From 3a5795b83d578cc542a92c94399946258cf1a2af Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Fri, 14 Apr 2017 10:30:30 -0700 Subject: 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 Signed-off-by: David S. Miller --- samples/bpf/map_perf_test_kern.c | 34 ++++++++++++++++++++-- samples/bpf/map_perf_test_user.c | 62 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 93 insertions(+), 3 deletions(-) (limited to 'samples') 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) diff --git a/samples/bpf/map_perf_test_user.c b/samples/bpf/map_perf_test_user.c index 2a12f48b5c6d..6ac778153315 100644 --- a/samples/bpf/map_perf_test_user.c +++ b/samples/bpf/map_perf_test_user.c @@ -25,6 +25,7 @@ #include "bpf_load.h" #define TEST_BIT(t) (1U << (t)) +#define MAX_NR_CPUS 1024 static __u64 time_get_ns(void) { @@ -44,6 +45,7 @@ enum test_type { LPM_KMALLOC, HASH_LOOKUP, ARRAY_LOOKUP, + INNER_LRU_HASH_PREALLOC, NR_TESTS, }; @@ -57,10 +59,14 @@ const char *test_map_names[NR_TESTS] = { [LPM_KMALLOC] = "lpm_trie_map_alloc", [HASH_LOOKUP] = "hash_map", [ARRAY_LOOKUP] = "array_map", + [INNER_LRU_HASH_PREALLOC] = "inner_lru_hash_map", }; static int test_flags = ~0; static uint32_t num_map_entries; +static uint32_t inner_lru_hash_size; +static int inner_lru_hash_idx = -1; +static int array_of_lru_hashs_idx = -1; static uint32_t max_cnt = 1000000; static int check_test_flags(enum test_type t) @@ -82,11 +88,42 @@ static void test_hash_prealloc(int cpu) static void do_test_lru(enum test_type test, int cpu) { + static int inner_lru_map_fds[MAX_NR_CPUS]; + struct sockaddr_in6 in6 = { .sin6_family = AF_INET6 }; const char *test_name; __u64 start_time; int i, ret; + if (test == INNER_LRU_HASH_PREALLOC) { + int outer_fd = map_fd[array_of_lru_hashs_idx]; + + assert(cpu < MAX_NR_CPUS); + + if (cpu) { + inner_lru_map_fds[cpu] = + bpf_create_map(BPF_MAP_TYPE_LRU_HASH, + sizeof(uint32_t), sizeof(long), + inner_lru_hash_size, 0); + if (inner_lru_map_fds[cpu] == -1) { + printf("cannot create BPF_MAP_TYPE_LRU_HASH %s(%d)\n", + strerror(errno), errno); + exit(1); + } + } else { + inner_lru_map_fds[cpu] = map_fd[inner_lru_hash_idx]; + } + + ret = bpf_map_update_elem(outer_fd, &cpu, + &inner_lru_map_fds[cpu], + BPF_ANY); + if (ret) { + printf("cannot update ARRAY_OF_LRU_HASHS with key:%u. %s(%d)\n", + cpu, strerror(errno), errno); + exit(1); + } + } + in6.sin6_addr.s6_addr16[0] = 0xdead; in6.sin6_addr.s6_addr16[1] = 0xbeef; @@ -96,6 +133,9 @@ static void do_test_lru(enum test_type test, int cpu) } else if (test == NOCOMMON_LRU_HASH_PREALLOC) { test_name = "nocommon_lru_hash_map_perf"; in6.sin6_addr.s6_addr16[7] = 1; + } else if (test == INNER_LRU_HASH_PREALLOC) { + test_name = "inner_lru_hash_map_perf"; + in6.sin6_addr.s6_addr16[7] = 2; } else { assert(0); } @@ -120,6 +160,11 @@ static void test_nocommon_lru_hash_prealloc(int cpu) do_test_lru(NOCOMMON_LRU_HASH_PREALLOC, cpu); } +static void test_inner_lru_hash_prealloc(int cpu) +{ + do_test_lru(INNER_LRU_HASH_PREALLOC, cpu); +} + static void test_percpu_hash_prealloc(int cpu) { __u64 start_time; @@ -203,6 +248,7 @@ const test_func test_funcs[] = { [LPM_KMALLOC] = test_lpm_kmalloc, [HASH_LOOKUP] = test_hash_lookup, [ARRAY_LOOKUP] = test_array_lookup, + [INNER_LRU_HASH_PREALLOC] = test_inner_lru_hash_prealloc, }; static void loop(int cpu) @@ -278,9 +324,25 @@ static void fixup_map(struct bpf_map_def *map, const char *name, int idx) { int i; + if (!strcmp("inner_lru_hash_map", name)) { + inner_lru_hash_idx = idx; + inner_lru_hash_size = map->max_entries; + } + + if (!strcmp("array_of_lru_hashs", name)) { + if (inner_lru_hash_idx == -1) { + printf("inner_lru_hash_map must be defined before array_of_lru_hashs\n"); + exit(1); + } + map->inner_map_idx = inner_lru_hash_idx; + array_of_lru_hashs_idx = idx; + } + if (num_map_entries <= 0) return; + inner_lru_hash_size = num_map_entries; + /* Only change the max_entries for the enabled test(s) */ for (i = 0; i < NR_TESTS; i++) { if (!strcmp(test_map_names[i], name) && -- cgit