On 11/16/22 23:03, Martin Wilck wrote:
On Wed, 2022-11-16 at 13:22 -0800, Bart Van Assche wrote:
On 11/16/22 02:32, Martin Wilck wrote:
[...]
Although I like the approach of this patch, how about the
implementation below?
I think the implementation below is a little cleaner but that might
be subjective ...
Fine with me, absolutely, and indeed better than mine. I don't quite
understand the last hook - sdev->handler_data isn't protected by rcu in
any way, is it? But that doesn't matter much.
Hi Martin,
I plan to leave out the sdev->handler_data assignment change and to follow
the approach from your patch for the sdev->handler_data pointer (check
whether or not it is NULL) since I'm no longer sure that the change I
proposed is safe.
Thanks,
Bart.
