On Mon, Nov 07, 2022 at 10:02:11AM +0100, Richard Biener wrote:
> The following adds a fold_builtins pass at -O0, keying off some
> unwanted optimization and setting pointer alignment of the result
> of __builtin_assume_alignment before removing the call. This
> allows libatomic calls to be elided at -O0 on s390 for
>
> __uint128_t foo(__uint128_t *p)
> {
> return __atomic_load_n((__uint128_t *)__builtin_assume_aligned (p, 16), 0);
> }
>
> not sure how to reliably test this though.
>
> Thoughts?
Do we really need a separate pass for it?
Can't we do it say during gimplification?
> PR tree-optimization/107389
> * tree-pass.h (pass_fold_builtins_O0): Declare.
> * passes.def (pass_fold_builtins_O0): Schedule.
> * tree-ssa-ccp.cc (pass_fold_builtins_O0): Define.
> (pass_fold_builtins): Template, add gate.
> (pass_fold_builtins::execute): Apply alignment to the
> target of __builtin_assume_aligned at -O0. Disable most
> optimizations for the -O0 pass instance.
Jakub
