Lockdep warnings on kexec (virtio_blk, hrtimers)

[Thomas Gleixner]

On Thu, Dec 12 2024 at 13:46, David Woodhouse wrote:
> On Thu, 2024-12-12 at 14:34 +0100, Thomas Gleixner wrote:
>> 
>> David, can you retest with the debug patch below? That should pin-point
>> the real culprit.
>
> B[    1.545489] ------------[ cut here ]------------
> [    1.546338] DEBUG_LOCKS_WARN_ON(suspend_syscore_active)
> [    1.546375] WARNING: CPU: 0 PID: 18 at kernel/locking/lockdep.c:4471 lockdep_hardirqs_on+0x13a/0x140

Now this starts to be completely mysterious:

> [    1.571399]  finish_task_switch.isra.0+0xc4/0x2d0
> [    1.572062]  __schedule+0x50a/0x1a10

How on earth did this end up in schedule() on the rcu_gp_kthread between
syscore_suspend() and syscore_resume()?

> [    1.575896]  rcu_gp_fqs_loop+0x10b/0x5b0
> [    1.576455]  ? _raw_spin_unlock_irq+0x28/0x50
> [    1.577072]  rcu_gp_kthread+0xf8/0x1b0
> [    1.577621]  kthread+0xd5/0x100
> [    1.578066]  ? __pfx_kthread+0x10/0x10

I tried to reproduce, but failed. Do you have a simple reproducer
recipe?

Thanks,

        tglx