diff mbox

[net-next] bpf: fix return in load_bpf_file

Message ID 20170704215129.4047623-1-brakmo@fb.com
State Superseded, archived
Delegated to: David Miller
Headers show

Commit Message

Lawrence Brakmo July 4, 2017, 9:51 p.m. UTC 
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().
---
 samples/bpf/bpf_load.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

Comments

Daniel Borkmann July 4, 2017, 10:14 p.m. UTC | #1 
On 07/04/2017 11:51 PM, Lawrence Brakmo wrote:
> 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().

Your SOB is missing.
Lawrence Brakmo July 4, 2017, 10:55 p.m. UTC | #2 
On 7/4/17, 3:14 PM, "Daniel Borkmann" <daniel@iogearbox.net> wrote:

    On 07/04/2017 11:51 PM, Lawrence Brakmo wrote:
    > 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().

    
    Your SOB is missing.

Oops, thanks. New patch forthcoming.
diff mbox

Patch

diff --git a/samples/bpf/bpf_load.c b/samples/bpf/bpf_load.c
index a4be7cf..899f403 100644
--- a/samples/bpf/bpf_load.c
+++ b/samples/bpf/bpf_load.c
@@ -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;