Subject: Re: [PATCH 6/8] kvm: svm: pass MSR_IA32_SPEC_CTRL and
 MSR_IA32_PRED_CMD down to guest
To: Tom Lendacky <thomas.lendacky@amd.com>,
        linux-kernel@vger.kernel.org, kvm@vger.kernel.org
Cc: rkrcmar@redhat.com, liran.alon@oracle.com, jmattson@google.com,
        aliguori@amazon.com, dwmw@amazon.co.uk, bp@alien8.de,
        x86@kernel.org
References: <20180109120311.27565-1-pbonzini@redhat.com>
 <20180109120311.27565-7-pbonzini@redhat.com>
 <18c3f6e4-84af-87c8-3441-d8b368ade1c8@amd.com>
From: Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <0337f197-a31d-db90-1563-363404298849@redhat.com>
Date: Thu, 11 Jan 2018 11:33:14 +0100
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101
 Thunderbird/52.5.0
MIME-Version: 1.0
In-Reply-To: <18c3f6e4-84af-87c8-3441-d8b368ade1c8@amd.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org

On 10/01/2018 21:13, Tom Lendacky wrote:
> On 1/9/2018 6:03 AM, Paolo Bonzini wrote:
>> Direct access to MSR_IA32_SPEC_CTRL and MSR_IA32_PRED_CMD is important
>> for performance.  Allow load/store of MSR_IA32_SPEC_CTRL, restore guest
>> IBRS on VM entry and set it to 0 on VM exit (because Linux does not use
>> it yet).
>>
>> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
>> ---
>>  arch/x86/kvm/svm.c | 42 ++++++++++++++++++++++++++++++++++++++++++
>>  1 file changed, 42 insertions(+)
>>
>> diff --git a/arch/x86/kvm/svm.c b/arch/x86/kvm/svm.c
>> index 31ace8d7774a..934a21e02e03 100644
>> --- a/arch/x86/kvm/svm.c
>> +++ b/arch/x86/kvm/svm.c
> 
> ...
> 
>> @@ -5015,6 +5045,18 @@ static void svm_vcpu_run(struct kvm_vcpu *vcpu)
>>  #endif
>>  		);
>>  
>> +	if (have_spec_ctrl) {
>> +		rdmsrl(MSR_IA32_SPEC_CTRL, svm->spec_ctrl);
>> +		if (svm->spec_ctrl != 0)
>> +			wrmsrl(MSR_IA32_SPEC_CTRL, 0);
>> +	}
>> +	/*
>> +	 * Speculative execution past the above wrmsrl might encounter
>> +	 * an indirect branch and use guest-controlled contents of the
>> +	 * indirect branch predictor; block it.
>> +	 */
>> +	asm("lfence");
> 
> This will end up needing to be an alternative macro based on the
> LFENCE_RDTSC or MFENCE_RDTSC features [1].  You'll probably just want to
> use the speculation barrier macro that ends up being defined to control
> the speculation here.

Yes.  Though, is there any processor that has spec_ctrl (which is none
on AMD) but needs a full fence?

Paolo

> Thanks,
> Tom
> 
> [1] https://marc.info/?l=linux-kernel&m=151545930207815&w=2
> 
>> +
>>  #ifdef CONFIG_X86_64
>>  	wrmsrl(MSR_GS_BASE, svm->host.gs_base);
>>  #else
>>