Irresponsive computer in certain situation

gee · 2010-03-20 01:13:31

Hello,

I have an issue with my Arch system.
I am running Arch amd64 testing, and sometimes the computer will freeze.
Then nothing will work, not even the special REISUB keys.

Usually that happens when my computer is running sabnzbd, but not right after starting it, maybe half hour later or so.
I have run 13 passes of memtest++ nothing... Hitachi's DFT on both drives, nothing... Did a full night of prime nothing...

I cannot find any info in any log either, so I really don't know what to look at. What should I be looking for?

Thanks

MadTux · 2010-03-20 07:45:17

You should give more details on your system.

If your system freezes, does it sometimes after come back or do you have to do a hard reboot? Can you keep a shell open with, for example, htop and see if any program shows unusual behaviour at the time the system freezes?

gee · 2010-03-21 21:24:21

It freezes and there is nothing that I can do about it, it does not come back after waiting, I have to reboot using the switch.

The htop idea is interesting, I'll try that.

As for my system not sure what I should give, but I use ext4 for my partitions, run KDE on a radeonhd with the ati driver.

Thank you.

Last edited by gee (2010-03-21 21:24:43)

gee · 2010-03-24 00:43:39

So there is nothing outputted on the screen but a black screen and the cursor, therefore I cannot use htop for that

ngoonee · 2010-03-24 01:08:58

I see this once in a while, completely different system. It used to happen on my self-compiled kernel, so I went back to Arch's stock kernel primarily. Then it happened there too, but when I re-compiled my self-compiled kernel (kernel26-ice) with a lot of config options OFF, it worked fine for days at a time.

No idea how to trace it down, primarily cos (for me) there's no reproducible 'cause' to the hard freezes.

gee · 2010-03-24 21:20:32

Interestingly this problem has not occurred that much if sabnzbd is not running, but when it runs it will freeze for sure (at some point).
Though sabnzbd should not touch any kernel thing no? I don't see why it has such a huge effect...

The computer has crashed a few times with it not running but only KDE, Konsole and Firefox... So it is not directly the issue, but it triggers it somehow...

Thank you

ewaller · 2010-03-24 21:50:52

gee wrote:

It freezes and there is nothing that I can do about it, it does not come back after waiting, I have to reboot using the switch.

Are any lights on your keyboard blinking?
If not, do the Caps Lock and Num Lock kights still respond their associated keys?
Is the TCP/IP stack working? (Can you ping your box from another machine?)

gee · 2010-03-25 20:18:52

The keyboard is dead, caps lock does not work anymore (it's the only led I have on it).
Now I have no tried using another machine to ping it // ssh in, that may work.
I remember in the past, with other machines being able to do that in situations similar, so I'll see, but I have no hope...

Thank you!

gee · 2010-03-27 20:19:40

The machine is not pingable, so no ssh either

ewaller · 2010-03-28 03:30:20

Sadly, it sounds like the system is crashing so hard that the kernel does not even have time to panic. This can be very difficult to diagnose.

You indicated it is not so bad if sabnzbd is not running. That tells me the problem is there and can occur, but it is more likely to occur when sabnzbd is running. I know nothing of that program, but Google tells me it is a Python based news reader. Seem like kind of a vanilla program to be crashing the system. I would have expected something that would cause some piece of hardware wake up and do something stupid (unsolicited interrupts, DMA, etc...)

Any chance that your system is sensitive to heat? Does sabnzbd drive up processor usage upwards significantly?
Is there *any* correlation between log entries and crashes?

After that, you might try blacklisting modules one-by-one to see if that helps.

You might try some extensive RAM tests.

Getting more esoteric, there are some kernel command line arguments that cause the system to allocate interrupts differently, no enable or disable local apic, etc...

If you are REALLY desperate, you might try running a proprietary OS from the Pacific Northwest and see if the hardware behaves with that kludge.

gee · 2010-03-30 18:15:26

Yesterday I tried something and did not use X.
When I came back home, I had a general fault protection crash on screen, not sure what that means.
The message involved something related to sound too (although no sound was playing), I can post a picture later.

While I was investigating that in firefox, I got a new crash, then ext4 failure, then all of my /home/gee ended in lost+found (is there any way to fix this?)
So I switched to a different HD to see if it is related, but no same crash type.

I'll look at CPU temp tonight.
I see nothing in log entries, as for RAM I did 13 passes of memtest86+, is that no enough?

I actually want to try a couple of other things, no broadcom blob tainting the kernel for wireless, maybe a different kernel, etc.

Thank you!

gee · 2010-03-31 18:10:28

well I disabled a bunch of modules and daemons and no crash at all.
I enabled back everything this morning but broadcom-wl and the crypto librarie that goes with it.
We'll see how it goes.

gee · 2010-04-01 17:41:33

I enabled the crypto librarie and no issue.
This morning I modprobed broadcol-wl but I am not using it (still using ethernet), we'll see what that does.

Trent · 2010-04-02 03:36:40

For some reason I keep reading the title of this post as "Irresponsible computer in certain situation."

gee · 2010-04-02 18:08:18

Yeah that too, since in theory it could have crashed when I was working, but that never happened...
That would have been very irresponsible of it!

Now, I know for sure it is wl, as it crashed again yesterday an hour after I started using wireless...
Damn!
I want to use wireless, bloody blob!

ewaller · 2010-04-02 19:17:30

Which kernel are you using? Recent kernels have native support for low power Broadcom chips such as the one in my system:

02:00.0 Network controller: Broadcom Corporation BCM4312 802.11b/g (rev 01)
        Subsystem: Hewlett-Packard Company BCM4312 802.11b/g Wireless LAN Controller
        Flags: bus master, fast devsel, latency 0, IRQ 18
        Memory at d8500000 (64-bit, non-prefetchable) [size=16K]
        Capabilities: <access denied>
        Kernel driver in use: b43-pci-bridge
        Kernel modules: ssb

You might look into dropping the Broadcom-wl blob, using ssb from the kernel (I assume you have it blacklisted now) and then grabbing new firmware for the device using the latest b43-fwcutter (the one from AUR, not from core)

gee · 2010-04-03 00:25:02

I am using 2.6.33 but my wireless card is not supported by the free driver yet (it is a draft n card).

gee · 2010-04-05 02:02:08

I sent an email to broadcom and they replied that they don't deal with end users...

ewaller · 2010-04-05 03:48:10

Well, I really hate to suggest it, but what about ndiswrapper. I understand the philosophical evils, but while waiting for a pure solution, we can be pragmatic.

ajonat · 2010-04-05 04:16:41

I've been using ndiswrapper + windows' driver for almost 2 years now (my card is not supported by the in-kernel driver) and had 0 problems. You should try it.

gee · 2010-04-06 00:17:54

Can I do that?
I think I tried once and could not find a driver in 64 bits.
Are you guys using a 64bits driver for a 64b arch?

Thanks

Arch Linux

#1 2010-03-20 01:13:31

Irresponsive computer in certain situation

#2 2010-03-20 07:45:17

Re: Irresponsive computer in certain situation

#3 2010-03-21 21:24:21

Re: Irresponsive computer in certain situation

#4 2010-03-24 00:43:39

Re: Irresponsive computer in certain situation

#5 2010-03-24 01:08:58

Re: Irresponsive computer in certain situation

#6 2010-03-24 21:20:32

Re: Irresponsive computer in certain situation

#7 2010-03-24 21:50:52

Re: Irresponsive computer in certain situation

#8 2010-03-25 20:18:52

Re: Irresponsive computer in certain situation

#9 2010-03-27 20:19:40

Re: Irresponsive computer in certain situation

#10 2010-03-28 03:30:20

Re: Irresponsive computer in certain situation

#11 2010-03-30 18:15:26

Re: Irresponsive computer in certain situation

#12 2010-03-31 18:10:28

Re: Irresponsive computer in certain situation

#13 2010-04-01 17:41:33

Re: Irresponsive computer in certain situation

#14 2010-04-02 03:36:40

Re: Irresponsive computer in certain situation

#15 2010-04-02 18:08:18

Re: Irresponsive computer in certain situation

#16 2010-04-02 19:17:30

Re: Irresponsive computer in certain situation

#17 2010-04-03 00:25:02

Re: Irresponsive computer in certain situation

#18 2010-04-05 02:02:08

Re: Irresponsive computer in certain situation

#19 2010-04-05 03:48:10

Re: Irresponsive computer in certain situation

#20 2010-04-05 04:16:41

Re: Irresponsive computer in certain situation

#21 2010-04-06 00:17:54

Re: Irresponsive computer in certain situation

Board footer