On Mon, Dec 29, 2003 at 06:53:59PM +0200, Sirotkin, Alexander wrote:

> Anybody tried to implement similar approach ?
> Any thoughts why this would (or would not) work ?

It's a nice idea in theory, but the reality is that as the code
currently stands, the upper layers 'hijack' the skb->destructor
without regards to whether it was set previously, causing your memory
to leak. See skb_set_owner_w(), skb_set_owner_r(). 

I wrote a patch to allow chaining of skb destructors, which was fairly
simple and would allow both the driver and the upper layers to set
their destructors. If there's any interet, I could probably resurrect
it.

Cheers, 
Muli
-- 
Muli Ben-Yehuda
http://www.mulix.org | http://mulix.livejournal.com/

"the nucleus of linux oscillates my world" - gccbot@#offtopic