From: Omar Sandoval <osandov@xxxxxx>
As part of simplifying all of the private data passed around for direct
I/O, bh->b_private will no longer be passed to dio_iodone_t. iocb is
still available there, however, so convert ext4 to use it. Note that
ext4_file_write_iter() also uses iocb->private, but
ext4_direct_IO_write() resets it to NULL after reading it.
Also note that the comment above ext4_should_dioread_nolock() is no
longer accurate. It seems that it should be possible to remove the data
journaling restriction now?
Cc: "Theodore Ts'o" <tytso@xxxxxxx>
Cc: Andreas Dilger <adilger.kernel@xxxxxxxxx>
Signed-off-by: Omar Sandoval <osandov@xxxxxx>
---
fs/ext4/inode.c | 10 ++++------
1 file changed, 4 insertions(+), 6 deletions(-)
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 18ad91b1c8f6..841d79919cef 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -884,18 +884,16 @@ static int ext4_dio_get_block_unwritten_async(struct kiocb *iocb,
/*
* When doing DIO using unwritten extents, we need io_end to convert
* unwritten extents to written on IO completion. We allocate io_end
- * once we spot unwritten extent and store it in b_private. Generic
- * DIO code keeps b_private set and furthermore passes the value to
- * our completion callback in 'private' argument.
+ * once we spot unwritten extent and store it in iocb->private.
*/
if (!ret && buffer_unwritten(bh_result)) {
- if (!bh_result->b_private) {
+ if (!iocb->private) {
ext4_io_end_t *io_end;
io_end = ext4_init_io_end(inode, GFP_KERNEL);
if (!io_end)
return -ENOMEM;
- bh_result->b_private = io_end;
+ iocb->private = io_end;
ext4_set_io_unwritten_flag(inode, io_end);
}
set_buffer_defer_completion(bh_result);
@@ -3617,7 +3615,7 @@ const struct iomap_ops ext4_iomap_ops = {
static int ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
ssize_t size, void *private)
{
- ext4_io_end_t *io_end = private;
+ ext4_io_end_t *io_end = iocb->private;
/* if not async direct IO just return */
if (!io_end)
--
2.18.0
