[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH 3/3] tools: introduce parameter max_ranges.



> From: Ian Campbell [mailto:ian.campbell@xxxxxxxxxx]
> Sent: Tuesday, January 19, 2016 11:19 PM
> 
> On Tue, 2016-01-19 at 15:04 +0000, Wei Liu wrote:
> 
> This patch doesn't seem to have been CCd to the tools maintainers, adding
> Ian too, I think everyone else was picked up along the way.
> 
> Please use ./scripts/get_maintainers.pl in the future.
> 
> > On Tue, Jan 19, 2016 at 02:47:40PM +0000, Paul Durrant wrote:
> > [...]
> > > > ranges so perhaps the parameter name could be
> > > > 'max_wp_memory_ranges'?
> > > > >
> > > >
> > > > What does "WP" mean? "Write Protected"?
> > > >
> > >
> > > Yes.
> > >
> > > > Is this parameter closely related to IOREQ server? Should it contain
> > > > "ioreq" somehow?
> > > >
> > >
> > > It is closely related but ioreq server is an implementation detail so
> > > do we want to expose it as a tunable? The concept we need to capture
> > > is that the toolstack can tune the limit of the maximum number of
> > > pages in the VM that can be set such that writes are emulated (but
> > > reads are as for normal ram). Or I guess we could get very specific
> > > and call it something like 'max_gtt_shadows'?
> >
> > I would prefer generic concept in this case ("wp"). Let's wait a bit for
> > other people to voice their opinion.
> >
> > Whichever one we pick it the meaning of the acronym needs to be clearly
> > documented...
> 
> I've got no ideas for a better name, "max_ranges" is clearly too generic
> though.
> 
> One thought -- does XenGT require some other configuration option to enable
> it or maybe a privilege which the target domain must necessarily have?
> Could we use something like one of those to cause the t/stack to just DTRT
> without the user having to micromanage the amount of pages which are
> allowed to have this property?
> 

Using "wp" is clear to me.

As a feature this write-protection has nothing to be GPU virtualization 
specific.
In the future the same mediated pass-through idea used in XenGT may be
used on other I/O devices which need to shadow some structure w/ requirement
to write-protect guest memory. So it's not good to tie this to either XenGT
or GTT.

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


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.