Re: [Xen-devel] [PATCH RFC] xen/grant_table: deprecate GNTTABOP_unmap_and_replace

[Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>]

On Mon, 22 Jul 2013, Konrad Rzeszutek Wilk wrote:
> On Sun, Jul 21, 2013 at 06:34:59PM +0100, Stefano Stabellini wrote:
> > GNTTABOP_unmap_and_replace has two issues:
> > - it unconditionally replaces the mapping passed in new_addr with 0;
> 
> OK, so the caller needs to save this. Perhaps you can also add a patch
> to the header file to mention this bug since this API call is quite
> "baked"
> 
> > - it doesn't support GNTMAP_contains_pte mappings on x86, returning a
> > general error instead of some forms of ENOSYS.
> 
> Is there a specific version of Xen that has this differently? If the
> generel error is replaced with a proper ENOSYS what is the fallout?
> 
> > 
> > Deprecate GNTTABOP_unmap_and_replace and introduce a new
> > GNTTABOP_unmap_and_replace (12) that returns GNTST_enosys for
> > GNTMAP_contains_pte requests and doesn't zero the mapping at new_addr.
> 
> Or just introduce v2 of said call and leave the old one as is. The
> Linux code can try the new one during bootup and if it fails use
> the fallback.
> 

Right, but if we have the fallback code in Linux we might as well use it
and avoid messing with the grant table interface here. That's why I
decided to drop this patch.

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