Hard freezes at seemingly random intervals

[vectorboson]

Hello, community. I posted this in another thread, but I now believe it's a separate issue, so I'm floating it separately to see if anyone has any ideas:

My entire system will freeze seemingly at random: sometimes when it's doing something (e.g. compiling, installing things), sometimes when it's just sitting around and I haven't touched it for several minutes - in any case it happens even with suspend and brightness reduction turned off. This has happened with a wide variety of applications open, and it does not seem to have anything to do with any individual programs. The system does not respond to SysRQ commands when it has locked up in this way. The machine is an essentially brand new HP Pavilion dm1-4010us, and it has done this in both a 64-bit installation upgraded from 11.10 and a 32-bit fresh installation of 12.04. I have it dual-booting Windows 7, and have not encountered any issues on Windows.

System specs are:
AMD E-450 Processor
4GB RAM, memtested and it's fine
Radeon HD 6320 Graphics
Broadcom BCM4313 Wireless controller, which has given me trouble before on 11.10

Though I am by no means an expert, I suspect the wireless controller may be related somehow: on 11.10, before I used the driver workaround in the networking support forum, I was getting much more frequent lockups along with unacceptably terrible wi-fi performance. The SysRQ keys DID work for these lockups. Once I did the driver override, wi-fi performance dramatically improved but I continued experiencing these freezes, and SysRQ no longer worked during them. I notice that the wi-fi fix appears to be applied by default in 12.04, as I did not have to do anything to get acceptable performance out of it when I reinstalled.

Two additional points:
-When this happens with audio playing, it stops immediately.
-This has occurred both with open-source video drivers and with flgrx (post-release updates) in use.

Anyone who has any theories or ideas for additional tests to perform, let me know and I'll try them as I'm able. I use Ubuntu for work and this is really making it hard to get anything done.

[pureblood]

Look at this post: http://bit.ly/KH1uJ7
It seems a lot of people are experiencing the same exact problem. Have you tried to keep the computer on without the wireless driver on to see if this prevents it from freezing? What wireless card/wireless driver do you use?

[pjbroad]

I have exactly the same freezing issue with my HP DM1-4125sa. This has the same CPU, Graphics and wifi as your HP dm1-4010us. The suggested launchpad thread appears to be filled with random freeze problem from many different hardware combinations and with hard and soft freezes too. My freeze is always hard, requiring a power off reboot. I cannot switch to a VT or use SysRQ.

I have the same random freeze with Debian Wheezy (testing), Ubuntu 12.04 and Ubuntu 12.10, Unity 2d/3d and Gnome. It does not appear to make a difference whether I use the proprietary or open source graphics driver so I'm tending to rule that out.

I've tried inducing the problem by loading the system with cpu, network, and graphics loads. Disconnecting/reconnecting wired network, wifi, power, suspend/resume etc but nothing. The only reliable way to induce the problem is to briefly forget about it. I must admit it's driving me crazy! There is never anything of note in the system logs.

[vectorboson]

I believe this problem, or at least my version of it, is caused by a bad disk I/O interaction between the motherboard and the Linux Kernel. I did some further tests, and found the following:

-it happens even with every possible wi-fi driver blacklisted.
-it does not happen with no programs running at all and the machine sitting completely idle, disconnected from the internet, even if left up for over 72 hours without suspending.
-it tends to happen quite quickly (within a few minutes) when I initiate a disk usage-heavy task like compiling software in a terminal, though even this is unreliable and it may stay up for over an hour.
-upgrading the Kernel to 3.4.x does not help.
-it happens even on KDE Fedora, so it's not a bug in Gnome, Unity, or Ubuntu.

A friend suggested that it could be a bug in the disk controller or host adapter that the manufacturer worked around for the Windows drivers but has not publicized so Linux can fix it. I'm not sure what else to do; I haven't been able to find any sign that it's a known issue with the Kernel team. I'd prefer not to have to buy another new computer to be able to run Linux, but I can't think of any further possible avenues for a workaround on our ends.

[MrArch]

Interesting, found your post via
http://ubuntuforums.org/showthread.php?t=1903054&page=2

I'm on a dm1-4000sd myself, running Arch Linux 64 bit and experiencing hard lockups too. When you say you blacklisted every Wifi driver iwconfig did not report any wlan0 devices right? (just to be sure).

The only other far fetched guess I could make is that we have a CPU affected by this problem:
http://it.slashdot.org/story/12/03/0...ds-matt-dillon

I'm currently running some compiler tests and trying out various HD stress situations to see if I can consistently reproduce the bug.

[vectorboson]

I didn't think to check the wlan0 devices and have since reformatted to a different distro (which also didn't work), but I do remember hitting the wireless switch on the keyboard and having it never go to "on", which seems to me like it would imply that no device was recognized. If it would be useful to investigate beyond that I could blacklist everything again in a few days when I'm back where the computer is.