diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c
index 87848ebe2bb7..4f7001f28936 100644
--- a/arch/x86/kernel/cpu/perf_event.c
+++ b/arch/x86/kernel/cpu/perf_event.c
@@ -190,6 +190,7 @@ static bool check_hw_exists(void)
 	u64 val, val_fail, val_new= ~0;
 	int i, reg, reg_fail, ret = 0;
 	int bios_fail = 0;
+	int reg_safe = -1;
 
 	/*
 	 * Check to see if the BIOS enabled any of the counters, if so
@@ -204,6 +205,8 @@ static bool check_hw_exists(void)
 			bios_fail = 1;
 			val_fail = val;
 			reg_fail = reg;
+		} else {
+			reg_safe = i;
 		}
 	}
 
@@ -221,12 +224,23 @@ static bool check_hw_exists(void)
 		}
 	}
 
+	/*
+	 * If all the counters are enabled, the below test will always
+	 * fail.  The tools will also become useless in this scenario.
+	 * Just fail and disable the hardware counters.
+	 */
+
+	if (reg_safe == -1) {
+		reg = reg_safe;
+		goto msr_fail;
+	}
+
 	/*
 	 * Read the current value, change it and read it back to see if it
 	 * matches, this is needed to detect certain hardware emulators
 	 * (qemu/kvm) that don't trap on the MSR access and always return 0s.
 	 */
-	reg = x86_pmu_event_addr(0);
+	reg = x86_pmu_event_addr(reg_safe);
 	if (rdmsrl_safe(reg, &val))
 		goto msr_fail;
 	val ^= 0xffffUL;
@@ -611,6 +625,7 @@ struct sched_state {
 	int	event;		/* event index */
 	int	counter;	/* counter index */
 	int	unassigned;	/* number of events to be assigned left */
+	int	nr_gp;		/* number of GP counters used */
 	unsigned long used[BITS_TO_LONGS(X86_PMC_IDX_MAX)];
 };
 
@@ -620,27 +635,29 @@ struct sched_state {
 struct perf_sched {
 	int			max_weight;
 	int			max_events;
-	struct perf_event	**events;
-	struct sched_state	state;
+	int			max_gp;
 	int			saved_states;
+	struct event_constraint	**constraints;
+	struct sched_state	state;
 	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 perf_event **events,
-			    int num, int wmin, int wmax)
+static void perf_sched_init(struct perf_sched *sched, struct event_constraint **constraints,
+			    int num, int wmin, int wmax, int gpmax)
 {
 	int idx;
 
 	memset(sched, 0, sizeof(*sched));
 	sched->max_events	= num;
 	sched->max_weight	= wmax;
-	sched->events		= events;
+	sched->max_gp		= gpmax;
+	sched->constraints	= constraints;
 
 	for (idx = 0; idx < num; idx++) {
-		if (events[idx]->hw.constraint->weight == wmin)
+		if (constraints[idx]->weight == wmin)
 			break;
 	}
 
@@ -687,7 +704,7 @@ static bool __perf_sched_find_counter(struct perf_sched *sched)
 	if (sched->state.event >= sched->max_events)
 		return false;
 
-	c = sched->events[sched->state.event]->hw.constraint;
+	c = sched->constraints[sched->state.event];
 	/* Prefer fixed purpose counters */
 	if (c->idxmsk64 & (~0ULL << INTEL_PMC_IDX_FIXED)) {
 		idx = INTEL_PMC_IDX_FIXED;
@@ -696,11 +713,16 @@ static bool __perf_sched_find_counter(struct perf_sched *sched)
 				goto done;
 		}
 	}
+
 	/* Grab the first unused counter starting with idx */
 	idx = sched->state.counter;
 	for_each_set_bit_from(idx, c->idxmsk, INTEL_PMC_IDX_FIXED) {
-		if (!__test_and_set_bit(idx, sched->state.used))
+		if (!__test_and_set_bit(idx, sched->state.used)) {
+			if (sched->state.nr_gp++ >= sched->max_gp)
+				return false;
+
 			goto done;
+		}
 	}
 
 	return false;
@@ -745,7 +767,7 @@ static bool perf_sched_next_event(struct perf_sched *sched)
 			if (sched->state.weight > sched->max_weight)
 				return false;
 		}
-		c = sched->events[sched->state.event]->hw.constraint;
+		c = sched->constraints[sched->state.event];
 	} while (c->weight != sched->state.weight);
 
 	sched->state.counter = 0;	/* start with first counter */
@@ -756,12 +778,12 @@ static bool perf_sched_next_event(struct perf_sched *sched)
 /*
  * Assign a counter for each event.
  */
-int perf_assign_events(struct perf_event **events, int n,
-			int wmin, int wmax, int *assign)
+int perf_assign_events(struct event_constraint **constraints, int n,
+			int wmin, int wmax, int gpmax, int *assign)
 {
 	struct perf_sched sched;
 
-	perf_sched_init(&sched, events, n, wmin, wmax);
+	perf_sched_init(&sched, constraints, n, wmin, wmax, gpmax);
 
 	do {
 		if (!perf_sched_find_counter(&sched))
@@ -788,9 +810,9 @@ int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
 		x86_pmu.start_scheduling(cpuc);
 
 	for (i = 0, wmin = X86_PMC_IDX_MAX, wmax = 0; i < n; i++) {
-		hwc = &cpuc->event_list[i]->hw;
+		cpuc->event_constraint[i] = NULL;
 		c = x86_pmu.get_event_constraints(cpuc, i, cpuc->event_list[i]);
-		hwc->constraint = c;
+		cpuc->event_constraint[i] = c;
 
 		wmin = min(wmin, c->weight);
 		wmax = max(wmax, c->weight);
@@ -801,7 +823,7 @@ int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
 	 */
 	for (i = 0; i < n; i++) {
 		hwc = &cpuc->event_list[i]->hw;
-		c = hwc->constraint;
+		c = cpuc->event_constraint[i];
 
 		/* never assigned */
 		if (hwc->idx == -1)
@@ -821,9 +843,26 @@ int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
 	}
 
 	/* slow path */
-	if (i != n)
-		unsched = perf_assign_events(cpuc->event_list, n, wmin,
-					     wmax, assign);
+	if (i != n) {
+		int gpmax = x86_pmu.num_counters;
+
+		/*
+		 * Do not allow scheduling of more than half the available
+		 * generic counters.
+		 *
+		 * This helps avoid counter starvation of sibling thread by
+		 * ensuring at most half the counters cannot be in exclusive
+		 * mode. There is no designated counters for the limits. Any
+		 * N/2 counters can be used. This helps with events with
+		 * specific counter constraints.
+		 */
+		if (is_ht_workaround_enabled() && !cpuc->is_fake &&
+		    READ_ONCE(cpuc->excl_cntrs->exclusive_present))
+			gpmax /= 2;
+
+		unsched = perf_assign_events(cpuc->event_constraint, n, wmin,
+					     wmax, gpmax, assign);
+	}
 
 	/*
 	 * In case of success (unsched = 0), mark events as committed,
@@ -840,7 +879,7 @@ int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign)
 			e = cpuc->event_list[i];
 			e->hw.flags |= PERF_X86_EVENT_COMMITTED;
 			if (x86_pmu.commit_scheduling)
-				x86_pmu.commit_scheduling(cpuc, e, assign[i]);
+				x86_pmu.commit_scheduling(cpuc, i, assign[i]);
 		}
 	}
 
@@ -1292,8 +1331,10 @@ static void x86_pmu_del(struct perf_event *event, int flags)
 		x86_pmu.put_event_constraints(cpuc, event);
 
 	/* Delete the array entry. */
-	while (++i < cpuc->n_events)
+	while (++i < cpuc->n_events) {
 		cpuc->event_list[i-1] = cpuc->event_list[i];
+		cpuc->event_constraint[i-1] = cpuc->event_constraint[i];
+	}
 	--cpuc->n_events;
 
 	perf_event_update_userpage(event);
diff --git a/arch/x86/kernel/cpu/perf_event.h b/arch/x86/kernel/cpu/perf_event.h
index 6ac5cb7a9e14..ef78516850fb 100644
--- a/arch/x86/kernel/cpu/perf_event.h
+++ b/arch/x86/kernel/cpu/perf_event.h
@@ -74,6 +74,7 @@ struct event_constraint {
 #define PERF_X86_EVENT_EXCL		0x0040 /* HT exclusivity on counter */
 #define PERF_X86_EVENT_DYNAMIC		0x0080 /* dynamic alloc'd constraint */
 #define PERF_X86_EVENT_RDPMC_ALLOWED	0x0100 /* grant rdpmc permission */
+#define PERF_X86_EVENT_EXCL_ACCT	0x0200 /* accounted EXCL event */
 
 
 struct amd_nb {
@@ -134,8 +135,6 @@ enum intel_excl_state_type {
 struct intel_excl_states {
 	enum intel_excl_state_type init_state[X86_PMC_IDX_MAX];
 	enum intel_excl_state_type state[X86_PMC_IDX_MAX];
-	int  num_alloc_cntrs;/* #counters allocated */
-	int  max_alloc_cntrs;/* max #counters allowed */
 	bool sched_started; /* true if scheduling has started */
 };
 
@@ -144,6 +143,11 @@ struct intel_excl_cntrs {
 
 	struct intel_excl_states states[2];
 
+	union {
+		u16	has_exclusive[2];
+		u32	exclusive_present;
+	};
+
 	int		refcnt;		/* per-core: #HT threads */
 	unsigned	core_id;	/* per-core: core id */
 };
@@ -172,7 +176,11 @@ struct cpu_hw_events {
 					     added in the current transaction */
 	int			assign[X86_PMC_IDX_MAX]; /* event to counter assignment */
 	u64			tags[X86_PMC_IDX_MAX];
+
 	struct perf_event	*event_list[X86_PMC_IDX_MAX]; /* in enabled order */
+	struct event_constraint	*event_constraint[X86_PMC_IDX_MAX];
+
+	int			n_excl; /* the number of exclusive events */
 
 	unsigned int		group_flag;
 	int			is_fake;
@@ -519,9 +527,7 @@ struct x86_pmu {
 	void		(*put_event_constraints)(struct cpu_hw_events *cpuc,
 						 struct perf_event *event);
 
-	void		(*commit_scheduling)(struct cpu_hw_events *cpuc,
-					     struct perf_event *event,
-					     int cntr);
+	void		(*commit_scheduling)(struct cpu_hw_events *cpuc, int idx, int cntr);
 
 	void		(*start_scheduling)(struct cpu_hw_events *cpuc);
 
@@ -717,8 +723,8 @@ static inline void __x86_pmu_enable_event(struct hw_perf_event *hwc,
 
 void x86_pmu_enable_all(int added);
 
-int perf_assign_events(struct perf_event **events, int n,
-			int wmin, int wmax, int *assign);
+int perf_assign_events(struct event_constraint **constraints, int n,
+			int wmin, int wmax, int gpmax, int *assign);
 int x86_schedule_events(struct cpu_hw_events *cpuc, int n, int *assign);
 
 void x86_pmu_stop(struct perf_event *event, int flags);
@@ -929,4 +935,8 @@ static inline struct intel_shared_regs *allocate_shared_regs(int cpu)
 	return NULL;
 }
 
+static inline int is_ht_workaround_enabled(void)
+{
+	return 0;
+}
 #endif /* CONFIG_CPU_SUP_INTEL */
diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 3998131d1a68..a1e35c9f06b9 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -1923,7 +1923,6 @@ intel_start_scheduling(struct cpu_hw_events *cpuc)
 	xl = &excl_cntrs->states[tid];
 
 	xl->sched_started = true;
-	xl->num_alloc_cntrs = 0;
 	/*
 	 * lock shared state until we are done scheduling
 	 * in stop_event_scheduling()
@@ -2000,6 +1999,11 @@ intel_get_excl_constraints(struct cpu_hw_events *cpuc, struct perf_event *event,
 	 * across HT threads
 	 */
 	is_excl = c->flags & PERF_X86_EVENT_EXCL;
+	if (is_excl && !(event->hw.flags & PERF_X86_EVENT_EXCL_ACCT)) {
+		event->hw.flags |= PERF_X86_EVENT_EXCL_ACCT;
+		if (!cpuc->n_excl++)
+			WRITE_ONCE(excl_cntrs->has_exclusive[tid], 1);
+	}
 
 	/*
 	 * xl = state of current HT
@@ -2008,18 +2012,6 @@ intel_get_excl_constraints(struct cpu_hw_events *cpuc, struct perf_event *event,
 	xl = &excl_cntrs->states[tid];
 	xlo = &excl_cntrs->states[o_tid];
 
-	/*
-	 * do not allow scheduling of more than max_alloc_cntrs
-	 * which is set to half the available generic counters.
-	 * this helps avoid counter starvation of sibling thread
-	 * by ensuring at most half the counters cannot be in
-	 * exclusive mode. There is not designated counters for the
-	 * limits. Any N/2 counters can be used. This helps with
-	 * events with specifix counter constraints
-	 */
-	if (xl->num_alloc_cntrs++ == xl->max_alloc_cntrs)
-		return &emptyconstraint;
-
 	cx = c;
 
 	/*
@@ -2106,7 +2098,7 @@ static struct event_constraint *
 intel_get_event_constraints(struct cpu_hw_events *cpuc, int idx,
 			    struct perf_event *event)
 {
-	struct event_constraint *c1 = event->hw.constraint;
+	struct event_constraint *c1 = cpuc->event_constraint[idx];
 	struct event_constraint *c2;
 
 	/*
@@ -2150,6 +2142,11 @@ static void intel_put_excl_constraints(struct cpu_hw_events *cpuc,
 
 	xl = &excl_cntrs->states[tid];
 	xlo = &excl_cntrs->states[o_tid];
+	if (hwc->flags & PERF_X86_EVENT_EXCL_ACCT) {
+		hwc->flags &= ~PERF_X86_EVENT_EXCL_ACCT;
+		if (!--cpuc->n_excl)
+			WRITE_ONCE(excl_cntrs->has_exclusive[tid], 0);
+	}
 
 	/*
 	 * put_constraint may be called from x86_schedule_events()
@@ -2188,8 +2185,6 @@ intel_put_shared_regs_event_constraints(struct cpu_hw_events *cpuc,
 static void intel_put_event_constraints(struct cpu_hw_events *cpuc,
 					struct perf_event *event)
 {
-	struct event_constraint *c = event->hw.constraint;
-
 	intel_put_shared_regs_event_constraints(cpuc, event);
 
 	/*
@@ -2197,19 +2192,14 @@ static void intel_put_event_constraints(struct cpu_hw_events *cpuc,
 	 * all events are subject to and must call the
 	 * put_excl_constraints() routine
 	 */
-	if (c && cpuc->excl_cntrs)
+	if (cpuc->excl_cntrs)
 		intel_put_excl_constraints(cpuc, event);
-
-	/* cleanup dynamic constraint */
-	if (c && (c->flags & PERF_X86_EVENT_DYNAMIC))
-		event->hw.constraint = NULL;
 }
 
-static void intel_commit_scheduling(struct cpu_hw_events *cpuc,
-				    struct perf_event *event, int cntr)
+static void intel_commit_scheduling(struct cpu_hw_events *cpuc, int idx, int cntr)
 {
 	struct intel_excl_cntrs *excl_cntrs = cpuc->excl_cntrs;
-	struct event_constraint *c = event->hw.constraint;
+	struct event_constraint *c = cpuc->event_constraint[idx];
 	struct intel_excl_states *xlo, *xl;
 	int tid = cpuc->excl_thread_id;
 	int o_tid = 1 - tid;
@@ -2639,8 +2629,6 @@ static void intel_pmu_cpu_starting(int cpu)
 		cpuc->lbr_sel = &cpuc->shared_regs->regs[EXTRA_REG_LBR];
 
 	if (x86_pmu.flags & PMU_FL_EXCL_CNTRS) {
-		int h = x86_pmu.num_counters >> 1;
-
 		for_each_cpu(i, topology_thread_cpumask(cpu)) {
 			struct intel_excl_cntrs *c;
 
@@ -2654,11 +2642,6 @@ static void intel_pmu_cpu_starting(int cpu)
 		}
 		cpuc->excl_cntrs->core_id = core_id;
 		cpuc->excl_cntrs->refcnt++;
-		/*
-		 * set hard limit to half the number of generic counters
-		 */
-		cpuc->excl_cntrs->states[0].max_alloc_cntrs = h;
-		cpuc->excl_cntrs->states[1].max_alloc_cntrs = h;
 	}
 }
 
diff --git a/arch/x86/kernel/cpu/perf_event_intel_ds.c b/arch/x86/kernel/cpu/perf_event_intel_ds.c
index 813f75d71175..7f73b3553e2e 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_ds.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_ds.c
@@ -706,9 +706,9 @@ void intel_pmu_pebs_disable(struct perf_event *event)
 
 	cpuc->pebs_enabled &= ~(1ULL << hwc->idx);
 
-	if (event->hw.constraint->flags & PERF_X86_EVENT_PEBS_LDLAT)
+	if (event->hw.flags & PERF_X86_EVENT_PEBS_LDLAT)
 		cpuc->pebs_enabled &= ~(1ULL << (hwc->idx + 32));
-	else if (event->hw.constraint->flags & PERF_X86_EVENT_PEBS_ST)
+	else if (event->hw.flags & PERF_X86_EVENT_PEBS_ST)
 		cpuc->pebs_enabled &= ~(1ULL << 63);
 
 	if (cpuc->enabled)
diff --git a/arch/x86/kernel/cpu/perf_event_intel_pt.c b/arch/x86/kernel/cpu/perf_event_intel_pt.c
index ffe666c2c6b5..123ff1bb2f60 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_pt.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_pt.c
@@ -151,7 +151,7 @@ static int __init pt_pmu_hw_init(void)
 
 		de_attr->attr.attr.name = pt_caps[i].name;
 
-		sysfs_attr_init(&de_attrs->attr.attr);
+		sysfs_attr_init(&de_attr->attr.attr);
 
 		de_attr->attr.attr.mode		= S_IRUGO;
 		de_attr->attr.show		= pt_cap_show;
@@ -615,7 +615,8 @@ static int pt_buffer_reset_markers(struct pt_buffer *buf,
 				   struct perf_output_handle *handle)
 
 {
-	unsigned long idx, npages, end;
+	unsigned long head = local64_read(&buf->head);
+	unsigned long idx, npages, wakeup;
 
 	if (buf->snapshot)
 		return 0;
@@ -634,17 +635,26 @@ static int pt_buffer_reset_markers(struct pt_buffer *buf,
 	buf->topa_index[buf->stop_pos]->stop = 0;
 	buf->topa_index[buf->intr_pos]->intr = 0;
 
-	if (pt_cap_get(PT_CAP_topa_multiple_entries)) {
-		npages = (handle->size + 1) >> PAGE_SHIFT;
-		end = (local64_read(&buf->head) >> PAGE_SHIFT) + npages;
-		/*if (end > handle->wakeup >> PAGE_SHIFT)
-		  end = handle->wakeup >> PAGE_SHIFT;*/
-		idx = end & (buf->nr_pages - 1);
-		buf->stop_pos = idx;
-		idx = (local64_read(&buf->head) >> PAGE_SHIFT) + npages - 1;
-		idx &= buf->nr_pages - 1;
-		buf->intr_pos = idx;
-	}
+	/* how many pages till the STOP marker */
+	npages = handle->size >> PAGE_SHIFT;
+
+	/* if it's on a page boundary, fill up one more page */
+	if (!offset_in_page(head + handle->size + 1))
+		npages++;
+
+	idx = (head >> PAGE_SHIFT) + npages;
+	idx &= buf->nr_pages - 1;
+	buf->stop_pos = idx;
+
+	wakeup = handle->wakeup >> PAGE_SHIFT;
+
+	/* in the worst case, wake up the consumer one page before hard stop */
+	idx = (head >> PAGE_SHIFT) + npages - 1;
+	if (idx > wakeup)
+		idx = wakeup;
+
+	idx &= buf->nr_pages - 1;
+	buf->intr_pos = idx;
 
 	buf->topa_index[buf->stop_pos]->stop = 1;
 	buf->topa_index[buf->intr_pos]->intr = 1;
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.c b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
index c635b8b49e93..dd319e59246b 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore.c
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.c
@@ -365,9 +365,8 @@ static int uncore_assign_events(struct intel_uncore_box *box, int assign[], int
 	bitmap_zero(used_mask, UNCORE_PMC_IDX_MAX);
 
 	for (i = 0, wmin = UNCORE_PMC_IDX_MAX, wmax = 0; i < n; i++) {
-		hwc = &box->event_list[i]->hw;
 		c = uncore_get_event_constraint(box, box->event_list[i]);
-		hwc->constraint = c;
+		box->event_constraint[i] = c;
 		wmin = min(wmin, c->weight);
 		wmax = max(wmax, c->weight);
 	}
@@ -375,7 +374,7 @@ static int uncore_assign_events(struct intel_uncore_box *box, int assign[], int
 	/* fastpath, try to reuse previous register */
 	for (i = 0; i < n; i++) {
 		hwc = &box->event_list[i]->hw;
-		c = hwc->constraint;
+		c = box->event_constraint[i];
 
 		/* never assigned */
 		if (hwc->idx == -1)
@@ -395,8 +394,8 @@ static int uncore_assign_events(struct intel_uncore_box *box, int assign[], int
 	}
 	/* slow path */
 	if (i != n)
-		ret = perf_assign_events(box->event_list, n,
-					 wmin, wmax, assign);
+		ret = perf_assign_events(box->event_constraint, n,
+					 wmin, wmax, n, assign);
 
 	if (!assign || ret) {
 		for (i = 0; i < n; i++)
diff --git a/arch/x86/kernel/cpu/perf_event_intel_uncore.h b/arch/x86/kernel/cpu/perf_event_intel_uncore.h
index 6c8c1e7e69d8..f789ec9a0133 100644
--- a/arch/x86/kernel/cpu/perf_event_intel_uncore.h
+++ b/arch/x86/kernel/cpu/perf_event_intel_uncore.h
@@ -97,6 +97,7 @@ struct intel_uncore_box {
 	atomic_t refcnt;
 	struct perf_event *events[UNCORE_PMC_IDX_MAX];
 	struct perf_event *event_list[UNCORE_PMC_IDX_MAX];
+	struct event_constraint *event_constraint[UNCORE_PMC_IDX_MAX];
 	unsigned long active_mask[BITS_TO_LONGS(UNCORE_PMC_IDX_MAX)];
 	u64 tags[UNCORE_PMC_IDX_MAX];
 	struct pci_dev *pci_dev;
diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h
index 61992cf2e977..d8a82a89f35a 100644
--- a/include/linux/perf_event.h
+++ b/include/linux/perf_event.h
@@ -92,8 +92,6 @@ struct hw_perf_event_extra {
 	int		idx;	/* index in shared_regs->regs[] */
 };
 
-struct event_constraint;
-
 /**
  * struct hw_perf_event - performance event hardware details:
  */
@@ -112,8 +110,6 @@ struct hw_perf_event {
 
 			struct hw_perf_event_extra extra_reg;
 			struct hw_perf_event_extra branch_reg;
-
-			struct event_constraint *constraint;
 		};
 		struct { /* software */
 			struct hrtimer	hrtimer;
diff --git a/kernel/events/core.c b/kernel/events/core.c
index 1a3bf48743ce..eddf1ed4155e 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -3442,7 +3442,6 @@ static void free_event_rcu(struct rcu_head *head)
 	if (event->ns)
 		put_pid_ns(event->ns);
 	perf_event_free_filter(event);
-	perf_event_free_bpf_prog(event);
 	kfree(event);
 }
 
@@ -3573,6 +3572,8 @@ static void __free_event(struct perf_event *event)
 			put_callchain_buffers();
 	}
 
+	perf_event_free_bpf_prog(event);
+
 	if (event->destroy)
 		event->destroy(event);
 
diff --git a/kernel/events/ring_buffer.c b/kernel/events/ring_buffer.c
index 232f00f273cb..725c416085e3 100644
--- a/kernel/events/ring_buffer.c
+++ b/kernel/events/ring_buffer.c
@@ -493,6 +493,20 @@ int rb_alloc_aux(struct ring_buffer *rb, struct perf_event *event,
 			rb->aux_pages[rb->aux_nr_pages] = page_address(page++);
 	}
 
+	/*
+	 * In overwrite mode, PMUs that don't support SG may not handle more
+	 * than one contiguous allocation, since they rely on PMI to do double
+	 * buffering. In this case, the entire buffer has to be one contiguous
+	 * chunk.
+	 */
+	if ((event->pmu->capabilities & PERF_PMU_CAP_AUX_NO_SG) &&
+	    overwrite) {
+		struct page *page = virt_to_page(rb->aux_pages[0]);
+
+		if (page_private(page) != max_order)
+			goto out;
+	}
+
 	rb->aux_priv = event->pmu->setup_aux(event->cpu, rb->aux_pages, nr_pages,
 					     overwrite);
 	if (!rb->aux_priv)