bpf: Adjust free target to avoid global starvation of LRU map

BPF_MAP_TYPE_LRU_HASH can recycle most recent elements well before the
map is full, due to percpu reservations and force shrink before
neighbor stealing. Once a CPU is unable to borrow from the global map,
it will once steal one elem from a neighbor and after that each time
flush this one element to the global list and immediately recycle it.

Batch value LOCAL_FREE_TARGET (128) will exhaust a 10K element map
with 79 CPUs. CPU 79 will observe this behavior even while its
neighbors hold 78 * 127 + 1 * 15 == 9921 free elements (99%).

CPUs need not be active concurrently. The issue can appear with
affinity migration, e.g., irqbalance. Each CPU can reserve and then
hold onto its 128 elements indefinitely.

Avoid global list exhaustion by limiting aggregate percpu caches to
half of map size, by adjusting LOCAL_FREE_TARGET based on cpu count.
This change has no effect on sufficiently large tables.

Similar to LOCAL_NR_SCANS and lru->nr_scans, introduce a map variable
lru->free_target. The extra field fits in a hole in struct bpf_lru.
The cacheline is already warm where read in the hot path. The field is
only accessed with the lru lock held.

Tested-by: Anton Protopopov <a.s.protopopov@gmail.com>
Signed-off-by: Willem de Bruijn <willemb@google.com>
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Link: https://lore.kernel.org/r/20250618215803.3587312-1-willemdebruijn.kernel@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

1 files changed, 35 insertions, 37 deletions

diff --git a/tools/testing/selftests/bpf/test_lru_map.c b/tools/testing/selftests/bpf/test_lru_map.c
index fda7589c5023..4ae83f4b7fc7 100644
--- a/tools/testing/selftests/bpf/test_lru_map.c
+++ b/tools/testing/selftests/bpf/test_lru_map.c
@@ -138,6 +138,12 @@ static int sched_next_online(int pid, int *next_to_try)
 	return ret;
 }
 
+/* Inverse of how bpf_common_lru_populate derives target_free from map_size. */
+static unsigned int __map_size(unsigned int tgt_free)
+{
+	return tgt_free * nr_cpus * 2;
+}
+
 /* Size of the LRU map is 2
  * Add key=1 (+1 key)
  * Add key=2 (+1 key)
@@ -231,11 +237,11 @@ static void test_lru_sanity0(int map_type, int map_flags)
 	printf("Pass\n");
 }
 
-/* Size of the LRU map is 1.5*tgt_free
- * Insert 1 to tgt_free (+tgt_free keys)
- * Lookup 1 to tgt_free/2
- * Insert 1+tgt_free to 2*tgt_free (+tgt_free keys)
- * => 1+tgt_free/2 to LOCALFREE_TARGET will be removed by LRU
+/* Verify that unreferenced elements are recycled before referenced ones.
+ * Insert elements.
+ * Reference a subset of these.
+ * Insert more, enough to trigger recycling.
+ * Verify that unreferenced are recycled.
  */
 static void test_lru_sanity1(int map_type, int map_flags, unsigned int tgt_free)
 {
@@ -257,7 +263,7 @@ static void test_lru_sanity1(int map_type, int map_flags, unsigned int tgt_free)
 	batch_size = tgt_free / 2;
 	assert(batch_size * 2 == tgt_free);
 
-	map_size = tgt_free + batch_size;
+	map_size = __map_size(tgt_free) + batch_size;
 	lru_map_fd = create_map(map_type, map_flags, map_size);
 	assert(lru_map_fd != -1);
 
@@ -266,13 +272,13 @@ static void test_lru_sanity1(int map_type, int map_flags, unsigned int tgt_free)
 
 	value[0] = 1234;
 
-	/* Insert 1 to tgt_free (+tgt_free keys) */
-	end_key = 1 + tgt_free;
+	/* Insert map_size - batch_size keys */
+	end_key = 1 + __map_size(tgt_free);
 	for (key = 1; key < end_key; key++)
 		assert(!bpf_map_update_elem(lru_map_fd, &key, value,
 					    BPF_NOEXIST));
 
-	/* Lookup 1 to tgt_free/2 */
+	/* Lookup 1 to batch_size */
 	end_key = 1 + batch_size;
 	for (key = 1; key < end_key; key++) {
 		assert(!bpf_map_lookup_elem_with_ref_bit(lru_map_fd, key, value));
@@ -280,12 +286,13 @@ static void test_lru_sanity1(int map_type, int map_flags, unsigned int tgt_free)
 					    BPF_NOEXIST));
 	}
 
-	/* Insert 1+tgt_free to 2*tgt_free
-	 * => 1+tgt_free/2 to LOCALFREE_TARGET will be
+	/* Insert another map_size - batch_size keys
+	 * Map will contain 1 to batch_size plus these latest, i.e.,
+	 * => previous 1+batch_size to map_size - batch_size will have been
 	 * removed by LRU
 	 */
-	key = 1 + tgt_free;
-	end_key = key + tgt_free;
+	key = 1 + __map_size(tgt_free);
+	end_key = key + __map_size(tgt_free);
 	for (; key < end_key; key++) {
 		assert(!bpf_map_update_elem(lru_map_fd, &key, value,
 					    BPF_NOEXIST));
@@ -301,17 +308,8 @@ static void test_lru_sanity1(int map_type, int map_flags, unsigned int tgt_free)
 	printf("Pass\n");
 }
 
-/* Size of the LRU map 1.5 * tgt_free
- * Insert 1 to tgt_free (+tgt_free keys)
- * Update 1 to tgt_free/2
- *   => The original 1 to tgt_free/2 will be removed due to
- *      the LRU shrink process
- * Re-insert 1 to tgt_free/2 again and do a lookup immeidately
- * Insert 1+tgt_free to tgt_free*3/2
- * Insert 1+tgt_free*3/2 to tgt_free*5/2
- *   => Key 1+tgt_free to tgt_free*3/2
- *      will be removed from LRU because it has never
- *      been lookup and ref bit is not set
+/* Verify that insertions exceeding map size will recycle the oldest.
+ * Verify that unreferenced elements are recycled before referenced.
  */
 static void test_lru_sanity2(int map_type, int map_flags, unsigned int tgt_free)
 {
@@ -334,7 +332,7 @@ static void test_lru_sanity2(int map_type, int map_flags, unsigned int tgt_free)
 	batch_size = tgt_free / 2;
 	assert(batch_size * 2 == tgt_free);
 
-	map_size = tgt_free + batch_size;
+	map_size = __map_size(tgt_free) + batch_size;
 	lru_map_fd = create_map(map_type, map_flags, map_size);
 	assert(lru_map_fd != -1);
 
@@ -343,8 +341,8 @@ static void test_lru_sanity2(int map_type, int map_flags, unsigned int tgt_free)
 
 	value[0] = 1234;
 
-	/* Insert 1 to tgt_free (+tgt_free keys) */
-	end_key = 1 + tgt_free;
+	/* Insert map_size - batch_size keys */
+	end_key = 1 + __map_size(tgt_free);
 	for (key = 1; key < end_key; key++)
 		assert(!bpf_map_update_elem(lru_map_fd, &key, value,
 					    BPF_NOEXIST));
@@ -357,8 +355,7 @@ static void test_lru_sanity2(int map_type, int map_flags, unsigned int tgt_free)
 	 * shrink the inactive list to get tgt_free
 	 * number of free nodes.
 	 *
-	 * Hence, the oldest key 1 to tgt_free/2
-	 * are removed from the LRU list.
+	 * Hence, the oldest key is removed from the LRU list.
 	 */
 	key = 1;
 	if (map_type == BPF_MAP_TYPE_LRU_PERCPU_HASH) {
@@ -370,8 +367,7 @@ static void test_lru_sanity2(int map_type, int map_flags, unsigned int tgt_free)
 					   BPF_EXIST));
 	}
 
-	/* Re-insert 1 to tgt_free/2 again and do a lookup
-	 * immeidately.
+	/* Re-insert 1 to batch_size again and do a lookup immediately.
 	 */
 	end_key = 1 + batch_size;
 	value[0] = 4321;
@@ -387,17 +383,18 @@ static void test_lru_sanity2(int map_type, int map_flags, unsigned int tgt_free)
 
 	value[0] = 1234;
 
-	/* Insert 1+tgt_free to tgt_free*3/2 */
-	end_key = 1 + tgt_free + batch_size;
-	for (key = 1 + tgt_free; key < end_key; key++)
+	/* Insert batch_size new elements */
+	key = 1 + __map_size(tgt_free);
+	end_key = key + batch_size;
+	for (; key < end_key; key++)
 		/* These newly added but not referenced keys will be
 		 * gone during the next LRU shrink.
 		 */
 		assert(!bpf_map_update_elem(lru_map_fd, &key, value,
 					    BPF_NOEXIST));
 
-	/* Insert 1+tgt_free*3/2 to  tgt_free*5/2 */
-	end_key = key + tgt_free;
+	/* Insert map_size - batch_size elements */
+	end_key += __map_size(tgt_free);
 	for (; key < end_key; key++) {
 		assert(!bpf_map_update_elem(lru_map_fd, &key, value,
 					    BPF_NOEXIST));
@@ -500,7 +497,8 @@ static void test_lru_sanity4(int map_type, int map_flags, unsigned int tgt_free)
 		lru_map_fd = create_map(map_type, map_flags,
 					3 * tgt_free * nr_cpus);
 	else
-		lru_map_fd = create_map(map_type, map_flags, 3 * tgt_free);
+		lru_map_fd = create_map(map_type, map_flags,
+					3 * __map_size(tgt_free));
 	assert(lru_map_fd != -1);
 
 	expected_map_fd = create_map(BPF_MAP_TYPE_HASH, 0,