EVALUATION
http://hg.openjdk.java.net/jdk7/hotspot/hotspot/rev/fb57d4cf76c2
|
|
|
SUGGESTED FIX
--- a/make/windows/makefiles/defs.make Fri May 07 15:13:00 2010 -0700
+++ b/make/windows/makefiles/defs.make Mon May 10 16:26:50 2010 +0200
@@ -55,7 +55,7 @@
MAKE_ARGS += Platform_arch_model=ia64
endif
-ifneq ($(shell $(ECHO) $(PROCESSOR_IDENTIFIER) | $(GREP) AMD64),)
+ifneq ($(shell $(ECHO) $(PROCESSOR_IDENTIFIER) | $(GREP) "AMD64\|Intel64"),)
ARCH_DATA_MODEL=64
PLATFORM=windows-amd64
VM_PLATFORM=windows_amd64
--- a/src/share/vm/gc_implementation/g1/heapRegion.cpp Fri May 07 15:13:00 2010 -0700
+++ b/src/share/vm/gc_implementation/g1/heapRegion.cpp Mon May 10 16:26:50 2010 +0200
@@ -290,7 +290,7 @@
// Recalculate the region size to make sure it's a power of
// 2. This means that region_size is the largest power of 2 that's
// <= what we've calculated so far.
- region_size = 1 << region_size_log;
+ region_size = (uintx)1 << region_size_log;
|
|
|
PUBLIC COMMENTS
Fix provided by Staffan Larsen.
|
|
|
EVALUATION
Suggested fix is ok.
|
|
|
SUGGESTED FIX
The change to heapRegion.cpp is also being made in a slightly different
way by Phil Race as part of the VS2010 work.
|
|
|
EVALUATION
Please note, later versions of Cygwin do not set PROCESSOR_IDENTIFIER at all!
http://www.gossamer-threads.com/lists/openssh/dev/48152
We may want to consider removing the dependency now.
|
|
|
EVALUATION
The suggested fix string "Intel64" matches what I see on Window 7.
Maybe the same on Vista/2008 but it won;t work for JDK's supported
build platform : Windows 2003 x64, where I see
"EM64T Family 6 Model 26 Stepping 5, GenuineIntel"
So lets look for both.
Regarding cygwin, the most recent shipping cygwin still leaves
that var alone. I know because that's what I am using right now.
[Let me qualify that. I'm not using ssh .. so maybe I'm
just not hitting it].
However due to the slow performance of 64 bit cygwin we'll
continue to use MKS for official builds.
And it looks to me as if the patch is ssh specific.
So I don't intend to hold up the VS2010 work that will include
this fix in order to find a new way to do this.
However I *do* agree that this should be made less fragile, but
not in a cygwin specific way.
|
|
|