Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Tue, 28 Jun 2022 10:15:32 -0700
From:   Jakub Kicinski <kuba@kernel.org>
To:     Eric Dumazet <edumazet@google.com>
Cc:     Albert Huang <huangjie.albert@bytedance.com>,
        "David S. Miller" <davem@davemloft.net>,
        Paolo Abeni <pabeni@redhat.com>,
        Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
        Menglong Dong <imagedong@tencent.com>,
        Petr Machata <petrm@nvidia.com>,
        Kumar Kartikeya Dwivedi <memxor@gmail.com>,
        Arnd Bergmann <arnd@arndb.de>,
        Antoine Tenart <atenart@kernel.org>,
        Phil Auld <pauld@redhat.com>,
        Frederic Weisbecker <frederic@kernel.org>,
        Xin Long <lucien.xin@gmail.com>,
        netdev <netdev@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] net : rps : supoort a single flow to use rps
Message-ID: <20220628101532.1be4a9df@kernel.org>
In-Reply-To: <CANn89iKfW8_OLN3veCaMDDLLPU1EP_eAcf03PJZJnLD+6Pv3vw@mail.gmail.com>
References: <20220628140044.65068-1-huangjie.albert@bytedance.com>
        <CANn89iKfW8_OLN3veCaMDDLLPU1EP_eAcf03PJZJnLD+6Pv3vw@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Tue, 28 Jun 2022 17:31:12 +0200 Eric Dumazet wrote:
> > how to use:
> > echo xxx > /sys/class/net/xxx/queues/rx-x/rps_cpus
> > echo 1 > /sys/class/net/xxx/queues/rx-x/rps_single_flow
> > and create a flow node with the function:
> > rps_flow_node_create  
> 
> Which part calls rps_flow_node_create() exactly ?
> 
> This seems to be very specialized to IPSEC.
> 
> Can IPSEC  use multiple threads for decryption ?

+1 Doesn't wireguard do something to spread the load across CPUs?
Maybe it can be generalized? I don't think a solution to "spray 
crypto around" belongs as part of RPS.