Re: Patch "md: call del_gendisk in control path" has been added to the 6.6-stable tree

[Yu Kuai <yukuai1@huaweicloud.com>]

Hi,

在 2025/08/18 17:39, Greg KH 写道:
> On Mon, Aug 18, 2025 at 03:13:11PM +0800, Yu Kuai wrote:
>> Hi,
>>
>> 在 2025/08/18 14:38, Greg KH 写道:
>>> On Mon, Aug 18, 2025 at 02:26:23PM +0800, Yu Kuai wrote:
>>>> Hi,
>>>>
>>>> 在 2025/08/18 13:55, Greg KH 写道:
>>>>> On Mon, Aug 18, 2025 at 09:03:39AM +0800, Yu Kuai wrote:
>>>>>> Hi,
>>>>>>
>>>>>> 在 2025/08/17 22:18, Sasha Levin 写道:
>>>>>>> This is a note to let you know that I've just added the patch titled
>>>>>>>
>>>>>>>         md: call del_gendisk in control path
>>>>>>>
>>>>>>> to the 6.6-stable tree which can be found at:
>>>>>>>         http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary
>>>>>>>
>>>>>>> The filename of the patch is:
>>>>>>>          md-call-del_gendisk-in-control-path.patch
>>>>>>> and it can be found in the queue-6.6 subdirectory.
>>>>>>>
>>>>>>> If you, or anyone else, feels it should not be added to the stable tree,
>>>>>>> please let <stable@vger.kernel.org> know about it.
>>>>>>>
>>>>>>>
>>>>>> This patch should be be backported to any stable kernel, this change
>>>>>> will break user tools mdadm:
>>>>>>
>>>>>> https://lore.kernel.org/all/f654db67-a5a5-114b-09b8-00db303daab7@redhat.com/
>>>>>
>>>>> Is it reverted in Linus's tree?
>>>>>
>>>>
>>>> No, we'll not revert it, this is an improvement. In order to keep user
>>>> tools compatibility, we added a switch in the kernel. As discussed in
>>>> the thread, for old tools + new kernel, functionality is the same,
>>>> however, there will be kernel warning about deprecated behaviour to
>>>> inform user upgrading user tools.
>>>>
>>>> However, I feel this new warning messages is not acceptable for
>>>> stable kernels.
>>>
>>> Why?  What is so special about stable kernels that taking the same
>>> functionality in newer kernels is not ok?
>>>
>>> Why not just "warn" the same here if you want to fix an issue where
>>> userspace should be also updating some tools.  As long as you aren't
>>> breaking anything, it should be fine, right?
>>
>> Yes, it's fine, just in downstream kernels, people won't be happy about
>> new warnings.
> 
> People are NEVER happy about new warnings.  So why are you warning them
> at all in newer kernels?

Again, only old user tools + new kernels will warn, and behave like old
tools + old kernels. We'll need both kernel and user tool to be updated
to fix this problem.

> 
>>> Or are you breaking existing workflows?  You should be able to take a
>>> new kernel without any userspace changes and all should work the same.
>>> Why make new kernel users change userspace tools at all?
>>
>> There is a pending fix in mdraid that will be pushed soon, with this
>> nothing will be broken.
> 
> Great, what is the git id?

It's just pushed to Jen's tree:

https://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux-block.git/commit/?h=block-6.17&id=25db5f284fb8f30222146ca15b3ab8265789da38
> 
>> And because user tools have problems in this case as well, both kernel
>> and user tools have to be fixed to make things better.
> 
> So Linus's tree right now doesn't work without the pending fix as well?

Sadly, yes, this is indeed a regression from rc1. We're using latest
user tools for test and didn't realize this problem in time.

Thanks,
Kuai

> 
> thanks,
> 
> greg k-h
> 
> .
>