[SOLVED] BUG: soft lockup. During boot.

paranoid times · 09-12-2011, 08:34 AM

I've been trying to figure out why my system has been occasionally powering down on me (As though it gets a cpu overheat.) In this search I found these lines in my messages file:

Code:

Sep 12 13:39:41 magicbane kernel: [   32.190267] BUG: soft lockup - CPU#0 stuck for 22s! [biosdevname:481]
Sep 12 13:39:41 magicbane kernel: [   60.247880] BUG: soft lockup - CPU#0 stuck for 23s! [biosdevname:481]

I don't believe it is related to the shutdown issue. But it does cause my system to take an abnormally long time to boot.

I've tried shutting acpi off at boot, also I read somewhere that removing the quiet option from grub helped. Seemed unlikely to me, but I tried, no effect on with either. Any other ideas?

Thanks,
Michael

H_TeXMeX_H · 09-12-2011, 10:03 AM

See:
http://bugs.centos.org/view.php?id=3582
http://muffinresearch.co.uk/archives...stuck-for-11s/

Try a different kernel version.

paranoid times · 09-12-2011, 12:55 PM

Looks like that worked. I reverted from:
2.6.40.4-5.fc15.i686
to
2.6.38.6-26.rc1.fc15.i686

And the messages are gone from the logs. Now other things are happening. But that is a story for a different post. Thank you for the assistance.

paranoid times · 10-02-2011, 09:23 AM

While the above did remove the error message, there was still a substantial pause during boot. The following appears to have removed both the error message and the pause: I was disabling devices that I wasn't using in BIOS. And after disabling the onboard LAN everything boots up right quick. I tested it a few times. With the ethernet enabled things pause during boot. Without it everything is fine. Not the best solution, but since I don't use the wired network here, it works fine for me.

H_TeXMeX_H · 10-02-2011, 09:38 AM

What onboard LAN do you have and what driver does it use ? It should be listed in 'lspci -k'. It could be a bug in the driver.

paranoid times · 10-02-2011, 09:57 AM

I think this is it:

Code:

03:00.0 Ethernet controller: Atheros Communications L2 Fast Ethernet (rev a0)
	Subsystem: ASUSTeK Computer Inc. Device 8233
	Kernel driver in use: atl2
	Kernel modules: atl2

Any way to test for bugs or the like?

H_TeXMeX_H · 10-02-2011, 10:10 AM

I have an atheros ethernet card and it does similar things, it uses the atl1c driver, which is staging. I notice that on ifconfig it sometimes reports bogus transfers, like negative numbers and such. That's a clear sign of a bug. Also, it is on IRQ 19 and it often disable this IRQ because of problems. That's another hint.

paranoid times · 10-02-2011, 10:50 AM

I might give that a try. I know on an older (Still) kernel for some older Ubuntu that I use to be running this didn't happen. So since I'm not using the ethernet at the moment I'm just going to leave it disabled and see if a future kernel won't fix it. Though thank you for the information. I suspect I will use it in the future.