https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107774
Bug ID: 107774
Summary: rtl failed to simplify subreg:vec_select to just
vec_select.
Product: gcc
Version: 13.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: rtl-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: crazylht at gmail dot com
Target Milestone: ---
When i'm looking at
https://gcc.gnu.org/pipermail/gcc-patches/2022-November/606373.html, I notice
there's some misoptimization in rtl.
It failed to simplift (subreg:SF (vec_select:V2SF (reg:V4SF) (parallel
[(const_int 2) (const_int 3)]) 0) to just (vec_select:SF (reg:V4SF) (parallel
[(const_int 2)])
The optimization is exposed after adjustment of ix86_can_change_mode_class, w/o
that, i don't have a case to reproduce this issue. But I can think backends
canonicalizing vec_select to subreg may have such issue.
