From: Vitaly Kuznetsov <vkuznets@redhat.com>
To: Sean Christopherson <seanjc@google.com>
Cc: kvm@vger.kernel.org, Paolo Bonzini <pbonzini@redhat.com>,
linux-kernel@vger.kernel.org, Wanpeng Li <wanpengli@tencent.com>,
Jim Mattson <jmattson@google.com>,
Maxim Levitsky <mlevitsk@redhat.com>
Subject: Re: [PATCH 3/7] KVM: nVMX: Ignore 'hv_clean_fields' data when eVMCS data is copied in vmx_get_nested_state()
Date: Wed, 12 May 2021 09:14:56 +0200 [thread overview]
Message-ID: <87eeecwhhr.fsf@vitty.brq.redhat.com> (raw)
In-Reply-To: <YJqytyu7+Q7+bqeG@google.com>
Sean Christopherson <seanjc@google.com> writes:
> On Tue, May 11, 2021, Vitaly Kuznetsov wrote:
>> 'Clean fields' data from enlightened VMCS is only valid upon vmentry: L1
>> hypervisor is not obliged to keep it up-to-date while it is mangling L2's
>> state, KVM_GET_NESTED_STATE request may come at a wrong moment when actual
>> eVMCS changes are unsynchronized with 'hv_clean_fields'. As upon migration
>> VMCS12 is used as a source of ultimate truce, we must make sure we pick all
>> the changes to eVMCS and thus 'clean fields' data must be ignored.
>>
>> Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
>> ---
>> arch/x86/kvm/vmx/nested.c | 43 +++++++++++++++++++++++----------------
>> 1 file changed, 25 insertions(+), 18 deletions(-)
>>
>> diff --git a/arch/x86/kvm/vmx/nested.c b/arch/x86/kvm/vmx/nested.c
>> index ea2869d8b823..7970a16ee6b1 100644
>> --- a/arch/x86/kvm/vmx/nested.c
>> +++ b/arch/x86/kvm/vmx/nested.c
>> @@ -1607,16 +1607,23 @@ static void copy_vmcs12_to_shadow(struct vcpu_vmx *vmx)
>> vmcs_load(vmx->loaded_vmcs->vmcs);
>> }
>>
>> -static int copy_enlightened_to_vmcs12(struct vcpu_vmx *vmx)
>> +static int copy_enlightened_to_vmcs12(struct vcpu_vmx *vmx, bool from_vmentry)
>> {
>> struct vmcs12 *vmcs12 = vmx->nested.cached_vmcs12;
>> struct hv_enlightened_vmcs *evmcs = vmx->nested.hv_evmcs;
>> + u32 hv_clean_fields;
>>
>> /* HV_VMX_ENLIGHTENED_CLEAN_FIELD_NONE */
>> vmcs12->tpr_threshold = evmcs->tpr_threshold;
>> vmcs12->guest_rip = evmcs->guest_rip;
>>
>> - if (unlikely(!(evmcs->hv_clean_fields &
>> + /* Clean fields data can only be trusted upon vmentry */
>> + if (likely(from_vmentry))
>> + hv_clean_fields = evmcs->hv_clean_fields;
>> + else
>> + hv_clean_fields = 0;
>
> ...
>
>> @@ -3503,7 +3510,7 @@ static int nested_vmx_run(struct kvm_vcpu *vcpu, bool launch)
>> return nested_vmx_failInvalid(vcpu);
>>
>> if (vmx->nested.hv_evmcs) {
>> - copy_enlightened_to_vmcs12(vmx);
>> + copy_enlightened_to_vmcs12(vmx, true);
>
> Rather than pass a bool, what about having the caller explicitly specify the
> clean fields? Then the migration path can have a comment about needing to
> assume all fields are dirty, and the normal path would be self-documenting.
> E.g. with evmcs captured in a local var:
>
> if (evmcs) {
> copy_enlightened_to_vmcs12(vmx, evmcs->hv_clean_fields);
> } else if (...) {
> }
>
I like the idea, thanks! Will incorporate into v2.
>> /* Enlightened VMCS doesn't have launch state */
>> vmcs12->launch_state = !launch;
>> } else if (enable_shadow_vmcs) {
>> @@ -6136,7 +6143,7 @@ static int vmx_get_nested_state(struct kvm_vcpu *vcpu,
>> copy_vmcs02_to_vmcs12_rare(vcpu, get_vmcs12(vcpu));
>> if (!vmx->nested.need_vmcs12_to_shadow_sync) {
>> if (vmx->nested.hv_evmcs)
>> - copy_enlightened_to_vmcs12(vmx);
>> + copy_enlightened_to_vmcs12(vmx, false);
>> else if (enable_shadow_vmcs)
>> copy_shadow_to_vmcs12(vmx);
>> }
>> --
>> 2.30.2
>>
>
--
Vitaly
next prev parent reply other threads:[~2021-05-12 7:15 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-05-11 11:19 [PATCH 0/7] KVM: nVMX: Fixes for nested state migration when eVMCS is in use Vitaly Kuznetsov
2021-05-11 11:19 ` [PATCH 1/7] KVM: nVMX: Introduce nested_evmcs_is_used() Vitaly Kuznetsov
2021-05-11 11:19 ` [PATCH 2/7] KVM: nVMX: Release enlightened VMCS on VMCLEAR Vitaly Kuznetsov
2021-05-11 11:19 ` [PATCH 3/7] KVM: nVMX: Ignore 'hv_clean_fields' data when eVMCS data is copied in vmx_get_nested_state() Vitaly Kuznetsov
2021-05-11 16:37 ` Sean Christopherson
2021-05-12 7:14 ` Vitaly Kuznetsov [this message]
2021-05-11 11:19 ` [PATCH 4/7] KVM: nVMX: Force enlightened VMCS sync from nested_vmx_failValid() Vitaly Kuznetsov
2021-05-11 11:19 ` [PATCH 5/7] KVM: nVMX: Reset eVMCS clean fields data from prepare_vmcs02() Vitaly Kuznetsov
2021-05-11 16:32 ` Sean Christopherson
2021-05-12 7:19 ` Vitaly Kuznetsov
2021-05-11 11:19 ` [PATCH 6/7] KVM: nVMX: Request to sync eVMCS from VMCS12 after migration Vitaly Kuznetsov
2021-05-11 11:19 ` [PATCH 7/7] KVM: selftests: evmcs_test: Test that KVM_STATE_NESTED_EVMCS is never lost Vitaly Kuznetsov
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=87eeecwhhr.fsf@vitty.brq.redhat.com \
--to=vkuznets@redhat.com \
--cc=jmattson@google.com \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=mlevitsk@redhat.com \
--cc=pbonzini@redhat.com \
--cc=seanjc@google.com \
--cc=wanpengli@tencent.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).