RISC-V architecture is marked as "unofficial"
Marc Deslauriers
marc.deslauriers at canonical.com
Mon May 5 14:31:22 UTC 2025
On 2025-05-05 10:23, Robie Basak wrote:
> On Sat, May 03, 2025 at 06:03:55AM +0530, Utkarsh Gupta wrote:
>> Hi Robie, others,
>> Yes - please consider marking RISC-V as an official architecture. As
>> far the release process goes, we already treat them as official
>> images.
>>
>> And as Colin said earlier in the thread, please consider doing this
>> for Questing onward and it'd be better to not touch the other stable
>> releases.
>
> Thanks! (and also to Colin and Dimitri). This gives me confidence that
> this change is fine to make.
>
> As an aside, I've been waiting over three days for a riscv64 build in
> Questing[1]. Right now the queue length is apparently 47 hours[2] while
> the other architectures have negligible queues. That does have a big
> impact wrt. proposed migration. Perhaps we should apply some expected
> standard that needs to be made before considering an architecture
> "official"? I don't think we've had anything like that before, but
> perhaps setting some quality expectations would be reasonable and useful
> for the project so as not to have yet another cause for development pace
> to slow?
>
Riscv64 build times are also a big challenge for the security team. We sometimes
have to skip riscv64 when issuing emergency security updates. While we do try
and complete the risc64 builds at a later time as much as possible, this still
results in riscv64 information being absent in our USNs, and our OVAL data.
I'm not sure we will be able to issue timely security updates if riscv64 becomes
an official architecture and we don't change how we build it.
Marc.
More information about the technical-board
mailing list