ext4: simplify parameters of add_new_gdb()
authorYongqiang Yang <xiaoqiangnk@gmail.com>
Thu, 28 Jul 2011 01:16:33 +0000 (21:16 -0400)
committerTheodore Ts'o <tytso@mit.edu>
Thu, 28 Jul 2011 01:16:33 +0000 (21:16 -0400)
add_new_gdb() only needs the block group number; there is no need to
pass a pointer to struct ext4_new_group_data to add_new_gdb().
Instead of filling in a pointer the struct buffer_head in
add_new_gdb(), it's simpler to have the caller fetch it from the
s_group_desc[] array.

[Fixed error path to handle the case where struct buffer_head *primary
 hasn't been set yet. -- Ted]

Signed-off-by: Yongqiang Yang <xiaoqiangnk@gmail.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
fs/ext4/resize.c

index 65e5cb6c094d971838d3ed68b6abbf87e79cab9d..9e453552f10eb41169e3cecbebc9417bbaa019e4 100644 (file)
@@ -394,15 +394,15 @@ static int verify_reserved_gdb(struct super_block *sb,
  * fail once we start modifying the data on disk, because JBD has no rollback.
  */
 static int add_new_gdb(handle_t *handle, struct inode *inode,
-                      struct ext4_new_group_data *input,
-                      struct buffer_head **primary)
+                      ext4_group_t group)
 {
        struct super_block *sb = inode->i_sb;
        struct ext4_super_block *es = EXT4_SB(sb)->s_es;
-       unsigned long gdb_num = input->group / EXT4_DESC_PER_BLOCK(sb);
+       unsigned long gdb_num = group / EXT4_DESC_PER_BLOCK(sb);
        ext4_fsblk_t gdblock = EXT4_SB(sb)->s_sbh->b_blocknr + 1 + gdb_num;
        struct buffer_head **o_group_desc, **n_group_desc;
        struct buffer_head *dind;
+       struct buffer_head *gdb_bh;
        int gdbackups;
        struct ext4_iloc iloc;
        __le32 *data;
@@ -425,11 +425,12 @@ static int add_new_gdb(handle_t *handle, struct inode *inode,
                return -EPERM;
        }
 
-       *primary = sb_bread(sb, gdblock);
-       if (!*primary)
+       gdb_bh = sb_bread(sb, gdblock);
+       if (!gdb_bh)
                return -EIO;
 
-       if ((gdbackups = verify_reserved_gdb(sb, *primary)) < 0) {
+       gdbackups = verify_reserved_gdb(sb, gdb_bh);
+       if (gdbackups < 0) {
                err = gdbackups;
                goto exit_bh;
        }
@@ -444,7 +445,7 @@ static int add_new_gdb(handle_t *handle, struct inode *inode,
        data = (__le32 *)dind->b_data;
        if (le32_to_cpu(data[gdb_num % EXT4_ADDR_PER_BLOCK(sb)]) != gdblock) {
                ext4_warning(sb, "new group %u GDT block %llu not reserved",
-                            input->group, gdblock);
+                            group, gdblock);
                err = -EINVAL;
                goto exit_dind;
        }
@@ -453,7 +454,7 @@ static int add_new_gdb(handle_t *handle, struct inode *inode,
        if (unlikely(err))
                goto exit_dind;
 
-       err = ext4_journal_get_write_access(handle, *primary);
+       err = ext4_journal_get_write_access(handle, gdb_bh);
        if (unlikely(err))
                goto exit_sbh;
 
@@ -492,8 +493,8 @@ static int add_new_gdb(handle_t *handle, struct inode *inode,
        }
        inode->i_blocks -= (gdbackups + 1) * sb->s_blocksize >> 9;
        ext4_mark_iloc_dirty(handle, inode, &iloc);
-       memset((*primary)->b_data, 0, sb->s_blocksize);
-       err = ext4_handle_dirty_metadata(handle, NULL, *primary);
+       memset(gdb_bh->b_data, 0, sb->s_blocksize);
+       err = ext4_handle_dirty_metadata(handle, NULL, gdb_bh);
        if (unlikely(err)) {
                ext4_std_error(sb, err);
                goto exit_inode;
@@ -503,7 +504,7 @@ static int add_new_gdb(handle_t *handle, struct inode *inode,
        o_group_desc = EXT4_SB(sb)->s_group_desc;
        memcpy(n_group_desc, o_group_desc,
               EXT4_SB(sb)->s_gdb_count * sizeof(struct buffer_head *));
-       n_group_desc[gdb_num] = *primary;
+       n_group_desc[gdb_num] = gdb_bh;
        EXT4_SB(sb)->s_group_desc = n_group_desc;
        EXT4_SB(sb)->s_gdb_count++;
        kfree(o_group_desc);
@@ -525,7 +526,7 @@ exit_sbh:
 exit_dind:
        brelse(dind);
 exit_bh:
-       brelse(*primary);
+       brelse(gdb_bh);
 
        ext4_debug("leaving with error %d\n", err);
        return err;
@@ -833,8 +834,16 @@ int ext4_group_add(struct super_block *sb, struct ext4_new_group_data *input)
                if (reserved_gdb && ext4_bg_num_gdb(sb, input->group) &&
                    (err = reserve_backup_gdb(handle, inode, input)))
                        goto exit_journal;
-       } else if ((err = add_new_gdb(handle, inode, input, &primary)))
-               goto exit_journal;
+       } else {
+               /*
+                * Note that we can access new group descriptor block safely
+                * only if add_new_gdb() succeeds.
+                */
+               err = add_new_gdb(handle, inode, input->group);
+               if (err)
+                       goto exit_journal;
+               primary = sbi->s_group_desc[gdb_num];
+       }
 
         /*
          * OK, now we've set up the new group.  Time to make it active.
@@ -944,7 +953,7 @@ int ext4_group_add(struct super_block *sb, struct ext4_new_group_data *input)
 exit_journal:
        if ((err2 = ext4_journal_stop(handle)) && !err)
                err = err2;
-       if (!err) {
+       if (!err && primary) {
                update_backups(sb, sbi->s_sbh->b_blocknr, (char *)es,
                               sizeof(struct ext4_super_block));
                update_backups(sb, primary->b_blocknr, primary->b_data,