Re: [PATCH 1/2] lockdep: warn about lockdep disabling after kernel taint

[Ingo Molnar <mingo@elte.hu>]

* Frederic Weisbecker <fweisbec@gmail.com> wrote:

> On Fri, Apr 10, 2009 at 02:12:43PM +0200, Ingo Molnar wrote:
> > 
> > * Frederic Weisbecker <fweisbec@gmail.com> wrote:
> > 
> > > Impact: provide useful missing info for developers
> > > 
> > > Kernel taint can occur in several situations such as warnings,
> > > load of prorietary or staging modules, bad page, etc...
> > > 
> > > But when such taint happens, a developer might still be working on
> > > the kernel, expecting that lockdep is still enabled. But a taint
> > > disables lockdep without ever warning about it.
> > > Such a kernel behaviour doesn't really help for kernel development.
> > > 
> > > This patch adds this missing warning.
> > > 
> > > Since the taint is done most of the time after the main message that
> > > explain the real source issue, it seems safe to warn about it inside
> > > add_taint() so that it appears at last, without hurting the main
> > > information.
> > > 
> > > Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
> > > 
> > > diff --git a/kernel/panic.c b/kernel/panic.c
> > > index 3fd8c5b..9e7420a 100644
> > > --- a/kernel/panic.c
> > > +++ b/kernel/panic.c
> > > @@ -213,8 +213,14 @@ unsigned long get_taint(void)
> > >  
> > >  void add_taint(unsigned flag)
> > >  {
> > > -	/* can't trust the integrity of the kernel anymore: */
> > > -	debug_locks = 0;
> > > +	/*
> > > +	 * Can't trust the integrity of the kernel anymore.
> > > +	 * We don't call directly debug_locks_off() because the issue
> > > +	 * is not necessarily serious enough to set oops_in_progress to 1
> > > +	 */
> > > +	if (xchg(&debug_locks, 0))
> > > +		printk(KERN_WARNING "Disabling lockdep due to kernel taint\n");
> > > +
> > 
> > nice idea - but please use the proper debug_locks_off() construct 
> > instead of an open-coded xchg(). Something like:
> > 
> > 	if (debug_locks_off())
> > 		printk(...);
> > 
> > should do the trick.
> > 
> > 	Ingo
> 
> Yeah, I first wanted to do so but was shy about the 
> oops_in_progress = 1 inside debug_locks_off(). Isn't it a problem?

hm, indeed. How about providing a __debug_locks_off() primitive that 
only does the xchg and none of the oops_in_progress and 
console_verbose() calls?

	Ingo