`update_fail_htlc.reason` is not long enough to decode

[nueno4]

1. Establish channel only 2 nodes(A and B)
2. A sends update_add_htlc(but, outgoing_cltv_value / cltv_expiry equal 0) to B
3. A and B exchange commitment_signed, revoke_and_ack.
4. B think WIRE_FINAL_EXPIRY_TOO_SOON (c-lightning log)
5. B sends update_fail_htlc to A

update_fail_htlc.len equal 2.
BOLT4's format(previous obscured) is longer than 2.

[rustyrussell]

As I read it final expiry_too soon has no data: it's just a type, which is 2 bytes?

[nueno4]

Maybe random 2bytes...

I think that expiry_too_soon shall be failure_len == 2.
But I receive update_fail_htlc reason length == 2, so cannot decode reason.

[cdecker]

@nayuta-ueno do you have an example of the error message so we can disect it together and clear up any misunderstanding?

If I read the spec correctly this is a valid error of type final_expiry_too_soon (aside from the HMAC of course)

0000000000000000000000000000000000000000000000000000000000000000 (hmac)
0002 (failure_len)
0011 (final_expiry_too_soon)
0004 (pad_len)
00000000 (padding)

I think maybe you are mixing final_expiry_too_soon and expiry_too_soon? The former is just 2 bytes which specify the type and an empty payload, while the latter requires a channel_update as payload.

[nueno4]

I think maybe you are mixing final_expiry_too_soon and expiry_too_soon?

it's my typo: expiry_too_soon ---> final_expiry_too_soon

======

update_fail_htlc log (using regtest)

c-lightning log:

lightningd(17687): peer 021b99745df3212c1dbbb8a06426b639faf3c3ce7f8b709a7853050206c8eb1114: got revoke 0: 1 changed
lightningd(17687): peer 021b99745df3212c1dbbb8a06426b639faf3c3ce7f8b709a7853050206c8eb1114: HTLC in 0 SENT_ADD_ACK_COMMIT->RCVD_ADD_ACK_REVOCATION
lightningd(17687): peer 021b99745df3212c1dbbb8a06426b639faf3c3ce7f8b709a7853050206c8eb1114: Expiry cltv too soon 487 < 487 + 6
lightningd(17687): peer 021b99745df3212c1dbbb8a06426b639faf3c3ce7f8b709a7853050206c8eb1114: failed htlc 0 code 0x0011 (WIRE_FINAL_EXPIRY_TOO_SOON)

I receive BOLT message:

0083eff68aae7d565ec2fb0c3a2ef944ac8b64747dbae860bf06f3e1b2b5b751b29f00000000000000000002ebad

decode

[update_fail_htlc]
channel-id: eff68aae7d565ec2fb0c3a2ef944ac8b64747dbae860bf06f3e1b2b5b751b29f
id: 0
len= 2
reason: ebad

I think that reason is encoded error packet like BOLT4 TestVector error packet for node x.

[cdecker]

Could it be that this was resolved by merging #858?

[ZmnSCPxj]

Not @nayuta-ueno, but I believe #857 turns out to be a duplicate of this issue, and as #858 resolves it, I think we can close this issue too.

[nueno4]

Thanks @ZmnSCPxj !
I can not make an environment to test right now...
I think its OK.