Re: [PATCH] virtiofs_submounts.py test: Note on vmlinuz param

[Max Reitz <mreitz@redhat.com>]

On 12.02.21 19:58, Cleber Rosa wrote:
> On Fri, Feb 12, 2021 at 04:16:49PM +0100, Max Reitz wrote:
>>  From the cancel message, it is not entirely clear why this parameter is
>> mandatory now, or that it will be optional in the future.  Add such a
>> more detailed explanation as a comment in the test source file.
>>
>> Suggested-by: Alex Bennée <alex.bennee@linaro.org>
>> Signed-off-by: Max Reitz <mreitz@redhat.com>
>> ---
>> I’ve uploaded a build of Linux 5.10 here:
>>    https://xanclic.moe/linux-5.10
>>
>> But I’ve decided against mentioning it in this new comment or the cancel
>> message, because, well, it’s my private server and I have limited
>> bandwidth.
>> ---
>>   tests/acceptance/virtiofs_submounts.py | 12 ++++++++++++
>>   1 file changed, 12 insertions(+)
>>
>> diff --git a/tests/acceptance/virtiofs_submounts.py b/tests/acceptance/virtiofs_submounts.py
>> index 949ca87a83..9a69b6b17b 100644
>> --- a/tests/acceptance/virtiofs_submounts.py
>> +++ b/tests/acceptance/virtiofs_submounts.py
>> @@ -228,6 +228,18 @@ class VirtiofsSubmountsTest(BootLinux):
>>       def setUp(self):
>>           vmlinuz = self.params.get('vmlinuz')
>>           if vmlinuz is None:
>> +            """
>> +            The Linux kernel supports FUSE auto-submounts only as of 5.10.
>> +            boot_linux.py currently provides Fedora 31, whose kernel is too
>> +            old, so this test cannot pass with the on-image kernel (you are
>> +            welcome to try, hence the option to force such a test with
>> +            -p vmlinuz='').  Therefore, for now the user must provide a
>> +            sufficiently new custom kernel, or effectively explicitly
>> +            request failure with -p vmlinuz=''.
>> +            Once an image with a sufficiently new kernel is available
>> +            (probably Fedora 34), we can make -p vmlinuz='' the default, so
>> +            that this parameter no longer needs to be specified.
>> +            """
>>               self.cancel('vmlinuz parameter not set; you must point it to a '
>>                           'Linux kernel binary to test (to run this test with ' \
>>                           'the on-image kernel, set it to an empty string)')
>> -- 
>> 2.29.2
>>
> 
> Hi Max,
> 
> This looks good to me, and I've also tested your kernel build and
> works like a charm.

Great :)

> As further work on top of this, it may be beneficial to have test
> documentation in a predictable place.  The possibilities that come to
> my mind:
> 
>   * docs/devel/testing.rst
>   * tests/acceptance/$test_file.py/data/README

Hm.  I think I’d prefer a dedicated file, i.e. the second one.  In any 
case, sounds good.

> On a different topic, the "https://avocado-project.org/data/assets" has
> enough bandwidth and can be used to hold this type asset.

I think it can’t hurt to put the kernel there, and then link to it in 
the cancel message.

> Alternatively,
> we can add a bit more automation to this test by letting people do something
> like:
> 
>   $ avocado assets register virtiofs-auto-submounts-vmlinuz /path/to/vmlinuz
> 
> And on the test:
> 
>   vmlinuz = self.fetch_asset('virtiofs-auto-submounts-vmlinuz')
> 
> And the test should cancel if that asset has not been previously registered.

Thanks, I wasn’t aware of this system.  Though I have no idea how 
“assets register” works, I can’t seem to find doc on it on
avocado-framework.readthedocs.io.  Is it global?  Is there a way to 
register an asset of a specific name only for a specific test?

Because I think this would make more sense if we tried to fetch the 
asset from e.g. https://avocado-project.org/data/assets, i.e. just put 
its name as linux-5.10.  Though perhaps it would also work with 
name='virtiofs-auto-submounts-vmlinuz', as you suggested.  But in any 
case, I’m a bit uneasy about a global namespace, which “assets register” 
seems to use.

Max