Hello,
On 2018-05-30 03:00, David Holmes wrote:
Hi Severin,
On 30/05/2018 7:15 PM, Severin Gehwolf wrote:
Hi,
Could I please get a review of this one-liner fix for a bootcycle-
images build of the Zero variant (on x86_64)? The proposed change is to
use the big java settings (over small) for the jdk.compiler's
annotation processing. Thoughts?
We set a number of the build steps to use "small Java" because we ran
into memory problems with high concurrency builds. I'm unclear what
impact this change has overall. Eril or Magnus should be clearer on that.
I think this is a reasonable change. It won't have a big impact since
it's only one process in the whole build. The big impact is mostly seen
when changing JVM sizing on many JVMs that tend to all run at the same time.
So, looks good.
/Erik
Thanks,
David
diff --git a/make/gensrc/Gensrc-jdk.internal.vm.compiler.gmk
b/make/gensrc/Gensrc-jdk.internal.vm.compiler.gmk
--- a/make/gensrc/Gensrc-jdk.internal.vm.compiler.gmk
+++ b/make/gensrc/Gensrc-jdk.internal.vm.compiler.gmk
@@ -103,7 +103,7 @@
$(GENSRC_DIR)/_gensrc_proc_done: $(PROC_SRCS) $(PROCESSOR_JARS)
$(call MakeDir, $(@D))
$(eval $(call
ListPathsSafely,PROC_SRCS,$(@D)/_gensrc_proc_files))
- $(JAVA_SMALL) $(NEW_JAVAC) \
+ $(JAVA) $(NEW_JAVAC) \
-XDignore.symbol.file \
--upgrade-module-path $(JDK_OUTPUTDIR)/modules --system
none \
$(ADD_EXPORTS) \
Bug: https://bugs.openjdk.java.net/browse/JDK-8203924
Testing: Zero linux-x86_64 bootcycle-images builds post-patch. Fails
before.
Thanks,
Severin
