I'm building from a crossdev generated package, and musl's /usr/bin/ldd is colliding with the system's version. Reproducible: Always
Duplicate of https://bugs.gentoo.org/show_bug.cgi?id=544128 Could you test my patch?
Patch seems to alleviate the ldd collision, but it installs /lib/ld-musl-armhf.so.1, and the following warning is emitted: /sbin/ldconfig: /lib64/ld-musl-armhf.so.1 is for unknown machine 40. /sbin/ldconfig: /lib64/ld-musl-armhf.so.1 is for unknown machine 40.
musl-1.1.7-r1 should be in the Felix's patch. (In reply to David Flogeras from comment #2) > Patch seems to alleviate the ldd collision, but it installs > /lib/ld-musl-armhf.so.1, and the following warning is emitted: > > /sbin/ldconfig: /lib64/ld-musl-armhf.so.1 is for unknown machine 40. > > /sbin/ldconfig: /lib64/ld-musl-armhf.so.1 is for unknown machine 40. I need more context here. First of all, why is this lib64? Where is that file installed exactly. Also musl doesn't use ldconfig so the message is spurious, but I'm still worried about where /lib64/ld-musl-armhf.so.1 lives.
OK I tried again with the now in-tree 1.1.7-r1 and don't get the previous warnings, but now I get this: QA Notice: Symbolic link /usr/armv6j-hardfloat-linux-musleabi/usr/bin/ldd points to /lib/ld-musl-* which does not exist. I'll attach a listing of all installed files in cross build musl-1.1.7-r1. Note as we previously discussed, I have to do the following when crossbuilding: CFLAGS="-O2 -pipe -march=armv6j -mfpu=vfp -mfloat-abi=hard -fomit-frame-pointer" emerge -1 cross-armv6j-hardfloat-linux-musleabi/musl
Created attachment 399942 [details] Files listing of installed crossdev build musl-1.17-r1
Please try the patch against -r1 of https://bugs.gentoo.org/show_bug.cgi?id=544128 This should fix it and is supposed to be musl-1.1.7-r2.
Sorry my local mirror didn't have -r2 yet. Tried again with -r2, same QA notice.
Created attachment 399952 [details, diff] patch against musl-1.1.7-r2 Sorry for messing it up repeatedly. Could you try this patch against musl-1.1.7-r2?
(In reply to Felix Janda from comment #8) > Created attachment 399952 [details, diff] [details, diff] > patch against musl-1.1.7-r2 > > Sorry for messing it up repeatedly. > > Could you try this patch against musl-1.1.7-r2? Actually I'm pretty sure that's correct so I committed 1.1.7-r3. The point is that both the sym link and the target are in sysroot. For native sysroot is unset and it works out. Test it out guys and let me know.
heh you _just_ beat me to the punch. The proposed patch works as I'd expect.
(In reply to Felix Janda from comment #8) > Created attachment 399952 [details, diff] [details, diff] > patch against musl-1.1.7-r2 > > Sorry for messing it up repeatedly. > > Could you try this patch against musl-1.1.7-r2? No need to apologize, thanks for the patches :)
reopen if there are still issues.