RFR [11]: Fix PCH build failure

[Roman Kennke]

Ok.

Roman

> There appears to be a weird header dependency issue somewhere in sh/jdk11.
> 
> In file included from
> /home/jenkins/workspace/nightly/shenandoah-jdk11/src/hotspot/share/compiler/compileBroker.cpp:72:0:
> /home/jenkins/workspace/nightly/shenandoah-jdk11/src/hotspot/share/jvmci/jvmciJavaClasses.hpp:260:15:
> error: using typedef-name ‘Value’ after ‘class’
>    start_class(Value)                                                                       \
>                ^
> /home/jenkins/workspace/nightly/shenandoah-jdk11/src/hotspot/share/jvmci/jvmciJavaClasses.hpp:320:7:
> note: in definition of macro ‘START_CLASS’
>  class name : AllStatic {                                                                   \
>        ^
> /home/jenkins/workspace/nightly/shenandoah-jdk11/src/hotspot/share/jvmci/jvmciJavaClasses.hpp:362:1:
> note: in expansion of macro ‘COMPILER_CLASSES_DO’
> 
> I did similar fix on aarch64 when doing backports, and it also fixes to be in x86_64.
> 
> --- a/src/hotspot/cpu/x86/x86_64.ad     Sat Nov 10 14:01:20 2018 +0100
> +++ b/src/hotspot/cpu/x86/x86_64.ad     Mon Nov 12 13:09:03 2018 +0100
> @@ -543,7 +543,7 @@
>  #include "gc/z/zBarrierSetAssembler.hpp"
>  #endif
>  #if INCLUDE_SHENANDOAHGC
> -#include "gc/shenandoah/shenandoahBarrierSetAssembler.hpp"
> +#include "gc/shenandoah/shenandoahBrooksPointer.hpp"
>  #endif
>  %}
> 
> Testing: x86_64 PCH/no-PCH build
> 
> Thanks,
> -Aleksey
>