[PATCH 6.12 105/172] net: mscc: ocelot: ocelot->ts_id_lock and ocelot_port->tx_skbs.lock are IRQ-safe

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



6.12-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Vladimir Oltean <vladimir.oltean@xxxxxxx>

[ Upstream commit 0c53cdb95eb4a604062e326636971d96dd9b1b26 ]

ocelot_get_txtstamp() is a threaded IRQ handler, requested explicitly as
such by both ocelot_ptp_rdy_irq_handler() and vsc9959_irq_handler().

As such, it runs with IRQs enabled, and not in hardirq context. Thus,
ocelot_port_add_txtstamp_skb() has no reason to turn off IRQs, it cannot
be preempted by ocelot_get_txtstamp(). For the same reason,
dev_kfree_skb_any_reason() will always evaluate as kfree_skb_reason() in
this calling context, so just simplify the dev_kfree_skb_any() call to
kfree_skb().

Also, ocelot_port_txtstamp_request() runs from NET_TX softirq context,
not with hardirqs enabled. Thus, ocelot_get_txtstamp() which shares the
ocelot_port->tx_skbs.lock lock with it, has no reason to disable hardirqs.

This is part of a larger rework of the TX timestamping procedure.
A logical subportion of the rework has been split into a separate
change.

Signed-off-by: Vladimir Oltean <vladimir.oltean@xxxxxxx>
Link: https://patch.msgid.link/20241205145519.1236778-4-vladimir.oltean@xxxxxxx
Signed-off-by: Jakub Kicinski <kuba@xxxxxxxxxx>
Stable-dep-of: b454abfab525 ("net: mscc: ocelot: be resilient to loss of PTP packets during transmission")
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
 drivers/net/ethernet/mscc/ocelot_ptp.c | 14 ++++++--------
 1 file changed, 6 insertions(+), 8 deletions(-)

diff --git a/drivers/net/ethernet/mscc/ocelot_ptp.c b/drivers/net/ethernet/mscc/ocelot_ptp.c
index 95a5267bc9ce..d732f99e6391 100644
--- a/drivers/net/ethernet/mscc/ocelot_ptp.c
+++ b/drivers/net/ethernet/mscc/ocelot_ptp.c
@@ -607,13 +607,12 @@ static int ocelot_port_add_txtstamp_skb(struct ocelot *ocelot, int port,
 					struct sk_buff *clone)
 {
 	struct ocelot_port *ocelot_port = ocelot->ports[port];
-	unsigned long flags;
 
-	spin_lock_irqsave(&ocelot->ts_id_lock, flags);
+	spin_lock(&ocelot->ts_id_lock);
 
 	if (ocelot_port->ptp_skbs_in_flight == OCELOT_MAX_PTP_ID ||
 	    ocelot->ptp_skbs_in_flight == OCELOT_PTP_FIFO_SIZE) {
-		spin_unlock_irqrestore(&ocelot->ts_id_lock, flags);
+		spin_unlock(&ocelot->ts_id_lock);
 		return -EBUSY;
 	}
 
@@ -630,7 +629,7 @@ static int ocelot_port_add_txtstamp_skb(struct ocelot *ocelot, int port,
 
 	skb_queue_tail(&ocelot_port->tx_skbs, clone);
 
-	spin_unlock_irqrestore(&ocelot->ts_id_lock, flags);
+	spin_unlock(&ocelot->ts_id_lock);
 
 	return 0;
 }
@@ -749,7 +748,6 @@ void ocelot_get_txtstamp(struct ocelot *ocelot)
 		u32 val, id, seqid, txport;
 		struct ocelot_port *port;
 		struct timespec64 ts;
-		unsigned long flags;
 
 		val = ocelot_read(ocelot, SYS_PTP_STATUS);
 
@@ -773,7 +771,7 @@ void ocelot_get_txtstamp(struct ocelot *ocelot)
 
 		/* Retrieve its associated skb */
 try_again:
-		spin_lock_irqsave(&port->tx_skbs.lock, flags);
+		spin_lock(&port->tx_skbs.lock);
 
 		skb_queue_walk_safe(&port->tx_skbs, skb, skb_tmp) {
 			if (OCELOT_SKB_CB(skb)->ts_id != id)
@@ -783,7 +781,7 @@ void ocelot_get_txtstamp(struct ocelot *ocelot)
 			break;
 		}
 
-		spin_unlock_irqrestore(&port->tx_skbs.lock, flags);
+		spin_unlock(&port->tx_skbs.lock);
 
 		if (WARN_ON(!skb_match))
 			goto next_ts;
@@ -792,7 +790,7 @@ void ocelot_get_txtstamp(struct ocelot *ocelot)
 			dev_err_ratelimited(ocelot->dev,
 					    "port %d received stale TX timestamp for seqid %d, discarding\n",
 					    txport, seqid);
-			dev_kfree_skb_any(skb);
+			kfree_skb(skb);
 			goto try_again;
 		}
 
-- 
2.39.5







[Index of Archives]     [Linux Kernel]     [Kernel Development Newbies]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite Hiking]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux