From 9fd63d05f3e8476282cd8c484eb34d3f6be54f40 Mon Sep 17 00:00:00 2001 From: Martin KaFai Lau Date: Fri, 14 Apr 2017 10:30:28 -0700 Subject: bpf: Allow bpf sample programs (*_user.c) to change bpf_map_def The current bpf_map_def is statically defined during compile time. This patch allows the *_user.c program to change it during runtime. It is done by adding load_bpf_file_fixup_map() which takes a callback. The callback will be called before creating each map so that it has a chance to modify the bpf_map_def. The current usecase is to change max_entries in map_perf_test. It is interesting to test with a much bigger map size in some cases (e.g. the following patch on bpf_lru_map.c). However, it is hard to find one size to fit all testing environment. Hence, it is handy to take the max_entries as a cmdline arg and then configure the bpf_map_def during runtime. This patch adds two cmdline args. One is to configure the map's max_entries. Another is to configure the max_cnt which controls how many times a syscall is called. Signed-off-by: Martin KaFai Lau Acked-by: Alexei Starovoitov Acked-by: Daniel Borkmann Signed-off-by: David S. Miller --- samples/bpf/bpf_load.c | 114 +++++++++++++++++++++++++----- samples/bpf/bpf_load.h | 13 ++++ samples/bpf/map_perf_test_user.c | 148 ++++++++++++++++++++++++--------------- 3 files changed, 201 insertions(+), 74 deletions(-) (limited to 'samples') diff --git a/samples/bpf/bpf_load.c b/samples/bpf/bpf_load.c index dcdce1270d38..0d449d8032d1 100644 --- a/samples/bpf/bpf_load.c +++ b/samples/bpf/bpf_load.c @@ -21,6 +21,7 @@ #include #include #include +#include #include "libbpf.h" #include "bpf_load.h" #include "perf-sys.h" @@ -37,15 +38,6 @@ int event_fd[MAX_PROGS]; int prog_cnt; int prog_array_fd = -1; -struct bpf_map_def { - unsigned int type; - unsigned int key_size; - unsigned int value_size; - unsigned int max_entries; - unsigned int map_flags; - unsigned int inner_map_idx; -}; - static int populate_prog_array(const char *event, int prog_fd) { int ind = atoi(event), err; @@ -193,11 +185,14 @@ static int load_and_attach(const char *event, struct bpf_insn *prog, int size) return 0; } -static int load_maps(struct bpf_map_def *maps, int len) +static int load_maps(struct bpf_map_def *maps, int len, + const char **map_names, fixup_map_cb fixup_map) { int i; for (i = 0; i < len / sizeof(struct bpf_map_def); i++) { + if (fixup_map) + fixup_map(&maps[i], map_names[i], i); if (maps[i].type == BPF_MAP_TYPE_ARRAY_OF_MAPS || maps[i].type == BPF_MAP_TYPE_HASH_OF_MAPS) { @@ -280,14 +275,64 @@ static int parse_relo_and_apply(Elf_Data *data, Elf_Data *symbols, return 0; } -int load_bpf_file(char *path) +static int cmp_symbols(const void *l, const void *r) +{ + const GElf_Sym *lsym = (const GElf_Sym *)l; + const GElf_Sym *rsym = (const GElf_Sym *)r; + + if (lsym->st_value < rsym->st_value) + return -1; + else if (lsym->st_value > rsym->st_value) + return 1; + else + return 0; +} + +static int get_sorted_map_names(Elf *elf, Elf_Data *symbols, int maps_shndx, + int strtabidx, char **map_names) +{ + GElf_Sym map_symbols[MAX_MAPS]; + int i, nr_maps = 0; + + for (i = 0; i < symbols->d_size / sizeof(GElf_Sym); i++) { + assert(nr_maps < MAX_MAPS); + if (!gelf_getsym(symbols, i, &map_symbols[nr_maps])) + continue; + if (map_symbols[nr_maps].st_shndx != maps_shndx) + continue; + nr_maps++; + } + + qsort(map_symbols, nr_maps, sizeof(GElf_Sym), cmp_symbols); + + for (i = 0; i < nr_maps; i++) { + char *map_name; + + map_name = elf_strptr(elf, strtabidx, map_symbols[i].st_name); + if (!map_name) { + printf("cannot get map symbol\n"); + return 1; + } + + map_names[i] = strdup(map_name); + if (!map_names[i]) { + printf("strdup(%s): %s(%d)\n", map_name, + strerror(errno), errno); + return 1; + } + } + + return 0; +} + +static int do_load_bpf_file(const char *path, fixup_map_cb fixup_map) { - int fd, i; + int fd, i, ret, maps_shndx = -1, strtabidx = -1; Elf *elf; GElf_Ehdr ehdr; GElf_Shdr shdr, shdr_prog; - Elf_Data *data, *data_prog, *symbols = NULL; - char *shname, *shname_prog; + Elf_Data *data, *data_prog, *data_maps = NULL, *symbols = NULL; + char *shname, *shname_prog, *map_names[MAX_MAPS] = { NULL }; /* reset global variables */ kern_version = 0; @@ -335,14 +380,33 @@ int load_bpf_file(char *path) } memcpy(&kern_version, data->d_buf, sizeof(int)); } else if (strcmp(shname, "maps") == 0) { - processed_sec[i] = true; - if (load_maps(data->d_buf, data->d_size)) - return 1; + maps_shndx = i; + data_maps = data; } else if (shdr.sh_type == SHT_SYMTAB) { + strtabidx = shdr.sh_link; symbols = data; } } + ret = 1; + + if (!symbols) { + printf("missing SHT_SYMTAB section\n"); + goto done; + } + + if (data_maps) { + if (get_sorted_map_names(elf, symbols, maps_shndx, strtabidx, + map_names)) + goto done; + + if (load_maps(data_maps->d_buf, data_maps->d_size, + (const char **)map_names, fixup_map)) + goto done; + + processed_sec[maps_shndx] = true; + } + /* load programs that need map fixup (relocations) */ for (i = 1; i < ehdr.e_shnum; i++) { if (processed_sec[i]) @@ -399,8 +463,22 @@ int load_bpf_file(char *path) load_and_attach(shname, data->d_buf, data->d_size); } + ret = 0; +done: + for (i = 0; i < MAX_MAPS; i++) + free(map_names[i]); close(fd); - return 0; + return ret; +} + +int load_bpf_file(char *path) +{ + return do_load_bpf_file(path, NULL); +} + +int load_bpf_file_fixup_map(const char *path, fixup_map_cb fixup_map) +{ + return do_load_bpf_file(path, fixup_map); } void read_trace_pipe(void) diff --git a/samples/bpf/bpf_load.h b/samples/bpf/bpf_load.h index c827827299b3..68f6b2d22507 100644 --- a/samples/bpf/bpf_load.h +++ b/samples/bpf/bpf_load.h @@ -6,6 +6,18 @@ #define MAX_MAPS 32 #define MAX_PROGS 32 +struct bpf_map_def { + unsigned int type; + unsigned int key_size; + unsigned int value_size; + unsigned int max_entries; + unsigned int map_flags; + unsigned int inner_map_idx; +}; + +typedef void (*fixup_map_cb)(struct bpf_map_def *map, const char *map_name, + int idx); + extern int map_fd[MAX_MAPS]; extern int prog_fd[MAX_PROGS]; extern int event_fd[MAX_PROGS]; @@ -25,6 +37,7 @@ extern int prog_cnt; * returns zero on success */ int load_bpf_file(char *path); +int load_bpf_file_fixup_map(const char *path, fixup_map_cb fixup_map); void read_trace_pipe(void); struct ksym { diff --git a/samples/bpf/map_perf_test_user.c b/samples/bpf/map_perf_test_user.c index 51cb8f238aa2..2a12f48b5c6d 100644 --- a/samples/bpf/map_perf_test_user.c +++ b/samples/bpf/map_perf_test_user.c @@ -24,7 +24,7 @@ #include "libbpf.h" #include "bpf_load.h" -#define MAX_CNT 1000000 +#define TEST_BIT(t) (1U << (t)) static __u64 time_get_ns(void) { @@ -34,17 +34,39 @@ static __u64 time_get_ns(void) return ts.tv_sec * 1000000000ull + ts.tv_nsec; } -#define HASH_PREALLOC (1 << 0) -#define PERCPU_HASH_PREALLOC (1 << 1) -#define HASH_KMALLOC (1 << 2) -#define PERCPU_HASH_KMALLOC (1 << 3) -#define LRU_HASH_PREALLOC (1 << 4) -#define NOCOMMON_LRU_HASH_PREALLOC (1 << 5) -#define LPM_KMALLOC (1 << 6) -#define HASH_LOOKUP (1 << 7) -#define ARRAY_LOOKUP (1 << 8) +enum test_type { + HASH_PREALLOC, + PERCPU_HASH_PREALLOC, + HASH_KMALLOC, + PERCPU_HASH_KMALLOC, + LRU_HASH_PREALLOC, + NOCOMMON_LRU_HASH_PREALLOC, + LPM_KMALLOC, + HASH_LOOKUP, + ARRAY_LOOKUP, + NR_TESTS, +}; + +const char *test_map_names[NR_TESTS] = { + [HASH_PREALLOC] = "hash_map", + [PERCPU_HASH_PREALLOC] = "percpu_hash_map", + [HASH_KMALLOC] = "hash_map_alloc", + [PERCPU_HASH_KMALLOC] = "percpu_hash_map_alloc", + [LRU_HASH_PREALLOC] = "lru_hash_map", + [NOCOMMON_LRU_HASH_PREALLOC] = "nocommon_lru_hash_map", + [LPM_KMALLOC] = "lpm_trie_map_alloc", + [HASH_LOOKUP] = "hash_map", + [ARRAY_LOOKUP] = "array_map", +}; static int test_flags = ~0; +static uint32_t num_map_entries; +static uint32_t max_cnt = 1000000; + +static int check_test_flags(enum test_type t) +{ + return test_flags & TEST_BIT(t); +} static void test_hash_prealloc(int cpu) { @@ -52,13 +74,13 @@ static void test_hash_prealloc(int cpu) int i; start_time = time_get_ns(); - for (i = 0; i < MAX_CNT; i++) + for (i = 0; i < max_cnt; i++) syscall(__NR_getuid); printf("%d:hash_map_perf pre-alloc %lld events per sec\n", - cpu, MAX_CNT * 1000000000ll / (time_get_ns() - start_time)); + cpu, max_cnt * 1000000000ll / (time_get_ns() - start_time)); } -static void do_test_lru(int lru_test_flag, int cpu) +static void do_test_lru(enum test_type test, int cpu) { struct sockaddr_in6 in6 = { .sin6_family = AF_INET6 }; const char *test_name; @@ -68,10 +90,10 @@ static void do_test_lru(int lru_test_flag, int cpu) in6.sin6_addr.s6_addr16[0] = 0xdead; in6.sin6_addr.s6_addr16[1] = 0xbeef; - if (lru_test_flag & LRU_HASH_PREALLOC) { + if (test == LRU_HASH_PREALLOC) { test_name = "lru_hash_map_perf"; in6.sin6_addr.s6_addr16[7] = 0; - } else if (lru_test_flag & NOCOMMON_LRU_HASH_PREALLOC) { + } else if (test == NOCOMMON_LRU_HASH_PREALLOC) { test_name = "nocommon_lru_hash_map_perf"; in6.sin6_addr.s6_addr16[7] = 1; } else { @@ -79,13 +101,13 @@ static void do_test_lru(int lru_test_flag, int cpu) } start_time = time_get_ns(); - for (i = 0; i < MAX_CNT; i++) { + for (i = 0; i < max_cnt; i++) { ret = connect(-1, (const struct sockaddr *)&in6, sizeof(in6)); assert(ret == -1 && errno == EBADF); } printf("%d:%s pre-alloc %lld events per sec\n", cpu, test_name, - MAX_CNT * 1000000000ll / (time_get_ns() - start_time)); + max_cnt * 1000000000ll / (time_get_ns() - start_time)); } static void test_lru_hash_prealloc(int cpu) @@ -104,10 +126,10 @@ static void test_percpu_hash_prealloc(int cpu) int i; start_time = time_get_ns(); - for (i = 0; i < MAX_CNT; i++) + for (i = 0; i < max_cnt; i++) syscall(__NR_geteuid); printf("%d:percpu_hash_map_perf pre-alloc %lld events per sec\n", - cpu, MAX_CNT * 1000000000ll / (time_get_ns() - start_time)); + cpu, max_cnt * 1000000000ll / (time_get_ns() - start_time)); } static void test_hash_kmalloc(int cpu) @@ -116,10 +138,10 @@ static void test_hash_kmalloc(int cpu) int i; start_time = time_get_ns(); - for (i = 0; i < MAX_CNT; i++) + for (i = 0; i < max_cnt; i++) syscall(__NR_getgid); printf("%d:hash_map_perf kmalloc %lld events per sec\n", - cpu, MAX_CNT * 1000000000ll / (time_get_ns() - start_time)); + cpu, max_cnt * 1000000000ll / (time_get_ns() - start_time)); } static void test_percpu_hash_kmalloc(int cpu) @@ -128,10 +150,10 @@ static void test_percpu_hash_kmalloc(int cpu) int i; start_time = time_get_ns(); - for (i = 0; i < MAX_CNT; i++) + for (i = 0; i < max_cnt; i++) syscall(__NR_getegid); printf("%d:percpu_hash_map_perf kmalloc %lld events per sec\n", - cpu, MAX_CNT * 1000000000ll / (time_get_ns() - start_time)); + cpu, max_cnt * 1000000000ll / (time_get_ns() - start_time)); } static void test_lpm_kmalloc(int cpu) @@ -140,10 +162,10 @@ static void test_lpm_kmalloc(int cpu) int i; start_time = time_get_ns(); - for (i = 0; i < MAX_CNT; i++) + for (i = 0; i < max_cnt; i++) syscall(__NR_gettid); printf("%d:lpm_perf kmalloc %lld events per sec\n", - cpu, MAX_CNT * 1000000000ll / (time_get_ns() - start_time)); + cpu, max_cnt * 1000000000ll / (time_get_ns() - start_time)); } static void test_hash_lookup(int cpu) @@ -152,10 +174,10 @@ static void test_hash_lookup(int cpu) int i; start_time = time_get_ns(); - for (i = 0; i < MAX_CNT; i++) + for (i = 0; i < max_cnt; i++) syscall(__NR_getpgid, 0); printf("%d:hash_lookup %lld lookups per sec\n", - cpu, MAX_CNT * 1000000000ll * 64 / (time_get_ns() - start_time)); + cpu, max_cnt * 1000000000ll * 64 / (time_get_ns() - start_time)); } static void test_array_lookup(int cpu) @@ -164,46 +186,38 @@ static void test_array_lookup(int cpu) int i; start_time = time_get_ns(); - for (i = 0; i < MAX_CNT; i++) + for (i = 0; i < max_cnt; i++) syscall(__NR_getpgrp, 0); printf("%d:array_lookup %lld lookups per sec\n", - cpu, MAX_CNT * 1000000000ll * 64 / (time_get_ns() - start_time)); + cpu, max_cnt * 1000000000ll * 64 / (time_get_ns() - start_time)); } +typedef void (*test_func)(int cpu); +const test_func test_funcs[] = { + [HASH_PREALLOC] = test_hash_prealloc, + [PERCPU_HASH_PREALLOC] = test_percpu_hash_prealloc, + [HASH_KMALLOC] = test_hash_kmalloc, + [PERCPU_HASH_KMALLOC] = test_percpu_hash_kmalloc, + [LRU_HASH_PREALLOC] = test_lru_hash_prealloc, + [NOCOMMON_LRU_HASH_PREALLOC] = test_nocommon_lru_hash_prealloc, + [LPM_KMALLOC] = test_lpm_kmalloc, + [HASH_LOOKUP] = test_hash_lookup, + [ARRAY_LOOKUP] = test_array_lookup, +}; + static void loop(int cpu) { cpu_set_t cpuset; + int i; CPU_ZERO(&cpuset); CPU_SET(cpu, &cpuset); sched_setaffinity(0, sizeof(cpuset), &cpuset); - if (test_flags & HASH_PREALLOC) - test_hash_prealloc(cpu); - - if (test_flags & PERCPU_HASH_PREALLOC) - test_percpu_hash_prealloc(cpu); - - if (test_flags & HASH_KMALLOC) - test_hash_kmalloc(cpu); - - if (test_flags & PERCPU_HASH_KMALLOC) - test_percpu_hash_kmalloc(cpu); - - if (test_flags & LRU_HASH_PREALLOC) - test_lru_hash_prealloc(cpu); - - if (test_flags & NOCOMMON_LRU_HASH_PREALLOC) - test_nocommon_lru_hash_prealloc(cpu); - - if (test_flags & LPM_KMALLOC) - test_lpm_kmalloc(cpu); - - if (test_flags & HASH_LOOKUP) - test_hash_lookup(cpu); - - if (test_flags & ARRAY_LOOKUP) - test_array_lookup(cpu); + for (i = 0; i < NR_TESTS; i++) { + if (check_test_flags(i)) + test_funcs[i](cpu); + } } static void run_perf_test(int tasks) @@ -260,6 +274,22 @@ static void fill_lpm_trie(void) assert(!r); } +static void fixup_map(struct bpf_map_def *map, const char *name, int idx) +{ + int i; + + if (num_map_entries <= 0) + return; + + /* Only change the max_entries for the enabled test(s) */ + for (i = 0; i < NR_TESTS; i++) { + if (!strcmp(test_map_names[i], name) && + (check_test_flags(i))) { + map->max_entries = num_map_entries; + } + } +} + int main(int argc, char **argv) { struct rlimit r = {RLIM_INFINITY, RLIM_INFINITY}; @@ -275,7 +305,13 @@ int main(int argc, char **argv) if (argc > 2) num_cpu = atoi(argv[2]) ? : num_cpu; - if (load_bpf_file(filename)) { + if (argc > 3) + num_map_entries = atoi(argv[3]); + + if (argc > 4) + max_cnt = atoi(argv[4]); + + if (load_bpf_file_fixup_map(filename, fixup_map)) { printf("%s", bpf_log_buf); return 1; } -- cgit