which doesn't exist nowby the way, I'm wondering about what's the race condition before
Before the commit you originally quoted, the calling thread could bepreempted and migrated to another CPU before get_online_cpus() thus
it's preemption mode related ,
on the 3.0.30-rt50, only config kernel with highest preemption level (Fully Preemptible Kernel (RT)) in cpu preemption model
will cause problem
and even i use the "Preemptible Kernel" or "Preemptible Kernel (Low-Latency Desktop)" the problem would not happen..
> I looked through the logs but the only worker depletion related patches which pop up are around CPU hotplugs, so I don't think they apply here. If the problem is relatively easy to reproduce && you can't move onto a newer kernel, I'm afraid bisection probably is the best option. Thanks!