Re: [Xen-devel] [PATCH v5 07/15] iommu: track reserved ranges using a rangeset

["Tian, Kevin" <kevin.tian@xxxxxxxxx>]

> From: Paul Durrant [mailto:Paul.Durrant@xxxxxxxxxx]
> Sent: Tuesday, August 7, 2018 4:16 PM
> 
> > -----Original Message-----
> > From: Tian, Kevin [mailto:kevin.tian@xxxxxxxxx]
> > Sent: 07 August 2018 04:04
> > To: Paul Durrant <Paul.Durrant@xxxxxxxxxx>; xen-
> devel@xxxxxxxxxxxxxxxxxxxx
> > Cc: Jan Beulich <jbeulich@xxxxxxxx>
> > Subject: RE: [PATCH v5 07/15] iommu: track reserved ranges using a
> rangeset
> >
> > > From: Paul Durrant [mailto:paul.durrant@xxxxxxxxxx]
> > > Sent: Saturday, August 4, 2018 1:22 AM
> > >
> > > Ranges that should be considered reserved in the IOMMU are not
> > > necessarily
> > > limited to RMRRs. If iommu_inclusive_mapping is set then any frame
> > > number
> > > falling within an E820 reserved region should also be considered as
> > > reserved in the IOMMU.
> >
> > I don't think it is a good extension. RMRR by definition requires
> > identity mapping in IOMMU page table, thus must be also reserved
> > in bfn address space (to avoid being used for other purpose).
> > Normal E820 reserved regions have no such requirement. Guest
> > can use same bfn location for any purpose. I'm not sure why we
> > want to add more limitations here.
> 
> Have you tried booting an average Intel based server without identity
> mapped E820 reserved regions? Good luck with that, because I think about
> half the servers on the planet will wedge up if you only give them RMRRs.
> My Dell R730 certainly will and I'd hazard a guess that any other Dell box
> we have with an iDRAC will too.
> Note the behaviour is predicated on 'iommu_inclusive_mapping' being set
> so if you feel lucky then you can turn it off :-)
> 

I know that fact... looks I misunderstood the patch - I thought you
blindly add all reserved regions in this patch bypassing the existing
control from inclusive_mapping. Now re-read the description and
code you are just giving background :-)

Thanks
Kevin

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