[linux.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;
	wait_queue_head_t *wq;

	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;
	wq = ext4_ioend_wq(io->inode);
	if (atomic_dec_and_test(&EXT4_I(io->inode)->i_ioend_count) &&
	    waitqueue_active(wq))
		wake_up_all(wq);
	kmem_cache_free(io_end_cachep, io);
}

/*
 * check a range of space and convert unwritten extents to written.
 */
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;
	wait_queue_head_t *wq;
	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 (list_empty(&io->list))
		return ret;

	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 */
		wq = ext4_ioend_wq(io->inode);
		if (atomic_dec_and_test(&EXT4_I(inode)->i_aiodio_unwritten) &&
		    waitqueue_active(wq)) {
			wake_up_all(wq);
		}
	}

	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;

	mutex_lock(&inode->i_mutex);
	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);
	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;
	do {
		bio = bio_alloc(GFP_NOIO, nvecs);
		nvecs >>= 1;
	} while (bio == NULL);

	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->io_end->flag |= EXT4_IO_END_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;
f7ad6d2e	73	wait_queue_head_t *wq;
bd2d0210 TT	74
	75	BUG_ON(!io);
	76	if (io->page)
	77	put_page(io->page);
83668e71 TT	78	for (i = 0; i < io->num_io_pages; i++)
83668e71 TT	79	put_io_page(io->pages[i]);
bd2d0210	80	io->num_io_pages = 0;
e9e3bcec	81	wq = ext4_ioend_wq(io->inode);
f7ad6d2e TT	82	if (atomic_dec_and_test(&EXT4_I(io->inode)->i_ioend_count) &&
	83	waitqueue_active(wq))
	84	wake_up_all(wq);
bd2d0210 TT	85	kmem_cache_free(io_end_cachep, io);
	86	}
	87
	88	/*
	89	* check a range of space and convert unwritten extents to written.
	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;
e9e3bcec	96	wait_queue_head_t *wq;
bd2d0210 TT	97	int ret = 0;
	98
	99	ext4_debug("ext4_end_io_nolock: io 0x%p from inode %lu,list->next 0x%p,"
	100	"list->prev 0x%p\n",
	101	io, inode->i_ino, io->list.next, io->list.prev);
	102
	103	if (list_empty(&io->list))
	104	return ret;
	105
	106	if (!(io->flag & EXT4_IO_END_UNWRITTEN))
	107	return ret;
	108
	109	ret = ext4_convert_unwritten_extents(inode, offset, size);
	110	if (ret < 0) {
	111	printk(KERN_EMERG "%s: failed to convert unwritten "
	112	"extents to written extents, error is %d "
	113	"io is still on inode %lu aio dio list\n",
	114	__func__, ret, inode->i_ino);
	115	return ret;
	116	}
	117
	118	if (io->iocb)
	119	aio_complete(io->iocb, io->result, 0);
	120	/* clear the DIO AIO unwritten flag */
e9e3bcec ES	121	if (io->flag & EXT4_IO_END_UNWRITTEN) {
	122	io->flag &= ~EXT4_IO_END_UNWRITTEN;
	123	/* Wake up anyone waiting on unwritten extent conversion */
	124	wq = ext4_ioend_wq(io->inode);
	125	if (atomic_dec_and_test(&EXT4_I(inode)->i_aiodio_unwritten) &&
	126	waitqueue_active(wq)) {
	127	wake_up_all(wq);
	128	}
	129	}
	130
bd2d0210 TT	131	return ret;
	132	}
	133
	134	/*
	135	* work on completed aio dio IO, to convert unwritten extents to extents
	136	*/
	137	static void ext4_end_io_work(struct work_struct *work)
	138	{
	139	ext4_io_end_t *io = container_of(work, ext4_io_end_t, work);
	140	struct inode *inode = io->inode;
	141	struct ext4_inode_info *ei = EXT4_I(inode);
	142	unsigned long flags;
	143	int ret;
	144
	145	mutex_lock(&inode->i_mutex);
	146	ret = ext4_end_io_nolock(io);
	147	if (ret < 0) {
	148	mutex_unlock(&inode->i_mutex);
	149	return;
	150	}
	151
	152	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	153	if (!list_empty(&io->list))
	154	list_del_init(&io->list);
	155	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
	156	mutex_unlock(&inode->i_mutex);
	157	ext4_free_io_end(io);
	158	}
	159
	160	ext4_io_end_t ext4_init_io_end(struct inode inode, gfp_t flags)
	161	{
b17b35ec	162	ext4_io_end_t *io = kmem_cache_zalloc(io_end_cachep, flags);
bd2d0210	163	if (io) {
f7ad6d2e TT	164	atomic_inc(&EXT4_I(inode)->i_ioend_count);
f7ad6d2e TT	165	io->inode = inode;
bd2d0210 TT	166	INIT_WORK(&io->work, ext4_end_io_work);
	167	INIT_LIST_HEAD(&io->list);
	168	}
	169	return io;
	170	}
	171
	172	/*
	173	* Print an buffer I/O error compatible with the fs/buffer.c. This
	174	* provides compatibility with dmesg scrapers that look for a specific
	175	* buffer I/O error message. We really need a unified error reporting
	176	* structure to userspace ala Digital Unix's uerf system, but it's
	177	* probably not going to happen in my lifetime, due to LKML politics...
	178	*/
	179	static void buffer_io_error(struct buffer_head *bh)
	180	{
	181	char b[BDEVNAME_SIZE];
	182	printk(KERN_ERR "Buffer I/O error on device %s, logical block %llu\n",
	183	bdevname(bh->b_bdev, b),
	184	(unsigned long long)bh->b_blocknr);
	185	}
	186
	187	static void ext4_end_bio(struct bio *bio, int error)
	188	{
	189	ext4_io_end_t *io_end = bio->bi_private;
	190	struct workqueue_struct *wq;
	191	struct inode *inode;
	192	unsigned long flags;
bd2d0210	193	int i;
d50bdd5a	194	sector_t bi_sector = bio->bi_sector;
bd2d0210 TT	195
bd2d0210 TT	196	BUG_ON(!io_end);
bd2d0210 TT	197	bio->bi_private = NULL;
	198	bio->bi_end_io = NULL;
	199	if (test_bit(BIO_UPTODATE, &bio->bi_flags))
	200	error = 0;
bd2d0210 TT	201	bio_put(bio);
bd2d0210 TT	202
bd2d0210 TT	203	for (i = 0; i < io_end->num_io_pages; i++) {
	204	struct page *page = io_end->pages[i]->p_page;
	205	struct buffer_head bh, head;
39db00f1 CW	206	loff_t offset;
39db00f1 CW	207	loff_t io_end_offset;
bd2d0210	208
39db00f1	209	if (error) {
bd2d0210	210	SetPageError(page);
39db00f1 CW	211	set_bit(AS_EIO, &page->mapping->flags);
	212	head = page_buffers(page);
	213	BUG_ON(!head);
	214
	215	io_end_offset = io_end->offset + io_end->size;
bd2d0210 TT	216
	217	offset = (sector_t) page->index << PAGE_CACHE_SHIFT;
	218	bh = head;
	219	do {
	220	if ((offset >= io_end->offset) &&
39db00f1 CW	221	(offset+bh->b_size <= io_end_offset))
	222	buffer_io_error(bh);
	223
bd2d0210 TT	224	offset += bh->b_size;
	225	bh = bh->b_this_page;
	226	} while (bh != head);
	227	}
	228
08da1193	229	put_io_page(io_end->pages[i]);
bd2d0210	230	}
bd2d0210	231	io_end->num_io_pages = 0;
f7ad6d2e TT	232	inode = io_end->inode;
	233
	234	if (error) {
	235	io_end->flag \|= EXT4_IO_END_ERROR;
	236	ext4_warning(inode->i_sb, "I/O error writing to inode %lu "
	237	"(offset %llu size %ld starting block %llu)",
	238	inode->i_ino,
	239	(unsigned long long) io_end->offset,
	240	(long) io_end->size,
	241	(unsigned long long)
d50bdd5a	242	bi_sector >> (inode->i_blkbits - 9));
f7ad6d2e	243	}
bd2d0210	244
b6168443 TT	245	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
	246	ext4_free_io_end(io_end);
	247	return;
	248	}
	249
bd2d0210 TT	250	/* Add the io_end to per-inode completed io list*/
	251	spin_lock_irqsave(&EXT4_I(inode)->i_completed_io_lock, flags);
	252	list_add_tail(&io_end->list, &EXT4_I(inode)->i_completed_io_list);
	253	spin_unlock_irqrestore(&EXT4_I(inode)->i_completed_io_lock, flags);
	254
	255	wq = EXT4_SB(inode->i_sb)->dio_unwritten_wq;
	256	/* queue the work to convert unwritten extents to written */
	257	queue_work(wq, &io_end->work);
	258	}
	259
	260	void ext4_io_submit(struct ext4_io_submit *io)
	261	{
	262	struct bio *bio = io->io_bio;
	263
	264	if (bio) {
	265	bio_get(io->io_bio);
	266	submit_bio(io->io_op, io->io_bio);
	267	BUG_ON(bio_flagged(io->io_bio, BIO_EOPNOTSUPP));
	268	bio_put(io->io_bio);
	269	}
7dc57615	270	io->io_bio = NULL;
bd2d0210	271	io->io_op = 0;
7dc57615	272	io->io_end = NULL;
bd2d0210 TT	273	}
	274
	275	static int io_submit_init(struct ext4_io_submit *io,
	276	struct inode *inode,
	277	struct writeback_control *wbc,
	278	struct buffer_head *bh)
	279	{
	280	ext4_io_end_t *io_end;
	281	struct page *page = bh->b_page;
	282	int nvecs = bio_get_nr_vecs(bh->b_bdev);
	283	struct bio *bio;
	284
	285	io_end = ext4_init_io_end(inode, GFP_NOFS);
	286	if (!io_end)
	287	return -ENOMEM;
	288	do {
	289	bio = bio_alloc(GFP_NOIO, nvecs);
	290	nvecs >>= 1;
	291	} while (bio == NULL);
	292
	293	bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
	294	bio->bi_bdev = bh->b_bdev;
	295	bio->bi_private = io->io_end = io_end;
	296	bio->bi_end_io = ext4_end_bio;
	297
bd2d0210 TT	298	io_end->offset = (page->index << PAGE_CACHE_SHIFT) + bh_offset(bh);
	299
	300	io->io_bio = bio;
721a9602	301	io->io_op = (wbc->sync_mode == WB_SYNC_ALL ? WRITE_SYNC : WRITE);
bd2d0210 TT	302	io->io_next_block = bh->b_blocknr;
	303	return 0;
	304	}
	305
	306	static int io_submit_add_bh(struct ext4_io_submit *io,
	307	struct ext4_io_page *io_page,
	308	struct inode *inode,
	309	struct writeback_control *wbc,
	310	struct buffer_head *bh)
	311	{
	312	ext4_io_end_t *io_end;
	313	int ret;
	314
	315	if (buffer_new(bh)) {
	316	clear_buffer_new(bh);
	317	unmap_underlying_metadata(bh->b_bdev, bh->b_blocknr);
	318	}
	319
	320	if (!buffer_mapped(bh) \|\| buffer_delay(bh)) {
	321	if (!buffer_mapped(bh))
	322	clear_buffer_dirty(bh);
	323	if (io->io_bio)
	324	ext4_io_submit(io);
	325	return 0;
	326	}
	327
	328	if (io->io_bio && bh->b_blocknr != io->io_next_block) {
	329	submit_and_retry:
	330	ext4_io_submit(io);
	331	}
	332	if (io->io_bio == NULL) {
	333	ret = io_submit_init(io, inode, wbc, bh);
	334	if (ret)
	335	return ret;
	336	}
	337	io_end = io->io_end;
	338	if ((io_end->num_io_pages >= MAX_IO_PAGES) &&
	339	(io_end->pages[io_end->num_io_pages-1] != io_page))
	340	goto submit_and_retry;
	341	if (buffer_uninit(bh))
	342	io->io_end->flag \|= EXT4_IO_END_UNWRITTEN;
	343	io->io_end->size += bh->b_size;
	344	io->io_next_block++;
	345	ret = bio_add_page(io->io_bio, bh->b_page, bh->b_size, bh_offset(bh));
	346	if (ret != bh->b_size)
	347	goto submit_and_retry;
	348	if ((io_end->num_io_pages == 0) \|\|
	349	(io_end->pages[io_end->num_io_pages-1] != io_page)) {
	350	io_end->pages[io_end->num_io_pages++] = io_page;
83668e71	351	atomic_inc(&io_page->p_count);
bd2d0210 TT	352	}
	353	return 0;
	354	}
	355
	356	int ext4_bio_write_page(struct ext4_io_submit *io,
	357	struct page *page,
	358	int len,
	359	struct writeback_control *wbc)
	360	{
	361	struct inode *inode = page->mapping->host;
	362	unsigned block_start, block_end, blocksize;
	363	struct ext4_io_page *io_page;
	364	struct buffer_head bh, head;
	365	int ret = 0;
	366
	367	blocksize = 1 << inode->i_blkbits;
	368
d50bdd5a	369	BUG_ON(!PageLocked(page));
bd2d0210	370	BUG_ON(PageWriteback(page));
bd2d0210 TT	371
	372	io_page = kmem_cache_alloc(io_page_cachep, GFP_NOFS);
	373	if (!io_page) {
	374	set_page_dirty(page);
	375	unlock_page(page);
	376	return -ENOMEM;
	377	}
	378	io_page->p_page = page;
83668e71	379	atomic_set(&io_page->p_count, 1);
bd2d0210	380	get_page(page);
a54aa761 TT	381	set_page_writeback(page);
a54aa761 TT	382	ClearPageError(page);
bd2d0210 TT	383
	384	for (bh = head = page_buffers(page), block_start = 0;
	385	bh != head \|\| !block_start;
	386	block_start = block_end, bh = bh->b_this_page) {
d50bdd5a	387
bd2d0210 TT	388	block_end = block_start + blocksize;
	389	if (block_start >= len) {
	390	clear_buffer_dirty(bh);
	391	set_buffer_uptodate(bh);
	392	continue;
	393	}
d50bdd5a	394	clear_buffer_dirty(bh);
bd2d0210 TT	395	ret = io_submit_add_bh(io, io_page, inode, wbc, bh);
	396	if (ret) {
	397	/*
	398	* We only get here on ENOMEM. Not much else
	399	* we can do but mark the page as dirty, and
	400	* better luck next time.
	401	*/
	402	set_page_dirty(page);
	403	break;
	404	}
	405	}
	406	unlock_page(page);
	407	/*
	408	* If the page was truncated before we could do the writeback,
	409	* or we had a memory allocation error while trying to write
	410	* the first buffer head, we won't have submitted any pages for
	411	* I/O. In that case we need to make sure we've cleared the
	412	* PageWriteback bit from the page to prevent the system from
	413	* wedging later on.
	414	*/
83668e71	415	put_io_page(io_page);
bd2d0210 TT	416	return ret;
bd2d0210 TT	417	}