https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92723
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|RESOLVED |REOPENED
Resolution|FIXED |---
Target Milestone|10.0 |8.4
Summary|[10 Regression] ICE in |[8/9 Regression] ICE in
|expand_shift_1, at |expand_shift_1, at
|expmed.c:2635 |expmed.c:2635
--- Comment #9 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Actually, reopening for backports to release branches. On the testcase that
got committed with just the rotate in there, this used to compile fine at -O3
on aarch64 with r200000 and ICEs already in r201445, so it is a regression,
just don't know which exact change caused it, but the rotate pattern changes
are before that, so most likely some aarch64 backend changes triggered it.
