RFR 8203454: Minimal, Zero builds fail after JDK-8202377 (Modularize C2 barriers)

[Erik Österlund]

Hi Aleksey,

Looks good.

Thanks,
/Erik

On 2018-05-19 13:52, Aleksey Shipilev wrote:
> Okay, this also works for Zero, Minimal and server:
>
> diff -r 5ec7380f671d make/hotspot/lib/JvmFeatures.gmk
> --- a/make/hotspot/lib/JvmFeatures.gmk	Fri May 18 18:31:28 2018 -0700
> +++ b/make/hotspot/lib/JvmFeatures.gmk	Sat May 19 13:36:32 2018 +0200
> @@ -41,7 +41,7 @@
>   else
>     JVM_EXCLUDES += opto libadt
>     JVM_EXCLUDE_FILES += bcEscapeAnalyzer.cpp ciTypeFlow.cpp
> -  JVM_EXCLUDE_PATTERNS += c2_ runtime_
> +  JVM_EXCLUDE_PATTERNS += c2_ runtime_ /c2/
>   endif
>
>   ifeq ($(call check-jvm-feature, zero), true)
>
>
> -Aleksey
>
> On 05/19/2018 12:50 PM, Erik Osterlund wrote:
>> Hi Aleksey,
>>
>> Thanks for fixing.
>>
>> I wonder if /c2/ could be added to the exclude pattern instead though, instead of enumerating all GCs.
>>
>> Thanks,
>> /Erik
>>
>>> On 19 May 2018, at 11:20, Aleksey Shipilev <shade at redhat.com> wrote:
>>>
>>> Bug:
>>>   https://bugs.openjdk.java.net/browse/JDK-8203454
>>>
>>> Other GCs that would eventually bring their own BarrierSetC2 files (e.g. Epsilon, Shenandoah, ZGC)
>>> would require excluding their "gc/<name>/c2" as well.
>>>
>>> Fix:
>>>
>>> diff -r 5ec7380f671d make/hotspot/lib/JvmFeatures.gmk
>>> --- a/make/hotspot/lib/JvmFeatures.gmk    Fri May 18 18:31:28 2018 -0700
>>> +++ b/make/hotspot/lib/JvmFeatures.gmk    Sat May 19 11:01:16 2018 +0200
>>> @@ -41,7 +41,7 @@
>>> else
>>>    JVM_EXCLUDES += opto libadt
>>>    JVM_EXCLUDE_FILES += bcEscapeAnalyzer.cpp ciTypeFlow.cpp
>>> -  JVM_EXCLUDE_PATTERNS += c2_ runtime_
>>> +  JVM_EXCLUDE_PATTERNS += c2_ runtime_ gc/shared/c2 gc/g1/c2
>>> endif
>>>
>>> ifeq ($(call check-jvm-feature, zero), true)
>>>
>>> Testing: {x86-minimal, x86_64-zero, x86_64-server} builds
>>>
>>> Thanks,
>>> -Aleksey
>>>
>