Red Hat Bugzilla – Bug 121865
Failure to recognize new version of Intel Pro/1000 (part PWLA8391MT) NIC
Last modified: 2007-11-30 17:10:41 EST
From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; X11; Linux i686) Opera
Description of problem:
Presence of recent Intel Pro/1000 MT card (Intel part PWLA8391MT,
with 82541 chip) causes Anaconda (and the kernel in general) to not
see either that particular NIC card or any of the built-in NICs that
have the same or Intel chipsets (e.g. 82546EB.)
Intel has reved the Pro/1000 to use a new chipset (82541). Things
work fine if the new-version of the Pro/1000 is removed from the
machine and the older version, Intel part number PWLA8390MT (with
82540 chipset), used in its place. It is only the new cards that
cause the system to fail to recognize *any* of the Intel NICs in the
In Anaconda this manifests itself by an inability to do kickstart/NFS
This problem affects Anaconda during the install as well as the full
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Install recent version of Intel Pro/1000MT (Part PWLA8391MT)
2. Try to install Fedora Core 2 test 3 from CD with command line of
Actual Results: Rather than getting to the screen that asks which of
my NIC's to use (which is what normally happens when the older Intel
NICs are installed) no network interfaces are found and Andaconda
goes forward and asks for languages etc, and eventually asks for a
install medium. If the user enters "nfs" and then selects driver
e1000 the machine comes back and says there is no matching hardware.
Expected Results: Anaconda should have asked me to pick one of my
NICs and then begun the kickstart.
Platform is a Supermicro Motherboard X5DPR-IG2+, dual Xeon, 512Meg,
40gig drive dual built-in intel NICs, one additional Pro/1000MT Intel
Nic in a PCI slot.
*** Bug 121908 has been marked as a duplicate of this bug. ***
I've done more digging - this may be a due to a single bad
motherboard or PCI slot.
I obtained two other chassises - each an exact duplicate of the first
- and they worked relatively OK with the new version of the
Pro/1000MT NIC. I say "relatively OK" because Anaconda flagged the
card as a generic e1000 class card rather than being more specific.
But at least I was able to reach my kickstart server.
My suggestions are these:
- Fix anaconda so that it knows more than this card is simply a
generic. I'd say that's a low priority thing.
- Close this bug report as one of those random things caused by
cosmic rays or things that go bump in the night.
Anaconda shows which driver it will use; the e1000 driver will drive
all e1000 class hw; the driver knows about the various models.
Anything beyond that is just cosmetics.....
If it's a duff slot then yeah it's best to close this bug; but if it
comes back do not hesitate to reopen the bug.