Odd behavior in fiber unittest - printf stops optimization crash

[Kai Nacke via digitalmars-d-ldc]

On Sunday, 1 March 2015 at 13:35:54 UTC, Kai Nacke wrote:
> A context switch with fiber_switchContext looks like a normal 
> function call to the compiler. The nonvolatile registers must 
> be preserved. That RDI/RSI are not saved is a bug. Compare with 
> the function from druntime 2.067: the bug is fixed there.

I pushed this to master. Maybe you wanna try it.

Regards,
Kai