--- /dev/null
+// SPDX-License-Identifier: GPL-2.0-only
+
+#include <test_progs.h>
+#include "test_lookup_and_delete.skel.h"
+
+#define START_VALUE 1234
+#define NEW_VALUE 4321
+#define MAX_ENTRIES 2
+
+static int duration;
+static int nr_cpus;
+
+static int fill_values(int map_fd)
+{
+       __u64 key, value = START_VALUE;
+       int err;
+
+       for (key = 1; key < MAX_ENTRIES + 1; key++) {
+               err = bpf_map_update_elem(map_fd, &key, &value, BPF_NOEXIST);
+               if (!ASSERT_OK(err, "bpf_map_update_elem"))
+                       return -1;
+       }
+
+       return 0;
+}
+
+static int fill_values_percpu(int map_fd)
+{
+       __u64 key, value[nr_cpus];
+       int i, err;
+
+       for (i = 0; i < nr_cpus; i++)
+               value[i] = START_VALUE;
+
+       for (key = 1; key < MAX_ENTRIES + 1; key++) {
+               err = bpf_map_update_elem(map_fd, &key, value, BPF_NOEXIST);
+               if (!ASSERT_OK(err, "bpf_map_update_elem"))
+                       return -1;
+       }
+
+       return 0;
+}
+
+static struct test_lookup_and_delete *setup_prog(enum bpf_map_type map_type,
+                                                int *map_fd)
+{
+       struct test_lookup_and_delete *skel;
+       int err;
+
+       skel = test_lookup_and_delete__open();
+       if (!ASSERT_OK_PTR(skel, "test_lookup_and_delete__open"))
+               return NULL;
+
+       err = bpf_map__set_type(skel->maps.hash_map, map_type);
+       if (!ASSERT_OK(err, "bpf_map__set_type"))
+               goto cleanup;
+
+       err = bpf_map__set_max_entries(skel->maps.hash_map, MAX_ENTRIES);
+       if (!ASSERT_OK(err, "bpf_map__set_max_entries"))
+               goto cleanup;
+
+       err = test_lookup_and_delete__load(skel);
+       if (!ASSERT_OK(err, "test_lookup_and_delete__load"))
+               goto cleanup;
+
+       *map_fd = bpf_map__fd(skel->maps.hash_map);
+       if (!ASSERT_GE(*map_fd, 0, "bpf_map__fd"))
+               goto cleanup;
+
+       return skel;
+
+cleanup:
+       test_lookup_and_delete__destroy(skel);
+       return NULL;
+}
+
+/* Triggers BPF program that updates map with given key and value */
+static int trigger_tp(struct test_lookup_and_delete *skel, __u64 key,
+                     __u64 value)
+{
+       int err;
+
+       skel->bss->set_pid = getpid();
+       skel->bss->set_key = key;
+       skel->bss->set_value = value;
+
+       err = test_lookup_and_delete__attach(skel);
+       if (!ASSERT_OK(err, "test_lookup_and_delete__attach"))
+               return -1;
+
+       syscall(__NR_getpgid);
+
+       test_lookup_and_delete__detach(skel);
+
+       return 0;
+}
+
+static void test_lookup_and_delete_hash(void)
+{
+       struct test_lookup_and_delete *skel;
+       __u64 key, value;
+       int map_fd, err;
+
+       /* Setup program and fill the map. */
+       skel = setup_prog(BPF_MAP_TYPE_HASH, &map_fd);
+       if (!ASSERT_OK_PTR(skel, "setup_prog"))
+               return;
+
+       err = fill_values(map_fd);
+       if (!ASSERT_OK(err, "fill_values"))
+               goto cleanup;
+
+       /* Lookup and delete element. */
+       key = 1;
+       err = bpf_map_lookup_and_delete_elem(map_fd, &key, &value);
+       if (!ASSERT_OK(err, "bpf_map_lookup_and_delete_elem"))
+               goto cleanup;
+
+       /* Fetched value should match the initially set value. */
+       if (CHECK(value != START_VALUE, "bpf_map_lookup_and_delete_elem",
+                 "unexpected value=%lld\n", value))
+               goto cleanup;
+
+       /* Check that the entry is non existent. */
+       err = bpf_map_lookup_elem(map_fd, &key, &value);
+       if (!ASSERT_ERR(err, "bpf_map_lookup_elem"))
+               goto cleanup;
+
+cleanup:
+       test_lookup_and_delete__destroy(skel);
+}
+
+static void test_lookup_and_delete_percpu_hash(void)
+{
+       struct test_lookup_and_delete *skel;
+       __u64 key, val, value[nr_cpus];
+       int map_fd, err, i;
+
+       /* Setup program and fill the map. */
+       skel = setup_prog(BPF_MAP_TYPE_PERCPU_HASH, &map_fd);
+       if (!ASSERT_OK_PTR(skel, "setup_prog"))
+               return;
+
+       err = fill_values_percpu(map_fd);
+       if (!ASSERT_OK(err, "fill_values_percpu"))
+               goto cleanup;
+
+       /* Lookup and delete element. */
+       key = 1;
+       err = bpf_map_lookup_and_delete_elem(map_fd, &key, value);
+       if (!ASSERT_OK(err, "bpf_map_lookup_and_delete_elem"))
+               goto cleanup;
+
+       for (i = 0; i < nr_cpus; i++) {
+               val = value[i];
+
+               /* Fetched value should match the initially set value. */
+               if (CHECK(val != START_VALUE, "map value",
+                         "unexpected for cpu %d: %lld\n", i, val))
+                       goto cleanup;
+       }
+
+       /* Check that the entry is non existent. */
+       err = bpf_map_lookup_elem(map_fd, &key, value);
+       if (!ASSERT_ERR(err, "bpf_map_lookup_elem"))
+               goto cleanup;
+
+cleanup:
+       test_lookup_and_delete__destroy(skel);
+}
+
+static void test_lookup_and_delete_lru_hash(void)
+{
+       struct test_lookup_and_delete *skel;
+       __u64 key, value;
+       int map_fd, err;
+
+       /* Setup program and fill the LRU map. */
+       skel = setup_prog(BPF_MAP_TYPE_LRU_HASH, &map_fd);
+       if (!ASSERT_OK_PTR(skel, "setup_prog"))
+               return;
+
+       err = fill_values(map_fd);
+       if (!ASSERT_OK(err, "fill_values"))
+               goto cleanup;
+
+       /* Insert new element at key=3, should reuse LRU element. */
+       key = 3;
+       err = trigger_tp(skel, key, NEW_VALUE);
+       if (!ASSERT_OK(err, "trigger_tp"))
+               goto cleanup;
+
+       /* Lookup and delete element 3. */
+       err = bpf_map_lookup_and_delete_elem(map_fd, &key, &value);
+       if (!ASSERT_OK(err, "bpf_map_lookup_and_delete_elem"))
+               goto cleanup;
+
+       /* Value should match the new value. */
+       if (CHECK(value != NEW_VALUE, "bpf_map_lookup_and_delete_elem",
+                 "unexpected value=%lld\n", value))
+               goto cleanup;
+
+       /* Check that entries 3 and 1 are non existent. */
+       err = bpf_map_lookup_elem(map_fd, &key, &value);
+       if (!ASSERT_ERR(err, "bpf_map_lookup_elem"))
+               goto cleanup;
+
+       key = 1;
+       err = bpf_map_lookup_elem(map_fd, &key, &value);
+       if (!ASSERT_ERR(err, "bpf_map_lookup_elem"))
+               goto cleanup;
+
+cleanup:
+       test_lookup_and_delete__destroy(skel);
+}
+
+static void test_lookup_and_delete_lru_percpu_hash(void)
+{
+       struct test_lookup_and_delete *skel;
+       __u64 key, val, value[nr_cpus];
+       int map_fd, err, i, cpucnt = 0;
+
+       /* Setup program and fill the LRU map. */
+       skel = setup_prog(BPF_MAP_TYPE_LRU_PERCPU_HASH, &map_fd);
+       if (!ASSERT_OK_PTR(skel, "setup_prog"))
+               return;
+
+       err = fill_values_percpu(map_fd);
+       if (!ASSERT_OK(err, "fill_values_percpu"))
+               goto cleanup;
+
+       /* Insert new element at key=3, should reuse LRU element 1. */
+       key = 3;
+       err = trigger_tp(skel, key, NEW_VALUE);
+       if (!ASSERT_OK(err, "trigger_tp"))
+               goto cleanup;
+
+       /* Clean value. */
+       for (i = 0; i < nr_cpus; i++)
+               value[i] = 0;
+
+       /* Lookup and delete element 3. */
+       err = bpf_map_lookup_and_delete_elem(map_fd, &key, value);
+       if (!ASSERT_OK(err, "bpf_map_lookup_and_delete_elem")) {
+               goto cleanup;
+       }
+
+       /* Check if only one CPU has set the value. */
+       for (i = 0; i < nr_cpus; i++) {
+               val = value[i];
+               if (val) {
+                       if (CHECK(val != NEW_VALUE, "map value",
+                                 "unexpected for cpu %d: %lld\n", i, val))
+                               goto cleanup;
+                       cpucnt++;
+               }
+       }
+       if (CHECK(cpucnt != 1, "map value", "set for %d CPUs instead of 1!\n",
+                 cpucnt))
+               goto cleanup;
+
+       /* Check that entries 3 and 1 are non existent. */
+       err = bpf_map_lookup_elem(map_fd, &key, &value);
+       if (!ASSERT_ERR(err, "bpf_map_lookup_elem"))
+               goto cleanup;
+
+       key = 1;
+       err = bpf_map_lookup_elem(map_fd, &key, &value);
+       if (!ASSERT_ERR(err, "bpf_map_lookup_elem"))
+               goto cleanup;
+
+cleanup:
+       test_lookup_and_delete__destroy(skel);
+}
+
+void test_lookup_and_delete(void)
+{
+       nr_cpus = bpf_num_possible_cpus();
+
+       if (test__start_subtest("lookup_and_delete"))
+               test_lookup_and_delete_hash();
+       if (test__start_subtest("lookup_and_delete_percpu"))
+               test_lookup_and_delete_percpu_hash();
+       if (test__start_subtest("lookup_and_delete_lru"))
+               test_lookup_and_delete_lru_hash();
+       if (test__start_subtest("lookup_and_delete_lru_percpu"))
+               test_lookup_and_delete_lru_percpu_hash();
+}