Merge branch 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

* 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (106 commits)
  perf kvm: Fix copy & paste error in description
  perf script: Kill script_spec__delete
  perf top: Fix a memory leak
  perf stat: Introduce get_ratio_color() helper
  perf session: Remove impossible condition check
  perf tools: Fix feature-bits rework fallout, remove unused variable
  perf script: Add generic perl handler to process events
  perf tools: Use for_each_set_bit() to iterate over feature flags
  perf tools: Unify handling of features when writing feature section
  perf report: Accept fifos as input file
  perf tools: Moving code in some files
  perf tools: Fix out-of-bound access to struct perf_session
  perf tools: Continue processing header on unknown features
  perf tools: Improve macros for struct feature_ops
  perf: builtin-record: Document and check that mmap_pages must be a power of two.
  perf: builtin-record: Provide advice if mmap'ing fails with EPERM.
  perf tools: Fix truncated annotation
  perf script: look up thread using tid instead of pid
  perf tools: Look up thread names for system wide profiling
  perf tools: Fix comm for processes with named threads
  ...

4 files changed, 315 insertions, 88 deletions

diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index 2bda212a001..5adce1040b1 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -484,18 +484,195 @@ static inline int is_x86_event(struct perf_event *event)
 	return event->pmu == &pmu;
 }
 
+/*
+ * Event scheduler state:
+ *
+ * Assign events iterating over all events and counters, beginning
+ * with events with least weights first. Keep the current iterator
+ * state in struct sched_state.
+ */
+struct sched_state {
+	int	weight;
+	int	event;		/* event index */
+	int	counter;	/* counter index */
+	int	unassigned;	/* number of events to be assigned left */
+	unsigned long used[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
+};
+
+/* Total max is X86_PMC_IDX_MAX, but we are O(n!) limited */
+#define	SCHED_STATES_MAX	2
+
+struct perf_sched {
+	int			max_weight;
+	int			max_events;
+	struct event_constraint	**constraints;
+	struct sched_state	state;
+	int			saved_states;
+	struct sched_state	saved[SCHED_STATES_MAX];
+};
+
+/*
+ * Initialize interator that runs through all events and counters.
+ */
+static void perf_sched_init(struct perf_sched *sched, struct event_constraint **c,
+			    int num, int wmin, int wmax)
+{
+	int idx;
+
+	memset(sched, 0, sizeof(*sched));
+	sched->max_events	= num;
+	sched->max_weight	= wmax;
+	sched->constraints	= c;
+
+	for (idx = 0; idx < num; idx++) {
+		if (c[idx]->weight == wmin)
+			break;
+	}
+
+	sched->state.event	= idx;		/* start with min weight */
+	sched->state.weight	= wmin;
+	sched->state.unassigned	= num;
+}
+
+static void perf_sched_save_state(struct perf_sched *sched)
+{
+	if (WARN_ON_ONCE(sched->saved_states >= SCHED_STATES_MAX))
+		return;
+
+	sched->saved[sched->saved_states] = sched->state;
+	sched->saved_states++;
+}
+
+static bool perf_sched_restore_state(struct perf_sched *sched)
+{
+	if (!sched->saved_states)
+		return false;
+
+	sched->saved_states--;
+	sched->state = sched->saved[sched->saved_states];
+
+	/* continue with next counter: */
+	clear_bit(sched->state.counter++, sched->state.used);
+
+	return true;
+}
+
+/*
+ * Select a counter for the current event to schedule. Return true on
+ * success.
+ */
+static bool __perf_sched_find_counter(struct perf_sched *sched)
+{
+	struct event_constraint *c;
+	int idx;
+
+	if (!sched->state.unassigned)
+		return false;
+
+	if (sched->state.event >= sched->max_events)
+		return false;
+
+	c = sched->constraints[sched->state.event];
+
+	/* Prefer fixed purpose counters */
+	if (x86_pmu.num_counters_fixed) {
+		idx = X86_PMC_IDX_FIXED;
+		for_each_set_bit_cont(idx, c->idxmsk, X86_PMC_IDX_MAX) {
+			if (!__test_and_set_bit(idx, sched->state.used))
+				goto done;
+		}
+	}
+	/* Grab the first unused counter starting with idx */
+	idx = sched->state.counter;
+	for_each_set_bit_cont(idx, c->idxmsk, X86_PMC_IDX_FIXED) {
+		if (!__test_and_set_bit(idx, sched->state.used))
+			goto done;
+	}
+
+	return false;
+
+done:
+	sched->state.counter = idx;
+
+	if (c->overlap)
+		perf_sched_save_state(sched);
+
+	return true;
+}
+
+static bool perf_sched_find_counter(struct perf_sched *sched)
+{
+	while (!__perf_sched_find_counter(sched)) {
+		if (!perf_sched_restore_state(sched))
+			return false;
+	}
+
+	return true;
+}
+
+/*
+ * Go through all unassigned events and find the next one to schedule.
+ * Take events with the least weight first. Return true on success.
+ */
+static bool perf_sched_next_event(struct perf_sched *sched)
+{
+	struct event_constraint *c;
+
+	if (!sched->state.unassigned || !--sched->state.unassigned)
+		return false;
+
+	do {
+		/* next event */
+		sched->state.event++;
+		if (sched->state.event >= sched->max_events) {
+			/* next weight */
+			sched->state.event = 0;
+			sched->state.weight++;
+			if (sched->state.weight > sched->max_weight)
+				return false;
+		}
+		c = sched->constraints[sched->state.event];
+	} while (c->weight != sched->state.weight);
+
+	sched->state.counter = 0;	/* start with first counter */
+
+	return true;
+}
+
+/*
+ * Assign a counter for each event.
+ */
+static int perf_assign_events(struct event_constraint **constraints, int n,
+			      int wmin, int wmax, int *assign)
+{
+	struct perf_sched sched;
+
+	perf_sched_init(&sched, constraints, n, wmin, wmax);
+
+	do {
+		if (!perf_sched_find_counter(&sched))
+			break;	/* failed */
+		if (assign)
+			assign[sched.state.event] = sched.state.counter;
+	} while (perf_sched_next_event(&sched));
+
+	return sched.state.unassigned;
+}
+
 int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
 {
 	struct event_constraint *c, *constraints[X86_PMC_IDX_MAX];
 	unsigned long used_mask[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
-	int i, j, w, wmax, num = 0;
+	int i, wmin, wmax, num = 0;
 	struct hw_perf_event *hwc;
 
 	bitmap_zero(used_mask, X86_PMC_IDX_MAX);
 
-	for (i = 0; i < n; i++) {
+	for (i = 0, wmin = X86_PMC_IDX_MAX, wmax = 0; i < n; i++) {
 		c = x86_pmu.get_event_constraints(cpuc, cpuc->event_list[i]);
 		constraints[i] = c;
+		wmin = min(wmin, c->weight);
+		wmax = max(wmax, c->weight);
 	}
 
 	/*
@@ -521,60 +698,12 @@ int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
 		if (assign)
 			assign[i] = hwc->idx;
 	}
-	if (i == n)
-		goto done;
-
-	/*
-	 * begin slow path
-	 */
-
-	bitmap_zero(used_mask, X86_PMC_IDX_MAX);
 
-	/*
-	 * weight = number of possible counters
-	 *
-	 * 1    = most constrained, only works on one counter
-	 * wmax = least constrained, works on any counter
-	 *
-	 * assign events to counters starting with most
-	 * constrained events.
-	 */
-	wmax = x86_pmu.num_counters;
+	/* slow path */
+	if (i != n)
+		num = perf_assign_events(constraints, n, wmin, wmax, assign);
 
 	/*
-	 * when fixed event counters are present,
-	 * wmax is incremented by 1 to account
-	 * for one more choice
-	 */
-	if (x86_pmu.num_counters_fixed)
-		wmax++;
-
-	for (w = 1, num = n; num && w <= wmax; w++) {
-		/* for each event */
-		for (i = 0; num && i < n; i++) {
-			c = constraints[i];
-			hwc = &cpuc->event_list[i]->hw;
-
-			if (c->weight != w)
-				continue;
-
-			for_each_set_bit(j, c->idxmsk, X86_PMC_IDX_MAX) {
-				if (!test_bit(j, used_mask))
-					break;
-			}
-
-			if (j == X86_PMC_IDX_MAX)
-				break;
-
-			__set_bit(j, used_mask);
-
-			if (assign)
-				assign[i] = j;
-			num--;
-		}
-	}
-done:
-	/*
 	 * scheduling failed or is just a simulation,
 	 * free resources if necessary
 	 */
@@ -1119,6 +1248,7 @@ static void __init pmu_check_apic(void)
 
 static int __init init_hw_perf_events(void)
 {
+	struct x86_pmu_quirk *quirk;
 	struct event_constraint *c;
 	int err;
 
@@ -1147,8 +1277,8 @@ static int __init init_hw_perf_events(void)
 
 	pr_cont("%s PMU driver.\n", x86_pmu.name);
 
-	if (x86_pmu.quirks)
-		x86_pmu.quirks();
+	for (quirk = x86_pmu.quirks; quirk; quirk = quirk->next)
+		quirk->func();
 
 	if (x86_pmu.num_counters > X86_PMC_MAX_GENERIC) {
 		WARN(1, KERN_ERR "hw perf events %d > max(%d), clipping!",
@@ -1171,12 +1301,18 @@ static int __init init_hw_perf_events(void)
 
 	unconstrained = (struct event_constraint)
 		__EVENT_CONSTRAINT(0, (1ULL << x86_pmu.num_counters) - 1,
-				   0, x86_pmu.num_counters);
+				   0, x86_pmu.num_counters, 0);
 
 	if (x86_pmu.event_constraints) {
+		/*
+		 * event on fixed counter2 (REF_CYCLES) only works on this
+		 * counter, so do not extend mask to generic counters
+		 */
 		for_each_event_constraint(c, x86_pmu.event_constraints) {
-			if (c->cmask != X86_RAW_EVENT_MASK)
+			if (c->cmask != X86_RAW_EVENT_MASK
+			    || c->idxmsk64 == X86_PMC_MSK_FIXED_REF_CYCLES) {
 				continue;
+			}
 
 			c->idxmsk64 |= (1ULL << x86_pmu.num_counters) - 1;
 			c->weight += x86_pmu.num_counters;
@@ -1566,3 +1702,15 @@ unsigned long perf_misc_flags(struct pt_regs *regs)
 
 	return misc;
 }
+
+void perf_get_x86_pmu_capability(struct x86_pmu_capability *cap)
+{
+	cap->version		= x86_pmu.version;
+	cap->num_counters_gp	= x86_pmu.num_counters;
+	cap->num_counters_fixed	= x86_pmu.num_counters_fixed;
+	cap->bit_width_gp	= x86_pmu.cntval_bits;
+	cap->bit_width_fixed	= x86_pmu.cntval_bits;
+	cap->events_mask	= (unsigned int)x86_pmu.events_maskl;
+	cap->events_mask_len	= x86_pmu.events_mask_len;
+}
+EXPORT_SYMBOL_GPL(perf_get_x86_pmu_capability);
diff --git a/arch/x86/kernel/cpu/perf_event.h b/arch/x86/kernel/cpu/perf_event.h
index b9698d40ac4..8944062f46e 100644
--- a/arch/x86/kernel/cpu/perf_event.h
+++ b/arch/x86/kernel/cpu/perf_event.h
@@ -45,6 +45,7 @@ struct event_constraint {
 	u64	code;
 	u64	cmask;
 	int	weight;
+	int	overlap;
 };
 
 struct amd_nb {
@@ -151,15 +152,40 @@ struct cpu_hw_events {
 	void				*kfree_on_online;
 };
 
-#define __EVENT_CONSTRAINT(c, n, m, w) {\
+#define __EVENT_CONSTRAINT(c, n, m, w, o) {\
 	{ .idxmsk64 = (n) },		\
 	.code = (c),			\
 	.cmask = (m),			\
 	.weight = (w),			\
+	.overlap = (o),			\
 }
 
 #define EVENT_CONSTRAINT(c, n, m)	\
-	__EVENT_CONSTRAINT(c, n, m, HWEIGHT(n))
+	__EVENT_CONSTRAINT(c, n, m, HWEIGHT(n), 0)
+
+/*
+ * The overlap flag marks event constraints with overlapping counter
+ * masks. This is the case if the counter mask of such an event is not
+ * a subset of any other counter mask of a constraint with an equal or
+ * higher weight, e.g.:
+ *
+ *  c_overlaps = EVENT_CONSTRAINT_OVERLAP(0, 0x09, 0);
+ *  c_another1 = EVENT_CONSTRAINT(0, 0x07, 0);
+ *  c_another2 = EVENT_CONSTRAINT(0, 0x38, 0);
+ *
+ * The event scheduler may not select the correct counter in the first
+ * cycle because it needs to know which subsequent events will be
+ * scheduled. It may fail to schedule the events then. So we set the
+ * overlap flag for such constraints to give the scheduler a hint which
+ * events to select for counter rescheduling.
+ *
+ * Care must be taken as the rescheduling algorithm is O(n!) which
+ * will increase scheduling cycles for an over-commited system
+ * dramatically.  The number of such EVENT_CONSTRAINT_OVERLAP() macros
+ * and its counter masks must be kept at a minimum.
+ */
+#define EVENT_CONSTRAINT_OVERLAP(c, n, m)	\
+	__EVENT_CONSTRAINT(c, n, m, HWEIGHT(n), 1)
 
 /*
  * Constraint on the Event code.
@@ -235,6 +261,11 @@ union perf_capabilities {
 	u64	capabilities;
 };
 
+struct x86_pmu_quirk {
+	struct x86_pmu_quirk *next;
+	void (*func)(void);
+};
+
 /*
  * struct x86_pmu - generic x86 pmu
  */
@@ -259,6 +290,11 @@ struct x86_pmu {
 	int		num_counters_fixed;
 	int		cntval_bits;
 	u64		cntval_mask;
+	union {
+			unsigned long events_maskl;
+			unsigned long events_mask[BITS_TO_LONGS(ARCH_PERFMON_EVENTS_COUNT)];
+	};
+	int		events_mask_len;
 	int		apic;
 	u64		max_period;
 	struct event_constraint *
@@ -268,7 +304,7 @@ struct x86_pmu {
 	void		(*put_event_constraints)(struct cpu_hw_events *cpuc,
 						 struct perf_event *event);
 	struct event_constraint *event_constraints;
-	void		(*quirks)(void);
+	struct x86_pmu_quirk *quirks;
 	int		perfctr_second_write;
 
 	int		(*cpu_prepare)(int cpu);
@@ -309,6 +345,15 @@ struct x86_pmu {
 	struct perf_guest_switch_msr *(*guest_get_msrs)(int *nr);
 };
 
+#define x86_add_quirk(func_)						\
+do {									\
+	static struct x86_pmu_quirk __quirk __initdata = {		\
+		.func = func_,						\
+	};								\
+	__quirk.next = x86_pmu.quirks;					\
+	x86_pmu.quirks = &__quirk;					\
+} while (0)
+
 #define ERF_NO_HT_SHARING	1
 #define ERF_HAS_RSP_1		2
 
diff --git a/arch/x86/kernel/cpu/perf_event_amd.c b/arch/x86/kernel/cpu/perf_event_amd.c
index aeefd45697a..0397b23be8e 100644
--- a/arch/x86/kernel/cpu/perf_event_amd.c
+++ b/arch/x86/kernel/cpu/perf_event_amd.c
@@ -492,7 +492,7 @@ static __initconst const struct x86_pmu amd_pmu = {
 static struct event_constraint amd_f15_PMC0  = EVENT_CONSTRAINT(0, 0x01, 0);
 static struct event_constraint amd_f15_PMC20 = EVENT_CONSTRAINT(0, 0x07, 0);
 static struct event_constraint amd_f15_PMC3  = EVENT_CONSTRAINT(0, 0x08, 0);
-static struct event_constraint amd_f15_PMC30 = EVENT_CONSTRAINT(0, 0x09, 0);
+static struct event_constraint amd_f15_PMC30 = EVENT_CONSTRAINT_OVERLAP(0, 0x09, 0);
 static struct event_constraint amd_f15_PMC50 = EVENT_CONSTRAINT(0, 0x3F, 0);
 static struct event_constraint amd_f15_PMC53 = EVENT_CONSTRAINT(0, 0x38, 0);
 
diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 121f1be4da1..3bd37bdf1b8 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -28,6 +28,7 @@ static u64 intel_perfmon_event_map[PERF_COUNT_HW_MAX] __read_mostly =
   [PERF_COUNT_HW_BRANCH_INSTRUCTIONS]	= 0x00c4,
   [PERF_COUNT_HW_BRANCH_MISSES]		= 0x00c5,
   [PERF_COUNT_HW_BUS_CYCLES]		= 0x013c,
+  [PERF_COUNT_HW_REF_CPU_CYCLES]	= 0x0300, /* pseudo-encoding */
 };
 
 static struct event_constraint intel_core_event_constraints[] __read_mostly =
@@ -45,12 +46,7 @@ static struct event_constraint intel_core2_event_constraints[] __read_mostly =
 {
 	FIXED_EVENT_CONSTRAINT(0x00c0, 0), /* INST_RETIRED.ANY */
 	FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */
-	/*
-	 * Core2 has Fixed Counter 2 listed as CPU_CLK_UNHALTED.REF and event
-	 * 0x013c as CPU_CLK_UNHALTED.BUS and specifies there is a fixed
-	 * ratio between these counters.
-	 */
-	/* FIXED_EVENT_CONSTRAINT(0x013c, 2),  CPU_CLK_UNHALTED.REF */
+	FIXED_EVENT_CONSTRAINT(0x0300, 2), /* CPU_CLK_UNHALTED.REF */
 	INTEL_EVENT_CONSTRAINT(0x10, 0x1), /* FP_COMP_OPS_EXE */
 	INTEL_EVENT_CONSTRAINT(0x11, 0x2), /* FP_ASSIST */
 	INTEL_EVENT_CONSTRAINT(0x12, 0x2), /* MUL */
@@ -68,7 +64,7 @@ static struct event_constraint intel_nehalem_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), /* CPU_CLK_UNHALTED.REF */
 	INTEL_EVENT_CONSTRAINT(0x40, 0x3), /* L1D_CACHE_LD */
 	INTEL_EVENT_CONSTRAINT(0x41, 0x3), /* L1D_CACHE_ST */
 	INTEL_EVENT_CONSTRAINT(0x42, 0x3), /* L1D_CACHE_LOCK */
@@ -90,7 +86,7 @@ static struct event_constraint intel_westmere_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), /* CPU_CLK_UNHALTED.REF */
 	INTEL_EVENT_CONSTRAINT(0x51, 0x3), /* L1D */
 	INTEL_EVENT_CONSTRAINT(0x60, 0x1), /* OFFCORE_REQUESTS_OUTSTANDING */
 	INTEL_EVENT_CONSTRAINT(0x63, 0x3), /* CACHE_LOCK_CYCLES */
@@ -102,7 +98,7 @@ static struct event_constraint intel_snb_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), /* CPU_CLK_UNHALTED.REF */
 	INTEL_EVENT_CONSTRAINT(0x48, 0x4), /* L1D_PEND_MISS.PENDING */
 	INTEL_UEVENT_CONSTRAINT(0x01c0, 0x2), /* INST_RETIRED.PREC_DIST */
 	INTEL_EVENT_CONSTRAINT(0xcd, 0x8), /* MEM_TRANS_RETIRED.LOAD_LATENCY */
@@ -125,7 +121,7 @@ static struct event_constraint intel_gen_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), /* CPU_CLK_UNHALTED.REF */
 	EVENT_CONSTRAINT_END
 };
 
@@ -1519,7 +1515,7 @@ static __initconst const struct x86_pmu intel_pmu = {
 	.guest_get_msrs		= intel_guest_get_msrs,
 };
 
-static void intel_clovertown_quirks(void)
+static __init void intel_clovertown_quirk(void)
 {
 	/*
 	 * PEBS is unreliable due to:
@@ -1545,19 +1541,60 @@ static void intel_clovertown_quirks(void)
 	x86_pmu.pebs_constraints = NULL;
 }
 
-static void intel_sandybridge_quirks(void)
+static __init void intel_sandybridge_quirk(void)
 {
 	printk(KERN_WARNING "PEBS disabled due to CPU errata.\n");
 	x86_pmu.pebs = 0;
 	x86_pmu.pebs_constraints = NULL;
 }
 
+static const struct { int id; char *name; } intel_arch_events_map[] __initconst = {
+	{ PERF_COUNT_HW_CPU_CYCLES, "cpu cycles" },
+	{ PERF_COUNT_HW_INSTRUCTIONS, "instructions" },
+	{ PERF_COUNT_HW_BUS_CYCLES, "bus cycles" },
+	{ PERF_COUNT_HW_CACHE_REFERENCES, "cache references" },
+	{ PERF_COUNT_HW_CACHE_MISSES, "cache misses" },
+	{ PERF_COUNT_HW_BRANCH_INSTRUCTIONS, "branch instructions" },
+	{ PERF_COUNT_HW_BRANCH_MISSES, "branch misses" },
+};
+
+static __init void intel_arch_events_quirk(void)
+{
+	int bit;
+
+	/* disable event that reported as not presend by cpuid */
+	for_each_set_bit(bit, x86_pmu.events_mask, ARRAY_SIZE(intel_arch_events_map)) {
+		intel_perfmon_event_map[intel_arch_events_map[bit].id] = 0;
+		printk(KERN_WARNING "CPUID marked event: \'%s\' unavailable\n",
+				intel_arch_events_map[bit].name);
+	}
+}
+
+static __init void intel_nehalem_quirk(void)
+{
+	union cpuid10_ebx ebx;
+
+	ebx.full = x86_pmu.events_maskl;
+	if (ebx.split.no_branch_misses_retired) {
+		/*
+		 * Erratum AAJ80 detected, we work it around by using
+		 * the BR_MISP_EXEC.ANY event. This will over-count
+		 * branch-misses, but it's still much better than the
+		 * architectural event which is often completely bogus:
+		 */
+		intel_perfmon_event_map[PERF_COUNT_HW_BRANCH_MISSES] = 0x7f89;
+		ebx.split.no_branch_misses_retired = 0;
+		x86_pmu.events_maskl = ebx.full;
+		printk(KERN_INFO "CPU erratum AAJ80 worked around\n");
+	}
+}
+
 __init int intel_pmu_init(void)
 {
 	union cpuid10_edx edx;
 	union cpuid10_eax eax;
+	union cpuid10_ebx ebx;
 	unsigned int unused;
-	unsigned int ebx;
 	int version;
 
 	if (!cpu_has(&boot_cpu_data, X86_FEATURE_ARCH_PERFMON)) {
@@ -1574,8 +1611,8 @@ __init int intel_pmu_init(void)
 	 * Check whether the Architectural PerfMon supports
 	 * Branch Misses Retired hw_event or not.
 	 */
-	cpuid(10, &eax.full, &ebx, &unused, &edx.full);
-	if (eax.split.mask_length <= ARCH_PERFMON_BRANCH_MISSES_RETIRED)
+	cpuid(10, &eax.full, &ebx.full, &unused, &edx.full);
+	if (eax.split.mask_length < ARCH_PERFMON_EVENTS_COUNT)
 		return -ENODEV;
 
 	version = eax.split.version_id;
@@ -1589,6 +1626,9 @@ __init int intel_pmu_init(void)
 	x86_pmu.cntval_bits		= eax.split.bit_width;
 	x86_pmu.cntval_mask		= (1ULL << eax.split.bit_width) - 1;
 
+	x86_pmu.events_maskl		= ebx.full;
+	x86_pmu.events_mask_len		= eax.split.mask_length;
+
 	/*
 	 * Quirk: v2 perfmon does not report fixed-purpose events, so
 	 * assume at least 3 events:
@@ -1608,6 +1648,8 @@ __init int intel_pmu_init(void)
 
 	intel_ds_init();
 
+	x86_add_quirk(intel_arch_events_quirk); /* Install first, so it runs last */
+
 	/*
 	 * Install the hw-cache-events table:
 	 */
@@ -1617,7 +1659,7 @@ __init int intel_pmu_init(void)
 		break;
 
 	case 15: /* original 65 nm celeron/pentium/core2/xeon, "Merom"/"Conroe" */
-		x86_pmu.quirks = intel_clovertown_quirks;
+		x86_add_quirk(intel_clovertown_quirk);
 	case 22: /* single-core 65 nm celeron/core2solo "Merom-L"/"Conroe-L" */
 	case 23: /* current 45 nm celeron/core2/xeon "Penryn"/"Wolfdale" */
 	case 29: /* six-core 45 nm xeon "Dunnington" */
@@ -1651,17 +1693,8 @@ __init int intel_pmu_init(void)
 		/* UOPS_EXECUTED.CORE_ACTIVE_CYCLES,c=1,i=1 */
 		intel_perfmon_event_map[PERF_COUNT_HW_STALLED_CYCLES_BACKEND] = 0x1803fb1;
 
-		if (ebx & 0x40) {
-			/*
-			 * Erratum AAJ80 detected, we work it around by using
-			 * the BR_MISP_EXEC.ANY event. This will over-count
-			 * branch-misses, but it's still much better than the
-			 * architectural event which is often completely bogus:
-			 */
-			intel_perfmon_event_map[PERF_COUNT_HW_BRANCH_MISSES] = 0x7f89;
+		x86_add_quirk(intel_nehalem_quirk);
 
-			pr_cont("erratum AAJ80 worked around, ");
-		}
 		pr_cont("Nehalem events, ");
 		break;
 
@@ -1701,7 +1734,7 @@ __init int intel_pmu_init(void)
 		break;
 
 	case 42: /* SandyBridge */
-		x86_pmu.quirks = intel_sandybridge_quirks;
+		x86_add_quirk(intel_sandybridge_quirk);
 	case 45: /* SandyBridge, "Romely-EP" */
 		memcpy(hw_cache_event_ids, snb_hw_cache_event_ids,
 		       sizeof(hw_cache_event_ids));
@@ -1738,5 +1771,6 @@ __init int intel_pmu_init(void)
 			break;
 		}
 	}
+
 	return 0;
 }