Re: [PATCH 0/2] dm thin: stop waiting for grace periods when activating and removing thins

From: Qiliang Yuan

Date: Thu Nov 19 2026 - 00:47:08 EST


On Thu, Nov 19, 2026 at 12:24:21AM -0500, Benjamin Marzinski wrote:
> I posted an alternate version of your [PATCH 2/2] that fixes the issue
> I mentioned. Can you verify that it works for you?
>
> I squashed it with your patch, since it your original patch did most of
> the work of fixing the issue, and it seemed easier to just have one
> patch to backport to fix it. I could refactor it into two patches: one
> the fixes the bug, and a second that just swaps the synchronize_rcu()
> for kfree_rcu(), if people would rather not mix the bug fix and feature.

Thanks. I had a v2 ready along the same lines and sent it before seeing
this:

https://lore.kernel.org/r/20261119-bug-dm-thin-rcu-gp-v2-0-055d627763c3@xxxxxxxxx

It carries the fix as its own patch 2/3, ahead of the kfree_rcu()
change in 3/3. It differs from your version in two places: it takes the
thin off the list in the thin_put() that drops the last reference,
under pool->lock, where yours does it in thin_dtr() after the
completion, which is simpler; and it does not skip dying thins in
requeue_bios() and error_retry_list_with_code(), which yours does.

Your Fixes tag is the right one. 17181fb7a0c3 moved list_del_rcu()
ahead of waiting for the references; v2 points at b10ebd34cccae, which
is wrong.

I have not run your version yet. I will run the KASAN reproducer and
the timing on it and report back.

Thanks,
Qiliang