On Wed, 2019-12-04 at 10:04 +0100, Jan Beulich wrote:
> On 03.12.2019 18:10, Andrew Cooper wrote:
> > --- a/xen/include/public/hvm/dm_op.h
> > +++ b/xen/include/public/hvm/dm_op.h
> > @@ -324,7 +324,7 @@ struct xen_dm_op_inject_event {
> > /* IN - error code (or ~0 to skip) */
> > uint32_t error_code;
> > uint32_t pad1;
> > - /* IN - CR2 for page faults */
> > + /* IN - type-specific extra data (%cr2 for #PF, pending_dbg
> > for #DB) */
> > uint64_aligned_t cr2;
> > };
>
> How about
>
> uint32_t error_code;
> uint32_t pad1;
> /* IN - type-specific extra data (%cr2 for #PF, pending_dbg for
> #DB) */
> #if defined(__XEN__) || defined(__XEN_TOOLS__)
> uint64_aligned_t extra;
> #else
> uint64_aligned_t cr2;
> #endif
> };
>
> or something along these lines (e.g. could also be an unnamed
> union guarded by a __GNUC__ check, or a __XEN_INTERFACE_VERSION__
> conditional), to have a less confusing name in Xen and the tools?
> Either way hypervisor bits
>
Honestly, this looks a little bit scary.
However, it might be a good opportunity to discuss upgrading to C11,
which supports anonymous unions and offers a cleaner (and safer) way to
handle this type of problems.
Best regards,
Petre
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
