Re: [PATCH 2/3] KVM: MTRR: simplify kvm_mtrr_get_guest_memory_type

[Alex Williamson <alex.williamson@redhat.com>]

Hi Paolo,

Something bad happened to this patch.  This is the version I provided
Tested-by for:

On Thu, 2015-07-16 at 03:25 +0800, Xiao Guangrong wrote:
> From: Xiao Guangrong <guangrong.xiao@intel.com>
> 
> kvm_mtrr_get_guest_memory_type never returns -1 which is implied
> in the current code since if @type = -1 (means no MTRR contains the
> range), iter.partial_map must be true
> 
> Simplify the code to indicate this fact
> 
> Signed-off-by: Xiao Guangrong <guangrong.xiao@intel.com>
> ---
>  arch/x86/kvm/mtrr.c | 19 ++++++++++---------
>  1 file changed, 10 insertions(+), 9 deletions(-)
> 
> diff --git a/arch/x86/kvm/mtrr.c b/arch/x86/kvm/mtrr.c
> index e275013..9e8bf13 100644
> --- a/arch/x86/kvm/mtrr.c
> +++ b/arch/x86/kvm/mtrr.c
> @@ -672,15 +672,16 @@ u8 kvm_mtrr_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn)
>  	if (iter.mtrr_disabled)
>  		return mtrr_disabled_type();
>  
> -	/* It is not covered by MTRRs. */
> -	if (iter.partial_map) {
> -		/*
> -		 * We just check one page, partially covered by MTRRs is
> -		 * impossible.
> -		 */
> -		WARN_ON(type != -1);
> -		type = mtrr_default_type(mtrr_state);
> -	}
> +	/* not contained in any MTRRs. */
> +	if (type == -1)
> +		return mtrr_default_type(mtrr_state);
> +
> +	/*
> +	 * We just check one page, partially covered by MTRRs is
> +	 * impossible.
> +	 */
> +	WARN_ON(iter.partial_map);
> +
>  	return type;
>  }
>  EXPORT_SYMBOL_GPL(kvm_mtrr_get_guest_memory_type);


This is the version that was committed for v4.2-rc4:

commit 3e5d2fdceda172554e681b68c853bf5d08205bbf
Author: Xiao Guangrong <guangrong.xiao@intel.com>
Date:   Thu Jul 16 03:25:55 2015 +0800

    KVM: MTRR: simplify kvm_mtrr_get_guest_memory_type
    
    kvm_mtrr_get_guest_memory_type never returns -1 which is implied
    in the current code since if @type = -1 (means no MTRR contains the
    range), iter.partial_map must be true
    
    Simplify the code to indicate this fact
    
    Signed-off-by: Xiao Guangrong <guangrong.xiao@intel.com>
    Tested-by: Alex Williamson <alex.williamson@redhat.com>
    Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

diff --git a/arch/x86/kvm/mtrr.c b/arch/x86/kvm/mtrr.c
index e275013..dc0a84a 100644
--- a/arch/x86/kvm/mtrr.c
+++ b/arch/x86/kvm/mtrr.c
@@ -672,15 +672,16 @@ u8 kvm_mtrr_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn)
 	if (iter.mtrr_disabled)
 		return mtrr_disabled_type();
 
-	/* It is not covered by MTRRs. */
-	if (iter.partial_map) {
-		/*
-		 * We just check one page, partially covered by MTRRs is
-		 * impossible.
-		 */
-		WARN_ON(type != -1);
-		type = mtrr_default_type(mtrr_state);
-	}
+	/*
+	 * We just check one page, partially covered by MTRRs is
+	 * impossible.
+	 */
+	WARN_ON(iter.partial_map);
+
+	/* not contained in any MTRRs. */
+	if (type == -1)
+		return mtrr_default_type(mtrr_state);
+
 	return type;
 }
 EXPORT_SYMBOL_GPL(kvm_mtrr_get_guest_memory_type);

The WARN_ON() now comes before the type == -1 test and I hit that at
*very* high frequency when trying to test device assignment.  Restoring
the ordering to what Xiao originally proposed resolves the problem.
Thanks,

Alex