Re: [PATCH bpf-next] seg6: Ensure that seg6_bpf_srh_states can only be accessed from input_action_end_bpf()

On 7/8/24 10:18 PM, Sebastian Andrzej Siewior wrote:

On 2024-07-08 17:03:58 [-0700], Martin KaFai Lau wrote:

quoted

quoted

diff --git a/net/core/filter.c b/net/core/filter.c
index 403d23faf22e1..ea5bc4a4a6a23 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c

@@ -6459,6 +6459,8 @@ BPF_CALL_4(bpf_lwt_seg6_store_bytes, struct sk_buff *, skb, u32, offset,
   	void *srh_tlvs, *srh_end, *ptr;
   	int srhoff = 0;
+	if (!bpf_net_ctx_seg6_state_avail())
+		return -EINVAL;

The syzbot stack shows that the seg6local bpf_prog can be run by test_run
like: bpf_prog_test_run_skb() => bpf_test_run(). "return -EINVAL;" will
reject and break the existing bpf prog doing test with test_run.

But wouldn't this be the case anyway because seg6_bpf_srh_states::srh
isn't assigned?

Good point.

I don't think the test_run for BPF_PROG_TYPE_LWT_SEG6LOCAL ever works. It seems 
no test_run selftest was ever added to exercise this code path since the 
lwt_seg6local_prog_ops was added in commit 004d4b274e2a ("ipv6: sr: Add 
seg6local action End.BPF").

I think the right thing to do here is to remove the test_run code path for 
BPF_PROG_TYPE_LWT_SEG6LOCAL instead of further patching it. A separate patch for 
proper test_run support is needed (cc: Mathieu Xhonneux).

To remove test_run for BPF_PROG_TYPE_LWT_SEG6LOCAL, this should do:

diff --git i/net/core/filter.c w/net/core/filter.c
index 403d23faf22e..db5e59f62b35 100644
--- i/net/core/filter.c
+++ w/net/core/filter.c

@@ -11053,7 +11053,6 @@ const struct bpf_verifier_ops lwt_seg6local_verifier_ops = {
  };

  const struct bpf_prog_ops lwt_seg6local_prog_ops = {
-	.test_run		= bpf_prog_test_run_skb,
  };

  const struct bpf_verifier_ops cg_sock_verifier_ops = {

Then the bpf_lwt_seg6_* helpers should stay in the input_action_end_bpf() code 
path only.