Re: [PATCH 2/2] perf, x86: Add Silvermont (22nm Atom) support

From: Stephane Eranian
Date: Thu Jul 18 2013 - 08:18:21 EST


On Thu, Jul 18, 2013 at 11:02 AM, Yan, Zheng <zheng.z.yan@xxxxxxxxx> wrote:
> From: "Yan, Zheng" <zheng.z.yan@xxxxxxxxx>
>
> Compare to old atom, Silvermont has offcore and has more events
> that support PEBS.
>
> Signed-off-by: Yan, Zheng <zheng.z.yan@xxxxxxxxx>
> ---
> arch/x86/kernel/cpu/perf_event.h | 2 +
> arch/x86/kernel/cpu/perf_event_intel.c | 158 ++++++++++++++++++++++++++++++
> arch/x86/kernel/cpu/perf_event_intel_ds.c | 26 +++++
> 3 files changed, 186 insertions(+)
>
> diff --git a/arch/x86/kernel/cpu/perf_event.h b/arch/x86/kernel/cpu/perf_event.h
> index 97e557b..cc16faa 100644
> --- a/arch/x86/kernel/cpu/perf_event.h
> +++ b/arch/x86/kernel/cpu/perf_event.h
> @@ -641,6 +641,8 @@ extern struct event_constraint intel_core2_pebs_event_constraints[];
>
> extern struct event_constraint intel_atom_pebs_event_constraints[];
>
> +extern struct event_constraint intel_slm_pebs_event_constraints[];
> +
> extern struct event_constraint intel_nehalem_pebs_event_constraints[];
>
> extern struct event_constraint intel_westmere_pebs_event_constraints[];
> diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
> index d312edf..e4bb30a 100644
> --- a/arch/x86/kernel/cpu/perf_event_intel.c
> +++ b/arch/x86/kernel/cpu/perf_event_intel.c
> @@ -164,6 +164,15 @@ static struct event_constraint intel_gen_event_constraints[] __read_mostly =
> EVENT_CONSTRAINT_END
> };
>
> +static struct event_constraint intel_slm_event_constraints[] __read_mostly =
> +{
> + FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */
> + FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */
> + FIXED_EVENT_CONSTRAINT(0x013c, 2), /* CPU_CLK_UNHALTED.REF */
> + FIXED_EVENT_CONSTRAINT(0x0300, 2), /* pseudo CPU_CLK_UNHALTED.REF */
> + EVENT_CONSTRAINT_END
> +};
> +
> static struct extra_reg intel_snb_extra_regs[] __read_mostly = {
> /* must define OFFCORE_RSP_X first, see intel_fixup_er() */
> INTEL_UEVENT_EXTRA_REG(0x01b7, MSR_OFFCORE_RSP_0, 0x3f807f8fffull, RSP_0),
> @@ -886,6 +895,140 @@ static __initconst const u64 atom_hw_cache_event_ids
> },
> };
>
> +static struct extra_reg intel_slm_extra_regs[] __read_mostly =
> +{
> + /* must define OFFCORE_RSP_X first, see intel_fixup_er() */
> + INTEL_UEVENT_EXTRA_REG(0x01b7, MSR_OFFCORE_RSP_0, 0x768005ffff, RSP_0),
> + INTEL_UEVENT_EXTRA_REG(0x02b7, MSR_OFFCORE_RSP_1, 0x768005ffff, RSP_1),
> + EVENT_EXTRA_END
> +};
> +
> +#define SLM_DMND_READ SNB_DMND_DATA_RD
> +#define SLM_DMND_WRITE SNB_DMND_RFO
> +#define SLM_DMND_PREFETCH (SNB_PF_DATA_RD|SNB_PF_RFO)
> +
> +#define SLM_SNP_ANY (SNB_SNP_NONE|SNB_SNP_MISS|SNB_NO_FWD|SNB_HITM)
> +#define SLM_LLC_ACCESS SNB_RESP_ANY
> +#define SLM_LLC_MISS (SLM_SNP_ANY|SNB_NON_DRAM)
> +
> +static __initconst const u64 slm_hw_cache_extra_regs
> + [PERF_COUNT_HW_CACHE_MAX]
> + [PERF_COUNT_HW_CACHE_OP_MAX]
> + [PERF_COUNT_HW_CACHE_RESULT_MAX] =
> +{
> + [ C(LL ) ] = {
> + [ C(OP_READ) ] = {
> + [ C(RESULT_ACCESS) ] = SLM_DMND_READ|SLM_LLC_ACCESS,
> + [ C(RESULT_MISS) ] = SLM_DMND_READ|SLM_LLC_MISS,
> + },
> + [ C(OP_WRITE) ] = {
> + [ C(RESULT_ACCESS) ] = SLM_DMND_WRITE|SLM_LLC_ACCESS,
> + [ C(RESULT_MISS) ] = SLM_DMND_WRITE|SLM_LLC_MISS,
> + },
> + [ C(OP_PREFETCH) ] = {
> + [ C(RESULT_ACCESS) ] = SLM_DMND_PREFETCH|SLM_LLC_ACCESS,
> + [ C(RESULT_MISS) ] = SLM_DMND_PREFETCH|SLM_LLC_MISS,
> + },
> + },
> +};
> +
> +static __initconst const u64 slm_hw_cache_event_ids
> + [PERF_COUNT_HW_CACHE_MAX]
> + [PERF_COUNT_HW_CACHE_OP_MAX]
> + [PERF_COUNT_HW_CACHE_RESULT_MAX] =
> +{
> + [ C(L1D) ] = {
> + [ C(OP_READ) ] = {
> + [ C(RESULT_ACCESS) ] = 0,
> + [ C(RESULT_MISS) ] = 0x0104, /* LD_DCU_MISS */
> + },
> + [ C(OP_WRITE) ] = {
> + [ C(RESULT_ACCESS) ] = 0,
> + [ C(RESULT_MISS) ] = 0,
> + },
> + [ C(OP_PREFETCH) ] = {
> + [ C(RESULT_ACCESS) ] = 0,
> + [ C(RESULT_MISS) ] = 0,
> + },
> + },
> + [ C(L1I ) ] = {
> + [ C(OP_READ) ] = {
> + [ C(RESULT_ACCESS) ] = 0x0380, /* ICACHE.ACCESSES */
> + [ C(RESULT_MISS) ] = 0x0280, /* ICACGE.MISSES */
> + },
> + [ C(OP_WRITE) ] = {
> + [ C(RESULT_ACCESS) ] = -1,
> + [ C(RESULT_MISS) ] = -1,
> + },
> + [ C(OP_PREFETCH) ] = {
> + [ C(RESULT_ACCESS) ] = 0,
> + [ C(RESULT_MISS) ] = 0,
> + },
> + },
> + [ C(LL ) ] = {
> + [ C(OP_READ) ] = {
> + /* OFFCORE_RESPONSE.ANY_DATA.LOCAL_CACHE */
> + [ C(RESULT_ACCESS) ] = 0x01b7,
> + /* OFFCORE_RESPONSE.ANY_DATA.ANY_LLC_MISS */
> + [ C(RESULT_MISS) ] = 0x01b7,
> + },
> + [ C(OP_WRITE) ] = {
> + /* OFFCORE_RESPONSE.ANY_RFO.LOCAL_CACHE */
> + [ C(RESULT_ACCESS) ] = 0x01b7,
> + /* OFFCORE_RESPONSE.ANY_RFO.ANY_LLC_MISS */
> + [ C(RESULT_MISS) ] = 0x01b7,
> + },
> + [ C(OP_PREFETCH) ] = {
> + /* OFFCORE_RESPONSE.PREFETCH.LOCAL_CACHE */
> + [ C(RESULT_ACCESS) ] = 0x01b7,
> + /* OFFCORE_RESPONSE.PREFETCH.ANY_LLC_MISS */
> + [ C(RESULT_MISS) ] = 0x01b7,
> + },
> + },
> + [ C(DTLB) ] = {
> + [ C(OP_READ) ] = {
> + [ C(RESULT_ACCESS) ] = 0,
> + [ C(RESULT_MISS) ] = 0x0804, /* LD_DTLB_MISS */
> + },
> + [ C(OP_WRITE) ] = {
> + [ C(RESULT_ACCESS) ] = 0,
> + [ C(RESULT_MISS) ] = 0,
> + },
> + [ C(OP_PREFETCH) ] = {
> + [ C(RESULT_ACCESS) ] = 0,
> + [ C(RESULT_MISS) ] = 0,
> + },
> + },
> + [ C(ITLB) ] = {
> + [ C(OP_READ) ] = {
> + [ C(RESULT_ACCESS) ] = 0x00c0, /* INST_RETIRED.ANY_P */
> + [ C(RESULT_MISS) ] = 0x0282, /* ITLB.MISSES */
> + },
> + [ C(OP_WRITE) ] = {
> + [ C(RESULT_ACCESS) ] = -1,
> + [ C(RESULT_MISS) ] = -1,
> + },
> + [ C(OP_PREFETCH) ] = {
> + [ C(RESULT_ACCESS) ] = -1,
> + [ C(RESULT_MISS) ] = -1,
> + },
> + },
> + [ C(BPU ) ] = {
> + [ C(OP_READ) ] = {
> + [ C(RESULT_ACCESS) ] = 0x00c4, /* BR_INST_RETIRED.ANY */
> + [ C(RESULT_MISS) ] = 0x00c5, /* BP_INST_RETIRED.MISPRED */
> + },
> + [ C(OP_WRITE) ] = {
> + [ C(RESULT_ACCESS) ] = -1,
> + [ C(RESULT_MISS) ] = -1,
> + },
> + [ C(OP_PREFETCH) ] = {
> + [ C(RESULT_ACCESS) ] = -1,
> + [ C(RESULT_MISS) ] = -1,
> + },
> + },
> +};
> +
> static inline bool intel_pmu_needs_lbr_smpl(struct perf_event *event)
> {
> /* user explicitly requested branch sampling */
> @@ -2180,6 +2323,21 @@ __init int intel_pmu_init(void)
> pr_cont("Atom events, ");
> break;
>
> + case 55: /* Atom 22nm "Silvermont" */
> + memcpy(hw_cache_event_ids, slm_hw_cache_event_ids,
> + sizeof(hw_cache_event_ids));
> + memcpy(hw_cache_extra_regs, slm_hw_cache_extra_regs,
> + sizeof(hw_cache_extra_regs));
> +
> + intel_pmu_lbr_init_atom();
> +
> + x86_pmu.event_constraints = intel_slm_event_constraints;
> + x86_pmu.pebs_constraints = intel_slm_pebs_event_constraints;
> + x86_pmu.extra_regs = intel_slm_extra_regs;
> + x86_pmu.er_flags |= ERF_HAS_RSP_1;
> + pr_cont("Silvermont events, ");
> + break;
> +
> case 37: /* 32 nm nehalem, "Clarkdale" */
> case 44: /* 32 nm nehalem, "Gulftown" */
> case 47: /* 32 nm Xeon E7 */
> diff --git a/arch/x86/kernel/cpu/perf_event_intel_ds.c b/arch/x86/kernel/cpu/perf_event_intel_ds.c
> index 3065c57..442fcc2 100644
> --- a/arch/x86/kernel/cpu/perf_event_intel_ds.c
> +++ b/arch/x86/kernel/cpu/perf_event_intel_ds.c
> @@ -517,6 +517,32 @@ struct event_constraint intel_atom_pebs_event_constraints[] = {
> EVENT_CONSTRAINT_END
> };
>
> +struct event_constraint intel_slm_pebs_event_constraints[] = {
> + INTEL_UEVENT_CONSTRAINT(0x0103, 0x1), /* REHABQ.LD_BLOCK_ST_FORWARD_PS */
> + INTEL_UEVENT_CONSTRAINT(0x0803, 0x1), /* REHABQ.LD_SPLITS_PS */
> + INTEL_UEVENT_CONSTRAINT(0x0204, 0x1), /* MEM_UOPS_RETIRED.L2_HIT_LOADS_PS */
> + INTEL_UEVENT_CONSTRAINT(0x0404, 0x1), /* MEM_UOPS_RETIRED.L2_MISS_LOADS_PS */
> + INTEL_UEVENT_CONSTRAINT(0x0804, 0x1), /* MEM_UOPS_RETIRED.DTLB_MISS_LOADS_PS */
> + INTEL_UEVENT_CONSTRAINT(0x2004, 0x1), /* MEM_UOPS_RETIRED.HITM_PS */
> + INTEL_UEVENT_CONSTRAINT(0x00c0, 0x1), /* INST_RETIRED.ANY_PS */
> + INTEL_UEVENT_CONSTRAINT(0x00c4, 0x1), /* BR_INST_RETIRED.ALL_BRANCHES_PS */
> + INTEL_UEVENT_CONSTRAINT(0x7ec4, 0x1), /* BR_INST_RETIRED.JCC_PS */
> + INTEL_UEVENT_CONSTRAINT(0xbfc4, 0x1), /* BR_INST_RETIRED.FAR_BRANCH_PS */
> + INTEL_UEVENT_CONSTRAINT(0xebc4, 0x1), /* BR_INST_RETIRED.NON_RETURN_IND_PS */
> + INTEL_UEVENT_CONSTRAINT(0xf7c4, 0x1), /* BR_INST_RETIRED.RETURN_PS */
> + INTEL_UEVENT_CONSTRAINT(0xf9c4, 0x1), /* BR_INST_RETIRED.CALL_PS */
> + INTEL_UEVENT_CONSTRAINT(0xfbc4, 0x1), /* BR_INST_RETIRED.IND_CALL_PS */
> + INTEL_UEVENT_CONSTRAINT(0xfdc4, 0x1), /* BR_INST_RETIRED.REL_CALL_PS */
> + INTEL_UEVENT_CONSTRAINT(0xfec4, 0x1), /* BR_INST_RETIRED.TAKEN_JCC_PS */
> + INTEL_UEVENT_CONSTRAINT(0x00c5, 0x1), /* BR_INST_MISP_RETIRED.ALL_BRANCHES_PS */
> + INTEL_UEVENT_CONSTRAINT(0x7ec5, 0x1), /* BR_INST_MISP_RETIRED.JCC_PS */
> + INTEL_UEVENT_CONSTRAINT(0xebc5, 0x1), /* BR_INST_MISP_RETIRED.NON_RETURN_IND_PS */
> + INTEL_UEVENT_CONSTRAINT(0xf7c5, 0x1), /* BR_INST_MISP_RETIRED.RETURN_PS */
> + INTEL_UEVENT_CONSTRAINT(0xfbc5, 0x1), /* BR_INST_MISP_RETIRED.IND_CALL_PS */
> + INTEL_UEVENT_CONSTRAINT(0xfec5, 0x1), /* BR_INST_MISP_RETIRED.TAKEN_JCC_PS */
> + EVENT_CONSTRAINT_END
> +};
> +
Drop those _PS (PEBS) suffixes, they are Intel tool specific and are
being deprecated.
That will make the lines shorter too.

> struct event_constraint intel_nehalem_pebs_event_constraints[] = {
> INTEL_PLD_CONSTRAINT(0x100b, 0xf), /* MEM_INST_RETIRED.* */
> INTEL_EVENT_CONSTRAINT(0x0f, 0xf), /* MEM_UNCORE_RETIRED.* */
> --
> 1.8.1.4
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/