Re: [Qemu-devel] Flatview rendering scalability issue

[Peter Xu]

On Tue, Mar 12, 2019 at 01:22:18PM +0100, Paolo Bonzini wrote:
> On 12/03/19 13:14, Peter Xu wrote:
> >> You're right.  However, the DMAR FlatView only has three sections so I
> >> suspect it's not a big deal if we keep it per-device.  You'd still have
> >> O(n) flatviews when the IOMMU is present and DMAR is enabled, but they
> >> would have a constant number of sections so the cost overall is still
> >> O(n) and not O(n^2).  If the IOMMU is present but DMAR is disabled, all
> >> VT-d address spaces would still share the same FlatView vtd_root_nodmar,
> >> and that is where the performance loss happens.
> >>
> >> The final scheme would be same as above with vtd_root_dmar replaced by
> >> vtd_root_dmar_%d.
> > 
> > Just to make sure I understand - do you mean to keep the DMAR-enabled
> > scenario as it is (I think it should be at least as slow as the perf
> > number provided by Sergio)
> 
> No, the DMAR-enabled scenario would not be (too) slow because the
> vtd_root_dmar FlatView only has three sections.  What makes things slows
> is not sharing the vtd_root_nodmar FlatView, which has hundreds of sections.

Ah I see the point!

> 
> > but we only fix up when guest doesn't
> > enable DMAR?  Asked since IIUC normally people won't specify "-device
> > intel-iommu" if they don't need the IOMMU functionality at all (and
> > DMAR is the major one), and if it's not specified then we should not
> > suffer from the performance degradation after all?
> 
> We would also suffer from performance degradation when using iommu=pt,
> which is the common case for VFIO for example.

You are right.  And AFAIU iommu=pt should be used no matter what (IIRC
people have discussed about making it the default) so indeed it can
cover a lot of cases.

> 
> > (But hmm... I think it could benefit users who only need IR/x2apic...)
> > 
> > Btw, how do you think about my previous proposal?  Do you think it's
> > doable somehow? (Make IOMMUMemoryRegion to have only MR pointer that
> > points to the shared MRs)
> 
> I'm not sure, but it would certainly be too complicated for 4.0.

Yeah if nodmar is the only bottleneck so far then I would completely
agree with the approach.  I'll draft something tomorrow and see
whether it works.

Thanks,

-- 
Peter Xu