net: dst: add net device refcount tracking to dst_entry

We want to track all dev_hold()/dev_put() to ease leak hunting.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

5 files changed, 29 insertions, 9 deletions

diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index b4f704337f65..afed3b10491b 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -3883,6 +3883,23 @@ static inline void dev_put_track(struct net_device *dev,
 	}
 }
 
+static inline void dev_replace_track(struct net_device *odev,
+				     struct net_device *ndev,
+				     netdevice_tracker *tracker,
+				     gfp_t gfp)
+{
+#ifdef CONFIG_NET_DEV_REFCNT_TRACKER
+	if (odev)
+		ref_tracker_free(&odev->refcnt_tracker, tracker);
+#endif
+	dev_hold(ndev);
+	dev_put(odev);
+#ifdef CONFIG_NET_DEV_REFCNT_TRACKER
+	if (ndev)
+		ref_tracker_alloc(&ndev->refcnt_tracker, tracker, gfp);
+#endif
+}
+
 /* Carrier loss detection, dial on demand. The functions netif_carrier_on
  * and _off may be called from IRQ context, but it is caller
  * who is responsible for serialization of these calls.
diff --git a/include/net/dst.h b/include/net/dst.h
index a057319aabef..6aa252c3fc55 100644
--- a/include/net/dst.h
+++ b/include/net/dst.h
@@ -77,6 +77,7 @@ struct dst_entry {
 #ifndef CONFIG_64BIT
 	atomic_t		__refcnt;	/* 32-bit offset 64 */
 #endif
+	netdevice_tracker	dev_tracker;
 };
 
 struct dst_metrics {
diff --git a/net/core/dst.c b/net/core/dst.c
index 497ef9b3fc6a..d16c2c9bfebd 100644
--- a/net/core/dst.c
+++ b/net/core/dst.c
@@ -49,7 +49,7 @@ void dst_init(struct dst_entry *dst, struct dst_ops *ops,
 	      unsigned short flags)
 {
 	dst->dev = dev;
-	dev_hold(dev);
+	dev_hold_track(dev, &dst->dev_tracker, GFP_ATOMIC);
 	dst->ops = ops;
 	dst_init_metrics(dst, dst_default_metrics.metrics, true);
 	dst->expires = 0UL;
@@ -117,7 +117,7 @@ struct dst_entry *dst_destroy(struct dst_entry * dst)
 
 	if (dst->ops->destroy)
 		dst->ops->destroy(dst);
-	dev_put(dst->dev);
+	dev_put_track(dst->dev, &dst->dev_tracker);
 
 	lwtstate_put(dst->lwtstate);
 
@@ -159,8 +159,8 @@ void dst_dev_put(struct dst_entry *dst)
 	dst->input = dst_discard;
 	dst->output = dst_discard_out;
 	dst->dev = blackhole_netdev;
-	dev_hold(dst->dev);
-	dev_put(dev);
+	dev_replace_track(dev, blackhole_netdev, &dst->dev_tracker,
+			  GFP_ATOMIC);
 }
 EXPORT_SYMBOL(dst_dev_put);
 
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 243a0c52be42..843a7a3699fe 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -1531,8 +1531,9 @@ void rt_flush_dev(struct net_device *dev)
 			if (rt->dst.dev != dev)
 				continue;
 			rt->dst.dev = blackhole_netdev;
-			dev_hold(rt->dst.dev);
-			dev_put(dev);
+			dev_replace_track(dev, blackhole_netdev,
+					  &rt->dst.dev_tracker,
+					  GFP_ATOMIC);
 		}
 		spin_unlock_bh(&ul->lock);
 	}
@@ -2819,7 +2820,7 @@ struct dst_entry *ipv4_blackhole_route(struct net *net, struct dst_entry *dst_or
 		new->output = dst_discard_out;
 
 		new->dev = net->loopback_dev;
-		dev_hold(new->dev);
+		dev_hold_track(new->dev, &new->dev_tracker, GFP_ATOMIC);
 
 		rt->rt_is_input = ort->rt_is_input;
 		rt->rt_iif = ort->rt_iif;
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index f0d29fcb2094..ba4dc94d76d6 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -182,8 +182,9 @@ static void rt6_uncached_list_flush_dev(struct net *net, struct net_device *dev)
 
 			if (rt_dev == dev) {
 				rt->dst.dev = blackhole_netdev;
-				dev_hold(rt->dst.dev);
-				dev_put(rt_dev);
+				dev_replace_track(rt_dev, blackhole_netdev,
+						  &rt->dst.dev_tracker,
+						  GFP_ATOMIC);
 			}
 		}
 		spin_unlock_bh(&ul->lock);