The current situation is that whenever any packet enters the
userspace, bfd_should_process_flow() looks at the UDP destination
port to figure out whether that is a BFD packet. This means that
UDP destination port cannot be wildcarded for all the other flows
too.
To optimize BFD for megaflows, we i
I don't know what to say, other than what I already said. Here it is
again. I imagine that you don't have a network device named
port1. Try adding one that you do (eth0?).
On Sun, Aug 04, 2013 at 06:48:01AM +0530, f 62 wrote:
> Ben,
>
> Here is the procedure I followed.
>
> 1. Created brid
On Sun, Aug 04, 2013 at 08:14:31PM -0700, Alex Wang wrote:
> This commit adds a new commmand "ovs-ofctl parse-ofpraw" to ovs.
> It allows user to parse undecoded raw OpenFlow messages into human
> readable form. It can parse the output log of "ovs-appctl log/controller"
> command.
>
> Signed-off-b
Thanks Ben for point that out,
I'd very much like to review it.
On Mon, Aug 5, 2013 at 10:37 AM, Ben Pfaff wrote:
> On Sun, Aug 04, 2013 at 08:14:31PM -0700, Alex Wang wrote:
> > This commit adds a new commmand "ovs-ofctl parse-ofpraw" to ovs.
> > It allows user to parse undecoded raw OpenFlo
After some consultation with Andy, this is the commit message I ended
up with. Jesse, Andy, is this correct?
--8<--cut here-->8--
From: Andy Zhou
Date: Sat, 3 Aug 2013 12:23:15 -0700
Subject: [PATCH] odp-util: Always export the priority and skb_ma
On Sat, Aug 03, 2013 at 06:42:03PM -0700, Ethan Jackson wrote:
> This allows us to move some minor functionality from ofproto-dpif to
> ofproto-dpif-xlate, where it's easier to ensure it's thread safe.
>
> Signed-off-by: Ethan Jackson
Acked-by: Ben Pfaff
On Mon, Aug 5, 2013 at 11:06 AM, Ben Pfaff wrote:
> After some consultation with Andy, this is the commit message I ended
> up with. Jesse, Andy, is this correct?
>
> --8<--cut here-->8--
>
> From: Andy Zhou
> Date: Sat, 3 Aug 2013 12:23:15 -0700
>
On Sat, Aug 03, 2013 at 06:42:04PM -0700, Ethan Jackson wrote:
> Signed-off-by: Ethan Jackson
In rule_destruct(), I think it would be better to destroy the mutex
following complete_operation(). Who knows, that function might want
to eventually look at the statistics.
Acked-by: Ben Pfaff
__
Looks good to me, tested with the ofp log snooped from controller.
On Tue, Jul 23, 2013 at 11:13 AM, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff ---
> NEWS |2 +
> utilities/ovs-ofctl.8.in |8 ++
> utilities/ovs-ofctl.c| 55
>
On Sat, Aug 03, 2013 at 06:42:05PM -0700, Ethan Jackson wrote:
> It's simple to hide it than to make it thread safe and ensure it stays
> that way in the long term.
>
> Signed-off-by: Ethan Jackson
I don't like how this duplicates logic in two places. Could we have
this code in just one place p
This looks good to me. My only comment is that I'd prefer we store
BFD_ETH_DST as an array of uint8_t in packets.h like we do for
eth_addr_stp and eth_addr_lacp.
Acked-by: Ethan Jackson
On Mon, Aug 5, 2013 at 12:47 AM, Gurucharan Shetty wrote:
> The current situation is that whenever any pack
On Mon, Aug 05, 2013 at 11:22:49AM -0700, Jesse Gross wrote:
> This information is factually correct but it's not really related to
> the change in this commit.
>
> The problem here is very simple: the current protocol allows a default
> value of zero if either mark or priority is not specified (t
On Sat, Aug 03, 2013 at 06:42:06PM -0700, Ethan Jackson wrote:
> Signed-off-by: Ethan Jackson
I wonder whether a rwlock is warranted here.
Acked-by: Ben Pfaff
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Sat, Aug 03, 2013 at 06:42:07PM -0700, Ethan Jackson wrote:
> This patch allows the lookup functions to take NULL as an argument as
> a convenience.
>
> Signed-off-by: Ethan Jackson
Acked-by: Ben Pfaff
___
dev mailing list
dev@openvswitch.org
http:
Good point, I made the change.
Ethan
On Mon, Aug 5, 2013 at 11:29 AM, Ben Pfaff wrote:
> On Sat, Aug 03, 2013 at 06:42:04PM -0700, Ethan Jackson wrote:
>> Signed-off-by: Ethan Jackson
>
> In rule_destruct(), I think it would be better to destroy the mutex
> following complete_operation(). Who
> I wonder whether a rwlock is warranted here.
Probably. Ill leave that for when we go back and optimize.
Ethan
X-CudaMail-Whitelist-To: dev@openvswitch.org
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
Ben's original commit message gives more background to the netlink protocol
changes we have been making. It would be nice to keep them as well.
On Mon, Aug 5, 2013 at 12:34 PM, Ben Pfaff wrote:
> On Mon, Aug 05, 2013 at 11:22:49AM -0700, Jesse Gross wrote:
> > This information is factually corr
On Mon, Aug 5, 2013 at 12:34 PM, Ben Pfaff wrote:
> On Mon, Aug 05, 2013 at 11:22:49AM -0700, Jesse Gross wrote:
>> This information is factually correct but it's not really related to
>> the change in this commit.
>>
>> The problem here is very simple: the current protocol allows a default
>> val
Those problems are different and not related to this change so it is
confusing to have it here.
On Mon, Aug 5, 2013 at 12:42 PM, Andy Zhou wrote:
> Ben's original commit message gives more background to the netlink protocol
> changes we have been making. It would be nice to keep them as well.
>
>
Signed-off-by: Pavithra Ramesh
Thanks,
Pavithra
- Original Message -
From: "Ethan Jackson"
To: "Pavithra Ramesh"
Cc: dev@openvswitch.org
Sent: Sunday, August 4, 2013 11:10:04 AM
Subject: Re: [ovs-dev] [PATCH] BFD: Populate ToS field in BFD packets.
This looks good, can I have your sig
I just saw this. Sorry about the delay!
Signed-off-by: Pavithra Ramesh
Thanks,
Pavithra
- Original Message -
From: "Ethan Jackson"
To: "Pavithra Ramesh"
Cc: dev@openvswitch.org
Sent: Thursday, August 1, 2013 12:41:00 PM
Subject: Re: [ovs-dev] [PATCH] BFD: Edit the unit test time/stop
On Sat, Aug 03, 2013 at 06:42:08PM -0700, Ethan Jackson wrote:
> ofproto_receive() is a slightly odd function which doesn't fit
> perfectly in either ofproto-dpif or ofproto-dpif-xlate. However, it's
> much easier to reason about its thread safety in ofproto-dpif-xlate,
> so this patch moves it th
looks good.
Acked-by: Andy Zhou
On Thu, Aug 1, 2013 at 2:29 PM, Ben Pfaff wrote:
> This API change is necessary for thread safety, to be added in an upcoming
> commit. Otherwise, the client would not be able to safely use the returned
> netdev because it could already have been destroyed.
>
Thanks, I've merged those patches which didn't have comments.
Ethan
On Mon, Aug 5, 2013 at 12:54 PM, Ben Pfaff wrote:
> On Sat, Aug 03, 2013 at 06:42:08PM -0700, Ethan Jackson wrote:
>> ofproto_receive() is a slightly odd function which doesn't fit
>> perfectly in either ofproto-dpif or ofproto-
On Thu, Aug 1, 2013 at 2:29 PM, Ben Pfaff wrote:
> This API change is necessary for thread safety, to be added in an upcoming
> commit. Otherwise, the client would not be able to actually use any of
> the returned netdevs because they could already have been destroyed.
>
> Signed-off-by: Ben Pfa
No problem. Thanks for writing these patches up. I've merged them
Ethan
On Mon, Aug 5, 2013 at 12:48 PM, Pavithra Ramesh wrote:
> Signed-off-by: Pavithra Ramesh
>
> Thanks,
> Pavithra
>
> - Original Message -
> From: "Ethan Jackson"
> To: "Pavithra Ramesh"
> Cc: dev@openvswitch.org
Acked-by: Andy Zhou
On Thu, Aug 1, 2013 at 2:29 PM, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
> ---
> lib/netdev-provider.h |6 --
> 1 file changed, 6 deletions(-)
>
> diff --git a/lib/netdev-provider.h b/lib/netdev-provider.h
> index 137b659..19230a1 100644
> --- a/lib/netdev-pro
On Mon, Aug 5, 2013 at 12:25 PM, Ethan Jackson wrote:
> This looks good to me. My only comment is that I'd prefer we store
> BFD_ETH_DST as an array of uint8_t in packets.h like we do for
> eth_addr_stp and eth_addr_lacp.
>
> Okay. I made the change and applied it to master and 1.12
> Acked-by
On Sat, Aug 03, 2013 at 06:42:09PM -0700, Ethan Jackson wrote:
> This patch makes packet_ins thread safe by handing responsibility for
> them to ofproto-dpif.
>
> Signed-off-by: Ethan Jackson
Acked-by: Ben Pfaff
We have a couple of instances of code like this:
if (ofproto->n_pins) {
Acked-by: Andy Zhou
On Thu, Aug 1, 2013 at 2:29 PM, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
> ---
> lib/netdev-linux.c |4 +---
> 1 file changed, 1 insertion(+), 3 deletions(-)
>
> diff --git a/lib/netdev-linux.c b/lib/netdev-linux.c
> index 29daef8..ba0d863 100644
> --- a/lib/netdev
Thanks Paul. As part of this patch, I moved your name to the list of
people who've contributed code in the AUTHORS file.
Acked-by: Ethan Jackson
On Fri, Aug 2, 2013 at 4:12 PM, Paul Ingram wrote:
> The remote opstate for a CFM interface is presumed to be up unless a CCM is
> received which si
When there is no incoming data traffic at the interface for a period,
BFD decay allows the bfd session to increase the min_rx. This is
helpful in that some interfaces usually idle for long time. And cpu
consumption can be reduced by processing fewer bfd control packets.
Signed-off-by: Alex Wang
-
This commit adds unit tests for BFD decay feature.
Signed-off-by: Alex Wang
---
tests/bfd.at | 189 +++---
1 file changed, 180 insertions(+), 9 deletions(-)
diff --git a/tests/bfd.at b/tests/bfd.at
index c54fff0..c1f78d2 100644
--- a/tests/bf
This commit truncates the trailing whitespace in the 'flag' field
out 'bfd/show' output. This is for the string matching in bfd
unit test.
Signed-off-by: Alex Wang
---
lib/bfd.c |3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/lib/bfd.c b/lib/bfd.c
index 0a0089b..0af0eeb
This commit adds a new boolean option "forwarding_if_rx" to bfd.
When forwarding_if_rx is true and BFD state is down, the forwarding
field in "ovs-appctl bfd/show" output will still be true as long as
there are incoming packets received. This is for indicating the
link liveness when the link is co
On Mon, Aug 05, 2013 at 12:46:43PM -0700, Jesse Gross wrote:
> On Mon, Aug 5, 2013 at 12:34 PM, Ben Pfaff wrote:
> > On Mon, Aug 05, 2013 at 11:22:49AM -0700, Jesse Gross wrote:
> >> This information is factually correct but it's not really related to
> >> the change in this commit.
> >>
> >> The
Also, here is a test, if it is needed, ;D
+
+AT_SETUP([ovs-ofctl ofp-parse])
+
+# Test the echo request/reply messages (0 payload).
+AT_CHECK([echo -n -e
\\x1\\x2\\x0\\x8\\x0\\x0\\x0\\x0\\x1\\x3\\x0\\x8\\x0\\x0\\x0\\x0 >
binary_ofp_msg])
+
+AT_CHECK([ovs-ofctl ofp-parse binary_ofp_msg], [0], [dnl
On Mon, Aug 5, 2013 at 2:04 PM, Alex Wang wrote:
> When there is no incoming data traffic at the interface for a period,
> BFD decay allows the bfd session to increase the min_rx. This is
> helpful in that some interfaces usually idle for long time. And cpu
> consumption can be reduced by processi
Let's just fold this into the bfd decay patch? I haven't read it yet.
Ethan
On Mon, Aug 5, 2013 at 2:04 PM, Alex Wang wrote:
> This commit adds unit tests for BFD decay feature.
>
> Signed-off-by: Alex Wang
> ---
> tests/bfd.at | 189
> +++
Sure, I'll do that.
On Mon, Aug 5, 2013 at 3:52 PM, Ethan Jackson wrote:
> Let's just fold this into the bfd decay patch? I haven't read it yet.
>
> Ethan
>
> On Mon, Aug 5, 2013 at 2:04 PM, Alex Wang wrote:
> > This commit adds unit tests for BFD decay feature.
> >
> > Signed-off-by: Alex Wa
Hi Ben,
I've had a "mostly working" implementation of groups for several months,
but sadly not enough time in my schedule for a proper upstreaming effort. I
don't expect to have time for at least a few more weeks.
Would it help if I sent my ofproto and action bucket changes? Those were
the most i
Thanks Ethan for the review.
I'll add 'in_decay' flag and set bfd->min_rx to decay_min_rx, when
receiving the FINAL packet.
On Mon, Aug 5, 2013 at 3:35 PM, Ethan Jackson wrote:
> On Mon, Aug 5, 2013 at 2:04 PM, Alex Wang wrote:
> > When there is no incoming data traffic at the interface for a
Thanks for writing this up, it's pretty close.
In bfd_check_rx() the assertion makes me uncomfortable. I'd prefer we
warn and reset bfd->rx_packets instead. Im worried that we could hit
the assertion transiently during a configuration change.
I'd prefer we don't use the bfd->detect_time for thi
When tracing a flow, it shows the "relevant fields" that were used
to determine the results. However, this currently only includes fields
that are used for computing the actions but not the flow lookup. This
can be confusing so this patch includes the wildcards from the classifer
lookup as well.
S
> I'll add 'in_decay' flag and set bfd->min_rx to decay_min_rx, when receiving
> the FINAL packet.
It should be sufficient to set bfd->poll_min_rx to the appropriate
value in the bfd_poll() function. The poll_min_rx is automatically
copied to the min_rx field when the FINAL packet comes in.
Etha
Open vSwitch Virtual Hackathon, Sep. 6-7, 2013
==
What?
-
Our goal is to spend two days working hard to complete and polish Open
vSwitch support for OpenFlow 1.1 and later. Open vSwitch already has
significant support for these later protocols, but
Hi Ben,
I realise that you've agreed to let Jesse review the datapath patch first,
but I wondered if I could get a little broad feedback on the approach on
the userspace side.
The general background to this patch is that we aim to keep the datapath
interface for MPLS actions simple. As such, when
Hello,
We added OVS into our continuous integration loop.
https://ci.linaro.org/jenkins/view/lng-ci/job/openvswitch-arm-native-build/
Currently, we pull from master branch and natively build & run the
testsuite on an ARM Cortex-A15 system. All 1280 tests were successful (SSL
tests were skipped).
Hi Casey,
I'll let Ben answer for himself but from my point of view I think
it would be helpful for you to send what you have and I would be more
than happy to look at helping move things forwards from where
you have got up to.
On Mon, Aug 05, 2013 at 03:55:20PM -0700, Casey Barker wrote:
> Hi Be
49 matches
Mail list logo
