On Fri, Jun 19, 2026 at 6:19 PM Tamar Christina <[email protected]> wrote:
>
> The PR reports a failure on -m32 that I can't seem to
> reproduce on my x86_64 machine..
>
> However cvise (I gave the compiler as many flags I
> could think of to prevent this) did give me a
> reproducer that has some UB in it wrt to signed
> shifts on 32-bit systems.
>
> So this fixed the UBs and hopefully that fixes
> the reported failure.
It fixed the failure for me.
> Bootstrapped Regtested on aarch64-none-linux-gnu,
> arm-none-linux-gnueabihf, x86_64-pc-linux-gnu
> -m32, -m64 and no issues.
>
> Pushed to master.
It isn't on master.
> Thanks,
> Tamar
>
> gcc/testsuite/ChangeLog:
>
> PR testsuite/125871
> * gcc.dg/vect/vect-early-break_144-pr125804.c: New test.
>
> -- inline copy of patch --
>
> diff --git a/gcc/testsuite/gcc.dg/vect/vect-early-break_144-pr125804.c
> b/gcc/testsuite/gcc.dg/vect/vect-early-break_144-pr125804.c
> index
> f8974023ee6a3216042b8656a14f1ac42be32a94..d8deb733f70e742db0a1eba24a63e95c335f3623
> 100644
> --- a/gcc/testsuite/gcc.dg/vect/vect-early-break_144-pr125804.c
> +++ b/gcc/testsuite/gcc.dg/vect/vect-early-break_144-pr125804.c
> @@ -21,11 +21,11 @@ int g;
> __attribute__((noipa))
> int h(bitmap i) {
> c j = *i->d;
> - unsigned k, l = j.b < 4;
> + unsigned k = 0, l = j.b < 4;
> for (; l < 2; l++) {
> long long m = j.bits[l];
> for (; k < 64; k++) {
> - long long n = (long long)1LL << k;
> + long long n = (long long)(1ull << k);
> if (m & n)
> goto o;
> }
> @@ -40,8 +40,8 @@ int main() {
>
> check_vect ();
>
> - c a;
> - a.bits[0] = 1ull << 63;
> + c a = { 4, { 0, 0 } };
> + a.bits[0] = -0x7fffffffffffffffLL - 1;
> f.d = &a;
> g = h(&f);
> return g == 0;
--
H.J.
