Re: [PATCH 1/1] net: core: 'ethtool' issue with querying phy settings

[Ben Hutchings]

On Sun, 2015-05-31 at 17:19 -0700, David Miller wrote:
> From: Ben Hutchings <ben@xxxxxxxxxxxxxxx>
> Date: Sun, 31 May 2015 20:54:06 +0100
> 
> > On Fri, 2015-05-22 at 16:15 -0400, David Miller wrote:
> >> From: Arun Parameswaran <aparames@xxxxxxxxxxxx>
> >> Date: Wed, 20 May 2015 14:35:30 -0700
> >> 
> >> > When trying to configure the settings for PHY1, using commands
> >> > like 'ethtool -s eth0 phyad 1 speed 100', the 'ethtool' seems to
> >> > modify other settings apart from the speed of the PHY1, in the
> >> > above case.
> >> > 
> >> > The ethtool seems to query the settings for PHY0, and use this
> >> > as the base to apply the new settings to the PHY1. This is
> >> > causing the other settings of the PHY 1 to be wrongly
> >> > configured.
> >> > 
> >> > The issue is caused by the '_ethtool_get_settings()' API, which
> >> > gets called because of the 'ETHTOOL_GSET' command, is clearing
> >> > the 'cmd' pointer (of type 'struct ethtool_cmd') by calling
> >> > memset. This clears all the parameters (if any) passed for the
> >> > 'ETHTOOL_GSET' cmd. So the driver's callback is always invoked
> >> > with 'cmd->phy_address' as '0'.
> >> > 
> >> > The '_ethtool_get_settings()' is called from other files in the
> >> > 'net/core'. So the fix is applied to the 'ethtool_get_settings()'
> >> > which is only called in the context of the 'ethtool'.
> >> > 
> >> > Signed-off-by: Arun Parameswaran <aparames@xxxxxxxxxxxx>
> >> > Reviewed-by: Ray Jui <rjui@xxxxxxxxxxxx>
> >> > Reviewed-by: Scott Branden <sbranden@xxxxxxxxxxxx>
> >> 
> >> Applied and queued up for -stable, thanks.
> > 
> > Please revert this.  This is an incompatible API change, not a bug fix.
> > The established semantics are that 'phyad' is filled in by the driver;
> > it is not a parameter to the ETHTOOL_GSET command.
> 
> But then how in the world can the user specify specific PHY ADs for
> a device that will respond to more than one?

ETHTOOL_SSET sets the current PHY address and ETHTOOL_GSET gets it.

If multiple PHYs need to be configured for a single link then the driver
should configure them all at the same time rather than making it the
administrator's problem.

What we can't support with the current API are:
- multiple physical links behind a single net device (different
  configuration possible for each link)
- multiple PHYs are needed for a single link, and the driver can't
  automatically decide which to use (multiple addresses to set)

Ben.

-- 
Ben Hutchings
Power corrupts.  Absolute power is kind of neat.
                           - John Lehman, Secretary of the US Navy 1981-1987

Attachment: signature.asc
Description: This is a digitally signed message part