Re: [PATCH 5.10 033/238] md: implement ->set_read_only to hook into BLKROSET processing

From: Li Nan
Date: Tue Mar 26 2024 - 21:18:20 EST




在 2024/3/26 16:46, Song Liu 写道:
Hi Li Nan,

Could you please look into this (back port 9674f54e41ff to older stable
kernels)? If there is no clean back port, I would recommend we not do
the back port.


There are some conflicts to back port, which are not related to the
modification of this patch. If necessary, let me know and I can adapt and
send it :)

Thanks,
Song

On Tue, Mar 26, 2024 at 12:40 AM Christoph Hellwig <hch@xxxxxx> wrote:

On Mon, Mar 25, 2024 at 07:26:43AM -0400, Sasha Levin wrote:
On Mon, Mar 25, 2024 at 02:04:35AM +0100, Christoph Hellwig wrote:
How did we end up backporting all these block layer API changes?

They were brought in as a dependency for 9674f54e41ff ("md: Don't clear
MD_CLOSING when the raid is about to stop").

It's possible to work around bringing them during backport, but I
preferred to bring the dependencies instead.

I really don't see what these giant backports bring us. If people
want all the changes they'd just be better off on a modern kernel
rather than these frankenkernels. The automatic backporting is
gettind way out of hand. If the MD maintainers want
9674f54e41ff, maybe they can send a tailor made backport?

.

--
Thanks,
Nan