Re: regression 3.11-rc1: rmmod hangs after tracing module

["Arend van Spriel" <arend@broadcom.com>]

On 07/23/2013 08:44 PM, Steven Rostedt wrote:
> On Tue, 2013-07-23 at 20:07 +0200, Arend van Spriel wrote:
>> Hi Steven,
>>
>> During testing of our brcmfmac wireless driver I had a trace-cmd running
>> along (trace-cmd record -e brcmfmac:*). After the test I stopped the
>> trace, unplugged my usb device and did a rmmod resulting in the attached
>> lockdep splat. Hope you have an idea what is going wrong here.
>
> Thanks for the report!
>
> Does this fix you bug?

Thanks Steve,

Tested your patch and it indeed fixes my deadlock.

Regards,
Arend

> -- Steve
>
> diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
> index 69cba47..882ec1d 100644
> --- a/kernel/trace/trace.c
> +++ b/kernel/trace/trace.c
> @@ -1224,18 +1224,17 @@ void tracing_reset_current(int cpu)
>   	tracing_reset(&global_trace.trace_buffer, cpu);
>   }
>
> +/* Must have trace_types_lock held */
>   void tracing_reset_all_online_cpus(void)
>   {
>   	struct trace_array *tr;
>
> -	mutex_lock(&trace_types_lock);
>   	list_for_each_entry(tr, &ftrace_trace_arrays, list) {
>   		tracing_reset_online_cpus(&tr->trace_buffer);
>   #ifdef CONFIG_TRACER_MAX_TRACE
>   		tracing_reset_online_cpus(&tr->max_buffer);
>   #endif
>   	}
> -	mutex_unlock(&trace_types_lock);
>   }
>
>   #define SAVED_CMDLINES 128
>
>
>