Stepping into the world of kernel oops debugging

kushalkoolwal · 12-29-2010, 08:10 PM

I have been using Linux for over 5 years now and I have always wanted to start learning about debugging kernel oops. Today upon using the Debian current Backport kernel on my PC, I got the following kernel oops:

Code:

[   24.155180] irq 17: nobody cared (try booting with the "irqpoll" option)
[   24.155284] Pid: 0, comm: swapper Not tainted 2.6.32-bpo.5-686 #1
[   24.155373] Call Trace:
[   24.155463]  [<c106cf75>] ? __report_bad_irq+0x24/0x69
[   24.155554]  [<c106cf7c>] ? __report_bad_irq+0x2b/0x69
[   24.155644]  [<c106d0a1>] ? note_interrupt+0xe7/0x13e
[   24.155734]  [<c106d5cf>] ? handle_fasteoi_irq+0x7a/0x97
[   24.155827]  [<c1004dd7>] ? handle_irq+0x17/0x1b
[   24.155915]  [<c1004659>] ? do_IRQ+0x38/0x89
[   24.156002]  [<c10037f0>] ? common_interrupt+0x30/0x38
[   24.156096]  [<c10400d8>] ? ftrace_raw_output_workqueue_insertion+0x80/0x8c
[   24.156192]  [<c104f06e>] ? tick_nohz_stop_sched_tick+0x34a/0x36e
[   24.156285]  [<c1002367>] ? cpu_idle+0x67/0xa4
[   24.156375]  [<c13bf7fc>] ? start_kernel+0x318/0x31d
[   24.156461] handlers:
[   24.156539] [<f7cc3855>] (usb_hcd_irq+0x0/0x71 [usbcore])
[   24.156755] Disabling IRQ #17

The system boots fine to the GNOME environment and everything seems to working (at least to the naked eye). I think this is a perfect opportunity to start learning how to do oops debugging. I goggled about this but most of the links that I found are either very old (for 2.4) or are incomplete.

If somebody can guide me in this process, I will really appreciate that.

Thanks

syg00 · 12-29-2010, 10:47 PM

Is that all of it ?. That doesn't look like an oops.
Just looks like tourist information in the log for you. Maybe an interrupt handler didn't register (properly).

Edit: I don't think google trawls lkml - have a look at this for some good info. Linus has also made a post re his technique a couple of years later - there is also a heap of files under ../Documentation in the source tree.

kushalkoolwal · 12-30-2010, 01:38 AM

Yes that's the only questionable out in the entire dmesg. If you want I can post the entire dmesg output.

For some reasons I always thought that messages like these are oops but after reading your reply it seems that it further needs to have some critical piece of information in order to qualify as oops, correct?

Out of curiosity, what kind of message exactly qualifies as oops? Is there an actual "oops" string somewhere in the dmesg that indicates that this is a kernel oops?

syg00 · 12-30-2010, 02:00 AM

Oh yeah - did you check out that link I posted ?.

kushalkoolwal · 12-30-2010, 12:04 PM

Got it now.

In any case is there a technique to debug the anomaly or as you said "tourist information" that I am seeing in my dmesg logs?