On Mon, Jan 12, 2009 at 11:56:07PM +1100, Herbert Xu (herbert@gondor.apana.org.au) wrote:
quoted text
> > As a long-term solution this sounds as the best case, but introduces
> > quite heavy overhead for the allocators. Right now we allocate
> > 1500+shared_info rounded up to the nearest power of the two (2k), but
> > then we will either need to have own network allocator (I have one :) or
> > allocate PAGE_SIZE+shared_info rounded up to the pwoer of the two (i.e.
> > 8k), which is unfeasible.
> 
> No that's not what I was suggesting.  The page split model allocates
> an skb with a very small head that accomodates only the headers.
> All payload is stored in the frags structure.
> 
> For 1500-byte packets, we can manage the payload area efficiently
> by dividing each allocated page into 2K chunks.  The page will
> then be automatically freed once all the 2K chunks on it have been
> freed through the page ref count mechanism.

That's the part I referred to as a network own allocator. We can have
multiple kmem_caches though for the popular MTUs and round up the
requested size otherwise.

-- 
	Evgeniy Polyakov
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html