ntdll: Don't modify LockCount in RtlLeaveCriticalSection if section is not acquired.

[Sebastian Lackner]

On 03.03.2017 19:30, Jacek Caban wrote:
> On 03.03.2017 19:13, Sebastian Lackner wrote:
>> On 03.03.2017 18:48, Jacek Caban wrote:
>>> Signed-off-by: Jacek Caban <jacek at codeweavers.com>
>>> @@ -151,6 +154,9 @@ static void InitFunctionPtrs(void)
>>>          pRtlIsCriticalSectionLocked = (void *)GetProcAddress(hntdll, "RtlIsCriticalSectionLocked");
>>>          pRtlIsCriticalSectionLockedByThread = (void *)GetProcAddress(hntdll, "RtlIsCriticalSectionLockedByThread");
>>>          pRtlInitializeCriticalSectionEx = (void *)GetProcAddress(hntdll, "RtlInitializeCriticalSectionEx");
>>> +        pRtlInitializeCriticalSection = (void *)GetProcAddress(hntdll, "RtlInitializeCriticalSection");
>>> +        pRtlEnterCriticalSection = (void *)GetProcAddress(hntdll, "RtlEnterCriticalSection");
>>> +        pRtlLeaveCriticalSection = (void *)GetProcAddress(hntdll, "RtlLeaveCriticalSection");
>> Those functions should not require dynamic loading. The *Ex function is only
>> loaded dynamically because it was introduced in later Windows versions.
> 
> Hmm, sure, I may change it. But what's the convention here? Given that
> basic functions such as NtReadFile are loaded dynamically, I was under
> an impression that we want to load dynamically all functions from ntdll.
> Now I see it's not the case for other functions. I recall considering to
> change that, but I think Alexandre preferred it that way (I'm not sure
> about that, I may remember it wrong).

When we still had a Windows 2000 testbot VM it was required, but I think
it is obsolete now and can be removed. Alexandre might know better if
there is any other reason why we should keep it.

Best regards,
Sebastian