From mboxrd@z Thu Jan  1 00:00:00 1970
From: Ed White <edmund.h.white@intel.com>
Subject: Re: [PATCH v2 06/12] VMX: add VMFUNC leaf 0 (EPTP
	switching) to emulator.
Date: Wed, 24 Jun 2015 13:29:34 -0700
Message-ID: <558B132E.4020701@intel.com>
References: <1434999372-3688-1-git-send-email-edmund.h.white@intel.com>
	<1434999372-3688-7-git-send-email-edmund.h.white@intel.com>
	<558AA6EB.4060900@citrix.com>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Return-path: <xen-devel-bounces@lists.xen.org>
In-Reply-To: <558AA6EB.4060900@citrix.com>
List-Unsubscribe: <http://lists.xen.org/cgi-bin/mailman/options/xen-devel>,
	<mailto:xen-devel-request@lists.xen.org?subject=unsubscribe>
List-Post: <mailto:xen-devel@lists.xen.org>
List-Help: <mailto:xen-devel-request@lists.xen.org?subject=help>
List-Subscribe: <http://lists.xen.org/cgi-bin/mailman/listinfo/xen-devel>,
	<mailto:xen-devel-request@lists.xen.org?subject=subscribe>
Sender: xen-devel-bounces@lists.xen.org
Errors-To: xen-devel-bounces@lists.xen.org
To: Andrew Cooper <andrew.cooper3@citrix.com>, xen-devel@lists.xen.org
Cc: Ravi Sahita <ravi.sahita@intel.com>, Wei Liu <wei.liu2@citrix.com>, Ian Jackson <ian.jackson@eu.citrix.com>, Tim Deegan <tim@xen.org>, Jan Beulich <jbeulich@suse.com>, tlengyel@novetta.com, Daniel De Graaf <dgdegra@tycho.nsa.gov>
List-Id: xen-devel@lists.xenproject.org

On 06/24/2015 05:47 AM, Andrew Cooper wrote:
>> +    case EXIT_REASON_VMFUNC:
>> +        if ( vmx_vmfunc_intercept(regs) == X86EMUL_OKAY )
> 
> This is currently an unconditional failure, and I don't see subsequent
> patches which alter vmx_vmfunc_intercept().  Shouldn't
> vmx_vmfunc_intercept() switch on eax and optionally call
> p2m_switch_vcpu_altp2m_by_id()?

If the VMFUNC instruction was valid, the hardware would have executed it.
The only time a VMFUNC exit occurs is if the hardware supports VMFUNC
and the hypervisor has enabled it, but the VMFUNC instruction is
invalid in some way and can't be executed (because EAX != 0, for example).

There are only two choices: crash the domain or inject #UD (which is the
closest analogue to what happens in the absence of a hypervisor and will
probably crash the OS in the domain). I chose the latter in the code I
originally wrote; Ravi chose the former in his patch. I don't have a
strong opinion either way, but I think these are the only two choices.

I hope this answers Jan's question in another email on the same subject.

Ed