> -----Original Message-----
> From: Flavio Leitner <f...@sysclose.org>
> Sent: Friday, July 9, 2021 3:24 PM
> To: Van Haaren, Harry <harry.van.haa...@intel.com>
> Cc: Amber, Kumar <kumar.am...@intel.com>; ovs-dev@openvswitch.org
> Subject: Re: [ovs-dev] [v3] dpif/dpcls: limit count subtable search info logs
>
> On Fri, Jul 09, 2021 at 01:43:08PM +0000, Van Haaren, Harry wrote:
> > > -----Original Message-----
> > <snip patch contents>
> >
> > > > +Optimizing Specific Subtable Search
> > > > +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> > > > +
> > > > +The DPCLS wildcarding engine can be "specialized" to handle specific
> > > > subtable
> > > > +searching even faster than the generic subtable search
> > > > implementation.Not all
> > > > +subtable searches are specialized, if you see a message like the
> > > > following in
> > > > +your OVS output, please run the below commands to inform OVS community
> > > of the
> > >
> > > The documentation is supposed to help users to understand what is
> > > happening and perhaps consider the next steps. The text above
> > > requires deeper knowledge of how OVS internally works and provides
> > > no indication of what is causing it or consequences like if it is
> > > harmful or not. Also, it commits to add new specialized lookups
> > > in the next release, which might not be realistic.
> > >
> > > What do you think about the suggestion below based on your proposal?
> > >
> > > ----8<---
> > > During the packet classification, the datapath can use specialized
> > > lookup tables to optimize the search. However, not all situations
> > > are optimized. If you see a message like the following one in the OVS
> > > logs, it means that there is no specialized implementation available
> > > for the current networking traffic. In this case, OVS will continue
> > > to process the traffic normally using a more generic lookup table."
> > >
> > > "Using non-specialized AVX512 lookup for subtable (4,1) and possibly
> > > others."
> > >
> > > (Note that the numbers 4 and 1 will likely be different in your logs)
> > >
> > > Additional specialized lookups can be added to OVS if the user
> > > provides that log message along with the command output as show
> > > below to the OVS mailing list. Note that the numbers in the log
> > > message ("subtable (X,Y)") need to match with the numbers in
> > > the provided command output ("dp-extra-info:miniflow_bits(X,Y)").
> > >
> > > "ovs-appctl dpctl/dump-flows -m", which results in output like this:
> > >
> > > ufid:82770b5d-ca38-44ff-8283-74ba36bd1ca5,
> > > skb_priority(0/0),skb_mark(0/0)
> > > ,ct_state(0/0),ct_zone(0/0),ct_mark(0/0),ct_label(0/0),recirc_id(0),
> > >
> > > dp_hash(0/0),in_port(pcap0),packet_type(ns=0,id=0),eth(src=00:00:00:00:00:
> > >
> > > 00/00:00:00:00:00:00,dst=ff:ff:ff:ff:ff:ff/00:00:00:00:00:00),eth_type(
> > >
> > > 0x8100),vlan(vid=1,pcp=0),encap(eth_type(0x0800),ipv4(src=127.0.0.1/0.0.0.0
> > >
> > > ,dst=127.0.0.1/0.0.0.0,proto=17/0,tos=0/0,ttl=64/0,frag=no),udp(src=53/0,
> > > dst=53/0)), packets:77072681, bytes:3545343326, used:0.000s, dp:ovs,
> > > actions:vhostuserclient0, dp-extra-info:miniflow_bits(4,1)
> > >
> > > ----8<---
> >
> > Good suggestion, I guess I'm too close to the code to even be able describe
> > or
> > document it from a users point of view!
> >
> > I like the suggested changes, thanks. Respin or update-on-apply all fine
> > with me. -
> Harry
>
> Due to this busy patching period, could you please respin?
> It helps maintainers, the 0day bot can do usual checking, etc.
Ah yes, good input from CI, we will do!
_______________________________________________
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
