From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alexander Graf Date: Wed, 30 Apr 2014 10:06:38 +0000 Subject: Re: [PATCH 2/6] KVM: PPC: Book3S PR: Emulate TIR register Message-Id: <5360CB2E.50104@suse.de> List-Id: References: <1398788262-3307-1-git-send-email-agraf@suse.de> <1398788262-3307-3-git-send-email-agraf@suse.de> <19971.1398837074@ale.ozlabs.ibm.com> In-Reply-To: <19971.1398837074@ale.ozlabs.ibm.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: Michael Neuling Cc: kvm-ppc@vger.kernel.org, kvm@vger.kernel.org On 30.04.14 07:51, Michael Neuling wrote: >> In parallel to the Processor ID Register (PIR) threaded POWER8 also adds a >> Thread ID Register (TID). Since PR KVM doesn't emulate more than one thread > s/TID/TIR/ above Oops :) > >> per core, we can just always expose 0 here. > I'm not sure if we ever do, but if we IPI ourselves using a doorbell, > we'll need to emulate the doorbell as well. Yeah, we don't. I'm surprised we don't myself, but I'd rather not implement the logic to emulate doorbells until I can easily verify it works - which means either have a Linux guest that does doorbells or have a test harness for kernel level test cases. Alex From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alexander Graf Subject: Re: [PATCH 2/6] KVM: PPC: Book3S PR: Emulate TIR register Date: Wed, 30 Apr 2014 12:06:38 +0200 Message-ID: <5360CB2E.50104@suse.de> References: <1398788262-3307-1-git-send-email-agraf@suse.de> <1398788262-3307-3-git-send-email-agraf@suse.de> <19971.1398837074@ale.ozlabs.ibm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: kvm-ppc@vger.kernel.org, kvm@vger.kernel.org To: Michael Neuling Return-path: In-Reply-To: <19971.1398837074@ale.ozlabs.ibm.com> Sender: kvm-ppc-owner@vger.kernel.org List-Id: kvm.vger.kernel.org On 30.04.14 07:51, Michael Neuling wrote: >> In parallel to the Processor ID Register (PIR) threaded POWER8 also adds a >> Thread ID Register (TID). Since PR KVM doesn't emulate more than one thread > s/TID/TIR/ above Oops :) > >> per core, we can just always expose 0 here. > I'm not sure if we ever do, but if we IPI ourselves using a doorbell, > we'll need to emulate the doorbell as well. Yeah, we don't. I'm surprised we don't myself, but I'd rather not implement the logic to emulate doorbells until I can easily verify it works - which means either have a Linux guest that does doorbells or have a test harness for kernel level test cases. Alex