Re: Documentation/NotInKernel Was Re: oh woe

Alan Cox (alan@lxorguk.ukuu.org.uk)
Tue, 2 Jun 1998 23:45:35 +0100 (BST)


> > good design to put into a kernel. But that decision was reversed by someone
> > that apparently thought "oohhh, it doesn't have this feature. Lets add it."
> > How can we keep a greatly-needed-void empty?
>
> Part of the problem is probably that decisions get made about what not to
> do, and there is no mechanism to pass this information to new people.

And part of it as I pointed out to Donald is that things change. Without
kernel side filtering you can't dump 100Mbit ethernet nicely or run high speed
security matching tools right.

Also someone took BPF and did it right - no funny BPF device files, and the
ability of any user to push filters onto their connections - eg a user run
irc server can filter its own connections now.

BPF=N has a 0 byte cost BPF=m costs about 30 bytes and 4 bytes per socket.
So it costs almost nothing if its there and modular and nothing if its not
there.

Kind of makes Donald's argument look a bit silly IMHO

Alan

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu