[mirror_ubuntu-bionic-kernel.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/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);
	BUG_ON(io->flag & EXT4_IO_END_UNWRITTEN);

	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;

	BUG_ON(!(io->flag & EXT4_IO_END_UNWRITTEN));

	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);

	ret = ext4_convert_unwritten_extents(inode, offset, size);
	if (ret < 0) {
		ext4_msg(inode->i_sb, KERN_EMERG,
			 "failed to convert unwritten extents to written "
			 "extents -- potential data loss!  "
			 "(inode %lu, offset %llu, size %zd, error %d)",
			 inode->i_ino, offset, size, ret);
	}
	io->flag &= ~EXT4_IO_END_UNWRITTEN;
	if (io->iocb)
		aio_complete(io->iocb, io->result, 0);

	if (io->flag & EXT4_IO_END_DIRECT)
		inode_dio_done(inode);
	/* Wake up anyone waiting on unwritten extent conversion */
	if (atomic_dec_and_test(&EXT4_I(inode)->i_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;

	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	if (io->flag & EXT4_IO_END_IN_FSYNC)
		goto requeue;
	if (list_empty(&io->list)) {
		spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
		goto free;
	}

	if (!mutex_trylock(&inode->i_mutex)) {
		bool was_queued;
requeue:
		was_queued = !!(io->flag & EXT4_IO_END_QUEUED);
		io->flag |= EXT4_IO_END_QUEUED;
		spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
		/*
		 * 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 (was_queued)
			yield();
		return;
	}
	list_del_init(&io->list);
	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
	(void) ext4_end_io_nolock(io);
	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))
		ext4_set_io_unwritten_flag(inode, io_end);
	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) {
			/*
			 * Comments copied from block_write_full_page_endio:
			 *
			 * The page straddles i_size.  It must be zeroed out on
			 * each and every writepage invocation because it may
			 * be mmapped.  "A file is mapped in multiples of the
			 * page size.  For a file that is not a multiple of
			 * the  page size, the remaining memory is zeroed when
			 * mapped, and writes to that region are not written
			 * out to the file."
			 */
			zero_user_segment(page, block_start, block_end);
			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
bd2d0210 TT	9	#include <linux/fs.h>
	10	#include <linux/time.h>
	11	#include <linux/jbd2.h>
	12	#include <linux/highuid.h>
	13	#include <linux/pagemap.h>
	14	#include <linux/quotaops.h>
	15	#include <linux/string.h>
	16	#include <linux/buffer_head.h>
	17	#include <linux/writeback.h>
	18	#include <linux/pagevec.h>
	19	#include <linux/mpage.h>
	20	#include <linux/namei.h>
	21	#include <linux/uio.h>
	22	#include <linux/bio.h>
	23	#include <linux/workqueue.h>
	24	#include <linux/kernel.h>
	25	#include <linux/slab.h>
	26
	27	#include "ext4_jbd2.h"
	28	#include "xattr.h"
	29	#include "acl.h"
	30	#include "ext4_extents.h"
	31
	32	static struct kmem_cache io_page_cachep, io_end_cachep;
	33
5dabfc78	34	int __init ext4_init_pageio(void)
bd2d0210 TT	35	{
	36	io_page_cachep = KMEM_CACHE(ext4_io_page, SLAB_RECLAIM_ACCOUNT);
	37	if (io_page_cachep == NULL)
	38	return -ENOMEM;
	39	io_end_cachep = KMEM_CACHE(ext4_io_end, SLAB_RECLAIM_ACCOUNT);
13195184	40	if (io_end_cachep == NULL) {
bd2d0210 TT	41	kmem_cache_destroy(io_page_cachep);
	42	return -ENOMEM;
	43	}
bd2d0210 TT	44	return 0;
	45	}
	46
5dabfc78	47	void ext4_exit_pageio(void)
bd2d0210 TT	48	{
	49	kmem_cache_destroy(io_end_cachep);
	50	kmem_cache_destroy(io_page_cachep);
	51	}
	52
f7ad6d2e TT	53	void ext4_ioend_wait(struct inode *inode)
f7ad6d2e TT	54	{
e9e3bcec	55	wait_queue_head_t *wq = ext4_ioend_wq(inode);
f7ad6d2e TT	56
	57	wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_ioend_count) == 0));
	58	}
	59
83668e71 TT	60	static void put_io_page(struct ext4_io_page *io_page)
	61	{
	62	if (atomic_dec_and_test(&io_page->p_count)) {
6268b325	63	end_page_writeback(io_page->p_page);
83668e71 TT	64	put_page(io_page->p_page);
	65	kmem_cache_free(io_page_cachep, io_page);
	66	}
	67	}
	68
bd2d0210 TT	69	void ext4_free_io_end(ext4_io_end_t *io)
	70	{
	71	int i;
	72
	73	BUG_ON(!io);
82e54229 DM	74	BUG_ON(io->flag & EXT4_IO_END_UNWRITTEN);
82e54229 DM	75
bd2d0210 TT	76	if (io->page)
bd2d0210 TT	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;
4e298021 TT	81	if (atomic_dec_and_test(&EXT4_I(io->inode)->i_ioend_count))
4e298021 TT	82	wake_up_all(ext4_ioend_wq(io->inode));
bd2d0210 TT	83	kmem_cache_free(io_end_cachep, io);
	84	}
	85
	86	/*
	87	* check a range of space and convert unwritten extents to written.
d73d5046 TM	88	*
	89	* Called with inode->i_mutex; we depend on this when we manipulate
	90	* io->flag, since we could otherwise race with ext4_flush_completed_IO()
bd2d0210 TT	91	*/
	92	int ext4_end_io_nolock(ext4_io_end_t *io)
	93	{
	94	struct inode *inode = io->inode;
	95	loff_t offset = io->offset;
	96	ssize_t size = io->size;
	97	int ret = 0;
	98
82e54229 DM	99	BUG_ON(!(io->flag & EXT4_IO_END_UNWRITTEN));
82e54229 DM	100
bd2d0210 TT	101	ext4_debug("ext4_end_io_nolock: io 0x%p from inode %lu,list->next 0x%p,"
	102	"list->prev 0x%p\n",
	103	io, inode->i_ino, io->list.next, io->list.prev);
	104
bd2d0210 TT	105	ret = ext4_convert_unwritten_extents(inode, offset, size);
bd2d0210 TT	106	if (ret < 0) {
b82e384c TT	107	ext4_msg(inode->i_sb, KERN_EMERG,
	108	"failed to convert unwritten extents to written "
	109	"extents -- potential data loss! "
	110	"(inode %lu, offset %llu, size %zd, error %d)",
	111	inode->i_ino, offset, size, ret);
bd2d0210	112	}
82e54229	113	io->flag &= ~EXT4_IO_END_UNWRITTEN;
bd2d0210 TT	114	if (io->iocb)
bd2d0210 TT	115	aio_complete(io->iocb, io->result, 0);
e9e3bcec	116
266991b1 JM	117	if (io->flag & EXT4_IO_END_DIRECT)
266991b1 JM	118	inode_dio_done(inode);
b82e384c	119	/* Wake up anyone waiting on unwritten extent conversion */
e27f41e1	120	if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
b82e384c	121	wake_up_all(ext4_ioend_wq(io->inode));
bd2d0210 TT	122	return ret;
	123	}
	124
	125	/*
	126	* work on completed aio dio IO, to convert unwritten extents to extents
	127	*/
	128	static void ext4_end_io_work(struct work_struct *work)
	129	{
	130	ext4_io_end_t *io = container_of(work, ext4_io_end_t, work);
	131	struct inode *inode = io->inode;
	132	struct ext4_inode_info *ei = EXT4_I(inode);
	133	unsigned long flags;
bd2d0210	134
d73d5046	135	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
491caa43 JM	136	if (io->flag & EXT4_IO_END_IN_FSYNC)
491caa43 JM	137	goto requeue;
d73d5046 TM	138	if (list_empty(&io->list)) {
	139	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
	140	goto free;
	141	}
d73d5046	142
8c0bec21	143	if (!mutex_trylock(&inode->i_mutex)) {
491caa43 JM	144	bool was_queued;
	145	requeue:
	146	was_queued = !!(io->flag & EXT4_IO_END_QUEUED);
	147	io->flag \|= EXT4_IO_END_QUEUED;
b82e384c	148	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
8c0bec21 JZ	149	/*
	150	* Requeue the work instead of waiting so that the work
	151	* items queued after this can be processed.
	152	*/
	153	queue_work(EXT4_SB(inode->i_sb)->dio_unwritten_wq, &io->work);
	154	/*
	155	* To prevent the ext4-dio-unwritten thread from keeping
	156	* requeueing end_io requests and occupying cpu for too long,
	157	* yield the cpu if it sees an end_io request that has already
	158	* been requeued.
	159	*/
491caa43	160	if (was_queued)
8c0bec21	161	yield();
8c0bec21 JZ	162	return;
8c0bec21 JZ	163	}
b82e384c	164	list_del_init(&io->list);
bd2d0210	165	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
b82e384c	166	(void) ext4_end_io_nolock(io);
bd2d0210	167	mutex_unlock(&inode->i_mutex);
d73d5046	168	free:
bd2d0210 TT	169	ext4_free_io_end(io);
	170	}
	171
	172	ext4_io_end_t ext4_init_io_end(struct inode inode, gfp_t flags)
	173	{
b17b35ec	174	ext4_io_end_t *io = kmem_cache_zalloc(io_end_cachep, flags);
bd2d0210	175	if (io) {
f7ad6d2e TT	176	atomic_inc(&EXT4_I(inode)->i_ioend_count);
f7ad6d2e TT	177	io->inode = inode;
bd2d0210 TT	178	INIT_WORK(&io->work, ext4_end_io_work);
	179	INIT_LIST_HEAD(&io->list);
	180	}
	181	return io;
	182	}
	183
	184	/*
	185	* Print an buffer I/O error compatible with the fs/buffer.c. This
	186	* provides compatibility with dmesg scrapers that look for a specific
	187	* buffer I/O error message. We really need a unified error reporting
	188	* structure to userspace ala Digital Unix's uerf system, but it's
	189	* probably not going to happen in my lifetime, due to LKML politics...
	190	*/
	191	static void buffer_io_error(struct buffer_head *bh)
	192	{
	193	char b[BDEVNAME_SIZE];
	194	printk(KERN_ERR "Buffer I/O error on device %s, logical block %llu\n",
	195	bdevname(bh->b_bdev, b),
	196	(unsigned long long)bh->b_blocknr);
	197	}
	198
	199	static void ext4_end_bio(struct bio *bio, int error)
	200	{
	201	ext4_io_end_t *io_end = bio->bi_private;
	202	struct workqueue_struct *wq;
	203	struct inode *inode;
	204	unsigned long flags;
bd2d0210	205	int i;
d50bdd5a	206	sector_t bi_sector = bio->bi_sector;
bd2d0210 TT	207
bd2d0210 TT	208	BUG_ON(!io_end);
bd2d0210 TT	209	bio->bi_private = NULL;
	210	bio->bi_end_io = NULL;
	211	if (test_bit(BIO_UPTODATE, &bio->bi_flags))
	212	error = 0;
bd2d0210 TT	213	bio_put(bio);
bd2d0210 TT	214
bd2d0210 TT	215	for (i = 0; i < io_end->num_io_pages; i++) {
	216	struct page *page = io_end->pages[i]->p_page;
	217	struct buffer_head bh, head;
39db00f1 CW	218	loff_t offset;
39db00f1 CW	219	loff_t io_end_offset;
bd2d0210	220
39db00f1	221	if (error) {
bd2d0210	222	SetPageError(page);
39db00f1 CW	223	set_bit(AS_EIO, &page->mapping->flags);
	224	head = page_buffers(page);
	225	BUG_ON(!head);
	226
	227	io_end_offset = io_end->offset + io_end->size;
bd2d0210 TT	228
	229	offset = (sector_t) page->index << PAGE_CACHE_SHIFT;
	230	bh = head;
	231	do {
	232	if ((offset >= io_end->offset) &&
39db00f1 CW	233	(offset+bh->b_size <= io_end_offset))
	234	buffer_io_error(bh);
	235
bd2d0210 TT	236	offset += bh->b_size;
	237	bh = bh->b_this_page;
	238	} while (bh != head);
	239	}
	240
6268b325	241	put_io_page(io_end->pages[i]);
bd2d0210	242	}
6268b325	243	io_end->num_io_pages = 0;
f7ad6d2e TT	244	inode = io_end->inode;
	245
	246	if (error) {
	247	io_end->flag \|= EXT4_IO_END_ERROR;
	248	ext4_warning(inode->i_sb, "I/O error writing to inode %lu "
	249	"(offset %llu size %ld starting block %llu)",
	250	inode->i_ino,
	251	(unsigned long long) io_end->offset,
	252	(long) io_end->size,
	253	(unsigned long long)
d50bdd5a	254	bi_sector >> (inode->i_blkbits - 9));
f7ad6d2e	255	}
bd2d0210	256
b6168443 TT	257	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
	258	ext4_free_io_end(io_end);
	259	return;
	260	}
	261
bd2d0210 TT	262	/* Add the io_end to per-inode completed io list*/
	263	spin_lock_irqsave(&EXT4_I(inode)->i_completed_io_lock, flags);
	264	list_add_tail(&io_end->list, &EXT4_I(inode)->i_completed_io_list);
	265	spin_unlock_irqrestore(&EXT4_I(inode)->i_completed_io_lock, flags);
	266
	267	wq = EXT4_SB(inode->i_sb)->dio_unwritten_wq;
	268	/* queue the work to convert unwritten extents to written */
	269	queue_work(wq, &io_end->work);
	270	}
	271
	272	void ext4_io_submit(struct ext4_io_submit *io)
	273	{
	274	struct bio *bio = io->io_bio;
	275
	276	if (bio) {
	277	bio_get(io->io_bio);
	278	submit_bio(io->io_op, io->io_bio);
	279	BUG_ON(bio_flagged(io->io_bio, BIO_EOPNOTSUPP));
	280	bio_put(io->io_bio);
	281	}
7dc57615	282	io->io_bio = NULL;
bd2d0210	283	io->io_op = 0;
7dc57615	284	io->io_end = NULL;
bd2d0210 TT	285	}
	286
	287	static int io_submit_init(struct ext4_io_submit *io,
	288	struct inode *inode,
	289	struct writeback_control *wbc,
	290	struct buffer_head *bh)
	291	{
	292	ext4_io_end_t *io_end;
	293	struct page *page = bh->b_page;
	294	int nvecs = bio_get_nr_vecs(bh->b_bdev);
	295	struct bio *bio;
	296
	297	io_end = ext4_init_io_end(inode, GFP_NOFS);
	298	if (!io_end)
	299	return -ENOMEM;
275d3ba6	300	bio = bio_alloc(GFP_NOIO, min(nvecs, BIO_MAX_PAGES));
bd2d0210 TT	301	bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
	302	bio->bi_bdev = bh->b_bdev;
	303	bio->bi_private = io->io_end = io_end;
	304	bio->bi_end_io = ext4_end_bio;
	305
bd2d0210 TT	306	io_end->offset = (page->index << PAGE_CACHE_SHIFT) + bh_offset(bh);
	307
	308	io->io_bio = bio;
721a9602	309	io->io_op = (wbc->sync_mode == WB_SYNC_ALL ? WRITE_SYNC : WRITE);
bd2d0210 TT	310	io->io_next_block = bh->b_blocknr;
	311	return 0;
	312	}
	313
	314	static int io_submit_add_bh(struct ext4_io_submit *io,
	315	struct ext4_io_page *io_page,
	316	struct inode *inode,
	317	struct writeback_control *wbc,
	318	struct buffer_head *bh)
	319	{
	320	ext4_io_end_t *io_end;
	321	int ret;
	322
	323	if (buffer_new(bh)) {
	324	clear_buffer_new(bh);
	325	unmap_underlying_metadata(bh->b_bdev, bh->b_blocknr);
	326	}
	327
	328	if (!buffer_mapped(bh) \|\| buffer_delay(bh)) {
	329	if (!buffer_mapped(bh))
	330	clear_buffer_dirty(bh);
	331	if (io->io_bio)
	332	ext4_io_submit(io);
	333	return 0;
	334	}
	335
	336	if (io->io_bio && bh->b_blocknr != io->io_next_block) {
	337	submit_and_retry:
	338	ext4_io_submit(io);
	339	}
	340	if (io->io_bio == NULL) {
	341	ret = io_submit_init(io, inode, wbc, bh);
	342	if (ret)
	343	return ret;
	344	}
	345	io_end = io->io_end;
	346	if ((io_end->num_io_pages >= MAX_IO_PAGES) &&
	347	(io_end->pages[io_end->num_io_pages-1] != io_page))
	348	goto submit_and_retry;
0edeb71d TM	349	if (buffer_uninit(bh))
0edeb71d TM	350	ext4_set_io_unwritten_flag(inode, io_end);
bd2d0210 TT	351	io->io_end->size += bh->b_size;
	352	io->io_next_block++;
	353	ret = bio_add_page(io->io_bio, bh->b_page, bh->b_size, bh_offset(bh));
	354	if (ret != bh->b_size)
	355	goto submit_and_retry;
	356	if ((io_end->num_io_pages == 0) \|\|
	357	(io_end->pages[io_end->num_io_pages-1] != io_page)) {
	358	io_end->pages[io_end->num_io_pages++] = io_page;
83668e71	359	atomic_inc(&io_page->p_count);
bd2d0210 TT	360	}
	361	return 0;
	362	}
	363
	364	int ext4_bio_write_page(struct ext4_io_submit *io,
	365	struct page *page,
	366	int len,
	367	struct writeback_control *wbc)
	368	{
	369	struct inode *inode = page->mapping->host;
	370	unsigned block_start, block_end, blocksize;
	371	struct ext4_io_page *io_page;
	372	struct buffer_head bh, head;
	373	int ret = 0;
	374
	375	blocksize = 1 << inode->i_blkbits;
	376
d50bdd5a	377	BUG_ON(!PageLocked(page));
bd2d0210	378	BUG_ON(PageWriteback(page));
bd2d0210 TT	379
	380	io_page = kmem_cache_alloc(io_page_cachep, GFP_NOFS);
	381	if (!io_page) {
	382	set_page_dirty(page);
	383	unlock_page(page);
	384	return -ENOMEM;
	385	}
	386	io_page->p_page = page;
83668e71	387	atomic_set(&io_page->p_count, 1);
bd2d0210	388	get_page(page);
a54aa761 TT	389	set_page_writeback(page);
a54aa761 TT	390	ClearPageError(page);
bd2d0210 TT	391
	392	for (bh = head = page_buffers(page), block_start = 0;
	393	bh != head \|\| !block_start;
	394	block_start = block_end, bh = bh->b_this_page) {
d50bdd5a	395
bd2d0210 TT	396	block_end = block_start + blocksize;
bd2d0210 TT	397	if (block_start >= len) {
5a0dc736 YY	398	/*
	399	* Comments copied from block_write_full_page_endio:
	400	*
	401	* The page straddles i_size. It must be zeroed out on
	402	* each and every writepage invocation because it may
	403	* be mmapped. "A file is mapped in multiples of the
	404	* page size. For a file that is not a multiple of
	405	* the page size, the remaining memory is zeroed when
	406	* mapped, and writes to that region are not written
	407	* out to the file."
	408	*/
	409	zero_user_segment(page, block_start, block_end);
bd2d0210 TT	410	clear_buffer_dirty(bh);
	411	set_buffer_uptodate(bh);
	412	continue;
	413	}
d50bdd5a	414	clear_buffer_dirty(bh);
bd2d0210 TT	415	ret = io_submit_add_bh(io, io_page, inode, wbc, bh);
	416	if (ret) {
	417	/*
	418	* We only get here on ENOMEM. Not much else
	419	* we can do but mark the page as dirty, and
	420	* better luck next time.
	421	*/
	422	set_page_dirty(page);
	423	break;
	424	}
	425	}
	426	unlock_page(page);
	427	/*
	428	* If the page was truncated before we could do the writeback,
	429	* or we had a memory allocation error while trying to write
	430	* the first buffer head, we won't have submitted any pages for
	431	* I/O. In that case we need to make sure we've cleared the
	432	* PageWriteback bit from the page to prevent the system from
	433	* wedging later on.
	434	*/
83668e71	435	put_io_page(io_page);
bd2d0210 TT	436	return ret;
bd2d0210 TT	437	}