Re: [PATCH] mptcp: add per-reason MIB counters for MPTCP_RST_EMPTCP resets

[Matthieu Baerts]

Hi Shardul,

On 21/04/2026 11:56, Shardul Bankar wrote:
> MPTCP_RST_EMPTCP (reset reason 1) is used as a catch-all for seven
> distinct error conditions across subflow setup, authentication, and
> data path validation.  The existing MPRstTx/MPRstRx counters only
> track aggregate reset volume, making it difficult to diagnose which
> code path is triggering subflow resets in production.
> 
> Add per-reason MIB counters for each MPTCP_RST_EMPTCP use site:
> 
>   MPRstMD5SIG         - MD5SIG on listener, incompatible with MPTCP
>   MPRstNoToken        - JOIN token lookup failed, no matching conn
>   MPRstNoMPJ          - SYN/ACK missing MP_JOIN option
>   MPRstHMAC           - HMAC validation failed during subflow join
>   MPRstFatalFallback  - fatal fallback during child socket creation
>   MPRstBadMap         - invalid data mapping on established subflow
>   MPRstNotEstablished - JOIN on not-fully-established msk
> 
> These counters are incremented alongside the existing reset_reason
> assignment and are visible via nstat as MPTcpExtMPRst*.  The
> aggregate MPRstTx/MPRstRx counters remain unchanged.
Thank you for this patch.

Even if the patch looks globally good, net-next is currently closed, and
only bug fixes are accepted, see:

https://docs.kernel.org/process/maintainer-netdev.html

pw-bot: defer

Instead, I suggest switching the discussions only on the MPTCP ML: I
will try to review this patch and only send a reply to you and the MPTCP
list, if that's OK.

Cheers,
Matt
-- 
Sponsored by the NGI0 Core fund.