Re: [Xen-devel] [PATCH 2/2] xenfb: Add comment documentation

[Owen Smith <owen.smith@xxxxxxxxxx>]

> -----Original Message-----
> From: David Vrabel
> Sent: 22 September 2014 11:16
> To: Owen Smith; xen-devel@xxxxxxxxxxxxx
> Cc: Stefano Stabellini; qemu-devel@xxxxxxxxxx; Ian Campbell
> Subject: Re: [Xen-devel] [PATCH 2/2] xenfb: Add comment documentation
> 
> On 22/09/14 10:04, Owen smith wrote:
> > Add documentation for page-ref, page-gref and event-channel.
> >
> > Signed-off-by: Owen smith <owen.smith@xxxxxxxxxx>
> > ---
> >  xen/include/public/io/fbif.h | 25 +++++++++++++++++++++++++
> >  1 file changed, 25 insertions(+)
> >
> > diff --git a/xen/include/public/io/fbif.h
> > b/xen/include/public/io/fbif.h index cc25aab..ba3f524 100644
> > --- a/xen/include/public/io/fbif.h
> > +++ b/xen/include/public/io/fbif.h
> > @@ -26,6 +26,31 @@
> >  #ifndef __XEN_PUBLIC_IO_FBIF_H__
> >  #define __XEN_PUBLIC_IO_FBIF_H__
> >
> > +/*
> > + * Frontend XenStore Nodes
> > + * -----------------------
> > + *
> > + * page-ref
> > + *   Values: <uint64_t>
> > + *   Optional, "page-gref" is used if "page-ref" is not set.
> > + *
> > + *   The MFN of a page of memory for the shared ring structures. If not
> > + *   present, "page-gref" must be set.page-ref" overrides "page-gref".
> 
> I think you mean GFN here, not MFN.

GFN would be more correct.

> 
> > + * page-gref
> > + *   Values: <uint32_t>
> > + *   Only required if "page-ref" is NOT set.
> > + *
> > + *   A grant reference to the memory page to be mapped for the shared
> ring
> > + *   structures. Must be present if "page-ref" is not present.
> 
> Should there be some negotiation for which of page-ref or page-gref is
> required/preferred by the backend?
> 
> The same applies to kbdif as well.
> 

I've not added negotiation, as I wanted minimum changes when "page-ref" is 
specified.
With the kbdif device only, not investigated the vfb, a backend only supporting 
"page-gref" 
will break a lot of frontends (I saw breaks in Ubuntu, which advertises both if 
I forced page-gref)

> David

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