filemap.c 65.7 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/*
 *	linux/mm/filemap.c
 *
 * Copyright (C) 1994-1999  Linus Torvalds
 */

/*
 * This file handles the generic file mmap semantics used by
 * most "normal" filesystems (but you don't /have/ to use this:
 * the NFS filesystem used to do this differently, for example)
 */
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/compiler.h>
#include <linux/fs.h>
16
#include <linux/uaccess.h>
L
Linus Torvalds 已提交
17
#include <linux/aio.h>
18
#include <linux/capability.h>
L
Linus Torvalds 已提交
19 20 21 22 23 24 25 26 27
#include <linux/kernel_stat.h>
#include <linux/mm.h>
#include <linux/swap.h>
#include <linux/mman.h>
#include <linux/pagemap.h>
#include <linux/file.h>
#include <linux/uio.h>
#include <linux/hash.h>
#include <linux/writeback.h>
28
#include <linux/backing-dev.h>
L
Linus Torvalds 已提交
29 30 31 32
#include <linux/pagevec.h>
#include <linux/blkdev.h>
#include <linux/security.h>
#include <linux/syscalls.h>
33
#include <linux/cpuset.h>
N
Nick Piggin 已提交
34
#include <linux/hardirq.h> /* for BUG_ON(!in_atomic()) only */
35
#include <linux/memcontrol.h>
36
#include <linux/mm_inline.h> /* for page_is_file_cache() */
37 38
#include "internal.h"

L
Linus Torvalds 已提交
39 40 41 42 43 44 45
/*
 * FIXME: remove all knowledge of the buffer layer from the core VM
 */
#include <linux/buffer_head.h> /* for generic_osync_inode */

#include <asm/mman.h>

46

L
Linus Torvalds 已提交
47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63
/*
 * Shared mappings implemented 30.11.1994. It's not fully working yet,
 * though.
 *
 * Shared mappings now work. 15.8.1995  Bruno.
 *
 * finished 'unifying' the page and buffer cache and SMP-threaded the
 * page-cache, 21.05.1999, Ingo Molnar <mingo@redhat.com>
 *
 * SMP-threaded pagemap-LRU 1999, Andrea Arcangeli <andrea@suse.de>
 */

/*
 * Lock ordering:
 *
 *  ->i_mmap_lock		(vmtruncate)
 *    ->private_lock		(__free_pte->__set_page_dirty_buffers)
64 65
 *      ->swap_lock		(exclusive_swap_page, others)
 *        ->mapping->tree_lock
L
Linus Torvalds 已提交
66
 *
67
 *  ->i_mutex
L
Linus Torvalds 已提交
68 69 70 71
 *    ->i_mmap_lock		(truncate->unmap_mapping_range)
 *
 *  ->mmap_sem
 *    ->i_mmap_lock
72
 *      ->page_table_lock or pte_lock	(various, mainly in memory.c)
L
Linus Torvalds 已提交
73 74 75 76 77
 *        ->mapping->tree_lock	(arch-dependent flush_dcache_mmap_lock)
 *
 *  ->mmap_sem
 *    ->lock_page		(access_process_vm)
 *
78 79
 *  ->i_mutex			(generic_file_buffered_write)
 *    ->mmap_sem		(fault_in_pages_readable->do_page_fault)
L
Linus Torvalds 已提交
80
 *
81
 *  ->i_mutex
L
Linus Torvalds 已提交
82 83 84 85 86 87 88 89 90 91
 *    ->i_alloc_sem             (various)
 *
 *  ->inode_lock
 *    ->sb_lock			(fs/fs-writeback.c)
 *    ->mapping->tree_lock	(__sync_single_inode)
 *
 *  ->i_mmap_lock
 *    ->anon_vma.lock		(vma_adjust)
 *
 *  ->anon_vma.lock
92
 *    ->page_table_lock or pte_lock	(anon_vma_prepare and various)
L
Linus Torvalds 已提交
93
 *
94
 *  ->page_table_lock or pte_lock
95
 *    ->swap_lock		(try_to_unmap_one)
L
Linus Torvalds 已提交
96 97 98
 *    ->private_lock		(try_to_unmap_one)
 *    ->tree_lock		(try_to_unmap_one)
 *    ->zone.lru_lock		(follow_page->mark_page_accessed)
99
 *    ->zone.lru_lock		(check_pte_range->isolate_lru_page)
L
Linus Torvalds 已提交
100 101 102 103 104 105 106 107 108 109 110 111 112
 *    ->private_lock		(page_remove_rmap->set_page_dirty)
 *    ->tree_lock		(page_remove_rmap->set_page_dirty)
 *    ->inode_lock		(page_remove_rmap->set_page_dirty)
 *    ->inode_lock		(zap_pte_range->set_page_dirty)
 *    ->private_lock		(zap_pte_range->__set_page_dirty_buffers)
 *
 *  ->task->proc_lock
 *    ->dcache_lock		(proc_pid_lookup)
 */

/*
 * Remove a page from the page cache and free it. Caller has to make
 * sure the page is locked and that nobody else uses it - or that usage
N
Nick Piggin 已提交
113
 * is safe.  The caller must hold the mapping's tree_lock.
L
Linus Torvalds 已提交
114 115 116 117 118 119 120 121
 */
void __remove_from_page_cache(struct page *page)
{
	struct address_space *mapping = page->mapping;

	radix_tree_delete(&mapping->page_tree, page->index);
	page->mapping = NULL;
	mapping->nrpages--;
122
	__dec_zone_page_state(page, NR_FILE_PAGES);
123
	BUG_ON(page_mapped(page));
124
	mem_cgroup_uncharge_cache_page(page);
125 126 127 128 129 130 131 132 133 134 135 136

	/*
	 * Some filesystems seem to re-dirty the page even after
	 * the VM has canceled the dirty bit (eg ext3 journaling).
	 *
	 * Fix it up by doing a final dirty accounting check after
	 * having removed the page entirely.
	 */
	if (PageDirty(page) && mapping_cap_account_dirty(mapping)) {
		dec_zone_page_state(page, NR_FILE_DIRTY);
		dec_bdi_stat(mapping->backing_dev_info, BDI_RECLAIMABLE);
	}
L
Linus Torvalds 已提交
137 138 139 140 141 142
}

void remove_from_page_cache(struct page *page)
{
	struct address_space *mapping = page->mapping;

M
Matt Mackall 已提交
143
	BUG_ON(!PageLocked(page));
L
Linus Torvalds 已提交
144

N
Nick Piggin 已提交
145
	spin_lock_irq(&mapping->tree_lock);
L
Linus Torvalds 已提交
146
	__remove_from_page_cache(page);
N
Nick Piggin 已提交
147
	spin_unlock_irq(&mapping->tree_lock);
L
Linus Torvalds 已提交
148 149 150 151 152 153 154
}

static int sync_page(void *word)
{
	struct address_space *mapping;
	struct page *page;

155
	page = container_of((unsigned long *)word, struct page, flags);
L
Linus Torvalds 已提交
156 157

	/*
158 159 160 161 162 163 164 165 166 167 168 169 170 171 172
	 * page_mapping() is being called without PG_locked held.
	 * Some knowledge of the state and use of the page is used to
	 * reduce the requirements down to a memory barrier.
	 * The danger here is of a stale page_mapping() return value
	 * indicating a struct address_space different from the one it's
	 * associated with when it is associated with one.
	 * After smp_mb(), it's either the correct page_mapping() for
	 * the page, or an old page_mapping() and the page's own
	 * page_mapping() has gone NULL.
	 * The ->sync_page() address_space operation must tolerate
	 * page_mapping() going NULL. By an amazing coincidence,
	 * this comes about because none of the users of the page
	 * in the ->sync_page() methods make essential use of the
	 * page_mapping(), merely passing the page down to the backing
	 * device's unplug functions when it's non-NULL, which in turn
H
Hugh Dickins 已提交
173
	 * ignore it for all cases but swap, where only page_private(page) is
174 175 176
	 * of interest. When page_mapping() does go NULL, the entire
	 * call stack gracefully ignores the page and returns.
	 * -- wli
L
Linus Torvalds 已提交
177 178 179 180 181 182 183 184 185
	 */
	smp_mb();
	mapping = page_mapping(page);
	if (mapping && mapping->a_ops && mapping->a_ops->sync_page)
		mapping->a_ops->sync_page(page);
	io_schedule();
	return 0;
}

M
Matthew Wilcox 已提交
186 187 188 189 190 191
static int sync_page_killable(void *word)
{
	sync_page(word);
	return fatal_signal_pending(current) ? -EINTR : 0;
}

L
Linus Torvalds 已提交
192
/**
193
 * __filemap_fdatawrite_range - start writeback on mapping dirty pages in range
194 195
 * @mapping:	address space structure to write
 * @start:	offset in bytes where the range starts
196
 * @end:	offset in bytes where the range ends (inclusive)
197
 * @sync_mode:	enable synchronous operation
L
Linus Torvalds 已提交
198
 *
199 200 201
 * Start writeback against all of a mapping's dirty pages that lie
 * within the byte offsets <start, end> inclusive.
 *
L
Linus Torvalds 已提交
202
 * If sync_mode is WB_SYNC_ALL then this is a "data integrity" operation, as
203
 * opposed to a regular memory cleansing writeback.  The difference between
L
Linus Torvalds 已提交
204 205 206
 * these two operations is that if a dirty page/buffer is encountered, it must
 * be waited upon, and not just skipped over.
 */
207 208
int __filemap_fdatawrite_range(struct address_space *mapping, loff_t start,
				loff_t end, int sync_mode)
L
Linus Torvalds 已提交
209 210 211 212
{
	int ret;
	struct writeback_control wbc = {
		.sync_mode = sync_mode,
213
		.nr_to_write = LONG_MAX,
214 215
		.range_start = start,
		.range_end = end,
L
Linus Torvalds 已提交
216 217 218 219 220 221 222 223 224 225 226 227
	};

	if (!mapping_cap_writeback_dirty(mapping))
		return 0;

	ret = do_writepages(mapping, &wbc);
	return ret;
}

static inline int __filemap_fdatawrite(struct address_space *mapping,
	int sync_mode)
{
228
	return __filemap_fdatawrite_range(mapping, 0, LLONG_MAX, sync_mode);
L
Linus Torvalds 已提交
229 230 231 232 233 234 235 236
}

int filemap_fdatawrite(struct address_space *mapping)
{
	return __filemap_fdatawrite(mapping, WB_SYNC_ALL);
}
EXPORT_SYMBOL(filemap_fdatawrite);

237
int filemap_fdatawrite_range(struct address_space *mapping, loff_t start,
238
				loff_t end)
L
Linus Torvalds 已提交
239 240 241
{
	return __filemap_fdatawrite_range(mapping, start, end, WB_SYNC_ALL);
}
242
EXPORT_SYMBOL(filemap_fdatawrite_range);
L
Linus Torvalds 已提交
243

244 245 246 247
/**
 * filemap_flush - mostly a non-blocking flush
 * @mapping:	target address_space
 *
L
Linus Torvalds 已提交
248 249 250 251 252 253 254 255 256
 * This is a mostly non-blocking flush.  Not suitable for data-integrity
 * purposes - I/O may not be started against all dirty pages.
 */
int filemap_flush(struct address_space *mapping)
{
	return __filemap_fdatawrite(mapping, WB_SYNC_NONE);
}
EXPORT_SYMBOL(filemap_flush);

257 258 259 260 261 262
/**
 * wait_on_page_writeback_range - wait for writeback to complete
 * @mapping:	target address_space
 * @start:	beginning page index
 * @end:	ending page index
 *
L
Linus Torvalds 已提交
263 264 265
 * Wait for writeback to complete against pages indexed by start->end
 * inclusive
 */
266
int wait_on_page_writeback_range(struct address_space *mapping,
L
Linus Torvalds 已提交
267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
				pgoff_t start, pgoff_t end)
{
	struct pagevec pvec;
	int nr_pages;
	int ret = 0;
	pgoff_t index;

	if (end < start)
		return 0;

	pagevec_init(&pvec, 0);
	index = start;
	while ((index <= end) &&
			(nr_pages = pagevec_lookup_tag(&pvec, mapping, &index,
			PAGECACHE_TAG_WRITEBACK,
			min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1)) != 0) {
		unsigned i;

		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];

			/* until radix tree lookup accepts end_index */
			if (page->index > end)
				continue;

			wait_on_page_writeback(page);
			if (PageError(page))
				ret = -EIO;
		}
		pagevec_release(&pvec);
		cond_resched();
	}

	/* Check for outstanding write errors */
	if (test_and_clear_bit(AS_ENOSPC, &mapping->flags))
		ret = -ENOSPC;
	if (test_and_clear_bit(AS_EIO, &mapping->flags))
		ret = -EIO;

	return ret;
}

309 310 311 312 313 314 315
/**
 * sync_page_range - write and wait on all pages in the passed range
 * @inode:	target inode
 * @mapping:	target address_space
 * @pos:	beginning offset in pages to write
 * @count:	number of bytes to write
 *
L
Linus Torvalds 已提交
316 317 318 319
 * Write and wait upon all the pages in the passed range.  This is a "data
 * integrity" operation.  It waits upon in-flight writeout before starting and
 * waiting upon new writeout.  If there was an IO error, return it.
 *
320
 * We need to re-take i_mutex during the generic_osync_inode list walk because
L
Linus Torvalds 已提交
321 322 323
 * it is otherwise livelockable.
 */
int sync_page_range(struct inode *inode, struct address_space *mapping,
324
			loff_t pos, loff_t count)
L
Linus Torvalds 已提交
325 326 327 328 329 330 331 332 333
{
	pgoff_t start = pos >> PAGE_CACHE_SHIFT;
	pgoff_t end = (pos + count - 1) >> PAGE_CACHE_SHIFT;
	int ret;

	if (!mapping_cap_writeback_dirty(mapping) || !count)
		return 0;
	ret = filemap_fdatawrite_range(mapping, pos, pos + count - 1);
	if (ret == 0) {
334
		mutex_lock(&inode->i_mutex);
L
Linus Torvalds 已提交
335
		ret = generic_osync_inode(inode, mapping, OSYNC_METADATA);
336
		mutex_unlock(&inode->i_mutex);
L
Linus Torvalds 已提交
337 338 339 340 341 342 343
	}
	if (ret == 0)
		ret = wait_on_page_writeback_range(mapping, start, end);
	return ret;
}
EXPORT_SYMBOL(sync_page_range);

344
/**
345
 * sync_page_range_nolock - write & wait on all pages in the passed range without locking
346 347 348 349 350
 * @inode:	target inode
 * @mapping:	target address_space
 * @pos:	beginning offset in pages to write
 * @count:	number of bytes to write
 *
351
 * Note: Holding i_mutex across sync_page_range_nolock() is not a good idea
L
Linus Torvalds 已提交
352 353 354
 * as it forces O_SYNC writers to different parts of the same file
 * to be serialised right until io completion.
 */
355 356
int sync_page_range_nolock(struct inode *inode, struct address_space *mapping,
			   loff_t pos, loff_t count)
L
Linus Torvalds 已提交
357 358 359 360 361 362 363 364 365 366 367 368 369 370
{
	pgoff_t start = pos >> PAGE_CACHE_SHIFT;
	pgoff_t end = (pos + count - 1) >> PAGE_CACHE_SHIFT;
	int ret;

	if (!mapping_cap_writeback_dirty(mapping) || !count)
		return 0;
	ret = filemap_fdatawrite_range(mapping, pos, pos + count - 1);
	if (ret == 0)
		ret = generic_osync_inode(inode, mapping, OSYNC_METADATA);
	if (ret == 0)
		ret = wait_on_page_writeback_range(mapping, start, end);
	return ret;
}
371
EXPORT_SYMBOL(sync_page_range_nolock);
L
Linus Torvalds 已提交
372 373

/**
374
 * filemap_fdatawait - wait for all under-writeback pages to complete
L
Linus Torvalds 已提交
375
 * @mapping: address space structure to wait for
376 377 378
 *
 * Walk the list of under-writeback pages of the given address space
 * and wait for all of them.
L
Linus Torvalds 已提交
379 380 381 382 383 384 385 386 387 388 389 390 391 392 393
 */
int filemap_fdatawait(struct address_space *mapping)
{
	loff_t i_size = i_size_read(mapping->host);

	if (i_size == 0)
		return 0;

	return wait_on_page_writeback_range(mapping, 0,
				(i_size - 1) >> PAGE_CACHE_SHIFT);
}
EXPORT_SYMBOL(filemap_fdatawait);

int filemap_write_and_wait(struct address_space *mapping)
{
394
	int err = 0;
L
Linus Torvalds 已提交
395 396

	if (mapping->nrpages) {
397 398 399 400 401 402 403 404 405 406 407 408
		err = filemap_fdatawrite(mapping);
		/*
		 * Even if the above returned error, the pages may be
		 * written partially (e.g. -ENOSPC), so we wait for it.
		 * But the -EIO is special case, it may indicate the worst
		 * thing (e.g. bug) happened, so we avoid waiting for it.
		 */
		if (err != -EIO) {
			int err2 = filemap_fdatawait(mapping);
			if (!err)
				err = err2;
		}
L
Linus Torvalds 已提交
409
	}
410
	return err;
L
Linus Torvalds 已提交
411
}
412
EXPORT_SYMBOL(filemap_write_and_wait);
L
Linus Torvalds 已提交
413

414 415 416 417 418 419
/**
 * filemap_write_and_wait_range - write out & wait on a file range
 * @mapping:	the address_space for the pages
 * @lstart:	offset in bytes where the range starts
 * @lend:	offset in bytes where the range ends (inclusive)
 *
420 421 422 423 424
 * Write out and wait upon file offsets lstart->lend, inclusive.
 *
 * Note that `lend' is inclusive (describes the last byte to be written) so
 * that this function can be used to write to the very end-of-file (end = -1).
 */
L
Linus Torvalds 已提交
425 426 427
int filemap_write_and_wait_range(struct address_space *mapping,
				 loff_t lstart, loff_t lend)
{
428
	int err = 0;
L
Linus Torvalds 已提交
429 430

	if (mapping->nrpages) {
431 432 433 434 435 436 437 438 439 440
		err = __filemap_fdatawrite_range(mapping, lstart, lend,
						 WB_SYNC_ALL);
		/* See comment of filemap_write_and_wait() */
		if (err != -EIO) {
			int err2 = wait_on_page_writeback_range(mapping,
						lstart >> PAGE_CACHE_SHIFT,
						lend >> PAGE_CACHE_SHIFT);
			if (!err)
				err = err2;
		}
L
Linus Torvalds 已提交
441
	}
442
	return err;
L
Linus Torvalds 已提交
443
}
444
EXPORT_SYMBOL(filemap_write_and_wait_range);
L
Linus Torvalds 已提交
445

446
/**
N
Nick Piggin 已提交
447
 * add_to_page_cache_locked - add a locked page to the pagecache
448 449 450 451 452
 * @page:	page to add
 * @mapping:	the page's address_space
 * @offset:	page index
 * @gfp_mask:	page allocation mode
 *
N
Nick Piggin 已提交
453
 * This function is used to add a page to the pagecache. It must be locked.
L
Linus Torvalds 已提交
454 455
 * This function does not add the page to the LRU.  The caller must do that.
 */
N
Nick Piggin 已提交
456
int add_to_page_cache_locked(struct page *page, struct address_space *mapping,
A
Al Viro 已提交
457
		pgoff_t offset, gfp_t gfp_mask)
L
Linus Torvalds 已提交
458
{
N
Nick Piggin 已提交
459 460 461 462 463
	int error;

	VM_BUG_ON(!PageLocked(page));

	error = mem_cgroup_cache_charge(page, current->mm,
K
KAMEZAWA Hiroyuki 已提交
464
					gfp_mask & GFP_RECLAIM_MASK);
B
Balbir Singh 已提交
465 466
	if (error)
		goto out;
L
Linus Torvalds 已提交
467

B
Balbir Singh 已提交
468
	error = radix_tree_preload(gfp_mask & ~__GFP_HIGHMEM);
L
Linus Torvalds 已提交
469
	if (error == 0) {
N
Nick Piggin 已提交
470 471 472 473
		page_cache_get(page);
		page->mapping = mapping;
		page->index = offset;

N
Nick Piggin 已提交
474
		spin_lock_irq(&mapping->tree_lock);
L
Linus Torvalds 已提交
475
		error = radix_tree_insert(&mapping->page_tree, offset, page);
N
Nick Piggin 已提交
476
		if (likely(!error)) {
L
Linus Torvalds 已提交
477
			mapping->nrpages++;
478
			__inc_zone_page_state(page, NR_FILE_PAGES);
N
Nick Piggin 已提交
479 480
		} else {
			page->mapping = NULL;
481
			mem_cgroup_uncharge_cache_page(page);
N
Nick Piggin 已提交
482 483
			page_cache_release(page);
		}
484

N
Nick Piggin 已提交
485
		spin_unlock_irq(&mapping->tree_lock);
L
Linus Torvalds 已提交
486
		radix_tree_preload_end();
B
Balbir Singh 已提交
487
	} else
488
		mem_cgroup_uncharge_cache_page(page);
489
out:
L
Linus Torvalds 已提交
490 491
	return error;
}
N
Nick Piggin 已提交
492
EXPORT_SYMBOL(add_to_page_cache_locked);
L
Linus Torvalds 已提交
493 494

int add_to_page_cache_lru(struct page *page, struct address_space *mapping,
A
Al Viro 已提交
495
				pgoff_t offset, gfp_t gfp_mask)
L
Linus Torvalds 已提交
496
{
497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514
	int ret;

	/*
	 * Splice_read and readahead add shmem/tmpfs pages into the page cache
	 * before shmem_readpage has a chance to mark them as SwapBacked: they
	 * need to go on the active_anon lru below, and mem_cgroup_cache_charge
	 * (called in add_to_page_cache) needs to know where they're going too.
	 */
	if (mapping_cap_swap_backed(mapping))
		SetPageSwapBacked(page);

	ret = add_to_page_cache(page, mapping, offset, gfp_mask);
	if (ret == 0) {
		if (page_is_file_cache(page))
			lru_cache_add_file(page);
		else
			lru_cache_add_active_anon(page);
	}
L
Linus Torvalds 已提交
515 516
	return ret;
}
517
EXPORT_SYMBOL_GPL(add_to_page_cache_lru);
L
Linus Torvalds 已提交
518

519
#ifdef CONFIG_NUMA
520
struct page *__page_cache_alloc(gfp_t gfp)
521 522 523
{
	if (cpuset_do_page_mem_spread()) {
		int n = cpuset_mem_spread_node();
524
		return alloc_pages_node(n, gfp, 0);
525
	}
526
	return alloc_pages(gfp, 0);
527
}
528
EXPORT_SYMBOL(__page_cache_alloc);
529 530
#endif

531 532 533 534 535 536
static int __sleep_on_page_lock(void *word)
{
	io_schedule();
	return 0;
}

L
Linus Torvalds 已提交
537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558
/*
 * In order to wait for pages to become available there must be
 * waitqueues associated with pages. By using a hash table of
 * waitqueues where the bucket discipline is to maintain all
 * waiters on the same queue and wake all when any of the pages
 * become available, and for the woken contexts to check to be
 * sure the appropriate page became available, this saves space
 * at a cost of "thundering herd" phenomena during rare hash
 * collisions.
 */
static wait_queue_head_t *page_waitqueue(struct page *page)
{
	const struct zone *zone = page_zone(page);

	return &zone->wait_table[hash_ptr(page, zone->wait_table_bits)];
}

static inline void wake_up_page(struct page *page, int bit)
{
	__wake_up_bit(page_waitqueue(page), &page->flags, bit);
}

H
Harvey Harrison 已提交
559
void wait_on_page_bit(struct page *page, int bit_nr)
L
Linus Torvalds 已提交
560 561 562 563 564 565 566 567 568
{
	DEFINE_WAIT_BIT(wait, &page->flags, bit_nr);

	if (test_bit(bit_nr, &page->flags))
		__wait_on_bit(page_waitqueue(page), &wait, sync_page,
							TASK_UNINTERRUPTIBLE);
}
EXPORT_SYMBOL(wait_on_page_bit);

569 570
/**
 * add_page_wait_queue - Add an arbitrary waiter to a page's wait queue
R
Randy Dunlap 已提交
571 572
 * @page: Page defining the wait queue of interest
 * @waiter: Waiter to add to the queue
573 574 575 576 577 578 579 580 581 582 583 584 585 586
 *
 * Add an arbitrary @waiter to the wait queue for the nominated @page.
 */
void add_page_wait_queue(struct page *page, wait_queue_t *waiter)
{
	wait_queue_head_t *q = page_waitqueue(page);
	unsigned long flags;

	spin_lock_irqsave(&q->lock, flags);
	__add_wait_queue(q, waiter);
	spin_unlock_irqrestore(&q->lock, flags);
}
EXPORT_SYMBOL_GPL(add_page_wait_queue);

L
Linus Torvalds 已提交
587
/**
588
 * unlock_page - unlock a locked page
L
Linus Torvalds 已提交
589 590 591 592 593 594 595
 * @page: the page
 *
 * Unlocks the page and wakes up sleepers in ___wait_on_page_locked().
 * Also wakes sleepers in wait_on_page_writeback() because the wakeup
 * mechananism between PageLocked pages and PageWriteback pages is shared.
 * But that's OK - sleepers in wait_on_page_writeback() just go back to sleep.
 *
N
Nick Piggin 已提交
596 597
 * The mb is necessary to enforce ordering between the clear_bit and the read
 * of the waitqueue (to avoid SMP races with a parallel wait_on_page_locked()).
L
Linus Torvalds 已提交
598
 */
H
Harvey Harrison 已提交
599
void unlock_page(struct page *page)
L
Linus Torvalds 已提交
600
{
N
Nick Piggin 已提交
601 602 603
	VM_BUG_ON(!PageLocked(page));
	clear_bit_unlock(PG_locked, &page->flags);
	smp_mb__after_clear_bit();
L
Linus Torvalds 已提交
604 605 606 607
	wake_up_page(page, PG_locked);
}
EXPORT_SYMBOL(unlock_page);

608 609 610
/**
 * end_page_writeback - end writeback against a page
 * @page: the page
L
Linus Torvalds 已提交
611 612 613
 */
void end_page_writeback(struct page *page)
{
614 615 616 617 618 619
	if (TestClearPageReclaim(page))
		rotate_reclaimable_page(page);

	if (!test_clear_page_writeback(page))
		BUG();

L
Linus Torvalds 已提交
620 621 622 623 624
	smp_mb__after_clear_bit();
	wake_up_page(page, PG_writeback);
}
EXPORT_SYMBOL(end_page_writeback);

625 626 627
/**
 * __lock_page - get a lock on the page, assuming we need to sleep to get it
 * @page: the page to lock
L
Linus Torvalds 已提交
628
 *
629
 * Ugly. Running sync_page() in state TASK_UNINTERRUPTIBLE is scary.  If some
L
Linus Torvalds 已提交
630 631 632 633
 * random driver's requestfn sets TASK_RUNNING, we could busywait.  However
 * chances are that on the second loop, the block layer's plug list is empty,
 * so sync_page() will then return in state TASK_UNINTERRUPTIBLE.
 */
H
Harvey Harrison 已提交
634
void __lock_page(struct page *page)
L
Linus Torvalds 已提交
635 636 637 638 639 640 641 642
{
	DEFINE_WAIT_BIT(wait, &page->flags, PG_locked);

	__wait_on_bit_lock(page_waitqueue(page), &wait, sync_page,
							TASK_UNINTERRUPTIBLE);
}
EXPORT_SYMBOL(__lock_page);

H
Harvey Harrison 已提交
643
int __lock_page_killable(struct page *page)
M
Matthew Wilcox 已提交
644 645 646 647 648 649
{
	DEFINE_WAIT_BIT(wait, &page->flags, PG_locked);

	return __wait_on_bit_lock(page_waitqueue(page), &wait,
					sync_page_killable, TASK_KILLABLE);
}
650
EXPORT_SYMBOL_GPL(__lock_page_killable);
M
Matthew Wilcox 已提交
651

652 653 654 655
/**
 * __lock_page_nosync - get a lock on the page, without calling sync_page()
 * @page: the page to lock
 *
656 657 658
 * Variant of lock_page that does not require the caller to hold a reference
 * on the page's mapping.
 */
H
Harvey Harrison 已提交
659
void __lock_page_nosync(struct page *page)
660 661 662 663 664 665
{
	DEFINE_WAIT_BIT(wait, &page->flags, PG_locked);
	__wait_on_bit_lock(page_waitqueue(page), &wait, __sleep_on_page_lock,
							TASK_UNINTERRUPTIBLE);
}

666 667 668 669 670
/**
 * find_get_page - find and get a page reference
 * @mapping: the address_space to search
 * @offset: the page index
 *
N
Nick Piggin 已提交
671 672
 * Is there a pagecache struct page at the given (mapping, offset) tuple?
 * If yes, increment its refcount and return it; if no, return NULL.
L
Linus Torvalds 已提交
673
 */
N
Nick Piggin 已提交
674
struct page *find_get_page(struct address_space *mapping, pgoff_t offset)
L
Linus Torvalds 已提交
675
{
N
Nick Piggin 已提交
676
	void **pagep;
L
Linus Torvalds 已提交
677 678
	struct page *page;

N
Nick Piggin 已提交
679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702
	rcu_read_lock();
repeat:
	page = NULL;
	pagep = radix_tree_lookup_slot(&mapping->page_tree, offset);
	if (pagep) {
		page = radix_tree_deref_slot(pagep);
		if (unlikely(!page || page == RADIX_TREE_RETRY))
			goto repeat;

		if (!page_cache_get_speculative(page))
			goto repeat;

		/*
		 * Has the page moved?
		 * This is part of the lockless pagecache protocol. See
		 * include/linux/pagemap.h for details.
		 */
		if (unlikely(page != *pagep)) {
			page_cache_release(page);
			goto repeat;
		}
	}
	rcu_read_unlock();

L
Linus Torvalds 已提交
703 704 705 706 707 708
	return page;
}
EXPORT_SYMBOL(find_get_page);

/**
 * find_lock_page - locate, pin and lock a pagecache page
709 710
 * @mapping: the address_space to search
 * @offset: the page index
L
Linus Torvalds 已提交
711 712 713 714 715 716
 *
 * Locates the desired pagecache page, locks it, increments its reference
 * count and returns its address.
 *
 * Returns zero if the page was not present. find_lock_page() may sleep.
 */
N
Nick Piggin 已提交
717
struct page *find_lock_page(struct address_space *mapping, pgoff_t offset)
L
Linus Torvalds 已提交
718 719 720 721
{
	struct page *page;

repeat:
N
Nick Piggin 已提交
722
	page = find_get_page(mapping, offset);
L
Linus Torvalds 已提交
723
	if (page) {
N
Nick Piggin 已提交
724 725 726 727 728 729
		lock_page(page);
		/* Has the page been truncated? */
		if (unlikely(page->mapping != mapping)) {
			unlock_page(page);
			page_cache_release(page);
			goto repeat;
L
Linus Torvalds 已提交
730
		}
N
Nick Piggin 已提交
731
		VM_BUG_ON(page->index != offset);
L
Linus Torvalds 已提交
732 733 734 735 736 737 738
	}
	return page;
}
EXPORT_SYMBOL(find_lock_page);

/**
 * find_or_create_page - locate or add a pagecache page
739 740 741
 * @mapping: the page's address_space
 * @index: the page's index into the mapping
 * @gfp_mask: page allocation mode
L
Linus Torvalds 已提交
742 743 744 745 746 747 748 749 750 751 752 753 754
 *
 * Locates a page in the pagecache.  If the page is not present, a new page
 * is allocated using @gfp_mask and is added to the pagecache and to the VM's
 * LRU list.  The returned page is locked and has its reference count
 * incremented.
 *
 * find_or_create_page() may sleep, even if @gfp_flags specifies an atomic
 * allocation!
 *
 * find_or_create_page() returns the desired page's address, or zero on
 * memory exhaustion.
 */
struct page *find_or_create_page(struct address_space *mapping,
755
		pgoff_t index, gfp_t gfp_mask)
L
Linus Torvalds 已提交
756
{
N
Nick Piggin 已提交
757
	struct page *page;
L
Linus Torvalds 已提交
758 759 760 761
	int err;
repeat:
	page = find_lock_page(mapping, index);
	if (!page) {
N
Nick Piggin 已提交
762 763 764
		page = __page_cache_alloc(gfp_mask);
		if (!page)
			return NULL;
N
Nick Piggin 已提交
765 766 767 768 769 770 771 772
		/*
		 * We want a regular kernel memory (not highmem or DMA etc)
		 * allocation for the radix tree nodes, but we need to honour
		 * the context-specific requirements the caller has asked for.
		 * GFP_RECLAIM_MASK collects those requirements.
		 */
		err = add_to_page_cache_lru(page, mapping, index,
			(gfp_mask & GFP_RECLAIM_MASK));
N
Nick Piggin 已提交
773 774 775 776 777
		if (unlikely(err)) {
			page_cache_release(page);
			page = NULL;
			if (err == -EEXIST)
				goto repeat;
L
Linus Torvalds 已提交
778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804
		}
	}
	return page;
}
EXPORT_SYMBOL(find_or_create_page);

/**
 * find_get_pages - gang pagecache lookup
 * @mapping:	The address_space to search
 * @start:	The starting page index
 * @nr_pages:	The maximum number of pages
 * @pages:	Where the resulting pages are placed
 *
 * find_get_pages() will search for and return a group of up to
 * @nr_pages pages in the mapping.  The pages are placed at @pages.
 * find_get_pages() takes a reference against the returned pages.
 *
 * The search returns a group of mapping-contiguous pages with ascending
 * indexes.  There may be holes in the indices due to not-present pages.
 *
 * find_get_pages() returns the number of pages which were found.
 */
unsigned find_get_pages(struct address_space *mapping, pgoff_t start,
			    unsigned int nr_pages, struct page **pages)
{
	unsigned int i;
	unsigned int ret;
N
Nick Piggin 已提交
805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832
	unsigned int nr_found;

	rcu_read_lock();
restart:
	nr_found = radix_tree_gang_lookup_slot(&mapping->page_tree,
				(void ***)pages, start, nr_pages);
	ret = 0;
	for (i = 0; i < nr_found; i++) {
		struct page *page;
repeat:
		page = radix_tree_deref_slot((void **)pages[i]);
		if (unlikely(!page))
			continue;
		/*
		 * this can only trigger if nr_found == 1, making livelock
		 * a non issue.
		 */
		if (unlikely(page == RADIX_TREE_RETRY))
			goto restart;

		if (!page_cache_get_speculative(page))
			goto repeat;

		/* Has the page moved? */
		if (unlikely(page != *((void **)pages[i]))) {
			page_cache_release(page);
			goto repeat;
		}
L
Linus Torvalds 已提交
833

N
Nick Piggin 已提交
834 835 836 837
		pages[ret] = page;
		ret++;
	}
	rcu_read_unlock();
L
Linus Torvalds 已提交
838 839 840
	return ret;
}

841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857
/**
 * find_get_pages_contig - gang contiguous pagecache lookup
 * @mapping:	The address_space to search
 * @index:	The starting page index
 * @nr_pages:	The maximum number of pages
 * @pages:	Where the resulting pages are placed
 *
 * find_get_pages_contig() works exactly like find_get_pages(), except
 * that the returned number of pages are guaranteed to be contiguous.
 *
 * find_get_pages_contig() returns the number of pages which were found.
 */
unsigned find_get_pages_contig(struct address_space *mapping, pgoff_t index,
			       unsigned int nr_pages, struct page **pages)
{
	unsigned int i;
	unsigned int ret;
N
Nick Piggin 已提交
858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876
	unsigned int nr_found;

	rcu_read_lock();
restart:
	nr_found = radix_tree_gang_lookup_slot(&mapping->page_tree,
				(void ***)pages, index, nr_pages);
	ret = 0;
	for (i = 0; i < nr_found; i++) {
		struct page *page;
repeat:
		page = radix_tree_deref_slot((void **)pages[i]);
		if (unlikely(!page))
			continue;
		/*
		 * this can only trigger if nr_found == 1, making livelock
		 * a non issue.
		 */
		if (unlikely(page == RADIX_TREE_RETRY))
			goto restart;
877

N
Nick Piggin 已提交
878
		if (page->mapping == NULL || page->index != index)
879 880
			break;

N
Nick Piggin 已提交
881 882 883 884 885 886 887 888 889 890 891
		if (!page_cache_get_speculative(page))
			goto repeat;

		/* Has the page moved? */
		if (unlikely(page != *((void **)pages[i]))) {
			page_cache_release(page);
			goto repeat;
		}

		pages[ret] = page;
		ret++;
892 893
		index++;
	}
N
Nick Piggin 已提交
894 895
	rcu_read_unlock();
	return ret;
896
}
897
EXPORT_SYMBOL(find_get_pages_contig);
898

899 900 901 902 903 904 905 906
/**
 * find_get_pages_tag - find and return pages that match @tag
 * @mapping:	the address_space to search
 * @index:	the starting page index
 * @tag:	the tag index
 * @nr_pages:	the maximum number of pages
 * @pages:	where the resulting pages are placed
 *
L
Linus Torvalds 已提交
907
 * Like find_get_pages, except we only return pages which are tagged with
908
 * @tag.   We update @index to index the next page for the traversal.
L
Linus Torvalds 已提交
909 910 911 912 913 914
 */
unsigned find_get_pages_tag(struct address_space *mapping, pgoff_t *index,
			int tag, unsigned int nr_pages, struct page **pages)
{
	unsigned int i;
	unsigned int ret;
N
Nick Piggin 已提交
915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947
	unsigned int nr_found;

	rcu_read_lock();
restart:
	nr_found = radix_tree_gang_lookup_tag_slot(&mapping->page_tree,
				(void ***)pages, *index, nr_pages, tag);
	ret = 0;
	for (i = 0; i < nr_found; i++) {
		struct page *page;
repeat:
		page = radix_tree_deref_slot((void **)pages[i]);
		if (unlikely(!page))
			continue;
		/*
		 * this can only trigger if nr_found == 1, making livelock
		 * a non issue.
		 */
		if (unlikely(page == RADIX_TREE_RETRY))
			goto restart;

		if (!page_cache_get_speculative(page))
			goto repeat;

		/* Has the page moved? */
		if (unlikely(page != *((void **)pages[i]))) {
			page_cache_release(page);
			goto repeat;
		}

		pages[ret] = page;
		ret++;
	}
	rcu_read_unlock();
L
Linus Torvalds 已提交
948 949 950

	if (ret)
		*index = pages[ret - 1]->index + 1;
N
Nick Piggin 已提交
951

L
Linus Torvalds 已提交
952 953
	return ret;
}
954
EXPORT_SYMBOL(find_get_pages_tag);
L
Linus Torvalds 已提交
955

956 957 958 959 960
/**
 * grab_cache_page_nowait - returns locked page at given index in given cache
 * @mapping: target address_space
 * @index: the page index
 *
961
 * Same as grab_cache_page(), but do not wait if the page is unavailable.
L
Linus Torvalds 已提交
962 963 964 965 966 967 968 969
 * This is intended for speculative data generators, where the data can
 * be regenerated if the page couldn't be grabbed.  This routine should
 * be safe to call while holding the lock for another page.
 *
 * Clear __GFP_FS when allocating the page to avoid recursion into the fs
 * and deadlock against the caller's locked page.
 */
struct page *
970
grab_cache_page_nowait(struct address_space *mapping, pgoff_t index)
L
Linus Torvalds 已提交
971 972 973 974
{
	struct page *page = find_get_page(mapping, index);

	if (page) {
N
Nick Piggin 已提交
975
		if (trylock_page(page))
L
Linus Torvalds 已提交
976 977 978 979
			return page;
		page_cache_release(page);
		return NULL;
	}
980
	page = __page_cache_alloc(mapping_gfp_mask(mapping) & ~__GFP_FS);
N
Nick Piggin 已提交
981
	if (page && add_to_page_cache_lru(page, mapping, index, GFP_NOFS)) {
L
Linus Torvalds 已提交
982 983 984 985 986 987 988
		page_cache_release(page);
		page = NULL;
	}
	return page;
}
EXPORT_SYMBOL(grab_cache_page_nowait);

989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012
/*
 * CD/DVDs are error prone. When a medium error occurs, the driver may fail
 * a _large_ part of the i/o request. Imagine the worst scenario:
 *
 *      ---R__________________________________________B__________
 *         ^ reading here                             ^ bad block(assume 4k)
 *
 * read(R) => miss => readahead(R...B) => media error => frustrating retries
 * => failing the whole request => read(R) => read(R+1) =>
 * readahead(R+1...B+1) => bang => read(R+2) => read(R+3) =>
 * readahead(R+3...B+2) => bang => read(R+3) => read(R+4) =>
 * readahead(R+4...B+3) => bang => read(R+4) => read(R+5) => ......
 *
 * It is going insane. Fix it by quickly scaling down the readahead size.
 */
static void shrink_readahead_size_eio(struct file *filp,
					struct file_ra_state *ra)
{
	if (!ra->ra_pages)
		return;

	ra->ra_pages /= 4;
}

1013
/**
C
Christoph Hellwig 已提交
1014
 * do_generic_file_read - generic file read routine
1015 1016 1017 1018 1019
 * @filp:	the file to read
 * @ppos:	current file position
 * @desc:	read_descriptor
 * @actor:	read method
 *
L
Linus Torvalds 已提交
1020
 * This is a generic file read routine, and uses the
1021
 * mapping->a_ops->readpage() function for the actual low-level stuff.
L
Linus Torvalds 已提交
1022 1023 1024 1025
 *
 * This is really ugly. But the goto's actually try to clarify some
 * of the logic when it comes to error handling etc.
 */
C
Christoph Hellwig 已提交
1026 1027
static void do_generic_file_read(struct file *filp, loff_t *ppos,
		read_descriptor_t *desc, read_actor_t actor)
L
Linus Torvalds 已提交
1028
{
C
Christoph Hellwig 已提交
1029
	struct address_space *mapping = filp->f_mapping;
L
Linus Torvalds 已提交
1030
	struct inode *inode = mapping->host;
C
Christoph Hellwig 已提交
1031
	struct file_ra_state *ra = &filp->f_ra;
1032 1033 1034 1035
	pgoff_t index;
	pgoff_t last_index;
	pgoff_t prev_index;
	unsigned long offset;      /* offset into pagecache page */
1036
	unsigned int prev_offset;
L
Linus Torvalds 已提交
1037 1038 1039
	int error;

	index = *ppos >> PAGE_CACHE_SHIFT;
1040 1041
	prev_index = ra->prev_pos >> PAGE_CACHE_SHIFT;
	prev_offset = ra->prev_pos & (PAGE_CACHE_SIZE-1);
L
Linus Torvalds 已提交
1042 1043 1044 1045 1046
	last_index = (*ppos + desc->count + PAGE_CACHE_SIZE-1) >> PAGE_CACHE_SHIFT;
	offset = *ppos & ~PAGE_CACHE_MASK;

	for (;;) {
		struct page *page;
1047
		pgoff_t end_index;
N
NeilBrown 已提交
1048
		loff_t isize;
L
Linus Torvalds 已提交
1049 1050 1051 1052 1053
		unsigned long nr, ret;

		cond_resched();
find_page:
		page = find_get_page(mapping, index);
1054
		if (!page) {
1055
			page_cache_sync_readahead(mapping,
1056
					ra, filp,
1057 1058 1059 1060 1061 1062
					index, last_index - index);
			page = find_get_page(mapping, index);
			if (unlikely(page == NULL))
				goto no_cached_page;
		}
		if (PageReadahead(page)) {
1063
			page_cache_async_readahead(mapping,
1064
					ra, filp, page,
1065
					index, last_index - index);
L
Linus Torvalds 已提交
1066
		}
1067 1068 1069 1070
		if (!PageUptodate(page)) {
			if (inode->i_blkbits == PAGE_CACHE_SHIFT ||
					!mapping->a_ops->is_partially_uptodate)
				goto page_not_up_to_date;
N
Nick Piggin 已提交
1071
			if (!trylock_page(page))
1072 1073 1074 1075 1076 1077
				goto page_not_up_to_date;
			if (!mapping->a_ops->is_partially_uptodate(page,
								desc, offset))
				goto page_not_up_to_date_locked;
			unlock_page(page);
		}
L
Linus Torvalds 已提交
1078
page_ok:
N
NeilBrown 已提交
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
		/*
		 * i_size must be checked after we know the page is Uptodate.
		 *
		 * Checking i_size after the check allows us to calculate
		 * the correct value for "nr", which means the zero-filled
		 * part of the page is not copied back to userspace (unless
		 * another truncate extends the file - this is desired though).
		 */

		isize = i_size_read(inode);
		end_index = (isize - 1) >> PAGE_CACHE_SHIFT;
		if (unlikely(!isize || index > end_index)) {
			page_cache_release(page);
			goto out;
		}

		/* nr is the maximum number of bytes to copy from this page */
		nr = PAGE_CACHE_SIZE;
		if (index == end_index) {
			nr = ((isize - 1) & ~PAGE_CACHE_MASK) + 1;
			if (nr <= offset) {
				page_cache_release(page);
				goto out;
			}
		}
		nr = nr - offset;
L
Linus Torvalds 已提交
1105 1106 1107 1108 1109 1110 1111 1112 1113

		/* If users can be writing to this page using arbitrary
		 * virtual addresses, take care about potential aliasing
		 * before reading the page on the kernel side.
		 */
		if (mapping_writably_mapped(mapping))
			flush_dcache_page(page);

		/*
1114 1115
		 * When a sequential read accesses a page several times,
		 * only mark it as accessed the first time.
L
Linus Torvalds 已提交
1116
		 */
1117
		if (prev_index != index || offset != prev_offset)
L
Linus Torvalds 已提交
1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134
			mark_page_accessed(page);
		prev_index = index;

		/*
		 * Ok, we have the page, and it's up-to-date, so
		 * now we can copy it to user space...
		 *
		 * The actor routine returns how many bytes were actually used..
		 * NOTE! This may not be the same as how much of a user buffer
		 * we filled up (we may be padding etc), so we can only update
		 * "pos" here (the actor routine has to update the user buffer
		 * pointers and the remaining count).
		 */
		ret = actor(desc, page, offset, nr);
		offset += ret;
		index += offset >> PAGE_CACHE_SHIFT;
		offset &= ~PAGE_CACHE_MASK;
J
Jan Kara 已提交
1135
		prev_offset = offset;
L
Linus Torvalds 已提交
1136 1137 1138 1139 1140 1141 1142 1143

		page_cache_release(page);
		if (ret == nr && desc->count)
			continue;
		goto out;

page_not_up_to_date:
		/* Get exclusive access to the page ... */
1144 1145 1146
		error = lock_page_killable(page);
		if (unlikely(error))
			goto readpage_error;
L
Linus Torvalds 已提交
1147

1148
page_not_up_to_date_locked:
N
Nick Piggin 已提交
1149
		/* Did it get truncated before we got the lock? */
L
Linus Torvalds 已提交
1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165
		if (!page->mapping) {
			unlock_page(page);
			page_cache_release(page);
			continue;
		}

		/* Did somebody else fill it already? */
		if (PageUptodate(page)) {
			unlock_page(page);
			goto page_ok;
		}

readpage:
		/* Start the actual read. The read will unlock the page. */
		error = mapping->a_ops->readpage(filp, page);

1166 1167 1168 1169 1170
		if (unlikely(error)) {
			if (error == AOP_TRUNCATED_PAGE) {
				page_cache_release(page);
				goto find_page;
			}
L
Linus Torvalds 已提交
1171
			goto readpage_error;
1172
		}
L
Linus Torvalds 已提交
1173 1174

		if (!PageUptodate(page)) {
1175 1176 1177
			error = lock_page_killable(page);
			if (unlikely(error))
				goto readpage_error;
L
Linus Torvalds 已提交
1178 1179 1180 1181 1182 1183 1184 1185 1186 1187
			if (!PageUptodate(page)) {
				if (page->mapping == NULL) {
					/*
					 * invalidate_inode_pages got it
					 */
					unlock_page(page);
					page_cache_release(page);
					goto find_page;
				}
				unlock_page(page);
1188
				shrink_readahead_size_eio(filp, ra);
1189 1190
				error = -EIO;
				goto readpage_error;
L
Linus Torvalds 已提交
1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207
			}
			unlock_page(page);
		}

		goto page_ok;

readpage_error:
		/* UHHUH! A synchronous read error occurred. Report it */
		desc->error = error;
		page_cache_release(page);
		goto out;

no_cached_page:
		/*
		 * Ok, it wasn't cached, so we need to create a new
		 * page..
		 */
N
Nick Piggin 已提交
1208 1209 1210 1211
		page = page_cache_alloc_cold(mapping);
		if (!page) {
			desc->error = -ENOMEM;
			goto out;
L
Linus Torvalds 已提交
1212
		}
N
Nick Piggin 已提交
1213
		error = add_to_page_cache_lru(page, mapping,
L
Linus Torvalds 已提交
1214 1215
						index, GFP_KERNEL);
		if (error) {
N
Nick Piggin 已提交
1216
			page_cache_release(page);
L
Linus Torvalds 已提交
1217 1218 1219 1220 1221 1222 1223 1224 1225
			if (error == -EEXIST)
				goto find_page;
			desc->error = error;
			goto out;
		}
		goto readpage;
	}

out:
1226 1227 1228
	ra->prev_pos = prev_index;
	ra->prev_pos <<= PAGE_CACHE_SHIFT;
	ra->prev_pos |= prev_offset;
L
Linus Torvalds 已提交
1229

1230
	*ppos = ((loff_t)index << PAGE_CACHE_SHIFT) + offset;
1231
	file_accessed(filp);
L
Linus Torvalds 已提交
1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271
}

int file_read_actor(read_descriptor_t *desc, struct page *page,
			unsigned long offset, unsigned long size)
{
	char *kaddr;
	unsigned long left, count = desc->count;

	if (size > count)
		size = count;

	/*
	 * Faults on the destination of a read are common, so do it before
	 * taking the kmap.
	 */
	if (!fault_in_pages_writeable(desc->arg.buf, size)) {
		kaddr = kmap_atomic(page, KM_USER0);
		left = __copy_to_user_inatomic(desc->arg.buf,
						kaddr + offset, size);
		kunmap_atomic(kaddr, KM_USER0);
		if (left == 0)
			goto success;
	}

	/* Do it the slow way */
	kaddr = kmap(page);
	left = __copy_to_user(desc->arg.buf, kaddr + offset, size);
	kunmap(page);

	if (left) {
		size -= left;
		desc->error = -EFAULT;
	}
success:
	desc->count = count - size;
	desc->written += size;
	desc->arg.buf += size;
	return size;
}

1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310
/*
 * Performs necessary checks before doing a write
 * @iov:	io vector request
 * @nr_segs:	number of segments in the iovec
 * @count:	number of bytes to write
 * @access_flags: type of access: %VERIFY_READ or %VERIFY_WRITE
 *
 * Adjust number of segments and amount of bytes to write (nr_segs should be
 * properly initialized first). Returns appropriate error code that caller
 * should return or zero in case that write should be allowed.
 */
int generic_segment_checks(const struct iovec *iov,
			unsigned long *nr_segs, size_t *count, int access_flags)
{
	unsigned long   seg;
	size_t cnt = 0;
	for (seg = 0; seg < *nr_segs; seg++) {
		const struct iovec *iv = &iov[seg];

		/*
		 * If any segment has a negative length, or the cumulative
		 * length ever wraps negative then return -EINVAL.
		 */
		cnt += iv->iov_len;
		if (unlikely((ssize_t)(cnt|iv->iov_len) < 0))
			return -EINVAL;
		if (access_ok(access_flags, iv->iov_base, iv->iov_len))
			continue;
		if (seg == 0)
			return -EFAULT;
		*nr_segs = seg;
		cnt -= iv->iov_len;	/* This segment is no good */
		break;
	}
	*count = cnt;
	return 0;
}
EXPORT_SYMBOL(generic_segment_checks);

1311
/**
H
Henrik Kretzschmar 已提交
1312
 * generic_file_aio_read - generic filesystem read routine
1313 1314 1315
 * @iocb:	kernel I/O control block
 * @iov:	io vector request
 * @nr_segs:	number of segments in the iovec
H
Henrik Kretzschmar 已提交
1316
 * @pos:	current file position
1317
 *
L
Linus Torvalds 已提交
1318 1319 1320 1321
 * This is the "read()" routine for all filesystems
 * that can use the page cache directly.
 */
ssize_t
1322 1323
generic_file_aio_read(struct kiocb *iocb, const struct iovec *iov,
		unsigned long nr_segs, loff_t pos)
L
Linus Torvalds 已提交
1324 1325 1326 1327 1328
{
	struct file *filp = iocb->ki_filp;
	ssize_t retval;
	unsigned long seg;
	size_t count;
1329
	loff_t *ppos = &iocb->ki_pos;
L
Linus Torvalds 已提交
1330 1331

	count = 0;
1332 1333 1334
	retval = generic_segment_checks(iov, &nr_segs, &count, VERIFY_WRITE);
	if (retval)
		return retval;
L
Linus Torvalds 已提交
1335 1336 1337

	/* coalesce the iovecs and go direct-to-BIO for O_DIRECT */
	if (filp->f_flags & O_DIRECT) {
1338
		loff_t size;
L
Linus Torvalds 已提交
1339 1340 1341 1342 1343 1344 1345 1346 1347
		struct address_space *mapping;
		struct inode *inode;

		mapping = filp->f_mapping;
		inode = mapping->host;
		if (!count)
			goto out; /* skip atime */
		size = i_size_read(inode);
		if (pos < size) {
1348 1349
			retval = filemap_write_and_wait_range(mapping, pos,
					pos + iov_length(iov, nr_segs) - 1);
1350 1351 1352 1353
			if (!retval) {
				retval = mapping->a_ops->direct_IO(READ, iocb,
							iov, pos, nr_segs);
			}
L
Linus Torvalds 已提交
1354 1355
			if (retval > 0)
				*ppos = pos + retval;
H
Hugh Dickins 已提交
1356 1357 1358 1359
			if (retval) {
				file_accessed(filp);
				goto out;
			}
1360
		}
L
Linus Torvalds 已提交
1361 1362
	}

H
Hugh Dickins 已提交
1363 1364
	for (seg = 0; seg < nr_segs; seg++) {
		read_descriptor_t desc;
L
Linus Torvalds 已提交
1365

H
Hugh Dickins 已提交
1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376
		desc.written = 0;
		desc.arg.buf = iov[seg].iov_base;
		desc.count = iov[seg].iov_len;
		if (desc.count == 0)
			continue;
		desc.error = 0;
		do_generic_file_read(filp, ppos, &desc, file_read_actor);
		retval += desc.written;
		if (desc.error) {
			retval = retval ?: desc.error;
			break;
L
Linus Torvalds 已提交
1377
		}
H
Hugh Dickins 已提交
1378 1379
		if (desc.count > 0)
			break;
L
Linus Torvalds 已提交
1380 1381 1382 1383 1384 1385 1386 1387
	}
out:
	return retval;
}
EXPORT_SYMBOL(generic_file_aio_read);

static ssize_t
do_readahead(struct address_space *mapping, struct file *filp,
1388
	     pgoff_t index, unsigned long nr)
L
Linus Torvalds 已提交
1389 1390 1391 1392 1393 1394 1395 1396 1397
{
	if (!mapping || !mapping->a_ops || !mapping->a_ops->readpage)
		return -EINVAL;

	force_page_cache_readahead(mapping, filp, index,
					max_sane_readahead(nr));
	return 0;
}

1398
SYSCALL_DEFINE(readahead)(int fd, loff_t offset, size_t count)
L
Linus Torvalds 已提交
1399 1400 1401 1402 1403 1404 1405 1406 1407
{
	ssize_t ret;
	struct file *file;

	ret = -EBADF;
	file = fget(fd);
	if (file) {
		if (file->f_mode & FMODE_READ) {
			struct address_space *mapping = file->f_mapping;
1408 1409
			pgoff_t start = offset >> PAGE_CACHE_SHIFT;
			pgoff_t end = (offset + count - 1) >> PAGE_CACHE_SHIFT;
L
Linus Torvalds 已提交
1410 1411 1412 1413 1414 1415 1416
			unsigned long len = end - start + 1;
			ret = do_readahead(mapping, file, start, len);
		}
		fput(file);
	}
	return ret;
}
1417 1418 1419 1420 1421 1422 1423
#ifdef CONFIG_HAVE_SYSCALL_WRAPPERS
asmlinkage long SyS_readahead(long fd, loff_t offset, long count)
{
	return SYSC_readahead((int) fd, offset, (size_t) count);
}
SYSCALL_ALIAS(sys_readahead, SyS_readahead);
#endif
L
Linus Torvalds 已提交
1424 1425

#ifdef CONFIG_MMU
1426 1427 1428 1429 1430
/**
 * page_cache_read - adds requested page to the page cache if not already there
 * @file:	file to read
 * @offset:	page index
 *
L
Linus Torvalds 已提交
1431 1432 1433
 * This adds the requested page to the page cache if it isn't already there,
 * and schedules an I/O to read in its contents from disk.
 */
H
Harvey Harrison 已提交
1434
static int page_cache_read(struct file *file, pgoff_t offset)
L
Linus Torvalds 已提交
1435 1436 1437
{
	struct address_space *mapping = file->f_mapping;
	struct page *page; 
1438
	int ret;
L
Linus Torvalds 已提交
1439

1440 1441 1442 1443 1444 1445 1446 1447 1448 1449
	do {
		page = page_cache_alloc_cold(mapping);
		if (!page)
			return -ENOMEM;

		ret = add_to_page_cache_lru(page, mapping, offset, GFP_KERNEL);
		if (ret == 0)
			ret = mapping->a_ops->readpage(file, page);
		else if (ret == -EEXIST)
			ret = 0; /* losing race to add is OK */
L
Linus Torvalds 已提交
1450 1451 1452

		page_cache_release(page);

1453 1454 1455
	} while (ret == AOP_TRUNCATED_PAGE);
		
	return ret;
L
Linus Torvalds 已提交
1456 1457 1458 1459
}

#define MMAP_LOTSAMISS  (100)

1460
/**
1461
 * filemap_fault - read in file data for page fault handling
N
Nick Piggin 已提交
1462 1463
 * @vma:	vma in which the fault was taken
 * @vmf:	struct vm_fault containing details of the fault
1464
 *
1465
 * filemap_fault() is invoked via the vma operations vector for a
L
Linus Torvalds 已提交
1466 1467 1468 1469 1470 1471
 * mapped memory region to read in file data during a page fault.
 *
 * The goto's are kind of ugly, but this streamlines the normal case of having
 * it in the page cache, and handles the special cases reasonably without
 * having a lot of duplicated code.
 */
N
Nick Piggin 已提交
1472
int filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
L
Linus Torvalds 已提交
1473 1474
{
	int error;
1475
	struct file *file = vma->vm_file;
L
Linus Torvalds 已提交
1476 1477 1478 1479
	struct address_space *mapping = file->f_mapping;
	struct file_ra_state *ra = &file->f_ra;
	struct inode *inode = mapping->host;
	struct page *page;
J
Jan Kara 已提交
1480
	pgoff_t size;
1481
	int did_readaround = 0;
N
Nick Piggin 已提交
1482
	int ret = 0;
L
Linus Torvalds 已提交
1483 1484

	size = (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
N
Nick Piggin 已提交
1485
	if (vmf->pgoff >= size)
1486
		return VM_FAULT_SIGBUS;
L
Linus Torvalds 已提交
1487 1488

	/* If we don't want any read-ahead, don't bother */
1489
	if (VM_RandomReadHint(vma))
L
Linus Torvalds 已提交
1490 1491 1492 1493 1494 1495
		goto no_cached_page;

	/*
	 * Do we have something in the page cache already?
	 */
retry_find:
N
Nick Piggin 已提交
1496
	page = find_lock_page(mapping, vmf->pgoff);
1497 1498 1499 1500 1501
	/*
	 * For sequential accesses, we use the generic readahead logic.
	 */
	if (VM_SequentialReadHint(vma)) {
		if (!page) {
1502
			page_cache_sync_readahead(mapping, ra, file,
1503 1504 1505 1506 1507 1508
							   vmf->pgoff, 1);
			page = find_lock_page(mapping, vmf->pgoff);
			if (!page)
				goto no_cached_page;
		}
		if (PageReadahead(page)) {
1509
			page_cache_async_readahead(mapping, ra, file, page,
1510 1511 1512 1513
							   vmf->pgoff, 1);
		}
	}

L
Linus Torvalds 已提交
1514 1515 1516 1517 1518 1519 1520 1521 1522
	if (!page) {
		unsigned long ra_pages;

		ra->mmap_miss++;

		/*
		 * Do we miss much more than hit in this file? If so,
		 * stop bothering with read-ahead. It will only hurt.
		 */
1523
		if (ra->mmap_miss > MMAP_LOTSAMISS)
L
Linus Torvalds 已提交
1524 1525 1526 1527 1528 1529 1530
			goto no_cached_page;

		/*
		 * To keep the pgmajfault counter straight, we need to
		 * check did_readaround, as this is an inner loop.
		 */
		if (!did_readaround) {
N
Nick Piggin 已提交
1531
			ret = VM_FAULT_MAJOR;
1532
			count_vm_event(PGMAJFAULT);
L
Linus Torvalds 已提交
1533 1534 1535 1536 1537 1538
		}
		did_readaround = 1;
		ra_pages = max_sane_readahead(file->f_ra.ra_pages);
		if (ra_pages) {
			pgoff_t start = 0;

N
Nick Piggin 已提交
1539 1540
			if (vmf->pgoff > ra_pages / 2)
				start = vmf->pgoff - ra_pages / 2;
L
Linus Torvalds 已提交
1541 1542
			do_page_cache_readahead(mapping, file, start, ra_pages);
		}
N
Nick Piggin 已提交
1543
		page = find_lock_page(mapping, vmf->pgoff);
L
Linus Torvalds 已提交
1544 1545 1546 1547 1548
		if (!page)
			goto no_cached_page;
	}

	if (!did_readaround)
1549
		ra->mmap_miss--;
L
Linus Torvalds 已提交
1550 1551

	/*
1552 1553
	 * We have a locked page in the page cache, now we need to check
	 * that it's up-to-date. If not, it is going to be due to an error.
L
Linus Torvalds 已提交
1554
	 */
1555
	if (unlikely(!PageUptodate(page)))
L
Linus Torvalds 已提交
1556 1557
		goto page_not_uptodate;

1558 1559
	/* Must recheck i_size under page lock */
	size = (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
N
Nick Piggin 已提交
1560
	if (unlikely(vmf->pgoff >= size)) {
1561
		unlock_page(page);
1562
		page_cache_release(page);
1563
		return VM_FAULT_SIGBUS;
1564 1565
	}

L
Linus Torvalds 已提交
1566 1567 1568
	/*
	 * Found the page and have a reference on it.
	 */
1569
	ra->prev_pos = (loff_t)page->index << PAGE_CACHE_SHIFT;
N
Nick Piggin 已提交
1570
	vmf->page = page;
N
Nick Piggin 已提交
1571
	return ret | VM_FAULT_LOCKED;
L
Linus Torvalds 已提交
1572 1573 1574 1575 1576 1577

no_cached_page:
	/*
	 * We're only likely to ever get here if MADV_RANDOM is in
	 * effect.
	 */
N
Nick Piggin 已提交
1578
	error = page_cache_read(file, vmf->pgoff);
L
Linus Torvalds 已提交
1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593

	/*
	 * The page we want has now been added to the page cache.
	 * In the unlikely event that someone removed it in the
	 * meantime, we'll just come back here and read it again.
	 */
	if (error >= 0)
		goto retry_find;

	/*
	 * An error return from page_cache_read can result if the
	 * system is low on memory, or a problem occurs while trying
	 * to schedule I/O.
	 */
	if (error == -ENOMEM)
N
Nick Piggin 已提交
1594 1595
		return VM_FAULT_OOM;
	return VM_FAULT_SIGBUS;
L
Linus Torvalds 已提交
1596 1597

page_not_uptodate:
1598
	/* IO error path */
L
Linus Torvalds 已提交
1599
	if (!did_readaround) {
N
Nick Piggin 已提交
1600
		ret = VM_FAULT_MAJOR;
1601
		count_vm_event(PGMAJFAULT);
L
Linus Torvalds 已提交
1602 1603 1604 1605 1606 1607 1608 1609 1610
	}

	/*
	 * Umm, take care of errors if the page isn't up-to-date.
	 * Try to re-read it _once_. We do this synchronously,
	 * because there really aren't any performance issues here
	 * and we need to check for errors.
	 */
	ClearPageError(page);
1611
	error = mapping->a_ops->readpage(file, page);
1612 1613 1614 1615 1616
	if (!error) {
		wait_on_page_locked(page);
		if (!PageUptodate(page))
			error = -EIO;
	}
1617 1618 1619
	page_cache_release(page);

	if (!error || error == AOP_TRUNCATED_PAGE)
1620
		goto retry_find;
L
Linus Torvalds 已提交
1621

1622
	/* Things didn't work out. Return zero to tell the mm layer so. */
1623
	shrink_readahead_size_eio(file, ra);
N
Nick Piggin 已提交
1624
	return VM_FAULT_SIGBUS;
1625 1626 1627
}
EXPORT_SYMBOL(filemap_fault);

L
Linus Torvalds 已提交
1628
struct vm_operations_struct generic_file_vm_ops = {
1629
	.fault		= filemap_fault,
L
Linus Torvalds 已提交
1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641
};

/* This is used for a general mmap of a disk file */

int generic_file_mmap(struct file * file, struct vm_area_struct * vma)
{
	struct address_space *mapping = file->f_mapping;

	if (!mapping->a_ops->readpage)
		return -ENOEXEC;
	file_accessed(file);
	vma->vm_ops = &generic_file_vm_ops;
N
Nick Piggin 已提交
1642
	vma->vm_flags |= VM_CAN_NONLINEAR;
L
Linus Torvalds 已提交
1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668
	return 0;
}

/*
 * This is for filesystems which do not implement ->writepage.
 */
int generic_file_readonly_mmap(struct file *file, struct vm_area_struct *vma)
{
	if ((vma->vm_flags & VM_SHARED) && (vma->vm_flags & VM_MAYWRITE))
		return -EINVAL;
	return generic_file_mmap(file, vma);
}
#else
int generic_file_mmap(struct file * file, struct vm_area_struct * vma)
{
	return -ENOSYS;
}
int generic_file_readonly_mmap(struct file * file, struct vm_area_struct * vma)
{
	return -ENOSYS;
}
#endif /* CONFIG_MMU */

EXPORT_SYMBOL(generic_file_mmap);
EXPORT_SYMBOL(generic_file_readonly_mmap);

1669
static struct page *__read_cache_page(struct address_space *mapping,
1670
				pgoff_t index,
L
Linus Torvalds 已提交
1671 1672 1673
				int (*filler)(void *,struct page*),
				void *data)
{
N
Nick Piggin 已提交
1674
	struct page *page;
L
Linus Torvalds 已提交
1675 1676 1677 1678
	int err;
repeat:
	page = find_get_page(mapping, index);
	if (!page) {
N
Nick Piggin 已提交
1679 1680 1681 1682 1683 1684 1685 1686
		page = page_cache_alloc_cold(mapping);
		if (!page)
			return ERR_PTR(-ENOMEM);
		err = add_to_page_cache_lru(page, mapping, index, GFP_KERNEL);
		if (unlikely(err)) {
			page_cache_release(page);
			if (err == -EEXIST)
				goto repeat;
L
Linus Torvalds 已提交
1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698
			/* Presumably ENOMEM for radix tree node */
			return ERR_PTR(err);
		}
		err = filler(data, page);
		if (err < 0) {
			page_cache_release(page);
			page = ERR_PTR(err);
		}
	}
	return page;
}

1699 1700 1701 1702 1703 1704 1705
/**
 * read_cache_page_async - read into page cache, fill it if needed
 * @mapping:	the page's address_space
 * @index:	the page index
 * @filler:	function to perform the read
 * @data:	destination for read data
 *
1706 1707
 * Same as read_cache_page, but don't wait for page to become unlocked
 * after submitting it to the filler.
1708 1709 1710 1711 1712
 *
 * Read into the page cache. If a page already exists, and PageUptodate() is
 * not set, try to fill the page but don't wait for it to become unlocked.
 *
 * If the page does not get brought uptodate, return -EIO.
L
Linus Torvalds 已提交
1713
 */
1714
struct page *read_cache_page_async(struct address_space *mapping,
1715
				pgoff_t index,
L
Linus Torvalds 已提交
1716 1717 1718 1719 1720 1721 1722 1723 1724
				int (*filler)(void *,struct page*),
				void *data)
{
	struct page *page;
	int err;

retry:
	page = __read_cache_page(mapping, index, filler, data);
	if (IS_ERR(page))
1725
		return page;
L
Linus Torvalds 已提交
1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741
	if (PageUptodate(page))
		goto out;

	lock_page(page);
	if (!page->mapping) {
		unlock_page(page);
		page_cache_release(page);
		goto retry;
	}
	if (PageUptodate(page)) {
		unlock_page(page);
		goto out;
	}
	err = filler(data, page);
	if (err < 0) {
		page_cache_release(page);
1742
		return ERR_PTR(err);
L
Linus Torvalds 已提交
1743
	}
1744
out:
1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762
	mark_page_accessed(page);
	return page;
}
EXPORT_SYMBOL(read_cache_page_async);

/**
 * read_cache_page - read into page cache, fill it if needed
 * @mapping:	the page's address_space
 * @index:	the page index
 * @filler:	function to perform the read
 * @data:	destination for read data
 *
 * Read into the page cache. If a page already exists, and PageUptodate() is
 * not set, try to fill the page then wait for it to become unlocked.
 *
 * If the page does not get brought uptodate, return -EIO.
 */
struct page *read_cache_page(struct address_space *mapping,
1763
				pgoff_t index,
1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776
				int (*filler)(void *,struct page*),
				void *data)
{
	struct page *page;

	page = read_cache_page_async(mapping, index, filler, data);
	if (IS_ERR(page))
		goto out;
	wait_on_page_locked(page);
	if (!PageUptodate(page)) {
		page_cache_release(page);
		page = ERR_PTR(-EIO);
	}
L
Linus Torvalds 已提交
1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787
 out:
	return page;
}
EXPORT_SYMBOL(read_cache_page);

/*
 * The logic we want is
 *
 *	if suid or (sgid and xgrp)
 *		remove privs
 */
1788
int should_remove_suid(struct dentry *dentry)
L
Linus Torvalds 已提交
1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803
{
	mode_t mode = dentry->d_inode->i_mode;
	int kill = 0;

	/* suid always must be killed */
	if (unlikely(mode & S_ISUID))
		kill = ATTR_KILL_SUID;

	/*
	 * sgid without any exec bits is just a mandatory locking mark; leave
	 * it alone.  If some exec bits are set, it's a real sgid; kill it.
	 */
	if (unlikely((mode & S_ISGID) && (mode & S_IXGRP)))
		kill |= ATTR_KILL_SGID;

1804
	if (unlikely(kill && !capable(CAP_FSETID) && S_ISREG(mode)))
1805
		return kill;
L
Linus Torvalds 已提交
1806

1807 1808
	return 0;
}
M
Mark Fasheh 已提交
1809
EXPORT_SYMBOL(should_remove_suid);
1810

1811
static int __remove_suid(struct dentry *dentry, int kill)
1812 1813 1814 1815 1816 1817 1818
{
	struct iattr newattrs;

	newattrs.ia_valid = ATTR_FORCE | kill;
	return notify_change(dentry, &newattrs);
}

1819
int file_remove_suid(struct file *file)
1820
{
1821
	struct dentry *dentry = file->f_path.dentry;
1822 1823 1824
	int killsuid = should_remove_suid(dentry);
	int killpriv = security_inode_need_killpriv(dentry);
	int error = 0;
1825

1826 1827 1828 1829 1830 1831
	if (killpriv < 0)
		return killpriv;
	if (killpriv)
		error = security_inode_killpriv(dentry);
	if (!error && killsuid)
		error = __remove_suid(dentry, killsuid);
1832

1833
	return error;
L
Linus Torvalds 已提交
1834
}
1835
EXPORT_SYMBOL(file_remove_suid);
L
Linus Torvalds 已提交
1836

N
Nick Piggin 已提交
1837
static size_t __iovec_copy_from_user_inatomic(char *vaddr,
L
Linus Torvalds 已提交
1838 1839
			const struct iovec *iov, size_t base, size_t bytes)
{
1840
	size_t copied = 0, left = 0;
L
Linus Torvalds 已提交
1841 1842 1843 1844 1845 1846

	while (bytes) {
		char __user *buf = iov->iov_base + base;
		int copy = min(bytes, iov->iov_len - base);

		base = 0;
1847
		left = __copy_from_user_inatomic(vaddr, buf, copy);
L
Linus Torvalds 已提交
1848 1849 1850 1851 1852
		copied += copy;
		bytes -= copy;
		vaddr += copy;
		iov++;

1853
		if (unlikely(left))
L
Linus Torvalds 已提交
1854 1855 1856 1857 1858
			break;
	}
	return copied - left;
}

N
Nick Piggin 已提交
1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874
/*
 * Copy as much as we can into the page and return the number of bytes which
 * were sucessfully copied.  If a fault is encountered then return the number of
 * bytes which were copied.
 */
size_t iov_iter_copy_from_user_atomic(struct page *page,
		struct iov_iter *i, unsigned long offset, size_t bytes)
{
	char *kaddr;
	size_t copied;

	BUG_ON(!in_atomic());
	kaddr = kmap_atomic(page, KM_USER0);
	if (likely(i->nr_segs == 1)) {
		int left;
		char __user *buf = i->iov->iov_base + i->iov_offset;
1875
		left = __copy_from_user_inatomic(kaddr + offset, buf, bytes);
N
Nick Piggin 已提交
1876 1877 1878 1879 1880 1881 1882 1883 1884
		copied = bytes - left;
	} else {
		copied = __iovec_copy_from_user_inatomic(kaddr + offset,
						i->iov, i->iov_offset, bytes);
	}
	kunmap_atomic(kaddr, KM_USER0);

	return copied;
}
N
Nick Piggin 已提交
1885
EXPORT_SYMBOL(iov_iter_copy_from_user_atomic);
N
Nick Piggin 已提交
1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902

/*
 * This has the same sideeffects and return value as
 * iov_iter_copy_from_user_atomic().
 * The difference is that it attempts to resolve faults.
 * Page must not be locked.
 */
size_t iov_iter_copy_from_user(struct page *page,
		struct iov_iter *i, unsigned long offset, size_t bytes)
{
	char *kaddr;
	size_t copied;

	kaddr = kmap(page);
	if (likely(i->nr_segs == 1)) {
		int left;
		char __user *buf = i->iov->iov_base + i->iov_offset;
1903
		left = __copy_from_user(kaddr + offset, buf, bytes);
N
Nick Piggin 已提交
1904 1905 1906 1907 1908 1909 1910 1911
		copied = bytes - left;
	} else {
		copied = __iovec_copy_from_user_inatomic(kaddr + offset,
						i->iov, i->iov_offset, bytes);
	}
	kunmap(page);
	return copied;
}
N
Nick Piggin 已提交
1912
EXPORT_SYMBOL(iov_iter_copy_from_user);
N
Nick Piggin 已提交
1913

N
Nick Piggin 已提交
1914
void iov_iter_advance(struct iov_iter *i, size_t bytes)
N
Nick Piggin 已提交
1915
{
N
Nick Piggin 已提交
1916 1917
	BUG_ON(i->count < bytes);

N
Nick Piggin 已提交
1918 1919
	if (likely(i->nr_segs == 1)) {
		i->iov_offset += bytes;
N
Nick Piggin 已提交
1920
		i->count -= bytes;
N
Nick Piggin 已提交
1921 1922 1923 1924
	} else {
		const struct iovec *iov = i->iov;
		size_t base = i->iov_offset;

1925 1926
		/*
		 * The !iov->iov_len check ensures we skip over unlikely
N
Nick Piggin 已提交
1927
		 * zero-length segments (without overruning the iovec).
1928
		 */
1929
		while (bytes || unlikely(i->count && !iov->iov_len)) {
N
Nick Piggin 已提交
1930
			int copy;
N
Nick Piggin 已提交
1931

N
Nick Piggin 已提交
1932 1933 1934
			copy = min(bytes, iov->iov_len - base);
			BUG_ON(!i->count || i->count < copy);
			i->count -= copy;
N
Nick Piggin 已提交
1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945
			bytes -= copy;
			base += copy;
			if (iov->iov_len == base) {
				iov++;
				base = 0;
			}
		}
		i->iov = iov;
		i->iov_offset = base;
	}
}
N
Nick Piggin 已提交
1946
EXPORT_SYMBOL(iov_iter_advance);
N
Nick Piggin 已提交
1947

1948 1949 1950 1951 1952 1953 1954 1955 1956 1957
/*
 * Fault in the first iovec of the given iov_iter, to a maximum length
 * of bytes. Returns 0 on success, or non-zero if the memory could not be
 * accessed (ie. because it is an invalid address).
 *
 * writev-intensive code may want this to prefault several iovecs -- that
 * would be possible (callers must not rely on the fact that _only_ the
 * first iovec will be faulted with the current implementation).
 */
int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes)
N
Nick Piggin 已提交
1958 1959
{
	char __user *buf = i->iov->iov_base + i->iov_offset;
1960 1961
	bytes = min(bytes, i->iov->iov_len - i->iov_offset);
	return fault_in_pages_readable(buf, bytes);
N
Nick Piggin 已提交
1962
}
N
Nick Piggin 已提交
1963
EXPORT_SYMBOL(iov_iter_fault_in_readable);
N
Nick Piggin 已提交
1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975

/*
 * Return the count of just the current iov_iter segment.
 */
size_t iov_iter_single_seg_count(struct iov_iter *i)
{
	const struct iovec *iov = i->iov;
	if (i->nr_segs == 1)
		return i->count;
	else
		return min(i->count, iov->iov_len - i->iov_offset);
}
N
Nick Piggin 已提交
1976
EXPORT_SYMBOL(iov_iter_single_seg_count);
N
Nick Piggin 已提交
1977

L
Linus Torvalds 已提交
1978 1979 1980
/*
 * Performs necessary checks before doing a write
 *
1981
 * Can adjust writing position or amount of bytes to write.
L
Linus Torvalds 已提交
1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039
 * Returns appropriate error code that caller should return or
 * zero in case that write should be allowed.
 */
inline int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk)
{
	struct inode *inode = file->f_mapping->host;
	unsigned long limit = current->signal->rlim[RLIMIT_FSIZE].rlim_cur;

        if (unlikely(*pos < 0))
                return -EINVAL;

	if (!isblk) {
		/* FIXME: this is for backwards compatibility with 2.4 */
		if (file->f_flags & O_APPEND)
                        *pos = i_size_read(inode);

		if (limit != RLIM_INFINITY) {
			if (*pos >= limit) {
				send_sig(SIGXFSZ, current, 0);
				return -EFBIG;
			}
			if (*count > limit - (typeof(limit))*pos) {
				*count = limit - (typeof(limit))*pos;
			}
		}
	}

	/*
	 * LFS rule
	 */
	if (unlikely(*pos + *count > MAX_NON_LFS &&
				!(file->f_flags & O_LARGEFILE))) {
		if (*pos >= MAX_NON_LFS) {
			return -EFBIG;
		}
		if (*count > MAX_NON_LFS - (unsigned long)*pos) {
			*count = MAX_NON_LFS - (unsigned long)*pos;
		}
	}

	/*
	 * Are we about to exceed the fs block limit ?
	 *
	 * If we have written data it becomes a short write.  If we have
	 * exceeded without writing data we send a signal and return EFBIG.
	 * Linus frestrict idea will clean these up nicely..
	 */
	if (likely(!isblk)) {
		if (unlikely(*pos >= inode->i_sb->s_maxbytes)) {
			if (*count || *pos > inode->i_sb->s_maxbytes) {
				return -EFBIG;
			}
			/* zero-length writes at ->s_maxbytes are OK */
		}

		if (unlikely(*pos + *count > inode->i_sb->s_maxbytes))
			*count = inode->i_sb->s_maxbytes - *pos;
	} else {
2040
#ifdef CONFIG_BLOCK
L
Linus Torvalds 已提交
2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051
		loff_t isize;
		if (bdev_read_only(I_BDEV(inode)))
			return -EPERM;
		isize = i_size_read(inode);
		if (*pos >= isize) {
			if (*count || *pos > isize)
				return -ENOSPC;
		}

		if (*pos + *count > isize)
			*count = isize - *pos;
2052 2053 2054
#else
		return -EPERM;
#endif
L
Linus Torvalds 已提交
2055 2056 2057 2058 2059
	}
	return 0;
}
EXPORT_SYMBOL(generic_write_checks);

2060 2061 2062 2063 2064 2065
int pagecache_write_begin(struct file *file, struct address_space *mapping,
				loff_t pos, unsigned len, unsigned flags,
				struct page **pagep, void **fsdata)
{
	const struct address_space_operations *aops = mapping->a_ops;

2066
	return aops->write_begin(file, mapping, pos, len, flags,
2067 2068 2069 2070 2071 2072 2073 2074 2075 2076
							pagep, fsdata);
}
EXPORT_SYMBOL(pagecache_write_begin);

int pagecache_write_end(struct file *file, struct address_space *mapping,
				loff_t pos, unsigned len, unsigned copied,
				struct page *page, void *fsdata)
{
	const struct address_space_operations *aops = mapping->a_ops;

2077 2078
	mark_page_accessed(page);
	return aops->write_end(file, mapping, pos, len, copied, page, fsdata);
2079 2080 2081
}
EXPORT_SYMBOL(pagecache_write_end);

L
Linus Torvalds 已提交
2082 2083 2084 2085 2086 2087 2088 2089 2090
ssize_t
generic_file_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 address_space *mapping = file->f_mapping;
	struct inode	*inode = mapping->host;
	ssize_t		written;
2091 2092
	size_t		write_len;
	pgoff_t		end;
L
Linus Torvalds 已提交
2093 2094 2095 2096

	if (count != ocount)
		*nr_segs = iov_shorten((struct iovec *)iov, *nr_segs, count);

2097 2098 2099
	write_len = iov_length(iov, *nr_segs);
	end = (pos + write_len - 1) >> PAGE_CACHE_SHIFT;

2100
	written = filemap_write_and_wait_range(mapping, pos, pos + write_len - 1);
2101 2102 2103 2104 2105 2106 2107
	if (written)
		goto out;

	/*
	 * After a write we want buffered reads to be sure to go to disk to get
	 * the new data.  We invalidate clean cached page from the region we're
	 * about to write.  We do this *before* the write so that we can return
2108
	 * without clobbering -EIOCBQUEUED from ->direct_IO().
2109 2110 2111 2112
	 */
	if (mapping->nrpages) {
		written = invalidate_inode_pages2_range(mapping,
					pos >> PAGE_CACHE_SHIFT, end);
2113 2114 2115 2116 2117 2118 2119
		/*
		 * If a page can not be invalidated, return 0 to fall back
		 * to buffered write.
		 */
		if (written) {
			if (written == -EBUSY)
				return 0;
2120
			goto out;
2121
		}
2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138
	}

	written = mapping->a_ops->direct_IO(WRITE, iocb, iov, pos, *nr_segs);

	/*
	 * Finally, try again to invalidate clean pages which might have been
	 * cached by non-direct readahead, or faulted in by get_user_pages()
	 * if the source of the write was an mmap'ed region of the file
	 * we're writing.  Either one is a pretty crazy thing to do,
	 * so we don't support it 100%.  If this invalidation
	 * fails, tough, the write still worked...
	 */
	if (mapping->nrpages) {
		invalidate_inode_pages2_range(mapping,
					      pos >> PAGE_CACHE_SHIFT, end);
	}

L
Linus Torvalds 已提交
2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150
	if (written > 0) {
		loff_t end = pos + written;
		if (end > i_size_read(inode) && !S_ISBLK(inode->i_mode)) {
			i_size_write(inode,  end);
			mark_inode_dirty(inode);
		}
		*ppos = end;
	}

	/*
	 * Sync the fs metadata but not the minor inode changes and
	 * of course not the data as we did direct DMA for the IO.
2151
	 * i_mutex is held, which protects generic_osync_inode() from
2152
	 * livelocking.  AIO O_DIRECT ops attempt to sync metadata here.
L
Linus Torvalds 已提交
2153
	 */
2154
out:
2155 2156
	if ((written >= 0 || written == -EIOCBQUEUED) &&
	    ((file->f_flags & O_SYNC) || IS_SYNC(inode))) {
2157 2158 2159 2160
		int err = generic_osync_inode(inode, mapping, OSYNC_METADATA);
		if (err < 0)
			written = err;
	}
L
Linus Torvalds 已提交
2161 2162 2163 2164
	return written;
}
EXPORT_SYMBOL(generic_file_direct_write);

N
Nick Piggin 已提交
2165 2166 2167 2168
/*
 * Find or create a page at the given pagecache position. Return the locked
 * page. This function is specifically for buffered writes.
 */
2169 2170
struct page *grab_cache_page_write_begin(struct address_space *mapping,
					pgoff_t index, unsigned flags)
N
Nick Piggin 已提交
2171 2172 2173
{
	int status;
	struct page *page;
2174 2175 2176
	gfp_t gfp_notmask = 0;
	if (flags & AOP_FLAG_NOFS)
		gfp_notmask = __GFP_FS;
N
Nick Piggin 已提交
2177 2178 2179 2180 2181
repeat:
	page = find_lock_page(mapping, index);
	if (likely(page))
		return page;

2182
	page = __page_cache_alloc(mapping_gfp_mask(mapping) & ~gfp_notmask);
N
Nick Piggin 已提交
2183 2184
	if (!page)
		return NULL;
2185 2186
	status = add_to_page_cache_lru(page, mapping, index,
						GFP_KERNEL & ~gfp_notmask);
N
Nick Piggin 已提交
2187 2188 2189 2190 2191 2192 2193 2194
	if (unlikely(status)) {
		page_cache_release(page);
		if (status == -EEXIST)
			goto repeat;
		return NULL;
	}
	return page;
}
2195
EXPORT_SYMBOL(grab_cache_page_write_begin);
N
Nick Piggin 已提交
2196

2197 2198 2199 2200 2201 2202 2203
static ssize_t generic_perform_write(struct file *file,
				struct iov_iter *i, loff_t pos)
{
	struct address_space *mapping = file->f_mapping;
	const struct address_space_operations *a_ops = mapping->a_ops;
	long status = 0;
	ssize_t written = 0;
N
Nick Piggin 已提交
2204 2205 2206 2207 2208 2209 2210
	unsigned int flags = 0;

	/*
	 * Copies from kernel address space cannot fail (NFSD is a big user).
	 */
	if (segment_eq(get_fs(), KERNEL_DS))
		flags |= AOP_FLAG_UNINTERRUPTIBLE;
2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241

	do {
		struct page *page;
		pgoff_t index;		/* Pagecache index for current page */
		unsigned long offset;	/* Offset into pagecache page */
		unsigned long bytes;	/* Bytes to write to page */
		size_t copied;		/* Bytes copied from user */
		void *fsdata;

		offset = (pos & (PAGE_CACHE_SIZE - 1));
		index = pos >> PAGE_CACHE_SHIFT;
		bytes = min_t(unsigned long, PAGE_CACHE_SIZE - offset,
						iov_iter_count(i));

again:

		/*
		 * Bring in the user page that we will copy from _first_.
		 * Otherwise there's a nasty deadlock on copying from the
		 * same page as we're writing to, without it being marked
		 * up-to-date.
		 *
		 * Not only is this an optimisation, but it is also required
		 * to check that the address is actually valid, when atomic
		 * usercopies are used, below.
		 */
		if (unlikely(iov_iter_fault_in_readable(i, bytes))) {
			status = -EFAULT;
			break;
		}

N
Nick Piggin 已提交
2242
		status = a_ops->write_begin(file, mapping, pos, bytes, flags,
2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259
						&page, &fsdata);
		if (unlikely(status))
			break;

		pagefault_disable();
		copied = iov_iter_copy_from_user_atomic(page, i, offset, bytes);
		pagefault_enable();
		flush_dcache_page(page);

		status = a_ops->write_end(file, mapping, pos, bytes, copied,
						page, fsdata);
		if (unlikely(status < 0))
			break;
		copied = status;

		cond_resched();

2260
		iov_iter_advance(i, copied);
2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296
		if (unlikely(copied == 0)) {
			/*
			 * If we were unable to copy any data at all, we must
			 * fall back to a single segment length write.
			 *
			 * If we didn't fallback here, we could livelock
			 * because not all segments in the iov can be copied at
			 * once without a pagefault.
			 */
			bytes = min_t(unsigned long, PAGE_CACHE_SIZE - offset,
						iov_iter_single_seg_count(i));
			goto again;
		}
		pos += copied;
		written += copied;

		balance_dirty_pages_ratelimited(mapping);

	} while (iov_iter_count(i));

	return written ? written : status;
}

ssize_t
generic_file_buffered_write(struct kiocb *iocb, const struct iovec *iov,
		unsigned long nr_segs, loff_t pos, loff_t *ppos,
		size_t count, ssize_t written)
{
	struct file *file = iocb->ki_filp;
	struct address_space *mapping = file->f_mapping;
	const struct address_space_operations *a_ops = mapping->a_ops;
	struct inode *inode = mapping->host;
	ssize_t status;
	struct iov_iter i;

	iov_iter_init(&i, iov, nr_segs, count, written);
2297
	status = generic_perform_write(file, &i, pos);
L
Linus Torvalds 已提交
2298 2299

	if (likely(status >= 0)) {
2300 2301 2302 2303 2304 2305 2306
		written += status;
		*ppos = pos + status;

		/*
		 * For now, when the user asks for O_SYNC, we'll actually give
		 * O_DSYNC
		 */
L
Linus Torvalds 已提交
2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319
		if (unlikely((file->f_flags & O_SYNC) || IS_SYNC(inode))) {
			if (!a_ops->writepage || !is_sync_kiocb(iocb))
				status = generic_osync_inode(inode, mapping,
						OSYNC_METADATA|OSYNC_DATA);
		}
  	}
	
	/*
	 * If we get here for O_DIRECT writes then we must have fallen through
	 * to buffered writes (block instantiation inside i_size).  So we sync
	 * the file data here, to try to honour O_DIRECT expectations.
	 */
	if (unlikely(file->f_flags & O_DIRECT) && written)
2320 2321
		status = filemap_write_and_wait_range(mapping,
					pos, pos + written - 1);
L
Linus Torvalds 已提交
2322 2323 2324 2325 2326

	return written ? written : status;
}
EXPORT_SYMBOL(generic_file_buffered_write);

2327
static ssize_t
L
Linus Torvalds 已提交
2328 2329 2330 2331
__generic_file_aio_write_nolock(struct kiocb *iocb, const struct iovec *iov,
				unsigned long nr_segs, loff_t *ppos)
{
	struct file *file = iocb->ki_filp;
2332
	struct address_space * mapping = file->f_mapping;
L
Linus Torvalds 已提交
2333 2334 2335 2336 2337 2338 2339 2340
	size_t ocount;		/* original count */
	size_t count;		/* after file limit checks */
	struct inode 	*inode = mapping->host;
	loff_t		pos;
	ssize_t		written;
	ssize_t		err;

	ocount = 0;
2341 2342 2343
	err = generic_segment_checks(iov, &nr_segs, &ocount, VERIFY_READ);
	if (err)
		return err;
L
Linus Torvalds 已提交
2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360

	count = ocount;
	pos = *ppos;

	vfs_check_frozen(inode->i_sb, SB_FREEZE_WRITE);

	/* We can write back this queue in page reclaim */
	current->backing_dev_info = mapping->backing_dev_info;
	written = 0;

	err = generic_write_checks(file, &pos, &count, S_ISBLK(inode->i_mode));
	if (err)
		goto out;

	if (count == 0)
		goto out;

2361
	err = file_remove_suid(file);
L
Linus Torvalds 已提交
2362 2363 2364
	if (err)
		goto out;

2365
	file_update_time(file);
L
Linus Torvalds 已提交
2366 2367 2368

	/* coalesce the iovecs and go direct-to-BIO for O_DIRECT */
	if (unlikely(file->f_flags & O_DIRECT)) {
2369 2370 2371 2372 2373
		loff_t endbyte;
		ssize_t written_buffered;

		written = generic_file_direct_write(iocb, iov, &nr_segs, pos,
							ppos, count, ocount);
L
Linus Torvalds 已提交
2374 2375 2376 2377 2378 2379 2380 2381
		if (written < 0 || written == count)
			goto out;
		/*
		 * direct-io write to a hole: fall through to buffered I/O
		 * for completing the rest of the request.
		 */
		pos += written;
		count -= written;
2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395
		written_buffered = generic_file_buffered_write(iocb, iov,
						nr_segs, pos, ppos, count,
						written);
		/*
		 * If generic_file_buffered_write() retuned a synchronous error
		 * then we want to return the number of bytes which were
		 * direct-written, or the error code if that was zero.  Note
		 * that this differs from normal direct-io semantics, which
		 * will return -EFOO even if some bytes were written.
		 */
		if (written_buffered < 0) {
			err = written_buffered;
			goto out;
		}
L
Linus Torvalds 已提交
2396

2397 2398 2399 2400 2401 2402
		/*
		 * We need to ensure that the page cache pages are written to
		 * disk and invalidated to preserve the expected O_DIRECT
		 * semantics.
		 */
		endbyte = pos + written_buffered - written - 1;
M
Mark Fasheh 已提交
2403 2404 2405 2406
		err = do_sync_mapping_range(file->f_mapping, pos, endbyte,
					    SYNC_FILE_RANGE_WAIT_BEFORE|
					    SYNC_FILE_RANGE_WRITE|
					    SYNC_FILE_RANGE_WAIT_AFTER);
2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421
		if (err == 0) {
			written = written_buffered;
			invalidate_mapping_pages(mapping,
						 pos >> PAGE_CACHE_SHIFT,
						 endbyte >> PAGE_CACHE_SHIFT);
		} else {
			/*
			 * We don't know how much we wrote, so just return
			 * the number of bytes which were direct-written
			 */
		}
	} else {
		written = generic_file_buffered_write(iocb, iov, nr_segs,
				pos, ppos, count, written);
	}
L
Linus Torvalds 已提交
2422 2423 2424 2425 2426
out:
	current->backing_dev_info = NULL;
	return written ? written : err;
}

2427 2428
ssize_t generic_file_aio_write_nolock(struct kiocb *iocb,
		const struct iovec *iov, unsigned long nr_segs, loff_t pos)
L
Linus Torvalds 已提交
2429 2430 2431 2432 2433 2434
{
	struct file *file = iocb->ki_filp;
	struct address_space *mapping = file->f_mapping;
	struct inode *inode = mapping->host;
	ssize_t ret;

2435 2436 2437 2438
	BUG_ON(iocb->ki_pos != pos);

	ret = __generic_file_aio_write_nolock(iocb, iov, nr_segs,
			&iocb->ki_pos);
L
Linus Torvalds 已提交
2439 2440

	if (ret > 0 && ((file->f_flags & O_SYNC) || IS_SYNC(inode))) {
2441
		ssize_t err;
L
Linus Torvalds 已提交
2442 2443 2444 2445 2446 2447 2448

		err = sync_page_range_nolock(inode, mapping, pos, ret);
		if (err < 0)
			ret = err;
	}
	return ret;
}
2449
EXPORT_SYMBOL(generic_file_aio_write_nolock);
L
Linus Torvalds 已提交
2450

2451 2452
ssize_t generic_file_aio_write(struct kiocb *iocb, const struct iovec *iov,
		unsigned long nr_segs, loff_t pos)
L
Linus Torvalds 已提交
2453 2454 2455 2456 2457 2458 2459 2460
{
	struct file *file = iocb->ki_filp;
	struct address_space *mapping = file->f_mapping;
	struct inode *inode = mapping->host;
	ssize_t ret;

	BUG_ON(iocb->ki_pos != pos);

2461
	mutex_lock(&inode->i_mutex);
2462 2463
	ret = __generic_file_aio_write_nolock(iocb, iov, nr_segs,
			&iocb->ki_pos);
2464
	mutex_unlock(&inode->i_mutex);
L
Linus Torvalds 已提交
2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476

	if (ret > 0 && ((file->f_flags & O_SYNC) || IS_SYNC(inode))) {
		ssize_t err;

		err = sync_page_range(inode, mapping, pos, ret);
		if (err < 0)
			ret = err;
	}
	return ret;
}
EXPORT_SYMBOL(generic_file_aio_write);

2477 2478 2479 2480 2481 2482 2483 2484 2485 2486
/**
 * try_to_release_page() - release old fs-specific metadata on a page
 *
 * @page: the page which the kernel is trying to free
 * @gfp_mask: memory allocation flags (and I/O mode)
 *
 * The address_space is to try to release any data against the page
 * (presumably at page->private).  If the release was successful, return `1'.
 * Otherwise return zero.
 *
2487 2488 2489
 * This may also be called if PG_fscache is set on a page, indicating that the
 * page is known to the local caching routines.
 *
2490
 * The @gfp_mask argument specifies whether I/O may be performed to release
2491
 * this page (__GFP_IO), and whether the call may block (__GFP_WAIT & __GFP_FS).
2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507
 *
 */
int try_to_release_page(struct page *page, gfp_t gfp_mask)
{
	struct address_space * const mapping = page->mapping;

	BUG_ON(!PageLocked(page));
	if (PageWriteback(page))
		return 0;

	if (mapping && mapping->a_ops->releasepage)
		return mapping->a_ops->releasepage(page, gfp_mask);
	return try_to_free_buffers(page);
}

EXPORT_SYMBOL(try_to_release_page);