Re: [Xen-devel] [PATCH v8.1] xSplice v1 design and implementation.

["Jan Beulich" <jbeulich@xxxxxxxx>]

>>> Konrad Rzeszutek Wilk <konrad@xxxxxxxxxx> 04/15/16 2:55 AM >>>
>On Thu, Apr 14, 2016 at 09:17:14AM -0600, Jan Beulich wrote:
>> >>> Andrew Cooper <andrew.cooper3@xxxxxxxxxx> 04/14/16 5:14 PM >>>
>> >On 14/04/16 15:26, Konrad Rzeszutek Wilk wrote:
>> >> @@ -312,8 +307,8 @@ struct xsplice_patch_func {
>> >>  };  
>> >>  </pre>
>> >>  
>> >> -The size of the structure is 64 bytes or 52 bytes if compiled under 
>> >> 32-bit
>> >> -hypervisors.
>> >> +The size of the structure is 64 bytes on 64-bit hypervisors. It will be
>> >> +52 on 32-bit hypervisors.
>> >
>> >I would go further an explicitly declare this API unstable, as patches
>> >must be built against an exact hypervisor source.  This also gives us
>> >leaway in the future.
>
>It will present a problem to the xsplice-tools.git which are outside
>the tree. Right now they have their own header file (which mirrors this
>one). But if we make it unstable and start mucking around then folks
>using the tool would need to pass parameters for 'what version' is
>this.
>
>I would rather not have this happen - and now that we do not need
>the #ifdef BITS_PER_LONG we can also remove the __XEN__ conditional.

I don't think that's the case: The mere use of pointer types in there makes it
so the type shouldn't be exposed to entities executing outside of hypervisor
context.

Jan


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