Re: [RESEND PATCH 4/4] can: c_can: cache frames to operate as a true FIFO

From: Dario Binacchi <dariobin@libero.it>
To: Marc Kleine-Budde <mkl@pengutronix.de>
Cc: linux-kernel@vger.kernel.org,
	Gianluca Falavigna <gianluca.falavigna@inwind.it>,
	Andrew Lunn <andrew@lunn.ch>,
	"David S. Miller" <davem@davemloft.net>,
	Jakub Kicinski <kuba@kernel.org>,
	Oliver Hartkopp <socketcan@hartkopp.net>,
	Vincent Mailhol <mailhol.vincent@wanadoo.fr>,
	Wolfgang Grandegger <wg@grandegger.com>,
	linux-can@vger.kernel.org, netdev@vger.kernel.org
Subject: Re: [RESEND PATCH 4/4] can: c_can: cache frames to operate as a true FIFO
Date: Sat, 7 Aug 2021 14:36:15 +0200 (CEST)	[thread overview]
Message-ID: <1265712151.254355.1628339776021@mail1.libero.it> (raw)
In-Reply-To: <20210806092523.hij5ejjq6wecbgfr@pengutronix.de>

> Il 06/08/2021 11:25 Marc Kleine-Budde <mkl@pengutronix.de> ha scritto:
> 
>  
> On 05.08.2021 22:16:06, Dario Binacchi wrote:
> > > > --- a/drivers/net/can/c_can/c_can.h
> > > > +++ b/drivers/net/can/c_can/c_can.h
> > > > @@ -200,6 +200,7 @@ struct c_can_priv {
> > > >  	atomic_t sie_pending;
> > > >  	unsigned long tx_dir;
> > > >  	int last_status;
> > > > +	spinlock_t tx_lock;
> > > 
> > > What does the spin lock protect?
> [...]
> > > > @@ -483,7 +469,11 @@ static netdev_tx_t c_can_start_xmit(struct sk_buff *skb,
> > > >  	if (c_can_get_tx_free(tx_ring) == 0)
> > > >  		netif_stop_queue(dev);
> > > >  
> > > > -	obj = idx + priv->msg_obj_tx_first;
> > > > +	spin_lock_bh(&priv->tx_lock);
> > > 
> > > What does the spin_lock protect? The ndo_start_xmit function is properly
> > > serialized by the networking core.
> > > 
> > 
> > The spin_lock protects the access to the IF_TX interface.
> 
> How? You only use the spin_lock in c_can_start_xmit(), but not anywhere
> else.
> 
> > Enabling the transmission of cached messages occur inside interrupt
> 
> The call chain is c_can_poll() -> c_can_do_tx(), and c_can_poll() is
> called from NAPI, which is not the IRQ handler.
> 
> > and the use of the IF_RX interface, which would avoid the use of the
> > spinlock, has not been validated by the tests.
> 
> What do you mean be has not been validated?

It's been a while since I submitted the series and I certainly got confused.

> 
> The driver already uses IF_RX to avoid concurrent access in
> c_can_do_tx() for c_can_inval_tx_object() [1], why not use IF_RX for
> c_can_object_put(), too?
> 
> [1] https://lore.kernel.org/r/20210302215435.18286-4-dariobin@libero.it

Right!

Thanks and Regards,
Dario

> 
> Marc
> 
> -- 
> Pengutronix e.K.                 | Marc Kleine-Budde           |
> Embedded Linux                   | https://www.pengutronix.de  |
> Vertretung West/Dortmund         | Phone: +49-231-2826-924     |
> Amtsgericht Hildesheim, HRA 2686 | Fax:   +49-5121-206917-5555 |