[PATCH] gre: fix ToS/DiffServ inherit bug

[PATCH] gre: fix ToS/DiffServ inherit bug

[Andreas Jaggi]

Fixes two bugs:
- ToS/DiffServ inheritance was unintentionally activated when using impair fixed ToS values
- ECN bit was lost during ToS/DiffServ inheritance

Signed-off-by: Andreas Jaggi <aj@open.ch>

--- vanilla-linux-2.6.29.4/net/ipv4/ip_gre.c	2009-05-19 01:52:34.000000000 +0200
+++ gre-dev-2/net/ipv4/ip_gre.c	2009-07-14 17:14:34.000000000 +0200
@@ -677,10 +677,10 @@
 	}
 
 	tos = tiph->tos;
-	if (tos&1) {
+	if (tos == 1) {
+		tos = 0;
 		if (skb->protocol == htons(ETH_P_IP))
 			tos = old_iph->tos;
-		tos &= ~1;
 	}
 
 	{

Re: [PATCH] gre: fix ToS/DiffServ inherit bug

[David Miller]

From: Andreas Jaggi <aj@open.ch>
Date: Tue, 14 Jul 2009 17:36:48 +0200

> Fixes two bugs:
> - ToS/DiffServ inheritance was unintentionally activated when using impair fixed ToS values
> - ECN bit was lost during ToS/DiffServ inheritance
> 
> Signed-off-by: Andreas Jaggi <aj@open.ch>
 ...
>  	tos = tiph->tos;
> -	if (tos&1) {
> +	if (tos == 1) {
> +		tos = 0;
>  		if (skb->protocol == htons(ETH_P_IP))
>  			tos = old_iph->tos;
> -		tos &= ~1;
>  	}

Note that this only works out because one wouldn't specify ECN
bits in the TOS to use for the tunnel.

But it demonstrates the outright danger of trying to use
"undefined" flag bits in protocol header fields to pass
private control flags into the driver from user interfaces.

Re: [PATCH] gre: fix ToS/DiffServ inherit bug

[David Miller]

From: David Miller <davem@davemloft.net>
Date: Tue, 14 Jul 2009 09:34:33 -0700 (PDT)

> From: Andreas Jaggi <aj@open.ch>
> Date: Tue, 14 Jul 2009 17:36:48 +0200
> 
>> Fixes two bugs:
>> - ToS/DiffServ inheritance was unintentionally activated when using impair fixed ToS values
>> - ECN bit was lost during ToS/DiffServ inheritance
>> 
>> Signed-off-by: Andreas Jaggi <aj@open.ch>

In case it wasn't clear, I've applied this patch.

Re: [PATCH] gre: fix ToS/DiffServ inherit bug

[Alexey Kuznetsov]

Hello!

On Tue, Jul 14, 2009 at 09:34:33AM -0700, David Miller wrote:
> Note that this only works out because one wouldn't specify ECN
> bits in the TOS to use for the tunnel.
> 
> But it demonstrates the outright danger of trying to use
> "undefined" flag bits in protocol header fields to pass
> private control flags into the driver from user interfaces.

Yes. My fault.

Well, in the form after the patch it will survive at least until
the next change of semantics of tos. :-)

Alexey