Re: [PATCH bpf-next] selftests/bpf: fix trampoline_count clean up logic

On Fri, Feb 21, 2020 at 2:21 AM Jakub Sitnicki [off-list ref] wrote:

On Thu, Feb 20, 2020 at 11:05 PM GMT, Andrii Nakryiko wrote:

quoted

Libbpf's Travis CI tests caught this issue. Ensure bpf_link and bpf_object
clean up is performed correctly.

Fixes: d633d57902a5 ("selftest/bpf: Add test for allowed trampolines count")
Cc: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Andrii Nakryiko <redacted>
---
 .../bpf/prog_tests/trampoline_count.c         | 25 +++++++++++++------
 1 file changed, 18 insertions(+), 7 deletions(-)

diff --git a/tools/testing/selftests/bpf/prog_tests/trampoline_count.c b/tools/testing/selftests/bpf/prog_tests/trampoline_count.c
index 1f6ccdaed1ac..781c8d11604b 100644
--- a/tools/testing/selftests/bpf/prog_tests/trampoline_count.c
+++ b/tools/testing/selftests/bpf/prog_tests/trampoline_count.c

@@ -55,31 +55,40 @@ void test_trampoline_count(void)
      /* attach 'allowed' 40 trampoline programs */
      for (i = 0; i < MAX_TRAMP_PROGS; i++) {
              obj = bpf_object__open_file(object, NULL);
-             if (CHECK(IS_ERR(obj), "obj_open_file", "err %ld\n", PTR_ERR(obj)))
+             if (CHECK(IS_ERR(obj), "obj_open_file", "err %ld\n", PTR_ERR(obj))) {
+                     obj = NULL;
                      goto cleanup;
+             }

              err = bpf_object__load(obj);
              if (CHECK(err, "obj_load", "err %d\n", err))
                      goto cleanup;
              inst[i].obj = obj;
+             obj = NULL;

              if (rand() % 2) {
-                     link = load(obj, fentry_name);
-                     if (CHECK(IS_ERR(link), "attach prog", "err %ld\n", PTR_ERR(link)))
+                     link = load(inst[i].obj, fentry_name);
+                     if (CHECK(IS_ERR(link), "attach prog", "err %ld\n", PTR_ERR(link))) {
+                             link = NULL;
                              goto cleanup;
+                     }
                      inst[i].link_fentry = link;
              } else {
-                     link = load(obj, fexit_name);
-                     if (CHECK(IS_ERR(link), "attach prog", "err %ld\n", PTR_ERR(link)))
+                     link = load(inst[i].obj, fexit_name);
+                     if (CHECK(IS_ERR(link), "attach prog", "err %ld\n", PTR_ERR(link))) {
+                             link = NULL;
                              goto cleanup;
+                     }
                      inst[i].link_fexit = link;
              }
      }

      /* and try 1 extra.. */
      obj = bpf_object__open_file(object, NULL);
-     if (CHECK(IS_ERR(obj), "obj_open_file", "err %ld\n", PTR_ERR(obj)))
+     if (CHECK(IS_ERR(obj), "obj_open_file", "err %ld\n", PTR_ERR(obj))) {
+             obj = NULL;
              goto cleanup;
+     }

      err = bpf_object__load(obj);
      if (CHECK(err, "obj_load", "err %d\n", err))

@@ -104,7 +113,9 @@ void test_trampoline_count(void)
 cleanup_extra:
      bpf_object__close(obj);
 cleanup:
-     while (--i) {
+     if (i >= MAX_TRAMP_PROGS)
+             i = MAX_TRAMP_PROGS - 1;
+     for (; i >= 0; i--) {
              bpf_link__destroy(inst[i].link_fentry);
              bpf_link__destroy(inst[i].link_fexit);
              bpf_object__close(inst[i].obj);

I'm not sure I'm grasping what the fix is about.

We don't access obj or link from cleanup, so what is the point of
setting them to NULL before jumping there?

Or is it all just an ancillary change to clamping the loop index
variable to (MAX_TRAMP_PROGS - 1)?

Yeah, mostly ancillary. But this is not just clamping to
MAX_TRAMP_PROGS-1. As Jiri pointed out, it's also handling a case of i
== 0.