ext4: Fix return value for direct io reads at or beyond eof

Message ID 1518470282-7948-1-git-send-email-ashish.samant@oracle.com
State New
Headers show
Series
  • ext4: Fix return value for direct io reads at or beyond eof
Related show

Commit Message

Ashish Samant Feb. 12, 2018, 9:18 p.m.
Currently, an unaligned dio read at eof on ext4 returns EINVAL from
do_blockdev_direct_IO. However this unaligned dio read at eof could be due
to previous short read and we should return zero.

Fix this by checking for this condition in ext4_direct_IO_read()
and returning zero, if true.

Signed-off-by: Ashish Samant <ashish.samant@oracle.com>
---
 fs/ext4/inode.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

Comments

Ashish Samant April 10, 2018, 6:33 p.m. | #1
Hi,

Does this patch look ok? Gentle reminder for review.

Thanks,
Ashish

On 02/12/2018 01:18 PM, Ashish Samant wrote:
> Currently, an unaligned dio read at eof on ext4 returns EINVAL from
> do_blockdev_direct_IO. However this unaligned dio read at eof could be due
> to previous short read and we should return zero.
>
> Fix this by checking for this condition in ext4_direct_IO_read()
> and returning zero, if true.
>
> Signed-off-by: Ashish Samant <ashish.samant@oracle.com>
> ---
>   fs/ext4/inode.c | 4 +++-
>   1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index 534a913..8b3dbc0 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -3806,7 +3806,7 @@ static ssize_t ext4_direct_IO_read(struct kiocb *iocb, struct iov_iter *iter)
>   	struct address_space *mapping = iocb->ki_filp->f_mapping;
>   	struct inode *inode = mapping->host;
>   	size_t count = iov_iter_count(iter);
> -	ssize_t ret;
> +	ssize_t ret = 0;
>   
>   	/*
>   	 * Shared inode_lock is enough for us - it protects against concurrent
> @@ -3814,6 +3814,8 @@ static ssize_t ext4_direct_IO_read(struct kiocb *iocb, struct iov_iter *iter)
>   	 * we are protected against page writeback as well.
>   	 */
>   	inode_lock_shared(inode);
> +	if (iocb->ki_pos >= i_size_read(inode))
> +		goto out_unlock;
>   	ret = filemap_write_and_wait_range(mapping, iocb->ki_pos,
>   					   iocb->ki_pos + count - 1);
>   	if (ret)

Patch

diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 534a913..8b3dbc0 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -3806,7 +3806,7 @@  static ssize_t ext4_direct_IO_read(struct kiocb *iocb, struct iov_iter *iter)
 	struct address_space *mapping = iocb->ki_filp->f_mapping;
 	struct inode *inode = mapping->host;
 	size_t count = iov_iter_count(iter);
-	ssize_t ret;
+	ssize_t ret = 0;
 
 	/*
 	 * Shared inode_lock is enough for us - it protects against concurrent
@@ -3814,6 +3814,8 @@  static ssize_t ext4_direct_IO_read(struct kiocb *iocb, struct iov_iter *iter)
 	 * we are protected against page writeback as well.
 	 */
 	inode_lock_shared(inode);
+	if (iocb->ki_pos >= i_size_read(inode))
+		goto out_unlock;
 	ret = filemap_write_and_wait_range(mapping, iocb->ki_pos,
 					   iocb->ki_pos + count - 1);
 	if (ret)