ext4: online defrag -- Add the EXT4_IOC_FREE_BLOCKS_INFO ioctl.
From: Akira Fujita <a-fujita@xxxxxxxxxxxxx>
The EXT4_IOC_FREE_BLOCKS_INFO ioctl gets free extents
information of the target block group.
This ioctl is used only in the force defrag (-f).
Defragger calculates the victim extents (which will be moved
into other block group in order to make contiguous free space for the target file)
based on used/unused extents information.
ext4_defrag_fblocks_distribution() was renamed as ext4_mb_search_free_extents()
and moved into mballoc.c. Because it uses ext4_buddy structure.
Signed-off-by: Akira Fujita <a-fujita@xxxxxxxxxxxxx>
Signed-off-by: Takashi Sato <t-sato@xxxxxxxxxxxxx>
---
fs/ext4/ext4.h | 27 +++++++++++++++++
fs/ext4/ioctl.c | 20 +++++++++++++
fs/ext4/mballoc.c | 81 +++++++++++++++++++++++++++++++++++++++++++++++++++++
3 files changed, 128 insertions(+), 0 deletions(-)
diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index 94c4186..f7b092d 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -304,6 +304,7 @@ struct ext4_new_group_data {
/* note ioctl 11 reserved for filesystem-independent FIEMAP ioctl */
#define EXT4_IOC_DEFRAG _IOW('f', 15, struct ext4_ext_defrag_data)
#define EXT4_IOC_SUPER_INFO _IOR('f', 16, struct ext4_super_block)
+#define EXT4_IOC_FREE_BLOCKS_INFO _IOWR('f', 17, struct ext4_extents_info)
/*
* ioctl commands in 32 bit emulation
@@ -321,12 +322,36 @@ struct ext4_new_group_data {
#define EXT4_IOC32_GETVERSION_OLD FS_IOC32_GETVERSION
#define EXT4_IOC32_SETVERSION_OLD FS_IOC32_SETVERSION
+/*
+ * The following four macros are used for the defrag force mode.
+ *
+ * DEFRAG_MAX_ENT: the maximum number of extents for exchanging between
+ * kernel-space and user-space per an ioctl
+ */
+#define DEFRAG_MAX_ENT 32
+
+struct ext4_extent_data {
+ ext4_lblk_t block; /* start logical block number */
+ ext4_fsblk_t start; /* start physical block number */
+ int len; /* blocks count */
+};
+
struct ext4_ext_defrag_data {
ext4_lblk_t start_offset; /* start offset to defrag in blocks */
ext4_lblk_t defrag_size; /* size of defrag in blocks */
ext4_fsblk_t goal; /* block offset for allocation */
};
+struct ext4_extents_info {
+ unsigned long long ino; /* inode number */
+ int max_entries; /* maximum extents count */
+ int entries; /* extent number/count */
+ ext4_lblk_t f_offset; /* file offset */
+ ext4_grpblk_t g_offset; /* group offset */
+ ext4_fsblk_t goal; /* block offset for allocation */
+ struct ext4_extent_data ext[DEFRAG_MAX_ENT];
+};
+
#define EXT4_TRANS_META_BLOCKS 4 /* bitmap + group desc + sb + inode */
/*
@@ -1069,6 +1094,8 @@ extern int ext4_mb_add_more_groupinfo(struct super_block *sb,
ext4_group_t i, struct ext4_group_desc *desc);
extern void ext4_mb_update_group_info(struct ext4_group_info *grp,
ext4_grpblk_t add);
+extern int ext4_mb_search_free_extents(struct inode *inode,
+ struct ext4_extents_info *ext_info);
/* inode.c */
diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c
index f8ce89b..5709574 100644
--- a/fs/ext4/ioctl.c
+++ b/fs/ext4/ioctl.c
@@ -256,6 +256,26 @@ setversion_out:
return 0;
}
+ case EXT4_IOC_FREE_BLOCKS_INFO: {
+ struct ext4_extents_info ext_info;
+ int err;
+
+ if (copy_from_user(&ext_info,
+ (struct ext4_extents_info __user *)arg,
+ sizeof(ext_info)))
+ return -EFAULT;
+
+ BUG_ON(ext_info.ino != inode->i_ino);
+
+ err = ext4_mb_search_free_extents(inode, &ext_info);
+
+ if (!err)
+ err = copy_to_user(
+ (struct ext4_extents_info __user *)arg,
+ &ext_info, sizeof(ext_info));
+ return err;
+ }
+
case EXT4_IOC_GROUP_ADD: {
struct ext4_new_group_data input;
struct super_block *sb = inode->i_sb;
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index dfe17a1..eebcf7c 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -4654,3 +4654,84 @@ error_return:
kmem_cache_free(ext4_ac_cachep, ac);
return;
}
+
+/**
+ * ext4_mb_search_free_extents
+ * - Search free extents in the block group where target inode is located
+ *
+ * @inode: target inode
+ * @ext_info: ext4_extents_info is used to store free extents
+ *
+ * This function returns 0 if succeed, otherwise returns error value.
+ */
+int
+ext4_mb_search_free_extents(struct inode *inode,
+ struct ext4_extents_info *ext_info)
+{
+ struct super_block *sb = inode->i_sb;
+ struct ext4_buddy e4b;
+ ext4_group_t group_no;
+ ext4_grpblk_t start, end;
+ ext4_fsblk_t start_block = 0;
+ int i, err;
+ int num = 0;
+ int len = 0;
+ int block_set = 0;
+ int extra_block = 0;
+
+ if (!sb) {
+ printk(KERN_ERR "ext4 defrag: Non-existent device\n");
+ return -ENOSPC;
+ }
+
+ group_no = (inode->i_ino - 1) / EXT4_INODES_PER_GROUP(sb);
+ start = ext_info->g_offset;
+ end = EXT4_BLOCKS_PER_GROUP(sb) - 1;
+
+ err = ext4_mb_load_buddy(sb, group_no, &e4b);
+ if (err)
+ return err;
+
+ /* We consider about the boot block if bs = 1k */
+ if (sb->s_blocksize == 1024)
+ extra_block = 1;
+
+ ext4_lock_group(sb, group_no);
+
+ for (i = start; i <= end ; i++) {
+ if (!mb_test_bit(i, e4b.bd_bitmap)) {
+ len++;
+ /*
+ * 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) +
+ extra_block;
+ block_set = 1;
+ }
+ } else if (len) {
+ ext_info->ext[num].start = start_block;
+ ext_info->ext[num].len = len;
+ num++;
+ len = 0;
+ block_set = 0;
+ if (num == ext_info->max_entries) {
+ ext_info->g_offset = i + 1;
+ break;
+ }
+ }
+ if (i == end && len) {
+ ext_info->ext[num].start = start_block;
+ ext_info->ext[num].len = len;
+ num++;
+ }
+ }
+ ext_info->entries = num;
+
+ ext4_unlock_group(sb, group_no);
+ ext4_mb_release_desc(&e4b);
+
+ return err;
+}
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
