Re: [PATCH 1/5] x86, perf: Fix LBR call stack save/restore

[Peter Zijlstra <peterz@infradead.org>]

On Tue, Oct 20, 2015 at 11:46:33AM -0700, Andi Kleen wrote:
> From: Andi Kleen <ak@linux.intel.com>
> 
> This fixes a bug added with the earlier 90405aa02. The bug
> could lead to lost LBR call stacks. When restoring the LBR
> state we need to use the TOS of the previous context, not
> the current context. To do that we need to save/restore the tos.

Current best practise also asks for:

Fixes: 90405aa02247 ("perf/x86/intel/lbr: Limit LBR accesses to TOS in callstack mode")
> Cc: <stable@vger.kernel.org> # 4.2+
> Signed-off-by: Andi Kleen <ak@linux.intel.com>

> --- a/arch/x86/kernel/cpu/perf_event_intel_lbr.c
> +++ b/arch/x86/kernel/cpu/perf_event_intel_lbr.c
> @@ -239,7 +239,7 @@ static void __intel_pmu_lbr_restore(struct x86_perf_task_context *task_ctx)
>  	}
>  
>  	mask = x86_pmu.lbr_nr - 1;
> -	tos = intel_pmu_lbr_tos();
> +	tos = task_ctx->tos;
>  	for (i = 0; i < tos; i++) {
>  		lbr_idx = (tos - i) & mask;
>  		wrmsrl(x86_pmu.lbr_from + lbr_idx, task_ctx->lbr_from[i]);
> @@ -247,6 +247,7 @@ static void __intel_pmu_lbr_restore(struct x86_perf_task_context *task_ctx)
>  		if (x86_pmu.intel_cap.lbr_format == LBR_FORMAT_INFO)
>  			wrmsrl(MSR_LBR_INFO_0 + lbr_idx, task_ctx->lbr_info[i]);
>  	}
> +	wrmsrl(x86_pmu.lbr_tos, tos);
>  	task_ctx->lbr_stack_state = LBR_NONE;
>  }

Any idea who much more expensive that wrmsr() is compared to the rdmsr()
it replaces?

If its significant we could think about having this behaviour depend on
callstacks.