KVM: x86: Fix uninitialized eoi_exit_bitmap usage in vcpu_load_eoi_exitmap()
author黄乐 <huangle1@jd.com>
Mon, 15 Nov 2021 14:08:29 +0000 (14:08 +0000)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 25 Nov 2021 08:48:43 +0000 (09:48 +0100)
commit c5adbb3af051079f35abfa26551107e2c653087f upstream.

In vcpu_load_eoi_exitmap(), currently the eoi_exit_bitmap[4] array is
initialized only when Hyper-V context is available, in other path it is
just passed to kvm_x86_ops.load_eoi_exitmap() directly from on the stack,
which would cause unexpected interrupt delivery/handling issues, e.g. an
*old* linux kernel that relies on PIT to do clock calibration on KVM might
randomly fail to boot.

Fix it by passing ioapic_handled_vectors to load_eoi_exitmap() when Hyper-V
context is not available.

Fixes: f2bc14b69c38 ("KVM: x86: hyper-v: Prepare to meet unallocated Hyper-V context")
Cc: stable@vger.kernel.org
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Huang Le <huangle1@jd.com>
Message-Id: <62115b277dab49ea97da5633f8522daf@jd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/x86/kvm/x86.c

index 66e4e9a..16c7465 100644 (file)
@@ -9429,12 +9429,16 @@ static void vcpu_load_eoi_exitmap(struct kvm_vcpu *vcpu)
        if (!kvm_apic_hw_enabled(vcpu->arch.apic))
                return;
 
-       if (to_hv_vcpu(vcpu))
+       if (to_hv_vcpu(vcpu)) {
                bitmap_or((ulong *)eoi_exit_bitmap,
                          vcpu->arch.ioapic_handled_vectors,
                          to_hv_synic(vcpu)->vec_bitmap, 256);
+               static_call(kvm_x86_load_eoi_exitmap)(vcpu, eoi_exit_bitmap);
+               return;
+       }
 
-       static_call(kvm_x86_load_eoi_exitmap)(vcpu, eoi_exit_bitmap);
+       static_call(kvm_x86_load_eoi_exitmap)(
+               vcpu, (u64 *)vcpu->arch.ioapic_handled_vectors);
 }
 
 void kvm_arch_mmu_notifier_invalidate_range(struct kvm *kvm,