[ovs-dev] [v13 08/12] dpif-netdev-unixctl.man: Document subtable-lookup-* CMDs

Flavio Leitner fbl at sysclose.org
Tue Jun 22 16:38:55 UTC 2021


On Tue, Jun 22, 2021 at 03:42:46PM +0000, Stokes, Ian wrote:
> > Hi Flavio,
> > 
> > Thanks for the review. My responses are inline.
> > 
> > Cian
> > 
> > > -----Original Message-----
> > > From: Flavio Leitner <fbl at sysclose.org>
> > > Sent: Monday 21 June 2021 19:22
> > > To: Ferriter, Cian <cian.ferriter at intel.com>
> > > Cc: ovs-dev at openvswitch.org; i.maximets at ovn.org
> > > Subject: Re: [ovs-dev] [v13 08/12] dpif-netdev-unixctl.man: Document
> > subtable-lookup-* CMDs
> > >
> > >
> > > Hi,
> > >
> > > This commit could be submitted outside of this patch-set as fix
> > > for commit 9ff7cabfd7 ("dpif-netdev: add subtable-lookup-prio-get
> > > command") and commit 3d018c3ea79d ("dpif-netdev: add subtable lookup
> > > prio set command.").
> > >
> > > This helps to get it merged sooner and reduce this patch-set size.
> > >
> > 
> > I'll remove this patch from the patchset and send to the mailing list separately.
> > I'll wait till the DPIF patchset has been merged to send this, since I don't want
> > there to be rebase conflicts (the DPIF patchset also modifies this part of lib/dpif-
> > netdev-unixctl.man).
> > 
> > I'll add the appropriate Fixes tags.
> 
> @Flavio, If there is an aim to reduce the overall patch number of
> the series then I would recommend the following patches be
> submitted separately also from this series as there is no
> dependency on them to enable DPIF with AVX512.
> 
> [v13 10/12] dpif-netdev/dpcls: Specialize more subtable signatures.
> [v13 11/12] dpdk: Cache result of CPU ISA checks.
> 
> 
> These two are quite small and I think could almost be applied now
> rather than as part of the series as they are modifying existing
> functionality (DPCLS AVX512 supported traffic types and DPDK flag
> caching).
> 
> Thoughts?

The idea is to get unrelated chunks merged sooner, if they make
sense of course, and then we have less patches to carry on. 

If the patches are going to arrive later or it causes more work
on following up patches, then I see no benefit.

It's a suggestion. I am happy to review either way.

Thanks,
fbl


More information about the dev mailing list