netdev - RE: EEE unsupported on enetc

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:
 <PAXPR04MB85106EB2145E199F8393BE6388122@PAXPR04MB8510.eurprd04.prod.outlook.com>
Date: Wed, 8 Jan 2025 01:49:26 +0000
From: Wei Fang <wei.fang@....com>
To: Andrew Lunn <andrew@...n.ch>
CC: Heiner Kallweit <hkallweit1@...il.com>, Vladimir Oltean
	<vladimir.oltean@....com>, Claudiu Manoil <claudiu.manoil@....com>, Clark
 Wang <xiaoning.wang@....com>, Russell King - ARM Linux
	<linux@...linux.org.uk>, "netdev@...r.kernel.org" <netdev@...r.kernel.org>,
	"imx@...ts.linux.dev" <imx@...ts.linux.dev>
Subject: RE: EEE unsupported on enetc



Best Regards,
Wei Fang
> On Tue, Jan 07, 2025 at 02:19:53AM +0000, Wei Fang wrote:
> > > > In enetc_phylink_connect() we have the following:
> > > >
> > > > /* disable EEE autoneg, until ENETC driver supports it */
> > > > memset(&edata, 0, sizeof(struct ethtool_keee));
> > > > phylink_ethtool_set_eee(priv->phylink, &edata);
> > > >
> > > > Is it a hw constraint (if yes, on all IP versions?) that EEE isn't
> > > > supported, or is just some driver code for lpi timer handling missing?
> > > > Any plans to fix EEE in this driver?
> > >
> > > Hi Heiner,
> > >
> > > Currently, there are two platforms use the enetc driver, one is
> > > LS1028A, whose ENETC version is v1.0, and the other is i.MX95, whose
> > > version is v4.1. As far as I know, the ENETC hardware of both
> > > platforms supports EEE, but the implementation is different. As the
> > > maintainer of i.MX platform, I definitely sure Clark will add the
> > > EEE support for i.MX95 in the future. But for LS1028A, it is not
> > > clear to me whether Vladimir has plans to support EEE.
> >
> > By the way, I am confirming with NETC architect internally whether
> > LS1028A ENETC supports dynamic LPI mode like i.MX95 (RM does not
> > indicate this, but the relevant registers exist). If it does, we can
> > add EEE support to LS1028A and i.MX95 together.
> 
> Do you know what the reset defaults are? Can you confirm it is disabled in the
> MAC by default. We have the issue that we suspect some MACs have EEE
> support enabled by default using some default LPI timer value. If we disable
> EEE advertisement in the PHY by default for MACs which don't say they
> support EEE, we potentially cause regressions for those which are active by
> default, but without any control plane.
> 

Which platform do you use? LS1028A or i.MX95?

>From the RM of LS1028A and i.MX95, the default value is 0, a value of 0 does 
not activate low power EEE transmission. I'm on a business trip now and don't
have a board available to confirm it. Or I will confirm it for you later when I
return to the office. Also you can find the address of the PM0_SLEEP_TIMER
register in RM and then read the value of the register through devmem2 to
confirm it.