[PATCH 3/3] configs: enable UNIT_TEST on qemu_arm_defconfig

[Heinrich Schuchardt]

On 11/3/24 17:12, Tom Rini wrote:
> On Sun, Nov 03, 2024 at 04:56:40PM +0100, Heinrich Schuchardt wrote:
>> On 11/3/24 16:50, Tom Rini wrote:
>>> On Sun, Nov 03, 2024 at 04:45:41PM +0100, Heinrich Schuchardt wrote:
>>>> On 11/3/24 08:12, Heinrich Schuchardt wrote:
>>>>> The lib_test_uuid_to_le test fails on 32-bit systems. But we never caught
>>>>> this in our CI because we never ran any of our C unit tests on 32-bit.
>>>>>
>>>>> Enable CONFIG_UNIT_TEST on qemu_arm_defconfig.
>>>>>
>>>>> Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt at canonical.com>
>>>>> ---
>>>>>     configs/qemu_arm_defconfig | 1 +
>>>>>     1 file changed, 1 insertion(+)
>>>>>
>>>>> diff --git a/configs/qemu_arm_defconfig b/configs/qemu_arm_defconfig
>>>>> index d042aea49bb..cc4f4540fd5 100644
>>>>> --- a/configs/qemu_arm_defconfig
>>>>> +++ b/configs/qemu_arm_defconfig
>>>>> @@ -67,3 +67,4 @@ CONFIG_TPM2_MMIO=y
>>>>>     CONFIG_USB_EHCI_HCD=y
>>>>>     CONFIG_USB_EHCI_PCI=y
>>>>>     CONFIG_TPM=y
>>>>> +CONFIG_UNIT_TEST=y
>>>>
>>>> Before merging this patch we also have to fix the lib_test_dynamic_uuid
>>>> which fails on 32-bit.
>>>
>>> Is it the test, or is it the UUID changes? I didn't go so far as to
>>> revert the UUID changes when I hit this on Pi 3 32b, just poked Caleb on
>>> irc.
>>>
>>
>> While on 64-bit the expected GUID is generated (both on sandbox and
>> qemu-riscv64_smode_defconfig), the generated GUID on 32-bit does not match
>> the expected value. The generated GUID should be the same irrespective of
>> the system.
>>
>> Once we have fixed 32-bit little endian, we should start testing big endian.
> 
> Right. What I mean is, did this work prior to Caleb's UUID series that
> was merged with:
> commit 35394e1ea77ba0ad971d9115bd965a2403c0e031
> Merge: 9eb0d731d800 7de51622a2cf
> Author: Tom Rini <trini at konsulko.com>
> Date:   Fri Sep 13 08:20:25 2024 -0600
> 
>      Merge tag 'efi-next-20241024' of https://source.denx.de/u-boot/custodians/u-boot-efi into n
> ext
> 
> Or has it always been wrong? To me, Patrick's report implies that it used to work.
> 

The functionality and the failing test were added by the series.

Best regards

Heinrich