Re: mii-diag strangeness

On 9 May 2002, Joe Rouvier wrote:
> On Wed, 2002-05-08 at 19:50, Donald Becker wrote:
> > On 8 May 2002, Joe Rouvier wrote:
> >
> > > I'm having a strange problem with mii-diag on IBM Netfinity 4000R's.  It
> > > seems that mii-diag works fine as a non-root user, but reads all nulls
> > > when run as root,
...
> I did some more checking on this.  It seems my earlier assumptions were
> incorrect.  Mii-diag fails on both the Network Engines and the IBMs, but
> only with the latest version, v2.02 is fine:

OK, that explains part of the problem.

The MII ioctl() constants were changed for the 2.4 kernel.  Yes, the
symbolic name that was supposed to represent a constant changed value
yet kept the same name.

The v2.02 'mii-diag' program always uses the symbolic constant, which
now changes depending on the system that it's compiled on.

The v2.03 'mii-diag' program tries both the new and old symbolic
values.  If the new value seems to return valid data, it uses the new,
otherwise it uses the old.

I've updated 'mii-diag' so that passing the '-v' flag will report which
value is being used.  Look for this in v2.04.

What driver and kernel version are you using?

> mii-diag.c:v2.03 11/5/2001 Donald Becker ([email protected])
> Basic registers of MII PHY #0:  0000 0000 0000 0000 0000 0000 0000 0000.
...
> mii-diag.c:v2.02 5/21/2001 Donald Becker ([email protected])
> Basic registers of MII PHY #1:  3000 782d 02a8 0154 05e1 41e1 0001 0000.

Note that 2.03 is trying to read PHY #0, not PHY #1.

-- 
Donald Becker				[email protected]
Scyld Computing Corporation		http://www.scyld.com
410 Severn Ave. Suite 210		Second Generation Beowulf Clusters
Annapolis MD 21403			410-990-9993