From: Frédéric Lécaille <flecaille@haproxy.com>
Date: Tue, 4 Apr 2023 13:47:16 +0000 (+0200)
Subject: BUG/MINOR: quic: Possible wrong PTO computing
X-Git-Tag: v2.8-dev7~106
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=0222cc636670c7d2fe8f02f41d2bd7e68f5e756c;p=thirdparty%2Fhaproxy.git

BUG/MINOR: quic: Possible wrong PTO computing

As timestamps based on now_ms values are used to compute the probing timeout,
they may wrap. So, use ticks API to compared them.

Must be backported to 2.7 and 2.6.
---

diff --git a/src/quic_loss.c b/src/quic_loss.c
index eff8e6c36c..bb518ca253 100644
--- a/src/quic_loss.c
+++ b/src/quic_loss.c
@@ -128,7 +128,7 @@ struct quic_pktns *quic_pto_pktns(struct quic_conn *qc,
 
 		p = &qc->pktns[i];
 		tmp_pto = tick_add(p->tx.time_of_last_eliciting, duration);
-		if (!tick_isset(lpto) || tmp_pto < lpto) {
+		if (!tick_isset(lpto) || tick_is_lt(tmp_pto, lpto)) {
 			lpto = tmp_pto;
 			pktns = p;
 		}