[Ocfs2-tools-devel] [PATCH 4/6] Fix filesize in create_file

[Goldwyn Rodrigues]

On Wed, Oct 26, 2011 at 2:21 AM, Sunil Mushran <sunil.mushran at oracle.com> wrote:
> I am confused. create_file() merely creates a file and adds it
> to a directory. It dies not allocate any space to it. The file size
> should be zero.

Hmm, I lost my previous work. However, AFAIR create_file used to
create a file with size as zero and with exactly one extent. I am
unable to reproduce it now.

Using the latest ocfs2-tools, fswreck is failing much before this
(local,noinline-data):

goldwyn at shrek:~/repos/ocfs2-tools> sudo ./fswreck/fswreck -c
INODE_SIZE /dev/sda10
fswreck 1.8.0
fswreck: Directory entry not found (corrupt.c,42)

This needs more time and effort and I will check after I get back from
my vacation.

>
> On 08/22/2011 03:36 PM, Goldwyn Rodrigues wrote:
>>
>> While creating a new file, the filesize is left at zero, even though
>> there is an extent in the file.
>> This fix is required because further extents added using
>> ocfs2_extend_allocation compute the offset based on current filesize.
>>
>> Signed-off-by: Goldwyn Rodrigues<rgoldwyn at suse.de>
>> ---
>>  fswreck/extent.c |   18 ++++++++++++++++++
>>  1 files changed, 18 insertions(+), 0 deletions(-)
>>
>> diff --git a/fswreck/extent.c b/fswreck/extent.c
>> index e36deb9..f6fade0 100644
>> --- a/fswreck/extent.c
>> +++ b/fswreck/extent.c
>> @@ -44,6 +44,12 @@ void create_file(ocfs2_filesys *fs, uint64_t blkno,
>> uint64_t *retblkno)
>>        errcode_t ret;
>>        uint64_t tmp_blkno = 0;
>>        char random_name[OCFS2_MAX_FILENAME_LEN];
>> +       char *buf;
>> +       struct ocfs2_dinode *di;
>> +
>> +       ret = ocfs2_malloc_block(fs->fs_io,&buf);
>> +       if (ret)
>> +               FSWRK_COM_FATAL(progname, errno);
>>
>>        memset(random_name, 0, sizeof(random_name));
>>        sprintf(random_name, "testXXXXXX");
>> @@ -70,8 +76,20 @@ void create_file(ocfs2_filesys *fs, uint64_t blkno,
>> uint64_t *retblkno)
>>        if (ret)
>>                FSWRK_COM_FATAL(progname, ret);
>>
>> +       /* Fix the i_size of the new file to greater than zero */
>> +       ret = ocfs2_read_inode(fs, tmp_blkno, buf);
>> +       if (ret)
>> +               FSWRK_COM_FATAL(progname, ret);
>> +       di = (struct ocfs2_dinode *)buf;
>> +
>> +       di->i_size = ocfs2_clusters_to_bytes(fs, 1) - 25;
>> +       ret = ocfs2_write_inode(fs, tmp_blkno, buf);
>> +       if (ret)
>> +               FSWRK_COM_FATAL(progname, ret);
>> +
>>        *retblkno = tmp_blkno;
>>
>> +       ocfs2_free(&buf);
>>        return;
>>  }
>>
>
>



-- 
Goldwyn