Turns out IPv6 only has two functions which are marked inline
and where gcc is able to generate tail-calls, bridging doesn't
have a single one.
[IPV6]: Uninline netfilter okfns
Uninline netfilter okfns for those cases where gcc can generate tail-calls.
Before:
text data bss dec hex filename
8994153 1016524 524652 10535329 a0c1a1 vmlinux
After:
text data bss dec hex filename
8992761 1016524 524652 10533937 a0bc31 vmlinux
-------------------------------------------------------
-1392
All cases have been verified to generate tail-calls with and without netfilter.
Signed-off-by: Patrick McHardy <[EMAIL PROTECTED]>
---
commit 2a33a3e84be9c078de91ddf7496dbf026fe86ce2
tree e9401bf6eaafb1efeca6bb6f0e8ef013e1932513
parent 39533dafc9b0458fecebd2da36d682837c6ff9a2
author Patrick McHardy <[EMAIL PROTECTED]> Mon, 15 Oct 2007 10:25:34 +0200
committer Patrick McHardy <[EMAIL PROTECTED]> Mon, 15 Oct 2007 10:25:34 +0200
net/ipv6/ip6_input.c | 2 +-
net/ipv6/ip6_output.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/net/ipv6/ip6_input.c b/net/ipv6/ip6_input.c
index af355df..fac6f7f 100644
--- a/net/ipv6/ip6_input.c
+++ b/net/ipv6/ip6_input.c
@@ -149,7 +149,7 @@ out:
*/
-static inline int ip6_input_finish(struct sk_buff *skb)
+static int ip6_input_finish(struct sk_buff *skb)
{
struct inet6_protocol *ipprot;
struct sock *raw_sk;
diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
index 011082e..13565df 100644
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -70,7 +70,7 @@ static __inline__ void ipv6_select_ident(struct sk_buff *skb,
struct frag_hdr *f
spin_unlock_bh(&ip6_id_lock);
}
-static inline int ip6_output_finish(struct sk_buff *skb)
+static int ip6_output_finish(struct sk_buff *skb)
{
struct dst_entry *dst = skb->dst;
