[dpdk-dev] [PATCH] vfio: support iommu group zero

[Alex Williamson]

On Wed, 2015-12-09 at 16:52 -0800, Stephen Hemminger wrote:
> On Wed, 09 Dec 2015 16:22:14 -0700
> Alex Williamson <alex.williamson at redhat.com> wrote:
> 
> > On Wed, 2015-12-09 at 15:12 -0800, Stephen Hemminger wrote:
> > > On Wed, 09 Dec 2015 23:49:59 +0100
> > > Thomas Monjalon <thomas.monjalon at 6wind.com> wrote:
> > > 
> > > > 2015-12-09 13:58, Stephen Hemminger:
> > > > > On Wed, 09 Dec 2015 22:12:33 +0100
> > > > > Thomas Monjalon <thomas.monjalon at 6wind.com> wrote:
> > > > > 
> > > > > > 2015-12-09 09:55, Stephen Hemminger:
> > > > > > > The current implementation of VFIO will not with the new no-IOMMU mode
> > > > > > > in 4.4 kernel. The original code assumed that IOMMU group zero would
> > > > > > > never be used. Group numbers are assigned starting at zero, and up
> > > > > > > until now the group numbers came from the hardware which is likely
> > > > > > > to use group 0 for system devices that are not used with DPDK.
> > > > > > > 
> > > > > > > The fix is to allow 0 as a valid group and rearrange code
> > > > > > > to split the return value from the group value.
> > > > > > > 
> > > > > > > Signed-off-by: Stephen Hemminger <stephen at networkplumber.org>
> > > > > > > ---
> > > > > > > Why was this ignored? It was originally sent on 26 Oct 15 back
> > > > > > > when IOMMU discussion was lively.
> > > > > > 
> > > > > > There was no review of this patch.
> > > > > > The patch has been marked as deferred recently when it was too late
> > > > > > to do such feature changes in DPDK code:
> > > > > > 	http://dpdk.org/dev/patchwork/patch/8035/
> > > > > 
> > > > > This is why as a fallback the MAINTAINER has to review the patch
> > > > > or direct a sub-maintainer to do it. I think almost 2 months is
> > > > > plenty of time for review.
> > > > 
> > > > 27 October was 3 days before the feature deadline.
> > > > And you have not pinged about it since then.
> > > > But that's true I have missed the importance of this patch.
> > > > Would it help to have it integrated today?
> > > > Are you sure it won't break something else?
> > > 
> > > Could the original VFIO submitter from Intel review it.
> > 
> > vfio group 0 has always been valid, but it's unlikely that you'd ever
> > hit it in regular usage since it will typically be the root bus device.
> > It's only with no-iommu mode in vfio that it's common, but that's
> > getting reverted for Linux v4.4, so that may change your priorities
> > about squeezing this in at the last minute.  Thanks,
> 
> Why, who objected? It was useful and working fine as far as I tested.
> Really wanted to get to the per-queue stuff, but that was harder to
> got working (even on regular IOMMU).

I objected because nobody has spoken up for the month that I've been
asking to be notified if a user has been tested.  You can ask for it
again in v4.5 with evidence showing that it works and meets your needs.