On Wed, Mar 13, 2019 at 11:53 AM Andrew Pinski <pins...@gmail.com> wrote: > This testcase seems generic, that is it has no ABI dependent values > attached to it. > Can it be moved to a more generic location instead? Maybe there are > other targets which get this incorrect also.
I'm not aware of any other target that passes structs the same way. RISC-V fails the testcase because it will decompose a struct containing only floats into fields and pass the fields as scalars. IA-64 and Aarch64 have support for HFA (homogenous FP aggregates), but this case doesn't qualify as an HFA because of the empty struct, and hence this case gets no special treatment for those targets and isn't broken. Jim