From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paolo Bonzini Subject: Re: [PATCH 3/4] KVM: i8254: remove unnecessary irqchip_in_kernel check Date: Thu, 30 Jul 2015 09:25:01 +0200 Message-ID: <55B9D14D.1040600@redhat.com> References: <1438176538-26191-1-git-send-email-pbonzini@redhat.com> <1438176538-26191-4-git-send-email-pbonzini@redhat.com> <20150730043645.GF15229@google.com> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Cc: linux-kernel@vger.kernel.org, kvm@vger.kernel.org, mtosatti@redhat.com, rkrcmar@redhat.com To: Steve Rutherford Return-path: In-Reply-To: <20150730043645.GF15229@google.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: kvm.vger.kernel.org On 30/07/2015 06:36, Steve Rutherford wrote: > On Wed, Jul 29, 2015 at 03:28:57PM +0200, Paolo Bonzini wrote: >> > The PIT is only created if irqchip_in_kernel returns true, so the >> > check is superfluous. > I poked around. Looks to me like the existence of an IOAPIC is not > checked on the creation of the in-kernel PIT. You're right, and presumably it's also legal to create the PIT before KVM_CREATE_IRQCHIP. > Userspace might limit itself to > that scenario (PIT implies IOAPIC in-kernel), but that isn't enforced at PIT > creation. I'll play with the "PIT without IOAPIC" scenario and see if something breaks badly with this patch. From reading the code it seems like it should not introduce any problems (oopses or similar), but I'll set this patch aside for now. Paolo