Re: [PATCH v6 07/13] KVM: s390: introduce struct kvm_s390_gisa_interrupt

[Halil Pasic <pasic@linux.ibm.com>]

On Tue, 29 Jan 2019 16:47:10 +0100
Michael Mueller <mimu@linux.ibm.com> wrote:

> 
> 
> On 29.01.19 14:22, Cornelia Huck wrote:
> > On Mon, 28 Jan 2019 17:50:54 +0100
> > Halil Pasic <pasic@linux.ibm.com> wrote:
> > 
> >> On Thu, 24 Jan 2019 13:59:33 +0100
> >> Michael Mueller <mimu@linux.ibm.com> wrote:
> >>
> >>> Use this struct analog to the kvm interruption structs
> >>> for kvm emulated floating and local interruptions.
> >>> Further fields will be added with this series as
> >>> required.
> >>>
> >>> Signed-off-by: Michael Mueller <mimu@linux.ibm.com>
> >>
> >> While looking at this I was asking myself what guards against invalid
> >> gisa pointer dereference e.g. when pending_irqs() is called (see below).
> >>
> >> AFAIU we set up gisa_int.origin only if we have
> >> css_general_characteristics.aiv. Opinions?
> > 
> > I think you're right that this is a (pre-existing) problem.
> > 
> >>> diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c
> >>> index 942cc7d33766..ee91d1de0036 100644
> >>> --- a/arch/s390/kvm/interrupt.c
> >>> +++ b/arch/s390/kvm/interrupt.c
> >>> @@ -246,7 +246,8 @@ static inline unsigned long pending_irqs_no_gisa(struct kvm_vcpu *vcpu)
> >>>   static inline unsigned long pending_irqs(struct kvm_vcpu *vcpu)
> >>>   {
> >>>   	return pending_irqs_no_gisa(vcpu) |
> >>> -		gisa_get_ipm(vcpu->kvm->arch.gisa) << IRQ_PEND_IO_ISC_7;
> >>> +		gisa_get_ipm(vcpu->kvm->arch.gisa_int.origin) <<
> >>
> >> Unconditional call to gisa_get_ipm(), and get ipm just accesses ->ipm.
> > 
> > All other callers of this function check for gisa != NULL first, so
> > either we should check here as well or move the check into the
> > gisa_get_ipm() function.
> 
> I suggest to use an explicit test like this.
> 
>   static inline unsigned long pending_irqs(struct kvm_vcpu *vcpu)
>   {
> -       return pending_irqs_no_gisa(vcpu) |
> -               gisa_get_ipm(vcpu->kvm->arch.gisa_int.origin) <<
> -                       IRQ_PEND_IO_ISC_7;
> +       struct kvm_s390_gisa_int *gi = &vcpu->kvm->arch.gisa_int;
> +       unsigned long pending_mask;
> +
> +       pending_mask = pending_irqs_no_gisa(vcpu);
> +       if (gi->origin)
> +               pending_mask |= gisa_get_ipm(gi->origin) << 
> IRQ_PEND_IO_ISC_7;
> +       return pending_mask;
>   }
> 

Works with me! Send a stand alone patch?

Regards,
Halil