Re: [PATCH 2/6] perf/x86: Fix data source decoding for Skylake

From: Peter Zijlstra
Date: Tue Jun 06 2017 - 06:08:47 EST


On Mon, Jun 05, 2017 at 03:48:34PM -0700, Andi Kleen wrote:

> +void __init intel_pmu_pebs_data_source_skl(void)
> +{
> + pebs_data_source[0x08] = OP_LH | P(LVLX, L4) | P(SNOOP, HIT);
> + pebs_data_source[0x09] = OP_LH | P(LVLX, L4) | P(LVLX, REMOTE) | P(SNOOP, HIT);
> + pebs_data_source[0x0b] = OP_LH | P(LVLX, RAM) | P(LVLX, REMOTE) | P(SNOOP, NONE);
> + pebs_data_source[0x0c] = OP_LH | P(LVL, NA) | P(LVLX, REMOTE) | P(SNOOPX, FWD);
> + pebs_data_source[0x0d] = OP_LH | P(LVL, NA) | P(LVLX, REMOTE) | P(SNOOP, HITM);
> +}

Not too happy about that..

P(LVLX, L4) | P(LVLX, REMOTE)

reads like something that should be PERF_MEM_LVL_REM_CCE1 or something
and

P(LVLX, RAM) | P(LVLX, REMOTE)

Should certainly be

PERF_MEM_LVL_REM_RAM1


This new generic 'REMOTE' has too much overlap with the existing things.

I realize the pickle you're in, but urgh..

Stephane?


> diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h
> index b1c0b187acfe..4b5deeada34b 100644
> --- a/include/uapi/linux/perf_event.h
> +++ b/include/uapi/linux/perf_event.h
> @@ -931,14 +931,18 @@ union perf_mem_data_src {
> mem_snoop:5, /* snoop mode */
> mem_lock:2, /* lock instr */
> mem_dtlb:7, /* tlb access */
> - mem_rsvd:31;
> + mem_lvlx:8, /* memory hierarchy level, ext */
> + mem_snoopx:2, /* snoop mode, ext */
> + mem_rsvd:21;
> };
> };
> #elif defined(__BIG_ENDIAN_BITFIELD)
> union perf_mem_data_src {
> __u64 val;
> struct {
> - __u64 mem_rsvd:31,
> + __u64 mem_rsvd:21,
> + mem_snoopx:2, /* snoop mode, ext */
> + mem_lvlx:8, /* memory hierarchy level, ext */
> mem_dtlb:7, /* tlb access */
> mem_lock:2, /* lock instr */
> mem_snoop:5, /* snoop mode */
> @@ -975,6 +979,13 @@ union perf_mem_data_src {
> #define PERF_MEM_LVL_UNC 0x2000 /* Uncached memory */
> #define PERF_MEM_LVL_SHIFT 5
>
> +#define PERF_MEM_LVLX_REMOTE 0x01 /* Remote */
> +#define PERF_MEM_LVLX_L4 0x02 /* L4 */
> +#define PERF_MEM_LVLX_RAM 0x04 /* Ram */
> +/* 5 free */
> +
> +#define PERF_MEM_LVLX_SHIFT 33
> +
> /* snoop mode */
> #define PERF_MEM_SNOOP_NA 0x01 /* not available */
> #define PERF_MEM_SNOOP_NONE 0x02 /* no snoop */
> @@ -983,6 +994,10 @@ union perf_mem_data_src {
> #define PERF_MEM_SNOOP_HITM 0x10 /* snoop hit modified */
> #define PERF_MEM_SNOOP_SHIFT 19
>
> +#define PERF_MEM_SNOOPX_FWD 0x01 /* forward */
> +/* 1 free */
> +#define PERF_MEM_SNOOPX_SHIFT 41
> +
> /* locked instruction */
> #define PERF_MEM_LOCK_NA 0x01 /* not available */
> #define PERF_MEM_LOCK_LOCKED 0x02 /* locked transaction */
> --
> 2.9.4
>