<div>Looks good to me.</div><div><br><div class="gmail_quote gmail_quote_container"><div dir="ltr" class="gmail_attr">Antonio Russo <<a href="mailto:aerusso@aerusso.net">aerusso@aerusso.net</a>>于2025年5月12日 周一20:10写道:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">On 2025-05-12 00:27, Yue Gui wrote:<br>
> <br>
> My solution to this issue:<br>
> The `test_lock_contention` test, which simulates high lock contention, is<br>
> timing out on RISC-V systems likely due to inherent performance differences<br>
> of the architecture compared to more established platforms like x86_64, for<br>
> which the current 5-second timeout is calibrated.<br>
> <br>
> To address this without masking potential regressions on other<br>
> architectures, I propose to conditionally adjust the timeout duration for<br>
> the RISC-V architecture. This is achieved by detecting the architecture<br>
> using `platform.machine()` within the test case and applying a more lenient<br>
> timeout (e.g., 15 seconds) specifically for RISC-V.The debdiff patch is in<br>
> the attachment.I have tested that locally,and it works well.Please let me<br>
> know whether this solution can be accepted.<br>
> <br>
> Gui-Yue<br>
> Best Regards<br>
> <br>
<br>
Hello Gui-Yue,<br>
<br>
Thanks for identifying this and authoring this patch! Would you be willing<br>
to submit this to upstream [1]?<br>
<br>
Best,<br>
Antonio Russo<br>
<br>
[1] <a href="https://github.com/element-hq/synapse" rel="noreferrer" target="_blank">https://github.com/element-hq/synapse</a><br>
</blockquote></div></div>