Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Tue, 16 May 2023 08:54:36 -0700
In-Reply-To: <20230516122159.3834044-1-dengqiao.joey@bytedance.com>
Mime-Version: 1.0
References: <20230516122159.3834044-1-dengqiao.joey@bytedance.com>
Message-ID: <ZGOnPMTEKqRq89jt@google.com>
Subject: Re: [PATCH] KVM: SVM: Update destination when updating pi irte
From:   Sean Christopherson <seanjc@google.com>
To:     "dengqiao.joey" <dengqiao.joey@bytedance.com>
Cc:     pbonzini@redhat.com, kvm@vger.kernel.org,
        linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="us-ascii"
Precedence: bulk

On Tue, May 16, 2023, dengqiao.joey wrote:
> Destination of irte will be cleard by IOMMU driver when updating irte.
> It will only be set correctly in vcpu_load. IOMMU will deliver the
> doorbell message to the wrong physical cpu before vcpu_load is executed.
> That means vcpu can not recognize interrupt delivery during the time of
> non-root mode.

I suspect this is actually a bug in amd_iommu_activate_guest_mode().  Does the
below fix the issue?  If so, can you give a Tested-by and/or Reviewed-by on the
posted patch[*]?

diff --git a/drivers/iommu/amd/iommu.c b/drivers/iommu/amd/iommu.c
index 5a505ba5467e..fbe77ee2d26c 100644
--- a/drivers/iommu/amd/iommu.c
+++ b/drivers/iommu/amd/iommu.c
@@ -3484,8 +3484,7 @@ int amd_iommu_activate_guest_mode(void *data)
 	struct irte_ga *entry = (struct irte_ga *) ir_data->entry;
 	u64 valid;
 
-	if (!AMD_IOMMU_GUEST_IR_VAPIC(amd_iommu_guest_ir) ||
-	    !entry || entry->lo.fields_vapic.guest_mode)
+	if (!AMD_IOMMU_GUEST_IR_VAPIC(amd_iommu_guest_ir) || !entry)
 		return 0;
 
 	valid = entry->lo.fields_vapic.valid;

[*] https://lore.kernel.org/all/20230419201154.83880-2-joao.m.martins@oracle.com