random freezes on HP DM1Z

[pmheideman]

I'm having random freezing issues on my hp dm1z. I can't seem to narrow them down to any particular process, but they happen once or twice a day, which basically makes the OS unusable. Often, the freeze is so severe that magic REISUB doesn't work, and I have to do a hard reset. Does anyone have any advice for tracking down the cause of this problem, or even solving it?

[tashirosgt]

Is this with FC 17? Is the freeze so hard that the numlock LED on the keyboard doesn't respond? What do the logs show? Check .xsession-errors in your home directory. Check /var/spool/abrt for directories that show crashes. And there are all the logs in /var/log, such a /var/log/messages.

Is the functionality of the sysrq key enabled in /etc/sysctl.conf ?
kernel.sysrq = 1

I've only tried FC 15, 16, 17 on desktop machines and I had intermittent crashes everytime I used an iOne brand USB mouse with them. Old Lenovo (IBM) brand USB mice work reliably. Try plugging in a mouse to the laptop.

Search for problems with the specific video hardware that you have. You can find out the Linux designation for it with lspci and a command like lspci -vs 01:00.0. (Use your VGA cards slot number instead of 01:00.0 ).

[pmheideman]

Thanks for the reply! Sysreq is now enabled, but I had a crash after I enabled it and the computer was still non-responsive.

/var/log/messages didn't have anything useful.

Jul 22 22:30:48 NotaBigTruck kernel: [  107.496568] sd 1:0:0:0: [sdb] Test WP failed, assume Write Enabled
Jul 22 22:30:48 NotaBigTruck kernel: [  107.499964] sd 1:0:0:0: [sdb] Asking for cache data failed
Jul 22 22:30:48 NotaBigTruck kernel: [  107.499981] sd 1:0:0:0: [sdb] Assuming drive cache: write through
Jul 22 22:31:40 NotaBigTruck kernel: [  159.134069] sd 1:0:0:0: [sdb] Test WP failed, assume Write Enabled
Jul 22 22:31:40 NotaBigTruck kernel: [  159.137286] sd 1:0:0:0: [sdb] Asking for cache data failed
Jul 22 22:31:40 NotaBigTruck kernel: [  159.137302] sd 1:0:0:0: [sdb] Assuming drive cache: write through
Jul 22 22:31:57 NotaBigTruck systemd[1]: Startup finished in 3s 542ms 768us (kernel) + 14s 101ms 723us (initrd) + 2min 37s 922ms 682us (userspace) = 2min 55s 567ms 173us.

You can see the crash happened at about 22:30, but there are no events logged here, as far as I can tell, that are out of the ordinary. Doesn't seem to be anything in .xsession-errors either. Which other logs should I pay attention to?

I'm also trying the solution described here, though my grub file doesn't have the line described here, so I'm not sure if I should simply add that line or what the best course of action is to disable the thermal module. I've already blacklisted the driver, and that didn't solve it.

---------- Post added at 11:00 PM ---------- Previous post was at 10:57 PM ----------

Also, the VGA device is ATI Technologies Inc Device 9806. Couldn't find any reports of hard freezes with it.

[tashirosgt]

You didn't say if the NUMLOCK LED works when the system freezes - if your machine has such an LED. Are you using Gnome3 or some other spin?

A problem that you'll have with logs is that some of them (like .xsession-errors) are rewritten when you reboot and unless you can get a console after the machine freezes (CLTR-ALT-F2) you won't be able to examine the current log. When the NUMLOCK LED fails to work on a system, I have never been able to get a console.

I don't want to over-sell the power of a diagnostic approach. I prefer to tackle problems that way, but often problems are solved by "replacing random parts" , by analogy to how automotive repairs often proceed. Other logs to examine are /var/log/boot.log, /var/log/Xorg.0.log, /var/log/secure, /var/log/cron and the older versions of these logs.

I looked at the link you provided. None of the posters helped us out by showing how to customize what is written to log files. This can be done, but I'd have to research how.

[pmheideman]

I don't have numlock, but I do have capslock and wifi LEDs, and both of those stop responding. I've also never been able to get into a console from the freeze. I'm using the standard distro of Fedora 17, I believe, which runs Gnome 3.4.

I figured out how to turn off the thermal module in grub (just added thermal.off=1 to the GRUB_CMDLINE_LINUX_DEFAULT list). No freezes yet, but I only did that just last night, so it will take a few days of testing to see whether or not that has really solved it.

[pmheideman]

So adding the thermal.off=1 got me about a week without any freezes. They have, however, just started up again. Now Magic REISUB works, so I'm not wearing out the hardware as much, but it's still unbelievably frustrating.

[tashirosgt]

I haven't used the SYSRQ key myself. Can you get a terminal without rebooting? That would let you look as .xsession-errors before it was overwritten by a new boot.

[pmheideman]

Once the freeze begins (which starts with a mouse speed slowdown and the two finger scroll ceasing to work, followed by total lockup after about 5 seconds), I can't bring up a terminal. I keep forgetting to try and switch using CTRL ALT F2, so next time I'll give that a try and see if I can switch before it locks up.

[tashirosgt]

When you use the SYSRQ key, you don't have to reboot. See if you can make CTRl-ALT-F2 work before rebooting.

[droidhacker]

Originally Posted by pmheideman

So adding the thermal.off=1 got me about a week without any freezes. They have, however, just started up again. Now Magic REISUB works, so I'm not wearing out the hardware as much, but it's still unbelievably frustrating.

So you're disabling thermal support in ACPI.... tell me, how HOT is the CPU when the thing shuts down? What you're describing sounds like a cooling problem. Fans running fast when it shuts down? Fans running AT ALL EVER?

Some laptops have software controlled cpu fans, and if the software controls aren't implemented in the OS, you could face overheating issues.

PLUS, it looks to me that that laptop has DUAL GRAPHICS. The AMD E2 is an APU with an HD 7310 (E2-1200) or 7340 (E2-1800), and it has a discrete HD "6300 series". That means that the second GPU is sitting there sucking up power and increasing heat DRAMATICALLY. There are typically a couple of ways to disable the discrete GPU, the simplest is to find the BIOS option and disable it there. It is probably called something like "switchable graphics". There is also something at the software level, I think vgaswitcheroo...??? Or something like that.