Re: [PATCH v4 1/3] perf/core: store context switch out type into Perf trace

[Alexey Budankov]

Hi,

On 29.03.2018 18:43, Mark Rutland wrote:
> Hi,
> 
> On Thu, Mar 29, 2018 at 06:02:48PM +0300, Alexey Budankov wrote:
>> diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h
>> index 912b85b52344..cd6ad7e13824 100644
>> --- a/include/uapi/linux/perf_event.h
>> +++ b/include/uapi/linux/perf_event.h
>> @@ -655,6 +655,10 @@ struct perf_event_mmap_page {
>>   * perf_event_attr::precise_ip.
>>   */
>>  #define PERF_RECORD_MISC_EXACT_IP		(1 << 14)
>> +/*
>> + * Indicates that thread was preempted in TASK_RUNNING state
>> + */
>> +#define PERF_RECORD_MISC_SWITCH_OUT_PREEMPT	(1 << 14)
>>  /*
>>   * Reserve the last bit to indicate some extended misc field
>>   */
> 
> Could you please add a description (in the header) describing why it's safe to
> reuse this bit? i.e. describe which events these are valid for, so it's clear
> they don't overlap.

It is currently employed only on x86 architecture when reporting precise events 
samples with exact bit set in pt_regs->flags. Apparently there is no conflict on 
this bit among different trace records so the message could be like this:

/*
 * These PERF_RECORD_MISC_* flags are safely reused for the following events:
 *
 *   PERF_RECORD_MISC_PRECISE_IP         - PERF_RECORD_SAMPLE of precise events
 *   PERF_RECORD_MISC_SWITCH_OUT_PREEMPT - PERF_RECORD_SWITCH* events
 */

Thanks,
Alexey

> 
> For bit 13 we have:
> 
> /*
>  * Following PERF_RECORD_MISC_* are used on different
>  * events, so can reuse the same bit position:
>  *
>  *   PERF_RECORD_MISC_MMAP_DATA  - PERF_RECORD_MMAP* events
>  *   PERF_RECORD_MISC_COMM_EXEC  - PERF_RECORD_COMM event
>  *   PERF_RECORD_MISC_SWITCH_OUT - PERF_RECORD_SWITCH* events
>  */
> #define PERF_RECORD_MISC_MMAP_DATA              (1 << 13)
> #define PERF_RECORD_MISC_COMM_EXEC              (1 << 13)
> #define PERF_RECORD_MISC_SWITCH_OUT             (1 << 13)
> 
> Thanks,
> Mark.
>