Re: [PATCH] exit: combine work under lock in synchronize_group_exit() and coredump_task_exit()

[Oleg Nesterov]

On 03/19, Mateusz Guzik wrote:
>
> This happens to test to for coredumping prior to handling kcov and
> kmsan, which afaics is harmless.
>
> I however was not comfortable lifting dumping prior to it, so it still
> happens after.

Well, I'd really prefer to move synchronize_group_exit(tsk, code) down,
after kmsan_task_exit() for the start.

Then the patch and resulting code will be simpler, coredump_task_exit()
can be even folded into synchronize_group_exit().

Oleg.