From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christian Borntraeger Subject: Re: [PATCH] kvm: notify host when guest panicked Date: Tue, 12 Jun 2012 09:49:28 +0200 Message-ID: <4FD6F488.7090303@de.ibm.com> References: <4FB9E4CE.8060709@cn.fujitsu.com> <4FD6E5EC.3090802@cn.fujitsu.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: kvm list , Avi Kivity , qemu-devel , "linux-kernel@vger.kernel.org" , "Daniel P. Berrange" , KAMEZAWA Hiroyuki , Jan Kiszka , Gleb Natapov To: Wen Congyang Return-path: In-Reply-To: <4FD6E5EC.3090802@cn.fujitsu.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: kvm.vger.kernel.org >> We have three solutions to implement this feature: >> 1. use vmcall >> 2. use I/O port >> 3. use virtio-serial. > >> We have decided to avoid touching hypervisor. The reason why I choose >> choose the I/O port is: >> 1. it is easier to implememt >> 2. it does not depend any virtual device >> 3. it can work when startint the kernel Havent looked deeply into that, butz wouldnt kvm_hypercall0 also fulfill all of these points? You just have to define the number for each architecture and then it should be possible to write a simple panic driver that works on all architectures. >> --- a/include/linux/kvm_para.h >> +++ b/include/linux/kvm_para.h >> @@ -20,6 +20,14 @@ >> #define KVM_HC_FEATURES 3 >> #define KVM_HC_PPC_MAP_MAGIC_PAGE 4 >> >> +#define KVM_PV_PORT (0x505UL) >> + ignoring the above, shouldnt a port number go into an architecture specific file? Christian From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([208.118.235.92]:59971) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1SeLrE-0006k9-Ru for qemu-devel@nongnu.org; Tue, 12 Jun 2012 03:49:50 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1SeLr8-0000Qw-PG for qemu-devel@nongnu.org; Tue, 12 Jun 2012 03:49:44 -0400 Received: from e06smtp12.uk.ibm.com ([195.75.94.108]:57430) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1SeLr8-0000Oj-Gk for qemu-devel@nongnu.org; Tue, 12 Jun 2012 03:49:38 -0400 Received: from /spool/local by e06smtp12.uk.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Tue, 12 Jun 2012 08:49:33 +0100 Received: from d06av02.portsmouth.uk.ibm.com (d06av02.portsmouth.uk.ibm.com [9.149.37.228]) by d06nrmr1307.portsmouth.uk.ibm.com (8.13.8/8.13.8/NCO v10.0) with ESMTP id q5C7nUaT1138816 for ; Tue, 12 Jun 2012 08:49:30 +0100 Received: from d06av02.portsmouth.uk.ibm.com (loopback [127.0.0.1]) by d06av02.portsmouth.uk.ibm.com (8.14.4/8.13.1/NCO v10.0 AVout) with ESMTP id q5C7nSov024526 for ; Tue, 12 Jun 2012 01:49:29 -0600 Message-ID: <4FD6F488.7090303@de.ibm.com> Date: Tue, 12 Jun 2012 09:49:28 +0200 From: Christian Borntraeger MIME-Version: 1.0 References: <4FB9E4CE.8060709@cn.fujitsu.com> <4FD6E5EC.3090802@cn.fujitsu.com> In-Reply-To: <4FD6E5EC.3090802@cn.fujitsu.com> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH] kvm: notify host when guest panicked List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Wen Congyang Cc: Gleb Natapov , kvm list , Jan Kiszka , "linux-kernel@vger.kernel.org" , qemu-devel , Avi Kivity , KAMEZAWA Hiroyuki >> We have three solutions to implement this feature: >> 1. use vmcall >> 2. use I/O port >> 3. use virtio-serial. > >> We have decided to avoid touching hypervisor. The reason why I choose >> choose the I/O port is: >> 1. it is easier to implememt >> 2. it does not depend any virtual device >> 3. it can work when startint the kernel Havent looked deeply into that, butz wouldnt kvm_hypercall0 also fulfill all of these points? You just have to define the number for each architecture and then it should be possible to write a simple panic driver that works on all architectures. >> --- a/include/linux/kvm_para.h >> +++ b/include/linux/kvm_para.h >> @@ -20,6 +20,14 @@ >> #define KVM_HC_FEATURES 3 >> #define KVM_HC_PPC_MAP_MAGIC_PAGE 4 >> >> +#define KVM_PV_PORT (0x505UL) >> + ignoring the above, shouldnt a port number go into an architecture specific file? Christian