On Fri, May 31, 2024 at 04:23:36PM +0200, Laurent Vivier wrote:

I think this needs a commit message.  In particular expanding on what
you mean by "buffer management" would be helpful.

> Signed-off-by: Laurent Vivier <lvivier@redhat.com>
> ---
>  tcp.c | 87 ++++++++++++++++++++++++++++++++++++-----------------------
>  1 file changed, 54 insertions(+), 33 deletions(-)
> 
> diff --git a/tcp.c b/tcp.c
> index 6f221995f3bc..a6f43010f58f 100644
> --- a/tcp.c
> +++ b/tcp.c
> @@ -1518,24 +1518,25 @@ static void tcp_update_seqack_from_tap(const struct ctx *c,
>  }
>  
>  /**
> - * tcp_send_flag() - Send segment with flags to tap (no payload)
> + * tcp_fill_flag_header() - Prepare header for flags-only segment (no payload)
>   * @c:		Execution context
>   * @conn:	Connection pointer
>   * @flags:	TCP flags: if not set, send segment only if ACK is due
> + * @th:		TCP header to update
> + * @data:	buffer to store TCP option
> + * @optlen:	size of the TCP option buffer
>   *
> - * Return: negative error code on connection reset, 0 otherwise
> + * Return: < 0 error code on connection reset,
> + *           0 if there is no flag to send
> + *	     1 otherwise
>   */
> -static int tcp_send_flag(struct ctx *c, struct tcp_tap_conn *conn, int flags)
> +static int tcp_fill_flag_header(struct ctx *c, struct tcp_tap_conn *conn,
> +				int flags, struct tcphdr *th, char *data,
> +				size_t *optlen)
>  {
> -	struct tcp_flags_t *payload;
>  	struct tcp_info tinfo = { 0 };
>  	socklen_t sl = sizeof(tinfo);
>  	int s = conn->sock;
> -	size_t optlen = 0;
> -	struct tcphdr *th;
> -	struct iovec *iov;
> -	size_t l4len;
> -	char *data;
>  
>  	if (SEQ_GE(conn->seq_ack_to_tap, conn->seq_from_tap) &&
>  	    !flags && conn->wnd_to_tap)
> @@ -1557,20 +1558,11 @@ static int tcp_send_flag(struct ctx *c, struct tcp_tap_conn *conn, int flags)
>  	if (!tcp_update_seqack_wnd(c, conn, flags, &tinfo) && !flags)
>  		return 0;
>  
> -	if (CONN_V4(conn))
> -		iov = tcp4_l2_flags_iov[tcp4_flags_used++];
> -	else
> -		iov = tcp6_l2_flags_iov[tcp6_flags_used++];
> -
> -	payload = iov[TCP_IOV_PAYLOAD].iov_base;
> -	th = &payload->th;
> -	data = payload->opts;
> -
>  	if (flags & SYN) {
>  		int mss;
>  
>  		/* Options: MSS, NOP and window scale (8 bytes) */
> -		optlen = OPT_MSS_LEN + 1 + OPT_WS_LEN;
> +		*optlen = OPT_MSS_LEN + 1 + OPT_WS_LEN;
>  
>  		*data++ = OPT_MSS;
>  		*data++ = OPT_MSS_LEN;
> @@ -1604,26 +1596,13 @@ static int tcp_send_flag(struct ctx *c, struct tcp_tap_conn *conn, int flags)
>  		flags |= ACK;
>  	}
>  
> -	th->doff = (sizeof(*th) + optlen) / 4;
> +	th->doff = (sizeof(*th) + *optlen) / 4;
>  
>  	th->ack = !!(flags & ACK);
>  	th->rst = !!(flags & RST);
>  	th->syn = !!(flags & SYN);
>  	th->fin = !!(flags & FIN);
>  
> -	if (CONN_V4(conn)) {
> -		l4len = tcp_fill_headers4(c, conn, iov[TCP_IOV_TAP].iov_base,
> -					  iov[TCP_IOV_IP].iov_base,
> -					  iov[TCP_IOV_PAYLOAD].iov_base, optlen,
> -					  NULL, conn->seq_to_tap);
> -	} else {
> -		l4len = tcp_fill_headers6(c, conn, iov[TCP_IOV_TAP].iov_base,
> -					  iov[TCP_IOV_IP].iov_base,
> -					  iov[TCP_IOV_PAYLOAD].iov_base, optlen,
> -					  conn->seq_to_tap);
> -	}
> -	iov[TCP_IOV_PAYLOAD].iov_len = l4len;
> -
>  	if (th->ack) {
>  		if (SEQ_GE(conn->seq_ack_to_tap, conn->seq_from_tap))
>  			conn_flag(c, conn, ~ACK_TO_TAP_DUE);
> @@ -1638,6 +1617,48 @@ static int tcp_send_flag(struct ctx *c, struct tcp_tap_conn *conn, int flags)
>  	if (th->fin || th->syn)
>  		conn->seq_to_tap++;
>  
> +	return 1;
> +}
> +
> +static int tcp_send_flag(struct ctx *c, struct tcp_tap_conn *conn, int flags)
> +{
> +	struct tcp_flags_t *payload;
> +	size_t optlen = 0;
> +	struct iovec *iov;
> +	size_t l4len;
> +	int ret;
> +
> +	if (CONN_V4(conn)) {
> +		iov = tcp4_l2_flags_iov[tcp4_flags_used++];

Hrm.. it does occur to me that if you avoided the previous patch,
keeping a single dispatcher for filling in the IP headers, then
setting the iov pointer would become the only v4/v6 dependent part of
this function, helping us down the road of unifying that more.

Obviously there would be a separate CONN_v4() check hidden inside
fill_headers(), but as I've said I'm already aiming to replace those
with more specific tests.

> +		payload = iov[TCP_IOV_PAYLOAD].iov_base;
> +
> +		ret = tcp_fill_flag_header(c, conn, flags, &payload->th,
> +					   payload->opts, &optlen);
> +		if (ret <= 0)
> +			return ret;
> +
> +		l4len = tcp_fill_headers4(c, conn, iov[TCP_IOV_TAP].iov_base,
> +					  iov[TCP_IOV_IP].iov_base,
> +					  iov[TCP_IOV_PAYLOAD].iov_base, optlen,
> +					  NULL, conn->seq_to_tap);
> +	} else {
> +		iov = tcp6_l2_flags_iov[tcp6_flags_used++];
> +
> +		payload = iov[TCP_IOV_PAYLOAD].iov_base;
> +
> +		ret = tcp_fill_flag_header(c, conn, flags, &payload->th,
> +					   payload->opts, &optlen);
> +		if (ret <= 0)
> +			return ret;
> +
> +		l4len = tcp_fill_headers6(c, conn, iov[TCP_IOV_TAP].iov_base,
> +					  iov[TCP_IOV_IP].iov_base,
> +					  iov[TCP_IOV_PAYLOAD].iov_base, optlen,
> +					  conn->seq_to_tap);
> +	}
> +	iov[TCP_IOV_PAYLOAD].iov_len = l4len;
> +
>  	if (flags & DUP_ACK) {
>  		struct iovec *dup_iov;
>  		int i;

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson