Re: [ovs-dev] [PATCH v5 1/9] datapath: add transport ports in route lookup for geneve

On Thu, Aug 9, 2018 at 4:13 PM, Qiuyu Xiao wrote: > I have one question. In "datapath/linux/compat/include/net/geneve.h", > USE_UPSTREAM_TUNNEL > decides whether to use Linux upstream kernel function or OVS kernel > function to transmit Geneve packet. Currently, it chooses Linux upstream >

Re: [ovs-dev] [PATCH] rhel: Install the network scripts in a new subpackage

On Thu, Aug 09, 2018 at 11:50:38AM +0200, Timothy Redaelli wrote: > On Wed, 8 Aug 2018 13:49:12 -0300 > Flavio Leitner wrote: > > > Actually, we also have dependencies in the systemd service > > to the network service: > > > > [Unit]

Re: [ovs-dev] infiniband (IPoIB) support

ср, 8 авг. 2018 г. в 12:28, Vasiliy Tolstov : > My question is - how can i do connectivity if vm attached to br-int > have external ip address for example 1.2.3.100 (provided by ovn dhcp), > and gateway server that have uplink to internet have ip 1.2.3.254. > As i understand logical router needs

Re: [ovs-dev] [PATCH 1/8] ip6_gre: Fix a bug that clears address bits

Hi Greg, Thanks for clarifying this bug, you did a much better job than I did. Sorry I didn't describe it clearly. I will created another patch, as you suggested. William, below is the crash dump you requested. [ 645.832991] general protection fault: [#1] SMP PTI [ 645.833033] Modules

Re: [ovs-dev] [PATCH 1/8] ip6_gre: Fix a bug that clears address bits

On 8/8/2018 11:32 AM, Yifeng Sun wrote: In compatable gre module, skb->cb is used as ovs_gso_cb. This bug clears the 16-23 bit in the address of ovs_gso_cb.tun_dst. Signed-off-by: Yifeng Sun --- datapath/linux/compat/ip6_gre.c | 1 - 1 file changed, 1 deletion(-) diff --git

Re: [ovs-dev] [PATCH v3 0/9] tests: Clean up syslog.

On Thu, Aug 09, 2018 at 04:51:44PM +0300, Ilya Maximets wrote: > On 09.08.2018 02:05, Ben Pfaff wrote: > > On Wed, Aug 08, 2018 at 03:35:31PM +0300, Ilya Maximets wrote: > >> Each run of the testsuite produces millions lines in a system > >> log. This is completely unnecessary and makes it

Re: [ovs-dev] [PATCH 1/8] ip6_gre: Fix a bug that clears address bits

On 8/9/2018 3:33 PM, William Tu wrote: On Thu, Aug 9, 2018 at 2:37 PM, Gregory Rose > wrote: On 8/9/2018 11:36 AM, Yifeng Sun wrote: Yes, I agree. It should be the way to go. Thanks, Yifeng If I'm reading the code correctly the OVS packet

Re: [ovs-dev] [PATCH v5 1/9] datapath: add transport ports in route lookup for geneve

On Thu, Aug 9, 2018 at 3:28 PM, Qiuyu Xiao wrote: > Hi William, > > ip_route_output_key() calls xfrm_lookup(). xfrm_lookup() needs L4 ports > so that the packet can match IPsec's security policy based on L4 ports. > IPsec security policy for Geneve selects udp packets with dst port 6081. > If no

Re: [ovs-dev] [PATCH] tests: Don't log to syslog during tests.

Thanks. I folded that in and applied this to master. Maybe some of your series is cleanups or improvements. If so, please feel free to re-send those parts of it. On Thu, Aug 09, 2018 at 05:02:29PM +0300, Ilya Maximets wrote: > The patch is good. > > The following incremental needed to make it

Re: [ovs-dev] [PATCH v5 3/6] debian and rhel: Create IPsec package.

On Thu, Aug 09, 2018 at 06:31:31PM -0400, Aaron Conole wrote: > Ben Pfaff writes: > > > On Thu, Aug 09, 2018 at 12:40:39PM -0700, Ansis Atteka wrote: > >> On Tue, 7 Aug 2018 at 09:43, Qiuyu Xiao wrote: > >> > > >> > Added rules and files to create debian and rpm ovs-ipsec packages. > >> > > >>

Re: [ovs-dev] [PATCH 1/8] ip6_gre: Fix a bug that clears address bits

On Thu, Aug 9, 2018 at 2:37 PM, Gregory Rose wrote: > On 8/9/2018 11:36 AM, Yifeng Sun wrote: > > Yes, I agree. It should be the way to go. > > Thanks, > Yifeng > > > If I'm reading the code correctly the OVS packet cmd will set the skb->cb > using the OVS_CB macro and then > pass that packet

Re: [ovs-dev] [PATCH v5 3/6] debian and rhel: Create IPsec package.

Ben Pfaff writes: > On Thu, Aug 09, 2018 at 12:40:39PM -0700, Ansis Atteka wrote: >> On Tue, 7 Aug 2018 at 09:43, Qiuyu Xiao wrote: >> > >> > Added rules and files to create debian and rpm ovs-ipsec packages. >> > >> > Signed-off-by: Qiuyu Xiao >> > Signed-off-by: Ansis Atteka >> >

Re: [ovs-dev] [PATCH v5 1/9] datapath: add transport ports in route lookup for geneve

Hi William, ip_route_output_key() calls xfrm_lookup(). xfrm_lookup() needs L4 ports so that the packet can match IPsec's security policy based on L4 ports. IPsec security policy for Geneve selects udp packets with dst port 6081. If no port information, the IPsec stack won't know the packet is a

Re: [ovs-dev] [PATCH v5 3/6] debian and rhel: Create IPsec package.

On Thu, Aug 09, 2018 at 12:40:39PM -0700, Ansis Atteka wrote: > On Tue, 7 Aug 2018 at 09:43, Qiuyu Xiao wrote: > > > > Added rules and files to create debian and rpm ovs-ipsec packages. > > > > Signed-off-by: Qiuyu Xiao > > Signed-off-by: Ansis Atteka > > Co-authored-by: Ansis Atteka > > Did

Re: [ovs-dev] [PATCH 5/8] system-traffic.at: Add erspan v1 tunnel test that doesn't depend on upstream gre module

On Wed, Aug 8, 2018 at 11:32 AM, Yifeng Sun wrote: > Introduce a new test that doesn't setup native erspan v1 tunnels but sends > simulated raw packets. > This test is supposed to only run for kernel version from 4.4.x to 4.15.x. > > Signed-off-by: Yifeng Sun > --- > Acked-by: William Tu

Re: [ovs-dev] [PATCH 7/8] system-traffic.at: Add ip6erspan v1 tunnel test that doesn't depend on upstream gre module

On Wed, Aug 8, 2018 at 11:32 AM, Yifeng Sun wrote: > Introduce a new test that doesn't setup native ip6erspan v1 tunnels but > sends > simulated raw packets. > This test is supposed to only run for kernel version from 4.4.x to 4.15.x. > > Signed-off-by: Yifeng Sun > --- > Acked-by: William Tu

Re: [ovs-dev] [PATCH 8/8] system-traffic.at: Add ip6erspan v2 tunnel test that doesn't depend on upstream gre module

On Wed, Aug 8, 2018 at 11:32 AM, Yifeng Sun wrote: > Introduce a new test that doesn't setup native ip6erspan v2 tunnels but > sends > simulated raw packets. > This test is supposed to only run for kernel version from 4.4.x to 4.15.x. > > Signed-off-by: Yifeng Sun > --- > Acked-by: William Tu

Re: [ovs-dev] [PATCH 1/8] ip6_gre: Fix a bug that clears address bits

On 8/9/2018 11:36 AM, Yifeng Sun wrote: Yes, I agree. It should be the way to go. Thanks, Yifeng If I'm reading the code correctly the OVS packet cmd will set the skb->cb using the OVS_CB macro and then pass that packet down to the ip6erspan_tunnel_xmit function.  The ip6erspan_tunnel_xmit

Re: [ovs-dev] [PATCH v5 1/9] datapath: add transport ports in route lookup for geneve

Hi Qiuyu, Can you explain a little more about why you need L4 ports for route lookup? Does Linux kernel support different route for different L4 ports? I thought it only uses L3 address for route lookup. On Mon, Aug 6, 2018 at 11:04 AM, Qiuyu Xiao wrote: > This patch adds transport ports

Re: [ovs-dev] [PATCH 6/8] system-traffic.at: Add erspan v2 tunnel test that doesn't depend on upstream gre module

Yes, exactly. Thanks, Yifeng On Thu, Aug 9, 2018 at 1:13 PM, William Tu wrote: > > > On Thu, Aug 9, 2018 at 12:52 PM, Yifeng Sun > wrote: > >> The above packet hex is supposed to be generated by linux native tunnel. >> But since native tunnel conflicts with openvswitch kernel module, so we >>

Re: [ovs-dev] [PATCH 6/8] system-traffic.at: Add erspan v2 tunnel test that doesn't depend on upstream gre module

On Thu, Aug 9, 2018 at 12:52 PM, Yifeng Sun wrote: > The above packet hex is supposed to be generated by linux native tunnel. > But since native tunnel conflicts with openvswitch kernel module, so we > don't add native tunnel but just send simulated packets to OVS. > > The ERSPAN implementation

Re: [ovs-dev] [PATCH 4/8] system-traffic.at: Add gre tunnel test that doesn't depend on upstream gre module

On Thu, Aug 9, 2018 at 12:46 PM, Yifeng Sun wrote: > Yes, in original test, ADD_NATIVE_TUNNEL will try to load upstream > gre modules. But since openvswitch is running in compatible model, > there is a conflict and the gre tests will always fail. So this new series > of tests abandoned using

Re: [ovs-dev] [PATCH 6/8] system-traffic.at: Add erspan v2 tunnel test that doesn't depend on upstream gre module

The above packet hex is supposed to be generated by linux native tunnel. But since native tunnel conflicts with openvswitch kernel module, so we don't add native tunnel but just send simulated packets to OVS. The ERSPAN implementation of OVS is still tested on br0 and br-underlay. In this test,

Re: [ovs-dev] [PATCH 3/8] system-traffic.at: Skip gre-related failed tests on kernel version from 4.4 to 4.15

Good question. I tried on RHEL7x but somehow these tests failed. For now, I have run these tests on 4.4, 4.15, 4.16 and they all passed. On Thu, Aug 9, 2018 at 11:59 AM, William Tu wrote: > > > On Wed, Aug 8, 2018 at 11:32 AM, Yifeng Sun > wrote: > >> Skip gre, erspan and ip6erspan related

Re: [ovs-dev] [PATCH 4/8] system-traffic.at: Add gre tunnel test that doesn't depend on upstream gre module

Yes, in original test, ADD_NATIVE_TUNNEL will try to load upstream gre modules. But since openvswitch is running in compatible model, there is a conflict and the gre tests will always fail. So this new series of tests abandoned using ADD_NATIVE_TUNNEL and doesn't need native gre modules any more.

Re: [ovs-dev] [PATCH v5 3/6] debian and rhel: Create IPsec package.

On Tue, 7 Aug 2018 at 09:43, Qiuyu Xiao wrote: > > Added rules and files to create debian and rpm ovs-ipsec packages. > > Signed-off-by: Qiuyu Xiao > Signed-off-by: Ansis Atteka > Co-authored-by: Ansis Atteka Did you test this patch on Fedora with SElinux enabled? ovs-monitor-ipsec daemon

Re: [ovs-dev] [PATCH 6/8] system-traffic.at: Add erspan v2 tunnel test that doesn't depend on upstream gre module

On Wed, Aug 8, 2018 at 11:32 AM, Yifeng Sun wrote: > Introduce a new test that doesn't setup native erspan v2 tunnels but sends > simulated raw packets. > This test is supposed to only run for kernel version from 4.4.x to 4.15.x. > > Signed-off-by: Yifeng Sun > --- > tests/system-traffic.at |

Re: [ovs-dev] [PATCH 3/8] system-traffic.at: Skip gre-related failed tests on kernel version from 4.4 to 4.15

On Wed, Aug 8, 2018 at 11:32 AM, Yifeng Sun wrote: > Skip gre, erspan and ip6erspan related tests on kernel version from > 4.4.x to 4.15.x because these tests will always fail. > > Signed-off-by: Yifeng Sun > --- > tests/system-traffic.at | 5 + > 1 file changed, 5 insertions(+) > > diff

Re: [ovs-dev] Problem in using ovs in raspberry pi - regd.

I understand that you have few Rasberry Pi's as switches and some as hosts. A picture/some more explanation of the topology would help. How many ports does the Rasberry Pi have. How many are connected to the OVS bridge? I understand you use the wireless interface of Rasberry Pi to connet to RYU.

Re: [ovs-dev] [PATCH 1/8] ip6_gre: Fix a bug that clears address bits

Yes, I agree. It should be the way to go. Thanks, Yifeng On Thu, Aug 9, 2018 at 11:17 AM, Gregory Rose wrote: > On 8/9/2018 11:03 AM, Yifeng Sun wrote: > > There is a difference regarding how skb_tunnel_info is stored in skb > between ovs and upstream kernel. In ovs's compatible gre module, >

Re: [ovs-dev] [PATCH 1/8] ip6_gre: Fix a bug that clears address bits

On 8/9/2018 11:03 AM, Yifeng Sun wrote: There is a difference regarding how skb_tunnel_info is stored in skb between ovs and upstream kernel. In ovs's compatible gre module, skb_tunnel_info is stored in skb->cb while in upstream kernel, it is referenced by skb->_skb_refdst. The upstream netdev

Re: [ovs-dev] [PATCH 1/8] ip6_gre: Fix a bug that clears address bits

There is a difference regarding how skb_tunnel_info is stored in skb between ovs and upstream kernel. In ovs's compatible gre module, skb_tunnel_info is stored in skb->cb while in upstream kernel, it is referenced by skb->_skb_refdst. The upstream netdev code should be okay. To fix this issue,

[ovs-dev] [PATCH 3/3] ovs-sandbox: Generate the SSL keys using the default key length

This commit removes the explicit set of 1024-bit RSA keys when the RSA keys are generated on "make sandbox" and so the default (2048-bit) is used. Signed-off-by: Timothy Redaelli --- tutorial/ovs-sandbox | 8 1 file changed, 4 insertions(+), 4 deletions(-) diff --git

[ovs-dev] [PATCH 2/3] ovn-architecture: Use the default key length in examples

This commit removes the explicit set of 1024-bit RSA keys on ovn-architecture examples and so the default (2048-bit) is used. Signed-off-by: Timothy Redaelli --- ovn/ovn-architecture.7.xml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/ovn/ovn-architecture.7.xml

[ovs-dev] [PATCH 1/3] tests: Use the default key length when generating RSA keys

This commit removes the explicit set of 1024-bit RSA keys when ovs-pki is launched and so the default (2048-bit) is used. Signed-off-by: Timothy Redaelli --- tests/ovs-vsctl.at | 4 ++-- tests/ovsdb-rbac.at | 8 2 files changed, 6 insertions(+), 6 deletions(-) diff --git

[ovs-dev] [PATCH 0/3] Use the default key length for RSA keys

Currently, 1024-bit RSA keys are generated for OVS tests, are suggested in ovn-architecture manpage examples and are used to generate the RSA keys inside the sandbox (make sandbox), but OpenSSL documentation suggests to use at least 2048-bit keys, since "fewer amount of bits is considered insecure

Re: [ovs-dev] [PATCH 1/8] ip6_gre: Fix a bug that clears address bits

On 8/8/2018 6:50 PM, William Tu wrote: thanks for the fix. On Wed, Aug 8, 2018 at 11:32 AM, Yifeng Sun wrote: In compatable gre module, skb->cb is used as ovs_gso_cb. This bug clears the 16-23 bit in the address of ovs_gso_cb.tun_dst. can you explain more about ovs_gso_cb?

Re: [ovs-dev] [PATCH v7 00/13] Support multi-segment mbufs

On 09/08/2018 12:44, Ilya Maximets wrote: > On 09.08.2018 11:38, Lam, Tiago wrote: >> Hi Ilya, >> >> On 09/08/2018 09:27, Ilya Maximets wrote: >>> Hmm. I found that this series modifies only dpdk related components >>> and doesn't pay any attention to others. >>> dp_packet API was modified to

Re: [ovs-dev] [PATCH] vswitch.xml: Update dpdk-init documentation.

Kevin Traynor writes: > dpdk-init is now a string. Add description of 'true' and 'try'. > > Fixes: 3e52fa5644cd ("dpdk: reflect status and version in the database") > Cc: acon...@redhat.com > Signed-off-by: Kevin Traynor > --- Acked-by: Aaron Conole D'oh! Thanks, Kevin.

[ovs-dev] [PATCH] vswitch.xml: Update dpdk-init documentation.

dpdk-init is now a string. Add description of 'true' and 'try'. Fixes: 3e52fa5644cd ("dpdk: reflect status and version in the database") Cc: acon...@redhat.com Signed-off-by: Kevin Traynor --- vswitchd/vswitch.xml | 14 ++ 1 file changed, 10 insertions(+), 4 deletions(-) diff --git

Re: [ovs-dev] [PATCH v7 11/13] dpdk-tests: Add uni-tests for multi-seg mbufs.

> In order to create a minimal environment that allows the tests to get > mbufs from an existing mempool, the following approach is taken: > - EAL is initialised (by using the main dpdk_init()) and a (very) small > mempool is instantiated (mimicking the logic in dpdk_mp_create()). > This

Re: [ovs-dev] [PATCH v7 07/13] dp-packet: Handle multi-seg mubfs in shift() func.

> In its current implementation dp_packet_shift() is also unaware of multi- > seg mbufs (that holds data in memory non-contiguously) and assumes that > data exists contiguously in memory, memmove'ing data to perform the shift. > > To add support for multi-seg mbuds a new set of functions was

Re: [ovs-dev] [PATCH v7 00/13] Support multi-segment mbufs

On 8/9/2018 12:44 PM, Ilya Maximets wrote: On 09.08.2018 11:38, Lam, Tiago wrote: Hi Ilya, On 09/08/2018 09:27, Ilya Maximets wrote: Hmm. I found that this series modifies only dpdk related components and doesn't pay any attention to others. dp_packet API was modified to respect the segmented

Re: [ovs-dev] [ovs-dev, Branch2.8] netdev-dpdk: Fix failure to configure flow control at netdev-init.

"Chandran, Sugesh" writes: > Hi Aron, > Thank you for pointing out. > I will send out the patch again with right branch convention to make the > robot happy. No need. We can ignore the bot in this case. A committer can do the right thing. Just an FYI for future commits. Thanks! > Regards

Re: [ovs-dev] [PATCH] tests: Don't log to syslog during tests.

The patch is good. The following incremental needed to make it complete: diff --git a/python/ovs/vlog.py b/python/ovs/vlog.py index d7a84d8..7d416c3 100644 --- a/python/ovs/vlog.py +++ b/python/ovs/vlog.py @@ -305,9 +305,11 @@ class Vlog(object): return logger =

Re: [ovs-dev] [PATCH v3 0/9] tests: Clean up syslog.

On 09.08.2018 02:05, Ben Pfaff wrote: > On Wed, Aug 08, 2018 at 03:35:31PM +0300, Ilya Maximets wrote: >> Each run of the testsuite produces millions lines in a system >> log. This is completely unnecessary and makes it difficult to >> use system logs on test / build servers. >> >> This series is

[ovs-dev] Interop ITX 2018 Attendees List

Hi, Good day to you, This is an outstanding offer for the Interop ITX 2018. I am writing to check if you would be interested in acquiring the list of *37,393 *attendees participated in Interop ITX 2018 for your marketing and sales initiatives. This is an opportunity to acquire the list of

Re: [ovs-dev] [ovs-dev, Branch2.8] netdev-dpdk: Fix failure to configure flow control at netdev-init.

Hi Aron, Thank you for pointing out. I will send out the patch again with right branch convention to make the robot happy. Regards _Sugesh > -Original Message- > From: Aaron Conole [mailto:acon...@bytheb.org] > Sent: Thursday, August 9, 2018 2:15 PM > To: 0-day Robot > Cc: Chandran,

Re: [ovs-dev] [ovs-dev, Branch2.8] netdev-dpdk: Fix failure to configure flow control at netdev-init.

Hi Sugesh (and robot), 0-day Robot writes: > Bleep bloop. Greetings Sugesh Chandran, I am a robot and I have tried out > your patch. > Thanks for your contribution. > > I encountered some error that I wasn't expecting. See the details below. > > > git-am: > Failed to merge in the changes. >

[ovs-dev] [PATCH Branch2.7] netdev-dpdk: Fix failure to configure flow control at netdev-init.

This patch backports the commit from the latest OVS master to OVS-2.7. Configuring flow control at ixgbe netdev-init is throwing error in port start. For eg: without this fix, user cannot configure flow control on ixgbe dpdk port as below, " ovs-vsctl add-port br0 dpdk0 -- set Interface

Re: [ovs-dev] [PATCH v7 00/13] Support multi-segment mbufs

On 09.08.2018 11:38, Lam, Tiago wrote: > Hi Ilya, > > On 09/08/2018 09:27, Ilya Maximets wrote: >> Hmm. I found that this series modifies only dpdk related components >> and doesn't pay any attention to others. >> dp_packet API was modified to respect the segmented packets, but >> there are many

Re: [ovs-dev] [ovs-dev, Branch2.8] netdev-dpdk: Fix failure to configure flow control at netdev-init.

Bleep bloop. Greetings Sugesh Chandran, I am a robot and I have tried out your patch. Thanks for your contribution. I encountered some error that I wasn't expecting. See the details below. git-am: Failed to merge in the changes. Patch failed at 0001 netdev-dpdk: Fix failure to configure flow

[ovs-dev] [PATCH Branch2.8] netdev-dpdk: Fix failure to configure flow control at netdev-init.

This patch backports the commit from the latest OVS master to OVS-2.8. Configuring flow control at ixgbe netdev-init is throwing error in port start. For eg: without this fix, user cannot configure flow control on ixgbe dpdk port as below, " ovs-vsctl add-port br0 dpdk0 -- set Interface

Re: [ovs-dev] OVS-DPDK public meeting

Next meeting: 21st August 16:00 UTC. Minutes of meeting 8th August 2018. Attendees: Eelco, Thomas, Matteo, Ian, Ophir, Johann, Yipeng, Tiago, Simon, Kevin. === GENERAL === - OVS 2.10 -- No info on anything that would make 2.10 slip -- No news is good news :-) - Updates to OVS to use

Re: [ovs-dev] [PATCH] rhel: Install the network scripts in a new subpackage

On Wed, 8 Aug 2018 13:49:12 -0300 Flavio Leitner wrote: > Actually, we also have dependencies in the systemd service > to the network service: > > [Unit] > > Description=Open vSwitch Forwarding Unit >

Re: [ovs-dev] [ovs-dev, v2] netdev: Retry getting interfaces on inconsistent dumps from kernel

Bleep bloop. Greetings Daniel Alvarez, I am a robot and I have tried out your patch. Thanks for your contribution. I encountered some error that I wasn't expecting. See the details below. checkpatch: ERROR: Too many signoffs; are you missing Co-authored-by lines? ERROR:

[ovs-dev] [PATCH v2] netdev: Retry getting interfaces on inconsistent dumps from kernel

This patch in glibc [0] is fixing a bug where we may be getting inconsistent dumps from the kernel when listing interfaces due to a race condition. This could happen if we try to retrieve them while interfaces are being added/removed from the system at the same time. For systems running against

Re: [ovs-dev] [PATCH v7 00/13] Support multi-segment mbufs

Hi Ilya, On 09/08/2018 09:27, Ilya Maximets wrote: > Hmm. I found that this series modifies only dpdk related components > and doesn't pay any attention to others. > dp_packet API was modified to respect the segmented packets, but > there are many places, where code uses packet data directly

Re: [ovs-dev] [PATCH v7 00/13] Support multi-segment mbufs

Hmm. I found that this series modifies only dpdk related components and doesn't pay any attention to others. dp_packet API was modified to respect the segmented packets, but there are many places, where code uses packet data directly using only dp_packet_data() pointer and dp_packet_size(). For

Re: [ovs-dev] [PATCHv3] netdev-dpdk: Fix failure to configure flow control at netdev-init.

Regards _Sugesh > -Original Message- > From: Stokes, Ian > Sent: Wednesday, August 8, 2018 2:08 PM > To: Chandran, Sugesh ; ovs- > d...@openvswitch.org > Subject: RE: [PATCHv3] netdev-dpdk: Fix failure to configure flow control at > netdev-init. > > > Configuring flow control at