From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753978Ab0CWR5k (ORCPT ); Tue, 23 Mar 2010 13:57:40 -0400 Received: from mail-fx0-f223.google.com ([209.85.220.223]:55397 "EHLO mail-fx0-f223.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753910Ab0CWR5j convert rfc822-to-8bit (ORCPT ); Tue, 23 Mar 2010 13:57:39 -0400 DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=subject:mime-version:content-type:from:in-reply-to:date:cc :content-transfer-encoding:message-id:references:to:x-mailer; b=ZwEEDCe3qw/Y7tR/Smd+Bt3pXkgx0gaWUi3PU5cA2yKyOttkiK+0NPRJN/p4AGE1GZ fCqDO1z+7qZphJAL6CgfZBvCCywb18WaoVcoGVZ8i/koVi0hejcPeUq3DgadgL7yGYCN /VTR4PbTGBJWMcKBXWnUZVF2s+mFYrwa798/4= Subject: Re: [Bugme-new] [Bug 15618] New: 2.6.18->2.6.32->2.6.33 huge regression in performance Mime-Version: 1.0 (Apple Message framework v1077) Content-Type: text/plain; charset=us-ascii From: Anton Starikov In-Reply-To: Date: Tue, 23 Mar 2010 18:57:34 +0100 Cc: Ingo Molnar , Andrew Morton , linux-mm@kvack.org, linux-kernel@vger.kernel.org, bugzilla-daemon@bugzilla.kernel.org, bugme-daemon@bugzilla.kernel.org, Peter Zijlstra Content-Transfer-Encoding: 8BIT Message-Id: <54F3A3FB-E99F-4278-AAAB-5B6A09247C4B@gmail.com> References: <20100323102208.512c16cc.akpm@linux-foundation.org> <20100323173409.GA24845@elte.hu> To: Linus Torvalds X-Mailer: Apple Mail (2.1077) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mar 23, 2010, at 6:45 PM, Linus Torvalds wrote: > > > On Tue, 23 Mar 2010, Ingo Molnar wrote: >> >> It shows a very brutal amount of page fault invoked mmap_sem spinning >> overhead. > > Isn't this already fixed? It's the same old "x86-64 rwsemaphores are using > the shit-for-brains generic version" thing, and it's fixed by > > 1838ef1 x86-64, rwsem: 64-bit xadd rwsem implementation > 5d0b723 x86: clean up rwsem type system > 59c33fa x86-32: clean up rwsem inline asm statements > > NOTE! None of those are in 2.6.33 - they were merged afterwards. But they > are in 2.6.34-rc1 (and obviously current -git). So Anton would have to > compile his own kernel to test his load. Thanks for info, I will try it now. > We could mark them as stable material if the load in question is a real > load rather than just a test-case. On one of the random page-fault > benchmarks the rwsem fix was something like a 400% performance > improvement, and it was apparently visible in real life on some crazy SGI > "initialize huge heap concurrently on lots of threads" load. It is not just a test-case, it is real-life code. With real-life problems on 2.6.32 and later :) Anton. From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail191.messagelabs.com (mail191.messagelabs.com [216.82.242.19]) by kanga.kvack.org (Postfix) with SMTP id 308136B01B3 for ; Tue, 23 Mar 2010 13:57:41 -0400 (EDT) Received: by fxm10 with SMTP id 10so1004732fxm.30 for ; Tue, 23 Mar 2010 10:57:38 -0700 (PDT) Subject: Re: [Bugme-new] [Bug 15618] New: 2.6.18->2.6.32->2.6.33 huge regression in performance Mime-Version: 1.0 (Apple Message framework v1077) Content-Type: text/plain; charset=us-ascii From: Anton Starikov In-Reply-To: Date: Tue, 23 Mar 2010 18:57:34 +0100 Content-Transfer-Encoding: quoted-printable Message-Id: <54F3A3FB-E99F-4278-AAAB-5B6A09247C4B@gmail.com> References: <20100323102208.512c16cc.akpm@linux-foundation.org> <20100323173409.GA24845@elte.hu> Sender: owner-linux-mm@kvack.org To: Linus Torvalds Cc: Ingo Molnar , Andrew Morton , linux-mm@kvack.org, linux-kernel@vger.kernel.org, bugzilla-daemon@bugzilla.kernel.org, bugme-daemon@bugzilla.kernel.org, Peter Zijlstra List-ID: On Mar 23, 2010, at 6:45 PM, Linus Torvalds wrote: >=20 >=20 > On Tue, 23 Mar 2010, Ingo Molnar wrote: >>=20 >> It shows a very brutal amount of page fault invoked mmap_sem spinning=20= >> overhead. >=20 > Isn't this already fixed? It's the same old "x86-64 rwsemaphores are = using=20 > the shit-for-brains generic version" thing, and it's fixed by >=20 > 1838ef1 x86-64, rwsem: 64-bit xadd rwsem implementation > 5d0b723 x86: clean up rwsem type system > 59c33fa x86-32: clean up rwsem inline asm statements >=20 > NOTE! None of those are in 2.6.33 - they were merged afterwards. But = they=20 > are in 2.6.34-rc1 (and obviously current -git). So Anton would have to=20= > compile his own kernel to test his load. Thanks for info, I will try it now. > We could mark them as stable material if the load in question is a = real=20 > load rather than just a test-case. On one of the random page-fault=20 > benchmarks the rwsem fix was something like a 400% performance=20 > improvement, and it was apparently visible in real life on some crazy = SGI=20 > "initialize huge heap concurrently on lots of threads" load. It is not just a test-case, it is real-life code. With real-life = problems on 2.6.32 and later :) Anton.= -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: email@kvack.org