e1000 driver loaded for PCI 8086:10b9 instead of e1000e

[WhoCares?]

On my desktop machine using Ubuntu 8.04 Hardy Heron, the installed PCIe Gigabit NIC Intel PRO/1000e using the 82572EI chipset is incorrectly identified as a 824xx type card. Thus, the standard e1000.ko module gets loaded instead of the e1000e.ko. Also, using the shipped e1000e module didn't give me network connectivity whereas the latest revision from e1000.sf.net did.

[Brian Murray (brian-murray)]

Thank you for taking the time to report this bug and helping to make Ubuntu better. Could you please add the PCI ID of your network adapter to the bug report? You can obtain it via the command 'sudo lspci -vvnn'. Thanks in advance.

[WhoCares? (stefan-whocares)]

Of course. Sorry, should have thought about that right away.

03:00.0 Ethernet controller [0200]: Intel Corporation 82572EI Gigabit Ethernet Controller (Copper) [8086:10b9] (rev 06)
 Subsystem: Intel Corporation PRO/1000 PT Desktop Adapter [8086:1083]
 Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
 Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR-
 Latency: 0, Cache Line Size: 4 bytes
 Interrupt: pin A routed to IRQ 28
 Region 0: Memory at dfee0000 (32-bit, non-prefetchable) [size=128K]
 Region 1: Memory at dfec0000 (32-bit, non-prefetchable) [size=128K]
 Region 2: I/O ports at bc00 [size=32]
 [virtual] Expansion ROM at dfd00000 [disabled] [size=128K]
 Capabilities: [c8] Power Management version 2
  Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA PME(D0+,D1-,D2-,D3hot+,D3cold+)
  Status: D0 PME-Enable- DSel=0 DScale=1 PME-
 Capabilities: [d0] Message Signalled Interrupts: Mask- 64bit+ Queue=0/0 Enable-
  Address: 00000000fee0300c Data: 41e9
 Capabilities: [e0] Express Endpoint IRQ 0
  Device: Supported: MaxPayload 256 bytes, PhantFunc 0, ExtTag-
  Device: Latency L0s <512ns, L1 <64us
  Device: AtnBtn- AtnInd- PwrInd-
  Device: Errors: Correctable- Non-Fatal- Fatal- Unsupported-
  Device: RlxdOrd+ ExtTag- PhantFunc- AuxPwr- NoSnoop+
  Device: MaxPayload 256 bytes, MaxReadReq 512 bytes
  Link: Supported Speed 2.5Gb/s, Width x1, ASPM L0s, Port 1
  Link: Latency L0s <4us, L1 <64us
  Link: ASPM Disabled RCB 64 bytes CommClk- ExtSynch-
  Link: Speed 2.5Gb/s, Width x1

[David Burgess (apt-get)]

Same problem here. Causes sine-wave-shaped network history graph in gnome system monitor at any speed and poor thin client performance when affected interface is used on client-facing interface of ltsp server. Blacklisting the e1000 driver corrected both problems.

$uname -r
2.6.24-16-generic

http://ubuntuforums.org/showthread.php?t=750910
http://marc.info/?l=ltsp-discuss&m=120900833431608&w=2

db

[Leann Ogasawara (leannogasawara)]

Hi David,

I just want to get some clarification. WhoCares, the original bug reporter, mentioned he needed the latest e1000e driver from sf.net to get things working. However you mentioned you just need to blacklist the e1000 and then indeed the e1000e driver shipped in Ubuntu's Hardy 2.6.24-16.30 kernel works for you and you did not need the latest version from sf.net?

Thanks.

[WhoCares? (stefan-whocares)]

Hi Leann,

that can be my fault and there's a chance that the driver shipping with Ubuntu is actually working. I might have to reinstall tonight anyway so I may be able to test this later. In that case I'll report back with more info.

-Stefan

[David Burgess (apt-get)]

I'm afraid my original description was incomplete. In fact I
downloaded the latest e1000e driver from sf.net before blacklisting
e1000, so I didn't try the shipping version of e1000e and so I don't
know whether would work with the hardware in question or not.

db

On Thu, May 1, 2008 at 2:36 PM, WhoCares? <email address hidden> wrote:
> Hi Leann,
>
> that can be my fault and there's a chance that the driver shipping with
> Ubuntu is actually working. I might have to reinstall tonight anyway so
> I may be able to test this later. In that case I'll report back with
> more info.
>
> -Stefan
>
>
>
> --
> e1000 driver loaded for PCI 8086:10b9 instead of e1000e
> https://bugs.launchpad.net/bugs/222831
> You received this bug notification because you are a direct subscriber
> of the bug.
>

[Leann Ogasawara (leannogasawara)]

Thanks for the updates guys. Definitely if you can test and verify if the e1000e driver shipped in Ubuntu Hardy does or does not work for you that would help determine what additional fixes we'll need to incorporate. Thanks.

[WhoCares? (stefan-whocares)]

Leann,
Sorry for being late to respond but private matters kept me from answering earlier.
I've now confirmed that the e1000e.ko shipping with Ubuntu 8.04 Hardy *DOESN'T* work for me, whereas the latest driver from e1000.sf.net does.
Log excerpt:

May 18 10:41:10 ubuntu-ws kernel: [ 348.097371] e1000e: Intel(R) PRO/1000 Network Driver - 0.2.0
May 18 10:41:10 ubuntu-ws kernel: [ 348.097378] e1000e: Copyright (c) 1999-2007 Intel Corporation.

This is all that will be in the logs after loading the e1000e.ko shipped with Ubuntu. After compiling version 0.2.9.5 from e1000.sf.net and manually loading the new kernel module, the log looks like this:

May 18 10:45:29 ubuntu-ws kernel: [ 606.857687] e1000e: Intel(R) PRO/1000 Network Driver - 0.2.9.5
May 18 10:45:29 ubuntu-ws kernel: [ 606.857693] e1000e: Copyright (c) 1999-2008 Intel Corporation.
May 18 10:45:29 ubuntu-ws kernel: [ 606.857789] ACPI: PCI Interrupt 0000:03:00.0[A] -> GSI 28 (level, low) -> IRQ 26
May 18 10:45:29 ubuntu-ws kernel: [ 607.034355] 0000:03:00.0: eth0: (PCI Express:2.5GB/s:Width x1) 00:1b:21:00:9e:f5
May 18 10:45:29 ubuntu-ws kernel: [ 607.034361] 0000:03:00.0: eth0: Intel(R) PRO/1000 Network Connection
May 18 10:45:29 ubuntu-ws kernel: [ 607.034441] 0000:03:00.0: eth0: MAC: 1, PHY: 4, PBA No: d50854-003
May 18 10:45:29 ubuntu-ws kernel: [ 607.238827] 0000:03:00.0: eth0: MSI interrupt test failed!
May 18 10:45:29 ubuntu-ws kernel: [ 607.238965] 0000:03:00.0: eth0: MSI interrupt test failed, using legacy interrupt.
May 18 10:45:29 ubuntu-ws kernel: [ 607.242007] ADDRCONF(NETDEV_UP): eth0: link is not ready
May 18 10:45:32 ubuntu-ws kernel: [ 610.070675] 0000:03:00.0: eth0: Link is Up 1000 Mbps Full Duplex, Flow Control: RX/TX
May 18 10:45:32 ubuntu-ws kernel: [ 610.072973] ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready

My best guess is that the e1000e.ko version 0.2.0 doesn't know my card.

[Tim Gardner (timg-tpi)]

Please keep an eye on upstream. I'll cherry-pick it as soon as support for this PCI ID appears.

[David Burgess (apt-get)]

I updated my kernel to 2.6.24-17-generic today and got the same result as WhoCares? with e1000e version 0.2.0, i.e., it didn't work.

db

[Tim Gardner (timg-tpi)]

Support for Intel Corporation 82572EI Gigabit Ethernet Controller (Copper) [8086:10b9] appeared in LUM -19.27.

[C Pirnat (histoplasmosis)]

It appears this bug still persists. I'm running Hardy 64bit and experience this behavior. I tried blacklisting the e1000 module but it still loaded. I then tried sudo rmmod e1000 and sudo modprobe e1000e and it loads although I get device errors trying to ifup the interface.

histo@xw8200:~$ uname -a
Linux xw8200 2.6.24-19-generic #1 SMP Fri Jul 11 21:01:46 UTC 2008 x86_64 GNU/Linux

Relevant output of sudo lspci -vvnn

10:03.0 Ethernet controller [0200]: Intel Corporation 82541GI Gigabit Ethernet Controller [8086:1076] (rev 05)
 Subsystem: Hewlett-Packard Company Unknown device [103c:12f1]
 Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B-
 Status: Cap+ 66MHz+ UDF- FastB2B- ParErr- DEVSEL=medium >TAbort- <TAbort- <MAbort- >SERR- <PERR-
 Latency: 64 (63750ns min), Cache Line Size: 64 bytes
 Interrupt: pin A routed to IRQ 17
 Region 0: Memory at d1100000 (32-bit, non-prefetchable) [size=128K]
 Region 1: Memory at d1120000 (32-bit, non-prefetchable) [size=128K]
 Region 2: I/O ports at 4000 [size=64]
 [virtual] Expansion ROM at d1160000 [disabled] [size=128K]
 Capabilities: [dc] Power Management version 2
  Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA PME(D0+,D1-,D2-,D3hot+,D3cold+)
  Status: D0 PME-Enable- DSel=0 DScale=1 PME-
 Capabilities: [e4] PCI-X non-bridge device
  Command: DPERE- ERO+ RBC=512 OST=1
  Status: Dev=00:00.0 64bit- 133MHz- SCD- USC- DC=simple DMMRBC=2048 DMOST=1 DMCRS=8 RSCEM- 266MHz- 533MHz-

[Aleksandr Koltsoff (czr)]

Running 2.6.24-19-generic on x86_64 with 82572EI (-17-generic didn't work either):
e1000 recognizes the PCI ID but does not detect link.
e1000e doesn't even recognize the chip nor register the ethdev (ifconfig -a doesn't show the device, with link or not).

uname -a:
Linux backup 2.6.24-19-generic #1 SMP Wed Aug 20 17:53:40 UTC 2008 x86_64 GNU/Linux

lspci -vvnn (for the PCIe NIC):
02:00.0 Ethernet controller [0200]: Intel Corporation 82572EI Gigabit Ethernet Controller (Copper) [8086:10b9] (rev 06)
        Subsystem: Intel Corporation PRO/1000 PT Desktop Adapter [8086:1083]
        Control: I/O+ Mem+ BusMaster- SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
        Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR-
        Interrupt: pin A routed to IRQ 5
        Region 0: Memory at fdee0000 (32-bit, non-prefetchable) [size=128K]
        Region 1: Memory at fdec0000 (32-bit, non-prefetchable) [size=128K]
        Region 2: I/O ports at 8c00 [size=32]
        Expansion ROM at fdea0000 [disabled] [size=128K]
        Capabilities: [c8] Power Management version 2
                Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA PME(D0+,D1-,D2-,D3hot+,D3cold+)
                Status: D0 PME-Enable- DSel=0 DScale=1 PME-
        Capabilities: [d0] Message Signalled Interrupts: Mask- 64bit+ Queue=0/0 Enable-
                Address: 0000000000000000 Data: 0000
        Capabilities: [e0] Express Endpoint IRQ 0
                Device: Supported: MaxPayload 256 bytes, PhantFunc 0, ExtTag-
                Device: Latency L0s <512ns, L1 <64us
                Device: AtnBtn- AtnInd- PwrInd-
                Device: Errors: Correctable- Non-Fatal- Fatal- Unsupported-
                Device: RlxdOrd+ ExtTag- PhantFunc- AuxPwr- NoSnoop+
                Device: MaxPayload 128 bytes, MaxReadReq 512 bytes
                Link: Supported Speed 2.5Gb/s, Width x1, ASPM L0s, Port 1
                Link: Latency L0s <4us, L1 <64us
                Link: ASPM Disabled RCB 64 bytes CommClk+ ExtSynch-
                Link: Speed 2.5Gb/s, Width x1

Somewhat ironically, the kernel from previous Ubuntu release worked with the e1000 driver without a glitch (other than the watchdog thread updating statistics less than per second leading to the "sinusoidal" network utilization problem, but that's not really an issue here).

[Martin Lindhe (martinlindhe)]

This bug is tagged as "Fix released", however I just had the same problem with Hardy Heron fully updated just last night.
I got it working by blacklisting "e1000" driver and installing e1000e driver 0.4.1.7 from http://sourceforge.net/projects/e1000

So I ask for this bug to be reopened. Or does it mean it's fixed in Interprid?

[WhoCares? (stefan-whocares)]

Well, since I opened this bug report, I'd like that add that the latest release of the e1000e driver (currently 0.4.1.7) again breaks the support for my network card. The older version 0.2.9.5 works like a charm however. This makes it quite hard to really fix this problem in whatever release of Ubuntu because chances are that some or other cards will remain "broken" no matter what driver release is going to be used. Would be nice if the Intel guys could fix up their end of the mess first.

[Martin Lindhe (martinlindhe)]

I apology if this is off-topic, but there are reports that the e1000e driver together with newer kernels (2.6.27-rc) can damage the hardware.
So if you hit this bug, I would advise you against trying out Intrepid Ibex.

http://www.heise-online.co.uk/news/Intel-e1000e-user-Don-t-install-the-new-SUSE-betas--/111583
http://bugzilla.kernel.org/show_bug.cgi?id=11382

[David Burgess (apt-get)]

This made it onto slashdot this morning.

http://linux.slashdot.org/article.pl?sid=08/09/23/133258

db

[David Burgess (apt-get)]

I did a fresh install of 8.04.1 x86_64 this week and the included e1000e driver (0.2.0) is still broken on this NIC (before and after updating linux-image through apt). I downloaded and compiled Intel.com's latest e1000e (0.4.1.7-NAPI) and that works.

[gcc (chris+ubuntu-qwirx)]

I have a machine with an Intel 82567V-2 (ICH10) onboard gigabit ethernet, and Hardy does not detect and cannot use this card even with the latest released kernel (2.6.24-24). I think this is the same problem as above, i.e. that the e1000e driver just does not recognise the PCI IDs for the device, and that if it did, it would work.

Loading the stock e1000 and e1000e drivers only shows the banner, nothing else:

Jun 5 10:55:47 pc1 kernel: [ 4367.997475] Intel(R) PRO/1000 Network Driver - version 7.3.20-k2-NAPI
Jun 5 10:55:47 pc1 kernel: [ 4367.997481] Copyright (c) 1999-2006 Intel Corporation.
Jun 5 10:55:48 pc1 kernel: [ 4369.070605] e1000e: Intel(R) PRO/1000 Network Driver - 0.2.0
Jun 5 10:55:48 pc1 kernel: [ 4369.070612] e1000e: Copyright (c) 1999-2007 Intel Corporation.

and no ethernet interface is created.

lspci -vvnn:

00:19.0 Ethernet controller [0200]: Intel Corporation 82567V-2 Gigabit Network Connection [8086:10ce]
 Subsystem: Hewlett-Packard Company Unknown device [103c:2a84]
 Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B-
 Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR-
 Latency: 0
 Interrupt: pin A routed to IRQ 3
 Region 0: Memory at fe9c0000 (32-bit, non-prefetchable) [size=128K]
 Region 1: Memory at fe9fd000 (32-bit, non-prefetchable) [size=4K]
 Region 2: I/O ports at d080 [size=32]
 Capabilities: [c8] Power Management version 2
  Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA PME(D0+,D1-,D2-,D3hot+,D3cold+)
  Status: D0 PME-Enable- DSel=0 DScale=1 PME-
 Capabilities: [d0] Message Signalled Interrupts: Mask- 64bit+ Queue=0/0 Enable-
  Address: 0000000000000000 Data: 0000
 Capabilities: [e0] #13 [0306]

It seems that the latest Intrepid kernels have support for this driver. It should work with e1000e as long as the correct PCI IDs are added.

May be related to:

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/222042
https://bugs.launchpad.net/debian/+bug/322737/+viewstatus
https://bugs.launchpad.net/bugs/352414
http://lists.debian.org/debian-boot/2009/02/msg00245.html

Unfortunately I don't know enough ubuntu-fu to work out which intrepid patch added these PCI IDs.