[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_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_unwritten_work))
		cancel_work_sync(&EXT4_I(inode)->i_unwritten_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);

	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;
	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);
	}
	ext4_clear_io_unwritten_flag(io);
	ext4_release_io_end(io);
	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;

	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. */
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));
	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))
		queue_work(wq, &ei->i_unwritten_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 *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);
	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;
	}
	return ret;
}

/*
 * work on completed aio dio IO, to convert unwritten extents to extents
 */
void ext4_end_io_work(struct work_struct *work)
{
	struct ext4_inode_info *ei = container_of(work, struct ext4_inode_info,
						  i_unwritten_work);
	ext4_do_flush_completed_IO(&ei->vfs_inode);
}

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);
	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->inode,
						io_end->offset, io_end->size);
			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);
	if (!io->io_end->size)
		io->io_end->offset = (bh->b_page->index << PAGE_CACHE_SHIFT)
				     + bh_offset(bh);
	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)
{
	ext4_io_end_t *io_end;
	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_end = io->io_end;
	if (test_clear_buffer_uninit(bh))
		ext4_set_io_unwritten_flag(inode, io_end);
	io_end->size += bh->b_size;
	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>
	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	31
0058f965	32	static struct kmem_cache *io_end_cachep;
bd2d0210	33
5dabfc78	34	int __init ext4_init_pageio(void)
bd2d0210	35	{
bd2d0210	36	io_end_cachep = KMEM_CACHE(ext4_io_end, SLAB_RECLAIM_ACCOUNT);
0058f965	37	if (io_end_cachep == NULL)
bd2d0210	38	return -ENOMEM;
bd2d0210 TT	39	return 0;
	40	}
	41
5dabfc78	42	void ext4_exit_pageio(void)
bd2d0210 TT	43	{
bd2d0210 TT	44	kmem_cache_destroy(io_end_cachep);
bd2d0210 TT	45	}
bd2d0210 TT	46
1ada47d9 TT	47	/*
	48	* This function is called by ext4_evict_inode() to make sure there is
	49	* no more pending I/O completion work left to do.
	50	*/
	51	void ext4_ioend_shutdown(struct inode *inode)
f7ad6d2e	52	{
e9e3bcec	53	wait_queue_head_t *wq = ext4_ioend_wq(inode);
f7ad6d2e TT	54
f7ad6d2e TT	55	wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_ioend_count) == 0));
1ada47d9 TT	56	/*
	57	* We need to make sure the work structure is finished being
	58	* used before we let the inode get destroyed.
	59	*/
	60	if (work_pending(&EXT4_I(inode)->i_unwritten_work))
	61	cancel_work_sync(&EXT4_I(inode)->i_unwritten_work);
f7ad6d2e TT	62	}
f7ad6d2e TT	63
4eec708d	64	static void ext4_release_io_end(ext4_io_end_t *io_end)
bd2d0210	65	{
4eec708d JK	66	BUG_ON(!list_empty(&io_end->list));
	67	BUG_ON(io_end->flag & EXT4_IO_END_UNWRITTEN);
	68
	69	if (atomic_dec_and_test(&EXT4_I(io_end->inode)->i_ioend_count))
	70	wake_up_all(ext4_ioend_wq(io_end->inode));
	71	if (io_end->flag & EXT4_IO_END_DIRECT)
	72	inode_dio_done(io_end->inode);
	73	if (io_end->iocb)
	74	aio_complete(io_end->iocb, io_end->result, 0);
	75	kmem_cache_free(io_end_cachep, io_end);
	76	}
bd2d0210	77
4eec708d JK	78	static void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
	79	{
	80	struct inode *inode = io_end->inode;
82e54229	81
4eec708d JK	82	io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
	83	/* Wake up anyone waiting on unwritten extent conversion */
	84	if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
	85	wake_up_all(ext4_ioend_wq(inode));
bd2d0210 TT	86	}
bd2d0210 TT	87
28a535f9 DM	88	/* check a range of space and convert unwritten extents to written. */
28a535f9 DM	89	static int ext4_end_io(ext4_io_end_t *io)
bd2d0210 TT	90	{
	91	struct inode *inode = io->inode;
	92	loff_t offset = io->offset;
	93	ssize_t size = io->size;
	94	int ret = 0;
	95
	96	ext4_debug("ext4_end_io_nolock: io 0x%p from inode %lu,list->next 0x%p,"
	97	"list->prev 0x%p\n",
	98	io, inode->i_ino, io->list.next, io->list.prev);
	99
bd2d0210 TT	100	ret = ext4_convert_unwritten_extents(inode, offset, size);
bd2d0210 TT	101	if (ret < 0) {
b82e384c TT	102	ext4_msg(inode->i_sb, KERN_EMERG,
	103	"failed to convert unwritten extents to written "
	104	"extents -- potential data loss! "
	105	"(inode %lu, offset %llu, size %zd, error %d)",
	106	inode->i_ino, offset, size, ret);
bd2d0210	107	}
4eec708d JK	108	ext4_clear_io_unwritten_flag(io);
4eec708d JK	109	ext4_release_io_end(io);
bd2d0210 TT	110	return ret;
	111	}
	112
28a535f9 DM	113	static void dump_completed_IO(struct inode *inode)
	114	{
	115	#ifdef EXT4FS_DEBUG
	116	struct list_head cur, before, *after;
	117	ext4_io_end_t io, io0, *io1;
28a535f9 DM	118
	119	if (list_empty(&EXT4_I(inode)->i_completed_io_list)) {
	120	ext4_debug("inode %lu completed_io list is empty\n",
	121	inode->i_ino);
	122	return;
	123	}
	124
	125	ext4_debug("Dump inode %lu completed_io list\n", inode->i_ino);
	126	list_for_each_entry(io, &EXT4_I(inode)->i_completed_io_list, list) {
	127	cur = &io->list;
	128	before = cur->prev;
	129	io0 = container_of(before, ext4_io_end_t, list);
	130	after = cur->next;
	131	io1 = container_of(after, ext4_io_end_t, list);
	132
	133	ext4_debug("io 0x%p from inode %lu,prev 0x%p,next 0x%p\n",
	134	io, inode->i_ino, io0, io1);
	135	}
	136	#endif
	137	}
	138
	139	/* Add the io_end to per-inode completed end_io list. */
4eec708d	140	static void ext4_add_complete_io(ext4_io_end_t *io_end)
bd2d0210	141	{
28a535f9 DM	142	struct ext4_inode_info *ei = EXT4_I(io_end->inode);
	143	struct workqueue_struct *wq;
	144	unsigned long flags;
	145
	146	BUG_ON(!(io_end->flag & EXT4_IO_END_UNWRITTEN));
	147	wq = EXT4_SB(io_end->inode->i_sb)->dio_unwritten_wq;
bd2d0210	148
d73d5046	149	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
84c17543 JK	150	if (list_empty(&ei->i_completed_io_list))
84c17543 JK	151	queue_work(wq, &ei->i_unwritten_work);
28a535f9 DM	152	list_add_tail(&io_end->list, &ei->i_completed_io_list);
	153	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
	154	}
d73d5046	155
84c17543	156	static int ext4_do_flush_completed_IO(struct inode *inode)
28a535f9 DM	157	{
28a535f9 DM	158	ext4_io_end_t *io;
002bd7fa	159	struct list_head unwritten;
28a535f9 DM	160	unsigned long flags;
	161	struct ext4_inode_info *ei = EXT4_I(inode);
	162	int err, ret = 0;
	163
28a535f9 DM	164	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	165	dump_completed_IO(inode);
	166	list_replace_init(&ei->i_completed_io_list, &unwritten);
	167	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
	168
	169	while (!list_empty(&unwritten)) {
	170	io = list_entry(unwritten.next, ext4_io_end_t, list);
	171	BUG_ON(!(io->flag & EXT4_IO_END_UNWRITTEN));
	172	list_del_init(&io->list);
	173
	174	err = ext4_end_io(io);
	175	if (unlikely(!ret && err))
	176	ret = err;
28a535f9 DM	177	}
	178	return ret;
	179	}
	180
	181	/*
	182	* work on completed aio dio IO, to convert unwritten extents to extents
	183	*/
84c17543	184	void ext4_end_io_work(struct work_struct *work)
28a535f9	185	{
84c17543 JK	186	struct ext4_inode_info *ei = container_of(work, struct ext4_inode_info,
	187	i_unwritten_work);
	188	ext4_do_flush_completed_IO(&ei->vfs_inode);
28a535f9 DM	189	}
28a535f9 DM	190
c278531d	191	int ext4_flush_unwritten_io(struct inode *inode)
28a535f9	192	{
c278531d DM	193	int ret;
	194	WARN_ON_ONCE(!mutex_is_locked(&inode->i_mutex) &&
	195	!(inode->i_state & I_FREEING));
84c17543	196	ret = ext4_do_flush_completed_IO(inode);
c278531d DM	197	ext4_unwritten_wait(inode);
c278531d DM	198	return ret;
bd2d0210 TT	199	}
	200
	201	ext4_io_end_t ext4_init_io_end(struct inode inode, gfp_t flags)
	202	{
b17b35ec	203	ext4_io_end_t *io = kmem_cache_zalloc(io_end_cachep, flags);
bd2d0210	204	if (io) {
f7ad6d2e TT	205	atomic_inc(&EXT4_I(inode)->i_ioend_count);
f7ad6d2e TT	206	io->inode = inode;
bd2d0210	207	INIT_LIST_HEAD(&io->list);
4eec708d	208	atomic_set(&io->count, 1);
bd2d0210 TT	209	}
	210	return io;
	211	}
	212
4eec708d JK	213	void ext4_put_io_end_defer(ext4_io_end_t *io_end)
	214	{
	215	if (atomic_dec_and_test(&io_end->count)) {
	216	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN) \|\| !io_end->size) {
	217	ext4_release_io_end(io_end);
	218	return;
	219	}
	220	ext4_add_complete_io(io_end);
	221	}
	222	}
	223
	224	int ext4_put_io_end(ext4_io_end_t *io_end)
	225	{
	226	int err = 0;
	227
	228	if (atomic_dec_and_test(&io_end->count)) {
	229	if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
	230	err = ext4_convert_unwritten_extents(io_end->inode,
	231	io_end->offset, io_end->size);
	232	ext4_clear_io_unwritten_flag(io_end);
	233	}
	234	ext4_release_io_end(io_end);
	235	}
	236	return err;
	237	}
	238
	239	ext4_io_end_t ext4_get_io_end(ext4_io_end_t io_end)
	240	{
	241	atomic_inc(&io_end->count);
	242	return io_end;
	243	}
	244
bd2d0210 TT	245	/*
	246	* Print an buffer I/O error compatible with the fs/buffer.c. This
	247	* provides compatibility with dmesg scrapers that look for a specific
	248	* buffer I/O error message. We really need a unified error reporting
	249	* structure to userspace ala Digital Unix's uerf system, but it's
	250	* probably not going to happen in my lifetime, due to LKML politics...
	251	*/
	252	static void buffer_io_error(struct buffer_head *bh)
	253	{
	254	char b[BDEVNAME_SIZE];
	255	printk(KERN_ERR "Buffer I/O error on device %s, logical block %llu\n",
	256	bdevname(bh->b_bdev, b),
	257	(unsigned long long)bh->b_blocknr);
	258	}
	259
	260	static void ext4_end_bio(struct bio *bio, int error)
	261	{
	262	ext4_io_end_t *io_end = bio->bi_private;
bd2d0210	263	struct inode *inode;
bd2d0210	264	int i;
0058f965	265	int blocksize;
d50bdd5a	266	sector_t bi_sector = bio->bi_sector;
bd2d0210 TT	267
bd2d0210 TT	268	BUG_ON(!io_end);
0058f965 JK	269	inode = io_end->inode;
0058f965 JK	270	blocksize = 1 << inode->i_blkbits;
bd2d0210 TT	271	bio->bi_private = NULL;
	272	bio->bi_end_io = NULL;
	273	if (test_bit(BIO_UPTODATE, &bio->bi_flags))
	274	error = 0;
0058f965 JK	275	for (i = 0; i < bio->bi_vcnt; i++) {
	276	struct bio_vec *bvec = &bio->bi_io_vec[i];
	277	struct page *page = bvec->bv_page;
bd2d0210	278	struct buffer_head bh, head;
0058f965 JK	279	unsigned bio_start = bvec->bv_offset;
	280	unsigned bio_end = bio_start + bvec->bv_len;
	281	unsigned under_io = 0;
	282	unsigned long flags;
	283
	284	if (!page)
	285	continue;
bd2d0210	286
39db00f1	287	if (error) {
bd2d0210	288	SetPageError(page);
39db00f1	289	set_bit(AS_EIO, &page->mapping->flags);
bd2d0210	290	}
0058f965 JK	291	bh = head = page_buffers(page);
	292	/*
	293	* We check all buffers in the page under BH_Uptodate_Lock
	294	* to avoid races with other end io clearing async_write flags
	295	*/
	296	local_irq_save(flags);
	297	bit_spin_lock(BH_Uptodate_Lock, &head->b_state);
	298	do {
	299	if (bh_offset(bh) < bio_start \|\|
	300	bh_offset(bh) + blocksize > bio_end) {
	301	if (buffer_async_write(bh))
	302	under_io++;
	303	continue;
	304	}
	305	clear_buffer_async_write(bh);
	306	if (error)
	307	buffer_io_error(bh);
	308	} while ((bh = bh->b_this_page) != head);
	309	bit_spin_unlock(BH_Uptodate_Lock, &head->b_state);
	310	local_irq_restore(flags);
	311	if (!under_io)
	312	end_page_writeback(page);
bd2d0210	313	}
0058f965	314	bio_put(bio);
f7ad6d2e TT	315
	316	if (error) {
	317	io_end->flag \|= EXT4_IO_END_ERROR;
	318	ext4_warning(inode->i_sb, "I/O error writing to inode %lu "
	319	"(offset %llu size %ld starting block %llu)",
	320	inode->i_ino,
	321	(unsigned long long) io_end->offset,
	322	(long) io_end->size,
	323	(unsigned long long)
d50bdd5a	324	bi_sector >> (inode->i_blkbits - 9));
f7ad6d2e	325	}
bd2d0210	326
4eec708d	327	ext4_put_io_end_defer(io_end);
bd2d0210 TT	328	}
	329
	330	void ext4_io_submit(struct ext4_io_submit *io)
	331	{
	332	struct bio *bio = io->io_bio;
	333
	334	if (bio) {
	335	bio_get(io->io_bio);
	336	submit_bio(io->io_op, io->io_bio);
	337	BUG_ON(bio_flagged(io->io_bio, BIO_EOPNOTSUPP));
	338	bio_put(io->io_bio);
	339	}
7dc57615	340	io->io_bio = NULL;
4eec708d JK	341	}
	342
	343	void ext4_io_submit_init(struct ext4_io_submit *io,
	344	struct writeback_control *wbc)
	345	{
	346	io->io_op = (wbc->sync_mode == WB_SYNC_ALL ? WRITE_SYNC : WRITE);
	347	io->io_bio = NULL;
7dc57615	348	io->io_end = NULL;
bd2d0210 TT	349	}
bd2d0210 TT	350
4eec708d JK	351	static int io_submit_init_bio(struct ext4_io_submit *io,
4eec708d JK	352	struct buffer_head *bh)
bd2d0210	353	{
bd2d0210 TT	354	int nvecs = bio_get_nr_vecs(bh->b_bdev);
	355	struct bio *bio;
	356
275d3ba6	357	bio = bio_alloc(GFP_NOIO, min(nvecs, BIO_MAX_PAGES));
bd2d0210 TT	358	bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
bd2d0210 TT	359	bio->bi_bdev = bh->b_bdev;
bd2d0210	360	bio->bi_end_io = ext4_end_bio;
4eec708d JK	361	bio->bi_private = ext4_get_io_end(io->io_end);
	362	if (!io->io_end->size)
	363	io->io_end->offset = (bh->b_page->index << PAGE_CACHE_SHIFT)
	364	+ bh_offset(bh);
bd2d0210	365	io->io_bio = bio;
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,
bd2d0210	371	struct inode *inode,
bd2d0210 TT	372	struct buffer_head *bh)
	373	{
	374	ext4_io_end_t *io_end;
	375	int ret;
	376
bd2d0210 TT	377	if (io->io_bio && bh->b_blocknr != io->io_next_block) {
	378	submit_and_retry:
	379	ext4_io_submit(io);
	380	}
	381	if (io->io_bio == NULL) {
4eec708d	382	ret = io_submit_init_bio(io, bh);
bd2d0210 TT	383	if (ret)
	384	return ret;
	385	}
4eec708d JK	386	ret = bio_add_page(io->io_bio, bh->b_page, bh->b_size, bh_offset(bh));
	387	if (ret != bh->b_size)
	388	goto submit_and_retry;
bd2d0210	389	io_end = io->io_end;
7b001d6a	390	if (test_clear_buffer_uninit(bh))
0edeb71d	391	ext4_set_io_unwritten_flag(inode, io_end);
4eec708d	392	io_end->size += bh->b_size;
bd2d0210	393	io->io_next_block++;
bd2d0210 TT	394	return 0;
	395	}
	396
	397	int ext4_bio_write_page(struct ext4_io_submit *io,
	398	struct page *page,
	399	int len,
	400	struct writeback_control *wbc)
	401	{
	402	struct inode *inode = page->mapping->host;
0058f965	403	unsigned block_start, blocksize;
bd2d0210 TT	404	struct buffer_head bh, head;
bd2d0210 TT	405	int ret = 0;
0058f965	406	int nr_submitted = 0;
bd2d0210 TT	407
	408	blocksize = 1 << inode->i_blkbits;
	409
d50bdd5a	410	BUG_ON(!PageLocked(page));
bd2d0210	411	BUG_ON(PageWriteback(page));
bd2d0210	412
a54aa761 TT	413	set_page_writeback(page);
a54aa761 TT	414	ClearPageError(page);
bd2d0210	415
0058f965 JK	416	/*
	417	* In the first loop we prepare and mark buffers to submit. We have to
	418	* mark all buffers in the page before submitting so that
	419	* end_page_writeback() cannot be called from ext4_bio_end_io() when IO
	420	* on the first buffer finishes and we are still working on submitting
	421	* the second buffer.
	422	*/
	423	bh = head = page_buffers(page);
	424	do {
	425	block_start = bh_offset(bh);
bd2d0210	426	if (block_start >= len) {
5a0dc736 YY	427	/*
	428	* Comments copied from block_write_full_page_endio:
	429	*
	430	* The page straddles i_size. It must be zeroed out on
	431	* each and every writepage invocation because it may
	432	* be mmapped. "A file is mapped in multiples of the
	433	* page size. For a file that is not a multiple of
	434	* the page size, the remaining memory is zeroed when
	435	* mapped, and writes to that region are not written
	436	* out to the file."
	437	*/
0058f965 JK	438	zero_user_segment(page, block_start,
0058f965 JK	439	block_start + blocksize);
bd2d0210 TT	440	clear_buffer_dirty(bh);
	441	set_buffer_uptodate(bh);
	442	continue;
	443	}
8a850c3f JK	444	if (!buffer_dirty(bh) \|\| buffer_delay(bh) \|\|
	445	!buffer_mapped(bh) \|\| buffer_unwritten(bh)) {
	446	/* A hole? We can safely clear the dirty bit */
	447	if (!buffer_mapped(bh))
	448	clear_buffer_dirty(bh);
	449	if (io->io_bio)
	450	ext4_io_submit(io);
	451	continue;
	452	}
0058f965 JK	453	if (buffer_new(bh)) {
	454	clear_buffer_new(bh);
	455	unmap_underlying_metadata(bh->b_bdev, bh->b_blocknr);
	456	}
	457	set_buffer_async_write(bh);
	458	} while ((bh = bh->b_this_page) != head);
	459
	460	/* Now submit buffers to write */
	461	bh = head = page_buffers(page);
	462	do {
	463	if (!buffer_async_write(bh))
	464	continue;
4eec708d	465	ret = io_submit_add_bh(io, inode, bh);
bd2d0210 TT	466	if (ret) {
	467	/*
	468	* We only get here on ENOMEM. Not much else
	469	* we can do but mark the page as dirty, and
	470	* better luck next time.
	471	*/
1ae48a63	472	redirty_page_for_writepage(wbc, page);
bd2d0210 TT	473	break;
bd2d0210 TT	474	}
0058f965	475	nr_submitted++;
1ae48a63	476	clear_buffer_dirty(bh);
0058f965 JK	477	} while ((bh = bh->b_this_page) != head);
	478
	479	/* Error stopped previous loop? Clean up buffers... */
	480	if (ret) {
	481	do {
	482	clear_buffer_async_write(bh);
	483	bh = bh->b_this_page;
	484	} while (bh != head);
bd2d0210 TT	485	}
bd2d0210 TT	486	unlock_page(page);
0058f965 JK	487	/* Nothing submitted - we have to end page writeback */
	488	if (!nr_submitted)
	489	end_page_writeback(page);
bd2d0210 TT	490	return ret;
bd2d0210 TT	491	}