Re: [PATCH v2 1/1] virtio-ccw: auto-manage VIRTIO_F_IOMMU_PLATFORM if PV

[David Gibson <david@gibson.dropbear.id.au>]

[-- Attachment #1: Type: text/plain, Size: 1344 bytes --]

On Wed, Jun 10, 2020 at 11:37:14PM +0200, Halil Pasic wrote:
> On Wed, 10 Jun 2020 14:25:54 +1000
> David Gibson <david@gibson.dropbear.id.au> wrote:
> 
> > > > I'm going to definitely have a good look at that. What I think special
> > > > about s390 is that F_ACCESS_PLATFORM is hurting us because all IO needs
> > > > to go through ZONE_DMA (this is a problem of the implementation that
> > > > stemming form a limitation of the DMA API, upstream didn't let me
> > > > fix it).   
> > > 
> > > My understanding is that power runs into similar issues, but I don't
> > > know much about power, so I might be entirely wrong :)  
> > 
> > Sort of, but not to the same extent, I think.
> 
> I'm curious what are the ramifications of a misguided hotplug on POWER?
> Does using F_ACCESS_PLATFORM when it isn't required have any
> significant drawbacks, or are you fine to just go with the safe option?

I expect it will have some performance impact, though it shouldn't be
*that* bad, at least if your guest kernel is ddw / large IOMMU window
capable.

Changing the default would require a machine type version bump, of
course.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]