summary refs log tree commit diff
path: root/samples/bpf/map_perf_test_kern.c
diff options
context:
space:
mode:
authorMartin KaFai Lau <kafai@fb.com>2017-04-14 10:30:30 -0700
committerDavid S. Miller <davem@davemloft.net>2017-04-17 13:55:52 -0400
commit3a5795b83d578cc542a92c94399946258cf1a2af (patch)
treef5d017cbdf7cd49aaa75c17c97abc4a3346f1237 /samples/bpf/map_perf_test_kern.c
parent695ba2651a2ecbb336145f9cc033c85b9c6a5cee (diff)
downloadlinux-3a5795b83d578cc542a92c94399946258cf1a2af.tar.gz
bpf: lru: Add map-in-map LRU example
This patch adds a map-in-map LRU example.
If we know only a subset of cores will use the
LRU, we can allocate a common LRU list per targeting core
and store it into an array-of-hashs.

It allows using the common LRU map with map-update performance
comparable to the BPF_F_NO_COMMON_LRU map but without wasting memory
on the unused cores that we know they will never access the LRU map.

BPF_F_NO_COMMON_LRU:
> map_perf_test 32 8 10000000 10000000 | awk '{sum += $3}END{print sum}'
9234314 (9.23M/s)

map-in-map LRU:
> map_perf_test 512 8 1260000 80000000 | awk '{sum += $3}END{print sum}'
9962743 (9.96M/s)

Notes that the max_entries for the map-in-map LRU test is 1260000 which
is the max_entries for each inner LRU map.  8 processes have been
started, so 8 * 1260000 = 10080000 (~10M) which is close to what is
used in the BPF_F_NO_COMMON_LRU test.

Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'samples/bpf/map_perf_test_kern.c')
-rw-r--r--samples/bpf/map_perf_test_kern.c34
1 files changed, 31 insertions, 3 deletions
diff --git a/samples/bpf/map_perf_test_kern.c b/samples/bpf/map_perf_test_kern.c
index 404ed53b8a53..245165817fbe 100644
--- a/samples/bpf/map_perf_test_kern.c
+++ b/samples/bpf/map_perf_test_kern.c
@@ -11,6 +11,7 @@
 #include "bpf_helpers.h"
 
 #define MAX_ENTRIES 1000
+#define MAX_NR_CPUS 1024
 
 struct bpf_map_def SEC("maps") hash_map = {
 	.type = BPF_MAP_TYPE_HASH,
@@ -34,6 +35,19 @@ struct bpf_map_def SEC("maps") nocommon_lru_hash_map = {
 	.map_flags = BPF_F_NO_COMMON_LRU,
 };
 
+struct bpf_map_def SEC("maps") inner_lru_hash_map = {
+	.type = BPF_MAP_TYPE_LRU_HASH,
+	.key_size = sizeof(u32),
+	.value_size = sizeof(long),
+	.max_entries = MAX_ENTRIES,
+};
+
+struct bpf_map_def SEC("maps") array_of_lru_hashs = {
+	.type = BPF_MAP_TYPE_ARRAY_OF_MAPS,
+	.key_size = sizeof(u32),
+	.max_entries = MAX_NR_CPUS,
+};
+
 struct bpf_map_def SEC("maps") percpu_hash_map = {
 	.type = BPF_MAP_TYPE_PERCPU_HASH,
 	.key_size = sizeof(u32),
@@ -154,13 +168,27 @@ int stress_lru_hmap_alloc(struct pt_regs *ctx)
 
 	test_case = dst6[7];
 
-	if (test_case == 0)
+	if (test_case == 0) {
 		ret = bpf_map_update_elem(&lru_hash_map, &key, &val, BPF_ANY);
-	else if (test_case == 1)
+	} else if (test_case == 1) {
 		ret = bpf_map_update_elem(&nocommon_lru_hash_map, &key, &val,
 					  BPF_ANY);
-	else
+	} else if (test_case == 2) {
+		void *nolocal_lru_map;
+		int cpu = bpf_get_smp_processor_id();
+
+		nolocal_lru_map = bpf_map_lookup_elem(&array_of_lru_hashs,
+						      &cpu);
+		if (!nolocal_lru_map) {
+			ret = -ENOENT;
+			goto done;
+		}
+
+		ret = bpf_map_update_elem(nolocal_lru_map, &key, &val,
+					  BPF_ANY);
+	} else {
 		ret = -EINVAL;
+	}
 
 done:
 	if (ret)