Re: [PATCH 1/5] x86/vioapic: move domain out of hvm_vioapic struct

["Jan Beulich" <JBeulich@suse.com>]

>>> On 23.02.17 at 12:52, <roger.pau@citrix.com> wrote:
> @@ -204,8 +205,7 @@ static void vioapic_write_indirect(
>              break;
>          }
>  
> -        vioapic_write_redirent(
> -            vioapic, redir_index, vioapic->ioregsel&1, val);
> +        vioapic_write_redirent(d, redir_index, vioapic->ioregsel&1, val);

Please correct obvious coding style violations in cases like this. Of
course here it can be taken care of while committing.

> @@ -224,7 +224,7 @@ static int vioapic_write(
>          break;
>  
>      case VIOAPIC_REG_WINDOW:
> -        vioapic_write_indirect(vioapic, val);
> +        vioapic_write_indirect(v->domain, val);

I'll assume this will clarify itself with later patches: At this point it
feels wrong to lose the IO-APIC pointer in places like this, as from
domain and value alone you won't be able to reconstruct it. I.e.
I would have considered it more natural if you simply added
another function parameter.

I also wonder whether in at least some of the cases the new
struct domain * parameters of functions couldn't be const-
qualified.

In any event I'll want to have looked at more of this series to
understand whether the patch here is okay in its current form.

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel