From: Yuanhan Liu <yuanhan.liu@linux.intel.com>
To: Gopakumar Choorakkot Edakkunni <gopakumar.c.e@gmail.com>
Cc: dev@dpdk.org
Subject: Re: [dpdk-dev] virtio "how to restart applications" - //dpdk.org/doc/virtio-net-pmd
Date: Fri, 17 Mar 2017 13:24:33 +0800 [thread overview]
Message-ID: <20170317052433.GZ18844@yliu-dev.sh.intel.com> (raw)
In-Reply-To: <CABK1yFBLxJWmEwUrq5FPg98mfKGiJP4_x71f6m2e+GxUFkj7Bg@mail.gmail.com>
On Thu, Mar 16, 2017 at 10:20:30PM -0700, Gopakumar Choorakkot Edakkunni wrote:
> >> When I was saying dpdk version, I meant the DPDK version with OVS.
>
> Oh I see! My apologies for the misuderstanding. The dpdk version used by host
> ovs should be dpdk2.2, the guest process uses dpdk16.07. The OVS process is not
> getting restarted, what is getting restarted is the guest process using
> dpdk16.07 - so the above clarifications you had about virtio being
> reset-before-opened on guest restart - does that still hold good or does that
> need the HOST side dpdk to be 16.04 or above ?
Yes, the HOST dpdk should be >= v16.04.
--yliu
>
> >> And yes, the fixes are not included in the DPDK required for OVS 2.4.
>
> Thanks for the info.
>
> Rgds,
> Gopa.
>
> On Thu, Mar 16, 2017 at 10:13 PM, Yuanhan Liu <yuanhan.liu@linux.intel.com>
> wrote:
>
> On Thu, Mar 16, 2017 at 09:56:01PM -0700, Gopakumar Choorakkot Edakkunni
> wrote:
> > Hi Yuanhan,
> >
> > Thanks for the confirmation about not having to do anything special to
> close
> > the ports on dpdk going down or coming up.
> >
> > As for the question about if I met any issue of ovs getting stuck - yes,
> my
> > guest process runs dpdk 16.07 as I mentioned earlier - and if I kill my
> guest
> > process, then the host OVS-dpdk on the host reports stall ! The OVS-dpdk
> and
> > emu versions I use are as below. But maybe that is because of the ovs
> missing
> > the fixes you mentioned ?
>
> When I was saying dpdk version, I meant the DPDK version with OVS.
>
> > ~# ovs-vswitchd --version
> > ovs-vswitchd (Open vSwitch) 2.4.1
>
> And yes, the fixes are not included in the DPDK required for OVS 2.4.
>
> --yliu
>
> > Compiled Nov 14 2016 06:53:31
> > # kvm --version
> > QEMU emulator version 2.2.0, Copyright (c) 2003-2008 Fabrice Bellard
> > ~#
> >
> >
> > Rgds,
> > Gopa.
> >
> > On Thu, Mar 16, 2017 at 9:35 PM, Yuanhan Liu <yuanhan.liu@linux.intel.com
> >
> > wrote:
> >
> > On Thu, Mar 16, 2017 at 07:48:28PM -0700, Gopakumar Choorakkot
> Edakkunni
> > wrote:
> > > Thanks a lot for the response Yuanhan. I am using dpdk v16.07. So
> what
> > you are
> > > saying is that in 16.07, we dont really need to call
> rte_eth_dev_close()
> > on
> > > exit,
> >
> > It's not about "don't really need", it's more like "it's hard to".
> Just
> > think that it may crash at any time.
> >
> > > because dpdk will ensure that it will do virtio reset before init
> when it
> > > comes up right ?
> >
> > No, It just handles the abnormal case well when guest APP restarts.
> >
> > > Regarding the vhost commits you mentioned - do we still need those
> fixes
> > if we
> > > have the "virtio reset before init" mechanism ?
> >
> > Yes, we still need them: just think some malicious guest may also
> forge
> > data like that.
> >
> > I'm a bit confused then. Have you actually met any issue (like got
> stucked)
> > with DPDK v16.07?
> >
> > --yliu
> >
> > > Or that is a seperate problem
> > > altogether (and hence we would need those fixes) ?
> > >
> > > Rgds,
> > > Gopa.
> > >
> > > On Thu, Mar 16, 2017 at 7:06 PM, Yuanhan Liu <
> yuanhan.liu@linux.intel.com
> > >
> > > wrote:
> > >
> > > On Thu, Mar 16, 2017 at 12:39:16PM -0700, Gopakumar Choorakkot
> > Edakkunni
> > > wrote:
> > > > So the doc says we should call rte_eth_dev_close() *before*
> going
> > down.
> > > And I
> > > > know that especially in dpdk-virtionet in the guest +
> ovs-dpdk in
> > the
> > > host,
> > > > the ovs ends up getting stalled/stuck (!!) if I dont close
> the port
> > > before
> > > > starting() it when the guest dpdk process comes back up.
> > >
> > > I'm assuming you were using an old version, something like dpdk
> v2.2?
> > > IIRC, DPDK v16.04 should have fixed your issue.
> > >
> > > > Considering that this not done properly can screw up the HOST
> ovs,
> > and I
> > > want
> > > > to do everything possible to avoid that, I want to be 200%
> sure
> > that I
> > > call
> > > > close even if my process gets a kill -9 .. So obviously the
> only
> > way of
> > > doing
> > > > that is to close the port when the dpdk process comes back up
> and
> > > *before* we
> > > > init the port. rte_eth_dev_close() is not capable of doing
> that as
> > it
> > > expects
> > > > the port parameters to be initialized etc.. before it can be
> > called.
> > >
> > > We do virtio reset before init, which is basically what
> > rte_eth_dev_close()
> > > mainly does. So I see no big issue here.
> > >
> > > The stuck issue is due to hugepage reset by the guest DPDK
> > application,
> > > leading all virtio vring elements being mem zeroed. The old
> vhost
> > doesn't
> > > handle it well, as a result, it got stuck. And here are some
> relevant
> > > commits:
> > >
> > > a436f53 vhost: avoid dead loop chain
> > > c687b0b vhost: check for ring descriptors overflow
> > > 623bc47 vhost: do sanity check for ring descriptor length
> > >
> > > --yliu
> > >
> > > > Any other
> > > > suggestions on what can be done to close on restart rather
> than
> > close on
> > > going
> > > > down ? Thought of bouncing this by the alias before I add a
> version
> > of
> > > close
> > > > myself that can do this close-on-restart
> > >
> > >
> >
> >
>
>
next prev parent reply other threads:[~2017-03-17 5:26 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-03-16 19:39 Gopakumar Choorakkot Edakkunni
2017-03-17 2:06 ` Yuanhan Liu
2017-03-17 2:48 ` Gopakumar Choorakkot Edakkunni
2017-03-17 4:35 ` Yuanhan Liu
2017-03-17 4:56 ` Gopakumar Choorakkot Edakkunni
2017-03-17 5:13 ` Yuanhan Liu
2017-03-17 5:20 ` Gopakumar Choorakkot Edakkunni
2017-03-17 5:24 ` Yuanhan Liu [this message]
2017-03-17 5:30 ` Gopakumar Choorakkot Edakkunni
2017-03-17 5:40 ` Yuanhan Liu
2017-03-17 5:50 ` Gopakumar Choorakkot Edakkunni
2017-03-18 21:32 ` Gopakumar Choorakkot Edakkunni
2017-03-18 21:37 ` Gopakumar Choorakkot Edakkunni
2017-03-18 23:43 ` Gopakumar Choorakkot Edakkunni
2017-03-22 5:32 ` Gopakumar Choorakkot Edakkunni
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20170317052433.GZ18844@yliu-dev.sh.intel.com \
--to=yuanhan.liu@linux.intel.com \
--cc=dev@dpdk.org \
--cc=gopakumar.c.e@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).