Re: [PATCH bpf-next v2] selftests/bpf:fix a resource leak

[Shuah Khan]

On 7/9/24 19:59, Zhu Jun wrote:

The requested resources should be closed before return
in main(), otherwise resource leak will occur

Signed-off-by: Zhu Jun <zhuj...@cmss.chinamobile.com>

How did you find this problem? Please give more details
on the tool used including the output from the tool.

---
Changes in v2:
  - check for cg_fd >= 0 and have just one out label

  tools/testing/selftests/bpf/test_sockmap.c | 11 +++++++----
  1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/tools/testing/selftests/bpf/test_sockmap.c 
b/tools/testing/selftests/bpf/test_sockmap.c
index a34e95040994..285a9a714666 100644
--- a/tools/testing/selftests/bpf/test_sockmap.c
+++ b/tools/testing/selftests/bpf/test_sockmap.c
@@ -2075,8 +2075,10 @@ int main(int argc, char **argv)

if (!cg_fd) {

                cg_fd = cgroup_setup_and_join(CG_PATH);
-               if (cg_fd < 0)
-                       return cg_fd;
+               if (cg_fd < 0) {
+                       err = cg_fd;
+                       goto out;
+               }

I don't this this improves the code - it makes it more complex.
If you want to improve this - add the same error message that
gets printed out for

case 'c':

in argument parsing.

fprintf(stderr, "ERROR: (%i) open cg path failed: %s\n",
                 cg_fd, optarg);

                cg_created = 1;
        }

@@ -2092,7 +2094,7 @@ int main(int argc, char **argv)

        if (err) {
                fprintf(stderr, "populate program: (%s) %s\n",
                        bpf_file, strerror(errno));
-               return 1;
+               goto out;

This looks good to me.

        }
        running = 1;

@@ -2109,7 +2111,8 @@ int main(int argc, char **argv)

                free(options.whitelist);
        if (options.blacklist)
                free(options.blacklist);
-       close(cg_fd);
+       if (cg_fd >= 0)

Why is this check necessary?

+               close(cg_fd);

        if (cg_created)
                cleanup_cgroup_environment();
        return err;

thanks,
-- Shuah