Re: WARN_ON_ONCE() hit at kernel/events/core.c:330

[Alexander Shishkin <alexander.shishkin@linux.intel.com>]

Peter Zijlstra <peterz@infradead.org> writes:

> On Wed, Apr 10, 2019 at 03:51:24PM +0200, Thomas-Mich Richter wrote:
>> Thanks for the fix with commit id 86071b11317550d994b55ce5e31aa06bcad783b5.
>> 
>> However doing an fgrep on the pending_disable member of struct perf_event
>> reveals two more hits in file kernel/events/ringbuffer.c when events
>> have auxiliary data.
>> 
>> Not sure if this needs to be addresses too, just wanted to let you know.
>
> *groan* indeed, and yes that would need fixing too. I completely missed
> the AUX stuff using that too.
>
> I think we can simply do the below on top, Alexander?

Looks good. FWIW, renaming pending_disable in the first place to match
the fact that it's a cpu# now, would have probably caught all these.

> ---
>  kernel/events/ring_buffer.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/kernel/events/ring_buffer.c b/kernel/events/ring_buffer.c
> index 678ccec60d8f..ab2b7b38adc5 100644
> --- a/kernel/events/ring_buffer.c
> +++ b/kernel/events/ring_buffer.c
> @@ -392,7 +392,7 @@ void *perf_aux_output_begin(struct perf_output_handle *handle,
>  		 * store that will be enabled on successful return
>  		 */
>  		if (!handle->size) { /* A, matches D */
> -			event->pending_disable = 1;
> +			event->pending_disable = smp_processor_id();
>  			perf_output_wakeup(handle);
>  			local_set(&rb->aux_nest, 0);
>  			goto err_put;
> @@ -480,7 +480,7 @@ void perf_aux_output_end(struct perf_output_handle *handle, unsigned long size)
>
>  	if (wakeup) {
>  		if (handle->aux_flags & PERF_AUX_FLAG_TRUNCATED)
> -			handle->event->pending_disable = 1;
> +			handle->event->pending_disable = smp_processor_id();
>  		perf_output_wakeup(handle);
>  	}

Thanks,
--
Alex