From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from rune.pobox.com (rune.pobox.com [208.210.124.79]) by ozlabs.org (Postfix) with ESMTP id 9757067B73 for ; Thu, 15 Jun 2006 08:40:55 +1000 (EST) Date: Wed, 14 Jun 2006 17:40:40 -0500 From: Nathan Lynch To: Paul Mackerras Subject: Re: [PATCH 2/3] powerpc: Instrument Hypervisor Calls: add wrappers Message-ID: <20060614224040.GH26750@localdomain> References: <20060614034756.GA6759@monkey.ibm.com> <20060614035258.GC6759@monkey.ibm.com> <20060614144230.GE26750@localdomain> <17552.36563.460775.131555@cargo.ozlabs.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <17552.36563.460775.131555@cargo.ozlabs.ibm.com> Cc: Bryan Rosenburg , Christopher Yeoh , linuxppc-dev@ozlabs.org List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Paul Mackerras wrote: > Nathan Lynch writes: > > > Without disabling preemption around the mfspr ... update_stats section > > in these hcall wrappers, you risk updating the stats on the wrong cpu. > > I think we are only looking for total counts and times anyway, so it > doesn't really matter which cpu updates the stats, as long as the time > gets accounted on some cpu. The use of per-cpu counters is just for > better cache behaviour. Makes sense... then perhaps the stats shouldn't be exported in per-cpu files in sysfs, and the kernel should calculate and report the totals to userspace in some other form.