[PATCH 1/4] perf/x86/intel: Add Tremont CPU PMU support

From: kan . liang
Date: Thu Mar 01 2018 - 12:06:43 EST

From: Kan Liang <kan.liang@xxxxxxxxx>

Add perf core PMU support for Intel Tremont CPU cores:
- The init code is based on Goldmont plus.
- There is a new cache event list, based on the Goldmont plus cache
event list.
- PDIR on fixed counter 0
- new topdown events
- OFFCORE_RESPONSE mask updates
- Adaptive PEBS? (no code change)

Signed-off-by: Kan Liang <kan.liang@xxxxxxxxx>
---
arch/x86/events/intel/core.c | 163 ++++++++++++++++++++++++++++++++++++
arch/x86/events/intel/ds.c | 7 ++
arch/x86/events/perf_event.h | 2 +
arch/x86/include/asm/intel-family.h | 1 +
4 files changed, 173 insertions(+)

diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c
index 3cc035e..7e25cf2 100644
--- a/arch/x86/events/intel/core.c
+++ b/arch/x86/events/intel/core.c
@@ -1875,6 +1875,140 @@ static __initconst const u64 glp_hw_cache_extra_regs
},
};

+EVENT_ATTR_STR(topdown-fe-bound, td_fe_bound_tnt, "event=0x71,umask=0x0");
+EVENT_ATTR_STR(topdown-retiring, td_retiring_tnt, "event=0x72,umask=0x0");
+EVENT_ATTR_STR(topdown-bad-spec, td_bad_spec_tnt, "event=0x73,umask=0x0");
+EVENT_ATTR_STR(topdown-be-bound, td_be_bound_tnt, "event=0x74,umask=0x0");
+
+static struct attribute *tnt_events_attrs[] = {
+ EVENT_PTR(td_fe_bound_tnt),
+ EVENT_PTR(td_retiring_tnt),
+ EVENT_PTR(td_bad_spec_tnt),
+ EVENT_PTR(td_be_bound_tnt),
+ NULL
+};
+
+static __initconst const u64 tnt_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)] = 0x81d0, /* MEM_UOPS_RETIRED.ALL_LOADS */
+ [C(RESULT_MISS)] = 0x0,
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = 0x82d0, /* MEM_UOPS_RETIRED.ALL_STORES */
+ [C(RESULT_MISS)] = 0x0,
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = 0x0,
+ [C(RESULT_MISS)] = 0x0,
+ },
+ },
+ [C(L1I)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = 0x0380, /* ICACHE.ACCESSES */
+ [C(RESULT_MISS)] = 0x0280, /* ICACHE.MISSES */
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = -1,
+ [C(RESULT_MISS)] = -1,
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = 0x0,
+ [C(RESULT_MISS)] = 0x0,
+ },
+ },
+ [C(LL)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = 0x1b7, /* OFFCORE_RESPONSE */
+ [C(RESULT_MISS)] = 0x1b7, /* OFFCORE_RESPONSE */
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = 0x1b7, /* OFFCORE_RESPONSE */
+ [C(RESULT_MISS)] = 0x1b7, /* OFFCORE_RESPONSE */
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = 0x0,
+ [C(RESULT_MISS)] = 0x0,
+ },
+ },
+ [C(DTLB)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = 0x81d0, /* MEM_UOPS_RETIRED.ALL_LOADS */
+ [C(RESULT_MISS)] = 0xe08, /* DTLB_LOAD_MISSES.WALK_COMPLETED */
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = 0x82d0, /* MEM_UOPS_RETIRED.ALL_STORES */
+ [C(RESULT_MISS)] = 0xe49, /* DTLB_STORE_MISSES.WALK_COMPLETED */
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = 0x0,
+ [C(RESULT_MISS)] = 0x0,
+ },
+ },
+ [C(ITLB)] = {
+ [C(OP_READ)] = {
+ [C(RESULT_ACCESS)] = 0x0,
+ [C(RESULT_MISS)] = 0x0481, /* ITLB.MISS */
+ },
+ [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.ALL_BRANCHES */
+ [C(RESULT_MISS)] = 0x00c5, /* BR_MISP_RETIRED.ALL_BRANCHES */
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = -1,
+ [C(RESULT_MISS)] = -1,
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = -1,
+ [C(RESULT_MISS)] = -1,
+ },
+ },
+};
+
+static __initconst const u64 tnt_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)] = GLM_DEMAND_READ|
+ GLM_LLC_ACCESS,
+ [C(RESULT_MISS)] = GLM_DEMAND_READ|
+ GLM_LLC_MISS,
+ },
+ [C(OP_WRITE)] = {
+ [C(RESULT_ACCESS)] = GLM_DEMAND_WRITE|
+ GLM_LLC_ACCESS,
+ [C(RESULT_MISS)] = GLM_DEMAND_WRITE|
+ GLM_LLC_MISS,
+ },
+ [C(OP_PREFETCH)] = {
+ [C(RESULT_ACCESS)] = 0x0,
+ [C(RESULT_MISS)] = 0x0,
+ },
+ },
+};
+
+static struct extra_reg intel_tnt_extra_regs[] __read_mostly = {
+ /* must define OFFCORE_RSP_X first, see intel_fixup_er() */
+ INTEL_UEVENT_EXTRA_REG(0x01b7, MSR_OFFCORE_RSP_0, 0xffffff9fffull, RSP_0),
+ INTEL_UEVENT_EXTRA_REG(0x02b7, MSR_OFFCORE_RSP_1, 0xffffff9fffull, RSP_1),
+ EVENT_EXTRA_END
+};
+
#define KNL_OT_L2_HITE BIT_ULL(19) /* Other Tile L2 Hit */
#define KNL_OT_L2_HITF BIT_ULL(20) /* Other Tile L2 Hit */
#define KNL_MCDRAM_LOCAL BIT_ULL(21)
@@ -4292,6 +4426,35 @@ __init int intel_pmu_init(void)
name = "goldmont_plus";
break;

+ /*TODO: test Adaptive PEBS, no extra implementation needed */
+ case INTEL_FAM6_ATOM_JACOBSVILLE:
+ memcpy(hw_cache_event_ids, tnt_hw_cache_event_ids,
+ sizeof(hw_cache_event_ids));
+ memcpy(hw_cache_extra_regs, tnt_hw_cache_extra_regs,
+ sizeof(hw_cache_extra_regs));
+
+ intel_pmu_lbr_init_skl();
+
+ x86_pmu.event_constraints = intel_slm_event_constraints;
+ /*TODO: test PDIR fixed counter */
+ x86_pmu.pebs_constraints = intel_tnt_pebs_event_constraints;
+ /* TODO: OFFCORE_RESPONSE mask updates for 10nm convered IDI */
+ x86_pmu.extra_regs = intel_tnt_extra_regs;
+ /*
+ * It's recommended to use CPU_CLK_UNHALTED.CORE_P + NPEBS
+ * for precise cycles.
+ */
+ x86_pmu.pebs_aliases = NULL;
+ x86_pmu.pebs_prec_dist = true;
+ x86_pmu.lbr_pt_coexist = true;
+ x86_pmu.flags |= PMU_FL_HAS_RSP_1;
+ x86_pmu.get_event_constraints = glp_get_event_constraints;
+ /*TODO: implement topdown metrics, base metrics interface same as SNC */
+ x86_pmu.cpu_events = tnt_events_attrs;
+
+ pr_cont("Tremont events, ");
+ break;
+
case INTEL_FAM6_WESTMERE:
case INTEL_FAM6_WESTMERE_EP:
case INTEL_FAM6_WESTMERE_EX:
diff --git a/arch/x86/events/intel/ds.c b/arch/x86/events/intel/ds.c
index 988b879..3b95e5c6 100644
--- a/arch/x86/events/intel/ds.c
+++ b/arch/x86/events/intel/ds.c
@@ -674,6 +674,13 @@ struct event_constraint intel_glp_pebs_event_constraints[] = {
EVENT_CONSTRAINT_END
};

+struct event_constraint intel_tnt_pebs_event_constraints[] = {
+ INTEL_FLAGS_UEVENT_CONSTRAINT(0x1c0, 0x100000000ULL), /* INST_RETIRED.PREC_DIST */
+ /* Allow all events as PEBS with no flags */
+ INTEL_ALL_EVENT_CONSTRAINT(0, 0x1),
+ EVENT_CONSTRAINT_END
+};
+
struct event_constraint intel_nehalem_pebs_event_constraints[] = {
INTEL_PLD_CONSTRAINT(0x100b, 0xf), /* MEM_INST_RETIRED.* */
INTEL_FLAGS_EVENT_CONSTRAINT(0x0f, 0xf), /* MEM_UNCORE_RETIRED.* */
diff --git a/arch/x86/events/perf_event.h b/arch/x86/events/perf_event.h
index 23e81e1..2713394 100644
--- a/arch/x86/events/perf_event.h
+++ b/arch/x86/events/perf_event.h
@@ -1002,6 +1002,8 @@ extern struct event_constraint intel_glm_pebs_event_constraints[];

extern struct event_constraint intel_glp_pebs_event_constraints[];

+extern struct event_constraint intel_tnt_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/include/asm/intel-family.h b/arch/x86/include/asm/intel-family.h
index b4d37ffa..aa4c026 100644
--- a/arch/x86/include/asm/intel-family.h
+++ b/arch/x86/include/asm/intel-family.h
@@ -63,6 +63,7 @@
#define INTEL_FAM6_ATOM_GOLDMONT 0x5C
#define INTEL_FAM6_ATOM_DENVERTON 0x5F /* Goldmont Microserver */
#define INTEL_FAM6_ATOM_GEMINI_LAKE 0x7A
+#define INTEL_FAM6_ATOM_JACOBSVILLE 0x86

/* Xeon Phi */

--
2.4.3