Netdev List
 help / color / mirror / Atom feed
From: Eric Dumazet <eric.dumazet@gmail.com>
To: Rui Ueyama <rui314@gmail.com>
Cc: netdev@vger.kernel.org
Subject: Re: [PATCH] net: filter: Convert the BPF VM to threaded code
Date: Sat, 30 Jul 2011 08:20:06 +0200	[thread overview]
Message-ID: <1312006806.2873.74.camel@edumazet-laptop> (raw)
In-Reply-To: <1312005899.2873.70.camel@edumazet-laptop>

Le samedi 30 juillet 2011 à 08:04 +0200, Eric Dumazet a écrit :

> We can remove one branch per BPF instruction with following patch :
> 
> diff --git a/net/core/filter.c b/net/core/filter.c
> index 36f975f..377f3ca 100644
> --- a/net/core/filter.c
> +++ b/net/core/filter.c
> @@ -119,16 +119,14 @@ unsigned int sk_run_filter(const struct sk_buff *skb,
>  	u32 tmp;
>  	int k;
>  
> +	fentry--;
>  	/*
>  	 * Process array of filter instructions.
>  	 */
> -	for (;; fentry++) {
> -#if defined(CONFIG_X86_32)
> +	for (;;) {
>  #define	K (fentry->k)
> -#else
> -		const u32 K = fentry->k;
> -#endif
>  
> +		fentry++;
>  		switch (fentry->code) {
>  		case BPF_S_ALU_ADD_X:
>  			A += X;
> 
> 

BTW, I tried to add one unreachable() in the default: branch, and gcc
4.5.2 generates interesting code :

It still does the compare and test, but both branches ends on same
location :

 348:	83 c3 08             	add    $0x8,%ebx
 34b:	66 83 3b 37          	cmpw   $0x37,(%ebx)
 34f:	76 07                	jbe    358 <sk_run_filter+0x28>
 351:	8d b4 26 00 00 00 00 	lea    0x0(%esi,%eiz,1),%esi
 358:	0f b7 03             	movzwl (%ebx),%eax
 35b:	ff 24 85 34 01 00 00 	jmp    *0x134(,%eax,4)


On a 64bit build and gcc 4.1.2, it even generates an infinite loop

 730:	66 83 3b 37          	cmpw   $0x37,(%rbx)
 734:	76 02                	jbe    738 <sk_run_filter+0x28>
 736:	eb fe                	jmp    736 <sk_run_filter+0x26>
 738:	0f b7 03             	movzwl (%rbx),%eax
 73b:	ff 24 c5 00 00 00 00 	jmpq   *0x0(,%rax,8)


diff --git a/net/core/filter.c b/net/core/filter.c
index 36f975f..89221e7 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -122,13 +122,11 @@ unsigned int sk_run_filter(const struct sk_buff *skb,
 	/*
 	 * Process array of filter instructions.
 	 */
-	for (;; fentry++) {
-#if defined(CONFIG_X86_32)
+	fentry--;
+	for (;;) {
 #define	K (fentry->k)
-#else
-		const u32 K = fentry->k;
-#endif
 
+		fentry++;
 		switch (fentry->code) {
 		case BPF_S_ALU_ADD_X:
 			A += X;
@@ -351,6 +349,7 @@ load_b:
 			continue;
 		}
 		default:
+			unreachable();
 			WARN_RATELIMIT(1, "Unknown code:%u jt:%u tf:%u k:%u\n",
 				       fentry->code, fentry->jt,
 				       fentry->jf, fentry->k);



  reply	other threads:[~2011-07-30  6:20 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-07-29  8:10 [PATCH] net: filter: Convert the BPF VM to threaded code Rui Ueyama
2011-07-29  9:30 ` Eric Dumazet
2011-07-30  5:09   ` Rui Ueyama
2011-07-30  6:04     ` Eric Dumazet
2011-07-30  6:20       ` Eric Dumazet [this message]
2011-07-30  9:55     ` Francois Romieu
2011-08-01 18:16 ` Hagen Paul Pfeifer
2011-08-01 18:37   ` Eric Dumazet
2011-08-02  0:57     ` David Miller
2011-08-09  5:00       ` Eric Dumazet
2011-08-09  8:29         ` Hagen Paul Pfeifer
2011-08-09  8:36           ` Eric Dumazet
2011-08-09  8:53             ` Hagen Paul Pfeifer

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1312006806.2873.74.camel@edumazet-laptop \
    --to=eric.dumazet@gmail.com \
    --cc=netdev@vger.kernel.org \
    --cc=rui314@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox