[PATCH] skbuff more likely/unlikely

[PATCH] skbuff more likely/unlikely

[Stephen Hemminger]

A couple more places where we can help by hinting the compiler
for 2.6.0-test6.  If we are pulling off header, is is likely there;
and skb alloc's succeed in the normal case.

Thought I saw an earlier similar patch, but here is my take on it.

diff -Nru a/include/linux/skbuff.h b/include/linux/skbuff.h
--- a/include/linux/skbuff.h	Thu Oct  2 10:01:36 2003
+++ b/include/linux/skbuff.h	Thu Oct  2 10:01:36 2003
@@ -885,7 +885,7 @@
  */
 static inline unsigned char *skb_pull(struct sk_buff *skb, unsigned int len)
 {
-	return (len > skb->len) ? NULL : __skb_pull(skb, len);
+	return unlikely(len > skb->len) ? NULL : __skb_pull(skb, len);
 }
 
 extern unsigned char *__pskb_pull_tail(struct sk_buff *skb, int delta);
@@ -901,7 +901,7 @@
 
 static inline unsigned char *pskb_pull(struct sk_buff *skb, unsigned int len)
 {
-	return (len > skb->len) ? NULL : __pskb_pull(skb, len);
+	return unlikely(len > skb->len) ? NULL : __pskb_pull(skb, len);
 }
 
 static inline int pskb_may_pull(struct sk_buff *skb, unsigned int len)
@@ -1052,7 +1052,7 @@
 					      int gfp_mask)
 {
 	struct sk_buff *skb = alloc_skb(length + 16, gfp_mask);
-	if (skb)
+	if (likely(skb))
 		skb_reserve(skb, 16);
 	return skb;
 }

Re: [PATCH] skbuff more likely/unlikely

[Mitchell Blank Jr]

Stephen Hemminger wrote:
> A couple more places where we can help by hinting the compiler
> for 2.6.0-test6.  If we are pulling off header, is is likely there;
> and skb alloc's succeed in the normal case.
> 
> Thought I saw an earlier similar patch, but here is my take on it.

Yes, my patch from a couple weeks ago does the same thing (but also
did a lot in skbuff.c)  I haven't had a chance to rediff and test
after the const parts went in.  Do you want to adopt the rest of the
changes?

Original patch:
  http://oss.sgi.com/projects/netdev/archive/2003-09/msg00036.html

-Mitch

Re: [PATCH] skbuff more likely/unlikely

[David S. Miller]

On Thu, 2 Oct 2003 19:34:31 -0700
Mitchell Blank Jr <mitch@sfgoth.com> wrote:

> Stephen Hemminger wrote:
> > A couple more places where we can help by hinting the compiler
> > for 2.6.0-test6.  If we are pulling off header, is is likely there;
> > and skb alloc's succeed in the normal case.
> > 
> > Thought I saw an earlier similar patch, but here is my take on it.
> 
> Yes, my patch from a couple weeks ago does the same thing (but also
> did a lot in skbuff.c)  I haven't had a chance to rediff and test
> after the const parts went in.  Do you want to adopt the rest of the
> changes?

I applied Stephen's patch here, you can post something relative
to that if you like.