Can't get i82557 working with 2.4.20

"Francesco Vivoli" <[email protected]> Fri, 28 Mar 2003 21:22:02 +0100
Newsgroups gmane.linux.drivers.eepro100.devel
Message-ID <001301c2f567$b1c130e0$0500a8c0@OBI5>
Hi there.
I have a problem here and hope someone can help me:)
I have a server shipped with two intel NIC recognized by a 2.2.20 kernel as:
___
eth1: OEM i82557/i82558 10/100 Ethernet, 00:20:ED:14:BA:87, IRQ 10.
Mar 26 12:50:52 sql kernel:   Board assembly 727095-004, Physical connectors
present: RJ45
Mar 26 12:50:52 sql kernel:   Primary interface chip i82555 PHY #1.
----
and in /proc/pci
___
 Ethernet controller: Intel 82557 (rev 8).
-----

The problem that I can't manage upgrading to any version of 2.4 (actually
2.4.20) using the shipped dirvers (eepro100, e100,the story is becoming a
bit confusing about which one sould be the right one)

What I have done is:
-downloaded netdrivers and compiled it.
-after the reboot the interfaces are up andf running but I notice these
messages in the log:
_______
eth1: OEM Intel i82559 rev 8 at 0xe0835000, 00:20:ED:14:BB:07, IRQ 26.
Mar 28 11:33:38 mail kernel:   Board assembly 727095-004, Physical
connectors present: RJ45
Mar 28 11:33:38 mail kernel:   Primary interface chip i82555 PHY #1.
Mar 28 11:33:38 mail kernel:
Mar 28 11:33:38 mail kernel:   ** The configuration EEPROM enables Sleep
Mode.
Mar 28 11:33:38 mail kernel:
Mar 28 11:33:38 mail kernel:   ** This will cause PCI bus errors!
Mar 28 11:33:38 mail kernel:   ** Update the configuration EEPROM with the
eepro100-diag program.
----------
Reading some post I have figured out the problem but not how to correct
this; I have to use eepro100-diag but I don't know the parameters (being
this a live system I can't afford-programmed EEPROM)...

Beside this error I get messages
IRQ 26 is physically blocked! (2050)Failing back to low-rate polling. The
effect is a bit strange: I can ping from the box but not to the box...
I have read that in this case disabling APIC support in the kernel should
work, it's true? And at what cost?

Sorry for the long post
Thanks all,
Francesco Vivoli