Re: Bug#1059698: Please do a maintainer upload to enable riscv64 cross-building
On 30.12.23 12:32, Helmut Grohne wrote:
> Control: tags -1 + wontfix
>
> Hi Jan,
>
> On Sat, Dec 30, 2023 at 12:16:29PM +0100, Jan Kiszka wrote:
>> It looks like this is caused by libkeyutils1:amd64 and
>> libkeyutils1:riscv64 having different versions and, thus, refuse to be
>> installed at the same time. I've resolved it locally by rebuilding
>> keyutils for riscv64 using the same version as for the other archs.
>
> Yes. keyutils was binNMUed on riscv64 and hence is not coinstallable
> with anything else.
>
>> Would be great to resolve that by performing a maintainer upload for
>> this package to sid for all architectures (not sure if
>> libkeyutils1:riscv64 could be "downgraded" from 1.6.3-2+b1 to 1.6.3-2, I
>> suspect not).
>
> The way to resolve this is to binNMU everything else to match the +b1 of
> riscv64. We're aware riscv64 generally is not version-synced with other
> architectures. The release team has decided to not solve this problem
> yet as it would cause very many binNMUs. When riscv64 enters testing, it
> will be synced, but not now.
>
> I also note that having crossbuild-essential-riscv64 is another missing
> prerequisite for cross building. See #1022540
Yes, we are carrying a workaround for that in Isar for almost 4 years
now [1].
>
> Thus I request more patience here. Sorry for the inconvenience.
>
Fixing this issue here already would allow to use sid or a snapshot of
it until riscv64 finally enters that state. We don't need all packages
in perfect sync, just enough to cross-build to usual suspects for
embedded: kernel, firmware, maybe a few board-specific packages.
In order to move forward a bit quicker, I will now need to propose an
Isar workaround to rebuild keyutils for riscv64 before creating the
cross-build sbuild-schroot. Fortunately not too complex, just not nice.
Our VisionFive2 image already builds again in about 30 min. thanks to that.
Jan
[1]
https://github.com/ilbers/isar/commit/94ad0ea915d02045a787762f66f70c7e52df899c
Reply to: