[PATCH] fix iptables on systems with discontiguous processor ids

[Eric Dumazet]

Henrik Nordstrom a écrit :
> On Mon, 10 Oct 2005, Harald Welte wrote:
> 
>> I don't know why they are incomplete ;)  My guess is that they're mostly
>> targeted to per_cpu users.  Basically to solve the 'iptables' problem,
>> there are two ways:
> 
> 
> Or three.
> 
>   3) Keep an NUM_CPUS array of pointers to the per-cpu tables, allocated 
> separately per CPU instead of allocating a single large blob for all 
> CPUs. These pointers are updated using RCU, and indexed using the 
> physical CPU id.
> 

Yes, this is what I suggested some weeks ago with a patch.

As vmalloc_node() is not yet part of kernel, I even coded a substitute for it.
(per_cpu/kmalloc are not an option since the allocations done in iptables can 
be very large)

http://marc.theaimsgroup.com/?l=linux-netdev&m=112733887410796&w=2


> Cost: One (most likely cached) indirect memory access on each packet, 
> and slightly more complex update procedure (num_possible_cpus 
> allocations and pointers to update instead of just one).
> 
> Regards
> Henrik
> 
>