@@ -786,7 +786,7 @@ void ocelot_get_txtstamp(struct ocelot *ocelot)
spin_unlock_irqrestore(&port->tx_skbs.lock, flags);
if (WARN_ON(!skb_match))
- continue;
+ goto next_ts;
if (!ocelot_validate_ptp_skb(skb_match, seqid)) {
dev_err_ratelimited(ocelot->dev,
@@ -804,7 +804,7 @@ void ocelot_get_txtstamp(struct ocelot *ocelot)
shhwtstamps.hwtstamp = ktime_set(ts.tv_sec, ts.tv_nsec);
skb_complete_tx_timestamp(skb_match, &shhwtstamps);
- /* Next ts */
+next_ts:
ocelot_write(ocelot, SYS_PTP_NXT_PTP_NXT, SYS_PTP_NXT);
}
}
This condition, theoretically impossible to trigger, is not really handled well. By "continuing", we are skipping the write to SYS_PTP_NXT which advances the timestamp FIFO to the next entry. So we are reading the same FIFO entry all over again, printing stack traces and eventually killing the kernel. No real problem has been observed here. This is part of a larger rework of the timestamp IRQ procedure, with this logical change split out into a patch of its own. We will need to "goto next_ts" for other conditions as well. Fixes: 9fde506e0c53 ("net: mscc: ocelot: warn when a PTP IRQ is raised for an unknown skb") Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com> --- drivers/net/ethernet/mscc/ocelot_ptp.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)