Re: [PATCH v3 bpf-next 1/3] perf: enable branch record for software events

[Peter Zijlstra <peterz@infradead.org>]

On Mon, Aug 30, 2021 at 02:41:04PM -0700, Song Liu wrote:

> diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c
> index ac6fd2dabf6a2..d28d0e12c112c 100644
> --- a/arch/x86/events/intel/core.c
> +++ b/arch/x86/events/intel/core.c
> @@ -2155,9 +2155,9 @@ static void __intel_pmu_disable_all(void)
>  
>  static void intel_pmu_disable_all(void)
>  {
> +	intel_pmu_lbr_disable_all();
>  	__intel_pmu_disable_all();
>  	intel_pmu_pebs_disable_all();
> -	intel_pmu_lbr_disable_all();
>  }

Hurmph... I'm not sure about that, I'd rather you sprinkle a few
__always_inline to ensure no actual function is called while you disable
things in the correct order.

You now still have a hole vs PMI.

> +static int
> +intel_pmu_snapshot_branch_stack(struct perf_branch_snapshot *br_snapshot)
> +{
> +	struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events);

Note that this requires preemption is disabled, then look at the
call-sites in your next patch and spot the problem...

> +
> +	intel_pmu_disable_all();
> +	intel_pmu_lbr_read();
> +	memcpy(br_snapshot->entries, cpuc->lbr_entries,
> +	       sizeof(struct perf_branch_entry) * x86_pmu.lbr_nr);
> +	br_snapshot->nr = x86_pmu.lbr_nr;
> +	intel_pmu_enable_all(0);
> +	return 0;
> +}
> +
>  /*
>   * Workaround for:
>   *   Intel Errata AAK100 (model 26)
> @@ -6283,9 +6297,15 @@ __init int intel_pmu_init(void)
>  			x86_pmu.lbr_nr = 0;
>  	}
>  
> -	if (x86_pmu.lbr_nr)
> +	if (x86_pmu.lbr_nr) {
>  		pr_cont("%d-deep LBR, ", x86_pmu.lbr_nr);
>  
> +		/* only support branch_stack snapshot for perfmon >= v2 */
> +		if (x86_pmu.disable_all == intel_pmu_disable_all)
								  {
> +			static_call_update(perf_snapshot_branch_stack,
> +					   intel_pmu_snapshot_branch_stack);

		}

> +	}
> +
>  	intel_pmu_check_extra_regs(x86_pmu.extra_regs);
>  
>  	/* Support full width counters using alternative MSR range */

> diff --git a/kernel/events/core.c b/kernel/events/core.c
> index 011cc5069b7ba..22807864e913b 100644
> --- a/kernel/events/core.c
> +++ b/kernel/events/core.c
> @@ -13437,3 +13437,6 @@ struct cgroup_subsys perf_event_cgrp_subsys = {
>  	.threaded	= true,
>  };
>  #endif /* CONFIG_CGROUP_PERF */
> +
> +DEFINE_STATIC_CALL_RET0(perf_snapshot_branch_stack,
> +			perf_snapshot_branch_stack_t);

I'll squint and accept 82 characters :-)