From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christian Borntraeger Date: Mon, 14 Jul 2008 07:44:42 +0000 Subject: Re: [PATCH 2/5] kvmtrace: make cycle calculation architecture aware Message-Id: <200807140944.42360.borntraeger@de.ibm.com> List-Id: References: <1215439013-11480-1-git-send-email-ehrhardt@linux.vnet.ibm.com> <48770CEF.6090102@de.ibm.com> <487A2210.5020602@qumranet.com> In-Reply-To: <487A2210.5020602@qumranet.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Avi Kivity Cc: carsteno@de.ibm.com, "Yang, Sheng" , kvm@vger.kernel.org, Hollis Blanchard , Christian Ehrhardt , xiantao.zhang@intel.com, kvm-ppc@vger.kernel.org Am Sonntag, 13. Juli 2008 schrieb Avi Kivity: > Carsten Otte wrote: > > Avi Kivity wrote: > >> What is the overhead of ktime_get()? > > I think I'd like an arch specific timestamp. This way we could use our > > clock-cycle-granularity-non-privileged-timestamp instruction ;-). If > > we need a common implementation, I don't think there's much difference > > between different syscalls in terms of overhead. > > This is all in-kernel, so no syscalls. Since I doubt you need sub-ns > granularity for kvmtrace, can we do without an arch hook? Yes. I just talked with our kernel maintainers and we do not want another special case for s390. If ktime_get is fine for blktrace, scheduler and almost any other kernel core component - it is fine for tracing. I also talked to Christian Ehrhardt and he will send an updated patch set soon. Christian From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christian Borntraeger Subject: Re: [PATCH 2/5] kvmtrace: make cycle calculation architecture aware Date: Mon, 14 Jul 2008 09:44:42 +0200 Message-ID: <200807140944.42360.borntraeger@de.ibm.com> References: <1215439013-11480-1-git-send-email-ehrhardt@linux.vnet.ibm.com> <48770CEF.6090102@de.ibm.com> <487A2210.5020602@qumranet.com> Mime-Version: 1.0 Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 7bit Cc: carsteno@de.ibm.com, "Yang, Sheng" , kvm@vger.kernel.org, Hollis Blanchard , Christian Ehrhardt , xiantao.zhang@intel.com, kvm-ppc@vger.kernel.org To: Avi Kivity Return-path: Received: from mtagate4.uk.ibm.com ([195.212.29.137]:58537 "EHLO mtagate4.uk.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754821AbYGNHoy (ORCPT ); Mon, 14 Jul 2008 03:44:54 -0400 In-Reply-To: <487A2210.5020602@qumranet.com> Content-Disposition: inline Sender: kvm-owner@vger.kernel.org List-ID: Am Sonntag, 13. Juli 2008 schrieb Avi Kivity: > Carsten Otte wrote: > > Avi Kivity wrote: > >> What is the overhead of ktime_get()? > > I think I'd like an arch specific timestamp. This way we could use our > > clock-cycle-granularity-non-privileged-timestamp instruction ;-). If > > we need a common implementation, I don't think there's much difference > > between different syscalls in terms of overhead. > > This is all in-kernel, so no syscalls. Since I doubt you need sub-ns > granularity for kvmtrace, can we do without an arch hook? Yes. I just talked with our kernel maintainers and we do not want another special case for s390. If ktime_get is fine for blktrace, scheduler and almost any other kernel core component - it is fine for tracing. I also talked to Christian Ehrhardt and he will send an updated patch set soon. Christian