Re: [patch 1/3] radix tree: RCU lockless read-side

[Nick Piggin <npiggin@suse.de>]

On Mon, Mar 13, 2006 at 08:34:53AM +0530, Balbir Singh wrote:
> On 3/11/06, Nick Piggin <nickpiggin@yahoo.com.au> wrote:
> > Balbir Singh wrote:
> > > <snip>
> > >
> > >>                if (slot->slots[i]) {
> > >>-                       results[nr_found++] = slot->slots[i];
> > >>+                       results[nr_found++] = &slot->slots[i];
> > >>                        if (nr_found == max_items)
> > >>                                goto out;
> > >>                }
> > >
> > >
> > > A quick clarification - Shouldn't accesses to slot->slots[i] above be
> > > protected using rcu_derefence()?
> > >
> >
> > I think we're safe here -- this is the _address_ of the pointer.
> > However, when dereferencing this address in _gang_lookup,
> > I think we do need rcu_dereference indeed.
> >
> 
> Yes, I saw the address operator, but we still derefence "slots" to get
> the address.
> 

OK, I reread what you wrote and I misunderstood you earlier I guess.
slot->slots[i] does dereference the pointer at the ith entry of slots,
but &slot->slots[i] does not, it will return the same thing as
slot->slots+i, which only dereferences 'slot' (which we've established
to be safe).

Even if &slot->slots[i] did, for some silly compiler, dereference the
pointer, we never actually see it or use it so it should be harmless.

Nick