Re: [RFC][PATCH 4/7] lockdep: Seperate lock ids for read/write acquires

[Steven Rostedt <rostedt@goodmis.org>]

On Mon, 2011-04-18 at 18:49 +0200, Peter Zijlstra wrote:
> On Mon, 2011-04-18 at 12:46 -0400, Steven Rostedt wrote:
>  
> > >  void lockdep_off(void)
> > > @@ -1988,6 +1988,9 @@ static void check_chain_key(struct task_
> > >               if (DEBUG_LOCKS_WARN_ON(id >= MAX_LOCKDEP_KEYS))
> > >                       return;
> > >  
> > > +             if (is_read(hlock->rw_state))
> > > +                     id += MAX_LOCKDEP_KEYS;
> > 
> > Again, isn't this about recursive reads? Or am I just confused ;) 
> 
> So what we do here is split off the write chain, the above could have
> been writeen if (!is_write()) to clarify that.

Or just add a comment ;)

-- Steve

> 
> Everything except recursive read validation will traverse both chains,
> the recursive read validation will only traverse the write chains and
> ignore the combined read/recursive-read chain.