Re: [E1000-devel] Lost Ticks and ICH8

[Brandeburg, J. <jes...@in...>]

Jason Gunthorpe wrote:
>> ugh, this doesn't look good.  please include output of ethtool -e
>> eth0, the firmware version is definitely not correct. (-1.-1dash-1)
>=20
> Hmm, more weirdness:
>=20
> $ ethtool -e eth0
> Cannot get EEPROM data: No such process
>=20
> $ strace ethtool -e eth0
> [..]
> ioctl(3, SIOCETHTOOL, 0x7fff0b702e80)   =3D -1 ESRCH (No such process)
>=20
> Also, I just noticed that the firmware-version field is giving
> different answers:
>=20
> bertha1{root}/tmp#ethtool -i eth0
> driver: e1000
> version: 7.1.9-k4
> firmware-version: 15.255-15
> bus-info: 0000:00:19.0

super odd

> bertha1{root}/tmp#exit
> exit
> bertha1{jgg}/tmp#sudo ethtool -i eth0
> driver: e1000
> version: 7.1.9-k4
> firmware-version: 0.0-0
> bus-info: 0000:00:19.0

ditto
=20
> So, I guess e1000_read_eeprom is failing with -1 - there is no error
> check at the call in e1000_get_drvinfo so firmware-version could be
> stack garbage.

good point, we'll look to that.
=20
> There is nothing is dmesg to give a clue. I'm guessing building
> the driver with a debug flag might help?

unfortunately the driver doesn't have much that can help us in its debug
code at this point.

> Is the BIOS is involved in this flash access somehow? Maybe the
> updated ethernet/management support modules in the Sep 25 bios will
> help. Also, I think I have the PXE module enabled, I was using it
> earlier..
>=20
>> I think the eeprom thing above is the best bet so far.  while
>> debugging, you may also want to try disabling active management
>> (iAMT) in the bios if is has that option.
>=20
> I'll see if it has that option and try it..

any luck?  I am expecting at the very least the BIOS changed behavior.

Jesse