rhel4 detects NICs in reverse on a 2950

[Matt Domsch]

On Wed, Sep 06, 2006 at 09:51:20AM -0700, Simon Gao wrote:
> Anyone knows when Dell is going to fix this problem permanently, instead 
> of just workaround?

It's not clear what would constitute a "permanent fix" vs a
"workaround". I'm open to suggestions. :-)  

What I'm investigating, and have some code I've been writing the last
few days, uses udev rules to ask the BIOS for what name it uses for a
particular device.  Then a udev PROGRAM="" rule can name the devices
"properly" (for BIOS's idea of "properly").  This is similar in
concept to what my name_eths program does, except that name_eths runs
a little too late (after udev has already assigned the 'wrong' name).

There are rumors of a future ACPI extension to provide this mapping
even more explicitly, but I haven't yet received any details about
that yet.  This would in theory use a udev PROGRAM="" method too.

Other ideas?  (Note, PCI Express, being a chip-internal switch rather than
physical bus topology, means we can't just move wires during board
layout to solve this trivially, and there are PCI bus device usage
considerations which is what causes our current boards to be laid out
as they are).  I don't want to add additional constraints to the board
layouts if I can avoid it (having been in that seat before, I know how
tricky layout can be already).

Thanks,
Matt

-- 
Matt Domsch
Software Architect
Dell Linux Solutions linux.dell.com & www.dell.com/linux
Linux on Dell mailing lists @ http://lists.us.dell.com