Re: [PATCH] Early init for security modules

[Chris Wright]

* Christoph Hellwig ([e-mail]) wrote:
> On Mon, May 12, 2003 at 10:20:00PM -0700, Chris Wright wrote:
> > This is too late.  Those are just for order in do_initcalls() which is
> > well after some kernel threads have been created and filesystems have been
> > mounted, etc.  This patch allows statically linked modules to catch
> > the creation of such kernel objects and give them all consistent labels.
> 
> Patch looks fine to me.  Could you please make the initcalls mandatory
> for security modules and remove the module exports for the regioster
> functions so peop can't do the crappy check for each module whether it's
> already initialized stuff the early selinux for LSM versions did?

I absolutely agree the preconditions aren't nice, but not all security modules
need them.  I don't think disabling dynamic loading needs to be a
requirement for the initcall.

thanks,
-chris
-- 
Linux Security Modules     http://lsm.immunix.org     http://lsm.bkbits.net