On Fri, Apr 28, 2017 at 06:27:28PM +0200, Sebastian Andrzej Siewior wrote:
> Upstream commit dc434e056fe1dada20df7ba07f32739d3a701adf
>
> The setup/remove_state/instance() functions in the hotplug core code are
> serialized against concurrent CPU hotplug, but unfortunately not serialized
> against themself.
>
> As a consequence a concurrent invocation of these function results in
> corruption of the callback machinery because two instances try to invoke
> callbacks on remote cpus at the same time. This results in missing callback
> invocations and initiator threads waiting forever on the completion.
>
> The obvious solution to replace get_cpu_online() with cpu_hotplug_begin()
> is not possible because at least one callsite calls into these functions
> from a get_online_cpu() locked region.
>
> Extend the protection scope of the cpuhp_state_mutex from solely protecting
> the state arrays to cover the callback invocation machinery as well.
>
> Fixes: 5b7aa87e0482 ("cpu/hotplug: Implement setup/removal interface")
> Reported-and-tested-by: Bart Van Assche <Bart.VanAssche@xxxxxxxxxxx>
> Signed-off-by: Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>
> Cc: hpa@xxxxxxxxx
> Cc: mingo@xxxxxxxxxx
> Cc: akpm@xxxxxxxxxxxxxxxxxxxx
> Cc: torvalds@xxxxxxxxxxxxxxxxxxxx
> Link: http://lkml.kernel.org/r/20170314150645.g4tdyoszlcbajmna@xxxxxxxxxxxxx
> Signed-off-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> ---
> kernel/cpu.c | 28 ++++++++++++++--------------
> 1 file changed, 14 insertions(+), 14 deletions(-)
Doesn't apply to 4.9-stable, can you provide a backport there if you
want it applied to that tree (I think you do...)
thanks,
greg k-h