Re: "PM / wakeup: Show wakeup sources stats in sysfs" causes boot warnings

[Tri Vo <trong@android.com>]

On Tue, Aug 13, 2019 at 3:35 PM Stephen Boyd <swboyd@chromium.org> wrote:
>
> Quoting Qian Cai (2019-08-13 14:32:56)
> > The linux-next commit "PM / wakeup: Show wakeup sources stats in sysfs" [1]
> > introduced some baddies during boot on several x86 servers. Reverted the commit
> > fixed the issue.
> >
> > [1] https://lore.kernel.org/lkml/20190807014846.143949-4-trong@android.com/
> >
> > [   39.195053][    T1] serio: i8042 KBD port at 0x60,0x64 irq 1
> > [   39.197347][    T1] kobject_add_internal failed for wakeup (error: -2 parent:
> > serio0)
> > [   39.199845][    T1] INFO: trying to register non-static key.
> > [   39.201582][    T1] the code is fine but needs lockdep annotation.
> > [   39.203477][    T1] turning off the locking correctness validator.
> > [   39.205399][    T1] CPU: 12 PID: 1 Comm: swapper/0 Not tainted 5.3.0-rc4-
> > next-20190813 #3
> > [   39.207938][    T1] Hardware name: HP ProLiant XL420 Gen9/ProLiant XL420
> > Gen9, BIOS U19 12/27/2015
> > [   39.210606][    T1] Call Trace:
> > [   39.210606][    T1]  dump_stack+0x62/0x9a
> > [   39.210606][    T1]  register_lock_class+0x95a/0x960
> > [   39.210606][    T1]  ? __platform_driver_probe+0xcd/0x230
> > [   39.210606][    T1]  ? __platform_create_bundle+0xc0/0xe0
> > [   39.210606][    T1]  ? i8042_init+0x4ec/0x578
> > [   39.210606][    T1]  ? do_one_initcall+0xfe/0x45a
> > [   39.219571][    T1]  ? kernel_init_freeable+0x614/0x6a7
> > [   39.219571][    T1]  ? kernel_init+0x11/0x138
> > [   39.219571][    T1]  ? ret_from_fork+0x35/0x40
> > [   39.219571][    T1]  ? is_dynamic_key+0xf0/0xf0
> > [   39.219571][    T1]  ? rwlock_bug.part.0+0x60/0x60
> > [   39.219571][    T1]  ? __debug_check_no_obj_freed+0x8e/0x250
> > [   39.219571][    T1]  __lock_acquire.isra.13+0x5f/0x830
> > [   39.229491][    T1]  ? __debug_check_no_obj_freed+0x152/0x250
> > [   39.229491][    T1]  lock_acquire+0x107/0x220
> > [   39.229491][    T1]  ? __pm_relax.part.2+0x21/0xa0
> > [   39.229491][    T1]  _raw_spin_lock_irqsave+0x35/0x50
> > [   39.229491][    T1]  ? __pm_relax.part.2+0x21/0xa0
> > [   39.229491][    T1]  __pm_relax.part.2+0x21/0xa0
> > [   39.239588][    T1]  wakeup_source_destroy.part.3+0x18/0x190
> > [   39.239588][    T1]  wakeup_source_register+0x43/0x50
>
> We shouldn't call wakeup_source_destroy() from the error path in
> wakeup_source_register() because that calls __pm_relax() and that takes
> a lock that isn't initialized until wakeup_source_add() is called. Can
> you try this patch?

Right, that makes sense. Thanks for sending a fix, Stephen!

What's the preferred procedure for merging this fix? Should we apply
this commit on top of pm tree? Or should I resend a new version of the
offending patch? Sorry, I'm still new to this.