filemap.c 76.3 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11
/*
 *	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)
 */
12
#include <linux/export.h>
L
Linus Torvalds 已提交
13
#include <linux/compiler.h>
14
#include <linux/dax.h>
L
Linus Torvalds 已提交
15
#include <linux/fs.h>
16
#include <linux/uaccess.h>
17
#include <linux/capability.h>
L
Linus Torvalds 已提交
18
#include <linux/kernel_stat.h>
19
#include <linux/gfp.h>
L
Linus Torvalds 已提交
20 21 22 23 24 25 26 27
#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
#include <linux/pagevec.h>
#include <linux/blkdev.h>
#include <linux/security.h>
32
#include <linux/cpuset.h>
N
Nick Piggin 已提交
33
#include <linux/hardirq.h> /* for BUG_ON(!in_atomic()) only */
34
#include <linux/hugetlb.h>
35
#include <linux/memcontrol.h>
36
#include <linux/cleancache.h>
37
#include <linux/rmap.h>
38 39
#include "internal.h"

R
Robert Jarzmik 已提交
40 41 42
#define CREATE_TRACE_POINTS
#include <trace/events/filemap.h>

L
Linus Torvalds 已提交
43 44 45
/*
 * FIXME: remove all knowledge of the buffer layer from the core VM
 */
46
#include <linux/buffer_head.h> /* for try_to_free_buffers */
L
Linus Torvalds 已提交
47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64

#include <asm/mman.h>

/*
 * 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:
 *
65
 *  ->i_mmap_rwsem		(truncate_pagecache)
L
Linus Torvalds 已提交
66
 *    ->private_lock		(__free_pte->__set_page_dirty_buffers)
67 68
 *      ->swap_lock		(exclusive_swap_page, others)
 *        ->mapping->tree_lock
L
Linus Torvalds 已提交
69
 *
70
 *  ->i_mutex
71
 *    ->i_mmap_rwsem		(truncate->unmap_mapping_range)
L
Linus Torvalds 已提交
72 73
 *
 *  ->mmap_sem
74
 *    ->i_mmap_rwsem
75
 *      ->page_table_lock or pte_lock	(various, mainly in memory.c)
L
Linus Torvalds 已提交
76 77 78 79 80
 *        ->mapping->tree_lock	(arch-dependent flush_dcache_mmap_lock)
 *
 *  ->mmap_sem
 *    ->lock_page		(access_process_vm)
 *
A
Al Viro 已提交
81
 *  ->i_mutex			(generic_perform_write)
82
 *    ->mmap_sem		(fault_in_pages_readable->do_page_fault)
L
Linus Torvalds 已提交
83
 *
84
 *  bdi->wb.list_lock
85
 *    sb_lock			(fs/fs-writeback.c)
L
Linus Torvalds 已提交
86 87
 *    ->mapping->tree_lock	(__sync_single_inode)
 *
88
 *  ->i_mmap_rwsem
L
Linus Torvalds 已提交
89 90 91
 *    ->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
 *    ->private_lock		(try_to_unmap_one)
 *    ->tree_lock		(try_to_unmap_one)
98 99
 *    ->zone_lru_lock(zone)	(follow_page->mark_page_accessed)
 *    ->zone_lru_lock(zone)	(check_pte_range->isolate_lru_page)
L
Linus Torvalds 已提交
100 101
 *    ->private_lock		(page_remove_rmap->set_page_dirty)
 *    ->tree_lock		(page_remove_rmap->set_page_dirty)
102
 *    bdi.wb->list_lock		(page_remove_rmap->set_page_dirty)
103
 *    ->inode->i_lock		(page_remove_rmap->set_page_dirty)
104
 *    ->memcg->move_lock	(page_remove_rmap->lock_page_memcg)
105
 *    bdi.wb->list_lock		(zap_pte_range->set_page_dirty)
106
 *    ->inode->i_lock		(zap_pte_range->set_page_dirty)
L
Linus Torvalds 已提交
107 108
 *    ->private_lock		(zap_pte_range->__set_page_dirty_buffers)
 *
109
 * ->i_mmap_rwsem
110
 *   ->tasklist_lock            (memory_failure, collect_procs_ao)
L
Linus Torvalds 已提交
111 112
 */

113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137
static int page_cache_tree_insert(struct address_space *mapping,
				  struct page *page, void **shadowp)
{
	struct radix_tree_node *node;
	void **slot;
	int error;

	error = __radix_tree_create(&mapping->page_tree, page->index, 0,
				    &node, &slot);
	if (error)
		return error;
	if (*slot) {
		void *p;

		p = radix_tree_deref_slot_protected(slot, &mapping->tree_lock);
		if (!radix_tree_exceptional_entry(p))
			return -EEXIST;

		mapping->nrexceptional--;
		if (!dax_mapping(mapping)) {
			if (shadowp)
				*shadowp = p;
		} else {
			/* DAX can replace empty locked entry with a hole */
			WARN_ON_ONCE(p !=
138
				dax_radix_locked_entry(0, RADIX_DAX_EMPTY));
139
			/* Wakeup waiters for exceptional entry lock */
140
			dax_wake_mapping_entry_waiter(mapping, page->index, p,
141 142 143
						      false);
		}
	}
144 145
	__radix_tree_replace(&mapping->page_tree, node, slot, page,
			     workingset_update_node, mapping);
146 147 148 149
	mapping->nrpages++;
	return 0;
}

150 151 152
static void page_cache_tree_delete(struct address_space *mapping,
				   struct page *page, void *shadow)
{
153 154 155 156
	int i, nr;

	/* hugetlb pages are represented by one entry in the radix tree */
	nr = PageHuge(page) ? 1 : hpage_nr_pages(page);
157

158 159 160
	VM_BUG_ON_PAGE(!PageLocked(page), page);
	VM_BUG_ON_PAGE(PageTail(page), page);
	VM_BUG_ON_PAGE(nr != 1 && shadow, page);
161

162
	for (i = 0; i < nr; i++) {
163 164 165 166 167 168
		struct radix_tree_node *node;
		void **slot;

		__radix_tree_lookup(&mapping->page_tree, page->index + i,
				    &node, &slot);

169
		VM_BUG_ON_PAGE(!node && nr != 1, page);
170

171 172 173
		radix_tree_clear_tags(&mapping->page_tree, node, slot);
		__radix_tree_replace(&mapping->page_tree, node, slot, shadow,
				     workingset_update_node, mapping);
174
	}
175 176 177 178 179 180 181 182 183 184 185 186

	if (shadow) {
		mapping->nrexceptional += nr;
		/*
		 * Make sure the nrexceptional update is committed before
		 * the nrpages update so that final truncate racing
		 * with reclaim does not see both counters 0 at the
		 * same time and miss a shadow entry.
		 */
		smp_wmb();
	}
	mapping->nrpages -= nr;
187 188
}

L
Linus Torvalds 已提交
189
/*
190
 * Delete a page from the page cache and free it. Caller has to make
L
Linus Torvalds 已提交
191
 * sure the page is locked and that nobody else uses it - or that usage
192
 * is safe.  The caller must hold the mapping's tree_lock.
L
Linus Torvalds 已提交
193
 */
J
Johannes Weiner 已提交
194
void __delete_from_page_cache(struct page *page, void *shadow)
L
Linus Torvalds 已提交
195 196
{
	struct address_space *mapping = page->mapping;
197
	int nr = hpage_nr_pages(page);
L
Linus Torvalds 已提交
198

R
Robert Jarzmik 已提交
199
	trace_mm_filemap_delete_from_page_cache(page);
200 201 202 203 204 205 206 207
	/*
	 * if we're uptodate, flush out into the cleancache, otherwise
	 * invalidate any existing cleancache entries.  We can't leave
	 * stale data around in the cleancache once our page is gone
	 */
	if (PageUptodate(page) && PageMappedToDisk(page))
		cleancache_put_page(page);
	else
208
		cleancache_invalidate_page(mapping, page);
209

210
	VM_BUG_ON_PAGE(PageTail(page), page);
211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230
	VM_BUG_ON_PAGE(page_mapped(page), page);
	if (!IS_ENABLED(CONFIG_DEBUG_VM) && unlikely(page_mapped(page))) {
		int mapcount;

		pr_alert("BUG: Bad page cache in process %s  pfn:%05lx\n",
			 current->comm, page_to_pfn(page));
		dump_page(page, "still mapped when deleted");
		dump_stack();
		add_taint(TAINT_BAD_PAGE, LOCKDEP_NOW_UNRELIABLE);

		mapcount = page_mapcount(page);
		if (mapping_exiting(mapping) &&
		    page_count(page) >= mapcount + 2) {
			/*
			 * All vmas have already been torn down, so it's
			 * a good bet that actually the page is unmapped,
			 * and we'd prefer not to leak it: if we're wrong,
			 * some other bad page check should catch it later.
			 */
			page_mapcount_reset(page);
231
			page_ref_sub(page, mapcount);
232 233 234
		}
	}

235 236
	page_cache_tree_delete(mapping, page, shadow);

L
Linus Torvalds 已提交
237
	page->mapping = NULL;
238
	/* Leave page->index set: truncation lookup relies upon it */
239

240 241
	/* hugetlb pages do not participate in page cache accounting. */
	if (!PageHuge(page))
242
		__mod_node_page_state(page_pgdat(page), NR_FILE_PAGES, -nr);
243
	if (PageSwapBacked(page)) {
244
		__mod_node_page_state(page_pgdat(page), NR_SHMEM, -nr);
245
		if (PageTransHuge(page))
246
			__dec_node_page_state(page, NR_SHMEM_THPS);
247 248 249
	} else {
		VM_BUG_ON_PAGE(PageTransHuge(page) && !PageHuge(page), page);
	}
250 251

	/*
252 253
	 * At this point page must be either written or cleaned by truncate.
	 * Dirty page here signals a bug and loss of unwritten data.
254
	 *
255 256 257
	 * This fixes dirty accounting after removing the page entirely but
	 * leaves PageDirty set: it has no effect for truncated page and
	 * anyway will be cleared before returning page into buddy allocator.
258
	 */
259
	if (WARN_ON_ONCE(PageDirty(page)))
J
Johannes Weiner 已提交
260
		account_page_cleaned(page, mapping, inode_to_wb(mapping->host));
L
Linus Torvalds 已提交
261 262
}

263 264 265 266 267 268 269 270 271
/**
 * delete_from_page_cache - delete page from page cache
 * @page: the page which the kernel is trying to remove from page cache
 *
 * This must be called only on pages that have been verified to be in the page
 * cache and locked.  It will never put the page into the free list, the caller
 * has a reference on the page.
 */
void delete_from_page_cache(struct page *page)
L
Linus Torvalds 已提交
272
{
273
	struct address_space *mapping = page_mapping(page);
274
	unsigned long flags;
275
	void (*freepage)(struct page *);
L
Linus Torvalds 已提交
276

M
Matt Mackall 已提交
277
	BUG_ON(!PageLocked(page));
L
Linus Torvalds 已提交
278

279
	freepage = mapping->a_ops->freepage;
280 281

	spin_lock_irqsave(&mapping->tree_lock, flags);
J
Johannes Weiner 已提交
282
	__delete_from_page_cache(page, NULL);
283
	spin_unlock_irqrestore(&mapping->tree_lock, flags);
284 285 286

	if (freepage)
		freepage(page);
287 288 289 290 291 292 293

	if (PageTransHuge(page) && !PageHuge(page)) {
		page_ref_sub(page, HPAGE_PMD_NR);
		VM_BUG_ON_PAGE(page_count(page) <= 0, page);
	} else {
		put_page(page);
	}
294 295 296
}
EXPORT_SYMBOL(delete_from_page_cache);

297
int filemap_check_errors(struct address_space *mapping)
298 299 300
{
	int ret = 0;
	/* Check for outstanding write errors */
301 302
	if (test_bit(AS_ENOSPC, &mapping->flags) &&
	    test_and_clear_bit(AS_ENOSPC, &mapping->flags))
303
		ret = -ENOSPC;
304 305
	if (test_bit(AS_EIO, &mapping->flags) &&
	    test_and_clear_bit(AS_EIO, &mapping->flags))
306 307 308
		ret = -EIO;
	return ret;
}
309
EXPORT_SYMBOL(filemap_check_errors);
310

L
Linus Torvalds 已提交
311
/**
312
 * __filemap_fdatawrite_range - start writeback on mapping dirty pages in range
313 314
 * @mapping:	address space structure to write
 * @start:	offset in bytes where the range starts
315
 * @end:	offset in bytes where the range ends (inclusive)
316
 * @sync_mode:	enable synchronous operation
L
Linus Torvalds 已提交
317
 *
318 319 320
 * 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
321
 * If sync_mode is WB_SYNC_ALL then this is a "data integrity" operation, as
322
 * opposed to a regular memory cleansing writeback.  The difference between
L
Linus Torvalds 已提交
323 324 325
 * these two operations is that if a dirty page/buffer is encountered, it must
 * be waited upon, and not just skipped over.
 */
326 327
int __filemap_fdatawrite_range(struct address_space *mapping, loff_t start,
				loff_t end, int sync_mode)
L
Linus Torvalds 已提交
328 329 330 331
{
	int ret;
	struct writeback_control wbc = {
		.sync_mode = sync_mode,
332
		.nr_to_write = LONG_MAX,
333 334
		.range_start = start,
		.range_end = end,
L
Linus Torvalds 已提交
335 336 337 338 339
	};

	if (!mapping_cap_writeback_dirty(mapping))
		return 0;

340
	wbc_attach_fdatawrite_inode(&wbc, mapping->host);
L
Linus Torvalds 已提交
341
	ret = do_writepages(mapping, &wbc);
342
	wbc_detach_inode(&wbc);
L
Linus Torvalds 已提交
343 344 345 346 347 348
	return ret;
}

static inline int __filemap_fdatawrite(struct address_space *mapping,
	int sync_mode)
{
349
	return __filemap_fdatawrite_range(mapping, 0, LLONG_MAX, sync_mode);
L
Linus Torvalds 已提交
350 351 352 353 354 355 356 357
}

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

358
int filemap_fdatawrite_range(struct address_space *mapping, loff_t start,
359
				loff_t end)
L
Linus Torvalds 已提交
360 361 362
{
	return __filemap_fdatawrite_range(mapping, start, end, WB_SYNC_ALL);
}
363
EXPORT_SYMBOL(filemap_fdatawrite_range);
L
Linus Torvalds 已提交
364

365 366 367 368
/**
 * filemap_flush - mostly a non-blocking flush
 * @mapping:	target address_space
 *
L
Linus Torvalds 已提交
369 370 371 372 373 374 375 376 377
 * 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);

378 379
static int __filemap_fdatawait_range(struct address_space *mapping,
				     loff_t start_byte, loff_t end_byte)
L
Linus Torvalds 已提交
380
{
381 382
	pgoff_t index = start_byte >> PAGE_SHIFT;
	pgoff_t end = end_byte >> PAGE_SHIFT;
L
Linus Torvalds 已提交
383 384
	struct pagevec pvec;
	int nr_pages;
385
	int ret = 0;
L
Linus Torvalds 已提交
386

387
	if (end_byte < start_byte)
388
		goto out;
L
Linus Torvalds 已提交
389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404

	pagevec_init(&pvec, 0);
	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);
405
			if (TestClearPageError(page))
L
Linus Torvalds 已提交
406 407 408 409 410
				ret = -EIO;
		}
		pagevec_release(&pvec);
		cond_resched();
	}
411
out:
412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434
	return ret;
}

/**
 * filemap_fdatawait_range - wait for writeback to complete
 * @mapping:		address space structure to wait for
 * @start_byte:		offset in bytes where the range starts
 * @end_byte:		offset in bytes where the range ends (inclusive)
 *
 * Walk the list of under-writeback pages of the given address space
 * in the given range and wait for all of them.  Check error status of
 * the address space and return it.
 *
 * Since the error status of the address space is cleared by this function,
 * callers are responsible for checking the return value and handling and/or
 * reporting the error.
 */
int filemap_fdatawait_range(struct address_space *mapping, loff_t start_byte,
			    loff_t end_byte)
{
	int ret, ret2;

	ret = __filemap_fdatawait_range(mapping, start_byte, end_byte);
435 436 437
	ret2 = filemap_check_errors(mapping);
	if (!ret)
		ret = ret2;
L
Linus Torvalds 已提交
438 439 440

	return ret;
}
441 442
EXPORT_SYMBOL(filemap_fdatawait_range);

443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464
/**
 * filemap_fdatawait_keep_errors - wait for writeback without clearing errors
 * @mapping: address space structure to wait for
 *
 * Walk the list of under-writeback pages of the given address space
 * and wait for all of them.  Unlike filemap_fdatawait(), this function
 * does not clear error status of the address space.
 *
 * Use this function if callers don't handle errors themselves.  Expected
 * call sites are system-wide / filesystem-wide data flushers: e.g. sync(2),
 * fsfreeze(8)
 */
void filemap_fdatawait_keep_errors(struct address_space *mapping)
{
	loff_t i_size = i_size_read(mapping->host);

	if (i_size == 0)
		return;

	__filemap_fdatawait_range(mapping, 0, i_size - 1);
}

L
Linus Torvalds 已提交
465
/**
466
 * filemap_fdatawait - wait for all under-writeback pages to complete
L
Linus Torvalds 已提交
467
 * @mapping: address space structure to wait for
468 469
 *
 * Walk the list of under-writeback pages of the given address space
470 471 472 473 474 475
 * and wait for all of them.  Check error status of the address space
 * and return it.
 *
 * Since the error status of the address space is cleared by this function,
 * callers are responsible for checking the return value and handling and/or
 * reporting the error.
L
Linus Torvalds 已提交
476 477 478 479 480 481 482 483
 */
int filemap_fdatawait(struct address_space *mapping)
{
	loff_t i_size = i_size_read(mapping->host);

	if (i_size == 0)
		return 0;

484
	return filemap_fdatawait_range(mapping, 0, i_size - 1);
L
Linus Torvalds 已提交
485 486 487 488 489
}
EXPORT_SYMBOL(filemap_fdatawait);

int filemap_write_and_wait(struct address_space *mapping)
{
490
	int err = 0;
L
Linus Torvalds 已提交
491

492 493
	if ((!dax_mapping(mapping) && mapping->nrpages) ||
	    (dax_mapping(mapping) && mapping->nrexceptional)) {
494 495 496 497 498 499 500 501 502 503 504 505
		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;
		}
506 507
	} else {
		err = filemap_check_errors(mapping);
L
Linus Torvalds 已提交
508
	}
509
	return err;
L
Linus Torvalds 已提交
510
}
511
EXPORT_SYMBOL(filemap_write_and_wait);
L
Linus Torvalds 已提交
512

513 514 515 516 517 518
/**
 * 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)
 *
519 520 521 522 523
 * 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
524 525 526
int filemap_write_and_wait_range(struct address_space *mapping,
				 loff_t lstart, loff_t lend)
{
527
	int err = 0;
L
Linus Torvalds 已提交
528

529 530
	if ((!dax_mapping(mapping) && mapping->nrpages) ||
	    (dax_mapping(mapping) && mapping->nrexceptional)) {
531 532 533 534
		err = __filemap_fdatawrite_range(mapping, lstart, lend,
						 WB_SYNC_ALL);
		/* See comment of filemap_write_and_wait() */
		if (err != -EIO) {
535 536
			int err2 = filemap_fdatawait_range(mapping,
						lstart, lend);
537 538 539
			if (!err)
				err = err2;
		}
540 541
	} else {
		err = filemap_check_errors(mapping);
L
Linus Torvalds 已提交
542
	}
543
	return err;
L
Linus Torvalds 已提交
544
}
545
EXPORT_SYMBOL(filemap_write_and_wait_range);
L
Linus Torvalds 已提交
546

547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565
/**
 * replace_page_cache_page - replace a pagecache page with a new one
 * @old:	page to be replaced
 * @new:	page to replace with
 * @gfp_mask:	allocation mode
 *
 * This function replaces a page in the pagecache with a new one.  On
 * success it acquires the pagecache reference for the new page and
 * drops it for the old page.  Both the old and new pages must be
 * locked.  This function does not add the new page to the LRU, the
 * caller must do that.
 *
 * The remove + add is atomic.  The only way this function can fail is
 * memory allocation failure.
 */
int replace_page_cache_page(struct page *old, struct page *new, gfp_t gfp_mask)
{
	int error;

566 567 568
	VM_BUG_ON_PAGE(!PageLocked(old), old);
	VM_BUG_ON_PAGE(!PageLocked(new), new);
	VM_BUG_ON_PAGE(new->mapping, new);
569 570 571 572 573

	error = radix_tree_preload(gfp_mask & ~__GFP_HIGHMEM);
	if (!error) {
		struct address_space *mapping = old->mapping;
		void (*freepage)(struct page *);
574
		unsigned long flags;
575 576 577 578

		pgoff_t offset = old->index;
		freepage = mapping->a_ops->freepage;

579
		get_page(new);
580 581 582
		new->mapping = mapping;
		new->index = offset;

583
		spin_lock_irqsave(&mapping->tree_lock, flags);
J
Johannes Weiner 已提交
584
		__delete_from_page_cache(old, NULL);
585
		error = page_cache_tree_insert(mapping, new, NULL);
586
		BUG_ON(error);
587 588 589 590 591

		/*
		 * hugetlb pages do not participate in page cache accounting.
		 */
		if (!PageHuge(new))
592
			__inc_node_page_state(new, NR_FILE_PAGES);
593
		if (PageSwapBacked(new))
594
			__inc_node_page_state(new, NR_SHMEM);
595
		spin_unlock_irqrestore(&mapping->tree_lock, flags);
596
		mem_cgroup_migrate(old, new);
597 598 599
		radix_tree_preload_end();
		if (freepage)
			freepage(old);
600
		put_page(old);
601 602 603 604 605 606
	}

	return error;
}
EXPORT_SYMBOL_GPL(replace_page_cache_page);

607 608 609 610
static int __add_to_page_cache_locked(struct page *page,
				      struct address_space *mapping,
				      pgoff_t offset, gfp_t gfp_mask,
				      void **shadowp)
L
Linus Torvalds 已提交
611
{
612 613
	int huge = PageHuge(page);
	struct mem_cgroup *memcg;
N
Nick Piggin 已提交
614 615
	int error;

616 617
	VM_BUG_ON_PAGE(!PageLocked(page), page);
	VM_BUG_ON_PAGE(PageSwapBacked(page), page);
N
Nick Piggin 已提交
618

619 620
	if (!huge) {
		error = mem_cgroup_try_charge(page, current->mm,
621
					      gfp_mask, &memcg, false);
622 623 624
		if (error)
			return error;
	}
L
Linus Torvalds 已提交
625

626
	error = radix_tree_maybe_preload(gfp_mask & ~__GFP_HIGHMEM);
627
	if (error) {
628
		if (!huge)
629
			mem_cgroup_cancel_charge(page, memcg, false);
630 631 632
		return error;
	}

633
	get_page(page);
634 635 636 637
	page->mapping = mapping;
	page->index = offset;

	spin_lock_irq(&mapping->tree_lock);
638
	error = page_cache_tree_insert(mapping, page, shadowp);
639 640 641
	radix_tree_preload_end();
	if (unlikely(error))
		goto err_insert;
642 643 644

	/* hugetlb pages do not participate in page cache accounting. */
	if (!huge)
645
		__inc_node_page_state(page, NR_FILE_PAGES);
646
	spin_unlock_irq(&mapping->tree_lock);
647
	if (!huge)
648
		mem_cgroup_commit_charge(page, memcg, false, false);
649 650 651 652 653 654
	trace_mm_filemap_add_to_page_cache(page);
	return 0;
err_insert:
	page->mapping = NULL;
	/* Leave page->index set: truncation relies upon it */
	spin_unlock_irq(&mapping->tree_lock);
655
	if (!huge)
656
		mem_cgroup_cancel_charge(page, memcg, false);
657
	put_page(page);
L
Linus Torvalds 已提交
658 659
	return error;
}
660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676

/**
 * add_to_page_cache_locked - add a locked page to the pagecache
 * @page:	page to add
 * @mapping:	the page's address_space
 * @offset:	page index
 * @gfp_mask:	page allocation mode
 *
 * This function is used to add a page to the pagecache. It must be locked.
 * This function does not add the page to the LRU.  The caller must do that.
 */
int add_to_page_cache_locked(struct page *page, struct address_space *mapping,
		pgoff_t offset, gfp_t gfp_mask)
{
	return __add_to_page_cache_locked(page, mapping, offset,
					  gfp_mask, NULL);
}
N
Nick Piggin 已提交
677
EXPORT_SYMBOL(add_to_page_cache_locked);
L
Linus Torvalds 已提交
678 679

int add_to_page_cache_lru(struct page *page, struct address_space *mapping,
A
Al Viro 已提交
680
				pgoff_t offset, gfp_t gfp_mask)
L
Linus Torvalds 已提交
681
{
682
	void *shadow = NULL;
683 684
	int ret;

685
	__SetPageLocked(page);
686 687 688
	ret = __add_to_page_cache_locked(page, mapping, offset,
					 gfp_mask, &shadow);
	if (unlikely(ret))
689
		__ClearPageLocked(page);
690 691 692 693 694
	else {
		/*
		 * The page might have been evicted from cache only
		 * recently, in which case it should be activated like
		 * any other repeatedly accessed page.
695 696 697
		 * The exception is pages getting rewritten; evicting other
		 * data from the working set, only to cache data that will
		 * get overwritten with something else, is a waste of memory.
698
		 */
699 700
		if (!(gfp_mask & __GFP_WRITE) &&
		    shadow && workingset_refault(shadow)) {
701 702 703 704 705 706
			SetPageActive(page);
			workingset_activation(page);
		} else
			ClearPageActive(page);
		lru_cache_add(page);
	}
L
Linus Torvalds 已提交
707 708
	return ret;
}
709
EXPORT_SYMBOL_GPL(add_to_page_cache_lru);
L
Linus Torvalds 已提交
710

711
#ifdef CONFIG_NUMA
712
struct page *__page_cache_alloc(gfp_t gfp)
713
{
714 715 716
	int n;
	struct page *page;

717
	if (cpuset_do_page_mem_spread()) {
718 719
		unsigned int cpuset_mems_cookie;
		do {
720
			cpuset_mems_cookie = read_mems_allowed_begin();
721
			n = cpuset_mem_spread_node();
722
			page = __alloc_pages_node(n, gfp, 0);
723
		} while (!page && read_mems_allowed_retry(cpuset_mems_cookie));
724

725
		return page;
726
	}
727
	return alloc_pages(gfp, 0);
728
}
729
EXPORT_SYMBOL(__page_cache_alloc);
730 731
#endif

L
Linus Torvalds 已提交
732 733 734 735 736 737 738 739 740 741
/*
 * 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.
 */
N
NeilBrown 已提交
742
wait_queue_head_t *page_waitqueue(struct page *page)
L
Linus Torvalds 已提交
743
{
744
	return bit_waitqueue(page, 0);
L
Linus Torvalds 已提交
745
}
N
NeilBrown 已提交
746
EXPORT_SYMBOL(page_waitqueue);
L
Linus Torvalds 已提交
747

H
Harvey Harrison 已提交
748
void wait_on_page_bit(struct page *page, int bit_nr)
L
Linus Torvalds 已提交
749 750 751 752
{
	DEFINE_WAIT_BIT(wait, &page->flags, bit_nr);

	if (test_bit(bit_nr, &page->flags))
753
		__wait_on_bit(page_waitqueue(page), &wait, bit_wait_io,
L
Linus Torvalds 已提交
754 755 756 757
							TASK_UNINTERRUPTIBLE);
}
EXPORT_SYMBOL(wait_on_page_bit);

758 759 760 761 762 763 764 765
int wait_on_page_bit_killable(struct page *page, int bit_nr)
{
	DEFINE_WAIT_BIT(wait, &page->flags, bit_nr);

	if (!test_bit(bit_nr, &page->flags))
		return 0;

	return __wait_on_bit(page_waitqueue(page), &wait,
766
			     bit_wait_io, TASK_KILLABLE);
767 768
}

769 770 771 772 773 774 775 776 777 778 779 780 781
int wait_on_page_bit_killable_timeout(struct page *page,
				       int bit_nr, unsigned long timeout)
{
	DEFINE_WAIT_BIT(wait, &page->flags, bit_nr);

	wait.key.timeout = jiffies + timeout;
	if (!test_bit(bit_nr, &page->flags))
		return 0;
	return __wait_on_bit(page_waitqueue(page), &wait,
			     bit_wait_io_timeout, TASK_KILLABLE);
}
EXPORT_SYMBOL_GPL(wait_on_page_bit_killable_timeout);

782 783
/**
 * add_page_wait_queue - Add an arbitrary waiter to a page's wait queue
R
Randy Dunlap 已提交
784 785
 * @page: Page defining the wait queue of interest
 * @waiter: Waiter to add to the queue
786 787 788 789 790 791 792 793 794 795 796 797 798 799
 *
 * 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
800
/**
801
 * unlock_page - unlock a locked page
L
Linus Torvalds 已提交
802 803 804 805
 * @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
806
 * mechanism between PageLocked pages and PageWriteback pages is shared.
L
Linus Torvalds 已提交
807 808
 * But that's OK - sleepers in wait_on_page_writeback() just go back to sleep.
 *
N
Nick Piggin 已提交
809 810
 * 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
811
 */
H
Harvey Harrison 已提交
812
void unlock_page(struct page *page)
L
Linus Torvalds 已提交
813
{
814
	page = compound_head(page);
815
	VM_BUG_ON_PAGE(!PageLocked(page), page);
N
Nick Piggin 已提交
816
	clear_bit_unlock(PG_locked, &page->flags);
817
	smp_mb__after_atomic();
L
Linus Torvalds 已提交
818 819 820 821
	wake_up_page(page, PG_locked);
}
EXPORT_SYMBOL(unlock_page);

822 823 824
/**
 * end_page_writeback - end writeback against a page
 * @page: the page
L
Linus Torvalds 已提交
825 826 827
 */
void end_page_writeback(struct page *page)
{
828 829 830 831 832 833 834 835 836
	/*
	 * TestClearPageReclaim could be used here but it is an atomic
	 * operation and overkill in this particular case. Failing to
	 * shuffle a page marked for immediate reclaim is too mild to
	 * justify taking an atomic operation penalty at the end of
	 * ever page writeback.
	 */
	if (PageReclaim(page)) {
		ClearPageReclaim(page);
837
		rotate_reclaimable_page(page);
838
	}
839 840 841 842

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

843
	smp_mb__after_atomic();
L
Linus Torvalds 已提交
844 845 846 847
	wake_up_page(page, PG_writeback);
}
EXPORT_SYMBOL(end_page_writeback);

848 849 850 851
/*
 * After completing I/O on a page, call this routine to update the page
 * flags appropriately
 */
852
void page_endio(struct page *page, bool is_write, int err)
853
{
854
	if (!is_write) {
855 856 857 858 859 860 861
		if (!err) {
			SetPageUptodate(page);
		} else {
			ClearPageUptodate(page);
			SetPageError(page);
		}
		unlock_page(page);
862
	} else {
863 864 865 866 867 868 869 870 871 872
		if (err) {
			SetPageError(page);
			if (page->mapping)
				mapping_set_error(page->mapping, err);
		}
		end_page_writeback(page);
	}
}
EXPORT_SYMBOL_GPL(page_endio);

873 874 875
/**
 * __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
876
 */
H
Harvey Harrison 已提交
877
void __lock_page(struct page *page)
L
Linus Torvalds 已提交
878
{
879 880
	struct page *page_head = compound_head(page);
	DEFINE_WAIT_BIT(wait, &page_head->flags, PG_locked);
L
Linus Torvalds 已提交
881

882
	__wait_on_bit_lock(page_waitqueue(page_head), &wait, bit_wait_io,
L
Linus Torvalds 已提交
883 884 885 886
							TASK_UNINTERRUPTIBLE);
}
EXPORT_SYMBOL(__lock_page);

H
Harvey Harrison 已提交
887
int __lock_page_killable(struct page *page)
M
Matthew Wilcox 已提交
888
{
889 890
	struct page *page_head = compound_head(page);
	DEFINE_WAIT_BIT(wait, &page_head->flags, PG_locked);
M
Matthew Wilcox 已提交
891

892
	return __wait_on_bit_lock(page_waitqueue(page_head), &wait,
893
					bit_wait_io, TASK_KILLABLE);
M
Matthew Wilcox 已提交
894
}
895
EXPORT_SYMBOL_GPL(__lock_page_killable);
M
Matthew Wilcox 已提交
896

897 898 899 900 901 902 903 904 905 906 907
/*
 * Return values:
 * 1 - page is locked; mmap_sem is still held.
 * 0 - page is not locked.
 *     mmap_sem has been released (up_read()), unless flags had both
 *     FAULT_FLAG_ALLOW_RETRY and FAULT_FLAG_RETRY_NOWAIT set, in
 *     which case mmap_sem is still held.
 *
 * If neither ALLOW_RETRY nor KILLABLE are set, will always return 1
 * with the page locked and the mmap_sem unperturbed.
 */
908 909 910
int __lock_page_or_retry(struct page *page, struct mm_struct *mm,
			 unsigned int flags)
{
911 912 913 914 915 916 917 918 919 920 921 922
	if (flags & FAULT_FLAG_ALLOW_RETRY) {
		/*
		 * CAUTION! In this case, mmap_sem is not released
		 * even though return 0.
		 */
		if (flags & FAULT_FLAG_RETRY_NOWAIT)
			return 0;

		up_read(&mm->mmap_sem);
		if (flags & FAULT_FLAG_KILLABLE)
			wait_on_page_locked_killable(page);
		else
923
			wait_on_page_locked(page);
924
		return 0;
925 926 927 928 929 930 931 932 933 934 935 936
	} else {
		if (flags & FAULT_FLAG_KILLABLE) {
			int ret;

			ret = __lock_page_killable(page);
			if (ret) {
				up_read(&mm->mmap_sem);
				return 0;
			}
		} else
			__lock_page(page);
		return 1;
937 938 939
	}
}

940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966
/**
 * page_cache_next_hole - find the next hole (not-present entry)
 * @mapping: mapping
 * @index: index
 * @max_scan: maximum range to search
 *
 * Search the set [index, min(index+max_scan-1, MAX_INDEX)] for the
 * lowest indexed hole.
 *
 * Returns: the index of the hole if found, otherwise returns an index
 * outside of the set specified (in which case 'return - index >=
 * max_scan' will be true). In rare cases of index wrap-around, 0 will
 * be returned.
 *
 * page_cache_next_hole may be called under rcu_read_lock. However,
 * like radix_tree_gang_lookup, this will not atomically search a
 * snapshot of the tree at a single point in time. For example, if a
 * hole is created at index 5, then subsequently a hole is created at
 * index 10, page_cache_next_hole covering both indexes may return 10
 * if called under rcu_read_lock.
 */
pgoff_t page_cache_next_hole(struct address_space *mapping,
			     pgoff_t index, unsigned long max_scan)
{
	unsigned long i;

	for (i = 0; i < max_scan; i++) {
967 968 969 970
		struct page *page;

		page = radix_tree_lookup(&mapping->page_tree, index);
		if (!page || radix_tree_exceptional_entry(page))
971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007
			break;
		index++;
		if (index == 0)
			break;
	}

	return index;
}
EXPORT_SYMBOL(page_cache_next_hole);

/**
 * page_cache_prev_hole - find the prev hole (not-present entry)
 * @mapping: mapping
 * @index: index
 * @max_scan: maximum range to search
 *
 * Search backwards in the range [max(index-max_scan+1, 0), index] for
 * the first hole.
 *
 * Returns: the index of the hole if found, otherwise returns an index
 * outside of the set specified (in which case 'index - return >=
 * max_scan' will be true). In rare cases of wrap-around, ULONG_MAX
 * will be returned.
 *
 * page_cache_prev_hole may be called under rcu_read_lock. However,
 * like radix_tree_gang_lookup, this will not atomically search a
 * snapshot of the tree at a single point in time. For example, if a
 * hole is created at index 10, then subsequently a hole is created at
 * index 5, page_cache_prev_hole covering both indexes may return 5 if
 * called under rcu_read_lock.
 */
pgoff_t page_cache_prev_hole(struct address_space *mapping,
			     pgoff_t index, unsigned long max_scan)
{
	unsigned long i;

	for (i = 0; i < max_scan; i++) {
1008 1009 1010 1011
		struct page *page;

		page = radix_tree_lookup(&mapping->page_tree, index);
		if (!page || radix_tree_exceptional_entry(page))
1012 1013 1014 1015 1016 1017 1018 1019 1020 1021
			break;
		index--;
		if (index == ULONG_MAX)
			break;
	}

	return index;
}
EXPORT_SYMBOL(page_cache_prev_hole);

1022
/**
1023
 * find_get_entry - find and get a page cache entry
1024
 * @mapping: the address_space to search
1025 1026 1027 1028
 * @offset: the page cache index
 *
 * Looks up the page cache slot at @mapping & @offset.  If there is a
 * page cache page, it is returned with an increased refcount.
1029
 *
1030 1031
 * If the slot holds a shadow entry of a previously evicted page, or a
 * swap entry from shmem/tmpfs, it is returned.
1032 1033
 *
 * Otherwise, %NULL is returned.
L
Linus Torvalds 已提交
1034
 */
1035
struct page *find_get_entry(struct address_space *mapping, pgoff_t offset)
L
Linus Torvalds 已提交
1036
{
N
Nick Piggin 已提交
1037
	void **pagep;
1038
	struct page *head, *page;
L
Linus Torvalds 已提交
1039

N
Nick Piggin 已提交
1040 1041 1042 1043 1044 1045
	rcu_read_lock();
repeat:
	page = NULL;
	pagep = radix_tree_lookup_slot(&mapping->page_tree, offset);
	if (pagep) {
		page = radix_tree_deref_slot(pagep);
N
Nick Piggin 已提交
1046 1047
		if (unlikely(!page))
			goto out;
1048
		if (radix_tree_exception(page)) {
1049 1050 1051
			if (radix_tree_deref_retry(page))
				goto repeat;
			/*
1052 1053 1054
			 * A shadow entry of a recently evicted page,
			 * or a swap entry from shmem/tmpfs.  Return
			 * it without attempting to raise page count.
1055 1056
			 */
			goto out;
1057
		}
1058 1059 1060 1061 1062 1063 1064 1065

		head = compound_head(page);
		if (!page_cache_get_speculative(head))
			goto repeat;

		/* The page was split under us? */
		if (compound_head(page) != head) {
			put_page(head);
N
Nick Piggin 已提交
1066
			goto repeat;
1067
		}
N
Nick Piggin 已提交
1068 1069 1070 1071 1072 1073 1074

		/*
		 * Has the page moved?
		 * This is part of the lockless pagecache protocol. See
		 * include/linux/pagemap.h for details.
		 */
		if (unlikely(page != *pagep)) {
1075
			put_page(head);
N
Nick Piggin 已提交
1076 1077 1078
			goto repeat;
		}
	}
N
Nick Piggin 已提交
1079
out:
N
Nick Piggin 已提交
1080 1081
	rcu_read_unlock();

L
Linus Torvalds 已提交
1082 1083
	return page;
}
1084
EXPORT_SYMBOL(find_get_entry);
L
Linus Torvalds 已提交
1085

1086 1087 1088 1089 1090 1091 1092 1093 1094
/**
 * find_lock_entry - locate, pin and lock a page cache entry
 * @mapping: the address_space to search
 * @offset: the page cache index
 *
 * Looks up the page cache slot at @mapping & @offset.  If there is a
 * page cache page, it is returned locked and with an increased
 * refcount.
 *
1095 1096
 * If the slot holds a shadow entry of a previously evicted page, or a
 * swap entry from shmem/tmpfs, it is returned.
1097 1098 1099 1100 1101 1102
 *
 * Otherwise, %NULL is returned.
 *
 * find_lock_entry() may sleep.
 */
struct page *find_lock_entry(struct address_space *mapping, pgoff_t offset)
L
Linus Torvalds 已提交
1103 1104 1105 1106
{
	struct page *page;

repeat:
1107
	page = find_get_entry(mapping, offset);
1108
	if (page && !radix_tree_exception(page)) {
N
Nick Piggin 已提交
1109 1110
		lock_page(page);
		/* Has the page been truncated? */
1111
		if (unlikely(page_mapping(page) != mapping)) {
N
Nick Piggin 已提交
1112
			unlock_page(page);
1113
			put_page(page);
N
Nick Piggin 已提交
1114
			goto repeat;
L
Linus Torvalds 已提交
1115
		}
1116
		VM_BUG_ON_PAGE(page_to_pgoff(page) != offset, page);
L
Linus Torvalds 已提交
1117 1118 1119
	}
	return page;
}
1120 1121 1122
EXPORT_SYMBOL(find_lock_entry);

/**
1123
 * pagecache_get_page - find and get a page reference
1124 1125
 * @mapping: the address_space to search
 * @offset: the page index
1126
 * @fgp_flags: PCG flags
1127
 * @gfp_mask: gfp mask to use for the page cache data page allocation
1128
 *
1129
 * Looks up the page cache slot at @mapping & @offset.
L
Linus Torvalds 已提交
1130
 *
1131
 * PCG flags modify how the page is returned.
1132
 *
1133 1134 1135
 * FGP_ACCESSED: the page will be marked accessed
 * FGP_LOCK: Page is return locked
 * FGP_CREAT: If page is not present then a new page is allocated using
1136 1137 1138
 *		@gfp_mask and added to the page cache and the VM's LRU
 *		list. The page is returned locked and with an increased
 *		refcount. Otherwise, %NULL is returned.
L
Linus Torvalds 已提交
1139
 *
1140 1141
 * If FGP_LOCK or FGP_CREAT are specified then the function may sleep even
 * if the GFP flags specified for FGP_CREAT are atomic.
L
Linus Torvalds 已提交
1142
 *
1143
 * If there is a page cache page, it is returned with an increased refcount.
L
Linus Torvalds 已提交
1144
 */
1145
struct page *pagecache_get_page(struct address_space *mapping, pgoff_t offset,
1146
	int fgp_flags, gfp_t gfp_mask)
L
Linus Torvalds 已提交
1147
{
N
Nick Piggin 已提交
1148
	struct page *page;
1149

L
Linus Torvalds 已提交
1150
repeat:
1151 1152 1153 1154 1155 1156 1157 1158 1159
	page = find_get_entry(mapping, offset);
	if (radix_tree_exceptional_entry(page))
		page = NULL;
	if (!page)
		goto no_page;

	if (fgp_flags & FGP_LOCK) {
		if (fgp_flags & FGP_NOWAIT) {
			if (!trylock_page(page)) {
1160
				put_page(page);
1161 1162 1163 1164 1165 1166 1167 1168 1169
				return NULL;
			}
		} else {
			lock_page(page);
		}

		/* Has the page been truncated? */
		if (unlikely(page->mapping != mapping)) {
			unlock_page(page);
1170
			put_page(page);
1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182
			goto repeat;
		}
		VM_BUG_ON_PAGE(page->index != offset, page);
	}

	if (page && (fgp_flags & FGP_ACCESSED))
		mark_page_accessed(page);

no_page:
	if (!page && (fgp_flags & FGP_CREAT)) {
		int err;
		if ((fgp_flags & FGP_WRITE) && mapping_cap_account_dirty(mapping))
1183 1184 1185
			gfp_mask |= __GFP_WRITE;
		if (fgp_flags & FGP_NOFS)
			gfp_mask &= ~__GFP_FS;
1186

1187
		page = __page_cache_alloc(gfp_mask);
N
Nick Piggin 已提交
1188 1189
		if (!page)
			return NULL;
1190 1191 1192 1193

		if (WARN_ON_ONCE(!(fgp_flags & FGP_LOCK)))
			fgp_flags |= FGP_LOCK;

1194
		/* Init accessed so avoid atomic mark_page_accessed later */
1195
		if (fgp_flags & FGP_ACCESSED)
1196
			__SetPageReferenced(page);
1197

1198 1199
		err = add_to_page_cache_lru(page, mapping, offset,
				gfp_mask & GFP_RECLAIM_MASK);
N
Nick Piggin 已提交
1200
		if (unlikely(err)) {
1201
			put_page(page);
N
Nick Piggin 已提交
1202 1203 1204
			page = NULL;
			if (err == -EEXIST)
				goto repeat;
L
Linus Torvalds 已提交
1205 1206
		}
	}
1207

L
Linus Torvalds 已提交
1208 1209
	return page;
}
1210
EXPORT_SYMBOL(pagecache_get_page);
L
Linus Torvalds 已提交
1211

1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228
/**
 * find_get_entries - gang pagecache lookup
 * @mapping:	The address_space to search
 * @start:	The starting page cache index
 * @nr_entries:	The maximum number of entries
 * @entries:	Where the resulting entries are placed
 * @indices:	The cache indices corresponding to the entries in @entries
 *
 * find_get_entries() will search for and return a group of up to
 * @nr_entries entries in the mapping.  The entries are placed at
 * @entries.  find_get_entries() takes a reference against any actual
 * pages it returns.
 *
 * The search returns a group of mapping-contiguous page cache entries
 * with ascending indexes.  There may be holes in the indices due to
 * not-present pages.
 *
1229 1230
 * Any shadow entries of evicted pages, or swap entries from
 * shmem/tmpfs, are included in the returned array.
1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247
 *
 * find_get_entries() returns the number of pages and shadow entries
 * which were found.
 */
unsigned find_get_entries(struct address_space *mapping,
			  pgoff_t start, unsigned int nr_entries,
			  struct page **entries, pgoff_t *indices)
{
	void **slot;
	unsigned int ret = 0;
	struct radix_tree_iter iter;

	if (!nr_entries)
		return 0;

	rcu_read_lock();
	radix_tree_for_each_slot(slot, &mapping->page_tree, &iter, start) {
1248
		struct page *head, *page;
1249 1250 1251 1252 1253
repeat:
		page = radix_tree_deref_slot(slot);
		if (unlikely(!page))
			continue;
		if (radix_tree_exception(page)) {
M
Matthew Wilcox 已提交
1254 1255 1256 1257
			if (radix_tree_deref_retry(page)) {
				slot = radix_tree_iter_retry(&iter);
				continue;
			}
1258
			/*
1259 1260 1261
			 * A shadow entry of a recently evicted page, a swap
			 * entry from shmem/tmpfs or a DAX entry.  Return it
			 * without attempting to raise page count.
1262 1263 1264
			 */
			goto export;
		}
1265 1266 1267 1268 1269 1270 1271 1272

		head = compound_head(page);
		if (!page_cache_get_speculative(head))
			goto repeat;

		/* The page was split under us? */
		if (compound_head(page) != head) {
			put_page(head);
1273
			goto repeat;
1274
		}
1275 1276 1277

		/* Has the page moved? */
		if (unlikely(page != *slot)) {
1278
			put_page(head);
1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290
			goto repeat;
		}
export:
		indices[ret] = iter.index;
		entries[ret] = page;
		if (++ret == nr_entries)
			break;
	}
	rcu_read_unlock();
	return ret;
}

L
Linus Torvalds 已提交
1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309
/**
 * 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)
{
1310 1311 1312 1313 1314 1315
	struct radix_tree_iter iter;
	void **slot;
	unsigned ret = 0;

	if (unlikely(!nr_pages))
		return 0;
N
Nick Piggin 已提交
1316 1317

	rcu_read_lock();
1318
	radix_tree_for_each_slot(slot, &mapping->page_tree, &iter, start) {
1319
		struct page *head, *page;
N
Nick Piggin 已提交
1320
repeat:
1321
		page = radix_tree_deref_slot(slot);
N
Nick Piggin 已提交
1322 1323
		if (unlikely(!page))
			continue;
1324

1325
		if (radix_tree_exception(page)) {
1326
			if (radix_tree_deref_retry(page)) {
M
Matthew Wilcox 已提交
1327 1328
				slot = radix_tree_iter_retry(&iter);
				continue;
1329
			}
1330
			/*
1331 1332 1333
			 * A shadow entry of a recently evicted page,
			 * or a swap entry from shmem/tmpfs.  Skip
			 * over it.
1334
			 */
1335
			continue;
N
Nick Piggin 已提交
1336
		}
N
Nick Piggin 已提交
1337

1338 1339 1340 1341 1342 1343 1344
		head = compound_head(page);
		if (!page_cache_get_speculative(head))
			goto repeat;

		/* The page was split under us? */
		if (compound_head(page) != head) {
			put_page(head);
N
Nick Piggin 已提交
1345
			goto repeat;
1346
		}
N
Nick Piggin 已提交
1347 1348

		/* Has the page moved? */
1349
		if (unlikely(page != *slot)) {
1350
			put_page(head);
N
Nick Piggin 已提交
1351 1352
			goto repeat;
		}
L
Linus Torvalds 已提交
1353

N
Nick Piggin 已提交
1354
		pages[ret] = page;
1355 1356
		if (++ret == nr_pages)
			break;
N
Nick Piggin 已提交
1357
	}
1358

N
Nick Piggin 已提交
1359
	rcu_read_unlock();
L
Linus Torvalds 已提交
1360 1361 1362
	return ret;
}

1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377
/**
 * 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)
{
1378 1379 1380 1381 1382 1383
	struct radix_tree_iter iter;
	void **slot;
	unsigned int ret = 0;

	if (unlikely(!nr_pages))
		return 0;
N
Nick Piggin 已提交
1384 1385

	rcu_read_lock();
1386
	radix_tree_for_each_contig(slot, &mapping->page_tree, &iter, index) {
1387
		struct page *head, *page;
N
Nick Piggin 已提交
1388
repeat:
1389 1390
		page = radix_tree_deref_slot(slot);
		/* The hole, there no reason to continue */
N
Nick Piggin 已提交
1391
		if (unlikely(!page))
1392
			break;
1393

1394
		if (radix_tree_exception(page)) {
1395
			if (radix_tree_deref_retry(page)) {
M
Matthew Wilcox 已提交
1396 1397
				slot = radix_tree_iter_retry(&iter);
				continue;
1398
			}
1399
			/*
1400 1401 1402
			 * A shadow entry of a recently evicted page,
			 * or a swap entry from shmem/tmpfs.  Stop
			 * looking for contiguous pages.
1403
			 */
1404
			break;
1405
		}
1406

1407 1408 1409 1410 1411 1412 1413
		head = compound_head(page);
		if (!page_cache_get_speculative(head))
			goto repeat;

		/* The page was split under us? */
		if (compound_head(page) != head) {
			put_page(head);
N
Nick Piggin 已提交
1414
			goto repeat;
1415
		}
N
Nick Piggin 已提交
1416 1417

		/* Has the page moved? */
1418
		if (unlikely(page != *slot)) {
1419
			put_page(head);
N
Nick Piggin 已提交
1420 1421 1422
			goto repeat;
		}

N
Nick Piggin 已提交
1423 1424 1425 1426 1427
		/*
		 * must check mapping and index after taking the ref.
		 * otherwise we can get both false positives and false
		 * negatives, which is just confusing to the caller.
		 */
1428
		if (page->mapping == NULL || page_to_pgoff(page) != iter.index) {
1429
			put_page(page);
N
Nick Piggin 已提交
1430 1431 1432
			break;
		}

N
Nick Piggin 已提交
1433
		pages[ret] = page;
1434 1435
		if (++ret == nr_pages)
			break;
1436
	}
N
Nick Piggin 已提交
1437 1438
	rcu_read_unlock();
	return ret;
1439
}
1440
EXPORT_SYMBOL(find_get_pages_contig);
1441

1442 1443 1444 1445 1446 1447 1448 1449
/**
 * 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
1450
 * Like find_get_pages, except we only return pages which are tagged with
1451
 * @tag.   We update @index to index the next page for the traversal.
L
Linus Torvalds 已提交
1452 1453 1454 1455
 */
unsigned find_get_pages_tag(struct address_space *mapping, pgoff_t *index,
			int tag, unsigned int nr_pages, struct page **pages)
{
1456 1457 1458 1459 1460 1461
	struct radix_tree_iter iter;
	void **slot;
	unsigned ret = 0;

	if (unlikely(!nr_pages))
		return 0;
N
Nick Piggin 已提交
1462 1463

	rcu_read_lock();
1464 1465
	radix_tree_for_each_tagged(slot, &mapping->page_tree,
				   &iter, *index, tag) {
1466
		struct page *head, *page;
N
Nick Piggin 已提交
1467
repeat:
1468
		page = radix_tree_deref_slot(slot);
N
Nick Piggin 已提交
1469 1470
		if (unlikely(!page))
			continue;
1471

1472
		if (radix_tree_exception(page)) {
1473
			if (radix_tree_deref_retry(page)) {
M
Matthew Wilcox 已提交
1474 1475
				slot = radix_tree_iter_retry(&iter);
				continue;
1476
			}
1477
			/*
1478 1479 1480 1481 1482 1483 1484 1485 1486
			 * A shadow entry of a recently evicted page.
			 *
			 * Those entries should never be tagged, but
			 * this tree walk is lockless and the tags are
			 * looked up in bulk, one radix tree node at a
			 * time, so there is a sizable window for page
			 * reclaim to evict a page we saw tagged.
			 *
			 * Skip over it.
1487
			 */
1488
			continue;
1489
		}
N
Nick Piggin 已提交
1490

1491 1492
		head = compound_head(page);
		if (!page_cache_get_speculative(head))
N
Nick Piggin 已提交
1493 1494
			goto repeat;

1495 1496 1497 1498 1499 1500
		/* The page was split under us? */
		if (compound_head(page) != head) {
			put_page(head);
			goto repeat;
		}

N
Nick Piggin 已提交
1501
		/* Has the page moved? */
1502
		if (unlikely(page != *slot)) {
1503
			put_page(head);
N
Nick Piggin 已提交
1504 1505 1506 1507
			goto repeat;
		}

		pages[ret] = page;
1508 1509
		if (++ret == nr_pages)
			break;
N
Nick Piggin 已提交
1510
	}
1511

N
Nick Piggin 已提交
1512
	rcu_read_unlock();
L
Linus Torvalds 已提交
1513 1514 1515

	if (ret)
		*index = pages[ret - 1]->index + 1;
N
Nick Piggin 已提交
1516

L
Linus Torvalds 已提交
1517 1518
	return ret;
}
1519
EXPORT_SYMBOL(find_get_pages_tag);
L
Linus Torvalds 已提交
1520

R
Ross Zwisler 已提交
1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546
/**
 * find_get_entries_tag - find and return entries that match @tag
 * @mapping:	the address_space to search
 * @start:	the starting page cache index
 * @tag:	the tag index
 * @nr_entries:	the maximum number of entries
 * @entries:	where the resulting entries are placed
 * @indices:	the cache indices corresponding to the entries in @entries
 *
 * Like find_get_entries, except we only return entries which are tagged with
 * @tag.
 */
unsigned find_get_entries_tag(struct address_space *mapping, pgoff_t start,
			int tag, unsigned int nr_entries,
			struct page **entries, pgoff_t *indices)
{
	void **slot;
	unsigned int ret = 0;
	struct radix_tree_iter iter;

	if (!nr_entries)
		return 0;

	rcu_read_lock();
	radix_tree_for_each_tagged(slot, &mapping->page_tree,
				   &iter, start, tag) {
1547
		struct page *head, *page;
R
Ross Zwisler 已提交
1548 1549 1550 1551 1552 1553
repeat:
		page = radix_tree_deref_slot(slot);
		if (unlikely(!page))
			continue;
		if (radix_tree_exception(page)) {
			if (radix_tree_deref_retry(page)) {
M
Matthew Wilcox 已提交
1554 1555
				slot = radix_tree_iter_retry(&iter);
				continue;
R
Ross Zwisler 已提交
1556 1557 1558 1559 1560 1561 1562 1563 1564
			}

			/*
			 * A shadow entry of a recently evicted page, a swap
			 * entry from shmem/tmpfs or a DAX entry.  Return it
			 * without attempting to raise page count.
			 */
			goto export;
		}
1565 1566 1567

		head = compound_head(page);
		if (!page_cache_get_speculative(head))
R
Ross Zwisler 已提交
1568 1569
			goto repeat;

1570 1571 1572 1573 1574 1575
		/* The page was split under us? */
		if (compound_head(page) != head) {
			put_page(head);
			goto repeat;
		}

R
Ross Zwisler 已提交
1576 1577
		/* Has the page moved? */
		if (unlikely(page != *slot)) {
1578
			put_page(head);
R
Ross Zwisler 已提交
1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591
			goto repeat;
		}
export:
		indices[ret] = iter.index;
		entries[ret] = page;
		if (++ret == nr_entries)
			break;
	}
	rcu_read_unlock();
	return ret;
}
EXPORT_SYMBOL(find_get_entries_tag);

1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612
/*
 * 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)
{
	ra->ra_pages /= 4;
}

1613
/**
C
Christoph Hellwig 已提交
1614
 * do_generic_file_read - generic file read routine
1615 1616
 * @filp:	the file to read
 * @ppos:	current file position
1617 1618
 * @iter:	data destination
 * @written:	already copied
1619
 *
L
Linus Torvalds 已提交
1620
 * This is a generic file read routine, and uses the
1621
 * mapping->a_ops->readpage() function for the actual low-level stuff.
L
Linus Torvalds 已提交
1622 1623 1624 1625
 *
 * This is really ugly. But the goto's actually try to clarify some
 * of the logic when it comes to error handling etc.
 */
1626 1627
static ssize_t do_generic_file_read(struct file *filp, loff_t *ppos,
		struct iov_iter *iter, ssize_t written)
L
Linus Torvalds 已提交
1628
{
C
Christoph Hellwig 已提交
1629
	struct address_space *mapping = filp->f_mapping;
L
Linus Torvalds 已提交
1630
	struct inode *inode = mapping->host;
C
Christoph Hellwig 已提交
1631
	struct file_ra_state *ra = &filp->f_ra;
1632 1633 1634 1635
	pgoff_t index;
	pgoff_t last_index;
	pgoff_t prev_index;
	unsigned long offset;      /* offset into pagecache page */
1636
	unsigned int prev_offset;
1637
	int error = 0;
L
Linus Torvalds 已提交
1638

1639 1640 1641 1642
	if (unlikely(*ppos >= inode->i_sb->s_maxbytes))
		return -EINVAL;
	iov_iter_truncate(iter, inode->i_sb->s_maxbytes);

1643 1644 1645 1646 1647
	index = *ppos >> PAGE_SHIFT;
	prev_index = ra->prev_pos >> PAGE_SHIFT;
	prev_offset = ra->prev_pos & (PAGE_SIZE-1);
	last_index = (*ppos + iter->count + PAGE_SIZE-1) >> PAGE_SHIFT;
	offset = *ppos & ~PAGE_MASK;
L
Linus Torvalds 已提交
1648 1649 1650

	for (;;) {
		struct page *page;
1651
		pgoff_t end_index;
N
NeilBrown 已提交
1652
		loff_t isize;
L
Linus Torvalds 已提交
1653 1654 1655 1656 1657
		unsigned long nr, ret;

		cond_resched();
find_page:
		page = find_get_page(mapping, index);
1658
		if (!page) {
1659
			page_cache_sync_readahead(mapping,
1660
					ra, filp,
1661 1662 1663 1664 1665 1666
					index, last_index - index);
			page = find_get_page(mapping, index);
			if (unlikely(page == NULL))
				goto no_cached_page;
		}
		if (PageReadahead(page)) {
1667
			page_cache_async_readahead(mapping,
1668
					ra, filp, page,
1669
					index, last_index - index);
L
Linus Torvalds 已提交
1670
		}
1671
		if (!PageUptodate(page)) {
1672 1673 1674 1675 1676
			/*
			 * See comment in do_read_cache_page on why
			 * wait_on_page_locked is used to avoid unnecessarily
			 * serialisations and why it's safe.
			 */
1677 1678 1679
			error = wait_on_page_locked_killable(page);
			if (unlikely(error))
				goto readpage_error;
1680 1681 1682
			if (PageUptodate(page))
				goto page_ok;

1683
			if (inode->i_blkbits == PAGE_SHIFT ||
1684 1685
					!mapping->a_ops->is_partially_uptodate)
				goto page_not_up_to_date;
1686 1687 1688
			/* pipes can't handle partially uptodate pages */
			if (unlikely(iter->type & ITER_PIPE))
				goto page_not_up_to_date;
N
Nick Piggin 已提交
1689
			if (!trylock_page(page))
1690
				goto page_not_up_to_date;
1691 1692 1693
			/* Did it get truncated before we got the lock? */
			if (!page->mapping)
				goto page_not_up_to_date_locked;
1694
			if (!mapping->a_ops->is_partially_uptodate(page,
1695
							offset, iter->count))
1696 1697 1698
				goto page_not_up_to_date_locked;
			unlock_page(page);
		}
L
Linus Torvalds 已提交
1699
page_ok:
N
NeilBrown 已提交
1700 1701 1702 1703 1704 1705 1706 1707 1708 1709
		/*
		 * 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);
1710
		end_index = (isize - 1) >> PAGE_SHIFT;
N
NeilBrown 已提交
1711
		if (unlikely(!isize || index > end_index)) {
1712
			put_page(page);
N
NeilBrown 已提交
1713 1714 1715 1716
			goto out;
		}

		/* nr is the maximum number of bytes to copy from this page */
1717
		nr = PAGE_SIZE;
N
NeilBrown 已提交
1718
		if (index == end_index) {
1719
			nr = ((isize - 1) & ~PAGE_MASK) + 1;
N
NeilBrown 已提交
1720
			if (nr <= offset) {
1721
				put_page(page);
N
NeilBrown 已提交
1722 1723 1724 1725
				goto out;
			}
		}
		nr = nr - offset;
L
Linus Torvalds 已提交
1726 1727 1728 1729 1730 1731 1732 1733 1734

		/* 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);

		/*
1735 1736
		 * When a sequential read accesses a page several times,
		 * only mark it as accessed the first time.
L
Linus Torvalds 已提交
1737
		 */
1738
		if (prev_index != index || offset != prev_offset)
L
Linus Torvalds 已提交
1739 1740 1741 1742 1743 1744 1745
			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...
		 */
1746 1747

		ret = copy_page_to_iter(page, offset, nr, iter);
L
Linus Torvalds 已提交
1748
		offset += ret;
1749 1750
		index += offset >> PAGE_SHIFT;
		offset &= ~PAGE_MASK;
J
Jan Kara 已提交
1751
		prev_offset = offset;
L
Linus Torvalds 已提交
1752

1753
		put_page(page);
1754 1755 1756 1757 1758 1759 1760 1761
		written += ret;
		if (!iov_iter_count(iter))
			goto out;
		if (ret < nr) {
			error = -EFAULT;
			goto out;
		}
		continue;
L
Linus Torvalds 已提交
1762 1763 1764

page_not_up_to_date:
		/* Get exclusive access to the page ... */
1765 1766 1767
		error = lock_page_killable(page);
		if (unlikely(error))
			goto readpage_error;
L
Linus Torvalds 已提交
1768

1769
page_not_up_to_date_locked:
N
Nick Piggin 已提交
1770
		/* Did it get truncated before we got the lock? */
L
Linus Torvalds 已提交
1771 1772
		if (!page->mapping) {
			unlock_page(page);
1773
			put_page(page);
L
Linus Torvalds 已提交
1774 1775 1776 1777 1778 1779 1780 1781 1782 1783
			continue;
		}

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

readpage:
1784 1785 1786 1787 1788 1789
		/*
		 * A previous I/O error may have been due to temporary
		 * failures, eg. multipath errors.
		 * PG_error will be set again if readpage fails.
		 */
		ClearPageError(page);
L
Linus Torvalds 已提交
1790 1791 1792
		/* Start the actual read. The read will unlock the page. */
		error = mapping->a_ops->readpage(filp, page);

1793 1794
		if (unlikely(error)) {
			if (error == AOP_TRUNCATED_PAGE) {
1795
				put_page(page);
1796
				error = 0;
1797 1798
				goto find_page;
			}
L
Linus Torvalds 已提交
1799
			goto readpage_error;
1800
		}
L
Linus Torvalds 已提交
1801 1802

		if (!PageUptodate(page)) {
1803 1804 1805
			error = lock_page_killable(page);
			if (unlikely(error))
				goto readpage_error;
L
Linus Torvalds 已提交
1806 1807 1808
			if (!PageUptodate(page)) {
				if (page->mapping == NULL) {
					/*
1809
					 * invalidate_mapping_pages got it
L
Linus Torvalds 已提交
1810 1811
					 */
					unlock_page(page);
1812
					put_page(page);
L
Linus Torvalds 已提交
1813 1814 1815
					goto find_page;
				}
				unlock_page(page);
1816
				shrink_readahead_size_eio(filp, ra);
1817 1818
				error = -EIO;
				goto readpage_error;
L
Linus Torvalds 已提交
1819 1820 1821 1822 1823 1824 1825 1826
			}
			unlock_page(page);
		}

		goto page_ok;

readpage_error:
		/* UHHUH! A synchronous read error occurred. Report it */
1827
		put_page(page);
L
Linus Torvalds 已提交
1828 1829 1830 1831 1832 1833 1834
		goto out;

no_cached_page:
		/*
		 * Ok, it wasn't cached, so we need to create a new
		 * page..
		 */
N
Nick Piggin 已提交
1835 1836
		page = page_cache_alloc_cold(mapping);
		if (!page) {
1837
			error = -ENOMEM;
N
Nick Piggin 已提交
1838
			goto out;
L
Linus Torvalds 已提交
1839
		}
1840
		error = add_to_page_cache_lru(page, mapping, index,
1841
				mapping_gfp_constraint(mapping, GFP_KERNEL));
L
Linus Torvalds 已提交
1842
		if (error) {
1843
			put_page(page);
1844 1845
			if (error == -EEXIST) {
				error = 0;
L
Linus Torvalds 已提交
1846
				goto find_page;
1847
			}
L
Linus Torvalds 已提交
1848 1849 1850 1851 1852 1853
			goto out;
		}
		goto readpage;
	}

out:
1854
	ra->prev_pos = prev_index;
1855
	ra->prev_pos <<= PAGE_SHIFT;
1856
	ra->prev_pos |= prev_offset;
L
Linus Torvalds 已提交
1857

1858
	*ppos = ((loff_t)index << PAGE_SHIFT) + offset;
1859
	file_accessed(filp);
1860
	return written ? written : error;
L
Linus Torvalds 已提交
1861 1862
}

1863
/**
A
Al Viro 已提交
1864
 * generic_file_read_iter - generic filesystem read routine
1865
 * @iocb:	kernel I/O control block
A
Al Viro 已提交
1866
 * @iter:	destination for the data read
1867
 *
A
Al Viro 已提交
1868
 * This is the "read_iter()" routine for all filesystems
L
Linus Torvalds 已提交
1869 1870 1871
 * that can use the page cache directly.
 */
ssize_t
A
Al Viro 已提交
1872
generic_file_read_iter(struct kiocb *iocb, struct iov_iter *iter)
L
Linus Torvalds 已提交
1873
{
A
Al Viro 已提交
1874
	struct file *file = iocb->ki_filp;
A
Al Viro 已提交
1875
	ssize_t retval = 0;
1876 1877 1878 1879
	size_t count = iov_iter_count(iter);

	if (!count)
		goto out; /* skip atime */
L
Linus Torvalds 已提交
1880

1881
	if (iocb->ki_flags & IOCB_DIRECT) {
A
Al Viro 已提交
1882 1883
		struct address_space *mapping = file->f_mapping;
		struct inode *inode = mapping->host;
1884
		struct iov_iter data = *iter;
1885
		loff_t size;
L
Linus Torvalds 已提交
1886 1887

		size = i_size_read(inode);
1888 1889
		retval = filemap_write_and_wait_range(mapping, iocb->ki_pos,
					iocb->ki_pos + count - 1);
1890 1891
		if (retval < 0)
			goto out;
A
Al Viro 已提交
1892

1893 1894 1895
		file_accessed(file);

		retval = mapping->a_ops->direct_IO(iocb, &data);
A
Al Viro 已提交
1896
		if (retval >= 0) {
1897
			iocb->ki_pos += retval;
A
Al Viro 已提交
1898
			iov_iter_advance(iter, retval);
1899
		}
1900

1901 1902 1903 1904 1905 1906
		/*
		 * Btrfs can have a short DIO read if we encounter
		 * compressed extents, so if there was an error, or if
		 * we've already read everything we wanted to, or if
		 * there was a short read because we hit EOF, go ahead
		 * and return.  Otherwise fallthrough to buffered io for
1907 1908
		 * the rest of the read.  Buffered reads will not work for
		 * DAX files, so don't bother trying.
1909
		 */
1910
		if (retval < 0 || !iov_iter_count(iter) || iocb->ki_pos >= size ||
1911
		    IS_DAX(inode))
1912
			goto out;
L
Linus Torvalds 已提交
1913 1914
	}

1915
	retval = do_generic_file_read(file, &iocb->ki_pos, iter, retval);
L
Linus Torvalds 已提交
1916 1917 1918
out:
	return retval;
}
A
Al Viro 已提交
1919
EXPORT_SYMBOL(generic_file_read_iter);
L
Linus Torvalds 已提交
1920 1921

#ifdef CONFIG_MMU
1922 1923 1924 1925
/**
 * page_cache_read - adds requested page to the page cache if not already there
 * @file:	file to read
 * @offset:	page index
1926
 * @gfp_mask:	memory allocation flags
1927
 *
L
Linus Torvalds 已提交
1928 1929 1930
 * 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.
 */
1931
static int page_cache_read(struct file *file, pgoff_t offset, gfp_t gfp_mask)
L
Linus Torvalds 已提交
1932 1933
{
	struct address_space *mapping = file->f_mapping;
1934
	struct page *page;
1935
	int ret;
L
Linus Torvalds 已提交
1936

1937
	do {
1938
		page = __page_cache_alloc(gfp_mask|__GFP_COLD);
1939 1940 1941
		if (!page)
			return -ENOMEM;

1942
		ret = add_to_page_cache_lru(page, mapping, offset, gfp_mask & GFP_KERNEL);
1943 1944 1945 1946
		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
1947

1948
		put_page(page);
L
Linus Torvalds 已提交
1949

1950
	} while (ret == AOP_TRUNCATED_PAGE);
1951

1952
	return ret;
L
Linus Torvalds 已提交
1953 1954 1955 1956
}

#define MMAP_LOTSAMISS  (100)

1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968
/*
 * Synchronous readahead happens when we don't even find
 * a page in the page cache at all.
 */
static void do_sync_mmap_readahead(struct vm_area_struct *vma,
				   struct file_ra_state *ra,
				   struct file *file,
				   pgoff_t offset)
{
	struct address_space *mapping = file->f_mapping;

	/* If we don't want any read-ahead, don't bother */
1969
	if (vma->vm_flags & VM_RAND_READ)
1970
		return;
1971 1972
	if (!ra->ra_pages)
		return;
1973

1974
	if (vma->vm_flags & VM_SEQ_READ) {
1975 1976
		page_cache_sync_readahead(mapping, ra, file, offset,
					  ra->ra_pages);
1977 1978 1979
		return;
	}

1980 1981
	/* Avoid banging the cache line if not needed */
	if (ra->mmap_miss < MMAP_LOTSAMISS * 10)
1982 1983 1984 1985 1986 1987 1988 1989 1990
		ra->mmap_miss++;

	/*
	 * Do we miss much more than hit in this file? If so,
	 * stop bothering with read-ahead. It will only hurt.
	 */
	if (ra->mmap_miss > MMAP_LOTSAMISS)
		return;

1991 1992 1993
	/*
	 * mmap read-around
	 */
1994 1995 1996
	ra->start = max_t(long, 0, offset - ra->ra_pages / 2);
	ra->size = ra->ra_pages;
	ra->async_size = ra->ra_pages / 4;
1997
	ra_submit(ra, mapping, file);
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
}

/*
 * Asynchronous readahead happens when we find the page and PG_readahead,
 * so we want to possibly extend the readahead further..
 */
static void do_async_mmap_readahead(struct vm_area_struct *vma,
				    struct file_ra_state *ra,
				    struct file *file,
				    struct page *page,
				    pgoff_t offset)
{
	struct address_space *mapping = file->f_mapping;

	/* If we don't want any read-ahead, don't bother */
2013
	if (vma->vm_flags & VM_RAND_READ)
2014 2015 2016 2017
		return;
	if (ra->mmap_miss > 0)
		ra->mmap_miss--;
	if (PageReadahead(page))
2018 2019
		page_cache_async_readahead(mapping, ra, file,
					   page, offset, ra->ra_pages);
2020 2021
}

2022
/**
2023
 * filemap_fault - read in file data for page fault handling
N
Nick Piggin 已提交
2024 2025
 * @vma:	vma in which the fault was taken
 * @vmf:	struct vm_fault containing details of the fault
2026
 *
2027
 * filemap_fault() is invoked via the vma operations vector for a
L
Linus Torvalds 已提交
2028 2029 2030 2031 2032
 * 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.
2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044
 *
 * vma->vm_mm->mmap_sem must be held on entry.
 *
 * If our return value has VM_FAULT_RETRY set, it's because
 * lock_page_or_retry() returned 0.
 * The mmap_sem has usually been released in this case.
 * See __lock_page_or_retry() for the exception.
 *
 * If our return value does not have VM_FAULT_RETRY set, the mmap_sem
 * has not been released.
 *
 * We never return with VM_FAULT_RETRY and a bit from VM_FAULT_ERROR set.
L
Linus Torvalds 已提交
2045
 */
N
Nick Piggin 已提交
2046
int filemap_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
L
Linus Torvalds 已提交
2047 2048
{
	int error;
2049
	struct file *file = vma->vm_file;
L
Linus Torvalds 已提交
2050 2051 2052
	struct address_space *mapping = file->f_mapping;
	struct file_ra_state *ra = &file->f_ra;
	struct inode *inode = mapping->host;
2053
	pgoff_t offset = vmf->pgoff;
L
Linus Torvalds 已提交
2054
	struct page *page;
2055
	loff_t size;
N
Nick Piggin 已提交
2056
	int ret = 0;
L
Linus Torvalds 已提交
2057

2058 2059
	size = round_up(i_size_read(inode), PAGE_SIZE);
	if (offset >= size >> PAGE_SHIFT)
2060
		return VM_FAULT_SIGBUS;
L
Linus Torvalds 已提交
2061 2062

	/*
2063
	 * Do we have something in the page cache already?
L
Linus Torvalds 已提交
2064
	 */
2065
	page = find_get_page(mapping, offset);
2066
	if (likely(page) && !(vmf->flags & FAULT_FLAG_TRIED)) {
L
Linus Torvalds 已提交
2067
		/*
2068 2069
		 * We found the page, so try async readahead before
		 * waiting for the lock.
L
Linus Torvalds 已提交
2070
		 */
2071
		do_async_mmap_readahead(vma, ra, file, page, offset);
2072
	} else if (!page) {
2073 2074 2075
		/* No page in the page cache at all */
		do_sync_mmap_readahead(vma, ra, file, offset);
		count_vm_event(PGMAJFAULT);
2076
		mem_cgroup_count_vm_event(vma->vm_mm, PGMAJFAULT);
2077 2078
		ret = VM_FAULT_MAJOR;
retry_find:
2079
		page = find_get_page(mapping, offset);
L
Linus Torvalds 已提交
2080 2081 2082 2083
		if (!page)
			goto no_cached_page;
	}

2084
	if (!lock_page_or_retry(page, vma->vm_mm, vmf->flags)) {
2085
		put_page(page);
2086
		return ret | VM_FAULT_RETRY;
2087
	}
2088 2089 2090 2091 2092 2093 2094

	/* Did it get truncated? */
	if (unlikely(page->mapping != mapping)) {
		unlock_page(page);
		put_page(page);
		goto retry_find;
	}
2095
	VM_BUG_ON_PAGE(page->index != offset, page);
2096

L
Linus Torvalds 已提交
2097
	/*
2098 2099
	 * 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
2100
	 */
2101
	if (unlikely(!PageUptodate(page)))
L
Linus Torvalds 已提交
2102 2103
		goto page_not_uptodate;

2104 2105 2106 2107
	/*
	 * Found the page and have a reference on it.
	 * We must recheck i_size under page lock.
	 */
2108 2109
	size = round_up(i_size_read(inode), PAGE_SIZE);
	if (unlikely(offset >= size >> PAGE_SHIFT)) {
2110
		unlock_page(page);
2111
		put_page(page);
2112
		return VM_FAULT_SIGBUS;
2113 2114
	}

N
Nick Piggin 已提交
2115
	vmf->page = page;
N
Nick Piggin 已提交
2116
	return ret | VM_FAULT_LOCKED;
L
Linus Torvalds 已提交
2117 2118 2119 2120 2121 2122

no_cached_page:
	/*
	 * We're only likely to ever get here if MADV_RANDOM is in
	 * effect.
	 */
2123
	error = page_cache_read(file, offset, vmf->gfp_mask);
L
Linus Torvalds 已提交
2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138

	/*
	 * 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
2139 2140
		return VM_FAULT_OOM;
	return VM_FAULT_SIGBUS;
L
Linus Torvalds 已提交
2141 2142 2143 2144 2145 2146 2147 2148 2149

page_not_uptodate:
	/*
	 * 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);
2150
	error = mapping->a_ops->readpage(file, page);
2151 2152 2153 2154 2155
	if (!error) {
		wait_on_page_locked(page);
		if (!PageUptodate(page))
			error = -EIO;
	}
2156
	put_page(page);
2157 2158

	if (!error || error == AOP_TRUNCATED_PAGE)
2159
		goto retry_find;
L
Linus Torvalds 已提交
2160

2161
	/* Things didn't work out. Return zero to tell the mm layer so. */
2162
	shrink_readahead_size_eio(file, ra);
N
Nick Piggin 已提交
2163
	return VM_FAULT_SIGBUS;
2164 2165 2166
}
EXPORT_SYMBOL(filemap_fault);

K
Kirill A. Shutemov 已提交
2167 2168
void filemap_map_pages(struct fault_env *fe,
		pgoff_t start_pgoff, pgoff_t end_pgoff)
2169 2170 2171
{
	struct radix_tree_iter iter;
	void **slot;
K
Kirill A. Shutemov 已提交
2172
	struct file *file = fe->vma->vm_file;
2173
	struct address_space *mapping = file->f_mapping;
K
Kirill A. Shutemov 已提交
2174
	pgoff_t last_pgoff = start_pgoff;
2175
	loff_t size;
2176
	struct page *head, *page;
2177 2178

	rcu_read_lock();
K
Kirill A. Shutemov 已提交
2179 2180 2181
	radix_tree_for_each_slot(slot, &mapping->page_tree, &iter,
			start_pgoff) {
		if (iter.index > end_pgoff)
2182 2183 2184 2185 2186 2187
			break;
repeat:
		page = radix_tree_deref_slot(slot);
		if (unlikely(!page))
			goto next;
		if (radix_tree_exception(page)) {
M
Matthew Wilcox 已提交
2188 2189 2190 2191 2192
			if (radix_tree_deref_retry(page)) {
				slot = radix_tree_iter_retry(&iter);
				continue;
			}
			goto next;
2193 2194
		}

2195 2196
		head = compound_head(page);
		if (!page_cache_get_speculative(head))
2197 2198
			goto repeat;

2199 2200 2201 2202 2203 2204
		/* The page was split under us? */
		if (compound_head(page) != head) {
			put_page(head);
			goto repeat;
		}

2205 2206
		/* Has the page moved? */
		if (unlikely(page != *slot)) {
2207
			put_page(head);
2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220
			goto repeat;
		}

		if (!PageUptodate(page) ||
				PageReadahead(page) ||
				PageHWPoison(page))
			goto skip;
		if (!trylock_page(page))
			goto skip;

		if (page->mapping != mapping || !PageUptodate(page))
			goto unlock;

2221 2222
		size = round_up(i_size_read(mapping->host), PAGE_SIZE);
		if (page->index >= size >> PAGE_SHIFT)
2223 2224 2225 2226
			goto unlock;

		if (file->f_ra.mmap_miss > 0)
			file->f_ra.mmap_miss--;
2227 2228 2229 2230 2231 2232 2233

		fe->address += (iter.index - last_pgoff) << PAGE_SHIFT;
		if (fe->pte)
			fe->pte += iter.index - last_pgoff;
		last_pgoff = iter.index;
		if (alloc_set_pte(fe, NULL, page))
			goto unlock;
2234 2235 2236 2237 2238
		unlock_page(page);
		goto next;
unlock:
		unlock_page(page);
skip:
2239
		put_page(page);
2240
next:
2241 2242 2243
		/* Huge page is mapped? No need to proceed. */
		if (pmd_trans_huge(*fe->pmd))
			break;
K
Kirill A. Shutemov 已提交
2244
		if (iter.index == end_pgoff)
2245 2246 2247 2248 2249 2250
			break;
	}
	rcu_read_unlock();
}
EXPORT_SYMBOL(filemap_map_pages);

2251 2252 2253
int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
{
	struct page *page = vmf->page;
A
Al Viro 已提交
2254
	struct inode *inode = file_inode(vma->vm_file);
2255 2256
	int ret = VM_FAULT_LOCKED;

2257
	sb_start_pagefault(inode->i_sb);
2258 2259 2260 2261 2262 2263 2264
	file_update_time(vma->vm_file);
	lock_page(page);
	if (page->mapping != inode->i_mapping) {
		unlock_page(page);
		ret = VM_FAULT_NOPAGE;
		goto out;
	}
2265 2266 2267 2268 2269 2270
	/*
	 * We mark the page dirty already here so that when freeze is in
	 * progress, we are guaranteed that writeback during freezing will
	 * see the dirty page and writeprotect it again.
	 */
	set_page_dirty(page);
2271
	wait_for_stable_page(page);
2272
out:
2273
	sb_end_pagefault(inode->i_sb);
2274 2275 2276 2277
	return ret;
}
EXPORT_SYMBOL(filemap_page_mkwrite);

2278
const struct vm_operations_struct generic_file_vm_ops = {
2279
	.fault		= filemap_fault,
2280
	.map_pages	= filemap_map_pages,
2281
	.page_mkwrite	= filemap_page_mkwrite,
L
Linus Torvalds 已提交
2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319
};

/* 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;
	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);

S
Sasha Levin 已提交
2320 2321 2322 2323 2324
static struct page *wait_on_page_read(struct page *page)
{
	if (!IS_ERR(page)) {
		wait_on_page_locked(page);
		if (!PageUptodate(page)) {
2325
			put_page(page);
S
Sasha Levin 已提交
2326 2327 2328 2329 2330 2331
			page = ERR_PTR(-EIO);
		}
	}
	return page;
}

2332
static struct page *do_read_cache_page(struct address_space *mapping,
2333
				pgoff_t index,
2334
				int (*filler)(void *, struct page *),
2335 2336
				void *data,
				gfp_t gfp)
L
Linus Torvalds 已提交
2337
{
N
Nick Piggin 已提交
2338
	struct page *page;
L
Linus Torvalds 已提交
2339 2340 2341 2342
	int err;
repeat:
	page = find_get_page(mapping, index);
	if (!page) {
2343
		page = __page_cache_alloc(gfp | __GFP_COLD);
N
Nick Piggin 已提交
2344 2345
		if (!page)
			return ERR_PTR(-ENOMEM);
2346
		err = add_to_page_cache_lru(page, mapping, index, gfp);
N
Nick Piggin 已提交
2347
		if (unlikely(err)) {
2348
			put_page(page);
N
Nick Piggin 已提交
2349 2350
			if (err == -EEXIST)
				goto repeat;
L
Linus Torvalds 已提交
2351 2352 2353
			/* Presumably ENOMEM for radix tree node */
			return ERR_PTR(err);
		}
2354 2355

filler:
L
Linus Torvalds 已提交
2356 2357
		err = filler(data, page);
		if (err < 0) {
2358
			put_page(page);
2359
			return ERR_PTR(err);
L
Linus Torvalds 已提交
2360 2361
		}

2362 2363 2364 2365 2366
		page = wait_on_page_read(page);
		if (IS_ERR(page))
			return page;
		goto out;
	}
L
Linus Torvalds 已提交
2367 2368 2369
	if (PageUptodate(page))
		goto out;

2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405
	/*
	 * Page is not up to date and may be locked due one of the following
	 * case a: Page is being filled and the page lock is held
	 * case b: Read/write error clearing the page uptodate status
	 * case c: Truncation in progress (page locked)
	 * case d: Reclaim in progress
	 *
	 * Case a, the page will be up to date when the page is unlocked.
	 *    There is no need to serialise on the page lock here as the page
	 *    is pinned so the lock gives no additional protection. Even if the
	 *    the page is truncated, the data is still valid if PageUptodate as
	 *    it's a race vs truncate race.
	 * Case b, the page will not be up to date
	 * Case c, the page may be truncated but in itself, the data may still
	 *    be valid after IO completes as it's a read vs truncate race. The
	 *    operation must restart if the page is not uptodate on unlock but
	 *    otherwise serialising on page lock to stabilise the mapping gives
	 *    no additional guarantees to the caller as the page lock is
	 *    released before return.
	 * Case d, similar to truncation. If reclaim holds the page lock, it
	 *    will be a race with remove_mapping that determines if the mapping
	 *    is valid on unlock but otherwise the data is valid and there is
	 *    no need to serialise with page lock.
	 *
	 * As the page lock gives no additional guarantee, we optimistically
	 * wait on the page to be unlocked and check if it's up to date and
	 * use the page if it is. Otherwise, the page lock is required to
	 * distinguish between the different cases. The motivation is that we
	 * avoid spurious serialisations and wakeups when multiple processes
	 * wait on the same page for IO to complete.
	 */
	wait_on_page_locked(page);
	if (PageUptodate(page))
		goto out;

	/* Distinguish between all the cases under the safety of the lock */
L
Linus Torvalds 已提交
2406
	lock_page(page);
2407 2408

	/* Case c or d, restart the operation */
L
Linus Torvalds 已提交
2409 2410
	if (!page->mapping) {
		unlock_page(page);
2411
		put_page(page);
2412
		goto repeat;
L
Linus Torvalds 已提交
2413
	}
2414 2415

	/* Someone else locked and filled the page in a very small window */
L
Linus Torvalds 已提交
2416 2417 2418 2419
	if (PageUptodate(page)) {
		unlock_page(page);
		goto out;
	}
2420 2421
	goto filler;

2422
out:
2423 2424 2425
	mark_page_accessed(page);
	return page;
}
2426 2427

/**
S
Sasha Levin 已提交
2428
 * read_cache_page - read into page cache, fill it if needed
2429 2430 2431
 * @mapping:	the page's address_space
 * @index:	the page index
 * @filler:	function to perform the read
2432
 * @data:	first arg to filler(data, page) function, often left as NULL
2433 2434
 *
 * Read into the page cache. If a page already exists, and PageUptodate() is
S
Sasha Levin 已提交
2435
 * not set, try to fill the page and wait for it to become unlocked.
2436 2437 2438
 *
 * If the page does not get brought uptodate, return -EIO.
 */
S
Sasha Levin 已提交
2439
struct page *read_cache_page(struct address_space *mapping,
2440
				pgoff_t index,
2441
				int (*filler)(void *, struct page *),
2442 2443 2444 2445
				void *data)
{
	return do_read_cache_page(mapping, index, filler, data, mapping_gfp_mask(mapping));
}
S
Sasha Levin 已提交
2446
EXPORT_SYMBOL(read_cache_page);
2447 2448 2449 2450 2451 2452 2453 2454

/**
 * read_cache_page_gfp - read into page cache, using specified page allocation flags.
 * @mapping:	the page's address_space
 * @index:	the page index
 * @gfp:	the page allocator flags to use if allocating
 *
 * This is the same as "read_mapping_page(mapping, index, NULL)", but with
2455
 * any new page allocations done using the specified allocation flags.
2456 2457 2458 2459 2460 2461 2462 2463 2464
 *
 * If the page does not get brought uptodate, return -EIO.
 */
struct page *read_cache_page_gfp(struct address_space *mapping,
				pgoff_t index,
				gfp_t gfp)
{
	filler_t *filler = (filler_t *)mapping->a_ops->readpage;

S
Sasha Levin 已提交
2465
	return do_read_cache_page(mapping, index, filler, NULL, gfp);
2466 2467 2468
}
EXPORT_SYMBOL(read_cache_page_gfp);

L
Linus Torvalds 已提交
2469 2470 2471
/*
 * Performs necessary checks before doing a write
 *
2472
 * Can adjust writing position or amount of bytes to write.
L
Linus Torvalds 已提交
2473 2474 2475
 * Returns appropriate error code that caller should return or
 * zero in case that write should be allowed.
 */
2476
inline ssize_t generic_write_checks(struct kiocb *iocb, struct iov_iter *from)
L
Linus Torvalds 已提交
2477
{
2478
	struct file *file = iocb->ki_filp;
L
Linus Torvalds 已提交
2479
	struct inode *inode = file->f_mapping->host;
J
Jiri Slaby 已提交
2480
	unsigned long limit = rlimit(RLIMIT_FSIZE);
2481
	loff_t pos;
L
Linus Torvalds 已提交
2482

2483 2484
	if (!iov_iter_count(from))
		return 0;
L
Linus Torvalds 已提交
2485

2486
	/* FIXME: this is for backwards compatibility with 2.4 */
2487
	if (iocb->ki_flags & IOCB_APPEND)
2488
		iocb->ki_pos = i_size_read(inode);
L
Linus Torvalds 已提交
2489

2490
	pos = iocb->ki_pos;
L
Linus Torvalds 已提交
2491

2492
	if (limit != RLIM_INFINITY) {
2493
		if (iocb->ki_pos >= limit) {
2494 2495
			send_sig(SIGXFSZ, current, 0);
			return -EFBIG;
L
Linus Torvalds 已提交
2496
		}
2497
		iov_iter_truncate(from, limit - (unsigned long)pos);
L
Linus Torvalds 已提交
2498 2499 2500 2501 2502
	}

	/*
	 * LFS rule
	 */
2503
	if (unlikely(pos + iov_iter_count(from) > MAX_NON_LFS &&
L
Linus Torvalds 已提交
2504
				!(file->f_flags & O_LARGEFILE))) {
2505
		if (pos >= MAX_NON_LFS)
L
Linus Torvalds 已提交
2506
			return -EFBIG;
2507
		iov_iter_truncate(from, MAX_NON_LFS - (unsigned long)pos);
L
Linus Torvalds 已提交
2508 2509 2510 2511 2512 2513 2514 2515 2516
	}

	/*
	 * 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..
	 */
2517 2518
	if (unlikely(pos >= inode->i_sb->s_maxbytes))
		return -EFBIG;
L
Linus Torvalds 已提交
2519

2520 2521
	iov_iter_truncate(from, inode->i_sb->s_maxbytes - pos);
	return iov_iter_count(from);
L
Linus Torvalds 已提交
2522 2523 2524
}
EXPORT_SYMBOL(generic_write_checks);

2525 2526 2527 2528 2529 2530
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;

2531
	return aops->write_begin(file, mapping, pos, len, flags,
2532 2533 2534 2535 2536 2537 2538 2539 2540 2541
							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;

2542
	return aops->write_end(file, mapping, pos, len, copied, page, fsdata);
2543 2544 2545
}
EXPORT_SYMBOL(pagecache_write_end);

L
Linus Torvalds 已提交
2546
ssize_t
2547
generic_file_direct_write(struct kiocb *iocb, struct iov_iter *from)
L
Linus Torvalds 已提交
2548 2549 2550 2551
{
	struct file	*file = iocb->ki_filp;
	struct address_space *mapping = file->f_mapping;
	struct inode	*inode = mapping->host;
2552
	loff_t		pos = iocb->ki_pos;
L
Linus Torvalds 已提交
2553
	ssize_t		written;
2554 2555
	size_t		write_len;
	pgoff_t		end;
A
Al Viro 已提交
2556
	struct iov_iter data;
L
Linus Torvalds 已提交
2557

A
Al Viro 已提交
2558
	write_len = iov_iter_count(from);
2559
	end = (pos + write_len - 1) >> PAGE_SHIFT;
2560

2561
	written = filemap_write_and_wait_range(mapping, pos, pos + write_len - 1);
2562 2563 2564 2565 2566 2567 2568
	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
2569
	 * without clobbering -EIOCBQUEUED from ->direct_IO().
2570 2571 2572
	 */
	if (mapping->nrpages) {
		written = invalidate_inode_pages2_range(mapping,
2573
					pos >> PAGE_SHIFT, end);
2574 2575 2576 2577 2578 2579 2580
		/*
		 * If a page can not be invalidated, return 0 to fall back
		 * to buffered write.
		 */
		if (written) {
			if (written == -EBUSY)
				return 0;
2581
			goto out;
2582
		}
2583 2584
	}

A
Al Viro 已提交
2585
	data = *from;
2586
	written = mapping->a_ops->direct_IO(iocb, &data);
2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597

	/*
	 * 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,
2598
					      pos >> PAGE_SHIFT, end);
2599 2600
	}

L
Linus Torvalds 已提交
2601
	if (written > 0) {
2602
		pos += written;
2603
		iov_iter_advance(from, written);
2604 2605
		if (pos > i_size_read(inode) && !S_ISBLK(inode->i_mode)) {
			i_size_write(inode, pos);
L
Linus Torvalds 已提交
2606 2607
			mark_inode_dirty(inode);
		}
2608
		iocb->ki_pos = pos;
L
Linus Torvalds 已提交
2609
	}
2610
out:
L
Linus Torvalds 已提交
2611 2612 2613 2614
	return written;
}
EXPORT_SYMBOL(generic_file_direct_write);

N
Nick Piggin 已提交
2615 2616 2617 2618
/*
 * Find or create a page at the given pagecache position. Return the locked
 * page. This function is specifically for buffered writes.
 */
2619 2620
struct page *grab_cache_page_write_begin(struct address_space *mapping,
					pgoff_t index, unsigned flags)
N
Nick Piggin 已提交
2621 2622
{
	struct page *page;
2623
	int fgp_flags = FGP_LOCK|FGP_WRITE|FGP_CREAT;
2624

2625
	if (flags & AOP_FLAG_NOFS)
2626 2627 2628
		fgp_flags |= FGP_NOFS;

	page = pagecache_get_page(mapping, index, fgp_flags,
2629
			mapping_gfp_mask(mapping));
2630
	if (page)
2631
		wait_for_stable_page(page);
N
Nick Piggin 已提交
2632 2633 2634

	return page;
}
2635
EXPORT_SYMBOL(grab_cache_page_write_begin);
N
Nick Piggin 已提交
2636

2637
ssize_t generic_perform_write(struct file *file,
2638 2639 2640 2641 2642 2643
				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
2644 2645 2646 2647 2648
	unsigned int flags = 0;

	/*
	 * Copies from kernel address space cannot fail (NFSD is a big user).
	 */
A
Al Viro 已提交
2649
	if (!iter_is_iovec(i))
N
Nick Piggin 已提交
2650
		flags |= AOP_FLAG_UNINTERRUPTIBLE;
2651 2652 2653 2654 2655 2656 2657 2658

	do {
		struct page *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;

2659 2660
		offset = (pos & (PAGE_SIZE - 1));
		bytes = min_t(unsigned long, PAGE_SIZE - offset,
2661 2662 2663
						iov_iter_count(i));

again:
2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678
		/*
		 * 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;
		}

J
Jan Kara 已提交
2679 2680 2681 2682 2683
		if (fatal_signal_pending(current)) {
			status = -EINTR;
			break;
		}

N
Nick Piggin 已提交
2684
		status = a_ops->write_begin(file, mapping, pos, bytes, flags,
2685
						&page, &fsdata);
2686
		if (unlikely(status < 0))
2687 2688
			break;

2689 2690
		if (mapping_writably_mapped(mapping))
			flush_dcache_page(page);
2691

2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702
		copied = iov_iter_copy_from_user_atomic(page, i, offset, bytes);
		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();

2703
		iov_iter_advance(i, copied);
2704 2705 2706 2707 2708 2709 2710 2711 2712
		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.
			 */
2713
			bytes = min_t(unsigned long, PAGE_SIZE - offset,
2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724
						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;
}
2725
EXPORT_SYMBOL(generic_perform_write);
L
Linus Torvalds 已提交
2726

2727
/**
2728
 * __generic_file_write_iter - write data to a file
2729
 * @iocb:	IO state structure (file, offset, etc.)
2730
 * @from:	iov_iter with data to write
2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743
 *
 * This function does all the work needed for actually writing data to a
 * file. It does all basic checks, removes SUID from the file, updates
 * modification times and calls proper subroutines depending on whether we
 * do direct IO or a standard buffered write.
 *
 * It expects i_mutex to be grabbed unless we work on a block device or similar
 * object which does not need locking at all.
 *
 * This function does *not* take care of syncing data in case of O_SYNC write.
 * A caller has to handle it. This is mainly due to the fact that we want to
 * avoid syncing under i_mutex.
 */
2744
ssize_t __generic_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
L
Linus Torvalds 已提交
2745 2746
{
	struct file *file = iocb->ki_filp;
2747
	struct address_space * mapping = file->f_mapping;
L
Linus Torvalds 已提交
2748
	struct inode 	*inode = mapping->host;
2749
	ssize_t		written = 0;
L
Linus Torvalds 已提交
2750
	ssize_t		err;
2751
	ssize_t		status;
L
Linus Torvalds 已提交
2752 2753

	/* We can write back this queue in page reclaim */
2754
	current->backing_dev_info = inode_to_bdi(inode);
2755
	err = file_remove_privs(file);
L
Linus Torvalds 已提交
2756 2757 2758
	if (err)
		goto out;

2759 2760 2761
	err = file_update_time(file);
	if (err)
		goto out;
L
Linus Torvalds 已提交
2762

2763
	if (iocb->ki_flags & IOCB_DIRECT) {
2764
		loff_t pos, endbyte;
2765

2766
		written = generic_file_direct_write(iocb, from);
L
Linus Torvalds 已提交
2767
		/*
2768 2769 2770 2771 2772
		 * If the write stopped short of completing, fall back to
		 * buffered writes.  Some filesystems do this for writes to
		 * holes, for example.  For DAX files, a buffered write will
		 * not succeed (even if it did, DAX does not handle dirty
		 * page-cache pages correctly).
L
Linus Torvalds 已提交
2773
		 */
2774
		if (written < 0 || !iov_iter_count(from) || IS_DAX(inode))
2775 2776
			goto out;

2777
		status = generic_perform_write(file, from, pos = iocb->ki_pos);
2778
		/*
2779
		 * If generic_perform_write() returned a synchronous error
2780 2781 2782 2783 2784
		 * 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.
		 */
2785
		if (unlikely(status < 0)) {
2786
			err = status;
2787 2788 2789 2790 2791 2792 2793
			goto out;
		}
		/*
		 * We need to ensure that the page cache pages are written to
		 * disk and invalidated to preserve the expected O_DIRECT
		 * semantics.
		 */
2794
		endbyte = pos + status - 1;
2795
		err = filemap_write_and_wait_range(mapping, pos, endbyte);
2796
		if (err == 0) {
2797
			iocb->ki_pos = endbyte + 1;
2798
			written += status;
2799
			invalidate_mapping_pages(mapping,
2800 2801
						 pos >> PAGE_SHIFT,
						 endbyte >> PAGE_SHIFT);
2802 2803 2804 2805 2806 2807 2808
		} else {
			/*
			 * We don't know how much we wrote, so just return
			 * the number of bytes which were direct-written
			 */
		}
	} else {
2809 2810 2811
		written = generic_perform_write(file, from, iocb->ki_pos);
		if (likely(written > 0))
			iocb->ki_pos += written;
2812
	}
L
Linus Torvalds 已提交
2813 2814 2815 2816
out:
	current->backing_dev_info = NULL;
	return written ? written : err;
}
2817
EXPORT_SYMBOL(__generic_file_write_iter);
2818 2819

/**
2820
 * generic_file_write_iter - write data to a file
2821
 * @iocb:	IO state structure
2822
 * @from:	iov_iter with data to write
2823
 *
2824
 * This is a wrapper around __generic_file_write_iter() to be used by most
2825 2826 2827
 * filesystems. It takes care of syncing the file in case of O_SYNC file
 * and acquires i_mutex as needed.
 */
2828
ssize_t generic_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
L
Linus Torvalds 已提交
2829 2830
{
	struct file *file = iocb->ki_filp;
2831
	struct inode *inode = file->f_mapping->host;
L
Linus Torvalds 已提交
2832 2833
	ssize_t ret;

A
Al Viro 已提交
2834
	inode_lock(inode);
2835 2836
	ret = generic_write_checks(iocb, from);
	if (ret > 0)
2837
		ret = __generic_file_write_iter(iocb, from);
A
Al Viro 已提交
2838
	inode_unlock(inode);
L
Linus Torvalds 已提交
2839

2840 2841
	if (ret > 0)
		ret = generic_write_sync(iocb, ret);
L
Linus Torvalds 已提交
2842 2843
	return ret;
}
2844
EXPORT_SYMBOL(generic_file_write_iter);
L
Linus Torvalds 已提交
2845

2846 2847 2848 2849 2850 2851 2852 2853 2854 2855
/**
 * 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.
 *
2856 2857 2858
 * This may also be called if PG_fscache is set on a page, indicating that the
 * page is known to the local caching routines.
 *
2859
 * The @gfp_mask argument specifies whether I/O may be performed to release
2860
 * this page (__GFP_IO), and whether the call may block (__GFP_RECLAIM & __GFP_FS).
2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876
 *
 */
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);