Re: [PATCH V4 4/7] perf/x86/intel: Support LBR event logging

["Liang, Kan" <kan.liang@linux.intel.com>]

On 2023-10-19 7:09 a.m., Peter Zijlstra wrote:
> On Wed, Oct 04, 2023 at 11:40:41AM -0700, kan.liang@linux.intel.com wrote:
> 
>> @@ -3905,6 +3915,44 @@ static int intel_pmu_hw_config(struct perf_event *event)
>>  	if (needs_branch_stack(event) && is_sampling_event(event))
>>  		event->hw.flags  |= PERF_X86_EVENT_NEEDS_BRANCH_STACK;
>>  
>> +	if (branch_sample_counters(event)) {
>> +		struct perf_event *leader, *sibling;
>> +
>> +		if (!(x86_pmu.flags & PMU_FL_LBR_EVENT) ||
>> +		    (event->attr.config & ~INTEL_ARCH_EVENT_MASK))
>> +			return -EINVAL;
>> +
>> +		/*
>> +		 * The event logging is not supported in the call stack mode
>> +		 * yet, since we cannot simply flush the LBR during e.g.,
>> +		 * multiplexing. Also, there is no obvious usage with the call
>> +		 * stack mode. Simply forbids it for now.
>> +		 *
>> +		 * If any events in the group enable the LBR event logging
>> +		 * feature, the group is treated as a LBR event logging group,
>> +		 * which requires the extra space to store the counters.
>> +		 */
>> +		leader = event->group_leader;
>> +		if (branch_sample_call_stack(leader))
>> +			return -EINVAL;
>> +		leader->hw.flags  |= PERF_X86_EVENT_BRANCH_COUNTERS;
> 
> (superfluous whitespace before operator)
> 
>> +
>> +		for_each_sibling_event(sibling, leader) {
>> +			if (branch_sample_call_stack(sibling))
>> +				return -EINVAL;
>> +		}
>> +
>> +		/*
>> +		 * Only applying the PERF_SAMPLE_BRANCH_COUNTERS doesn't
>> +		 * require any branch stack setup.
>> +		 * Clear the bit to avoid unnecessary branch stack setup.
>> +		 */
>> +		if (0 == (event->attr.branch_sample_type &
>> +			  ~(PERF_SAMPLE_BRANCH_PLM_ALL |
>> +			    PERF_SAMPLE_BRANCH_COUNTERS)))
>> +			event->hw.flags  &= ~PERF_X86_EVENT_NEEDS_BRANCH_STACK;
>> +	}
> 
> Does this / should this check the number of group members vs supported
> number of lbr counters?

Sure, I will add the check here for the numbers, so perf can error out
earlier.

Thanks,
Kan