Re: [Patch] lto_input_mode_table: on offload targets, check whether mode is actually available [PR127589]

[Richard Biener]

On Fri, Sep 25, 2026 at 1:56 PM Tobias Burnus <tbur...@baylibre.com> wrote:
>
> Tobias Burnus wrote:
> >>> Thus, how about the patch plus adding a comment like above?
> >> That would be nice, yes.   But then the check could also be moved into
> >> that 'else'?
> > You mean as follows?
>
> Or rather like this – with additional tests, esp. for _Complex
> and _Decimal.

Yes.  Though you can possibly simplify that switch () by using
scalar_mode::includes_p (mr).

Richard.

> [Successfully Regtested with amdgcn + nvptx offloading enabled
> on a system with nvptx GPU.]
>
> Tobias