Messages in this thread |  | | | Date | Tue, 11 Nov 2008 16:11:51 +0100 | | From | "Dmitry Adamushko" <> | | Subject | Re: [Bug #11989] Suspend failure on NForce4-based boards due to chanes in stop_machine |
2008/11/11 Vegard Nossum <vegard.nossum@gmail.com>:
> On Tue, Nov 11, 2008 at 11:52 AM, Ingo Molnar <mingo@elte.hu> wrote:
>> [ Cc:-ed workqueue/locking/suspend-race-condition experts. ]
>>
>> Seems like the new kernel/stop_machine.c logic has a race for the test
>> sequence above. (Below is the bisected commit again, maybe the race is
>> visible via email review as well.)
>
> I try again.
>
> I think that the test for stop_machine_data in stop_cpu() should not
> have been moved from __stop_machine().
Do you mean the following test?
if (!active_cpus) {
if (cpu == first_cpu(cpu_online_map))
smdata = &active;
} else {
if (cpu_isset(cpu, *active_cpus))
smdata = &active;
}
> Because now cpu_online_map may
> change in-between calls to stop_cpu() (if the callback tries to
> online/offline CPUs), and the end result may be different.
take_cpu_down() may not run earlier than stop_cpu() on all the cpus
have completed the STOPMACHINE_DISABLE_IRQ step, iow. "state ==
STOPMACHINE_RUN". By that moment, 'smdata' has been set up on all
cpus... if this is the case you had in mind.
>
> Maybe?
>
>
> Vegard
>
--
Best regards,
Dmitry Adamushko
|  |