Re: 2.6.35-rc2-git1 - lib/idr.c:605 invoked rcu_dereference_check() without protection!

["Paul E. McKenney" <paulmck@linux.vnet.ibm.com>]

On Tue, Jun 08, 2010 at 12:28:15AM -0400, Miles Lane wrote:
> On Mon, Jun 7, 2010 at 8:12 PM, Paul E. McKenney
> <paulmck@linux.vnet.ibm.com> wrote:
> > On Mon, Jun 07, 2010 at 02:23:17PM -0400, Miles Lane wrote:
> >> [    2.677955] [ INFO: suspicious rcu_dereference_check() usage. ]
> >> [    2.679089] ---------------------------------------------------
> >> [    2.680276] lib/idr.c:605 invoked rcu_dereference_check() without protection!
> >> [    2.681499]
> >> [    2.681500] other info that might help us debug this:
> >> [    2.681501]
> >> [    2.685509]
> >> [    2.685510] rcu_scheduler_active = 1, debug_locks = 1
> >> [    2.688221] 1 lock held by swapper/1:
> >> [    2.689587]  #0:  (mtd_table_mutex){+.+...}, at:
> >> [<ffffffff812bea45>] register_mtd_user+0x1a/0x69
> >> [    2.691096]
> >> [    2.691098] stack backtrace:
> >> [    2.694059] Pid: 1, comm: swapper Not tainted 2.6.35-rc2-git1 #8
> >> [    2.695601] Call Trace:
> >> [    2.697243]  [<ffffffff81064e9c>] lockdep_rcu_dereference+0x9d/0xa5
> >> [    2.698868]  [<ffffffff811b9c86>] idr_get_next+0x60/0x124
> >> [    2.700556]  [<ffffffff812be779>] __mtd_next_device+0x1b/0x1d
> >> [    2.702238]  [<ffffffff812bea7c>] register_mtd_user+0x51/0x69
> >> [    2.703964]  [<ffffffff816cca45>] init_mtdchar+0xb3/0xd3
> >> [    2.705686]  [<ffffffff816cc992>] ? init_mtdchar+0x0/0xd3
> >> [    2.707470]  [<ffffffff810001ef>] do_one_initcall+0x59/0x14e
> >> [    2.709255]  [<ffffffff816a768a>] kernel_init+0x144/0x1ce
> >> [    2.711082]  [<ffffffff81003054>] kernel_thread_helper+0x4/0x10
> >> [    2.712862]  [<ffffffff813ca480>] ? restore_args+0x0/0x30
> >> [    2.714647]  [<ffffffff816a7546>] ? kernel_init+0x0/0x1ce
> >> [    2.716415]  [<ffffffff81003050>] ? kernel_thread_helper+0x0/0x10
> >
> > This looks like a new one!  Does the following patch take care of it?
> >
> >                                                        Thanx, Paul
> >
> > ------------------------------------------------------------------------
> >
> > commit 2d54a6c31b72c902b09d365e9c66205a5c07e549
> > Author: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> > Date:   Mon Jun 7 17:09:45 2010 -0700
> >
> >    idr: fix RCU lockdep splat in idr_get_next()
> >
> >    Convert to rcu_dereference_raw() given that many callers may have many
> >    different locking models.
> >
> >    Located-by: Miles Lane <miles.lane@gmail.com>
> >    Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
> >
> > diff --git a/lib/idr.c b/lib/idr.c
> > index 2eb1dca..f099f25 100644
> > --- a/lib/idr.c
> > +++ b/lib/idr.c
> > @@ -599,7 +599,7 @@ void *idr_get_next(struct idr *idp, int *nextidp)
> >        /* find first ent */
> >        n = idp->layers * IDR_BITS;
> >        max = 1 << n;
> > -       p = rcu_dereference(idp->top);
> > +       p = rcu_dereference_raw(idp->top);
> >        if (!p)
> >                return NULL;
> >
> > @@ -607,7 +607,7 @@ void *idr_get_next(struct idr *idp, int *nextidp)
> >                while (n > 0 && p) {
> >                        n -= IDR_BITS;
> >                        *paa++ = p;
> > -                       p = rcu_dereference(p->ary[(id >> n) & IDR_MASK]);
> > +                       p = rcu_dereference_raw(p->ary[(id >> n) & IDR_MASK]);
> >                }
> >
> >                if (p) {
> >
> 
> Tested.  Looks good!

Thank you very much for both locating this one and for testing the fix!
I have added your Tested-by.

							Thanx, Paul