Debian freezing

I've also posted this on a Debian forum, with no luck so far. I am running Debian Wheezy with the 2.7 release of LinuxCNC. My computer is a Dell Optiplex 360 with 2G ram (in one slot). I have disconnected the DVD drive, all it has is keyboard, display and mouse. I am configured for dual boot with Windows 7. Note that Ubuntu 10.04 did the same thing.

If I am logged in, but the machine is idle, in 30-40 minutes the screen blanks and neither keyboard or mouse will bring it back. The only recourse that works is power cycling the machine. This happens with either the onboard graphics or with a Matrox G550 video card. I have used memtest to test the ram, no problems. I have removed and reseated the ram and cpu. I cleaned off the thermal grease on the cpu and new (little dab) applied.

The same sort of thing happened in Windows, when I thought to try it. Reading about Windows freezes, I set all my usb root hub devices so that the system was not allowed to power them down. Windows now stays live (well, it goes into screensaver, but recovers normally) for more than 24 hours. So I conclude it is probably not a hardware issue. A usb issue makes sense as both keyboard and mouse are usb.

If I am actively using the keyboard/mouse, it delays the freeze. When it freezes, I cannot successfully reboot immediately, I have to wait some time (30 min?).

I have changed the battery and updated to the most recent bios. The Wheezy screensaver is off, power management is set to "don't do it".

The usb issue in Windows made sense as the keyboard and mouse are both usb. If the machine went to sleep, and the usb ports were also asleep, it would not notice my inputs.
I put this in /etc/default/grub:
GRUB_CMDLINE_LINUX_DEFAULT="quiet splash usbcore.autosuspend=-1"
and ran update-grub. When I rebooted I verified that autosuspend was set to -1 (it was originally 2). This did not help.

I put these lines in my ~/.xinitrc (I did not have one before):
xset s off # don't activate screensaver
xset -dpms # disable DPMS (Energy Star) features.
xset s noblank # don't blank the video device
exec /etc/alternatives/x-session-manager # start lxde

I find it perplexing that the screen still blanks after all of these settings. It also still freezes.

After the freeze this morning I let the machine cool off for about 30 minutes and then rebooted. Immediately after login I started xsensors; I had to figure out the details of lm-sensors, so that took a few more minutes. Xsensors didn't say if it was updating; I ran 'watch sensors' which claimed to be updating every two seconds. Both started at 38.0/37.0C (each core) and stayed there, not a bit of change until the machine froze and the screen blanked. I was not watching it every second--five minutes before the freeze, 38.0/37.0C. When I came back five minutes later, blank screen and frozen. It doesn't seem physically reasonable that the temp stayed constant to 0.1C, but I don't know where to go with that.

I have the same install on my office laptop. It has no problems.
Any ideas will be greatly appreciated.

Steve

Andy, thank you for some great leads.
The caps-lock and num-lock keys DO light their respective LEDs when "frozen". I'm not sure what to make of this clue, but a definite advance.

Digging through /var/log/kern.log I found lines saying the machine was trying to come out of hibernation but could not find a state file. This sounded like a big clue, so I have been experimenting with different ways to force it to not hibernate. No luck so far. Here are the lines from kern.log:

Sep 14 14:10:12 steve-cnc kernel: [ 3.111673] generic-usb 0003:046D:C05A.0003: input,hidraw2: USB HID v1.11 Mouse [Logitech USB Optical Mouse] on usb-0000:00:1d.2-2/input0
Sep 14 14:10:12 steve-cnc kernel: [ 6.301458] sd 2:0:0:0: [sda] 312500000 512-byte logical blocks: (160 GB/149 GiB)
Sep 14 14:10:12 steve-cnc kernel: [ 6.301507] sd 2:0:0:0: [sda] Write Protect is off
Sep 14 14:10:12 steve-cnc kernel: [ 6.301510] sd 2:0:0:0: [sda] Mode Sense: 00 3a 00 00
Sep 14 14:10:12 steve-cnc kernel: [ 6.301530] sd 2:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
Sep 14 14:10:12 steve-cnc kernel: [ 6.348141] sda: sda1 sda2 sda3 < sda5 sda6 >
Sep 14 14:10:12 steve-cnc kernel: [ 6.348535] sd 2:0:0:0: [sda] Attached SCSI disk
Sep 14 14:10:12 steve-cnc kernel: [ 6.349879] sd 2:0:0:0: Attached scsi generic sg0 type 0
Sep 14 14:10:12 steve-cnc kernel: [ 11.867122] PM: Starting manual resume from disk
Sep 14 14:10:12 steve-cnc kernel: [ 11.867125] PM: Hibernation image partition 8:6 present
Sep 14 14:10:12 steve-cnc kernel: [ 11.867127] PM: Looking for hibernation image.
Sep 14 14:10:12 steve-cnc kernel: [ 11.882516] PM: Image not found (code -22)
Sep 14 14:10:12 steve-cnc kernel: [ 11.882519] PM: Hibernation image not present or could not be loaded.
Sep 14 14:10:12 steve-cnc kernel: [ 11.927992] EXT4-fs (sda5): mounted filesystem with ordered data mode. Opts: (null)
Sep 14 14:10:12 steve-cnc kernel: [ 13.745109] input: Power Button as /devices/LNXSYSTM:00/LNXSYBUS:00/PNP0C0C:00/input/input3
Sep 14 14:10:12 steve-cnc kernel: [ 13.745116] ACPI: Power Button [VBTN]
Sep 14 14:10:12 steve-cnc kernel: [ 13.745176] input: Power Button as /devices/LNXSYSTM:00/LNXPWRBN:00/input/input4
Sep 14 14:10:12 steve-cnc kernel: [ 13.745179] ACPI: Power Button [PWRF]

I followed these instructions to prevent hibernation:
www.linuxquestions.org/questions/debian-...-down-applet-877889/
but nothing changed. I have something else to try, but it requires an xconf.org file which I don't have. I found how to create one, but it requires single user mode, which I have not been able to figure out how to get into.

Thanks again.
Steve

I was out this evening. Before I left, I logged in and typed the xset -dpms command. When I got back, the screen was blank and the system could not be woken ("frozen" just like before). Numlock and Capslock still lit their LEDs.

Figuring out ssh will take a little work.

Steve

I followed these instructions to prevent hibernation:
www.linuxquestions.org/questions/debian-...-down-applet-877889/

That looks to be about how to remove the command from an applet menu, not how to stop it happening.

I run Wheezy and Jessie on all my machines and have not had this problem.

I do however disable ACPI features such as suspend to disk and hibernate, wake on LAN etc. etc. as a matter of routine in BIOS.

If I build a kernel I disable them at source in the config.

I would suspect the problem is either from the power management or the monitor itself. Some 'smart' monitors can be difficult to get displaying anything if they lose a signal, or never have one (start up with monitor off then switch on later when computer has no longer displaying anything).

First step, go into BIOS and turn everything related to power management off.

regards

When the screen is blank and machine does not respond to keyboard or mouse, I can ssh in from another computer. It's alive!
How do I figure out what is causing the problem, now that I can actually find something out?

Thanks,
Steve

Since you can login from another computer, maintain a session on the other
computer and monitor syslog continuously:

and you might get some clues