Re: [PATCH net 1/2] net: bpfilter: restart bpfilter_umh when error occurred

On Wed, 19 Dec 2018 at 07:51, David Miller [off-list ref] wrote:

From: Taehee Yoo <ap420073@gmail.com>
Date: Mon, 17 Dec 2018 16:31:04 +0900

quoted

But I don't know how cleanup callback is invoked when
bpfilter_umh process is killed.

I am suggesting that a new piece of generic infrastructure might
be needed, but it would need to be carefully designed.

The task_struct would get a piece of state, and at exit() time
the kernel would check that state and use it to invoke exit()
time cleanups for UMH type processes.

Is the idea clearer now?

This idea is clear and I will try to make an infrastructure code
to clean up when UMH type process is killed.

Thanks a lot for the suggestion!