https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61309
--- Comment #3 from Kai-Uwe Eckhardt <kuehro at gmx dot de> --- Hello John, > FYI, this is still happening. > So if/when DragonFly features libgcc_s from 5.0 in base, these errors will go > away? To be honest I am not quite sure, but I think it is not the presence of libgcc_s.so itself but the presence of __cpu_model within. Mabey a linker script similar to the one in PR 40134 would help. > If so, doesn't that imply the linking is making assumptions it should not > make? Cilk is Linux centric in other places as as well, so I wouldn't wonder. I don't have a Linux system at hand right now to compare the defined symbols. Actually I am giving up on gcc and NetBSD. I have tried to push some changes upstream but gcc has become too complex for my skills and the number of upstream NetBSD developers is too low. Good luck with Dragonfly. Kind regards, Kai-Uwe
