RFR 8239782: CC_INTERP is only used by Zero interpreter

Magnus Ihse Bursie magnus.ihse.bursie at oracle.com
Tue Jun 23 09:54:16 UTC 2020

On 2020-06-23 06:17, David Holmes wrote:
> Hi Coleen,
> Cleanup is looking good but a few comments:
> - if the bytecodeInterpreter is also zero-only can we rename its files 
> too? (I really find it hard to figure out which files are really 
> needed/used for a given build.)
> - you are excluding shared templateInterpreter*.* from the zero build so
> "#ifndef ZERO" is always true in those files.
> - are the platform specific templateInterpreter* files already 
> excluded from a zero build? Otherwise they should be added to the 
> exclude list.
> - how were you able to completely delete:
>   - src/hotspot/share/interpreter/cppInterpreter.cpp
>   - src/hotspot/share/interpreter/cppInterpreterGenerator.cpp
It is inlined into src/hotspot/cpu/zero/zeroInterpreter_zero.cpp.

Anyway, build changes look good.

Coleen, in the future, please cc build-dev at ojn whenever you make changes 
to the makefiles.

> ?
> Thanks,
> David
> ----
> On 23/06/2020 1:36 am, coleen.phillimore at oracle.com wrote:
>> Summary: Change CC_INTERP conditional to ZERO and remove in places 
>> where unnecessary. Fix build to exclude compilers and rename 
>> CppInterpreter to ZeroInterpreter. The "C++ Interpreter" has been 
>> removed from the code a while ago.
>> The motivation is to remove CC_INTERP conditionals from common code 
>> for the most part.  The C++ interpreter used to work with C1 and C2.  
>> Some of the hooks are still present (can be cleaned out or 
>> implemented correctly later) but I removed some other unconditionally 
>> false code in order to remove interactions with common code.  Also it 
>> appeared that Zero was creating method counters when it was never 
>> using them.  I removed this too, hoping it would make zero faster, 
>> but nope, it's still slow.
>> I also renamed cppInterpreter and CppInterpreter to zeroInterpreter 
>> and ZeroInterpreter, respectively, and moved some code to cpu/zero. 
>> Thus ends pass 10? of cleaning up this code.
>> Tested with tier1 on Oracle platforms and built these: 
>> linux-arm32,linux-ppc64le-debug,linux-s390x-debug,linux-x64-zero,linux-x64-zero-debug. 
>> If you work on Zero, can you give this a test run with your favorite 
>> platform and review?
>> open webrev at 
>> http://cr.openjdk.java.net/~coleenp/2020/8239782.01/webrev
>> bug link https://bugs.openjdk.java.net/browse/JDK-8239782
>> Thanks,
>> Coleen

More information about the hotspot-runtime-dev mailing list