#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_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_error.h"
#include "xfs_rw.h"
#include "xfs_iomap.h"
+#include "xfs_vnodeops.h"
#include <linux/mpage.h>
#include <linux/pagevec.h>
#include <linux/writeback.h>
-STATIC void xfs_count_page_state(struct page *, int *, int *, int *);
+STATIC void
+xfs_count_page_state(
+ struct page *page,
+ int *delalloc,
+ int *unmapped,
+ int *unwritten)
+{
+ struct buffer_head *bh, *head;
+
+ *delalloc = *unmapped = *unwritten = 0;
+
+ bh = head = page_buffers(page);
+ do {
+ if (buffer_uptodate(bh) && !buffer_mapped(bh))
+ (*unmapped) = 1;
+ else if (buffer_unwritten(bh))
+ (*unwritten) = 1;
+ else if (buffer_delay(bh))
+ (*delalloc) = 1;
+ } while ((bh = bh->b_this_page) != head);
+}
#if defined(XFS_RW_TRACE)
void
int tag,
struct inode *inode,
struct page *page,
- int mask)
+ unsigned long pgoff)
{
xfs_inode_t *ip;
- vnode_t *vp = LINVFS_GET_VP(inode);
+ bhv_vnode_t *vp = vn_from_inode(inode);
loff_t isize = i_size_read(inode);
loff_t offset = page_offset(page);
int delalloc = -1, unmapped = -1, unwritten = -1;
(void *)ip,
(void *)inode,
(void *)page,
- (void *)((unsigned long)mask),
+ (void *)pgoff,
(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)delalloc),
(void *)((unsigned long)unmapped),
(void *)((unsigned long)unwritten),
- (void *)NULL,
+ (void *)((unsigned long)current_pid()),
(void *)NULL);
}
#else
-#define xfs_page_trace(tag, inode, page, mask)
+#define xfs_page_trace(tag, inode, page, pgoff)
#endif
+STATIC struct block_device *
+xfs_find_bdev_for_inode(
+ struct xfs_inode *ip)
+{
+ struct xfs_mount *mp = ip->i_mount;
+
+ if (XFS_IS_REALTIME_INODE(ip))
+ return mp->m_rtdev_targp->bt_bdev;
+ else
+ return mp->m_ddev_targp->bt_bdev;
+}
+
/*
* Schedule IO completion handling on a xfsdatad if this was
- * the final hold on this ioend.
+ * the final hold on this ioend. If we are asked to wait,
+ * flush the workqueue.
*/
STATIC void
xfs_finish_ioend(
- xfs_ioend_t *ioend)
+ xfs_ioend_t *ioend,
+ int wait)
{
- if (atomic_dec_and_test(&ioend->io_remaining))
+ if (atomic_dec_and_test(&ioend->io_remaining)) {
queue_work(xfsdatad_workqueue, &ioend->io_work);
+ if (wait)
+ flush_workqueue(xfsdatad_workqueue);
+ }
}
/*
for (bh = ioend->io_buffer_head; bh; bh = next) {
next = bh->b_private;
- bh->b_end_io(bh, ioend->io_uptodate);
+ bh->b_end_io(bh, !ioend->io_error);
}
-
- vn_iowake(ioend->io_vnode);
+ if (unlikely(ioend->io_error)) {
+ vn_ioerror(XFS_I(ioend->io_inode), ioend->io_error,
+ __FILE__,__LINE__);
+ }
+ vn_iowake(XFS_I(ioend->io_inode));
mempool_free(ioend, xfs_ioend_pool);
}
/*
+ * Update on-disk file size now that data has been written to disk.
+ * The current in-memory file size is i_size. If a write is beyond
+ * eof i_new_size will be the intended file size until i_size is
+ * updated. If this write does not extend all the way to the valid
+ * file size then restrict this update to the end of the write.
+ */
+STATIC void
+xfs_setfilesize(
+ xfs_ioend_t *ioend)
+{
+ xfs_inode_t *ip = XFS_I(ioend->io_inode);
+ xfs_fsize_t isize;
+ xfs_fsize_t bsize;
+
+ ASSERT((ip->i_d.di_mode & S_IFMT) == S_IFREG);
+ ASSERT(ioend->io_type != IOMAP_READ);
+
+ if (unlikely(ioend->io_error))
+ return;
+
+ bsize = ioend->io_offset + ioend->io_size;
+
+ xfs_ilock(ip, XFS_ILOCK_EXCL);
+
+ isize = MAX(ip->i_size, ip->i_new_size);
+ isize = MIN(isize, bsize);
+
+ if (ip->i_d.di_size < isize) {
+ ip->i_d.di_size = isize;
+ ip->i_update_core = 1;
+ ip->i_update_size = 1;
+ mark_inode_dirty_sync(ioend->io_inode);
+ }
+
+ xfs_iunlock(ip, XFS_ILOCK_EXCL);
+}
+
+/*
* 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)
+ struct work_struct *work)
{
- xfs_ioend_t *ioend = data;
+ xfs_ioend_t *ioend =
+ container_of(work, xfs_ioend_t, io_work);
+ xfs_setfilesize(ioend);
xfs_destroy_ioend(ioend);
}
*/
STATIC void
xfs_end_bio_written(
- void *data)
+ struct work_struct *work)
{
- xfs_ioend_t *ioend = data;
+ xfs_ioend_t *ioend =
+ container_of(work, xfs_ioend_t, io_work);
+ xfs_setfilesize(ioend);
xfs_destroy_ioend(ioend);
}
*/
STATIC void
xfs_end_bio_unwritten(
- void *data)
+ struct work_struct *work)
{
- xfs_ioend_t *ioend = data;
- vnode_t *vp = ioend->io_vnode;
+ xfs_ioend_t *ioend =
+ container_of(work, xfs_ioend_t, io_work);
+ struct xfs_inode *ip = XFS_I(ioend->io_inode);
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);
+ if (likely(!ioend->io_error)) {
+ if (!XFS_FORCED_SHUTDOWN(ip->i_mount)) {
+ int error;
+ error = xfs_iomap_write_unwritten(ip, offset, size);
+ if (error)
+ ioend->io_error = error;
+ }
+ xfs_setfilesize(ioend);
+ }
+ xfs_destroy_ioend(ioend);
+}
+
+/*
+ * IO read completion for regular, written extents.
+ */
+STATIC void
+xfs_end_bio_read(
+ struct work_struct *work)
+{
+ xfs_ioend_t *ioend =
+ container_of(work, xfs_ioend_t, io_work);
+
xfs_destroy_ioend(ioend);
}
* 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_error = 0;
ioend->io_list = NULL;
ioend->io_type = type;
- ioend->io_vnode = LINVFS_GET_VP(inode);
+ ioend->io_inode = inode;
ioend->io_buffer_head = NULL;
ioend->io_buffer_tail = NULL;
- atomic_inc(&ioend->io_vnode->v_iocount);
+ atomic_inc(&XFS_I(ioend->io_inode)->i_iocount);
ioend->io_offset = 0;
ioend->io_size = 0;
if (type == IOMAP_UNWRITTEN)
- INIT_WORK(&ioend->io_work, xfs_end_bio_unwritten, ioend);
+ INIT_WORK(&ioend->io_work, xfs_end_bio_unwritten);
else if (type == IOMAP_DELAY)
- INIT_WORK(&ioend->io_work, xfs_end_bio_delalloc, ioend);
+ INIT_WORK(&ioend->io_work, xfs_end_bio_delalloc);
+ else if (type == IOMAP_READ)
+ INIT_WORK(&ioend->io_work, xfs_end_bio_read);
else
- INIT_WORK(&ioend->io_work, xfs_end_bio_written, ioend);
+ INIT_WORK(&ioend->io_work, xfs_end_bio_written);
return ioend;
}
xfs_iomap_t *mapp,
int flags)
{
- vnode_t *vp = LINVFS_GET_VP(inode);
+ xfs_inode_t *ip = XFS_I(inode);
int error, nmaps = 1;
- VOP_BMAP(vp, offset, count, flags, mapp, &nmaps, error);
+ error = xfs_iomap(ip, offset, count,
+ flags, mapp, &nmaps);
if (!error && (flags & (BMAPI_WRITE|BMAPI_ALLOCATE)))
- VMODIFY(vp);
+ xfs_iflags_set(ip, XFS_IMODIFIED);
return -error;
}
-STATIC inline int
+STATIC_INLINE int
xfs_iomap_valid(
xfs_iomap_t *iomapp,
loff_t offset)
/*
* BIO completion handler for buffered IO.
*/
-STATIC int
+STATIC void
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);
+ ioend->io_error = test_bit(BIO_UPTODATE, &bio->bi_flags) ? 0 : error;
/* 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;
+
+ xfs_finish_ioend(ioend, 0);
}
STATIC void
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);
+ clear_page_dirty_for_io(page);
+ set_page_writeback(page);
unlock_page(page);
- if (!buffers) {
+ /* If no buffers on the page are to be written, finish it here */
+ 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)
* 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.
+ * buffer_heads, and then submit them for I/O on the second pass.
*/
STATIC void
xfs_submit_ioend(
}
if (bio)
xfs_submit_ioend_bio(ioend, bio);
- xfs_finish_ioend(ioend);
+ xfs_finish_ioend(ioend, 0);
} while ((ioend = next) != NULL);
}
unlock_buffer(bh);
} while ((bh = next_bh) != NULL);
- vn_iowake(ioend->io_vnode);
+ vn_iowake(XFS_I(ioend->io_inode));
mempool_free(ioend, xfs_ioend_pool);
} while ((ioend = next) != NULL);
}
}
STATIC void
+xfs_map_buffer(
+ struct buffer_head *bh,
+ xfs_iomap_t *mp,
+ xfs_off_t offset,
+ uint block_bits)
+{
+ sector_t bn;
+
+ ASSERT(mp->iomap_bn != IOMAP_DADDR_NULL);
+
+ bn = (mp->iomap_bn >> (block_bits - BBSHIFT)) +
+ ((offset - mp->iomap_offset) >> block_bits);
+
+ ASSERT(bn || (mp->iomap_flags & IOMAP_REALTIME));
+
+ bh->b_blocknr = bn;
+ set_buffer_mapped(bh);
+}
+
+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;
+ xfs_map_buffer(bh, iomapp, offset, block_bits);
bh->b_bdev = iomapp->iomap_target->bt_bdev;
set_buffer_mapped(bh);
clear_buffer_delay(bh);
/* First sum forwards in this page */
do {
- if (mapped != buffer_mapped(bh))
+ if (!buffer_uptodate(bh) || (mapped != buffer_mapped(bh)))
return total;
total += bh->b_size;
} while ((bh = bh->b_this_page) != head);
for (i = 0; i < pagevec_count(&pvec); i++) {
struct page *page = pvec.pages[i];
- size_t pg_offset, len = 0;
+ size_t pg_offset, pg_len = 0;
if (tindex == tlast) {
pg_offset =
} else
pg_offset = PAGE_CACHE_SIZE;
- if (page->index == tindex && !TestSetPageLocked(page)) {
- len = xfs_probe_page(page, pg_offset, mapped);
+ if (page->index == tindex && trylock_page(page)) {
+ pg_len = xfs_probe_page(page, pg_offset, mapped);
unlock_page(page);
}
- if (!len) {
+ if (!pg_len) {
done = 1;
break;
}
- total += len;
+ total += pg_len;
tindex++;
}
acceptable = (type == IOMAP_UNWRITTEN);
else if (buffer_delay(bh))
acceptable = (type == IOMAP_DELAY);
- else if (buffer_mapped(bh))
- acceptable = (type == 0);
+ else if (buffer_dirty(bh) && buffer_mapped(bh))
+ acceptable = (type == IOMAP_NEW);
else
break;
} while ((bh = bh->b_this_page) != head);
if (page->index != tindex)
goto fail;
- if (TestSetPageLocked(page))
+ if (!trylock_page(page))
goto fail;
if (PageWriteback(page))
goto fail_unlock_page;
/*
* page_dirty is initially a count of buffers on the page before
- * EOF and is decrememted as we move each into a cleanable state.
+ * EOF and is decremented as we move each into a cleanable state.
*
* Derivation:
*
page_dirty--;
count++;
} else {
- type = 0;
+ type = IOMAP_NEW;
if (buffer_mapped(bh) && all_bh && startio) {
lock_buffer(bh);
xfs_add_to_ioend(inode, bh, offset,
struct backing_dev_info *bdi;
bdi = inode->i_mapping->backing_dev_info;
+ wbc->nr_to_write--;
if (bdi_write_congested(bdi)) {
wbc->encountered_congestion = 1;
done = 1;
- } else if (--wbc->nr_to_write <= 0) {
+ } else if (wbc->nr_to_write <= 0) {
done = 1;
}
}
- xfs_start_page_writeback(page, wbc, !page_dirty, count);
+ xfs_start_page_writeback(page, !page_dirty, count);
}
return done;
* page if possible.
* The bh->b_state's cannot know if any of the blocks or which block for
* that matter are dirty due to mmap writes, and therefore bh uptodate is
- * only vaild if the page itself isn't completely uptodate. Some layers
+ * only valid if the page itself isn't completely uptodate. Some layers
* may clear the page dirty flag prior to calling write page, under the
* assumption the entire page will be written out; by not writing out the
* whole page the page can be reused before all valid dirty data is
pgoff_t end_index, last_index, tlast;
ssize_t size, len;
int flags, err, iomap_valid = 0, uptodate = 1;
- int page_dirty, count = 0, trylock_flag = 0;
+ int page_dirty, count = 0;
+ int trylock = 0;
int all_bh = unmapped;
- /* wait for other IO threads? */
- if (startio && (wbc->sync_mode == WB_SYNC_NONE && wbc->nonblocking))
- trylock_flag |= BMAPI_TRYLOCK;
+ if (startio) {
+ if (wbc->sync_mode == WB_SYNC_NONE && wbc->nonblocking)
+ trylock |= BMAPI_TRYLOCK;
+ }
/* Is this page beyond the end of the file? */
offset = i_size_read(inode);
/*
* page_dirty is initially a count of buffers on the page before
- * EOF and is decrememted as we move each into a cleanable state.
+ * EOF and is decremented as we move each into a cleanable state.
*
* Derivation:
*
bh = head = page_buffers(page);
offset = page_offset(page);
- flags = -1;
- type = 0;
+ flags = BMAPI_READ;
+ type = IOMAP_NEW;
/* TODO: cleanup count and page_dirty */
*
* Third case, an unmapped buffer was found, and we are
* in a path where we need to write the whole page out.
- */
+ */
if (buffer_unwritten(bh) || buffer_delay(bh) ||
((buffer_uptodate(bh) || PageUptodate(page)) &&
!buffer_mapped(bh) && (unmapped || startio))) {
- /*
+ int new_ioend = 0;
+
+ /*
* Make sure we don't use a read-only iomap
*/
- if (flags == BMAPI_READ)
+ if (flags == BMAPI_READ)
iomap_valid = 0;
if (buffer_unwritten(bh)) {
type = IOMAP_UNWRITTEN;
- flags = BMAPI_WRITE|BMAPI_IGNSTATE;
+ flags = BMAPI_WRITE | BMAPI_IGNSTATE;
} else if (buffer_delay(bh)) {
type = IOMAP_DELAY;
- flags = BMAPI_ALLOCATE;
- if (!startio)
- flags |= trylock_flag;
+ flags = BMAPI_ALLOCATE | trylock;
} else {
type = IOMAP_NEW;
- flags = BMAPI_WRITE|BMAPI_MMAP;
+ flags = BMAPI_WRITE | BMAPI_MMAP;
}
if (!iomap_valid) {
+ /*
+ * if we didn't have a valid mapping then we
+ * need to ensure that we put the new mapping
+ * in a new ioend structure. This needs to be
+ * done to ensure that the ioends correctly
+ * reflect the block mappings at io completion
+ * for unwritten extent conversion.
+ */
+ new_ioend = 1;
if (type == IOMAP_NEW) {
size = xfs_probe_cluster(inode,
page, bh, head, 0);
if (startio) {
xfs_add_to_ioend(inode, bh, offset,
type, &ioend,
- !iomap_valid);
+ new_ioend);
} else {
set_buffer_dirty(bh);
unlock_buffer(bh);
* That means it must already have extents allocated
* underneath it. Map the extent by reading it.
*/
- if (!iomap_valid || type != 0) {
+ if (!iomap_valid || flags != BMAPI_READ) {
flags = BMAPI_READ;
size = xfs_probe_cluster(inode, page, bh,
head, 1);
iomap_valid = xfs_iomap_valid(&iomap, offset);
}
- type = 0;
+ /*
+ * We set the type to IOMAP_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;
if (!test_and_set_bit(BH_Lock, &bh->b_state)) {
ASSERT(buffer_mapped(bh));
if (iomap_valid)
SetPageUptodate(page);
if (startio)
- xfs_start_page_writeback(page, wbc, 1, count);
+ xfs_start_page_writeback(page, 1, count);
if (ioend && iomap_valid) {
offset = (iomap.iomap_offset + iomap.iomap_bsize - 1) >>
return err;
}
-STATIC int
-__linvfs_get_block(
- struct inode *inode,
- sector_t iblock,
- unsigned long blocks,
- struct buffer_head *bh_result,
- int create,
- int direct,
- bmapi_flags_t flags)
-{
- vnode_t *vp = LINVFS_GET_VP(inode);
- xfs_iomap_t iomap;
- xfs_off_t offset;
- ssize_t size;
- int retpbbm = 1;
- int error;
-
- offset = (xfs_off_t)iblock << inode->i_blkbits;
- if (blocks)
- size = (ssize_t) min_t(xfs_off_t, LONG_MAX,
- (xfs_off_t)blocks << inode->i_blkbits);
- else
- size = 1 << inode->i_blkbits;
-
- VOP_BMAP(vp, offset, size,
- create ? flags : BMAPI_READ, &iomap, &retpbbm, error);
- if (error)
- return -error;
-
- if (retpbbm == 0)
- return 0;
-
- if (iomap.iomap_bn != IOMAP_DADDR_NULL) {
- xfs_daddr_t bn;
- xfs_off_t delta;
-
- /* 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)) {
- delta = offset - iomap.iomap_offset;
- delta >>= inode->i_blkbits;
-
- bn = iomap.iomap_bn >> (inode->i_blkbits - BBSHIFT);
- bn += delta;
- BUG_ON(!bn && !(iomap.iomap_flags & IOMAP_REALTIME));
- bh_result->b_blocknr = bn;
- set_buffer_mapped(bh_result);
- }
- if (create && (iomap.iomap_flags & IOMAP_UNWRITTEN)) {
- if (direct)
- bh_result->b_private = inode;
- set_buffer_unwritten(bh_result);
- set_buffer_delay(bh_result);
- }
- }
-
- /* If this is a realtime file, data might be on a new device */
- bh_result->b_bdev = iomap.iomap_target->bt_bdev;
-
- /* If we previously allocated a block out beyond eof and
- * we are now coming back to use it then we will need to
- * flag it as new even if it has a disk address.
- */
- if (create &&
- ((!buffer_mapped(bh_result) && !buffer_uptodate(bh_result)) ||
- (offset >= i_size_read(inode)) || (iomap.iomap_flags & IOMAP_NEW)))
- set_buffer_new(bh_result);
-
- if (iomap.iomap_flags & IOMAP_DELAY) {
- BUG_ON(direct);
- if (create) {
- set_buffer_uptodate(bh_result);
- set_buffer_mapped(bh_result);
- set_buffer_delay(bh_result);
- }
- }
-
- if (blocks) {
- ASSERT(iomap.iomap_bsize - iomap.iomap_delta > 0);
- offset = min_t(xfs_off_t,
- iomap.iomap_bsize - iomap.iomap_delta,
- (xfs_off_t)blocks << inode->i_blkbits);
- bh_result->b_size = (u32) min_t(xfs_off_t, UINT_MAX, offset);
- }
-
- return 0;
-}
-
-int
-linvfs_get_block(
- struct inode *inode,
- sector_t iblock,
- struct buffer_head *bh_result,
- int create)
-{
- return __linvfs_get_block(inode, iblock, 0, bh_result,
- create, 0, BMAPI_WRITE);
-}
-
-STATIC int
-linvfs_get_blocks_direct(
- struct inode *inode,
- sector_t iblock,
- unsigned long max_blocks,
- struct buffer_head *bh_result,
- int create)
-{
- return __linvfs_get_block(inode, iblock, max_blocks, bh_result,
- create, 1, BMAPI_WRITE|BMAPI_DIRECT);
-}
-
-STATIC void
-linvfs_end_io_direct(
- struct kiocb *iocb,
- loff_t offset,
- ssize_t size,
- void *private)
-{
- xfs_ioend_t *ioend = iocb->private;
-
- /*
- * Non-NULL private data means we need to issue a transaction to
- * convert a range from unwritten to written extents. This needs
- * to happen from process contect but aio+dio I/O completion
- * happens from irq context so we need to defer it to a workqueue.
- * This is not nessecary for synchronous direct I/O, but we do
- * it anyway to keep the code uniform and simpler.
- *
- * The core direct I/O code might be changed to always call the
- * completion handler in the future, in which case all this can
- * go away.
- */
- if (private && size > 0) {
- ioend->io_offset = offset;
- ioend->io_size = size;
- xfs_finish_ioend(ioend);
- } else {
- ASSERT(size >= 0);
- xfs_destroy_ioend(ioend);
- }
-
- /*
- * blockdev_direct_IO can return an error even afer the I/O
- * completion handler was called. Thus we need to protect
- * against double-freeing.
- */
- iocb->private = NULL;
-}
-
-STATIC ssize_t
-linvfs_direct_IO(
- int rw,
- struct kiocb *iocb,
- const struct iovec *iov,
- loff_t offset,
- unsigned long nr_segs)
-{
- struct file *file = iocb->ki_filp;
- struct inode *inode = file->f_mapping->host;
- vnode_t *vp = LINVFS_GET_VP(inode);
- xfs_iomap_t iomap;
- int maps = 1;
- int error;
- ssize_t ret;
-
- VOP_BMAP(vp, offset, 0, BMAPI_DEVICE, &iomap, &maps, error);
- if (error)
- return -error;
-
- iocb->private = xfs_alloc_ioend(inode, IOMAP_UNWRITTEN);
-
- ret = blockdev_direct_IO_own_locking(rw, iocb, inode,
- iomap.iomap_target->bt_bdev,
- iov, offset, nr_segs,
- linvfs_get_blocks_direct,
- linvfs_end_io_direct);
-
- if (unlikely(ret <= 0 && iocb->private))
- xfs_destroy_ioend(iocb->private);
- return ret;
-}
-
-
-STATIC sector_t
-linvfs_bmap(
- struct address_space *mapping,
- sector_t block)
-{
- struct inode *inode = (struct inode *)mapping->host;
- vnode_t *vp = LINVFS_GET_VP(inode);
- int error;
-
- vn_trace_entry(vp, "linvfs_bmap", (inst_t *)__return_address);
-
- VOP_RWLOCK(vp, VRWLOCK_READ);
- VOP_FLUSH_PAGES(vp, (xfs_off_t)0, -1, 0, FI_REMAPF, error);
- VOP_RWUNLOCK(vp, VRWLOCK_READ);
- return generic_block_bmap(mapping, block, linvfs_get_block);
-}
-
-STATIC int
-linvfs_readpage(
- struct file *unused,
- struct page *page)
-{
- return mpage_readpage(page, linvfs_get_block);
-}
-
-STATIC int
-linvfs_readpages(
- struct file *unused,
- struct address_space *mapping,
- struct list_head *pages,
- unsigned nr_pages)
-{
- return mpage_readpages(mapping, pages, nr_pages, linvfs_get_block);
-}
-
-STATIC void
-xfs_count_page_state(
- struct page *page,
- int *delalloc,
- int *unmapped,
- int *unwritten)
-{
- struct buffer_head *bh, *head;
-
- *delalloc = *unmapped = *unwritten = 0;
-
- bh = head = page_buffers(page);
- do {
- if (buffer_uptodate(bh) && !buffer_mapped(bh))
- (*unmapped) = 1;
- else if (buffer_unwritten(bh) && !buffer_delay(bh))
- clear_buffer_unwritten(bh);
- else if (buffer_unwritten(bh))
- (*unwritten) = 1;
- else if (buffer_delay(bh))
- (*delalloc) = 1;
- } while ((bh = bh->b_this_page) != head);
-}
-
-
/*
* writepage: Called from one of two places:
*
*/
STATIC int
-linvfs_writepage(
+xfs_vm_writepage(
struct page *page,
struct writeback_control *wbc)
{
* then mark the page dirty again and leave the page
* as is.
*/
- if (PFLAGS_TEST_FSTRANS() && need_trans)
+ if (current_test_flags(PF_FSTRANS) && need_trans)
goto out_fail;
/*
}
STATIC int
-linvfs_invalidate_page(
- struct page *page,
- unsigned long offset)
+xfs_vm_writepages(
+ struct address_space *mapping,
+ struct writeback_control *wbc)
{
- xfs_page_trace(XFS_INVALIDPAGE_ENTER,
- page->mapping->host, page, offset);
- return block_invalidatepage(page, offset);
+ xfs_iflags_clear(XFS_I(mapping->host), XFS_ITRUNCATED);
+ return generic_writepages(mapping, wbc);
}
/*
* free them and we should come back later via writepage.
*/
STATIC int
-linvfs_release_page(
+xfs_vm_releasepage(
struct page *page,
gfp_t gfp_mask)
{
.nr_to_write = 1,
};
- xfs_page_trace(XFS_RELEASEPAGE_ENTER, inode, page, gfp_mask);
+ xfs_page_trace(XFS_RELEASEPAGE_ENTER, inode, page, 0);
+
+ if (!page_has_buffers(page))
+ return 0;
xfs_count_page_state(page, &delalloc, &unmapped, &unwritten);
if (!delalloc && !unwritten)
/* If we are already inside a transaction or the thread cannot
* do I/O, we cannot release this page.
*/
- if (PFLAGS_TEST_FSTRANS())
+ if (current_test_flags(PF_FSTRANS))
return 0;
/*
}
STATIC int
-linvfs_prepare_write(
+__xfs_get_blocks(
+ struct inode *inode,
+ sector_t iblock,
+ struct buffer_head *bh_result,
+ int create,
+ int direct,
+ bmapi_flags_t flags)
+{
+ xfs_iomap_t iomap;
+ xfs_off_t offset;
+ ssize_t size;
+ int niomap = 1;
+ int error;
+
+ offset = (xfs_off_t)iblock << inode->i_blkbits;
+ ASSERT(bh_result->b_size >= (1 << inode->i_blkbits));
+ size = bh_result->b_size;
+ error = xfs_iomap(XFS_I(inode), offset, size,
+ create ? flags : BMAPI_READ, &iomap, &niomap);
+ if (error)
+ return -error;
+ if (niomap == 0)
+ return 0;
+
+ if (iomap.iomap_bn != IOMAP_DADDR_NULL) {
+ /*
+ * 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 (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;
+
+ /*
+ * If we previously allocated a block out beyond eof and we are now
+ * coming back to use it then we will need to flag it as new even if it
+ * has a disk address.
+ *
+ * With sub-block writes into unwritten extents we also need to mark
+ * the buffer as new so that the unwritten parts of the buffer gets
+ * correctly zeroed.
+ */
+ if (create &&
+ ((!buffer_mapped(bh_result) && !buffer_uptodate(bh_result)) ||
+ (offset >= i_size_read(inode)) ||
+ (iomap.iomap_flags & (IOMAP_NEW|IOMAP_UNWRITTEN))))
+ set_buffer_new(bh_result);
+
+ if (iomap.iomap_flags & IOMAP_DELAY) {
+ BUG_ON(direct);
+ if (create) {
+ set_buffer_uptodate(bh_result);
+ set_buffer_mapped(bh_result);
+ set_buffer_delay(bh_result);
+ }
+ }
+
+ 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);
+ }
+
+ return 0;
+}
+
+int
+xfs_get_blocks(
+ struct inode *inode,
+ sector_t iblock,
+ struct buffer_head *bh_result,
+ int create)
+{
+ return __xfs_get_blocks(inode, iblock,
+ bh_result, create, 0, BMAPI_WRITE);
+}
+
+STATIC int
+xfs_get_blocks_direct(
+ struct inode *inode,
+ sector_t iblock,
+ struct buffer_head *bh_result,
+ int create)
+{
+ return __xfs_get_blocks(inode, iblock,
+ bh_result, create, 1, BMAPI_WRITE|BMAPI_DIRECT);
+}
+
+STATIC void
+xfs_end_io_direct(
+ struct kiocb *iocb,
+ loff_t offset,
+ ssize_t size,
+ void *private)
+{
+ xfs_ioend_t *ioend = iocb->private;
+
+ /*
+ * Non-NULL private data means we need to issue a transaction to
+ * convert a range from unwritten to written extents. This needs
+ * to happen from process context but aio+dio I/O completion
+ * happens from irq context so we need to defer it to a workqueue.
+ * This is not necessary for synchronous direct I/O, but we do
+ * it anyway to keep the code uniform and simpler.
+ *
+ * Well, if only it were that simple. Because synchronous direct I/O
+ * requires extent conversion to occur *before* we return to userspace,
+ * we have to wait for extent conversion to complete. Look at the
+ * iocb that has been passed to us to determine if this is AIO or
+ * not. If it is synchronous, tell xfs_finish_ioend() to kick the
+ * workqueue and wait for it to complete.
+ *
+ * The core direct I/O code might be changed to always call the
+ * completion handler in the future, in which case all this can
+ * go away.
+ */
+ ioend->io_offset = offset;
+ ioend->io_size = size;
+ if (ioend->io_type == IOMAP_READ) {
+ xfs_finish_ioend(ioend, 0);
+ } else if (private && size > 0) {
+ xfs_finish_ioend(ioend, is_sync_kiocb(iocb));
+ } else {
+ /*
+ * A direct I/O write ioend starts it's life in unwritten
+ * state in case they map an unwritten extent. This write
+ * didn't map an unwritten extent so switch it's completion
+ * handler.
+ */
+ INIT_WORK(&ioend->io_work, xfs_end_bio_written);
+ xfs_finish_ioend(ioend, 0);
+ }
+
+ /*
+ * blockdev_direct_IO can return an error even after the I/O
+ * completion handler was called. Thus we need to protect
+ * against double-freeing.
+ */
+ iocb->private = NULL;
+}
+
+STATIC ssize_t
+xfs_vm_direct_IO(
+ int rw,
+ struct kiocb *iocb,
+ const struct iovec *iov,
+ loff_t offset,
+ unsigned long nr_segs)
+{
+ struct file *file = iocb->ki_filp;
+ struct inode *inode = file->f_mapping->host;
+ struct block_device *bdev;
+ ssize_t ret;
+
+ bdev = xfs_find_bdev_for_inode(XFS_I(inode));
+
+ if (rw == WRITE) {
+ iocb->private = xfs_alloc_ioend(inode, IOMAP_UNWRITTEN);
+ ret = blockdev_direct_IO_own_locking(rw, iocb, inode,
+ bdev, iov, offset, nr_segs,
+ xfs_get_blocks_direct,
+ xfs_end_io_direct);
+ } else {
+ iocb->private = xfs_alloc_ioend(inode, IOMAP_READ);
+ ret = blockdev_direct_IO_no_locking(rw, iocb, inode,
+ bdev, iov, offset, nr_segs,
+ xfs_get_blocks_direct,
+ xfs_end_io_direct);
+ }
+
+ if (unlikely(ret != -EIOCBQUEUED && iocb->private))
+ xfs_destroy_ioend(iocb->private);
+ return ret;
+}
+
+STATIC int
+xfs_vm_write_begin(
struct file *file,
+ struct address_space *mapping,
+ loff_t pos,
+ unsigned len,
+ unsigned flags,
+ struct page **pagep,
+ void **fsdata)
+{
+ *pagep = NULL;
+ return block_write_begin(file, mapping, pos, len, flags, pagep, fsdata,
+ xfs_get_blocks);
+}
+
+STATIC sector_t
+xfs_vm_bmap(
+ struct address_space *mapping,
+ sector_t block)
+{
+ struct inode *inode = (struct inode *)mapping->host;
+ struct xfs_inode *ip = XFS_I(inode);
+
+ xfs_itrace_entry(XFS_I(inode));
+ xfs_ilock(ip, XFS_IOLOCK_SHARED);
+ xfs_flush_pages(ip, (xfs_off_t)0, -1, 0, FI_REMAPF);
+ xfs_iunlock(ip, XFS_IOLOCK_SHARED);
+ return generic_block_bmap(mapping, block, xfs_get_blocks);
+}
+
+STATIC int
+xfs_vm_readpage(
+ struct file *unused,
+ struct page *page)
+{
+ return mpage_readpage(page, xfs_get_blocks);
+}
+
+STATIC int
+xfs_vm_readpages(
+ struct file *unused,
+ struct address_space *mapping,
+ struct list_head *pages,
+ unsigned nr_pages)
+{
+ return mpage_readpages(mapping, pages, nr_pages, xfs_get_blocks);
+}
+
+STATIC void
+xfs_vm_invalidatepage(
struct page *page,
- unsigned int from,
- unsigned int to)
+ unsigned long offset)
{
- return block_prepare_write(page, from, to, linvfs_get_block);
+ xfs_page_trace(XFS_INVALIDPAGE_ENTER,
+ page->mapping->host, page, offset);
+ block_invalidatepage(page, offset);
}
-struct address_space_operations linvfs_aops = {
- .readpage = linvfs_readpage,
- .readpages = linvfs_readpages,
- .writepage = linvfs_writepage,
+const struct address_space_operations xfs_address_space_operations = {
+ .readpage = xfs_vm_readpage,
+ .readpages = xfs_vm_readpages,
+ .writepage = xfs_vm_writepage,
+ .writepages = xfs_vm_writepages,
.sync_page = block_sync_page,
- .releasepage = linvfs_release_page,
- .invalidatepage = linvfs_invalidate_page,
- .prepare_write = linvfs_prepare_write,
- .commit_write = generic_commit_write,
- .bmap = linvfs_bmap,
- .direct_IO = linvfs_direct_IO,
+ .releasepage = xfs_vm_releasepage,
+ .invalidatepage = xfs_vm_invalidatepage,
+ .write_begin = xfs_vm_write_begin,
+ .write_end = generic_write_end,
+ .bmap = xfs_vm_bmap,
+ .direct_IO = xfs_vm_direct_IO,
.migratepage = buffer_migrate_page,
};