[PATCH 1/2] Hyperv detection - code cleanup

[PATCH 1/2] Hyperv detection - code cleanup

[Ky Srinivasan]

From: K. Y. Srinivasan <ksrinivasan@novell.com>
Subject: Code cleanup - consolidate all hyperv related definitions into one
header file.


Signed-off-by: K. Y. Srinivasan <ksrinivasan@novell.com>

Index: linux.trees.git/arch/x86/include/asm/hyperv.h
===================================================================
--- linux.trees.git.orig/arch/x86/include/asm/hyperv.h	2010-05-18 10:14:02.000000000 -0600
+++ linux.trees.git/arch/x86/include/asm/hyperv.h	2010-05-18 14:11:55.000000000 -0600
@@ -190,4 +190,11 @@
 #define HV_STATUS_INVALID_HYPERCALL_INPUT	3
 #define HV_STATUS_INVALID_ALIGNMENT		4
 
+struct ms_hyperv_info {
+	u32 features;
+	u32 hints;
+};
+
+extern struct ms_hyperv_info ms_hyperv;
+
 #endif
Index: linux.trees.git/arch/x86/include/asm/mshyperv.h
===================================================================
--- linux.trees.git.orig/arch/x86/include/asm/mshyperv.h	2010-05-18 10:14:02.000000000 -0600
+++ /dev/null	1970-01-01 00:00:00.000000000 +0000
@@ -1,14 +0,0 @@
-#ifndef _ASM_X86_MSHYPER_H
-#define _ASM_X86_MSHYPER_H
-
-#include <linux/types.h>
-#include <asm/hyperv.h>
-
-struct ms_hyperv_info {
-	u32 features;
-	u32 hints;
-};
-
-extern struct ms_hyperv_info ms_hyperv;
-
-#endif
Index: linux.trees.git/arch/x86/kernel/cpu/mshyperv.c
===================================================================
--- linux.trees.git.orig/arch/x86/kernel/cpu/mshyperv.c	2010-05-18 12:46:50.000000000 -0600
+++ linux.trees.git/arch/x86/kernel/cpu/mshyperv.c	2010-05-18 14:11:55.000000000 -0600
@@ -15,7 +15,6 @@
 #include <asm/processor.h>
 #include <asm/hypervisor.h>
 #include <asm/hyperv.h>
-#include <asm/mshyperv.h>
 
 struct ms_hyperv_info ms_hyperv;
 EXPORT_SYMBOL(ms_hyperv);

Re: [PATCH 1/2] Hyperv detection - code cleanup

[H. Peter Anvin]

On 05/19/2010 10:12 AM, Ky Srinivasan wrote:
> 
> From: K. Y. Srinivasan <ksrinivasan@novell.com>
> Subject: Code cleanup - consolidate all hyperv related definitions into one
> header file.
> 

How will this affect KVM?  I was under the impression that this was
separated out as a constants-only header file for a reason.

	-hpa

Re: [PATCH 1/2] Hyperv detection - code cleanup

[Ky Srinivasan]

>>> On 5/19/2010 at  1:51 PM, in message <4BF4252E.6090609@zytor.com>, "H. Peter
Anvin" <hpa@zytor.com> wrote: 
> On 05/19/2010 10:12 AM, Ky Srinivasan wrote:
>> 
>> From: K. Y. Srinivasan <ksrinivasan@novell.com>
>> Subject: Code cleanup - consolidate all hyperv related definitions into one
>> header file.
>> 
> 
> How will this affect KVM?  I was under the impression that this was
> separated out as a constants-only header file for a reason.

I was under the impression that hyperv.h was for all public definitions of  Hyperv. I am hoping Avi can shed some light on KVM implications (if any). 

Regards,

K. Y
> 
> 	-hpa

Re: [PATCH 1/2] Hyperv detection - code cleanup

[Avi Kivity]

On 05/19/2010 08:51 PM, H. Peter Anvin wrote:
> On 05/19/2010 10:12 AM, Ky Srinivasan wrote:
>    
>> From: K. Y. Srinivasan<ksrinivasan@novell.com>
>> Subject: Code cleanup - consolidate all hyperv related definitions into one
>> header file.
>>
>>      
> How will this affect KVM?  I was under the impression that this was
> separated out as a constants-only header file for a reason.
>
>    

It's exported, so any types need to be compatible with userspace (i.e. 
__u32).

-- 
Do not meddle in the internals of kernels, for they are subtle and quick to panic.

Re: [PATCH 1/2] Hyperv detection - code cleanup

[H. Peter Anvin]

On 05/19/2010 11:47 AM, Avi Kivity wrote:
> On 05/19/2010 08:51 PM, H. Peter Anvin wrote:
>> On 05/19/2010 10:12 AM, Ky Srinivasan wrote:
>>    
>>> From: K. Y. Srinivasan<ksrinivasan@novell.com>
>>> Subject: Code cleanup - consolidate all hyperv related definitions into one
>>> header file.
>>>
>>>      
>> How will this affect KVM?  I was under the impression that this was
>> separated out as a constants-only header file for a reason.
> 
> It's exported, so any types need to be compatible with userspace (i.e. 
> __u32).
> 

... and certainly no prototypes for kernel-only functionality.
Right, that's what I thought.  So NAK on this patch, too.

	-hpa