Skip to contentSkip to conversation

selftests/bpf: Fix resource leak in serial_test_wq on attach failure

PATCH Applied Kery Qi 3 participants 5 messages Jan 21 – Jan 22

Subsystems (3)
lore
Share & tools
lore

Kery Qi

author
#0 lore raw

PATCH 1/1 selftests/bpf: Fix resource leak in serial_test_wq on attach failure

When wq__attach() fails, serial_test_wq() returns early without calling wq__destroy(), leaking the skeleton resources allocated by wq__open_and_load(). This causes ASAN leak reports in selftests runs.

Fix this by jumping to a common clean_up label that calls wq__destroy() on all exit paths after successful open_and_load.

Note that the early return after wq__open_and_load() failure is correct and doesn't need fixing, since that function returns NULL on failure (after internally cleaning up any partial allocations).

Fixes 8290dba51910 ("selftests/bpf: wq: add bpf_wq_start() checks")
---
 tools/testing/selftests/bpf/prog_tests/wq.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
tools/testing/selftests/bpf/prog_tests/wq.c +3 -2
index 15c67d23128b..84831eecc935 100644
@@ -16,12 +16,12 @@ void serial_test_wq(void)
 	/* re-run the success test to check if the timer was actually executed */
 
 	wq_skel = wq__open_and_load();
-	if (!ASSERT_OK_PTR(wq_skel, "wq_skel_load"))
+	if (!ASSERT_OK_PTR(wq_skel, "wq__open_and_load"))
 		return;
 
 	err = wq__attach(wq_skel);
 	if (!ASSERT_OK(err, "wq_attach"))
-		return;
+		goto clean_up;
 
 	prog_fd = bpf_program__fd(wq_skel->progs.test_syscall_array_sleepable);
 	err = bpf_prog_test_run_opts(prog_fd, &topts);
@@ -31,6 +31,7 @@ void serial_test_wq(void)
 	usleep(50); /* 10 usecs should be enough, but give it extra */
 
 	ASSERT_EQ(wq_skel->bss->ok_sleepable, (1 << 1), "ok_sleepable");
+clean_up:
 	wq__destroy(wq_skel);
 }
 
Signature · 1 line
2.34.1
reviewer Acked-by on v1 1/1
#1 lore raw
Quoted 7 lines · Kery Qi

When wq__attach() fails, serial_test_wq() returns early without calling wq__destroy(), leaking the skeleton resources allocated by wq__open_and_load(). This causes ASAN leak reports in selftests runs.

Fix this by jumping to a common clean_up label that calls wq__destroy() on all exit paths after successful open_and_load.

Quoted 6 lines

Note that the early return after wq__open_and_load() failure is correct and doesn't need fixing, since that function returns NULL on failure (after internally cleaning up any partial allocations).

Fixes 8290dba51910 ("selftests/bpf: wq: add bpf_wq_start() checks")
author
#2 lore raw

Hi,

Thank you for the review.

You are completely right regarding wq__open_and_load(): if it fails, it returns NULL and performs its own cleanup, so an immediate return is correct there.

However, my concern is about the next step: wq__attach(wq_skel).

If wq__open_and_load() succeeds, wq_skel is allocated. If wq__attach() subsequently fails, the original code returns immediately without calling wq__destroy(wq_skel), which causes a memory leak of the skeleton object.

The proposed goto clean_up is intended to ensure wq__destroy(wq_skel) is called specifically when wq__attach() fails.

Does that make sense?

Best regards,

Kery

Quoted 20 lines · Yonghong Song <[email protected]> 于2026年1月21日周三 23:45写道:
Quoted 7 lines · Kery Qi

When wq__attach() fails, serial_test_wq() returns early without calling wq__destroy(), leaking the skeleton resources allocated by wq__open_and_load(). This causes ASAN leak reports in selftests runs.

Fix this by jumping to a common clean_up label that calls wq__destroy() on all exit paths after successful open_and_load.

Quoted 6 lines

Note that the early return after wq__open_and_load() failure is correct and doesn't need fixing, since that function returns NULL on failure (after internally cleaning up any partial allocations).

Fixes 8290dba51910 ("selftests/bpf: wq: add bpf_wq_start() checks")
reviewer
#3 lore raw
Quoted 15 lines · Kery Qi

Hi,

Thank you for the review.

You are completely right regarding wq__open_and_load(): if it fails, it returns NULL and performs its own cleanup, so an immediate return is correct there.

However, my concern is about the next step: wq__attach(wq_skel).

If wq__open_and_load() succeeds, wq_skel is allocated. If wq__attach() subsequently fails, the original code returns immediately without calling wq__destroy(wq_skel), which causes a memory leak of the skeleton object.

Quoted 4 lines

The proposed goto clean_up is intended to ensure wq__destroy(wq_skel) is called specifically when wq__attach() fails.

Does that make sense?

Yes.

Quoted 24 lines

Best regards,

Kery

Quoted 18 lines · Yonghong Song <[email protected]> 于2026年1月21日周三 23:45写道:
Quoted 7 lines · Kery Qi

When wq__attach() fails, serial_test_wq() returns early without calling wq__destroy(), leaking the skeleton resources allocated by wq__open_and_load(). This causes ASAN leak reports in selftests runs.

Fix this by jumping to a common clean_up label that calls wq__destroy() on all exit paths after successful open_and_load.

Quoted 6 lines

Note that the early return after wq__open_and_load() failure is correct and doesn't need fixing, since that function returns NULL on failure (after internally cleaning up any partial allocations).

Fixes 8290dba51910 ("selftests/bpf: wq: add bpf_wq_start() checks")
#4 lore raw
In reply to Kery Qi
This patch was applied to bpf/bpf-next.git (master) by Andrii Nakryiko <[email protected]>:

Hello:

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

Quoted 8 lines · On Wed, 21 Jan 2026 17:41:16 +0800 you wrote:

When wq__attach() fails, serial_test_wq() returns early without calling wq__destroy(), leaking the skeleton resources allocated by wq__open_and_load(). This causes ASAN leak reports in selftests runs.

Fix this by jumping to a common clean_up label that calls wq__destroy() on all exit paths after successful open_and_load.

[...]

Here is the summary with links:
- [bpf] selftests/bpf: Fix resource leak in serial_test_wq on attach failure

    https://git.kernel.org/bpf/bpf-next/c/a32ae2658471

You are awesome, thank you!

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