From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christian Borntraeger Subject: Re: [PATCH] KVM: disable halt_poll_ns as default for s390x Date: Fri, 18 Sep 2015 13:35:49 +0200 Message-ID: <55FBF715.9030902@de.ibm.com> References: <1442572493-51400-1-git-send-email-dahi@linux.vnet.ibm.com> <55FBED5E.4060600@de.ibm.com> <55FBF5B2.6090907@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-15 Content-Transfer-Encoding: 7bit Cc: wanpeng.li@hotmail.com, dmatlack@google.com, rkrcmar@redhat.com To: Paolo Bonzini , David Hildenbrand , kvm@vger.kernel.org Return-path: Received: from e06smtp15.uk.ibm.com ([195.75.94.111]:56990 "EHLO e06smtp15.uk.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750853AbbIRLfy (ORCPT ); Fri, 18 Sep 2015 07:35:54 -0400 Received: from /spool/local by e06smtp15.uk.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Fri, 18 Sep 2015 12:35:52 +0100 Received: from b06cxnps4074.portsmouth.uk.ibm.com (d06relay11.portsmouth.uk.ibm.com [9.149.109.196]) by d06dlp02.portsmouth.uk.ibm.com (Postfix) with ESMTP id 9E3E6219005C for ; Fri, 18 Sep 2015 12:35:21 +0100 (BST) Received: from d06av05.portsmouth.uk.ibm.com (d06av05.portsmouth.uk.ibm.com [9.149.37.229]) by b06cxnps4074.portsmouth.uk.ibm.com (8.14.9/8.14.9/NCO v10.0) with ESMTP id t8IBZoD241877558 for ; Fri, 18 Sep 2015 11:35:50 GMT Received: from d06av05.portsmouth.uk.ibm.com (localhost [127.0.0.1]) by d06av05.portsmouth.uk.ibm.com (8.14.4/8.14.4/NCO v10.0 AVout) with ESMTP id t8IBZnWc027022 for ; Fri, 18 Sep 2015 05:35:50 -0600 In-Reply-To: <55FBF5B2.6090907@redhat.com> Sender: kvm-owner@vger.kernel.org List-ID: Am 18.09.2015 um 13:29 schrieb Paolo Bonzini: > > > On 18/09/2015 12:54, Christian Borntraeger wrote: >>> -/* halt polling only reduces halt latency by 5-7 us, 500us is enough */ >>> -static unsigned int halt_poll_ns = 500000; >>> +/* Architectures should define their poll value according to the halt latency */ >>> +static unsigned int halt_poll_ns = KVM_HALT_POLL_NS_DEFAULT; >> >> Yes, I prefer this over disabling it via Kconfig. There are benchmarks which >> benefit from polling on s390. Furthermore it seems that the latency >> strongly depends on timing of the architecture so making it per arch is >> probably the right thing to do. > > Perhaps a #ifndef is better than replicating the 500us default in all > architectures? Or should the default be 0? I prefer to not have an ifdef in .c files. I would assume that over time architectures will provide their own number, e.g. for my uperf cases that got better, 50000 was enough. 500000 just increased cpu overhead. Christian