E1000 driver for 82562V-2 (8086:10c0) device

[Michael Morgan]

Binary package hint: linux-source-2.6.20

Dell is awaiting a back-port for the E1000 driver for Feisty.

The driver at version 7.5.5 works fine, however, includes too many changes
to be moved to a previous ABI.

[Ben Collins (ben-collins)]

Packages for the FI can be found here. The meta package linux-backports-modules-generic will ensure that upgrades work smoothly. Upgrades to gutsy, or new installs of gutsy will work without problem. Non factory installs of feisty will require the user to install the linux-backports-modules-generic package in order to get working e1000.

This package will be uploaded to feisty-proposed, and later pulled into feisty-updates, and built against latest kernel source.

http://people.ubuntu.com/~bcollins/e1000-ich9/

[Michael Morgan (mikmorg)]

Ben,
I tested this new driver on our FI, and it seems to be functional. Thank you, and good work!!

I hate to request this, given that you have done so much work so quickly, but it seems we may
be forced into shipping with the 2.6.20-16 kernel. For this case, could we get the same back-port
against the -16 kernel?

[Michael Morgan (mikmorg)]

The supplied backport gives this error in certain actions, such as loading and running dhclient:

e1000_request_irq: Unable to allocate MSI interrupt
Error: -22

It still functions perfectly, however this wouldn't be the best user experience.

[Michael Morgan (mikmorg)]

In response to my own submission, the submitted backport actually works perfectly in the -16 kernel, as well as the -15 kernel.
Apparently the ABI did not change enough to break anything. However, I still think it would be best to get the -16 version
back-ported if possible.

[Michael Morgan (mikmorg)]

Please ignore my last post.. I was looking at the wrong terminal.

[Ben Collins (ben-collins)]

New -16 based package is at the same URL.

[Ben Collins (ben-collins)]

I found out why this message occurs. We disabled MSI by default in feisty because of too many problems. So the card reports it supports MSI, but the allocation of the MSI fails since the kernel disabled it. If a user wants, they can pass "msi" on the kernel command line to enable it.

[Amit Bhutani (amit.bhutani)]

Ben- Will this latest provided package also lower the printk on the MSI error message to KERN_INFO ??

> -----Original Message-----
> From: <email address hidden> [mailto:<email address hidden>] On Behalf Of
> Ben Collins
> Sent: Thursday, June 21, 2007 1:14 PM
> To: WW Linux Engineering
> Subject: [Bug 121187] Re: E1000 driver for 82562V-2 (8086:10c0) device
>
> New -16 based package is at the same URL.
>
> --
> E1000 driver for 82562V-2 (8086:10c0) device
> https://bugs.launchpad.net/bugs/121187
> You received this bug notification because you are a direct subscriber
> of the bug.

[Michael Morgan (mikmorg)]

Testing -16 today

[Michael Morgan (mikmorg)]

e1000 confirmed on -16. Thanks!!

[Henrik Nilsen Omma (henrik)]

We are introducing a series a new teams under the canonical-* namespace that only have Canonical employees as members. Examples are canonical-kernel-team and canonical-support. This is especially important WRT some bugs marked private.

The distro-qa team is being renamed canonical-qa and the subscribed bugs are being transferred. When subscribing the Canonical Ubuntu QA team to bugs in the future, please make sure to use 'canonical-qa'.

Sorry for the bug mail noise.

[Ben Collins (ben-collins)]

linux-backports-modules-2.6.20 is currently processing through feisty-proposed -> feisty-updates.

[Martin Pitt (pitti)]

In gutsy this is included in l-u-m.

[Martin Pitt (pitti)]

Accepted into feisty-proposed. Please go ahead with QA testing.

[Ben Collins (ben-collins)]

linux-backports-modules-2.6.20 (2.6.20-16.11) feisty-proposed; urgency=low

  * ABI bump
  * Add e1000-ich9 driver
    LP: #121187

    This driver adds three new PCI device IDs for ICH9 mac types. It contains
    only these PCI IDs, which do not exist in the primary e1000 driver found
    in our kernel package. So there are no conflicts, just new hardware
    support.

  * Add actual build infrastructure (not just an empty package anymore).

 -- Ben Collins <email address hidden> Wed, 20 Jun 2007 09:10:39 -0400

[Michael Vogt (mvo)]

I do not have the required hardware to do a sru verfication. We need help from the community or the hardware verification lab.

[Tim Gardner (timg-tpi)]

I can verify that the e1000-ich9 module works.

[Martin Pitt (pitti)]

Thanks, Tim. I marked this as verification-done then.

[Martin Pitt (pitti)]

Unfortunately Soyuz' copy-package.py is broken ATM, I'll consult Celso today.

[Martin Pitt (pitti)]

Ah, so the reason was that the packages were still in NEW, sorry. I just accepted them.

However, Tim, that means that you can't possibly have tested the feisty-proposed packages, since they did not exist yet. Can you please test the actual packages? Thank you!

[Tim Gardner (timg-tpi)]

Martin,

I retested by completely removing linux-backports-modules-generic*, then reinstalled from my mirror, then rebooted. I acquired the following information using ssh over ethernet.

rtg@e530-nvidia:~$ dpkg -l | grep backports
ii linux-backports-modules-2.6.20-16-generic 2.6.20-16.11 Ubuntu supplied Linux modules for version 2.
ii linux-backports-modules-generic 2.6.20.16.28.1 Backported drivers for generic kernel image

This is on a Dell Inspiron 530 with ICH9.

rtg@e530-nvidia:~$ lsmod | grep e1000
e1000_ich9 186432 0

[Martin Pitt (pitti)]

Thank you for testing the packages. verification-done, this will go to -updates on August 08 (after 7 days in -proposed).

[Martin Pitt (pitti)]

Copied to -updates.

[Martin Pitt (pitti)]

Whoops, sorry, wrong bug. This one still has to mature a bit.

[Martin Pitt (pitti)]

In feisty-updates now.

[Timothy R. Chavez (timrchavez)]

The bug task for the somerville project has been removed by an automated script. This bug has been cloned on that project and is available here: https://bugs.launchpad.net/bugs/1306045