Re: [mpls-linux-devel] mpls-linux on 64 bit kernel

[Jorge B. [DTI2] <jo...@dt...>]

El 18/05/2010 0:15, Chris Robson escribió:
> 
> Ashan
> 
> I'm currently trying to test this but having lock up problems.  However,
> I have cornered the problem to the following code snip.  Note the
> conditional compile statement.  This maybe the cause of the x86_64
> system locking up, but (and a big but) it also locks up with my x86_32
> systems as well.  Will be collecting more data tomorrow but if anyone
> has any ideas I'm all ears.  FYI, the lock up occurs when
> mpls_opcode_peek is called from mpls_skb_recv().
> 
> int mpls_opcode_peek(struct sk_buff *skb)
> {
>     u32 shim;
> 
> #define CAN_WE_ASSUME_32BIT_ALIGNED 0
> #if CAN_WE_ASSUME_32BIT_ALIGNED
>     shim = ntohl(*((u32*)&skb->network_header));
> #else
>     memcpy(&shim,skb->network_header,sizeof(u32));
>     shim = ntohl(shim);
> #endif
> 
>     if (!(MPLSCB(skb)->flag)) {
>         MPLSCB(skb)->ttl  = shim & 0xFF;
>         MPLSCB(skb)->flag = 1;
>     }
>     MPLSCB(skb)->bos   = (shim >> 8 ) & 0x1;
>     MPLSCB(skb)->exp   = (shim >> 9 ) & 0x7;
>     MPLSCB(skb)->label = (shim >> 12) & 0xFFFFF;
> 
>     return MPLS_RESULT_RECURSE;
> }

	Ok, that rang a bell. The problem is that in arches with BITS_PER_LONG > 32,
some of the members of the skbuff are not pointers (network_header,
transport_header, ...) but offsets.

	I'll cook a patch and send it to you to test.

	Regards,
		Jorge
-- 
==============================================================
Jorge Boncompte - Ingenieria y Gestion de RED
DTI2 - Desarrollo de la Tecnologia de las Comunicaciones
--------------------------------------------------------------
C/ Abogado Enriquez Barrios, 5   14004 CORDOBA (SPAIN)
Tlf: +34 957 761395 / FAX: +34 957 450380
==============================================================
- Sin pistachos no hay Rock & Roll...
- Without wicker a basket cannot be made.
==============================================================