2008-03-06 00:02:14

by Akira Fujita

Subject: [RFC][PATCH 2/3] ext4 online defrag (ver 0.7)

From: Akira Fujita <[email protected]>

Change the name of functions (ext4_ext_xxx -> ext4_defrag_xxx)
and some cleanups.

Signed-off-by: Akira Fujita <[email protected]>
Signed-off-by: Takashi Sato <[email protected]>
--
fs/ext4/defrag.c | 1060 +++++++++++++++------------------------
fs/ext4/extents.c | 5 +-
fs/ext4/ioctl.c | 5 +-
fs/ext4/mballoc.c | 3 +-
include/linux/ext4_fs.h | 13 +-
include/linux/ext4_fs_extents.h | 13 -
6 files changed, 424 insertions(+), 675 deletions(-)

diff --git a/fs/ext4/defrag.c b/fs/ext4/defrag.c
index d22bec9..c86a9e2 100644
--- a/fs/ext4/defrag.c
+++ b/fs/ext4/defrag.c
@@ -1,207 +1,45 @@
-#include <linux/module.h>
-#include <linux/fs.h>
-#include <linux/time.h>
-#include <linux/jbd2.h>
-#include <linux/highuid.h>
-#include <linux/pagemap.h>
+/*
+ * Copyright (c) 2008, NEC Software Tohoku, Ltd.
+ * Written by Takashi Sato <[email protected]>
+ * Akira Fujita <[email protected]>
+ *
+ * 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 will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ */
+
+/* Online defragmentation for EXT4 */
+
#include <linux/quotaops.h>
-#include <linux/string.h>
-#include <linux/slab.h>
-#include <linux/falloc.h>
-#include <asm/uaccess.h>
#include <linux/ext4_jbd2.h>
#include <linux/ext4_fs_extents.h>
#include "group.h"

-#define DIO_CREDITS (EXT4_RESERVE_TRANS_BLOCKS + 32)
#define EXT_SET_EXTENT_DATA(src, dest) do { \
dest.block = le32_to_cpu(src->ee_block); \
dest.start = ext_pblock(src); \
dest.len = le16_to_cpu(src->ee_len); \
} while (0)

-/*
- * this structure is used to gather extents from the tree via ioctl
- */
-struct ext4_extent_buf {
- ext4_fsblk_t start;
- int buflen;
- void *buffer;
- void *cur;
- int err;
-};
-
-/*
- * this structure is used to collect stats info about the tree
- */
-struct ext4_extent_tree_stats {
- int depth;
- int extents_num;
- int leaf_num;
-};
-
-int ext4_ext_walk_space(struct inode *inode, ext4_lblk_t block,
- ext4_lblk_t num, ext_prepare_callback func,
- void *cbdata)
-{
- struct ext4_ext_path *path = NULL;
- struct ext4_ext_cache cbex;
- struct ext4_extent *ex;
- ext4_lblk_t next, start = 0, end = 0;
- ext4_lblk_t last = block + num;
- int depth, exists, err = 0;
-
- BUG_ON(func == NULL);
- BUG_ON(inode == NULL);
-
- while (block < last && block != EXT_MAX_BLOCK) {
- num = last - block;
- /* find extent for this block */
- path = ext4_ext_find_extent(inode, block, path);
- if (IS_ERR(path)) {
- err = PTR_ERR(path);
- path = NULL;
- break;
- }
-
- depth = ext_depth(inode);
- BUG_ON(path[depth].p_hdr == NULL);
- ex = path[depth].p_ext;
- next = ext4_ext_next_allocated_block(path);
-
- exists = 0;
- if (!ex) {
- /* there is no extent yet, so try to allocate
- * all requested space */
- start = block;
- end = block + num;
- } else if (le32_to_cpu(ex->ee_block) > block) {
- /* need to allocate space before found extent */
- start = block;
- end = le32_to_cpu(ex->ee_block);
- if (block + num < end)
- end = block + num;
- } else if (block >= le32_to_cpu(ex->ee_block)
- + ext4_ext_get_actual_len(ex)) {
- /* need to allocate space after found extent */
- start = block;
- end = block + num;
- if (end >= next)
- end = next;
- } else if (block >= le32_to_cpu(ex->ee_block)) {
- /*
- * some part of requested space is covered
- * by found extent
- */
- start = block;
- end = le32_to_cpu(ex->ee_block)
- + ext4_ext_get_actual_len(ex);
- if (block + num < end)
- end = block + num;
- exists = 1;
- } else {
- BUG();
- }
- BUG_ON(end <= start);
-
- if (!exists) {
- cbex.ec_block = start;
- cbex.ec_len = end - start;
- cbex.ec_start = 0;
- cbex.ec_type = EXT4_EXT_CACHE_GAP;
- } else {
- cbex.ec_block = le32_to_cpu(ex->ee_block);
- cbex.ec_len = ext4_ext_get_actual_len(ex);
- cbex.ec_start = ext_pblock(ex);
- cbex.ec_type = EXT4_EXT_CACHE_EXTENT;
- }
-
- BUG_ON(cbex.ec_len == 0);
- err = func(inode, path, &cbex, cbdata);
- ext4_ext_drop_refs(path);
-
- if (err < 0)
- break;
- if (err == EXT_REPEAT)
- continue;
- else if (err == EXT_BREAK) {
- err = 0;
- break;
- }
-
- if (ext_depth(inode) != depth) {
- /* depth was changed. we have to realloc path */
- kfree(path);
- path = NULL;
- }
-
- block = cbex.ec_block + cbex.ec_len;
- }
-
- if (path) {
- ext4_ext_drop_refs(path);
- kfree(path);
- }
-
- return err;
-}
-
-static int
-ext4_ext_store_extent_cb(struct inode *inode,
- struct ext4_ext_path *path,
- struct ext4_ext_cache *newex,
- struct ext4_extent_buf *buf)
-{
-
- if (newex->ec_type != EXT4_EXT_CACHE_EXTENT)
- return EXT_CONTINUE;
-
- if (buf->err < 0)
- return EXT_BREAK;
- if (buf->cur - buf->buffer + sizeof(*newex) > buf->buflen)
- return EXT_BREAK;
-
- if (!copy_to_user(buf->cur, newex, sizeof(*newex))) {
- buf->err++;
- buf->cur += sizeof(*newex);
- } else {
- buf->err = -EFAULT;
- return EXT_BREAK;
- }
- return EXT_CONTINUE;
-}
-
-static int
-ext4_ext_collect_stats_cb(struct inode *inode,
- struct ext4_ext_path *path,
- struct ext4_ext_cache *ex,
- struct ext4_extent_tree_stats *buf)
-{
- int depth;
-
- if (ex->ec_type != EXT4_EXT_CACHE_EXTENT)
- return EXT_CONTINUE;
-
- depth = ext_depth(inode);
- buf->extents_num++;
- if (path[depth].p_ext == EXT_FIRST_EXTENT(path[depth].p_hdr))
- buf->leaf_num++;
- return EXT_CONTINUE;
-}
-
/**
- * ext4_ext_next_extent - search for next extent and set it to "extent"
+ * ext4_defrag_next_extent - Search for the next extent and set it to "extent"
+ *
* @inode: inode of the the original file
- * @path: this will obtain data for next extent
- * @extent: pointer to next extent we have just gotten
+ * @path: this will obtain data for the next extent
+ * @extent: pointer to the next extent we have just gotten
*
- * This function returns 0 or 1(last_entry) if succeeded, otherwise
- * returns -EIO
+ * This function returns 0 or 1(last entry) if succeeded, otherwise
+ * returns -EIO.
*/
static int
-ext4_ext_next_extent(struct inode *inode,
- struct ext4_ext_path *path,
- struct ext4_extent **extent)
+ext4_defrag_next_extent(struct inode *inode,
+ struct ext4_ext_path *path,
+ struct ext4_extent **extent)
{
int ppos;
int leaf_ppos = path->p_depth;
@@ -231,7 +69,7 @@ ext4_ext_next_extent(struct inode *inode,
path[ppos+1].p_hdr =
ext_block_hdr(path[ppos+1].p_bh);

- /* halfway index block */
+ /* Halfway index block */
while (++cur_ppos < leaf_ppos) {
path[cur_ppos].p_idx =
EXT_FIRST_INDEX(path[cur_ppos].p_hdr);
@@ -253,39 +91,39 @@ ext4_ext_next_extent(struct inode *inode,
return 0;
}
}
- /* last_extent */
+ /* We found the last extent */
return 1;
}

/**
- * ext4_ext_extents_info() - get extents information
- *
- * @ext_info: pointer to ext4_extents_info
- * @ext_info->ino describe an inode which is used to get extent
- * information
- * @ext_info->max_entries: defined by DEFRAG_MAX_ENT
- * @ext_info->entries: amount of extents (output)
- * @ext_info->ext[]: array of extent (output)
- * @ext_info->offset: starting block offset of targeted extent
- * (file relative)
+ * ext4_defrag_extents_info - Get extents information
*
- * @sb: for iget()
+ * @sb: for ext4_iget()
+ * @ext_info: pointer to ext4_extents_info
+ * @ext_info->ino describe an inode which is used to get
+ * extent information
+ * @ext_info->max_entries: defined by DEFRAG_MAX_ENT
+ * @ext_info->entries: amount of extents (output)
+ * @ext_info->ext[]: array of extent (output)
+ * @ext_info->offset: starting block offset of targeted extent
+ * (file relative)
*
- * This function returns 0 if next extent(s) exists,
- * or returns 1 if next extent doesn't exist, otherwise returns error value.
+ * This function returns 0 if the next extent(s) exists,
+ * or returns 1 if the next extent doesn't exist,
+ * otherwise returns error value.
*/
-static int ext4_ext_extents_info(struct ext4_extents_info *ext_info,
- struct super_block *sb)
+static int ext4_defrag_extents_info(struct super_block *sb,
+ struct ext4_extents_info *ext_info)
{
struct ext4_ext_path *path = NULL;
struct ext4_extent *ext = NULL;
struct inode *inode = NULL;
ext4_lblk_t offset = ext_info->f_offset;
int max_entries = ext_info->max_entries;
- int is_last_extent = 0;
int depth = 0;
int entries = 0;
int err = 0;
+ int ret = 0;

inode = ext4_iget(sb, ext_info->ino);
if (IS_ERR(inode))
@@ -293,7 +131,7 @@ static int ext4_ext_extents_info(struct ext4_extents_info *ext_info,

down_write(&EXT4_I(inode)->i_data_sem);

- /* if a file doesn't exist*/
+ /* Return -ENOENT if a file does not exist */
if ((!inode->i_nlink) || (inode->i_ino < 11) ||
!S_ISREG(inode->i_mode)) {
ext_info->entries = 0;
@@ -309,7 +147,7 @@ static int ext4_ext_extents_info(struct ext4_extents_info *ext_info,
}
depth = ext_depth(inode);

- /* if file size is 0, skip this one. */
+ /* Skip the 0 size file */
if (path[depth].p_ext == NULL) {
ext_info->entries = 0;
goto out;
@@ -324,9 +162,9 @@ static int ext4_ext_extents_info(struct ext4_extents_info *ext_info,
* more the number of extents than 'max_entries'.
*/
while (entries < max_entries) {
- is_last_extent = ext4_ext_next_extent(inode, path, &ext);
- /* found next extent (not the last one)*/
- if (is_last_extent == 0) {
+ ret = ext4_defrag_next_extent(inode, path, &ext);
+ if (ret == 0) {
+ /* Found the next extent (it means not the last one) */
EXT_SET_EXTENT_DATA(ext, ext_info->ext[entries]);
entries++;

@@ -341,27 +179,27 @@ static int ext4_ext_extents_info(struct ext4_extents_info *ext_info,
ext_info->f_offset =
le32_to_cpu(ext->ee_block) +
le32_to_cpu(ext->ee_len);
- /* check the extent is the last one or not*/
- is_last_extent =
- ext4_ext_next_extent(inode, path, &ext);
- if (is_last_extent == 1) {
- err = is_last_extent;
- } else if (is_last_extent < 0) {
- /*ERR*/
- err = is_last_extent;
+ /* Check the extent is the last one or not */
+ ret =
+ ext4_defrag_next_extent(inode, path, &ext);
+ if (ret == 1) {
+ err = ret;
+ } else if (ret < 0) {
+ /* Failed to get the next extent */
+ err = ret;
goto out;
}
break;
}

- /* the extent is the last one */
- } else if (is_last_extent == 1) {
+ } else if (ret == 1) {
+ /* The extent is the last one */
ext_info->f_offset = 0;
- err = is_last_extent;
+ err = ret;
break;
} else {
- /* ERR */
- err = is_last_extent;
+ /* Failed to get the next extent */
+ err = ret;
goto out;
}
}
@@ -379,20 +217,21 @@ out:
}

/**
- * ext4_ext_defrag_reserve - reserve blocks for defrag
+ * ext4_defrag_reserve_blocks - Reserve blocks for defrag
+ *
* @inode target inode
* @goal block reservation goal
* @len blocks count to reserve
*
* This function returns 0 if succeeded, otherwise
- * returns error value
+ * returns error value.
*/

static int
-ext4_ext_defrag_reserve(struct inode *inode, ext4_fsblk_t goal, int len)
+ext4_defrag_reserve_blocks(struct inode *inode, ext4_fsblk_t goal, int len)
{
struct super_block *sb = NULL;
- handle_t *handle = NULL;
+ handle_t *handle;
struct buffer_head *bitmap_bh = NULL;
struct ext4_block_alloc_info *block_i;
struct ext4_reserve_window_node *my_rsv = NULL;
@@ -413,16 +252,14 @@ ext4_ext_defrag_reserve(struct inode *inode, ext4_fsblk_t goal, int len)
if (S_ISREG(inode->i_mode) && (!EXT4_I(inode)->i_block_alloc_info)) {
ext4_init_block_alloc_info(inode);
} else if (!S_ISREG(inode->i_mode)) {
- printk(KERN_ERR "ext4_ext_defrag_reserve:"
- " incorrect file type\n");
- err = -1;
+ printk(KERN_ERR "ext4 defrag: Invalid file type\n");
+ err = -EINVAL;
goto out;
}

sb = inode->i_sb;
if (!sb) {
- printk(KERN_ERR "ext4_ext_defrag_reserve: "
- "nonexistent device\n");
+ printk(KERN_ERR "ext4 defrag: Non-existent device\n");
err = -ENXIO;
goto out;
}
@@ -430,13 +267,13 @@ ext4_ext_defrag_reserve(struct inode *inode, ext4_fsblk_t goal, int len)
&grp_target_blk);

block_i = EXT4_I(inode)->i_block_alloc_info;
+ /* Block reservation should be enabled */
+ BUG_ON(!block_i);
+
+ windowsz = block_i->rsv_window_node.rsv_goal_size;
+ /* Goal size should be set */
+ BUG_ON(!windowsz);

- if (!block_i || ((windowsz =
- block_i->rsv_window_node.rsv_goal_size) == 0)) {
- printk(KERN_ERR "ex4_ext_defrag_reserve: unable to reserve\n");
- err = -1;
- goto out;
- }

my_rsv = &block_i->rsv_window_node;

@@ -454,7 +291,9 @@ ext4_ext_defrag_reserve(struct inode *inode, ext4_fsblk_t goal, int len)
err = alloc_new_reservation(my_rsv, grp_target_blk, sb,
group_no, bitmap_bh);
if (err < 0) {
- printk(KERN_ERR "defrag: reservation faild\n");
+ printk(KERN_ERR "ext4 defrag: Block reservation failed."
+ "offset [%d], bg[%lu]\n",
+ grp_target_blk, group_no);
ext4_discard_reservation(inode);
goto out;
} else {
@@ -476,16 +315,17 @@ out:
}

/**
- * ext4_ext_block_within_rsv - Is target extent reserved ?
- * @ inode inode of target file
- * @ ex_start start physical block number of the extent
- * which already moved
- * @ ex_len block length of the extent which already moved
+ * ext4_defrag_block_within_rsv - Is target extent reserved ?
+ *
+ * @ inode inode of target file
+ * @ ex_start start physical block number of the extent
+ * which already moved
+ * @ ex_len block length of the extent which already moved
*
* This function returns 0 if succeeded, otherwise
- * returns error value
+ * returns error value.
*/
-static int ext4_ext_block_within_rsv(struct inode *inode,
+static int ext4_defrag_block_within_rsv(struct inode *inode,
ext4_fsblk_t ex_start, int ex_len)
{
struct super_block *sb = inode->i_sb;
@@ -495,15 +335,15 @@ static int ext4_ext_block_within_rsv(struct inode *inode,
struct ext4_reserve_window_node *rsv;

block_i = EXT4_I(inode)->i_block_alloc_info;
- if (block_i && block_i->rsv_window_node.rsv_goal_size > 0) {
- rsv = &block_i->rsv_window_node;
- if (rsv_is_empty(&rsv->rsv_window)) {
- printk(KERN_ERR "defrag: Can't defrag due to"
- " the empty reservation\n");
- return -ENOSPC;
- }
- } else {
- printk(KERN_ERR "defrag: No i_block_alloc_info\n");
+ /* Block reservation should be enabled */
+ BUG_ON(!block_i);
+
+ /* Goal size should be set */
+ BUG_ON(!block_i->rsv_window_node.rsv_goal_size);
+
+ rsv = &block_i->rsv_window_node;
+ if (rsv_is_empty(&rsv->rsv_window)) {
+ printk(KERN_ERR "ext4 defrag: Reservation window is empty\n");
return -ENOSPC;
}

@@ -512,7 +352,7 @@ static int ext4_ext_block_within_rsv(struct inode *inode,
if (!goal_in_my_reservation(&rsv->rsv_window, grp_blk, group_no, sb)
|| !goal_in_my_reservation(&rsv->rsv_window, grp_blk + ex_len - 1,
group_no, sb)){
- printk(KERN_ERR "defrag: %d or %d in bg %lu is "
+ printk(KERN_ERR "ext4 defrag: %d or %d in bg %lu is "
"not in rsv_window\n", grp_blk,
grp_blk + ex_len - 1, group_no);
return -ENOSPC;
@@ -521,13 +361,14 @@ static int ext4_ext_block_within_rsv(struct inode *inode,
}

/*
- * ext4_ext_fblocks_reserve() -
- * reserve free blocks by ext4_ext_defrag_reserve()
+ * ext4_defrag_reserve_fblocks - Reserve free blocks
+ * with ext4_defrag_reserve_blocks
+ *
* @inode: To get a block group number
* @ext_info: freeblocks distribution which stored extent-like style
- * @ext_info->ext[] an array of struct ext4_extents_data
+ * @ext_info->ext[] an array of struct ext4_extents_data
*/
-static int ext4_ext_fblocks_reserve(struct inode *inode,
+static int ext4_defrag_reserve_fblocks(struct inode *inode,
struct ext4_extents_info *ext_info)
{
ext4_fsblk_t ex_start = 0;
@@ -539,22 +380,24 @@ static int ext4_ext_fblocks_reserve(struct inode *inode,
ex_start = ext_info->ext[i].start;
len = ext_info->ext[i].len;

- ret = ext4_ext_defrag_reserve(inode, ex_start, len);
+ ret = ext4_defrag_reserve_blocks(inode, ex_start, len);
if (ret < 0) {
- printk(KERN_ERR "defrag: failed "
- "ext4_ext_defrag_reserve\n");
- goto ERR;
+ printk(KERN_ERR "ext4 defrag: "
+ "Block reservation failed. offset [%llu], "
+ "length [%d]\n", ex_start, len);
+ goto err;
}
- ret = ext4_ext_block_within_rsv(inode, ex_start, len);
+ ret = ext4_defrag_block_within_rsv(inode, ex_start, len);
if (ret < 0) {
- printk(KERN_ERR "defrag: failed "
- "ext4_ext_block_within_rsv\n");
- goto ERR;
+ printk(KERN_ERR "ext4 defrag: "
+ "Reservation window is not set. "
+ "offset [%llu], length [%d]\n", ex_start, len);
+ goto err;
}
}
return ret;

-ERR:
+err:
down_write(&EXT4_I(inode)->i_data_sem);
ext4_discard_reservation(inode);
up_write(&EXT4_I(inode)->i_data_sem);
@@ -562,33 +405,34 @@ ERR:
}

/**
- * ext4_ext_defrag_victim - Create free space for defrag
- * @filp target file
- * @ex_info target extents array to move
+ * ext4_defrag_move_victim - Create free space for defrag
+ *
+ * @filp target file
+ * @ext_info target extents array to move
*
* This function returns 0 if succeeded, otherwise
- * returns error value
+ * returns error value.
*/
-static int ext4_ext_defrag_victim(struct file *target_filp,
- struct ext4_extents_info *ex_info)
+static int ext4_defrag_move_victim(struct file *target_filp,
+ struct ext4_extents_info *ext_info)
{
struct inode *target_inode = target_filp->f_dentry->d_inode;
struct super_block *sb = target_inode->i_sb;
struct file victim_file;
struct dentry victim_dent;
struct inode *victim_inode;
- ext4_fsblk_t goal = ex_info->goal;
+ ext4_fsblk_t goal = ext_info->goal;
int ret = 0;
int i = 0;
struct ext4_extent_data ext;
ext4_group_t group;
ext4_grpblk_t grp_off;

- /* Setup dummy entent data */
+ /* Setup dummy extent data */
ext.len = 0;

/* Get the inode of the victim file */
- victim_inode = ext4_iget(sb, ex_info->ino);
+ victim_inode = ext4_iget(sb, ext_info->ino);
if (IS_ERR(victim_inode))
return PTR_ERR(victim_inode);

@@ -600,30 +444,33 @@ static int ext4_ext_defrag_victim(struct file *target_filp,
/* Set the goal appropriate offset */
if (goal == -1) {
ext4_get_group_no_and_offset(victim_inode->i_sb,
- ex_info->ext[0].start, &group, &grp_off);
+ ext_info->ext[0].start, &group, &grp_off);
goal = ext4_group_first_block_no(sb, group + 1);
}

- for (i = 0; i < ex_info->entries; i++) {
+ for (i = 0; i < ext_info->entries; i++) {
/* Move original blocks to another block group */
- ret = ext4_ext_defrag(&victim_file, ex_info->ext[i].block,
- ex_info->ext[i].len, goal, DEFRAG_FORCE_VICTIM, &ext);
+ ret = ext4_defrag(&victim_file, ext_info->ext[i].block,
+ ext_info->ext[i].len, goal, DEFRAG_FORCE_VICTIM, &ext);
if (ret < 0) {
- printk(KERN_ERR "defrag: failed ext4_ext_defrag\n");
- goto ERR;
+ printk(KERN_ERR "ext4 defrag: "
+ "Moving victim file failed. ino [%lu]\n",
+ ext_info->ino);
+ goto err;
}

/* Sync journal blocks before reservation */
ret = ext4_force_commit(sb);
if (ret) {
- printk(KERN_ERR "defrag: failed ext4_force_commit (%d)\n", ret);
- goto ERR;
+ printk(KERN_ERR "ext4 defrag: "
+ "ext4_force_commit failed(%d)\n", ret);
+ goto err;
}
}

iput(victim_inode);
return 0;
-ERR:
+err:
down_write(&EXT4_I(target_inode)->i_data_sem);
ext4_discard_reservation(target_inode);
up_write(&EXT4_I(target_inode)->i_data_sem);
@@ -632,19 +479,19 @@ ERR:
}

/**
- * ext4_ext_fblocks_distribution - Search free block distribution
- * @filp target file
- * @ex_info ext4_extents_info
+ * ext4_defrag_fblocks_distribution - Search free blocks distribution
+ *
+ * @inode target file
+ * @ext_info ext4_extents_info
*
* This function returns 0 if succeeded, otherwise
- * returns error value
+ * returns error value.
*/
-static int ext4_ext_fblocks_distribution(struct inode *inode,
+static int ext4_defrag_fblocks_distribution(struct inode *inode,
struct ext4_extents_info *ext_info)
{
struct buffer_head *bitmap_bh = NULL;
struct super_block *sb = inode->i_sb;
- struct ext4_super_block *es;
handle_t *handle;
ext4_group_t group_no;
ext4_grpblk_t start, end;
@@ -654,18 +501,21 @@ static int ext4_ext_fblocks_distribution(struct inode *inode,
int i = 0;
int err = 0;
int block_set = 0;
+ int extra_block = 0;

if (!sb) {
- printk(KERN_ERR "ext4_ext_fblock_distribution: "
- "nonexitent device\n");
+ printk(KERN_ERR "ext4 defrag: Non-existent device\n");
return -ENOSPC;
}
- es = EXT4_SB(sb)->s_es;

group_no = (inode->i_ino - 1) / EXT4_INODES_PER_GROUP(sb);
start = ext_info->g_offset;
end = EXT4_BLOCKS_PER_GROUP(sb) - 1;

+ /* We consider about the boot block if bs = 1k */
+ if (sb->s_blocksize == 1024)
+ extra_block = 1;
+
handle = ext4_journal_start(inode, 1);
if (IS_ERR(handle)) {
err = PTR_ERR(handle);
@@ -686,10 +536,14 @@ static int ext4_ext_fblocks_distribution(struct inode *inode,
for (i = start; i <= end ; i++) {
if (bitmap_search_next_usable_block(i, bitmap_bh, i + 1) >= 0) {
len++;
- /* if the free block is the first one in a region */
+ /*
+ * Reset start_block if the free block is
+ * the head of region.
+ */
if (!block_set) {
start_block =
- i + group_no * EXT4_BLOCKS_PER_GROUP(sb);
+ i + group_no * EXT4_BLOCKS_PER_GROUP(sb) +
+ extra_block;
block_set = 1;
}
} else if (len) {
@@ -721,7 +575,7 @@ out:
return err;
}

-int ext4_ext_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,
+int ext4_defrag_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,
unsigned long arg)
{
int err = 0;
@@ -729,37 +583,7 @@ int ext4_ext_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,
cmd == EXT4_IOC_FIBMAP))
return -EINVAL;

- if (cmd == EXT4_IOC_GET_EXTENTS) {
- struct ext4_extent_buf buf;
-
- if (copy_from_user(&buf, (void *) arg, sizeof(buf)))
- return -EFAULT;
-
- buf.cur = buf.buffer;
- buf.err = 0;
- down_write(&EXT4_I(inode)->i_data_sem);
- err = ext4_ext_walk_space(inode, buf.start, EXT_MAX_BLOCK,
- (void *)ext4_ext_store_extent_cb, &buf);
- up_write(&EXT4_I(inode)->i_data_sem);
- if (err == 0)
- err = buf.err;
- } else if (cmd == EXT4_IOC_GET_TREE_STATS) {
- struct ext4_extent_tree_stats buf;
-
- down_write(&EXT4_I(inode)->i_data_sem);
- buf.depth = ext_depth(inode);
- buf.extents_num = 0;
- buf.leaf_num = 0;
- err = ext4_ext_walk_space(inode, 0, EXT_MAX_BLOCK,
- (void *)ext4_ext_collect_stats_cb, &buf);
- up_write(&EXT4_I(inode)->i_data_sem);
- if (!err)
- err = copy_to_user((void *) arg, &buf, sizeof(buf));
- } else if (cmd == EXT4_IOC_GET_TREE_DEPTH) {
- down_write(&EXT4_I(inode)->i_data_sem);
- err = ext_depth(inode);
- up_write(&EXT4_I(inode)->i_data_sem);
- } else if (cmd == EXT4_IOC_FIBMAP) {
+ if (cmd == EXT4_IOC_FIBMAP) {
ext4_fsblk_t __user *p = (ext4_fsblk_t __user *)arg;
ext4_fsblk_t block = 0;
struct address_space *mapping = filp->f_mapping;
@@ -799,7 +623,7 @@ int ext4_ext_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,

BUG_ON(ext_info.ino != inode->i_ino);

- err = ext4_ext_fblocks_distribution(inode, &ext_info);
+ err = ext4_defrag_fblocks_distribution(inode, &ext_info);

if (!err)
err = copy_to_user((struct ext4_extents_info *)arg,
@@ -812,7 +636,7 @@ int ext4_ext_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,
sizeof(ext_info)))
return -EFAULT;

- err = ext4_ext_extents_info(&ext_info, inode->i_sb);
+ err = ext4_defrag_extents_info(inode->i_sb, &ext_info);
if (err >= 0) {
if (copy_to_user((struct ext4_extents_info __user *)arg,
&ext_info, sizeof(ext_info)))
@@ -826,7 +650,7 @@ int ext4_ext_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,
sizeof(ext_info)))
return -EFAULT;

- err = ext4_ext_fblocks_reserve(inode, &ext_info);
+ err = ext4_defrag_reserve_fblocks(inode, &ext_info);
} else if (cmd == EXT4_IOC_MOVE_VICTIM) {
struct ext4_extents_info ext_info;

@@ -835,7 +659,7 @@ int ext4_ext_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,
sizeof(ext_info)))
return -EFAULT;

- err = ext4_ext_defrag_victim(filp, &ext_info);
+ err = ext4_defrag_move_victim(filp, &ext_info);

} else if (cmd == EXT4_IOC_BLOCK_RELEASE) {
down_write(&EXT4_I(inode)->i_data_sem);
@@ -848,7 +672,7 @@ int ext4_ext_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,
(struct ext4_ext_defrag_data __user *)arg,
sizeof(defrag)))
return -EFAULT;
- err = ext4_ext_defrag(filp, defrag.start_offset,
+ err = ext4_defrag(filp, defrag.start_offset,
defrag.defrag_size, defrag.goal, defrag.flag,
&defrag.ext);
}
@@ -857,7 +681,7 @@ int ext4_ext_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,
}

/**
- * ext4_ext_merge_across - merge extents across leaf block
+ * ext4_defrag_merge_across_blocks - Merge extents across leaf block
*
* @handle journal handle
* @inode target file's inode
@@ -871,7 +695,7 @@ int ext4_ext_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,
* This function returns 0 if succeed, otherwise returns error value.
*/
static int
-ext4_ext_merge_across_blocks(handle_t *handle, struct inode *inode,
+ext4_defrag_merge_across_blocks(handle_t *handle, struct inode *inode,
struct ext4_extent *o_start,
struct ext4_extent *o_end, struct ext4_extent *start_ext,
struct ext4_extent *new_ext, struct ext4_extent *end_ext,
@@ -943,7 +767,10 @@ ext4_ext_merge_across_blocks(handle_t *handle, struct inode *inode,
o_end->ee_len = end_ext->ee_len;
ext4_ext_store_pblock(o_end, ext_pblock(end_ext));

- /* If new_ext was first block */
+ /*
+ * Set 0 to the extent block if new_ext was
+ * the first block.
+ */
if (!new_ext->ee_block)
eblock = 0;
else
@@ -951,7 +778,7 @@ ext4_ext_merge_across_blocks(handle_t *handle, struct inode *inode,

new_flag = 1;
} else {
- printk(KERN_ERR "Unexpected case \n");
+ printk(KERN_ERR "ext4 defrag: Unexpected merge case\n");
return -EIO;
}

@@ -960,12 +787,12 @@ ext4_ext_merge_across_blocks(handle_t *handle, struct inode *inode,
if (IS_ERR(org_path)) {
err = PTR_ERR(org_path);
org_path = NULL;
- goto ERR;
+ goto out;
}
err = ext4_ext_insert_extent_defrag(handle, inode,
org_path, new_ext, defrag_flag);
if (err)
- goto ERR;
+ goto out;
}

if (end_flag) {
@@ -974,14 +801,14 @@ ext4_ext_merge_across_blocks(handle_t *handle, struct inode *inode,
if (IS_ERR(org_path)) {
err = PTR_ERR(org_path);
org_path = NULL;
- goto ERR;
+ goto out;
}
err = ext4_ext_insert_extent_defrag(handle, inode,
org_path, end_ext, defrag_flag);
if (err)
- goto ERR;
+ goto out;
}
-ERR:
+out:
if (org_path) {
ext4_ext_drop_refs(org_path);
kfree(org_path);
@@ -992,23 +819,23 @@ ERR:
}

/**
- * ext4_ext_merge_inside_block - merge new extent to the extent block
+ * ext4_defrag_merge_inside_block - Merge new extent to the extent block
*
- * @handle journal handle
- * @inode target file's inode
- * @o_start first original extent to be defraged
- * @o_end last original extent to be merged
- * @start_ext first new extent to be merged
- * @new_ext middle of new extent to be merged
- * @end_ext last new extent to be merged
- * @eh extent header of target leaf block
- * @replaced the number of blocks which will be replaced with new_ext
- * @range_to_move used to dicide how to merge
+ * @handle journal handle
+ * @inode target file's inode
+ * @o_start first original extent to be defraged
+ * @o_end last original extent to be merged
+ * @start_ext first new extent to be merged
+ * @new_ext middle of new extent to be merged
+ * @end_ext last new extent to be merged
+ * @eh extent header of target leaf block
+ * @replaced the number of blocks which will be replaced with new_ext
+ * @range_to_move used to decide how to merge
*
* This function always returns 0.
-*/
+ */
static int
-ext4_ext_merge_inside_block(handle_t *handle, struct inode *inode,
+ext4_defrag_merge_inside_block(handle_t *handle, struct inode *inode,
struct ext4_extent *o_start, struct ext4_extent *o_end,
struct ext4_extent *start_ext, struct ext4_extent *new_ext,
struct ext4_extent *end_ext, struct ext4_extent_header *eh,
@@ -1047,23 +874,23 @@ ext4_ext_merge_inside_block(handle_t *handle, struct inode *inode,
}

/**
- * ext4_ext_merge_extents - merge new extent
+ * ext4_defrag_merge_extents - Merge new extent
*
- * @handle journal handle
- * @inode target file's inode
- * @org_path path indicates first extent to be defraged
- * @o_start first original extent to be defraged
- * @o_end last original extent to be defraged
- * @start_ext first new extent to be merged
- * @new_ext middle of new extent to be merged
- * @end_ext last new extent to be merged
- * @replaced the number of blocks which will be replaced with new_ext
- * @flag defrag mode (e.g. -f)
+ * @handle journal handle
+ * @inode target file's inode
+ * @org_path path indicates first extent to be defraged
+ * @o_start first original extent to be defraged
+ * @o_end last original extent to be defraged
+ * @start_ext first new extent to be merged
+ * @new_ext middle of new extent to be merged
+ * @end_ext last new extent to be merged
+ * @replaced the number of blocks which will be replaced with new_ext
+ * @flag defrag mode (e.g. -f)
*
* This function returns 0 if succeed, otherwise returns error value.
*/
static int
-ext4_ext_merge_extents(handle_t *handle, struct inode *inode,
+ext4_defrag_merge_extents(handle_t *handle, struct inode *inode,
struct ext4_ext_path *org_path,
struct ext4_extent *o_start, struct ext4_extent *o_end,
struct ext4_extent *start_ext, struct ext4_extent *new_ext,
@@ -1073,8 +900,9 @@ ext4_ext_merge_extents(handle_t *handle, struct inode *inode,
unsigned need_slots, slots_range;
int range_to_move, depth, ret;

- /* The extents need to be inserted
- * start_extent + new_extent + end_extent
+ /*
+ * The extents need to be inserted
+ * start_extent + new_extent + end_extent.
*/
need_slots = (le16_to_cpu(start_ext->ee_len) ? 1 : 0) +
(le16_to_cpu(end_ext->ee_len) ? 1 : 0) +
@@ -1096,18 +924,18 @@ ext4_ext_merge_extents(handle_t *handle, struct inode *inode,
return ret;
}

- /* expansion */
+ /* Expansion */
if ((range_to_move > 0) &&
(range_to_move > le16_to_cpu(eh->eh_max)
- le16_to_cpu(eh->eh_entries))) {

- ret = ext4_ext_merge_across_blocks(handle, inode, o_start,
+ ret = ext4_defrag_merge_across_blocks(handle, inode, o_start,
o_end, start_ext, new_ext,
end_ext, flag);
if (ret < 0)
return ret;
} else {
- ret = ext4_ext_merge_inside_block(handle, inode, o_start,
+ ret = ext4_defrag_merge_inside_block(handle, inode, o_start,
o_end, start_ext, new_ext, end_ext,
eh, replaced, range_to_move);
if (ret < 0)
@@ -1129,18 +957,19 @@ ext4_ext_merge_extents(handle_t *handle, struct inode *inode,
}

/**
- * ext4_ext_defrag_leaf_block - Defragmentation for one leaf extent block.
- * @handle journal handle
- * @org_inode target inode
- * @org_path path indicates first extent to be defraged
- * @dext destination extent
- * @from start offset on the target file
- * @flag defrag mode (e.g. -f)
+ * ext4_defrag_leaf_block - Defragmentation for one leaf extent block
+ *
+ * @handle journal handle
+ * @org_inode target inode
+ * @org_path path indicates first extent to be defraged
+ * @dext destination extent
+ * @from start offset on the target file
+ * @flag defrag mode (e.g. -f)
*
* This function returns 0 if succeed, otherwise returns error value.
*/
static int
-ext4_ext_defrag_leaf_block(handle_t *handle, struct inode *org_inode,
+ext4_defrag_leaf_block(handle_t *handle, struct inode *org_inode,
struct ext4_ext_path *org_path, struct ext4_extent *dext,
ext4_lblk_t *from, int flag)
{
@@ -1166,7 +995,8 @@ ext4_ext_defrag_leaf_block(handle_t *handle, struct inode *org_inode,
new_phys_end = ext_pblock(&new_ext)
+ le16_to_cpu(new_ext.ee_len) - 1;

- /* First original extent
+ /*
+ * First original extent
* dest |---------------|
* org |---------------|
*/
@@ -1194,12 +1024,14 @@ ext4_ext_defrag_leaf_block(handle_t *handle, struct inode *org_inode,
new_ext.ee_len = 0;
}
}
- for (;;) {
+
+ for (;;) {
/* The extent for destination must be found. */
BUG_ON(!oext || lblock != le32_to_cpu(oext->ee_block));
lblock += le16_to_cpu(oext->ee_len);

- /* Middle of original extent
+ /*
+ * Middle of original extent
* dest |-------------------|
* org |-----------------|
*/
@@ -1209,7 +1041,8 @@ ext4_ext_defrag_leaf_block(handle_t *handle, struct inode *org_inode,
+ le16_to_cpu(oext->ee_len) - 1)
replaced += le16_to_cpu(oext->ee_len);

- /* Last original extent
+ /*
+ * Last original extent
* dest |----------------|
* org |---------------|
*/
@@ -1230,13 +1063,14 @@ ext4_ext_defrag_leaf_block(handle_t *handle, struct inode *org_inode,
- le16_to_cpu(end_ext.ee_len);
}

- /* Detected the block end, reached the number of replaced
- * blocks to dext->ee_len. Then, merge the extent.
+ /*
+ * Detected the block end, reached the number of replaced
+ * blocks to dext->ee_len. Then merge the extent.
*/
if (oext == EXT_LAST_EXTENT(org_path[depth].p_hdr) ||
new_end <= le32_to_cpu(oext->ee_block)
+ le16_to_cpu(oext->ee_len) - 1) {
- ret = ext4_ext_merge_extents(handle, org_inode,
+ ret = ext4_defrag_merge_extents(handle, org_inode,
org_path, o_start, o_end, &start_ext,
&new_ext, &end_ext, replaced, flag);
if (ret < 0)
@@ -1249,19 +1083,19 @@ ext4_ext_defrag_leaf_block(handle_t *handle, struct inode *org_inode,
return 0;
}

- /* re-calculate new_ext */
+ /* Re-calculate new_ext */
new_ext.ee_len = cpu_to_le32(le16_to_cpu(new_ext.ee_len)
- replaced);
new_ext.ee_block =
cpu_to_le32(le32_to_cpu(new_ext.ee_block)
+ replaced);
ext4_ext_store_pblock(&new_ext, ext_pblock(&new_ext)
- + replaced);
+ + replaced);
replaced = 0;
start_ext.ee_len = end_ext.ee_len = 0;
o_start = NULL;

- /* All expected blocks are replaced */
+ /* All expected blocks are replaced. */
if (le16_to_cpu(new_ext.ee_len) <= 0) {
if (DQUOT_ALLOC_BLOCK(org_inode, len))
return -EDQUOT;
@@ -1269,7 +1103,7 @@ ext4_ext_defrag_leaf_block(handle_t *handle, struct inode *org_inode,
}
}

- /* Get next extent for original. */
+ /* Get the next extent for original. */
if (org_path)
ext4_ext_drop_refs(org_path);
org_path = ext4_ext_find_extent(org_inode, lblock, org_path);
@@ -1290,22 +1124,24 @@ ext4_ext_defrag_leaf_block(handle_t *handle, struct inode *org_inode,
}

/**
- * ext4_ext_replace_branches - replace original extents with new extents.
- * @org_inode Original inode
- * @dest_inode temporary inode
- * @from_page Page offset
- * @count_page Page count to be replaced
- * @flag defrag mode (e.g. -f)
+ * ext4_defrag_replace_branches - Replace original extents with new extents
+ *
+ * @handle journal handle
+ * @org_inode original inode
+ * @dest_inode temporary inode
+ * @from_page page offset of org_inode
+ * @dest_from_page page offset of dest_inode
+ * @count_page page count to be replaced
+ * @flag defrag mode (e.g. -f)
*
* This function returns 0 if succeed, otherwise returns error value.
* Replace extents for blocks from "from" to "from + count - 1".
*/
static int
-ext4_ext_replace_branches(struct inode *org_inode, struct inode *dest_inode,
- pgoff_t from_page, pgoff_t dest_from_page,
- pgoff_t count_page, int flag)
+ext4_defrag_replace_branches(handle_t *handle, struct inode *org_inode,
+ struct inode *dest_inode, pgoff_t from_page,
+ pgoff_t dest_from_page, pgoff_t count_page, int flag)
{
- handle_t *handle = NULL;
struct ext4_ext_path *org_path = NULL;
struct ext4_ext_path *dest_path = NULL;
struct ext4_extent *oext, *dext, *swap_ext;
@@ -1314,7 +1150,6 @@ ext4_ext_replace_branches(struct inode *org_inode, struct inode *dest_inode,
int err = 0;
int depth;
int replaced_count = 0;
- unsigned jnum;

from = (ext4_lblk_t)from_page <<
(PAGE_CACHE_SHIFT - dest_inode->i_blkbits);
@@ -1322,12 +1157,6 @@ ext4_ext_replace_branches(struct inode *org_inode, struct inode *dest_inode,
(PAGE_CACHE_SHIFT - dest_inode->i_blkbits);
dest_off = (ext4_lblk_t)dest_from_page <<
(PAGE_CACHE_SHIFT - dest_inode->i_blkbits);
- jnum = ext4_ext_writepage_trans_blocks(org_inode, count) + 3;
- handle = ext4_journal_start(org_inode, jnum);
- if (IS_ERR(handle)) {
- err = PTR_ERR(handle);
- goto out;
- }

/* Get the original extent for the block "from" */
org_path = ext4_ext_find_extent(org_inode, from, NULL);
@@ -1361,7 +1190,7 @@ ext4_ext_replace_branches(struct inode *org_inode, struct inode *dest_inode,
ext4_ext_store_pblock(&tmp_ext2, ext_pblock(oext) + org_diff);
tmp_ext2.ee_block = tmp_ext.ee_block;

- /* adjust extent length when blocksize != pagesize */
+ /* Adjust extent length when blocksize != pagesize */
if (tmp_ext.ee_len <= (oext->ee_len - org_diff)) {
tmp_ext2.ee_len = tmp_ext.ee_len;
} else {
@@ -1370,22 +1199,23 @@ ext4_ext_replace_branches(struct inode *org_inode, struct inode *dest_inode,
}
swap_ext = &tmp_ext2;

- /* loop for the destination extents */
+ /* Loop for the destination extents */
while (1) {
/* The extent for destination must be found. */
BUG_ON(!dext || dest_off != le32_to_cpu(dext->ee_block));

- /* loop for the original extent blocks */
- err = ext4_ext_defrag_leaf_block(handle, org_inode,
+ /* Loop for the original extent blocks */
+ err = ext4_defrag_leaf_block(handle, org_inode,
org_path, dext, &from, flag);
if (err < 0)
goto out;

- /* We need the function which fixes extent information for
+ /*
+ * We need the function which fixes extent information for
* inserting.
- * e.g. ext4_ext_merge_extents().
+ * e.g. ext4_defrag_merge_extents()
*/
- err = ext4_ext_defrag_leaf_block(handle, dest_inode,
+ err = ext4_defrag_leaf_block(handle, dest_inode,
dest_path, swap_ext, &dest_off, -1);
if (err < 0)
goto out;
@@ -1444,7 +1274,7 @@ ext4_ext_replace_branches(struct inode *org_inode, struct inode *dest_inode,
ext4_ext_store_pblock(&tmp_ext2, ext_pblock(oext) + org_diff);
tmp_ext2.ee_block = tmp_ext.ee_block;

- /* adjust extent length when blocksize != pagesize */
+ /* Adjust extent length when blocksize != pagesize */
if (tmp_ext.ee_len <= (oext->ee_len - org_diff)) {
tmp_ext2.ee_len = tmp_ext.ee_len;
} else {
@@ -1455,8 +1285,6 @@ ext4_ext_replace_branches(struct inode *org_inode, struct inode *dest_inode,
}

out:
- if (handle)
- ext4_journal_stop(handle);
if (org_path) {
ext4_ext_drop_refs(org_path);
kfree(org_path);
@@ -1470,18 +1298,19 @@ out:
}

/**
- * ext4_ext_alloc_blocks - allocate contiguous blocks to temporary inode
- * @dest_inode temporary inode for multiple block allocation
- * @org_inode original inode
- * @iblock file related offset
- * @total_blocks contiguous blocks count
- * @goal block offset for allocation
- * @phase phase of create free space mode
+ * ext4_defrag_alloc_blocks - Allocate contiguous blocks to temporary inode
+ *
+ * @dest_inode temporary inode for multiple block allocation
+ * @org_inode original inode
+ * @iblock file related offset
+ * @total_blocks contiguous blocks count
+ * @goal block offset for allocation
+ * @phase phase of the force defrag mode
*
* If succeed, fuction returns count of extent we got,
* otherwise returns err.
*/
-static int ext4_ext_alloc_blocks(struct inode *dest_inode,
+static int ext4_defrag_alloc_blocks(struct inode *dest_inode,
struct inode *org_inode, ext4_lblk_t iblock,
ext4_fsblk_t total_blocks, ext4_fsblk_t goal, int phase)
{
@@ -1525,7 +1354,7 @@ static int ext4_ext_alloc_blocks(struct inode *dest_inode,
ar.excepted_group = -1;
}

- /* Find first extent. */
+ /* Find first extent */
dest_path = ext4_ext_find_extent(dest_inode, iblock, dest_path);
if (IS_ERR(dest_path)) {
err = PTR_ERR(dest_path);
@@ -1536,11 +1365,11 @@ static int ext4_ext_alloc_blocks(struct inode *dest_inode,
ar.inode = dest_inode;
ar.flags = EXT4_MB_HINT_DATA | EXT4_MB_HINT_RESERVED
| EXT4_MB_HINT_NOPREALLOC;
- if (goal) {
+
+ if (goal)
ar.goal = goal;
- } else {
+ else
ar.goal = ext4_ext_find_goal(dest_inode, dest_path, iblock);
- }

ar.logical = iblock;
ar.lleft = 0;
@@ -1572,23 +1401,22 @@ static int ext4_ext_alloc_blocks(struct inode *dest_inode,
(phase == DEFRAG_FORCE_TRY)) {
ext4_free_blocks(handle, org_inode, newblock,
ar.len, metadata);
- /* go to force mode */
+ /* -ENOSPC triggers DEFRAG_FORCE_VICTIM phase. */
err = -ENOSPC;
goto out;
} else {
/*
- * If ext4_mb_new_blocks() allcates
- * the block which used to be the metadata block,
- * its dirty buffer_head causes the overwriting
- * with old metadata.
+ * Dirty buffer_head causes the overwriting
+ * if ext4_mb_new_blocks() allocates the block
+ * which used to be the metadata block.
* We should call unmap_underlying_metadata()
* to clear the dirty flag.
*/
for (len_cnt = 0; len_cnt < ar.len; len_cnt++) {
bh = sb_find_get_block(org_sb,
- newblock + len_cnt);
+ newblock + len_cnt);
unmap_underlying_metadata(org_sb->s_bdev,
- newblock + len_cnt);
+ newblock + len_cnt);
}

alloc_total += ar.len;
@@ -1596,12 +1424,18 @@ static int ext4_ext_alloc_blocks(struct inode *dest_inode,
goal, &goal_grp_no, &goal_blk_off);
ext4_get_group_no_and_offset(dest_inode->i_sb,
newblock, &dest_grp_no, &dest_blk_off);
- /* We can't allocate at the same block group */
+
+ /* Only the force defrag mode */
switch (phase) {
case DEFRAG_FORCE_VICTIM:
+ /*
+ * We can't allocate new blocks in the same
+ * block group.
+ */
if (dest_grp_no == org_grp_no) {
- printk(KERN_ERR "defrag: Can't allocate"
- " in same block group\n");
+ printk(KERN_ERR "ext4 defrag: "
+ "Failed to allocate victim file"
+ " to other block group\n");
ext4_free_blocks(handle, org_inode,
newblock, ar.len, metadata);
err = -ENOSPC;
@@ -1609,12 +1443,15 @@ static int ext4_ext_alloc_blocks(struct inode *dest_inode,
}
break;
case DEFRAG_FORCE_GATHER:
- /* Maybe reserved blocks are already used by
- other process */
+ /*
+ * Maybe reserved blocks are already used by
+ * other process.
+ */
if (dest_grp_no != goal_grp_no
|| alloc_total != total_blocks) {
- printk(KERN_ERR "defrag: Already used"
- " the specified blocks\n");
+ printk(KERN_ERR "ext4 defrag: "
+ "Reserved blocks are already "
+ "used by other process\n");
ext4_free_blocks(handle, org_inode,
newblock, ar.len, metadata);
err = -EIO;
@@ -1645,11 +1482,14 @@ static int ext4_ext_alloc_blocks(struct inode *dest_inode,
}

out:
- /* Faild case: We have to remove halfway blocks */
- if (err)
+ if (err) {
+ /* Faild case: We have to remove halfway blocks */
err2 = ext4_ext_remove_space(dest_inode, 0);
-
- /* Successful case */
+ if (err2)
+ printk(KERN_ERR "ext4 defrag: "
+ "Failed to remove temporary inode blocks\n");
+ }
+out2:
if (dest_path) {
ext4_ext_drop_refs(dest_path);
kfree(dest_path);
@@ -1658,107 +1498,27 @@ out:
ext4_ext_drop_refs(org_path);
kfree(org_path);
}
-out2:
- ext4_journal_stop(handle);
-
- if (err2) {
- return err2;
- } else if (err) {
- return err;
- }
- /* return extents count */
- return count;
-}
-
-/**
- * ext4_ext_defrag_partial - defrag original file partially
- * @filp: pointer to file
- * @org_offset: page index on original file
- * @dest_offset: page index on temporary file
- * @flag: defrag mode (e.g. -f)
- *
- * This function returns 0 if succeeded, otherwise returns error value
- */
-static int
-ext4_ext_defrag_partial(struct inode *tmp_inode, struct file *filp,
- pgoff_t org_offset, pgoff_t dest_offset, int flag)
-{
- struct inode *inode = filp->f_dentry->d_inode;
- struct address_space *mapping = inode->i_mapping;
- struct page *page;
- pgoff_t offset_in_page = PAGE_SIZE;
- int ret = 0;
-
- up_write(&EXT4_I(inode)->i_data_sem);
- page = read_cache_page(inode->i_mapping, org_offset,
- (filler_t *)inode->i_mapping->a_ops->readpage, NULL);
- down_write(&EXT4_I(inode)->i_data_sem);
-
- if (IS_ERR(page)) {
- ret = PTR_ERR(page);
- return ret;
- }
-
- lock_page(page);

- /*
- * try_to_release_page() doesn't call relasepage in writeback mode.
- * We should care about the order of writing to the same file
- * by multiple defrag processes.
- * It needs to call wait_on_page_writeback() to wait for the
- * writeback of the page.
- */
- if (PageWriteback(page))
- wait_on_page_writeback(page);
-
- /* release old bh and drop refs */
- try_to_release_page(page, 0);
- ret = ext4_ext_replace_branches(inode, tmp_inode, org_offset,
- dest_offset, 1, flag);
- if (ret < 0)
- goto ERR;
-
- /* Clear the inode cache not to refer to the old data. */
- ext4_ext_invalidate_cache(inode);
-
- if (org_offset == ((inode->i_size - 1) >> PAGE_CACHE_SHIFT)) {
- offset_in_page = (inode->i_size & (PAGE_CACHE_SIZE - 1));
- /*
- * If org_offset is the last page and i_size is
- * multiples of PAGE_CACHE_SIZE, set PAGE_CACHE_SIZE to
- * offset_in_page not to be 0.
- */
- if (offset_in_page == 0)
- offset_in_page = PAGE_CACHE_SIZE;
- }
-
- up_write(&EXT4_I(inode)->i_data_sem);
- ret = mapping->a_ops->prepare_write(filp, page,
- 0, offset_in_page);
- down_write(&EXT4_I(inode)->i_data_sem);
- if (ret)
- goto ERR;
+ ext4_journal_stop(handle);

- ret = mapping->a_ops->commit_write(filp, page,
- 0, offset_in_page);
-ERR:
- unlock_page(page);
- page_cache_release(page);
+ /* Return extents count or err value */
+ return (!err ? count : err);

- return (ret < 0 ? ret : 0);
}

/**
- * ext4_ext_defrag_partial2 - defrag_partial with write_{begin, end}
+ * ext4_defrag_partial - Defrag a file per page
+ *
+ * @tmp_inode: the inode which has blocks to swap with original
* @filp: pointer to file
* @org_offset: page index on original file
* @dest_offset: page index on temporary file
* @flag: defrag mode (e.g. -f)
*
- * This function returns 0 if succeeded, otherwise returns error value
+ * This function returns 0 if succeeded, otherwise returns error value.
*/
static int
-ext4_ext_defrag_partial2(struct inode *tmp_inode, struct file *filp,
+ext4_defrag_partial(struct inode *tmp_inode, struct file *filp,
pgoff_t org_offset, pgoff_t dest_offset, int flag)
{
struct inode *inode = filp->f_dentry->d_inode;
@@ -1766,7 +1526,9 @@ ext4_ext_defrag_partial2(struct inode *tmp_inode, struct file *filp,
struct buffer_head *bh;
struct page *page;
const struct address_space_operations *a_ops = mapping->a_ops;
+ handle_t *handle;
pgoff_t offset_in_page = PAGE_SIZE;
+ int jblocks;
int ret = 0;
int blocksize = inode->i_sb->s_blocksize;
int blocks_per_page = 0;
@@ -1776,15 +1538,26 @@ ext4_ext_defrag_partial2(struct inode *tmp_inode, struct file *filp,
unsigned int w_flags = 0;
void *fsdata;

+ /*
+ * It needs twice the amount of ordinary journal buffers because
+ * inode and tmp_inode may change each different metadata blocks.
+ */
+ jblocks = ext4_writepage_trans_blocks(inode) * 2;
+ handle = ext4_journal_start(inode, jblocks);
+ if (IS_ERR(handle)) {
+ ret = PTR_ERR(handle);
+ return ret;
+ }
+
if (segment_eq(get_fs(), KERNEL_DS))
w_flags |= AOP_FLAG_UNINTERRUPTIBLE;

if (org_offset == ((inode->i_size - 1) >> PAGE_CACHE_SHIFT)) {
offset_in_page = (inode->i_size & (PAGE_CACHE_SIZE - 1));
/*
- * If org_offset is the last page and i_size is
- * multiples of PAGE_CACHE_SIZE, set PAGE_CACHE_SIZE to
- * offset_in_page not to be 0.
+ * Set PAGE_CACHE_SIZE to offset_in_page not be 0
+ * if org_offset is the last page and i_size is
+ * multiples of PAGE_CACHE_SIZE.
*/
if (offset_in_page == 0)
offset_in_page = PAGE_CACHE_SIZE;
@@ -1796,7 +1569,7 @@ ext4_ext_defrag_partial2(struct inode *tmp_inode, struct file *filp,
down_write(&EXT4_I(inode)->i_data_sem);

if (unlikely(ret < 0))
- goto ERR;
+ goto out;

if (!PageUptodate(page)) {
mapping->a_ops->readpage(filp, page);
@@ -1813,15 +1586,15 @@ ext4_ext_defrag_partial2(struct inode *tmp_inode, struct file *filp,
if (PageWriteback(page))
wait_on_page_writeback(page);

- /* release old bh and drop refs */
+ /* Release old bh and drop refs */
try_to_release_page(page, 0);
- ret = ext4_ext_replace_branches(inode, tmp_inode, org_offset,
- dest_offset, 1, flag);
+ ret = ext4_defrag_replace_branches(handle, inode, tmp_inode,
+ org_offset, dest_offset, 1, flag);

if (ret < 0)
- goto ERR;
+ goto out;

- /* Clear the inode cache not to refer to the old data. */
+ /* Clear the inode cache not to refer to the old data */
ext4_ext_invalidate_cache(inode);

if (!page_has_buffers(page))
@@ -1837,7 +1610,7 @@ ext4_ext_defrag_partial2(struct inode *tmp_inode, struct file *filp,
down_write(&EXT4_I(inode)->i_data_sem);

if (ret < 0)
- goto ERR;
+ goto out;

if (bh->b_this_page != NULL)
bh = bh->b_this_page;
@@ -1847,13 +1620,16 @@ ext4_ext_defrag_partial2(struct inode *tmp_inode, struct file *filp,
offset_in_page, page, fsdata);

if (unlikely(ret < 0))
- goto ERR;
-ERR:
+ goto out;
+out:
+ ext4_journal_stop(handle);
+
return (ret < 0 ? ret : 0);
}

/**
- * ext4_ext_new_extent_tree - allocate contiguous blocks
+ * ext4_defrag_new_extent_tree - Allocate contiguous blocks
+ *
* @inode: inode of the original file
* @tmp_inode: inode of the temporary file
* @path: the structure holding some info about
@@ -1862,15 +1638,15 @@ ERR:
* @tar_blocks: the number of blocks to allocate
* @iblock: file related offset
* @goal: block offset for allocaton
- * @flag: phase of create free space mode
+ * @flag: phase of the force defrag mode
*
* This function returns the value as below:
- * 0(succeeded)
+ * 0(succeeded)
* 1(not improved)
* negative value(error)
*/
static int
-ext4_ext_new_extent_tree(struct inode *inode, struct inode *tmp_inode,
+ext4_defrag_new_extent_tree(struct inode *inode, struct inode *tmp_inode,
struct ext4_ext_path *path, ext4_lblk_t tar_start,
ext4_lblk_t tar_blocks, ext4_lblk_t iblock,
ext4_fsblk_t goal, int flag)
@@ -1885,12 +1661,12 @@ ext4_ext_new_extent_tree(struct inode *inode, struct inode *tmp_inode,
eh = ext_inode_hdr(tmp_inode);
eh->eh_depth = 0;

- /* allocate contiguous blocks */
- sum_tmp = ext4_ext_alloc_blocks(tmp_inode, inode, iblock,
+ /* Allocate contiguous blocks */
+ sum_tmp = ext4_defrag_alloc_blocks(tmp_inode, inode, iblock,
tar_blocks, goal, flag);
if (sum_tmp < 0) {
ret = sum_tmp;
- goto ERR;
+ goto out;
}

depth = ext_depth(inode);
@@ -1904,45 +1680,46 @@ ext4_ext_new_extent_tree(struct inode *inode, struct inode *tmp_inode,
last_extent) {

if ((sum_org == sum_tmp) && !goal) {
- /* not improved */
- if (!(ret =
- ext4_ext_remove_space(tmp_inode, 0)))
+ /* Not improved */
+ ret = ext4_ext_remove_space(tmp_inode, 0);
+ if (!ret)
ret = 1;
} else if (sum_org < sum_tmp &&
flag != DEFRAG_FORCE_VICTIM) {
- /* fragment increased */
- if (!(ret =
- ext4_ext_remove_space(tmp_inode, 0)))
+ /* Fragment increased */
+ ret = ext4_ext_remove_space(tmp_inode, 0);
+ if (!ret)
ret = -ENOSPC;
- printk("defrag failed due to no space\n");
- }
+ printk(KERN_ERR "ext4 defrag: "
+ "Insufficient free blocks\n");
+ }
break;
}
- if ((last_extent =
- ext4_ext_next_extent(tmp_inode,
- path, &ext)) < 0) {
+ last_extent = ext4_defrag_next_extent(tmp_inode, path, &ext);
+ if (last_extent < 0) {
ret = last_extent;
break;
}
}
-ERR:
+out:
return ret;
}

/**
- * ext4_ext_defrag - defrag whole file
- * @filp: pointer to file
- * @from: starting offset to defrag in blocks
- * @defrag_size: size of defrag in blocks
- * @goal: block offset for allocation
- * @flag: phase of create free space mode
- * @ext: extent to be moved (only -f)
+ * ext4_defrag - Defrag the specified range of a file
+ *
+ * @filp: pointer to file
+ * @from: starting offset to defrag in blocks
+ * @defrag_size: size of defrag in blocks
+ * @goal: block offset for allocation
+ * @flag: phase of the force defrag mode
+ * @ext: extent to be moved (only -f)
*
* This function returns the number of blocks if succeeded, otherwise
- * returns error value
+ * returns error value.
*/
int
-ext4_ext_defrag(struct file *filp, ext4_lblk_t block_start,
+ext4_defrag(struct file *filp, ext4_lblk_t block_start,
ext4_lblk_t defrag_size, ext4_fsblk_t goal,
int flag, struct ext4_extent_data *ext)
{
@@ -1958,20 +1735,26 @@ ext4_ext_defrag(struct file *filp, ext4_lblk_t block_start,
pgoff_t page_offset = 0, dest_offset = 0, seq_end_page = 0;
int ret = 0, depth = 0, last_extent = 0, seq_extents = 0;

- /* Check goal offset if goal offset was given from userspace. */
+ /* ext4 defrag needs mballoc mount option. */
+ if (!test_opt(inode->i_sb, MBALLOC)) {
+ printk(KERN_ERR "ext4 defrag: multiblock allocation "
+ "is disabled\n");
+ return -EINVAL;
+ }
+
+ /* Check goal offset if goal offset was given from userspace */
if (((0 < goal) && (ext4_blocks_count(es) < goal)) && (goal != -1)) {
- printk(KERN_ERR "defrag: incorrect goal number %llu, "
- "you can set goal until %llu\n", goal,
- ext4_blocks_count(es));
- ret = -EINVAL;
- goto ERR1;
+ printk(KERN_ERR "ext4 defrag: Invalid goal offset %llu, "
+ "you can set goal offset up to %llu\n", goal,
+ ext4_blocks_count(es));
+ return -EINVAL;
}

- /* Setup for fixed blocks mode */
if (ext->len) {
+ /* Setup for the force defrag mode */
if (ext->len < defrag_size) {
- printk("Cannot defrag due to the insufficient"
- " specified free blocks\n");
+ printk(KERN_ERR "ext4 defrag: "
+ "Invalid length of extent\n");
return -EINVAL;
}
flag = DEFRAG_FORCE_GATHER;
@@ -1988,51 +1771,51 @@ ext4_ext_defrag(struct file *filp, ext4_lblk_t block_start,
if (IS_ERR(path)) {
ret = PTR_ERR(path);
path = NULL;
- goto ERR2;
+ goto out;
}

- /* get path structure to check hole */
+ /* Get path structure to check the hole */
holecheck_path = ext4_ext_find_extent(inode, block_start, NULL);
if (IS_ERR(holecheck_path)) {
ret = PTR_ERR(holecheck_path);
holecheck_path = NULL;
- goto ERR2;
+ goto out;
}

depth = ext_depth(inode);
ext_cur = holecheck_path[depth].p_ext;
if (ext_cur == NULL)
- goto ERR2;
+ goto out;

/*
- * if block_start was within the hole, get proper extent whose ee_block
- * is beyond block_start
+ * Get proper extent whose ee_block is beyond block_start
+ * if block_start was within the hole.
*/
if (le32_to_cpu(ext_cur->ee_block) +
le32_to_cpu(ext_cur->ee_len) - 1 < block_start) {
- if ((last_extent =
- ext4_ext_next_extent(inode, holecheck_path,
- &ext_cur)) < 0) {
+ last_extent = ext4_defrag_next_extent(inode, holecheck_path,
+ &ext_cur);
+ if (last_extent < 0) {
ret = last_extent;
- goto ERR2;
+ goto out;
}
- if ((last_extent =
- ext4_ext_next_extent(inode, path,
- &ext_dummy)) < 0) {
+ last_extent = ext4_defrag_next_extent(inode, path, &ext_dummy);
+ if (last_extent < 0) {
ret = last_extent;
- goto ERR2;
+ goto out;
}
}
seq_extents = 1;
seq_start = ext_cur->ee_block;

- /* no blocks existed within designated range */
+ /* No blocks within the specified range. */
if (le32_to_cpu(ext_cur->ee_block) > block_end) {
- printk("nothing done due to the lack of contiguous blocks\n");
- goto ERR2;
+ printk(KERN_INFO "ext4 defrag: The specified range of file"
+ " may be the hole\n");
+ goto out;
}

- /* adjust start blocks */
+ /* Adjust start blocks */
add_blocks = min(ext_cur->ee_block +
ext_cur->ee_len, block_end + 1) -
max(ext_cur->ee_block, block_start);
@@ -2046,7 +1829,7 @@ ext4_ext_defrag(struct file *filp, ext4_lblk_t block_start,
2 * EXT4_QUOTA_INIT_BLOCKS(inode->i_sb) + 1);
if (IS_ERR(handle)) {
ret = PTR_ERR(handle);
- goto ERR1;
+ goto out;
}
tmp_inode = ext4_new_inode(handle,
inode->i_sb->s_root->d_inode, S_IFREG);
@@ -2054,7 +1837,7 @@ ext4_ext_defrag(struct file *filp, ext4_lblk_t block_start,
ret = -ENOMEM;
ext4_journal_stop(handle);
tmp_inode = NULL;
- goto ERR1;
+ goto out;
}

i_size_write(tmp_inode, i_size_read(inode));
@@ -2063,14 +1846,14 @@ ext4_ext_defrag(struct file *filp, ext4_lblk_t block_start,
ext4_orphan_add(handle, tmp_inode);
ext4_journal_stop(handle);

- /* adjust tail blocks */
+ /* Adjust tail blocks */
if (seq_start + seq_blocks - 1 > block_end)
seq_blocks = block_end - seq_start + 1;

ext_prev = ext_cur;
- if ((last_extent =
- ext4_ext_next_extent(inode, holecheck_path,
- &ext_cur)) < 0) {
+ last_extent = ext4_defrag_next_extent(inode, holecheck_path,
+ &ext_cur);
+ if (last_extent < 0) {
ret = last_extent;
break;
}
@@ -2078,8 +1861,9 @@ ext4_ext_defrag(struct file *filp, ext4_lblk_t block_start,
seq_extents++;
add_blocks = le16_to_cpu(ext_cur->ee_len);

- /* found hole or reached the tail of either a designated range
- * or the file
+ /*
+ * Extend the length of contiguous block (seq_blocks)
+ * if extents are contiguous.
*/
if ((le32_to_cpu(ext_prev->ee_block) +
le16_to_cpu(ext_prev->ee_len) ==
@@ -2093,13 +1877,13 @@ ext4_ext_defrag(struct file *filp, ext4_lblk_t block_start,
continue;
}

- /* found an isolated block */
+ /* Found an isolated block */
if ((seq_extents == 1) && !goal) {
seq_start = ext_cur->ee_block;
goto CLEANUP;
}

- ret = ext4_ext_new_extent_tree(inode, tmp_inode, path,
+ ret = ext4_defrag_new_extent_tree(inode, tmp_inode, path,
seq_start, seq_blocks, block_start, goal, flag);

if (ret < 0) {
@@ -2118,40 +1902,26 @@ ext4_ext_defrag(struct file *filp, ext4_lblk_t block_start,
dest_offset = 0;
seq_start = le32_to_cpu(ext_cur->ee_block);

- /* Discard all preallocations.
+ /*
+ * Discard all preallocations.
* This is provisional solution.
* When true ext4_mb_return_to_preallocation() is
* implemented, this will be removed.
*/
ext4_mb_discard_inode_preallocations(inode);

- if (inode->i_mapping->a_ops->write_begin) {
- while (page_offset <= seq_end_page) {
- /* replace original branches for new branches */
- ret = ext4_ext_defrag_partial2(tmp_inode,
- filp, page_offset,
- dest_offset, flag);
- if (ret < 0)
- goto ERR2;
-
- page_offset++;
- dest_offset++;
- }
- } else {
- while (page_offset <= seq_end_page) {
- /* replace original branches for new branches */
- ret = ext4_ext_defrag_partial(tmp_inode,
- filp, page_offset,
- dest_offset, flag);
- if (ret < 0)
- goto ERR2;
-
- page_offset++;
- dest_offset++;
- }
+ while (page_offset <= seq_end_page) {
+ /* Swap original branches with new branches */
+ ret = ext4_defrag_partial(tmp_inode, filp,
+ page_offset, dest_offset, flag);
+ if (ret < 0)
+ goto out;
+
+ page_offset++;
+ dest_offset++;
}

- /* decrease buffer counter */
+ /* Decrease buffer counter */
if (holecheck_path)
ext4_ext_drop_refs(holecheck_path);
holecheck_path =
@@ -2164,7 +1934,7 @@ ext4_ext_defrag(struct file *filp, ext4_lblk_t block_start,
depth = holecheck_path->p_depth;

CLEANUP:
- /* decrease buffer counter */
+ /* Decrease buffer counter */
if (path)
ext4_ext_drop_refs(path);
path = ext4_ext_find_extent(inode, seq_start, path);
@@ -2186,7 +1956,7 @@ CLEANUP:
}
}

-ERR2:
+out:
if (path) {
ext4_ext_drop_refs(path);
kfree(path);
@@ -2195,7 +1965,7 @@ ERR2:
ext4_ext_drop_refs(holecheck_path);
kfree(holecheck_path);
}
-ERR1:
+
up_write(&EXT4_I(inode)->i_data_sem);
mutex_unlock(&inode->i_mutex);

diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index ff18c70..767e550 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -1531,11 +1531,10 @@ repeat:
le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max));
}

- if (defrag) {
+ if (defrag)
defrag_goal = ext_pblock(newext);
- } else {
+ else
defrag_goal = 0;
- }
/*
* There is no free space in the found leaf.
* We're gonna add a new leaf in the tree.
diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c
index 2f7524f..ed3876b 100644
--- a/fs/ext4/ioctl.c
+++ b/fs/ext4/ioctl.c
@@ -231,9 +231,6 @@ flags_err:

return err;
}
- case EXT4_IOC_GET_EXTENTS:
- case EXT4_IOC_GET_TREE_STATS:
- case EXT4_IOC_GET_TREE_DEPTH:
case EXT4_IOC_FIBMAP:
case EXT4_IOC_DEFRAG:
case EXT4_IOC_GROUP_INFO:
@@ -242,7 +239,7 @@ flags_err:
case EXT4_IOC_RESERVE_BLOCK:
case EXT4_IOC_MOVE_VICTIM:
case EXT4_IOC_BLOCK_RELEASE: {
- return ext4_ext_ioctl(inode, filp, cmd, arg);
+ return ext4_defrag_ioctl(inode, filp, cmd, arg);
}
case EXT4_IOC_GROUP_ADD: {
struct ext4_new_group_data input;
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index 7f1ff75..b07f34f 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -2046,9 +2046,8 @@ repeat:
group = 0;

if (ac->ac_excepted_group != -1 &&
- group == ac->ac_excepted_group) {
+ group == ac->ac_excepted_group)
continue;
- }

/* quick check to skip empty groups */
grp = ext4_get_group_info(ac->ac_sb, group);
diff --git a/include/linux/ext4_fs.h b/include/linux/ext4_fs.h
index 84631ec..03b4154 100644
--- a/include/linux/ext4_fs.h
+++ b/include/linux/ext4_fs.h
@@ -300,9 +300,6 @@ struct ext4_new_group_data {
#define EXT4_IOC_GETRSVSZ _IOR('f', 5, long)
#define EXT4_IOC_SETRSVSZ _IOW('f', 6, long)
#define EXT4_IOC_MIGRATE _IO('f', 7)
-#define EXT4_IOC_GET_EXTENTS _IOR('f', 7, long)
-#define EXT4_IOC_GET_TREE_DEPTH _IOR('f', 8, long)
-#define EXT4_IOC_GET_TREE_STATS _IOR('f', 9, long)
#define EXT4_IOC_FIBMAP _IOW('f', 9, ext4_fsblk_t)
#define EXT4_IOC_DEFRAG _IOW('f', 10, struct ext4_ext_defrag_data)
#define EXT4_IOC_GROUP_INFO _IOW('f', 11, struct ext4_group_data_info)
@@ -310,7 +307,7 @@ struct ext4_new_group_data {
#define EXT4_IOC_EXTENTS_INFO _IOW('f', 13, struct ext4_extents_info)
#define EXT4_IOC_RESERVE_BLOCK _IOW('f', 14, struct ext4_extents_info)
#define EXT4_IOC_MOVE_VICTIM _IOW('f', 15, struct ext4_extents_info)
-#define EXT4_IOC_BLOCK_RELEASE _IO('f', 16)
+#define EXT4_IOC_BLOCK_RELEASE _IO('f', 8)

/*
* ioctl commands in 32 bit emulation
@@ -1174,10 +1171,10 @@ extern void ext4_inode_table_set(struct super_block *sb,
/* extents.c */
extern handle_t *ext4_ext_journal_restart(handle_t *handle, int needed);
/* defrag.c */
-extern int ext4_ext_defrag(struct file *filp, ext4_lblk_t block_start,
- ext4_lblk_t defrag_size, ext4_fsblk_t goal,
- int flag, struct ext4_extent_data *ext);
-extern int ext4_ext_ioctl(struct inode *, struct file *, unsigned int,
+extern int ext4_defrag(struct file *filp, ext4_lblk_t block_start,
+ ext4_lblk_t defrag_size, ext4_fsblk_t goal,
+ int flag, struct ext4_extent_data *ext);
+extern int ext4_defrag_ioctl(struct inode *, struct file *, unsigned int,
unsigned long);

static inline ext4_fsblk_t ext4_blocks_count(struct ext4_super_block *es)
diff --git a/include/linux/ext4_fs_extents.h b/include/linux/ext4_fs_extents.h
index 7f0140f..6fb42b1 100644
--- a/include/linux/ext4_fs_extents.h
+++ b/include/linux/ext4_fs_extents.h
@@ -124,19 +124,6 @@ struct ext4_ext_path {
#define EXT4_EXT_CACHE_GAP 1
#define EXT4_EXT_CACHE_EXTENT 2

-/*
- * to be called by ext4_ext_walk_space()
- * negative retcode - error
- * positive retcode - signal for ext4_ext_walk_space(), see below
- * callback must return valid extent (passed or newly created)
- */
-typedef int (*ext_prepare_callback)(struct inode *, struct ext4_ext_path *,
- struct ext4_ext_cache *,
- void *);

2008-03-06 05:26:34

by Andreas Dilger

Subject: Re: [RFC][PATCH 2/3] ext4 online defrag (ver 0.7)

On Mar 06, 2008 09:01 +0900, Akira Fujita wrote:
> Change the name of functions (ext4_ext_xxx -> ext4_defrag_xxx)
> and some cleanups.
>
> -int ext4_ext_walk_space(struct inode *inode, ext4_lblk_t block,
> - ext4_lblk_t num, ext_prepare_callback func,
> - void *cbdata)
> -{

We were just going to be using this function for the FIEMAP support.
I think there are several other functions which are not specific
to defrag, so they shouldn't be renamed to be defrag specific.

Cheers, Andreas
--
Andreas Dilger
Sr. Staff Engineer, Lustre Group
Sun Microsystems of Canada, Inc.

2008-03-06 06:25:17

by Akira Fujita

Subject: Re: [RFC][PATCH 2/3] ext4 online defrag (ver 0.7)

Hi Andreas,
> On Mar 06, 2008 09:01 +0900, Akira Fujita wrote:
>
>> Change the name of functions (ext4_ext_xxx -> ext4_defrag_xxx)
>> and some cleanups.
>>
>> -int ext4_ext_walk_space(struct inode *inode, ext4_lblk_t block,
>> - ext4_lblk_t num, ext_prepare_callback func,
>> - void *cbdata)
>> -{
>>
>
> We were just going to be using this function for the FIEMAP support.
> I think there are several other functions which are not specific
> to defrag, so they shouldn't be renamed to be defrag specific.
>

ext4_ext_walk_space is no longer used by ext4 online defrag.
So I just removed from defrag.c not renamed.

Shouldn't I remove ext4_ext_walk_space from defrag
for support FIEMAP?
I think it should be moved to extents.c

Regards, Akira

--
Akira Fujita <[email protected]>

2008-03-06 16:41:31

by Eric Sandeen

Subject: Re: [RFC][PATCH 2/3] ext4 online defrag (ver 0.7)

Akira Fujita wrote:
> Hi Andreas,
>> On Mar 06, 2008 09:01 +0900, Akira Fujita wrote:
>>
>>> Change the name of functions (ext4_ext_xxx -> ext4_defrag_xxx)
>>> and some cleanups.
>>>
>>> -int ext4_ext_walk_space(struct inode *inode, ext4_lblk_t block,
>>> - ext4_lblk_t num, ext_prepare_callback func,
>>> - void *cbdata)
>>> -{
>>>
>> We were just going to be using this function for the FIEMAP support.
>> I think there are several other functions which are not specific
>> to defrag, so they shouldn't be renamed to be defrag specific.
>>
>
> ext4_ext_walk_space is no longer used by ext4 online defrag.
> So I just removed from defrag.c not renamed.
>
> Shouldn't I remove ext4_ext_walk_space from defrag
> for support FIEMAP?
> I think it should be moved to extents.c

I agree. I'm working on FIEMAP, and I'll put it back into extents.c
along with the FIEMAP patch, if you don't need it for defrag.

-Eric

2008-03-06 23:46:53

by Akira Fujita

Subject: Re: [RFC][PATCH 2/3] ext4 online defrag (ver 0.7)

Hello Eric,

>>>> Change the name of functions (ext4_ext_xxx -> ext4_defrag_xxx)
>>>> and some cleanups.
>>>>
>>>> -int ext4_ext_walk_space(struct inode *inode, ext4_lblk_t block,
>>>> - ext4_lblk_t num, ext_prepare_callback func,
>>>> - void *cbdata)
>>>> -{
>>>>
>>>>
>>> We were just going to be using this function for the FIEMAP support.
>>> I think there are several other functions which are not specific
>>> to defrag, so they shouldn't be renamed to be defrag specific.
>>>
>>>
>> ext4_ext_walk_space is no longer used by ext4 online defrag.
>> So I just removed from defrag.c not renamed.
>>
>> Shouldn't I remove ext4_ext_walk_space from defrag
>> for support FIEMAP?
>> I think it should be moved to extents.c
>>
>
> I agree. I'm working on FIEMAP, and I'll put it back into extents.c
> along with the FIEMAP patch, if you don't need it for defrag.
>
>
OK. Please put ext4_ext_walk_space in extents.c.
Thank you for coping. :-)

Cheers Akira

2008-03-08 02:13:53

by Mingming Cao

Subject: Re:[PATCH 3/5] ext4: online defrag -- Move the file data to the new blocks

ext4: online defrag -- Move the file data to the new blocks
>
> From: Akira Fujita <[email protected]>
>
> Move the blocks on the temporary inode to the original inode
> by a page.
> 1. Read the file data from the old blocks to the page
> 2. Move the block on the temporary inode to the original inode
> 3. Write the file data on the page into the new blocks
>
> Signed-off-by: Akira Fujita <[email protected]>
> Signed-off-by: Takashi Sato <[email protected]>
>
This patch is a bit of too big to review, will do it later....it would helpful to add more
comments as always, and small patches are prefered...

Mingming

> --
> fs/ext4/defrag.c | 738 +++++++++++++++++++++++++++++++++++++++++++++++++++++
> fs/ext4/extents.c | 2 +-
> fs/ext4/inode.c | 3 +-
> 3 files changed, 740 insertions(+), 3 deletions(-)
>
> diff --git a/fs/ext4/defrag.c b/fs/ext4/defrag.c
> index 6121705..19be87a 100644
> --- a/fs/ext4/defrag.c
> +++ b/fs/ext4/defrag.c
> @@ -127,6 +127,623 @@ int ext4_defrag_ioctl(struct inode *inode, struct file *filp, unsigned int cmd,
> }
>
> /**
> + * ext4_defrag_merge_across_blocks - Merge extents across leaf block
> + *
> + * @handle journal handle
> + * @inode target file's inode
> + * @o_start first original extent to be defraged
> + * @o_end last original extent to be defraged
> + * @start_ext first new extent to be merged
> + * @new_ext middle of new extent to be merged
> + * @end_ext last new extent to be merged
> + * @flag defrag mode (e.g. -f)
> + *
> + * This function returns 0 if succeed, otherwise returns error value.
> + */
> +static int
> +ext4_defrag_merge_across_blocks(handle_t *handle, struct inode *inode,
> + struct ext4_extent *o_start,
> + struct ext4_extent *o_end, struct ext4_extent *start_ext,
> + struct ext4_extent *new_ext, struct ext4_extent *end_ext,
> + int flag)
> +{
> + struct ext4_ext_path *org_path = NULL;
> + ext4_lblk_t eblock = 0;
> + int err = 0;
> + int new_flag = 0;
> + int end_flag = 0;
> + int defrag_flag;
> +
> + if (flag == DEFRAG_FORCE_VICTIM)
> + defrag_flag = 1;
> + else
> + defrag_flag = 0;
> +
> + if (le16_to_cpu(start_ext->ee_len) &&
> + le16_to_cpu(new_ext->ee_len) &&
> + le16_to_cpu(end_ext->ee_len)) {
> +
> + if ((o_start) == (o_end)) {
> +
> + /* start_ext new_ext end_ext
> + * dest |---------|-----------|--------|
> + * org |------------------------------|
> + */
> +
> + end_flag = 1;
> + } else {
> +
> + /* start_ext new_ext end_ext
> + * dest |---------|----------|---------|
> + * org |---------------|--------------|
> + */
> +
> + o_end->ee_block = end_ext->ee_block;
> + o_end->ee_len = end_ext->ee_len;
> + ext4_ext_store_pblock(o_end, ext_pblock(end_ext));
> + }
> +
> + o_start->ee_len = start_ext->ee_len;
> + new_flag = 1;
> +
> + } else if ((le16_to_cpu(start_ext->ee_len)) &&
> + (le16_to_cpu(new_ext->ee_len)) &&
> + (!le16_to_cpu(end_ext->ee_len)) &&
> + ((o_start) == (o_end))) {
> +
> + /* start_ext new_ext
> + * dest |--------------|---------------|
> + * org |------------------------------|
> + */
> +
> + o_start->ee_len = start_ext->ee_len;
> + new_flag = 1;
> +
> + } else if ((!le16_to_cpu(start_ext->ee_len)) &&
> + (le16_to_cpu(new_ext->ee_len)) &&
> + (le16_to_cpu(end_ext->ee_len)) &&
> + ((o_start) == (o_end))) {
> +
> + /* new_ext end_ext
> + * dest |--------------|---------------|
> + * org |------------------------------|
> + */
> +
> + o_end->ee_block = end_ext->ee_block;
> + o_end->ee_len = end_ext->ee_len;
> + ext4_ext_store_pblock(o_end, ext_pblock(end_ext));
> +
> + /*
> + * Set 0 to the extent block if new_ext was
> + * the first block.
> + */
> + if (!new_ext->ee_block)
> + eblock = 0;
> + else
> + eblock = le32_to_cpu(new_ext->ee_block);
> +
> + new_flag = 1;
> + } else {
> + printk(KERN_ERR "ext4 defrag: Unexpected merge case\n");
> + return -EIO;
> + }
> +
> + if (new_flag) {
> + org_path = ext4_ext_find_extent(inode, eblock, NULL);
> + if (IS_ERR(org_path)) {
> + err = PTR_ERR(org_path);
> + org_path = NULL;
> + goto out;
> + }
> + err = ext4_ext_insert_extent_defrag(handle, inode,
> + org_path, new_ext, defrag_flag);
> + if (err)
> + goto out;
> + }
> +
> + if (end_flag) {
> + org_path = ext4_ext_find_extent(inode,
> + le32_to_cpu(end_ext->ee_block) - 1, org_path);
> + if (IS_ERR(org_path)) {
> + err = PTR_ERR(org_path);
> + org_path = NULL;
> + goto out;
> + }
> + err = ext4_ext_insert_extent_defrag(handle, inode,
> + org_path, end_ext, defrag_flag);
> + if (err)
> + goto out;
> + }
> +out:
> + if (org_path) {
> + ext4_ext_drop_refs(org_path);
> + kfree(org_path);
> + }
> +
> + return err;
> +
> +}
> +
> +/**
> + * ext4_defrag_merge_inside_block - Merge new extent to the extent block
> + *
> + * @handle journal handle
> + * @inode target file's inode
> + * @o_start first original extent to be defraged
> + * @o_end last original extent to be merged
> + * @start_ext first new extent to be merged
> + * @new_ext middle of new extent to be merged
> + * @end_ext last new extent to be merged
> + * @eh extent header of target leaf block
> + * @replaced the number of blocks which will be replaced with new_ext
> + * @range_to_move used to decide how to merge
> + *
> + * This function always returns 0.
> + */
> +static int
> +ext4_defrag_merge_inside_block(handle_t *handle, struct inode *inode,
> + struct ext4_extent *o_start, struct ext4_extent *o_end,
> + struct ext4_extent *start_ext, struct ext4_extent *new_ext,
> + struct ext4_extent *end_ext, struct ext4_extent_header *eh,
> + ext4_fsblk_t replaced, int range_to_move)
> +{
> + int i = 0;
> + unsigned len;
> +
> + /* Move the existing extents */
> + if (range_to_move && o_end < EXT_LAST_EXTENT(eh)) {
> + len = EXT_LAST_EXTENT(eh) - (o_end + 1) + 1;
> + len = len * sizeof(struct ext4_extent);
> + memmove(o_end + 1 + range_to_move, o_end + 1, len);
> + }
> +
> + /* Insert start entry */
> + if (le16_to_cpu(start_ext->ee_len))
> + o_start[i++].ee_len = start_ext->ee_len;
> +
> + /* Insert new entry */
> + if (le16_to_cpu(new_ext->ee_len)) {
> + o_start[i].ee_block = new_ext->ee_block;
> + o_start[i].ee_len = cpu_to_le16(replaced);
> + ext4_ext_store_pblock(&o_start[i++], ext_pblock(new_ext));
> + }
> +
> + /* Insert end entry */
> + if (end_ext->ee_len)
> + o_start[i] = *end_ext;
> +
> + /* Increment the total entries counter on the extent block */
> + eh->eh_entries
> + = cpu_to_le16(le16_to_cpu(eh->eh_entries) + range_to_move);
> +
> + return 0;
> +}
> +
> +/**
> + * ext4_defrag_merge_extents - Merge new extent
> + *
> + * @handle journal handle
> + * @inode target file's inode
> + * @org_path path indicates first extent to be defraged
> + * @o_start first original extent to be defraged
> + * @o_end last original extent to be defraged
> + * @start_ext first new extent to be merged
> + * @new_ext middle of new extent to be merged
> + * @end_ext last new extent to be merged
> + * @replaced the number of blocks which will be replaced with new_ext
> + * @flag defrag mode (e.g. -f)
> + *
> + * This function returns 0 if succeed, otherwise returns error value.
> + */
> +static int
> +ext4_defrag_merge_extents(handle_t *handle, struct inode *inode,
> + struct ext4_ext_path *org_path,
> + struct ext4_extent *o_start, struct ext4_extent *o_end,
> + struct ext4_extent *start_ext, struct ext4_extent *new_ext,
> + struct ext4_extent *end_ext, ext4_fsblk_t replaced, int flag)
> +{
> + struct ext4_extent_header *eh;
> + unsigned need_slots, slots_range;
> + int range_to_move, depth, ret;
> +
> + /*
> + * The extents need to be inserted
> + * start_extent + new_extent + end_extent.
> + */
> + need_slots = (le16_to_cpu(start_ext->ee_len) ? 1 : 0) +
> + (le16_to_cpu(end_ext->ee_len) ? 1 : 0) +
> + (le16_to_cpu(new_ext->ee_len) ? 1 : 0);
> +
> + /* The number of slots between start and end */
> + slots_range = o_end - o_start + 1;
> +
> + /* Range to move the end of extent */
> + range_to_move = need_slots - slots_range;
> + depth = org_path->p_depth;
> + org_path += depth;
> + eh = org_path->p_hdr;
> +
> + if (depth) {
> + /* Register to journal */
> + ret = ext4_journal_get_write_access(handle, org_path->p_bh);
> + if (ret)
> + return ret;
> + }
> +
> + /* Expansion */
> + if ((range_to_move > 0) &&
> + (range_to_move > le16_to_cpu(eh->eh_max)
> + - le16_to_cpu(eh->eh_entries))) {
> +
> + ret = ext4_defrag_merge_across_blocks(handle, inode, o_start,
> + o_end, start_ext, new_ext,
> + end_ext, flag);
> + if (ret < 0)
> + return ret;
> + } else {
> + ret = ext4_defrag_merge_inside_block(handle, inode, o_start,
> + o_end, start_ext, new_ext, end_ext,
> + eh, replaced, range_to_move);
> + if (ret < 0)
> + return ret;
> + }
> +
> + if (depth) {
> + ret = ext4_journal_dirty_metadata(handle, org_path->p_bh);
> + if (ret)
> + return ret;
> + } else {
> + ret = ext4_mark_inode_dirty(handle, inode);
> + if (ret < 0)
> + return ret;
> + }
> +
> + return 0;
> +
> +}
> +
> +/**
> + * ext4_defrag_leaf_block - Defragmentation for one leaf extent block
> + *
> + * @handle journal handle
> + * @org_inode target inode
> + * @org_path path indicates first extent to be defraged
> + * @dext destination extent
> + * @from start offset on the target file
> + * @flag defrag mode (e.g. -f)
> + *
> + * This function returns 0 if succeed, otherwise returns error value.
> + */
> +static int
> +ext4_defrag_leaf_block(handle_t *handle, struct inode *org_inode,
> + struct ext4_ext_path *org_path, struct ext4_extent *dext,
> + ext4_lblk_t *from, int flag)
> +{
> + unsigned long depth;
> + ext4_fsblk_t replaced = 0;
> + struct ext4_extent *oext, *o_start = NULL, *o_end = NULL, *prev_ext;
> + struct ext4_extent new_ext, start_ext, end_ext;
> + ext4_lblk_t new_end, lblock;
> + unsigned short len;
> + ext4_fsblk_t new_phys_end;
> + int ret;
> +
> + depth = ext_depth(org_inode);
> + start_ext.ee_len = end_ext.ee_len = 0;
> + o_start = o_end = oext = org_path[depth].p_ext;
> + ext4_ext_store_pblock(&new_ext, ext_pblock(dext));
> + new_ext.ee_len = dext->ee_len;
> + len = le16_to_cpu(new_ext.ee_len);
> + new_ext.ee_block = cpu_to_le32(*from);
> + lblock = le32_to_cpu(oext->ee_block);
> + new_end = le32_to_cpu(new_ext.ee_block)
> + + le16_to_cpu(new_ext.ee_len) - 1;
> + new_phys_end = ext_pblock(&new_ext)
> + + le16_to_cpu(new_ext.ee_len) - 1;
> +
> + /*
> + * First original extent
> + * dest |---------------|
> + * org |---------------|
> + */
> + if (le32_to_cpu(new_ext.ee_block) >
> + le32_to_cpu(oext->ee_block) &&
> + le32_to_cpu(new_ext.ee_block) <
> + le32_to_cpu(oext->ee_block)
> + + le16_to_cpu(oext->ee_len)) {
> + start_ext.ee_len = cpu_to_le32(le32_to_cpu(new_ext.ee_block)
> + - le32_to_cpu(oext->ee_block));
> + replaced += le16_to_cpu(oext->ee_len)
> + - le16_to_cpu(start_ext.ee_len);
> + } else if (oext > EXT_FIRST_EXTENT(org_path[depth].p_hdr)) {
> + /* We can merge previous extent. */
> + prev_ext = oext - 1;
> + if (((ext_pblock(prev_ext) + le32_to_cpu(prev_ext->ee_len))
> + == ext_pblock(&new_ext))
> + && (le32_to_cpu(prev_ext->ee_block)
> + + le32_to_cpu(prev_ext->ee_len)
> + == le32_to_cpu(new_ext.ee_block))) {
> + o_start = prev_ext;
> + start_ext.ee_len = cpu_to_le32(
> + le16_to_cpu(prev_ext->ee_len)
> + + le16_to_cpu(new_ext.ee_len));
> + new_ext.ee_len = 0;
> + }
> + }
> +
> + for (;;) {
> + /* The extent for destination must be found. */
> + BUG_ON(!oext || lblock != le32_to_cpu(oext->ee_block));
> + lblock += le16_to_cpu(oext->ee_len);
> +
> + /*
> + * Middle of original extent
> + * dest |-------------------|
> + * org |-----------------|
> + */
> + if (le32_to_cpu(new_ext.ee_block) <=
> + le32_to_cpu(oext->ee_block) &&
> + new_end >= le32_to_cpu(oext->ee_block)
> + + le16_to_cpu(oext->ee_len) - 1)
> + replaced += le16_to_cpu(oext->ee_len);
> +
> + /*
> + * Last original extent
> + * dest |----------------|
> + * org |---------------|
> + */
> + if (new_end >= le32_to_cpu(oext->ee_block) &&
> + new_end < le32_to_cpu(oext->ee_block)
> + + le16_to_cpu(oext->ee_len) - 1) {
> + end_ext.ee_len
> + = cpu_to_le16(le32_to_cpu(oext->ee_block)
> + + le16_to_cpu(oext->ee_len) - 1 - new_end);
> + ext4_ext_store_pblock(&end_ext, (ext_pblock(o_end)
> + + cpu_to_le16(oext->ee_len)
> + - cpu_to_le16(end_ext.ee_len)));
> + end_ext.ee_block
> + = cpu_to_le32(le32_to_cpu(o_end->ee_block)
> + + le16_to_cpu(oext->ee_len)
> + - le16_to_cpu(end_ext.ee_len));
> + replaced += le16_to_cpu(oext->ee_len)
> + - le16_to_cpu(end_ext.ee_len);
> + }
> +
> + /*
> + * Detected the block end, reached the number of replaced
> + * blocks to dext->ee_len. Then merge the extent.
> + */
> + if (oext == EXT_LAST_EXTENT(org_path[depth].p_hdr) ||
> + new_end <= le32_to_cpu(oext->ee_block)
> + + le16_to_cpu(oext->ee_len) - 1) {
> + ret = ext4_defrag_merge_extents(handle, org_inode,
> + org_path, o_start, o_end, &start_ext,
> + &new_ext, &end_ext, replaced, flag);
> + if (ret < 0)
> + return ret;
> +
> + /* All expected blocks are replaced */
> + if (le16_to_cpu(new_ext.ee_len) <= 0) {
> + if (DQUOT_ALLOC_BLOCK(org_inode, len))
> + return -EDQUOT;
> + return 0;
> + }
> +
> + /* Re-calculate new_ext */
> + new_ext.ee_len = cpu_to_le32(le16_to_cpu(new_ext.ee_len)
> + - replaced);
> + new_ext.ee_block =
> + cpu_to_le32(le32_to_cpu(new_ext.ee_block)
> + + replaced);
> + ext4_ext_store_pblock(&new_ext, ext_pblock(&new_ext)
> + + replaced);
> + replaced = 0;
> + start_ext.ee_len = end_ext.ee_len = 0;
> + o_start = NULL;
> +
> + /* All expected blocks are replaced. */
> + if (le16_to_cpu(new_ext.ee_len) <= 0) {
> + if (DQUOT_ALLOC_BLOCK(org_inode, len))
> + return -EDQUOT;
> + return 0;
> + }
> + }
> +
> + /* Get the next extent for original. */
> + if (org_path)
> + ext4_ext_drop_refs(org_path);
> + org_path = ext4_ext_find_extent(org_inode, lblock, org_path);
> + if (IS_ERR(org_path)) {
> + ret = PTR_ERR(org_path);
> + org_path = NULL;
> + return ret;
> + }
> + depth = ext_depth(org_inode);
> + oext = org_path[depth].p_ext;
> + if (oext->ee_block + oext->ee_len <= lblock)
> + return -ENOENT;
> +
> + o_end = oext;
> + if (!o_start)
> + o_start = oext;
> + }
> +}
> +
> +/**
> + * ext4_defrag_replace_branches - Replace original extents with new extents
> + *
> + * @handle journal handle
> + * @org_inode original inode
> + * @dest_inode temporary inode
> + * @from_page page offset of org_inode
> + * @dest_from_page page offset of dest_inode
> + * @count_page page count to be replaced
> + * @flag defrag mode (e.g. -f)
> + *
> + * This function returns 0 if succeed, otherwise returns error value.
> + * Replace extents for blocks from "from" to "from + count - 1".
> + */
> +static int
> +ext4_defrag_replace_branches(handle_t *handle, struct inode *org_inode,
> + struct inode *dest_inode, pgoff_t from_page,
> + pgoff_t dest_from_page, pgoff_t count_page, int flag)
> +{
> + struct ext4_ext_path *org_path = NULL;
> + struct ext4_ext_path *dest_path = NULL;
> + struct ext4_extent *oext, *dext, *swap_ext;
> + struct ext4_extent tmp_ext, tmp_ext2;
> + ext4_lblk_t from, count, dest_off, diff, org_diff;
> + int err = 0;
> + int depth;
> + int replaced_count = 0;
> +
> + from = (ext4_lblk_t)from_page <<
> + (PAGE_CACHE_SHIFT - dest_inode->i_blkbits);
> + count = (ext4_lblk_t)count_page <<
> + (PAGE_CACHE_SHIFT - dest_inode->i_blkbits);
> + dest_off = (ext4_lblk_t)dest_from_page <<
> + (PAGE_CACHE_SHIFT - dest_inode->i_blkbits);
> +
> + /* Get the original extent for the block "from" */
> + org_path = ext4_ext_find_extent(org_inode, from, NULL);
> + if (IS_ERR(org_path)) {
> + err = PTR_ERR(org_path);
> + org_path = NULL;
> + goto out;
> + }
> +
> + /* Get the destination extent for the head */
> + dest_path = ext4_ext_find_extent(dest_inode, dest_off, NULL);
> + if (IS_ERR(dest_path)) {
> + err = PTR_ERR(dest_path);
> + dest_path = NULL;
> + goto out;
> + }
> + depth = ext_depth(dest_inode);
> + dext = dest_path[depth].p_ext;
> + /* When dext is too large, pick up the target range. */
> + diff = dest_off - le32_to_cpu(dext->ee_block);
> + ext4_ext_store_pblock(&tmp_ext, ext_pblock(dext) + diff);
> + tmp_ext.ee_block = cpu_to_le32(le32_to_cpu(dext->ee_block) + diff);
> + tmp_ext.ee_len = cpu_to_le16(le16_to_cpu(dext->ee_len) - diff);
> + if (count < le16_to_cpu(tmp_ext.ee_len))
> + tmp_ext.ee_len = cpu_to_le16(count);
> + dext = &tmp_ext;
> +
> + depth = ext_depth(org_inode);
> + oext = org_path[depth].p_ext;
> + org_diff = from - le32_to_cpu(oext->ee_block);
> + ext4_ext_store_pblock(&tmp_ext2, ext_pblock(oext) + org_diff);
> + tmp_ext2.ee_block = tmp_ext.ee_block;
> +
> + /* Adjust extent length when blocksize != pagesize */
> + if (tmp_ext.ee_len <= (oext->ee_len - org_diff)) {
> + tmp_ext2.ee_len = tmp_ext.ee_len;
> + } else {
> + tmp_ext2.ee_len = oext->ee_len - org_diff;
> + tmp_ext.ee_len = tmp_ext2.ee_len;
> + }
> + swap_ext = &tmp_ext2;
> +
> + /* Loop for the destination extents */
> + while (1) {
> + /* The extent for destination must be found. */
> + BUG_ON(!dext || dest_off != le32_to_cpu(dext->ee_block));
> +
> + /* Loop for the original extent blocks */
> + err = ext4_defrag_leaf_block(handle, org_inode,
> + org_path, dext, &from, flag);
> + if (err < 0)
> + goto out;
> +
> + /*
> + * We need the function which fixes extent information for
> + * inserting.
> + * e.g. ext4_defrag_merge_extents()
> + */
> + err = ext4_defrag_leaf_block(handle, dest_inode,
> + dest_path, swap_ext, &dest_off, -1);
> + if (err < 0)
> + goto out;
> +
> + replaced_count += le16_to_cpu(dext->ee_len);
> + dest_off += le16_to_cpu(dext->ee_len);
> + from += le16_to_cpu(dext->ee_len);
> +
> + /* Already moved the expected blocks */
> + if (replaced_count >= count)
> + break;
> +
> + if (org_path)
> + ext4_ext_drop_refs(org_path);
> + org_path = ext4_ext_find_extent(org_inode, from, NULL);
> + if (IS_ERR(org_path)) {
> + err = PTR_ERR(org_path);
> + org_path = NULL;
> + goto out;
> + }
> + depth = ext_depth(org_inode);
> + oext = org_path[depth].p_ext;
> + if (oext->ee_block + oext->ee_len <= from) {
> + err = 0;
> + goto out;
> + }
> +
> + if (dest_path)
> + ext4_ext_drop_refs(dest_path);
> + dest_path = ext4_ext_find_extent(dest_inode, dest_off, NULL);
> + if (IS_ERR(dest_path)) {
> + err = PTR_ERR(dest_path);
> + dest_path = NULL;
> + goto out;
> + }
> + depth = ext_depth(dest_inode);
> + dext = dest_path[depth].p_ext;
> + if (dext->ee_block + dext->ee_len <= dest_off) {
> + err = 0;
> + goto out;
> + }
> +
> + /* When dext is too large, pick up the target range. */
> + diff = dest_off - le32_to_cpu(dext->ee_block);
> + ext4_ext_store_pblock(&tmp_ext, ext_pblock(dext) + diff);
> + tmp_ext.ee_block =
> + cpu_to_le32(le32_to_cpu(dext->ee_block) + diff);
> + tmp_ext.ee_len = cpu_to_le16(le16_to_cpu(dext->ee_len) - diff);
> +
> + if ((count - replaced_count) < le16_to_cpu(tmp_ext.ee_len))
> + tmp_ext.ee_len = count - replaced_count ;
> +
> + dext = &tmp_ext;
> +
> + org_diff = from - le32_to_cpu(oext->ee_block);
> + ext4_ext_store_pblock(&tmp_ext2, ext_pblock(oext) + org_diff);
> + tmp_ext2.ee_block = tmp_ext.ee_block;
> +
> + /* Adjust extent length when blocksize != pagesize */
> + if (tmp_ext.ee_len <= (oext->ee_len - org_diff)) {
> + tmp_ext2.ee_len = tmp_ext.ee_len;
> + } else {
> + tmp_ext2.ee_len = oext->ee_len - org_diff;
> + tmp_ext.ee_len = tmp_ext2.ee_len;
> + }
> + swap_ext = &tmp_ext2;
> + }
> +
> +out:
> + if (org_path) {
> + ext4_ext_drop_refs(org_path);
> + kfree(org_path);
> + }
> + if (dest_path) {
> + ext4_ext_drop_refs(dest_path);
> + kfree(dest_path);
> + }
> +
> + return err;
> +}
> +
> +/**
> * ext4_defrag_alloc_blocks - Allocate contiguous blocks to temporary inode
> *
> * @dest_inode temporary inode for multiple block allocation
> @@ -336,6 +953,127 @@ out2:
> }
>
> /**
> + * ext4_defrag_partial - Defrag a file per page
> + *
> + * @tmp_inode: the inode which has blocks to swap with original
> + * @filp: pointer to file
> + * @org_offset: page index on original file
> + * @dest_offset: page index on temporary file
> + * @flag: defrag mode (e.g. -f)
> + *
> + * This function returns 0 if succeeded, otherwise returns error value.
> + */
> +static int
> +ext4_defrag_partial(struct inode *tmp_inode, struct file *filp,
> + pgoff_t org_offset, pgoff_t dest_offset, int flag)
> +{
> + struct inode *inode = filp->f_dentry->d_inode;
> + struct address_space *mapping = inode->i_mapping;
> + struct buffer_head *bh;
> + struct page *page;
> + const struct address_space_operations *a_ops = mapping->a_ops;
> + handle_t *handle;
> + pgoff_t offset_in_page = PAGE_SIZE;
> + int jblocks;
> + int ret = 0;
> + int blocksize = inode->i_sb->s_blocksize;
> + int blocks_per_page = 0;
> + int i = 0;
> + long long offs = org_offset << PAGE_CACHE_SHIFT;
> + unsigned long blk_off = 0;
> + unsigned int w_flags = 0;
> + void *fsdata;
> +
> + /*
> + * It needs twice the amount of ordinary journal buffers because
> + * inode and tmp_inode may change each different metadata blocks.
> + */
> + jblocks = ext4_writepage_trans_blocks(inode) * 2;
> + handle = ext4_journal_start(inode, jblocks);
> + if (IS_ERR(handle)) {
> + ret = PTR_ERR(handle);
> + return ret;
> + }
> +
> + if (segment_eq(get_fs(), KERNEL_DS))
> + w_flags |= AOP_FLAG_UNINTERRUPTIBLE;
> +
> + if (org_offset == ((inode->i_size - 1) >> PAGE_CACHE_SHIFT)) {
> + offset_in_page = (inode->i_size & (PAGE_CACHE_SIZE - 1));
> + /*
> + * Set PAGE_CACHE_SIZE to offset_in_page not be 0
> + * if org_offset is the last page and i_size is
> + * multiples of PAGE_CACHE_SIZE.
> + */
> + if (offset_in_page == 0)
> + offset_in_page = PAGE_CACHE_SIZE;
> + }
> +
> + up_write(&EXT4_I(inode)->i_data_sem);
> + ret = a_ops->write_begin(filp, mapping, offs,
> + offset_in_page, w_flags, &page, &fsdata);
> + down_write(&EXT4_I(inode)->i_data_sem);
> +
> + if (unlikely(ret < 0))
> + goto out;
> +
> + if (!PageUptodate(page)) {
> + mapping->a_ops->readpage(filp, page);
> + lock_page(page);
> + }
> +
> + /*
> + * try_to_release_page() doesn't call relasepage in writeback mode.
> + * We should care about the order of writing to the same file
> + * by multiple defrag processes.
> + * It needs to call wait_on_page_writeback() to wait for the
> + * writeback of the page.
> + */
> + if (PageWriteback(page))
> + wait_on_page_writeback(page);
> +
> + /* Release old bh and drop refs */
> + try_to_release_page(page, 0);
> + ret = ext4_defrag_replace_branches(handle, inode, tmp_inode,
> + org_offset, dest_offset, 1, flag);
> +
> + if (ret < 0)
> + goto out;
> +
> + /* Clear the inode cache not to refer to the old data */
> + ext4_ext_invalidate_cache(inode);
> +
> + if (!page_has_buffers(page))
> + create_empty_buffers(page, 1 << inode->i_blkbits, 0);
> +
> + blocks_per_page = PAGE_SIZE / blocksize;
> + blk_off = org_offset * blocks_per_page;
> +
> + bh = page_buffers(page);
> + for (i = 0; i < blocks_per_page; i++) {
> + up_write(&EXT4_I(inode)->i_data_sem);
> + ret = ext4_get_block(inode, blk_off++, bh, 0);
> + down_write(&EXT4_I(inode)->i_data_sem);
> +
> + if (ret < 0)
> + goto out;
> +
> + if (bh->b_this_page != NULL)
> + bh = bh->b_this_page;
> + }
> +
> + ret = a_ops->write_end(filp, mapping, offs, offset_in_page,
> + offset_in_page, page, fsdata);
> +
> + if (unlikely(ret < 0))
> + goto out;
> +out:
> + ext4_journal_stop(handle);
> +
> + return (ret < 0 ? ret : 0);
> +}
> +
> +/**
> * ext4_defrag_new_extent_tree - Allocate contiguous blocks
> *
> * @inode: inode of the original file
> diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
> index f8828ff..dd8dc46 100644
> --- a/fs/ext4/extents.c
> +++ b/fs/ext4/extents.c
> @@ -1171,7 +1171,7 @@ ext4_ext_search_right(struct inode *inode, struct ext4_ext_path *path,
> * allocated block. Thus, index entries have to be consistent
> * with leaves.
> */
> -static ext4_lblk_t
> +ext4_lblk_t
> ext4_ext_next_allocated_block(struct ext4_ext_path *path)
> {
> int depth;
> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index 0f252db..695877e 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -991,8 +991,7 @@ int ext4_get_blocks_wrap(handle_t *handle, struct inode *inode, sector_t block,
> up_write((&EXT4_I(inode)->i_data_sem));
> return retval;
> }
> -
> -static int ext4_get_block(struct inode *inode, sector_t iblock,
> +int ext4_get_block(struct inode *inode, sector_t iblock,
> struct buffer_head *bh_result, int create)
> {
> handle_t *handle = ext4_journal_current_handle();
>
>

2008-03-18 09:39:04

by Solofo.Ramangalahy

Subject: [RFC][PATCH 2/3] ext4 online defrag (ver 0.7)

Hello,

Akira Fujita writes:
> - printk(KERN_ERR "defrag: failed ext4_ext_defrag\n");
> - goto ERR;
> + printk(KERN_ERR "ext4 defrag: "
> + "Moving victim file failed. ino [%lu]\n",
> + ext_info->ino);

This triggers this warning:
fs/ext4/defrag.c: In function 'ext4_defrag_move_victim':
fs/ext4/defrag.c:458: warning: format '%lu' expects type 'long unsigned int', but argument 2 has type 'long long unsigned int'

(ext4-online-defrag-free-space-fragmentation.patch from ext4-patch-queue 36c86f09d6ac97c2797a1a94d76e2fc1dc3b03f8)

--
solofo

2008-03-19 20:56:05

by Akira Fujita

Subject: Re: [RFC][PATCH 2/3] ext4 online defrag (ver 0.7)

Hello Solofo,
> Hello,
>
> Akira Fujita writes:
> > - printk(KERN_ERR "defrag: failed ext4_ext_defrag\n");
> > - goto ERR;
> > + printk(KERN_ERR "ext4 defrag: "
> > + "Moving victim file failed. ino [%lu]\n",
> > + ext_info->ino);
>
> This triggers this warning:
> fs/ext4/defrag.c: In function 'ext4_defrag_move_victim':
> fs/ext4/defrag.c:458: warning: format '%lu' expects type 'long unsigned int', but argument 2 has type 'long long unsigned int'
>
> (ext4-online-defrag-free-space-fragmentation.patch from ext4-patch-queue 36c86f09d6ac97c2797a1a94d76e2fc1dc3b03f8)
>
Thank you for notification.
And sorry for the delay fix, it has already been pointed out by Mingming.

Mingming, could you replace attached file with ext4 patch queue's ?

Regards Akira,


Signed-off-by: Akira Fujita <[email protected]>
---
defrag.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
--- linux-2.6.25-rc5-ext4-org/fs/ext4/defrag.c 2008-03-19 07:16:40.000000000 +0900
+++ linux-2.6.25-rc5-full/fs/ext4/defrag.c 2008-03-19 08:31:01.000000000 +0900
@@ -454,7 +454,7 @@ static int ext4_defrag_move_victim(struc
ext_info->ext[i].len, goal, DEFRAG_FORCE_VICTIM, &ext);
if (ret < 0) {
printk(KERN_ERR "ext4 defrag: "
- "Moving victim file failed. ino [%lu]\n",
+ "Moving victim file failed. ino [%llu]\n",
ext_info->ino);
goto err;
}

Attachments:

2008-03-19 21:09:46

by Akira Fujita

Subject: Re: [RFC][PATCH 2/3] ext4 online defrag (ver 0.7)

Hello Solofo,
> Hello,
>
> Akira Fujita writes:
> > - printk(KERN_ERR "defrag: failed ext4_ext_defrag\n");
> > - goto ERR;
> > + printk(KERN_ERR "ext4 defrag: "
> > + "Moving victim file failed. ino [%lu]\n",
> > + ext_info->ino);
>
> This triggers this warning:
> fs/ext4/defrag.c: In function 'ext4_defrag_move_victim':
> fs/ext4/defrag.c:458: warning: format '%lu' expects type 'long unsigned int', but argument 2 has type 'long long unsigned int'
>
> (ext4-online-defrag-free-space-fragmentation.patch from ext4-patch-queue 36c86f09d6ac97c2797a1a94d76e2fc1dc3b03f8)
>
Thank you for notification.
And sorry for the delay fix, it has already been pointed out by Mingming.

Regards Akira,

Signed-off-by: Akira Fujita <[email protected]>
--- linux-2.6.25-rc5-ext4-org/fs/ext4/defrag.c 2008-03-19 07:16:40.000000000 +0900
+++ linux-2.6.25-rc5-full/fs/ext4/defrag.c 2008-03-19 08:31:01.000000000 +0900
@@ -454,7 +454,7 @@ static int ext4_defrag_move_victim(struc
ext_info->ext[i].len, goal, DEFRAG_FORCE_VICTIM, &ext);
if (ret < 0) {
printk(KERN_ERR "ext4 defrag: "
- "Moving victim file failed. ino [%lu]\n",
+ "Moving victim file failed. ino [%llu]\n",
ext_info->ino);
goto err;
}