System freeze related to "rcu_sched detected stalls on CPUs"

[fredbear5150]

Just to add to my previous post, I note the comments about the nouveau driver possibly being the cause.

I should add that with my newly built system, which was running perfectly stable a few months ago under an older kernel, I did replace the radeon card in it then for a nouveau (NVIDIA) card now - so there seems to be a common factor there.

I have just reconfigured the system and put the old ATI card back in, it will be interesting to see if it is now stable. The next couple of hours will tell.

[fredbear5150]

Just to confirm that having replaced the NVIDIA card using the nouveau driver with an ATI card using the radeon driver, the system has been up and running and entirely stable for several hours now - it would have crashed out by this time were the original problem still present.

Therefore I think the conclusion that it's a nouveau driver problem is correct. I haven't tried the closed NVIDIA drivers, I don't need to use either those or the closed ATI drivers in my general computing work.

It would be interesting to see what the actual problem is in the nouveau driver that seems to be causing this issue.

[fredbear5150]

My final post on the issue but just to confirm that I left the PC doing an "rsync" overnight from a NAS drive to its local hard disk and it was still happily running this morning - with the NVIDIA card it would have definitely have crashed, this is the longest time the PC has been up since the recent rebuild.

[TigerJr]

What kernel are you use before 4.1.12 version?
_________________
Do not use gentoo, it die

[ElDavo]

Finally - Someone with my same issue!
It is so strange. System hangs randomly. Sometimes I can move the mouse on X. This usually happens with Internet browsing.
Intel Pentium N3520 with its integrated Intel GPU Card. Sometimes, the browser hangs and I see the very same dmesg messages you posted. No CPU usage, but high load average.

Something I found really interesting is that with the default Gentoo Linux kernel this bug does not happen, but when I start customizing the configuration, it happens.
I struggled for months to understand what was causing the issue, and I believe I found it, even though I had many false positives in the past.

Please tell us what option did you choose in General Setup > Timers Subsystem > Timer Tick Handling.
The default is tickless idle and I have no problems with that.
I have this problem with Full dynticks system (tickless) ( CONFIG_NO_HZ_FULL ) .
The strange fact is that this option doesn't change anything by default because you have to put your cpu numbers in the nohz_full= boot parameter, yet changing the option causes a complete kernel recompilation.