Created attachment 558422 [details] build.log.xz This is a regression from 18.1.9 (same USE flags), which has since been removed entirely.
mesa-18.2.7 builds fine on all my ppc/ppc64 machines. I guess this only happens with LE-profiles: [13] default/linux/powerpc/ppc64/17.0/64bit-userland/little-endian (exp) [14] default/linux/powerpc/ppc64/17.0/64bit-userland/little-endian/systemd (exp)
The problem is that the Meson build system never got the hack I added to autotools in https://cgit.freedesktop.org/mesa/mesa/commit/?id=7d1195c1e4d071fe796bf5f210c468ea1cc86225 Cc'ing Dylan.
(In reply to ernsteiswuerfel from comment #1) > mesa-18.2.7 builds fine on all my ppc/ppc64 machines. I guess this only > happens with LE-profiles: > > [13] default/linux/powerpc/ppc64/17.0/64bit-userland/little-endian (exp) > [14] default/linux/powerpc/ppc64/17.0/64bit-userland/little-endian/systemd > (exp) No, I don't think that's true. This bug only occurs when you have opencl enabled, which I suspect you do not.
I'm pretty sure this is related to not using gnu++11. I don't know how to inject meson arguments into a build, but you could try adding '-Dcpp_std=gnu++11` to verify the issue. There is an MR opened to add the gnu++11 overrides, I'll mention Matt there. https://gitlab.freedesktop.org/mesa/mesa/merge_requests/29
Created attachment 558626 [details] patched-build.log.xz Build fails also with https://gitlab.freedesktop.org/mesa/mesa/merge_requests/29
That was my fault. it's helpful to compare 'ppc' against the host machine cpu family, not the operating system family. Can you try again?
>>> media-libs/mesa-18.2.7 merged.
Fixed in 18.3.2.