[PATCH 1/3] perf, x86: Add new cache events table for Haswell

[PATCH 1/3] perf, x86: Add new cache events table for Haswell

[Andi Kleen]

From: Andi Kleen <ak@linux.intel.com>

Haswell offcore events are quite different from Sandy Bridge.
Add a new table to handle Haswell properly.

Note that the offcore bits listed in the SDM are not quite correct
(this is currently being fixed). An uptodate list of bits is
in the patch.

The basic setup is similar to Sandy Bridge. The prefetch columns
have been removed, as prefetch counting is not very reliable
on Haswell. One L1 event that is not in the event list anymore
has been also removed.

- data reads do not include code reads (comparable to earlier Sandy
Bridge tables)
- data counts include speculative execution (except L1 write, dtlb, bpu)
- remote node access includes both remote memory, remote cache, remote mmio.
- prefetches are not included in the counts for consistency
(different from Sandy Bridge, which includes prefetches in the remote node)

The events with additional caveats have references to the specification update.

v2: Change name of variable.
v3: Based on older Broadwell patch, but now just for Haswell.
    Various fixes with events being more similar to the older
    Sandy Bridge tables.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
 arch/x86/kernel/cpu/perf_event_intel.c | 197 ++++++++++++++++++++++++++++++++-
 1 file changed, 195 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 498b6d9..387f580 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -415,6 +415,199 @@ static __initconst const u64 snb_hw_cache_event_ids
 
 };
 
+/*
+ * Notes on the events:
+ * - data reads do not include code reads (comparable to earlier tables)
+ * - data counts include speculative execution (except L1 write, dtlb, bpu)
+ * - remote node access includes both remote memory, remote cache, remote mmio.
+ * - prefetches are not included in the counts.
+ * The events with additional caveats have references to the specification update.
+ */
+
+#define HSW_DEMAND_DATA_RD		BIT(0)
+#define HSW_DEMAND_RFO			BIT(1)
+#define HSW_PF_L2_RFO			BIT(5)
+#define HSW_PF_L3_RFO			BIT(8)
+#define HSW_ALL_RFO			(HSW_DEMAND_RFO| \
+					 HSW_PF_L2_RFO|HSW_PF_L3_RFO)
+#define HSW_ANY_RESPONSE		BIT(16)
+#define HSW_SUPPLIER_NONE		BIT(17)
+#define HSW_L3_MISS_LOCAL		BIT(22)
+#define HSW_L3_MISS_REMOTE_HOP0		BIT(27)
+#define HSW_L3_MISS_REMOTE_HOP1		BIT(28)
+#define HSW_L3_MISS_REMOTE_HOP2P	BIT(29)
+#define HSW_L3_MISS			(HSW_L3_MISS_LOCAL| \
+					 HSW_L3_MISS_REMOTE_HOP0|HSW_L3_MISS_REMOTE_HOP1| \
+					 HSW_L3_MISS_REMOTE_HOP2P)
+#define HSW_SNOOP_NONE			BIT(31)
+#define HSW_SNOOP_NOT_NEEDED		BIT(32)
+#define HSW_SNOOP_MISS			BIT(33)
+#define HSW_SNOOP_HIT_NO_FWD		BIT(34)
+#define HSW_SNOOP_HIT_WITH_FWD		BIT(35)
+#define HSW_SNOOP_HITM			BIT(36)
+#define HSW_SNOOP_NON_DRAM		BIT(37)
+#define HSW_ANY_SNOOP			(HSW_SNOOP_NONE| \
+					 HSW_SNOOP_NOT_NEEDED|HSW_SNOOP_MISS| \
+					 HSW_SNOOP_HIT_NO_FWD|HSW_SNOOP_HIT_WITH_FWD| \
+					 HSW_SNOOP_HITM|HSW_SNOOP_NON_DRAM)
+
+static __initconst const u64 hsw_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, HSM30 */
+		[ C(RESULT_MISS)   ] = 0x151, 	/* L1D.REPLACEMENT */
+	},
+	[ C(OP_WRITE) ] = {
+		[ C(RESULT_ACCESS) ] = 0x82d0, 	/* MEM_UOPS_RETIRED.ALL_STORES, HSM30 */
+		[ C(RESULT_MISS)   ] = 0x0,
+	},
+	[ C(OP_PREFETCH) ] = {
+		[ C(RESULT_ACCESS) ] = 0x0,
+		[ C(RESULT_MISS)   ] = 0x0,
+	},
+ },
+ [ C(L1I ) ] = {
+	[ C(OP_READ) ] = {
+		[ C(RESULT_ACCESS) ] = 0x0,
+		[ C(RESULT_MISS)   ] = 0x280, 	/* 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, HSM30 */
+		[ C(RESULT_MISS)   ] = 0x108, 	/* DTLB_LOAD_MISSES.MISS_CAUSES_A_WALK */
+	},
+	[ C(OP_WRITE) ] = {
+		[ C(RESULT_ACCESS) ] = 0x82d0, 	/* MEM_UOPS_RETIRED.ALL_STORES, HSM30 */
+		[ C(RESULT_MISS)   ] = 0x149, 	/* DTLB_STORE_MISSES.MISS_CAUSES_A_WALK */
+	},
+	[ C(OP_PREFETCH) ] = {
+		[ C(RESULT_ACCESS) ] = 0x0,
+		[ C(RESULT_MISS)   ] = 0x0,
+	},
+ },
+ [ C(ITLB) ] = {
+	[ C(OP_READ) ] = {
+		[ C(RESULT_ACCESS) ] = 0x6085, 	/* ITLB_MISSES.STLB_HIT */
+		[ C(RESULT_MISS)   ] = 0x185, 	/* ITLB_MISSES.MISS_CAUSES_A_WALK */
+	},
+	[ 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) ] = 0xc4, 	/* BR_INST_RETIRED.ALL_BRANCHES */
+		[ C(RESULT_MISS)   ] = 0xc5, 	/* 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,
+	},
+ },
+ [ C(NODE) ] = {
+	[ 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,
+	},
+ },
+};
+
+static __initconst const u64 hsw_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) ] = HSW_DEMAND_DATA_RD|
+				       HSW_ANY_RESPONSE|HSW_ANY_SNOOP|
+				       HSW_SUPPLIER_NONE,
+		[ C(RESULT_MISS)   ] = HSW_DEMAND_DATA_RD|
+				       HSW_L3_MISS|HSW_ANY_SNOOP|
+				       HSW_SUPPLIER_NONE,
+	},
+	[ C(OP_WRITE) ] = {
+		[ C(RESULT_ACCESS) ] = HSW_DEMAND_RFO|
+				       HSW_ANY_RESPONSE|HSW_ANY_SNOOP|
+				       HSW_SUPPLIER_NONE,
+		[ C(RESULT_MISS)   ] = HSW_DEMAND_RFO|HSW_L3_MISS|
+				       HSW_ANY_SNOOP|HSW_SUPPLIER_NONE,
+	},
+	[ C(OP_PREFETCH) ] = {
+		[ C(RESULT_ACCESS) ] = 0x0,
+		[ C(RESULT_MISS)   ] = 0x0,
+	},
+ },
+ [ C(NODE) ] = {
+	[ C(OP_READ) ] = {
+		[ C(RESULT_ACCESS) ] = HSW_DEMAND_DATA_RD|
+				       HSW_L3_MISS_LOCAL|HSW_SUPPLIER_NONE|
+				       HSW_ANY_SNOOP,
+		[ C(RESULT_MISS)   ] = HSW_DEMAND_DATA_RD|
+				       HSW_L3_MISS_REMOTE_HOP0|HSW_L3_MISS_REMOTE_HOP1|
+				       HSW_L3_MISS_REMOTE_HOP2P|HSW_SUPPLIER_NONE|
+				       HSW_ANY_SNOOP,
+	},
+	[ C(OP_WRITE) ] = {
+		[ C(RESULT_ACCESS) ] = HSW_DEMAND_RFO|
+				       HSW_L3_MISS_LOCAL|HSW_SUPPLIER_NONE|
+				       HSW_ANY_SNOOP,
+		[ C(RESULT_MISS)   ] = HSW_DEMAND_RFO|
+				       HSW_L3_MISS_REMOTE_HOP0|HSW_L3_MISS_REMOTE_HOP1|
+				       HSW_L3_MISS_REMOTE_HOP2P|HSW_SUPPLIER_NONE|
+				       HSW_ANY_SNOOP,
+	},
+	[ C(OP_PREFETCH) ] = {
+		[ C(RESULT_ACCESS) ] = 0x0,
+		[ C(RESULT_MISS)   ] = 0x0,
+	},
+ },
+};
+
 static __initconst const u64 westmere_hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
@@ -2546,8 +2739,8 @@ __init int intel_pmu_init(void)
 	case 69: /* 22nm Haswell ULT */
 	case 70: /* 22nm Haswell + GT3e (Intel Iris Pro graphics) */
 		x86_pmu.late_ack = true;
-		memcpy(hw_cache_event_ids, snb_hw_cache_event_ids, sizeof(hw_cache_event_ids));
-		memcpy(hw_cache_extra_regs, snb_hw_cache_extra_regs, sizeof(hw_cache_extra_regs));
+		memcpy(hw_cache_event_ids, hsw_hw_cache_event_ids, sizeof(hw_cache_event_ids));
+		memcpy(hw_cache_extra_regs, hsw_hw_cache_extra_regs, sizeof(hw_cache_extra_regs));
 
 		intel_pmu_lbr_init_snb();
 
-- 
1.9.3

[PATCH 2/3] perf, x86: Add Broadwell core support

[Andi Kleen]

From: Andi Kleen <ak@linux.intel.com>

Add Broadwell support for Broadwell to perf.

The basic support is very similar to Haswell. We use the new cache
event list added for Haswell earlier. The only differences
are a few bits related to remote nodes. To avoid an extra,
mostly identical, table these are patched up in the initialization code.

The constraint list has one new event that needs to be handled over Haswell.

Includes code and testing from Kan Liang.

v2: Remove unnamed model numbers.
v3: Rename cache event list to hsw_*. Change names.
v4: Use symbolic names for cache events. Improve comments and description.
    Fix sparse warnings (Fengguang Wu)
    Add Xeon D model number.
    Remove cache event table (in separate patch)
    Patch up remote node differences (Kan Liang)
Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
 arch/x86/kernel/cpu/perf_event_intel.c | 48 ++++++++++++++++++++++++++++++++++
 1 file changed, 48 insertions(+)

diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 387f580..7c480e8 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -220,6 +220,15 @@ static struct event_constraint intel_hsw_event_constraints[] = {
 	EVENT_CONSTRAINT_END
 };
 
+struct event_constraint intel_bdw_event_constraints[] = {
+	FIXED_EVENT_CONSTRAINT(0x00c0, 0),	/* INST_RETIRED.ANY */
+	FIXED_EVENT_CONSTRAINT(0x003c, 1),	/* CPU_CLK_UNHALTED.CORE */
+	FIXED_EVENT_CONSTRAINT(0x0300, 2),	/* CPU_CLK_UNHALTED.REF */
+	INTEL_UEVENT_CONSTRAINT(0x148, 0x4),	/* L1D_PEND_MISS.PENDING */
+	INTEL_EVENT_CONSTRAINT(0xa3, 0x4),	/* CYCLE_ACTIVITY.* */
+	EVENT_CONSTRAINT_END
+};
+
 static u64 intel_pmu_event_map(int hw_event)
 {
 	return intel_perfmon_event_map[hw_event];
@@ -433,12 +442,16 @@ static __initconst const u64 snb_hw_cache_event_ids
 #define HSW_ANY_RESPONSE		BIT(16)
 #define HSW_SUPPLIER_NONE		BIT(17)
 #define HSW_L3_MISS_LOCAL		BIT(22)
+#define BDW_L3_MISS_LOCAL		BIT(26)
 #define HSW_L3_MISS_REMOTE_HOP0		BIT(27)
 #define HSW_L3_MISS_REMOTE_HOP1		BIT(28)
 #define HSW_L3_MISS_REMOTE_HOP2P	BIT(29)
 #define HSW_L3_MISS			(HSW_L3_MISS_LOCAL| \
 					 HSW_L3_MISS_REMOTE_HOP0|HSW_L3_MISS_REMOTE_HOP1| \
 					 HSW_L3_MISS_REMOTE_HOP2P)
+#define BDW_L3_MISS			(BDW_L3_MISS_LOCAL| \
+					 HSW_L3_MISS_REMOTE_HOP0|HSW_L3_MISS_REMOTE_HOP1| \
+					 HSW_L3_MISS_REMOTE_HOP2P)
 #define HSW_SNOOP_NONE			BIT(31)
 #define HSW_SNOOP_NOT_NEEDED		BIT(32)
 #define HSW_SNOOP_MISS			BIT(33)
@@ -2759,6 +2772,41 @@ __init int intel_pmu_init(void)
 		pr_cont("Haswell events, ");
 		break;
 
+	case 61: /* 14nm Broadwell Core-M */
+	case 86: /* 14nm Broadwell Xeon D */
+		x86_pmu.late_ack = true;
+		memcpy(hw_cache_event_ids, hsw_hw_cache_event_ids, sizeof(hw_cache_event_ids));
+		memcpy(hw_cache_extra_regs, hsw_hw_cache_extra_regs, sizeof(hw_cache_extra_regs));
+
+		/* L3_MISS_LOCAL_DRAM is BIT(26) in Broadwell */
+		hw_cache_extra_regs[C(LL)][C(OP_READ)][C(RESULT_MISS)] = HSW_DEMAND_DATA_RD |
+									 BDW_L3_MISS|HSW_ANY_SNOOP|
+									 HSW_SUPPLIER_NONE;
+		hw_cache_extra_regs[C(LL)][C(OP_WRITE)][C(RESULT_MISS)] = HSW_ALL_RFO|BDW_L3_MISS|
+									  HSW_ANY_SNOOP|HSW_SUPPLIER_NONE;
+		hw_cache_extra_regs[C(NODE)][C(OP_READ)][C(RESULT_ACCESS)] = HSW_DEMAND_DATA_RD|
+									     BDW_L3_MISS_LOCAL|HSW_SUPPLIER_NONE|
+									     HSW_ANY_SNOOP;
+		hw_cache_extra_regs[C(NODE)][C(OP_WRITE)][C(RESULT_ACCESS)] = HSW_DEMAND_RFO|
+									      BDW_L3_MISS_LOCAL|HSW_SUPPLIER_NONE|
+									      HSW_ANY_SNOOP;
+
+		intel_pmu_lbr_init_snb();
+
+		x86_pmu.event_constraints = intel_bdw_event_constraints;
+		x86_pmu.pebs_constraints = intel_hsw_pebs_event_constraints;
+		x86_pmu.extra_regs = intel_snbep_extra_regs;
+		x86_pmu.pebs_aliases = intel_pebs_aliases_snb;
+		/* all extra regs are per-cpu when HT is on */
+		x86_pmu.er_flags |= ERF_HAS_RSP_1;
+		x86_pmu.er_flags |= ERF_NO_HT_SHARING;
+
+		x86_pmu.hw_config = hsw_hw_config;
+		x86_pmu.get_event_constraints = hsw_get_event_constraints;
+		x86_pmu.cpu_events = hsw_events_attrs;
+		pr_cont("Broadwell events, ");
+		break;
+
 	default:
 		switch (x86_pmu.version) {
 		case 1:
-- 
1.9.3

[PATCH 3/3] perf, x86: Add INST_RETIRED.ALL workarounds

[Andi Kleen]

From: Andi Kleen <ak@linux.intel.com>

On Broadwell INST_RETIRED.ALL cannot be used with any period
that doesn't have the lowest 6 bits cleared. And the period
should not be smaller than 128.

Add a new callback to enforce this, and set it for Broadwell.

This is erratum BDM57 and BDM11.

How does this handle the case when an app requests a specific
period with some of the bottom bits set

The apps thinks it is sampling at X occurences per sample, when it is
in fact at X - 63 (worst case).

Short answer:

Any useful instruction sampling period needs to be 4-6 orders
of magnitude larger than 128, as an PMI every 128 instructions
would instantly overwhelm the system and be throttled.
So the +-64 error from this is really small compared to the
period, much smaller than normal system jitter.

Long answer:

<write up by Peter:>

IFF we guarantee perf_event_attr::sample_period >= 128.

Suppose we start out with sample_period=192; then we'll set period_left
to 192, we'll end up with left = 128 (we truncate the lower bits). We
get an interrupt, find that period_left = 64 (>0 so we return 0 and
don't get an overflow handler), up that to 128. Then we trigger again,
at n=256. Then we find period_left = -64 (<=0 so we return 1 and do get
an overflow). We increment with sample_period so we get left = 128. We
fire again, at n=384, period_left = 0 (<=0 so we return 1 and get an
overflow). And on and on.

So while the individual interrupts are 'wrong' we get then with
interval=256,128 in exactly the right ratio to average out at 192. And
this works for everything >=128.

So the num_samples*fixed_period thing is still entirely correct +- 127,
which is good enough I'd say, as you already have that error anyhow.

So no need to 'fix' the tools, al we need to do is refuse to create
INST_RETIRED:ALL events with sample_period < 128.

v2: Use correct event name in description. Use EVENT() macro.
v3: Use INTEL_ARCH_EVENT_MASK. Error out for events with too small period.
v4: Expand description.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
 arch/x86/kernel/cpu/perf_event.c       |  9 +++++++++
 arch/x86/kernel/cpu/perf_event.h       |  1 +
 arch/x86/kernel/cpu/perf_event_intel.c | 19 +++++++++++++++++++
 3 files changed, 29 insertions(+)

diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index 143e5f5..66451a6 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -445,6 +445,12 @@ int x86_pmu_hw_config(struct perf_event *event)
 	if (event->attr.type == PERF_TYPE_RAW)
 		event->hw.config |= event->attr.config & X86_RAW_EVENT_MASK;
 
+	if (event->attr.sample_period && x86_pmu.limit_period) {
+		if (x86_pmu.limit_period(event, event->attr.sample_period) >
+				event->attr.sample_period)
+			return -EINVAL;
+	}
+
 	return x86_setup_perfctr(event);
 }
 
@@ -982,6 +988,9 @@ int x86_perf_event_set_period(struct perf_event *event)
 	if (left > x86_pmu.max_period)
 		left = x86_pmu.max_period;
 
+	if (x86_pmu.limit_period)
+		left = x86_pmu.limit_period(event, left);
+
 	per_cpu(pmc_prev_left[idx], smp_processor_id()) = left;
 
 	/*
diff --git a/arch/x86/kernel/cpu/perf_event.h b/arch/x86/kernel/cpu/perf_event.h
index 4e6cdb0..6ce77bd 100644
--- a/arch/x86/kernel/cpu/perf_event.h
+++ b/arch/x86/kernel/cpu/perf_event.h
@@ -449,6 +449,7 @@ struct x86_pmu {
 	struct x86_pmu_quirk *quirks;
 	int		perfctr_second_write;
 	bool		late_ack;
+	unsigned	(*limit_period)(struct perf_event *event, unsigned l);
 
 	/*
 	 * sysfs attrs
diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 7c480e8..96c0898 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -2111,6 +2111,24 @@ hsw_get_event_constraints(struct cpu_hw_events *cpuc, struct perf_event *event)
 	return c;
 }
 
+/*
+ * Broadwell:
+ * The INST_RETIRED.ALL period always needs to have lowest
+ * 6bits cleared (BDM57). It shall not use a period smaller
+ * than 100 (BDM11). We combine the two to enforce
+ * a min-period of 128.
+ */
+static unsigned bdw_limit_period(struct perf_event *event, unsigned left)
+{
+	if ((event->hw.config & INTEL_ARCH_EVENT_MASK) ==
+			X86_CONFIG(.event=0xc0, .umask=0x01)) {
+		if (left < 128)
+			left = 128;
+		left &= ~0x3fu;
+	}
+	return left;
+}
+
 PMU_FORMAT_ATTR(event,	"config:0-7"	);
 PMU_FORMAT_ATTR(umask,	"config:8-15"	);
 PMU_FORMAT_ATTR(edge,	"config:18"	);
@@ -2804,6 +2822,7 @@ __init int intel_pmu_init(void)
 		x86_pmu.hw_config = hsw_hw_config;
 		x86_pmu.get_event_constraints = hsw_get_event_constraints;
 		x86_pmu.cpu_events = hsw_events_attrs;
+		x86_pmu.limit_period = bdw_limit_period;
 		pr_cont("Broadwell events, ");
 		break;
 
-- 
1.9.3

Re: [PATCH 1/3] perf, x86: Add new cache events table for Haswell

[Peter Zijlstra]

On Mon, Feb 09, 2015 at 11:17:44AM -0800, Andi Kleen wrote:
> +++ b/arch/x86/kernel/cpu/perf_event_intel.c
> @@ -415,6 +415,199 @@ static __initconst const u64 snb_hw_cache_event_ids
>  
>  };
>  
> +/*
> + * Notes on the events:
> + * - data reads do not include code reads (comparable to earlier tables)
> + * - data counts include speculative execution (except L1 write, dtlb, bpu)
> + * - remote node access includes both remote memory, remote cache, remote mmio.
> + * - prefetches are not included in the counts.
> + * The events with additional caveats have references to the specification update.
> + */

It would be good to have the prefetch note from the changelog in a
comment here or at the extra_regs table; i.e. prefetches are not
included because they're 'broken'.

> +
> +#define HSW_DEMAND_DATA_RD		BIT(0)
> +#define HSW_DEMAND_RFO			BIT(1)
> +#define HSW_PF_L2_RFO			BIT(5)
> +#define HSW_PF_L3_RFO			BIT(8)
> +#define HSW_ALL_RFO			(HSW_DEMAND_RFO| \
> +					 HSW_PF_L2_RFO|HSW_PF_L3_RFO)
> +#define HSW_ANY_RESPONSE		BIT(16)
> +#define HSW_SUPPLIER_NONE		BIT(17)
> +#define HSW_L3_MISS_LOCAL		BIT(22)
> +#define HSW_L3_MISS_REMOTE_HOP0		BIT(27)
> +#define HSW_L3_MISS_REMOTE_HOP1		BIT(28)
> +#define HSW_L3_MISS_REMOTE_HOP2P	BIT(29)
> +#define HSW_L3_MISS			(HSW_L3_MISS_LOCAL| \
> +					 HSW_L3_MISS_REMOTE_HOP0|HSW_L3_MISS_REMOTE_HOP1| \
> +					 HSW_L3_MISS_REMOTE_HOP2P)
> +#define HSW_SNOOP_NONE			BIT(31)
> +#define HSW_SNOOP_NOT_NEEDED		BIT(32)
> +#define HSW_SNOOP_MISS			BIT(33)
> +#define HSW_SNOOP_HIT_NO_FWD		BIT(34)
> +#define HSW_SNOOP_HIT_WITH_FWD		BIT(35)
> +#define HSW_SNOOP_HITM			BIT(36)
> +#define HSW_SNOOP_NON_DRAM		BIT(37)
> +#define HSW_ANY_SNOOP			(HSW_SNOOP_NONE| \
> +					 HSW_SNOOP_NOT_NEEDED|HSW_SNOOP_MISS| \
> +					 HSW_SNOOP_HIT_NO_FWD|HSW_SNOOP_HIT_WITH_FWD| \
> +					 HSW_SNOOP_HITM|HSW_SNOOP_NON_DRAM)

Just in case someone is crazy enough to run a 32bit kernel on HSW, this
needs to be BIT_ULL() -- also for compile testing, gcc tends to complain
about things like (1UL << 32) for ILP32 targets.

> +static __initconst const u64 hsw_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) ] = HSW_DEMAND_DATA_RD|
> +				       HSW_ANY_RESPONSE|HSW_ANY_SNOOP|
> +				       HSW_SUPPLIER_NONE,
> +		[ C(RESULT_MISS)   ] = HSW_DEMAND_DATA_RD|
> +				       HSW_L3_MISS|HSW_ANY_SNOOP|
> +				       HSW_SUPPLIER_NONE,
> +	},
> +	[ C(OP_WRITE) ] = {
> +		[ C(RESULT_ACCESS) ] = HSW_DEMAND_RFO|
> +				       HSW_ANY_RESPONSE|HSW_ANY_SNOOP|
> +				       HSW_SUPPLIER_NONE,
> +		[ C(RESULT_MISS)   ] = HSW_DEMAND_RFO|HSW_L3_MISS|
> +				       HSW_ANY_SNOOP|HSW_SUPPLIER_NONE,
> +	},
> +	[ C(OP_PREFETCH) ] = {
> +		[ C(RESULT_ACCESS) ] = 0x0,
> +		[ C(RESULT_MISS)   ] = 0x0,
> +	},
> + },
> + [ C(NODE) ] = {
> +	[ C(OP_READ) ] = {
> +		[ C(RESULT_ACCESS) ] = HSW_DEMAND_DATA_RD|
> +				       HSW_L3_MISS_LOCAL|HSW_SUPPLIER_NONE|
> +				       HSW_ANY_SNOOP,
> +		[ C(RESULT_MISS)   ] = HSW_DEMAND_DATA_RD|
> +				       HSW_L3_MISS_REMOTE_HOP0|HSW_L3_MISS_REMOTE_HOP1|
> +				       HSW_L3_MISS_REMOTE_HOP2P|HSW_SUPPLIER_NONE|
> +				       HSW_ANY_SNOOP,
> +	},
> +	[ C(OP_WRITE) ] = {
> +		[ C(RESULT_ACCESS) ] = HSW_DEMAND_RFO|
> +				       HSW_L3_MISS_LOCAL|HSW_SUPPLIER_NONE|
> +				       HSW_ANY_SNOOP,
> +		[ C(RESULT_MISS)   ] = HSW_DEMAND_RFO|
> +				       HSW_L3_MISS_REMOTE_HOP0|HSW_L3_MISS_REMOTE_HOP1|
> +				       HSW_L3_MISS_REMOTE_HOP2P|HSW_SUPPLIER_NONE|
> +				       HSW_ANY_SNOOP,
> +	},
> +	[ C(OP_PREFETCH) ] = {
> +		[ C(RESULT_ACCESS) ] = 0x0,
> +		[ C(RESULT_MISS)   ] = 0x0,
> +	},
> + },
> +};

Now the other tables create little helpers like:

#define HSW_DMND_READ	(HSW_DMND_DATA_RD)
#define HSW_DMND_WRITE	(HSW_DMND_RFO)

#define HSW_L3_ACCESS	(HSW_ANY_RESPONSE)
#define HSW_L3_MISS	(HSW_L3_MISS)

And compose the tables values using those:

	HSW_DMND_READ|HSW_L3_ACCESS

Please do so here too.

Now; when comparing these value to the SNB for example I note that you
include ANY_SNOOP and SUPPLIER_NONE in L3_ACCESS, SNB and other do not,
please explain.

Re: [PATCH 1/3] perf, x86: Add new cache events table for Haswell

[Andi Kleen]

> Just in case someone is crazy enough to run a 32bit kernel on HSW, this
> needs to be BIT_ULL() -- also for compile testing, gcc tends to complain
> about things like (1UL << 32) for ILP32 targets.

Ok.

> Now the other tables create little helpers like:
> 
> #define HSW_DMND_READ	(HSW_DMND_DATA_RD)
> #define HSW_DMND_WRITE	(HSW_DMND_RFO)
> 
> #define HSW_L3_ACCESS	(HSW_ANY_RESPONSE)
> #define HSW_L3_MISS	(HSW_L3_MISS)
> 
> And compose the tables values using those:
> 
> 	HSW_DMND_READ|HSW_L3_ACCESS
> 
> Please do so here too.

I'm trying to stay with the official documented bit names. No such bit names exist.
If we make up our own names nobody else can read it anymore.

> 
> Now; when comparing these value to the SNB for example I note that you
> include ANY_SNOOP and SUPPLIER_NONE in L3_ACCESS, SNB and other do not,
> please explain.

You're supposed to set a snoop and supplier qualifier.
AFAIK SNB should set them too. It may work without them due to some
quirk.

-Andi

-- 
ak@linux.intel.com -- Speaking for myself only

Re: [PATCH 1/3] perf, x86: Add new cache events table for Haswell

[Peter Zijlstra]

On Tue, Feb 10, 2015 at 09:11:35AM -0800, Andi Kleen wrote:

> > Now the other tables create little helpers like:
> > 
> > #define HSW_DMND_READ	(HSW_DMND_DATA_RD)
> > #define HSW_DMND_WRITE	(HSW_DMND_RFO)
> > 
> > #define HSW_L3_ACCESS	(HSW_ANY_RESPONSE)
> > #define HSW_L3_MISS	(HSW_L3_MISS)
> > 
> > And compose the tables values using those:
> > 
> > 	HSW_DMND_READ|HSW_L3_ACCESS
> > 
> > Please do so here too.
> 
> I'm trying to stay with the official documented bit names. No such bit names exist.
> If we make up our own names nobody else can read it anymore.

Well, its a simple matter of looking up the bit definitions; we define
those helpers in terms of the official names after all.

By having the 4 helpers {r,w}x{access,miss} you avoid some repetition
and decrease the room for mistakes.

> > Now; when comparing these value to the SNB for example I note that you
> > include ANY_SNOOP and SUPPLIER_NONE in L3_ACCESS, SNB and other do not,
> > please explain.
> 
> You're supposed to set a snoop and supplier qualifier.
> AFAIK SNB should set them too. It may work without them due to some
> quirk.

OK, then we should fix the others. Thanks.

[PATCH 2/3] perf, x86: Add Broadwell core support

[Andi Kleen]

From: Andi Kleen <ak@linux.intel.com>

Add Broadwell support for Broadwell to perf.

The basic support is very similar to Haswell. We use the new cache
event list added for Haswell earlier. The only differences
are a few bits related to remote nodes. To avoid an extra,
mostly identical, table these are patched up in the initialization code.

The constraint list has one new event that needs to be handled over Haswell.

Includes code and testing from Kan Liang.

v2: Remove unnamed model numbers.
v3: Rename cache event list to hsw_*. Change names.
v4: Use symbolic names for cache events. Improve comments and description.
    Fix sparse warnings (Fengguang Wu)
    Add Xeon D model number.
    Remove cache event table (in separate patch)
    Patch up remote node differences (Kan Liang)
v5: Update offcore bits
Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
 arch/x86/kernel/cpu/perf_event_intel.c | 47 ++++++++++++++++++++++++++++++++++
 1 file changed, 47 insertions(+)

diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 5cd7e2b..259e06b 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -220,6 +220,15 @@ static struct event_constraint intel_hsw_event_constraints[] = {
 	EVENT_CONSTRAINT_END
 };
 
+struct event_constraint intel_bdw_event_constraints[] = {
+	FIXED_EVENT_CONSTRAINT(0x00c0, 0),	/* INST_RETIRED.ANY */
+	FIXED_EVENT_CONSTRAINT(0x003c, 1),	/* CPU_CLK_UNHALTED.CORE */
+	FIXED_EVENT_CONSTRAINT(0x0300, 2),	/* CPU_CLK_UNHALTED.REF */
+	INTEL_UEVENT_CONSTRAINT(0x148, 0x4),	/* L1D_PEND_MISS.PENDING */
+	INTEL_EVENT_CONSTRAINT(0xa3, 0x4),	/* CYCLE_ACTIVITY.* */
+	EVENT_CONSTRAINT_END
+};
+
 static u64 intel_pmu_event_map(int hw_event)
 {
 	return intel_perfmon_event_map[hw_event];
@@ -454,6 +463,12 @@ static __initconst const u64 snb_hw_cache_event_ids
 					 HSW_L3_MISS_REMOTE_HOP1|HSW_L3_MISS_REMOTE_HOP2P)
 #define HSW_LLC_ACCESS			HSW_ANY_RESPONSE
 
+#define BDW_L3_MISS_LOCAL		BIT(26)
+#define BDW_L3_MISS			(BDW_L3_MISS_LOCAL| \
+					 HSW_L3_MISS_REMOTE_HOP0|HSW_L3_MISS_REMOTE_HOP1| \
+					 HSW_L3_MISS_REMOTE_HOP2P)
+
+
 static __initconst const u64 hsw_hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
@@ -2757,6 +2772,38 @@ __init int intel_pmu_init(void)
 		pr_cont("Haswell events, ");
 		break;
 
+	case 61: /* 14nm Broadwell Core-M */
+	case 86: /* 14nm Broadwell Xeon D */
+		x86_pmu.late_ack = true;
+		memcpy(hw_cache_event_ids, hsw_hw_cache_event_ids, sizeof(hw_cache_event_ids));
+		memcpy(hw_cache_extra_regs, hsw_hw_cache_extra_regs, sizeof(hw_cache_extra_regs));
+
+		/* L3_MISS_LOCAL_DRAM is BIT(26) in Broadwell */
+		hw_cache_extra_regs[C(LL)][C(OP_READ)][C(RESULT_MISS)] = HSW_DEMAND_READ |
+									 BDW_L3_MISS|HSW_SNOOP_DRAM;
+		hw_cache_extra_regs[C(LL)][C(OP_WRITE)][C(RESULT_MISS)] = HSW_DEMAND_WRITE|BDW_L3_MISS|
+									  HSW_SNOOP_DRAM;
+		hw_cache_extra_regs[C(NODE)][C(OP_READ)][C(RESULT_ACCESS)] = HSW_DEMAND_READ|
+									     BDW_L3_MISS_LOCAL|HSW_SNOOP_DRAM;
+		hw_cache_extra_regs[C(NODE)][C(OP_WRITE)][C(RESULT_ACCESS)] = HSW_DEMAND_WRITE|
+									      BDW_L3_MISS_LOCAL|HSW_SNOOP_DRAM;
+
+		intel_pmu_lbr_init_snb();
+
+		x86_pmu.event_constraints = intel_bdw_event_constraints;
+		x86_pmu.pebs_constraints = intel_hsw_pebs_event_constraints;
+		x86_pmu.extra_regs = intel_snbep_extra_regs;
+		x86_pmu.pebs_aliases = intel_pebs_aliases_snb;
+		/* all extra regs are per-cpu when HT is on */
+		x86_pmu.er_flags |= ERF_HAS_RSP_1;
+		x86_pmu.er_flags |= ERF_NO_HT_SHARING;
+
+		x86_pmu.hw_config = hsw_hw_config;
+		x86_pmu.get_event_constraints = hsw_get_event_constraints;
+		x86_pmu.cpu_events = hsw_events_attrs;
+		pr_cont("Broadwell events, ");
+		break;
+
 	default:
 		switch (x86_pmu.version) {
 		case 1:
-- 
1.9.3

[PATCH 2/3] perf, x86: Add Broadwell core support

[Andi Kleen]

From: Andi Kleen <ak@linux.intel.com>

Add Broadwell support for Broadwell to perf.

The basic support is very similar to Haswell. We use the new cache
event list added for Haswell earlier. The only differences
are a few bits related to remote nodes. To avoid an extra,
mostly identical, table these are patched up in the initialization code.

The constraint list has one new event that needs to be handled over Haswell.

Includes code and testing from Kan Liang.

v2: Remove unnamed model numbers.
v3: Rename cache event list to hsw_*. Change names.
v4: Use symbolic names for cache events. Improve comments and description.
    Fix sparse warnings (Fengguang Wu)
    Add Xeon D model number.
    Remove cache event table (in separate patch)
    Patch up remote node differences (Kan Liang)
Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
 arch/x86/kernel/cpu/perf_event_intel.c | 50 ++++++++++++++++++++++++++++++++++
 1 file changed, 50 insertions(+)

diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 02ab31d..035f8d1 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -220,6 +220,15 @@ static struct event_constraint intel_hsw_event_constraints[] = {
 	EVENT_CONSTRAINT_END
 };
 
+struct event_constraint intel_bdw_event_constraints[] = {
+	FIXED_EVENT_CONSTRAINT(0x00c0, 0),	/* INST_RETIRED.ANY */
+	FIXED_EVENT_CONSTRAINT(0x003c, 1),	/* CPU_CLK_UNHALTED.CORE */
+	FIXED_EVENT_CONSTRAINT(0x0300, 2),	/* CPU_CLK_UNHALTED.REF */
+	INTEL_UEVENT_CONSTRAINT(0x148, 0x4),	/* L1D_PEND_MISS.PENDING */
+	INTEL_EVENT_CONSTRAINT(0xa3, 0x4),	/* CYCLE_ACTIVITY.* */
+	EVENT_CONSTRAINT_END
+};
+
 static u64 intel_pmu_event_map(int hw_event)
 {
 	return intel_perfmon_event_map[hw_event];
@@ -452,6 +461,12 @@ static __initconst const u64 snb_hw_cache_event_ids
 #define HSW_L3_MISS_REMOTE		(HSW_L3_MISS_REMOTE_HOP0|\
 					 HSW_L3_MISS_REMOTE_HOP1|HSW_L3_MISS_REMOTE_HOP2P)
 
+#define BDW_L3_MISS_LOCAL		BIT(26)
+#define BDW_L3_MISS			(BDW_L3_MISS_LOCAL| \
+					 HSW_L3_MISS_REMOTE_HOP0|HSW_L3_MISS_REMOTE_HOP1| \
+					 HSW_L3_MISS_REMOTE_HOP2P)
+
+
 static __initconst const u64 hsw_hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
@@ -2759,6 +2774,41 @@ __init int intel_pmu_init(void)
 		pr_cont("Haswell events, ");
 		break;
 
+	case 61: /* 14nm Broadwell Core-M */
+	case 86: /* 14nm Broadwell Xeon D */
+		x86_pmu.late_ack = true;
+		memcpy(hw_cache_event_ids, hsw_hw_cache_event_ids, sizeof(hw_cache_event_ids));
+		memcpy(hw_cache_extra_regs, hsw_hw_cache_extra_regs, sizeof(hw_cache_extra_regs));
+
+		/* L3_MISS_LOCAL_DRAM is BIT(26) in Broadwell */
+		hw_cache_extra_regs[C(LL)][C(OP_READ)][C(RESULT_MISS)] = HSW_DEMAND_READ |
+									 BDW_L3_MISS|HSW_ANY_SNOOP|
+									 HSW_SUPPLIER_NONE;
+		hw_cache_extra_regs[C(LL)][C(OP_WRITE)][C(RESULT_MISS)] = HSW_DEMAND_WRITE|BDW_L3_MISS|
+									  HSW_ANY_SNOOP|HSW_SUPPLIER_NONE;
+		hw_cache_extra_regs[C(NODE)][C(OP_READ)][C(RESULT_ACCESS)] = HSW_DEMAND_READ|
+									     BDW_L3_MISS_LOCAL|HSW_SUPPLIER_NONE|
+									     HSW_ANY_SNOOP;
+		hw_cache_extra_regs[C(NODE)][C(OP_WRITE)][C(RESULT_ACCESS)] = HSW_DEMAND_WRITE|
+									      BDW_L3_MISS_LOCAL|HSW_SUPPLIER_NONE|
+									      HSW_ANY_SNOOP;
+
+		intel_pmu_lbr_init_snb();
+
+		x86_pmu.event_constraints = intel_bdw_event_constraints;
+		x86_pmu.pebs_constraints = intel_hsw_pebs_event_constraints;
+		x86_pmu.extra_regs = intel_snbep_extra_regs;
+		x86_pmu.pebs_aliases = intel_pebs_aliases_snb;
+		/* all extra regs are per-cpu when HT is on */
+		x86_pmu.er_flags |= ERF_HAS_RSP_1;
+		x86_pmu.er_flags |= ERF_NO_HT_SHARING;
+
+		x86_pmu.hw_config = hsw_hw_config;
+		x86_pmu.get_event_constraints = hsw_get_event_constraints;
+		x86_pmu.cpu_events = hsw_events_attrs;
+		pr_cont("Broadwell events, ");
+		break;
+
 	default:
 		switch (x86_pmu.version) {
 		case 1:
-- 
1.9.3

Re: [PATCH 2/3] perf, x86: Add Broadwell core support

[Peter Zijlstra]

[-- Attachment #1: Type: text/plain, Size: 305 bytes --]

On Tue, Aug 12, 2014 at 06:45:14PM -0700, Andi Kleen wrote:
> +	case 61: /* Broadwell */
> +	case 71:
> +	case 79:

Same deal; please put a descriptive comment per model number.

https://git.kernel.org/cgit/linux/kernel/git/peterz/queue.git/commit/?h=perf/core&id=a42c1bf50d297fcb96a3c71b0fdc2da6dbe47030

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

[PATCH 2/3] perf, x86: Add Broadwell core support

[Andi Kleen]

From: Andi Kleen <ak@linux.intel.com>

Add Broadwell support for Broadwell Client to perf.
This is very similar to Haswell.  It uses a new cache event table,
because there were various changes there.

The constraint list has one new event that needs to be handled over Haswell.

The PEBS event list is the same, so we reuse Haswell's.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
 arch/x86/kernel/cpu/perf_event_intel.c | 152 +++++++++++++++++++++++++++++++++
 1 file changed, 152 insertions(+)

diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index ef6c8b7..dd0ea95 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -220,6 +220,15 @@ static struct event_constraint intel_hsw_event_constraints[] = {
 	EVENT_CONSTRAINT_END
 };
 
+struct event_constraint intel_bdw_event_constraints[] = {
+	FIXED_EVENT_CONSTRAINT(0x00c0, 0),	/* INST_RETIRED.ANY */
+	FIXED_EVENT_CONSTRAINT(0x003c, 1),	/* CPU_CLK_UNHALTED.CORE */
+	FIXED_EVENT_CONSTRAINT(0x0300, 2),	/* CPU_CLK_UNHALTED.REF */
+	INTEL_UEVENT_CONSTRAINT(0x148, 0x4),	/* L1D_PEND_MISS.PENDING */
+	INTEL_EVENT_CONSTRAINT(0xa3, 0x4),	/* CYCLE_ACTIVITY.* */
+	EVENT_CONSTRAINT_END
+};
+
 static u64 intel_pmu_event_map(int hw_event)
 {
 	return intel_perfmon_event_map[hw_event];
@@ -415,6 +424,126 @@ static __initconst const u64 snb_hw_cache_event_ids
 
 };
 
+static __initconst const u64 bdw_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)   ] = 0x151, 	/* L1D.REPLACEMENT */
+	},
+	[ 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) ] = 0x0,
+		[ C(RESULT_MISS)   ] = 0x280, 	/* 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) ] = {
+		/* OFFCORE_RESPONSE:ALL_DATA_RD|ALL_CODE_RD */
+		[ C(RESULT_ACCESS) ] = 0x1b7,
+		/* OFFCORE_RESPONSE:ALL_DATA_RD|ALL_CODE_RD|SUPPLIER_NONE|
+                   L3_MISS|ANY_SNOOP */
+		[ C(RESULT_MISS)   ] = 0x1b7,
+	},
+	[ C(OP_WRITE) ] = {
+		[ C(RESULT_ACCESS) ] = 0x1b7, 	/* OFFCORE_RESPONSE:ALL_RFO */
+		/* OFFCORE_RESPONSE:ALL_RFO|SUPPLIER_NONE|L3_MISS|ANY_SNOOP */
+		[ C(RESULT_MISS)   ] = 0x1b7,
+	},
+	[ 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)   ] = 0x108, 	/* DTLB_LOAD_MISSES.MISS_CAUSES_A_WALK */
+	},
+	[ C(OP_WRITE) ] = {
+		[ C(RESULT_ACCESS) ] = 0x82d0, 	/* MEM_UOPS_RETIRED.ALL_STORES */
+		[ C(RESULT_MISS)   ] = 0x149, 	/* DTLB_STORE_MISSES.MISS_CAUSES_A_WALK */
+	},
+	[ C(OP_PREFETCH) ] = {
+		[ C(RESULT_ACCESS) ] = 0x0,
+		[ C(RESULT_MISS)   ] = 0x0,
+	},
+ },
+ [ C(ITLB) ] = {
+	[ C(OP_READ) ] = {
+		[ C(RESULT_ACCESS) ] = 0x6085, 	/* ITLB_MISSES.STLB_HIT */
+		[ C(RESULT_MISS)   ] = 0x185, 	/* ITLB_MISSES.MISS_CAUSES_A_WALK */
+	},
+	[ 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) ] = 0xc4, 	/* BR_INST_RETIRED.ALL_BRANCHES */
+		[ C(RESULT_MISS)   ] = 0xc5, 	/* 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 bdw_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) ] = {
+		/* OFFCORE_RESPONSE:ALL_DATA_RD|ALL_CODE_RD */
+		[ C(RESULT_ACCESS) ] = 0x2d5,
+		/* OFFCORE_RESPONSE:ALL_DATA_RD|ALL_CODE_RD|SUPPLIER_NONE|
+                   L3_MISS|ANY_SNOOP */
+		[ C(RESULT_MISS)   ] = 0x3fbc0202d5,
+	},
+	[ C(OP_WRITE) ] = {
+		[ C(RESULT_ACCESS) ] = 0x122, 	/* OFFCORE_RESPONSE:ALL_RFO */
+		/* OFFCORE_RESPONSE:ALL_RFO|SUPPLIER_NONE|L3_MISS|ANY_SNOOP */
+		[ C(RESULT_MISS)   ] = 0x3fbc020122,
+	},
+	[ C(OP_PREFETCH) ] = {
+		[ C(RESULT_ACCESS) ] = 0x0,
+		[ C(RESULT_MISS)   ] = 0x0,
+	},
+ },
+};
+
 static __initconst const u64 westmere_hw_cache_event_ids
 				[PERF_COUNT_HW_CACHE_MAX]
 				[PERF_COUNT_HW_CACHE_OP_MAX]
@@ -2564,6 +2693,29 @@ __init int intel_pmu_init(void)
 		pr_cont("Haswell events, ");
 		break;
 
+	case 61: /* Broadwell */
+	case 71:
+	case 79:
+		x86_pmu.late_ack = true;
+		memcpy(hw_cache_event_ids, bdw_hw_cache_event_ids, sizeof(hw_cache_event_ids));
+		memcpy(hw_cache_extra_regs, bdw_hw_cache_extra_regs, sizeof(hw_cache_extra_regs));
+
+		intel_pmu_lbr_init_snb();
+
+		x86_pmu.event_constraints = intel_bdw_event_constraints;
+		x86_pmu.pebs_constraints = intel_hsw_pebs_event_constraints;
+		x86_pmu.extra_regs = intel_snbep_extra_regs;
+		x86_pmu.pebs_aliases = intel_pebs_aliases_snb;
+		/* all extra regs are per-cpu when HT is on */
+		x86_pmu.er_flags |= ERF_HAS_RSP_1;
+		x86_pmu.er_flags |= ERF_NO_HT_SHARING;
+
+		x86_pmu.hw_config = hsw_hw_config;
+		x86_pmu.get_event_constraints = hsw_get_event_constraints;
+		x86_pmu.cpu_events = hsw_events_attrs;
+		pr_cont("Broadwell events, ");
+		break;
+
 	default:
 		switch (x86_pmu.version) {
 		case 1:
-- 
1.9.3