Xorg uses 100% CPU after 5-10 minutes

I originally thought this was related to Firefox3 but I can reproduce without running Firefox.

Strace output consists of only what's attached.

Hi, did you capture any earlier data with strace. What you've included isn't a lot of help on it's own because it is not possible to determine what device the file descriptor '9' corresponds to.

Thanks

I'm having trouble capturing an strace from early enough in the Xorg processes life to see what file descriptor '9' is.

Trying to strace the parent gdm process with -ffF causes X to crash immediately.

Trying with 'startx' from the console throws the error "xf86EnableIOPorts: failed to set IOPL" which I haven't been able to figure out in 10 minutes of Googling just now.

Any pointers on this?

Right, so attacking this a different way.

$ sudo ls -l /proc/7102/fd/9
lrwx------ 1 root root 64 2008-08-28 11:41 /proc/7102/fd/9 -> /dev/dri/card0

I'll test disabling DRI.

Regarding the error message "xf86EnableIOPorts: failed to set IOPL", that is probably a very strong clue. This error message occurs in ./hw/xfree86/os-support/linux/lnx_video.c in function xf86EnableIO(), line 566 due to a failure in this call:

        if (ioperm(0, 1024, 1) || iopl(3)) { ...

That's a pretty low level call... almost makes me wonder if it's a hardware issue?

Fwiw, *sometimes* high X resource loads can just be due to weird X client apps (like a panel applet you might have installed from somewhere). To rule out this possibility, could you burn and boot an Intrepid Alpha-4 LiveCD? If you can reproduce the issue there, it'll strengthen the case that this may be a hardware or kernel issue. If you can't, then we may want to look more closely at the software.

Hi Bryce

"xf86EnableIOPorts: failed to set IOPL" only occurs if I try 'startx' as a non-root user, so it's probably a permissions issue.

I've burnt a LiveCD and am testing it now. The first boot seemed to freeze at the desktop, panels still blank. The mouse cursor could be moved around the screen but clicking did nothing and no keyboard input had any effect. The second boot got me to a working desktop but I've just gone to launch a terminal and it's frozen again, moving cursor but no other response.

Ok, got it.

LiveCD Intrepid Alpha-4

Stracing Xorg
rt_sigreturn(0xe) = -1 EINTR (Interrupted system call)
ioctl(10, 0x40046457, 0x27d636c) = -1 EINTR (Interrupted system call)
--- SIGALRM (Alarm clock) @ 0 (0) ---
rt_sigreturn(0xe) = -1 EINTR (Interrupted system call)
ioctl(10, 0x40046457, 0x27d636c) = -1 EINTR (Interrupted system call)
--- SIGALRM (Alarm clock) @ 0 (0) ---
rt_sigreturn(0xe) = -1 EINTR (Interrupted system call)
ioctl(10, 0x40046457, 0x27d636c) = -1 EINTR (Interrupted system call)
--- SIGALRM (Alarm clock) @ 0 (0) ---
rt_sigreturn(0xe) = -1 EINTR (Interrupted system call)

root@ubuntu:~# ls -l /proc/6791/fd/10
lr-x------ 1 root root 64 2008-08-31 21:59 /proc/6791/fd/10 -> /dev/dri/card0

Working my way through http://dri.freedesktop.org/wiki/DriTroubleshooting

'dmesg | grep drm' output has an error (attached) that I can't fathom, however Xorg.0.log claims "Direct rendering enabled" so it can't be too fatal.

The next two attachments are the standard output and standard error when running 'glxinfo'.

There are no more troubleshooting steps in that document. But running 'glxgears' freezes the system instantly, Xorg at 99% CPU. strace attaches to the Xorg process but then freezes itself.

Hi Marty,

Given that it sounds like it's DRI related, maybe a next step would be to experiment with different settings of AGPMode? This is a pretty common issue with the -ati driver.

In your /etc/X11/xorg.conf set AGPMode to various values like this:

Section "Device"
   ...
   Option "AGPMode" "2"
EndSection

Possible values include 1, 2, 4, 8. You can see what it's currently set to by looking in /var/log/Xorg.0.log. It's worthwhile to test all four values even once you find one that works. If this takes care of it for you, we may be able to establish a quirk to set it for your hardware combo, if you can provide the following data:

  * AGPMode value(s) that work
  * Make/Model of laptop or motherboard
  * Output of lspci -vvnn
  * Is the system all factory hardware, or have any parts been replaced?
  * Is there an AGP Mode in the system BIOS?
    - If so, is it set to the factory default?

For more details on this, please see the "ATI AGP Mode Quirk" section at https://wiki.ubuntu.com/X/Quirks

Apologies for the delay in response. The card had been in a work PC and was swapped out for something more stable till I had time to revisit this.

The issue is no longer present with an up to date Intrepid install (only issue is occasional artefacts around the cursor), so I'll resolve this.