https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122456
--- Comment #8 from Andrew Pinski <pinskia at gcc dot gnu.org> --- (In reply to Andrew Pinski from comment #7) > (In reply to Andrew Pinski from comment #6) > > (In reply to Andrew Macleod from comment #5) > > > It does not reproduce on x86_64 either... > > > > Hmm, that might give me an idea. Maybe someone is misusing char? thinking it > > is signed but on aarch64 it is unsigned by default. > > Maybe one thing to try is to force unsigned char by default for x86_64 or > > force signed char by default for aarch64. > > > > That is change: > > #define DEFAULT_SIGNED_CHAR 0 > > > > in aarch64/aarch64.h to 1. > > > > or define it to 0 in i386/i386.h. > > If I get sometime tomorrow I will do this experiment unless someone beats me > to it. That experiment failed with the same failure as reported here.
