Portage reports QA notices after building wine-1.5.23. [ebuild R ] app-emulation/wine-1.5.23 USE="X alsa cups gecko jpeg lcms ldap mono mp3 ncurses nls openal opengl osmesa oss perl png prelink pulseaudio samba ssl threads truetype udisks win32 win64 xcomposite xinerama xml -capi -custom-cflags -fontconfig -gnutls -gphoto2 -gsm (-gstreamer) -odbc -opencl -scanner (-selinux) {-test} -v4l" 0 kB Reproducible: Always
Created attachment 337808 [details] QA Notice
Created attachment 337810 [details] emerge --info
Source of the change is following commit: http://source.winehq.org/git/wine.git/commitdiff/c14bdaf1ddb7d0e5587f63f1216b61c9ecb7a8c3 AFAICT, there's no opt-out. The interesting thing about this is that those entries seem to be empty.
Thanks for reporting. Fixed by patching winebuild to respect the build-time value of $CCAS instead of just automagically looking for clang. Patch submitted upstream at http://www.winehq.org/pipermail/wine-patches/2013-February/122040.html +*wine-1.5.23-r1 (04 Feb 2013) + + 04 Feb 2013; Alexandre Rostovtsev <tetromino@gentoo.org> + +wine-1.5.23-r1.ebuild, +files/wine-1.5.23-winebuild-AS.patch, + wine-9999.ebuild: + Use the correct system 'as' command as the assembler; fixes text relocation in + wine binaries on systems with sys-devel/clang installed (bug #455308, thanks + to Dennis Schridde and Rafał Mużyło).
TBH, I'd be more interested in hearing *why* such effect happens.
(In reply to comment #5) > TBH, I'd be more interested in hearing *why* such effect happens. @clang maintainers, any idea why clang-3.2 produces textrels in wine binaries when used as a linker?
(In reply to comment #6) > (In reply to comment #5) > > TBH, I'd be more interested in hearing *why* such effect happens. > > @clang maintainers, any idea why clang-3.2 produces textrels in wine > binaries when used as a linker? More exactly, why does it produce *bogus* entries - scanelf says "has TEXTREL markings but doesnt appear to have any real TEXTREL's !?" and indeed readelf says the size of the entry is 0.