[ovs-dev] [dpdk-latest PATCH] netdev-dpdk: Prefix network structures with rte_.

Ilya Maximets i.maximets at samsung.com
Tue Jun 11 16:30:14 UTC 2019


On 11.06.2019 19:10, David Marchand wrote:
> 
> 
> On Tue, Jun 11, 2019 at 5:36 PM Ilya Maximets <i.maximets at samsung.com <mailto:i.maximets at samsung.com>> wrote:
> 
>     On 11.06.2019 18:21, Ian Stokes wrote:
>     > On 6/11/2019 3:40 PM, Aaron Conole wrote:
>     >> Ian Stokes <ian.stokes at intel.com <mailto:ian.stokes at intel.com>> writes:
>     >>
>     >>> On 6/10/2019 3:57 PM, Ian Stokes wrote:
>     >>>> On 6/6/2019 12:36 PM, David Marchand wrote:
>     >>>>>
>     >>>>>
>     >>>>> On Thu, Jun 6, 2019 at 1:25 PM Ian Stokes <ian.stokes at intel.com <mailto:ian.stokes at intel.com>
>     >>>>> <mailto:ian.stokes at intel.com <mailto:ian.stokes at intel.com>>> wrote:
>     >>>>>
>     >>>>>      On 6/4/2019 12:14 PM, David Marchand wrote:
>     >>>>>       > On Tue, Jun 4, 2019 at 11:29 AM David Marchand
>     >>>>>       > <david.marchand at redhat.com <mailto:david.marchand at redhat.com> <mailto:david.marchand at redhat.com <mailto:david.marchand at redhat.com>>
>     >>>>>      <mailto:david.marchand at redhat.com <mailto:david.marchand at redhat.com>
>     >>>>>      <mailto:david.marchand at redhat.com <mailto:david.marchand at redhat.com>>>> wrote:
>     >>>>>       >
>     >>>>>       >     Following a rework of dpdk network structures names [1],
>     >>>>>      update the
>     >>>>>       >     concerned parts.
>     >>>>>       >
>     >>>>>       >     Ran Olivier script:
>     >>>>>       >     sh prefix-net-rte.sh $(find -name "*dpdk*.c")
>     >>>>>       >     sh prefix-net-rte.sh $(find -name "*dpdk*.h")
>     >>>>>       >     sh prefix-net-rte.sh $(find -name "*rte*.c")
>     >>>>>       >     sh prefix-net-rte.sh $(find -name "*rte*.h")
>     >>>>>       >
>     >>>>>       >     Plus an extra pass following further changes [2]:
>     >>>>>       >     old=RTE_IPv4
>     >>>>>       >     new=RTE_IPV4
>     >>>>>       >     git grep -lw $old | xargs sed -i -e "s/\<$old\>/$new/g"
>     >>>>>       >
>     >>>>>       >     old=RTE_ETHER_TYPE_IPv4
>     >>>>>       >     new=RTE_ETHER_TYPE_IPV4
>     >>>>>       >     git grep -lw $old | xargs sed -i -e "s/\<$old\>/$new/g"
>     >>>>>       >
>     >>>>>       >     old=RTE_ETHER_TYPE_IPv6
>     >>>>>       >     new=RTE_ETHER_TYPE_IPV6
>     >>>>>       >     git grep -lw $old | xargs sed -i -e "s/\<$old\>/$new/g"
>     >>>>>       >
>     >>>>>       >     1: http://mails.dpdk.org/archives/dev/2019-May/132612.html
>     >>>>>       >     2: https://git.dpdk.org/dpdk/commit/?id=0c9da7555da8
>     >>>>>       >
>     >>>>>       >
>     >>>>>       > Olivier noticed that I had used an early version of his patch.
>     >>>>>       > The published one handles the update on RTE_IPv4.
>     >>>>>       > I tried the last version which gives the same result anyway.
>     >>>>>       > So the extra pass is unnecessary.
>     >>>>>       >
>     >>>>>       > I can send a v2 to update the commitlog accordingly.
>     >>>>>       >
>     >>>>>
>     >>>>>      Hi David,
>     >>>>>
>     >>>>>      thanks for this, upon inspection the patch looks fine and I can
>     >>>>> confirm
>     >>>>>      that dpdk-latest is now building with Master of DPDK again.
>     >>>>>
>     >>>>>      I'm just in the process of running a few smoke tests to make sure
>     >>>>>      there's no issues functionally (I don't expect to see any as the
>     >>>>>      changes
>     >>>>>      seem straight forward).
>     >>>>>
>     >>>>>      WRT the v2, what exactly do you want to change in the commit? If it's
>     >>>>>      trivial I can amend it before committing.
>     >>>>>
>     >>>>>
>     >>>>>
>     >>>>> I just stripped the useless part in the commitlog and put a link to
>     >>>>> Olivier mail which contained his script.
>     >>>>> You can see the commitlog here:
>     >>>>> https://github.com/david-marchand/ovs/commit/9d367de7d323c28f7c89d590ff60373c47ffa073
>     >>>>>
>     >>>>>
>     >>>>>
>     >>>>>
>     >>>>>      I'll be applying this to dpdk-latest and dpdk-hwol branches but
>     >>>>> not ovs
>     >>>>>      master (master is still using DPDK 18.11.1 currently so no need for
>     >>>>>      these changes until it moves to 19.11).
>     >>>>>
>     >>>>>
>     >>>>>
>     >>>>> Yes, makes sense.
>     >>>>> Thanks Ian.
>     >>>>
>     >>>> Thanks David, validated and pushed to dpdk-latest and dpdk-wol.
>     >>>
>     >>> Good catch actually. In the past we had previously tracked the latest
>     >>> DPDK release to compile against, but now that dpdk-latest is used with
>     >>> the UNH DPDK CI it probably makes more sense to track DPDK master at
>     >>> that's what dpdk-latest looks to enable compilation of.
>     >>>
>     >>> I'm not against this, would be interested in what peoples thoughts are
>     >>> and if so we can modify travis for the dpdk-latest branch.
>     >>
>     >> At least for this part (per-branch), the travis yml is read on a
>     >> per-branch basis.  If it changes on one branch, only that branch will
>     >> be affected.  Is this what you mean?
>     >
>     > Yes, travis would be changed to track master just for dpdk-latest is my understanding. OVS master and OVS release branches should still only track the DPDK LTS release they are currently validated against in their travis yml.
> 
>     Makes sense. Broken travis builds on dpdk-latest branch are annoying.
> 
> 
> I have changes that apply to both master and dpdk-latest branch, then a change for the switch to dpdk master branch in dpdk-latest.
> 
> I can post the changes for master on top of your patch that disables kni/igb_uio.
> We merge yours and mine in master, then pull master into dpdk-latest.
> 
> Then I would only send the switch to dpdk master branch for the dpdk-latest ovs branch.
> 
> Is this ok this way ?

OK for me in general. However, I think that it's time for 'git pull --rebase'
on dpdk-latest branch since it's already a bit diverged and some patches, including
travis patches, will have conflicts.

So, I agree with your scheme, but with rebase instead of pull.

BTW, I'm leaving the office now and will back only on Monday due to public holidays
and a small PTO.

Best regards, Ilya Maximets.


More information about the dev mailing list