how to stop a lockup

batkins · Tux's lil' helper Joined: 20 May 2003 Posts: 127

i've been using gentoo for a little while now, but it has an alarming tendency to lock up when I'm running X. normally, i just hit ctrl-alt-f1, log in as root, and kill the offending process (usually gtk-gnutella or mplayer)

every now and then, tho, this doesn't work and i'm forced to hard-reset my box. is there any definite way to stop a misbehaving process? in xp pro, ctrl-alt-del and task manager pretty much always worked. but i can't seem to rely on my current method (switching to a virtual console and killing the process).

how do other gentoo users deal with this?

thanks for your time!

JM_ · n00b Joined: 09 Apr 2003 Posts: 13 Location: Bucharest

burmashave · Tux's lil' helper Joined: 01 Dec 2002 Posts: 82

batkins · Tux's lil' helper Joined: 20 May 2003 Posts: 127

um, how could i have set up my box in such a way that it would lock up?

also, given linux's famous stability, i would expect the kernel to accomodate "poor setups."

Frozenwings · n00b Joined: 02 Apr 2003 Posts: 47

Hmm...my box was locked two times for unknown reasons...(not counting
two other freezes because of nvidia drivers). But in those, my keyboard
was somehow locked up and it won't respond to any alt+ctrl+backspace,
alt+ctrl+Fx or anything like that. I fear these lockups, as I don't have another
box to kill the process from (in these two locks, processes in the background
were active, even xmms). Any way to recover from these?

dsd · Posted: Sun Jun 08, 2003 6:56 pm Post subject:

make sure that you have DMA enabled on your hard disk.
this is done through hdparm. Search it on these forums if you need help.
for some reason, after a fresh install, it wasnt enabled on my slower box, and it kept crashing.

enabling DMA stopped this. and bloody hell it runs fast now

Frozenwings · n00b Joined: 02 Apr 2003 Posts: 47

If that was for me, my disk is tuned-up.
Looks like the failures were related to gqview, I'll go ask in the multimedia
forum as it's not behaving as it should lately.

sn4ip3r · Posted: Wed Jun 11, 2003 9:36 am Post subject:

I have several PC's at home and when I X locks up, it also disables mouse and keyboard so I can't ctrl-alt-F1 or smth. but SSH access is still enabled and I can kill X from another PC, but even then it doesn't always fix things, it might lock the whole PC so even SSH doesn't work anymore.

puddpunk · l33t Joined: 20 Jul 2002 Posts: 681 Location: New Zealand

What you could also look at doing (and is of great help to me) is enabling the "Magic SysRq Key" in the kernel config, and using that to safely unmount your system etc... The SysRq key is the PrintScrn key.

It almost always works, unless your system is REALLY hard locked.

If my system jams, I press: Alt + SysRq + s to sync my disks (i.e. flush all buffered data onto the disk, so I don't lose anything), then I press: Alt + SysRq + u, which unmounts my filesystems (and I think remounts them ReadOnly, but I can't remember. Anyway, it stops the long fsck at boot) then I press Alt + SysRq + b for a reboot, which is fine because all my filesystems are unmounted, or safely mounted.

Just a tip for the hackers amoung you

lots more cool things can be done with the Magic SysRq key, there is a readme file about it in the kernel documentation directory!

Cheers,
Chris.

bsolar · Bodhisattva Joined: 12 Jan 2003 Posts: 2764

TinheadNed · Posted: Thu Jul 17, 2003 3:26 pm Post subject: Lock ups

My laptop has never stopped locking up in X since I installed gentoo. Redhat was fine. Changing optimisations hasn't made any difference. These lockups can occur between seconds and hours after starting, and seem linked to using the mouse (as keyboard only is fine). When it locks, the keyboard is locked, the mouse is locked, the acpi events from the sleep button (on the keyboard) is locked. The only thing that works is network devices, and the lid switch. Magic Sysrq does not work.

Any ideas for that?

xunil · Posted: Thu Jul 17, 2003 7:58 pm Post subject:

In the years I've been using Linux, almost every true system lock-up was caused by hardware. Hardware is the direct cause when your hardware is just plain bad or flaky (overheating due to overclocking or packing too many heat-producing devices -- hard drives -- in an area w/o enough airflow or underpowering, both being very common among lamers who build their own computers w/o doing enough research). Hardware can indirectly be the cause when a driver is bad, but this usually comes w/ a kernel oops and traceback. In the few cases when hardware was not the true problem, it's userspace interacting poorly w/ kernelspace, but often this can be another indirect hardware issue: most userspace software that interacts w/ kernelspace is doing so to gain access to data provided by hardware (like X or webcam programs).

To prevent lock-ups due to hardware, I follow a few guidelines:

1) Before buying any hardware, I Google for it's name and the word "Linux" together. Also, searching one of the many LKML archives for the same keywords is very helpful.
2) Buy name-brand. I know that Asus motherboard is $20 more expensive than that off-brand one, but it's worth it. Trust me. This is *especially* true of RAM. Don't buy bleeding edge; chances are Linux won't support it well if it supports it at all.
3) Are you running the latest *stable* versions of tricky software like the kernel, X, or your webcam program? Being way behind is just as risky as being bleeding edge.
4) Check for BIOS updates frequently. If your motherboard's manufacturer has a notification list for BIOS updates, get on it. BIOS changelogs have a way of being terribly incomplete, too, so apply the stable updates. If other hardware has updateable BIOS/flash, make sure you're up to date w/ that hardware as well.

SeJo · Posted: Fri Jul 18, 2003 6:00 am Post subject:

odinbpl · n00b Joined: 24 Nov 2002 Posts: 19 Location: London

Whilst playing with new kernels (ck-sources with boot splash and nvidia), I sometimes experienced a similar X server lockup. I used to ssh into my box using a laptop and shut it down, but now I've installed ACPID (available on portage) and configured the power button to do a software powerdown.

One note: The version of acpid that I installed needed reconfiguring to recognise the power button event (I think it changed in kernel 2.4.20+), if you install it check that it works ;-)

_________________
--
Steven Capper

jago25_98 · Apprentice Joined: 23 Aug 2002 Posts: 164

Assuming you have in your kernel - Kernel Hacking :: Magic SysRq Key
compiled in...

To get out of some lock ups you can use the system rescue key.

It's on your keyboard, honest.

Found it?

Ok, it doubles as the print screen key.

To use it hold alt+sysRq+ ->

R (raw)
S ?
E (kill current process, very useful)
I ?
U (unmount mounted filesystems to avoid damage)
B (force reboot)

To remember that sequence - "Rasing Silly Elephants Is Utterly Boring"

TinheadNed · Posted: Wed Jul 23, 2003 7:52 pm Post subject:

Magic Sysrq doesn't work here when X locks. Otherwise it's very useful, and if you get the keys wrong, the set of commands will be printed to the terminal. Or look in the kernel documentation