Re: [Xen-devel] [PATCH for 4.6] VT-d: Create IOMMU mappings for RMRR regions if shared EPT is not being used

["Chen, Tiejun" <tiejun.chen@xxxxxxxxx>]

On 8/27/2015 4:40 PM, Malcolm Crossley wrote:
> On 27/08/15 03:59, Chen, Tiejun wrote:
> > This kind of issue is already gone.
> >
> > https://www.mail-archive.com/xen-devel@xxxxxxxxxxxxx/msg32464.html
> There is a bug in the code you refer to above which results in no IOMMU page 
> table
> mappings being created if the guest domain is not sharing it's EPT page tables 
> with
> the IOMMU.
>
> set_identity_p2m_entry only configures the EPT page tables and does not 
> configure
> the IOMMU page tables.
Okay, I got what you mean.

Instead, could you insert iommu_{map,unmap_page() into 
{set,clear}_identity_p2m_entry()? I think this can make 
{set,clear}_identity_p2m_entry approachable in all circumstances.
Kevin and Jan,

Is this fine?

Thanks
Tiejun

> We had a real world regression (with xen 4.6-rc1) on a Intel Haswell system with
> integrated graphics.
>
> The patch below resolves the regression.
>
> Malcolm
>
> > Thanks
> > Tiejun
> >
> > On 8/26/2015 11:49 PM, Malcolm Crossley wrote:
> > > Add RMRR 1:1 IOMMU mappings to IOMMU page tables if EPT page table are not being
> > > shared with the IOMMU.
> > >
> > > This is a regression in behaviour versus Xen 4.5.
> > >
> > > Signed-off-by: Malcolm Crossley <malcolm.crossley@xxxxxxxxxx>
> > > ---
> > >   xen/drivers/passthrough/vtd/iommu.c | 23 ++++++++++++++++++++---
> > >   1 file changed, 20 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/xen/drivers/passthrough/vtd/iommu.c 
> > > b/xen/drivers/passthrough/vtd/iommu.c
> > > index 836aed5..89de741 100644
> > > --- a/xen/drivers/passthrough/vtd/iommu.c
> > > +++ b/xen/drivers/passthrough/vtd/iommu.c
> > > @@ -1839,8 +1839,16 @@ static int rmrr_identity_mapping(struct domain *d, 
> > > bool_t map,
> > >
> > >               while ( base_pfn < end_pfn )
> > >               {
> > > -                if (   (d, base_pfn) )
> > > -                    ret = -ENXIO;
> > > +                if ( iommu_use_hap_pt(d) )
> > > +                {
> > > +                    if ( clear_identity_p2m_entry(d, base_pfn) )
> > > +                            ret = -ENXIO;
> > > +                }
> > > +                else
> > > +                {
> > > +                    if ( intel_iommu_unmap_page(d, base_pfn) )
> > > +                        ret = -ENXIO;
> > > +                }
> > >                   base_pfn++;
> > >               }
> > >
> > > @@ -1855,7 +1863,16 @@ static int rmrr_identity_mapping(struct domain *d, 
> > > bool_t map,
> > >
> > >       while ( base_pfn < end_pfn )
> > >       {
> > > -        int err = set_identity_p2m_entry(d, base_pfn, p2m_access_rw, flag);
> > > +        int err;
> > > +        if ( iommu_use_hap_pt(d) )
> > > +        {
> > > +            err = set_identity_p2m_entry(d, base_pfn, p2m_access_rw, flag);
> > > +        }
> > > +        else
> > > +        {
> > > +            err = intel_iommu_map_page(d, base_pfn, base_pfn,
> > > +                                       IOMMUF_readable|IOMMUF_writable);
> > > +        }
> > >
> > >           if ( err )
> > >               return err;
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel