Re: [PATCH RFC v1] pidfd: fix a race in setting exit_state for pidfd polling

[Oleg Nesterov]

On 07/17, Joel Fernandes (Google) wrote:
>
> exit_notify
>   do_notify_parent
>     do_notify_pidfd
>   tsk->exit_state = EXIT_DEAD

OOPS. Somehow I thought it sets exit_state before do_notify_parent(),
I didn't even bother to verify this when I reviewed pidfd_poll()...

Sorry, can't read the patch today, will do tomorrow.

Oleg.