From mboxrd@z Thu Jan 1 00:00:00 1970 From: Avi Kivity Subject: Re: [PATCH] kvm: notify host when guest paniced Date: Wed, 29 Feb 2012 12:46:55 +0200 Message-ID: <4F4E021F.80206@redhat.com> References: <4F4AF1FB.6000903@cn.fujitsu.com> <4F4CB926.6050600@redhat.com> <4F4D7F5E.5040202@cn.fujitsu.com> <4F4DF4C6.90609@redhat.com> <20120229095842.GF5050@redhat.com> <4F4DF86C.5010407@redhat.com> <4F4DFE7C.3030501@cn.fujitsu.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Cc: qemu-devel , KAMEZAWA Hiroyuki , kvm list , linux-kernel@vger.kernel.org To: Wen Congyang Return-path: In-Reply-To: <4F4DFE7C.3030501@cn.fujitsu.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: qemu-devel-bounces+gceq-qemu-devel=gmane.org@nongnu.org Sender: qemu-devel-bounces+gceq-qemu-devel=gmane.org@nongnu.org List-Id: kvm.vger.kernel.org On 02/29/2012 12:31 PM, Wen Congyang wrote: > At 02/29/2012 06:05 PM, Avi Kivity Wrote: > > On 02/29/2012 11:58 AM, Daniel P. Berrange wrote: > >>> > >>> How about using a virtio-serial channel for this? You can transfer any > >>> amount of information (including the dump itself). > >> > >> When the guest OS has crashed, any dumps will be done from the host > >> OS using libvirt's core dump mechanism. The guest OS isn't involved > >> and is likely too dead to be of any use anyway. Likewise it is > >> quite probably too dead to work a virtio-serial channel or any > >> similarly complex device. We're really just after the simplest > >> possible notification that the guest kernel has paniced. > > > > If it's alive enough to panic, it's alive enough to kexec its kdump > > kernel. After that it can do anything. > > > > Guest-internal dumps are more useful IMO that host-initiated dumps. In > > Yes, guest-internal dump is better than host dump. But the user may not > start guest-internal dump or guest-internal dump failed. So we need the > following feature: > 1. If the guest-internal dump does not work, the guest's status is 'crashed'. > And then the user does the host dump. > 2. If the guest-internal dump is working, the guest's status should be > 'dumping'. The user see this status and know the guest has paniced, and > the guest-internal dump is working. I agree. There is room for host dump, and we do want notification about what the guest is doing. The question is whether we should reuse virtio-serial for guest-host communication in this case. It's more complicated, but allows us to avoid touching the hypervisor. -- error compiling committee.c: too many arguments to function