Re: [ovs-dev] [PATCH branch-2.13 0/3] Backport ct-offload state flag support

On Fri, Apr 09, 2021 at 09:35:32AM -0400, Aaron Conole wrote: > Backport the recent fixes related to ct-offload state flags being ignored > to fix HWOL. > > This pulls in the skip_hw flag refactor (note it already existed, so the > refactor exposes it to the upper layer - if you consider this a

Re: [ovs-dev] [PATCH branch-2.15 0/2] Backport ct-offload state flag support

On Fri, Apr 09, 2021 at 09:31:50AM -0400, Aaron Conole wrote: > Backport the recent fixes related to ct-offload state flags being ignored > to fix HWOL. > > Paul Blakey (2): > compat: Add ct_state flags definitions > netdev-offload-tc: Probe for support for any of the ct_state flags

Re: [ovs-dev] [PATCH branch-2.14 2/3] netdev-offload-tc: Probe for support for any of the ct_state flags

Marcelo Leitner writes: > On Fri, Apr 09, 2021 at 09:35:04AM -0400, Aaron Conole wrote: >> +static int >> +probe_insert_ct_state_rule(int ifindex, uint16_t ct_state, struct tcf_id >> *id) >> +{ >> +int prio = TC_RESERVED_PRIORITY_MAX + 1; >> +struct tc_flower flower; >> + >> +

Re: [ovs-dev] [PATCH ovn v2] ofctrl: Send all flow modifications in a bundle.

On 4/15/21 7:15 PM, Han Zhou wrote: > > > On Tue, Apr 13, 2021 at 1:23 AM Ilya Maximets > wrote: >> >> If some OF rules needs to be replaced due to logical flow changes, >> ovn-controller deletes old rules first and adds new ones later. >> In a complex scenario with

Re: [ovs-dev] [PATCH branch-2.14 2/3] netdev-offload-tc: Probe for support for any of the ct_state flags

On Fri, Apr 09, 2021 at 09:35:04AM -0400, Aaron Conole wrote: > +static int > +probe_insert_ct_state_rule(int ifindex, uint16_t ct_state, struct tcf_id *id) > +{ > +int prio = TC_RESERVED_PRIORITY_MAX + 1; > +struct tc_flower flower; > + > +memset(, 0, sizeof flower); > +

Re: [ovs-dev] [PATCH ovn v2] ofctrl: Send all flow modifications in a bundle.

On Tue, Apr 13, 2021 at 1:23 AM Ilya Maximets wrote: > > If some OF rules needs to be replaced due to logical flow changes, > ovn-controller deletes old rules first and adds new ones later. > In a complex scenario with big number of flows a lot of time > can pass between these events leading to

[ovs-dev] NVIDIA Roadmap for v2.16

Hello, This is the list of topics we are working on for the v2.16 release: OVS-DPDK Hardware Offloads: * VXLAN decap v6 submitted: https://patchwork.ozlabs.org/project/openvswitch/list/?series=237293 * IP fragmentation match v1 submitted:

Re: [ovs-dev] [PATCH v3 ovn] controller: do not claim localports

> On Wed, Apr 14, 2021 at 12:55 PM Lorenzo Bianconi > wrote: > > > > Localports should not be binded to any specific controller but should be > > available to each hv however in the current codebase consider_iface_claim > > routine is claiming the localport for a specific chassis. > > Fix the

Re: [ovs-dev] [PATCH v3 ovn] controller: do not claim localports

On Wed, Apr 14, 2021 at 12:55 PM Lorenzo Bianconi wrote: > > Localports should not be binded to any specific controller but should be > available to each hv however in the current codebase consider_iface_claim > routine is claiming the localport for a specific chassis. > Fix the issue skipping

Re: [ovs-dev] [PATCH] flow: Count and dump invalid IP packets.

On 15 Apr 2021, at 15:25, David Marchand wrote: On Thu, Apr 15, 2021 at 1:45 PM Flavio Leitner wrote: +COVERAGE_DEFINE(ipv4_check_too_short); +COVERAGE_DEFINE(ipv4_check_length_error); +COVERAGE_DEFINE(ipv6_check_too_short); +COVERAGE_DEFINE(ipv6_check_length_error); The check keyword is

Re: [ovs-dev] [PATCH v2 ovn] Support 802.11ad EthType for localnet ports

On Thu, Apr 1, 2021 at 1:06 PM Ihar Hrachyshka wrote: > > In some environments, hardware serving the fabric network doesn't > support double 802.1q (0x8100) VLAN tags, but does support 802.11ad > (0x8a88) EthType for two layer VLAN traffic. Specifically, Cisco > hardware UCS VIC was identified

Re: [ovs-dev] [PATCH] flow: Count and dump invalid IP packets.

On Thu, Apr 15, 2021 at 1:12 PM Flavio Leitner wrote: > > > +static void > > > +dump_invalid_packet(struct dp_packet *packet, const char *reason) > > > +{ > > > +static struct vlog_rate_limit rl = VLOG_RATE_LIMIT_INIT(1, 5); > > > +struct ds ds = DS_EMPTY_INITIALIZER; > > > +size_t

Re: [ovs-dev] [PATCH] flow: Count and dump invalid IP packets.

On Thu, Apr 15, 2021 at 1:45 PM Flavio Leitner wrote: > > > > > +COVERAGE_DEFINE(ipv4_check_too_short); > > > > > +COVERAGE_DEFINE(ipv4_check_length_error); > > > > > +COVERAGE_DEFINE(ipv6_check_too_short); > > > > > +COVERAGE_DEFINE(ipv6_check_length_error); > > > > > > > > The check keyword is

Re: [ovs-dev] [PATCH] ipf.c: Fix userspace datapath crash caused by IP fragments

王亮 Liang Wang writes: > Hi, Conole. Thank you for pointing out the mistakes in my last patch. Yes, I > did missed some something important. > The correct and complete way to fix this problem is like the following patch. > In fact, we have found and fixed this > crash problem one year ago; and

Re: [ovs-dev] [PATCH ovn] Support 802.11ad EthType for localnet ports

Hi Mark, thanks for comments. I figured the issues you pointed out and there's a new version up for review. Please take a look. Ihar On Wed, Mar 31, 2021 at 9:37 PM Mark Michelson wrote: > > Hi Ihar, > > See comments in-line > > On 3/29/21 9:29 PM, Ihar Hrachyshka wrote: > > In some

Re: [ovs-dev] [PATCH] flow: Count and dump invalid IP packets.

On Thu, Apr 15, 2021 at 01:32:06PM +0200, Eelco Chaudron wrote: > > > On 15 Apr 2021, at 13:12, Flavio Leitner wrote: > > > Hi Eelco, > > > > On Thu, Apr 15, 2021 at 10:07:24AM +0200, Eelco Chaudron wrote: > > > > > > > > > On 14 Apr 2021, at 17:50, David Marchand wrote: > > > > > > >

Re: [ovs-dev] [PATCH] flow: Count and dump invalid IP packets.

On 15 Apr 2021, at 13:12, Flavio Leitner wrote: Hi Eelco, On Thu, Apr 15, 2021 at 10:07:24AM +0200, Eelco Chaudron wrote: On 14 Apr 2021, at 17:50, David Marchand wrote: Skipping further processing of invalid IP packets helps avoid crashes but it does not help to figure out if the

Re: [ovs-dev] [PATCH] flow: Count and dump invalid IP packets.

Hi Eelco, On Thu, Apr 15, 2021 at 10:07:24AM +0200, Eelco Chaudron wrote: > > > On 14 Apr 2021, at 17:50, David Marchand wrote: > > > Skipping further processing of invalid IP packets helps avoid crashes > > but it does not help to figure out if the malformed packets are still > > present on

Re: [ovs-dev] [RFC 0/1] Userspace deferral of work

> -Original Message- > From: David Marchand > Sent: Wednesday 14 April 2021 17:02 > To: Ferriter, Cian > Cc: ovs-dev@openvswitch.org > Subject: Re: [ovs-dev] [RFC 0/1] Userspace deferral of work > > On Wed, Apr 14, 2021 at 5:44 PM Ferriter, Cian > wrote: > > > > I am sending this RFC

Re: [ovs-dev] [PATCH] flow: Count and dump invalid IP packets.

On 14 Apr 2021, at 17:50, David Marchand wrote: Skipping further processing of invalid IP packets helps avoid crashes but it does not help to figure out if the malformed packets are still present on the network. Add coverage counters for IPv4 and IPv6 sanity checks so that we know there are

Re: [ovs-dev] [PATCH] ipf.c: Fix userspace datapath crash caused by IP fragments

Hi, Conole. Thank you for pointing out the mistakes in my last patch. Yes, I did missed some something important. The correct and complete way to fix this problem is like the following patch. In fact, we have found and fixed this crash problem one year ago; and this patch has worked very well in