file.c 20.9 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 27 28 29 30 31
#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>
32
#include <linux/version.h>
C
Chris Mason 已提交
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82
#include "ctree.h"
#include "disk-io.h"
#include "transaction.h"
#include "btrfs_inode.h"
#include "ioctl.h"
#include "print-tree.h"


static int btrfs_copy_from_user(loff_t pos, int num_pages, int write_bytes,
				struct page **prepared_pages,
				const char __user * buf)
{
	long page_fault = 0;
	int i;
	int offset = pos & (PAGE_CACHE_SIZE - 1);

	for (i = 0; i < num_pages && write_bytes > 0; i++, offset = 0) {
		size_t count = min_t(size_t,
				     PAGE_CACHE_SIZE - offset, write_bytes);
		struct page *page = prepared_pages[i];
		fault_in_pages_readable(buf, count);

		/* Copy data from userspace to the current page */
		kmap(page);
		page_fault = __copy_from_user(page_address(page) + offset,
					      buf, count);
		/* Flush processor's dcache for this page */
		flush_dcache_page(page);
		kunmap(page);
		buf += count;
		write_bytes -= count;

		if (page_fault)
			break;
	}
	return page_fault ? -EFAULT : 0;
}

static void btrfs_drop_pages(struct page **pages, size_t num_pages)
{
	size_t i;
	for (i = 0; i < num_pages; i++) {
		if (!pages[i])
			break;
		unlock_page(pages[i]);
		mark_page_accessed(pages[i]);
		page_cache_release(pages[i]);
	}
}

83 84
static int insert_inline_extent(struct btrfs_trans_handle *trans,
				struct btrfs_root *root, struct inode *inode,
85 86 87
				u64 offset, size_t size,
				struct page **pages, size_t page_offset,
				int num_pages)
88 89 90
{
	struct btrfs_key key;
	struct btrfs_path *path;
91 92 93
	struct extent_buffer *leaf;
	char *kaddr;
	unsigned long ptr;
94
	struct btrfs_file_extent_item *ei;
95
	struct page *page;
96 97 98
	u32 datasize;
	int err = 0;
	int ret;
99 100
	int i;
	ssize_t cur_size;
101 102 103 104 105 106 107 108 109 110

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

	btrfs_set_trans_block_group(trans, inode);

	key.objectid = inode->i_ino;
	key.offset = offset;
	btrfs_set_key_type(&key, BTRFS_EXTENT_DATA_KEY);
111
	datasize = btrfs_file_extent_calc_inline_size(offset + size);
112

113 114
	ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
	if (ret < 0) {
115 116 117
		err = ret;
		goto fail;
	}
118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201
	if (ret == 1) {
		path->slots[0]--;
		leaf = path->nodes[0];
		ei = btrfs_item_ptr(leaf, path->slots[0],
				    struct btrfs_file_extent_item);

		if (btrfs_file_extent_type(leaf, ei) !=
		    BTRFS_FILE_EXTENT_INLINE) {
			goto insert;
		}
		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
		ret = 0;
	}
	if (ret == 0) {
		u32 found_size;
		u64 found_start;

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

		if (btrfs_file_extent_type(leaf, ei) !=
		    BTRFS_FILE_EXTENT_INLINE) {
			err = ret;
			btrfs_print_leaf(root, leaf);
			printk("found wasn't inline offset %Lu inode %lu\n",
			       offset, inode->i_ino);
			goto fail;
		}
		found_start = key.offset;
		found_size = btrfs_file_extent_inline_len(leaf,
					  btrfs_item_nr(leaf, path->slots[0]));

		if (found_size < offset + size) {
			btrfs_release_path(root, path);
			ret = btrfs_search_slot(trans, root, &key, path,
						offset + size - found_size -
						found_start, 1);
			BUG_ON(ret != 0);
			ret = btrfs_extend_item(trans, root, path,
						offset + size - found_size -
						found_start);
			if (ret) {
				err = ret;
				goto fail;
			}
			leaf = path->nodes[0];
			ei = btrfs_item_ptr(leaf, path->slots[0],
					    struct btrfs_file_extent_item);
		}
	} else {
insert:
		btrfs_release_path(root, path);
		ret = btrfs_insert_empty_item(trans, root, path, &key,
					      datasize);
		if (ret) {
			err = ret;
			printk("got bad ret %d\n", ret);
			goto fail;
		}
		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, BTRFS_FILE_EXTENT_INLINE);
	}
	ptr = btrfs_file_extent_inline_start(ei) + offset;

	cur_size = size;
	i = 0;
	while (size > 0) {
		page = pages[i];
		kaddr = kmap_atomic(page, KM_USER0);
		cur_size = min(PAGE_CACHE_SIZE - page_offset, size);
		write_extent_buffer(leaf, kaddr + page_offset, ptr, cur_size);
		kunmap_atomic(kaddr, KM_USER0);
		page_offset = 0;
		ptr += cur_size;
		size -= cur_size;
		if (i >= num_pages) {
			printk("i %d num_pages %d\n", i, num_pages);
		}
		i++;
	}
202
	btrfs_mark_buffer_dirty(leaf);
203 204 205 206 207
fail:
	btrfs_free_path(path);
	return err;
}

C
Chris Mason 已提交
208 209 210 211 212 213 214 215 216
static int dirty_and_release_pages(struct btrfs_trans_handle *trans,
				   struct btrfs_root *root,
				   struct file *file,
				   struct page **pages,
				   size_t num_pages,
				   loff_t pos,
				   size_t write_bytes)
{
	int err = 0;
217
	int i;
C
Chris Mason 已提交
218
	struct inode *inode = file->f_path.dentry->d_inode;
219 220
	struct extent_map *em;
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
221 222
	u64 hint_byte;
	u64 num_bytes;
223 224 225
	u64 start_pos;
	u64 end_of_last_block;
	u64 end_pos = pos + write_bytes;
226
	u32 inline_size;
227
	loff_t isize = i_size_read(inode);
C
Chris Mason 已提交
228

229 230 231
	em = alloc_extent_map(GFP_NOFS);
	if (!em)
		return -ENOMEM;
C
Chris Mason 已提交
232

233
	em->bdev = inode->i_sb->s_bdev;
C
Chris Mason 已提交
234

235
	start_pos = pos & ~((u64)root->sectorsize - 1);
236 237
	num_bytes = (write_bytes + pos - start_pos +
		    root->sectorsize - 1) & ~((u64)root->sectorsize - 1);
C
Chris Mason 已提交
238

239
	down_read(&BTRFS_I(inode)->root->snap_sem);
240 241
	end_of_last_block = start_pos + num_bytes - 1;

242
	lock_extent(em_tree, start_pos, end_of_last_block, GFP_NOFS);
243 244 245 246 247 248 249
	mutex_lock(&root->fs_info->fs_mutex);
	trans = btrfs_start_transaction(root, 1);
	if (!trans) {
		err = -ENOMEM;
		goto out_unlock;
	}
	btrfs_set_trans_block_group(trans, inode);
250 251
	inode->i_blocks += num_bytes >> 9;
	hint_byte = 0;
252 253

	if ((end_of_last_block & 4095) == 0) {
254
		printk("strange end of last %Lu %zu %Lu\n", start_pos, write_bytes, end_of_last_block);
255 256 257 258 259 260 261 262 263
	}
	set_extent_uptodate(em_tree, start_pos, end_of_last_block, GFP_NOFS);

	/* FIXME...EIEIO, ENOSPC and more */

	/* insert any holes we need to create */
	if (inode->i_size < start_pos) {
		u64 last_pos_in_file;
		u64 hole_size;
264
		u64 mask = root->sectorsize - 1;
265 266
		last_pos_in_file = (isize + mask) & ~mask;
		hole_size = (start_pos - last_pos_in_file + mask) & ~mask;
C
Chris Mason 已提交
267

268
		if (last_pos_in_file < start_pos) {
C
Chris Mason 已提交
269 270 271
			err = btrfs_drop_extents(trans, root, inode,
						 last_pos_in_file,
						 last_pos_in_file + hole_size,
272
						 last_pos_in_file,
273
						 &hint_byte);
C
Chris Mason 已提交
274 275 276
			if (err)
				goto failed;

277 278 279 280 281 282
			err = btrfs_insert_file_extent(trans, root,
						       inode->i_ino,
						       last_pos_in_file,
						       0, 0, hole_size);
		}
		if (err)
C
Chris Mason 已提交
283
			goto failed;
284 285 286 287 288 289
	}

	/*
	 * either allocate an extent for the new bytes or setup the key
	 * to show we are doing inline data in the extent
	 */
290 291
	inline_size = end_pos;
	if (isize >= BTRFS_MAX_INLINE_DATA_SIZE(root) ||
292
	    inline_size > 16384 ||
293
	    inline_size >= BTRFS_MAX_INLINE_DATA_SIZE(root)) {
294
		u64 last_end;
295

296 297 298
		for (i = 0; i < num_pages; i++) {
			struct page *p = pages[i];
			SetPageUptodate(p);
299
			set_page_dirty(p);
C
Chris Mason 已提交
300
		}
301 302 303 304
		last_end = pages[num_pages -1]->index << PAGE_CACHE_SHIFT;
		last_end += PAGE_CACHE_SIZE - 1;
		set_extent_delalloc(em_tree, start_pos, end_of_last_block,
				 GFP_NOFS);
305
	} else {
306
		u64 aligned_end;
307
		/* step one, delete the existing extents in this range */
308 309
		aligned_end = (pos + write_bytes + root->sectorsize - 1) &
			~((u64)root->sectorsize - 1);
C
Chris Mason 已提交
310
		err = btrfs_drop_extents(trans, root, inode, start_pos,
311
					 aligned_end, end_pos, &hint_byte);
C
Chris Mason 已提交
312 313
		if (err)
			goto failed;
314
		err = insert_inline_extent(trans, root, inode, start_pos,
315 316
					   end_pos - start_pos, pages, 0,
					   num_pages);
317 318 319 320 321
		BUG_ON(err);
	}
	if (end_pos > isize) {
		i_size_write(inode, end_pos);
		btrfs_update_inode(trans, root, inode);
C
Chris Mason 已提交
322 323
	}
failed:
324 325 326
	err = btrfs_end_transaction(trans, root);
out_unlock:
	mutex_unlock(&root->fs_info->fs_mutex);
327
	unlock_extent(em_tree, start_pos, end_of_last_block, GFP_NOFS);
328
	free_extent_map(em);
329
	up_read(&BTRFS_I(inode)->root->snap_sem);
C
Chris Mason 已提交
330 331 332
	return err;
}

333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350
int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end)
{
	struct extent_map *em;
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;

	while(1) {
		em = lookup_extent_mapping(em_tree, start, end);
		if (!em)
			break;
		remove_extent_mapping(em_tree, em);
		/* once for us */
		free_extent_map(em);
		/* once for the tree*/
		free_extent_map(em);
	}
	return 0;
}

C
Chris Mason 已提交
351 352 353 354 355 356 357 358 359 360 361
/*
 * 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.
 */
int btrfs_drop_extents(struct btrfs_trans_handle *trans,
		       struct btrfs_root *root, struct inode *inode,
362
		       u64 start, u64 end, u64 inline_end, u64 *hint_byte)
C
Chris Mason 已提交
363 364 365
{
	int ret;
	struct btrfs_key key;
366
	struct extent_buffer *leaf;
C
Chris Mason 已提交
367 368 369 370 371 372 373 374 375 376 377
	int slot;
	struct btrfs_file_extent_item *extent;
	u64 extent_end = 0;
	int keep;
	struct btrfs_file_extent_item old;
	struct btrfs_path *path;
	u64 search_start = start;
	int bookend;
	int found_type;
	int found_extent;
	int found_inline;
C
Chris Mason 已提交
378
	int recow;
C
Chris Mason 已提交
379

380 381
	btrfs_drop_extent_cache(inode, start, end - 1);

C
Chris Mason 已提交
382 383 384 385
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;
	while(1) {
C
Chris Mason 已提交
386
		recow = 0;
C
Chris Mason 已提交
387 388 389 390 391 392 393 394 395 396 397 398
		btrfs_release_path(root, path);
		ret = btrfs_lookup_file_extent(trans, root, path, inode->i_ino,
					       search_start, -1);
		if (ret < 0)
			goto out;
		if (ret > 0) {
			if (path->slots[0] == 0) {
				ret = 0;
				goto out;
			}
			path->slots[0]--;
		}
399
next_slot:
C
Chris Mason 已提交
400 401 402 403 404
		keep = 0;
		bookend = 0;
		found_extent = 0;
		found_inline = 0;
		extent = NULL;
405
		leaf = path->nodes[0];
C
Chris Mason 已提交
406
		slot = path->slots[0];
407
		ret = 0;
408
		btrfs_item_key_to_cpu(leaf, &key, slot);
C
Chris Mason 已提交
409 410 411
		if (key.offset >= end || key.objectid != inode->i_ino) {
			goto out;
		}
412
		if (btrfs_key_type(&key) > BTRFS_EXTENT_DATA_KEY) {
C
Chris Mason 已提交
413 414
			goto out;
		}
C
Chris Mason 已提交
415 416 417 418
		if (recow) {
			search_start = key.offset;
			continue;
		}
419 420 421
		if (btrfs_key_type(&key) == BTRFS_EXTENT_DATA_KEY) {
			extent = btrfs_item_ptr(leaf, slot,
						struct btrfs_file_extent_item);
422
			found_type = btrfs_file_extent_type(leaf, extent);
423 424
			if (found_type == BTRFS_FILE_EXTENT_REG) {
				extent_end = key.offset +
425
				     btrfs_file_extent_num_bytes(leaf, extent);
426 427
				found_extent = 1;
			} else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
428 429
				struct btrfs_item *item;
				item = btrfs_item_nr(leaf, slot);
430 431
				found_inline = 1;
				extent_end = key.offset +
432
				     btrfs_file_extent_inline_len(leaf, item);
433 434 435
			}
		} else {
			extent_end = search_start;
C
Chris Mason 已提交
436 437 438
		}

		/* we found nothing we can drop */
439 440 441 442
		if ((!found_extent && !found_inline) ||
		    search_start >= extent_end) {
			int nextret;
			u32 nritems;
443
			nritems = btrfs_header_nritems(leaf);
444 445 446 447
			if (slot >= nritems - 1) {
				nextret = btrfs_next_leaf(root, path);
				if (nextret)
					goto out;
C
Chris Mason 已提交
448
				recow = 1;
449 450 451 452
			} else {
				path->slots[0]++;
			}
			goto next_slot;
C
Chris Mason 已提交
453 454 455 456
		}

		/* FIXME, there's only one inline extent allowed right now */
		if (found_inline) {
457
			u64 mask = root->sectorsize - 1;
C
Chris Mason 已提交
458 459 460 461 462 463
			search_start = (extent_end + mask) & ~mask;
		} else
			search_start = extent_end;

		if (end < extent_end && end >= key.offset) {
			if (found_extent) {
464 465 466 467
				u64 disk_bytenr =
				    btrfs_file_extent_disk_bytenr(leaf, extent);
				u64 disk_num_bytes =
				    btrfs_file_extent_disk_num_bytes(leaf,
468 469 470 471
								      extent);
				read_extent_buffer(leaf, &old,
						   (unsigned long)extent,
						   sizeof(old));
472
				if (disk_bytenr != 0) {
C
Chris Mason 已提交
473
					ret = btrfs_inc_extent_ref(trans, root,
474
					         disk_bytenr, disk_num_bytes);
C
Chris Mason 已提交
475 476 477
					BUG_ON(ret);
				}
			}
478 479
			if (!found_inline)
				bookend = 1;
C
Chris Mason 已提交
480 481 482 483 484 485
		}
		/* truncate existing extent */
		if (start > key.offset) {
			u64 new_num;
			u64 old_num;
			keep = 1;
486
			WARN_ON(start & (root->sectorsize - 1));
C
Chris Mason 已提交
487
			if (found_extent) {
488 489 490 491 492 493 494 495
				new_num = start - key.offset;
				old_num = btrfs_file_extent_num_bytes(leaf,
								      extent);
				*hint_byte =
					btrfs_file_extent_disk_bytenr(leaf,
								      extent);
				if (btrfs_file_extent_disk_bytenr(leaf,
								  extent)) {
C
Chris Mason 已提交
496
					inode->i_blocks -=
497
						(old_num - new_num) >> 9;
C
Chris Mason 已提交
498
				}
499 500
				btrfs_set_file_extent_num_bytes(leaf, extent,
								new_num);
501
				btrfs_mark_buffer_dirty(leaf);
502 503 504 505 506 507 508 509
			} else if (end > extent_end &&
				   key.offset < inline_end &&
				   inline_end < extent_end) {
				u32 new_size;
				new_size = btrfs_file_extent_calc_inline_size(
						   inline_end - key.offset);
				btrfs_truncate_item(trans, root, path,
						    new_size);
C
Chris Mason 已提交
510 511 512 513
			}
		}
		/* delete the entire extent */
		if (!keep) {
514 515 516
			u64 disk_bytenr = 0;
			u64 disk_num_bytes = 0;
			u64 extent_num_bytes = 0;
C
Chris Mason 已提交
517
			if (found_extent) {
518 519
				disk_bytenr =
				      btrfs_file_extent_disk_bytenr(leaf,
520
								     extent);
521 522
				disk_num_bytes =
				      btrfs_file_extent_disk_num_bytes(leaf,
523
								       extent);
524 525 526 527 528
				extent_num_bytes =
				      btrfs_file_extent_num_bytes(leaf, extent);
				*hint_byte =
					btrfs_file_extent_disk_bytenr(leaf,
								      extent);
C
Chris Mason 已提交
529 530
			}
			ret = btrfs_del_item(trans, root, path);
531
			/* TODO update progress marker and return */
C
Chris Mason 已提交
532 533 534
			BUG_ON(ret);
			btrfs_release_path(root, path);
			extent = NULL;
535 536
			if (found_extent && disk_bytenr != 0) {
				inode->i_blocks -= extent_num_bytes >> 9;
C
Chris Mason 已提交
537
				ret = btrfs_free_extent(trans, root,
538 539
							disk_bytenr,
							disk_num_bytes, 0);
C
Chris Mason 已提交
540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558
			}

			BUG_ON(ret);
			if (!bookend && search_start >= end) {
				ret = 0;
				goto out;
			}
			if (!bookend)
				continue;
		}
		/* create bookend, splitting the extent in two */
		if (bookend && found_extent) {
			struct btrfs_key ins;
			ins.objectid = inode->i_ino;
			ins.offset = end;
			btrfs_set_key_type(&ins, BTRFS_EXTENT_DATA_KEY);
			btrfs_release_path(root, path);
			ret = btrfs_insert_empty_item(trans, root, path, &ins,
						      sizeof(*extent));
559

560
			leaf = path->nodes[0];
561
			if (ret) {
562 563
				btrfs_print_leaf(root, leaf);
				printk("got %d on inserting %Lu %u %Lu start %Lu end %Lu found %Lu %Lu keep was %d\n", ret , ins.objectid, ins.type, ins.offset, start, end, key.offset, extent_end, keep);
564
			}
C
Chris Mason 已提交
565
			BUG_ON(ret);
566 567 568 569 570 571
			extent = btrfs_item_ptr(leaf, path->slots[0],
						struct btrfs_file_extent_item);
			write_extent_buffer(leaf, &old,
					    (unsigned long)extent, sizeof(old));

			btrfs_set_file_extent_offset(leaf, extent,
572 573 574 575 576
				    le64_to_cpu(old.offset) + end - key.offset);
			WARN_ON(le64_to_cpu(old.num_bytes) <
				(extent_end - end));
			btrfs_set_file_extent_num_bytes(leaf, extent,
							extent_end - end);
577
			btrfs_set_file_extent_type(leaf, extent,
C
Chris Mason 已提交
578
						   BTRFS_FILE_EXTENT_REG);
579

C
Chris Mason 已提交
580
			btrfs_mark_buffer_dirty(path->nodes[0]);
581
			if (le64_to_cpu(old.disk_bytenr) != 0) {
C
Chris Mason 已提交
582
				inode->i_blocks +=
583 584
				      btrfs_file_extent_num_bytes(leaf,
								  extent) >> 9;
C
Chris Mason 已提交
585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604
			}
			ret = 0;
			goto out;
		}
	}
out:
	btrfs_free_path(path);
	return ret;
}

/*
 * this gets pages into the page cache and locks them down
 */
static int prepare_pages(struct btrfs_root *root,
			 struct file *file,
			 struct page **pages,
			 size_t num_pages,
			 loff_t pos,
			 unsigned long first_index,
			 unsigned long last_index,
605
			 size_t write_bytes)
C
Chris Mason 已提交
606 607 608 609 610
{
	int i;
	unsigned long index = pos >> PAGE_CACHE_SHIFT;
	struct inode *inode = file->f_path.dentry->d_inode;
	int err = 0;
611 612
	u64 start_pos;

613
	start_pos = pos & ~((u64)root->sectorsize - 1);
C
Chris Mason 已提交
614 615 616 617 618 619 620

	memset(pages, 0, num_pages * sizeof(struct page *));

	for (i = 0; i < num_pages; i++) {
		pages[i] = grab_cache_page(inode->i_mapping, index + i);
		if (!pages[i]) {
			err = -ENOMEM;
621
			BUG_ON(1);
C
Chris Mason 已提交
622
		}
C
Chris Mason 已提交
623 624
		cancel_dirty_page(pages[i], PAGE_CACHE_SIZE);
		wait_on_page_writeback(pages[i]);
625
		set_page_extent_mapped(pages[i]);
626
		WARN_ON(!PageLocked(pages[i]));
C
Chris Mason 已提交
627 628 629 630 631 632 633 634 635 636 637 638 639
	}
	return 0;
}

static ssize_t btrfs_file_write(struct file *file, const char __user *buf,
				size_t count, loff_t *ppos)
{
	loff_t pos;
	size_t num_written = 0;
	int err = 0;
	int ret = 0;
	struct inode *inode = file->f_path.dentry->d_inode;
	struct btrfs_root *root = BTRFS_I(inode)->root;
640 641
	struct page **pages = NULL;
	int nrptrs;
C
Chris Mason 已提交
642 643 644
	struct page *pinned[2];
	unsigned long first_index;
	unsigned long last_index;
645 646 647

	nrptrs = min((count + PAGE_CACHE_SIZE - 1) / PAGE_CACHE_SIZE,
		     PAGE_CACHE_SIZE / (sizeof(struct page *)));
C
Chris Mason 已提交
648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664
	pinned[0] = NULL;
	pinned[1] = NULL;
	if (file->f_flags & O_DIRECT)
		return -EINVAL;
	pos = *ppos;
	vfs_check_frozen(inode->i_sb, SB_FREEZE_WRITE);
	current->backing_dev_info = inode->i_mapping->backing_dev_info;
	err = generic_write_checks(file, &pos, &count, S_ISBLK(inode->i_mode));
	if (err)
		goto out;
	if (count == 0)
		goto out;
	err = remove_suid(file->f_path.dentry);
	if (err)
		goto out;
	file_update_time(file);

665
	pages = kmalloc(nrptrs * sizeof(struct page *), GFP_KERNEL);
C
Chris Mason 已提交
666 667 668 669 670 671 672 673 674 675 676 677 678

	mutex_lock(&inode->i_mutex);
	first_index = pos >> PAGE_CACHE_SHIFT;
	last_index = (pos + count) >> PAGE_CACHE_SHIFT;

	/*
	 * there are lots of better ways to do this, but this code
	 * makes sure the first and last page in the file range are
	 * up to date and ready for cow
	 */
	if ((pos & (PAGE_CACHE_SIZE - 1))) {
		pinned[0] = grab_cache_page(inode->i_mapping, first_index);
		if (!PageUptodate(pinned[0])) {
C
Chris Mason 已提交
679
			ret = btrfs_readpage(NULL, pinned[0]);
C
Chris Mason 已提交
680 681 682 683 684 685 686 687 688
			BUG_ON(ret);
			wait_on_page_locked(pinned[0]);
		} else {
			unlock_page(pinned[0]);
		}
	}
	if ((pos + count) & (PAGE_CACHE_SIZE - 1)) {
		pinned[1] = grab_cache_page(inode->i_mapping, last_index);
		if (!PageUptodate(pinned[1])) {
C
Chris Mason 已提交
689
			ret = btrfs_readpage(NULL, pinned[1]);
C
Chris Mason 已提交
690 691 692 693 694 695 696 697 698
			BUG_ON(ret);
			wait_on_page_locked(pinned[1]);
		} else {
			unlock_page(pinned[1]);
		}
	}

	while(count > 0) {
		size_t offset = pos & (PAGE_CACHE_SIZE - 1);
699 700
		size_t write_bytes = min(count, nrptrs *
					(size_t)PAGE_CACHE_SIZE -
701
					 offset);
C
Chris Mason 已提交
702 703 704
		size_t num_pages = (write_bytes + PAGE_CACHE_SIZE - 1) >>
					PAGE_CACHE_SHIFT;

705
		WARN_ON(num_pages > nrptrs);
C
Chris Mason 已提交
706 707 708
		memset(pages, 0, sizeof(pages));
		ret = prepare_pages(root, file, pages, num_pages,
				    pos, first_index, last_index,
709
				    write_bytes);
710 711
		if (ret)
			goto out;
C
Chris Mason 已提交
712 713 714

		ret = btrfs_copy_from_user(pos, num_pages,
					   write_bytes, pages, buf);
715 716 717 718
		if (ret) {
			btrfs_drop_pages(pages, num_pages);
			goto out;
		}
C
Chris Mason 已提交
719 720 721 722

		ret = dirty_and_release_pages(NULL, root, file, pages,
					      num_pages, pos, write_bytes);
		btrfs_drop_pages(pages, num_pages);
723 724
		if (ret)
			goto out;
C
Chris Mason 已提交
725 726 727 728 729 730

		buf += write_bytes;
		count -= write_bytes;
		pos += write_bytes;
		num_written += write_bytes;

731
		balance_dirty_pages_ratelimited_nr(inode->i_mapping, num_pages);
732
		btrfs_btree_balance_dirty(root, 1);
C
Chris Mason 已提交
733 734 735 736
		cond_resched();
	}
	mutex_unlock(&inode->i_mutex);
out:
737
	kfree(pages);
C
Chris Mason 已提交
738 739 740 741 742 743 744 745 746 747 748 749 750 751
	if (pinned[0])
		page_cache_release(pinned[0]);
	if (pinned[1])
		page_cache_release(pinned[1]);
	*ppos = pos;
	current->backing_dev_info = NULL;
	return num_written ? num_written : err;
}

static int btrfs_sync_file(struct file *file,
			   struct dentry *dentry, int datasync)
{
	struct inode *inode = dentry->d_inode;
	struct btrfs_root *root = BTRFS_I(inode)->root;
752
	int ret = 0;
C
Chris Mason 已提交
753 754 755
	struct btrfs_trans_handle *trans;

	/*
756 757
	 * check the transaction that last modified this inode
	 * and see if its already been committed
C
Chris Mason 已提交
758 759
	 */
	mutex_lock(&root->fs_info->fs_mutex);
760 761 762 763 764 765 766 767 768 769 770 771
	if (!BTRFS_I(inode)->last_trans)
		goto out;
	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);

	/*
772 773
	 * ok we haven't committed the transaction yet, lets do a commit
	 */
C
Chris Mason 已提交
774 775 776 777 778 779 780
	trans = btrfs_start_transaction(root, 1);
	if (!trans) {
		ret = -ENOMEM;
		goto out;
	}
	ret = btrfs_commit_transaction(trans, root);
out:
781
	mutex_unlock(&root->fs_info->fs_mutex);
C
Chris Mason 已提交
782 783 784
	return ret > 0 ? EIO : ret;
}

C
Chris Mason 已提交
785
static struct vm_operations_struct btrfs_file_vm_ops = {
786 787 788 789 790 791
#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23)
	.nopage         = filemap_nopage,
	.populate       = filemap_populate,
#else
	.fault		= filemap_fault,
#endif
C
Chris Mason 已提交
792 793 794 795 796 797 798 799 800 801
	.page_mkwrite	= btrfs_page_mkwrite,
};

static int btrfs_file_mmap(struct file	*filp, struct vm_area_struct *vma)
{
	vma->vm_ops = &btrfs_file_vm_ops;
	file_accessed(filp);
	return 0;
}

C
Chris Mason 已提交
802 803 804
struct file_operations btrfs_file_operations = {
	.llseek		= generic_file_llseek,
	.read		= do_sync_read,
C
Chris Mason 已提交
805
	.aio_read       = generic_file_aio_read,
C
Chris Mason 已提交
806
	.write		= btrfs_file_write,
C
Chris Mason 已提交
807
	.mmap		= btrfs_file_mmap,
C
Chris Mason 已提交
808 809
	.open		= generic_file_open,
	.fsync		= btrfs_sync_file,
C
Christoph Hellwig 已提交
810
	.unlocked_ioctl	= btrfs_ioctl,
C
Chris Mason 已提交
811
#ifdef CONFIG_COMPAT
C
Christoph Hellwig 已提交
812
	.compat_ioctl	= btrfs_ioctl,
C
Chris Mason 已提交
813 814 815
#endif
};