Updated review with a regression test. I'm not familiar with jtreg so
tried pattern matching on existing tests and translated the reproducer
from the bug report. I suspect that this test should not be run in the
same jvm instance as other tests as it destructively changes the system
properties. Is that something that can be expressed to jtreg?
http://cr.openjdk.java.net/~erikj/8030781/webrev.01/
/Erik
On 2014-01-07 14:34, Erik Joelsson wrote:
Please review this late fix for JDK 8.
System.c needs to be compiled with the macro RELEASE set to something
meaningful. In the new build, since JDK-8001895, it gets set to the
empty string. The reason for this is that the variable RELEASE is only
defined in the makefiles, in spec.gmk, while the CFLAGS_JDKLIB
variable is defined in configure. The fix is simple and posted below
inline. With this fix, the value for RELEASE is evaluated in the
makefile rather than in configure.
diff -r 1e1f86d5d4e2 common/autoconf/toolchain.m4
--- a/common/autoconf/toolchain.m4
+++ b/common/autoconf/toolchain.m4
@@ -936,7 +936,7 @@
fi
CCXXFLAGS_JDK="$CCXXFLAGS_JDK
-DARCH='\"$OPENJDK_TARGET_CPU_LEGACY\"' -D$OPENJDK_TARGET_CPU_LEGACY"
- CCXXFLAGS_JDK="$CCXXFLAGS_JDK -DRELEASE='\"$RELEASE\"'"
+ CCXXFLAGS_JDK="$CCXXFLAGS_JDK -DRELEASE='\"\$(RELEASE)\"'"
CCXXFLAGS_JDK="$CCXXFLAGS_JDK \
-I${JDK_OUTPUTDIR}/include \
/Erik
