[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/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 <linux/backing-dev.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);
}

/*
 * 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)
{
	printk_ratelimited(KERN_ERR "Buffer I/O error on device %pg, logical block %llu\n",
		       bh->b_bdev,
			(unsigned long long)bh->b_blocknr);
}

static void ext4_finish_bio(struct bio *bio)
{
	int i;
	struct bio_vec *bvec;

	bio_for_each_segment_all(bvec, bio, i) {
		struct page *page = bvec->bv_page;
#ifdef CONFIG_EXT4_FS_ENCRYPTION
		struct page *data_page = NULL;
#endif
		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;

#ifdef CONFIG_EXT4_FS_ENCRYPTION
		if (!page->mapping) {
			/* The bounce data pages are unmapped. */
			data_page = page;
			fscrypt_pullback_bio_page(&page, false);
		}
#endif

		if (bio->bi_error) {
			SetPageError(page);
			mapping_set_error(page->mapping, -EIO);
		}
		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) + bh->b_size > bio_end) {
				if (buffer_async_write(bh))
					under_io++;
				continue;
			}
			clear_buffer_async_write(bh);
			if (bio->bi_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) {
#ifdef CONFIG_EXT4_FS_ENCRYPTION
			if (data_page)
				fscrypt_restore_control_page(data_page);
#endif
			end_page_writeback(page);
		}
	}
}

static void ext4_release_io_end(ext4_io_end_t *io_end)
{
	struct bio *bio, *next_bio;

	BUG_ON(!list_empty(&io_end->list));
	BUG_ON(io_end->flag & EXT4_IO_END_UNWRITTEN);
	WARN_ON(io_end->handle);

	for (bio = io_end->bio; bio; bio = next_bio) {
		next_bio = bio->bi_private;
		ext4_finish_bio(bio);
		bio_put(bio);
	}
	kmem_cache_free(io_end_cachep, io_end);
}

/*
 * Check a range of space and convert unwritten extents to written. Note that
 * we are protected from truncate touching same part of extent tree by the
 * fact that truncate code waits for all DIO to finish (thus exclusion from
 * direct IO is achieved) and also waits for PageWriteback bits. Thus we
 * cannot get to ext4_ext_truncate() before all IOs overlapping that range are
 * completed (happens from ext4_free_ioend()).
 */
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_forced_shutdown(EXT4_SB(inode->i_sb))) {
		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 ext4_sb_info *sbi = EXT4_SB(io_end->inode->i_sb);
	struct workqueue_struct *wq;
	unsigned long flags;

	/* Only reserved conversions from writeback should enter here */
	WARN_ON(!(io_end->flag & EXT4_IO_END_UNWRITTEN));
	WARN_ON(!io_end->handle && sbi->s_journal);
	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
	wq = sbi->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);
	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);
}

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) {
		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;
}

/* BIO completion function for page writeback */
static void ext4_end_bio(struct bio *bio)
{
	ext4_io_end_t *io_end = bio->bi_private;
	sector_t bi_sector = bio->bi_iter.bi_sector;

	BUG_ON(!io_end);
	bio->bi_end_io = NULL;

	if (bio->bi_error) {
		struct inode *inode = io_end->inode;

		ext4_warning(inode->i_sb, "I/O error %d writing to inode %lu "
			     "(offset %llu size %ld starting block %llu)",
			     bio->bi_error, inode->i_ino,
			     (unsigned long long) io_end->offset,
			     (long) io_end->size,
			     (unsigned long long)
			     bi_sector >> (inode->i_blkbits - 9));
		mapping_set_error(inode->i_mapping, bio->bi_error);
	}

	if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
		/*
		 * Link bio into list hanging from io_end. We have to do it
		 * atomically as bio completions can be racing against each
		 * other.
		 */
		bio->bi_private = xchg(&io_end->bio, bio);
		ext4_put_io_end_defer(io_end);
	} else {
		/*
		 * Drop io_end reference early. Inode can get freed once
		 * we finish the bio.
		 */
		ext4_put_io_end_defer(io_end);
		ext4_finish_bio(bio);
		bio_put(bio);
	}
}

void ext4_io_submit(struct ext4_io_submit *io)
{
	struct bio *bio = io->io_bio;

	if (bio) {
		int io_op_flags = io->io_wbc->sync_mode == WB_SYNC_ALL ?
				  REQ_SYNC : 0;
		bio_set_op_attrs(io->io_bio, REQ_OP_WRITE, io_op_flags);
		submit_bio(io->io_bio);
	}
	io->io_bio = NULL;
}

void ext4_io_submit_init(struct ext4_io_submit *io,
			 struct writeback_control *wbc)
{
	io->io_wbc = wbc;
	io->io_bio = NULL;
	io->io_end = NULL;
}

static int io_submit_init_bio(struct ext4_io_submit *io,
			      struct buffer_head *bh)
{
	struct bio *bio;

	bio = bio_alloc(GFP_NOIO, BIO_MAX_PAGES);
	if (!bio)
		return -ENOMEM;
	wbc_init_bio(io->io_wbc, bio);
	bio->bi_iter.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 page *page,
			    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, page, bh->b_size, bh_offset(bh));
	if (ret != bh->b_size)
		goto submit_and_retry;
	wbc_account_io(io->io_wbc, page, 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,
			bool keep_towrite)
{
	struct page *data_page = NULL;
	struct inode *inode = page->mapping->host;
	unsigned block_start;
	struct buffer_head *bh, *head;
	int ret = 0;
	int nr_submitted = 0;
	int nr_to_submit = 0;

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

	if (keep_towrite)
		set_page_writeback_keepwrite(page);
	else
		set_page_writeback(page);
	ClearPageError(page);

	/*
	 * Comments copied from block_write_full_page:
	 *
	 * 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."
	 */
	if (len < PAGE_SIZE)
		zero_user_segment(page, len, PAGE_SIZE);
	/*
	 * 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) {
			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);
			clean_bdev_bh_alias(bh);
		}
		set_buffer_async_write(bh);
		nr_to_submit++;
	} while ((bh = bh->b_this_page) != head);

	bh = head = page_buffers(page);

	if (ext4_encrypted_inode(inode) && S_ISREG(inode->i_mode) &&
	    nr_to_submit) {
		gfp_t gfp_flags = GFP_NOFS;

	retry_encrypt:
		data_page = fscrypt_encrypt_page(inode, page, PAGE_SIZE, 0,
						page->index, gfp_flags);
		if (IS_ERR(data_page)) {
			ret = PTR_ERR(data_page);
			if (ret == -ENOMEM && wbc->sync_mode == WB_SYNC_ALL) {
				if (io->io_bio) {
					ext4_io_submit(io);
					congestion_wait(BLK_RW_ASYNC, HZ/50);
				}
				gfp_flags |= __GFP_NOFAIL;
				goto retry_encrypt;
			}
			data_page = NULL;
			goto out;
		}
	}

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

	/* Error stopped previous loop? Clean up buffers... */
	if (ret) {
	out:
		if (data_page)
			fscrypt_restore_control_page(data_page);
		printk_ratelimited(KERN_ERR "%s: ret = %d\n", __func__, ret);
		redirty_page_for_writepage(wbc, page);
		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>
bd2d0210 TT	10	#include <linux/time.h>
bd2d0210 TT	11	#include <linux/highuid.h>
	12	#include <linux/pagemap.h>
	13	#include <linux/quotaops.h>
	14	#include <linux/string.h>
	15	#include <linux/buffer_head.h>
	16	#include <linux/writeback.h>
	17	#include <linux/pagevec.h>
	18	#include <linux/mpage.h>
	19	#include <linux/namei.h>
	20	#include <linux/uio.h>
	21	#include <linux/bio.h>
	22	#include <linux/workqueue.h>
	23	#include <linux/kernel.h>
	24	#include <linux/slab.h>
1ae48a63	25	#include <linux/mm.h>
c9af28fd	26	#include <linux/backing-dev.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
b0857d30 JK	47	/*
	48	* Print an buffer I/O error compatible with the fs/buffer.c. This
	49	* provides compatibility with dmesg scrapers that look for a specific
	50	* buffer I/O error message. We really need a unified error reporting
	51	* structure to userspace ala Digital Unix's uerf system, but it's
	52	* probably not going to happen in my lifetime, due to LKML politics...
	53	*/
	54	static void buffer_io_error(struct buffer_head *bh)
	55	{
a1c6f057 DM	56	printk_ratelimited(KERN_ERR "Buffer I/O error on device %pg, logical block %llu\n",
a1c6f057 DM	57	bh->b_bdev,
b0857d30 JK	58	(unsigned long long)bh->b_blocknr);
	59	}
	60
	61	static void ext4_finish_bio(struct bio *bio)
	62	{
	63	int i;
2c30c71b	64	struct bio_vec *bvec;
b0857d30	65
2c30c71b	66	bio_for_each_segment_all(bvec, bio, i) {
b0857d30	67	struct page *page = bvec->bv_page;
2058f83a MH	68	#ifdef CONFIG_EXT4_FS_ENCRYPTION
2058f83a MH	69	struct page *data_page = NULL;
2058f83a	70	#endif
b0857d30 JK	71	struct buffer_head bh, head;
	72	unsigned bio_start = bvec->bv_offset;
	73	unsigned bio_end = bio_start + bvec->bv_len;
	74	unsigned under_io = 0;
	75	unsigned long flags;
	76
	77	if (!page)
	78	continue;
	79
2058f83a MH	80	#ifdef CONFIG_EXT4_FS_ENCRYPTION
	81	if (!page->mapping) {
	82	/* The bounce data pages are unmapped. */
	83	data_page = page;
a7550b30	84	fscrypt_pullback_bio_page(&page, false);
2058f83a MH	85	}
	86	#endif
	87
4246a0b6	88	if (bio->bi_error) {
b0857d30	89	SetPageError(page);
5114a97a	90	mapping_set_error(page->mapping, -EIO);
b0857d30 JK	91	}
	92	bh = head = page_buffers(page);
	93	/*
	94	* We check all buffers in the page under BH_Uptodate_Lock
	95	* to avoid races with other end io clearing async_write flags
	96	*/
	97	local_irq_save(flags);
	98	bit_spin_lock(BH_Uptodate_Lock, &head->b_state);
	99	do {
	100	if (bh_offset(bh) < bio_start \|\|
	101	bh_offset(bh) + bh->b_size > bio_end) {
	102	if (buffer_async_write(bh))
	103	under_io++;
	104	continue;
	105	}
	106	clear_buffer_async_write(bh);
4246a0b6	107	if (bio->bi_error)
b0857d30 JK	108	buffer_io_error(bh);
	109	} while ((bh = bh->b_this_page) != head);
	110	bit_spin_unlock(BH_Uptodate_Lock, &head->b_state);
	111	local_irq_restore(flags);
2058f83a MH	112	if (!under_io) {
2058f83a MH	113	#ifdef CONFIG_EXT4_FS_ENCRYPTION
a7550b30 JK	114	if (data_page)
a7550b30 JK	115	fscrypt_restore_control_page(data_page);
2058f83a	116	#endif
b0857d30	117	end_page_writeback(page);
2058f83a	118	}
b0857d30 JK	119	}
	120	}
	121
97a851ed	122	static void ext4_release_io_end(ext4_io_end_t *io_end)
bd2d0210	123	{
b0857d30 JK	124	struct bio bio, next_bio;
b0857d30 JK	125
97a851ed JK	126	BUG_ON(!list_empty(&io_end->list));
97a851ed JK	127	BUG_ON(io_end->flag & EXT4_IO_END_UNWRITTEN);
6b523df4	128	WARN_ON(io_end->handle);
97a851ed	129
b0857d30 JK	130	for (bio = io_end->bio; bio; bio = next_bio) {
	131	next_bio = bio->bi_private;
	132	ext4_finish_bio(bio);
	133	bio_put(bio);
	134	}
97a851ed JK	135	kmem_cache_free(io_end_cachep, io_end);
	136	}
	137
a115f749 JK	138	/*
	139	* Check a range of space and convert unwritten extents to written. Note that
	140	* we are protected from truncate touching same part of extent tree by the
	141	* fact that truncate code waits for all DIO to finish (thus exclusion from
	142	* direct IO is achieved) and also waits for PageWriteback bits. Thus we
	143	* cannot get to ext4_ext_truncate() before all IOs overlapping that range are
	144	* completed (happens from ext4_free_ioend()).
	145	*/
28a535f9	146	static int ext4_end_io(ext4_io_end_t *io)
bd2d0210 TT	147	{
	148	struct inode *inode = io->inode;
	149	loff_t offset = io->offset;
	150	ssize_t size = io->size;
6b523df4	151	handle_t *handle = io->handle;
bd2d0210 TT	152	int ret = 0;
	153
	154	ext4_debug("ext4_end_io_nolock: io 0x%p from inode %lu,list->next 0x%p,"
	155	"list->prev 0x%p\n",
	156	io, inode->i_ino, io->list.next, io->list.prev);
	157
6b523df4 JK	158	io->handle = NULL; /* Following call will use up the handle */
6b523df4 JK	159	ret = ext4_convert_unwritten_extents(handle, inode, offset, size);
0db1ff22	160	if (ret < 0 && !ext4_forced_shutdown(EXT4_SB(inode->i_sb))) {
b82e384c TT	161	ext4_msg(inode->i_sb, KERN_EMERG,
	162	"failed to convert unwritten extents to written "
	163	"extents -- potential data loss! "
	164	"(inode %lu, offset %llu, size %zd, error %d)",
	165	inode->i_ino, offset, size, ret);
bd2d0210	166	}
97a851ed JK	167	ext4_clear_io_unwritten_flag(io);
97a851ed JK	168	ext4_release_io_end(io);
bd2d0210 TT	169	return ret;
	170	}
	171
2e8fa54e	172	static void dump_completed_IO(struct inode inode, struct list_head head)
28a535f9 DM	173	{
	174	#ifdef EXT4FS_DEBUG
	175	struct list_head cur, before, *after;
	176	ext4_io_end_t io, io0, *io1;
28a535f9	177
2e8fa54e	178	if (list_empty(head))
28a535f9	179	return;
28a535f9	180
2e8fa54e JK	181	ext4_debug("Dump inode %lu completed io list\n", inode->i_ino);
2e8fa54e JK	182	list_for_each_entry(io, head, list) {
28a535f9 DM	183	cur = &io->list;
	184	before = cur->prev;
	185	io0 = container_of(before, ext4_io_end_t, list);
	186	after = cur->next;
	187	io1 = container_of(after, ext4_io_end_t, list);
	188
	189	ext4_debug("io 0x%p from inode %lu,prev 0x%p,next 0x%p\n",
	190	io, inode->i_ino, io0, io1);
	191	}
	192	#endif
	193	}
	194
	195	/* Add the io_end to per-inode completed end_io list. */
97a851ed	196	static void ext4_add_complete_io(ext4_io_end_t *io_end)
bd2d0210	197	{
28a535f9	198	struct ext4_inode_info *ei = EXT4_I(io_end->inode);
78371a45	199	struct ext4_sb_info *sbi = EXT4_SB(io_end->inode->i_sb);
28a535f9 DM	200	struct workqueue_struct *wq;
	201	unsigned long flags;
	202
7b7a8665 CH	203	/* Only reserved conversions from writeback should enter here */
7b7a8665 CH	204	WARN_ON(!(io_end->flag & EXT4_IO_END_UNWRITTEN));
78371a45	205	WARN_ON(!io_end->handle && sbi->s_journal);
d73d5046	206	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
78371a45	207	wq = sbi->rsv_conversion_wq;
7b7a8665 CH	208	if (list_empty(&ei->i_rsv_conversion_list))
	209	queue_work(wq, &ei->i_rsv_conversion_work);
	210	list_add_tail(&io_end->list, &ei->i_rsv_conversion_list);
28a535f9 DM	211	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
28a535f9 DM	212	}
d73d5046	213
2e8fa54e JK	214	static int ext4_do_flush_completed_IO(struct inode *inode,
2e8fa54e JK	215	struct list_head *head)
28a535f9 DM	216	{
28a535f9 DM	217	ext4_io_end_t *io;
002bd7fa	218	struct list_head unwritten;
28a535f9 DM	219	unsigned long flags;
	220	struct ext4_inode_info *ei = EXT4_I(inode);
	221	int err, ret = 0;
	222
28a535f9	223	spin_lock_irqsave(&ei->i_completed_io_lock, flags);
2e8fa54e JK	224	dump_completed_IO(inode, head);
2e8fa54e JK	225	list_replace_init(head, &unwritten);
28a535f9 DM	226	spin_unlock_irqrestore(&ei->i_completed_io_lock, flags);
	227
	228	while (!list_empty(&unwritten)) {
	229	io = list_entry(unwritten.next, ext4_io_end_t, list);
	230	BUG_ON(!(io->flag & EXT4_IO_END_UNWRITTEN));
	231	list_del_init(&io->list);
	232
	233	err = ext4_end_io(io);
	234	if (unlikely(!ret && err))
	235	ret = err;
28a535f9 DM	236	}
	237	return ret;
	238	}
	239
	240	/*
2e8fa54e	241	* work on completed IO, to convert unwritten extents to extents
28a535f9	242	*/
2e8fa54e JK	243	void ext4_end_io_rsv_work(struct work_struct *work)
	244	{
	245	struct ext4_inode_info *ei = container_of(work, struct ext4_inode_info,
	246	i_rsv_conversion_work);
	247	ext4_do_flush_completed_IO(&ei->vfs_inode, &ei->i_rsv_conversion_list);
	248	}
	249
bd2d0210 TT	250	ext4_io_end_t ext4_init_io_end(struct inode inode, gfp_t flags)
bd2d0210 TT	251	{
b17b35ec	252	ext4_io_end_t *io = kmem_cache_zalloc(io_end_cachep, flags);
bd2d0210	253	if (io) {
f7ad6d2e	254	io->inode = inode;
bd2d0210	255	INIT_LIST_HEAD(&io->list);
97a851ed	256	atomic_set(&io->count, 1);
bd2d0210 TT	257	}
	258	return io;
	259	}
	260
97a851ed JK	261	void ext4_put_io_end_defer(ext4_io_end_t *io_end)
	262	{
	263	if (atomic_dec_and_test(&io_end->count)) {
	264	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN) \|\| !io_end->size) {
	265	ext4_release_io_end(io_end);
	266	return;
	267	}
	268	ext4_add_complete_io(io_end);
	269	}
	270	}
	271
	272	int ext4_put_io_end(ext4_io_end_t *io_end)
	273	{
	274	int err = 0;
	275
	276	if (atomic_dec_and_test(&io_end->count)) {
	277	if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
6b523df4 JK	278	err = ext4_convert_unwritten_extents(io_end->handle,
	279	io_end->inode, io_end->offset,
	280	io_end->size);
	281	io_end->handle = NULL;
97a851ed JK	282	ext4_clear_io_unwritten_flag(io_end);
	283	}
	284	ext4_release_io_end(io_end);
	285	}
	286	return err;
	287	}
	288
	289	ext4_io_end_t ext4_get_io_end(ext4_io_end_t io_end)
	290	{
	291	atomic_inc(&io_end->count);
	292	return io_end;
	293	}
	294
822dbba3	295	/* BIO completion function for page writeback */
4246a0b6	296	static void ext4_end_bio(struct bio *bio)
bd2d0210 TT	297	{
bd2d0210 TT	298	ext4_io_end_t *io_end = bio->bi_private;
4f024f37	299	sector_t bi_sector = bio->bi_iter.bi_sector;
bd2d0210 TT	300
bd2d0210 TT	301	BUG_ON(!io_end);
bd2d0210	302	bio->bi_end_io = NULL;
0058f965	303
4246a0b6	304	if (bio->bi_error) {
b0857d30 JK	305	struct inode *inode = io_end->inode;
b0857d30 JK	306
9503c67c	307	ext4_warning(inode->i_sb, "I/O error %d writing to inode %lu "
f7ad6d2e	308	"(offset %llu size %ld starting block %llu)",
4246a0b6	309	bio->bi_error, inode->i_ino,
f7ad6d2e TT	310	(unsigned long long) io_end->offset,
	311	(long) io_end->size,
	312	(unsigned long long)
d50bdd5a	313	bi_sector >> (inode->i_blkbits - 9));
4246a0b6	314	mapping_set_error(inode->i_mapping, bio->bi_error);
f7ad6d2e	315	}
822dbba3 JK	316
	317	if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
	318	/*
	319	* Link bio into list hanging from io_end. We have to do it
	320	* atomically as bio completions can be racing against each
	321	* other.
	322	*/
	323	bio->bi_private = xchg(&io_end->bio, bio);
	324	ext4_put_io_end_defer(io_end);
	325	} else {
	326	/*
	327	* Drop io_end reference early. Inode can get freed once
	328	* we finish the bio.
	329	*/
	330	ext4_put_io_end_defer(io_end);
	331	ext4_finish_bio(bio);
	332	bio_put(bio);
	333	}
bd2d0210 TT	334	}
	335
	336	void ext4_io_submit(struct ext4_io_submit *io)
	337	{
	338	struct bio *bio = io->io_bio;
	339
	340	if (bio) {
95fe6c1a	341	int io_op_flags = io->io_wbc->sync_mode == WB_SYNC_ALL ?
70fd7614	342	REQ_SYNC : 0;
95fe6c1a	343	bio_set_op_attrs(io->io_bio, REQ_OP_WRITE, io_op_flags);
4e49ea4a	344	submit_bio(io->io_bio);
bd2d0210	345	}
7dc57615	346	io->io_bio = NULL;
97a851ed JK	347	}
	348
	349	void ext4_io_submit_init(struct ext4_io_submit *io,
	350	struct writeback_control *wbc)
	351	{
5a33911f	352	io->io_wbc = wbc;
97a851ed	353	io->io_bio = NULL;
7dc57615	354	io->io_end = NULL;
bd2d0210 TT	355	}
bd2d0210 TT	356
97a851ed JK	357	static int io_submit_init_bio(struct ext4_io_submit *io,
97a851ed JK	358	struct buffer_head *bh)
bd2d0210	359	{
bd2d0210 TT	360	struct bio *bio;
bd2d0210 TT	361
b54ffb73	362	bio = bio_alloc(GFP_NOIO, BIO_MAX_PAGES);
a1d8d9a7 TT	363	if (!bio)
a1d8d9a7 TT	364	return -ENOMEM;
001e4a87	365	wbc_init_bio(io->io_wbc, bio);
4f024f37	366	bio->bi_iter.bi_sector = bh->b_blocknr * (bh->b_size >> 9);
bd2d0210	367	bio->bi_bdev = bh->b_bdev;
bd2d0210	368	bio->bi_end_io = ext4_end_bio;
97a851ed	369	bio->bi_private = ext4_get_io_end(io->io_end);
bd2d0210	370	io->io_bio = bio;
bd2d0210 TT	371	io->io_next_block = bh->b_blocknr;
	372	return 0;
	373	}
	374
	375	static int io_submit_add_bh(struct ext4_io_submit *io,
bd2d0210	376	struct inode *inode,
2058f83a	377	struct page *page,
bd2d0210 TT	378	struct buffer_head *bh)
bd2d0210 TT	379	{
bd2d0210 TT	380	int ret;
bd2d0210 TT	381
bd2d0210 TT	382	if (io->io_bio && bh->b_blocknr != io->io_next_block) {
	383	submit_and_retry:
	384	ext4_io_submit(io);
	385	}
	386	if (io->io_bio == NULL) {
97a851ed	387	ret = io_submit_init_bio(io, bh);
bd2d0210 TT	388	if (ret)
	389	return ret;
	390	}
2058f83a	391	ret = bio_add_page(io->io_bio, page, bh->b_size, bh_offset(bh));
97a851ed JK	392	if (ret != bh->b_size)
97a851ed JK	393	goto submit_and_retry;
001e4a87	394	wbc_account_io(io->io_wbc, page, bh->b_size);
bd2d0210	395	io->io_next_block++;
bd2d0210 TT	396	return 0;
	397	}
	398
	399	int ext4_bio_write_page(struct ext4_io_submit *io,
	400	struct page *page,
	401	int len,
1c8349a1 NJ	402	struct writeback_control *wbc,
1c8349a1 NJ	403	bool keep_towrite)
bd2d0210	404	{
2058f83a	405	struct page *data_page = NULL;
bd2d0210	406	struct inode *inode = page->mapping->host;
18017479	407	unsigned block_start;
bd2d0210 TT	408	struct buffer_head bh, head;
bd2d0210 TT	409	int ret = 0;
0058f965	410	int nr_submitted = 0;
937d7b84	411	int nr_to_submit = 0;
bd2d0210	412
d50bdd5a	413	BUG_ON(!PageLocked(page));
bd2d0210	414	BUG_ON(PageWriteback(page));
bd2d0210	415
1c8349a1 NJ	416	if (keep_towrite)
	417	set_page_writeback_keepwrite(page);
	418	else
	419	set_page_writeback(page);
a54aa761	420	ClearPageError(page);
bd2d0210	421
eeece469	422	/*
f8409abd	423	* Comments copied from block_write_full_page:
eeece469 JK	424	*
	425	* The page straddles i_size. It must be zeroed out on each and every
	426	* writepage invocation because it may be mmapped. "A file is mapped
	427	* in multiples of the page size. For a file that is not a multiple of
	428	* the page size, the remaining memory is zeroed when mapped, and
	429	* writes to that region are not written out to the file."
	430	*/
09cbfeaf KS	431	if (len < PAGE_SIZE)
09cbfeaf KS	432	zero_user_segment(page, len, PAGE_SIZE);
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 TT	443	if (block_start >= len) {
	444	clear_buffer_dirty(bh);
	445	set_buffer_uptodate(bh);
	446	continue;
	447	}
8a850c3f JK	448	if (!buffer_dirty(bh) \|\| buffer_delay(bh) \|\|
	449	!buffer_mapped(bh) \|\| buffer_unwritten(bh)) {
	450	/* A hole? We can safely clear the dirty bit */
	451	if (!buffer_mapped(bh))
	452	clear_buffer_dirty(bh);
	453	if (io->io_bio)
	454	ext4_io_submit(io);
	455	continue;
	456	}
0058f965 JK	457	if (buffer_new(bh)) {
0058f965 JK	458	clear_buffer_new(bh);
e64855c6	459	clean_bdev_bh_alias(bh);
0058f965 JK	460	}
0058f965 JK	461	set_buffer_async_write(bh);
937d7b84	462	nr_to_submit++;
0058f965 JK	463	} while ((bh = bh->b_this_page) != head);
0058f965 JK	464
0058f965	465	bh = head = page_buffers(page);
2058f83a	466
937d7b84 TT	467	if (ext4_encrypted_inode(inode) && S_ISREG(inode->i_mode) &&
937d7b84 TT	468	nr_to_submit) {
c9af28fd TT	469	gfp_t gfp_flags = GFP_NOFS;
	470
	471	retry_encrypt:
9c4bb8a3 DG	472	data_page = fscrypt_encrypt_page(inode, page, PAGE_SIZE, 0,
9c4bb8a3 DG	473	page->index, gfp_flags);
2058f83a MH	474	if (IS_ERR(data_page)) {
2058f83a MH	475	ret = PTR_ERR(data_page);
4762cc3f	476	if (ret == -ENOMEM && wbc->sync_mode == WB_SYNC_ALL) {
c9af28fd TT	477	if (io->io_bio) {
	478	ext4_io_submit(io);
	479	congestion_wait(BLK_RW_ASYNC, HZ/50);
	480	}
	481	gfp_flags \|= __GFP_NOFAIL;
	482	goto retry_encrypt;
	483	}
2058f83a MH	484	data_page = NULL;
	485	goto out;
	486	}
	487	}
	488
	489	/* Now submit buffers to write */
0058f965 JK	490	do {
	491	if (!buffer_async_write(bh))
	492	continue;
2058f83a MH	493	ret = io_submit_add_bh(io, inode,
2058f83a MH	494	data_page ? data_page : page, bh);
bd2d0210 TT	495	if (ret) {
	496	/*
	497	* We only get here on ENOMEM. Not much else
	498	* we can do but mark the page as dirty, and
	499	* better luck next time.
	500	*/
bd2d0210 TT	501	break;
bd2d0210 TT	502	}
0058f965	503	nr_submitted++;
1ae48a63	504	clear_buffer_dirty(bh);
0058f965 JK	505	} while ((bh = bh->b_this_page) != head);
	506
	507	/* Error stopped previous loop? Clean up buffers... */
	508	if (ret) {
2058f83a MH	509	out:
2058f83a MH	510	if (data_page)
a7550b30	511	fscrypt_restore_control_page(data_page);
2058f83a MH	512	printk_ratelimited(KERN_ERR "%s: ret = %d\n", __func__, ret);
2058f83a MH	513	redirty_page_for_writepage(wbc, page);
0058f965 JK	514	do {
	515	clear_buffer_async_write(bh);
	516	bh = bh->b_this_page;
	517	} while (bh != head);
bd2d0210 TT	518	}
bd2d0210 TT	519	unlock_page(page);
0058f965 JK	520	/* Nothing submitted - we have to end page writeback */
	521	if (!nr_submitted)
	522	end_page_writeback(page);
bd2d0210 TT	523	return ret;
bd2d0210 TT	524	}