Intel PRO/100+ Server Adapter w/ Redhat

[Donald Becker]

On Tue, 14 Sep 1999, Erik Barker wrote:

> I'm currently running a server with 2 identical Intel PRO/100+ (i82559)
> cards running on a Redhat 5.2 box with the 2.2.12 kernel.  I replaced the
> default (1.06) eepro100.c file in the kernel directory with the newest ver.
> 1.09l and recompiled fine.  I rebooted and got the following information:
> 
> eth0: Intel PCI EtherExpress Pro100 at 0xc4802000, 00:90:27:9F:A0:CE, IRQ
> 10.
>   Primary interface chip i82555 PHY #1.
...
> It looks like its finding an incorrect interface chip version because I'm
> running the i82559 (so says the box and chips), not the i82555.  Is there
> something I have to change in the driver source to force it to use the
> i82559 or is this display output simply cosmetic?

The i82555 is a transceiver chip.  The i82558 and i82559 are a i82557 and a
i82555 on a single chip, with features added over time.

> Another inconsistency is that I compiled the eepro100-diag program with the
> libmii.c file and although I needed to take down the interface to get an
> actual diagnostic, it found the card as an i82557???

They all show up as the same primary PCI ID.  That's why the driver didn't
identify the chip as a i82559, but as a generic eepro100 board.

> Here's some of the output from dmesg:
> 
> eth0: Transmit timed out: status 0050  0000 at 316120/316132 command
> 000c0000.
> eth0: Tx ring dump,  Tx queue 316132 / 316120:
..
> eth0:   3 400ca000.
> eth0:  =4 000ca000.
> eth0:   5 000ca000.
...
> eth0:   23 000ca000.
> eth0: * 24 000c0000.
> eth0:   25 000ca000.
> eth0:   26 000ca000.

This is exactly the bug reported on this list -- the chip never marks this
packet as having been transmitted.  But all subsequent packets are marked as
having been transmitted.

> eth0: Trying to restart the transmitter...

And this doesn't restart properly...

OK, I've ditched the restart attempt, and always restart the whole chip.

Please try v1.09p, which adds the following features:
  Support for CardBus and hot-swap PCI implementations
  Support for suspend, resume, power-down, power-up even for PCI cards.

http://cesdis.gsfc.nasa.gov/linux/drivers/kern-2.3/index.html
You'll need the following files:
ftp://cesdis.gsfc.nasa.gov/pub/linux/drivers/kern-2.3/eepro100.c
ftp://cesdis.gsfc.nasa.gov/pub/linux/drivers/kern-2.3/pci-netif.c
ftp://cesdis.gsfc.nasa.gov/pub/linux/drivers/kern-2.3/pci-netif.h
ftp://cesdis.gsfc.nasa.gov/pub/linux/drivers/kern-2.3/k_compat.h

Donald Becker					  becker@cesdis.gsfc.nasa.gov
USRA-CESDIS, Center of Excellence in Space Data and Information Sciences.
Code 930.5, Goddard Space Flight Center,  Greenbelt, MD.  20771
301-286-0882	     http://cesdis.gsfc.nasa.gov/people/becker/whoiam.html