Re: [PATCH] x86/irq: Cure live lock in irq_force_complete_move()

From: Thomas Gleixner
Date: Sat Mar 12 2016 - 03:29:10 EST

On Fri, 11 Mar 2016, Luck, Tony wrote:

> With this patch applied my system survives me doing
> several rounds of:
> # echo 0 | tee /sys/devices/system/cpu/cpu*/online
> # echo 1 | tee /sys/devices/system/cpu/cpu*/online
> whereas without the patch the first of those went to
> [152455.129604] NMI watchdog: Watchdog detected hard LOCKUP on cpu 96
> [152455.136943] Kernel panic - not syncing: Hard LOCKUP
> I'm not sure we care to optimize the cpu offline path, but I'll note here
> that taking all (but one) cpus offline took 52 seconds (for a hundred
> and mumble logical cpus). Bringing them all back is just 4 seconds.

Yes. I noticed that as well. No idea where we waste all that time. I've put
that on the todo list of our ongoing hotplug refactoring work.