Re: [PATCH] KVM: VMX: replace move_msr_up with swap macro

[Gustavo A. R. Silva]

Hi Paolo,

Quoting Paolo Bonzini <pbonzini@xxxxxxxxxx>:

> ----- Original Message -----
> > From: "Gustavo A. R. Silva" <garsilva@xxxxxxxxxxxxxx>
> > To: "Paolo Bonzini" <pbonzini@xxxxxxxxxx>, "Radim KrÄmÃÅ"  
> > <rkrcmar@xxxxxxxxxx>, "Thomas Gleixner"
> > <tglx@xxxxxxxxxxxxx>, "Ingo Molnar" <mingo@xxxxxxxxxx>, "H. Peter  
> > Anvin" <hpa@xxxxxxxxx>, x86@xxxxxxxxxx
> > Cc: kvm@xxxxxxxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx, "Gustavo A.  
> > R. Silva" <garsilva@xxxxxxxxxxxxxx>
> > Sent: Friday, November 3, 2017 11:58:19 PM
> > Subject: [PATCH] KVM: VMX: replace move_msr_up with swap macro
> >
> > Function move_msr_up is used to _manually_ swap MSR entries in MSR array.
> > This function can be removed and replaced using the swap macro instead.
> >
> > This code was detected with the help of Coccinelle.
>
> I think move_msr_up should instead change into a function like
>
>    void mark_msr_for_save(struct vcpu_vmx *vmx, int index)
>    {
>        swap(vmx->guest_msrs[index], vmx->guest_msrs[vmx->save_nmsrs]);
>        vmx->save_nmsrs++;
>    }
>
> Using swap is useful, but it is also hiding what's going on exactly
> (in addition, using ++ inside a macro argument might be calling for
> trouble).

Thanks for your comments.

I'll work on v2 based on your feedback.

--
Gustavo A. R. Silva

> Paolo
>
> > Signed-off-by: Gustavo A. R. Silva <garsilva@xxxxxxxxxxxxxx>
> > ---
> > The new lines are over 80 characters, but I think in this case that is
> > preferable over splitting them.
> >
> >  arch/x86/kvm/vmx.c | 24 ++++++------------------
> >  1 file changed, 6 insertions(+), 18 deletions(-)
> >
> > diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
> > index e6c8ffa..210e491 100644
> > --- a/arch/x86/kvm/vmx.c
> > +++ b/arch/x86/kvm/vmx.c
> > @@ -2544,18 +2544,6 @@ static bool vmx_invpcid_supported(void)
> >  	return cpu_has_vmx_invpcid() && enable_ept;
> >  }
> >
> > -/*
> > - * Swap MSR entry in host/guest MSR entry array.
> > - */
> > -static void move_msr_up(struct vcpu_vmx *vmx, int from, int to)
> > -{
> > -	struct shared_msr_entry tmp;
> > -
> > -	tmp = vmx->guest_msrs[to];
> > -	vmx->guest_msrs[to] = vmx->guest_msrs[from];
> > -	vmx->guest_msrs[from] = tmp;
> > -}
> > -
> >  static void vmx_set_msr_bitmap(struct kvm_vcpu *vcpu)
> >  {
> >  	unsigned long *msr_bitmap;
> > @@ -2600,28 +2588,28 @@ static void setup_msrs(struct vcpu_vmx *vmx)
> >  	if (is_long_mode(&vmx->vcpu)) {
> >  		index = __find_msr_index(vmx, MSR_SYSCALL_MASK);
> >  		if (index >= 0)
> > -			move_msr_up(vmx, index, save_nmsrs++);
> > +			swap(vmx->guest_msrs[index], vmx->guest_msrs[save_nmsrs++]);
> >  		index = __find_msr_index(vmx, MSR_LSTAR);
> >  		if (index >= 0)
> > -			move_msr_up(vmx, index, save_nmsrs++);
> > +			swap(vmx->guest_msrs[index], vmx->guest_msrs[save_nmsrs++]);
> >  		index = __find_msr_index(vmx, MSR_CSTAR);
> >  		if (index >= 0)
> > -			move_msr_up(vmx, index, save_nmsrs++);
> > +			swap(vmx->guest_msrs[index], vmx->guest_msrs[save_nmsrs++]);
> >  		index = __find_msr_index(vmx, MSR_TSC_AUX);
> >  		if (index >= 0 && guest_cpuid_has(&vmx->vcpu, X86_FEATURE_RDTSCP))
> > -			move_msr_up(vmx, index, save_nmsrs++);
> > +			swap(vmx->guest_msrs[index], vmx->guest_msrs[save_nmsrs++]);
> >  		/*
> >  		 * MSR_STAR is only needed on long mode guests, and only
> >  		 * if efer.sce is enabled.
> >  		 */
> >  		index = __find_msr_index(vmx, MSR_STAR);
> >  		if ((index >= 0) && (vmx->vcpu.arch.efer & EFER_SCE))
> > -			move_msr_up(vmx, index, save_nmsrs++);
> > +			swap(vmx->guest_msrs[index], vmx->guest_msrs[save_nmsrs++]);
> >  	}
> >  #endif
> >  	index = __find_msr_index(vmx, MSR_EFER);
> >  	if (index >= 0 && update_transition_efer(vmx, index))
> > -		move_msr_up(vmx, index, save_nmsrs++);
> > +		swap(vmx->guest_msrs[index], vmx->guest_msrs[save_nmsrs++]);
> >
> >  	vmx->save_nmsrs = save_nmsrs;
>
>
> > --
> > 2.7.4