[PATCH] coresight: Don't immediately close events that are run on invalid CPU/sink combos

James Clark james.clark at arm.com
Wed Sep 22 05:53:42 PDT 2021



On 22/09/2021 10:24, Suzuki K Poulose wrote:
> On 21/09/2021 14:02, James Clark wrote:
[...]
>> -    event->hw.state = 0;
>> -
>>       /* Finally enable the tracer */
>>       if (source_ops(csdev)->enable(csdev, event, CS_MODE_PERF))
>>           goto fail_disable_path;
>>   +out:
>> +    /* Tell the perf core the event is alive */
>> +    event->hw.state = 0;
>>       /* Save the event_data for this ETM */
>>       ctxt->event_data = event_data;
>> -out:
>>       return;
> 
> Hi James,
> 
> I think this problematic. The "out" label is reached from the "failed"
> path. So, we would reset the STOPPED state. I think we should simply
> return when we reach the "fail" label.
> 
> Please could you resend this patch with the changes.
> 
> Kind regards
> Suzuki
> 

Oops yes I missed that, I've resubmitted. It probably would have just about
worked because from my testing it seems like the event isn't re-enabled
if perf_aux_output_end() is called from start regardless of the STOPPED/RUNNING
state.

Thanks
James



More information about the linux-arm-kernel mailing list