Skip to contentSkip to conversation

Follow-up fix for potential error pointer dereference in propagate_to_outer_instance()

PATCH Superseded Shardul Bankar 2 participants 4 messages Oct 16, 2025 – Oct 20, 2025 BPF

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover Follow-up fix for potential error pointer dereference in propagate_to_outer_instance()

Hi BPF maintainers,

This patch follows up on my previous submission:

[PATCH v2 bpf] bpf: Fix memory leak in __lookup_instance error path

During the review and CI discussions for that patch, a potential issue was identified in propagate_to_outer_instance(), where get_outer_instance() may return an ERR_PTR (e.g. -ENOMEM) that is not currently checked before use.

This patch adds a simple IS_ERR() guard and returns the error code to prevent dereferencing the error pointer.

Thanks,
Shardul

---

Shardul Bankar (1):
bpf: liveness: Handle ERR_PTR from get_outer_instance() in
propagate_to_outer_instance()

 kernel/bpf/liveness.c | 2 ++
 1 file changed, 2 insertions(+)
Signature · 1 line
2.34.1
#1 lore raw

PATCH 1/1 bpf: liveness: Handle ERR_PTR from get_outer_instance() in propagate_to_outer_instance()

propagate_to_outer_instance() calls get_outer_instance() and then uses the returned pointer to reset/commit stack write marks. When get_outer_instance() fails (e.g., __lookup_instance() returns -ENOMEM), it may return an ERR_PTR. Without a check, the code dereferences this error pointer.

Protect the call with IS_ERR() and propagate the error.

Fixes b3698c356ad9 ("bpf: callchain sensitive stack liveness tracking

using CFG")

---
 kernel/bpf/liveness.c | 2 ++
 1 file changed, 2 insertions(+)
kernel/bpf/liveness.c +2 -0
index 3c611aba7f52..ae31f9ee4994 100644
@@ -522,6 +522,8 @@ static int propagate_to_outer_instance(struct bpf_verifier_env *env,
 
 	this_subprog_start = callchain_subprog_start(callchain);
 	outer_instance = get_outer_instance(env, instance);
+	if (IS_ERR(outer_instance))
+		return PTR_ERR(outer_instance);
 	callsite = callchain->callsites[callchain->curframe - 1];
 
 	reset_stack_write_marks(env, outer_instance, callsite);
Signature · 1 line
2.34.1
maintainer
#2 lore raw

Re: [PATCH bpf 1/1] bpf: liveness: Handle ERR_PTR from get_outer_instance() in propagate_to_outer_instance()

Quoted 5 lines · Shardul Bankar

propagate_to_outer_instance() calls get_outer_instance() and then uses the returned pointer to reset/commit stack write marks. When get_outer_instance() fails (e.g., __lookup_instance() returns -ENOMEM), it may return an ERR_PTR. Without a check, the code dereferences this error pointer.

Quoted 7 lines

Protect the call with IS_ERR() and propagate the error.

Fixes b3698c356ad9 ("bpf: callchain sensitive stack liveness tracking

using CFG")

---

This was brought up already in [1]. This is not a bug as check before propagate_to_outer_instance() call in update_instance() guarantees that outer instance exists.

We can land this change to avoid confusion, but the fixes tag is unnecessary.

[1] https://lore.kernel.org/bpf/[email protected]/

Quoted 16 lines
 kernel/bpf/liveness.c | 2 ++
 1 file changed, 2 insertions(+)
kernel/bpf/liveness.c +2 -0
index 3c611aba7f52..ae31f9ee4994 100644
@@ -522,6 +522,8 @@ static int propagate_to_outer_instance(struct bpf_verifier_env *env,
 
 	this_subprog_start = callchain_subprog_start(callchain);
 	outer_instance = get_outer_instance(env, instance);
+	if (IS_ERR(outer_instance))
+		return PTR_ERR(outer_instance);
 	callsite = callchain->callsites[callchain->curframe - 1];
 
 	reset_stack_write_marks(env, outer_instance, callsite);
#3 lore raw

Re: [PATCH bpf 1/1] bpf: liveness: Handle ERR_PTR from get_outer_instance() in propagate_to_outer_instance()

Quoted 6 lines · Eduard Zingerman

We can land this change to avoid confusion, but the fixes tag is unnecessary.

[1]
https://lore.kernel.org/bpf/[email protected]/

Thanks, Eduard, for the clarification and reference.

That makes sense — since update_instance() guarantees the outer instance’s existence, this case wouldn’t be hit in normal execution.

I agree, the Fixes: tag can be dropped. The check mainly serves to make the intent explicit and avoid future confusion.

I’ve resent the patch without the Fixes: tag.
Link:
https://lore.kernel.org/all/[email protected]/

Thanks again,
Shardul