diff --git a/arch/arm/kernel/perf_event_v6.c b/arch/arm/kernel/perf_event_v6.c
index be42c4f66a40..1ae99deeec54 100644
--- a/arch/arm/kernel/perf_event_v6.c
+++ b/arch/arm/kernel/perf_event_v6.c
@@ -233,7 +233,7 @@ armv6_pmcr_counter_has_overflowed(unsigned long pmcr,
 	return ret;
 }
 
-static inline u32 armv6pmu_read_counter(struct perf_event *event)
+static inline u64 armv6pmu_read_counter(struct perf_event *event)
 {
 	struct hw_perf_event *hwc = &event->hw;
 	int counter = hwc->idx;
@@ -251,7 +251,7 @@ static inline u32 armv6pmu_read_counter(struct perf_event *event)
 	return value;
 }
 
-static inline void armv6pmu_write_counter(struct perf_event *event, u32 value)
+static inline void armv6pmu_write_counter(struct perf_event *event, u64 value)
 {
 	struct hw_perf_event *hwc = &event->hw;
 	int counter = hwc->idx;
@@ -411,6 +411,12 @@ armv6pmu_get_event_idx(struct pmu_hw_events *cpuc,
 	}
 }
 
+static void armv6pmu_clear_event_idx(struct pmu_hw_events *cpuc,
+				     struct perf_event *event)
+{
+	clear_bit(event->hw.idx, cpuc->used_mask);
+}
+
 static void armv6pmu_disable_event(struct perf_event *event)
 {
 	unsigned long val, mask, evt, flags;
@@ -491,11 +497,11 @@ static void armv6pmu_init(struct arm_pmu *cpu_pmu)
 	cpu_pmu->read_counter	= armv6pmu_read_counter;
 	cpu_pmu->write_counter	= armv6pmu_write_counter;
 	cpu_pmu->get_event_idx	= armv6pmu_get_event_idx;
+	cpu_pmu->clear_event_idx = armv6pmu_clear_event_idx;
 	cpu_pmu->start		= armv6pmu_start;
 	cpu_pmu->stop		= armv6pmu_stop;
 	cpu_pmu->map_event	= armv6_map_event;
 	cpu_pmu->num_events	= 3;
-	cpu_pmu->max_period	= (1LLU << 32) - 1;
 }
 
 static int armv6_1136_pmu_init(struct arm_pmu *cpu_pmu)
@@ -542,11 +548,11 @@ static int armv6mpcore_pmu_init(struct arm_pmu *cpu_pmu)
 	cpu_pmu->read_counter	= armv6pmu_read_counter;
 	cpu_pmu->write_counter	= armv6pmu_write_counter;
 	cpu_pmu->get_event_idx	= armv6pmu_get_event_idx;
+	cpu_pmu->clear_event_idx = armv6pmu_clear_event_idx;
 	cpu_pmu->start		= armv6pmu_start;
 	cpu_pmu->stop		= armv6pmu_stop;
 	cpu_pmu->map_event	= armv6mpcore_map_event;
 	cpu_pmu->num_events	= 3;
-	cpu_pmu->max_period	= (1LLU << 32) - 1;
 
 	return 0;
 }
diff --git a/arch/arm/kernel/perf_event_v7.c b/arch/arm/kernel/perf_event_v7.c
index 57f01e059f39..a4fb0f8b8f84 100644
--- a/arch/arm/kernel/perf_event_v7.c
+++ b/arch/arm/kernel/perf_event_v7.c
@@ -743,7 +743,7 @@ static inline void armv7_pmnc_select_counter(int idx)
 	isb();
 }
 
-static inline u32 armv7pmu_read_counter(struct perf_event *event)
+static inline u64 armv7pmu_read_counter(struct perf_event *event)
 {
 	struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu);
 	struct hw_perf_event *hwc = &event->hw;
@@ -763,7 +763,7 @@ static inline u32 armv7pmu_read_counter(struct perf_event *event)
 	return value;
 }
 
-static inline void armv7pmu_write_counter(struct perf_event *event, u32 value)
+static inline void armv7pmu_write_counter(struct perf_event *event, u64 value)
 {
 	struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu);
 	struct hw_perf_event *hwc = &event->hw;
@@ -1058,6 +1058,12 @@ static int armv7pmu_get_event_idx(struct pmu_hw_events *cpuc,
 	return -EAGAIN;
 }
 
+static void armv7pmu_clear_event_idx(struct pmu_hw_events *cpuc,
+				     struct perf_event *event)
+{
+	clear_bit(event->hw.idx, cpuc->used_mask);
+}
+
 /*
  * Add an event filter to a given event. This will only work for PMUv2 PMUs.
  */
@@ -1167,10 +1173,10 @@ static void armv7pmu_init(struct arm_pmu *cpu_pmu)
 	cpu_pmu->read_counter	= armv7pmu_read_counter;
 	cpu_pmu->write_counter	= armv7pmu_write_counter;
 	cpu_pmu->get_event_idx	= armv7pmu_get_event_idx;
+	cpu_pmu->clear_event_idx = armv7pmu_clear_event_idx;
 	cpu_pmu->start		= armv7pmu_start;
 	cpu_pmu->stop		= armv7pmu_stop;
 	cpu_pmu->reset		= armv7pmu_reset;
-	cpu_pmu->max_period	= (1LLU << 32) - 1;
 };
 
 static void armv7_read_num_pmnc_events(void *info)
@@ -1638,6 +1644,7 @@ static void krait_pmu_clear_event_idx(struct pmu_hw_events *cpuc,
 	bool venum_event = EVENT_VENUM(hwc->config_base);
 	bool krait_event = EVENT_CPU(hwc->config_base);
 
+	armv7pmu_clear_event_idx(cpuc, event);
 	if (venum_event || krait_event) {
 		bit = krait_event_to_bit(event, region, group);
 		clear_bit(bit, cpuc->used_mask);
@@ -1967,6 +1974,7 @@ static void scorpion_pmu_clear_event_idx(struct pmu_hw_events *cpuc,
 	bool venum_event = EVENT_VENUM(hwc->config_base);
 	bool scorpion_event = EVENT_CPU(hwc->config_base);
 
+	armv7pmu_clear_event_idx(cpuc, event);
 	if (venum_event || scorpion_event) {
 		bit = scorpion_event_to_bit(event, region, group);
 		clear_bit(bit, cpuc->used_mask);
@@ -2030,6 +2038,7 @@ static struct platform_driver armv7_pmu_driver = {
 	.driver		= {
 		.name	= "armv7-pmu",
 		.of_match_table = armv7_pmu_of_device_ids,
+		.suppress_bind_attrs = true,
 	},
 	.probe		= armv7_pmu_device_probe,
 };
diff --git a/arch/arm/kernel/perf_event_xscale.c b/arch/arm/kernel/perf_event_xscale.c
index 88d1a76f5367..f6cdcacfb96d 100644
--- a/arch/arm/kernel/perf_event_xscale.c
+++ b/arch/arm/kernel/perf_event_xscale.c
@@ -292,6 +292,12 @@ xscale1pmu_get_event_idx(struct pmu_hw_events *cpuc,
 	}
 }
 
+static void xscalepmu_clear_event_idx(struct pmu_hw_events *cpuc,
+				     struct perf_event *event)
+{
+	clear_bit(event->hw.idx, cpuc->used_mask);
+}
+
 static void xscale1pmu_start(struct arm_pmu *cpu_pmu)
 {
 	unsigned long flags, val;
@@ -316,7 +322,7 @@ static void xscale1pmu_stop(struct arm_pmu *cpu_pmu)
 	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
-static inline u32 xscale1pmu_read_counter(struct perf_event *event)
+static inline u64 xscale1pmu_read_counter(struct perf_event *event)
 {
 	struct hw_perf_event *hwc = &event->hw;
 	int counter = hwc->idx;
@@ -337,7 +343,7 @@ static inline u32 xscale1pmu_read_counter(struct perf_event *event)
 	return val;
 }
 
-static inline void xscale1pmu_write_counter(struct perf_event *event, u32 val)
+static inline void xscale1pmu_write_counter(struct perf_event *event, u64 val)
 {
 	struct hw_perf_event *hwc = &event->hw;
 	int counter = hwc->idx;
@@ -370,11 +376,11 @@ static int xscale1pmu_init(struct arm_pmu *cpu_pmu)
 	cpu_pmu->read_counter	= xscale1pmu_read_counter;
 	cpu_pmu->write_counter	= xscale1pmu_write_counter;
 	cpu_pmu->get_event_idx	= xscale1pmu_get_event_idx;
+	cpu_pmu->clear_event_idx = xscalepmu_clear_event_idx;
 	cpu_pmu->start		= xscale1pmu_start;
 	cpu_pmu->stop		= xscale1pmu_stop;
 	cpu_pmu->map_event	= xscale_map_event;
 	cpu_pmu->num_events	= 3;
-	cpu_pmu->max_period	= (1LLU << 32) - 1;
 
 	return 0;
 }
@@ -679,7 +685,7 @@ static void xscale2pmu_stop(struct arm_pmu *cpu_pmu)
 	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
-static inline u32 xscale2pmu_read_counter(struct perf_event *event)
+static inline u64 xscale2pmu_read_counter(struct perf_event *event)
 {
 	struct hw_perf_event *hwc = &event->hw;
 	int counter = hwc->idx;
@@ -706,7 +712,7 @@ static inline u32 xscale2pmu_read_counter(struct perf_event *event)
 	return val;
 }
 
-static inline void xscale2pmu_write_counter(struct perf_event *event, u32 val)
+static inline void xscale2pmu_write_counter(struct perf_event *event, u64 val)
 {
 	struct hw_perf_event *hwc = &event->hw;
 	int counter = hwc->idx;
@@ -739,11 +745,11 @@ static int xscale2pmu_init(struct arm_pmu *cpu_pmu)
 	cpu_pmu->read_counter	= xscale2pmu_read_counter;
 	cpu_pmu->write_counter	= xscale2pmu_write_counter;
 	cpu_pmu->get_event_idx	= xscale2pmu_get_event_idx;
+	cpu_pmu->clear_event_idx = xscalepmu_clear_event_idx;
 	cpu_pmu->start		= xscale2pmu_start;
 	cpu_pmu->stop		= xscale2pmu_stop;
 	cpu_pmu->map_event	= xscale_map_event;
 	cpu_pmu->num_events	= 5;
-	cpu_pmu->max_period	= (1LLU << 32) - 1;
 
 	return 0;
 }
diff --git a/arch/arm64/kernel/perf_event.c b/arch/arm64/kernel/perf_event.c
index 33147aacdafd..dfff5ed5c625 100644
--- a/arch/arm64/kernel/perf_event.c
+++ b/arch/arm64/kernel/perf_event.c
@@ -446,9 +446,16 @@ static struct attribute_group armv8_pmuv3_events_attr_group = {
 };
 
 PMU_FORMAT_ATTR(event, "config:0-15");
+PMU_FORMAT_ATTR(long, "config1:0");
+
+static inline bool armv8pmu_event_is_64bit(struct perf_event *event)
+{
+	return event->attr.config1 & 0x1;
+}
 
 static struct attribute *armv8_pmuv3_format_attrs[] = {
 	&format_attr_event.attr,
+	&format_attr_long.attr,
 	NULL,
 };
 
@@ -465,6 +472,21 @@ static struct attribute_group armv8_pmuv3_format_attr_group = {
 #define	ARMV8_IDX_COUNTER_LAST(cpu_pmu) \
 	(ARMV8_IDX_CYCLE_COUNTER + cpu_pmu->num_events - 1)
 
+/*
+ * We must chain two programmable counters for 64 bit events,
+ * except when we have allocated the 64bit cycle counter (for CPU
+ * cycles event). This must be called only when the event has
+ * a counter allocated.
+ */
+static inline bool armv8pmu_event_is_chained(struct perf_event *event)
+{
+	int idx = event->hw.idx;
+
+	return !WARN_ON(idx < 0) &&
+	       armv8pmu_event_is_64bit(event) &&
+	       (idx != ARMV8_IDX_CYCLE_COUNTER);
+}
+
 /*
  * ARMv8 low level PMU access
  */
@@ -503,34 +525,68 @@ static inline int armv8pmu_counter_has_overflowed(u32 pmnc, int idx)
 	return pmnc & BIT(ARMV8_IDX_TO_COUNTER(idx));
 }
 
-static inline int armv8pmu_select_counter(int idx)
+static inline void armv8pmu_select_counter(int idx)
 {
 	u32 counter = ARMV8_IDX_TO_COUNTER(idx);
 	write_sysreg(counter, pmselr_el0);
 	isb();
-
-	return idx;
 }
 
-static inline u32 armv8pmu_read_counter(struct perf_event *event)
+static inline u32 armv8pmu_read_evcntr(int idx)
+{
+	armv8pmu_select_counter(idx);
+	return read_sysreg(pmxevcntr_el0);
+}
+
+static inline u64 armv8pmu_read_hw_counter(struct perf_event *event)
+{
+	int idx = event->hw.idx;
+	u64 val = 0;
+
+	val = armv8pmu_read_evcntr(idx);
+	if (armv8pmu_event_is_chained(event))
+		val = (val << 32) | armv8pmu_read_evcntr(idx - 1);
+	return val;
+}
+
+static inline u64 armv8pmu_read_counter(struct perf_event *event)
 {
 	struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu);
 	struct hw_perf_event *hwc = &event->hw;
 	int idx = hwc->idx;
-	u32 value = 0;
+	u64 value = 0;
 
 	if (!armv8pmu_counter_valid(cpu_pmu, idx))
 		pr_err("CPU%u reading wrong counter %d\n",
 			smp_processor_id(), idx);
 	else if (idx == ARMV8_IDX_CYCLE_COUNTER)
 		value = read_sysreg(pmccntr_el0);
-	else if (armv8pmu_select_counter(idx) == idx)
-		value = read_sysreg(pmxevcntr_el0);
+	else
+		value = armv8pmu_read_hw_counter(event);
 
 	return value;
 }
 
-static inline void armv8pmu_write_counter(struct perf_event *event, u32 value)
+static inline void armv8pmu_write_evcntr(int idx, u32 value)
+{
+	armv8pmu_select_counter(idx);
+	write_sysreg(value, pmxevcntr_el0);
+}
+
+static inline void armv8pmu_write_hw_counter(struct perf_event *event,
+					     u64 value)
+{
+	int idx = event->hw.idx;
+
+	if (armv8pmu_event_is_chained(event)) {
+		armv8pmu_write_evcntr(idx, upper_32_bits(value));
+		armv8pmu_write_evcntr(idx - 1, lower_32_bits(value));
+	} else {
+		armv8pmu_write_evcntr(idx, value);
+	}
+}
+
+static inline void armv8pmu_write_counter(struct perf_event *event, u64 value)
 {
 	struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu);
 	struct hw_perf_event *hwc = &event->hw;
@@ -541,22 +597,43 @@ static inline void armv8pmu_write_counter(struct perf_event *event, u32 value)
 			smp_processor_id(), idx);
 	else if (idx == ARMV8_IDX_CYCLE_COUNTER) {
 		/*
-		 * Set the upper 32bits as this is a 64bit counter but we only
-		 * count using the lower 32bits and we want an interrupt when
-		 * it overflows.
+		 * The cycles counter is really a 64-bit counter.
+		 * When treating it as a 32-bit counter, we only count
+		 * the lower 32 bits, and set the upper 32-bits so that
+		 * we get an interrupt upon 32-bit overflow.
 		 */
-		u64 value64 = 0xffffffff00000000ULL | value;
-
-		write_sysreg(value64, pmccntr_el0);
-	} else if (armv8pmu_select_counter(idx) == idx)
-		write_sysreg(value, pmxevcntr_el0);
+		if (!armv8pmu_event_is_64bit(event))
+			value |= 0xffffffff00000000ULL;
+		write_sysreg(value, pmccntr_el0);
+	} else
+		armv8pmu_write_hw_counter(event, value);
 }
 
 static inline void armv8pmu_write_evtype(int idx, u32 val)
 {
-	if (armv8pmu_select_counter(idx) == idx) {
-		val &= ARMV8_PMU_EVTYPE_MASK;
-		write_sysreg(val, pmxevtyper_el0);
+	armv8pmu_select_counter(idx);
+	val &= ARMV8_PMU_EVTYPE_MASK;
+	write_sysreg(val, pmxevtyper_el0);
+}
+
+static inline void armv8pmu_write_event_type(struct perf_event *event)
+{
+	struct hw_perf_event *hwc = &event->hw;
+	int idx = hwc->idx;
+
+	/*
+	 * For chained events, the low counter is programmed to count
+	 * the event of interest and the high counter is programmed
+	 * with CHAIN event code with filters set to count at all ELs.
+	 */
+	if (armv8pmu_event_is_chained(event)) {
+		u32 chain_evt = ARMV8_PMUV3_PERFCTR_CHAIN |
+				ARMV8_PMU_INCLUDE_EL2;
+
+		armv8pmu_write_evtype(idx - 1, hwc->config_base);
+		armv8pmu_write_evtype(idx, chain_evt);
+	} else {
+		armv8pmu_write_evtype(idx, hwc->config_base);
 	}
 }
 
@@ -567,6 +644,16 @@ static inline int armv8pmu_enable_counter(int idx)
 	return idx;
 }
 
+static inline void armv8pmu_enable_event_counter(struct perf_event *event)
+{
+	int idx = event->hw.idx;
+
+	armv8pmu_enable_counter(idx);
+	if (armv8pmu_event_is_chained(event))
+		armv8pmu_enable_counter(idx - 1);
+	isb();
+}
+
 static inline int armv8pmu_disable_counter(int idx)
 {
 	u32 counter = ARMV8_IDX_TO_COUNTER(idx);
@@ -574,6 +661,16 @@ static inline int armv8pmu_disable_counter(int idx)
 	return idx;
 }
 
+static inline void armv8pmu_disable_event_counter(struct perf_event *event)
+{
+	struct hw_perf_event *hwc = &event->hw;
+	int idx = hwc->idx;
+
+	if (armv8pmu_event_is_chained(event))
+		armv8pmu_disable_counter(idx - 1);
+	armv8pmu_disable_counter(idx);
+}
+
 static inline int armv8pmu_enable_intens(int idx)
 {
 	u32 counter = ARMV8_IDX_TO_COUNTER(idx);
@@ -581,6 +678,11 @@ static inline int armv8pmu_enable_intens(int idx)
 	return idx;
 }
 
+static inline int armv8pmu_enable_event_irq(struct perf_event *event)
+{
+	return armv8pmu_enable_intens(event->hw.idx);
+}
+
 static inline int armv8pmu_disable_intens(int idx)
 {
 	u32 counter = ARMV8_IDX_TO_COUNTER(idx);
@@ -593,6 +695,11 @@ static inline int armv8pmu_disable_intens(int idx)
 	return idx;
 }
 
+static inline int armv8pmu_disable_event_irq(struct perf_event *event)
+{
+	return armv8pmu_disable_intens(event->hw.idx);
+}
+
 static inline u32 armv8pmu_getreset_flags(void)
 {
 	u32 value;
@@ -610,10 +717,8 @@ static inline u32 armv8pmu_getreset_flags(void)
 static void armv8pmu_enable_event(struct perf_event *event)
 {
 	unsigned long flags;
-	struct hw_perf_event *hwc = &event->hw;
 	struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu);
 	struct pmu_hw_events *events = this_cpu_ptr(cpu_pmu->hw_events);
-	int idx = hwc->idx;
 
 	/*
 	 * Enable counter and interrupt, and set the counter to count
@@ -624,22 +729,22 @@ static void armv8pmu_enable_event(struct perf_event *event)
 	/*
 	 * Disable counter
 	 */
-	armv8pmu_disable_counter(idx);
+	armv8pmu_disable_event_counter(event);
 
 	/*
 	 * Set event (if destined for PMNx counters).
 	 */
-	armv8pmu_write_evtype(idx, hwc->config_base);
+	armv8pmu_write_event_type(event);
 
 	/*
 	 * Enable interrupt for this counter
 	 */
-	armv8pmu_enable_intens(idx);
+	armv8pmu_enable_event_irq(event);
 
 	/*
 	 * Enable counter
 	 */
-	armv8pmu_enable_counter(idx);
+	armv8pmu_enable_event_counter(event);
 
 	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
@@ -647,10 +752,8 @@ static void armv8pmu_enable_event(struct perf_event *event)
 static void armv8pmu_disable_event(struct perf_event *event)
 {
 	unsigned long flags;
-	struct hw_perf_event *hwc = &event->hw;
 	struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu);
 	struct pmu_hw_events *events = this_cpu_ptr(cpu_pmu->hw_events);
-	int idx = hwc->idx;
 
 	/*
 	 * Disable counter and interrupt
@@ -660,16 +763,38 @@ static void armv8pmu_disable_event(struct perf_event *event)
 	/*
 	 * Disable counter
 	 */
-	armv8pmu_disable_counter(idx);
+	armv8pmu_disable_event_counter(event);
 
 	/*
 	 * Disable interrupt for this counter
 	 */
-	armv8pmu_disable_intens(idx);
+	armv8pmu_disable_event_irq(event);
 
 	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
 }
 
+static void armv8pmu_start(struct arm_pmu *cpu_pmu)
+{
+	unsigned long flags;
+	struct pmu_hw_events *events = this_cpu_ptr(cpu_pmu->hw_events);
+
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
+	/* Enable all counters */
+	armv8pmu_pmcr_write(armv8pmu_pmcr_read() | ARMV8_PMU_PMCR_E);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
+}
+
+static void armv8pmu_stop(struct arm_pmu *cpu_pmu)
+{
+	unsigned long flags;
+	struct pmu_hw_events *events = this_cpu_ptr(cpu_pmu->hw_events);
+
+	raw_spin_lock_irqsave(&events->pmu_lock, flags);
+	/* Disable all counters */
+	armv8pmu_pmcr_write(armv8pmu_pmcr_read() & ~ARMV8_PMU_PMCR_E);
+	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
+}
+
 static irqreturn_t armv8pmu_handle_irq(struct arm_pmu *cpu_pmu)
 {
 	u32 pmovsr;
@@ -694,6 +819,11 @@ static irqreturn_t armv8pmu_handle_irq(struct arm_pmu *cpu_pmu)
 	 */
 	regs = get_irq_regs();
 
+	/*
+	 * Stop the PMU while processing the counter overflows
+	 * to prevent skews in group events.
+	 */
+	armv8pmu_stop(cpu_pmu);
 	for (idx = 0; idx < cpu_pmu->num_events; ++idx) {
 		struct perf_event *event = cpuc->events[idx];
 		struct hw_perf_event *hwc;
@@ -718,6 +848,7 @@ static irqreturn_t armv8pmu_handle_irq(struct arm_pmu *cpu_pmu)
 		if (perf_event_overflow(event, &data, regs))
 			cpu_pmu->disable(event);
 	}
+	armv8pmu_start(cpu_pmu);
 
 	/*
 	 * Handle the pending perf events.
@@ -731,32 +862,42 @@ static irqreturn_t armv8pmu_handle_irq(struct arm_pmu *cpu_pmu)
 	return IRQ_HANDLED;
 }
 
-static void armv8pmu_start(struct arm_pmu *cpu_pmu)
+static int armv8pmu_get_single_idx(struct pmu_hw_events *cpuc,
+				    struct arm_pmu *cpu_pmu)
 {
-	unsigned long flags;
-	struct pmu_hw_events *events = this_cpu_ptr(cpu_pmu->hw_events);
+	int idx;
 
-	raw_spin_lock_irqsave(&events->pmu_lock, flags);
-	/* Enable all counters */
-	armv8pmu_pmcr_write(armv8pmu_pmcr_read() | ARMV8_PMU_PMCR_E);
-	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
+	for (idx = ARMV8_IDX_COUNTER0; idx < cpu_pmu->num_events; idx ++) {
+		if (!test_and_set_bit(idx, cpuc->used_mask))
+			return idx;
+	}
+	return -EAGAIN;
 }
 
-static void armv8pmu_stop(struct arm_pmu *cpu_pmu)
+static int armv8pmu_get_chain_idx(struct pmu_hw_events *cpuc,
+				   struct arm_pmu *cpu_pmu)
 {
-	unsigned long flags;
-	struct pmu_hw_events *events = this_cpu_ptr(cpu_pmu->hw_events);
+	int idx;
 
-	raw_spin_lock_irqsave(&events->pmu_lock, flags);
-	/* Disable all counters */
-	armv8pmu_pmcr_write(armv8pmu_pmcr_read() & ~ARMV8_PMU_PMCR_E);
-	raw_spin_unlock_irqrestore(&events->pmu_lock, flags);
+	/*
+	 * Chaining requires two consecutive event counters, where
+	 * the lower idx must be even.
+	 */
+	for (idx = ARMV8_IDX_COUNTER0 + 1; idx < cpu_pmu->num_events; idx += 2) {
+		if (!test_and_set_bit(idx, cpuc->used_mask)) {
+			/* Check if the preceding even counter is available */
+			if (!test_and_set_bit(idx - 1, cpuc->used_mask))
+				return idx;
+			/* Release the Odd counter */
+			clear_bit(idx, cpuc->used_mask);
+		}
+	}
+	return -EAGAIN;
 }
 
 static int armv8pmu_get_event_idx(struct pmu_hw_events *cpuc,
 				  struct perf_event *event)
 {
-	int idx;
 	struct arm_pmu *cpu_pmu = to_arm_pmu(event->pmu);
 	struct hw_perf_event *hwc = &event->hw;
 	unsigned long evtype = hwc->config_base & ARMV8_PMU_EVTYPE_EVENT;
@@ -770,13 +911,20 @@ static int armv8pmu_get_event_idx(struct pmu_hw_events *cpuc,
 	/*
 	 * Otherwise use events counters
 	 */
-	for (idx = ARMV8_IDX_COUNTER0; idx < cpu_pmu->num_events; ++idx) {
-		if (!test_and_set_bit(idx, cpuc->used_mask))
-			return idx;
-	}
+	if (armv8pmu_event_is_64bit(event))
+		return	armv8pmu_get_chain_idx(cpuc, cpu_pmu);
+	else
+		return armv8pmu_get_single_idx(cpuc, cpu_pmu);
+}
 
-	/* The counters are all in use. */
-	return -EAGAIN;
+static void armv8pmu_clear_event_idx(struct pmu_hw_events *cpuc,
+				     struct perf_event *event)
+{
+	int idx = event->hw.idx;
+
+	clear_bit(idx, cpuc->used_mask);
+	if (armv8pmu_event_is_chained(event))
+		clear_bit(idx - 1, cpuc->used_mask);
 }
 
 /*
@@ -851,6 +999,9 @@ static int __armv8_pmuv3_map_event(struct perf_event *event,
 				       &armv8_pmuv3_perf_cache_map,
 				       ARMV8_PMU_EVTYPE_EVENT);
 
+	if (armv8pmu_event_is_64bit(event))
+		event->hw.flags |= ARMPMU_EVT_64BIT;
+
 	/* Onl expose micro/arch events supported by this PMU */
 	if ((hw_event_id > 0) && (hw_event_id < ARMV8_PMUV3_MAX_COMMON_EVENTS)
 	    && test_bit(hw_event_id, armpmu->pmceid_bitmap)) {
@@ -957,10 +1108,10 @@ static int armv8_pmu_init(struct arm_pmu *cpu_pmu)
 	cpu_pmu->read_counter		= armv8pmu_read_counter,
 	cpu_pmu->write_counter		= armv8pmu_write_counter,
 	cpu_pmu->get_event_idx		= armv8pmu_get_event_idx,
+	cpu_pmu->clear_event_idx	= armv8pmu_clear_event_idx,
 	cpu_pmu->start			= armv8pmu_start,
 	cpu_pmu->stop			= armv8pmu_stop,
 	cpu_pmu->reset			= armv8pmu_reset,
-	cpu_pmu->max_period		= (1LLU << 32) - 1,
 	cpu_pmu->set_event_filter	= armv8pmu_set_event_filter;
 
 	return 0;
diff --git a/drivers/perf/arm-cci.c b/drivers/perf/arm-cci.c
index 0d09d8e669cd..1bfeb160c5b1 100644
--- a/drivers/perf/arm-cci.c
+++ b/drivers/perf/arm-cci.c
@@ -53,6 +53,16 @@ enum {
 	CCI_IF_MAX,
 };
 
+#define NUM_HW_CNTRS_CII_4XX	4
+#define NUM_HW_CNTRS_CII_5XX	8
+#define NUM_HW_CNTRS_MAX	NUM_HW_CNTRS_CII_5XX
+
+#define FIXED_HW_CNTRS_CII_4XX	1
+#define FIXED_HW_CNTRS_CII_5XX	0
+#define FIXED_HW_CNTRS_MAX	FIXED_HW_CNTRS_CII_4XX
+
+#define HW_CNTRS_MAX		(NUM_HW_CNTRS_MAX + FIXED_HW_CNTRS_MAX)
+
 struct event_range {
 	u32 min;
 	u32 max;
@@ -633,8 +643,7 @@ static void cci_pmu_sync_counters(struct cci_pmu *cci_pmu)
 {
 	int i;
 	struct cci_pmu_hw_events *cci_hw = &cci_pmu->hw_events;
-
-	DECLARE_BITMAP(mask, cci_pmu->num_cntrs);
+	DECLARE_BITMAP(mask, HW_CNTRS_MAX);
 
 	bitmap_zero(mask, cci_pmu->num_cntrs);
 	for_each_set_bit(i, cci_pmu->hw_events.used_mask, cci_pmu->num_cntrs) {
@@ -940,7 +949,7 @@ static void pmu_write_counters(struct cci_pmu *cci_pmu, unsigned long *mask)
 static void cci5xx_pmu_write_counters(struct cci_pmu *cci_pmu, unsigned long *mask)
 {
 	int i;
-	DECLARE_BITMAP(saved_mask, cci_pmu->num_cntrs);
+	DECLARE_BITMAP(saved_mask, HW_CNTRS_MAX);
 
 	bitmap_zero(saved_mask, cci_pmu->num_cntrs);
 	pmu_save_counters(cci_pmu, saved_mask);
@@ -1245,7 +1254,7 @@ static int validate_group(struct perf_event *event)
 {
 	struct perf_event *sibling, *leader = event->group_leader;
 	struct cci_pmu *cci_pmu = to_cci_pmu(event->pmu);
-	unsigned long mask[BITS_TO_LONGS(cci_pmu->num_cntrs)];
+	unsigned long mask[BITS_TO_LONGS(HW_CNTRS_MAX)];
 	struct cci_pmu_hw_events fake_pmu = {
 		/*
 		 * Initialise the fake PMU. We only need to populate the
@@ -1403,6 +1412,11 @@ static int cci_pmu_init(struct cci_pmu *cci_pmu, struct platform_device *pdev)
 	char *name = model->name;
 	u32 num_cntrs;
 
+	if (WARN_ON(model->num_hw_cntrs > NUM_HW_CNTRS_MAX))
+		return -EINVAL;
+	if (WARN_ON(model->fixed_hw_cntrs > FIXED_HW_CNTRS_MAX))
+		return -EINVAL;
+
 	pmu_event_attr_group.attrs = model->event_attrs;
 	pmu_format_attr_group.attrs = model->format_attrs;
 
@@ -1455,8 +1469,8 @@ static __maybe_unused struct cci_pmu_model cci_pmu_models[] = {
 #ifdef CONFIG_ARM_CCI400_PMU
 	[CCI400_R0] = {
 		.name = "CCI_400",
-		.fixed_hw_cntrs = 1,	/* Cycle counter */
-		.num_hw_cntrs = 4,
+		.fixed_hw_cntrs = FIXED_HW_CNTRS_CII_4XX, /* Cycle counter */
+		.num_hw_cntrs = NUM_HW_CNTRS_CII_4XX,
 		.cntr_size = SZ_4K,
 		.format_attrs = cci400_pmu_format_attrs,
 		.event_attrs = cci400_r0_pmu_event_attrs,
@@ -1475,8 +1489,8 @@ static __maybe_unused struct cci_pmu_model cci_pmu_models[] = {
 	},
 	[CCI400_R1] = {
 		.name = "CCI_400_r1",
-		.fixed_hw_cntrs = 1,	/* Cycle counter */
-		.num_hw_cntrs = 4,
+		.fixed_hw_cntrs = FIXED_HW_CNTRS_CII_4XX, /* Cycle counter */
+		.num_hw_cntrs = NUM_HW_CNTRS_CII_4XX,
 		.cntr_size = SZ_4K,
 		.format_attrs = cci400_pmu_format_attrs,
 		.event_attrs = cci400_r1_pmu_event_attrs,
@@ -1497,8 +1511,8 @@ static __maybe_unused struct cci_pmu_model cci_pmu_models[] = {
 #ifdef CONFIG_ARM_CCI5xx_PMU
 	[CCI500_R0] = {
 		.name = "CCI_500",
-		.fixed_hw_cntrs = 0,
-		.num_hw_cntrs = 8,
+		.fixed_hw_cntrs = FIXED_HW_CNTRS_CII_5XX,
+		.num_hw_cntrs = NUM_HW_CNTRS_CII_5XX,
 		.cntr_size = SZ_64K,
 		.format_attrs = cci5xx_pmu_format_attrs,
 		.event_attrs = cci5xx_pmu_event_attrs,
@@ -1521,8 +1535,8 @@ static __maybe_unused struct cci_pmu_model cci_pmu_models[] = {
 	},
 	[CCI550_R0] = {
 		.name = "CCI_550",
-		.fixed_hw_cntrs = 0,
-		.num_hw_cntrs = 8,
+		.fixed_hw_cntrs = FIXED_HW_CNTRS_CII_5XX,
+		.num_hw_cntrs = NUM_HW_CNTRS_CII_5XX,
 		.cntr_size = SZ_64K,
 		.format_attrs = cci5xx_pmu_format_attrs,
 		.event_attrs = cci5xx_pmu_event_attrs,
diff --git a/drivers/perf/arm-ccn.c b/drivers/perf/arm-ccn.c
index b416ee18e6bb..4b15c36f4631 100644
--- a/drivers/perf/arm-ccn.c
+++ b/drivers/perf/arm-ccn.c
@@ -1485,17 +1485,9 @@ static int arm_ccn_probe(struct platform_device *pdev)
 	platform_set_drvdata(pdev, ccn);
 
 	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
-	if (!res)
-		return -EINVAL;
-
-	if (!devm_request_mem_region(ccn->dev, res->start,
-			resource_size(res), pdev->name))
-		return -EBUSY;
-
-	ccn->base = devm_ioremap(ccn->dev, res->start,
-				resource_size(res));
-	if (!ccn->base)
-		return -EFAULT;
+	ccn->base = devm_ioremap_resource(ccn->dev, res);
+	if (IS_ERR(ccn->base))
+		return PTR_ERR(ccn->base);
 
 	res = platform_get_resource(pdev, IORESOURCE_IRQ, 0);
 	if (!res)
diff --git a/drivers/perf/arm_pmu.c b/drivers/perf/arm_pmu.c
index a6347d487635..7f01f6f60b87 100644
--- a/drivers/perf/arm_pmu.c
+++ b/drivers/perf/arm_pmu.c
@@ -28,6 +28,14 @@
 static DEFINE_PER_CPU(struct arm_pmu *, cpu_armpmu);
 static DEFINE_PER_CPU(int, cpu_irq);
 
+static inline u64 arm_pmu_event_max_period(struct perf_event *event)
+{
+	if (event->hw.flags & ARMPMU_EVT_64BIT)
+		return GENMASK_ULL(63, 0);
+	else
+		return GENMASK_ULL(31, 0);
+}
+
 static int
 armpmu_map_cache_event(const unsigned (*cache_map)
 				      [PERF_COUNT_HW_CACHE_MAX]
@@ -114,8 +122,10 @@ int armpmu_event_set_period(struct perf_event *event)
 	struct hw_perf_event *hwc = &event->hw;
 	s64 left = local64_read(&hwc->period_left);
 	s64 period = hwc->sample_period;
+	u64 max_period;
 	int ret = 0;
 
+	max_period = arm_pmu_event_max_period(event);
 	if (unlikely(left <= -period)) {
 		left = period;
 		local64_set(&hwc->period_left, left);
@@ -136,12 +146,12 @@ int armpmu_event_set_period(struct perf_event *event)
 	 * effect we are reducing max_period to account for
 	 * interrupt latency (and we are being very conservative).
 	 */
-	if (left > (armpmu->max_period >> 1))
-		left = armpmu->max_period >> 1;
+	if (left > (max_period >> 1))
+		left = (max_period >> 1);
 
 	local64_set(&hwc->prev_count, (u64)-left);
 
-	armpmu->write_counter(event, (u64)(-left) & 0xffffffff);
+	armpmu->write_counter(event, (u64)(-left) & max_period);
 
 	perf_event_update_userpage(event);
 
@@ -153,6 +163,7 @@ u64 armpmu_event_update(struct perf_event *event)
 	struct arm_pmu *armpmu = to_arm_pmu(event->pmu);
 	struct hw_perf_event *hwc = &event->hw;
 	u64 delta, prev_raw_count, new_raw_count;
+	u64 max_period = arm_pmu_event_max_period(event);
 
 again:
 	prev_raw_count = local64_read(&hwc->prev_count);
@@ -162,7 +173,7 @@ again:
 			     new_raw_count) != prev_raw_count)
 		goto again;
 
-	delta = (new_raw_count - prev_raw_count) & armpmu->max_period;
+	delta = (new_raw_count - prev_raw_count) & max_period;
 
 	local64_add(delta, &event->count);
 	local64_sub(delta, &hwc->period_left);
@@ -227,11 +238,10 @@ armpmu_del(struct perf_event *event, int flags)
 
 	armpmu_stop(event, PERF_EF_UPDATE);
 	hw_events->events[idx] = NULL;
-	clear_bit(idx, hw_events->used_mask);
-	if (armpmu->clear_event_idx)
-		armpmu->clear_event_idx(hw_events, event);
-
+	armpmu->clear_event_idx(hw_events, event);
 	perf_event_update_userpage(event);
+	/* Clear the allocated counter */
+	hwc->idx = -1;
 }
 
 static int
@@ -360,6 +370,7 @@ __hw_perf_event_init(struct perf_event *event)
 	struct hw_perf_event *hwc = &event->hw;
 	int mapping;
 
+	hwc->flags = 0;
 	mapping = armpmu->map_event(event);
 
 	if (mapping < 0) {
@@ -402,7 +413,7 @@ __hw_perf_event_init(struct perf_event *event)
 		 * is far less likely to overtake the previous one unless
 		 * you have some serious IRQ latency issues.
 		 */
-		hwc->sample_period  = armpmu->max_period >> 1;
+		hwc->sample_period  = arm_pmu_event_max_period(event) >> 1;
 		hwc->last_period    = hwc->sample_period;
 		local64_set(&hwc->period_left, hwc->sample_period);
 	}
@@ -654,14 +665,9 @@ static void cpu_pm_pmu_setup(struct arm_pmu *armpmu, unsigned long cmd)
 	int idx;
 
 	for (idx = 0; idx < armpmu->num_events; idx++) {
-		/*
-		 * If the counter is not used skip it, there is no
-		 * need of stopping/restarting it.
-		 */
-		if (!test_bit(idx, hw_events->used_mask))
-			continue;
-
 		event = hw_events->events[idx];
+		if (!event)
+			continue;
 
 		switch (cmd) {
 		case CPU_PM_ENTER:
diff --git a/drivers/perf/arm_pmu_platform.c b/drivers/perf/arm_pmu_platform.c
index 971ff336494a..96075cecb0ae 100644
--- a/drivers/perf/arm_pmu_platform.c
+++ b/drivers/perf/arm_pmu_platform.c
@@ -160,7 +160,7 @@ static int pmu_parse_irqs(struct arm_pmu *pmu)
 static int armpmu_request_irqs(struct arm_pmu *armpmu)
 {
 	struct pmu_hw_events __percpu *hw_events = armpmu->hw_events;
-	int cpu, err;
+	int cpu, err = 0;
 
 	for_each_cpu(cpu, &armpmu->supported_cpus) {
 		int irq = per_cpu(hw_events->irq, cpu);
diff --git a/drivers/perf/hisilicon/hisi_uncore_pmu.c b/drivers/perf/hisilicon/hisi_uncore_pmu.c
index 44df61397a38..9efd2413240c 100644
--- a/drivers/perf/hisilicon/hisi_uncore_pmu.c
+++ b/drivers/perf/hisilicon/hisi_uncore_pmu.c
@@ -350,19 +350,21 @@ void hisi_uncore_pmu_disable(struct pmu *pmu)
 
 /*
  * Read Super CPU cluster and CPU cluster ID from MPIDR_EL1.
- * If multi-threading is supported, SCCL_ID is in MPIDR[aff3] and CCL_ID
- * is in MPIDR[aff2]; if not, SCCL_ID is in MPIDR[aff2] and CCL_ID is
- * in MPIDR[aff1]. If this changes in future, this shall be updated.
+ * If multi-threading is supported, CCL_ID is the low 3-bits in MPIDR[Aff2]
+ * and SCCL_ID is the upper 5-bits of Aff2 field; if not, SCCL_ID
+ * is in MPIDR[Aff2] and CCL_ID is in MPIDR[Aff1].
  */
 static void hisi_read_sccl_and_ccl_id(int *sccl_id, int *ccl_id)
 {
 	u64 mpidr = read_cpuid_mpidr();
 
 	if (mpidr & MPIDR_MT_BITMASK) {
+		int aff2 = MPIDR_AFFINITY_LEVEL(mpidr, 2);
+
 		if (sccl_id)
-			*sccl_id = MPIDR_AFFINITY_LEVEL(mpidr, 3);
+			*sccl_id = aff2 >> 3;
 		if (ccl_id)
-			*ccl_id = MPIDR_AFFINITY_LEVEL(mpidr, 2);
+			*ccl_id = aff2 & 0x7;
 	} else {
 		if (sccl_id)
 			*sccl_id = MPIDR_AFFINITY_LEVEL(mpidr, 2);
diff --git a/include/linux/perf/arm_pmu.h b/include/linux/perf/arm_pmu.h
index ad5444491975..10f92e1d8e7b 100644
--- a/include/linux/perf/arm_pmu.h
+++ b/include/linux/perf/arm_pmu.h
@@ -25,6 +25,12 @@
  */
 #define ARMPMU_MAX_HWEVENTS		32
 
+/*
+ * ARM PMU hw_event flags
+ */
+/* Event uses a 64bit counter */
+#define ARMPMU_EVT_64BIT		1
+
 #define HW_OP_UNSUPPORTED		0xFFFF
 #define C(_x)				PERF_COUNT_HW_CACHE_##_x
 #define CACHE_OP_UNSUPPORTED		0xFFFF
@@ -87,14 +93,13 @@ struct arm_pmu {
 					 struct perf_event *event);
 	int		(*set_event_filter)(struct hw_perf_event *evt,
 					    struct perf_event_attr *attr);
-	u32		(*read_counter)(struct perf_event *event);
-	void		(*write_counter)(struct perf_event *event, u32 val);
+	u64		(*read_counter)(struct perf_event *event);
+	void		(*write_counter)(struct perf_event *event, u64 val);
 	void		(*start)(struct arm_pmu *);
 	void		(*stop)(struct arm_pmu *);
 	void		(*reset)(void *);
 	int		(*map_event)(struct perf_event *event);
 	int		num_events;
-	u64		max_period;
 	bool		secure_access; /* 32-bit ARM only */
 #define ARMV8_PMUV3_MAX_COMMON_EVENTS 0x40
 	DECLARE_BITMAP(pmceid_bitmap, ARMV8_PMUV3_MAX_COMMON_EVENTS);