These were useful in isolating a number of bugs.
Signed-off-by: Simon Horman
---
The "ofproto - flow table configuration (OpenFlow 1.2)]" test
will need to be updated for the MPLS changes that are outstanding
as they change the following:
wild=0xf -> wild=0x1f
match=0xf
Allow only OpenFlow 12 as a flow dump protocol for OpenFlow12.
Allow OpenFlow10 and NXM as flow dump protocols for other OpenFlow versions.
This only changes the behaviour prior to this patch in the case of OpenFlow12
which seemed broken as an attempt would be made to set NXM as the
flow dump forma
Open Flow and 1.1 and 1.2 support some of the same
flow format capabilities as NXM.
Signed-off-by: Simon Horman
---
lib/ofp-util.c | 30 +++---
lib/ofp-util.h |3 +++
tests/ovs-ofctl.at |2 +-
3 files changed, 19 insertions(+), 16 deletions(-)
diff --gi
--allowed-ofp-versions allows configuration of the versions
that may be used when establishing an OpenFlow connection.
The default is 'OpenFlow10' which is consistent with
the behaviour prior to this patch.
The useful values at this time are:
'OpenFlow10', 'OpenFlow12' and 'OpenFlow10,OpenFlow12'
--allowed-ofp-versions allows configuration of the versions
that may be used when establishing an OpenFlow connection.
The default is 'OpenFlow10' which is consistent with
the behaviour prior to this patch.
The useful values at this time are:
'OpenFlow10', 'OpenFlow12' and 'OpenFlow10,OpenFlow12'
Only set the default format for ovs-ofctl monitor if
OpenFlow 1.0 is the prevailing version. IMHO that is
the only case where it makes sense.
Signed-off-by: Simon Horman
---
utilities/ovs-ofctl.c | 36 +++-
1 file changed, 23 insertions(+), 13 deletions(-)
diff
Signed-off-by: Simon Horman
---
lib/automake.mk |3 +++
lib/ofp-version-opt.c | 43 +++
lib/ofp-version-opt.h | 28
lib/ofp-version.man | 26 ++
4 files changed, 100 insertions(+)
creat
Signed-off-by: Simon Horman
---
lib/ofp-util.c| 14 --
lib/ofp-util.h|2 ++
ofproto/connmgr.c |9 -
3 files changed, 22 insertions(+), 3 deletions(-)
diff --git a/lib/ofp-util.c b/lib/ofp-util.c
index 2ec6d9d..b6cb4a8 100644
--- a/lib/ofp-util.c
+++ b/lib/of
Paramatise the allowed OpenFlow verion for snoops.
This is in preparation for making the allow OpenFlow verions to
be configured.
Signed-off-by: Simon Horman
---
ofproto/connmgr.c | 20 +---
ofproto/connmgr.h |2 +-
ofproto/ofproto.c |2 +-
ofproto/ofproto.h |8
Allow allowed Open Flow versions to be passed rather
than relying on hard-coded defaults.
This is in preparation for allowing configuration of the
allowed OpenFlow versions.
Signed-off-by: Simon Horman
---
ofproto/connmgr.c | 34 +-
ofproto/connmgr.h |3 ++-
Versions may be configured using a comma delimited list as
the value for 'openflow-versions' the 'other-config' column
of the Bridge table.
Currently list elements other than 'OpenFlow10' and 'OpenFlow12' will
be ignored.
If the list is empty, then OpenFlow10 is used.
This default is consistent w
This enables the use of the OpenFlow 12 flow format.
Signed-off-by: Simon Horman
---
lib/ofp-util.c |2 +-
lib/ofp-util.h |4 ++--
tests/learn.at |2 +-
3 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/lib/ofp-util.c b/lib/ofp-util.c
index f7111ce..2ec6d9d 100644
--- a/
Add allowed OpenFlow versions to struct rconn to allow
reconnect to use these parameters rather than hard-coded defaults.
This is in preparation for allowing configuration of the
allowed OpenFlow versions.
Signed-off-by: Simon Horman
---
lib/rconn.c| 15 ++-
lib/rc
This is in preparation for allowing configuration of the
allowed OpenFlow versions.
Signed-off-by: Simon Horman
---
ofproto/connmgr.c | 19 ---
1 file changed, 12 insertions(+), 7 deletions(-)
diff --git a/ofproto/connmgr.c b/ofproto/connmgr.c
index d63a166..17d9fb5 100644
---
Replace minimum version with bitmap of allowed versions.
This is in preparation for allowing the range of allowed OpenFlow
versions to be configured.
Signed-off-by: Simon Horman
---
lib/rconn.c|5 ++-
lib/vconn-provider.h | 26 ++-
lib/vconn-stream.c
The motivation for this is to avoid avoids needing to provide a minimu
version parameter to pvconn_accept(). This will in turn avoid the need to
pass version information to connmgr_run() when the range of accetable
OpenFlow versions is made configurable.
Signed-off-by: Simon Horman
---
lib/vcon
Allow encoding and decoding of version bitmap in hello messages
as specified in Open Flow 1.3.1.
Signed-off-by: Simon Horman
---
include/openflow/openflow-common.h | 11
lib/ofp-print.c| 30 -
lib/ofp-util.c | 119
Data structure to store allowed versions as a bitmap.
Bit offsets correspond to ofp_version numbers which in turn
correspond to wire-protocol numbers for Open Flow versions..
E.g. (1 << OFP10_VERSION) is the mask for Open Flow 1.1.
If the bit for a version is set then it is allowed, otherwise it i
Hi,
This series adds run-time configuration of allowed OpenFlow versions
to ovs-vswtichd, ovs-ofctl and ovs-controller; and adds tests for Open Flow
1.2 messages that utilise the run-time configuration of allowed OpenFlow
versions.
Substantial portions have been re-written since my previous post.
On Wed, Oct 17, 2012 at 06:03:10PM -0700, Pravin Shelar wrote:
> On Wed, Oct 17, 2012 at 4:12 PM, Ben Pfaff wrote:
> > In tnl_set_config(), with port_table_lookup() moved to the end, do we
> > need to back out the ip_mc_inc_group() that we've done just above in
> > the multicast case?
> >
> I am n
Sent from my iPad
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Wed, Oct 17, 2012 at 4:12 PM, Ben Pfaff wrote:
> On Sun, Oct 14, 2012 at 06:41:47PM -0700, Pravin B Shelar wrote:
>> Following patch allow port create without any parameters. This is
>> required for compatibility with future flow based tunneling
>> where each type would have single vport with a
On Wed, Oct 17, 2012 at 09:32:40PM +1300, Joe Stringer wrote:
> On 17 October 2012 15:24, Simon Horman wrote:
> > Hi Ben,
> >
> > sorry for not replying earlier. I have done some work in this area
> > but I am not comfortable with publishing it as the Open Flow 1.3.1 spec
> > has not been publishe
With a previous commit 953cc1888321346e6c4175fbc0a3f66d34913ea3, the
ovs logs were re-formatted such that they can be input into the
ofproto/trace. For ARP packets, ofproto/trace expects arp_tpa/arp_spa
instead of nw_dst/nw_src.
Signed-off-by: Mehak Mahajan
---
lib/match.c|3 +++
te
The packet's nw_proto field overlaps with the ARP opcode field. The
nw_proto for ICMP packets is 1. Hence when the packet is an ARP
request (opcode 1), we erroneously print the ICMP code and ICMP type
for ARP frames as well. With this commit, the ICMP code and ICMP
type will be printed only if t
flow_format() logs packets contents. However, the format used is not
the format accepted by ofproto/trace. Hence it becomes difficult to
trace the packets using the debugs printed. With this commit, the
logging of the packet contents is done in a format that is accepted
by ofproto/trace. This w
On Wed, Oct 17, 2012 at 04:09:52PM -0700, Pravin Shelar wrote:
> On Wed, Oct 17, 2012 at 3:52 PM, Ben Pfaff wrote:
> > On Sun, Oct 14, 2012 at 06:41:25PM -0700, Pravin B Shelar wrote:
> >> Following patch adds start offset for sw_flow-key, so that we can
> >> skip tunneling information in key for
On Sun, Oct 14, 2012 at 06:41:47PM -0700, Pravin B Shelar wrote:
> Following patch allow port create without any parameters. This is
> required for compatibility with future flow based tunneling
> where each type would have single vport with all parameter
> wild carded. So that all packet for all p
On Wed, Oct 17, 2012 at 3:59 PM, Ben Pfaff wrote:
> On Wed, Oct 17, 2012 at 03:54:54PM -0700, Pravin Shelar wrote:
>> On Wed, Oct 17, 2012 at 3:15 PM, Ben Pfaff wrote:
>> > On Sun, Oct 14, 2012 at 06:41:06PM -0700, Pravin B Shelar wrote:
>> > In ovs_tnl_send(), can this case happen? I'm a little
On Wed, Oct 17, 2012 at 3:52 PM, Ben Pfaff wrote:
> On Sun, Oct 14, 2012 at 06:41:25PM -0700, Pravin B Shelar wrote:
>> Following patch adds start offset for sw_flow-key, so that we can
>> skip tunneling information in key for non-tunnel flows.
>>
>> Signed-off-by: Pravin B Shelar
>> CC: Kyle Mes
The following OF packets may produce the wrong result as follows.
It depends on how ovs-vswitchd serves OF packets. Sending the OF packets
in single TCP packet will increase the possiblity.
OF request:
feature request
port_mod with config LINK DOWN
barrier
feature request
The replies:
f
On Wed, Oct 17, 2012 at 03:54:54PM -0700, Pravin Shelar wrote:
> On Wed, Oct 17, 2012 at 3:15 PM, Ben Pfaff wrote:
> > On Sun, Oct 14, 2012 at 06:41:06PM -0700, Pravin B Shelar wrote:
> > In ovs_tnl_send(), can this case happen? I'm a little surprised:
> > if (!daddr) {
> >
On Wed, Oct 17, 2012 at 3:15 PM, Ben Pfaff wrote:
> On Sun, Oct 14, 2012 at 06:41:06PM -0700, Pravin B Shelar wrote:
>> This patch was posted by Kyle. I fixed few issues found in earlier
>> version.
>
> I'm not sure that the tun_key handling is correct in the sampling
> case. ovs_execute_actions(
On Sun, Oct 14, 2012 at 06:41:25PM -0700, Pravin B Shelar wrote:
> Following patch adds start offset for sw_flow-key, so that we can
> skip tunneling information in key for non-tunnel flows.
>
> Signed-off-by: Pravin B Shelar
> CC: Kyle Mestery
> Cc: Simon Horman
> Cc: Jesse Gross
Does this a
On Sun, Oct 14, 2012 at 06:41:06PM -0700, Pravin B Shelar wrote:
> This patch was posted by Kyle. I fixed few issues found in earlier
> version.
I'm not sure that the tun_key handling is correct in the sampling
case. ovs_execute_actions() sets OVS_CB(skb)->tun_key to NULL and
passes &tun_key (a l
On Oct 14, 2012, at 8:41 PM, Pravin B Shelar wrote:
> Following patch allow port create without any parameters. This is
> required for compatibility with future flow based tunneling
> where each type would have single vport with all parameter
> wild carded. So that all packet for all ports are rec
On Oct 14, 2012, at 8:41 PM, Pravin B Shelar wrote:
> Following patch adds start offset for sw_flow-key, so that we can
> skip tunneling information in key for non-tunnel flows.
>
> Signed-off-by: Pravin B Shelar
> CC: Kyle Mestery
> Cc: Simon Horman
> Cc: Jesse Gross
I like this approach.
action keyword dec_ttl is already used for NX dec_ttl. So dec_nw_ttl is
introduced to disambiguate them.
Signed-off-by: Isaku Yamahata
---
lib/ofp-actions.c| 25 +
lib/ofp-actions.h|4 ++--
lib/ofp-parse.c | 12 +---
lib/ofp-util.d
On Oct 17, 2012, at 1:53 PM, Pravin Shelar wrote:
> On Wed, Oct 17, 2012 at 7:00 AM, Kyle Mestery (kmestery)
> wrote:
>> On Oct 14, 2012, at 8:41 PM, Pravin B Shelar wrote:
>>> This patch was posted by Kyle. I fixed few issues found in earlier
>>> version.
>>>
>>> V7:
>>> - Fix according to com
Sorry it took me so long to get back to this.
> The reason to additionally check if the format is IP or not is because else
> we will print the icmp code and icmp type for non icmp packets which can be
> quite misleading. For example for the ARP request packets. The arp_op field
> in an ARP packet
When I wrote this function I didn't think that port 0 was important (it's
not a valid OpenFlow port number) so I used a return value of 0 to indicate
an error. However, my assumption turns out to be wrong, so this commit
changes the interface to use the return value only for error reporting
and st
Commit c6100d92142b (ovs-ofctl: Accept port keywords, OF1.1 port numbers,
reject port number 0.) intentionally prohibited using port 0, which is not
a valid OpenFlow port number, in OpenFlow commands and actions. However,
since then we have discovered that the NVP controller uses port 0 in
"resubm
On Wed, Oct 17, 2012 at 01:21:45PM -0700, Ethan Jackson wrote:
> Actually, ovs_vswitchd is a silly name for that variable. I've actually
> decided to fold this in instead.
That's even better, thanks.
___
dev mailing list
dev@openvswitch.org
http://openv
Thanks.
On Wed, Oct 17, 2012 at 01:18:45PM -0700, Ethan Jackson wrote:
>
> Good idea, I had been bitten by that once myself. I"ve folded in
> the following incremental. I'll merge shortly. Thanks for the
> review.
>
> Ethan
>
> ---
> utilities/ovs-parse-backtrace.in | 2 ++
> 1 file changed
Actually, ovs_vswitchd is a silly name for that variable. I've actually
decided to fold this in instead.
Ethan
---
utilities/ovs-parse-backtrace.in | 16 +---
1 file changed, 9 insertions(+), 7 deletions(-)
diff --git a/utilities/ovs-parse-backtrace.in b/utilities/ovs-parse-backtra
Good idea, I had been bitten by that once myself. I"ve folded in the following
incremental. I'll merge shortly. Thanks for the review.
Ethan
---
utilities/ovs-parse-backtrace.in | 2 ++
1 file changed, 2 insertions(+)
diff --git a/utilities/ovs-parse-backtrace.in b/utilities/ovs-parse-backt
> [adding back ovs-dev]
Oops, sorry about that.
> A user would never want to change it at runtime. Developers might
> want it for testing, that's all, without having to do a rebuild.
True.
> Oh well, it's a corner case, we can always change it later if it
> proves useful.
Agreed.
I'll merge
[adding back ovs-dev]
On Wed, Oct 17, 2012 at 12:58:14PM -0700, Ethan Jackson wrote:
> > This looks good, but I wonder whether it would be better to make this a
> > command-line option or a runtime ovs-appctl feature, so that one doesn't
> > need to reconfigure and rebuild everything to test chang
On Wed, Oct 17, 2012 at 7:00 AM, Kyle Mestery (kmestery)
wrote:
> On Oct 14, 2012, at 8:41 PM, Pravin B Shelar wrote:
>> This patch was posted by Kyle. I fixed few issues found in earlier
>> version.
>>
>> V7:
>> - Fix according to comments posted on V6.
>> V6:
>> - Fix more comments addressed fr
This patch adds support of OF11+ pop_vlan.
Signed-off-by: Isaku Yamahata
---
lib/ofp-actions.c |6 +-
lib/ofp-actions.h |2 +-
lib/ofp-parse.c |1 +
lib/ofp-util.def |2 +-
4 files changed, 8 insertions(+), 3 deletions(-)
diff --git a/lib/ofp-actions.c b/lib/ofp-actions.c
On Thu, Oct 18, 2012 at 03:18:16AM +0900, Isaku Yamahata wrote:
> On Thu, Oct 18, 2012 at 03:09:02AM +0900, Isaku Yamahata wrote:
> > Hi. The patch looks reasonable.
> >
> > If it isn't used as decorator, returning class is not necessary.
> > I'll send a patch to fix it.
>
> Here's the incrementa
On Thu, Oct 18, 2012 at 03:09:02AM +0900, Isaku Yamahata wrote:
> Hi. The patch looks reasonable.
>
> If it isn't used as decorator, returning class is not necessary.
> I'll send a patch to fix it.
Here's the incremental patch to c38f8724aeb994fd840fa4283a07e38c3c114d68
>From 02683397771860a32d5
Hi. The patch looks reasonable.
If it isn't used as decorator, returning class is not necessary.
I'll send a patch to fix it.
On Wed, Oct 17, 2012 at 10:04:49AM -0700, Justin Pettit wrote:
> I went ahead and pushed the change, since it was blocking our internal
> testing. Please let me know if
I went ahead and pushed the change, since it was blocking our internal testing.
Please let me know if you object to the patch, and we can work on a revised
version.
--Justin
On Oct 17, 2012, at 1:28 AM, Justin Pettit wrote:
> Hi, Isaku. Can you confirm that this change looks reasonable to
On Tue, Oct 16, 2012 at 06:23:37PM -0700, Ethan Jackson wrote:
> Often when developing, it would be convenient to force OVS to
> either enable or disable cached timing so that a feature can be
> tested under both strategies.
>
> Signed-off-by: Ethan Jackson
This looks good, but I wonder whether
On Tue, Oct 16, 2012 at 05:29:08PM -0700, Justin Pettit wrote:
> Commit 8cc820 (python/ovs/stream: teach stream.py tcp socket) made a
> change that used class decorators. Unfortunately, they were not
> introduced until Python 2.6. XenServer uses Python 2.4, so the change
> caused some Python-base
On Tue, Oct 16, 2012 at 08:02:44PM -0700, Ethan Jackson wrote:
> The new ovs-parse-backtrace utility makes the output of ovs-appctl
> backtrace more human readable by removing duplicate traces and
> converting addresses to function names.
>
> Signed-off-by: Ethan Jackson
Looks good.
My only sug
On Wed, Oct 17, 2012 at 09:32:40PM +1300, Joe Stringer wrote:
> On 17 October 2012 15:24, Simon Horman wrote:
> > Hi Ben,
> >
> > sorry for not replying earlier. I have done some work in this area
> > but I am not comfortable with publishing it as the Open Flow 1.3.1 spec
> > has not been publishe
On Oct 14, 2012, at 8:41 PM, Pravin B Shelar wrote:
> This patch was posted by Kyle. I fixed few issues found in earlier
> version.
>
> V7:
> - Fix according to comments posted on V6.
> V6:
> - Fix more comments addressed from Jesse.
> V5:
> - Address another round of comments from Jesse.
> V4:
>
Hi all
I have a problem as following:
I setup KVM (libvirt-0.8.7) with OvS (1.7.1) by Fake bridge but when I create
Guest error:
LibvirtException: Failed to add tap interface to bridge 'VirBr85': No such
process
Please help me.
Thanks
___
dev maili
On 17 October 2012 15:24, Simon Horman wrote:
> Hi Ben,
>
> sorry for not replying earlier. I have done some work in this area
> but I am not comfortable with publishing it as the Open Flow 1.3.1 spec
> has not been published yet.
Just FYI, the public can access it on the ONF website under docume
Hi, Isaku. Can you confirm that this change looks reasonable to you? If so,
I'll go ahead and push it.
Thanks,
--Justin
On Oct 16, 2012, at 5:36 PM, Reid Price wrote:
> Thanks Justin!
>
> Signed-off-by: Reid Price
>
> On Tue, Oct 16, 2012 at 5:29 PM, Justin Pettit wrote:
>> Commit 8cc
62 matches
Mail list logo
