Re: [xen-unstable test] 187507: regressions - FAIL

[Jan Beulich <jbeulich@suse.com>]

On 06.09.2024 12:07, Andrew Cooper wrote:
> On 06/09/2024 11:01 am, Andrew Cooper wrote:
>> On 06/09/2024 7:08 am, Jan Beulich wrote:
>>> On 06.09.2024 06:41, osstest service owner wrote:
>>>> flight 187507 xen-unstable real [real]
>>>> http://logs.test-lab.xenproject.org/osstest/logs/187507/
>>>>
>>>> Regressions :-(
>>>>
>>>> Tests which did not succeed and are blocking,
>>>> including tests which could not be run:
>>>>  build-i386-xsm                6 xen-build                fail REGR. vs. 187498
>>>>  build-i386                    6 xen-build                fail REGR. vs. 187498
>>> This is a result of "x86emul: introduce a struct cpu_policy * local in
>>> x86_emulate()", and me not noticing the issue because in my tree (for
>>> AMX in particular) I have several more uses of the variable. I'd really
>>> like to avoid undoing the change, but adding __maybe_unused also seems
>>> bogus to me, as does adding a seemingly stray (void)cp somewhere. Good
>>> alternative ideas, anyone?
>> __maybe_unused as at least accurate, and its less fragile than a (void)cp;
> 
> Interestingly, Gitlab's x86_32 build test missed this.
> 
> https://gitlab.com/xen-project/people/andyhhp/xen/-/jobs/7762103169 passed.
> 
> I wonder if there's anything we should have done to get better coverage.

From the build log:

make -C x86_emulator install
make[6]: Entering directory '/builds/xen-project/people/andyhhp/xen/tools/tests/x86_emulator'
make[6]: Nothing to be done for 'install'.
make[6]: Leaving directory '/builds/xen-project/people/andyhhp/xen/tools/tests/x86_emulator'

Iirc one of your colleagues (Alejandro?) had a patch to actually install the
test harness binary. I didn't like this very much, but the above may be a
good reason to have it despite my slight dislike. And I think I had indicated
already that if everyone else thinks this wants installing, so be it.

Jan