ntdll: On macOS, don't use clock_gettime() even if it's available.

[Ken Thomases]

On Jun 1, 2017, at 3:19 PM, Michael Müller <michael at fds-team.de> wrote:
> 
> Not sure which one of two commands is causing the lower overhead, but
> shouldn't you just build Wine using the 10.11 SDK if you plan to deploy
> it on 10.11?

No.  That's not the way Mac software is typically built.  The whole point of a deployment target[1] is that it's often different from the SDK against which you're building.  You typically build against the newest SDK.  (Apple used to ship a few old SDKs with each version of Xcode but stopped doing that and encourages everyone to use the latest.)

[1] https://developer.apple.com/library/content/documentation/DeveloperTools/Conceptual/cross_development/Configuring/configuring.html#//apple_ref/doc/uid/10000163i-CH1-SW1

Ideally, the code would check for the run-time availability of clock_gettime() and use it when it's available (assuming a build against the 10.12 SDK) and fall back to mach_absolute_time() otherwise.  Alexandre expressed a preference for the approach in my patch, instead.

> This could theoretically also apply to other parts of Wine
> and I don't think it makes sense to add backwards compatibility hacks
> everywhere.

Theoretically, it could.  In practice, it's not been a big problem.

-Ken