Re: [PATCH] ipv4: only produce one record in fib_seq_show

Andi Kleen (ak@suse.de)
20 Oct 2002 07:58:26 +0200


"David S. Miller" <davem@redhat.com> writes:

> Note that this would prevent ipv4 from being hacked into being
> modular, but there are no plans at all to make modular ipv4 for 2.6.x
> at all so this is a valid transformation/cleanup.

(if ipv4 was made modular) Just reserve the per-cpu statistics space in the
main kernel when CONFIG_IPV4_MODULE is set. Then you can use it from the module
as needed.

> kernel/timer.c's main data structures desperately want to be per-cpu
> or allocated at boot time also. It, as has been noted often on this
> list, is actually more bloat than the ipv4 statistics stuff. :-)

.. and many more data structures that are still cacheline padded too.

(my favourite is in balance_dirty_pages_ratelimit which costs 4K overall
for a single per cpu counter)

-Andi
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/