Re: [Xen-devel] [PATCH] iommu: specify page_count rather than page_order to iommu_map/unmap()...

[Paul Durrant <Paul.Durrant@xxxxxxxxxx>]

> -----Original Message-----
> From: Andrew Cooper
> Sent: 18 January 2019 17:41
> To: Paul Durrant <Paul.Durrant@xxxxxxxxxx>; xen-devel@xxxxxxxxxxxxxxxxxxxx
> Cc: Chao Gao <chao.gao@xxxxxxxxx>; Sander Eikelenboom
> <linux@xxxxxxxxxxxxxx>; Jan Beulich <jbeulich@xxxxxxxx>; Wei Liu
> <wei.liu2@xxxxxxxxxx>; Roger Pau Monne <roger.pau@xxxxxxxxxx>; George
> Dunlap <George.Dunlap@xxxxxxxxxx>; Ian Jackson <Ian.Jackson@xxxxxxxxxx>;
> Julien Grall <julien.grall@xxxxxxx>; Konrad Rzeszutek Wilk
> <konrad.wilk@xxxxxxxxxx>; Stefano Stabellini <sstabellini@xxxxxxxxxx>; Tim
> (Xen.org) <tim@xxxxxxx>; Jun Nakajima <jun.nakajima@xxxxxxxxx>; Kevin Tian
> <kevin.tian@xxxxxxxxx>
> Subject: Re: [PATCH] iommu: specify page_count rather than page_order to
> iommu_map/unmap()...
> 
> On 18/01/2019 16:03, Paul Durrant wrote:
> > diff --git a/xen/include/xen/iommu.h b/xen/include/xen/iommu.h
> > index cdc8021cbd..82fb86c7ff 100644
> > --- a/xen/include/xen/iommu.h
> > +++ b/xen/include/xen/iommu.h
> > @@ -111,17 +111,17 @@ enum
> >  #define IOMMU_FLUSHF_modified (1u << _IOMMU_FLUSHF_modified)
> >
> >  int __must_check iommu_map(struct domain *d, dfn_t dfn, mfn_t mfn,
> > -                           unsigned int page_order, unsigned int flags,
> > +                           unsigned int page_count, unsigned int flags,
> >                             unsigned int *flush_flags);
> 
> I'd take the opportunity to make page_count an unsigned long, as we can
> now sensibly issue a single call for an entire BAR, and some graphics
> card BARs are getting to be a ludicrous size.

The 1G order is still vetoed by mmio_order 
(http://xenbits.xen.org/gitweb/?p=xen.git;a=blob;f=xen/arch/x86/mm/p2m.c#l2228) 
so the BAR limit for a single mapping operation is still going to 2M. So I 
think we're a way off needing a 64-bit count.

> 
> Otherwise, LGTM and can also be fixed on commit.
> 

Bear in mind that, if you want to move to a 64-bit count, then the flush 
operations should be adjusted accordingly; which is a non-trivial amount of 
code churn.

  Paul

> ~Andrew
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel