From mboxrd@z Thu Jan 1 00:00:00 1970 From: Raghavendra K T Subject: Re: [PATCH RFC V4 0/5] kvm : Paravirt-spinlock support for KVM guests Date: Wed, 25 Jan 2012 23:15:36 +0530 Message-ID: <4F203FC0.70907@linux.vnet.ibm.com> References: <20120114182501.8604.68416.sendpatchset@oc5400248562.ibm.com> <3EC1B881-0724-49E3-B892-F40BEB07D15D@suse.de> <20120116142014.GA10155@linux.vnet.ibm.com> <4F146EA5.3010106@linux.vnet.ibm.com> <4F15AF9E.9000907@linux.vnet.ibm.com> <1485A122-9D48-46E3-A01E-E37B5C9EC54A@suse.de> <4F15BFAE.7060500@linux.vnet.ibm.com> <4F1FC370.5020506@linux.vnet.ibm.com> <20120125163552.GB23999@phenom.dumpdata.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; Format="flowed" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20120125163552.GB23999@phenom.dumpdata.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: virtualization-bounces@lists.linux-foundation.org Errors-To: virtualization-bounces@lists.linux-foundation.org To: Konrad Rzeszutek Wilk Cc: Jeremy Fitzhardinge , Greg Kroah-Hartman , linux-doc@vger.kernel.org, Peter Zijlstra , Jan Kiszka , Srivatsa Vaddagiri , Paul Mackerras , "H. Peter Anvin" , Xen , Dave Jiang , KVM , Glauber Costa , X86 , Ingo Molnar , Avi Kivity , Rik van Riel , Stefano Stabellini , Sasha Levin , Sedat Dilek , Thomas Gleixner , Virtualization , LKML , Dave Hansen , Suzuki Poulose List-Id: virtualization@lists.linuxfoundation.org On 01/25/2012 10:05 PM, Konrad Rzeszutek Wilk wrote: >> So it seems we have worst case overhead of around 8%. But we see >> improvement of at-least 15% once when little more time is spent in >> critical section. > > Is this with collecting the histogram information about spinlocks? We found > that if you enable that for production runs it makes them quite slower. > Ok. Are you referring to CONFIG_KVM_DEBUG_FS/CONFIG_XEN_DEBUG_FS?. Not it was not enabled. But then may be I was not precise. This test was only on native. So it should have not affected IMO. It is nice to know that histogram affects, since I was always under the impression that it does not affect much on guest too. My experiments had almost always enabled that. Let me know if you referred to something else.. From mboxrd@z Thu Jan 1 00:00:00 1970 From: Raghavendra K T Subject: Re: [PATCH RFC V4 0/5] kvm : Paravirt-spinlock support for KVM guests Date: Wed, 25 Jan 2012 23:15:36 +0530 Message-ID: <4F203FC0.70907@linux.vnet.ibm.com> References: <20120114182501.8604.68416.sendpatchset@oc5400248562.ibm.com> <3EC1B881-0724-49E3-B892-F40BEB07D15D@suse.de> <20120116142014.GA10155@linux.vnet.ibm.com> <4F146EA5.3010106@linux.vnet.ibm.com> <4F15AF9E.9000907@linux.vnet.ibm.com> <1485A122-9D48-46E3-A01E-E37B5C9EC54A@suse.de> <4F15BFAE.7060500@linux.vnet.ibm.com> <4F1FC370.5020506@linux.vnet.ibm.com> <20120125163552.GB23999@phenom.dumpdata.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; Format="flowed" Content-Transfer-Encoding: 7bit Cc: Jeremy Fitzhardinge , Greg Kroah-Hartman , linux-doc@vger.kernel.org, Peter Zijlstra , Jan Kiszka , Srivatsa Vaddagiri , Paul Mackerras , "H. Peter Anvin" , Xen , Dave Jiang , KVM , Glauber Costa , X86 , Ingo Molnar , Avi Kivity , Rik van Riel , Stefano Stabellini , Sasha Levin , Sedat Dilek , Thomas Gleixner , Virtualization , LKML , Dave Hansen , Suzuki Poulose Return-path: In-Reply-To: <20120125163552.GB23999@phenom.dumpdata.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: virtualization-bounces@lists.linux-foundation.org Errors-To: virtualization-bounces@lists.linux-foundation.org List-Id: kvm.vger.kernel.org On 01/25/2012 10:05 PM, Konrad Rzeszutek Wilk wrote: >> So it seems we have worst case overhead of around 8%. But we see >> improvement of at-least 15% once when little more time is spent in >> critical section. > > Is this with collecting the histogram information about spinlocks? We found > that if you enable that for production runs it makes them quite slower. > Ok. Are you referring to CONFIG_KVM_DEBUG_FS/CONFIG_XEN_DEBUG_FS?. Not it was not enabled. But then may be I was not precise. This test was only on native. So it should have not affected IMO. It is nice to know that histogram affects, since I was always under the impression that it does not affect much on guest too. My experiments had almost always enabled that. Let me know if you referred to something else..