linux-kernel - Re: [PATCH] KVM: x86: Fully initialize 'struct kvm_lapic_irq' in kvm_pv_kick_cpu

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Fri, 08 Jul 2022 14:44:21 +0200
From:   Vitaly Kuznetsov <vkuznets@...hat.com>
To:     Sean Christopherson <seanjc@...gle.com>
Cc:     kvm@...r.kernel.org, Paolo Bonzini <pbonzini@...hat.com>,
        Wanpeng Li <wanpengli@...cent.com>,
        Jim Mattson <jmattson@...gle.com>,
        linux-hyperv@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] KVM: x86: Fully initialize 'struct kvm_lapic_irq' in
 kvm_pv_kick_cpu_op()

Sean Christopherson <seanjc@...gle.com> writes:

> On Tue, Jun 28, 2022, Vitaly Kuznetsov wrote:
>> 'vector' and 'trig_mode' fields of 'struct kvm_lapic_irq' are left
>> uninitialized in kvm_pv_kick_cpu_op(). While these fields are normally
>> not needed for APIC_DM_REMRD, they're still referenced by
>> __apic_accept_irq() for trace_kvm_apic_accept_irq(). Fully initialize
>> the structure to avoid consuming random stack memory.
>> 
>> Fixes: a183b638b61c ("KVM: x86: make apic_accept_irq tracepoint more generic")
>> Reported-by: syzbot+d6caa905917d353f0d07@...kaller.appspotmail.com
>> Signed-off-by: Vitaly Kuznetsov <vkuznets@...hat.com>
>> ---
>>  arch/x86/kvm/x86.c | 18 ++++++++++--------
>>  1 file changed, 10 insertions(+), 8 deletions(-)
>> 
>> diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
>> index 567d13405445..8a98608dad4f 100644
>> --- a/arch/x86/kvm/x86.c
>> +++ b/arch/x86/kvm/x86.c
>> @@ -9340,15 +9340,17 @@ static int kvm_pv_clock_pairing(struct kvm_vcpu *vcpu, gpa_t paddr,
>>   */
>>  static void kvm_pv_kick_cpu_op(struct kvm *kvm, int apicid)
>>  {
>> -	struct kvm_lapic_irq lapic_irq;
>> -
>> -	lapic_irq.shorthand = APIC_DEST_NOSHORT;
>> -	lapic_irq.dest_mode = APIC_DEST_PHYSICAL;
>> -	lapic_irq.level = 0;
>> -	lapic_irq.dest_id = apicid;
>> -	lapic_irq.msi_redir_hint = false;
>> +	struct kvm_lapic_irq lapic_irq = {
>> +		.vector = 0,
>> +		.delivery_mode = APIC_DM_REMRD,
>> +		.dest_mode = APIC_DEST_PHYSICAL,
>> +		.level = false,
>> +		.trig_mode = 0,
>> +		.shorthand = APIC_DEST_NOSHORT,
>> +		.dest_id = apicid,
>> +		.msi_redir_hint = false
>> +	};
>
> What if we rely on the compiler to zero-initialize omitted fields?  E.g.
>
> 	/*
> 	 * All other fields are unused for APIC_DM_REMRD, but may be consumed by
> 	 * common code, e.g. for tracing.  Defer initialization to the compiler.
> 	 */
> 	struct kvm_lapic_irq lapic_irq = {
> 		.delivery_mode = APIC_DM_REMRD,
> 		.dest_mode = APIC_DEST_PHYSICAL,
> 		.shorthand = APIC_DEST_NOSHORT,
> 		.dest_id = apicid,
> 	};
>
> KVM doesn't actually care about the vector, level, trig_mode, etc... for its magic
> magic DM_REMRD, i.e. using 0/false is completely arbitrary.  

Honestly, I just always forget whether zero-initializing of omitted
fields is a de-facto or de-jure standard. We don't care much whether
these fields are zero of not but some random values in tracing may
confuse an unprepared reader.

I'll send v2 with your amendment.

>
>>  
>> -	lapic_irq.delivery_mode = APIC_DM_REMRD;
>>  	kvm_irq_delivery_to_apic(kvm, NULL, &lapic_irq, NULL);
>>  }
>>  
>> -- 
>> 2.35.3
>> 
>

-- 
Vitaly