From aa5e2af660fc6e35b9518d68dd7e1bb736e9f7e7 Mon Sep 17 00:00:00 2001 From: "Daniel T. Lee" Date: Sat, 21 Mar 2020 19:04:24 +0900 Subject: [PATCH] samples, bpf: Refactor perf_event user program with libbpf bpf_link The bpf_program__attach of libbpf(using bpf_link) is much more intuitive than the previous method using ioctl. bpf_program__attach_perf_event manages the enable of perf_event and attach of BPF programs to it, so there's no neeed to do this directly with ioctl. In addition, bpf_link provides consistency in the use of API because it allows disable (detach, destroy) for multiple events to be treated as one bpf_link__destroy. Also, bpf_link__destroy manages the close() of perf_event fd. This commit refactors samples that attach the bpf program to perf_event by using libbbpf instead of ioctl. Also the bpf_load in the samples were removed and migrated to use libbbpf API. Signed-off-by: Daniel T. Lee Signed-off-by: Daniel Borkmann Acked-by: Andrii Nakryiko Link: https://lore.kernel.org/bpf/20200321100424.1593964-3-danieltimlee@gmail.com --- samples/bpf/Makefile | 4 +- samples/bpf/sampleip_user.c | 98 +++++++++++++++++++---------- samples/bpf/trace_event_user.c | 139 +++++++++++++++++++++++++++-------------- 3 files changed, 159 insertions(+), 82 deletions(-) diff --git a/samples/bpf/Makefile b/samples/bpf/Makefile index ff00614..424f6fe 100644 --- a/samples/bpf/Makefile +++ b/samples/bpf/Makefile @@ -88,8 +88,8 @@ xdp2-objs := xdp1_user.o xdp_router_ipv4-objs := xdp_router_ipv4_user.o test_current_task_under_cgroup-objs := bpf_load.o $(CGROUP_HELPERS) \ test_current_task_under_cgroup_user.o -trace_event-objs := bpf_load.o trace_event_user.o $(TRACE_HELPERS) -sampleip-objs := bpf_load.o sampleip_user.o $(TRACE_HELPERS) +trace_event-objs := trace_event_user.o $(TRACE_HELPERS) +sampleip-objs := sampleip_user.o $(TRACE_HELPERS) tc_l2_redirect-objs := bpf_load.o tc_l2_redirect_user.o lwt_len_hist-objs := bpf_load.o lwt_len_hist_user.o xdp_tx_iptunnel-objs := xdp_tx_iptunnel_user.o diff --git a/samples/bpf/sampleip_user.c b/samples/bpf/sampleip_user.c index b0f115f..4372d2d 100644 --- a/samples/bpf/sampleip_user.c +++ b/samples/bpf/sampleip_user.c @@ -10,21 +10,23 @@ #include #include #include -#include #include #include #include -#include +#include #include -#include "bpf_load.h" #include "perf-sys.h" #include "trace_helpers.h" +#define __must_check +#include + #define DEFAULT_FREQ 99 #define DEFAULT_SECS 5 #define MAX_IPS 8192 #define PAGE_OFFSET 0xffff880000000000 +static int map_fd; static int nr_cpus; static void usage(void) @@ -34,9 +36,10 @@ static void usage(void) printf(" duration # sampling duration (seconds), default 5\n"); } -static int sampling_start(int *pmu_fd, int freq) +static int sampling_start(int freq, struct bpf_program *prog, + struct bpf_link *links[]) { - int i; + int i, pmu_fd; struct perf_event_attr pe_sample_attr = { .type = PERF_TYPE_SOFTWARE, @@ -47,26 +50,30 @@ static int sampling_start(int *pmu_fd, int freq) }; for (i = 0; i < nr_cpus; i++) { - pmu_fd[i] = sys_perf_event_open(&pe_sample_attr, -1 /* pid */, i, + pmu_fd = sys_perf_event_open(&pe_sample_attr, -1 /* pid */, i, -1 /* group_fd */, 0 /* flags */); - if (pmu_fd[i] < 0) { + if (pmu_fd < 0) { fprintf(stderr, "ERROR: Initializing perf sampling\n"); return 1; } - assert(ioctl(pmu_fd[i], PERF_EVENT_IOC_SET_BPF, - prog_fd[0]) == 0); - assert(ioctl(pmu_fd[i], PERF_EVENT_IOC_ENABLE, 0) == 0); + links[i] = bpf_program__attach_perf_event(prog, pmu_fd); + if (IS_ERR(links[i])) { + fprintf(stderr, "ERROR: Attach perf event\n"); + links[i] = NULL; + close(pmu_fd); + return 1; + } } return 0; } -static void sampling_end(int *pmu_fd) +static void sampling_end(struct bpf_link *links[]) { int i; for (i = 0; i < nr_cpus; i++) - close(pmu_fd[i]); + bpf_link__destroy(links[i]); } struct ipcount { @@ -128,14 +135,17 @@ static void print_ip_map(int fd) static void int_exit(int sig) { printf("\n"); - print_ip_map(map_fd[0]); + print_ip_map(map_fd); exit(0); } int main(int argc, char **argv) { + int opt, freq = DEFAULT_FREQ, secs = DEFAULT_SECS, error = 1; + struct bpf_object *obj = NULL; + struct bpf_program *prog; + struct bpf_link **links; char filename[256]; - int *pmu_fd, opt, freq = DEFAULT_FREQ, secs = DEFAULT_SECS; /* process arguments */ while ((opt = getopt(argc, argv, "F:h")) != -1) { @@ -163,38 +173,58 @@ int main(int argc, char **argv) } /* create perf FDs for each CPU */ - nr_cpus = sysconf(_SC_NPROCESSORS_CONF); - pmu_fd = malloc(nr_cpus * sizeof(int)); - if (pmu_fd == NULL) { - fprintf(stderr, "ERROR: malloc of pmu_fd\n"); - return 1; + nr_cpus = sysconf(_SC_NPROCESSORS_ONLN); + links = calloc(nr_cpus, sizeof(struct bpf_link *)); + if (!links) { + fprintf(stderr, "ERROR: malloc of links\n"); + goto cleanup; } - /* load BPF program */ snprintf(filename, sizeof(filename), "%s_kern.o", argv[0]); - if (load_bpf_file(filename)) { - fprintf(stderr, "ERROR: loading BPF program (errno %d):\n", - errno); - if (strcmp(bpf_log_buf, "") == 0) - fprintf(stderr, "Try: ulimit -l unlimited\n"); - else - fprintf(stderr, "%s", bpf_log_buf); - return 1; + obj = bpf_object__open_file(filename, NULL); + if (IS_ERR(obj)) { + fprintf(stderr, "ERROR: opening BPF object file failed\n"); + obj = NULL; + goto cleanup; + } + + prog = bpf_object__find_program_by_name(obj, "do_sample"); + if (!prog) { + fprintf(stderr, "ERROR: finding a prog in obj file failed\n"); + goto cleanup; } + + /* load BPF program */ + if (bpf_object__load(obj)) { + fprintf(stderr, "ERROR: loading BPF object file failed\n"); + goto cleanup; + } + + map_fd = bpf_object__find_map_fd_by_name(obj, "ip_map"); + if (map_fd < 0) { + fprintf(stderr, "ERROR: finding a map in obj file failed\n"); + goto cleanup; + } + signal(SIGINT, int_exit); signal(SIGTERM, int_exit); /* do sampling */ printf("Sampling at %d Hertz for %d seconds. Ctrl-C also ends.\n", freq, secs); - if (sampling_start(pmu_fd, freq) != 0) - return 1; + if (sampling_start(freq, prog, links) != 0) + goto cleanup; + sleep(secs); - sampling_end(pmu_fd); - free(pmu_fd); + error = 0; +cleanup: + sampling_end(links); /* output sample counts */ - print_ip_map(map_fd[0]); + if (!error) + print_ip_map(map_fd); - return 0; + free(links); + bpf_object__close(obj); + return error; } diff --git a/samples/bpf/trace_event_user.c b/samples/bpf/trace_event_user.c index 356171b..b6cd358 100644 --- a/samples/bpf/trace_event_user.c +++ b/samples/bpf/trace_event_user.c @@ -6,22 +6,25 @@ #include #include #include -#include -#include -#include #include #include #include -#include #include #include +#include #include -#include "bpf_load.h" #include "perf-sys.h" #include "trace_helpers.h" +#define __must_check +#include + #define SAMPLE_FREQ 50 +static int pid; +/* counts, stackmap */ +static int map_fd[2]; +struct bpf_program *prog; static bool sys_read_seen, sys_write_seen; static void print_ksym(__u64 addr) @@ -91,10 +94,10 @@ static void print_stack(struct key_t *key, __u64 count) } } -static void int_exit(int sig) +static void err_exit(int err) { - kill(0, SIGKILL); - exit(0); + kill(pid, SIGKILL); + exit(err); } static void print_stacks(void) @@ -102,7 +105,7 @@ static void print_stacks(void) struct key_t key = {}, next_key; __u64 value; __u32 stackid = 0, next_id; - int fd = map_fd[0], stack_map = map_fd[1]; + int error = 1, fd = map_fd[0], stack_map = map_fd[1]; sys_read_seen = sys_write_seen = false; while (bpf_map_get_next_key(fd, &key, &next_key) == 0) { @@ -114,7 +117,7 @@ static void print_stacks(void) printf("\n"); if (!sys_read_seen || !sys_write_seen) { printf("BUG kernel stack doesn't contain sys_read() and sys_write()\n"); - int_exit(0); + err_exit(error); } /* clear stack map */ @@ -136,43 +139,52 @@ static inline int generate_load(void) static void test_perf_event_all_cpu(struct perf_event_attr *attr) { - int nr_cpus = sysconf(_SC_NPROCESSORS_CONF); - int *pmu_fd = malloc(nr_cpus * sizeof(int)); - int i, error = 0; + int nr_cpus = sysconf(_SC_NPROCESSORS_ONLN); + struct bpf_link **links = calloc(nr_cpus, sizeof(struct bpf_link *)); + int i, pmu_fd, error = 1; + + if (!links) { + printf("malloc of links failed\n"); + goto err; + } /* system wide perf event, no need to inherit */ attr->inherit = 0; /* open perf_event on all cpus */ for (i = 0; i < nr_cpus; i++) { - pmu_fd[i] = sys_perf_event_open(attr, -1, i, -1, 0); - if (pmu_fd[i] < 0) { + pmu_fd = sys_perf_event_open(attr, -1, i, -1, 0); + if (pmu_fd < 0) { printf("sys_perf_event_open failed\n"); - error = 1; goto all_cpu_err; } - assert(ioctl(pmu_fd[i], PERF_EVENT_IOC_SET_BPF, prog_fd[0]) == 0); - assert(ioctl(pmu_fd[i], PERF_EVENT_IOC_ENABLE) == 0); + links[i] = bpf_program__attach_perf_event(prog, pmu_fd); + if (IS_ERR(links[i])) { + printf("bpf_program__attach_perf_event failed\n"); + links[i] = NULL; + close(pmu_fd); + goto all_cpu_err; + } } - if (generate_load() < 0) { - error = 1; + if (generate_load() < 0) goto all_cpu_err; - } + print_stacks(); + error = 0; all_cpu_err: - for (i--; i >= 0; i--) { - ioctl(pmu_fd[i], PERF_EVENT_IOC_DISABLE); - close(pmu_fd[i]); - } - free(pmu_fd); + for (i--; i >= 0; i--) + bpf_link__destroy(links[i]); +err: + free(links); if (error) - int_exit(0); + err_exit(error); } static void test_perf_event_task(struct perf_event_attr *attr) { - int pmu_fd, error = 0; + struct bpf_link *link = NULL; + int pmu_fd, error = 1; /* per task perf event, enable inherit so the "dd ..." command can be traced properly. * Enabling inherit will cause bpf_perf_prog_read_time helper failure. @@ -183,21 +195,25 @@ static void test_perf_event_task(struct perf_event_attr *attr) pmu_fd = sys_perf_event_open(attr, 0, -1, -1, 0); if (pmu_fd < 0) { printf("sys_perf_event_open failed\n"); - int_exit(0); + goto err; } - assert(ioctl(pmu_fd, PERF_EVENT_IOC_SET_BPF, prog_fd[0]) == 0); - assert(ioctl(pmu_fd, PERF_EVENT_IOC_ENABLE) == 0); - - if (generate_load() < 0) { - error = 1; + link = bpf_program__attach_perf_event(prog, pmu_fd); + if (IS_ERR(link)) { + printf("bpf_program__attach_perf_event failed\n"); + link = NULL; + close(pmu_fd); goto err; } + + if (generate_load() < 0) + goto err; + print_stacks(); + error = 0; err: - ioctl(pmu_fd, PERF_EVENT_IOC_DISABLE); - close(pmu_fd); + bpf_link__destroy(link); if (error) - int_exit(0); + err_exit(error); } static void test_bpf_perf_event(void) @@ -282,29 +298,60 @@ static void test_bpf_perf_event(void) int main(int argc, char **argv) { struct rlimit r = {RLIM_INFINITY, RLIM_INFINITY}; + struct bpf_object *obj = NULL; char filename[256]; + int error = 1; snprintf(filename, sizeof(filename), "%s_kern.o", argv[0]); setrlimit(RLIMIT_MEMLOCK, &r); - signal(SIGINT, int_exit); - signal(SIGTERM, int_exit); + signal(SIGINT, err_exit); + signal(SIGTERM, err_exit); if (load_kallsyms()) { printf("failed to process /proc/kallsyms\n"); - return 1; + goto cleanup; + } + + obj = bpf_object__open_file(filename, NULL); + if (IS_ERR(obj)) { + printf("opening BPF object file failed\n"); + obj = NULL; + goto cleanup; + } + + prog = bpf_object__find_program_by_name(obj, "bpf_prog1"); + if (!prog) { + printf("finding a prog in obj file failed\n"); + goto cleanup; + } + + /* load BPF program */ + if (bpf_object__load(obj)) { + printf("loading BPF object file failed\n"); + goto cleanup; } - if (load_bpf_file(filename)) { - printf("%s", bpf_log_buf); - return 2; + map_fd[0] = bpf_object__find_map_fd_by_name(obj, "counts"); + map_fd[1] = bpf_object__find_map_fd_by_name(obj, "stackmap"); + if (map_fd[0] < 0 || map_fd[1] < 0) { + printf("finding a counts/stackmap map in obj file failed\n"); + goto cleanup; } - if (fork() == 0) { + pid = fork(); + if (pid == 0) { read_trace_pipe(); return 0; + } else if (pid == -1) { + printf("couldn't spawn process\n"); + goto cleanup; } + test_bpf_perf_event(); - int_exit(0); - return 0; + error = 0; + +cleanup: + bpf_object__close(obj); + err_exit(error); } -- 2.7.4