Re: [PATCH] pid: Allow creation of pidfds to threads

[Christian Brauner <brauner@kernel.org>]

On Thu, Mar 31, 2022 at 09:02:32PM +0200, Alois Wohlschlager wrote:
> The pidfd_open() syscall now allows retrieving pidfds to processes which
> are not thread group leaders. Like standard pidfds so far, these may be
> used to retrieve file descriptors from the target thread using
> pidfd_getfd(), as well as for killing the target thread group using
> pidfd_send_signal().
> However, unlike pidfds referencing thread group leaders, they do not
> support polling for process exit. Attempts to do so signal an error
> condition instead of blocking indefinitely.
> 
> Since the semantics of pidfd_getfd() and pidfd_send_signal() are not
> very useful within a thread group, these thread pidfds can only be
> created using pidfd_open(), not via clone().
> 
> Signed-off-by: Alois Wohlschlager <alois1@gmx-topmail.de>
> ---

Hey Alois,

We originally blocked this because it is not as easy as simply allowing
pidfds to be created for non-thread-group leaders.
For a start, pidfd_poll() currently doens't work if pidfd_task() isn't a
thread-group leader and you'll just hang for CLONE_PIDFD | CLONE_THREAD.
So at least that needs to be adapated as well and there's likely a bunch
of other corner-cases I'm forgetting about.
Do you have a concrete use-case you want this for?

Christian