tools/bpf: fix test_sockmap failure
authorYonghong Song <yhs@fb.com>
Thu, 21 Jun 2018 17:02:20 +0000 (10:02 -0700)
committerDaniel Borkmann <daniel@iogearbox.net>
Thu, 21 Jun 2018 22:31:42 +0000 (00:31 +0200)
On one of our production test machine, when running
bpf selftest test_sockmap, I got the following error:
  # sudo ./test_sockmap
  libbpf: failed to create map (name: 'sock_map'): Operation not permitted
  libbpf: failed to load object 'test_sockmap_kern.o'
  libbpf: Can't get the 0th fd from program sk_skb1: only -1 instances
  ......
  load_bpf_file: (-1) Operation not permitted
  ERROR: (-1) load bpf failed

The error is due to not-big-enough rlimit
  struct rlimit r = {10 * 1024 * 1024, RLIM_INFINITY};

The test already includes "bpf_rlimit.h", which sets current
and max rlimit to RLIM_INFINITY. Let us just use it.

Signed-off-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
tools/testing/selftests/bpf/test_sockmap.c

index 05c8cb7..9e78df2 100644 (file)
@@ -1413,18 +1413,12 @@ out:
 
 int main(int argc, char **argv)
 {
-       struct rlimit r = {10 * 1024 * 1024, RLIM_INFINITY};
        int iov_count = 1, length = 1024, rate = 1;
        struct sockmap_options options = {0};
        int opt, longindex, err, cg_fd = 0;
        char *bpf_file = BPF_SOCKMAP_FILENAME;
        int test = PING_PONG;
 
-       if (setrlimit(RLIMIT_MEMLOCK, &r)) {
-               perror("setrlimit(RLIMIT_MEMLOCK)");
-               return 1;
-       }
-
        if (argc < 2)
                return test_suite();