lib: Fix RTO update on timeout

This fixes the RTO doubling on timeout according to Karn/Partridge.
Exponentially increasing RTO when it times out (e.g. doubling)
fixes the problem that a sudden increase in real RTT starves the sRTT
updates by never getting out of backoff as retransmitted packets can't
update RTT.

Added an parameter to make it less aggressive, default is doubling.

Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>

3 files changed, 9 insertions, 12 deletions

diff --git a/src/lib/CMakeLists.txt b/src/lib/CMakeLists.txt
index 4f7d5d18..196bb905 100644
--- a/src/lib/CMakeLists.txt
+++ b/src/lib/CMakeLists.txt
@@ -200,6 +200,8 @@ set(FRCT_LINUX_RTT_ESTIMATOR TRUE CACHE BOOL
   "Use Linux RTT estimator formula instead of the TCP RFC formula")
 set(FRCT_RTO_MDEV_MULTIPLIER 2 CACHE STRING
   "Multiplier for deviation term in the RTO: RTO = sRTT + (mdev << X)")
+set(FRCT_RTO_INC_FACTOR 0 CACHE STRING
+  "Divisor for RTO increase after timeout: RTO += RTX >> X, 0: Karn/Partridge")
 set(FRCT_RTO_MIN 250 CACHE STRING
   "Minimum Retransmission Timeout (RTO) for FRCT (us)")
 set(FRCT_TICK_TIME 5000 CACHE STRING
diff --git a/src/lib/config.h.in b/src/lib/config.h.in
index e413a0c9..fd54dcd8 100644
--- a/src/lib/config.h.in
+++ b/src/lib/config.h.in
@@ -76,6 +76,7 @@
 #define RQ_SIZE             (@FRCT_REORDER_QUEUE_SIZE@)
 #define START_WINDOW        (@FRCT_START_WINDOW@)
 #define RTO_MIN             (@FRCT_RTO_MIN@ * 1000)
+#define RTO_DIV             (@FRCT_RTO_INC_FACTOR@)
 #define MDEV_MUL            (@FRCT_RTO_MDEV_MULTIPLIER@)
 
 #define TICTIME             (@FRCT_TICK_TIME@ * 1000)        /* ns */
diff --git a/src/lib/timerwheel.c b/src/lib/timerwheel.c
index f30f867d..cf4e1eef 100644
--- a/src/lib/timerwheel.c
+++ b/src/lib/timerwheel.c
@@ -36,7 +36,6 @@ struct rxm {
         struct frct_pci *    pkt;
         size_t               len;
         time_t               t0;      /* Time when original was sent (us). */
-        size_t               mul;     /* RTO multiplier.                   */
         struct frcti *       frcti;
         int                  fd;
         int                  flow_id; /* Prevent rtx when fd reused.       */
@@ -166,9 +165,7 @@ static void timerwheel_move(void)
                                 struct flow *        f;
                                 uint32_t             snd_lwe;
                                 uint32_t             rcv_lwe;
-                                time_t               rto;
                                 size_t               lvl = 0;
-                                time_t               act;
 
                                 r = list_entry(p, struct rxm, next);
 
@@ -188,8 +185,6 @@ static void timerwheel_move(void)
 
                                 snd_lwe = snd_cr->lwe;
                                 rcv_lwe = rcv_cr->lwe;
-                                rto     = r->frcti->rto;
-                                act     = ts_to_ns(r->frcti->rcv_cr.act);
 
                                 pthread_rwlock_unlock(&r->frcti->lock);
 
@@ -203,20 +198,20 @@ static void timerwheel_move(void)
 
                                 pthread_rwlock_wrlock(&r->frcti->lock);
 
-                                if (r->frcti->probe
-                                    && (r->frcti->rttseq == r->seqno))
+                                if (r->seqno == r->frcti->rttseq) {
+                                        r->frcti->rto +=
+                                                f->frcti->rto >> RTO_DIV;
                                         r->frcti->probe = false;
+                                }
 #ifdef PROC_FLOW_STATS
                                 r->frcti->n_rtx++;
 #endif
-                                pthread_rwlock_unlock(&r->frcti->lock);
+                                rslot = r->frcti->rto >> RXMQ_RES;
 
-                                if (ts_to_ns(now) - act > (rto << 2))
-                                        rto <<= r->mul++;
+                                pthread_rwlock_unlock(&r->frcti->lock);
 
                                 /* Schedule at least in the next time slot. */
                                 slot = ts_to_ns(now) >> RXMQ_RES;
-                                rslot = rto >> RXMQ_RES;
 
                                 while (rslot >= RXMQ_SLOTS) {
                                         ++lvl;
@@ -326,7 +321,6 @@ static int timerwheel_rxm(struct frcti *       frcti,
         clock_gettime(PTHREAD_COND_CLOCK, &now);
 
         r->t0    = ts_to_ns(now);
-        r->mul   = 0;
         r->seqno = seqno;
         r->frcti = frcti;
         r->len  = shm_du_buff_len(sdb);