[PATCH 1/2] ext4_get_blocks_wrap take the truncate_mutex early.

[PATCH 1/2] ext4_get_blocks_wrap take the truncate_mutex early.

[Aneesh Kumar K.V]

When doing a migrate from ext3 to ext4 inode we need to make sure the test
for inode type and walking inode data happens inside  lock. To make this
happen move truncate_mutex early before checking the i_flags.


This actually should enable us to remove the verify_chain().

I haven't done any performance benchmark with this change.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
---
 fs/ext4/extents.c       |    7 +++--
 fs/ext4/inode.c         |   69 +++++-----------------------------------------
 include/linux/ext4_fs.h |   15 +++++++---
 3 files changed, 23 insertions(+), 68 deletions(-)

diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index b9ce241..8193e97 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -1977,6 +1977,10 @@ void ext4_ext_release(struct super_block *sb)
 #endif
 }
 
+/*
+ * Need to be called with
+ * mutex_lock(&EXT4_I(inode)->truncate_mutex);
+ */
 int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
 			ext4_fsblk_t iblock,
 			unsigned long max_blocks, struct buffer_head *bh_result,
@@ -1991,7 +1995,6 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
 	__clear_bit(BH_New, &bh_result->b_state);
 	ext_debug("blocks %d/%lu requested for inode %u\n", (int) iblock,
 			max_blocks, (unsigned) inode->i_ino);
-	mutex_lock(&EXT4_I(inode)->truncate_mutex);
 
 	/* check in cache */
 	goal = ext4_ext_in_cache(inode, iblock, &newex);
@@ -2127,8 +2130,6 @@ out2:
 		ext4_ext_drop_refs(path);
 		kfree(path);
 	}
-	mutex_unlock(&EXT4_I(inode)->truncate_mutex);
-
 	return err ? err : allocated;
 }
 
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 8416fa2..255db76 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -243,13 +243,6 @@ static inline void add_chain(Indirect *p, struct buffer_head *bh, __le32 *v)
 	p->bh = bh;
 }
 
-static int verify_chain(Indirect *from, Indirect *to)
-{
-	while (from <= to && from->key == *from->p)
-		from++;
-	return (from > to);
-}
-
 /**
  *	ext4_block_to_path - parse the block number into array of offsets
  *	@inode: inode in question (we are only interested in its superblock)
@@ -344,10 +337,11 @@ static int ext4_block_to_path(struct inode *inode,
  *		(pointer to last triple returned, *@err == 0)
  *	or when it gets an IO error reading an indirect block
  *		(ditto, *@err == -EIO)
- *	or when it notices that chain had been changed while it was reading
- *		(ditto, *@err == -EAGAIN)
  *	or when it reads all @depth-1 indirect blocks successfully and finds
  *	the whole chain, all way to the data (returns %NULL, *err == 0).
+ *
+ *      Need to be called with
+ *      mutex_lock(&EXT4_I(inode)->truncate_mutex)
  */
 static Indirect *ext4_get_branch(struct inode *inode, int depth, int *offsets,
 				 Indirect chain[4], int *err)
@@ -365,9 +359,6 @@ static Indirect *ext4_get_branch(struct inode *inode, int depth, int *offsets,
 		bh = sb_bread(sb, le32_to_cpu(p->key));
 		if (!bh)
 			goto failure;
-		/* Reader: pointers */
-		if (!verify_chain(chain, p))
-			goto changed;
 		add_chain(++p, bh, (__le32*)bh->b_data + *++offsets);
 		/* Reader: end */
 		if (!p->key)
@@ -375,10 +366,6 @@ static Indirect *ext4_get_branch(struct inode *inode, int depth, int *offsets,
 	}
 	return NULL;
 
-changed:
-	brelse(bh);
-	*err = -EAGAIN;
-	goto no_block;
 failure:
 	*err = -EIO;
 no_block:
@@ -782,6 +769,10 @@ err_out:
  * return > 0, # of blocks mapped or allocated.
  * return = 0, if plain lookup failed.
  * return < 0, error case.
+ *
+ *
+ * Need to be called with
+ * mutex_lock(&EXT4_I(inode)->truncate_mutex)
  */
 int ext4_get_blocks_handle(handle_t *handle, struct inode *inode,
 		sector_t iblock, unsigned long maxblocks,
@@ -819,18 +810,6 @@ int ext4_get_blocks_handle(handle_t *handle, struct inode *inode,
 		while (count < maxblocks && count <= blocks_to_boundary) {
 			ext4_fsblk_t blk;
 
-			if (!verify_chain(chain, partial)) {
-				/*
-				 * Indirect block might be removed by
-				 * truncate while we were reading it.
-				 * Handling of that case: forget what we've
-				 * got now. Flag the err as EAGAIN, so it
-				 * will reread.
-				 */
-				err = -EAGAIN;
-				count = 0;
-				break;
-			}
 			blk = le32_to_cpu(*(chain[depth-1].p + count));
 
 			if (blk == first_block + count)
@@ -838,44 +817,13 @@ int ext4_get_blocks_handle(handle_t *handle, struct inode *inode,
 			else
 				break;
 		}
-		if (err != -EAGAIN)
-			goto got_it;
+		goto got_it;
 	}
 
 	/* Next simple case - plain lookup or failed read of indirect block */
 	if (!create || err == -EIO)
 		goto cleanup;
 
-	mutex_lock(&ei->truncate_mutex);
-
-	/*
-	 * If the indirect block is missing while we are reading
-	 * the chain(ext4_get_branch() returns -EAGAIN err), or
-	 * if the chain has been changed after we grab the semaphore,
-	 * (either because another process truncated this branch, or
-	 * another get_block allocated this branch) re-grab the chain to see if
-	 * the request block has been allocated or not.
-	 *
-	 * Since we already block the truncate/other get_block
-	 * at this point, we will have the current copy of the chain when we
-	 * splice the branch into the tree.
-	 */
-	if (err == -EAGAIN || !verify_chain(chain, partial)) {
-		while (partial > chain) {
-			brelse(partial->bh);
-			partial--;
-		}
-		partial = ext4_get_branch(inode, depth, offsets, chain, &err);
-		if (!partial) {
-			count++;
-			mutex_unlock(&ei->truncate_mutex);
-			if (err)
-				goto cleanup;
-			clear_buffer_new(bh_result);
-			goto got_it;
-		}
-	}
-
 	/*
 	 * Okay, we need to do block allocation.  Lazily initialize the block
 	 * allocation info here if necessary
@@ -917,7 +865,6 @@ int ext4_get_blocks_handle(handle_t *handle, struct inode *inode,
 	*/
 	if (!err && extend_disksize && inode->i_size > ei->i_disksize)
 		ei->i_disksize = inode->i_size;
-	mutex_unlock(&ei->truncate_mutex);
 	if (err)
 		goto cleanup;
 
diff --git a/include/linux/ext4_fs.h b/include/linux/ext4_fs.h
index de1f9f7..ebd6f86 100644
--- a/include/linux/ext4_fs.h
+++ b/include/linux/ext4_fs.h
@@ -988,11 +988,18 @@ ext4_get_blocks_wrap(handle_t *handle, struct inode *inode, sector_t block,
 			unsigned long max_blocks, struct buffer_head *bh,
 			int create, int extend_disksize)
 {
-	if (EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL)
-		return ext4_ext_get_blocks(handle, inode, block, max_blocks,
+	int retval;
+	mutex_lock(&EXT4_I(inode)->truncate_mutex);
+	if (EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL) {
+		retval = ext4_ext_get_blocks(handle, inode, block, max_blocks,
 					bh, create, extend_disksize);
-	return ext4_get_blocks_handle(handle, inode, block, max_blocks, bh,
-					create, extend_disksize);
+	} else {
+		retval = ext4_get_blocks_handle(handle, inode, block, max_blocks,
+					bh, create, extend_disksize);
+	}
+	mutex_unlock(&EXT4_I(inode)->truncate_mutex);
+
+	return retval;
 }
 
 
-- 
1.5.2.2.571.ge1341-dirty

[PATCH 2/2] Add EXT4_IOC_MIGRATE ioctl

[Aneesh Kumar K.V]

From: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>

The below patch add ioctl for migrating ext3 indirect block mapped inode
to ext4 extent mapped inode.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
---
 fs/ext4/Makefile        |    2 +-
 fs/ext4/ioctl.c         |    3 +
 fs/ext4/migrate.c       |  655 +++++++++++++++++++++++++++++++++++++++++++++++
 include/linux/ext4_fs.h |    4 +
 4 files changed, 663 insertions(+), 1 deletions(-)
 create mode 100644 fs/ext4/migrate.c

diff --git a/fs/ext4/Makefile b/fs/ext4/Makefile
index ae6e7e5..d5fd80b 100644
--- a/fs/ext4/Makefile
+++ b/fs/ext4/Makefile
@@ -6,7 +6,7 @@ obj-$(CONFIG_EXT4DEV_FS) += ext4dev.o
 
 ext4dev-y	:= balloc.o bitmap.o dir.o file.o fsync.o ialloc.o inode.o \
 		   ioctl.o namei.o super.o symlink.o hash.o resize.o extents.o \
-		   ext4_jbd2.o
+		   ext4_jbd2.o migrate.o
 
 ext4dev-$(CONFIG_EXT4DEV_FS_XATTR)	+= xattr.o xattr_user.o xattr_trusted.o
 ext4dev-$(CONFIG_EXT4DEV_FS_POSIX_ACL)	+= acl.o
diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c
index 500567d..6a6d72b 100644
--- a/fs/ext4/ioctl.c
+++ b/fs/ext4/ioctl.c
@@ -248,6 +248,9 @@ flags_err:
 		return err;
 	}
 
+	case EXT4_IOC_MIGRATE:
+		return ext4_ext_migrate(inode, filp, cmd, arg);
+
 	default:
 		return -ENOTTY;
 	}
diff --git a/fs/ext4/migrate.c b/fs/ext4/migrate.c
new file mode 100644
index 0000000..523de12
--- /dev/null
+++ b/fs/ext4/migrate.c
@@ -0,0 +1,655 @@
+/*
+ * Copyright IBM Corporation, 2007
+ * Author Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of version 2.1 of the GNU Lesser General Public License
+ * as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ */
+
+#include <linux/module.h>
+#include <linux/ext4_jbd2.h>
+#include <linux/ext4_fs_extents.h>
+
+struct list_blocks_struct {
+	int first_block, last_block;
+	ext4_fsblk_t first_pblock, last_pblock;
+};
+
+/* will go away */
+static void ext4_ext_store_pblock(struct ext4_extent *ex, ext4_fsblk_t pb)
+{
+	ex->ee_start = cpu_to_le32((unsigned long) (pb & 0xffffffff));
+	ex->ee_start_hi = cpu_to_le16((unsigned long) ((pb >> 31) >> 1) & 0xffff);
+}
+static int ext4_get_credit(struct inode *inode, struct ext4_ext_path *path)
+{
+	int depth, needed;
+
+	/* probably there is space in leaf? */
+	if (path) {
+		depth = ext_depth(inode);
+		if (le16_to_cpu(path[depth].p_hdr->eh_entries)
+				< le16_to_cpu(path[depth].p_hdr->eh_max))
+			/*
+			 * Since we are adding all the extent entries together
+			 * we already have credit alloted for writing this block
+			 */
+			return 0;
+	}
+
+	/*
+	 * given 32-bit logical block (4294967296 blocks), max. tree
+	 * can be 4 levels in depth -- 4 * 340^4 == 53453440000.
+	 * Let's also add one more level for imbalance.
+	 */
+	depth = 5;
+
+	/* allocation of new data block(s) */
+	needed = 2;
+
+	/*
+	 * tree can be full, so it would need to grow in depth:
+	 * we need one credit to modify old root, credits for
+	 * new root will be added in split accounting
+	 */
+	needed += 1;
+
+	/*
+	 * Index split can happen, we would need:
+	 *    allocate intermediate indexes (bitmap + group)
+	 *  + change two blocks at each level, but root (already included)
+	 */
+	needed += (depth * 2) + (depth * 2);
+
+
+	return needed;
+
+}
+
+static int finish_range(handle_t *handle, struct inode *inode,
+				struct list_blocks_struct *lb)
+
+{
+	int retval = 0, needed;
+	struct ext4_extent newext;
+	struct ext4_ext_path *path;
+	if (lb->first_pblock == 0)
+		return 0;
+
+	/* Add the extent to temp inode*/
+	newext.ee_block = cpu_to_le32(lb->first_block);
+	newext.ee_len   = cpu_to_le16(lb->last_block - lb->first_block +1);
+	ext4_ext_store_pblock(&newext, lb->first_pblock);
+	path = ext4_ext_find_extent(inode, lb->first_block, NULL);
+
+	if (IS_ERR(path)) {
+		retval = PTR_ERR(path);
+		goto err_out;
+	}
+
+	/* Calculate the credit needed to inserting this extent */
+	needed = ext4_get_credit(inode, path);
+
+	/*
+	 * Make sure the credit we accumalated is not really high
+	 */
+
+	if (needed && handle->h_buffer_credits >= EXT4_RESERVE_TRANS_BLOCKS) {
+
+		/*
+		 * +2 is needed for orignal inode write and
+		 * super block modification
+		 */
+
+		ext4_mark_inode_dirty(handle, inode);
+		retval = ext4_journal_restart(handle, needed + 2);
+		if (retval)
+			goto err_out;
+
+	}
+
+	if (needed && (retval = ext4_journal_extend(handle, needed)) != 0) {
+		/*
+		 * IF not able to extend the journal restart the journal
+		 *
+		 * +2 is needed for orignal inode write and
+		 * super block modification
+		 */
+		ext4_mark_inode_dirty(handle, inode);
+		retval = ext4_journal_restart(handle, needed + 2);
+		if (retval)
+			goto err_out;
+	}
+
+	retval = ext4_ext_insert_extent(handle, inode, path, &newext);
+
+err_out:
+	lb->first_pblock = 0;
+	return retval;
+}
+static int update_extent_range(handle_t *handle, struct inode *inode,
+				ext4_fsblk_t pblock, int blk_num,
+				struct list_blocks_struct *lb)
+{
+	int retval;
+
+	/*
+	 * See if we can add on to the existing range (if it exists)
+	 */
+	if (lb->first_pblock &&
+		(lb->last_pblock+1 == pblock) &&
+		(lb->last_block+1 == blk_num)) {
+		lb->last_pblock = pblock;
+		lb->last_block = blk_num;
+		return 0;
+	}
+	/*
+	 * Start a new range.
+	 */
+	retval = finish_range(handle, inode, lb);
+	lb->first_pblock = lb->last_pblock = pblock;
+	lb->first_block = lb->last_block = blk_num;
+
+	return retval;
+
+}
+
+static int update_ind_extent_range(handle_t *handle, struct inode *inode,
+					ext4_fsblk_t pblock, int *blk_nump,
+					struct list_blocks_struct *lb)
+{
+	struct buffer_head *bh;
+	__le32 *i_data;
+	int i, retval = 0;
+	int blk_count = *blk_nump;
+	unsigned long max_entries = inode->i_sb->s_blocksize >> 2;
+
+	if (!pblock) {
+		/* Only update the file block number */
+		*blk_nump += max_entries;
+		return 0;
+	}
+
+	bh = sb_bread(inode->i_sb, pblock);
+	if (!bh)
+		return -EIO;
+
+	i_data = (__le32 *)bh->b_data;
+
+	for (i = 0; i < max_entries; i++, blk_count++) {
+		if (i_data[i]) {
+			retval = update_extent_range(handle, inode,
+						le32_to_cpu(i_data[i]),
+						blk_count, lb);
+			if (retval)
+				break;
+		}
+	}
+
+	/* Update the file block number */
+	*blk_nump = blk_count;
+	brelse(bh);
+	return retval;
+
+}
+static int update_dind_extent_range(handle_t *handle, struct inode *inode,
+					ext4_fsblk_t pblock, int *blk_nump,
+					struct list_blocks_struct *lb)
+{
+	struct buffer_head *bh;
+	__le32 *i_data;
+	int i, retval = 0;
+	int blk_count = *blk_nump;
+	unsigned long max_entries = inode->i_sb->s_blocksize >> 2;
+
+	if (!pblock) {
+		/* Only update the file block number */
+		*blk_nump += max_entries * max_entries;
+		return 0;
+	}
+
+	bh = sb_bread(inode->i_sb, pblock);
+	if (!bh)
+		return -EIO;
+
+	i_data = (__le32 *)bh->b_data;
+
+	for (i = 0; i < max_entries; i++) {
+		if (i_data[i]) {
+			retval = update_ind_extent_range(handle, inode,
+						le32_to_cpu(i_data[i]),
+						&blk_count, lb);
+			if (retval)
+				break;
+		} else {
+			/* Only update the file block number */
+			blk_count += max_entries;
+		}
+	}
+
+	/* Update the file block number */
+	*blk_nump = blk_count;
+	brelse(bh);
+	return retval;
+
+}
+static int update_tind_extent_range(handle_t *handle, struct inode *inode,
+					ext4_fsblk_t pblock, int *blk_nump,
+					struct list_blocks_struct *lb)
+{
+	struct buffer_head *bh;
+	__le32 *i_data;
+	int i, retval = 0;
+	int blk_count = *blk_nump;
+	unsigned long max_entries = inode->i_sb->s_blocksize >> 2;
+
+	if (!pblock) {
+		/* Only update the file block number */
+		*blk_nump += max_entries * max_entries * max_entries;
+		return 0;
+	}
+
+	bh = sb_bread(inode->i_sb, pblock);
+	if (!bh)
+		return -EIO;
+
+	i_data = (__le32 *)bh->b_data;
+
+	for (i = 0; i < max_entries; i++) {
+		if (i_data[i]) {
+			retval = update_dind_extent_range(handle, inode,
+						le32_to_cpu(i_data[i]),
+						&blk_count, lb);
+			if (retval)
+				break;
+		} else {
+			/* Only update the file block number */
+			blk_count += max_entries * max_entries;
+		}
+	}
+
+	/* Update the file block number */
+	*blk_nump = blk_count;
+	brelse(bh);
+	return retval;
+
+}
+
+
+static int free_dind_blocks(handle_t *handle,
+				struct inode *inode, __le32 i_data)
+{
+	int i;
+	__le32 *tmp_idata;
+	struct buffer_head *bh;
+	unsigned long max_entries = inode->i_sb->s_blocksize >> 2;
+
+	bh = sb_bread(inode->i_sb, le32_to_cpu(i_data));
+	if (!bh)
+		return -EIO;
+
+	tmp_idata = (__le32 *)bh->b_data;
+	for (i = 0; i < max_entries; i++) {
+		if (tmp_idata[i]) {
+			ext4_free_blocks(handle, inode,
+					le32_to_cpu(tmp_idata[i]), 1);
+		}
+	}
+	brelse(bh);
+	ext4_free_blocks(handle, inode, le32_to_cpu(i_data), 1);
+
+	return 0;
+
+
+}
+
+static int free_tind_blocks(handle_t *handle,
+				struct inode *inode, __le32 i_data)
+{
+	int i, retval = 0;
+	__le32 *tmp_idata;
+	struct buffer_head *bh;
+	unsigned long max_entries = inode->i_sb->s_blocksize >> 2;
+
+	bh = sb_bread(inode->i_sb, le32_to_cpu(i_data));
+	if (!bh)
+		return -EIO;
+
+	tmp_idata = (__le32 *)bh->b_data;
+
+	for (i = 0; i < max_entries; i++) {
+		if (tmp_idata[i]) {
+			retval = free_dind_blocks(handle,
+					inode, tmp_idata[i]);
+			if (retval) {
+				brelse(bh);
+				return retval;
+			}
+		}
+	}
+	brelse(bh);
+	ext4_free_blocks(handle, inode, le32_to_cpu(i_data), 1);
+
+	return 0;
+
+
+}
+
+static int free_ind_block(handle_t *handle, struct inode *inode)
+{
+	int retval;
+	struct ext4_inode_info *ei = EXT4_I(inode);
+
+	if (ei->i_data[EXT4_IND_BLOCK]) {
+
+		ext4_free_blocks(handle, inode,
+				le32_to_cpu(ei->i_data[EXT4_IND_BLOCK]), 1);
+
+	}
+
+	if (ei->i_data[EXT4_DIND_BLOCK]) {
+		retval = free_dind_blocks(handle, inode,
+						ei->i_data[EXT4_DIND_BLOCK]);
+		if (retval)
+			return retval;
+	}
+
+	if (ei->i_data[EXT4_TIND_BLOCK]) {
+		retval = free_tind_blocks(handle, inode,
+						ei->i_data[EXT4_TIND_BLOCK]);
+		if (retval)
+			return retval;
+	}
+
+
+	return 0;
+}
+static int ext4_ext_swap_inode_data(handle_t *handle, struct inode *inode,
+				struct inode *tmp_inode, int retval)
+{
+	struct ext4_inode_info *ei = EXT4_I(inode);
+	struct ext4_inode_info *tmp_ei = EXT4_I(tmp_inode);
+
+
+	retval = free_ind_block(handle, inode);
+	if (retval)
+		goto err_out;
+
+	/*
+	 * We have the extent map build with the tmp inode.
+	 * Now copy the i_data across
+	 */
+	ei->i_flags |= EXT4_EXTENTS_FL;
+	memcpy(ei->i_data, tmp_ei->i_data, sizeof(ei->i_data));
+
+	/*
+	 * Update i_blocks with the new blocks that got
+	 * allocated while adding extents for extent index
+	 * blocks.
+	 *
+	 * While converting to extents we need not
+	 * update the orignal inode i_blocks for extent blocks
+	 * via quota APIs. The quota update happened via tmp_inode already.
+	 */
+	spin_lock(&inode->i_lock);
+	inode->i_blocks += tmp_inode->i_blocks;
+	spin_unlock(&inode->i_lock);
+
+	ext4_mark_inode_dirty(handle, inode);
+
+err_out:
+
+	return retval;
+}
+
+/* Will go away */
+static ext4_fsblk_t idx_pblock(struct ext4_extent_idx *ix)
+{
+	ext4_fsblk_t block;
+
+	block = le32_to_cpu(ix->ei_leaf);
+	block |= ((ext4_fsblk_t) le16_to_cpu(ix->ei_leaf_hi) << 31) << 1;
+	return block;
+}
+
+static int free_ext_idx(handle_t *handle, struct inode *inode,
+					struct ext4_extent_idx *ix)
+{
+	int i, retval = 0;
+	ext4_fsblk_t block;
+	struct buffer_head *bh;
+	struct ext4_extent_header *eh;
+
+
+	block = idx_pblock(ix);
+	bh = sb_bread(inode->i_sb, block);
+	if (!bh)
+		return -EIO;
+
+	eh = (struct ext4_extent_header *)bh->b_data;
+	if (eh->eh_depth == 0) {
+
+		brelse(bh);
+		ext4_free_blocks(handle, inode, block, 1);
+
+	} else {
+
+		ix = EXT_FIRST_INDEX(eh);
+		for (i = 0; i < le16_to_cpu(eh->eh_entries); i++, ix++) {
+			retval = free_ext_idx(handle, inode, ix);
+			if (retval)
+				return retval;
+		}
+
+	}
+
+	return retval;
+
+}
+/*
+ * Free the extent meta data blocks only
+ */
+static int free_ext_block(handle_t *handle, struct inode *inode)
+{
+	int i, retval = 0;
+	struct ext4_inode_info *ei = EXT4_I(inode);
+	struct ext4_extent_header *eh = (struct ext4_extent_header *)ei->i_data;
+	struct ext4_extent_idx *ix;
+	if (eh->eh_depth == 0) {
+		/*
+		 * No extra blocks allocated for extent meta data
+		 */
+		return 0;
+	}
+	ix = EXT_FIRST_INDEX(eh);
+	for (i = 0; i < le16_to_cpu(eh->eh_entries); i++, ix++) {
+		retval = free_ext_idx(handle, inode, ix);
+		if (retval)
+			return retval;
+	}
+
+	return retval;
+
+}
+int ext4_ext_migrate(struct inode * inode, struct file * filp,
+				unsigned int cmd, unsigned long arg)
+{
+	handle_t *handle;
+	int retval = 0, i;
+	__le32 *i_data;
+	int blk_count = 0;
+	struct ext4_inode_info *ei;
+	struct inode *tmp_inode = NULL;
+	struct list_blocks_struct lb;
+	unsigned long max_entries;
+
+
+	if (!test_opt(inode->i_sb, EXTENTS)) {
+		/*
+		 * if mounted with noextents
+		 * we don't allow the migrate
+		 */
+		return -EINVAL;
+	}
+
+	if ((EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL))
+		return -EINVAL;
+
+	mutex_lock(&EXT4_I(inode)->truncate_mutex);
+
+
+	handle = ext4_journal_start(inode,
+					EXT4_DATA_TRANS_BLOCKS(inode->i_sb) +
+					EXT4_INDEX_EXTRA_TRANS_BLOCKS + 3 +
+					2 * EXT4_QUOTA_INIT_BLOCKS(inode->i_sb)
+					+ 1);
+	if (IS_ERR(handle)) {
+		retval = PTR_ERR(handle);
+		goto err_out;
+	}
+
+	tmp_inode = ext4_new_inode(handle,
+				inode->i_sb->s_root->d_inode,
+				S_IFREG);
+
+	if (IS_ERR(tmp_inode)) {
+		retval = -ENOMEM;
+		ext4_journal_stop(handle);
+		tmp_inode = NULL;
+		goto err_out;
+	}
+
+	i_size_write(tmp_inode, i_size_read(inode));
+	/*
+	 * We don't want the inode to be reclaimed
+	 * if we got interrupted in between. We have
+	 * this tmp inode carrying reference to the
+	 * data blocks of the original file. We set
+	 * the i_nlink to zero at the last stage after
+	 * switching the original file to extent format
+	 */
+	tmp_inode->i_nlink = 1;
+
+	ext4_ext_tree_init(handle, tmp_inode);
+	ext4_orphan_add(handle, tmp_inode);
+	ext4_journal_stop(handle);
+
+	ei = EXT4_I(inode);
+        i_data = ei->i_data;
+	memset(&lb, 0, sizeof(lb));
+
+	/* 32 bit block address 4 bytes */
+	max_entries = inode->i_sb->s_blocksize >> 2;
+
+	/*
+	 * start with one credit accounted for writing the
+	 * i_data field of the original inode and
+	 * one credit for superblock modification.
+	 *
+	 * For the tmp_inode we already have commited the
+	 * trascation that created the inode. Later as and
+	 * when we add extents we extent the journal
+	 */
+	handle = ext4_journal_start(inode, 2);
+	for (i = 0; i < EXT4_NDIR_BLOCKS; i++, blk_count++) {
+
+		if (i_data[i]) {
+			retval = update_extent_range(handle, tmp_inode,
+						le32_to_cpu(i_data[i]),
+						blk_count, &lb);
+			if (retval)
+				goto err_out;
+		}
+	}
+
+	if (i_data[EXT4_IND_BLOCK]) {
+		retval = update_ind_extent_range(handle, tmp_inode,
+					le32_to_cpu(i_data[EXT4_IND_BLOCK]),
+					&blk_count, &lb);
+			if (retval)
+				goto err_out;
+	} else {
+		blk_count +=  max_entries;
+	}
+
+	if (i_data[EXT4_DIND_BLOCK]) {
+		retval = update_dind_extent_range(handle, tmp_inode,
+					le32_to_cpu(i_data[EXT4_DIND_BLOCK]),
+					&blk_count, &lb);
+			if (retval)
+				goto err_out;
+	} else {
+		blk_count += max_entries * max_entries;
+	}
+
+
+	if (i_data[EXT4_TIND_BLOCK]) {
+		retval = update_tind_extent_range(handle, tmp_inode,
+					le32_to_cpu(i_data[EXT4_TIND_BLOCK]),
+					&blk_count, &lb);
+			if (retval)
+				goto err_out;
+	}
+
+	/*
+	 * Build the last extent
+	 */
+	retval = finish_range(handle, tmp_inode, &lb);
+
+err_out:
+	if (retval) {
+		/*
+		 * Failure case delete the extent information with the
+		 * tmp_inode
+		 */
+		free_ext_block(handle, tmp_inode);
+
+	} else {
+
+		retval = ext4_ext_swap_inode_data(handle, inode,
+							tmp_inode, retval);
+	}
+
+	/*
+	 * Mark the tmp_inode as of size zero
+	 */
+	i_size_write(tmp_inode, 0);
+
+
+	/*
+	 * set the  i_blocks count to zero
+	 * so that the ext4_delete_inode does the
+	 * right job
+	 *
+	 * FIXME!! do we need to take the lock ?
+	 */
+	spin_lock(&tmp_inode->i_lock);
+	tmp_inode->i_blocks = 0;
+	spin_unlock(&tmp_inode->i_lock);
+
+	/* Reset the extent details */
+	ext4_ext_tree_init(handle, tmp_inode);
+
+	/*
+	 * Set the i_nlink to zero so that
+	 * generic_drop_inode really deletes the
+	 * inode
+	 */
+	tmp_inode->i_nlink = 0;
+
+	ext4_journal_stop(handle);
+
+	mutex_unlock(&EXT4_I(inode)->truncate_mutex);
+
+	if (tmp_inode)
+		iput(tmp_inode);
+
+
+	return retval;
+}
diff --git a/include/linux/ext4_fs.h b/include/linux/ext4_fs.h
index ebd6f86..f66fb7c 100644
--- a/include/linux/ext4_fs.h
+++ b/include/linux/ext4_fs.h
@@ -242,6 +242,7 @@ struct ext4_new_group_data {
 #endif
 #define EXT4_IOC_GETRSVSZ		_IOR('f', 5, long)
 #define EXT4_IOC_SETRSVSZ		_IOW('f', 6, long)
+#define EXT4_IOC_MIGRATE		_IO('f', 7)
 
 /*
  * ioctl commands in 32 bit emulation
@@ -872,6 +873,9 @@ extern int ext4_ioctl (struct inode *, struct file *, unsigned int,
 		       unsigned long);
 extern long ext4_compat_ioctl (struct file *, unsigned int, unsigned long);
 
+/* migrate.c */
+extern int ext4_ext_migrate (struct inode *, struct file *, unsigned int,
+		       unsigned long);
 /* namei.c */
 extern int ext4_orphan_add(handle_t *, struct inode *);
 extern int ext4_orphan_del(handle_t *, struct inode *);
-- 
1.5.2.2.571.ge1341-dirty