Status of building problem with GCC 6

[David Holmes]

Please include the build team on this discussion - cc'd

Thanks,
David

On 4/07/2016 2:36 AM, Kim Barrett wrote:
>> On Jul 1, 2016, at 3:13 PM, Kim Barrett <kim.barrett at oracle.com> wrote:
>>
>>> On Jul 1, 2016, at 10:17 AM, Yasumasa Suenaga <yasuenag at gmail.com> wrote:
>>> For HotSpot, I think JDK-8156980 should be fixed at first.
>>> I've proposed changes as below:
>>>
>>> -----------
>>> diff -r ba08710f3b6c make/lib/CompileJvm.gmk
>>> --- a/make/lib/CompileJvm.gmk   Mon Jun 27 09:35:18 2016 +0200
>>> +++ b/make/lib/CompileJvm.gmk   Tue Jun 28 12:10:09 2016 +0900
>>> @@ -187,6 +187,11 @@
>>>
>>> JVM_OPTIMIZATION ?= HIGHEST_JVM
>>>
>>> +JVM_CXXFLAGS := $(JVM_CFLAGS)
>>> +ifeq ($(TOOLCHAIN_TYPE), gcc)
>>> +  JVM_CXXFLAGS += -std=gnu++98 -fno-delete-null-pointer-checks -fno-lifetime-dse
>>> +endif
>>> +
>>> ################################################################################
>>> # Now set up the actual compilation of the main hotspot native library
>>>
>>> @@ -202,6 +207,7 @@
>>>     CFLAGS := $(JVM_CFLAGS), \
>>>     CFLAGS_DEBUG_SYMBOLS := $(JVM_CFLAGS_SYMBOLS), \
>>>     CXXFLAGS_DEBUG_SYMBOLS := $(JVM_CFLAGS_SYMBOLS), \
>>> +    CXXFLAGS := $(JVM_CXXFLAGS), \
>>>     vm_version.cpp_CXXFLAGS := $(CFLAGS_VM_VERSION), \
>>>     DISABLED_WARNINGS_clang := delete-non-virtual-dtor dynamic-class-memaccess \
>>>         empty-body format logical-op-parentheses parentheses \
>>> —————
>>
>> Please keep in mind that my understanding of the new build system is
>> pretty weak.  That said, this doesn't look to me like the right way to
>> fix JDK-8156980.
>>
>> There is already code that is supposed to deal with both the -std
>> option and the additional code generation options, which seems to work
>> for some of the packages that we build, but is not affecting the
>> Hotspot build for some reason.  Adding a completely separate way to
>> deal with this just for Hotspot seems contrary to the unification
>> effort that was part of the new build system.
>>
>> -fno-lifetime-dse is a relatively recent option, and needs to be
>> conditionalized.  And discussion during the review of JDK-8151841 led
>> to the additional code generation options being limited to gcc6+.
>> The existing code mentioned above is conditionalizing that way, except
>> it's just not working for Hotspot.
>
> I looked into why the additional code generation options aren’t being
> added, and think I found 3-4 different problems, each of which would
> interfere with that part.  I’m working on a patch, which I hope to send
> to Yasumasa in next couple of days for a gcc6 test, since I don’t have
> easy access to that compiler right now.
>
> I’ve not (re)explored why -std=gnu++98 isn’t making it to Hotspot builds
> yet, though what I’ve learned in the process of investigating the code
> generation options problem has given me a possible lead.
>
>