Re: [Xen-devel] [RFC v3]Proposal to allow setting up shared memory areas between VMs from xl config file

[Stefano Stabellini <sstabellini@xxxxxxxxxx>]

On Fri, 21 Jul 2017, Julien Grall wrote:
> > >   @x86_cacheattr        can be 'uc', 'wc', 'wt', 'wp', 'wb' or 'suc'.
> > > Default
> > >                         is 'wb'.
> > 
> > Also here, I would write:
> > 
> > @x86_cacheattr  Only 'wb' (write-back) is supported today.
> > 
> > Like you wrote later, begin and end addresses need to be multiple of 4K.
> 
> This is not true. The addresses should be a multiple of the hypervisor page
> granularity.
> 
> I will not be possible to map a 4K chunk in stage-2 when the hypervisor is
> using 16K or 64K page granularity.

Yes, but there are no 16K or 64K hypervisor pages now. So far, we have
not really attemped to say "granularity" for hypevisor pages rather than
4K, given that 4K has always been a solid assumption. But this doc could
be the right time to start doing that :-)

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