diff --git a/include/zephyr/net/net_if.h b/include/zephyr/net/net_if.h
index 35f5e8577ffd6..cfba714d6631a 100644
--- a/include/zephyr/net/net_if.h
+++ b/include/zephyr/net/net_if.h
@@ -94,6 +94,11 @@ struct net_if_addr {
 		struct {
 			/** Duplicate address detection (DAD) timer */
 			sys_snode_t dad_node;
+
+			/** DAD needed list node */
+			sys_snode_t dad_need_node;
+
+			/** DAD start time */
 			uint32_t dad_start;
 
 			/** How many times we have done DAD */
@@ -104,6 +109,11 @@ struct net_if_addr {
 		struct {
 			/** Address conflict detection (ACD) timer. */
 			sys_snode_t acd_node;
+
+			/** ACD needed list node */
+			sys_snode_t acd_need_node;
+
+			/** ACD timeout value. */
 			k_timepoint_t acd_timeout;
 
 			/** ACD probe/announcement counter. */
diff --git a/subsys/net/ip/net_if.c b/subsys/net/ip/net_if.c
index 3c1f4aa5382ce..597ca39bb4aec 100644
--- a/subsys/net/ip/net_if.c
+++ b/subsys/net/ip/net_if.c
@@ -1324,8 +1324,9 @@ void net_if_ipv6_start_dad(struct net_if *iface,
 
 void net_if_start_dad(struct net_if *iface)
 {
-	struct net_if_addr *ifaddr;
+	struct net_if_addr *ifaddr, *next;
 	struct net_if_ipv6 *ipv6;
+	sys_slist_t dad_needed;
 	struct in6_addr addr = { };
 	int ret;
 
@@ -1357,6 +1358,8 @@ void net_if_start_dad(struct net_if *iface)
 	/* Start DAD for all the addresses that were added earlier when
 	 * the interface was down.
 	 */
+	sys_slist_init(&dad_needed);
+
 	ARRAY_FOR_EACH(ipv6->unicast, i) {
 		if (!ipv6->unicast[i].is_used ||
 		    ipv6->unicast[i].address.family != AF_INET6 ||
@@ -1366,9 +1369,21 @@ void net_if_start_dad(struct net_if *iface)
 			continue;
 		}
 
-		net_if_ipv6_start_dad(iface, &ipv6->unicast[i]);
+		sys_slist_prepend(&dad_needed, &ipv6->unicast[i].dad_need_node);
 	}
 
+	net_if_unlock(iface);
+
+	/* Start DAD for all the addresses without holding the iface lock
+	 * to avoid any possible mutex deadlock issues.
+	 */
+	SYS_SLIST_FOR_EACH_CONTAINER_SAFE(&dad_needed,
+					  ifaddr, next, dad_need_node) {
+		net_if_ipv6_start_dad(iface, ifaddr);
+	}
+
+	return;
+
 out:
 	net_if_unlock(iface);
 }
@@ -1415,7 +1430,10 @@ void net_if_ipv6_dad_failed(struct net_if *iface, const struct in6_addr *addr)
 	net_if_ipv6_addr_rm(iface, addr);
 
 	if (IS_ENABLED(CONFIG_NET_IPV6_PE) && iface->pe_enabled) {
+		net_if_unlock(iface);
+
 		net_ipv6_pe_start(iface, addr, timeout, preferred_lifetime);
+		return;
 	}
 
 out:
@@ -1519,6 +1537,8 @@ void net_if_start_rs(struct net_if *iface)
 		goto out;
 	}
 
+	net_if_unlock(iface);
+
 	NET_DBG("Starting ND/RS for iface %p", iface);
 
 	if (!net_ipv6_start_rs(iface)) {
@@ -1534,6 +1554,7 @@ void net_if_start_rs(struct net_if *iface)
 		}
 	}
 
+	return;
 out:
 	net_if_unlock(iface);
 }
@@ -1941,6 +1962,7 @@ struct net_if_addr *net_if_ipv6_addr_add(struct net_if *iface,
 {
 	struct net_if_addr *ifaddr = NULL;
 	struct net_if_ipv6 *ipv6;
+	bool do_dad = false;
 
 	net_if_lock(iface);
 
@@ -1992,8 +2014,7 @@ struct net_if_addr *net_if_ipv6_addr_add(struct net_if *iface,
 			 */
 			join_mcast_nodes(iface,
 					 &ipv6->unicast[i].address.in6_addr);
-
-			net_if_ipv6_start_dad(iface, &ipv6->unicast[i]);
+			do_dad = true;
 		} else {
 			/* If DAD is not done for point-to-point links, then
 			 * the address is usable immediately.
@@ -2007,9 +2028,17 @@ struct net_if_addr *net_if_ipv6_addr_add(struct net_if *iface,
 			sizeof(struct in6_addr));
 
 		ifaddr = &ipv6->unicast[i];
-		goto out;
+		break;
 	}
 
+	net_if_unlock(iface);
+
+	if (ifaddr != NULL && do_dad) {
+		net_if_ipv6_start_dad(iface, ifaddr);
+	}
+
+	return ifaddr;
+
 out:
 	net_if_unlock(iface);
 
@@ -4179,7 +4208,9 @@ void net_if_ipv4_start_acd(struct net_if *iface, struct net_if_addr *ifaddr)
 
 void net_if_start_acd(struct net_if *iface)
 {
+	struct net_if_addr *ifaddr, *next;
 	struct net_if_ipv4 *ipv4;
+	sys_slist_t acd_needed;
 	int ret;
 
 	net_if_lock(iface);
@@ -4201,6 +4232,11 @@ void net_if_start_acd(struct net_if *iface)
 
 	ipv4->conflict_cnt = 0;
 
+	/* Start ACD for all the addresses that were added earlier when
+	 * the interface was down.
+	 */
+	sys_slist_init(&acd_needed);
+
 	/* Start ACD for all the addresses that were added earlier when
 	 * the interface was down.
 	 */
@@ -4212,9 +4248,21 @@ void net_if_start_acd(struct net_if *iface)
 			continue;
 		}
 
-		net_if_ipv4_start_acd(iface, &ipv4->unicast[i].ipv4);
+		sys_slist_prepend(&acd_needed, &ipv4->unicast[i].ipv4.acd_need_node);
 	}
 
+	net_if_unlock(iface);
+
+	/* Start ACD for all the addresses without holding the iface lock
+	 * to avoid any possible mutex deadlock issues.
+	 */
+	SYS_SLIST_FOR_EACH_CONTAINER_SAFE(&acd_needed,
+					  ifaddr, next, acd_need_node) {
+		net_if_ipv4_start_acd(iface, ifaddr);
+	}
+
+	return;
+
 out:
 	net_if_unlock(iface);
 }
@@ -4306,7 +4354,8 @@ struct net_if_addr *net_if_ipv4_addr_add(struct net_if *iface,
 
 		if (!(l2_flags_get(iface) & NET_L2_POINT_TO_POINT) &&
 		    !net_ipv4_is_addr_loopback(addr)) {
-			net_if_ipv4_start_acd(iface, ifaddr);
+			/* ACD is started after the lock is released. */
+			;
 		} else {
 			ifaddr->addr_state = NET_ADDR_PREFERRED;
 		}
@@ -4314,7 +4363,12 @@ struct net_if_addr *net_if_ipv4_addr_add(struct net_if *iface,
 		net_mgmt_event_notify_with_info(NET_EVENT_IPV4_ADDR_ADD, iface,
 						&ifaddr->address.in_addr,
 						sizeof(struct in_addr));
-		goto out;
+
+		net_if_unlock(iface);
+
+		net_if_ipv4_start_acd(iface, ifaddr);
+
+		return ifaddr;
 	}
 
 out: