From mboxrd@z Thu Jan 1 00:00:00 1970 From: Zhang Yanfei Subject: Re: [PATCH 0/3] x86: clear vmcss on all cpus when doing kdump if necessary Date: Fri, 19 Oct 2012 12:51:00 +0800 Message-ID: <5080DC34.1080503@cn.fujitsu.com> References: <5077BB7B.4030203@cn.fujitsu.com> <507C2F18.8080904@redhat.com> <507E17C0.9070305@cn.fujitsu.com> <507E8575.4040908@redhat.com> <507F5762.3060208@cn.fujitsu.com> <507FE020.7040805@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: x86@kernel.org, kexec@lists.infradead.org, linux-doc@vger.kernel.org, mtosatti@redhat.com, linux-kernel@vger.kernel.org, "kvm@vger.kernel.org" , "Hao, Xudong" , "Nakajima, Jun" To: Avi Kivity Return-path: In-Reply-To: <507FE020.7040805@redhat.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: kvm.vger.kernel.org =E4=BA=8E 2012=E5=B9=B410=E6=9C=8818=E6=97=A5 18:55, Avi Kivity =E5=86=99= =E9=81=93: > On 10/18/2012 03:12 AM, Zhang Yanfei wrote: >> =E4=BA=8E 2012=E5=B9=B410=E6=9C=8817=E6=97=A5 18:16, Avi Kivity =E5=86= =99=E9=81=93: >>> On 10/17/2012 04:28 AM, Zhang Yanfei wrote: >>>> =E4=BA=8E 2012=E5=B9=B410=E6=9C=8815=E6=97=A5 23:43, Avi Kivity =E5= =86=99=E9=81=93: >>>>> On 10/12/2012 08:40 AM, Zhang Yanfei wrote: >>>>>> Currently, kdump just makes all the logical processors leave VMX= operation by >>>>>> executing VMXOFF instruction, so any VMCSs active on the logical= processors may >>>>>> be corrupted. But, sometimes, we need the VMCSs to debug guest i= mages contained >>>>>> in the host vmcore. To prevent the corruption, we should VMCLEAR= the VMCSs before >>>>>> executing the VMXOFF instruction. >>>>> >>>>> How have you verified that VMXOFF doesn't flush cached VMCSs alre= ady? >>>>> >>>> >>>> I tried some tests, for example, I made copies for every vmcs, and= in the kdump >>>> path, I backed up all the loaded vmcs into the copies before vmxof= f. >>>> After generating the vmcore, I retrieve the vmcss and their copies= , and compare them, >>>> no differences. >>>> >>>> Another test is using VMCLEAR to clear all the loaded vmcs before = VMXOFF, >>>> and compare the vmcss and their copies, there are indeed differenc= es between the >>>> vmcs and its copy. >>>> >>>> I know the tests may be not so convincing, for example, I used mem= cpy to back up >>>> the vmcss and it is an ordinary memory operation. But to ensure th= e non-corruption >>>> of the vmcss in the vmcore, I think we should VMCLEAR the vmcss be= fore VMXOFF just >>>> as the Intel spec says. >>> >>> Sorry, I was unclear -- I was referring to the spec, I wasn't sure >>> whether VMXOFF is defined to flush VMCSes or whether it just invali= dates >>> on-chip caches so that it won't flush them out in the future, corru= pting >>> memory. We don't want to depend on actual behaviour as it may chan= ge >>> with future version. >>> >>> Copying some Intel folk, maybe they can clarify it. >>> >> >> Yes, the Intel spec says "may be" about the VMCS-corruption thing. F= rom >> chapter 24.10.1 in Intel=C2=AE 64 and IA-32 Architectures Software D= eveloper=E2=80=99s >> Manual Volume 3C:System Programming Guide, Part 3, there is the desc= ription: >> >> "If a logical processor leaves VMX operation, any VMCSs active on th= at logical >> processor may be corrupted (see below). To prevent such corruption o= f a VMCS that >> may be used either after a return to VMX operation or on another log= ical processor, >> software should VMCLEAR that VMCS before executing the VMXOFF instru= ction or >> removing power from the processor (e.g., as part of a transition to = the S3 and S4 >> power states)." >> >> Our purpose is to make sure the VMCSs in the vmcore are updated and = non-corrupted. So >> according to the description above, no matter whether VMXOFF is defi= ned to flush >> VMCSs or whether it just invalidates on-chip caches, we'd better VMC= LEAR the >> VMCSs before executing the VMXOFF. >=20 > Ok, that's clear then. So all we need is to remove the sysctl and cl= ear > VMCSs unconditionally. >=20 OK, I'll make the new patch and resend it again. Thanks Zhang Yanfei