https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88309
Andrew Pinski <pinskia at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |willschm at gcc dot gnu.org Summary|ICE: Floating point |[11/12/13/14 Regression] |exception (in |ICE: Floating point |is_miss_rate_acceptable), |exception (in |target assigning alignent |is_miss_rate_acceptable), |of 4 bits(!) to vector |target assigning alignent | |of 4 bits(!) to vector Known to fail| |9.1.0 Target Milestone|--- |11.5 Known to work| |8.5.0 --- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> --- Found it: /* In GIMPLE the type of the MEM_REF specifies the alignment. The required alignment (power) is 4 bytes regardless of data type. */ tree align_ltype = build_aligned_type (lhs_type, 4); That should be 4*8 instead of just 4. There are 2 build_aligned_type in rs6000-builtins.cc which uses the wrong alignment; thinking it was the alignment argument was bytes rather than bits. Introduced by r9-2375-g3f7a77cd20d07c which means this is a regression.