diff --git a/kernel/sched/walt/walt.c b/kernel/sched/walt/walt.c
index fc9d05e02fbe..cdab278259e1 100644
--- a/kernel/sched/walt/walt.c
+++ b/kernel/sched/walt/walt.c
@@ -3752,7 +3752,7 @@ static void android_rvh_enqueue_task(void *unused, struct rq *rq, struct task_st
 	wts->last_enqueued_ts = wallclock;
 	sched_update_nr_prod(rq->cpu, true);
 
-	if (fair_policy(p->policy)) {
+	if (walt_fair_task(p)) {
 		wts->misfit = !task_fits_max(p, rq->cpu);
 		inc_rq_walt_stats(rq, p);
 	}
@@ -3772,7 +3772,7 @@ static void android_rvh_dequeue_task(void *unused, struct rq *rq, struct task_st
 
 	sched_update_nr_prod(rq->cpu, false);
 
-	if (fair_policy(p->policy))
+	if (walt_fair_task(p))
 		dec_rq_walt_stats(rq, p);
 
 	walt_dec_cumulative_runnable_avg(rq, p);
diff --git a/kernel/sched/walt/walt.h b/kernel/sched/walt/walt.h
index fc186cf98c6b..f16314775ce6 100644
--- a/kernel/sched/walt/walt.h
+++ b/kernel/sched/walt/walt.h
@@ -858,4 +858,13 @@ static inline struct task_group *css_tg(struct cgroup_subsys_state *css)
 	return css ? container_of(css, struct task_group, css) : NULL;
 }
 
+/*
+ * The policy of a RT boosted task (via PI mutex) still indicates it is
+ * a fair task, so use prio check as well. The prio check alone is not
+ * sufficient since idle task also has 120 priority.
+ */
+static inline bool walt_fair_task(struct task_struct *p)
+{
+	return p->prio >= MAX_RT_PRIO && !is_idle_task(p);
+}
 #endif /* _WALT_H */
diff --git a/kernel/sched/walt/walt_lb.c b/kernel/sched/walt/walt_lb.c
index fb290403daae..c5eb0e5a31e8 100644
--- a/kernel/sched/walt/walt_lb.c
+++ b/kernel/sched/walt/walt_lb.c
@@ -144,7 +144,7 @@ static void walt_lb_check_for_rotation(struct rq *src_rq)
 		if (is_reserved(i))
 			continue;
 
-		if (!rq->misfit_task_load)
+		if (!rq->misfit_task_load || !walt_fair_task(rq->curr))
 			continue;
 
 		wts = (struct walt_task_struct *) rq->curr->android_vendor_data1;
@@ -167,7 +167,7 @@ static void walt_lb_check_for_rotation(struct rq *src_rq)
 		if (is_reserved(i))
 			continue;
 
-		if (rq->curr->prio < MAX_RT_PRIO)
+		if (!walt_fair_task(rq->curr))
 			continue;
 
 		if (rq->nr_running > 1)
@@ -191,8 +191,7 @@ static void walt_lb_check_for_rotation(struct rq *src_rq)
 	dst_rq = cpu_rq(dst_cpu);
 
 	double_rq_lock(src_rq, dst_rq);
-	if (dst_rq->curr->prio >= MAX_RT_PRIO && dst_rq->curr != dst_rq->idle &&
-		src_rq->curr->prio >= MAX_RT_PRIO && src_rq->curr != src_rq->idle) {
+	if (walt_fair_task(dst_rq->curr)) {
 		get_task_struct(src_rq->curr);
 		get_task_struct(dst_rq->curr);
 
@@ -494,7 +493,7 @@ static void walt_lb_tick(void *unused, struct rq *rq)
 
 	if (static_branch_unlikely(&walt_disabled))
 		return;
-	if (!rq->misfit_task_load)
+	if (!rq->misfit_task_load || !walt_fair_task(p))
 		return;
 
 	if (p->state != TASK_RUNNING || p->nr_cpus_allowed == 1)