https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97192
Richard Biener <rguenth at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Resolution|--- |DUPLICATE Status|UNCONFIRMED |RESOLVED --- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> --- Looks like the same case as for PR97085 for aarch64: - _19 = VEC_COND_EXPR <_18, { -1, -1, -1, -1, -1, -1, -1, -1 }, { 0, 0, 0, 0, 0, 0, 0, 0 }>; - _20 = {a.2_2, a.2_2, a.2_2, a.2_2, a.2_2, a.2_2, a.2_2, a.2_2}; - _21 = _19 == _20; + _21 = VEC_COND_EXPR <_18, { 0, 0, 0, 0, 0, 0, 0, 0 }, { -1, -1, -1, -1, -1, -1, -1, -1 }>; _22 = VEC_COND_EXPR <_21, { -1, -1, -1, -1, -1, -1, -1, -1 }, { 0, 0, 0, 0, 0, 0, 0, 0 }>; so we compute the VEC_COND_EXPR mask using a VEC_COND_EXPR. And it's indeed fixed by the fix for said PR. I will add the testcase. *** This bug has been marked as a duplicate of bug 97085 ***