https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82141
--- Comment #32 from Iain Sandoe <iains at gcc dot gnu.org> --- (In reply to simon from comment #28) > (In reply to Iain Sandoe from comment #23) > > > --with-checking=release --disable-libada succeeds for me too. > > Because the bug appears when building libada. right. > I haven’t tried with --checking=release. makes no difference. The stage1 compiler builds the files without an uncaught exception (presumably, that's got sjlj exception handling); and that allows building of the library and thence gnattools. stage2 and stage3 gnat1 both show the same issue. --- so on Darwin15.6 I've built libada (using the stage1 compiler for the 4 files that fail) make -C gcc gnattools proceeds without error ===== make check-ada acats: lots of fails (raised RTSFIND.RE_NOT_AVAILABLE) gnat: many fails because it says the adalibs are not installed correctly (raised TYPES.UNRECOVERABLE_ERROR : targparm.adb:175) ---- debugging unwinder bugs is always "fun". If Eric cannot reproduce this, then I will try Darwin14.5 (the only difference we have at present, I think is Darwin15.6 c.f. Darwin15.5). it doesn't make any difference if the libstdc++ is statically or dynamically linked, in fact I'm looking through the symbol bindings to see if there's anything that look suspicious
