[RESEND]

On Tue, Jul 12, 2022 at 5:34 AM Steven Rostedt <rostedt@goodmis.org> wrote:
>
> On Mon, 11 Jul 2022 10:58:28 -1000
> Tejun Heo <tj@kernel.org> wrote:
>
> > I don't think lockdep would be able to track CPU1 -> CPU2 dependency here
> > unfortunately.
> >
> > > AFAIU:
> > >
> > >
> > > CPU0                                     CPU1                                   CPU2
> > >
> > > // attach task to a different
> > > // cpuset cgroup via sysfs
> > > __acquire(cgroup_threadgroup_rwsem)
> > >
> > >                                          // pring up CPU2 online
> > >                                          __acquire(cpu_hotplug_lock)
> > >                                          // wait for CPU2 to come online
>
> Should there be some annotation here that tells lockdep that CPU1 is now
> blocked on CPU2?
>
> Then this case would be caught by lockdep.
>
> -- Steve
>
>
> > >                                                                                 // bringup cpu online
> > >                                                                                 // call cpufreq_online() which tries to create sugov kthread
> > > __acquire(cpu_hotplug_lock)                                                     copy_process()
> > >                                                                                    cgroup_can_fork()
> > >                                                                                       cgroup_css_set_fork()
> > >                                                                                       __acquire(cgroup_threadgroup_rwsem)
> > > // blocks forever                        // blocks forever                            // blocks forever
> > >

Indeed, It's caused by threads instead of cpus.

Our soc contains two cpufreq policy.0-5 belongs to policy0, 6-7
belongs to policy1.

when cpu6/7 online

Thread-A                                                      Thread-B

cgroup_file_write                                          device_online
cgroup1_tasks_write                                              ...
__cgroup1_procs_write                                      _cpu_up
write(&cgroup_threadgroup_rwsem); <<     cpus_write_lock();<<
cgroup_attach_task                                               ......
cgroup_migrate_execute                             cpuhp_kick_ap
cpuset_attach                                              //wakeup cpuhp
cpus_read_lock()                                        //waitingfor cpuhp


cpuhp/6                                                        kthreadd

cpuhp_thread_fun
cpuhp_invoke_callback
cpuhp_cpufreq_online
cpufreq_online
 sugov_init
 __kthread_create_on_node                 copy_process
 //blocked, waiting for kthreadd             cgroup_can_fork
                                                        cgroup_css_set_fork
                           __acquires(&cgroup_threadgroup_rwsem)
                                                        //blocked

So it's logic is:
Thread-A --->Thread-B---->cpuhp--->kthreadd---->Thread-A

When cpu offline, the sugov thread would stop, so it would waiting
cgroup_threadgroup_rwsem when kthread_stop();
It's logic is:
Thread-A --->Thread-B---->cpuhp--->sugov---->Thread-A

As Qais said:
> if there's anything else that creates a kthread when a cpu goes online/offline
> then we'll hit the same problem again.

Indeed, only the cpuhp thread create/destroy kthread would cause the case.

I have put the test script in the mail, and I have tested it without
monkey test, the deadlock still occurs..


Thanks!
xuewen.yan