As reported by @Tristan971 in GH #2116, the congestion control window could be zero
due to an inversion in the code about the reduction factor to be applied.
On a new loss event, it must be applied to the slow start threshold and the window
should never be below ->min_cwnd (2*max_udp_payload_sz).
Same issue in both newReno and cubic algorithm. Furthermore in newReno, only the
threshold was decremented.
Must be backported to 2.6 and 2.7.
else {
c->last_w_max = path->cwnd;
}
- path->cwnd = (CUBIC_BETA * path->cwnd) >> CUBIC_BETA_SCALE_SHIFT;
- c->ssthresh = QUIC_MAX(path->cwnd, path->min_cwnd);
+ c->ssthresh = (CUBIC_BETA * path->cwnd) >> CUBIC_BETA_SCALE_SHIFT;
+ path->cwnd = QUIC_MAX(c->ssthresh, (uint32_t)path->min_cwnd);
c->state = QUIC_CC_ST_RP;
TRACE_LEAVE(QUIC_EV_CONN_CC, cc->qc, NULL, cc);
}
path = container_of(cc, struct quic_path, cc);
nr->recovery_start_time = now_ms;
- nr->ssthresh = QUIC_MAX(path->cwnd >> 1, path->min_cwnd);
+ nr->ssthresh = path->cwnd >> 1;
+ path->cwnd = QUIC_MAX(nr->ssthresh, (uint32_t)path->min_cwnd);
nr->state = QUIC_CC_ST_RP;
}