Re: [regression] suspend stress test stalls within 30 minutes

[Kalle Valo]

Kalle Valo <kvalo@xxxxxxxxxx> writes:

> Borislav Petkov <bp@xxxxxxxxx> writes:
>
>> On Sat, May 11, 2024 at 08:48:47PM +0200, Borislav Petkov wrote:
>>> On Sat, May 11, 2024 at 09:22:43PM +0300, Kalle Valo wrote:
>>> > Here's the diff between broken and working .config:
>>> > 
>>> > $ diffconfig broken.config works.config 
>>> > -CALL_PADDING y
>>> > -CALL_THUNKS y
>>> > -CALL_THUNKS_DEBUG n
>>> > -HAVE_CALL_THUNKS y
>>> > -MITIGATION_CALL_DEPTH_TRACKING y
>>> > -MITIGATION_GDS_FORCE y
>>> > -MITIGATION_IBPB_ENTRY y
>>> > -MITIGATION_IBRS_ENTRY y
>>> > -MITIGATION_PAGE_TABLE_ISOLATION y
>>> > -MITIGATION_RETHUNK y
>>> > -MITIGATION_RETPOLINE y
>>> > -MITIGATION_RFDS y
>>> > -MITIGATION_SLS y
>>> > -MITIGATION_SPECTRE_BHI y
>>
>> ... and if it started with -rc4, I'd try this one first.
>
> I'm not sure if this bug started with -rc4, let's say somewhere between
> -rc3 and -rc6. I'll start with disabling MITIGATION_SPECTRE_BHI and then
> I'll disable them one by one.

It is starting to look like that MITIGATION_IBRS_ENTRY is causing my
problems. If I enable MITIGATION_IBRS_ENTRY I see the suspend stall
relatively quickly. And then I disable the config I have not been able
to reproduce the problem so far. But as this issue happens randomly I
want to do some more tests to be sure.

-- 
https://patchwork.kernel.org/project/linux-wireless/list/

https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches