buffer.c 88.8 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
/*
 *  linux/fs/buffer.c
 *
 *  Copyright (C) 1991, 1992, 2002  Linus Torvalds
 */

/*
 * Start bdflush() with kernel_thread not syscall - Paul Gortmaker, 12/95
 *
 * Removed a lot of unnecessary code and simplified things now that
 * the buffer cache isn't our primary cache - Andrew Tridgell 12/96
 *
 * Speed up hash, lru, and free list operations.  Use gfp() for allocating
 * hash table, use SLAB cache for buffer heads. SMP threading.  -DaveM
 *
 * Added 32k buffer block sizes - these are required older ARM systems. - RMK
 *
 * async buffer flushing, 1999 Andrea Arcangeli <[email protected]>
 */

#include <linux/kernel.h>
#include <linux/syscalls.h>
#include <linux/fs.h>
#include <linux/mm.h>
#include <linux/percpu.h>
#include <linux/slab.h>
27
#include <linux/capability.h>
Linus Torvalds's avatar
Linus Torvalds committed
28 29 30 31
#include <linux/blkdev.h>
#include <linux/file.h>
#include <linux/quotaops.h>
#include <linux/highmem.h>
32
#include <linux/export.h>
Linus Torvalds's avatar
Linus Torvalds committed
33 34 35 36
#include <linux/writeback.h>
#include <linux/hash.h>
#include <linux/suspend.h>
#include <linux/buffer_head.h>
37
#include <linux/task_io_accounting_ops.h>
Linus Torvalds's avatar
Linus Torvalds committed
38 39 40 41 42
#include <linux/bio.h>
#include <linux/notifier.h>
#include <linux/cpu.h>
#include <linux/bitops.h>
#include <linux/mpage.h>
Ingo Molnar's avatar
Ingo Molnar committed
43
#include <linux/bit_spinlock.h>
44
#include <trace/events/block.h>
Linus Torvalds's avatar
Linus Torvalds committed
45 46 47 48 49

static int fsync_buffers_list(spinlock_t *lock, struct list_head *list);

#define BH_ENTRY(list) list_entry((list), struct buffer_head, b_assoc_buffers)

50
void init_buffer(struct buffer_head *bh, bh_end_io_t *handler, void *private)
Linus Torvalds's avatar
Linus Torvalds committed
51 52 53 54
{
	bh->b_end_io = handler;
	bh->b_private = private;
}
55
EXPORT_SYMBOL(init_buffer);
Linus Torvalds's avatar
Linus Torvalds committed
56

57 58
inline void touch_buffer(struct buffer_head *bh)
{
59
	trace_block_touch_buffer(bh);
60 61 62 63
	mark_page_accessed(bh->b_page);
}
EXPORT_SYMBOL(touch_buffer);

64
void __lock_buffer(struct buffer_head *bh)
Linus Torvalds's avatar
Linus Torvalds committed
65
{
66
	wait_on_bit_lock_io(&bh->b_state, BH_Lock, TASK_UNINTERRUPTIBLE);
Linus Torvalds's avatar
Linus Torvalds committed
67 68 69
}
EXPORT_SYMBOL(__lock_buffer);

70
void unlock_buffer(struct buffer_head *bh)
Linus Torvalds's avatar
Linus Torvalds committed
71
{
72
	clear_bit_unlock(BH_Lock, &bh->b_state);
73
	smp_mb__after_atomic();
Linus Torvalds's avatar
Linus Torvalds committed
74 75
	wake_up_bit(&bh->b_state, BH_Lock);
}
76
EXPORT_SYMBOL(unlock_buffer);
Linus Torvalds's avatar
Linus Torvalds committed
77

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
/*
 * Returns if the page has dirty or writeback buffers. If all the buffers
 * are unlocked and clean then the PageDirty information is stale. If
 * any of the pages are locked, it is assumed they are locked for IO.
 */
void buffer_check_dirty_writeback(struct page *page,
				     bool *dirty, bool *writeback)
{
	struct buffer_head *head, *bh;
	*dirty = false;
	*writeback = false;

	BUG_ON(!PageLocked(page));

	if (!page_has_buffers(page))
		return;

	if (PageWriteback(page))
		*writeback = true;

	head = page_buffers(page);
	bh = head;
	do {
		if (buffer_locked(bh))
			*writeback = true;

		if (buffer_dirty(bh))
			*dirty = true;

		bh = bh->b_this_page;
	} while (bh != head);
}
EXPORT_SYMBOL(buffer_check_dirty_writeback);

Linus Torvalds's avatar
Linus Torvalds committed
112 113 114 115 116 117 118
/*
 * Block until a buffer comes unlocked.  This doesn't stop it
 * from becoming locked again - you have to lock it yourself
 * if you want to preserve its state.
 */
void __wait_on_buffer(struct buffer_head * bh)
{
119
	wait_on_bit_io(&bh->b_state, BH_Lock, TASK_UNINTERRUPTIBLE);
Linus Torvalds's avatar
Linus Torvalds committed
120
}
121
EXPORT_SYMBOL(__wait_on_buffer);
Linus Torvalds's avatar
Linus Torvalds committed
122 123 124 125 126

static void
__clear_page_buffers(struct page *page)
{
	ClearPagePrivate(page);
127
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
128 129 130
	page_cache_release(page);
}

131
static void buffer_io_error(struct buffer_head *bh, char *msg)
Linus Torvalds's avatar
Linus Torvalds committed
132 133
{
	char b[BDEVNAME_SIZE];
134 135 136 137

	if (!test_bit(BH_Quiet, &bh->b_state))
		printk_ratelimited(KERN_ERR
			"Buffer I/O error on dev %s, logical block %llu%s\n",
Linus Torvalds's avatar
Linus Torvalds committed
138
			bdevname(bh->b_bdev, b),
139
			(unsigned long long)bh->b_blocknr, msg);
Linus Torvalds's avatar
Linus Torvalds committed
140 141 142
}

/*
143 144 145 146 147 148
 * End-of-IO handler helper function which does not touch the bh after
 * unlocking it.
 * Note: unlock_buffer() sort-of does touch the bh after unlocking it, but
 * a race there is benign: unlock_buffer() only use the bh's address for
 * hashing after unlocking the buffer, so it doesn't actually touch the bh
 * itself.
Linus Torvalds's avatar
Linus Torvalds committed
149
 */
150
static void __end_buffer_read_notouch(struct buffer_head *bh, int uptodate)
Linus Torvalds's avatar
Linus Torvalds committed
151 152 153 154 155 156 157 158
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
		/* This happens, due to failed READA attempts. */
		clear_buffer_uptodate(bh);
	}
	unlock_buffer(bh);
159 160 161 162 163 164 165 166 167
}

/*
 * Default synchronous end-of-IO handler..  Just mark it up-to-date and
 * unlock the buffer. This is what ll_rw_block uses too.
 */
void end_buffer_read_sync(struct buffer_head *bh, int uptodate)
{
	__end_buffer_read_notouch(bh, uptodate);
Linus Torvalds's avatar
Linus Torvalds committed
168 169
	put_bh(bh);
}
170
EXPORT_SYMBOL(end_buffer_read_sync);
Linus Torvalds's avatar
Linus Torvalds committed
171 172 173 174 175 176

void end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
177
		buffer_io_error(bh, ", lost sync page write");
Linus Torvalds's avatar
Linus Torvalds committed
178 179 180 181 182 183
		set_buffer_write_io_error(bh);
		clear_buffer_uptodate(bh);
	}
	unlock_buffer(bh);
	put_bh(bh);
}
184
EXPORT_SYMBOL(end_buffer_write_sync);
Linus Torvalds's avatar
Linus Torvalds committed
185 186 187 188 189 190 191 192 193 194 195 196 197

/*
 * Various filesystems appear to want __find_get_block to be non-blocking.
 * But it's the page lock which protects the buffers.  To get around this,
 * we get exclusion from try_to_free_buffers with the blockdev mapping's
 * private_lock.
 *
 * Hack idea: for the blockdev mapping, i_bufferlist_lock contention
 * may be quite high.  This code could TryLock the page, and if that
 * succeeds, there is no need to take private_lock. (But if
 * private_lock is contended then so is mapping->tree_lock).
 */
static struct buffer_head *
198
__find_get_block_slow(struct block_device *bdev, sector_t block)
Linus Torvalds's avatar
Linus Torvalds committed
199 200 201 202 203 204 205 206 207 208 209
{
	struct inode *bd_inode = bdev->bd_inode;
	struct address_space *bd_mapping = bd_inode->i_mapping;
	struct buffer_head *ret = NULL;
	pgoff_t index;
	struct buffer_head *bh;
	struct buffer_head *head;
	struct page *page;
	int all_mapped = 1;

	index = block >> (PAGE_CACHE_SHIFT - bd_inode->i_blkbits);
210
	page = find_get_page_flags(bd_mapping, index, FGP_ACCESSED);
Linus Torvalds's avatar
Linus Torvalds committed
211 212 213 214 215 216 217 218 219
	if (!page)
		goto out;

	spin_lock(&bd_mapping->private_lock);
	if (!page_has_buffers(page))
		goto out_unlock;
	head = page_buffers(page);
	bh = head;
	do {
220 221 222
		if (!buffer_mapped(bh))
			all_mapped = 0;
		else if (bh->b_blocknr == block) {
Linus Torvalds's avatar
Linus Torvalds committed
223 224 225 226 227 228 229 230 231 232 233 234 235
			ret = bh;
			get_bh(bh);
			goto out_unlock;
		}
		bh = bh->b_this_page;
	} while (bh != head);

	/* we might be here because some of the buffers on this page are
	 * not mapped.  This is due to various races between
	 * file io on the block device and getblk.  It gets dealt with
	 * elsewhere, don't buffer_error if we had some unmapped buffers
	 */
	if (all_mapped) {
236 237
		char b[BDEVNAME_SIZE];

Linus Torvalds's avatar
Linus Torvalds committed
238 239
		printk("__find_get_block_slow() failed. "
			"block=%llu, b_blocknr=%llu\n",
240 241 242 243
			(unsigned long long)block,
			(unsigned long long)bh->b_blocknr);
		printk("b_state=0x%08lx, b_size=%zu\n",
			bh->b_state, bh->b_size);
244 245
		printk("device %s blocksize: %d\n", bdevname(bdev, b),
			1 << bd_inode->i_blkbits);
Linus Torvalds's avatar
Linus Torvalds committed
246 247 248 249 250 251 252 253 254
	}
out_unlock:
	spin_unlock(&bd_mapping->private_lock);
	page_cache_release(page);
out:
	return ret;
}

/*
255
 * Kick the writeback threads then try to free up some ZONE_NORMAL memory.
Linus Torvalds's avatar
Linus Torvalds committed
256 257 258
 */
static void free_more_memory(void)
{
259
	struct zone *zone;
260
	int nid;
Linus Torvalds's avatar
Linus Torvalds committed
261

262
	wakeup_flusher_threads(1024, WB_REASON_FREE_MORE_MEM);
Linus Torvalds's avatar
Linus Torvalds committed
263 264
	yield();

265
	for_each_online_node(nid) {
266 267 268 269
		(void)first_zones_zonelist(node_zonelist(nid, GFP_NOFS),
						gfp_zone(GFP_NOFS), NULL,
						&zone);
		if (zone)
270
			try_to_free_pages(node_zonelist(nid, GFP_NOFS), 0,
271
						GFP_NOFS, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
272 273 274 275 276 277 278 279 280 281
	}
}

/*
 * I/O completion handler for block_read_full_page() - pages
 * which come unlocked at the end of I/O.
 */
static void end_buffer_async_read(struct buffer_head *bh, int uptodate)
{
	unsigned long flags;
282
	struct buffer_head *first;
Linus Torvalds's avatar
Linus Torvalds committed
283 284 285 286 287 288 289 290 291 292 293
	struct buffer_head *tmp;
	struct page *page;
	int page_uptodate = 1;

	BUG_ON(!buffer_async_read(bh));

	page = bh->b_page;
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
		clear_buffer_uptodate(bh);
294
		buffer_io_error(bh, ", async page read");
Linus Torvalds's avatar
Linus Torvalds committed
295 296 297 298 299 300 301 302
		SetPageError(page);
	}

	/*
	 * Be _very_ careful from here on. Bad things can happen if
	 * two buffer heads end IO at almost the same time and both
	 * decide that the page is now completely done.
	 */
303 304 305
	first = page_buffers(page);
	local_irq_save(flags);
	bit_spin_lock(BH_Uptodate_Lock, &first->b_state);
Linus Torvalds's avatar
Linus Torvalds committed
306 307 308 309 310 311 312 313 314 315 316 317
	clear_buffer_async_read(bh);
	unlock_buffer(bh);
	tmp = bh;
	do {
		if (!buffer_uptodate(tmp))
			page_uptodate = 0;
		if (buffer_async_read(tmp)) {
			BUG_ON(!buffer_locked(tmp));
			goto still_busy;
		}
		tmp = tmp->b_this_page;
	} while (tmp != bh);
318 319
	bit_spin_unlock(BH_Uptodate_Lock, &first->b_state);
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
320 321 322 323 324 325 326 327 328 329 330

	/*
	 * If none of the buffers had errors and they are all
	 * uptodate then we can set the page uptodate.
	 */
	if (page_uptodate && !PageError(page))
		SetPageUptodate(page);
	unlock_page(page);
	return;

still_busy:
331 332
	bit_spin_unlock(BH_Uptodate_Lock, &first->b_state);
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
333 334 335 336 337 338 339
	return;
}

/*
 * Completion handler for block_write_full_page() - pages which are unlocked
 * during I/O, and which have PageWriteback cleared upon I/O completion.
 */
340
void end_buffer_async_write(struct buffer_head *bh, int uptodate)
Linus Torvalds's avatar
Linus Torvalds committed
341 342
{
	unsigned long flags;
343
	struct buffer_head *first;
Linus Torvalds's avatar
Linus Torvalds committed
344 345 346 347 348 349 350 351 352
	struct buffer_head *tmp;
	struct page *page;

	BUG_ON(!buffer_async_write(bh));

	page = bh->b_page;
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
353
		buffer_io_error(bh, ", lost async page write");
Linus Torvalds's avatar
Linus Torvalds committed
354
		set_bit(AS_EIO, &page->mapping->flags);
355
		set_buffer_write_io_error(bh);
Linus Torvalds's avatar
Linus Torvalds committed
356 357 358 359
		clear_buffer_uptodate(bh);
		SetPageError(page);
	}

360 361 362 363
	first = page_buffers(page);
	local_irq_save(flags);
	bit_spin_lock(BH_Uptodate_Lock, &first->b_state);

Linus Torvalds's avatar
Linus Torvalds committed
364 365 366 367 368 369 370 371 372 373
	clear_buffer_async_write(bh);
	unlock_buffer(bh);
	tmp = bh->b_this_page;
	while (tmp != bh) {
		if (buffer_async_write(tmp)) {
			BUG_ON(!buffer_locked(tmp));
			goto still_busy;
		}
		tmp = tmp->b_this_page;
	}
374 375
	bit_spin_unlock(BH_Uptodate_Lock, &first->b_state);
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
376 377 378 379
	end_page_writeback(page);
	return;

still_busy:
380 381
	bit_spin_unlock(BH_Uptodate_Lock, &first->b_state);
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
382 383
	return;
}
384
EXPORT_SYMBOL(end_buffer_async_write);
Linus Torvalds's avatar
Linus Torvalds committed
385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412

/*
 * If a page's buffers are under async readin (end_buffer_async_read
 * completion) then there is a possibility that another thread of
 * control could lock one of the buffers after it has completed
 * but while some of the other buffers have not completed.  This
 * locked buffer would confuse end_buffer_async_read() into not unlocking
 * the page.  So the absence of BH_Async_Read tells end_buffer_async_read()
 * that this buffer is not under async I/O.
 *
 * The page comes unlocked when it has no locked buffer_async buffers
 * left.
 *
 * PageLocked prevents anyone starting new async I/O reads any of
 * the buffers.
 *
 * PageWriteback is used to prevent simultaneous writeout of the same
 * page.
 *
 * PageLocked prevents anyone from starting writeback of a page which is
 * under read I/O (PageWriteback is only ever set against a locked page).
 */
static void mark_buffer_async_read(struct buffer_head *bh)
{
	bh->b_end_io = end_buffer_async_read;
	set_buffer_async_read(bh);
}

413 414
static void mark_buffer_async_write_endio(struct buffer_head *bh,
					  bh_end_io_t *handler)
Linus Torvalds's avatar
Linus Torvalds committed
415
{
416
	bh->b_end_io = handler;
Linus Torvalds's avatar
Linus Torvalds committed
417 418
	set_buffer_async_write(bh);
}
419 420 421 422 423

void mark_buffer_async_write(struct buffer_head *bh)
{
	mark_buffer_async_write_endio(bh, end_buffer_async_write);
}
Linus Torvalds's avatar
Linus Torvalds committed
424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478
EXPORT_SYMBOL(mark_buffer_async_write);


/*
 * fs/buffer.c contains helper functions for buffer-backed address space's
 * fsync functions.  A common requirement for buffer-based filesystems is
 * that certain data from the backing blockdev needs to be written out for
 * a successful fsync().  For example, ext2 indirect blocks need to be
 * written back and waited upon before fsync() returns.
 *
 * The functions mark_buffer_inode_dirty(), fsync_inode_buffers(),
 * inode_has_buffers() and invalidate_inode_buffers() are provided for the
 * management of a list of dependent buffers at ->i_mapping->private_list.
 *
 * Locking is a little subtle: try_to_free_buffers() will remove buffers
 * from their controlling inode's queue when they are being freed.  But
 * try_to_free_buffers() will be operating against the *blockdev* mapping
 * at the time, not against the S_ISREG file which depends on those buffers.
 * So the locking for private_list is via the private_lock in the address_space
 * which backs the buffers.  Which is different from the address_space 
 * against which the buffers are listed.  So for a particular address_space,
 * mapping->private_lock does *not* protect mapping->private_list!  In fact,
 * mapping->private_list will always be protected by the backing blockdev's
 * ->private_lock.
 *
 * Which introduces a requirement: all buffers on an address_space's
 * ->private_list must be from the same address_space: the blockdev's.
 *
 * address_spaces which do not place buffers at ->private_list via these
 * utility functions are free to use private_lock and private_list for
 * whatever they want.  The only requirement is that list_empty(private_list)
 * be true at clear_inode() time.
 *
 * FIXME: clear_inode should not call invalidate_inode_buffers().  The
 * filesystems should do that.  invalidate_inode_buffers() should just go
 * BUG_ON(!list_empty).
 *
 * FIXME: mark_buffer_dirty_inode() is a data-plane operation.  It should
 * take an address_space, not an inode.  And it should be called
 * mark_buffer_dirty_fsync() to clearly define why those buffers are being
 * queued up.
 *
 * FIXME: mark_buffer_dirty_inode() doesn't need to add the buffer to the
 * list if it is already on a list.  Because if the buffer is on a list,
 * it *must* already be on the right one.  If not, the filesystem is being
 * silly.  This will save a ton of locking.  But first we have to ensure
 * that buffers are taken *off* the old inode's list when they are freed
 * (presumably in truncate).  That requires careful auditing of all
 * filesystems (do it inside bforget()).  It could also be done by bringing
 * b_inode back.
 */

/*
 * The buffer's backing address_space's private_lock must be held
 */
479
static void __remove_assoc_queue(struct buffer_head *bh)
Linus Torvalds's avatar
Linus Torvalds committed
480 481
{
	list_del_init(&bh->b_assoc_buffers);
482 483 484 485
	WARN_ON(!bh->b_assoc_map);
	if (buffer_write_io_error(bh))
		set_bit(AS_EIO, &bh->b_assoc_map->flags);
	bh->b_assoc_map = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527
}

int inode_has_buffers(struct inode *inode)
{
	return !list_empty(&inode->i_data.private_list);
}

/*
 * osync is designed to support O_SYNC io.  It waits synchronously for
 * all already-submitted IO to complete, but does not queue any new
 * writes to the disk.
 *
 * To do O_SYNC writes, just queue the buffer writes with ll_rw_block as
 * you dirty the buffers, and then use osync_inode_buffers to wait for
 * completion.  Any other dirty buffers which are not yet queued for
 * write will not be flushed to disk by the osync.
 */
static int osync_buffers_list(spinlock_t *lock, struct list_head *list)
{
	struct buffer_head *bh;
	struct list_head *p;
	int err = 0;

	spin_lock(lock);
repeat:
	list_for_each_prev(p, list) {
		bh = BH_ENTRY(p);
		if (buffer_locked(bh)) {
			get_bh(bh);
			spin_unlock(lock);
			wait_on_buffer(bh);
			if (!buffer_uptodate(bh))
				err = -EIO;
			brelse(bh);
			spin_lock(lock);
			goto repeat;
		}
	}
	spin_unlock(lock);
	return err;
}

Al Viro's avatar
Al Viro committed
528
static void do_thaw_one(struct super_block *sb, void *unused)
529 530
{
	char b[BDEVNAME_SIZE];
Al Viro's avatar
Al Viro committed
531 532 533 534
	while (sb->s_bdev && !thaw_bdev(sb->s_bdev, sb))
		printk(KERN_WARNING "Emergency Thaw on %s\n",
		       bdevname(sb->s_bdev, b));
}
535

Al Viro's avatar
Al Viro committed
536 537 538
static void do_thaw_all(struct work_struct *work)
{
	iterate_supers(do_thaw_one, NULL);
539
	kfree(work);
540 541 542 543 544 545 546 547 548 549
	printk(KERN_WARNING "Emergency Thaw complete\n");
}

/**
 * emergency_thaw_all -- forcibly thaw every frozen filesystem
 *
 * Used for emergency unfreeze of all filesystems via SysRq
 */
void emergency_thaw_all(void)
{
550 551 552 553 554 555 556
	struct work_struct *work;

	work = kmalloc(sizeof(*work), GFP_ATOMIC);
	if (work) {
		INIT_WORK(work, do_thaw_all);
		schedule_work(work);
	}
557 558
}

Linus Torvalds's avatar
Linus Torvalds committed
559
/**
560
 * sync_mapping_buffers - write out & wait upon a mapping's "associated" buffers
561
 * @mapping: the mapping which wants those buffers written
Linus Torvalds's avatar
Linus Torvalds committed
562 563 564 565
 *
 * Starts I/O against the buffers at mapping->private_list, and waits upon
 * that I/O.
 *
566 567 568
 * Basically, this is a convenience function for fsync().
 * @mapping is a file or directory which needs those buffers to be written for
 * a successful fsync().
Linus Torvalds's avatar
Linus Torvalds committed
569 570 571
 */
int sync_mapping_buffers(struct address_space *mapping)
{
572
	struct address_space *buffer_mapping = mapping->private_data;
Linus Torvalds's avatar
Linus Torvalds committed
573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604

	if (buffer_mapping == NULL || list_empty(&mapping->private_list))
		return 0;

	return fsync_buffers_list(&buffer_mapping->private_lock,
					&mapping->private_list);
}
EXPORT_SYMBOL(sync_mapping_buffers);

/*
 * Called when we've recently written block `bblock', and it is known that
 * `bblock' was for a buffer_boundary() buffer.  This means that the block at
 * `bblock + 1' is probably a dirty indirect block.  Hunt it down and, if it's
 * dirty, schedule it for IO.  So that indirects merge nicely with their data.
 */
void write_boundary_block(struct block_device *bdev,
			sector_t bblock, unsigned blocksize)
{
	struct buffer_head *bh = __find_get_block(bdev, bblock + 1, blocksize);
	if (bh) {
		if (buffer_dirty(bh))
			ll_rw_block(WRITE, 1, &bh);
		put_bh(bh);
	}
}

void mark_buffer_dirty_inode(struct buffer_head *bh, struct inode *inode)
{
	struct address_space *mapping = inode->i_mapping;
	struct address_space *buffer_mapping = bh->b_page->mapping;

	mark_buffer_dirty(bh);
605 606
	if (!mapping->private_data) {
		mapping->private_data = buffer_mapping;
Linus Torvalds's avatar
Linus Torvalds committed
607
	} else {
608
		BUG_ON(mapping->private_data != buffer_mapping);
Linus Torvalds's avatar
Linus Torvalds committed
609
	}
610
	if (!bh->b_assoc_map) {
Linus Torvalds's avatar
Linus Torvalds committed
611 612 613
		spin_lock(&buffer_mapping->private_lock);
		list_move_tail(&bh->b_assoc_buffers,
				&mapping->private_list);
614
		bh->b_assoc_map = mapping;
Linus Torvalds's avatar
Linus Torvalds committed
615 616 617 618 619
		spin_unlock(&buffer_mapping->private_lock);
	}
}
EXPORT_SYMBOL(mark_buffer_dirty_inode);

620 621 622 623 624 625 626
/*
 * Mark the page dirty, and set it dirty in the radix tree, and mark the inode
 * dirty.
 *
 * If warn is true, then emit a warning if the page is not uptodate and has
 * not been truncated.
 */
627
static void __set_page_dirty(struct page *page,
628 629
		struct address_space *mapping, int warn)
{
630 631 632
	unsigned long flags;

	spin_lock_irqsave(&mapping->tree_lock, flags);
633 634
	if (page->mapping) {	/* Race with truncate? */
		WARN_ON_ONCE(warn && !PageUptodate(page));
635
		account_page_dirtied(page, mapping);
636 637 638
		radix_tree_tag_set(&mapping->page_tree,
				page_index(page), PAGECACHE_TAG_DIRTY);
	}
639
	spin_unlock_irqrestore(&mapping->tree_lock, flags);
640 641 642
	__mark_inode_dirty(mapping->host, I_DIRTY_PAGES);
}

Linus Torvalds's avatar
Linus Torvalds committed
643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669
/*
 * Add a page to the dirty page list.
 *
 * It is a sad fact of life that this function is called from several places
 * deeply under spinlocking.  It may not sleep.
 *
 * If the page has buffers, the uptodate buffers are set dirty, to preserve
 * dirty-state coherency between the page and the buffers.  It the page does
 * not have buffers then when they are later attached they will all be set
 * dirty.
 *
 * The buffers are dirtied before the page is dirtied.  There's a small race
 * window in which a writepage caller may see the page cleanness but not the
 * buffer dirtiness.  That's fine.  If this code were to set the page dirty
 * before the buffers, a concurrent writepage caller could clear the page dirty
 * bit, see a bunch of clean buffers and we'd end up with dirty buffers/clean
 * page on the dirty page list.
 *
 * We use private_lock to lock against try_to_free_buffers while using the
 * page's buffer list.  Also use this to protect against clean buffers being
 * added to the page after it was set dirty.
 *
 * FIXME: may need to call ->reservepage here as well.  That's rather up to the
 * address_space though.
 */
int __set_page_dirty_buffers(struct page *page)
{
670
	int newly_dirty;
671
	struct address_space *mapping = page_mapping(page);
672 673 674

	if (unlikely(!mapping))
		return !TestSetPageDirty(page);
Linus Torvalds's avatar
Linus Torvalds committed
675 676 677 678 679 680 681 682 683 684 685

	spin_lock(&mapping->private_lock);
	if (page_has_buffers(page)) {
		struct buffer_head *head = page_buffers(page);
		struct buffer_head *bh = head;

		do {
			set_buffer_dirty(bh);
			bh = bh->b_this_page;
		} while (bh != head);
	}
686
	newly_dirty = !TestSetPageDirty(page);
Linus Torvalds's avatar
Linus Torvalds committed
687 688
	spin_unlock(&mapping->private_lock);

689 690 691
	if (newly_dirty)
		__set_page_dirty(page, mapping, 1);
	return newly_dirty;
Linus Torvalds's avatar
Linus Torvalds committed
692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
}
EXPORT_SYMBOL(__set_page_dirty_buffers);

/*
 * Write out and wait upon a list of buffers.
 *
 * We have conflicting pressures: we want to make sure that all
 * initially dirty buffers get waited on, but that any subsequently
 * dirtied buffers don't.  After all, we don't want fsync to last
 * forever if somebody is actively writing to the file.
 *
 * Do this in two main stages: first we copy dirty buffers to a
 * temporary inode list, queueing the writes as we go.  Then we clean
 * up, waiting for those writes to complete.
 * 
 * During this second stage, any subsequent updates to the file may end
 * up refiling the buffer on the original inode's dirty list again, so
 * there is a chance we will end up with a buffer queued for write but
 * not yet completed on that list.  So, as a final cleanup we go through
 * the osync code to catch these locked, dirty buffers without requeuing
 * any newly dirty buffers for write.
 */
static int fsync_buffers_list(spinlock_t *lock, struct list_head *list)
{
	struct buffer_head *bh;
	struct list_head tmp;
718
	struct address_space *mapping;
Linus Torvalds's avatar
Linus Torvalds committed
719
	int err = 0, err2;
720
	struct blk_plug plug;
Linus Torvalds's avatar
Linus Torvalds committed
721 722

	INIT_LIST_HEAD(&tmp);
723
	blk_start_plug(&plug);
Linus Torvalds's avatar
Linus Torvalds committed
724 725 726 727

	spin_lock(lock);
	while (!list_empty(list)) {
		bh = BH_ENTRY(list->next);
728
		mapping = bh->b_assoc_map;
729
		__remove_assoc_queue(bh);
730 731 732
		/* Avoid race with mark_buffer_dirty_inode() which does
		 * a lockless check and we rely on seeing the dirty bit */
		smp_mb();
Linus Torvalds's avatar
Linus Torvalds committed
733 734
		if (buffer_dirty(bh) || buffer_locked(bh)) {
			list_add(&bh->b_assoc_buffers, &tmp);
735
			bh->b_assoc_map = mapping;
Linus Torvalds's avatar
Linus Torvalds committed
736 737 738 739 740
			if (buffer_dirty(bh)) {
				get_bh(bh);
				spin_unlock(lock);
				/*
				 * Ensure any pending I/O completes so that
Christoph Hellwig's avatar
Christoph Hellwig committed
741 742 743 744
				 * write_dirty_buffer() actually writes the
				 * current contents - it is a noop if I/O is
				 * still in flight on potentially older
				 * contents.
Linus Torvalds's avatar
Linus Torvalds committed
745
				 */
Jens Axboe's avatar
Jens Axboe committed
746
				write_dirty_buffer(bh, WRITE_SYNC);
747 748 749 750 751 752 753

				/*
				 * Kick off IO for the previous mapping. Note
				 * that we will not run the very last mapping,
				 * wait_on_buffer() will do that for us
				 * through sync_buffer().
				 */
Linus Torvalds's avatar
Linus Torvalds committed
754 755 756 757 758 759
				brelse(bh);
				spin_lock(lock);
			}
		}
	}

760 761 762 763
	spin_unlock(lock);
	blk_finish_plug(&plug);
	spin_lock(lock);

Linus Torvalds's avatar
Linus Torvalds committed
764 765 766
	while (!list_empty(&tmp)) {
		bh = BH_ENTRY(tmp.prev);
		get_bh(bh);
767 768 769 770 771 772 773
		mapping = bh->b_assoc_map;
		__remove_assoc_queue(bh);
		/* Avoid race with mark_buffer_dirty_inode() which does
		 * a lockless check and we rely on seeing the dirty bit */
		smp_mb();
		if (buffer_dirty(bh)) {
			list_add(&bh->b_assoc_buffers,
774
				 &mapping->private_list);
775 776
			bh->b_assoc_map = mapping;
		}
Linus Torvalds's avatar
Linus Torvalds committed
777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806
		spin_unlock(lock);
		wait_on_buffer(bh);
		if (!buffer_uptodate(bh))
			err = -EIO;
		brelse(bh);
		spin_lock(lock);
	}
	
	spin_unlock(lock);
	err2 = osync_buffers_list(lock, list);
	if (err)
		return err;
	else
		return err2;
}

/*
 * Invalidate any and all dirty buffers on a given inode.  We are
 * probably unmounting the fs, but that doesn't mean we have already
 * done a sync().  Just drop the buffers from the inode list.
 *
 * NOTE: we take the inode's blockdev's mapping's private_lock.  Which
 * assumes that all the buffers are against the blockdev.  Not true
 * for reiserfs.
 */
void invalidate_inode_buffers(struct inode *inode)
{
	if (inode_has_buffers(inode)) {
		struct address_space *mapping = &inode->i_data;
		struct list_head *list = &mapping->private_list;
807
		struct address_space *buffer_mapping = mapping->private_data;
Linus Torvalds's avatar
Linus Torvalds committed
808 809 810 811 812 813 814

		spin_lock(&buffer_mapping->private_lock);
		while (!list_empty(list))
			__remove_assoc_queue(BH_ENTRY(list->next));
		spin_unlock(&buffer_mapping->private_lock);
	}
}
815
EXPORT_SYMBOL(invalidate_inode_buffers);
Linus Torvalds's avatar
Linus Torvalds committed
816 817 818 819 820 821 822 823 824 825 826 827 828 829

/*
 * Remove any clean buffers from the inode's buffer list.  This is called
 * when we're trying to free the inode itself.  Those buffers can pin it.
 *
 * Returns true if all buffers were removed.
 */
int remove_inode_buffers(struct inode *inode)
{
	int ret = 1;

	if (inode_has_buffers(inode)) {
		struct address_space *mapping = &inode->i_data;
		struct list_head *list = &mapping->private_list;
830
		struct address_space *buffer_mapping = mapping->private_data;
Linus Torvalds's avatar
Linus Torvalds committed
831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924

		spin_lock(&buffer_mapping->private_lock);
		while (!list_empty(list)) {
			struct buffer_head *bh = BH_ENTRY(list->next);
			if (buffer_dirty(bh)) {
				ret = 0;
				break;
			}
			__remove_assoc_queue(bh);
		}
		spin_unlock(&buffer_mapping->private_lock);
	}
	return ret;
}

/*
 * Create the appropriate buffers when given a page for data area and
 * the size of each buffer.. Use the bh->b_this_page linked list to
 * follow the buffers created.  Return NULL if unable to create more
 * buffers.
 *
 * The retry flag is used to differentiate async IO (paging, swapping)
 * which may not fail from ordinary buffer allocations.
 */
struct buffer_head *alloc_page_buffers(struct page *page, unsigned long size,
		int retry)
{
	struct buffer_head *bh, *head;
	long offset;

try_again:
	head = NULL;
	offset = PAGE_SIZE;
	while ((offset -= size) >= 0) {
		bh = alloc_buffer_head(GFP_NOFS);
		if (!bh)
			goto no_grow;

		bh->b_this_page = head;
		bh->b_blocknr = -1;
		head = bh;

		bh->b_size = size;

		/* Link the buffer to its page */
		set_bh_page(bh, page, offset);
	}
	return head;
/*
 * In case anything failed, we just free everything we got.
 */
no_grow:
	if (head) {
		do {
			bh = head;
			head = head->b_this_page;
			free_buffer_head(bh);
		} while (head);
	}

	/*
	 * Return failure for non-async IO requests.  Async IO requests
	 * are not allowed to fail, so we have to wait until buffer heads
	 * become available.  But we don't want tasks sleeping with 
	 * partially complete buffers, so all were released above.
	 */
	if (!retry)
		return NULL;

	/* We're _really_ low on memory. Now we just
	 * wait for old buffer heads to become free due to
	 * finishing IO.  Since this is an async request and
	 * the reserve list is empty, we're sure there are 
	 * async buffer heads in use.
	 */
	free_more_memory();
	goto try_again;
}
EXPORT_SYMBOL_GPL(alloc_page_buffers);

static inline void
link_dev_buffers(struct page *page, struct buffer_head *head)
{
	struct buffer_head *bh, *tail;

	bh = head;
	do {
		tail = bh;
		bh = bh->b_this_page;
	} while (bh);
	tail->b_this_page = head;
	attach_page_buffers(page, head);
}

925 926 927 928 929 930 931 932 933 934 935 936
static sector_t blkdev_max_block(struct block_device *bdev, unsigned int size)
{
	sector_t retval = ~((sector_t)0);
	loff_t sz = i_size_read(bdev->bd_inode);

	if (sz) {
		unsigned int sizebits = blksize_bits(size);
		retval = (sz >> sizebits);
	}
	return retval;
}

Linus Torvalds's avatar
Linus Torvalds committed
937 938 939
/*
 * Initialise the state of a blockdev page's buffers.
 */ 
940
static sector_t
Linus Torvalds's avatar
Linus Torvalds committed
941 942 943 944 945 946
init_page_buffers(struct page *page, struct block_device *bdev,
			sector_t block, int size)
{
	struct buffer_head *head = page_buffers(page);
	struct buffer_head *bh = head;
	int uptodate = PageUptodate(page);
947
	sector_t end_block = blkdev_max_block(I_BDEV(bdev->bd_inode), size);
Linus Torvalds's avatar
Linus Torvalds committed
948 949 950 951 952 953 954 955

	do {
		if (!buffer_mapped(bh)) {
			init_buffer(bh, NULL, NULL);
			bh->b_bdev = bdev;
			bh->b_blocknr = block;
			if (uptodate)
				set_buffer_uptodate(bh);
956 957
			if (block < end_block)
				set_buffer_mapped(bh);
Linus Torvalds's avatar
Linus Torvalds committed
958 959 960 961
		}
		block++;
		bh = bh->b_this_page;
	} while (bh != head);
962 963 964 965 966

	/*
	 * Caller needs to validate requested block against end of device.
	 */
	return end_block;
Linus Torvalds's avatar
Linus Torvalds committed
967 968 969 970 971
}

/*
 * Create the page-cache page that contains the requested block.
 *
972
 * This is used purely for blockdev mappings.
Linus Torvalds's avatar
Linus Torvalds committed
973
 */
974
static int
Linus Torvalds's avatar
Linus Torvalds committed
975
grow_dev_page(struct block_device *bdev, sector_t block,
976
	      pgoff_t index, int size, int sizebits, gfp_t gfp)
Linus Torvalds's avatar
Linus Torvalds committed
977 978 979 980
{
	struct inode *inode = bdev->bd_inode;
	struct page *page;
	struct buffer_head *bh;
981 982
	sector_t end_block;
	int ret = 0;		/* Will call free_more_memory() */
983
	gfp_t gfp_mask;
Linus Torvalds's avatar
Linus Torvalds committed
984

985 986
	gfp_mask = (mapping_gfp_mask(inode->i_mapping) & ~__GFP_FS) | gfp;

987 988 989 990 991 992 993 994 995
	/*
	 * XXX: __getblk_slow() can not really deal with failure and
	 * will endlessly loop on improvised global reclaim.  Prefer
	 * looping in the allocator rather than here, at least that
	 * code knows what it's doing.
	 */
	gfp_mask |= __GFP_NOFAIL;

	page = find_or_create_page(inode->i_mapping, index, gfp_mask);
Linus Torvalds's avatar
Linus Torvalds committed
996
	if (!page)
997
		return ret;
Linus Torvalds's avatar
Linus Torvalds committed
998

999
	BUG_ON(!PageLocked(page));
Linus Torvalds's avatar
Linus Torvalds committed
1000 1001 1002 1003

	if (page_has_buffers(page)) {
		bh = page_buffers(page);
		if (bh->b_size == size) {
1004
			end_block = init_page_buffers(page, bdev,
1005 1006
						(sector_t)index << sizebits,
						size);
1007
			goto done;
Linus Torvalds's avatar
Linus Torvalds committed
1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026
		}
		if (!try_to_free_buffers(page))
			goto failed;
	}

	/*
	 * Allocate some buffers for this page
	 */
	bh = alloc_page_buffers(page, size, 0);
	if (!bh)
		goto failed;

	/*
	 * Link the page to the buffers and initialise them.  Take the
	 * lock to be atomic wrt __find_get_block(), which does not
	 * run under the page lock.
	 */
	spin_lock(&inode->i_mapping->private_lock);
	link_dev_buffers(page, bh);
1027 1028
	end_block = init_page_buffers(page, bdev, (sector_t)index << sizebits,
			size);
Linus Torvalds's avatar
Linus Torvalds committed
1029
	spin_unlock(&inode->i_mapping->private_lock);
1030 1031
done:
	ret = (block < end_block) ? 1 : -ENXIO;
Linus Torvalds's avatar
Linus Torvalds committed
1032 1033 1034
failed:
	unlock_page(page);
	page_cache_release(page);
1035
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
1036 1037 1038 1039 1040 1041
}

/*
 * Create buffers for the specified block device block's page.  If
 * that page was dirty, the buffers are set dirty also.
 */
1042
static int
1043
grow_buffers(struct block_device *bdev, sector_t block, int size, gfp_t gfp)
Linus Torvalds's avatar
Linus Torvalds committed
1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
{
	pgoff_t index;
	int sizebits;

	sizebits = -1;
	do {
		sizebits++;
	} while ((size << sizebits) < PAGE_SIZE);

	index = block >> sizebits;

1055 1056 1057 1058 1059 1060 1061 1062 1063
	/*
	 * Check for a block which wants to lie outside our maximum possible
	 * pagecache index.  (this comparison is done using sector_t types).
	 */
	if (unlikely(index != block >> sizebits)) {
		char b[BDEVNAME_SIZE];

		printk(KERN_ERR "%s: requested out-of-range block %llu for "
			"device %s\n",
1064
			__func__, (unsigned long long)block,
1065 1066 1067
			bdevname(bdev, b));
		return -EIO;
	}
1068

Linus Torvalds's avatar
Linus Torvalds committed
1069
	/* Create a page with the proper size buffers.. */
1070
	return grow_dev_page(bdev, block, index, size, sizebits, gfp);
Linus Torvalds's avatar
Linus Torvalds committed
1071 1072
}

1073 1074 1075
struct buffer_head *
__getblk_slow(struct block_device *bdev, sector_t block,
	     unsigned size, gfp_t gfp)
Linus Torvalds's avatar
Linus Torvalds committed
1076 1077
{
	/* Size must be multiple of hard sectorsize */
1078
	if (unlikely(size & (bdev_logical_block_size(bdev)-1) ||
Linus Torvalds's avatar
Linus Torvalds committed
1079 1080 1081
			(size < 512 || size > PAGE_SIZE))) {
		printk(KERN_ERR "getblk(): invalid block size %d requested\n",
					size);
1082 1083
		printk(KERN_ERR "logical block size: %d\n",
					bdev_logical_block_size(bdev));
Linus Torvalds's avatar
Linus Torvalds committed
1084 1085 1086 1087 1088

		dump_stack();
		return NULL;
	}

1089 1090 1091
	for (;;) {
		struct buffer_head *bh;
		int ret;
Linus Torvalds's avatar
Linus Torvalds committed
1092 1093 1094 1095

		bh = __find_get_block(bdev, block, size);
		if (bh)
			return bh;
1096

1097
		ret = grow_buffers(bdev, block, size, gfp);
1098 1099 1100 1101
		if (ret < 0)
			return NULL;
		if (ret == 0)
			free_more_memory();
Linus Torvalds's avatar
Linus Torvalds committed
1102 1103
	}
}
1104
EXPORT_SYMBOL(__getblk_slow);
Linus Torvalds's avatar
Linus Torvalds committed
1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130

/*
 * The relationship between dirty buffers and dirty pages:
 *
 * Whenever a page has any dirty buffers, the page's dirty bit is set, and
 * the page is tagged dirty in its radix tree.
 *
 * At all times, the dirtiness of the buffers represents the dirtiness of
 * subsections of the page.  If the page has buffers, the page dirty bit is
 * merely a hint about the true dirty state.
 *
 * When a page is set dirty in its entirety, all its buffers are marked dirty
 * (if the page has buffers).
 *
 * When a buffer is marked dirty, its page is dirtied, but the page's other
 * buffers are not.
 *
 * Also.  When blockdev buffers are explicitly read with bread(), they
 * individually become uptodate.  But their backing page remains not
 * uptodate - even if all of its buffers are uptodate.  A subsequent
 * block_read_full_page() against that page will discover all the uptodate
 * buffers, will set the page uptodate and will perform no I/O.
 */

/**
 * mark_buffer_dirty - mark a buffer_head as needing writeout
1131
 * @bh: the buffer_head to mark dirty
Linus Torvalds's avatar
Linus Torvalds committed
1132 1133 1134 1135 1136 1137 1138
 *
 * mark_buffer_dirty() will set the dirty bit against the buffer, then set its
 * backing page dirty, then tag the page as dirty in its address_space's radix
 * tree and then attach the address_space's inode to its superblock's dirty
 * inode list.
 *
 * mark_buffer_dirty() is atomic.  It takes bh->b_page->mapping->private_lock,
1139
 * mapping->tree_lock and mapping->host->i_lock.
Linus Torvalds's avatar
Linus Torvalds committed
1140
 */
1141
void mark_buffer_dirty(struct buffer_head *bh)
Linus Torvalds's avatar
Linus Torvalds committed
1142
{
1143
	WARN_ON_ONCE(!buffer_uptodate(bh));
1144

1145 1146
	trace_block_dirty_buffer(bh);

1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158
	/*
	 * Very *carefully* optimize the it-is-already-dirty case.
	 *
	 * Don't let the final "is it dirty" escape to before we
	 * perhaps modified the buffer.
	 */
	if (buffer_dirty(bh)) {
		smp_mb();
		if (buffer_dirty(bh))
			return;
	}

1159 1160
	if (!test_set_buffer_dirty(bh)) {
		struct page *page = bh->b_page;
1161 1162 1163 1164 1165
		if (!TestSetPageDirty(page)) {
			struct address_space *mapping = page_mapping(page);
			if (mapping)
				__set_page_dirty(page, mapping, 0);
		}
1166
	}
Linus Torvalds's avatar
Linus Torvalds committed
1167
}
1168
EXPORT_SYMBOL(mark_buffer_dirty);
Linus Torvalds's avatar
Linus Torvalds committed
1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182

/*
 * Decrement a buffer_head's reference count.  If all buffers against a page
 * have zero reference count, are clean and unlocked, and if the page is clean
 * and unlocked then try_to_free_buffers() may strip the buffers from the page
 * in preparation for freeing it (sometimes, rarely, buffers are removed from
 * a page but it ends up not being freed, and buffers may later be reattached).
 */
void __brelse(struct buffer_head * buf)
{
	if (atomic_read(&buf->b_count)) {
		put_bh(buf);
		return;
	}
Arjan van de Ven's avatar
Arjan van de Ven committed
1183
	WARN(1, KERN_ERR "VFS: brelse: Trying to free free buffer\n");
Linus Torvalds's avatar
Linus Torvalds committed
1184
}
1185
EXPORT_SYMBOL(__brelse);
Linus Torvalds's avatar
Linus Torvalds committed
1186 1187 1188 1189 1190 1191 1192 1193

/*
 * bforget() is like brelse(), except it discards any
 * potentially dirty data.
 */
void __bforget(struct buffer_head *bh)
{
	clear_buffer_dirty(bh);
1194
	if (bh->b_assoc_map) {
Linus Torvalds's avatar
Linus Torvalds committed
1195 1196 1197 1198
		struct address_space *buffer_mapping = bh->b_page->mapping;

		spin_lock(&buffer_mapping->private_lock);
		list_del_init(&bh->b_assoc_buffers);
1199
		bh->b_assoc_map = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
1200 1201 1202 1203
		spin_unlock(&buffer_mapping->private_lock);
	}
	__brelse(bh);
}
1204
EXPORT_SYMBOL(__bforget);
Linus Torvalds's avatar
Linus Torvalds committed
1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237

static struct buffer_head *__bread_slow(struct buffer_head *bh)
{
	lock_buffer(bh);
	if (buffer_uptodate(bh)) {
		unlock_buffer(bh);
		return bh;
	} else {
		get_bh(bh);
		bh->b_end_io = end_buffer_read_sync;
		submit_bh(READ, bh);
		wait_on_buffer(bh);
		if (buffer_uptodate(bh))
			return bh;
	}
	brelse(bh);
	return NULL;
}

/*
 * Per-cpu buffer LRU implementation.  To reduce the cost of __find_get_block().
 * The bhs[] array is sorted - newest buffer is at bhs[0].  Buffers have their
 * refcount elevated by one when they're in an LRU.  A buffer can only appear
 * once in a particular CPU's LRU.  A single buffer can be present in multiple
 * CPU's LRUs at the same time.
 *
 * This is a transparent caching front-end to sb_bread(), sb_getblk() and
 * sb_find_get_block().
 *
 * The LRUs themselves only need locking against invalidate_bh_lrus.  We use
 * a local interrupt disable for that.
 */

1238
#define BH_LRU_SIZE	16
Linus Torvalds's avatar
Linus Torvalds committed
1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269

struct bh_lru {
	struct buffer_head *bhs[BH_LRU_SIZE];
};

static DEFINE_PER_CPU(struct bh_lru, bh_lrus) = {{ NULL }};

#ifdef CONFIG_SMP
#define bh_lru_lock()	local_irq_disable()
#define bh_lru_unlock()	local_irq_enable()
#else
#define bh_lru_lock()	preempt_disable()
#define bh_lru_unlock()	preempt_enable()
#endif

static inline void check_irqs_on(void)
{
#ifdef irqs_disabled
	BUG_ON(irqs_disabled());
#endif
}

/*
 * The LRU management algorithm is dopey-but-simple.  Sorry.
 */
static void bh_lru_install(struct buffer_head *bh)
{
	struct buffer_head *evictee = NULL;

	check_irqs_on();
	bh_lru_lock();
1270
	if (__this_cpu_read(bh_lrus.bhs[0]) != bh) {
Linus Torvalds's avatar
Linus Torvalds committed
1271 1272 1273 1274 1275 1276 1277
		struct buffer_head *bhs[BH_LRU_SIZE];
		int in;
		int out = 0;

		get_bh(bh);
		bhs[out++] = bh;
		for (in = 0; in < BH_LRU_SIZE; in++) {
1278 1279
			struct buffer_head *bh2 =
				__this_cpu_read(bh_lrus.bhs[in]);
Linus Torvalds's avatar
Linus Torvalds committed
1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293

			if (bh2 == bh) {
				__brelse(bh2);
			} else {
				if (out >= BH_LRU_SIZE) {
					BUG_ON(evictee != NULL);
					evictee = bh2;
				} else {
					bhs[out++] = bh2;
				}
			}
		}
		while (out < BH_LRU_SIZE)
			bhs[out++] = NULL;
1294
		memcpy(this_cpu_ptr(&bh_lrus.bhs), bhs, sizeof(bhs));
Linus Torvalds's avatar
Linus Torvalds committed
1295 1296 1297 1298 1299 1300 1301 1302 1303 1304
	}
	bh_lru_unlock();

	if (evictee)
		__brelse(evictee);
}

/*
 * Look up the bh in this cpu's LRU.  If it's there, move it to the head.
 */
1305
static struct buffer_head *
1306
lookup_bh_lru(struct block_device *bdev, sector_t block, unsigned size)
Linus Torvalds's avatar
Linus Torvalds committed
1307 1308
{
	struct buffer_head *ret = NULL;
1309
	unsigned int i;
Linus Torvalds's avatar
Linus Torvalds committed
1310 1311 1312 1313

	check_irqs_on();
	bh_lru_lock();
	for (i = 0; i < BH_LRU_SIZE; i++) {
1314
		struct buffer_head *bh = __this_cpu_read(bh_lrus.bhs[i]);
Linus Torvalds's avatar
Linus Torvalds committed
1315

1316 1317
		if (bh && bh->b_blocknr == block && bh->b_bdev == bdev &&
		    bh->b_size == size) {
Linus Torvalds's avatar
Linus Torvalds committed
1318 1319
			if (i) {
				while (i) {
1320 1321
					__this_cpu_write(bh_lrus.bhs[i],
						__this_cpu_read(bh_lrus.bhs[i - 1]));
Linus Torvalds's avatar
Linus Torvalds committed
1322 1323
					i--;
				}
1324
				__this_cpu_write(bh_lrus.bhs[0], bh);
Linus Torvalds's avatar
Linus Torvalds committed
1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340
			}
			get_bh(bh);
			ret = bh;
			break;
		}
	}
	bh_lru_unlock();
	return ret;
}

/*
 * Perform a pagecache lookup for the matching buffer.  If it's there, refresh
 * it in the LRU and mark it as accessed.  If it is not present then return
 * NULL
 */
struct buffer_head *
1341
__find_get_block(struct block_device *bdev, sector_t block, unsigned size)
Linus Torvalds's avatar
Linus Torvalds committed
1342 1343 1344 1345
{
	struct buffer_head *bh = lookup_bh_lru(bdev, block, size);

	if (bh == NULL) {
1346
		/* __find_get_block_slow will mark the page accessed */
1347
		bh = __find_get_block_slow(bdev, block);
Linus Torvalds's avatar
Linus Torvalds committed
1348 1349
		if (bh)
			bh_lru_install(bh);
1350
	} else
Linus Torvalds's avatar
Linus Torvalds committed
1351
		touch_buffer(bh);
1352

Linus Torvalds's avatar
Linus Torvalds committed
1353 1354 1355 1356 1357
	return bh;
}
EXPORT_SYMBOL(__find_get_block);

/*
1358
 * __getblk_gfp() will locate (and, if necessary, create) the buffer_head
Linus Torvalds's avatar
Linus Torvalds committed
1359 1360 1361
 * which corresponds to the passed block_device, block and size. The
 * returned buffer has its reference count incremented.
 *
1362 1363
 * __getblk_gfp() will lock up the machine if grow_dev_page's
 * try_to_free_buffers() attempt is failing.  FIXME, perhaps?
Linus Torvalds's avatar
Linus Torvalds committed
1364 1365
 */
struct buffer_head *
1366 1367
__getblk_gfp(struct block_device *bdev, sector_t block,
	     unsigned size, gfp_t gfp)
Linus Torvalds's avatar
Linus Torvalds committed
1368 1369 1370 1371 1372
{
	struct buffer_head *bh = __find_get_block(bdev, block, size);

	might_sleep();
	if (bh == NULL)
1373
		bh = __getblk_slow(bdev, block, size, gfp);
Linus Torvalds's avatar
Linus Torvalds committed
1374 1375
	return bh;
}
1376
EXPORT_SYMBOL(__getblk_gfp);
Linus Torvalds's avatar
Linus Torvalds committed
1377 1378 1379 1380

/*
 * Do async read-ahead on a buffer..
 */
1381
void __breadahead(struct block_device *bdev, sector_t block, unsigned size)
Linus Torvalds's avatar
Linus Torvalds committed
1382 1383
{
	struct buffer_head *bh = __getblk(bdev, block, size);
Andrew Morton's avatar
Andrew Morton committed
1384 1385 1386 1387
	if (likely(bh)) {
		ll_rw_block(READA, 1, &bh);
		brelse(bh);
	}
Linus Torvalds's avatar
Linus Torvalds committed
1388 1389 1390 1391
}
EXPORT_SYMBOL(__breadahead);

/**
1392
 *  __bread_gfp() - reads a specified block and returns the bh
1393
 *  @bdev: the block_device to read from
Linus Torvalds's avatar
Linus Torvalds committed
1394 1395
 *  @block: number of block
 *  @size: size (in bytes) to read
1396 1397
 *  @gfp: page allocation flag
 *
Linus Torvalds's avatar
Linus Torvalds committed
1398
 *  Reads a specified block, and returns buffer head that contains it.
1399 1400
 *  The page cache can be allocated from non-movable area
 *  not to prevent page migration if you set gfp to zero.
Linus Torvalds's avatar
Linus Torvalds committed
1401 1402 1403
 *  It returns NULL if the block was unreadable.
 */
struct buffer_head *
1404 1405
__bread_gfp(struct block_device *bdev, sector_t block,
		   unsigned size, gfp_t gfp)
Linus Torvalds's avatar
Linus Torvalds committed
1406
{
1407
	struct buffer_head *bh = __getblk_gfp(bdev, block, size, gfp);
Linus Torvalds's avatar
Linus Torvalds committed
1408

Andrew Morton's avatar
Andrew Morton committed
1409
	if (likely(bh) && !buffer_uptodate(bh))
Linus Torvalds's avatar
Linus Torvalds committed
1410 1411 1412
		bh = __bread_slow(bh);
	return bh;
}
1413
EXPORT_SYMBOL(__bread_gfp);
Linus Torvalds's avatar
Linus Torvalds committed
1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430

/*
 * invalidate_bh_lrus() is called rarely - but not only at unmount.
 * This doesn't race because it runs in each cpu either in irq
 * or with preempt disabled.
 */
static void invalidate_bh_lru(void *arg)
{
	struct bh_lru *b = &get_cpu_var(bh_lrus);
	int i;

	for (i = 0; i < BH_LRU_SIZE; i++) {
		brelse(b->bhs[i]);
		b->bhs[i] = NULL;
	}
	put_cpu_var(bh_lrus);
}
1431 1432 1433 1434 1435

static bool has_bh_in_lru(int cpu, void *dummy)
{
	struct bh_lru *b = per_cpu_ptr(&bh_lrus, cpu);
	int i;
Linus Torvalds's avatar
Linus Torvalds committed
1436
	
1437 1438 1439 1440 1441 1442 1443 1444
	for (i = 0; i < BH_LRU_SIZE; i++) {
		if (b->bhs[i])
			return 1;
	}

	return 0;
}

Peter Zijlstra's avatar
Peter Zijlstra committed
1445
void invalidate_bh_lrus(void)
Linus Torvalds's avatar
Linus Torvalds committed
1446
{
1447
	on_each_cpu_cond(has_bh_in_lru, invalidate_bh_lru, NULL, 1, GFP_KERNEL);
Linus Torvalds's avatar
Linus Torvalds committed
1448
}
Nick Piggin's avatar
Nick Piggin committed
1449
EXPORT_SYMBOL_GPL(invalidate_bh_lrus);
Linus Torvalds's avatar
Linus Torvalds committed
1450 1451 1452 1453 1454

void set_bh_page(struct buffer_head *bh,
		struct page *page, unsigned long offset)
{
	bh->b_page = page;
1455
	BUG_ON(offset >= PAGE_SIZE);
Linus Torvalds's avatar
Linus Torvalds committed
1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468
	if (PageHighMem(page))
		/*
		 * This catches illegal uses and preserves the offset:
		 */
		bh->b_data = (char *)(0 + offset);
	else
		bh->b_data = page_address(page) + offset;
}
EXPORT_SYMBOL(set_bh_page);

/*
 * Called when truncating a buffer on a page completely.
 */
1469 1470 1471 1472 1473 1474

/* Bits that are cleared during an invalidate */
#define BUFFER_FLAGS_DISCARD \
	(1 << BH_Mapped | 1 << BH_New | 1 << BH_Req | \
	 1 << BH_Delay | 1 << BH_Unwritten)

1475
static void discard_buffer(struct buffer_head * bh)
Linus Torvalds's avatar
Linus Torvalds committed
1476
{
1477 1478
	unsigned long b_state, b_state_old;

Linus Torvalds's avatar
Linus Torvalds committed
1479 1480 1481
	lock_buffer(bh);
	clear_buffer_dirty(bh);
	bh->b_bdev = NULL;
1482 1483 1484 1485 1486 1487 1488 1489
	b_state = bh->b_state;
	for (;;) {
		b_state_old = cmpxchg(&bh->b_state, b_state,
				      (b_state & ~BUFFER_FLAGS_DISCARD));
		if (b_state_old == b_state)
			break;
		b_state = b_state_old;
	}
Linus Torvalds's avatar
Linus Torvalds committed
1490 1491 1492 1493
	unlock_buffer(bh);
}

/**
1494
 * block_invalidatepage - invalidate part or all of a buffer-backed page
Linus Torvalds's avatar
Linus Torvalds committed
1495 1496
 *
 * @page: the page which is affected
1497 1498
 * @offset: start of the range to invalidate
 * @length: length of the range to invalidate
Linus Torvalds's avatar
Linus Torvalds committed
1499 1500
 *
 * block_invalidatepage() is called when all or part of the page has become
1501
 * invalidated by a truncate operation.
Linus Torvalds's avatar
Linus Torvalds committed
1502 1503 1504 1505 1506 1507 1508
 *
 * block_invalidatepage() does not have to release all buffers, but it must
 * ensure that no dirty buffer is left outside @offset and that no I/O
 * is underway against any of the blocks which are outside the truncation
 * point.  Because the caller is about to free (and possibly reuse) those
 * blocks on-disk.
 */
1509 1510
void block_invalidatepage(struct page *page, unsigned int offset,
			  unsigned int length)
Linus Torvalds's avatar
Linus Torvalds committed
1511 1512 1513
{
	struct buffer_head *head, *bh, *next;
	unsigned int curr_off = 0;
1514
	unsigned int stop = length + offset;
Linus Torvalds's avatar
Linus Torvalds committed
1515 1516 1517 1518 1519

	BUG_ON(!PageLocked(page));
	if (!page_has_buffers(page))
		goto out;

1520 1521 1522 1523 1524
	/*
	 * Check for overflow
	 */
	BUG_ON(stop > PAGE_CACHE_SIZE || stop < length);

Linus Torvalds's avatar
Linus Torvalds committed
1525 1526 1527 1528 1529 1530
	head = page_buffers(page);
	bh = head;
	do {
		unsigned int next_off = curr_off + bh->b_size;
		next = bh->b_this_page;

1531 1532 1533 1534 1535 1536
		/*
		 * Are we still fully in range ?
		 */
		if (next_off > stop)
			goto out;

Linus Torvalds's avatar
Linus Torvalds committed
1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551
		/*
		 * is this block fully invalidated?
		 */
		if (offset <= curr_off)
			discard_buffer(bh);
		curr_off = next_off;
		bh = next;
	} while (bh != head);

	/*
	 * We release buffers only if the entire page is being invalidated.
	 * The get_block cached value has been unconditionally invalidated,
	 * so real IO is not possible anymore.
	 */
	if (offset == 0)
1552
		try_to_release_page(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
1553
out:
1554
	return;
Linus Torvalds's avatar
Linus Torvalds committed
1555 1556 1557
}
EXPORT_SYMBOL(block_invalidatepage);

1558

Linus Torvalds's avatar
Linus Torvalds committed
1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615
/*
 * We attach and possibly dirty the buffers atomically wrt
 * __set_page_dirty_buffers() via private_lock.  try_to_free_buffers
 * is already excluded via the page lock.
 */
void create_empty_buffers(struct page *page,
			unsigned long blocksize, unsigned long b_state)
{
	struct buffer_head *bh, *head, *tail;

	head = alloc_page_buffers(page, blocksize, 1);
	bh = head;
	do {
		bh->b_state |= b_state;
		tail = bh;
		bh = bh->b_this_page;
	} while (bh);
	tail->b_this_page = head;

	spin_lock(&page->mapping->private_lock);
	if (PageUptodate(page) || PageDirty(page)) {
		bh = head;
		do {
			if (PageDirty(page))
				set_buffer_dirty(bh);
			if (PageUptodate(page))
				set_buffer_uptodate(bh);
			bh = bh->b_this_page;
		} while (bh != head);
	}
	attach_page_buffers(page, head);
	spin_unlock(&page->mapping->private_lock);
}
EXPORT_SYMBOL(create_empty_buffers);

/*
 * We are taking a block for data and we don't want any output from any
 * buffer-cache aliases starting from return from that function and
 * until the moment when something will explicitly mark the buffer
 * dirty (hopefully that will not happen until we will free that block ;-)
 * We don't even need to mark it not-uptodate - nobody can expect
 * anything from a newly allocated buffer anyway. We used to used
 * unmap_buffer() for such invalidation, but that was wrong. We definitely
 * don't want to mark the alias unmapped, for example - it would confuse
 * anyone who might pick it with bread() afterwards...
 *
 * Also..  Note that bforget() doesn't lock the buffer.  So there can
 * be writeout I/O going on against recently-freed buffers.  We don't
 * wait on that I/O in bforget() - it's more efficient to wait on the I/O
 * only if we really need to.  That happens here.
 */
void unmap_underlying_metadata(struct block_device *bdev, sector_t block)
{
	struct buffer_head *old_bh;

	might_sleep();

1616
	old_bh = __find_get_block_slow(bdev, block);
Linus Torvalds's avatar
Linus Torvalds committed
1617 1618 1619 1620 1621 1622 1623 1624 1625
	if (old_bh) {
		clear_buffer_dirty(old_bh);
		wait_on_buffer(old_bh);
		clear_buffer_req(old_bh);
		__brelse(old_bh);
	}
}
EXPORT_SYMBOL(unmap_underlying_metadata);

1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647
/*
 * Size is a power-of-two in the range 512..PAGE_SIZE,
 * and the case we care about most is PAGE_SIZE.
 *
 * So this *could* possibly be written with those
 * constraints in mind (relevant mostly if some
 * architecture has a slow bit-scan instruction)
 */
static inline int block_size_bits(unsigned int blocksize)
{
	return ilog2(blocksize);
}

static struct buffer_head *create_page_buffers(struct page *page, struct inode *inode, unsigned int b_state)
{
	BUG_ON(!PageLocked(page));

	if (!page_has_buffers(page))
		create_empty_buffers(page, 1 << ACCESS_ONCE(inode->i_blkbits), b_state);
	return page_buffers(page);
}

Linus Torvalds's avatar
Linus Torvalds committed
1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671
/*
 * NOTE! All mapped/uptodate combinations are valid:
 *
 *	Mapped	Uptodate	Meaning
 *
 *	No	No		"unknown" - must do get_block()
 *	No	Yes		"hole" - zero-filled
 *	Yes	No		"allocated" - allocated on disk, not read in
 *	Yes	Yes		"valid" - allocated and up-to-date in memory.
 *
 * "Dirty" is valid only with the last case (mapped+uptodate).
 */

/*
 * While block_write_full_page is writing back the dirty buffers under
 * the page lock, whoever dirtied the buffers may decide to clean them
 * again at any time.  We handle that by only looking at the buffer
 * state inside lock_buffer().
 *
 * If block_write_full_page() is called for regular writeback
 * (wbc->sync_mode == WB_SYNC_NONE) then it will redirty a page which has a
 * locked buffer.   This only can happen if someone has written the buffer
 * directly, with submit_bh().  At the address_space level PageWriteback
 * prevents this contention from occurring.
1672 1673
 *
 * If block_write_full_page() is called with wbc->sync_mode ==
Jens Axboe's avatar
Jens Axboe committed
1674 1675
 * WB_SYNC_ALL, the writes are posted using WRITE_SYNC; this
 * causes the writes to be flagged as synchronous writes.
Linus Torvalds's avatar
Linus Torvalds committed
1676 1677
 */
static int __block_write_full_page(struct inode *inode, struct page *page,
1678 1679
			get_block_t *get_block, struct writeback_control *wbc,
			bh_end_io_t *handler)
Linus Torvalds's avatar
Linus Torvalds committed
1680 1681 1682 1683
{
	int err;
	sector_t block;
	sector_t last_block;