kernel-packages team mailing list archive

[Launchpad Bug Tracker <1315615@xxxxxxxxxxxxxxxxxx>]

You have been subscribed to a public bug:

Dmesg part of the crash:

May  2 22:40:28 fenrir1 kernel: [1496015.997789] irq 48: nobody cared (try booting with the "irqpoll" option)
May  2 22:40:28 fenrir1 kernel: [1496015.997853] Pid: 973, comm: irq/48-eth0 Tainted: G        W    3.2.0-60-lowlatency-pae #62-Ubuntu
May  2 22:40:28 fenrir1 kernel: [1496015.997854] Call Trace:
May  2 22:40:28 fenrir1 kernel: [1496015.997859]  [<c15a32f5>] ? printk+0x2d/0x2f
May  2 22:40:28 fenrir1 kernel: [1496015.997863]  [<c10c4aa9>] __report_bad_irq+0x29/0xd0
May  2 22:40:28 fenrir1 kernel: [1496015.997865]  [<c10c4e64>] note_interrupt+0x104/0x150
May  2 22:40:28 fenrir1 kernel: [1496015.997868]  [<c10c38b8>] ? irq_forced_thread_fn+0x38/0x50
May  2 22:40:28 fenrir1 kernel: [1496015.997870]  [<c10c3817>] irq_thread+0x1b7/0x1e0
May  2 22:40:28 fenrir1 kernel: [1496015.997872]  [<c10c3880>] ? irq_thread_fn+0x40/0x40
May  2 22:40:28 fenrir1 kernel: [1496015.997874]  [<c10c3660>] ? irq_finalize_oneshot+0x20/0x20
May  2 22:40:28 fenrir1 kernel: [1496015.997877]  [<c107c99d>] kthread+0x6d/0x80
May  2 22:40:28 fenrir1 kernel: [1496015.997879]  [<c107c930>] ? kthread_worker_fn+0x160/0x160
May  2 22:40:28 fenrir1 kernel: [1496015.997882]  [<c15c05be>] kernel_thread_helper+0x6/0x10
May  2 22:40:28 fenrir1 kernel: [1496015.997883] handlers:
May  2 22:40:28 fenrir1 kernel: [1496015.997938] [<c10c2eb0>] irq_default_primary_handler threaded [<f84cd760>] e1000_msix_other
May  2 22:40:28 fenrir1 kernel: [1496015.998108] Disabling IRQ #48
May  2 22:40:30 fenrir1 kernel: [1496017.778701] e1000e 0000:02:00.0: eth0: Reset adapter

Server is a HP Proliant DL120G6 with 2 integrated gigabit nic which uses e1000e driver for "Intel Corporation 82574L Gigabit Network Connection".
This happened in kernel 3.2.0-60-lowlatency. Before that i tried 3.2.0-33-generic-pae and similar thing happened. So i'm quite sure it exists in all kernel versions.
I tried to manually compile e1000e driver from sourceforge but that just did work terribly (didn't crash just didn't work well).
We use this machine as pppoe concetrator and important thing is that this crash never happened on the interface that is configured as link (have an ip) but only on interface that is setup as pppoe server (kernel mode) so it maybe related to this.
When this happens eth0 interface is down and as you can imagine all pppoe connections are lost while eth1 works just normally.
I am quite aware this will be very hard to reproduce and this doesn't happen
right away - server can work for months before crashing.

ProblemType: Bug
DistroRelease: Ubuntu 12.04
Package: linux-image-3.2.0-60-lowlatency-pae 3.2.0-60.62
ProcVersionSignature: Ubuntu 3.2.0-60.62-lowlatency-pae 3.2.55
Uname: Linux 3.2.0-60-lowlatency-pae i686
ApportVersion: 2.0.1-0ubuntu17.6
Architecture: i386
Date: Sat May  3 09:10:58 2014
MarkForUpload: True
ProcEnviron:
 LANGUAGE=en_US:en
 TERM=xterm
 PATH=(custom, no user)
 LANG=en_US.UTF-8
 SHELL=/bin/bash
SourcePackage: linux-lowlatency
UpgradeStatus: No upgrade log present (probably fresh install)

** Affects: linux (Ubuntu)
     Importance: Low
         Status: Incomplete


** Tags: apport-bug i386 precise
-- 
nic driver/irq crash
https://bugs.launchpad.net/bugs/1315615
You received this bug notification because you are a member of Kernel Packages, which is subscribed to linux in Ubuntu.