https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125311
--- Comment #6 from anlauf at gcc dot gnu.org ---
Oops, copy and paste... I meant:
(In reply to anlauf from comment #5)
> The problem is that -fcheck= is handled in gfc_handle_option(), which comes
> before gfc_post_options where we handle -fcheck=bounds .
^^^-fbounds-check
> So the issue is that currently
>
> -fcheck=bounds -fcheck=no-bounds
equals
-fbounds-check -fcheck=no-bounds
> equals
>
> -fcheck=no-bounds -fcheck=bounds
>
> :-(
>
> It looks like the problem is in the driver.
