[Bug target/88765] powerpc64le-linux-gnu sub-optimal code generation for builtin atomic ops

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88765

--- Comment #1 from Segher Boessenkool <segher at gcc dot gnu.org> ---
So in the first example, why is the branch not forwarded already?