Re: obs-studio FTBFS on loong64
Hi,
On Mon, Nov 25, 2024 at 12:46 AM John Paul Adrian Glaubitz
<glaubitz@physik.fu-berlin.de> wrote:
>
> Hello,
>
> On Sun, 2024-11-24 at 20:37 +0800, Bo YU wrote:
> > Now luajit[0] can be built on loong64, so the obs-studio[1] have a
> > chance to try build on loong64, but it failed now:
> >
> > ```
> > ...
> > /usr/include/simde/x86/sse2.h: At top level:
> > /usr/include/simde/x86/sse2.h:260:24: error: conflicting types for
> > ‘__m128i’; have ‘simde__m128i’
> > 260 | typedef simde__m128i __m128i;
> > | ^~~~~~~
> > In file included from /usr/include/simde/simde-features.h:766,
> > from /usr/include/simde/simde-common.h:45,
> > from /usr/include/simde/x86/mmx.h:30,
> > from /usr/include/simde/x86/sse.h:33:
> >
> > ```
> > see https://buildd.debian.org/status/fetch.php?pkg=obs-studio&arch=loong64&ver=30.2.3%2Bdfsg-2&stamp=1732241930&raw=0
> >
> > The obs-studio is a key package for obs-* packages so I am reporting
> > this to here to reminder loong64 porter.
> >
> > Could you have a look at this?
>
> This is very obviously not a loong64-specific issue but a generic bug as the
> build process tries to unconditionally build x86 code on any target.
>
hmm, right. Maybe I am wrong, but I believe there are some asm
implementation on loongarch needed on simde:
https://github.com/simd-everywhere/simde/blob/master/simde/x86/sse2.h
> Please report this issue upstream!
>
I’m happy to report this upstream if needed, but I believe the
porters' expertise in handling architecture-specific issues could be
valuable here.
BR,
Bo
> Thanks,
> Adrian
>
> --
> .''`. John Paul Adrian Glaubitz
> : :' : Debian Developer
> `. `' Physicist
> `- GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Reply to: