[prev in list] [next in list] [prev in thread] [next in thread] 

List:       coreutils
Subject:    Re: Supporting reflinked/cow files in coreutils
From:       Jie Liu <jeff.liu () oracle ! com>
Date:       2012-10-09 10:16:32
Message-ID: 5073F980.60909 () oracle ! com
[Download RAW message or body]

On 10/09/12 15:05, Jim Meyering wrote:
> Jeff liu wrote:
>> Hi Philipp,
>>
>> Thanks for taking a look at this feature, and sorry for my late response!
>>
>> Hi Jim,
>>
>> Recently, I have discussed with a colleague(Bo Liu CC-ed) who is focus
>> on Btrfs development, and we
>> decided to teach Btrfs aware of reflinked extents per file for du(1) purpose. :)
Hi Jim,

> Hi Jeff,
>
> When do you expect that support to be in upstream kernels?
It depends on Bo's time as well as whether we can get a quick review
from Btrfs upstream folks or not.
>
> ...
>>> If you and/or Jeff are motivated enough, a good first step would be to
>>> post a rebased patch to bug-coreutils, preferably after you've switched
>>> out that rbtree.[ch] for gnulib's code.  I hesitate even to suggest
>>> that you rebase, because we might end up deciding it's not yet worth
>>> the cost, but let's hear what others have to say.  Maybe enough people
>>> use du and are bothered because the space used by their reflinked files
>>> is over-reported.
>> We can find time to rebase it according to your suggestions if you like.
> Depending on when btrfs support is expected, you may want to wait.
It's better to me.
> I would want to be able to test against both file system types.
Ok, I'll submit a test case ogether with du(1) patches.


Thanks,
-Jeff


[prev in list] [next in list] [prev in thread] [next in thread]