[PATCH net-next] net: relax expensive skb_unclone() in iptunnel_handle_offloads()

[PATCH net-next] net: relax expensive skb_unclone() in iptunnel_handle_offloads()

[Eric Dumazet]

From: Eric Dumazet <edumazet@google.com>

Locally generated TCP GSO packets having to go through a GRE/SIT/IPIP
tunnel have to go through an expensive skb_unclone()

Reallocating skb->head is a lot of work.

Test should really check if a 'real clone' of the packet was done.

TCP does not care if the original gso_type is changed while the packet
travels in the stack.

This adds skb_header_unclone() which is a variant of skb_clone()
using skb_header_cloned() check instead of skb_cloned().

This variant can probably be used from other points.

Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 include/linux/skbuff.h    |   10 ++++++++++
 net/ipv4/ip_tunnel_core.c |    2 +-
 2 files changed, 11 insertions(+), 1 deletion(-)

diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index c84a5a1078c5528bf6fc84573f63f3c6f470ce8f..c413c588a24f854be9e4df78d8a6872b6b1ff9f3 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -1325,6 +1325,16 @@ static inline int skb_header_cloned(const struct sk_buff *skb)
 	return dataref != 1;
 }
 
+static inline int skb_header_unclone(struct sk_buff *skb, gfp_t pri)
+{
+	might_sleep_if(gfpflags_allow_blocking(pri));
+
+	if (skb_header_cloned(skb))
+		return pskb_expand_head(skb, 0, 0, pri);
+
+	return 0;
+}
+
 /**
  *	skb_header_release - release reference to header
  *	@skb: buffer to operate on
diff --git a/net/ipv4/ip_tunnel_core.c b/net/ipv4/ip_tunnel_core.c
index 786fa7ca28e0db00865cdf9fab472836866b7d6d..9118b0e640ba3a96850cf28aa8caab4e615254d9 100644
--- a/net/ipv4/ip_tunnel_core.c
+++ b/net/ipv4/ip_tunnel_core.c
@@ -157,7 +157,7 @@ int iptunnel_handle_offloads(struct sk_buff *skb,
 	}
 
 	if (skb_is_gso(skb)) {
-		err = skb_unclone(skb, GFP_ATOMIC);
+		err = skb_header_unclone(skb, GFP_ATOMIC);
 		if (unlikely(err))
 			return err;
 		skb_shinfo(skb)->gso_type |= gso_type_mask;

Re: [PATCH net-next] net: relax expensive skb_unclone() in iptunnel_handle_offloads()

[David Miller]

From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Sat, 30 Apr 2016 10:19:29 -0700

> From: Eric Dumazet <edumazet@google.com>
> 
> Locally generated TCP GSO packets having to go through a GRE/SIT/IPIP
> tunnel have to go through an expensive skb_unclone()
> 
> Reallocating skb->head is a lot of work.
> 
> Test should really check if a 'real clone' of the packet was done.
> 
> TCP does not care if the original gso_type is changed while the packet
> travels in the stack.
> 
> This adds skb_header_unclone() which is a variant of skb_clone()
> using skb_header_cloned() check instead of skb_cloned().
> 
> This variant can probably be used from other points.
> 
> Signed-off-by: Eric Dumazet <edumazet@google.com>

Nice observation, applied, thanks Eric.