]> pilppa.com Git - linux-2.6-omap-h63xx.git/commitdiff
ext4: Add EXT4_IOC_MIGRATE ioctl
authorAneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Tue, 29 Jan 2008 04:58:26 +0000 (23:58 -0500)
committerTheodore Ts'o <tytso@mit.edu>
Tue, 29 Jan 2008 04:58:26 +0000 (23:58 -0500)
The below patch add ioctl for migrating ext3 indirect block mapped inode
to ext4 extent mapped inode.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
fs/ext4/Makefile
fs/ext4/extents.c
fs/ext4/ioctl.c
fs/ext4/migrate.c [new file with mode: 0644]
include/linux/ext4_fs.h
include/linux/ext4_fs_extents.h

index ae6e7e502ac9c0d5585370ba0c051410ab766ad8..d5fd80bc0d04d9b6496b649c1e37a9ee7f140f09 100644 (file)
@@ -6,7 +6,7 @@ obj-$(CONFIG_EXT4DEV_FS) += ext4dev.o
 
 ext4dev-y      := balloc.o bitmap.o dir.o file.o fsync.o ialloc.o inode.o \
                   ioctl.o namei.o super.o symlink.o hash.o resize.o extents.o \
-                  ext4_jbd2.o
+                  ext4_jbd2.o migrate.o
 
 ext4dev-$(CONFIG_EXT4DEV_FS_XATTR)     += xattr.o xattr_user.o xattr_trusted.o
 ext4dev-$(CONFIG_EXT4DEV_FS_POSIX_ACL) += acl.o
index 03d1bbb78a2ffac7f86016db6769578f6e107fe4..01eda5c5281e38c0f5a69af66037d52566ce4f04 100644 (file)
@@ -61,7 +61,7 @@ static ext4_fsblk_t ext_pblock(struct ext4_extent *ex)
  * idx_pblock:
  * combine low and high parts of a leaf physical block number into ext4_fsblk_t
  */
-static ext4_fsblk_t idx_pblock(struct ext4_extent_idx *ix)
+ext4_fsblk_t idx_pblock(struct ext4_extent_idx *ix)
 {
        ext4_fsblk_t block;
 
@@ -75,7 +75,7 @@ static ext4_fsblk_t idx_pblock(struct ext4_extent_idx *ix)
  * stores a large physical block number into an extent struct,
  * breaking it into parts
  */
-static void ext4_ext_store_pblock(struct ext4_extent *ex, ext4_fsblk_t pb)
+void ext4_ext_store_pblock(struct ext4_extent *ex, ext4_fsblk_t pb)
 {
        ex->ee_start_lo = cpu_to_le32((unsigned long) (pb & 0xffffffff));
        ex->ee_start_hi = cpu_to_le16((unsigned long) ((pb >> 31) >> 1) & 0xffff);
index c0e5b8cf635c65aec9db57e0163308ae4808ed29..2ed7c37f897e79f0b9c6d6af3a5c2aad79fa7f59 100644 (file)
@@ -254,6 +254,9 @@ flags_err:
                return err;
        }
 
+       case EXT4_IOC_MIGRATE:
+               return ext4_ext_migrate(inode, filp, cmd, arg);
+
        default:
                return -ENOTTY;
        }
diff --git a/fs/ext4/migrate.c b/fs/ext4/migrate.c
new file mode 100644 (file)
index 0000000..ec7cb56
--- /dev/null
@@ -0,0 +1,560 @@
+/*
+ * Copyright IBM Corporation, 2007
+ * Author Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of version 2.1 of the GNU Lesser General Public License
+ * as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ */
+
+#include <linux/module.h>
+#include <linux/ext4_jbd2.h>
+#include <linux/ext4_fs_extents.h>
+
+/*
+ * The contiguous blocks details which can be
+ * represented by a single extent
+ */
+struct list_blocks_struct {
+       ext4_lblk_t first_block, last_block;
+       ext4_fsblk_t first_pblock, last_pblock;
+};
+
+static int finish_range(handle_t *handle, struct inode *inode,
+                               struct list_blocks_struct *lb)
+
+{
+       int retval = 0, needed;
+       struct ext4_extent newext;
+       struct ext4_ext_path *path;
+       if (lb->first_pblock == 0)
+               return 0;
+
+       /* Add the extent to temp inode*/
+       newext.ee_block = cpu_to_le32(lb->first_block);
+       newext.ee_len   = cpu_to_le16(lb->last_block - lb->first_block + 1);
+       ext4_ext_store_pblock(&newext, lb->first_pblock);
+       path = ext4_ext_find_extent(inode, lb->first_block, NULL);
+
+       if (IS_ERR(path)) {
+               retval = PTR_ERR(path);
+               goto err_out;
+       }
+
+       /*
+        * Calculate the credit needed to inserting this extent
+        * Since we are doing this in loop we may accumalate extra
+        * credit. But below we try to not accumalate too much
+        * of them by restarting the journal.
+        */
+       needed = ext4_ext_calc_credits_for_insert(inode, path);
+
+       /*
+        * Make sure the credit we accumalated is not really high
+        */
+       if (needed && handle->h_buffer_credits >= EXT4_RESERVE_TRANS_BLOCKS) {
+               retval = ext4_journal_restart(handle, needed);
+               if (retval)
+                       goto err_out;
+       }
+       if (needed) {
+               retval = ext4_journal_extend(handle, needed);
+               if (retval != 0) {
+                       /*
+                        * IF not able to extend the journal restart the journal
+                        */
+                       retval = ext4_journal_restart(handle, needed);
+                       if (retval)
+                               goto err_out;
+               }
+       }
+       retval = ext4_ext_insert_extent(handle, inode, path, &newext);
+err_out:
+       lb->first_pblock = 0;
+       return retval;
+}
+
+static int update_extent_range(handle_t *handle, struct inode *inode,
+                               ext4_fsblk_t pblock, ext4_lblk_t blk_num,
+                               struct list_blocks_struct *lb)
+{
+       int retval;
+       /*
+        * See if we can add on to the existing range (if it exists)
+        */
+       if (lb->first_pblock &&
+               (lb->last_pblock+1 == pblock) &&
+               (lb->last_block+1 == blk_num)) {
+               lb->last_pblock = pblock;
+               lb->last_block = blk_num;
+               return 0;
+       }
+       /*
+        * Start a new range.
+        */
+       retval = finish_range(handle, inode, lb);
+       lb->first_pblock = lb->last_pblock = pblock;
+       lb->first_block = lb->last_block = blk_num;
+
+       return retval;
+}
+
+static int update_ind_extent_range(handle_t *handle, struct inode *inode,
+                                  ext4_fsblk_t pblock, ext4_lblk_t *blk_nump,
+                                  struct list_blocks_struct *lb)
+{
+       struct buffer_head *bh;
+       __le32 *i_data;
+       int i, retval = 0;
+       ext4_lblk_t blk_count = *blk_nump;
+       unsigned long max_entries = inode->i_sb->s_blocksize >> 2;
+
+       if (!pblock) {
+               /* Only update the file block number */
+               *blk_nump += max_entries;
+               return 0;
+       }
+
+       bh = sb_bread(inode->i_sb, pblock);
+       if (!bh)
+               return -EIO;
+
+       i_data = (__le32 *)bh->b_data;
+       for (i = 0; i < max_entries; i++, blk_count++) {
+               if (i_data[i]) {
+                       retval = update_extent_range(handle, inode,
+                                               le32_to_cpu(i_data[i]),
+                                               blk_count, lb);
+                       if (retval)
+                               break;
+               }
+       }
+
+       /* Update the file block number */
+       *blk_nump = blk_count;
+       put_bh(bh);
+       return retval;
+
+}
+
+static int update_dind_extent_range(handle_t *handle, struct inode *inode,
+                                   ext4_fsblk_t pblock, ext4_lblk_t *blk_nump,
+                                   struct list_blocks_struct *lb)
+{
+       struct buffer_head *bh;
+       __le32 *i_data;
+       int i, retval = 0;
+       ext4_lblk_t blk_count = *blk_nump;
+       unsigned long max_entries = inode->i_sb->s_blocksize >> 2;
+
+       if (!pblock) {
+               /* Only update the file block number */
+               *blk_nump += max_entries * max_entries;
+               return 0;
+       }
+       bh = sb_bread(inode->i_sb, pblock);
+       if (!bh)
+               return -EIO;
+
+       i_data = (__le32 *)bh->b_data;
+       for (i = 0; i < max_entries; i++) {
+               if (i_data[i]) {
+                       retval = update_ind_extent_range(handle, inode,
+                                               le32_to_cpu(i_data[i]),
+                                               &blk_count, lb);
+                       if (retval)
+                               break;
+               } else {
+                       /* Only update the file block number */
+                       blk_count += max_entries;
+               }
+       }
+
+       /* Update the file block number */
+       *blk_nump = blk_count;
+       put_bh(bh);
+       return retval;
+
+}
+
+static int update_tind_extent_range(handle_t *handle, struct inode *inode,
+                                    ext4_fsblk_t pblock, ext4_lblk_t *blk_nump,
+                                    struct list_blocks_struct *lb)
+{
+       struct buffer_head *bh;
+       __le32 *i_data;
+       int i, retval = 0;
+       ext4_lblk_t blk_count = *blk_nump;
+       unsigned long max_entries = inode->i_sb->s_blocksize >> 2;
+
+       if (!pblock) {
+               /* Only update the file block number */
+               *blk_nump += max_entries * max_entries * max_entries;
+               return 0;
+       }
+       bh = sb_bread(inode->i_sb, pblock);
+       if (!bh)
+               return -EIO;
+
+       i_data = (__le32 *)bh->b_data;
+       for (i = 0; i < max_entries; i++) {
+               if (i_data[i]) {
+                       retval = update_dind_extent_range(handle, inode,
+                                               le32_to_cpu(i_data[i]),
+                                               &blk_count, lb);
+                       if (retval)
+                               break;
+               } else
+                       /* Only update the file block number */
+                       blk_count += max_entries * max_entries;
+       }
+       /* Update the file block number */
+       *blk_nump = blk_count;
+       put_bh(bh);
+       return retval;
+
+}
+
+static int free_dind_blocks(handle_t *handle,
+                               struct inode *inode, __le32 i_data)
+{
+       int i;
+       __le32 *tmp_idata;
+       struct buffer_head *bh;
+       unsigned long max_entries = inode->i_sb->s_blocksize >> 2;
+
+       bh = sb_bread(inode->i_sb, le32_to_cpu(i_data));
+       if (!bh)
+               return -EIO;
+
+       tmp_idata = (__le32 *)bh->b_data;
+       for (i = 0; i < max_entries; i++) {
+               if (tmp_idata[i])
+                       ext4_free_blocks(handle, inode,
+                                       le32_to_cpu(tmp_idata[i]), 1);
+       }
+       put_bh(bh);
+       ext4_free_blocks(handle, inode, le32_to_cpu(i_data), 1);
+       return 0;
+}
+
+static int free_tind_blocks(handle_t *handle,
+                               struct inode *inode, __le32 i_data)
+{
+       int i, retval = 0;
+       __le32 *tmp_idata;
+       struct buffer_head *bh;
+       unsigned long max_entries = inode->i_sb->s_blocksize >> 2;
+
+       bh = sb_bread(inode->i_sb, le32_to_cpu(i_data));
+       if (!bh)
+               return -EIO;
+
+       tmp_idata = (__le32 *)bh->b_data;
+       for (i = 0; i < max_entries; i++) {
+               if (tmp_idata[i]) {
+                       retval = free_dind_blocks(handle,
+                                       inode, tmp_idata[i]);
+                       if (retval) {
+                               put_bh(bh);
+                               return retval;
+                       }
+               }
+       }
+       put_bh(bh);
+       ext4_free_blocks(handle, inode, le32_to_cpu(i_data), 1);
+       return 0;
+}
+
+static int free_ind_block(handle_t *handle, struct inode *inode)
+{
+       int retval;
+       struct ext4_inode_info *ei = EXT4_I(inode);
+
+       if (ei->i_data[EXT4_IND_BLOCK])
+               ext4_free_blocks(handle, inode,
+                               le32_to_cpu(ei->i_data[EXT4_IND_BLOCK]), 1);
+
+       if (ei->i_data[EXT4_DIND_BLOCK]) {
+               retval = free_dind_blocks(handle, inode,
+                                               ei->i_data[EXT4_DIND_BLOCK]);
+               if (retval)
+                       return retval;
+       }
+
+       if (ei->i_data[EXT4_TIND_BLOCK]) {
+               retval = free_tind_blocks(handle, inode,
+                                               ei->i_data[EXT4_TIND_BLOCK]);
+               if (retval)
+                       return retval;
+       }
+       return 0;
+}
+
+static int ext4_ext_swap_inode_data(handle_t *handle, struct inode *inode,
+                               struct inode *tmp_inode, int retval)
+{
+       struct ext4_inode_info *ei = EXT4_I(inode);
+       struct ext4_inode_info *tmp_ei = EXT4_I(tmp_inode);
+
+       retval = free_ind_block(handle, inode);
+       if (retval)
+               goto err_out;
+
+       /*
+        * One credit accounted for writing the
+        * i_data field of the original inode
+        */
+       retval = ext4_journal_extend(handle, 1);
+       if (retval != 0) {
+               retval = ext4_journal_restart(handle, 1);
+               if (retval)
+                       goto err_out;
+       }
+
+       /*
+        * We have the extent map build with the tmp inode.
+        * Now copy the i_data across
+        */
+       ei->i_flags |= EXT4_EXTENTS_FL;
+       memcpy(ei->i_data, tmp_ei->i_data, sizeof(ei->i_data));
+
+       /*
+        * Update i_blocks with the new blocks that got
+        * allocated while adding extents for extent index
+        * blocks.
+        *
+        * While converting to extents we need not
+        * update the orignal inode i_blocks for extent blocks
+        * via quota APIs. The quota update happened via tmp_inode already.
+        */
+       spin_lock(&inode->i_lock);
+       inode->i_blocks += tmp_inode->i_blocks;
+       spin_unlock(&inode->i_lock);
+
+       ext4_mark_inode_dirty(handle, inode);
+err_out:
+       return retval;
+}
+
+static int free_ext_idx(handle_t *handle, struct inode *inode,
+                                       struct ext4_extent_idx *ix)
+{
+       int i, retval = 0;
+       ext4_fsblk_t block;
+       struct buffer_head *bh;
+       struct ext4_extent_header *eh;
+
+       block = idx_pblock(ix);
+       bh = sb_bread(inode->i_sb, block);
+       if (!bh)
+               return -EIO;
+
+       eh = (struct ext4_extent_header *)bh->b_data;
+       if (eh->eh_depth != 0) {
+               ix = EXT_FIRST_INDEX(eh);
+               for (i = 0; i < le16_to_cpu(eh->eh_entries); i++, ix++) {
+                       retval = free_ext_idx(handle, inode, ix);
+                       if (retval)
+                               break;
+               }
+       }
+       put_bh(bh);
+       ext4_free_blocks(handle, inode, block, 1);
+       return retval;
+}
+
+/*
+ * Free the extent meta data blocks only
+ */
+static int free_ext_block(handle_t *handle, struct inode *inode)
+{
+       int i, retval = 0;
+       struct ext4_inode_info *ei = EXT4_I(inode);
+       struct ext4_extent_header *eh = (struct ext4_extent_header *)ei->i_data;
+       struct ext4_extent_idx *ix;
+       if (eh->eh_depth == 0)
+               /*
+                * No extra blocks allocated for extent meta data
+                */
+               return 0;
+       ix = EXT_FIRST_INDEX(eh);
+       for (i = 0; i < le16_to_cpu(eh->eh_entries); i++, ix++) {
+               retval = free_ext_idx(handle, inode, ix);
+               if (retval)
+                       return retval;
+       }
+       return retval;
+
+}
+
+int ext4_ext_migrate(struct inode *inode, struct file *filp,
+                               unsigned int cmd, unsigned long arg)
+{
+       handle_t *handle;
+       int retval = 0, i;
+       __le32 *i_data;
+       ext4_lblk_t blk_count = 0;
+       struct ext4_inode_info *ei;
+       struct inode *tmp_inode = NULL;
+       struct list_blocks_struct lb;
+       unsigned long max_entries;
+
+       if (!test_opt(inode->i_sb, EXTENTS))
+               /*
+                * if mounted with noextents we don't allow the migrate
+                */
+               return -EINVAL;
+
+       if ((EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL))
+               return -EINVAL;
+
+       down_write(&EXT4_I(inode)->i_data_sem);
+       handle = ext4_journal_start(inode,
+                                       EXT4_DATA_TRANS_BLOCKS(inode->i_sb) +
+                                       EXT4_INDEX_EXTRA_TRANS_BLOCKS + 3 +
+                                       2 * EXT4_QUOTA_INIT_BLOCKS(inode->i_sb)
+                                       + 1);
+       if (IS_ERR(handle)) {
+               retval = PTR_ERR(handle);
+               goto err_out;
+       }
+       tmp_inode = ext4_new_inode(handle,
+                               inode->i_sb->s_root->d_inode,
+                               S_IFREG);
+       if (IS_ERR(tmp_inode)) {
+               retval = -ENOMEM;
+               ext4_journal_stop(handle);
+               tmp_inode = NULL;
+               goto err_out;
+       }
+       i_size_write(tmp_inode, i_size_read(inode));
+       /*
+        * We don't want the inode to be reclaimed
+        * if we got interrupted in between. We have
+        * this tmp inode carrying reference to the
+        * data blocks of the original file. We set
+        * the i_nlink to zero at the last stage after
+        * switching the original file to extent format
+        */
+       tmp_inode->i_nlink = 1;
+
+       ext4_ext_tree_init(handle, tmp_inode);
+       ext4_orphan_add(handle, tmp_inode);
+       ext4_journal_stop(handle);
+
+       ei = EXT4_I(inode);
+       i_data = ei->i_data;
+       memset(&lb, 0, sizeof(lb));
+
+       /* 32 bit block address 4 bytes */
+       max_entries = inode->i_sb->s_blocksize >> 2;
+
+       /*
+        * start with one credit accounted for
+        * superblock modification.
+        *
+        * For the tmp_inode we already have commited the
+        * trascation that created the inode. Later as and
+        * when we add extents we extent the journal
+        */
+       handle = ext4_journal_start(inode, 1);
+       for (i = 0; i < EXT4_NDIR_BLOCKS; i++, blk_count++) {
+               if (i_data[i]) {
+                       retval = update_extent_range(handle, tmp_inode,
+                                               le32_to_cpu(i_data[i]),
+                                               blk_count, &lb);
+                       if (retval)
+                               goto err_out;
+               }
+       }
+       if (i_data[EXT4_IND_BLOCK]) {
+               retval = update_ind_extent_range(handle, tmp_inode,
+                                       le32_to_cpu(i_data[EXT4_IND_BLOCK]),
+                                       &blk_count, &lb);
+                       if (retval)
+                               goto err_out;
+       } else
+               blk_count +=  max_entries;
+       if (i_data[EXT4_DIND_BLOCK]) {
+               retval = update_dind_extent_range(handle, tmp_inode,
+                                       le32_to_cpu(i_data[EXT4_DIND_BLOCK]),
+                                       &blk_count, &lb);
+                       if (retval)
+                               goto err_out;
+       } else
+               blk_count += max_entries * max_entries;
+       if (i_data[EXT4_TIND_BLOCK]) {
+               retval = update_tind_extent_range(handle, tmp_inode,
+                                       le32_to_cpu(i_data[EXT4_TIND_BLOCK]),
+                                       &blk_count, &lb);
+                       if (retval)
+                               goto err_out;
+       }
+       /*
+        * Build the last extent
+        */
+       retval = finish_range(handle, tmp_inode, &lb);
+err_out:
+       /*
+        * We are either freeing extent information or indirect
+        * blocks. During this we touch superblock, group descriptor
+        * and block bitmap. Later we mark the tmp_inode dirty
+        * via ext4_ext_tree_init. So allocate a credit of 4
+        * We may update quota (user and group).
+        *
+        * FIXME!! we may be touching bitmaps in different block groups.
+        */
+       if (ext4_journal_extend(handle,
+                       4 + 2*EXT4_QUOTA_TRANS_BLOCKS(inode->i_sb)) != 0)
+               ext4_journal_restart(handle,
+                               4 + 2*EXT4_QUOTA_TRANS_BLOCKS(inode->i_sb));
+       if (retval)
+               /*
+                * Failure case delete the extent information with the
+                * tmp_inode
+                */
+               free_ext_block(handle, tmp_inode);
+       else
+               retval = ext4_ext_swap_inode_data(handle, inode,
+                                                       tmp_inode, retval);
+
+       /*
+        * Mark the tmp_inode as of size zero
+        */
+       i_size_write(tmp_inode, 0);
+
+       /*
+        * set the  i_blocks count to zero
+        * so that the ext4_delete_inode does the
+        * right job
+        *
+        * We don't need to take the i_lock because
+        * the inode is not visible to user space.
+        */
+       tmp_inode->i_blocks = 0;
+
+       /* Reset the extent details */
+       ext4_ext_tree_init(handle, tmp_inode);
+
+       /*
+        * Set the i_nlink to zero so that
+        * generic_drop_inode really deletes the
+        * inode
+        */
+       tmp_inode->i_nlink = 0;
+
+       ext4_journal_stop(handle);
+
+       up_write(&EXT4_I(inode)->i_data_sem);
+
+       if (tmp_inode)
+               iput(tmp_inode);
+
+       return retval;
+}
index b6092940c8b84d21b676e4dac8302d95bbead62a..213974f7c5d8c61c42d41ba84084d055284900c8 100644 (file)
@@ -243,6 +243,7 @@ struct ext4_new_group_data {
 #endif
 #define EXT4_IOC_GETRSVSZ              _IOR('f', 5, long)
 #define EXT4_IOC_SETRSVSZ              _IOW('f', 6, long)
+#define EXT4_IOC_MIGRATE               _IO('f', 7)
 
 /*
  * ioctl commands in 32 bit emulation
@@ -983,6 +984,9 @@ extern int ext4_ioctl (struct inode *, struct file *, unsigned int,
                       unsigned long);
 extern long ext4_compat_ioctl (struct file *, unsigned int, unsigned long);
 
+/* migrate.c */
+extern int ext4_ext_migrate(struct inode *, struct file *, unsigned int,
+                      unsigned long);
 /* namei.c */
 extern int ext4_orphan_add(handle_t *, struct inode *);
 extern int ext4_orphan_del(handle_t *, struct inode *);
index 023683b9d8832d79ffee115abb0a5356e5d73535..1cfb4854c964ef2673c229ff5a79202409e5aafa 100644 (file)
@@ -212,6 +212,8 @@ static inline int ext4_ext_get_actual_len(struct ext4_extent *ext)
                (le16_to_cpu(ext->ee_len) - EXT_INIT_MAX_LEN));
 }
 
+extern ext4_fsblk_t idx_pblock(struct ext4_extent_idx *);
+extern void ext4_ext_store_pblock(struct ext4_extent *, ext4_fsblk_t);
 extern int ext4_extent_tree_init(handle_t *, struct inode *);
 extern int ext4_ext_calc_credits_for_insert(struct inode *, struct ext4_ext_path *);
 extern int ext4_ext_try_to_merge(struct inode *inode,