radix-tree: replace preallocated node array with linked list

Currently we use per-cpu array to hold pointers to preallocated nodes.
Let's replace it with linked list.  On x86_64 it saves 256 bytes in
per-cpu ELF section which may translate into freeing up 2MB of memory for
NR_CPUS==8192.

[akpm@linux-foundation.org: fix comment, coding style]
Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
diff --git a/lib/radix-tree.c b/lib/radix-tree.c
index 061550d..f9ebe1c 100644
--- a/lib/radix-tree.c
+++ b/lib/radix-tree.c
@@ -65,7 +65,8 @@
  */
 struct radix_tree_preload {
 	int nr;
-	struct radix_tree_node *nodes[RADIX_TREE_PRELOAD_SIZE];
+	/* nodes->private_data points to next preallocated node */
+	struct radix_tree_node *nodes;
 };
 static DEFINE_PER_CPU(struct radix_tree_preload, radix_tree_preloads) = { 0, };
 
@@ -197,8 +198,9 @@
 		 */
 		rtp = this_cpu_ptr(&radix_tree_preloads);
 		if (rtp->nr) {
-			ret = rtp->nodes[rtp->nr - 1];
-			rtp->nodes[rtp->nr - 1] = NULL;
+			ret = rtp->nodes;
+			rtp->nodes = ret->private_data;
+			ret->private_data = NULL;
 			rtp->nr--;
 		}
 		/*
@@ -257,17 +259,20 @@
 
 	preempt_disable();
 	rtp = this_cpu_ptr(&radix_tree_preloads);
-	while (rtp->nr < ARRAY_SIZE(rtp->nodes)) {
+	while (rtp->nr < RADIX_TREE_PRELOAD_SIZE) {
 		preempt_enable();
 		node = kmem_cache_alloc(radix_tree_node_cachep, gfp_mask);
 		if (node == NULL)
 			goto out;
 		preempt_disable();
 		rtp = this_cpu_ptr(&radix_tree_preloads);
-		if (rtp->nr < ARRAY_SIZE(rtp->nodes))
-			rtp->nodes[rtp->nr++] = node;
-		else
+		if (rtp->nr < RADIX_TREE_PRELOAD_SIZE) {
+			node->private_data = rtp->nodes;
+			rtp->nodes = node;
+			rtp->nr++;
+		} else {
 			kmem_cache_free(radix_tree_node_cachep, node);
+		}
 	}
 	ret = 0;
 out:
@@ -1463,15 +1468,16 @@
 {
        int cpu = (long)hcpu;
        struct radix_tree_preload *rtp;
+       struct radix_tree_node *node;
 
        /* Free per-cpu pool of perloaded nodes */
        if (action == CPU_DEAD || action == CPU_DEAD_FROZEN) {
                rtp = &per_cpu(radix_tree_preloads, cpu);
                while (rtp->nr) {
-                       kmem_cache_free(radix_tree_node_cachep,
-                                       rtp->nodes[rtp->nr-1]);
-                       rtp->nodes[rtp->nr-1] = NULL;
-                       rtp->nr--;
+			node = rtp->nodes;
+			rtp->nodes = node->private_data;
+			kmem_cache_free(radix_tree_node_cachep, node);
+			rtp->nr--;
                }
        }
        return NOTIFY_OK;