summaryrefslogtreecommitdiffstats
path: root/fs/xfs/xfs_aops.c
diff options
context:
space:
mode:
Diffstat (limited to 'fs/xfs/xfs_aops.c')
-rw-r--r--fs/xfs/xfs_aops.c96
1 files changed, 50 insertions, 46 deletions
diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
index e4fff5898c1c..29e7e5dd5178 100644
--- a/fs/xfs/xfs_aops.c
+++ b/fs/xfs/xfs_aops.c
@@ -1259,13 +1259,28 @@ xfs_vm_releasepage(
* the DIO. There is only going to be one reference to the ioend and its life
* cycle is constrained by the DIO completion code. hence we don't need
* reference counting here.
+ *
+ * Note that for DIO, an IO to the highest supported file block offset (i.e.
+ * 2^63 - 1FSB bytes) will result in the offset + count overflowing a signed 64
+ * bit variable. Hence if we see this overflow, we have to assume that the IO is
+ * extending the file size. We won't know for sure until IO completion is run
+ * and the actual max write offset is communicated to the IO completion
+ * routine.
+ *
+ * For DAX page faults, we are preparing to never see unwritten extents here,
+ * nor should we ever extend the inode size. Hence we will soon have nothing to
+ * do here for this case, ensuring we don't have to provide an IO completion
+ * callback to free an ioend that we don't actually need for a fault into the
+ * page at offset (2^63 - 1FSB) bytes.
*/
+
static void
xfs_map_direct(
struct inode *inode,
struct buffer_head *bh_result,
struct xfs_bmbt_irec *imap,
- xfs_off_t offset)
+ xfs_off_t offset,
+ bool dax_fault)
{
struct xfs_ioend *ioend;
xfs_off_t size = bh_result->b_size;
@@ -1278,6 +1293,13 @@ xfs_map_direct(
trace_xfs_gbmap_direct(XFS_I(inode), offset, size, type, imap);
+ if (dax_fault) {
+ ASSERT(type == XFS_IO_OVERWRITE);
+ trace_xfs_gbmap_direct_none(XFS_I(inode), offset, size, type,
+ imap);
+ return;
+ }
+
if (bh_result->b_private) {
ioend = bh_result->b_private;
ASSERT(ioend->io_size > 0);
@@ -1292,7 +1314,8 @@ xfs_map_direct(
ioend->io_size, ioend->io_type,
imap);
} else if (type == XFS_IO_UNWRITTEN ||
- offset + size > i_size_read(inode)) {
+ offset + size > i_size_read(inode) ||
+ offset + size < 0) {
ioend = xfs_alloc_ioend(inode, type);
ioend->io_offset = offset;
ioend->io_size = size;
@@ -1354,7 +1377,8 @@ __xfs_get_blocks(
sector_t iblock,
struct buffer_head *bh_result,
int create,
- bool direct)
+ bool direct,
+ bool dax_fault)
{
struct xfs_inode *ip = XFS_I(inode);
struct xfs_mount *mp = ip->i_mount;
@@ -1402,10 +1426,12 @@ __xfs_get_blocks(
if (error)
goto out_unlock;
+ /* for DAX, we convert unwritten extents directly */
if (create &&
(!nimaps ||
(imap.br_startblock == HOLESTARTBLOCK ||
- imap.br_startblock == DELAYSTARTBLOCK))) {
+ imap.br_startblock == DELAYSTARTBLOCK) ||
+ (IS_DAX(inode) && ISUNWRITTEN(&imap)))) {
if (direct || xfs_get_extsz_hint(ip)) {
/*
* xfs_iomap_write_direct() expects the shared lock. It
@@ -1450,6 +1476,12 @@ __xfs_get_blocks(
goto out_unlock;
}
+ if (IS_DAX(inode) && create) {
+ ASSERT(!ISUNWRITTEN(&imap));
+ /* zeroing is not needed at a higher layer */
+ new = 0;
+ }
+
/* trim mapping down to size requested */
if (direct || size > (1 << inode->i_blkbits))
xfs_map_trim_size(inode, iblock, bh_result,
@@ -1467,7 +1499,8 @@ __xfs_get_blocks(
set_buffer_unwritten(bh_result);
/* direct IO needs special help */
if (create && direct)
- xfs_map_direct(inode, bh_result, &imap, offset);
+ xfs_map_direct(inode, bh_result, &imap, offset,
+ dax_fault);
}
/*
@@ -1514,7 +1547,7 @@ xfs_get_blocks(
struct buffer_head *bh_result,
int create)
{
- return __xfs_get_blocks(inode, iblock, bh_result, create, false);
+ return __xfs_get_blocks(inode, iblock, bh_result, create, false, false);
}
int
@@ -1524,7 +1557,17 @@ xfs_get_blocks_direct(
struct buffer_head *bh_result,
int create)
{
- return __xfs_get_blocks(inode, iblock, bh_result, create, true);
+ return __xfs_get_blocks(inode, iblock, bh_result, create, true, false);
+}
+
+int
+xfs_get_blocks_dax_fault(
+ struct inode *inode,
+ sector_t iblock,
+ struct buffer_head *bh_result,
+ int create)
+{
+ return __xfs_get_blocks(inode, iblock, bh_result, create, true, true);
}
static void
@@ -1623,45 +1666,6 @@ xfs_end_io_direct_write(
__xfs_end_io_direct_write(inode, ioend, offset, size);
}
-/*
- * For DAX we need a mapping buffer callback for unwritten extent conversion
- * when page faults allocate blocks and then zero them. Note that in this
- * case the mapping indicated by the ioend may extend beyond EOF. We most
- * definitely do not want to extend EOF here, so we trim back the ioend size to
- * EOF.
- */
-#ifdef CONFIG_FS_DAX
-void
-xfs_end_io_dax_write(
- struct buffer_head *bh,
- int uptodate)
-{
- struct xfs_ioend *ioend = bh->b_private;
- struct inode *inode = ioend->io_inode;
- ssize_t size = ioend->io_size;
-
- ASSERT(IS_DAX(ioend->io_inode));
-
- /* if there was an error zeroing, then don't convert it */
- if (!uptodate)
- ioend->io_error = -EIO;
-
- /*
- * Trim update to EOF, so we don't extend EOF during unwritten extent
- * conversion of partial EOF blocks.
- */
- spin_lock(&XFS_I(inode)->i_flags_lock);
- if (ioend->io_offset + size > i_size_read(inode))
- size = i_size_read(inode) - ioend->io_offset;
- spin_unlock(&XFS_I(inode)->i_flags_lock);
-
- __xfs_end_io_direct_write(inode, ioend, ioend->io_offset, size);
-
-}
-#else
-void xfs_end_io_dax_write(struct buffer_head *bh, int uptodate) { }
-#endif
-
static inline ssize_t
xfs_vm_do_dio(
struct inode *inode,