Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Thu, 29 Aug 2019 15:49:01 +0200
From:   Jiri Pirko <jiri@resnulli.us>
To:     Andrew Lunn <andrew@lunn.ch>
Cc:     Horatiu Vultur <horatiu.vultur@microchip.com>,
        alexandre.belloni@bootlin.com, UNGLinuxDriver@microchip.com,
        davem@davemloft.net, allan.nielsen@microchip.com,
        ivecera@redhat.com, f.fainelli@gmail.com, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH v3 1/2] net: core: Notify on changes to dev->promiscuity.
Message-ID: <20190829134901.GJ2312@nanopsycho>
References: <1567070549-29255-1-git-send-email-horatiu.vultur@microchip.com>
 <1567070549-29255-2-git-send-email-horatiu.vultur@microchip.com>
 <20190829095100.GH2312@nanopsycho>
 <20190829132611.GC6998@lunn.ch>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20190829132611.GC6998@lunn.ch>
User-Agent: Mutt/1.11.4 (2019-03-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Thu, Aug 29, 2019 at 03:26:11PM CEST, andrew@lunn.ch wrote:
>> NACK
>> 
>> This is invalid usecase for switchdev infra. Switchdev is there for
>> bridge offload purposes only.
>
>Hi Jiri
>
>I would argue this is for bridge offload. In another email, you say
>promisc is promisc. Does that mean the Mellonox hardware forwards
>every frame ingressing a port to the CPU by default as soon as it is
>enslaved to a bridge and promisc mode turned on? Or course not. At the
>moment, every switchdev driver wrongly implement promisc mode.
>
>This patchset is about correctly implementing promisc mode, so that
>applications can use it as expected. And that means configuring the
>hardware bridge to also forward a copy of frames to the CPU.

Wait, I believe there has been some misundestanding. Promisc mode is NOT
about getting packets to the cpu. It's about setting hw filters in a way
that no rx packet is dropped. For normal nics it means that all packets
get to the cpu, but that is just because it is the only direction they
can make.

If you want to get packets from the hw forwarding dataplane to cpu, you
should not use promisc mode for that. That would be incorrect.

If you want to get packets from the hw forwarding dataplane to cpu, you
should use tc trap action. It is there exactly for this purpose.

Promisc is for setting rx filters.


>
>I see trap as a different use case. tcpdump/pcap is not going to use
>traps.
>
>	Andrew