Whamcloud - gitweb
resize2fs: use blk64_t and location getters for free_gdp_blocks()
authorDarrick J. Wong <darrick.wong@oracle.com>
Mon, 9 Sep 2013 14:40:36 +0000 (10:40 -0400)
committerTheodore Ts'o <tytso@mit.edu>
Mon, 9 Sep 2013 14:40:48 +0000 (10:40 -0400)
free_gdp_blocks needs to be taught to use 64-bit fields and the appropriate
getters, otherwise it'll truncate high block numbers (when, say, resizing a
>16T fs) and mark the low numbered group descriptor blocks as free.  Yikes.

Reported-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
resize/resize2fs.c

index 204b10a..3b48206 100644 (file)
@@ -273,34 +273,36 @@ static void fix_uninit_block_bitmaps(ext2_filsys fs)
  */
 static void free_gdp_blocks(ext2_filsys fs,
                            ext2fs_block_bitmap reserve_blocks,
-                           struct ext2_group_desc *gdp)
+                           ext2_filsys old_fs,
+                           dgrp_t group)
 {
-       blk_t   blk;
+       blk64_t blk;
        int     j;
 
-       if (gdp->bg_block_bitmap &&
-           (gdp->bg_block_bitmap < ext2fs_blocks_count(fs->super))) {
-               ext2fs_block_alloc_stats(fs, gdp->bg_block_bitmap, -1);
-               ext2fs_mark_block_bitmap2(reserve_blocks,
-                                        gdp->bg_block_bitmap);
+       blk = ext2fs_block_bitmap_loc(old_fs, group);
+       if (blk &&
+           (blk < ext2fs_blocks_count(fs->super))) {
+               ext2fs_block_alloc_stats2(fs, blk, -1);
+               ext2fs_mark_block_bitmap2(reserve_blocks, blk);
        }
 
-       if (gdp->bg_inode_bitmap &&
-           (gdp->bg_inode_bitmap < ext2fs_blocks_count(fs->super))) {
-               ext2fs_block_alloc_stats(fs, gdp->bg_inode_bitmap, -1);
-               ext2fs_mark_block_bitmap2(reserve_blocks,
-                                        gdp->bg_inode_bitmap);
+       blk = ext2fs_inode_bitmap_loc(old_fs, group);
+       if (blk &&
+           (blk < ext2fs_blocks_count(fs->super))) {
+               ext2fs_block_alloc_stats2(fs, blk, -1);
+               ext2fs_mark_block_bitmap2(reserve_blocks, blk);
        }
 
-       if (gdp->bg_inode_table == 0 ||
-           (gdp->bg_inode_table >= ext2fs_blocks_count(fs->super)))
+       blk = ext2fs_inode_table_loc(old_fs, group);
+       if (blk == 0 ||
+           (blk >= ext2fs_blocks_count(fs->super)))
                return;
 
-       for (blk = gdp->bg_inode_table, j = 0;
+       for (j = 0;
             j < fs->inode_blocks_per_group; j++, blk++) {
                if (blk >= ext2fs_blocks_count(fs->super))
                        break;
-               ext2fs_block_alloc_stats(fs, blk, -1);
+               ext2fs_block_alloc_stats2(fs, blk, -1);
                ext2fs_mark_block_bitmap2(reserve_blocks, blk);
        }
 }
@@ -467,11 +469,8 @@ retry:
                 * and free any blocks associated with their metadata
                 */
                for (i = fs->group_desc_count;
-                    i < old_fs->group_desc_count; i++) {
-                       free_gdp_blocks(fs, reserve_blocks,
-                                       ext2fs_group_desc(old_fs,
-                                               old_fs->group_desc, i));
-               }
+                    i < old_fs->group_desc_count; i++)
+                       free_gdp_blocks(fs, reserve_blocks, old_fs, i);
                retval = 0;
                goto errout;
        }