A question of rtable creation

["KONG, Gangfeng" <gkong@IPRG.nokia.com>]

Hi,

Here, I was lost by line code in ip_route_output_slow():
	rth = dst_alloc(&ipv4_dst_ops);

It tries to allocate a rtable entry with ipv4_dst_ops, which is defined as
	struct dst_ops
	{
         	unsigned short          family;
         	unsigned short          protocol;
         	unsigned                gc_thresh;

         	int                     (*gc)(void);
         	struct dst_entry *  (*check)(struct dst_entry *, __u32 cookie);
         	struct dst_entry *     (*reroute)(struct dst_entry *,
                                            struct sk_buff *);
         	void                    (*destroy)(struct dst_entry *);
         	struct dst_entry *      (*negative_advice)(struct dst_entry *);
         	void                    (*link_failure)(struct sk_buff *);
         	int                     entry_size;

         	atomic_t                entries;
         	kmem_cache_t            *kmem_cachep;
	};

And it's initialized as :
	struct dst_ops ipv4_dst_ops = {
         	family:                 AF_INET,
         	protocol:               __constant_htons(ETH_P_IP),
         	gc:                     rt_garbage_collect,
         	check:                  ipv4_dst_check,
         	reroute:                ipv4_dst_reroute,
         	destroy:                ipv4_dst_destroy,
         	negative_advice:        ipv4_negative_advice,
         	link_failure:           ipv4_link_failure,
         	entry_size:             sizeof(struct rtable),
	};

It means:
	ipv4_dst_ops.kmem_cachep == NULL

Then in dst_alloc(), I found:
	void * dst_alloc(struct dst_ops * ops)
	{
         		struct dst_entry * dst;

         		if (ops->gc && atomic_read(&ops->entries) > ops->gc_thresh) {
                 		if (ops->gc())
                         		return NULL;
         		}
????====>	dst = kmem_cache_alloc(ops->kmem_cachep, SLAB_ATOMIC);
		...........
	}

This is equal to:
         dst = kmem_cache_alloc(NULL, SLAB_ATOMIC);

My question is how can it work, not crash ?? I found similar situation in 
several other
place, it seems an acceptable practice. Anyone could shed me some light?

Thanks
GF



--
Kernelnewbies: Help each other learn about the Linux kernel.
Archive:       http://mail.nl.linux.org/kernelnewbies/
FAQ:           http://kernelnewbies.org/faq/