On Tue, 6 Oct 2020 06:49:28 -0400
"Michael S. Tsirkin" wrote:
> On Tue, Oct 06, 2020 at 11:58:50AM +0200, Greg Kurz wrote:
> > On Mon, 5 Oct 2020 10:18:03 -0400
> > "Michael S. Tsirkin" wrote:
> >
> > > On Mon, Sep 28, 2020 at 09:37:18AM +0200, Greg Kurz wrote:
> > > > On Mon, 28 Sep 2020 16:23
On Tue, Oct 06, 2020 at 11:58:50AM +0200, Greg Kurz wrote:
> On Mon, 5 Oct 2020 10:18:03 -0400
> "Michael S. Tsirkin" wrote:
>
> > On Mon, Sep 28, 2020 at 09:37:18AM +0200, Greg Kurz wrote:
> > > On Mon, 28 Sep 2020 16:23:43 +1000
> > > David Gibson wrote:
> > >
> > > > On Fri, Sep 25, 2020 at
On Mon, 5 Oct 2020 10:18:03 -0400
"Michael S. Tsirkin" wrote:
> On Mon, Sep 28, 2020 at 09:37:18AM +0200, Greg Kurz wrote:
> > On Mon, 28 Sep 2020 16:23:43 +1000
> > David Gibson wrote:
> >
> > > On Fri, Sep 25, 2020 at 07:29:43PM +0200, Greg Kurz wrote:
> > > > When a vIOMMU is present, any ad
On Mon, Sep 28, 2020 at 09:37:18AM +0200, Greg Kurz wrote:
> On Mon, 28 Sep 2020 16:23:43 +1000
> David Gibson wrote:
>
> > On Fri, Sep 25, 2020 at 07:29:43PM +0200, Greg Kurz wrote:
> > > When a vIOMMU is present, any address comming from the guest is an IO
> > > virtual address, including those
On Mon, 28 Sep 2020 16:23:43 +1000
David Gibson wrote:
> On Fri, Sep 25, 2020 at 07:29:43PM +0200, Greg Kurz wrote:
> > When a vIOMMU is present, any address comming from the guest is an IO
> > virtual address, including those of the vrings. The backend's accesses
> > to the vrings happen through
On Fri, Sep 25, 2020 at 07:29:43PM +0200, Greg Kurz wrote:
> When a vIOMMU is present, any address comming from the guest is an IO
> virtual address, including those of the vrings. The backend's accesses
> to the vrings happen through vIOMMU translation : the backend hence
> only logs the final gue
Cc'ing Jason since this was detected using vhost-net.
On Fri, 25 Sep 2020 19:29:43 +0200
Greg Kurz wrote:
> When a vIOMMU is present, any address comming from the guest is an IO
> virtual address, including those of the vrings. The backend's accesses
> to the vrings happen through vIOMMU transla
When a vIOMMU is present, any address comming from the guest is an IO
virtual address, including those of the vrings. The backend's accesses
to the vrings happen through vIOMMU translation : the backend hence
only logs the final guest physical address, not the IO virtual one.
It thus doesn't make s