On Dec 21, 2011, at 8:28 PM, Pravin B Shelar wrote:
> diff --git a/datapath/linux/compat/include/net/dst.h 
> b/datapath/linux/compat/include/net/dst.h
> index f481a9d..5341daa 100644
> --- a/datapath/linux/compat/include/net/dst.h
> +++ b/datapath/linux/compat/include/net/dst.h
> @@ -14,4 +14,26 @@ static inline void skb_dst_drop(struct sk_buff *skb)
> 
> #endif
> 
> +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20)
> +/*
> + * On these kernels we have a fast mechanism to tell if the ARP cache for a
> + * particular destination has changed.
> + **/
> +#define HAVE_HH_SEQ
> +#endif
> +
> +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,27)
> +/*
> + * On these kernels we have a fast mechanism to tell if the routing table
> + * has changed.
> + **/
> +#define HAVE_RT_GENID

The symbols that we define shouldn't go in compatibility code.  We've had 
problems in the past where systems that integrated OVS need our symbols so they 
integrate those too but then we can't compile newer versions of OVS there due 
to duplicate symbols.  That's why we have the checksum and vlan files directly 
in the main source directory even though they are strictly compatibility code.  
So we should only have things that are upstream or we plan to propose upstream 
here.  In retrospect, we should do this for the genl_exec() stuff as well.
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev

Reply via email to