[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/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/aio.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_end_cachep;

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

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

/*
 * This function is called by ext4_evict_inode() to make sure there is
 * no more pending I/O completion work left to do.
 */
void ext4_ioend_shutdown(struct inode *inode)
{
	wait_queue_head_t *wq = ext4_ioend_wq(inode);

	wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_ioend_count) == 0));
	/*
	 * We need to make sure the work structure is finished being
	 * used before we let the inode get destroyed.
	 */
	if (work_pending(&EXT4_I(inode)->i_rsv_conversion_work))
		cancel_work_sync(&EXT4_I(inode)->i_rsv_conversion_work);
	if (work_pending(&EXT4_I(inode)->i_unrsv_conversion_work))
		cancel_work_sync(&EXT4_I(inode)->i_unrsv_conversion_work);
}

static void ext4_release_io_end(ext4_io_end_t *io_end)
{
	BUG_ON(!list_empty(&io_end->list));
	BUG_ON(io_end->flag & EXT4_IO_END_UNWRITTEN);
	WARN_ON(io_end->handle);

	if (atomic_dec_and_test(&EXT4_I(io_end->inode)->i_ioend_count))
		wake_up_all(ext4_ioend_wq(io_end->inode));
	if (io_end->flag & EXT4_IO_END_DIRECT)
		inode_dio_done(io_end->inode);
	if (io_end->iocb)
		aio_complete(io_end->iocb, io_end->result, 0);
	kmem_cache_free(io_end_cachep, io_end);
}

static void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
{
	struct inode *inode = io_end->inode;

	io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
	/* 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));
}

/* 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;
	handle_t *handle = io->handle;
	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);

	io->handle = NULL;	/* Following call will use up the handle */
	ret = ext4_convert_unwritten_extents(handle, 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);
	}
	ext4_clear_io_unwritten_flag(io);
	ext4_release_io_end(io);
	return ret;
}

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

	if (list_empty(head))
		return;

	ext4_debug("Dump inode %lu completed io list\n", inode->i_ino);
	list_for_each_entry(io, head, 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. */
static 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));
	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	if (io_end->handle) {
		wq = EXT4_SB(io_end->inode->i_sb)->rsv_conversion_wq;
		if (list_empty(&ei->i_rsv_conversion_list))
			queue_work(wq, &ei->i_rsv_conversion_work);
		list_add_tail(&io_end->list, &ei->i_rsv_conversion_list);
	} else {
		wq = EXT4_SB(io_end->inode->i_sb)->unrsv_conversion_wq;
		if (list_empty(&ei->i_unrsv_conversion_list))
			queue_work(wq, &ei->i_unrsv_conversion_work);
		list_add_tail(&io_end->list, &ei->i_unrsv_conversion_list);
	}
	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
}

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

	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	dump_completed_IO(inode, head);
	list_replace_init(head, &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;
	}
	return ret;
}

/*
 * work on completed IO, to convert unwritten extents to extents
 */
void ext4_end_io_rsv_work(struct work_struct *work)
{
	struct ext4_inode_info *ei = container_of(work, struct ext4_inode_info,
						  i_rsv_conversion_work);
	ext4_do_flush_completed_IO(&ei->vfs_inode, &ei->i_rsv_conversion_list);
}

void ext4_end_io_unrsv_work(struct work_struct *work)
{
	struct ext4_inode_info *ei = container_of(work, struct ext4_inode_info,
						  i_unrsv_conversion_work);
	ext4_do_flush_completed_IO(&ei->vfs_inode, &ei->i_unrsv_conversion_list);
}

int ext4_flush_unwritten_io(struct inode *inode)
{
	int ret, err;

	WARN_ON_ONCE(!mutex_is_locked(&inode->i_mutex) &&
		     !(inode->i_state & I_FREEING));
	ret = ext4_do_flush_completed_IO(inode,
					 &EXT4_I(inode)->i_rsv_conversion_list);
	err = ext4_do_flush_completed_IO(inode,
					 &EXT4_I(inode)->i_unrsv_conversion_list);
	if (!ret)
		ret = err;
	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_LIST_HEAD(&io->list);
		atomic_set(&io->count, 1);
	}
	return io;
}

void ext4_put_io_end_defer(ext4_io_end_t *io_end)
{
	if (atomic_dec_and_test(&io_end->count)) {
		if (!(io_end->flag & EXT4_IO_END_UNWRITTEN) || !io_end->size) {
			ext4_release_io_end(io_end);
			return;
		}
		ext4_add_complete_io(io_end);
	}
}

int ext4_put_io_end(ext4_io_end_t *io_end)
{
	int err = 0;

	if (atomic_dec_and_test(&io_end->count)) {
		if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
			err = ext4_convert_unwritten_extents(io_end->handle,
						io_end->inode, io_end->offset,
						io_end->size);
			io_end->handle = NULL;
			ext4_clear_io_unwritten_flag(io_end);
		}
		ext4_release_io_end(io_end);
	}
	return err;
}

ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end)
{
	atomic_inc(&io_end->count);
	return io_end;
}

/*
 * 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;
	int blocksize;
	sector_t bi_sector = bio->bi_sector;

	BUG_ON(!io_end);
	inode = io_end->inode;
	blocksize = 1 << inode->i_blkbits;
	bio->bi_private = NULL;
	bio->bi_end_io = NULL;
	if (test_bit(BIO_UPTODATE, &bio->bi_flags))
		error = 0;
	for (i = 0; i < bio->bi_vcnt; i++) {
		struct bio_vec *bvec = &bio->bi_io_vec[i];
		struct page *page = bvec->bv_page;
		struct buffer_head *bh, *head;
		unsigned bio_start = bvec->bv_offset;
		unsigned bio_end = bio_start + bvec->bv_len;
		unsigned under_io = 0;
		unsigned long flags;

		if (!page)
			continue;

		if (error) {
			SetPageError(page);
			set_bit(AS_EIO, &page->mapping->flags);
		}
		bh = head = page_buffers(page);
		/*
		 * We check all buffers in the page under BH_Uptodate_Lock
		 * to avoid races with other end io clearing async_write flags
		 */
		local_irq_save(flags);
		bit_spin_lock(BH_Uptodate_Lock, &head->b_state);
		do {
			if (bh_offset(bh) < bio_start ||
			    bh_offset(bh) + blocksize > bio_end) {
				if (buffer_async_write(bh))
					under_io++;
				continue;
			}
			clear_buffer_async_write(bh);
			if (error)
				buffer_io_error(bh);
		} while ((bh = bh->b_this_page) != head);
		bit_spin_unlock(BH_Uptodate_Lock, &head->b_state);
		local_irq_restore(flags);
		if (!under_io)
			end_page_writeback(page);
	}
	bio_put(bio);

	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));
	}

	ext4_put_io_end_defer(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;
}

void ext4_io_submit_init(struct ext4_io_submit *io,
			 struct writeback_control *wbc)
{
	io->io_op = (wbc->sync_mode == WB_SYNC_ALL ?  WRITE_SYNC : WRITE);
	io->io_bio = NULL;
	io->io_end = NULL;
}

static int io_submit_init_bio(struct ext4_io_submit *io,
			      struct buffer_head *bh)
{
	int nvecs = bio_get_nr_vecs(bh->b_bdev);
	struct bio *bio;

	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_end_io = ext4_end_bio;
	bio->bi_private = ext4_get_io_end(io->io_end);
	io->io_bio = bio;
	io->io_next_block = bh->b_blocknr;
	return 0;
}

static int io_submit_add_bh(struct ext4_io_submit *io,
			    struct inode *inode,
			    struct buffer_head *bh)
{
	int ret;

	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_bio(io, bh);
		if (ret)
			return ret;
	}
	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;
	io->io_next_block++;
	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, blocksize;
	struct buffer_head *bh, *head;
	int ret = 0;
	int nr_submitted = 0;

	blocksize = 1 << inode->i_blkbits;

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

	set_page_writeback(page);
	ClearPageError(page);

	/*
	 * In the first loop we prepare and mark buffers to submit. We have to
	 * mark all buffers in the page before submitting so that
	 * end_page_writeback() cannot be called from ext4_bio_end_io() when IO
	 * on the first buffer finishes and we are still working on submitting
	 * the second buffer.
	 */
	bh = head = page_buffers(page);
	do {
		block_start = bh_offset(bh);
		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_start + blocksize);
			clear_buffer_dirty(bh);
			set_buffer_uptodate(bh);
			continue;
		}
		if (!buffer_dirty(bh) || buffer_delay(bh) ||
		    !buffer_mapped(bh) || buffer_unwritten(bh)) {
			/* A hole? We can safely clear the dirty bit */
			if (!buffer_mapped(bh))
				clear_buffer_dirty(bh);
			if (io->io_bio)
				ext4_io_submit(io);
			continue;
		}
		if (buffer_new(bh)) {
			clear_buffer_new(bh);
			unmap_underlying_metadata(bh->b_bdev, bh->b_blocknr);
		}
		set_buffer_async_write(bh);
	} while ((bh = bh->b_this_page) != head);

	/* Now submit buffers to write */
	bh = head = page_buffers(page);
	do {
		if (!buffer_async_write(bh))
			continue;
		ret = io_submit_add_bh(io, inode, 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;
		}
		nr_submitted++;
		clear_buffer_dirty(bh);
	} while ((bh = bh->b_this_page) != head);

	/* Error stopped previous loop? Clean up buffers... */
	if (ret) {
		do {
			clear_buffer_async_write(bh);
			bh = bh->b_this_page;
		} while (bh != head);
	}
	unlock_page(page);
	/* Nothing submitted - we have to end page writeback */
	if (!nr_submitted)
		end_page_writeback(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>
a27bb332	21	#include <linux/aio.h>
bd2d0210 TT	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>
1ae48a63	27	#include <linux/mm.h>
bd2d0210 TT	28
	29	#include "ext4_jbd2.h"
	30	#include "xattr.h"
	31	#include "acl.h"
bd2d0210	32
0058f965	33	static struct kmem_cache *io_end_cachep;
bd2d0210	34
5dabfc78	35	int __init ext4_init_pageio(void)
bd2d0210	36	{
bd2d0210	37	io_end_cachep = KMEM_CACHE(ext4_io_end, SLAB_RECLAIM_ACCOUNT);
0058f965	38	if (io_end_cachep == NULL)
bd2d0210	39	return -ENOMEM;
bd2d0210 TT	40	return 0;
	41	}
	42
5dabfc78	43	void ext4_exit_pageio(void)
bd2d0210 TT	44	{
bd2d0210 TT	45	kmem_cache_destroy(io_end_cachep);
bd2d0210 TT	46	}
bd2d0210 TT	47
1ada47d9 TT	48	/*
	49	* This function is called by ext4_evict_inode() to make sure there is
	50	* no more pending I/O completion work left to do.
	51	*/
	52	void ext4_ioend_shutdown(struct inode *inode)
f7ad6d2e	53	{
e9e3bcec	54	wait_queue_head_t *wq = ext4_ioend_wq(inode);
f7ad6d2e TT	55
f7ad6d2e TT	56	wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_ioend_count) == 0));
1ada47d9 TT	57	/*
	58	* We need to make sure the work structure is finished being
	59	* used before we let the inode get destroyed.
	60	*/
2e8fa54e JK	61	if (work_pending(&EXT4_I(inode)->i_rsv_conversion_work))
	62	cancel_work_sync(&EXT4_I(inode)->i_rsv_conversion_work);
	63	if (work_pending(&EXT4_I(inode)->i_unrsv_conversion_work))
	64	cancel_work_sync(&EXT4_I(inode)->i_unrsv_conversion_work);
f7ad6d2e TT	65	}
f7ad6d2e TT	66
97a851ed	67	static void ext4_release_io_end(ext4_io_end_t *io_end)
bd2d0210	68	{
97a851ed JK	69	BUG_ON(!list_empty(&io_end->list));
97a851ed JK	70	BUG_ON(io_end->flag & EXT4_IO_END_UNWRITTEN);
6b523df4	71	WARN_ON(io_end->handle);
97a851ed JK	72
	73	if (atomic_dec_and_test(&EXT4_I(io_end->inode)->i_ioend_count))
	74	wake_up_all(ext4_ioend_wq(io_end->inode));
	75	if (io_end->flag & EXT4_IO_END_DIRECT)
	76	inode_dio_done(io_end->inode);
	77	if (io_end->iocb)
	78	aio_complete(io_end->iocb, io_end->result, 0);
	79	kmem_cache_free(io_end_cachep, io_end);
	80	}
	81
	82	static void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
	83	{
	84	struct inode *inode = io_end->inode;
82e54229	85
97a851ed JK	86	io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
	87	/* Wake up anyone waiting on unwritten extent conversion */
	88	if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
	89	wake_up_all(ext4_ioend_wq(inode));
bd2d0210 TT	90	}
bd2d0210 TT	91
28a535f9 DM	92	/* check a range of space and convert unwritten extents to written. */
28a535f9 DM	93	static int ext4_end_io(ext4_io_end_t *io)
bd2d0210 TT	94	{
	95	struct inode *inode = io->inode;
	96	loff_t offset = io->offset;
	97	ssize_t size = io->size;
6b523df4	98	handle_t *handle = io->handle;
bd2d0210 TT	99	int ret = 0;
	100
	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
6b523df4 JK	105	io->handle = NULL; /* Following call will use up the handle */
6b523df4 JK	106	ret = ext4_convert_unwritten_extents(handle, inode, offset, size);
bd2d0210	107	if (ret < 0) {
b82e384c TT	108	ext4_msg(inode->i_sb, KERN_EMERG,
	109	"failed to convert unwritten extents to written "
	110	"extents -- potential data loss! "
	111	"(inode %lu, offset %llu, size %zd, error %d)",
	112	inode->i_ino, offset, size, ret);
bd2d0210	113	}
97a851ed JK	114	ext4_clear_io_unwritten_flag(io);
97a851ed JK	115	ext4_release_io_end(io);
bd2d0210 TT	116	return ret;
	117	}
	118
2e8fa54e	119	static void dump_completed_IO(struct inode inode, struct list_head head)
28a535f9 DM	120	{
	121	#ifdef EXT4FS_DEBUG
	122	struct list_head cur, before, *after;
	123	ext4_io_end_t io, io0, *io1;
28a535f9	124
2e8fa54e	125	if (list_empty(head))
28a535f9	126	return;
28a535f9	127
2e8fa54e JK	128	ext4_debug("Dump inode %lu completed io list\n", inode->i_ino);
2e8fa54e JK	129	list_for_each_entry(io, head, list) {
28a535f9 DM	130	cur = &io->list;
	131	before = cur->prev;
	132	io0 = container_of(before, ext4_io_end_t, list);
	133	after = cur->next;
	134	io1 = container_of(after, ext4_io_end_t, list);
	135
	136	ext4_debug("io 0x%p from inode %lu,prev 0x%p,next 0x%p\n",
	137	io, inode->i_ino, io0, io1);
	138	}
	139	#endif
	140	}
	141
	142	/* Add the io_end to per-inode completed end_io list. */
97a851ed	143	static void ext4_add_complete_io(ext4_io_end_t *io_end)
bd2d0210	144	{
28a535f9 DM	145	struct ext4_inode_info *ei = EXT4_I(io_end->inode);
	146	struct workqueue_struct *wq;
	147	unsigned long flags;
	148
	149	BUG_ON(!(io_end->flag & EXT4_IO_END_UNWRITTEN));
d73d5046	150	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
2e8fa54e JK	151	if (io_end->handle) {
	152	wq = EXT4_SB(io_end->inode->i_sb)->rsv_conversion_wq;
	153	if (list_empty(&ei->i_rsv_conversion_list))
	154	queue_work(wq, &ei->i_rsv_conversion_work);
	155	list_add_tail(&io_end->list, &ei->i_rsv_conversion_list);
	156	} else {
	157	wq = EXT4_SB(io_end->inode->i_sb)->unrsv_conversion_wq;
	158	if (list_empty(&ei->i_unrsv_conversion_list))
	159	queue_work(wq, &ei->i_unrsv_conversion_work);
	160	list_add_tail(&io_end->list, &ei->i_unrsv_conversion_list);
	161	}
28a535f9 DM	162	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
28a535f9 DM	163	}
d73d5046	164
2e8fa54e JK	165	static int ext4_do_flush_completed_IO(struct inode *inode,
2e8fa54e JK	166	struct list_head *head)
28a535f9 DM	167	{
28a535f9 DM	168	ext4_io_end_t *io;
002bd7fa	169	struct list_head unwritten;
28a535f9 DM	170	unsigned long flags;
	171	struct ext4_inode_info *ei = EXT4_I(inode);
	172	int err, ret = 0;
	173
28a535f9	174	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
2e8fa54e JK	175	dump_completed_IO(inode, head);
2e8fa54e JK	176	list_replace_init(head, &unwritten);
28a535f9 DM	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;
28a535f9 DM	187	}
	188	return ret;
	189	}
	190
	191	/*
2e8fa54e	192	* work on completed IO, to convert unwritten extents to extents
28a535f9	193	*/
2e8fa54e JK	194	void ext4_end_io_rsv_work(struct work_struct *work)
	195	{
	196	struct ext4_inode_info *ei = container_of(work, struct ext4_inode_info,
	197	i_rsv_conversion_work);
	198	ext4_do_flush_completed_IO(&ei->vfs_inode, &ei->i_rsv_conversion_list);
	199	}
	200
	201	void ext4_end_io_unrsv_work(struct work_struct *work)
28a535f9	202	{
84c17543	203	struct ext4_inode_info *ei = container_of(work, struct ext4_inode_info,
2e8fa54e JK	204	i_unrsv_conversion_work);
2e8fa54e JK	205	ext4_do_flush_completed_IO(&ei->vfs_inode, &ei->i_unrsv_conversion_list);
28a535f9 DM	206	}
28a535f9 DM	207
c278531d	208	int ext4_flush_unwritten_io(struct inode *inode)
28a535f9	209	{
2e8fa54e JK	210	int ret, err;
2e8fa54e JK	211
c278531d DM	212	WARN_ON_ONCE(!mutex_is_locked(&inode->i_mutex) &&
c278531d DM	213	!(inode->i_state & I_FREEING));
2e8fa54e JK	214	ret = ext4_do_flush_completed_IO(inode,
	215	&EXT4_I(inode)->i_rsv_conversion_list);
	216	err = ext4_do_flush_completed_IO(inode,
	217	&EXT4_I(inode)->i_unrsv_conversion_list);
	218	if (!ret)
	219	ret = err;
c278531d DM	220	ext4_unwritten_wait(inode);
c278531d DM	221	return ret;
bd2d0210 TT	222	}
	223
	224	ext4_io_end_t ext4_init_io_end(struct inode inode, gfp_t flags)
	225	{
b17b35ec	226	ext4_io_end_t *io = kmem_cache_zalloc(io_end_cachep, flags);
bd2d0210	227	if (io) {
f7ad6d2e TT	228	atomic_inc(&EXT4_I(inode)->i_ioend_count);
f7ad6d2e TT	229	io->inode = inode;
bd2d0210	230	INIT_LIST_HEAD(&io->list);
97a851ed	231	atomic_set(&io->count, 1);
bd2d0210 TT	232	}
	233	return io;
	234	}
	235
97a851ed JK	236	void ext4_put_io_end_defer(ext4_io_end_t *io_end)
	237	{
	238	if (atomic_dec_and_test(&io_end->count)) {
	239	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN) \|\| !io_end->size) {
	240	ext4_release_io_end(io_end);
	241	return;
	242	}
	243	ext4_add_complete_io(io_end);
	244	}
	245	}
	246
	247	int ext4_put_io_end(ext4_io_end_t *io_end)
	248	{
	249	int err = 0;
	250
	251	if (atomic_dec_and_test(&io_end->count)) {
	252	if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
6b523df4 JK	253	err = ext4_convert_unwritten_extents(io_end->handle,
	254	io_end->inode, io_end->offset,
	255	io_end->size);
	256	io_end->handle = NULL;
97a851ed JK	257	ext4_clear_io_unwritten_flag(io_end);
	258	}
	259	ext4_release_io_end(io_end);
	260	}
	261	return err;
	262	}
	263
	264	ext4_io_end_t ext4_get_io_end(ext4_io_end_t io_end)
	265	{
	266	atomic_inc(&io_end->count);
	267	return io_end;
	268	}
	269
bd2d0210 TT	270	/*
	271	* Print an buffer I/O error compatible with the fs/buffer.c. This
	272	* provides compatibility with dmesg scrapers that look for a specific
	273	* buffer I/O error message. We really need a unified error reporting
	274	* structure to userspace ala Digital Unix's uerf system, but it's
	275	* probably not going to happen in my lifetime, due to LKML politics...
	276	*/
	277	static void buffer_io_error(struct buffer_head *bh)
	278	{
	279	char b[BDEVNAME_SIZE];
	280	printk(KERN_ERR "Buffer I/O error on device %s, logical block %llu\n",
	281	bdevname(bh->b_bdev, b),
	282	(unsigned long long)bh->b_blocknr);
	283	}
	284
	285	static void ext4_end_bio(struct bio *bio, int error)
	286	{
	287	ext4_io_end_t *io_end = bio->bi_private;
bd2d0210	288	struct inode *inode;
bd2d0210	289	int i;
0058f965	290	int blocksize;
d50bdd5a	291	sector_t bi_sector = bio->bi_sector;
bd2d0210 TT	292
bd2d0210 TT	293	BUG_ON(!io_end);
0058f965 JK	294	inode = io_end->inode;
0058f965 JK	295	blocksize = 1 << inode->i_blkbits;
bd2d0210 TT	296	bio->bi_private = NULL;
	297	bio->bi_end_io = NULL;
	298	if (test_bit(BIO_UPTODATE, &bio->bi_flags))
	299	error = 0;
0058f965 JK	300	for (i = 0; i < bio->bi_vcnt; i++) {
	301	struct bio_vec *bvec = &bio->bi_io_vec[i];
	302	struct page *page = bvec->bv_page;
bd2d0210	303	struct buffer_head bh, head;
0058f965 JK	304	unsigned bio_start = bvec->bv_offset;
	305	unsigned bio_end = bio_start + bvec->bv_len;
	306	unsigned under_io = 0;
	307	unsigned long flags;
	308
	309	if (!page)
	310	continue;
bd2d0210	311
39db00f1	312	if (error) {
bd2d0210	313	SetPageError(page);
39db00f1	314	set_bit(AS_EIO, &page->mapping->flags);
bd2d0210	315	}
0058f965 JK	316	bh = head = page_buffers(page);
	317	/*
	318	* We check all buffers in the page under BH_Uptodate_Lock
	319	* to avoid races with other end io clearing async_write flags
	320	*/
	321	local_irq_save(flags);
	322	bit_spin_lock(BH_Uptodate_Lock, &head->b_state);
	323	do {
	324	if (bh_offset(bh) < bio_start \|\|
	325	bh_offset(bh) + blocksize > bio_end) {
	326	if (buffer_async_write(bh))
	327	under_io++;
	328	continue;
	329	}
	330	clear_buffer_async_write(bh);
	331	if (error)
	332	buffer_io_error(bh);
	333	} while ((bh = bh->b_this_page) != head);
	334	bit_spin_unlock(BH_Uptodate_Lock, &head->b_state);
	335	local_irq_restore(flags);
	336	if (!under_io)
	337	end_page_writeback(page);
bd2d0210	338	}
0058f965	339	bio_put(bio);
f7ad6d2e TT	340
	341	if (error) {
	342	io_end->flag \|= EXT4_IO_END_ERROR;
	343	ext4_warning(inode->i_sb, "I/O error writing to inode %lu "
	344	"(offset %llu size %ld starting block %llu)",
	345	inode->i_ino,
	346	(unsigned long long) io_end->offset,
	347	(long) io_end->size,
	348	(unsigned long long)
d50bdd5a	349	bi_sector >> (inode->i_blkbits - 9));
f7ad6d2e	350	}
bd2d0210	351
97a851ed	352	ext4_put_io_end_defer(io_end);
bd2d0210 TT	353	}
	354
	355	void ext4_io_submit(struct ext4_io_submit *io)
	356	{
	357	struct bio *bio = io->io_bio;
	358
	359	if (bio) {
	360	bio_get(io->io_bio);
	361	submit_bio(io->io_op, io->io_bio);
	362	BUG_ON(bio_flagged(io->io_bio, BIO_EOPNOTSUPP));
	363	bio_put(io->io_bio);
	364	}
7dc57615	365	io->io_bio = NULL;
97a851ed JK	366	}
	367
	368	void ext4_io_submit_init(struct ext4_io_submit *io,
	369	struct writeback_control *wbc)
	370	{
	371	io->io_op = (wbc->sync_mode == WB_SYNC_ALL ? WRITE_SYNC : WRITE);
	372	io->io_bio = NULL;
7dc57615	373	io->io_end = NULL;
bd2d0210 TT	374	}
bd2d0210 TT	375
97a851ed JK	376	static int io_submit_init_bio(struct ext4_io_submit *io,
97a851ed JK	377	struct buffer_head *bh)
bd2d0210	378	{
bd2d0210 TT	379	int nvecs = bio_get_nr_vecs(bh->b_bdev);
	380	struct bio *bio;
	381
275d3ba6	382	bio = bio_alloc(GFP_NOIO, min(nvecs, BIO_MAX_PAGES));
bd2d0210 TT	383	bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
bd2d0210 TT	384	bio->bi_bdev = bh->b_bdev;
bd2d0210	385	bio->bi_end_io = ext4_end_bio;
97a851ed	386	bio->bi_private = ext4_get_io_end(io->io_end);
bd2d0210	387	io->io_bio = bio;
bd2d0210 TT	388	io->io_next_block = bh->b_blocknr;
	389	return 0;
	390	}
	391
	392	static int io_submit_add_bh(struct ext4_io_submit *io,
bd2d0210	393	struct inode *inode,
bd2d0210 TT	394	struct buffer_head *bh)
bd2d0210 TT	395	{
bd2d0210 TT	396	int ret;
bd2d0210 TT	397
bd2d0210 TT	398	if (io->io_bio && bh->b_blocknr != io->io_next_block) {
	399	submit_and_retry:
	400	ext4_io_submit(io);
	401	}
	402	if (io->io_bio == NULL) {
97a851ed	403	ret = io_submit_init_bio(io, bh);
bd2d0210 TT	404	if (ret)
	405	return ret;
	406	}
97a851ed JK	407	ret = bio_add_page(io->io_bio, bh->b_page, bh->b_size, bh_offset(bh));
	408	if (ret != bh->b_size)
	409	goto submit_and_retry;
bd2d0210	410	io->io_next_block++;
bd2d0210 TT	411	return 0;
	412	}
	413
	414	int ext4_bio_write_page(struct ext4_io_submit *io,
	415	struct page *page,
	416	int len,
	417	struct writeback_control *wbc)
	418	{
	419	struct inode *inode = page->mapping->host;
0058f965	420	unsigned block_start, blocksize;
bd2d0210 TT	421	struct buffer_head bh, head;
bd2d0210 TT	422	int ret = 0;
0058f965	423	int nr_submitted = 0;
bd2d0210 TT	424
	425	blocksize = 1 << inode->i_blkbits;
	426
d50bdd5a	427	BUG_ON(!PageLocked(page));
bd2d0210	428	BUG_ON(PageWriteback(page));
bd2d0210	429
a54aa761 TT	430	set_page_writeback(page);
a54aa761 TT	431	ClearPageError(page);
bd2d0210	432
0058f965 JK	433	/*
	434	* In the first loop we prepare and mark buffers to submit. We have to
	435	* mark all buffers in the page before submitting so that
	436	* end_page_writeback() cannot be called from ext4_bio_end_io() when IO
	437	* on the first buffer finishes and we are still working on submitting
	438	* the second buffer.
	439	*/
	440	bh = head = page_buffers(page);
	441	do {
	442	block_start = bh_offset(bh);
bd2d0210	443	if (block_start >= len) {
5a0dc736 YY	444	/*
	445	* Comments copied from block_write_full_page_endio:
	446	*
	447	* The page straddles i_size. It must be zeroed out on
	448	* each and every writepage invocation because it may
	449	* be mmapped. "A file is mapped in multiples of the
	450	* page size. For a file that is not a multiple of
	451	* the page size, the remaining memory is zeroed when
	452	* mapped, and writes to that region are not written
	453	* out to the file."
	454	*/
0058f965 JK	455	zero_user_segment(page, block_start,
0058f965 JK	456	block_start + blocksize);
bd2d0210 TT	457	clear_buffer_dirty(bh);
	458	set_buffer_uptodate(bh);
	459	continue;
	460	}
8a850c3f JK	461	if (!buffer_dirty(bh) \|\| buffer_delay(bh) \|\|
	462	!buffer_mapped(bh) \|\| buffer_unwritten(bh)) {
	463	/* A hole? We can safely clear the dirty bit */
	464	if (!buffer_mapped(bh))
	465	clear_buffer_dirty(bh);
	466	if (io->io_bio)
	467	ext4_io_submit(io);
	468	continue;
	469	}
0058f965 JK	470	if (buffer_new(bh)) {
	471	clear_buffer_new(bh);
	472	unmap_underlying_metadata(bh->b_bdev, bh->b_blocknr);
	473	}
	474	set_buffer_async_write(bh);
	475	} while ((bh = bh->b_this_page) != head);
	476
	477	/* Now submit buffers to write */
	478	bh = head = page_buffers(page);
	479	do {
	480	if (!buffer_async_write(bh))
	481	continue;
97a851ed	482	ret = io_submit_add_bh(io, inode, bh);
bd2d0210 TT	483	if (ret) {
	484	/*
	485	* We only get here on ENOMEM. Not much else
	486	* we can do but mark the page as dirty, and
	487	* better luck next time.
	488	*/
1ae48a63	489	redirty_page_for_writepage(wbc, page);
bd2d0210 TT	490	break;
bd2d0210 TT	491	}
0058f965	492	nr_submitted++;
1ae48a63	493	clear_buffer_dirty(bh);
0058f965 JK	494	} while ((bh = bh->b_this_page) != head);
	495
	496	/* Error stopped previous loop? Clean up buffers... */
	497	if (ret) {
	498	do {
	499	clear_buffer_async_write(bh);
	500	bh = bh->b_this_page;
	501	} while (bh != head);
bd2d0210 TT	502	}
bd2d0210 TT	503	unlock_page(page);
0058f965 JK	504	/* Nothing submitted - we have to end page writeback */
	505	if (!nr_submitted)
	506	end_page_writeback(page);
bd2d0210 TT	507	return ret;
bd2d0210 TT	508	}