diff --git a/tools/include/linux/build_bug.h b/tools/include/linux/build_bug.h
index ce365d2..cc7070c 100644
--- a/tools/include/linux/build_bug.h
+++ b/tools/include/linux/build_bug.h
@@ -79,9 +79,4 @@
 #define __static_assert(expr, msg, ...) _Static_assert(expr, msg)
 #endif // static_assert
 
-#ifdef __GENKSYMS__
-/* genksyms gets confused by _Static_assert */
-#define _Static_assert(expr, ...)
-#endif
-
 #endif	/* _LINUX_BUILD_BUG_H */
diff --git a/tools/include/uapi/linux/kvm.h b/tools/include/uapi/linux/kvm.h
index 886802b..374c678 100644
--- a/tools/include/uapi/linux/kvm.h
+++ b/tools/include/uapi/linux/kvm.h
@@ -251,6 +251,7 @@
 #define KVM_EXIT_X86_RDMSR        29
 #define KVM_EXIT_X86_WRMSR        30
 #define KVM_EXIT_DIRTY_RING_FULL  31
+#define KVM_EXIT_AP_RESET_HOLD    32
 
 /* For KVM_EXIT_INTERNAL_ERROR */
 /* Emulate instruction failed. */
@@ -573,6 +574,7 @@
 #define KVM_MP_STATE_CHECK_STOP        6
 #define KVM_MP_STATE_OPERATING         7
 #define KVM_MP_STATE_LOAD              8
+#define KVM_MP_STATE_AP_RESET_HOLD     9
 
 struct kvm_mp_state {
 	__u32 mp_state;
diff --git a/tools/lib/perf/tests/test-cpumap.c b/tools/lib/perf/tests/test-cpumap.c
index c8d4509..c70e9e0 100644
--- a/tools/lib/perf/tests/test-cpumap.c
+++ b/tools/lib/perf/tests/test-cpumap.c
@@ -27,5 +27,5 @@
 	perf_cpu_map__put(cpus);
 
 	__T_END;
-	return 0;
+	return tests_failed == 0 ? 0 : -1;
 }
diff --git a/tools/lib/perf/tests/test-evlist.c b/tools/lib/perf/tests/test-evlist.c
index 6d8ebe0..e2ac0b7 100644
--- a/tools/lib/perf/tests/test-evlist.c
+++ b/tools/lib/perf/tests/test-evlist.c
@@ -208,13 +208,13 @@
 	char path[PATH_MAX];
 	int id, err, pid, go_pipe[2];
 	union perf_event *event;
-	char bf;
 	int count = 0;
 
 	snprintf(path, PATH_MAX, "%s/kernel/debug/tracing/events/syscalls/sys_enter_prctl/id",
 		 sysfs__mountpoint());
 
 	if (filename__read_int(path, &id)) {
+		tests_failed++;
 		fprintf(stderr, "error: failed to get tracepoint id: %s\n", path);
 		return -1;
 	}
@@ -229,6 +229,7 @@
 	pid = fork();
 	if (!pid) {
 		int i;
+		char bf;
 
 		read(go_pipe[0], &bf, 1);
 
@@ -266,7 +267,7 @@
 	perf_evlist__enable(evlist);
 
 	/* kick the child and wait for it to finish */
-	write(go_pipe[1], &bf, 1);
+	write(go_pipe[1], "A", 1);
 	waitpid(pid, NULL, 0);
 
 	/*
@@ -409,5 +410,5 @@
 	test_mmap_cpus();
 
 	__T_END;
-	return 0;
+	return tests_failed == 0 ? 0 : -1;
 }
diff --git a/tools/lib/perf/tests/test-evsel.c b/tools/lib/perf/tests/test-evsel.c
index 135722a..0ad82d7 100644
--- a/tools/lib/perf/tests/test-evsel.c
+++ b/tools/lib/perf/tests/test-evsel.c
@@ -131,5 +131,5 @@
 	test_stat_thread_enable();
 
 	__T_END;
-	return 0;
+	return tests_failed == 0 ? 0 : -1;
 }
diff --git a/tools/lib/perf/tests/test-threadmap.c b/tools/lib/perf/tests/test-threadmap.c
index 7dc4d6f..3844714 100644
--- a/tools/lib/perf/tests/test-threadmap.c
+++ b/tools/lib/perf/tests/test-threadmap.c
@@ -27,5 +27,5 @@
 	perf_thread_map__put(threads);
 
 	__T_END;
-	return 0;
+	return tests_failed == 0 ? 0 : -1;
 }
diff --git a/tools/perf/examples/bpf/5sec.c b/tools/perf/examples/bpf/5sec.c
index 65c4ff6..e6b6181 100644
--- a/tools/perf/examples/bpf/5sec.c
+++ b/tools/perf/examples/bpf/5sec.c
@@ -39,7 +39,7 @@
    Copyright (C) 2018 Red Hat, Inc., Arnaldo Carvalho de Melo <acme@redhat.com>
 */
 
-#include <bpf/bpf.h>
+#include <bpf.h>
 
 #define NSEC_PER_SEC	1000000000L
 
diff --git a/tools/perf/tests/shell/stat+shadow_stat.sh b/tools/perf/tests/shell/stat+shadow_stat.sh
index 249dfe4..ebebd35 100755
--- a/tools/perf/tests/shell/stat+shadow_stat.sh
+++ b/tools/perf/tests/shell/stat+shadow_stat.sh
@@ -9,31 +9,29 @@
 
 test_global_aggr()
 {
-	local cyc
-
 	perf stat -a --no-big-num -e cycles,instructions sleep 1  2>&1 | \
 	grep -e cycles -e instructions | \
 	while read num evt hash ipc rest
 	do
 		# skip not counted events
-		if [[ $num == "<not" ]]; then
+		if [ "$num" = "<not" ]; then
 			continue
 		fi
 
 		# save cycles count
-		if [[ $evt == "cycles" ]]; then
+		if [ "$evt" = "cycles" ]; then
 			cyc=$num
 			continue
 		fi
 
 		# skip if no cycles
-		if [[ -z $cyc ]]; then
+		if [ -z "$cyc" ]; then
 			continue
 		fi
 
 		# use printf for rounding and a leading zero
-		local res=`printf "%.2f" $(echo "scale=6; $num / $cyc" | bc -q)`
-		if [[ $ipc != $res ]]; then
+		res=`printf "%.2f" $(echo "scale=6; $num / $cyc" | bc -q)`
+		if [ "$ipc" != "$res" ]; then
 			echo "IPC is different: $res != $ipc  ($num / $cyc)"
 			exit 1
 		fi
@@ -42,32 +40,32 @@
 
 test_no_aggr()
 {
-	declare -A results
-
 	perf stat -a -A --no-big-num -e cycles,instructions sleep 1  2>&1 | \
 	grep ^CPU | \
 	while read cpu num evt hash ipc rest
 	do
 		# skip not counted events
-		if [[ $num == "<not" ]]; then
+		if [ "$num" = "<not" ]; then
 			continue
 		fi
 
 		# save cycles count
-		if [[ $evt == "cycles" ]]; then
-			results[$cpu]=$num
+		if [ "$evt" = "cycles" ]; then
+			results="$results $cpu:$num"
 			continue
 		fi
 
+		cyc=${results##* $cpu:}
+		cyc=${cyc%% *}
+
 		# skip if no cycles
-		local cyc=${results[$cpu]}
-		if [[ -z $cyc ]]; then
+		if [ -z "$cyc" ]; then
 			continue
 		fi
 
 		# use printf for rounding and a leading zero
-		local res=`printf "%.2f" $(echo "scale=6; $num / $cyc" | bc -q)`
-		if [[ $ipc != $res ]]; then
+		res=`printf "%.2f" $(echo "scale=6; $num / $cyc" | bc -q)`
+		if [ "$ipc" != "$res" ]; then
 			echo "IPC is different for $cpu: $res != $ipc  ($num / $cyc)"
 			exit 1
 		fi
diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
index 062383e..c4ed3dc 100644
--- a/tools/perf/util/header.c
+++ b/tools/perf/util/header.c
@@ -3323,6 +3323,14 @@
 	attr_offset = lseek(ff.fd, 0, SEEK_CUR);
 
 	evlist__for_each_entry(evlist, evsel) {
+		if (evsel->core.attr.size < sizeof(evsel->core.attr)) {
+			/*
+			 * We are likely in "perf inject" and have read
+			 * from an older file. Update attr size so that
+			 * reader gets the right offset to the ids.
+			 */
+			evsel->core.attr.size = sizeof(evsel->core.attr);
+		}
 		f_attr = (struct perf_file_attr){
 			.attr = evsel->core.attr,
 			.ids  = {
diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c
index f841f35..1e9d3f9 100644
--- a/tools/perf/util/machine.c
+++ b/tools/perf/util/machine.c
@@ -2980,7 +2980,7 @@
 
 pid_t machine__get_current_tid(struct machine *machine, int cpu)
 {
-	int nr_cpus = min(machine->env->nr_cpus_online, MAX_NR_CPUS);
+	int nr_cpus = min(machine->env->nr_cpus_avail, MAX_NR_CPUS);
 
 	if (cpu < 0 || cpu >= nr_cpus || !machine->current_tid)
 		return -1;
@@ -2992,7 +2992,7 @@
 			     pid_t tid)
 {
 	struct thread *thread;
-	int nr_cpus = min(machine->env->nr_cpus_online, MAX_NR_CPUS);
+	int nr_cpus = min(machine->env->nr_cpus_avail, MAX_NR_CPUS);
 
 	if (cpu < 0)
 		return -EINVAL;
diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
index 50ff979..25adbcc 100644
--- a/tools/perf/util/session.c
+++ b/tools/perf/util/session.c
@@ -2404,7 +2404,7 @@
 {
 	int i, err = -1;
 	struct perf_cpu_map *map;
-	int nr_cpus = min(session->header.env.nr_cpus_online, MAX_NR_CPUS);
+	int nr_cpus = min(session->header.env.nr_cpus_avail, MAX_NR_CPUS);
 
 	for (i = 0; i < PERF_TYPE_MAX; ++i) {
 		struct evsel *evsel;
diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat-shadow.c
index 9012651..12eafd1 100644
--- a/tools/perf/util/stat-shadow.c
+++ b/tools/perf/util/stat-shadow.c
@@ -8,6 +8,7 @@
 #include "evlist.h"
 #include "expr.h"
 #include "metricgroup.h"
+#include "cgroup.h"
 #include <linux/zalloc.h>
 
 /*
@@ -28,6 +29,7 @@
 	enum stat_type type;
 	int ctx;
 	int cpu;
+	struct cgroup *cgrp;
 	struct runtime_stat *stat;
 	struct stats stats;
 	u64 metric_total;
@@ -57,6 +59,9 @@
 	if (a->ctx != b->ctx)
 		return a->ctx - b->ctx;
 
+	if (a->cgrp != b->cgrp)
+		return (char *)a->cgrp < (char *)b->cgrp ? -1 : +1;
+
 	if (a->evsel == NULL && b->evsel == NULL) {
 		if (a->stat == b->stat)
 			return 0;
@@ -100,7 +105,8 @@
 					      bool create,
 					      enum stat_type type,
 					      int ctx,
-					      struct runtime_stat *st)
+					      struct runtime_stat *st,
+					      struct cgroup *cgrp)
 {
 	struct rblist *rblist;
 	struct rb_node *nd;
@@ -110,10 +116,15 @@
 		.type = type,
 		.ctx = ctx,
 		.stat = st,
+		.cgrp = cgrp,
 	};
 
 	rblist = &st->value_list;
 
+	/* don't use context info for clock events */
+	if (type == STAT_NSECS)
+		dm.ctx = 0;
+
 	nd = rblist__find(rblist, &dm);
 	if (nd)
 		return container_of(nd, struct saved_value, rb_node);
@@ -191,12 +202,18 @@
 	reset_stat(st);
 }
 
+struct runtime_stat_data {
+	int ctx;
+	struct cgroup *cgrp;
+};
+
 static void update_runtime_stat(struct runtime_stat *st,
 				enum stat_type type,
-				int ctx, int cpu, u64 count)
+				int cpu, u64 count,
+				struct runtime_stat_data *rsd)
 {
-	struct saved_value *v = saved_value_lookup(NULL, cpu, true,
-						   type, ctx, st);
+	struct saved_value *v = saved_value_lookup(NULL, cpu, true, type,
+						   rsd->ctx, st, rsd->cgrp);
 
 	if (v)
 		update_stats(&v->stats, count);
@@ -210,82 +227,86 @@
 void perf_stat__update_shadow_stats(struct evsel *counter, u64 count,
 				    int cpu, struct runtime_stat *st)
 {
-	int ctx = evsel_context(counter);
 	u64 count_ns = count;
 	struct saved_value *v;
+	struct runtime_stat_data rsd = {
+		.ctx = evsel_context(counter),
+		.cgrp = counter->cgrp,
+	};
 
 	count *= counter->scale;
 
 	if (evsel__is_clock(counter))
-		update_runtime_stat(st, STAT_NSECS, 0, cpu, count_ns);
+		update_runtime_stat(st, STAT_NSECS, cpu, count_ns, &rsd);
 	else if (evsel__match(counter, HARDWARE, HW_CPU_CYCLES))
-		update_runtime_stat(st, STAT_CYCLES, ctx, cpu, count);
+		update_runtime_stat(st, STAT_CYCLES, cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, CYCLES_IN_TX))
-		update_runtime_stat(st, STAT_CYCLES_IN_TX, ctx, cpu, count);
+		update_runtime_stat(st, STAT_CYCLES_IN_TX, cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, TRANSACTION_START))
-		update_runtime_stat(st, STAT_TRANSACTION, ctx, cpu, count);
+		update_runtime_stat(st, STAT_TRANSACTION, cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, ELISION_START))
-		update_runtime_stat(st, STAT_ELISION, ctx, cpu, count);
+		update_runtime_stat(st, STAT_ELISION, cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, TOPDOWN_TOTAL_SLOTS))
 		update_runtime_stat(st, STAT_TOPDOWN_TOTAL_SLOTS,
-				    ctx, cpu, count);
+				    cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, TOPDOWN_SLOTS_ISSUED))
 		update_runtime_stat(st, STAT_TOPDOWN_SLOTS_ISSUED,
-				    ctx, cpu, count);
+				    cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, TOPDOWN_SLOTS_RETIRED))
 		update_runtime_stat(st, STAT_TOPDOWN_SLOTS_RETIRED,
-				    ctx, cpu, count);
+				    cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, TOPDOWN_FETCH_BUBBLES))
 		update_runtime_stat(st, STAT_TOPDOWN_FETCH_BUBBLES,
-				    ctx, cpu, count);
+				    cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, TOPDOWN_RECOVERY_BUBBLES))
 		update_runtime_stat(st, STAT_TOPDOWN_RECOVERY_BUBBLES,
-				    ctx, cpu, count);
+				    cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, TOPDOWN_RETIRING))
 		update_runtime_stat(st, STAT_TOPDOWN_RETIRING,
-				    ctx, cpu, count);
+				    cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, TOPDOWN_BAD_SPEC))
 		update_runtime_stat(st, STAT_TOPDOWN_BAD_SPEC,
-				    ctx, cpu, count);
+				    cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, TOPDOWN_FE_BOUND))
 		update_runtime_stat(st, STAT_TOPDOWN_FE_BOUND,
-				    ctx, cpu, count);
+				    cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, TOPDOWN_BE_BOUND))
 		update_runtime_stat(st, STAT_TOPDOWN_BE_BOUND,
-				    ctx, cpu, count);
+				    cpu, count, &rsd);
 	else if (evsel__match(counter, HARDWARE, HW_STALLED_CYCLES_FRONTEND))
 		update_runtime_stat(st, STAT_STALLED_CYCLES_FRONT,
-				    ctx, cpu, count);
+				    cpu, count, &rsd);
 	else if (evsel__match(counter, HARDWARE, HW_STALLED_CYCLES_BACKEND))
 		update_runtime_stat(st, STAT_STALLED_CYCLES_BACK,
-				    ctx, cpu, count);
+				    cpu, count, &rsd);
 	else if (evsel__match(counter, HARDWARE, HW_BRANCH_INSTRUCTIONS))
-		update_runtime_stat(st, STAT_BRANCHES, ctx, cpu, count);
+		update_runtime_stat(st, STAT_BRANCHES, cpu, count, &rsd);
 	else if (evsel__match(counter, HARDWARE, HW_CACHE_REFERENCES))
-		update_runtime_stat(st, STAT_CACHEREFS, ctx, cpu, count);
+		update_runtime_stat(st, STAT_CACHEREFS, cpu, count, &rsd);
 	else if (evsel__match(counter, HW_CACHE, HW_CACHE_L1D))
-		update_runtime_stat(st, STAT_L1_DCACHE, ctx, cpu, count);
+		update_runtime_stat(st, STAT_L1_DCACHE, cpu, count, &rsd);
 	else if (evsel__match(counter, HW_CACHE, HW_CACHE_L1I))
-		update_runtime_stat(st, STAT_L1_ICACHE, ctx, cpu, count);
+		update_runtime_stat(st, STAT_L1_ICACHE, cpu, count, &rsd);
 	else if (evsel__match(counter, HW_CACHE, HW_CACHE_LL))
-		update_runtime_stat(st, STAT_LL_CACHE, ctx, cpu, count);
+		update_runtime_stat(st, STAT_LL_CACHE, cpu, count, &rsd);
 	else if (evsel__match(counter, HW_CACHE, HW_CACHE_DTLB))
-		update_runtime_stat(st, STAT_DTLB_CACHE, ctx, cpu, count);
+		update_runtime_stat(st, STAT_DTLB_CACHE, cpu, count, &rsd);
 	else if (evsel__match(counter, HW_CACHE, HW_CACHE_ITLB))
-		update_runtime_stat(st, STAT_ITLB_CACHE, ctx, cpu, count);
+		update_runtime_stat(st, STAT_ITLB_CACHE, cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, SMI_NUM))
-		update_runtime_stat(st, STAT_SMI_NUM, ctx, cpu, count);
+		update_runtime_stat(st, STAT_SMI_NUM, cpu, count, &rsd);
 	else if (perf_stat_evsel__is(counter, APERF))
-		update_runtime_stat(st, STAT_APERF, ctx, cpu, count);
+		update_runtime_stat(st, STAT_APERF, cpu, count, &rsd);
 
 	if (counter->collect_stat) {
-		v = saved_value_lookup(counter, cpu, true, STAT_NONE, 0, st);
+		v = saved_value_lookup(counter, cpu, true, STAT_NONE, 0, st,
+				       rsd.cgrp);
 		update_stats(&v->stats, count);
 		if (counter->metric_leader)
 			v->metric_total += count;
 	} else if (counter->metric_leader) {
 		v = saved_value_lookup(counter->metric_leader,
-				       cpu, true, STAT_NONE, 0, st);
+				       cpu, true, STAT_NONE, 0, st, rsd.cgrp);
 		v->metric_total += count;
 		v->metric_other++;
 	}
@@ -422,11 +443,12 @@
 }
 
 static double runtime_stat_avg(struct runtime_stat *st,
-			       enum stat_type type, int ctx, int cpu)
+			       enum stat_type type, int cpu,
+			       struct runtime_stat_data *rsd)
 {
 	struct saved_value *v;
 
-	v = saved_value_lookup(NULL, cpu, false, type, ctx, st);
+	v = saved_value_lookup(NULL, cpu, false, type, rsd->ctx, st, rsd->cgrp);
 	if (!v)
 		return 0.0;
 
@@ -434,11 +456,12 @@
 }
 
 static double runtime_stat_n(struct runtime_stat *st,
-			     enum stat_type type, int ctx, int cpu)
+			     enum stat_type type, int cpu,
+			     struct runtime_stat_data *rsd)
 {
 	struct saved_value *v;
 
-	v = saved_value_lookup(NULL, cpu, false, type, ctx, st);
+	v = saved_value_lookup(NULL, cpu, false, type, rsd->ctx, st, rsd->cgrp);
 	if (!v)
 		return 0.0;
 
@@ -446,16 +469,15 @@
 }
 
 static void print_stalled_cycles_frontend(struct perf_stat_config *config,
-					  int cpu,
-					  struct evsel *evsel, double avg,
+					  int cpu, double avg,
 					  struct perf_stat_output_ctx *out,
-					  struct runtime_stat *st)
+					  struct runtime_stat *st,
+					  struct runtime_stat_data *rsd)
 {
 	double total, ratio = 0.0;
 	const char *color;
-	int ctx = evsel_context(evsel);
 
-	total = runtime_stat_avg(st, STAT_CYCLES, ctx, cpu);
+	total = runtime_stat_avg(st, STAT_CYCLES, cpu, rsd);
 
 	if (total)
 		ratio = avg / total * 100.0;
@@ -470,16 +492,15 @@
 }
 
 static void print_stalled_cycles_backend(struct perf_stat_config *config,
-					 int cpu,
-					 struct evsel *evsel, double avg,
+					 int cpu, double avg,
 					 struct perf_stat_output_ctx *out,
-					 struct runtime_stat *st)
+					 struct runtime_stat *st,
+					 struct runtime_stat_data *rsd)
 {
 	double total, ratio = 0.0;
 	const char *color;
-	int ctx = evsel_context(evsel);
 
-	total = runtime_stat_avg(st, STAT_CYCLES, ctx, cpu);
+	total = runtime_stat_avg(st, STAT_CYCLES, cpu, rsd);
 
 	if (total)
 		ratio = avg / total * 100.0;
@@ -490,17 +511,15 @@
 }
 
 static void print_branch_misses(struct perf_stat_config *config,
-				int cpu,
-				struct evsel *evsel,
-				double avg,
+				int cpu, double avg,
 				struct perf_stat_output_ctx *out,
-				struct runtime_stat *st)
+				struct runtime_stat *st,
+				struct runtime_stat_data *rsd)
 {
 	double total, ratio = 0.0;
 	const char *color;
-	int ctx = evsel_context(evsel);
 
-	total = runtime_stat_avg(st, STAT_BRANCHES, ctx, cpu);
+	total = runtime_stat_avg(st, STAT_BRANCHES, cpu, rsd);
 
 	if (total)
 		ratio = avg / total * 100.0;
@@ -511,18 +530,15 @@
 }
 
 static void print_l1_dcache_misses(struct perf_stat_config *config,
-				   int cpu,
-				   struct evsel *evsel,
-				   double avg,
+				   int cpu, double avg,
 				   struct perf_stat_output_ctx *out,
-				   struct runtime_stat *st)
-
+				   struct runtime_stat *st,
+				   struct runtime_stat_data *rsd)
 {
 	double total, ratio = 0.0;
 	const char *color;
-	int ctx = evsel_context(evsel);
 
-	total = runtime_stat_avg(st, STAT_L1_DCACHE, ctx, cpu);
+	total = runtime_stat_avg(st, STAT_L1_DCACHE, cpu, rsd);
 
 	if (total)
 		ratio = avg / total * 100.0;
@@ -533,18 +549,15 @@
 }
 
 static void print_l1_icache_misses(struct perf_stat_config *config,
-				   int cpu,
-				   struct evsel *evsel,
-				   double avg,
+				   int cpu, double avg,
 				   struct perf_stat_output_ctx *out,
-				   struct runtime_stat *st)
-
+				   struct runtime_stat *st,
+				   struct runtime_stat_data *rsd)
 {
 	double total, ratio = 0.0;
 	const char *color;
-	int ctx = evsel_context(evsel);
 
-	total = runtime_stat_avg(st, STAT_L1_ICACHE, ctx, cpu);
+	total = runtime_stat_avg(st, STAT_L1_ICACHE, cpu, rsd);
 
 	if (total)
 		ratio = avg / total * 100.0;
@@ -554,17 +567,15 @@
 }
 
 static void print_dtlb_cache_misses(struct perf_stat_config *config,
-				    int cpu,
-				    struct evsel *evsel,
-				    double avg,
+				    int cpu, double avg,
 				    struct perf_stat_output_ctx *out,
-				    struct runtime_stat *st)
+				    struct runtime_stat *st,
+				    struct runtime_stat_data *rsd)
 {
 	double total, ratio = 0.0;
 	const char *color;
-	int ctx = evsel_context(evsel);
 
-	total = runtime_stat_avg(st, STAT_DTLB_CACHE, ctx, cpu);
+	total = runtime_stat_avg(st, STAT_DTLB_CACHE, cpu, rsd);
 
 	if (total)
 		ratio = avg / total * 100.0;
@@ -574,17 +585,15 @@
 }
 
 static void print_itlb_cache_misses(struct perf_stat_config *config,
-				    int cpu,
-				    struct evsel *evsel,
-				    double avg,
+				    int cpu, double avg,
 				    struct perf_stat_output_ctx *out,
-				    struct runtime_stat *st)
+				    struct runtime_stat *st,
+				    struct runtime_stat_data *rsd)
 {
 	double total, ratio = 0.0;
 	const char *color;
-	int ctx = evsel_context(evsel);
 
-	total = runtime_stat_avg(st, STAT_ITLB_CACHE, ctx, cpu);
+	total = runtime_stat_avg(st, STAT_ITLB_CACHE, cpu, rsd);
 
 	if (total)
 		ratio = avg / total * 100.0;
@@ -594,17 +603,15 @@
 }
 
 static void print_ll_cache_misses(struct perf_stat_config *config,
-				  int cpu,
-				  struct evsel *evsel,
-				  double avg,
+				  int cpu, double avg,
 				  struct perf_stat_output_ctx *out,
-				  struct runtime_stat *st)
+				  struct runtime_stat *st,
+				  struct runtime_stat_data *rsd)
 {
 	double total, ratio = 0.0;
 	const char *color;
-	int ctx = evsel_context(evsel);
 
-	total = runtime_stat_avg(st, STAT_LL_CACHE, ctx, cpu);
+	total = runtime_stat_avg(st, STAT_LL_CACHE, cpu, rsd);
 
 	if (total)
 		ratio = avg / total * 100.0;
@@ -662,56 +669,61 @@
 	return x;
 }
 
-static double td_total_slots(int ctx, int cpu, struct runtime_stat *st)
+static double td_total_slots(int cpu, struct runtime_stat *st,
+			     struct runtime_stat_data *rsd)
 {
-	return runtime_stat_avg(st, STAT_TOPDOWN_TOTAL_SLOTS, ctx, cpu);
+	return runtime_stat_avg(st, STAT_TOPDOWN_TOTAL_SLOTS, cpu, rsd);
 }
 
-static double td_bad_spec(int ctx, int cpu, struct runtime_stat *st)
+static double td_bad_spec(int cpu, struct runtime_stat *st,
+			  struct runtime_stat_data *rsd)
 {
 	double bad_spec = 0;
 	double total_slots;
 	double total;
 
-	total = runtime_stat_avg(st, STAT_TOPDOWN_SLOTS_ISSUED, ctx, cpu) -
-		runtime_stat_avg(st, STAT_TOPDOWN_SLOTS_RETIRED, ctx, cpu) +
-		runtime_stat_avg(st, STAT_TOPDOWN_RECOVERY_BUBBLES, ctx, cpu);
+	total = runtime_stat_avg(st, STAT_TOPDOWN_SLOTS_ISSUED, cpu, rsd) -
+		runtime_stat_avg(st, STAT_TOPDOWN_SLOTS_RETIRED, cpu, rsd) +
+		runtime_stat_avg(st, STAT_TOPDOWN_RECOVERY_BUBBLES, cpu, rsd);
 
-	total_slots = td_total_slots(ctx, cpu, st);
+	total_slots = td_total_slots(cpu, st, rsd);
 	if (total_slots)
 		bad_spec = total / total_slots;
 	return sanitize_val(bad_spec);
 }
 
-static double td_retiring(int ctx, int cpu, struct runtime_stat *st)
+static double td_retiring(int cpu, struct runtime_stat *st,
+			  struct runtime_stat_data *rsd)
 {
 	double retiring = 0;
-	double total_slots = td_total_slots(ctx, cpu, st);
+	double total_slots = td_total_slots(cpu, st, rsd);
 	double ret_slots = runtime_stat_avg(st, STAT_TOPDOWN_SLOTS_RETIRED,
-					    ctx, cpu);
+					    cpu, rsd);
 
 	if (total_slots)
 		retiring = ret_slots / total_slots;
 	return retiring;
 }
 
-static double td_fe_bound(int ctx, int cpu, struct runtime_stat *st)
+static double td_fe_bound(int cpu, struct runtime_stat *st,
+			  struct runtime_stat_data *rsd)
 {
 	double fe_bound = 0;
-	double total_slots = td_total_slots(ctx, cpu, st);
+	double total_slots = td_total_slots(cpu, st, rsd);
 	double fetch_bub = runtime_stat_avg(st, STAT_TOPDOWN_FETCH_BUBBLES,
-					    ctx, cpu);
+					    cpu, rsd);
 
 	if (total_slots)
 		fe_bound = fetch_bub / total_slots;
 	return fe_bound;
 }
 
-static double td_be_bound(int ctx, int cpu, struct runtime_stat *st)
+static double td_be_bound(int cpu, struct runtime_stat *st,
+			  struct runtime_stat_data *rsd)
 {
-	double sum = (td_fe_bound(ctx, cpu, st) +
-		      td_bad_spec(ctx, cpu, st) +
-		      td_retiring(ctx, cpu, st));
+	double sum = (td_fe_bound(cpu, st, rsd) +
+		      td_bad_spec(cpu, st, rsd) +
+		      td_retiring(cpu, st, rsd));
 	if (sum == 0)
 		return 0;
 	return sanitize_val(1.0 - sum);
@@ -722,15 +734,15 @@
  * the ratios we need to recreate the sum.
  */
 
-static double td_metric_ratio(int ctx, int cpu,
-			      enum stat_type type,
-			      struct runtime_stat *stat)
+static double td_metric_ratio(int cpu, enum stat_type type,
+			      struct runtime_stat *stat,
+			      struct runtime_stat_data *rsd)
 {
-	double sum = runtime_stat_avg(stat, STAT_TOPDOWN_RETIRING, ctx, cpu) +
-		runtime_stat_avg(stat, STAT_TOPDOWN_FE_BOUND, ctx, cpu) +
-		runtime_stat_avg(stat, STAT_TOPDOWN_BE_BOUND, ctx, cpu) +
-		runtime_stat_avg(stat, STAT_TOPDOWN_BAD_SPEC, ctx, cpu);
-	double d = runtime_stat_avg(stat, type, ctx, cpu);
+	double sum = runtime_stat_avg(stat, STAT_TOPDOWN_RETIRING, cpu, rsd) +
+		runtime_stat_avg(stat, STAT_TOPDOWN_FE_BOUND, cpu, rsd) +
+		runtime_stat_avg(stat, STAT_TOPDOWN_BE_BOUND, cpu, rsd) +
+		runtime_stat_avg(stat, STAT_TOPDOWN_BAD_SPEC, cpu, rsd);
+	double d = runtime_stat_avg(stat, type, cpu, rsd);
 
 	if (sum)
 		return d / sum;
@@ -742,34 +754,33 @@
  * We allow two missing.
  */
 
-static bool full_td(int ctx, int cpu,
-		    struct runtime_stat *stat)
+static bool full_td(int cpu, struct runtime_stat *stat,
+		    struct runtime_stat_data *rsd)
 {
 	int c = 0;
 
-	if (runtime_stat_avg(stat, STAT_TOPDOWN_RETIRING, ctx, cpu) > 0)
+	if (runtime_stat_avg(stat, STAT_TOPDOWN_RETIRING, cpu, rsd) > 0)
 		c++;
-	if (runtime_stat_avg(stat, STAT_TOPDOWN_BE_BOUND, ctx, cpu) > 0)
+	if (runtime_stat_avg(stat, STAT_TOPDOWN_BE_BOUND, cpu, rsd) > 0)
 		c++;
-	if (runtime_stat_avg(stat, STAT_TOPDOWN_FE_BOUND, ctx, cpu) > 0)
+	if (runtime_stat_avg(stat, STAT_TOPDOWN_FE_BOUND, cpu, rsd) > 0)
 		c++;
-	if (runtime_stat_avg(stat, STAT_TOPDOWN_BAD_SPEC, ctx, cpu) > 0)
+	if (runtime_stat_avg(stat, STAT_TOPDOWN_BAD_SPEC, cpu, rsd) > 0)
 		c++;
 	return c >= 2;
 }
 
-static void print_smi_cost(struct perf_stat_config *config,
-			   int cpu, struct evsel *evsel,
+static void print_smi_cost(struct perf_stat_config *config, int cpu,
 			   struct perf_stat_output_ctx *out,
-			   struct runtime_stat *st)
+			   struct runtime_stat *st,
+			   struct runtime_stat_data *rsd)
 {
 	double smi_num, aperf, cycles, cost = 0.0;
-	int ctx = evsel_context(evsel);
 	const char *color = NULL;
 
-	smi_num = runtime_stat_avg(st, STAT_SMI_NUM, ctx, cpu);
-	aperf = runtime_stat_avg(st, STAT_APERF, ctx, cpu);
-	cycles = runtime_stat_avg(st, STAT_CYCLES, ctx, cpu);
+	smi_num = runtime_stat_avg(st, STAT_SMI_NUM, cpu, rsd);
+	aperf = runtime_stat_avg(st, STAT_APERF, cpu, rsd);
+	cycles = runtime_stat_avg(st, STAT_CYCLES, cpu, rsd);
 
 	if ((cycles == 0) || (aperf == 0))
 		return;
@@ -804,7 +815,8 @@
 			scale = 1e-9;
 		} else {
 			v = saved_value_lookup(metric_events[i], cpu, false,
-					       STAT_NONE, 0, st);
+					       STAT_NONE, 0, st,
+					       metric_events[i]->cgrp);
 			if (!v)
 				break;
 			stats = &v->stats;
@@ -930,12 +942,15 @@
 	print_metric_t print_metric = out->print_metric;
 	double total, ratio = 0.0, total2;
 	const char *color = NULL;
-	int ctx = evsel_context(evsel);
+	struct runtime_stat_data rsd = {
+		.ctx = evsel_context(evsel),
+		.cgrp = evsel->cgrp,
+	};
 	struct metric_event *me;
 	int num = 1;
 
 	if (evsel__match(evsel, HARDWARE, HW_INSTRUCTIONS)) {
-		total = runtime_stat_avg(st, STAT_CYCLES, ctx, cpu);
+		total = runtime_stat_avg(st, STAT_CYCLES, cpu, &rsd);
 
 		if (total) {
 			ratio = avg / total;
@@ -945,12 +960,11 @@
 			print_metric(config, ctxp, NULL, NULL, "insn per cycle", 0);
 		}
 
-		total = runtime_stat_avg(st, STAT_STALLED_CYCLES_FRONT,
-					 ctx, cpu);
+		total = runtime_stat_avg(st, STAT_STALLED_CYCLES_FRONT, cpu, &rsd);
 
 		total = max(total, runtime_stat_avg(st,
 						    STAT_STALLED_CYCLES_BACK,
-						    ctx, cpu));
+						    cpu, &rsd));
 
 		if (total && avg) {
 			out->new_line(config, ctxp);
@@ -960,8 +974,8 @@
 					ratio);
 		}
 	} else if (evsel__match(evsel, HARDWARE, HW_BRANCH_MISSES)) {
-		if (runtime_stat_n(st, STAT_BRANCHES, ctx, cpu) != 0)
-			print_branch_misses(config, cpu, evsel, avg, out, st);
+		if (runtime_stat_n(st, STAT_BRANCHES, cpu, &rsd) != 0)
+			print_branch_misses(config, cpu, avg, out, st, &rsd);
 		else
 			print_metric(config, ctxp, NULL, NULL, "of all branches", 0);
 	} else if (
@@ -970,8 +984,8 @@
 					((PERF_COUNT_HW_CACHE_OP_READ) << 8) |
 					 ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) {
 
-		if (runtime_stat_n(st, STAT_L1_DCACHE, ctx, cpu) != 0)
-			print_l1_dcache_misses(config, cpu, evsel, avg, out, st);
+		if (runtime_stat_n(st, STAT_L1_DCACHE, cpu, &rsd) != 0)
+			print_l1_dcache_misses(config, cpu, avg, out, st, &rsd);
 		else
 			print_metric(config, ctxp, NULL, NULL, "of all L1-dcache accesses", 0);
 	} else if (
@@ -980,8 +994,8 @@
 					((PERF_COUNT_HW_CACHE_OP_READ) << 8) |
 					 ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) {
 
-		if (runtime_stat_n(st, STAT_L1_ICACHE, ctx, cpu) != 0)
-			print_l1_icache_misses(config, cpu, evsel, avg, out, st);
+		if (runtime_stat_n(st, STAT_L1_ICACHE, cpu, &rsd) != 0)
+			print_l1_icache_misses(config, cpu, avg, out, st, &rsd);
 		else
 			print_metric(config, ctxp, NULL, NULL, "of all L1-icache accesses", 0);
 	} else if (
@@ -990,8 +1004,8 @@
 					((PERF_COUNT_HW_CACHE_OP_READ) << 8) |
 					 ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) {
 
-		if (runtime_stat_n(st, STAT_DTLB_CACHE, ctx, cpu) != 0)
-			print_dtlb_cache_misses(config, cpu, evsel, avg, out, st);
+		if (runtime_stat_n(st, STAT_DTLB_CACHE, cpu, &rsd) != 0)
+			print_dtlb_cache_misses(config, cpu, avg, out, st, &rsd);
 		else
 			print_metric(config, ctxp, NULL, NULL, "of all dTLB cache accesses", 0);
 	} else if (
@@ -1000,8 +1014,8 @@
 					((PERF_COUNT_HW_CACHE_OP_READ) << 8) |
 					 ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) {
 
-		if (runtime_stat_n(st, STAT_ITLB_CACHE, ctx, cpu) != 0)
-			print_itlb_cache_misses(config, cpu, evsel, avg, out, st);
+		if (runtime_stat_n(st, STAT_ITLB_CACHE, cpu, &rsd) != 0)
+			print_itlb_cache_misses(config, cpu, avg, out, st, &rsd);
 		else
 			print_metric(config, ctxp, NULL, NULL, "of all iTLB cache accesses", 0);
 	} else if (
@@ -1010,27 +1024,27 @@
 					((PERF_COUNT_HW_CACHE_OP_READ) << 8) |
 					 ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) {
 
-		if (runtime_stat_n(st, STAT_LL_CACHE, ctx, cpu) != 0)
-			print_ll_cache_misses(config, cpu, evsel, avg, out, st);
+		if (runtime_stat_n(st, STAT_LL_CACHE, cpu, &rsd) != 0)
+			print_ll_cache_misses(config, cpu, avg, out, st, &rsd);
 		else
 			print_metric(config, ctxp, NULL, NULL, "of all LL-cache accesses", 0);
 	} else if (evsel__match(evsel, HARDWARE, HW_CACHE_MISSES)) {
-		total = runtime_stat_avg(st, STAT_CACHEREFS, ctx, cpu);
+		total = runtime_stat_avg(st, STAT_CACHEREFS, cpu, &rsd);
 
 		if (total)
 			ratio = avg * 100 / total;
 
-		if (runtime_stat_n(st, STAT_CACHEREFS, ctx, cpu) != 0)
+		if (runtime_stat_n(st, STAT_CACHEREFS, cpu, &rsd) != 0)
 			print_metric(config, ctxp, NULL, "%8.3f %%",
 				     "of all cache refs", ratio);
 		else
 			print_metric(config, ctxp, NULL, NULL, "of all cache refs", 0);
 	} else if (evsel__match(evsel, HARDWARE, HW_STALLED_CYCLES_FRONTEND)) {
-		print_stalled_cycles_frontend(config, cpu, evsel, avg, out, st);
+		print_stalled_cycles_frontend(config, cpu, avg, out, st, &rsd);
 	} else if (evsel__match(evsel, HARDWARE, HW_STALLED_CYCLES_BACKEND)) {
-		print_stalled_cycles_backend(config, cpu, evsel, avg, out, st);
+		print_stalled_cycles_backend(config, cpu, avg, out, st, &rsd);
 	} else if (evsel__match(evsel, HARDWARE, HW_CPU_CYCLES)) {
-		total = runtime_stat_avg(st, STAT_NSECS, 0, cpu);
+		total = runtime_stat_avg(st, STAT_NSECS, cpu, &rsd);
 
 		if (total) {
 			ratio = avg / total;
@@ -1039,7 +1053,7 @@
 			print_metric(config, ctxp, NULL, NULL, "Ghz", 0);
 		}
 	} else if (perf_stat_evsel__is(evsel, CYCLES_IN_TX)) {
-		total = runtime_stat_avg(st, STAT_CYCLES, ctx, cpu);
+		total = runtime_stat_avg(st, STAT_CYCLES, cpu, &rsd);
 
 		if (total)
 			print_metric(config, ctxp, NULL,
@@ -1049,8 +1063,8 @@
 			print_metric(config, ctxp, NULL, NULL, "transactional cycles",
 				     0);
 	} else if (perf_stat_evsel__is(evsel, CYCLES_IN_TX_CP)) {
-		total = runtime_stat_avg(st, STAT_CYCLES, ctx, cpu);
-		total2 = runtime_stat_avg(st, STAT_CYCLES_IN_TX, ctx, cpu);
+		total = runtime_stat_avg(st, STAT_CYCLES, cpu, &rsd);
+		total2 = runtime_stat_avg(st, STAT_CYCLES_IN_TX, cpu, &rsd);
 
 		if (total2 < avg)
 			total2 = avg;
@@ -1060,21 +1074,19 @@
 		else
 			print_metric(config, ctxp, NULL, NULL, "aborted cycles", 0);
 	} else if (perf_stat_evsel__is(evsel, TRANSACTION_START)) {
-		total = runtime_stat_avg(st, STAT_CYCLES_IN_TX,
-					 ctx, cpu);
+		total = runtime_stat_avg(st, STAT_CYCLES_IN_TX, cpu, &rsd);
 
 		if (avg)
 			ratio = total / avg;
 
-		if (runtime_stat_n(st, STAT_CYCLES_IN_TX, ctx, cpu) != 0)
+		if (runtime_stat_n(st, STAT_CYCLES_IN_TX, cpu, &rsd) != 0)
 			print_metric(config, ctxp, NULL, "%8.0f",
 				     "cycles / transaction", ratio);
 		else
 			print_metric(config, ctxp, NULL, NULL, "cycles / transaction",
 				      0);
 	} else if (perf_stat_evsel__is(evsel, ELISION_START)) {
-		total = runtime_stat_avg(st, STAT_CYCLES_IN_TX,
-					 ctx, cpu);
+		total = runtime_stat_avg(st, STAT_CYCLES_IN_TX, cpu, &rsd);
 
 		if (avg)
 			ratio = total / avg;
@@ -1087,28 +1099,28 @@
 		else
 			print_metric(config, ctxp, NULL, NULL, "CPUs utilized", 0);
 	} else if (perf_stat_evsel__is(evsel, TOPDOWN_FETCH_BUBBLES)) {
-		double fe_bound = td_fe_bound(ctx, cpu, st);
+		double fe_bound = td_fe_bound(cpu, st, &rsd);
 
 		if (fe_bound > 0.2)
 			color = PERF_COLOR_RED;
 		print_metric(config, ctxp, color, "%8.1f%%", "frontend bound",
 				fe_bound * 100.);
 	} else if (perf_stat_evsel__is(evsel, TOPDOWN_SLOTS_RETIRED)) {
-		double retiring = td_retiring(ctx, cpu, st);
+		double retiring = td_retiring(cpu, st, &rsd);
 
 		if (retiring > 0.7)
 			color = PERF_COLOR_GREEN;
 		print_metric(config, ctxp, color, "%8.1f%%", "retiring",
 				retiring * 100.);
 	} else if (perf_stat_evsel__is(evsel, TOPDOWN_RECOVERY_BUBBLES)) {
-		double bad_spec = td_bad_spec(ctx, cpu, st);
+		double bad_spec = td_bad_spec(cpu, st, &rsd);
 
 		if (bad_spec > 0.1)
 			color = PERF_COLOR_RED;
 		print_metric(config, ctxp, color, "%8.1f%%", "bad speculation",
 				bad_spec * 100.);
 	} else if (perf_stat_evsel__is(evsel, TOPDOWN_SLOTS_ISSUED)) {
-		double be_bound = td_be_bound(ctx, cpu, st);
+		double be_bound = td_be_bound(cpu, st, &rsd);
 		const char *name = "backend bound";
 		static int have_recovery_bubbles = -1;
 
@@ -1121,43 +1133,43 @@
 
 		if (be_bound > 0.2)
 			color = PERF_COLOR_RED;
-		if (td_total_slots(ctx, cpu, st) > 0)
+		if (td_total_slots(cpu, st, &rsd) > 0)
 			print_metric(config, ctxp, color, "%8.1f%%", name,
 					be_bound * 100.);
 		else
 			print_metric(config, ctxp, NULL, NULL, name, 0);
 	} else if (perf_stat_evsel__is(evsel, TOPDOWN_RETIRING) &&
-			full_td(ctx, cpu, st)) {
-		double retiring = td_metric_ratio(ctx, cpu,
-						  STAT_TOPDOWN_RETIRING, st);
-
+		   full_td(cpu, st, &rsd)) {
+		double retiring = td_metric_ratio(cpu,
+						  STAT_TOPDOWN_RETIRING, st,
+						  &rsd);
 		if (retiring > 0.7)
 			color = PERF_COLOR_GREEN;
 		print_metric(config, ctxp, color, "%8.1f%%", "retiring",
 				retiring * 100.);
 	} else if (perf_stat_evsel__is(evsel, TOPDOWN_FE_BOUND) &&
-			full_td(ctx, cpu, st)) {
-		double fe_bound = td_metric_ratio(ctx, cpu,
-						  STAT_TOPDOWN_FE_BOUND, st);
-
+		   full_td(cpu, st, &rsd)) {
+		double fe_bound = td_metric_ratio(cpu,
+						  STAT_TOPDOWN_FE_BOUND, st,
+						  &rsd);
 		if (fe_bound > 0.2)
 			color = PERF_COLOR_RED;
 		print_metric(config, ctxp, color, "%8.1f%%", "frontend bound",
 				fe_bound * 100.);
 	} else if (perf_stat_evsel__is(evsel, TOPDOWN_BE_BOUND) &&
-			full_td(ctx, cpu, st)) {
-		double be_bound = td_metric_ratio(ctx, cpu,
-						  STAT_TOPDOWN_BE_BOUND, st);
-
+		   full_td(cpu, st, &rsd)) {
+		double be_bound = td_metric_ratio(cpu,
+						  STAT_TOPDOWN_BE_BOUND, st,
+						  &rsd);
 		if (be_bound > 0.2)
 			color = PERF_COLOR_RED;
 		print_metric(config, ctxp, color, "%8.1f%%", "backend bound",
 				be_bound * 100.);
 	} else if (perf_stat_evsel__is(evsel, TOPDOWN_BAD_SPEC) &&
-			full_td(ctx, cpu, st)) {
-		double bad_spec = td_metric_ratio(ctx, cpu,
-						  STAT_TOPDOWN_BAD_SPEC, st);
-
+		   full_td(cpu, st, &rsd)) {
+		double bad_spec = td_metric_ratio(cpu,
+						  STAT_TOPDOWN_BAD_SPEC, st,
+						  &rsd);
 		if (bad_spec > 0.1)
 			color = PERF_COLOR_RED;
 		print_metric(config, ctxp, color, "%8.1f%%", "bad speculation",
@@ -1165,11 +1177,11 @@
 	} else if (evsel->metric_expr) {
 		generic_metric(config, evsel->metric_expr, evsel->metric_events, NULL,
 				evsel->name, evsel->metric_name, NULL, 1, cpu, out, st);
-	} else if (runtime_stat_n(st, STAT_NSECS, 0, cpu) != 0) {
+	} else if (runtime_stat_n(st, STAT_NSECS, cpu, &rsd) != 0) {
 		char unit = 'M';
 		char unit_buf[10];
 
-		total = runtime_stat_avg(st, STAT_NSECS, 0, cpu);
+		total = runtime_stat_avg(st, STAT_NSECS, cpu, &rsd);
 
 		if (total)
 			ratio = 1000.0 * avg / total;
@@ -1180,7 +1192,7 @@
 		snprintf(unit_buf, sizeof(unit_buf), "%c/sec", unit);
 		print_metric(config, ctxp, NULL, "%8.3f", unit_buf, ratio);
 	} else if (perf_stat_evsel__is(evsel, SMI_NUM)) {
-		print_smi_cost(config, cpu, evsel, out, st);
+		print_smi_cost(config, cpu, out, st, &rsd);
 	} else {
 		num = 0;
 	}