[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <VI1P190MB073490ED820B9BF0356059398F709@VI1P190MB0734.EURP190.PROD.OUTLOOK.COM>
Date: Thu, 9 Dec 2021 10:46:02 +0000
From: Volodymyr Mytnyk <volodymyr.mytnyk@...ision.eu>
To: Jamal Hadi Salim <jhs@...atatu.com>,
"netdev@...r.kernel.org" <netdev@...r.kernel.org>
CC: Taras Chornyi <taras.chornyi@...ision.eu>,
Mickey Rachamim <mickeyr@...vell.com>,
Serhiy Pshyk <serhiy.pshyk@...ision.eu>,
Volodymyr Mytnyk <vmytnyk@...vell.com>,
Taras Chornyi <tchornyi@...vell.com>,
"David S. Miller" <davem@...emloft.net>,
Jakub Kicinski <kuba@...nel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH net-next] net: prestera: flower template support
Hi Jamal,
>
> > Hi Jamal,
> >
> >>
> >>> From: Volodymyr Mytnyk<vmytnyk@...vell.com>
> >>>
> >>> Add user template explicit support. At this moment, max TCAM rule size
> >>> is utilized for all rules, doesn't matter which and how much flower
> >>> matches are provided by user. It means that some of TCAM space is
> >>> wasted, which impacts the number of filters that can be offloaded.
> >>>
> >>> Introducing the template, allows to have more HW offloaded filters.
> >>>
> >>> Example:
> >>> tc qd add dev PORT clsact
> >>> tc chain add dev PORT ingress protocol ip \
> >>> flower dst_ip 0.0.0.0/16
> >>
> >> "chain" or "filter"?
> >
> > tc chain add ... flower [tempalte] is the command to add explicitly chain with a given template
> >
>
> I guess you are enforcing the template on chain 0. My brain
> was expecting chain id to be called out.
>
chain 0 is the default chain id for "tc chain" & "tc filter" command,
so, that's why I did not mention it in the command line. Please note,
this patch adds only template support. Chains are not supported yet,
and will be added later.
>
> > tc filter ... is the command to add a filter itself in that chain
> >
>
> Got it.
>
>
> >> You are not using tc priority? Above will result in two priorities (the 0.0.0.0 entry will be more important) and in classical flower approach two different tables.
> >> I am wondering how you map the table to the TCAM.
> >> Is the priority sorting entirely based on masks in hardware?
> >
> > Kernel tc filter priority is used as a priority for HW rule (see flower implementation).
>
> The TCAM however should be able to accept many masks - is the idea
> here to enforce some mask per chain and then have priority being the
> priorities handle conflict? What happens when you explicitly specify
> priority. If you dont specify it the kernel provides it and essentially
> resolution is based on the order in which the rules are entered..
The HW rule insert/delete into TCAM is done by the FW itself. It means,
that the FW will take care about prio and (re)order the rule based on the
priority provided by user/kernel. So, kernel driver just need to provide
prio to the FW when adding the rule into the HW.
>
> cheers,
> jamal
Thanks and Regards,
Volodymyr
Powered by blists - more mailing lists