As I used the #gentoo-soc-musl channel as a mini-blog Blueknight suggested to me that I’d do small daily blog posts instead.
Today I’ve worked on a lot of different and seemingly unrelated stuff. To start with I fixed plocate to use fprintf + exit instead of error. Upstream wanted a proper patch. This was not as trivial as just replacing since the code used the error_message_count variable. Still pretty straight forward though.<\/p>\r\n
The next thing I did was to take a look at cross compiling for aarch64-gentoo-linux-musl on x86-64, but I failed to compile the stage1 gcc, both with crossdev and https:\/\/github.com\/richfelker\/musl-cross-make.
Then I tried fchroot instead but glib needed patching to build correctly with [static-libs]. It used ngettext() without any reason so I removed it, will publish it once I know it’s okay. The fchroot::guru ebuild had dependencies on qemu_softmmu_* which is just wrong, so I changed it to _user_ and will push to ::guru soon. As neither aarch64 or arm is mandatory I think it makes sense to make USE-flags for them too.<\/p>\r\n
I successfully got into a Gentoo aarch64 musl root with fchroot and started emerging some router-specific software. I noticed that miniupnpd didn’t have any keywords for arm64 so I added ~arm64.
I also took a look at qtwebengine and worked on some things, like the bundled Chromium (oh god no) using sys\/cdefs.h which is an internal glibc header. Qtwebengine seems like a monster and will probably take a while to fix, especially upstreaming the changes.
After messing with qtwebengine I got tired of build-fix-cry-build-cycle so I continued testing some stuff on arm64 musl.
<\/p>\r\n","protected":false},"excerpt":{"rendered":"
As I used the #gentoo-soc-musl channel as a mini-blog Blueknight suggested to me that I’d do small daily blog posts instead. Today I’ve worked on a lot of different and seemingly unrelated stuff. To start with I fixed plocate to … Continue reading