[llvm] [X86] Resolve FIXME: Add FPCW as a rounding control register (PR #82452)

[via llvm-commits]

AtariDreams wrote:

> > In tests, this seems to only cause the wait instruction to disappear, which I do not see as necessary in these particular cases.
> 
> I think the wait instructions are intended. The affected tests are all for constrained intrinsics which care the precise exception. Per to SDM
> 
> > Synchronization problems occur in the time between the moment when the exception is signaled and when it is
> > actually handled. Because of concurrent execution, integer or system instructions can be executed during this
> > time.
> 
> We don't know what instructions a function call may execute, so should emit `wait` for conservative.

In that case, let's fix the wait instruction to do that.

https://github.com/llvm/llvm-project/pull/82452