public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Nick Piggin <nickpiggin@yahoo.com.au>
To: linux-kernel <linux-kernel@vger.kernel.org>
Subject: [patch 5/5] radix tree: shrink
Date: Sun, 30 Oct 2005 11:57:58 +1100	[thread overview]
Message-ID: <43641A96.60007@yahoo.com.au> (raw)
In-Reply-To: <43641906.9030806@yahoo.com.au>

[-- Attachment #1: Type: text/plain, Size: 316 bytes --]

5/5

It always really bothered me that we don't shrink the tree when
truncating it. No actual performance results (though it wouldn't
be hard to concoct some brainless microbenchmark), but it is
simply nicer to do this IMO and will result in more consistent
performance in corner cases.

-- 
SUSE Labs, Novell Inc.


[-- Attachment #2: radix-tree-shrink.patch --]
[-- Type: text/plain, Size: 2414 bytes --]

Actually shrink the height of a radix tree when it is truncated.

Signed-off-by: Nick Piggin <npiggin@suse.de>

Index: linux-2.6/lib/radix-tree.c
===================================================================
--- linux-2.6.orig/lib/radix-tree.c
+++ linux-2.6/lib/radix-tree.c
@@ -251,7 +251,7 @@ int radix_tree_insert(struct radix_tree_
 	shift = (height-1) * RADIX_TREE_MAP_SHIFT;
 
 	offset = 0;			/* uninitialised var warning */
-	while (height > 0) {
+	do {
 		if (slot == NULL) {
 			/* Have to add a child node.  */
 			if (!(slot = radix_tree_node_alloc(root)))
@@ -269,18 +269,16 @@ int radix_tree_insert(struct radix_tree_
 		slot = node->slots[offset];
 		shift -= RADIX_TREE_MAP_SHIFT;
 		height--;
-	}
+	} while (height > 0);
 
 	if (slot != NULL)
 		return -EEXIST;
 
-	if (node) {
-		node->count++;
-		node->slots[offset] = item;
-		BUG_ON(tag_get(node, 0, offset));
-		BUG_ON(tag_get(node, 1, offset));
-	} else
-		root->rnode = item;
+	BUG_ON(!node);
+	node->count++;
+	node->slots[offset] = item;
+	BUG_ON(tag_get(node, 0, offset));
+	BUG_ON(tag_get(node, 1, offset));
 
 	return 0;
 }
@@ -678,6 +676,29 @@ radix_tree_gang_lookup_tag(struct radix_
 EXPORT_SYMBOL(radix_tree_gang_lookup_tag);
 
 /**
+ *	radix_tree_shrink    -    shrink height of a radix tree to minimal
+ *	@root		radix tree root
+ */
+static inline void radix_tree_shrink(struct radix_tree_root *root)
+{
+	/* try to shrink tree height */
+	while (root->height > 1 &&
+			root->rnode->count == 1 &&
+			root->rnode->slots[0]) {
+		struct radix_tree_node *to_free = root->rnode;
+
+		root->rnode = to_free->slots[0];
+		root->height--;
+		/* must only free zeroed nodes into the slab */
+		tag_clear(to_free, 0, 0);
+		tag_clear(to_free, 1, 0);
+		to_free->slots[0] = NULL;
+		to_free->count = 0;
+		radix_tree_node_free(to_free);
+	}
+}
+
+/**
  *	radix_tree_delete    -    delete an item from a radix tree
  *	@root:		radix tree root
  *	@index:		index key
@@ -753,8 +774,13 @@ void *radix_tree_delete(struct radix_tre
 	/* Now free the nodes we do not need anymore */
 	for (pathp = orig_pathp; pathp->node; pathp--) {
 		pathp->node->slots[pathp->offset] = NULL;
-		if (--pathp->node->count)
+		pathp->node->count--;
+
+		if (pathp->node->count) {
+			if (pathp->node == root->rnode)
+				radix_tree_shrink(root);
 			goto out;
+		}
 
 		/* Node with zero slots in use so free it */
 		radix_tree_node_free(pathp->node);

  reply	other threads:[~2005-10-30  0:56 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-10-30  0:41 [patches] lockless pagecache prep round 1 Nick Piggin
2005-10-30  0:43 ` [patch 1/5] i386 generic cmpxchg Nick Piggin
2005-10-30  0:44   ` [patch 2/5] atomic: atomic_cmpxchg Nick Piggin
2005-10-30  0:45     ` [patch 3/5] atomic: atomic_inc_not_zero Nick Piggin
2005-10-30  0:45       ` [patch 4/5] rcu file: use atomic primitives Nick Piggin
2005-10-30  0:47         ` [patch 5/5] atomic: dec_and_lock " Nick Piggin
2005-10-31 19:05       ` [patch 3/5] atomic: atomic_inc_not_zero Christoph Lameter
2005-11-01  4:34         ` Nick Piggin
2005-11-14 16:29       ` Paul Jackson
2005-11-14 21:48         ` Andrew Morton
2005-11-14 22:02           ` Paul Jackson
2005-11-14 23:09             ` Andrew Morton
2005-11-15  8:57               ` Nick Piggin
2005-10-30 20:12   ` [patch 1/5] i386 generic cmpxchg Zwane Mwaikambo
2005-10-31  1:29     ` Nick Piggin
2005-10-31  6:06       ` Zwane Mwaikambo
2005-10-31 19:09       ` Christoph Lameter
2005-10-30  0:48 ` [patche 1/5] radix tree: lookup_slot Nick Piggin
2005-10-30  0:49   ` [patch 2/5] radix tree: use prealloc Nick Piggin
2005-10-30  0:50     ` [patch 3/5] radix tree: cleanup Nick Piggin
2005-10-30  0:51       ` [patch 4/5] radix tree: clear_tags bail Nick Piggin
2005-10-30  0:57         ` Nick Piggin [this message]
2005-10-30  1:04 ` [patch 1/5] radix tree: lookup_slot Nick Piggin

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=43641A96.60007@yahoo.com.au \
    --to=nickpiggin@yahoo.com.au \
    --cc=linux-kernel@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox