xfs_aops.c

linux2.6.16版本

/*
 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
 * All Rights Reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it would be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write the Free Software Foundation,
 * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */
#include "xfs.h"
#include "xfs_bit.h"
#include "xfs_log.h"
#include "xfs_inum.h"
#include "xfs_sb.h"
#include "xfs_ag.h"
#include "xfs_dir.h"
#include "xfs_dir2.h"
#include "xfs_trans.h"
#include "xfs_dmapi.h"
#include "xfs_mount.h"
#include "xfs_bmap_btree.h"
#include "xfs_alloc_btree.h"
#include "xfs_ialloc_btree.h"
#include "xfs_dir_sf.h"
#include "xfs_dir2_sf.h"
#include "xfs_attr_sf.h"
#include "xfs_dinode.h"
#include "xfs_inode.h"
#include "xfs_alloc.h"
#include "xfs_btree.h"
#include "xfs_error.h"
#include "xfs_rw.h"
#include "xfs_iomap.h"
#include <linux/mpage.h>
#include <linux/pagevec.h>
#include <linux/writeback.h>

STATIC void xfs_count_page_state(struct page *, int *, int *, int *);

#if defined(XFS_RW_TRACE)
void
xfs_page_trace(
	int		tag,
	struct inode	*inode,
	struct page	*page,
	int		mask)
{
	xfs_inode_t	*ip;
	vnode_t		*vp = LINVFS_GET_VP(inode);
	loff_t		isize = i_size_read(inode);
	loff_t		offset = page_offset(page);
	int		delalloc = -1, unmapped = -1, unwritten = -1;

	if (page_has_buffers(page))
		xfs_count_page_state(page, &delalloc, &unmapped, &unwritten);

	ip = xfs_vtoi(vp);
	if (!ip->i_rwtrace)
		return;

	ktrace_enter(ip->i_rwtrace,
		(void *)((unsigned long)tag),
		(void *)ip,
		(void *)inode,
		(void *)page,
		(void *)((unsigned long)mask),
		(void *)((unsigned long)((ip->i_d.di_size >> 32) & 0xffffffff)),
		(void *)((unsigned long)(ip->i_d.di_size & 0xffffffff)),
		(void *)((unsigned long)((isize >> 32) & 0xffffffff)),
		(void *)((unsigned long)(isize & 0xffffffff)),
		(void *)((unsigned long)((offset >> 32) & 0xffffffff)),
		(void *)((unsigned long)(offset & 0xffffffff)),
		(void *)((unsigned long)delalloc),
		(void *)((unsigned long)unmapped),
		(void *)((unsigned long)unwritten),
		(void *)NULL,
		(void *)NULL);
}
#else
#define xfs_page_trace(tag, inode, page, mask)
#endif

/*
 * Schedule IO completion handling on a xfsdatad if this was
 * the final hold on this ioend.
 */
STATIC void
xfs_finish_ioend(
	xfs_ioend_t		*ioend)
{
	if (atomic_dec_and_test(&ioend->io_remaining))
		queue_work(xfsdatad_workqueue, &ioend->io_work);
}

/*
 * We're now finished for good with this ioend structure.
 * Update the page state via the associated buffer_heads,
 * release holds on the inode and bio, and finally free
 * up memory.  Do not use the ioend after this.
 */
STATIC void
xfs_destroy_ioend(
	xfs_ioend_t		*ioend)
{
	struct buffer_head	*bh, *next;

	for (bh = ioend->io_buffer_head; bh; bh = next) {
		next = bh->b_private;
		bh->b_end_io(bh, ioend->io_uptodate);
	}

	vn_iowake(ioend->io_vnode);
	mempool_free(ioend, xfs_ioend_pool);
}

/*
 * Buffered IO write completion for delayed allocate extents.
 * TODO: Update ondisk isize now that we know the file data
 * has been flushed (i.e. the notorious "NULL file" problem).
 */
STATIC void
xfs_end_bio_delalloc(
	void			*data)
{
	xfs_ioend_t		*ioend = data;

	xfs_destroy_ioend(ioend);
}

/*
 * Buffered IO write completion for regular, written extents.
 */
STATIC void
xfs_end_bio_written(
	void			*data)
{
	xfs_ioend_t		*ioend = data;

	xfs_destroy_ioend(ioend);
}

/*
 * IO write completion for unwritten extents.
 *
 * Issue transactions to convert a buffer range from unwritten
 * to written extents.
 */
STATIC void
xfs_end_bio_unwritten(
	void			*data)
{
	xfs_ioend_t		*ioend = data;
	vnode_t			*vp = ioend->io_vnode;
	xfs_off_t		offset = ioend->io_offset;
	size_t			size = ioend->io_size;
	int			error;

	if (ioend->io_uptodate)
		VOP_BMAP(vp, offset, size, BMAPI_UNWRITTEN, NULL, NULL, error);
	xfs_destroy_ioend(ioend);
}

/*
 * Allocate and initialise an IO completion structure.
 * We need to track unwritten extent write completion here initially.
 * We'll need to extend this for updating the ondisk inode size later
 * (vs. incore size).
 */
STATIC xfs_ioend_t *
xfs_alloc_ioend(
	struct inode		*inode,
	unsigned int		type)
{
	xfs_ioend_t		*ioend;

	ioend = mempool_alloc(xfs_ioend_pool, GFP_NOFS);

	/*
	 * Set the count to 1 initially, which will prevent an I/O
	 * completion callback from happening before we have started
	 * all the I/O from calling the completion routine too early.
	 */
	atomic_set(&ioend->io_remaining, 1);
	ioend->io_uptodate = 1; /* cleared if any I/O fails */
	ioend->io_list = NULL;
	ioend->io_type = type;
	ioend->io_vnode = LINVFS_GET_VP(inode);
	ioend->io_buffer_head = NULL;
	ioend->io_buffer_tail = NULL;
	atomic_inc(&ioend->io_vnode->v_iocount);
	ioend->io_offset = 0;
	ioend->io_size = 0;

	if (type == IOMAP_UNWRITTEN)
		INIT_WORK(&ioend->io_work, xfs_end_bio_unwritten, ioend);
	else if (type == IOMAP_DELAY)
		INIT_WORK(&ioend->io_work, xfs_end_bio_delalloc, ioend);
	else
		INIT_WORK(&ioend->io_work, xfs_end_bio_written, ioend);

	return ioend;
}

STATIC int
xfs_map_blocks(
	struct inode		*inode,
	loff_t			offset,
	ssize_t			count,
	xfs_iomap_t		*mapp,
	int			flags)
{
	vnode_t			*vp = LINVFS_GET_VP(inode);
	int			error, nmaps = 1;

	VOP_BMAP(vp, offset, count, flags, mapp, &nmaps, error);
	if (!error && (flags & (BMAPI_WRITE|BMAPI_ALLOCATE)))
		VMODIFY(vp);
	return -error;
}

STATIC inline int
xfs_iomap_valid(
	xfs_iomap_t		*iomapp,
	loff_t			offset)
{
	return offset >= iomapp->iomap_offset &&
		offset < iomapp->iomap_offset + iomapp->iomap_bsize;
}

/*
 * BIO completion handler for buffered IO.
 */
STATIC int
xfs_end_bio(
	struct bio		*bio,
	unsigned int		bytes_done,
	int			error)
{
	xfs_ioend_t		*ioend = bio->bi_private;

	if (bio->bi_size)
		return 1;

	ASSERT(ioend);
	ASSERT(atomic_read(&bio->bi_cnt) >= 1);

	/* Toss bio and pass work off to an xfsdatad thread */
	if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
		ioend->io_uptodate = 0;
	bio->bi_private = NULL;
	bio->bi_end_io = NULL;

	bio_put(bio);
	xfs_finish_ioend(ioend);
	return 0;
}

STATIC void
xfs_submit_ioend_bio(
	xfs_ioend_t	*ioend,
	struct bio	*bio)
{
	atomic_inc(&ioend->io_remaining);

	bio->bi_private = ioend;
	bio->bi_end_io = xfs_end_bio;

	submit_bio(WRITE, bio);
	ASSERT(!bio_flagged(bio, BIO_EOPNOTSUPP));
	bio_put(bio);
}

STATIC struct bio *
xfs_alloc_ioend_bio(
	struct buffer_head	*bh)
{
	struct bio		*bio;
	int			nvecs = bio_get_nr_vecs(bh->b_bdev);

	do {
		bio = bio_alloc(GFP_NOIO, nvecs);
		nvecs >>= 1;
	} while (!bio);

	ASSERT(bio->bi_private == NULL);
	bio->bi_sector = bh->b_blocknr * (bh->b_size >> 9);
	bio->bi_bdev = bh->b_bdev;
	bio_get(bio);
	return bio;
}

STATIC void
xfs_start_buffer_writeback(
	struct buffer_head	*bh)
{
	ASSERT(buffer_mapped(bh));
	ASSERT(buffer_locked(bh));
	ASSERT(!buffer_delay(bh));
	ASSERT(!buffer_unwritten(bh));

	mark_buffer_async_write(bh);
	set_buffer_uptodate(bh);
	clear_buffer_dirty(bh);
}

STATIC void
xfs_start_page_writeback(
	struct page		*page,
	struct writeback_control *wbc,
	int			clear_dirty,
	int			buffers)
{
	ASSERT(PageLocked(page));
	ASSERT(!PageWriteback(page));
	set_page_writeback(page);
	if (clear_dirty)
		clear_page_dirty(page);
	unlock_page(page);
	if (!buffers) {
		end_page_writeback(page);
		wbc->pages_skipped++;	/* We didn't write this page */
	}
}

static inline int bio_add_buffer(struct bio *bio, struct buffer_head *bh)
{
	return bio_add_page(bio, bh->b_page, bh->b_size, bh_offset(bh));
}

/*
 * Submit all of the bios for all of the ioends we have saved up, covering the
 * initial writepage page and also any probed pages.
 *
 * Because we may have multiple ioends spanning a page, we need to start
 * writeback on all the buffers before we submit them for I/O. If we mark the
 * buffers as we got, then we can end up with a page that only has buffers
 * marked async write and I/O complete on can occur before we mark the other
 * buffers async write.
 *
 * The end result of this is that we trip a bug in end_page_writeback() because
 * we call it twice for the one page as the code in end_buffer_async_write()
 * assumes that all buffers on the page are started at the same time.
 *
 * The fix is two passes across the ioend list - one to start writeback on the
 * bufferheads, and then the second one submit them for I/O.
 */
STATIC void
xfs_submit_ioend(
	xfs_ioend_t		*ioend)
{
	xfs_ioend_t		*head = ioend;
	xfs_ioend_t		*next;
	struct buffer_head	*bh;
	struct bio		*bio;
	sector_t		lastblock = 0;

	/* Pass 1 - start writeback */
	do {
		next = ioend->io_list;
		for (bh = ioend->io_buffer_head; bh; bh = bh->b_private) {
			xfs_start_buffer_writeback(bh);
		}
	} while ((ioend = next) != NULL);

	/* Pass 2 - submit I/O */
	ioend = head;
	do {
		next = ioend->io_list;
		bio = NULL;

		for (bh = ioend->io_buffer_head; bh; bh = bh->b_private) {

			if (!bio) {
 retry:
				bio = xfs_alloc_ioend_bio(bh);
			} else if (bh->b_blocknr != lastblock + 1) {
				xfs_submit_ioend_bio(ioend, bio);
				goto retry;
			}

			if (bio_add_buffer(bio, bh) != bh->b_size) {
				xfs_submit_ioend_bio(ioend, bio);
				goto retry;
			}

			lastblock = bh->b_blocknr;
		}
		if (bio)
			xfs_submit_ioend_bio(ioend, bio);
		xfs_finish_ioend(ioend);
	} while ((ioend = next) != NULL);
}

/*
 * Cancel submission of all buffer_heads so far in this endio.
 * Toss the endio too.  Only ever called for the initial page
 * in a writepage request, so only ever one page.
 */
STATIC void
xfs_cancel_ioend(
	xfs_ioend_t		*ioend)
{
	xfs_ioend_t		*next;
	struct buffer_head	*bh, *next_bh;

	do {
		next = ioend->io_list;
		bh = ioend->io_buffer_head;
		do {
			next_bh = bh->b_private;
			clear_buffer_async_write(bh);
			unlock_buffer(bh);
		} while ((bh = next_bh) != NULL);

		vn_iowake(ioend->io_vnode);
		mempool_free(ioend, xfs_ioend_pool);
	} while ((ioend = next) != NULL);
}

/*
 * Test to see if we've been building up a completion structure for
 * earlier buffers -- if so, we try to append to this ioend if we
 * can, otherwise we finish off any current ioend and start another.
 * Return true if we've finished the given ioend.
 */
STATIC void
xfs_add_to_ioend(
	struct inode		*inode,
	struct buffer_head	*bh,
	xfs_off_t		offset,
	unsigned int		type,
	xfs_ioend_t		**result,
	int			need_ioend)
{
	xfs_ioend_t		*ioend = *result;

	if (!ioend || need_ioend || type != ioend->io_type) {
		xfs_ioend_t	*previous = *result;

		ioend = xfs_alloc_ioend(inode, type);
		ioend->io_offset = offset;
		ioend->io_buffer_head = bh;
		ioend->io_buffer_tail = bh;
		if (previous)
			previous->io_list = ioend;
		*result = ioend;
	} else {
		ioend->io_buffer_tail->b_private = bh;
		ioend->io_buffer_tail = bh;
	}

	bh->b_private = NULL;
	ioend->io_size += bh->b_size;
}

STATIC void
xfs_map_at_offset(
	struct buffer_head	*bh,
	loff_t			offset,
	int			block_bits,
	xfs_iomap_t		*iomapp)
{
	xfs_daddr_t		bn;
	int			sector_shift;

	ASSERT(!(iomapp->iomap_flags & IOMAP_HOLE));
	ASSERT(!(iomapp->iomap_flags & IOMAP_DELAY));
	ASSERT(iomapp->iomap_bn != IOMAP_DADDR_NULL);

	sector_shift = block_bits - BBSHIFT;
	bn = (iomapp->iomap_bn >> sector_shift) +
	      ((offset - iomapp->iomap_offset) >> block_bits);

	ASSERT(bn || (iomapp->iomap_flags & IOMAP_REALTIME));
	ASSERT((bn << sector_shift) >= iomapp->iomap_bn);

	lock_buffer(bh);
	bh->b_blocknr = bn;
	bh->b_bdev = iomapp->iomap_target->bt_bdev;
	set_buffer_mapped(bh);
	clear_buffer_delay(bh);
	clear_buffer_unwritten(bh);