Re: [PATCH v5 3/3] generic/471: add syncfs test

[cgxu519 <cgxu519@gmx.com>]

Hi Amir,

It seems godown will cause side effect on ext4/tune2fs, mount an ext4 
filesystem after running godown,
then tune2fs -l <dev> complains like below. so definitely it will not 
pass d_type check.


[root@x201 xfstests-dev]# tune2fs -l /dev/mapper/test-test1
tune2fs 1.44.2 (14-May-2018)
tune2fs: Superblock checksum does not match superblock while trying to 
open /dev/mapper/test-test1
Couldn't find valid filesystem superblock.


Thanks,
Chengguang


On 08/30/2018 11:12 AM, cgxu519 wrote:
> Hi Amir,
>
> I think I tested both xfs/ext4. Let me have a look at the code for 
> detail.
>
>
> Thanks,
>
> Chengguang
>
>
> On 08/26/2018 10:53 PM, Amir Goldstein wrote:
>> On Fri, Dec 15, 2017 at 9:48 AM Chengguang Xu <cgxu519@icloud.com> 
>> wrote:
>>> Inspired by syncfs bug of overlayfs which does not sync dirtyinodes in
>>> underlying filesystem.
>>> Run syncfs and shutdown filesystem(or underlying filesystem of 
>>> overlayfs)
>>> to check syncfs result.
>>>
>>> Signed-off-by: Chengguang Xu <cgxu519@icloud.com>
>>> ---
>> Hi Chengguang,
>>
>> I just noticed that this test (now generic/474) is skipped for 
>> overlay over ext4
>> for a strange looking reason - it seems that in the overlay_mount() 
>> call from
>> _scratch_cycle_mount() the function hits
>> _notrun "upper fs needs to support d_type".
>>
>> This doesn't happen on my test system with overlay over xfs.
>> Did you happen to test this with overlay over ext4 at the time?
>> Will you have time to look into this?
>>
>> Thanks,
>> Amir.
>