Re: [PATCH v4 01/27] lockdep: Teach lockdep about "USED" <- "IN-NMI" inversions

[Frederic Weisbecker]

On Fri, Feb 21, 2020 at 09:25:11PM +0100, Peter Zijlstra wrote:
> On Fri, Feb 21, 2020 at 10:08:55AM -0500, Steven Rostedt wrote:
> > On Fri, 21 Feb 2020 14:34:17 +0100
> > Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> > 
> > > --- a/kernel/locking/lockdep.c
> > > +++ b/kernel/locking/lockdep.c
> > > @@ -379,13 +379,13 @@ void lockdep_init_task(struct task_struc
> > >  
> > >  void lockdep_off(void)
> > >  {
> > > -	current->lockdep_recursion++;
> > > +	current->lockdep_recursion += BIT(16);
> > >  }
> > >  EXPORT_SYMBOL(lockdep_off);
> > >  
> > >  void lockdep_on(void)
> > >  {
> > > -	current->lockdep_recursion--;
> > > +	current->lockdep_recursion -= BIT(16);
> > >  }
> > >  EXPORT_SYMBOL(lockdep_on);
> > >  
> > 
> > > +
> > > +static bool lockdep_nmi(void)
> > > +{
> > > +	if (current->lockdep_recursion & 0xFFFF)
> > 
> > Nitpick, but the association with bit 16 and this mask really should be
> > defined as a macro somewhere and not have hard coded numbers.
> 
> Right, I suppose I can do something like:
> 
> #define LOCKDEP_RECURSION_BITS	16
> #define LOCKDEP_OFF (1U << LOCKDEP_RECURSION_BITS)
> #define LOCKDEP_RECURSION_MASK (LOCKDEP_OFF - 1)

With that I'd say

Reviewed-by: Frederic Weisbecker <frederic@xxxxxxxxxx>