From: Allison Henderson <achender@linux.vnet.ibm.com>
Subject: [PATCH 2/6] ext4: add operations on status extent tree
Date: Wed, 18 Apr 2012 00:07:14 -0700
Message-ID: <1334732838-25102-3-git-send-email-achender@linux.vnet.ibm.com>
References: <1334732838-25102-1-git-send-email-achender@linux.vnet.ibm.com>
Cc: Allison Henderson <achender@linux.vnet.ibm.com>,
	Yongqiang Yang <xiaoqiangnk@gmail.com>
To: linux-ext4@vger.kernel.org
In-Reply-To: <1334732838-25102-1-git-send-email-achender@linux.vnet.ibm.com>
Sender: linux-ext4-owner@vger.kernel.org

This patch adds operations on a status extent tree.

Signed-off-by: Yongqiang Yang <xiaoqiangnk@gmail.com>
Signed-off-by: Allison Henderson <achender@linux.vnet.ibm.com>
---
:100644 100644 56fd8f86.. 55482c1... M	fs/ext4/Makefile
:000000 100644 0000000... 938e935... A	fs/ext4/status_extents.c
:100644 100644 cbf96ed... 415befd... M	fs/ext4/status_extents.h
 fs/ext4/Makefile         |    2 +-
 fs/ext4/status_extents.c |  417 ++++++++++++++++++++++++++++++++++++++++++++++
 fs/ext4/status_extents.h |   18 ++
 3 files changed, 436 insertions(+), 1 deletions(-)

diff --git a/fs/ext4/Makefile b/fs/ext4/Makefile
index 56fd8f8..55482c1 100644
--- a/fs/ext4/Makefile
+++ b/fs/ext4/Makefile
@@ -7,7 +7,7 @@ obj-$(CONFIG_EXT4_FS) += ext4.o
 ext4-y	:= balloc.o bitmap.o dir.o file.o fsync.o ialloc.o inode.o page-io.o \
 		ioctl.o namei.o super.o symlink.o hash.o resize.o extents.o \
 		ext4_jbd2.o migrate.o mballoc.o block_validity.o move_extent.o \
-		mmp.o indirect.o
+		mmp.o indirect.o status_extents.o
 
 ext4-$(CONFIG_EXT4_FS_XATTR)		+= xattr.o xattr_user.o xattr_trusted.o
 ext4-$(CONFIG_EXT4_FS_POSIX_ACL)	+= acl.o
diff --git a/fs/ext4/status_extents.c b/fs/ext4/status_extents.c
new file mode 100644
index 0000000..938e935
--- /dev/null
+++ b/fs/ext4/status_extents.c
@@ -0,0 +1,417 @@
+/*
+ *  fs/ext4/status_extents.c
+ *
+ * Written by Yongqiang Yang <xiaoqiangnk@gmail.com>
+ *
+ * Ext4 status extents core functions.
+ */
+#include <linux/rbtree.h>
+#include "ext4.h"
+#include "status_extents.h"
+#include "ext4_extents.h"
+
+/*
+ * status extent implementation for ext4.
+ *
+ *
+ * ==========================================================================
+ * Status extents encompass delayed extents and extent locks
+ *
+ * 1. Why delayed extent implementation ?
+ *
+ * Without delayed extent, ext4 identifies a delayed extent by looking up
+ * page cache, this has several deficiencies - complicated, buggy, and
+ * inefficient code.
+ *
+ * FIEMAP, SEEK_HOLE/DATA, bigalloc, punch hole and writeout all need to know if
+ * a block or a range of blocks are belonged to a delayed extent.
+ *
+ * Let us have a look at how they do without delayed extents implementation.
+ *   --	FIEMAP
+ *	FIEMAP looks up page cache to identify delayed allocations from holes.
+ *
+ *   --	SEEK_HOLE/DATA
+ *	SEEK_HOLE/DATA has the same problem as FIEMAP.
+ *
+ *   --	bigalloc
+ *	bigalloc looks up page cache to figure out if a block is already
+ *	under delayed allocation or not to determine whether quota reserving
+ *	is needed for the cluster.
+ *
+ *   -- punch hole
+ *	punch hole looks up page cache to identify a delayed extent.
+ *
+ *   --	writeout
+ *	Writeout looks up whole page cache to see if a buffer is mapped, If
+ *	there are not very many delayed buffers, then it is time comsuming.
+ *
+ * With delayed extents implementation, FIEMAP, SEEK_HOLE/DATA, bigalloc and
+ * writeout can figure out if a block or a range of blocks is under delayed
+ * allocation(belonged to a delayed extent) or not by searching the delayed
+ * extent tree.
+ *
+ *
+ * ==========================================================================
+ * 2. ext4 delayed extents impelmentation
+ *
+ *   --	delayed extent
+ *	A delayed extent is a range of blocks which are contiguous logically and
+ *	under delayed allocation.  Unlike extent in ext4, delayed extent in ext4
+ *	is a in-memory struct, there is no corresponding on-disk data.  There is
+ *	no limit on length of delayed extent, so a delayed extent can contain as
+ *	many blocks as they are contiguous logically.
+ *
+ *   --	delayed extent tree
+ *	Every inode has a delayed extent tree and all under delayed allocation
+ *	blocks are added to the tree as dealyed extents.  Delayed extents in
+ *	the tree are ordered by logical block no.
+ *
+ *   --	operations on a delayed extent tree
+ *	There are three operations on a delayed extent tree: find next delayed
+ *	extent, adding a space(a range of blocks) and removing a space.
+ *
+ *   --	race on a delayed extent tree
+ *	Delayed extent tree is protected inode->i_data_sem like extent tree.
+ *
+ *
+ * ==========================================================================
+ * 3. performance analysis
+ *   --	overhead
+ *	1. Apart from operations on a delayed extent tree, we need to
+ *	down_write(inode->i_data_sem) in delayed write path to maintain delayed
+ *	extent tree, this can have impact on parallel read-write and write-write
+ *
+ *	2. There is a cache extent for write access, so if writes are not very
+ *	random, adding space operaions are in O(1) time.
+ *
+ *   -- gain
+ *	3. Code is much simpler, more readable, more maintainable and
+ *      more efficient.
+ */
+
+static struct kmem_cache *ext4_se_cachep;
+
+int __init ext4_init_se(void)
+{
+	ext4_se_cachep = KMEM_CACHE(status_extent, SLAB_RECLAIM_ACCOUNT);
+	if (ext4_se_cachep == NULL)
+		return -ENOMEM;
+	return 0;
+}
+
+void ext4_exit_se(void)
+{
+	if (ext4_se_cachep)
+		kmem_cache_destroy(ext4_se_cachep);
+}
+
+void ext4_se_init_tree(struct ext4_se_tree *tree)
+{
+	tree->root = RB_ROOT;
+	tree->cache_se = NULL;
+}
+
+#ifdef SE_DEBUG
+static void ext4_se_print_tree(struct inode *inode)
+{
+	struct ext4_se_tree *tree;
+	struct rb_node *node;
+
+	printk(KERN_DEBUG "status extents for inode %lu:", inode->i_ino);
+	tree = &EXT4_I(inode)->i_se_tree;
+	node = rb_first(&tree->root);
+	while (node) {
+		struct status_extent *se;
+		se = rb_entry(node, struct status_extent, rb_node);
+		printk(KERN_DEBUG " [%u/%u)", se->start, se->len);
+		node = rb_next(node);
+	}
+	printk(KERN_DEBUG "\n");
+}
+#else
+#define ext4_se_print_tree(inode)
+#endif
+
+static inline ext4_lblk_t status_extent_end(struct status_extent *se)
+{
+	if (se->start + se->len < se->start)
+		return (ext4_lblk_t)-1;
+	return se->start + se->len;
+}
+
+/*
+ * search through the tree for an delayed_extent with a given offset.  If
+ * it can't be found, try to find next extent.
+ */
+static struct status_extent *__se_tree_search(struct rb_root *root,
+						ext4_lblk_t offset)
+{
+	struct rb_node *node = root->rb_node;
+	struct status_extent *se = NULL;
+
+	while (node) {
+		se = rb_entry(node, struct status_extent, rb_node);
+		if (offset < se->start)
+			node = node->rb_left;
+		else if (offset >= status_extent_end(se))
+			node = node->rb_right;
+		else
+			return se;
+	}
+
+	if (se && offset < se->start)
+		return se;
+
+	if (se && offset >= status_extent_end(se)) {
+		node = rb_next(&se->rb_node);
+		return node ? rb_entry(node, struct status_extent, rb_node) :
+			      NULL;
+	}
+
+	return NULL;
+}
+
+/*
+ * ext4_se_find_extent: find the 1st delayed extent covering @se->start
+ * if it exists, otherwise, the next extent after @se->start.
+ *
+ * @inode: the inode which owns delayed extents
+ * @se:
+ *
+ * Returns next block beyond the found extent.
+ * Delayed extent is returned via @se.
+ */
+ext4_lblk_t ext4_se_find_extent(struct inode *inode, struct status_extent *se)
+{
+	struct ext4_se_tree *tree;
+	struct status_extent *se1;
+	struct rb_node *node;
+
+	se->len = 0;
+	tree = &EXT4_I(inode)->i_se_tree;
+	se1 = __se_tree_search(&tree->root, se->start);
+	if (se1) {
+		tree->cache_se = se1;
+		se->start = se1->start;
+		se->len = se1->len;
+		node = rb_next(&se1->rb_node);
+		if (node) {
+			se1 = rb_entry(node, struct status_extent, rb_node);
+			return se1->start;
+		}
+	}
+
+	return EXT_MAX_BLOCKS;
+}
+
+static struct status_extent *
+ext4_se_alloc_extent(ext4_lblk_t start, ext4_lblk_t len)
+{
+	struct status_extent *se;
+	se = kmem_cache_alloc(ext4_se_cachep, GFP_NOFS);
+	if (se == NULL)
+		return NULL;
+	se->start = start;
+	se->len = len;
+	return se;
+}
+
+static void ext4_se_free_extent(struct status_extent *se)
+{
+	kmem_cache_free(ext4_se_cachep, se);
+}
+
+static void ext4_se_try_to_merge_left(struct ext4_se_tree *tree,
+				      struct status_extent *se)
+{
+	struct status_extent *se1;
+	struct rb_node *node;
+
+	node = rb_prev(&se->rb_node);
+	if (!node)
+		return;
+
+	se1 = rb_entry(node, struct status_extent, rb_node);
+	if (status_extent_end(se1) == se->start) {
+		se1->len += se->len;
+		rb_erase(&se->rb_node, &tree->root);
+		if (se == tree->cache_se)
+			tree->cache_se = se1;
+		ext4_se_free_extent(se);
+	}
+}
+
+static void ext4_se_try_to_merge_right(struct ext4_se_tree *tree,
+				       struct status_extent *se)
+{
+	struct status_extent *se1;
+	struct rb_node *node;
+
+	node = rb_next(&se->rb_node);
+	if (!node)
+		return;
+
+	se1 = rb_entry(node, struct status_extent, rb_node);
+	if (se1->start == status_extent_end(se)) {
+		se->len += se1->len;
+		rb_erase(node, &tree->root);
+		if (se1 == tree->cache_se)
+			tree->cache_se = se;
+		ext4_se_free_extent(se1);
+	}
+}
+
+/*
+ * ext4_se_add_space: adds a space to a delayed extent tree.
+ * Caller holds inode->i_data_sem.
+ *
+ * ext4_se_add_space is callyed by ext4_dealyed_write_begin and
+ * ext4_se_remove_space.
+ *
+ * Return 0 on success, error code on failure.
+ */
+int ext4_se_add_space(struct inode *inode, ext4_lblk_t offset, ext4_lblk_t len)
+{
+	struct ext4_se_tree *tree = &EXT4_I(inode)->i_se_tree;
+	struct rb_node **p = &tree->root.rb_node;
+	struct rb_node *parent = NULL;
+	struct status_extent *se;
+	ext4_lblk_t end = offset + len;
+
+	BUG_ON(end <= offset);
+
+	se = tree->cache_se;
+	se_debug("add [%u/%u) to status extent tree of inode %lu\n",
+		 offset, len, inode->i_ino);
+
+	if (se && status_extent_end(se) == offset) {
+		se_debug("cached by [%u/%u)\n", se->start, se->len);
+		se->len += len;
+		ext4_se_try_to_merge_right(tree, se);
+		goto out;
+	} else if (se && se->start == end) {
+		se_debug("cached by [%u/%u)\n", se->start, se->len);
+		se->start = offset;
+		se->len += len;
+		ext4_se_try_to_merge_left(tree, se);
+		goto out;
+	} else if (se && se->start <= offset &&
+		   status_extent_end(se) >= end) {
+		se_debug("cached by [%u/%u)\n", se->start, se->len);
+		goto out;
+	}
+
+	while (*p) {
+		parent = *p;
+		se = rb_entry(parent, struct status_extent, rb_node);
+
+		if (offset < se->start) {
+			if (end == se->start) {
+				se->len += len;
+				se->start = offset;
+				goto out;
+			}
+			p = &(*p)->rb_left;
+		} else if (offset >= status_extent_end(se)) {
+			if (status_extent_end(se) == offset) {
+				se->len += len;
+				goto out;
+			}
+			p = &(*p)->rb_right;
+		} else
+			goto out;
+	}
+
+	se = ext4_se_alloc_extent(offset, len);
+	if (!se)
+		return -ENOMEM;
+	rb_link_node(&se->rb_node, parent, p);
+	rb_insert_color(&se->rb_node, &tree->root);
+
+out:
+	tree->cache_se = se;
+	ext4_se_print_tree(inode);
+
+	return 0;
+}
+
+/*
+ * ext4_se_remove_space() removes a space from a delayed extent tree.
+ * Caller holds inode->i_data_sem.
+ *
+ * Return 0 on success, error code on failure.
+ */
+int ext4_se_remove_space(struct inode *inode, ext4_lblk_t offset,
+			 ext4_lblk_t len)
+{
+	struct rb_node *node;
+	struct ext4_se_tree *tree;
+	struct status_extent *se;
+	struct status_extent orig_se;
+	ext4_lblk_t len1, len2, end;
+
+	se_debug("remove [%u/%u) from status extent tree of inode %lu\n",
+		 offset, len, inode->i_ino);
+
+	end = offset + len;
+	BUG_ON(end <= offset);
+	tree = &EXT4_I(inode)->i_se_tree;
+	se = __se_tree_search(&tree->root, offset);
+	if (!se)
+		goto out;
+
+	/* Simply invalidate cache_se. */
+	tree->cache_se = NULL;
+
+	orig_se.start = se->start;
+	orig_se.len = se->len;
+	len1 = offset > se->start ? offset - se->start : 0;
+	len2 = status_extent_end(se) > end ?
+	       status_extent_end(se) - end : 0;
+	if (len1 > 0)
+		se->len = len1;
+	if (len2 > 0) {
+		if (len1 > 0) {
+			int err;
+			err = ext4_se_add_space(inode, end, len2);
+			if (err) {
+				se->start = orig_se.start;
+				se->len = orig_se.len;
+				return err;
+			}
+		} else {
+			se->start = end;
+			se->len = len2;
+		}
+		goto out;
+	}
+
+	if (len1 > 0) {
+		node = rb_next(&se->rb_node);
+		if (!node)
+			se = rb_entry(node, struct status_extent, rb_node);
+		else
+			se = NULL;
+	}
+
+	while (se && status_extent_end(se) <= end) {
+		node = rb_next(&se->rb_node);
+		rb_erase(&se->rb_node, &tree->root);
+		ext4_se_free_extent(se);
+		if (!node) {
+			se = NULL;
+			break;
+		}
+		se = rb_entry(node, struct status_extent, rb_node);
+	}
+
+	if (se && se->start < end) {
+		len1 = status_extent_end(se) - end;
+		se->start = end;
+		se->len = len1;
+	}
+
+out:
+	ext4_se_print_tree(inode);
+	return 0;
+}
diff --git a/fs/ext4/status_extents.h b/fs/ext4/status_extents.h
index cbf96ed..415befd 100644
--- a/fs/ext4/status_extents.h
+++ b/fs/ext4/status_extents.h
@@ -8,6 +8,13 @@
 #ifndef _EXT4_STATUS_EXTENTS_H
 #define _EXT4_STATUS_EXTENTS_H
 
+#define SE_DEBUG
+#ifdef SE_DEBUG
+#define se_debug(a...)         printk(a)
+#else
+#define se_debug(a...)
+#endif
+
 struct status_extent {
 	struct rb_node rb_node;
 	ext4_lblk_t start;	/* first block extent covers */
@@ -19,4 +26,15 @@ struct ext4_se_tree {
 	struct status_extent *cache_se;	/* recently accessed extent */
 };
 
+extern int __init ext4_init_se(void);
+extern void ext4_exit_se(void);
+extern void ext4_se_init_tree(struct ext4_se_tree *tree);
+
+extern int ext4_se_add_space(struct inode *inode, ext4_lblk_t start,
+				ext4_lblk_t len);
+extern int ext4_se_remove_space(struct inode *inode, ext4_lblk_t start,
+				ext4_lblk_t len);
+extern ext4_lblk_t ext4_se_find_extent(struct inode *inode,
+				struct status_extent *se);
+
 #endif /* _EXT4_STATUS_EXTENTS_H */
-- 
1.7.1