[PATCH v4] Return bytes transferred for partial direct I/O

[PATCH v4] Return bytes transferred for partial direct I/O

[Goldwyn Rodrigues]

From: Goldwyn Rodrigues <rgoldwyn@suse.com>

In case direct I/O encounters an error midway, it returns the error.
Instead it should be returning the number of bytes transferred so far.

Test case for filesystems (with ENOSPC):
1. Create an almost full filesystem
2. Create a file, say /mnt/lastfile, until the filesystem is full.
3. Direct write() with count > sizeof /mnt/lastfile.

Result: write() returns -ENOSPC. However, file content has data written
in step 3.

Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>

Changes since v1:
 - incorporated iomap and block devices

Changes since v2:
 - realized that file size was not increasing when performing a (partial)
   direct I/O because end_io function was receiving the error instead of
   size. Fixed.

Changes since v3:
 - [hch] initialize transferred with dio->size and use transferred instead
   of dio->size.

---
 fs/block_dev.c |  2 +-
 fs/direct-io.c |  4 +---
 fs/iomap.c     | 20 ++++++++++----------
 3 files changed, 12 insertions(+), 14 deletions(-)

diff --git a/fs/block_dev.c b/fs/block_dev.c
index 93d088ffc05c..1c6640ffb929 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -421,7 +421,7 @@ __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, int nr_pages)
 
 	if (!ret)
 		ret = blk_status_to_errno(dio->bio.bi_status);
-	if (likely(!ret))
+	if (likely(dio->size))
 		ret = dio->size;
 
 	bio_put(&dio->bio);
diff --git a/fs/direct-io.c b/fs/direct-io.c
index 5fa2211e49ae..0fc1789498ae 100644
--- a/fs/direct-io.c
+++ b/fs/direct-io.c
@@ -255,8 +255,6 @@ static ssize_t dio_complete(struct dio *dio, ssize_t ret, bool is_async)
 		ret = dio->page_errors;
 	if (ret == 0)
 		ret = dio->io_error;
-	if (ret == 0)
-		ret = transferred;
 
 	if (dio->end_io) {
 		int err;
@@ -284,7 +282,7 @@ static ssize_t dio_complete(struct dio *dio, ssize_t ret, bool is_async)
 	}
 
 	kmem_cache_free(dio_cache, dio);
-	return ret;
+	return transferred ? transferred : ret;
 }
 
 static void dio_aio_complete_work(struct work_struct *work)
diff --git a/fs/iomap.c b/fs/iomap.c
index 269b24a01f32..0e046e0a32ff 100644
--- a/fs/iomap.c
+++ b/fs/iomap.c
@@ -713,29 +713,29 @@ struct iomap_dio {
 static ssize_t iomap_dio_complete(struct iomap_dio *dio)
 {
 	struct kiocb *iocb = dio->iocb;
-	ssize_t ret;
+	ssize_t err;
+	ssize_t transferred = dio->size;
 
 	if (dio->end_io) {
-		ret = dio->end_io(iocb,
-				dio->error ? dio->error : dio->size,
+		err = dio->end_io(iocb,
+				transferred ? transferred : dio->error,
 				dio->flags);
 	} else {
-		ret = dio->error;
+		err = dio->error;
 	}
 
-	if (likely(!ret)) {
-		ret = dio->size;
+	if (likely(transferred)) {
 		/* check for short read */
-		if (iocb->ki_pos + ret > dio->i_size &&
+		if (iocb->ki_pos + transferred > dio->i_size &&
 		    !(dio->flags & IOMAP_DIO_WRITE))
-			ret = dio->i_size - iocb->ki_pos;
-		iocb->ki_pos += ret;
+			transferred = dio->i_size - iocb->ki_pos;
+		iocb->ki_pos += transferred;
 	}
 
 	inode_dio_end(file_inode(iocb->ki_filp));
 	kfree(dio);
 
-	return ret;
+	return transferred ? transferred : err;
 }
 
 static void iomap_dio_complete_work(struct work_struct *work)
-- 
2.14.1

Re: [PATCH v4] Return bytes transferred for partial direct I/O

[Goldwyn Rodrigues]

Al,

A gentle reminder... Could you please include this patch if you have no
objections to it?

On 10/01/2017 02:51 PM, Goldwyn Rodrigues wrote:
> From: Goldwyn Rodrigues <rgoldwyn@suse.com>
> 
> In case direct I/O encounters an error midway, it returns the error.
> Instead it should be returning the number of bytes transferred so far.
> 
> Test case for filesystems (with ENOSPC):
> 1. Create an almost full filesystem
> 2. Create a file, say /mnt/lastfile, until the filesystem is full.
> 3. Direct write() with count > sizeof /mnt/lastfile.
> 
> Result: write() returns -ENOSPC. However, file content has data written
> in step 3.
> 
> Signed-off-by: Goldwyn Rodrigues <rgoldwyn@suse.com>
> Reviewed-by: Christoph Hellwig <hch@lst.de>
> 
> Changes since v1:
>  - incorporated iomap and block devices
> 
> Changes since v2:
>  - realized that file size was not increasing when performing a (partial)
>    direct I/O because end_io function was receiving the error instead of
>    size. Fixed.
> 
> Changes since v3:
>  - [hch] initialize transferred with dio->size and use transferred instead
>    of dio->size.
> 
> ---
>  fs/block_dev.c |  2 +-
>  fs/direct-io.c |  4 +---
>  fs/iomap.c     | 20 ++++++++++----------
>  3 files changed, 12 insertions(+), 14 deletions(-)
> 
> diff --git a/fs/block_dev.c b/fs/block_dev.c
> index 93d088ffc05c..1c6640ffb929 100644
> --- a/fs/block_dev.c
> +++ b/fs/block_dev.c
> @@ -421,7 +421,7 @@ __blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, int nr_pages)
>  
>  	if (!ret)
>  		ret = blk_status_to_errno(dio->bio.bi_status);
> -	if (likely(!ret))
> +	if (likely(dio->size))
>  		ret = dio->size;
>  
>  	bio_put(&dio->bio);
> diff --git a/fs/direct-io.c b/fs/direct-io.c
> index 5fa2211e49ae..0fc1789498ae 100644
> --- a/fs/direct-io.c
> +++ b/fs/direct-io.c
> @@ -255,8 +255,6 @@ static ssize_t dio_complete(struct dio *dio, ssize_t ret, bool is_async)
>  		ret = dio->page_errors;
>  	if (ret == 0)
>  		ret = dio->io_error;
> -	if (ret == 0)
> -		ret = transferred;
>  
>  	if (dio->end_io) {
>  		int err;
> @@ -284,7 +282,7 @@ static ssize_t dio_complete(struct dio *dio, ssize_t ret, bool is_async)
>  	}
>  
>  	kmem_cache_free(dio_cache, dio);
> -	return ret;
> +	return transferred ? transferred : ret;
>  }
>  
>  static void dio_aio_complete_work(struct work_struct *work)
> diff --git a/fs/iomap.c b/fs/iomap.c
> index 269b24a01f32..0e046e0a32ff 100644
> --- a/fs/iomap.c
> +++ b/fs/iomap.c
> @@ -713,29 +713,29 @@ struct iomap_dio {
>  static ssize_t iomap_dio_complete(struct iomap_dio *dio)
>  {
>  	struct kiocb *iocb = dio->iocb;
> -	ssize_t ret;
> +	ssize_t err;
> +	ssize_t transferred = dio->size;
>  
>  	if (dio->end_io) {
> -		ret = dio->end_io(iocb,
> -				dio->error ? dio->error : dio->size,
> +		err = dio->end_io(iocb,
> +				transferred ? transferred : dio->error,
>  				dio->flags);
>  	} else {
> -		ret = dio->error;
> +		err = dio->error;
>  	}
>  
> -	if (likely(!ret)) {
> -		ret = dio->size;
> +	if (likely(transferred)) {
>  		/* check for short read */
> -		if (iocb->ki_pos + ret > dio->i_size &&
> +		if (iocb->ki_pos + transferred > dio->i_size &&
>  		    !(dio->flags & IOMAP_DIO_WRITE))
> -			ret = dio->i_size - iocb->ki_pos;
> -		iocb->ki_pos += ret;
> +			transferred = dio->i_size - iocb->ki_pos;
> +		iocb->ki_pos += transferred;
>  	}
>  
>  	inode_dio_end(file_inode(iocb->ki_filp));
>  	kfree(dio);
>  
> -	return ret;
> +	return transferred ? transferred : err;
>  }
>  
>  static void iomap_dio_complete_work(struct work_struct *work)
> 

-- 
Goldwyn