Re: [Xen-devel] [RFC v2 06/15] vt-d: Extend struct iremap_entry to support VT-d Posted-Interrupts

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 16.06.15 at 08:18, <feng.wu@xxxxxxxxx> wrote:
>> From: Jan Beulich [mailto:JBeulich@xxxxxxxx]
>> Sent: Tuesday, June 09, 2015 10:25 PM
>> >>> On 08.05.15 at 11:07, <feng.wu@xxxxxxxxx> wrote:
>> > @@ -302,16 +302,28 @@ struct iremap_entry {
>> >              vector  : 8,
>> >              res_2   : 8,
>> >              dst     : 32;
>> > -    }lo;
>> > -  };
>> > -  union {
>> > -    u64 hi_val;
>> > +        u64 sid     : 16,
>> > +            sq      : 2,
>> > +            svt     : 2,
>> > +            res_3   : 44;
> 
> For this, are we worth using u32 for this one?
> 
>       u32 sid: 16,
>           sq: 2,
>           svt: 2,
>           res_3: 12;
>       u32 res_4;
> 
> 
>> > +    }remap;
>> >      struct {
>> > +        u64 p       : 1,
>> > +            fpd     : 1,
>> > +            res_1   : 6,
>> > +            avail   : 4,
>> > +            res_2   : 2,
>> > +            urg     : 1,
>> > +            im      : 1,
>> > +            vector  : 8,
>> > +            res_3   : 14,
>> > +            pda_l   : 26;
> 
> Also this one?
> 
>       ...
>               res_3: 8
>       u32 res_4: 6,
>           pda_l: 26;
> 
> 
>> >          u64 sid     : 16,
>> >              sq      : 2,
>> >              svt     : 2,
>> > -            res_1   : 44;
> 
> This one?

Yes in all three case, subject to ...

>> > -    }hi;
>> > +            res_4   : 12,
>> > +            pda_h   : 32;
>> > +    }post;
>> >    };
>> >  };
>> 
>> Same here - unless the VT-d maintainers disagree, I think using u32
>> wherever possible would be preferable over using u64, as well as
>> avoiding bitfields for members filling an entire word.

... the maintainers' not disagreeing.

Jan


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