Re: [Patch] cpufreq: fix a deadlock during shutting down

[David Rientjes]

On Tue, 9 Feb 2010, Amerigo Wang wrote:

> Benjamin reported that, the machine deadlocks right after printing the
> following when doing a shutdown:
> 
> halt/4071 is trying to acquire lock:
>  (s_active){++++.+}, at: [<c0000000001ef868>] .sysfs_addrm_finish+0x58/0xc0
> 
> but task is already holding lock:
>  (&per_cpu(cpu_policy_rwsem, cpu)){+.+.+.}, at: [<c0000000004cd6ac>] .lock_policy_rwsem_write+0x84/0xf4
> 
> which lock already depends on the new lock.
> 
> the existing dependency chain (in reverse order) is:
> 
> <nothing else ... machine deadlocked here>
> 
> 
> This is because we are trying to kobject_put() a kobject while
> we are holding cpu policy rwsem. So just move kobject_put()
> down after releasing the rwsem.
> 
> Totally untested.
> 
> Reported-by: Xiaotian Feng <xtfeng@xxxxxxxxx>
> Reported-by: Benjamin Herrenschmidt <benh@xxxxxxxxxxxxxxxxxxx>
> Signed-off-by: WANG Cong <amwang@xxxxxxxxxx>
> Cc: Dave Jones <davej@xxxxxxxxxx>
> Cc: Thomas Renninger <trenn@xxxxxxx>
> Cc: Prarit Bhargava <prarit@xxxxxxxxxx>
> Cc: Venkatesh Pallipadi <venkatesh.pallipadi@xxxxxxxxx>
> 
> ---
> diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
> index 67bc2ec..222b35f 100644
> --- a/drivers/cpufreq/cpufreq.c
> +++ b/drivers/cpufreq/cpufreq.c
> @@ -1113,6 +1113,7 @@ static int __cpufreq_remove_dev(struct sys_device *sys_dev)
>  	unsigned int cpu = sys_dev->id;
>  	unsigned long flags;
>  	struct cpufreq_policy *data;
> +	struct kobject *kobj;
>  #ifdef CONFIG_SMP
>  	struct sys_device *cpu_sys_dev;
>  	unsigned int j;
> @@ -1192,7 +1193,7 @@ static int __cpufreq_remove_dev(struct sys_device *sys_dev)
>  	if (cpufreq_driver->target)
>  		__cpufreq_governor(data, CPUFREQ_GOV_STOP);
>  
> -	kobject_put(&data->kobj);
> +	kobj = &data->kobj;
>  
>  	/* we need to make sure that the underlying kobj is actually
>  	 * not referenced anymore by anybody before we proceed with

NACK.

If you read this comment, it says:

	/* we need to make sure that the underlying kobj is actually
	 * not referenced anymore by anybody before we proceed with
	 * unloading.
	 */

That would suggest that the wait_for_completion(&data->kobj_unregister); 
would never actually return if you're holding a reference to it in your 
patch since it only completes when the last reference is dropped (the 
->release function is cpufreq_sysfs_release()).
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/