[Bug middle-end/86122] [8/9 Regression] ICE in useless_type_conversion_p, at gimple-expr.c:87

[glisse at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86122

--- Comment #5 from Marc Glisse <glisse at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #2)
> if we want unsigned_type_for to support complex integer types or not.

I think we do (seems super easy). Testing utype can't hurt indeed.

(In reply to Jakub Jelinek from comment #3)
> Note, other uses of unsigned_type_for in match.pd seems to be guarded with
> INTEGRAL_TYPE_P check or are on expressions that can't appear with complex
> operands.

This transform is specifically written in a way that works for vectors (and
should work with complex if unsigned_type_for was extended), it would be sad to
disable it.