https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118097
--- Comment #12 from Xi Ruoyao <xry111 at gcc dot gnu.org> --- (In reply to David Binderman from comment #10) > (In reply to Sam James from comment #9) > > Ah, sorry, I see it on the original with -O2. I don't see it on the reduced > > one (though it was invalid anyway). OK. > > It looks as if my reduction was invalid. My cvise script was > as follows: > > /usr/bin/gcc -w bug1071.c && (ulimit -t 10 ; ./a.out 1 | grep "g_90 : > F239C101") \ > && /home/dcb40b/gcc/results/bin/gcc -w -O2 -Werror=uninitialized \ > -Werror=aggressive-loop-optimizations -Werror=sequence-point bug1071.c \ > && (ulimit -t 10; ./a.out 1 | grep "g_90 : B01CC67C") > > I am a little bit surprised the Werror-uninitialized didn't catch > the local variable read of __trans_tmp_5. AFAIK -w suppresses -Werror=uninitialized. Without -w and with -Werror=uninitialized at least GCC 14.2.0 rejects the code (if trunk does not reject it, we have another regression :)
