migrate.c 15.5 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/*
 * Copyright IBM Corporation, 2007
 * Author Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms of version 2.1 of the GNU Lesser General Public License
 * as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it would be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
 *
 */

#include <linux/module.h>
16
#include <linux/slab.h>
17 18
#include "ext4_jbd2.h"
#include "ext4_extents.h"
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

/*
 * The contiguous blocks details which can be
 * represented by a single extent
 */
struct list_blocks_struct {
	ext4_lblk_t first_block, last_block;
	ext4_fsblk_t first_pblock, last_pblock;
};

static int finish_range(handle_t *handle, struct inode *inode,
				struct list_blocks_struct *lb)

{
	int retval = 0, needed;
	struct ext4_extent newext;
	struct ext4_ext_path *path;
	if (lb->first_pblock == 0)
		return 0;

	/* Add the extent to temp inode*/
	newext.ee_block = cpu_to_le32(lb->first_block);
	newext.ee_len   = cpu_to_le16(lb->last_block - lb->first_block + 1);
	ext4_ext_store_pblock(&newext, lb->first_pblock);
	path = ext4_ext_find_extent(inode, lb->first_block, NULL);

	if (IS_ERR(path)) {
		retval = PTR_ERR(path);
47
		path = NULL;
48 49 50 51 52 53 54 55 56
		goto err_out;
	}

	/*
	 * Calculate the credit needed to inserting this extent
	 * Since we are doing this in loop we may accumalate extra
	 * credit. But below we try to not accumalate too much
	 * of them by restarting the journal.
	 */
57 58
	needed = ext4_ext_calc_credits_for_single_extent(inode,
		    lb->last_block - lb->first_block + 1, path);
59 60 61 62

	/*
	 * Make sure the credit we accumalated is not really high
	 */
63 64
	if (needed && ext4_handle_has_enough_credits(handle,
						EXT4_RESERVE_TRANS_BLOCKS)) {
65 66 67
		retval = ext4_journal_restart(handle, needed);
		if (retval)
			goto err_out;
68
	} else if (needed) {
69
		retval = ext4_journal_extend(handle, needed);
70
		if (retval) {
71 72 73 74 75 76 77 78
			/*
			 * IF not able to extend the journal restart the journal
			 */
			retval = ext4_journal_restart(handle, needed);
			if (retval)
				goto err_out;
		}
	}
79
	retval = ext4_ext_insert_extent(handle, inode, path, &newext, 0);
80
err_out:
81 82 83 84
	if (path) {
		ext4_ext_drop_refs(path);
		kfree(path);
	}
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
	lb->first_pblock = 0;
	return retval;
}

static int update_extent_range(handle_t *handle, struct inode *inode,
				ext4_fsblk_t pblock, ext4_lblk_t blk_num,
				struct list_blocks_struct *lb)
{
	int retval;
	/*
	 * See if we can add on to the existing range (if it exists)
	 */
	if (lb->first_pblock &&
		(lb->last_pblock+1 == pblock) &&
		(lb->last_block+1 == blk_num)) {
		lb->last_pblock = pblock;
		lb->last_block = blk_num;
		return 0;
	}
	/*
	 * Start a new range.
	 */
	retval = finish_range(handle, inode, lb);
	lb->first_pblock = lb->last_pblock = pblock;
	lb->first_block = lb->last_block = blk_num;

	return retval;
}

static int update_ind_extent_range(handle_t *handle, struct inode *inode,
				   ext4_fsblk_t pblock, ext4_lblk_t *blk_nump,
				   struct list_blocks_struct *lb)
{
	struct buffer_head *bh;
	__le32 *i_data;
	int i, retval = 0;
	ext4_lblk_t blk_count = *blk_nump;
	unsigned long max_entries = inode->i_sb->s_blocksize >> 2;

	if (!pblock) {
		/* Only update the file block number */
		*blk_nump += max_entries;
		return 0;
	}

	bh = sb_bread(inode->i_sb, pblock);
	if (!bh)
		return -EIO;

	i_data = (__le32 *)bh->b_data;
	for (i = 0; i < max_entries; i++, blk_count++) {
		if (i_data[i]) {
			retval = update_extent_range(handle, inode,
						le32_to_cpu(i_data[i]),
						blk_count, lb);
			if (retval)
				break;
		}
	}

	/* Update the file block number */
	*blk_nump = blk_count;
	put_bh(bh);
	return retval;

}

static int update_dind_extent_range(handle_t *handle, struct inode *inode,
				    ext4_fsblk_t pblock, ext4_lblk_t *blk_nump,
				    struct list_blocks_struct *lb)
{
	struct buffer_head *bh;
	__le32 *i_data;
	int i, retval = 0;
	ext4_lblk_t blk_count = *blk_nump;
	unsigned long max_entries = inode->i_sb->s_blocksize >> 2;

	if (!pblock) {
		/* Only update the file block number */
		*blk_nump += max_entries * max_entries;
		return 0;
	}
	bh = sb_bread(inode->i_sb, pblock);
	if (!bh)
		return -EIO;

	i_data = (__le32 *)bh->b_data;
	for (i = 0; i < max_entries; i++) {
		if (i_data[i]) {
			retval = update_ind_extent_range(handle, inode,
						le32_to_cpu(i_data[i]),
						&blk_count, lb);
			if (retval)
				break;
		} else {
			/* Only update the file block number */
			blk_count += max_entries;
		}
	}

	/* Update the file block number */
	*blk_nump = blk_count;
	put_bh(bh);
	return retval;

}

static int update_tind_extent_range(handle_t *handle, struct inode *inode,
				     ext4_fsblk_t pblock, ext4_lblk_t *blk_nump,
				     struct list_blocks_struct *lb)
{
	struct buffer_head *bh;
	__le32 *i_data;
	int i, retval = 0;
	ext4_lblk_t blk_count = *blk_nump;
	unsigned long max_entries = inode->i_sb->s_blocksize >> 2;

	if (!pblock) {
		/* Only update the file block number */
		*blk_nump += max_entries * max_entries * max_entries;
		return 0;
	}
	bh = sb_bread(inode->i_sb, pblock);
	if (!bh)
		return -EIO;

	i_data = (__le32 *)bh->b_data;
	for (i = 0; i < max_entries; i++) {
		if (i_data[i]) {
			retval = update_dind_extent_range(handle, inode,
						le32_to_cpu(i_data[i]),
						&blk_count, lb);
			if (retval)
				break;
		} else
			/* Only update the file block number */
			blk_count += max_entries * max_entries;
	}
	/* Update the file block number */
	*blk_nump = blk_count;
	put_bh(bh);
	return retval;

}

230 231 232 233
static int extend_credit_for_blkdel(handle_t *handle, struct inode *inode)
{
	int retval = 0, needed;

234
	if (ext4_handle_has_enough_credits(handle, EXT4_RESERVE_TRANS_BLOCKS+1))
235 236 237 238 239 240 241
		return 0;
	/*
	 * We are freeing a blocks. During this we touch
	 * superblock, group descriptor and block bitmap.
	 * So allocate a credit of 3. We may update
	 * quota (user and group).
	 */
Dmitry Monakhov's avatar
Dmitry Monakhov committed
242
	needed = 3 + EXT4_MAXQUOTAS_TRANS_BLOCKS(inode->i_sb);
243 244 245 246 247 248 249

	if (ext4_journal_extend(handle, needed) != 0)
		retval = ext4_journal_restart(handle, needed);

	return retval;
}

250 251 252 253 254 255 256 257 258 259 260 261 262 263
static int free_dind_blocks(handle_t *handle,
				struct inode *inode, __le32 i_data)
{
	int i;
	__le32 *tmp_idata;
	struct buffer_head *bh;
	unsigned long max_entries = inode->i_sb->s_blocksize >> 2;

	bh = sb_bread(inode->i_sb, le32_to_cpu(i_data));
	if (!bh)
		return -EIO;

	tmp_idata = (__le32 *)bh->b_data;
	for (i = 0; i < max_entries; i++) {
264 265
		if (tmp_idata[i]) {
			extend_credit_for_blkdel(handle, inode);
266 267 268 269
			ext4_free_blocks(handle, inode, 0,
					 le32_to_cpu(tmp_idata[i]), 1,
					 EXT4_FREE_BLOCKS_METADATA |
					 EXT4_FREE_BLOCKS_FORGET);
270
		}
271 272
	}
	put_bh(bh);
273
	extend_credit_for_blkdel(handle, inode);
274 275 276
	ext4_free_blocks(handle, inode, 0, le32_to_cpu(i_data), 1,
			 EXT4_FREE_BLOCKS_METADATA |
			 EXT4_FREE_BLOCKS_FORGET);
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
	return 0;
}

static int free_tind_blocks(handle_t *handle,
				struct inode *inode, __le32 i_data)
{
	int i, retval = 0;
	__le32 *tmp_idata;
	struct buffer_head *bh;
	unsigned long max_entries = inode->i_sb->s_blocksize >> 2;

	bh = sb_bread(inode->i_sb, le32_to_cpu(i_data));
	if (!bh)
		return -EIO;

	tmp_idata = (__le32 *)bh->b_data;
	for (i = 0; i < max_entries; i++) {
		if (tmp_idata[i]) {
			retval = free_dind_blocks(handle,
					inode, tmp_idata[i]);
			if (retval) {
				put_bh(bh);
				return retval;
			}
		}
	}
	put_bh(bh);
304
	extend_credit_for_blkdel(handle, inode);
305 306 307
	ext4_free_blocks(handle, inode, 0, le32_to_cpu(i_data), 1,
			 EXT4_FREE_BLOCKS_METADATA |
			 EXT4_FREE_BLOCKS_FORGET);
308 309 310
	return 0;
}

311
static int free_ind_block(handle_t *handle, struct inode *inode, __le32 *i_data)
312 313 314
{
	int retval;

315 316 317
	/* ei->i_data[EXT4_IND_BLOCK] */
	if (i_data[0]) {
		extend_credit_for_blkdel(handle, inode);
318 319 320 321
		ext4_free_blocks(handle, inode, 0,
				le32_to_cpu(i_data[0]), 1,
				 EXT4_FREE_BLOCKS_METADATA |
				 EXT4_FREE_BLOCKS_FORGET);
322
	}
323

324 325 326
	/* ei->i_data[EXT4_DIND_BLOCK] */
	if (i_data[1]) {
		retval = free_dind_blocks(handle, inode, i_data[1]);
327 328 329 330
		if (retval)
			return retval;
	}

331 332 333
	/* ei->i_data[EXT4_TIND_BLOCK] */
	if (i_data[2]) {
		retval = free_tind_blocks(handle, inode, i_data[2]);
334 335 336 337 338 339 340
		if (retval)
			return retval;
	}
	return 0;
}

static int ext4_ext_swap_inode_data(handle_t *handle, struct inode *inode,
341
						struct inode *tmp_inode)
342
{
343 344
	int retval;
	__le32	i_data[3];
345 346 347 348 349 350 351 352
	struct ext4_inode_info *ei = EXT4_I(inode);
	struct ext4_inode_info *tmp_ei = EXT4_I(tmp_inode);

	/*
	 * One credit accounted for writing the
	 * i_data field of the original inode
	 */
	retval = ext4_journal_extend(handle, 1);
353
	if (retval) {
354 355 356 357 358
		retval = ext4_journal_restart(handle, 1);
		if (retval)
			goto err_out;
	}

359 360 361 362 363
	i_data[0] = ei->i_data[EXT4_IND_BLOCK];
	i_data[1] = ei->i_data[EXT4_DIND_BLOCK];
	i_data[2] = ei->i_data[EXT4_TIND_BLOCK];

	down_write(&EXT4_I(inode)->i_data_sem);
364
	/*
365
	 * if EXT4_STATE_EXT_MIGRATE is cleared a block allocation
366 367 368
	 * happened after we started the migrate. We need to
	 * fail the migrate
	 */
369
	if (!ext4_test_inode_state(inode, EXT4_STATE_EXT_MIGRATE)) {
370 371 372 373
		retval = -EAGAIN;
		up_write(&EXT4_I(inode)->i_data_sem);
		goto err_out;
	} else
374
		ext4_clear_inode_state(inode, EXT4_STATE_EXT_MIGRATE);
375 376 377 378
	/*
	 * We have the extent map build with the tmp inode.
	 * Now copy the i_data across
	 */
379
	ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS);
380 381 382 383 384 385 386 387 388 389 390 391 392 393
	memcpy(ei->i_data, tmp_ei->i_data, sizeof(ei->i_data));

	/*
	 * Update i_blocks with the new blocks that got
	 * allocated while adding extents for extent index
	 * blocks.
	 *
	 * While converting to extents we need not
	 * update the orignal inode i_blocks for extent blocks
	 * via quota APIs. The quota update happened via tmp_inode already.
	 */
	spin_lock(&inode->i_lock);
	inode->i_blocks += tmp_inode->i_blocks;
	spin_unlock(&inode->i_lock);
394
	up_write(&EXT4_I(inode)->i_data_sem);
395

396 397 398 399 400
	/*
	 * We mark the inode dirty after, because we decrement the
	 * i_blocks when freeing the indirect meta-data blocks
	 */
	retval = free_ind_block(handle, inode, i_data);
401
	ext4_mark_inode_dirty(handle, inode);
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
err_out:
	return retval;
}

static int free_ext_idx(handle_t *handle, struct inode *inode,
					struct ext4_extent_idx *ix)
{
	int i, retval = 0;
	ext4_fsblk_t block;
	struct buffer_head *bh;
	struct ext4_extent_header *eh;

	block = idx_pblock(ix);
	bh = sb_bread(inode->i_sb, block);
	if (!bh)
		return -EIO;

	eh = (struct ext4_extent_header *)bh->b_data;
	if (eh->eh_depth != 0) {
		ix = EXT_FIRST_INDEX(eh);
		for (i = 0; i < le16_to_cpu(eh->eh_entries); i++, ix++) {
			retval = free_ext_idx(handle, inode, ix);
			if (retval)
				break;
		}
	}
	put_bh(bh);
430
	extend_credit_for_blkdel(handle, inode);
431 432
	ext4_free_blocks(handle, inode, 0, block, 1,
			 EXT4_FREE_BLOCKS_METADATA | EXT4_FREE_BLOCKS_FORGET);
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
	return retval;
}

/*
 * Free the extent meta data blocks only
 */
static int free_ext_block(handle_t *handle, struct inode *inode)
{
	int i, retval = 0;
	struct ext4_inode_info *ei = EXT4_I(inode);
	struct ext4_extent_header *eh = (struct ext4_extent_header *)ei->i_data;
	struct ext4_extent_idx *ix;
	if (eh->eh_depth == 0)
		/*
		 * No extra blocks allocated for extent meta data
		 */
		return 0;
	ix = EXT_FIRST_INDEX(eh);
	for (i = 0; i < le16_to_cpu(eh->eh_entries); i++, ix++) {
		retval = free_ext_idx(handle, inode, ix);
		if (retval)
			return retval;
	}
	return retval;

}

460
int ext4_ext_migrate(struct inode *inode)
461 462 463 464 465 466 467 468 469
{
	handle_t *handle;
	int retval = 0, i;
	__le32 *i_data;
	ext4_lblk_t blk_count = 0;
	struct ext4_inode_info *ei;
	struct inode *tmp_inode = NULL;
	struct list_blocks_struct lb;
	unsigned long max_entries;
470
	__u32 goal;
471

472 473 474 475 476 477
	/*
	 * If the filesystem does not support extents, or the inode
	 * already is extent-based, error out.
	 */
	if (!EXT4_HAS_INCOMPAT_FEATURE(inode->i_sb,
				       EXT4_FEATURE_INCOMPAT_EXTENTS) ||
478
	    (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)))
479 480
		return -EINVAL;

481 482 483 484 485 486
	if (S_ISLNK(inode->i_mode) && inode->i_blocks == 0)
		/*
		 * don't migrate fast symlink
		 */
		return retval;

487 488 489
	handle = ext4_journal_start(inode,
					EXT4_DATA_TRANS_BLOCKS(inode->i_sb) +
					EXT4_INDEX_EXTRA_TRANS_BLOCKS + 3 +
Dmitry Monakhov's avatar
Dmitry Monakhov committed
490
					EXT4_MAXQUOTAS_INIT_BLOCKS(inode->i_sb)
491 492 493
					+ 1);
	if (IS_ERR(handle)) {
		retval = PTR_ERR(handle);
494
		return retval;
495
	}
496 497
	goal = (((inode->i_ino - 1) / EXT4_INODES_PER_GROUP(inode->i_sb)) *
		EXT4_INODES_PER_GROUP(inode->i_sb)) + 1;
498
	tmp_inode = ext4_new_inode(handle, inode->i_sb->s_root->d_inode,
499
				   S_IFREG, 0, goal);
500 501 502
	if (IS_ERR(tmp_inode)) {
		retval = -ENOMEM;
		ext4_journal_stop(handle);
503
		return retval;
504 505 506
	}
	i_size_write(tmp_inode, i_size_read(inode));
	/*
507 508
	 * Set the i_nlink to zero so it will be deleted later
	 * when we drop inode reference.
509
	 */
510
	tmp_inode->i_nlink = 0;
511 512 513 514 515 516 517 518 519 520 521 522 523

	ext4_ext_tree_init(handle, tmp_inode);
	ext4_orphan_add(handle, tmp_inode);
	ext4_journal_stop(handle);

	/*
	 * start with one credit accounted for
	 * superblock modification.
	 *
	 * For the tmp_inode we already have commited the
	 * trascation that created the inode. Later as and
	 * when we add extents we extent the journal
	 */
524
	/*
525 526 527 528 529 530
	 * Even though we take i_mutex we can still cause block
	 * allocation via mmap write to holes. If we have allocated
	 * new blocks we fail migrate.  New block allocation will
	 * clear EXT4_STATE_EXT_MIGRATE flag.  The flag is updated
	 * with i_data_sem held to prevent racing with block
	 * allocation.
531 532
	 */
	down_read((&EXT4_I(inode)->i_data_sem));
533
	ext4_set_inode_state(inode, EXT4_STATE_EXT_MIGRATE);
534 535
	up_read((&EXT4_I(inode)->i_data_sem));

536
	handle = ext4_journal_start(inode, 1);
537 538 539 540 541 542 543 544 545 546
	if (IS_ERR(handle)) {
		/*
		 * It is impossible to update on-disk structures without
		 * a handle, so just rollback in-core changes and live other
		 * work to orphan_list_cleanup()
		 */
		ext4_orphan_del(NULL, tmp_inode);
		retval = PTR_ERR(handle);
		goto out;
	}
547 548 549 550 551 552 553

	ei = EXT4_I(inode);
	i_data = ei->i_data;
	memset(&lb, 0, sizeof(lb));

	/* 32 bit block address 4 bytes */
	max_entries = inode->i_sb->s_blocksize >> 2;
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
	for (i = 0; i < EXT4_NDIR_BLOCKS; i++, blk_count++) {
		if (i_data[i]) {
			retval = update_extent_range(handle, tmp_inode,
						le32_to_cpu(i_data[i]),
						blk_count, &lb);
			if (retval)
				goto err_out;
		}
	}
	if (i_data[EXT4_IND_BLOCK]) {
		retval = update_ind_extent_range(handle, tmp_inode,
					le32_to_cpu(i_data[EXT4_IND_BLOCK]),
					&blk_count, &lb);
			if (retval)
				goto err_out;
	} else
		blk_count +=  max_entries;
	if (i_data[EXT4_DIND_BLOCK]) {
		retval = update_dind_extent_range(handle, tmp_inode,
					le32_to_cpu(i_data[EXT4_DIND_BLOCK]),
					&blk_count, &lb);
			if (retval)
				goto err_out;
	} else
		blk_count += max_entries * max_entries;
	if (i_data[EXT4_TIND_BLOCK]) {
		retval = update_tind_extent_range(handle, tmp_inode,
					le32_to_cpu(i_data[EXT4_TIND_BLOCK]),
					&blk_count, &lb);
			if (retval)
				goto err_out;
	}
	/*
	 * Build the last extent
	 */
	retval = finish_range(handle, tmp_inode, &lb);
err_out:
	if (retval)
		/*
		 * Failure case delete the extent information with the
		 * tmp_inode
		 */
		free_ext_block(handle, tmp_inode);
597 598 599 600 601 602 603 604 605
	else {
		retval = ext4_ext_swap_inode_data(handle, inode, tmp_inode);
		if (retval)
			/*
			 * if we fail to swap inode data free the extent
			 * details of the tmp inode
			 */
			free_ext_block(handle, tmp_inode);
	}
606 607 608 609

	/* We mark the tmp_inode dirty via ext4_ext_tree_init. */
	if (ext4_journal_extend(handle, 1) != 0)
		ext4_journal_restart(handle, 1);
610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628

	/*
	 * Mark the tmp_inode as of size zero
	 */
	i_size_write(tmp_inode, 0);

	/*
	 * set the  i_blocks count to zero
	 * so that the ext4_delete_inode does the
	 * right job
	 *
	 * We don't need to take the i_lock because
	 * the inode is not visible to user space.
	 */
	tmp_inode->i_blocks = 0;

	/* Reset the extent details */
	ext4_ext_tree_init(handle, tmp_inode);
	ext4_journal_stop(handle);
629
out:
630
	unlock_new_inode(tmp_inode);
631
	iput(tmp_inode);
632 633 634

	return retval;
}