Skip to contentSkip to conversation

selftests/bpf: Fix RELEASE build failure with gcc14

PATCH Applied Yonghong Song 3 participants 4 messages Jun 17, 2025

Subsystems (3)
lore
Share & tools
lore

Yonghong Song

author reviewer
#0 lore raw

PATCH 1/1 selftests/bpf: Fix RELEASE build failure with gcc14

With gcc14, when building with RELEASE=1, I hit four below compilation failure:

Error 1:
In file included from test_loader.c:6:
test_loader.c: In function ‘run_subtest’: test_progs.h:194:17:
error: ‘retval’ may be used uninitialized in this function
[-Werror=maybe-uninitialized]

    194 |                 fprintf(stdout, ##format);           \
        |                 ^~~~~~~
  test_loader.c:958:13: note: ‘retval’ was declared here
    958 |         int retval, err, i;
        |             ^~~~~~

The uninitialized var 'retval' actaully could cause incorrect result.

Error 2:
In function ‘test_fd_array_cnt’:
prog_tests/fd_array.c:71:14: error: ‘btf_id’ may be used uninitialized in this

      function [-Werror=maybe-uninitialized]
     71 |         fd = bpf_btf_get_fd_by_id(id);
        |              ^~~~~~~~~~~~~~~~~~~~~~~~
  prog_tests/fd_array.c:302:15: note: ‘btf_id’ was declared here
    302 |         __u32 btf_id;
        |               ^~~~~~

Changing ASSERT_GE to ASSERT_EQ can fix the compilation error. Otherwise, there is no functionality change.

Error 3:
prog_tests/tailcalls.c: In function ‘test_tailcall_hierarchy_count’:
prog_tests/tailcalls.c:1402:23: error: ‘fentry_data_fd’ may be used uninitialized

      in this function [-Werror=maybe-uninitialized]
     1402 |                 err = bpf_map_lookup_elem(fentry_data_fd, &i, &val);
          |                       ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The code is correct. The change intends to slient gcc errors.

Error 4: (this error only happens on arm64)
In file included from prog_tests/log_buf.c:4:
prog_tests/log_buf.c: In function ‘bpf_prog_load_log_buf’:
./test_progs.h:390:22: error: ‘log_buf’ may be used uninitialized [-Werror=maybe-uninitialized]

    390 |         int ___err = libbpf_get_error(___res);             \
        |                      ^~~~~~~~~~~~~~~~~~~~~~~~
  prog_tests/log_buf.c:158:14: note: in expansion of macro ‘ASSERT_OK_PTR’
    158 |         if (!ASSERT_OK_PTR(log_buf, "log_buf_alloc"))
        |              ^~~~~~~~~~~~~
  In file included from selftests/bpf/tools/include/bpf/bpf.h:32,
                 from ./test_progs.h:36:
  selftests/bpf/tools/include/bpf/libbpf_legacy.h:113:17:
    note: by argument 1 of type ‘const void *’ to ‘libbpf_get_error’ declared here
    113 | LIBBPF_API long libbpf_get_error(const void *ptr);
        |                 ^~~~~~~~~~~~~~~~

Adding a pragma to disable maybe-uninitialized fixed the issue.

---
 tools/testing/selftests/bpf/prog_tests/fd_array.c  | 2 +-
 tools/testing/selftests/bpf/prog_tests/log_buf.c   | 4 ++++
 tools/testing/selftests/bpf/prog_tests/tailcalls.c | 2 +-
 tools/testing/selftests/bpf/test_loader.c          | 6 +++---
 4 files changed, 9 insertions(+), 5 deletions(-)

NOTE: I found these issues when I tried to compare usdt.test.o binaries
when building selftests with and without RELEASE=1.

tools/testing/selftests/bpf/prog_tests/fd_array.c +1 -1
index 9add890c2d37..241b2c8c6e0f 100644
@@ -312,7 +312,7 @@ static void check_fd_array_cnt__referenced_btfs(void)
 
 	/* btf should still exist when original file descriptor is closed */
 	err = get_btf_id_by_fd(extra_fds[0], &btf_id);
-	if (!ASSERT_GE(err, 0, "get_btf_id_by_fd"))
+	if (!ASSERT_EQ(err, 0, "get_btf_id_by_fd"))
 		goto cleanup;
 
 	Close(extra_fds[0]);
tools/testing/selftests/bpf/prog_tests/log_buf.c +4 -0
index 169ce689b97c..d6f14a232002 100644
@@ -7,6 +7,10 @@
 #include "test_log_buf.skel.h"
 #include "bpf_util.h"
 
+#if !defined(__clang__)
+#pragma GCC diagnostic ignored "-Wmaybe-uninitialized"
+#endif
+
 static size_t libbpf_log_pos;
 static char libbpf_log_buf[1024 * 1024];
 static bool libbpf_log_error;
tools/testing/selftests/bpf/prog_tests/tailcalls.c +1 -1
index 66a900327f91..0ab36503c3b2 100644
@@ -1195,7 +1195,7 @@ static void test_tailcall_hierarchy_count(const char *which, bool test_fentry,
 					  bool test_fexit,
 					  bool test_fentry_entry)
 {
-	int err, map_fd, prog_fd, main_data_fd, fentry_data_fd, fexit_data_fd, i, val;
+	int err, map_fd, prog_fd, main_data_fd, fentry_data_fd = 0, fexit_data_fd = 0, i, val;
 	struct bpf_object *obj = NULL, *fentry_obj = NULL, *fexit_obj = NULL;
 	struct bpf_link *fentry_link = NULL, *fexit_link = NULL;
 	struct bpf_program *prog, *fentry_prog;
tools/testing/selftests/bpf/test_loader.c +3 -3
index 9551d8d5f8f9..2c7e9729d5fe 100644
@@ -1103,9 +1103,9 @@ void run_subtest(struct test_loader *tester,
 			}
 		}
 
-		do_prog_test_run(bpf_program__fd(tprog), &retval,
-				 bpf_program__type(tprog) == BPF_PROG_TYPE_SYSCALL ? true : false);
-		if (retval != subspec->retval && subspec->retval != POINTER_VALUE) {
+		err = do_prog_test_run(bpf_program__fd(tprog), &retval,
+				       bpf_program__type(tprog) == BPF_PROG_TYPE_SYSCALL ? true : false);
+		if (!err && retval != subspec->retval && subspec->retval != POINTER_VALUE) {
 			PRINT_FAIL("Unexpected retval: %d != %d\n", retval, subspec->retval);
 			goto tobj_cleanup;
 		}
Signature · 1 line
2.47.1
#1 lore raw
Quoted 9 lines · Yonghong Song

With gcc14, when building with RELEASE=1, I hit four below compilation failure:

Error 1:
In file included from test_loader.c:6:
test_loader.c: In function ‘run_subtest’: test_progs.h:194:17:
error: ‘retval’ may be used uninitialized in this function
[-Werror=maybe-uninitialized]

Quoted 7 lines
    194 |                 fprintf(stdout, ##format);           \
        |                 ^~~~~~~
  test_loader.c:958:13: note: ‘retval’ was declared here
    958 |         int retval, err, i;
        |             ^~~~~~

The uninitialized var 'retval' actaully could cause incorrect result.

actually

Quoted 13 lines

Error 2:
In function ‘test_fd_array_cnt’:
prog_tests/fd_array.c:71:14: error: ‘btf_id’ may be used uninitialized in this

      function [-Werror=maybe-uninitialized]
     71 |         fd = bpf_btf_get_fd_by_id(id);
        |              ^~~~~~~~~~~~~~~~~~~~~~~~
  prog_tests/fd_array.c:302:15: note: ‘btf_id’ was declared here
    302 |         __u32 btf_id;
        |               ^~~~~~

Changing ASSERT_GE to ASSERT_EQ can fix the compilation error. Otherwise, there is no functionality change.

Quoted 8 lines

Error 3:
prog_tests/tailcalls.c: In function ‘test_tailcall_hierarchy_count’:
prog_tests/tailcalls.c:1402:23: error: ‘fentry_data_fd’ may be used uninitialized

      in this function [-Werror=maybe-uninitialized]
     1402 |                 err = bpf_map_lookup_elem(fentry_data_fd, &i, &val);
          |                       ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The code is correct. The change intends to slient gcc errors.

to silence.

Fixed the typos while applying.
Pls use spell check.

author reviewer
#2 lore raw
Quoted 38 lines · Alexei Starovoitov
Quoted 14 lines · Yonghong Song

With gcc14, when building with RELEASE=1, I hit four below compilation failure:

Error 1:
In file included from test_loader.c:6:
test_loader.c: In function ‘run_subtest’: test_progs.h:194:17:

       error: ‘retval’ may be used uninitialized in this function
    [-Werror=maybe-uninitialized]
     194 |                 fprintf(stdout, ##format);           \
         |                 ^~~~~~~
   test_loader.c:958:13: note: ‘retval’ was declared here
     958 |         int retval, err, i;
         |             ^~~~~~
Quoted 1 line

The uninitialized var 'retval' actaully could cause incorrect result.

actually

Quoted 20 lines

Error 2:
In function ‘test_fd_array_cnt’:
prog_tests/fd_array.c:71:14: error: ‘btf_id’ may be used uninitialized in this

       function [-Werror=maybe-uninitialized]
      71 |         fd = bpf_btf_get_fd_by_id(id);
         |              ^~~~~~~~~~~~~~~~~~~~~~~~
   prog_tests/fd_array.c:302:15: note: ‘btf_id’ was declared here
     302 |         __u32 btf_id;
         |               ^~~~~~

Changing ASSERT_GE to ASSERT_EQ can fix the compilation error. Otherwise, there is no functionality change.

Error 3:
prog_tests/tailcalls.c: In function ‘test_tailcall_hierarchy_count’:
prog_tests/tailcalls.c:1402:23: error: ‘fentry_data_fd’ may be used uninitialized

       in this function [-Werror=maybe-uninitialized]
      1402 |                 err = bpf_map_lookup_elem(fentry_data_fd, &i, &val);
           |                       ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Quoted 5 lines
Quoted 1 line

The code is correct. The change intends to slient gcc errors.

to silence.

Fixed the typos while applying.
Pls use spell check.

Sorry about typo's. Will pay attention to spell check for later patches.

#3 lore raw
In reply to Yonghong Song
This patch was applied to bpf/bpf-next.git (master) by Alexei Starovoitov <[email protected]>:

Hello:

This patch was applied to bpf/bpf-next.git (master) by Alexei Starovoitov <[email protected]>:

Quoted 8 lines · On Mon, 16 Jun 2025 21:49:56 -0700 you wrote:

With gcc14, when building with RELEASE=1, I hit four below compilation failure:

Error 1:
In file included from test_loader.c:6:
test_loader.c: In function ‘run_subtest’: test_progs.h:194:17:
error: ‘retval’ may be used uninitialized in this function
[-Werror=maybe-uninitialized]

Quoted 7 lines
    194 |                 fprintf(stdout, ##format);           \
        |                 ^~~~~~~
  test_loader.c:958:13: note: ‘retval’ was declared here
    958 |         int retval, err, i;
        |             ^~~~~~

[...]

Here is the summary with links:
- [bpf-next] selftests/bpf: Fix RELEASE build failure with gcc14

    https://git.kernel.org/bpf/bpf-next/c/a633dab4b4d2

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html