From mboxrd@z Thu Jan 1 00:00:00 1970 From: ebiederm-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org (Eric W. Biederman) Subject: Re: [RFC][PATCH 0/8][v2]: Enable multiple mounts of devpts Date: Wed, 03 Sep 2008 06:12:06 -0700 Message-ID: References: <20080821031028.GB30205@us.ibm.com> <48ACDDC7.3000704@zytor.com> <48AD991F.9010906@fr.ibm.com> <48AD9A97.6000807@zytor.com> <48AD9DCD.3060306@fr.ibm.com> <48ADD7D3.7080400@fr.ibm.com> <48B7BB3C.5080404@fr.ibm.com> <20080902030426.GB12277@us.ibm.com> <20080902155211.GF8524@us.ibm.com> <48BE7C98.1040004@fr.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <48BE7C98.1040004-NmTC/0ZBporQT0dZR+AlfA@public.gmane.org> (Cedric Le Goater's message of "Wed, 03 Sep 2008 14:01:28 +0200") 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: Cedric Le Goater Cc: kyle-hoO6YkzgTuCM0SS3m2neIg@public.gmane.org, Dave Hansen , bastian-yyjItF7Rl6lg9hUCZPvPmw@public.gmane.org, "H. Peter Anvin" , containers-qjLDD68F18O7TbgM5vRIOg@public.gmane.org, alan-qBU/x9rampVanCEyBjwyrvXRex20P6io@public.gmane.org, xemul-GEFAQzZX7r8dnm+yROfE0A@public.gmane.org List-Id: containers.vger.kernel.org Cedric Le Goater writes: > ok. complete isolation would require 2 steps. I guess this is > acceptable because mq uses a fs > > allowing the host to see the child's /dev/mqueue is also 'a nice > to have' feature. unfortunately, we can't do that for all namespaces, > for sysvipc for example. So I'm wondering if we should allow it > at all ? Definitely. One of the lessons from the people doing monitoring is that it really is best done through a filesystem interface. You don't have to have it mounted and there are times you don't want to be able to mount a view into another namespace but having the option is nice. I'm torn because the more I look at the way posix message queues are implemented the more it looks like new versions of sys_open and sys_unlink should never have been written and it should have been a user space convention to always mount mqueuefs on /dev/mqueue. Just doing newinstance and having a pointer in nsproxy will get the job done, but it feels like we may have the opportunity to correct a blunder in the initial implementation. Eric