[linux-2.6-block.git] / fs / ext4 / page-io.c

/*
 * linux/fs/ext4/page-io.c
 *
 * This contains the new page_io functions for ext4
 *
 * Written by Theodore Ts'o, 2010.
 */

#include <linux/module.h>
#include <linux/fs.h>
#include <linux/time.h>
#include <linux/jbd2.h>
#include <linux/highuid.h>
#include <linux/pagemap.h>
#include <linux/quotaops.h>
#include <linux/string.h>
#include <linux/buffer_head.h>
#include <linux/writeback.h>
#include <linux/pagevec.h>
#include <linux/mpage.h>
#include <linux/namei.h>
#include <linux/uio.h>
#include <linux/bio.h>
#include <linux/workqueue.h>
#include <linux/kernel.h>
#include <linux/slab.h>

#include "ext4_jbd2.h"
#include "xattr.h"
#include "acl.h"
#include "ext4_extents.h"

static struct kmem_cache *io_page_cachep, *io_end_cachep;

int __init ext4_init_pageio(void)
{
	io_page_cachep = KMEM_CACHE(ext4_io_page, SLAB_RECLAIM_ACCOUNT);
	if (io_page_cachep == NULL)
		return -ENOMEM;
	io_end_cachep = KMEM_CACHE(ext4_io_end, SLAB_RECLAIM_ACCOUNT);
	if (io_end_cachep == NULL) {
		kmem_cache_destroy(io_page_cachep);
		return -ENOMEM;
	}
	return 0;
}

void ext4_exit_pageio(void)
{
	kmem_cache_destroy(io_end_cachep);
	kmem_cache_destroy(io_page_cachep);
}

void ext4_ioend_wait(struct inode *inode)
{
	wait_queue_head_t *wq = ext4_ioend_wq(inode);

	wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_ioend_count) == 0));
}

static void put_io_page(struct ext4_io_page *io_page)
{
	if (atomic_dec_and_test(&io_page->p_count)) {
		end_page_writeback(io_page->p_page);
		put_page(io_page->p_page);
		kmem_cache_free(io_page_cachep, io_page);
	}
}

void ext4_free_io_end(ext4_io_end_t *io)
{
	int i;

	BUG_ON(!io);
	if (io->page)
		put_page(io->page);
	for (i = 0; i < io->num_io_pages; i++)
		put_io_page(io->pages[i]);
	io->num_io_pages = 0;
	if (atomic_dec_and_test(&EXT4_I(io->inode)->i_ioend_count))
		wake_up_all(ext4_ioend_wq(io->inode));
	kmem_cache_free(io_end_cachep, io);
}

/*
 * check a range of space and convert unwritten extents to written.
 *
 * Called with inode->i_mutex; we depend on this when we manipulate
 * io->flag, since we could otherwise race with ext4_flush_completed_IO()
 */
int ext4_end_io_nolock(ext4_io_end_t *io)
{
	struct inode *inode = io->inode;
	loff_t offset = io->offset;
	ssize_t size = io->size;
	int ret = 0;

	ext4_debug("ext4_end_io_nolock: io 0x%p from inode %lu,list->next 0x%p,"
		   "list->prev 0x%p\n",
		   io, inode->i_ino, io->list.next, io->list.prev);

	if (!(io->flag & EXT4_IO_END_UNWRITTEN))
		return ret;

	ret = ext4_convert_unwritten_extents(inode, offset, size);
	if (ret < 0) {
		printk(KERN_EMERG "%s: failed to convert unwritten "
			"extents to written extents, error is %d "
			"io is still on inode %lu aio dio list\n",
		       __func__, ret, inode->i_ino);
		return ret;
	}

	if (io->iocb)
		aio_complete(io->iocb, io->result, 0);
	/* clear the DIO AIO unwritten flag */
	if (io->flag & EXT4_IO_END_UNWRITTEN) {
		io->flag &= ~EXT4_IO_END_UNWRITTEN;
		/* Wake up anyone waiting on unwritten extent conversion */
		if (atomic_dec_and_test(&EXT4_I(inode)->i_aiodio_unwritten))
			wake_up_all(ext4_ioend_wq(io->inode));
	}

	return ret;
}

/*
 * work on completed aio dio IO, to convert unwritten extents to extents
 */
static void ext4_end_io_work(struct work_struct *work)
{
	ext4_io_end_t		*io = container_of(work, ext4_io_end_t, work);
	struct inode		*inode = io->inode;
	struct ext4_inode_info	*ei = EXT4_I(inode);
	unsigned long		flags;
	int			ret;

	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	if (list_empty(&io->list)) {
		spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
		goto free;
	}
	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);

	if (!mutex_trylock(&inode->i_mutex)) {
		/*
		 * Requeue the work instead of waiting so that the work
		 * items queued after this can be processed.
		 */
		queue_work(EXT4_SB(inode->i_sb)->dio_unwritten_wq, &io->work);
		/*
		 * To prevent the ext4-dio-unwritten thread from keeping
		 * requeueing end_io requests and occupying cpu for too long,
		 * yield the cpu if it sees an end_io request that has already
		 * been requeued.
		 */
		if (io->flag & EXT4_IO_END_QUEUED)
			yield();
		io->flag |= EXT4_IO_END_QUEUED;
		return;
	}
	ret = ext4_end_io_nolock(io);
	if (ret < 0) {
		mutex_unlock(&inode->i_mutex);
		return;
	}

	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	if (!list_empty(&io->list))
		list_del_init(&io->list);
	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
	mutex_unlock(&inode->i_mutex);
free:
	ext4_free_io_end(io);
}

ext4_io_end_t *ext4_init_io_end(struct inode *inode, gfp_t flags)
{
	ext4_io_end_t *io = kmem_cache_zalloc(io_end_cachep, flags);
	if (io) {
		atomic_inc(&EXT4_I(inode)->i_ioend_count);
		io->inode = inode;
		INIT_WORK(&io->work, ext4_end_io_work);
		INIT_LIST_HEAD(&io->list);
	}
	return io;
}

/*
 * Print an buffer I/O error compatible with the fs/buffer.c.  This
 * provides compatibility with dmesg scrapers that look for a specific
 * buffer I/O error message.  We really need a unified error reporting
 * structure to userspace ala Digital Unix's uerf system, but it's
 * probably not going to happen in my lifetime, due to LKML politics...
 */
static void buffer_io_error(struct buffer_head *bh)
{
	char b[BDEVNAME_SIZE];
	printk(KERN_ERR "Buffer I/O error on device %s, logical block %llu\n",
			bdevname(bh->b_bdev, b),
			(unsigned long long)bh->b_blocknr);
}

static void ext4_end_bio(struct bio *bio, int error)
{
	ext4_io_end_t *io_end = bio->bi_private;
	struct workqueue_struct *wq;
	struct inode *inode;
	unsigned long flags;
	int i;
	sector_t bi_sector = bio->bi_sector;

	BUG_ON(!io_end);
	bio->bi_private = NULL;
	bio->bi_end_io = NULL;
	if (test_bit(BIO_UPTODATE, &bio->bi_flags))
		error = 0;
	bio_put(bio);

	for (i = 0; i < io_end->num_io_pages; i++) {
		struct page *page = io_end->pages[i]->p_page;
		struct buffer_head *bh, *head;
		loff_t offset;
		loff_t io_end_offset;

		if (error) {
			SetPageError(page);
			set_bit(AS_EIO, &page->mapping->flags);
			head = page_buffers(page);
			BUG_ON(!head);

			io_end_offset = io_end->offset + io_end->size;

			offset = (sector_t) page->index << PAGE_CACHE_SHIFT;
			bh = head;
			do {
				if ((offset >= io_end->offset) &&
				    (offset+bh->b_size <= io_end_offset))
					buffer_io_error(bh);

				offset += bh->b_size;
				bh = bh->b_this_page;
			} while (bh != head);
		}

		put_io_page(io_end->pages[i]);
	}
	io_end->num_io_pages = 0;
	inode = io_end->inode;

	if (error) {
		io_end->flag |= EXT4_IO_END_ERROR;
		ext4_warning(inode->i_sb, "I/O error writing to inode %lu "
			     "(offset %llu size %ld starting block %llu)",
			     inode->i_ino,
			     (unsigned long long) io_end->offset,
			     (long) io_end->size,
			     (unsigned long long)
			     bi_sector >> (inode->i_blkbits - 9));
	}

	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
		ext4_free_io_end(io_end);
		return;
	}

	/* Add the io_end to per-inode completed io list*/
	spin_lock_irqsave(&EXT4_I(inode)->i_completed_io_lock, flags);
	list_add_tail(&io_end->list, &EXT4_I(inode)->i_completed_io_list);
	spin_unlock_irqrestore(&EXT4_I(inode)->i_completed_io_lock, flags);

	wq = EXT4_SB(inode->i_sb)->dio_unwritten_wq;
	/* queue the work to convert unwritten extents to written */
	queue_work(wq, &io_end->work);
}

void ext4_io_submit(struct ext4_io_submit *io)
{
	struct bio *bio = io->io_bio;

	if (bio) {
		bio_get(io->io_bio);
		submit_bio(io->io_op, io->io_bio);
		BUG_ON(bio_flagged(io->io_bio, BIO_EOPNOTSUPP));
		bio_put(io->io_bio);
	}
	io->io_bio = NULL;
	io->io_op = 0;
	io->io_end = NULL;
}

static int io_submit_init(struct ext4_io_submit *io,
			  struct inode *inode,
			  struct writeback_control *wbc,
			  struct buffer_head *bh)
{
	ext4_io_end_t *io_end;
	struct page *page = bh->b_page;
	int nvecs = bio_get_nr_vecs(bh->b_bdev);
	struct bio *bio;

	io_end = ext4_init_io_end(inode, GFP_NOFS);
	if (!io_end)
		return -ENOMEM;
	bio = bio_alloc(GFP_NOIO, min(nvecs, BIO_MAX_PAGES));
	bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
	bio->bi_bdev = bh->b_bdev;
	bio->bi_private = io->io_end = io_end;
	bio->bi_end_io = ext4_end_bio;

	io_end->offset = (page->index << PAGE_CACHE_SHIFT) + bh_offset(bh);

	io->io_bio = bio;
	io->io_op = (wbc->sync_mode == WB_SYNC_ALL ?  WRITE_SYNC : WRITE);
	io->io_next_block = bh->b_blocknr;
	return 0;
}

static int io_submit_add_bh(struct ext4_io_submit *io,
			    struct ext4_io_page *io_page,
			    struct inode *inode,
			    struct writeback_control *wbc,
			    struct buffer_head *bh)
{
	ext4_io_end_t *io_end;
	int ret;

	if (buffer_new(bh)) {
		clear_buffer_new(bh);
		unmap_underlying_metadata(bh->b_bdev, bh->b_blocknr);
	}

	if (!buffer_mapped(bh) || buffer_delay(bh)) {
		if (!buffer_mapped(bh))
			clear_buffer_dirty(bh);
		if (io->io_bio)
			ext4_io_submit(io);
		return 0;
	}

	if (io->io_bio && bh->b_blocknr != io->io_next_block) {
submit_and_retry:
		ext4_io_submit(io);
	}
	if (io->io_bio == NULL) {
		ret = io_submit_init(io, inode, wbc, bh);
		if (ret)
			return ret;
	}
	io_end = io->io_end;
	if ((io_end->num_io_pages >= MAX_IO_PAGES) &&
	    (io_end->pages[io_end->num_io_pages-1] != io_page))
		goto submit_and_retry;
	if (buffer_uninit(bh) && !(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
		io_end->flag |= EXT4_IO_END_UNWRITTEN;
		atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
	}
	io->io_end->size += bh->b_size;
	io->io_next_block++;
	ret = bio_add_page(io->io_bio, bh->b_page, bh->b_size, bh_offset(bh));
	if (ret != bh->b_size)
		goto submit_and_retry;
	if ((io_end->num_io_pages == 0) ||
	    (io_end->pages[io_end->num_io_pages-1] != io_page)) {
		io_end->pages[io_end->num_io_pages++] = io_page;
		atomic_inc(&io_page->p_count);
	}
	return 0;
}

int ext4_bio_write_page(struct ext4_io_submit *io,
			struct page *page,
			int len,
			struct writeback_control *wbc)
{
	struct inode *inode = page->mapping->host;
	unsigned block_start, block_end, blocksize;
	struct ext4_io_page *io_page;
	struct buffer_head *bh, *head;
	int ret = 0;

	blocksize = 1 << inode->i_blkbits;

	BUG_ON(!PageLocked(page));
	BUG_ON(PageWriteback(page));

	io_page = kmem_cache_alloc(io_page_cachep, GFP_NOFS);
	if (!io_page) {
		set_page_dirty(page);
		unlock_page(page);
		return -ENOMEM;
	}
	io_page->p_page = page;
	atomic_set(&io_page->p_count, 1);
	get_page(page);
	set_page_writeback(page);
	ClearPageError(page);

	for (bh = head = page_buffers(page), block_start = 0;
	     bh != head || !block_start;
	     block_start = block_end, bh = bh->b_this_page) {

		block_end = block_start + blocksize;
		if (block_start >= len) {
			clear_buffer_dirty(bh);
			set_buffer_uptodate(bh);
			continue;
		}
		clear_buffer_dirty(bh);
		ret = io_submit_add_bh(io, io_page, inode, wbc, bh);
		if (ret) {
			/*
			 * We only get here on ENOMEM.  Not much else
			 * we can do but mark the page as dirty, and
			 * better luck next time.
			 */
			set_page_dirty(page);
			break;
		}
	}
	unlock_page(page);
	/*
	 * If the page was truncated before we could do the writeback,
	 * or we had a memory allocation error while trying to write
	 * the first buffer head, we won't have submitted any pages for
	 * I/O.  In that case we need to make sure we've cleared the
	 * PageWriteback bit from the page to prevent the system from
	 * wedging later on.
	 */
	put_io_page(io_page);
	return ret;
}
Commit	Line	Data
bd2d0210 TT	1	/*
	2	* linux/fs/ext4/page-io.c
	3	*
	4	* This contains the new page_io functions for ext4
	5	*
	6	* Written by Theodore Ts'o, 2010.
	7	*/
	8
	9	#include <linux/module.h>
	10	#include <linux/fs.h>
	11	#include <linux/time.h>
	12	#include <linux/jbd2.h>
	13	#include <linux/highuid.h>
	14	#include <linux/pagemap.h>
	15	#include <linux/quotaops.h>
	16	#include <linux/string.h>
	17	#include <linux/buffer_head.h>
	18	#include <linux/writeback.h>
	19	#include <linux/pagevec.h>
	20	#include <linux/mpage.h>
	21	#include <linux/namei.h>
	22	#include <linux/uio.h>
	23	#include <linux/bio.h>
	24	#include <linux/workqueue.h>
	25	#include <linux/kernel.h>
	26	#include <linux/slab.h>
	27
	28	#include "ext4_jbd2.h"
	29	#include "xattr.h"
	30	#include "acl.h"
	31	#include "ext4_extents.h"
	32
	33	static struct kmem_cache io_page_cachep, io_end_cachep;
	34
5dabfc78	35	int __init ext4_init_pageio(void)
bd2d0210 TT	36	{
	37	io_page_cachep = KMEM_CACHE(ext4_io_page, SLAB_RECLAIM_ACCOUNT);
	38	if (io_page_cachep == NULL)
	39	return -ENOMEM;
	40	io_end_cachep = KMEM_CACHE(ext4_io_end, SLAB_RECLAIM_ACCOUNT);
13195184	41	if (io_end_cachep == NULL) {
bd2d0210 TT	42	kmem_cache_destroy(io_page_cachep);
	43	return -ENOMEM;
	44	}
bd2d0210 TT	45	return 0;
	46	}
	47
5dabfc78	48	void ext4_exit_pageio(void)
bd2d0210 TT	49	{
	50	kmem_cache_destroy(io_end_cachep);
	51	kmem_cache_destroy(io_page_cachep);
	52	}
	53
f7ad6d2e TT	54	void ext4_ioend_wait(struct inode *inode)
f7ad6d2e TT	55	{
e9e3bcec	56	wait_queue_head_t *wq = ext4_ioend_wq(inode);
f7ad6d2e TT	57
	58	wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_ioend_count) == 0));
	59	}
	60
83668e71 TT	61	static void put_io_page(struct ext4_io_page *io_page)
	62	{
	63	if (atomic_dec_and_test(&io_page->p_count)) {
	64	end_page_writeback(io_page->p_page);
	65	put_page(io_page->p_page);
	66	kmem_cache_free(io_page_cachep, io_page);
	67	}
	68	}
	69
bd2d0210 TT	70	void ext4_free_io_end(ext4_io_end_t *io)
	71	{
	72	int i;
	73
	74	BUG_ON(!io);
	75	if (io->page)
	76	put_page(io->page);
83668e71 TT	77	for (i = 0; i < io->num_io_pages; i++)
83668e71 TT	78	put_io_page(io->pages[i]);
bd2d0210	79	io->num_io_pages = 0;
4e298021 TT	80	if (atomic_dec_and_test(&EXT4_I(io->inode)->i_ioend_count))
4e298021 TT	81	wake_up_all(ext4_ioend_wq(io->inode));
bd2d0210 TT	82	kmem_cache_free(io_end_cachep, io);
	83	}
	84
	85	/*
	86	* check a range of space and convert unwritten extents to written.
d73d5046 TM	87	*
	88	* Called with inode->i_mutex; we depend on this when we manipulate
	89	* io->flag, since we could otherwise race with ext4_flush_completed_IO()
bd2d0210 TT	90	*/
	91	int ext4_end_io_nolock(ext4_io_end_t *io)
	92	{
	93	struct inode *inode = io->inode;
	94	loff_t offset = io->offset;
	95	ssize_t size = io->size;
	96	int ret = 0;
	97
	98	ext4_debug("ext4_end_io_nolock: io 0x%p from inode %lu,list->next 0x%p,"
	99	"list->prev 0x%p\n",
	100	io, inode->i_ino, io->list.next, io->list.prev);
	101
bd2d0210 TT	102	if (!(io->flag & EXT4_IO_END_UNWRITTEN))
	103	return ret;
	104
	105	ret = ext4_convert_unwritten_extents(inode, offset, size);
	106	if (ret < 0) {
	107	printk(KERN_EMERG "%s: failed to convert unwritten "
	108	"extents to written extents, error is %d "
	109	"io is still on inode %lu aio dio list\n",
	110	__func__, ret, inode->i_ino);
	111	return ret;
	112	}
	113
	114	if (io->iocb)
	115	aio_complete(io->iocb, io->result, 0);
	116	/* clear the DIO AIO unwritten flag */
e9e3bcec ES	117	if (io->flag & EXT4_IO_END_UNWRITTEN) {
	118	io->flag &= ~EXT4_IO_END_UNWRITTEN;
	119	/* Wake up anyone waiting on unwritten extent conversion */
4e298021 TT	120	if (atomic_dec_and_test(&EXT4_I(inode)->i_aiodio_unwritten))
4e298021 TT	121	wake_up_all(ext4_ioend_wq(io->inode));
e9e3bcec ES	122	}
e9e3bcec ES	123
bd2d0210 TT	124	return ret;
	125	}
	126
	127	/*
	128	* work on completed aio dio IO, to convert unwritten extents to extents
	129	*/
	130	static void ext4_end_io_work(struct work_struct *work)
	131	{
	132	ext4_io_end_t *io = container_of(work, ext4_io_end_t, work);
	133	struct inode *inode = io->inode;
	134	struct ext4_inode_info *ei = EXT4_I(inode);
	135	unsigned long flags;
	136	int ret;
	137
d73d5046 TM	138	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	139	if (list_empty(&io->list)) {
	140	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
	141	goto free;
	142	}
	143	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
	144
8c0bec21 JZ	145	if (!mutex_trylock(&inode->i_mutex)) {
	146	/*
	147	* Requeue the work instead of waiting so that the work
	148	* items queued after this can be processed.
	149	*/
	150	queue_work(EXT4_SB(inode->i_sb)->dio_unwritten_wq, &io->work);
	151	/*
	152	* To prevent the ext4-dio-unwritten thread from keeping
	153	* requeueing end_io requests and occupying cpu for too long,
	154	* yield the cpu if it sees an end_io request that has already
	155	* been requeued.
	156	*/
	157	if (io->flag & EXT4_IO_END_QUEUED)
	158	yield();
	159	io->flag \|= EXT4_IO_END_QUEUED;
	160	return;
	161	}
bd2d0210 TT	162	ret = ext4_end_io_nolock(io);
	163	if (ret < 0) {
	164	mutex_unlock(&inode->i_mutex);
	165	return;
	166	}
	167
	168	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	169	if (!list_empty(&io->list))
	170	list_del_init(&io->list);
	171	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
	172	mutex_unlock(&inode->i_mutex);
d73d5046	173	free:
bd2d0210 TT	174	ext4_free_io_end(io);
	175	}
	176
	177	ext4_io_end_t ext4_init_io_end(struct inode inode, gfp_t flags)
	178	{
b17b35ec	179	ext4_io_end_t *io = kmem_cache_zalloc(io_end_cachep, flags);
bd2d0210	180	if (io) {
f7ad6d2e TT	181	atomic_inc(&EXT4_I(inode)->i_ioend_count);
f7ad6d2e TT	182	io->inode = inode;
bd2d0210 TT	183	INIT_WORK(&io->work, ext4_end_io_work);
	184	INIT_LIST_HEAD(&io->list);
	185	}
	186	return io;
	187	}
	188
	189	/*
	190	* Print an buffer I/O error compatible with the fs/buffer.c. This
	191	* provides compatibility with dmesg scrapers that look for a specific
	192	* buffer I/O error message. We really need a unified error reporting
	193	* structure to userspace ala Digital Unix's uerf system, but it's
	194	* probably not going to happen in my lifetime, due to LKML politics...
	195	*/
	196	static void buffer_io_error(struct buffer_head *bh)
	197	{
	198	char b[BDEVNAME_SIZE];
	199	printk(KERN_ERR "Buffer I/O error on device %s, logical block %llu\n",
	200	bdevname(bh->b_bdev, b),
	201	(unsigned long long)bh->b_blocknr);
	202	}
	203
	204	static void ext4_end_bio(struct bio *bio, int error)
	205	{
	206	ext4_io_end_t *io_end = bio->bi_private;
	207	struct workqueue_struct *wq;
	208	struct inode *inode;
	209	unsigned long flags;
bd2d0210	210	int i;
d50bdd5a	211	sector_t bi_sector = bio->bi_sector;
bd2d0210 TT	212
bd2d0210 TT	213	BUG_ON(!io_end);
bd2d0210 TT	214	bio->bi_private = NULL;
	215	bio->bi_end_io = NULL;
	216	if (test_bit(BIO_UPTODATE, &bio->bi_flags))
	217	error = 0;
bd2d0210 TT	218	bio_put(bio);
bd2d0210 TT	219
bd2d0210 TT	220	for (i = 0; i < io_end->num_io_pages; i++) {
	221	struct page *page = io_end->pages[i]->p_page;
	222	struct buffer_head bh, head;
39db00f1 CW	223	loff_t offset;
39db00f1 CW	224	loff_t io_end_offset;
bd2d0210	225
39db00f1	226	if (error) {
bd2d0210	227	SetPageError(page);
39db00f1 CW	228	set_bit(AS_EIO, &page->mapping->flags);
	229	head = page_buffers(page);
	230	BUG_ON(!head);
	231
	232	io_end_offset = io_end->offset + io_end->size;
bd2d0210 TT	233
	234	offset = (sector_t) page->index << PAGE_CACHE_SHIFT;
	235	bh = head;
	236	do {
	237	if ((offset >= io_end->offset) &&
39db00f1 CW	238	(offset+bh->b_size <= io_end_offset))
	239	buffer_io_error(bh);
	240
bd2d0210 TT	241	offset += bh->b_size;
	242	bh = bh->b_this_page;
	243	} while (bh != head);
	244	}
	245
08da1193	246	put_io_page(io_end->pages[i]);
bd2d0210	247	}
bd2d0210	248	io_end->num_io_pages = 0;
f7ad6d2e TT	249	inode = io_end->inode;
	250
	251	if (error) {
	252	io_end->flag \|= EXT4_IO_END_ERROR;
	253	ext4_warning(inode->i_sb, "I/O error writing to inode %lu "
	254	"(offset %llu size %ld starting block %llu)",
	255	inode->i_ino,
	256	(unsigned long long) io_end->offset,
	257	(long) io_end->size,
	258	(unsigned long long)
d50bdd5a	259	bi_sector >> (inode->i_blkbits - 9));
f7ad6d2e	260	}
bd2d0210	261
b6168443 TT	262	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
	263	ext4_free_io_end(io_end);
	264	return;
	265	}
	266
bd2d0210 TT	267	/* Add the io_end to per-inode completed io list*/
	268	spin_lock_irqsave(&EXT4_I(inode)->i_completed_io_lock, flags);
	269	list_add_tail(&io_end->list, &EXT4_I(inode)->i_completed_io_list);
	270	spin_unlock_irqrestore(&EXT4_I(inode)->i_completed_io_lock, flags);
	271
	272	wq = EXT4_SB(inode->i_sb)->dio_unwritten_wq;
	273	/* queue the work to convert unwritten extents to written */
	274	queue_work(wq, &io_end->work);
	275	}
	276
	277	void ext4_io_submit(struct ext4_io_submit *io)
	278	{
	279	struct bio *bio = io->io_bio;
	280
	281	if (bio) {
	282	bio_get(io->io_bio);
	283	submit_bio(io->io_op, io->io_bio);
	284	BUG_ON(bio_flagged(io->io_bio, BIO_EOPNOTSUPP));
	285	bio_put(io->io_bio);
	286	}
7dc57615	287	io->io_bio = NULL;
bd2d0210	288	io->io_op = 0;
7dc57615	289	io->io_end = NULL;
bd2d0210 TT	290	}
	291
	292	static int io_submit_init(struct ext4_io_submit *io,
	293	struct inode *inode,
	294	struct writeback_control *wbc,
	295	struct buffer_head *bh)
	296	{
	297	ext4_io_end_t *io_end;
	298	struct page *page = bh->b_page;
	299	int nvecs = bio_get_nr_vecs(bh->b_bdev);
	300	struct bio *bio;
	301
	302	io_end = ext4_init_io_end(inode, GFP_NOFS);
	303	if (!io_end)
	304	return -ENOMEM;
275d3ba6	305	bio = bio_alloc(GFP_NOIO, min(nvecs, BIO_MAX_PAGES));
bd2d0210 TT	306	bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
	307	bio->bi_bdev = bh->b_bdev;
	308	bio->bi_private = io->io_end = io_end;
	309	bio->bi_end_io = ext4_end_bio;
	310
bd2d0210 TT	311	io_end->offset = (page->index << PAGE_CACHE_SHIFT) + bh_offset(bh);
	312
	313	io->io_bio = bio;
721a9602	314	io->io_op = (wbc->sync_mode == WB_SYNC_ALL ? WRITE_SYNC : WRITE);
bd2d0210 TT	315	io->io_next_block = bh->b_blocknr;
	316	return 0;
	317	}
	318
	319	static int io_submit_add_bh(struct ext4_io_submit *io,
	320	struct ext4_io_page *io_page,
	321	struct inode *inode,
	322	struct writeback_control *wbc,
	323	struct buffer_head *bh)
	324	{
	325	ext4_io_end_t *io_end;
	326	int ret;
	327
	328	if (buffer_new(bh)) {
	329	clear_buffer_new(bh);
	330	unmap_underlying_metadata(bh->b_bdev, bh->b_blocknr);
	331	}
	332
	333	if (!buffer_mapped(bh) \|\| buffer_delay(bh)) {
	334	if (!buffer_mapped(bh))
	335	clear_buffer_dirty(bh);
	336	if (io->io_bio)
	337	ext4_io_submit(io);
	338	return 0;
	339	}
	340
	341	if (io->io_bio && bh->b_blocknr != io->io_next_block) {
	342	submit_and_retry:
	343	ext4_io_submit(io);
	344	}
	345	if (io->io_bio == NULL) {
	346	ret = io_submit_init(io, inode, wbc, bh);
	347	if (ret)
	348	return ret;
	349	}
	350	io_end = io->io_end;
	351	if ((io_end->num_io_pages >= MAX_IO_PAGES) &&
	352	(io_end->pages[io_end->num_io_pages-1] != io_page))
	353	goto submit_and_retry;
32c80b32 TM	354	if (buffer_uninit(bh) && !(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
	355	io_end->flag \|= EXT4_IO_END_UNWRITTEN;
	356	atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
	357	}
bd2d0210 TT	358	io->io_end->size += bh->b_size;
	359	io->io_next_block++;
	360	ret = bio_add_page(io->io_bio, bh->b_page, bh->b_size, bh_offset(bh));
	361	if (ret != bh->b_size)
	362	goto submit_and_retry;
	363	if ((io_end->num_io_pages == 0) \|\|
	364	(io_end->pages[io_end->num_io_pages-1] != io_page)) {
	365	io_end->pages[io_end->num_io_pages++] = io_page;
83668e71	366	atomic_inc(&io_page->p_count);
bd2d0210 TT	367	}
	368	return 0;
	369	}
	370
	371	int ext4_bio_write_page(struct ext4_io_submit *io,
	372	struct page *page,
	373	int len,
	374	struct writeback_control *wbc)
	375	{
	376	struct inode *inode = page->mapping->host;
	377	unsigned block_start, block_end, blocksize;
	378	struct ext4_io_page *io_page;
	379	struct buffer_head bh, head;
	380	int ret = 0;
	381
	382	blocksize = 1 << inode->i_blkbits;
	383
d50bdd5a	384	BUG_ON(!PageLocked(page));
bd2d0210	385	BUG_ON(PageWriteback(page));
bd2d0210 TT	386
	387	io_page = kmem_cache_alloc(io_page_cachep, GFP_NOFS);
	388	if (!io_page) {
	389	set_page_dirty(page);
	390	unlock_page(page);
	391	return -ENOMEM;
	392	}
	393	io_page->p_page = page;
83668e71	394	atomic_set(&io_page->p_count, 1);
bd2d0210	395	get_page(page);
a54aa761 TT	396	set_page_writeback(page);
a54aa761 TT	397	ClearPageError(page);
bd2d0210 TT	398
	399	for (bh = head = page_buffers(page), block_start = 0;
	400	bh != head \|\| !block_start;
	401	block_start = block_end, bh = bh->b_this_page) {
d50bdd5a	402
bd2d0210 TT	403	block_end = block_start + blocksize;
	404	if (block_start >= len) {
	405	clear_buffer_dirty(bh);
	406	set_buffer_uptodate(bh);
	407	continue;
	408	}
d50bdd5a	409	clear_buffer_dirty(bh);
bd2d0210 TT	410	ret = io_submit_add_bh(io, io_page, inode, wbc, bh);
	411	if (ret) {
	412	/*
	413	* We only get here on ENOMEM. Not much else
	414	* we can do but mark the page as dirty, and
	415	* better luck next time.
	416	*/
	417	set_page_dirty(page);
	418	break;
	419	}
	420	}
	421	unlock_page(page);
	422	/*
	423	* If the page was truncated before we could do the writeback,
	424	* or we had a memory allocation error while trying to write
	425	* the first buffer head, we won't have submitted any pages for
	426	* I/O. In that case we need to make sure we've cleared the
	427	* PageWriteback bit from the page to prevent the system from
	428	* wedging later on.
	429	*/
83668e71	430	put_io_page(io_page);
bd2d0210 TT	431	return ret;
bd2d0210 TT	432	}