[PATCH] KVM: x86: fix 32 bit build

[PATCH] KVM: x86: fix 32 bit build

[Maxim Levitsky]

Now that kvm->stat.nx_lpage_splits is 64 bit, use DIV_ROUND_UP_ULL
when doing division.

Fixes: 7ee093d4f3f5 ("KVM: switch per-VM stats to u64")
Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
---
 arch/x86/kvm/mmu/mmu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
index 720ceb0a1f5c..97372225f183 100644
--- a/arch/x86/kvm/mmu/mmu.c
+++ b/arch/x86/kvm/mmu/mmu.c
@@ -6054,7 +6054,7 @@ static void kvm_recover_nx_lpages(struct kvm *kvm)
 	write_lock(&kvm->mmu_lock);
 
 	ratio = READ_ONCE(nx_huge_pages_recovery_ratio);
-	to_zap = ratio ? DIV_ROUND_UP(kvm->stat.nx_lpage_splits, ratio) : 0;
+	to_zap = ratio ? DIV_ROUND_UP_ULL(kvm->stat.nx_lpage_splits, ratio) : 0;
 	for ( ; to_zap; --to_zap) {
 		if (list_empty(&kvm->arch.lpage_disallowed_mmu_pages))
 			break;
-- 
2.26.3

Re: [PATCH] KVM: x86: fix 32 bit build

[Sean Christopherson]

On Wed, Jun 16, 2021, Maxim Levitsky wrote:
> Now that kvm->stat.nx_lpage_splits is 64 bit, use DIV_ROUND_UP_ULL
> when doing division.

I went the "cast to an unsigned long" route.  I prefer the cast approach because
to_zap is also an unsigned long, i.e. using DIV_ROUND_UP_ULL() could look like a
truncation bug.  In practice, nx_lpage_splits can't be more than an unsigned long
so it's largely a moot point, I just like the more explicit "this is doing
something odd".

https://lkml.kernel.org/r/20210615162905.2132937-1-seanjc@google.com

> Fixes: 7ee093d4f3f5 ("KVM: switch per-VM stats to u64")
> Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
> ---
>  arch/x86/kvm/mmu/mmu.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
> index 720ceb0a1f5c..97372225f183 100644
> --- a/arch/x86/kvm/mmu/mmu.c
> +++ b/arch/x86/kvm/mmu/mmu.c
> @@ -6054,7 +6054,7 @@ static void kvm_recover_nx_lpages(struct kvm *kvm)
>  	write_lock(&kvm->mmu_lock);
>  
>  	ratio = READ_ONCE(nx_huge_pages_recovery_ratio);
> -	to_zap = ratio ? DIV_ROUND_UP(kvm->stat.nx_lpage_splits, ratio) : 0;
> +	to_zap = ratio ? DIV_ROUND_UP_ULL(kvm->stat.nx_lpage_splits, ratio) : 0;
>  	for ( ; to_zap; --to_zap) {
>  		if (list_empty(&kvm->arch.lpage_disallowed_mmu_pages))
>  			break;
> -- 
> 2.26.3
> 

Re: [PATCH] KVM: x86: fix 32 bit build

[Maxim Levitsky]

On Wed, 2021-06-16 at 15:59 +0000, Sean Christopherson wrote:
> On Wed, Jun 16, 2021, Maxim Levitsky wrote:
> > Now that kvm->stat.nx_lpage_splits is 64 bit, use DIV_ROUND_UP_ULL
> > when doing division.
> 
> I went the "cast to an unsigned long" route.  I prefer the cast approach because
> to_zap is also an unsigned long, i.e. using DIV_ROUND_UP_ULL() could look like a
> truncation bug.  In practice, nx_lpage_splits can't be more than an unsigned long
> so it's largely a moot point, I just like the more explicit "this is doing
> something odd".
> 
> https://lkml.kernel.org/r/20210615162905.2132937-1-seanjc@google.com
> 
> > Fixes: 7ee093d4f3f5 ("KVM: switch per-VM stats to u64")
> > Signed-off-by: Maxim Levitsky <mlevitsk@redhat.com>
> > ---
> >  arch/x86/kvm/mmu/mmu.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
> > index 720ceb0a1f5c..97372225f183 100644
> > --- a/arch/x86/kvm/mmu/mmu.c
> > +++ b/arch/x86/kvm/mmu/mmu.c
> > @@ -6054,7 +6054,7 @@ static void kvm_recover_nx_lpages(struct kvm *kvm)
> >  	write_lock(&kvm->mmu_lock);
> >  
> >  	ratio = READ_ONCE(nx_huge_pages_recovery_ratio);
> > -	to_zap = ratio ? DIV_ROUND_UP(kvm->stat.nx_lpage_splits, ratio) : 0;
> > +	to_zap = ratio ? DIV_ROUND_UP_ULL(kvm->stat.nx_lpage_splits, ratio) : 0;
> >  	for ( ; to_zap; --to_zap) {
> >  		if (list_empty(&kvm->arch.lpage_disallowed_mmu_pages))
> >  			break;
> > -- 
> > 2.26.3
> > 
Cool, makes sense.

I didn't notice your patch (I did look at the list but
since the subject didn't mention the build breakage I didn't notice).

I just wanted to send this patch to avoid someone else
spending time figuring it out.

Thanks,
Best regards,
	Maxim Levitsky