From: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
To: Maxim Levitsky <mlevitsk@redhat.com>,
linux-kernel@vger.kernel.org, kvm@vger.kernel.org
Cc: pbonzini@redhat.com, seanjc@google.com, joro@8bytes.org,
jon.grimm@amd.com, wei.huang2@amd.com, terry.bowman@amd.com,
kernel test robot <lkp@intel.com>
Subject: Re: [PATCH v4 10/15] KVM: SVM: Introduce helper functions to (de)activate AVIC and x2AVIC
Date: Wed, 11 May 2022 22:37:31 +0700 [thread overview]
Message-ID: <3fa12834-d144-54d8-0bf8-8a72e726db99@amd.com> (raw)
In-Reply-To: <a60d885cf4b0b11aca730273ff317546362bff83.camel@redhat.com>
Maxim,
On 5/9/22 8:42 PM, Maxim Levitsky wrote:
>...
>
> So I did some testing, and reviewed this code again with regard to nesting,
> and now I see that it has CVE worthy bug, so have to revoke my Reviewed-By.
>
> This is what happens:
>
> On nested VM entry, *request to inhibit AVIC is done*, and then nested msr bitmap
> is calculated, still with all X2AVIC msrs open,
>
> 1. nested_svm_vmrun -> enter_svm_guest_mode -> kvm_make_request(KVM_REQ_APICV_UPDATE, vcpu);
> 2. nested_svm_vmrun -> nested_svm_vmrun_msrpm
>
> But the nested guest will be entered without AVIC active
> (since we don't yet support nested avic and it is optional anyway), thus if the nested guest
> also doesn't intercept those msrs, it will gain access to the *host* x2apic msrs. Ooops.
Shouldn't this be changed to intercept the x2APIC msrs because of the following logic?
kvm_make_request(KVM_REQ_APICV_UPDATE, vcpu)
kvm_vcpu_update_apicv(vcpu)
static_call(kvm_x86_refresh_apicv_exec_ctrl)(vcpu)
avic_deactivate_vmcb()
svm_set_x2apic_msr_interception(true)
> I think the easist way to fix this for now, is to make nested_svm_vmrun_msrpm
> never open access to x2apic msrs regardless of the host bitmap value, but in the long
> term the whole thing needs to be refactored.
Agree.
> Another thing I noted is that avic_deactivate_vmcb should not touch avic msrs
> when avic_mode == AVIC_MODE_X1, it is just a waste of time.
We can add the check.
> Also updating these msr intercepts is pointless if the guest doesn't use x2apic.
We can also add the check.
Best Regards,
Suravee
next prev parent reply other threads:[~2022-05-11 15:37 UTC|newest]
Thread overview: 30+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-05-08 2:39 [PATCH v4 00/15] Introducing AMD x2AVIC and hybrid-AVIC modes Suravee Suthikulpanit
2022-05-08 2:39 ` [PATCH v4 01/15] x86/cpufeatures: Introduce x2AVIC CPUID bit Suravee Suthikulpanit
2022-05-08 2:39 ` [PATCH v4 02/15] KVM: x86: lapic: Rename [GET/SET]_APIC_DEST_FIELD to [GET/SET]_XAPIC_DEST_FIELD Suravee Suthikulpanit
2022-05-09 14:27 ` Gupta, Pankaj
2022-05-08 2:39 ` [PATCH v4 03/15] KVM: SVM: Detect X2APIC virtualization (x2AVIC) support Suravee Suthikulpanit
2022-05-09 10:15 ` Maxim Levitsky
2022-05-08 2:39 ` [PATCH v4 04/15] KVM: SVM: Update max number of vCPUs supported for x2AVIC mode Suravee Suthikulpanit
2022-05-09 13:35 ` Gupta, Pankaj
2022-05-08 2:39 ` [PATCH v4 05/15] KVM: SVM: Update avic_kick_target_vcpus to support 32-bit APIC ID Suravee Suthikulpanit
2022-05-08 2:39 ` [PATCH v4 06/15] KVM: SVM: Do not support updating APIC ID when in x2APIC mode Suravee Suthikulpanit
2022-05-08 2:39 ` [PATCH v4 07/15] KVM: SVM: Adding support for configuring x2APIC MSRs interception Suravee Suthikulpanit
2022-05-08 2:39 ` [PATCH v4 08/15] KVM: x86: Deactivate APICv on vCPU with APIC disabled Suravee Suthikulpanit
2022-05-09 10:18 ` Maxim Levitsky
2022-05-09 10:30 ` Maxim Levitsky
2022-05-09 10:19 ` Maxim Levitsky
2022-05-08 2:39 ` [PATCH v4 09/15] KVM: SVM: Refresh AVIC configuration when changing APIC mode Suravee Suthikulpanit
2022-05-09 10:22 ` Maxim Levitsky
2022-05-08 2:39 ` [PATCH v4 10/15] KVM: SVM: Introduce helper functions to (de)activate AVIC and x2AVIC Suravee Suthikulpanit
2022-05-09 13:42 ` Maxim Levitsky
2022-05-11 15:37 ` Suravee Suthikulpanit [this message]
2022-05-11 16:26 ` Maxim Levitsky
2022-05-08 2:39 ` [PATCH v4 11/15] KVM: SVM: Do not throw warning when calling avic_vcpu_load on a running vcpu Suravee Suthikulpanit
2022-05-09 13:40 ` Gupta, Pankaj
2022-05-08 2:39 ` [PATCH v4 12/15] KVM: SVM: Introduce hybrid-AVIC mode Suravee Suthikulpanit
2022-05-09 11:23 ` Suthikulpanit, Suravee
2022-05-08 2:39 ` [PATCH v4 13/15] KVM: x86: Warning APICv inconsistency only when vcpu APIC mode is valid Suravee Suthikulpanit
2022-05-09 13:38 ` Gupta, Pankaj
2022-05-08 2:39 ` [PATCH v4 14/15] KVM: SVM: Use target APIC ID to complete x2AVIC IRQs when possible Suravee Suthikulpanit
2022-05-08 2:39 ` [PATCH v4 15/15] KVM: SVM: Add AVIC doorbell tracepoint Suravee Suthikulpanit
2022-05-09 10:28 ` [PATCH v4 00/15] Introducing AMD x2AVIC and hybrid-AVIC modes Maxim Levitsky
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=3fa12834-d144-54d8-0bf8-8a72e726db99@amd.com \
--to=suravee.suthikulpanit@amd.com \
--cc=jon.grimm@amd.com \
--cc=joro@8bytes.org \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=lkp@intel.com \
--cc=mlevitsk@redhat.com \
--cc=pbonzini@redhat.com \
--cc=seanjc@google.com \
--cc=terry.bowman@amd.com \
--cc=wei.huang2@amd.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).