Re: [PULL 0/2] Seabios 1.17.0 20250611 patches

[Stefan Hajnoczi <stefanha@gmail.com>]

On Thu, Jun 12, 2025 at 12:36 AM Thomas Huth <thuth@redhat.com> wrote:
>
> On 11/06/2025 21.09, Stefan Hajnoczi wrote:
> > On Wed, Jun 11, 2025 at 2:36 PM Thomas Huth <thuth@redhat.com> wrote:
> >>
> >> On 11/06/2025 09.50, Gerd Hoffmann wrote:
> >>> The following changes since commit bc98ffdc7577e55ab8373c579c28fe24d600c40f:
> >>>
> >>>     Merge tag 'pull-10.1-maintainer-may-2025-070625-1' of https://gitlab.com/stsquad/qemu into staging (2025-06-07 15:08:55 -0400)
> >>>
> >>> are available in the Git repository at:
> >>>
> >>>     https://gitlab.com/kraxel/qemu.git tags/seabios-1.17.0-20250611-pull-request
> >>>
> >>> for you to fetch changes up to cba36cf3881e907553ba2de38abd5edf7f952de1:
> >>>
> >>>     seabios: update binaries to 1.17.0 (2025-06-11 09:45:00 +0200)
> >>>
> >>> ----------------------------------------------------------------
> >>> seabios: update to 1.17.0 release
> >>>
> >>> ----------------------------------------------------------------
> >>>
> >>> Gerd Hoffmann (2):
> >>>     seabios: update submodule to 1.17.0
> >>>     seabios: update binaries to 1.17.0
> >>
> >>    Hi Gerd, hi Stefan,
> >>
> >> I'm now getting this when doing a git pull:
> >>
> >> Fetching submodule roms/seabios
> >> fatal: remote error: upload-pack: not our ref
> >> b52ca86e094d19b58e2304417787e96b940e39c6
> >> Errors during submodule fetch:
> >>          roms/seabios
> >
> > GitLab CI didn't detect this, probably because the tests don't build
> > SeaBIOS from source and use the binaries instead. Given infinite CI
> > resources we should rebuild all ROMs from source to catch problems
> > like this one.
> >
> > I wanted to mention this in case anyone wants to tighten up the CI to
> > catch these issues for SeaBIOS and other ROMs.
> >
> >> Looking at
> >> https://gitlab.com/qemu-project/seabios/-/commits/master?ref_type=HEADS
> >> there is a problem with the mirroring:
> >>
> >> "This project is mirrored from https://review.coreboot.org/seabios. Pull
> >> mirroring failed 4 months ago.
> >> Repository mirroring has been paused due to too many failed attempts. It can
> >> be resumed by a project maintainer or owner.
> >> Last successful update 4 months ago. This branch has diverged from upstream. "
> >>
> >> Could you please fix this?
> >
> > `git clone https://review.coreboot.org/seabios` works on my machine,
> > but I manually clicked the "update" button and GitLab failed again.
> >
> > In the GitLab UI there is a button to add a new mirror repo. I thought
> > maybe we can delete the old repo and add a new one, but the push/pull
> > direction dropdown list is disabled. Maybe that's because only 1 repo
> > can be a mirror source for pull, I'm worried that deleting the
> > existing failed repo will leave us with no way to add a new repo that
> > supports pull.
> >
> > Does anyone know how to go about fixing this failed mirror?
>
> Maybe do a manual push --force to our mirror to sync them again? ... but it
> would be good to know how this could have happened at all - maybe seabios
> force-pushed their master branch at one point in time? Or did we commit
> something to the mirror that was not in the upstream repository?
>
> Anyway, I just noticed that it seems to be working now again - how did you
> fix it?

I didn't fix it. GitLab's mirror operation is still failing. Your
suggestion of force pushing should fix it, but I want to understand
why a commit was lost first.

The issue is that QEMU's mirror and upstream have diverged. The
following commit is only in QEMU's mirror repo:

commit 44693a974cd90917f81a7d0310df4b592edd7e09
Author: Christopher Lentocha <christopherericlentocha@gmail.com>
Date:   Tue Jan 21 11:59:14 2025 -0500

    Fix AHCI Disk Detection when using EDK2 CSM

It is not possible to fast-forward to upstream's master due to this difference.

Gerd: Do you know what happened to this commit upstream? According to
the mailing list archives you merged it, but I don't see it on the
master branch:
https://patchew.org/Seabios/clqcnplppk34e7eyiquorweq5dvmbsgm2qrk7ozhzybvxviboa@4i2yfqplktar/

Stefan