Not what I hopped for but what I suspected. Thanks for the quick answer! Best regards, Arne
> Am 06.07.2016 um 06:24 schrieb Tristan Gingold <tging...@free.fr>: > >> On 05/07/16 21:56, Richter, Arne wrote: >> Hi all, >> >> I'm currently trying to build a test-pipeline using Python to describe >> my test cases. Now I want to add GHDL tests into my pipeline by >> compiling them with a C wrapper into a shared library. >> So far so good. With the prebuilt OSX version everything works (except >> the procedure's). But if I now try to fresh build version under linux, >> the same makefile fails. >> >> I repeatedly get the error: >> >> /usr/bin/ld: e~tb_rand.o: relocation R_X86_64_32 against `.bss' can not >> be used when making a shared object; recompile with -fPIC >> >> e~tb_rand.o: error adding symbols: Bad value >> >> Got anyone a clue for me? I'm currently not even sure if automatically >> build libraries might cause the problem. > > The problem is that to create a shared library on x86-64 linux files > must be compiled with -fPIC. > > By default, this is not the case when using ghdl -a or ghdl -e. > You could add the -fPIC switch at that point (I haven't tested that), but > then you will hit the next issue: the program is also linked with libgrt.a, > which is not compiled with -fPIC too. > > So you have to rebuild libgrt.a with the -fPIC option. Maybe ghdl should be > enhanced to also have a libgrt.so. > > It might be simpler to slightly redesign your test-pipeline architecture not > to rely on that mechanism. > > Regards, > Tristan. > > > _______________________________________________ > Ghdl-discuss mailing list > Ghdl-discuss@gna.org > https://mail.gna.org/listinfo/ghdl-discuss _______________________________________________ Ghdl-discuss mailing list Ghdl-discuss@gna.org https://mail.gna.org/listinfo/ghdl-discuss
