inode.c 93.2 KB
Newer Older
C
Chris Mason 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

19
#include <linux/kernel.h>
20
#include <linux/bio.h>
C
Chris Mason 已提交
21
#include <linux/buffer_head.h>
S
Sage Weil 已提交
22
#include <linux/file.h>
C
Chris Mason 已提交
23 24 25 26 27 28 29 30 31 32 33 34 35
#include <linux/fs.h>
#include <linux/pagemap.h>
#include <linux/highmem.h>
#include <linux/time.h>
#include <linux/init.h>
#include <linux/string.h>
#include <linux/smp_lock.h>
#include <linux/backing-dev.h>
#include <linux/mpage.h>
#include <linux/swap.h>
#include <linux/writeback.h>
#include <linux/statfs.h>
#include <linux/compat.h>
36
#include <linux/bit_spinlock.h>
37
#include <linux/version.h>
J
Josef Bacik 已提交
38
#include <linux/xattr.h>
J
Josef Bacik 已提交
39
#include <linux/posix_acl.h>
C
Chris Mason 已提交
40 41 42 43 44 45
#include "ctree.h"
#include "disk-io.h"
#include "transaction.h"
#include "btrfs_inode.h"
#include "ioctl.h"
#include "print-tree.h"
46
#include "volumes.h"
47
#include "ordered-data.h"
C
Chris Mason 已提交
48 49 50 51 52 53 54 55 56

struct btrfs_iget_args {
	u64 ino;
	struct btrfs_root *root;
};

static struct inode_operations btrfs_dir_inode_operations;
static struct inode_operations btrfs_symlink_inode_operations;
static struct inode_operations btrfs_dir_ro_inode_operations;
J
Josef Bacik 已提交
57
static struct inode_operations btrfs_special_inode_operations;
C
Chris Mason 已提交
58 59 60 61
static struct inode_operations btrfs_file_inode_operations;
static struct address_space_operations btrfs_aops;
static struct address_space_operations btrfs_symlink_aops;
static struct file_operations btrfs_dir_file_operations;
62
static struct extent_io_ops btrfs_extent_io_ops;
C
Chris Mason 已提交
63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80

static struct kmem_cache *btrfs_inode_cachep;
struct kmem_cache *btrfs_trans_handle_cachep;
struct kmem_cache *btrfs_transaction_cachep;
struct kmem_cache *btrfs_bit_radix_cachep;
struct kmem_cache *btrfs_path_cachep;

#define S_SHIFT 12
static unsigned char btrfs_type_by_mode[S_IFMT >> S_SHIFT] = {
	[S_IFREG >> S_SHIFT]	= BTRFS_FT_REG_FILE,
	[S_IFDIR >> S_SHIFT]	= BTRFS_FT_DIR,
	[S_IFCHR >> S_SHIFT]	= BTRFS_FT_CHRDEV,
	[S_IFBLK >> S_SHIFT]	= BTRFS_FT_BLKDEV,
	[S_IFIFO >> S_SHIFT]	= BTRFS_FT_FIFO,
	[S_IFSOCK >> S_SHIFT]	= BTRFS_FT_SOCK,
	[S_IFLNK >> S_SHIFT]	= BTRFS_FT_SYMLINK,
};

81 82
static void btrfs_truncate(struct inode *inode);

83 84 85
int btrfs_check_free_space(struct btrfs_root *root, u64 num_required,
			   int for_del)
{
86 87
	u64 total;
	u64 used;
88
	u64 thresh;
89
	unsigned long flags;
90 91
	int ret = 0;

92 93 94
	spin_lock_irqsave(&root->fs_info->delalloc_lock, flags);
	total = btrfs_super_total_bytes(&root->fs_info->super_copy);
	used = btrfs_super_bytes_used(&root->fs_info->super_copy);
95
	if (for_del)
96
		thresh = total * 90;
97
	else
98 99 100
		thresh = total * 85;

	do_div(thresh, 100);
101 102 103

	if (used + root->fs_info->delalloc_bytes + num_required > thresh)
		ret = -ENOSPC;
104
	spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags);
105 106 107
	return ret;
}

108
static int cow_file_range(struct inode *inode, u64 start, u64 end)
109 110 111 112
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_trans_handle *trans;
	u64 alloc_hint = 0;
113
	u64 num_bytes;
114
	u64 cur_alloc_size;
115
	u64 blocksize = root->sectorsize;
116
	u64 orig_num_bytes;
117
	struct btrfs_key ins;
118 119 120
	struct extent_map *em;
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
	int ret = 0;
121

122
	trans = btrfs_join_transaction(root, 1);
123
	BUG_ON(!trans);
124 125
	btrfs_set_trans_block_group(trans, inode);

126
	num_bytes = (end - start + blocksize) & ~(blocksize - 1);
127
	num_bytes = max(blocksize,  num_bytes);
128
	orig_num_bytes = num_bytes;
129

130 131 132
	if (alloc_hint == EXTENT_MAP_INLINE)
		goto out;

133
	BUG_ON(num_bytes > btrfs_super_total_bytes(&root->fs_info->super_copy));
134
	mutex_lock(&BTRFS_I(inode)->extent_mutex);
135
	btrfs_drop_extent_cache(inode, start, start + num_bytes - 1);
136
	mutex_unlock(&BTRFS_I(inode)->extent_mutex);
137

138 139
	while(num_bytes > 0) {
		cur_alloc_size = min(num_bytes, root->fs_info->max_extent);
140 141 142
		ret = btrfs_reserve_extent(trans, root, cur_alloc_size,
					   root->sectorsize, 0, 0,
					   (u64)-1, &ins, 1);
143 144 145 146
		if (ret) {
			WARN_ON(1);
			goto out;
		}
147 148 149 150 151
		em = alloc_extent_map(GFP_NOFS);
		em->start = start;
		em->len = ins.offset;
		em->block_start = ins.objectid;
		em->bdev = root->fs_info->fs_devices->latest_bdev;
152
		mutex_lock(&BTRFS_I(inode)->extent_mutex);
153
		set_bit(EXTENT_FLAG_PINNED, &em->flags);
154 155 156 157 158 159 160 161 162 163 164
		while(1) {
			spin_lock(&em_tree->lock);
			ret = add_extent_mapping(em_tree, em);
			spin_unlock(&em_tree->lock);
			if (ret != -EEXIST) {
				free_extent_map(em);
				break;
			}
			btrfs_drop_extent_cache(inode, start,
						start + ins.offset - 1);
		}
165
		mutex_unlock(&BTRFS_I(inode)->extent_mutex);
166

167
		cur_alloc_size = ins.offset;
168 169 170
		ret = btrfs_add_ordered_extent(inode, start, ins.objectid,
					       ins.offset);
		BUG_ON(ret);
171 172 173 174 175
		if (num_bytes < cur_alloc_size) {
			printk("num_bytes %Lu cur_alloc %Lu\n", num_bytes,
			       cur_alloc_size);
			break;
		}
176 177 178
		num_bytes -= cur_alloc_size;
		alloc_hint = ins.objectid + ins.offset;
		start += cur_alloc_size;
179 180 181
	}
out:
	btrfs_end_transaction(trans, root);
182 183 184 185 186 187 188 189 190
	return ret;
}

static int run_delalloc_nocow(struct inode *inode, u64 start, u64 end)
{
	u64 extent_start;
	u64 extent_end;
	u64 bytenr;
	u64 cow_end;
191
	u64 loops = 0;
192
	u64 total_fs_bytes;
193
	struct btrfs_root *root = BTRFS_I(inode)->root;
194
	struct btrfs_block_group_cache *block_group;
195 196 197 198 199 200 201 202
	struct extent_buffer *leaf;
	int found_type;
	struct btrfs_path *path;
	struct btrfs_file_extent_item *item;
	int ret;
	int err;
	struct btrfs_key found_key;

203
	total_fs_bytes = btrfs_super_total_bytes(&root->fs_info->super_copy);
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
	path = btrfs_alloc_path();
	BUG_ON(!path);
again:
	ret = btrfs_lookup_file_extent(NULL, root, path,
				       inode->i_ino, start, 0);
	if (ret < 0) {
		btrfs_free_path(path);
		return ret;
	}

	cow_end = end;
	if (ret != 0) {
		if (path->slots[0] == 0)
			goto not_found;
		path->slots[0]--;
	}

	leaf = path->nodes[0];
	item = btrfs_item_ptr(leaf, path->slots[0],
			      struct btrfs_file_extent_item);

	/* are we inside the extent that was found? */
	btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
	found_type = btrfs_key_type(&found_key);
	if (found_key.objectid != inode->i_ino ||
229
	    found_type != BTRFS_EXTENT_DATA_KEY)
230 231 232 233 234
		goto not_found;

	found_type = btrfs_file_extent_type(leaf, item);
	extent_start = found_key.offset;
	if (found_type == BTRFS_FILE_EXTENT_REG) {
235 236 237 238
		u64 extent_num_bytes;

		extent_num_bytes = btrfs_file_extent_num_bytes(leaf, item);
		extent_end = extent_start + extent_num_bytes;
239 240
		err = 0;

241 242 243
		if (loops && start != extent_start)
			goto not_found;

244 245 246 247 248 249 250 251
		if (start < extent_start || start >= extent_end)
			goto not_found;

		cow_end = min(end, extent_end - 1);
		bytenr = btrfs_file_extent_disk_bytenr(leaf, item);
		if (bytenr == 0)
			goto not_found;

252
		if (btrfs_cross_ref_exists(root, &found_key, bytenr))
253
			goto not_found;
254 255 256 257
		/*
		 * we may be called by the resizer, make sure we're inside
		 * the limits of the FS
		 */
258 259 260
		block_group = btrfs_lookup_block_group(root->fs_info,
						       bytenr);
		if (!block_group || block_group->ro)
261 262
			goto not_found;

263
		start = extent_end;
264
	} else {
265 266 267 268 269 270 271 272
		goto not_found;
	}
loop:
	if (start > end) {
		btrfs_free_path(path);
		return 0;
	}
	btrfs_release_path(root, path);
273
	loops++;
274 275 276
	goto again;

not_found:
277
	btrfs_release_path(root, path);
278 279
	cow_file_range(inode, start, end);
	start = end + 1;
280 281 282 283 284 285 286
	goto loop;
}

static int run_delalloc_range(struct inode *inode, u64 start, u64 end)
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int ret;
287

Y
Yan 已提交
288 289
	if (btrfs_test_opt(root, NODATACOW) ||
	    btrfs_test_flag(inode, NODATACOW))
290 291 292
		ret = run_delalloc_nocow(inode, start, end);
	else
		ret = cow_file_range(inode, start, end);
293

294 295 296
	return ret;
}

297
int btrfs_set_bit_hook(struct inode *inode, u64 start, u64 end,
298
		       unsigned long old, unsigned long bits)
299
{
300
	unsigned long flags;
301
	if (!(old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
302
		struct btrfs_root *root = BTRFS_I(inode)->root;
303
		spin_lock_irqsave(&root->fs_info->delalloc_lock, flags);
304
		BTRFS_I(inode)->delalloc_bytes += end - start + 1;
305
		root->fs_info->delalloc_bytes += end - start + 1;
306
		spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags);
307 308 309 310 311
	}
	return 0;
}

int btrfs_clear_bit_hook(struct inode *inode, u64 start, u64 end,
312
			 unsigned long old, unsigned long bits)
313
{
314
	if ((old & EXTENT_DELALLOC) && (bits & EXTENT_DELALLOC)) {
315
		struct btrfs_root *root = BTRFS_I(inode)->root;
316 317 318
		unsigned long flags;

		spin_lock_irqsave(&root->fs_info->delalloc_lock, flags);
319 320 321 322
		if (end - start + 1 > root->fs_info->delalloc_bytes) {
			printk("warning: delalloc account %Lu %Lu\n",
			       end - start + 1, root->fs_info->delalloc_bytes);
			root->fs_info->delalloc_bytes = 0;
323
			BTRFS_I(inode)->delalloc_bytes = 0;
324 325
		} else {
			root->fs_info->delalloc_bytes -= end - start + 1;
326
			BTRFS_I(inode)->delalloc_bytes -= end - start + 1;
327
		}
328
		spin_unlock_irqrestore(&root->fs_info->delalloc_lock, flags);
329 330 331 332
	}
	return 0;
}

333 334 335 336 337 338 339 340 341 342
int btrfs_merge_bio_hook(struct page *page, unsigned long offset,
			 size_t size, struct bio *bio)
{
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
	struct btrfs_mapping_tree *map_tree;
	u64 logical = bio->bi_sector << 9;
	u64 length = 0;
	u64 map_length;
	int ret;

343
	length = bio->bi_size;
344 345
	map_tree = &root->fs_info->mapping_tree;
	map_length = length;
346
	ret = btrfs_map_block(map_tree, READ, logical,
347
			      &map_length, NULL, 0);
348

349 350 351 352 353 354
	if (map_length < length + size) {
		return 1;
	}
	return 0;
}

355
int __btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
356
			  int mirror_num)
357 358 359
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int ret = 0;
360

361
	ret = btrfs_csum_one_bio(root, inode, bio);
362
	BUG_ON(ret);
363

364
	return btrfs_map_bio(root, rw, bio, mirror_num, 1);
365 366 367 368 369 370 371 372
}

int btrfs_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
			  int mirror_num)
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int ret = 0;

373 374
	ret = btrfs_bio_wq_end_io(root->fs_info, bio, 0);
	BUG_ON(ret);
375

376
	if (!(rw & (1 << BIO_RW))) {
377 378 379 380
		if (!btrfs_test_opt(root, NODATASUM) &&
		    !btrfs_test_flag(inode, NODATASUM)) {
			btrfs_lookup_bio_sums(root, inode, bio);
		}
381 382
		goto mapit;
	}
383

384 385 386
	return btrfs_wq_submit_bio(BTRFS_I(inode)->root->fs_info,
				   inode, rw, bio, mirror_num,
				   __btrfs_submit_bio_hook);
387
mapit:
388
	return btrfs_map_bio(root, rw, bio, mirror_num, 0);
389
}
C
Chris Mason 已提交
390

391
static noinline int add_pending_csums(struct btrfs_trans_handle *trans,
392 393 394 395 396 397 398
			     struct inode *inode, u64 file_offset,
			     struct list_head *list)
{
	struct list_head *cur;
	struct btrfs_ordered_sum *sum;

	btrfs_set_trans_block_group(trans, inode);
399
	list_for_each(cur, list) {
400 401 402 403 404 405 406 407 408
		sum = list_entry(cur, struct btrfs_ordered_sum, list);
		mutex_lock(&BTRFS_I(inode)->csum_mutex);
		btrfs_csum_file_blocks(trans, BTRFS_I(inode)->root,
				       inode, sum);
		mutex_unlock(&BTRFS_I(inode)->csum_mutex);
	}
	return 0;
}

409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425
struct btrfs_writepage_fixup {
	struct page *page;
	struct btrfs_work work;
};

/* see btrfs_writepage_start_hook for details on why this is required */
void btrfs_writepage_fixup_worker(struct btrfs_work *work)
{
	struct btrfs_writepage_fixup *fixup;
	struct btrfs_ordered_extent *ordered;
	struct page *page;
	struct inode *inode;
	u64 page_start;
	u64 page_end;

	fixup = container_of(work, struct btrfs_writepage_fixup, work);
	page = fixup->page;
C
Chris Mason 已提交
426
again:
427 428 429 430 431 432 433 434 435 436 437
	lock_page(page);
	if (!page->mapping || !PageDirty(page) || !PageChecked(page)) {
		ClearPageChecked(page);
		goto out_page;
	}

	inode = page->mapping->host;
	page_start = page_offset(page);
	page_end = page_offset(page) + PAGE_CACHE_SIZE - 1;

	lock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end, GFP_NOFS);
C
Chris Mason 已提交
438 439 440 441

	/* already ordered? We're done */
	if (test_range_bit(&BTRFS_I(inode)->io_tree, page_start, page_end,
			     EXTENT_ORDERED, 0)) {
442
		goto out;
C
Chris Mason 已提交
443 444 445 446 447 448 449 450 451 452
	}

	ordered = btrfs_lookup_ordered_extent(inode, page_start);
	if (ordered) {
		unlock_extent(&BTRFS_I(inode)->io_tree, page_start,
			      page_end, GFP_NOFS);
		unlock_page(page);
		btrfs_start_ordered_extent(inode, ordered, 1);
		goto again;
	}
453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492

	set_extent_delalloc(&BTRFS_I(inode)->io_tree, page_start, page_end,
			    GFP_NOFS);
	ClearPageChecked(page);
out:
	unlock_extent(&BTRFS_I(inode)->io_tree, page_start, page_end, GFP_NOFS);
out_page:
	unlock_page(page);
	page_cache_release(page);
}

/*
 * There are a few paths in the higher layers of the kernel that directly
 * set the page dirty bit without asking the filesystem if it is a
 * good idea.  This causes problems because we want to make sure COW
 * properly happens and the data=ordered rules are followed.
 *
 * In our case any range that doesn't have the EXTENT_ORDERED bit set
 * hasn't been properly setup for IO.  We kick off an async process
 * to fix it up.  The async helper will wait for ordered extents, set
 * the delalloc bit and make it safe to write the page.
 */
int btrfs_writepage_start_hook(struct page *page, u64 start, u64 end)
{
	struct inode *inode = page->mapping->host;
	struct btrfs_writepage_fixup *fixup;
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int ret;

	ret = test_range_bit(&BTRFS_I(inode)->io_tree, start, end,
			     EXTENT_ORDERED, 0);
	if (ret)
		return 0;

	if (PageChecked(page))
		return -EAGAIN;

	fixup = kzalloc(sizeof(*fixup), GFP_NOFS);
	if (!fixup)
		return -EAGAIN;
493

494 495 496 497 498 499 500 501
	SetPageChecked(page);
	page_cache_get(page);
	fixup->work.func = btrfs_writepage_fixup_worker;
	fixup->page = page;
	btrfs_queue_worker(&root->fs_info->fixup_workers, &fixup->work);
	return -EAGAIN;
}

502
static int btrfs_finish_ordered_io(struct inode *inode, u64 start, u64 end)
503 504 505 506 507 508 509 510 511 512 513
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_trans_handle *trans;
	struct btrfs_ordered_extent *ordered_extent;
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	u64 alloc_hint = 0;
	struct list_head list;
	struct btrfs_key ins;
	int ret;

	ret = btrfs_dec_test_ordered_pending(inode, start, end - start + 1);
514
	if (!ret)
515 516
		return 0;

517
	trans = btrfs_join_transaction(root, 1);
518 519 520 521 522 523 524 525 526 527 528 529 530

	ordered_extent = btrfs_lookup_ordered_extent(inode, start);
	BUG_ON(!ordered_extent);

	lock_extent(io_tree, ordered_extent->file_offset,
		    ordered_extent->file_offset + ordered_extent->len - 1,
		    GFP_NOFS);

	INIT_LIST_HEAD(&list);

	ins.objectid = ordered_extent->start;
	ins.offset = ordered_extent->len;
	ins.type = BTRFS_EXTENT_ITEM_KEY;
531

532 533 534 535
	ret = btrfs_alloc_reserved_extent(trans, root, root->root_key.objectid,
					  trans->transid, inode->i_ino,
					  ordered_extent->file_offset, &ins);
	BUG_ON(ret);
536 537

	mutex_lock(&BTRFS_I(inode)->extent_mutex);
538

539 540 541 542 543 544 545 546 547 548 549 550
	ret = btrfs_drop_extents(trans, root, inode,
				 ordered_extent->file_offset,
				 ordered_extent->file_offset +
				 ordered_extent->len,
				 ordered_extent->file_offset, &alloc_hint);
	BUG_ON(ret);
	ret = btrfs_insert_file_extent(trans, root, inode->i_ino,
				       ordered_extent->file_offset,
				       ordered_extent->start,
				       ordered_extent->len,
				       ordered_extent->len, 0);
	BUG_ON(ret);
551

552 553 554
	btrfs_drop_extent_cache(inode, ordered_extent->file_offset,
				ordered_extent->file_offset +
				ordered_extent->len - 1);
555 556
	mutex_unlock(&BTRFS_I(inode)->extent_mutex);

557 558 559 560 561 562 563
	inode->i_blocks += ordered_extent->len >> 9;
	unlock_extent(io_tree, ordered_extent->file_offset,
		    ordered_extent->file_offset + ordered_extent->len - 1,
		    GFP_NOFS);
	add_pending_csums(trans, inode, ordered_extent->file_offset,
			  &ordered_extent->list);

564
	btrfs_ordered_update_i_size(inode, ordered_extent);
565
	btrfs_remove_ordered_extent(inode, ordered_extent);
566

567 568 569 570 571 572 573 574 575 576
	/* once for us */
	btrfs_put_ordered_extent(ordered_extent);
	/* once for the tree */
	btrfs_put_ordered_extent(ordered_extent);

	btrfs_update_inode(trans, root, inode);
	btrfs_end_transaction(trans, root);
	return 0;
}

577 578 579 580 581 582
int btrfs_writepage_end_io_hook(struct page *page, u64 start, u64 end,
				struct extent_state *state, int uptodate)
{
	return btrfs_finish_ordered_io(page->mapping->host, start, end);
}

583 584 585 586 587 588 589 590
struct io_failure_record {
	struct page *page;
	u64 start;
	u64 len;
	u64 logical;
	int last_mirror;
};

591 592 593
int btrfs_io_failed_hook(struct bio *failed_bio,
			 struct page *page, u64 start, u64 end,
			 struct extent_state *state)
594 595 596 597 598 599
{
	struct io_failure_record *failrec = NULL;
	u64 private;
	struct extent_map *em;
	struct inode *inode = page->mapping->host;
	struct extent_io_tree *failure_tree = &BTRFS_I(inode)->io_failure_tree;
600
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
601 602 603
	struct bio *bio;
	int num_copies;
	int ret;
604
	int rw;
605 606 607 608 609 610 611 612 613 614 615
	u64 logical;

	ret = get_state_private(failure_tree, start, &private);
	if (ret) {
		failrec = kmalloc(sizeof(*failrec), GFP_NOFS);
		if (!failrec)
			return -ENOMEM;
		failrec->start = start;
		failrec->len = end - start + 1;
		failrec->last_mirror = 0;

616 617 618 619 620 621 622
		spin_lock(&em_tree->lock);
		em = lookup_extent_mapping(em_tree, start, failrec->len);
		if (em->start > start || em->start + em->len < start) {
			free_extent_map(em);
			em = NULL;
		}
		spin_unlock(&em_tree->lock);
623 624 625 626 627 628 629 630 631 632 633

		if (!em || IS_ERR(em)) {
			kfree(failrec);
			return -EIO;
		}
		logical = start - em->start;
		logical = em->block_start + logical;
		failrec->logical = logical;
		free_extent_map(em);
		set_extent_bits(failure_tree, start, end, EXTENT_LOCKED |
				EXTENT_DIRTY, GFP_NOFS);
634 635
		set_state_private(failure_tree, start,
				 (u64)(unsigned long)failrec);
636
	} else {
637
		failrec = (struct io_failure_record *)(unsigned long)private;
638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664
	}
	num_copies = btrfs_num_copies(
			      &BTRFS_I(inode)->root->fs_info->mapping_tree,
			      failrec->logical, failrec->len);
	failrec->last_mirror++;
	if (!state) {
		spin_lock_irq(&BTRFS_I(inode)->io_tree.lock);
		state = find_first_extent_bit_state(&BTRFS_I(inode)->io_tree,
						    failrec->start,
						    EXTENT_LOCKED);
		if (state && state->start != failrec->start)
			state = NULL;
		spin_unlock_irq(&BTRFS_I(inode)->io_tree.lock);
	}
	if (!state || failrec->last_mirror > num_copies) {
		set_state_private(failure_tree, failrec->start, 0);
		clear_extent_bits(failure_tree, failrec->start,
				  failrec->start + failrec->len - 1,
				  EXTENT_LOCKED | EXTENT_DIRTY, GFP_NOFS);
		kfree(failrec);
		return -EIO;
	}
	bio = bio_alloc(GFP_NOFS, 1);
	bio->bi_private = state;
	bio->bi_end_io = failed_bio->bi_end_io;
	bio->bi_sector = failrec->logical >> 9;
	bio->bi_bdev = failed_bio->bi_bdev;
665
	bio->bi_size = 0;
666
	bio_add_page(bio, page, failrec->len, start - page_offset(page));
667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701
	if (failed_bio->bi_rw & (1 << BIO_RW))
		rw = WRITE;
	else
		rw = READ;

	BTRFS_I(inode)->io_tree.ops->submit_bio_hook(inode, rw, bio,
						      failrec->last_mirror);
	return 0;
}

int btrfs_clean_io_failures(struct inode *inode, u64 start)
{
	u64 private;
	u64 private_failure;
	struct io_failure_record *failure;
	int ret;

	private = 0;
	if (count_range_bits(&BTRFS_I(inode)->io_failure_tree, &private,
			     (u64)-1, 1, EXTENT_DIRTY)) {
		ret = get_state_private(&BTRFS_I(inode)->io_failure_tree,
					start, &private_failure);
		if (ret == 0) {
			failure = (struct io_failure_record *)(unsigned long)
				   private_failure;
			set_state_private(&BTRFS_I(inode)->io_failure_tree,
					  failure->start, 0);
			clear_extent_bits(&BTRFS_I(inode)->io_failure_tree,
					  failure->start,
					  failure->start + failure->len - 1,
					  EXTENT_DIRTY | EXTENT_LOCKED,
					  GFP_NOFS);
			kfree(failure);
		}
	}
702 703 704
	return 0;
}

705 706
int btrfs_readpage_end_io_hook(struct page *page, u64 start, u64 end,
			       struct extent_state *state)
707
{
708
	size_t offset = start - ((u64)page->index << PAGE_CACHE_SHIFT);
709
	struct inode *inode = page->mapping->host;
710
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
711
	char *kaddr;
712
	u64 private = ~(u32)0;
713
	int ret;
714 715
	struct btrfs_root *root = BTRFS_I(inode)->root;
	u32 csum = ~(u32)0;
716
	unsigned long flags;
717

Y
Yan 已提交
718 719
	if (btrfs_test_opt(root, NODATASUM) ||
	    btrfs_test_flag(inode, NODATASUM))
720
		return 0;
Y
Yan 已提交
721
	if (state && state->start == start) {
722 723 724 725 726
		private = state->private;
		ret = 0;
	} else {
		ret = get_state_private(io_tree, start, &private);
	}
727
	local_irq_save(flags);
728 729 730 731
	kaddr = kmap_atomic(page, KM_IRQ0);
	if (ret) {
		goto zeroit;
	}
732 733 734
	csum = btrfs_csum_data(root, kaddr + offset, csum,  end - start + 1);
	btrfs_csum_final(csum, (char *)&csum);
	if (csum != private) {
735 736 737
		goto zeroit;
	}
	kunmap_atomic(kaddr, KM_IRQ0);
738
	local_irq_restore(flags);
739 740 741 742

	/* if the io failure tree for this inode is non-empty,
	 * check to see if we've recovered from a failed IO
	 */
743
	btrfs_clean_io_failures(inode, start);
744 745 746
	return 0;

zeroit:
747 748 749
	printk("btrfs csum failed ino %lu off %llu csum %u private %Lu\n",
	       page->mapping->host->i_ino, (unsigned long long)start, csum,
	       private);
750 751
	memset(kaddr + offset, 1, end - start + 1);
	flush_dcache_page(page);
752
	kunmap_atomic(kaddr, KM_IRQ0);
753
	local_irq_restore(flags);
754 755
	if (private == 0)
		return 0;
756
	return -EIO;
757
}
758

759 760 761 762 763 764 765 766 767
/*
 * This creates an orphan entry for the given inode in case something goes
 * wrong in the middle of an unlink/truncate.
 */
int btrfs_orphan_add(struct btrfs_trans_handle *trans, struct inode *inode)
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int ret = 0;

768
	spin_lock(&root->list_lock);
769 770 771

	/* already on the orphan list, we're good */
	if (!list_empty(&BTRFS_I(inode)->i_orphan)) {
772
		spin_unlock(&root->list_lock);
773 774 775 776 777
		return 0;
	}

	list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list);

778
	spin_unlock(&root->list_lock);
779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796

	/*
	 * insert an orphan item to track this unlinked/truncated file
	 */
	ret = btrfs_insert_orphan_item(trans, root, inode->i_ino);

	return ret;
}

/*
 * We have done the truncate/delete so we can go ahead and remove the orphan
 * item for this particular inode.
 */
int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode)
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int ret = 0;

797
	spin_lock(&root->list_lock);
798 799

	if (list_empty(&BTRFS_I(inode)->i_orphan)) {
800
		spin_unlock(&root->list_lock);
801 802 803 804 805
		return 0;
	}

	list_del_init(&BTRFS_I(inode)->i_orphan);
	if (!trans) {
806
		spin_unlock(&root->list_lock);
807 808 809
		return 0;
	}

810
	spin_unlock(&root->list_lock);
811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905

	ret = btrfs_del_orphan_item(trans, root, inode->i_ino);

	return ret;
}

/*
 * this cleans up any orphans that may be left on the list from the last use
 * of this root.
 */
void btrfs_orphan_cleanup(struct btrfs_root *root)
{
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	struct btrfs_item *item;
	struct btrfs_key key, found_key;
	struct btrfs_trans_handle *trans;
	struct inode *inode;
	int ret = 0, nr_unlink = 0, nr_truncate = 0;

	/* don't do orphan cleanup if the fs is readonly. */
	if (root->inode->i_sb->s_flags & MS_RDONLY)
		return;

	path = btrfs_alloc_path();
	if (!path)
		return;
	path->reada = -1;

	key.objectid = BTRFS_ORPHAN_OBJECTID;
	btrfs_set_key_type(&key, BTRFS_ORPHAN_ITEM_KEY);
	key.offset = (u64)-1;

	trans = btrfs_start_transaction(root, 1);
	btrfs_set_trans_block_group(trans, root->inode);

	while (1) {
		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
		if (ret < 0) {
			printk(KERN_ERR "Error searching slot for orphan: %d"
			       "\n", ret);
			break;
		}

		/*
		 * if ret == 0 means we found what we were searching for, which
		 * is weird, but possible, so only screw with path if we didnt
		 * find the key and see if we have stuff that matches
		 */
		if (ret > 0) {
			if (path->slots[0] == 0)
				break;
			path->slots[0]--;
		}

		/* pull out the item */
		leaf = path->nodes[0];
		item = btrfs_item_nr(leaf, path->slots[0]);
		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);

		/* make sure the item matches what we want */
		if (found_key.objectid != BTRFS_ORPHAN_OBJECTID)
			break;
		if (btrfs_key_type(&found_key) != BTRFS_ORPHAN_ITEM_KEY)
			break;

		/* release the path since we're done with it */
		btrfs_release_path(root, path);

		/*
		 * this is where we are basically btrfs_lookup, without the
		 * crossing root thing.  we store the inode number in the
		 * offset of the orphan item.
		 */
		inode = btrfs_iget_locked(root->inode->i_sb,
					  found_key.offset, root);
		if (!inode)
			break;

		if (inode->i_state & I_NEW) {
			BTRFS_I(inode)->root = root;

			/* have to set the location manually */
			BTRFS_I(inode)->location.objectid = inode->i_ino;
			BTRFS_I(inode)->location.type = BTRFS_INODE_ITEM_KEY;
			BTRFS_I(inode)->location.offset = 0;

			btrfs_read_locked_inode(inode);
			unlock_new_inode(inode);
		}

		/*
		 * add this inode to the orphan list so btrfs_orphan_del does
		 * the proper thing when we hit it
		 */
906
		spin_lock(&root->list_lock);
907
		list_add(&BTRFS_I(inode)->i_orphan, &root->orphan_list);
908
		spin_unlock(&root->list_lock);
909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942

		/*
		 * if this is a bad inode, means we actually succeeded in
		 * removing the inode, but not the orphan record, which means
		 * we need to manually delete the orphan since iput will just
		 * do a destroy_inode
		 */
		if (is_bad_inode(inode)) {
			btrfs_orphan_del(trans, inode);
			iput(inode);
			continue;
		}

		/* if we have links, this was a truncate, lets do that */
		if (inode->i_nlink) {
			nr_truncate++;
			btrfs_truncate(inode);
		} else {
			nr_unlink++;
		}

		/* this will do delete_inode and everything for us */
		iput(inode);
	}

	if (nr_unlink)
		printk(KERN_INFO "btrfs: unlinked %d orphans\n", nr_unlink);
	if (nr_truncate)
		printk(KERN_INFO "btrfs: truncated %d orphans\n", nr_truncate);

	btrfs_free_path(path);
	btrfs_end_transaction(trans, root);
}

C
Chris Mason 已提交
943 944 945
void btrfs_read_locked_inode(struct inode *inode)
{
	struct btrfs_path *path;
946
	struct extent_buffer *leaf;
C
Chris Mason 已提交
947
	struct btrfs_inode_item *inode_item;
948
	struct btrfs_timespec *tspec;
C
Chris Mason 已提交
949 950 951
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_key location;
	u64 alloc_group_block;
J
Josef Bacik 已提交
952
	u32 rdev;
C
Chris Mason 已提交
953 954 955 956 957
	int ret;

	path = btrfs_alloc_path();
	BUG_ON(!path);
	memcpy(&location, &BTRFS_I(inode)->location, sizeof(location));
958

C
Chris Mason 已提交
959
	ret = btrfs_lookup_inode(NULL, root, path, &location, 0);
960
	if (ret)
C
Chris Mason 已提交
961 962
		goto make_bad;

963 964 965 966 967 968 969 970
	leaf = path->nodes[0];
	inode_item = btrfs_item_ptr(leaf, path->slots[0],
				    struct btrfs_inode_item);

	inode->i_mode = btrfs_inode_mode(leaf, inode_item);
	inode->i_nlink = btrfs_inode_nlink(leaf, inode_item);
	inode->i_uid = btrfs_inode_uid(leaf, inode_item);
	inode->i_gid = btrfs_inode_gid(leaf, inode_item);
971
	btrfs_i_size_write(inode, btrfs_inode_size(leaf, inode_item));
972 973 974 975 976 977 978 979 980 981 982 983 984 985 986

	tspec = btrfs_inode_atime(inode_item);
	inode->i_atime.tv_sec = btrfs_timespec_sec(leaf, tspec);
	inode->i_atime.tv_nsec = btrfs_timespec_nsec(leaf, tspec);

	tspec = btrfs_inode_mtime(inode_item);
	inode->i_mtime.tv_sec = btrfs_timespec_sec(leaf, tspec);
	inode->i_mtime.tv_nsec = btrfs_timespec_nsec(leaf, tspec);

	tspec = btrfs_inode_ctime(inode_item);
	inode->i_ctime.tv_sec = btrfs_timespec_sec(leaf, tspec);
	inode->i_ctime.tv_nsec = btrfs_timespec_nsec(leaf, tspec);

	inode->i_blocks = btrfs_inode_nblocks(leaf, inode_item);
	inode->i_generation = btrfs_inode_generation(leaf, inode_item);
J
Josef Bacik 已提交
987
	inode->i_rdev = 0;
988 989
	rdev = btrfs_inode_rdev(leaf, inode_item);

990 991
	BTRFS_I(inode)->index_cnt = (u64)-1;

992
	alloc_group_block = btrfs_inode_block_group(leaf, inode_item);
C
Chris Mason 已提交
993 994
	BTRFS_I(inode)->block_group = btrfs_lookup_block_group(root->fs_info,
						       alloc_group_block);
Y
Yan 已提交
995
	BTRFS_I(inode)->flags = btrfs_inode_flags(leaf, inode_item);
996 997
	if (!BTRFS_I(inode)->block_group) {
		BTRFS_I(inode)->block_group = btrfs_find_block_group(root,
998 999
						 NULL, 0,
						 BTRFS_BLOCK_GROUP_METADATA, 0);
1000
	}
C
Chris Mason 已提交
1001 1002 1003 1004 1005 1006
	btrfs_free_path(path);
	inode_item = NULL;

	switch (inode->i_mode & S_IFMT) {
	case S_IFREG:
		inode->i_mapping->a_ops = &btrfs_aops;
1007
		inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
1008
		BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
C
Chris Mason 已提交
1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021
		inode->i_fop = &btrfs_file_operations;
		inode->i_op = &btrfs_file_inode_operations;
		break;
	case S_IFDIR:
		inode->i_fop = &btrfs_dir_file_operations;
		if (root == root->fs_info->tree_root)
			inode->i_op = &btrfs_dir_ro_inode_operations;
		else
			inode->i_op = &btrfs_dir_inode_operations;
		break;
	case S_IFLNK:
		inode->i_op = &btrfs_symlink_inode_operations;
		inode->i_mapping->a_ops = &btrfs_symlink_aops;
1022
		inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
C
Chris Mason 已提交
1023
		break;
J
Josef Bacik 已提交
1024 1025 1026
	default:
		init_special_inode(inode, inode->i_mode, rdev);
		break;
C
Chris Mason 已提交
1027 1028 1029 1030 1031 1032 1033 1034
	}
	return;

make_bad:
	btrfs_free_path(path);
	make_bad_inode(inode);
}

1035 1036
static void fill_inode_item(struct extent_buffer *leaf,
			    struct btrfs_inode_item *item,
C
Chris Mason 已提交
1037 1038
			    struct inode *inode)
{
1039 1040
	btrfs_set_inode_uid(leaf, item, inode->i_uid);
	btrfs_set_inode_gid(leaf, item, inode->i_gid);
1041
	btrfs_set_inode_size(leaf, item, BTRFS_I(inode)->disk_i_size);
1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062
	btrfs_set_inode_mode(leaf, item, inode->i_mode);
	btrfs_set_inode_nlink(leaf, item, inode->i_nlink);

	btrfs_set_timespec_sec(leaf, btrfs_inode_atime(item),
			       inode->i_atime.tv_sec);
	btrfs_set_timespec_nsec(leaf, btrfs_inode_atime(item),
				inode->i_atime.tv_nsec);

	btrfs_set_timespec_sec(leaf, btrfs_inode_mtime(item),
			       inode->i_mtime.tv_sec);
	btrfs_set_timespec_nsec(leaf, btrfs_inode_mtime(item),
				inode->i_mtime.tv_nsec);

	btrfs_set_timespec_sec(leaf, btrfs_inode_ctime(item),
			       inode->i_ctime.tv_sec);
	btrfs_set_timespec_nsec(leaf, btrfs_inode_ctime(item),
				inode->i_ctime.tv_nsec);

	btrfs_set_inode_nblocks(leaf, item, inode->i_blocks);
	btrfs_set_inode_generation(leaf, item, inode->i_generation);
	btrfs_set_inode_rdev(leaf, item, inode->i_rdev);
Y
Yan 已提交
1063
	btrfs_set_inode_flags(leaf, item, BTRFS_I(inode)->flags);
1064
	btrfs_set_inode_block_group(leaf, item,
C
Chris Mason 已提交
1065 1066 1067
				    BTRFS_I(inode)->block_group->key.objectid);
}

1068
int noinline btrfs_update_inode(struct btrfs_trans_handle *trans,
C
Chris Mason 已提交
1069 1070 1071 1072 1073
			      struct btrfs_root *root,
			      struct inode *inode)
{
	struct btrfs_inode_item *inode_item;
	struct btrfs_path *path;
1074
	struct extent_buffer *leaf;
C
Chris Mason 已提交
1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086
	int ret;

	path = btrfs_alloc_path();
	BUG_ON(!path);
	ret = btrfs_lookup_inode(trans, root, path,
				 &BTRFS_I(inode)->location, 1);
	if (ret) {
		if (ret > 0)
			ret = -ENOENT;
		goto failed;
	}

1087 1088
	leaf = path->nodes[0];
	inode_item = btrfs_item_ptr(leaf, path->slots[0],
C
Chris Mason 已提交
1089 1090
				  struct btrfs_inode_item);

1091 1092
	fill_inode_item(leaf, inode_item, inode);
	btrfs_mark_buffer_dirty(leaf);
1093
	btrfs_set_inode_last_trans(trans, inode);
C
Chris Mason 已提交
1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109
	ret = 0;
failed:
	btrfs_free_path(path);
	return ret;
}


static int btrfs_unlink_trans(struct btrfs_trans_handle *trans,
			      struct btrfs_root *root,
			      struct inode *dir,
			      struct dentry *dentry)
{
	struct btrfs_path *path;
	const char *name = dentry->d_name.name;
	int name_len = dentry->d_name.len;
	int ret = 0;
1110
	struct extent_buffer *leaf;
C
Chris Mason 已提交
1111
	struct btrfs_dir_item *di;
1112
	struct btrfs_key key;
1113
	u64 index;
C
Chris Mason 已提交
1114 1115

	path = btrfs_alloc_path();
1116 1117 1118 1119 1120
	if (!path) {
		ret = -ENOMEM;
		goto err;
	}

C
Chris Mason 已提交
1121 1122 1123 1124 1125 1126 1127 1128 1129 1130
	di = btrfs_lookup_dir_item(trans, root, path, dir->i_ino,
				    name, name_len, -1);
	if (IS_ERR(di)) {
		ret = PTR_ERR(di);
		goto err;
	}
	if (!di) {
		ret = -ENOENT;
		goto err;
	}
1131 1132
	leaf = path->nodes[0];
	btrfs_dir_item_key_to_cpu(leaf, di, &key);
C
Chris Mason 已提交
1133
	ret = btrfs_delete_one_dir_name(trans, root, path, di);
1134 1135
	if (ret)
		goto err;
C
Chris Mason 已提交
1136 1137
	btrfs_release_path(root, path);

1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148
	ret = btrfs_del_inode_ref(trans, root, name, name_len,
				  dentry->d_inode->i_ino,
				  dentry->d_parent->d_inode->i_ino, &index);
	if (ret) {
		printk("failed to delete reference to %.*s, "
		       "inode %lu parent %lu\n", name_len, name,
		       dentry->d_inode->i_ino,
		       dentry->d_parent->d_inode->i_ino);
		goto err;
	}

C
Chris Mason 已提交
1149
	di = btrfs_lookup_dir_index_item(trans, root, path, dir->i_ino,
1150
					 index, name, name_len, -1);
C
Chris Mason 已提交
1151 1152 1153 1154 1155 1156 1157 1158 1159
	if (IS_ERR(di)) {
		ret = PTR_ERR(di);
		goto err;
	}
	if (!di) {
		ret = -ENOENT;
		goto err;
	}
	ret = btrfs_delete_one_dir_name(trans, root, path, di);
1160
	btrfs_release_path(root, path);
C
Chris Mason 已提交
1161 1162 1163 1164 1165

	dentry->d_inode->i_ctime = dir->i_ctime;
err:
	btrfs_free_path(path);
	if (!ret) {
1166
		btrfs_i_size_write(dir, dir->i_size - name_len * 2);
1167
		dir->i_mtime = dir->i_ctime = CURRENT_TIME;
C
Chris Mason 已提交
1168
		btrfs_update_inode(trans, root, dir);
1169 1170 1171
#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,18)
		dentry->d_inode->i_nlink--;
#else
C
Chris Mason 已提交
1172
		drop_nlink(dentry->d_inode);
1173
#endif
1174
		ret = btrfs_update_inode(trans, root, dentry->d_inode);
C
Chris Mason 已提交
1175 1176 1177 1178 1179 1180 1181 1182 1183
		dir->i_sb->s_dirt = 1;
	}
	return ret;
}

static int btrfs_unlink(struct inode *dir, struct dentry *dentry)
{
	struct btrfs_root *root;
	struct btrfs_trans_handle *trans;
1184
	struct inode *inode = dentry->d_inode;
C
Chris Mason 已提交
1185
	int ret;
1186
	unsigned long nr = 0;
C
Chris Mason 已提交
1187 1188

	root = BTRFS_I(dir)->root;
1189 1190 1191 1192 1193

	ret = btrfs_check_free_space(root, 1, 1);
	if (ret)
		goto fail;

C
Chris Mason 已提交
1194
	trans = btrfs_start_transaction(root, 1);
1195

C
Chris Mason 已提交
1196 1197
	btrfs_set_trans_block_group(trans, dir);
	ret = btrfs_unlink_trans(trans, root, dir, dentry);
1198 1199 1200 1201

	if (inode->i_nlink == 0)
		ret = btrfs_orphan_add(trans, inode);

1202
	nr = trans->blocks_used;
1203

1204
	btrfs_end_transaction_throttle(trans, root);
1205
fail:
1206
	btrfs_btree_balance_dirty(root, nr);
C
Chris Mason 已提交
1207 1208 1209 1210 1211 1212
	return ret;
}

static int btrfs_rmdir(struct inode *dir, struct dentry *dentry)
{
	struct inode *inode = dentry->d_inode;
1213
	int err = 0;
C
Chris Mason 已提交
1214 1215 1216
	int ret;
	struct btrfs_root *root = BTRFS_I(dir)->root;
	struct btrfs_trans_handle *trans;
1217
	unsigned long nr = 0;
C
Chris Mason 已提交
1218

1219
	if (inode->i_size > BTRFS_EMPTY_DIR_SIZE) {
Y
Yan 已提交
1220
		return -ENOTEMPTY;
1221
	}
Y
Yan 已提交
1222

1223 1224 1225 1226
	ret = btrfs_check_free_space(root, 1, 1);
	if (ret)
		goto fail;

C
Chris Mason 已提交
1227 1228 1229
	trans = btrfs_start_transaction(root, 1);
	btrfs_set_trans_block_group(trans, dir);

1230 1231 1232 1233
	err = btrfs_orphan_add(trans, inode);
	if (err)
		goto fail_trans;

C
Chris Mason 已提交
1234 1235 1236
	/* now the directory is empty */
	err = btrfs_unlink_trans(trans, root, dir, dentry);
	if (!err) {
1237
		btrfs_i_size_write(inode, 0);
C
Chris Mason 已提交
1238
	}
1239

1240
fail_trans:
1241
	nr = trans->blocks_used;
1242
	ret = btrfs_end_transaction_throttle(trans, root);
1243
fail:
1244
	btrfs_btree_balance_dirty(root, nr);
1245

C
Chris Mason 已提交
1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257
	if (ret && !err)
		err = ret;
	return err;
}

/*
 * this can truncate away extent items, csum items and directory items.
 * It starts at a high offset and removes keys until it can't find
 * any higher than i_size.
 *
 * csum items that cross the new i_size are truncated to the new size
 * as well.
1258 1259 1260
 *
 * min_type is the minimum key type to truncate down to.  If set to 0, this
 * will kill all the items on this inode, including the INODE_ITEM_KEY.
C
Chris Mason 已提交
1261 1262 1263
 */
static int btrfs_truncate_in_trans(struct btrfs_trans_handle *trans,
				   struct btrfs_root *root,
1264 1265
				   struct inode *inode,
				   u32 min_type)
C
Chris Mason 已提交
1266 1267 1268 1269
{
	int ret;
	struct btrfs_path *path;
	struct btrfs_key key;
1270
	struct btrfs_key found_key;
C
Chris Mason 已提交
1271
	u32 found_type;
1272
	struct extent_buffer *leaf;
C
Chris Mason 已提交
1273 1274
	struct btrfs_file_extent_item *fi;
	u64 extent_start = 0;
1275
	u64 extent_num_bytes = 0;
C
Chris Mason 已提交
1276
	u64 item_end = 0;
1277
	u64 root_gen = 0;
1278
	u64 root_owner = 0;
C
Chris Mason 已提交
1279 1280
	int found_extent;
	int del_item;
1281 1282
	int pending_del_nr = 0;
	int pending_del_slot = 0;
1283
	int extent_type = -1;
1284
	u64 mask = root->sectorsize - 1;
C
Chris Mason 已提交
1285

1286
	btrfs_drop_extent_cache(inode, inode->i_size & (~mask), (u64)-1);
C
Chris Mason 已提交
1287
	path = btrfs_alloc_path();
1288
	path->reada = -1;
C
Chris Mason 已提交
1289
	BUG_ON(!path);
1290

C
Chris Mason 已提交
1291 1292 1293
	/* FIXME, add redo link to tree so we don't leak on crash */
	key.objectid = inode->i_ino;
	key.offset = (u64)-1;
1294 1295
	key.type = (u8)-1;

1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306
	btrfs_init_path(path);
search_again:
	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
	if (ret < 0) {
		goto error;
	}
	if (ret > 0) {
		BUG_ON(path->slots[0] == 0);
		path->slots[0]--;
	}

C
Chris Mason 已提交
1307 1308
	while(1) {
		fi = NULL;
1309 1310 1311
		leaf = path->nodes[0];
		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
		found_type = btrfs_key_type(&found_key);
C
Chris Mason 已提交
1312

1313
		if (found_key.objectid != inode->i_ino)
C
Chris Mason 已提交
1314
			break;
1315

1316
		if (found_type < min_type)
C
Chris Mason 已提交
1317 1318
			break;

1319
		item_end = found_key.offset;
C
Chris Mason 已提交
1320
		if (found_type == BTRFS_EXTENT_DATA_KEY) {
1321
			fi = btrfs_item_ptr(leaf, path->slots[0],
C
Chris Mason 已提交
1322
					    struct btrfs_file_extent_item);
1323 1324
			extent_type = btrfs_file_extent_type(leaf, fi);
			if (extent_type != BTRFS_FILE_EXTENT_INLINE) {
1325
				item_end +=
1326
				    btrfs_file_extent_num_bytes(leaf, fi);
1327 1328 1329 1330 1331
			} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
				struct btrfs_item *item = btrfs_item_nr(leaf,
							        path->slots[0]);
				item_end += btrfs_file_extent_inline_len(leaf,
									 item);
C
Chris Mason 已提交
1332
			}
1333
			item_end--;
C
Chris Mason 已提交
1334 1335 1336 1337 1338 1339
		}
		if (found_type == BTRFS_CSUM_ITEM_KEY) {
			ret = btrfs_csum_truncate(trans, root, path,
						  inode->i_size);
			BUG_ON(ret);
		}
1340
		if (item_end < inode->i_size) {
1341 1342 1343 1344
			if (found_type == BTRFS_DIR_ITEM_KEY) {
				found_type = BTRFS_INODE_ITEM_KEY;
			} else if (found_type == BTRFS_EXTENT_ITEM_KEY) {
				found_type = BTRFS_CSUM_ITEM_KEY;
1345 1346 1347 1348
			} else if (found_type == BTRFS_EXTENT_DATA_KEY) {
				found_type = BTRFS_XATTR_ITEM_KEY;
			} else if (found_type == BTRFS_XATTR_ITEM_KEY) {
				found_type = BTRFS_INODE_REF_KEY;
1349 1350 1351 1352
			} else if (found_type) {
				found_type--;
			} else {
				break;
C
Chris Mason 已提交
1353
			}
1354
			btrfs_set_key_type(&key, found_type);
1355
			goto next;
C
Chris Mason 已提交
1356
		}
1357
		if (found_key.offset >= inode->i_size)
C
Chris Mason 已提交
1358 1359 1360 1361 1362 1363
			del_item = 1;
		else
			del_item = 0;
		found_extent = 0;

		/* FIXME, shrink the extent if the ref count is only 1 */
1364 1365 1366 1367
		if (found_type != BTRFS_EXTENT_DATA_KEY)
			goto delete;

		if (extent_type != BTRFS_FILE_EXTENT_INLINE) {
C
Chris Mason 已提交
1368
			u64 num_dec;
1369
			extent_start = btrfs_file_extent_disk_bytenr(leaf, fi);
C
Chris Mason 已提交
1370
			if (!del_item) {
1371 1372 1373
				u64 orig_num_bytes =
					btrfs_file_extent_num_bytes(leaf, fi);
				extent_num_bytes = inode->i_size -
1374
					found_key.offset + root->sectorsize - 1;
1375 1376
				extent_num_bytes = extent_num_bytes &
					~((u64)root->sectorsize - 1);
1377 1378 1379
				btrfs_set_file_extent_num_bytes(leaf, fi,
							 extent_num_bytes);
				num_dec = (orig_num_bytes -
1380 1381 1382
					   extent_num_bytes);
				if (extent_start != 0)
					dec_i_blocks(inode, num_dec);
1383
				btrfs_mark_buffer_dirty(leaf);
C
Chris Mason 已提交
1384
			} else {
1385 1386 1387
				extent_num_bytes =
					btrfs_file_extent_disk_num_bytes(leaf,
									 fi);
C
Chris Mason 已提交
1388
				/* FIXME blocksize != 4096 */
1389
				num_dec = btrfs_file_extent_num_bytes(leaf, fi);
C
Chris Mason 已提交
1390 1391
				if (extent_start != 0) {
					found_extent = 1;
1392
					dec_i_blocks(inode, num_dec);
C
Chris Mason 已提交
1393
				}
1394 1395
				root_gen = btrfs_header_generation(leaf);
				root_owner = btrfs_header_owner(leaf);
C
Chris Mason 已提交
1396
			}
1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410
		} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
			if (!del_item) {
				u32 newsize = inode->i_size - found_key.offset;
				dec_i_blocks(inode, item_end + 1 -
					    found_key.offset - newsize);
				newsize =
				    btrfs_file_extent_calc_inline_size(newsize);
				ret = btrfs_truncate_item(trans, root, path,
							  newsize, 1);
				BUG_ON(ret);
			} else {
				dec_i_blocks(inode, item_end + 1 -
					     found_key.offset);
			}
C
Chris Mason 已提交
1411
		}
1412
delete:
C
Chris Mason 已提交
1413
		if (del_item) {
1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425
			if (!pending_del_nr) {
				/* no pending yet, add ourselves */
				pending_del_slot = path->slots[0];
				pending_del_nr = 1;
			} else if (pending_del_nr &&
				   path->slots[0] + 1 == pending_del_slot) {
				/* hop on the pending chunk */
				pending_del_nr++;
				pending_del_slot = path->slots[0];
			} else {
				printk("bad pending slot %d pending_del_nr %d pending_del_slot %d\n", path->slots[0], pending_del_nr, pending_del_slot);
			}
C
Chris Mason 已提交
1426 1427 1428 1429 1430
		} else {
			break;
		}
		if (found_extent) {
			ret = btrfs_free_extent(trans, root, extent_start,
1431
						extent_num_bytes,
1432
						root_owner,
1433 1434
						root_gen, inode->i_ino,
						found_key.offset, 0);
C
Chris Mason 已提交
1435 1436
			BUG_ON(ret);
		}
1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459
next:
		if (path->slots[0] == 0) {
			if (pending_del_nr)
				goto del_pending;
			btrfs_release_path(root, path);
			goto search_again;
		}

		path->slots[0]--;
		if (pending_del_nr &&
		    path->slots[0] + 1 != pending_del_slot) {
			struct btrfs_key debug;
del_pending:
			btrfs_item_key_to_cpu(path->nodes[0], &debug,
					      pending_del_slot);
			ret = btrfs_del_items(trans, root, path,
					      pending_del_slot,
					      pending_del_nr);
			BUG_ON(ret);
			pending_del_nr = 0;
			btrfs_release_path(root, path);
			goto search_again;
		}
C
Chris Mason 已提交
1460 1461 1462
	}
	ret = 0;
error:
1463 1464 1465 1466
	if (pending_del_nr) {
		ret = btrfs_del_items(trans, root, path, pending_del_slot,
				      pending_del_nr);
	}
C
Chris Mason 已提交
1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478
	btrfs_free_path(path);
	inode->i_sb->s_dirt = 1;
	return ret;
}

/*
 * taken from block_truncate_page, but does cow as it zeros out
 * any bytes left in the last page in the file.
 */
static int btrfs_truncate_page(struct address_space *mapping, loff_t from)
{
	struct inode *inode = mapping->host;
1479
	struct btrfs_root *root = BTRFS_I(inode)->root;
1480 1481 1482
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	struct btrfs_ordered_extent *ordered;
	char *kaddr;
1483
	u32 blocksize = root->sectorsize;
C
Chris Mason 已提交
1484 1485 1486 1487
	pgoff_t index = from >> PAGE_CACHE_SHIFT;
	unsigned offset = from & (PAGE_CACHE_SIZE-1);
	struct page *page;
	int ret = 0;
1488
	u64 page_start;
1489
	u64 page_end;
C
Chris Mason 已提交
1490 1491 1492 1493 1494

	if ((offset & (blocksize - 1)) == 0)
		goto out;

	ret = -ENOMEM;
1495
again:
C
Chris Mason 已提交
1496 1497 1498
	page = grab_cache_page(mapping, index);
	if (!page)
		goto out;
1499 1500 1501 1502

	page_start = page_offset(page);
	page_end = page_start + PAGE_CACHE_SIZE - 1;

C
Chris Mason 已提交
1503
	if (!PageUptodate(page)) {
1504
		ret = btrfs_readpage(NULL, page);
C
Chris Mason 已提交
1505
		lock_page(page);
1506 1507 1508 1509 1510
		if (page->mapping != mapping) {
			unlock_page(page);
			page_cache_release(page);
			goto again;
		}
C
Chris Mason 已提交
1511 1512
		if (!PageUptodate(page)) {
			ret = -EIO;
1513
			goto out_unlock;
C
Chris Mason 已提交
1514 1515
		}
	}
1516
	wait_on_page_writeback(page);
1517 1518 1519 1520 1521 1522 1523 1524 1525

	lock_extent(io_tree, page_start, page_end, GFP_NOFS);
	set_page_extent_mapped(page);

	ordered = btrfs_lookup_ordered_extent(inode, page_start);
	if (ordered) {
		unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
		unlock_page(page);
		page_cache_release(page);
1526
		btrfs_start_ordered_extent(inode, ordered, 1);
1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539
		btrfs_put_ordered_extent(ordered);
		goto again;
	}

	set_extent_delalloc(&BTRFS_I(inode)->io_tree, page_start,
			    page_end, GFP_NOFS);
	ret = 0;
	if (offset != PAGE_CACHE_SIZE) {
		kaddr = kmap(page);
		memset(kaddr + offset, 0, PAGE_CACHE_SIZE - offset);
		flush_dcache_page(page);
		kunmap(page);
	}
1540
	ClearPageChecked(page);
1541 1542
	set_page_dirty(page);
	unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
C
Chris Mason 已提交
1543

1544
out_unlock:
C
Chris Mason 已提交
1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563
	unlock_page(page);
	page_cache_release(page);
out:
	return ret;
}

static int btrfs_setattr(struct dentry *dentry, struct iattr *attr)
{
	struct inode *inode = dentry->d_inode;
	int err;

	err = inode_change_ok(inode, attr);
	if (err)
		return err;

	if (S_ISREG(inode->i_mode) &&
	    attr->ia_valid & ATTR_SIZE && attr->ia_size > inode->i_size) {
		struct btrfs_trans_handle *trans;
		struct btrfs_root *root = BTRFS_I(inode)->root;
1564
		struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
C
Chris Mason 已提交
1565

1566
		u64 mask = root->sectorsize - 1;
1567
		u64 hole_start = (inode->i_size + mask) & ~mask;
1568
		u64 block_end = (attr->ia_size + mask) & ~mask;
C
Chris Mason 已提交
1569
		u64 hole_size;
1570
		u64 alloc_hint = 0;
C
Chris Mason 已提交
1571

1572
		if (attr->ia_size <= hole_start)
C
Chris Mason 已提交
1573 1574
			goto out;

1575 1576 1577 1578
		err = btrfs_check_free_space(root, 1, 0);
		if (err)
			goto fail;

C
Chris Mason 已提交
1579 1580
		btrfs_truncate_page(inode->i_mapping, inode->i_size);

1581
		hole_size = block_end - hole_start;
1582 1583
		btrfs_wait_ordered_range(inode, hole_start, hole_size);
		lock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS);
C
Chris Mason 已提交
1584 1585 1586

		trans = btrfs_start_transaction(root, 1);
		btrfs_set_trans_block_group(trans, inode);
1587
		mutex_lock(&BTRFS_I(inode)->extent_mutex);
C
Chris Mason 已提交
1588
		err = btrfs_drop_extents(trans, root, inode,
1589
					 hole_start, block_end, hole_start,
1590
					 &alloc_hint);
C
Chris Mason 已提交
1591

1592 1593 1594
		if (alloc_hint != EXTENT_MAP_INLINE) {
			err = btrfs_insert_file_extent(trans, root,
						       inode->i_ino,
1595
						       hole_start, 0, 0,
S
Sage Weil 已提交
1596
						       hole_size, 0);
1597
			btrfs_drop_extent_cache(inode, hole_start,
1598
						(u64)-1);
1599
			btrfs_check_file(root, inode);
1600
		}
1601
		mutex_unlock(&BTRFS_I(inode)->extent_mutex);
C
Chris Mason 已提交
1602
		btrfs_end_transaction(trans, root);
1603
		unlock_extent(io_tree, hole_start, block_end - 1, GFP_NOFS);
1604 1605
		if (err)
			return err;
C
Chris Mason 已提交
1606 1607 1608
	}
out:
	err = inode_setattr(inode, attr);
J
Josef Bacik 已提交
1609 1610 1611

	if (!err && ((attr->ia_valid & ATTR_MODE)))
		err = btrfs_acl_chmod(inode);
1612
fail:
C
Chris Mason 已提交
1613 1614
	return err;
}
1615

C
Chris Mason 已提交
1616 1617 1618 1619
void btrfs_delete_inode(struct inode *inode)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_root *root = BTRFS_I(inode)->root;
1620
	unsigned long nr;
C
Chris Mason 已提交
1621 1622 1623 1624
	int ret;

	truncate_inode_pages(&inode->i_data, 0);
	if (is_bad_inode(inode)) {
1625
		btrfs_orphan_del(NULL, inode);
C
Chris Mason 已提交
1626 1627
		goto no_delete;
	}
C
Chris Mason 已提交
1628
	btrfs_wait_ordered_range(inode, 0, (u64)-1);
1629

1630
	btrfs_i_size_write(inode, 0);
C
Chris Mason 已提交
1631
	trans = btrfs_start_transaction(root, 1);
1632

C
Chris Mason 已提交
1633
	btrfs_set_trans_block_group(trans, inode);
1634
	ret = btrfs_truncate_in_trans(trans, root, inode, 0);
1635 1636
	if (ret) {
		btrfs_orphan_del(NULL, inode);
1637
		goto no_delete_lock;
1638 1639 1640
	}

	btrfs_orphan_del(trans, inode);
1641

1642
	nr = trans->blocks_used;
1643
	clear_inode(inode);
1644

C
Chris Mason 已提交
1645
	btrfs_end_transaction(trans, root);
1646
	btrfs_btree_balance_dirty(root, nr);
C
Chris Mason 已提交
1647
	return;
1648 1649

no_delete_lock:
1650
	nr = trans->blocks_used;
1651
	btrfs_end_transaction(trans, root);
1652
	btrfs_btree_balance_dirty(root, nr);
C
Chris Mason 已提交
1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668
no_delete:
	clear_inode(inode);
}

/*
 * this returns the key found in the dir entry in the location pointer.
 * If no dir entries were found, location->objectid is 0.
 */
static int btrfs_inode_by_name(struct inode *dir, struct dentry *dentry,
			       struct btrfs_key *location)
{
	const char *name = dentry->d_name.name;
	int namelen = dentry->d_name.len;
	struct btrfs_dir_item *di;
	struct btrfs_path *path;
	struct btrfs_root *root = BTRFS_I(dir)->root;
1669
	int ret = 0;
C
Chris Mason 已提交
1670

1671 1672 1673 1674 1675 1676
	if (namelen == 1 && strcmp(name, ".") == 0) {
		location->objectid = dir->i_ino;
		location->type = BTRFS_INODE_ITEM_KEY;
		location->offset = 0;
		return 0;
	}
C
Chris Mason 已提交
1677 1678
	path = btrfs_alloc_path();
	BUG_ON(!path);
1679

1680
	if (namelen == 2 && strcmp(name, "..") == 0) {
1681 1682 1683 1684 1685
		struct btrfs_key key;
		struct extent_buffer *leaf;
		int slot;

		key.objectid = dir->i_ino;
Y
Yan 已提交
1686
		key.offset = (u64)-1;
1687
		btrfs_set_key_type(&key, BTRFS_INODE_REF_KEY);
Y
Yan 已提交
1688 1689
		if (ret < 0 || path->slots[0] == 0)
			goto out_err;
1690 1691 1692 1693
		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
		BUG_ON(ret == 0);
		ret = 0;
		leaf = path->nodes[0];
Y
Yan 已提交
1694
		slot = path->slots[0] - 1;
1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706

		btrfs_item_key_to_cpu(leaf, &key, slot);
		if (key.objectid != dir->i_ino ||
		    key.type != BTRFS_INODE_REF_KEY) {
			goto out_err;
		}
		location->objectid = key.offset;
		location->type = BTRFS_INODE_ITEM_KEY;
		location->offset = 0;
		goto out;
	}

C
Chris Mason 已提交
1707 1708
	di = btrfs_lookup_dir_item(NULL, root, path, dir->i_ino, name,
				    namelen, 0);
1709 1710
	if (IS_ERR(di))
		ret = PTR_ERR(di);
C
Chris Mason 已提交
1711
	if (!di || IS_ERR(di)) {
1712
		goto out_err;
C
Chris Mason 已提交
1713
	}
1714
	btrfs_dir_item_key_to_cpu(path->nodes[0], di, location);
C
Chris Mason 已提交
1715 1716 1717
out:
	btrfs_free_path(path);
	return ret;
1718 1719 1720
out_err:
	location->objectid = 0;
	goto out;
C
Chris Mason 已提交
1721 1722 1723 1724 1725 1726 1727 1728 1729
}

/*
 * when we hit a tree root in a directory, the btrfs part of the inode
 * needs to be changed to reflect the root directory of the tree root.  This
 * is kind of like crossing a mount point.
 */
static int fixup_tree_root_location(struct btrfs_root *root,
			     struct btrfs_key *location,
1730 1731
			     struct btrfs_root **sub_root,
			     struct dentry *dentry)
C
Chris Mason 已提交
1732 1733 1734 1735 1736 1737 1738 1739
{
	struct btrfs_root_item *ri;

	if (btrfs_key_type(location) != BTRFS_ROOT_ITEM_KEY)
		return 0;
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return 0;

1740 1741 1742
	*sub_root = btrfs_read_fs_root(root->fs_info, location,
					dentry->d_name.name,
					dentry->d_name.len);
C
Chris Mason 已提交
1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758
	if (IS_ERR(*sub_root))
		return PTR_ERR(*sub_root);

	ri = &(*sub_root)->root_item;
	location->objectid = btrfs_root_dirid(ri);
	btrfs_set_key_type(location, BTRFS_INODE_ITEM_KEY);
	location->offset = 0;

	return 0;
}

static int btrfs_init_locked_inode(struct inode *inode, void *p)
{
	struct btrfs_iget_args *args = p;
	inode->i_ino = args->ino;
	BTRFS_I(inode)->root = args->root;
1759
	BTRFS_I(inode)->delalloc_bytes = 0;
1760
	BTRFS_I(inode)->disk_i_size = 0;
1761
	BTRFS_I(inode)->index_cnt = (u64)-1;
1762 1763
	extent_map_tree_init(&BTRFS_I(inode)->extent_tree, GFP_NOFS);
	extent_io_tree_init(&BTRFS_I(inode)->io_tree,
1764
			     inode->i_mapping, GFP_NOFS);
1765 1766
	extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree,
			     inode->i_mapping, GFP_NOFS);
1767
	btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree);
1768
	mutex_init(&BTRFS_I(inode)->csum_mutex);
1769
	mutex_init(&BTRFS_I(inode)->extent_mutex);
C
Chris Mason 已提交
1770 1771 1772 1773 1774 1775 1776 1777 1778 1779
	return 0;
}

static int btrfs_find_actor(struct inode *inode, void *opaque)
{
	struct btrfs_iget_args *args = opaque;
	return (args->ino == inode->i_ino &&
		args->root == BTRFS_I(inode)->root);
}

1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792
struct inode *btrfs_ilookup(struct super_block *s, u64 objectid,
			    u64 root_objectid)
{
	struct btrfs_iget_args args;
	args.ino = objectid;
	args.root = btrfs_lookup_fs_root(btrfs_sb(s)->fs_info, root_objectid);

	if (!args.root)
		return NULL;

	return ilookup5(s, objectid, btrfs_find_actor, (void *)&args);
}

C
Chris Mason 已提交
1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814
struct inode *btrfs_iget_locked(struct super_block *s, u64 objectid,
				struct btrfs_root *root)
{
	struct inode *inode;
	struct btrfs_iget_args args;
	args.ino = objectid;
	args.root = root;

	inode = iget5_locked(s, objectid, btrfs_find_actor,
			     btrfs_init_locked_inode,
			     (void *)&args);
	return inode;
}

static struct dentry *btrfs_lookup(struct inode *dir, struct dentry *dentry,
				   struct nameidata *nd)
{
	struct inode * inode;
	struct btrfs_inode *bi = BTRFS_I(dir);
	struct btrfs_root *root = bi->root;
	struct btrfs_root *sub_root = root;
	struct btrfs_key location;
1815
	int ret, do_orphan = 0;
C
Chris Mason 已提交
1816 1817 1818

	if (dentry->d_name.len > BTRFS_NAME_LEN)
		return ERR_PTR(-ENAMETOOLONG);
1819

C
Chris Mason 已提交
1820
	ret = btrfs_inode_by_name(dir, dentry, &location);
1821

C
Chris Mason 已提交
1822 1823
	if (ret < 0)
		return ERR_PTR(ret);
1824

C
Chris Mason 已提交
1825 1826
	inode = NULL;
	if (location.objectid) {
1827 1828
		ret = fixup_tree_root_location(root, &location, &sub_root,
						dentry);
C
Chris Mason 已提交
1829 1830 1831 1832
		if (ret < 0)
			return ERR_PTR(ret);
		if (ret > 0)
			return ERR_PTR(-ENOENT);
1833

C
Chris Mason 已提交
1834 1835 1836 1837 1838 1839 1840 1841 1842
		inode = btrfs_iget_locked(dir->i_sb, location.objectid,
					  sub_root);
		if (!inode)
			return ERR_PTR(-EACCES);
		if (inode->i_state & I_NEW) {
			/* the inode and parent dir are two different roots */
			if (sub_root != root) {
				igrab(inode);
				sub_root->inode = inode;
1843
				do_orphan = 1;
C
Chris Mason 已提交
1844 1845 1846 1847 1848 1849 1850 1851
			}
			BTRFS_I(inode)->root = sub_root;
			memcpy(&BTRFS_I(inode)->location, &location,
			       sizeof(location));
			btrfs_read_locked_inode(inode);
			unlock_new_inode(inode);
		}
	}
1852 1853 1854 1855

	if (unlikely(do_orphan))
		btrfs_orphan_cleanup(sub_root);

C
Chris Mason 已提交
1856 1857 1858 1859 1860 1861 1862 1863 1864
	return d_splice_alias(inode, dentry);
}

static unsigned char btrfs_filetype_table[] = {
	DT_UNKNOWN, DT_REG, DT_DIR, DT_CHR, DT_BLK, DT_FIFO, DT_SOCK, DT_LNK
};

static int btrfs_readdir(struct file *filp, void *dirent, filldir_t filldir)
{
1865
	struct inode *inode = filp->f_dentry->d_inode;
C
Chris Mason 已提交
1866 1867 1868 1869
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_item *item;
	struct btrfs_dir_item *di;
	struct btrfs_key key;
1870
	struct btrfs_key found_key;
C
Chris Mason 已提交
1871 1872 1873
	struct btrfs_path *path;
	int ret;
	u32 nritems;
1874
	struct extent_buffer *leaf;
C
Chris Mason 已提交
1875 1876 1877 1878 1879 1880 1881 1882
	int slot;
	int advance;
	unsigned char d_type;
	int over = 0;
	u32 di_cur;
	u32 di_total;
	u32 di_len;
	int key_type = BTRFS_DIR_INDEX_KEY;
1883 1884 1885
	char tmp_name[32];
	char *name_ptr;
	int name_len;
C
Chris Mason 已提交
1886 1887 1888 1889

	/* FIXME, use a real flag for deciding about the key type */
	if (root->fs_info->tree_root == root)
		key_type = BTRFS_DIR_ITEM_KEY;
1890

1891 1892 1893 1894 1895 1896 1897 1898 1899 1900
	/* special case for "." */
	if (filp->f_pos == 0) {
		over = filldir(dirent, ".", 1,
			       1, inode->i_ino,
			       DT_DIR);
		if (over)
			return 0;
		filp->f_pos = 1;
	}

C
Chris Mason 已提交
1901
	key.objectid = inode->i_ino;
1902 1903 1904 1905 1906 1907
	path = btrfs_alloc_path();
	path->reada = 2;

	/* special case for .., just use the back ref */
	if (filp->f_pos == 1) {
		btrfs_set_key_type(&key, BTRFS_INODE_REF_KEY);
Y
Yan 已提交
1908
		key.offset = (u64)-1;
1909
		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
Y
Yan 已提交
1910
		if (ret < 0 || path->slots[0] == 0) {
1911 1912 1913
			btrfs_release_path(root, path);
			goto read_dir_items;
		}
Y
Yan 已提交
1914 1915 1916
		BUG_ON(ret == 0);
		leaf = path->nodes[0];
		slot = path->slots[0] - 1;
1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929
		btrfs_item_key_to_cpu(leaf, &found_key, slot);
		btrfs_release_path(root, path);
		if (found_key.objectid != key.objectid ||
		    found_key.type != BTRFS_INODE_REF_KEY)
			goto read_dir_items;
		over = filldir(dirent, "..", 2,
			       2, found_key.offset, DT_DIR);
		if (over)
			goto nopos;
		filp->f_pos = 2;
	}

read_dir_items:
C
Chris Mason 已提交
1930 1931
	btrfs_set_key_type(&key, key_type);
	key.offset = filp->f_pos;
1932

C
Chris Mason 已提交
1933 1934 1935 1936 1937
	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
	if (ret < 0)
		goto err;
	advance = 0;
	while(1) {
1938 1939
		leaf = path->nodes[0];
		nritems = btrfs_header_nritems(leaf);
C
Chris Mason 已提交
1940 1941 1942 1943 1944 1945
		slot = path->slots[0];
		if (advance || slot >= nritems) {
			if (slot >= nritems -1) {
				ret = btrfs_next_leaf(root, path);
				if (ret)
					break;
1946 1947
				leaf = path->nodes[0];
				nritems = btrfs_header_nritems(leaf);
C
Chris Mason 已提交
1948 1949 1950 1951 1952 1953 1954
				slot = path->slots[0];
			} else {
				slot++;
				path->slots[0]++;
			}
		}
		advance = 1;
1955 1956 1957 1958
		item = btrfs_item_nr(leaf, slot);
		btrfs_item_key_to_cpu(leaf, &found_key, slot);

		if (found_key.objectid != key.objectid)
C
Chris Mason 已提交
1959
			break;
1960
		if (btrfs_key_type(&found_key) != key_type)
C
Chris Mason 已提交
1961
			break;
1962
		if (found_key.offset < filp->f_pos)
C
Chris Mason 已提交
1963
			continue;
1964 1965

		filp->f_pos = found_key.offset;
C
Chris Mason 已提交
1966 1967 1968
		advance = 1;
		di = btrfs_item_ptr(leaf, slot, struct btrfs_dir_item);
		di_cur = 0;
1969
		di_total = btrfs_item_size(leaf, item);
C
Chris Mason 已提交
1970
		while(di_cur < di_total) {
1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987
			struct btrfs_key location;

			name_len = btrfs_dir_name_len(leaf, di);
			if (name_len < 32) {
				name_ptr = tmp_name;
			} else {
				name_ptr = kmalloc(name_len, GFP_NOFS);
				BUG_ON(!name_ptr);
			}
			read_extent_buffer(leaf, name_ptr,
					   (unsigned long)(di + 1), name_len);

			d_type = btrfs_filetype_table[btrfs_dir_type(leaf, di)];
			btrfs_dir_item_key_to_cpu(leaf, di, &location);
			over = filldir(dirent, name_ptr, name_len,
				       found_key.offset,
				       location.objectid,
C
Chris Mason 已提交
1988
				       d_type);
1989 1990 1991 1992

			if (name_ptr != tmp_name)
				kfree(name_ptr);

C
Chris Mason 已提交
1993 1994
			if (over)
				goto nopos;
J
Josef Bacik 已提交
1995 1996
			di_len = btrfs_dir_name_len(leaf, di) +
				btrfs_dir_data_len(leaf, di) +sizeof(*di);
C
Chris Mason 已提交
1997 1998 1999 2000
			di_cur += di_len;
			di = (struct btrfs_dir_item *)((char *)di + di_len);
		}
	}
2001 2002 2003 2004
	if (key_type == BTRFS_DIR_INDEX_KEY)
		filp->f_pos = INT_LIMIT(typeof(filp->f_pos));
	else
		filp->f_pos++;
C
Chris Mason 已提交
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
nopos:
	ret = 0;
err:
	btrfs_free_path(path);
	return ret;
}

int btrfs_write_inode(struct inode *inode, int wait)
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_trans_handle *trans;
	int ret = 0;

	if (wait) {
2019
		trans = btrfs_join_transaction(root, 1);
C
Chris Mason 已提交
2020 2021 2022 2023 2024 2025 2026
		btrfs_set_trans_block_group(trans, inode);
		ret = btrfs_commit_transaction(trans, root);
	}
	return ret;
}

/*
2027
 * This is somewhat expensive, updating the tree every time the
C
Chris Mason 已提交
2028 2029 2030 2031 2032 2033 2034 2035 2036
 * inode changes.  But, it is most likely to find the inode in cache.
 * FIXME, needs more benchmarking...there are no reasons other than performance
 * to keep or drop this code.
 */
void btrfs_dirty_inode(struct inode *inode)
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_trans_handle *trans;

2037
	trans = btrfs_join_transaction(root, 1);
C
Chris Mason 已提交
2038 2039 2040 2041 2042
	btrfs_set_trans_block_group(trans, inode);
	btrfs_update_inode(trans, root, inode);
	btrfs_end_transaction(trans, root);
}

2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110
static int btrfs_set_inode_index_count(struct inode *inode)
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_key key, found_key;
	struct btrfs_path *path;
	struct extent_buffer *leaf;
	int ret;

	key.objectid = inode->i_ino;
	btrfs_set_key_type(&key, BTRFS_DIR_INDEX_KEY);
	key.offset = (u64)-1;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
	if (ret < 0)
		goto out;
	/* FIXME: we should be able to handle this */
	if (ret == 0)
		goto out;
	ret = 0;

	/*
	 * MAGIC NUMBER EXPLANATION:
	 * since we search a directory based on f_pos we have to start at 2
	 * since '.' and '..' have f_pos of 0 and 1 respectively, so everybody
	 * else has to start at 2
	 */
	if (path->slots[0] == 0) {
		BTRFS_I(inode)->index_cnt = 2;
		goto out;
	}

	path->slots[0]--;

	leaf = path->nodes[0];
	btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);

	if (found_key.objectid != inode->i_ino ||
	    btrfs_key_type(&found_key) != BTRFS_DIR_INDEX_KEY) {
		BTRFS_I(inode)->index_cnt = 2;
		goto out;
	}

	BTRFS_I(inode)->index_cnt = found_key.offset + 1;
out:
	btrfs_free_path(path);
	return ret;
}

static int btrfs_set_inode_index(struct inode *dir, struct inode *inode)
{
	int ret = 0;

	if (BTRFS_I(dir)->index_cnt == (u64)-1) {
		ret = btrfs_set_inode_index_count(dir);
		if (ret)
			return ret;
	}

	BTRFS_I(inode)->index = BTRFS_I(dir)->index_cnt;
	BTRFS_I(dir)->index_cnt++;

	return ret;
}

C
Chris Mason 已提交
2111 2112
static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
				     struct btrfs_root *root,
2113
				     struct inode *dir,
2114 2115
				     const char *name, int name_len,
				     u64 ref_objectid,
C
Chris Mason 已提交
2116 2117 2118 2119 2120
				     u64 objectid,
				     struct btrfs_block_group_cache *group,
				     int mode)
{
	struct inode *inode;
2121
	struct btrfs_inode_item *inode_item;
2122
	struct btrfs_block_group_cache *new_inode_group;
C
Chris Mason 已提交
2123
	struct btrfs_key *location;
2124
	struct btrfs_path *path;
2125 2126 2127 2128
	struct btrfs_inode_ref *ref;
	struct btrfs_key key[2];
	u32 sizes[2];
	unsigned long ptr;
C
Chris Mason 已提交
2129 2130 2131
	int ret;
	int owner;

2132 2133 2134
	path = btrfs_alloc_path();
	BUG_ON(!path);

C
Chris Mason 已提交
2135 2136 2137 2138
	inode = new_inode(root->fs_info->sb);
	if (!inode)
		return ERR_PTR(-ENOMEM);

2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152
	if (dir) {
		ret = btrfs_set_inode_index(dir, inode);
		if (ret)
			return ERR_PTR(ret);
	} else {
		BTRFS_I(inode)->index = 0;
	}
	/*
	 * index_cnt is ignored for everything but a dir,
	 * btrfs_get_inode_index_count has an explanation for the magic
	 * number
	 */
	BTRFS_I(inode)->index_cnt = 2;

2153 2154
	extent_map_tree_init(&BTRFS_I(inode)->extent_tree, GFP_NOFS);
	extent_io_tree_init(&BTRFS_I(inode)->io_tree,
2155
			     inode->i_mapping, GFP_NOFS);
2156 2157
	extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree,
			     inode->i_mapping, GFP_NOFS);
2158
	btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree);
2159
	mutex_init(&BTRFS_I(inode)->csum_mutex);
2160
	mutex_init(&BTRFS_I(inode)->extent_mutex);
2161
	BTRFS_I(inode)->delalloc_bytes = 0;
2162
	BTRFS_I(inode)->disk_i_size = 0;
C
Chris Mason 已提交
2163
	BTRFS_I(inode)->root = root;
2164

C
Chris Mason 已提交
2165 2166 2167 2168
	if (mode & S_IFDIR)
		owner = 0;
	else
		owner = 1;
2169
	new_inode_group = btrfs_find_block_group(root, group, 0,
2170
				       BTRFS_BLOCK_GROUP_METADATA, owner);
2171 2172 2173 2174 2175
	if (!new_inode_group) {
		printk("find_block group failed\n");
		new_inode_group = group;
	}
	BTRFS_I(inode)->block_group = new_inode_group;
Y
Yan 已提交
2176
	BTRFS_I(inode)->flags = 0;
2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190

	key[0].objectid = objectid;
	btrfs_set_key_type(&key[0], BTRFS_INODE_ITEM_KEY);
	key[0].offset = 0;

	key[1].objectid = objectid;
	btrfs_set_key_type(&key[1], BTRFS_INODE_REF_KEY);
	key[1].offset = ref_objectid;

	sizes[0] = sizeof(struct btrfs_inode_item);
	sizes[1] = name_len + sizeof(*ref);

	ret = btrfs_insert_empty_items(trans, root, path, key, sizes, 2);
	if (ret != 0)
2191 2192
		goto fail;

2193 2194 2195
	if (objectid > root->highest_inode)
		root->highest_inode = objectid;

C
Chris Mason 已提交
2196 2197 2198 2199 2200 2201
	inode->i_uid = current->fsuid;
	inode->i_gid = current->fsgid;
	inode->i_mode = mode;
	inode->i_ino = objectid;
	inode->i_blocks = 0;
	inode->i_mtime = inode->i_atime = inode->i_ctime = CURRENT_TIME;
2202 2203 2204
	inode_item = btrfs_item_ptr(path->nodes[0], path->slots[0],
				  struct btrfs_inode_item);
	fill_inode_item(path->nodes[0], inode_item, inode);
2205 2206 2207 2208

	ref = btrfs_item_ptr(path->nodes[0], path->slots[0] + 1,
			     struct btrfs_inode_ref);
	btrfs_set_inode_ref_name_len(path->nodes[0], ref, name_len);
2209
	btrfs_set_inode_ref_index(path->nodes[0], ref, BTRFS_I(inode)->index);
2210 2211 2212
	ptr = (unsigned long)(ref + 1);
	write_extent_buffer(path->nodes[0], name, ptr, name_len);

2213 2214 2215
	btrfs_mark_buffer_dirty(path->nodes[0]);
	btrfs_free_path(path);

C
Chris Mason 已提交
2216 2217 2218 2219 2220 2221 2222
	location = &BTRFS_I(inode)->location;
	location->objectid = objectid;
	location->offset = 0;
	btrfs_set_key_type(location, BTRFS_INODE_ITEM_KEY);

	insert_inode_hash(inode);
	return inode;
2223
fail:
2224 2225
	if (dir)
		BTRFS_I(dir)->index_cnt--;
2226 2227
	btrfs_free_path(path);
	return ERR_PTR(ret);
C
Chris Mason 已提交
2228 2229 2230 2231 2232 2233 2234 2235
}

static inline u8 btrfs_inode_type(struct inode *inode)
{
	return btrfs_type_by_mode[(inode->i_mode & S_IFMT) >> S_SHIFT];
}

static int btrfs_add_link(struct btrfs_trans_handle *trans,
2236 2237
			    struct dentry *dentry, struct inode *inode,
			    int add_backref)
C
Chris Mason 已提交
2238 2239 2240 2241
{
	int ret;
	struct btrfs_key key;
	struct btrfs_root *root = BTRFS_I(dentry->d_parent->d_inode)->root;
2242
	struct inode *parent_inode = dentry->d_parent->d_inode;
2243

C
Chris Mason 已提交
2244 2245 2246 2247 2248 2249 2250
	key.objectid = inode->i_ino;
	btrfs_set_key_type(&key, BTRFS_INODE_ITEM_KEY);
	key.offset = 0;

	ret = btrfs_insert_dir_item(trans, root,
				    dentry->d_name.name, dentry->d_name.len,
				    dentry->d_parent->d_inode->i_ino,
2251 2252
				    &key, btrfs_inode_type(inode),
				    BTRFS_I(inode)->index);
C
Chris Mason 已提交
2253
	if (ret == 0) {
2254 2255 2256 2257 2258
		if (add_backref) {
			ret = btrfs_insert_inode_ref(trans, root,
					     dentry->d_name.name,
					     dentry->d_name.len,
					     inode->i_ino,
2259 2260
					     parent_inode->i_ino,
					     BTRFS_I(inode)->index);
2261
		}
2262 2263
		btrfs_i_size_write(parent_inode, parent_inode->i_size +
				   dentry->d_name.len * 2);
2264
		parent_inode->i_mtime = parent_inode->i_ctime = CURRENT_TIME;
C
Chris Mason 已提交
2265 2266 2267 2268 2269 2270 2271
		ret = btrfs_update_inode(trans, root,
					 dentry->d_parent->d_inode);
	}
	return ret;
}

static int btrfs_add_nondir(struct btrfs_trans_handle *trans,
2272 2273
			    struct dentry *dentry, struct inode *inode,
			    int backref)
C
Chris Mason 已提交
2274
{
2275
	int err = btrfs_add_link(trans, dentry, inode, backref);
C
Chris Mason 已提交
2276 2277 2278 2279 2280 2281 2282 2283 2284
	if (!err) {
		d_instantiate(dentry, inode);
		return 0;
	}
	if (err > 0)
		err = -EEXIST;
	return err;
}

J
Josef Bacik 已提交
2285 2286 2287 2288 2289
static int btrfs_mknod(struct inode *dir, struct dentry *dentry,
			int mode, dev_t rdev)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_root *root = BTRFS_I(dir)->root;
2290
	struct inode *inode = NULL;
J
Josef Bacik 已提交
2291 2292 2293
	int err;
	int drop_inode = 0;
	u64 objectid;
2294
	unsigned long nr = 0;
J
Josef Bacik 已提交
2295 2296 2297 2298

	if (!new_valid_dev(rdev))
		return -EINVAL;

2299 2300 2301 2302
	err = btrfs_check_free_space(root, 1, 0);
	if (err)
		goto fail;

J
Josef Bacik 已提交
2303 2304 2305 2306 2307 2308 2309 2310 2311
	trans = btrfs_start_transaction(root, 1);
	btrfs_set_trans_block_group(trans, dir);

	err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
	if (err) {
		err = -ENOSPC;
		goto out_unlock;
	}

2312
	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
2313 2314
				dentry->d_name.len,
				dentry->d_parent->d_inode->i_ino, objectid,
J
Josef Bacik 已提交
2315 2316 2317 2318 2319
				BTRFS_I(dir)->block_group, mode);
	err = PTR_ERR(inode);
	if (IS_ERR(inode))
		goto out_unlock;

J
Josef Bacik 已提交
2320 2321 2322 2323 2324 2325
	err = btrfs_init_acl(inode, dir);
	if (err) {
		drop_inode = 1;
		goto out_unlock;
	}

J
Josef Bacik 已提交
2326
	btrfs_set_trans_block_group(trans, inode);
2327
	err = btrfs_add_nondir(trans, dentry, inode, 0);
J
Josef Bacik 已提交
2328 2329 2330 2331 2332
	if (err)
		drop_inode = 1;
	else {
		inode->i_op = &btrfs_special_inode_operations;
		init_special_inode(inode, inode->i_mode, rdev);
2333
		btrfs_update_inode(trans, root, inode);
J
Josef Bacik 已提交
2334 2335 2336 2337 2338
	}
	dir->i_sb->s_dirt = 1;
	btrfs_update_inode_block_group(trans, inode);
	btrfs_update_inode_block_group(trans, dir);
out_unlock:
2339
	nr = trans->blocks_used;
2340
	btrfs_end_transaction_throttle(trans, root);
2341
fail:
J
Josef Bacik 已提交
2342 2343 2344 2345
	if (drop_inode) {
		inode_dec_link_count(inode);
		iput(inode);
	}
2346
	btrfs_btree_balance_dirty(root, nr);
J
Josef Bacik 已提交
2347 2348 2349
	return err;
}

C
Chris Mason 已提交
2350 2351 2352 2353 2354
static int btrfs_create(struct inode *dir, struct dentry *dentry,
			int mode, struct nameidata *nd)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_root *root = BTRFS_I(dir)->root;
2355
	struct inode *inode = NULL;
C
Chris Mason 已提交
2356 2357
	int err;
	int drop_inode = 0;
2358
	unsigned long nr = 0;
C
Chris Mason 已提交
2359 2360
	u64 objectid;

2361 2362 2363
	err = btrfs_check_free_space(root, 1, 0);
	if (err)
		goto fail;
C
Chris Mason 已提交
2364 2365 2366 2367 2368 2369 2370 2371 2372
	trans = btrfs_start_transaction(root, 1);
	btrfs_set_trans_block_group(trans, dir);

	err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
	if (err) {
		err = -ENOSPC;
		goto out_unlock;
	}

2373
	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
2374 2375 2376
				dentry->d_name.len,
				dentry->d_parent->d_inode->i_ino,
				objectid, BTRFS_I(dir)->block_group, mode);
C
Chris Mason 已提交
2377 2378 2379 2380
	err = PTR_ERR(inode);
	if (IS_ERR(inode))
		goto out_unlock;

J
Josef Bacik 已提交
2381 2382 2383 2384 2385 2386
	err = btrfs_init_acl(inode, dir);
	if (err) {
		drop_inode = 1;
		goto out_unlock;
	}

C
Chris Mason 已提交
2387
	btrfs_set_trans_block_group(trans, inode);
2388
	err = btrfs_add_nondir(trans, dentry, inode, 0);
C
Chris Mason 已提交
2389 2390 2391 2392
	if (err)
		drop_inode = 1;
	else {
		inode->i_mapping->a_ops = &btrfs_aops;
2393
		inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
C
Chris Mason 已提交
2394 2395
		inode->i_fop = &btrfs_file_operations;
		inode->i_op = &btrfs_file_inode_operations;
2396 2397
		extent_map_tree_init(&BTRFS_I(inode)->extent_tree, GFP_NOFS);
		extent_io_tree_init(&BTRFS_I(inode)->io_tree,
2398
				     inode->i_mapping, GFP_NOFS);
2399 2400
		extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree,
				     inode->i_mapping, GFP_NOFS);
2401
		mutex_init(&BTRFS_I(inode)->csum_mutex);
2402
		mutex_init(&BTRFS_I(inode)->extent_mutex);
2403
		BTRFS_I(inode)->delalloc_bytes = 0;
2404
		BTRFS_I(inode)->disk_i_size = 0;
2405
		BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
2406
		btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree);
C
Chris Mason 已提交
2407 2408 2409 2410 2411
	}
	dir->i_sb->s_dirt = 1;
	btrfs_update_inode_block_group(trans, inode);
	btrfs_update_inode_block_group(trans, dir);
out_unlock:
2412
	nr = trans->blocks_used;
2413
	btrfs_end_transaction_throttle(trans, root);
2414
fail:
C
Chris Mason 已提交
2415 2416 2417 2418
	if (drop_inode) {
		inode_dec_link_count(inode);
		iput(inode);
	}
2419
	btrfs_btree_balance_dirty(root, nr);
C
Chris Mason 已提交
2420 2421 2422 2423 2424 2425 2426 2427 2428
	return err;
}

static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
		      struct dentry *dentry)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_root *root = BTRFS_I(dir)->root;
	struct inode *inode = old_dentry->d_inode;
2429
	unsigned long nr = 0;
C
Chris Mason 已提交
2430 2431 2432 2433 2434 2435
	int err;
	int drop_inode = 0;

	if (inode->i_nlink == 0)
		return -ENOENT;

2436 2437 2438
#if LINUX_VERSION_CODE <= KERNEL_VERSION(2,6,18)
	inode->i_nlink++;
#else
C
Chris Mason 已提交
2439
	inc_nlink(inode);
2440
#endif
2441 2442 2443
	err = btrfs_check_free_space(root, 1, 0);
	if (err)
		goto fail;
2444 2445 2446 2447
	err = btrfs_set_inode_index(dir, inode);
	if (err)
		goto fail;

C
Chris Mason 已提交
2448
	trans = btrfs_start_transaction(root, 1);
2449

C
Chris Mason 已提交
2450 2451
	btrfs_set_trans_block_group(trans, dir);
	atomic_inc(&inode->i_count);
2452

2453
	err = btrfs_add_nondir(trans, dentry, inode, 1);
2454

C
Chris Mason 已提交
2455 2456
	if (err)
		drop_inode = 1;
2457

C
Chris Mason 已提交
2458 2459
	dir->i_sb->s_dirt = 1;
	btrfs_update_inode_block_group(trans, dir);
2460
	err = btrfs_update_inode(trans, root, inode);
2461

2462 2463
	if (err)
		drop_inode = 1;
C
Chris Mason 已提交
2464

2465
	nr = trans->blocks_used;
2466
	btrfs_end_transaction_throttle(trans, root);
2467
fail:
C
Chris Mason 已提交
2468 2469 2470 2471
	if (drop_inode) {
		inode_dec_link_count(inode);
		iput(inode);
	}
2472
	btrfs_btree_balance_dirty(root, nr);
C
Chris Mason 已提交
2473 2474 2475 2476 2477
	return err;
}

static int btrfs_mkdir(struct inode *dir, struct dentry *dentry, int mode)
{
2478
	struct inode *inode = NULL;
C
Chris Mason 已提交
2479 2480 2481 2482
	struct btrfs_trans_handle *trans;
	struct btrfs_root *root = BTRFS_I(dir)->root;
	int err = 0;
	int drop_on_err = 0;
2483
	u64 objectid = 0;
2484
	unsigned long nr = 1;
C
Chris Mason 已提交
2485

2486 2487 2488 2489
	err = btrfs_check_free_space(root, 1, 0);
	if (err)
		goto out_unlock;

C
Chris Mason 已提交
2490 2491
	trans = btrfs_start_transaction(root, 1);
	btrfs_set_trans_block_group(trans, dir);
2492

C
Chris Mason 已提交
2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503
	if (IS_ERR(trans)) {
		err = PTR_ERR(trans);
		goto out_unlock;
	}

	err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
	if (err) {
		err = -ENOSPC;
		goto out_unlock;
	}

2504
	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
2505 2506
				dentry->d_name.len,
				dentry->d_parent->d_inode->i_ino, objectid,
C
Chris Mason 已提交
2507 2508 2509 2510 2511
				BTRFS_I(dir)->block_group, S_IFDIR | mode);
	if (IS_ERR(inode)) {
		err = PTR_ERR(inode);
		goto out_fail;
	}
2512

C
Chris Mason 已提交
2513
	drop_on_err = 1;
J
Josef Bacik 已提交
2514 2515 2516 2517 2518

	err = btrfs_init_acl(inode, dir);
	if (err)
		goto out_fail;

C
Chris Mason 已提交
2519 2520 2521 2522
	inode->i_op = &btrfs_dir_inode_operations;
	inode->i_fop = &btrfs_dir_file_operations;
	btrfs_set_trans_block_group(trans, inode);

2523
	btrfs_i_size_write(inode, 0);
C
Chris Mason 已提交
2524 2525 2526
	err = btrfs_update_inode(trans, root, inode);
	if (err)
		goto out_fail;
2527

2528
	err = btrfs_add_link(trans, dentry, inode, 0);
C
Chris Mason 已提交
2529 2530
	if (err)
		goto out_fail;
2531

C
Chris Mason 已提交
2532 2533 2534 2535 2536 2537 2538
	d_instantiate(dentry, inode);
	drop_on_err = 0;
	dir->i_sb->s_dirt = 1;
	btrfs_update_inode_block_group(trans, inode);
	btrfs_update_inode_block_group(trans, dir);

out_fail:
2539
	nr = trans->blocks_used;
2540
	btrfs_end_transaction_throttle(trans, root);
2541

C
Chris Mason 已提交
2542 2543 2544
out_unlock:
	if (drop_on_err)
		iput(inode);
2545
	btrfs_btree_balance_dirty(root, nr);
C
Chris Mason 已提交
2546 2547 2548
	return err;
}

2549 2550
static int merge_extent_mapping(struct extent_map_tree *em_tree,
				struct extent_map *existing,
2551 2552
				struct extent_map *em,
				u64 map_start, u64 map_len)
2553 2554 2555
{
	u64 start_diff;

2556 2557 2558 2559 2560 2561 2562
	BUG_ON(map_start < em->start || map_start >= extent_map_end(em));
	start_diff = map_start - em->start;
	em->start = map_start;
	em->len = map_len;
	if (em->block_start < EXTENT_MAP_LAST_BYTE)
		em->block_start += start_diff;
	return add_extent_mapping(em_tree, em);
2563 2564
}

2565
struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page,
2566
				    size_t pg_offset, u64 start, u64 len,
2567 2568 2569 2570
				    int create)
{
	int ret;
	int err = 0;
2571
	u64 bytenr;
2572 2573 2574 2575
	u64 extent_start = 0;
	u64 extent_end = 0;
	u64 objectid = inode->i_ino;
	u32 found_type;
2576
	struct btrfs_path *path = NULL;
2577 2578
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_file_extent_item *item;
2579 2580
	struct extent_buffer *leaf;
	struct btrfs_key found_key;
2581 2582
	struct extent_map *em = NULL;
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
2583
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
2584 2585 2586
	struct btrfs_trans_handle *trans = NULL;

again:
2587 2588
	spin_lock(&em_tree->lock);
	em = lookup_extent_mapping(em_tree, start, len);
2589 2590
	if (em)
		em->bdev = root->fs_info->fs_devices->latest_bdev;
2591 2592
	spin_unlock(&em_tree->lock);

2593
	if (em) {
2594 2595 2596
		if (em->start > start || em->start + em->len <= start)
			free_extent_map(em);
		else if (em->block_start == EXTENT_MAP_INLINE && page)
2597 2598 2599
			free_extent_map(em);
		else
			goto out;
2600
	}
2601
	em = alloc_extent_map(GFP_NOFS);
2602
	if (!em) {
2603 2604
		err = -ENOMEM;
		goto out;
2605
	}
2606
	em->bdev = root->fs_info->fs_devices->latest_bdev;
2607 2608
	em->start = EXTENT_MAP_HOLE;
	em->len = (u64)-1;
2609 2610 2611 2612 2613 2614

	if (!path) {
		path = btrfs_alloc_path();
		BUG_ON(!path);
	}

2615 2616
	ret = btrfs_lookup_file_extent(trans, root, path,
				       objectid, start, trans != NULL);
2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627
	if (ret < 0) {
		err = ret;
		goto out;
	}

	if (ret != 0) {
		if (path->slots[0] == 0)
			goto not_found;
		path->slots[0]--;
	}

2628 2629
	leaf = path->nodes[0];
	item = btrfs_item_ptr(leaf, path->slots[0],
2630 2631
			      struct btrfs_file_extent_item);
	/* are we inside the extent that was found? */
2632 2633 2634
	btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
	found_type = btrfs_key_type(&found_key);
	if (found_key.objectid != objectid ||
2635 2636 2637 2638
	    found_type != BTRFS_EXTENT_DATA_KEY) {
		goto not_found;
	}

2639 2640
	found_type = btrfs_file_extent_type(leaf, item);
	extent_start = found_key.offset;
2641 2642
	if (found_type == BTRFS_FILE_EXTENT_REG) {
		extent_end = extent_start +
2643
		       btrfs_file_extent_num_bytes(leaf, item);
2644
		err = 0;
2645
		if (start < extent_start || start >= extent_end) {
2646 2647
			em->start = start;
			if (start < extent_start) {
2648
				if (start + len <= extent_start)
2649
					goto not_found;
2650
				em->len = extent_end - extent_start;
2651
			} else {
2652
				em->len = len;
2653 2654 2655
			}
			goto not_found_em;
		}
2656 2657
		bytenr = btrfs_file_extent_disk_bytenr(leaf, item);
		if (bytenr == 0) {
2658
			em->start = extent_start;
2659
			em->len = extent_end - extent_start;
2660
			em->block_start = EXTENT_MAP_HOLE;
2661 2662
			goto insert;
		}
2663 2664
		bytenr += btrfs_file_extent_offset(leaf, item);
		em->block_start = bytenr;
2665
		em->start = extent_start;
2666
		em->len = extent_end - extent_start;
2667 2668
		goto insert;
	} else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
2669
		u64 page_start;
2670
		unsigned long ptr;
2671
		char *map;
2672 2673 2674
		size_t size;
		size_t extent_offset;
		size_t copy_size;
2675

2676 2677
		size = btrfs_file_extent_inline_len(leaf, btrfs_item_nr(leaf,
						    path->slots[0]));
2678 2679
		extent_end = (extent_start + size + root->sectorsize - 1) &
			~((u64)root->sectorsize - 1);
2680
		if (start < extent_start || start >= extent_end) {
2681 2682
			em->start = start;
			if (start < extent_start) {
2683
				if (start + len <= extent_start)
2684
					goto not_found;
2685
				em->len = extent_end - extent_start;
2686
			} else {
2687
				em->len = len;
2688 2689 2690
			}
			goto not_found_em;
		}
2691 2692 2693 2694
		em->block_start = EXTENT_MAP_INLINE;

		if (!page) {
			em->start = extent_start;
2695
			em->len = size;
2696 2697
			goto out;
		}
2698

2699 2700 2701
		page_start = page_offset(page) + pg_offset;
		extent_offset = page_start - extent_start;
		copy_size = min_t(u64, PAGE_CACHE_SIZE - pg_offset,
2702 2703
				size - extent_offset);
		em->start = extent_start + extent_offset;
2704 2705
		em->len = (copy_size + root->sectorsize - 1) &
			~((u64)root->sectorsize - 1);
2706 2707
		map = kmap(page);
		ptr = btrfs_file_extent_inline_start(item) + extent_offset;
2708
		if (create == 0 && !PageUptodate(page)) {
2709
			read_extent_buffer(leaf, map + pg_offset, ptr,
2710 2711 2712 2713 2714 2715 2716 2717
					   copy_size);
			flush_dcache_page(page);
		} else if (create && PageUptodate(page)) {
			if (!trans) {
				kunmap(page);
				free_extent_map(em);
				em = NULL;
				btrfs_release_path(root, path);
2718
				trans = btrfs_join_transaction(root, 1);
2719 2720
				goto again;
			}
2721
			write_extent_buffer(leaf, map + pg_offset, ptr,
2722 2723
					    copy_size);
			btrfs_mark_buffer_dirty(leaf);
2724 2725
		}
		kunmap(page);
2726 2727
		set_extent_uptodate(io_tree, em->start,
				    extent_map_end(em) - 1, GFP_NOFS);
2728 2729 2730 2731 2732 2733 2734
		goto insert;
	} else {
		printk("unkknown found_type %d\n", found_type);
		WARN_ON(1);
	}
not_found:
	em->start = start;
2735
	em->len = len;
2736
not_found_em:
2737
	em->block_start = EXTENT_MAP_HOLE;
2738 2739
insert:
	btrfs_release_path(root, path);
2740 2741
	if (em->start > start || extent_map_end(em) <= start) {
		printk("bad extent! em: [%Lu %Lu] passed [%Lu %Lu]\n", em->start, em->len, start, len);
2742 2743 2744
		err = -EIO;
		goto out;
	}
2745 2746 2747

	err = 0;
	spin_lock(&em_tree->lock);
2748
	ret = add_extent_mapping(em_tree, em);
2749 2750 2751 2752
	/* it is possible that someone inserted the extent into the tree
	 * while we had the lock dropped.  It is also possible that
	 * an overlapping map exists in the tree
	 */
2753
	if (ret == -EEXIST) {
2754
		struct extent_map *existing;
2755 2756 2757

		ret = 0;

2758
		existing = lookup_extent_mapping(em_tree, start, len);
2759 2760 2761 2762 2763
		if (existing && (existing->start > start ||
		    existing->start + existing->len <= start)) {
			free_extent_map(existing);
			existing = NULL;
		}
2764 2765 2766 2767 2768
		if (!existing) {
			existing = lookup_extent_mapping(em_tree, em->start,
							 em->len);
			if (existing) {
				err = merge_extent_mapping(em_tree, existing,
2769 2770
							   em, start,
							   root->sectorsize);
2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785
				free_extent_map(existing);
				if (err) {
					free_extent_map(em);
					em = NULL;
				}
			} else {
				err = -EIO;
				printk("failing to insert %Lu %Lu\n",
				       start, len);
				free_extent_map(em);
				em = NULL;
			}
		} else {
			free_extent_map(em);
			em = existing;
2786
			err = 0;
2787 2788
		}
	}
2789
	spin_unlock(&em_tree->lock);
2790
out:
2791 2792
	if (path)
		btrfs_free_path(path);
2793 2794
	if (trans) {
		ret = btrfs_end_transaction(trans, root);
2795
		if (!err) {
2796
			err = ret;
2797
		}
2798 2799 2800 2801 2802 2803 2804 2805 2806
	}
	if (err) {
		free_extent_map(em);
		WARN_ON(1);
		return ERR_PTR(err);
	}
	return em;
}

2807
#if 0 /* waiting for O_DIRECT reads */
2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824
static int btrfs_get_block(struct inode *inode, sector_t iblock,
			struct buffer_head *bh_result, int create)
{
	struct extent_map *em;
	u64 start = (u64)iblock << inode->i_blkbits;
	struct btrfs_multi_bio *multi = NULL;
	struct btrfs_root *root = BTRFS_I(inode)->root;
	u64 len;
	u64 logical;
	u64 map_length;
	int ret = 0;

	em = btrfs_get_extent(inode, NULL, 0, start, bh_result->b_size, 0);

	if (!em || IS_ERR(em))
		goto out;

2825
	if (em->start > start || em->start + em->len <= start) {
2826
	    goto out;
2827
	}
2828 2829 2830 2831 2832 2833

	if (em->block_start == EXTENT_MAP_INLINE) {
		ret = -EINVAL;
		goto out;
	}

2834 2835 2836
	len = em->start + em->len - start;
	len = min_t(u64, len, INT_LIMIT(typeof(bh_result->b_size)));

2837 2838
	if (em->block_start == EXTENT_MAP_HOLE ||
	    em->block_start == EXTENT_MAP_DELALLOC) {
2839
		bh_result->b_size = len;
2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851
		goto out;
	}

	logical = start - em->start;
	logical = em->block_start + logical;

	map_length = len;
	ret = btrfs_map_block(&root->fs_info->mapping_tree, READ,
			      logical, &map_length, &multi, 0);
	BUG_ON(ret);
	bh_result->b_blocknr = multi->stripes[0].physical >> inode->i_blkbits;
	bh_result->b_size = min(map_length, len);
2852

2853 2854 2855 2856 2857 2858 2859
	bh_result->b_bdev = multi->stripes[0].dev->bdev;
	set_buffer_mapped(bh_result);
	kfree(multi);
out:
	free_extent_map(em);
	return ret;
}
2860
#endif
2861 2862 2863 2864 2865

static ssize_t btrfs_direct_IO(int rw, struct kiocb *iocb,
			const struct iovec *iov, loff_t offset,
			unsigned long nr_segs)
{
2866 2867
	return -EINVAL;
#if 0
2868 2869 2870 2871 2872 2873 2874 2875
	struct file *file = iocb->ki_filp;
	struct inode *inode = file->f_mapping->host;

	if (rw == WRITE)
		return -EINVAL;

	return blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
				  offset, nr_segs, btrfs_get_block, NULL);
2876
#endif
2877 2878
}

2879
static sector_t btrfs_bmap(struct address_space *mapping, sector_t iblock)
C
Chris Mason 已提交
2880
{
2881
	return extent_bmap(mapping, iblock, btrfs_get_extent);
C
Chris Mason 已提交
2882 2883
}

2884
int btrfs_readpage(struct file *file, struct page *page)
2885
{
2886 2887
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
2888
	return extent_read_full_page(tree, page, btrfs_get_extent);
2889
}
2890

2891
static int btrfs_writepage(struct page *page, struct writeback_control *wbc)
C
Chris Mason 已提交
2892
{
2893
	struct extent_io_tree *tree;
2894 2895 2896 2897 2898 2899 2900


	if (current->flags & PF_MEMALLOC) {
		redirty_page_for_writepage(wbc, page);
		unlock_page(page);
		return 0;
	}
2901
	tree = &BTRFS_I(page->mapping->host)->io_tree;
2902
	return extent_write_full_page(tree, page, btrfs_get_extent, wbc);
2903 2904
}

2905 2906
int btrfs_writepages(struct address_space *mapping,
		     struct writeback_control *wbc)
2907
{
2908 2909
	struct extent_io_tree *tree;
	tree = &BTRFS_I(mapping->host)->io_tree;
2910 2911 2912
	return extent_writepages(tree, mapping, btrfs_get_extent, wbc);
}

2913 2914 2915 2916
static int
btrfs_readpages(struct file *file, struct address_space *mapping,
		struct list_head *pages, unsigned nr_pages)
{
2917 2918
	struct extent_io_tree *tree;
	tree = &BTRFS_I(mapping->host)->io_tree;
2919 2920 2921
	return extent_readpages(tree, mapping, pages, nr_pages,
				btrfs_get_extent);
}
2922
static int __btrfs_releasepage(struct page *page, gfp_t gfp_flags)
2923
{
2924 2925
	struct extent_io_tree *tree;
	struct extent_map_tree *map;
2926
	int ret;
2927

2928 2929
	tree = &BTRFS_I(page->mapping->host)->io_tree;
	map = &BTRFS_I(page->mapping->host)->extent_tree;
2930
	ret = try_release_extent_mapping(map, tree, page, gfp_flags);
2931 2932 2933 2934
	if (ret == 1) {
		ClearPagePrivate(page);
		set_page_private(page, 0);
		page_cache_release(page);
C
Chris Mason 已提交
2935
	}
2936
	return ret;
C
Chris Mason 已提交
2937 2938
}

2939 2940 2941 2942 2943
static int btrfs_releasepage(struct page *page, gfp_t gfp_flags)
{
	return __btrfs_releasepage(page, gfp_flags);
}

2944
static void btrfs_invalidatepage(struct page *page, unsigned long offset)
C
Chris Mason 已提交
2945
{
2946
	struct extent_io_tree *tree;
2947 2948 2949
	struct btrfs_ordered_extent *ordered;
	u64 page_start = page_offset(page);
	u64 page_end = page_start + PAGE_CACHE_SIZE - 1;
C
Chris Mason 已提交
2950

2951
	wait_on_page_writeback(page);
2952
	tree = &BTRFS_I(page->mapping->host)->io_tree;
2953 2954 2955 2956 2957 2958 2959 2960 2961
	if (offset) {
		btrfs_releasepage(page, GFP_NOFS);
		return;
	}

	lock_extent(tree, page_start, page_end, GFP_NOFS);
	ordered = btrfs_lookup_ordered_extent(page->mapping->host,
					   page_offset(page));
	if (ordered) {
2962 2963 2964 2965
		/*
		 * IO on this page will never be started, so we need
		 * to account for any ordered extents now
		 */
2966 2967 2968
		clear_extent_bit(tree, page_start, page_end,
				 EXTENT_DIRTY | EXTENT_DELALLOC |
				 EXTENT_LOCKED, 1, 0, GFP_NOFS);
2969 2970
		btrfs_finish_ordered_io(page->mapping->host,
					page_start, page_end);
2971 2972 2973 2974 2975 2976 2977 2978 2979
		btrfs_put_ordered_extent(ordered);
		lock_extent(tree, page_start, page_end, GFP_NOFS);
	}
	clear_extent_bit(tree, page_start, page_end,
		 EXTENT_LOCKED | EXTENT_DIRTY | EXTENT_DELALLOC |
		 EXTENT_ORDERED,
		 1, 1, GFP_NOFS);
	__btrfs_releasepage(page, GFP_NOFS);

C
Chris Mason 已提交
2980
	ClearPageChecked(page);
2981 2982 2983 2984 2985
	if (PagePrivate(page)) {
		ClearPagePrivate(page);
		set_page_private(page, 0);
		page_cache_release(page);
	}
C
Chris Mason 已提交
2986 2987
}

2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004
/*
 * btrfs_page_mkwrite() is not allowed to change the file size as it gets
 * called from a page fault handler when a page is first dirtied. Hence we must
 * be careful to check for EOF conditions here. We set the page up correctly
 * for a written page which means we get ENOSPC checking when writing into
 * holes and correct delalloc and unwritten extent mapping on filesystems that
 * support these features.
 *
 * We are not allowed to take the i_mutex here so we have to play games to
 * protect against truncate races as the page could now be beyond EOF.  Because
 * vmtruncate() writes the inode size before removing pages, once we have the
 * page lock we can determine safely if the page is beyond EOF. If it is not
 * beyond EOF, then the page is guaranteed safe against truncation until we
 * unlock the page.
 */
int btrfs_page_mkwrite(struct vm_area_struct *vma, struct page *page)
{
3005
	struct inode *inode = fdentry(vma->vm_file)->d_inode;
3006
	struct btrfs_root *root = BTRFS_I(inode)->root;
3007 3008 3009 3010
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	struct btrfs_ordered_extent *ordered;
	char *kaddr;
	unsigned long zero_start;
3011
	loff_t size;
3012
	int ret;
3013
	u64 page_start;
3014
	u64 page_end;
3015

3016 3017 3018 3019 3020
	ret = btrfs_check_free_space(root, PAGE_CACHE_SIZE, 0);
	if (ret)
		goto out;

	ret = -EINVAL;
3021
again:
3022 3023
	lock_page(page);
	size = i_size_read(inode);
3024 3025
	page_start = page_offset(page);
	page_end = page_start + PAGE_CACHE_SIZE - 1;
3026

3027
	if ((page->mapping != inode->i_mapping) ||
3028
	    (page_start >= size)) {
3029 3030 3031
		/* page got truncated out from underneath us */
		goto out_unlock;
	}
3032 3033 3034 3035 3036
	wait_on_page_writeback(page);

	lock_extent(io_tree, page_start, page_end, GFP_NOFS);
	set_page_extent_mapped(page);

3037 3038 3039 3040
	/*
	 * we can't set the delalloc bits if there are pending ordered
	 * extents.  Drop our locks and wait for them to finish
	 */
3041 3042 3043 3044
	ordered = btrfs_lookup_ordered_extent(inode, page_start);
	if (ordered) {
		unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
		unlock_page(page);
3045
		btrfs_start_ordered_extent(inode, ordered, 1);
3046 3047 3048 3049 3050 3051 3052
		btrfs_put_ordered_extent(ordered);
		goto again;
	}

	set_extent_delalloc(&BTRFS_I(inode)->io_tree, page_start,
			    page_end, GFP_NOFS);
	ret = 0;
3053 3054

	/* page is wholly or partially inside EOF */
3055
	if (page_start + PAGE_CACHE_SIZE > size)
3056
		zero_start = size & ~PAGE_CACHE_MASK;
3057
	else
3058
		zero_start = PAGE_CACHE_SIZE;
3059

3060 3061 3062 3063 3064 3065
	if (zero_start != PAGE_CACHE_SIZE) {
		kaddr = kmap(page);
		memset(kaddr + zero_start, 0, PAGE_CACHE_SIZE - zero_start);
		flush_dcache_page(page);
		kunmap(page);
	}
3066
	ClearPageChecked(page);
3067 3068
	set_page_dirty(page);
	unlock_extent(io_tree, page_start, page_end, GFP_NOFS);
3069 3070 3071

out_unlock:
	unlock_page(page);
3072
out:
3073 3074 3075
	return ret;
}

C
Chris Mason 已提交
3076 3077 3078 3079 3080
static void btrfs_truncate(struct inode *inode)
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int ret;
	struct btrfs_trans_handle *trans;
3081
	unsigned long nr;
3082
	u64 mask = root->sectorsize - 1;
C
Chris Mason 已提交
3083 3084 3085 3086 3087 3088 3089

	if (!S_ISREG(inode->i_mode))
		return;
	if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
		return;

	btrfs_truncate_page(inode->i_mapping, inode->i_size);
C
Chris Mason 已提交
3090
	btrfs_wait_ordered_range(inode, inode->i_size & (~mask), (u64)-1);
C
Chris Mason 已提交
3091 3092 3093

	trans = btrfs_start_transaction(root, 1);
	btrfs_set_trans_block_group(trans, inode);
3094
	btrfs_i_size_write(inode, inode->i_size);
C
Chris Mason 已提交
3095

3096 3097 3098
	ret = btrfs_orphan_add(trans, inode);
	if (ret)
		goto out;
C
Chris Mason 已提交
3099
	/* FIXME, add redo link to tree so we don't leak on crash */
3100 3101
	ret = btrfs_truncate_in_trans(trans, root, inode,
				      BTRFS_EXTENT_DATA_KEY);
C
Chris Mason 已提交
3102
	btrfs_update_inode(trans, root, inode);
3103

3104 3105 3106 3107 3108
	ret = btrfs_orphan_del(trans, inode);
	BUG_ON(ret);

out:
	nr = trans->blocks_used;
3109
	ret = btrfs_end_transaction_throttle(trans, root);
C
Chris Mason 已提交
3110
	BUG_ON(ret);
3111
	btrfs_btree_balance_dirty(root, nr);
C
Chris Mason 已提交
3112 3113
}

3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138
/*
 * Invalidate a single dcache entry at the root of the filesystem.
 * Needed after creation of snapshot or subvolume.
 */
void btrfs_invalidate_dcache_root(struct btrfs_root *root, char *name,
				  int namelen)
{
	struct dentry *alias, *entry;
	struct qstr qstr;

	alias = d_find_alias(root->fs_info->sb->s_root->d_inode);
	if (alias) {
		qstr.name = name;
		qstr.len = namelen;
		/* change me if btrfs ever gets a d_hash operation */
		qstr.hash = full_name_hash(qstr.name, qstr.len);
		entry = d_lookup(alias, &qstr);
		dput(alias);
		if (entry) {
			d_invalidate(entry);
			dput(entry);
		}
	}
}

3139 3140 3141
int btrfs_create_subvol_root(struct btrfs_root *new_root,
		struct btrfs_trans_handle *trans, u64 new_dirid,
		struct btrfs_block_group_cache *block_group)
C
Chris Mason 已提交
3142 3143 3144
{
	struct inode *inode;

3145
	inode = btrfs_new_inode(trans, new_root, NULL, "..", 2, new_dirid,
3146
				new_dirid, block_group, S_IFDIR | 0700);
3147
	if (IS_ERR(inode))
3148
		return PTR_ERR(inode);
C
Chris Mason 已提交
3149 3150
	inode->i_op = &btrfs_dir_inode_operations;
	inode->i_fop = &btrfs_dir_file_operations;
3151
	new_root->inode = inode;
C
Chris Mason 已提交
3152 3153

	inode->i_nlink = 1;
3154
	btrfs_i_size_write(inode, 0);
3155

3156
	return btrfs_update_inode(trans, new_root, inode);
C
Chris Mason 已提交
3157 3158
}

3159
unsigned long btrfs_force_ra(struct address_space *mapping,
3160 3161 3162
			      struct file_ra_state *ra, struct file *file,
			      pgoff_t offset, pgoff_t last_index)
{
3163
	pgoff_t req_size = last_index - offset + 1;
3164 3165 3166 3167 3168 3169 3170 3171 3172 3173

#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23)
	offset = page_cache_readahead(mapping, ra, file, offset, req_size);
	return offset;
#else
	page_cache_sync_readahead(mapping, ra, file, offset, req_size);
	return offset + req_size;
#endif
}

C
Chris Mason 已提交
3174 3175 3176 3177 3178 3179 3180
struct inode *btrfs_alloc_inode(struct super_block *sb)
{
	struct btrfs_inode *ei;

	ei = kmem_cache_alloc(btrfs_inode_cachep, GFP_NOFS);
	if (!ei)
		return NULL;
3181
	ei->last_trans = 0;
3182
	btrfs_ordered_inode_tree_init(&ei->ordered_tree);
J
Josef Bacik 已提交
3183 3184
	ei->i_acl = BTRFS_ACL_NOT_CACHED;
	ei->i_default_acl = BTRFS_ACL_NOT_CACHED;
3185
	INIT_LIST_HEAD(&ei->i_orphan);
C
Chris Mason 已提交
3186 3187 3188 3189 3190
	return &ei->vfs_inode;
}

void btrfs_destroy_inode(struct inode *inode)
{
3191
	struct btrfs_ordered_extent *ordered;
C
Chris Mason 已提交
3192 3193 3194
	WARN_ON(!list_empty(&inode->i_dentry));
	WARN_ON(inode->i_data.nrpages);

J
Josef Bacik 已提交
3195 3196 3197 3198 3199 3200 3201
	if (BTRFS_I(inode)->i_acl &&
	    BTRFS_I(inode)->i_acl != BTRFS_ACL_NOT_CACHED)
		posix_acl_release(BTRFS_I(inode)->i_acl);
	if (BTRFS_I(inode)->i_default_acl &&
	    BTRFS_I(inode)->i_default_acl != BTRFS_ACL_NOT_CACHED)
		posix_acl_release(BTRFS_I(inode)->i_default_acl);

3202
	spin_lock(&BTRFS_I(inode)->root->list_lock);
3203 3204 3205 3206 3207
	if (!list_empty(&BTRFS_I(inode)->i_orphan)) {
		printk(KERN_ERR "BTRFS: inode %lu: inode still on the orphan"
		       " list\n", inode->i_ino);
		dump_stack();
	}
3208
	spin_unlock(&BTRFS_I(inode)->root->list_lock);
3209

3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221
	while(1) {
		ordered = btrfs_lookup_first_ordered_extent(inode, (u64)-1);
		if (!ordered)
			break;
		else {
			printk("found ordered extent %Lu %Lu\n",
			       ordered->file_offset, ordered->len);
			btrfs_remove_ordered_extent(inode, ordered);
			btrfs_put_ordered_extent(ordered);
			btrfs_put_ordered_extent(ordered);
		}
	}
3222
	btrfs_drop_extent_cache(inode, 0, (u64)-1);
C
Chris Mason 已提交
3223 3224 3225
	kmem_cache_free(btrfs_inode_cachep, BTRFS_I(inode));
}

3226 3227 3228
#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,26)
static void init_once(void *foo)
#elif LINUX_VERSION_CODE > KERNEL_VERSION(2,6,23)
3229 3230
static void init_once(struct kmem_cache * cachep, void *foo)
#else
C
Chris Mason 已提交
3231 3232
static void init_once(void * foo, struct kmem_cache * cachep,
		      unsigned long flags)
3233
#endif
C
Chris Mason 已提交
3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253
{
	struct btrfs_inode *ei = (struct btrfs_inode *) foo;

	inode_init_once(&ei->vfs_inode);
}

void btrfs_destroy_cachep(void)
{
	if (btrfs_inode_cachep)
		kmem_cache_destroy(btrfs_inode_cachep);
	if (btrfs_trans_handle_cachep)
		kmem_cache_destroy(btrfs_trans_handle_cachep);
	if (btrfs_transaction_cachep)
		kmem_cache_destroy(btrfs_transaction_cachep);
	if (btrfs_bit_radix_cachep)
		kmem_cache_destroy(btrfs_bit_radix_cachep);
	if (btrfs_path_cachep)
		kmem_cache_destroy(btrfs_path_cachep);
}

3254
struct kmem_cache *btrfs_cache_create(const char *name, size_t size,
3255
				       unsigned long extra_flags,
3256 3257 3258
#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,26)
				       void (*ctor)(void *)
#elif LINUX_VERSION_CODE > KERNEL_VERSION(2,6,23)
3259 3260
				       void (*ctor)(struct kmem_cache *, void *)
#else
3261
				       void (*ctor)(void *, struct kmem_cache *,
3262 3263 3264
						    unsigned long)
#endif
				     )
3265 3266 3267 3268 3269 3270 3271 3272 3273
{
	return kmem_cache_create(name, size, 0, (SLAB_RECLAIM_ACCOUNT |
				 SLAB_MEM_SPREAD | extra_flags), ctor
#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23)
				 ,NULL
#endif
				);
}

C
Chris Mason 已提交
3274 3275
int btrfs_init_cachep(void)
{
3276
	btrfs_inode_cachep = btrfs_cache_create("btrfs_inode_cache",
3277 3278
					  sizeof(struct btrfs_inode),
					  0, init_once);
C
Chris Mason 已提交
3279 3280
	if (!btrfs_inode_cachep)
		goto fail;
3281 3282 3283 3284
	btrfs_trans_handle_cachep =
			btrfs_cache_create("btrfs_trans_handle_cache",
					   sizeof(struct btrfs_trans_handle),
					   0, NULL);
C
Chris Mason 已提交
3285 3286
	if (!btrfs_trans_handle_cachep)
		goto fail;
3287
	btrfs_transaction_cachep = btrfs_cache_create("btrfs_transaction_cache",
C
Chris Mason 已提交
3288
					     sizeof(struct btrfs_transaction),
3289
					     0, NULL);
C
Chris Mason 已提交
3290 3291
	if (!btrfs_transaction_cachep)
		goto fail;
3292
	btrfs_path_cachep = btrfs_cache_create("btrfs_path_cache",
3293
					 sizeof(struct btrfs_path),
3294
					 0, NULL);
C
Chris Mason 已提交
3295 3296
	if (!btrfs_path_cachep)
		goto fail;
3297
	btrfs_bit_radix_cachep = btrfs_cache_create("btrfs_radix", 256,
3298
					      SLAB_DESTROY_BY_RCU, NULL);
C
Chris Mason 已提交
3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311
	if (!btrfs_bit_radix_cachep)
		goto fail;
	return 0;
fail:
	btrfs_destroy_cachep();
	return -ENOMEM;
}

static int btrfs_getattr(struct vfsmount *mnt,
			 struct dentry *dentry, struct kstat *stat)
{
	struct inode *inode = dentry->d_inode;
	generic_fillattr(inode, stat);
3312
	stat->blksize = PAGE_CACHE_SIZE;
3313
	stat->blocks = inode->i_blocks + (BTRFS_I(inode)->delalloc_bytes >> 9);
C
Chris Mason 已提交
3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330
	return 0;
}

static int btrfs_rename(struct inode * old_dir, struct dentry *old_dentry,
			   struct inode * new_dir,struct dentry *new_dentry)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_root *root = BTRFS_I(old_dir)->root;
	struct inode *new_inode = new_dentry->d_inode;
	struct inode *old_inode = old_dentry->d_inode;
	struct timespec ctime = CURRENT_TIME;
	int ret;

	if (S_ISDIR(old_inode->i_mode) && new_inode &&
	    new_inode->i_size > BTRFS_EMPTY_DIR_SIZE) {
		return -ENOTEMPTY;
	}
3331

3332 3333 3334 3335
	ret = btrfs_check_free_space(root, 1, 0);
	if (ret)
		goto out_unlock;

C
Chris Mason 已提交
3336
	trans = btrfs_start_transaction(root, 1);
3337

C
Chris Mason 已提交
3338 3339 3340 3341 3342 3343
	btrfs_set_trans_block_group(trans, new_dir);

	old_dentry->d_inode->i_nlink++;
	old_dir->i_ctime = old_dir->i_mtime = ctime;
	new_dir->i_ctime = new_dir->i_mtime = ctime;
	old_inode->i_ctime = ctime;
3344

C
Chris Mason 已提交
3345 3346 3347 3348 3349 3350 3351 3352 3353
	ret = btrfs_unlink_trans(trans, root, old_dir, old_dentry);
	if (ret)
		goto out_fail;

	if (new_inode) {
		new_inode->i_ctime = CURRENT_TIME;
		ret = btrfs_unlink_trans(trans, root, new_dir, new_dentry);
		if (ret)
			goto out_fail;
3354 3355 3356 3357 3358
		if (new_inode->i_nlink == 0) {
			ret = btrfs_orphan_add(trans, new_inode);
			if (ret)
				goto out_fail;
		}
C
Chris Mason 已提交
3359
	}
3360 3361 3362 3363
	ret = btrfs_set_inode_index(new_dir, old_inode);
	if (ret)
		goto out_fail;

3364
	ret = btrfs_add_link(trans, new_dentry, old_inode, 1);
C
Chris Mason 已提交
3365 3366 3367 3368
	if (ret)
		goto out_fail;

out_fail:
3369
	btrfs_end_transaction_throttle(trans, root);
3370
out_unlock:
C
Chris Mason 已提交
3371 3372 3373 3374 3375 3376 3377 3378 3379 3380
	return ret;
}

static int btrfs_symlink(struct inode *dir, struct dentry *dentry,
			 const char *symname)
{
	struct btrfs_trans_handle *trans;
	struct btrfs_root *root = BTRFS_I(dir)->root;
	struct btrfs_path *path;
	struct btrfs_key key;
3381
	struct inode *inode = NULL;
C
Chris Mason 已提交
3382 3383 3384 3385 3386
	int err;
	int drop_inode = 0;
	u64 objectid;
	int name_len;
	int datasize;
3387
	unsigned long ptr;
C
Chris Mason 已提交
3388
	struct btrfs_file_extent_item *ei;
3389
	struct extent_buffer *leaf;
3390
	unsigned long nr = 0;
C
Chris Mason 已提交
3391 3392 3393 3394

	name_len = strlen(symname) + 1;
	if (name_len > BTRFS_MAX_INLINE_DATA_SIZE(root))
		return -ENAMETOOLONG;
3395 3396 3397 3398 3399

	err = btrfs_check_free_space(root, 1, 0);
	if (err)
		goto out_fail;

C
Chris Mason 已提交
3400 3401 3402 3403 3404 3405 3406 3407 3408
	trans = btrfs_start_transaction(root, 1);
	btrfs_set_trans_block_group(trans, dir);

	err = btrfs_find_free_objectid(trans, root, dir->i_ino, &objectid);
	if (err) {
		err = -ENOSPC;
		goto out_unlock;
	}

3409
	inode = btrfs_new_inode(trans, root, dir, dentry->d_name.name,
3410 3411
				dentry->d_name.len,
				dentry->d_parent->d_inode->i_ino, objectid,
C
Chris Mason 已提交
3412 3413 3414 3415 3416
				BTRFS_I(dir)->block_group, S_IFLNK|S_IRWXUGO);
	err = PTR_ERR(inode);
	if (IS_ERR(inode))
		goto out_unlock;

J
Josef Bacik 已提交
3417 3418 3419 3420 3421 3422
	err = btrfs_init_acl(inode, dir);
	if (err) {
		drop_inode = 1;
		goto out_unlock;
	}

C
Chris Mason 已提交
3423
	btrfs_set_trans_block_group(trans, inode);
3424
	err = btrfs_add_nondir(trans, dentry, inode, 0);
C
Chris Mason 已提交
3425 3426 3427 3428
	if (err)
		drop_inode = 1;
	else {
		inode->i_mapping->a_ops = &btrfs_aops;
3429
		inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
C
Chris Mason 已提交
3430 3431
		inode->i_fop = &btrfs_file_operations;
		inode->i_op = &btrfs_file_inode_operations;
3432 3433
		extent_map_tree_init(&BTRFS_I(inode)->extent_tree, GFP_NOFS);
		extent_io_tree_init(&BTRFS_I(inode)->io_tree,
3434
				     inode->i_mapping, GFP_NOFS);
3435 3436
		extent_io_tree_init(&BTRFS_I(inode)->io_failure_tree,
				     inode->i_mapping, GFP_NOFS);
3437
		mutex_init(&BTRFS_I(inode)->csum_mutex);
3438
		mutex_init(&BTRFS_I(inode)->extent_mutex);
3439
		BTRFS_I(inode)->delalloc_bytes = 0;
3440
		BTRFS_I(inode)->disk_i_size = 0;
3441
		BTRFS_I(inode)->io_tree.ops = &btrfs_extent_io_ops;
3442
		btrfs_ordered_inode_tree_init(&BTRFS_I(inode)->ordered_tree);
C
Chris Mason 已提交
3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457
	}
	dir->i_sb->s_dirt = 1;
	btrfs_update_inode_block_group(trans, inode);
	btrfs_update_inode_block_group(trans, dir);
	if (drop_inode)
		goto out_unlock;

	path = btrfs_alloc_path();
	BUG_ON(!path);
	key.objectid = inode->i_ino;
	key.offset = 0;
	btrfs_set_key_type(&key, BTRFS_EXTENT_DATA_KEY);
	datasize = btrfs_file_extent_calc_inline_size(name_len);
	err = btrfs_insert_empty_item(trans, root, path, &key,
				      datasize);
3458 3459 3460 3461
	if (err) {
		drop_inode = 1;
		goto out_unlock;
	}
3462 3463 3464 3465 3466
	leaf = path->nodes[0];
	ei = btrfs_item_ptr(leaf, path->slots[0],
			    struct btrfs_file_extent_item);
	btrfs_set_file_extent_generation(leaf, ei, trans->transid);
	btrfs_set_file_extent_type(leaf, ei,
C
Chris Mason 已提交
3467 3468
				   BTRFS_FILE_EXTENT_INLINE);
	ptr = btrfs_file_extent_inline_start(ei);
3469 3470
	write_extent_buffer(leaf, symname, ptr, name_len);
	btrfs_mark_buffer_dirty(leaf);
C
Chris Mason 已提交
3471
	btrfs_free_path(path);
3472

C
Chris Mason 已提交
3473 3474
	inode->i_op = &btrfs_symlink_inode_operations;
	inode->i_mapping->a_ops = &btrfs_symlink_aops;
3475
	inode->i_mapping->backing_dev_info = &root->fs_info->bdi;
3476
	btrfs_i_size_write(inode, name_len - 1);
3477 3478 3479
	err = btrfs_update_inode(trans, root, inode);
	if (err)
		drop_inode = 1;
C
Chris Mason 已提交
3480 3481

out_unlock:
3482
	nr = trans->blocks_used;
3483
	btrfs_end_transaction_throttle(trans, root);
3484
out_fail:
C
Chris Mason 已提交
3485 3486 3487 3488
	if (drop_inode) {
		inode_dec_link_count(inode);
		iput(inode);
	}
3489
	btrfs_btree_balance_dirty(root, nr);
C
Chris Mason 已提交
3490 3491
	return err;
}
3492

3493 3494 3495 3496 3497
static int btrfs_set_page_dirty(struct page *page)
{
	return __set_page_dirty_nobuffers(page);
}

3498 3499 3500
#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,26)
static int btrfs_permission(struct inode *inode, int mask)
#else
Y
Yan 已提交
3501 3502
static int btrfs_permission(struct inode *inode, int mask,
			    struct nameidata *nd)
3503
#endif
Y
Yan 已提交
3504 3505 3506
{
	if (btrfs_test_flag(inode, READONLY) && (mask & MAY_WRITE))
		return -EACCES;
J
Josef Bacik 已提交
3507
	return generic_permission(inode, mask, btrfs_check_acl);
Y
Yan 已提交
3508
}
C
Chris Mason 已提交
3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519

static struct inode_operations btrfs_dir_inode_operations = {
	.lookup		= btrfs_lookup,
	.create		= btrfs_create,
	.unlink		= btrfs_unlink,
	.link		= btrfs_link,
	.mkdir		= btrfs_mkdir,
	.rmdir		= btrfs_rmdir,
	.rename		= btrfs_rename,
	.symlink	= btrfs_symlink,
	.setattr	= btrfs_setattr,
J
Josef Bacik 已提交
3520
	.mknod		= btrfs_mknod,
J
Josef Bacik 已提交
3521 3522 3523 3524
	.setxattr	= generic_setxattr,
	.getxattr	= generic_getxattr,
	.listxattr	= btrfs_listxattr,
	.removexattr	= generic_removexattr,
Y
Yan 已提交
3525
	.permission	= btrfs_permission,
C
Chris Mason 已提交
3526 3527 3528
};
static struct inode_operations btrfs_dir_ro_inode_operations = {
	.lookup		= btrfs_lookup,
Y
Yan 已提交
3529
	.permission	= btrfs_permission,
C
Chris Mason 已提交
3530 3531 3532 3533 3534
};
static struct file_operations btrfs_dir_file_operations = {
	.llseek		= generic_file_llseek,
	.read		= generic_read_dir,
	.readdir	= btrfs_readdir,
3535
	.unlocked_ioctl	= btrfs_ioctl,
C
Chris Mason 已提交
3536
#ifdef CONFIG_COMPAT
3537
	.compat_ioctl	= btrfs_ioctl,
C
Chris Mason 已提交
3538
#endif
S
Sage Weil 已提交
3539
	.release        = btrfs_release_file,
C
Chris Mason 已提交
3540 3541
};

3542
static struct extent_io_ops btrfs_extent_io_ops = {
3543
	.fill_delalloc = run_delalloc_range,
3544
	.submit_bio_hook = btrfs_submit_bio_hook,
3545
	.merge_bio_hook = btrfs_merge_bio_hook,
3546
	.readpage_end_io_hook = btrfs_readpage_end_io_hook,
3547
	.writepage_end_io_hook = btrfs_writepage_end_io_hook,
3548
	.writepage_start_hook = btrfs_writepage_start_hook,
3549
	.readpage_io_failed_hook = btrfs_io_failed_hook,
3550 3551
	.set_bit_hook = btrfs_set_bit_hook,
	.clear_bit_hook = btrfs_clear_bit_hook,
3552 3553
};

C
Chris Mason 已提交
3554 3555 3556
static struct address_space_operations btrfs_aops = {
	.readpage	= btrfs_readpage,
	.writepage	= btrfs_writepage,
3557
	.writepages	= btrfs_writepages,
3558
	.readpages	= btrfs_readpages,
C
Chris Mason 已提交
3559 3560
	.sync_page	= block_sync_page,
	.bmap		= btrfs_bmap,
3561
	.direct_IO	= btrfs_direct_IO,
3562 3563
	.invalidatepage = btrfs_invalidatepage,
	.releasepage	= btrfs_releasepage,
3564
	.set_page_dirty	= btrfs_set_page_dirty,
C
Chris Mason 已提交
3565 3566 3567 3568 3569
};

static struct address_space_operations btrfs_symlink_aops = {
	.readpage	= btrfs_readpage,
	.writepage	= btrfs_writepage,
C
Chris Mason 已提交
3570 3571
	.invalidatepage = btrfs_invalidatepage,
	.releasepage	= btrfs_releasepage,
C
Chris Mason 已提交
3572 3573 3574 3575 3576 3577
};

static struct inode_operations btrfs_file_inode_operations = {
	.truncate	= btrfs_truncate,
	.getattr	= btrfs_getattr,
	.setattr	= btrfs_setattr,
J
Josef Bacik 已提交
3578 3579 3580 3581
	.setxattr	= generic_setxattr,
	.getxattr	= generic_getxattr,
	.listxattr      = btrfs_listxattr,
	.removexattr	= generic_removexattr,
Y
Yan 已提交
3582
	.permission	= btrfs_permission,
C
Chris Mason 已提交
3583
};
J
Josef Bacik 已提交
3584 3585 3586
static struct inode_operations btrfs_special_inode_operations = {
	.getattr	= btrfs_getattr,
	.setattr	= btrfs_setattr,
Y
Yan 已提交
3587
	.permission	= btrfs_permission,
J
Josef Bacik 已提交
3588 3589 3590 3591
	.setxattr	= generic_setxattr,
	.getxattr	= generic_getxattr,
	.listxattr	= btrfs_listxattr,
	.removexattr	= generic_removexattr,
J
Josef Bacik 已提交
3592
};
C
Chris Mason 已提交
3593 3594 3595 3596
static struct inode_operations btrfs_symlink_inode_operations = {
	.readlink	= generic_readlink,
	.follow_link	= page_follow_link_light,
	.put_link	= page_put_link,
Y
Yan 已提交
3597
	.permission	= btrfs_permission,
C
Chris Mason 已提交
3598
};
新手
引导
客服 返回
顶部