[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200117085823.GV2827@hirez.programming.kicks-ass.net>
Date: Fri, 17 Jan 2020 09:58:23 +0100
From: Peter Zijlstra <peterz@...radead.org>
To: kan.liang@...ux.intel.com
Cc: mingo@...nel.org, linux-kernel@...r.kernel.org, ak@...ux.intel.com
Subject: Re: [RESEND PATCH] perf/x86/intel: Fix inaccurate period in context
switch for auto-reload
On Thu, Jan 16, 2020 at 10:31:54AM -0800, kan.liang@...ux.intel.com wrote:
> From: Kan Liang <kan.liang@...ux.intel.com>
>
> Perf doesn't take the left period into account when auto-reload is
> enabled with fixed period sampling mode in context switch.
> Here is the ftrace when recording PEBS event with fixed period.
>
> #perf record -e cycles:p -c 2000000 -- ./triad_loop
>
> //Task is scheduled out
> triad_loop-17222 [000] d... 861765.878032: write_msr:
> MSR_CORE_PERF_GLOBAL_CTRL(38f), value 0 //Disable global counter
> triad_loop-17222 [000] d... 861765.878033: write_msr:
> MSR_IA32_PEBS_ENABLE(3f1), value 0 //Disable PEBS
> triad_loop-17222 [000] d... 861765.878033: write_msr:
> MSR_P6_EVNTSEL0(186), value 40003003c //Disable the counter
> triad_loop-17222 [000] d... 861765.878033: rdpmc: 0, value
> fffffff82840 //Read value of the counter
> triad_loop-17222 [000] d... 861765.878034: write_msr:
> MSR_CORE_PERF_GLOBAL_CTRL(38f), value 1000f000000ff //Re-enable global
> counter
This is unreadable garbage, please don't wrap trace output.
Powered by blists - more mailing lists