netdev - Re: RE: RE: [RESEND] [PATCH 1/1] can: m_can: Control tx and rx flow to avoid communication stall

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20230714-sublime-vanilla-239fd808904c-mkl@pengutronix.de>
Date: Fri, 14 Jul 2023 09:41:59 +0200
From: Marc Kleine-Budde <mkl@...gutronix.de>
To: Kumari Pallavi <kumari.pallavi@...el.com>
Cc: "rcsekar@...sung.com" <rcsekar@...sung.com>,
	"Sangannavar, Mallikarjunappa" <mallikarjunappa.sangannavar@...el.com>,
	"Nikula, Jarkko" <jarkko.nikula@...el.com>,
	"linux-can@...r.kernel.org" <linux-can@...r.kernel.org>,
	"netdev@...r.kernel.org" <netdev@...r.kernel.org>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"Thokala, Srikanth" <srikanth.thokala@...el.com>
Subject: Re: RE: RE: [RESEND] [PATCH 1/1] can: m_can: Control tx and rx flow
 to avoid communication stall

On 14.07.2023 07:26:01, Kumari Pallavi wrote:
> > On 07.07.2023 05:38:09, Kumari Pallavi wrote:
> > > > >  			if (netif_queue_stopped(dev) &&
> > > > >  			    !m_can_tx_fifo_full(cdev))
> > > > >  				netif_wake_queue(dev);
> > > > > @@ -1787,6 +1787,7 @@ static netdev_tx_t m_can_start_xmit(struct
> > > > > sk_buff
> > > > *skb,
> > > > >  		}
> > > > >  	} else {
> > > > >  		cdev->tx_skb = skb;
> > > > > +		m_can_write(cdev, M_CAN_IE, IR_ALL_INT & (IR_TEFN));
> > > >
> > > > - What's the purpose of  "()" around IR_TEFN?
> > > > - "IR_ALL_INT & (IR_TEFN)" is equal to IR_TEFN, isn't it?
> > > > - This basically disables all other interrupts, is this what you want to
> > > >   do?
> > > > - What happens if the bus is busy with high prio CAN frames and you want
> > > >   to send low prio ones? You will not get any RX-IRQ, this doesn't look
> > > >   correct to me.
> > > >
> > >
> > > Even though the RX interrupt is disabled (in IE), if there is an TX
> > > interrupt and the RF0N bit is set (in IR), the RX packet will still be
> > > serviced because the TX and RX share the same IRQ handler.
> > 
> > If the bus is busy with high prio CAN frames and the m_can wants to send a low
> > prio frame, the m_can will not be able to send it's CAN frame, there will be not
> > TX interrupt. If there are enough high prio CAN frames the RX buffer will
> > overflow.
> 
> Sorry for late reply, I agree let me see if I can try to simulate this
> scenario using CAN analyzer. I already stressed the current solution
> for more than 10 days and didn't observe any issue. However, I will
> try to incorporate this scenario for stress as well and come back.

From my point of view, this is a conceptional problem of this
workaround.

To test this issue, saturate the bus with CAN frames ID 0x0. Then on the
m_can send CAN frames with ID=0x7ff.

Marc

-- 
Pengutronix e.K.                 | Marc Kleine-Budde          |
Embedded Linux                   | https://www.pengutronix.de |
Vertretung Nürnberg              | Phone: +49-5121-206917-129 |
Amtsgericht Hildesheim, HRA 2686 | Fax:   +49-5121-206917-9   |

Download attachment "signature.asc" of type "application/pgp-signature" (489 bytes)