ntdll: Add NtCreateKey tests

[Robert Shearman]

James Hawkins wrote:

>On 03 Mar 2005 11:02:24 +0100, Alexandre Julliard <julliard at winehq.org> wrote:
>  
>
>>James Hawkins <truiken at gmail.com> writes:
>>
>>    
>>
>>>I've been discerning the behavior of RegCreateKey and NtCreateKey when
>>>creating a key directly under HKLM or HKU, and this test reveals that
>>>NtCreateKey does check for this case and returns a
>>>STATUS_INVALID_PARAMETER.
>>>      
>>>
>>This is going to break registry initialisation. Do you really have an
>>app that depends on that?
>>
>>    
>>
>
>What are the steps I need to take to break registry initialisation?  I
>did wonder how anything in the real win32 registry could be created
>though if you can't create a key directly under HKLM or HKU.  If this
>really does break initialisation, can we implement a different
>create_key that has little restrictions and would only be used
>internally, or would this just be scrapped entirely?
>  
>

It looks like Windows only uses RegLoadKey to create keys under HKLM and 
HKU. Maybe we should do the same.

Rob