Re: [Xen-devel] [PATCH 05 of 10] Introduce a grant_entry_v2 structure

[Keir Fraser <keir.fraser@xxxxxxxxxxxxx>]

On 05/10/2009 10:36, "Steven Smith" <Steven.Smith@xxxxxxxxxxxxx> wrote:

>> It would seem that this should be a GUEST_HANDLE, or is there any
>> particular reason to not declare it so?
> Mostly that GUEST_HANDLE isn't 32/64 invariant, and I didn't want to
> have to introduce yet more compat shims.
> 
> If there's a strong consensus that GUEST_HANDLE is preferable then it
> could be switched easily enough, but it seems kind of silly to
> continue introducing more ABIs which aren't 64-bit clean.

I think now we have guest handles we should just use them, and the guest
access macros, consistently. Despite it requiring us to have a (pretty
trivial) compat shim.

This is despite choosing the opposite for the grant_entry_v2 struct and
xen_pfn_t versus uint64_t. There's less machinery and magic around
xen_pfn_t, and also compat shims for shared-memory structures are just plain
annoying.

 -- Keir



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel