On Tue, Oct 27, 2020 at 4:24 AM Christophe Lyon via Gcc-patches <gcc-patches@gcc.gnu.org> wrote: > > Hi, > > On Mon, 26 Oct 2020 at 22:51, Andrew MacLeod via Gcc-patches > <gcc-patches@gcc.gnu.org> wrote: > > > > In the core of gori_compute::logical_combine we are suppose to combine > > the calculated true and false ranges on each side of the operation. > > > > when encountering > > > > [0,0] = c_3 | c_4 > > > > we know we only need to consider the FALSE values of the range carried > > by c_3 and c_4, but it can be EITHER of those ranges, so we need to > > union them together to get the correct result. > > > > The code was performing an intersection instead, and in this particualr > > case, we knew the range carried thru c_3 was alwasy [0,0] and it was > > always varying through c_4.... instead of returning varying, we were > > returning [0,0] which then caused some folding which was incorrect. > > > > Fixed by correctly calling union... > > > > Bootstrapped on x86_64-pc-linux-gnu, no regressions, and pushed. > > > > I think you need to update the testcase and declare > long long g = 4073709551615 > instead of just long, as it causes a warning on 32-bit targets: > /gcc/testsuite/gcc.dg/pr97567.c:7:12: warning: overflow in conversion > from 'long long int' to 'long int' changes value from '4073709551615' > to '2080555007' [-Woverflow] >
The testcase is an infinite loop on ILP32 targets. -- H.J.
