netdev - Re: [RFC PATCH net-next 2/6] net: ethernet: add mac-phy interrupt support with reset complete handling

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

Message-ID: <027d8f7b-6932-4d9d-b2f9-5369806a79a3@lunn.ch>
Date: Wed, 13 Sep 2023 04:39:41 +0200
From: Andrew Lunn <andrew@...n.ch>
To: Parthiban Veerasooran <Parthiban.Veerasooran@...rochip.com>
Cc: davem@...emloft.net, edumazet@...gle.com, kuba@...nel.org,
	pabeni@...hat.com, robh+dt@...nel.org,
	krzysztof.kozlowski+dt@...aro.org, conor+dt@...nel.org,
	corbet@....net, steen.hegelund@...rochip.com, rdunlap@...radead.org,
	horms@...nel.org, casper.casan@...il.com, netdev@...r.kernel.org,
	devicetree@...r.kernel.org, linux-kernel@...r.kernel.org,
	linux-doc@...r.kernel.org, horatiu.vultur@...rochip.com,
	Woojung.Huh@...rochip.com, Nicolas.Ferre@...rochip.com,
	UNGLinuxDriver@...rochip.com, Thorsten.Kummermehr@...rochip.com
Subject: Re: [RFC PATCH net-next 2/6] net: ethernet: add mac-phy interrupt
 support with reset complete handling

> +static int oa_tc6_sw_reset(struct oa_tc6 *tc6)
> +{
> +	long timeleft;
> +	u32 regval;
> +	int ret;
> +
> +	/* Perform software reset with both protected and unprotected control
> +	 * commands because the driver doesn't know the current status of the
> +	 * MAC-PHY.
> +	 */
> +	regval = SW_RESET;
> +	reinit_completion(&tc6->rst_complete);
> +	ret = oa_tc6_perform_ctrl(tc6, OA_TC6_RESET, &regval, 1, true, false);
> +	if (ret) {
> +		dev_err(&tc6->spi->dev, "RESET register write failed\n");
> +		return ret;
> +	}
> +
> +	ret = oa_tc6_perform_ctrl(tc6, OA_TC6_RESET, &regval, 1, true, true);
> +	if (ret) {
> +		dev_err(&tc6->spi->dev, "RESET register write failed\n");
> +		return ret;
> +	}
> +	timeleft = wait_for_completion_interruptible_timeout(&tc6->rst_complete,
> +							     msecs_to_jiffies(1));
> +	if (timeleft <= 0) {
> +		dev_err(&tc6->spi->dev, "MAC-PHY reset failed\n");
> +		return -ENODEV;
> +	}

This seems a bit messy and complex. I assume reset is performed once
during probe, and never again? So i wonder if it would be cleaner to
actually just poll for the reset to complete? You can then remove all
this completion code, and the interrupt handler gets simpler?

> +	/* Register MAC-PHY interrupt service routine */
> +	ret = devm_request_irq(&spi->dev, spi->irq, macphy_irq, 0, "macphy int",
> +			       tc6);
> +	if ((ret != -ENOTCONN) && ret < 0) {
> +		dev_err(&spi->dev, "Error attaching macphy irq %d\n", ret);
> +		goto err_macphy_irq;
> +	}

Why is -ENOTCONN special? A comment would be good here.

> -void oa_tc6_deinit(struct oa_tc6 *tc6)
> +int oa_tc6_deinit(struct oa_tc6 *tc6)
>  {
> -	kfree(tc6);
> +	int ret;
> +
> +	devm_free_irq(&tc6->spi->dev, tc6->spi->irq, tc6);
> +	ret = kthread_stop(tc6->tc6_task);
> +	if (!ret)
> +		kfree(tc6);
> +	return ret;
>  }

What is the MAC driver supposed to do if this fails?

But this problem probably goes away once you use a threaded interrupt
handler.

w> +/* Open Alliance TC6 Standard Control and Status Registers */
> +#define OA_TC6_RESET	0x0003		/* Reset Control and Status Register */
> +#define OA_TC6_STS0	0x0008		/* Status Register #0 */

Please use the same name as the standard. It use STATUS0, so
OA_TC6_STATUS0. Please make sure all your defines follow the standard.

> +
> +/* RESET register field */
> +#define SW_RESET	BIT(0)		/* Software Reset */

It is pretty normal to put #defines for a register members after the
#define for the register itself:

#define OA_TC6_RESET	0x0003		/* Reset Control and Status Register */
#define OA_TC6_RESET_SWRESET	BIT(0)

#define OA_TC6_STATUS0	0x0008		/* Status Register #0 */
#define OA_TC6_STATUS0_RESETC		BIT(6)		/* Reset Complete */

The naming like this also helps avoid mixups.

    Andrew