On 25/11/2020 11:03, Paolo Abeni wrote:
On Wed, 2020-11-25 at 11:01 +0800, Geliang Tang wrote:
> This patch added the mibs for ADD_ADDR with port:
> MPTCP_MIB_PORTADD for receiving of the ADD_ADDR suboption with a port
> number. MPTCP_MIB_PORTSYNRX, MPTCP_MIB_PORTSYNACKRX, MPTCP_MIB_PORTACKRX,
> for receiving of the MP_JOIN's SYN or SYN/ACK or ACK with a port number
> which is different from the msk's port number.
> Signed-off-by: Geliang Tang <geliangtang(a)gmail.com>
> diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
> index cb72327c006a..6193e53d4af1 100644
> --- a/net/mptcp/subflow.c
> +++ b/net/mptcp/subflow.c
> @@ -196,6 +196,7 @@ static void subflow_init_req(struct request_sock *req,
> subflow_req->mp_join = 0;
We likely want to increment some 'bad port' MIB counter in case of
@Christoph: looks like mptcp.org
does not have MIBs for these things,
any plan to add them? any suggestions on relevant names/things to
, we have one MIB counter related to MP_JOIN on a different
I don't think any PMs in mptcp.org
announce an address with a different
port but we have "ndiffport" that sends MP_JOIN to a different port. It
works because we accept MP_JOIN sent to any port (hack).
I think it is not a bad thing to have these new counters :)
But maybe we should prefix the PORTSYN/SYNACK/ACK ones with JOIN? (e.g.
MPTCP_MIB_JOINPORTSYNRX and MPJoinPortSynRx ; or with "Port" at the end)
It is still an MP_JOIN we receive but to a different port.
Do you mean we might not need them?
Tessares | Belgium | Hybrid Access Solutions