Re: [PATCH v2 2/2] btrfs: automatically remove the subvolume qgroup

[Qu Wenruo <quwenruo.btrfs@gmx.com>]

在 2024/5/3 00:33, David Sterba 写道:
> On Wed, May 01, 2024 at 07:57:08AM +0930, Qu Wenruo wrote:
>>
>>
>> 在 2024/5/1 07:48, Boris Burkov 写道:
>>> On Wed, May 01, 2024 at 07:35:09AM +0930, Qu Wenruo wrote:
>> [...]
>>>>>
>>>>> I don't see how a compat bit would work here, we use them for feature
>>>>> compatibility and for general access to data (full or read-only). What
>>>>> we do with individual behavioral changes are sysfs files. They're
>>>>> detectable by scripts and can be also used for configuration. In this
>>>>> case enabling/disabling autoclean of the qgroups.
>>>
>>> This was my initial thought too, but your compat bit idea is interesting
>>> since it persists? I vote sysfs since it has good
>>> infrastructure/momentum already for similar config.
>>
>> Sysfs is another way which we are already utilizing for qgroups, like
>> drop_subtree_threshold.
>>
>> The problem is as mentioned already, it's not persistent, thus it needs
>> a user space daemon to set it for every fs after mount.
>
> My idea of using sysfs is to export the information that the
> autocleaning feature is present and if we make it on by default then
> there's no need for additional step to enable it. The feedback about
> that was that it should have been default so we're going to make that
> change, but with sysfs export also provide a fallback to disable it in
> case it breaks things for somebody.
>
>> I'm totally fine to go sysfs for now, but I really hope to a persistent
>> solution.
>> Maybe a dedicated config tree?
>
> No, we already have a way to store data in the trees or in the
> properties so no new tree.

That means a on-disk format change.
IIRC everytime we introduce a new TEMP objectid, it should at least be
compat or compat_ro bit change.

Or older kernel won't understand nor follow the new TEMP key.

Thanks,
Qu