net-gro: reset skb->truesize in napi_reuse_skb()

edumazet · davem330 · commit e33d0ba8047b · 2014-04-03T16:17:52.000-04:00
Recycling skb always had been very tough...

This time it appears GRO layer can accumulate skb-&gt;truesize
adjustments made by drivers when they attach a fragment to skb.

skb_gro_receive() can only subtract from skb-&gt;truesize the used part
of a fragment.

I spotted this problem seeing TcpExtPruneCalled and
TcpExtTCPRcvCollapsed that were unexpected with a recent kernel, where
TCP receive window should be sized properly to accept traffic coming
from a driver not overshooting skb-&gt;truesize.

Signed-off-by: Eric Dumazet &lt;edumazet@google.com&gt;
Signed-off-by: David S. Miller &lt;davem@davemloft.net&gt;
diff --git a/net/core/dev.c b/net/core/dev.c
@@ -4043,6 +4043,7 @@ static void napi_reuse_skb(struct napi_struct *napi, struct sk_buff *skb)
 	skb->vlan_tci = 0;
 	skb->dev = napi->dev;
 	skb->skb_iif = 0;
+	skb->truesize = SKB_TRUESIZE(skb_end_offset(skb));
 
 	napi->skb = skb;
 }

Original file line number	Diff line number	Diff line change
`@@ -4043,6 +4043,7 @@ static void napi_reuse_skb(struct napi_struct napi, struct sk_buff skb)`
`4043`	`4043`	`skb->vlan_tci = 0;`
`4044`	`4044`	`skb->dev = napi->dev;`
`4045`	`4045`	`skb->skb_iif = 0;`
	`4046`	`+ skb->truesize = SKB_TRUESIZE(skb_end_offset(skb));`
`4046`	`4047`
`4047`	`4048`	`napi->skb = skb;`
`4048`	`4049`	`}`