Re: [PATCH v2 1/2] rcu: don't bind offline cpu

[Peter Zijlstra]

On Thu, 2011-05-19 at 10:34 +0200, Peter Zijlstra wrote:
> On Thu, 2011-05-19 at 15:06 +0900, KOSAKI Motohiro wrote:
> > 
> > The right way is, explicit two phase cpu bindings (1) bind boot
> > (or any other online) cpu at CPU_UP_PREPARE (2) bind correct
> > target cpu at CPU_ONLINE. This patch does it. 
> 
> I'm not sure that is in-fact correct. From what I understood, RCU could
> have need of this thread before the ONLINE callback and expects it to be
> affine at that time.

Right, so we're waking that thread from the timer tick, so we must have
that thread set-up and ready _before_ we enable interrupts for the first
time. CPU_ONLINE is _WAAAAAY_ too late for that.

> What was wrong with delaying the wakeup to STARTING?

Hrmm right, so I thought STARTING was right after we marked the cpu
online, but its right before. Bummer.

Also Paul:

rcu_cpu_kthread()
  rcu_process_callbacks()
    __rcu_process_callbacks()
      rcu_do_batch()
        invoke_rcu_cpu_kthread()

Why wake the thread again if its already running?


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/