Re: xen | Failed pipeline for staging | 08aacc39

[Stefano Stabellini <sstabellini@xxxxxxxxxx>]

On Thu, 8 Aug 2024, Alejandro Vallejo wrote:
> On Thu Aug 8, 2024 at 2:02 PM BST, Jan Beulich wrote:
> > On 08.08.2024 14:43, GitLab wrote:
> > > 
> > > 
> > > Pipeline #1405649318 has failed!
> > > 
> > > Project: xen ( https://gitlab.com/xen-project/hardware/xen )
> > > Branch: staging ( 
> > > https://gitlab.com/xen-project/hardware/xen/-/commits/staging )
> > > 
> > > Commit: 08aacc39 ( 
> > > https://gitlab.com/xen-project/hardware/xen/-/commit/08aacc392d86d4c7dbebdb5e664060ae2af72057
> > >  )
> > > Commit Message: x86/emul: Fix misaligned IO breakpoint behaviou...
> > > Commit Author: Matthew Barnes
> > > Committed by: Jan Beulich ( https://gitlab.com/jbeulich )
> > > 
> > > 
> > > Pipeline #1405649318 ( 
> > > https://gitlab.com/xen-project/hardware/xen/-/pipelines/1405649318 ) 
> > > triggered by Jan Beulich ( https://gitlab.com/jbeulich )
> > > had 4 failed jobs.
> > > 
> > > Job #7535428747 ( 
> > > https://gitlab.com/xen-project/hardware/xen/-/jobs/7535428747/raw )
> > > 
> > > Stage: build
> > > Name: qemu-system-aarch64-6.0.0-arm64-export
> > > Job #7535428873 ( 
> > > https://gitlab.com/xen-project/hardware/xen/-/jobs/7535428873/raw )
> > > 
> > > Stage: build
> > > Name: alpine-3.18-gcc-debug-arm64-static-shared-mem
> > > Job #7535428869 ( 
> > > https://gitlab.com/xen-project/hardware/xen/-/jobs/7535428869/raw )
> > > 
> > > Stage: build
> > > Name: alpine-3.18-gcc-debug-arm64-staticmem
> > > Job #7535429434 ( 
> > > https://gitlab.com/xen-project/hardware/xen/-/jobs/7535429434/raw )
> > > 
> > > Stage: test
> > > Name: qemu-smoke-dom0less-arm32-gcc
> >
> > All Arm failures when the three commits under test only touch x86 code.
> > How can that be? And Stefano, note how this would needlessly have blocked
> > a merge request, if we were already using that model you're proposing to
> > switch to.
> >
> > Jan
> 
> I'd argue it the other way around. It would (may?) have prevented reaching 
> that
> situation in the first place.

Thanks, Alejandro. That's exactly the right mindset and precisely why we
should transition to pre-commit testing. Recently, we encountered a few
broken tests caused by easily avoidable errors from untested commits.

As it happens, due to a change in the ARM GitLab runner on August 10, I
added two new temporary ARM GitLab runners. However, one of them appears
to be experiencing intermittent network issues, I might have to disable
it.