From mboxrd@z Thu Jan 1 00:00:00 1970 From: Thomas Gleixner Subject: Re: [ANNOUNCE] 3.0-rt4 Date: Thu, 28 Jul 2011 11:37:46 +0200 (CEST) Message-ID: References: <201107281224.44323@zigzag.lvk.cs.msu.su> <1311842677.5890.209.camel@twins> <201107281251.16892@zigzag.lvk.cs.msu.su> Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Cc: Peter Zijlstra , LKML , linux-rt-users , "Paul E. McKenney" , Steven Rostedt , Jason Wessel , lasaine@lvk.cs.msu.su To: "Nikita V. Youshchenko" Return-path: Received: from www.linutronix.de ([62.245.132.108]:57801 "EHLO Galois.linutronix.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754797Ab1G1Jhx (ORCPT ); Thu, 28 Jul 2011 05:37:53 -0400 In-Reply-To: <201107281251.16892@zigzag.lvk.cs.msu.su> Sender: linux-rt-users-owner@vger.kernel.org List-ID: On Thu, 28 Jul 2011, Nikita V. Youshchenko wrote: > > On Thu, 2011-07-28 at 12:24 +0400, Nikita V. Youshchenko wrote: > > > (*) we need >1G of memory for data buffers, and have too much legacy > > > software components so moving to 64bit system is not practical within > > > available resource > > > > You can run a complete 32bit userspace on a 64bit kernel without any > > problems. > > ... expect shared data between kernel and userspace (structures passed over > ioctl()s, structures in mmapped buffers). Unfortunately, in scope of our > project (that lasts since 2001 or so), we do have such issues :(. Aside of that being horrid, it's trivial to solve with proper data types. u32 and friends are there for a reason. But feel free to get the highmem thing working again :) Thanks, tglx