#include <linux/pagevec.h>
#include <linux/writeback.h>
+/*
+ * Types of I/O for bmap clustering and I/O completion tracking.
+ */
+enum {
+ IO_READ, /* mapping for a read */
+ IO_DELAY, /* mapping covers delalloc region */
+ IO_UNWRITTEN, /* mapping covers allocated but uninitialized data */
+ IO_NEW /* just allocated */
+};
/*
* Prime number of hash buckets since address is used as the key.
STATIC struct block_device *
xfs_find_bdev_for_inode(
- struct xfs_inode *ip)
+ struct inode *inode)
{
+ struct xfs_inode *ip = XFS_I(inode);
struct xfs_mount *mp = ip->i_mount;
if (XFS_IS_REALTIME_INODE(ip))
xfs_fsize_t isize;
ASSERT((ip->i_d.di_mode & S_IFMT) == S_IFREG);
- ASSERT(ioend->io_type != IOMAP_READ);
+ ASSERT(ioend->io_type != IO_READ);
if (unlikely(ioend->io_error))
return 0;
if (atomic_dec_and_test(&ioend->io_remaining)) {
struct workqueue_struct *wq;
- wq = (ioend->io_type == IOMAP_UNWRITTEN) ?
+ wq = (ioend->io_type == IO_UNWRITTEN) ?
xfsconvertd_workqueue : xfsdatad_workqueue;
queue_work(wq, &ioend->io_work);
if (wait)
* For unwritten extents we need to issue transactions to convert a
* range to normal written extens after the data I/O has finished.
*/
- if (ioend->io_type == IOMAP_UNWRITTEN &&
+ if (ioend->io_type == IO_UNWRITTEN &&
likely(!ioend->io_error && !XFS_FORCED_SHUTDOWN(ip->i_mount))) {
error = xfs_iomap_write_unwritten(ip, ioend->io_offset,
* We might have to update the on-disk file size after extending
* writes.
*/
- if (ioend->io_type != IOMAP_READ) {
+ if (ioend->io_type != IO_READ) {
error = xfs_setfilesize(ioend);
ASSERT(!error || error == EAGAIN);
}
struct inode *inode,
loff_t offset,
ssize_t count,
- xfs_iomap_t *mapp,
+ struct xfs_bmbt_irec *imap,
int flags)
{
int nmaps = 1;
+ int new = 0;
- return -xfs_iomap(XFS_I(inode), offset, count, flags, mapp, &nmaps);
+ return -xfs_iomap(XFS_I(inode), offset, count, flags, imap, &nmaps, &new);
}
STATIC int
-xfs_iomap_valid(
- xfs_iomap_t *iomapp,
- loff_t offset)
+xfs_imap_valid(
+ struct inode *inode,
+ struct xfs_bmbt_irec *imap,
+ xfs_off_t offset)
{
- return offset >= iomapp->iomap_offset &&
- offset < iomapp->iomap_offset + iomapp->iomap_bsize;
+ offset >>= inode->i_blkbits;
+
+ return offset >= imap->br_startoff &&
+ offset < imap->br_startoff + imap->br_blockcount;
}
/*
STATIC void
xfs_map_buffer(
+ struct inode *inode,
struct buffer_head *bh,
- xfs_iomap_t *mp,
- xfs_off_t offset,
- uint block_bits)
+ struct xfs_bmbt_irec *imap,
+ xfs_off_t offset)
{
sector_t bn;
+ struct xfs_mount *m = XFS_I(inode)->i_mount;
+ xfs_off_t iomap_offset = XFS_FSB_TO_B(m, imap->br_startoff);
+ xfs_daddr_t iomap_bn = xfs_fsb_to_db(XFS_I(inode), imap->br_startblock);
- ASSERT(mp->iomap_bn != IOMAP_DADDR_NULL);
+ ASSERT(imap->br_startblock != HOLESTARTBLOCK);
+ ASSERT(imap->br_startblock != DELAYSTARTBLOCK);
- bn = (mp->iomap_bn >> (block_bits - BBSHIFT)) +
- ((offset - mp->iomap_offset) >> block_bits);
+ bn = (iomap_bn >> (inode->i_blkbits - BBSHIFT)) +
+ ((offset - iomap_offset) >> inode->i_blkbits);
- ASSERT(bn || (mp->iomap_flags & IOMAP_REALTIME));
+ ASSERT(bn || XFS_IS_REALTIME_INODE(XFS_I(inode)));
bh->b_blocknr = bn;
set_buffer_mapped(bh);
STATIC void
xfs_map_at_offset(
+ struct inode *inode,
struct buffer_head *bh,
- loff_t offset,
- int block_bits,
- xfs_iomap_t *iomapp)
+ struct xfs_bmbt_irec *imap,
+ xfs_off_t offset)
{
- ASSERT(!(iomapp->iomap_flags & IOMAP_HOLE));
- ASSERT(!(iomapp->iomap_flags & IOMAP_DELAY));
+ ASSERT(imap->br_startblock != HOLESTARTBLOCK);
+ ASSERT(imap->br_startblock != DELAYSTARTBLOCK);
lock_buffer(bh);
- xfs_map_buffer(bh, iomapp, offset, block_bits);
- bh->b_bdev = iomapp->iomap_target->bt_bdev;
+ xfs_map_buffer(inode, bh, imap, offset);
+ bh->b_bdev = xfs_find_bdev_for_inode(inode);
set_buffer_mapped(bh);
clear_buffer_delay(bh);
clear_buffer_unwritten(bh);
bh = head = page_buffers(page);
do {
if (buffer_unwritten(bh))
- acceptable = (type == IOMAP_UNWRITTEN);
+ acceptable = (type == IO_UNWRITTEN);
else if (buffer_delay(bh))
- acceptable = (type == IOMAP_DELAY);
+ acceptable = (type == IO_DELAY);
else if (buffer_dirty(bh) && buffer_mapped(bh))
- acceptable = (type == IOMAP_NEW);
+ acceptable = (type == IO_NEW);
else
break;
} while ((bh = bh->b_this_page) != head);
struct inode *inode,
struct page *page,
loff_t tindex,
- xfs_iomap_t *mp,
+ struct xfs_bmbt_irec *imap,
xfs_ioend_t **ioendp,
struct writeback_control *wbc,
int startio,
xfs_off_t end_offset;
unsigned long p_offset;
unsigned int type;
- int bbits = inode->i_blkbits;
int len, page_dirty;
int count = 0, done = 0, uptodate = 1;
xfs_off_t offset = page_offset(page);
if (buffer_unwritten(bh) || buffer_delay(bh)) {
if (buffer_unwritten(bh))
- type = IOMAP_UNWRITTEN;
+ type = IO_UNWRITTEN;
else
- type = IOMAP_DELAY;
+ type = IO_DELAY;
- if (!xfs_iomap_valid(mp, offset)) {
+ if (!xfs_imap_valid(inode, imap, offset)) {
done = 1;
continue;
}
- ASSERT(!(mp->iomap_flags & IOMAP_HOLE));
- ASSERT(!(mp->iomap_flags & IOMAP_DELAY));
+ ASSERT(imap->br_startblock != HOLESTARTBLOCK);
+ ASSERT(imap->br_startblock != DELAYSTARTBLOCK);
- xfs_map_at_offset(bh, offset, bbits, mp);
+ xfs_map_at_offset(inode, bh, imap, offset);
if (startio) {
xfs_add_to_ioend(inode, bh, offset,
type, ioendp, done);
page_dirty--;
count++;
} else {
- type = IOMAP_NEW;
+ type = IO_NEW;
if (buffer_mapped(bh) && all_bh && startio) {
lock_buffer(bh);
xfs_add_to_ioend(inode, bh, offset,
xfs_cluster_write(
struct inode *inode,
pgoff_t tindex,
- xfs_iomap_t *iomapp,
+ struct xfs_bmbt_irec *imap,
xfs_ioend_t **ioendp,
struct writeback_control *wbc,
int startio,
for (i = 0; i < pagevec_count(&pvec); i++) {
done = xfs_convert_page(inode, pvec.pages[i], tindex++,
- iomapp, ioendp, wbc, startio, all_bh);
+ imap, ioendp, wbc, startio, all_bh);
if (done)
break;
}
loff_t offset = page_offset(page);
ssize_t len = 1 << inode->i_blkbits;
- if (!xfs_is_delayed_page(page, IOMAP_DELAY))
+ if (!xfs_is_delayed_page(page, IO_DELAY))
goto out_invalidate;
if (XFS_FORCED_SHUTDOWN(ip->i_mount))
int unmapped) /* also implies page uptodate */
{
struct buffer_head *bh, *head;
- xfs_iomap_t iomap;
+ struct xfs_bmbt_irec imap;
xfs_ioend_t *ioend = NULL, *iohead = NULL;
loff_t offset;
unsigned long p_offset = 0;
unsigned int type;
__uint64_t end_offset;
- pgoff_t end_index, last_index, tlast;
+ pgoff_t end_index, last_index;
ssize_t size, len;
- int flags, err, iomap_valid = 0, uptodate = 1;
+ int flags, err, imap_valid = 0, uptodate = 1;
int page_dirty, count = 0;
int trylock = 0;
int all_bh = unmapped;
bh = head = page_buffers(page);
offset = page_offset(page);
flags = BMAPI_READ;
- type = IOMAP_NEW;
+ type = IO_NEW;
/* TODO: cleanup count and page_dirty */
* the iomap is actually still valid, but the ioend
* isn't. shouldn't happen too often.
*/
- iomap_valid = 0;
+ imap_valid = 0;
continue;
}
- if (iomap_valid)
- iomap_valid = xfs_iomap_valid(&iomap, offset);
+ if (imap_valid)
+ imap_valid = xfs_imap_valid(inode, &imap, offset);
/*
* First case, map an unwritten extent and prepare for
* Make sure we don't use a read-only iomap
*/
if (flags == BMAPI_READ)
- iomap_valid = 0;
+ imap_valid = 0;
if (buffer_unwritten(bh)) {
- type = IOMAP_UNWRITTEN;
+ type = IO_UNWRITTEN;
flags = BMAPI_WRITE | BMAPI_IGNSTATE;
} else if (buffer_delay(bh)) {
- type = IOMAP_DELAY;
+ type = IO_DELAY;
flags = BMAPI_ALLOCATE | trylock;
} else {
- type = IOMAP_NEW;
+ type = IO_NEW;
flags = BMAPI_WRITE | BMAPI_MMAP;
}
- if (!iomap_valid) {
+ if (!imap_valid) {
/*
* if we didn't have a valid mapping then we
* need to ensure that we put the new mapping
* for unwritten extent conversion.
*/
new_ioend = 1;
- if (type == IOMAP_NEW) {
+ if (type == IO_NEW) {
size = xfs_probe_cluster(inode,
page, bh, head, 0);
} else {
}
err = xfs_map_blocks(inode, offset, size,
- &iomap, flags);
+ &imap, flags);
if (err)
goto error;
- iomap_valid = xfs_iomap_valid(&iomap, offset);
+ imap_valid = xfs_imap_valid(inode, &imap,
+ offset);
}
- if (iomap_valid) {
- xfs_map_at_offset(bh, offset,
- inode->i_blkbits, &iomap);
+ if (imap_valid) {
+ xfs_map_at_offset(inode, bh, &imap, offset);
if (startio) {
xfs_add_to_ioend(inode, bh, offset,
type, &ioend,
* That means it must already have extents allocated
* underneath it. Map the extent by reading it.
*/
- if (!iomap_valid || flags != BMAPI_READ) {
+ if (!imap_valid || flags != BMAPI_READ) {
flags = BMAPI_READ;
size = xfs_probe_cluster(inode, page, bh,
head, 1);
err = xfs_map_blocks(inode, offset, size,
- &iomap, flags);
+ &imap, flags);
if (err)
goto error;
- iomap_valid = xfs_iomap_valid(&iomap, offset);
+ imap_valid = xfs_imap_valid(inode, &imap,
+ offset);
}
/*
- * We set the type to IOMAP_NEW in case we are doing a
+ * We set the type to IO_NEW in case we are doing a
* small write at EOF that is extending the file but
* without needing an allocation. We need to update the
* file size on I/O completion in this case so it is
* the same case as having just allocated a new extent
* that we are writing into for the first time.
*/
- type = IOMAP_NEW;
+ type = IO_NEW;
if (trylock_buffer(bh)) {
ASSERT(buffer_mapped(bh));
- if (iomap_valid)
+ if (imap_valid)
all_bh = 1;
xfs_add_to_ioend(inode, bh, offset, type,
- &ioend, !iomap_valid);
+ &ioend, !imap_valid);
page_dirty--;
count++;
} else {
- iomap_valid = 0;
+ imap_valid = 0;
}
} else if ((buffer_uptodate(bh) || PageUptodate(page)) &&
(unmapped || startio)) {
- iomap_valid = 0;
+ imap_valid = 0;
}
if (!iohead)
if (startio)
xfs_start_page_writeback(page, 1, count);
- if (ioend && iomap_valid) {
- offset = (iomap.iomap_offset + iomap.iomap_bsize - 1) >>
- PAGE_CACHE_SHIFT;
- tlast = min_t(pgoff_t, offset, last_index);
- xfs_cluster_write(inode, page->index + 1, &iomap, &ioend,
- wbc, startio, all_bh, tlast);
+ if (ioend && imap_valid) {
+ xfs_off_t end_index;
+
+ end_index = imap.br_startoff + imap.br_blockcount;
+
+ /* to bytes */
+ end_index <<= inode->i_blkbits;
+
+ /* to pages */
+ end_index = (end_index - 1) >> PAGE_CACHE_SHIFT;
+
+ /* check against file size */
+ if (end_index > last_index)
+ end_index = last_index;
+
+ xfs_cluster_write(inode, page->index + 1, &imap, &ioend,
+ wbc, startio, all_bh, end_index);
}
if (iohead)
trace_xfs_writepage(inode, page, 0);
+ /*
+ * Refuse to write the page out if we are called from reclaim context.
+ *
+ * This is primarily to avoid stack overflows when called from deep
+ * used stacks in random callers for direct reclaim, but disabling
+ * reclaim for kswap is a nice side-effect as kswapd causes rather
+ * suboptimal I/O patters, too.
+ *
+ * This should really be done by the core VM, but until that happens
+ * filesystems like XFS, btrfs and ext4 have to take care of this
+ * by themselves.
+ */
+ if (current->flags & PF_MEMALLOC)
+ goto out_fail;
+
/*
* We need a transaction if:
* 1. There are delalloc buffers on the page
int direct,
bmapi_flags_t flags)
{
- xfs_iomap_t iomap;
+ struct xfs_bmbt_irec imap;
xfs_off_t offset;
ssize_t size;
- int niomap = 1;
+ int nimap = 1;
+ int new = 0;
int error;
offset = (xfs_off_t)iblock << inode->i_blkbits;
return 0;
error = xfs_iomap(XFS_I(inode), offset, size,
- create ? flags : BMAPI_READ, &iomap, &niomap);
+ create ? flags : BMAPI_READ, &imap, &nimap, &new);
if (error)
return -error;
- if (niomap == 0)
+ if (nimap == 0)
return 0;
- if (iomap.iomap_bn != IOMAP_DADDR_NULL) {
+ if (imap.br_startblock != HOLESTARTBLOCK &&
+ imap.br_startblock != DELAYSTARTBLOCK) {
/*
* For unwritten extents do not report a disk address on
* the read case (treat as if we're reading into a hole).
*/
- if (create || !(iomap.iomap_flags & IOMAP_UNWRITTEN)) {
- xfs_map_buffer(bh_result, &iomap, offset,
- inode->i_blkbits);
- }
- if (create && (iomap.iomap_flags & IOMAP_UNWRITTEN)) {
+ if (create || !ISUNWRITTEN(&imap))
+ xfs_map_buffer(inode, bh_result, &imap, offset);
+ if (create && ISUNWRITTEN(&imap)) {
if (direct)
bh_result->b_private = inode;
set_buffer_unwritten(bh_result);
* If this is a realtime file, data may be on a different device.
* to that pointed to from the buffer_head b_bdev currently.
*/
- bh_result->b_bdev = iomap.iomap_target->bt_bdev;
+ bh_result->b_bdev = xfs_find_bdev_for_inode(inode);
/*
* If we previously allocated a block out beyond eof and we are now
if (create &&
((!buffer_mapped(bh_result) && !buffer_uptodate(bh_result)) ||
(offset >= i_size_read(inode)) ||
- (iomap.iomap_flags & (IOMAP_NEW|IOMAP_UNWRITTEN))))
+ (new || ISUNWRITTEN(&imap))))
set_buffer_new(bh_result);
- if (iomap.iomap_flags & IOMAP_DELAY) {
+ if (imap.br_startblock == DELAYSTARTBLOCK) {
BUG_ON(direct);
if (create) {
set_buffer_uptodate(bh_result);
}
}
+ /*
+ * If this is O_DIRECT or the mpage code calling tell them how large
+ * the mapping is, so that we can avoid repeated get_blocks calls.
+ */
if (direct || size > (1 << inode->i_blkbits)) {
- ASSERT(iomap.iomap_bsize - iomap.iomap_delta > 0);
- offset = min_t(xfs_off_t,
- iomap.iomap_bsize - iomap.iomap_delta, size);
- bh_result->b_size = (ssize_t)min_t(xfs_off_t, LONG_MAX, offset);
+ xfs_off_t mapping_size;
+
+ mapping_size = imap.br_startoff + imap.br_blockcount - iblock;
+ mapping_size <<= inode->i_blkbits;
+
+ ASSERT(mapping_size > 0);
+ if (mapping_size > size)
+ mapping_size = size;
+ if (mapping_size > LONG_MAX)
+ mapping_size = LONG_MAX;
+
+ bh_result->b_size = mapping_size;
}
return 0;
*/
ioend->io_offset = offset;
ioend->io_size = size;
- if (ioend->io_type == IOMAP_READ) {
+ if (ioend->io_type == IO_READ) {
xfs_finish_ioend(ioend, 0);
} else if (private && size > 0) {
xfs_finish_ioend(ioend, is_sync_kiocb(iocb));
* didn't map an unwritten extent so switch it's completion
* handler.
*/
- ioend->io_type = IOMAP_NEW;
+ ioend->io_type = IO_NEW;
xfs_finish_ioend(ioend, 0);
}
struct block_device *bdev;
ssize_t ret;
- bdev = xfs_find_bdev_for_inode(XFS_I(inode));
+ bdev = xfs_find_bdev_for_inode(inode);
iocb->private = xfs_alloc_ioend(inode, rw == WRITE ?
- IOMAP_UNWRITTEN : IOMAP_READ);
+ IO_UNWRITTEN : IO_READ);
ret = blockdev_direct_IO_no_locking(rw, iocb, inode, bdev, iov,
offset, nr_segs,
git.openpandora.org / pandora-kernel.git / blobdiff