Kernel panic, can't understand dmesg

HappyClown · 2015-10-24 19:35:44

My machine has been having frequent kernel panics, not at boot. I did the dmesg -w > myfile trick and I think I've found the problem, since the last five lines of the file are

[   37.016106] cfg80211: Calling CRDA to update world regulatory domain
[   40.178179] cfg80211: Exceeded CRDA call max attempts. Not calling CRDA
[  658.678092] fuse init (API version 7.23)
[ 1677.956952] xhci_hcd 0000:00:14.0: ERROR Transfer event TRB DMA ptr not part of current TD ep_index 3 comp_code 1
[ 1677.956961] xhci_hcd 0000:00:14.0: Looking for event-dma 0000000225317670 trb-start 0000000225317680 trb-end 0000000225317680 seg-start 0000000225317000 seg-end 0000000225317ff0

I don't understand what the log is saying, though. Does it have something to do with Xorg??

ukhippo · 2015-10-25 00:25:33

The “[number]” is the time stamp - seconds since boot. Not everything in dmesg is an error; some are just informational messages.
xhci_hcd has nothing to do with Xorg, it's the USB 3.0 driver.
You need to get the kernel panic listing, that will help determine whether the xhci_hcd errors are relevant or not.
As for the USB errors, do you have something plugged into your USB ports?

HappyClown · 2015-10-25 00:39:18

Yes, my sound card and a flash drive. How do I get the kernel panic listing?

ukhippo · 2015-10-25 01:36:01

Easiest way is to take a picture of the panic. If you have a serial port, you can use serialconsole logging.

HappyClown · 2015-10-25 03:24:31

I can't, the whole machine freezes up without displaying anything. All I get is the frozen image of my window manager in graphics mode when it happens. I don't have a serial port.

HappyClown · 2015-10-25 20:40:15

An interesting find, just today a new version of the LTS kernel was released. While updating pacman gave me two notes:

==> WARNING: Possibly missing firmware for module: aic94xx
==> WARNING: Possibly missing firmware for module: wd719x

Could this be the problem?

ukhippo · 2015-10-26 00:19:03

See https://bbs.archlinux.org/viewtopic.php?id=162969

Since your problem seems to be USB3 related, try determining if the problem is with
• xhci_hcd
• your USB devices
• USB power management (if used)

HappyClown · 2015-10-26 04:11:18

The only two devices that could have caused these issues (id they're USB3 related) seems to be a memory stick, FAT32 formatted or this thing: http://www.numark.com/product/mixtrack-quad. I think the only USB3 port on my machine was the one the mem stick was connected to, the rest should be USB2. I'll look into xhci_hcd and see if that's an issue, while I wait for my machine to panic again. The warnings seem fairly normal, my CPU temp is also normal.

After doing a bit of research, people are saying that the source code for xhci_hcd is terrible and that we're better off just using the USB2 module instead. Thoughts?

Last edited by HappyClown (2015-10-26 04:15:20)

Fuxino · 2015-10-26 08:26:19

HappyClown wrote:

I can't, the whole machine freezes up without displaying anything. All I get is the frozen image of my window manager in graphics mode when it happens. I don't have a serial port.

If you can trigger the kernel panic while not under X you may be able to see something printed on the screen before the system freezes.

ukhippo · 2015-10-26 12:34:27

HappyClown wrote:

After doing a bit of research, people are saying that the source code for xhci_hcd is terrible and that we're better off just using the USB2 module instead. Thoughts?

You need to be methodical in trying to determine where the issue is. Disabling USB3 in BIOS/firmware and using "legacy" mode should be one of the steps to try.

HappyClown · 2015-10-27 02:42:33

Thanks, I unplugged the USB3 device, the panic still happened. I've tried triggering it in console mode before, no luck. That's why I asked if it had something to do with X. I'll keep trying, the kernel panics about once a day, so it'll be hard to test this stuff. I don't know if my BIOS allows disabling firmware or running in legacy mode. Thanks, your suggestions are super helpful. I'd be lost without a few pointers. I guess I'll be going in all-out console mode tomorrow, partying like it's 1985 and I'm too cheap to run MSDOS! (sorry, bad joke.)

To be honest, my machine was built for Windows and Windows only. It was impossible to run Linux until a new BIOS version came out, I had to make a bunch of changes to CMOS before I installed centOS, and then finally, I installed ArchLinux.

HappyClown · 2015-10-27 22:39:15

I just found something while going through the wiki. from https://wiki.archlinux.org/index.php/Netconsole: "The module is mainly used for capturing kernel panic output from a headless machine." Can I use this while running Xorg to capture the listing?

ewaller · 2015-10-27 22:46:11

It sure looks like it. I would image that the only problem is that the tail end of the log might not get sent from the kernel that is in the process of panicking.

Arch Linux

#1 2015-10-24 19:35:44

Kernel panic, can't understand dmesg

#2 2015-10-25 00:25:33

Re: Kernel panic, can't understand dmesg

#3 2015-10-25 00:39:18

Re: Kernel panic, can't understand dmesg

#4 2015-10-25 01:36:01

Re: Kernel panic, can't understand dmesg

#5 2015-10-25 03:24:31

Re: Kernel panic, can't understand dmesg

#6 2015-10-25 20:40:15

Re: Kernel panic, can't understand dmesg

#7 2015-10-26 00:19:03

Re: Kernel panic, can't understand dmesg

#8 2015-10-26 04:11:18

Re: Kernel panic, can't understand dmesg

#9 2015-10-26 08:26:19

Re: Kernel panic, can't understand dmesg

#10 2015-10-26 12:34:27

Re: Kernel panic, can't understand dmesg

#11 2015-10-27 02:42:33

Re: Kernel panic, can't understand dmesg

#12 2015-10-27 22:39:15

Re: Kernel panic, can't understand dmesg

#13 2015-10-27 22:46:11

Re: Kernel panic, can't understand dmesg

Board footer