From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Serge E. Hallyn" Subject: Re: [PATCH] Add a choosepid() syscall as a simpler alternative to clone_with_pids() Date: Mon, 2 Nov 2009 16:10:16 -0600 Message-ID: <20091102221016.GA13766@us.ibm.com> References: <1257194293-12099-1-git-send-email-danms@us.ibm.com> <4AEF45BB.4080609@librato.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Content-Disposition: inline In-Reply-To: <4AEF45BB.4080609-RdfvBDnrOixBDgjK7y7TUQ@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: Oren Laadan Cc: containers-qjLDD68F18O7TbgM5vRIOg@public.gmane.org, Dan Smith , ebiederm-aS9lmoZGLiVWk0Htik3J/w@public.gmane.org List-Id: containers.vger.kernel.org Quoting Oren Laadan (orenl-RdfvBDnrOixBDgjK7y7TUQ@public.gmane.org): > Hi Dan, > > Two comments: > > 1. We already tried a similar approach over a year ago - selecting pids > using a /proc interface instead of a dedicated syscall - and received > chilling reactions. (IIRC posted by Nadia Derbey). The chilling reactions were due to (1) general objections to the functionality of clone_with_pids() itself, (2) new procfiles, the way alloc_pid() was being extended (i.e. that was solved by Suka as well as (1)) and especially (3) using a generic, multiplexed 'next_id' file. I don't think we can extrapolate from those objections to this patchset being objectionable. > 2. What do you expect to gain by splitting the work into two separate > system calls ? Two very simple syscalls which are both trivially correct and trivial for userspace to use. -serge