[rds-devel] [net-next, v2, 7/7] net/rds: Trigger rds_send_ping() more than once

[Jakub Kicinski]

This is an AI-generated review of your patch. The human sending this
email has considered the AI review valid, or at least plausible.

For local reproduction steps see https://netdev-ai.bots.linux.dev/ai-local.html
---
net/rds: Trigger rds_send_ping() more than once

This commit ensures rds_send_ping() is triggered after reconnect to
properly communicate multipath capabilities, especially after an RDS
module reload. It also validates source port modulo during fan-out.

> diff --git a/net/rds/send.c b/net/rds/send.c
> index 599c2cfb7a1d..68f6b12c60ef 100644
> --- a/net/rds/send.c
> +++ b/net/rds/send.c
> @@ -1339,6 +1339,24 @@ int rds_sendmsg(struct socket *sock, struct msghdr *msg, size_t payload_len)
>  		cpath = &conn->c_path[0];
>  	}
>
> +	/* Check to see if we have talked to this peer before.
> +	 * If not, this means conn->c_npaths will be zero,
> +	 * and we should initiate a connection request to the
> +	 * peer right away.
> +	 */
> +	if (conn->c_trans->t_mp_capable &&
> +	    !rds_conn_path_up(&conn->c_path[0])) {

This isn't a bug, but the comment says "c_npaths will be zero" while
the code checks "!rds_conn_path_up()". These conditions are different:
c_npaths == 0 means no paths negotiated yet, while !rds_conn_path_up()
is true for any state other than RDS_CONN_UP (DOWN, CONNECTING,
DISCONNECTING, ERROR, RESETTING).

The commit message mentions "after a re-connect to any peer" which
aligns with the code, but the comment specifically mentions c_npaths.
Should the comment be updated to reflect the actual logic?

This was flagged in v1 review:
https://lore.kernel.org/netdev/20260126174754.1393215-1-horms@kernel.org/

[ ... ]