Re: [PATCH v6 4/4] tcp: Update data retransmission timeout

[Yumei Huang <yuhuang@redhat.com>]

On Fri, Oct 24, 2025 at 7:04 AM Stefano Brivio <sbrivio@redhat.com> wrote:
>
> On Fri, 17 Oct 2025 14:28:38 +0800
> Yumei Huang <yuhuang@redhat.com> wrote:
>
> > Use an exponential backoff timeout for data retransmission according
> > to RFC 2988 and RFC 6298. Set the initial RTO to one second as discussed
> > in Appendix A of RFC 6298.
> >
> > Also combine the macros defining the initial RTO for both SYN and ACK.
> >
> > Signed-off-by: Yumei Huang <yuhuang@redhat.com>
> > Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
> > ---
> >  tcp.c | 27 ++++++++++++---------------
> >  1 file changed, 12 insertions(+), 15 deletions(-)
> >
> > diff --git a/tcp.c b/tcp.c
> > index 9385132..dc0ec6c 100644
> > --- a/tcp.c
> > +++ b/tcp.c
> > @@ -179,16 +179,14 @@
> >   *
> >   * Timeouts are implemented by means of timerfd timers, set based on flags:
> >   *
> > - * - SYN_TIMEOUT_INIT: if no ACK is received from tap/guest during handshake
> > - *   (flag ACK_FROM_TAP_DUE without ESTABLISHED event) within this time, resend
> > - *   SYN. It's the starting timeout for the first SYN retry. If this persists
> > - *   for more than TCP_MAX_RETRIES or (tcp_syn_retries +
> > - *   tcp_syn_linear_timeouts) times in a row, reset the connection
> > - *
> > - * - ACK_TIMEOUT: if no ACK segment was received from tap/guest, after sending
> > - *   data (flag ACK_FROM_TAP_DUE with ESTABLISHED event), re-send data from the
> > - *   socket and reset sequence to what was acknowledged. If this persists for
> > - *   more than TCP_MAX_RETRIES times in a row, reset the connection
> > + * - RTO_INIT: if no ACK segment was received from tap/guest, either during
> > + *   handshake (flag ACK_FROM_TAP_DUE without ESTABLISHED event) or after
> > + *   sending data (flag ACK_FROM_TAP_DUE with ESTABLISHED event), re-send data
> > + *   from the socket and reset sequence to what was acknowledged. This is the
> > + *   timeout for the first retry, in seconds. If this persists too many times
> > + *   in a row, reset the connection: TCP_MAX_RETRIES for established
> > + *   connections, or (tcp_syn_retries + tcp_syn_linear_timeouts) during the
> > + *   handshake.
> >   *
> >   * - FIN_TIMEOUT: if a FIN segment was sent to tap/guest (flag ACK_FROM_TAP_DUE
> >   *   with TAP_FIN_SENT event), and no ACK is received within this time, reset
> > @@ -342,8 +340,7 @@ enum {
> >  #define WINDOW_DEFAULT                       14600           /* RFC 6928 */
> >
> >  #define ACK_INTERVAL                 10              /* ms */
> > -#define SYN_TIMEOUT_INIT             1               /* s */
> > -#define ACK_TIMEOUT                  2
> > +#define RTO_INIT                     1               /* s, RFC 6298 */
> >  #define FIN_TIMEOUT                  60
> >  #define ACT_TIMEOUT                  7200
> >
> > @@ -588,13 +585,13 @@ static void tcp_timer_ctl(const struct ctx *c, struct tcp_tap_conn *conn)
> >       } else if (conn->flags & ACK_FROM_TAP_DUE) {
> >               if (!(conn->events & ESTABLISHED)) {
> >                       if (conn->retries < c->tcp.syn_linear_timeouts)
> > -                             it.it_value.tv_sec = SYN_TIMEOUT_INIT;
> > +                             it.it_value.tv_sec = RTO_INIT;
> >                       else
> > -                             it.it_value.tv_sec = SYN_TIMEOUT_INIT <<
> > +                             it.it_value.tv_sec = RTO_INIT <<
> >                                       (conn->retries - c->tcp.syn_linear_timeouts);
> >               }
> >               else
> > -                     it.it_value.tv_sec = ACK_TIMEOUT;
> > +                     it.it_value.tv_sec = RTO_INIT << conn->retries;
>
> Same as on 3/4, but here it's clearly more convenient: just assign
> RTO_INIT, and multiply as needed in the if / else clauses.

I guess we can't just assign RTO_INIT.  Maybe assign it only when
retries==0, otherwise multiply as it.it_value.tv_sec <<=1.  But it
seems more complicated. What do you think?

>
> >       } else if (CONN_HAS(conn, SOCK_FIN_SENT | TAP_FIN_ACKED)) {
> >               it.it_value.tv_sec = FIN_TIMEOUT;
> >       } else {
>
> The rest of the series looks good to me.
>
> It might be slightly more practical to factor in directly the RTO
> clamp, and I don't think it's complicated now that you have the helper
> from 2/4, but it's not a strong preference from my side, as the series
> makes sense in any case.

Reading tcp_rto_max_ms can be easy with the helper. My concern is
about the way we get the total time for retries.

I used to do it like this in v2,
https://archives.passt.top/passt-dev/20251010074700.22177-4-yuhuang@redhat.com/:

+#define RETRY_ELAPSED(timeout_init, retries) \
+ ((timeout_init) * ((1 << ((retries) + 1)) - 2))

Though the formula is not quite right, we could refine it as below:

#define RETRY_ELAPSED(retries) ((RTO_INIT) * ((1 << ((retries) + 1)) - 1))

Does it make sense to get the time this way?

>
> --
> Stefano
>


-- 
Thanks,

Yumei Huang