Hi Andrew, Mingli,

On 16 Jun 2026, at 21:50, Andrew Geissler via lists.openembedded.org <geissonator=gmail.com@lists.openembedded.org> wrote:

The move to qemu 11 and the enforcement of 64 bit machines broke ppc64le
machines. Add ppc64le wherever there is a ppc64 machine.
Interestingly this patch from Andrew conflicts with a patch from Mingli Yu which changes ppc64 to powerpc64[1].

Without triple-checking I’m more liable to believe your patch as I expect that you’re actually running PPC64LE servers (and also touches more places where COMPATIBLE_HOST is set), but I’m also wondering what Mingli’s patch is doing because it must have been sent for a reason.

Can you both compare each-others patches so we can understand what happened?

Cheers,
Ross

[1] "qemu: Correct the ppc64 setting in COMPATIBLE_HOST”, https://lists.openembedded.org/g/openembedded-core/message/238884

Yep I tested my patch on our POWER9 ppc64le based machine and verified it allowed me to build the latest OpenBMC distro. I submitted this patch because our CI process that utilizes a ppc64le stopped working when we picked up the latest oe-core in OpenBMC.

I poked around a bit and when I look at a POWER server that not running in little endian mode, I don't actually see ppc64 or powerpc64 from Mingli patch. As far as I can tell it's just powerpc (uname -p). So I don't have a lot vested in the non-ppc64le portion. If Mingli can confirm it tested fine for their config then I'm good if we just merge both patches, otherwise please merge mine at least :)