Re: [PATCH RFC 1/3] ext4:Add a function merging extent right and left.

[Yongqiang Yang <xiaoqiangnk@gmail.com>]

On Fri, Apr 15, 2011 at 3:03 AM, Mingming Cao <cmm@us.ibm.com> wrote:
> On Wed, 2011-04-13 at 22:40 -0700, Yongqiang Yang wrote:
>> 1] Rename ext4_ext_try_to_merge() to ext4_ext_try_to_merge_right().
>>
>> 2] Add a new function ext4_ext_try_to_merge() which tries to merge
>>    an extent both left and right.
>>
>> 3] Use the new function in ext4_ext_convert_unwritten_endio() and
>>    ext4_ext_insert_extent().
>>
>> Signed-off-by: Yongqiang Yang <xiaoqiangnk@gmail.com>
>> ---
>>  fs/ext4/extents.c |   65 ++++++++++++++++++++++++++++------------------------
>>  1 files changed, 35 insertions(+), 30 deletions(-)
>>
>> diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
>> index dd2cb50..11f30d2 100644
>> --- a/fs/ext4/extents.c
>> +++ b/fs/ext4/extents.c
>> @@ -1563,7 +1563,7 @@ ext4_can_extents_be_merged(struct inode *inode, struct ext4_extent *ex1,
>>   * Returns 0 if the extents (ex and ex+1) were _not_ merged and returns
>>   * 1 if they got merged.
>>   */
>> -static int ext4_ext_try_to_merge(struct inode *inode,
>> +static int ext4_ext_try_to_merge_right(struct inode *inode,
>>                                struct ext4_ext_path *path,
>>                                struct ext4_extent *ex)
>>  {
>> @@ -1603,6 +1603,31 @@ static int ext4_ext_try_to_merge(struct inode *inode,
>>  }
>>
>>  /*
>> + * This function tries to merge the @ex extent to neighbours in the tree.
>> + * return 1 if merge left else 0.
>> + */
>> +static int ext4_ext_try_to_merge(struct inode *inode,
>> +                               struct ext4_ext_path *path,
>> +                               struct ext4_extent *ex) {
>> +     struct ext4_extent_header *eh;
>> +     unsigned int depth;
>> +     int merge_done = 0;
>> +     int ret = 0;
>> +
>> +     depth = ext_depth(inode);
>> +     BUG_ON(path[depth].p_hdr == NULL);
>> +     eh = path[depth].p_hdr;
>> +
>> +     if (ex > EXT_FIRST_EXTENT(eh))
>> +             merge_done = ext4_ext_try_to_merge_right(inode, path, ex - 1);
>> +
>> +     if (!merge_done)
>> +             ret = ext4_ext_try_to_merge_right(inode, path, ex);
>> +
>
> Is there any reason not to merge right after merge left? The old code
> does both, I think.
What does merge left do?  Actually it does as merge right, it merge
ex-1 to right until no more merges can be done.  So if merge to right
happens then, ex,ex+1 has been tried to merge also.

>
>> +     return ret;
>> +}
>> +
>> +/*
>>   * check if a portion of the "newext" extent overlaps with an
>>   * existing extent.
>>   *
>> @@ -3039,6 +3064,7 @@ fix_extent_len:
>>       ext4_ext_dirty(handle, inode, path + depth);
>>       return err;
>>  }
>> +
>>  static int ext4_convert_unwritten_extents_endio(handle_t *handle,
>>                                             struct inode *inode,
>>                                             struct ext4_ext_path *path)
>> @@ -3047,46 +3073,25 @@ static int ext4_convert_unwritten_extents_endio(handle_t *handle,
>>       struct ext4_extent_header *eh;
>>       int depth;
>>       int err = 0;
>> -     int ret = 0;
>>
>>       depth = ext_depth(inode);
>>       eh = path[depth].p_hdr;
>>       ex = path[depth].p_ext;
>>
>> +     ext_debug("ext4_convert_unwritten_extents_endio: inode %lu, logical"
>> +             "block %llu, max_blocks %u\n", inode->i_ino,
>> +             (unsigned long long)le32_to_cpu(ex->ee_block),
>> +             ext4_ext_get_actual_len(ex));
>> +
>>       err = ext4_ext_get_access(handle, inode, path + depth);
>>       if (err)
>>               goto out;
>>       /* first mark the extent as initialized */
>>       ext4_ext_mark_initialized(ex);
>>
>> -     /*
>> -      * We have to see if it can be merged with the extent
>> -      * on the left.
>> -      */
>> -     if (ex > EXT_FIRST_EXTENT(eh)) {
>> -             /*
>> -              * To merge left, pass "ex - 1" to try_to_merge(),
>> -              * since it merges towards right _only_.
>> -              */
>> -             ret = ext4_ext_try_to_merge(inode, path, ex - 1);
>> -             if (ret) {
>> -                     err = ext4_ext_correct_indexes(handle, inode, path);
>> -                     if (err)
>> -                             goto out;
>> -                     depth = ext_depth(inode);
>> -                     ex--;
>> -             }
>> -     }
>> -     /*
>> -      * Try to Merge towards right.
>> -      */
>> -     ret = ext4_ext_try_to_merge(inode, path, ex);
>> -     if (ret) {
>> -             err = ext4_ext_correct_indexes(handle, inode, path);
>> -             if (err)
>> -                     goto out;
>> -             depth = ext_depth(inode);
>> -     }
>> +     /* correct indexes is nt needed becasue borders are not changed */
>> +     ext4_ext_try_to_merge(inode, path, ex);
>> +
>
> Ah, so you discovered an issue -- currently ext4 can't merge across the
> index block borders. that's a pity. This might need to fix up,
Yes, currently borders are changed only in one case that storing an
extent to an empty leaf.

I think the patch which enable merging across index block should be
independent on these patches.


> especially with split is going to be heavily used in punch hole,
> snapshots, direct IO handling holes.
What does direct IO here refer to?  It seems I am missing something.

These patches involves ext4_ext_convrt_to_initialized() in buffer
write case, and split_unwritten_extents() and
convert_to_initialized_endio() in direct IO case.

>
>
>>       /* Mark modified extent as dirty */
>>       err = ext4_ext_dirty(handle, inode, path + depth);
>>  out:
>
>
>



-- 
Best Wishes
Yongqiang Yang
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html