[deliverable/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/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 <linux/mm.h>

#include "ext4_jbd2.h"
#include "xattr.h"
#include "acl.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(!list_empty(&io->list));
	BUG_ON(io->flag & EXT4_IO_END_UNWRITTEN);

	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. */
static int ext4_end_io(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);

	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);
	}
	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(inode));
	return ret;
}

static void dump_completed_IO(struct inode *inode)
{
#ifdef	EXT4FS_DEBUG
	struct list_head *cur, *before, *after;
	ext4_io_end_t *io, *io0, *io1;
	unsigned long flags;

	if (list_empty(&EXT4_I(inode)->i_completed_io_list)) {
		ext4_debug("inode %lu completed_io list is empty\n",
			   inode->i_ino);
		return;
	}

	ext4_debug("Dump inode %lu completed_io list\n", inode->i_ino);
	list_for_each_entry(io, &EXT4_I(inode)->i_completed_io_list, list) {
		cur = &io->list;
		before = cur->prev;
		io0 = container_of(before, ext4_io_end_t, list);
		after = cur->next;
		io1 = container_of(after, ext4_io_end_t, list);

		ext4_debug("io 0x%p from inode %lu,prev 0x%p,next 0x%p\n",
			    io, inode->i_ino, io0, io1);
	}
#endif
}

/* Add the io_end to per-inode completed end_io list. */
void ext4_add_complete_io(ext4_io_end_t *io_end)
{
	struct ext4_inode_info *ei = EXT4_I(io_end->inode);
	struct workqueue_struct *wq;
	unsigned long flags;

	BUG_ON(!(io_end->flag & EXT4_IO_END_UNWRITTEN));
	wq = EXT4_SB(io_end->inode->i_sb)->dio_unwritten_wq;

	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	if (list_empty(&ei->i_completed_io_list)) {
		io_end->flag |= EXT4_IO_END_QUEUED;
		queue_work(wq, &io_end->work);
	}
	list_add_tail(&io_end->list, &ei->i_completed_io_list);
	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
}

static int ext4_do_flush_completed_IO(struct inode *inode,
				      ext4_io_end_t *work_io)
{
	ext4_io_end_t *io;
	struct list_head unwritten, complete, to_free;
	unsigned long flags;
	struct ext4_inode_info *ei = EXT4_I(inode);
	int err, ret = 0;

	INIT_LIST_HEAD(&complete);
	INIT_LIST_HEAD(&to_free);

	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	dump_completed_IO(inode);
	list_replace_init(&ei->i_completed_io_list, &unwritten);
	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);

	while (!list_empty(&unwritten)) {
		io = list_entry(unwritten.next, ext4_io_end_t, list);
		BUG_ON(!(io->flag & EXT4_IO_END_UNWRITTEN));
		list_del_init(&io->list);

		err = ext4_end_io(io);
		if (unlikely(!ret && err))
			ret = err;

		list_add_tail(&io->list, &complete);
	}
	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	while (!list_empty(&complete)) {
		io = list_entry(complete.next, ext4_io_end_t, list);
		io->flag &= ~EXT4_IO_END_UNWRITTEN;
		/* end_io context can not be destroyed now because it still
		 * used by queued worker. Worker thread will destroy it later */
		if (io->flag & EXT4_IO_END_QUEUED)
			list_del_init(&io->list);
		else
			list_move(&io->list, &to_free);
	}
	/* If we are called from worker context, it is time to clear queued
	 * flag, and destroy it's end_io if it was converted already */
	if (work_io) {
		work_io->flag &= ~EXT4_IO_END_QUEUED;
		if (!(work_io->flag & EXT4_IO_END_UNWRITTEN))
			list_add_tail(&work_io->list, &to_free);
	}
	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);

	while (!list_empty(&to_free)) {
		io = list_entry(to_free.next, ext4_io_end_t, list);
		list_del_init(&io->list);
		ext4_free_io_end(io);
	}
	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);
	ext4_do_flush_completed_IO(io->inode, io);
}

int ext4_flush_unwritten_io(struct inode *inode)
{
	int ret;
	WARN_ON_ONCE(!mutex_is_locked(&inode->i_mutex) &&
		     !(inode->i_state & I_FREEING));
	ret = ext4_do_flush_completed_IO(inode, NULL);
	ext4_unwritten_wait(inode);
	return ret;
}

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 inode *inode;
	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;
	}

	ext4_add_complete_io(io_end);
}

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) {
		redirty_page_for_writepage(wbc, 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;
		}
		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.
			 */
			redirty_page_for_writepage(wbc, page);
			break;
		}
		clear_buffer_dirty(bh);
	}
	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>
1ae48a63	26	#include <linux/mm.h>
bd2d0210 TT	27
	28	#include "ext4_jbd2.h"
	29	#include "xattr.h"
	30	#include "acl.h"
bd2d0210 TT	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);
28a535f9	74	BUG_ON(!list_empty(&io->list));
82e54229 DM	75	BUG_ON(io->flag & EXT4_IO_END_UNWRITTEN);
82e54229 DM	76
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
28a535f9 DM	85	/* check a range of space and convert unwritten extents to written. */
28a535f9 DM	86	static int ext4_end_io(ext4_io_end_t *io)
bd2d0210 TT	87	{
	88	struct inode *inode = io->inode;
	89	loff_t offset = io->offset;
	90	ssize_t size = io->size;
	91	int ret = 0;
	92
	93	ext4_debug("ext4_end_io_nolock: io 0x%p from inode %lu,list->next 0x%p,"
	94	"list->prev 0x%p\n",
	95	io, inode->i_ino, io->list.next, io->list.prev);
	96
bd2d0210 TT	97	ret = ext4_convert_unwritten_extents(inode, offset, size);
bd2d0210 TT	98	if (ret < 0) {
b82e384c TT	99	ext4_msg(inode->i_sb, KERN_EMERG,
	100	"failed to convert unwritten extents to written "
	101	"extents -- potential data loss! "
	102	"(inode %lu, offset %llu, size %zd, error %d)",
	103	inode->i_ino, offset, size, ret);
bd2d0210	104	}
bd2d0210 TT	105	if (io->iocb)
bd2d0210 TT	106	aio_complete(io->iocb, io->result, 0);
e9e3bcec	107
266991b1 JM	108	if (io->flag & EXT4_IO_END_DIRECT)
266991b1 JM	109	inode_dio_done(inode);
b82e384c	110	/* Wake up anyone waiting on unwritten extent conversion */
e27f41e1	111	if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
8d8c1825	112	wake_up_all(ext4_ioend_wq(inode));
bd2d0210 TT	113	return ret;
	114	}
	115
28a535f9 DM	116	static void dump_completed_IO(struct inode *inode)
	117	{
	118	#ifdef EXT4FS_DEBUG
	119	struct list_head cur, before, *after;
	120	ext4_io_end_t io, io0, *io1;
	121	unsigned long flags;
	122
	123	if (list_empty(&EXT4_I(inode)->i_completed_io_list)) {
	124	ext4_debug("inode %lu completed_io list is empty\n",
	125	inode->i_ino);
	126	return;
	127	}
	128
	129	ext4_debug("Dump inode %lu completed_io list\n", inode->i_ino);
	130	list_for_each_entry(io, &EXT4_I(inode)->i_completed_io_list, list) {
	131	cur = &io->list;
	132	before = cur->prev;
	133	io0 = container_of(before, ext4_io_end_t, list);
	134	after = cur->next;
	135	io1 = container_of(after, ext4_io_end_t, list);
	136
	137	ext4_debug("io 0x%p from inode %lu,prev 0x%p,next 0x%p\n",
	138	io, inode->i_ino, io0, io1);
	139	}
	140	#endif
	141	}
	142
	143	/* Add the io_end to per-inode completed end_io list. */
	144	void ext4_add_complete_io(ext4_io_end_t *io_end)
bd2d0210	145	{
28a535f9 DM	146	struct ext4_inode_info *ei = EXT4_I(io_end->inode);
	147	struct workqueue_struct *wq;
	148	unsigned long flags;
	149
	150	BUG_ON(!(io_end->flag & EXT4_IO_END_UNWRITTEN));
	151	wq = EXT4_SB(io_end->inode->i_sb)->dio_unwritten_wq;
bd2d0210	152
d73d5046	153	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
28a535f9 DM	154	if (list_empty(&ei->i_completed_io_list)) {
	155	io_end->flag \|= EXT4_IO_END_QUEUED;
	156	queue_work(wq, &io_end->work);
d73d5046	157	}
28a535f9 DM	158	list_add_tail(&io_end->list, &ei->i_completed_io_list);
	159	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
	160	}
d73d5046	161
28a535f9 DM	162	static int ext4_do_flush_completed_IO(struct inode *inode,
	163	ext4_io_end_t *work_io)
	164	{
	165	ext4_io_end_t *io;
	166	struct list_head unwritten, complete, to_free;
	167	unsigned long flags;
	168	struct ext4_inode_info *ei = EXT4_I(inode);
	169	int err, ret = 0;
	170
	171	INIT_LIST_HEAD(&complete);
	172	INIT_LIST_HEAD(&to_free);
	173
	174	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	175	dump_completed_IO(inode);
	176	list_replace_init(&ei->i_completed_io_list, &unwritten);
	177	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
	178
	179	while (!list_empty(&unwritten)) {
	180	io = list_entry(unwritten.next, ext4_io_end_t, list);
	181	BUG_ON(!(io->flag & EXT4_IO_END_UNWRITTEN));
	182	list_del_init(&io->list);
	183
	184	err = ext4_end_io(io);
	185	if (unlikely(!ret && err))
	186	ret = err;
	187
	188	list_add_tail(&io->list, &complete);
	189	}
28a535f9 DM	190	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	191	while (!list_empty(&complete)) {
	192	io = list_entry(complete.next, ext4_io_end_t, list);
	193	io->flag &= ~EXT4_IO_END_UNWRITTEN;
	194	/* end_io context can not be destroyed now because it still
	195	* used by queued worker. Worker thread will destroy it later */
	196	if (io->flag & EXT4_IO_END_QUEUED)
	197	list_del_init(&io->list);
	198	else
	199	list_move(&io->list, &to_free);
	200	}
	201	/* If we are called from worker context, it is time to clear queued
	202	* flag, and destroy it's end_io if it was converted already */
	203	if (work_io) {
	204	work_io->flag &= ~EXT4_IO_END_QUEUED;
	205	if (!(work_io->flag & EXT4_IO_END_UNWRITTEN))
	206	list_add_tail(&work_io->list, &to_free);
8c0bec21	207	}
bd2d0210	208	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
28a535f9 DM	209
	210	while (!list_empty(&to_free)) {
	211	io = list_entry(to_free.next, ext4_io_end_t, list);
	212	list_del_init(&io->list);
	213	ext4_free_io_end(io);
	214	}
	215	return ret;
	216	}
	217
	218	/*
	219	* work on completed aio dio IO, to convert unwritten extents to extents
	220	*/
	221	static void ext4_end_io_work(struct work_struct *work)
	222	{
	223	ext4_io_end_t *io = container_of(work, ext4_io_end_t, work);
	224	ext4_do_flush_completed_IO(io->inode, io);
	225	}
	226
c278531d	227	int ext4_flush_unwritten_io(struct inode *inode)
28a535f9	228	{
c278531d DM	229	int ret;
	230	WARN_ON_ONCE(!mutex_is_locked(&inode->i_mutex) &&
	231	!(inode->i_state & I_FREEING));
	232	ret = ext4_do_flush_completed_IO(inode, NULL);
	233	ext4_unwritten_wait(inode);
	234	return ret;
bd2d0210 TT	235	}
	236
	237	ext4_io_end_t ext4_init_io_end(struct inode inode, gfp_t flags)
	238	{
b17b35ec	239	ext4_io_end_t *io = kmem_cache_zalloc(io_end_cachep, flags);
bd2d0210	240	if (io) {
f7ad6d2e TT	241	atomic_inc(&EXT4_I(inode)->i_ioend_count);
f7ad6d2e TT	242	io->inode = inode;
bd2d0210 TT	243	INIT_WORK(&io->work, ext4_end_io_work);
	244	INIT_LIST_HEAD(&io->list);
	245	}
	246	return io;
	247	}
	248
	249	/*
	250	* Print an buffer I/O error compatible with the fs/buffer.c. This
	251	* provides compatibility with dmesg scrapers that look for a specific
	252	* buffer I/O error message. We really need a unified error reporting
	253	* structure to userspace ala Digital Unix's uerf system, but it's
	254	* probably not going to happen in my lifetime, due to LKML politics...
	255	*/
	256	static void buffer_io_error(struct buffer_head *bh)
	257	{
	258	char b[BDEVNAME_SIZE];
	259	printk(KERN_ERR "Buffer I/O error on device %s, logical block %llu\n",
	260	bdevname(bh->b_bdev, b),
	261	(unsigned long long)bh->b_blocknr);
	262	}
	263
	264	static void ext4_end_bio(struct bio *bio, int error)
	265	{
	266	ext4_io_end_t *io_end = bio->bi_private;
bd2d0210	267	struct inode *inode;
bd2d0210	268	int i;
d50bdd5a	269	sector_t bi_sector = bio->bi_sector;
bd2d0210 TT	270
bd2d0210 TT	271	BUG_ON(!io_end);
bd2d0210 TT	272	bio->bi_private = NULL;
	273	bio->bi_end_io = NULL;
	274	if (test_bit(BIO_UPTODATE, &bio->bi_flags))
	275	error = 0;
bd2d0210 TT	276	bio_put(bio);
bd2d0210 TT	277
bd2d0210 TT	278	for (i = 0; i < io_end->num_io_pages; i++) {
	279	struct page *page = io_end->pages[i]->p_page;
	280	struct buffer_head bh, head;
39db00f1 CW	281	loff_t offset;
39db00f1 CW	282	loff_t io_end_offset;
bd2d0210	283
39db00f1	284	if (error) {
bd2d0210	285	SetPageError(page);
39db00f1 CW	286	set_bit(AS_EIO, &page->mapping->flags);
	287	head = page_buffers(page);
	288	BUG_ON(!head);
	289
	290	io_end_offset = io_end->offset + io_end->size;
bd2d0210 TT	291
	292	offset = (sector_t) page->index << PAGE_CACHE_SHIFT;
	293	bh = head;
	294	do {
	295	if ((offset >= io_end->offset) &&
39db00f1 CW	296	(offset+bh->b_size <= io_end_offset))
	297	buffer_io_error(bh);
	298
bd2d0210 TT	299	offset += bh->b_size;
	300	bh = bh->b_this_page;
	301	} while (bh != head);
	302	}
	303
6268b325	304	put_io_page(io_end->pages[i]);
bd2d0210	305	}
6268b325	306	io_end->num_io_pages = 0;
f7ad6d2e TT	307	inode = io_end->inode;
	308
	309	if (error) {
	310	io_end->flag \|= EXT4_IO_END_ERROR;
	311	ext4_warning(inode->i_sb, "I/O error writing to inode %lu "
	312	"(offset %llu size %ld starting block %llu)",
	313	inode->i_ino,
	314	(unsigned long long) io_end->offset,
	315	(long) io_end->size,
	316	(unsigned long long)
d50bdd5a	317	bi_sector >> (inode->i_blkbits - 9));
f7ad6d2e	318	}
bd2d0210	319
b6168443 TT	320	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
	321	ext4_free_io_end(io_end);
	322	return;
	323	}
	324
28a535f9	325	ext4_add_complete_io(io_end);
bd2d0210 TT	326	}
	327
	328	void ext4_io_submit(struct ext4_io_submit *io)
	329	{
	330	struct bio *bio = io->io_bio;
	331
	332	if (bio) {
	333	bio_get(io->io_bio);
	334	submit_bio(io->io_op, io->io_bio);
	335	BUG_ON(bio_flagged(io->io_bio, BIO_EOPNOTSUPP));
	336	bio_put(io->io_bio);
	337	}
7dc57615	338	io->io_bio = NULL;
bd2d0210	339	io->io_op = 0;
7dc57615	340	io->io_end = NULL;
bd2d0210 TT	341	}
	342
	343	static int io_submit_init(struct ext4_io_submit *io,
	344	struct inode *inode,
	345	struct writeback_control *wbc,
	346	struct buffer_head *bh)
	347	{
	348	ext4_io_end_t *io_end;
	349	struct page *page = bh->b_page;
	350	int nvecs = bio_get_nr_vecs(bh->b_bdev);
	351	struct bio *bio;
	352
	353	io_end = ext4_init_io_end(inode, GFP_NOFS);
	354	if (!io_end)
	355	return -ENOMEM;
275d3ba6	356	bio = bio_alloc(GFP_NOIO, min(nvecs, BIO_MAX_PAGES));
bd2d0210 TT	357	bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
	358	bio->bi_bdev = bh->b_bdev;
	359	bio->bi_private = io->io_end = io_end;
	360	bio->bi_end_io = ext4_end_bio;
	361
bd2d0210 TT	362	io_end->offset = (page->index << PAGE_CACHE_SHIFT) + bh_offset(bh);
	363
	364	io->io_bio = bio;
721a9602	365	io->io_op = (wbc->sync_mode == WB_SYNC_ALL ? WRITE_SYNC : WRITE);
bd2d0210 TT	366	io->io_next_block = bh->b_blocknr;
	367	return 0;
	368	}
	369
	370	static int io_submit_add_bh(struct ext4_io_submit *io,
	371	struct ext4_io_page *io_page,
	372	struct inode *inode,
	373	struct writeback_control *wbc,
	374	struct buffer_head *bh)
	375	{
	376	ext4_io_end_t *io_end;
	377	int ret;
	378
	379	if (buffer_new(bh)) {
	380	clear_buffer_new(bh);
	381	unmap_underlying_metadata(bh->b_bdev, bh->b_blocknr);
	382	}
	383
	384	if (!buffer_mapped(bh) \|\| buffer_delay(bh)) {
	385	if (!buffer_mapped(bh))
	386	clear_buffer_dirty(bh);
	387	if (io->io_bio)
	388	ext4_io_submit(io);
	389	return 0;
	390	}
	391
	392	if (io->io_bio && bh->b_blocknr != io->io_next_block) {
	393	submit_and_retry:
	394	ext4_io_submit(io);
	395	}
	396	if (io->io_bio == NULL) {
	397	ret = io_submit_init(io, inode, wbc, bh);
	398	if (ret)
	399	return ret;
	400	}
	401	io_end = io->io_end;
	402	if ((io_end->num_io_pages >= MAX_IO_PAGES) &&
	403	(io_end->pages[io_end->num_io_pages-1] != io_page))
	404	goto submit_and_retry;
0edeb71d TM	405	if (buffer_uninit(bh))
0edeb71d TM	406	ext4_set_io_unwritten_flag(inode, io_end);
bd2d0210 TT	407	io->io_end->size += bh->b_size;
	408	io->io_next_block++;
	409	ret = bio_add_page(io->io_bio, bh->b_page, bh->b_size, bh_offset(bh));
	410	if (ret != bh->b_size)
	411	goto submit_and_retry;
	412	if ((io_end->num_io_pages == 0) \|\|
	413	(io_end->pages[io_end->num_io_pages-1] != io_page)) {
	414	io_end->pages[io_end->num_io_pages++] = io_page;
83668e71	415	atomic_inc(&io_page->p_count);
bd2d0210 TT	416	}
	417	return 0;
	418	}
	419
	420	int ext4_bio_write_page(struct ext4_io_submit *io,
	421	struct page *page,
	422	int len,
	423	struct writeback_control *wbc)
	424	{
	425	struct inode *inode = page->mapping->host;
	426	unsigned block_start, block_end, blocksize;
	427	struct ext4_io_page *io_page;
	428	struct buffer_head bh, head;
	429	int ret = 0;
	430
	431	blocksize = 1 << inode->i_blkbits;
	432
d50bdd5a	433	BUG_ON(!PageLocked(page));
bd2d0210	434	BUG_ON(PageWriteback(page));
bd2d0210 TT	435
	436	io_page = kmem_cache_alloc(io_page_cachep, GFP_NOFS);
	437	if (!io_page) {
1ae48a63	438	redirty_page_for_writepage(wbc, page);
bd2d0210 TT	439	unlock_page(page);
	440	return -ENOMEM;
	441	}
	442	io_page->p_page = page;
83668e71	443	atomic_set(&io_page->p_count, 1);
bd2d0210	444	get_page(page);
a54aa761 TT	445	set_page_writeback(page);
a54aa761 TT	446	ClearPageError(page);
bd2d0210 TT	447
	448	for (bh = head = page_buffers(page), block_start = 0;
	449	bh != head \|\| !block_start;
	450	block_start = block_end, bh = bh->b_this_page) {
d50bdd5a	451
bd2d0210 TT	452	block_end = block_start + blocksize;
bd2d0210 TT	453	if (block_start >= len) {
5a0dc736 YY	454	/*
	455	* Comments copied from block_write_full_page_endio:
	456	*
	457	* The page straddles i_size. It must be zeroed out on
	458	* each and every writepage invocation because it may
	459	* be mmapped. "A file is mapped in multiples of the
	460	* page size. For a file that is not a multiple of
	461	* the page size, the remaining memory is zeroed when
	462	* mapped, and writes to that region are not written
	463	* out to the file."
	464	*/
	465	zero_user_segment(page, block_start, block_end);
bd2d0210 TT	466	clear_buffer_dirty(bh);
	467	set_buffer_uptodate(bh);
	468	continue;
	469	}
	470	ret = io_submit_add_bh(io, io_page, inode, wbc, bh);
	471	if (ret) {
	472	/*
	473	* We only get here on ENOMEM. Not much else
	474	* we can do but mark the page as dirty, and
	475	* better luck next time.
	476	*/
1ae48a63	477	redirty_page_for_writepage(wbc, page);
bd2d0210 TT	478	break;
bd2d0210 TT	479	}
1ae48a63	480	clear_buffer_dirty(bh);
bd2d0210 TT	481	}
	482	unlock_page(page);
	483	/*
	484	* If the page was truncated before we could do the writeback,
	485	* or we had a memory allocation error while trying to write
	486	* the first buffer head, we won't have submitted any pages for
	487	* I/O. In that case we need to make sure we've cleared the
	488	* PageWriteback bit from the page to prevent the system from
	489	* wedging later on.
	490	*/
83668e71	491	put_io_page(io_page);
bd2d0210 TT	492	return ret;
bd2d0210 TT	493	}