Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
Date:   Wed, 30 Mar 2022 21:59:22 +0000
From:   Sean Christopherson <seanjc@google.com>
To:     "Maciej S. Szmigiero" <mail@maciej.szmigiero.name>
Cc:     Paolo Bonzini <pbonzini@redhat.com>,
        Vitaly Kuznetsov <vkuznets@redhat.com>,
        Wanpeng Li <wanpengli@tencent.com>,
        Jim Mattson <jmattson@google.com>,
        Joerg Roedel <joro@8bytes.org>,
        Tom Lendacky <thomas.lendacky@amd.com>,
        Brijesh Singh <brijesh.singh@amd.com>,
        Jon Grimm <Jon.Grimm@amd.com>,
        David Kaplan <David.Kaplan@amd.com>,
        Boris Ostrovsky <boris.ostrovsky@oracle.com>,
        Liam Merwick <liam.merwick@oracle.com>, kvm@vger.kernel.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH 3/5] KVM: nSVM: Don't forget about L1-injected events
Message-ID: <YkTSul0CbYi/ae0t@google.com>
References: <cover.1646944472.git.maciej.szmigiero@oracle.com>
 <a28577564a7583c32f0029f2307f63ca8869cf22.1646944472.git.maciej.szmigiero@oracle.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <a28577564a7583c32f0029f2307f63ca8869cf22.1646944472.git.maciej.szmigiero@oracle.com>
Precedence: bulk

On Thu, Mar 10, 2022, Maciej S. Szmigiero wrote:
> From: "Maciej S. Szmigiero" <maciej.szmigiero@oracle.com>
> 
> In SVM synthetic software interrupts or INT3 or INTO exception that L1
> wants to inject into its L2 guest are forgotten if there is an intervening
> L0 VMEXIT during their delivery.
> 
> They are re-injected correctly with VMX, however.
> 
> This is because there is an assumption in SVM that such exceptions will be
> re-delivered by simply re-executing the current instruction.
> Which might not be true if this is a synthetic exception injected by L1,
> since in this case the re-executed instruction will be one already in L2,
> not the VMRUN instruction in L1 that attempted the injection.
> 
> Leave the pending L1 -> L2 event in svm->nested.ctl.event_inj{,err} until
> it is either re-injected successfully or returned to L1 upon a nested
> VMEXIT.
> Make sure to always re-queue such event if returned in EXITINTINFO.
> 
> The handling of L0 -> {L1, L2} event re-injection is left as-is to avoid
> unforeseen regressions.
> 
> Signed-off-by: Maciej S. Szmigiero <maciej.szmigiero@oracle.com>
> ---

...

> @@ -3627,6 +3632,14 @@ static void svm_complete_interrupts(struct kvm_vcpu *vcpu)
>  	if (!(exitintinfo & SVM_EXITINTINFO_VALID))
>  		return;
>  
> +	/* L1 -> L2 event re-injection needs a different handling */
> +	if (is_guest_mode(vcpu) &&
> +	    exit_during_event_injection(svm, svm->nested.ctl.event_inj,
> +					svm->nested.ctl.event_inj_err)) {
> +		nested_svm_maybe_reinject(vcpu);

Why is this manually re-injecting?  More specifically, why does the below (out of
sight in the diff) code that re-queues the exception/interrupt not work?  The
re-queued event should be picked up by nested_save_pending_event_to_vmcb12() and
propagatred to vmcb12.