2

I've been having this problem for several months. I'm running Linux on my personal computer. At random times, upwards of 3 times per day, my system totally freezes and becomes unresponsive to any input except a hard power-off. No mouse cursor movement, no SysRq magic keys, even pressing Num Lock doesn't toggle the LED on my keyboard.

I understand hardware problems are often to blame for things like this. However:

  • PassMark MemTest86 succeeded with 0 errors or warnings.
  • smartctl and badblocks succeeded with 0 errors or warnings.
  • I have never observed any excessive temperatures or fan problems.
  • My graphics card is AMD (i.e., this is not one of the issues associated with NVIDIA cards.)
  • I ran Windows on this hardware for years without comparable problems.

I have set up kdump in an attempt to diagnose the issue. The dump kernel executes when I induce a kernel panic using echo "c" > /proc/sysrq-trigger, but when a wild freeze occurs it just stays frozen indefinitely without executing kdump or rebooting.

My sysctl configuration has these variables set:

kernel.hardlockup_panic = 1
kernel.hung_task_check_count = 4194304
kernel.hung_task_check_interval_secs = 0
kernel.hung_task_panic = 1
kernel.hung_task_timeout_secs = 10
kernel.hung_task_warnings = 10
kernel.nmi_watchdog = 1
kernel.panic = 60
kernel.panic_on_io_nmi = 1
kernel.panic_on_oops = 1
kernel.panic_on_rcu_stall = 1
kernel.panic_on_unrecovered_nmi = 1
kernel.panic_on_warn = 1
kernel.softlockup_panic = 1
kernel.soft_watchdog = 1
kernel.unknown_nmi_panic = 1
kernel.watchdog = 1
kernel.watchdog_cpumask = 0-3       # my system has 4 cores
kernel.watchdog_thresh = 10

I have observed these freezes on Ubuntu 18.04 running Linux kernels 4.19 and 5.0, and on Arch Linux running Linux kernels 4.19 and 5.3. I'm running the latest Intel microcode package. There is nothing corresponding to these freezes in any Xorg log or .xsession-errors

I'm out of ideas. What should I try next?

2
  • 1
    I'm having the exact same problem. Wondering why it's happening. At least in my case, perhaps it has something to do with the graphics card or driver, because when it happens, my screen glitches in strange ways, like numerous light-coloured dots appear everywhere. The only difference of my setup in comparison to yours is that I have only the integrated GPU. My processor is a 4th gen Intel (I think it's an i5-4590, or something). Commented Nov 13, 2019 at 23:25
  • There are currently some open kernel bugs related to power management of some (older) Intel processors and transitions between idle states. So far these have seemed the closest match to my problem and the workarounds suggested by the bug threads have been somewhat promising, namely adding intel_idle.max_cstate=1 to my cmdline. see bugzilla.kernel.org/…
    – squeevee
    Commented Nov 14, 2019 at 22:28

1 Answer 1

1

This freezing closely matches an open bug in the Linux Kernel (https://bugzilla.kernel.org/show_bug.cgi?id=109051) related to power management and idle states of certain Intel processors. My system uses an i3-3220T CPU manufactured ca 2013, which appears to be approximately the same generation as the processors affected by the bug.

The workaround suggested by the bug thread is to add intel_idle.max_cstate=1 to my boot cmdline. Having done this, my system has not frozen in a month.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .