CAKE's autorate-ingress path intends to limit shaper reconfiguration to once per 250 ms, but last_reconfig_time is only checked and never updated. Since the field stays zero, every qualifying capacity-estimate window can call cake_reconfigure(), causing avoidable rate churn and scheduler work under bursty traffic. Store the current timestamp when autorate actually reconfigures the qdisc so the guard enforces the intended interval. Fixes: 7298de9cd725 ("sch_cake: Add ingress mode") Assisted-by: Codex:gpt-5.5 Signed-off-by: Giuseppe Piscitelli --- v3: resend via git send-email; v2 went out through a mail client that mangled whitespace and does not apply v2: spell out my full name in From/Signed-off-by (Paolo Abeni) net/sched/sch_cake.c | 1 + 1 file changed, 1 insertion(+) diff --git a/net/sched/sch_cake.c b/net/sched/sch_cake.c index f25f60978..dc9326702 100644 --- a/net/sched/sch_cake.c +++ b/net/sched/sch_cake.c @@ -1907,6 +1907,7 @@ static s32 cake_enqueue(struct sk_buff *skb, struct Qdisc *sch, ktime_add_ms(q->last_reconfig_time, 250))) { q->config->rate_bps = (q->avg_peak_bandwidth * 15) >> 4; + q->last_reconfig_time = now; cake_reconfigure(sch); } } base-commit: 24ef02f934eeb48830cff6b739abc3c62b1d107b -- 2.54.0