NIC on Intel STL2 - bad work with eepro100 & e100

Hello.

1. Hardware:
Intel Server Board STL2.
Network: onboard NIC "Intel(R) 82559 single chip PCI LAN controller"

Software:
Linux kernel up-to version 2.4.20pre9 + arp hidden patch.

2. dmesg:

...
Intel(R) PRO/100 Network Driver - version 2.1.15-k1 Copyright (c) 2002 Intel Corporation

e100: hw init failed
e100: Failed to initialize, instance #0
...

3. cat /proc/pci

...
Bus 0, device 3, function 0:

    Ethernet controller: Intel Corp. 82557/8/9 [Ethernet Pro 100] (rev 8).
      IRQ 18.
      Master Capable.  Latency=66.  Min Gnt=8.Max Lat=56.
      Non-prefetchable 32 bit memory at 0xfb101000 [0xfb101fff].
      I/O at 0x5400 [0x543f].
      Non-prefetchable 32 bit memory at 0xfb000000 [0xfb0fffff].

...

ps. eepro100 driver work, but bad.

dmesg:
...
<4>eepro100.c:v1.09j-t 9/29/99 Donald Becker http://www.scyld.com/network/eepro100.html
<4>eepro100.c: $Revision: 1.36 $ 2000/11/17 Modified by Andrey V. Savochkin <[email protected] <6>eth0: OEM i82557/i82558 10/100 Ethernet, 00:D0:B7:B6:B2:8B, IRQ 18.

<6>  Board assembly 000000-000, Physical connectors present: RJ45
<6>  Primary interface chip i82555 PHY #1.
<6>  General self-test: passed.
<6>  Serial sub-system self-test: passed.
<6>  Internal registers self-test: passed.
<6>  ROM checksum self-test: passed (0x04f4518b).

...
...
eepro100: wait_for_cmd_done timeout!
eepro100: wait_for_cmd_done timeout!
eepro100: wait_for_cmd_done timeout!
eepro100: wait_for_cmd_done timeout!
eepro100: wait_for_cmd_done timeout!
eepro100: wait_for_cmd_done timeout!
...

And periodically freeze.

bye.
Andrey Nekrasov, SpyLOG.

Replying to Andrey Nekrasov:

I have almost the same behavior here except that eepro100 works without freezes. Maybe spylog servers' NIC load is much more higher than mine :)

Intel support guys advised me to test both versions of e100 driver.

ftp://aiedownload.intel.com/df-support/2896/eng/e100-2.1.15.tar.gz ftp://aiedownload.intel.com/df-support/2896/eng/e100-1.8.38.tar.gz

Unfortunately, I haven't tested it yet. My investigation shows that the following fragment of e100_main.c code fails:

<code from="e100_hw_init">

     if (!e100_wait_exec_cmplx(bdp, 0, SCB_RUC_LOAD_BASE))
            return false;

</code>

we returning false here.

And I am stalled for now till the end of next week until the ACM ICPC Qfinal is over :(

Paul P 'Stingray' Komkoff 'Greatest' Jr /// (icq)23200764 /// (http)stingr.net
  When you're invisible, the only one really watching you is you (my keychain)

I believe the problem is due to the sleep mode bit,
I think the errata of Intel-STL2 documents it. We had a similar problem with our

STL2 based board,
clearing the sleep mode fixed it.