Re: MiniOS build regressions

[Julien Grall <julien@xxxxxxx>]

Hi,

On 06/12/2023 21:58, Andrew Cooper wrote:
> Following the chaos this morning and a scattering of fixes, I finally
> got back to a working Gitlab with:
>
> https://gitlab.com/xen-project/people/andyhhp/xen/-/pipelines/1097843454
>
> This contains a full revert of the DMI patch, and removal of the Xilinx
> hardware runner, both handled in other threads but noted here for anyone
> looking at the test results.
>
> It also contains 3 MiniOS fixes, mixed between repos:
>
> "tools/libs/evtchn: drop assert()s in stubdom" in xen [1]
> "Mini-OS: don't use objcopy --dump-section" in minios [2]
> "Mini-OS: export main_thread" [3]
>
> and the result with all of these passes.
>
> However, it is my understanding that we don't actually want to take
> patch 3, instead preferring patch 1 as an alternative.
Looking at patch 1, I feel that 3 is better temporarily until we have 
proper locking. With 1, we are just setting ourself for another failure.
> But there is a concern which has been raised over patch 1.
>
>
> So what are we going to do?
>
> If there isn't an answer promptly, I will revert the most recent bump to
> MINIOS_UPSTREAM_REVISION to unbreak Gitlab CI testing for everyone else.
I would say we should unblock gitlab CI ASAP. So whichever is the easiest.

Cheers,

--
Julien Grall