Re: [PATCH] smp: Fix missed destroy_work_on_stack() calls in smp_call_on_cpu()

[Paul E. McKenney]

On Thu, Jul 04, 2024 at 02:52:13PM +0800, Zqiang wrote:
> For builts with CONFIG_DEBUG_OBJECTS_WORK=y kernels, the sscs.work
> defined using INIT_WORK_ONSTACK() will be initialized by
> debug_object_init_on_stack() for debug check in __init_work().
> This commit therefore invoke destroy_work_on_stack() to free
> sscs.work debug objects before smp_call_on_cpu() returns.
> 
> Signed-off-by: Zqiang <qiang.zhang1211@xxxxxxxxx>

I have queued this for testing.  My guess is that it should go to
mainline some other way, so:

Tested-by: Paul E. McKenney <paulmck@xxxxxxxxxx>

> ---
>  kernel/smp.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/kernel/smp.c b/kernel/smp.c
> index cc13e73a887c..61f10f982341 100644
> --- a/kernel/smp.c
> +++ b/kernel/smp.c
> @@ -1135,6 +1135,7 @@ int smp_call_on_cpu(unsigned int cpu, int (*func)(void *), void *par, bool phys)
>  
>  	queue_work_on(cpu, system_wq, &sscs.work);
>  	wait_for_completion(&sscs.done);
> +	destroy_work_on_stack(&sscs.work);
>  
>  	return sscs.ret;
>  }
> -- 
> 2.17.1
>