WARNING - OLD ARCHIVES

This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
   
 
 
Xen 
 
Home Products Support Community News
 
   
 

xen-devel

Re: [Xen-devel] 32/64-bit hypercall interface - padding


On 3 Oct 2005, at 20:16, Hollis Blanchard wrote:

However, since we're not actually changing the size of any types, this
change isn't essential to rush through before 3.0, though it might be
nice as it should be very low risk.

I'm working on this patch now.

However, ureg_t alone will not alleviate the padding issues that the dom0_ops structures have. Since we're insisting on using types that change size (and alignment requirements), only reordering the fields will help there. Poor structure layout impacts memory usage and cache footprint, as you mention
above.

Well, at least this patch will be benign, but on its own it is also *totally* pointless. You can't use it to get 64-bit struct arrangement on 32-bit builds because it will leave pointers as 32-bit aligned and sized fields. We could macro up pointer fields I suppose:
#define XENIF_PTR(type, name) type name
So that macro can be overridden to get 64-bit sized and aligned pointer fields?

And ureg_t is a name that's bound to clash with something down the road. Maybe xenreg_t, or even just be explicit about what it is and call it xenif_ulong_t?

 -- Keir


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