Re: linux-next: manual merge of the zonefs tree with the vfs-idmapping tree

[Damien Le Moal <damien.lemoal@opensource.wdc.com>]

On 1/24/23 17:32, Christian Brauner wrote:
> On Tue, Jan 24, 2023 at 09:11:47AM +0900, Damien Le Moal wrote:
>> On 1/24/23 09:07, Stephen Rothwell wrote:
>>> Hi Damien,
>>>
>>> On Tue, 24 Jan 2023 08:30:29 +0900 Damien Le Moal <damien.lemoal@opensource.wdc.com> wrote:
>>>>
>>>> OK. I think I will merge the 3 patches that create the conflict and rebase
>>>> the patches. I need that for retesting at least. But given the size of the
>>>> conflict resolution, I may push that as an update to my for-6.3/for-next
>>>> branch. Let me see...
>>>>
>>>>> Alternatively, just leave the fix up to Linus (but mention it to him
>>>>> when you send your pull requests).  
>>>>
>>>> Understood. Let me retest first :)
>>>
>>> When I said "merge", I meant literally "git merge <some stable branch
>>> from the vfs-mapping tree that contains the conflicting commit>" not
>>> cherry pick the commits i.e. you would need to coordinate with
>>> Christian about having a common branch (or making sure that the part of
>>> his tree you pull in is immutable).
>>
>> Yep, cherry picking did not work :)
>> I did a merge test and came up with the same resolution as you did. It
>> looks good. It looks big but is in fact fairly simple. I will keep it as
>> is and signal it to Linus when I send my PR.
> 
> I don't rebase branches after they're in -next as soon as someone
> wants to depend on them.

I understand that. Nobody does.

> 
>>
>> But retesting everything to be sure there are no issues.
>>
>> Christian,
>>
>> Next time, when you touch an fs, please cc the maintainer for acks. I had
>> that zonefs series ready for a while and we could have coordinated for the
>> conflict...
> 
> I understand merge conflicts aren't pleasant as I'm dealing with them on
> a regular basis myself and I'm sorry that this is causing churn for you.

The conflict is fine, I can handle. I was more surprised that patches
touching zonefs were applied without acks from me.

> 
> Similar to other large branches such as the initial folio conversion
> that affected a lot of filesystems and other branches of mine it simply
> becomes impractical to generate a massive recipients list.

Fair enough.

> All filesystems were touched in non-semantical ways and simply replace a
> vfs type.

That I saw :)

> 
> One of linux-next's tasks is to find generated merge conflicts so that
> we can coordinate. As usual, I will send a list of merge conflicts
> caused by one of our branches to Linus and point him to the relevant
> threads that Steven reported with proposed resolutions as he prefers to
> fix them up himself.
> 
> Sorry for the inconvenience.

All solved now. My morning was only a little more busy than expected
dealing with this :)
Thanks.

-- 
Damien Le Moal
Western Digital Research