Re: [PATCH v2] perf/x86/intel: Add support for MISPREDICT bit on Knights Landing cpus

From: Jacek Tomaka
Date: Mon Jul 30 2018 - 12:34:08 EST


Ah, right:Â
/*
Â* Due to lack of segmentation in Linux the effective address (offset)
Â* is the same as the linear address, allowing us to merge the LIP and EIP
Â* LBR formats.
Â*/

Yeah,ÂLBR_FORMAT_EIP_FLAGS is ok as well. Would it be preffered?

On Tue, Jul 31, 2018 at 12:29 AM, Jacek Tomaka <jacekt@xxxxxxx> wrote:
I do not understand the difference between linear address vs effective address but LBR_FORMAT_EIP_FLAGS implies effective address, no?

On Tue, Jul 31, 2018 at 12:17 AM, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
On Mon, Jul 30, 2018 at 10:28:13PM +0800, Jacek Tomaka wrote:
> From: Jacek Tomaka <jacek.tomaka@xxxxxxxxx>
>
> Problem: perf did not show branch predicted/mispredicted bit in brstack.
>
> Output of perf -F brstack for profile collected
>
> Before:
> 0x4fdbcd/0x4fdc03/-/-/-/0
> 0x45f4c1/0x4fdba0/-/-/-/0
> 0x45f544/0x45f4bb/-/-/-/0
> 0x45f555/0x45f53c/-/-/-/0
> 0x7f66901cc24b/0x45f555/-/-/-/0
> 0x7f66901cc22e/0x7f66901cc23d/-/-/-/0
> 0x7f66901cc1ff/0x7f66901cc20f/-/-/-/0
> 0x7f66901cc1e8/0x7f66901cc1fc/-/-/-/0
>
> After:
> 0x4fdbcd/0x4fdc03/P/-/-/0
> 0x45f4c1/0x4fdba0/P/-/-/0
> 0x45f544/0x45f4bb/P/-/-/0
> 0x45f555/0x45f53c/P/-/-/0
> 0x7f66901cc24b/0x45f555/P/-/-/0
> 0x7f66901cc22e/0x7f66901cc23d/P/-/-/0
> 0x7f66901cc1ff/0x7f66901cc20f/P/-/-/0
> 0x7f66901cc1e8/0x7f66901cc1fc/P/-/-/0
>
> Cause:
> As mentioned in Software Development Manual vol 3, 17.4.8.1,
> IA32_PERF_CAPABILITIES[5:0] indicates the format of the address that is
> stored in the LBR stack. Knights Landing reports 1 (LBR_FORMAT_LIP) as
> its format. Despite that, registers containing FROM address of the branch,
> do have MISPREDICT bit but because of the format indicated in
> IA32_PERF_CAPABILITIES[5:0], LBR did not read MISPREDICT bit.
>
> Solution:
> Teach LBR about above Knights Landing quirk and make it read MISPREDICT bit.
> ---
>Â arch/x86/events/intel/lbr.c | 6 +++++-
>Â 1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/arch/x86/events/intel/lbr.c b/arch/x86/events/intel/lbr.c
> index cf372b9055..043aa09f3a 100644
> --- a/arch/x86/events/intel/lbr.c
> +++ b/arch/x86/events/intel/lbr.c
> @@ -19,7 +19,7 @@ enum {
>Â Â Â ÂLBR_FORMAT_MAX_KNOWNÂ Â = LBR_FORMAT_TIME,
>Â };

> -static const enum {
> +static enum {
>Â Â Â ÂLBR_EIP_FLAGSÂ Â Â Â Â Â= 1,
>Â Â Â ÂLBR_TSXÂ Â Â Â Â Â Â Â Â= 2,
>Â } lbr_desc[LBR_FORMAT_MAX_KNOWN + 1] = {
> @@ -1230,4 +1230,8 @@ void intel_pmu_lbr_init_knl(void)

>Â Â Â Âx86_pmu.lbr_sel_mask = LBR_SEL_MASK;
>   Âx86_pmu.lbr_sel_map = snb_lbr_sel_map;
> +
> +Â Â Â/* Knights Landing does have MISPREDICT bit */
> +Â Â Âif (x86_pmu.intel_cap.lbr_format == LBR_FORMAT_LIP)
> +Â Â Â Â Â Â Âlbr_desc[LBR_FORMAT_LIP] |= LBR_EIP_FLAGS;
>Â }

So why not set lbr_format to LBR_FORMAT_EIP_FLAGS ?



--
Jacek Tomaka
Geophysical Software Developer




DownUnder GeoSolutions

76 Kings Park Road
West Perth 6005 WA, Australia
tel +61 8 9287 4143
jacekt@xxxxxxx
www.dug.com



--
Jacek Tomaka
Geophysical Software Developer




DownUnder GeoSolutions

76 Kings Park Road
West Perth 6005 WA, Australia
tel +61 8 9287 4143
jacekt@xxxxxxx
www.dug.com