From mboxrd@z Thu Jan 1 00:00:00 1970 From: Daniel Lezcano Subject: Re: [PATCH 0/1][V3] Handle reboot in a child pid namespace Date: Mon, 05 Dec 2011 00:08:23 +0100 Message-ID: <4EDBFD67.1040009@free.fr> References: <1323030290-22216-1-git-send-email-daniel.lezcano@free.fr> <20111204212756.GB16362@khazad-dum.debian.net> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20111204212756.GB16362-ZGHd14iZgfaRjzvQDGKj+xxZW9W5cXbT@public.gmane.org> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: containers-bounces-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org Errors-To: containers-bounces-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org To: Henrique de Moraes Holschuh Cc: containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org, oleg-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, akpm-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org List-Id: containers.vger.kernel.org On 12/04/2011 10:27 PM, Henrique de Moraes Holschuh wrote: > On Sun, 04 Dec 2011, Daniel Lezcano wrote: >> * V3 >> - removed lock and serialization of pid_ns_reboot >> * V2 >> - added a lock for the pid namespace to prevent racy call >> to the 'reboot' syscall >> - Moved 'reboot' command assigned in zap_pid_ns_processes >> instead of wait_task_zombie >> - added tasklist lock around force_sig >> - added do_exit in pid_ns_reboot >> - used task_active_pid_ns instead of declaring a new variable in sys_reboot >> - moved code up before POWER_OFF changed to HALT in sys_reboot > Daniel, can you address Miquel's concern? Is it a valid concern, or > not? I assume CAP_REBOOT functionality is still in place inside the > container, so it really does look like userspace would need to know > whether it should drop CAP_REBOOT or not, in order to automatically use > the new feature. Hmm, I missed its email. I think it is worth to have such ability to detect how behaves the reboot syscall vs the pid ns. At present, if we call 'reboot' in a child pid namespace, that will affect the host, we are changing this behavior with this patch. I don't think there is any application doing a shutdown from a child pid namespace, that don't makes sense as the shutdown is invoked after killing all the processes on the system and that could only be done from the init_pid_ns. I would like to address this in a separate patch in order to discuss the best way to do that. Adding a fake 'reboot' parameter returning EINVAL or 0 seems a good solution to detect at runtime if the shutdown is correctly supported inside a container.