On Tue, Nov 13, 2012 at 04:09:05PM +0000, Sasha Levin wrote:
> On 11/13/2012 07:16 AM, Will Deacon wrote:
> > On Tue, Nov 13, 2012 at 04:37:38AM +0000, Sasha Levin wrote:
> >> On 11/12/2012 06:57 AM, Will Deacon wrote:
> >>> +struct kvm_mem_bank {
> >>> + struct list_head list;
> >>> + unsigned long guest_phys_addr;
> >>> + void *host_addr;
> >>> + unsigned long size;
> >>> +};
> >>
> >> Can we just reuse struct kvm_userspace_memory_region here? We're also
> >> using different
> >> data types for guest_phys_addr and size than whats in
> >> kvm_userspace_memory_region - that
> >> can't be good.
> >
> > I looked briefly at doing that when I wrote the multi-bank stuff, but I hit
> > a couple of issues:
> >
> > - kvmtool itself tends to use void * for host addresses, rather than
> > the __u64 userspace_addr in kvm_userspace_memory_region
> >
> > - kvm_userspace_memory_region is a superset of what we need (not the
> > end of the world I guess)
> >
> > so you end up casting address types a fair amount. Still, I'll revisit it
> > and see if I can come up with something cleaner.
>
> That's a good point. We used void* while the kernel side is using u64, which
> looks odd.
>
> In that case, let's get everything moved to u64 (obviously not in the scope of
> this patch series).
Ok, I'll update the size field to match in this patch series, then we can
tackle the address discrepancy separately.
Cheers,
Will
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
