On Thu, 2004-09-23 at 08:07, Herbert Xu wrote:

On Wed, Sep 22, 2004 at 08:08:40AM -0400, jamal wrote:

quoted

Since netlink is being used for a lot of things now, it may be time to
obsolete those assumptions. 

I'm not against changes.  But so far I haven't seen anything concrete
about what these new things are yet :)

So lets start by using the following logic:
1) If you made the socket buffers small enough compared to message
size/arrival rate, then an overrun will happen
corrollary:
2) If you made the message size/arrival fast enough relative to socket
size, an overrun will happen

If you agree that #1 is equivalent to #2, then you can experiment by #1
to see the issue.

quoted

Note also, theres a lot of wastage of that scarce sock buffer via page
sized skbs - its not as trivial to fix, but would go some way to improve
overrunning of the socket.

I wasn't aware of this scarcity problem.  Can you elaborate?

The NLM_GOODSIZE allocation done for each netlink skb,
In the case of the dump you dont apriori know how much space you need,
so it is fair. 

cheers,
jamal