Re: [PATCH v4] x86/p2m: use large pages for MMIO mappings

[Ian Campbell <ian.campbell@citrix.com>]

On Mon, 2016-01-25 at 07:16 -0700, Jan Beulich wrote:
> > > > On 25.01.16 at 15:05, <ian.campbell@citrix.com> wrote:
> > On Mon, 2016-01-25 at 06:54 -0700, Jan Beulich wrote:
> > > > > > On 25.01.16 at 13:16, <ian.campbell@citrix.com> wrote:
> > > > On Fri, 2016-01-22 at 08:42 -0700, Jan Beulich wrote:
> > > > > +#define MAP_MMIO_MAX_ITER 64 /* pretty arbitrary */
> > > > > +
> > > > 
> > > > I suppose no existing in-tree code exceeds that (or there'd be more
> > > > patch
> > > > here).
> > > 
> > > There simply is no in-tree user other than the domctl on x86.
> > 
> > Right, I meant callers of the domctl (via libxc)
> 
> Then I don't understand the question. The domctl clearly can be
> invoked with higher MFN counts; the #define only establishes
> the cut off point for the code to return "partial success", directing
> the caller to re-invoke the operation after updating inputs. I.e.
> nothing else than an implementation detail which could be changed
> without affecting any well behaved caller.

Sorry, I'd confused myself into thinking the eventual caller would need to
handle the E2BIG, whereas you (of course) deal with it in the libxc
wrapper.

Ian.