Re: [PATCH v3] tests/functional: Convert most Aspeed machine tests

["Cédric Le Goater" <clg@redhat.com>]

On 10/21/24 11:47, Daniel P. Berrangé wrote:
> On Mon, Oct 21, 2024 at 11:31:03AM +0200, Cédric Le Goater wrote:
>> Hello Thomas,
>>
>>
>> On 10/21/24 10:53, Thomas Huth wrote:
>>> On 18/10/2024 09.17, Cédric Le Goater wrote:
>>>> This is a simple conversion of the tests with some cleanups and
>>>> adjustments to match the new test framework. Replace the zephyr image
>>>> MD5 hashes with SHA256 hashes while at it.
>>>>
>>>> The SDK tests depend on a ssh class from avocado.utils which is
>>>> difficult to replace. To be addressed separately.
>>>>
>>>> Signed-off-by: Cédric Le Goater <clg@redhat.com>
>>>> ---
>>>
>>>    Hi Cédric,
>>>
>>> I'm seeing errors when I try to run "make check-functional" with this patch multiple times, e.g.:
>>>
>>>     File "/home/thuth/devel/qemu/tests/functional/test_arm_aspeed.py", line 111, in test_arm_ast2400_palmetto_openbmc_v2_9_0
>>>       image_path = self.ASSET_PALMETTO_FLASH.fetch()
>>>                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>>>     File "/home/thuth/devel/qemu/tests/functional/qemu_test/asset.py", line 104, in fetch
>>>       raise Exception("Asset cache is invalid and downloads disabled")
>>> Exception: Asset cache is invalid and downloads disabled
>>>
>>> Could it be that the tests alter the file in the download cache?
>>
>> yes. the flash image is updated by Linux. I don't reproduce the issue
>> on my build though. Are you using special options ?
> 
> NB, this would only happen if invoked via 'make check-functional' as
> that has the logic to run asset downloads seprately from the tests.
> If you were just directly invoking the test program you wouldn't see
> this error message, as it'll just re-download.
> 
> You would have to run the test twice too I guess, first time downloads
> and modifies the image, second time sees the modified image and
> complains.

.cache/qemu/download/3e13bbbc28e424865dc42f35ad672b10f2e82cdb11846bb28fa625b48beafd0d

is indeed updated. check-functional-arm doesn't complain (which I was using),
check-functional does complain.
  
I will add '-snapshot' to the machine options.

Thanks,

C.