[PATCH] generic/730: skip on ext4 when dax is enabled

[PATCH] generic/730: skip on ext4 when dax is enabled

[Disha Goel]

The generic/730 test utilizes the scsi_debug driver to create a
synthetic SCSI block device for testing. However, scsi_debug does
not support DAX capabilities.

When xfstests is configured with DAX mount options, ext4 strictly
validates the block device during the mount phase. This results
in a mount failure:
  EXT4-fs (sdc): DAX unsupported by block device.

Skip this test for ext4 when the 'dax' mount option is present to
avoid false failures.

Signed-off-by: Disha Goel <disgoel@linux.ibm.com>
---
 tests/generic/730 | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/tests/generic/730 b/tests/generic/730
index 6b5d3196..b42f5f63 100755
--- a/tests/generic/730
+++ b/tests/generic/730
@@ -20,6 +20,9 @@ _cleanup()
 . ./common/filter
 . ./common/scsi_debug
 
+if [ "$FSTYP" == "ext4" ] && echo "$MOUNT_OPTIONS" | grep -q "dax"; then
+	_notrun "ext4 does not support dax mounts on scsi_debug"
+fi
 
 # We don't actually use the test device, but we need a block based fs
 _require_test
-- 
2.45.1

Re: [PATCH] generic/730: skip on ext4 when dax is enabled

[Zorro Lang]

On Wed, Mar 18, 2026 at 01:48:12PM +0530, Disha Goel wrote:
> The generic/730 test utilizes the scsi_debug driver to create a
> synthetic SCSI block device for testing. However, scsi_debug does
> not support DAX capabilities.
> 
> When xfstests is configured with DAX mount options, ext4 strictly
> validates the block device during the mount phase. This results
> in a mount failure:
>   EXT4-fs (sdc): DAX unsupported by block device.
> 
> Skip this test for ext4 when the 'dax' mount option is present to
> avoid false failures.
> 
> Signed-off-by: Disha Goel <disgoel@linux.ibm.com>
> ---

I think this case isn't a DAX related test case, even give it "-o dax=always",
it won't test scsi_debug with that either. So why don't we simply skip the test
if dax is detected in MOUNT_OPTIONS?

  _exclude_scratch_mount_option dax

Any thoughts on this?

Thanks,
Zorro

>  tests/generic/730 | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/tests/generic/730 b/tests/generic/730
> index 6b5d3196..b42f5f63 100755
> --- a/tests/generic/730
> +++ b/tests/generic/730
> @@ -20,6 +20,9 @@ _cleanup()
>  . ./common/filter
>  . ./common/scsi_debug
>  
> +if [ "$FSTYP" == "ext4" ] && echo "$MOUNT_OPTIONS" | grep -q "dax"; then
> +	_notrun "ext4 does not support dax mounts on scsi_debug"
> +fi
>  
>  # We don't actually use the test device, but we need a block based fs
>  _require_test
> -- 
> 2.45.1
> 

Re: [PATCH] generic/730: skip on ext4 when dax is enabled

[Christoph Hellwig]

On Fri, Mar 20, 2026 at 11:56:44PM +0800, Zorro Lang wrote:
> I think this case isn't a DAX related test case, even give it "-o dax=always",
> it won't test scsi_debug with that either. So why don't we simply skip the test
> if dax is detected in MOUNT_OPTIONS?
> 
>   _exclude_scratch_mount_option dax
> 
> Any thoughts on this?

Agreed to all of the above.

Re: [PATCH] generic/730: skip on ext4 when dax is enabled

[Disha Goel]

On 23/03/26 11:37 am, Christoph Hellwig wrote:
> On Fri, Mar 20, 2026 at 11:56:44PM +0800, Zorro Lang wrote:
>> I think this case isn't a DAX related test case, even give it "-o dax=always",
>> it won't test scsi_debug with that either. So why don't we simply skip the test
>> if dax is detected in MOUNT_OPTIONS?
>>
>>    _exclude_scratch_mount_option dax
>>
>> Any thoughts on this?
> 
> Agreed to all of the above.
> 
Thanks for the review. That makes sense.

I will send a v2 using _exclude_scratch_mount_option dax.

-- 
Regards,
Disha