks_count + add);
1069 1070 1071 1072 1073
	err2 = ext4_journal_stop(handle);
	if (!err && err2)
		err = err2;

	if (err)
1074
		goto exit_put;
1075

1076
	if (test_opt(sb, DEBUG))
L
Laurent Vivier 已提交
1077 1078
		printk(KERN_DEBUG "EXT4-fs: extended group to %llu blocks\n",
		       ext4_blocks_count(es));
1079 1080
	update_backups(sb, EXT4_SB(sb)->s_sbh->b_blocknr, (char *)es,
		       sizeof(struct ext4_super_block));
1081 1082
exit_put:
	return err;
1083
} /* ext4_group_extend */