From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: [PATCH 08/28] kvm tools: Fix KVM_RUN exit code check Date: Thu, 22 Dec 2011 12:21:47 +0200 Message-ID: <4EF304BB.3010808@redhat.com> References: <4EDD8E82.1010909@ozlabs.org> <1323159768.3882.9.camel@lappy> <4EE028E5.7080207@ozlabs.org> <1323322304.3904.6.camel@lappy> <4EF30088.6090606@redhat.com> <1324549080.3878.1.camel@lappy> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: Matt Evans , kvm@vger.kernel.org, kvm-ppc@vger.kernel.org To: Sasha Levin Return-path: In-Reply-To: <1324549080.3878.1.camel@lappy> Sender: kvm-ppc-owner@vger.kernel.org List-Id: kvm.vger.kernel.org On 12/22/2011 12:18 PM, Sasha Levin wrote: > On Thu, 2011-12-22 at 12:03 +0200, Avi Kivity wrote: > > On 12/08/2011 07:31 AM, Sasha Levin wrote: > > > > > > > > It sounds like Alex is considering KVM PPC's return value in this case (and > > > > updating api.txt if appropriate) -- what say you on this patch? It actually > > > > brings kvmtool's KVM_RUN return val check in line with QEMU's (also "< 0") and > > > > nothing PPC will run without it, currently. (I'm about to repost a new series, > > > > will include it for these reasons, until I hear more complaint ;) ) > > > > > > '<0' is fine as it's what api.txt says :) > > > > What? ioctls return -1 on error, not <0. > > <0 as opposed to the !=0 check we had there before. > > Theres no harm in checking for <0 even if the only possible negative > result is -1. > > Yes, but the documentation should say -1. Which ioctl is this? -- error compiling committee.c: too many arguments to function