#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>
/*
* 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);
+ }
}
/*
xfs_fsize_t bsize;
ip = xfs_vtoi(ioend->io_vnode);
+ if (!ip)
+ return;
ASSERT((ip->i_d.di_mode & S_IFMT) == S_IFREG);
ASSERT(ioend->io_type != IOMAP_READ);
ip->i_d.di_size = isize;
ip->i_update_core = 1;
ip->i_update_size = 1;
+ mark_inode_dirty_sync(vn_to_inode(ioend->io_vnode));
}
xfs_iunlock(ip, XFS_ILOCK_EXCL);
size_t size = ioend->io_size;
if (likely(!ioend->io_error)) {
- bhv_vop_bmap(vp, offset, size, BMAPI_UNWRITTEN, NULL, NULL);
+ xfs_bmap(xfs_vtoi(vp), offset, size,
+ BMAPI_UNWRITTEN, NULL, NULL);
xfs_setfilesize(ioend);
}
xfs_destroy_ioend(ioend);
xfs_iomap_t *mapp,
int flags)
{
- bhv_vnode_t *vp = vn_from_inode(inode);
+ xfs_inode_t *ip = XFS_I(inode);
int error, nmaps = 1;
- error = bhv_vop_bmap(vp, offset, count, flags, mapp, &nmaps);
+ error = xfs_bmap(ip, offset, count,
+ flags, mapp, &nmaps);
if (!error && (flags & (BMAPI_WRITE|BMAPI_ALLOCATE)))
- VMODIFY(vp);
+ xfs_iflags_set(ip, XFS_IMODIFIED);
return -error;
}
bio->bi_end_io = NULL;
bio_put(bio);
- xfs_finish_ioend(ioend);
+ xfs_finish_ioend(ioend, 0);
return 0;
}
}
if (bio)
xfs_submit_ioend_bio(ioend, bio);
- xfs_finish_ioend(ioend);
+ xfs_finish_ioend(ioend, 0);
} while ((ioend = next) != NULL);
}
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 =
pg_offset = PAGE_CACHE_SIZE;
if (page->index == tindex && !TestSetPageLocked(page)) {
- len = xfs_probe_page(page, pg_offset, mapped);
+ 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++;
}
else if (buffer_delay(bh))
acceptable = (type == IOMAP_DELAY);
else if (buffer_dirty(bh) && buffer_mapped(bh))
- acceptable = (type == 0);
+ acceptable = (type == IOMAP_NEW);
else
break;
} while ((bh = bh->b_this_page) != head);
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,
bh = head = page_buffers(page);
offset = page_offset(page);
- flags = -1;
- type = IOMAP_READ;
+ 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)) {
}
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 != IOMAP_READ) {
+ 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 = IOMAP_READ;
+ /*
+ * 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)
struct address_space *mapping,
struct writeback_control *wbc)
{
- struct bhv_vnode *vp = vn_from_inode(mapping->host);
-
- if (VN_TRUNC(vp))
- VUNTRUNCATE(vp);
+ xfs_iflags_clear(XFS_I(mapping->host), XFS_ITRUNCATED);
return generic_writepages(mapping, wbc);
}
int direct,
bmapi_flags_t flags)
{
- bhv_vnode_t *vp = vn_from_inode(inode);
xfs_iomap_t iomap;
xfs_off_t offset;
ssize_t size;
offset = (xfs_off_t)iblock << inode->i_blkbits;
ASSERT(bh_result->b_size >= (1 << inode->i_blkbits));
size = bh_result->b_size;
- error = bhv_vop_bmap(vp, offset, size,
+ error = xfs_bmap(XFS_I(inode), offset, size,
create ? flags : BMAPI_READ, &iomap, &niomap);
if (error)
return -error;
* 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);
+ xfs_finish_ioend(ioend, 0);
} else if (private && size > 0) {
- xfs_finish_ioend(ioend);
+ xfs_finish_ioend(ioend, is_sync_kiocb(iocb));
} else {
/*
* A direct I/O write ioend starts it's life in unwritten
* handler.
*/
INIT_WORK(&ioend->io_work, xfs_end_bio_written);
- xfs_finish_ioend(ioend);
+ xfs_finish_ioend(ioend, 0);
}
/*
{
struct file *file = iocb->ki_filp;
struct inode *inode = file->f_mapping->host;
- bhv_vnode_t *vp = vn_from_inode(inode);
xfs_iomap_t iomap;
int maps = 1;
int error;
ssize_t ret;
- error = bhv_vop_bmap(vp, offset, 0, BMAPI_DEVICE, &iomap, &maps);
+ error = xfs_bmap(XFS_I(inode), offset, 0,
+ BMAPI_DEVICE, &iomap, &maps);
if (error)
return -error;
sector_t block)
{
struct inode *inode = (struct inode *)mapping->host;
- bhv_vnode_t *vp = vn_from_inode(inode);
+ struct xfs_inode *ip = XFS_I(inode);
- vn_trace_entry(vp, __FUNCTION__, (inst_t *)__return_address);
- bhv_vop_rwlock(vp, VRWLOCK_READ);
- bhv_vop_flush_pages(vp, (xfs_off_t)0, -1, 0, FI_REMAPF);
- bhv_vop_rwunlock(vp, VRWLOCK_READ);
+ vn_trace_entry(vn_from_inode(inode), __FUNCTION__,
+ (inst_t *)__return_address);
+ xfs_rwlock(ip, VRWLOCK_READ);
+ xfs_flush_pages(ip, (xfs_off_t)0, -1, 0, FI_REMAPF);
+ xfs_rwunlock(ip, VRWLOCK_READ);
return generic_block_bmap(mapping, block, xfs_get_blocks);
}