[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PULL 47/47] tests/tcg/multiarch: add vma-pthread.c
From: |
Alex Bennée |
Subject: |
Re: [PULL 47/47] tests/tcg/multiarch: add vma-pthread.c |
Date: |
Thu, 05 Jan 2023 09:10:17 +0000 |
User-agent: |
mu4e 1.9.11; emacs 29.0.60 |
Richard Henderson <richard.henderson@linaro.org> writes:
> On 1/4/23 16:37, Richard Henderson wrote:
>> On 1/4/23 16:26, Richard Henderson wrote:
>>> From the failures I see on the gitlab merge job, I think I need to
>>> resubmit with this new test adjusted to loop less,
>>>
>>>> + for (i = 0; i < 50000; i++) {
>>>
>>> here.
>>>
>>> The failing jobs are --enable-debug, and take about 115 seconds to
>>> run manually on our aarch64 test host, exceeding the 90 second
>>> timeout.
>>>
>>> I'll cut this down to 10000 loops and double-check times before
>>> resubmitting.
>> Hmm. Even this only reduced the runtime to 98 seconds.
>
> Bah. The testcase didn't rebuild as expected. Building from clean,
> the 10k loop completes in 20 seconds with optimization disabled.
>
> I do wonder what the build time / test time trade-off is here, and
> whether we should be doing much -O0 testing in CI...
I think the main argument for --enable-debug is less about the -O0 and
more about the extra asserts. Can we have -O3 with --enable-debug-tcg?
>
>
> r~
--
Alex Bennée
Virtualisation Tech Lead @ Linaro