bpf: fix return in load_bpf_file
authorLawrence Brakmo <brakmo@fb.com>
Tue, 4 Jul 2017 22:57:50 +0000 (15:57 -0700)
committerDavid S. Miller <davem@davemloft.net>
Wed, 5 Jul 2017 08:05:28 +0000 (09:05 +0100)
The function load_bpf_file ignores the return value of
load_and_attach(), so even if load_and_attach() returns an error,
load_bpf_file() will return 0.

Now, load_bpf_file() can call load_and_attach() multiple times and some
can succeed and some could fail. I think the correct behavor is to
return error on the first failed load_and_attach().

v2: Added missing SOB

Signed-off-by: Lawrence Brakmo <brakmo@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
samples/bpf/bpf_load.c

index a4be7cf..899f403 100644 (file)
@@ -567,8 +567,12 @@ static int do_load_bpf_file(const char *path, fixup_map_cb fixup_map)
                    memcmp(shname, "perf_event", 10) == 0 ||
                    memcmp(shname, "socket", 6) == 0 ||
                    memcmp(shname, "cgroup/", 7) == 0 ||
-                   memcmp(shname, "sockops", 7) == 0)
-                       load_and_attach(shname, data->d_buf, data->d_size);
+                   memcmp(shname, "sockops", 7) == 0) {
+                       ret = load_and_attach(shname, data->d_buf,
+                                             data->d_size);
+                       if (ret != 0)
+                               goto done;
+               }
        }
 
        ret = 0;