ext4: Fix sub-block zeroing for writes into preallocated extents
authorAneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Wed, 13 May 2009 22:36:58 +0000 (18:36 -0400)
committerTheodore Ts'o <tytso@mit.edu>
Wed, 13 May 2009 22:36:58 +0000 (18:36 -0400)
We need to mark the buffer_head mapping preallocated space as new
during write_begin. Otherwise we don't zero out the page cache content
properly for a partial write. This will cause file corruption with
preallocation.

Now that we mark the buffer_head new we also need to have a valid
buffer_head blocknr so that unmap_underlying_metadata() unmaps the
correct block.

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
fs/ext4/extents.c
fs/ext4/inode.c

index e4033215834074fe724987098aac17d36ab3ac85..172656c2a3bd591a91f701b535d530e2f9a73023 100644 (file)
@@ -2875,6 +2875,8 @@ int ext4_ext_get_blocks(handle_t *handle, struct inode *inode,
                                if (allocated > max_blocks)
                                        allocated = max_blocks;
                                set_buffer_unwritten(bh_result);
+                               bh_result->b_bdev = inode->i_sb->s_bdev;
+                               bh_result->b_blocknr = newblock;
                                goto out2;
                        }
 
index e91f978c7f12848616855876628189a44a6885f8..d4b634ae06bdce9aee783633a2bd816df54ebcc0 100644 (file)
@@ -2323,6 +2323,13 @@ static int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
                set_buffer_delay(bh_result);
        } else if (ret > 0) {
                bh_result->b_size = (ret << inode->i_blkbits);
+               /*
+                * With sub-block writes into unwritten extents
+                * we also need to mark the buffer as new so that
+                * the unwritten parts of the buffer gets correctly zeroed.
+                */
+               if (buffer_unwritten(bh_result))
+                       set_buffer_new(bh_result);
                ret = 0;
        }