From mboxrd@z Thu Jan 1 00:00:00 1970 From: KOSAKI Motohiro Subject: Re: [PATCH v5] add MAP_UNLOCKED mmap flag Date: Thu, 14 Jan 2010 16:30:51 +0900 (JST) Message-ID: <20100114162327.673E.A69D9226@jp.fujitsu.com> References: <20100114155229.6735.A69D9226@jp.fujitsu.com> <20100114072210.GK18808@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20100114072210.GK18808-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> Sender: linux-api-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Gleb Natapov Cc: kosaki.motohiro-+CUm20s59erQFUHtdCDX3A@public.gmane.org, linux-mm-Bw31MaZKKs3YtjvyW6yDsg@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-api-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, akpm-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org List-Id: linux-api@vger.kernel.org > On Thu, Jan 14, 2010 at 04:02:42PM +0900, KOSAKI Motohiro wrote: > > > On Thu, Jan 14, 2010 at 09:31:03AM +0900, KOSAKI Motohiro wrote: > > > > > If application does mlockall(MCL_FUTURE) it is no longer possible to mmap > > > > > file bigger than main memory or allocate big area of anonymous memory > > > > > in a thread safe manner. Sometimes it is desirable to lock everything > > > > > related to program execution into memory, but still be able to mmap > > > > > big file or allocate huge amount of memory and allow OS to swap them on > > > > > demand. MAP_UNLOCKED allows to do that. > > > > > > > > > > Signed-off-by: Gleb Natapov > > > > > --- > > > > > > > > > > I get reports that people find this useful, so resending. > > > > > > > > This description is still wrong. It doesn't describe why this patch is useful. > > > > > > > I think the text above describes the feature it adds and its use > > > case quite well. Can you elaborate what is missing in your opinion, > > > or suggest alternative text please? > > > > My point is, introducing mmap new flags need strong and clearly use-case. > > All patch should have good benefit/cost balance. the code can describe the cost, > > but the benefit can be only explained by the patch description. > > > > I don't think this poor description explained bit benefit rather than cost. > > you should explain why this patch is useful and not just pretty toy. > > > The benefit is that with this patch I can lock all of my application in > memory except some very big memory areas. My use case is that I want to > run virtual machine in such a way that everything related to machine > emulator is locked into the memory, but guest address space can be > swapped out at will. Guest address space is so huge that it is not > possible to allocated it locked and then unlock. I was very surprised > that current Linux API has no way to do it hence this patch. It may look > like a pretty toy to you until some day you need this and has no way to > do it. Hmm.. Your answer didn't match I wanted. few additional questions. - Why don't you change your application? It seems natural way than kernel change. - Why do you want your virtual machine have mlockall? AFAIK, current majority virtual machine doesn't. - If this feature added, average distro user can get any benefit? I mean, many application developrs want to add their specific feature into kernel. but if we allow it unlimitedly, major syscall become the trushbox of pretty toy feature soon.