Re: [PULL 00/30] Rust, wasm changes for 2025-05-06

[Paolo Bonzini <pbonzini@redhat.com>]

On 5/8/25 19:05, Thomas Huth wrote:
> On 08/05/2025 18.08, Paolo Bonzini wrote:
>> On 5/8/25 16:26, Stefan Hajnoczi wrote:
>>> On Tue, May 6, 2025 at 11:30 AM Paolo Bonzini <pbonzini@redhat.com> 
>>> wrote:
>>>>
>>>> The following changes since commit 
>>>> a9e0c9c0f14e19d23443ac24c8080b4708d2eab8:
>>>>
>>>>    Merge tag 'pull-9p-20250505' of https://github.com/cschoenebeck/ 
>>>> qemu into staging (2025-05-05 11:26:59 -0400)
>>>>
>>>> are available in the Git repository at:
>>>>
>>>>    https://gitlab.com/bonzini/qemu.git tags/for-upstream
>>>>
>>>> for you to fetch changes up to 
>>>> e6b9b79c3076777b791f72ebdbc9d37ad8005fe9:
>>>>
>>>>    gitlab: Enable CI for wasm build (2025-05-06 16:02:04 +0200)
>>>>
>>>> ----------------------------------------------------------------
>>>> * ci: enable RISC-V cross jobs
>>>> * rust: bump minimum supported version to 1.77
>>>> * rust: enable uninlined_format_args lint
>>>> * initial Emscripten support
>>>> * small fixes
>>>
>>> I'm not sure why, but the following CI failure seems to be caused by
>>> this pull request:
>>> https://gitlab.com/qemu-project/qemu/-/jobs/9974291215#L4684
>>>
>>> Please take a look, thanks!
>>
>> It is a transient failure; it reproduces 30% of the time with
>>
>>     meson test --repeat 100 func-hppa-hppa_seabios
> 
> Please not that you should use -j1 when using --repeat ... otherwise the 
> same test will run in parallel multiple times and the instances might 
> destroy the output files of other tests while running.

Yes, I noticed that later but it also fails with

for i in $(seq 1 100); do
   meson test --no-rebuild func-hppa-hppa_seabios
done

>> even without the pull request (commit 
>> a9e0c9c0f14e19d23443ac24c8080b4708d2eab8).
>>
>> Before finding this I had already sent the first half (which should be 
>> safe since it's all Rust code that isn't compiled on that runner---and 
>> for hppa targets in general), but if you still have the merge commit 
>> perhaps you can push it?
> 
> Looking at the log and tests/functional/test_hppa_seabios.py, it seems 
> like this test just fires up the hppa BIOS, checks for some strings and 
> that's it. However, it seems like qemu-system-hppa quits completely if 
> the BIOS cannot boot from any device - so it's likely a race: QEMU quits 
> too fast, while the test still tries to talk to it to shut down QEMU 
> gracefully?
> 
> If I'm right this could likely be fixed by simply adding "-no-shutdown" 
> to the QEMU options here...?
Good idea!

Paolo