Re: mm: deadlock between get_online_cpus/pcpu_alloc

From: Thomas Gleixner
Date: Thu Feb 09 2017 - 06:44:48 EST


On Wed, 8 Feb 2017, Christoph Lameter wrote:
> On Wed, 8 Feb 2017, Thomas Gleixner wrote:
>
> > There is a world outside yours. Hotplug is actually used frequently for
> > power purposes in some scenarios.
>
> The usual case does not inolve hotplug.

We do not care about your definition of "usual". The kernel serves _ALL_
use cases.

> > It will improve nothing. The stop machine context is extremly limited and
> > you cannot do complex things there at all. Not to talk about the inability
> > of taking a simple mutex which would immediately deadlock the machine.
>
> You do not need to do complex things. Basically flipping some cpu mask
> bits will do it. stop machine ensures that code is not
> executing on the processors when the bits are flipped. That will ensure
> that there is no need to do any get_online_cpu() nastiness in critical VM
> paths since we are guaranteed not to be executing them.

And how does that solve the problem at hand? Not at all:

CPU 0 CPU 1

for_each_online_cpu(cpu)
==> cpu = 1
stop_machine()
set_cpu_online(1, false)
queue_work(cpu1)

Thanks,

tglx