Re: [PATCH v4] KVM: x86: Fix KVM_GET_CPUID2 ioctl to return cpuid entries count

From: Paolo Bonzini <pbonzini@redhat.com>
To: Sean Christopherson <seanjc@google.com>,
	Valeriy Vdovin <valeriy.vdovin@virtuozzo.com>
Cc: linux-kernel@vger.kernel.org, Denis Lunev <den@openvz.org>,
	Jonathan Corbet <corbet@lwn.net>,
	Vitaly Kuznetsov <vkuznets@redhat.com>,
	Wanpeng Li <wanpengli@tencent.com>,
	Jim Mattson <jmattson@google.com>, Joerg Roedel <joro@8bytes.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
	x86@kernel.org, "H. Peter Anvin" <hpa@zytor.com>,
	Shuah Khan <shuah@kernel.org>,
	Aaron Lewis <aaronlewis@google.com>,
	Alexander Graf <graf@amazon.com>,
	Andrew Jones <drjones@redhat.com>,
	Oliver Upton <oupton@google.com>,
	Like Xu <like.xu@linux.intel.com>,
	kvm@vger.kernel.org, linux-doc@vger.kernel.org,
	linux-kselftest@vger.kernel.org
Subject: Re: [PATCH v4] KVM: x86: Fix KVM_GET_CPUID2 ioctl to return cpuid entries count
Date: Mon, 3 May 2021 16:54:26 +0200	[thread overview]
Message-ID: <0d68dbc3-8462-7763-fbad-f3b895fcf6e6@redhat.com> (raw)
In-Reply-To: <YIoFFl72VSeuhCRt@google.com>

On 29/04/21 03:00, Sean Christopherson wrote:
> On Wed, Apr 28, 2021, Valeriy Vdovin wrote:
>> It's very explicit by the code that it was designed to receive some
>> small number of entries to return E2BIG along with the corrected number.
> 
> LOL, saying KVM_GET_CPUID2 was "designed" is definitely giving the KVM
> forefathers the benefit of the doubt.

I was going to make a different joke, i.e. that KVM_GET_CPUID2 was 
indeed designed the way Valeriy described, but that design was forgotten 
soon after.

Really, this ioctl has been such a trainwreck that I think the only good 
solution here is to drop it.

Paolo

>> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
>> index efc7a82ab140..3f941b1f4e78 100644
>> --- a/arch/x86/kvm/x86.c
>> +++ b/arch/x86/kvm/x86.c
>> @@ -4773,14 +4773,17 @@ long kvm_arch_vcpu_ioctl(struct file *filp,
>>   		r = -EFAULT;
>>   		if (copy_from_user(&cpuid, cpuid_arg, sizeof(cpuid)))
>>   			goto out;
>> +
>>   		r = kvm_vcpu_ioctl_get_cpuid2(vcpu, &cpuid,
>>   					      cpuid_arg->entries);
>> -		if (r)
>> +
>> +		if (r && r != -E2BIG)
>>   			goto out;
>> -		r = -EFAULT;
>> -		if (copy_to_user(cpuid_arg, &cpuid, sizeof(cpuid)))
>> +
>> +		if (copy_to_user(cpuid_arg, &cpuid, sizeof(cpuid))) {
>> +			r = -EFAULT;
>>   			goto out;
> 
> As I pointed out[*], copying the number of entries but not the entries themselves
> is wrong.  All of my feedback on v1 still stands.
> 
> [*] https://lkml.kernel.org/r/YIl4M/GgaYvwNuXv@google.com
> 
>> -		r = 0;
>> +		}
>>   		break;
>>   	}
>>   	case KVM_GET_MSRS: {
> 