[bug report] PM / reboot: Eliminate race between reboot and suspend

[Rafael J. Wysocki]

On Thu, Jan 21, 2021 at 10:14 AM Pingfan Liu <kernelfans at gmail.com> wrote:
>
> On Wed, Jan 20, 2021 at 5:30 PM Baoquan He <bhe at redhat.com> wrote:
> >
> > Hi,
> >
> > On 11/18/20 at 09:59pm, Dan Carpenter wrote:
> > > Hello Pingfan Liu,
> > >
> > > The patch 55f2503c3b69: "PM / reboot: Eliminate race between reboot
> > > and suspend" from Jul 31, 2018, leads to the following static checker
> > > warning:
> > >
> > >       kernel/power/main.c:27 lock_system_sleep()
> > >       warn: called with lock held.  '&system_transition_mutex'
> >
> > This is a good finding. I think we can simply remove the lock/unlock
> > pair of system_transition_mutex in kernel_kexec() function. The dead
> > lock should be easily triggered, but it hasn't caused any failure report
> > because the feature 'kexec jump' is almost not used by anyone as far as
> > I know. We may need to find out who is using it and where it's used
> > through an inquiry. Before that, we can just remove the lock operation
> > inside CONFIG_KEXEC_JUMP ifdeffery scope. Thanks.
> >
> >
> > diff --git a/kernel/kexec_core.c b/kernel/kexec_core.c
> > index 80905e5aa8ae..a0b6780740c8 100644
> > --- a/kernel/kexec_core.c
> > +++ b/kernel/kexec_core.c
> > @@ -1134,7 +1134,6 @@ int kernel_kexec(void)
> >
> >  #ifdef CONFIG_KEXEC_JUMP
> >         if (kexec_image->preserve_context) {
> > -               lock_system_sleep();
> >                 pm_prepare_console();
> >                 error = freeze_processes();
> >                 if (error) {
> > @@ -1197,7 +1196,6 @@ int kernel_kexec(void)
> >                 thaw_processes();
> >   Restore_console:
> >                 pm_restore_console();
> > -               unlock_system_sleep();
>
> This should work since the only caller syscall_reboot has already
> placed kernel_kexec() under the protection of system_transition_mutex.
>
> Thanks for the fix.
>
> Reviewed-by: Pingfan Liu <kernelfans at gmail.com>

OK, so can anyone please submit that patch formally (Cc linux-pm, please)?