file.c 37.3 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.
 */

C
Chris Mason 已提交
19 20 21 22 23 24 25 26
#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/backing-dev.h>
#include <linux/mpage.h>
27
#include <linux/falloc.h>
C
Chris Mason 已提交
28 29 30 31
#include <linux/swap.h>
#include <linux/writeback.h>
#include <linux/statfs.h>
#include <linux/compat.h>
32
#include <linux/slab.h>
C
Chris Mason 已提交
33 34 35 36 37 38
#include "ctree.h"
#include "disk-io.h"
#include "transaction.h"
#include "btrfs_inode.h"
#include "ioctl.h"
#include "print-tree.h"
39 40
#include "tree-log.h"
#include "locking.h"
41
#include "compat.h"
C
Chris Mason 已提交
42 43


C
Chris Mason 已提交
44 45 46
/* simple helper to fault in pages and copy.  This should go away
 * and be replaced with calls into generic code.
 */
C
Chris Mason 已提交
47
static noinline int btrfs_copy_from_user(loff_t pos, int num_pages,
J
Josef Bacik 已提交
48
					 size_t write_bytes,
49
					 struct page **prepared_pages,
50
					 struct iov_iter *i)
C
Chris Mason 已提交
51
{
52
	size_t copied = 0;
J
Josef Bacik 已提交
53
	size_t total_copied = 0;
54
	int pg = 0;
C
Chris Mason 已提交
55 56
	int offset = pos & (PAGE_CACHE_SIZE - 1);

57
	while (write_bytes > 0) {
C
Chris Mason 已提交
58 59
		size_t count = min_t(size_t,
				     PAGE_CACHE_SIZE - offset, write_bytes);
60
		struct page *page = prepared_pages[pg];
61 62 63 64 65 66 67 68 69
		/*
		 * Copy data from userspace to the current page
		 *
		 * Disable pagefault to avoid recursive lock since
		 * the pages are already locked
		 */
		pagefault_disable();
		copied = iov_iter_copy_from_user_atomic(page, i, offset, count);
		pagefault_enable();
70

C
Chris Mason 已提交
71 72
		/* Flush processor's dcache for this page */
		flush_dcache_page(page);
73 74 75 76 77 78 79 80 81 82 83 84 85

		/*
		 * if we get a partial write, we can end up with
		 * partially up to date pages.  These add
		 * a lot of complexity, so make sure they don't
		 * happen by forcing this copy to be retried.
		 *
		 * The rest of the btrfs_file_write code will fall
		 * back to page at a time copies after we return 0.
		 */
		if (!PageUptodate(page) && copied < count)
			copied = 0;

86 87
		iov_iter_advance(i, copied);
		write_bytes -= copied;
88
		total_copied += copied;
C
Chris Mason 已提交
89

90
		/* Return to btrfs_file_aio_write to fault page */
J
Josef Bacik 已提交
91
		if (unlikely(copied == 0))
92
			break;
93 94 95 96 97 98 99

		if (unlikely(copied < PAGE_CACHE_SIZE - offset)) {
			offset += copied;
		} else {
			pg++;
			offset = 0;
		}
C
Chris Mason 已提交
100
	}
101
	return total_copied;
C
Chris Mason 已提交
102 103
}

C
Chris Mason 已提交
104 105 106
/*
 * unlocks pages after btrfs_file_write is done with them
 */
C
Chris Mason 已提交
107
static noinline void btrfs_drop_pages(struct page **pages, size_t num_pages)
C
Chris Mason 已提交
108 109 110
{
	size_t i;
	for (i = 0; i < num_pages; i++) {
C
Chris Mason 已提交
111 112 113 114
		/* page checked is some magic around finding pages that
		 * have been modified without going through btrfs_set_page_dirty
		 * clear it here
		 */
C
Chris Mason 已提交
115
		ClearPageChecked(pages[i]);
C
Chris Mason 已提交
116 117 118 119 120 121
		unlock_page(pages[i]);
		mark_page_accessed(pages[i]);
		page_cache_release(pages[i]);
	}
}

C
Chris Mason 已提交
122 123 124 125 126 127 128 129
/*
 * after copy_from_user, pages need to be dirtied and we need to make
 * sure holes are created between the current EOF and the start of
 * any next extents (if required).
 *
 * this also makes the decision about creating an inline extent vs
 * doing real data extents, marking pages dirty and delalloc as required.
 */
J
Josef Bacik 已提交
130 131 132 133 134 135
static noinline int dirty_and_release_pages(struct btrfs_root *root,
					    struct file *file,
					    struct page **pages,
					    size_t num_pages,
					    loff_t pos,
					    size_t write_bytes)
C
Chris Mason 已提交
136 137
{
	int err = 0;
138
	int i;
139
	struct inode *inode = fdentry(file)->d_inode;
140
	u64 num_bytes;
141 142 143 144
	u64 start_pos;
	u64 end_of_last_block;
	u64 end_pos = pos + write_bytes;
	loff_t isize = i_size_read(inode);
C
Chris Mason 已提交
145

146
	start_pos = pos & ~((u64)root->sectorsize - 1);
147 148
	num_bytes = (write_bytes + pos - start_pos +
		    root->sectorsize - 1) & ~((u64)root->sectorsize - 1);
C
Chris Mason 已提交
149

150
	end_of_last_block = start_pos + num_bytes - 1;
151 152
	err = btrfs_set_extent_delalloc(inode, start_pos, end_of_last_block,
					NULL);
J
Josef Bacik 已提交
153 154
	if (err)
		return err;
J
Josef Bacik 已提交
155

C
Chris Mason 已提交
156 157 158 159 160
	for (i = 0; i < num_pages; i++) {
		struct page *p = pages[i];
		SetPageUptodate(p);
		ClearPageChecked(p);
		set_page_dirty(p);
161
	}
J
Josef Bacik 已提交
162 163 164 165 166 167 168

	/*
	 * we've only changed i_size in ram, and we haven't updated
	 * the disk i_size.  There is no need to log the inode
	 * at this time.
	 */
	if (end_pos > isize)
169
		i_size_write(inode, end_pos);
170
	return 0;
C
Chris Mason 已提交
171 172
}

C
Chris Mason 已提交
173 174 175 176
/*
 * this drops all the extents in the cache that intersect the range
 * [start, end].  Existing extents are split as required.
 */
177 178
int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
			    int skip_pinned)
179 180
{
	struct extent_map *em;
181 182
	struct extent_map *split = NULL;
	struct extent_map *split2 = NULL;
183
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
184
	u64 len = end - start + 1;
185 186
	int ret;
	int testend = 1;
187
	unsigned long flags;
C
Chris Mason 已提交
188
	int compressed = 0;
189

190
	WARN_ON(end < start);
191
	if (end == (u64)-1) {
192
		len = (u64)-1;
193 194
		testend = 0;
	}
C
Chris Mason 已提交
195
	while (1) {
196 197 198 199
		if (!split)
			split = alloc_extent_map(GFP_NOFS);
		if (!split2)
			split2 = alloc_extent_map(GFP_NOFS);
200
		BUG_ON(!split || !split2);
201

202
		write_lock(&em_tree->lock);
203
		em = lookup_extent_mapping(em_tree, start, len);
204
		if (!em) {
205
			write_unlock(&em_tree->lock);
206
			break;
207
		}
208 209
		flags = em->flags;
		if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) {
210
			if (testend && em->start + em->len >= start + len) {
211
				free_extent_map(em);
C
Chris Mason 已提交
212
				write_unlock(&em_tree->lock);
213 214
				break;
			}
215 216
			start = em->start + em->len;
			if (testend)
217 218
				len = start + len - (em->start + em->len);
			free_extent_map(em);
C
Chris Mason 已提交
219
			write_unlock(&em_tree->lock);
220 221
			continue;
		}
C
Chris Mason 已提交
222
		compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
223
		clear_bit(EXTENT_FLAG_PINNED, &em->flags);
224
		remove_extent_mapping(em_tree, em);
225 226 227 228 229

		if (em->block_start < EXTENT_MAP_LAST_BYTE &&
		    em->start < start) {
			split->start = em->start;
			split->len = start - em->start;
230
			split->orig_start = em->orig_start;
231
			split->block_start = em->block_start;
C
Chris Mason 已提交
232 233 234 235 236 237

			if (compressed)
				split->block_len = em->block_len;
			else
				split->block_len = split->len;

238
			split->bdev = em->bdev;
239
			split->flags = flags;
240
			split->compress_type = em->compress_type;
241 242 243 244 245 246 247 248 249 250 251 252 253
			ret = add_extent_mapping(em_tree, split);
			BUG_ON(ret);
			free_extent_map(split);
			split = split2;
			split2 = NULL;
		}
		if (em->block_start < EXTENT_MAP_LAST_BYTE &&
		    testend && em->start + em->len > start + len) {
			u64 diff = start + len - em->start;

			split->start = start + len;
			split->len = em->start + em->len - (start + len);
			split->bdev = em->bdev;
254
			split->flags = flags;
255
			split->compress_type = em->compress_type;
256

C
Chris Mason 已提交
257 258 259
			if (compressed) {
				split->block_len = em->block_len;
				split->block_start = em->block_start;
260
				split->orig_start = em->orig_start;
C
Chris Mason 已提交
261 262 263
			} else {
				split->block_len = split->len;
				split->block_start = em->block_start + diff;
264
				split->orig_start = split->start;
C
Chris Mason 已提交
265
			}
266 267 268 269 270 271

			ret = add_extent_mapping(em_tree, split);
			BUG_ON(ret);
			free_extent_map(split);
			split = NULL;
		}
272
		write_unlock(&em_tree->lock);
273

274 275 276 277 278
		/* once for us */
		free_extent_map(em);
		/* once for the tree*/
		free_extent_map(em);
	}
279 280 281 282
	if (split)
		free_extent_map(split);
	if (split2)
		free_extent_map(split2);
283 284 285
	return 0;
}

C
Chris Mason 已提交
286 287 288 289 290 291 292 293 294
/*
 * this is very complex, but the basic idea is to drop all extents
 * in the range start - end.  hint_block is filled in with a block number
 * that would be a good hint to the block allocator for this file.
 *
 * If an extent intersects the range but is not entirely inside the range
 * it is either truncated or split.  Anything entirely inside the range
 * is deleted from the tree.
 */
Y
Yan, Zheng 已提交
295 296
int btrfs_drop_extents(struct btrfs_trans_handle *trans, struct inode *inode,
		       u64 start, u64 end, u64 *hint_byte, int drop_cache)
C
Chris Mason 已提交
297
{
Y
Yan, Zheng 已提交
298
	struct btrfs_root *root = BTRFS_I(inode)->root;
299
	struct extent_buffer *leaf;
Y
Yan, Zheng 已提交
300
	struct btrfs_file_extent_item *fi;
C
Chris Mason 已提交
301
	struct btrfs_path *path;
302
	struct btrfs_key key;
Y
Yan, Zheng 已提交
303 304 305 306 307 308 309 310 311
	struct btrfs_key new_key;
	u64 search_start = start;
	u64 disk_bytenr = 0;
	u64 num_bytes = 0;
	u64 extent_offset = 0;
	u64 extent_end = 0;
	int del_nr = 0;
	int del_slot = 0;
	int extent_type;
C
Chris Mason 已提交
312
	int recow;
313
	int ret;
C
Chris Mason 已提交
314

C
Chris Mason 已提交
315 316
	if (drop_cache)
		btrfs_drop_extent_cache(inode, start, end - 1, 0);
317

C
Chris Mason 已提交
318 319 320
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
Y
Yan, Zheng 已提交
321

C
Chris Mason 已提交
322
	while (1) {
C
Chris Mason 已提交
323
		recow = 0;
C
Chris Mason 已提交
324 325 326
		ret = btrfs_lookup_file_extent(trans, root, path, inode->i_ino,
					       search_start, -1);
		if (ret < 0)
Y
Yan, Zheng 已提交
327 328 329 330 331 332 333
			break;
		if (ret > 0 && path->slots[0] > 0 && search_start == start) {
			leaf = path->nodes[0];
			btrfs_item_key_to_cpu(leaf, &key, path->slots[0] - 1);
			if (key.objectid == inode->i_ino &&
			    key.type == BTRFS_EXTENT_DATA_KEY)
				path->slots[0]--;
C
Chris Mason 已提交
334
		}
Y
Yan, Zheng 已提交
335
		ret = 0;
336
next_slot:
337
		leaf = path->nodes[0];
Y
Yan, Zheng 已提交
338 339 340 341 342 343 344 345
		if (path->slots[0] >= btrfs_header_nritems(leaf)) {
			BUG_ON(del_nr > 0);
			ret = btrfs_next_leaf(root, path);
			if (ret < 0)
				break;
			if (ret > 0) {
				ret = 0;
				break;
346
			}
Y
Yan, Zheng 已提交
347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369
			leaf = path->nodes[0];
			recow = 1;
		}

		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
		if (key.objectid > inode->i_ino ||
		    key.type > BTRFS_EXTENT_DATA_KEY || key.offset >= end)
			break;

		fi = btrfs_item_ptr(leaf, path->slots[0],
				    struct btrfs_file_extent_item);
		extent_type = btrfs_file_extent_type(leaf, fi);

		if (extent_type == BTRFS_FILE_EXTENT_REG ||
		    extent_type == BTRFS_FILE_EXTENT_PREALLOC) {
			disk_bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
			num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
			extent_offset = btrfs_file_extent_offset(leaf, fi);
			extent_end = key.offset +
				btrfs_file_extent_num_bytes(leaf, fi);
		} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
			extent_end = key.offset +
				btrfs_file_extent_inline_len(leaf, fi);
370
		} else {
Y
Yan, Zheng 已提交
371
			WARN_ON(1);
372
			extent_end = search_start;
C
Chris Mason 已提交
373 374
		}

Y
Yan, Zheng 已提交
375 376
		if (extent_end <= search_start) {
			path->slots[0]++;
377
			goto next_slot;
C
Chris Mason 已提交
378 379
		}

Y
Yan, Zheng 已提交
380 381 382 383
		search_start = max(key.offset, start);
		if (recow) {
			btrfs_release_path(root, path);
			continue;
C
Chris Mason 已提交
384
		}
Y
Yan Zheng 已提交
385

Y
Yan, Zheng 已提交
386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
		/*
		 *     | - range to drop - |
		 *  | -------- extent -------- |
		 */
		if (start > key.offset && end < extent_end) {
			BUG_ON(del_nr > 0);
			BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE);

			memcpy(&new_key, &key, sizeof(new_key));
			new_key.offset = start;
			ret = btrfs_duplicate_item(trans, root, path,
						   &new_key);
			if (ret == -EAGAIN) {
				btrfs_release_path(root, path);
				continue;
Y
Yan Zheng 已提交
401
			}
Y
Yan, Zheng 已提交
402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420
			if (ret < 0)
				break;

			leaf = path->nodes[0];
			fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
					    struct btrfs_file_extent_item);
			btrfs_set_file_extent_num_bytes(leaf, fi,
							start - key.offset);

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

			extent_offset += start - key.offset;
			btrfs_set_file_extent_offset(leaf, fi, extent_offset);
			btrfs_set_file_extent_num_bytes(leaf, fi,
							extent_end - start);
			btrfs_mark_buffer_dirty(leaf);

			if (disk_bytenr > 0) {
421
				ret = btrfs_inc_extent_ref(trans, root,
Y
Yan, Zheng 已提交
422 423 424 425
						disk_bytenr, num_bytes, 0,
						root->root_key.objectid,
						new_key.objectid,
						start - extent_offset);
426
				BUG_ON(ret);
Y
Yan, Zheng 已提交
427
				*hint_byte = disk_bytenr;
428
			}
Y
Yan, Zheng 已提交
429
			key.offset = start;
Y
Yan Zheng 已提交
430
		}
Y
Yan, Zheng 已提交
431 432 433 434 435 436
		/*
		 *  | ---- range to drop ----- |
		 *      | -------- extent -------- |
		 */
		if (start <= key.offset && end < extent_end) {
			BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE);
Y
Yan Zheng 已提交
437

Y
Yan, Zheng 已提交
438 439 440
			memcpy(&new_key, &key, sizeof(new_key));
			new_key.offset = end;
			btrfs_set_item_key_safe(trans, root, path, &new_key);
Y
Yan Zheng 已提交
441

Y
Yan, Zheng 已提交
442 443 444 445 446 447 448 449
			extent_offset += end - key.offset;
			btrfs_set_file_extent_offset(leaf, fi, extent_offset);
			btrfs_set_file_extent_num_bytes(leaf, fi,
							extent_end - end);
			btrfs_mark_buffer_dirty(leaf);
			if (disk_bytenr > 0) {
				inode_sub_bytes(inode, end - key.offset);
				*hint_byte = disk_bytenr;
C
Chris Mason 已提交
450
			}
Y
Yan, Zheng 已提交
451
			break;
C
Chris Mason 已提交
452
		}
453

Y
Yan, Zheng 已提交
454 455 456 457 458 459 460 461
		search_start = extent_end;
		/*
		 *       | ---- range to drop ----- |
		 *  | -------- extent -------- |
		 */
		if (start > key.offset && end >= extent_end) {
			BUG_ON(del_nr > 0);
			BUG_ON(extent_type == BTRFS_FILE_EXTENT_INLINE);
462

Y
Yan, Zheng 已提交
463 464 465 466 467 468 469 470 471
			btrfs_set_file_extent_num_bytes(leaf, fi,
							start - key.offset);
			btrfs_mark_buffer_dirty(leaf);
			if (disk_bytenr > 0) {
				inode_sub_bytes(inode, extent_end - start);
				*hint_byte = disk_bytenr;
			}
			if (end == extent_end)
				break;
C
Chris Mason 已提交
472

Y
Yan, Zheng 已提交
473 474
			path->slots[0]++;
			goto next_slot;
Z
Zheng Yan 已提交
475 476
		}

Y
Yan, Zheng 已提交
477 478 479 480 481 482 483 484 485 486 487 488
		/*
		 *  | ---- range to drop ----- |
		 *    | ------ extent ------ |
		 */
		if (start <= key.offset && end >= extent_end) {
			if (del_nr == 0) {
				del_slot = path->slots[0];
				del_nr = 1;
			} else {
				BUG_ON(del_slot + del_nr != path->slots[0]);
				del_nr++;
			}
Z
Zheng Yan 已提交
489

Y
Yan, Zheng 已提交
490
			if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
491
				inode_sub_bytes(inode,
Y
Yan, Zheng 已提交
492 493 494 495
						extent_end - key.offset);
				extent_end = ALIGN(extent_end,
						   root->sectorsize);
			} else if (disk_bytenr > 0) {
Z
Zheng Yan 已提交
496
				ret = btrfs_free_extent(trans, root,
Y
Yan, Zheng 已提交
497 498
						disk_bytenr, num_bytes, 0,
						root->root_key.objectid,
499
						key.objectid, key.offset -
Y
Yan, Zheng 已提交
500
						extent_offset);
Z
Zheng Yan 已提交
501
				BUG_ON(ret);
Y
Yan, Zheng 已提交
502 503 504
				inode_sub_bytes(inode,
						extent_end - key.offset);
				*hint_byte = disk_bytenr;
Z
Zheng Yan 已提交
505 506
			}

Y
Yan, Zheng 已提交
507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523
			if (end == extent_end)
				break;

			if (path->slots[0] + 1 < btrfs_header_nritems(leaf)) {
				path->slots[0]++;
				goto next_slot;
			}

			ret = btrfs_del_items(trans, root, path, del_slot,
					      del_nr);
			BUG_ON(ret);

			del_nr = 0;
			del_slot = 0;

			btrfs_release_path(root, path);
			continue;
C
Chris Mason 已提交
524
		}
Y
Yan, Zheng 已提交
525 526

		BUG_ON(1);
C
Chris Mason 已提交
527
	}
Y
Yan, Zheng 已提交
528 529 530 531

	if (del_nr > 0) {
		ret = btrfs_del_items(trans, root, path, del_slot, del_nr);
		BUG_ON(ret);
Y
Yan Zheng 已提交
532
	}
Y
Yan, Zheng 已提交
533 534

	btrfs_free_path(path);
C
Chris Mason 已提交
535 536 537
	return ret;
}

Y
Yan Zheng 已提交
538
static int extent_mergeable(struct extent_buffer *leaf, int slot,
539 540
			    u64 objectid, u64 bytenr, u64 orig_offset,
			    u64 *start, u64 *end)
Y
Yan Zheng 已提交
541 542 543 544 545 546 547 548 549 550 551 552 553 554 555
{
	struct btrfs_file_extent_item *fi;
	struct btrfs_key key;
	u64 extent_end;

	if (slot < 0 || slot >= btrfs_header_nritems(leaf))
		return 0;

	btrfs_item_key_to_cpu(leaf, &key, slot);
	if (key.objectid != objectid || key.type != BTRFS_EXTENT_DATA_KEY)
		return 0;

	fi = btrfs_item_ptr(leaf, slot, struct btrfs_file_extent_item);
	if (btrfs_file_extent_type(leaf, fi) != BTRFS_FILE_EXTENT_REG ||
	    btrfs_file_extent_disk_bytenr(leaf, fi) != bytenr ||
556
	    btrfs_file_extent_offset(leaf, fi) != key.offset - orig_offset ||
Y
Yan Zheng 已提交
557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580
	    btrfs_file_extent_compression(leaf, fi) ||
	    btrfs_file_extent_encryption(leaf, fi) ||
	    btrfs_file_extent_other_encoding(leaf, fi))
		return 0;

	extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi);
	if ((*start && *start != key.offset) || (*end && *end != extent_end))
		return 0;

	*start = key.offset;
	*end = extent_end;
	return 1;
}

/*
 * Mark extent in the range start - end as written.
 *
 * This changes extent type from 'pre-allocated' to 'regular'. If only
 * part of extent is marked as written, the extent will be split into
 * two or three.
 */
int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
			      struct inode *inode, u64 start, u64 end)
{
Y
Yan, Zheng 已提交
581
	struct btrfs_root *root = BTRFS_I(inode)->root;
Y
Yan Zheng 已提交
582 583 584 585
	struct extent_buffer *leaf;
	struct btrfs_path *path;
	struct btrfs_file_extent_item *fi;
	struct btrfs_key key;
Y
Yan, Zheng 已提交
586
	struct btrfs_key new_key;
Y
Yan Zheng 已提交
587 588 589
	u64 bytenr;
	u64 num_bytes;
	u64 extent_end;
590
	u64 orig_offset;
Y
Yan Zheng 已提交
591 592
	u64 other_start;
	u64 other_end;
Y
Yan, Zheng 已提交
593 594 595
	u64 split;
	int del_nr = 0;
	int del_slot = 0;
596
	int recow;
Y
Yan Zheng 已提交
597 598 599 600 601 602 603
	int ret;

	btrfs_drop_extent_cache(inode, start, end - 1, 0);

	path = btrfs_alloc_path();
	BUG_ON(!path);
again:
604
	recow = 0;
Y
Yan, Zheng 已提交
605
	split = start;
Y
Yan Zheng 已提交
606 607
	key.objectid = inode->i_ino;
	key.type = BTRFS_EXTENT_DATA_KEY;
Y
Yan, Zheng 已提交
608
	key.offset = split;
Y
Yan Zheng 已提交
609 610

	ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
611 612
	if (ret < 0)
		goto out;
Y
Yan Zheng 已提交
613 614 615 616 617 618 619 620 621
	if (ret > 0 && path->slots[0] > 0)
		path->slots[0]--;

	leaf = path->nodes[0];
	btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
	BUG_ON(key.objectid != inode->i_ino ||
	       key.type != BTRFS_EXTENT_DATA_KEY);
	fi = btrfs_item_ptr(leaf, path->slots[0],
			    struct btrfs_file_extent_item);
Y
Yan, Zheng 已提交
622 623
	BUG_ON(btrfs_file_extent_type(leaf, fi) !=
	       BTRFS_FILE_EXTENT_PREALLOC);
Y
Yan Zheng 已提交
624 625 626 627 628
	extent_end = key.offset + btrfs_file_extent_num_bytes(leaf, fi);
	BUG_ON(key.offset > start || extent_end < end);

	bytenr = btrfs_file_extent_disk_bytenr(leaf, fi);
	num_bytes = btrfs_file_extent_disk_num_bytes(leaf, fi);
629
	orig_offset = key.offset - btrfs_file_extent_offset(leaf, fi);
630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678
	memcpy(&new_key, &key, sizeof(new_key));

	if (start == key.offset && end < extent_end) {
		other_start = 0;
		other_end = start;
		if (extent_mergeable(leaf, path->slots[0] - 1,
				     inode->i_ino, bytenr, orig_offset,
				     &other_start, &other_end)) {
			new_key.offset = end;
			btrfs_set_item_key_safe(trans, root, path, &new_key);
			fi = btrfs_item_ptr(leaf, path->slots[0],
					    struct btrfs_file_extent_item);
			btrfs_set_file_extent_num_bytes(leaf, fi,
							extent_end - end);
			btrfs_set_file_extent_offset(leaf, fi,
						     end - orig_offset);
			fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
					    struct btrfs_file_extent_item);
			btrfs_set_file_extent_num_bytes(leaf, fi,
							end - other_start);
			btrfs_mark_buffer_dirty(leaf);
			goto out;
		}
	}

	if (start > key.offset && end == extent_end) {
		other_start = end;
		other_end = 0;
		if (extent_mergeable(leaf, path->slots[0] + 1,
				     inode->i_ino, bytenr, orig_offset,
				     &other_start, &other_end)) {
			fi = btrfs_item_ptr(leaf, path->slots[0],
					    struct btrfs_file_extent_item);
			btrfs_set_file_extent_num_bytes(leaf, fi,
							start - key.offset);
			path->slots[0]++;
			new_key.offset = start;
			btrfs_set_item_key_safe(trans, root, path, &new_key);

			fi = btrfs_item_ptr(leaf, path->slots[0],
					    struct btrfs_file_extent_item);
			btrfs_set_file_extent_num_bytes(leaf, fi,
							other_end - start);
			btrfs_set_file_extent_offset(leaf, fi,
						     start - orig_offset);
			btrfs_mark_buffer_dirty(leaf);
			goto out;
		}
	}
Y
Yan Zheng 已提交
679

Y
Yan, Zheng 已提交
680 681 682 683 684 685 686 687 688
	while (start > key.offset || end < extent_end) {
		if (key.offset == start)
			split = end;

		new_key.offset = split;
		ret = btrfs_duplicate_item(trans, root, path, &new_key);
		if (ret == -EAGAIN) {
			btrfs_release_path(root, path);
			goto again;
Y
Yan Zheng 已提交
689
		}
Y
Yan, Zheng 已提交
690
		BUG_ON(ret < 0);
Y
Yan Zheng 已提交
691

Y
Yan, Zheng 已提交
692 693
		leaf = path->nodes[0];
		fi = btrfs_item_ptr(leaf, path->slots[0] - 1,
Y
Yan Zheng 已提交
694 695
				    struct btrfs_file_extent_item);
		btrfs_set_file_extent_num_bytes(leaf, fi,
Y
Yan, Zheng 已提交
696 697 698 699 700 701 702 703
						split - key.offset);

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

		btrfs_set_file_extent_offset(leaf, fi, split - orig_offset);
		btrfs_set_file_extent_num_bytes(leaf, fi,
						extent_end - split);
Y
Yan Zheng 已提交
704 705
		btrfs_mark_buffer_dirty(leaf);

Y
Yan, Zheng 已提交
706 707 708
		ret = btrfs_inc_extent_ref(trans, root, bytenr, num_bytes, 0,
					   root->root_key.objectid,
					   inode->i_ino, orig_offset);
Y
Yan Zheng 已提交
709 710
		BUG_ON(ret);

Y
Yan, Zheng 已提交
711 712 713 714
		if (split == start) {
			key.offset = start;
		} else {
			BUG_ON(start != key.offset);
Y
Yan Zheng 已提交
715
			path->slots[0]--;
Y
Yan, Zheng 已提交
716
			extent_end = end;
Y
Yan Zheng 已提交
717
		}
718
		recow = 1;
Y
Yan Zheng 已提交
719 720
	}

Y
Yan, Zheng 已提交
721 722
	other_start = end;
	other_end = 0;
723 724 725 726 727 728 729
	if (extent_mergeable(leaf, path->slots[0] + 1,
			     inode->i_ino, bytenr, orig_offset,
			     &other_start, &other_end)) {
		if (recow) {
			btrfs_release_path(root, path);
			goto again;
		}
Y
Yan, Zheng 已提交
730 731 732 733 734 735 736
		extent_end = other_end;
		del_slot = path->slots[0] + 1;
		del_nr++;
		ret = btrfs_free_extent(trans, root, bytenr, num_bytes,
					0, root->root_key.objectid,
					inode->i_ino, orig_offset);
		BUG_ON(ret);
Y
Yan Zheng 已提交
737
	}
Y
Yan, Zheng 已提交
738 739
	other_start = 0;
	other_end = start;
740 741 742 743 744 745 746
	if (extent_mergeable(leaf, path->slots[0] - 1,
			     inode->i_ino, bytenr, orig_offset,
			     &other_start, &other_end)) {
		if (recow) {
			btrfs_release_path(root, path);
			goto again;
		}
Y
Yan, Zheng 已提交
747 748 749 750 751 752 753 754 755
		key.offset = other_start;
		del_slot = path->slots[0];
		del_nr++;
		ret = btrfs_free_extent(trans, root, bytenr, num_bytes,
					0, root->root_key.objectid,
					inode->i_ino, orig_offset);
		BUG_ON(ret);
	}
	if (del_nr == 0) {
756 757
		fi = btrfs_item_ptr(leaf, path->slots[0],
			   struct btrfs_file_extent_item);
Y
Yan, Zheng 已提交
758 759 760
		btrfs_set_file_extent_type(leaf, fi,
					   BTRFS_FILE_EXTENT_REG);
		btrfs_mark_buffer_dirty(leaf);
761
	} else {
762 763
		fi = btrfs_item_ptr(leaf, del_slot - 1,
			   struct btrfs_file_extent_item);
764 765 766 767 768
		btrfs_set_file_extent_type(leaf, fi,
					   BTRFS_FILE_EXTENT_REG);
		btrfs_set_file_extent_num_bytes(leaf, fi,
						extent_end - key.offset);
		btrfs_mark_buffer_dirty(leaf);
Y
Yan, Zheng 已提交
769

770 771 772
		ret = btrfs_del_items(trans, root, path, del_slot, del_nr);
		BUG_ON(ret);
	}
Y
Yan, Zheng 已提交
773
out:
Y
Yan Zheng 已提交
774 775 776 777
	btrfs_free_path(path);
	return 0;
}

778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798
/*
 * on error we return an unlocked page and the error value
 * on success we return a locked page and 0
 */
static int prepare_uptodate_page(struct page *page, u64 pos)
{
	int ret = 0;

	if ((pos & (PAGE_CACHE_SIZE - 1)) && !PageUptodate(page)) {
		ret = btrfs_readpage(NULL, page);
		if (ret)
			return ret;
		lock_page(page);
		if (!PageUptodate(page)) {
			unlock_page(page);
			return -EIO;
		}
	}
	return 0;
}

C
Chris Mason 已提交
799
/*
C
Chris Mason 已提交
800 801 802
 * this gets pages into the page cache and locks them down, it also properly
 * waits for data=ordered extents to finish before allowing the pages to be
 * modified.
C
Chris Mason 已提交
803
 */
C
Chris Mason 已提交
804
static noinline int prepare_pages(struct btrfs_root *root, struct file *file,
805 806 807
			 struct page **pages, size_t num_pages,
			 loff_t pos, unsigned long first_index,
			 unsigned long last_index, size_t write_bytes)
C
Chris Mason 已提交
808
{
809
	struct extent_state *cached_state = NULL;
C
Chris Mason 已提交
810 811
	int i;
	unsigned long index = pos >> PAGE_CACHE_SHIFT;
812
	struct inode *inode = fdentry(file)->d_inode;
C
Chris Mason 已提交
813
	int err = 0;
814
	int faili = 0;
815
	u64 start_pos;
816
	u64 last_pos;
817

818
	start_pos = pos & ~((u64)root->sectorsize - 1);
819
	last_pos = ((u64)index + num_pages) << PAGE_CACHE_SHIFT;
C
Chris Mason 已提交
820

Y
Yan Zheng 已提交
821
	if (start_pos > inode->i_size) {
822
		err = btrfs_cont_expand(inode, i_size_read(inode), start_pos);
Y
Yan Zheng 已提交
823 824 825 826
		if (err)
			return err;
	}

827
again:
C
Chris Mason 已提交
828 829 830
	for (i = 0; i < num_pages; i++) {
		pages[i] = grab_cache_page(inode->i_mapping, index + i);
		if (!pages[i]) {
831 832 833 834 835 836 837 838 839 840 841 842 843 844
			faili = i - 1;
			err = -ENOMEM;
			goto fail;
		}

		if (i == 0)
			err = prepare_uptodate_page(pages[i], pos);
		if (i == num_pages - 1)
			err = prepare_uptodate_page(pages[i],
						    pos + write_bytes);
		if (err) {
			page_cache_release(pages[i]);
			faili = i - 1;
			goto fail;
C
Chris Mason 已提交
845
		}
C
Chris Mason 已提交
846
		wait_on_page_writeback(pages[i]);
C
Chris Mason 已提交
847
	}
848
	err = 0;
849
	if (start_pos < inode->i_size) {
850
		struct btrfs_ordered_extent *ordered;
851 852 853
		lock_extent_bits(&BTRFS_I(inode)->io_tree,
				 start_pos, last_pos - 1, 0, &cached_state,
				 GFP_NOFS);
C
Chris Mason 已提交
854 855
		ordered = btrfs_lookup_first_ordered_extent(inode,
							    last_pos - 1);
856 857 858 859
		if (ordered &&
		    ordered->file_offset + ordered->len > start_pos &&
		    ordered->file_offset < last_pos) {
			btrfs_put_ordered_extent(ordered);
860 861 862
			unlock_extent_cached(&BTRFS_I(inode)->io_tree,
					     start_pos, last_pos - 1,
					     &cached_state, GFP_NOFS);
863 864 865 866 867 868 869 870 871 872 873
			for (i = 0; i < num_pages; i++) {
				unlock_page(pages[i]);
				page_cache_release(pages[i]);
			}
			btrfs_wait_ordered_range(inode, start_pos,
						 last_pos - start_pos);
			goto again;
		}
		if (ordered)
			btrfs_put_ordered_extent(ordered);

874
		clear_extent_bit(&BTRFS_I(inode)->io_tree, start_pos,
875
				  last_pos - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
876
				  EXTENT_DO_ACCOUNTING, 0, 0, &cached_state,
877
				  GFP_NOFS);
878 879 880
		unlock_extent_cached(&BTRFS_I(inode)->io_tree,
				     start_pos, last_pos - 1, &cached_state,
				     GFP_NOFS);
881
	}
882
	for (i = 0; i < num_pages; i++) {
883
		clear_page_dirty_for_io(pages[i]);
884 885 886
		set_page_extent_mapped(pages[i]);
		WARN_ON(!PageLocked(pages[i]));
	}
C
Chris Mason 已提交
887
	return 0;
888 889 890 891 892 893 894 895
fail:
	while (faili >= 0) {
		unlock_page(pages[faili]);
		page_cache_release(pages[faili]);
		faili--;
	}
	return err;

C
Chris Mason 已提交
896 897
}

J
Josef Bacik 已提交
898 899 900
static noinline ssize_t __btrfs_buffered_write(struct file *file,
					       struct iov_iter *i,
					       loff_t pos)
901
{
902 903 904
	struct inode *inode = fdentry(file)->d_inode;
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct page **pages = NULL;
C
Chris Mason 已提交
905 906
	unsigned long first_index;
	unsigned long last_index;
J
Josef Bacik 已提交
907 908
	size_t num_written = 0;
	int nrptrs;
909
	int ret = 0;
910

J
Josef Bacik 已提交
911
	nrptrs = min((iov_iter_count(i) + PAGE_CACHE_SIZE - 1) /
912 913
		     PAGE_CACHE_SIZE, PAGE_CACHE_SIZE /
		     (sizeof(struct page *)));
914
	pages = kmalloc(nrptrs * sizeof(struct page *), GFP_KERNEL);
J
Josef Bacik 已提交
915 916
	if (!pages)
		return -ENOMEM;
917

C
Chris Mason 已提交
918
	first_index = pos >> PAGE_CACHE_SHIFT;
J
Josef Bacik 已提交
919
	last_index = (pos + iov_iter_count(i)) >> PAGE_CACHE_SHIFT;
C
Chris Mason 已提交
920

J
Josef Bacik 已提交
921
	while (iov_iter_count(i) > 0) {
C
Chris Mason 已提交
922
		size_t offset = pos & (PAGE_CACHE_SIZE - 1);
J
Josef Bacik 已提交
923
		size_t write_bytes = min(iov_iter_count(i),
924
					 nrptrs * (size_t)PAGE_CACHE_SIZE -
925
					 offset);
Y
Yan, Zheng 已提交
926 927
		size_t num_pages = (write_bytes + offset +
				    PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
J
Josef Bacik 已提交
928 929
		size_t dirty_pages;
		size_t copied;
C
Chris Mason 已提交
930

931
		WARN_ON(num_pages > nrptrs);
932

933 934 935 936
		/*
		 * Fault pages before locking them in prepare_pages
		 * to avoid recursive lock
		 */
J
Josef Bacik 已提交
937
		if (unlikely(iov_iter_fault_in_readable(i, write_bytes))) {
938
			ret = -EFAULT;
J
Josef Bacik 已提交
939
			break;
940 941 942 943
		}

		ret = btrfs_delalloc_reserve_space(inode,
					num_pages << PAGE_CACHE_SHIFT);
944
		if (ret)
J
Josef Bacik 已提交
945
			break;
946

947 948 949 950 951
		/*
		 * This is going to setup the pages array with the number of
		 * pages we want, so we don't really need to worry about the
		 * contents of pages from loop to loop
		 */
C
Chris Mason 已提交
952 953
		ret = prepare_pages(root, file, pages, num_pages,
				    pos, first_index, last_index,
954
				    write_bytes);
J
Josef Bacik 已提交
955
		if (ret) {
956 957
			btrfs_delalloc_release_space(inode,
					num_pages << PAGE_CACHE_SHIFT);
J
Josef Bacik 已提交
958
			break;
J
Josef Bacik 已提交
959
		}
C
Chris Mason 已提交
960

961
		copied = btrfs_copy_from_user(pos, num_pages,
J
Josef Bacik 已提交
962
					   write_bytes, pages, i);
963 964 965 966 967 968 969 970 971 972 973 974 975 976

		/*
		 * if we have trouble faulting in the pages, fall
		 * back to one page at a time
		 */
		if (copied < write_bytes)
			nrptrs = 1;

		if (copied == 0)
			dirty_pages = 0;
		else
			dirty_pages = (copied + offset +
				       PAGE_CACHE_SIZE - 1) >>
				       PAGE_CACHE_SHIFT;
977

J
Josef Bacik 已提交
978 979 980 981 982 983 984
		/*
		 * If we had a short copy we need to release the excess delaloc
		 * bytes we reserved.  We need to increment outstanding_extents
		 * because btrfs_delalloc_release_space will decrement it, but
		 * we still have an outstanding extent for the chunk we actually
		 * managed to copy.
		 */
985 986 987 988 989 990 991 992 993 994
		if (num_pages > dirty_pages) {
			if (copied > 0)
				atomic_inc(
					&BTRFS_I(inode)->outstanding_extents);
			btrfs_delalloc_release_space(inode,
					(num_pages - dirty_pages) <<
					PAGE_CACHE_SHIFT);
		}

		if (copied > 0) {
J
Josef Bacik 已提交
995 996 997 998 999 1000 1001 1002 1003
			ret = dirty_and_release_pages(root, file, pages,
						      dirty_pages, pos,
						      copied);
			if (ret) {
				btrfs_delalloc_release_space(inode,
					dirty_pages << PAGE_CACHE_SHIFT);
				btrfs_drop_pages(pages, num_pages);
				break;
			}
1004
		}
C
Chris Mason 已提交
1005 1006 1007

		btrfs_drop_pages(pages, num_pages);

J
Josef Bacik 已提交
1008 1009 1010 1011 1012 1013 1014
		cond_resched();

		balance_dirty_pages_ratelimited_nr(inode->i_mapping,
						   dirty_pages);
		if (dirty_pages < (root->leafsize >> PAGE_CACHE_SHIFT) + 1)
			btrfs_btree_balance_dirty(root, 1);
		btrfs_throttle(root);
1015

1016 1017
		pos += copied;
		num_written += copied;
J
Josef Bacik 已提交
1018
	}
C
Chris Mason 已提交
1019

J
Josef Bacik 已提交
1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062
	kfree(pages);

	return num_written ? num_written : ret;
}

static ssize_t __btrfs_direct_write(struct kiocb *iocb,
				    const struct iovec *iov,
				    unsigned long nr_segs, loff_t pos,
				    loff_t *ppos, size_t count, size_t ocount)
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = fdentry(file)->d_inode;
	struct iov_iter i;
	ssize_t written;
	ssize_t written_buffered;
	loff_t endbyte;
	int err;

	written = generic_file_direct_write(iocb, iov, &nr_segs, pos, ppos,
					    count, ocount);

	/*
	 * the generic O_DIRECT will update in-memory i_size after the
	 * DIOs are done.  But our endio handlers that update the on
	 * disk i_size never update past the in memory i_size.  So we
	 * need one more update here to catch any additions to the
	 * file
	 */
	if (inode->i_size != BTRFS_I(inode)->disk_i_size) {
		btrfs_ordered_update_i_size(inode, inode->i_size, NULL);
		mark_inode_dirty(inode);
	}

	if (written < 0 || written == count)
		return written;

	pos += written;
	count -= written;
	iov_iter_init(&i, iov, nr_segs, count, written);
	written_buffered = __btrfs_buffered_write(file, &i, pos);
	if (written_buffered < 0) {
		err = written_buffered;
		goto out;
C
Chris Mason 已提交
1063
	}
J
Josef Bacik 已提交
1064 1065 1066 1067 1068 1069 1070 1071
	endbyte = pos + written_buffered - 1;
	err = filemap_write_and_wait_range(file->f_mapping, pos, endbyte);
	if (err)
		goto out;
	written += written_buffered;
	*ppos = pos + written_buffered;
	invalidate_mapping_pages(file->f_mapping, pos >> PAGE_CACHE_SHIFT,
				 endbyte >> PAGE_CACHE_SHIFT);
C
Chris Mason 已提交
1072
out:
J
Josef Bacik 已提交
1073 1074
	return written ? written : err;
}
1075

J
Josef Bacik 已提交
1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145
static ssize_t btrfs_file_aio_write(struct kiocb *iocb,
				    const struct iovec *iov,
				    unsigned long nr_segs, loff_t pos)
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = fdentry(file)->d_inode;
	struct btrfs_root *root = BTRFS_I(inode)->root;
	loff_t *ppos = &iocb->ki_pos;
	ssize_t num_written = 0;
	ssize_t err = 0;
	size_t count, ocount;

	vfs_check_frozen(inode->i_sb, SB_FREEZE_WRITE);

	mutex_lock(&inode->i_mutex);

	err = generic_segment_checks(iov, &nr_segs, &ocount, VERIFY_READ);
	if (err) {
		mutex_unlock(&inode->i_mutex);
		goto out;
	}
	count = ocount;

	current->backing_dev_info = inode->i_mapping->backing_dev_info;
	err = generic_write_checks(file, &pos, &count, S_ISBLK(inode->i_mode));
	if (err) {
		mutex_unlock(&inode->i_mutex);
		goto out;
	}

	if (count == 0) {
		mutex_unlock(&inode->i_mutex);
		goto out;
	}

	err = file_remove_suid(file);
	if (err) {
		mutex_unlock(&inode->i_mutex);
		goto out;
	}

	/*
	 * If BTRFS flips readonly due to some impossible error
	 * (fs_info->fs_state now has BTRFS_SUPER_FLAG_ERROR),
	 * although we have opened a file as writable, we have
	 * to stop this write operation to ensure FS consistency.
	 */
	if (root->fs_info->fs_state & BTRFS_SUPER_FLAG_ERROR) {
		mutex_unlock(&inode->i_mutex);
		err = -EROFS;
		goto out;
	}

	file_update_time(file);
	BTRFS_I(inode)->sequence++;

	if (unlikely(file->f_flags & O_DIRECT)) {
		num_written = __btrfs_direct_write(iocb, iov, nr_segs,
						   pos, ppos, count, ocount);
	} else {
		struct iov_iter i;

		iov_iter_init(&i, iov, nr_segs, count, num_written);

		num_written = __btrfs_buffered_write(file, &i, pos);
		if (num_written > 0)
			*ppos = pos + num_written;
	}

	mutex_unlock(&inode->i_mutex);
1146

1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159
	/*
	 * we want to make sure fsync finds this change
	 * but we haven't joined a transaction running right now.
	 *
	 * Later on, someone is sure to update the inode and get the
	 * real transid recorded.
	 *
	 * We set last_trans now to the fs_info generation + 1,
	 * this will either be one more than the running transaction
	 * or the generation used for the next transaction if there isn't
	 * one running right now.
	 */
	BTRFS_I(inode)->last_trans = root->fs_info->generation + 1;
J
Josef Bacik 已提交
1160 1161 1162
	if (num_written > 0 || num_written == -EIOCBQUEUED) {
		err = generic_write_sync(file, pos, num_written);
		if (err < 0 && num_written > 0)
1163 1164
			num_written = err;
	}
J
Josef Bacik 已提交
1165
out:
C
Chris Mason 已提交
1166 1167 1168 1169
	current->backing_dev_info = NULL;
	return num_written ? num_written : err;
}

C
Chris Mason 已提交
1170
int btrfs_release_file(struct inode *inode, struct file *filp)
1171
{
1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183
	/*
	 * ordered_data_close is set by settattr when we are about to truncate
	 * a file from a non-zero size to a zero size.  This tries to
	 * flush down new bytes that may have been written if the
	 * application were using truncate to replace a file in place.
	 */
	if (BTRFS_I(inode)->ordered_data_close) {
		BTRFS_I(inode)->ordered_data_close = 0;
		btrfs_add_ordered_operation(NULL, BTRFS_I(inode)->root, inode);
		if (inode->i_size > BTRFS_ORDERED_OPERATIONS_FLUSH_LIMIT)
			filemap_flush(inode->i_mapping);
	}
S
Sage Weil 已提交
1184 1185
	if (filp->private_data)
		btrfs_ioctl_trans_end(filp);
1186 1187 1188
	return 0;
}

C
Chris Mason 已提交
1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199
/*
 * fsync call for both files and directories.  This logs the inode into
 * the tree log instead of forcing full commits whenever possible.
 *
 * It needs to call filemap_fdatawait so that all ordered extent updates are
 * in the metadata btree are up to date for copying to the log.
 *
 * It drops the inode mutex before doing the tree log commit.  This is an
 * important optimization for directories because holding the mutex prevents
 * new operations on the dir while we write to disk.
 */
1200
int btrfs_sync_file(struct file *file, int datasync)
C
Chris Mason 已提交
1201
{
1202
	struct dentry *dentry = file->f_path.dentry;
C
Chris Mason 已提交
1203 1204
	struct inode *inode = dentry->d_inode;
	struct btrfs_root *root = BTRFS_I(inode)->root;
1205
	int ret = 0;
C
Chris Mason 已提交
1206 1207
	struct btrfs_trans_handle *trans;

1208
	trace_btrfs_sync_file(file, datasync);
1209 1210 1211 1212 1213 1214 1215

	/* we wait first, since the writeback may change the inode */
	root->log_batch++;
	/* the VFS called filemap_fdatawrite for us */
	btrfs_wait_ordered_range(inode, 0, (u64)-1);
	root->log_batch++;

C
Chris Mason 已提交
1216
	/*
1217 1218
	 * check the transaction that last modified this inode
	 * and see if its already been committed
C
Chris Mason 已提交
1219
	 */
1220 1221
	if (!BTRFS_I(inode)->last_trans)
		goto out;
1222

1223 1224 1225 1226 1227
	/*
	 * if the last transaction that changed this file was before
	 * the current transaction, we can bail out now without any
	 * syncing
	 */
1228 1229 1230 1231 1232 1233 1234 1235 1236 1237
	mutex_lock(&root->fs_info->trans_mutex);
	if (BTRFS_I(inode)->last_trans <=
	    root->fs_info->last_trans_committed) {
		BTRFS_I(inode)->last_trans = 0;
		mutex_unlock(&root->fs_info->trans_mutex);
		goto out;
	}
	mutex_unlock(&root->fs_info->trans_mutex);

	/*
1238 1239
	 * ok we haven't committed the transaction yet, lets do a commit
	 */
1240
	if (file->private_data)
S
Sage Weil 已提交
1241 1242
		btrfs_ioctl_trans_end(file);

1243 1244 1245
	trans = btrfs_start_transaction(root, 0);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
C
Chris Mason 已提交
1246 1247
		goto out;
	}
1248

1249
	ret = btrfs_log_dentry_safe(trans, root, dentry);
C
Chris Mason 已提交
1250
	if (ret < 0)
1251
		goto out;
C
Chris Mason 已提交
1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262

	/* we've logged all the items and now have a consistent
	 * version of the file in the log.  It is possible that
	 * someone will come in and modify the file, but that's
	 * fine because the log is consistent on disk, and we
	 * have references to all of the file's extents
	 *
	 * It is possible that someone will come in and log the
	 * file again, but that will end up using the synchronization
	 * inside btrfs_sync_log to keep things safe.
	 */
1263
	mutex_unlock(&dentry->d_inode->i_mutex);
C
Chris Mason 已提交
1264

1265 1266
	if (ret != BTRFS_NO_LOG_SYNC) {
		if (ret > 0) {
1267
			ret = btrfs_commit_transaction(trans, root);
1268 1269 1270 1271 1272 1273 1274 1275 1276
		} else {
			ret = btrfs_sync_log(trans, root);
			if (ret == 0)
				ret = btrfs_end_transaction(trans, root);
			else
				ret = btrfs_commit_transaction(trans, root);
		}
	} else {
		ret = btrfs_end_transaction(trans, root);
1277
	}
1278
	mutex_lock(&dentry->d_inode->i_mutex);
C
Chris Mason 已提交
1279
out:
1280
	return ret > 0 ? -EIO : ret;
C
Chris Mason 已提交
1281 1282
}

1283
static const struct vm_operations_struct btrfs_file_vm_ops = {
1284
	.fault		= filemap_fault,
C
Chris Mason 已提交
1285 1286 1287 1288 1289
	.page_mkwrite	= btrfs_page_mkwrite,
};

static int btrfs_file_mmap(struct file	*filp, struct vm_area_struct *vma)
{
M
Miao Xie 已提交
1290 1291 1292 1293 1294
	struct address_space *mapping = filp->f_mapping;

	if (!mapping->a_ops->readpage)
		return -ENOEXEC;

C
Chris Mason 已提交
1295
	file_accessed(filp);
M
Miao Xie 已提交
1296 1297 1298
	vma->vm_ops = &btrfs_file_vm_ops;
	vma->vm_flags |= VM_CAN_NONLINEAR;

C
Chris Mason 已提交
1299 1300 1301
	return 0;
}

1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335
static long btrfs_fallocate(struct file *file, int mode,
			    loff_t offset, loff_t len)
{
	struct inode *inode = file->f_path.dentry->d_inode;
	struct extent_state *cached_state = NULL;
	u64 cur_offset;
	u64 last_byte;
	u64 alloc_start;
	u64 alloc_end;
	u64 alloc_hint = 0;
	u64 locked_end;
	u64 mask = BTRFS_I(inode)->root->sectorsize - 1;
	struct extent_map *em;
	int ret;

	alloc_start = offset & ~mask;
	alloc_end =  (offset + len + mask) & ~mask;

	/* We only support the FALLOC_FL_KEEP_SIZE mode */
	if (mode & ~FALLOC_FL_KEEP_SIZE)
		return -EOPNOTSUPP;

	/*
	 * wait for ordered IO before we have any locks.  We'll loop again
	 * below with the locks held.
	 */
	btrfs_wait_ordered_range(inode, alloc_start, alloc_end - alloc_start);

	mutex_lock(&inode->i_mutex);
	ret = inode_newsize_ok(inode, alloc_end);
	if (ret)
		goto out;

	if (alloc_start > inode->i_size) {
1336 1337
		ret = btrfs_cont_expand(inode, i_size_read(inode),
					alloc_start);
1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413
		if (ret)
			goto out;
	}

	ret = btrfs_check_data_free_space(inode, alloc_end - alloc_start);
	if (ret)
		goto out;

	locked_end = alloc_end - 1;
	while (1) {
		struct btrfs_ordered_extent *ordered;

		/* the extent lock is ordered inside the running
		 * transaction
		 */
		lock_extent_bits(&BTRFS_I(inode)->io_tree, alloc_start,
				 locked_end, 0, &cached_state, GFP_NOFS);
		ordered = btrfs_lookup_first_ordered_extent(inode,
							    alloc_end - 1);
		if (ordered &&
		    ordered->file_offset + ordered->len > alloc_start &&
		    ordered->file_offset < alloc_end) {
			btrfs_put_ordered_extent(ordered);
			unlock_extent_cached(&BTRFS_I(inode)->io_tree,
					     alloc_start, locked_end,
					     &cached_state, GFP_NOFS);
			/*
			 * we can't wait on the range with the transaction
			 * running or with the extent lock held
			 */
			btrfs_wait_ordered_range(inode, alloc_start,
						 alloc_end - alloc_start);
		} else {
			if (ordered)
				btrfs_put_ordered_extent(ordered);
			break;
		}
	}

	cur_offset = alloc_start;
	while (1) {
		em = btrfs_get_extent(inode, NULL, 0, cur_offset,
				      alloc_end - cur_offset, 0);
		BUG_ON(IS_ERR(em) || !em);
		last_byte = min(extent_map_end(em), alloc_end);
		last_byte = (last_byte + mask) & ~mask;
		if (em->block_start == EXTENT_MAP_HOLE ||
		    (cur_offset >= inode->i_size &&
		     !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))) {
			ret = btrfs_prealloc_file_range(inode, mode, cur_offset,
							last_byte - cur_offset,
							1 << inode->i_blkbits,
							offset + len,
							&alloc_hint);
			if (ret < 0) {
				free_extent_map(em);
				break;
			}
		}
		free_extent_map(em);

		cur_offset = last_byte;
		if (cur_offset >= alloc_end) {
			ret = 0;
			break;
		}
	}
	unlock_extent_cached(&BTRFS_I(inode)->io_tree, alloc_start, locked_end,
			     &cached_state, GFP_NOFS);

	btrfs_free_reserved_data_space(inode, alloc_end - alloc_start);
out:
	mutex_unlock(&inode->i_mutex);
	return ret;
}

1414
const struct file_operations btrfs_file_operations = {
C
Chris Mason 已提交
1415 1416
	.llseek		= generic_file_llseek,
	.read		= do_sync_read,
1417
	.write		= do_sync_write,
C
Chris Mason 已提交
1418
	.aio_read       = generic_file_aio_read,
C
Chris Mason 已提交
1419
	.splice_read	= generic_file_splice_read,
1420
	.aio_write	= btrfs_file_aio_write,
C
Chris Mason 已提交
1421
	.mmap		= btrfs_file_mmap,
C
Chris Mason 已提交
1422
	.open		= generic_file_open,
1423
	.release	= btrfs_release_file,
C
Chris Mason 已提交
1424
	.fsync		= btrfs_sync_file,
1425
	.fallocate	= btrfs_fallocate,
C
Christoph Hellwig 已提交
1426
	.unlocked_ioctl	= btrfs_ioctl,
C
Chris Mason 已提交
1427
#ifdef CONFIG_COMPAT
C
Christoph Hellwig 已提交
1428
	.compat_ioctl	= btrfs_ioctl,
C
Chris Mason 已提交
1429 1430
#endif
};