I already posted in debian-user, but figured deb-kernel would be more appropriate.
I have an i7-3612Qe system that has been giving kernel panics when running a custom video streaming application pretty consistently after a couple minutes. After the kernel panic,
the machine reboots. When I turn off turbo mode in the BIOS, the panics happen less frequently. Other BIOS settings are all default, no overclocking or anything fancy.
-Can be reproduced by running streaming application using ~250% CPU, temps are a little high, they float around 68-71 degrees
-sysbench runs fine with 8 threads, throttles CPU up to ~800%, no kernel panics, temps remain below 70 degrees
-MemTest did not report any errors
-Intel Processor Diagnotic tool passed
-Tried swapping RAM
-Reproducible on multiple machines, not just a single processor (possibly eliminates it being a bad single proc)
-Able to mitigate most of the kernel panics and reboots by disabling Turbo mode (this is unacceptable, just including this for debugging purposes)
-Also able to mitigate kernel panics and reboots by changing the cpu frequency sacling_governor to conservative, from ondemand. Conservative should "gracefully increase and decreases
the CPU speed rather than jumping to max speed the moment there is any load on the CPU" (https://www.kernel.org/doc/Documentation/cpu-freq/governors.txt)
Here is what I could copy down from the kernel panic on the monitor, sometimes the messages vary slightly, bu the TSC and PROCESSOR messages are almost always the same.
[Hardware Error]: TSC 6e496d96062
[Hardware Error]: PROCESSOR 0:306a9 TIME 1418929330 SOCKET 0 APIC 3 microcode 12
[Hardware Error]: Run the above through 'mcelog --ascii'
[Hardware Error]: CPU 0: Machine Check Exception: 5 Bank 4: b200000000100402
[Hardware Error]: RIP !INEXACT! 10:<ffffffff811ee04c> {intel_idle+0xb9/0x119}
[Hardware Error]: TSC 148c99828a0
[Hardware Error]: PROCESSOR 0:306a9 TIME 1418929330 SOCKET 0 APIC 3 microcode 12
[Hardware Error]: Run the above through 'mcelog --ascii'
[Hardware Error]: Some CPUs didn't answer in synchronization
[Hardware Error]: Machine check: Processor context corrupt
Kernel panic - not synching: Fatal machine check on current CPU
Pid: 0, comm: swapper/3 Tained: P M 0 3.2.0-4-amd64 #1 Debian 3.2.51-1
Call Trace: ...
Also, here is a post from superuser on some suggestions that I tried: cpu
- Kernel Panic from overheating? - Super User
Here is my discussion on intel where someone recommended I post to the debian list
https://communities.intel.com/thread/58372?sr=stream
I am looking to actually debug and fix this, but not finding a lot out there. Any ideas for what to try next? |