[bpf-next,v2,01/13] bpf: in bpf_skb_adjust_room avoid copy in tx fast path
diff mbox series

Message ID 20190322151504.89983-2-willemdebruijn.kernel@gmail.com
State Changes Requested
Headers show
Series
  • bpf tc tunneling
Related show

Commit Message

Willem de Bruijn March 22, 2019, 3:14 p.m. UTC
From: Willem de Bruijn <willemb@google.com>

bpf_skb_adjust_room calls skb_cow on grow.

This expensive operation can be avoided in the fast path when the only
other clone has released the header. This is the common case for TCP,
where one headerless clone is kept on the retransmit queue.

It is safe to do so even when touching the gso fields in skb_shinfo.
Regular tunnel encap with iptunnel_handle_offloads takes the same
optimization.

The tcp stack unclones in the unlikely case that it accesses these
fields through headerless clones packets on the retransmit queue (see
__tcp_retransmit_skb).

If any other clones are present, e.g., from packet sockets,
skb_cow_head returns the same value as skb_cow().

Signed-off-by: Willem de Bruijn <willemb@google.com>
---
 net/core/filter.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Patch
diff mbox series

diff --git a/net/core/filter.c b/net/core/filter.c
index d2511fe46db3..d21e1acdde29 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -2971,7 +2971,7 @@  static int bpf_skb_net_grow(struct sk_buff *skb, u32 len_diff)
 	if (skb_is_gso(skb) && !skb_is_gso_tcp(skb))
 		return -ENOTSUPP;
 
-	ret = skb_cow(skb, len_diff);
+	ret = skb_cow_head(skb, len_diff);
 	if (unlikely(ret < 0))
 		return ret;