[PATCH v12 2/2] xfs: add FALLOC_FL_WRITE_ZEROES to XFS code base

[PATCH v12 2/2] xfs: add FALLOC_FL_WRITE_ZEROES to XFS code base

[Lukas Herbolt]

Add support for FALLOC_FL_WRITE_ZEROES if the underlying device
enable the unmap write zeroes operation.

Co-developed-by: Pankaj Raghav <p.raghav@samsung.com>
Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
Signed-off-by: Lukas Herbolt <lukas@herbolt.com>
---
 v12 changes:
	split from xfs_falloc_zero_range() into separate function
 fs/xfs/xfs_file.c | 49 ++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 48 insertions(+), 1 deletion(-)

diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c
index fd049a1fc9c6..ede7be05d83e 100644
--- a/fs/xfs/xfs_file.c
+++ b/fs/xfs/xfs_file.c
@@ -1318,6 +1318,49 @@ xfs_falloc_zero_range(
 	return xfs_falloc_setsize(file, new_size);
 }
 
+static int
+xfs_falloc_write_zero_range(
+	struct file		*file,
+	int			mode,
+	loff_t			offset,
+	loff_t			len,
+	struct xfs_zone_alloc_ctx *ac)
+{
+	struct inode		*inode = file_inode(file);
+	struct xfs_inode	*ip = XFS_I(inode);
+	unsigned int		blksize = i_blocksize(inode);
+	loff_t			new_size = 0;
+	int			error;
+
+	if (xfs_is_always_cow_inode(ip) ||
+	    !bdev_write_zeroes_unmap_sectors(
+		    xfs_inode_buftarg(ip)->bt_bdev))
+		return -EOPNOTSUPP;
+
+	trace_xfs_zero_file_space(ip);
+
+	error = xfs_falloc_newsize(file, mode, offset, len, &new_size);
+	if (error)
+		return error;
+
+	error = xfs_free_file_space(ip, offset, len, ac);
+	if (error)
+		return error;
+
+	len = round_up(offset + len, blksize) - round_down(offset, blksize);
+	offset = round_down(offset, blksize);
+	error = xfs_alloc_file_space(ip, offset, len, XFS_BMAPI_PREALLOC);
+	if (error)
+		return error;
+
+	error = xfs_falloc_setsize(file, new_size);
+	if (error)
+		return error;
+
+	return xfs_alloc_file_space(ip, offset, len,
+			XFS_BMAPI_CONVERT | XFS_BMAPI_ZERO);
+}
+
 static int
 xfs_falloc_unshare_range(
 	struct file		*file,
@@ -1377,7 +1420,7 @@ xfs_falloc_allocate_range(
 		(FALLOC_FL_ALLOCATE_RANGE | FALLOC_FL_KEEP_SIZE |	\
 		 FALLOC_FL_PUNCH_HOLE |	FALLOC_FL_COLLAPSE_RANGE |	\
 		 FALLOC_FL_ZERO_RANGE |	FALLOC_FL_INSERT_RANGE |	\
-		 FALLOC_FL_UNSHARE_RANGE)
+		 FALLOC_FL_UNSHARE_RANGE | FALLOC_FL_WRITE_ZEROES)
 
 STATIC long
 __xfs_file_fallocate(
@@ -1423,6 +1466,10 @@ __xfs_file_fallocate(
 	case FALLOC_FL_ZERO_RANGE:
 		error = xfs_falloc_zero_range(file, mode, offset, len, ac);
 		break;
+	case FALLOC_FL_WRITE_ZEROES:
+		error = xfs_falloc_write_zero_range(file, mode, offset,
+				len, ac);
+		break;
 	case FALLOC_FL_UNSHARE_RANGE:
 		error = xfs_falloc_unshare_range(file, mode, offset, len);
 		break;
-- 
2.53.0

Re: [PATCH v12 2/2] xfs: add FALLOC_FL_WRITE_ZEROES to XFS code base

[Darrick J. Wong]

On Tue, Mar 10, 2026 at 08:42:46PM +0100, Lukas Herbolt wrote:
> Add support for FALLOC_FL_WRITE_ZEROES if the underlying device
> enable the unmap write zeroes operation.
> 
> Co-developed-by: Pankaj Raghav <p.raghav@samsung.com>
> Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
> Signed-off-by: Lukas Herbolt <lukas@herbolt.com>
> ---
>  v12 changes:
> 	split from xfs_falloc_zero_range() into separate function
>  fs/xfs/xfs_file.c | 49 ++++++++++++++++++++++++++++++++++++++++++++++-
>  1 file changed, 48 insertions(+), 1 deletion(-)
> 
> diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c
> index fd049a1fc9c6..ede7be05d83e 100644
> --- a/fs/xfs/xfs_file.c
> +++ b/fs/xfs/xfs_file.c
> @@ -1318,6 +1318,49 @@ xfs_falloc_zero_range(
>  	return xfs_falloc_setsize(file, new_size);
>  }
>  
> +static int
> +xfs_falloc_write_zero_range(
> +	struct file		*file,
> +	int			mode,
> +	loff_t			offset,
> +	loff_t			len,
> +	struct xfs_zone_alloc_ctx *ac)
> +{
> +	struct inode		*inode = file_inode(file);
> +	struct xfs_inode	*ip = XFS_I(inode);
> +	unsigned int		blksize = i_blocksize(inode);
> +	loff_t			new_size = 0;
> +	int			error;
> +
> +	if (xfs_is_always_cow_inode(ip) ||
> +	    !bdev_write_zeroes_unmap_sectors(
> +		    xfs_inode_buftarg(ip)->bt_bdev))
> +		return -EOPNOTSUPP;
> +
> +	trace_xfs_zero_file_space(ip);

I think we want to have a separate tracepoint for write-zeroes, but TBH
the fallocate tracepoints all kinda suck in that we don't record the
range or any of the flags so... there's a minor cleanup to be done
separately.

In the meantime this looks ok provided it doesn't break fsx/fsstress?
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>

--D

> +
> +	error = xfs_falloc_newsize(file, mode, offset, len, &new_size);
> +	if (error)
> +		return error;
> +
> +	error = xfs_free_file_space(ip, offset, len, ac);
> +	if (error)
> +		return error;
> +
> +	len = round_up(offset + len, blksize) - round_down(offset, blksize);
> +	offset = round_down(offset, blksize);
> +	error = xfs_alloc_file_space(ip, offset, len, XFS_BMAPI_PREALLOC);
> +	if (error)
> +		return error;
> +
> +	error = xfs_falloc_setsize(file, new_size);
> +	if (error)
> +		return error;
> +
> +	return xfs_alloc_file_space(ip, offset, len,
> +			XFS_BMAPI_CONVERT | XFS_BMAPI_ZERO);
> +}
> +
>  static int
>  xfs_falloc_unshare_range(
>  	struct file		*file,
> @@ -1377,7 +1420,7 @@ xfs_falloc_allocate_range(
>  		(FALLOC_FL_ALLOCATE_RANGE | FALLOC_FL_KEEP_SIZE |	\
>  		 FALLOC_FL_PUNCH_HOLE |	FALLOC_FL_COLLAPSE_RANGE |	\
>  		 FALLOC_FL_ZERO_RANGE |	FALLOC_FL_INSERT_RANGE |	\
> -		 FALLOC_FL_UNSHARE_RANGE)
> +		 FALLOC_FL_UNSHARE_RANGE | FALLOC_FL_WRITE_ZEROES)
>  
>  STATIC long
>  __xfs_file_fallocate(
> @@ -1423,6 +1466,10 @@ __xfs_file_fallocate(
>  	case FALLOC_FL_ZERO_RANGE:
>  		error = xfs_falloc_zero_range(file, mode, offset, len, ac);
>  		break;
> +	case FALLOC_FL_WRITE_ZEROES:
> +		error = xfs_falloc_write_zero_range(file, mode, offset,
> +				len, ac);
> +		break;
>  	case FALLOC_FL_UNSHARE_RANGE:
>  		error = xfs_falloc_unshare_range(file, mode, offset, len);
>  		break;
> -- 
> 2.53.0
> 
> 

Re: [PATCH v12 2/2] xfs: add FALLOC_FL_WRITE_ZEROES to XFS code base

[Dave Chinner]

On Tue, Mar 10, 2026 at 08:42:46PM +0100, Lukas Herbolt wrote:
> Add support for FALLOC_FL_WRITE_ZEROES if the underlying device
> enable the unmap write zeroes operation.
> 
> Co-developed-by: Pankaj Raghav <p.raghav@samsung.com>
> Signed-off-by: Pankaj Raghav <p.raghav@samsung.com>
> Signed-off-by: Lukas Herbolt <lukas@herbolt.com>
> ---
>  v12 changes:
> 	split from xfs_falloc_zero_range() into separate function
>  fs/xfs/xfs_file.c | 49 ++++++++++++++++++++++++++++++++++++++++++++++-
>  1 file changed, 48 insertions(+), 1 deletion(-)

Ugh, I just reviewed v11 and pointed out how unnecessarily complex
this new "three phase" method is.  Please given people a chance to
review the code by waiting a couple of days before posting a new
version....

My comments against v11 are still relevant; working around low level
truncate issues when calling truncate to transactionally update the
file size is not necessary is less than ideal.

-Dave.
-- 
Dave Chinner
dgc@kernel.org