From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753291Ab1DVInk (ORCPT ); Fri, 22 Apr 2011 04:43:40 -0400 Received: from mail-pz0-f46.google.com ([209.85.210.46]:33947 "EHLO mail-pz0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752485Ab1DVInj convert rfc822-to-8bit (ORCPT ); Fri, 22 Apr 2011 04:43:39 -0400 DomainKey-Signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type:content-transfer-encoding; b=B1GOvuMiZi1hRuq12TJZHHUqMvlo60kTEoPZGJicVboQ6Aj3k9Mc5Mu51aPAeOVfgb nRyIP0Qibfo2VZrg+b7bHvN/lR+BWu+J/Q25wR6PpnyGBgoev3uaOt3Z0Q/E9u1lz1UZ alP93BCKOQpIhqm/Ix6mPugagi3yaSvdvY400= MIME-Version: 1.0 In-Reply-To: <20110422081838.GA24011@elte.hu> References: <1303398203-2918-1-git-send-email-dzickus@redhat.com> <1303398203-2918-5-git-send-email-dzickus@redhat.com> <20110422081838.GA24011@elte.hu> Date: Fri, 22 Apr 2011 12:43:38 +0400 Message-ID: Subject: Re: [PATCH 4/4] perf, x86: Add PERF_COUNT_HW_NMI_WATCHDOG event From: Cyrill Gorcunov To: Ingo Molnar Cc: Don Zickus , "x86@kernel.org" , LKML , Cyrill Gorcunov , Peter Zijlstra Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8BIT Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org P4 cant move events between counters in compare with architectural event so when nmi-watchdog is used perf top is blocked because it needs same counters but the counters already borrowed for watchdog. as result we need a different event and counters for watchdog. Ingo, nonsleeping cycles is from intel sdm, so when someone start reading this code in future will find this term in sdm fast. or i could put the reference from sdm itself here, hm? On Friday, April 22, 2011, Ingo Molnar wrote: > > * Don Zickus wrote: > >> From: Cyrill Gorcunov >> >> Due to restriction and specifics of Netburst PMU we need >> a separated event for NMI watchdog. Note that on all other >> than P4 PMU cpus this event is a simple alias for PERF_COUNT_HW_CPU_CYCLES. >> >> This event is only used by x86 for now so no other archs involved. > > hm, there's no explanation exactly why this is needed and what the alternatives > are. > > This sentence: > >> +  /* we use that named non-sleeping calls */ > > does not parse for me. > > Thanks, > >        Ingo >