https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67529
--- Comment #4 from Eric <ericns1 at spirilis dot net> --- Okay, I think I've figured out what I'm looking at here. The inheritance has to do with the use of virtual functions - so the JSR vs BSR argument in here is a red herring, overridden virtual functions use JSR because it's doing a lookup inside the vtable. GCC 5.2.0 is providing the vtable with a general gcc/newlib-derived ldscript I have. The GNURX example which failed on my RX210, otoh, does not appear to have the vtable when I disassemble using rx-elf-objdump -C -DS and search for 'vtable'. Probably an incomplete ldscript there. Need to figure out what sections vtable et al fall under... then I will go complain to the KPIT folks. Sorry for the wild goose chase :-)
