Re: [PATCH] KVM: x86: Initialize tdp_level during vCPU creation

From: Paolo Bonzini <pbonzini@redhat.com>
To: Sean Christopherson <sean.j.christopherson@intel.com>
Cc: Vitaly Kuznetsov <vkuznets@redhat.com>,
	Wanpeng Li <wanpengli@tencent.com>,
	Jim Mattson <jmattson@google.com>, Joerg Roedel <joro@8bytes.org>,
	kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
	syzbot+904752567107eefb728c@syzkaller.appspotmail.com
Subject: Re: [PATCH] KVM: x86: Initialize tdp_level during vCPU creation
Date: Wed, 27 May 2020 18:56:19 +0200	[thread overview]
Message-ID: <078365dd-64ff-4f3c-813c-3d9fc955ed1a@redhat.com> (raw)
In-Reply-To: <20200527162933.GE24461@linux.intel.com>

On 27/05/20 18:29, Sean Christopherson wrote:
> On Wed, May 27, 2020 at 06:17:57PM +0200, Paolo Bonzini wrote:
>> On 27/05/20 10:54, Sean Christopherson wrote:
>>> Initialize vcpu->arch.tdp_level during vCPU creation to avoid consuming
>>> garbage if userspace calls KVM_RUN without first calling KVM_SET_CPUID.
>>>
>>> Fixes: e93fd3b3e89e9 ("KVM: x86/mmu: Capture TDP level when updating CPUID")
>>> Reported-by: syzbot+904752567107eefb728c@syzkaller.appspotmail.com
>>> Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
>>> ---
>>>  arch/x86/kvm/x86.c | 1 +
>>>  1 file changed, 1 insertion(+)
>>>
>>> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
>>> index b226fb8abe41b..01a6304056197 100644
>>> --- a/arch/x86/kvm/x86.c
>>> +++ b/arch/x86/kvm/x86.c
>>> @@ -9414,6 +9414,7 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
>>>  	fx_init(vcpu);
>>>  
>>>  	vcpu->arch.maxphyaddr = cpuid_query_maxphyaddr(vcpu);
>>> +	vcpu->arch.tdp_level = kvm_x86_ops.get_tdp_level(vcpu);
>>>  
>>>  	vcpu->arch.pat = MSR_IA32_CR_PAT_DEFAULT;
>>
>> Queued, it is probably a good idea to add a selftests testcase for this
>> (it's even okay if it doesn't use the whole selftests infrastructure and
>> invokes KVM_CREATE_VM/KVM_CREATE_VCPU/KVM_RUN manually).
> 
> Ya.  syzbot is hitting a #GP due to NULL pointer during debugfs on the exact
> same sequence.  I haven't been able to reproduce that one (have yet to try
> syzbot's exact config), but it's another example of a "dumb" test hitting
> meaningful bugs.

Saw that, it's mine. :)

Paolo