Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.38 as permitted sender) client-ip=23.128.96.38;
Date:   Wed, 22 Nov 2023 17:54:53 +0100
From:   Andrew Lunn <andrew@lunn.ch>
To:     =?iso-8859-1?Q?K=F6ry?= Maincent <kory.maincent@bootlin.com>
Cc:     "David S. Miller" <davem@davemloft.net>,
        Eric Dumazet <edumazet@google.com>,
        Jakub Kicinski <kuba@kernel.org>,
        Paolo Abeni <pabeni@redhat.com>,
        Jonathan Corbet <corbet@lwn.net>,
        Luis Chamberlain <mcgrof@kernel.org>,
        Russ Weight <russ.weight@linux.dev>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        "Rafael J. Wysocki" <rafael@kernel.org>,
        Rob Herring <robh+dt@kernel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@linaro.org>,
        Conor Dooley <conor+dt@kernel.org>,
        Thomas Petazzoni <thomas.petazzoni@bootlin.com>,
        netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
        linux-doc@vger.kernel.org, devicetree@vger.kernel.org
Subject: Re: [PATCH net-next 9/9] net: pse-pd: Add PD692x0 PSE controller
 driver
Message-ID: <04f59e77-134b-45b2-8759-84b8e22c30d5@lunn.ch>
References: <20231116-feature_poe-v1-0-be48044bf249@bootlin.com>
 <20231116-feature_poe-v1-9-be48044bf249@bootlin.com>
 <2ff8bea5-5972-4d1a-a692-34ad27b05446@lunn.ch>
 <20231122171112.59370d21@kmaincent-XPS-13-7390>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20231122171112.59370d21@kmaincent-XPS-13-7390>
Precedence: bulk

> > > +static int pd692x0_sendrecv_msg(struct pd692x0_priv *priv,
> > > +				struct pd692x0_msg *msg,
> > > +				struct pd692x0_msg_content *buf)
> > > +{
> > > +	struct device *dev = &priv->client->dev;
> > > +	int ret;
> > > +
> > > +	ret = pd692x0_send_msg(priv, msg);
> > > +	if (ret)
> > > +		return ret;
> > > +
> > > +	ret = pd692x0_recv_msg(priv, msg, buf);  
> > 
> > So this function takes at least 10 seconds?
> 
> No, on normal communication it takes a bit more than 30ms.

So i think the first step is to refactor this code to make it clear
what the normal path is, and what the exception path is, and the
timing of each.

> > > +	msg.content.sub[2] = id;
> > > +	ret = pd692x0_sendrecv_msg(priv, &msg, &buf);  
> > 
> > So this is also 10 seconds? 
> > 
> > Given its name, it looks like this is called via ethtool? Is the
> > ethtool core holding RTNL? It is generally considered bad to hold RTNL for
> > that long.
> 
> Yes it is holding RTNL lock. Should I consider another behavior in case of
> communication loss to not holding RTNL lock so long?

How often does it happen? On the scale of its a theoretical
possibility, through to it happens every N calls? Also, does it happen
on cold boot and reboot?

If its never supposed to happen, i would keep holding RTNL, and add a
pr_warn() that the PSE has crashed and burned, waiting for it to
reboot. If this is likely to happen on the first communication with
the device, we might want to do a dummy transfer during probe to get
is synchronized before we start using it with the RTNL held.

   Andrew