https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123052
--- Comment #2 from Iain Sandoe <iains at gcc dot gnu.org> --- Yes... it is unfortunate (and also the reason that we cannot switch to using .cfi_xxx instructions for the unwind data). There are multiple interacting pieces - including what GCC emits for DWARF-2 c.f. what clang emits Some versions of clang do not emit the correct "DWARF-fallback" flag for cases that they cannot encode although I think that is now fixed. Whether the dyld linker does better remains to be seen- but that's academic for folks on linux since it is closed source. I guess ld64 would be fixable, someone just needs to do it. I have multiple WIP branches that attempt to address these incompatibilities - but progress is slow - and this is not a regression - it has never worked.
