From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Morton Subject: Re: [PATCH 14/38] Remove struct mm_struct::exe_file et al Date: Sun, 31 May 2009 16:50:26 -0700 Message-ID: <20090531165026.376a914c.akpm@linux-foundation.org> References: <20090526113618.GJ28083@us.ibm.com> <20090526162415.fb9cefef.akpm@linux-foundation.org> <20090531215427.GA29534@x200.localdomain> <20090531151953.8f8b14b5.akpm@linux-foundation.org> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: 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: Linus Torvalds Cc: xemul-bzQdu9zFT3WakBO8gow8eQ@public.gmane.org, containers-cunTk1MwBs9QetFLy7KEm3xJsTq8ys+cHZ5vskTnxNA@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, dave-23VcF4HTsmIX0ybBhKVfKdBPR1lH4CV8@public.gmane.org, mingo-X9Un+BFzKDI@public.gmane.org, Alexey Dobriyan List-Id: containers.vger.kernel.org On Sun, 31 May 2009 16:15:50 -0700 (PDT) Linus Torvalds wrote: > > > On Sun, 31 May 2009, Andrew Morton wrote: > > > > What I always find a bit weird is that an MM container is used as the > > central point for a number of sched obects. But it's logical, given > > that the never-before-stated definition of a heavyweight process is > > "thing which share a VM". > > It has nothing to do with "heavy-weight process" or anything else. > > The thing is, from a scheduling standpoint, one of the primary performance > concerns in the TLB switch. > > And there's a 1:1 relationship between TLB switch and MM container, modulo > the issue of kernel tasks (and those obviously "borrow" approproate MM > structs to avoid the switch). That's all an obscure performance-oriented internal implementation detail. > So it's not weird at all. It's very direct, and a very straightforward and > obvious relationship. It's arbitrary! If we were to gain more performance benefit by aggregating processes under, say, the fs_struct then that's the way the kernel would have been implemented.