On 04.12.2023 17:26, Nicola Vetrini wrote: > On 2023-12-01 17:57, Nicola Vetrini wrote: >> On 2023-11-30 17:41, Jan Beulich wrote: >>> On 29.11.2023 16:24, Nicola Vetrini wrote: >>>> The comment referred to the declaration for do_mca, which >>>> now is part of hypercall-defs.h, therefore the comment is stale. >>> >>> If the comments were stale, the #include-s should also be able to >>> disappear? > >>>> --- a/xen/arch/x86/include/asm/hypercall.h >>>> +++ b/xen/arch/x86/include/asm/hypercall.h >>>> @@ -12,7 +12,7 @@ >>>> #include <xen/types.h> >>>> #include <public/physdev.h> >>>> #include <public/event_channel.h> >>>> -#include <public/arch-x86/xen-mca.h> /* for do_mca */ >>>> +#include <public/arch-x86/xen-mca.h> >>>> #include <asm/paging.h> >>> >>> Here otoh I'm not even sure this public header (or the others) is >>> (are) >>> really needed. >>> >> >> I confirm this. It build even without this header. > > It does appear to be needed after all. I did two differential pipeline > runs, and some jobs fail to compile when I remove the header (e.g., > [1]). Looking trough the build log, it's not entirely clear what is the > relationship, but it seems related to some use of this struct defined in > xen-mca.h: > > typedef struct xen_mc xen_mc_t; > DEFINE_XEN_GUEST_HANDLE(xen_mc_t);
That do_mca()'s parameter type, so in a way the comment is still correct then. Jan
