linux-kernel - Re: [PATCH net-next v2 3/3] net: phy: micrel: Fix forced link mode for KSZ886X switches

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: <20231013122646.bjiy6soo3pdquk53@skbuf>
Date:   Fri, 13 Oct 2023 15:26:46 +0300
From:   Vladimir Oltean <olteanv@...il.com>
To:     Oleksij Rempel <o.rempel@...gutronix.de>
Cc:     Woojung Huh <woojung.huh@...rochip.com>,
        UNGLinuxDriver@...rochip.com, Andrew Lunn <andrew@...n.ch>,
        Florian Fainelli <f.fainelli@...il.com>,
        "David S. Miller" <davem@...emloft.net>,
        Eric Dumazet <edumazet@...gle.com>,
        Jakub Kicinski <kuba@...nel.org>,
        Paolo Abeni <pabeni@...hat.com>,
        Heiner Kallweit <hkallweit1@...il.com>, kernel@...gutronix.de,
        linux-kernel@...r.kernel.org, Russell King <linux@...linux.org.uk>,
        netdev@...r.kernel.org,
        Alexander Stein <alexander.stein@...tq-group.com>
Subject: Re: [PATCH net-next v2 3/3] net: phy: micrel: Fix forced link mode
 for KSZ886X switches

Hi Oleksij,

On Thu, Oct 12, 2023 at 08:55:02AM +0200, Oleksij Rempel wrote:
> Address a link speed detection issue in KSZ886X PHY driver when in
> forced link mode. Previously, link partners like "ASIX AX88772B"
> with KSZ8873 could fall back to 10Mbit instead of configured 100Mbit.
> 
> The issue arises as KSZ886X PHY continues sending Fast Link Pulses (FLPs)
> even with autonegotiation off, misleading link partners in autoneg mode,
> leading to incorrect link speed detection.
> 
> Now, when autonegotiation is disabled, the driver sets the link state
> forcefully using KSZ886X_CTRL_FORCE_LINK bit. This action, beyond just
> disabling autonegotiation, makes the PHY state more reliably detected by
> link partners using parallel detection, thus fixing the link speed
> misconfiguration.
> 
> With autonegotiation enabled, link state is not forced, allowing proper
> autonegotiation process participation.
> 
> Signed-off-by: Oleksij Rempel <o.rempel@...gutronix.de>
> ---

Have you considered denying "ethtool -s swpN autoneg off" in "net.git"
(considering that it doesn't work properly), and re-enabling it in
"net-next.git"?

>  drivers/net/phy/micrel.c | 32 +++++++++++++++++++++++++++++---
>  1 file changed, 29 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/net/phy/micrel.c b/drivers/net/phy/micrel.c
> index 927d3d54658e..599ebf54fafe 100644
> --- a/drivers/net/phy/micrel.c
> +++ b/drivers/net/phy/micrel.c
> @@ -1729,9 +1729,35 @@ static int ksz886x_config_aneg(struct phy_device *phydev)
>  {
>  	int ret;
>  
> -	ret = genphy_config_aneg(phydev);
> -	if (ret)
> -		return ret;
> +	if (phydev->autoneg != AUTONEG_ENABLE) {
> +		ret = genphy_setup_forced(phydev);
> +		if (ret)
> +			return ret;

__genphy_config_aneg() will call genphy_setup_forced() as appropriate,
and additionally it will resync the master-slave resolution to a forced
value, if needed. So I think it's better to call genphy_config_aneg()
from the common code path, and just use the "if (phydev->autoneg)" test
to keep the vendor-specific register in sync with the autoneg setting.

> +
> +		/* When autonegotation is disabled, we need to manually force
> +		 * the link state. If we don't do this, the PHY will keep
> +		 * sending Fast Link Pulses (FLPs) which are part of the
> +		 * autonegotiation process. This is not desired when
> +		 * autonegotiation is off.
> +		 */
> +		ret = phy_set_bits(phydev, MII_KSZPHY_CTRL,
> +				   KSZ886X_CTRL_FORCE_LINK);
> +		if (ret)
> +			return ret;
> +	} else {
> +		/* If we had previously forced the link state, we need to
> +		 * clear KSZ886X_CTRL_FORCE_LINK bit now. Otherwise, the PHY
> +		 * will not perform autonegotiation.
> +		 */
> +		ret = phy_clear_bits(phydev, MII_KSZPHY_CTRL,
> +				     KSZ886X_CTRL_FORCE_LINK);
> +		if (ret)
> +			return ret;
> +
> +		ret = genphy_config_aneg(phydev);
> +		if (ret)
> +			return ret;
> +	}
>  
>  	/* The MDI-X configuration is automatically changed by the PHY after
>  	 * switching from autoneg off to on. So, take MDI-X configuration under
> -- 
> 2.39.2
>