Skip to contentSkip to conversation

bpf: Compare iterator types during state pruning

PATCH Superseded Ning Ding 2 participants 6 messages Aug 7 – Aug 11

Subsystems (3)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Ning Ding

author
#0 lore raw

cover bpf: Compare iterator types during state pruning

Iterator stack slots can be marked MEM_RCU or PTR_UNTRUSTED. The STACK_ITER check in stacksafe() does not compare this type, so state pruning can treat these states as equal and prune an unsafe path.

Compare the type and add a test where RCU protection has a gap.

Ning Ding (2):
bpf: Compare iterator types during state pruning selftests/bpf: Test RCU iterator state pruning

 kernel/bpf/states.c                           |  3 ++-
 .../selftests/bpf/progs/iters_task_failure.c  | 27 +++++++++++++++++++
 2 files changed, 29 insertions(+), 1 deletion(-)
Signature · 1 line
2.43.0
author
#1 lore raw

PATCH 1/2 bpf: Compare iterator types during state pruning

An iterator stack slot can be MEM_RCU or PTR_UNTRUSTED. These states must not be equal, or the verifier can prune an unsafe path.

Compare the pointer type for STACK_ITER slots.

Fixes dfab99df147b ("bpf: teach the verifier to enforce css_iter and task_iter in RCU CS")
---
 kernel/bpf/states.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
kernel/bpf/states.c +2 -1
index ea2153cf28d0a..4e6aafad33bd2 100644
@@ -812,7 +812,8 @@ static bool stacksafe(struct bpf_verifier_env *env, struct bpf_func_state *old,
 			 * infinite loop check triggering, see
 			 * iter_active_depths_differ()
 			 */
-			if (old_reg->iter.btf != cur_reg->iter.btf ||
+			if (old_reg->type != cur_reg->type ||
+			    old_reg->iter.btf != cur_reg->iter.btf ||
 			    old_reg->iter.btf_id != cur_reg->iter.btf_id ||
 			    old_reg->iter.state != cur_reg->iter.state ||
 			    /* ignore {old_reg,cur_reg}->iter.depth, see above */
Signature · 1 line
2.43.0
author
#3 lore raw

PATCH 2/2 selftests/bpf: Test RCU iterator state pruning

In reply to Ning Ding

Add a path where RCU protection reaches zero and then starts again. The iterator is untrusted after this gap and must be rejected.

---
 .../selftests/bpf/progs/iters_task_failure.c  | 27 +++++++++++++++++++
 1 file changed, 27 insertions(+)
tools/testing/selftests/bpf/progs/iters_task_failure.c +27 -0
index fe3663dedbe14..bac394ae77f44 100644
@@ -61,6 +61,33 @@ int BPF_PROG(iter_tasks_lock_and_unlock)
 	return 0;
 }
 
+SEC("?fentry.s/" SYS_PREFIX "sys_getpgid")
+__failure __msg("expected an RCU CS when using bpf_iter_task_next")
+__flag(BPF_F_TEST_STATE_FREQ)
+int BPF_PROG(iter_tasks_rcu_state_pruning)
+{
+	struct bpf_iter_task it;
+
+	bpf_rcu_read_lock();
+	bpf_iter_task_new(&it, NULL, BPF_TASK_ITER_ALL_PROCS);
+
+	if (likely(bpf_get_prandom_u32())) {
+		/* Keep the outer RCU lock active. */
+		bpf_rcu_read_lock();
+		bpf_rcu_read_unlock();
+	} else {
+		/* Create an unprotected gap. */
+		bpf_rcu_read_unlock();
+		bpf_rcu_read_lock();
+	}
+
+	bpf_iter_task_next(&it);
+	bpf_iter_task_destroy(&it);
+	bpf_rcu_read_unlock();
+
+	return 0;
+}
+
 SEC("?fentry.s/" SYS_PREFIX "sys_getpgid")
 __failure __msg("expected an RCU CS when using bpf_iter_css_next")
 int BPF_PROG(iter_css_lock_and_unlock)
Signature · 1 line
2.43.0
maintainer
#4 lore raw

Re: [PATCH bpf-next 2/2] selftests/bpf: Test RCU iterator state pruning

In reply to Ning Ding

On Thu, 2026-08-06 at 17:43 -0700, Ning Ding wrote:

...

tools/testing/selftests/bpf/progs/iters_task_failure.c int BPF_PROG(iter_tasks_lock_and_unlock) · line 74

Quoted 11 lines
tools/testing/selftests/bpf/progs/iters_task_failure.c +11 -0
+SEC("?fentry.s/" SYS_PREFIX "sys_getpgid")
+__failure __msg("expected an RCU CS when using bpf_iter_task_next")
+__flag(BPF_F_TEST_STATE_FREQ)
+int BPF_PROG(iter_tasks_rcu_state_pruning)
+{
+	struct bpf_iter_task it;
+
+	bpf_rcu_read_lock();
+	bpf_iter_task_new(&it, NULL, BPF_TASK_ITER_ALL_PROCS);
+
+	if (likely(bpf_get_prandom_u32())) {

Could you please convert this test to inline assembly? The presence of 'likely' might not be sufficient to guarantee the codegen pattern, so we usually resort to the assembly for tests where that is important.

Quoted 19 lines
tools/testing/selftests/bpf/progs/iters_task_failure.c +16 -0
+		/* Keep the outer RCU lock active. */
+		bpf_rcu_read_lock();
+		bpf_rcu_read_unlock();
+	} else {
+		/* Create an unprotected gap. */
+		bpf_rcu_read_unlock();
+		bpf_rcu_read_lock();
+	}
+
+	bpf_iter_task_next(&it);
+	bpf_iter_task_destroy(&it);
+	bpf_rcu_read_unlock();
+
+	return 0;
+}
+

 SEC("?fentry.s/" SYS_PREFIX "sys_getpgid")
 __failure __msg("expected an RCU CS when using bpf_iter_css_next") int BPF_PROG(iter_css_lock_and_unlock)

author
#5 lore raw

Re: [PATCH bpf-next 2/2] selftests/bpf: Test RCU iterator state pruning

Quoted 4 lines

Could you please convert this test to inline assembly? The presence of 'likely' might not be sufficient to guarantee the codegen pattern, so we usually resort to the assembly for tests where that is important.

Sure, I will fix and resubmit.