diff options
Diffstat (limited to 'lib')
-rw-r--r-- | lib/diameter/src/base/diameter_peer_fsm.erl | 19 |
1 files changed, 17 insertions, 2 deletions
diff --git a/lib/diameter/src/base/diameter_peer_fsm.erl b/lib/diameter/src/base/diameter_peer_fsm.erl index 9ff6845ab7..8bfdb3ae39 100644 --- a/lib/diameter/src/base/diameter_peer_fsm.erl +++ b/lib/diameter/src/base/diameter_peer_fsm.erl @@ -642,7 +642,9 @@ rcv('DPA' = N, diameter_peer:close(TPid), {stop, N}; -%% Ignore anything else, an unsolicited DPA in particular. +%% Ignore anything else, an unsolicited DPA in particular. Note that +%% dpa_timeout deals with the case in which the peer sends the wrong +%% identifiers in DPA. rcv(N, #diameter_packet{header = H}, _) when N == 'CER'; N == 'CEA'; @@ -820,7 +822,7 @@ build_answer(Type, errors = Es} = Pkt, S) -> - {RC, FailedAVP} = result_code(H, Es), + {RC, FailedAVP} = result_code(Type, H, Es), {answer(Type, RC, FailedAVP, S), post(Type, RC, Pkt, S)}. inband_security([]) -> @@ -890,6 +892,19 @@ set(['answer-message' | _] = Ans, FailedAvp) -> set([_|_] = Ans, FailedAvp) -> Ans ++ FailedAvp. +%% result_code/3 + +%% Be lenient with errors in DPR since there's no reason to be +%% otherwise. Rejecting may cause the peer to missinterpret the error +%% as meaning that the connection should not be closed, which may well +%% lead to more problems than any errors in the DPR. + +result_code('DPR', _, _) -> + {2001, []}; + +result_code('CER', H, Es) -> + result_code(H, Es). + %% result_code/2 result_code(#diameter_header{is_error = true}, _) -> |