linux-kernel - Re: Error during --arch x86

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <65050389-2b80-4355-be2e-604c1c5b6b9a@linuxfoundation.org>
Date: Thu, 13 Mar 2025 14:19:06 -0600
From: Shuah Khan <skhan@...uxfoundation.org>
To: David Gow <davidgow@...gle.com>
Cc: Brendan Higgins <brendan.higgins@...ux.dev>, Rae Moar <rmoar@...gle.com>,
 linux-kselftest@...r.kernel.org, linux-kernel@...r.kernel.org,
 Shuah Khan <skhan@...uxfoundation.org>
Subject: Re: Error during --arch x86_64 kunit test run

On 3/12/25 20:21, David Gow wrote:
> On Thu, 13 Mar 2025 at 07:51, Shuah Khan <skhan@...uxfoundation.org> wrote:
>>
>> On 3/12/25 16:52, David Gow wrote:
>>> Hi Shuah,
>>>
>>> On Thu, 13 Mar 2025 at 05:14, Shuah Khan <skhan@...uxfoundation.org> wrote:
>>>>
>>>> David, Brendan, Rae,
>>>>
>>>> I am seeing the following error when I run
>>>>
>>>> ./tools/testing/kunit/kunit.py run --arch x86_64
>>>>
>>>> ERROR:root:ld:arch/x86/realmode/rm/realmode.lds:236: undefined symbol `sev_es_trampoline_start' referenced in expression
>>>>
>>>> I isolated it to dependency on CONFIG_AMD_MEM_ENCRYPT
>>>>
>>>
>>> That's interesting. I recall seeing this issue briefly about a year
>>> ago on an internal branch, and we worked around it there by enabling
>>> CONFIG_AMD_MEM_ENCRYPT (which worked).
>>>
>>> But I've been totally unable to reproduce it this morning: are you
>>> seeing this on any particular branch / config / environment?
>>
>> linux_next - I did allmodconfig build prior to running the tests.
>> That means I had to run make ARCH=x86_64 mrproper before running
>>
>> ./tools/testing/kunit/kunit.py run --arch x86_64
>>
>> I tried something different checking out a fresh
>> linux_next repo and running ./tools/testing/kunit/kunit.py run --arch x86_64
>>
>> No errors on
>>
>> ./tools/testing/kunit/kunit.py run --arch x86_64
>>
>> I will try this again and let you know. Can you try this as well.
>>
>> - Clean linux_next and run tests
>> - Run tests after buidling allmodconfig and mrproper
>>
> 
> Thanks, Shuah.
> 
> Alas, I've tried both on a clean linux-next (next-20250312), and after
> doing an allmodconfig build (and make mrproper), and am still unable
> to reproduce this here.
> 
> I also dug up the old 5.10-based version we saw something similar on,
> and wasn't able to reproduce it there, either. (That had been built
> with clang, but using clang on upstream doesn't seem to reproduce it
> either.)
> 
> I'll continue to play around with different setups, but none of the
> obvious things seem to work. Worst-case, I don't think it'd be a
> _problem_ to make EFI + AMD_MEM_ENCRYPT the default for the
> x86_64/qemu KUnit configuration.
> 

I tried again starting with a clean linux_next repo next-20250313

- Ran tests first
- Built allmodconfig
- Ran mrproper
- Ran tests

Saw no problems. Let's watch and see if we can narrow the problem.
For now I was able to test the pr I will be sending soon.

thanks,
-- Shuah