https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83105
--- Comment #3 from nsz at gcc dot gnu.org --- (In reply to Jakub Jelinek from comment #1) > Any reason why this should be a supportable configuration? I want hard > float, but don't care what CPU/FPU I target... i think there should be a baseline target that provides the widest possible compatibility when one wants to distribute a binary. (in which case the user should not try to hardcode a cpu/fpu name in their gcc build script) i think it should work for --with-float=hard and arm*-gnueabihf target, but if not then this configuration should be rejected earlier. (to me it makes sense to raise the baseline of arm*-gnueabihf to armv5te but leave arm*-gnueabi at armv5t) i don't use many different toolchain build scripts, but from that small set 2 got broken by this change, one of them is the glibc test script which required a workaround like https://sourceware.org/ml/libc-alpha/2017-12/msg00611.html i think requiring such change is suboptimal.
