Subject: Re: Kernel source tree specifications
To: None <ianzag@megasignal.com>
From: der Mouse <mouse@Rodents.Montreal.QC.CA>
List: tech-kern
Date: 06/18/2003 16:31:29
> [...adding device drivers...]

I don't have much to say to most of this.  But...

> For field busses like CANBUS or BITBUS IMO datagram socket scheme
> would be good.

> While i have done all underlying hardware dependant board raw
> interface, i do not know at this moment how to implement kernel <->
> userland interface.  Can you suggest me a good example of alike
> "non-standard" interface ?

Well, you mentioned sockets.  I added a new socket family to my kernel;
while it isn't what you want directly, it may be able to serve as a
model on which to build something.  (They're timer sockets.  The
functionality is basically like setitimer(), but appearing as fd input
rather than as signals.  This means that a single process can have
multiple outstanding timer events without needing hairy internal code
to support them all.  I did it partly for the functionality and partly
as an exercise in adding a new socket family.  Actually, if I've
understood kqueue correctly, it could replace much of this, but that's
neither here nor there at the moment.)

If you're interested, drop me a line and I can go into more detail
and/or send you the code.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign                  mouse@netbsd.org
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B