Re: Bulk CPU Hotplug (Was Re: [PATCH] Do not force shutdown/reboot to boot cpu.)

[Oleg Nesterov <oleg@redhat.com>]

On 04/15, Robin Holt wrote:
>
> On Sat, Apr 13, 2013 at 06:30:22PM +0200, Oleg Nesterov wrote:
> > On 04/12, Robin Holt wrote:
> > >
> > > +void migrate_to_boot_cpu(void)
> > > +{
> > > +	/* The boot cpu is always logical cpu 0 */
> > > +	int reboot_cpu_id = 0;
> > > +
> > > +	/* Make certain the cpu I'm about to reboot on is online */
> > > +	if (!cpu_online(reboot_cpu_id))
> > > +		reboot_cpu_id = smp_processor_id();
> > > +
> > > +	/* Make certain I only run on the appropriate processor */
> > > +	set_cpus_allowed_ptr(current, cpumask_of(reboot_cpu_id));
> >
> > This is only theoretical, but perhaps it makes sense to set
> > PF_THREAD_BOUND before set_cpus_allowed_ptr() ? To prevent the
> > race with another thread doing sched_setaffinity().
>
> I don't quite understand this comment.  We are migrating our own thread.
> How does setting PF_THREAD_BOUND have any effect?

Suppose that another thread does, say, sys_sched_setaffinity(our_pid)
right after set_cpus_allowed_ptr(). If we set PF_THREAD_BOUND the
next set_cpus_allowed() will fail.

Oleg.