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

["Balbir Singh" <bsingharora@gmail.com>]

<snip>

>
> But we should have already rcu_dereference()ed "slot", right
> (in the loop above this one)? That means we are now able to
> dereference it, and the data at the other end will be valid.
>

Yes, but my confusion is about the following piece of code

<begin code>

       for ( ; height > 1; height--) {

               for (i = (index >> shift) & RADIX_TREE_MAP_MASK ;
                               i < RADIX_TREE_MAP_SIZE; i++) {
-                       if (slot->slots[i] != NULL)
+                       __s = rcu_dereference(slot->slots[i]);
+                       if (__s != NULL)
                               break;
                       index &= ~((1UL << shift) - 1);
                       index += 1UL << shift;
@@ -531,14 +550,14 @@ __lookup(struct radix_tree_root *root, v
                       goto out;

               shift -= RADIX_TREE_MAP_SHIFT;
-               slot = slot->slots[i];
+               slot = __s;
       }

       /* Bottom level: grab some items */
       for (i = index & RADIX_TREE_MAP_MASK; i < RADIX_TREE_MAP_SIZE; i++) {
               index++;
               if (slot->slots[i]) {
-                       results[nr_found++] = slot->slots[i];
+                       results[nr_found++] = &slot->slots[i];
                       if (nr_found == max_items)
                               goto out;
               }
<end code>

In the for loop, lets say __s is *not* NULL, we break from the loop.
In the loop below
slot->slots[i] is derefenced without rcu, __s is not used. Is that not
inconsistent?

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>