2003-05-20 17:34:54

by Downing, Thomas

[permalink] [raw]
Subject: [Patch] e100 driver patch for 2.5 - option to restore old behavio r

These patches add a module parameter to restore older
EEPROM behavior to the Intel e100 NIC driver.

In 2.4.x (and early 2.5.x) the e100 driver would print
a warning if there was an error with the EEPROM checksum,
but would continue with device initialization.

Sometime during 2.5.x, the behavior changed, so that any
EEPROM checksum error fails the driver initialization.

We have lots of boxen that for some reason always fail the
checksum, but the E100 NIC operates just fine. Others might
have the same issue.

So here is are two patches that adds a new module parameter
'PromXsumFail' as boolean to the E100 module. The default
is 'true', so that the behavior is the same as the unpatched
version. The patch is against 2.5.69. The first patch file
'patch-e100-src-2.5.69' is for the module source. The second
file 'patch-e100-doc-2.5.69' is for the e100.txt file in the
Documentation directory.

Patches have been tested on:

1. Custom netbooted P3 PIIX4 with embedded 82559.
2. Dell PC - Intel chipset, Pro/100+ PCI, UP P3.
2. Dell PC - Intel chipset, Pro/100+ PCI, SMP P4.

If you think it useful, please apply.

Thanks
Thomas Downing
Principal Member Technical Staff
IPC Information Systems, Inc.
777 Commerce Drive
Fairfield, CT, USA 06432-5500
voice: 203.339.7030


Attachments:
patch-e100-doc-2.5.69 (578.00 B)
patch-e100-src-2.5.69 (3.35 kB)
Download all attachments

2003-05-20 20:47:35

by Feldman, Scott

[permalink] [raw]
Subject: RE: [Patch] e100 driver patch for 2.5 - option to restore old behavio r

> These patches add a module parameter to restore older
> EEPROM behavior to the Intel e100 NIC driver.

We want the check to fail to detect mis-programmed eeproms. The
checksum test isn't conclusive, but it does provide an indication
something is wrong.

> We have lots of boxen that for some reason always fail the
> checksum, but the E100 NIC operates just fine. Others might
> have the same issue.

The assumption is that the eeprom was programmed correctly in
manufacturing, so your nics were 1) mis-programmed during manufacturing,
or 2) modified later using some tool. Follow-up with
[email protected] to make sure 1) isn't the case.

-scott