Subject: Re: static functions
To: Martin Husemann <martin@duskware.de>
From: =?iso-8859-1?q?J=F6rn=20Seger?= <joern.seger@udo.edu>
List: tech-kern
Date: 05/10/2004 14:21:57
Hi Martin and all

Am Montag, 10. Mai 2004 12:30 schrieb Martin Husemann:
> No. They are static to avoid pollution of the kernel namespace and to
> make them "more private" to the driver/module. You can, of course, still
> pass function pointers to them around, which is what happens for interrupt
> handlers and callouts.
>
> Most (all?) of them could be made non-static, but why would you want to do
> that?

ok, maybe I'm completely on the wrong track. I have had some strange=20
behaviours on a testimplementation of a new protocol, which sends keep aliv=
e=20
packets periodicly. I experienced, that the kernel crashes (without direct=
=20
connection to my code and sadly without a panic or trace). When I used a=20
static version of my periodicly called function, this problem disappeared.=
=20
But when there is no real (fundamental) difference between static and=20
"normal" function calls, I really need to have a look at my code again!

The problem takes place, when I copy a lot of data e.g. via "scp". It's rea=
lly=20
hard to debug things like these, does someone had a similar problem and som=
e=20
ideas how to recognize/debug them!?

Thanks - J=F6rn