[Ocfs2-tools-devel] [PATCH 2/3] O2image: Cleanup for redundant bit mapping.

[tristan]

Srinivas Eeda wrote:
> ocfs2_image_mark_bitmap sets the bit in memory. It sets some bits 
> redundantly so that in the case of corrupt fs it might help. Are you 
> running into any issue ?
>

No, I didn't ever run into any issue so far;)

It did do redundant mappings if you output the blkno as debugging. 
However, it's neither a performance killer nor bug producer as you said, 
and it may help when bit did got set in extent_alloc and btree was using 
corresponding blk in a corrupted fs.

>
> On 5/24/2010 5:25 AM, Tristan Ye wrote:
>> traverse_extents() should not be trying to map btree extent blocks since
>> we've already mapped them when doing scans for extent_alloc chains.
>>
>> Am I missing anything?
>>
>> Signed-off-by: Tristan Ye <tristan.ye at oracle.com>
>> ---
>>  o2image/o2image.c |    1 -
>>  1 files changed, 0 insertions(+), 1 deletions(-)
>>
>> diff --git a/o2image/o2image.c b/o2image/o2image.c
>> index 6f32486..9bac71a 100644
>> --- a/o2image/o2image.c
>> +++ b/o2image/o2image.c
>> @@ -106,7 +106,6 @@ static errcode_t traverse_extents(ocfs2_filesys 
>> *ofs,
>>  
>>      for (i = 0; i < el->l_next_free_rec; ++i) {
>>          rec = &(el->l_recs[i]);
>> -        ocfs2_image_mark_bitmap(ofs, rec->e_blkno);
>>          if (el->l_tree_depth) {
>>              ret = ocfs2_read_extent_block(ofs, rec->e_blkno, buf);
>>              if (ret)
>>